From da46162f31a24fa1690aca8fbe72cf892f360a4a Mon Sep 17 00:00:00 2001
From: redoules <guillaume.redoules@gadz.org>
Date: Sun, 16 Dec 2018 09:37:42 +0100
Subject: [PATCH] added article

added an article on the mayer multiple
---
 blog/Statistics_10days-day0.html              |   2 +-
 blog/Statistics_10days-day1.html              |   2 +-
 blog/Statistics_10days-day2.html              |   2 +-
 blog/Statistics_10days-day3.html              |   2 +-
 blog/Statistics_10days-day4.html              |   2 +-
 blog/Statistics_10days-day5.html              |   2 +-
 blog/Statistics_10days-day6.html              |   2 +-
 blog/Statistics_10days-day7.html              |   2 +-
 blog/Statistics_10days-day8.html              |   2 +-
 blog/Statistics_10days-day9.html              |   2 +-
 .../Installing_bitcoind_on_raspberry_pi.html  |   2 +-
 cryptocurrencies/blockchain_bad.html          |   2 +-
 cryptocurrencies/mayer_multiple.html          | 414 ++++++++++++++++++
 dl/csv_logger_keras.html                      |   2 +-
 dl/keras_info.html                            |   2 +-
 dl/keras_input.html                           |  10 +-
 dl/keras_install.html                         |   2 +-
 dl/keras_load.html                            |  10 +-
 dl/model_checkpoint_keras.html                |   6 +-
 dl/tensorboard_keras.html                     |   6 +-
 images/mayer_multiple/output_5_1.png          | Bin 0 -> 214503 bytes
 images/mayer_multiple/output_7_1.png          | Bin 0 -> 92772 bytes
 index.html                                    |   4 +-
 jupyter/clear_cell.html                       |   2 +-
 jupyter/remote_run_notebook.html              |   2 +-
 linux/Reloading_.bashrc.html                  |   2 +-
 linux/Reloading_fstab.html                    |   2 +-
 linux/bashrc.html                             |   2 +-
 linux/directory_size.html                     |   2 +-
 linux/free_disk_space_linux.html              |   2 +-
 linux/get_ip_linux.html                       |   2 +-
 linux/linux_process_information.html          |   2 +-
 linux/mount_nfs_share_fstab.html              |   2 +-
 linux/share_nfs_share.html                    |   2 +-
 linux/simple_bash_function.html               |   2 +-
 linux/version_kernel.html                     |   2 +-
 .../LeastSquareRegressionLine.html            |   2 +-
 .../Multiple_Linear_Regression.html           |   2 +-
 ...he recommandation engine for articles.html |   2 +-
 mathematics/Bayes.html                        |   2 +-
 mathematics/Binomial.html                     |   2 +-
 mathematics/Geometric.html                    |   2 +-
 mathematics/Number_edges_Complete_graph.html  |   2 +-
 .../Premutations_and_combinations.html        |   2 +-
 mathematics/cond_prob.html                    |   2 +-
 mathematics/normal.html                       |   2 +-
 mathematics/pearson.html                      |   2 +-
 mathematics/poisson.html                      |   2 +-
 mathematics/spearman.html                     |   2 +-
 pages/about.html                              |   2 +-
 python/Creating_a_sqlite_database.html        |   2 +-
 ...of_eigenvalues_from_a_list_of_tensors.html |   2 +-
 python/Iterating_over_a_dataframe.html        |   2 +-
 python/Moving_average_pandas.html             |   2 +-
 python/Opening_SQLite_database.html           |   2 +-
 python/Opening_file.html                      |   2 +-
 python/Optimized_numpy_random_intel.html      |   2 +-
 ..._data_from_a_sql_database_with_pandas.html |   2 +-
 ...plotlib_figure_with_a_high_resolution.html |   2 +-
 ...notebook_for_plotting_with_matplotlib.html |   2 +-
 ...ng_data_to_a_sql_database_with_pandas.html |   2 +-
 python/counting.html                          |   2 +-
 python/dask_distributed_parallelism.html      |   2 +-
 python/dask_infiniband.html                   |   2 +-
 python/design_own_libs.html                   |   2 +-
 python/dict_comprehension.html                |   2 +-
 python/download_page.html                     |   2 +-
 python/ensure_dir.html                        |   2 +-
 python/list_files_directory.html              |   2 +-
 python/logplot.html                           |   2 +-
 python/randint.html                           |   2 +-
 python/reverse.html                           |   2 +-
 python/sorting.html                           |   2 +-
 python/stock_pandas.html                      |   2 +-
 python/unique.html                            |   2 +-
 ...ting_all_python_package_with_anaconda.html |   2 +-
 search.html                                   |   2 +-
 sitemap.xml                                   |   9 +-
 sql/Sorting_results.html                      |   2 +-
 sql/WHERE_SQL_keywords.html                   |   2 +-
 sql/display_table.html                        |   2 +-
 sql/display_table_filter.html                 |   2 +-
 tipuesearch_content.json                      |   2 +-
 83 files changed, 504 insertions(+), 103 deletions(-)
 create mode 100644 cryptocurrencies/mayer_multiple.html
 create mode 100644 images/mayer_multiple/output_5_1.png
 create mode 100644 images/mayer_multiple/output_7_1.png

diff --git a/blog/Statistics_10days-day0.html b/blog/Statistics_10days-day0.html
index d4f2005..7cedc93 100644
--- a/blog/Statistics_10days-day0.html
+++ b/blog/Statistics_10days-day0.html
@@ -185,7 +185,7 @@ Mean : 43117.75, Median : 44627.5, Mode : 4978
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/blog/Statistics_10days-day1.html b/blog/Statistics_10days-day1.html
index 9437a6f..b339772 100644
--- a/blog/Statistics_10days-day1.html
+++ b/blog/Statistics_10days-day1.html
@@ -219,7 +219,7 @@ Q1 : 6.0, Q2 : 12, Q3 : 16.0
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/blog/Statistics_10days-day2.html b/blog/Statistics_10days-day2.html
index 95f165f..d1c9115 100644
--- a/blog/Statistics_10days-day2.html
+++ b/blog/Statistics_10days-day2.html
@@ -429,7 +429,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/blog/Statistics_10days-day3.html b/blog/Statistics_10days-day3.html
index 358ddc3..956f457 100644
--- a/blog/Statistics_10days-day3.html
+++ b/blog/Statistics_10days-day3.html
@@ -256,7 +256,7 @@ we call B the event "a blue ball is drawn" and R the event "a red ball is drawn"
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/blog/Statistics_10days-day4.html b/blog/Statistics_10days-day4.html
index 8a272cb..694eaea 100644
--- a/blog/Statistics_10days-day4.html
+++ b/blog/Statistics_10days-day4.html
@@ -311,7 +311,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/blog/Statistics_10days-day5.html b/blog/Statistics_10days-day5.html
index c21d4b5..5ce76a3 100644
--- a/blog/Statistics_10days-day5.html
+++ b/blog/Statistics_10days-day5.html
@@ -331,7 +331,7 @@ Between 20 and 22 hours?</p>
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/blog/Statistics_10days-day6.html b/blog/Statistics_10days-day6.html
index a1fc83e..3622cbc 100644
--- a/blog/Statistics_10days-day6.html
+++ b/blog/Statistics_10days-day6.html
@@ -270,7 +270,7 @@ B = 515.68
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/blog/Statistics_10days-day7.html b/blog/Statistics_10days-day7.html
index 6c6336f..167cb2b 100644
--- a/blog/Statistics_10days-day7.html
+++ b/blog/Statistics_10days-day7.html
@@ -272,7 +272,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/blog/Statistics_10days-day8.html b/blog/Statistics_10days-day8.html
index 53f14c1..5f443b4 100644
--- a/blog/Statistics_10days-day8.html
+++ b/blog/Statistics_10days-day8.html
@@ -277,7 +277,7 @@ $$</div>
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/blog/Statistics_10days-day9.html b/blog/Statistics_10days-day9.html
index 529d64a..331b950 100644
--- a/blog/Statistics_10days-day9.html
+++ b/blog/Statistics_10days-day9.html
@@ -275,7 +275,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/cryptocurrencies/Installing_bitcoind_on_raspberry_pi.html b/cryptocurrencies/Installing_bitcoind_on_raspberry_pi.html
index 5c69191..9c2081a 100644
--- a/cryptocurrencies/Installing_bitcoind_on_raspberry_pi.html
+++ b/cryptocurrencies/Installing_bitcoind_on_raspberry_pi.html
@@ -286,7 +286,7 @@ maxuploadtarget=5000
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/cryptocurrencies/blockchain_bad.html b/cryptocurrencies/blockchain_bad.html
index 9a25365..930e70d 100644
--- a/cryptocurrencies/blockchain_bad.html
+++ b/cryptocurrencies/blockchain_bad.html
@@ -161,7 +161,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/cryptocurrencies/mayer_multiple.html b/cryptocurrencies/mayer_multiple.html
new file mode 100644
index 0000000..0b6e32a
--- /dev/null
+++ b/cryptocurrencies/mayer_multiple.html
@@ -0,0 +1,414 @@
+<!DOCTYPE html>
+<html lang="fr">
+
+<head>
+      <meta charset="utf-8">
+    <meta http-equiv="X-UA-Compatible" content="IE=edge">
+    <meta name="viewport" content="width=device-width, initial-scale=1">
+    <!-- The above 3 meta tags *must* come first in the head; any other head content must come *after* these tags -->
+    <meta name="description" content="Data Science for Political and Social Phenomena">
+    <meta name="author" content="Guillaume Redoulès">
+    <link rel="icon" href="../favicon.ico">
+
+    <title>Computing the Mayer multiple - Cryptocurrencies</title>
+
+    <!-- JQuery -->
+    <script src="https://ajax.googleapis.com/ajax/libs/jquery/1.11.3/jquery.min.js"></script>
+    <script>
+        window.jQuery || document.write('<script src="../theme/js/jquery.min.js"><\/script>')
+    </script>
+
+    <!-- Bootstrap core CSS -->
+    <link rel="stylesheet" href="../theme/css/bootstrap.css" />
+    <!-- IE10 viewport hack for Surface/desktop Windows 8 bug -->
+    <link rel="stylesheet" type="text/css" href="../theme/css/ie10-viewport-bug-workaround.css" />
+    <!-- Custom styles for this template -->
+    <link rel="stylesheet" type="text/css" href="../theme/css/style.css" />
+    <link rel="stylesheet" type="text/css" href="../theme/css/notebooks.css" />
+    <link href='https://fonts.googleapis.com/css?family=PT+Serif:400,700|Roboto:400,500,700' rel='stylesheet' type='text/css'>
+
+    <!-- HTML5 shim and Respond.js for IE8 support of HTML5 elements and media queries -->
+    <!--[if lt IE 9]>
+      <script src="https://oss.maxcdn.com/html5shiv/3.7.2/html5shiv.min.js"></script>
+      <script src="https://oss.maxcdn.com/respond/1.4.2/respond.min.js"></script>
+    <![endif]-->
+        
+
+    <meta name="tags" content="Visualization" />
+
+
+</head>
+
+<body>
+
+    <div class="navbar navbar-fixed-top">
+        <div class="container">
+            <div class="navbar-header">
+                <button type="button" class="navbar-toggle" data-toggle="collapse" data-target=".navbar-collapse">
+                    <span class="icon-bar"></span>
+                    <span class="icon-bar"></span>
+                    <span class="icon-bar"></span>
+                </button>
+                <a class="navbar-brand" href="..">Guillaume Redoulès</a>
+            </div>
+            <div class="navbar-collapse collapse" id="searchbar">
+
+                <ul class="nav navbar-nav navbar-right">
+                    <li class="dropdown">
+                        <a href="#" class="dropdown-toggle" data-toggle="dropdown" role="button" aria-haspopup="true" aria-expanded="false">About<span class="caret"></span></a>
+                        <ul class="dropdown-menu">
+                            <li><a href="../pages/about.html">About Guillaume</a></li>
+                            <li><a href="https://github.com/redoules">GitHub</a></li>
+                            <li><a href="https://www.linkedin.com/in/guillaume-redoul%C3%A8s-33923860/">LinkedIn</a></li>
+                        </ul>
+                    </li>
+                    <li class="dropdown">
+                        <a href="#" class="dropdown-toggle" data-toggle="dropdown" role="button" aria-haspopup="true" aria-expanded="false">Data Science<span class="caret"></span></a>
+                        <ul class="dropdown-menu">
+                            <li><a href="..#Blog">Blog</a></li>
+                            <li><a href="..#Python">Python</a></li>
+                            <li><a href="..#Bash">Bash</a></li>
+                            <li><a href="..#SQL">SQL</a></li>
+                            <li><a href="..#Mathematics">Mathematics</a></li>
+                            <li><a href="..#Machine_Learning">Machine Learning</a></li>
+                            <li><a href="..#Projects">Projects</a></li>
+                        </ul>
+                    </li>
+                    <li class="dropdown">
+                        <a href="#" class="dropdown-toggle" data-toggle="dropdown" role="button" aria-haspopup="true" aria-expanded="false">Projects<span class="caret"></span></a>
+                        <ul class="dropdown-menu">
+                            <li><a href="https://github.com/redoules/redoules.github.io">Notes (Github)</a></li>
+                        </ul>
+                    </li>
+
+                    <!--<li class="dropdown">
+                        <a href="../feeds/blog.rss.xml">Blog RSS</a>
+                    </li>-->
+
+
+                </ul>
+
+                <form class="navbar-form" action="../search.html" onsubmit="return validateForm(this.elements['q'].value);">
+                    <div class="form-group" style="display:inline;">
+                        <div class="input-group" style="display:table;">
+                            <span class="input-group-addon" style="width:1%;"><span class="glyphicon glyphicon-search"></span></span>
+                            <input class="form-control search-query" name="q" id="tipue_search_input" placeholder="e.g. scikit KNN, pandas merge" required autocomplete="off" type="text">
+                        </div>
+                    </div>
+                </form>
+
+            </div>
+            <!--/.nav-collapse -->
+        </div>
+    </div>
+
+
+
+    <!-- end of header section -->
+    <div class="container">
+<!-- <div class="alert alert-warning" role="alert">
+    Did you find this page useful? Please do me a quick favor and <a href="#" class="alert-link">endorse me for data science on LinkedIn</a>.
+</div> -->
+<section id="content" class="body">
+    <header>
+    <h1>
+      Computing the Mayer multiple
+    </h1>
+<ol class="breadcrumb">
+    <li>
+        <time class="published" datetime="2018-12-16T08:51:00+01:00">
+            16 décembre 2018
+        </time>
+    </li>
+    <li>Cryptocurrencies</li>
+    <li>Visualization</li>
+</ol>
+</header>
+<div class='article_content'>
+<p>I've learnt about the Mayer mutliple from <a href="https://www.theinvestorspodcast.com/bitcoin-mayer-multiple/">The Inverstor Podcast</a>. The Mayer multiple is the ratio of the bitcoin price divided by the 200-day moving average. It is designed to understand the price of bitcoin without taking in account the short term volatility. It helps investors filter out their emotions during a bull run.</p>
+<p>Let's see how to compute the Mayer mutliple in python.</p>
+<p>First, we need to import the data, we will use Quandl to download data from coinbase</p>
+<div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">quandl</span>
+<span class="n">btc_price_data</span> <span class="o">=</span> <span class="n">quandl</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s2">&quot;BCHARTS/COINBASEEUR&quot;</span><span class="p">)</span>
+<span class="n">btc_price_data</span><span class="o">.</span><span class="n">tail</span><span class="p">()</span>
+</pre></div>
+
+
+<div>
+<style scoped>
+    .dataframe tbody tr th:only-of-type {
+        vertical-align: middle;
+    }
+
+    .dataframe tbody tr th {
+        vertical-align: top;
+    }
+
+    .dataframe thead th {
+        text-align: right;
+    }
+</style>
+<table border="1" class="dataframe">
+  <thead>
+    <tr style="text-align: right;">
+      <th></th>
+      <th>Open</th>
+      <th>High</th>
+      <th>Low</th>
+      <th>Close</th>
+      <th>Volume (BTC)</th>
+      <th>Volume (Currency)</th>
+      <th>Weighted Price</th>
+    </tr>
+    <tr>
+      <th>Date</th>
+      <th></th>
+      <th></th>
+      <th></th>
+      <th></th>
+      <th></th>
+      <th></th>
+      <th></th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <th>2018-12-12</th>
+      <td>2966.00</td>
+      <td>3076.71</td>
+      <td>2952.05</td>
+      <td>3026.00</td>
+      <td>1447.627465</td>
+      <td>4.372890e+06</td>
+      <td>3020.728514</td>
+    </tr>
+    <tr>
+      <th>2018-12-13</th>
+      <td>3025.19</td>
+      <td>3028.06</td>
+      <td>2861.15</td>
+      <td>2886.91</td>
+      <td>2125.242928</td>
+      <td>6.261750e+06</td>
+      <td>2946.369017</td>
+    </tr>
+    <tr>
+      <th>2018-12-14</th>
+      <td>2886.91</td>
+      <td>2919.00</td>
+      <td>2800.32</td>
+      <td>2835.50</td>
+      <td>2527.558347</td>
+      <td>7.256959e+06</td>
+      <td>2871.134083</td>
+    </tr>
+    <tr>
+      <th>2018-12-15</th>
+      <td>2835.49</td>
+      <td>2865.00</td>
+      <td>2781.47</td>
+      <td>2830.45</td>
+      <td>1267.004758</td>
+      <td>3.568614e+06</td>
+      <td>2816.575409</td>
+    </tr>
+    <tr>
+      <th>2018-12-16</th>
+      <td>2830.45</td>
+      <td>2830.45</td>
+      <td>2830.44</td>
+      <td>2830.45</td>
+      <td>0.144249</td>
+      <td>4.082886e+02</td>
+      <td>2830.447385</td>
+    </tr>
+  </tbody>
+</table>
+</div>
+
+<p>Next we need to compute the 200 days moving average of the price of bitcoin</p>
+<div class="highlight"><pre><span></span><span class="n">moving_averages</span> <span class="o">=</span> <span class="n">btc_price_data</span><span class="p">[[</span><span class="s2">&quot;Open&quot;</span> <span class="p">,</span><span class="s2">&quot;High&quot;</span> <span class="p">,</span><span class="s2">&quot;Low&quot;</span><span class="p">,</span><span class="s2">&quot;Close&quot;</span><span class="p">]]</span><span class="o">.</span><span class="n">rolling</span><span class="p">(</span><span class="n">window</span><span class="o">=</span><span class="mi">200</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">()</span>
+<span class="n">moving_averages</span><span class="o">.</span><span class="n">tail</span><span class="p">()</span>
+</pre></div>
+
+
+<div>
+<style scoped>
+    .dataframe tbody tr th:only-of-type {
+        vertical-align: middle;
+    }
+
+    .dataframe tbody tr th {
+        vertical-align: top;
+    }
+
+    .dataframe thead th {
+        text-align: right;
+    }
+</style>
+<table border="1" class="dataframe">
+  <thead>
+    <tr style="text-align: right;">
+      <th></th>
+      <th>Open</th>
+      <th>High</th>
+      <th>Low</th>
+      <th>Close</th>
+    </tr>
+    <tr>
+      <th>Date</th>
+      <th></th>
+      <th></th>
+      <th></th>
+      <th></th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <th>2018-12-12</th>
+      <td>5507.88295</td>
+      <td>5611.72570</td>
+      <td>5380.60150</td>
+      <td>5491.43610</td>
+    </tr>
+    <tr>
+      <th>2018-12-13</th>
+      <td>5491.44155</td>
+      <td>5595.14440</td>
+      <td>5363.77840</td>
+      <td>5474.38560</td>
+    </tr>
+    <tr>
+      <th>2018-12-14</th>
+      <td>5474.39910</td>
+      <td>5577.89585</td>
+      <td>5347.27500</td>
+      <td>5457.99125</td>
+    </tr>
+    <tr>
+      <th>2018-12-15</th>
+      <td>5457.94930</td>
+      <td>5559.49145</td>
+      <td>5330.78235</td>
+      <td>5439.75570</td>
+    </tr>
+    <tr>
+      <th>2018-12-16</th>
+      <td>5439.71660</td>
+      <td>5540.89425</td>
+      <td>5313.62955</td>
+      <td>5422.17700</td>
+    </tr>
+  </tbody>
+</table>
+</div>
+
+<p>Finally, we can compute the ratio and plot it.</p>
+<div class="highlight"><pre><span></span><span class="o">%</span><span class="n">matplotlib</span> <span class="n">inline</span> 
+<span class="kn">import</span> <span class="nn">matplotlib.pyplot</span> <span class="kn">as</span> <span class="nn">plt</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">rcParams</span><span class="p">[</span><span class="s1">&#39;savefig.dpi&#39;</span><span class="p">]</span> <span class="o">=</span> <span class="mi">300</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">rcParams</span><span class="p">[</span><span class="s1">&#39;figure.dpi&#39;</span><span class="p">]</span> <span class="o">=</span> <span class="mi">163</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">rcParams</span><span class="p">[</span><span class="s1">&#39;figure.autolayout&#39;</span><span class="p">]</span> <span class="o">=</span> <span class="bp">False</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">rcParams</span><span class="p">[</span><span class="s1">&#39;figure.figsize&#39;</span><span class="p">]</span> <span class="o">=</span> <span class="mi">20</span><span class="p">,</span> <span class="mi">12</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">rcParams</span><span class="p">[</span><span class="s1">&#39;font.size&#39;</span><span class="p">]</span> <span class="o">=</span> <span class="mi">26</span>
+
+
+<span class="n">mayer_multiple</span> <span class="o">=</span> <span class="n">btc_price_data</span><span class="o">/</span><span class="n">moving_averages</span>
+<span class="n">mayer_multiple</span><span class="p">[</span><span class="s2">&quot;High&quot;</span><span class="p">]</span><span class="o">.</span><span class="n">plot</span><span class="p">()</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">title</span><span class="p">(</span><span class="s2">&quot;Mayer Mutliple over time&quot;</span><span class="p">)</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">ylabel</span><span class="p">(</span><span class="s2">&quot;Mayer Mutliple&quot;</span><span class="p">)</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">xlabel</span><span class="p">(</span><span class="s2">&quot;Time&quot;</span><span class="p">)</span>
+<span class="k">print</span><span class="p">(</span><span class="n">f</span><span class="s2">&quot;Mayer multiple {mayer_multiple.iloc[-1][&#39;High&#39;]}&quot;</span><span class="p">)</span>
+<span class="k">print</span><span class="p">(</span><span class="n">f</span><span class="s2">&quot;Mayer multiple average {mayer_multiple.mean()[&#39;High&#39;]}&quot;</span><span class="p">)</span>
+</pre></div>
+
+
+<div class="highlight"><pre><span></span>Mayer multiple 0.5108290958630005
+Mayer multiple average 1.3789102045356179
+</pre></div>
+
+
+<p><img alt="png" src="../images/mayer_multiple/output_5_1.png"></p>
+<p>Lastly, I wanted to plot the distribution of the Mayer multiple</p>
+<div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">numpy</span> <span class="kn">as</span> <span class="nn">np</span>
+<span class="n">x</span> <span class="o">=</span> <span class="n">mayer_multiple</span><span class="p">[</span><span class="s2">&quot;High&quot;</span><span class="p">]</span><span class="o">.</span><span class="n">values</span>
+<span class="n">x</span> <span class="o">=</span> <span class="n">x</span><span class="p">[</span><span class="o">~</span><span class="n">np</span><span class="o">.</span><span class="n">isnan</span><span class="p">(</span><span class="n">x</span><span class="p">)]</span>
+<span class="n">n</span><span class="p">,</span> <span class="n">a</span><span class="p">,</span> <span class="n">patches</span> <span class="o">=</span> <span class="n">plt</span><span class="o">.</span><span class="n">hist</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="mi">100</span><span class="p">,</span> <span class="n">facecolor</span><span class="o">=</span><span class="s1">&#39;green&#39;</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">0.75</span><span class="p">,</span> <span class="n">density</span><span class="o">=</span><span class="bp">True</span><span class="p">)</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">axvline</span><span class="p">(</span><span class="n">x</span><span class="o">=</span><span class="mf">2.4</span><span class="p">,</span> <span class="n">color</span><span class="o">=</span><span class="s2">&quot;red&quot;</span><span class="p">)</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">annotate</span><span class="p">(</span><span class="s1">&#39;We are here today&#39;</span><span class="p">,</span> 
+             <span class="n">xy</span><span class="o">=</span><span class="p">(</span><span class="n">mayer_multiple</span><span class="o">.</span><span class="n">iloc</span><span class="p">[</span><span class="o">-</span><span class="mi">1</span><span class="p">][</span><span class="s2">&quot;High&quot;</span><span class="p">],</span> 
+                 <span class="n">n</span><span class="p">[(</span><span class="n">np</span><span class="o">.</span><span class="n">abs</span><span class="p">(</span><span class="n">bins</span><span class="o">-</span><span class="n">mayer_multiple</span><span class="o">.</span><span class="n">iloc</span><span class="p">[</span><span class="o">-</span><span class="mi">1</span><span class="p">][</span><span class="s2">&quot;High&quot;</span><span class="p">]))</span><span class="o">.</span><span class="n">argmin</span><span class="p">()]),</span> 
+             <span class="n">xytext</span><span class="o">=</span><span class="p">(</span><span class="n">mayer_multiple</span><span class="o">.</span><span class="n">iloc</span><span class="p">[</span><span class="o">-</span><span class="mi">1</span><span class="p">][</span><span class="s2">&quot;High&quot;</span><span class="p">]</span><span class="o">*</span><span class="mi">3</span><span class="p">,</span><span class="n">n</span><span class="o">.</span><span class="n">max</span><span class="p">()</span><span class="o">/</span><span class="mi">2</span><span class="p">),</span>
+            <span class="n">arrowprops</span><span class="o">=</span><span class="nb">dict</span><span class="p">(</span><span class="n">facecolor</span><span class="o">=</span><span class="s1">&#39;black&#39;</span><span class="p">,</span> <span class="n">shrink</span><span class="o">=</span><span class="mf">0.05</span><span class="p">),</span>
+            <span class="p">)</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">title</span><span class="p">(</span><span class="s2">&quot;Distribution of the Mayer mutliple&quot;</span><span class="p">)</span>
+<span class="n">plt</span><span class="o">.</span><span class="n">plot</span><span class="p">()</span>
+</pre></div>
+
+
+<div class="highlight"><pre><span></span>[]
+</pre></div>
+
+
+<p><img alt="png" src="../images/mayer_multiple/output_7_1.png"></p>
+</div>
+    <aside>
+    <div class="bug-reporting__panel">
+        <h3>Find an error or bug? Have a suggestion?</h3>
+        <p>Everything on this site is avaliable on GitHub. Head on over and <a href='https://github.com/redoules/redoules.github.io/issues/new'>submit an issue.</a> You can also message me directly by <a href='mailto:guillaume.redoules@gadz.org'>email</a>.</p>
+    </div>
+    </aside>
+</section>
+
+    </div>
+    <!-- start of footer section -->
+    <footer class="footer">
+        <div class="container">
+            <p class="text-muted">
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <br/>
+                Copyright &copy; Guillaume Redoulès,
+                    <time datetime="2018">2018</time>.
+                </center>
+            </p>
+        </div>
+    </footer>
+
+    <!-- This jQuery line finds any span that contains code highlighting classes and then selects the parent <pre> tag and adds a border. This is done as a workaround to visually distinguish the code inputs and outputs -->
+    <script>
+        $( ".hll, .n, .c, .err, .k, .o, .cm, .cp, .c1, .cs, .gd, .ge, .gr, .gh, .gi, .go, .gp, .gs, .gu, .gt, .kc, .kd, .kn, .kp, .kr, .kt, .m, .s, .na, .nb, .nc, .no, .nd, .ni, .ne, .nf, .nl, .nn, .nt, .nv, .ow, .w, .mf, .mh, .mi, .mo, .sb, .sc, .sd, .s2, .se, .sh, .si, .sx, .sr, .s1, .ss, .bp, .vc, .vg, .vi, .il" ).parent( "pre" ).css( "border", "1px solid #DEDEDE" );
+    </script>
+
+    
+    <!-- Load Google Analytics -->
+    <script>
+        /*
+        (function(i, s, o, g, r, a, m) {
+            i['GoogleAnalyticsObject'] = r;
+            i[r] = i[r] || function() {
+                (i[r].q = i[r].q || []).push(arguments)
+            }, i[r].l = 1 * new Date();
+            a = s.createElement(o),
+                m = s.getElementsByTagName(o)[0];
+            a.async = 1;
+            a.src = g;
+            m.parentNode.insertBefore(a, m)
+        })(window, document, 'script', '//www.google-analytics.com/analytics.js', 'ga');
+
+        ga('create', 'UA-66582-32', 'auto');
+        ga('send', 'pageview');
+        */
+    </script>
+    <!-- End of Google Analytics -->
+
+    <!-- Bootstrap core JavaScript
+      ================================================== -->
+    <!-- Placed at the end of the document so the pages load faster -->
+    <script src="../theme/js/bootstrap.min.js"></script>
+    <!-- IE10 viewport hack for Surface/desktop Windows 8 bug -->
+    <script src="../theme/js/ie10-viewport-bug-workaround.js"></script>
+
+
+</body>
+
+</html>
\ No newline at end of file
diff --git a/dl/csv_logger_keras.html b/dl/csv_logger_keras.html
index 37f061f..f3a9c1e 100644
--- a/dl/csv_logger_keras.html
+++ b/dl/csv_logger_keras.html
@@ -279,7 +279,7 @@ Epoch 5/5
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/dl/keras_info.html b/dl/keras_info.html
index 3855037..3683bdd 100644
--- a/dl/keras_info.html
+++ b/dl/keras_info.html
@@ -161,7 +161,7 @@ _________________________________________________________________
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/dl/keras_input.html b/dl/keras_input.html
index 56778e9..ab20172 100644
--- a/dl/keras_input.html
+++ b/dl/keras_input.html
@@ -133,13 +133,7 @@
 </pre></div>
 
 
-<div class="highlight"><pre><span></span>Using TensorFlow backend.
-
-
-
-
-
-(None, 28, 28)
+<div class="highlight"><pre><span></span>(None, 28, 28)
 </pre></div>
 </div>
     <aside>
@@ -155,7 +149,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/dl/keras_install.html b/dl/keras_install.html
index 76cbebb..36d65d3 100644
--- a/dl/keras_install.html
+++ b/dl/keras_install.html
@@ -152,7 +152,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/dl/keras_load.html b/dl/keras_load.html
index 5b93e98..89efe95 100644
--- a/dl/keras_load.html
+++ b/dl/keras_load.html
@@ -138,13 +138,7 @@
 </pre></div>
 
 
-<div class="highlight"><pre><span></span>Using TensorFlow backend.
-
-
-
-
-
-keras.engine.sequential.Sequential
+<div class="highlight"><pre><span></span>keras.engine.sequential.Sequential
 </pre></div>
 </div>
     <aside>
@@ -160,7 +154,7 @@ keras.engine.sequential.Sequential
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/dl/model_checkpoint_keras.html b/dl/model_checkpoint_keras.html
index 2edd7ea..b1806ec 100644
--- a/dl/model_checkpoint_keras.html
+++ b/dl/model_checkpoint_keras.html
@@ -132,10 +132,6 @@ Let's first import some libraries</p>
 </pre></div>
 
 
-<div class="highlight"><pre><span></span>Using TensorFlow backend.
-</pre></div>
-
-
 <p>In this example, we will be using the fashion MNIST dataset to do some basic computer vision, where we will train a Keras neural network to classify items of clothing.</p>
 <p>In order to import the data we will be using the built in function in Keras : </p>
 <div class="highlight"><pre><span></span><span class="n">keras</span><span class="o">.</span><span class="n">datasets</span><span class="o">.</span><span class="n">fashion_mnist</span><span class="o">.</span><span class="n">load_data</span><span class="p">()</span>
@@ -239,7 +235,7 @@ Epoch 5/5
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/dl/tensorboard_keras.html b/dl/tensorboard_keras.html
index 3bb5561..ff69ace 100644
--- a/dl/tensorboard_keras.html
+++ b/dl/tensorboard_keras.html
@@ -133,10 +133,6 @@
 </pre></div>
 
 
-<div class="highlight"><pre><span></span>Using TensorFlow backend.
-</pre></div>
-
-
 <p>In this example, we will be using the fashion MNIST dataset to do some basic computer vision, where we will train a Keras neural network to classify items of clothing.</p>
 <p>In order to import the data we will be using the built in function in Keras : </p>
 <div class="highlight"><pre><span></span><span class="n">keras</span><span class="o">.</span><span class="n">datasets</span><span class="o">.</span><span class="n">fashion_mnist</span><span class="o">.</span><span class="n">load_data</span><span class="p">()</span>
@@ -284,7 +280,7 @@ Epoch 5/5
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/images/mayer_multiple/output_5_1.png b/images/mayer_multiple/output_5_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..2be2211e5bd5fe657cf3b511eeb9e0be4329d0b6
GIT binary patch
literal 214503
zcmeFZhg(zG7dFa_qd1~6jH8G&g#d#JsE8WsC?G}zEQAhGX-X(k0)!CaSO7&qg91{5
z4TB+}8hQ~yAX1_r2~rgT5kirYgr2*TIOBZ3`wx8gdG2|hagv;~+uCct>s{~Ohg+wu
zEjDe~u|YyYViW9y*;xsR4TTaC>mICM2YxA*P+9^1`z81o%yvDvBG-Gwg73dxIpG*A
zAt5It{O_lGF<F<ukI)cvhmdmtULj!@f;=VsE`(h14G8hQd~s)}XHf9v0RNo_v=3<O
z?%jDQB;<;|j?Rw<v;%^?byWW_TPq>4Qvzmo)HXbQyboRKY451<dElK{lf+NId+lm6
z-16A`w|y55)gM>?bvyp4%;^ewYR?mA^V*-@hCZE0+V!V-ij3T~zf#u3LLUFM{g}kx
z>vwrvgLqhz#yBRtmmz0RyfW`f_U5T2Lq7T}W@3c`=M$<Xt@;KQLUTx<``@qEK0U}!
zQvZCDkT@Uwy9eml_v^goFOEO`^F_j8!%@3`zepq<ySMe<Z&$N^`@gK%RXs8_)B{P0
zFy5T`w4Xt6BDP_tk*ARp-()iBs>xNx=YtD0z}T+t-1F}!BrH{$gpMKKvcWYYR1nVF
zO&PUmaP9CROEL->-ATVrelei5r&`Mg8xeCGB1Gl7j)+WjKG@*Ghb^xdJnNv|r|iG2
z#^%j!-aAz|<0Z8sltd36(hym%t5=EPR55go9Ug-3*&DvOiyR+7p{}JDc-C^68C&%c
z%4Or+pZ2B6V3w0S^EQE^hg3ub&wo%<EZJLX?CVux6U2IYyM)2J9sf>L_1=4;LFDJh
z$5W`Lc*6@a)W10y`LDYZZvCH6TnWc#x}2VHPaLm#-2E3@lR~||%@)&qZR}@ZJ%9Z!
z?xC;t7_>S&KHgc~MRH$vK8F1oVng7Ysp{{pz%k26-Dlx`O^x@P-Q-_;or8Px$7vLj
zwJUEC)No#JeIcmo^~g*c#C{`m^#P;%@7eThYG?^WddGpe1qdRQ?UH}1vVc9Gw6>*_
zd1hggREo&^u3DWGxznNK=EO?Brs?Yk1fT;E@c|jLhWU>oA+#E4=xQQ`+CxX@Y8f?_
zw4uLSb-%dJk;Oa(lM8Ob88{i#d+&)K%*`YZeSQlGqOl3uk5@3sgB?fN=91`bNO3`t
zzg}%6<~<ua6FmNPJ_S-D7qnajN7bf}Tx}iGGiqdSJ8pwJ%&0CCt$_2puaUtB^J%tl
zHT1;2CrPYF!Kq%B^l3^kNjl9Ajy0m0Y?3Q?M@~E`Vj!M8o8fp-q=i<MsOa_<udt+x
zNc;e3<OT$H#PSQHdl!tb&*Liwwi=k!eU9uYfS3?i)v!qd%YRh=ZU3cj1p)XRybGR{
z4nbJW<iU(>;RvTzYOXznYM=q#O}n1ht48}WYFu*oRFEIO8K0iPB8Plw=G%2bn;kg0
zPDl*VdDb1@t3z{5R7hn#h8W;oQdoOn1XuWg6GPvkE7il{63Q|<#zg~k*WU?g0}JYL
z-P{*aKWcNC(iiYXvmXKXwyrMZUcMqQI!GIu6-?EnL7+Qn4`JFTW`-fkHbFt~4C?c&
zw$6w?9rjCV=l~?)1ixybZFDl{I(|HP5{bKv4|uV(C^%P+8Co7Dhhi^Ls1TDR))&~S
zDc&G!sZuG0wQe-3t?Wo2n>!0LPGUVEmHTQ+^_wFnUPBTTB71fFx}{+yyc~RbG7FJ2
zHY(-T2*%{z(xI$7?-=$w5#D)ui(3)N2^pgXRqs<m<v~pFtj~~#lib<65TY8|8NM?6
zAXp)NQXms`fF?8QMm%g?7*?M7w&p%-4x*v4@a=PJSdN#`fF+nJ`J>Lm0_k7Cr)>Z|
z!pjl)n`hfeiw|qjf(I;(==9n5q{aG;vw~yz<V|vm98$#U=k#+OE5Bo;;hirsFsGON
zVFbMPAJd;hYQhN(OnjvZZ4YT|MvgIyhtzz~jT|5?F2ADMr|`FpzA3BgZ(3d??O+53
z_jWI7pzZMN=Ta)CE?VFdAZQpId%jz5yqKT`b2IYw<ulJAC!%45OqK~z>N1%XOIqwD
zBh%}%Yi#fucAW<6F5H`KUn>aZUWb&BNjY9A0xKBdkjGaVoLgKA#R{_LKali;-YgV*
zwB!x!^?Ky}k)WqRCy^+sBPIv>u(~aZq_J_BI+EgN#c#OTXEn=}$M{jGSp4!aJi;<)
z>DLzvm+-Q-osAlf&Y4@QZIs|}Vw*dm)&=QP;}Tt4y)+Ap((IEzsa{}2lf*AO!m*d4
z^&<DXo(>8V5GGOO_Ye<WG2)xQY0tw7vq)~brW4N3m!fzrclay5_V#KEOAwr4-s5@5
z(n;h*F)S+h%4+SN2i31g>YL)NJ?gvJ!{%`A8<|;U0(26lTJ2r+f>b;nI+>u3K8nw@
z<gd#MOux^vZ*a9oPB5-Riy@zIEzoK_i<vX#>P|4BQbTpX685#jXkmm)`E|SR1k{+L
zGgC6Bkf~2Dz%#ntDAa*V6zm~a*%FfzK^`xp3)EeSr6!g$R7SV1SN81fkUnxWx@{xM
zeug|}a+h6yIj)!yrR~~CuA#A>ke@H{vn|xoION2Gqv7uGR}7{5tQH6kPrJSW`R?%&
zvnOU_-=DNCL`p@H{_diJSB%L=tPa?!e&S7sMD2Ur0!DX^q@n`zCbtsNW{I4j!B(|s
zf;QGJ#&gcc`BI3oF9i!Ql&F~;be2BdLtH{%I1@hwjWWmMQeK=#`Ux1_%Fj9dOU{ob
zmp*NhL@u30PUP9mR{Ky^m?0)PE6HSqi=KZwA?Y?VH$zM+667yhm1`(QKBw}$uWP?}
zHWD6MemG|3Sc7QPlZnnq9r=)HLIpQUWVVB1+(TEXxynB2TN!_h>8^J|hFQ%#-erm)
zlW-X?MpAf|Gx(5%Y(j~%IvR_dsJ|R_-wi%+VrG@_W$49<OTvNmTMa9`=X2oP51Kwo
z?M@hbJX5W#b1$dot$SkR3o7p`w3E~6gmJ_3voH_%#|x++K*;$hrF}V?c-XF!s{7G%
z3zD+pfw%GC8AVv)JJ$8BFv#3$M4KtnuZGcm3)Y|N;fBbwfIy#7c^@@%v99Hlxpi|l
zt-RZvDy+(P&g>KHvnt^;x{dIW+O#Q;!qew7tMJP~<x69mgR_}A)M6XtMCa(SR!OZO
z?c+$wgi)ypBGt&C!e+2W13i(ykMr5eEAP|{Nsc{#6O!kwxzFGwH&SFDTd{p$@J{Ii
zmCuImCN8%;@XYKCj^sSxGVqx;Gtrv3`>QjAp%@7I-JInZ$`)Upi`o+A?nmj<rV;T9
zy3J1`dLcB76SAXv#O|XL#;ZK<C?cktJ5~p)He$`IN+$Rb8j=V@&O~yfI@75lpxnRu
zFyR~xqqZ>r`OJG{Tf~44`w6uehYuqzRYL<LA?W7EiL5TtST96*TrTP)(l6<;Yvxe9
zz#!#T#YgF5Ekq*678<}mIdhBAogsS~LrnH8KJ2v51edYUmuZr$y)mMymi{iUZJA(r
z#TP%2a*lHMifFJW<R_G;gsh3?Yc+FcSEJ2|nFiPfDhRr=-Ccb=NM2^R=PU|_rxP)i
z(^=O|5RWGB7az9f$Bj!upuRWtP0}Vm%;}_FSq)QLu*Uanl*3r#nbr;gYS`Mpk-8Vc
zfh(xOj8h*PJeoYg5AVjog3jQlG@$!vTJ2bW<$KR{eyOy<A1+}ioq=bjcwqN{IHf}y
z;B7?QpPUK}#m{!2{xpX4alSM`4yUo5Mku7vR}8nb$-vTzZnVxS4Z=MhV!PUY&0u#?
zAXt$!xqfV*GyYu1MDqZQzkZ=dy$pkod5)_&fmgPg$z^ock;YQd+Tq(c*qZd1qgN+d
zVM=-pqX?Mn;)*qXDhpH93#Hk34O<uM_)%6KlZvhLE1AzZz7-YM@rIhNOa?l_i#{Dd
z#7rbmsJ!eDe*4J`)jaDUCwx&m)>_9Zs3Rnk9+OoOw%EIyCh3fn&L+jFI}V(5LJFWO
zwY1rezQgRzs}*X{gS4s+$|ERP&}Z=og#)itT}x+YN2@p6tCyk6&yU<6B}BbvhdCny
z@<=+HSC>r=RCY|qLo%&{8qUY>Sr^=^>e@K`a3uU3d}S#%*zQ&Tlp2`AyQjpAt3*0W
z%67TQzfrjltQ9!c`<hb%?c3uKO*pFml~j8BLs)-i?$CH62oZtWJ`|&UuIocA7jchP
z4Jgzw+fKP5Z2-q547Z0Ye)L8Y5I0Lm`pDjhXRjFkNvvB&byL|SDUEDY9b~HM>uJU(
z*s6WG1s!4dl%a(;G@<3GmwmZ+j%UgAoHB-Us{)xvqM49SVI}T}6EoK!Q#qs@45K2f
zoIb>7mnNBar5ar)!(@+V{@Ijzx9<#E9gbq<;`}L!1Xh|6Q5l3K?ND=0gC)M#zr;OR
z4SfNhxcIa2^}oIbCoX&ODsf?p(xYxeHTua9+b&%nPoDjD4I&7<8SIwH(jkrM)~%Dg
z(AB8Ea2+D8aE9$kq0;%Ca)XMaZ!Xo9uTjo~u5$!)bRvjO_>xHSY5avcnaZNbDON7A
zL-;*Q$8T0RBl)LhDn|QLbXz9cvDoOfQAAWvZFgu`5MEYe!4}?`FXfDs%rg1S@GzXJ
zR)LE1qik|LhxfMH{@BfOGTo?#XAfBzf`Ev=whmZY5$`{+sx#Ujl3xKLu{pjG1ePS0
zbeMdNVPBMyPrN)4^UPsDK@zes+wu)^eUo^JUcx3&g<<Z{-$(CVbU6IJdXn_acNAKU
zdZS-7f3X~;-~_iJlM)*?6{y4DE0%C09okAe>Xy7$<|(D$PT{w7poofIZV7Hy<@?$;
zMKriUFHmYu1qlcRilaYEhUJoK<hH&$i6`mTkfID+nG_vvdMGDk*G+UeJ-7J;Q*At!
z6yc4C)f;`&2<I|Li;Q%pu05Pf4S5pR*GHiqB3F!^-V<U=2unV8D4}tJ$HrL)Ws(--
z&bW?|#(Ln~Yk(f+BE2&a1Qx3Ly!?flbQp4?di0trwmOsbD;(9E{p~W!8phnU$hY{|
zV?tvvx*awdkj6+5+<@+K6Qeu-@X#sF4H<WhyUx$4#R27d(ozHJuKW0a3vGd;W=<;J
z&Ka|Wb6-Iclw1s2+?|ll4X#7mCsT5VzF8-$L(55o({LgmMCRut?L1O)YoJzIpXp59
zm-&vw-Yn8;w$Axf2pk1Wcd|Tl%i^)F1mJ!5S;Gro28^PDTA<pBErF5cnapVPMZiX|
zddENQ<o^yxN3^;qwVX5ZiLzZk6thuZ&*(za`?s)uz1FAdue4}G9c!Pv>8{T?MIlwE
zv$7$|Q_$f+3Uz7j4!3jJ8jl;7Gp?o{8Ve*Pe#L}LF%u)hdsQgZ7$PV3NQ)jU{v}D)
zuCo}n?eO|-CI|bTdB#95{B|Ry<m5~*DFO!PuD#97sm6pi|Dw03zq=;jSH*{)QjF7B
zfm^D}J_nB~oxxetHMLx?hP*uVISNXtY5Qh<8^+{Le<kUodoRMN&qz@^G@N{3cU9tY
z`@H3NpAHedhJO0`>)ryRx=b=D&-(Nv?)j3z=HO8Ll6_~7dUi78B}w`P#xMfGoHp^N
zQ0BgJFT13(&cM0-Yd?A&Aih{~N;jBX3+KKFaRq6BfvW_LD6NJLfn&q2LHqBq=n>H%
zJlzRIPIx`*6H)VcQI6{ybp5Zl8h}q#QF-p}Nzg-e${h<05A}Dnn!^2ZNc=lvV|wPT
zdpd3gFDdm@dfqP70YvFid1DtH(48#{UnFXC-Sob&*)|<F>!Z=Uy}LlZ0im5sbZHNM
z7hvU#xluzV*=s$Xbow>|se997gQRd#?7lux^mY7geB=$p1P)KOinsvJNMsT5E<M5T
zt~(<w+VYFhU6+=I>(?Q!$J{+K+(pvIrgp=*$#PY)D<h>CSNO`g^24DUgO_hua%RuL
zHgn`SQtgvZsG$z{fv<AaAiK&Lli!j%5dr7kgwQADs)u6>9Mez31%+43PmL@?Y$&8=
zVkz(KsNmDQUBxm#EYXlc%1yZ#Z*mg1Z|-v9ww}jgH&Ye48D!<B`+*gO{*Uq}%lCyZ
z&q`sU1w((R4I~4Ybf-Uh+Fz<y?0GE@WLX8HN~4A?1Mp6ch6_`PG&azoynFr;qr25R
z!(G<fdu}L~NhOsZq!~nS?1Q*hT00{TTjMN-3o10;tgmr}vu!YeaO{4XrA7M8@?|95
zXU$}C#}2MuI1<i9L6n(!Sv`r-WhgxH)M?Ih!Vv!w#pudY_$DU5s6%hGKYU-VAc^&u
zP#|yundU3fF4CAHtpB}kFVi#J35S5^e8|dtxN;Wmda(5-93(eGA=)c%Nn=u#xd$7w
zR7Xmu{;=8KbxAyMh<M<x`GaG-8Z)cUCKoZfU3CrXgT3Hfqs?d|^f|nuY5YI|HM9b@
zVUzyT>O${%8IVd_mzQR&yHHQ$ZOk<MGV~>r8NB=sqLXsz!n2GA9$n=;m|(zj2TtG9
zBA@#_uX!M8e=as6I?Ohz!zaYnt8VEmz9z#o!{iiB_VqJK%;y8XuU3LCQmAR_=t%lE
zDr|L$9K7-8Kplm^cclT9uyVR5B4*}Md0{cbAHPa}Yapo@+W4|405ABcyZi}~mv#}2
zU46x#$2uYT=|hG(>F!!D_zxignFS5`hw0;EWKsnC+DO;o6&vNQx^-2_LY#+)<)N(j
zc$2C(!$31hSefso`o1q1n>DJlICW-b>cFzhCoLM?=1c%F(=zBUde^O?ZE`ni<45J9
z{NNb|qi+^B9>ZfZrj>MwmvcdM?f+_gk7ZyQZ&FMRZQhzYA$8R+<Ffhtg&xulp4Zh|
z8_Oklea(kY1kI6yPl+ac@g)SS`1=457E(J50l|UOyUqY@sZsMR+*@P8f&Q&SvbN;1
zV%jA4NDIV7)<_?QAoiIkR_^!GCXL;JWRA%Zm`KjK*{=~am&(Y^d71RweXE6XoV<gL
zSdHwW50gT^vdL1!SGs_gO>1Fgg-Xv%%P!NN#L7^f8bz+|=WDcI6rY5WNK)+szalM>
zewi@jW<w%-9x!;pFDRvq+@8*T0U4LfheoBg82MUr2M};<!n6K;>yFl5;P>{!;yD%m
z@6RkW#%#ovkrv-<-<PZSX#C9Bo>7c%FkV5g5#2!jz-zv26ja~UTdru==?PmbH=Iml
zecAGWBlS=z_0zEr*no{kbZATJ*&QX(If=_dW_xzfpCX)RdnHp^A$fIANMjW+ZN<QM
z5R))oYv5!o)PMicmv_base>qycy0YmU!S+~X8r@rUzat%Bw54i|8r5v+h}~sI6j%L
zP~&Mgd9@;j(Y+?IH_tn~-9I+rt1uNjr{Ty<vF1Rd5*ghIEl*a7?9=X5U!}b6MzDB7
z^zwH4HN;q#o%>vnzPb|buoLs)P&a^jWl+Ku&u>BJmRJ?UUsyG!Jv4G}v^mX=pinCy
z=+pFX9Fot{FT;y&&dLCSG@L&<fRBkG)e4@vSRJ=!ZCDT^-6vxpczF<laQvolIQ4PE
z@d}Y&?iKmvP8wHHF$QJbqtWNiiTTtX9EopaYuh3xJ~O)AL8^I)QdLb2bwW<aQ>d2C
zNaibsPgcTQjFbB4dQ=rjUvK5;wdyv~vpLryBiewK183E!5eOH4TNqAf;8-0PqR)VS
z;WoYHNkjQ*xVIV<+w{H)C|{AT_FuCNOSEtM`eZu))v6m2KY)skXLNK6G1>{aq*i80
z|Mm!u;-(W7sKf#9UX|+-&WJyGDU5C_4S+?q61?YuajzJgPXsXuUv#WrJ)P#RT{pEa
zW{FgoQgxLE9K^?(x_IodH|33j`&!4tpR0gsl>g*cq5KIHzwY43=1EB>fR!`4rfpS_
zm4sE-trTjNGt$O~Qe{B93A2XHEWrr(bDg^t+gE;%tdS$?Ux`K{J_Iip0F_h7T|&>}
ziTsT<L{c>Blj1M|2$G$($doC*LT2&MMDD8JW<5qC1xTSAp$s6Tu9+QwN_)YO)ps>(
zUBQz>?c47bBdmu)VY~@BXS48Wd05nY-O$Egk=&|LuNp|B_1SDH1Fyi;U48_7{g>fM
z!s<ZZ-Z5$?bF>ts=69&{JW{h4xi-?pu+a&-tD`ez=<)u%f$v}|CNnMpW)_W@5=MO2
zL{|ZfryMkyw(b4je!di{J3dmAW&(lg(>858(zln2NMw<AHG+tqsh%C|i@%Nmypnbe
zh<49S6ut!yh2&+7sM1<7BbA>ZoT<Q31s}3Tz{vyStC$F+A3y(75@ClkyKLtnYF=JY
zD_rVVzxOLaJDwS{5gB}s>(&W~w2NAGkw<7yBll<JIh%t$@ICxHR^3@qwmX;V*H-1W
zDePyH^C+!6Y`9j1M?7`0JU6#NVQAGq_&c{q73=2})vw<sZC$Nw=Sp=t^$@AN8OYR0
zfx8*oZ|brRnG}~;ndLoiLS-~iR?-PWYzTVlMza$#<Ej6Vm>t-Rr9;n!T!LmAz1b?L
zrOkvq9RJSu&$v>-@lL~#pJutAVwZ*w9$#CujVj=+Rn^#V3*H$I$(xXiI*s(hXR0k;
z%VVmchajAH4H!Q^(o`L?7bWizAL#cA^D9u+O}B_PSVH*?5GGL8?_?+{N`8q~Rt{>j
zCyl)_*Y;02`gPV1MkwhHrhgqaYQ#ZkLw4n*)X*u2^qAag1*zNkOGE?KmgDm&fblaN
zHEg)8unh=dQE#*beZYTOo5!<TqKmqZGCDW@>Q}lFbQ0J%Zo@uaClCT*23ZLav#k!U
zAOQmELJ<8MhBWr(SLUk^Zq8}bOZ$*@-`D&rY*H0i021f@UXm=GkW;5-iYhiKrn0_<
z^zA5_bE!rJ?vWqGsV$7|ryJ??4aL)8K}XB~YIz7NRvVRyGFGMGQUdQ{vPm6IND!%N
zcGb~gGm~3S&Y|4<?y~}j4(-o>dTfqtc6+VQU4;y^5;B_N0sTI-9_}|<7RvdV`F5lz
zR-$LCYJ;6F#G__>+~A_`N2brVJyq+XtK_`1NsB;WoZE;y*955bz)MnfW7^c*bBHAH
zmX54`q|`{AXc@<gb}=OCD%COT91}f&B<BhvKs|gvsz%5Hg<=R6<yo)!kW~fif50+@
zqbC0d?$wc6U$_~_z0J`IxJcxL9h}=o(3{9)l@d}x4jW)qkfeCdrEbSLc#ktO&vJ%S
zv1uYLw`fMw(G*be_nRI5YF%S-$F<_>!eW<>!80XQ<1wUO9J*$mTjH0aj2<=U0h)zf
zX?0#OhhI1&e>Qr$``Usi-6N}tqsG3e=5O)w2{tox!hMWZ$(%8RIJd6fbWY-8o&{*2
zec&0!Lke*lh9EQ!e)%$Qe5BY0@2!Roq@#f9pGYX!LUd_{t2XrGS&8Qu8A+#y=Zd}P
zJCXSGd#ofQ=BW<oZ^z3%>}FO*07ouBgdWE#ZJXvQ!%Ij$<NoA|@N@W@mT!}HH9XAj
z0l75|k^p2UcevFSvLY9X4a7?kv!QV~=i!C(ZoNV>&xsW;xR|Z_`m0uSPiNWk>|IXC
z63dydgfFxBWe0f1;h2|;KR?sWAvFV8<HStmt~!$er==l6;|>!dk%b0%({NDr;fv$V
zaIXD1hUY_g=OYT$#+z>a#sX3{bk8!W{S5ly-H2cG;|fGRt@a7n)mS=iLt%d@dp?^w
zM<T^Z_Kjp7@R+Ajs0b2i4@TQWo91TZ-cb!b|8Zbw4IDN00Ck}#P$+znRk4ECLTG<4
z`+EDfQ)Pe;Z!1kQwjkEXb8wH&#p=QQ^@;jeH%9RCW|$InsTOe0M8isUs=a+@u_iOn
zA3tH_ZldpLmE?`b%mbfkQ<2qulm9pr|D4JQ!`Zlh;P1zjckE^^PyRf@ajzLqH@G<7
zHcRl$*O;qXcbwv>hFuyy3L_jGEcXT*T}7l-&v(>FOnsMEgkLTBzc$@x3&(2GN_X8U
zvF*%m4AXvhJt7|Q0J;|Gj9h}0aSVO9FG;%i^nP7K^6VI$|L`QxG#Le2w8NO_sY8CQ
zNh5b*4cHbQV{NXxT5`g6gG=$Vd&xxyswRtW>*ck6ec$f#Nhgm_fiSsu6br5{Ox|0K
zCnGYSOlrckzseb0=mH}L%%q5&dby%Non~)w$L|GmTg$uuuB&Ih8MEkF2J_*BD!h}a
z8E_WJw&$&>$I)o|v<8T|Jqv)|M2-I130XJ~$GSJ5=58h{=X3NOmPlhuVB#BInw*&7
zt<~|)T3AvrY(y{q?7g>bWyI0<{oUDWqdLT+C-U4^hgPB-;Y_uKlk{&!AP%R@KHw|w
z=Pg{0+w5kLHzgNkhMYKL8m~MMyb4syv<q4Tuy*E|RpJYZRN7q;;9_M&fO{P6UalSX
zERsT6MeA+Tt#6Hpr_WBeTeKB0{tU1Vn!YzwA!(0OhavtNtA~<@hKz(!ZXWv48+Es1
z>$g_RvFyr@EY*hSn~my+*8gSfp+sGGeC;rwFl5N-TX9B~5ZtTGe+!lT=PUqF9|uhQ
z;%DzPM}1nbrB7pZ1^HcDhz^aVi@fG4yg~`*Z$@`DEPeti01i5N2RVxI>&u~r&flTR
z(c=k2ym{_hk*%%87*^FV{`E-nP_Q=@L7(2kJ`?#{6lT)UhvM-Y#PFr}Jk&@~l_Ql{
z>Au+#imrDU=q0-npV!ELvTm-7NnCA2Ys_K}u7LRrv8Lib{3MWX3nXy;^B^{9?p5A*
zwl6g)u;r3h3TJC&+#Bz1PdkB}cnq-t>8v)b(d#AHJOpyQcQ}7C$AT!ZQk(5MS5=&p
z7m!)fn5QFR^eFKW0tXsH#@(P@x!Pq@oqme?$`?FLF@t3tOwtn(UI4#eK%NFcd(ffI
zk_q7i7|6X>_q{iDLZ-cJ{#;)2W~eZcH9#8Ml0zjZ2?(pXWa-ong&ISmzQM}*`Yx*x
zLuLejoYCDP*bEfTGmEALOm*}W$X(t=p{w}nWiPzP0q^r;AEf^Z+dM9}bQ<nz90OfE
zmMzLwlTD_SeEUY*Ayesf`qtL%eEz~PNRQe3B%g67X8+gTz@ts<;GBi73Q~1d{jh6S
z@N#X@HV~sySb1I)-CHF6Jv0kUbY>cho%o6{y@g%+K%jeZ9Ji@_u)x~3Gw*I`%v9w%
zhCR%<M|b&ot!gC^oE8YiXJ%wIOs-RWpu>@?9x`NiOH4lUpGxhVf|NPZyFe_)LtFOW
zwW>jEHs%YubR(}raECYwqAc`7?70!{i&Un@Cx&k5vNjU>f}LKnkr(sGwo|%Y5T?H`
zcs{-G=`1F`XNV~y%A^-}i85)jHv+JO1hAaF$SZ8eK7X<S7v+f!!h7@YWH`YQX+t~u
zlUPzl?qOhy10CXU!!?q`o^Qn~82EiSx}LWQ=Cz@YH-DTy{pk9f{Cj1f2JfGEOQB*s
zL+twmMt&2}d_<_5{>Aaj6l!CugMi+l>|v}yGih-!ft`iRLZE<Zjb1kz?t(0F(>s_S
zP&;&BK`45ct~0XcP-9tFqEXb4I0CLo`=b0FK=akRoa?F$Zu)*2jbKAe9o-*dnfj>*
z@3N0IzY1Htc0Mni{WHl2gGYe;%SHp;_ZT_hLS-m@a>4{4{ie`NFDp!QY(X*rphvI3
zu39Dpm&W>gPliDjeNMro5G=84BQAOV1eQ@ph=!tJ(yGy@^k#w`T=|h_U=bJ#X>15)
z4XSp2fxQkS%dkXWxxjasww=_DkCPu{kAn^L4{{-1ih~?D_ZC{~DXIG5hdvcVyVKIt
zY-#`z03L;^MWI$7X83NyVwPQ<kkXEw5BF^mPsz^jQ$kcPEA$@Mz*iPiiv^``0kDB|
z;s7mxe!=gY+>Jmtqgu5AOWCLlzP|ATmLUGf@F>i<&uOW^k?lk2^Y|9GN`Ee;VWJI7
zqiS<3;44VHZbWndU^2pFodUL&9Px>XhQXVgIUTYDaBu*S6~4pgUe1S=#B{Kcwo(eR
z?#VQNoB8{MbGvD{w7PDui#3(~IIBX6_xR@!gdlTAN!QL_c!Lm~9f93)qf;KnlrXij
zQtxq~S7m$q68ap%7Oz7xC**qe&@!ov{{8DCfiM>e@Lq6grZvPQrWeUb<@9D>UFO_$
zWGadVzoUd+C=2FSVjqQ6UP<NQ5|<<Bnl_d}^je)npgHJ`KR^0);w8*2#i)Ydmd-*X
z8v1C*Dbe@fmyb6{X=K>KnR_oO)+DiFA=nFdD>R@$I=ml}@No4C+_m`Zhoi)n3Zj&}
zi6-sL^pq_GmpDl^X-Uvn=zZ+oO@mY2yI^1)3Ag73k5~;An`(4u?(A2L=Ta4}4_VQs
z4U@?%O9~Zd#q69ifN9$lRuLIXIb$u*TtCVYHP`n+#LN>jHyPbh9TTpyn<_5-28nPh
z$L8m4LE@Kz?9EDN=;wfSYu70&iRdvIhYr6F$w1qjoRNj=9jr9|`eL;df_uQ~SA9{r
zoA&GV+|uB4aF8Mc-YFZiaiC_Z5R#X{)?bG#D+b%1kjzs-4)n`5CxSwd6L&}TU*Slk
zT<4d($Clr)c*6sINKWhmIihafth{RJkZHqa8{zg}a$@9e+NV+8eDzNLWjLz)XGT+s
zh)*B~QkNxlgzM_p<;!@s7yW&=HUFgSktKiR#1O1HiPa8~R4enuSE!>e;j1ll1sCxJ
z22DVQ#THR{t9N9}4Fdn>l$KL@9i#dK@f)@%>dklcI~l0b*ol$2Z-IVzQ?Nzik7xPA
zz;5`uIH$uLMoznt@p7%vkVz_Q0;UZ%<ag3a+Sj%Qq)&d}T0qc$MN7W6IW<Fn<bIDz
z(!NZo+U+X)&RBYTB$?$1J9P0L_i$pbDs3VZ8hQn9Vl$&bp`J@zb}rBF>L2~toqs~M
z+eNcU@F?UcQ?C{PzpMuOJU&w{ux%O^KcMzYxPSRnbIFOI9&m&usA%iRO9-7KRo7OF
zJYd+=&3*<)@l>{*g(CteM#ePBc0o~FwOpiIqRS^I11%bJR)BEENaq?sTp!Q(!zTZ*
znc%?T22M!aegB%VV18}mNw)`P_fRykVbjFqHy7S%PN8}~m&%D>RGXpTm&b1m<MK%T
zhO}q**k6Yop&@)5ICt1w8(_tN6DGX8@s)}1%?`dblaia~t~j^jMl>Nq`*Eot_FMpg
z#a90ldlD}&{GvCsSOpR6r%ix;UDf%c<(tF6*_GFQ4LqxvueFWO+pL35!Cj3y#ztDA
z9@)aFjCd<PG`wX4F+u+4UiVZ8Nc+pe{;G<)M(S`zB4B*4<iJX=Ge|iyy8~=DWcASf
ztKN)CK9^c!a`8U08?vfT>*%;y^hrWIwB)e$LYB;NgI8(Lbcj+VEzMFnV|=zlz}cyj
zL2Vn@d&%;gToA895-e8|A*RZsn`if}_cAeBN$@6oq`?S=ZO{HDo`K5}r7o`asEh^t
z0_T=N8lOv1+v*;{l<eRG8Rx3X3K3x6&#F9cL8aQl^%2WEXH2m7F>kVA6VdHHixTH@
z0&m+XD?IpAA!{U?QOl2S)L57+WWHH06$(=fb+KS-y8VVzJdB?`?Ffz*fjyF^pX~Lj
zNZkZhfJJ%`3}1uD8h%d7%^?XM<?i|PQYv-iX-FSk=`?(H!-2v!+fIXaRFEMq`T!H0
zSeV_CGxn}*FWCZ))pFfAVm|6tH~xZTZ8h^Al9^fXk%KqPCfRLR(0RBrvq-NEPBhe{
zl?Cr!dII6UB;AZTF6ovw{iZ1Z<T|odDPZFSm@EZiW5q9>HJ|sv4_F6{UmMHY14kXj
z$3PYyLI&Jsf>OkRp<68X`)<G9Mh1J2$)n}PR)g0<j=rjZclLzIS7_7pOpS9hF4yjn
z@6;Sa!BL&p*aKjR7L(#n&OC-_sFj5@;1RGHT!#^1g#qU-lNMX<=sB`UW2&&p`>ebJ
zpQ<WVUA^nUrjAsO4_yDEUFTMC6elD>Gb#jca)P;u7YEZm6@)r;Ijk*Xo#HoeoXcJp
zy?xvr7EdC1=h~gWqBNdN`@A%&m<aZI9nU@JB23DBP9Tvi;GIu@_Qq`g<%Nl1!^@#i
z|BSSwDg0b;e2Rq~jk12k;DN(i3w03bTvCNyXB;!)e54+{vzQcd8DE#UY=)PGEW|*(
zpJUd1gnwdTmLrDUfZsV7ICUCkmI5*@k$(sfj#O3w42<(C1e^^!NHZQ)TC-dgl7VkK
zG4nS>8S!zbk=W_9^bAtdsT-Mjbf51_APm0Xyqf<E0V$*@zDFJUCvA1~O?C1n<It!!
z`blu)Epw(BYuD~%aD=u6Z>U1UCiWWB?&4pC<mCi+tUV46wq?M%8ERUnb71%O5lhvm
zE-af=6ys84gOAaG>boB9seqKh@N3%FChMFiv4~XZW6rgd=|uiIo(w2M6in%nb}xn<
zf)iX%98c5U7!Jc1C8j@`<ls)iTiY|x=PptDj`YzDq)!G-g)Ffsc!d;?nh-t^qNDER
z_Hf&nosVGFc_c?uM*c35t0!5AB=hr|d?*tcEay>u!^z1x(&9AS?f2J8FT`S{#JEqM
zrx52Qu0Dxx`}P|WUm2t)D_q6EihqjK1qoz_4o^XF6n{8x<e2D&#5Qwr&37+TRxaTi
zc^hl0Nx2TKT=Ct#w?z>5x_j27T`+Y>$GP?9_ZxWejU9kZAn{uwqD@@jh#$rNx%<JX
z=P>OSjK_fvtL2m`h{|5QDOSK7)(Fq~p^^9)Zu$(36d~BH4fHW^o&Umf<ca!yC=Ss1
zL_^PHQ;ZrELIc6?5N%aeTfkkbEj&;Q9{ql<*^ZupUmpI0x#9}PUhIz(Zkq!yDy|ZX
zSQ5t<o$v@(o{H#aiEU5dorO#Z%MVT!`m}ynb;&pt+{nK?p3G9;l3t@q<0tweT7o3J
z4*4J|{^W^F+I7i4$pZtQFv+`ijqK}0p>lZL$UN+3gYkS%9dOS2s?{L)-uhM+VaO<3
znySHVHlF`HPED2^lRz@Fu@mo2rCa+kQ&sQwhz16Z4NB>>#48#U!#k^INt#XXUUWJc
zOl>PEeXG@GFje_aogdHHw4Sk0q&7~k*?eT@;i^%qV1C7bku>(JY4GxNBl}yOG-74#
z&!@5j5SI9WsM$2$0pS?C&ikXAokx~u=_WNKe`sE+u+L%-5Sd_KwfQfOMroTxUxfes
z-*@a=+_nG9&!+$XofKOD{QLi@cMAS@-W=d;mq^dh@$5%FF4#Wa@-(I<on13sw9s%f
znSR+5JWekDE{KEA=V#|BCAeJyM^>E6hAv2s^wDt+yn(Q(mt)Iy-fgr!#*23$w6)2u
z8HbJdwX-hjph%9H#rHQZ&cj{M6UCKZ<^t4U)lpxzqDNz@51}XnQA@IUf~k(`>6J%J
z@CM84AhGNtEIXLW30;1lIe=;?JICw}$W&g1vx_GTBHF-C)p;#3n1bi*-ttJwi?dK!
zgdU1w>R&n~y|S=m?_z+`0yn=05b()it3jw#VcVd(<T&3HSN&}UIbtUd5TH}cK)@5^
z+lmG}!_Y6D9cxbqM81{|^H}L0))#JwkuClKk=LXy<L9az`?d;Sl2c6Yx0vf>*->F`
zo3H&n8i9^qY>Wq|yu35`*I)Wfl+dLGq;tYxa~}Ket0|z*yN)t+aAD=~<O*i_*BK;{
zv_?4k|5So$%~RzR_gyC^iyKh-aqnsLFOD^#tHYm)v5ocA(3+7q&1;44xiC~k%n*#_
zJbYtwj#~xRq8j~S>lS8rl`4m?T@%{v37*O~`_E{q-RbdUum_4xZgt_PD2i%P`%%ky
zHjwX?fpgSTi~R;Ds+_G%FAJN~+koMy;_YmaX|alqOYsSXeqi7}E<v5Fza7BON&k7E
z2cus&+3$){=!w38&?#R|y~P1f<h%UnNj708$5zoX<_1DM6y+bC@EzE{JfA;F+ViQ{
zoqkc)HEN<BC2Usv$0Ktnw?1RYgL3?ZFU&aYj{8D$SBV8s@TA?__#H|f86;JX(CpuB
zCI6q&VICQVP5R?Igg;&7ib;pCg{`yWO%Z)OMT|lPm?Mqc|CytwI!*d{@#Fz<dwu_D
z@18mdhWG2ykS^TJ86>Ih0du?fA4jWh%jpT5gSc399fEAdeClLcxd764b@)FHBJ8mu
z!xg1gms{r5^Z&N!ilQT1)`LRG-wQg5Occ=)@F{=yNauYjDWWc~T;#brKP>!Js?fUo
zKFanRT<nei=`N3OXY{$Eqs3Z89`LEBPN1T_$v>(Lo%`i<$y5qCZVZLJ)dgX?5Je-3
z5Wj5%9wy(O&@*f*_p68(zGIU1@c$?Zsy!D|=5O|m>m&BBtw-%X3kt&cM+?Wpumd$U
zqo0VulSXGDbHBJmYVEn5{ouyg^BbHE-$kFiRv-m^`S^UAMbBsAX>9{dS!+#eXu-Mp
zp9Uc+zsbJoIu`ne(z~l>8=McU|5ZWe-W${0^~a7$f87AH=Z~fr8FZ{PD_0nF^mP<T
zdxu4o)3H?~>%Jo0qAIK<cJe`6C0iHt7F<=?j9+qQEu+cXg#JINeC#`DJ(lJC{IOZ5
zFc;x!^h&tO_wD|1YdF;n<>snwY=pTF7u{3;aj#QPtFVo}(XrW5Q3}lK#P6bQbzw_?
z28UgZuuJJ)ndj!@gn5e^Y%mk4W8n7OQ>QJ3ZV~sXqSmg|{Pu4-VGoOR-+a{e)+$wM
z(-QuE=az`Vf?{9I%U<l)%4vQ}^fp?ZY*7)qTgrOjSip7F?Lip^#wFkE6qnumzh$ZF
z4q7^Lg#_+=U75%;PTdgBBDnt2VQ3c;3i7)>?GCwj_KnVX(kqH)W;BFA&G=EjE%F#6
zN4wBC|FG+Fzo_xj|K7aTxu9msa75N}N8AeLSHATAzxs!4INLlHw1+fd&Pc0hHfL1m
z#tD1iuJX@2c@bxZIgTC~C2yFLyrf+@=E7wfDD$szF6q1^ge~dM7B0-o-@VuQzxBH?
zRgeae*z<V~%d1z^g>`P<vJ2Qqc)gE9?c#XKI(ox(?Toq{wF~<*l9NsLgFD4u5v!(M
zHy*SfR53PCQPN*v7pV+-DF^~)%*+gxWHMu85i}?+(wQx0?-EQ;*L){CXi(ezv*mMv
zp|}57GH6~=5)EWX#bcrt_G<B6Y&hI3GRmydP{sJL3Qv1}MfrAYiFv)WS$z!+-RcZ4
z*_kz$B{Q8RufHrkoNiM3#2p}xu$M(McN$)igHFZ8-6R(UEd)94hZP*w+J;be5)EhT
zE>3jTgX%w3qSWJ=c2k|EGSlAHQB^uwbBa)QiA&>x>vemJdd{Jc@i4O+-@WFu@tKSQ
z=7GGRj}aerA9fbTW>~8i*gJbvAw}qEtIjsLu_t0@?zV#Bs)H&~yR+sR9<<l$DT$;^
zaBPc2X2g)a`<QkJkBn`aJoz}X(EqRQmnn>AB8QG#s%ks2P1JdbLnD>uj5v}-UI)x)
zZ=q6KGsXZ4?i61vpV$(<?w!muQf4|>W;#b^x;i!-Cp4~&S-rJc@9o%RvwF5gJ#L3g
zo>MScCm|M?i|IReN9@?2ccgDEbo088a?`fACnmzit&qDo|2qHDbhz2FhuQKav*n!h
zS$A*4&}}j$s)Oz_)A6zMga_@i$LskP^?US{1XRr|CMzAsN~JICHcknS3hxoVR|&4y
z`RQA8+H&F~v$<TxMf^n^(IFCRYj*|?RZk2F&;axPSKbYB8b3ES+^r2AnkU%=74$r4
z&ruoFl9_(1D2;YC+o<{7g|BX<?&>_D`PBG*#7lE<awzQfM+c9)N8%m%8s<wqpt0?b
zX$#U{b_ccDAk;2W6xlCh=gVWmNkfzN&mXu&bY_aXB#~x29C5Wf@@0^ln~H~4S{xbE
z;q@va%&vb(4nV^D!@D@KtaOv#WlTy(ChhrXZP7p^t~hc#JFD!PBFfAe!>>5j`yCxU
zjMD8V0zj?jAKxGk%`)^#aCUj#-&;wxMR@f2y(P|C-Xyo0GgdN2R5)KwiBN9aF~;=5
z`_!Jy0gj`%^4Dq^qIsb1O-{(zme|#ES*I|GA07?v&RI3-8(Sd#aplY+@xxawNsXkO
zII=9_Oh%`YLB{hgq#`F*w4guyRv1tB&MjN1t0I6kD!g3IR>AwUa`T&<FykS@3Aof7
zAo5N_VV}qJN(3(d>GO!^)$((%WF2*s3fr2Unw-emz}du5dGj91?CeSD<MoG8z1$OK
zl@;s`@ldZ?zVz9}nVjpFb!1a7GcO*vRdqS|KKDC_I-DxO{_Jw<<jdIDskm5KRPRbj
zP@Aj($4e_oI~(C=78z)^TpJsH!OT!c<z4K&hgsxql_(9Bei<1Pd6}9A!GwTO7jZjZ
zVQ`AHBWF(z`=v>!HyBEz2&Ei;b+iC${)$<pvC3*)e8s?n;MFf~HeVxkZpR+ZEt~Q(
zi?lYYJfN~VPxGruJG(d;JKvu^yTvG7S!^KZ(~M$AUhb==@<`wPusw0)m|!t<?uQC?
z&1q)60^RgKIr>mDE8vvo$_X>W-}_`Z2^RH-=P=Ro(@$gPl~o4A&YtAg)rjBqlu&%v
zeR|CQgOdkEyZWgUZd~XpVO0i2;<?}ESOUqKD~z4Li&^Grs1BlJru(4mxEejwJYlO$
zzzIz(+HCn|`fP?)=J)!T*(9x(lYiJdh@AB;<gNAni=jbFU&QOwVZCGHaFxt7U3HKP
zWj7XM4|ta<Hr}5MFslc(Pset}#j=mr*U3zq%S^wP8xY);nLd4TG+{V>%550Tr?SD&
zB-?ekOKb}lYsa2Tm-DjQbhZ>Slnk`anwgE;#HBo(jg38$?kvJdsa>4#2Z0JVm5!Sg
z3<r<SRL9N_s16cU24#cVkb=;iSqJaN>fDX3*(M|SkQ%dcKXyJ-b<jC?!B9MjUHGv(
z_qb={$lQCVAcexoiBp5Lx}l!QHc`NI7utNInT!lVMh2&&Vw`ezF&7|a5?vG2uR54;
zyuLZ84X4XqZYxq@#Kxx28nT!3g4+av3NhqNhuQZ5)|%K>xV1KW`3<T@Wssj6<SQPS
zHj$L;eV<BB(y_3L8yXt7N_O5CGdtsE1K#JqN^tb(BMijF3d)CgX)@E}vEfHlR?nLm
zmJIDbJQ&_OmvinUA~ATCVwJ%{2ek#14f1RBvgX#xOt;0(x5URRuRLhCA;>n0<Qa*5
zk)GK5)Jr;rZp*`u(VwcCBF3&vW~qT@3Vti}hqWyv%1oEgF)k-ZrOoRRv9SU}JJ}GE
z!z%qrXBYA5vyyoN^9%RDUmy_wU4Xqy_k&l@gZVPAuTZ&|hvL`B%OJMMnDj5*65I0%
zWsEZBhPh9zoL50&t5Es{A~*Az*oXg{2Cdn{UM8vzHmf|{V~71yr9b8DZv9F5yXmt!
z7WMqxvLz~(0LYc_N;|{&`D4aIbB5&&@-PTw+ce_H)P1~v7V&yV3l;V<fc@LC^Cuen
za7R?CsRO#*u{B#{TyZ(yEEbJ{A%#~i;$d=u&wki`So6|I@~@6LBY`R^ys#u4tWw)e
z5hK~<p?E)iHo&63gj}|i_u@f&>eBIDY!GBBD~59ECp3+Er#^}6y3*IZixW~9SvC^{
zI^)TtjjC{F)@+pa9)??Q0R~k9D_@<x+#xevq!<PQtXu!=sP7vCM(^sN8(lmbSCRI`
zNw#s_5Kfg&p;8U2?67O~9GWQtj!R(Wj9M^zSKn3<-g+69kMP*Mi0|>%r6Qt9AEx<&
zdXGxhR?2Kc0$2tXF%Xl@6HWToHI*Yg`Y3nrp`il3k}3X=f{S0rAm)F(J>Ze?P!{2#
zg-kHIdx4Jxup%mZ+!U1M*jZ+Fa2$ObYKlg92LD1eGq<|Q=~OpBc4<fNyXK6cWJ7)^
zuv|Y?G^aa`F#?@Rto5gajjl!d3vB-Z?$^|unFDqi&;Kyj^2k7_cu*|zi2KL;Upe%K
zdWf+8Ai(+%lSTppbXs~24PHhOUM@8eO%as<_sR;%+S@Wm(%tPC#~6c0Jo@sq3Y9hp
z$*-uph{6%xz&LkruH(tk1)1r!ieXrQ(f{KA=hV(*Ta84FvxlDv83}-&T=~XP7)dpD
zahje#P?wq3VD_%$8q+=3tR4?RBnC7?5z9p$;y!FoPY+f-rghfIEL%p#6J3yeeS|X1
z=zue;UWy{x{@+1F%e)(#jOtxI(YAoaX11VUnpiip$feYZfsgDmaWz*~x-ChGfAC*5
z=?ARM>lmEh4ydC!!=UKE_;c{Yo$L2d*Uz5RvDCE6n-0&r9a~c}GAZEx-nh{4;Oyi)
zmuIOkw4j-lCW~E$HbwvrctC}xKe_ss*>V?ZS*$%v6cc@Kk`iG)H^>u9%*&E_#5l6_
zxE3O5PmV&Ex1G>f4SzM2I}U7<Nl3r(tb!kLFJwGTb&w)6{a2%@@phR3OkQJ+1z0)8
z_s%Z1J!mhQ;_lSU%DWRAW>N1=K{p}jQ2G>iUOb+SrkzeFfK|vZkGpu?r^;m$8uCjv
zhwdmb2@DI>AqsgDUC5gTW9R>l4L_JQ7yY2!p7?+ed>;fUBZI6t&}^K{biy-(iy0tP
zjigVx^vEH4z0K+e;$suC9z^ZUn&WC_$rAWhnprAYb6YjDaB?yxiZXP>?0Z~laMY|G
zL8M+wNH`b%UTA&xkHq2Thn}c{wVk{m&K$%8d!=UK=q($ulYHqm<`II83{<MWumIOG
z{zGQ!Zmfkc`h=^fgsKkODa2%C%F5{6Dsms%t+E=~wlE=Yz=``57i$tYvj+sBgjmF_
z*aWlLfeEP0^cAz^C$Zt*(%q}IR91Vvti^2kYBhDuM;hd98*3(xTol?AQ8*TXtSM_+
zoL5v{1eyQ7b;lSJ=&vkh)uOH0@~8CK1ecBl7b4)k_O4%+1b}4>Yf)OnsF45DL~SQb
z-_6sgbF(W`vze#DnbvNhtP#ik5JtB~9k(HNzW49rWc%sq*!k}C*^=81V!)l$@$_Gp
zZGJLNuQRrJZ{eZ1t`2v@E+-G{pqPT~2Pn_L$;;rtKphn<QthX20kRw$?%B4$32w7D
z>W5busH`r<hI^a2a3{z$f5;3d$kgbtm)-4+ggFzqqKQ5?$%(Mr7XkDYR=p~>#+0GJ
za2G%KA8ap-6NAC;9ka@Bs@_3yyW>?ExNS1`&qf5AMP3jQ9pJXBB1W}c=J3#w^;i`Z
zrG(f<l|iQJpdfze{jA^}6wxf`X@l16Qt7|));$uSxvq*zR>*HMd9ktdxY%Tidcf7_
zw_;`Mr)!u;&016kJ2YyH<z#Apm#O(pX6hYOz<{@j3Fg%=Iiz@V2F+-1VG>9_hqi|v
zfK9)iJV-D!bi0Thj2Exs-wJtuvW30EnN?oyVFzUdwS|F@ZWARo>g(HuQ~-Tv{}AFW
ziB}S$?A@w@G^?jv&+u&W5kc4=AejQHrSSUf)BTgr$-sm-GUXoCL20OVWjLOmC`z5q
z{G0B}GnHIiT|KHCnj!%6UTquX+w?vfs1pIx;eS%s{6O(!ig06VK?iL{jqWTd2KnPc
z&=p?3E}a`3H4*{19;^DQ!m0FKTR`2;t6fy#xQZQS=l29f;!rNhZq3J{ZH~C1u*_JK
zX_G3|>dB@4t;#EN8E+>hKBJuWCIlh}MA4_~-~9gWo}M_erumt;)6FEGUIC(dp=z2S
zZK723ZPn(jeE-7VZzqZh7fLVtby5qOBE+P3`;XE{otM^U56U_=53fMk^El&~`9aRq
z(kFhNk@d><lFeeR;_g2>M^`E1v(F!if4qHO-BiBO)?}6SrF*EQdlwSaUfr}fcSdfO
z5g|<M>r{V?>K&Naj}GT3F=WA1Bf$leUI?$Z)@V*P4Cp)tXc%!bn;()3%FJ2(u-T^#
zjtR1k$vXCJAf;S#+Qlgf*t@3Fu`KF7mr?sFa9~TAi1C?f{Ah+`XUd2JA_yu>X#sLW
z%enET@=z^9g|j8h+uq<ZsQ-}lkNOS%%6LAM!u?oD+E#;OebCsEkYNuc81GKNA+(k4
zPn~^k>+I1Npj9YBsC70!3O?%Dh2*Os9f#*zEi#=JRvQV%w_}I0xwb|(?tXRfsM4+J
zh5^YSKky$d>im$;^&FW%g(ut0vG=)0POp5{$c(Q;b82O#M|&DMB(5hj<KaCjfTHIj
zXRqDzkE$YXpHb4}q%<ICY9x#=ui6Yi-2f}#$U5@FZg*v9BYyq|G@ksYNbxa{C_hcv
zi^w%}^}w<|OoG1-@m7TsL_rZ9uw}Chp(~3#H2Ob-f>WZ&fH91vn=^XYI)%s0di7JZ
zqXhMeA`ktqk$rEHHT6tCF3{S?0&s()ZBv4|i}z9J^vf1ai;qS9Kl86DByT^=P(k`=
z7uqd^Q@L$qFos5JSx1{|%-Gn=1&cv!%1XCNj!5wO_K)E8y5S*wmU9umBaRFLdRt(a
z6@chb6(kyfETut(v+QOA0R1ZAzYbhzt*-gPe2)huNTJR7E$3E;IRg!sa*T_=$9?@j
zO4MCs4B7nTnNz92v2wU^<b;g9g>5;WU%OXVV`C-7nd<Ey&R*!FRrunM4p{IRl$#we
z&`QQ|^P_|T=w`<RAO2LRR^cRbo8y#&&<D5c|3HO3yEu@imu`~7yx|{XoDtzj-1JhT
zzp}YmeTJh^h?ZSc8|C_l3esPz76g=idw+Ne$?mjM$X?*WM*<>Ot~r6BXE;89$5#{u
z1@_cV`ZO)R?hgWL%W%V+|6OBxO*`H~D0>V0jeVlYI+2V69<9(pT1AmpQ+GjAxq)Bx
zFS|v(JoKZNo)ix&O_eM#yxHM*z`Rm`Wy;7T^by`>JPSU?c#3XO6+Zm>@vr~rn0EnB
z^w4Ub{vC<A&mFBRoaRb_l}QXxe%rS&vmYbt;9;~|#L`Vg8Xr(L-Y(&$Ud~e!jH@R)
zBt9FT1xi)=4RXg*#lqrkj+vdC9d+`Xx;HzD48Ji|Z1}6uB_=b|^Y3u;Mbn*vA%j85
zfrY>QGB-=`Zyj%nnAtu2PK9$u*TYYz&^UDI&&jr=3uCpDW6c>DF@%42zxUrdYIt8&
z4lUl|M};(duX(-l^j-5k>Je8DznkO|>E;F|y{t^W;_ZxDtIYry@~4^oOC2B*Jw>O9
zJhJP%M;`omU3pqsf|+%wX-8t>Rcck!;>QcY_fa_S-|5u_Me~atB?*K6S{_DS6YCuT
z5_^`-YzJ5qZ7PP0zf5zrMuv5jBqfs1FitW$uA=lWSG%B>vgHaVt>TV1EjnQCqu9&$
zxbM@18|dq-{wZeUM!{`7DB)rbJXzy%p9(VA2o<1z%X^^9_5We(OTeLC+y9;Fq?4jK
zS;`)jV@rkXM3z!fWEs1REE9^xzEqA(L}X2tBne}i?0X>-*)w*EESYemu`mDoJLtUc
z`}=pDb9HrHGvDv?+{@?w+{^Qf_gbF7wE9~5x6eC{Wa4gvJ6LW3K<kyq2EtRY+Ls?Q
zk5KYih}kEITEpaZ=WyymqWsPNp%Ri2&2zR3-e=GGd}{_D=}N66dG^K?vZ0d{O&$<B
zE(F<~dI*)?yvejgMSJ;m3peU$Go*Jq$ns8Nwk(!?ayBEB<t_!(DhiGr>`F2^;p<mg
z=xOCtT0C`7JE4kH+Vy|5+rL=x1io~Y0~7M$!=Dt;W*e+e;=IpuRxn>7j>j{5(QXVh
z(~I45b6ZkDUAboc-fGEodu^e#-zaz5A+7sXV@)bG#0;2*=xKA^`5WOpcA{bbV}j3<
zGpLGs`otZ#6dH^mDI0*=L0u<6#Iv#v!!%(`2p)mb-J7)N=Jv*%Mb7baA!7MR4Wlk5
zb?B3puQ|XX)hkNe)Jdf`vI@AjQk;J^=5VqFk>slkg}3o1@)8S3rGCUhPnR4yeRxMw
z_8(}5&3J!gGh{tK!hk*;i$>S=aWA=h{}Z9@v~umP@aw9p*Pn#$B+o;L@oF7}ly!O;
zaA#mT1{g0vBb8>GB=Sa;B3KI@U1GSis99&;>0N$Q^KyapQ7=XI6}gDT^r|1nW{1ZW
za;J26Oe;TaF~?-d0h!5zQtgwt#!TFHiT*iyYc0ux*?ZB4?<Xx$HTe04rwKtx8(G?p
zOE>R&>2;Fzwb{_Rnpi6`NR58Jw8(?eHRyc$iWlYHCV2mHm1_l_1v<pjbaVEHlbd{0
zr61{|nb6qT$<i<!fKVS7F>!Fn8W9p*HDu2;?q>VCC{soM!E@ikrAIBx8+v&@zaVF8
z&D;*B62fpr+{C?~;bZ^G7#uHHV28kDOmZ>%`^*f5^iKwcOyphZ!D{4+=@lkdu|veo
zdc4FI&SC9#AHM%x<?j`hwI)<V{ywx5*Y`V$>vzJbW7DN-WU3f$nzeC-k%n6STd4Is
zxzb&b7qe&5Ao=2-?=wGw$rR0(o$IWKxKa@Nve_0l@C2)h4KH9u0`<P9+o<LF*MwF7
z;rq+hu{Dx^Sm|{hM~e|jrAnp`GMK(+z0Y2;c|d4f2^||vZAPjOzFVkj`?@|e>PFbT
zHvJm+C~6pRv|cXabf><d)A#HL1e+5?*2bBHsP!5U!@qtfLOgx61h>DG&$4{7MX-)&
z6=F=E8E&uYa+^AqiqIy1?xxJfcw_)TVH+A+CwA-k-$Z+{7#i;%(c73aF~Nivu%-<5
zn}+DOm|;x0&1t^<uM1?E6_6SVvrZ68k6RKNj6TR-C<voIm%}JHg;SsNJ1TTSMe)DT
z9^@9=WmfZdWr3KogXALvka$QK&QDS&#1<lDmU`xjjVO_PPM_um{~_$VcHvPDSo8Z+
z0s0p5lfi_#udAR=JBM(bo22W&5PCyr+%a!;@Z5hzBiq+Gx%Rtz3j2F8xXbU{JI|3X
zC}0N64-%(8v7rslk^LE<1|B1;4I)@Rw(W6JzB%leGJPnby0BMEzRvMxYYdfi<RYkN
zpqkXk9A%kplgjlDY3xU^cC@U4^eYG`l~(BNKYfxFO_Mkql>C%zAP9K2sp->}74}P=
zf%l{A!$gfuk!2Z16+FQ`f%3Mp8_CEVQ_+)GS?UKY#w^eKaDD^IB@-!W`EM^TnLB5!
z-^4bcvY<vUh9Bg&lzQI493o3eH<HbS(|kX7dr1Tli@o$cFFMuqEeODtA^;1k#2+T`
zWyx0p&+iPwouCDLp)E()ub0Wm%t*XEMQKqv`N%*4BWj<c^D1)qZKxi3julN=D4JrW
z1yHCn($d^puUxzL`q&3%p3pdyvpjbpJdOd}z~EdbtbclPd2;<2E1GH3Z+>ln3s9uz
zRg{`O$J>)e1<K^uOTZ75UBV?(@d?+LVzbrA{-MRxYhXiLJmJ?j2eKv&dyaSU_bX+9
zr9pGV;Q?W#gHLDbl$4vC3i%>*t!P+vNMw6Itmwm$uB6h0H+0#`XfH)U(CV|~$w93-
z>yOX8K#}X{BUl>L9-_q(5cGawp3!VhOt3R0=#IiW_ox^XOjcQm8+2qU0B&i0qC>yL
zHFc$xPq^<><!Y~`@JYC>&K%<~=tg;dm<Ta);Z;NwOP4FbsHkr{x<zlJK)(>Z@BpuI
zzVlKEw`Klhi$Gli#O~X`8{%|=ILaFP=3hm`=D#D;=wh||HVoPZW=`d@Ea$W9mvMu)
z^(GB{VniDU6Q&MG@(tgsu>S6uYfR`LT3jc8r6m8iA>=ES4s0&<4&kJk-qatdvCx8f
zb_Clq`&Ab)9~l9JEJf3_o4nNsZ@p4r#1l-osk`#SMH1Wxd`5-rZ#P28{wE*Ir)R5&
zYq4e4(h+L<h3ORs2$>INGja%8?LEN+(=Xpb2Egjv%(h8|$~Jk4>ihECIZV_Djln6r
zGt|?AlD5ZCmjB#2@-weScdEts_ZS&n&5&5Ct(n=_HGsAWnsEaxK5#9v2Q;uZ_bs(Y
z%<h8+ls>t>d<#y*<u0x<Bo+8gxAY}Iow1!3My_0L@!JEcrK4*(3R&iSSF1n_Prnkx
zk%=e(t0kg!7av8cV)D!4D3bzf;2WXJwMH{%<PlnYj9om?BDrh1W$&%I#JeR?Z|+)^
z;C{bhE&NDd?-@q)cvBeFBF)y@ryG@6x#+=EtGzlZR`O3vpQ!y-xAkpY{Kf&=3KB2$
zCR-@T-v{_pLfqFXynu!HimZZ_yHmkd)&cFX<;PHIs=l%f6}C)f?A4fOy_PtyWR^pL
z@h&4vFVANtPiC6^=&zgEm`?Iuit@X`kJ;oVh8+$`wD-`}=oGM_qX;*{bJ?evrX)vS
zSUHZ*RF1LCR+X<64JVIt)^?Q-i|4G0y9F;p<}^A20?s|WO>KG;o`*#pxh?rlIx6dl
zBo4BTpYa^CZe}jjV`SaU+C}n6MJVl6@)3G^z)vp<?63@HnP!`kQe);iLcw80w({bO
zTMDnS7kJHy@%qs~9ryhEl)V^{HG*4(3@`s*r>eUMHV3v;tQ480R31^e9SpW{;7HN+
zl=*t)b4mL1hNht>HR0`&)B_L$V^sbZYbvUZ?EJd<EG7GSdRs)*m~wPBBR~B%K*;><
zzpozHrGE2{;tZl5@s%RL+>!sYq)z$nSFRdanGbH;R|&qDklPmbmb~JOk|D1x6fUZC
zR(N*^1Rs2XX&YE}Zog{kR;P+^Uw->laNbB*jlw?SdowWlO@<J)jSa2>wqV=sdGFSe
z?Qt(FM2K3O8RxxUA1ICsAl$qPKB@3dx**B;xWXmk^d>ArVUPyPt-0)sj^pyYe0)V`
z428S+AGPbNY(~_StwNGTGImNwnsg@X@6JA%{hxz+TaHCExe-FbsJZ{Wo`7s>=X?<N
zPdnU^Zt5PZd?Km$+=pH>40VX+7h6b?+scxKpn89=1rCfR>u*AOU|4}|C9|yX8XGbQ
zh~J7xB6p4{%IX)|jV{CL|I&jro_jZ%1IZ~htP94%!JuwNY;6?Ytan8CT)Ua=@9!<)
zBVJS6;fildqztBQ_6GfXeeBGAA^xPnTjkV7C^a4>vq@0Bt{&a;A;Sa2(-Kp^>w~}P
zb}qk>`}Rez!5_NpE9YrsMs-Vxtfh*DrXZ4n93@N1M;&Ka(N5pboFmK8AW`|(M$JCT
zRbG}R&FAMR8o&bwRY)L>3UL~P+P95R0j{=py!%+Sc4_6Gvj_R>q9v;mZ&q4`_(QG9
zVUyb&-6R|qZ~27Jz41gM!6QTq+?v2KI^K6})~@`wKb0<ync9%B_`oOJ&!n$H&R;?>
zr-bBbb7p9<1tk3W$1e$I<=hzMHHN5U5^_>&N~sYsv~{hy)Yd+l=kC6+>5$bb21ngX
zp&s`|e!XAG!_B+LtwMBgIg^J%Eq5QYH;;H{e<+PO=?8)Nlc9A5tyXvF=(gw+f(ZhG
zUpV|`-G(<-sTw2xN#$Ha^V)vtGrA*(8(@{-aAFaP$I<-5@3U#KXyVrbLrfDxC)TQ4
zHe9O&cIK5<8wsKjj}TM95%SC-`={yc0ZlugMwPDm!lXg6SxY8!z|ZGCQ8?3->VKlL
z#vfgPwNVRc^`w8*6SXJNm-9{=R0bO|0>y`)Y8CBW2o!Qs!u;<Brdt2vdm;Nx4}9sI
z<7nqQ{X=Q<_=T5Ap_21%K*l!5DE_q#zPBCEg2<kOZ6+j5Cg+ul$AlkU<*xoe%1nl-
zgs+7_ZjJ~m*qVm}6^w5H0!$CAc(Zo@?@B>=&Q|tn8i6YMo~@O~0|+Oz^e8RUFVhSJ
znOt854-F~CEzF$*ST1JX7P@KpXFKm}`RnREe8N7YtjiF~s-?e*iV_E3!~crJ|FLJj
zx06Q|g^b$5!=e=PM<(Yp_)ju+MnecUB6+%PPFVfxO7LheiKqE0WSs-GV_7CoRP+y}
z5~+iMiLzwr?#+(xoo!F)<ykYSa_HTQW)dZ~`_D`E6tM5xXAT|W*q>F~<zwD$@0&2f
zC{$#voodyfZS*J{ShLvGd)0GA##-0$(>Azo+4=Ju1%Jw$wjCrAS(ZME-3zy^y80qb
zncO{uen6NL@|p&CJH)#!4#*S=m3A***{90nENtqKBfDedN@1c!J1Z-ipKtmjftGr0
zm8I_nWUK!p!5M!^gxFp+FDYmha63a|VoFN=X1gZIllJyQ?As64)A-=&soORg8mf6j
z_*v1PL?JNHNq>TZD<Q9{1M)+jrj?a109GdWUto3q9|=bXAEA6Teb2Vaqi>$DsHd;1
zKa5;9!6clitQ6lI&5+oZU@0qFdnXwJ&R+Dlb8;q8dGDk<lg`116(9{7l%b)`zU`%p
zC%n&kLp+?#9i^gw#t9dW6Hfch<my_BV;hAJE4&qpr2q5LY^R{g1l{0o$}3Mko<SB5
z&d3c355Qg@jTC*}B9(@(yBJ1_wr2rk+23}|+gAQb47%4{-Y*0-H(&q-O69Qe+qoS^
z7#5mHL$^2WDsd?9kukhVgZ8Q*X83sL#&-zpZ0NpiANsF7D>0T<kc2R)OjEb?>i3RZ
z>r6a%TtN#d&d$GTj8S#k901|mQiNpY)6d^Dv1rK}hy-BpSRz*P(4op@uOm8-k3hB)
zGB7H%)gkfU>LpktL)mdLNjR908Wz=LjK*~4z56Swq>|J-1En}Q)7AC)w(=n!I9yU)
z@(YmE?xzLE&JkIlE|5w;b&*)ny7$pJpUWi^B^XF8E021vkZ3|{{UtP6DAOGw6KQ*)
zFTq_l!hk(NhPC#9Kt(1akquP;`uIf9F71&NkIsDtKu~Vp3uSqqzHWw3lmpKaWUVTq
zxvl)R{(31f^+jA89YEJxE030CU$t0-^W?z<`m)9_Doykjw|=es*!{HGwsICCH!$}z
zVnx}ShlL;M>IJt3Uuc1TOE|T86UTn(AOq5?2*2o5HT~rx=L<Bmo(Y)n2$OXD1o7^_
zrO#lR)a%P$)SPjn*i(XSXP(CtRtM3UUrO!E@h4QI5=X0}?)zmA&G7JN4H?j9Gs0NU
z^%`KhB=_z!*Mq#<*Khu<7EBIoLi<+V3nCDoJ3C`Dn2HIhVQf^tI0JD@rGTF>(o#c{
z_;y{{&{o4AReuDlHyU;KpDO^t@bKx6S@)7Na|9Ql7Sg3c?$ekv^E(4hIsVLXWmjUl
zj%Ejb5e980z9zK<;h)`Jt@>~f1>=U^e0;vg##d(@62@^b3c`sJdpzRTMbCeo6_pf<
zA9{pe_!^(o-TMKM7t%pfY>Z4!cAmN+CRHHoBGjcG=>lgTPfy^GD^MdprA41)6mkCZ
zQ6nE~)p8*|MWxV`=9UnA1%vNKAm6Y9YIY}T*2%f+=(}nd5rkGOH!25a4dQ28Vbo!q
zZ+L}&xJkyw$scKT-W8pk&`W7Rf$56ciRChGIFat1(3|pjIpGmbeLwx<SR4AOmA*cV
zR)5xnPbWhk*iiFZFyFAvu)71*Oo18YH}<t_MMrZe+{L#M7e9o%D9D1EHCL;3Q?CtS
z@NR@DwdJo}Loj5v6hGJE^ba$#$-ClrDjqA2bI#dI8bO0AL#Q6p?i5xl9(;>snn`bp
z$z@Uo+(4AweCR0v5Pbf(*^SmSGxv&(YJ2C~@r&t^?HszA^j+@#!WUsMfNzi;wP9)^
za@}ENfq!^PFtzGV7BY(rXlV#du~3g5(}Ubm`sg0Qe3aCECefX;pOVa%eK%@C5^j(@
z7hZo~Kzgv!5n;G;Gmd&?3QnE40HXwFRm+d@txvEpNrvG4B6XI<vVCDK2+hk<zptlg
zW<KRC%|;}mryOxjseAUB_=Mxy=lCefFfCF@;~!W4Cag8m=2w^Ti!itlg_@=38y4rg
zE*08kY|e}_{}VL?6M=s3^D5=}d~N;5Jr8dft9rM^tqgLZhL}-eq4)~vb+^@W{2~Wx
zwhUW2&hs*|a<WvLHtHg6Y;ZKzMYZG=xBQ<(qGbt-K~dH2_lRZ4e<;<JF#j{bAh8UV
z`v8NC(>`#ASDfbamEc>~YhKC|=avh`_n~I<@a`I_-rAbq82AypVBvba(Hq&0b#7L2
zZW{Nt6ikKbi4AnUsh;X86~7o*vV_*wK+O)~-JLYQ@zmYkwSvo<Nl>k2N{w8vh@VT7
zy48zF=xkuNwnK|dPKYn#xc*c(0pc|-WJisGpOR_|7xq%{7?VlSX#@~mp-x`zrdB)W
z+-#l%^BFQIY+C$WRS4do?sj+YD9%9UZ6<gO%$XF7O~c&eI(-(t0OqRhsCxUTt_Fve
zCGns(nxZxZQk<JHk~5bPFG?gzP6=!K?jC_r+RJOvgbsZfIkB}EBr7=oH-RJC#aDJ*
zzU;w>JWG9Msx}O721T}C%vlznZiAU?23f}qCFkaxD$;Mx%`laHQFT?TYgSiNto+>)
z%|T;Yqvo%9ta%Px6~tKA7~L*V10t8v3+0CVZEC?sn+DV18>WXDp-IJ-V=#vx@ep6t
ze!EMaZ@t$#6!+((uT8zF7iyMG%+D1OS#`z4^4(OS@d65t72T8HDXe{Ww-u~E0g$QJ
z`$9fJ-VNTq?_n6udk}`YS|+I?p%t=xie)47QsePTx0vd^bXC2JN9L8HbXuI6XufUp
zI=i5e6E!E<_Bbr+GS)l-A%~OvuNB+~r$mi{3E07ScxJULjEs$+gJE3X9GlzUmC!CL
zN+mJ=qw(XY*%~f4fGhJ`V)clJ(49z|e7(!1`D*skp&LL`$d$)OZBA-Y>t-&rtQ<q6
zY(5u{oB^Oc(Y9$y8z#u=D~Th8a;VvyoLe803|&@Wa3z+u;oostruv=B6XTHoffHeg
z!iBpZ<g7oySF)iu`z!#(_2c3&%l1W48^(FisM+Drvcab=iskQ;hgeW<a*i8BVNNJp
z17m%jeLh##Dx3;){MK)J4H9V)PojN+kD}G~-{Lc?7@JR`P(}C@{hFp`DjSqtn&05v
zwd+kc-s*)iAP+OSWWaAYiU-(Wg7ThqjdP}ngWvu~=hj00^a&(DpvNY5z{k27VW|3L
zB=;k{(_}Fvy!iTAPbRaxv3c{d2UZ)N^$tlezS1Vi$$>ml`kdR*Vwe)JJ%LFz{~>-T
zZMmcI<+#CLX^>(zxv;}0`e20cVWbWUHN=K8JvBbmuvYc^S{)3;ZDfDxuJZVRB$%+Y
zi0mpmf^+T7ruvf%U?)*n9lzZv*Zj_@_ReoLtpRoT!*o2Q`F#2KDzitXrAFiZin|WV
z&i@nFr{ui9etjaY&u`<`x(xbsM>$919$PBL#c(L{Cw_z7MkT$K)+-RIm;e|Sj1zkp
zy=^JT*t0}xBuqcSbn{<)W`|LJ@_d!De6Dib>m#Vy{c8!PHFKb_2}#ZstC-cN$daoi
z?T|=%n^H<GP0`s`djqN!D;sxDf0q5UF`>0QWoW>Pw%LBcp8wAvv4@wR)k1t|-}G2K
zQ_h#}$L7N}w3<WL>r8_t87aj-OQ%7sG_9}hd!JrOlKtipUQj_-HZ23EqVG<@_;{li
z|8No<bZ}{Y?673Q{k?c99V<H@5wrNzMpH1$*ewL_K4^`U!RY7~Dnb{U4P@%QLBc#x
zVaT0726f+nt3ptuX^B9(c^8Z-(N+Tu((vKFQ$6#4+_JP*^*+tV$p51IM+wXpyY{=@
zxm=%^RpK)f7s^Cez0oq5Kz|`(ka}-J0e5G?t0`S3we8e-6?M|S9Y>B_yeoab?hNn8
zcSP1hJFKj{cFx3A=N6*U_8-dK$*E|_!&xT01Aj8GM6~BT@AtaXfes<fheMj%PO&$6
zE%cezzws*{=tTECHTAisr?2Oyh~nyasu#6X-qAx<wegvIMn`ugr8K;tvO<er_)YXh
zE`l>5H$JF?6z>8RkT$y+9}(F@u}GL;eaK7sVXILKrxakw%||em=Yn8@-7WcnAl#au
zL)i(EQ9;X27_iP-3Bn6#ow<a_%21)CFuXuk;dOWvTA1ev?a^}=g`1Q<^a{STkwmDw
zNVHbRPU*TtJibexpu~m(gnh$5d<=Elrp~2Vy%zqNbWza9u6d~qz5fSYP500~xvV#g
zxZkigwFuf;yw4yi)Nr=(e3BsA7g!B`ZLtvC&!9~HbT+I&6k)Nsxx3IDPh23ht*^fv
zPmP2lIqT}Gt99Kx|E%hwX5|D^4aVY!q}Hx-+sg4Zgyd>CUdp=aX5y%Aw}R~8Y)FQ!
zcC*axn#t<PLUsKcg#JI{kD)CES#rE=rEay5uJcpQ(qehC#6LEC8ZfuqA~T)?V`L%t
zjND1!S#JbWo#iG81FsDY7z??(dg7?hbbD!C&?_mN?2o$bFj;@B?IE7Vj-De9ANW1`
z>}_8=$yO2~ABxd?*rM$NUdj`aOqC3u84HRh!+miSHG6rE>#Dq7-t@c3Lervz3?G`g
z>t18|v%H5SMA8&*&`4Z7ERtr~p-fic$5y?vhH1{bH}()uzznh#YBmPTy-toCTQ=B;
zV4)ZZruL=fPNj7cHfos*Y1agBTDaG$le5+>$mfP5`-VtD4SO43yDO{ilp)(b{vBQM
z7z88or!x5%%!g0m7YBj@6M0Z#aU)kU;w&M}O?X#GDV%*Pcyklk&g$ilSF4ln>C4>^
zjtSLV5;xSU+vi#KMV(wq^FL%=Y33Q|jUbuOFL`m)mZ!AuC;s}rDL*#ncy8$c8^*ej
zIrp9YU(I4!`a6}$^Mmy-*6vBM!rs*QNTu=X#W&W($E6>*kqLh5%!O;^K{)@-UuV1f
zG;{BMMkad<3&#5Wq|yC*qdM{G?0;dnu$U>PCR5l>Fx*lj<CA}5jmw4eoN4TQvf#~?
z4q<YI!g#q%fCKF?75(eId&W51a{eGX{sH=utUsD-62zvO&wid*Sbv`|s0xokoW!~J
zHF_v%<j}gYYX57)jUN#=VN&(yYscs{F@c^F@FBWe(i)B!E>OC0R0zR^=l5GWbqvRP
z_wPanhY<YlxxrBR=7y%z(Hjpv_VbRHv!JTXXItlVG+PwkzN&hCV~s;A_cHRR&tVxq
zGGsJ<VpWv-*wIh(GH&7YTu$b82hCl?uFH@o@)s0YJ86DkxC<TAA&ezgd1a^I7pZYn
zbKHFg{fa_1)W-U*K3e!@6g0SFI(QS4#YjN5Xr)*Gau2&mlI4-RiR`u*2e$S%YjM9+
z_B2vQn6HvJ_GLr9B~P`sMMJgIeAy3Y6R+_^$5RVVxYV7iN^tT0@w$&Dz0mB8myWS6
z4qgf1MOmfcxb+b_tyhOI4pLV>^d4p)dL9H@H#~Zu&^Vr>TFW3ru7U#(*&d8X$5-Jz
zmQmd~@Te|Ms|d(cT<ezoJRUh9?oU^LHT=LwR2$D8V*8^Eey6`a?7f&4%jLjZ&z^vo
zhySnjHpBUk-vv{fnTr-|0KvcmszP%U;N;0@P>vM0m7$W!n{^F1(86hX$sFk9mD$#s
zt6b5vY0n$aGhUt;U!E_Wy2mU0=2-sT%;d$g_$TTJ*`MD-Y$lrr<K3orX>^5_h2QCr
zQm=JVt6d68D;MMQ9WL%udKf=9qfslSCx1s<b=6+;Tj7O1psdU&B_5QXjcBE)jg24;
z$;_P9Hbzg5AKI6>Gt=KP^xoM6Z>}B8-W}bTd(t4EeGkpZPP~J6B2s#@V4=EtP1PGr
zs{=@Qj85suNwG_U18|<=6@JklzbJK;>-rQ7L4S>xN;@;0vI56d(u1gC5AjNu1buFI
z2J2{I;cU-CycFndBO<90hFgC=piziRI#Cz6#=I@O5@5I2aCh{6&%RfuA1~Cq-&+NJ
z6C;V7{St{zADB;k=opO88Cn8%3~5W1MlHkz9aWQ1d;z}r7~ga#UK>u~ynkysVyWuw
z(!3NR&Oto_XUt#_{32>rIM1&fj#W9OMsmX<)N4m>&lP()!Z7bOdP9EXAe0~O_)c1m
zX0In{${0u==)3EFY-+|cS369=sQAaGGep<D-+I=opFtz7nh%9>Xx=GjUQdO&V_}(b
z^Z0gmEifHqkB{(4k0$?_a&fC<98rK1K{DgQ9qhDYP`9M@lHcN`i^KIY!VYK1mUG_D
z^$vpAp&F(~JE#24{_}gTOK)4$Ym<y34^XlpK~=4VvHZtL7UE8Le9Tz9l61mVnjLCf
zWtloW#<})3cV2R5lk3MToo#w-#k*MzPRaGg`GrdM1|-ZspwfukoB&DI7)YEs2MPkk
z<AhtrFbNL-1wvBsK_=;rmpDX`DLmlqY6res{@q2MhxnWp8NaK|OGlLr5u^XS>j8@>
zt|nfR*nTL4kjoCvUUu(jCSs6x>pJZ`=`uyeS05H1zV2hHs;c?zEq;-JurWnJzQlv5
z1@+nhyfhoC0$<FDis2hh$GaP=u3lALg)!=5<NeRti+yGJd_CFwGJ4kG)Q+fN>P#pe
z4uP`4)<f|yOTUCaq*_~tcXvb5UtfN{b*tu{(Hvx5t+|ux<k{91&oma}wTcdp9m*ed
zvZaL*E<YF*7kSyPYtKvS9qh9)j{s^^t%AZNq{Pp8@ePAgTrNMh=61bbu2eEdb2*WF
zmTGI(CnSk`iKn$YLkOWVhG**1U))mxM4NZK*2AM&1E?bMg#0xPc`@BBqiYX#H--=f
z=DSDyGuYnlllf{<vhO0BeEt@2`LJ2Q$3&R7&hD!_tEt>Fw8xlYA14`c7psYjU&CFJ
z9kwWm?5pepAeIZ@%{-2bsMT#9Rj)p;mv0lxzdNAsWR9DrvOYDIreYSy-&qZ0q<YVW
zO3F^D@MkidgtxG6UIIi%$3IFgAHdbMwiwAkuel`gQ_W-ZAhD&{p<siJH$#sX6g(X4
z`>Z|mto7Ae;nm~O`a3U>U9nEnliIfjA*O%+y7eohhtl`$AH(}5YdM+oQ<&$@4lOl0
zg%tTMI<J19Kk?peqbxA3$TC_#8=oZv&OsC-^;Qc+CXmhCU9V*0L7>mCZk^_Tbj{k!
zx!Ec?iK^xGeb(+w-`eboSZJl+h+I-^uPfRNM6F_*d#S<YqId|wYvnrwjiPN!ZSQ=b
z*8>lELg^rvQo(n|rjI&${g?QKMPhtH2q3gFkO1YEKhpGQ@lq&K_M?Z$&o_gIl?vp;
zgfI_;IM7_qx!0WS9PM6(<6y5PJ3y5d$y@h#oYcWJ&d1j~)s*Wq68G*prOQ%}vDOyO
zecb@c<XHTh*dZiAWUL58w7L1Sop4?U;ehe=DoWTX=P=xZ5!4x3ss!=f4~}SGJ6(R-
zMxGh}mtvy8?<ZK=GbjmQFl&d;5Y2ifTW<D=1eRL;X~;dwi_N^pi!G`sYKb@f3;a9x
z{I()Ku%y3r=BIeNt$zBk&xLUNo2Qd2hmGLZpl<ul+j~W_f8^Xm;;wDc`|<2S5dkrH
zYpCj0aXLK1-J%&)aXqX++}GYCI)tEYI+WN`h-L&90Mz)}?$r4X!~(SJl15L#TU$A|
zt+k=ZZ?Kc+dw^b-rZTy->rzeK<q$%escA7gEY{;i6fHU|^spiQJHtcUy8iV0u3!(E
zrt{uL?nP%hviV)*r=%P7kc<vH{m!blv``EVwH89yIl~X@4Bq-(+q>a4`I*riX!@FO
zN8w9z-WQNBu^0i3XtYMLqx-UEOERTWOCS}#8?kkpipAHHOVv~#W%9T0US-$MX`3Fz
zC{RJu_pEg)uOtI0s#GTj1q;AifX+f88vgJv0d}nk7W7h=R~CG3yLbup-SE&W*u^s@
z{X46Mj5rN8Wv6Bn=IIAD;Pb2j+g}X)K%ec_Yt^WyTr<+mzHdIO?2aiv#m)eC=V8)Y
zXTiW0<NZg`z=lACtkJ)|^hauT3<o;xlYLiacXQokWpYRMWY{%iPgfX;XWB@;ahIPV
zDVu-*LYDxbNR~V9V^eu4`Rr(~wYB6IY1vNN)_o#5lPMRUutZ;g6>2s~5$89vn0?!-
zVvL}mu`$L|uMY0+<^Ik@G%L*xr{)U}J&_;2RWFz;4Q4EX!b}xjR3-}+JOs<X<-DaY
z=1Yw8cY5szJ&w{@yjV4U%GM$$;&lEvc~W3>s7K8f!*#wE0lxoyon6s;e?rIJom3%q
zw3yjYqS2M;V|ueAErEMXb_QEral|-$KG>j@*?+v!uNg%0{P4DYTpM^_LFlkoD{Fn#
zm1N0JIjLdNGpTL8B9fNTnxU?5c%lZp={I)t%1)=FAdr9UV?=@&y;HNkGxHA?)6&6b
z4fAL%y}%e=0D>QCe_lp_VAe+~^D>fj{D*<$B|3V+Whn*f<hReA*Xo+4(r+e>UOS9Y
z_!PcZB<P1k8^lSRIX_mLmQPv!!UM;qOkz$r&PbW;OLaYAk5TB%P|0LRTLM;VBbt`W
zntiP1wPsm9{xI`)ih&ObL7uq|2!pggBB5W7{CPh?AA3B6Q1Ld`c}=%Vl9+fyB{O#w
zU=DIol93~3?bXIgl{)aw>52nY%c41WN8=49$o_O!)o2_9j6ubJN<YXoPK_Un3!}Eb
zQYUv1CMRSseP%c095wIRy%(L*mp%ARPSN?8?^>0gdvgjR>+{>LUn5C;m3O1;ZKO&7
zE`-3A!vISFnQ?;5)NW!#J%2vTbm2tkm-V)GI3X)PuxC;`JzOs=%6{dsm<(vdzg@n~
z{`r%Fj*Jb%pU=Y(PkQsOz1}AbxC+Y#w6@#=n8@-q4a0#`*~sWg5rc^DP77`{oC)Vp
z_>0i%utH*n#a&nz0$T#?J<9WfLLAVva`5cHVzLIoI{;nKomY>%2k$V{Bo^u%3?Wcg
z<R?zSr*F{o1F8+L!RW|kF;7XXsgnnvmxfat_JdOqGJK#j29_Yv20_s%>ssHgzBa_8
zTL1Ek0+GNxsB^4wVNsvmch;s3d(6Oe*YD^R-5$lp^yP^sm;me3e>EMPI{t(J#Rok<
ziCzaR0&2FK7vuqI1Uh0wccB}L9w}xotLq~#ocyO9yg1c6Ts1SD7kk-4Qc?o>rDMmf
zJ8@{x<SGbS?v=~V>V>slD{OJjkS%?6hxD6^&U%7_5XmUMENEGhO$S*W&$a?0jps7%
z7-i+1yyB;s{;pJ=oN2svE+n|~8ug8ve6{o8NDef|^4~;V4h%IBf_3ZW&k;d1IzGBn
zw|9sn8-F_t7t5dR!1Rj9r0-8UEE40qSYDz|o_wtCbuI*%p}f4kMvU~452bxKy0ki+
z5Q)$~%<iAKY|SH*lyOdnCe8l#YheX_WLhSN-wsAknGf_kT!$n!K_?^EV9%d$p|rNS
zc}3gM!s?fk_13?2Ma2j&8)TAy3(hog@4XU=yE!d1*(GrwwEpilFF9f=Jd8>X?T*ur
z7;rsu{!V>KL0<idnEPO*PW9s(DN^r)t7fRPPJ^r6*XCQUXOxc=f!RhW+c~B)L^d0x
z-Kg5cTJ~dGz?s}c49AsK#Tb^)t-6O%HPbz)<3m60xz9v&r6_%zwaW1`*cYgBJ~N{<
zoW<~XiG+4Odj@7RNsaxrwSJgECiN!i;Pc(i@>8O8_$hKqp4KyYS-#$%(<!=#cLt|T
zhQ;f+vd2b%sa-jOz0=pp6QYIXhRuC;|MeK=1DujX*504Hb@$S-aG)ENL5m~GWDO-(
z+|4B}o{~QauQF1wMxxc#-jRMb9rq=P^AZT>k=XoIZm{wTzr7teQ|9U=wofAKish;E
z`E7RfJq7c&OAR4|6D#p9^2@aad0mwwHTMa^nHiARrqF3w7#4p{K5<EwDIv~hfqqaA
zR2{eZGhGl?UjEf$WFtT_vh8@LPiI?}n^2%7eB$OE^+p57?~+8{pFhZz56Eut_qF%c
z!yi~QtU>D+Qw{qEEJ6z^!*Q>X5#Q$*wypG11AX??8Din9AyVl8&!ftjLgU+(DRM?9
zcB6Tgx;uSxI!UFa8H{4506pqF_)>Rspb<g+pIu&#8E9^z(>X=z<<SzLySDF7oLqWg
zI5qt76jN5^!Ut?#i9&Z#;;4bgoe%MEb<mOOWUZ#=m7V`%#yZCf%KBhypE$HcNu_K@
zjhu>xS3TjO96lQp2fOAgTx0nMO;NY3m-GTM-y8)q8kTQ!%ndI9!a%ct*3BPJLcHkr
zT^dm1DM<(gb!^@X@RI{wwBn^qb}VBlx_L*VaU_IbAZ)%7%EUm-egSG!xqUL%;z59U
zufD%B89Y6imJuD_y8rCtX~D)f`fViHSpBeqR#GY1t}Y-17%SO~pTe*GX%|`vibz2a
zTkaE}u+yRHkCaa`mDr~?5h!wtNdy*>d;TD@8cjXW)nu994Qr_5rF15!lP^B2a35|U
z+xnb3Kn;CtlGq?^D59s)l^-f&q4=I1jWxPDo}3+C;C_S_pMIhC%exjuKH|Hs?IX^o
zb&`6;Ao<9Wt{UVLHHX(I8rfU<e0Vc`$Y(<<gpip&9LsDKFnU_#<H4!j!stFpWpX95
ztg|V<4F1}ko`Dikm%L6#=>T4eMQpvJLWLQ??0%x+kG5G}N~YPZg^3k`9EW-f&1551
zjZW=r2*3;*yV##38Fi6lvlx&QJ<{CUgyy&;F?(tk+E^EWw=hp#^^TmK2&q)z(EB<{
z-|Rsr&pY|k*$kB13jx%;E^7W{R{Z2Sor!(X7j8SIe|XZYmS(!vX6y7>0XF>~MjQ0w
z`xzpu(|&~zb@G%BT|u8C=wiCNv}!JRgMV;t7Mn7f<2@^WJxkm-&MQ-lRBG>BcQrkn
zCHmrt)%HS;OGNS+VkvUt^1qqQ*9##8xwmce0LzV+GvFsE4f$an!>N+WpE8r=wy?CQ
zRQZp`%6-D<4TEU&&Le{L5Zg|YNxddv1t(PHLOk|CNE_LvvlyL@(bo`vce86HbHCv{
zJLNi0k6w5mHe6phmfSs2Xq9#6*@QE1^r*pMOq7g&s?NuUXJW0mnk=nP+z08|{-T``
zD(T=?yTGqxW0fIqC}UweQySH>_C&iv%*=fw3k8H)`(8$hFa;BkdfH@BUEhTs^cnF>
zpUg}qnYNgO_o7b$yZqVFJf5EmD`q483G8SCt?WGpWen0HF_vFq3Tbdt$_oh?1lxZf
za9Hd7p6S781mAzg)pQF<rn~lmeVK7#*7E5c2c!~BVF7^|w}f+x5t|Pn6ddXOQ3v#Z
z?jtZd_zPlg#PGecw{62&$!Q%V*~*XtSK%|IQ&)F>&vq5AOY3)|bk7Az5VN_T>KXdh
zyc2WIe0vm&sC`0W3?U4mzX-#H<hKRQO&tw&`ek33pF9{n-U<JlGVea*WbL!Az&8BR
z$ZN=bvaseyBg?G)7X+||dg63Gl}y;u5}WhCFt5~l#c1Wmp4W#k30nLVcf4)UQUNDy
z$fjSm(8LkLqHd(17l_d-W-&DJ|8iY|Scvr}cnq(CWUIe#^UBkL5v3d_MIYRA4n353
z@?)vOUsA-uer>0dl>^#dp9>qtLkd&Eaiegx;Ex-C2TcQ)ccUrnX#V##g|Mmr5Ki0)
zKnxVh|KM?k#@{{-ta(oik999AnZi0SrXG5t4{3OExJ}<bNz1jDDwm{QhU6LbTi5zk
zlc3g~@Y%NP(`&c(<7H--`LR87GbJY#6XzGD8=A63s(_6b19>Rl6+#&GfNIy>rL7zI
zEL#5l`*W|>Gt)F?UarJkJ{R*aoQ+bB$2V~eiet~jqc|gtmJU+skBm@41h-SQH;b67
z?P(xF2gN3t%8`nV#O7tjn^6vQao~y!{!Y8`>6o3b%kFhLdDpoNzb$L7^C@zy)%LBq
zIR7|}`6k2@Pok#2p=X`o@LiE_AMI$mxs>n<qziwlhhrm|`uE31*PN}1*o|hm>^`N>
zUNGj<ZaO})-;d(LpNhZMHXNf9<S3`+G)@I8*6s|Cf|g1;^UiOVp{UnO<x;9R*lY#Q
z;!L8|Sdm~LNt}isA<p`ChB&|Xw@;2<-hEm68T<04g$jxGpNtiv0tld`Ye6m-J=SB>
zz)e=hfWF^-|F5!ni#ae3X?9Y?>EH2WM%#~>L~@T8WTl<Dos~j>1*um`x%Q74YB&t(
zdG4GMfwy1Yk}@yfd#5!M8y^9QUdI*y@I#jr9yuC^;i8a2q!gLtdA4Srz|~EXmGFaa
z{Sql&+C(}xI~;eiPMth`fgLRg(AdD6sikiy<N@Uv`j282l?xEc@4nhv<N?%3hzG7z
z?lNIizpSQwo)_ZERSq<?7QjOT4vxLLCnFr4P{?SjmQVmY^r9uW>|0iJR<$-+_JmuX
zPNVY6^*<s4>W-K5OA!l^hQh%R|GEfaVKm72={^WV&tQbbI@c#;5LO%VYmbaw8Kr%`
zr8(i$zm>_?>8K>6b(Cp*E+;V~>2=a+kr;VajKhf!veFjWlT4cwY1?>UryoXpIoZ*{
zZ$H%>a>B5v8cFU&TQ0Gr;g55mz3B6^=e49C1NXn>z2(8GwO`)d8%PMVLyZ1P<ACK)
zP-?+^n~ql9?yjjyUP{IXmbFYYq}iIcLv4xS1#Z=KjI{m6{o3y}`rVDCaz{E`v89mg
zz>grj4k2(sj3fa0T(1ER7R#nY@@z%y=;JmVqfYr!#$^Y{TKrfUlIObzAI*|46m;i$
z9<ox_4<QJrO_JcGXk+T4T%Y|v9UtgtWX)F$oj!*ku}kpR(!cf<xETo=$^AxdhFBAP
zIYWbySok~I;3=fczM$++7W|Yng%R@NFx%ik6iRI4rCdEXg{cWCIbZ%Jp#XouAJ&?e
zBKD^&dgGkK3fwP_U$2?8WKNnPw`SUI<VgzN*)I}3%-QC65;=a8wAkB%hvdU>CVymU
zXz5u>{igi4t{c7k>^R0f=w2I?rT%%TOkN+PbZNCi8?e+F3NWKh7U|4>HpWe(hWx`f
zVyW-c-OPC_#b^mxY)Zm@THnXQdcVNSo%QVKiy@wC2X~{35Eomq6->Kc?OKKeUjPt!
zLEL_k$l+C;M5VYVI_zbColB8|IODPvsPOoQ0fapLJ+ZEumK|^Z5I$Yx>0hySC)9Pa
zG&3BR8fARYDI06STb;+^Pq0%4yc}_PN@FhnKH!`Lk#)zbDV+m$x850_Ht(B$T&Ywv
zDASGw0#Z1su5a;_ISoI*!$-EqVQT-_%cfzFBInSKwRGA^nzIqd3rtv_Y11&mtgjb~
zQK&jjYDmmDcM+*jUVr0^Rpvmi{EyRI8+un^n{@y)1p!syCb6uvpo`?`JAzl%FM42R
zr&n)HhK5pV_hJI;UnHaQvng)xZa`J|-;jon^)8!t9KWkhKDZ0GX-(eoP8SFgJ-Rpy
zf)1xe$|Z}~COSSq%};3)9<1-JU!w{A`GuN?P@4SG33_|k#r@SNIRws3tzPvf1n{;O
zBEWRp_Od<GZq|$70-{&=u{MOZz7-_4>;97Y2ho_iu~O#)>3;X=T03=UK2Z0ndytnw
zCL8Gttg@aV^$P7qo9+ScYuzs6@?Ce%v_FNEJ-FguLZ;SF>Q!u~bZIW6lN(BnGD0sR
zl?DjpQ#(ct>&FTrOv2#X)gOMrRD|(~uDwlZNUhcVJ}*UIhSaOjP6-|f!go&t6MvF1
zqC*ooZ$-t~3*<arN^Adry?*!rV9RHLR0(p1+Y{i<S#M!3;vgV9u*XP~wko3nMnjXD
zcbskZqFP}E9fKUfC6LrTm-*KNZ&?&6a4SwSzdJ7gKWhhp+cos3C;0wlcyHA(bKcni
zLYu=>6Hmq4rjil2!P&gUB;W6DBgTH~+Aow&BIl7l*0eS62L1@OUE*=;t7WNU9r~Im
zxi4K2yU;lfA@t~$KZrV4aF`Ey@0<^_wvT81u1RDC`;iMF46S6sFjZ1H_sk+GYB6%f
z>7kS7sQ=~t5O6+!YTe9%&#J0^{X;TlK*u5<0)xD=+E}Mb7%r=$mn}Bo<?5=#e$UL*
z(;tw&I>xZLJGtVDGI>^=9Pd)T@PU$o^V{ee8m!HQ#c6SE(QwciKgFm7C@E$NN<!3k
zs6In}AYG*<;TVOtUv^782hcK#=fvK*I2X%s1il(~`=>a<Md6a&*Zv*%MyqY}!g051
zF<)|4KLidJRIQTT|H+=4UG-{yOiq&1f|eIWGgHVsYj4qx;LPsc1Xkd`#`bXfy25XL
zV*djT$Gs5XKijT$BIyg=S=H}OB_c&*xV{nLUV%Nt^%TF2>|*Z}9={*=f@c{vsQxL`
z`n|;H9fb<Y<R~2+MlZnlUY+D1sZ?5-tYv9NVRF7Xzk4NdZsMT9xD1pR@9JLIs+Hlp
zc^GfTJRyv#mtH^g?Js2zi@fjcH$U>`C{yR0Y`P$5Lxd25O1vIosDQX?LQ<kqME&2<
zL-UKH9*j^%`+QAyId#e<QpZzYwI3$IRBNYDd6jdmJfrbLz1`Kw4(Rk-bD0m_EYkV8
z+l$hmOm05KfzEMadSO1dY9|4Gr)*ZlfuOZ!^irMtPM*v$*4;nIgYj9MZJcE@lW#%#
ze({P7EN~me09;SUhJs72T0)=wMizn=buTYg^P7okKibGgaNOIDeT3%nhH8L67cEnq
z=j<{MVH8w||Fqx8na`W6-vjm3!af*j(9^-$S|>m+;6Xqzn^7<`M)V^OG*@7?A!t1E
z%mCCZ7izW>)iz*iI`^|CKEYF6wLB|-dVz0vE>xU0S8})Gc|XZ$Uxxa3Cq)HEe#$6E
z$9^J|O#{9f0?A5?X)mA;nJiow%UkEI4+-L>AWf<GITgtd;0}bi_Z!!sLF`3q2Oj4m
z58t>eS=ZEYWi;m;T&yR~=Ub7R#f-{<>$bdt&e0g3p&fL&U`Lc#W%9)?ia%jI6ooNm
z!<eE=UVSqNg_JF?z$+Zc!JLtk^rNP!HMmGyYI!y}_}LsMjBZq?cNm6ArrtS``V)p=
z-*z+Pk^}S~&#_HCo?a~~&dQN{?@H5_R)8}EZB@gY*}>Aj{FKzDZ@bZrzx55C4a4bu
z`f6D3sLLo)IlM|d-}EZsT{)R!pVgZe;Q**#>cv{H=iw1)I{eDy$aW*uRaaGS<wx{}
z8rp(z{dn_|8~^amvS7K5HJADxSP(49?h51n=r>UOAM3cDIhcQ+Fy7S{IM@v7j#}A)
zi~&}R>0z8k)FC8oW^SW7ow3)PkzW{O88hn>Q6Bqg7HFRlPOax39=*zC;UmFUDFAcW
zeB5}vi$Hga*n6`BXVhwgbv29PExc&^6V3#snVpbIVtw{~H?H$33nL@Zm8l}(MZvr7
zLP%RPU-a)SFY;~P`J2|u#j<|0=SL~-{Mb+qC;|A7Ave}*IZh+{6XqL!X61L6s-x;{
zs_O0BytD$BH-41y`)j=1<Nm;!vt0=nD#`8s6j(2)o8WvVx1U(5zju==)(9|&@f_GA
z>-I<;Za8v0@+kipsr2+D2XM#a3)&_<K^ls#)YcZUW&uG2n%(cI`yncwL_mDK`~evG
zEg{J^((+pBC^HIfZ%iAHH;<v5$Wct(U|(r*QIM#c0wjm`KKk*k2Z&GKVWrf5doCcI
zKSGwzwk}bdlZ$lo$vq6!^ShBPPv=5ux5=Y#BxXcUY5D!)JHORei#Anrq+~PQcrDh~
zVYu&ncrfD|m-%4@lT>*0rFs*8f?(>nTJ89IvtkLg+EM%>+%UQFs-h}>t^m5$c2tH}
z6}0yMUA-%e^wBlxb=BIcs%eKXT-=Ii_g2xk7A|p8so9<$M=U5h&cJfkdnQLw-u|z(
z@FCyv+VL2%C$xB+Rv_i}b9c=$FC}Ezg8>jM;+Y#iC6i$&t@yC*+L!pbvKz%MMXd8t
ze8WrP{j~ess@sKZt~S0U7r3T}OI`+g+OJN&PSOv<;XX+b_ZF+@u7A$c=KQua`YpNQ
ztTK7Hzl~$x$a_~9fbv0l=5zv5GrW`}K~cCz@^OoMl6G`-(wUJpnkq$7FMO$X6okRC
zU3VL%vgn_xLA((k%&50NXcaKV+*izhyoGC&fs6Go7>#*Xe&!>p%l<yWz>yZ!Kem_q
z6OuT+nqJHB4aU1vo*D5m6B0F#on1$?MHk)vQm6B}p~Q<%P{q&+qxLi3C-gzNGvx%2
z)F;y3ke-Rc`skwL`4%yL=Ee+>7|}rIoc`B|Ni>@JkqpmXh>9MP3G|8fCytNjY9r6Q
zn;IeT(_|WpaS#PIe*0#Gik`^hQY)QMmbf=&rLVvA{vBqsv``>9qb<O6C9!KV&Ol&|
z19mF?)MF#vJv)DOR_6O$bURTz7I_*S%khAnU7GijT%pjC-dyn6`Epo66yKTBdN~JE
zPUD$9Q@J9ArB?0QhvJ})RZ%D@{m&92jP!7*8&r6o<;O-p2Q-U)>Oqfg^jk|vs8O8x
zOEiYVh$dV7ha&8Z>-!pwx|8je{sbvc|K9hXTqp_W>-R}#v9RwbTA>fjFysbIkUuP8
zQ*r0*+s;26Qb$Xy!|xLk<`)f5IO%T~d-@ZmU?9%o!59~&!o<k8mvbI2UmRs>u|)U?
z9o<D69qHVD6N7zpTxXyV_nsYf+qc5ywZk>RGek!Qj;u$^w+fC5OIWAqLN!lOniO}@
z=>r6asc6vRUg#yuRVmJQzpO?E%<2&{S>fZWYZAXC$fsx(Eq75cAq1~2_8o3OO?lHV
zfV{7o825=}bW<Tj8tR{^JMR+&u5ubX@qR59nq-7hT5EiS6j96Pd_CDNWWzYSzk8*q
zAD7IGsi6Hx=1Ji>Mxr9@!cy~F_TlN7i~KLV|JdMZy5)yFn(iW2=gh4v;FWQXVbr5V
z4p58+5`REh-%uJNpujFP{U|SG$1${Asyf-%@<booX7Tr=I6SqoBtM>S*zehw^>x>}
z(xIncwro&8F^us$Y_{&Cj?C+8<JG)ar(1}09^5QKu!67aD4Igv`KIWqJtlz}PKVH#
z#*cFSJ6#?Khq~+W)Gdr^z<+(=5+?MSGTG^Z;DFQ=7hHZW$Y;hm(uGw*t<>bxR$u!y
zy3sXVZALnT5aaZT9_<wW{6c}TEoRCN0si{0N0_86b~jY{8Wb^B?nz$tDhIkr4|Wqm
zkfJg38>!+g38V2z=LOf)RlV)LuO)avJXL`+6wb|k9o=x4LL8=zndie@D>pvKrz2e_
zU@Vcc_R_gNfAdnr)@JEPl2!vXhVveqb9nj8;sW>A=p$Bu;1N#aZ}GL(_#a->a1J!v
zUmtp9EJQ%}X9@|lt3O=F21x91z$akFZNnKZRW3i0dZ+$^RHDu@9E*93%Dd`0;n}~h
zh{k-ce0i-v7r37|PhYc`=(kXqcK`~eQ#NH99G`z9GlUTOpg5yGaOx?RKf9VhrPp8}
zMvn*T8^!ZdltTz^iRQt>S8_Jt<72-+L1tVDk578|Of(%ORYL%wqEJrSf@@R=xdRjm
z+>a@f`qKq9dl7Y8alR;T{cQ6RzW|pKvi0f=S%~ejCSlZu5MPtt?^DKpjp|ZoO1;w3
ztCn1N<a_D%23fvLR3}GW^}6;UPsOTXW<l?AD!AdI6HoCWf6KwH=ZFs4?7Eo7z8eWN
z=gbWBJL^~=H?ECPF8MZGT5T%sdAHo3FgkFOVv3s8<{u8!(L^q57UUZ=n_Vu5Y!^_g
z^+(OZ34&DA>{GeE?)}-*4RU=;$=^1_ktNdUe`ssJSJ!uriP6?&;QM}|t|8}F97A)~
zVvnj88aMc-%cz&UJA^}dFs4v&dosx=ETeC$<tM$)ZOrICII(}2VPp2$u)ns$`*Hb=
z?8QnkuP@yr3{dXRm|2zSaHtQUYMGSWo*Qzb(#9jI2@-$8O=#v9;8YaeU01dLO_}F-
zl_1{<)!MG_-fEU>(|+b<b9%mn;)DU}W7S%NuPK8EAp{|2BxBEyO~VzdlPP_j%4Fk~
zJ0mWMeg><9MSkmTt|rQ1ILXmCQikq&Yy3Je>Dh%vsS3)&#m~CWP;i83#*>eGWfA&+
zY=R`uEButF1+MW!6RIYqB_^Hn@Pg_)pUiJaTgi;W^{wOYo1WXr;7j{j<1`;nX|g<?
zsmxxC59A-#lNo=kjr#{#4xRAC>z-=z5r4un`d#R>)V>qo`w%(KL82DFZm}wr+fQ1)
zr)|Px?AsS8_E@xtRiIw>j9X;Fx5APBWix?R78>~#M>yhj<{N7;$7cGSDQm2oM=Q&d
zAXFU(2hD(D$&52y_WMeEVWtf<tk+2!t7#fT8+9>?E<<xnLnI@v^`|$&a0DpFed;dZ
zoAeFRaA4Z~Ds9!*FY70FRp0wk3N!;x7xo(guTn2sPiYTvlxt=KXf5)bJM<@PJpI$t
z)yN*WLo^7Vb5U@Of?uqJ>rH3BwEHm;R}5x&G-@}Vy7;dBR7aZlbh*n70@tg)sx8!Y
zH0RVGG$<VIOj`fx)8dqN4)WX#7_QOX_bi*ui0WMFL7`0lR`lqG%qK3s`wgI#TjkW{
zN%C7|av)EYL}B?^88Wnk9*PgRj>_Y1WO1>bnOD@TDAZGkf!K7*8B$d@NNSi-NpNF3
zjPM=jt31kQ#vF(ZS%Vu<Avu6+!N+<+U<(j?a8Wh)z%!}$%dkm9j#CMh3V%hGmY5>K
zVa@A1tgWtqbC8>^w6^k;dyXo3HLmiAC9XpZYNiZCGf+{>!LIkwGnKj0Gfjf2#S7#R
z_wOTJ^9ru7K9M}FGfPcr10Tk*yvU?v4QdvM^~|ftW}t!UweS-dgLr8Usd{gIK1CF2
z2F4v-nt2>;i;;EF38!$1IHdYx=k^|ZO%KCZFk|gI8oW?c&fO(GBP~_9s_i!(4qmhI
zDcvFI#-|4fz(-H5Da(*b7r2vLLYbl<y_hb;@h9G1dHOVDN_x67*B1@|$Qf07rbx9|
zf}rAs7}zkPpgy>Lw|j{Oe-;lOjH;9WDDrcJLxnyTVFh;@$DCHhTOP70j6^R!vKjFG
zsZ4gC6?Y1bkzbj2Z}dtEf<p>xZ&yaJ=JN!1Xjr@|WE#s+tQY#Q_p)Hs5LB!*nq_n}
zZ+pD)vvB!lqc9$NKH_+cOxXf?@<BIMOSn0wNTbn!HZSz?6848KZR^te8>V6+i0TRF
zcB7q`+0iZ81tVpZKXr$N71s{y?iWvu94=d5WzewV-sj@Iky8N^gK39LUOg!&OlOHL
z@%^@@I9@R}#UzhgtM01&+T-@|rb#`nMRdiVz%ihoZzS04^c}aubUSKqmrzt;jD?d^
z!XXvR5n}3_A1`eB+0B=RZ5~DY%?$874(=+lDO9?Vb0Sv=2>3Uqz!kD-pPld;zyy8e
zd}<3il$Y{!x~XJx)zdA1x_6%U0sTiD?1wXHt7nAD+c-0wl_Ejp4ihOK?8D~uO*EfC
zBGV|-fKs8in*X3?!%C>Wh1qU@n3gesPVcfhS*|&&&SgeX>CJ7fE`=h(jIq>SA}h39
z__7_IA=0{`{HsLZ$FUH}s6{L^4#Z+APNgB~iWYV?-_0qvN+xXdYRqH(pgm3B3|4oT
z^_4Cu`T>;LC{pGh&S@UZU`I>2j%6@VU<_b@MwdFa1{a{F!L(n?9tPc2QeXf(X1$tY
zn5SypbY(|g7aS=O$BraFx1$u+)0?ssU-$SVnZtdBZr?}4&H#q8-&&YV2~0#^vdeUA
zPpsDPAmr7MdY3qd=`})`wEEJo;@OJv2>~f$sF>f+ANs2i8~;o%-WzgXF^r{p=D?|s
z1P6X>S!CsuLH~|tqgQzhsV)Jpeg>{Q%g%uoEEMyZ262wxWyE}%6W(xjpBlhf`V5s0
z@KTnb^cO^Hc0b)b@IVJmi#{<XmxXctrik^<83opj;meOhqott8Yr_GgvAsa=o^Z5=
zsR>LvOIS&NTsh&1#}^k?a9}DFjTR;Y6S<Q%?w7%(dW}4#N{ZBPMIWsnX%w_M)B``T
zqPUJ{FmQnUurzxTzx~2X8V-1l<nKN$ATz%D`(>t02Sc&cTnHrPsi&E45O}=S%0@Dk
z=I*)oE>TM$cek1ZI%QvS9PVn|+caU23iatpeWeo)^TQ1x70<_ay2U@NTKc^u7~1N^
z)JsJ_aKXd3?&^na&jKVGpBYZ4RJ8q2M{++7+CtvkdcliT^)?e}S$e_Da0asBDb(yc
zygM%+45}OR4STA}vmn2}Qv@e|hWe43D3qO253m!k#-f_^wWdY%HSXgY*|i;s|BtEf
zfNDBx-o}ckfXJ#8DGG?vk={iR4Jarj5UN6`q4y$16hY}ARjM?Bgc1y)gVLfjX+b)I
z0wL0*_wOd)|NVYv_sH&^<+izVXP$XxrkZuY2M`U)Mj0t#2Ea&tp@KJu0+aG_-p<cH
z2EY|4wp(i*0z`G7aRx&|f@ncte8jg+rDlplSEbDo$d^kFJMrl8a_RvZvC%MPJUd7-
zD^Al}5#SHxZa#e17ZjQhpiH&I{U(ijO%5}sWMmRFw^KqkQyA^`4*=lepfXJ6pQ(bl
z4XqO%yPHzK`*x<O-fP7y>o={@NLWZn$9(NX*B}f8ds--U23T><0D=t7Q9#y7PR!xi
z<v07Le^gnH_*t+jCBOA?fUy~jQKO68^h*K#!Tn&m-hj+>w|&Tqn@f2T{;nsKW3mwZ
zb$#-)neF6$>VUtSat)}g4vK+yX*h*&uYpqdqas7=?>UaxqSM(7Ob>$}pzAi1u#t83
zi}>(Oyr8$=?y$o4Q^S!;xs@pyMPQ2fgd>^D4&x*cBTS>Ij(h_FByCBc<(<qCwxdQf
z1H9?gAq$n}X0}@opA}V=jJ@IB8K$FiI%MKMe8_>TnzciQgDUt4<=w1X_J6$qYaCRP
z4+*JHbMdozd$>>Pm8B|pCE0@)GMz15d;JN`{$79RNjisZ1~21ufU;aa1#?(k4@6p4
z_s;2l#M@z8;&)!ot>1}PGBS1p*(2c}t0gCfpbBYK<#PIU^R^PwIU7^=02SH&+?9L4
z6l#hcIJEHz*_P+)J$SP8j1VG23kMhJ(o;0OLFV*i*gDx~Ou60csY$#L4kE*XK1-dR
zgD*WI;uq`ZAJ)t_CV+d@m|(Bm<jFxiJh|4<LTg&M6t3`#DI-a`?APu;>k0{x>Fk)}
zKo8!egoVE7cR~2D<r$M13W;y%Pr+O|1Cgzy9zwmH1H+&}k&FpQY)K;*7;sa5?xmf;
zCi@><eVpcYP0C7mWf%kLxzV$(jAD|7N4TZujhA0Jz5Ttp11HsLkr}<#-u@B^1U(Z7
zdOSiJcFHi5(QPq*pnTix%AJu)xFX&MFjm{%2eA{k$7*zHgaD2cKuDUGOS$O-QkLD6
zck6*sI3Q~&b6udhBZJrlDhn`>@*6Y?Jr~qCCS^>QlbyyBT>RX7Uy>$55s}M4BWnv{
z-M-jVM&E<Y=8lSP<_NouhaN3}^|M()96AA`aeL8U6~HTGK%U8IGwL7MbfY46;wTmw
z$!{|BC%jIfknHtL>&K>l`LBlmdY$<7v^bajg%fv?JHJxcBHq8~m*|uIEdvs#Qob-8
zUU+*wTe#e`_yS<sIIO;2-puJJZSCp=Y)gQ)o-Wjr$0$M-pjOJGK6CwmY&JOrs9v2b
zZI|G_LtB79W3v6nGQ4hp1oI2|fR}UaMabrMh$l$SaFd!bGs;bJ83Ps~DA@p&eGY7~
zg&!Pb==aByFm9HzD6urZ;m!FZ5;Lpxw}=Q3D}$-LS|EYLPs^?XjN=q6lMw?vFQ8jx
zI{9^@Fr%u=pgF;$=jYlhlL+I#cb<=NkIQ-60z>xj;_|uPPzaj(CRK0$cDO1<bK8gR
z38n*NYbYe*>qy%i&})s1UQ2}FckJav%6Dh!bJ!6c<%)PH?>*joe=JbmVwt@0O5oLb
z)Nm;gwO_J58Bi#(?ex3F7`L)s238#&vRNirCAj}5bYU2~Z}{_g7}0<nM6&>N7U!+z
zkP2w&zKEtO^Cco#?rBgAOtvg#pdrRNVz)reW_M(GdJmuYaEEH%_ixgGb`>LvzAG}v
z1WT_eXBiIK7?$Z0G>O81oK+IaEy9hfQg-f?xIRhC%c@G}w3rfn2pA^H1_8!d^`(t5
zw8!Wk2Qb_)gW~A1tBqqn0HTnfZ^u!{#w+?kLUKfKoMi@YDI2rJh=7lh36nzSNLghZ
zV#LKH;!X0V#EAgjQAo&+{pdM#-HaL7?Xs0K=*y@5e|*0bdh7mLSFybukh*ZBxks7-
zpmvoJ!RY+t8I}p{_(-xSsanfjmzTS4e31@R@&LE5JVcU>PM2eHvU32GN#LhjzAs~!
zlc#93>FH9GMH-U=+UY4z*aQdRAVa@gmv{a8r0T6zmt2`0Qlfze6UL;XU##su@QdR2
zmmC-KvBi(0Fh6ir{cI}!LSgC#x)*>MA<F-dgq4e5uLGhbY=2rW@^-nRDgg`!GId;}
zW)LC_lOHy-lh@eN{;gEYwo(=l`1i1>`M!VJL6Cm98ED9IN_66!pSNGZ_pshEV3s-6
zN`x4Y6<cn>ay=RxC&=#&3jy#osMKM3Us=N5c)|ZiF(}r-cYWVnSr*_4!MAd@MRL@5
za$N5#*L<AJ<fU6)mZCij2?#eo*OYz|<(B2_%X<&$Vd1Y|1;SvGa3A*2TLap~cod~Y
zp#sCfXtlYSzhK88#@)6()7QtiGftqPqE5`piWBTtVPQ8Z-~mdB{0!M#2N`ASf%#hu
zCe_8Y7wAB{5eJYJynOXQYGo<;qkX|1sz%Rp9<Ql-0RESBV2h}!683K8biC{7f7b)5
zkNAVD5%;cVT>Vz9XwE00jt0_+I2QZ{>ATCZ$c=IRQ4vRX6g`2HYEWtJ)5iJ}8lzG}
zkV0ZgNC1-RH8bejRk`<P?o+AC2%riV*=<B(Qk%DLf{@?SQqjvG1hh9Ave^;hss4wy
z7+ROTq?2>i#up0wFI9p}p5&jF?00z34+<n5#(72jzOWD)0<y^J!OQR_gz~y6$93Q%
z0Z1q_WD|7A0;@f^J_#ld(38=gc1N<(3IY6As&bbDJK)lLiLJWn`(!Orp#?BiBMGt$
zqFE!(dq|nid7sYI8C_nSxBVVG965m9S=LclPM@swUzC{Q9$TFi4F<ME$!#Vobnd%4
zkh|$`d7$&N-_&Z&Mp^bUoj>SuQNJ?Y>W&`>G6;QvUwm}Sck$u=wUsqH&>iaS8e3?H
zzsz{%!%<(lW!fo=j5XqWR=}A9t`)S>v0$CEBcFEU3E&SFvlGYi7yWLzHly>^k>5)1
ziqQ0e3S5=_XVGb&qStf}hNqU}H{>q?3!81b!=ovDtUHht>dgBLKqT^mzBmX&o-Hwc
z1JI-p@V-;JbOM7d{k@ByQ|DHk%z>|c=N3N1({aX2p=_n9;11$#$Ra2o6EwIRY5^J-
zg5G3B$-(ssGtbS^rjjxZK$|WtRZ#SzK)#op+}z%-2lkQ&CTQzAWj!;nfl4=s!qiG^
zI?ulStvQaHn>qW40k8o)v`+$G3@}9j5xLU$6Z;(V*nVm0??^-axJXi*AvaHC>sD{9
zX4M6{gAa3=6^zPzpF=iF8&8Y4bS{R6fGoS>?05Sh+Jp+(l%aE~+x(O|f3Whd2b37l
zjo72Lx}!>X4nRR=?Y;hk)+c)hW!3|rG>wdjhDuXTh+brf%#=mppElE7OYL=Px&sBD
z4VD{Uy?%J?Z@4cI(n_q|U)<BXa8BrWl(MSqG`l|f<{W&Bk2m?oy>zZEI6-P>c7|^`
zI}7<MIJYxz6(^l^ooieD@+}yQv-zY@B8>e`6z*inkLHE;*Hoi&spC6YmQ~{gMRSC!
z_31N7ltxDVI*aPTcL#RobgQ}<R$ePARkb9BvRI@uz(+|(4pdQ%Q`w7X#TaCay;NOx
zTV>2l-Rh4(B!KdPW`$k8tjpDU_Mu7+{$5m^3N(_wZH!(<z;>R{zTDn0MRE=Tyl&&!
z3$y)iSsS6uG$`-iT})W@M$B*d>Dr?BA={VC+I_h8;<9Ca$+ftd`tQ{DEsD2vh!%~m
zu|m425{+V3RS@%^4E=vSgDl3b4#dctp{V7gZNuI`ymPV-9`$&<;#)_>_-XYRyp~lb
zPnd$@+DLrcF&Ct?3tv{3f>Vc!iL`tgJ3@9+6peo%+ll_Jg8v@E8}Ou{9|XLq&Z8}6
z_;GWi9+y;W_pCd|AN%<s941IvSzzgJuGZlm*WMeuP#Cvr4k>2fY??#k<@pJhVisH$
z$HvDLm|MhKX0LGWHAZ<Un4|W-Nm9~Jmc<xz6QuA%SQtWK^i~rO1V2=OS<k``4HVMB
zYp<ebBN@!<5K*6D2#uh<2SZo?zn#u{@lqUp8q=}Wcqi*(XnBiqKD}p(xj5%nRSsSu
zMxAlXIk&=F_c+G(n!9th&Z1kg=s+h(h5Yp)TyluFzrrD1Bx4kuPfOx3*XU$*H6I}u
zg^_6-!S8ST>1j4g5U2N?9?F@uM?$s<QG<(U{C=KO1a}x_-2sVVD}8pKp4NYtlC$<f
zPf*moG3zi!j|oyq67cCgaO2JpeR{1h?a~GKBkmTL=n}nbw+G$htRaIJ-6sIYM~G3(
zQh_d#srdt4sy@UkUMe}Mn7xqGJcVBP?C&uqSewqG-|X>2nuC={XTU^i=ln}?{00oc
z#?aidKCC*Pw?2jP2sia<gokbULFyTA?+9}3%`yawID?)=T(ydtx1O%s#hkV=Co>ZO
z(S=2VS@P*M#8Ut&%cm*z0OXkIO)C&ez2C4?HyZaI%2}D#`gJhF<>FK8L2(6_>ne49
z1{{d|X=*fxH<Fe$-@*tS;fXqhtIOgC^#ZNJVqwDLw+I`Tc1K3|32X)kpPkpTCy*F_
zZUR=x;-EqDs1F_mUqMwYljw;Xu&%2JTC~8UjA)ELppnK8M&WGIuTY&3V#og8o+w-2
z#(c6&;!s<;0`K75ae$+=YtwbFl)qJCq$dsE9_w@()_ZItJUhI9IWAx)hGWcPJWRMa
zZq*O6959$3w|W{eFPomb)fN*N!Q)S?pd>roATjK^1TzC;NwipMNV$={^t@rR*Skq@
zDKY93PPxmg$F!PBtB1?YpF^{@J)BRU+!lYpfvB9Ou1gyq(n|FukM0>Jf=fPEvG^qf
z>m@u<2mTnmV%FZqxmTI3-SDSAc69@VQLJyQe1@v!B%C`|7Qt-R8Rj8MnmD)}lTHQ#
z;in&>Ri#~?e-)|yk4+;=Y%bfeQ0&XyGCn?{PG0n-#E|5bByY!_hB9^d`wjj)Q+-He
zp07{JX=^yYF43-6_wVn_FfyDy-7NVMshHW{Pv_v0)8_Q>o$&O%n0<ZC9ZWeB3Nvnj
z8f9Aicw;*D+J*uGd1z+=YF!jFwdl(kJ-0Te#cN&5V(4-jUtOZ&l^0)-CD-DRzw};w
z{diIoUH7w-xXO>c+OK)N0PD9?X8yYLMAB^_EY1$$$hZJ{o=%63l~ZJ=A;V)zmg_I?
zv5iqa!rwDkFP8@2jYvlyIgyqTE`riono;@d1ULN9ll_R?j-05S$Z$6jfts`%$DIuX
zg6Wj2wh|?9$yc>F5P&aBefGjL_gAN&UQjR*DQ?Q9>*2AdHJ3l`&)zrrpo&_Mihh~~
zCtZ)6&rC2b!WVp6-Om=xi&nCDubA47-#}p^A9#Q9*kD$v8C?APNTE_8BL8WaS6X&`
zoGa(k^#QJ0yuOc9O;&OR;@U=Di!@oLfvs|^GV?R;!r8SVgEdzCut3<1<~Ly`*o=za
z(|*mbXCEqyWkjzv)V$W?vc5Iw&w;qbj`QAUBAdMw6HFTt-!@cl+g$az9Ww?GV}#ss
z@~=JU8;p-zrOf&PRR|Nts%?j4(;qAA%BM5+^#-G#Rm?PoOfnxa9)(QR<spp~w^ym}
z_oNZ5;IhHlqbhiwaHI%7U&*&J@LY6kJq*7WT#}VYM>>Vba3m&xop{-Qoh;k~6)%zX
zH>}%|>L5LWkyZxv*9jJu-U$C4ddslNd*K&|`2mAn54NF*d3rOJzEw#^=Y#%IQtk-p
z(|dpl(1f3X;H&LHL24olqIvytw}M|c^@-v#X8Ve7cC<}@sC?Hw^<TVkRM0YO{J4-t
zr$&_*jUnj;E>OPa5Z3f;qA`JcA~6Z!!U|jjIUBK@{cPcBG-mxuFowmfeVB7EyT%Tw
zV2!H1jkH^BV{pmJj#ND>ZCahL60#P(9AQ&XBl~NgyG4pjSp)|{*Pn{)5SoOPS}8Aj
z_eu~?m{fjDNR!Jo8~k+v?=89U7vCRnL$*CCs)0ue`t)`|XN0LC|8OsR>{Wm+o;pNx
zCxq7>67fR{QLE={PCX3IPM2ue%@nrC=v@vM&I=dzze$+ytl}bEL17Nequgj;3QtRm
zCF^yaY_B5=ZE?8UW348JG6BN9V`W9E+uMnUE*DTHSBXt0Ixdi8uq8`vJKbc9Pwrs5
zKNDHZY_Ik-N9C6jNv#!NWQ!X(`!0Wq76W-m@r81}rvWpXB&z{RkA7)G5c6FH-HG~7
zcpbNEV^~F7c3%rmo7W<{7*=kZwL7SKTlvdb9OUs4xYw}~Pg)YEV&wS<<pF~RF?AiJ
zbb$L}Zu_V}lUTO0t8rvc8nD4Tzj9kYkix$SsSHQD0YDS>Bl@TG{je8rK1y>_Dqp#Y
zONM}?T~yrLxP=XVMZgZedv8nA?IyV=c7WFfndi|zZ@1Jq9ZW}Ln<T7i#eA?isgc0B
zSCP#o(P9|qnx4JGWi(_`S5n{pHroy2v2buzPOr8s|82HO>^qn8K@r6K;>OO86D^5i
z5Px2R*Re7WPTy+Gy4p*7%Xmh!1<k2S+P}e5SIHU-zy!SVTR3yyU!T>s(;Q?`G7fso
zp1$^0Lbn?sP%Kx8%A`|n&?#7Bvi`BK_?;xC$j$}754vhpi|N`czYBWQD=@LD{F1BQ
z3;K#(QkdIIKS=WYlr>Ba$3qy^9HzT5HR7>^+DOdST|97|^iy*mpo?8C-np8E$Y3uN
zzQId45mR^Vsk?d)5rs)-hBNzy5gx_(2|mS-;)irSDkIz=UI;tCGgFmMdQ2dyr`!GR
z_N3EPkxgYV&Y10=|JGVOe4%c<d&^hS@iTtYZmP}I=lWkc35JQbEoUdv7ZvoK0rpdg
zWhBD$SnQL0K`RbKPqjb)I~&dN>el$<B>BnbJv2fl?tC{NX>FENaNb9xvh_AcR@Xcy
zGOt>e_-6=PL;P<Mnva#8iy3>0+Uv_0PKruU_5Q-JBC-=+<}2K?D+e)&a+QWhWAzc0
z<wq)QVx`qYdc3A)TfMZP%WrS4f#_QHqHF}m6!{4#eWm0j!yKDO{_cxjqX(<&H-m%Z
ze?&F$D1xhGvxc{{;k*4+K_o&KkZZBuuI*&z6<t2fbo2m2K<L7eR%ELYNx4!j_unN~
zWw-B*dX18*I|ebvmu@e{#uz^f^J>0O&H>rbLo71(R8_ed_6R=M28xg^YE>S>R*E<N
zv&wGvP`ihhYmXN)p1(A{=7Yj4vQ2uDkrIl4%Ad4WmsMY;(;9p9z9@TdwkD3o(<*51
z7VpotZ5QsOMGN<&?}sK3ikNrgJ=l~~j4ve$^aFhyqdj{Kr%y$NPCH`5KZ-q%t~xDu
zSq#;RwkHTJ4zK&a{yiVl!UC6zdbc{dqj0;$_Cb|yj~8S(#+{RklO@horALjEpAFp!
zEWVj6MsisqAgVi)!VdJtdQ%31BUg!hGqI+f69}V|uthIVEzPO%gVJE}GideblIi}E
zTqs4>Z@=7@ERmY5*(WU^#9H9C_rn=j`y~04Y|jovhOTtD3!@MvQ49X@tAZ$uj}jAb
zm`R_IhiP55wLLBzY4MtNs^tSJ$^m6>j*1ihd5~tBbLqj*2!FV%?W~UDQ^2wlVzh5t
zyDo+n;M#V8Q=dc29}z{9TUg;3yb6uU*>g5ZJ(2o|1J|AthdWD=YxMO>*dUMgY2!R*
zny~hzXhVM-R<lgCDe#0Yrb?%|QhWWhB0m2jv__w7kLcH87T<LIjreds3GNSaovU>E
zsSf0c$|{i#(_9b@lToc&`%AElJ3^${zk;PWBr=qPv*v=N$BWL={S{Q*rh`b(n;sdU
zRk0glCB|(Dl*W0fu_}1aKA+iYeOmj**rE4-K`xiTGbbHqWCL6q5ju!V4!iej-=*3P
z&OcDIlUT-$PV-7|4^?6$Gt}CvN+r+&7ZLM?+;Xqk@k1VJ-Wvf#PyL>*2Qk7o_zCI|
z4@i|#Pe+V>jH{-mil)A%P7i*+>k?!`bvrEkz#ipm8&yi0GWbG2^5djmso;})nJg_N
z(|^(GR)!b)HNVB`M6wfIsm>u|%z$w`s<Am-Jd-4>lBE2wdS_?sd1bW<zMP%t4Q1P&
z%b8x)(mh*cFD6&>g7mO9gk8tIeE~ykaQeC*kdYto?-(|fL=A#IGsC(Z=EWS?+Bi=e
zk7qY!jXB0K*Zk!@C5qNZm|IAFmAP;G01l)DJZkn)VpV$Sv1(tfL~!~y-3D=U!>uE8
zqG;6~E!}h8%ViPNmI0&{2sK_07Cd^;DH6r^+n7vPK+a;Zfm~HILVeub*(+o1ktW}y
z+^W%<*kt7{NSD$#lJOZm^-jq9=L?+}WCN!*5!Kae1uxxiPlSBheo<uyM{TYv+!vu@
zaZ~@#YtH7G<VzCX7{G|t8y7l~rKjq@RQFZz-|i4)>M?mhJS(KC)8<Q7U01Ltq25;T
zxa{!J$?vVN_Fu`l0LEMF6O>dqGP)TgAVe*eMT^me&Vtvo!}{>C{!XqIM_LVqMbrVh
zW#F_e+-+Eoi%<@uL9c%Yx2PODXSWdlTK>jr>uXNv*HZ8cUDVHq2<Sz*@sHYQ*2XN_
zr!xZ{7CC)_g5ndIk4654A~=88)O_t*?x%f)_r=}cW9hp??41L}%Sg9#e#ytlz>3y(
z8eHJ&k^tU-Zs!e=<yq5Huz`fwj$vk_w517$y@;Ll;+ex(WdNu$MgFy5<70Od#LA&V
zgN3WLb@v4@OfUzjv?qMh)xnW_KY2DZfA|4H5VH&*c-+%7b_-}14>+bhhk9xudb@h(
zjbDnM^(mO32P4>Orq6raZw))+hSz6Ezq%ykGLiYl(ID+_?pDfeI-|ai=0v=dMRx<Z
zNG@sKN3NHB`ke08OXZ@A;Yh6x9?4R&(<?U96rcw$1({KB14VqtsHSeZk~t_I0nEBJ
zgmstiXn8eRLS;j#|2{QqfAnkxiP2cwPINlMOlD$&LEB>t@4Otw+y(a?uP-+Tq5%f!
z%j|{q(HJFPOJR_cEV(M2sqjon*c9B=iuAnhQy{saI6E2JhN&t)dx_;Mr?h<T)$h1u
zb!EAYQ(|PnY52f+6THh*r`QD1EO?gnM5kHNBi%}oNCh;Ha0bM+uuBCWDsxXmXuW*x
zcU`MEHIP({bU3n%NA%^x7~mc0`#Ld`jwOxIVMthFM*w=_?dj3`r{;r6d-b$ina1ST
z;~T&7(p$r(j;<Yrg?{G9SPReEIBpx4lVZ7D9=p<y)6ZYS<Ep-zr{$^q$f+jD(ob=8
z%ZJ8Ik*PnDKhL4<|6Hvr{<z5uGo#CytYb$UnYrh)Qv3XCKE5xR<%|%f`=`}}uKdlJ
zXuGg63`S7l#W965q)NpT5OTfO0gDsmeg=J7?5EuYcq&t75c1kFU5YFnx4<Nj(?jbU
zrjR{k1rPUL;DrWm5J+n%d2pGi4I%vmzQhw-QJ}iQ)_x!|(*BRQL%LLXwz6$GEelo*
zjc*)Y2uxlA`yF>YSzq|rH7_0VafBmD$6*mGoHK>GU?TyL`f>RitYWH1{=404C()iw
zziqE}Nx#EED4q)ZQUjzKM3AZpacBGY^ObcJH`*6vmb__j9Kpf_46r3o<n7!kvkIV?
zFk07$?wKEWrS#oiJozZ42t9zDSm|WQqF3idNX(C38sF!pvvYv|ZX&aNqD)W7@5c+v
zkwdj$xDr8=voy7~m$uKD7?0Q+v2LwKyP6OaZ<Kz-Hf9c)aTIcb8}90Dx9)$XOn|RM
zpEhav>wN)Y$KrkpCqZE^_R1WlOG>IB^VqpoeWEoy{GLU1Z8aOQ4I<};Vb3p3mwY%#
zm4D*{9i!R1RZLm}D2POs^*f5d`H7wKB%`um8W}l{X+VKPIh+luzd02ewbrLvu;}#Y
zoWEK|Jzdu&{HuXScc`+ithWb^9u#T#)^0topXt?@`)#Q)i(T}4$XtDIZ?mxLqWbC6
z6Xm6<^Ya6tPko!gL*=fA&NM^>Y18i<gZf8d9*`o)p!D#_A#?(MWGFMdC5fxj`M6qB
zlv-QxR-JhyzkiSx-NG>{fxZlkvMQ>Z&sM*HJ&yq={wo=qJ~-C9Q@{DiZ_R-%zR%Nt
zw}sZPH|qWwzK8-r9FtK4JMj$?uoiNoF__3Vi|KHu@85RlMBDyx0ldGT0o6-MlXwv7
z(Q#yBP$W79IXn6oD1;;JnF~2nQ(me-{VV1&GRq86MEXcuo3<@EPz+$y_JOI`?7nGE
zt_w(NBB;r5?XCJ(*GugsY{-f19y@+68x<{9<*A5wvt%3W@cZybG>|*x1{52I(LSP$
z9PU0CydmssuqfJ?R~rs2nbF|TT}8Z>%LRD&&KDiZpM`AOqg}Hr8iu|3aqr}Ji%oKD
zZ(kk)N6^cvB8ZFzQ{XQ8CAZ<QxHk6qJp0vfWY3%5P!6JfX#tp;wk%Us%`k3`3SfJh
zrKN(jGMVIe@G{k*a}EX<G6T2syIQ6GfN>7?PJ50W$OU#=e~D~i&}b;oFC>oi2g)Bb
zi9y1b78XbL5;>V*c}5Q+#FqA3#s6kqJzKI=Am*^7a2ZykAX23WB`o;{fkv124eE=v
zYuu_CobQrN8@`z4ox>u5U__Ck$29ivI3^Ub*0Y5QU6eb;^8U94q@4;4wSjt#FaC**
zG>-qz4`LaJ)ViJ>(|>y8;D^P+_x|qZXI6`+5%6$+kSJLN(hExs&co1|ldt^g)uY51
z5%Y*GL-OkW_x`vz=g=tPpI7F&8I4#6ITpPK0O1My6(fwX?%0Nm&tFx=813`cC0saZ
z7H=E){B;}?-s1d(qmD_G7?V*Gd!dIyWofT3(4v>AYFONl)^hQGzu)ZaZz#hePIE-W
z0n5PZOzr#?*w%(g>FkI;>$6Ba%DIkWD(*Q9KNJy;R79mS6qn?&&+ab-B5P|EC_a71
z@vwAo{@O2-a0xyM1DHk$JL0HV)e#$qWKm4LUf%uLt8Rx<6tKqzBJ-wi;+Q;=j~9v+
z8s2Hx*u|%D-*M%OwHgx`m>U`^40Df?)XozsmFdziY?W9EE-D_#kO;U8Kp8%?r@uv*
zN?3$tYzM)Z2{xL$1m_#7F)HZ!kL<LCZcvk57;r)shIJ!nr&9ITUTb)S)&>d6ffTtF
zdbryOXurN($(xlI=VW-Kk9}Z{5b&CISa7ESmr~kovWjtl^g3xdB^tjRqtFL(;bO&C
z<q6!5Pj!<gVIH9@@#h`jkD3jzp@*U$VER(_2gAx{$?4lS0PaN0raZEoC|=?c^k3AP
zC1ro0>#h0aBB)raiOJ>Na(3}f0}~I>$+17-FGy?nV_6hE=E;c?OdqyUcL1t!`0MX(
zn*UhbZo=1I8L|ya)`aV5KFwgf;{jgEZk*|9Sfn{92UWP$$sJh{j-(>Tqbb<?fC7YA
z_DMoTdB?S7QfrQUSPGHP)KliMPt0S5?H0wph}IfPkNQJ?X%&8r;<D6er_$@66BlaO
z>_0a=rh?}He^UdBdXWnUg4jhLX*XS^KQ5pM&2SI{!c?B2i2tok>?#Znni^$Cj99}T
zfp<FxU|3+cDK1Dok1qPbxOuetfXWHuyK=E%x8B<Z*?3c-jfa+vFK+!ZB;k?%x;AS?
zOt45xdR1w~tVe-JdP^B?v^Nv1`aF8t+tZ(7>8}L3<$B|`$ds+qOJsFUCx9gN5p1!m
zqUfWVK?A0KjYoh;0Asu$_KRl{yq&(_E6c?KC!^I<FpUnasD6_4+CGh1OPc?%&VzdF
z$iv#dty&o8E^<H0ExD3$pxC5S!r?s)S9CCj`}i4Ml-G%b(pghbTZ23D>g5PQnfkI<
z^u~{xvpR;kEDsyehuSbGLPa>p#BH!<q29Ebm;saLTY=?gQ6)Av=+?>`*S<@7IaPzi
zIAsp$WXO5T3@!Er(QzH(TG8>auX@VdXcRl~wcu>%GVw8vhv<rypQ?#>kc+1FZ6|Lu
zu(hq=!UwzYk6PYiU#iDA#ua{6gdU!81F!q$b(UJMoByz03fI;4^|bx5;;|`fkJ#)S
z_`|ka+}gcLQEM^ueTF8GZ>|OC6k+>@f{Q)4xdcFAs~fS&$%%IRJ5pjZENhA9;>LN;
z5OjbCaqju~aO9l>8(;?6f+SqI6xqqsuu*!x7=VzHPP9Snaa6zlPTI4>oN3Z|FDTI(
z+MFv@)x1HsVmMs^Scs|6$1C8KZHy`7UNHPy1zi-tCBMDrYC32K2g=Esg&DRn^IJ_v
zqjxS<0y(F#zwms*%$v&l#L|ao%(KD4nfL;?xI#CL;exhNhu*{Y^k-nb&cFMV`weqf
zYH9!B?E-TwOHptO1l#=g*WB|{Ftcd<&qY;K(R0F5OSK9If)}UCLHy;0D-|EsAGLS}
zsdAGm%uyDdy=HVO`Pj+AZS`D7W|b>%hH*S@_TKKyF#66#-b)U?DwhtCJBt6Q#&g$i
zRBQ>vhHVMB7!tfwbyFSxys<@-7C0@6h9dUF(Ds*jcyy>39%%wA><TCWsKt9L^}IF1
z4cyt)NEQ5P>?xzU-O7tYI40m}gI%LAJcNS4Pft((YWu91l%uT8oA2pTm5?z8&AjAo
zWboEQ&w0J#(TjfX+QM<Q+pKFR0fgi%P(gKN>Y0oynzeoK$FlfLn&Bg_WH*3Lyx8x5
z4jM9jI1K}-lGV*)%&=Ht4>t4RA$!ukJ?;Mgl<)<2!ldo(Y3ty&ONHoE8$J+a!qg6&
zwJyq;0R%YlQ&3|0R8BfArx_he(C59ZgooNSHM6VI5@>OYs&J&*n~(9a!bVu89^BA1
z$I}@iOfcqnA1gTOl^C$9(|&(DTp6F_|F-YM*_;Wr5y8Ak5NgEO>maYPhTM9GDCpDA
z%*Nz2O}tFGzHG(&RHq29bCdF5-!LC!l2ebH6NQaa?+#MBfR`dIE0*^bquFYtv8UNe
zoq3*Iv<fuCA5-NqJ}GV@co94ZMSWHh;~*+Ud63>Ui&@;QkY8=M*H6TkrMtaE=2vWZ
z40QOBD{)W5{w&>l-O8}i7~|Je*`|-EcQY>k=Rag!{I9`h)Mda;G<NCa8mBc@`-wE2
zoxvCeqr3dDxDy5*o}RN*c02p1yn!C+dd$FjBAoAo+2E@%Yq1yh(*V}U20AMIz<3Dg
zt8n(eZ4NR;8oTcYZt51yWzpKtvamgW0P1x<y<mS+OCO}YAwfw$k9xvZWSyfrNX~xC
zNk~8Vd}VEAeD8Uev;EQl2~cA7LzB&p_OjB`FfI^#>9<L_Y1b$H%=>@y=J+Z8wr=;2
zQ|;PtI!Z>pSk=ZMc$gmw?zxw*K@i%~+szvA<Uo}e=x4xf{rSK%N{lspc2@JQ?%(k7
zu4w^9JY>%G@j(WvpBkN}4-s=1_jHBCtX3AA*o`Id>X1OGP{xK6r=JvpB+PT$^4~bY
z8x+o7I84J2QP{Dh^~?niKD-?iz4|~F?}k8wcEmrQ16YI!29IL6bv3JqEn@f9hnJw5
zV;5|SJUxlO$^}O%#XW#=8r9D{RQnUBIbo)`tfDklMgRKwoa(caSxVR{kR2Bkt|8qu
zdJHR93dMqJ7Itj!qqHJE<cSO0C#&mpQ(J~oKV~lTrW)wb+jAgBuE2l0kX(|6z$0W3
z{4B!0Xn0XBcGWmW^3UzoRNKX^Kp@G-R3v;&)t}xpS$`d>&IX~BpU$|pSed8qR&G*R
zA{W@5X;7}jT^OfbY@oq`cp4>^QOa;<(FOzkar-jtkAkpwAaZACX7bvzzU;i{{%SS_
zzLR9Qd&R!A-kfZ`k?djc{9j856RfC8`D-%$gISXr{Tukun|NX3Bunyfhcjo;Djd0t
ze!iax{nLBE6VSd%|2*X#kRzz>IdcuQU-|vu=NO6qXfk{jeE2ULwUf-gJAT+U6?}zD
z;5{<snSx@8{r@}<u+~?%M5s(S3I$(P_1>HM1r!S0NWEE>I(H?bw|?d|W}O}J&H7vQ
z%4<OFu+J$r{=v8~@f|}W1%*MbU_;Y()QdQW<5wtQE<mr;RVd=Ef@4ozRs@#nHATE~
zunJ!D2;mI1tBZJ8zLM@;D-rz!>@w+V2sjHI$=K45-0ZQuP#u(;y~pqnuY^N!?vxVK
z@6`~uGq6+|EBK=pgUk;zM)lV>es0Ls%Jr3IQnK)XaD~A9=1b(UaAaM^hW!;q2a0(c
z5abD@gpJ-<HB1Ocj<FXOM`P}i9i4z?@kK>^Y#m|rP1w<HQ=EPbw1|6Pev&fMF3FsG
zWV_9OMXYw>U^inQRJzIVmXkbFF%L=x@OM0m68nW7C=H=HFH@EIAxHTkHVr49SMQw0
zPRw?-hhwTfY5g6+)?K{-UprO50*pN0d_1_aUI+AMmY>6ssf{0<EbD(c!1)io?WdnR
zcChVNwsVUg!H?pe6hZ-a2bpVDB+#m3_^+xR1$z0!Mw9C7NV!jR6`lMb7OH;UI=h$S
z@<sGFPBYc!EY;@P6)h)+kFycX(D&Fy4xF?w%UKu~uaPUTlFTorwKSgCXGcV^^L|gT
zslC(}7M#LYa&x3BZd4Oop#>`XuOL~EbOaO0!yxK0?H{!oZj(}Cw|=mLc|&s)E1=k(
zpdI7q%Cc+Y%NuG$#jO4;K1B(`ZU9rVU&<Ch%P^K#k-I;Q4$RXQB}dPq<;#<@nqBVN
z2CL(%;soq~(XRNYI_A^6kZ+QL!A5Cn8G`*)!?>ymMW|)^dtUMNJo`eM?~4E!n6KX3
zNzpt6Q@}~-Dnfgg_{`chRJTI|?k%uv9Hy*-r|EGJ8vDCrT!_&<ofGxyq#<@4Z1dJR
z^lPK_Dn|HPkVkAC$cg|J^AS0e0)NhXy_!|YQ-SgVOh699bQLbva`U}aE$tbu6MOkm
z-~|M-lAwk1eNZ%r8#<dMf3O`_-dYjf-NMl2nm<(8L4JLwjh5)z<r&QURC_d7&_4<e
zXw2HD8h_Mef65kvIL*i(Tr0|OB>U1}Hy>g^ArfT}9&u!F%VW4YG%WD$J$5g%)9V5r
zmptfDW_J#~J+z$WF<Njvw|n~*WFvhXj~#J#-8^ycK8rIF4XsMy70UxiR%V-pR;dL&
zLg1I&v5Ybsz!5sp2Ej(VDFD{F>)&0XuG@I;a9Zo;h8i#sfA8IZN*t<_P%A4%sMy_X
z%PK>><iY&9aIN&FxbFq4+5vq~)`fK7z_m)!y2%0xOI+;%Of7@6HRE=0qtr&=g_Sq7
zD*_70&&qK&Ku!4|U7gn7nBW|f4Z3<ha_HvgV7Mhs7{Sd^dlkESvQ#fJ`U_pX<593i
zQgJ=rwKvh~!(MfUS}1!RSACtXBFkac*rj<(5T|$FP%1g}eDqwiHsh^nw+L=qe*T6!
z5C!pUe==|nrWK}B_Xor)&Y*Me1|lng+R4|~`fMGs5-dHV$j9`2b7x`$T%kwJ!v^*y
zuDSgU+xhneeFO@_mrGcrVJOTunasY24LcvCA_0Y_<eY&|$MF5zD;~#{1$VjM{khCr
zAQ_HH3Yd2FkLs6u@j*N1@bNKyztPJTp3MSL03o#9Y)TTCP5Sn5JAocaaGyIjm@k8t
zJ>{d=`A?;)J##&~r($*!9OWG^xUU|(-vi2a_1mzzsDr~Lo~nag_u<<NrRo44Wmcv3
zHo5Bfa*Q(J80Gi(_a&S=h84rJLYwNlYX8VsbI$$>@yu3=!W`xVG?+5qi|W7nx0m>L
zZ(o*!pvwuG9`A5B9vrsi0eaH&SL(X9S+?t)I~G!xuBwc-$H9FEEXS7t7%goW;|sl~
zB;e!wy`A8Y`KV>cKTlB3=VA8)V#&H3tpkk{=@QV4*o+p;zN+7)epAF3j%>K|HZD1>
zO5aL7U9R+pBHny_rlZB)Cu*%F3t+*~Asb!U3x0WeC^A}i8MgT(b<DRW)ZF+ML5##_
z*g4qSx@KD~aEseoH=Kf4iyvB1(i1$~If{oV2Y(4IpFB>#3fklmD+KsDK3@NG2PzOC
z7w?u7Y^Q_7(5#{^MO#d;m+>Z$=Cv>HBW7JR4CDGhRg_=YG4Wr~AL#wh4F7h3QX{zH
z^^aM+0a`2)pglSsV-c1E3EUEix4w_z^w%HeP15ZRZTM%^&0yvEA>Pe)t;6`}w?}>e
z<?G8J+%?8eSWs^R*H;DDAUhRUTEsuJPbl)Nk)gbArBaB48P@edl-oWN%q&<O?S0VE
ztm?=-#_OOkb{@SCibL8h>U@$8rY`R-P>Rs0=2|35f^<xw+u;3`?X?`ZQXSpz{}=Zg
zSfUO!gb=^QN8`w1`Le7<-)?F<q9Zd*?MU_=Ki~g*v_}+L9J)5A3~#@gC^61sC*}%1
zEyK)EQ*@dMiCpf%d0%ih1iEbFE~v4Tj77K=YAASnJ>XE5kp88IAv(b4LB2bvj=K7d
zngr;k8bZ_^MPsoks|!lvF7v>p_wI1PuhrK~Bk>7ZcEkt35Z5_$$@t14r(iE5+5f$O
zL;rmPMv$YP;QRBmS)e4>Ig}bZw^_8sghg7@_0PeX3GV?|(6!eb)h1m(v8}_>@hN`q
zq$0j_t`9dyoBcYdhHG)>h6%}W0aF>PN&}|fUY@+ajJFR5@uz>_qZa+PR~S<DT#AMd
zXA}xVnEV9@1Q0Mi+69YbpIfUr%8s+5OczJqfvC_2OYdQj<YbUsFAjygz4uZ{nes&o
zN74dFX`{<>sIq><wm<ZmlQ7c%<-+|pjX!8okqbL_=SEah-0oTiIX}V=1<_6(t95*d
zwA6reOU4i`gsQxYxc`2K23Y-7q4dfzrTE+b6p<M<2&^q8xE@Gi$x;sHR95uZK8UYT
zAl+~0D>DA#lS|*k7^m#es}g8nz13Otc+t^Pt+Rfb-8vY}p|SJi&4<qR`w9qo11}Z0
zaV6!Z5Pjz%eQ;iy82N-pO$|5u=Ae-PFv#mFq6bkgTORy*v~lMaGmQ4&xs!pxVVPAu
z*I<)@e+MG;Buxb$RWjzxQ{>^dXhf3ewa|kJ8sFYWhz_rPP&<lbo*8AR;I(Bx%%tdB
zkGV6i_)TK4K}BsYu<eTpIq8>Wn%`OufhNxCE1s+`7E-JA9}iOLUxeG~xvx_vc(kQ?
z<?8{a5Q};)=@bI)3E_`w27($zivZS}6&SvT++guA%v-FOl7{pE6(YVkzC;NFVAu*E
z`Tcw_)cerKn;&(sxi_AXm83#nD6CSyqhXS5axlwLHMLq`y!VGblj!^sxqta_7_?T2
z3xKj~hD}*5^k-LGm|;08WecUBgDLr50kxP0!hF<&@95!cV(1%F!5&mxW`_sVaO{6R
z;qS9KP_6e0QCFs$oY7vOzfzTvl~^mP!<vH#Vpp~EGU<5=^&2(DUOQBKp}Ofh(m$!C
z>;3>2=g*&IrGekCgRTEwEe00r@3;01)hj#Kr{1RvuE-uA^>!^IgDzFe54;1jl2{jJ
z)G=e*CAV9E;XU=*=+Lz<nLSEH1Va4I#o*4~iy9%KdQ?A%g!ZG-YJdu5gQ=c>_Nte_
zFIa!i!E1~EL~-k0>Wtc}@RqMUvR{YVz<?ke<#n=ZV?soxV;Fn2>*i%S+3d<QUyiTl
z>uI+nDzT+I>u3KPZZr3vJZh2TapCrxrMA3kVCMi$s@mToUAMa$Uw$@jy2hbu3m2s#
zRexLkoP8x!H>sifhbt&`!C!9N3e?zka+o##RFzh-f$wB=TL(eLUJn$_%x!b#spyuy
zw7Ae!oTdU@Cz}M9M}9EF!i5)^HpE%*`aAJd_W#~wZ&TEo;c4IeHWjD@jt7$9VxDTO
zMqCIMf4|XuT}7~V?T6!O>DD0uHIo7!y7%KZhOK+9wHAz;3IgkUE9f?W0hviAPTccK
zsl8;&Vhu;pUUM;)%kcSySFykkii@`*31O(+AE31yG-u2-vP{>pYKKXyV87dc6!yuY
z-VMS%jwP-AerM@zmiMt2VePy769i|1B-p6z-YKaKn-kh}SiW9>vjl^uS1M0utwaUS
zWfJRDr)+h8w&gzCh}BqxN|>?J&N5YmMHMpprGiHlExMcu#w%8+Kgz=-ti(um_uG>p
zCt<WpkN!OnG|?+Y$9->ggBBvw_7L3_^U;k}=t`Y#FLL0L|AUwdza)dWVLnvo8)kI(
zI8-p+X1-obuop$ocQe84?n$t9pXF8%r5?h~nG|WG+-k|)@B+N(0q7ZBYPSCm^i0xP
z3c5ji?hlJ({3D#C+wu36G4ji}+`ONEb+G9SEf;3e4$!Onj>0OfzO_syaxLI4$!;3B
z?1c%ZL7`u84L`*_)VBZj4tEd+<bz|`44-9=Y1dRM)B~*_qg~KP$$H4oDFB+kBg(%D
z_78|`Pgw_bHj~FDyF(%bxRQIBWw=K7zFW9>`%2}{7-MBq6gw-|RiJ^$DFEwn)Nv9O
zs|)&A#c3wxBTs>4Dvhoz<yvlworK9jL14;kbUrV(t^tvMSO&adGP!B()|g24S`&#o
zbGIioh9&~1KT}Vux;+y0e;A6H?#+z>AXnV+u~jjGYBr*1zv*S(4zJ>>y;ouNlNVp?
zUW0%Mn5CX8x)=5zKC5b^dFKoDi^wb;tTu|^YvQR7?b+~O34eQ^+bdc&!z0~$)p=!|
zfca!^&;{PaUQ01P+>jebp~R9t@;e#rf9e!Y3h()@lnhvhK~%kyveTnh6<7fp<PJG#
z*56TU9Qzkv4|}N~TG?k7we~r*Iwm`G_@2?oE%BLAr^Y$Q(-joE2`IMvXD(8M_mjiZ
z?nDAHH@e>__c4?2(?1>x%Wkz(FKHn$0``jp)cAf%v;#duZrcv`-3L7@#U?+0QKrYa
zF3qS{wE=T<;-7zPE1-UW!N&vsI0nD31QCp0ZQ4hE%=Who+X{a}O+&gU;5!|)VRT&X
z{VYGud*a2pVdtWzA}!wh7p>(tFLsBRMneNaL5&J0v%+W^#;T$~WT#K-K^;@j`2MC@
z$8h~tX{$DhPP5VWmaf5zMIa`Q)7Po&;Nfl?)c`)f+_F?j2FQDD1Okx)i1`_=+Q&U7
zY_3k-j`15|SSf!)^Y3bG1N&>pHxWk-K>Ax=(?-~2R@<lNni!_2RHxev_q_~(x_S^S
zj<eMbN1`^=pPldeS+rQYLuf4ud;Y#bsFFJK&5w~ih|^Qv{0Vjhy6t=MfeWv~>s+!r
zNQAm<p?dYJxd>}9?tN#=_EUtXo9S7|`Iuk_R&mtjoFC|#vqstQMkX{4Msk(O=Tn2-
z?I%LO(o5C35w0Nq8PkhPW^QyzTn`%DpKQCLQupX-qG}v8N(>T(=>P;~t2;WbJY%Da
zJn#Sd*@sV&K_H6U{t-XoCwyA$Dvj7(Bvezo=RmAm@#v=Bi3AY=c7$~Ia8Bx|3RIWR
z)KwOk@K@~Lb1UM7ota=Vzx3q7D@XZ6=F~hB3dFj6bYtX~8t%mvhzgjxze9+k-A%)~
zZK`VH#QH5ccn=Bup|)f7Zu_~M8y~-iTL0tY3#qNA0B3oP3Epu!SVyBM`#&nHXeCLB
zlSxy5yUeD3>B-btR9nM4Bp1`agrqm}Ryq&+RyP;OFE9O$w!gw~L&7sPY_0qN_8)Kl
zdKardvkM7%OkAmKfsm_OTpl!}1P;@1jC_^~FY(1<(sxI)Xn_E@B``k(8igtGx`8i|
zc~Gqr;nsAOijXfVQ)YI&-<SxGzs`wIzlQyN6@7o_s{KOz@x=q;PC6sO{lzmLUEpLt
zbmTyI)bpB!q4Sr01xOp$AZqZ+A=PB`VDWQzQNlwT_M@RDum;FkU4_*szuyk5ebCI>
z=F$mKbgmEZMQK266jW{0x$*NP?EZ6@_a{%Vx9`qap3tY;lLZ{5ZgKSMQ!uvr6gKj>
zz>()v#NXk4p@?7D=m*6(u(&!B3feF!y^HJoGwf>548dyrZ{Q8!y+HPc?z6@9RfMf{
z`3|het%RNsMGJ77skU5C6N*vhKq%Zcvt3aR)xz=gW=8khwNp>Y2X8s+fOsbHYLkV4
zBu@3cpamp}otO#LJf?`(+t``={h!@9X>IKMDjbCe;}zonhjQHs@OZwIIXyqV(?@T&
zvT@o9;1E{6Ikj<dptpv3s@ed(I8(YkuyJ#%qdZ!?Xi#zMJXr{HY6Sc@<|Pai-2ZtT
zR0aN_>k>o0ouYhSu${!`y;-%!AjOgVoo}eR^e*NsScGcN8{0b|g$SR<Uul=dMI#a!
zy5kq}`ct08#nPf=orH^Hc>EK2#<vL5e_VjbXNu53o5V}~`bHRU;$Fg^Omo+0WcKiI
zhL+qbzp_Jw{!WkSNtg-8B4qe|EYlGeo8qcP;p63pS13u0uC?Fm-+2_la(okVT0JW$
zHyXbb3~KvC@i|szdsKcolisSX2=&)vNmTg_QU_BrqQ$=30GNc9JGCO+r1JU>xsG5c
zqNv`D$*fMR=MiPB&DE7E$x6C^m(Nneh~G2!TpYB-JRk7y-jLj_OCX;7O|srcfL1|8
zeOlz=-z(t{(bepQ2YVvl)*Cwv9Z86#X%4F-ZRk%BFW%pyWbFg2iB;|3WVN~h>lk)K
zMqFQjA|BKUCv77;WRztzNwFnSE5^@gjJ}+UKu^!`Pw$?Tj)%;u){Ugdd!RC^?F|v?
zi-5IO`^t>wBnTZ|gwH22$0PfYs{Kp}L#!u=S0^K1feQ~>oe1hWju|vggWTKawh#?t
z>r;I$IW{Kq-x6sL)-@Ie<Y*-XSS#vM`hdGB{FClfRy0NiK!$)|GNJ-ik0H&g1ULQH
zAWwLjD+WY*Z#-Qe|G(T3iVK}=MF2Cbnh)OtKqUa&Tg#}9LwuJ^;^j@4jLux-w|{01
zcyN_~K=mZ31&INGWs+G^sqK~CX8ppvqW`qrj%VWaeWy!#I|yCAdvlzx@p0hM|1zKm
zJsyZ8u&PTt9k4x3=pqw!&e;k?K$*<0w<OLmn1PIjtpFxeA+|E~V|TlEJ!ajEo{^I@
zUGSZ}sZ;v`=c2nI+Z*J}`=duqayp<sI!+wFzj>nd6<up1HTv~0y*P;bM%_|y^!<KZ
zvf#icU2D`%{L~mbZvw5!`2(VVt#0x2*4EW+lO5%il}VerKi1N<(3`<>&nI_xa^bZ6
zS)p*Qe%9>89Jx{I$ubj?pp@W1HE^$xFyTmACpS6>48j{T;t)g`T!wVm!qPYmlP3pa
zuq7Xg*A`NWqx-3w)-Fa!O@OXbSk)*D+dez*v!>;bG7x-;3iQB*-r|;U+RS%5v$$-L
z)_B?8SN+rD&`hL4?M5Ic;isC^-Z-0=vclf>%p3?8TUHWi(B2!5-hcwE?1cc8t_8P8
z{X+TUqI@X!t;$`^9n$DRt$>6J$i(3Rgq!t2TPFa1<c^phk8pf{G`H_%rp=(FGU^f>
zHGtxtoui#}GxHr}Hq42}D9qcXfzA6Y_k)d^3ca2FUg(XU^76+gIe0z9GebovVDqI7
zbdC)3d#6v0SI#)?trF7<)nw|}5p}mapRF~q!E>U;eyD;vKMAqwf$A}5&XSX`8fxDE
z!m7)BNk5^YX#~<jx*~T?j^gALawYikprB5ZVJ*~{WIJrtXh<q`b}wq7)b@)AsJrjK
zn^4>pr{V(h_LK@jzJ2xSKX*X7F%dUm?Qof;MUPDaxKk-2Q6a}>2SMyne{u!XGQ9=b
zV)WvM;^MS@W<^JpqTB{Iw(gmUfaM3=6DuB<?%40T!d2U+zq3wt*gqQlm!VB1V9cTn
z<jQKfAww&t=6jYw&~mCf#Mp1$(`8!+Tf9+2^C0Quv#hUHT(HKX@>Iy`cfVQL$SiLf
z2@iv3&3a9<tNKGC3L0m;7b+hOQSQ&xS-Lt^yF_913net*iK6}t5`#*TF43H(^ZqvB
z8yB`aa)+47VeF@LYkqw$l6Pq&sF6Gb1h|8%_K5iel}{{3C&r+6i~SELWTt3=@R~!7
z2ZSW#POcl)Z~J->H0oT9yK3$4d5n|{4M<-$`u;mz*c-TA^t*XMIm#RUb9&BWO8XlF
zCAQQ0;o$B<K+a#z)RTD~reThgatDFxnwxDq0<(B}7<i`6UV<Ua!@BcWYmsbwF9mrR
zLuW@F_T%69LqU@X*8Nz{A6U|i^t5BopzFLshy`jxai91J1=*A+Oim~5ks=KsJU(Bs
z`-v0a0^il_<#J7%W&HEIK#>Y(RfY8egB@=^ri-N^z?S@mQK<+On)Iv~eU}-I{K^XA
zG*h6b?OxJkle4B~mfJErPW7?f{?_nAE`pkdowZN#WM#1-s_=2qt=KSvQTlF`b54OX
z7oj%YOY>l#ZB(l0=jp$}PO(mb$gC0o^z#4RB2D+H@S#E8Rq=l?OfQa2Th0-rZ)p0M
z!IRZwG_-Fp>QsWdwJAuguQsp0<K|u)ulyE14FX5gdo}LUCVIo`X79JkcZx231HqB1
z6hVo_>X`KBmTCr3Vk%K?|5@#j^jE;?I=3U7rQy|oN?+79jO<RrHl8aE{HKV2<9y<N
zWsjAB8f`IniG8ej+?9&AuE_-t`&`=FzR%iCBHMwaQ8iHaS#0<WI)dX)D1ojI$7E6X
z-O75@#rR62w;jy~(Ejnphh2!%f4kv5N1^QkKuG_G?l1vIDZQ?oB*LI=(#0Q*OXjKU
zh!;YWdXnJwoX?8x_oAgHO6`sxY|8I-=nAODXfB5c%>Va(MZG9@9Xrf4C6GXq?z1o6
zt6sDqQCdYk5)|dzC|UVzM49j6P)OKkPO~RmrqLFTlC3jjR`#i8LM&;M!6$U>8HvlO
zQu1{imkjzyey9mY0)*moFOEs`>zsq=pC(_L;C^qf3#M_|>6`us>bfavO(s}2c!zg+
z8^Urd?(R_eOHqYIP(10s8WmvPbJ`(U_Nx7t0;lKMo-aR=Q%LI~hsHAwNef`e8@UWU
zL)fx`GiX4V?G;6<Yrih({sGI25#wwfGT<oOxf+qFvjNy2zSUZegPvi~(xE+<8l8_1
z{-#wtzNxV<Hd+4d#uRApv-R1b5>gt{-D!N!;rZ93G`93dEy*1<DV!vke){C7B?CBN
zDk@OIVz3aWPpqu@B&>N`CDAGdt-itylzQkZ!=ai46?{&UXL@7oF_O)*QVjmNa0`vZ
zfb!VCOgMn;%fS7_^<p@~UITV@tjW7!@A;44E2Kdh0w+ImgFi{%!v}g%7{kIsZB*gx
zr@#EoS+}n#%7?W&-yW-})j~dW<P2@R1*fJ)3p7_R{?UVA(i%~B2kG&|`zTN$PXRMT
z>fYTld@}Sa4=^C^$4}}%YE?=Y<$0%r?2{{A+tmi$S6ZSbINwI==aTB5Qc(O4f68{X
znI9f3^b`<GRsS9$VU18j2@PTHdIfU8u0m0+CVCOo2^%egYql%y82h!QoKYpVl{&$c
z-tmLTjada?gqC|fd*+%?w=aV}NdC%srhq>thCWKYIm>&@jW@FmFfoUi^xFEX3NWAV
z;$QLrIAhCY@}_xehEEahP@<ZfBs55K0$*a3shI3ZY_0CHp^fw<(4!Yj=gssy>UcPE
z$`SMuy3Rz%2bX-n=R}Z6ZAq`w7H?(D*!KR;>O`~AUP~)$|DJ|}&-4{(QA-kyz~G(8
zSNx6ypjj<!DP`;rq!RTr|7;-@9?xEgVWq^P5>@cs-$2arxjboXJ?-Z!M{iUB6`^ms
zLJR3REeP1DP9OqoUn;%a(sXZv8eQIHqPo6b2auIHAZAi~4MFoXYHB3*gypu9N$qUX
z!YZpfHr6fBruo~iVR2#o1>$wa&^fQ*Q?L`qyMub*FOj0?2#up|7&ifxDp;gNRz7Sk
z>;;G{ND~xEa{*sK-<eNAgM)?M0urg<ZcRo=Dnj$SlYRK!r+l}%{8ESaD^nlHv0L4@
zjJT3w+})h|fs^n;3G^l!#Cc&8R_%+mIEY{Sp;7$@70>f?`?<cnSSk4RygY9*x)ZQC
zxt84C32KOcf51W9gj?H^bs2DtxKgA4A60K24|Vsyk9Xf~<d#ZBWGSQw*|R6YSc`09
zOLk){+4oYE5Jt!z#uj5A%h-w<WZwoemXKvkC<YnZ_l)%ZeSWXM?)zcl%sH>~T(9eT
zTJs!FWsuYVF%|J=9iUNEg9<Dp)>Lgm6ALFN3hjGnm2w=c(`tr1nneRqz=jTj4C<fw
zqTPC-bno2{@;hjHNdIv9=z|95qr(Ca&|aTS@|w&>{3}Eo{B2O3gpOSY!b!6q3<5vd
zcSh?D5!(C!wAgMh$x6d-T3xX}hzs_SbUi=EGj(VGM5MOFDCK6`pP5jl@52I5<)lsj
zEhWC=Th2)Vo$1%omYB51_xyo9i08Tl3iG~q!P~q1Y4?4RLoSHy*V4Q?18Q?YyXSc|
z9Mu@m4Nxe_Yx47H&0%>411PsD@$z#9DR#?t{vzxTWLf%zf_muQYGKJ;c5fCy1vW>j
z>3*?8;e7{_RwTI6!AGNtxZ<_Y@*G|f0TR?*GSX~uJYZW=|A*WL2k3%qDCCO=T%otp
zEZ+x80mz(sE)CE}4)Fi|pArb%lQ+0Yx>=hF?y{5KNt>XrykOjj)r3fUdzNek$-45a
z(3KlBE7|24=ynm{QANj$=ac6v0u_!?l!O|MCpWWLFGQ$-*17rW&0tGd37jo79B7Y)
z6w!uIS5~#LlH+I=4>kLBy~&-rqzIdl$&k+T`_%si&&P*144{gcpxhttQTVH(uMO}n
zX2|C>@s4<K7N|R1q<mxEc9!#i7VppClVPsmpWXqMcudfHWxxc!Y1Wk}JK#U-M-Vxw
zjk*SU(Ez?v4S!q-tUqel-bo>K!Nog$gOfk1C_7%u&w}CbdMUTauR09TL_s0*JaQ7>
z*ALsGAIGj6m?-8vytzm_y!1m~PBGUbgtHHsGyNcErq5Pyd4nDN9$TRQ^Yk*zA__pM
z0sZwtAf=2SfjlYbbW(70)?HE2z{RKb{%MBPG=NmteP#me@KWiSh2)a)t!6L(81cy|
zwD@To7UNdz{#L0;QB{KqbsHW{(93|n*kf^owuYZ>e;v=RAq?)^MmTh>Pm>Iu2r8t#
zRy`jD)aQZI?wKvvru||08<#MRCpgOrfoq%|XGQO1ZKh71Xya>VwBB_DME?6g22iMN
zu~BswmpIN@xVc+r!rdm_quvcU)o!#6Bfk<t`UDk@RB2i^Th6!a(KDydzAY*9ZN6el
zUDUf5mKk@6?0O;yb3URsw;AB-HVV&)D^0Lde+r(?qQ^NyK8deY(Fo^IW3Sg205phC
zJI0=+M!?quR8dvCq~nWXuZV-ReTg?QrMl#M2b{hDybKn!_KxEM269^QmWja%=~{l!
zn{j2Y9(OB%taocJsOlt8ht8cz51XPhS&AwOJ$;4f_CD4#puD}tPFl5>8kv6lmjx%;
z!|!;^kfG7b1f6Ugh2()Ka1&dw?|37HT_mxxFttAtaKL;eJuLEh{-Mypg^QF9s`8v9
zF(dbME)b~A1M<CEIYp>_9JL<*){gR|_<K^&id{PS#aT0F#uan5k)MN({i5*5UzO!^
zO8=I=V@Q`6@Y&9o&H_lQdH}LR3+P?f+7Q?M;B|WDu}}F8mXrgZZ8Rxc5=T~mQ0O4y
z=zoh0mNIcE+znd&T-bMYQNk`FSnDvcznS;DOJ6<!XuUnp&?$+f^Ze=WVaY1E0#|Mj
z8Ek%KOZa??8a%xNc!4Hr2tA=uSEq=JO3Rwq%6zHB;m<P+TI)=2<B>&-)?L6s468d6
zyl(!k*lyFk`g8ASTw{GeyDywq37qYl3xEQU*LGMW>$x9Nbf7rWWDILE>34v*#CtS_
zm_WzxlR9{I?I28`VIqO8_hX)+ek1TaHU8GH<MEJRTWNC0)Hoi>{{=WR>7rbdsS22H
zvkJHu+^{=WV3S&3c*`fSgI(hWx7V5WDX{eSM3bi7(R7=VIS>u1RV-?$61HW=Hb!I>
zakX;}qV%A?@*SMRxeYu^;uF-mi&}4+v~H4e=h{{k=tEQg@^LY;OF9m(1zVCqp}ip#
zTSyZRF*!sDb#l{&dzTTyA&ixEj@**6>J(^UKZq&7ztUu^b`!s_dBS)%kzHr=(nyOd
zo3W#uD#0IA-MT;o<6ynC66hut7gL9Mu7Yznq@z^$3yG}+Q+tvvLXD$qBm7$i{)Gxo
zHK~zAG8F0(5WXo4IA;;ahjR5@9`FM9n4u;HW?mRdGEyVP6M6M)woP?xS|Ht@757tD
zEChuG-|hEAH0tzQXnZJi*C1VjDDV@{T6MO9msAXUGwMky_&D{sMGH#peC!T&^`)T1
zP5c(t1H%2o<3H6_#uxWraF55i2o;aP+RE)YXRcJr$543B6PNAoc?TXgz0Lv+4_0G3
zZRXvaYW9Ko$qni8H5c=2(N-`S%)|-+oP&1ve~kOS<Fd0{*`6dU8fYvACIw>CyVy%c
ze2%BE&CDpZ@!T2t_n&{C#l+v_{{H>4vjHOY=Ga)=ckY-7;WPhU;g<T5!j^C(?AO01
zgd*j(8`mcU3UZtqHlvplCivn97nl3-lRax>(GPqDi);nRAPl+C!Wb3^Cw#K_vMvP5
z)kQP%B8e60o4|+;4YLXb$ZO^vwuplrR}=Hl5-`**6KqL?dntxYKm;MbCE#-I!%~p*
zYLxfL3hX-uW~+3)h+l^$@s2<z1&GC%Ir^cy;%E$;7=tuz=j6H~XfoD4%EPE#mzMFg
zErJG!K?Sd^tw|FI9-~<2OIBjN`aa_TtB-f+iu6`KPkME||6)uHQ064vvz4DudJ}(}
z_S5c7T9LSdN%?6&a{_8WcT0fL_!Iq;bh`FeK?NPOFp|(4j*y-hpuoa->NHxgi<CgV
z2C)#wUItPE1ZKSev%gl-{;C7-u}d%)J(na-SR^M*>1XwIZElC|!dyqHWpvfKqE>U>
zTF%eynu5#x{Y3mlr2-}G7|+jeYk=47s8_9%1=Xrv)QA2v_O*;m+&ALE$`F9>?Ve8c
zYt(lKmx<Y%bwG`~paScV0#se04!&bXInxOSTn>rI$vBTTX7>!?xpnM2A5`mKy8{!>
zsQB8=r0ZAq<`r+-v?>?O$~8rb$3gbqxBO&qH=oY_ItNdT|0z)}_(Cr_^0p#wN_lVN
zJ)uq73}BebCx9Y{-(w*Ecz#=}Be~1q;=fBeZyCzF2Zh@fU6Plb<~McF>sY!BhPXqY
z`;uk)ygF>D2vpK=_&RPgK4P$%bpHPi8A6Gn9&+$4?3gL3?5{(1L{?0^zhA_qTb1b|
z@J#MSD;T3ZSLuE=K>LmOj*0=0k^z39hCi49SS;r5n}PMFBLTg&Z+DCkySe&X3IrLA
z052gXYprhF%Ia!-nA=%)7cSVYxj%K+@)N_hI|iWsU!mKrID8-?ormZQ(ff0jv1D7w
zO2O`{L)-Ugb|*38jc27V%jgbGxkT6=D5;{Y)6@+Zd)x8ou-wNwF`kYFq80Aa3gRgl
z)$aMATMNdxbBTs-bEn}tM?|dL2m!5n#$EGHN&eZ!7Xqsi9RZu_=)4fbmNeX!oV2;e
zk#)?tuy|Ms5`sN&zN80A``}D>tI!*EJn&eu3S_^m#{R}f%z1mW<i8+ot;~FMiCk9q
z#;eC{pI*XeU^G9j$nN3B+>*s%*Yab<%fbMBTn-$PrmeY7Y7+UT_w%CXW}(}ld3s}>
zx>;jzPqaZYXAurzJI@EWfhJY4=pZm?M8zOz|6xisc<rg-_z{wh5`YpK!SrzWXdrG7
zQ>oF<0a@i>`#xwx{V{m7vVX4*(7l9>---a{3ws9cqF2qBep%z(?GM-TV#G_k9t*5Q
zOgX!7%{_0$$_~_p0bk(ln&S8sP-bopw=KffJdc*yv+b$ix)GoU^1LW%E494Qb`aX$
z=^w1n*9HxgD{EefZy!v}i#m{+!adKWbPO=NM<Fd@$p(<w{pQ0>d3qvYLww@P#iPCr
z|4ilDBej-dbEQ)W{bL3=P@B=WNjpu)c)y*+!DROfKCDhgZh(QEt!wg)w{m4cb!kA`
zaISPhXB)H*448AUWV4t10?*|E=$J?DG1j^AsG|5ijx+3=-odu=hUZ^V_vWcVH3haj
z_)6IJPCtpe0n$H^Xe;AgmKAk?t^}ZyILNl30-H&_N5y{sI^sNuQ+ZJB1V-A!QI-e7
z!W(I%KXj94tak>?<8=khF+bw{X2#U@m-kieL>9x*^C`wWH`oRC<2&voZZ?3DBy2Uf
z$~%GP_9ixA|FvLJ65UQ3s8XHb0XB7hlH;1*=nkzQ^*l=1#jIn7m<O&An;fsyQ_evi
zTzF|H$6&m(=`(cl@9T3#%?<#2zLYxNuW`t84|gB52p+5L1R0e6Y|y1M_uDYn_FNa0
zKm3F)uQtqw1#Ge<b=&Ldbc(uwq2>Qr9Gz!Y#3g)p5siwcc~+&09~>AhE*Pyvkdl*7
zAmO_VuE?B=RfeV=x=hkwdV;#DUOyJTHpj?XDBaZ1Fa%#IMyQRLe>XJUWXBWuupboi
zq7D5csUHN-nXs(VfyXq(T)LN;De+9E264ifBD>*tzrE^cF--L2GmdK3D)&;xi^VMT
z8r@+Y<zU}@Y2Kri8_`@GZokC|Q$=dYm3e99Foi^9|Bw-IsA^U*fGlcwyrnPC6<I%3
zZm<`8=eC~}KoJvn4}1#r!8o8tzWWEFkP~`@M6XeiR(CAaP!u80MP!FqRN`>+LcV{N
zb_Cc~6<p~1-Z^=fqlBKZMm=v+G{pY0bN;U%)|fY34uL~XplL)emj;@4W=dZaai5Cw
zRhUd3t|*9Q<-u3h#_|vyPZ_-Ktx<VQCIjkt&22_DX{iLh+JI|%8Mmyko58P6q&|A)
zx{htI7}d`GQ{gLZ9Gn%>vC>Qto&({-{#9Zna-WB&B5+JU>f#$Mf!(&G{ZeonDgi4A
z7}|N7vO`yCuR@dxj`5swT^bM~_Vm=*nv?`>H<>>eaI^GHg#YNCl$SsZK1}}jVCXgk
z$M@Yxd~E#E?a4`SP<`$~lQ9{4i_g~7DR_~Y4r&ALPS%+a_B6{ThkF$ujex5-y6`v5
zZ?tCfBtmq{)dA%g)eR0I*)Xxj($ot1MooMo5PXe8^iShZW$X~Md;G;@Lkl5?g1e7u
zH-^>_@_fYSIJ|11oXX+;=7)OV2tG_O3z|<2WDGpHYtF;@iGd3yFi)CLi*(Pfbu6q7
zUsJWq^H}TtBpsED)~|~dzYZ2$IZVqh<U7$wPuW~aLvW{MBoDQ;w#s%eUI0W)iODi<
zkQcVmKf_926|wOeYgATkw?Vva?zi09?U60^Zw!1$pv+J@g%mm}EYyLAvqz$S?Ef>h
z@TRJRxm_@5q3?iL@Tj4{k9xO+h<TM~9yPc*RJkX<|LJJZZ^tvvMO>>a{=4tu4tZ!H
z76(1TW0-|WN*~O*7HQsc_jvKh*bMDuIWNZ$z_2$m`F$#r#S-(LlcZ5XfS&&mhnzR9
zQiQ8~6*x^?Q;?l8W?wOK0R42F!S@r!GK3Bb<}!sw%@|LKos!n&z@60Pt&Ine{(@g`
z<+pli8<wQ}WjBr6!J_S9qKibQ9?3H8elnH$_$aO5t3?L(orG0qM3itw3{ZTr(7-_G
z6!o)1e=hb%r*2Iqk_!=4P?&`!YWKe@2f8ib-=RPaVd)3w^+8RHGy&aFOWv`{<0air
z#$6nIaW*&8+2xTqej<0Wp7i?T_GyoM>tu}+s{@l#vWv#-CDf7GrpQ>dc$YNqjhn;y
z;~YTKe(uMDXp*atkV1U-;&{fY`OeQhC9Zp$DQWI|3>S`7o_%A^J0hEOY7<R0Q%_Zx
z9Jv61pHD!nq*1$nE@!mcEaZ)pSqxbRA#ab`Reblq{6u9IXc^F)5>G3v-XQ)Ay#&xs
zh3y7X?b4uZ-S{gh(a|Kh{`+o5lgo2<X^D$9MOgp~0B)a|U-4?(dCJVhZb?H|z6`GT
z*Y~@ub|Q?<!yOg@#HZbwNV&B}@5iRKsa@3}8oJmm1!b;O`v?3DWv3xGfGQC6u#JN6
zZjlC<*VeG_b`e3>iIUv#2d>+T{q<5n5QZ!QArzsTR9+o=gpYoRbob{)$3m04^iBNM
zsxzb&W@&@2$l{2tw*U?)EWKO2WFMFfk%+?ak%0n>peY-hmHqlaRt>>IkiEaVevB!v
z-~DTM#bhN5h-S8WHsnET4B3^>4_$Z<2^(ok?(QbdFr;2jR!Hw${1)oaQ$Kp)h1=Qm
zbfYPBhUTAga%tm>p*Bi2-!Xk2m$eV6TM98A(a7LRocVta1n`BZPp^k<Md<tcPDvd)
z*}qI1z@*8pZj;?HRrZH={>n`|%?%hom<5|E4zRZjtdKIrV@Usu+q@7;p`k_7-1^Gp
z<$l%R4THe>pZVr)EmdDsKKB<-;0yPZRU3ufA$I;Ld41*Qt7G*MSi~jxk8o7LH+v&z
z8jsqlrM(*O;EvAu^U)XCMMYV`G^)@7<DL=I?dZtGRV}$tH=v5b)X=?mE9tb*=L5t}
z7iSSb#JrE#iBzNTQstMLm{vFbe2F8>Ibk5$<2(XUR($yWMi-#e(w0lGQqg`Ea;}O9
z4=1q#q57H8b!J_SRZw)?RRFJ9hR;u=7d@}X_QcfvDmOy9<#<(Ix#uM-Y6c<a>d`4W
zNzaiW3Oaz@*0kkS%sMGVk*K0w+LgUWsKMDE8U>oK6mi5+L<hU9^oL=Enf{&1?d^iN
zNgeT-6U0M9IPrqbI{xAt9%BeEPoQ$Y9k&hlC96(){gH{+Mlg=z&A(_MPN1RoggBQg
zIyhd?)gYXfrs+SY;AX1}V$H?HBJ9-WdKYs6FD|oM7G`MHTIOw~^B_k2f#rzHM<&zD
zY3(m01jSD1(0C+)T~&5FTACLt&S=g*A9}+jX1zAirFS^Pl;+#^rkxHm=oKUrqnLtR
z{+E?mG|MAwQ-iA^l%+2jDwg_bKMx`cu&+gG(k8wmv{A&@0&@2Nv{5?1Qf1hArLg4{
zG#WT&tv&u%6pWLVcVGR8BOqTMGd79xM>Lm575$p#gc(-;6vuTs`6%Kdw872pHyd{9
zK548Il?NiC9QdvqL>Eh6{v_*Omk+CPi~qq~!uOXRu9fHvy#I^syYKBx#emFtbQFL3
zxOGxUa|n>dJETDe={wE-#8XaKRx>BeqFxER6%6bRb*SEmh*m=b=!$qxSYxE4ijIco
zr`P(oRqQ<D%HATR1C9IU8k*=$^dCm*M~ZG&lRW(S?%^wRr{%Vsx&11+S}aSdo>-OU
z*QcN2F_d85v-5mDk0h3bi*si&mZR;J#yIJ@#fUd*K5~N~yL_Ox!5q)3)X@uK<tL7t
zZKv*9$$rL~1Coo9cb*`dx=+xitY_nykKU}t*~g4wxsnF<GhLmtvb^=wKYOQ;Zo5P$
zB>Qj)rIA~yL$R?*JLF~hFQ0CUH7Zw{?f(o*A#Zuhnn<o(ZtLM~V`0~MgAQu>IWoqo
z=WSPKXD*93pPu+~BK7At8W5<>IEb3vBX|X80ZA6M(BZFObo%K{Wzcbb^GP{qqM+>I
z-iBsw?=e<b$xUPwdQFn79XirREoK~v671wGbolwV(rr(Z*Q%F}ts8GK^Sn&o{LS1s
zofGCWhKoFL!#ZAA#b3g&W&YQNV;4^{g2cn<RaA=of6fuM^QV;FgN!^KJMqif02emE
zHgm#Q(!*OFQMK8+wx%5gW-HC}rfqL{gp~*0ERmd}FW1g(p%bAdf_JhDcg|0(23<q0
zA8yB;4jGS6tR7<x&p2Tfgqg}kc2)?VOopE73ymiw@zt?tst~atVFabB9XX#j^H1U{
zAs|%320#Y00Bl7_-2cyKJ}r?X)VPDhN0OOJ0+O+Dp+BchyB)8IS_~ekXJN{t+s6NV
z9OFrN%4luoWZgAe1frU?HOX}>t{(m{S8_QpjaSL(I~9lpT$DPM37&ZicbK*wA$^LC
z5ZxlgN@Q!C0Om#p9J-*Lr<1jczvRZCn=&dZ#v1+>Bp6rP7GE2jF)|ey1oL<Jfd>Sj
zU7Fkb(`%cHxN#*A|C3&KHBdnKl|DPuu6bS2K!=O;xMR@^kkq|IP(K2+XWEPAo`i0W
z&&h`rp{HG}JzH9-ThzOFn+*(KI(Qs^{0s%NhFk9h+`DpHEVrTY(afhcQJ&>+YHBZW
zdY{yQ_-Gv5Ok3XFTQlNlfTxCiT{0&<zs#9Kbc4rOZ9Q0_9c8%1>izYT7-IDD&}7j1
zd%MxjAE~;dcRzaFXv3hIfNL-&{!qpbVEI6*rorkbEw|mt%P#w%wU4@nJ8(CTzeGT=
zoG-+krEykF#~5Q52#m(ueYcdDay!OY<DWb!qQ|u@!U<-WoBli}&V|08K9+deMaqHq
z78DXQ)I%?=5pQ9>_cHGq`PWJ9g=uop-j2pea&KARj!1QsJ|e7+={cbG0&y1>9paA{
zl7RFMTly@0)o+WQur^@fyOWwOEvesn0J^%f^#9SoSpaO`o56d?jhjXT#A)eKp~I`+
zd_#&XOSO3l9o};nDje|{f7AGygRlZLt?cICruD?vN+yS+aR3hTg21m21}9p^tbfjz
z)ZA5Q^5G2LZmyJU13fW5fn()8z}{fhSeaHL@V{&4e;A~!J=pZNP<Yt%>0oQ<(Zl{n
zNJ-T6aR}k1wwO4`f8&OOW`90X6@_jkNjAGOTY0Ht4CwZI*a6~a29b^+ze0PJJ0Ivv
zfnVWxGbw4{Qr<7;VKDX(^>`)BS<cUK$`yb%T}NhDbeX_Q2%JNr{Le0~*PfNc#Pnd}
zFVRa8{$Tc2+-N}QnJxhtms#tQ3nt(dEKvYux|@LfqP(IY@&TL*+D@rjFaeA@4EM~J
z#1*%bfgsU$BX#%MX1_3u^w8N%#=Hct)851{!_Iwn!LBL@a=}UlPG)5+57pQ!O#OnD
zEFneW){(V$`{{TZ?(~at!%^Ra^LPUF`cLu%wqx=#ByU-DUW6zJ5z&Z^gjyJmc`XC6
zr`XSnLc61x4{gqUA*y8068m!TfZCZVIo4Apvzc@Q<v59l5%!`_z`Pv1-j+hp5B1&8
z>x4=;XYr@vg;fTl?-C$gQRV|6)_a}ZUH-X-01V(Dn(j*U;nVAz(H-5&@?F`dSnl>@
z1SA7#*RHmTCTCVP`_<07NA?GO`q$D4W355P3@*XT=>ajG@^<UvTwH#m1J%R652@em
zzFDe@u?_YeYou_nW26)}kroonMV$S>`59whh%|&yLXbcb)X>llIdu!G94XwyMSPF+
zE)003kC3iB^JDiNa)s5(xn%XspCZ#!jC4RBtM#}N7p%jT;2D?X8`b&n;vFRi=j!JV
ztat!{0psOA`TA~=u3HScAFCzSI4b4B?>Nr~^fU_;9*pE&;~P^zBu#UYngUml*AT|Q
zRy;DyyV{K=RZD4Q<YD!sWsL3rn;c^`>8m;W1_a4peZ$4%Om^_p2=-1Ps3_jUzOzUq
z#~{5Sc(!Ifp*XS|C{#xYAH7l}1ZW&V5R12Dc1XB>(X=DPLKM5Bn}77s;IXuw1shmb
zk#-Gxu{`jjzY+Pn(FbiV2exINuPmroPc#Uxzkw767rTMZDwc19(#RNMf`3;FKt_EF
z^SX6FCw{ldy@~-88Oz>w1H`M=F=J9Qb9T+ZT*_&h8wXB+x<E628Tz^>B?)AXh);cC
zk0Ob<yRaW@FjM_HWD{ssSS@8Ard}*PQxOAOgu1Z08gc-g+fhq|$CI1Jd9jsec-!%O
zE9Zl`NJb*rO^quz8fMc^l0WwKNk^TsRI>w&$Jle5H%o+;j~VO5i_7GxzhLXp&h@nX
zQ2J)ddBxChc^*gnRoSfx;tHDSzUIcbQ~zgq3V_jj5>k@rk&R?SNb{^~;CX?z>Rxm`
zZ)#m}BKe+`X7=TKzxTQ2vOkxf<k^k92z8L*@q}Qs`-9^=az9&5z#?l!e<&V)hjp*-
zk27OEtKDz-i8W;w=jYv9`FuU92sZROstD9<^9t776=buUcD}}8R#fNLVf#CQ#wLrf
zf5b4vQK=peouWW}5>e-Tb8fsu`85qNy+Erw0S-o>SkRQ)&bcS6Gwm^?P*`U9^VfNC
ze+H?2q}+``S7~cdjzSVmU_c842x{-<qc<#JOAr_w3ze}%1@=R?ufUc%xXt*9vFAv2
z`u%l|;=gqxg%wbt2}Vt+$jKOw2=oR!Y$^83z7_cJQP@(3I!3d<^(Z^x8j&k%Q2HRI
z06}Ct8pmioE~N|x^D<k>E`<SvXBz9Gn7}~$BHU&d|BP;nz}u>uuA(WHSWqeHfm!4%
z1W2&h4>#MOd_R1gMNkoXHlQ+h&4CFBbk3xofA9f*Ams<8Rm9<aOzFA`6G8iB7`pv1
zhoKTDpBwsdv4LlKRnV%F=SrlTex$G*Vz2)5v_(0taX3LLa11KLMf60-pDp#tMoyZn
zaUNdF6T<18et4jU#JhN#<>7OxQz1dokAat2rxrcfWEIR>VmURJ=j*uFKTE5RRme&%
zm^uJLo^s4OaEXLAycs5T2vo_6JQ<@eh7{$Pn^3KXJr77vNiUF+>KLywUr*=SlgG+%
z-Rgd!$n1Y2W)MMsLtLhH)f)KoxArYi=wQS4Q$=Vgh)b-8to<(bT^H=HY3#lA=1r$h
zt`If0apJiCB2K5b4P%=FJC|%?k>&9*mJ<w7ER03>I9Rjk*zoyjiZJe67>t|GeVf!V
zn+IO^=VlMSNQX2DjsaNs_-ByH{o$qI06$H?lG%Ls=i~eCC{prU@}$Piy>N7KFi^gS
zBZ7Dy^`#=ci>=LYc!%Tr`_eqS(_d8&>xyFOR$LR0X0$_!1ngQLWsd(GG*Y(_Gzt{~
zs)jy6P~beKJO}Q^7cJ+^?4CB~Y{M6U2;E3EMfz&tHcPR>1N$FZcJ3F%pX8Q*vbZPb
z>-<%8i6M{0s?tSIr*CGO?d*E7$g2o+1n7hQ_lISGispp5Z%IN#|5*b~wZWOn5hgV}
z`>p*r96r0yXtI?j?Z0>Xego>N*Z-VpsWT9Y2E{@8w+=SS?oU{a?dG4PYxrf0;GnzB
zsvJ1Bm0e8vS}3P0KO6TlFl;<pKw=|(WKm&qS8+c4Vby0jx_+MKH&PY}L7T3>XwSO@
z;fk|-Q??%qTwr&RaQj0>EfPIuobn1SAwbN0dQYW)j)g%L5mf$phoQvQ28Wk|aL^f5
z0wj2$F)#IX05O{y*o8DDvllO*y&KPJ>?oh@sV3B1CoKcV#4cwd))YS>=sc4<c_pMs
z3&v+gle$rq*)_=F!LB0+`0TcHSMbI5+&|^V)0XhP1IA+etr=DNhE5w*<tm1w*CJd$
zrwx}PRLrS=%HC_|0zTk%^n}33!+{4keI9(Qe~a8VYLQkjdAyefGy1!tGj9NdEAceX
zR}iUvg@cM_oK^d|P30S)VosMjc~{C{{fkDm=bKJzF^RZ>Df=@C1*Kr{CK#@?!mvC$
z?|q`OSiQe^>Ko&AJU%jrR2-NN_Gr=V%AK3xhlRAXD{c4y^#)sUIn*R`g;xLa(|eOu
zxZ+u;@ur6Q$<E+q7Th6OQ0!PR7eBP;PFBl6n9mb1jy@hAY|ZT&#4pag{rq*%Nul1a
z_(-sU6XIg!8<ddlnSxmW${RJ^^D5vZEz{=&ugC^Zs*Gb`w-f?f<5?5OuYKTMK4A>b
zi26!Yg7CS^PA7cJ&Qbal5qoZp!?Z&kg*JvnAG_&Cf#MEL)*!)Jy6|kCO{@wIXHCTf
zB%CVO&Z(eGXWMx=i!9Xdt#?2RSJ*Gno8qXO)0%AqX0GYPpfCBd8XIz0uS;%db0;#C
zUlR?5ukg8LeO{hgMnp?u7XlRmvnMuE$Nh%Q3>WkzSMpse^TT{9e_kkN)&0|BKvSBs
z9=K!QDME%%+)PGzlwn(Zq2XK6p{2{MJmZTgl1`DMk5MR3>WpLRL-Xa=*%h8S4e`0m
zEjY(3(#P5xkpA{E>%LWxM;1<QvdIP6>VkBP^0h=x^{CnG^C;4ncHnw7kG=Fax!#Hi
zpcWi+ztfg`B?L!x5S^m0pN+a+JMvj7{p-y&nw8mkP_JH$@V0Lr{thjU^_+5hu-hF`
zV*2IPSqXpvN%%v#MFF{|xmD@oFQc36QD|}Yj^%S}3}L0qc?LCFf0p@vij6Jlm*zkl
z9#h5ed94~Wt21tk7PR(gsv`uodln<2(@>)$EM^6OFD;-CaGYag|JR%_J>a}TvSAM3
zjSRBmxMYzVD-IgVxNeXXr}$gl+a!FUJ+H+z_y0Ryt!Czd8^mE?#2yN?mfPEfZ{^)J
z+9aM^^A0U_K3t6;^?eVwS7yb?kbtcpc=PG5+7>WgwW-xFzm*LzrW5xA7Vm0dhyH8y
z_d$LyufnyL2?ChXE3l+hv2rY!2+0Qt48N<f=JO#SMqF;5UK*PX#}67_wl*t{y%5E9
zR^V8s-l*?T<Da&gV+63yK6xQ6+4DTRF!;->5WLajz1lN4vkyV<2PF<2P?$Fv!~yI5
znSx0d9TQ&+Cp6yWg4xRFl_59FzeT4flQn}$-(JnRzj({-u;14oRiuJiODB5YeXXb+
zmG@eZE*d)z_y}MUQ<)}LVN;pt#M!<%^Og2<CX~nNuG}mpvWvT?yuAbsr_-t1(97b;
z0th^xBZV6pGtg-W%xPw;;&^g!qS4Mv9Nt5qwZi_^BXc}dhP4^PtKP|gzxHSIde4K~
zr;&!$u1p7QJ~cOMEFgz~iSfvC%2a{w%X`Dobb6rqA8x^K(^Qvr7QM}9D+D-H0ePA-
z0xUYxG9{e>SDb3*fRD$^q<?)|oA;J{(l~nwKS%;Gi?}8$Of=kES$1HrRD}BXDwCm-
z<w4N?cPsO`EYX<-%vKnlz`MG=#L>w(vNxLTfw|7d`qE-gF?0w#!6s75WKp$04E<LK
zF@LPKpm}S7aAMkKIowuqa0!05{<i+@{a_H||4xGJE0HsF%=0NVEaRi$t7-HCZ>{2m
zv5XJjBIj2K@6CC4eD;A2?)pq6PNa|E{Y=L=e~_KF&R9vC0QnGx^5qG@K}h-x<UVg6
z`X>|)fRig-iyVmaH(*P<;_ONTx9Em1ZpR?`|Gn6w?)MyBWgc96;yA1$<D6-&?r<CN
zrto3vWAEeKRv^?J`T;t60%&^x5?X*Il<4$}Cu@JLAVQzRFq>7wk3ralxOU<fv>;ZW
zZ}0NR7}Kz*7sG5`lV?m|R+45;g9V&iTn30<-!+8>utWwatjtyX4~Fv7-X#vQ`KYYF
zP9JKDdS29m_%RI!!5xcB)96ktGBMF;<1}n38+|!ar#E(^_Ni~-1HLxQboMq5-<1=@
z{lM+d`9DoA#rfUUhBbT_V=-~5A9o-99ZLk{uIxPCe>v8E{u0*e3G;k~mfCt=H0(@T
z(@aTNw2nV<Zldv3lw6u0&ZcFH2oeF(j`daow5Yo;Le!7&l4H;%VESwnZwqmO1~hos
z1$NTaY&(J_xC0f3eZ;c58aaQ13Vw>Z>>^#cAfxSpskAjD0fv<Q$9AxO2f*VIBzmjg
zY$QOp8=-G_@3`6A@iFw`LtC7HGyYk(y${hzd(<~2W7*8=BW+X{b9UzPUA13XjMjla
z2Cet1;yi~;yBAYP_3@sv#x(xbmxv!w4W5^^*4KIKGmfVCSz+KkrFIk53{itm)JRy4
zz-70<tzYhx;g>p#T$>zpHfejD6sV4oE31@V(yB%Dd2yg|g?XY;B1ov|ZN{J!itVJ~
zP4)qx^DXuV&2exFR*l3Y@C|M-TbXkaU4YdpLS__BxvJAj-;*lJUdeg0{K6e7_U9|y
zx7yRt1~!xe+SzN>s=0J8C4dGQVdFp8M_SRKsXkHkCDZ3XuV6Pk-|Mt5E4QjGlyt_;
zVr;k#PbiRM#ZVjHD*>c$l+%)iN}x`)!Z^P+5WE%7t7PmllCk#7eu}?-_HLwug^67j
zSC(2zU(^2dlHb10WPZ=yYQ_1r<6M0sJ0*z<MF+)rR4#bg^S;BE#vaZrNy;9~kSYza
zNGZ|qi$_<TEf%`qA(!o=M$&%!F5{HLVU6{UmR49myMtKc6;Rr86Q(K@TjkM#p><0O
z1JnYs&gAy*TF-He<ecergF;^>+w%#fFRiL^Xm(xwEmmX64UezP*@QF$oC^m_EE>0e
zzvxpL-;%0jaJN5;amAX==5y=Z8qjON$Phf|hf>6QFVOojvyAk^YPjQ#o0?du8(GAF
zK3y(f^LwS=f{-}uvrXT{y!J07<DqK>(VM5?Gn8=MQaiY=n0x(2<+3h`_J4-&r@ggV
za9ADPr?QBF3dF^bcY4|*OUWpJ3x!qB<=Kg>Q9v$(X0=Y6ePYn`v+t$HR`7nv-tRRX
z`!U+`)7)8j@gKm%;h758&lG-8F#pZR3HoV`Ux<VakS&0;9aCDf|NdK_9wWtKQ!Xd3
zTMM;VR&1nt+akVpWvP2Z;vQkR3Y3-I<?Z6rgxG+3ymXZDh<vh1>K`vm(nReec0=3=
zYM3tO0h1YFF>b5QC4{`djRm!_P400UR9f*;{#O>-cf+=kmuM`<TX^Q8!~W+J=W+wI
z0viBmzuq!0*3;Vd%B$aP5SORR$?!lw?6M-gG(0?GdnyV4-~E1tfSJ-|N}<a_ASiED
zz%d&fY(w+GJ_Z2FF|i-jLfH%dwNw5~`Q?eaEX>STFeQ7PSDO4G#e2{ub?C|bP^iQN
zN1azhq;bvr@r{30zVuwvi*k*v!V1TJU5pP7ZkpJR=UvCvZI;x0`Q0}#v6-oGs|LDE
zOJ`Z>%q0FKVJjHZqgc;{m`v=ge+-yze=n2PbAWR!QI}rmbW9}Upl>pPOtv$E><cmb
zCxRo+BM<E*`^Z=>jZZk(J$#G*fg>BQl9A-$#d1IRJt&tp+PpWKLSj*SA88#=d)-3^
zau#o>r7UA|UDe2qyxl|Gr5?6db2bC^Xm=NWyxsdwr_i5ChhBN`d%|tSawVH9RHJc|
z_hNceQ}8Z5WyB;SQ)qwD9b7x0oWBWQ?n?|>m!hO(_+j>b+YHamdwcsSa3C&)6cr08
zcu8<9%F4y!0(DX?`ery}#ED7lIyT~x876BrV?A3FXXIxX`D-czL;!#!YwI39GH)pH
zF#0NHAHKLHCs0|2{}Mb2j~q34=_Yo}suQ3QZ{EmH+PDGnHi8M^ydGYV(tW&F41x|v
zwwM3st2^suaC<UZ!+ilRQAD)F+EzO{_2x0j62e5Hi&j88+C+A9mlDVHLB8>1^c&yY
zrq$oIA6y1{=(j2K+!t8{2wZT#x%4Y=r?N97L|}?P8IcEHuXLdtjT``^TQjz^D~2Zg
zw3a5Y2o@l9heeE@i_T94b@HW*U-zO?Z9*cKMfPWU>&NZ`l@01?po-_Ha^?V2TY1V7
zXAaq9v68^~-CVbgtX1zf<fcD%6U^K?j>A`=+4R4he)}%imxk1M^U*~8UwX-Vh?ev}
zdCFqC=Re;iU0U1-69Wa2#g}=f{=oo9M$crU&F~}nnj4TwS)*M}n9G<yaV>lh|2%+S
zU$D1funoh0LA<>C+#`2^xu&%@(bt}@Fn=5LQJ?uIbEq$=6Z&!%;7P+h-`3jM>~wvY
z1!~4Ndz@m)iFFJ}SnuG0^6>h@lI$D+3ue2}T(r$C=M?GdkcQO*L460$IIn}!$7x0=
zO<8);Qog}=f1WuDj%G{-`>hpP7@Y5g=V6^qe4hSx9~Q$7M2->Uv=%OPJGw-nrXW?2
zMzK2#!2QG41}5ci>#qOWQhKsZQ{Fx6)NXT>bh|D2=W$~@5wUX2&a>_ruK1J29CG59
zZU<ZYd1+Z8n)g$qRSf~{OqO@AHPg4Y1b~UCI}SB7qe6r)L4;z+!eGrNJi95vwoSrw
zOW1d8>&zwspfK%l<ugX=(J%0sN*p_i_}*aoEQI)Us~TmKgM9}_2+|r=1`Ci5L_3ch
zS-kNhS6)m$IA$H(8Q<KDs)ioX-A*~BEy7Eo?uAZe&mL07W~dtJoHeMgo6boL7E97v
zwE2Xo7%&9CU-=eFe9f^vLJa@7#iRrV!_~jM3s|BX&pm263oCAd3Km+wriABzFH7DY
ztJSjCH9MX(M5Pv;r2!Z58bnf^Vq(>qkL2Z!(nyY}#Bn)L<zU2%t+Xy4#q3{k@c%eX
z2hKoVpYQ;ciuMAfXaR^2Spd_?T;WcyIK~wJE_j7+_^`?`IBwnc53g9`Cqb8Nkiki3
zG&KOTaK)l`eWRfzo#?`GxsTuZKk3NJrvdh+9O#OI$b6Sn6N^T5{Yj?Er9?Hj3OhI!
zSku};FT?49k8E3Z9uOwT&Xg=KmV0Fex(ajstIO^PFf`ESv&^YL<pNFOw}U2VWr*Kp
z;lQ|ycMBC0nu~1R{M7oFuE4zuw$RRs2B4lUAXc{gpONfUA_(0W#n}e&0oq@v%2_!`
z1&di}i%YkJ_=z2brhm)p-k|oT(6=G6j>U`D{>Gck=l}MUp+qb<xnIvj1q!F78`Ye_
zRi8F1a*Xj5oTtAx#MPGE6%XEc=;TVAa%BVP|D3V*%sKjRnofFJ|IZ14-&0jZJX=$}
z3)m*)fbN%olt)GR9ll|fb5EiUx}3eI@Ci*CL(INPg&NhUmDU$vo(%S#r9kCt^NXcE
zru}I+JjbBdL)b<Sm}@B3*u`8i9oWHKq0t{qYeeROEw!otJq@wfr9x6sbGODAFBjE0
z`&;zMT17g$LyZCA>MBe_GNCxWDb|xYF&V(E|6l>K*a7@pL=opX26DFXw9q3a<(iE1
ze&;>H*N@KU^7;Mn2=^k~3MrcC*+QQzvueEKc>k$)_;%_UYdG67>{8|AK~b8&z%FKp
z!^n(MKn~*HEnv0MY(kc(VfY^XLL&DcqXKKHvY|4tB?m|by-}rMKgpo%tGC7}>E_iv
z{i4I7f;}@Zus~3VgmpD6;Dot>!s2+9-Syz@PGzX7VE02CQ-s(&T{#AhKu>c#FH|UN
zoa7C+zw*<ZFQC@^_;E^l(JipK7p&9L%?93yxJz|OfBJDt+i2lcWdWD&g2N3aYoHOX
z0&ohaBs){D_EO=$@4twH&X_>ZBFKpah!Kcl1|wMk;ujnqWF`YCk-`$+=3#|K;~DYq
z_UEzhwD_3^*bo!vOBaNXzqC0_N^^yYV(aTbVbS?NPp0bE8y;084BIC~907<^LMY3z
zeIFt<9KjTiEuepLmF0omHs$n;r}ss9UiDC+;lhWxHSL=pN5N}?gYsy+P*lww27Nz(
zww?;?Zsl|OPK68|mWbpETF(x+<ag5-pdUwu!>+vWg?*exA^{k+30%|A4(PUkRcDii
zUy@7THAsdqcFY0ACR7bXEY}fx#-E<ec;g+-6%MyL|DWnu@kn5UUQR6o0H5JlO~V3W
zN&~$1m6>^ZOt}~FS9&7zK1fIevn~d7JIdQ#a40kr`aBlk#`yqo;@p>izO7&DGagaR
zwWBSq)g7IFv;qpLR7K#N%}Eal-c3Rn73XJn@!uki{bJnPlD-;V<lK^vDtFWJlc-b>
zLtWcMYyP2+^-enmeFiM>dZ<gqo4sn}e1g%27Nkc4D=UM92ev@BqhfGq-IW-!EaYh8
zmZXMX8}k|+AQWtISVxAdQ=}llR}1(bHE>kpf`q^S1^S8Gpa7k1c#~&fI1Xs24UhXl
zwpu*^$`BVT9!C0_)N|pb8n3P~z1NVSa&p&kZ)P&DPqYW9g<fJYHdB1o4f+OYmo*$(
zA7y=?u?^^%dGpLi{-d88)k~}b*;$oww3@PEUmK7TumK)OG)Za+bP3s>yTOSO!)||i
zVr;1t3$wbfH|yMg@!o7I39TX|SK!ibVe1w9^SIGl7grKM8>CYaViD&7rQ04ihj^>8
z+$_U$K>SSfPaNbAN1o<xPHjgoW=~2LGnv*tX!4zxM-P?Wdy=C&>f2)jH1^FDDAN;n
z=0KC$V>CCtJlgF!(yD#&xbgg4gc34av_hOZwiOCZWK>B_;%QstyRfh`_bI|}Ng(I>
zG{mA}>CLAX>{3!!Ajz6NH$)TlD#!OPsM=i9HSX~+vUI9$s0FXaxu_m{`DIe@W3e2=
zhLs=?S2dCgdvFs`v%!f}-*~5D>eY?A04G-@JlBO-HgJ2V52YzyM!9;v<Y{Pw%3Q$x
zrhW{RQ>3YlS4vBdydXr{)mc69^feCvC<^@XYzE`(-?WUebx%`W&_`)#Xa9mP5XqWN
ztGR3)@>qJr8%@IsFmtF)S|qJ+B1)fN9qR;yBBP3QuR?!QY;!fAy2+{AJl{+KF=CZd
z%+O?;UwBD_=gp+P+=-D7Sh4igM#JrysTXB40DscmHUHLa0`!1PRIk1_^9E%O7lw>}
zmvPT~Y$^vl(;=mp3z=oOAV3r_=vPo@F`so@jweH5OJ9DfP}6EE&iNN;?S6e}l(<7=
z3Pu8`#1r9Ll<*AyT<1~!#k+UVlJUTQ^tEAvWc^o8czRdC{JgXcotFD_Oq<}Ntu>0c
zoQ`zE7E@3fNhZ)C=<P$}0r96kF&D^1^P62A*+waGC8wvY$$A$ffs$qmlXb&wxjG6^
zPx>k4qUn>d=6oIfFlI<Wf&EO7vOC{8l$})z=X0RQaX+xIPUY175HN8OQ5-{-MCkJS
zEGR-;Dshdk&hQ_s{z%BqydD#|JJ=U#zvm;8!T`ihc{!vXbXW?&`avG70Va?1B`)+P
z1}{lbf-T$|<S7-^1|FlUP(YPsHHKZbPU|=>2|L}}_CeY1%14!$p$>ZIiF$X=U>VWG
z0m=0G`>SrE6%!xxfR_<isSbY#A~IIeeUlX9b&RQV4Eo{I>g!ZALdUCkAA#uTKhT3-
zzuW6guZz$<g`3PF09GICp>tta;VKbK6K&!r@^S-pssV0z;-G+od4CNuk6VtDxQ5tk
zfwBe+9ta%)T>b=uj`qS5hi}`!x0<p%z4<)bh-H6}zl&#7ynb<>8(P?Y9(a~f++K}2
zRFRQ{0O>V=w_i`J$PAq}jRHg~LojF{0}v?+?L|a|V5mrk$ev0+_va}+W1>Y|Oh81!
z&vJ72>&!fP8diNwzBoVD^CPP@TheLXHay4UGcp&=8o_MgUqL{I8b-Q)L0w)#L~kaR
zy5)DkyYTu$8SV{Gp{&9+c0<9aL$8lef+Z@-7$#`~%YdVHJz+}?FO2?pl>1-HVlRPc
za9ez)vaeyc)}K68J$8=SfDHsknO0Wl%a4uq#L6EZKat7_>-IbP#4?$|UJ2zHBW`lF
zO}h7sKR2kY8qYs5sBpi0*8=e*h{?lSh1|?9F`KvIKb=~>_Nj*&0a&@@<?(Dyd;AJN
z8_%M1aYAPI*Uyc~7mU`L0OX)$+t6fn4qOry&@-_1>#pdbF6|Eg|K!BqTbjK~9%5ad
z$Z#RVAq~e$%~mk53;^dX+-zKwHful|mel&b94l|8-pSM1^m7d5UIT5fMNZV!!;xJ2
zrxGPVlkLtsgREF!E$lJsq0GaqFjQWW7%0g`6!RT3<{fbNtN9%MABYG5p0lDstEGDz
zF0j4tQ-0?53Z2`&CqaLJhs9X0F)({@cA+oeE1Cq5PdL}ZYe4V&d5~F@n<xR1RKs&R
z-0y#i?EX?00K|6tr0EaGAl;$Y?CB=jEsuiC0<=U<>=OBlyDT1W=M1}aSXAu-OwDrO
zbUin*BRXRsc)yIFby{~!k9hZE0Y3|^Q|$4C|2#eYatkjArct*ajL>n7@yw3^^7kN$
z0+_^mSLWBR%eaWIyePF|J3HR!ZL|cx#?c+L@cW(fxZFdsQT33_*{C+=>`%-2+B$%;
z$u$6@DVOeQw*)grqEZWV6%8hLbOqAw<!A4<x2nWpoK=JARS&YCp`lwKwACy@R;&9}
zAo<nDwGi<hD<UM!-TuOWiY6D-E*aW-*e!Zb0fA7%E*Ar_PcLrEcX(xCc(QT%gr{<d
zkeKJZehTo2GuaTf+mbo_5}P>d`dQ%4<b(y2-mLCee(6gZvRJN91rF2^-IRc!rx{ot
z?acvri6M{6QTc@fU_^Ike9b&%ngM<-Yw5Kr7r?=m95|F5&`G50u^fJU-^bnM=Dz=&
z>U^$&TrqWSu)^7i`Fd7svSdj@@(P1uwr@Ng4+lq5!IT>Hvb?7(?hO|S=PLEDXn-M-
zW>A{K;vrhdO3e54*FL&nOak>o;A5Xa329{~QI*{FoM~;%%VnW|+yztXbZ0))(D`9X
zY{IJVWWeHMkhY!d(IIg(xohe)AuWU&f>OW=B+WOK%I#hE&X}T`bJl>Wmt&}{MiM}r
zxe!E=8UD-c<Mr-OagQNpX%j~XouG0JQQs!0X4COXjJXo;Txs(QFva1sRGa~<WO~iN
zCFkfCBPY9@cxHizmp<?1$h=c41NQ-x5o8i}%cK?W05Rhym#B*t%l@_)!eZ>Cy;}@`
zC5r}EK09jcJp4@Uwg+~Dsgxj_WC{~~Lj5Rsh904Y{N?vw$nUlhcj`fDnDR5l#dxC}
zuQZ;9uPc*L2>iL@F@!b#`eUqq!e=vRd~GS)ueG(Q<Yja;FgO>8Anfg-ZI05vLB(H5
zOUK2cFL1nO@ca+JwdFQf=_?_Ei>q=Bq7|R6VArve@|Zy*zZvjde6vL_v9aAay~QK?
zC90@rMDA^dWJZPGPkYxC(i1G8d^aj@P3rSch3W2Inc&%*N1FaV+m&N_eaq;hJT;^~
z(p7YjdAV~hC3>V?49Qh|ra!Y&G&q@mdxwJ8FFH&b??lz_v%Xw^=EB~tvk78mC6~r~
zgp=nk$bm*O*y4q|OV5ik0>4Ls&dusT#`dR6)OkX3!gu|QCSn-z#pYnThrPMigd--b
zD+9n;&~kU_3JVJdsM1IJeB<(dEj&llw;0&?Pa`pw-#$McS;NBv7oUV=06c#Av~W(v
zT5=lykB%WnDwMm&-xlOE%J@iWA0c$)lfNT%Dw8qC0NBV-ol<m4ph$6Qh!L?IM-Cry
zYJvJg`>WH|Ss_Bu-TXgR`^dwACb<CJ;MGu$*}UvniH3c8N#a+5h6^QDJ=o%%B_$@?
zyKIwBiqM#C6II!@H>Rn5fP!Ikr8>~3G7NB*#>~Gl&(T>eyc%5p_X$1k6l^H99gNR{
zw4YnbHrbxR_cL|$1poGNqw3L``31>bg=2h&9|i607Se!>s0cBbK?C@PbRv}J^#~B#
zPn?Yv2KM{jO(?~kO2x6(I8f~t%TDIFB!aDkT}j0YxEw8v^?1;~qK*fQNe5J5EP6v6
zHZsN=I6#OJPDEJ-Qcn{+AHKm(c9W{*x1R4*3in$y%=yW=_uNlXy#T<1fSjdYchf&+
zS5;+wgb3-W=k3vQ_{s=Q;X(&%{yhK|yA8_N7_XJ2BF0>}3s=DWqLa&-EUz<-yWHBz
z&9V3M^obXt#vJCwu|p2^f!O8dl_XUiEgxtlzc2{g>^@ww);f7CdE6By3&nh5?Ba!9
z`{si3coJ#wrxupYo_Bj6fh?S*FJ*!3tM<#omN=lW=$Dl!q2rzYUTUO2G24+dAuAco
z!#(>mE$Lxx3cs6*aS<is<<p)^qa#a6@&RssT{DGmn*efB`%R9ApKN@_wYJyl{{j3}
zEZ=br4WMIwTSh;A>3F+zqScB$2-Iw-O>ga<b0BbA%0@^o_|EVEjPEmdz2w#7y5qN8
z6W7M%2WZRZipU>h`z}NQEgn1*gD{xb=L?$mFNIYkrT<w{*ly4BP2gbtGpF>u=ODQ)
zM)=xQqU0=n85F_+Vg-Q2VPCz*ptiby!P~R#FCC#&{iTp5scTdA8}0ngD&59nUPi6E
zK8etf*&!OZu+s?*Gb{rMOyKM<_h%i~bVFkXHl8VKVV}=QEF3d#MGOIe>->{vYaF*R
z<!2_Lol;f{psb=+J;0KjI(u#!fa(V5{V>lIU^y=Go_+3<NXFlq%5*(TLni)mJ?pzK
zRSd*(0eZdivz(zI?tx;AxCX#2?auO-3{d%BD}h6vep>_8zRAgdIL-DvX!$Q&ry&~R
zr!;~ccJK6?d?*u!q0CJ8bsrfZHuS9yKtSL%Gnm`4hPu@FH_i7GsnJ8w?!CkkzcgU3
zKswN=!|~w(7YRg0@z-m}J`=O79~79y<Rn2-5ZBk7x$GM6XpM4+@oc;mwjq+(YhH3r
zy9S|xJHE%6$4ahugq2VjLnQ_`_vbBp{9EbEy*w|n4}(IXmJHpM5#DZX1F-}hg(uLa
z^>$VAI<Jt=+)FTj_@`vKZ2}lF2_&_}WnR{i>lA{}=0Po9XjOb=AZNOTLB|O=^?u}6
zK{+mA1|;D;4INOEF1>IDsp=@?xc=Bh>Q6z!a7r_`!f)mzM{j%C!=MQ^L+5^u-Ou78
zw(zb<|5O7URJ=T$dA%xY{WCsdxx(A3+Uvjmw%4<@invZwZP9<G(CB2Lv(|D_3H?do
z55JX{+5jWGK@(LJWCo(6FwR2m96%ve{hZ%Id&gTDPY$y)P0rYx_~o{wpi5)hbGs*#
z8j_f7(@lMhCC-|}Fqpl=l?43or+j|SyI!^nFz28PeUH$xz1THb3iL&^KrN%#v^rZM
zxGsVBfVzDLJtW=c7w1B$M{jiKPt4W0U`oCc4lA+uu~Y}NF?DkNbJAVXcvRx)g%XQx
z@_p!~O{WFGs0+xH0CEjrg_R9jJ%sAA1^J#yRaSmJfO_X~s0?{Dt`5>}lkLapHqO6G
zl0obIjRe3zVp6@jl13djo{2F?wq2s5fj9oDN_*{jej}%8-!GHIO_Y6U);QPyNqI_6
zGQo6jccX+M?Z4E?j8`YREd+vFJ~YJ|Ao_s^=CA(2(g_yfrZS(`4?z0V^LUYmQw{Ti
zWrhyj+}lR8IJq}Vj_W;#AA${OZe>(Z)`NqTC#6hjzM55)nQ-{bXk{LE^^!S=m2BqS
zZI;>ISaE^-{^nUt@hINJBV*p;!DZ*;AqOF$s;#ZkB(|RM>Uc$&Ny~WX5pxuDue_Vt
z-dbR;Dy*Ao)YWajR!5)q^j_4LTx!z^UjKMp&eK^vsWw%ldnH6UuNvqTMn`vBTf<$g
zej(nUL2<8tk@>oR(x9y=4UDdr0U$3@oz47vJIbeQD^He#Q13+{Xf`TNk_%Xz27=#8
zv-|v=(f|fE4RHj_2?s8}v|C-PB7Y%gN=}}0j2KU+H`^k(OOa@k3?cw@qy5ocLookH
zE^P_po45b-`WdHE!gf$T^2(DO>Y=Em*zE^=Q-g;*aBi#TdLc1qFfa~?6Tko>J&tE~
zRm=NV+fnLNRphja5rY>5*r<M0OK2@1@s#-{Z!>*4r3JIM|FSZmh6!b}f_3iMgBJdu
z+q23MRs6-XVMq04i-IRr)dV~r%Yk6tX`<35M~BWP7QYUborT(+O80MUrQ%&$m_7tE
zuY!c7!6RS8(D~O1<BY(qTZ(cIoqCymuqT~XXp_$1f1$<&$1+;WfqDaDyVC-JoU(Sz
z*ui?;-9*XGHsWX`DK;;#QrX_U_0e5$QU){bal|YYf^yzwne(mtP<tBzJ88Dd9zc->
ze^&T)<%J1WBG6^=INn2hV8hL%V}PHa&d_<UKS*hF_$n!P#;3$}t7KT*{D1G0Yrym~
zM#`EHTbqt|RW<Rb3&uFnWsx?}1bFpDC!xy}S)*A_Qb70OccZD^dEHlE=_~er4^$a=
z{%0@n%l<Q?TAfiSHLJic(*Y27OKaie*JL0|^}gDonh_*w)Y3r|G6z7P9o#KI;ADH%
zjJy*MvySJZDNnF>T8el%EGQ`O;v!?WogyD++hVuITVL&7X;7;8T^}_419?q%Na5GP
zet$rLYavL<tsZ)mQmt0rO>Sp#I)ny%%-T-BowvWSqR^=W`u^1Wo%PMs=qg+V;EGoV
zS_Q9ljQKACf}m{XM~xGV2xODr`Z}eZwVE_8RIBc7R#E%G>i^O8<?&F4-`iTGP1+Sn
zg-9V;v$t3V*|TP;v5zqcS;m@D%ATy*vM+-$_N7wRu?<GHk%SNuV;S4;jQW1x_w&A=
zUw_r7Kjwa(=f2N<&ULPHU2@ZX{p>Bnv=Q3md~b4}6$e^D&wf&~d!LW2m*urZG!+IW
z5?Gbrrib0tvVgWuad~OSqNn{e<8JF|CJ(A#F0!c}o;EdFh-9AbD*JHa3$&Y=2hZi;
z;jLtQf8Wl-bc}$w@to^4-)@zl)PbCGLEyNC^Bct88=M{O8T`l4@p<w*IB1f2Yx2Ij
z^{li&rLK#K@sQa>+O2@~8Dxr3_{mSIE%;x!`gu)^25qbZ;~28}{Y~#AUwc0na_{7Q
z3loo-$n_b>d7OW*>ts$ZgUpj+ztd{^?0OY@JfjQNtNiU7q1|{zYjX<7yy{mFX@p((
z#}{xk87!++1yY!WVJJ7CP;1{u7z}9B#OasqvgD`BY*q3r-uAk00Sh~LJ9GJ@In8m4
ziJC6Pg0#ndUm~+y#dx16{g*?Vb$(pXc5iupRzt$b+ISr=$={2@l&2C>J~>Q=0z){w
zJEO*1xqapJ+G3z=xjNI+Gb`dU-x&*MPX$Jmach+84OiW=-&{?s!CvU5U|9{AYk0qw
zW{JtpedwDKEwpq5?lhM5KBo3|H)CPj<%x^ly_X;uUq049IA~-oEF&4FOGAvtuy(l|
z-)(G+OTwH*SF~m~?j9;k-vxcYiO<8ruSnf+&5p%k|8z}U{BHnGfmoHlhp^UaYu9+H
z>+&RXamS}A>I0Gg=*1^}`*qt8v+}_!i2K0(L_lu?yvCkkQH7qRVH9Wv%&n{}Ti}4V
zJ1LzAOGm>6LL0Oh5i-M<FmWzcwTS)5*rJTQP7rOzgWS0?@r}{`kC~8x3rp%%xAa*}
zR6n0h0mCmURPeU~nhu+8S-v6Y^fwxcC!5(mR%1@Ll5RN`|0S_&SHR87(~Z|=hPsMC
zs&^ycw_mQk85r=z#<aMRTi#!TIG}>bEy#akyO;L0d@Qx#Yb_&K464vv&<PZ(uPWX@
zP+>&;@Eu)LRt;1Aia7;*+lLM0zhE$-dVY38Vn+DskTb<{<D*J`m>}N9cyJO{=DGFR
zHFE%{Dp3b70_jvgCb5D-T%QTh(KU5lRJEQ?aADEp2z$0XCX+FOk+E|AxvpwGun-pq
zTk)tlW%|#v?^eoBxqDyhq}|ZX0J_H^)958t2I<EIIUvYk#NTLmLHsLo2wax0Top&E
z!<yZP#Wu}Q<rBMcY23Oi9y~pV|IU-9rYrjA;vK`u8PCkJboLf`b)0h4)*le8Uk^x)
z$|(z?<k$88OR;?(O#Z4OQ%4w_zYRi~!&kY70w|D47>kL-Ab4n$9NTT6$<;nwdLzzV
z61ub3g<1M50H|g)$p1PxUEUtjrsiKFEei((oBBABsrAAL%ABufXGKC8?(@8Lxq>bX
zYIZ!k$s<WVvTxlJCcqV=_JOh-L^;+j+j72P`%p1!3C(9SDS~c(xqlVlti34WUskKb
z2UF0xTF{O<jWnr^Ji&_e529%Xl<Mq#&yRmE!#|mV`j1(H4{_X4m$to%x2#8g51)W;
zJrkxZYdlJlOk~y*){3i31l_?yPjdHaVpHz`n$oV$etKX2d1rF@nN_V!%q{f{;J@6j
z*=3dS@+@;C&?OBD{m!9|ENd+j|BN(KKA_L>K`7U&2*c+HtvFoXU$|#C1CkBhU!Cm4
zx?)?t)viOTG-fgA%}>x*1jeqJ2yM3rzmBEpxTl1B-Fl`=N=vf2NqinCI4n|MRe!3>
zcv1ptRN!}UOctDs)!$I8dKB(OHk`&-|7$iM-igAIH?kBdX$@^-6X@xX?ONM&KKNQl
zE?ny{_2}9(W@EZyg}>Nk5X5yLY7vUJ`vZ{u)U%cAL*iE1rB;q{RNq`W)&n+4H1Q;l
zGgkMv=I4qs?E515hV^FGg6oc)8(h`&Iy1&|{>RftCk=TbL}itYf^Q}l59&1`lI8Xv
z-y1>a3*t$^qn`uf>w@!FBFvjeIO)V;_QJ=BF+*>3iDwOGCN|tx$Kv~9;uV){(_&WJ
zXs*V-wu|-zw8}u$eL|k^d)^@P*@vu3+$dRefPeSVR#DyA$ySqx-$dV5z)i4LKFh3<
zeq7zs?<==+$<L~u1B>%|@wgIy;#QTT!0wn8s>{;BNilUNR3o#Q9aKH`WpkLVwe1sW
ztGQWCxmH9XgDQs(G42go4bQ@qWRg2S-W@Gy8JF6yQ1zLsJ~{+!iHZKB-YR``IkiEQ
z#9aJx<obbs7{ECMb<kLHab_*CO;x~p|1@b@?8ol+yf(&)MH7#Sc?lo&5&hbUk3VUK
z7nk^*yR3}zR1|IO7dq9Yab<Gf)m9xMh(j@kC6wgesrAm*3t#-4>}!C@2)*4r(8%&G
zOr+scdcmg~DWB&%FuiiQ%^cY^?{6=83f*1%S-7-Ex<@)fs5VRPYZqhT(xDTZrDucP
z1fIHXsi<|1S@9)0T`B53)#T+J{8`)m`)?nU`DeqI6O>W$b5?4}`HTx8YmPpD*)j_^
zL<AhX;!+33(CW^+G8x19Bb?HNFyH5*R4Q%|SvlifJ;=AMdxk5hBB+@LWog%c@ocol
zdCSCeNl)@x7vqSJeZ|s)_-=#6J5kLK(265QQZ7u9>3t7P6KPW;mzB{Q59Zcf)dLI-
zUP15tY~Y)j$4mC1y5SdYK<*zXIAzyra<IdA0_2Ew4ClLrpXAc>r&{wL-VLffyO2$)
zU#S6gyQp7E>S8X=eH6qPp;U^iaHF%Xir+5{ZJ4W%58n@&d+;)A^UhW8Vo~#9=6M~%
z_vSXS<FwC|WSITfywNa@<a(IW1<6*0Z?{L(ebzC5J=~ldf*Orqisw2b`tP9{;s*{*
z&uv+uEG14|rT%%t{PT&(fgJH{hv}2dhu$n5glGC~0|Up5^u?)MhT*;R7k}(Fz^XuE
zmR{uhAqg^1PlGtCa_&P>8E<;3MJhR;?JXR~=Ss8I|H@uCH*NiiwHp=v&a+s9!!?;i
ze0JwZ`(kHUS;;TS^cLAQ$!ba*(-gZ`@f|JuH4BwnZWynxQ#NE(YcpZZhXzk_DJeb6
zHLUO6(9{&<f^w$9&P&evg_~Tl$tWD@CR}J9E;O{R3C^!oB#x2#nWEeY{~oGdx*$V+
zZZrS+kCT_ZzN?=Xlm!aHUoJp9>42J%VfBsc7cu7X5{=YvWSODI!)MuIew^&G4|S%z
zyGXxt@GM&zK2;>!lH0YkGNURems<cGwkGQJ=eHT6WMTrG{pPD*_YzC&#A{GV!aLRt
z$CwJ$PW0a+n}8X>H85%6WMfnNnLa$%rt0m|CByh7oaH4H9`r0GDqlP|lJy@E2O7oH
zj?8{pAHnQQ_~p5Mw(w?Z_rBZVs#S=S;p8heMO8sLF<&BNtETuXRfwhtiB`k4mTOGg
z>_sTZjnvPc?T*8^edx+w@r_Ez?lmp0c$xCYurd>DE0&M@7HoMvtTkD9v$i<Gj_$&4
zPA5NUfpS~qwH;IRC1fD!Ixe?T1%zCcEST^Netoy*8Ap=1ch^*-GC2npRMO#@@3R;z
zOxZcJ7%)s%hB;rE^=wxKzb-=<;W^$~n;_fRFLtIE$xt2i*?X`ruFIoOcUFPqZ)YYL
zR`}ZF9(XTqGy<^jSAareV4M(k9))f;#1`R~E+8@x?)j+$T<9p_ix{&WYpD*PwB&Pw
zDv%Hk&uxT~#o&VUMp6I1<<v)CWJ<+&dJxOQPg1=ysCcqlU%=fnck|vHV3jUab#Ycd
zx_6*(E}HDU7D-rSf4nrMv1LuR*jM=O=-|2O1dg%oVHQcFYauR9ZvfX~Br0q1fL^AI
z%0S_LLe|ce1!Fai>EkmgEd!R2>%mjm4o^^}CEbNjskn~gBd#}2d2je_ta^LNKiTnY
z@6%q!t+fYcS2WEKv;}4|jtx~Nz?9tMAO?L|F~N3_Vah^`BG0~dA{e%eJ^T9*a`bGh
z0n@Q@qEo-+M3ok}C$989H}&!lj4`@V{P1d0&jD6^h70FJ`RvDa--6Asr{twp2FG#C
z;82EMLMc&5AVzumL#N#ba20JH+YKY{w{MWlB?jzIA9St5WL!b_2%Mb`=S0u3MBqz)
z3=hvPaa}rd&L^s!SYkas{~0~g;(2+koEO@G@(@RF2IbQ4ESL$pPah~8Ha`pN`ud_&
zY%F0$Iw8sBLqsio&v2F7PyNjUmsvh}F!j8Ct3VP|i=jaWnP`zBJrs_ufONKmG+RnK
zd;ar3cU2vdwN%<P#!XJ$*$N}FnR`rf5tA>g=ex5tqh~4ID)$X1{VoNoE0`97K)2nB
z-}5Ex&5C29v$HKRE}7n{h>V_acrpuPxGLP8IFmQS`o!k$^fy)rvv#wdQaO3ht6c%o
zaF#=@t?i+<F!p$2rg>piM?hh;3uR8(8n|PPIye!}e()5Zb#qDK662R1$zM<cOtN(%
zpYOTW#n-9PC-B`m$h~1eC}L~PA={gYAh5aoMI*2)o%VTbjavUs2EpHT)UUlEnR;N0
zL}fj%2+akKp#jim=4FYk9EmBFn%KQD<=Nh|Ay%a!AW<zORI71yuBmPnlac^k?ELnD
z#fo&e^O}+IU4Vd8Vy%Xor~C{2K#}{Bijj2JK;<8AQ-XpNY)Hn=#@Hypta@BRo$X@5
z!LId){4Mfw&y)0m&31#9{dq;1*ICV_C64M1Ye-qeB+5MESt!Gw_vR^yST7AN72SGt
zCyZ)0h0}w3Xrc%k`uAQF3Ff#(dm#985g-vpSq~i->CX6=Qg_W1`|{22FQ2+vPN%l%
z^4SO226VOcN}kasGN-=YKV0IexE$*4$>*vLwi`LjN6qEk-hz?D+0(oQ<_<)g>3wz2
z{K)5~_o>86@5q`lE*dbAf#Dgr$QC9S@h?x>zRqJvsR6M=*FW0#e0@6aHW-z2B>UuH
zmn1g6SGV`<Q!#`mf^Z-`D4YIKiZ5!HnpuZ8cKH#{V^&7(-B`g|m<gR@{5wk5Ey3UZ
zPY(NPR{Q#c%9>IeZ%cPg$J?dG-T@Lu3&SODd4|Mrv}m#J6l?Ir)6YxdccR**Edply
z$gg9(%-S`LXN6ZG*0qs%N}Uz@uGj@<Rtx5Cc0T2_Sq=-kM2dbpP|u6FVajSNe7w6q
z_~Oje=`F9XnjoKN-VeJh)rBLGWUUl5Vs`PfK|74!rUK4Fv6_8u;^j{+xE~Z*tk6Ri
zr#xeZG+A3(o1x6;-I#2v?k`c6jJ-<_Hf;j@o$SolCqO!!_91MI<(F-P1igW}N<P)4
z-!$_NgNu$a+A2^qU2oNnnM_yG5HrQTC#|tuK;6igJ2^DuN4BgBbJa3d8yHb(zn3kV
zF=ID6E7Sf58-H#(R=)WCyDE~i`7m-Vq>aB3;CO>uzJsFO+7d=WA*E60A*kPx-V<2L
z&+jZ#Sus@x!Fz;JR`KOQ*~IQXEgjlOx?$lB(oM#pyvY4P7rW*Aj^$O{1*hI_R<R#a
z;*pu(xBHZ4OT3<L<2K=&w}|JF1dxcG&~^QrN7B)~T<lUjG|BE=$S1_M|CHz09p<UQ
zo&(w9*)h|06b6IpR(~wEdgQ~glcm;mRQZ_6AgsG|I~Y5^-2W)<0)v7Tv;Lpb<UI@Y
z22mD%xsnzB=D1b^OI@UrTf^b>F#NgF>lLViU{-@;ePF$HHA(LDr}oPy&9VCHzH=TR
zIw9e8U^UCH?2B;s11}p#GV?9Le_N--_RQ&P;r*P|w}_TVf@gg+%Esnji7+0#1~)_X
zu&2KL<C(5}NyT<a3j}44h>8k~?u@qjKVNGo&aqO$ddy&_Pl~xc$u(j+Xg%$_Nm%Rx
zeNK@Zx)BC(T|9lVE$2|0LAfe$n=l{|Z2a?lGsoIa1vas#e<`z23&mG{-eSUYMiTrJ
zQ%3&0B^%-&x@Tm++!q~rQ_#1s9;8{^M@4rs1^?$-%ytea>Tyl#fe!wp0)hdgy2R?=
z8m9^VOSVH7zkb|NO)0taQZgt&tm>C=HFf>&-7`JE<Zs=8Smz&H!GNTaHn5bp>Kldv
zD+CNhUDoZ&!o6GKNohDn_omY=QadU`HWmd9P(8A-9q+^Jq939e;8~)us72q-KKjUT
zIJTgDmWE51oH4#t)SqtpX&)D{UZOk=-FMilcb?8mnneXB^)=jX#$9Iq!BnX5#J8#)
zFB(w`JxKn>e%c_h!?5?QzX7zn%=Ro>;h#N*16+{CfIPD-%#(@M3#%^u+x2wHDAVB=
zdF5^?d*Y3{YsQEa_i0D<-0zvn8C^CO)f*o@sot`K*|%+_x+g@{@}{kI&zTk5I~3)8
zEszk}3aO+nwlq{Y4Y!%JPv<3vov;`u->mo%l=~4b`ifvnb?xcb&lxQ7cWTZ1_Z7lZ
zWTmNbpn5+<=f+bbmlrgz#579<`!_qfqbz(CNb;>m;tNu*TX_lY=+ff3cvS^b7I>7|
zx$TaJ#w%Vnb2h2XQmjrhuUNNxKx&)_nMeiM;g}O_T(px7q7cvBpFf~wCR3*gqMOxI
zRVOM&EHd;QbkfH5d|ILeoxInE*5Alsj{m;z)%rsF+1}o{zL*%3ug6ISmj_g&v_7-|
zf+bpc4h2n)*{ml@JV~!%ss+cO4D%y<LshiE0}kOVk!bYZ+?^Cuh36eFUJ-rT0`q=$
zXbo}o0xZNtZho7l7}+Q=9tG{%_>&W!g5oWS+YCqz^a*JRbKHg)f?1i$gt*nbM^ETg
zBdfLXxk&*|=?8)3y{8r~B2~AUeDEyFwwwvwlu~<Q9!Fh_Tl;ml_cD}z#}m<t=zx52
z7kSMVoY7k}*=X=+k7QyZI(4W^@?EPsdy8uAe7MN&7aETtb>>Xay6`5!e6lG`t44ve
zb9mGWXWb%2b}#U*A86s83_Nf8@UVf3QFVs=;L=a`Yms0l+_DhE3fDHK9c?jV9q{vW
z+u%7A?Q>W=v&0;*0lLy`*`1+Ve|6U~9Zu8|`V7}X@k{7%T_yL*REE3a9aH{>iI*mI
zgG!4xD+R{`Ump4=g%h1|G^?1rifipJ6ku(W!dg_;{7FA>;!@X4T+Wy?GmhjkOg*?U
z?5M|c^x*M>o4Iqp$v3G~tNdPs+fCJExWn4@^@5Xko<#?dKiu2;Nf=lTx54Zc)#quP
z5@wu}xpt436d~!ak)WV&jx+2ACGqbMZ3Y}r&!CLu(}n^7cP|vUQy6n}TZ}0%eFa<d
zgYIc)x(qIV>Co3A^&=v;(X?`$t>W`sd>ss}7@-6roqSXCbIKw~&%y!f&_oiX>!UCK
zH-cH#8|E;1Z5^SNf0kspw}w21zMQDEc$8-R=;rg9hhJ9vsp<kf=}gs1?4DyW*IROd
zAIkdahUZdyTZ>Kb`fJBXlmXVy=Mf6ED0=M>;X)%!*7h}4TpNH1ti_dNB0bMBI4-nJ
zhqK0Xe2@HP6Z-tE!cTYb%w%l_eH+AO=9Bvi$4v+wN8>9l;g1`>+)EGzeKwcD;jVC-
zocEszaTVL;XaD$Ti#AgKa~vMi>#8z6z=_s!;KQ{?I>pgMQrb@;)pCA*_#5JUSeBa-
zWs}+MWI|(129I{4dojPCrjO^qSx8q>5|BXJv^h>QPl(RN;|q=Fu=807ZoSf2&%Szw
zvVM=!lZnP)h}meE8Ae$gLqAG%ck<o}ES-!}ve=~;_U*0sT|p}la@&lhkC}YF)MFLu
zd?l<|D$j`nrRaD&9#<~Q{AxQgx8~o0)58X7{Z-M0|C}Djsh=J5k)*8xl6?m#f-@ti
zmk|RX>>^Ts?|xu_&f~@H5CC#V<@4krYCp!sB0Y}d#?Yf7|F@#VReLwAy|s;v(dM~1
zqeScz@}o(2oZ;8Z+VXXs_)_;9<S`#D6abmQTvqFbfbFiTpAeS6nQ*ch$DjfV#u7As
zMh46@vrB%>PpNZ&PhMKVP!IQObG9X{UPY(=M~(XS;!#H!Cxd9K_YQS`cv5zIBs{5q
z9N3d%#l`$lpq9&&2|G%KYa^rH$WRs?cs=diu7BXV?T3)FGD#uid3{TXGs3h)cQW_Q
z43t)s!QX|yS`!kSA1IPPgbS1JttLk$CSzq+c2+MGzUCE?l0yf*!GM(es_NIrW@hyR
ziz?zsp5v+Vq}3c|%d^~Cx~!Y4dKF9K42z%I*L6MX=6Qp2A3Zo>Yi<k`n6u`BtIJDS
z4MWur#m8UTc%gv-KD!cE@%-=fPUd~b&7Yz8$Da+?7*R7^;cZciA_%|OiZuK!B8<|2
z_Ir@vczd8ASz6-nCrCOpR5|(r%D<ECRqa)qQpWCozVilUQX&6z-&l1dO{=#V<b_oI
zA=ol8G<LEuHnr+8ZSVdJ>8{rz?<#JP*oORAf-?5=OR_;E26vWA@kQCj_GIxe7N(s{
z_2JZ^-~`^F?815|g{(6@GD`o57M^7)%rpItz9MkZ%y2)hGfN!<y&Z2uj~i3Tr~Xp7
z@F<YF1IY|NQooCmVKO+m%IFbGX_v?)B%$d9oA}?=Usbe|?#M6obkx4Mn+w72wR+n9
zOZjE^)Kdo-c2T{(ANkiT)*o=vvO?jX@A|QKcVNbbgKki6mJPIiA-t*?P0rV)*;qI?
z9Ntgr&U%?`9qjKk1rvPj&P<<qrDCTb%Ca_pm8qzz*4)&Z4NxG3)cnaMRCPt-b#v*#
zul2{Mx9P#)R!CTB>(Mc~bC&H7zBqw<GUrc8vD%pjzZndfRj4S0`YGfrsGawp(tf@5
z4-Ws}G&93S!i!<L3%W8_mL}perv;2nzi95@(Zy<)ym5f8coR(<E0(j-y?#H&_0fRu
z=4njtn}_vR`<B9-;as{+G-yv&Elk9@J_FClWFHBmlms`n*3qtsJd>n$Q-_*7Pm+Y^
ze9%P8%CUTl?-(-SxdMsW6}!#8f#u50WUn61z%LyEw*cCkJ&6dpuCBgYe*s@?0&2)y
zGlKP1!>3+j^Z!XKnkD=^9XL4!b$FwOl+c_Ab1soVKo<WpW#?74nIK5?PPSq<fe#AG
zv5deLMm4IkpFI^f|5zIN@KaRSHUFa%uNU+tKnwC>_dS)r+o3AC_0568jKG^aQcCLW
z%zmGk$m@`;`A&FQ;!-N|G~*tW71U#l^d}rxaDUk%nHCF^KRbjtt8WvVSeSAv22&nx
z%1!ekwl3I|3XaF0IK};UV`ymJdY_tCX_2zD6l^{&z17V{{ef!B(~mpj)8|-cw^6iL
zAe~6w)2?CmN1ML2(@xI*vyD?i5VVUy9!8uw*Erf`Q!f5#?*j<xx5}zAN_F||nx#y>
zp2?!ND{t}rSV#`T%aqRrCF^^7Y|}WR6Ky3x5YE63+z<sVD^pFC-Q!aWE@B?UfAh|R
zN-<Dy`4m$3?rhHA_0ctRC#X)(40~ums%$<EVa>P3X^eh91B&|loQ8`O*ry0FUE$LH
z0_`a+ZBEt*7CwVe5lR2UiSqK-?>iLn%Q_mgt@jl#kiwNk3ll1{r59Epidzin3L3tX
zr}eOLD!9LSCE=L%+-e)RSu!iR$EE7=vfhi=$PA){gg;iMY9oSb#&U~(rG<N01-jMX
zAFJCaZj<z2Gpk9UDVqbOKgk*J2IC9L48)AF@42J*B#A}<84G*LGh0EYbkYH6cUi?f
zGjU6klQl<-mw!<Vi7c&BQBH>U>!8Mcbw99EkL=_N2^}DnRd;LM3Kg|B2GyYB6BJn<
zMrl?d*;=)SR_u8|{Wg~I#@Xo0i>EtZ`tMBe_FJkim~aG%O3z2P+Wef4+S20gRoNaw
zr!A6j6w!-tf0cTS+md~2fZZhC)n1=i%H`Q-SUI*bU8O?2%fj?2Va&+;UA!|cb%0a4
zwAl1tDWh>3Y%bTWLZ%yx4q1*nmnb+-Q&@{LuWNk?+Yz0$bP*!#N#8Ue*0a35pwuty
z2Oz9eS}k?<R)wQc6z62v#s;;$^j^3dxvP-{8U*_Fr*M<j#CPL%Jg#OZlgACiRMVvA
z2K=vy6^bf6p?*$G&?cc(^-#Bz-Y2_SQnxmmCaK-+i5+3dWhUO8y!7dNQ;}YO_83fa
zbkp9unO7_*T_Xvg0C%?wW$LL41DivZI5sEd0@5KRYIv%^X=H-H<Q$iAy4?%s?{ikH
zNE18Uh6SSb+~0kbw6n&moqx!6ZkFJL5>7SBA5Q9d6`L*2OIOr6K6tYtA}Dv{(vwt3
zIEBL)l>u9&j(cKH9Ac@SJdNskSasyl!z;--_Jf(vm)TmbR8q`pCq!On*%g^XVq@co
z;WlC73$#C$r2{kg<(c^Nxg2UB=Ev1qG3^;O`Qg20d`%VpCC4oh5%D`!n{(UbfJ`pO
zYmHOI`J3gb$1cLxs?=Dsy-8%^^!#FoRTmt1Puf%<VgdzW+LIB%S0kSndiy`7?;>Li
zam&-kw$^T2N*q1-b6U?dwJWOhE{xAu=2UlrO4A*l_!1%8`ve>F`0pYtb$!}k3^J`Z
z>!oZ;ZB;{PNxdOS3P>Wtod!Iy(&s^YRrEWfQ3dkF?rDJtaU=9BHu;d@DaR+d$y986
zS!>*A*L}R$GwD06nX<FXH{YJ4jYIgbh-NG;1_=sWtM^bR${;`s`d$64>|=MbW;^i#
zt81K|N<>@K%!x1l`fyi4-KI+}(c`eGQ?2=Th>{Fvy$X%I7luR3(}Tif*8CS13CAR9
zpCj#?KXaEZOK|&@57EllI<@3WPjbN!5@5z7V7#ZkvED=&_4MR$Gi_irC=Y>0tSPBL
zncs0sPf<HyQT-c+jC4)9^Ise@r+6pQKD``Y7S$wAFkO3eCI_QAKa-<NZ0GWnC$f?H
zJzDO7&{$M`bPx+ue*frz!S#XOf2V=hqNN7uK2a))jXfm$=cN#(uqX;jFgoQgj1YIW
ztZ(8NyD&&6S&qQY2wa{$ja<r?iV;t6*|e5=v(LC3xtD%iXF}nQwd3Y+!%>?ruWi(m
z&^4!3U`CcHujH`#vs7!IwN=%><}`-7cxg}5wJF)0vZWpqL}H{<IDW~3)Snn};q%mr
zAEpBTG#?09K5mVMIUz&y(I&rRDk<?n>UvnX?URB5X(Bq80;w^kdX1iYGTSOIhnz$g
zp;IZ)C~;MyOF>ucwuCdQEiw#D97R7MIaXP34(;hP=VdM150xT%Q)AQvW6=u!IX(9|
za%3A>fK~}gN&nJ~oOiTi_W(&4W<e%|J7uKyD(!D!HECrwt5zr84c9e5$y(!vjML9A
ztv<h(J_9e@Ea7YMa%PRgoINazB)joLknxM<;|sv#HWP(((}{A2RhsVwrpy&jUjDAV
z{mDtBP%s5K)}*s^)uZ6Wxf~ec_DJM5T^Qa>NeNG_34?AX94L{I!1Kp*Yah4v{V82I
z6xcfF*%_T5hChGLdoG4#En3!pl<`hqY9RGn$+51#<f6#C?&@2{;n1nxeZt<fXMVLV
zyyoJh6l$gjhaWB^jmuK+N~GDjz&Dp;b2az&9-p_qkmlE*kRuKT5gH+T>W@~8k274`
zaSh?#IjKT<uNMzAmU6m<0GI<w2b(UA>F$huv2Vv4I&UkJ4B1+b>3UXRZihSNf{mPb
zB)Ug69cCGKx^?Mdf^w1jSPYm&6yqDhc!PI4B0#&9wmTPYi|`x7XM0iv5#Hx*((p^L
zEfslXZ^zPRS~T7cDf&xw4ef&NEr6HZDf-vaO35XN5O$VELgmAmo1P;t7i`3Lk0+NX
zU|b!{{71f+yN?Iul9Z7F#qQ5Rj2f#L{QTZR2<oeixwFyzIuDaOakcR-W=;uWh18wy
z_OAIRRqKOeF?7>Vm>P|Bn&rLd_^VH!u=h^2Qr~HW141@*@dRpSE?Yd3FPz4tp{)z~
zm~s%CUdhmqEevBs8W=5`oMlg!SDR%y0zz}CZ6}9ZTjgy7yG^y(G??BcC$W6m5yV#8
zO8pSXYEX1PqeIv)H(dCrdNxK}tg)ih<@PQlLVYyHN*bi46CTyWyX&9Ah4rwTqxbd>
zx0fG_D*jo@UZz_2J|Yq?rh^ra=72s@79x^{=lYc63}IVyAhBGY#ze7Py!2fb6@UEF
zA%u$#mR7W9vQ>tF;ViA#_xV#qn&Ob{OHY(>=Mlmf82Q;(dik9ccBs5>CWFEpL|qZA
z85Zn1Hw`vOtQ=(ReH`-<?$5yzCSjJfvr}qE@4jcO2IYu;Cp_2&F*J<3fCA(v){a;H
z(?aWm{fFCDYZX0Hw!J++cxVQlLm3j|K`3F;iqyXzWl1A_Iu0sHW!>5!?;*{CHXa<^
z^%{?5wITOmGQr8{mQrc8o1g9#Yrw7ObZ#2m(l>5cmZVw@HpWU{k$ny3j<<PX!*zMq
zNl0kHOyALV|0A!v?4CW-<2?SmWPxPBkuz{}Uw08LuKW@;rv08AF3TyD;>|T=iyP7&
z*~W;uCNLm5H2d+rlnP|#5pBz&GhG>T9@m7}80nox@yyc^X}G&*PWUL>U<f6GhX@7F
zy1E(-Z?oE_%!$r&%-;lFchv&-!G`(<6z=XPh#B<zxY}+T`E|7!Dbhr0kC26YI6+eW
zg2fv(Bl@!KYDA7P`)2~AP#ouJTb{rS`aOR?NBQ(T|Hcs0*;ZACz{O{m+j_rDpQK%5
zyjdw+;cn|X%N>Cxbr2XRHakIW6W$$J%D3nJZNq`eD&(VE1;>JNX+5cXx1&#%8?qJ}
zbL|qco4}obVoUgVs9=>9-KEm_=)ZJ#{7V&G*tkBY`-fh$n$K(#$D4nAtX8DtVK7U(
zKgeuO6;Z;4qPBm1W?kd}K;CS>5<B_EUaPU%7Q7g*Nq@0CZl5wALG}M5mT%wZa%DlB
zWxwl_(z^M1Qir)(=o{@5LY?Eq3y0bkK#360k14V;Tc*nq(@>Gfztdkyq7F_pWBBM~
z<5k!{_1Rb9>|7lN`a!1HFx992?ZZ!4Q>T{zN~S~QpuyNKp3Jczj!t7(J`=x$N+tHS
zhie?UK~gZ~n0_dgRs)%Ec(H}IXk$gerFW6mqPr!JvNCZ{S$6?yYL$vV4`?4EsXVew
zc8ldxm8s9ZO*5A(Z+Hu~=e!8$Uo*4~A`}5v#2IxPanahihp=BP*k;+gdc^Fz0n=5Q
zgcS6rw189=Z8dy=x@i|3yRtEwN4BZ;A9!JS4=@rQ5f=4q7xAqjo<kHXfmx%kXXbO`
zz>Wu=@71YYkMJVqw?6qTNYR)Jq3nM<7vl%E)geC|)JNfSta#=erTtdbf5Ki=)!S0K
z**jRQ$q;Y$b<B#tOVPwFGVuHgeRqn+)udOw0?n%J0F&)T0P%s5Vb@D>To%V`(4#mh
z`mE<IL=FT83IWR8iWAH9Z|ubdSiW!URQ>HOyTY1J!-SG_3rcC@YXrHuhxh!DtLn!e
zZ%<KkU5!+lY{v|wa#!vVS`f`OO>M6ksd~q9d<VfaDtE3?%4<<>t9U>>gGL1pQY)GQ
zoJ~<Nojm?&{6#EB-??dksxQo-&GU@kgW}6yE&wC_s%l(7i2^K2S2ie@5nA59`~ISh
z(E>NM1+?7iqeD|)bMZwK%V|}n{!YDoVh)7SK4vk~zRCUpPxG>GjzfM(`b79Qj{+~6
z5;g3m#Jap_F&<x0e0R%%BEQ>}?&oQ4bI>Q$?T5=Y!Q0lVb2@|Am!!?3ac@{&n=^B>
zbH3om$4$+4N2@&Rt`fgp*Qc-8?MHDYFrGWqQ16!uiR!~Jq>mnJP_Eq2tKEM+1+K~x
zV$Ok8h6h3q;qIThz$M>YoF%J#fp#bO81#8NeFh^#^N^_h&P?0)gk_syPVH8w?oc`I
zvJhtMiNcWk#vZ&_<dh+X4*v^KAKT#FA|6tl)Q0%#uKk5=Ky=*F1Ui}l_v0U6|7-iQ
zk!Rqz#1Y#&sSwfOHCd(NK0^Mg(r9=YIX+Qqab<G1LEUb6dyF-6&t0LL3-8CLUx!ld
z2@WKHn!PSY5KWaUZfog;U#g=iQ6zCHjFOM-QZZt^7E;MTtBieO*8X2+k+lJ~juW*b
zlY7KWjv3cdQs_0|>j_Oi=4!9zeTDun4~B`}VKFy;>k?0Y=ZJ`W(KKU2h*)8I^_q6F
zS&#bN?o=QKilE=@6Kz*S2K?$J9ex*cCi?I`_geM`O{;6yVHr@a<n(u*N2+$5TDXLe
zjat0owkxCLpf_##;STBtZ{OU(hkj+beS-?ZiM-0>v9_5(=`)>aR26%=t75YMm9v(W
z$_Yh!Gk*7(XF1nG))bG?$|D~4p%~#p(Pu?7iJOu-Z;NcJ&XX_z*tdrLa_4wI8k7qZ
zN|j_>C$_8l<NMpbTqrnoa$cKr4F6hL*|xJY-ljxTc6L9YfW)fh@bc06iJ5!b5ycvE
zDu6_Ri}?<^4+jU!X9{E%WkW2}{dS}Xn7aA$>lYQ!01dtB^|xdF1>}L|iq**!dE7Ld
z#^WxrNuB$$FKYEsb}1NsJ>Ge##n6#t&~bc8xLi%R@FJvd^1(Yy>;b@L47`{l9R9~q
z;9?N5aliR(k06rZU%6OYox8z@P9tS5PM-L)N1qF`Qmn4EHQzv5U9&+v=lT-Cs*kU+
zzy=F!6{l7=Y$eo2al+|HdhlQ$KtVbgT_7m{bRN~-U0CX3!j`*|T8V1nN>>r(NmryU
zMuTo#7#S!-Z7z}u$~N6>)t1M_E$%jW1*-gAWxPS9d3$y$WbmO{*6uk953oo?DqVh4
zqMQzM2hT)JVG;`v)$42Dpck_9^nZ>qZ%ru%Gy#djmr6Z*tgF%}Y2hq5yb7mheJ$K{
z)%eG0_!5LxPHGG1(78zmw_P{|;)~RrkbI$eD;Kw#Yw*syk6~C<<v^VB97>FqYhPvj
zFZU*i@_)EDJ2eid8I!Vm?-g7V+A!}m)0|FwtI%y0cAi*|c#<3LWS9UpQg_1M(*1(_
z(YzTt&79HV$8e12+ZR7b9Rd}D{Xc*b5c}W2-y<&J<X7mc&E|dP^&&57W*U5Dib?ga
z+<<ISu&ufbvguZcihD%uX3=VpZM=*m#3`@QdQ(Z{az^IPhp=YGn8}(EaYlf@l6V+`
zgQkFjVrzB<EBO<Zh*z7d`t_MNc)uLqbqgC&x&9t@Z3#5Df%{vyGhQm_X2JCS^>^bh
z#O$sgRlgJlec9vvEFdD>v{Qv7v-}{$`o%Y)tLKj&bd4>)ewJ2@CF$OE!hp2vP2zL}
z`a%5Ow<cg|H*aS;Y`As+F@c~)F7)3=l%7Iv&e@sLLO}F4>41}<SE8)d`$0eLIV6d;
zg!f@(dF3`Kha_vo)|;-kP)4E>&m;Fb@8D_svs6zQb>+p?gk@nbXsLw?>x1VRyHb2d
zr&NuAQ-5KJ)b#G2<CHyS&yfJAZ-=PsM<dpT#jb_`uO7y7MWBZJ_Q=Qp`L9=*sm#vi
zo>0^pkTl7J!^>Xj*^>+W)4a$<V=LAjpbT|3mG)4UtyrTv9}=JaAE~Xz%}r${7)9w0
z?Q~V{9sc8!Fm<+>fF0k*MlJ14%#jOAzD5_CYB9r@Fj=>uZK@LmBe`Mt_uy+BKFUmi
z(%dmYsQw4qoWg*6$+oL%eXFMK{pt~jqN$6%fKoY|Zuohvba)wn*eU(RBaDB=Lp0WS
zt4p^Ar4}};yIJ3p3;{OkpC;J9Sq#v^&>DS|CTYRL`{<p%hes0lM7V55XUVn);Xk^4
z%z1IOD@&GWFRH#Jd9r0-5uYQVBYeB%jL&2YI4le_8V=e#N#6;iuvf8v+7dh9Zg$@4
z3?n#k9`E>+{=!@2coVvcEe6a5*fG~L2iOPm8J5oUU~4Po=T_QVB}Bs;LGSj&Z;(_T
z{qHNCn@OCAn=K@5)|_E2Bn2ZXZuIGb923CxEzR$xVY%H=`kw~&AOkxggR~qUs;$gt
zse_!t@hx6gZ2h<95aqUuB{}<9Pz9ypRSy)KmV4>6C+&io#h;}=-_zXF$kHr1%vk94
z>#mi>S<|&#=kU~Y$gi;OMca>I{Vw6zxb@Po3r}RMc~k%U9ILNF8w=>?(p`-A`KfgG
z!VdpqWteVqsE-xt_i1FkDYmH)FblLPCIz1?Ut!{8REZT2KN-2Gj4+LxeN4+p^sdB6
z-w^M7TXNs?`RpKx3*laf(r8flZYWoDn2zqD$2ju<rTWgAP}i$ij*1y7<CGmL+MJJ$
z=gOIb)(91yAwNL`abz=#DRKMSFK9e$qwV|qTWqjhD20(ZKT?qR!CFgC1{uVe)yHhk
z7R-&GP!i27%S$QGoZ+ZF32=&(I<1O1`dUUEcJP=Nh<<Mm<mldY$9RWf+}n@0Oh`@<
z^?%jp;sp`4ovsfbf3o<JM@yoPe;2)XyKb~r*R)V(Z8{H6OqW)?EW-~m%!NjJ^qdz{
zhql#)8$%v5UnK4$R49As%}n&k35<2@vmtlN*HuAr^8!=>(ADwFO*WI$Gbz-aEBp0X
zwZ$m~Qyr$SjD<lxXSKs8yCybk4>nAIH}0Q&4oD)u9oc75%RPADZ%P-bBhlDD6(=M5
z<&U4+I9}Z`PMv)V`)=RTqRsX)fBfEU5F%#!wx+7J%fHA$-SkTXoW1e%L}_=Wt>2^Q
zm`RNfKQCh{G@vb@0sVz&fk+8l+jEAOHs-B+4E#QRf0DZ?m(9s%6Eq$__Y%<8-9;G}
zs0!aDeiql~8LM(k+jS#3P+*_#!|*S+^;4dk30X-;lhX=;KBEUKQuOG`NiUaymlj%w
zE_-nnQ?qwZAirVhRiUlTA=>`-q45qjdR)#%ItM@Xq&iRhbj!Gs1^5l&GJtl;o9{10
zsi#{KVTZ<)T539=v+s1;eWe84+Lh2=M`43hziz|(bZ~pf#rhqPX4(V*tKwg$NF<wh
zL)ZJNb9QrpM-UE`QGRDDOn9~jx-32huejiRaoNyVGz~^fi|ug=P%B>*3Sm3bVFbyr
zEALum9ixEcoqU-@6mCaETOln>)@>jQAOAVD59FDX%DQ&naf{V^#R%3k_gl~m3{vg5
z8J<~@P*X=X(Lx0XHB#2w9`aY&b=)yqp`6QqfQyhjut2CZj~$TuC?+i%`~%)fQug3g
zW#P9Qc<NDZ=o}lNQ-PK1Q-RON-0mH$()3d+_{kTQ*9+aw2efQR29^z6<@O0RnWn`>
zDATRLwUx!d#fiz{tm2a$IQ@=j%!%V!Usg-wWp9vv^@h!D^vTio4FwczLM$!e7wIS&
z#2X~;;4QSsS!-$Z%Q27-qTn12-m?nTKjoC3SgAPX{>qqklkj|oi&Hw_aa%UMtpkgL
zS7mj^yHQ$HxizPuN~=_){oWI*^x{iz0*Pi3uDlN(_+s2gA<r^zJK~u3<gZQDg2r>x
zcCgz)NV9iJK1d$JzFNLjQa8VP#0>n~qTUwpOt?Y9C9qr;fgRZ%S?CUjJtR9DC)0IG
z`!~-Un)v$>H=}#%U`JlrIAeU?rfI%GVbqtD@#-F}E6dY}E_<l{?wQ{X>FLkGghgrE
zp30Qum`}~Cb$<Pr?Q|Of+t&1olGU4#L+8s!XYq~ZXhOqOg<zL6D;RePZot~8Us~mH
zA)f{hT$%v$5MH#%4-d9>{o$~pgRr~xz(@GbU%eTPIN+~Tw~Gz&Oy*u9t=3@rm8dRA
zRrA6(fr2Oy_p5?|`o+quJ+#S6Rjx<?QmSipW#_uNVj3!g4+2X^=}SGF!R*Un%G2?~
z{keQ=!BEsx7aE(obv;W6-9Z;Yyk;7TM`fQw@nXML0f!MK2_l5LMaRt90x}}h*3<*K
zBhQ<jaxqRQ06PSqs5R*u|1i|MfVdFE=I(0eF0fhDWghc3<(X7N5-L0{g8**k{0H0F
zv<jm7tJ6HNw5u^qPjcI(-2$wrrNvMrXfWsEso$2_JBQRQy_fQ5tck6?#=mR`yP%)G
zmjPdJ>1|9xc9@6=xO_jny6p5RJknZoT6ALHM`dEg1y*Ct=-QdboMAv~M4vi!$2HxE
z^^v#NQFHpL!B=VDAxnc_q_fqQUtYq|B5AK{PPX*sa!!W6Rc-3rSySCM_x9Jtb*CJm
zPxlJuBv_C&AnbUXD8Px+5?zsaU%iII*9y!>!F%(Ocnw_?Zm*Q{a{+bGzJQh3p$LD!
z`GKuJ3)~dHR)%X`uZ%9xZVbbF<{by<ERi-wC2uIdKm1H>x!462;#GAyyuR;)fqMs@
zt}fuw58Put(WP}Y3SEXm;^S=IMo9gflcMh9B2O0>QheP{ILI{oH@C69SK?lYq25Ch
zp22IDLHM;T-nKo*GkHCne138d0`uA_vhd?wJuTvz)ipDC=nrGEmku*dY#Pp#6mH(P
z5(t8fk>DscRB2?oG1EqHa^%c3${YBAF500(__JgOB8M;}+TUEg=J7r#w@9-L%!3>I
zZ2yQ9lDue{ML+2<5l{h3OLuv!9}84>XJ;KQ1<TTI)6Ba)dkZ&9>u;qUQ)~!Kt&O*%
zN6p+v^eR7^3qZa>TRCs3N=+=6lKN!{uHCeiIe<igRH=vuISYz-jVoC=hP4*j6{f)^
zOMaxmUOxJs;3RGh$qhddXjM7Tn4nl1U`z0SeskU+MNrPlFtuH6bZ^W4E;|}~0a8QA
zxM|VH%0^XDr5gs_Xqy4%;7dfGQr5L4DK*TA?^M(cbG_YWuFH$<*hRT;l;B^9Z0g@U
z2&<vfsQokCo)MUfuda_4Q=WQgsa~#Uf!jzH9N&*9RkGC;&5ac(miTBbCA?;Wnu)R$
zeL#DL7H_R{EX`=Y>qxgas|pN{2?;2EK2v2b4M#R8%V(DeYlP|ct~73eJmL$w?wsIp
zrE!NX>v!KCWfd=nQk_@~w#G!Prv2IN?`n%@wpWj=zw2Fi&Ib+i%`uS)g81!Mxt7DV
zGaYAa&F^I30NmBP%q)6xNJHinxLF#p-n>>%gNMq`+}b@(50ls4^M3=_%a;rWTup{+
z!$SAw)_|<13<@p1O0iR$f=~!Ul`$S;M@u3axg(>&)x`UngSd=~+C`|;iC^85OfhTY
z7rC_hG``Z9>?RJgFKMMj^&^a+Y~Re->&<hhf)&=%jlM6a#q}<U9@1gmI$`@cuf8mm
z#rZ^OeuN>3l$3AmoZMzyo%EBUsb7>o^Vs~*4AyHgBV=-Xcuw<!l1tUvbqS>WK*Tgi
z_I&iw92F*8$yLAH{jZPic^70J073Ri6+4AIspKA$`ihIM$F**{AoVOFjtlfYhO}`G
zJ2Tm3`yN3*`jJL{^+UXgM?DdEAaoi#Hec_wU-<I$;&4+*dG|{wQ7ti;xq`fEw-o=;
zzVxM1*}2=s-jh_?Q%O4-sXrgRWPq|zwI8+N6<MIU3)`Wim%gf*^k++wJ+scr3MWOq
zeXlxmXT+v<&mUB9+mL5_oOi8acti<t*u(qPtn0F4`aP=;HpddIk@Q!uFPG^^#qh7J
z(q1r~qd_aZ0k=LX1VLSzm!;D<@tmqishN|`z1=+JA{qR%rWTvr&wBkZ6WxOGAiFu^
znCR!-I~7{P%q|XI6BhmZP5LJP7iNRT+@NKfPCsv?>IC}E9RNp!KHBUtu$ltUM*?Mm
zl`pxg9{UJ%rsyMcbKgI8IRoAN<M;$P!~GgX83jHsU#`UBN3)#B$h)+qUX428-F}Wv
zE_lvvFsl674<)HIhR4fy-EPzQD8JyU=Kk?6dA(LwtTIT+G4llUy9tAEWT)->GOMdO
zV%(zQD{d!xwDt*`?#Sfz7P3nA4ApUe9F!Yv^kheh0(XIw-ja0Xp87}Q2B?KAlOo@C
z8w{}U9He`BgkoI$t(I<k_{A2i9qFQ<c}?i*g&*fXY`&9@ARsur&sWK}-FumoaO-_}
z=l0j+|I1G3=;C#moua2rHu0P>_dYD=T+U2H5Fh$_hC++nFiSx<LcPdp*wKEs0n_rt
z9s}$iq;{T|c%(EL(pZhRkPrO377i<Z-C=eT*ur7gA0?fl(=dTpvWn>?5g~1X_K^Jl
z$2Bm;I<%qVcO<cZJN*0uR)gzzG^Phd$zdylWiS3b1T>(5iAk-ToYd9WPy49|>#u9q
zXaD<<ikWn@U(;QdyG&=c`I4)Od-pqfNKoQ6S~PrLH$p-ty3HvKDLq3_b@{(Y0Gd1G
zbR9oKVDn~w3Ywe^MRV5``QQBl#Tw7=9dZd=)sXI`3&1bJ*Yt7o0jQ2}b2(|*=hp}Q
zcOSy(d{#dX+l^pqqQ8k=pviK|%}`+&h7yEzu}U{eIZwk4aVFh5kcIn14K!amaYz_g
zC>t>vq32Mz1<2jPJlk(=0DoFuxbTp;{0=nWF!BLUVc5#^;hg5i@8pd$agdqOUCnT~
z4!``#Hzg&(1`?|zna|;5BN2rsP;KzFHQiPNN{U9DSAkpY=Bkytc{;~OpBgWUyo9Hl
z5u}%N<$zQ@vlj$?s`XKm+t-h`Y;Bp?0GQ1VuAV%vtQ)cC{RSE<Y}6A?)_T@B5&7jw
zdWmiu<tNy-x6tud(9?@f9-$$`RgAfii6zODwVq7&(ae+!?U9aYYzO$?wtYzb@ib<0
zEsOw~Kvu&K6B67WDAIm>wS0|TYXrsxnpw>mkC@}!9Ov!NJ%@*%{dGF_PvpdGjnEC<
zByrdueR6A7-gu6Mm9@C#%VDwWSST1Fsd`3x=^ZfEbbVp4-Y0+aafaI3+E$N<7<%!o
z_)pg0=>?aWj-cgP(nl#MprY3v1pK<OWu5suzngEi>GR5<cnqxXWs$|OVD*AgoOJ6m
zcDFAv?tOhNwA+v8%F+F<vnRs@I--@*9D{9um$n7~xM+PCVo7@W%xO|hO_=gQgz>KT
zZ`}eSdh$4K(f1<Ki3t}+O8g%h)bAmPZ6dX0v5N&XEaFS}rI|^#C8^S)jNi-WXp7g;
zM(|sS)DwcRJ*usSsPd=Ga)@8DvDglmcY<cZ3B=lsN0r~3iXjJA^xmj`PJw$loYV8$
z5?{r1fIWh!IOnP=Adb`CqE<NKfwNNnaTlHC7j(%Be?R)Dg_7^x&G`gCW%b%a^D4)7
zs7H?v35X-~q#cc+G^LZbe!1ra`{~->Z5sz)r@p_f>`fh${V6y+GjwYQmEShLa?T(#
zU^ye?46$+xRYcUKPQ~$ej9I;z`xNNZpN%?4b0AI-7tRFo+n8sY>=GPpIZfC}VjEti
zoHlyx1*9vs#Lbkx1E)&Pp5j@NhPMnVWjfg$I_dtKbaIF@;@}-9B1&#_cQqq8#^p+u
zI!^+_0|k$bDP49Ytx}E+RjjV~ase&$q3q}@pc*(`I|4zFZ)g{?f(k@&)?lmg>XGj~
z?!&Ltv!H#LB;FeKY3-c%U{Bn!2OyFe&ot<%?(O6D3B5eidAn(4V9U`FQ8FNXHJx4B
zL+<%nb#f1Rq%5Ozkg_r|xcu#DK)BC79OGg*-4`AwyQu@UUcB5tDkyT^T>b#3(b0A3
zHmA_7hA0!*&IIk!&{~6uix76Z)?V@7$y<HA$cW{0vw#Sw$jZ^l^?kNHY#zkjrJPFK
z&tN0Dlg=!U<&?_L*f{pzzvilcE~<F~=F)WP5Z{jIH7Q>O$o*Y-W=-#fX~?zovo_1{
zN9c8N?ik;<JU6?bD(m1njtiVOHxwgR8YR}Eo+})?gctLtR5Hn!R$~2~yXIUbesL?_
z@+EjJ2IqY$6`5m)WE`&bFZ0ju8gKfnl=%Hcln8gyWcGhM+nG515f1E|jfVXU29*rR
z>+M0GwW(j(LMA+MkPcel#y<1Vb?*hKq`sz|%S*xG!e>-$Z)m65(TYGJn3rB84Tw%U
zWZ-_oY%BCH`88e~(4^~rk#2o;!IftmJ$~lQg8yz)C#qKU=DWKIzc1TrzgpOmIioTz
z-Bwn{^^QX=dbNRxz1t>AUA8e+&B)lZYlAwPXLq&4TR3@mPW+wv`rI-13Jj-J{KVU;
z+qLdhD&fV|sDf{(3DH9T3~8C9v+}n-K51uM+4+6o&><lG?C+=to(P>;@TCb&SD^;l
z=3;Av2CA1vrldMYj<A;z&SOtQxEs@-oq&VrU&G#XEqqPisyXNQEY<<~f6i${-MO{d
zlj=MZv@X8W%op3h<C_LdbdEu+%4|zB-eqeX2jV_a>qVF+xvcnezJAq^fp%-ZyFks`
zj>AivQfo~d3>GtsYf6uo4!U`nsFiuUn36~1>Xq7MvN(jQw|-i;VI7~y+6^htuBL|t
zzim2@Fvteiy|JQyi!`)C0KhHLVy7=)VG1%ttZ{892p8>)&RCiNpNqg?_NOyVU5!&N
zt1&H$fWHZKg`EA_+aEkmWIRl#Aqq3ysYzRr0sP)YM`(!p_*oQA=Fe+1bu#7owS`E$
z$j_2#v5wS3N#nwjzV#zzE?dZsK6ab)=fv|@=Js4Wt}?z`Z(}WGjjBNDs6kekwSKZe
zPybZ9Ft=_h+ZpfTrRb<*b!Az!7UD7Sdd|roSEhl8k4lBB!Cz3JQ{x=wh4gUwOEkk8
zt#`xazfEVAzf9@O4mg|>R;)~1;2OF(2?hk(T;!8<I<5K_(l~CFGx&VZDy~0z$@zLv
z?kl5rZc&~+tCyXI^Gqo0wHK(JyI`>Y`sHRZYdX26GVVR^Mn2up-D>)L6DI^$wfp}@
zA273w^(c*YDL?-)l4r-q)z98X#+0t4>Addol*IOQ9__SgF<h7p*)(tYXv1J|V5awE
z#XxuAQl^?g|5QwAb&f`;2x~YaLu)E?@Xp^akHh435k;=n#Esm1Eo3`*rDt?2?igI^
z3Aj4V7@i<pKZp!0Z#z}T<wa*p_C=PXbC1C;n|hr0sK$EcW_(Gmh*kN4UVdWV<pH6t
zfkOu)9*5HT6b5i<$24?mj0`Zsc23`a1u%o`5iHOpyIf>qVuJm5R?PRma@;9Kr*V1E
zGxYxOW<w}=Nu-52n>g$u5M9@%CT9DS#fDsKH`^z?GAa{)X<m4PU8n+rUffbiIyNKB
z0?lan^-SmYNV}HR;JlAjEr;C6DfdrnuRNZlt4cq%PHryWei}y&K&C>hrOG56bUyM9
z5G3X&v=CrI`Sp}zoZvV;SHCEo&L9ZL(kHT>QE`2e8;{nk{z0%vupm7awn3nDtnvtC
z-D&t#g`&$o*XYb<=Y<Q^9=-cOlT-vSt9&%Cp>okh)>T-D?JAKuPu!F_--wp!eE20S
z4^p&)m0Kfw4jZL__U!Aop6#w$5ZV+isRA|C*XP`gH=cz!oV+%)(Z4uf7tiftDR8t8
z{0?EBb)ZrCD@$}j6Myj%_B-gB(vFU!I-m`t{sm4L$9Wa*{)RE%a3UnQ9To1BmTJm^
ze9Vg6>^$Xo^HTi|x;Nj+<Gjn|sTB-`48D>^DejXoD$*_YT-PIM;%Go{TMId9?@mvr
z(+>MIrEZ?!bVc(Xv~DuOU|bk}>7ySuWXgD$oU0>&6&z_W^?ckW|AI~fW%$-o&HG!2
zbb`FnLob_`D=Y>LRa?t-^Xo1QlV<}j#V3%TQM!imr5zAOoRbl0eQ^{RFmA5=bz_2G
zzh!_m9nY3TmQU|&?3ayA(xLUkKfKk>RsfnAP1aA5sEo*e?esh~;*pWzP&x@8zG*@P
zp>|A=VDBkptcu7*I-aPeLtmE;)(d|-b7j^J)p5FI!0D0IlG)6H`g8;|ot25CT^n3y
zFVUHmEG-?HHb`8@^HW#y7l#-UB00Z=_r3OcEn))wS@|b@rscG*coG%*F)$PFDTi!(
zb_qCVmnt6A#X|i?r(9PZ-NtU>gc%@%JGRNZSfR|2m{zy@-Q>pE|8N1CihIIFzIGpw
zNIzySqCo5Aav`_{T7l^=E*k5y<+}C$SD7Babu6jO$DR3LbQZG&Kf2n5m6pr;lqZ?r
zAnhadKgBPBQ<k2RA=e~^6Rj;;#XH7?H{k_0iswOsVywsny{Y#*<gx;fTMLG!8Y@->
z?Tp81o<=Ty5i^1Ov{sAVbutU0NDCJNbyCEQ<?2~3NWdN2YBwKLQvFAj({M5?8}kE9
zl?-bsBTe3E_6ey+;4JEpMm7bFJx3MT9PN!g#<)<63Csz!Z7Vk^tXS9(1KHTotY0;+
zXd(`JPWO27(_a9aD+OEd)47tG!EElljOx8&`{!QO^$`_Jm7J-^=qh#9ROWx!`pT#%
z*YE8?4j@RVh#;+`A|WX)qBMfi4bluD4TFRM7<6|i4AMF<^Z<&|9YYKu(lH<~0}=!O
zdpzg&;a%%xEk7Leta;|S_r2q~uD!QWkq`^r=bT?)pkhZFrW@8;Z#D*qnGt42-EA0o
z(E&<sQks@%%by|={x&+eL$NKU(1G}BB_R1_(>vAO_}=88ZlrmHyyH-GXBo5u`Yy5Z
zL9|jaaz9&q$NRTz27DYh{<wW6(x#zjReQLKm0WJycExkicVxhNTZc*1!x|n4<kkwG
z87&NM+m-YSh<@Lh)z(L={ml<PWq?-szjWb#to^b@$g*hgVzQ(C{tzS&EpeP7Ju{de
zBv+cOf)4x<Diw#dsJ}jRpO!948x?2EpF-1ulzd$cd+h!bSC;HX^xdLFg{{5wd&hLF
z#q4J+s9Qhwhw}na!`d2$wiYVx)%$WHfYmEuPlISELuQsCw-c((!MqTH`4iK`+G?=%
zmo6bkwEbj?$vgQ@#BmqoQ$u{fKIg#8_RoCZar&`90x2l71{HMfQ<#&wB0O}S1X6f;
zhs@vr9}V;{gn{Gw+JHrF@EK!w{-<@x9_)?qeUF=z%91L_d_LD|zA}61enLd7)RqYp
zpda2BG7y4|W<{RkMQh|H!1pXWsckSy1?A9DL>mfq(&P&{y$gtW<*o?&;fW(ut6i=a
zy$thW*8WdC=w56b=lc>V>0VCA0YcUUv>F(IlGr^r45S`3YOtIFGj2A?_Gzc6z(<80
zv=I7;?By5WBg$zeNq@+KxjFpS)t_#Pg0JM-UxAU*0>u7u#=gq7pcJSxK)WSD`XUV(
zp)?@iRaq9u81jaJ{IKPuUwNvo2*?jd3(JpUp+ZkU+j#B*a7a`Vh#1}{8ij5S3QFk1
zmDfCz_tx}lsja3H<Si%16!$oxn&JgWHqeNk@lczJY%dok)936obAkiXnd4a4w*Etq
zs#u!4{!OeZiUZ=t?91gr2V?<qpIUt&0mqR*o}@S`bEop*Ffw5aY+taCI*hyHp#6h_
zF0Z1VA~55_b)+=KS*&zzD`)^Puzu&4`>+p;hB!~e8=x6CX2W`NW^I0422IJUVAA^5
z+r{gTh@#5JV3RB%wr0KG%_2QefKhxM4QAq)mz{&`DS3Zq(aoNqxW>G<^}hPSA$69=
zqjlGtF7=?}y9ZNseF%&fq&<3BwjtBhSG>W<$lk%T!D!E}h-jJf9xo+;#*J1rkX^$i
zgz{zyq5E613Nc6b&i;rsmX;5fn0mzreQmd9vCX^;Ytg;zWe#h-*Bx$7)fKPVXr9o+
zle(BXn|MLuy8m^`f=g<gB=;eVt?hL$f|BjK)`#~G3D=?i4)3|m8mE$Rfi0G>FJiKd
z$7Gup*xtS1W+TBIxm7|%^~M!tvRR_hTURqkbh_AQEPKP9fpfp2?vaX1dqQG(v#M|`
zPF!`^kq0YEy)Y1AV=lH%N+EI`|DcS|HCx*3_PSt?{n8U#ll{_iz8fRr6w=HPktvzu
z4>ef%a(Swb2o5_Wah?|$#9BOfV~4FFw6A&5FS$kQU>fo68FiRvsa$zH^59R&#q4Xu
zrnrvG`osuEJDLD<m-BE1ir6gcxUnOYWL#nV*c5f<)}}PF3-jf5@yY?*)P%04=w+GS
zt>9ZdL+ezWjdWID?(J>#NH{*Zr&(lVKkY#X67~ea0c6u_H|xbn;LZ_RBTaMSeFEHl
z2{>nU*<@6qFX!lboX^_=x~QWt5-l}!VLrh<^;XlQ*==3}>R``mxEzCDG*55Tg$t0d
zU+Uaiz6E3^{%qLC-0mO?{=)6qn%I#09#mpcImhFhBjX%FJoQ*X@74Y>fyfzS3W#m|
zob6dif~%eOu}}i^iDY}lJh_R0oRD^fR(7oG`rSiLXnjYqiL(-V(zpIqnI?1TqM)Wl
z3e4T8w4v=z_y<omrw#3`TtjTn{xirc<3Zhg#hH*7>}S8jWr7}Qw-fY+dA;xt|5c3X
znra-mCOc<Pa0gosuh6vJ)fw(faF5f!^Q?1+1Hxn=@2u%mDBL5Kcl|DCkaTNx!ziHX
zZQg1^_%CJWf4c14;@0d{fT@w!2Sz#UlN7Q87=kP9A2Ty?#gURcwb``Fb3qs{!Q96d
z-$CS=w--mSTN!+keGuo{#Ip4uI;^9xE5)H5PVZ`mt<m~T2M(}o-+3r0uZDdqsbgs{
z`~vUN^D@lR;Jb7Q7)<p&_gZ7<OmVJ4t&aSB+S3%_xoF^s6Q42>#%}@+s?*hu7#e{~
zsZKKIf(pT%72GnG%M*AtkKPo#S=!p96ov4|<-9@0kEQP|h7v@WO-c!CO6cF6<P?z3
z%$ixJ_U07#<KyZMi2yFBG%+Vr8EcU5z`e+AamtL^{UMcAp?OvE3eY2oXvhggmhF{O
z@Y?!^Z_c?5lU6bv!}VcH{&c4V{q?g@51MM<@TXm^d(R4Ve*530o(bO5Aao#$){VC1
zjr2;%>GE?VFtb-ChZvFJ#yrKIZ&n~8b*@SP{nk?GDY?(Ld+>Jxtx8)OC4xC(UW9sq
zS^Yr$=mYmU70H!oJhx7LM<JERer>_gyf`AIh{8lnSr1s-T{Qi*-86b<yFwrr4*NDW
zJ6dSAZCUi}cKoz|M{?0>v;dqDD0D?&)NN{J3?I&9)r_VJ8UWDS!PX`s?ms+OUPcnD
znK+Wd<gw0VDFBtTnM&c}_)~?xh+AJC??pS0+=_f2Q?-Jo4b-wlh8i!B#98cfmTHp3
zN_2Kh-vaL*Qdqe7sw~3hwj}k!nYs-Bnp8qX>Y`!jg^ku}FR_j`l043`KGKj{0Y|Ql
zMNUAxi=-eAXt_E~CI3{8qTwHN)U@DTl@-8@zJ1+SnDJRL^)*cyungY1J>xeteFI!g
zm0Y}RXRF<&P*2cD%o3BAJ!O=(Hy>Gq9DI7A_pamQsB{x2^yn5B-^gZTf#>RHsI5eF
zS*w~I4FcA)=#LqAOmtoMZr<$TEVW@8i?C@L5Q*|luuSl2EFdrZRtjHow6=OMJoPJz
zsr7-a+vtUWEE5)!8KKCOfluta_0vL;w{G(Vn?(+tQb3s4+)&E3c67cZ6EK?rac3}3
zD(~V{5Ld%mKeC3if2>GnsKv^=gu0DsF&-jvYMh$Nah~1J>}8qKlyOISAYNGtkrHI#
zq6W2-7wzSQ>af2oV^7sv8K`8zF#$P1d>6Z&@1z0;%tV4%>D+sJ6UaMlm}uKXD!|(%
zQIF&f)*MdnX17eXKif-~6?QIh&d6oPzaUL}h7in8E|15*(__(J8<4X}($)?vNk&G4
z+&?J_SWJV(t1}HYwZ_k>A@hK2h|;p+t}xrOEMn`M=?Y%yEgl=6t=4};aorF?AM0lD
z<^?D8ZU#OiO4yr^hK=ZFOMXv~x@ohunO(m{ICeccn3-=N?2@^hhlwMHOAj2JALW`$
z<8B#2yh?ZL#rDa9?Q2>N{?*>`OFUGxwP$L7s$R;2(k7IbJ_)_&8|I50dqGg;Cg=~(
zz@6QNa7-kvB~sXbV&||)e4208BE(oC30V>(wfi;u8XrJarlC&U%r&V%T6T^-$QJ5<
z7k+dp8eDzqH$UiMX9HG)7mlYAkFRP|u18AcbRi}oytkospI838yJ(T3f+pH~d3~Fc
zUHR`iJEjO;3e?otFqiY6=(3q>O6bJnG7E)|rUStzPp(DetKH7uR{S#Jccx3PLP9NM
z#sv)@w?BTqYWsr|90`O6&hJP{-S!EbM{tBqSt2f>fKD@`Y!h2bdi)_FL~u<;JsR#j
ztMrE9-W^<lV7=a!XYsepz^AJxhlvs3E1ix~uMzp#JiBVVN~D~_Cr&5~^Y~~5ocCR6
zxbu1<U2|DS4cIItVR-B_&?msqyURuc;iYPkt%^tXT)tQK`!~C^WnUSdZc;_zq8;=+
z+*Bu11_Z-CKTDk5HI=iwz;X?9Wd-WbcJ)}i<j%D|GYGH{9iF9nIg3*<?AIY)27bYh
zlK3eR>d1o9(Ya<=Id`p>RAsl<-1~&B6r9+Icr1fo5=5jL$K!wEe*DBd!Kdrn5xwa*
z1aBM_q6SPY2}`(>bzrA62{e{#nu31ek+dDa`{V~>2P#+3(Azrh<Nr)I9HbbO&>>jw
zguN}F9!{8Kfyeva*s05saWgHH-0~O*>}~ZQ?Mf~l=xO7d>CDJg)I5VN-CSRNU-D77
zB(%}R#xjRKt&^2jnSni3MsS<JFe{3A5II}jGC({AU01TdR#m)CSJ`Ya+^RoiPJ(_x
z=t~IUT@xA+w<229&j+}I?4LIe>cl~?Ut&)M!Lr_hmT)Xllr!j!P6BB`xygOK@);Xn
z+(U^*p*UAiCttBN^H`TZ{H4e1jNP)m77aEjor8!RZ3X$oHWO0Wy4+eY#f^rQqgm?8
z!o}2(g=`7)yBFm)X$yHr4Hj=ce>05Btu!kWRzz4tk12%PQp*4NrGz1no}ICj26qJ?
zn+8STr$GdY2ZL}G2gc{BZwn8Q<yjTB^j@{2Iob?o1#4G{wIu%?GHGyWRFVs7OB7pS
z%dPG`*C)S~K2{Ia4Lx4x&>C4}xe}I;#C~r#D2pn<{HRM?05=$++^}BVm0`hnH-DvU
z2YkJbVW8m2kntt>=RUR)ub{JPzQ<ALbJ|764taKeDoBqpsL1fpX9nSXbT}1UV;ez*
zFa(*kFR>Y`^^IK9+Ghm<rmxE_wQcZVE(D(PsPYyOtM)Mlvk3(*n2^3%n|mzMeA~Hf
zSZ6LUg?sG+okw3?=+A+UvZ>L-o|@TDGO&*&xKx^>X*A`jP~@G(0~^Q>y)}W65Ij%q
zD{!1E${w%(ZL2Ee@o+4oZhNNIph1`)gYwiT%QoHW31%7Nuq32AX9ylL(?PCi)BwU&
z2KuPHKSM9*$$$yH8ktEDmnTSo0%L^@2=9F<Mj5em10h5!%rz8u(I^z&*n>cJd%(b3
ztp5hM4v2qDve)RUc9Nx+e3;(x9fh)OZ7$pgH-c|tGQ0d}@XS5VS-J<89BTUE0;AR~
zpz6x|{?BbqRl2N>bwu@UGQC~4$V-z`E4q`jT=_1AOQHVI*px$x+$&FM9#F^$8OEH`
zFfJ5k(DIj|KAd@U*!^VX^RN-?UvS`QPmhmDn@BUab4$UT`jM(Sd*RCx_>=2~3x9M<
z3p~_i+Qj?Bv_?*NfeW^Ms57Nyzu_gvS6VaehN-DRC^Xb}`OU@VPRaioCxuY=iEATG
zKr%z>F6dm=d$*!j`cJrQn}}<)s>OAp9{lb#Gg{yEnS%-=?i{tP<L^+$^&>In_e5Gq
zUcrC8I+TCvEN7Ap0qTy~hPOLy%yujyA~!?*+)U$;g+ZMwRgZoPA6$Gxu91%44h^qW
z3}F9VP+^aYNOvMuRe9Bm(3w<OZf(O`pIl5s$oGHn5DY;9tSzCvrwqJ#@5lye>!&sC
z%;HV|*pmV`q@sVEgw522^3Z0#T+_xjt`KcV>(8SPo)4cE=V>8Zv?6R)6Duu}rlm^J
zie==?L2=0!fOsbB6QKj<J7NJHv#<Yo+P~+pF~is^NVGt46P5$;zw=y%H*o!Pk?+G#
zP9^ku80M-EW@cD+aSuNM+Da7aIquCHpM_Aw>R5la>{*R&x6fF6?@T<4dpnB)oidze
z$LqZ69P#3P;(g<p@WVLg409BPh=YmF1nq4``6oz{ShpZB=8!FB?U`z)pB1=3ji=3y
z-zzC0Oj*ci0}XANx}~N`Q;?{Xr<s%ww4CB><c$W014h+B0USm@JjI6A_OBQlsu+FU
zJT917qGQYp7Rf!4OW_1r7I1R6poYaadc5|Sa$lNE`IDw_?rk>c3vA@1<Xq3Kl+DWc
z$-#M%CaIby+Mo+lq<ujHK{2Q;Ha0~(0Z3uqMMq|GW)nKK2A$!(j)PpqteTShV%mG-
zKiEmbzLXb1_*|jgRZ@s8*Z$^$ah>taF$$q<;aY*YJwQZ}PTj^%B<L?~w?I~Xn2vgV
z-|MW_Z$_65DaC*)#Sj&r=3eiJ@-z7+0Wy;jBX{Fls_eWB=UM~oh(i=VItv_Mw9CHx
zvbpzcoWy^9pkuQkOmfpBl^e@$7$<=bPI0ixtDbr0@DcXy4jYzsV`gO_p8K$SGt?vK
z#+}3IY5^il$WL>qFk)?dDy==-v3qjDWnUsPKSS?;D}t?PnBH{$v6H)$&|SrPYBD0O
zC<gn;qR=KH6z8M#BIi_LMEv0DYW<5uG=10EWmA$`J@+Ez_<%F&RrJDqF`M+uGA!z`
z{9*TFix3xOOG^$&Tygn0yB(X8o&2@G?)=%e4?C~jk0&oo1;|-*sTb*#jspM2n|K7V
zzBr5Hgr>@Onr?Nuu6_Rl23?x3(@8gG6%P!WAoo=aN5mzJufO$usR)?e>SSp)*vbxM
zBMi53EjYf8crJPRS9Y~XkxW-3Y<l%?eQtO%oSslZ9y~bl>}|mr|Dd;`AXw<9@knRl
zvYqeBirc6N_1HhtF6-3l8c0a93nhXaB_z$J;#IssVI_ADm=TrJPl8`-=gYqhc+YNV
zjf{4LAk}qT6*H!#s?e2<?w0D$$RQDNV36P?yQN^C59eW6B{XiV&|&N_v9|Hs9i)${
zc=5Nyy~Bp`m4}Dk|3pcL_&fY1{UJ&3?&mm*T=+B1_tELADR51<*J7Tc6v9prWF~vW
zK|f=_QfolBg5FD+!u<OO^5ePNA+FFWnWGQa`lujbCr-8E2pP8aHz{+~mU5fKEG_JM
z@LD|fXzT(-0>*dAeI}^q)>!fi?KlK-6xg#_K1Pgu&uhO<T-K8HNIN|Xd6XUeq7EPc
zugFcn<57af(%3n-mK+;56CxQC_O$0f#udIqag)z#jx28xXPkmj&;2TQ8Tn$=>_<Eh
zRc5Q|DWL=LOAR51)r)@IDT8g9v@2LV3B=D$zIb``mQ^}yOkyr01K@lhVK$9>Ycoe*
z6J--Iz-IjCnl!`&dpJlTUVCsU1n4fy{sc<A>HbmsJe>L=8ibmunPIWjJqDMTieW?X
z&EL$0|5XiZs42WEQ+ay^xo3H$B+%M2xP-)pkYh{raA67r+d#RNW^+fa%3%-$DVHg%
z?xxj(g|!oZ-Sr6<w58&+IKpq=Gah_B^N_}!IE8~_!7i+5Q9ehrZTO>1LlaOpw@&=h
zgib@%FCY-yC*N(DBFd8`<i8uv3+h<}x_A+EmC){Hamk9Rt;P#F7xE5N{#EYk%Ix~W
z=YPq`4?MV%Sp9GqrjANDEv#$ndXIuT_uty*47>kD3b9-n_;b1*EN4oWPblcu)oKY{
z`+PYiWCOfKQEFx2r{Es)rzPBL;lXaA+IkU6;Qs9PNqT;j9p;3tk8d?QnGXfrx9@*0
z90Q>KiG5tD;8CZr<ivh3KDx(e_N3uP^m#}-9H)Y|l`mF@xfi|KzCo`wxS|n}UWMH0
zuek83u<>l5Ge_xV8Qgj5W!WK^Flp`cpOigmX-VsqbB50HLLhJ0>jJlRx^mhRR}t+h
zodoK(I<*C^PO2!VJedpoU62H@MkLIS9p}v0HiqVX@qJ|BBHd^FP2Yq25$QR|m3wzn
z)Mob^RQ&SCr^2|@XY$iZNZs8g<|-=X8nml6&H==c(Oc;nM8pR5{Il|Qy$9;U&Zuj%
z4yfHK%w&*nKAsEv(j}!9CxI$wSg<NwI(Dl0y;e3AhQ>N4iUX{GKjX7@10Q%C`B~9O
zYGu*bGm9<C|C-o+p22N=PI3#Z)L~!6`@W7Zlj!ck+?3EK*(a5_<h5$&WZ2rhNN(qa
zolp$|Uj!costaQ{Gfz~3^M^gV?JGK`Cy6*X#-TGS7L?Ef1otY-iSde-ZKpBt0JDLo
zS(i`%aJL<r1?#t)q2+G)_o2<tD%_<#{4N1iifw9buYytqQkKE4Z)N>H6N`}9XddGM
zQZqrwVl>Y43|xxc{d?X}NwlQl%@+C-k|<DJE(&||hHh`Y|4%e5b9Y39N^9hK&4eu^
zj$<g`o1dq|9G~kzwpl7Q!c;Kh$@Ml7ZhJaqiGTKFs3Bn>a^zf+L>LBOqsstD1Z-!4
zk)CI5FLZ*~<0UfBily3UmmTL#6;COUx}6ITZCp<x*7fa@l>+`?Z~^-5O3nLa7j@gh
z6ZfM2ER#gfP^2P07>Sa)H6ot4a-HX9ya+y?LUR2pBi$6cDwlwW<e_}UqC`xNn(Ld7
ze{o6NGdKw(B47COI3^UKBFwS*nJn|%)^iFIDE<{7;Wk?iC5m}J>_%}y#Vc(g5Znb-
zz3qcpIdJcHXK8?bz3tA-^8-_?8&-R(aWzT(XjoC`3z-wh3)|;K7X75JNQcmt3TI>_
zVF&i=Zzq9R<a2tWF1=m3+n88uOQ|OxUnf}55aMB`^2Px8L8j6&dS=*vfZKf)LqReK
z{dZyX_V`{9#`U`4%`-k%qu!GK3IDBWQ$m|_Vj_ltJ*d>`2w@4l>iHi}`1Vw1J2NKX
zZW#{ko+xn}Fv#@{C`%7cae=^G#p*rQaJPcD-Fgg`7kwBYQAxrbc5<L`P)%>CHW%tz
zCH_Za_|z8F4evHe@l^~PcE}zaY(s*K3RJJJLGTr!)<JoRZ=BxC(K$Vy5h~BHxB9gq
zHZE9$ykvhtnFO*sFx4pouKOUgrFER)-A(2JYJ@v<oq1>j2)3?rUxsh}SyCZj$5IxX
zrF}_ekT4$Y7mzr<S;Xd?&B;F|@$FL}c(4WUGd!4rVZRy{hyEh;nuaQMkmZS#jf)@s
z=dJF8IbW?8K_t$>jAh`P@88bH!;?o6YtNQ`1)2`(+BCMv33g6t6NBIlSEwD^<PS@a
zbUk-~Y8&yp1f7zA4A)_k6k^?U_^r(hpGgy$fWb4c&v=87^Uw4vIV#Ar8(tC!@-&e!
z@v*h&GUu;jwkZ*44XgCi=f`9X+vqA+`fZXZwvZl)Xm&$#VqU0w|1Ojaq8#99)d=(I
z)uIj9oqM_yu{0u%YOmB4uh?4|x1Y9J_USZR)?E$Ivv%aq@T7)VFMhiNW=6(Hex!dq
z?^}%jL)O~goQ6H5a3wCxVOj~njXoj_j5HF@ICZPuJ{dkfxK0i^T<p7KN8FbxZlG}d
zAYmWW=IGJ=>RY+XMcNG%A~k;mbLsw93tk6sf%SQv7MS=-aJZA51hsOh!#FietG-Ap
z-^n;(l+6Jx4!QDD!fBli3<R;GL+pcnUgRdVtbd=NC4}&eBzS(_t;?-;B%foZ?*4oD
zKg3diDc6h4gaFP9{%s?nE$*_znO~5{gM24*3e2&%|C9Qu+tFp!MWnX5b6`_2cs=)=
zZ*;!9|BNMK=PKyq134O^Kt17Cf+-JMVztijNJ35muE{xrdq)>>TJw=-1<b{X5nS~x
zE<6s1llTr={rF$&iu;Nx0z*}@a;a~x(CytnA+2HC(!ZS(*CV|P>YBVHxi2Ua4#JXV
zms?w0^UkupjHA07(W@<_N_F@Hge<;)UPOOQKvS}E%Vlk-{*1uLw()i~1pt*_X7$0E
zVNL;&Ws^#0oNfZqE#$qvX!r2CR=iJW%<;Jd2Q^dPwT6XbTJQ~M)D=av6PZPml46=a
z{Y4`}LE#knaXJ$=Q*n;mM0;fPV`eP>Wu4Xak)qBu5vqMth^|XnbTRvQ=c_^n^ug9M
z{dp%agzs+3;8scwy=lpvE72RzRMkLi;bP~4K*B$JhBgv&%y%^UD6lNS_dR}Ur05C}
zifumh5=UGGIKPHj@?17pH1Qm(5FQh1MOgC=8Z%zdWH(w6?l}s8*hV7LGyf}(UzhMD
zy}X$Ks>fU#CMkv>k+3oEMpOHGbcv1yia#*OG+8IP`BzqxVC_FnjTi1hk1}IhN<vT9
z_KsQytVV0uLy^{%q>#dPaqWiybB?dQY5-&l*Rq4}gxgLEUv-t1I|-EUv$&qU_oKsi
z=Pc;{c9WyLHZy2`uG(;_b#D&J#@;wQ7EUkn&k`#Qcy%s}vNe;da~-HGP4rF9m0a&B
z=uwl)N0p;e7QpRpmua!T4@&zzHE;$H4dSHP1Pyh6mI%*O8>Z&#vE(XWYuYTjFaVNq
z59xCaK8)aLIE`{X0{bcXO_ht0W0hi81UsRw9_zzW?3&?592i&uC-Lz+r`m5ZfZ6(2
zkQS`fGVz=nq#JbHX5Q==*aMH#A#a%}2KLE&B#_R;<^DMLChzhXd_f9#$jH03qu)yx
z-T{+OV*)PW=se=a`w}JIjSj)86X-cWD0~t63C76&HF5H-^^E^%0iLp5zjD8_pu8j0
zl*d<XSFI#jTyuCbp_?M`#6^9Crme1ByiZq2m5MsQUI4H944lEE$b6}av<mRY^7~`;
zr4+-K`=FirGF)NjCoiVCFH$C1EzRr7&87>Gk2mEzGfMkbAN(W8NZxqAp2dW&8c%a9
zUCWAyw3K4o9BkY+tEBJ<>g%Gp3<;B4qkae~;SJYZXsTezc&J5a2KB4$D-d0$9M(CT
zbM8WvoO*bXOv6$jp0mwC#Y^-{6MNSoJg=$@J5!B_You+|2h!o`sHa*9HcUe5jA}<l
z>Su8apEY@cjfTs)t|i$pX}_A5cjg|5emk$`STa_juzK^C+U90?$RQo#bT^z72EtqW
zX<PPn?a~h-*INHD#!M;058dss)IUVOAxIx_cQ#0925r1Qo*H9|DV}btJOcE@`Ho+(
z0n`9Sp$LHK!?kk=b0NO>q$I6QdGQQ<ABJ-q<Ht%I(q0%r3QMP<idLg+Z*Q)$%?1&l
z9dhNe5uL~rD7OW#b3TB%MYZ?RzuqrPu*-1h74K_FgCClL8PjH7tg!bsOykReO|6vD
zE?~Tr{^JZK^cFZ)j#;0(8>bekjeDoy^1d5Z28-jL|6`)ZNG7BFIQ+vTc;&QWU}yJR
zXMd;;`Z&ffylRUAk~esnQjg>Ghl+e-sYJbpWpMqLIV$1z6-d;LT-jrPCdKU^olgO9
z{CXzri%{BCJM8BBtImMc2IQz}90S+WjAah(k>}Im+I!p_XKxA^5@?t1E|x~vWZ`o4
zM#TA6j^cfAcDao7s#QGaM=#>8dUE^K0I>Z`2@Q|YMH6H*!4al0eft99LA=!_ve{qt
z>D#R6DZ!cftm{UlZrT>zcot%+c<)Xcym){R^8WPq6PKsci+M%|vj`0X@j8;$PrJ*1
zXo1h3)TaRpQd?J~MbKoy`z_nq!G|r)#3zODf%)|B`o;TD2^H~4f}VG*E|-__zkBel
zn(W28=BIe%bfV1UC|k_UoV&%h10M;_Op@x##yWVMl7{W@UeSiy`9`0NFXQZ|yinty
zIc-1R3<R6?-aspFqkT#sMY%G3kTX%-m^+y`O$)PVI6xa@m@WwbG|l>D_t;CUe<Nb6
z<GbWEC10DHJDG07@PLkQ#D7@<^LfMvZg93A3OE+)js-tTU$t)iGW+IfN|;mdL$EI@
zrAY8>guZL8Y%Sp1=$9Nd^syg1P;Iq-zZf(5s%#bkTw9^=jHGcK{wLrI%;*3L3V_&T
z>F7l;h7uAswErG|j89gr<*St!iVQ)(3a^Um|MkkAfnLuefftz6Ka6m<!zWAc?anaW
z^8_z$PcJ+8Uma=^`O7kX9;>9JPQ=;p5c_#f^}g)=i@L#|L2Hh{@HgH1lp=*|<?=Qr
z7c3pejuLB$N6G(fsLo9m1U&}lXh!tsVqxslpI>Wi#1W=1%ib4}8yMvrWx4zf)DL4l
z<Uhoo!SuW^HfnPrJ=aX7OX0*y<gFk<R>ccI4t9^1IAl6a>GPr)nCL`VO!oZUh+z(*
z@)Cdw5zZN^+~osaM%gw@RVx1C_m&a8G(4l7rt@EbT6Ql0QC!$F+YUo-PAy`&vTc07
zj9v_*<V3}f>7~SSmJ(1HX1j7+^5n};6cOhU?ey>(DPNviMr}%$ClA+wK<*qwr3a+1
zxaMOk-Dy#lt(FQ!3!~dP=Ac}KG)XtN{h6sKJHGS9;5d@D*pMQEw(%T)@T_RjbP6Tr
zKS-Mati^K4>HiYi|0(92u0y7Uvjk#Fhjm5F9478oyiv6>WtQZ}hV8;YfjEk)V;FRf
z&YC8%+qtbh(q{dgBrRS(l{&33p%4(~>j900aTVXmk>UAtL`4KVKZ04lsqQGp?E6Eb
z|HNU0&GKr?G9DA3tw4#{#DtmMM4$F!yzgBbBZ%!QWHK`G7m+rG@DN{IUp(tE)O(YH
z3SKWZ240tMm)U*3qn?p(x)$>%5(s$h9}{Bl?aI(r6Y44?sU%9p&`5#V1j|TXzFljX
zsuyjL!npxD%FXEfCu@jFe^d=TCeHn&%M^J47kb`;FS~ZsTDV7yKdoV|D_@%IX#vJ_
zN#)fOqj^u7L&uN20O?^8e%U;V5ASb2u`nY(Sy()0=es8(DYl_&Wot2%ZBq{87jKs<
zzFcAtibKjKwIpf_0$wW&Q(SDIfO+cznp1dRJ`+DvGOX>fF>5Fn&T!F}A5uX(`XjkL
zi~=l(EIT;Is&ywLeoPx^=6VnE9`^?Ca$4U41&<2z>AbV&GtQ#a=bk`xsQ~^%a5qZf
zBIy#LHaH~rum=~6N%6l6`VL&sQT>$(#efV_{;=hK{k)L|bMKpKVDJ6XolY0Gol@D;
zZF&DEob}tuT;SS7h2|UIvRn6J@{-=$gL=1Sqh>G*UQQWcI%%<3fw5dXe!@vEKT=A9
zr*<IMijN3lmzB%MXT?%Lthw(G)y?{D!`u`TTqvw0N>%2b?_H}tA-PAnlVTqdxwG;|
z;hfr&<Xn*TJ^n}VE52Xh-zFUmQjl_FM6e#mMz8Zj)caB-8=qVNP*sE=^%z_KTo*5t
zYolQIG<5rrhM_B1$JGN3a=0cGJ0$4_6l{c<GNxE`xwo4l4T^T)W#(fx%S%G%hCbNJ
z$1m>tY8GiALJiX$Lel`I^!y#4R`6=5loNJq)Ci}?d<(abX=vwmN(#9WD-7il*dFUF
zcX^ZNqHX;<Pay0I*Q$1QE)ggL*>o}SfW0ka2tg{|7sI$E%rewBUd<}S+>W8@*;OP1
z3^BVMYQ7hYcy^i|4JFuU&RWtzG#*b3cdmEYitzpHn(3S}UZ}##PUN8CPHS0cdB8!N
zH>@VX)DaR0jr{{};265awS$kOgg4^*iVc4|8z>OVBLmtmSMETparT+kKYQ+cp^Yn8
zuU0|h;P`b2pv?npJnZ`B{;d|7D1pDb7GX20X&S#&VVmF=yZcP!YdXAm`4L!!{2i>~
zES?sHHUK>8h3@F?wXv>tMRgcJ2y>n<In7$qLo~GS8iF~>VSLBowy{04pem%mh=K_M
zPUf+u=N=g4NJ<C3faN_53PYN(0*gOFFUaDmB#sj>2B)ILC)KXFjZzPVgt3A`#L@O*
zvO<uq@BAVa#;@;UbUCCwCa;kD)qe>x@gjG1BohV}t+YqPRf9yBIQ+!@H>@}&J^?mY
z@9p>BXH0<RV|ZJ1U#oDkQ|5#dYQ$eM=O$MGNQTJgvngA>V83wiCdbuovy>x&BafEn
z>e|vNDDto`{mcMKdD8WQl}0T`H1bxHeNEvEc*CdmW>v5o#WfS-f|}Z*AfJKg@*h-A
z7F8<lYj|PImIt#-ja)2)1BvAelUB|4oqEGJkb+*YWnrXr^8jE)Q7lSU>nD1S93Yy5
zpTZZ&*SeRf+l+6mp`^D?=Z07>w5QS$=H9~&t&6z3L<#`Pv-O#MJnQP|%`O!~%~C7>
zx-+|xFF9{=<}kcg>3Vu4O&NNOx`3x(X+A5f;aE!mf|?nr%3*15XqAGac#bZhHoqHW
z7tVONB<pOWo>}gm^V4%wtDUWS)DkctEzLqsVqHe{P|s%Qxg_FfW>T0n$eeIoWN<td
zSa;?P)zQ>x_k{X$XwFv9)cN*jjRK)!Z*ksuVgEuR3vp=o?)*@n_R*+vW&#Ez!rLgf
zdXV%90m25N+_f*?o@)D)jZ+=}bu_SbD$J}3N?LPV3fyHo4Y+egci|7dCAHaH@q9ph
z7hqP(YdfZPmo|~!+lo3_t=^+75to>bI&I{~ebhtak_2zosSYb22C!+#Mamu}0LA9f
zwxat8s17CGNcyLM=Hn?MaZ`+v`moWwS55RGC;}hUXU!;g6h9ZF;(xNs<EA*wQ`}<n
z8?a>KDazR<xm8Hnr<g|`2SM8py?xP}8H@YN%Gof?Ss6gudOri4#eUuiG?nKnp?6uo
zp1l}1wlitiDSX9tt;0mIV_7PxXQ?{2Z;Dev<Yt?=HvZwS4?h9T`T>oVhSQy#JTzgf
z{qu|E(KycqmUmM8kk-$4StH!++0=JeJdu=9isQLMpCa;9X+NH^{yd4$--CsZ7cv7y
zh$q*oCYg!0-g*pm^Dh?SyNLox$X$~ajVW8Ec0QW6Pt3cktN<M~2YBy@7pqsKbH?*m
z?dm2~8^-O;!URvuXZB3?03`V<jfdeKqEMMDBdCF`u@{oZqN6PPwHtK8Y&F;ostzB(
zvsFpr3%5%JYIF1a8d;C#W&2078uVj=+fMyZ`<bC5xwWj)YoUw~eu?=sWV}QuV-P_)
zWf#ZV(&}o_;JGs7?i;b$(`drd#)5b*sf`eK=D<=3fU4NyK%w$sZpee}`t87nPv1s$
zfY};-`*5Kd*nv#AU@n`#dEmR*x~DMWMY(%vw)$s@a0k_;uu;W)!dhtKgAIkibbxxY
zT8M-u%+`H)JOX`wgp(-h&hAjpfxHt~Me0;<ZN6W=3pAL(;n6`Uqc69K%wBh1z8Klp
z|K;pln}~u0*gXe7Jfvrz=p0o)YAOhoqF%V9PNxQ&7Vm?9We187y<xHhh1+W`>h)-W
z!(!)*7BFP!&R4nGQVH9+N-McfW0mQJJkhUKDr0>LUzU;7iaEZrypwG9ND}E2?t35>
z6hWByXkin){g#c0EMxX-eK4=e0O|xtHvbKv^TVS#6pNJE2xr-_8=k?tU$q-f^fS^S
zQFQ3Rn4;UEj2q{Ouacdv$MjVh{HB}CVE%Y-speMIqw3D9c1R(z$!P|(V`%NeST}60
z)9)fdRa~KJ&Dvb0w6EP6s1H52t%Guj;ao<!IIvJETTHLkLcA8_j*u*V;2E$Q;fd#@
zdC?@K-C49_?75t4&zO#OX*A!<-GGE!6FEHqS?OC8d?miZR?ouJUgj3}WS@C;h5cU(
ziptlpoEwnXf^oL9cpIj^1W7d+{$k}PMclOo%%CYCo2&Q+hTi1FV7T#un7P)K0S8JT
z+Rv5bxF*_p^s9M)5!>z(YT1!2%mC7)dwD*sszO$IG{+eHpPFT`zM#o3pMB=}gMn?g
zMDCduhuqnqdGm+#@KXf-4SLIYL`>GybwJN3=NfoO6JHYfldpUDO1h>bYBdS}@UbCA
z1N$j;5eU<JfNID|BFMe8qL|M22myo|U|niWXLzYi4;><$qwnqL0+wK;bdjS~&?^Tu
zJac;#CyQ${rV@=RMnXAqh|o)r*1EA-mLb5|8YJ59ekW?ViYhH`w{=R;0pI=(<f8=Q
zJB0A2M5&<vn;`TqR?LY1HB8+o>gSGPJT7EJ{5V4Ha_78mP%cm|IiM<$e4?wi9|6cY
zlTP(DN32$|(zYknT=0t+l+b|v6)4J0SoLVc4Zf%F-T4MX?KMf8N}^0XPUfo;j+9MH
zn|}xNuT=hS5ETkAHaG+qhRi!b31}cD(4f|At-JWHJ+XFgvU7dlOVY|cWAuWT%g|_^
z<kohs?iZU-m!}VdJjAa(OGW@V5IX$Gzk`&;cR$N?7BrJyJG$XH0~>&0owgor$vQqN
zZ{CL&ddZysa<0`lF~qV%sxy!s()#@p4p3u6SuOxapC2~fqKh)GD%b$^E#Gy+sL_m>
zeM$7q<l@f*x_tJQ6Bzi`GZ)G}Ae`vl{<&N5EmOYp2!N&lJzE6kwsK)*J)X`Y4=8)`
zfv5){n5)w^<lcNsPPRypole%!Rom0ik+vG!67!GB-g;&eRLbW~M0@ul7FdgY2Aq3M
zEfp0kYZix8NF$(*hr{t*nC$$t9N5z8ZweYn+Qp|kuV>Hp7+u|F0m6xIV;Sg$%1VPo
zCt!qkiA=5uOlK^xo?2Y)*S*~V=r@36!_Fgq*_Ef_$kYeanAi0CR(NJ^IK!PY=%RhB
z!@#WdJ>y+J<-w0UDY}EYJ&SU9ms`^delkAjtc(*n5xT27<-HS9sP-b3epgnUCr6V{
z2%4}~`dTJps18*2-S*O3*0R@pAPdTO9l-1s+W<Zzpi*Y;anR@S0IqRES5@sxz<pJj
zwwexcJIeAL`65I6{?TJE;Q0s4gGx?U;0z@+b&1F0+tMjJ#^5x-J5V_M*#j~xZ!arb
zq|U)~ljz5CN{EPGfz;=t>Dz<kkk#C3tRGzH(tlLcm1oN?NonxF$6q=`TE+Vsd1mCl
zt=#2YOH2ShPiOeDx+YJykTbmCS^_gS8TTRg4{jvKnR7B|-SAGxk<j*DQENYEx0h8I
zj|Y<K8c9^l^+(-ucn!_n{gO$E-PBd}(-FAAx~zx1a%fQD!;o0L#8;LUwXQDh2+uVj
z(kU&ys)Odr%;Z)#5>Ed>tjbpWkn4JMs4Tgyq=<LdJAV2wWtgvxy{I{bY<!gj$$NX`
zv+8?5RKK}BgdR7ZGPZBrmH;3x8A`~)&*ha{Xz)V^u8ky(Tz1$TwA%gPj$qd1(R8s=
z1HY4R-`|;x!F={EsSh{EyIqSVgJ_vs9q$Powcw8vF+fSUg|bgxuc{6u1&Jfh`@!k;
z{%hbaledXn!PSorv&D=19D+dwAihQnYdP9z-2trQaK?h-y`hH-lvp=yLC?_TUKgxl
z$`re+A4r3Yg*n#M+^s(6RfxRmO@yXsjY(+2bAB&0AEwyD98^o;<A1~}n-xFH_&>Q9
z8iaQeUG&*%ngR(-^(`yWP8evo5A(di3SxU?1}9)Dv|~Ouw*!9frDdv{NiC6=4Vkz0
zl3n%fO69uJ7~o=$MsFooxg>1wm}Sb0U3)i-pnwbp4{Qz9o*}sEvHr5Y<Lk*8V#Q4W
zorNqugn_}}9scKm(a_^kcwUuRq<HHRznt!EVo4_JCVvS7EW0#jRzJY^vkJP%zq(Ug
z55ssnE$Lt<`CM#I@q0!zW*?{_x0`(H*^NW*k;Lka(~5CojIrYxGXzb24p+c_T0cMk
zaG?ux=xQf;V0VVc95K^I=8HKRZd77om!Ib(I(M2$m+}T;#0wesYt}yKty2X40|=X-
zfDxPEquR_wj1Wi@TL@Z$*_INW9|5NsUetSP(Vn|-d>&%`tN3>zuN0l(>0gNBBTuN&
z5=q{JYB-t6M)yn!&g@b5>!+gS&Ia#+%0I=lqVo4DFsX0UhzCSj#)gR)uEBl;IG#3G
zZ?%int|UmZ7s5f>Lin|<_D6n_C2qJ4Q40idVlF4rxkcAkrRT>Ld|d|7oz)>a7*17S
zC5?{$HS5Y}jbA0>w|k^+4}~xIh;l%*9uVOSFG`qRB8!ajzqrs9VhX?T2wDq-BlN!~
zQXa+dY2hFITBA$k#wwySn@~;~x)FDpL{=W4jgzSm>>F=CNqrN20BLW2);rai(efTe
z;=fPM?J-5R4Aefb^^c4jI<;1D_&*Wv=soNNy=8pbqx9Qmfim5TJyXpt_jkOzzh-RC
zT#)NT2PtT41L-U50JQ;%8(#P4CmdJTX?j?Q49~sh!q5RVx|(wk-io~P_+;-Rq7A%@
zCE8wo8v>^d)tT_Ka;rB>PDvD9>o8+;(v)~-nyC_Rb`)Cls^g3$WbccS@8NjPXz&15
zu>CXSXF)pL*#p3T_W|RD9tZ4SeN|IP@bB+nKzQ|2L+#!D2m-dIbt_EO(tQ{)Ku|Y&
zPUup~tQp^Q8RM}EzJK<I;E2a}SV|Y~fLv`3kdzH;707j}A|b;$*sLq+_@(qiRt6xc
zCV~x%PX*72Wfdp~Zi0!-*1c1hsgE9V`|m1_+Vx;F{y8au&$YEc@sK{E8{ScvH!vGE
zp3c+hk^vX|JfaroYdm}=%jUE_LPV{=XAYJ2%^dZ5NR@s}rMCe-z-hW#uqtczsrY3O
zYP^xn4Oz5;PW%HCa~E-0qZ^e#D%ker_3Ta}a%yPWO?Fuz)?Lq$A8r)mvr*MDleJgB
z73A)*+cHADe=FRVK=%lc;uDhxqTp+><*(qY+3!;6V@T`WMeR%x#@}{l3g;kc0cGP_
z|86+qCnMfPlYvdwF)8*S158FkbrI4Cph)WLEq7(}B!7ZztIfsEZ6mrwD!~PwY4{Kf
zsfX(y*`bz$KbPyAHTqe3$>xY4OABRtxJdHE!-q{Zh{=Et(>2D#!rlwLhVOYP#P%h!
z?6@|riIupPYk}fzWTWNSAM}ePbK_szP6bx;1)k;BunwWyKmoLI0Ez*(GrxuWL}q%3
zwGU3^X~W@`@6=cVELSl!B)i$G?TIw&Yhw~?ImDxGv8a=fS`tW<8;}S!PjIRd8PwL*
zpS=uWcQ$|u-3~Q-)Y(ILk%7+9Z`9Rp6;!7JSGP(6LHjy@T-@+@Ph!-w59{@*`NkM^
zz-N>W#nh(e<5cOxIua4T*B>b{e$B%B4p95PF@PGphud91<bV#3)`elI*(NQ6)6tcu
zKmyKZC4EbMDx^SQ;m+Li&9SK9nfcxetY~$hNf^Ey&F6y|upDeBV7r3<lX$_WUuQt9
zr)6wa4A%E?x2C}3tD~X37ND_((mU?2k|6JQohgZRZ;9Z+YS|>R3#-^X&8_Z&tpUh0
z2OR<y>24>VcD32H42YTQ%CJa^siSH|!7HpWJi{F4AUpzDvz?|XvUV_0gy4#_v9Tmb
zR+@h;0-$8A`$r=FPo%K^%+7$CI<VxkW)$|BLmWQ7PkYL%bGAF9Sx`Nb&evt+MNA?)
zBP37c|D*LP_DV@35A($Krb`^{L<{&F+h4%7`Iqs7wjT%&_o{X<qgvV%J;zV~iXtGy
zcv?4}qaYo4Jhto(QoSe8Se{N;Q~a3Oa&`D?8Rc|sFZ9*NAX8Rf;v4sqx@3ZZ($(M{
zz%o!^p-jrdZ3lqg!8eM%`tl0ZUOf4-j1M3uc{S4?fmo;uE8mftK8sGerkh-FJidi=
zYw^jG-|$qlQWXF_2N}Zh2#Xfj{5`M6is|wo50gTf=50vp>t~0>G)F(}_aus69t1<~
zb^j_(uE9o1u?Y?TB@6S2p44_K`8{7x!>uU}DT!jFY}RxI(mV?(l$?l+N6IraXBlD>
z+rD~1+a&siEoBGS9U0z&8L338*+01L@q)#~f>v3`am%VlfM9>^Sgs;I*4aUuGcf2g
zkaWTP982ATM!s(Lm9}64)bgF=Hi7{w;J-4`NEC|j){?pxZ8ZlbJD|zQ1Fry!4%=}D
z6p-64PML-Fs;4;k4FsUItxQH7IF}#2>x=ts!cu@q671nM0)T$D$;&c|sV5}8&IZj<
z$MKo?mQ;Mp83Z8IQ9Su+@^8|&0no+?ZO#p?b3hk@N#N`ok8jwxlmalc`*Pk)V6|$W
zMKRB^BpXmRmq&+K<fpl<xd6TDiQmHY;H?9*1dRP7WyR+mRN0c{UU90#$e6}op=fKm
zyWc}0i(il5q|YM$(pnrSc|?|hwnQ(Xww8TM--thDx`HMP2%5$KYN#EDe{kEG_2lVY
z;@eh-aY1KamAMCHmN+s+SI)tDu61FyTkI??+wb%Yw==%ut)9Qvh!yj}*H^iAD1IS)
zYTjLyBo?uti6U`26!XyA!1l0-fK_3ZgXB%*gdiVs9R483G*>ozruS!B85LY2J`=yh
zxKOr>+0NF-7MS|Lma9FI4g41JPzk)wH2)>%MDJN67B%E|GE^~nQ%p5ysdI*m6WHI`
z3401MZGGRA!b*zQwpoA9fpoXUk%rF+cH_i7h^y3}Nof+Asx^eFEgSZqg+<JUcAYdu
z**cWZ)eu&Dv}>>+H$x)JLvX3W|2C7APddtOMO`y!IKqt<B!FIvQn4gyn|~l8GL{Mx
z`3KXrmP2*v(+At*StT1OkBEX}@RO9$m28<UTPs;l_|@!(TIv{Bi{Oy6l~vqtjE?jf
zlgb$mJz2H~3@f&n_|L7s6K~KS+~AO?zQi|Y>;eqanIsd`*$v-{uQs$ZgUj@7O$3(_
zSdYmD(M}}gwQV>U6myUjzy&&f8mzx^TU6PVOb0R#Xla5u-0}!%20MHmD^T|jPI|kL
z2h$d{4Su+(ggTAVzkBPlrc;;WX@S_;-avz*AdU*cUxq4|Hw_*)#Xoq$m_6q`&pHE}
z-6V>xh{!XHfN;Kc48Cn~i#s%h8wluwe~o8Wmltau$Tb>;&U)EZN1i8si<iW>Yi=yg
zlja+m4Z7$q+i82)$q&v#F*%0z9>r~gLdGIDSAygJ+L0<OZY*<K#V{7|&vQzX^S8io
zGspDbUuq@rCIQ`xRMKn_{W*G>*nIU1d`3V-Lqv>S%4&4ZIb+|hinU<0WCb8m2`O8j
ziPXJm16NnDhp~b-2j~PLB11sUkz~>6Q<5TS=&*U#7onU`U=-_v(YTj0J0$92OA7vh
z$QPp33*F8zkM|Aj2T>^vTSw{^T!Y1l&_?*3J||v7x_qbUn(Xa7=hom~V0Ki(Y`2W{
z{La9bxsqXQPd<=g9$1PSJqfjSt>G!HnWX@Vr+;?@aD91w+LR8bh45EzJ=2n!L-tGF
zod2N5yQX<o_aFk`mUiF)AR6fa5sjQZU?wbaF-~>gVxL2oRp{!2l-;ePawgdI|9!1@
zdBi%`J*8c3^7Ev?ZW1L;u&q&q$O&5~Enq;Av)(OlvaRVM^-tInhu{G(CPMCo!z=vp
zz>Q^?WTIZ2;K2Us;c4}BrE$LPIyDxB;BfTwU!oRO+zgf|piIxONDVyUkYwv0aUy|C
zK6pxrIw%@|j(m+-Yz7cC4pl2?l%6<39CZsJcVvoK*19!LA(Wmv&G?oFSJbCkjeVAi
zoxrL#jLx#ChyJC@nC?_KL6{h?KB9HlETyS)!zu$X1<(`d%Pwwq*-w^|#Rl>V-`+6$
zsSY@f)``m?YQoI>o1`YmrVD|-HW{?%E_khGj1;M@x5RlaNQ)&Nlq>kB0B5`CEo(sk
z48v|@v^A9L(Q?mB(LZ@M_-oeUX$uOb&xYEk<Ob3>HEyk$ou23;Olk3w{K_bq52$%|
zi-Ko~U!m0+fX(RYl2I4mmdOA$%rG-RO-);h+^QpqT4L*Se@!_}wpi}tol41aHgVuv
zG@dx}Gn7b==1?I_bw+=FpAI>ES$53jNerB{P_ELV+h;98$MTObHMPCz3!UG3(4DZp
zgyNE#*ZODpZarLn&+^7cqwEhlmMEve6>n*_-TLf?qkxNs2_HQ5V%?p)=kh>?VHFTi
z8d5c{>($u&?PHL;N%EaZ;sZk>em!Gq@N;&-X9I2LkS2wUXpC9o%7hc`1@M=x{}BCR
z-A_MhP(@DH2orv}4LW*ouWy2ik<LFNp|;5IvYot%<3RBkn^De(<f;<4<F?2DK6u>N
zMt5kZ_|KbQDswt$6PXLyr^fgVn!)KTYVHAG&|^YVU)Z^96qhl&vPQabAYkX*09ecE
z*qo{2vp|@?SL3&ZVR_(AaT&epe-B=SbsR*to!7-%&u%ZINi+gt8$4sDBJuO0!xIHu
zG03HTPN=;p7c`Qmb80l+#V*@?O|W&9H|D@XPKWMfufFDtBmV9@55!h=;&R<>u!W;F
zo83;(l53+3$jo}+M$vx)Y1#(Iii(6l|L+J~v*oA&Jv2cA3EbIWq+JEAMVju7*|db!
zb_P>%9%CZ23xP0gT-7cfJgFXz^8u1iU7eIL(7^>N_<xAySo^o?_ccl$lX;lA=O+e4
zXc4f98I9k%2zp>ZPB_v=<T6<P(>h@TyzT=tKn2lNz%E=&VXL^+V&=X6q`b_k+2$>E
zTW61Vi}R8Xrey<kr=m7y@$0M6w7USn*Ug!jEsH(fEnKVtbbY{{Xtr2w4H0q$j)<Kp
z!Up87yg4HHZ_PE2nJ<A~hP?$GRhTCiD!6m1?v{FTNKhR9Emjr(mO^T+C)PF98NRPm
zfurPF0b&5_D5?(dTF9I<X#zBgz__yJqxQmvlTq@SkjZm|k@V=mpxH?vy$s&F)t8+_
z)m?L-f?rL6H=)QOTH-7RwHr0Ls668vc(UcHaX~axC-=#JFHEPR@3LLCl~skPqs#8z
zKD=Uxm57^%QU1jv7Lp>UFf~Y<M+x0CLiPgbK|c=(HEzr%du`0U@4kaWXOjIk^bNqQ
za{E6^irxTt8x*E4gg@;J#FTCi-7aH2w#iJ`dBx}i({@Sd+_sX*8XpJW{3$3`i7GNc
z+EX%~dF{x}1oW5vq9NK=Yn@oBP#$nduE&o&%0>&kbi5h?XgbyQ_eJ)DW<wEotqR;a
zfcp{(sA4re6ocT{kjOUOnt(?jtJ#rbk|z3f1B?05m6rn+VJe2TK+|BhsDi%l-N*`J
zynbTGoPYHy6QeBT=WB8QaG!;2sOxsFhv^*)j{2uc|Hc;UrUs$%WZ$91JeP8y9{TM>
z9Pke+x4$s!E~-{|T*pIRQzsbr)PQ38+JTwj&)#mV&nL;Ag=fD&%Ydp@8aUfK-(feJ
z`@LQ&SUtBd9P5-}t}sb3-IW1G@Wl#MaCW*NSw1;t3QnSRVvyk-xc3{?XU0}ubN27o
z@mAxxDF?i6;6uUg0Fcf+P8Y)m+aC71jHMJ+WLNu%se^rNkb5D~XQj&!-YWp1*;^jB
zr}?5WY*Ga!4?bDc5N3R}_kD`ER-@JUsReWS#Y7f>vyPNJ@50`s&!c?4Gn(fxPL_ZP
zo$`917~oYji~G2OQw+uMjIch1Q$n<K=Z(#Y6pfqobozAdbCu*~Lduetv~(dcwC0Jd
zzp@NoiYPQ4HZG>Kl(BN`FRRi1<8JeB5*OHRKnJ1k1`5GSKw>di8ljK*bizc2oW78o
z{1pLQN2|P@ItP|K3_F7M$V1Sv;kTx5;kmx}vPtFU9qKl8w+sGI(RGaRtj;Y-^&d!~
zFpzxDTe*&k$?9Rp_SSoRSAeUp&weHMFU_-schnm3I9rmN2{}AR<hyHp!oB7Hj^=$t
z1#Ulmci`Se^)h{`^FmQ&Tu1WZ)w`EWcT_Hbzxr`C<_CUJKy_?N{)B!5aM3gK#-K=%
zJ0qI&B&bgIB85k5=e*iqQ*xo)C-tShl5);gi!DrRS0+|z&qFlQoHLH5w`*4cb@pg0
zlNGl<Yi3y7ZIw2_OFTU$EelSGd0@i5RI<eX-hBrwgR*O0#8s1(4rmuc%arETz*5Gr
zApqyPqk>RG+D7PbJ<)|_*~?MTIe)z2oCM<0gS~YD+}Wvfv5<*zWvr#|yXGSu&83>#
z?+Jh_cz;(0wD8UaJuTV_anK>6gln^AM$RtDl|=Lp_@DeSg96^M@DG%Dm0u58i5+8M
zI}ya^5HO4L)c`X(n?c7#UK*|cYaA}-s`?XEEB#nkcV_g0X(RT<_!RL0^gWDcKgk+Z
zj=C}SjVR_!eiFDU#l0Q0xWxM-?GNbZ`Ef)z>i&pH#IUC-E1IC5TwuOMRJed2dJDH!
zg?U+m9N5YCpEIU6L~{zzs0J>00y05YPBx>_4FiA<+6~Wg`F$0%DYMCi^$Z?RtC$7d
zFs(W^z69{M9|wQXTwv0Z6i+Q%1g^X`D<9yW|9wPf*6-l|W9luSs(ilhVGsmGLRz|%
zF6l<3K|q>Iigb53Dk&`@B^}B|;L_bKCDL8e-Tj{npYQK|pS4^|!3)pSnSJ)&XNJqh
zga-Z#;3fqGS}0U(R1QR{_Q@{d$db0F{zLKP#xt%}R$5V3vUbK8;DT})!c-Nu#;L^V
zn``;taByQF{VyN{t^1rdrn|Si*}1&hR2l<`C;Yes@C+9H5~{yHI}0EWZ~@u7mLyIF
z&MtvON~>8=0Sj2bXEuV7x!f=wmmaP@Qt3Rc1c`#Zs$~`@Oh+!M9)NF6!ii(2qUyZ0
zgcm@IkP+HTvZ{69YI}zH#_2ySJstx)VJoTo0EEE6ZGqbWn$Hg}!2{)k@W}I#8zj#a
z+m6qEJ}Lg_+)$MY3bO8|PW)`wB!1ZXJFmIuD60}K*=)9ipq47n?k~&RpzN0Q?7-mo
zu>3NNj~d~bo&p&a{7{ARQ>nPiS<g{)pS?flaFPu?<A5std64j;^1f6dS_u$RZ?phc
zzMU;EKFAc@O1~J=L0HK&^8Q&wvfT3^+t<kc-V=!gK#MC3t_||mEnw;DSq>VqUkH@_
zS)q@-9=f*FIGYV>*WSXT+}he&_(rba*d(ovZgrK|?Rm|x@iT8;_|E-S+mfXa$})3r
z4eO$S4`JDaOTh9rL;Ngtr18~$)fUCeGSJ5?Ho`Gb`D7d6Y5rxS!l(i;k^;naCW_a4
zWK3L*v$J1r&gbmZgL6`!B6K7Jm%iIft^QP0Hmx=^Nj=N1;T)Ym@g1DkPe+E(`BvT+
z8c{}p?Cjeh>o-vbk4sPtB}v+Swgk4_bQxFDK5KB_o{d7P<qjG=UCbEsJKcqwsMQRf
z+X{9}K3CQhquDfv0NF6gd+ouD!^*AMZUfR3BNmH4h0Rp#>1=3fz1st197^beoSZ=z
z+Q-)r78?(inQgC6m~uL3EH(^JTWkYx=sBYih>Yos`FkTzXDgoTTsDZNOlPcMTWN_j
zHty8d)ZAEyF#8FddaJe@k!hfk>-#YWW>Hh4kyT1Jx!smi)czD`Hv$*D?f+#7f0{Rp
z>_LkSf0UPvdjg?k;9L}H@MJ1yM;sx9gix>d<G%<Qvzo_7o^~AlD1$|pdxi|rLx<rg
zyQ`I5V8I&P@`)(OECPmE5D-qrT#QWhadvc-8NUxxS;cm}qp{Mm_{%t5pNU5jRq@Wi
zh8JNzUa$OHtVs)tnbJJYlms3pvRD(_tiWfio}_l8bB2BQor)JSkz<nf#u9f=b53&h
zISiAD1QHqo9`mE(nD~&6a^!jfw?g`MVJ5m}^LOA7O%GJaHvFZ&A)~<_JQauL*$7QC
z+Ik5+5P+)0ef0jpg@#qG{VBMCOGj5l=c)g#&*h^X9}fy^#Z$@wT54hv@;R-51lQ59
zONPe+LT*zy%g;L{?bYsI;e$M{+@Z<#|Mg58B@;IGr;lQsbyA}T1e`k_7Hqz!H!mwH
zafXOuP$9Ue)v|_mHiVuwiU_Zi8gXr-@SNj$-@HBh-6h|0%J%&y-^px->D9$d7B(?9
z_RyM#hZzg`jQ#8BsNo;`zgfaxGW6m#E8mR?Hb#VZ1GHcOF(KLM{O)DA1fGEb%8cH#
zHO3%N++KHE%r$detQYS)>&%UFxAGIu50;jA`3CsE%uOo>mfU{PV^_|QBTg|2Ae^xV
z)7pJY->dO5VfbEUi1-FZdGJKgvUpuW2U%KD!r62*fqTi!jvVKd+Y%rC9|#EYr2f|j
z;t5nb%`#8@e~FkOON-yVO36nt4KS@j7my)-hdat;M{<ZT8#=+)Jgn4VHrJTE#5I@g
z*XNQysx0-%5h%rqnoA1^mjdK-skp`HzB>t!uGf=ln^KcNXx0z-tIx)G6U6eALpy!<
zCJ~)a9VP^*iM?Dk-4+`(MS`JcBp{dLVNKcbYJGUuG0J5<d)m!immT)-$x59W^FBI0
zB;<8V`Uw3AnQi6VVa1oqg*mScyiUfcLMjqM?DmR|@4oH6vO(B>7aKxXt2?E|*V2TG
zS3e+w`RaUbj=Apr|8Q-dA|q}SfICf+A?Mv^J3ko~eG4Wc{cL}Io(EGeXjdYfXilq^
z#C0M<PqfOIv7siP9lo;Q4{?KYoMQuY<~K>?{XBJM?~;>{pjw#evIH0CU0h;CUXm7A
zmcDw5ENX^}{cvny_@|}6RAEcYBG>a?ZLNW1>W2`LnoH#gTXiALEDb7teblhJT8%I7
zZN*?!x3d|vH<6T?FSAN&DJr28pY)UEg<N|f?Nw$b7VyD?KU^H^|C^#u4ez@{uxn!v
z*BBoe(Ii4xq|?iWwG6_CC?(OlI`#ACsn@JW94FqTlC3}}PwCM#R~ETtdC#Q<9wKo$
z+87%H&S`<*g?(K+X*1|?3<b9|B;KNlPr2mD?pp^F)o1PToEa+nZs>Qz5g{CkQuuzy
zKqO7-?p2&=?`P?t6^fnlwp-1@<F`LGhu%MKFV563Nq!h*$J{xa)Gvee$^6Y`^h3x9
zBjt-`zi4hI2F%jKoUa3#r+=4<($qpO10EFP5fzvki+^XjnYVXZMiH0bLIzW-VSk!j
ztxRs$Vt`S$Gp&J#ka(6}?Cv)keUZy~9cGPBbH0Il_XP>F!24;M0esV20d&<pI-N?<
z<#P)i5lXvlX(9pvn`x7wlaUQFSfS>>Ei3ujwz;yX%(d1CHT(U=Hg5sbo@w~EQ;H_R
zkmKF{M=pmn@Z-WvSu0WJd`&1$Mx8@Db605>x|0dP<dgVc)H3Tg9}JuF5&L&`Hygm&
z)ORoOc-Q{h9HmusZ8=kia`f?atHNSzDP{udE}3oEUPwsCFU*3!T}g&$vp;LxUlWi1
z(ho$gx63y-blAv`is=8HQ|j2ZZPaI~R;fe7M@&t93u33z7+FV9R^ehv;9qy4O3#<E
zi_T)-W$&8K#q$i~k8aXPGdx~_Cjt`%1X23uN5n;x1zvknqs$Pb_Z|iOTu9-si*|mf
zZrO<p5oCZHu6)SA<s3(#&KAL5!$Ls6!H9?eFXx@g*T=ndCi7kex$}`u@~`>d5C_Pc
z{-Zw`ODKtt56K*W_oZL&>{JfXvHMNMi!Qma@cDdr%nk{Wic)6d2Hn5&DDvao7v6`u
z?x}*m1t}G3fMwsjPJn=37~|rrj5eB34i0S8?vx(Zw~?lBRdZ#G-2SoiI<55+vap{5
zE{0~UB)Fr3mpCwmAyM;NAny)8otk*2a%=)l6y#i%ZyH>K!uNKLYyq^Fp}x4D*zddw
zMAk8jy{*D<<yecb>GZo72$4$crtH(v`M;<>)GP&$W<(xL1&x#i6FX$a@m~*YE60as
zR`-16etY=Xr4GCdp87x*Plo*rL)qEOMTyM!<R8_SN@QDWa$6dv2$mceyOw;Ba95&)
z#KJLHu~&<YCW{Rkg)MV-+Wys_7_dq4@5W(hjgxpe0t&V(_&<l%-rX%$siJt_zHz+<
zd2|n6SpVTgdSm7YPEvs6<0$E-O|Q*sk&1H(uW!=gcr)<W7BGgj*iKenwVLDe6uEiV
z<3MT@eRz-<VzJfrm8Eayk5%}*Oi0M{DJ9qW+S^I2^P=2215Q+?JX1;AxA*IU>b)_|
z{k-^u@Bu0zfyZ$`S<on@aXvGtU5>_i7>gH`(CHpsp4k`JLCz2o|BOxn!B%95XKV8z
z+6i`hpc@uX=+sIfftuUVdCHQ##~{;_F0<ko?tjKF<QA4k^#u)1L+BSHd5#y82X+Hm
zS?b5G_f3ZFj{^fAMTXst0lo7Zl!&Un|IJQ<yV+?BKNxXdOe0Gyyp4~{F)pW~KJF#C
zap5uAqK^-xq)KnXwte&<qY|c3spBw|m0WVQ5Vh}NGnO9_Lr?rD(4R&BmK^LG7b#;F
zF7zkE5!S{L-2AXPY|Kj38K3?~kwGCHwKe@B(=b!Q->}oNIV}<FIiaHG|BK^jEFLrk
z?K`zNI$RFb!e{gLxFUSh=m}{9^!Kmqc^}rZ#?S#(;B0wNz|JBjG&(p#(J~~d?~{OM
zJd&?|Ho5q^eaXO}6gLVg%9GvEEX|wpO0wYLvEKFBf*&R<C;GjddhN*S@jJV6yLYuS
z?=`pbO5q(x*UoUF^Sc7U3<vv@M&3w(&gj<b0?5UT(@34r)zOpT43!_$d>nWCP39m+
zcObBrc3oVZ5KkX3ihw`*b#(<ie_E;F{~?p|J_5fq&v_Aag#s~=_owT?c;kzo-4@=d
zh&l$wC6gLu3bI*~{Z3zBjx=mIqRP{^!Trrse>xHV(5R?#l={dins>vYJPx-K;SU8F
zeWn;gJAM}8H*B}nT6y16oMOUEzm0bhTfuPa1EKXNcW<o-*6E0qTNZ4FkzCNt5rw?k
zl=ZJ=kk(xkq_5~Efn28ARtXR3{M;LgA#m{WfjA*czrLH0AMW=osUk988Y|zQ%w3Zs
zQJQXJ3$V_LcOz*$3)kWbRHQa%BQA1)-C{seC~LgzA@$z8l8rW_#2h_EuJ6C8u}xbk
zPn*&Zwp@B1R>rc@AU4{=!$dyyHgYO$Dncm=>tnqjGiEzKm=nSf*&cgGc{TtmsWZT4
z5K7>r=BGu(s91@5T33dJx>3_^RC*Y<f0|6dID2`wP_f$?JwP=pG;+<xB!2F!pCt3H
zRhcBBCR6~CDI-H>8pJe8p<rGBudl&q*QBA#`~p^F=%rg@85>^f=Vu-?R)&*Jr-frb
z_pjx1-@E$4&a@PQM^>vHlOr#ADtz6uC(IaJGfTSGPu^X2V>U(!*DN6OklM7TvTaFP
z6=9HRSa1mF=ZOXCZ=QN^xwfx!0i+yySqLJ@C%xeI$iv$`23B8Z&K<86r$;Skk4w=4
z{VA?jpNa^xaxW0((alecgyYLw{nr|fII*np%E>I=uF2d-Fh}zHF-m%Qhg*!mw!E$A
z7XeiX&j&%_9p!IELivoG3)UYo<?P&I`&te)%G1Ag*2LM<zMCKN|HJkttQbpKcoa&#
z`%14evwni2w)cbI?k`m?nCWv^1h<AsuQp^NTMW<lc2Y0+pXzSstnlBDW2d9i?h*@!
zd#IiJX+D2WlMVkYolkwGY&Cv#z<fW?$KM~-cnAS+D`tXpNo?Z8zt;wMm7RA@h;W04
zQ01;Q?q(9m%0iihXaw}eTIk=DhycOy5kG!J*+Yw-U9Z<Qc-F^bUD^|-&BwRX0Zccr
zhp3xgZA--Ea?`%HI^4S)c`xl6t=CNpwLiVU{d4vGwC9(0SyIkhqnenEOs)7-JUZJ$
z?nwHpS^n|talf^hEOX9I-+b@BQN}ByVOfW{{IPf28!y6=qwYdPNC#5ZSQs8bt>9~#
z$BGS}!xnW6N$~H`d){|#RsuInKo?Av0_(udBfV*l56BQ+QVXJ-AxRCr1egi87BID^
z;R?e9I}zul?MPcrdN4_C*VIXGLHumn_RS;lyV^T5$4I27_B09D1-OkV*YrZTsH8K<
zC5<=al8BZE*6|Z60f}}Y{kR&EDXU(KFP;2lSJ?9+^{lb7mzpbc?`$_f>S`mEX0bsr
zt0m6B;3ELM|J$CFjlZEAWyO@wZ0+%PB>y+b3GcUv4-X+L5!tz_5l{yKg~1g_rAAq2
z+kxoutk@`*NwmxBAjigBRZA1$(dlqK^;D6u-wKhfykC{tJXEvO=d8Rol}E>yoLGmn
z4Bg}cPWe(%5`O>2P7Z#evJ^vFba)ADxE;T?p079BFo}9oq7|HCa-R#|^f%Lb>rdhT
z8~WI3hQEE5rg_Q~dA(n7SnTa6_3<8WO5i~rqeQ%)!^sgF<q6-d!-l^tc!t97U>DH`
z<}OROOB4u;>kkt(UqUxu%XtkAFhL7fwzIfex_)eK5mp{+b92<U)Xw())o_z(9wT)7
zdshEX{qO(;-MwM9H=RFHD3B4|kFs#HxGXdGN*My7mT!7E7Rqb;TLxIuA^{K~8DnPN
zyR35tt$=Pl2McAyI=GWY!k$ZFsaFs+htevj8zEKd#g1rAhVHa$l@Pt%OD6r+&dgQ}
z;i|(=_4YM6V~J@7FOdvVk_rEp07{N(*&5qyuWX3iHz@3`L5c4opdeE5<T*3HR7|>J
zlzB`FPS(Sa){(OK7^6yOWy~TI8mwKYBJ|z%FB2`Y)g-tSx{rr$bUW^=l7U)E7=<1W
zhG>H|af8!>F+$Vd6baqh;<rL(X6kgv0b*>#$;(B!4mAN4!4E<9c0G+_{Gwwa*Q9;#
zVigHN2J6q(8>3|41lh(OG(5h38yF%rmy2t|H~E)oXRZ{6XQ0DeKR523iQ)BAaN*=p
z)zZ0e)!I;B)kaZTMk@!8=yT}yT5I9Jjt1A0nd(erumQjP-{u=u=bkF{l7bNM=sco+
zQ9Pv})rT(}%l{V6n2bC+ziy8d>uxtooR~Jb5ZepP=B*#(m-1tyZn9uouy`yIRiMU*
zmnYw;#0&d%S%;_*d{lZhE2(Nf7N#;?or$;axxVUilI`p;W$lzP(~?dYp}><+7o)y(
zc-Dqav%^=^w(5>T0x{Cih&^yaPWVWJ#sct{-JD<;*o;+3Z(Fu>r@d~Tw@1kPU92JV
zBvcxL9TK5wpMcYMUt=QwbE~CW2Gg77W7lvpkS>L~P6U6~Wb5jTVofFZTBLM0hIoBy
zjQF2De|&6yAJsp?H~AZcf_|DiWGKmYkAaq(QED}53MT&H8^?3Q{#{xav-EL=O~r4I
z7C$6<o%2g|v@2<&F~Ie8%@HVT(V}|vKz~x1iJYmMYjNw3v!9ZrMXqd#fy>XKcys8w
z8!{F+F5{X|K};pZQMMEvK0U2bAQmYb<@^r$4XCzC?mDA$BI6-+yqb)WJ~ql!j<HM8
z$)r&vU-3daD^s_cKJ1bzr(Eu2(AGH>?|<%iJ7n3xkLCCI9Lx73VTESeP&tZGNp11P
z#LS|C1VXW+B`x_)4{5%dFJsev_R~R?WzAhu@)z6)Cdi|SaR*R{vF-s1AWX>gK`d;o
zjZr5sRyekyuQxT!c<<-ub83$Mk{Vg;EjaPu;@lVU44em?zuUY|Q0l(bk<*TdgpbSd
zIqdqUYi}ZYBK>=7*jOZA;Vrn|yAXwKt^|Ej)hcP7+w#qT0$T4G{S`<3>mHUXgusq~
z?;>=pS4*_I=VOIu9Zt0ij$SB!f9M&kw(u(f@3B*B&q021z`Q>$yqZ9*NLMeSS#drq
zba=>%l{Tcn&9;*->6Yd+uWc8fQBvy{ZT9(d>q_$s+fI8%5c&T2bPL|qcE`g(3qGHF
zP~)65M5*|zF8zOAfEoK8;V}x9$!l=4`(cpz9a770S9%X7xiLDx-<tC+iz8i0d@?kf
zE2U2ORG|Y_owm{CKWD;x^Zs%`zVUzqglgL(%_=lPFFhlOH3tT;&bZ&8u)uI~iZx%5
zKrBAP_)+aZECBJ`ef=b-$^wG@U9bL`g_I9pkTiDs*v}Zl*qcjOOnk?OsK|;Cd1_zJ
zIt&0)Bh#RQA6LMj`dJGQ{VAH6rnx3bHux5J$(EG)$i)-{fX&{p5BgaJ1ED0}Cj+!0
zz3cYgTBFeM+t%PtY3I~;Lga{|?$-z)Hwo{Q<M`};GTJUPX!SuT4!(jz%&%TXfSS^;
z`XtCybxf&G?-HNEdn|4A=uemBNzyb2BQYetxt}kU_~DQJZ*ghV<WmN`CxOsF%K1iA
zVu$0oN`^>6bogltvm%r%y<Xi=(riqezsdAgCdXruQd|V%CSs#!lt8D^(vFP;R3FLe
zo1RC<#q=#OoTavH4bBfxyh}@gFO8jP)4HJvZljqE1r+e6k3L1ISpiqbF7q|6PM2rm
zJ@#C&Iw}J}k)1_;+HRv{;Udp9y?5i;=-8<{^3jnXe&?vqtUs*dJHo)T#JM^*JiG0L
zSu2wl#Cy{=^6o~+cS|{jLP|1vglzfup5?}R=c$39|6GF`9)?~+fycNW`O%5A+1}T?
zo~bjsPyZ;McWumcF*be><kt_Dm;hAOW6N^1!*Fmh#3ZY`kkM-Y`nAhU)y3|V{Tx$T
zxct%sB*T$JkDG*7X_QYgJ!@?p7Sat$pR0P_Br>nc<H=rw8`v9Mau6qERhFDpvEYs0
z5J2KhtPy@-`9a7xce$7KA1yebAvYPl>UisMn2~{wZU~*NmVc?-pLQdRnd`HuJp9&+
zi}F;TVu3FdVQ3Hbnl@2~W_at~T2tOmd*dkeChg*4XhZetkw017*zw%qHEQDBWibu;
zFI+HyJ-zfNg8{@_>h6m#_VS^1(B{-Uy<7ZYa&M5H!jYRvZ-0%@gK60OkJII^LS1O|
zIuIpP4lo${$xq^qo~J;@CW4DZs>#yE7<G&_-*mTGxuxc4eb;1V8k!%rMM3yvu^;wU
zhVlgwtwU{Y>(8@u+z5Q0Jgx;&UKx;)Q+P1^*?r`<O)ODjTAXfKYvTMMbrwVv@1_J?
zxIK7>uTla%(d-4W{qfa<woPvLf1{G$Qq!2fpI)6L8FF`#TWlPA)j>-Ld4<ZI#xaU!
zXU8?8z<|G*9<%y;w%QlYPSX_e9|Flv8K{(!jdrn;tQ{b&6J_}?yO(_NUqlBt*uvt~
z!P!jdW3RF_82Uk!3#js$S)@j2y2?~*=T94RjhNIDw*a`m{q#FtUVG86q|ovLLZ+{s
zC%FKEj7T>074WjM1Wj(Pt_(;(vI>S-*taB&ViiWaECGX{$RHE1jk8&MLG=iIL)E%q
zTkQr@DRPYAL=bus?LG5LByUmpO);g_Pvt0=)07WzHeho{7`6K`lF^$%tNr2Lyztct
zJ@TvV;W%;f))*-`nyWe5y_HdA`{ieeD!@#0dBdA(-ilO2fp@CTI6pAf*9yt`Fvotk
zr%7;qC&Kt6q*ulTyE$IUC`4$v&+3((?}=Gjt>_Wv<mm%jqlf->%xNK&;8&-4`@Q(8
z)Q^Aw@?I)Ug7k5{8U9Ie%utYHNV0%6(1(>0l{}Dy5O2B1zQI<M5|ChL5N2n<{+O_@
ztD}B3VG*!UK}Rlhro{M{bW<#?Bz($v(WsCmXYWlpB(&jdsZjuxQ_aS^WbpbfNLkBz
z1CttWdq_#oR@!A0qFbP3RKOv<p^Reg6#D=b5)e?aT*S&K(E#NpLiie=y_6!W!gsWi
zT0XAMU3TgK_>~v9!+RU4W*KTH<f~exL}Q}mNH+f#K0&blx-9Jwp0>QOg5EZ=^|uk4
zu3~Nq+;_`TjPh*PMu>N`i<=9bE9U<0?@_~>Yni)mE;Mr?03G48e%2MYk(+z~*SVv3
zF5etwfm}`t*hnM`=)t{|a@dsyQ#w&@m^_iA_TstvqaOvRgb^7YUr>o^=BO-e$r7EV
zh375)NG4oBO?Y!?QG}5V(lZBJ^<<Z|542%8JP-tt?-5Yvu(}xhw+;O`?MajUBJI8)
z4ZXcxr~Tm0tY&Dt`!xyR$q&`HnLidQTDO~BZl@zf?f4L*=-jhlbulhdwm7iZ#l0fC
z@1GbZX!p-T37+iux!gu~s*KpSZ<N;#twzr#`+`io>Kyu5yaDAd{tdaC{^<>by1GSk
zwl4kXix;C{lZ*Qoo8&bP&o2Sp6mh@*KL*pcL{@ARp1cK27iVit6Clrf-OfgB(+T#!
z`;tM2rL~gRfy(dXyk2gJ*VCY{E(jFViYnp2MHiY#F%5y7rp@xQUj=MUi4sNzHo*tC
zyO~F-fIo6J`dnAy!o}_3Q<V_UcR;z>vZ&^5q5r*V7u#Rdj=r;Kmj8j+9);<4@%Rw)
zli~g@`3DHbw8VQoWWDSxbjse=@*aXUK~L9gxTow7lTyR>cE<?t7yVejb8PM2!y5Ct
zSYAMC#sFU<U>x^Kmfcg|yHM;V^W10B0{?AqgN)*4E~zT%sv^}Uq*Bo?=ga7DGN;`q
zPp3&~l#IBnk13Y8CTDl5R+9a%97j$CtjASetC43Y5mx)f2Gc<R9>v$C>N7JEBA|Fo
zKvshKx)QQ_ICHwHVNU1M*cp#lsCCH_KZ?su&ZYz|Dv1av3?v(}yH0U`?1de-b*BL=
zXM|K=P1=mOTN_o+9Y*ma;$Lz)#<gub$00UWdsZy&H_vW<vr(L`F|Aw_UjMlcIXg};
z6U!miUJ||=PHaTliIj%cm4M;x$BwRR-8zQybs8R}wbNa)vEZB(;S}k+@E<01AK2Z4
z$>34L*3*-2i<S2`3|MmEC(l$Z)!|GxcxP4_fSD7N#bU-Y{?wsH*n9(^z}7Kv=ozj_
zoNH%tpS`5}>lkxBYsHn@A_r|A*dK&=xg2-lLNgks_K*bEzl(88Hhi=1JFahMu_~os
zr80MEe@jA$@T9a)wB-uv6oh7gN<dk150Xs_fe2zOT>%U$O|}5B6OsJ^lAirP{lsa9
zQC!C%Q-J7<D^&>Vl;?~GA>={jz?7ct^m&Uusl^6<6brpvv%xO|gu)<In)}t<8uRXL
z|1gARyod4811RSDrhyQr^nO*}x-u86$IkZo9h_ThR<eS_Ih-YLq@OxWllvF5o?6z5
z``@|;Jir=TWVqvOl<a;GPf&_Ot2<FVAl6kMpnz;@-*yeBu|#T=xEW8Gh4aI)5=6!;
zs88`o^rxvl@sog>9aoqhxQ+80Eq84%2}H0k!@}6+D@&z&&0|+C=(pHVuzgboIU>#z
z+Ql0l0_JXR345=j0mZ>j<!g7|wkQay=y7;!h3HR>kKK`{Z#NLF1t&79W|MRXY4iu?
zSpI9}N_|h#i$F1W_zbkh^tFsU$n9*$RZ*tbVJOU-3;pg(cZccdYwjmFC&Tc^9Nb~=
z$XmdMIQN2}6FdQi^06mNLjzQ6dQN>N6aW_*67xqR9_7dcU9`Vj7XCo^WjN);61a5=
zS}%@6`;h@W`FwS(HmfDe^_9OaY1m*+r|1!k0A$%J{-whk{yM&^pDBRove}#DP?bTq
z=Hb|vHQmKnX(`16Pc|Ipwz2JZ<u5^Mf`|ZRv+5`@Ej>GncRIY%&wH+FnI<}6y6>U;
z6dBHWxSzg@b7!Vwz#N#VKxTOCnRfW!a<-o>fI_x+5X^$UyM>JAk1g^Ve8)g|Ti|b}
z$4(@<-s9oX|Ke^jNK)Ky>VGZqZ;ty+#_ey>yzNHt_TBZ&8Z%#7JqO1H*&j}{5B7cy
z^zoEm=|Ux;o@EI*XIQW?ju^JEU;0Ds;HT-bTKs@nBIfTduJMYFW~@6e4DI#aQwrs>
z(n&pY&MExPS<&gwaj?@jVUz`1Qut8_&ZiIUOs>0Y#NlK?)j^RCm{}K85e#Y(&^*1;
zVuVvYcG+p(DLwbI4@P_L$X6cX)$T$s#DbFt7*E_De*xltyl)`X@AByRsBHnkY20EN
zTx&P<Q@Temc=Dew)+~&VF^7o3b@GkbPf2zR0+bs4L^R%Xs{ymc{{Rh5E+UxT*&v6d
zHA%Ez4Fl`|@zr-Djx{OgZD<>xF1M29>cx>eiO`3#m`}+<T}h^YFPW?HFnL*+f)r_3
zW@=5X)W<ttapCaf!4jh*pwH>-jLU0<%qu^dTDWA|^AEj0;s0mPV1W7r33y7<TZzx+
z{HJK<w~>${vI7ykZ_AMYP#|&^(x`o6ntUW?tmtSSm^N2yj>eiCPqSmkHyJ1j<VA;x
zfhB+sqMxnaUcZ+ODl;~=F{99hX-olaaUtfov)!j&BO=cZ^7#jUXx2H)T|$B#A=nTB
zjX{Tba>Izr7FTidlC?^4Fb1u1qT1^<RcP*(ch?<UZ=qJqYvvtb+P@r^A%P&y2U<WD
zE82GsEM7&Fc}{%%RwvA!0*hO;ZzpBgyS!QWm;dAn=iKPj!q_A_{`n`1p0U54z6fGQ
z)BC1izvq>tDUkK}@K!pS-A%HfJ#XcrpYS9zLvZhs)`NuUX%e7c=X3MGA>1e<KfG)K
z3FV#jw<9Sjy%l&U9-Up`KxhCKOyBwI)eLC+`A$1h917I}|3+O=3VGVxh#lTnck7x9
zlL({7qsAN!mW1Bm6k=H}_fpiL7cP1MwIILs&ePC$EFX4N0aKJw+H*P?1bzjTEeeYb
zvi<VcV&u>WHNbweeve_XJr$fWuN?o7qJT$c&1$m7_+3mRmSImDhjr4{n3DLHk@zE6
z57IOdEht%OYMjCvb-N_pEECu(sO)_pl2_fy)RjaEj$1@{^P7D!mWbb%;<Eo!oKKSP
zvK*&E*mV!jlamqGh3^R`%`CMJs~K#49WOYq{V?F~33PVy-S|#sFh&Aj3<Sk?Z|!8x
z`!ZZo|F`|{+%dxA&oH#>4*E>aIu1g$xP?icgsbis9?~sr{rQY-P|D7YC{>SVD@Y8|
zMHEmNw;V9T8K0bqKE&hsNi`%>xlz12+chQ9PER&ADP25p>o`rmWk|~n3h~4$g+9k1
zRP4pkkwbi^D8_-a<h<ysl7dNG?)(~*!hSdw_KhlHvuZwxBlD$rVg|ZvM|WXr+HV4y
zl5PWSYKnL*5OL&mwkGq$R;QMPC+|pbxwe3IDJVxWi|1|}s5AHb)78b9E|a-n5zwCG
zg?jOdJp9}qaQoSR`zsHPL2n|+r;$Nlot@)zN^N_BEOFQ|uDCMy$t!2TEYFa09m{|t
zgTHhE=E&odQc}(3%Ods{ud}F#iIc}T#vOo3fGFPboop2@3ju<)aZZWCQ)EeS4R)s5
z)!rWqbMC3)eRC!OW>;?)<%~ss@Y5}%geQL^DL~1@CP!Bypg<aYJ8K!x%Kh}ndu@j1
zY`5|&)9Vt1!Eu)BWH~PL4;7h_LBRH*#qkmYSdG99b%|0!0EFbL{Y)Piw){)4$eWqz
zGG9J#P`im2HCJ5rXs+7c(HuqLewHf!P-6tc2hwar1dU*?XtR%zZXp`_-c%q=6z*;S
zLiJT4<_+(vJ(u(8Qnk&<x7gGWD*}SglP-5ryyMBJHqcbR@e1=QPe<lTjs7DiF=m>~
zq~vxK8%2*38`9{I(jb0f&ANhICDvQ{E4z&rBMiULRIuhUW$~sgCE%0?66#RpD~sLW
z(rj$};+!Wlm*jZSHF(62rT|bqL$nOfFF#FGS8b<+F8V@@*lhJnaRJkbKcM564Vh+@
z!%`}^$}&g}UWwy?Ghsl(=UCpemJJ%(-Iq5tU1}~c#UKv4vsh7QF2vpTjm{9d$49>{
z1qZdG6cTGk-14`za9Q}yE}W#d3wb&%1Dcmtv=sLiP!_w5t_Z+J^?9=Vzozm4`2#eS
zh8HsZYj;B$c#XREYbXaEVeOjydZN;b`(R(XP=kt&AJZijY1{JPD|e>(3j>15bal=U
z2@yxt?00FW9BUFr0ia#DX=?K;dgWnD!;Y~*M_1@^h)u6#%3Bw2#aoy9O!=n&Y_6vS
z8vh2=CeWe>eK+{i)Sc%`!?xYGYCM3!M+9@!ZOQ(Fv(q5hbAt|TBD8$vg!Jot){E0L
zV=W$}Uq#z@f8-K!o$*e~h5WaY$s{cvoAYrGczecwmRXA{2b?Drj9r0W)L?8PYI->)
zL8<OuzkgsBpKIi4i=$WpY45x#Eg~mW93bESuM=qQ#~aL!hyF&&e|j!}T+r_8iwvSN
zJ?%=8=QHD+H8Ur<xWRqIugF>ZEYhQIC#HKTnIeSYE)*k@-XYV9@J?r9d^wJ5-p`)$
zyg;Me_8s7VJ6AZK7MB(XL}>DA*I!D*$MS-x56aoLqXWwYA{aLkB$$scJQVsA+f*x=
zU#CPi3f_UDkk=+j*7`SErTkrq876?dB@PDny9<99@d4_K`ySAaxTp3<Z=|kvu8qPv
zMfpc_zfIla#^yVc+h`WB@m=A>QBq>jE&#&S)v2NdFf%KX9UN3;#3nF5Y#^>hLT!LM
zqksiq9*Cld15{|J=f-v^Uk#rh#IrnsCE2hZ@#`WEl=IKzo0m6jW4FBI$`q>>exN_2
zy6WUD=SuC2_*LxatLc}A5F|jV{@lf`3?IggPGKL^_PhDZY<<yu4x*;Kx3`}jFNgzI
zr?0ZyxBft?^O6jJvc-@XGK_Kc`#^PZAJAQynW<^_Xz;5-Z%iOmL73<0h@XZ7Gn~0A
z>FKiRSvDV;3Ql$-{$@XAm-gs5QgAQs-abH<$x3ArXBT&Ce(6rG;rjPtt^C_JDsBfd
zo3QDbIfS<W;A7o7y`3JjP;=k=3Be{E@}K?!q`s-&t{_h#1kH)^GMEhp@`>*N4H~Ls
zw0dV{A+To?45IpjWe;AR)Q(5|X?Bw(SQ&&j4!9cEz$km&Bhqdp5d$muiCbuD5^fo&
z|6%u6634Pt7g8?SXaMErLJ;|t@eEbnxd##{E0(K<OUUYP1{q}`QLvCIzY8l*yHyFf
z@1@eLCDh-{7H6AxErou8X7km#%~hYZmOJgwy7eD>*ihch^2{M#7CPAyKPW%PLtG!s
zb-VHAp66l4g-|mbW3ryQ^9$HlY=gTFG#3i9<|YN9i$3vt@<$+BBg_MP5Ib>M)#;T=
zc5sQ<Z-C_!qL}u^lzJAN<*9z`>hckB9>fL;aS9y(j!>hNOtG>vZ{KFSf>BFJ;mQLU
z1X!#6w>MkGk#TEW$*Pvz^9O^=N{X=@(2DX*GOT(pJm5S!)#D>NHQ;*s$lJ(3n79V@
z!PPFAWe3S6@m@T=0b}a>IUrn(4!ftK)cSf-efDJeYpEV>#*ouX4Dq^1T?6(*a86*>
z%=^w>L96}o4o5XN8H(Q^aoVr$xXQ#c|I1iXc-D9Qr%H|^LQ5l-2fE9oBFCXX-_0fI
z`RqR1SnBw)6%hYNagNQquKjb)i*-YaU?48>$wUL484*>fP*-Klg%;d|oiX$04UHl=
z^74J_5J98U-rdKSDdzZbv6bI^i(z^w3H)u)b10go@IFz^<rP~g37Q?5c~cgNBK>RV
zVYmvAfc_LmP<<1E2qsdZa}oAHwpc-gk#g87yp9Odc#~-#d)sIRsI;rD6HC3T#>%LW
z-WuVC5>nRFrjyQ2ZT!{ifAv|0*VnI>`&LfC==3e#d>Sr|ewIg~5rXeOfFTIf0Wn}t
z@+h>d44bO%cBt}fu=AfXAAGOjPh#5uOGjWF+3UD(w#1Tno&JZ%>oEoleLo+;P}0PI
z%g4a!5gW07^d0-Gzg;x|Yn(!v+krY8LSew8Cp0^}+t%OHk9`k?GWUbvQefC)gYRiY
z<}Z-^FMjz7?Oik%vryGn=E|4&ZMQstQ?rxw)k^}r{Q4zx&Wv3bM_=W1kK4??9rqru
zl~?1fegj>`W`&~sd|X$B#;T50;D$`uS(eK@COttUoo|NTV=~}^9Pk2L`CD&EcZ&t~
zE{{7A5hzG)Teb(bJ8A|t#YNv`s#c_Ky~<NZQ#E)=XD2Zxd+51)_<NMGah0gdQpn|y
z`Z?X&l}Lko;O*#0!M_M=-TNNk5_!`ri&q)NeQC#lQ9jJ{8PQ*vqArG;Q@Mt!Y-a1(
zkQpjLt-Qn18sO6w1LX4-ViWnbMEH^RVI&7`nw<A65|`Q=WI!ap>5>oa?7S_-iPGgn
zRV~fwF2#x}n+NE?kVrhjdVMxtm2<zV+$n~1SCk#+@j-1LV)~Cf3Ta}m{lQvkhxA{9
z^N0F#z4nBXQjOL)0+A8W<zQeVR8-Mz>a0{RGqUoi&1ASN({vqK1WZ=dX1mY0eHs2~
zqbEY}hS!zuC-b9vzbe^j9-=ZWer_j;vBWPXf%>kkpM^-7PA{$gz0>e@Y}#c0zD8oK
zCq#w+@W*2sV#0e?B?3zopitaCQ+J^ShWKc3djkp<%(P2`XWgv^pKAQwa=;i(M1AzC
zf!RM<KfBI#E_%7Ro3EHhZUT3oKfhyF-%B|LVveygwhe=w=zR*8;W6-23|Tn^U2ST*
z$+#IR{bonu=j4c8atW(#CaJYKqkJ{Bk<A=eRS)7oIV_r0xoTVwP&8k~G!EDVWulKu
z<F?1$q3YATL}i01+BJGl{S6x3)VzT=sNAq-Kfv5C6)#C2w6x9edfA=KR@5zXz#$$@
z%WYudteY@*VMXxk;!lrrv7)7-7?<MJ``ij!p;s2CvJjJZqxPaz1jQtp`)qFP035++
z&O7ztJKLiN<Ul<#y>S#K_g;Mje2ohz>vo>yXy0Ozv)vUzX#CPI0#CZZ6LhD6hw*>1
zczm`!QrdnEb`uKH1B!7{77)-dV6rufWouUKsz21a;>%$?uBS#zLj-AutVuBujWboR
zrU{YfHFrrL_dW|<wpR3@VVbLm_2cgj_yG0#S-Uq&_2c{i%rv+pf;ixJ`Hn__xmmP{
zQa}~_Y=9FT{9L1R|7M(>Jd9J0CNAQ0on3wdSlV`w3${<>5Pi_6>C4xgZp$3bw?E&S
z=KS9{w>PX;hlX7n_0=0vw8AsSbf0LCbYX5icNZn$3sPI5(sH!gf?E43Wi)U;1_zaX
z5&A$b5x||i|2Y}NJ0t*E+@9TQwwlSVfaU_dtfR^{D9Wg%I8cuZ%PIzH#(`>fTKB4p
z9=+8US^~d|PwRXq^EvJt|0LlMIo!-RJPL^Pfmldol|AJ;TL6bhxeapJPNa}KOnJTj
zwpIo>VAd#qkqI1!f?Tr&FQ6AI`mgfol3!W5&E!YL$m|#=KUAzd$~COfbn3Nl-{d|U
zQRl?({H3GIx3zhvl{L6aVXBx=)ZNc*>~qy`N&d(abu%TVPSK^(X_SwOd;a;p0d|*q
zu!*Bs7|&?W{u=jl9F_&?2UKV-SydGvg*-rO88Ae`N19&aj$wK!3z+U+*b4_YHJ9`f
zLFZoL5oay@%b{_=GF8L=wX|aN)GlfEb!jl$T=lZ`tG#)Bv+LK3=T_-8WuN^p+^-(d
z?DoXS+p&&&-a|rsgw4vB+nmVw#jRs3`iC^+S4GN3fRg=&u6=dTLloAzO!2<1EDMFT
zvh#9XBI}=d+`;K?BFbHkKXEo>G8~<1+!P@nZr}Yo#n51B7NnKACX!XIhKM%<OwZ)+
zW7~3+c43PI<cB!@>il&{t&n{$F)hwf^9SgU$DaA>e{Mimds)@sU6MqVFG_?<PN%J#
z#m94HDNcj?mEfH>G<XLFUvB<9SBX{!<qG8H=o4c<*Vk|>M;Oy7xq%&O-#043m>3^7
zvgwuA=Bb7P_9@^8;r0~7qU*Li{HGrSP}<7V)T4UbExb#CRq!C1J4RR0ab8NJ8m^&)
zIv`Oz*A-ww6|MCa_(zyC5Ay65EVfkI?_=Xet^W=xqCdw!dRsAU{Ow4@hMmR**eABG
zx*4x7dP_6D`Z4lKbngoY9py~8FHcsB$g|vA%CH~CKBN02%_0`|^mTKsd$aK)k>Gsg
zf@s#uWL_<rg5K@shRI+4QAkaTkAH}Cs^I^h2NnbW&R$dslG<-Nlrz}Z|5h=mF0~IR
zt@P!DjwaWx@RQRPMY}vRtoaSAM^@jw<!TqMEhL7!+E>9YP6|zkOsM?m_Lw0|fhX^8
z7*Sq$zn?0e$As($REVE_x0?56pzLA;&{eiqzrLFhUY!SSma6xPzZS(4+%UO58AB!f
zVKFXTllMMFQ1#Mn98E0qp>Pc!O`}$)#EuIYq!;N`9;JOaeC7bfhKmFX&Mz4xB?R+7
zWDx0G5rdnFZYam@LEmFsM4)dBH$1ljMkwnONdMS-H?gP4yLlKoDJJ`>|MLPU1glSf
zkG4g3G&N2YW+2=5O=4k7`M!PXT__bX=j}nDmKQcN4ODfAn^uV~F2dHY9W_#`t&&gI
zJ6>VJu7DJOdS+!@4+Q>SDapYW{GoTL*y)w*+aC{j)jHO>s?;j*E`z1D@gkZevO?it
z0se%&I3Q5leP@yHF(5{_7K~#YADG)r3~^I;LGj+2?j}0B=61Rk`D*zWD2JEhJ$G6S
z7v3GM@}~B^A~_dq#bu=s+UUK>ZhGw~tZ}oTS_dm2z)a|zdl3YjVV1@$a!SwkU0xSb
zawFRLY?A8o$fH?<;NO<{e(puIVzR-PFp>m<&`V&k+#mHdsojq3!Ax6CwJ!hM64IFm
zthv^Fe>iqpn|Knq_P5#m6_61UX_4wQmD7?^L*Seih4-nDE;rlUEpEN;OJ1evigCSh
z=S&-}U?P0rK~VBm34rIxQ%V;NwyJC8iQDE0<@Eq*e4aOlhZOMir_?><o<2<U!CV>R
z#K?dn4j=?pB0q*{$cv4FjyzlZQJwW9*JOc$mj%6w5qO>O5o-Qt*T#5wH&ur@a4|sH
zNKn-+Nam`5q%L@|H+@2k@Jd0N!uP6!O80X41e?WYdH+#<{6h$xgaaUha*gR0cX?Ub
z08n|LG+m5PH0wf8mNako*w?<Bt36hJz*nQt95%B+_f)-^ax(<D{2lhQk{vD&z3gu&
zMA}I%H>7j8-J!sj46I_if{O5LCru+uqfP{|>Eq`GUZXU<q)4u;BeB*gRN?`5>jm0J
z4=U2Onszo68;eLbMeLerHlvn~>yN8$evrX^S^3**szPb6duxe+MED;n8%?BsW_)a8
zOm<~@_GUszB^{i~A>Bj{%!CkG0rEy)k;t136==0f33c6OpJ3PI%qpq#o4e26l|@A$
z8GL6e-{(G^EG~s1CHMLypI`!gw0e7_wZYe^NMY!o_7|Qj1oh`%jFaroQ*@uVucHV9
zcnVywZ=RlFaUb0kiKUPIYXgRwwAC-5YLa@uu1M_m$BRCA-Ul=b@6)-<PQ9kz|3;!o
zJEb(xsOk6>OGF(&<$kJEguIE(2;fK}t2;guy8fL9&W-s=KkH6FTKJIPrb8hmx4?I-
zdogZ$byA;IwN=9M?V=QGFE2{*qgB;XtWej_;rzEr7GeLt$s+WZKQ=Rj)3Cf2|B?pq
z<Gj1tL#AR0%>FgT?}-s8eIJQIb}6mQ=Ejo@?iU!|ojCAMkBdgff$pScfTXxx35>q}
zg!+>xGxvbz4a|b~ey^pRd0~-%0pE((`=W400|Q}PZ{)F`C<nM#GYw96e`sKdw%fLA
z6GTm>X3y*@;g63SL}39kiI!_DaG6_;bje3j47vRJ7RLnVW2+X~bV`aUHa~X~vNn2N
zhE!dD7{g*Dcd8clohrQq0HeAZM>^Sa!QN}HjSk<4wTUERR(SklPUD3ClbbyTU)au5
zS9OfCD$@L2f`P;@@z3muUf(r`I<utZ8K{jY5I;W-<j(m{YcSEdb8tphdK$(Ec-3H`
z+%`TIF3b3;x&Ms(L(zMxoFLA400D?Km7#pA-lrIwB(kE1JmG<(OX~Hw%8&?jvp2y6
z#Xqxb932K=PL-wY_o~|);4gE_f6TU(zwhNMS-`I=9!@l+Sb?hBztQX#*<?*3+WQui
z`2l@@N-m(=z)hZ0;cc%7^Rq}GKD`8r?X=z6PG^MDc3XEv3HEO{j~VrM<+X#us}ib8
zqL@5KP0W|?Xy5w(*~Wf5A;QxzXMQp-C_VVcO4_6rpKTY~u9-F+iK;K^e~KkvSXFuu
zOU?Jc4j3}zkU89RO+aO>r0bf<lAlE=7|suJZp{aQ!GG>t>pZ2gJ6-bYm{Xu~fx1j+
zf`-`UGDhp&nh<Y+!?uVg21#jbwo@tORVIdMaO*`VOt9Pl?dfmyY&l@%i#LB2=(|_j
z(D_rR|2w5^t<VEdQ%E5qywf;SQBgWT_gSX7{=E8chb4#kepYWTaD1Bz-qd9X;k1T5
z+P!sKbO9`GldZPK=dI^Ri!Q{|Zq3=)2GUs?3yC^}Mz)4w?$EjQj>A0S{47XXYo+jV
zVxhX$#T16t%^qFVWqbXiLxlDxAJN7~e$~DrD!9%f0|l15hx}6FpoAj5dhDOm)rjyh
zU%hsQRH4e1;yrgsO<eE>Wo0^WguH>@&?DM~-Uj$vK%AUpX%HKpRJ5?oW@kYJ>8@>(
zBKJo7_D;jTs^Xmv@n30}!$OKI(YK`3H7DIX<pofNSLGRi=RD&M#a9r%7H>EiqW$4f
z#x?A0iWK2kdJyN(pY7Zj^+UaB!-+QJk{mJLvhqSW!TVTn8kA_nRrz)QCIz}OwP_FW
z<#*u!xBZ$4<B#vda6pHtJjLbwe6|%#_-=PVa{)F{ZRT8f^~Jr;brk}S-|kz$Tee{T
zsYt>2`7%Xu$_WSsA-r~=m~or>@QLa>4s(Z<X(8*|{ZByU^N@*os9gXqJeFb%9-P1I
zJCtK*66HtFy7M%F;vtqSF@8I<eWy~Jd@V8+^HILGz)(<jHrn+M5++^zAo;4p5Kh^G
z2Ll7Xi*T_5hU9AbU|pn>5I?Wkd~h)+r+x7ysM9^13Lb@v%#mCtA(xz;%^b~OY!Efl
zI1yzX{<O`5L*YQS#b0$;r_x@HU#R=tGhc*%_UAYE>pxdQ=G>N7YA3gay)^Uhe4-lj
zI!VFFO2F^Qt{vg|m$(G@oEsY6T+|PathUj0>PW9lW~dCmJR%rzqLrnZ0iNtQP(~YT
zh!@qJ++lIXIQ2!eG0mg?7H~F;f`p2gtI_wnXvpi1<&lq!cCr}UCN!Ia|FBdST?X@W
zf<|1!?V`&S2>=YBkd0QiG~OnrYe&9Y3}$zU8!;~s6x$|pdU%Fb@`z7!|2==d-o~|m
z7A|P{t^%WBS*c*+mzJlwsqYplWcafu0Og6}=W*)p7yl&B24>8bKz8}w)({OAhc5F)
zLHBEvO7zIW!8<DtLG8W<dG5|ImX+;A*C5%bQhC&yIbF3BlvJrCEf%XSjLUKA=E|6O
zI~|5P4aGi{GPAmpB9Xu>BHe9p-Ak{DO=+G2D0=8rrLh8~H-J<$-zjNcIkA}${xcp9
z_L6kdXA84&FsV_f1d^?2C30hpCGC+m_n!z)A>SGAp#MZ<dedj18pc%TED%Znv?X(1
zzJd`(<8RsNKip^v`qrhKiny@!xH91ba-YmvIWvZX9_<K$JHHb|8y=D)zWw1JZVJ-1
z8Q?_}ED%6Me^XIkS4seCc`=+B5MDvDU}R3Fvu6-mKK-I*#)$ze1R%!BmA)A_7{fr|
z8|8XhW6eD^IJMiB4Q(^@BNKZH_pgjAPlP&Wk=xCBVWLU1dpz5ARd)%Q>IGgu4Fn*C
z0vVKVntiWSXqr!i1hPbQk;y)6HFCB%EhOajGvji}sBjG?5N-Jv>)>}+{yM4l;2%)+
zpl=v9upheKLs>xi=<N?Rk=6ECxJB_MO>od}<Cu0Ry?tW^*J1ZOhuxgp-9Z||cYoB-
z{bPXXsTBd<d@Z4Zl&z!p;3iPFauR6~n)=KXnluUUUKOc}=NYo!v&tdwrJsWfa;BKw
zL7~R%_}Ei>(q>SeV7^ZFsMwPiHtKdP_B8t%`ys^SP+D)a`3dL!<b<Ocu72SCZ85OZ
z@`tlOQ)~krU#uCls(~hx+jEjOq5ZJ7cYBv36Nd7GY&cz=>8mMuk?B9#!m>1SyH?cT
z<l@ECwD+LyXsL)I0qfFJeOqHIpY+i)H!ZTN+mY)mo>6lk94uv9Gi>C%==nDff%$_w
zy3ZcLPP`sMFhJNr_BH_qzisCyXX#M>z^uo=Xo079<jH}9XL}vctIjY~6n}s5^@EVv
zrt!pT9d$H#wp&F^?6fSA^g2u!>Dc!s>77j^BT;D}*n1&gc`1>xdRadK?%Id-wmDnJ
zbzJfn)j+5Jud*tH{+~N<Mi5Px=YafWTXK|={2-Rk4Lj!$rLEUMz+YF!RjsgL=|5eJ
zg$J$wDZUvCdFG&#cDxN;S>c7In_ux5B7t`vnB*m<?SM$pI)>G<ZuVngc)7PusaGNJ
zD<ZybZT?$OxF8H?z}~-ITQ5}KbN~2C>DbCnA>qBBn+qWixefnefK-fLUXs6(j=)QW
z5;iPUO}C?F>+tK^Gm!N!IbT&uF7OJ9LWXJ<uI;#!-2J*7wS&6-gBv+eZ?8ukg}gT3
zUoG||`VJ&V|Ej~z`srhiQDg-1(QKSsinr0Q7FZC40?gUQV&%H9&JyoS0m%6LnCnXV
zZuZpkS9eP)Bqa0OP&#_F_G?mOpn>p#6pS-+fgARH^nC>bdE6W`6r`jHINl}%exM4|
z2lF)j6Vd+%U}IW<f)&k{{LRzJk7mcy(5{Ky3KI7>k^ze(((DjTmA$dcZNl;$C4$<7
z?LI&2qX)>%eoZb0HVFeIw?4JZ%ea{fWu-sGrxiIjUh>Nmc_$^NLB~a@pS<m4OqRdk
z%~L`23~FO|09NeRuE3Sq<_mM+Kp&)hV#3O-h+nx|Ymehh%x$_tw{&Fz8BF<Qdd$&W
zpj=l&bMgTiG4p4DSwC;TRxKSAl}5292RYH1g>(u%*aIx)e(-UpkFVQ^tM<aznPTOn
z_W#N%vD<k*Xdr<YX!pp4#!9L2Hu*}S73lM=KY~d-@e7K&><IehLrI1<7y&oY?(1f-
zMTmRH*rhViJ^-u97`zUE{p2_%ar$t&Vno;l0cGbzgzWA5x?AQiRkUXIF2-u0h9Xi-
zsTQIKf^>Yk&Xkb$=1lksurS`Q#V##te10hdO%GnqX!n^bZU_2Bn>^u-%m=v~^dyA=
z*+3yLS9elm$TkSBe1#pWX0P#B58j#s!-jUfwm_zsY7_=5#Qk6gKj&<|c6Bk;j;(Vn
z0Uv9fnD&E42~=Ob_BLot>(5`B>&m*#zXMyq2ie~9BT-9)0SRdJE#Z}H?Rrdd`_mgQ
z&Srlygst8l$Dv&v^|gkPFez3rAT{MA_V}~IH)U`J26Jq%Okw0;MHC9!C(j3XFU}6z
zt3AT@wxe#GWON@EXpDAgr|`fUH&t<+!>$zzj3mbHiy4&QW=84*STP?)k+)Ogd)WH$
zlFb=3lMK#2J)yXG1UlPW`=!~x7hVRI(>2Z-n2xcFS0WWU1aK#o_wL+fNlqOZLb0n3
z_xD(qv1vf{hic53Wcmk~wiSt%KEX`5t(Z0+{bMEG_oJefwE(G8CwteyLN;mOeBTp;
zwp;tUhuG7*Ka`sX63wk&^w^PB*_<WzA-Me001x;H6Y7xs7mtuZ8r%ZCS~mRDM3*a1
z(XUnkc{^cB6jxLtKD27jK&<Pix)_~4wg@;aLr~D>zJHPf9MSDSxWNQ2DD|`cpWW_@
zK@SIec9`A4nHBxFt$Am~P$C3-n1Ur4IJ_bzLFj4)^}jXPL7WAOt|~rN#FW<#_|bC>
zFZ+p1rpZER0Xj8yXavM1sA5A#Frl%^NE^`!q8`b&Hr_^KH@!<bV()mr<xc38F-Y|d
zrf=&i;6-nIzj+9OPJcFY+4**q{?lVnuaUd*T<U$kvTUC6((Du3b02HFh`%BiXPqm+
zWf8u(KUeREKE;R<Sa`WvL)92_W5>(-O6(^dn!ww!h)OK3$i*&kyQyy1l*8JnlEoM$
zhv8wt{rOsHa0657x1CdugLLN!K6B%3O)l603zMXxMebvOzk*-R7XBGZ&sCjxAn*qZ
zLTfBDaJeV*gcTeF!+hO`ay2()lE3-6Fle;H)}(ajEmB^>KzF;St@fjOYe+LsDH*)L
z6i`?BU%**OzuCnKO_RCqr(e4WZ~_Up0p}e!QL2~jGEEa>`B!|-w|KTf*vUzNl)F$M
zHJE@REtFP);6eam5~lFG4}OZhucq(FJC|h8s<naF6ie5(vx=Ndbh}E%=~(1qI|x)r
z^?(BEmjK_dZ8*IKR=CAN4un@O(iD@wUAa!j_f|1hL<SBYHG$Hs4x3&&#w3I#XG>i!
z8WIrn?9Z2U23;&4|M+k`bJo)mxaX~!HNVcHzKAogqx8^KNj_}&y`}%Fqu+JU0WpmT
zq+3g=gpma&Ge71Sw#xXFr?qo*lA_=MmQU?J&X(Ot5a!!cx%2-YSKl4hWY)ZmqGCZ{
zMWllYtOC-cgLDM}14{3`hhCMgC@4}xHT2$lZvh3VQUjqGA|f?}PU!HR=<fTzzwh~D
z+3VWuQ|6pA<(_+Huq$$nJ%LBuiDAjVGV1)^hWhq2t+KYi4`7CLZy0iyfm#!5siMxJ
zChW!+obs&utBr+Ic>uP5uDNzCcx;~z0{1?oI%;5-J6o>@Jit@G#fQi4Xj2ATy84Q?
z`M);4#dBzQ04mJLw`eY@Kb-5w0w|hx8VlUAda!JO=YjjTZpfD}W|+9rH^a80^)9x2
z1nh=LI%Mp1@S6XGrQ66?mS$ZM(2FDPs+l73KRswJbrE#y3nCC75KI1jx$Vy6)*7g+
ze`o&Ge1OVri->5R%j)Z9i(tzBzD`knnbtXgr_Z&k<BG}ZNX{nQ)i#p>YB*#&N=RBI
zzqx5I#z(<XWUnpsmpp$GGTh<bG_j=rUa#WN1)KGx6wo)p`+~{t6q;}xLu-R{Mn*U<
zc2&iY{>5<RubVPjNywmGOX4z3=6J!mP3*!PV1!3AFxi2O@XzY#kbC!XN&HL42+b~V
zH>Y97lbRm^o#S|QYuWtax2XwGIZRd)y;}|FIxml=`!;dfctAPoCkm`17m_WChwCXn
zgjh>H{|5-*lV-B0c`MS*LH>?zmk!j<_Pj5PThqMhoQ}}1kQ&i+60oBJNv0)9szzMp
z!+#QVG^I#@A>NcW)oUJn94}=zCVh1Oq88j%R_Tq(_cOo~1XZ=HYsR9&<N>X=cM!Gt
zygvkQ-BSMi^M-C;iZAV=O<hwXrBGhi;{M17+n2b~D(Ai4k-$Mifjigvx3hCf2|b5)
z&#JsIvkqe+r(IGCwU+ubJ{`KAw<fRLiF)!SSXS3E?pE*pA=z<U>K-1yg~c^qyvk+x
zT%O&sCddBi?J!mgS$qUP?xj;-JX;EeStXlAxzb$OsOM<+(JQ8uby2ZHC6TM~l8f6X
zAn~9Bf1~<$9OTrp)eZE5G;@9j-N|?KT&{{NUK2`}q#)CYZ>AsXVq0fj{`T_$+obrk
zfw_IK<0mQkYn=xP^Wscq5^K@qN*;a2s1<w{Y9+vECKop!?(N{#u;%TWf;c(u^gsK_
zal>-=bT5%qA^8Tb#i=5^g7)y|PZtD^ERuGEPWKE=Xw*jZ!y=wx{j8<&>sV_V$*8`T
zl5(UoTVC-mZW)x^YZ=s2N$Z~KZzZM5A0L1p2b$coen(A2^umND*eYCUd;}MVX7U=2
zJ(e{7<X2ceNtz<bA=ejCQ-11)%LqIfAEYgyNy<nw<~P*tXUa%xuI{fh*mlW_th}Uq
z!sVr~lH%I<$nWZ%>(%pJbc}30rjf(b7@fP;AqwV|3=%?gofD)Ipci!G8hMc~?jv+>
zd^v{PKOKW{Z_K66*~5M((KBJe{@_7Js84^3UJk9G-Fn%d?q{Ta*k6f;8{cS*8xw%E
z)HUgmK=s^*V}0l44;Ci{AdH1Ipf}IPMP-;F&vyyqbzI>H>_Y=QD&cPA_^9^jX|z#=
zfv<^2Go$_rX{{dZimx$qXfORSs0(eeH?M(f#pa4;@mH@&8@BC>Qo2>!L?w9>8j2Co
z+*kA@v}{dcVbONF@-COH)5FbT@L|k(o^*`MXlrF+J5$!^TSS^vzuM~a^=yE05D_h1
z0VH1{cXL#J>h!3iv4|&;)x+a{&Yr8Hk{rm4s?d|80gi!@K4YVNjGtBQ-wQcUVFNTV
z8ElkpzFn~vB-{apdJ>QHbQ9KhLNXT<EI6gOWX!8#R9@<?HIH79*gfFaOk5!U5+_)_
zyr7_3YNb`zRb<z<4Z5a~FO@v@2^&a>#ZDu<QI`#UyrZ3bJZwuMgsLy1os2cB9K9{V
zCHI!-PL{rw=<R>>+G!7P(DH-sAe!?XTrLWv$&cK;s~EGT?y)$@26tXP5$?2FLSU^{
z)T=&P(2R66-}F19jJXNL9{6D!c0;Q3Dhs;iF3B~_w`!YuTdecdc^sStHjlnMS(&1(
zOlgA>TFw!*_1qTy%nZhDE>h{yJ;->D<uLIZu8H>WX*hEWd#e4Rw;Z4#e+J9!yeEv^
zYvhAkIQPvbJ_x}xLpzd<Vmn7Z-&<<rmtQGkX4+`+L~M#akuhaX>L0q7OSJ?mI4E>*
zVdw8*qQF-RfliQ!6)On$GhAzr!sm=BEnLYKi6gcCn}(Jbv9m)`>$>SpX;bODU35fP
zees=6)K&D$h5Y<)9_3@z^%i=eJS-<eH1C7cj07P_h3ta+{4Wv)X);DuZb{Ugmb3N<
z<Rl2kEpLAeWmBXlLoKVW<B&<y9Ov~63?0i(4`*dcq@I^iuj}6SvHvwr#gu>DSCug(
z?&d-_qN;0ZCt$pPD3T|wIMy7(I_IYDepZQw*>Xq<$z#bz3!$bZ<FWp`o!lp<xyUTZ
zyr*$v3WN@=+{wv#Jr@h6v}W%6ghBR<rIdAhnw@@cO%=wT&3JiK?dkPOJ-LkNOS<Ym
z&8eu(^V^XJ8Ns_h50fuT#XPI4{wlwnX$98dhs6_sGzU69<5Y?<?WQj5gM1igc7$h%
zr|R4*vt7y4hKSD!^{X4&$tI^ZpFA_%K8izF&N&3T_b;=w&vfyIMVY39wPl+$AN{#v
zkZWhGxlX4`_--slUqcU_>0~HiEB&P#HDAYO+w&H2Q8#D63YE7mqP%$G-9R?x8C-#i
zMI{T!ph7#_tok0&>GECpmED2uTMXZGi0p%)V!voOa6w~PTlQV7qjo1fLUvN9g)Y9W
zXqqyb$(L$w68_=q=TYAG^;{M|3)BLq5^5>h;`aCa3WKt;gKmp@nS*FyF>BMzcTbHn
zM@bGvZ$In{Klk#rw{!y}p|<dI;H3Wrs+&6}c7Ea9YU~wqtRkN6)lB|Ym(GPxLz(O4
zwV+XgT*EW{(GvWma=MC}VpBBDBP!XTeSva!?TjJ|p$qU(OgCpB5s}D;9M+D5__~Db
zcXO43d{R%={h=Rb&hVmevAYfw*22eO-`tmarUys!&1H&w259q+t7k~wleP~{Fa%(K
zAPN_Z&}Tj5<V(wfAv?1OJ-@|wkQV=AMENCMWmlNM!!PBZBv84pzUm@yHAmRHVwmTy
zkCuEvuXmcl(nQ^*rj~=JL8<y)Po?%37t$%Pas1bI_Z{3cV`rUCeK%4hTKw0q&Iht}
z!(kNkE|iMKRJr$7@bAn$=8OM2oMFpj=k5%M?xHLrBkfHFs$m&B=v&Nu=2JvBeQZo4
zC?Y=Uy()WIkuI7EkjGgq*}AZeqAXV`eoIP45p^4?-lQ)T;1}Y5dKeLi>B`FaOxpsQ
zhxb0a${5@a`LOzBoHXRD`FhFkAvS@jpm{n`I*D<s$s5=zU$08Uc)|eVy#oHHij+_c
z;N}dEl<}Yxq#C6$_0DE#gX+mCt4%Fsjb7o!We(hX=o$4%hVl;g0B4`(LuaROC!gVc
zL|uhm3g6#IucN0e>LFOYCef+E)lv@E%lY<K?WNyWpAaqaBtUpbCr(k@NxsU%iX$Fw
zp*x4;x0X-ot!y{nGMwxb+9Le8>$EYxj#<g^(w4>y`Cy(x_<$?vhkzt`E3us(+S73>
zh^>RhK+~K}mjB6uN{Q6a_5qlkk)m3Pl#&$7?9y5rPf;z7>BGgb#*L96x^8AaVfsDM
zjOd3(JOeXj`N_ssi}%X|BZ+c%uTkbA#MpdON-E7O(8)~HJxXw7kQLb^P)W8sXaNp$
z;AeTl*<>;hVKc@CxWCp4ID163$hzlCveV53RhLiXM5}6im~Z`xD2%5(y<MK$V2CBZ
zRGUyLKGf=vOHVGG-&2{5%G{iMA-ev_yy|@I_}aSU_nP^qK?6sv^M@Tyl0#h~D_IXi
zPQh5dewA8I?%*@auR|J%sre1&{Bmne4xe>Se!^+tv1u#x;y$twrPbB-jv_i`)6B+2
zKBu)Z)@MH%!Dt3@_dRtiF<r8c>(EHIm>7RS5(^3H_{Q1M_h|3Dh{rEKOWS&Ul-zBK
zJM5qKaqnN(UY?!zY4XDaIP|@65c~dP`qbID1hE&{zq0XCbB01B;f<s~5TOw;5m76v
zkA3&%yf|mpp$J$SuK15{vWC0iR10n^PlJkW<D*KtpqG3l3B7o4>nveFOIYc^xAlps
z<X-|g1~y4*;nV0sdpm6-yFt@zFHfzDTt_Xn)mjOI_euRi%<-Cn*cvVc5Bxp@gi~od
z+(|?y?<ip{UZOJ>e80dQo+seA(<LMoKlWK1><^U~L(G{w1P?qIxGEN49QN&*hQHmL
z2qDkfYo+WZW13Z=UzpB53VYf)X<GZtK_ZJ$NU!9a0v53}Ww$1M?mFRBH9g{j-jQxN
z<?Ib3>42j(wix$CN~|V&mE2^Ir)OkOijV!*rc|SSB~Xcl#Tna+zBxH1x-cPT+o2?V
z&YczQzhb>^{mTWQP8oW*OJa8-gNptXfq3-J-=#A#ED+~$1Qj$}c<;DAhkrRX%jdVA
z(&h28`EUv&wdME15<W6$jL87vtZ_D|W!S8+^YU@m^Di7ojbcFx>=p@oT}nB7cRd6s
z!gniNeFwC`v4lJ%xB}m^c7>ewzB2as2%i#*g;Bivp#h^yW;{1-`yo0Y{?n(MV&!z8
zS<#R0bb8{kbUEV$TO>jY8fp^}wZjC#SRJ1txa$4xMtA)uLx60WTf$xExIY&H7H3?}
zX?>W}SU3M?vPDhf_r{ZonnLkgp?ZzzfVJulkrP}5zhDk8gn1QDqH7(lM1J(y(Fx27
z<L|Uu$qBcFH5wzPv4X_sYakAe=$&pmf)^=fusFJRR-p^QIi`rSYa2^AHCu%}i|l1M
z`T{0kCD2WhbV>&#NpQ$<w)&Jxis^u<vvxO`>9h0~B45}i{ykg#>!pN8moC8>)fgLP
z+M=l&A00HF&QNd^mCLoY1HMzdL)mjhmCDjY(J$0H?v0PmEz+b2y1yyTddLWr+1^Fp
zaa<Xa>du?uT-z0b(fZkCdAR?)MbD54!v1V52YJmMogCe(REgDgS{s#DZd!3EQij$v
z_8rZ<-+4Tr$7X}681mOfQ53Pxvvwg53HzH!`Yr5>?~E;c1XU@YLX~svh(5VA5SK4d
z`F99e`+_`;MgP*nx<;+>sK(2L_T4mggvn2VrndH;CQ%;CwMzY*129t>({&_jcG#yW
zC}DL*cR2f!wL+V#s#XymmA|E<^KEi=IH0u}5{VJg9xRpkb*;j4?8oYmaN2ZsE9O<a
z5JT~HI1fVta?TAK-T)?S&J^A<#Q!*g3qF*~qsq4H>0kgxMG2+^)%&%1NoyEY<GBRK
zlwG9>@KMN_nR!vL9$*Grd(^~Q=qB-kD*RL)MKJiJ6Xp(LRU+88)A3za_m-W~q|}8a
zD&-aEr(<t07k$iKfO=V02{9EWuyRemC`fCm1sf)Rs5I0H;}?KGBtyUQtUx$>ejwAE
z(qXgRp*Vcl!AlA(fgG^?z~FDS6MKWF7s97JtXy&jq_|WQRCtEYS{l?<%S3NFV@fQ;
zr}R+yAu}^s$g65Kr-5e@0{7UAN^om|mKRDQM#Vj{XO^c2l*7kcv{jH^#uM4_TiQdL
zPiQ|3zicEd3sLe2jPgsseDi^`+~}KPL!Rg1N>PH}u4?`zT<K~15hvzF_|fY>ivi*u
zO+^(Lo(YQfaeq=^W)Dui3!HWN1U+VF>jjnd5AC#IV+a0qi=n37^}qZdLTuY#i>2_h
z3;#0rGqHUbm+j@HX^WiomZ;3Eo)&fFaT=9Sb%>h2ann#lQjB?uN!aU|RU1B@C49;?
z`-Pfy{n}TbTT9h%G#to=t?skm|L*-gs1Lui@X9=vf|}AbAf&4(>$(F+@6%PGN8hRA
z;)8n(=yY9RbZH5hmKdF4iMW0cEM<c5fgp&1viQDvyb{d+Y;7j}Xsph2iMniELp@2q
zs!`xxB%LhiX)a*>ESIO7qlcQ;FTks224h&l+*++^wN!UoQLU(tlEO@obU8zHP|aWO
zO{$*Xa}ej*L3Ok2<rj}$9#xF@uOv~cj9!e{uC_UD@CG+9LK~FE`4njmPjDyQ@xvDn
z5Bq!}NRuczvr+%G-FEa)nlk>$^C&?+_sccHm9QYGfNEMbEj5cY<y?;|lPK{&vm?-;
zqc#|UQL=9ENYQv7>&-Z5cDj6`Hdf==y}Y#)*FTS#8)0;5^wx?_HC}f*QI3d=8yE2V
zjM8V;FVd$D8xD6<u=O=kYMb*d7YEq2VuDolU6Ui=8zVKV<nx<97Yv@hs_CT@(|=NO
zXCiiAr~m#{($pt*zK)fD&*#~64d<p4{EEnui84ZbOiw<3)QJR@VExNaynhh~P>l0t
zKbdpenJFFUoXYjAi(fAu{i=SA)uT4s-#6a~uao4-<S>%O99vjfU|6M)gH;tCymIi`
z$+%v^2&4837MrBR!l`1UxLE-=y^<C2RCSbVyK5<O9f#dSkn>=Q*xwZ(aP>5D7j5N=
z40po#XwhqR!8(+4X71QGd`gfoWKmLIO@Z>hB%7qxo{`|#xVO)7=DWEmu}9vv=AUN1
zYu!Pv+UrRai6Hlw;wh-JuKm5Z{HEvCLkMuQl0hVnK-v9XlOZofVk!#{`BuP>0wXuH
zT$|18qYazW;1oqPNwb_cY;3L@u@lCl1QO<M5%Jm^0>7s9=$pOY{Bq>{mb=iVMC+;Z
zkjrifqj$Bk&r056#Axl7e946R9Q{#W=!rK}|J}?9N)$oP!AUn=-BhWI?op}5u1vCk
zy_^NC$H}kIf<w?||4l@aHY{g?M!zi0+RI)@E^@{-IMvYWslrNfp`GLSLi=?DTeb1D
zTIm_c?Mbs>m$UeHviknUCVyx+{7NrAy0W#@M6C*t0#UK;yguJOzs<RahL38b`)ak9
zI8_5~y0WidC&7BOGfR!Cr%Ca1WFS|pw5990c)qYT2hJ{}vd!(RNA|f_SMZZc{oR+5
zv=BZ#RwcjGg4R6jtfKeh8r1Kj^V_5UH2Z%4TD(n?K@^>U5P$&ob@bbg?38?qPhGLJ
zaJOR9!_umcyEAKcCx<15Q@-^^iz)GWX$E{b)9H?~24STs{A%L^w3BnqAb{xpmN8~T
z80A9L=DcvZsxXE%POC2e^JGSKN4H$wZalAzY%<rXomxwD3Il^h|A#M6j)G0^bMdLv
zbV))>pgDgK3bNnpQcU>PfCfPouSXRF<E2wE%6U{v(?B3jm`O?s(2l*`1313V^BT;>
zi8VR#7wRfJJ+_<q&N0JNr9@SMk3H0vbX-zC*|T}L)i9@O<KLnQJj${Zsn2yS_CD_%
z1Dg4tPERbLnwkpFpLasaXDh;PMGT_8Y+fTC@R-~=FM(KTtJv09LDYFJf3EX7(sPoV
zG@L9}JzJJ**_nUmI>B;;>Fdp{4FtnNTnZK4I+l(19t<T-N7y~n0(a_*$LUD#hJIoh
z9a$URJ8KYK^3SRaHsa+M>dd`3S9yNM`pkWs0n)9ET41?;*cHrT=j^#u`t>V3cAdP1
zE;!>WrEB#|*$4y2vek;=mft&@i0E-bNytZN9h2?QvCHINQYt|?(Sj0uB7sxX8DO5H
zdX}4uUn5($-XCvi=@#ZP!SXtNuD1v7OI56?u2SKBmKTFFf~w?b5#=iRgM>ANt5pV5
zy=m@GGyB8Q)w?;Y7f5(8Q=X$<M?vV*+fW(3H%u*-5epGLJ6!r6or$FAT4(zq(X}T3
z+|#`k+ru(d6_nIkY8e$qLID{%lDXyUQ)%NHp~f|cJ<VzQwYm}FYaK=^71?JUv4Qq{
z)$DDeyF`J_KE4S)9PhARVmF*N!x`6y8|!Tun=(d^<TYpu)RNdQK3SGCHxcOp>EQnD
z&ZX5WqB6!802p%L&(uY6$gk7t-u2NP?|IaLIh8rT1%Gy0fF1nrg}iQFQV5UIIKVM7
zk4n>MDa>%86u+nA5Vd1+S*}YkyPsbHBW9K(y6KJB*hM9^@Z3c<PqLylSDH-@>b>)d
zl2`~1<L`#~Dp&axa<T7t2FBa__BaIjRE%{DGLy~wMvTQuq`}D<Tf^Jn^GRvtt$FP&
za^?D38y*8abT|nW2cuW5p+EGeId)D|lXJl?ZLz*m$<2^Z0tP=Xe^-13%l52O*sRD%
zC=2~qv5a(6*j-d7>tli<pQvx`em`u7oljs+$Sz>{0vydqCG8jTnPlUlL3qFIFIO79
zJ)`Y|GXT;7BH^8a5N(2&Z?n6%zBWeCo>iHOFpcDxZ2HBB5!|(|L!<3f9eUmCSyer^
zq|?`p9x5)T#TSLlOlHlC7sY<EV6NUvN=*ibWj)}C;S)?g`yLtI{r665Bl;ATP;5xm
z56kWk$)dEVsfFjV$=~a?qY18r;W(T&0=QotyUjnf6ghS%110w0e?qemjGGGGSJ9rs
z`5OK=o!8qh)6aZUiqHlcm^)2(>00GlneAA3fL@8<5p=#{G@6@%x&BzQo&x*ac`{7a
zIBKl$HkV8T`xVo5FqsP=3HcaA)wpo{CyQ~_GI-&~yjvF-lGe(xQc|K%(tO;*S$igD
zH{StY%UJW{nxAD;s2j|87L#gUQO$3`Hc3S-N6$jho6Psgdll6H&enYta6J5%PVKax
z2L6d9)CWet2pC7z6yMp;JV4kuAGW6t+gZekNu5<+(sj$aX#uHw>wwk&Gbf@^?jiU{
z>5*4?^Syd>jdsQm>Gy&?Oj+QVUNW`v<8NhE{-lUc-?T+nS1@^XWtu^HRign+LO(!L
zU}<d3aMMh}ar135lW~H!5F>7ruf<4x577((&fhP3`nct#P5-t6ZZ1B@ry)AO@w2vq
z*`MqOi0O}=)+D}0Acd5<MT62p?sWVZlsw`n>MMLMioSVnM;^{*um3NyONMy~R^n+d
zL~#g!(}znYenW_=d%u9vglIIfcr!=f>*{He5Prjt0mRC+vawIxK&{%pzF?Fx!syw8
zAA7R0kyzP<_eCSgxg|iPQ-Hr1Kpj4)Rop)$E?&f{xV}~xI}Yu&;+LIiMdOC+Qm3;6
zoE~h4(+0cm!CWLHg?{fHfEVoLxOLeEKtA61X|_|P3K^#a)kOFjdCu;YojSNY=onn2
zd`H^R&Yez||3TV4;p#ur)*IyrLEeu<VA#QwSpJyR!|lZdaK^%9k~J}CUawg0WLO!<
zIZSDzGc&^;bBTV|(rxFph`i4D2xlj26iQHIsX89g*OV$f_C(pd-JtJyy7kbjBS68z
z`5sCvA-Hgk=!0KkzfE~n*MvJ}m*3;`S2Lxz@1Hj(3WBb@7ic-Hsq&ltfTzLW@-OOg
znGRL^lU6RUD%|Ua&wy!e0ub8H4}J?^vlAu%j%|n-+;)rap3I{C2-}QD&`8zq^jUZx
zgm%^T2`hh3DbV&cCI+6Zk^`2=YrwL|V(^()Qc8-~%(+TH2ti6~(XtokaPAHYxNIkz
z3!-ps*jU*%Vfr7qh5{JkZx<owBldkABefw8r`y^0<ehelzY)?Uu}c?pkvC4j7{751
zfO{2O8NT*|a2D^yq>hJz8;CslmB+S}!^1U7Ec|U0U|CG~JE+gGaEd%p^J7Tu`f4$$
zm{owsN8puA)TJcQm07p)M*gEQiFm)UO`Z5Dl`t}W%B3AXSb5-3tY$K8n{9RLfTt}D
zsE6$RC3l$|@$<x$*9}vD<0ORHXhcM}vML!G%@!`_I9vL55b*w{mSyLlS_a$86K6x>
zKEE~(rP};|xmkmFVDwxPGvO<NGj=-rep|Q_o;$jRmv=`L=Hdf!-|yGtDvdP59-M59
zAl(nbWKS?P{8^qeVce$Ejau5H1B4+z$@ZCUqJ%(m=qt#L>>{<j_G@{3Hi*Ib12r$#
zVv}I$0sihCb?cezuL$@jFPQuE`?}rBfKLp>X(cFJap5zKSku*XGl>$s2dIASygLU^
zzP=F0xGaBo9bUZqN5Pr?W$~nypVT}o+ZHCTq!c}Rr@&~$QIQ^kv}dU9IY$)e&(AHa
zJLeixlUF$@TsvnI<H_J~RSaLW2%+gN_<g8B_C?`?jEq*@!<bdV6D`g4+dKt=ici=|
zlv$uHF`%^s5fKHM3`)j~CiqUl;}Kfe0>cr%zMAX%_Ooe$erEFTt9Pww<X>m=C>>U3
zyPz4SOldlgV5h9huaE+0Vm>J#hBM%Y1*N6toV|LLq=hSrwhfkC1F8gWkSJCj^9Rl%
z8ggR5x0Pk2aH~FEtP-mHyQ}L`O2B<3$a~GO{;00$gU>c{MbIc~LW+dcQ10w(L-j_m
zX;{2exp#0jk}ZdPCHwHTDGeVC!Lm#OW4Py>AfmX)bhItk&b#&y;`T&J1q^>9O$UMO
zZk_mmnx%OXA%aQ}ii4iK7E@FMzNJ;<z>sQb^26?`-)R`|)sTpEc@*WfO7G@eGkf`=
zmzG&l-C_(=IynQ^PC%5RF+D<S@~?+@Ipj8$h?jJZ8io*@*j{o9HqBP{Rv9_)d-V0D
zV&C=leR8A#bLh-B;D(y?K_Ek1*E(0?FPe`HP2GvK<!JvaZvf`~XA^*<(qVK-7lIs3
zNIiW@@T?=Ad_C-ZM@fv=r`HWpj?#pUaLJ&&e{XDQo!ZZ-ZR}qpCnyQeWL5Tt_>osI
zQFAWKwDTh94#4Jbdmg*+Ufl|o^6x2fhOJC>udF(I`5c2}%ua8vYmLXt-5=9F1VpDC
z(<qVrrr0(@1cUR=gHl)%4lk+01>L{60^;4Z68*5OXpRCMwdQPAMSfO4Rkc@%N5@mV
zVb`)G6#sdD@aLrOFY0<;8&v>GJCYsuZai(_H<_lh=x3A%jUz%@*YjT!5?4gGAG|3J
z8sxH#$yQ*K5a@PWrflbRgt_~@H#AbjL<x)w^mjc&J^-mDOSl!7L!0yrdHgAR6L!k<
z6CtFiN*5eG6TnXxD)i@~aQI*dN~y0_3e{W`sudC;mP=~~ljbFcG2MFOAP@w<qEN9P
z47nZJ4D-yd%GlXYi-aKEl>qdk%<5-5WHn9#R_*DbWg^=6A!K`YZOsX8_0g0j?S-}T
zZX<V-8{3J_$oSoqd;dO=DDbsuWDPsYsv9TBrZ~V9RQ&q$7~Cqrxn^u$%dJUj2_Eyv
z7s!$<Agbn{bO_-jz1|GVre<~Eg;ZIV-=nlNj*6u=JARicTio?2cUh2h&QsA^k#21F
zIN0>~n?QfoUjvl?d0-onVs+O(T+7GKqQu6Ye4@1Wf^LaJF$i?w6gjl`R?-l(S5rpZ
zpF<nrP=3#ugBRi!XA56Xv%dqBh&>E(!z@96MA*lT8!3>)c5=I;^rhn7j=$%B9zk^J
zV>{Da6VAP%ooUh4$D_Vf(4;tba=r;I*e#Y!r%OQL;klyks|e~NnEOrf`;G=!sBD;&
z6MW%RGI@O+MixSIR%}Wm&phzowK1U(Sv~zK(*(7U&=}`%O7OL=0YSl}U^o0BK|&u+
zZx?kz3kxKC(?MdLNF)v@HrCQB{HKqNRjlB-`Xk6Yp5f5~c|42%oK5f}-i;e`ARk@A
zI2%-aR1`#AQ*65#qFoYR%qJuV{GR*$<!8VWEKUtg@jd1L{jx%Ym%VtWGZ{7$@gtch
zgtnFUjT>Qq|G8+S$(p3V!N`rS5zvSl1C3O#S6TtL&75BHny8h?&-1ldxeK~dk{V^g
zwM=o-bBy`6gxku2*l_2ADGduNRmeomh(JG{V(y+@Ty`zR+(#CX|Jf|(V!^D_+Q#pu
zk<-SM9Nkl;WRcqR<YOeOgOKpZHP$0tZ!<9C$*_pi<Opr3*|=55={_EV^;qYeCUGU;
z2yIcaQB-~&-iezY(aEC%{bY*ta;tEvo>%`Ua_|#U1_kM;F|gk!M{~Ip=q%v5thNCj
z&>_Df__PNOq!&Ulr630F^_vszX$P%?P+*ua?L_xsgTnZ}qg1Wyne;P)DWZX0uFNdU
z4WY%*y>oNA#ux8EBQJCFw`~Le6<q0QqN^2%=>c;<2wJ(7vRCBgF&V_X1ReB!P&Te%
z>){LrkamrN6iQI1n_94Tl%U`pEk2?4I<$$cg~Sv^^V6IzC|P$ixKm)p-?fW!fl;CD
zVe22hG<bP!C(fJV|NBKWcW=lER3d6{`S2+;ESp-28@co>q?a!Fp6|f-*tbh6m1zL-
z259(Lt@|Y%tlmj@@!8Bv)}?)TJl=GV0o*9i2DB&j6a4$z%MB-k7VTSBzV`p$r~EvY
z(Xv*sj;hK1O6B~`X=a9{ikdCEy#%C_M=0lf>##uG2Ke1Pj<c~eKMMTzTSCCi$}Dqj
z57T};`(453An~_Ye*YtwLP44ZYJCmrmL#6(M%Hy33U9w*;o*7dl@|c3YPUMm%>hVr
z$=(S3y6rh(N#khEiu(onZ4CaN82Z@tKRdjGb;{hYPTw#k9Pm{y=zgq<X_QC<9(;p&
z2xj~pEm_mLXM2S_DnTgND82H{Z<U&yY33t&;W9g0hmKq90DAOe*J`{e&FqvR8+%?I
zXgU7xHN+2C8rwQiH3cOsKN3IY{o=eckb7MH7%k!zTY8MXI#yeMKB57nxu`5z?M~|-
zSl%OV1#@g32Y?+4U3ByDQLyYb{~d?46@bx-hrLW@->A%rdBnbik+))={bvCK)t_gw
zb+6#zt_>}A?uV_Hn!WLtPYxSDo!jeCswBCnOAYBp0WayutMcT2|15bWN?crHaw6AZ
zKKxE38zFuRRHp(5Ow-_f#Rl~Cj!OM)Xg>HRTlm2&o)HS|$Kpi)^DrX!!KrQzk6K^>
z4q8>7K`x)J@NiC+q78eMy5?E;`i?It1-E0!M~9{|09br+mHE@Be7PDG#ow7%=atIq
z_zS%DujImLUx<0tr}aIL{Kqc?LNj^4Ob2?$I~P^-W2|G~6yF<S;~46t&3%g@EF+Xh
z)^CVeTa;O_uF6p!kG(S&sI8A`!77uuJN{l}FhtmI{wf~|+d0bnMet&X_p^W30qBos
zW(GgB{Y$tb=~4haq2ou{t*V&SbJHMdt(?pp%i$Lpc`%5a6WpCzaK+#O^wz7ayy2iN
zH2g<6lmy}AV?Y}y(QKA)_54bU|JAxfk&Qw?77g>O8~e5DD(TT429B<0&m4(>Vr*|P
zLI(JN?CyNdVVavql`<pKtpc$3+6qFzau#xT=qSYwFs?nyVEP~PLxB|R)*Ggoh#^P-
z)LJpn95TdEm7ZDccRRl|8hyy!NkD;mRUwyY%Ixfz-~UmNR|&5OdC!+*N<;fX#LF{r
zbLxIF{eL7NEv3AAryG9bZud$JDo&~QStSW0|M%>B!Tw~{VLckIzQtRO7jz?qg7s(!
zuOiX_jDs=b75-G_cWxo#U9<o9UTzcS#I-?mx|GCw$|JwNR0raa3JO_r*3LQaY%N+W
z^j@>F3kyn`G{e9g&dUrAHb}2jO!oStR>kWB5sr}_KFx)Wxfkh(cm*z;c@iZjj_}Tz
zvpGa=d^ZOQ>f;7MW&Lwq3d{lvJ{;Uuu8}W~n#>yUSgt<b(wJi~{*8NNCv(R?Wls9(
zKkn;@5WP9Sd(Y^AV;YE20#B&I-}Vyi^n)ZYNcC|r7J4Yb^b8y&+~-!Kx=N>wi^?9I
zMEBB7xsOkNJ)F6xLf>6*o`~WtSQk6%#LX)(qwtOIb8=y?9{))}{(d2vTR+Vm9+LJv
zipP+Gp-z`i3#WKVPw@OzV=JyCf{kR_0M!Iq*rVjss^fr7w^wD&-VBi8dLbXQU&$?E
zk-Hw}11Vn&PvokeZOR$H*Mt%L<93e{j$Y>r3`c{JWLWV*r})0PIM$QR6Te2&wfo`y
zDs)4l#6%{R^s84)CJyg^+y8by*+=Vr1uoZCOfKxd%7(^_v^|Pt0Mk1qm`+<5y;~o{
z*5REj*ukNgmi&UIJS7SKqY1*o8Hjkfgp&}{nWp1i5X`x!#kk?*136g<PdSXV7CMwI
zR$IJC5?OM&?miC@5pfVf8d=%T6Nm>Qe`SMEAWDFvqf{_DnczVDx!T&4sxPQT>b)tx
z;vzI@u8J-z<blJO%Tp3=tp9o5{{C~n?#;Y$hEYI$YN@JH1Hi5dc?&=&Tn0;|ia`pn
z+YeaPx9WR!#E&}!`aKmnWmoB(XgU)wtN6(zZsz=z{6D9I$ZZ0cACPUu;6@#%bX4WF
z2kB;;B+5jTZ9lvol;TMz%rOm|sS1AHwfOguqi{__%@Xzc5~{U$tdcPfu?bj$^n*KI
zVKt49ysdDbUw?W)M0AG&ctS>3`28I&ERuR)$P}IAomKw*Kp=*1bQU#Aitn$B`d0!r
z_l<srPo;pEw^M|*x`5+(x<>*!DpwblwDU(HiJfl%x=a}Wv3{;*mv`m<-LbN%zqoQ{
zD5?n<Y`$}mWshzn1&%yYiHP0;*dzL-JWRM_KPA|7=g_*3H>?82eYFC69Vr0y_W3=m
z{?AJU-Q#-(%=_l3odsh;Qf2_UN_sHB1p$D&OSm)%j&`r<`As|YY2)i;ra;pHtl-$h
z0%(&ix_XHJN*NTP{2_8|p8x9Z&xBicv-|eIk|!;OZ{noX)-^n;pto~LiKL^{v1-DP
zfeULHY}))_d5gQLQ}Hk?`<~CMUS`FsyVI-8N4ygM6-dy1Usi-9_=ej=ijQS9Sg-|D
z7Py|`7t2cTs|rf8s^o0G&e(A=q)BvFo#!<WsA;m~QiCXmAwgPc2QPe!kWeL6TqbJ*
zY^MYXq4F4eH5sKg&sRa}qU@W*yig*d^TXZVS09qj-D>CkMiCIdH|ySdng3L|op;Nd
zFK?R-eZy>AOOa>KlwjWfxh}mhGgvap8NF~AU;NYD+`TjL83x}gF%~33Pt*>{s$ww_
z$LU}_TvRH#;jpr6dB31*6W`K7hMOw)V(sIEvASK%<B)>FT+f!4EZG{a>ap|E0Y8N=
z5H=(g2olssGz~rH6}@{v6b3;Sudgu#n|_Pk31emH01-^??$n)sUj#4?Wa{NnqmSuy
zo8ZQ<l07f4OFF{Lw$FpA+NtG<1J?xHuT<g}%w8wQzWh2t`l>mlB5doDwZ@{*cTJ(1
z<u}Hyd6%q1R{4YsdD&D63?(I37>SP^k@Lu%ZZT=n_IkyUB_=X@daGmmc3%Gpq<LcQ
zE2px1w-}A%svoP#Cv5VwmI#y1IU)rxOWya|c~@a;u$@*IOg4FMJ=Ye&G~e()-Z;_h
zn__kWK$7L<?(>?4Di!UGEq!4>2ohm&?-GEVSWPlHJTfTb;by;LGkWoo-uP{(lk5iY
z{a>+fazK~t&A6P>qsLa5=_B@yj1#(wv_Yu9i=ZnxoOu=c(ey8SqW6FM%h8C`DeHaj
zI!bk0RaL#j)@=u0xC27a_?^t`(la$`Wp6ZZm;@|dnL~YX5DfqY<*R!EJL0V9Nghf<
zy6(?4v_8U+(XLT&MIhXhz{e=KQekKBrVP_sW9t@+ZK*CW*yBR8De5)U>76xNzXpj<
zvaMq0!cbO_FTJ?@;eu}aOTWJ1^okX$1hUa;8w1DQ+utq|sTdhB1?KZYwk;Lw6t{v!
zEt0bbl@C=FF$wM|UqW>D6dT(F`g3gt|409{C8{Ofa%~pwGfLwyvkzW~+t=6C1|wZM
zCe!Z?mPxHPp;5DA=m2zjZf%6)+PR&yQW@0f%q4)=`PTJeQ(k;R_N0T9Z6x_v$ak;f
zOkyJG9iV+i6LSyhvsIsl{;Dg?fei>iww)st{%4&kb0aTPTHq|ZS2RowQ~jMZtDYel
zIwo%?H{MFn^1MQ~UgS6&aN>i#-G4c+QC>L;K-5Sq+xj&mr5_l0yz9gAm>%l_0WR3>
zmCKcP3!fo<<wjT)$rV@nbHsdD{RRa3uRTNl+LHZO+(agc^eOX~8#+9$h8N$N&+a&K
z^nnz7ou?%{bG!~jX7mgXPZt#wMu(I*IWMIG2w>F5QMEX{81N>3UD@3BE40OKENLo7
zjUCb{X1=B+Iy@(XI%fu7xPcT1KRrf%Y<P5Qfk8QQ#{dcdhsx}`ign6|N!eVm6}ASp
z8s);Mo!8k%tcJI3Gi~9jkLNc3>wc0-w(|<K+^>%l6nrg@dfs@+mp<h}PH<*|Q2aXv
zVTkR~aOb(y>lC17LV3<Z2HqMyHYcbFK$o~z*A>DY7!OC=IR%#$%AgS25~pK0^5gkP
zw-Mmw4Fj^<NvA-+9)OSpA=~`9wkrG({&s<W5u^a+Gvqg6h+DKR+!*{Ru!YZjP6g)i
zpDS<)E_7F<8!j6qm|c3#IKDI@0>o<`FYg|SP2o0IKt9ZUtt=ioKSPJvXZ+aNYgLdu
zr3Icw%rhd7`m)njJ6st&^8n%Fda&(tRDCLKIJ6HKI<Xd!n-0isI$C7*F10i$`2F{B
zz*+a_Yq>9B^FWYKs3Ur5DtqGTa}G0%pJ~j%!I&%8{^azOGp|UpXXfdy%cnKtM*X8g
z{qvn=qJG&xllV<f_g<+q`|+d-1YW(^>AR^&=h4~Cm31ij#c%=*tfM(La6JCn^~1HV
zk&#O~J_##_QpGUjJniZ-5sfT?M)IHQ@M~D~=kQ{~B!%GuEOumxwVUokyk}ji;W({Q
z5nxu0wdxmGPIDf*mp;E94wwT_Gyl-PsdEoIYe%|6Wrk=>L6#n7h^Gng%|E{BU}0F>
zA{z(6A#lKRI_%H0UM3e#;cn-Rh5cx~93+oI9DSCm%dh>KGnP|xh~AP2l1MRfX?%1z
zNL{y76ma%bKSF>M?;aTQJB$49J4&~913#J0O()(uI8Bh1ZMQUf7jRn|7oF&L5~y;9
zQ&Hub(?4N#eJf+XnHj~lxZQ@vOGj&>d5>F9zS%2|{5l^H<)LaBW8mnte(&1j^gCZ}
z7xi1+=i!>EGn{d@6RMP~IZR*o1RoO6uuEzvnf08#i~HeIMW7c8$L3M-EZ1FqRE{ME
z=kbLN=(8z_ef-AN%9&R81VYM!$KN2AGoVS9`~kWof_L!jxkHiBq(*6rW<T*GWs`lg
z{{JdKBI#Cue7@}*qTK)<<ragoaq83Oy<mThJK;vw<JaR@TPh0-*hG88d}LKx2Ruab
zp&l2*JzhR*scbMXHc%TWa??%FJKOpG;t=&D#&;Sek0OgKK*vGgYBb*;lkOw{LL*Eg
zDjvjYDX(#cR1j}3?Kohd*S8OXlZXYm2z(f)Be>|3(8%k}OJBSzWaikxluHMfWf;kf
z`V;deu(@_xwN=NzSA3F~;Tb*Javom8FMTrS!oYsv;f+qq+yqk+A|)>VW6@I8va0kH
zeHAm<^S0T-Zx5y@=A||hI7m;UV)KWu`ey-mlFmc|-mK$SUn~cS#ANjIlIG|IsSTvk
z>$c**bs;FQE@ckvk%!-dvl>xDWyZ!+n?M7-vwh9go~KKv%Vw*uYhE@;z5^Vaa~Zve
zOR6PO<Lm3Cs6{5jEbLjik^^Jv`$Z9k=IV+ECP$qOS#KgLgevj5d`A{LtBrDnHuedu
zIFPFL2++wPRFrt#krx^3$H|yWpsn*?f;7{L+;P(uK0?WFwr3a7$_dDJBxm0lHuhY#
zpHxYn?TeMLmoSzeoNxb&Ky3K;l5TB)x>OMIoFJUp4N_L6UtK1(E%<;+*VqdW0QdKP
zrS5FwaWKH~X<_5?rQN(k-u{4=>y@?DO~j}AH-XQ$I;aw%ZB!Vbi+7m6pv&hBSf=M>
z+9Uzr>!2Y%n^7T&!2ghC@Y~XLxI#0)6jwT|t0>b2GQz6ad_MOJ?Ci#gmA$a@kl};%
z1Eu!?K1_x$3VBXF{OV>G;R`?Z=D}*e_@o}-V+AT}aPA>z>r)^N=#Dz$Mh3LhKWX(f
zw{9Cmx`5)B6^xXcVc&*B&c#Fe|NDuG?>`QM+|!8Eno&AZVXt{-qu@afb$Ibnp=?M%
zFa1<?QRZGr*1aTW(+oiJYPou-H?5NcT74<P?QJaq9fY_bqwnak0G)0LcscJ8k~uU}
znGi$%w!<J}64h3{21uJAdF?EAy&>q+eLZxYdjevoF0$I*=gH$m@#o0Qeg!(&H6J&9
zK=<%1X*Si<g6h4Y*9F*6sEy=_B~&*E#hedl(ql7)R(|DqQ+4Yfm6dkB8*ay@*zbiC
zsAZd)B<J-;2sEja<Y)wLD_Dp?hbP{?5NP`TH}d=Ma>_}YJTE-zY&{v9)nA=KYtq(;
zUf1@x6CW5aW;|X<(Wf7xu7$Z(&RE#s>Y4%I*qN6jBxk(t0%F9Ndum5C7L1>kFN-2T
z)=2WZmn%;Q;otNOPHzqKcUruC<XA6MiKy7k7h=SdsGijxnmk_gv2-(`$2v($DAUG!
zj!^m7?F<YIon3loXMxt`+X3zJbzc-J2W-EV+}%*>KkP|rTte)f#~TY<XLl~<EG>%<
zj4Txo1$eP8`y@OYY0-v^EoYrbsXI9q4Sj~A{FB6alYg^j|NT6Z%OIYi_pv$Vta;Ft
zoWfAH{KOW413ZFEsRq+rsCv%WOfRKzZzGfiD$^~mbq2+ZE}-BPeM-b9cl5%Gx0Nde
z=O%o*-?-=!s9JuJzRs?7m8EFr{XGwr#DxtF<*1qlw8R4=%Chm%6J;-wu$Go8HSRjl
z%dXbrCHOR@lfk~}op5X66;Op%lUuE|*E%nS0ii~tuY5-8+0Wy^@x_X`Be~QnF^woe
z0MGrdJ9>8|*A~C;WUUHVWR2@3h6voP1O-Rs_Ep`lpXicOSQQBj!yIGdVkyic1!|Q6
zb=T%IK@l&t($XY;H5Wdilf8NM^GtJ3{eHA^5{Z4q`(FN69^mhi6RNYd)A3EB8mI}I
zun$?FvOdt;L#>r*x;2U4)^IB<fbNpy=0-dn13X9sPC~OEu6@J%V51c1rca8VY#m=O
zFjkou#NYU7Z--qX4LJc^z!Y6yvqh4-!vL#?Pt>(;1J>QC%iejw_5gg1aq_bo-GGzw
zss^`Z{8jWZD3t<KGXxQ)r-RNSB{3mWKw*j3Idoo@803SC`m(a`oH899qOxSd^K#is
z0Mc<VSJ^(}r2Qnu@faC$?lY2F=mN7B08$jy2qa=D^X$_d|AOBv&?)uFY?lW1b6w+a
zZQ9qM1n5PGP;^6}G&lczV8|<~s#-i%b|P2;6n3e<=dQqf?j_@wLowRI2}OL*=GH(t
z7UAX2nFA@N0E)??>Ojmm6wp<wfR+MzSV*n;>y7WD`+-LSPiSf{s%ZU5vi>7p-U~Vj
z7uRi<SJVK+N1#)we%o~Ldikv6yE@~^6*3k8tfW7e(aQvp?_uqpmQdQpD<ISiz`SKg
zn~vvMQ6JPM;PQtMg_HrL>1B<VW{v<<srWlr;i)(Rc!`&)D-o?tdfKO+?^b|D5Dei*
z3e<Q8q&WEuB!-oD>G23ec{}zqUG=Z?Rqfu)ndYpiDV9kBr3@wQl+|aibDBVVdryy^
zirYo3=toXrq-T*~OvbNn10<Y6O6sUc|NMe;CZui;Q3(mbieb{UaYe(bpkflp3zl8y
zrLpppWM<qC80=P1tr3#2K6Zv^Gz2l%&dV8qkC09vfMq~<&%A6oo`2_k0{sYz0NB1s
zuz{+w{pOqfba5H=Gy$xGQ?=ae_RnvLj9tem{}=xur2Tc;xzX>kK}E=G^Vf`ubTKK}
zpIG=~<7pP5Hom}|10n?`Acpbh3RieBZG;HmD&V;mo9ha7#UlU%&~&|`qOAEdNNhyK
z)8Ol;sUyZPD27;DPa!7BEt!H-j3)A~?=9&)AaMZoeC@g<d};4%{dWsOc77))z)|aC
z-<(e|@Xi`vu59=k^Fh?bIpNiy-l6A8MWb2gWyV6Nre;hln>z@+%lA}qH*PrExfR_m
z!Q@VtfGkajoxUKTAx<O?F$M!a{TyUeE3)+dSQe+)v^pqBB~V!bdSUpmMjDiC63CeA
z6TkL#sIu8?R5h!LX_45)Fw%Ny6Try=Y^TMh^=tG1O-p!o)#eg8mBlVRMP{@nr^egy
z#FVY8YbAr;+E&SQ@go-ha(64&6#((>fS&y~b+#SIRJSx2G*6EcTuqXA92t#|{k(t6
z;n)+79>w_Sxww8_iz-?%jp;)2y)heO4`u1{1L$Jp`1nBD5N<B{{k~2Th}|0=rIet-
z*3S6<l_~eNnI-o=z$HP2HkotKP0v>gOaP2eEv$bCRLUTn2e(o*P1gfB@~Z&FgV7Vb
z>oUkw!NjK7co1iPU%RH<&vrqnWokZnH7(g>2mlCV+gv9=`7I%q*vJ?MMAQ^vc2(s%
zui*Mr{Qw=n(Jc7Hg3LM@C1M7pWH2Gr^U`EN*=?oD>WxLGqITpblSgI;N4&}QT+ku3
zs>9`aA``DCRe)-DkwMt{+5T<lxIF6fRTd_I;#xE7mH=}N_}Zi6Jc;R?RuPwNhT`5N
z<&}xHJO#logS#XsYz6fJpk{8Ac?O;L*YXEMTMp~V?zXkvLbtnMjmOgS5_K-_I@p}K
zAiGrkryaNLPSAM>;PQZBKmNvL5^*93JgmLwZ2dFPET^fkz|c#g!KlpFG8(TTgbg{H
z%EZl|N;`tFLTXC^<F0By343OjWJPw($;)SVXC<gt<->RxW{M3d5I0+1RLTC*NJy^H
zaiBYDahZz{Oi*C2ZQ)1j$&w5IPb|ZGAzKGj4l<QZp^ex&E($-0*Q&c<AK9})YR>g3
z5qqtYslW?pF*oV^+Zt+^pX8r%#DB^LAA?`}9V|zJ9pTy_9;Cm%Ry@K}mXe|Rj5Ep>
zeh<w1QS?eRT)kG=_}vJNpU^|MWIq<<;^z7kmBs(DDsyv`DD9NAQc2Co?#&kRhI@W3
zf17u<<I79B>&A`Zw>v<V1*TH{fKuwN1Yf#oby4h{@<+bPP5N~bwOT$A?=;BAMR#|B
zY5br`vePv9O)JCl)CZvW>tanJ7Ev{vP@Pu>j+It7J%qAySvzc9P4}eC^CAZ>Pe|%a
z7ha6V0cQ^c(12v<(Q~q1@NMhV$G{`O?R$ky@tR1Jjt{54092z9Y-Wj?!$6}6=f_PP
za6nGJz;O0$O3u!+!4FrNAzKV{AL#fVIL@yIYVfjR<h5?c4xNnrkF_+B&hfN8hLR%p
z>soUGFfn~c5kx4SZTm4rWJ~&k4x>DniHzEaHr~OW;`TXB^(HBc_}OwO==|Dkjy6J4
zo5f`V+wCP)Bg5mwp`L|gq<FJZ3YeP)@yDI~3J0}NeS^lmI2@s<4ZXiO)?V9xLryz@
zM3?R2Tk*)zI#0J>Lm)#dDMs^sXc8bZoq&CfQysX!ihl$`S7Le#L5RBXt3wuts_Gb~
zz0C}7P*6gmS5wz1Yy}e2_qw3IMxg^7c;NvRz$;>Q5;Md!Spc)t{UU#B4XVnsqFLmX
zDIpJVO9U>c<}@uqD?vP6(RbzjXnNPP_uE|)vhQ@72!%GFqEy>a>CHl%1^XRtd7-4=
zwCvV3-E=M~#8Y~I_t*YDbM_cN_|(ZARQVtax4dzyvutD}hQ1)aHI1N1erKZom1`SC
z)D+u$)RH@=M#hRiaE6WSni=o|Wv6cHHw1EcHTC<%f>a{Q#D0i#gW;TY+%(hVRZQM1
z`ozR-K<~%>3MQ)z-fkN~b1dKR#}xo1z~A-YrlwG>?J~YM&;^{~#*-R4KDWnZ3+LRe
zP2LofDb~OQ9<)zTh2vznQVl1~?He6FCptzd)^_#s{<<bZ=3cb0GNQ*dJsV(5`{~pD
zJYWeNYq+@42OVNuM+;{2xvUp1=yoytoE?n;V{zoQ2RNTE>bPi%9{X6&?HPVl-bu1-
z+Qrs+9;}(ujA&N*ocRliAu|fix$;Q3xatk3by_`!S>t}O+5+7KifIJJ`T!gDg5hD(
z-k4cK;BA!EQZ1+@-EJP9axOPYA{&Et4gak;BFqENAAP$awFd<LjkSq8F*y&E<Cguk
zTmKGJ$XY%w3>~4Yu`nxDqny*2`JwS5SgvPtIDAw$Xb6HM2%S(S`k%|>@)BM56)>0G
z?n-4ukJ8nNIt#N1o^HDuGu3|fn-<qeIdjBU!CG?yk&pGKoGWTDf9Rp57|&#cP{7k9
z`MXbAFcv$E{f^afrS_;w9M!o+_NQ0`kiVI^z4r#P%40DGyK}m>66kkS6k?pj7Aypf
zs)15S>V-ah5MJ!j&ePpE|5UCGPn{6EG;1C(7<n|h%o3Y4b%Jq9>VIA-ff~m@8Y+P&
zap5W|gydAw{_$#*=DTC90mUE{>oHJ(E*Y<57z5O5n>qlSu7S+-{q_u>vgBJ^$HrpT
z$)Larxs{T#c-S#|em-~SICXq1|5*r%aml_7G%{ccSeM}D>?zG84GDe7aBhzLJFoKD
z=C>|a&zFQ+lE%ek7rIZ&8~tZnOmH{33Sg?nKxywbNe-d{7)AeAxSjNCHxD@}Qg5tF
zIsH}@H~$yaA|gF!m{3YK4)jDPv&G5TD%$#`{-P;<lRoD5l|Y}EpA>AvQ!uvaN{d&6
zqY*I^F1hs9qPCtZtSVD=wtx|0m()ocjuh|_EXDpMT07U0U=Lk~MF0Hz99h(b*Ly+5
zL2g8_yX>&rAI{d1ub3W*yu5F=(da%WId-rqgOWY@u}Qq?MT;Mqa?@0hWhv4ZN`e^I
z+eM$w&M+2o?W{z2m8eByvXaI%dy^JoyXK$kd$q3luB3m7j*0uY5TK9*;Q;gHak1`o
za;^3A3^ks6SP3b=7d}ZUmwF=-{hit@Y*s@rDsMkTt8dX*=H7-XtfzaEe38GlV_)8K
z=N1vfOLV6|nR+R*Y?S`9QWE5F=%S#xSb`MV67*MhDq6UMJW{YU3SO8`A@w@lJ3_7W
zu1i`9luK3?={gy!XP7fqP%OX6L<*}-<AhE&&_%mJU4SrR@-z)$Kl>K{%FY<JE5U<^
z7912m-rlYJ`ow!yIiVL<-8WyZYD3=MVC(RXt^4rw`k1-hWa@)ccXbx5fRtC@q}i6y
z=6?<VNYH=zNlz65J^uJOQ_QT<z)jO2%0u%>?Qv>Hl&=0@AUr5TaccS7Elf8&#$^k$
z+bKvjbuh;1aduCLCS%ms3I3=wCSiPFJ9qiZyNjJChnLI~>IU|5dvFXs9bX*l3WnB)
zzuZmZ^U6+<*Ls#JZSx%RTBmRR_zg0I>R#rfX*zt8LrL*;WEWi_b;9Aii|Ed5c!lQ*
zRy)4HD(uWNzOP-?jRl)=Ynz(J?8sxnp>ehNuKDqo*Fp~BGdVT_o|q}VQ(s!gJV`Ih
z6ofAhc11X5pf4QEl*uS>oY^;%(T@Z$wpZ*A4=cEvnp`ogI=wq@FU1&S6<(y~v^LnB
zW~ZLH`{s7;w*8jPoBx9Y@CWkrvnO#0A+`&G8h5qJmL^Xslt+ax$*zzqgt=&@3U;r!
zPwehF={>SvdU$$c+D+5ifCcI)|A5z$|BRS??oiXReSn93SJj4R0Ye^MocN678E6u2
zmJpd{zW{)d!u4cB85AQJ5S8Q993fSoA&>E$et3cwv|h$K=?X2l=>`=ezjsg3`JCmm
zmm}7VJ+(4i9`W8B9!!<6XlU^y4KI$#85sp0iNciXTGQBQ_sX_N+=BbF=~R`2E^NFF
zR>~;DN3wS2#)A~H9^W;hs*C@Ry*Ces`fdM*siX+0q^xBvyX=HYNVdwpjD6o_H^x$l
zvW1X6Wy#K1#!lJyWtg!mVla&C8Qbsj`Tp+vZu;K;KhJSIPsist^wGTMeJ$s8F0b=-
zp4Xp{4v`g6?G#7xQq0suq}NggqbmM4=-n~X)ag0UT!<2|@Gt~r`Smcz;L78<xbZ`E
zo!e71<m8Tp>Ji|A&)jGc<mxq@*yZ`Ku%D`BqT{~El$zDI@|VZMB!)-ErSAovkF`1$
zCec%d5AhQ3<MmE1UY{y%E7HBZ9Ruf4;+(N})|XOe9@qWszB_t4ZsFIQI{@moLOfV5
z?n}(h?CSJ(mkOuIl&rpTUD*}a`mUOcS8d2(LDN}C;}tG@B;W2xl8xAZgW0W8+Yf%7
z>pQ6dV8+*qe`NNF0<V9Q?U_%J@+GZ6+`iR0-o?pYIg>i;IS%Bkd~}#Ci7TQ2tiHAz
zJyzG-kP=e+spZ<fwAvy5CB|-vp15i~?=5)>qb)Xdz$egq-@xoX%C8tY*}q--TxykT
ztMx@lyY<s|9*S~JcZ@{8qStRPX!kY=zaFAOL>!Tv4VlUgqouJECx7dQq?_dVm95us
z&-4tqu_mIrt|!;0sLx^n(|1q&)4=y9m%L8*sQkmF`PaGpaI?phwhY`L>&9wt8?VTO
zUUYo1@Euio<viRoiB7Lfs4ZY`aiUm8IiRBMVFHR~nF>*a?0UO$evSRdoe2OSDp7u*
z)5FlJtY*!1!;uklhrLt2-tFh75Lx%5)@-%C0-levlZLS5K+GxEYEyTFoC+(9nEdhn
zxZnH&!xxY(ura-6Qa$^e-__LpR%NznxpCVz24<}GX=0#DKC1uIlHU<gS?jL=;y;4b
zKwB{&F&5x%zofm{37HUcO1LNqeuUv1Py!OokUj@Spd<Uv=yR0~LGVYl+Ui|KWOcS(
zE>go%E^rusgC6yVFNfJTu!(1d)pK|_<%*eB-?K@>^Od4j?Cv;i`lR5yt@$)vQCs!J
z2kor;Tg{w1_Ati=|L2=@5-ZJOkxuzhBzoYU$<BH4vIg1B`5!qV9pjgvGac?ynkIG~
z((^BZNj7{^&hP6#5w5=vj$cRbK$@6LnVJZ^(leP(ul~(`I`&`*6>20PEWWZv{m7CK
z$dLQ3=wKmzU0+CSGvaoKzEl1C9YcG|2!+ph{OQ_qoH5e~f_c2qMmR5WdX&A96ErSJ
zbP!GE+B9p-VFg=4%vtjQ|8;JEdwBY)(lZ;Ej&tW8BQIKrh``&(Uw^B?Y|m-ne_KId
zFKgyXVL-!djpD4NOLKfiJx>1!O_Lt53r~o*q^Nbe+f$enH1PcvH>b|QenW5kwaM*l
ziEU%C;=!0_jh06TdwUuUjou3!%dC#|bZ;MyTnZ=A+Z56$jviq~chCV&-=ll~aC-kG
zYeEnGM(5wxtKO%O|0Ic;w(ilwq{^W2IKbihA1*+LWMfmldx9M7B`5uT1j=h^sO&9=
zb(v68dZp-cUk)WG-~IKQcpi!z_IV97QXv<=TrJ({n9OF!`W!t9U3r$Lu<x||_KMA&
zn1Om`gJR(XNVJVZuO_Kpx0?bb=%lkA+3?gkU8Q{-N&^fXH0eciLyol9$B`j5-!Znj
zwLKMM;YA!3_oX*oG7soibT3fKUtxO7HMM%pW9p^`QG@S$et{H&*c@4y*~_$XbY9L*
zrH!Lj(baYmgPi8NN!@qb-qz{UJ{WR4?3;T295dGZ{#bBrM-!mm{c8=tPoY<`Fc2A=
z>pZGeKO-z~MR1{QsAIWgqP)DPIDyLRQH+uJsHg;eDlxuU4!`iSme<WVz$D>ph}lg<
z3Y!@G@uz4S>z{F~j0%7P<bS;Jw6M6M^5kk4tSz{j7=om=a9knNYYWkT9iV#XGjV=y
zG+|+QufoUrV01yiOqgr?!Yf|h!nyU8Bkd`CJ}B0jC@w-m`|=rF3#+3m7KGRKf%iho
z$1@O2PR>zY+$qq1qWAe3ox~!iC~N!D0~Q*?#yF0Yx$e{H1p&wWKgm_14>xBGCVGzG
zWu145`ueY#W{u#-{a8C{_8x`2$oP6bq}VFPSjOw&wyWy91wfv#?0=_#(nvaWxXgHt
zyIIxnwE1R0@rXg&L=>s!N)(B5y{#I}zO+V7$i_;`4QvJXV~^wbvr5n8tZ)6;$sHQ7
zh4m)OrRBT|Ve$30=IQVqn`Oz8P;{1N4Ew=HB^$jvv;aJq-&6L>IN9q~2m8=UEWcFp
z@U+kk;2@pYtnE4m=?a8=KeK~|wR1szS<Uw?W0OafU6Vg#lQ-&ZYvxk(fe{z$m3l1W
zJZTO67zN_fH0iPb(Bi;eaz<Qr^083_y&n-uG_0?JXB~w_HiKE8N(Y$iAmC@^{yAc8
z-kJGOj0J~miM<mjLH)d1uQYz%DtX}hyQA`o!8=|bvkaigFWS8QGjv2JTKJu1uhLXs
z?{aB>@zEHzc}ZDl@lBKPS-n~y!6>h-<uEVj5v}9(a!HjR{rT}%)w|61k%gu$D3jx=
ze(DH??P^&kEogtB<xyun@3?23vH`YsYaZ&g82r+)X0OqBC@M|dL{BGGwqmt5HbYWY
zqq_^v<94FigVdb{YOXk#$FU_e7rE=Kh&*KFeivH&`w!6-koyTm@|Z_;7Y&#^j496-
z^z>ODTG<j5yHIqiUt7CI8RGoPB7Jh=6L-9dS_!wFDD%>jrYmRgyUE(1`I|(x^{#<j
zO_ttg<4Z6}%~lZ%lhf)`NU5w+?N^75#iE5*O$KXC??fHLf(`7+vt>pXftMDCjbBt$
zXBYqy?2DGzbVo1fYB+%TJ$Na%QPo$sz2!bTVLah`c6dbWRzh>O*U?WBMtj6rh|nY@
z#D;7${mrsp@rvl}4Tbg1&$Y=FpID%VnBPl6AzMGE<Da4(fOD5F<7V?)`e*Nff_8Fm
z_DsO<yY=2*R3)xE@$zik7K#W>lAX$J_<B=`voV~w>Fqxp-hVBLRcJCwyJJ;+&v=jC
zZqvhQ``4<+$euZski?3pqwYKB@RPN$5%#sBSB1w{akOwJT)`%+$?C}ZuZ@jAN#+L&
z34UeksO1#nc}?Tzrf$-2%5pX%GDZQY%l`TL?U4SH?Ur60XU{@je%s)x8q8F?$^u)(
zTRxXu;p(h>-9lfRofB74TE=gRbFq(V*zag~1x^;kP)FRBcw$lo(U%)ppL4gQ(DnVu
zo3v^4r()TX?rYZ8Ra>#iXtwf%$3w=RqU=4X8gcmr;Xe^7QSw$v8VSQRwj*^64+@-r
z`It4<5=9i;J2id#UvmaOKinicWRmur2p-p;$s)h{6C2sSLp3q2BhshcKu&wr21%cA
zt9We@nSo*&QIU(C)3g<W#7iMT=W#n7zT2ocp#nz~@Yu6&Xh_kii7w-uacUcp0<jgl
z7Z#BL-n}+vOc7HY^|n5p!Yf=)ZR6YTIIS$x>ooa)qI-QFq`$N7A0IlHy|hh27Ew8U
zNgHktK*BH8yOcnNI?)@`*&nZ7-BW`$HWx~txS2xZhU`o9&XA&RTH2+JR{6IIpY;J*
z@=FEu#*YSPeeF-=v^0xJmYsVL^XI?smVPh_1jiio*7{?*JvZhhuGd)Uz42p8xO#<>
zS+tPO+Qeb*6gI0OU9vZ;(7MJ`<DE!<a(qi_xN70j?xS}!pMA{K-5!LF%)#w_e9Y`I
zuSrO_HlUXla^zh7-c`7PstDe%{*Hxf!%r+6C<cFIgeJ|7e_&r~Op~xU^UzUrDT-My
zCq%aFX}{l(ONB`ZLO58&LRVHRe9UZ4!}y|jYJk5<LJi*RN2|I~%{?97h+v$fXfg`D
zK;@+a+pB5y-A^L(u@y}nIRGyIjg$Q&2$p-E<a38mnQCFI!?AZ;^_WIz+VxKxOHTSL
z>_4tV;dDRniXl_~nN05C&OK!Wi}?lI&F#CDwPSaVCgZ}~``c02!u+mFs0K%m*-3YB
z*Z4-_i?|5L8`a#;5~D+f6f=G*XxOnH>@Wz@b^u&Y{+U`x!V&f|el<*ych{D3CtY+4
zKfTFSQOTu49_GjOOVz)legHElp;g`XJIFDI3QqZo)z`{6U$6m$VKDp<$$2cQU;P3+
zO>0&sgI~hbQiTrySIl5$cmJB%COPOBw5=Q$OKtz$O=?x&C`}sY#>*WhA_<gD>ia|X
zu<B@$87Xqm-bMj69duLVY}<^{n`&p5Tc0>p9Xr=#ywYSmT??1r9_mY|lr6?7zrDPW
zt)b+}V=^`>(tz3GPm?M*`qjBG2Gqu*`qQTJpPx?o?`j{;?y*V^41I2_wNCaFJrlk!
z>>kM!?=kL(Z()B!JlMIDJ<r~@-mu_wh)nm4q$`<vHglp>%n#5Qoh~N?dx1gY4Z#lI
zAKzZQlIT&B0jBjSA^Lw7+V2zRi`!A&?swM9qwnWKr>;pG>`kwWQD`r;>KjItZ{MOz
ztuOrOI8}NB>tJJ%>P!Yao}dLMrJ?mcnZv1LM<iAHXz{z-i?4^vfkkHj?x_UHaILFq
z;m^5!;N3dQ;$s+TJ7ruKQ{X=5U3VXqvy0Gg<7aX}H_F1*y&8QdKZGe{Im=W{ToX%K
z5ai3{PKmjS^48(yjc=z3Gd$c%l&je?P9IPVsMp(dwtb{~d=Ml$H#%Qpt7_@=jZTTH
zTTREqKikr2?puosw-skV4T$W3iMbd>=p7l>Ok{Ya8{|6AR6U^O<+^*w`1)xgE%~GE
z3m0-nuAEJE?1?DwxqNP04Ut!;BB64#1w<Uc``ivnZ!DOp^VXd4b%}DS*(4D`4kryU
z%}hBk<VdVa&ba6LAdp47&R^o$_J&gMvm(~2`$w~J7Oix8>gag7;^8YV;xgByE7fP9
z?#AD0ICm=-2O+A!GfCeObka8vERt5y%;^8@JX7+(Ns-6XqpDCRVA}t|iGxwI79oHI
zy3S*Va<Te!=>V?ltwUC=BIS*AS=oN^J{Kv43-x9@>%j5HO5w|w^Qo}0jZoF-r{jJ>
znzMHgFa4ov1SYiR^^%Vw%ePgOp+79b6h7On;>6l~$^av1wLJfhm>j7Aluc>hpvI%$
zzP)1q;ec-0ylzg>{!v-Y@nw3e`85)2oS??L?Snm^^Fi(W5Vv?I@mfQ9|0wtnWZf&>
za{<&#&RsA~Hg1mD8Xr`f8#P6Anj&m(FhsKf-?_NMN@{Gt&PeDAEm56|>DQ_q7mgx8
z@QC4waylHbs@T?Tyz<NI7@_Lbs9(&vQ+})0R_B;rHiWG(zcWlWu9X*?4-B;Lb_gr}
z`)j-ERQD@=G%`^l+2}#3FZ-3ki?7SND_rT)YnxPBYZ?ET_H7d*qt}*xRfS?oQW8Sa
zr{&^Jbkn>R01$8MpMe&i9=SFEgiy@1_szXWUNRmgl8E5zpKf0q0_VxxPig8`Z#U2)
zLM(u;&OQdpZThC~Z&O@|*lXzhdPMCE#YoWLo9I`|avsh-=1xek)@u_>f$-&+`<g$o
z&wIZ8%UX1q`N*!cNKf0Lhg8lKghKK$JZoAp?$ZZ@FAFMyW$L%Gy0<z&u0$w-KjlD`
z)fR}Ll>t^NB$f0T1p(0U>>hA}s<U^7SRJL6CVN{dybLC&Ht$#~LTw9gci_Q_k9b`O
zNZitibXP&g3%pEOgo`E>6?7;KskIfcb(>!obEh~?iO*cP_ektuQ3A5#tiB8!H@hf%
zK@sT8-Uz-UcvCAiU1BzA{8aIDY2||$&6VB7O&wl3431MVO-yxd=1C!QoqYwLv7zv?
ze$G_($fnvcw71J8w{n0pxmQ$gqUFNHnGWB_a_HaV;X&1ep~>w}L&IH?porA~TA&)a
z;DS>TZETWS@ze9Hk?~O~6r3KsD8QE+IZS<yLcdKY%fmyMU2iGg)$ba5g?o2xm|Ib@
zys_FKz1F@v6f4Td^TaVjM-Nc9@ITKUF1*AnqasAkV;`3xxA~nR*@`l?Pot(hGss|R
zrhTF?5WqI<RnBv5w7=M1;tR3t9gCCLXz9J_@mtYbmDwpZ)tHNzgMcVR^z5OQQNa(F
zB4PLf)Y_Oee{9C6^of1GTf{mrIntajkQKZ=7lRr~ELYT-8|qy--Z)!Kl@~tk4Qdg<
z4ptgDj-zbCqbi~%C-FzyT5YbcCvvwd_V5$jS91rfVW|lgJ{DzTng)v+uc1~42J==m
z+xt|-NN9ilFuKS%uid$}1HJ~lH%?ctN0rfZdZ1ksCxOboA^(A1kE&wl7F`5~tJr1a
zRWv|0B{q|}fl%L{wH%UlRVti@=c#me!+AH0(OzjgeuFle(uY5gMa?T0!|g4$vwOo`
z7HsYyQ#O*n`82}Tzli!_T@Me&<l=YZkNtNBKJ+09c$DdBudwyhrh3AD<VkiLS{lt_
zZ8^U5^$*KJ8a3A>ik|@kfA|CCQR5_^>r`K&AXc2Ova|E<A&tcKE_Y7R4N>>tD|8Hx
zzeNTn1`}xF+utTT=q?10j7f=iEhpyw3-n82OjzFWG4usKmr_i(DbmZyFr%^FEE;Se
z{ElH>XVjrnB?G-g2VI0hg{}SPzT-A?{k@Tp?R*TC`g)%r<0*j6WKZ18G537v%crkt
zbDioHE53c-Ygw*wvcPd0A~`*lBU|U^(6V34AbPkGUxELo<$+(?M|C(MC~9%M`Caa-
zZ=Wd*`G{0Ywhx=#jRkJZGIY%_i{A)_D%eXkv$Fsw#i3Ur3fS%~)QDyLL}cYIhpkRi
zkKkrw!yn-)5l3(}NS{QO?{aY>Q!}SjRG{kM9`7|(ddvBq%_*;fVX6MaoZwa?8hk;5
zrLj?3)=OtYGvCY`_TINXsUQFFIJC82xwCp=aHf6w?ml>4X84gGRfJVfXXFLywU9pE
z-Zj^Fw|@F#Y6P+QQ8Yx=*FpzryJoz&xi~8yTqDkRk-pyvu>@Vy%`L9?$vs$78*^wT
zH3~{xnXP#MeE2W;NTaF4Wmof~qTcOspak{U{b4&B!5dXp?X<#yoh`<Ke<;z+(7dW{
zS7uu*P$sWwK_u9cW;In&4Zh)rDZ1Qam%hEajOP{h`Dlgdeg6jlJkZv(pl=7J<T9TK
zzmjYKu7n*Z(62`&p2BA?q0Br5#o&19GR}@=5QRmPW3WN$kdMsu&<ek_b&q8ld^?kZ
z?1+&{C`}neNxNn;o1b}dZ}`4+eX$nM&`*pHk$r5q3y+ux+_TKkl?w?6*&oM#pb|YK
z=0ZGE2KN1T;7<E<S0S6s>h_z_k*`D?fn!#uch!P1;zWoZ1IU=Ypjwu%#kXsdbRyfq
z_yg0|Cp=Zs(*lbsuG{cS(!V;%w}5x`T%y1fnOApVZ|G7$ky(J*7y1ua2G8V$R*}r~
zjS1SQ<s)7SWF6~`J%ZHzI~E|fc^l@11<>S|N|UOe_EmnQf(YGrKPnw9C#9x~D)&f!
zQ7~P7*#crXn}B2@KbEdWv_6Nr31srNONP|1k)P4CuD_QKoEg}TFuW}vV0~2675e=F
z>(5}Dj@?`8&PJaa{j5GSHLkKiYT?4BFu>@{{PO#?dl~<9y=nvmhODChdsGkX$m`qI
zUrwmo^uI~GY&E6t?0AzAI5{Kxg$a622w`~SA^fLMYWw^EfZ_y?)?-PxyD?PsbDW+A
zFAG8ZM%T{IY={2-#3Bz7di%<`#1{o+<<TQNGs)S>s6HeVtz~;ZS{>j#B71MXO1~(0
zq&2eF44|cFUIDtv3O3)*E^ra<+dGiEy524q{pDz;eu5@KAr_G#R=#`TO9N|Z5DtJ|
zgDODaMo>uS$@pXmFTkE%Vx{dvgdc+h7#@j=<*hhDRu?eypgAB$i*7D0=%%BT69uuk
z=lQMRh`F)J4#}jX7oeQPjWe7wTyaF;t5aFL$?+G&X}7=ZYX|VH>dhU41^t(v)Z$?E
zH*Z62&+lMAH%=E&9dFPmKM|V7DoU{{1wGMyDOWn+nC9g~qE~FB{iztzIDBWfUUB!f
zuy|(9ZUvoNLYmf>ijnH)qmFHp<Uq0<(?9s#*I><erUYGVb2w&_1$0zcxu0oj8JWHF
zXt=$?xI)D_N!epb0C_zUOmECn)5}J+2kdtn1;@I8_EwJLxw6i(86jSxOPgMfA`4S$
zwRN_o0yFmQYG?H(pl5k@`*-{5Mca2%baET({W1inCx#1|rCsmAUdpviW_ieOWJ^>O
z@N|EG9ueI~jqzN(18zlFuwu3MQL8beOtvg~cLNN|@1OtQP)WC)x(RlW_pf19Cpy7U
zSK{xEPv?Wd2o^%my*|7B{ajRa=5TQtw_EsnE@-svgw2X~6T)b-VsEcObZ#Srt}gwK
zQrx}X-XNUQm)jS|mZ}qL-Qz4Xl<|lSr5!fLdiTmxL@sPFT?X9LtQG><d@jV}CVx~-
zt3cZJsb#ISj_GK=u$#yD`(j2k&F$BO=43Iuf~IS??ZwCAy`RMLpOLh${Ja%o@y*SM
zCO-b|p*)|lHgD{@8U>~Ni4WObQzrrXQS`ZByVzTsr#-jfW~@JJ;4C}N!%_m7bEz;S
zTvnwBSJgCb=m<cyq*KMNDNCLmVj%(ACc>Vzs<Pf_yD1mX>V|ccWsS-GAodPAwvM+~
z{%IHVFQplH<tNZ>4V(Gl>yfVm)5c~b9isyJvF(a=j(R?o?3vf;!WS<$!c-dr_C5&e
z7%XZ9IpZoN8q6K#zZo`&sUnPF@q$IXTl!8jGn3Nu``><&w?Q6EVq7Cjp6?hK&Wc6m
zliyL)JKqEA?kd|#rPo<x#Pr0n`KT5j;w}OmaFy_kec4UZ$x=ZyUpqb+YesHV^?rSq
z#>g&Q`{`~sxVJWw)GI|GQkt)}0oMF{%%Sr3ggG2|-5T;vHI_I7F&>Sg+~CSP{BnGX
zD>H#Nuqyx$v$?*H#5gH0?e`TAfnx)()RVkf=QO3j=ie$31=6R_=tb)jYQs;!gxC0M
zB??Z%Sj2gwYJmY~8H{!=zWs|!s<|-9l#YH?bZLLhuuyCBdA3VPr!`_?Ydy?9V*TVD
z>>l#*B77kbEuYc=R-kb-Zd{?YNNtWNkVtfz+AQ7nJqk-${3IXz&j>8C2Rq&li+IWI
z{cAVC&nDZ%m+u^<50M)m)Nzi2k=VB|ea#hQeX-c17(U&1{2_-H-HTu<j1EhJsBwoO
zvFS>~1m4QCl9)W8*EHnSUgMjoA?068Mn%Zq){XBf1bP^6Zk>r1tUd)pH6`^6c>x}K
zVz*;g(?cKC`zRCH|Gf(Wn_0_pFR8aFU&Z}M8yCN?*dORx2fVEGTDxd9_uyfni>Or{
z^mPQJ<c2W_iVb6amQQRsixm!ga6is*cNXg1tJwQIZDjCWByv~Gi<>Uu(xnv37p@Tn
z4HA<bWF{6yNc(^r41@Q?SrD=0MXVyfI{#A*PIQl~d%K2rt`bLV`BTj?<)gJphj^cg
zTONQ$n4KUACAePhE`Yjznc{p=<6$0wO6M3(d}|lx*XDZTinelzaj?JoC9hZOWFPo$
zrY)3r&pl^z6(fpiJ)x3GADULD;;6MxyO9${EV@S7B5cS(sgG|b!*9cSn2_wTb1$W=
zz!5U8Stz(Y)$L#nP;0eH_$7-}$SGK<f_6X+_ZunVcPi`YAybE@^A?vZ<n2B}ul;8q
z1Fyu#QlmS@g#_lK#vIAuR*{!1T9-zySadmsMeK@raXJfqcz5p>xDCq3Tbd}hi+&eR
zl{B6dR2@mEF?jmzLFs}QXc8+z7ttmHe|~*i=!dHW4@J%Xd3c}XLP^q$9Kfja{9KJ!
z7O&p`@&wBG35M~*vuo~cE9rPnRkVokwWn-!^vDk1g{xP5Bj#YeU)gYHE>katiNNtl
z;JxQ^9c`8*D`;n4Bs`Z@bv?p!RgQ1LE#wq#*PUJzSlcz+?x<<3=6oad_2HQZ**Ee-
z(5K(!)wvcQ|07Y9e?1RiSl7vEq-<8y$Yt7|^`Q>mwyRh0;d3aQf5n-c9k0+%n#Hhs
z4nHP<`e?>2Wk=KGh7m4AAE0WFE1+<owSrZW`D12&8Xn$FgEkO?#+F}Jc~~+t+>xYK
zLjOpMvQ{+_H)?HmKr0(q9ZvWqe9m+)3&3@w@q!Tr)<u;d)2737P@dJm{igP-cOt#s
zAxJ{r=0$ELUOQ2gpzXz<E14ivcpvp!K^x4zYpKLuPcL3DJKAF>^x-G6*Fcvr8ByKi
z+Dw{0T>~1hL`xDXmD%UutW`dJQ4nGP-NMpFE_z$fykgF`25L%E*NBN6?j(WAwGmEg
znKCKxJ^StL2Y;<2(B>^jb3q8k;kio{3BUBg;J5ung7H|Bf%J+g#Vgk5QiW5xkouSS
ze5tgjz<-JJx$e7&q@Ndr^!XQDqSKq2lO?^xTihWU3DWagGlMxm&e?G1%&9}|4Y#0R
zq`5r@^O0#s89*977!f35-_3$RKW2C^d9(}PGHvD2Nc}*12$c_z43)*w8r82U!Z{E%
z(!t=)68PPTJN}2dn9LLefF*JQj+10I@_X?k<cjW}xP@<HT9v-Ot8g7mSL0-S*Hs51
zAOTz$bCY44@yq{AiYT+{ZZSwCqf)O~IG*1zXp35LNE$E)JJTD)n3DK*(tX~dW&ADz
zmG1MdLiTRv?(u=;)TZxCrb*Wa$7^Y;HQx=QY>}uoAsI0>I|_Ycd%Y!a-cLOuF;ZV~
zE0y9~{9nF{YjvV+yem}Q?43x%n@x}}I`0@`Zb}Ka0!LjKc!%98g#`x7wSB)+j`6>Y
zR&CDdE}RYdgdfjNr8}qC_xs=-WySWjcF|<bwAi$kDWXTB=}T^<?ZYONJsyPK2zU|d
zYRArW%ME)uopj4d=E%ip$egp~Dmb)~TH5i}*V`wSYcJ`$x6dWQ3Tprhrft*~alJhP
za&zj?uAc2SUBq%L$X_`tzr;7cDr&<%vghg0pWW^FzZ$OhpnB91KT9frCXB#a_oUU7
zpOn^12@?6#eFiy_Bi^~jmbz+FKBTIR^PR$?9>K`{vmskS#&rcn)A!{+51OLAwkeiD
zDrHB6goZBs&}bZbc5Cd|XX={Nis%W5ktf`3-UHjukj8ITDnO2#G49Az{>2kOPwWV@
zN#=5U#Z|%qG=J*qSD95K>N#TsLQ2VKl$%NAY`*HBn=KBKO(`mf!S7?!AW7_7C-h!m
z7bNUGwkf%|n?)Ad?}+sbc1Tw9=|1^5vI*Z_9#m5rLgD!ee=j#bb{qHDta*Gmm3)=n
ze$CrGy<vY*h5}eWLzt~veSloW3Fc>99gvfx>R6Os;~vyk*oiEp?L#N50S2ud_BsG*
zH}mG#Pg3mh>GcJO0<KsPN?2Oc04qPfL!J*3>TRx;O!u{KjO@+Q>3tsj;vhP4YnN!@
zEZv{^-oTAJ=0VMsKr7j}JSuE1FMrMYbk&hE66@-#a$lSOK1yl6Wa1>#^o5JR`OXa3
zbxw8Wp-McT(led4kN-E?JenJXoro?r8q#fY1E7Nc!v#=nI~dPnfUV5xD-EgIbH4{s
zxd&Bpl>6e^+E~%VQI$ONK`S?F<<|!Q`}G*UVo?YZ&Xyv9nkGjt6xq0f&$22_QV75Z
z6cN#U(u;z1(5VOV#<9S>+EH9OUO~IP3Kxd#2COpLHCaWHhL)x!H&T6erG6%xidVXd
zvzy}EYhs1(oBiPw19Q(80RUK1q~vyjWO4N?saGAJKwf)~+8_Fd>n-U*iz&<9qkLfI
zI_YO$o81%vi`5eFm3371LuM)wGOSqiPw)ZInH!?Fy%lFdckNdg_9}mDUAS<)Jy960
z_NJ>&MfzYV!gu{ivdjinCt@#wvI2_&A@~xEH-6iz&MnUWV3^oZmPud%M7Ryx05<C{
zt$fw6KlwP$!pFT%D0ALYPv3Akq1zM<GW)~4vH^Pg$%Xg(j$su>0P@WFo|B??go!i9
z4^2&6i3DM#82?}|Hct(j&YCR>_$qyz9b}o3#qId&QAV=-frQLnB4lXexj%*<xS?_%
zg)(WJuY6rqVplrz`EpRKYJv4o2i?rwkj~iHC6t!3@>USV2p;5G*-q#K=qGF^gb6zA
zbRAq14T>|2dHDx{+pYiNaG9QA=o?G1o#G>P$TOYy0y@#j87Cq0i$hb`I?i#tyy9uh
zRt|3jexpM9r)%%j&J-<W<@KA|O8B;PFRW_!<M$X<hgK1(;c@fp{oTFSOxXi9e<-`@
zwE=6#T5x9JDikjE$I&B-DQ1)MwEg|EP}<kT;Im+n7>S;n&{W>-P!UB@&xqpaiis;0
zy)Kt6QtdF>?c+~gW|T0EqQ1;!&tOQ%$=H;DW>oHmySqX%0Azf}EY#yfSNcaOi={!B
zpgl(=Ow(iTXJx<k2w!<)heRnHHrG#VM;kkf0>1l6>~ie4(PZ22n#l`pjqcp&St{WV
zEbUY&YoI!_PYTSri7$Z;Bf{s;vu{+H3BG7-w64{!RTnKwqSsR{ljiDt8*J1Lio7Ef
z<^cMLQ?01bs*YdxVwyI;Zd`IBV2FA11;{#XwT?eJ_HsMSlvRLFt&hsURuIL@3H{7Z
zt)0yVR!;t+VX}p8%Rm{;RdMmVy^kY$7K}DOe4*%d>+ZM5CQkpfpa1vb0ZpGVQJg2#
zwkmxZzkCP+?RF`^R)ep&1HlVl_kK~3@ikPXHLgqLbF@d^yhV+c4uidp^w3pubrj9M
z=Fi;=UI?LCpcLY@EZcsch%?cq!wFW^u5+omTKVx|n`?mhCuW(3fZ<Jmp5|`|aX7yu
zvf?#P>b`?ZlUqw|vLQUuN<%<y%ZoVruQ`vSz5v!I1q7rxz-ZdPMHNmR0z1{_ew0RR
ze*<)X+!D``^cvmEwK12rocnfaPW)0Fooau^7|z9@I5JuPEn0J7ksHlBD9vo(yeBqQ
zzWVCZ;G#>;6Shku(VpNegNg{xF~`z%L;NKo0XhOF?_P`rh=lg?uU?<ac%<3CVy|Ux
zUEoo{BL&IQTWS-h8?oij@PF)gqaBo{>8BW^ime=ES<|SFnP1&Q^2{yVJAN;X(v$?L
z(}`oReE)-=#H=NUJq=g8lhzvTD~|97T-6MoC(*3n44^ii=ACLzLkM1kpK0O-e(O5`
zQ}d)p+y@{4^@04K#fBHSyUOB{m$U(0f^8{43~g_6>uMD@$iunDZ$IAx##QSlhY_7L
zFs>t?8M*o%>*#s<eq)ajnvA^t`~1bVx+KHuR8Lm%CD*ZS5afB_&Z!p0W;9vDB85q<
zjZb{?WpfZ6|3kq-M|l96bI}lTUq3>mBG{-K`*qTWN2#CLM#C|K|HtII>F6@(BS@sS
z;bTblO3mNsEFScd5c#KT_P?d;0th<j?aUJt1KhpJ=6B!>paYulf#n>qKQi0<=<P~o
z>0;%E%~v}q=kjs_I^eGrP)pGmRb=CI)dP$+X1ST(erpTEKiJN%cO$vi6ECIyZe~t1
zN$9GwZDKIszfush0)+FQwB5m>y{}aO;O+PqK6;ih^LdY?^vN>77cceEU!Yi?pjs7a
z+nW|1xdKT1E-}@EzPoRlZgB@ygPunJ9xV}3NfpSEg(BiG79Ht0%^Gf0_V_5xV_h$d
z2C2u)2~`Qm@_r?QjyfRga@<GRx&h$&1d7KGiCE`0C4d2@+yq|01C49>5+8muBrX9p
zBi=K4ew(u8)Yv*84(f}^qPp<kzIuWzbN8yO_d%e3t@+pKi6Y%w2`|r41;+NBcraqj
zxPd2Zl3jKtPTw{D@REsi4NC+BQ#h5d_M7jMm*RV^KpuPLIcV1e>i02}<;%%bxXPYY
z$7St)fHP@6TMUy{OtRKvwHKxv$u<S|x~tWyViF(Z=cMRdPPSjWJ}{l_#M*=Zjk7x3
z6$%NTNO+I;m`i%AG%gZYZ31|3t_&+wfE)V0$@tYDi`4XK00T%q>VWJPA%SQvDmFw6
zUjxY%_mV8dn2PA_{t3G6#A6@GI>~dzao<bydQ+69I4)t4Mp05LofFdQJ@D&@_1eDF
z98yI36Clfi2s@+7y5?7<y<mxIHfq&7Z4tJE^u@Q`YG)IQAuJ(Nq9n4!fn!|^05(aW
zs<2nG)sFEztwKfMyxe=CyiCSB7iA=D4CK|^Fec84A)O3<w9v|mUvxIovJr@JX=pC2
zCj<Q30;U9r8efh~j^2xcuLHNZt>fE@WDXCio7+JCbjeMJXK0F9+89J0Z3o-=U9al{
z06Ih-Ru^CG!j=c5PC0Ztep^d2M^fwvNA%24Eqh{vX=CF@BpdGR2PT4uBk+u~o>sP0
zy5tw|WQ|$+(ZcLr{Dd;xEFr>qWm)5f@R#E0MkG9YAi>J;nZL;`Gh6Ro_se=8uw*;{
zDIDy;I|q#AOEFF4e93+44JfP6?@V=dx?`uK3ejsulG!O25~drn&001K@;i~!**I$F
z<KZmwt)=d(Dhnv$o}2dh&m(r<(Na|a+@j0sbNzKU+5nnAnsx-90VLxZj5)Sq)jOfc
zNWVZhr}=p;pbb_)??vJ4fV`txeD_B@YMv{V@*VYEZkt5m)#k+AK7cz$oo;Wne(U@C
zL^tE>O~5pGii<V`n8e|)^~3lR#uDPwuy`Lp`##h5qL(K9{Nr%`t5q*DO5%Z#*+Mdm
zq)@p47J=EASAd+E9~?ntLMf3Wsnh7Q{)H{KqSiQM!WSe)6MsU8H9{K6{_s?qO5)=!
zZEf{XB?u5ou`i+d8AKvC>uvGBNJQQ~{QO+PfsR2k@xmmU!Tt+dhvY8J?|(FM<HEcp
zo2^)m7ZB4Qk9b;>5|&UN+{)X)aLvh}4^9=Q)gN9W!JpNNPgRR$33Synh_k=m==UY{
z2KR#SU;c(w`p}AdxA$f=<=pR~ng?U6q7sarTJsdlb?e!^lU8afkq?Xw--%p%46?ur
zCiV!+g8`6}?nd)PH_Mb>%$u2e=A(A@rZ&->c2a859VBK2l8mK^C=NTjKnPWsOoFrl
zSA@N#^gO82Xj?xxwN_#em>~OPFce17twdHy>U#3&?6&U(?g<-rrhtbWsOi-UH=FyN
z)x7HKr|t&$;VQfYhRtLWc$hLCV=rFqfnXPq+#+06zy69SIHQMYs8c~?^_EKrlz}v7
z`PQ}eP4YXo&q&^BGFQ7a)MfW~IZIb5%2Dg<35pSVG?zCjN9OZ75ofi~z5N<B6Q<#S
z?krOHbrRUn_iGUqNowWM74_Elj$PI~VVDf*WgL|{YUdJz%^MLoD*RU+N#Ifp(E1^K
zUj*BLRspQg*C3$(AW0B&HoX#O^n~ra4UVq{Nm_%pP4W<`Giku`TJC@AdAhiHt)2Js
zIk>{-r-!}HqFCSu4em0}6dU;O#7O#{-2yreWU?%WcCNp9=<#J_vg7ePntL0tUoY&s
zX-|nd+B2XJna4rc4MylFJ1sJ1aGHZg`tT2}XY#E`PydrLHWE8H)tlGVJpkydya@Ca
z3yw+5<gy{8G0&l<GL(&LE9-juWm^lFopC;({W^sk%S>jJHSW-+*3|5@^B0Xc>hMX<
z!I;)oGy=rU!b70br<&ZmRos^IvRd)m<cjP{fH`X3u(oC_nK!XE&7p+v?&Fa5pOD|E
zvXN`=`Lq2X5s}J=DJNtqpl@fOoO-)^-%r@qix&Xl%HZ$4WOe{Jnq8&{)m`oAGr>o1
zNKb;khQk5et38bqNBisLeUHIUrwVhsQhzbAF9A}RB}k40CYqT>_lwJbBtt{cO~84f
z$^=-B?JHgO+*0mLbfl-<l;8M(pwO6l<A0i_f=VC@fKycj1F9nSwzE_%z)+u${u3q&
zOTi=&A#7Z~e8T7AO>7KzVSomC66drv2d962y5#2N2+BqVTEuQInJ7oiyDrRIr3~&7
znZQ51EExO(xChYpBhS<WbeRvg7hdgQs47+Q(}A522w3PJo~1WFkLD}n)&z8kkv&7W
zblerZSww-O^l5XuOLyn{a;WIq0c+t9Cx=-Ar1NVL1-p*6@7NgaRtKZf?tbhyYH<0s
zFDwSvUa$(k)^#>;$?Ak!4tR(0an=!4N9)qPc8iY>W}C=K$1}S*VbG?`?;8JUc?h=Q
zHn903TGg*~HQxP`;iL_~oCm4oX4^Hi)I}7C@y!oFPNKYOxOlC>Jg#B4PYaKt=&txU
zU_W`((U&bWo+HtrukoGHlATDU7=Vi#^`*N*^{QW=qSLQ&00*a-5i35oV`q{_P=QF)
z^j`N_JUgf|lA%7gq|q1vYxy{G8?b|g-Sd34hF+5ZLnF->GF@KRa!t&ATh%6DKlvpd
zqbjux{10;{f8ro}BBAbYpLCm!-0!3A&3rG}d@1+TQD#UmcmEVem=GZ`x@l4~KDCIs
z5#w-(%uba=`QLGnyz*wbRnAdgKRyd;>^uPSJvBC_K@Qn1*nCBEQ*OXjv9x@XFv!>Y
zSS(TsN;)ESQTsJ?<>?39QOUF`CJYdv5Iy}l(KJmL%1SzNfd8Ttvm@#O8mL|*qi+Yc
zS8|XB7$6oUJfd-B@G>-M<l4F10D(_p#XKY6@Z~toHC-drLrw3i&48UM;vV)f4$uVq
z9|ODv1Qs}dPhstyE?9tvETzxDH355yZAnYR{yWIE-sMyL=A&GO<ehun_s)*Io#_GW
z=$-*!dIA?@>Yz?A{WMW_GpIvz^{XpW|2pu|j|p93-y2Z?oXI#$CU)$QQ?3M*n+)eG
zN<~bhk0J|Nv;0pTJd*s&Vgy$A&X?F*%2x_D?2pWyF#VavMlu1`qo5k}?$h8GcAE-2
zqrMUhPw$Ry8$~Ltxrn~AY~LVa4pdJ#8$gPT8;%*Yep&%~Cy~I&RT_hr*@%M`OW2R1
zLc5j6elz5+%<gM9w&!|?u!d|lM}J`&8x!`SC*dIMPD7NPBGEA}CGg$_>Q-Vb>pL3=
zcxicu5evXEg-AR*!7n`g?xQd>RVoNJKzA+<fbEl|8YjyZxe+zFf8pKIYb$108ej&*
zZ|4IauW#e`j7P<Ha%$ZNKRxJPz3yb+ybZzy5;j6s50OeW4XDj|bW++N4euVO^xioK
zF01Iy0Fo}~v`7}=!w<N#tF~lxbK9GWPCRD<ExI3j<&Lf?eGntvPzOzzYRb!bl#Lra
zOHm^~mxK~{hs%ts9x~T@H<T^_J4Z&r;a^SkL5;v#5DCmHkim@_syf6eDHOIURBOXs
zqh(dI^xpg^)OcKGj)=&K-rjqO`bDc+Kx9=D+}iM}dcmYo5=zSB27v)Qw?-qzy&G?L
zh-=0yX(aGIQOh=<w<8Z10TneyGLtg$wP)K1)1gG!hJY$2LXvoX69FlGl#iiceh!i5
zR}@WlY;lXX%A#+dBZD)ALLN(jq&-RSwBQUaQOpl%Odq|czq1Cyl3@|t!9Z#4VOe=+
zlG6NOs{v(-CPor6c!;uFyPn&dRCJ_g2g@75wmyxUq5&t-d3ZC<3c!;1l_OnN?B3?4
z0!8<!CKZ>~7&Ga^y|9X<SbCn#lZpgaz;gkqdh?Vx8NYWTy-NetBh_smZg9s$5QIS|
zuc=xXA-{U1!>VivI5>rOrWQQMw|7M5Yg|=ZzhJm~cdoTlKyCtj6^S#o1|IzcG-7?}
z8=Oc$L-VJVjgoDAmz&wmqmzXcxKdfL8+CG37X>+4_#7mHScFtAwo1dqFQs$(4+Bd?
z^sKq-EvTbuVCw`N8Vn-kGt^j?FTFiixu1(}J!aGj-~d4`55ckS>UPIs`WaEUwvW?p
zMp-~jU$+NmsKIZ>vs*&6u9TGM125o=YtFdBgMGo;3NizF8!BQ)rbZA~zm|rM#5U7o
z)MxpEO8>d8=--0~v@I<cR<`GI+v<v@)S9<k1oZMYM_@Xp>Wb_PF+za8n43|izFX&M
zXgpD_3_6?@qp%<lup`^b2frV?&R6eO)vOzUOd<mzq{;#5sX#W}CqGBbYm?cP_LIyb
z3W6GbE1;t1f<S;WLbG4K-_p5%yfW5YV+pG;q~zO(Tq3AKBqBnBLf9oQW*K#$*2}9}
ziro=&;e9$+X)R!$0m+~^z{s2dpnb79R20w|5DPdp*5JaKn2BOcO*8t%omQDn3X?bL
zY?qhmp{$@$9g<X36ANr^9Gi$|Fz7X?Q14?oHtaxmjbz>S->U%rqCyWUMgXaxPaIeu
zpy?!5Ma>U-NxkxA8LlH?&W$x{5{WpW>C;3i2>Di-Zf~&0Y_~7IB?pZuPeL@Tc6?q_
z0;s~pF3ACUer&n2$qFILW1<aM8eDxF7oZb0Nq1=)s0V=KwDd`yWCN;ZRQDSnmsPXA
z^44~OlSn0dkHl!q?7IJ|Jt!AD1@f*k{fo2?>TZ%-Xe?L6dgUC%cm7fPPsc0QJ`bkn
z{yB@L3))F>MfgaW?vv!ctlzGkWv0Y1!y-UMYF0dxzsXStf1|dnVT31Ym=FnsW!5q#
zB_wBn+P9d@dR6B$da9s9lT(OcBy!jA1c`+muOE_8q&bu|1Sr2mY0s;DEeeVzXv9Di
zsF%;fIdumxgtq`+?HSM+wrU|<&xI@y=@$wm89jZFDX(;idwj{sU~TK0q~n06_v|CT
zL(d)(VlJr;K7PXVnLFW{WIS@;5G4yICP?iP_2}mVriVFD^u__d3;~xTp@osSY1lx7
ze?nXakk$LADzqRc8Qn%}hm%kmlsLFfURptOhfHu<$2MJ)h^_x4kOF?&ClEava@OF7
zuPmY!`3jyE=4M?aTFBc?F~|j!Jr!`Q*_8nIyJQll&I}d)6;ty7KxP1okI7V>p!47}
zKG!4+s@MKp^fK)vWJW(kFK2DC>cvIw!1vK{KZewgeIxvAn@P0+yqWaFa~Y83+J5%~
zGSsb0(sR<M%ulX;|MO3vo)w@or8m|%jZPdanNtCGOr-nqi3{17HZ|j`06pDC2;6{&
z%Lek^u9#xV+Q3apnMF^Sba>Q$$L)LF5*Bwq`8**n_T=*?WLC;rUeoC{*GTN$L|e5f
zc*FQLW}Br!#;8@PxkF-eg5)!JBv23Jr$Fk1)pT064^&~}L9DHrnr<6@yxIA39DU{P
z=l}XpUQtWB0A@EF+;johDZL0@u{J=w0zt>!&-MT9C3!_12?7{(FSSKOj3&lT-X~sA
zW_Voa2aK3tp^nq*_bFowxGu{c!D_8v2VK8t+VMF4#;v!Z<XmKC3IjBPCCi^0Y09uk
zIu|~;>9z)%J&^aBqG7G!{uug}EHu!BNS^xIwdV|0HT#E$uDH}ZTyE0RlaUd~_TgQ1
z*%4U&@}!19P9L-%)-MsmrWBLuW4Bvqzy?<{mq(4Ob;lR%9g<zv6Cqczhy(xnpZ{#r
z#@&)tQIeINuaW8&Q>C{7B!r&155OlL+dKc|!%j#Fw4`U$bvC$Yafk0j7qwHJ`AB(p
z3`g@r?LWU)DP|!zBSRNE$`K|N+d<~iO5vsQ-@Y<8iU(3>GQKwD`hW$SuJfTI4|fN!
zLjN<slc}uyZr;`3+^0--ZgC0I(&w~~Z6zP`>-;1VP|K2E<A}FK6Fk6{i;48_*g4o)
z6LVV<L9CCr<7DPGJPj*f(c^SUnrP575_21w9)kbtJq}-gU1XMACT8Sz&aqS$Un>?P
z{*{2G=Y|`3O+3DOfud9o`L=LKg<`wr6yXbPaig|lF(Qq|uFZ@V95c1Q=qUYmuQU9@
zA@_c)#qXAQE{k!%9}_&2DdIU<7v15D$$MH<yXqtZL(PYWvwKB_ug4Up^r!1gPB+u#
zrGEh25rW51qpo8Lb4N#0BM6rtGtPROyB)e4v*3xYk<Voik?ot8_1FoX7^3UthMzjv
z_0cKZPMYCwjEG8i9dpo9g7y5pVdZTx6jw{OIl|Bfhm8q5&U-aTz6PtxyiA$Ok7$}F
z)YP@a(9;t{%0ab3$$_#&jlDg!zPi$cFC*FuNfnzLCNDZ`vTal8*E`RR`#x@<1J}|R
z$`{t+fbTqcxN@rQiuk1}w4{-*&chT+zf0o0LIwfcTlXF^oM9NwnpN-XY4pr>Ao$*A
z^v6|e^(Y{8(Ah)84=AmuHgJ}Pu;RzgVx_Z{wLZ<4cTA!R#S(k%@~A#^jGv}1rT*pl
zeO?83{PlH{=}3P+>nus)UM-)tD;Fj4^iUSO|LkAFovF7o07)nX)7SIIcDc5vH_X&a
z?7pS1-a*w8|K4>rDqpCcfxx(D0$Q6#pPTF0jQl*2gO?mEIC-Uk1F`IcM}9Ycb%%P}
zDROfV4&KuHmiNYK1ihETEt45WTxU=%kNsX6*10h`A$@~hL80C1^V{Xi9PqwRp`&!-
zm%YDV2-831H)TP1pq?QZ_CBt^DS$dcsp6eV0#AF6)vZGN_BJE@4tlz4z}{0|*0WZ6
z$V1dXyfVJPo^x8A6i2~xN|bnqbed#xa(279sgv5Zu@9|HAW_1Lo>R}wcc%Q2A_1}o
zGM4P9)9C09vR~C@$KUL9EJxT9J0zjgM0d+u*W;Q76Mbx~j;yVYq=j^>7aSA!uswdP
zCG?i;=r^Kg)N*4iFfol*R_@<1?e#}D2w&47^Q#=@{S#;4(sH_TJ~(QUBoI2~_PN4n
zw;;r!abed$C(;E+Bi5_WDAEXCNzP&heawP=Xx<c^xHsZaFDs-&q)yc5w7zB3q-<@d
zM6kp$9HTyAS$;o09PTX&F?Dc+(H2P@rJVNE-Q1s^aQv2>=52hH-yxGN&3~z)Okcg%
z7%X-q`l+^h>SA-^UVZ14cZu9V)<lR3Z~QiOsrT9K*gq8zo9&R1X9rB&dzW`KAgA~D
z$tzxmz4jR`>PFo?{;XWgE(Fz;>!sy@d+I%OAY(wz=F$hv<vi22G9sxfZq+a#R#dO|
zBv=hL)v2rW&P|x8vGn7+J4BviX19q#aMdG8nBuA*YJA47ip6<3J}nPXIH?#^erei=
z#6oQyNc!c#T78Cq9pNs*FUw`695e5B4LDvLvWIG3?&Sm{nTd9c46v09wp>tc@-n#V
z@#LYy*lZQSw3Hx-d5*9#jM1*{=B_O*M9ylbL?MYv#bWi7R2ND4zqkOQqeR}`VL3-B
zi3xol7q#CM5a0DZYe}fhgjxR?{Y=3Au7UWxj?PYr=s#anbQl<+&mimedOmG%;Ww+5
zZV}-KkUQar2n-WFpW)`NRD;)fooR1tD*MdpJ}`6NC&_b|zRZE111T5BupB1w*yn_*
zE$I!f=yV{0u@s%uxMnEQRmz#(zG0xfQ=*;J4JJ{vkB~U3rF;)l3Nwc)K6L*`s8eFH
z=YPQ}=hvam^$TmtMb&YvYU0V{$Np|q_F2_k;Qiy(z3bhQPj=Gv84NBsOa#HL?Vdio
zE5xD}8fq-xGQ%=k6j~+Cazsbn_$Jg@zQvwp_9k^H(3J56Q}j@H_$!igt@p$ROL(pp
zdbthf1bn69Ysd}}K^j_|GWA*=ogKa9;z-)nP}94f8kjHn<VeNgv|mN9n{wc45GhfJ
zPJ7Xt&?utxt{wyH41a>H&*U5TTRMBtA4d68-dG~a?Z}Uni1l+PH~>3UR!R1T(G|6o
z4z4RTJq*};W8FC}ebr9%-|L&e;H&G4!U+ayyMU{<pyrf4zIoT}c$L^2`vdCwy{OyQ
zkh@o;sAHU+x-=AyUwnvgU%ax~HC7-uzxz!(kpE!|YY?sV0z|qfityXcUZtd3FUncz
z{ju-RO<e0Rja{~(yc1mk6?Jku*rALIn7Gm}m5e*XSLpBbk!{L}$boX@+`;J*cgD42
z>+jRluDVu5DN6DzpQgGhhzxo%2{&&DFaJuZz^)r?liu*c?>x%5MzDAklA1@Fht$9M
zzwb9mBub!g!PKQh;^(|Ye*G5ZhhA|x_3MBVtwFGWY`xUru=IZuZC8}k^sjx0^2Dsh
z8=nd}Wo==lwm^|6aVmv4g_=6*wA$JD3h$*&mf7K+gFbieghT@bv`D|uBY!ELmrxg3
zS$#p%ESj%svqezG&ESq96LC*!Zwje*_aHfwTgGP(Wr=o97ddUlP$|4RRwESYWjT&>
zp8a{E?NSwHcC{n!6RBQ(T|JGbLWsQ(-gud&SSy4A;rMMufxVOFo?w_Mk(+l0YXM>8
z3edJ(a1K<j@WCgR>CqC7PFM;YjhLP~nrpV}!yJr~t!<8xGs77h&>vGxx?yusNvm%k
z5xgN2Cj_KOpFPfLIs=|iA>nTx(v|L!=)>#c6qXY&-x677Ku`FJVMnl}Xf57I$4k`o
z9v@c<Cm4>Qq+PhIGB8RXZTp*qAFnw%?0DtmIVsPNE2-JueS5RxCkqf$NFY&4NvtP_
zTDDxG!2p~U^)PcC%Y{|OTeS(G7pdQR0qOz4u-QCP-xN%3CyTZ-3&cKPH9P&iK|_4T
z?|Hvy`&!b7{$Fz}I+>&Y*uw81A)ej)P2aAz2Z%I3$Gj*UXJ~Z&uOsF=*_H%^_p&K<
z>5mI<{Z*3ZWRvbEy0tcgN6bXI|DI$&nB=}um^uBaklv3yIThlSFMJwU>>oOuUqPSP
zi^iVk9)$CKmXO#n4l{OiI|xSJ=3${cg`1eoEw)}*bni-$o4$t)h_$w|>hrYx>nShr
zlx5<c^7n67)fq2QQ>80JJA>~KLf*3P3k`(@(gLxt|F2?^%D86lcA!CBI%q-uCKN+V
z3TC&TIL4({36VHvpdOpWv}@cl?+(?yGlnkCDS)1yIW20lzD3iNn<`Ft`jwIlR4h^%
zbBHjN;{L|pvDAZHL2fCb-oh>q!g3S>(Y*z}$mcg)<R8`vvWTWS-5kM4(vBAODjUUu
zey6;~gd5ol0oM9O55wt55^Z3h1-@gb*WYQcMTBNiuWN&oK6{ma@1B*>WNwYx;GOS`
zO&g)&A}o=Sp@rz(?9gkGp)XlyrJAUMcqi(%x)b*zC{a3peg92zUr5mX=tPWS3%6~3
z-f}~e1hwnUUqas{9}g`vHDxr#h_<GY)a8<oq~=isvgR?f%u0Sg^_}bc)$iIS)FszT
zSY3!p)_uF~UiQB4mB&pe-3zCHSN7D(pM^z~bQ-n5zFX|iF^6u7z2lb28iy_54QlWC
ze#lA8-`!IysmV<EtT5^SxR^pSly_;X%+I#^(Rf=}O^+w)_g+U6t9SMsTVDc~JpVbJ
zLm+2N9X<TS&{M`|v*DkXNt#R(ecY|4$KtwFr>mQ41Jh6ETkE2Hnl=L&?`#eyYODPD
zcnn)FRdix~QS-g9P{*JgBGS`P;)zqrfm%%wP0T^YL^HiT##?!5Sa7gU?UL>g<49MC
z9H<=R_ju@^UGj7;I??6xX{QfZg!tF)SG)<GH*l;N(e6&*Uu#_=)O1NYmD9fVMfE%5
z_Cju0wYch!8YH&9cQ)<5zgo67>guTtG}e|S9GX(>onb5);mKvQ6-fL$M#@YR_$#zy
zQ<64i`D>+Oe82r20kLhrdQ<S%JD1P@9Z|&$17Drc{b9@H{41PbQ+lTUzk;{_e_;O+
z*8jJx*om~i9KK9l>T6?T^YyBM!yq<~P{f2Wg}S9LN=r+ldkjZ&@2FagG`$UtYZ7pg
zE5@?)#z@*j<<;$P<{CK64VjeoD!=?%$+yipQ7NqJusLK>QmA~`S;<Rrb)ruAngceq
zR6-yrF-%J9BM>y<8!ToL_pF?zur-gC?@m<x*C(C@z#Wxewu<sooSnc=iu^xt0N!-;
z7@_a~bu3PI(EJ57{D1veWZJ&{=6#|TlIFs}byhK%dpB7=p*`NQDGjZ-{Q@z!Egg%Q
z^LL4+Ln!1QpKSGdtupxdrW{SJtPo4(>i?nY&Eui`!uN4w>}#l$HDpT>W6PdMk$p?n
zEEBSo?0b}vtt@Hm$-XZcF_@`DF_!F%F;ha;nPLpa@O^rJKHu-__xsBq<C!_<KG%KU
z*L~gRJSW5%@D63<e}N?bDQs3%ycjIRUC7dXRUEPXsQsVon7VDs1Am6O`P?ZA^rNY;
zFWQZ-%jF-Fy>5>i{D<U=<XjkRcgS{Klc(i$xOxx+N!^XQUQ-vJ2d<jmt2p+=%ae7n
zos+I<xQdTnCGxk&k6JTkX`^UMctmW|x>OCoaEqs!1i=#MEqKI3?z)RLsoghcZ2kpV
zh|Mrd3^>RK8hb_jxl(Zb3^{A^JdYy7CbJf+I>&5Pq%X$@_Pg|jtU?9wJaTzfmWDi#
z^i&yC>CV!rfz_ZZ>&13=tc=cm23%s+S>R6pwpCo6J0*{fAG&f90dZRA&Q8Oj*t#!5
z0@rR!N`@F1EP!W))!EdmHM-TjTGzDWcj8aLvH5UR<Dy5L!tjj@<ipCdyMP#{Tw>UV
zq9Zb^f<|7zCX{y0YI@e_8uZ1;w4pQUM3^qu7T;sNT+UGma-u&9nU<5Oc~?6}hyt54
za#dhh@a%W5pSPg;8yub9=kc=euK@dwA9-*nkflAse45W8O%mU}!0EDFX6!x9qdSVx
z^cx*NQhhc{V<PN1G?kWTz+k5FpX*Ab=>$!~zsXN6dDNA>g6%5?X$66Tj0`vc!nQei
zcl?g}cB5I4A#hE~bC^2IzD)Z0vUg~IG*W2?%vFAS#}OccLiAea!%sIykSA-`TSV?s
z6-h5J-QITp4Hwid;q7-{^^qRXI-xIHo2|%ks95z}32}j$1eYjwPnP^!PxnoE&gW-a
zG||iGKk$R)#YyAwI{?TM-`;z40LnQtRl;z&uJ|4!2l(5(VwCUzH)u$fZ`>sZ9%kdE
z5g<KmnSrd!2H;mNTWuGmE)wn&;+@3nr%*4VK%X&)(bbTVAJ*k*5hGa$wA~16EEcGV
zJ!>C#!nxmpIVxF?xj~1B@QJON-=@jX4teUeOPSI?zCV1Fa}yuO9wZ;c5o8L@m_IQI
z&iwz47B<TfE*<0-;_be9>`bVDM^yOV&DZ@m8D5@I`eV-74h?1flwmdRo!5%}((M_R
z?ngxWl5ZH|8y4wdZD7EH{{~F11~0*8g~KIDeuIW&-@(nXa>ZajNy*Y!!=`mPpGa|c
zS3*k8+*_(`GT(=Se!Y1a(h4RzY=XKlZIE9j{!@maf=;ZxVJ&PS$at0x#h_CHmkzC|
z&8qx*&SgEYe|wyu7@BuzZ@t;ePr5|#HjWa0`-rdG%-aFftqai|t2;E#7x)JjgT6;-
zY^|;|f~-iWQnqu(@y(u{eXpTvobn^Pf3QIfWIl{3u~{w{yF`Auw7sqhPzxX7e2t%`
z31Q$-kg8`i^Br=Ow)(L>Ap#yz$v;27z|W%z%B-mW{rFTx`i;&HB9$D4;6jdW4T!xB
z(eWi*qRq+qHo)?m-tBIUMe1!CP5r5hcTNzF_xE;!&8YP!=P&M<qr;STxE|yg_xuC3
z0x-_204Qj6vZ45_R^E;8IeR^gv$O{}R|ZBqXvV>LEl6c`@v@9|yfHkd^-VkWa7M;4
z)0NSw1TI%c*#>FlEBS1>Og-G_jmc4}S>{GNLUuOD2X~;k(2}L?d`^HHjYrx*KAJOd
zRdY=hZ3ImWt;a{gl&p^&(G!qhjN85Zg4%{bu1bz_ZGqX^NZ&J(=tldH14&X7{9tH3
z<}>=u*1i0=eEC*a?rzZRn|9_&Gmy@mxxvlpN>F=4!VMY7oekU4_0%0;zQ%+l_`&hY
z&T#*OY~^AAS^vc~55NyHV6%ezmz(24yklCZTRC@6R1l6n+rTn1Sp!0IG~;Z>c;F1H
zKfm?irl?ssEeM)1oo!VcPH4|zeK7VW%}#p2nWb9-g1WC%IH?KX4Eqmfs^adTpmTs{
z>k1T+PDuv!vhQn?^o*Ii<NG(WHe?~UP`~Pfd;(HnhUgm!(iRBFd_$yfPV3h7p4y?1
zoY14?^%xLG#G}i@fjljuB-b3<{Cws~d5}v19A82@Oac22#V^I@*%6D>OGCB2<ldW~
z2j6NN1u&=pV9*T$V9?RvGD9&(chUHOc^b~1J`WsaJIeXp8-;oo&0V-M4E%VtI|VE|
z^DFs2V1zVZFEWvyYU#cm`FB<!KF_wUnpsk}#vdchBM5S35MjPtGw8GE@WYqz5guXM
zp_5uZEfhB6cmCdq$wePGVYY68*@AvLhv2G1VS)piF()J3b|Sa{bi&KyXaio2#Ty6f
zo$=It-s(OicTV?>s)5JVEoU^sG|`8>HR!5o{(!i<2SHmoAZdnn$cRQws(mV;Zgt_*
zhEMktw+??#jxUhG3zNfc&=19P8})&^&*epa0w&A%zmsM9#3zI0#wXuwjoh*Ogr{?%
zcQSVPQ4>-32N(aqu&;7zeK?`KI0g0vtxGT32V|6UrZnP&jHcKm-@sX`Q0x7>&|+?j
zc5;Yh3zyeQ5gIGA6gGD2rHQfr5dtkO;qJq)lTjWLyx`pSDNZPBQr|#`DY{50O4*y^
zjY9qLF+FJza|9-}uKTmg7KLj6MsNpR&H?Q@)c2OFDg`Ggg(I${U+CgOB(}DHw1Y=D
zow=pp9a&NyQR}0(#X{gA-yfu)S-ecXyG}L*&IH=44FEY)g``Lbp0fUI8)3mTsY<7F
zjwdQ~H}TdzX^-$4)W1~oFEMiP{IMB)@sK)k%L0B5iMj+CExY($??9w<Erh%3;>eE|
z^;UsiE5>eo#`GIk8$2lfK)5^}pF9z{Dbe3pR0wx1(0A~dS~5i6MjP6~`8uxX3WeWS
z`NJvQb_=)q0SV3i$33sB1n-KW`*R54NxVYb;GqP2&y^jeoqS)!?@<|Sp&@x*!_9>a
zkUO{Sb2Zb-I|)vNohf3flvdayXd?{9Su5xQ<gJ)eV6v<^$K4UR$wcVN>U69<wsjrU
zE|)NBfiUvT0IeS%yAft#ANbWvr9uh-2kig`3j!RJjQSV)c_r8U-8W-7Nk1`1d9@{?
z^u4S{@i&!Ba5p=cGtyv1b&q>iWyyNhs0b#~);cAng?b(y;cswR32+4y6|v2hsgfHx
zu`wL`0pz1{kB!QiVsLcLVNXQV-$Ue51zO_cx64)?*R~tyRt$?SnpzAC(KJZb{ecd<
z{D2Djtr_Z8&z0fope^ry6#4*Z@pMfETX1{50pkP^0!@^Lvhlm&UPbXe3<l{Cwf9?N
zihq_|+L^dIzQD_awP={s_t4`kUCer-p;mXo5M6#4BbMHfz1~^^(yExBo&>4Xq^A7<
zC>P*}QXXM7D6f@t^h`WnNhkD#kHkfgvj_-yumZ@OwHY_w(WW0ACMG<vTkdMjo&<-W
z;P5<4cgj$0RX+9?%BWA?HmcZXb32L8Ct$8Avfkd^<<=Tj0q%R<6fYfSk+l@(qHC)y
zU)o291rHeX(LNFxYYP5(a_~#Ltb0^GFV<a4H^m@Lk`5sW(|vvreHWTx2llm}nEhET
z!eVZ`dbAG<*9s55lYc{=uGiS7iCxA&)Rw%!>_!9a_U$b^Nzk4?y4#~#=paR20L#+;
zuCEiHO)UjhIl4`4#~G_23;%$9J<zXn1bvwk0s3_L>hrwAY#@<@1FNkwHHnD5KIhw=
zfeS3_+m?o|_+Raf8^x4>ORA6SJETqBf(f^I{&3qK0ECcRxqV9%RAo88dNuu5ZbQ79
zo{>AnFVcT@yJFGXH;Pg#!mq4NGa~e#H{+#(gH)IEOQ%#0m*PK5xJyWv-FE(m@AMOy
zi#5-D4f7zfj%R7C;ReuH4T#Y_4$m3>?|~J5F*c*hb%`tf@%)B;Wf-C03r_1jzN)L-
zb)w!FUSW@Dx*oTmQ4;Re^T*!xhfzRBN0cWs6}BY)6Hv{qr<T~oe}0_0Rs>i=Z{iv0
zbq>Ns8htV7Ob|Q9o4ss9R_;baIY+hW<<;xxHgR`r=E?GHrT;aC`aeIN$$1&dE*YT9
zy)R7qiqWd+I>5>A9n&>G<z-GA_x>COHnMkw3m{Fj$0iYhrS*f60TUA;uE!XFG~Nr8
zUJwt(UlvDXz>Ez0h9rK@!ouwhg!p*O(_y%_YS%z)9@KIoY0SBHKv5uO2e?Y3@{F}T
zYjc>D6hbOJclPZs^{H=9%t2iFBiAmP76aMZ^dTq7IBR-eE=cHV4jfgJccbLo6MNxJ
zKyCxMn9dz~!}`A*TqRi$N)q@Uv>azmZ&A>)Xb`$m86hi7uftJj$4m%82KsIo^dZnY
z2~<#&1=BG0nyz>(L$n6uWW0g<qL05R*cWMexH6?201ISZV0_3lP^r~y7h`_J`T;ci
zx_{66`hklBi`EOJ>b--tINsaYp8rd}>41m`UezL`qXE6y5;@!R8g6gU7ZuNKHScfn
zGa)_xM#ATK>jxDpS<~Z_x+R>^%pVk*lg0;_n#D*`gzq8ahZ^pbRr)A;Ky+K5MIi*v
zrNc_-PS~RnF2fGvo8HDcNJONcnFT6^ccKLHd7{u5jzKV{H+*}`GoX7aT;HHD4=xA&
zKwtmASpq0|Ytm~cdSoDR)--KI#`X8F*8iL)tmbfqhfJ8LG8XgmV7;JSce*B{rY@Q2
zf{v>+PUi@E)+OM?f4bH!eI|4KS+ATagN3tKG>9?_-R|TmR)*#@lsVbVv>8wsUDyj<
z8g}0^6iUhuKd%f8bs*mjG|^yNVB6$|+?!%0a1hJ{)V{B`t{=xT6Tb1ZEG22&UcU-t
z4&8qw|09q&F2g$o)a4)wwZ{R1NVRLM`<#-YqCOhEa-=laT__NAZ%*x%-2>h|jCPo^
z9!^=!d;tT_SSH`=+a3VvGA?o)@kyqr`0UPwZ0g)`NdME4^J`Dkpzl?P^CqfdyUxy_
zF?-xC8r7{WdGLOj4U;X4^d0z|lcF%TFV8Aa*GjLIG2$eB>mKs}PQ~$P;6jZ;NNwB0
zd#vtCtM&Gsv4-Riuzto+i-6ZXLAA@q_Z*qA`?+Yf_#q#s{pmy}Vm*RSs3xEgKr{0(
zNB6VGVlExjskDyfOI3@t%`S?*MuRUbMN?o`I7ceBA9lT5d>>hl-MHG!+I;3tT)d|E
z^Sa=pZVj8!%MC+y%#$J<TCw^?7gZGh$M?#=0U2W@lWUL2M_nfO9NQ7Lrb<FuqlzZ?
zEC@bR8-Qo}yz#*{Qin0!bo;7aCO?2wvmx%jgc0(8fhz$yfx0<oQy%+srQa=$4HQOl
z1Ht+5034q!{*#mE(debnlE;4wmA#?kD+P;Dq2v!Q|7`s|cNd}<GFYoMrx_nLlPuL;
zek?M3Cj5zk-;46sKx~fwNvJFDRc>i3NT17_O8fp6qs6`N9ul6^&$xFJZOPN(YAdZ!
zR^hX5AQ%`q+7#bpk5VY@Gpdtu6z)MrZe{R15)V@HqJFeho~N*+gWg9DeU6Mu>MH)M
z%(SUm?)NZKziw3A#~&hOQHMQToAS^uwHrO9%Uc<j*yVE}VBpYVXU&JB+MiR1mq*id
zRcrjh|Bcyq0b_g|OP%fc){zishf<Rs@Mk{qFxa%5`W96>emX=w^aMnDw@zVHcHh>P
zapdIV(KPD2jn0#$5ibewnb<*M#~p-`nd9En5Pd33vmz;yP&D-bZU-3X*8dntf<*0N
z5!j-A=Xs=OyPY#IA?|+Nz|ST(SNV|IQaR97q?_B?_O$%$KP&N?$-Gv^8}{iB4%bM?
zf=7^GHE@RoG{Ko+!Gim-y)e{1K*v^2Gwo*I3VJFh_4lh<mm4|kL~Es;66B-U8qwjK
zf|)Q{LMGh_dS8{D#6S>3qd+D;3rElU-RAKLqSj19jrA@laP#>k=Cyq0F|fMGz5%qc
z8sV&THCyVw##TiaD@}5c?Fu<<wEX?hfagl)dTjKIWuX6K%kxNz#3GRB8&c3TB*<h(
zr|=aMXu2$9rTf1TzdQvn<w27%3+}A?)A5dNRwwp_>uc6=CGfg@7d$yuNqmoS<n1ml
zL?3%favPwly%hzkd50hU-fzLe*+?0+n>+%nHEMK5;AA?5w-?)#2eu`0_F_0I8Ui^k
zSJmhDYI}V;uRI?YtZzoAK-X&#=zqB;%3nwPwUtzv_M3634QK8wrQ$tvNcGMLx>GsS
z?*rRDqZj6FuUWM{G|mLt8>2D9YNh)KxIGp3l#DXSTB(H1K7E|M&YF|);XuDV0y>?G
zc2GXeJoJC}0`UCl#Avl`!;p%lb|HPA;So0sx>>``t(~*Rujy)t7iYoon^3n4!(nX;
z{R;HGY?~%?^KLo)>DvnQj><(HE<7u|gD)&}F_@FW*<uNI!NF%#!FZZ1|2X*>Sw@m8
ziYLg+Ao@G|!<sjnhM^N1(%3xtDkqs4wVHx#8z4MYuMbq67PV3P)Iq;(Em}Jg%tZ3*
zZ#vt;NRlBq_R|U2^pmXO08eKExW^>E_L!!K=1ZuuJRGo-ujlR4&4%Y;`aLr-TDsyo
zY~3$zmPJ|dIv1bgsd}*DfN=HkZ~KlYIQ$Jsn2?DHEDG0Q@Z&1Hb=&yapB8EbIs$!S
z73y*K4LnRGw}(0C0;zJH_T46+h=U3xjPg9H<Qr4|-8HvT7rVCyQ-Ai@6eomYCcS~F
z*LHE0m_1#akd29QUgXRnJ_Cv~v3JU$4&jyt(H9`Z$A4>>UIAJFf8*CnK8yc~nfp1Q
zl|<RPdcCE7TkX2G;ueR|J^@A@GdNJ&2Qan_FW>B2Mw{8<@;jup!21K3?$2)b1e)yq
zwZucK{EQAg1;DOXW3Ot-fR4lJO4^imGS+FIO+gVmOG<<Ff0xe1-RN$lWF{;r?W|fK
za!ZooH7=guk|7MFU=fcPp5aIF{P81P6lnXj(V863@3m4@)!9119&bJ!XJr=|B}IyG
zv+r-(d?MZK3|(~RD826fzsdSPQK4ZSU&3(2N4hm=3>*86x1YFmtBy($8#L}e0Yytf
zntq2OQ(*qGaw?0df8;b97|fVlUcu%3QC8pU5W+m%Rejsc06h-e#)J0-B?C@$NZFOM
zJsF(qZ?u&l67J@NFFE@?^2`D))Ho}E9ZNad*OXvx7L;5q-cy_2&ITQ`XsJU23*2d<
z9EEIt)>Dt(WT9i}NsI*2dW8e6zI?NqG*we*lur~2vq$|;mAegtMbV#dMZ<Wb6&USo
zbwiAt0vYc%X*h?)e@4s26Ak+K5+rTm!Zi7M`O=MGRBH{_uo`NQxfuU#$Cy?+;D2PE
z122?22kawX?*<()h3k?K7_A1b>bla%(}SC>teG<pR{o$hWgW$)$PGiWGNf6cf%+}d
zI7U+ZtWn=9^n-*<OXxG^W&N|p)OUxPvXF*Mreig69U-r%ALQHd7_IX_Ho-oh);Yz4
zHD9{9uY=z3!CYY6KkKy&6Sa$*;r~@9>sTQS$ql}>Qv?)P|0#k06<C*m({;*+l=Sm>
z)XZ&#+dc}zDISh=17Lg_jFum8H*DF(6Y9d*U3b`aE!6c+XG)xb+KAk+k#+zUMb~TU
z(9)BEd%H^v*$k2==@@Sw(^y-m=ak*|?m=Dh;R@hhn_OX2uNwzk>N10Df5-LU4XUo`
zRa0`^y*)bG>EThQXF2zu<P`Bt+Y^7R|5-$v$lk86tE~#8i+u@KvkrCYOufmwams7p
zHO_xW;3A-~hyoh3&myoK76^fB8A|US<|rY3Ey-<J<DeNx&-#|Z)BYEi>tsRutY|?{
zSBq8zG{Ytn&tz8h1YHNj^Ld8w0E11Nki}ThhXit0x_uK>mV}>rtML*D6jV)u3jE*?
zrW;Wk?CB+3QWt<PDX%Bdr>8!%9DLTd6t699Z4O{*+Ay>L)V?w$YG+S8@ifr#<|Q{W
zPa+sxST-%DA~-$>?-MajtVj8+S1y^B+~){)Gq@aLSHpC@iAt#f@~2eOj6ZhPIX|X<
zyMKE<j;G}s$;bE;CA1!g!khXKfwCvCmq(jv6CAB5?!tjf!4*WOWL>lmly$+LSHC2V
zPz;?Z-SEOTx|4VaQE2o05liXD%XG2q-Kr4XSMV7X;85h|Vvddd06Ve#;-FRT`5a;*
zrwler!O>N_B323O-z)p=s4Zt>H*(p?JFP}LQR(~^h;LnE;V7(q{V!L}?cmWyE`{2`
zqcf!B_1QS3SpA1RwVOKdP7U$=u(X$lR!svdm7gAknHs7ei~HzQLtTm(gzgD<6#z?6
zgiXmY3zQ7H=9QWy4We2?A5IAe1QP_%j`7@6jyX?0jF45ODU9-sTx!hBVW5()n3*Ss
z3**s{y!mhgtrw49bu*_o*z|H~DL{y|E)eA+J)Oa`)zWmy&3k*76h<)SonB9w3`1Vg
z90)a2Z-G_W__<EZog$JsADxQ}?mmtZ^V(=>@><!Dj@Uk7Djv5n(|#y8SbFHFv{PR>
zqfo8zz90i9xG&>nvINy-<{HHaO1MW6l#N?WXhn)PMN27YXG9pet#cA({&u9FG#-hm
zK=^OU4|92hd4!Rhx)f4jxcE`aL;t(TI9=BL^O7RHyy5Cz>=Qi+QDZRp#td^1v*fj{
zZ0BQ+%y)i<TottGx}ht~iv@7et(Aj{PH!fE>iwMAm>7R*2>c*Lc6#j`p#j}fz!i5s
zJQmv6=mNpb(oO-gL*ltOPHs{;=Ex<cM|Y~V;=t|f;Z6wRplj=9-y)h(dERyWmJ{>7
z9O*tGF1{|GccP<u{XolD30JmG4zXJN0<>xl1}aA**(4o$4kBq#W2KGUY+9i7uu-s&
zUjSQ3=z?>$yLerBgqXMsV|ThzB5;uDsS^6=FQ&WuOVo22NE4n}>v^m2UEKL;uX_4L
zP_z;x@Xfa2UyXHV=50~1i@rzh3k{ilcG3by3!CcA+1feQUmzb#1a;Tfa_4L@x)#9s
z+q=Y6zl?U$g>#W)1?rG#wy2ZK5K$c*Fr9LH{y9_rK_tB{U1*u&Q1=0H_2yu$d<pY0
zkUZ^)_~(xkkFTKx??&luDG?liw$nt|AS*gmM(uFJ-xw4+gc-UMZ>d|rPNOPs_;V;U
ze0K820*c_KVkmPOPK8eV=Wq_fO8VW5|0WfUg06U7S!~N`y@yRs0Cdho4S3S}BNa*6
z<;i*k8}X**3z!0Uw5-MUF|8>0tfWq7^TZ72c=XjoL`)n(;z_B$8o3Q?#p_UF?(?pS
zzNv|KqVfz<Ew|@%IHwnR|Gsgz1+kih!)Qg<dYSB4Do}SS|GZbTLU7~L`i+~e>bxbh
z5r{d?JfW4*(pqmG)9BK3FK&$f>8g$O$aw~bh!?X*Kkxt5S$c6*M;kH<$HXc4^yagB
z1W(j=EN$OY0ObGS2%bk5mZ6ip0%{pIWgTg>C}^YKREjERc;KD6R)Vff(X})bDCA6#
zJG8fm>*0D@9=*La!uetGc~$+nPn+<m4dGKWXm~1+{r4&snisjkouDqt*`w-jH+$kU
z!I`Bk<l8+lpRWTub<K)Q^6Q{MuX@<MLY|SGJ>$ym+15;j8EK<wC-L?m<V39AmIgs!
z@azMu?I=zZkmh55#?Lz@P%P$mEXcy-TZA~Di)c@BuZPl(VD@pQ)G#EtC9Xstm;R$d
zb;TX48k*NNG>e;crYVRQ>&?AJE^bGqjjQVDJCMW6JJVI$hcyBg&xcEqLIqMT-u5WK
z<3XY$n9-0wkWI#;tiLf6LP$JY^d;#|w)d0f!1raqKFiF|zeDrDBW@?_1oT|mh@+!a
zv}9@KBy|EJ5>Ycm-ba2tq#*Fp1F*6C!BB!%PHbn!+{WV#CFy&%+SA!%$7bgr?mtf5
z5$1{p8VUV~H9BU`6$RlYYgB%!N3eA@5?F+$xBrN5q=fi-{LUO8a3}F5$(PoXf~=-^
zjOySQ$Q<dI?kf;oaQACo&l_u)jnW8BLI}^F{q=5PYCydPu&q2o(~D^#KhubHnxBr@
zs1K5#VFJFN(eWo$R&4=b34%Xzhh!ZKR0Qd_84yHLXZf|NuCK9j&G$oOWa~6y&V~Cz
zfosN#ZbFY)!{xnhQgF(4x7h~6zKRp>4&E`FhFI^W<_NA1`P5HZ1X|}$jY4zgevg7g
z7`5^rOGK*x{Yu4jcIxVb6|7RQ?OF1j?*P+~vYZNj^S3C>h>ZXIP>Gy-Z4qVuLb3Zj
z;@To29akZ*7^)U(e^Ph2|ITMYy^%^WiC3_@t*~@RIfOorxePfegcBJS&3!zs>a0a8
z5L9^nkk$(uc<rGVA#u#n%?Ix+d*xI?BK$m=0f*gI_>ITdMh{9)yuEp)zgl~bsP+81
zV$mJ<hNGuh-CRrdNv;OBpJX3SJN+K5`}6*5>y1wXvT_$^Zr6vPgiJ%~JPvsTX3nS9
zmQ3Xs^u#gtx)cRgueH%1y^q|wn$B89NAWq1wnDvG@(9lU`2i{QIf4Inu$>CKAFCJP
zyg-5sx}ff34VkITv}oft`caNvsMykjYCnc(adCvxDEi>Z<n@oN=!zV!4_t$9F*h|;
z#Z?<Lh_CU(T&*>D%(7}Rye|Iaos&%19hC_`NiRo;cl08|_U+CVXq3HcexKpqDs)L4
zdAOZ|aVmnlf?V!6bk_8?xNw&?Jk;|S^Lkl-+!%f)$hd9?#BbG`SiR^vdI?`un~4>u
zZNe-c$GiaaU#Z|ue?g>4(SuDFv&<GG(`QxmTV*>FGh@fd5a=<F)D&vQ|F8$mEaP7l
zkR5`YE4uVuxvC{*|6)wxUt!yU?RM>d&5rTIsEVnAt%<1zU2_N{pZ8nZZw#;j#&3{C
zSvh-cYthvqlOZw?VwqpbqF)+spr2*WNN8j1)SBu^>^x;vw06+3Zdk3ITmac=Lr)}#
z+L&`&_v)?_UuuN9T4+zp-8fEbmFO#ow}u3V*g}hf94X81&=DbI*+;tQ_NmpDo;qMY
zEuFHD7o0adf!xx$oq!%m!~42j3)=Q%_h$1WVlaW0OrviV!G5su>qNS4$-&3#!q$vu
z$o22@%+)D(Lylu)>nz?2oo-^<YCDXngMH{IYCG&J^j`wD8Sq2>rk9;r7ryLIdD9^p
zMz-!)l8fY>JyEar_~%Oe!d~H`Gca0b!r3eW&6O8yM!J|LM=I;hLWD@^m%e*B>56#O
zz0S=!BUakN#*pLx%4zC0&39yvXswR=5B1o8s2@=ZRY<-XI^!08)%GcWdhW|(`lHa+
zla+$3nLq{|&hDm~?k)_E8w#f08*J-ZCn6_Pr*`8&q}qdGh?bDSi(Ww?0iBGpX{TM`
zp!5J63U$G+o$IPG2gJj`$&mV`q>6ts(o~ZMZpFMjrXX$%xP49gI2Vp*C}um7ko<n6
zD&Pp4$({(PyrV~DCD9WM(aB0X3!$t}Y#n<GvZ~FjaW`e_TCj^#O)HzT3#UhjACS}E
z{=pshxpsiZc%c0y@0g1=jvlh}4s9=keWnryv!MU=S=`;Spu?O}C^5CPhVJlTpbJUv
zQuMOh7Xf$`i*0hRWzLnCi|pNk9f~$_zf)Cpo^?&}kC@0fLZ;Q`*5Qpa@oirxXEb77
z(fsFKJp)H0<y#95=Uly3g5$vn*|PZS63V@r&*KH-9pfhwJY?#w$0$SMoQK(*>u}H&
z?~v{Yy@o}giC__ID@}ga`x_4=gZ5oRY_UELNY=m8aOpUeb6m9|8^<x{?iPDVw#%Lr
zjAD?Go#C9V*bRhBTqfO7l!1`?j&BE9g>Xk8+kAazn?>jh5B77U2ZaJADn&T&IHB6N
zLh{EtX6lrt$%2&^M{m>o%-Rns8QQ*La;K8sSAnR2LNfa9juQ0&;Q)Od9uWkMRfQ;w
zw0?T*$kWn%Sg$sfER%irxvZR)F|Agie+5vMx-Q@S*X7)^t-CN;r-a@o)}=y%%hyyU
zWGUJ?aG%V#Es>vZg%sz}p_pSKpEJL2FElZ5Rbd!JD-?rJjv?E5DxAAB3CFSjju-XX
zzFIZtOL0J0=B(t#`YmPG<1MCoiO!<H#OL14BtQ(OX7UwUlh?)!L#NDjp8YLS8r)Ca
zY?JtjawO*sgMFv)!1GNB!?M`CO^_d}lll46z?N|ZZ|h7uxvt9?%oNX~;22mx1@~;?
zpnKeDI+KR{SL*pkWsxC3%>elq&_y4vC#yAZsZfagcfyBl2AwT%Z(^^IB8;&@Wv_&K
zPla0>Xh{#T)e7gr!;Jc{@riZ8Q)E%Vnp)s+JRp=y({tA@?%W_un$1|QSBi9O0WFVC
zoHsVR#@#kgxE8gMgp6z-Sk|u2o(u`@8CDy>**;!ho*FfuFI1BXPIvyYRiLQz$e-iG
z*$)A&%eQt6bu)&6)>~+ysf>@A@49DLgUE)1BpnhMT+YyNE93la^cS+VaxWh@1JLRl
zu5#H4DM7BkJkf_waV|rq)IO`yab?18Y)}mP<az$M6P$BWpT+nbg{hzY5GcJNGyEDb
zwDo&e^4*S;!!iP;2e=u<na8rpX*&9q^9j{8_?33zWBiRd`O^ZBz74@`1<>9-`)dw0
zv7*D64W*qdVABfb$OalK0Woh1OfnB$ZuiymXPu(l^Ga)F>*LI(=8PYG|9^IV0ifC9
zf6$zNxUI3aancsH@(5+?aT4cB2^#TvtvvebqM^RNviU_DZsX@AUr)X8{F<1Pqvd()
zdAL@VSkfKI8$6s=Im#(g#yjEbg;5`3<D4EA^$>~Lamm!<Sjus?YZL9zEUKw}FJ|O{
z>&(54douzR%|t=fLP){m;LwWHl!mV-wwH8unM0*Fsf2yrdjq&pX0t0(1+*7a2{D*W
z_`yOJ@!0ZH82-8VK3N9hSi$Fiy388$3W4hDn_o=i@{P*hTaH?|)U%>`{?+~M*PQ{`
zr-HBZd&lDKl<{w9(XT&NVZn909xM?;KW-B4Gq${_wfxd@EghNq&Ei7csWLg)YmI7^
zGoTj_rom9e;T`5OX&FsdN-%*#Jlk0J{>$WvK;~0&@3E$#g}QniPgiY!%%K}Oh1{VJ
z;d4Xr9`W2DBwcEJ#c6E5Kge%p6Mr3cQ2t4vq}5r(@mzp`XIQ*@DpR~@7j-yzPD*U$
z#kEaO*W5UFKj|pEoyJ`H=sO3lB!5HC=YwEh8oHr;3-+v<_Y8vg_lIFaO*5j(FE2XG
z(*756D2$(s#vED^ZjV^a8om9JJY0{0my(@ABclX@RDZCqwzkH^X*T;nmPcP^@;RIn
zjG<gfpbh&CGvQ<G+3?yOj|9bk+rf(OXS}C{te9i@+K0wBrjuSn=3Ebok$9Cne_r>-
z-}F((T=&MJWm#fWw+7)dVO3v^;W5Yf3QUu}%?q~7VV?!(v>OTa`mo0GUmtW-d$_F6
zwN`;l7H|cM1{L+3+#uZxg~U4}^{rxxLA{}T!zH!b-`@*=KGkA;F}p+i`~{79S82Nh
z)YH4NI`mq`<u<OjILbtSZC82Q!hajq*8W+W_kyw~NW`Bxw*gaN>ScwDo|Q0AP&6vq
zY)c?seI!lQyZ75u&Eug7*f)w^ZnbQDgNUQ)tU24H*UpC>kjl1gY_SvVCcVgNxxct&
zWNIW~``tH`vuunlB4jkx8Y=p}c*rbOe;Z`REPsWzP_jPHp<bHk;BaEb(5An?6>)v=
z1^U<kF<TQ&ag4V|QqLE8mz*jpTc0m3cb*H2(=@E46sBLJKz<=pjg^e@Q3c!MN6qjC
z6Nc7wtIK-e{Y9tt8?esuT906XFn+$bdUj$4_ErMQqmJpz{+Q<pw()uIYidy-ytrMh
z*Nwt-Y%CfwkL$eRKR`mEz504MbIlxBXW>D|VrfC5Ewu~;?=U<XXU-hbYZz$dm9FOQ
z(Q$K*HQeD?R+xsCL?4VTloy=BjR=Lbik2hk-JV-Z)%5Jxp9cHt-dVjxEVHti>m5qw
z52+lS4>0IOp8AX1vHu#zf2Jy@Xp4KTy@cdy1alkJz9)++hvRuQy3}%A5yU|H;z?ue
z36hsMHCWfMSqsgSv{>~Qr}l&r7#|(>-Eq<Yi&n?eDo)d~bUuX<!6S^Z%&p61`IC2c
zXlo%6gv35B-UAn-gpy8bMRwi=1ESUPf6jLE^w&#Czm1I(SGf6n`r1C8zrgV~d)UuX
zPI&bF3k>TjmAbakz5MCwtAro9N$>D>#&Q#jg)Td*(YiNZvBg*(9tpcShv{nMe&=3~
zPF*Zt7f{G1PGR=w%^0W4W-G_yfzw2lxdHk%S(n0E1;0`9H+QM=Rh1<-Q?9q;c@lr?
z+ho*Y|3ReD#u@AN3&}2q*QQ;>JKGdng(8etPE@ee*iU)5Qn6cp(-)~aY)5b2vm`Zd
z(P~Bp=#$HXv8t{<DLL1-6W2D5E@u<vFPY}H<8FNm!wagXkgNxLsmthBi+$6}`nzl)
zW6|4uIosrR1G}3rs_>nyYi-fJr)<mC{{w~B|C+&UWvWj#wsYs2{<&XOmVVh|rz<X>
z>Ht5e==yv1M(448dDr~U=)A(Mw>fL~83`q)=bRtYOj2xqCmKAU9&8^QhoP=?yqjyd
zko21g@S8iG@BCV}e%|kU?e&n)Z11XJXkH-w%#+6!=CFgx^`cED-L%e`8Tux3qESlJ
z7IW;&BRU@!z#{+8Y0e*P>!%s^B7tDTZM;XMe?fPOh1&a)<YnY|aN^oZoT1YN1-xx0
zb}n?hnd`4<JWesmzjkntrDW8UzutM~V7p8Ta6LF{G$o9OeQesTn|N#+hPoc=WCPdm
z)|m-}J}j6Uh#%$gIQ6+NCWEddl+yGXD`${fy*^vmkNaZq^%0-`6G@-aZ&}vSB1`GN
z0{UKq-v2z6mv&r%HlLs#+nAd<#D)jh)sf4o0?KWxJa7>mc0Z>}aXz(ipdjv`W@O;f
zHh+#?)<bgj?}*w;;GUczyD0b^Jp>EdKSeTgb1u7~_$K2Enu?=^hl%DQ40CIIr8OZR
z<L8ayNBQ6v{>BTSXBKGXlV<-JW)6UgK7;K<XpIhFnL7aTfhmgQJk>g8J1S;XzRPyY
z=U!sPgF%IO_6chD)}Qg+gP0mqQ*#}20y3~<P+Q=xLXp%zbk+JGZSV%Y)bHr~==C%T
zE;ERdJ)tuGzWTDt_oGty7DKd<x_Xh4!LBJ2aQ-|{`MovF*s}dUkE!`DJw24YhF?AP
z{8Hcvr2iHIFbUHLkujFl$>Fe5cM6#a_s`~7t1m9_k|@;CJs-|YYT4~F)Cjp<`|$eV
z5v$u)cO(XL_}BRF9KPZd?zX~1t%t}zM+<h4C>@d|A>7?0SIezflTMA#u^<NFRD8;|
z!nbJCDkeSgw<fRZ@hSIH4O*q4a-`)aMThsxT6bGxM!zQWyKD8-`vYdda2;&_#Tjm$
zV~`hnqWYG}5iTS@M{!Ij0iO)&R&)u^PGMA&=dg3K(Uc`CYDuA)^9ul93XxV=2nh%5
zgc9@H7lbi-TkOl8&@CncAL&L~?ZEqY1eN2IgDwLX?X*8JUJN}7LoaS0S3TBzRxRG6
zrJDEs1M&^4KoXDsF8vHxt&%b)qe&8nM4f%_PdMUE%GNdb6OD0@c42J``)hJDTo@9?
z`NLb3Q7iB2o{3cqyLFK+NHv5ap>0*YdOft^B9Cq5s<O-DUq;r#nTl;^Fn-(+s>rM1
z<tykmmIeh$yJr+o2e^u^s*C!0`EOFSlI?F#a|>4yizpC@N1SCvPn?edh`0P8g%Aq+
zL`dsK8P{&%{}{<Q(hU0-_w0@Kf0^j;<tEk^op}2C5_l~{4T7b=$G#32t1w}XRdb`S
zZCTOa<x&0C=H+M#BM_oh%Iw><6NtTKaLzY#0nj~DcauLI^jM|EuoJZ($deKz_i&|)
zSY(mQRCLO*&A@Xgq7&XTiHW^=Sw2b=Xde2C5X{k42@lV_m-oFnnm+8Mye$v?gx;-U
z5$ij8*{ZAFa^@3CW*qk5EL|XjyiW*0Yl{35P7|7C4j1~uW(HP;v~l8Sl@5!a>41{X
z7R`lzzE&kBUhw79?tuW|YS@WkSEJ}T|K;AK-=UGVOSelL!3Dh7bKO^SDHq2P$&$}m
z8&=PFpg<XqDXIBAqGFfywmkze3`Krns6NbL!BouJT$bjXxwuq?ku1|67uF*mM3FPZ
zaReu-*fzlDn5mNxZsyNW&v?1S%-Q%!D|5UY4AhRwnU^x~TC=4t266xW@yxr_zWoZ`
z8TR@WQQqL0EKT=>=hc><m$-W0c+1&lg&<qG>@U~O;c!IL)LWrqn*m(+Tp4E<Fm{h9
z)CUpV>?|cTpx+4Cy6i!F(c#YmPyrnUtUzxkd9>#*U?yrt$x%6$BY?MKAGJ{=44TI4
z@)g=o{-B43{rJVtg-R0sa8{}K9?zg(ZO1zN_aEQuJc2nXrH{fCuMGZ6b@z<9{?O3p
z*EU)ot%F_MpqkgQh{HqB7q-_tpWO8CL|>UPh>$yVb6+J%Oavq%$+XN*&kYU?qe$m*
zv~*HoOhblKYxmO?%>B$mHCmnxzU_N0Wp1*3#dI~Dm+wA1ETQ8DF6M!0Qfimgy)SUh
zx%6LVp+oSZ$I=C3-Xh9F;INo-vpv5-l{S@l3VhD6@6s3%gqM;UR23`c{M`5AJ7d!E
zC^J_mh4_Km?1!udYy07c0N*Z45Sd0qhl%Cu{ZBH;Vge#&0AX^MEb~@Zn*M&MQd`5@
z<ttrnm1dH~a^SRfBJ%a(=yTh6TEdJN-kGJes=Y^tBq@H5s6}8QkPS{f+{0<->gstP
z#S$8}d+VOwHFRFPKm0&MY8Tj@aE7N(!+3w)Vs{3Z{Hc7r;r05mfL+q8(|LfN`&j;?
z=h$ypSTxs%^sDX?r!)RWqUDDD5|Ejy8$fc5mPUvqxTB*P?lRAcKp*U8OZ?$?pJGXR
z<eIY>CvsO4hi!fAm1H>l8zB4CL^NxXhiPQUi`%5ak3I|^`1aMCi;;1?a0I*MN1HU@
zUGD#LN5Gj&W8Zp9#tzzGerAG-p?X&yR0GsJxr>^y9qW_r&tI0ec8$7d(HI>rCRMrm
z{vQHYUIGYg_<Bz>xA|H69?b{MKI~U|rli9YZ<H`v*+d;zyyKZO&k5ysTiOoeb(F;y
z&tC=}E9GyeDG=mN_eLVHUU==iN8CZ;KaeF)PnGejlniwj-F*^jdujF6uJs7!LsIij
zCrW-eRbwI;K-seIg7ckJdWUDH_{M+w)+1h}dmiq8-_37R`nCI0fo?D0hI548u|1lz
z&yHo;)<d)E`qBA@i*hv1QAxX(`s#crdc+gT4Q;DdO4VV(;>dN_oOpUi{U3JNbrl)A
zLir2Pby(P$C22ZrCp8MV>CX<UU|{%r&J2N65NqwEy$X+2=BmCT#k63q2fFGmqS&T_
z@hk9iUH!a5X^zR`R_HqG;|S2r9UyCp`Re(&@2e70*J4DU#kiiI1<y8ehImn9c}PT?
zOlt;|U$&C`uy<?M?j7}_=GEuG6Y-0azz;FT-o6}=K*A-<L{qQ{c%F}b*e+@mz}t9s
z%T@$G2<-<%qMR;m5Yd#BKgp|=$A>o>5Uwzn%H)BSf3Ys@pK+G1wte(u$TKflB|BIu
zW#JhcFQoknTFXgKuzxcraLvOY)m9#KT)4Pu>scO2g(`X%GanJ9HSl=w9%c6DuS<}P
zp_(yAzot`d`T~K@OFmYcsRpm}5}qv33$b@IhQC9`jp0tdyE{JfW@mE=?8z?rkmCMW
zo4eZCsQ#QT)9fM+Uj<T86n!)I7mAPrdY|ELr~^N9g?Y+k06;|RULyWgC}o6@f2maE
zw(A4Xj5R~ngJRb1S&Ur*LMuB!;_0`yUmMlw>c>f&aK!L8H5A{zEJ@H!7kJ|IZYd&A
z?CAtQ-!a<l(oQ&VL|AZvCKHBw`vt*fc{|c#rxVA0>Uy0kT`ZFjOE*J!0aA2z`^2xM
zLEq-Ni%@~d)-;evdB8h3Ixz!^8#EDpjD8K>J55M|BE0S~UIXGn5p?goyJOOIcfF{b
zt68IrTy$o>AtBF^Cj*#mQ_*Brj|FZm?sHElMp8W=62@P&xgx1e0=+YrconN|Ee);v
z<n=evFX}xxKKZ?ke^}wH|K!E#ZXHkP5R&>+N-7>}d_$U=j<z!heWul%t2u`XrfANO
zEiQbE)$-u~yKJ_XW*=I4yR{AWaYvdS50VFqJfUdb^L}^g%6%x7PJzLNK}#YgrV3tE
z#Pob6UF7}0l6mHW|Gz?6;);5Ve6zw2RI%}X;m%V4RP#n2(AJ!k4_GOmCU7rnH0+J8
zkR-hu6KlyYUIKEN>dh2}TEqmHi?+ou&x!%BEI&X-58=#5MS)7MQ-5CGI~ie(9)5=A
zL9d{nq0f)E{XScM@H7;=bNcIEY#6F3H<Z<qKQhSAY-!v5ri?`0DU|`tS@_&EH1u?k
z1St-!$!Zj8O=y1N(xIggYgR6p9MrYL*3=#k#`ryiW!!)~%WI_iL<(Ol&V6UF_qzRo
zwTq8eO4Z+nBL|q7M0#dLk$}0zu}+XpF$?i^eV)D^Lv#Kd2yKMf96&F>M$ih0i0e;U
z6GqE36W7{XvT=X+#I}z`OBcS4ECYtD9}J^J#e3jlbi+;>k@(V-AVpVz#Y6X`#7*X-
zWax2psxPd&KB<z^Vwev|?#HhXV1VrgloQAjEL$j^+4Hjl7&~~NAW$yB5N>o(gg_eW
zY3T=|`AyHgv|8WN(pZ8L#@s|8(Y&O`5bWuEQIpGal#+@WKRjrg$qqh@FZ+eee2rDX
zkD6M_mHW@v`G=td_t}Dyka3z<%-|H%yHl(_#e3MDJ6!HtW!FqBJ-V)sLP{U>ujm2h
zgnp%Cv9!kcp)m+%9kV+tLO0F{NOO$6Tbty5GZ*PRY1u8jFZ7D`w!4+V%|u*}J51DP
z{Q8J+24f%IEor}C7qmea);=D7{=TC9ttGyRfb^)VmyH~s04^(Pg4CQlALd#(NPLZY
zSA8Ce$cJwk`KcV=cT;m8THtwc<#RD-{p(gZrX!swOVzo4&`NYSdx;nj$2(6lGBz;-
z<ag?8mr7ez!CVTuYIxI%Pz4Qr1&_gFbo`*^p_K3a-xeZ%HT5ADKc@}<WNHSJ?s@A3
z^?2g*%$q(Z0Z2>wn!oTqef#wnWLeiOEHF5KB=zM0m-S!Vmu~T4^R&&*>Pt|>b9e*r
zb#x*Rp_y=^s_6~$^kGwr-c97W0d%yQ9*ZL7&D>1`XmZ{uxuy$3fY>i~Kl*G1(H@Zh
zH?f@0FPcu%p5IMBkYc;BW`L!rwC&(%le@Hf#FeX#uSbqtPsf4H2e#0c(BQ2P{9+o<
zly$BZjfblpZ5i|^wt?I<-QwJ-Khnp|KcD`%kqRS_1G}gVlG+&Q&g>rW_Yp0zMWgTE
zbY#L>A7B?*HRcj4{$js>n~u+a11y=?TQWa5iU<2kpTq3ht24u_SFSK2m;&AeHc)C~
z_}v>{BBXBuT8+k8hr&>XJy0m)^etVlhFxohCj1QJF;I+(LG|)4xbV*oVfG|<o%?0!
zKf1CB1weJ_%t9%;BOOJLDS!*CqW93TX=FnY6VFG~Y-EX#sb7^}%M?i_DCV$M%RGX;
zc?@%Ng-X0&ATho-`NSBzLkr_08|Ie?%mrIYTH%jm(Fy1Yn9dqs%Ae|G%%%d`0UZcj
zk^98dRc%)>D#XtWRx@_VQ_T0d;q(4cdLt`_58B4+P3#k^$oME&?mbtgM#Z-4T?}Dy
zFqUp)ZctkCj(HM@mSp(+o5T5^Hld-_GvL=&^_N;~Mz<Q;U0@^P>E@0IW%)}+Y%V7(
zBtwFQI4tDcU43uwG6K2r9E+L%*CIce-MibKqJ^;qA^3{-pOnK|sV2*-W{6YT32=W#
zAub{2?nLo1f7w4L8YlSY`>0;t+iJL%33v9KVJWW=RhJ#-&*t`F&gRc?p)JuZeV%gy
zrya5mWu@MTTzopSii^u5%J=QQnOlOz9@r<0S~K)`Qq`>@hv%~F=v_FYu^1QSu!f>6
zL6(M-<YwX=6WEVK$RC>NBh_Z@^GTPg1zcr(1O0A?*3DPO;}aEG?r$ksez5jx#qiA4
zwgLEDK}r#49HS72AK1@u_qM3{re3Rb5F3dZLVXL@hTjdZ0qWb~6t1hpGG2X*Fm&%R
zfx|uOhmJVf3wG6wsgS#yTb*GRgX5-O+}t+mq>ce?Q}+ehJpy>mBVwpk%*~773BkG{
zmhZKs&mH#p7W@tVYn<6|CujKH6Up)Y-CIj#zB1u4^Q}|g&T41E8v>zC()7-Y=AKXp
zn<gq=eD(@+lqa&6{fRk~(p;DrTlv=7D}>=0OVVME7mz-*Q{kND#|IBsu)XYGW*1lt
zNw=mSi5m9*5bEnyBzZxwKE34P-F?k*{kt#rG`=rhfkhRCyo~u4+GKx8GuVNxa?J=P
zuQy+OZ5&ZeKYoM(%%YI}z8B0>9jn!%X*J!&M;jF?FIIu*Uwjdv_^OD!pEG<H5Qhr*
z{Gp4?sH9UxUoeLlzcCE3hXs5$`|4ej6uP|Y1U7fUHtl%xE-E*Uwq3}J;|*Te_Sp!g
zG=X(1!(T7^^U&=vWn68OuKWvKznqyrXWaLCvlD}HweI=6)X3{z{Ib8|k)<n;fwk8?
zQ(gKoc8YB=@$V-7h`TrY!N#HxJ9&JL2QJwP6|*A`86l4Mxv-b1@_ov;u)s8=*<4CQ
zfSfQ8A>0zQd&Yn+a;}jA=B}Fga#|zV{c(5Zo2#yT;rz0E!P4-NUL@6>;3-Zg)xrG&
zff~eQD7Up+MNEvS<-<FvfLhA7nNwFbMzyWkc3}R@{#~rx-^X#1-Nz-LpG+|A<4UF~
zx@^Ss?K_HkB%*er;yF{E)tnSO#cTGGY&?R?%)ALkuou}+CyR{%88Rs`zpYuDMF_a5
zUfCtDaO}HHe__S3Ik^joG(>!m?xg1O70#3h5kAM8#t+Ndb#wBKcFfQLVpN>z>SeV_
zq&6wX=4{Qv2^~IQ!k_KCYWX7{r^(dG^V`ge^wHuyqGhuqeB$9W6!4Oech|hzg}S+R
z-T|86BT7@jT(3d${CWJh-ftM4)$N(wVxEb&W{BU{tm69-eJ50Ykhz&|IxoBJ*@{sL
zi1>)2P_?n>sKKqTTL5SY!DCPWd6n1aUn2~)729m%g<)SRkf4q7lIO6KnYm<?kUD2L
z`0d+AA2)*EQYWGRDB>xIWAWFYPrMI(liU+z(O~omy**fszJq6O)J316Mi=qKxj1N1
zc)tLSBWzPbN{78PHR=3*ZqJxLX+-LJLjytb^nb<k3oCr%zh3RpAZ@Qb-4ZOm5(O%-
zoDBmvN=q$1dlUBmSqwDhHQ4Op%8Xn0v?#s2)LFYlv)|EReYIu3ez-mTs9GCN`)u?`
zwXXJ%>O()XN$#+jw>OL!xzXbr{b;<*c!`2(6F*opZ-hvEzV})u&`f>Ao}OAkWD*!z
zGig#_mN%iO7^n*$oMSX&L|n!{SR+o%Sgg39g-<)`{X)%rE1wG4p#|36NJc8lr*Bm6
zT(AFi9Gtjflt4{E=cP2l^G>kKktr9bSER2gJPcdfb^7Yq!I8K9CzgPObml#eq|k4h
zdM!T=lz1AyLc?1pXZ(G&xultrHs}?j0Ns3j^8|ePvhxG<nj|i*d?reI!nt5xj@o!*
zfP>4&aemqomH6gjq`aco@l4i8a`xWNYE-;hP9w6XyhLa-@|#e*Lobp`xJ@0#x;2`+
zdvq(&S(o*0@QaNSZl^I)^w+Kv2&quDBTSrnY(}t0Z?7Sbp9cdS8o{6;2A$ghR93Wa
zcFy;^#MFs41uP|cRQk@7UM}I_n19}u5h*<0t=#lZg!xqjncdo(7L2w|AP=6LBvkQf
zUZNqdYf2cLov{>O`cv#5(A2;3WODBfh3WU$+jniuB)b<+e$S;iadeB^y^7y5@xRP{
zw}}%t87%q%(oVrw;Hd&l5=uHkbPI(euHyvdOsRLhg3aDpt<#KLQg-js=`&w9qBz6X
zqfd&S{C@NMU(66Ds~bmJkh(PHG%kfJ&kjVL*DT%hUfVI3VSMZ<Ea`ZV%Y;j#KcPo#
zSrmBTLw>(md5Ng-7ba@~&qRj_x2ppkcce-F@I26~4PDccL#akn=m|5YY1CI%*8pk#
zaDHt1+h!D(UZ;0faIhwG0>zpEYwn331r^U(1YTotb0)kThb(>18fgNn78B#Fg*6r`
zfRJH?;{syFjAdrxZzQxKr#)7Ol*T#|xCz!RVmYV4+>kePpxsO}m{I+1mp;5!O@4JE
z!7g?Q25&Gtm&bYTp$&HkrD@~=*3Hj11meguTRiLQ&s{criZ6NyG|)w57VMB1@+xlt
z-kKJjd21t}j|M%2Jt4(`dT&3`6S=dx#p##5?uvQ~595n#BEH}1f9B0M{y|xOJsIAU
zdzVqi_y`4*k}m_nX2Fz-omnln!o3+OgncAyi$;tJAvDZYnzp#=gDlHhq^4Yd_J0U_
z?{GM~u5Db25=03ig6Jg#L6k&`7A=S*dS?(}bVeH`dT$Y(5Q#*IUPqZkf>9Hqj4_7j
zZS*>Z-{rpF=Xvkv`+eW>{FURFjBD0jd+l?zbM0$BcC1Q8^>Mj=+@=<gGm)H)quKM{
zRrrL-$0i*D8BqPX_Zit_7y4r4mtq91gf#^&i&$y5@$m#vGz^DaZ4HK_q4zH^(sf<2
zp;<P6E?Vh_WAW4Yf!fEWE_E9)+EEU$HmbPb{&9yaVR=d6TiT0GrjlVr?)ngDt1BEG
zJt{G)4PO^%<dw8dTd@afUSxk&DERKxRfc)RRz~b)O%sT<Kucg&t_}|f0`u}koL+|Z
zzVy9iW9K#7&ry`kX6my1WCQo?kG~fX)qnieB+%evV5@vq$8d;dE<ciuwGJJPRck&f
z#Rbq6ZL!^S!))=q4ho*^Yp3`{{U4YkMHW0Xd9qJvYjZM(4OY3;zgy8T(qNAVKe{_U
z&1s>i{PL4%elTx!B8)~Fshc=>m5>6&SlpL|%ncou@ZH9}vULtKLrKn!slxgmPlh^P
zAKIEx?ei5dtQ3B-8GOYUjb_q&euxUa$m~OeFK2H=EGOJJqAfR$(q3MC1Sk2f`c)II
ziB<T0#zy&z`7d`sNm5mwE?lNkn~WlaUB8%M*|ekf47?)Kht?CckbNxuVGF`%=ukB<
zNw@-dXS&Uoml}W04NevUqOgWXU)O;&0V;s+yr6Oyh%`(*4#87EE>7<Do72nC`7svd
zR}}S!wxjV$4ybI5L<`hPd4=^4bAu1HOkwdxG97Zg@9X;06-*3CQ&$9j=00?2_?!M8
zq>P4(TTn49#I^BAKvgysIC1)-CH_v9zr&v?;;zi16Y}*s>HR*yE&`(L7ni#a;rgml
zqjYh>IP#+R!$ez$COrASCi7_Rn6u6qI903k#eHR$s<aeC2GDLKS?dH%(_hj!XWrc8
zz>xWg*!Xd>&M8qi9y8dcxjj=cztwxeCB|_sY|Z^=rcNLG-W|?F<SfaME}?Lj<DuXM
zao$~(yRFxp%x^I@Eb@LY@x;2&5KItvhSr!QE~V8TI{riqwT5J7vtd5r&dxNBJU(W(
zhZm>klM!*$cd+ib_)FETg9?!yFs5RsaUum0mmdp<JKCy;(PZy>$EEb&nc3dA$StJv
z=k;ss*OBK2VIs2_N2Gu#g+_wPk{v|)b)A>ZicszmU0XrjQ#mr3I)e}^pd3}1T3<(6
z<S!teNtpK#^$xhU9Ju={ld%8!6M-t#m-gNFZ7^Ox`<cRybQ$-b)o%4%Kg9RL*H)Li
z39>e1U2;15JMD1_<kRYtN1}sEw<A}sj5O8@uSMP$9gGi1d2LJ<);?mic(Ok{a6d*e
zBLla@MBD05QB7?}Jjdy)*Y<m0$o+_Skz?%feJTF$-vv?(VkpvBw@DIcJYh7=)`H;=
z=$x!`k>(O6uxI||#Ip2{{g`aa#mVi+Bk2@q{Gwr2*<V<;HJ{aFlI})N<DW^Om?jI2
z=_Im(`vD6<@yP7l^^Oys3BJm@gbi_C431o3_zR9AUrzv0@K_hxO`<M@{wp_)gdz9x
zFKhO&RD=sC7DL?h{7Rc6OS5)a-Dk%Lz-}efrz{OG7xwz~1W;Iig&0>otUU{P=xC)t
z*6r{jC+V3(>3Yn&%nE4CUHyK#oGdzU^%lKDjGle+$BF2$x`7Ra*@jgCz^82`9zq$A
zkYDJ(B1k5{rHQV6rH0=CDuG1@ypLw?d|<9DRNs*DyrRYt6e1W%>qq8Jzm59bHxAe(
z{d=HyUe+%KpBFF;OSoSC75nbq<Md~@Z@Nh~SH4(M{$?^s>Bm{EdFmv?h9t*OwT+^O
z2f&(}S+1|E@~-`%n5B@AExgx<BXEA|&HB#$?hSgjm`-AL$Pv3VBwBgk&ZTZ%0ZY8p
zP4WG#GefRWP~`se(L=EsE_YtK4MXYxF$vUfbd_6ed+nC;0D)7W9AIG&DJ-oFf%3S%
zv?oSR@5S;!ztLdhyJwweTM>A(*q`&j^?8s&kDrvG*}_BxdXt9^8#s{CA$<eB3GOzy
zA+7ot#`NsvgY-=t_iA&2aZ2F03syVkf%4R204S7m)N_~`X8l(sMQd4kmPRh9>vH*p
z61~u12S<9@T$k=i3kEj7hwdra=nbs2u2ghs%>z!p#{e7SMM4^<dAfUA-4HI!EGUkO
zV)Zl_!j1_4>Cnhn^UTH3mdlmSrp=Y;oazi{hvaI)Lcw7)wAr^gs7}&eI9UETP=Y9a
zsgajxIqE)@JLcWimcU&Rd(?TSns@GH6{$I?`L*d;?D9b4ZO)}zPHms`ZZS!!d|d~!
zh`3oSH{u+R8c?8NrEbHZlZTh@+p(kZ1r?&*4jyLFKYvshW-aff0=fQU4b<}qesFwQ
z(d@wv&~-5={arOA1xYDZJSPw{JgVZuP3!UPD{6_2AC*T1l!Do-S+*@_k*n*vQ%4FA
zcAuls(9wYnB)kCLy&8<=ZxrkxZ>DaZLWev=1w^deA2~ce1+q~4l9QbS&AsBdH}`9F
zPs><YAlJ6t>smF<l{PeLnSXT-4k!HD_h+9Km%#l|otgj)E&bBa1Ro=6ETfIZGKei%
zo29Rri-Nu3EJjY*$O8C!*tM&~jpAZX<5sO3gtHGNWJzQufEQxZ>lC-Za*-+wkI(43
z9m<YLfoc{ewJ*!Hynz|druFzFz&7%xgVYLE#i0?7&eSzYp|yy*uW-1<kr-e0e)gbP
z!KYU3M{n5eE%CbK7M#V;Ht1cabr;*}2dF(8F7ophICC9aL?)oQaSc@Z@60|RL$!A&
zsnRnk1k6uOi~gr`>4c54;^WChxh1Lhk20dXtd!|gN6E~!D1x$`Ot1>C9Fl3j@)0C|
zJoM@7SoVlwCGeu7TS${YTq7jLc7Pg(g*lh5I|T+o7Xwoo)y7hmdM1oI$bjy5WE$Tt
z*HD{E3?a+O08_56;kkvwTM3!vW%7ta_1I-Zhid21pQ~T~(%!qvW2!yVTE1~laR{Ax
zzrfhHd(eet){jeD<-t3!0CcF^<IUu)IcB&~X!+nqK3p_yE??m1Ca?iFYT}&H`v<(=
z4egmqPjO+t4D+L(<@ZmekYa_I|0qDL9_A*+Tq-T;zDSwg>P7SG34MOE7tv`JjrUh1
zM>nJRg+!g5V{Xn`J~}4HukP8IzAV24Fx9DoMqMYf7uNt~J+l&22MSCKys`(*vttoS
zdDWxwp%d4YHYNB-%H^0RGi-yl3}-vh_zrVN1fSmL5QyT$KnEG?lPd`-%bPfHRQ^U_
zAxtr6szdeA!drIaaN#Hg14L3<8hhy&m0$q+1yj!1X^1YgtTG_v{yzt#ypEi6@5-*v
z;oz|h@hZ4mx<Da1IvYQRW_P`-Zs(=^<jMP#2wHhvaT(flvo|xc)lu2tNk1LW0weJ{
z`cTE|CQ}Lbn%|d<88Qq(6xC?iR*we5+=$pU?WIV6FC3l`0e2d(Z{LE1G{UT0l!k6i
zq>h<8ZOgfp%PkqiSSsBrWOXusk=H%z!wszdT~Av;!0&-Zu{y7?y3kBQ6s=N)%IAbs
zZz?e<KRl}hN?uPP^L_^vq?79Sh8S%%>keznmB}Hr)g~5_WfCIzo;!cTyUOk%a`TgV
zObgdVq3%~M9aK-+<BP|1Z-{J`nNkf4h7Gv^e7qlBFE0<^Bhkr)%LBi5M$AMHS(PKc
zf!B`>$Xx4H5xQt~F-|Ol((=XdK*3Xr7uj#lE-(WHSMmrdGJ{*$kLlcFvn$Cfzz#wK
z-}y4{`;fZxyVLC1Vbn(g0T)8Afg0ro#|v@}>LuS7Eqint()Pwy4jfUOjOVfT&%$S!
z3I^U!td!T<@Fuqd_6_LshqZ^6O1>?+gh=&pwZ9+|mOIH+k<2w~PsoeTWZk$Hsz)6C
zq3~_V9&TivLv1|M-heLM5!nK<HkcKVko`Sy+vv;t`>7COcOJSRFVlyP?=~1Hfv4SK
zDz*fwOP;?GH;Uht)H%l{I3_~zjycG0$X9@s#GI&2-6)n8aMJp^p3VVKK*G}AijbY_
z{3;!aS2f}i7gt(o4AK$dPJW-5gItSXn*jzBlEy+si7q(210%EXy#opk4tm)e%uh$g
z7vU^nl9%z#Z6~vs0M0>5fSePyvixC!P&xMjZkMp~P2uTKGA)cF372iFyOj+;bZDHu
zG67`cF;}ap6Lqd~E-_1>*pI30nmDk;haN}lW;GVf4u^!Z@^Fd-oxotSnp>N<X1T1v
z*230iTyS0qt+`nmS`=>(5}lm=F_C!Y<6Ro%-#{oA6JX;&Mrg%FG+sAzLA`;+UOZV|
z5RL=x`k`XgxbjJ)Xfm#OI_%^A1G1{j@cBJ|H?oe+RQp%cja&|myzU0)mQzf&=lALk
z!U~DZ-gNRcYNMF^-i;A30#9yH2lQo`h0KAwh3VWm+Yg?VhFfAgN+N3)VU)j!9Qu8Q
z{n@wm<0^EY?vG(;^03&Z66aGE{-Qw_#aVhd>1sJ$-1#wWYP%B`mT2gT!Xmv(WDd&c
z^ll=h7;y=h^D;zJh28^-UFl&y<Y3?)6&62$@hJ1FZGx(AlYePUlyn>vB%YIJ>_E^=
z<(Cb^$mwK{w9;ONX#PJ2ZnWUugGD4{G3(bKd-!zU>T_$xm1{3|h8fe5&RGLRFv3ov
zA;REQY)Ij|0%_u^D+TH{`~4>Kt_0MOqrxW()0FyTB_5orFd!YKMXT^{1j>rl6%q#0
z4##(R%TZKpVIk7gQIwoSPG;o4ga!O8Vh(?~;w()cwNlWT7c^ZJP978{72zea`{H$!
zJK$nqZ8v)rZWgvV>Z7vi6I#6kP)hFkvx@4KEu%9yYK1z{=wauH_N%|NHBiczRkfIa
zpSqQqOQX_Urz0%-15oCovGi$t?~T40LYZ9K<x%I*kn7Aj{b8F>c=<FwH|IH%ZGc0=
zwNzSQ)*cz>aM%gI<8V0X`_|%C?`@t&_J|s@C^u^dvl8D5iZ(%LCr`e_l_0j`6j%r+
zS@Yulx`R~8+5-|l=52Z|aDZ>O7oLT93*gmELqWB|*CRf0jWp990EKCtknxKC0m6k4
z0xS<{m!~0LD4tMDsltswLHSC}QqrypJUQlWr6}Zx7>%#Po(DrNLa-eqV%xq-U!=xT
zi5(`0Ex`_=T0RTQgE#NfiInN`ZwAT=Apa<SoM^KcAs4%M#CmgHo18PHG4hj8{x%w=
z)_3YH6gghkQAkWxa#bRC<CJ+E;48ey(M29H?2l|=kLbgMrQZY(=^<Lh{>24Q5{wUB
zCRQX6k*cE%6$%>-bE8)h<^KNq%~$V7HShbvzaLvPp7;||em<j{y#v3<M%I}@Z<GI1
z^#BhhxDJLPKa+u4V996*F6<qKwZS*Dbdal?VGJ+<V6_EG1Bk_w2G;rj<5okZ3gsu^
zuXQ7`>;^WTa<NP~lg~F=@aK(!-fZcz_lM7z5hd$wS!u~k>DIJ=Eseb!Z&Y$<D(J@3
zPF=bM5H&<*(eR0p2517h@wkQOy?L2cj23qyrM@CVn~3T`wh(TH;=^Y0+m91@*NWMV
zQyB<E@f7+T;D3E#wH_*(SOjH1mT{L?V-G+&sVW9vQ*K%($o+tMd%bmt@IaaXxCZcE
zuwXUh;nAa67WifYTk<-00ogk7)`wm$e$@}DfWm=TpZ4Hgm+9~#t4czt@VnkNs%k=a
z)`8%f((Fym`lf*kx~q2Q=4BxqV@;{qn^|8r%WBj)Td!U7q!d%X<U9E?ZZli++G@Wh
zoPBDImYT>2H|xl-^x8&}R|{_cVQ8*(JC&yiE+jr{g3&C7?IB3JwW-e=;J?1#_yJ`|
zf0d0Na0#zpQyXU0H~A}rrzfeR)CVt(bTnxps4pJ2FFm0}`gRg@U-@xW-|P}1VqZ^0
zev<M6*k!nJD^g3!Gv2X`y2?dxWZg&gJHAFHDMKeTVn}@OtRyzL4Q92B?SvfVPUFqs
z>!IjSZO<_VvtptRk_~4Ep{Bj_h!wtn!H#?uWo2fqD26163!2}MS@%I<ctBSp)|PM6
z+3aWLj9cndON@q~ovPG0(3B(PN|7Y%ez;R%tV$6v@$CiY$H&0bs`pKaz=Uiz;=F4v
zx;HCr;uuX~#G?1>zG}v0Nq%?(9H1{;^_B&{0~l*8zM8J?-4nZa?gX()8>txfyFP&4
zx{{>9?sdBF5-Lq(nL0H3J?9haUwI+0V#YR+vcaIqBf~q$IbL|;heFbr<!?C-jc4Xf
z+A*%m04f(;!qF=UJv*<0H7Q6Z?PM%@nj&#?_(~Nfe+Kj9O3Om99J~^afJdY8Fl;cy
z-?Oc#tUNUQIg&PPv!#M?Hd1v-no}^cUKkt?xG(L!;@bla<Jw+p`~=c;(2dln(V45P
zjt=W@a8ph*!kP1T?j1V8m&4>o^(e0jUtDh%f*wKRJYcNN9b9)(47{9`pCn%4QY90p
z>^6=jKMx?b`4mOobKi~ft&wdT<77Rtjz9UaP5?%aE~Pfqn5{A3)dYdNq7$kX`GrW6
z`kXzq>SNVo0mIZvBDt{vL+5j;h!v5)?1PD<O32URMW#sYR;8y<ii8XZf2;Z}OcZ#N
zxtFr{Ex`E#qZmFwpxZQgB-g!!J!-`{Vb3XwF=tP#g#gFdg<@ks(JNA1ra(SNxg${H
z0#-MHko)cDjK>V_!Ka_|er$e*Qx3Sa_+koay%PVakEZd#3ScgkVCHQ{U)Tcldmcp5
zo!tFxcKfdZL!E5%y`-pQiMpO#q$WTRABuUe+6iTlN%p(F5Fxfq%#_(TyyqO4h4joZ
zqqltpP>TWOFc)@(c7;g_Bycj&>*6Ck0x8@>vnzit`A)0fM3`wGEM;(;5v^xT9q>as
z1I;-lxMh(&^htp;4xD5&E@)5-z{CB^LVg56YPBy7fO<suu$^(_-836~i~rc4MFOQ^
z#uVGi3BZPEYcIr&5jb69Bxn~58m|xO>_fY8*D8eMb9GH!2N+-H*8~boB<klw_-Yk}
zKajf)57**<1r>fG_(%2R9>!FZB>m#bG10d5snh+2ucNrqQj93UkW`&s4$|biIkGUc
zsdW9e&cxj(6HsA$5LW15EF{m8OzWuSjR=bIag3{K)YBMhYlqxiWK}HT-eL?vN?T5s
zfT0EGJ-Jx*Orv#mS}JhY{KGMW@u1sRh7T}K4&j4T5t(_(EO#2^qiiz}>7kpqrg&9{
zxe`THfV!1+hb;1g@87GxcGzw(a`-qtRQAqPt|Xq|9f!BR?BxkpYdXeOz5@*7V`=mT
zj#JeND$uZQJNv}AF~(a8R!T)L+i>=2-$&5{V@5QrxA<yCZ~yMdp*WDIQkI<9YwQ(|
zc!vYKuh@>i2hi;~<dZ2=rc#&TQpCP|-HeuRk}^gmRlbXe;^C5TN;H0*cAoA-$0>UC
z%=7U$KS4aoM^E_$Bl9E9DGgKm8i{QaYqD?mu002CdE_;8KNGW3zS&~K6+gu?l`%Jf
zmVYn|p&i-}A2qM7h0!z^Zwl1hvKRBPdMd-JH?aAthMIHfuY%n{iS|Pzg(Xw5ig!BH
zEn|U2j^Zr-rK3Lwl-v4)cR@Fi>f$BB^P^{BO>qEdX60GdZaExM%u24>4K*LKK|0xU
zZmYl&$qW1=Y+aIG9AZD4gAD`&E_*6B4wQ2;m(=7}_$pyI4JS+dJiKQ_56`gY>~CI~
zd{cD$_tLN5<B_U=GNVJc2mvOfmvAcf!4X<iV|lqIrs##MG?tuO)u@Qjm2;|p9G(hK
zTJ=IsGTT(oPzGIXj<h8bWT__(W0SpivM)JXyWb)A((3%|iiBBJ21otEiwk<=0$Y{{
zg}BOO$r`=@w#)LQWO4uBtM&3Ohq^^#^j-8xBhfQg@81-86<#l_nz+J;X*U3T=l3_>
z?NCwsz$u~jrH&JBxWDk5T!{`mqw%9uF+mJY<DuR#>96>I6uOj*oaBccTX6Z<bGlg$
z+XRkbpJZA)vU`TP&<IPmW0KUdgO`vCdy%FWz6BD21kH^v@YbzW7xnX&0LNH7RJr~t
zz-V^AQ7vQ>5ycE!+=7G}wc^-A>%y!)Zm}5Iv@siZvVpafMDf^;h?~8gRW(Qkvoc{0
z(k`>b<Osr{qx{6Pho5zFjOpcLfdmieX~uq817#lHy4+e}@AtrB(|`gcxZgJ*0sYoj
zZR~PWdC_aICZHx3Dez0^P~JY!^vcG_(~Eq#u!GdS)gR9@bMax6VW>Md@*Fy65vy|Y
zEUJ2(mu&@0jlGe!Ph{wdgXt~E(P$VGA=%S?puaHcIlBr~t<5Z?4XfHBP*}$reW>L_
zHMrk2uA1AUt=5v=Qi~BB(INIy-hZe|w<4x>6RGg62v6&t0>fm#9&#lPCO0=}m69!!
zCBkJPiKaqf$t4r)1d1PfAE!Pq5vBAaEEtIroFFy?#%w8LGVGQ2#Yg;r@`~xt#C#V3
zlYY^)j&w+Cca5uQ#~vm=yV9{iF!JRS$w(uuWHB%vWq28or;lhK;!WVD5td$7P#tR>
z2z$ZWH|A=QYS#IlrSq$H;=NaY32C6hE(@af*?}Xw=3P0ycWly^5>yHc$t8aYeOS%E
z{ctuME)*tBfNI9Yqw%@e;7TAascDT(LzFx^;+%DjQ@u~T+T6Gk#QEL${v@wzuKST9
zQE5psvZ$Txn9IEkM#CTemjCS#aqwfYt0X^oew&5zH<W{U9gouClB>}1i&`*6sLJSC
zt)XAUH#)>(@^z$0#EMDDjy*ti5b?I>H~YG-T%YOMxHXtUG?Unz96vGvRW)NXMh-~j
zRUKXhm`^}5ko09AovQ|X^T`v@$)mL3_99DezFKJ~70E7m@iXM&rgml`3;E@8`+<Ux
zaLQ69=U}|nD5a?=Sy3x492=Yu1Fk~8SjGbFF~~Ij38b@8Y;WxajPv88BcD)70pBpf
zW0vh+1{3qhzmliMr+uOB*XdNHvJ}`UgJiCy&BxZa&MR@5eUJQ5e0o*oTy>U1!t68K
zrVvT{8OgD@yYs2j8(VmA*5sGjgVll@$#;RY%-nrSlG4Nz(k2D8dr9{dNC8uR?TQ5f
z?mFi1=jLI%$q2_(7yR4)7`fUzoLx+1O2ntTzz!e0&xIa2T<-AvDkIJ^=ci@<O+~Yu
z5H9g5`Hs2Z92-$CKgG=4p5x##<!t2v%8|pH04)2LoQMH?T)Q7PUdoM8e}yRtD^IVc
zr>juV5_5zISJ%hZMtowt>^EI&#(l~@Xu+|4=Q!bo4`G9!ombACx#WrPQSHh{-1hp(
z4F3w3K#g@jW6>oJzH+yycjf8QO$ii8inJc|eS*-^0;pV%c__F?8)n8>s@=-$R%>55
z)NKS#-%FZyYfLVueJ%^%hJ-4MR|>8h{(RKG7FC?~H^88=FzI;Ui#N5afNf;!VDL*>
z(UbkLaFDwTbo{2}q?*KjVF$=PtTKt1`e|s}qy7<Y{IxRYptoLY#H*XSC*>o<%b7R6
zP;l}o0d?g1WEZ#sO&jxZB?*-NL9Nql(3@U@o!ot&p)dESDu^ye9_v4N%rRuaFl0fI
z?nHh2#46vdkHs7szdq>`Mw7uszK4eRF6=FQm7a#GmTRjfb0=!g7xV2a|81a&UjPH0
z{V{F_p*6|a+bd2c4=_>8=Z6WGB>-TnvqFMtXTAj6h)b}Qw<8Vf_k6D9<7)2^y>K{K
zG7yDBu}r?PbrI2fbMX_+ZoQ$}LfiiOAB*#DAD0d5NxEl7jCz(JK}A4Acl5DF7GPOi
zt}WKZ{n_j2ey^t!pz<26Wh?C~@1`|sEZlsb9+CMr*iz2<HtY<;W8q^A%>3^rbvHBM
zumKnPEI;RnGYeBm@hsNEDncRJ;kR=yuuNzR6h};f`sJ=|uB`p<t~qKu5#X)hp5G+|
z91ScBCT}F3yaM7q06kPUS6=<k>ihx*IxdKh`F|C<QawJ&$_D44q)iB(oGTn$kPU9g
zs~)aKCa~qo3PBau2i%ptiT@EO{j@W6%bw<wFk_R6x>?#Eudg09ipIY^aW0~54+j{}
zFRhD1zfg~7Bw}0O$W^GK&F`p@<t74_2AC_P^Ucj^E9D9z1ejU3%I!+WD~ENR-`EXM
z7IFzr7Ro&`tQ~t5p!(V)Rsm>tQ#IlMwoZ8<!g2+ax%Oha^^?3$&6Y$F!%t5!ca}M_
zgl8a^V5Ns{ptCV$Z>AAnV2R%s_Fr!)m;oDD_GD}`kfIm@Owevm->Ge0Bq=QfX0}SS
zcfSanG_c1xW@O7pW>JMWJ`A}E5UlQPcVasL0+emXKiimP&g}+TU)TKQuweKX21-}5
z*p>$NA?SL<iVlceZBwi+V`SaVW<#)4mxdIWiMmjs%Cjm60VYaVC#feV25vOeVHrLh
znsLc&mI1R1b5skbg?chqdvcI(S6j!PrSuy*94gGtM<@jc3EC0vxU`T?P5_a{(Wfyx
zTg$(wQf_M_2t#1uEItiebs#macr*WZZ16#s41~BExmckFN&-lcGCqiO;GO_D^Xa-z
z-PXzZG~R`4&ntLFCK4UY>0X<?nG+&vAx#8Crdi0Zsz3X4V+edpduM&w#H5bi?;a=*
zF#cnEpGsFYDd(8=e_hMXL1xAPHGy})m@GvQ96g-f`Ih3BnXwojpQ^kD@mMlp!4+2W
zimMemPe{?!XDr9usoyRXu?pr=%K0z#C~leE_M@Sbsb#g9tkv(hIysEK_}v|VlWP<+
ze&oV-X8&i4;Kz-tiS*<~^xeMXex%#?s-LaiB45UDMW~z$KQg0<f1q~D%CHLPkpX4d
z1uRVt(hlhAL}5;sUlmUb_>v_NH}dVLA2`4R!-kuI`dEdiqVtRvRSa0k`^_F)nHqG#
zvU7%e3p6cXPi#rmRRYs0%LK!`1@G;@5K{XbSv8dpDi4q%k3OPK0&U&hviztI2q8r4
zFx<$=t4vyEC)ld57OMFA?=eWBJxYpn=t`qth|ZIb?BsIq6k1lKK$v((;NU1F6oQpf
zPace_C$w8opIvDtH)ixBm}K%3tf>yF^gh>EZS5rFC%}-rU3<W>1Qgi*9fGEe=f0Y+
z-J!T7eq8ml2(U=KS;g|C)anZB3ie`S0izQm2n_G)3k|q{tre~Htdju@v=#50eD@`U
zpp#hF`>4~W*#gO-7=BX*6KN}3$Li-PUSm3cPTgR`KdS;iWcWP9av^1S`sHlfSgV3H
zy!Xs60ao}c)|Wzzapd?7IHg;SSD!!7q#cOIGSm*+KQ}Ih9nQxP<W->PX6a__vFGJ2
zF6p$18YrCxov9o0i9vv>KF|96lIZQwGhWUOh0rP8<xa40U%OTq9g9xslc~XJ4L$VX
z-L`>TD9KLU`Thjp)>b?R3>4tGR$tCEQaeGrH&^$fTo5;8`!*;xs5dmORCY~vV)VkK
z13+1f0Kbf+HqO)<n7?iho46hYYS|yX{w;MNs`)Otc8>*|DRdZy+w}T5t`hrZ-ua$X
z3`lHFBX9I8;aNl21?f4Z-(9sEah}$>`XIk`y9)X}l36@9*ouFO|2s%yUgC?{Ev5|R
z{Ea&(AYb~m*W1zb!9tzBF(m6)b}?fXCOD}x0HBE36fZcwK&2z_Ir#Z@>k2an!(YL#
zX!VxZQspkKMIx5~TB|hhRD`ViF7HNQ@a;v%1;<dD!Gk@G-%L({-!5sWCM73V*8#%i
zV#>H9I_iwokEm!B9J~BA0Dj64wyqsZQbU?Y#=4H})aNtgbWpc5<l_`nHwEru+rDEh
z`i)l9-X-O+Vx*VJZB=%X+v+!aMh9JTsT(C^TNR0S8`~eGxZE_p?I-bLoEY4Lb|ap7
z{6qjSyWE;2AKzZGQ5@Om14lwV$WfC3TxURX<ai*pTg~8JU8F-Jd#caAUtH`q@2aFD
z2b(Y(3q4!^-oOU8!?<Rpe?a@2pTlenVRP9Ngs6;yYn!fJRbfW7#1u#3QpS=K*(izc
zlcD2h8)rn-?XQi01aaVIJYluq3upk_2Xb#5i#y)DQ9UN19_<M$0sry>Qu5DW|1?*D
z_9WWjcZ6~I3=vrnK$UztzPT{PKUgaQYBGQ>nRh|G=%|COTfXc1cFGYh8Va7q-y>;{
z0mdB)ZUX%?MyFug5`TTkk^_^b`8#m+!40BWS@;9EQ5X{~a}E<N^zZ}F%HOu&qF;A=
zcN0RZJV0d+V$1-DrGyIVE=Htcv$CqGjTgbQx8uMq<9e7(iOh^9;$~_YhGPSS2N)JJ
zu+bUAXpt)(EIS(>Rt3Yk!`5|tmKk~R@F=(|Tst%)Rv3fjWX<skGYbXs-*3A0YB8%R
z-MM9eEy?SwfjV=4uC)R~<M~TO-Htwe+0iUU@)x8sBeL~sFEyQW33{$8eJD9hq$Rdc
zB%OKldO%ir<a8WP7UoD-Zs;X@i~kS%mD?mDJdKj4UZqpK-(LVm*`TAx>2jL$-FI-y
z^Jr_po+YAJ0|f@B*9!*_onVEV%7@YFgD%puE^)D902RJ{8o$)?Dw&p^6Q{5qP^fYR
zzM&###y8J7>%gUQu#CoU_%^A<d`@X}w5S3eamy)q{6K53m_Kzqw=9sJYX#yK7x|&<
zd$ZbuKw+ScCX2=g=D&=m^#KBw;@T`EhdbBy;l6=xWI?!;E{~zkUBWLUQbbZ@)Le#7
zD}Vy96ay+2PjXxWD|}1#1~s`P=iYIP&VAP~j_{0*YqZZEPKXV*fUV`12;)Bt&}5ku
zSo>ALE=tpVD8=YYcHW}65`n{(zwvPmaC0knKp0efHu!2DO;9&l(fqWdGK84(qso-4
z=t$!V$pQ?}4FFaau^YIJb)79-OlMbrSbmJc$*HQtfzs@C;LU-22g87Ef`>b1_ALR-
zioLIh3m6S_vfO7F3D^U3ydKvh9Vflb*)6$?*`%A*on?S17SV%gi(P?E%ZnQTp+$z#
zII>2FbBGjh5g|`JE!elcxp<H1uM#Uf5v!)IzUh|)v-<V*aO(=II!Ldd`Hm;W5e+ea
zD~>zAEadg!K2TIZ*sSzMgwaZ<-Es>IEb;g?_9-3hc^Q%f5Fw9}>Q}`8ND5F~GU51c
z&IyK;=m1MI1(pWX|KeQ%7+wv?beb3-jp;=9eV$QALp@;Y47O4L2u}zBxTKon)@u3p
za<qM_G!y9CTJ{ezzU$s$A}odBEb>2@#XKC5#@|uizwOY-<$lE`MqQxx!5M+fcl3pg
z1Kx>5?c^ahQJW~Q_~X2^;5B|yRDSR68aem4Y%n^kaajsqFxj|K^y1>>Ljr2Tf+=ik
zDVJW{B3z23Irc>LmYd>8^9rFQRU2PBJIRgft@12>uVI}%Xi=}D@@eQK`T~np<ZC2>
za&8;N32S*d@RC>N+d7yv1YU10DTRfx6J{<I#%!uTjwK8Qrr$g_D*|6P-Y0EJXvWdH
zlO^~|T@`cd?SKGdP@>mlbocMl7SC&-@|C7;SR@4<9X4)iS*Nppe^rAbqOfJ$i`@v^
zd(*I<I)@zj^Cl3`QzJl;Asg}C(?At7r7q<HM3(l5`k7mIm<OETtH6jJRi!A|%o4WH
zkhYw5Q;qb+i@3Ds$Jeb<`yI*m9FFL-DP6;|R-CAQ?{biNS*&8CiLJP)JfymTQEA!P
zO@~H*6|}H^Uqthj<Ez`SpukmysMz|&D(N}{lzu6Y0^Ao-mDlHo;em3UH+}6^zIvF5
ztwKtpY;wJDE8Ph;U+L?5u~`b#O_6?o*dFsEj~m>??qudIF%pCWT!l1OWmATBD5M#8
zjy<=7bapes1y$97+51nokj>aE%V@36{B`W|aQ(Y94=8`lw?DvK(2v@^lV`Jum(&h*
z7P7hA4+{(Dk-bD7W}mU<JxcJy=MM-s0bH<%02m*)OFZ<QE}O+ty&cKzWJriUW65UC
z5eGP4iyPCv_fp*4`)ci$-Img!#S0RrzzumoiEA2rzy~=@25P44fiyZzJ!o|x{(6!0
zXCq>T<9gO71}C_`BtR7^Z#!Uui&`&{3NlY9B!dag$z&w(AciOpduyw2_Zl;v$qtha
z>1=u3h_kUK+`eI*e+T>ZxygDK(h}(D?VC(70jE>}dnuvMZ^Z3Dnp;olGgJ?u)AGF?
z@md_<WEdIX2!IwoAPhjG;Hll3$6ks_<P<o4DdvdA9`Hd9Q--Q3JA?-OtMx`U#V-?3
zkQ$>2qu-N&LJ7XRq$2rXB_w$ZjuFVM7$aPxv`Z@~+Xe@uw1ysqP=ZA#^2vtW_(+GU
zh7|Q*LZq$TBv>Nu&DuM54H$~owTHeB(<9uvgUO_okGl^{FsO|IQPZ^i%SSpdn2|>S
zU+vS)5nmt-H&E}uH3tG?xMc|&fy#h{R8F#-;P@PlL4Lr6d6XPP)QhOPy$h4jS2H9w
zBzsKWB}sLc$S~qt@8DZdO0pI^*{JhScQx<!jQ5O2-8Huoka5VhvxMeMmyJ4=?Hj;M
zq(i~IIt2;T4Vml50^0%B&#l=d0v`*)gmuG#Iah$79v@Jo>#LmaqVEC1*kjZr|4wBV
zVEn)UVH`J31Se7!`;SyrV;$%)%SRHK0IIqoXM8G0Aa`Y&q!PTjh*c!%NK<eCZim4U
zIyBmqy+;!-9}zneGA$R9zn7zMA!OAUsmlhFvQhvuKrDW*S{?3G6K<csa6sjeaxUGx
zENdPBascci`4lL?9~x7=oH1wzbOQ{n5hEdh{ptLJaQnzKb3Ah84$wL%>(h!TinwWR
z$?QjZU-6~uk8!IlQdjPTw&yqJ88yS#unHq_zX~HLuZ9OHMsg0Buu}8YK4N_Y3OMGd
z=B^n-j(FeYV9l@rDmsH;dfP9l0Cjxl0tMT<r(IjC3-3%f7#kdaR2NHepToQ9T~ZPS
z7kMJNU~R&?(D?+R=7jlo#R7scuM-{$CKo4Ah{P1Xj%%Jxs!Vz(5P!E%u$Qa+Cx<q~
z$RK!6T{WdCZJ=UW(9I$52wJcQa#(2R|KqSI+9{*IT(A<Y)Tktde~KA{ENM2{=LX9_
zAXr_+-;jfR|1>c2qo}y(-oDbfJzEkLy-D_a2#>$>cyMPj?}O}xrNEXf3xo2?20H^E
zwW!z9*>SGCCwb#jMir<I$N4_$Qt=7`kCSG9-Y#d^jD^RV!2(@cG&EYC;U2gQ=D2Nt
z&6%u)FoT14Wu|9(Br>l+=H08wm%{9oc9d@oq`UM>;XSa}>HL%D{}QgFw{2jQ%p}uC
z=Gha_;?pdVdt397OU{m)wui5xP)QczmIE{WeN^R|&kPhZiv+$b3{Wofrv<4vZ@#Ya
z+v;C8F(&zwp;?f=pu-%8e>LtI<dGKNZPy~+rG|F~3G`ts44S@_Ir8I9y`sRWJCz6-
z7X$~2BUeQWipe=0(&t5)rGqEaydcY1)@N(D@!yanLF7!1i-fdzr)0LH2ngz2OTFj7
zB^3Qlx#py|j|yBpkNH&=V33w+Uump!`?o1<&eG7~Q<<s2QgZ*(71N=%eB=jaSE0uB
z48C!ekQ<}H7aRyoQXzYPHq)ja--G-ja(;v;|4DP%X;Dp~9nRmp+CscTlG)Lfs<z1?
zWzhW7mf3U6gECN{bM4@?3<Q(o1zA%5hp<bJmzwzY1|lombjEFFo`CE>n>tH7S6WP)
z;s(a2h|Qcc8^*DdrnszI6$R~?$!mM6QD(usDmA8vr}u~Tb{P>fary@Nw-wLT(j4st
zqMqyPrmgd~K%6%}&hKoj68Sh+CMw2{RyMHkwLDMDO~6Hw#v1;M3sB+FS=v*j2WmDx
zZ_-pKFY72Pn~~4|{iKJQOs#uZw>z6Ve4~;X<n{aWmSbs$7X#uDiIDPIn1(tJp&zXM
zKB@y&76-y@fN~$Cp46rLFFhI`)*ZJwPZGrutmrv!exnj%46;vyIULKk$#jum=rw0D
zK_40pnKXkH5nG01vZck|jpyZSAm>dT6Ii4X%aOh&_O~f+EPuZV;VdI{qp^qYq;{uS
zFBxa5>51ckJlV~Y->?(>HfAa-ig4s_u}Twt;)yt}9-~rh+7r+7YvV>V8I4sXK`THq
zN0U36*EZfJxCOF{4{g+syS`m8WhVXDq_MlDsgSpjXmF5JfhXJ3J<TIULIzX9`?A9~
z_^}flu!h45s%*vVhP?%+6uhj#!D0IunAvK0|16OdIT{jGI5f3pfVkIt{<bV&Zn8(q
zpz;>TE9LhQcuNJktdY72{M(q#wv0su$OBqR`#=7JBpY+604m_t8E<N*F^RT(k@1mD
zL)){9DssK4);ASWW0BtgQNQaNsGl~h?U`6pm?3=lt2<d*Z})2jV|n`ga+Q{66*NFz
zXugL(5o*<d^^tddxWsmd8i(TVXm$u8_9pNy3T<D{r%vZb>%(7(cWegRo97y6d)h=1
zuB4~8*$&LT$sXg`8oc-OJfDX~$1{D$A54Zj?&e{?B_Y1}$2-jt`?2%dsiIVH+otl8
zt_W!NKoi^9Q#fiH3`k^>#;&(JOo@}rn6v%1e}GC0I{CI|c)0se5-hC)#rSrv!F)?{
zT-;z*z@YfppJ^zN17W|gAwG5hA01f8KGSJzLRYys*S!?)1SGSR9RW0fuBvc0bp)Sw
zkt`1!C&d6}J*J5*u<?li30U%rGo;zY>mjzOQ^t1I?KEfJEHIU?;d&ILq3XT$BTCMj
zpEW!99KWQc@F-?2M3=3F2L@S;1}ktPwA0OHRXi%Fwro#Ic?$InjIgyAx%@3gX*kDb
z>)`wyGb8)vn+lct+qs$@m1Q&H3T^>Cl&s{;nI5P2zi4<@@mQrTbb^G#)9>VF%s~RC
z4tIGl*$Wp2V;RrKrI707Ww-Wr%XlURQo0hLoqULkvNA<v^n`@{xu#~v9cLj~;dAiG
z*L;UyJf?1Zs+6Y%5(SBy;dEBaEw4RvopcIG&MttB4PPB1Ik12_7J$qeSJS<Yz(?!T
z`P`aNBTP0w@-6t3YfoH{9b_>cRwt8L2a*)-+HAa?zF<{*h>b26Z{xPb!?bP~-8VL`
zu{k{hrKu<33w+eVshDyu?B-gGk_!R3HV=0s3p92|DRB<~(*k1ipn;^Bxzq&_X2UYS
zt{aL?nS3p%Z(bGq@7Vm(J(wp>ote+O4m&i1WxL{Z;<tR|G8!PUB1?gNgKzSs`A#|_
z6{PX#p(&V5VU7eQvi0~&pYvANdd}hA)!Z|%znbrk>#xXjw}J)d^fTAAqbgjBMZ@-i
zMEbPLoPW<-yxW(93BYq%<T|E<k>jV%9@q}lc|Ocp{p07YHJH>eAF8(UU<_or)S!@4
z(ck*f$<b{T)@y@l7*~)&4D<NAq?tJga2PoVSS~v@4^Arb_-m#u^Y~jtp)z@$x9(=o
zJgCi?%jZ-CfLz4@?09Exg6E_^@;u32;7ya}?ku{qR>EKm!85GAi>;`%6l>W!&KhI$
z4OELV5Etp2s_gq%h+@?2=qOWYC_ArBECXH`qsTPLBwnD}z52v?laJXIxkJxv=-Fxs
zoF)V2;1HwQ#_J=$KL)X~bXsY;rY;1UIVa1Z6qt4oP|U7SPdVrCBTz3+EilL_9_Rl&
z&C2q4e7|YYSF=NadB&tQy~E5IqPc6Q*&&U9inl%Y-dXnLLiDFix0l_1Vc_OeS($)5
z($%Mab6TJZb&LYZ#=49$FJp($*@_49v8O{+2;b|-kKE#gs!OY(75u3_*1$_v7xP|L
z%5xy%CCml%qHNQ6CT=P!@FJFZPx^Qu6=kAd%Hq?54Aazk{OyO@22J|if5~XxmbOe#
z%$?vr$)ytFf5L<*wLMS-JP&r?2RvuMtjqV&F2C02{rFVRM#*~*<be@rQ~7MQ0zX}l
zwP07ic84-qSS>4@Z(dwCgsL|ECP>0-4R$&thiG&e1@7n2%!B>o1E%au|5J|jTI!8Q
zY_IKY8vN@Mw7ndgCvm$K7CcsMi{pD~)4}1(x2=v2gwy08bN;2>Qc!%#jt7f~Y=`^Z
z!h!`iucc?}qdR9C3Z3VUATaaZsxc;GAt1h|J~22ck=?C+J0L_90iIBNJW5@QXijj(
zH}cSE>^4SXkTjM38^aDapN9Ogok=%OTbQE^k@#r#?ykc@a{4rgrRQgwoM2bTZ9ORP
zhRu^Bo=oigdOI|2xeAA>%~?1~4;rAx<+ui(YTL2LH$6P5;aRQ<srbkSx{Q9n1v-uP
zX6sEqT3}ufaH9@zf*bs$fFoR4#kVJ3b?^%~#F9EF3$s(gmwT@?v3U#ZOmV&5JIDE&
z9TJF!o*r;O4WuX~I4Ek2MOTDj;d9D4o;qpuDqtq!2Oget$0Dfc2|31LT!ql-D9I_=
zM)C0cd`~p*$)6ukaMLH}sg{@GBJ-w-!lX!2q$z;K|Ni~L8<!T-;F@##rNep8rB)ba
zYNfMeZ@K;+(OTi0)3_%kJVzBMKBKv=EneZfjbMJu*(umXLA9l|%grqEEaV>Yw9(SQ
zV7?aPrn6hZ<Cf){zTg+7e6(8G5U7KjS;+c`*($Yf2JTwEYu1t0@L`;N)D(6;WVOTG
z6Bux|f6Xq}^{cO3SIDg(zw-I%@2fRir7y<nN|%ksR?EuTSw%OZ68KX+9!fYI3~AoZ
zx5;=9YW}bkOVxx8dXa`R91~*Nm61RCy?=5H!+)eA9_Sp}#wp4N_pNbvFN2Sp?E2IU
z+6B5o;2|8&CV8g;#?G6y=#H7ld3mYe(}bP*9S^J0wCQ^Y`5VcrE@f-p&YOM)?Xrjo
zet&BNbW@on%2Za$KBTJW8my>;UmugriPYsnd3Rs#;*uUzued`(r8Zx+aW!UuKh^&s
z$GRqnZIYMX1S#62yL(Mjr1$BHAmU{HC|DekUbYs;wlB??htb?E-dT4-X+?1zeMT%X
zy)S)jWl*KNJ3s{h)Tx3(aNAvkQT_N3>T3;niCFWC{2@yGjP*<E0)6AdA)O@r``hK@
zbs%6jy}xtF#wtlc{~h`#$w2h}Oy?_n+FFy{<3ZKxSOG}68eY-=uv(k1w$`z-VJ}^d
z53%g9S?rzG#*dgV8ly2BGmve0+2%fX_96{4^mO_o<z_}vFWPVVq9!r$!Th(A=V@*C
zoP{)hANz|}@a!N_@{Kx&qh)JW>20oZ;~;_KQgMdWe9QaMDk|Ofc_PdpmVcga-oMc3
z!M|Nt{6N1PG{wh+?66za@UUx4(=wwWexyhI%oT5+hS82HZXS*ApcGO)(@&GO-&8)T
z#}CcVOSOM6GN??#1-qpc>g+D<n1SVnMm1dv7IICU3&n4*C&e%M?Yf8aObk!`z>k{1
zK4izwu-gBvARzd6{kJ;Q^I)=&QkH+C&;No=SZu`x(ijxXiLgq{HB3LVNxLvKaARk$
zTLt$w&XF=x;Ooo2d8eKUgR587_Eu?o?2apWhM94uZRCYKM>@OJ<5P*Hl!vkp;5#?J
z^j7~-GlS%b&S+Y>Pxdw^jIxEhgu6U^_phV~{Z~>%{co}UcbBArT@o!@W77mecqLQK
z-F}0EIWR-t=03iPnZgmsDF)pP!*xV+l83GCe$mtMJUKq8sTm&C9Max>$3w$jSoc0X
zdgrEV;lkT_F~cZSR=WfO<~zc~`WR5ge%ECt6#VaBRs2gC?Ee>K6adO7#pvx0@aTOi
z!)P>GX4}BJ4Hy;jir~UCe482bN5iLqa*PvY0rqKydb?>mX62N4v<w2$G=96SdaLfs
zm0!TADe#uN@66O^B*Ac;=_t^9kL+?ozFo@j<=-;!|0RRa|3!vxJNE1Zd-?|KuPWsq
zAaH5P)2%L|<qA(`au@7seSZPQO3zHO-ze(dG(Oe*>_;JQi))(IKuXvsGsUhqs;2N|
zr3F8t>cyBrZhO+_&a#JSQ(Oek(ZCc5{<DB$dztR$l3JGyH{g;WAC(Sa!#f;-t%J+z
zq)5kp{GZDhe~;M}n6zUSf4Jy%3)C6(@I11U-i!^if98}TkB@FwlBz5D@7pPB2(q^_
z80&v!TxnC)Bx2?r&m%qXG$OqLgEHbr1sWmz<VH6KfNX<Vv*2E^bHH+M4Hd}=pXTk%
z1zdMeRT9o=zQ-;e-!Lwsy}QAa=}`H}C=GDb4f*XaWTr2(ZFK{v03tnIju#nxqsivI
zl5cIi(&6!HA@9jfP2oH<X?HyST-6z1-a0QEfrHCOK6maYp1EZ=DEa>2$j$zDR{sAi
z8B9K)H+ewuMK)Njpve6Ts99~&EFH~=-$8RE>+I%gDgZAP&(?`N`KZ}Z4x~lnQ|&j5
z!fuMDl{Wa6k58$j%RNN!JII*G<j=8*3L~nF#t_i6Vd5+y=4h2Zzz|ptA3>vm8ih>B
zr5QxaKgHM-fLMT-_9vMMnZxFBEe(NG-BSX^I|vdJvu-(N!(hK3w@($MZj5c|?k@3Q
zawLpiXT8-5u}%x*Kf#SpRq?kxM0E0=obfR6pZw-wm-RIRsuY1~XO9fzC6N;sa-k-v
z9xLO2IAoEfO1^15jpMGFoSjUdn>>0U&jB+4OH@2NC}p`;n?B8r$g~Hipv8fBB~qzz
zx@+k-B#R#z<@x{;&HVhIVg}d%#e#9~)Q4RptH~}yPgAVS464k6ntG^l3i8207S5Z$
zm}|GMF^l*W4pm=QwHFxTPn-5}VKEtWr@J`gxkz1D)hx))^H3(N;GX}<6z;MTeo75s
z_fMu}>`$@-HlU<aBq~k=VCFj}zFN<h!WB90xgMi`#OFm>6>s-XZ4UZ=C{qBE;zph-
z?OiraSE=Cj-;*KY3`?L*+tV@Txc&XCJqd;deja~Y0}J`XO4=?y5bLRJ$~C}k{H5+c
z(`={jI=bW#Qu}I4MKrb4`?iZ4G+~RM%Erd0A~mrx;679Rpq=9^dB#5r+)RynMq}w|
zavTT=adUFT0vpNb=ge{L!a-2WaaWVZw1pN;g_g2lKf&+tao~4GOml~1naGgfvmlR@
zc7B6{A|A}!1yp*Q{YITVnDO(fVz-bp@o70wW=OIUrLOI)E1OX`1gBhcEc#odA1$)`
zN`G0&O8kHdJBV#X%}Xb_tD@nMS@T>_<*RIJ5kbKdPbY)+IacgZeEKv$f{DN7oRp@8
z3t^VtCWS~ZUmG48tAXbsWcKY@K^H1%X7)Fa`Bcu?fp^4p>fw%i`?TjVS#FxUDHZKu
z>!7GkNg)%Km8wqduU_dvW(MsJib+=k-v7U(0UQVGtICFG(5A(0u#0F}*|52>0#xw`
z(hcr0b)L}kEal1MX;F$YxIOcBfl+hD%s_N%7_Z{?-}b;7FFQF`-Lm{`d31L{RNhF|
zECf|k*2;xoPZD1KbA{j`|NPHP6i>_JaYJvdv`m9BmZ3!sSXn{-0<y9}{W}K>_3QtK
z9a$d&Il#|Xl@`(ni(12oQm*6IJP%fiWh|dsSzHX!TSGvst)P@^5~Y7F*&ER!JCOH3
zkIIf#=zaJzJ_CSempQv0UqKZy1*zb3{(hOP&jdo>tASNLto_On*f%voh9qmTS_%F8
z59t6~zV;usObOU>7W7%#I>`7zbNAs$S%9`P*sK#K;@(xn_g~JzUw?}ZIWuN{``As>
zvax!z;v?mBWrJD$xI$%vukD#>YSd!7$4b1C)W1!~3-a%C_4gto=lc~1CN-QlA8IN{
zAXcg9zm)#LF(Jl<?j<y_<i>3ShW~cqtLB%&g;>q=?-bP4=Hk>ArmP^NC<EXX^1ZsA
zT|8Bv%F^RQm_irz|2jym3)5^4y8mISv4C+HO6%_K%rB1$4}5vD+csEmk8uvTy&BD3
zn6c2IiOpZ|o$s<7H4XGIXh%hQ84EH;C*1_8NdFBtKs*->L@)RBYOg_4{a3;(h|%{_
zGrHDvnO$w9D$lBO78W%X%3?%r%<ZhB90lM<{ZCVp?VFE-?;@D^fXvtGS&Yr+|6*z0
zU8;uE<+wz4CO4bIWCI^hJTX(;b`O7xV-%_oEsyZ&i8jw@pxkQG^aL`zowA2Mx>khM
zjLgv;fSl}q%h>_Q2?U-0hny-O8UQ>v<0*X8V3wtG4hnjrJtXjriKoRmitCI7Kq`fE
zP*nWpa<Ww?7XtEPjJ*(LMCkIkyU2_bdCzW57APRAH08m%5<dU)$@1c<87d#j49%RS
zH9h;OBAXL20U>G2$;sJqzzeWtLCpq8Xj;(j@#Gkf5r0ef@c~(yTv=Igz*MqX)>qG7
znf)IeEdRp0|3A2-0z3BjNW-&|N=<UU$B5s5Yd>^+ikDd=1*3?_PtUvA>9crz@**bW
zAp&ABM%J@j`C_b0nM#{?{^wNyyW{@0_p5-tGyaFY3teLOUGv`av&g}6W>>{SX&|y2
zpVCPWa*g5&O6HNi$qqR<_7MCQQm`;*aG+<0=>d`vgGy3k3g&26X3)QsfcC^vOMLtf
zwC)4;FWqjQtxrVr8Q>zbT<s1al?^anfA=U`Xt@XObB&JL@ionkhO%JklU@os1A|hU
z1x5lhz!k<QNqMXMZLu-H35oiTV-9375AK((y>Q+v=4l#y8Y2Xp5~lB+0*DWq3U&d0
zY31S!bMu<IeE!}B_}<?bKe`z;PZ=HnJnT}gvHk5^TIZ+PfUx(UzU5NdTu}|rL8@|v
zdMY7Ba|uLvTAsARQke(NyXY)4Ehx~So&SN|2eKx?@GkZ4B6?EfzjR4Igf0^L+b*m?
zw}C3bBgkbF!MHgRna?Ec*ep{(6?qJb;)0$xZ=Nu_8XOkaBxjVFHjEInDu`2~#e1Xv
z$9K0b0!Q`9e;n0pVB;w$nVQ|?!9bm7z)*tW&uZ}rO5B|1qG2JyEW>1#kowMlF?xA3
zlS4vyNrNv}*TPy~b*l=Xp|=gnXT$~vJMoczZBJc4kjX3n=h5S>lCOQU+G;G-1>Jcg
zkCmkX<Z~HQ!?@ve977WI--g2uESCL07F%_R+W_Wh-p*KnGS6sw3U)>AzFjcUbhR`{
z)`>F9NF4SkVof6dFH*Jx2dHc5|7-7Bqnf(1aI95EQJAUiqE2`?<0aUsg4HgeJZ-hY
z1w_}hS0sUk`lyc}579s*Ax!DWPz$LiVw4Ctx?1%@0KpK#8*z#m)KI7a6JE+)bpb-g
zK!PEJ%(<6fm-A!R%->ma{)F?|XP<rc+26i<-vk39EPR|cxg9`sVqrnG_<+CnpWUI%
z*Nz$P`kiQ9L$5o|gSY!?$68(I5~$Aw9S{Ru<J=#eE{=7-nByMn9^>HO-EgG1?Rz;l
z1;B^eVa)hb3M=A4ZU5@z?GSDE-!oNsnY&WDBLF`Fvl?Gm-?A-onzZXf8w7T{?%9Do
z4<11{%tdtvgvPGRF+=^#ffn38CI{+g&MW%43hL*rH$T7NyE-Qmnxc>cuB5L-iEY!o
z2Q_c$9aqIHqWL{tQMHIRbU1m!i3W@7`()!n+50t#XulM}rn7IotYv+-@qee5CF){k
z@HsM#JcMORp(HT5@AvP%+$=33h5yKbs&+~>S|i|3R*Ci(d0`JN=LNqe+qbeK4=f@7
z@Jojf(^uuZuZWG>NhDw1b6!C9*~^q_JIp05l=HV=sa?SyR*5JMvfK{K8sAfWJf?`o
z0$yCD+IoZDt$J-?Cav;K#(!LEm)KdDg1!I>e9+@P2n=C=4kkhDSwfD^`Kw^`q9m>J
zsT&B~sB#m>ULtNI*TfA$s%FI8h%{rEtwvZuT+M&XM4Q&A`T?H_Sb6O!_YsdM!vyJN
zr=i7C!i-^0N7G3bhc+H9&!xC_-5`_uGL(m{Cj6+SR@ClgE<P!>HQ>QXQnm0fxK^g)
z*Wvj-fj?_d7#eLtYf_&z7ZR+}(Kyz#=~OTq2GrUe!e|DM4{V=c(MiM4bzRK*b*i;u
zm?q=shi+83iJp!yZjq9icf%+VS&xtb-41b|k4}205MdV+dM3vusguo-S+|0Y>PEB-
zJ)XB>Rj9C8&Yjy*Qhh_C4d;y4gV|JIJ6p7K-%g-o@R*>3#qm(|43k^Qy4qqtEhh*7
z%?uv=HxuJeMik+OP?#&+Fx3NQBSGk)w4O%&MdD(8`atL{O}UtG2D5skr!RtyP81S)
zGDLinx>$Zw`6brhQOs6`FgY+$cjklC+an{0T3rc70ydW&I3-1<@BSc;gI~b91_BSv
z&SU_7e_cC&W7}w1Qlz06abcLxA!$*^6S9N#Jxq%MpLrP9@IA-NJdBhy{;dt>eU_jD
z#e{%#*{&cGPvyW;k*yXH4lpf0;ir0{M9awK^OxH%zTdhHs&4Tm-%P<tP!eWn!;E?l
z<B-NX%utIs4_>ZtP-u$%Oh&7t4kmRn#DR@?dEaF<%hRkxN)#WKglL(T`*__^ozLOD
zD@{iprN1$D18D|CXz6HyG|-j~96Wg+oz9h=3HDa`AWA(~MBS?V!v4N}6eBe5eoxi0
zcJ-i`D94O<^B6Z7C}xGv^AVsup=(fa=)yVwxUtL>unWhpCA^<|^4a7{(p}u;`@*!d
zdJ?_LmJolUCkQ#910OB*mEQ4pSKq)-?EtoO1dkp_Ku<Vl;vW~tJrCWrTd=Hb?j{wC
zs}3HUd=8Ug2k8kmfJpL7nuAsSPzMi<mEjQ1_kuh|{oM31_fKG^8A^uEA9F(|{)h>y
z`<w7l!D>ctQ+jq%2rmEmR4_@tnz-#ke$pJqQ!t~mPLPbr7`(0+cT)jooO7EpSD=|0
z*-3&2TuYeYFG!Cr-Xhuake=45J`ox1$qPH3+jJ(eh?ol)cP6RizqW8fVftN$J8^;f
z4_JJ?I<0DEBxa_eQqJwD4!TJ~EM7il-2q|6JKqz&El{!3Fa|m0aCKfJ71uB979;uF
zOc@?r*$qutJ!m*4P6|0FX*YE3tQpLts~e*MJ*iUQD?ZVap<Sw*=wos$o$zPz;ru_m
zobNr@mddVHnVS1x9C%k%U8{+nIILvXw$}K1@g94N4*uu}F2yRwZzBp_P7`@~hIYR<
z4;fk6Q&)llr>*n^ivmAoN%SwpyV2m<Rz-Pz-Zi0VsqVAgm|VI)=2Ofg+>iv>57En$
z$mRVK>{fPYO=GIF+S(F4t%xacNicq;(cVN9gF`G^L;L=iDLp-*{&}#85?s!gxn<;B
zn(Mj0|4@o2_KYsomG^EFN{zf{>($C5U~)_EoIobmiq7Xnvm0BZ*sj!Q?!r!Y?<vJi
zN_DrTZ;;z}Vwoj!q$ccb8twHJ==Kjk1&)$iY;wHcAv<eXMM@1YWy}GSBia7+8~gH=
zn-@&*Mi=iI$}YA%h*hdg>?ngawRTW)sg06S{5vlR@nBxhrvhgOcCMw2&)UR;b`+w8
z-jn)1${br9w47Yz!XmqJ#5>@l+t2|h%8BHb911ByVc3!!L#s^n>5bPNl*t{PX2Zya
zp<ivbLpXOrNl$?924p+UY^8eE{nC>r=ErA`ixOo;X`K#>$==X{1&Z6CCL_(aS{2^m
zh8NjEza@M9>K6F#=kd3ER$vTZa@Y!Xz!6!0E!0x$D0M1M&`qIa5br|9b={RJg|J3$
z62nkDS106*1C3TJp5=i0XEpPd^4<c~I2^Z0iJmQH6Vx*SXZ$xZx%85(LZi)v>0-GC
zzPW1xNJY3I4=q;;LsyV2<#doyQ>wV8unUM8v?)1T-@jfZVwiu_XiLQneyVzLgRhD=
zJPFMtKolMdlDVzE3yTI_$4gn9>0Jf_aqfTeSIIXkNosG))>|m5w4v*zxEM25BJ4t9
zgQd?3Nq$jcc^!6UZELk7MlTHWNLTNP6QrMBcD>%HMM4UQDC{H*(IkiD&kIGm2kQu4
zN|C(qokr^gpEf}NqV!YU2eIE04Twji!2>N1Gt`kA{2Gx&Ka@(2w~*tLTkNb-ljGm{
tFFcM{{g40GC<)SN|KgYZJGnfuEq<`PBJmBuIwm<8{WxlS`A6|z{u6#gNVNa}

literal 0
HcmV?d00001

diff --git a/images/mayer_multiple/output_7_1.png b/images/mayer_multiple/output_7_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..6c77bbb4dea1b605c441358c8782d2d29e0173f6
GIT binary patch
literal 92772
zcmeFZXH=8h*Do4CKty2&6i_Kj?<k>HO8^skmyVQ3uSyTaMg#;6%}`Z(Z_-QR20<Wz
z(pyLb3DQgG^@go`|M&Zz4|j|+?z!i~InRhDdGf3>=Ui*e_M5!a)6uwa?#ej;0B`|v
z4{QJcoJ#}%XuQwTkndQXjd3Kuo$<W`F*-}WLe4rM$lvL_?wR=l0GENM|0rHZ<~WgW
z%0E)O|H#l2_Q?N%k0ZeS!6Pp>&qr>~cGvtIeSDogJ+4WLONxU;uQ@$><fR}X@t+68
zJ$)WZ@LU8_0|3_m5b$lIfV3r?f1RjtV)?qiRpmXVi)X0VUSFfJZ97l%hwqioSYu$v
zZjYQoZ%pj;TQv?^=BV$IFtywI;3wf9p4gi_Id}FP^mW>R6`rec#Y=g!bl|;0nlBzt
z8q69joeIXS6fP_n77fQ@!AM@nzh6Z)d;<)({(Xo1dElWz!GC=WsU@8F-|j$i{#&a*
zvkm@B*GS$6|3f!73XJ|+tG7nN|E<;kFccGiBJ{tuqKk*}{nu8{{=c10v<}$qGSJM-
zFV!z(K^L56wRe8!>W8u;VR$+Y{v#zpd<W<}&TY4h2!Mi{M7U#LA@FK#&w<`|9aT?0
zf3)Wmi1UOXSzaXJr!*sfx;mGjZ$49rZ=yM+0@k|l56Alpg}8^QtQngwjyiOGzxRvC
ze&N1foGdCvj56)EzLv0Iuw=9xIb6-bP-C8BGQ$VL2O{te#M!KNVMxCnGjNBxg>uq&
zyyBkKOT<is)xqdy8#ZM0ZR>UJn)jB;+cVEm^=>i0yLYDeErS<TXYS{j0c5CZmMd@@
zh+XM5I6;u0hze|VxRb!^ETh(Dg7L<v55%gaAS2b=a3B))W?a=H?2+bQ({z4gCb<ka
zT@%i4EGQ>g1Nhc15dK8$J_0XE;mTQR5OONql^vCC;}y<Bv!Oy;0^BB4EnSpQ!B>bQ
z;f|fLfmm(&@u`_5CGI3Z-_<!bq?9Z7^ir4qb}0Fm(-BzCiTow@eh2qk1#Hz)+_6+s
z2`B*_S@LiE3k8@Xi*X6cSXv$0Qh-OH8;{r{2?Rp+AuP7wL^)ineX#a)YUEyoek~Q}
zAWn)93GQRs04rF^H6DohrKW~Tl?YkDYB%;OKU<uXm_M~&4zGS(YyZtHk-wT|t6_i1
zVPV;TpgK1OqbrvwXDO%i(t!G4YA_1$SmuNRT`mrwZXY&2lHL@VJ;n-Lq+l$N;(Ci(
zO><D{KYzcA4GHD3yG!V_+WFn-=ikCzSdPLTVs1#pAowsc7@mj@K(OS=YMGdzv1Vp6
z%H}d`@{%OU4O&^qQjVSU8NcaNQy=FP5d@fhA7jn%*5O+JGb?&g+Trx7<5iyNOQ^ud
z9W%<&8*2lWVL3|Y`5-AbQT6BV%go+5WkbO}t>r06{jVi`17R5{D6TH7fB}>G2}fN(
zMsg`#hzfUx<n}dktJnjhVH+eWPgjb`xzYpBGHid;fWgN-Xv#_4%H(Rj|Ci_?=HOp@
zH2Y>ZmOgbpq?S_UtV1`kh{^_mIMCshBMZoNrfc7q|H^YIj&9+_ofV(gh*6r+9F2jp
z@JxUCq~EgP`YoF~lB6zdh;+Gtms)qOE0s+wVnT;slH3cdwQIlVuU<O7DMkd<BI;Zt
zXGBYiZHB+I{hmtp@2SvrW3$2|DUxg4YNXfNJHcdX+lnpB&><i6?-{;E(S_t{7y=}I
z+QlI5Q!15d3}|EK8rbgs>Lfb#*XbcejoK3-6eL_dls;htT4)QB=gEIPmm^%t@h>X!
zPnT9u@rfVIdi2}h6KFHZ^aGss^RciC%gjBUjtMi8F{2MqUSo`pE(`;%0|*hAqljRt
zt5l1!nv^3_-Hr(b)_pX+03Djp6x7I;e6L@^nU{&Ug33KV;CdX@(OtqMR>=V~f-QP^
za5G`Y$rbvf!(9mdSt6&2)w3Jx>96d^%57%_Jq!jY0{pW=YFsHzDd#Bj03DS0`$J5|
zEDNFXQHarJ8=j$p<;vxk9EC{}09k;k;dV4AJU)CzlJs!k<@PYQodyAd!Bb2a^qMVF
zj!=bB_kKvrj6KtmjEer?0|OT`C^eb+QLZq!sqfWFybtjZIX*%IS{T{mQ6oBhgv);b
z7pWvts`euidq%jXclkl8H@s{~2*ic(4yw4^wAxjuyr%?3DdjO(-ZP9*P6{OPQ{jb*
zon^WYLm*o@k7L+mTBT~~?Rt+oHo5ppRZd3sbkvBh!mxKI7Q{0R?m8&1Y}9DG3nH}v
zBk!A=q21a-X<2|1fJVLE`UwM-)eAxgvr(kgA&iaS$V)UJ6u^#tt3vXAiP#F)pkVx-
z<XWvQxa#iJ9URO*d$8PyiA#y9H@pP2pp46D3zZ=BVuBJDvE4#=2v;Z8ao}Zn7_fq&
zTyeHxPbJ|fgsMw^S(4C)iHlg(D@A#-+r<nCWo~68;H`I@+V4mDgsMjR9EOnJZ5&%#
zd8Q*K_Z5SVx`>SI(~1X!u-rGs>R>nBdRQo+oWm=)Ydw%^)_-YL@L&^&*?IUqO|p>a
z-y21mkHX8>W~J$R=BDZPQHAR6rTZGej(rrqovJ%#Bue~WSJ4kjHwYwjsX9&hJ=^fR
z$)6>ky)TvDjyx&FsmJ&;6M7P|E5_N5L}m4mG{<yx14{Pd9NxuyNi&Thsu9hCvSa&~
z54bGal^Zuwj$Z6eZZG<)daBGEMTpP@FqY+M%{4kkcBs1-=WuX6z55EZd{nszcgXPe
zqh_G>V4udXD-xtJ?g}w^`%|VhVKi$biWwF9fa`x)wIs(|3ck!bpkh70UT;Ma!c;eM
zc7{u}ea61~CfeU3c5ip3zO47<`_-A$jjw-33eusD+P#=LGHd7dcHF4h0hUs22ahkW
z-6qyszfH_>#G~DE+_wwde7_>>J<nQBWZ}`2t}+v%?tx{=v2w@z82a!&L*(U1W%bfC
z43srjZ)Uh#v<IBL-Fez7GGpBnKC-R6A9~j7-fUHd=N{FXf-R2Wqy)aIo6y#6zaO(j
zHFFpve3En|T#<&epV$ho)nod806*!Y-OlM-q=)wZTx45nJY*tNV_Z0zxGZX7sw^T`
z55CTP41rmRbvG!{?Ej_A>DV*ls6wR(2PBIwX%doQC$dV!etBewP2|;90Ab?%j^f-!
z0w3mzRmbe~-<KqPH`8Coo?(2BLC=J_j2UY{@3TJ(qU^Z3w6l%`F9oPq(5%@Kjxb!Q
zE`s#tr}jq&l})zM9ZN6nX`zHv`?8!#!PSuz3@Lnods=!1p&dW6G@xO?l%0X`0!>Xa
z2XJ|rfR&0`*NEgr4sP=gm!R6Yx;AS;kx3crF3c*%2CPsAH@!j1ZJHAp?mnvG%rTfN
z0It8u^~r<jk#HfxYv-UF`e>7^wAwIcpah$S8vt1G037lRX?rZ(^_2eUI3R$}i`>Qg
zg3;Qqa1`>4(g-NEFtRm>rQ9!W%jOIAs`}4bW3ZAWcMFZ>S@!cvsmM{4;t2yL&#+5C
zTJ<*f$a!9nE|pHc6#5$X@D&-146uSa=iEaHn_g^p<kF5c&|Aujd+kwB*f42~DS&b9
z01>co!}~lBw@fUe%%Ctr)x}^YRuilg<<oP>50!2Wbw3w!UXGtzChcJit$y$7>Vxvt
zjaPz<iDXUhp^91j80FGZggz4z9pt#3QX({hRgFP3KIWW!3vQ$0o)*B!3SA11tgjhm
z*p-_#U-)j-E<o0jJ(Rz-(!PpDfuJEN8$#W$2`*MIfcgFL6eqPDd=`qFe=?-ixN8qp
z>!23mKBn8f(#rR)q+DjL1+i-f7X8`x$e0zVOBt$GAUOP7J3<27_K@4|E$`b1DMQ=|
z+o*k8Dd+8ahzitaunpYCAv7?8PW+C1MKq^uI=`z1Uix}FY~dcbm<7m{8zLKr+-wy0
z@&X_F%D%yIb91Ug5J8t)`CPHpEVH-M>{-k!cPvqX)Lu?;xmywXOh-v3!xC5D_k}KL
zzHD6N?HLMBrc{ka1fL)4h)CSJ3K|c_5!Dh#gZFlCaHzL+N7zutamwV&ubd4L<34Vz
zALdF$j9!tu9561|fl5FG0Quy%dwCv_?h)D$qxq8U^bxE2$neN5Tf#F0TgR%b7?U>B
z;Y$oiW66cJtA77MmWH$nu!5I+3R>``!B7L|ae<u`Shkw4N@lDx|8HC0aM&nz->#cK
z4uN<qa7mXjlOjX$^ZG+`@MNlwec&e($|wY@VPU(vhSdr+yW|Ql^(zf<(;2y#PlZF<
zi-8#Z8Kpew&bSUhs{i1uvd~Le(2@T~#Ap=qV^ud>1OD9ZZLm_ndAO?_Ev3Y2IeJGO
zTv9oWNm<D2Fkl*Ks1eDb#0xUs#y)u}vbe*UTZO1&ZH_x3o71|H^S&=8nRkz08?Tud
z=`GLlwMGtFkgV1}KuJQVy)24@XvD7I+Jgq&aq0l+wR&%>Dz!Eu=39(P_a`l<!qTuO
z9!#^fv@KFQ`ycAXPaHJt-RMJVo`!xrE1mcgeRqttv7BNp-K=DM%qUOgVgL^g&t*eN
zb{@7@>ugTh<HFKvbo(zuf{naleZH6{%n!z8-j{lX2w-qLHz3n3bm>AEecA;}YD@ab
z7e4mRJOpbDLyU>GI}0vJ5{IWz9`~Tpq12%g2Z(1GJ!@24yJkBYWsx3*NRNm!Ns*v!
zp+5xf($C`Ssn-CuseuilOlvj?4%A|Q0N2k$E5}3V+D*N>Th<>!)oK(CdyN%)N&Ar$
zk!o!js6f(}+hy4hNkyd36lzhpOX^G5J6<*%xZqvGU1mdz$2eO3T)fJ=?tsRAEze+4
zlBu>P)G^mEf)rjPH1Nh#7c%IYY8RwTm;kPS#U5^bx$Wu%TvzWJBs8Z%A>u86)Ul1e
z_L&O3=#>}PpmKw|rVm@B9<xe~$wR3=m=Pw>Vo(^Hda}Xr#DK3(?D}I10s<D+<dG1N
z_wznMcXGcD{s`>$1~qqK6i+X*=-OEn@5Ah+uy(a;E+nw37&4~1dar<`R(kmW3&VT|
zR`|lKCsqllz`Nj(D?g2IuK<DGaag4+{PBLd;?@GP)ksMBY=Ae{M7nN)OaSLByalzv
zQF|#z7&fs7m+0D?zI3;31SU-=!dM9}t7aM5G{3tX#z810TN9|FSPtM$9=mRHsdp&t
z;p+`~ODGEWKm_`Ev&MPFKhJ1ix%~3XgTb=aP;@_AvR>)qnrx%QRkQVv>!XMOby#SQ
z<JFQBhTIIjhUr};#<~|9xF){YpbsTijukNPffbq^bHi_>s+`!>Q466eb9$w^y?JS=
zPkdCGS!aXl#Xg->1HB)>QY^_XbvE^HM<eBe06^A2*Jb=Aoj{^`n*!#*TxO4_Ty^a;
z_RuvGKK06nXAXk#*wptH0wv#kI6jP!%a}6k;Yvek9=y!q@ywgNVLfIOaU{6f{kfV|
zy-oZPNO?<e_#N-(UfgQ$nd1v`O-aQAS`x<+3E*cbrfO{!sJ$)mC+U2k78?dW?i%}J
za5<jB6VNhN>FGw@6pQ55FACXqBBUq4^EFReOGFZm&j&DjJu<w!Z^CVNw@c+DE&-fp
zJ;e#M!D7unHx@Vvr@<pEGNVoWl-E26%&5IQsOqAgaA(W&AW8gwLSc!b^nicu-XF8X
zBg?+`yj*>cVWgB$%l<jPyPeo7RQy+F-~ol_O2wW$s(w~(1~qd%;aPX0l?Ei>_#Q07
zF!%vjiE<a<mD$DbPjUQos?#6t%=AI!_Cer}djw_74v3TsRlQx%uxGL`0AB^~`L8Ov
zH4P-7x_PES7<8YrIi(U;fGHX-Jr^;z<<x+hHxfFO&$%)QOn?@Q^zA(1mp`UXXloze
zb5kU!3gK@)-DR6SfO&kd)?<=6amO5w4;E%xYZM&gYZ1Gq<IK>qJj>{i8z5ch`G?ws
zml#FC&c@~@CGODIs7x;2_Q0kTi%ieb4@(o><4k!xb{4RrhWV5Pd*@l#XW5qiO(0O%
z#G)FoLSgoLxW9p}INSmp66>RU?0CH0Vr9T|GL??XtBqN=>pC#MI)mo^5M+gZ?MMtI
z+qEYG+`YNHf_+;_u;4_zmoKv@^X^%%Td?EOk?53#+`e~dx-OSmJ!#3#BB$j+M&=X7
z{tUe!YQFae80_I)_bgb=^N0E^(JocoO(1TaF6=7f)nc;2BSjAj8W%tHKFy6Ch(ig&
zI?_9tMxr8tqB+7w68JMK=_sxR+4mQ(f%>|zaSp_cgwz`+It4y;dOgkh$9%Sn*h4Ri
zJ3_Z$8EGhi7>SQ9ZoMTUw>i#C%7W1Selzl*uq0F;Wd@z)m}Gkgw;lUHe$id+JN<cr
zLyufl&Q3~z7*P-LyLVwNX-M1PP|dKiR|r~$OI|p8ll>D}!@}s5G!(v&FM-m5=!n`2
z>Iz&<>4Mef@CNmyB<mNd4;?a%-UQ(-H3g}^U|m6^Yta5-S&RKfTE5%W##NaiwkvlE
zvd6mXn`_EMNqis0VR1yy`05kZI#XCni`Xeuaap{Zp;?Yqtg>`YRMCi6g7~M^v<|&g
zoEfON9_-9GlXj(?5Xe7|S93TPpDfi`Nk3HCwckv;^pFvHslGlIi>)DfC6@xG=-p5#
zMHtZnM9MqjJ)N%_YEK0^)U;FMw!eUQTP94JY&ik28#{+SF#P1<pk2KrV4T9(iPfb2
zx%ep|GT9{*FbM*kyeH2JT0M+C6!!9j+6DGd04pHqo7)do9&dPKWKetCd&285qBCl2
z+e*sWKRZOV)a&}%YpkXDUo1$KB;D$wqNv8xTs34W#JA#)w;KQfda#!AB)x`yHZCeo
z=+XOVv2cNr{jb8#9#3X7BQ#({5m&k#04h)o7K-#yaaL46VF>8?`5;!DJA=CEqhov-
zm4)<(!O$}U!LVBA<0A`0BqrGG!Y&w$Jph!PE}`z<KK|G@nRG7;>IUyXsN>OUM6+a-
z*c&fvR3HS_5-<76+Tzag9q$`!X}6hZU|$6!eewLA^%7>e{UHiD{T>pj{qEE^UUe8i
zjWH=ZFk+E!xlm--t%7?Y`i}AMjP72ET#U$*p_>|8jg}-eel&evq%EO*3Fxgj9v`bi
z%bAX<2KQQYV!OlhC~aKF-0+Ax8#LU-EtZD!4JvGDhg#cb3zTo;>!zhF<e{>fV?Dc_
zhdGKY7*9lHX7PrMAer&1PMKBQpfJY6e2oDaZ8cJ;SV^&J0O8#W-)mmU*brPuXAjAl
zY0^fcz-O#G1*3Z5+ZNemOEoAA@OvP%mA&wEe>oenbj-VTAe%|)=Nhc{E*aZ&6Ldls
zVDA|J%3@so@x`^>JtTw`K?;v_gq7LX<`IMIOr^hHnmXI$lyh>-y!I8V6c~O}=l#+5
zWV4VP>Q&z*2{%!JA!vy&*y!*){ZciTsjuE2_Ax20vs7d)(T5Fz!WfLG(R<B@ZiJac
z?V4Pr-P6FU*!^E2+5RQDqMT(N=ICjb>Syn%B&$q1J#Bj~tUOn_G;=~J(S@BIU~o8t
zr#^o0mK8?)mN4()YW+<sG5_*KV3GIYe6p$E*}FQ1c%#)Ien*8K>lH4bE?e$u;GqSK
zQw2*8B1I>-xHP+Q0Vf!_Oy;aFxam1JUL572&F62vV|aUGZ?Gsz{6WY|jfFn0h=&Xt
ze_+mO+szl5%Z#~xRpFVQ8f3nDa#ra*<kn?q{Mykzlyt;Bz23U5OqFCjyklG<#VzyB
zpZZTt=Q8L73ag|7l31<lW2DS1l~S!mZ3O=jX^5l<F3>hp&4&vwhNYtLSIaZqh28Nq
z+wMi1ika9DMK7nEK=ML3Du<gM0t5E1G3H{+XEDAL!t%oo#5aS?Ycg3%ZdG<Ma}LDI
zcVoQI%%-#?cSz0AJD27?Bzn@o^<h)@AfhE4jEjb_3vYA7s9okhK9wZ(wE9gRZin%$
z&2;gf^c(ZJomz)i{@w%Y$w@UmJly8v9S;V&_T%}ld|qm#3+SZO6>O#pFosPj-Qmgv
zxC|u>=ViEM&j%}c?La3IF-5cm+w!J{TR*GBQV-D&4oc-Rw&&po>)(@V(lwSEnYHg2
zTTo}8uzifeb)`hojx#Z4D*U`SxG*@SZ9Mkgfk<~tEW+U}{P4oy$?##5c&=?9_9Rzp
z57*t2wXLdVLm8xONh(&!>xIkw<qQ7wU)a06U3RMs+@=U%WY5kbUYTP~Br0=Hmr}KF
z%W~Vv5;78rw+j<Wb(;S?_?+O-s$V){G8(zTp*o`Q(R7f#y&FU-j6Fz8Vw9chdmy%G
zpGqZ34g|ID8aW;<V2(6?nnn-IrP9j2=x3jBAIh?kX}b$ZcPdO6<_tM>*R;T=w2e0l
zbrO7WK(>xnU$XY;2mkgys%HL7su>cJ4HR(B>e_#k+idf!B?*O2^ZSEG!RhsUMX)oH
z!DgVc(^q}o#(wKbQcdkcn8lyLXluqnVRltX?uO#_!y0|0?N$647rGl&$k!DDd=-Vs
zhWWZ%lePSv?@W3;=yY-v#RM26)!d^cOolx#1FK!bEu3>};O04d<qvY3JS)N+k7ih0
zs9SRXoO__0B14%msKHttl}pRWF&Og&OCMn`+!M$Cq5EmLH62QGccP#pQoqPAi0Ubv
zB`!<l?k5Wk2)~UPXow^r(qm;^btDHu4?PGQcm&$=)kC*HR^eXHQ^R*+y<^6;E}Bc@
zZb7EFAcog>@}VS-aE&;H_SYG-lezud`~24*IiLa+wz>$%G~}Rc{}abw8JnKiz1;B3
zM+Wnk61S>_wTcQI2N>Q_!A-xq#G_QRd8@W6HNkFEpXC;Gz{L=V&eE*<f-#P!pEozO
zq`&FeO5tu=3aTS^9AbzoOA-ym>mRJ%E1NGh-3`!4Tzc2+;;KsLxiCe+v1M2!e!(k~
zVn2|-9G0_;quLyZ^8~PN-bX96@?GwYawL!bOdr5KZM|7K8fXq1&YiJxa+%wVU_Jzq
zl%RiFNk~K?hG>?VkISxDi&gBCeTzi^Zm95?^=(>4dhGgo*FFc2-Aruz{b7C8Dne?O
z5rFm!RwNu`AtT>3rmJyz#)nW0B^6t?I?1rhFQ_ylYw+>9%T6)3hkEnSJ=ijTC6ncl
z{d~llbrV?+vG7VboSuHpew!&%zK+egG8}o<kgEWRbf=3|%aRMmW-L(yAkrMi_Rd~X
z4Niew0~c_u+#_fC3%04QwL)d0e>*?6U9K`K6cB1<W>dtV;9R}4#J_M!Fw&b{Pt$8G
z@V!Z-B_@eV!++4!oh$OwHmSpSI8!rtdI6In+*aL{$GGTIOsAMK)iIU<tT5|d{L9&w
zes|eG@x)EmA*{L$t4I;YQQ0?Vh}J3NyMlv))HXR2iLI%$oC@)%JPOt=C$hPko%Nt)
z@{}Qp99zETnii&L5f^*6as3l>_VLi#3k#`(Tu?YTa%Pi{dj+s$>8#>#{_b)3)TJk~
zlq<?F*2$d6t^oZ?<iR>Z(BIV4djs_ZjE7aPb5Ah=Z+>%v4fH7#)5wWcM_IKhH06|-
zN=s{xP??VhX?G|^*@ufnBJBTA{h7~|FCbZh|IT*TIrl2Qu>UX%*7Z}zf#^ftLeA2*
z9i{CCHahcO`-}8~kT1=f)tRzGwSFa1VMP$iW%K2IhfgL1*VroSmperieal-Z#OAQj
zxZG3`xR_CO<kcFsO7toqEdS$BZh2;(O5)Z-JtopYHIJU=;Oc&m`*ZVYf7XrxGY-`>
zNq5zS{LJyE!B(-}p^is`We=o4OmZKpdL_%UEEjit`f@*zc$aAgkFCG$?R_k`W2s5B
zE*}v%HE63*zmk_#_9cPib43p0)itSFo|HPVh&);2L=$7xR6hX6u3E=YI?bT4vpatM
z{uLa9n0xvPUYpX-9lF=}$Y_0IkYWG+dlPqO7GYnVCEvxYzT3Hwk)I8d9q%)K;#Naj
zq`fAf6P`4Tc@ovnii2aFn;>Q;W@vxMqm@maVkHI8$s`2K6E%GdT;D**-8fA}lEZ^#
zKyPjEBs*}d27oVlg!TXr3cH(1zwE7Ecpn#s=%JmL;N}E3pTh~zXcc>C_S+yj2JGr|
ztj#Q*f4gld{rRSr@X)!3Fxc`H+xzZl5rHGc^5u}AuD9)$#lI+5`-k%Mz;S_eqae1m
zgQRVAmKMiO@@Wq65Dw~@Ec_AUO#3y$9EsPHB=Ijmba#IGh(4ZPN!*g{hi_c@h<7Xf
zEQ!!m_-X&#{EoU<ga>kbs;HFt^W0)pbf5Ooq7PT(w;fUq@0$VIm>iqxu1J5kLVxjl
z<O8d@NPXoGU^CJ_$)v%2*jTGI3bD*H4IuO-wEf9QB|*(e=SJnGRlmZqMY<wa@Nr#;
z@Pc#El*(Gp)~0$PS5|WOp}t~<NA6Kd$VDsDCst#pw0}>*7As3h=Z^>Oq+pk-ZT#fh
zMNbcgNBdXH<5W&r;HRSOdA{?bNDL|DG-s_nS!k)Pej2cW_2i1V%Sst`X=}lz*nwpq
zj6~fV=h1Y06%T}NfzuccUrWMg=uq;dse~%R#wkxg9fNa6!Nqt{Zab}k>#lPC#r&+2
zLVsfcid7q1+g>Cd`62GpcstX=i_4oTTVxW6yw%O7=&P*(GslyIXz?BI{MET5{sUZn
z&hk&hOHT13!@@Nyna<maemBZB<F<mU42C*#D@leyZvL0ow$`7?pb>S<Ybu1xT0~2E
zo5Yh0Nz$JPxr?Jk8YC6qdTeN^F}EF@fJ!8KI>08R=9&j%^M=9~`9qZ4CvS5_e^&^6
zF5V|n4R>X-7aeU2e(Yh1rf}tlH{Vyx^!R4<DI;6FH}6!@w(U_F{Kb_Hn3(&VwrEtM
z1JOTxL=Sy)tXl>X9sV?&oGa(zu`>fJ$#W00X`~q%rE&(puM+}w=u^~gw-nPA%f48b
zY8uz>^<JKEr{hnJ84rX_$+>XRgC~0CdsQ2;O;H9yWyx<;WiL1kZ?Bta3TKop926f=
zel5B(?Mt-0t4x2tvBS?MW%KSm5pWxGgpDCmR3P!FZv1SSa(QK_rtOZ(K>)*+O9AbL
zD}PiyR(k9$Dz>@!BT;*LQqgtJU(91hPGw?0ZF6hdPBHVZWXx|wL{Dsu(Dz)V(Ye>C
zlM70(z8B3+q6{7^-XKT-y>A(Iy4<XTjZ?%pS;`(v>~H0cyA{AW2S#Os`0f?d*339r
zW?H!y>Z#TmPZVqCG!Z{iv8lm^g<pA4Duw2pSewt=nRQ&>;^WS^z<$n6?$vta8&n5X
zo@t7!;rLD?ZVdB{>cyFVChOpQGnK5Mv%{#wYe=VP4Jq#;T}1$ST|)X##@^?+vkwex
ziGdiS5v%0ji*@^&>~k3!YwD$|KrZU3*QhD4j12GQ9n;#6frjoigN3xU@E2KL_@C+)
zLv!Ate$7nnhm$7mcePbj)S1({LbQ^;1w*%*BU;J27(*5O@k9osd*z{RGq??}Z?}@?
z@>3e)8KDYb)#JS1aITJ`u3#f27*NJ0|8+`*=Fl{5OE(sw=s<ja8b6A~#3mATVZ@E=
z)2N&o=bP-B$c6TGMX#h%lJZ}?=8^oZTW*1iY29#@oqhb5G__&Yl!w0HLxFNu$F4rw
zfs9F`Dwef4%nERQ6MN|H|HcJSd)W|OU10m3H`d8-^4@A{lY8cO!AaSMZ!&5t4VcgN
zY*pK1d-WybFCb}#jalJnuf>93E-qlIUYSAAn%vU`J~9#^^MkV4&-2j!c_5}tl0>BF
z7=5+Qm&kd(>8XXPXf+MrOaGIkk>SFGDXl==7vI-;oeh3ZJ$V4%%5b{sh5xqG%SD#a
z-*PjI8t%ZX8A8~11Hi@fN@+|dt>Nix#ll~<F*kwh;n+jXxlb-Ci@NXei|tamFMP$C
z<0B?zVH4G2?lb#K!$rkAz82Nh%h6>yyrbIve9W9>zNkQ7a0p8PE-hC&d``C8vOA(L
zMzMx-n{vg<*+Qql3w`qLWwvk3;&oa{Qa@r>k}u*>DLkK-9Q$(@WHjB{D8U|*USA)I
z&8vzy;jHX?@XFrB4N_&#J`!E^C_PN9nJe_@a{Nh)lK-DGr>iObH(mtR%4|MfN)akq
zaEs;*%9IN4!NP4PMpzBUh2w-0wdM|%0|f~5J+4w_MH+1isIjDrQZeHfK3{$Np7%L|
zEw^jM&i7{d?YONOsMeD7@VPLKfw$567JU8W{fwv++j7wj^(d;Tfv*>#0=Kg%%^Z@v
z7qbn*kAA{qxZ+U*_sqxga+cLlV`^K2uM~b}^+asy2;&sE{d%y>K_8dZjw}z>fp!HP
zx^GZiZ0~KHhXb`$0^@aro1M<jRM)?qTDP3vJB1!C^dX&Tubt9W)W8q)Elbjl?ZHy(
z(}{!1`moF(k~^wiq+F<6f@Errljh!L4|}%oos)Tqbcn?Q4`P8f=*o%5qa$~Wi*o%&
z5KgJR0rxE!fzci%%gqgQZ0#gp!=6m@@fDd0K^C%0nK;FGafr#U;ocPcy?D3dH7Kc8
z0V??XW|-+iA_OBL+;d<ujLxn+wj_K3{z;y@|Hf@LZOFvq#*rWN6TD$mlv+V1$nm{F
zS7!2r3v_?`+?&c=qvbhAi}B~OT~A{loCzpJioWb-sG~)-$g4hh!;&bs32jY@ROZcF
zrl!Wjc)xOl`K!kLu{&Nc$D<UCpFP`ZsF?f1R}Y##7jD;6OoowJ1ov>Sa9TgG(wvZC
zj<)kQAD;6w1w_W)j7>?;H_hV8F`u(kd^Bp>(;>W=$q5XLUm|DXsTd30PCBtofzRYz
zItE-~<{~JZRjL=g*HS{>FK)`(a={rk6wB$WN%JhgiYYM98}jyG5&WH9i%NlpdpeE%
zI7iTAKKHO#k|cteiI`2)<^%$db<&$^94a2T(H9##D>{aUy3l>GnmRMt(%QmOXjA*~
zaZ1`KJAQ{uU_2_>Uw(>PwxX}3!d`Qtto=p?OKc`pHE%<Qj^t~#pY)guaJQWtY?I&Q
zr_HY>Z&pSbIs&~S;X@%Tb(J0a)q;#+CB-JqkL1@nfipji)BUL91hnr?RHIG>fa`c|
zIkw&C^BNyvm7lJg3<%be{DUkA=l2`1bTTQt+9RpqVix0!z1W&_H|e3UjI$v*{UM6B
zB$dWOzVc!jFTJ2U9Y@LqIm_7;`vthRDZ6kLBIz~5!IS6pHpg{2%bgr?TY?LQvbCSz
z%S)2%780$v*nzdHF2#tm3mse9n7)Q9uI*xNGr>!eq$>*tm@IC)bVz&ETL&T^<~!VN
zd`OD)biA(eI11a;5KX~tXSgubr#N8g8rIT(7-opmXdB0ngeN#qsOtHogU3v>P2=6n
z8cP;e?!%_e3?he{`E5p@Y+xSBrYDkSvP62&o-G6X^9N$?8M<UQCtKz3bR840k+RHl
z?1wo)nJg+~3$DKgk_jIa_D3~I*12<k=<a&a=i#gv8XhWMT)gU2*xvDIlvNaaw86Vz
zn5yEp@)6Bb$TtmOANVd;6g3`Pqfl-dIkQAFS>L~nMwt}}H;sytX*6ijH$yWlcQg}g
z-7>cE4~dNYz<Z3YCc`!Z-`Nffm@@glXC-b;XtYJmuVm^3o-Is{i@S2mu;Snul+?;?
zm#W|K`o)c5*A4#5-Xr4s<}dMMN6rU0nS23*uz;mHhK_knSjM?Dedkbf@E~W*Yq9_b
zf1hl**SJ@x*rAZ)f-zN4U%KAC_)T(*<?+E%R#Mx?Z{ZdB`P#DzuMq(^w&q2`(okWX
z?;34PhVc^Q9JXksMKnT#YU0&+A3MDV8jHOe!6V$`N-42omzFy|IUv&09pUMBvsai{
zo>T+p`A~#DXl}|`UgUUH@yB57>Xu?i8Ec^gz|1+=kacd!c=QcsPi%Rh(BPbxv*S^_
z_VbxH;5Kq7FlJmAElEnobS4s$v3vOXW6F@CEIk!dGxV7NR}K4nPxGzfSNru(`m!>L
zm5ZtUwN?0n)xpc3-jf~KxUZo-3Od#21{Wz#6gqE_>CS%cWPI{8@2@3n8IRpo9`|%f
zr)+YS1ZrDa7{Bh`Uz&N+wcp6d799oWw#z=#%^&gONGx^GV_|48*RVxwt&ZYg|EhMe
zCtFTj)bj?L(A=EmSK4=e=1-|9J0HHKV2(9rSA%CxL@*#fr$&v3**KFGpx@w4me=FK
z%`{%g5E@BWNs`ur@t4)MfZp1oXV@ko(CS;1zS)$@=QZ)dVjHT_`lrWuO1G)os!!#}
z23r|o6cr@>Syy`Cng4fs%;&_KJdL(2g)15hrpqaZ8Cku4UzHxMjqC5{-_IZU5nsc<
zwxX~26461?XqrL>335J;xeh;{pPp(11Xf30KlR~%LsPZ2z{RXUu6MkY?q4hCr7<87
zX}lBc=L#faX&S>E9FNEW0+oY&cmHI>+eegt+LbG;eFCcP!1GN-8<2Z>Yw3#}V(w*E
z%FF7HUrOfqkC35PfoIkZpcDHTTJ?D)Zcd6NvH^4NsH{+u1S7PHr8bFi*Ra)`AnyBT
z9Fv3dq!-I!6fOs$L3)}Kh1049rXp~Au#!FUlM{F7NdNW|eFYEUOteRW%8O0DAXD4s
z3uI`EG+jt#5kQipKu}I30$P}W7XH#1gON3^8>X?NJI9!2^`)%jasTaD?Mn{0=`o=G
z0bHewtYX(}#8N{(wq_qhNHLah7&%+h$|;dy5}!`DLN@j1Us%U0<a{_Bd?+$Y96rgK
z{G+s<f}EDjP<TUF;JgCF#~}nB-F!ZO4p<?!HVGx^(%j1~F~-u_&cE}f6YOl^;vhcP
z-on-Vs3IqFW}j}N$RP(bIh@MPy?rGceTjRToX(02bmwCE5sE@}N7T6Z_eCN6&ky*6
zmjEHJ_5)W1$j5<K{vP~wl2CrcD!2_}*sR`m1!EdvPMx>6jr#o4h|mC?pclCOMYIbW
zU!Xg|f%!zH@0iORTznN+7K^BTjH@MXX)X*wZLR=!IC2LN0f04a+cZYGtwGDkW8Iv+
z(!?!|heR#~hFNq3NZ9?Ve($oPpAtXNI}b6c%AGOx7;pYHHpg0d=Q;4^;L}UMbp!P<
zz2aOy?H10hb4x{)WekSZRrS_NXY%<PZ0JIq5-<vDFai=fVS>>Ide@fHzCr}6wRP}J
zzqi^EJ5>@<zp#aa6^Q%&li+|rFgrt<2E=Z{AtRZb_R|Y$8-1E)17IWrtwK3~5?8CE
zG<>rbED{Q@HJ_JM-j9nY5^+BP0)yk~#~19jkci#;;K^J(E7|ZiMAD4pyZKzdVRVOk
zZ9b_+SK82bhcAk(wheNF*dw1e*&H*jWhHHgMI&~k2NJe(ATlcFM$0QZeF&G1BJ=Ad
zNs<J-h0GWy89}pOO34CDrHrJAa~@mbLG2ZV9$222B+(Ee6NxyU=_BpxQsc3r2wxHR
zML?!|i@k^J!YyD09Nfgb%gjgN*{y#tXe$Yd;1TI@1!#cV{GM)(8GTfKfs~UMqh+Xr
zmz<mWDoF|=><#9*PciL205=sv7Os$m#sX|qZ+j@nh-pqT$g!k6UOgyVejicBh<wck
zbd_SHJRekiVR|>Ih78xCvu6*!KK>eMDgVW@NL!HcTD6r#3lGSZkxL5<I@Yw1spncF
zi(3aA^QcdWdpHKsq1HAMF%X^%aE<FfJ?x~k|8|g4^cOYSRC}wm0j~NU7v0WJzA^ah
zq}U~edMZa75Cl7Z^!1j2Z7_i+Vg6fOo)OO;g_jHH$ft-OHSd^^Dqo!{JO{!IzDEUG
z#g`uxbN09z=9-6@g46DyGPceU_M!fBFqk`Z;PXmfLf2iz;*TCSgg3zTRg7#(+Y5`A
zhy$vVJF9)RjuPEikry$mAdH2e@pEFsg<0a<L4^O1@Ov3U%M~3|N}z{v8Bc_AWX!S&
z%LaMFTn0)|9<%22wKvlQxPhC1yC&T&70NlnBu(PwZ5(P$CZ}+p<V`5Ra3*eZMA%3T
z!BmY?D^<8<vX>e?vqJKQ4r8~$!=IPB6AF8q51DfmX6O<MbCUrh1}G*KGI%rTY@Q%U
z!ktq}E_R=a_kgnR6+%~^lYQ6GP^{j#Dr_na*|Z_GVEj$WZ1ug-yVXQQmHb*fB0$8;
z7PehnIH}~e?nO`y*Vr8WA%X!l(D{d=0)u?ncj7G{Utin0`Pth5+(aYyuCmiydt#Ai
z8Xa+WzvS%|panpYN*9oEUhb%hNV?he^j0%xY=iPz7GhVI&{~ZWNQoTIIMZ~gt|N_D
zTGmO1<qA+RE`4jd={QbO)#J=ph&IqCy@2}X!HD`+pJnb6Tjo-!$dZqYV1yA@gPH}&
z`1%Y;!F{x|=7}oR)QEmDPbc;+_cU5?gq&Hr2T2R{t-{GllKPG|Ie-=vjPI5vbTQc>
z<c%xF{mb4q@VrWKA|2t4L>jF(V2ic8WkgJ<IEO?5BkFAuf&;$RP{_ObD5O#?#VqVQ
zJR&qVKG+OqEp4l`O11q#rDPknDck&=E&)7d{OuJcS_tx|h-pIuEnnaQIb6rG%fq+s
z54WZ7dR66q@be@KwO9S%X*U+F-nJ+N>1i&eI}};#)2965DM-gzMB)EV%en4qb3E_K
z`KEVcEt>RxXg4P3Bm<QL^PCKKA>)@OOIO1vkMBSZaqOdXb-G-w@JQsXY@j1?gNz~a
zlJwf?g1H(p741YLxBz*Eg--Z;dy^|V5MoLh)AB54fDFSHx3dX1;cXZ}3VKi})tC{h
zej=e1lB6H7g{8astGbH@xk{mzW-T{(x>}{l_!WWSY)cKdd-BP6R*^To+1NvEa0oC!
z$1(n~;7}KSkT>8cX5&V;PuI?OmIUy!`PR6nVJQWFlF_)_>tiG|*^&5a<{URjl?|Bs
zE#nKCny-FG8R#wU#Wb6;%mn=~*)Zv&ghL_Il<`Y+{rE3fL!N1P0&Q228RIBdfF?ME
zbyv&_IJ-Z?fKe9eoM~=h4b=|xDyYitZ$D0O9Zo4paa5r-6ke5(vH5^a*hokmre1>*
z5`pW61l9z+)`$V;J6zO8ZTR?89uZ3#rQo!=j_ZdYK5{-nY{KnA$i-bJa2f>le45*1
zPqAZv(nkds<pZz$#AbgAV9)LPr7=@7?89YY#7=Ui^WGwW(IrX0={kt<>LxioACHg=
zHvsD2?ZTeCLL?6a?+;%Py!8FWx>%Rdmn?<X<br{uNiyj3Cv3jM5kp2<>4U{~eC6oj
zWIM7<y{#o-A=%yVR&1qb+AWAyT5W&DemshmZ1;q(QusTMJ)&DHhg}{_Oqd}iT(@o%
zbRbIi>}~w9@N%Z833~_wD>29=?1}if-`aTzPAN_{<P@wk?Ur#h<lGo_4-XB46l5sX
zcJ1<pyxREM$zfgt39?aqEDXZ`0ZcYv(wI|RoqT!BY5C^GIt=PKQDc83li>2|NJrvC
zgWw(W-P#crT7r6UNG@w=Mu<{}-a|P%mZqmT>W%1^l7dYX(vx5&FIe@Mzz2J;g8Ua4
zUCH|iQuM4Qm5Th1%wDmAqKHSKsQLp+9l_;4UtvTfNgO<z>wTC!g{X&#<diwUTGKIw
zc%A)DJt|-zpTiMmz{IO_RlaaK#C8XJ_*mm;tMc7WH};S>$K{(`p?)Y$fELUe^rFSo
z85PJ-PM-s{8tJ9vOhe8548HALC~-nHtEhDewP!^*j(He1c%Gccj}R~@9D6*=Ya54c
z`p~L!kc5)egE2YzhQKmzz7Lq9U%FjzUR+xZW~~8XQATHphtVo=gZh^9(bgIC>^Iww
zSw9jFkJt!<u?<a>0d6qP5&f|AuIUr~LRTA=<$zNv&3?xWUyb-cb8;9d<7j1iuyC>&
zKXAd-dX5rL+inb56SR#apDC_W<W#;q$VLkS+r{f}Tc|k_4+ICgj~_qtD9zJ~-ox@=
z&mse#ObPE3ga$OcmC7|?#|}T<nTx(Sk_0i*`T=5>^1JB;2W(refyih}-5xR+&M-F_
z&XZ#>n4e^ro6%weJI;N2%<>$2D986;L$7^Dxt`!iG{f8tPd4bKm|63MI{u_K;|Qww
zg2fvOM}>N}l_pVfec(;8lJl|6+GAU0oCpaqhyXYcZ(?K;gqj=W(0z|VYwpm;yL~b5
z!^9}HxwphI`6K-{)SROsx-sN<oi^kd?y4L-&0fq%+hFys(4`cOb!OW<tWxNYckQAe
z^ey?!P~^I0hZSSU^NpQMlvep^iP`>d%lyJa`PmmMvkBmr146R>Xv!wW06NE@?-P#1
zBT3Q_Di914M|8nZf$A`oG|A;TQg|Q5!?@99X$-4iVjgEa3e8c0K8Z)+u~=HYT>B8|
zS=eqTxpb=N=6s$6YR{Og-l-CHN^E2ew%;sE`kv%AIE_?TAyL7UeR)PQSK6EtQaj_Y
zSXDm`i=I1p2@oj_8zSo1Fzrd%YLs!Loa3{IlO#oRyvQT_-eiCfgPR&erNIf|QToAX
z0w)8qd4rnCH*b9BNF*air3kTP+<F`tecG38c~um#df?rSoz-Sqv5H`&6fh`#|KhRo
zEIgEF_Im;ch!l=(3ix3wjl?zq0xr(_a&BH(g+UWZ=x(g3^d+F_eVx6@V6k)?&*Imp
zuy@vcM|MzyJ+LE@yt*YYR)QUve27o7!7+yfEg9Jq*Nv4pSc@E&8tG@u<|Spy61v0A
zE0HlHWU#}sQXOZ?LIRmb69$}T13A{z2zd$Tx<_t|@AN-OMkT|D0}%qCT`KQ~fCD|w
z$<%vWL~#e=9;%+vOO}Ac$n*FGQmh4b`4h-<M)lysVEa26Wcdy9w|ix3oNn`v_te^8
zn5Tj-BX<DgA`*t;5m8sYTx`n|q!B3?dZ{9lO2C0=Y<Sk<E`nULp&Wq|x+GFdIZQj8
z&6}1^fB3=xK3mR@axfrc#6R@TM9}JOf7TAK7L<J<@t*hN;Tbu0IcN5lr9~--I%UTw
zY@}W<g$2b`ibXx9pQRsoSr0RiN(VUX;rycyLu7wNq1?4<>7Kl`4D~QIWnB_7Ac$p+
z9-6q2MRxdZey*Y|w~EX41(9BR#}te)0sR4-w+sAgIrzRmG9<`g780@#L{ev!pwpAi
z7$HpO%((1OI^zDFz`6zfo|7rM(YHPv-W;<E>x>CP$!OY;hcv?R$FC4|0$Y*yr!r9|
zN}T~N<iL3X>7W-goG>%*m#|7CB-zjVK>G)sdp1M}bEtZ#z<{$<k92#M&`6}BA=9RZ
z#$L0R^z8n$RM+G0vMd-k4suQ6U?^4d4N$VNwJ%p!3_at!5M4ZHo!D&9i9uKWdItuW
zfZvUt_BR1f#%)~neb}|NVHJKsXl`KkaCvW2)I&b=_|K?gvS`d>eg&z?LgIzEfZG})
z!V083Q&BoOV$Q@1@xPLq08Mh?(_c@JAL<uHAf&^aMSAeE?|)TriuwJkD)W*qwM@xJ
z4ynJw3*<JrpML4a(`Ar7J)5|d46s2tTs!?rt}eCxPg$t>tXkXCa4V6X%wFsw|K+&9
z9_W|)S5fGh)%;@g<=wBmRHxxXUPubduK?od=YC-_Xq{rTr2lXhs}$_>?N<S%cHe&r
zEKh+m+;((p288e~epxEdRLkE5umiu54yWCi4{}edVfIBz8ZMOcmD75SBqfImgLFCU
zoyb3uU!dQ<@#(jZqs=e3e|)nRr#yB)U-4_ep5lMk&^ot^(k8)(p9dIz4dsmF2IX-)
z1%oYVi)nUg6;693zYIn^`fQ)elmbdQ3`LPrbg3r8!J+ZwpI9<qMWlj!m&FyxId;%>
zZV$^!ht7ps^wDx2N7P?U$wa-W|M?;MeoM$+(?m(lxJxnmc)z%wn8J)G^d8XmT8D12
z$V{=B*uZ^eG|dyl>j+OW7DsHt;KODQ)9EyL$m2VwrJ#c+6Hu;?zGXskQs{u*$GNE<
z4fA_*nX1crzlVWDW0NdAI_2lmNV1E=_w6spOnTj237J!@>K1QYOUP4tj`PQ-Bi_1s
zsU8U}HIc+lj^Z;hI*QsdG>~YWWy|e(-LRA#&14-Jwo{VrtH)re-k+#hEnlNjAAie?
zYxujYH(PCasj4r-?<`VEpM~X=K~n#Pr@wjjf9sNet)0LA|JN4&b@zWmne_kC>1>{i
zHI6t<ENmzd`l4yUf6!jJXk>IPeDHvp26E{VSZnCJ*XZkah!1Zqs>i{Gt}ZbQ_Q({I
zi8oTiU_Cu5gTbK4{C8m2>MPF++n&5R%S}^H=M!1WwH+(nA>Y?hQolWwhWCuwnmsDp
z>wQ?}tD<7BLoVnCAPmT9_`ffGVRC)`@9P>g=kJdI0Jz}dKVQNH6Dj}s@<$H&j|u#J
zeK0!x3j@EdTM1B#zd!%K4RWz5=i1*`fZuKY*ADo9=paZpvsE(b_6(|fYn^_Xq$j@o
z6FrB!s(*{KY6kKzng3hN@BSwx`+q2-|7SPC|DWvQ9joM&Eh!A<4|GD(@{TEYDkCRf
z|5)HP<=Zp&4|}Lw*<)b59RK6H*{-6$rjIMq6>0hQN7KsjcLj$A4WtIY{55*?Q7oiX
z(+7e)8NjjjDKh__J^)aDOf(QKDSV1llJ5jULd83elDvByAu6dqI3T?XD;ZKhVvL61
z@rUZjIo`3bt2WKmg~tk=HoWp#jh?AD29kI+1ojen<NEHrd9>8X%D30e>$$wro!52d
zpAp<d6}77C3ONmp!;FmaP@mgKRqc%>(JZ}&pP`UYq0Wx@kgCGaz?AGNePjQELOjGr
zABpd&q0i~OQ$a%T=JD^j7%O{xu#DuBm!Mbs)>`%I&isd?_$Hao+QJ3A@sL&Sm;W4v
zZlduC(4vJ}WQcrc5pqfPX(BDJd`Y2BQDJbeEndrbXa&-%wb9PoSZw$CZhS~Xp|!${
zrG%!B4l=TMdcz8N_y8GaJfsXm(f>35(SY0_gLxh1<)c~NhZ0G0vd;<|f>|<hgMM5w
zb%$&S7;kM<=Rpe>E<uhpk<}{BoB5&_HT|uEh5zZnHY=sjs6p&Rmdv+6Yh(w1;4Ne$
zKUt1X<JmKRgz?V5S>#Imfh`;l+{o@GZS$(i`fTU$N?`vSFkek^7~zf6+F0PlU#%zh
z7(X5~r!*cKKpw7tSs#Lgs&-z7_;uf**o}Z>#6QnVVF<KD)|T-7OS~2Cr~$Zm_no~B
z91i~di@yiN$DIY5Ctx~sH@-=Lo)25Xk@CYBVy$|6*s-W^rgCj_umtz!M{RslPGeXl
zd$1;w6rW~e9EN<3|M9k@6nDL|v2fuN4&Eayi4*DUgt!oP4I#1fD_m~UXx>g|2$|FN
z@o8_LGcD;V)D)_W<MK+S{xzL-?W62|PlxX>vl$mw@Du9qH}f9tj8y4y9{u=YYHe)H
zn0C9}xS$Xxc`Wrc*>5McRyg698F!S$+rcjnC+w;~tZTcRJnvO@U=AmFb-4r+Jkl4H
zA+fz{@pdQn#>)*|mj7@aLnC8cu!-cAxVN#k5~}NuuSwy<!F`cos2|f2GmU)m9z%2^
zD<kns@;Ab4{Lzo)=^GAWc?Ed)dcPhQj=S<j6fPV?td+brTY3A;4waMR7h566B+R!X
z<Y5<7@;@3gwUsxg|1<ZzP;^db^~(MOWSFYYVR7|}&8x!k>z&;+u^yTmU5QHzD~q7R
z9h`KhrU~0JAH?s<TXaFAgvIyL5x)8$+v8MYW0A>!tz$1k<$_1UQ<*<Bs7M_qL%k%1
z_#X(~U`A};&y{G?iq((2$qd-c+DXc}>75K&uSfPsRSV~LpCwl_8QGO@bTXSZF#f}d
z(NP(-f{%@!(2d!MENJBi)3&}Z60ulav!2S87O~C(DvbkP`2dtFgZ4&v#9m}$@Xwc7
z-8Z*)*qgiH|8#<(sTP9fM2mBwR^>-7(q~BCXuR1u<~q9f=FOgdcT=d~K+U#zK_mLj
zj|nP`?w)_`f2Py|BZZ``_P`hUg$Dv-o`j!aemgy`_mB2?WiMcp4+{4LVz7N?R%u*U
zg}YX4jCU+%tp6F-mTF8Zr$3Ve73NtggzSjLqCY0(;NY%E5=|^;Gxk#4<+pHjql6W_
zkas3*#K-qva>3*ZCYlU2Bb7o3KO1;e-}<I$#9+}XPjPs*x6As*JGfyr<7$6oaL-Lc
z$z#QTMg*jOB{$@km+4H3*hArPzP*+BFNYs4FSU3*-?3Z>Q3m~Mj4v-F$YWT=9}a(6
zr=3GNepBHgt@hgPIP#F-$#ou?Z}vnb;i{d>RHU4LFPYP3{DF=yG~-WR|A3QTA$jZj
zz8oHaM&WzsR_h1%Mq3;gR}_(!6;Fcy*7yMOyk3Zt2WyRQ^82xS0S6yYB8fJPV3u-M
zS^{Yr%d9>rC;O_@c;>v=7EPr4hn!JxO|gIYP2<Nqym<m+&V*eN$dm#etXeaHS=zt6
z+~PsKW7)fxP@mMbX;|oDp{vup1Ks%#g;iD8GmXrR6My{RrM<w8Py4ZlgCB$#9*ydC
zZY7bMKXwQm;VZq`m7taCTA=mMs2KiA{Y<0%14xFl&-TX3@wif}n`1}Bm^)}(*+iJ5
zVa}&F4_j@No7s;Gew8k!^7&C;;>v&449vb&x!_}AsB-6O414Av&NKC428I85tc!YM
z$FcH@mHIT2kgwQ(p;E?(+#nXRVWSbNiu^G`WP6DHDl%(&IM@rc*NwGWklwbRGbEY|
z|LZV%>GsyZ+4%oq@6F?(-rxV>&goRAw47EdOC3tqC`*>1T?ko*5RxSOGS(SOd#4gL
zLbg#DJ42R<u~b5`Wf^1)6`2giI>zq4-qpe9e13oY?)&lm-uL(SxIg~r(do?l^?tps
z*R?#K&+7sTRVA{sZahM$w3(iRjhQj&kddO$0zHQg9V&Ib*MH~GzNUXZ_7^<H_qZo;
zy{hpTmpY+=zMJ^goBhgxymOGBcG$iCYjl^@24nxmbD95o>xJP^n_SgZo?K>J?<NUN
z-DZ{em!#zPy}$l(Bif^W=Rl{`Wq<0H)(zGg_*ml;VX-n#{$bQiVq>b}k*Z4kfTGJR
zTS(1=BRi&*moOAm!-;&&%H|7~bfsU8NVN-JubSfiulL~uk76(5OWzNRwI4j|cG+v;
zVo6HaR~mS#O%~3fvP;61e!6Dq4jlgMN07l?p{bMj=SCF)N>!Wbep9~s5;S?W-ZkLk
zzpTum<+xuTixYyINqYYzaAD)J`<LM$Mta(#ug*fb_^R(%N8CL{Wm@cvP_<i45kN^%
zo>81TT29gZPTbKXETpG;M(+7%7+%mMqVt^LN!$kBfP+;bl$kF+S4?h?;o*iB81}{{
zXpYY4tLyOxB{AjPtFI{SQ}Cuf-!VJ;YM?^M;$M?UgeiCj|D&D(NTIwIk8dCEyFK<k
z#{{Vd^&5Ek&Y@@0E!p~~dt)#3vSW8e2;|)tSI4Ion&K-r@y@ZoAXxn8SBH8H5dwl!
zd%Vvz_IkHi2=7Naca30mps)L7|M6X(8_jU`_)?Qzw#~i*zX)7-|4wz5oRq-K%wkMk
z)3O@{+YCOwo#e<b<>IS_k1w2=O~KT(;X*p3r532ECL_)!=$ZXK%vTM@bic1|p6;CO
z>MGJM<Bw7m9f%P-<|<wdOo-3U{k^9ZuVUUM!1cDHxK35xUC+B9Ri#gvaZap}0uK|f
zQd=D~q2o3nYB=wNr>w78@V$+zC|M}n&QeFC_EvpfsU|SM8X~z1EH0Ootz`&TPZ3r)
zTou1Mh@>-`%bNdUz!9i&_eHZ`F`ClOa}pm<2^dicV(;89<Ga@cl@LCEVdncvRmrdE
zRWqP|Kvf(XjO=j~tX#0X9kICZHLZ(yI67T|>-M`7Yl585j0UpiJGOBM;pMXeJV)9O
zSKZ}~Z7kG6vy;V(G)h(nS$w1yzd7qAG({GybT$*uk_@$FKps=L`f0LwO!C*2iz|<1
zTgMd#g!SALw29SnJCEj*i^s*#=#09nts2G3RU!BeOxo05^^yzdg;Yv?`*d7tyGRvj
z-&C0pW#a%UgHnG%>lthQ96IK44>wt{U7@Nnjk!U6Uf*ls>Ymnmo6lWN-}+t{?e{6S
zTb9RncNq{H{m8DK&0lN`&0{Z{X+UWLqpEU|V!EbNNHhBsa#AHVM*D=qtv_=*RjVGW
zvt(aPIETzKcf}VUsS*s~&7%5h?(ioW8B&vW40&L75HhDHg*NT}=!L^~7tT=ZYnEYx
zbNZng<lXeS$EjFde8^~1dE@2hCkq!2wa%I00Y-V)*SHUUNn@$Te7-T{q)y1I@gNH;
zlu%Sl!j>iG3OR!tmv0x&^a$x_j^30_pSp}cL78dXX+5otKM@jLd`H#@NlgA6@#O~%
z*Xs*(=NPMR)EG@tce?PJ$i2_!n3pP%(yFJy_ul;!9*z>*V=DG?{Eql6-x?2&uua6W
z8uT>(Y<4asR;Ii&kIC}d@58t?LpnQ3N@H?f3}dH+g?&!qi$gk4TqJzy{;J#R^IrJi
zeF_xH&{HXWVTFR!&d{E}#Fcf~kA%L{A2a-g*q%lqhRx@DDd4(kiOboBG7tc=*Mr+N
zAv;uezq#bf4tcQT*w+Fs-DNLjsTHrL^_S@=N0&(_EB-6glAqrBABE(nA^s;QSpT_`
z75{exzyFx)Kj!-B11tXP<>dc)-i3FF;A}`s#m_q6PvCirCVkH0DVu!f*z~!=DWWMM
z6<f>e>}Bp1;9X!rkfL>YaVY!SH~ej&73n!ufLYnkW*0`k?N#TGy39!#Qp%RE88H_Y
zHug-BX|kR<yJcBrc^ML!3t$qMIZeOQO37sgyJm5%6|7m$r``+NTZg*%eJ>R+L7f#|
z{`%n>#!+c_-;2d3mOk(0Z`v}_N!~Yg8h>zK!QLv(R_W^d?q0QS7U*JyD(8*7#T1G&
z;kAT$8+|DHHUHv+pS11FiH%D5S;rmg5*8+>?7X_33ZdBv4sm*Y;%I7+Izise_lt7Q
z;sYyaq!dl!kk)94P(Tx1HFDBCih4Mt19x$r$WjUDv|Rk)wZSHUrL6hcCo@Ar+B^}Z
zewj-xjBA7K8~gb#;tbI^VFc-JP}h#Z@Gd@kQ7O@&uq3E{xb*c-`dpcU%_4qYk#D6z
z>`zD)t>RJ((uua&vG~jiVj~;ljh|cH%k9Z27~8mHGl#Ax8m;wf1u;4zOMt3ZJH?&P
z03<;I%EjFjAHv{Wvi=^^7Nu97%r+sVO*N`5^l%-Q(z|GC8lE@o@`4afs6Y*GA95k0
zXN`JEWKV|>L;K0n_S#;-yDHZTj*)fTON|0+IOk`Tk5qNz;TyG$sCPq(QYfnOx=ZH2
zc=MxW@k$5hRAz%ZKgwmiSV$li6TLUICqiAt6m4TjCG_+CzIeo^1mV!St~PxQ9Ttn$
z8+qAQM{AUtigoge&1tbldrZj%4HdU6BL+;7&^lf8V02G`V5P8@cfTgySE8y*c*+vL
zJEY@q6~FpMZMRQynV{tvquJ@K`L<g-7f*7{Bf&t?qQ35U1jatLkHux<9czmBJy4~F
z_bs9<Sejax`<Bl=2q_Y;@>^Z<!V2G%#(cyoxtRC75Jz&n{8G7!kVan?8@0%FD`DQe
zn|-^pvzw|u!D7WzW_pP0i0X~>t|6BE{*S6vDS{R>*(wqA*M;?qXYN=l@_U^+dN8VI
z_<Fp3NVIg?j5|lbgR?!YXrhqi?}V2R^8NZrsNT%u$l@~>%k)HJV!8LFT%Ssz`*jzU
zIcRzTw(JwEBt2c`F&9;yXfQ60gpO1O^OpJr=^B3N5-PB~IG>!Vcy+lW+kzP@^+N_w
zQmQ0Msq&CibpyRib2R92w;&^8$)lbU^!j{@u3q;4azQ`6Lk)Oxz3kHel~N!!-a}LQ
z)d>nN?if6$Mxjo_&+~Fg*9HB>eXj`77mew3{jLv|Qb>_rPeZ@4`tco7F5`#C>XyWy
zvzjUHSkTRMqV4ZfDOWD$k990x1P*@=4a)BGm1^9=3b+zmLr2c_j}I|Ai<i6NtVW8v
ziIt8BDusf}R5yH{>Z$pn`c(t^pKtjSooA1?S{=kXGa95&X`E3RTyv0Mv*eK%u^PlC
z1;r|-m18-DGmSoXjfqLYHKj7GOMS>!Jx?@}zF&Mn;(20fyBHF=^@Qd30E=u*PF;C;
zmN_Yg_q|%@YS=HK?z!`6Kry9MZb)j;KCg1~(-2|70*xA;8V_`qop{oyUpUjUvr9}}
z4?R=o(%K)wd)RF@%$U9u%xy>Dp+02;YM4;HfP#wc8`IL#)VkEvGDFI<RL_i5eLP5N
zVvKzXK6%S%`a}EbX8GjQZoKl<d!&ab>#?<Hb|jJ2Krb!*a<Rz=i!WZ?8#}~uKI;u#
z4Fi09Dl<-K0adjIAKgkPJB+<aoe}cU#fRG*!6EG$(9LKAe=cRD-Q_*RM)9I`^Y_4u
zhs*2r<n&U;Z#|x(F`im32$9TZJ512rLIr!OsOp(oZu)q7$iU;CZ1s&Mm%1bn!F^0v
zb16d7(|myS>_&TZ2`yau^le%wsa~N@wOzVOaEEHz2wkYsgQJl1V41cfqXeR$QUv$q
zm&r4oph9(8Uinn4&YyX1sV(f~uR+XV$3339hk3~n5SpqMS2lK+$kZ9_!Yij!g8Nu$
zt<g(6XdAvmNLA3M&n2V`WnZAr)d>wnyDVJSA2hEh*3ogp;ti-PLzEQ}%KNnE7xfA4
z7dX5be&A_Op-`*1zoho)ZJ`CRDsIZm9fiupi@UZ^*;%~09~*X?Z=p+xcI%H{d4Byx
zME#F1d#sK<I<xur{k*3)(^83oH`j~Y^)x9L5w_vA<+?4hD&d5ZOx$77Kem6lzE|&L
zxzPG^n|I&R(9Djz^~WCR{UV?L_*0$b-=_3!plGO0vB*BdYiMq!w#b6T@X2IE8e1%k
ziBGfzsSyWC6d0k@P-;1@c&Q!aa=DG%y7ZqbvN*m7ZQOB$Bf(&`4OA2SJ_gkyKdc|V
zu4RlaEA}f+5`X+N^5<X6dhTbIw`}8<b?LV(Yu`UgUW%t{%U2Rw#$fm8F4fl+_x_t-
zA(J~;wrm#Qzx#DywEw(f#s7?}|Ib|@G5nvqSh3>h|2g^Uztm|BW;obiJX?9XS=6$$
zoK&1)mw{i(v;%uYfw3J&=0n`$`&X8cCrXshDv)`M`jHyHah~6p?)Z9;^b}L#%gLv`
zl&(;qDk#a!zfoY=7|J>XW|v~I>{+HO-|y<-^qCZ9Yiev1#=JOnX~Apk{BGSop6<YT
zIervnI&-7%-mx=Y=IkNQs}pO|LQfK{0xON2Q~3*DtMOl$qcQhc1sWNWqA;C>_s#y^
z)QW5Dym(}hJgTPGD4)1|#bgyz6<z80d4ys}%B<tFQueF!>DCBef-5cWOZW%@vm5Je
z?nzfXNL%P=xzSkKFG3v(l-6#aH(Xr#RbhKy@)ML)g|{8Ma9_A{vN<}&ET-f7FlU^(
zlWMI=@Nr;|iJ32T;m!-raE9q*gV<zAQi;jjTYlCH(bxzKp&*<KmQLx^d$=n6V2Zzc
z-Bjz{=zx4Ijy)k=aX8H)UHeE6!6rz$6UFeQrF#9WxUi3jPN!2uGCFqP=uzy<kQJsb
zDz$>Ai`K9?H1npBVUpyLfB^PN_}5LG;pj1=N*;MbZm&uywoN5ZDT*=s-p+6du(y<J
zs}Z@;rRcePf4DLpO0z6b&aa8P(`-77&tWMF3`zwL9MJn6(K&@+QEX}YU<sew+jMDf
zjK%n0t(*GgSYPJU2y=23gA{}D_YcW2H>YQPn!Rlmc%t;;65itd>?tE^&EZ{-3UGK$
zj7sH_Q*y5Jv0`&^NyeC?EOW*dZG4TrUlfH^F2ugNG3HgP!0qSiF4$SvGuvkxy$GFF
zq>JeGHFQ^L)-BKs*VHMf>iS_{6O$Y~_Af4zML0=R3&w;+TjbXkgi9X54fAxH<|}Y7
ze&D=dTw{EtAa}IcVMdW%6y~6X+x%aSb<6DYSMvA)sftM)RfJ1&ZfJYp!dv{<-HeNJ
zi%VqfpBQ5Qsus~*$y1-qZxmxjCl`8o*n60H_<2~fUo+>ZYQ53q^iRcIuz!8Ar*n)b
zU*W~4U4GjVn%Od!YGLX&n7|xoDy0Q@4-arz1<D&ZM`Ai(KI9MzdS)@Sr#=*W#XrjV
zz1~f&G?LoA`26{;GZWbBRvmMxtp-Lx(uCrN3LIlMf41BeLs>>qMxp1{p(RWlbTqD!
z?GntawF<Q0w~Bswo1tUuHec>I;BJEnD_(NgBfk!I^{)y#WRWpn|GLYA?=NYcWES;$
zfbPPkR0?13)L7b@Kxu*;W43*NCvFr!cF_4!#b;mJtwR^=0$pDoOzOlfze?)nP^Yei
z4FCA)7-sOsdCKJN3Z9uXOlr76S7}eau;D@;mbKsWh=0{omjkkn-|LTS#Uhun>R)vG
zTKryO^*s8bF-|?#6&Mm!A{COrP?}`AwwX&mrh;cE8dG9_lj7TiA2XGkoQ*Nw7AlEj
z=wa$+Bq^T5Vu?wPp3A&q7h*F(YV;TiQ>Q7n>xrcIyjPkUI@ZFiy1$ZV_uP#5j3Tpb
z{L%x7cB?*a<3&PwO)xmweN(N1sy-^C(C78noK%61Ps!E`z5Y!WTg@&ta65+oixux_
zaN45B;;B(#7yK($SNb|uP2G#?y84_Vd4ytnDrae3**&6Wn@CP0r6B=dR>ziFR!z~#
z`E`8JMfbdx;!SOjZr`q_7;}BIl245(R)L4*rM?vIV{qD;+ILymp9ZG4{|(B)J6f(K
zyZiaNR85hcK7Fj0P$(VR;Wc$fy28wKme`n<ue8jA1i&d=Yw=?beI3V}H0uOjmYc~P
z>OT0dUWR<83_nJ3`1DZ+O_n&ake@*s+h8N1u#mI8W+^kZ+(KSEh98S!4ZK!h7@6|=
z_|NckPdYlHgU`FoH~b^C-2I~SduFFMs4RS5Wm9Ps+}^I(HT-myjXZk6K5A*u<(llD
zqGkfYrf&Y_V4cfVQ~GkH?|-oh+-h3Xx~4Z;-q}NPF(9)1PZI4f2D--X2+$p$W;X5S
z9m7#Gxawj{sKu#CXkkc>9Jwn?R<%E4P{{9JSRU}8x;GjlxUx68)o{qnrSCZu8E02g
zK*6Vh7G|78-Lg9IB-uUqp)`Nz7w;VAU=%NhOS8_Rq)OqTbX`gvJo!a|AsZF5EKQNE
zM4>)Lt5!Va@*kqT4V;0jZ_=Q$>D#MPysG8vuVSfQPHsKr)L4lls?LF~^ZO?hQ>Lf(
zg<IrQ#Vnrdn&2zx?V*|PI0i<6(*9QhUF$P8?zB=aC@d;mz?+&YGK#DZA7(_8(W-ep
zqxBD^3AxvcS$cC=_9G}Rk(l|BE1pY-4q89sHNDCz&<1VOQRTSlf}4LGPq#Jv)B*)Y
znp|<)M6^k9>a@I7U>8Jf;pI9X_GkWGR=h?r`o+o9*X_g$*han4r3Ino@F%KbvX&Xk
z>((bp#Z~Z-(#=(&!rYkS*Ec)o$GFkZiayx+{`+kA_mjMuri~=$Vx8ExBvf#$*Ih9m
z2b0G=g$)+IeIiL$SfB;cRE+$x3$qnV6p4yC*s~s^F*QZ$(+pZizYl^!HY#*k6vtt!
zc)Ib-5Na*emfD><bpFs#3)2`=SMP{9xM;77T7#zzu;K+e>?e-O3lFGV;3tkY_Y@f1
zo95GIIC+y|S;9sO-Z#6QGU{S6B{oV*i{ZGU-?7X5;8An{dhSE?m=A4(hO<1Q=mpl@
zqff5lf~j8@6jj6QrF5B9a*`P3<--<8RZTA`sWlczA?U|QGkh7I+)B;SW9>|J>J-(&
z;Y$O_&^ev|5M%ZxdT6P_q{q%F&D~I7a9HC?a&~o7z06u#v@4o@7!3}db{D;gH1F?m
z+5_uh;-Tkp$wQO9uyH;!O~r6dm;GFUapib@21_Z894IZBUYABN0VIN^=O|`QyLq75
zNmKy??s^4MS08=RF4x%E0InJaK9!~}48QI&0oc@X{bF;t+T!uADgBfFFX!URS^XdX
zqscoQ`9Ey>KMW4s;Qy}-?*B90>_4H0<lF!G;z?criTg-<pft-6m5aUTQP1Runz~_4
zKx{N-eyY=PsRQDoM&S<)l0MCT#vEYkt_cWEXXSsIFsf0oD8bPjJ9)Zmm_^KcRI9Y6
zG*{vJ3r2p)g<06j=&`WvgKojhRcTN2UM(OH{qB_|90dcJ@r}|IJLepgv7T%*zR4Sm
z&GvyKkC=T_)r=A=E<qL=HTY==1MJ?2yGDT_XEf8x45dY9*O-bf6M=b&Kq~pHit@32
z+9g)xhtmGZ*vfTkouybSc9~V+7$Z8??6U)MAK$;OZTLto8w>9GJcavwi!Tyq&VDUj
z(WfM{G(mMSB_>ECST;7L(*#9Uq$uVnW*6e%;A&}HX{3Wb^W#jsbVae3um0aluDDie
zwZLC86UN+1{Yo`Ti)|}2%kRF!tfk(@)CEfxFzl(}sio<XyQ~1peBJ09Gk$JUrn%`{
z&UThJD|*aNY#FEJwf5}SB$7S$?D4Z_hfp~NUa=Sz-@2&|CfU<A7+v@Rd(2}&f$?DD
z)+e2>qsQJetEi@79is!%6<(x^YD+uX@Uib;sCo1Ci{m=I(H0*+dku#5r*RvC9T%>o
zod7}#2+;zSfF0i(t_EsMFu7>o;=`}@4_q}+WXyihuIu8$)OC^k15V|Xs8o3Cvb(r1
z%!9q4Yv7pXy+os5eIB3zVf4V%six0gdQEIf&GGbE5_CTC>;r!Dpv1t$qtuMP1VfKp
zxu?-uM%Ap-p;09kUw_RiEZH3;KBLr;n`XNBuPZ9Wb^Ea1AAoHRFqit_ucS2%3M(sk
zD$&6fF1Ch71FR)x{wmvU0dSm729QMmDF9;Ii@O`F0)-8NGwRY=W-$h40mEL7Y}Hhb
zk-$?-XI30%X=`e~>-Kfd^zCDClZN>gT&t#R*93H#0|0(mXdO6`&xE4}lU$ZS{>6ub
zhy|<xQyT^wyY%%k!##asXoDmIdJa^mH_{bVxZI_WXFWJc#9a<_EoGL97tB@wD)q#~
z?(<JC^bH)zDmGuD>}|%N!d<s&;OTzJbVM`=3-sK~_7YY;^FFhi%1AqCXrVgEnlb1y
zi*bF*D9ykw@O0mUy*b!Cmad@QE$|&0Tt?6G)eH*d{qwPJZj2eDY(W2#V+0gn)z~a1
z2fReN1B8LYX#jg4Fj-tq1#0X}=?Xz0L=G7F#e2B8;hu)qN1pB}YN@%~lq{x><771(
zo+dz&jK=ue#sfPhCHvb2j!ZL+(u5^n4*FA9QG*>k3KSTA;buyc+nbv56_?0o8^I4(
z{3`o$;7IPqTM-Y6$k>mbz4ZzU*gBurS8g<(Ae1IzH9S7!P5r$CM_x)CT*^r461sho
zJFTMi=dce-H{yKQf|xo}FRwHuP)!UMVvCi6Nqw&tROeOx%zhhr>Rg_tM7Gf_{HeYV
zcA6Zs+O~ky8Uu71pJ^WTl%`Ubh(x<xANlGU{onF#ox9QIMbx<Ju=$1qgQHe$<W?6-
zv3}h@<=r!qTwlPzs1$H|v4K7eOtWU=t~Hyc)P4zC%4y%twC?xkjlW*xsffY%YKmH8
zyzeH|v7ESTHznF8YGRM!Z84MPWm50ChtDh;_4{A;#BLU?wYg&{6P9~kPa&#oW}n9)
ztucew+M)7IGV+?!4ZV+I6DC$EFB|Seo_l8lKZEhHCC_tw*qL=%3bSbm6s@MSqHQS`
zMlIH`mAKrO^tbP?c{KEUXz$iCDLZwO{UUFBho#l|&Nj{FT}*QOXt##lzDjY4ntcEK
zJ1<!Ox+M=;b6Q<bHma;tlX*3=_xg>P_Y%Uq%`5+x>tiD<dC+XNXSs>BpPTTC5s3w9
zNAJ3I?Oemg|1Z`r=fYi!x@XA~DkKqV(Aw*AKLK~Psly~o-zOpN=5+b5F~6?(l>PMJ
zrO^z|D1Y@>#*#C_2G?zj(kOl_6?k-R^8v?;n8~7qx<p$i@wOD@9bcaL7amReeu$+%
zrh?9;aP!>JHP=f|sF1tcBt*A~>@9@Th_x#u3y8kkyYx}yZ%5|^4VZi3y3UvgrQELl
ztcmBB(!!u$Yk!sHC?pGS`?l1w{QlscuFazCYnD7jA&E!ZEHOvS^t_s6!nY3?bXxM9
zKCI4Z=Y9WZo6)Q8e|^g}-_yy>T6Zj;JQ}~5a3HE|C!Swc3!@8f;k#owsS$agDocUa
zKC(CBSN^4u<L^uK<d}*Y*uf9Ky<_=Q&O+?Mfc}L>GM<&K5G~TTwn>KJV|m%=O6pgl
z%AJHNe7lnOyWgBIETy6EOKg)M8zeXHUSD=#f1jrMrgq*s{CK);B!<gVR9)|eB~RC`
zI}OeB)psm^(=-w4^8uG9*+y1(3@&~5)%RuYXMUOT`=v~Z5NaS8d{?3U#ZFtEioc37
zEhBr+hp)kO_2UWf3)({x3hQbH5+o9B>%2y@Rq*^F<4-MlOojY}Z3zi6iMChFhE9TS
zo?&trMyu9rQreqB3d#A0v38#=xvQ~nPD}4e6r*BxDm47Y`cU&Bxv`A-+0n3rLJGS@
ziw&DpODi-AFY{D*s^j*>gg*IG<b>WS^-!VP{nK3`LNOuZ*K8xz=uiLQ>m4iB{b_Lj
z4O*`;CcwYIDm+10tEuBl@#BBpz)@ZO^sF^UABRfxS*c5`eX>+Fkk7e{2^wI>c=p?7
zCWP=$DD1N>N@!2CEx8!kdn#D?ADQBm`N(Vu?4)7JZ-Y%lqJmkY`zD<7!9Jh*q!Gy`
z8P(A76ic2v35EE``6;xKo`*VFT~s(Fcix+>FBCI3W?!1%5#aAAqjk9WRp}ys|FPL~
z$)hNP`!}p36Gd#hVl8sGe3k~!zJJ2y<^jXDf9z+)v*FFldhI_}dcLe(|NUS7_~%Ei
z{k}{H`2LZjx0ZG0zyHgg6mQG#9@AlWspPLJa%VM8yq?_o;}yuC*P?bu{qv!t%O(PT
zeDdGdJx*L&<NM$K__q!Jb#B?r!1vFb{I`YwTsEQbW5It|f|InhVQCiNf9uHqZ*2{X
z_rG%k|F)e+21QxllffsHw8h&V)1j}ZFovn~l0A=pw$lHn>Hh6a*ZwQ4`~UO^|4o<u
zpY1F;0sQSBq40mUb&g&?bN1{LOP(;&+qZAu@Nak(`{vCXan*Q%fLT||;GY#?8_&N*
z%bp(@;#KWi8Ld@o)|lp}uUftOy^L1y%t4dfcb-jgmOR+8TKVE(zB7DEp1FgiHue0Z
z@x@|$t@z7r)DNo;^Os3&+qpBP0K@1_j#cq9F(o^pn-aB9%u=_n8h0xV)6YL|HOJM)
z&+pv1(}d!gK{PLLv!=UN^6|4*UyGt=TUE;Vj9%9Cdv)!XRD%?~y=Q+ChEJzY{77uq
zqRzUz=k7BQo!k67TgIU9Qu(^Om0!BMLmxlhYhhtQo*`M3n%mmihGF4jG}JGj3|$A>
zq#y2*eXkeL^H^;$#$7bB-~628^$w3;(aZkzdEHBw&o|_pI=&5io0vq(IkbIhkmYqe
zzHcdiK3e$syi3ShE-w9Z=N=JD?5Ls<iG7}3Vo_&El}(>SmvZ#AH;a1?wB)%T7?|6v
z6Qf8Rq>(A~R!J!)gAv|cA9VbA-Mg9^17F`FqB&7xJ7Nnb&SLexhWY!ATitR$Hic-#
z6|9WkE@s$#bCb%``1~uiyR5HVDWt0I-Fu8tA4_ijsJ94;&q*yky6vsISID>W2i$p<
zWp{Y<CI)F@qN1W|gKutgX`DIZ(!VR{SJJWqv~Km!Q#c*Bw6t8eZrzJ!qs$8;O)_RI
zZUF((`1m+)UUKq*Q>Rax@b|Oi4Aib0Epw_>H-F|$d4QJ=D`SOq2+cu!N*7o*+>?R#
z@9UjE|Mb^ie|2l7uMNFoZEaf{DI12RrlPm++QoEJU5w6`nLo?wiWQ@3aF@MxG4?5>
z3k7<xBT4pqzHJglnbDE5_WR4EnoIm}a}3pm0;~fp_8zP`@}+Jer>x@P?$F3cyA+(g
zrmdbHzPCP>Bx~Ptl~EU?=sTIEr>~DVO+Y{ZI>Q9b?r&{1^P&@3bV{0bjG}pglmExk
zfKtR^9clTt92Q;Cd-z!$0hNYf)Kyhhs{QiIFYA8){T#fInrWs4rbtDn8IF#Oh<s3h
z52yKz^@>PJYE!(1T1L7nL&L)M!;7eCMmj@9V?m1U-#+Qpk1^`|D+L0Y;lU?QpC+HU
zyRok(;&7pRA6_F=l=$h>Mdafdrewp`OfzbsN5AhhYqWK0jKMD3q<Z!9o7);^&o(#2
zt9g8WvnuqUg?XqbTJP%BWJNDl`pi(N1)1(?T2gBnndm=1Wm8^J!DPe+v|L}g#(j3u
z)$FCqF%=aRQHzoyyXNHQy|qzk@OEZ<UfS&3+?XSwL(Z{1VHMYo11c)!I1Y<i6(a00
z^5qor7Q{|ZMD&}V9a|ZWB`A7maP1I3ci}>GzH?V5a)OQmjE0VmQC}kw4Uv8O&Ye^+
zA9ncl8fvm`JaPh0Sg3eme#U6Q7Xp)3sYOsWhON7$GO{eodiwkO8+rTs`b>(vv%@4z
z6Sbn``ob;z8~U1(Jf=pi;e#5-jvY%iC1<`VD`SV3&YpsB3b*8uG%MR~-<nB{R={in
z5A~aw5ZSly6ga%m?RDET{Z-uiTeM&!wd9?dFMX$8AU4S`&N)r-9BhOe4}J0^0j{+V
z!a)?g3cPb04^Q%bt;hq14(T;CG!)KF(CCGoU?<FMjPEIud9gu;iL6d(Od5P9K|e|R
zYce>nCJH+~+%EgWgU=mV@8#sP#yiTAr}D({<Hv&pRL$GsSG>{^>3#6jx4*cka@4oq
zFTU>KjgZ)(&F4ojD?7aN9chWM;CW`S^5|bBGc2W~nG1#P>f1#PK76eXC$)fU#6d3C
z#gD&UBk)aNVNxv!(~@Dz(0e)rKQWD3eX!IaMF-4TjN^C~>Nuqtky>85cISGJT(Y*Y
zQSviS(vFe2QuFlBt!HZL>RQ0<Wb9kS=O(*@72SINszx!JQ{o{#q^Je+kB*IDaACo!
z^XCs14IdAREG;dqi&YI^*VD+(tskDB$b&V_hPnueXCvawi?IqG^p$Hiev9$yYzW+W
z*d1;-iQ(9m^}N%&djo2IvRgYg#A3_Vt<)kPjHcgMZ>?QRdW4XOpx}w2Vh*7bH-w5&
z@l&iP`1||sI^y!oJYdcPoqX=KeU3StUO(|{!>zSj$KiBG{LCW`S<75}wPJ&SO2?qj
z>P`IV!79@S)G}<F5+NFjrWp`}eP6>ZTIy97eCg)Z;Rn4Gqf30ot~}f=udO;eeDl!Z
z!!ZOhc)~6eHkVa0<CIiC5*8w?BYNaWTN+AVUmu-#c4rblj@6;y2TmI)V~f`xY88VG
zr+^!Nw`?ExAER5q$M&AMb9HW-t&J+Nt`3Wovb@nwb`Hqfv17-TkKa(XjY1oEk9_D;
zsEbyJSHSdsqYRbkLzrmB8aE2iDkvWPiBJc%Buw+&`#QQSgVV5V=KFW=-suXc_`Za^
z#n3ybHZkxqewy8H%G|pD?1xXUZx*{fG_9+6*cjtA-xa%cix9U3?z^wK^rv1Zc%dDA
z#Py9#?fcGeMm4*&zW<M-4O@5YNPPBek0zaEIMJHpIoQs7T~Gb~{riQ}U5_6>=C=^!
z>~9w9pc`PP$8AyUW_`P^lHwhb7ryB|Gr)(58^=K&xH|mlb=XC1%ilL_K%YGus4_RM
zPyUtIDDb>S#cw*<IOj?`q{(pMS=EsJT6%D|c~=|a`d4ejgocI^9}B9b>c*drJY;>p
z`g3}Wk{9!yFyVv6{3!Q6VPTxIj{h{f)nYq8f2NUzAGPnAITl-}GWluc?Yno8pDW0I
z&kGl51m$A&hFv|drD>}&=Hy5>6>^V+c~P6#*aa~qB~!>cXafTS!gW0xB(KfQaEz+M
zBxG?D4&lf?k5xeJNd=$Q(9|4ze^&lRTibbAhqiO^ryshBoD)^QyOFQ)+Ak~J88Kca
zzecmBp^l_LeN^H*5IWz6=i9y=IR(~wwfgi(rNH(rTjDy4abj|EEnn-!`+_&T?h05a
zW)+QTZ`gV0q_MHFcC0C+J&*pDTD@2~$H?j-N45Pr1c#JRabvxD6erqf`Uf$6=JFGH
ze%f!%bI0-1VFd+aB`=nHVX)cJ@%D4JU`TYn!$;mqzS}1UFGH1NGpXOJB${N7g0ztB
zH{+S-@KNuuO`ZExhkZn13IxI#vqE=~@n=uvhX%6kTP4AaVv>?|_9zn1;VXfN4f8%V
ze$26<W?QKtfM|<@C)&Xw-6YSBcyrArthJo8lFO%SfByNWn?tvId5PalGZePqn>hl$
z!`Br(m<bT#-pvfSfI?0uPS*PUw&3ZWiPqA2G!n(F(|jtv=QnTOJk2Rx7<B!88sgR9
z^5ul;U#fRE2*>)yNtzXY`_>T8&;Izr`ayfHZ9M!0M{IMF*H9-Ciq)k@_soqLtJ*am
zpgDeI9r2saH451I87iGYxTM*Ibx9j}q+aPM`Om(DB+M954nJto0s(28qB{Cc5bE&o
z%b&<$T`Y$s@dGs8kN)=UYHf^CtDZ&nrQ5sY9KIO@cRX)wBx*8V@{xjrgWK~RqpZX}
zGpT9dtosB6VwGov_wQ#y3Cqu+*v(7~-oJOR0cxLPFt?cDSx6y<a4p-nZA;K2YKXvv
zA=LvAh(p;UaJZ)KVAV{h&J=L#G~M_+kUHYthl-{7PK}OoI7W7MsoDhPl(%nB4t5kY
zP2dzUy@C9G<GPL?U!2;zcdvPbHLIfl+x<{poUO!qdE)cu&+FE%eM};e{AUK``Hr~k
zsh0>55fNdwWygPqsM3IB45H*50VMB3>atB;bi?x#cXI&`6B83n_yZ<~E$Mz7dV7&i
zj*;{Bp%n0xZ_n@Y>%mtL_RN{T;^N{A0B<7CLS1NtAj!qD?P#s+caH^k$y|LzyUqkO
z&V&H<@-A}t_`;5spJ7K$CBAxfl$ey%|MuQilVWV1<9T@qK`6Iq0K<h|L#&pwt-DnH
zi%KTHaA_8wda%Q;G2tGRzsr${LA(labc&Zf*syib3$}Oa)TzGV&t)40njoO_?5Id(
z7u8P;MC=4B+CoXqKgIyFGocnD$<xTW-~7qhSe3a#@{NPWFA@MoJp=oT?%8v+6c7cK
zQRCEo1q{va{plCTtY;nDb6tH1kbwGf>Qyk_%14fzhhm;;m=UQFCY}<n7HkL9q#eLx
zP-XE9i-K&my4B0@-j;pM$-08FS3ewu(t8d7#ATj42=XhtyGd;C-q)i$UCQzMG-WQo
z`(tjVYr&Ig0;}46cy=A~YiPU*@E6qhhZFpM0+`;K>^UVsM(|5Mejxgz4X&8^>K_Ap
zk9v+gEdW+T)4Eb%R^n^i-;z<SYY$fT_4C6t|0yVlxupRGDbKcPKNv@}-u_6t9n&sZ
z1{}|f;+eOB^4sdw%yMoEgN6AyaLUWLd5CQ`f(+(3U1W#-W?2`nt3dJXU!AfJeuUUp
zYM#9jvN0yqZ^G+$Jw2DTVpYtXOXu>fabNCj6>g4K+av*eue;#Tp+n6;qj>X#3Esyc
zPa?2#P@yu2FK5G6&462*V|_=}X-89HV`B*c^FDLk0t+r}N+^_>MZjDhQoO78mqSuc
zu_*P2+X*||WpbV8h|9$wS`FF0Rol5_DiJE>!7?fF;04$z`@I<HD&&JOtDGyfdZqqz
zZVp>LE9AR;yQR+U+PwMs=d!!~01-TRo^<5ehTPdCklEJO#_Q1}1L*i5QYrFjP8i!n
zdp;CaM(t{W_sAD!S83^;4ZC}@h9Fg=3*BvLR@{;oKCIlNd~x>y@Qd`->v^6XySX|A
z+<9aD0%S0vYj;K8z{ebgAb{K2f@(p{ZP`{aa}ymqwrzW2Rc3em_>DmZ3*gzd9XonQ
z*iarA!<ekI5r=jud8W@#x0agZ8)*g<ISZeC`pm1#uXT%LvG+(E!GGEo!n=Ipf*OQu
zvb0r28-JR4aiPleH$jiVcIt#5#gQggJl?3O(3x{{^#;8v{JxkHfAd|kb_oa#0ptxX
zsZKc74fUiiK3L_&obU8_lMVuV%loc*g005CJUG~!jG%kC)Kue~#HOwjP5@`}n<3pL
zgFhC~e@=6q(p==bJ+WC?QnWvabcw;i+1asrfR%bdk!I&54jpRL!))8(;TiVyDNj;e
zX&iWN{x8{(MIF0J%t)p4g=I%mgaDcygV-?y#_IUU;oFDjaGSB3hkY~$+w)`r_kcuj
z_VAUO4<|OEdYTh!fyZp%Q#63$mg`5~anL+H&%PC_Y=Om2Q)!tqQ&YBpkRlZ@iDe&>
zgnkk+GD^<!&2JJHUVhLc!I@gpjpKDx=f(?*=VoV{b8K{woCy#vk9T2$=KR1Q*&P_2
z(E_gKsr3B^5A>nvI0AeTMW-gD=qK6rynCEqIe6Tn6$2@UCUFxuL95bGQT*hjH){CR
z%GM2U0kIr^z<m(G%HkMAv<R4?bzO)QA%h{!>FZ6NA752oUT*v4?Y-6wsds?UK*h|j
zG{^SsbEFqi^_&~xZUeYw!(NouKheMgR=%^5FBS4udkJ?Jwm(CeJkojf#ful?9hiEA
zVuV@2eNq9qv*Gp>*WWfwHBfghW``M)%pGT@*ibkT%q3adl%yj-8%&B(^k@XvYtAs0
zzdJ+&Y-r(EZ|>S#v-2+PcnmuIyfpX{+SSz+5*ei;aA*7u36mpv1#*A>c^p!H^Nm$p
zJwrpz3aoE7o!Y_tC~?%;re+{;(X7Wb2oJ}FI#yFsFbE~z5#zb>pAz_|{Z|lJG6rcr
zjP|8Ho9j+Hn`_f1qZB>jsIokHV88)PV$1}R8O3bLh+<3`%|hL$a=HSXQ@Wt++<*8G
z{q=ol42zm;s}C_(v!$W{h;3)Uf?v{%xBjI|ajQ4-esk%+WDdn4_u}hcp`6vYIe?7`
zLE&Xc^fwHKhT!pT4hcnZU~iM0D-RKvm4pLn$_@g>4(P?#wkYS4zKOso>yDSMyQDAO
zY6=d33hHno2gsbq%T+CPfSNt#XMHs1#_CmD!LS2T9d<U8!=Lj>U0=!{fEcnmY39h6
z{{A!og{MxQd`TAsAtly-I)&I7cZw^+Bu@jp)w9qDTszNq$_>OZMzW2KO@9Wt(+yV<
z_bp(76Ce?lwlBr)>wDzL@I*9oD`YL8{~w)xgR4vZnm-=oJM}djQe4RQRRNJ0foU+Q
z^-%m|-M^hn;_nYcGA-x<oU0fHST9MyfAaNc4%wsMFeyMJ0U!i+v`Scl4KzpO@ZqM6
zf`MQA_i2VTW9m^1!=J!so7PVO=Vg^b;lrZCub*@Ti|EGPG9f!T`b0ynCYB)eeS=$T
zI{>h+Ee-hY2)=BtY`Q1SG4ilY5L6vA=00_V-{XdgZBhbY&tbH8+_x^3D=!E~AF_V`
zEnuDzAkGFhb3j6(wtF5w-V>t3{CrnnBDs?0T<RE+bUypibbY`)hF<$P=!vA62{>S`
z@?C^8LcLUuq()u3bZH#&{Y&>COP+jkdMUWXmMvSNnb}a}=7!5Qg-yea(Kp^N9;w*9
zZqp`mGjO&O<Y}}1P$To~;$n)}{8W8F?@Q|<p-?d+Z7>*_zA+*km~8u<K}6mGi24#`
zqdQh59NX3+tQFbg{%rg{q-*=OEX({#sD@Ta*ftacw2w#?zry!VM7kT5yDJ{%Y>1_l
z?T&55Hv=@L#l9w$Ouz5rU%7gHRx=)tmvQb?4ri$A=(KWM_*$l@PPAF2p3jb}p8Lu4
z-4~%e<G{0s*8L6R<L=cN4T69EX%KgEEv=H#c@YbI(y_B>s<72%n}f$bgF??i6Plxk
zRb>zwDsv1ggw{1bAzz)5chI~jIpVNQAw4Gx{4p8ol7bybrTO)-Y3s{|y=QDdnM~A)
zbxPq3eAI_(?Lu(|-ZlQ|<|Y#+=nx*1B=SRkR6FtEGb8tbHr?yjuZI=pkW$RfNl8h4
zAkrKdAZSS!Pqf>IO+%qdYHl_l&xDB?)qVZs=(J`Ne>0>f2{x2m%65~II1q1<0u*H8
z>6t$`Gj%%Gx_SqwCKY<YK|w8o!OC9@>k+*G!ZU0dc>fa!S2~95+~q`(&9kW|gw-Fh
z*>=#pV}g6_x+uEi*TE8JwMcnqr<n(PJg;{Yf!3u4(qL{g0PTLapYzSz92dR#_U+pu
z2M)Z`tGCzh0~!nhR5@>6$LofopsyYq)Lt8c<k;mq_K*lQI>CnEQ_;uo2|n!T%*V^k
z&Cu<7vt#R4tCTsYZ~UBxK|!8Arj|~K!xTHtJlcCgnJK0Ya!pqA{rC1CU%Fb6okPuv
zyj^Lp1_1~|h68OP$EH5N88<P|r@O}#&njK;U-jE>=Q9~qxIrL-h`@!4wC}Bn2$LZ0
z(|D3a|MpFkl7T<<71f&pEP&QB!x<X#tOur$f|z3ZHWxRPZdEf5unR_e3)J?(31wi!
zhCuA-Dypg{;kGYV;T7zFRJ@prc7Mx~HF<IQ-bOyfo<?!cW6h@@?n;50C&~VisE;<N
z{jG;bLnvLZ2Mxi#o`Bj|pTl+Q^r=${zP>;X#L1DRopJ_EGL9>a08Q)D3qAM<;ZgLK
z*G}}p6o5+F9GAB&pB!3?mmp;J`^0>pmBBNe;2Qj#ESyAtD~SXz_UWJi;tbPz1}vHi
z&VgbLicp@B#DFHiNEssCj6sdI+O3W88tchhUzWAuJ$7^~#9}BLl+=_91#bMD{F$Qj
zrT)c$ER|ViZ3FpMAnqA%QXamh<`?TXVNAoaBt8~=yd7c>3}blMg={0e{)CS5$SoR!
zaraXBeGwrcAvFnL8eko_!b+f8PHVl9M2rCYLik2=C}b)&nm;;vuWGBt7QB7{?rlDR
z&GVz^h>i=qs%#><<CiVaCjeAk@mW?QTrvYT()+srR5KA+zWly%012vdM*eGgqx>iH
z^D{$=z=?^hj9P?&=QJa1473X|kc-4xhnoRl<N}7{Q}w@)l9HmKr9}mJAwlCIWpP=>
zC}1kO6_g85bK6HoM#3V==GDNRz=^)WV+u^T9z9^;MQ$6se74*QNw~jV%*4)zu5#f;
zKna3#URY}e_ZGZiF>HXOIB~|tQDKiCKfiVBH%*kP>Oo1!6F{aY9-0$zZ&?^YW#x_+
zr<%~PM}5#ppTJi<p;e+&-l6ScN&#h1-YXbXXk16-W2oGNU8NS_*9|}yNd8wW4thu9
ztX}rC+jv4*r2@P{)L8{2n^1uY#!RnTPmv%JHA3BZ*8TW)m{$)U{uKDhE7XqFLMCo^
z+ExATU6#XrxIQ$2ObEJv|NZ9G8tUr0Q0N_bYMiz**;n#%pq$YP>Ak1DfVs4{a2+;?
zgcE_m`k{<B09W<^K{yykK;c@t`3Wiv=;9#SQtOPl7$mPPqFUP86D9#(Upr4is7XYD
z!F>oeoH2b+L`V)(p#uT_9eyQqb8{XugJhfx1Ov_;Bx6mw=}`9Q*XnA}>_MPWV4nJn
z>b;kcAQYqQ)2Qd%SyW&&Ef3&hizppXmIosnrwv!b91-vQ_Sh7<23Kk$iC`%)X=z>9
z5I@JAeV78wW*Sni+w6;DKl=)R*DW|5kof&Z%DKa)t%6aB>$Z!&l=;~tb6Vdb`(ys*
zQ{ti~r<-*c&<0SGce**0JdBfPa5lx7#m*emNN#S(sHWxG3YJMJ&qE&!?-*d?B*Ah_
zw6&Wa??cpR51>v+_Ay`kM5E3zujWD0=Q!XkZB9mfitfoq=9*<vxn4u&nhS5dB`{;-
z1v#|eqLtdNR)RYLk2A-WSopEgngL$1&d#@5dHu}YhdSwEUC!B~xZ%XVX8|siJH&G0
zhWW>C#ZU1$7I>mT?UcxS@?_iHq0b$!hs6fu$_rCLw-g1GQR$bU`sOVPH(>x=bsV6e
z8TdT1xy*notr@09NZ+t|Q0`H$oEPf??OZnfV8V~;`KkBkD8w~W7%qZD1~t*PtHdv?
z(CiNx=uy@8=i0Z{ZE?xr3XV)n0eRP$iDWWBR-;9A+be?wyr#1|dmhrtDSv|C4~IVo
z7w_@q?OGD2v@;_-<}-uA7;(&U4sC4=BFo(@OqC*FykdvIA#lEfrc<ciwQ8o)#Uqht
zCazmH$?&Zf=Bu?LXSecoe#^Pq@F(;y5R1RBtDAUZpinRtfdcVaQ$kM85M;wbmXGJ_
zCIIAV(4~;9|8bscyI20*JKG#2cddQZ8O)kve!jl$&|UiK^AK>Nv6BE)D4Vs^fwt^2
z&oSe(XU`H^UxtzMOA6kzcC=6JgLA;~++L349T;Y_*)J<IhD~<J-X4jj%=nJ+izx>k
z&z|D@;_gnL-58496$YZ`bLbBBj!H1cXh5h*Kx@AygW~@^IeOQDkms<#vRq7#i^2Q5
z%?V;eL<B%<tss*hf}?R#W(ZD@2?zvgeOSNwVbV^%s+t=2z7hF2MDH%Uvmw9oeV9br
z8h-DiP;y#}e8zG?UWYv+{2A$8oo&kX)ryo!xmLQ8(XozXt&2%RG05$ls4kep((@}S
zm(e2ELnycC>!7|$o#EZJD;0#~tI&Ly?n{czb?#C@@~f;Y8n%`n`w>#48;)Q<H946K
zQeoIM<W9bMhqDCbkE6Z_#H?2y3pD!qU2jJl432PGehe+Gk`xy|3xZPEhUW!jyi><Z
zUFeV2>pr-D-)&kEwD0L}@xkJ)S0Vh%?9D&hLBvV4LV1fJ9V2FCWkG8uW5aH_t8n*W
zg@Bz;2lHpl&&^PZVh*BEbj>E8@c7mjMkI4Y-7YZ$oZd3kLP^fS)tTlyq(E)m{!Nza
ze!Tsr-EwEG-#@7$_n$s_5~fgYT=c?W)uBTNI<Qh<tzY^5?T-CHa`&+I{5sR?p{L<_
z!WQ%v8P_i-J-&T-{gu!;TLYPe474sp@MI)(sh}x$-k;Uf)Winm=R8&!^c{X@%-`bD
z>7SUu^w#%&b2DHKOf!(aR1n3}U|aZsm9d$3G&YpI%ceWP%XcYz4^vf1CVB7OItR&J
zneBP@N`8Fmpr>YKaF|q0bgF@Cijl4iXuZ{8M+5`}5Kf-qf)|3K-@ZHECnmmilz?M3
zfHq<Bs;OfaG|m8ZX7lDfdi2Lb9_hG8``=pxby`6m$vfzSlAGH$)_BVxeegK(j^&Z{
zlD%H<WMyTk8Dou&jpLv*nlimQeUhM3BRyI2G<NJT1sb6Uy}VLSh>4#E_b@q51ZszN
zb9+HlgCtQb_kHkc#ftlhm;N3>IB+CL?z%ZnX}W`)D8K#$FgMKfE^q`8=#B-oBfZv`
z7zy?E0!)(s!kkx<4TS*uFRf)0zw$RXQe1_Mmc;q2S9(J<GKmJgk}$2jQ{}a@COgoc
z^;5O85Pz89cqtYDj^f4u$TVRlAmR0NcU#R&LUE@}L$E0mzC!bY3i5V`q8F6Zd<LK(
z>fljZL&LDvHx#4sp{_2y2M-@2ND9-8bO(GcK#6zm8?5`0LdffWHq$$f^RAL7ISALj
ze~;+QZoKxOP?l+V1f4v6x_5F6iBf=sysd7Y`Te%QhaLSOLrZc*?SkOSp~7hvgGckT
z=`UU5)nM2Flo({XgDGETZ*1^!Zgzl1)&bdA<d*HI1;|ezauVxs1Jh7%F757ttTe8%
zP#Yz0Y6nV&=j0?<P85OTUBDhIbG#d^AQyxoX@fLV+_Mj=BNSJsSmz{A{!*bnekTno
zGXptk0i969zjB`yO3{f`F^~B<+Hh*?-)V9$H1zW=qs2ZyhBRvIw3m1L_OGKJZg(|E
z#dplkox%fWyilTzNRT^<5q-TQEm^_!%PmKgxOmMV0cx`OTJx{rfBp6Q_q+^^0m)mo
zPX3_%W;cWAlI!yM&GD!E7m$YHmc$UJp!x@5_qXRPFwC1|P)vJrj2qaDf5CzOh|5t#
z;cOo4jB14p8cGbC;r_ED8#pt$K5wRGqG!wm4|66c#jeNxbJNClfRE5ut|TU?=0EB6
zub-Tm>5s4>RE7KeB!#(u3n@D;eHT*5UH(n4Z?179|J9CCIS7DWtk@{S#9G0y80!HA
zt@>*}&z<McBRS6mDJRTys`Dy%jdjY@)KrRlU!xNT8rz8odfq8!K?i;w*8X!4VA~rR
z`BLC4@9XR9^%j_Eq;-~&H!L~)G*EjV(?9j60&#C^0b2P@*MjoMMnmKN4)OD!bh~yQ
zzH-E`*#6_~3y9}8v=}$ZfCM8>t|!mbczRTOIZd&1+CUMm?%D@tX#yGP34xFaZFti<
zpD*`>4Kgnj_(uFhH>bG%zDwyKr^BVtf*{@2NHfexg?z$DMpP7}Qvv`qhETN=l)PMG
zSVTk$q|O3*&VgYLhuqxQ)P(3X<Y{PRffS8ZG%UeRe!hza{;cT-3Z8lCDwzbRohi`c
z51WoV^`PhlJNv+knW9Gt5NpNPgsTLgUgc%bR`Hv;0yLtvEX)lBQ0QaO?1_Xv9Fa9N
zL;@~#fztu`l-81THsZ^K@`D``ji72WC#F1OKN6IO0J5}aS>B{EEObEhN7M}yCTM^y
zplXpoF0amp=<cl+|8c_yuz!!&JaFCiX%>=`IC&mvdf&KtlU}IJl41fm7ze6IkqJ4B
z3}t-ya87d$G^l3KLONKcZM>)<RwrQ!T3pSb1)_F}bU}2|R>yb_b!LD7mh;E{!l@0r
zWZvaN`~;9!)|@hiU_)k2H2q-sV|u!u#AW&C5De$wp-V(8ftjq}Y3SD)c+ow=i6x-v
zLh#mS5Qa)-G6zrh0Yg3wLj<L)0w~V9-rg^XDvs?kkUna^8bec@-Kcr6z_&ARKqZGh
zSEQ|@LoPm5EYC(#l_`v34B9w?I*&}M`2PL<->23rUi#lNLO^sgfHn6f7vvDf@56wQ
z<{;>9WWRLb=XzyQGEN;wG*Mz=n!%`vr>4IMsv+YJn&FZ@8C`6s>XmZmWkK}~$CX!B
z_EU{4*cIoeoZ#w3q0!?=k9uATaHAgxiFv<QJan-2K=YMg!~XkHk_dXxI5x@*0jGj9
z<A~u&o&Gb9G%U<%Yy&MIub&r!8#3L9iK@-y^mHnS7GZ@+<dpD)1<t~JtRpgQCb=Mv
zycmSN7brQGPe+rm<9$t)dexA6;S57=d#3YR&710Cl*|-`$sn<wx4XMZ0P&0=Jpl-v
zs~`4|=IPc+;Duq&t|EGg$KRt^F7|tW&eus)NUf)3eS>R|kfmHq&WSDtaxH6+W+*`>
z3C_9F#@7h=e$Z3JO{63JQHao&bQ;tnvqA%C?COHfv(5n*4}D8yZb1_TJ!s1Vb#)L7
z?$BA@cAlq2U)HYq1N2xWpgY2b-+TuesU=K9xl<M(K{P|Vgqm)wzgymE{Bm<t1@vSu
z;b`NKM*1@;tj2?-vtK-i0NndQ+@<O#1;N(7Ry{Sx0B-%bddoqx!UR~v4g^|$wgn9C
z@%!{_26W8)-qLkb1IjFs2Z!%NI<|0pnx6n}KpG^h^GrlEKyf;{-Lu@GWFSLEKwhsZ
zeD55C=V-J08shIlklLj<x96&FQk_L6vtpQIbxK25XD3IrQ4?6`J_|v=hz*I?7MeQR
z>@zT81h+d5QHNCyJ`4?=M+_^NU1EdA!u?9-XM~FW9id!qo`;i#qck9UAU#YU$FL&b
z`6&i6Mnv?7@hKG4sr!HZ^#PLXld!Nb_`Pk6I$P*dV?W>KMkakcn9b^tEJoi)XS!^g
z+zBHZ1VpqjRkO&(ahy<d<4w#)`1L9mUb^J@!yEzDPum&SM+_f-wQ|#WCXhBbMye5s
zuwOem8S+W*M=@xYtzElzmx5cuE_tVPXvg@$cJhB=n4$r%u;I5Q(pG-~VHBCNV)r$v
zfe;sJ8|gEl1Y>J35|x-~Mo9%_Y1g64fBDa`TW5LbjxZyVMHi2N91S9+F2r~P%v5|e
zkdl^=(1MDBh3z1Ymgh(_gtp|!fmeO^7zA8hz6x|PVWz-LM`bF(+~@0dP!COjg5~*I
zU%uQ%80zkh2hY$0Fq6+qE5?ywpmZF>)g$XWg@uLf-~P23X$udoZUqexxqA2M8FBNX
zm&jPqrSkNJw|u*IrvX9gS6!G2F8gB|X=4;%Y>G$Uiy7I&6r%xX714yxCeQ#!(!B&5
zrYX#z^*s&F%j-0+1(GXV3DS(!icyqR)3KT7_h?1>>>!qybEI`p@FNhu{ha=nCoN8)
z9`LUFO`)HfmIV0<>mGyXzd)7w>%Rm7Mg!6GjI^|L{_Zhk2n+?|1EXv<8XEZMQhz^C
zXOf}UT?m7qNW&K{3(n_=O#bPAE|~S?(7n>YI3|VeDPP+_@sz>zo<$nhu;#go7oS1-
zPC#@xKospcHo;$`kuDzKAsDed@Ye_(Ix2{m0M*+_%rHHRSJnO7BXi%8Qych|&qJa=
z2N`x7Xqj*}@Fq=APZ|(}2M&fz3|OAoxWH&o6F{&pv`()23|T<W7#WU&W)AWgr2b^^
zZDb}9l+sW<J`Peb6A~s2Mu`ubmQdmWM|_`Q*}63z<UHdbXIj1%GLz!#>)V`ed>E3)
z85n9rh5`YQgP0o!5XTmZe>=s?`T43dgB>2NX2^ihmoHz$D-nnZ!#`%6SwJ<_t*xyS
z=Zmpy#5a-=z0(TPbJMHqE7_yf2N4qhQMv={;7N3JI?~#Q{>R3Z1Q3-Pp?nXzd1nK$
z$FiwLvYrFytFxico(3+D2p<6c+-Z*F6p%(0JqMmbAN49OiUSaV34Fy6#1~D>1!?T~
zRixH!;nxDi3GxHdQci}~An^@J!DF;Xoln8-Bn%Uge|!31yI3OhQEZ`cISp%cL<%27
z_5n$rw07(MBff!jLyI5zqoZRnf2{jH?fbz)9)*2D9B=1ewOuqc0kRS@APJ%>#@l`@
zFmOH6JcdL#*6uOZ1n0^y%rN=rgspl1-fbEOuY3rzcaN>O%ypsRCxW;k(q}}74^0JV
z2VAl~G)cuj^*L!n=i%JZqt`%N3ALgO<{<+Skf*taFHZ>y37tgd`ape;gE~S}*$)L3
zQo8%3kDh@a3~58x#7l|S2L%IgMUY2jy@xN-q#U7ZfWSWYQb2=OVbBByA0C>H1wyPN
z**Vv?$srg-zoBX9VU7(PhMp8^tmZHlbToS82|UQW&tM9sAJP&T<<D#uK{_?sY-BbL
znMj1GlR{MR;c3ng36l4R32qUS+^eo99LPV^!fRf}LmPe!<TzpPSoZAP_3-}v(Md%g
zGhn*Ake+pAYk?s|Q7-gEfHWS5#@4ur1<aB2@j2r98MDZQ3sM7tkGoSUA<3jeHrg>c
zKtOaSFzIvX)@UFjG=p?V6bP11LpiTT#A9TN6*}3ZiHvj4k?Fi4Dxvd_KmK?I&og0Q
zl07X^0&bQ819GGZ_wQSP1_uh-J21vOf?eP?_dbTeNL1$)dV=~^<aQBVhI04+qU+7W
zn!L7n;q-W_=Ty<t+fzjlX=?-{G6(`PQ>kb`K*4~_Sb+!uA|N0`fYw%P0nsFY$Y?bp
zVNyn!u_zE`L4wRGVGM(YdCXlqK=9n}`SM5Wqi^2qz1O_fZ>@b%bP>u5Go7R?mcovL
zMm~a!fp=T1RhbtsJ^UbKg=>!hl-y+yV?k~c@MR^6CUy{LrT`AGYxV-fuG<~g&Ij{f
z*^QFN$$@Xhrm7yd0C)l65#M`~J#eMb;N79JF67pRCqwwfAVuAy{h*g2;)#&ZK#8ap
z$YyOnc+V0Np&5XF2QZiK$f94j=gqQ}S?h*Gt)&~=>2@eQc?~&?830qH!bUvi0H9HZ
zAe4|(M)iG&zGgnqRU^Sq1%qZ{1!*PZw*LI8drA6bXedT&`C}!*b3jFGIx=LrvW^yT
z>e<h)0KOpa0uU5be+BUDm~Rpww2);$DiEPledWpj+PRYr0Z<xbC@+Ht5in6W^aik?
zsN4@egU^yr>0e-RQB1BMF8TlzV9`|%@Qjf641k*-;g7t_J4YfNh5?q52ap-ymfDdi
zV6z8cL@tAqxWDNN%FiIU8{@k?aFGzSYW(dpJ!IA_hFD=eL#_ZFoql6x^6iGfI3%{C
zxT_QMWrIR^Kd{gDkV}*wJBNAq_`g%X`xWf4CHe1K>iQ|5GD2PTE92m@Qx311L_)<a
zA?7pBv`K}mH$o2~(*Or@6`TkNT|l~!`9O;6gq|KUD9Yqsn{(&R!NJ&Ej7-u4<&AK?
zvi4;*z<EfQgGZSH0Rh4=q2xV8O>Y>6dl0f-WiCN0(7cd8=Ldi!k0%)kajIlCa1^Rq
z1L)bLd}THcBzzsi9tHIj(B4SpjY8E1T#A#YP9bcJ%-C!4X%>rx=*iBsO@pr@aAg}r
zCkUvE0!&ykf<1$ak_m;jG=Rn+{Q$@N-t)>c1xF&01YU_CJLlQ24_Gu5D7S(_0@R$(
z%+wEu4h@eEodow3BsQa>qT=uU<!2wIq>vy=vC4OgRU!AL151t~Y&$@&5%daS+f&U>
zfu#qqUCua>ri#GWV6WH^G3JHvgBgz=)j+-;!LK0jhZ1%MxW;E;6?FjMzEI$i0NN}$
zN`|i!=(qo|h3&lvzBVoDKhggZ{UL1ZAJ1xj!+iCR6R&>HcU{_gE_%<GVcXA>UjL9G
z{=qyw`CmVTx&8Y;HHXZfocrnPLtlLv{4bS@eX0kt<vWI!mJUp3{7yVAHKaLJ@2H)`
zy|1&JIMh94Y$K=a*XZp8HNaJraI?(4e6A`|GD$?`?RQ;nIYyu`B1@KfCL8s|-KVpc
z01erD#OC?Okc14lKie{JJWqT?HDq=9%>83H1Y_Z=>x@0E^cqb~QQ_?{uZU2UGD}%)
z(kXj?!||qsdOM1#MNQd&rK$&l+u;~@%LhEp*Csoh|Ek$e&nI5(e4x^maD#2pA$=H3
z1#=+YyOFg``{OS$34Pc4BY??kEB1(GhNFhVP_g^_qjr{BtfyS^D3gJGfgK$XY>5c^
z5pNuQxYhwF>YNEnwSlF|Ye2=lQllB^rqCCEuQwCkh=qd2tP_!&o@3lQS2pL9EEr;U
z{rB~06*~?9?HUip*zmHaB7>(N?$z*h+a`u8l?oR`*E%RG6g<Jc4~1^od97o?ij)x*
zGJi8<SoZ)jwOeU-6V~UP9pe&pmMqa~uz1`-kY2}3s8%lzs+PMfEIhoeei-~*(V9Qe
zw|1P+XQ2OMp8i?Lp&EJtLL#e~SX^8@1^0s&HAJ9)68$RNpz8^fX+&MNq()HF?>c%?
zQ>fjsoI%rCC5ho&XR0c;Ly^^q*FWg5!n%$RwlF9pC$A06YTo;ev~m6Kd+hGrh2RWr
z1|uOFWlkU*?9gQ0#y>8bS31a=-z};0n9h6k3(3NsK%h~Bj<J?#ZlAuAw6!6$wc(9T
zTFeYATdIr?(8l68WLk+@-Hk^S%4{ZfNLr0S=PmjA2-4NbIggDK?nA`qSSMRBcJ*8I
zV_R?KM;fFt1Kd}MnnjnW64_%Sv1xQBQy;5#n584Vx8NTI;Rf05VLC|}%+X2)oz&S(
zD!}#`W{-c>n|+enB)xE+K&X~l(8oEsw3ebbVni>H(x2%=ivc~sw-X2E-F~)%m~_?0
z$EQdzc;fnL>r)H@K_Z6JIv!NzaF)>4iFZEe&o$6LW6VoqGPNxY)KXvANt$>CQNNb)
zTzZbRH_Rpy2usnXi7w2wsTt!wEqLjHM>Mge@;G#8qEekJ7I#h+sFn9KnPN$fzUIMa
z7yI<37K{mmq$hocQ<<YR*uKVE5d;34F6#0lCbNxOu-SOLRi=+**0&em`?fkyoC<f8
zIpeFamRc-v>b?ZdYwZcOGA6U24R@Z<rj0LqT|FR8gWF?o<JmRY?<cG9y-A!uwI@21
zS*bqi0tyO_;kvc-DTX1no64!j>hUzLF`4RzSi!P#Gg4}bEMo>e#}GFcZMr+k)b0?Y
z+tgqofGSI?ck?7$P)7)aj8bZXT@)uCO9-ngKb=j$4R|x?;s&^;DAV0Bru&aD>W?tS
zqBs@U?bh9ax&*?cee7pb;@+vAeb~~A^`^zgar*JT_-tp~fCz)GM_YCX4LJ6g$y9Ec
z>(Y%p#MozVfTQ%-8~AIpzQD`5dM>o#%bru)ZX0ELgiWPUmrvpbBD1b~;|5BwOHwXv
zEB0wjE6TyZ&o+6X80MeCO|^AfAJGhP+hR=1%nTNca8AK-Nj)Btc*SS07IqonQqoWB
z;uxx|2l)M$qwEQ7i!10Xx8k+md}vX^_u3n{%xrBjFO|ddGIqz<(KzI#*3CwJQL#?-
z!xT@=fSEAzfQ=J;G0KT-LrFb|*KSKnH?8%?uNJ;y9L4XPB<5Ap!aY2LZEyp3gPkcy
zSv1*39ZLeCt!X`-2840ALz;^-upk5;?0EXh-vl>M7F81$%XzSSXj|rR=+vtwiMJnG
zma%=~<Od+Vcw>^MOFuSnT6Mh!fOxGt+`ytSSG%#vHukA$Efu>b_jxX!nkplj<4ujQ
zinlj~?Nq(~_H}y}u`jBAvqyX@?qN$*j$c^L0zTOCz*LKNV_*KSa|NT*28DG5gV;Du
z@9r4$+|9vw3O|WP&6*Ji;@FP#R}A|Y)7|VpR*k4%$~Ia|rw#f|_Amb#l{UrC6|40)
zM`$Ct4sJ5^^*_$J@W-307estc7~;Zc-Q6Ed=f2WE<Kn53Mdvlw78tb5Wsy#PMoGO;
z_=bCX^+NpC?4kVQ)&)!^KI<y_bt(TgjIY7sDn-E7g8$t3?qSPnS9;`Bu~<!6PEwOi
z?~L(h^AT_S2mSM-$f=v+HNo&dqZsc#DN4NRYZp5zL$)!d{WId82sc=oY@@_Ck-l4+
zTL~KEWzIzv{B3}Py9T(`n-{5ueqZw>mv7hh!R<$!)5R?vCUgC<GG3)wC9AGcIVGR*
z{(ls+Tfe-KtZWsp)qZw1XXV=9pEhJ-)__kyER$(B(vZrX`HYsAZ>vVl_6_eQhKbeM
zbyqLk8|ZFlnzi)W)p6IN3AEiE-}j#H40&i3FY9usIP08C8ZonG#&_1rE+q!XNbhUU
z+tc^6`m^%QOAK|DN>X$A0oL&Z5vRGCrc=T%NDGh9;a^7&kC&U<<N78Gd}kkIsaDHP
zlE#*bdK>?rpVnur6ePdua20Wi<n{-xdQFhpa4uxn&nB@2#~qmcszVHZE~<4D4u;Yo
zX1jVm?f?6(xocrLnM|hCu{)(6l0zm(hRVznsE!H$e@|bL9OA0p>9MHNt1HZMPXOH-
z!zLfLwflcJ61a4{FZY+RGQ|nYM#^<G+_B=$OwEBvPQ6l7Js^^LMh>rSUBLm-dse%#
z!1cSpy4^%W^}6?(8(4sx<u8V8@o;VhX=W{RlOQ=2cV=6YR*yk&Zo3m;<t0-Ill6e9
z)UaLWpsl|F*Y>=rUYHee^eZ@CoJUvWM|m7(l-yw-XmW?;_S`VW_mv~z)k8fQQ-JJ4
z^&*q~A<LhzEOPKv+&-T-=OjGV!=x>hz-1oB*~zaK<a|z~F+0CwRmsY&41J|<)BeHj
zDhRo)YpiPC@wg;Txj^1_h(W%*4gu@-1&%$#F_yDP((zYL1kYDW`aapaw&xxD9<~G@
zW*j-=WfW#;ZBEF;M+}!P9WM|1a*}^ia)@kwOAf9QWttoE26x9~27f}A>pIjFFOH_Q
z_Vgs!FaMRrw~Pw<cwMp{dPkK*GuGbf@NREZW_-wGe^8%TY{rCbJSY3x5<U?ho#`;~
z=VBc;E${dww;gt_;7B+1>h`R5f;3D$6j2xdOri|N5*49%RS)fUp`B~$yrM$su&vr2
zsS$V2;0px8%Bofd)3p^#CVP$$8T4uk>_a;50yU}Nd9FKwP&`x68m8U(!ObOlWf8uA
zBONlmij)3LfnIOUopKs!F%Qw19PlVx7z|qOmn3^Dt*v<&P9o~8fj^eTkd<@4%L$G8
zW7&rKwQQpamuPIC#Q#~x+Z5fvKI`T=(r191T8slnRpCPSn$Kwv4DXs->Q!nyegM*n
z9^zJOE7u(oNjv94AV`7_mOH=DX7Ex1^q4A>dH-#sG1-Xs%kG%LM>AJuqL!_xUDolF
z1Og$AINjzlgH4(BT(Dp40A_MHdx*iIn;Em@ipQwXra@+SuTD&^$^m5CBU5ypj{m39
z^?c_e(^_S`oQJ2&&s0Mu(<;z^FM;5m<Ya%)KJWQ*s;R*UF{KyFe_K6ZX1DLDTvqE_
z?4E-iMXv_78{+sKNk~&&e*uu%80Zv$OJCnsXG1HSm1UPlvAcfw<r$}S^{<c{;WoEL
z;4_e7rk)vre3depO>&ca8Fcx%;tNXJcR8|nsS-|5HCEM}F!Ri`w#hC4)b&GJ&~a8f
zHNE!-Kd|~KGG<TcJn)GEsAmFMjAwXf3xbaul?<*F`8C%FUl!-MwEK0Ww3?R_a$T;r
zKrvql7^52y<#{<f)rXVgy=rkIqddQ3YiE67CYN&lKBpNQn9eB;ESt~Kl-L^;ZMr|5
zIm&nLXm2c^Hu<lrA+MFmoG{>}VIv(oCg*nJd*KFo%FEtp&eQvC+H{TxK0k=HP=l?;
zBJZ*M+9%8HIb-8YvN3N_n-?vm21~xuU<W6)#TAkg3m<Rl#6z0ofqCP-qnCIIlR?sw
zi_JGp2^nQ>>q_uO*)O6|nJosm7;}QpX{yo`@1je#+fPNWzN+e>IZm3^<|u1ut$eWP
z;UCWE`$w@pbuQhZy9FzGh*6)3JPn8MpdSBS()>$luTGbgw#H7koCc^C(k3P*w%%Xc
zc*VPcrA4u`25QnQZT#wm1A{!Q26$qwPg=L&eT{i_Vn-tn`k%r%>3@>|3*Y)T6eDQ1
zz8#FaKaK?UP9-81-nte^V0LWRXNZ8RIKpD%jSUpbB@5vC(s!n{y$rhg!#Cy$`we+)
ztmJ7gKOGOs_jSP%)-k4X<ZNIJU5%CDQ#t@BHuGvVQf;f<tLEicw$MNxys5K%k!=?^
zexq|^u`1a;yACl*7(GoF&VjE<i|?sguBI*95(xG+*;z6CDQwrl4h7cyk)^{6f>>n{
ztApp&nHrGAHKMRX=LyPMEIZ3t{Wb}jTdd<a_5f8ghVuYh;MgH<#?PShekw-%a?BmF
zz-oYUnQ9;h=lHU>io4y}E?=ThJW)j2(y%5q4#St1o?Ja1;CaV?FihwA-(i()b@t2f
z<qJ-l5CwKXZ;)4ma{HUcPRLn%_xcYxmCmq9$s4*;)^;swli@;=*Tm1yZ~R9P^@R_6
z%I^VSPJDfOkb;}PVA6_SBU17M@*!%~`@GK>;PSHBpflq`Ru?+p?c%N|?uCyYO%PM+
z)k4G>jRY~q*4J`89q4rs8114;_mxT^-J98MBZ$qO{uK-<w-gBdfTVX%?T@uz8MW9T
z>Bk=gLGIR4Q-j+&tMg;es`vXtt%2l}k;k?woc|CumBnAo+N<n(UQc9=L=MB^9fJ3M
zG=LVK5rKLKw2h5MT5+)kB<X7h<H)_b>w5D--ZjDYuCvLI&V|5!XT%PJT^eIgm!U04
z<<MH_9jYyef)$2A-9sK_D1sbx35s|VR+a!huY*FPH)I2x0XZY0yyn}45}Im^Q9Ko@
zGXtr{p-YlW7aD<YJ1worr9J@x0gC=Sj_@3q4Jb|z{jnd?J#~=GRfJk+rzIo{m%Kq{
zoS^z7a%|1l4JFuHq@W9sbgA|6YD$2Jy)^#g2-NVVdM1Nap>&n_BM$x*KGWA#QQ+Qu
zZu=g&dUU2V6YWA@wp5%O71I7r0M`<}^ZhHq&)+VDQZo<I1FDc-qKv1PFQ%2(stv`{
zAgz9^dHuGj8%ddheK}k!g2bT`fTKFEw|sp-pJTr|V9$Ffxl)t7K6lLZ<71SZ*ow;K
zyuv1moIQWBokWDn77p49>L-dPg)|WD6{I^v6~KaC?&Yt}4M71QJpl@~fdg=QP0++5
zetF%Exaw1cwpUsCt%c29P$K$U8A&T+MDGr#n7!*9t9!i!-3k?&KRQ0%&|Ws17qmK+
zWI`sBGtw<eM+vb<FX{I#H2SH$*@@nS!OW8diHO_=s7xhZ->H5Gk|CK1c~HO{`YTMX
zwo07by}>Q-7)9!~@Rd+PD-=^=aAuE>DIQvRDXo1sEwiY0W9ywYf&<l$GW7zeg)=nS
z)PW?Nl_5navun~Vb9fnKiVU=1c|7@EumDVR?Hac@w0#1vj-w~Jcor&61PxO0czzvx
zA-m1oT6u#67xt6Un5*Qk3%b1wA??svlRSM#Zzl_mi?nft*p{^wbT+DU)5l7t8Vg}F
zq=9e=4Hdu<IF~_IoS4^UPI@A-_WMtN(DxM_t6ux^!j$*RuTLTG5-2bbZT&bUMH)GY
z8%ig1RT2HX1Ebr2C?QtG-O8hIAiC2M8TQ8G{o(0_gZDQ$3uCUo#;po4aby~@W4=$c
z4oQHd8f$)fX>;xU?;A4Se6bgm*ekxWhSc2ZCrSQM@cIf!MEBg+`mxF+b6?P18`6UD
z*^3za=?Ul({bdZY(uL3sK_ZqS{}m)L2evHa*2^c>e}5jb=o&peJ$c(0wzjqlwH9Sm
zL;2D-8wA_B7aeJG#;hi>$=CwpdI6da%jRFn57ybQdX1+;xoYs&?Q%7Vfs0LR3IQV;
zE@0WqF1;oG9U=UcDR?s5mknH<2t_|<=(-RdLg$0DR`e~=pv<CC=2bsGRbZKVqBhNb
z`NN)#6HurT?CK+bdUkg3dH?M?Fga1DBq_P(Y#5&4#X1<nA=xik>EE6{xvsty-Y%GF
z!E32llx5oqefKu-Se_seg(RG2q>^gCF14cn5uC5C$Mg3$pz9G?PH6i{<j+U)6*g4d
z_f?2#9Ykf;oHf7{+y|OwnX>_tl(=2)`FGL@S53iZXg%}e8q+vbgpXZC8dtr)a450v
z;!TNYcr4kM&0*01;79`~M$s?MYtQOR3$$#7!ZA>7W3WG@WWfcp0kDhk2wn<~V22X)
z%{3WojMKq!%Mdsw$a4_nNnt6yXEoMdnQXSsvs*5``1Q*LB!jKS8X3N9(C&$F01PlE
z8$Vy+0NGb?cXxC{b5;j*M|dM6-?+#vU**R8uq`b~8F^wG8fM4tHM^)AC~ygOYK)x?
zJPQ*xYW6{3v+!b9u)~IwrF|!m#0w{3n!#}()>Y&&up2uau8VlRKDoyUOrOh!wN(%m
z>@NM+{Q<)dN+v0jb;|8FW*S^~a-cP;E6%X3vo@F8wq_WQaiKKjUsPO7+69@!+N0I`
zXi3W)Y2k_OSUc6jKSMXm0xzaK{?ctVDF)rj5^ykffWIgigIu;q-g=&7s|K>9p@a&7
zY2JDKJE-b6k%;3(+|0s{e{auuu~l3sIiiQx6<t6k2=qGBzDd!JlF5Aj_lDTq_+IeB
z@0oNes7`INf1NcCN{j8akRQ(T0wDd6c;CD}R0-S8>aFw4p}lLSdAbka!%=!W)nE!n
zkmq<hXmX??avD^G{)xhQ!HG!}45<a?%P#$SJ?pZyZ%X&oRwJZ!pahlJX(_q-@#YlL
zrcl*)PqqwP*vNn6TC7R_)1RT!e>}I9_gJ+wbvF4bG#bWgQv%B7N+(5bZP?1@V1alj
z{8N*^JhcFR;C;Z!wKsM;$GIa{)xDr6Uq~9JAgAtd2Q+h+AkK?bO7Uew?KKHDqzvc*
z@&?799v&GL$j74Zi!b*$g2e^x16sckml98fRI%vBct%-)gH`bzf;tBAXi4?(fT;!v
zShNA@{>q&iLb6;StjUNXAQ|jup#}=>lt^Aq?@XrO0AMZ}L01U*IK6hAgfJxdV9X;B
z2iQN<_0%Sw_y3Qjz5q!WBpv@cVwsKQFE{by`)_Mx`m!J4X<h)JLH28tC^iC5=;~XD
zcJIR>QVK_zX|BX^Fg|K<cuJlhUJ4a;C4P+tJBa1W1M)f0_0YbbZcd15qD*)fj%C;F
zSRbkw8uh`50sJA++|?DA6+Dx+@b{jwJH5TVz{gOqK+@q?(k~(8;~NJAZ1=@1OuiXP
zzP_^-kn3OoK1Go0Pwug$N2qi@5t8f5x{Fzf%wyb|zkr%PlVq|=S~mFj@^8pzt-wH>
z3h4%4Uub~UoPT#6JX{;^24G9hZ{HZ#fRiQex<2K1hu;AW<OVnp=8(OHR%w8`n2e_b
zPsnL$X?1wj$qF5ok2(TxBsl`v$ukM+d7yk@j>COYRw-M!fbD`Esi;BKJ_q?7AiL>N
zIi0R)0;=vM#@_`GAmf4WniOI9U2Fcrg|-q9T(u8>9PnaW5+S&9P0K(WtFx4&BP5w_
zpTJa!U9Fbza|B@80=SA68eE_q*%-=M|6{PPIj!lwr_6g`cEP!@0@Vj2rh<kv`dCl^
zT&p{Q68WNEKr&^3K^#h2olfxvM+Y<?gknl0cKTRutFh4jUGUUL)K%7mZf$?4V8sDm
z4Y&((MKlQ3RS}6dd<)EC%9^db!Gs6|1q8FnxoK%KC?4<zCKif}#M;lV<N_up=>2y4
z8jCVsRHr2>_`oWsq6<nG!AI8)D^yId{l)jt)IymBfRc^pt8epiP|?rk%8M^!kGYr%
ze|~gLVnZK%00;C8L_`O!w*aH7!=#eJ4P~Klc}?khAxn+)+v-TtDz3D&G%s7~%)L9H
zIfmp{Kgf%11Vqg{1hlpI`0^VQ4m2G#GGgqkG)O><dgKarwwDQ&3#MUIpfN&Ui~>5e
z34&3vz4ZN_FTXB?gq85<f2>1>`rSbT+`@!kd7NtbLPH^dKS)jyk~-SRwe~BUN&s-v
z;|Dj^*`;AU9=gtgcDp#xCr9A1^n@9RaI}G4{d>4}?He9$=qwNU?V)>C+Cx?>d?C37
z9ySzjuYg0KetoUn2q`oL<O7~oq%2M&S{H(Bue+D=p8Ck?3>R0;6te5MDkP3!E$S}F
zQqNcGBtqV@bj?yHCMLQ;C0xiYq8AXiR6O+o4M#-WI!+F;*5Hy?2gzQLT62b|&viq?
zR#b4;&dc~Y#DVL<2u%AK1bGF7VWf~`j|3W4Asw)7$CDQ7iC)W}szJYl?(_ybJzzu*
zFcb2Cy5M+4Rgj%V?&ZQvR*0<z7nTu-G!eM{1smE0lhD3qr?<S%u9GtqM$j!WW%SSO
zbj9IUXn<@ep%X%AZ-`(A4eb|A1tBzqgD8ad=UVr5Zg@7KFDem*FF45tIG^kNblZ;u
z@p)Fl3=~n-5PcyP&@3d^i)z8M2el|et+C#`_#*+X=f=}bLKa&EUM_&Ng@J>^6@lk&
zj$_a@r(17&&7~L9dION}V~%3e60Yr}>*IAaUd4CL$R?yK?$#+kUJ$~cUH-&hx-^~+
z5sxGEigrL2M`&mi<^@J6?Kk3Npo+I~BJIma*(V7zcO!$aC}9ipo4SO{eeeP;R}}I|
z{?oOmyobUy%%Z_xL18ptK@*|<sa(i5dN+|;R&bQHeSXX@o^on%#o2B=-95iU3h7vA
zdl%|hg=v9R(rcND)J#&7ZeVSXEIF&eg<^FLRJ%*VC3VV#G+RqhC~)eFeeFDSxU<u8
zV;NCOfcuHR_TiQqBmhha1ikTf$+;nggx8^F*tR>aSnjz@<TCGWr)8W+kqg9UQ4+py
zA8fuiNw8XMeTZA|V7#VuQ_q~uc)DNtQjZ#_IK_*9|0Xny=>nA$`r~Az>dMiAgkbCF
zlBbW%8~^wkpp01|TBr8=Kgudz&!=4bw_pX$qqd!KfgxAO`<GeQ^K$H_?zcMvCb9#3
z6CoX*jvhr)xs~y|iClfR+2G)aMq*~d%n-EwBilSRYEdq9$T4Y;*8=%#snSE*vEpS{
z^j3a!i9(TbQ&aG)(uSmm4hX_{hvq}22j}bqLMDpm%O1U{6H^Oz%wPGKuQqt^lGw<4
zX#SL`B>4T`Pd3Tbx~KF+O9#*WG@h<pa6zdn_-a?UEzwOVWjD5=Z6)_7mT^bwS0=iT
z1pRz*JYA%(?4nASU(VHtu}vZWcoA{p)ibz*Fl6E9;HA92$44^eD02tQ#qVb9>QLv7
zKeyW?*dD=q4<z~z#_sc~`5yV}XEjJO5#~*t15(89m!6xNCJ`Bi&@cP|b%!)fA=o8X
zqERQsH2?UOuy9eWzl5}{j1kZ_xe<UR`-11+TRwI8F(W?nwQa{J-y>?KuOwJ+Jtc>^
zp(>!4uD~!u63!bc^zDsS?kjDI3m;i~%U{ngzFKFl(?6g8O84fNe`BJ1UD=0sGM&*P
z3h`|VO%r>9`6mYBk4ZP_Ob3nc=Z@#v+0DIZcjCCN4y^}ty0m2+6keOnL*|=wII~LR
zrj{=aNdxB(Fx@__%JS~(gcvcIt{dS!hA&7kT@LF#SXvl3*yBIXyGbHOSQ5`jNouQ+
zJo24|bh#|Z1zqm`RY$HSezK93Gp7mKmq5No$Y1DkS9D!|BIM%cwWB^4*|&VnLu;IU
z#6K8cntZrpn(uZ$Vy@#jAx6l=mj#EC{B@~epm*P^xq)8{C7&cH2V}hNn0?sgJf6|+
zv7yp2G6MhliTiuA?cUruj|Q^>Ey#nwBaJGO9a5Lr`&&133C5g`sB*Rr9*HVUsc-lz
zWe>PO^YW^VQX!@IBK@mE%@?{lJpUw@$;_aZXtN$+`)Nnc@EoaKv~C-N*uED5kWP~{
zAhWjU_<uLI`mPu8cj01<oA+B4%OMf&F|+kNQD`t1Dy2+j18v`3M=Pp`{riVgY0Slw
zR5NP59OKRM!%^q+gqe;ng1Q?RbVKv$O7}6}5S&v$T=o|j(Vutg6dW%|3??N$U%1t2
zNpj2AmkM$7s>;;ZH2>-8Uq}EzMOlvZ46j}GYm=QU%f4)Wipyl~lws7nTzd7b&lpCV
z8;_OIJZ!X7WE=Yulxe7@caxGp^9KbX5$Y}aQ6;2ra71i`JF%7z#i+Yi=mRXo{T`<p
zUu2>9a}2hfK=8SO8`##MEIu7ANc|wV&My9k_uZ5iJvc)y<yOTX&kpd4^z33RCZ|U?
zvfr4q?V;fiidyNgzQyN1*_#(Y-43NZvZ#n*NH5B{K6PBBb_cV^{l|19(-QI)Z=?}<
zU(=L4RyO4>8RssEcBzOrS=>)w{ek6z4^D8jqdvpu8}kyel7`u{hT1wmu%@BVafq>;
zDa+UvDNVKM9{E2|FD5r>P1_*Zm<rJ;_hm1V?R2d7=Q`<!?|m)gi~Gm^Lq4pf&dQW+
z+=ErirF*R!Htvxw)Mm{P2oxFC@A%mu7SY_$>qi2iE5<bXnC&%x0~~?o55Sou4oz0B
z;Sl49+q<tNk1*nUbb<Dj^^^Smxx)T#jh5mk1?I->Pcq&7o1<fMmmb|%FJWMsbMtS*
zxpW?{(`U%H#W`Bx2KwwrR!wU!5eVZnC;Kn>gE2OV@ti-feXf6zQlJQMnDq<AW_4B9
z7jzIm|LW_OZsq<=&49Ltc{%2B&#jPT_N(si6Yi>$(F40WT-SSe(Ny{f88A2oxb0AV
zvZq?p(mzq_k^qPk$e@3z9O&CpOige9uzbU`R(6XmC}+ESb0-E9HC3CmK7vW|Ny&^u
zdh<ZF$jcb)x-L3<%A)vJpsNOJo3%~A53Hjrtiue=c+T_PF;Y@$$^7xHQF=?qbuQ_`
zy*IklYL9D2RD(ZNLGIV}rzqLkz8`42lH4}ZA__m{x}U%AXk}H-n9hX)k<48`FY1D9
z<1VXP=|4VTGDT=5rLP!!s5S}}KHv*f1;k7bW^SLTji(Jdjx00sva3p5&*bX_EV<3+
zi$L9NBd?FST;N-#c-cyFXejf{1>69O$&@q7o-~vjw<ENr8R9}N#CGpo(zV(Z7zl8l
z9akX#ek1kENckFM3uB8%`4-Lp&PuxDu3ciJpLKN86z+JGk9b;)=2);hW&+3PwmMMM
z_9WLJ+mAZs%GvqcwL>OO8>)_zyByIgvyfREOjK7jOtdzYe`27rz%}A^mIt{r<;c>=
z&4rKXpPe(0TT<~=dB@BDa-NWf8sh;yH#n3gY#jVm4&AFwH|C4vskMaT1)N@<>JC3f
z0jDgaSbmN_zlu~)-ry6rrMvD*7rXx5KwnScBp>KhKHT#la!P#kHz*B)j<=)Uk}rH?
zD_x+sbG*OX8Z6r7w3AIoYBN%L?@JG9T56CoGT62G_TT&#v4I7U>`}QS6)wz7VG`bC
zo*VH6oQTVJfgX0+hba^&<Kh^D8qr(z2?Ub#Lfr0{XsjABlndTixTa6FS?p3zq2=_O
zzd9|&$8*l9tmwo&CjDd~G$O;fz7Yo2l#D=Xcbo2Pg5_-6rw~^+YCG2d`7g`AT+xm8
z2~JI-n+EsY7%Bz$*F)-9v||{SD`ly~YQ^t2;Z<YHjl2ldw9Ypm8#@m(I4_d4?*%&d
z(5Bhod3ZSHuXbhRc+3?EkHz8a<@P66X@Gg4)dSe&u6CnpL-XFK{hiedd))(i?^5Ek
zmOeCbsblv!AF&K3(~!CkfP#;WC_pLQuq6BVuEyuGk$uzG#?wE=g*S03rHOen=E4XX
za~)r-&&cxdoV{sk(BSvOLHuGsm~$F&r#h%n=@Adlk=NLqZO@Hm2vom={uTs+j-l5T
z9OKqe1xRXq9HTTjH?q95-#z;h9JicyQue}<O4A>&4+tH%3WrvKd3udE2S~ukjs34J
z8q>sT$pvM-rUAw{CswG6Wm$pr0-ytyVoWoihy#ex{`xztWtN9F2E%_yr=AL#xjdmc
z*+@UgFgI_%msv8gC9CeGs*#(}1-O1rqxXD~zT{AoGM(?HlcaGObdU;bm13iv6A>A#
zzzR87sKZji4@$@0aqPjG(cEUfn#)Y$06MTVh0dd?`2G`<{Jr7sBP@xb?z=k6gS+x_
zs!E-0q$H23rIkN!>l0sRQ3pcH?Q<i;+KqdZ>E+KOr;dsY6twm^cHEm{WFBF3cf9l7
zeQ?2(3Y3y#lzh0K%w*oeL*cp~Z`bqTz(cAHfY5F;=++kb=W$Le7Q-Kxuy&L$F=tz+
zcIkj?-?Zl1lRm^rIj8M%i~9OfP!Gz2!>8g(GyL!q?Z&u8*SRu%21UzHK`l3OYEpGR
z97^})bl%%Pd(Rrg#&Bc^goJyXH`qz?r)O=ncxysiBZJ=S%=`zY`H_ElY)n^EkydHg
z=t_H@t7DYxu5Lxz>^%T(3Pbt|hxXg&T<k-B?Rk@=?>945c>M%{5GU1<^NNu|n?5!Z
zZ(7?<>RxiAuB0)Gq**!z^EGd<OVY8=vYWu?^*7uj+t9P8JK<F3wMV|}w=-4Fy!W)6
zDk+jM%@_LMk;KE6va-RlRMlIvZG~AwJ9poFH1l-!SJT?%U*x6?3~&G%b*;lrx_d6v
zpx6h%GJFLc6nqU$gg|zs$c){<-AS?4&(QvF*#_P!^l{#$^F_&#Q*C0kl7Qn#7IR)r
zWipGXLCKMI#2pS%tMg0Pkk}Z0Fd2}9YzDncz4_lz!u(Vnw|dXoq*&Q-NIqlFn<%%k
z$-$0~65@+RYXgiip%>FXopZqr01hlVx5h5#!KvvPt#}g^df!g6N%qCCsW?NN>dvdN
z-<~t@2L~#Ly07;G&zgo7e)aSuwYPiedf?Q%v2Ea#+67+CUwNBv8JLzQ4DTw=vH?sc
zRkt;fTgq`uxRKpH-oJe72CwV=9qq<)on+a?$-dTr^XfjyVc>z(|1G`8c>)JhQzT5<
z8}BRlpU7YMsIs?oa9HTND!j=fonO2Hy3RI-$@FB<_qaM|^;IwEDSLPQ#bo->O4432
ze#bY8zf)hCk)*_vRgz^P-9gn<n`6~$IeF~XDSmEuHmYRSHmgBuvKRZAtLqR8x^L1b
zdxhWBn{b0`$sYPpHWB5I`CD__-~Xka{qCPGC!hXhZq3N|c#wSh+)w9|$9uc13i5WI
z_gk^B?)Fjo_sO$w&)oW!m_azQ^IuM9kNsQe;(onQ-NC$xmG3?Iq@Wlq*Pk7fSp6xu
zaB{9Ru`hsqGkqV|YOeJnr3o7*=G-(d(!TFc<1;NI!zm%Dc{>MM31%c)I|IwRjy!#9
z_XPFrm<p*UTFP@FGpk1mo$;n|P7z%{+c=k!%=6V8;><U$L3H2@XPJ&jl%`L{3^H0M
z8tGa#Bud#uj}ZS;^4vWl!~dnsoe3^IukNX^6};OB_7Qv9t0dcar}7Kz<D@!De*IXh
zGx38;aQwPjc-KB-kvh$`WA8HdGP2h*#C=1p?+xdQFimU8(y^@<NqFj{$k;K8N=9X8
z?H5V^z&QLjV0MC-Gu(aBa%Fn7%CflBw_wa?%x*-itL<;cHIc`J|Fl9lT>2nprc!*c
z(NHn7axp^0d&!WTLJ!VlcYb`qSjnPipA251w_yDPl8LZVBi9W7VSCBiQj5O0&YrRA
zs-v_jN50u?)9h3}*+qT*z52{xTol`CQl8NbAeV^z+lLrun0d1^kLuc|vscNKdaf#i
zj>F%ceA-6ieE4F2O4jXRCkjh<=NA}-`{~xt<O)J2W;K0U%_5*4mvmo^SkHPSA(f}F
zUc18_um!9fk1N5JI;d9k<KQz1c4B?AX6p1?to79keBX?j2fyrp9BJ2Aw`jBlu05+b
z^lW)`pm3o+&KgIbST5EePW06i)+=`y2igoVb(&|#?(2yG)NCECe&VSQAtSF?K6Ygc
zC}@8y*=+h8&M5^;o^)CY+@c07aN2^@@Q}ZH*<)os!*b!8{jM4NEK-ogY{<}R?80p9
zUZvnOhTkA;27DC6!pF1TS#bg9+(q5@{54y~5O<1g74kIpuH?AW%HIw9Gn1lBXAOr=
zRqMoZWVUD;+f&q8oSoe5q>OUW;folIK69e2r*LY!-{2=nt={9L7Gs)MMUuoFT(Pms
zMa8vW!RW`WYNj5;Yj@NwE->iEzrVU|fRmP1V^*?NT{8;Y^&7h&V^+ysio5^pR)7X$
zdV3x_Pu5-rZ)y<x`e$}55W{Jy)!lX*#v#}G)9|0&yl2){a9SU`G#q<R4dOPcEzQIg
z+ry<lu#6<wMCn2%UGe>^e;DA(V>t0Ut+~{a@{VicGK_E|UP`H3NT#Ixiadi(jxoJG
zM$*MO8Q=!C<F&i%tb?ODQ<R2nn0NfSmk}nMK_znI_FbjThf>02JLX<el(;=PioT}h
zT7RtXE%aQny&)ryors+@$bMWo`x+l?W3Vu8oFvOQBg=rM16Asov5(OqqJKqR453gH
z)YI+lNw#ky+hy^9*)B4zwPn!DWf&!o=!&Z+buQHIWYGC{ITKjvSzi9P_|;r?QD*7g
zFEIK^iFsOu+?L{O5StOZww^<zzw^IQoV)Iw!%g!w!Z8^CUJnUHGZt44J+$J=!i%CC
zQzG_LIns!{%ZFIsSlWNrqr*z0P0x-sa;ZU2B&*yjH3BU8Ltn&SOu|aO{b*%vU>Ogc
z7*6;^qwK@1TmU@9#>~tusan!F^HyB<6_umFvQI}-36|Fy|2|Sq_U})3-_@zIRPN0W
zk5kY5owa@}4j2Ol-jt+OK+krkmX8m}Zu7Y2{wM!5@21|Tm-VK#jri5%O7;{!bN@Tf
zU>6&e$yJZB_gHDRRhBy$iwS*A95IP;{`7lT&cyFGbgFz029~;Hjje7gPPwX<0BJ__
z37y?>8o0_eLsEv(oy6&$4*thTSfM5h{9>8TxO9RvV@j)($zEL6-PQVQ8nq;aUF7z3
z8z_p*fTF-`j-es1qJ976c;(=<PqCEFu`_XM8N%hiv+xn&ymHO3KdmUka>A<&{qZ<S
zO09?;@b`b@=u#{e5<N`&YH_-Jk9N1h&;n{2NoEI>oWNmp;R1VSpWs<mQl9Cd1Qn9i
zV1k;-gN-!@D>+C<t(s=cyueCFn?e&r1pc;xqBKj+GUV5h`&bj{LM7G=zRa9dTE-yR
zQ#n+X>;P6@R(ZDS`%Vk#LUp_Wu9a4A<aL(1LLEzVEWox>&loGpv3S@fw`;O^FFk4!
zw$SsBJ$_$cZv5^P7UT43AuMOGLm9Lc|NV6Rwf-HE8~Mj%dy#9Y(=o0Ykq`k?hV5>%
zl041RrONM`h5XY)qw!S#cjCJ^W%9i-Z=5YJrGz)6`!&?&Z}aGCys}0B%t!GZRtS%y
ze?PA&tKq};-9sZ?i5~HJuk5TH8%MkqJ-cqno>@;JV&sj{qgEfgOPieDt8LE-7(uFk
zwKub4>3M0Za(<fd@EuwYf%~AlC6BGS+;h{r>D#$SzN&$}EB9_)Qk>$%cD#QoVRe5!
zbAwSyFojgn^y{K1u2=OdKMqSg$t_;lsQ?%2T@`<(*Tri)ZCx-b2F)Oe8LEA4^~F%m
z$N1kZ7v9!ZS!zfWn$1$UWqXw_o)JE(+lN+A;T$k_GjgC&zB@KiSu4V;sLIEwPp3y!
zL$c7@aO17IeK_=})Tmu+La|)2Or-7Hh5mHBhH-6t>yYE<bn4~_G^{;&;nUkck7yDv
z+FkqjdzEFEWD&)=?2>bpF|Ysl`nl@e?E85OrXuVbPbaZ{d{0l}1feIwtM(qXG_p^l
zxGbQZZn}8@oexkFwRG6Bofeo?rn+l|s~3?Y<)-v~j!9fX#%u>uc&^D511P{WzhE-b
zvWH7f`cF&@i4ons->Jm<jHdsO^d^T0_OBn^n|~}!{zAM@Qx<9WdFozr{N0gLCMF}i
z=j-kv=7tIbvb}GUXWt-~V}d`rf_INr_NeoWU76_U2$*G$?^sWFV^+=}2B4<<dl>Cb
zC_mJ6B;c_p+gz!jcFYgT+ZociLL0hc2YRV~DRJVOR9K@mUAZaUB|dM7>)RBr+He|z
zpxsjIzDMXU4`KHePY9%j>wQ5Ab8^pBM_i(neR*%k_>C($J>BD-dp9^`>pjFN4_i9-
zum@*D%Crp0ww=dGVSB9|XI*lXT$b>gl)A15OAS3oT%ZmhJ(nKfMd(krwJS87^`Fun
z5gYQHTc^|xGCI<5F<@1%@dJfRykfKVHhoLmZ;BMB#+7pVC&%8lTwQ-w;W-^*c+qd!
z8CEeG@`Xw!%bZ(PmNdVFc(s%!7TxyoX9pl6c<nWL&E*rj6LS<~_FKL_PO7f-s}GhQ
zk2(5H)n|`N2p+v0$)Eq!?a7iDv21N!GgcW(&O?1Y(90J6c;mDl2=?`LYM_zdMA`C$
z=HkjJaiUT8#h(Y`wWY1co}Cm*Kt=N-bg+Nq@NPw+BhU=tPetV9Yd99y2UXXH$0>fM
z>W581t{i^yAB|4Qp8xyRh+RhytN_=+eOsE{Me>^xGK>C~0=oW|=%ZCgenIj7XuV6N
zQLKDszFy7UIKR^`ZIiC=e?%0xNS?`Vv%K2-muBEl4*a~f1Giq?$GnQHRwm?XhxxPU
z=Pv{Xyl=G(TK$l2Tb6Hm7C*cEP%EceeD?Hb%&ISme*Xre)6AQ1G1){VMPu_U+0SLG
zoY=+APH|iEU7R*<au&TUAeJv%vp@<-z=ro*P#@nq==eAL8>@Tk346>9dBIzwh?4;-
z+na3fDc7-&HFl2A`PJD~k40rZv)$yY$~eS=l_xTkk(d_FjRu+($7xLl*FQSi!Kbg}
ztdxvf3pLvf9l?7`d}P=Ivtb@hW~w@E$4R66&6I|w)naoiA8nr2T>)LccVAek6KJ-X
zmrT%X<v$ONO+`}_otN)@_UK7;u-~V$@1M#FJmzN`U@=<c99cXm))%$Xn{V%{9+a}l
zFLzG~9!?4L@tauwJHdbXRHSO-{gNq8z1k(?$7AY@O)78%OmNx~{Ilf`WA=`(DrLlo
z;~zCN9Lo#FW1fAp@)m(x(MP_D+zxV}nE3qtt!~dx5&~}qxmQE7OO7!QeevMTjI2G1
zad1RB+o5?TeVo-}(VK9pZ0U4=`mHV_2ttU&y1mx<LZuPD9|^(^W|T>0{a@aDoE9i`
z;zakrkNF+^ge77(-E)%+55o`{t~+n~7(NZZ7Ew!YcgPg;S0`|``~5aL*C<H4R)lg#
zCRo%EOX{|=Ro(vdPpwtN*+6GvXT8g3FO5Mj-Qqf<l$L|E>+EeJ$X1-PY)+q+Mt^}=
zpD?shC_;un!691KF8bt<Y;ETdTcxqHf7e`1BDuS@e<Qb_d=PDDy13J9s-j)8hQjx<
zM<HE;x=N05LX`Wi?)-%J>&Fy|6+D{rzh-!#*V-AWw*N81%S6GWapL5XqpB3n<8-^C
zB2RW^c>!rw+hg)-#JH=_ZFjg7d|($wm+P9b(8pg28T#YE?aMZYB$PbQI45aqEG0h8
zIsrQte5Mab!<L$v&|xB%67OXI6Tkb8lWHf-6cMkMv~q+Dhzb4%v9;CIR1#m49MBr)
z1!Ihq8#_m860#?6EjicEAKg6D5>dgk2<5~9FhU(nq7@&w8ZlX0COr|C5ZPVGU~Qa$
z5h9fC0U3+TdIYLLNF*?W<HvE{jrVA&#U}2X+ww-nt(gb_MT^@|=In-gGw7BO)@-al
zuRc71bL#y1OGS4CUxEWj3i=dgUys_Aj=%(#bH;nkvgbYp65A~%@0%JJwywJum^ecu
zEDhOEKVELLUm+m$NUA6MH+N~rMi=jFpvES%Hvfj0C2s7OZToBBTEQ5EDS$PoU5OLu
zHzSF5A-Cw7A%a2s359Eozy$YUMuy7v!>n+=;(=Cs9W6$kkYgO442zAyazr};SUB}y
z7v)e^U=LSpk3qG&n%uUPwjVoy$<A&Bm!aQR;Sqz4!YqfTe@n;TnGBvQhDF>jBsiB7
zeWlLCHE%AW3#r55;cYteKi&pxx+b(QfJ<2(95A~qANZIQex0{|3=U6_B6w$kz*Ul~
z?<IKvHfbN3nDC(I0T7bdPV;T>-9~N~T1DLu&-36G@SC!o>Uh;5%hmq;rM|xBX91&e
zb1>T4<D7807f+z`#j+Yy`{EJ25J88;i3!jLu<(9Y$UVS0<SaO>Q?4Qu83oY+*9^7E
zk|{6fDj>qV$B1my%MJK2MPd>KHh8bOF;w^kQ4Mr6vBI@P0BytvjHVWT_!Bf(L#&cC
zLh~xYXr@gb-g=(~$DyAJx9AK&Ujd16b}=PnF{SYR4e|fNKL|b!qDXa#{tmrS*(IEV
z@+@e<6CXHpPvLviEg=Vu@d&Rg)O4XQ6nxRexDM~&+NZ-IpZ0{@MVMPpHVg$xfQb%7
zm<_Wj7(0;L6?*OijQC_h+e93IP4Yb1fLW@t3Hk($k7*g3_DS@sqOHhXTkHc|D$Rdt
zP+5)+1H)8^2|nfs3VnDZey;de|08(o*KRP)VWDE1_8p)`h+BLfpmhhBX_>VEEbbL@
zM8#R;c`2Mgvqv^?hfh4&B)~7&_;xsTz&LJep@|gzb8QKobEpvsVUGGwxZn;75bFT$
zJG8liA;Ag&(x?HK=v8|IF6c44^AHId;+x>|xI~Nc#Wq=^4Uck%6B3OBFQEY<wxDb#
zV*&A6m8E=ON1az2!NLhN%V%PLd`&nmPn^<OJO_7Hb{RaFmgn>2y#eC5c?O3B+X`>9
z@&?@9z_(e~PVNU-fWD=*dmaG+W`jn}l4_rO_FLWm&pPXf%2n;0g+EU>C^c!)U@dC)
zi!Cw0Itd!8a0O!x;Jr7~S`}9~XonYv+1BI+7|{!Pj50vy)WX=7OTU9&6nsPnkhI;n
zYekX+utCW*L%spfzFmh%JPr%NE4^SO84;$n$QrGq7BM?4E^R|JR%w7vqC0aXLt2H<
z#34?EJ1}L?8URQ688u+|Gray6xEVqVb^O4!a&M1wWy$`A+Djcx&H!iKWVG=A+FnOl
z`0@_8=OI>amsYOl1GuO1lR&JXGeeks!ITTMiQXF+r89-hMH9eyyN;6pS01Yq5)6}P
zHdR{1Y2;ig{8JWCsnz=f-8_m$9#>hCuT_d8IJ_f3oIN(R^`lXNPMas#FLky9?&Ib$
zVBft;Qb=WbLl2aYc`)*|NT2DwgptcU)sQLGg{5--68J%rCEf_1zt5j@_p^;{MZBX|
zfzYl9;tafE+b`%MDO~2&<p7j!)5lh@x+;Pw{ICZ>P2rXR6*+peDGw=F)XnhihRBCD
z3?kzGh=7_VK_9*mG~odZPG8Weh!nx{>LfpS(_}o105eeW9coHIjT1praeZ#_GXOF>
zW?}0?KxoxXoui+Y`2w!Rb;w0pM$vDg+x#mAf2Mr3JqQ%0#fFx|hzYdLcX{lfL8T-X
zY7YW&dwPxysx%fx6my_6_$rJ@>1+8-z-mxjxvU~kJfb3?ObSuwBBED+I%=0etOIDy
zR-x2mxN4w;latWZBx-PD6(lVa;Twgki`>mqV8b&G_cTVQPR&@)=1JRjkvkNukX}M?
z?QEEYrmqGgX=<U#B!*_Jo+`BAnBXWl4qY9%tK=kVI6+^D$brFE<wu~=uzbe4d{d4v
z3CM@a9RoORy#PGVUml9cL8b|HpLF~jCcCKsll=rdTya_vX1(kK!xFKbfnaGcU?TxQ
z?<Y2rGZ@kT7=y~b`~qev@c>*Vj{t>YM7WDf3$CYVf;I^D=0U{iql0(=)3(10;Yo*3
z#XX)s@MLqL6-t0wRq1m3?9}2iNl3SgHlj8HXVB!Z0@SU7S{t&024ubhRup_HlCPj6
zkOWx%&;iMX)2l=L<)nxl1lF|iABggQC1dwG2x^alPpCFBwQtYB0GJECyyyZBjJ<M?
zpZU|G<Od5I&y&8;Td~PL!^LatbG&e(h`Y`{pe;i{ArEP6{{EeJh=Yxq5eHL7%@*Em
z)47MR)q?IeIk*w4WOXKMGn*bdh*XK)OF(@l04;GyM#YGLp1X*EtvMjvn^L43C@=z{
z55^wHdDMP?b6*Q6;HK5FCf;?)W!+68P|+*ZpX3d~rA!TVcC*>t(1o*Ux)D(ngO}%{
zr8bIRm)qY8B;$!?S&+XmT$YKBDk5j+@pqw!aR6B9CR8V(s{$LxcZ3m0{wzZ~w1!7C
z<605b?&J-W1N53A-5_oQqZtuw&<hZ~$7*Abd;Yw-ki=I*Z4EFRx?AYNU{*+D4yb4R
zEsT|m(&10(G{J0|x5r6ve+CQ2^LjP^xNWmQP*HIXsHUF>!X}NhZ}4Jkl1d{u!oBn`
zxuQeJ&IqG$%#QRmsyz)FF|@A*dx4y}+Y)gI1qxzXCUn3R+>l_(Tdag{U;#vZ0@yPg
zgat>aT}ZW!=UZo%Esi2{vS}4VUm$xIemE`BF6!ivT&<6vefx24UKYh2sVg^V5ZF|i
zf&|8J+qoni*QcjK&9zp-QgY^V!g72UI`tvV5D&mS0l#EsW<m%ldVDov-B7Cwc3jeF
zmcywt^x|7gn7HPo1Cx9Y=xcww8Ky;^(mFU5Q}G|<$A@1gC0eglowmm)H>x8TJTmC0
z?Pimd4_-}KeG&RKx1%_*!AP<&HR{C35;Qpm0TSIzs1NmzpvdHaF@)R{?t^Y6e&VAP
zOE$D(^3YTRWOFt%L=Jz5O`E#loXWA=upMfwH`p}V5ks$2)TA~mi=7{?9e{#;9d_Hn
z4pm5u_fwCU@EoW~?N%{6daq)dyHQ%{cGyIXS%D^m>Cile)F|MJUvKIRy{=@sfgvtX
zTI~SqDBgETglubRu;4+Rq6X!#t+xGKW&~AKe`?H}8DM4DOI=@~a)p;0)lW+82X5G-
z($cvJY>c#87JKo6wGyU6_d+nAcSJMrqPbQf)02%lrcuiiV${z#h_c1%Fw`Q9dC#yR
zWw+WB>6~GaEXf#C<BUAHvFfi?t7T|GkCL)O?mdcvy${qtDm%9HmSl%C>n3g42{&*F
zuT3D0dtBO$8J34M*WzU4U}oKffo=O3`7BHmrJ@Eh)E~5&k#x8w`V<{w0oER`SP=R3
zP3gi1bT7zxhm9Mlctx46YO*ZxRnzbLX|$jpSkP&GiImZlr@jx9`v@f(N;r+0!7Kd{
zIcT;#m@I@!{)`Dru8O=oX@Ju>Sg1mgr?0lqu@+ILy>Eh;ccH@?h6vEhzBS<Iv}4h9
zC4;F{OtP)KChvsDY8R@oC0_aa--U34)uypJm)2*<V9=l8cRQ|JYSRE}^rOUj5DXUD
z7tNWZYPP;#xkR<GGFZr%O_A7ZfXg!ErP}4k99+0Yjlp*Oz@1!}#*!NsKai#64S5~W
z)&C6D-_K|Dqmo|Ehxgw`c%gx0FvC(O#xB6+@tIGm5bX#)6~oHDn9NVVq{|=la07hF
z8!P#~%M^Di6iFnZb`!;6Vo$kGC*tv$FmuR;WZOPlA<}J6>hV_Cc8K9d<4oZVZjMh%
zsg<!8-{^C9Nfj!xq8RjVs5W%Rzx5~89?*F+R@`shn|5I09unP6CdxicX!1%=fn$2_
z=mBKekqX{q&Pwa@#Pe1Lv57N`mWUaw+GD*3(-v>6lFsmourL-)OPaMe;^sbo?P$m1
zo!s_={*;?1MzWsdttMfsrBWsG`48wH{-ui9+E6QPF~}}2&I4(bTzM@CeWsf+93+aN
zSO&A7W>Wt27;DXAXORtbGkgeCz->-GUC}YXeIs74`WM~Q*T~D@X2uY;q>x<{VCDsd
z7$aV}y4yIFlR!!O4`y*9)jFxRL9$S-Zjw8F;Dz(B!K}JRmOx2@Xn!Wr{~RtrtY}3o
z@OvAe&Weqr5lEEdENKn5@q3)#8j?I-s%x?4cXE@`^E@W*dPo{&r&lh9$S~*)5<4)9
z%#>p3IU9;e@F$b<nxb#HVoniBD%#ulk=K=0c0I1vA+D1WQ_j>eIjkj>V1g@6W_QLM
zsA|M#XIC!XqST0as{B{0uk#|wiR!JY3ZiR6-LCFtWWp<sgE<|PX8|3f%VU!v(c8*D
zIrOJz47nxf@n@UQbTy4U+IV@wv|nm2O@8`dI=koyt6kl#w1l3mXRmnbUY~Jx;LVvV
zkLP}r*Ijd?Z?y-%sF1SGzijETSYzZr+enQ2GRiShr}MzJvbl=zzU<atdN*E*T{(}`
z!f7wZn;C(mj4<(^{0>J&W6RsW^L{iuzW-_wG42t&=n#t{k#E=9eg<>xXkLml%OaXX
zOj@*LYZv67Ly6OYin;ywP+6d5K$sK9l;cAZ&IeqkvLx79?)R2T?buqkwfz0%xhtMc
zb2!FWi^3jEaKrrp5^u_cuW7jIgrExUg-y}1euE`$-W4ht^&3Sag_el%ly3jh^u9OP
zT=wEw+`uQShDWY)0e<3No~Li-Rg*GyDc6_cSQ_f-O-8)am2Q~%ki^oZjw9-R=gf6V
zA+_a^)gv1yQEX9EHQU|?D4n%WxF*00w4TPy{4VP#Uz-(U0D~XsR-{U$h8roSwH|nN
ziApwmWdeqU*^|GQ3_X|0zJIfP>dHjnXyTg*XFfeL_L$UC`6a$v$nflDK&lWhZIOC4
z%Z@$i@%>UJh;OdSQI;HjzsL9Gw>~lGs?usdunehVbjKxZ4?bGTU!AoDAAFBfgjH>~
zYR|jUvs%^{=WT>}>&P|FtdrQ&b61)iknPESGh(K+G~E$f(*Ic=O_Z{jcD{&sGj9SL
zS#YPP>p(lv@uML_L%OgerWA*NEUk8&C9aaL1)9qq-<XxB-Ttm9a5lTQYUP?Hx7_cN
zidE9<6DpX8Z>M%?CjCBiq`tP=0@9ER0~*Anf(hEue$~|>C@cv+p+;?OCWRVTF9iO@
z?rc3VdnwMd;ep2xHYADji)k%h#4BK_GqBhq&aHATz4eo4<)`2+3O{KE^gJvaxKse~
znlGg6Rwui4luQy+g<e+kS*|opL8&6Hq{fQ}<}(bZqjQW57p`AS!^Y4}?e21Nt+<0K
zZt&`^k|7QE56XcAn38s|eOy&_SLk~|PP%A1Q**YmDrD{sUc;z1UTM&A)Wd8u&<$mS
zvdJHIT3LwZ#gGSmwzd3;4O1<NcBbt0K0$TQS^avZoW@aAw(pI0oHMO`oyR7`dG@?`
zKBV^^HV~eE&F$pw#8L%$&FK^U>EpY)w3Yg_l`cuoR7WbV9}MQ9S;(=YRdd;8j~UB(
zrRgS@aUtG4v5vu1GIoJ=0}7Jo4fe)y2JowjEXiqJzQsWpg`%Ny=?45r9HpF;A)e&B
z$ag#b>S}dE!#^FaUHeca?NvQAae9t9p|9q2hNVsJl`bxEg#+2Nuc41Z3>f_ZmiFzY
z!DL2DZ)TUh$7$Xuwl6J@ZL|RT)RsDVaKR2YFh{ew589fnRY><T%9u%uJ6`9|J@rdj
zA7+Q17B|4FDQb@MJ|yt@dXjPkl9YCzB`GoD+(=Uwvb0aWPyB^;XS)LXH&*2xAE){b
zGxpsFz`5@T;2MrFf>u-$f)<l@RB1^zFNsuq%n_giCIB7ahtL5#JEd}Z)nxh~FUD<1
zb?9n~B&>UXVXyi2M+LE60dJN<Q}bY=otippW?`n9H5td*MGNx(=C7^&iMxv7r4^Qo
zISk977a!6CAH09iaWMAN<Grz~MI)c5>YI}SG-h&c5r7J=xDYRSu8ei9<k>v7H7&lc
zZZ=N!sI$Vv!qbvI%yd@bL}GXGyWZvHRvj{Aw9DJ(_bsnp&#8VLxS37Z5z00>oROz(
zVPM(*ePCU*+_RJN+&ZgT$sI|jdBlYo>SUVAr>RoP_=<ni&8d~G$&%?%_urjVe!QuB
z^-O=d%aTc5x?fZ6Z!Y4J9FcW?(G77aJzw-98{0+B&38BK=$gr1y=u6yC6m2)4L9%@
zd(%BP&lcy@WW}ZKxpkqlE+C|L>SAE|7nl2#3Up|wSa6@%eys6+@bvW9LX}pl@?)Ky
z=hwY#OtLEIG|KUEyTd)%bVab7e_fn4E=|{Yqu}>+MFIZWh+0ZL;(xaZa4rGlIagYg
zsZ`1oo>93Hyrub9RAx}!Uqi7kkxy;Nt}@;mXPMFn_5QpW)vq_W-R<`OrS8qcn!L8Y
z;dp4Fo(h~E>wpN<8lVUWGKkDtTQQ6k1Y{6U5M-7yGUKs|76eTY2oR`RKmuVD5W>`=
zXaEU<gfT=Vf)EH2h5#YqTeq!1pXYnu>;10x`lkHRM$LWS`(AtRwb$@lza>0ycO;$l
zT7n&=;P6U4sx|QO*rmnGYV54;;)!YBnQ0gQ<Pi;y|32yXbAHovz9IZ;HC_|6NPYFF
z9{=*borI8;Z1dspP4;QmE*NyfgY*SKcm3(Qz=Wy<Fa~uxWc=DA9hj-%FBq6eoZo8p
zYr!6iM=k|?;=IsM)a~oboRRSgB!}M+8U~qJ*9OR5#6tFBVLjIv|NAS~yl$r`2je4D
z?HF<7sMCnjEYZpKW3;cd?c%$U!+@wU8C`jJFw@pl3NSf6ZY8x1#VJ2WNi=(ss>bfj
z7~v?TkM*l419O~b=6mr3MOueM!gSY$->3Bw9w8m+U|fHWQ;#0iD`xhqW54S!PhTj|
znVr*N^_k_3;*qTqBXfXire91p=O&60VDlmBotQpMPz<<yKZu^n$oJ@!3ekutpl;Z?
zN2OQZ=~jXlfFJHT$G&0R_M7O-(uoTK<{l4;9VYV?jUL;L*BCDJD2kVwPYaLK<81t}
z>Tul1AyiB2oTOoH7OlP(OfX9mPU>-^&>QQHV)MUx)bhH;8%NIhTjL%`d&mx*o2G7?
z!-HR<?;*H(Uph*U8k4nqZW_&yq3bg$Ey6<SakRtqiKs7|zgXx`k!@Fvi!!q_s;)R?
z+4s1l!yv+QG47cVflqA2wYt)TDtBE*+E*z2f7>B0&45gG38^alfQ^Kr+{y8or&U$-
zI;AaDbl?OnXrWJ8w*iowZ74&p!#nprGl0-1N1IVQ8iHjlei>**%Y+CTK0F(Fk(1%E
z%bgnsF7$OGpLX<Z*hlpTJ7wS*Zqy{op{Y#L;rI!qcfW=n3{;@r&62_%Q)A09I<j5U
zE~iYEExdDNkAm^l{GsCQ?2g)n(VDTBY+3*8D#5)WuMv~+3x5M`9F3D2qAll|2RkGz
z%$X^v6;_q>r{1=1qfR(qT0KFovI(lH$*z&FecVEC>T`50zbU1Uj<`o(`6J(jr{gi5
zvyg(z?K&5Q!TBE6<hEltCpAJIEZ76|2OvHk+s89;tjY$zHGya6XU*z!EDNhKNcCFr
z+C0Usq#u%xl54_-$1Id6<vr4pX&!@+8d0eHGTGOvu9^v@M;||;j!i^RDLXI$lVwKw
zR9gVpU1)PXWxCve9uMtzOlJvRw6gIncon6*=(1D|a_K(l#*!@D)L^^Qh`sOG8eIYc
zUW@B(UyR60wGaHmc{|>j=&>d-WZ&x;IGjyD?KVOJ2^?ik=O2~3Ao6M1@A<uiu=)3a
zX<qn-pS@dQrRsFuJ=RhQ+E>p{j>vkJm&)YK4_LRovTifMNAIUTkqBH)aAd0uW`=3{
zrvofpRyohv#uOZtik`%nhp8I*sllg}-W6dQi5>|qP`*kIJLW1{R23_aDWF-Jb6;4u
zCFo@TFzDTwt^g%JY&}0U3+_l`$7ju5H9ixh)%i8Px9rLypkJz&$^IHM9T5XQxtsDJ
z@mDmeo78^tZ2)#(X7}==m&bwAHZ|A@Cnq2AExSF1uNV=AOMdqZ-6y~NlW`KIRsp^I
zK)f{7I_D&}2Q8@xG{hm8R>c7YY7WLF(<5;gwf-=*P<&iZo_Y+k$ha8ehtvG%F;g%B
zRuy_cb(?%L_^4V017lfSWi)*0AnN+B1#!aAoiX93=!Q)WVjgRCI6!12nHxt&nx^+)
z{+dwK^{iO6y9yZc!mxE)HAYySUbFY{n<^#YeyUsm=fNAX=Mor~p4W4`<d-WQ*=-8P
z!AWC$c*vhoB)oj1>V#e%wMb%Ha;qKLI$4NtY@&S9q9XvC)Am2c<}}Ak-?sL~y>c|R
zGlgP-+6lpD-Mc~}a9Fyd{Ue=$+aG?wVS61*y_L)8D@T>y)!>~8th+Hu5gJq{AMzTV
z?u^<d{<o0LpidK7gKt==d5*nxL#Z{pP3(D|*z4GI+rtjmj`QopJTSHh4ka)Ayzi&)
zz=~%g^#fJnl>wH)!8pS~l~e>NM@5&y+n{-VttR$pU45z$qLuq;nph)zmQl2g^0jV;
zYfY3y<l3?IC?;xub3F0%idhw144-|*G$IbCqX5jz*=UM~U~9fF&Vb13ME5B!JCYYx
z(n?}5II}&T2C1y82;ejoOnx!>WimPnO~~<hR=nn@a3LmQksw~0=iqu<fWfhS$yqk@
zQ|x+e>bRLwx~Je~9)tvN!1YT=^DxJa<Ka%#=(RLfdlkKaA9WZdY6r2WSoZJ3`fDRd
z?5UsXUb&{vqlJe657H;d7fO^7pKcmgNMAXL3preWJr8CUt>brD+#7ahv09Q|#Mf?L
zJM#Y+92U~xL?uIR$AS^!gWz`X)w!=W-1u!fuG&F-QM8pS{}&d71y~SDeQ^5J+*nMv
zcF6|L4z=49LalB;!(sPw#yFEAq@68f`#dNv*?Ja13J=U;!~NEA)2QovAmzCG%188)
zahzAqW8p{i5+NIbK>rQBggTDI@sq%A<^p=D;eUo+LLDa{Km8`fb?q|eF)Mk#nB$#l
z)0#HLr^i&z&2X|eC)m|fA?DxwMOQ<<KF>MGO15_KPOHQAn%oQoKeaO|U{j=R+b)RU
zs&|m1gD)T>pfT5?!Gf0<IH<`yq&b?O<hve4L|sSl2)kQz>J|!zW8X|W)SN$C`D=;C
zjPZaz0Mm64xMciVJ3vp^4!zixHBN}7=gEKU>p1>}jp*m!J|x=>zIyAVdL{<_YRf<K
zs=s>(9S&mrie@U3>#L?o;amQ>Q~t*vzgRl*`pltU>`pi9z55w$es0^<t<cqgN*$Py
zdtx`c*F5+vPYOr#BJ(8|8)s&O#Gmo_y9;OeA3@sjj8sHI-wR>{wR3~|A++=v?`-Yx
z(sDx7$gsJtkLgCJPu&rb&v%vHFY@POAM5z0LpGqQZ@58&KRh6%AK|y@mzf`bS&@&m
zh&1lbQcdyqM&NyM2Ky5l=F3tKu46<|M<J?w;@Y5VCHG6^kUp;&Q;6Iiib|$Vg?M?*
zcug_d{65)+Q+qxxrlI<PkTjPHE#KH`fcLTXD|&TCglNRh%x=;GB|kud=!*{5BXN&S
z+)^E*h6d+dhO;b+bRg8bBq`b1(c_t4Q~(S0M3ZH#>p>&mpl20V?mW$vSLZG@dOTi^
zyL2&ZvTvwV8+mpa1+pOVUp=H}Mm2>8avrZFzaqcL!7>f+=~uewEB9DsjBHqJQ@?y1
z<it={x|4CZuWaoyETO2q)h1&(#yZ;<i5rLq)KnYt)6X1gg3m^(abC2F=^XCVd(D%0
zey7w(gW*ou<Yt~lFsMl=h>m_Bi&ebcuH4;}*L2FRufJq(4E^Q6+A_aJOXicDC?d#u
z^OQd9_Z-Q9T|ffdnU|ZNbi}j_o9kV)-k7B_eTbXKpL+yNAhIEuvphhHg$fKNb5O@5
zHzqW)l|TCl+Sr`jqqimgTrk4pPXnHxN@&LgA^8H<b9*_0$hcg_=516)L2kRR8v5SU
zFOUOY{F$p*H5%0%U)3c<FKvo3xMA}l8h`vEGzCHm<NwUl_9W+~Im8`Oc2~Hy5sXKL
zW04*A!$)p4V9x1)99rh*Ve*<*S%l-l54Sh3sb6Kzwm?pKYI0>|g<BKHuhD&%YJ(&;
zEM>Arnc>kAG1Pk-y8Ka_MINf_7_{GLQ+8c$iv|dcPqud0;;W7bMs^ZWqx#w>3=p7B
zU|AncK+>okMUYnjBn8C9&lSj~NvY+&3ICK{N9pDJK89U=)&jbj9(?@zKlLB3srfm7
zH%)kj_iB7c&|Te;p!^Av06dEUAg-Ncmj=C$cm+tG09+|#EQb&!98StTyn#eSy`Dgx
zZx#!z!z}32LPOw6UNWKFW4dpxl26*Fm_7Y;#P8k%((3w&ti9`O=qV(3=K*c5ZdKGh
z!;WW!BvTt{12yZ&5R@Aq082+jpnp+2FvSfz$1L;$v`I!oDTJ1rb(3$kk=;V|AA<Xq
z;gOuu7y_7r_A>yH3F*{zm~~R_Hms&>2+QqRm^i>#)%QT>Sak$0B@HtGq3V>FDR%^9
z@U`jsVC@hI$dm<*w$>w5Opu15A7MXyklcm<N`P|g;}o1oS^jcdlJ|*62~ix?J<u?w
z!tM9}tO7F(;KGq_K+NWgWY0EEtS;xiOOQQ#@H>)R-)?#29v}=lG7c?wB^adw_oC@~
zW-$J%r?3=0XhGYO?3#vYe<axizy*XM5nAnWhsTX-HeT=7v&ee9fn?eK%+Chamj}@k
z0n3B11R*?Q8yoE?zwSDqCwAWis-aR3NZtIIo{eDgN78{h!gA9HULYeds}5jfVjI|J
zKaw6f`aUQImK-&Emxv1p?rlTcbm=0q@d8j2sR(cd$&$ltgmz<==L?L9A`(Y_-8A00
zKgY1OQ7|1i4*XYv04hT8EPu<bqjuYZR_Z7BI?hS1eUL;_ObA)x38a^aeIqPm7O)wf
zv`T4jUZ1?PS}v%5f9T?(fPkKifEZ4vQ*ZXkc4(q~AH@7%9W#ZH=$-y&bHD^d?5ThU
z>w?rM!qEw&Lo(~o6x0cAP+#H7fG=R}m!r<*@ahq^$XgXi>k0%}g@Bu&uUy|cz5u0e
z5cyHM2iSe32q+{ANi5&01O!f&Z%M5J(6(+s4pF;fVXe8<Abert2IIQquT5#d%<*@K
zgIoH280WzmV9fK#0fta7;K+kC%Pr}2V1Cp>SW$uBY$IHTI&nSc4In4<-rZEpd7i~w
z9Yg380W5Ij?F|(mxdJZoP3FMv2t>g03_#@}Di7f#hwSWfyK4?WbVi|x%ccXj`S~mD
zVG-2PAtPQb?VSTH(hguvKo|CrHwq|D$=`p%N1`YOK-yZ?{(go{BlZmw(qrCTfDARl
zlL)&x0y1|S=opkb7t~0ZS^pzXn@Ho~tg7}t7{SX#msA8(f^ghztlZ}Jzk(<p)<jY#
zRRqtw{TyoYPFX6+$MLXn=`~G)X#{xZtru=Gx2tpJ%;MY1omds=_|DyA<6U~XI@@&C
z;@0Y3u6XVEcs5PLIqLR`B+yoFD$~G1CzRg~w`aoBjFzcZW6*ImU>zKd+U*ZQ+I{!1
zakZ|Q+SyEPW2c@YY{;j^e#F^?b{*R|>>)z>aeM3gO#f0<yG+!nvHlK_jLgiZiMuwG
zY}rr<aP>YoLDuHQPLB)+Eyxr)A;4&$ZARcYa~0_g#hVmjD1cntPQOP^aW#r|$}kr|
z!2oP~m+4!AaG7l!&8Tn$NuP8X3Ao@A(^xkXaxP|v0mI-JU=G|69L$CuDq{m?(CrCW
z!&jtbpZ}t4_uHa&)UW_3Y7v4mR+47t#SLq?zY27+`nUd!N80Uc{t>dPDPn)3tzg>y
zfxc%G_@m4h5$uzQRpGNr)_wVbOU&!74hR_@a$64oZ(Q6kM*G8H>ecTXEH#K#u5#ZE
ztURHg&;<s>7YRo5_p(8meT)QrG9{+!kZJg8rDlG-+KQRRFKucHq9gR52rluoI%V_|
z764eYHb2A=8T-^>qxwS)yLg@qA{#(1SvsmU&T(eIxLUda8i{8lfMwG%c}LC==VTn|
zyCOT|yYtoE4g-HriHO;K|NI17L6vPsMzT}j#0Mjh)5dVErv%aa3*9xx7eX`^GByFq
zP!vT_AlHZUO?2{jEe#G7KmXC<!pZ{APWG6`RBHLAl|w;*_19lz=J|H>-;DT0_?cDy
zirX7Ld5$n!A*FOSZyo-Oa?=4jrTO+lfi^a7mNUFe(8mv|6nUlObO_rA2;ozY!ms~T
z0)axoK;PG^vTHmh^2{n%6pO%-GCChTmaU^WT5v0HV=7+nLVo%1SwNhM*+Cbag)#Hh
z5%%j{GTDRznz_A|{l+FlH9=)dd@>+b8cUh!L&h;tlXOoS@Wpl!9mYG_=Dn;o05ho9
zg>c)8XPK>yD*XNmgXo@?|9+OcJ<4dhB)pV@dSI#86u2|D>aqI(GUv_D0UU9(h|a7u
zGCSyuyCoR_=*%??fBd0lvMH~hvoUd~oMf0da>VN)Nsp&Hz#gq_f^Z$!qk;kxuQldC
zQ{KAK6wd@}JifI$U?g*geerWrU@4ea=hn6&=FL>lbJ<N3;)hJgo`Eu-<7v*1&?(Hh
z%cI2;G(wKMa4KWo{KM$hPYd&UCS3xhKm8_4wv(LB$Ru0&`KQ{{nI&KD{Yv-!HL<+Z
z?u|pTIT_h@2@AZ1@x`J5TFhFk>u7OMPwNHFa)@Tc*d}Wf6`lq!rO&#VYkX%UF>0=V
zIt45-et&}|U(L&_L)Lu*@`btxVBz@b#kt&714KhLF2ua}k7@9dvIOm(ny%I+%rwdk
z4bOkA19}fDU{Wf$b$5{pCQoigM}#1IN;|4_)7o(a`8W3r)`y26Zj(v65fl)$9RiR$
ztnz@#$D8Co2nfgKUuDlOUPM&*i<E$X?6CJr37)O(UcVJlHu>Q3)v)6~`1-`XFm$a|
z#5#wVbTeV;T3dOpJ)JVwk9E{<`E-<{WRS;~_H&oghF4xX&CSUP>VxPx2{Q!U@fXye
zCpOijpx&cls)X%z6#50~H8bv${Z?9?dLa6DH|%BalpNY9a6mn{0zxgU8WTR{ffiw;
z4cbdz2uqpiJQNphz8J%4+z7U!wn6<)Jj0o}J;ot+nQJiLkN}E9@3Qqr=Sec7e?ylx
z{{%X(|CV7abJQXDCC=}0-0{(3K~9rvvashtk73CsJ2d_t?A@{~#|4Hha@YAlPGBnJ
zHIz+=L8ljuz1KEpzEHb;K%G+v<gA)8WUOM1%Sbr}KbbZ>H||MmS=?j_yBz_+PMtVx
z^yrY;K}Xzza-UA;;P}rm8f=<Zp4WXj`A>_rOu6f`fAm`zM7OupF1QB1px<(mewWI@
z!9MT!T}`il9bS)e+XaS4St1(_ahbm1J5f!+fv{AgQrffU#yr~Y*f?hRg=YS_b5S2$
zR~es$t2K^e#bf1=8H~RJTEcBtHF)PLwU;rnPP0%a*pO3h@0qKw@6Ovhv`Km=0hCpi
z=qRY{b%>;XJk_D1$+O<?goOUW-`w-~vk+=&KA6AQSG>B~HoQ<v!40bTcI3Eg@WU!A
zdBY7(giom_)b3i)JmLGY<c6~I(p6q$%KO6X=|-B4)yG=4PHS%zbiq)eR?}@)dHB;o
zWT=Nr2Zq7S-X(`?aSgg$7g@JypY0Zho1Un3^yLU98*1HO{&{XUMtzU<TzfY)r5TvJ
zUQGlQeP6=*Ipx<r>Qpi1-p+llngWGVqh*E~ywzGl+04(I_Rk4v<flH2wlIg_H#mdY
z58J_9%8qkGRLF$v5Mz_DF2Yt*Z+qo9*IWa~wpH=TY2U@ie5kw99it^vW1AXWHxd)5
z2Z+4Bs|f1>T7$<1lCiwsA%rBEol!eZd-2=VO?CVz^+8xaN&ik^u3bj7oit~t3vs?j
z!1)eCobSI#NMU9TFj-$fO6o(Dr02l1^t?EmxChzYHo>oaliRj5VUL<BZQ4Rp|3u_f
zCR7#FPPYvOujz0Cp#phjslzZXU@^PgYtyMhfq+~5S3aXH(mhc@fpkV~Nf7-}M2`l$
zs#3$Vd!wWag$n->cE$d#{cFo70|f8M5Z&lgd8kaNb@P;1ZaHqi0ps`2&t*sChbPzQ
z!r_Q%V(qvYPpfl}ewZJ!1oCAyt2W$`cqFFL{6+c1ON$A?r9k>YC%|$~W)~9X>U94e
zOF|(5jrgzJEM_w2V}H|ek{#++d5A{hNF7U~a7=GfHGWXh7Ju$E-<Q+Kfm1qe#vGza
zIfjuM<HNcGd_q*sIt&n1<^(~C-3m_hKx(&Hy61wG9Tc8Lvsj2y{=3s1&$t64RX_9h
zDAJnl5zSp4pU&z1Pgk$aB}9FOx_B+uTfHKb8x=h3kf@)A?2YCw(Eq!?uc8n07Wl#Z
zf<P<A;8X;UvAs@{o$WvnPEBl{zD}6F>nH#F>CIHv+0M(_qybU;e+4uZB(ybgdvANB
zptEqrm<knY3Z(oKg6d`LOb7NuKsrH@km+Fx6t9kW7TuBk%bTo!?UrEAIONLjI!J95
z%UZh=-i#4rJ=aV&|NHrVxw{O35}KeRj^lc8dsnpr8D~r(E0TbZuB8u7IlPnsKAbx&
z9==|kItMw(-&U{FRt>1PFeN9s1V>+2vr2OexAky^=!a29_Ab3tampjazgd1miOYb+
zi4;4Bzk6Sq+$p3nHyH~k;n3|gc<y$5y&^TMdTwef#&TeBB|zz&nlIT*@5tM5rMwQ)
zif^bB5!|$18Ew@yK^OxFnZBVuRZaEewc(o@iDl|Ht(B(Yp=mHnJn-^yBjGGkR9Lq5
z^T$1K#3N;s8dwTDZU-=>z1e_v8=Llc0!YcG)L=XE!r-F9(4oB4bj4-!#5-d0wd&Nh
z@A9(G`Zg3i^>fRBIxT|&Vz}d<*xF2C_5`8}Ftne~)jIF_v>Ey>LdBo%%S#pDRXRI&
znZ}*7#uWj)@1jw!rF7rDVWt(AXx(;CrcEroSrS4z2&J$?Q>;KDF8t;FA9cZlhJc9~
zmU_x-?_J%}+z0Ep7F4M7Gbhu$)EsmlS~vK1jdL!m8_l!;u;?t(ve5zHp5`%ad<x$A
zl&OP@uYc_4ZWKF(WqNg=70w`Au_^x$fOjIoamJ?HHh*8Mu!}ghnG&t2%E$n{U8Kpg
z<+`};!Yi9Opr%H!bx7y~8X?}w9fG#K!pTzM0Vru%>%M$_lL^)8@adgVL+;-N91vuo
zX>x9Yo9NQQD&aBm!lWa+p^C0rMGvc@XZer=t%5^y>fg*4DxhnFoz8_lCS$lXOpdKD
z@szJ&5q+iHk^Q2IeiyR0)iD+TTP!723B7Yv`eWU1*YC$i|N2g9ywB0BrEfr$_St?q
z^d7|}$VPsL35GhG0DH_lbW-TVC=y?p-V2=>UF^I$qjzdOzykeMWmNA5M5XeX+sb-#
z5eQI04rgvCfE^7iGTX`wb*R#NBx~n5mfTjfSg#|~+E+lE7PyfYfH6bS;mC8RP9V1{
zxF&T)<`IOeqF;+N4YFeqcN6LzgSu$Z;K+9ZjG7D0qnifa$m^~?YMJX;Tkc32aR0SE
zY2DJ&E9lX=+o);<lzgYxoP7mv7e9k6+jD$MqN`5xB6JmO)yW9-Dz@Y{;r814Hk8rp
zg1o->Aqy}-dU#azQH8(c?(+avcCaHmE-3rP#Kh=A<K{W&?Y7#s0-cxVlCkys)c}89
z0gc2@j3*N{4nJ#}UfLVpb`JcSQ#(IRS99n4Glh>*HMFU@;$>BOR7yCyh#u)G8jD_y
z#kf?C_Dt71vipDm1JH{B`+UuCHQOp;e3bHf@?2l1`wc!)LI}}~UynfbP!`evOdx;+
z7BqB!2kv|)f|YI^w<Kz7E7r4rHYaLV1#Nopq=)zHqixY<6M#g%5+Eb6IB2TTMgG-`
z^8@|HDPKVVZb+<uP>p)(xPinJ{b187cNlsFUFF0r6uG|OGuMO&eLVrX<nB~+8D8o*
zZl$aBOxRm4r4!~n3iJ|xe&Q7(^)g}xzhSRcYo4p3Yv-l1aVnSF=$O3J7y$N7n{&RH
zmBS?Ob`!*-+pk~V*{|w9P$9nyc4A;OQ=hrqnQC1T#-^VbUofAfPBmo{y`Ne5ej?NZ
zrh~NPVbH-+>8#I1{1~Tv&klaGQg<cUEI9aT#~rucZ`geCiLc9~DNyKF5&W)=xzs2W
z3R|yr8+vgF+nHhaH=;M{_d|b!23}9l+r`cVt`xry+*To^h=(K$xiv{jCxrg9-(q25
z;T^20wcFWYU2=O<BjWVcMtM@k=<)R><yV>`@h(UG`N|86?|sS8#Z<PVl=)h>6p)OU
zo_xcW+y-OGs^hjPhd{9#c0P$S%g`OmcCVNTS)6{a`$=Rd{UV#ZZg(*U*Y5Ab1A(S#
zCCS>fu!_Eun{ZkwPoMgwXg>+V*)!<v#;XL;AnhpF@1<x^45ax+d%H%^+LeQsKEbe1
zyQM^<rXp=L#T=T**QR!e6yLH6cKdRm<&8XMQOZp)=E!bQCC)+?THj2`RXI;ZEw5N9
z&+lMf@qxVgH0wry<2|!(GXp?-Rj5MJKl_|REYeh+=;f=xMXnFAQ{LC>A>{b4N^NxR
zcJ6lb_r-}hJ#UP=z0iUm?)zT&6F=)1-MH_q)EX1!!$0bUWw14Fjrdz>YVZYI9Keeo
zJ(Y5MGBGnlL*VVn9x1?tj>OD6p?y&@Yu%}TL<XtEh@FXgNFQW=zetif67mC(>L^eH
zCCjRmsL!rn&USr|3HHQknknU;40a)hG1OoDLuLU|Du>O|RJ;v&us1u9a5mB)VZpDO
z_Lsqc3O`0%jT-wiat4O+3ntQDb2p}LP(lU3`5NBw<SJE}LuWciRNqP7v~Jsi5ni}G
z4fIHo+uj*mJ+hw)gt6`kQUAK=M6TLjliK)pN9l7YU@hJ_Et2$zj|)R-$LM`EsmA?e
z!d0UF+f-`+!Vs<`mhluKu!+wW6l;~KR`}suMYCF`YGOkmZXB&U<!uxfii7sHYQ<F}
z-5vl*ajWT2pw24{lsy1O_qtW+cHyI42|a$0x?Y4r-7aK(F<$>((KHWM5p&pDn0Ac*
ztSZ&`YvCd4j6|6WKHJ7OruZDRdCtXfpQ2p>0ynK%o|h_x=`;%tz)?l&FPP%gQmw~;
zFUIeP^frp4?}&hNm@;HF$!51hxeHRx^9^1>JoMaOjku7&MQ`twjzd?AdGMe!g`s63
zCaKnKuIPxlZ-(MreFZ$xSvEd9m(}dZcBrBk#o63LczpnIux6E0e|LW6*hu(MT%vSG
znj>3ywE<@^jFt|6f`rtEqvu}8P&v>IitWOyd*!H)`&77(>QW+pcXv^;!UZ6Eeq+CP
z${d20x*i|Sk2NVqnBZ3(+54;LiN0iZICoA`A>tHHSu~;@Y>3rPOUW<xA1w$_5?nxf
z+i_o}#6Ehk4)jJnB5%*@{#`bNNw>IfsDGJr1G?KOQG2B>3$yU`_vv+E7YkP47!{&R
zEx2-5>)*vmzB$rW7Z^P%jnp0J$986o%;&E{cfP5mjZN)8@k8PK&0}=MN2$g^w-;`y
zzN;xo4L*-k^LXugRURX}<Q2GU`e%T{q65t}y5=a*EtjD2WdAW=*b}^MRNbt@{%zg`
z#5AN+&B@O7U2&1LVC_v_C`ZSL_*n!@+%N>S(PJG8Ru0Xa+q-Rbftki)IkIU9;Us1;
zeI*yusMSpGN2~3A8}#cAo#9Kzp{6!ad>{fl65%xhoxUXpBG&bgMd;i7Zjaah8Gn>k
z!IScc2{T(JE9F7wXC{8iuz8VzaXI4~<94xNaG&d555bdt#D6$yoUTa~Rv>Nep3^?n
zZlU_63}xCH)Tr|X#TEzi%;6xrn9;EI4V!nk!yh_0e*m1uKw$hdAHNz4hx|RFPeP2>
z?FGVdz0V)LhW=e0eNDoJ<lIW$NR$^+a|0(1(n}60;FdA|t1TojGl~-dw@6?uZJ~Ri
zC;T&FmU^;vzfHB4z;JJ&#U#rFKj8p9H)8b2W;#_&%ps6Xci62#e4IBA#ohc;?t8Ug
zenX0uO_j*GpMR1-mp9z=ptOhJy_VWtS?%V*m@4{$c!<*}Q(bkdou*#f${O^j)GB)3
zalKcaPMXhGmOuu3pwcyq3$6Ad3XGZ1rxbc&teHm4BW<Ce&YSTz_0!!^Er_091JNy^
za~T6&)@_f`A>ps<_t~04`B5K!uWs|=AX*J9FOQ1_J|3mV=sFS;Uw3syf!>{tok4=u
z(2YRYwtp&jHp$E5kDs^7WPd7HM!7A8kxs7&j?r`b)OkOn-9^u=9cvkG1!^g-$Gy-+
z)DRoLdXBu1c8^FZ7l=Bz8vbT;R0K)DLeHY*Ej*+%xvNr2XO@ndiA>mO;2KD{Z2D>7
z@o)O>z@iI-%ArNHiNOUn7K%E2$of$|?-E5F)>bUXs_v(1$V0bHgOK=Xag6XGeWd{&
zw4(_`boRLJXIkYzA~E*4+T~+lC^jOUnVPs?@R5Fld8uz?*P*CxKfHfsk%wrgz{s}2
zpV0+qhevqAm!fB)j<j61fs&6Ds6OHrZ>a{fZZ|O89oM(@6<)l(kOjx}MK|c;1!c4^
zS5!Y0P@~jC>+~`Ng~#ZwHJa!0Qt`MBDK|FBkxeI3{eVhtA2kdU>l;7)0^{Q1+Yk#f
zH&J@GgjSPs1FI`ga&{fZgF2cJRt%6tK`btJTiF617$~QHU@^rjgfEHDeTq|otIG?&
z%0Ue|;Zz=0#a{S$rwit|IE+SD$nQ;dzGpR`>|fW@C>c66%M{NNxmp|t3~@nG$4!;S
z(FEgWI`DRh?64PW*QKOdi(QThiPEfoH;=A8Kur|w=a_NZ9NFVobY-D++Z)w)spxbC
zsp2I*&Hoc;Bg|Hk3i04w7XoW+3w<12>oeq^YgLpAuq(KJ(u`Z}IXeB=+SDEwXU2Ud
z;umNU>`7kG1R>&BW7cf|G8_#*{A;R)A=Mt2ZTQ@&@eByXYi@nq@pp;}@NYm3TDR3=
z0?aiQm4*t?%it!0y~@j70aTzY(%*FYYR7U)6+O4ga4vi)4qxwcyFpGtw<49BoYuJ#
zL|*uEXnBN#E>P5MfPSp#L|n8HCXgJ)?ujW&4L*bG^N+HSVfjlc=wkJO0P@r41R*)i
z6#pB2rNz2!t*yD{mXR`*XR2h3;mToTO*M?D#pt-(9^K!R>#(T7<CiYiJ?k)z8Y)u#
z67LrXd}uN?(+eYLD<CmW^N7Hi;EOFP<f$&0MGJ1WfuJuH4IN8Cr#2EZpi=lHSPA+#
z%2M%D8S3Y#YO8|SR>j!Pmd^Jq9=fTsR7ZW!QYF+-#4~<;29*fbz73C4HB_k-v55Ba
zA(x@id~^Zbk?$xK(5GWWy*cEf@9vEQ)Mec8$jMJH1ALA>+NTju2hS=-r|8^RK8*vM
zzY)Fjw81SQdY6Lku?^=56&?fbvIWkO-6T&SvbJNwS@|#J<aP7tE18b$>&e@6RvJ{P
zUO0np_3?3x(QOZy(_2;aTelR~j||j_`kZ@4W%NiZ#8e72<8sk+2;ZC=4ud(Y=s*OK
zNaLODeH+U2D%GrlBU+M>HmT8>;k}M(m|&ppLF;NzgN9s86m+!!7O9{}$WV-9_9;Y!
z4za7z%%R1)JbGQk^QQIp1Lh|2h{E|@K4hL^?Zu(Qv{#P4xE(_-ecR?X_EhtCFOfv+
z`bT{g!Z7Q9g80q)cLa9u=@XCeynOon3*zDH9|+^Fe?PkURyQCPzW#v-)cW^tAaJ(+
zfrQEW_y5I*;E)t_{{mp1n%oBxX^z9?AAwV%bZXqpxxPb=YsRyI$;eID5QM?{XY(6C
zrg#1C|DPH3|09yw{E(B~ntx7nm+t0_HR`&lVw_z-%7YolbhFrrCRziu6<Hg|X_^g*
zl|YsS@%6`!q<@KL^Bp&@jsM*_+<f2v+ZM|I=0nhi0Ih0HhWugdvx;5onM2frYjFv%
zn&TPiNWY>_pUu+w|LV^9p9*AYZHO{=afm~*oM#fA{QT&|(SLt)rEAZzZ%!6yUHSR5
znB7Hf|0c1v-2LLf!~U;-H2m#5Q$N$9=I}E&&KsKjBy!5{a`%CM{qW6+vMY+;bn6`Z
z{^UQl-u@YlrOWZ%?qpSQ49BNpb7@{lni^dGv?0%#J;HaIn=<4Dev;t7Ht*FRK6V@a
zOSCUHZ&qAO8WGh@Ix_myOGO`MA?*j%FDRo-36u)0&Lv@-z`Hi?a-^oe%uhItD%<Kz
zbHlC<oqIX8=@#*u*SFSYwf4U>0{-%m_)l1XqyN`0|9|@uw5}i0ZoVwqt9~izLIEOT
z^&K;TDs%(dU!r4Y@(jrACSsT&v$@lO5*To{F91cqO5C*8yZ%SwyuVqSU)-$oGdEp+
zWAhCB|L_v(xpvE6Nt_&(G;a7Xun&}&L4buPAK`^{t$Z?{Z*5zEuC>SE@q4pa;Q3o{
zh33iABzab-Gmek0b%FxzXlUDjk2avLF&T0Wp_YHH^NtZIH6!-{rKw-)9je0wmGvKn
zzf33ok^ExLxpZ{-B%L=Sr!rJ09CGNZ`X$S$uj#6~hmu8}ZCE<m%8Z+VR+M>tukUoP
zG!C*9zIBvccbHnF$!biaF?AF-ynvE~9ug2x4MayoV43RAvL(#8{b*(u>mET1mu$Ew
zg6Xv4{=pHxst?{c=k40}B}SK%^hl&wBdRfa`Q_fn5KKrm54ObF%SV{mGD13)IF7b5
zj<#z`R56@^qFBIp77DkBUJ4Q3xTYdfLWfdm0*FQ7)_pG)N<7brlgVmAFT2?a&e+Q6
zrPTlV-tgb}-iosPZjzo7)d^={NEzC<@H^&U327$V`@H5)w$BtJpprk6_3-S^kG*g@
zw!J-`*?p3Vkf$&VpPhezTlK@)+xG@lc6?$*gsyOgUwR)PleTMm2h`7`-b=O&{{4V<
znqBW9gebaL>9|Z9%bql8l21EONLstiO2?a1dfq?K@K!O5Tq)idmaR4q8I}sQT{BjH
zNe8`LgR!&>z6`Vm2I^EAdU>jORMLVwWE)mG;6+hh)1>O$>S)+c#VXc@;Pz{H|I%QE
z^O9dKxHG=`y{(~7=Fg<bmzW%5jhG9#TuW{|`k7~N=ge(Z5J5{Q5pfddnya9={%p4@
z)$;}2UKhGkA2<2p4H_x5Ne+|Eu?poaFk-^6O%0S>B8wvx5Ne=7HLKZUm!40`$-o?+
z5-!RwOJZdE8BM(mISxlHjJJ!zKg$p>al@)q=nW8cQo*9~*^EYklR`ugZcj$+aU2E9
z3#sGFFa*ow5aBE*)&ZMKU~R>eNMGiR7EK0GsOwl1)R~wNe~_!bNB8NvrWlND4`W>J
z{W2rQ0cWBZ(cV07h&S+`L}Lp{$uE`yezjtAT;FqKqhhgf`>3P%)sg0Tf1Cj}y-;u(
z=Omfbq)V*{<9SKUvp7NqI>kaGl~1$neM#9okLTt_3g-!gsRMOQu2UR)f5oF%*DMwV
z=WM7haJLOhRN`i}+TgR*)rDdV!5`C~z30SWpn;J4@!6uuCq(pV+Zrpbp@<DES7)r`
z^ijRkzFhQj-j%{3`Q;HJYdvy;Do&`?z8k=36i~$SA+5Jv|7I^E9UU>!{Aw#<YO(ew
z+pv>3IoXW5eN@fxe-}}Vb6Eo-HYVuh-(~U_uVN(RcE`D3<AM{)bFx@%S7zf}ZL3<&
ztj}fhFInT3c`4Kn2lLb>Irq+*NREAuDUsS8_x9j<3w8%uQ9P-so?={)cv;Pid-KYy
zqF>FG5O$E&?u<GGYAHwS9;XrUMe{kgS#R_|5y8OSi>o7a?n*>Z1O5}nyNhwrRS%1_
zWq9fwDo|&;Q}$8K#PTgM!E654BYwi+#ah=q*u&GU{j&z)>9gxYi{GK0jx6zSsV$=U
zMz0t%#Bb++vhkMJjrt{na%euX4;?Z3^`V4hRg7%ai}4iV)$+$;y^I+Q_sx~U>9I}n
zyBRTkI&uwdBA(=W<ic=y{)=mj<OWheE~`_-CXmfx=YJ7%3g;`<S`+10kdj0j&1P1S
z0(34ouaCs-T1QdgeC+<wnGA)C*tl=0c_KDEq5?<a1-rk*irWd#lO58LNIF{=)|r1O
z+&R%;eEr(y>=$tD2wa<UY>KxnDdxCs@3!-S&+ey8({00sL<|c_mZrh&pO2*`<%+Fe
z<t7eSg}z_>fZ<+|dJ?<V+^>pLnGBywG^!S982f!Jov<GCLY18u>pdOW*4%$fsxeGp
z858WWYN>gnL8c%n$yMWOKgF=x=0vO4ot`|x>cDj1sd4G`r<{q8(7rq3={D8FXb2Q7
zH?F$cmz_}FNBFU2Dj>z*FIHuw+Fw98TYa@}Lz?uijx5un$q#4~N%NS16KaQcj;Cq<
z5~qH=G{hh4s`K%A<;xK{b3YOT*B|7zMI!5Xhr3CuvGIc?`<V}6UjFaH(qjDDkBsXZ
zF>S+sCE8d$Mi}p)2`xuUMfv?Nz9nbE|MY`=-@KL9FJ1a&t!Z+3rb9e|-p$=|W|yK-
zNNlch3ES)JoPpw;1}&r^d3?VEvnh_<#@wFfe|AkGIE1}2)6M+-#U<dclc$nBcLd5>
z<Rzpe#d)8YEtQipK7V3X()0b&pTp2JX7U%ODQ~&~TO3UfBGi^LR%R}kz4%A`A-Z>b
z?}b~UXZ-2jZb~Iq!7uD3qTgJXzfjR#{IdClb7XH`yXCoeDzn+eEWPQOFNWUCY;CWO
zZ@Q{nd!kaW_^R}%9VQ^#wk$&HZfx)KrQ(s#rF`r6Mll?uBZ(K}hBP{tlWAD2uRjyz
zIub<BRc!XDuoNji=&RN21ATS>_N7cZp<~}z<DFRtM$rY8@$x=m)xAE4#d6irY_-Yg
z4afOQ^_2YXa;FlSkK)U6hOJjv5$~$hr>EWT)LEaGGSu9~0}bIGtIwsL8vA&xX2Sn6
z?u|PApWSjUzZ==>;vzQBUc8r2mXT@-{wTn(*5JWHhU`%PssU~JzudgPP)jP{NPTb8
zd1j=?>MzLM+r?TX3Q47eV^eU=PnOi^|8|kjfd&t^smG)}lm@<K_7M%?O{pg%$@kpN
zTIU9D!bQWugDaZSgN31gU&<|GU-McpmeRY8<*8BmNihW?^<&>AHFOej_1Lz4_h*au
zt{Ti7)wS@g(JjAM7qwIT6{_QA*dDhqhX-5ivi{4LD~yaj%e)zBsV4rm^GtuYDI7<*
z`q#E?-CKWUP6V?ko<vhF@K`limnO>hqKiddoW6$9qo|2$+upA*obs!ESqZg!htl{G
zNewlWQr#|m9Y0YjfV;Xv(u<bRl}|MMud3qhomnv}DPAn5(Z_AIG(U;5ea1X^x_RmJ
zGdnmjf)hR}n=V~r1F8T^eDy>}GU+UyHB*qZVPO>CuGP95&Vy>)OKy|*Y{OVz-;>kU
z5Uwbl=bXg4rnAT*HrOJP4XC}v=S$%!)8hL_-N&bXtrix`>Zvd6a@t5{qWb?eHweua
zT(s50J#ZoUH<tUal($#(6HW1N+|91f4VBy*cgba0)hw^}$=3#D1=X;#&4nF{I^nD9
zQfbc2^s<>)-al%^uqYz|4UeH{{=xU0d!B6(b>C1MuguOAkz{gvNOd;2K&gz#vsl+g
z;%?;FHkQt)GqE-FE70{{7G5cr*3^w%zp2~R4szimikc<2<Vv9?c7U}sd}o%~|6_4s
zC{GbzAM7?+f{|@XyiLOdzi(bj4BzcFHA`ch%U08w%{ipL@C&Cgv|!Uo(|ec(uf&Sj
zKiXtxYvQkS($-L^MLfBb!@1YmTYhBR)>d-0-kWl$@sIv;!zPkt|1nE1dEG#_NrqH_
zv*B%fbOf_`iD0WezsFryhREtO$zS73glt%=^yd03Q`bzEjL0bOB(yD*2`WR`hK1vH
zZ8K-wW)XVwYN8kdIyz2EJ?wj&aUp3`a_0k*Wjo7`tF10b;&5cf9;s2+G&KZ&kVWZN
z--RZkq20-=D-&PV@k_&Tn~nlm$eSeU$6X->ng^f4wTR`fDNrRtc~3=X`U(J1GSH`<
z6QKzL8G>S&v0xY6nzs7FI74uQ_+;=GXz_<tSkAmMA$cL#uT=W-$Qw)!j^%~lYxrR}
zJ&ZRRo=LNRvL-;MNFk<%mHhHDNBBqcT%#_vR6*DOb#1z0trga_VyHfYOsb2*1)dBZ
zLO(;-9;DtSOxf`}IgOq-C$^N`BTc5Fd8QD6cE<x^rV>b^s0-tyJb5uxvm=RAiV3Ef
z`*Z-ELOLsvP)`f`5!W=sNl{g68TiRegzZyW+=F80lO)W7trggF7)+xco|YqN2EFEJ
z^2_G9Ca^-)lxYsgZ!fBmdz!}xQ8wPV)4h@&mup_CMcLzO6qZAAP97CzV-8XQkIiv9
ziV+@IW?73tXZyA+)0*=+c&h*Sv+x+fQJjwavWLR1>lDvmznZ-ftH$UEH>_(zR&;gn
z$|O3_D!ArOf-{jN?o=9fR)JDvq=grKx=%-8d65(QA|T}aVbbJ5Oimi6=L1IU@_TK<
zs#Fs=KjC%kX#?Bd35;c~%@UKtN$FHN-29M4LSr#tbn}=i%dO^_LLP5BLoodGf=&ae
z&L1~p#eHn<LpTJL%rNhFSnu@7y{NGZHgxYQtuiIj<STTFIi$P;O@k@^2Ge-V2#y*R
zRqB+-)06l?RlN9tg>yk~cXQp4kGY;I^*C-=4KLbWY8Bk2e>DKe&A!xhC|HZ`rBCeh
z@pdEnarMY)*BiX|LC(e6mWJWtHNRL`M0-t-4V{3ag7HgK0FK#l*&VtKen)Mg@VDEv
zVr&`v@Sb#i3#rLg^mrI=#yo2~#opHBLPh(Oq}S!~5s?Td+zgw8r5Ti;eMlNtSf16r
zX8eG{)GFvltre3v8EZYM^YP}fx(2aXR;~k{c>>$sdzc&owR6^F`oqMGB{!85%kJ1N
za^!}^S>#pphU&6jHetX?#O7!W^YI1+6txV!a@|uf>z+M*h*&yFlUdLzc2rtp`jxT5
z@=IIBPbWbZSr)EvgEAe9q0egl83G$jr$un(P{~a5GDf!CXQ&U$OT-)K;r$D->;;Uh
zR^s(;a<)+#>lkL_$d4KAxagwm-S0>wyvFi%UQgxRb?8ZT371$jsqdA!yII!OmS)Di
zgI2TRx)Y{`%NZx~s!n}!5Q$Ae>E;26ynIsRpgHm;4agynCq+i$nmSom-*`*k#Zeg%
zE_SwRLv`|VrXOFWl-z^zeV@@$a!LsU9(}KESgfs#Zp&jE8#57KA6juw^GG@PrsxSy
zx<^IB+zfhvT0IvBeULb@(b%}4gH($`3aI;6=(YWLr<A}xF<kZ#!wbvPvM4;Rdo~Rc
z4de35uO)I;o|-3*zT(6RR6@>sktUy*S08Em(=)x&Y8_8F5<RfB!^u_#&(bEm`l{!M
z3bpifZ;E;Dj^=9rRXy>t>tj`i*M`u6!z-^N<R9i;(XR5Md!qv0g72vMqT(x`zE|BS
zol0xAR#}b8BREXz_@fNL6ExPG`v@{LEIFL>T`HLlJ0GKPE7Taz`kfKO<iz%FbK%$2
zQAA^95BX1#m(sC3NqNY?<245y7H06#hw%!7Y@Bl@wFKXk!J0z5*7W-6?=qZyYu-q`
zv`A++UQkY9S94dU>~h8xBdl?F3O}Ka|4YQXr@!0wo+C_^g!bJ@sNaXzd1u+lvS>oQ
zW_+AZuFU1tdze@XiKET;+tRPT;Wm3!giyaCL<hYw!sOg#<(|ly`_gvdp@ZU9RP%`?
zXm@!Byi!g&mA}`-qoRGj?*jM2gt2A!ewwi=)wgJJX9z7c>Q#4h{=g$JG`5E-<W4uL
zQzbpQQXkIZhCzpIOR)rXDQU)ifRPAQUIOoJT6e5&J8VfCzezpuseP2%yNl)@&o(!<
z*8@Loci|YFW{S9!E=>z9k(cdo3EPsu{iHHtgt4Q2iV;z`8v9-~d4Oa}@eRkGqt(cy
zp1LSS%ICBp6AC}|g5nl+5gl}dl|ta9m_vBAuHw6-$ny%Wj{VF=^x7e+S~$->_NVzg
zXm`aeA+6jsA6ep;2oM+D)o-beBfIier1^faK+o_fqGGH(t#yQ!*rqtvAy-pwt$8NF
z#o3l;&ZQ8dFV#1RT?lR;%gs1mGL)YtdieQ?k{%I|h2F9=gld)_pab7?lqEm(vo#}_
zqc#M;!}IO73(e+@=u0zKp2eBi3fLTBi`00%KL|s_Z~o&Gr32pR-~ES}y)E|nZ-t?8
ze4K&julME^>i8Atl+NE!15Cq8DY+aGm7zxSJ$M%!D*<nCgc`(%$>qcbD^SWy?~?Qm
zQ|C<bbLotxGNv?E=Ox4SBzFnjq@x}Kph==0CVqP8F-(<6y&|5-;q)_%W8+WcWleCf
zUfro$i`OQ+0vd8FzJ?(k_nU$}Fm}+CTg4D8ablN`?YS!&#&f+;L3q*Sm4M-n_Mnf5
zgG$<RIFcy3s9=#-mI=D8<UXJO6jpioKSFs`gy{V<lUF9>LMZ#<&r$$Y3#@aDN(yb7
z<2u1ftm9{1P}IpQP|!W+wklc2zc#eBtV(|cST+5=i?t;l`7stXR8D1UHP@|N#)aUA
zizso}8m!d>2U>r5;ZnAl%Mn;NDV>)h5d}Ye4c^KhA}@9P6{#o7(F(aTv9WF|Ber2Y
zZ&&wpm>0((5!k9?-UqJjumSb8%(+siT+uQl62_Cq<miBbqQLAIaZ9Mq#0k%{hiK9d
zND@>J#U=q5GE(|Cs}{^=D#h}XPjav7(<I#Zhr@`XXS|%mP1Ocge#H)+vA^(`4q0$*
z|5XNZNC$wrMhR2MGznJ&jhL}-LolE$_)b+}d-^F_cmJ4-Fu#sfia7=G5xq!twUZi;
z_TvMuiF~-*X&&6&ywpzN@2A}0RiJZs?~eQ4w)f!MEvV*n!8zPLHOb*X4Av9W=<q8N
zsL|n9z9y}KO<gJqh%%}%g*P;TH|(9uxHFqRrEmu83PdL&HU;**{^8rV$KTRovuWZO
z7d`bbQyhOrK>>wAW02@hsaXC2sz!isVVdaDKb0R_id?dhlaq379l!459iQct2jhMK
z<NjwMX+O@u&u#KJLolNBy`=VCJ6vVKjjhd|QgsT>c-AeOCwuFJXzGLf8xX+4a0~Cu
zZr_>t^k>ICugC1XiK4Ib8@0fTXrHey%-a$Wa2Gt!l<w32EP5xzB=6bXbgAvJd(&mZ
z=(fiwzj#tz3{GFtyn#$9&H8r!M@)&#ZmeJn%FWSz#@D6tIT~L0f!%qcs9V?ciPnSw
zX=5E2>M7-NOOdFoeCso+JZY)8XIjVLEXN#_$*7yr3fj`Yd#+9h))`U6ro<S9GG`y$
zKf3m)O9gJ_Zg&BOH`cjVYAx}a=e;!1-IWJGm<fn5gJ<r{VrkzjMv1mI-Xr;lmTx%|
znAj*Bt|JBLU+o7bfMd2{?P%Vfj5<B4*Us!@$31Sw%GH)4A9j!?A>eDMzR+ks&{6v(
zj89P94_fbBf~x+O*9zAIW^beAG9Y}SsO#QT4%S=PioE6x<6A8d0>sE>vbJD=H>e5#
z2*D9JWn8ohglg+`sjgU_xLMl{x9u<ZE1HX$;fh-_avb;cC(7o&;c2@LF&bVw|FszA
zufVZom`9SZ;NC|czjfCb-s?=s4(9?%FrcI@ZIp`BY(!fFIzEuu*h&2UQEVGG%QP^H
zwR)%a0e*NN^gkCqrS}`#g1%j@;xzhfS>EDThXh@MVvckzjFUBQNS5^#etZOMjQKkg
z_9`0bst}(kg1__ryl+AT7HZT9TOy8i4bvFrwsL_X=)R_)71Z|pF^o9GE|h5WuPQCS
z$HyVTbZL8y)NXz#b|&lCl-2ZWZF`14ewBxoD<XxY6Sp;s?xyVQcxxt}6)M=${~2s~
zhT;AeIiol|s+`K<yJ$qV<Uu_{Gx!(P)4gwO!-`0HWKw4l$+&}6O+9|gwDe>}v9^V(
z_!P$XktGf1xQCuZG(Gnt{FV^y?jDz$_0lZqY3;kDw`wi+ipykD<PSKx_Z;7}Y{}-j
zJ#{;shZub?zFAaoi9;Q};<)FGK186=YZp9L3hMZSKaN%Ys48A|;m5v-2-yD0_ehq&
z?R3^iFo`zKVWp>C*1rIn^5tPSVvbmqR`FI)mU<9rSX)hc>)O!LclI|mPiylA9qK#6
z>V51FIHwC-@*>#19b0nZTyXfJ$!zgR_kX+HxnA;l)SDHNvaO)bav(^L29m4<N%J*-
zg_;{qqE%W)?IJ%4v}{*{GWBzeOFzT@pzf|Jt>zcxZoQaqe7yTFWtyG~+tx6UhIY+7
zwaPYWP02X5N?d`1K9|*t0dyPhmSGv<hD0jVSgtgkkX?im_9T}jHNLfM*S51)4=;Rr
zytVCJ)>Dk7Oh)5DD!3=oyRv$X!G4Px`K7dnk9--p1u=H#bXJ%AIPPKImc<$$_Pp>_
zcUi*TimA`y;}&iAkLq-MC+t_Vaxc|a+oh<hOf6&xaw&X&u>*&3zfs`$)CuQ3&{R5N
zq{;14wC$~-9dV6saC`4q=G`Y3Z}DrI=pqSX`lTjgvA<E)V((C=OkfTM+4Je3TolAj
zz0k|GXgM-z+`jjYSqJ>vC7YsZ%Q%ALp2ctUK?m=qyr(C<P0_w1_u;&;FGG;2l)E-a
zs*A(b$k#HOkH<EWT*At}6^+m8PtS@Q4SE17BMYp8sUK|@!1VO?Ska;I`*sm%D45y;
z^6z>?*i?SS?KlH3x7GLPXI{blH)0;a%5McScd48prkyQwgdvx0j}wEDA7<2*_Q07?
zW{)VCL*e=f4l(WlQ%rtw!2WV|hTtVhPm%g#?Xv#?P8w?nUw{71%`L^HYVreXS;;UC
zX~gKqlU7R11Fjxs-NocoG>5h#d)?Q9K0$y@SCmgtqj}W47a@@XEx95XU@xs7;Ii?~
zXKW?e;GG$@i&&e9Nc)R?{CUfY6#u$J+2y|EB$|)@+=COXrG!(f*Cshl7F^%`Jt<Be
z2T{^)@At4^^6w!g&O=fla7Jxwid!DX2raO*NqR;Ptox|$J>(wEyfjhzDXi-&hLt(@
z1Nyi)Mp03>iL}Vn4jlo#<S-EynGY}S0>{ZO?>R-dZ4K1$;Ut&0{dZ=Sy}pi<iap#?
zQ+SzmfBTj*A;<gvQ_gO+^nbe`=C&Q)@2;&aB)THt?5sON)nzmskP}5zjn@nM$9?Bp
zNMYhwN$B)#=0KasQ#rgJ*1o;t?_rU$@I(c8qL0+dk7IV4zr?WNX`Wxly{Y4Oqh}!K
zA7)Fe==Vd>2j7XfC!Fr>>-c7|_3U3;-PU&DHhRq!=d=yB8ZK~}LsL6kgvc+dZBI#$
zW&ymf12*mm6%U4&i)~eND7+@$p*$>d44!KA=Tlp$LAKiAiPwoof;?8YLm1mUP)e$4
z?)ysAnZV<HxgM*7<=e9&|59yvr+WqXdOhk(ksk08A993K=PUKTgJtGbJfUZ(1}Y9L
z9cT#~;Wo*^aH^V5!vjU^E)sKKuNQ9t<>F{$pb^^!5kBvP(%ORuWu$U@B}!3IzxP+U
zO~HfQ4;g8}!DSA6qBx0EXN5Di2p(>p=i{?`8MzLAc0>1JZ~d(8QZNo$t(aoGM=I|9
zKSE%Flbo)FMY_OSi6U<m!F;~N@cI`&K*T-WubF(KWUrK5o#5Z7mYqQMbZDer<jm(V
zRQc1rhPYJ++`S>^*dtT|Ls0nCvkEbKN6E{7z6?V!{d`GgXU0`HIEN1u_src-yRQ9f
z;n#NUXU<(3)H+r)c>~Ss|D8?7-_y~E620>fh1zO}`Qg}UJS+2>=a#aOv0%YKF!M?|
zFT4h<z-3sW-L}09eEmLZpM-gw0scguZR3x5aN=B)F0-U1g7(I`Q=Tw_ujqneY9;^i
zaTsSrKsQ!78)yh$D~*n*QSHT()_PNe?0?_7f7D(@>&y6sMPOZ%NU{nO6q1I#mZb4n
zm;RcsLeiuqLF)kVh`PWEBRc>I=8M+)uq+PKuYY_SQ97eP;F6AtIw=3<$F^ra%GLKM
zoWHj@!izXb33J0oq~z>x=V9l$8voc~JKJ?<Hbq5aYkU};(I+3lZhpmp_>$81xrQ2z
zTmMpiY562eq>M;W+}t${^Prm#8n_b<V~Z*!_WWM>5>_*RT^1g=MKy+)13Tg4%82|g
zA<r@9^CHp^x^}Jk)eekN{p(N^ozb<RDAjn~udgel)QVeR^W-x(Q4pzyZLh!Pz%nQG
z%0xS|P-f0vwC({_BYi=nZ-Rhmleg#gk0uKLh=*zHClw=#;QGQnm`2)v3zAA>w})Ia
zP;O!OUt=&5Ek)g4oW1P<p+=r<Z?bPI*@+d?-u#MP$LA7TMciN)HCvzVC2J1sdmkXu
z1Ihh|oY=nem6@3^qr0W+ENSrNjIWdSBsSUH_0e-@SiO%4JDdu<;I#Ms9;BSXX*@#h
z$_n<f>+P>OCBdqNcxZXz8RZ8g4xYs~$#<=O?u<Ho3>&9GRTj&KjQK3(i)E;8lt^m2
zxT_qMu@APMifvdl&ba+Ux@)N^SCt{qNqpghcpiu!`_bw+9vz6Y5o=}kG1eYG{{eR7
zRXa~&jy%YH*GJz-M?>gq8}=&v>girQE*pZilf>w(dc;DJU5e<DI6s_EN*>{iO8%S8
zG7m@ydeS^^f?J(caS?Nq={SE&_)qt~TsX<L>OY@=_=dQ7gMLfkzy9H;0u%fL62qv1
z6hVeC8fSv#xvCyfguhPvGpbYvA@QQgGO5<6lek=If^b3R3Cx;DB@$1X@`p9A-JnI)
z7j2!n!}<?tPhHo@54J=ilpNWB7d;j_G2M0*8Z3}<8WwKx&JO66V9~XL=f{Qa5L_Hu
zLPa@Xo3yEN;^xHLEUW6uSo>aY!f?ey(fr&u^c^S)%mf|43IuAxW63=k!t=P0YM+-?
z$N@2M{r{-uY-5_Z<2YVt)J%bn5R<WjNOq-J9Ksk>P%M$D^Rz1wa)5woeHa~))&-oU
z1FlRWW;ht$F$skNN!@UV3}jJak3upG<Cu&r#e%GbE#7cEtUwvhrQPo`v#}R@vv+%a
z|4aV&yWivIzW-~nfc5MLqnNe(`eL6kRfRq;F+{#QI`gOCASgR@SsXWAj*GR1<P+Vy
z#9ynmip!>aId)LU96=YuV!ub7_o>y79G!WdXX-+u%O9;X{H%K#H)PkvP1UwfSW$dG
zlS@tBL-9C9U|LJs_F*CxsAqpi5)zy$fA`gJ6mO*ElJ0^{Td~J}q-}0DpSJWaPhbo6
z{KZ%7>y`tuY%z{7uM!3Q_*4Ed=TQ1k{m^foTg~#ydX6A_6<(Nwj&{WV+4k`F-ujcz
z;ZaQY+vB4Idj?x*^*38<=(YYgOYbGYvz}J10)&KIU?5#BNLMP>(tAFTzD>785Bs(2
zi*|u4MtphVnyrHE0H}O)#Zdxot*q&gHGO`Do09fWwC4-TU09+KtoI<gQy3r&D^f4Z
zxsJdLy{Ee?QpnVhu}Yr_$Dl9j{>7R=O<GX*UPg#Ys#r>@&$?L6Rw84y>~QhJj&L%5
zhr+k6c8;I!GyX|)eetQ|Nm&|Q2jxCZq_E@k$XE<Vcs8S>l`IK=7Sr^ztGVCZX!X2w
zA+j}9K+Cw|(RlZ$8_^D02ky8Ihs@|+uZ10-%Rg&?5hzvnttdt_CBhcwfVJdgsZHg9
zRje3wkf>SPKv#jt8}O&a);0=_YiqFhGI!Hr0EEgnrnQu&$z&AIVS|@OSD%!(i&hqQ
z+Qlw$wWEyj>|^+$D}qX%z(~s<d~m&N?9psko5l|70Ho1$OC5&uY%p;uu0nFRmcmXn
zqKoU()}3f)JMDS@=!z7IT6RNOTH})OW{J0^L9J12xwMq<uJfInVN&057xF2zC7I;o
zzU52A%&(4_li9pS$(pWc#z+=qmWD%4Y;d?AX1BedD1j)?1s@;IgV*Z^2Y{whnopvk
z-m?SAIflAT&pZ9CrZpkjDPcH{DKl>5n5wY@giFfxn|NQU&cLE55E)&p>MYM|$}uF2
ziDEp*+DQ4dq=L0MBsATqo#z+52%)Wq2)^S5c5~nnj341rhm=KImZgMTmNQ~iUbFPn
z?0}1z&toUqZk`S(3cO8$0;3#JA`g_rk3u-wT@~@f?Fu3@D%sEMg+H`8V=Jir5saC%
zi5N3sdxHriYAcw#0oCaIw6Y!FJ8<jJVi=h*Xb;MWAK~SvVfF~7>$tO5YMmvC0Fv#I
zS|^t{K$jxYp3}^vCsI5CWZ+gDtkiS}Y<`6@<r3RMVnP{!gRe0kqvieQLum<vasJtd
zfWl#4b>w`9nQl~VPvhDhHBc4^{!!ybi5oXQ{Pn$zmnzw)AlxW_gYx?KvVZ6q|9=X?
gf7fRIhi4%?-di?%H@)l}aB~JikH`*x6|cDR2j$QOwEzGB

literal 0
HcmV?d00001

diff --git a/index.html b/index.html
index 6d3e1bd..749cef5 100644
--- a/index.html
+++ b/index.html
@@ -693,7 +693,7 @@
         </div>
         <div class="panel-body">
           <ul>
-              <li>Coming soon.</li>
+              <li><a href="./cryptocurrencies/mayer_multiple.html">Computing the Mayer multiple</a></li>
           </ul>
         </div>
       </div>
@@ -721,7 +721,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/jupyter/clear_cell.html b/jupyter/clear_cell.html
index a370f76..4ce277d 100644
--- a/jupyter/clear_cell.html
+++ b/jupyter/clear_cell.html
@@ -149,7 +149,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/jupyter/remote_run_notebook.html b/jupyter/remote_run_notebook.html
index d0aa852..3e1e92b 100644
--- a/jupyter/remote_run_notebook.html
+++ b/jupyter/remote_run_notebook.html
@@ -211,7 +211,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/Reloading_.bashrc.html b/linux/Reloading_.bashrc.html
index 55a5437..77864f4 100644
--- a/linux/Reloading_.bashrc.html
+++ b/linux/Reloading_.bashrc.html
@@ -144,7 +144,7 @@ However, you can use the following command to do it </p>
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/Reloading_fstab.html b/linux/Reloading_fstab.html
index 249b0a2..ee01002 100644
--- a/linux/Reloading_fstab.html
+++ b/linux/Reloading_fstab.html
@@ -143,7 +143,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/bashrc.html b/linux/bashrc.html
index 2d80d6c..a4caaf7 100644
--- a/linux/bashrc.html
+++ b/linux/bashrc.html
@@ -203,7 +203,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/directory_size.html b/linux/directory_size.html
index ae18c90..50a0f81 100644
--- a/linux/directory_size.html
+++ b/linux/directory_size.html
@@ -148,7 +148,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/free_disk_space_linux.html b/linux/free_disk_space_linux.html
index 31f9a2c..1b9abdd 100644
--- a/linux/free_disk_space_linux.html
+++ b/linux/free_disk_space_linux.html
@@ -167,7 +167,7 @@ none               0     0     0    - /config
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/get_ip_linux.html b/linux/get_ip_linux.html
index f5674b2..02b6ee2 100644
--- a/linux/get_ip_linux.html
+++ b/linux/get_ip_linux.html
@@ -256,7 +256,7 @@ wifi0     Link encap:UNSPEC  HWaddr 5C-51-4F-41-7A-AD-00-00-00-00-00-00-00-00-00
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/linux_process_information.html b/linux/linux_process_information.html
index 5d77d5c..2ea3e94 100644
--- a/linux/linux_process_information.html
+++ b/linux/linux_process_information.html
@@ -172,7 +172,7 @@ USER;PID;%CPU;%MEM;VSZ;RSS;TTY;STAT;START;TIME;COMMAND</p>
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/mount_nfs_share_fstab.html b/linux/mount_nfs_share_fstab.html
index 576aafc..42dea0a 100644
--- a/linux/mount_nfs_share_fstab.html
+++ b/linux/mount_nfs_share_fstab.html
@@ -158,7 +158,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/share_nfs_share.html b/linux/share_nfs_share.html
index 2c16450..37085aa 100644
--- a/linux/share_nfs_share.html
+++ b/linux/share_nfs_share.html
@@ -151,7 +151,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/simple_bash_function.html b/linux/simple_bash_function.html
index 24cc76d..a5f8525 100644
--- a/linux/simple_bash_function.html
+++ b/linux/simple_bash_function.html
@@ -208,7 +208,7 @@ The secret number is 126
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/linux/version_kernel.html b/linux/version_kernel.html
index 2152db9..840866a 100644
--- a/linux/version_kernel.html
+++ b/linux/version_kernel.html
@@ -160,7 +160,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/machine-learning/LeastSquareRegressionLine.html b/machine-learning/LeastSquareRegressionLine.html
index c1cde1f..0200e73 100644
--- a/machine-learning/LeastSquareRegressionLine.html
+++ b/machine-learning/LeastSquareRegressionLine.html
@@ -234,7 +234,7 @@ Where Y=a+b*X
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/machine-learning/Multiple_Linear_Regression.html b/machine-learning/Multiple_Linear_Regression.html
index 9dc34ef..51f168c 100644
--- a/machine-learning/Multiple_Linear_Regression.html
+++ b/machine-learning/Multiple_Linear_Regression.html
@@ -248,7 +248,7 @@ b_2\\
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/machine-learning/Source code for the recommandation engine for articles.html b/machine-learning/Source code for the recommandation engine for articles.html
index 9a735ef..cda5cc5 100644
--- a/machine-learning/Source code for the recommandation engine for articles.html	
+++ b/machine-learning/Source code for the recommandation engine for articles.html	
@@ -1033,7 +1033,7 @@ Top dislike features:
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/mathematics/Bayes.html b/mathematics/Bayes.html
index d3934bc..763c88a 100644
--- a/mathematics/Bayes.html
+++ b/mathematics/Bayes.html
@@ -196,7 +196,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/mathematics/Binomial.html b/mathematics/Binomial.html
index 12c572a..929fe17 100644
--- a/mathematics/Binomial.html
+++ b/mathematics/Binomial.html
@@ -256,7 +256,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/mathematics/Geometric.html b/mathematics/Geometric.html
index ac45801..836beb1 100644
--- a/mathematics/Geometric.html
+++ b/mathematics/Geometric.html
@@ -233,7 +233,7 @@ So :
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/mathematics/Number_edges_Complete_graph.html b/mathematics/Number_edges_Complete_graph.html
index 2c33ff1..c209259 100644
--- a/mathematics/Number_edges_Complete_graph.html
+++ b/mathematics/Number_edges_Complete_graph.html
@@ -195,7 +195,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/mathematics/Premutations_and_combinations.html b/mathematics/Premutations_and_combinations.html
index b6072e0..3c7e787 100644
--- a/mathematics/Premutations_and_combinations.html
+++ b/mathematics/Premutations_and_combinations.html
@@ -203,7 +203,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/mathematics/cond_prob.html b/mathematics/cond_prob.html
index 43cb0ce..be43cfa 100644
--- a/mathematics/cond_prob.html
+++ b/mathematics/cond_prob.html
@@ -197,7 +197,7 @@ and B are considered to be independent if event A has no effect on the probabili
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/mathematics/normal.html b/mathematics/normal.html
index c838ba9..aafac47 100644
--- a/mathematics/normal.html
+++ b/mathematics/normal.html
@@ -221,7 +221,7 @@ where <span class="math">\(erf\)</span> is the error function:
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/mathematics/pearson.html b/mathematics/pearson.html
index 2c79806..8c9ea86 100644
--- a/mathematics/pearson.html
+++ b/mathematics/pearson.html
@@ -206,7 +206,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/mathematics/poisson.html b/mathematics/poisson.html
index 370ce69..6fed328 100644
--- a/mathematics/poisson.html
+++ b/mathematics/poisson.html
@@ -222,7 +222,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/mathematics/spearman.html b/mathematics/spearman.html
index 88128bf..ff4ad46 100644
--- a/mathematics/spearman.html
+++ b/mathematics/spearman.html
@@ -223,7 +223,7 @@ $$</div>
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/pages/about.html b/pages/about.html
index f4eed0e..c2aa0c4 100644
--- a/pages/about.html
+++ b/pages/about.html
@@ -145,7 +145,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Creating_a_sqlite_database.html b/python/Creating_a_sqlite_database.html
index e4fe6e8..e154797 100644
--- a/python/Creating_a_sqlite_database.html
+++ b/python/Creating_a_sqlite_database.html
@@ -192,7 +192,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Efficient_extraction_of_eigenvalues_from_a_list_of_tensors.html b/python/Efficient_extraction_of_eigenvalues_from_a_list_of_tensors.html
index 123620c..c86e7b4 100644
--- a/python/Efficient_extraction_of_eigenvalues_from_a_list_of_tensors.html
+++ b/python/Efficient_extraction_of_eigenvalues_from_a_list_of_tensors.html
@@ -241,7 +241,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Iterating_over_a_dataframe.html b/python/Iterating_over_a_dataframe.html
index 6d07d4c..911bb46 100644
--- a/python/Iterating_over_a_dataframe.html
+++ b/python/Iterating_over_a_dataframe.html
@@ -235,7 +235,7 @@ At line 5 there is a plum which is purple and contains 28 kcal
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Moving_average_pandas.html b/python/Moving_average_pandas.html
index 2c12e18..0905d2c 100644
--- a/python/Moving_average_pandas.html
+++ b/python/Moving_average_pandas.html
@@ -378,7 +378,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Opening_SQLite_database.html b/python/Opening_SQLite_database.html
index ef1ffb3..3a4e172 100644
--- a/python/Opening_SQLite_database.html
+++ b/python/Opening_SQLite_database.html
@@ -168,7 +168,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Opening_file.html b/python/Opening_file.html
index b3bb7d2..94699dd 100644
--- a/python/Opening_file.html
+++ b/python/Opening_file.html
@@ -192,7 +192,7 @@ If you try to open a non existing file in read mode 'r', a FileNotFoundError wil
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Optimized_numpy_random_intel.html b/python/Optimized_numpy_random_intel.html
index 5dc9937..2192546 100644
--- a/python/Optimized_numpy_random_intel.html
+++ b/python/Optimized_numpy_random_intel.html
@@ -176,7 +176,7 @@ conda install numpy --channel intel --override-channels
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Reading_data_from_a_sql_database_with_pandas.html b/python/Reading_data_from_a_sql_database_with_pandas.html
index 03569bb..fa8d028 100644
--- a/python/Reading_data_from_a_sql_database_with_pandas.html
+++ b/python/Reading_data_from_a_sql_database_with_pandas.html
@@ -237,7 +237,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Saving_a_matplotlib_figure_with_a_high_resolution.html b/python/Saving_a_matplotlib_figure_with_a_high_resolution.html
index bf71b51..479141e 100644
--- a/python/Saving_a_matplotlib_figure_with_a_high_resolution.html
+++ b/python/Saving_a_matplotlib_figure_with_a_high_resolution.html
@@ -186,7 +186,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Setting_up_the_notebook_for_plotting_with_matplotlib.html b/python/Setting_up_the_notebook_for_plotting_with_matplotlib.html
index d71dfe1..d59d322 100644
--- a/python/Setting_up_the_notebook_for_plotting_with_matplotlib.html
+++ b/python/Setting_up_the_notebook_for_plotting_with_matplotlib.html
@@ -171,7 +171,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/Writing_data_to_a_sql_database_with_pandas.html b/python/Writing_data_to_a_sql_database_with_pandas.html
index d725ca4..400096d 100644
--- a/python/Writing_data_to_a_sql_database_with_pandas.html
+++ b/python/Writing_data_to_a_sql_database_with_pandas.html
@@ -339,7 +339,7 @@ However, exporting to a sql database can prove very useful. Indeed, having a wel
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/counting.html b/python/counting.html
index 302945f..8ff9673 100644
--- a/python/counting.html
+++ b/python/counting.html
@@ -181,7 +181,7 @@ The number 4 appears 1 times in the list
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/dask_distributed_parallelism.html b/python/dask_distributed_parallelism.html
index 5523543..efbec15 100644
--- a/python/dask_distributed_parallelism.html
+++ b/python/dask_distributed_parallelism.html
@@ -338,7 +338,7 @@ http://127.0.0.1:8787/status</p>
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/dask_infiniband.html b/python/dask_infiniband.html
index 37e6c19..f37cfb2 100644
--- a/python/dask_infiniband.html
+++ b/python/dask_infiniband.html
@@ -148,7 +148,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/design_own_libs.html b/python/design_own_libs.html
index fa1c851..24282b6 100644
--- a/python/design_own_libs.html
+++ b/python/design_own_libs.html
@@ -232,7 +232,7 @@ The list of dependencies required to test, build and generate the doc are listed
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/dict_comprehension.html b/python/dict_comprehension.html
index dae6ff6..a50458e 100644
--- a/python/dict_comprehension.html
+++ b/python/dict_comprehension.html
@@ -164,7 +164,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/download_page.html b/python/download_page.html
index a295454..0bdcbda 100644
--- a/python/download_page.html
+++ b/python/download_page.html
@@ -218,7 +218,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/ensure_dir.html b/python/ensure_dir.html
index 52f93e4..60a2a9b 100644
--- a/python/ensure_dir.html
+++ b/python/ensure_dir.html
@@ -177,7 +177,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/list_files_directory.html b/python/list_files_directory.html
index 68b122e..7fd6375 100644
--- a/python/list_files_directory.html
+++ b/python/list_files_directory.html
@@ -213,7 +213,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/logplot.html b/python/logplot.html
index 58c9921..cc2ea74 100644
--- a/python/logplot.html
+++ b/python/logplot.html
@@ -248,7 +248,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/randint.html b/python/randint.html
index 8ae4d78..7cc8e1d 100644
--- a/python/randint.html
+++ b/python/randint.html
@@ -156,7 +156,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/reverse.html b/python/reverse.html
index 0364841..fe3a466 100644
--- a/python/reverse.html
+++ b/python/reverse.html
@@ -173,7 +173,7 @@ Reversed array : [4 3 2 1 0]
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/sorting.html b/python/sorting.html
index cd4f4e0..2ac827c 100644
--- a/python/sorting.html
+++ b/python/sorting.html
@@ -192,7 +192,7 @@ Sorted array : [0.13876208 0.19939047 0.40021786 0.43734158 0.46015169]
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/stock_pandas.html b/python/stock_pandas.html
index 0537cdc..afa4778 100644
--- a/python/stock_pandas.html
+++ b/python/stock_pandas.html
@@ -381,7 +381,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/unique.html b/python/unique.html
index 6e9cc5d..ffa47e8 100644
--- a/python/unique.html
+++ b/python/unique.html
@@ -171,7 +171,7 @@ Numpy array with unique values : [10 20 30 40 50 60]
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/python/updating_all_python_package_with_anaconda.html b/python/updating_all_python_package_with_anaconda.html
index ca54965..94496d2 100644
--- a/python/updating_all_python_package_with_anaconda.html
+++ b/python/updating_all_python_package_with_anaconda.html
@@ -151,7 +151,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/search.html b/search.html
index 419b826..143b6a8 100644
--- a/search.html
+++ b/search.html
@@ -112,7 +112,7 @@
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/sitemap.xml b/sitemap.xml
index 1cd0533..b0751b2 100644
--- a/sitemap.xml
+++ b/sitemap.xml
@@ -5,7 +5,7 @@ xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
 
 <url>
 <loc>redoules.github.io/</loc>
-<lastmod>2018-12-05T21:25:38-00:00</lastmod>
+<lastmod>2018-12-16T09:36:30-00:00</lastmod>
 <changefreq>daily</changefreq>
 <priority>0.5</priority>
 </url>
@@ -17,6 +17,13 @@ xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
 <priority>0.5</priority>
 </url>
 
+<url>
+<loc>redoules.github.io/cryptocurrencies/mayer_multiple.html</loc>
+<lastmod>2018-12-16T08:51:00+01:00</lastmod>
+<changefreq>monthly</changefreq>
+<priority>0.5</priority>
+</url>
+
 <url>
 <loc>redoules.github.io/dl/keras_load.html</loc>
 <lastmod>2018-12-05T21:19:00+01:00</lastmod>
diff --git a/sql/Sorting_results.html b/sql/Sorting_results.html
index 31b4ae7..5e44431 100644
--- a/sql/Sorting_results.html
+++ b/sql/Sorting_results.html
@@ -344,7 +344,7 @@ Done.
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/sql/WHERE_SQL_keywords.html b/sql/WHERE_SQL_keywords.html
index f526668..1cfd350 100644
--- a/sql/WHERE_SQL_keywords.html
+++ b/sql/WHERE_SQL_keywords.html
@@ -631,7 +631,7 @@ Done.
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/sql/display_table.html b/sql/display_table.html
index 325edf3..4070dd2 100644
--- a/sql/display_table.html
+++ b/sql/display_table.html
@@ -216,7 +216,7 @@ Done.
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/sql/display_table_filter.html b/sql/display_table_filter.html
index 08d62b2..89c3b10 100644
--- a/sql/display_table_filter.html
+++ b/sql/display_table_filter.html
@@ -346,7 +346,7 @@ Done.
     <footer class="footer">
         <div class="container">
             <p class="text-muted">
-                <center>This project contains 75 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
+                <center>This project contains 76 pages and is available on <a href="https://github.com/redoules/redoules.github.io">GitHub</a>.
                 <br/>
                 Copyright &copy; Guillaume Redoulès,
                     <time datetime="2018">2018</time>.
diff --git a/tipuesearch_content.json b/tipuesearch_content.json
index 810f8dc..06db80c 100644
--- a/tipuesearch_content.json
+++ b/tipuesearch_content.json
@@ -1 +1 @@
-{"pages":[{"title":"About Guillaume Redoulès","text":"I am a data scientist and a mechanical engineer working on numerical methods for stress computations in the field of rocket propulsion. Prior to that, I've got a MSc in Computational Fluid Dynamics and aerodynamics from Imperial College London. Email: guillaume.redoules@gadz.org Linkedin: Guillaume Redoulès Curriculum Vitae Experience Thermomecanical method and tools engineer , Ariane Group , 2015 - Present In charge of tools and methods related to thermomecanical computations. Focal point for machine learning. Education MSc Advanced Computational Methods for Aeronautics, Flow Management and Fluid-Structure Interaction , Imperial College London, London. 2013 Dissertation: \"Estimator design for fluid flows\" Fields: Aeronautics, aerodynamics, computational fluid dynamics, numerical methods Arts et Métiers Paristech , France, 2011 Generalist engineering degree Fields: Mechanics, electrical engineering, casting, machining, project management, finance, IT, etc.","tags":"pages","url":"redoules.github.io/pages/about.html","loc":"redoules.github.io/pages/about.html"},{"title":"Load a saved model","text":"A keras model can be loaded from a hdf5 file. Becareful, a keras generated hdf5 can contain either : the model weights (obtained by the .save_weights() method) the model weights and the model architecture (obtained by the .save() method) In our case, we will load both the model and its weights with the load_model function from the keras.models module from keras.models import load_model model = load_model ( \"my_model.h5\" ) type ( model ) Using TensorFlow backend. keras.engine.sequential.Sequential","tags":"DL","url":"redoules.github.io/dl/keras_load.html","loc":"redoules.github.io/dl/keras_load.html"},{"title":"Get model input shape","text":"Information on the model can be accessed by the .input_shape parameter of the model object from keras.models import load_model model = load_model ( \"my_model.h5\" ) model . input_shape Using TensorFlow backend. (None, 28, 28)","tags":"DL","url":"redoules.github.io/dl/keras_input.html","loc":"redoules.github.io/dl/keras_input.html"},{"title":"Get model info and number of parameters","text":"Information on the model can be accessed by the .summary() method from keras.models import load_model model = load_model ( \"my_model.h5\" ) model . summary () _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= flatten_2 (Flatten) (None, 784) 0 _________________________________________________________________ dense_3 (Dense) (None, 128) 100480 _________________________________________________________________ dense_4 (Dense) (None, 10) 1290 ================================================================= Total params: 101,770 Trainable params: 101,770 Non-trainable params: 0 _________________________________________________________________","tags":"DL","url":"redoules.github.io/dl/keras_info.html","loc":"redoules.github.io/dl/keras_info.html"},{"title":"Saving the models weights after each epoch","text":"Let's see how we can save the model weights after every epoch. Let's first import some libraries import keras import numpy as np Using TensorFlow backend. In this example, we will be using the fashion MNIST dataset to do some basic computer vision, where we will train a Keras neural network to classify items of clothing. In order to import the data we will be using the built in function in Keras : keras . datasets . fashion_mnist . load_data () The model is a very simple neural network consisting in 2 fully connected layers. The model loss function is chosen in order to have a multiclass classifier : \"sparse_categorical_crossentropy\" Let's define a simple feedforward network. ##get and preprocess the data fashion_mnist = keras . datasets . fashion_mnist ( train_images , train_labels ), ( test_images , test_labels ) = fashion_mnist . load_data () train_images = train_images / 255.0 test_images = test_images / 255.0 ## define the model model = keras . Sequential ([ keras . layers . Flatten ( input_shape = ( 28 , 28 )), keras . layers . Dense ( 128 , activation = \"relu\" ), keras . layers . Dense ( 10 , activation = \"softmax\" ) ]) model . compile ( optimizer = \"adam\" , loss = \"sparse_categorical_crossentropy\" , metrics = [ \"acc\" ]) In order to automatically save the model weights to a hdf5 format after every epoch we need to import the ModelCheckpoint callback located in keras . callbacks from keras.callbacks import ModelCheckpoint We now need to define the ModelCheckPoint callback, it takes 7 arguments : filepath: string, path to save the model file. monitor: quantity to monitor verbose: verbosity mode, 0 or 1 save_best_only: if save_best_only=True , the latest best model according to the quantity monitored will not be overwritten. mode: one of {auto, min, max} If save_best_only=True , the decision to overwrite the current save file is made based on either the maximization or the minimization of the monitored quantity. save_weights_only: if True, then only the model's weights will be saved ( model.save_weights(filepath) ), else the full model is saved ( model.save(filepath) ). period: Interval (number of epochs) between checkpoints. The callback has to be added to the callbacks list in the fit method. save_to_hdf5 = ModelCheckpoint ( filepath = \"my_model.h5\" , monitor = 'acc' , verbose = 0 , save_best_only = True , save_weights_only = False , mode = 'auto' , period = 1 ) model . fit ( train_images , train_labels , epochs = 5 , callbacks = [ save_to_hdf5 ]) Epoch 1/5 60000/60000 [==============================] - 29s 478us/step - loss: 0.3975 - acc: 0.8575 Epoch 2/5 60000/60000 [==============================] - 98s 2ms/step - loss: 0.3498 - acc: 0.8721 Epoch 3/5 60000/60000 [==============================] - 95s 2ms/step - loss: 0.3213 - acc: 0.8825 Epoch 4/5 60000/60000 [==============================] - 64s 1ms/step - loss: 0.3021 - acc: 0.8887: 6s - loss: Epoch 5/5 60000/60000 [==============================] - 61s 1ms/step - loss: 0.2855 - acc: 0.8953: 4s - loss: 0.284 - ETA: 3s - loss: - ETA: &lt;keras.callbacks.History at 0x1d70aa62f28&gt; import os if \"my_model.h5\" in os . listdir () : print ( \"The model is saved to my_model.h5\" ) else : print ( 'no model saved to disk' ) The model is saved to my_model.h5","tags":"DL","url":"redoules.github.io/dl/model_checkpoint_keras.html","loc":"redoules.github.io/dl/model_checkpoint_keras.html"},{"title":"Logging the training progress in a CSV","text":"Let's see how we can log the progress and various metrics during the training process to a csv file. Let's first import some libraries import keras import numpy as np In this example, we will be using the fashion MNIST dataset to do some basic computer vision, where we will train a Keras neural network to classify items of clothing. In order to import the data we will be using the built in function in Keras : keras . datasets . fashion_mnist . load_data () The model is a very simple neural network consisting in 2 fully connected layers. The model loss function is chosen in order to have a multiclass classifier : \"sparse_categorical_crossentropy\" Let's define a simple feedforward network. ##get and preprocess the data fashion_mnist = keras . datasets . fashion_mnist ( train_images , train_labels ), ( test_images , test_labels ) = fashion_mnist . load_data () train_images = train_images / 255.0 test_images = test_images / 255.0 ## define the model model = keras . Sequential ([ keras . layers . Flatten ( input_shape = ( 28 , 28 )), keras . layers . Dense ( 128 , activation = \"relu\" ), keras . layers . Dense ( 10 , activation = \"softmax\" ) ]) model . compile ( optimizer = \"adam\" , loss = \"sparse_categorical_crossentropy\" , metrics = [ \"accuracy\" , 'mae' ]) In order to stream to a csv file the epoch results and metrics, we define a CSV logger. It is a callback located in keras . callbacks Let's first import it from keras.callbacks import CSVLogger We now need to define the callback by specifiying a file to be written to, the separator and whether to append to the file or erase it every time. The callback has to be added to the callbacks list in the fit method. csv_logger = CSVLogger ( filename = \"my_csv.csv\" , separator = ';' , append = False ) model . fit ( train_images , train_labels , epochs = 5 , callbacks = [ csv_logger ]) Epoch 1/5 60000/60000 [==============================] - 9s 148us/step - loss: 0.5020 - acc: 0.8234 - mean_absolute_error: 4.4200 Epoch 2/5 60000/60000 [==============================] - 8s 138us/step - loss: 0.3765 - acc: 0.8630 - mean_absolute_error: 4.4200 Epoch 3/5 60000/60000 [==============================] - 8s 129us/step - loss: 0.3371 - acc: 0.8789 - mean_absolute_error: 4.4200 Epoch 4/5 60000/60000 [==============================] - 8s 133us/step - loss: 0.3129 - acc: 0.8843 - mean_absolute_error: 4.4200 Epoch 5/5 60000/60000 [==============================] - 9s 151us/step - loss: 0.2952 - acc: 0.8916 - mean_absolute_error: 4.4200 &lt;keras.callbacks.History at 0x1582adc6780&gt; The results are stored in the my_csv.csv file and contain the epoch results import pandas as pd pd . read_csv ( \"my_csv.csv\" , sep = \";\" ) .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } epoch acc loss mean_absolute_error 0 0 0.823400 0.502013 4.42 1 1 0.863050 0.376516 4.42 2 2 0.878867 0.337097 4.42 3 3 0.884317 0.312893 4.42 4 4 0.891583 0.295157 4.42","tags":"DL","url":"redoules.github.io/dl/csv_logger_keras.html","loc":"redoules.github.io/dl/csv_logger_keras.html"},{"title":"Random integer","text":"The random module comes with the randint function that returns a pseudo random number between 2 values : # Return a random integer N such that a <= N <= b. random . randint ( a , b ) #### import random min_int = 126 max_int = 211 print ( f \"My pseudo random number between {min_int} and {max_int} : {random.randint(min_int, max_int)}\" ) My pseudo random number between 126 and 211 : 206","tags":"Python","url":"redoules.github.io/python/randint.html","loc":"redoules.github.io/python/randint.html"},{"title":"Using tensorboard with Keras","text":"general workflow Let's see how we can get tensorboard to work with a Keras-based Tensorflow code. import tensorflow as tf import keras import numpy as np Using TensorFlow backend. In this example, we will be using the fashion MNIST dataset to do some basic computer vision, where we will train a Keras neural network to classify items of clothing. In order to import the data we will be using the built in function in Keras : keras . datasets . fashion_mnist . load_data () The model is a very simple neural network consisting in 2 fully connected layers. The model loss function is chosen in order to have a multiclass classifier : \"sparse_categorical_crossentropy\" Finally, let's train the model for 5 epochs ##get and preprocess the data fashion_mnist = keras . datasets . fashion_mnist ( train_images , train_labels ), ( test_images , test_labels ) = fashion_mnist . load_data () train_images = train_images / 255.0 test_images = test_images / 255.0 ## define the model model = keras . Sequential ([ keras . layers . Flatten ( input_shape = ( 28 , 28 )), keras . layers . Dense ( 128 , activation = \"relu\" ), keras . layers . Dense ( 10 , activation = \"softmax\" ) ]) model . compile ( optimizer = \"adam\" , loss = \"sparse_categorical_crossentropy\" , metrics = [ \"accuracy\" ]) model . fit ( train_images , train_labels , epochs = 5 ) Epoch 1/5 60000/60000 [==============================] - 9s 143us/step - loss: 0.4939 - acc: 0.8254 Epoch 2/5 60000/60000 [==============================] - 11s 182us/step - loss: 0.3688 - acc: 0.8661 Epoch 3/5 60000/60000 [==============================] - 10s 169us/step - loss: 0.3305 - acc: 0.8798 Epoch 4/5 60000/60000 [==============================] - 21s 350us/step - loss: 0.3079 - acc: 0.8874 Epoch 5/5 60000/60000 [==============================] - 18s 302us/step - loss: 0.2889 - acc: 0.8927 &lt;keras.callbacks.History at 0x235c1bc1be0&gt; During the training we can see the process, including the loss and the accuracy in the output. test_loss , test_acc = model . evaluate ( test_images , test_labels ) print ( f \"Test accuracy : {test_acc}\" ) 10000/10000 [==============================] - 1s 67us/step Test accuracy : 0.8763 When the model finishes training, we get an accuracy of about 87%, and we output some sample predictions predictions = model . predict ( test_images ) print ( predictions [ 0 ]) [1.8075149e-05 3.6810281e-08 6.3094416e-07 5.1111499e-07 1.6264809e-06 3.5973577e-04 1.0840570e-06 3.1453002e-02 1.7062060e-06 9.6816361e-01] This kind of process only gives us minimal information during the training process. Setting up tensorboard To make it easier to understand, debug, and optimize TensorFlow programs, a suite of visualization tools called TensorBoard is included. You can use TensorBoard to visualize your TensorFlow graph, plot quantitative metrics about the execution of your graph, and show additional data like images that pass through it. When TensorBoard is fully configured, it looks like this: Let's start by importing the time library and tensorboard itself. It can be found in tensorflow.python.keras.callbacks. from time import time from tensorflow.python.keras.callbacks import TensorBoard After having imported our data and defined our model, we specify a log directory where the training information will get written to. #keep in mind that we already imported the data and defined the model. tensorboard = TensorBoard ( log_dir = f \"logs/{time()}\" ) Finally, to tell Keras to call back to TensorBoard we refer to the instant of TensorBoard we created. model . compile ( optimizer = \"adam\" , loss = \"sparse_categorical_crossentropy\" , metrics = [ \"accuracy\" ]) Now, we need to execute the TensorBoard command pointing at the log directory previously specified. tensorboard --logdir = logs/ TensorBoard will return a http address TensorBoard 1.12.0 at http://localhost:6006 (Press CTRL+C to quit) Now, if we retrain again, we can take a look in TensorBoard and start investigating the loss and accuracy model . fit ( train_images , train_labels , epochs = 5 , callbacks = [ tensorboard ]) Epoch 1/5 60000/60000 [==============================] - 41s 684us/step - loss: 0.4990 - acc: 0.8241 Epoch 2/5 60000/60000 [==============================] - 49s 812us/step - loss: 0.3765 - acc: 0.8648 Epoch 3/5 60000/60000 [==============================] - 46s 765us/step - loss: 0.3392 - acc: 0.8766 Epoch 4/5 60000/60000 [==============================] - 48s 794us/step - loss: 0.3135 - acc: 0.8836 Epoch 5/5 60000/60000 [==============================] - 49s 813us/step - loss: 0.2971 - acc: 0.8897 &lt;keras.callbacks.History at 0x235be1c76d8&gt; TensorBoard also give access to a dynamic visualization of the graph","tags":"DL","url":"redoules.github.io/dl/tensorboard_keras.html","loc":"redoules.github.io/dl/tensorboard_keras.html"},{"title":"Install keras using conda","text":"Keras with the Tensorflow backend can be installed by running the following conda command conda install -c conda-forge keras tensorflow If you want a Intel CPU optimized version, install tensorflow-mkl conda install -c conda-forge keras tensorflow-mkl A GPU compatible version is also available conda install -c conda-forge keras tensorflow-gpu","tags":"DL","url":"redoules.github.io/dl/keras_install.html","loc":"redoules.github.io/dl/keras_install.html"},{"title":"Day 9 - Multiple Linear Regression","text":"Problem Here is a simple equation: $$Y=a+b_1\\cdot f_1++b_2\\cdot f_2+...++b_m\\cdot f_m$$ $$Y=a+\\sum_{i=1}&#94;m b_i\\cdot f_i$$ for \\((m+1)\\) read constants \\((a,f_1, f_2, ..., f_m)\\) . We can say that the value of \\(Y\\) depends on \\(m\\) features. We study this equation for \\(n\\) different feature sets \\((f_1, f_2, ..., f_m)\\) and records each respective value of \\(Y\\) . If we have \\(q\\) new feature sets, and without accounting for bias and variance trade-offs,what is the value of \\(Y\\) for each of the sets? Python implementation import numpy as np m = 2 n = 7 x_1 = [ 0.18 , 0.89 ] y_1 = 109.85 x_2 = [ 1.0 , 0.26 ] y_2 = 155.72 x_3 = [ 0.92 , 0.11 ] y_3 = 137.66 x_4 = [ 0.07 , 0.37 ] y_4 = 76.17 x_5 = [ 0.85 , 0.16 ] y_5 = 139.75 x_6 = [ 0.99 , 0.41 ] y_6 = 162.6 x_7 = [ 0.87 , 0.47 ] y_7 = 151.77 q_1 = [ 0.49 , 0.18 ] q_2 = [ 0.57 , 0.83 ] q_3 = [ 0.56 , 0.64 ] q_4 = [ 0.76 , 0.18 ] With scikit learn X = np . array ([ x_1 , x_2 , x_3 , x_4 , x_5 , x_6 , x_7 ]) Y = np . array ([ y_1 , y_2 , y_3 , y_4 , y_5 , y_6 , y_7 ]) X_q = np . array ([ q_1 , q_2 , q_3 , q_4 ]) from sklearn import linear_model lm = linear_model . LinearRegression () lm . fit ( X , Y ) lm . predict ( X_q ) array([105.21455835, 142.67095131, 132.93605469, 129.70175405]) without scikit learn (but with numpy) from numpy.linalg import inv #center X_R = X - np . mean ( X , axis = 0 ) a = np . mean ( Y ) Y_R = Y - a #calculate b B = inv ( X_R . T @X_R ) @X_R.T@Y_R #predict X_new_R = X_q - np . mean ( X , axis = 0 ) Y_new_R = X_new_R @B Y_new = Y_new_R + a Y_new array([105.21455835, 142.67095131, 132.93605469, 129.70175405]) if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day9.html","loc":"redoules.github.io/blog/Statistics_10days-day9.html"},{"title":"Multiple Linear Regression","text":"If \\(Y\\) is linearly dependent only on \\(X\\) , then we can use the ordinary least square regression line, \\(\\hat{Y}=a+bX\\) . However, if \\(Y\\) shows linear dependency on \\(m\\) variables \\(X_1\\) , \\(X_2\\) , ..., \\(X_m\\) , then we need to find the values of \\(a\\) and \\(m\\) other constants ( \\(b_1\\) , \\(b_2\\) , ..., \\(b_m\\) ). We can then write the regression equation as: $$\\hat{Y}=a+\\sum_{i=1}&#94;{m}b_iX_i$$ Matrix Form of the Regression Equation Let's consider that \\(Y\\) depends on two variables, \\(X_1\\) and \\(X_2\\) . We write the regression relation as \\(\\hat{Y}=a+b_1X_1+b_2X_2\\) . Consider the following matrix operation: $$\\begin{bmatrix} 1 & X_1 & X_2\\\\ \\end{bmatrix}\\cdot\\begin{bmatrix} a \\\\ b_1\\\\ b_2\\\\ \\end{bmatrix}=a+b_1X_1+b_2X_2$$ We define two matrices, \\(X\\) and \\(B\\) as: $$X=\\begin{bmatrix}1 & X_1 & X_2\\\\\\end{bmatrix}$$ $$B=\\begin{bmatrix}a \\\\b_1\\\\b_2\\\\\\end{bmatrix}$$ Now, we rewrite the regression relation as \\(\\hat{Y}=X\\cdot B\\) . This transforms the regression relation into matrix form. Generalized Matrix Form We will consider that \\(Y\\) shows a linear relationship with \\(m\\) variables, \\(X_1\\) , \\(X_2\\) , ..., \\(X_m\\) . Let's say that we made \\(n\\) observations on different tuples \\((x_1, x_2, ..., x_m)\\) : \\(y_1=a+b_1\\cdot x_{1,1} + b_2\\cdot x_{2,1} + ... + b_m\\cdot x_{m,1}\\) \\(y_2=a+b_2\\cdot x_{1,2} + b_2\\cdot x_{2,2} + ... + b_m\\cdot x_{m,2}\\) \\(...\\) \\(y_n=a+b_n\\cdot x_{1,n} + b_2\\cdot x_{2,n} + ... + b_m\\cdot x_{m,n}\\) Now, we can find the matrices: $$X=\\begin{bmatrix}1 & x_{1,1} & x_{2,1} & x_{3,1} & ... & x_{m,1} \\\\1 & x_{1,2} & x_{2,2} & x_{3,2} & ... & x_{m,2} \\\\1 & x_{1,3} & x_{2,3} & x_{3,3} & ... & x_{m,3} \\\\... & ... & ... & ... & ... & ... \\\\1 & x_{1,n} & x_{2,n} & x_{3,n} & ... & x_{m,n} \\\\\\end{bmatrix}$$ $$Y=\\begin{bmatrix}y_1 \\\\y_2\\\\y_3\\\\...\\\\y_n\\\\\\end{bmatrix}$$ Finding the Matrix B We know that \\(Y=X\\cdot B\\) $$\\Rightarrow X&#94;T\\cdot Y=X&#94;T\\cdot X \\cdot B$$ $$\\Rightarrow (X&#94;T\\cdot X)&#94;{-1}\\cdot X&#94;T \\cdot Y=I\\cdot B$$ $$\\Rightarrow B= (X&#94;T\\cdot X)&#94;{-1}\\cdot X&#94;T \\cdot Y$$ Finding the Value of Y Suppose we want to find the value of for some tuple \\(Y\\) , then \\((x_1, x_2, ..., x_m)\\) , $$Y=\\begin{bmatrix} 1 & x_1 & x_2 & ... & x_m\\\\ \\end{bmatrix}\\cdot B$$ Multiple Regression in Python We can use the fit function in the sklearn.linear_model.LinearRegression class. from sklearn import linear_model x = [[ 5 , 7 ], [ 6 , 6 ], [ 7 , 4 ], [ 8 , 5 ], [ 9 , 6 ]] y = [ 10 , 20 , 60 , 40 , 50 ] lm = linear_model . LinearRegression () lm . fit ( x , y ) a = lm . intercept_ b = lm . coef_ print ( f \"Linear regression coefficients between Y and X : a={a}, b_0={b[0]}, b_1={b[1]}\" ) Linear regression coefficients between Y and X : a=51.953488372092984, b_0=6.65116279069768, b_1=-11.162790697674419 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Machine Learning","url":"redoules.github.io/machine-learning/Multiple_Linear_Regression.html","loc":"redoules.github.io/machine-learning/Multiple_Linear_Regression.html"},{"title":"Least Square Regression Line","text":"Linear Regression If our data shows a linear relationship between \\(X\\) and \\(Y\\) , then the straight line which best describes the relationship is the regression line. The regression line is given by \\(\\hat{Y}\\) =a+bX$. Finding the value of b The value of \\(b\\) can be calculated using either of the following formulae: \\(b=\\frac{n\\sum(x_iy_i)-(\\sum x_i)(\\sum y_i)}{n\\sum(x_i&#94;2)-(\\sum x_i)&#94;2}\\) \\(b=\\rho\\frac{\\sigma_Y}{\\sigma_X}\\) , where \\(\\rho\\) is the Pearson correlation coefficient, \\(\\sigma_X\\) Finding the value of a \\(a=\\bar{y}-b\\cdot\\bar{x}\\) , where \\(\\bar{x}\\) is the mean of \\(X\\) and \\(\\bar{y}\\) is the mean of \\(Y\\) . Coefficient of determination ( \\(R&#94;2\\) ) The coefficient of determination can be computer with : \\(R&#94;2 = \\frac{SSR}{SST}=1-\\frac{SSE}{SST}\\) Where : \\(SST\\) is the total Sum of Squares : \\(SST=\\sum (y_i-\\bar{y})&#94;2\\) \\(SSR\\) is the regression Sum of Squares : \\(SSR=\\sum (\\hat{y_i}-\\bar{y})&#94;2\\) \\(SSE\\) is the error Sum of Squares : \\(SSE=\\sum (\\hat{y_i}-y)&#94;2\\) If \\(SSE\\) is small, we can assume that our fit is good. Linear Regression in Python We can use the fit function in the sklearn.linear_model.LinearRegression class. from sklearn import linear_model import numpy as np xl = [ 1 , 2 , 3 , 4 , 5 ] x = np . asarray ( xl ) . reshape ( - 1 , 1 ) y = [ 2 , 1 , 4 , 3 , 5 ] lm = linear_model . LinearRegression () lm . fit ( x , y ) print ( f 'a = {lm.intercept_}' ) print ( f 'b = {lm.coef_[0]}' ) print ( \"Where Y=a+b*X\" ) a = 0.5999999999999996 b = 0.8000000000000002 Where Y=a+b*X if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Machine Learning","url":"redoules.github.io/machine-learning/LeastSquareRegressionLine.html","loc":"redoules.github.io/machine-learning/LeastSquareRegressionLine.html"},{"title":"Day 8 - Least Square Regression Line","text":"Least Square Regression Line Problem A group of five students enrolls in Statistics immediately after taking a Math aptitude test. Each student's Math aptitude test score, \\(x\\) , and Statistics course grade, \\(y\\) , can be expressed as the following list \\((x,y)\\) of points: \\((95, 85)\\) \\((85, 95)\\) \\((80, 70)\\) \\((70, 65)\\) \\((60, 70)\\) If a student scored an 80 on the Math aptitude test, what grade would we expect them to achieve in Statistics? Determine the equation of the best-fit line using the least squares method, then compute and print the value of \\(y\\) when \\(x=80\\) . X = [ 95 , 85 , 80 , 70 , 60 ] Y = [ 85 , 95 , 70 , 65 , 70 ] n = len ( X ) def cov ( X , Y , n ): x_mean = 1 / n * sum ( X ) y_mean = 1 / n * sum ( Y ) return 1 / n * sum ([( X [ i ] - x_mean ) * ( Y [ i ] - y_mean ) for i in range ( n )]) def stdv ( X , mu_x , n ): return ( sum ([( x - mu_x ) ** 2 for x in X ]) / n ) ** 0.5 def pearson_1 ( X , Y , n ): std_x = stdv ( X , 1 / n * sum ( X ), n ) std_y = stdv ( Y , 1 / n * sum ( Y ), n ) return cov ( X , Y , n ) / ( std_x * std_y ) b = pearson_1 ( X , Y , n ) * stdv ( Y , sum ( Y ) / n , n ) / stdv ( X , sum ( X ) / n , n ) a = sum ( Y ) / n - b * sum ( X ) / n print ( f \"If a student scored 80 on the math test, he would most likely score a {round(a+80*b,3)} in statistics\" ) If a student scored 80 on the math test, he would most likely score a 78.288 in statistics Pearson correlation coefficient Problem The regression line of \\(y\\) on \\(x\\) is \\(3x+4y+8=0\\) , and the regression line of \\(x\\) on \\(y\\) is \\(4x+3y+7=0\\) . What is the value of the Pearson correlation coefficient? Mathematical explanation The initial equation system is : $$ \\left\\{\\begin{array}{ r @{{}={}} r >{{}}c<{{}} r >{{}}c<{{}} r } 3x+4y+8=0 & (1)\\\\ 4x+3y+7=0 & (2)\\\\ \\end{array} \\right. $$ So we can rewrite the 2 lines this way : $$ \\left\\{\\begin{array}{ r @{{}={}} r >{{}}c<{{}} r >{{}}c<{{}} r } y=-2+(\\frac{-3}{4})x & (1)\\\\ x=-\\frac{7}{4}+(-\\frac{3}{4})y & (2)\\\\ \\end{array} \\right. $$ so \\(b_1=-\\frac{3}{4}\\) and \\(b_2=-\\frac{3}{4}\\) When we apply the Pearson's coefficient formula : let \\(p\\) be the pearson coefficient let \\(\\sigma_X\\) be the standard deviation of \\(x\\) let \\(\\sigma_Y\\) be the standard deviation of \\(y\\) We hence have $$ \\left\\{\\begin{array}{ r @{{}={}} r >{{}}c<{{}} r >{{}}c<{{}} r } p=b_1\\left(\\frac{\\sigma_X}{\\sigma_Y}\\right) & (1)\\\\ p=b_2\\left(\\frac{\\sigma_Y}{\\sigma_X}\\right) & (2)\\\\ \\end{array} \\right. $$ by multiplying theses 2 equations together we get $$p&#94;2=b_1\\cdot b_2$$ $$p&#94;2=\\left(-\\frac{3}{4}\\right)\\left(-\\frac{3}{4}\\right)$$ $$p&#94;2=\\left(-\\frac{9}{16}\\right)$$ finally we get \\(p=\\left(-\\frac{3}{4}\\right)\\) or \\(p=\\left(\\frac{3}{4}\\right)\\) Since \\(X\\) and \\(Y\\) are negatively correlated we have \\(p=\\left(-\\frac{3}{4}\\right)\\) if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day8.html","loc":"redoules.github.io/blog/Statistics_10days-day8.html"},{"title":"Spearman's Rank Correlation Coefficient","text":"A rank correlation is any of several statistics that measure an ordinal association—the relationship between rankings of different ordinal variables or different rankings of the same variable, where a \"ranking\" is the assignment of the ordering labels \"first\", \"second\", \"third\", etc. to different observations of a particular variable. A rank correlation coefficient measures the degree of similarity between two rankings, and can be used to assess the significance of the relation between them. We have two random variables \\(X\\) and \\(Y\\) : \\(X=\\{x_i, x_2, x_3, ..., x_n\\}\\) \\(Y=\\{y_i, y_2, y_3, ..., y_n\\}\\) if \\(Rank_X\\) and \\(Rank_Y\\) denote the respective ranks of each data point, then the Spearman's rank correlation coefficient, \\(r_s\\) , is the Pearson correlation coefficient of \\(Rank_X\\) and \\(Rank_Y\\) . What does it means? The Spearman's rank correlation coefficientis is a nonparametric measure of rank correlation (statistical dependence between the rankings of two variables). It assesses how well the relationship between two variables can be described using a monotonic function. The Spearman correlation between two variables is equal to the Pearson correlation between the rank values of those two variables; while Pearson's correlation assesses linear relationships, Spearman's correlation assesses monotonic relationships (whether linear or not). If there are no repeated data values, a perfect Spearman correlation of +1 or −1 occurs when each of the variables is a perfect monotone function of the other. A Spearman correlation of 1 results when the two variables being compared are monotonically related, even if their relationship is not linear. This means that all data-points with greater x-values than that of a given data-point will have greater y-values as well. In contrast, this does not give a perfect Pearson correlation. Example \\(X=\\{0.2, 1.3, 0.2, 1.1, 1.4, 1.5\\}\\) \\(Y=\\{1.9, 2.2, 3.1, 1.2, 2.2, 2.2\\}\\) $$ Rank_X \\quad \\begin{bmatrix} X: & 0.2 & 1.3 & 0.2 & 1.1 & 1.4 & 1.5 \\\\ Rank: & 1 & 3 & 1 & 2 & 4 & 5 \\end{bmatrix} \\quad $$ so, \\(Rank_X = \\{1, 3, 1, 2, 4, 5\\}\\) similarly, \\(Rank_Y=\\{2,3,4,1,3,3\\}\\) \\(r_s\\) equals the Pearson correlation coefficient of \\(Rank_X\\) and \\(Rank_Y\\) , meaning that \\(r=0.158114\\) Special case : \\(X\\) and \\(Y\\) don't contain duplicates $$r_s=1-\\frac{6\\sum d_i&#94;2}{n(n&#94;2-1)}$$ Where, \\(d_i\\) is the difference between the respective values of \\(Rank_X\\) and \\(Rank_Y\\) . if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/spearman.html","loc":"redoules.github.io/mathematics/spearman.html"},{"title":"Pearson correlation coefficient","text":"Covariance This is a measure of how two random variables change together, or the strength of their correlation. Consider two random variables, \\(X\\) and \\(Y\\) , each with \\(n\\) values (i.e., \\(x_1\\) , \\(x_2\\) , \\(...\\) , \\(x_n\\) and \\(y_1\\) , \\(y_2\\) , \\(...\\) , \\(y_n\\) ). The covariance of \\(X\\) and \\(Y\\) can be found using either of the following equivalent formulas: $$cov(X,Y)=\\frac{1}{n}\\sum_{i=1}&#94;{n}(x_i-\\bar{x})\\cdot(y_i-\\bar{y})$$ or $$cov(X,Y)=\\frac{1}{n&#94;2}\\sum_{i=1}&#94;{n}\\sum_{j=1}&#94;{n}\\frac{1}{2}(x_i-x_j)\\cdot(y_i-y_j))$$ $$cov(X,Y)=\\frac{1}{n&#94;2}\\sum_{i}\\sum_{j\\gt i}&#94;{n}(x_i-x_j)\\cdot(y_i-y_j)$$ where, \\(\\bar{x}\\) is the mean of \\(X\\) (or \\(\\mu_X\\) ) and \\(\\bar{y}\\) is the mean of \\(Y\\) (or \\(\\mu_Y\\) ) Pearson correlation coefficient The pearson correlation coefficient, \\(\\rho_{X,Y}\\) , is given by : $$\\rho_{X,Y}=\\frac{cov(X,Y)}{\\sigma_X\\sigma_Y}=\\frac{\\sum_{i}(x_i-\\bar{x})(y_i-\\bar{y})}{n\\sigma_X\\sigma_Y}$$ Here, \\(\\sigma_X\\) is the standard deviation of \\(X\\) and \\(\\sigma_Y\\) is the standard deviation of \\(Y\\) . You may also see \\(\\rho_{X,Y}\\) written as \\(r_{X,Y}\\) . The pearson correlation coefficient is a measure of the linear correlation between two variables X and Y. if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/pearson.html","loc":"redoules.github.io/mathematics/pearson.html"},{"title":"Day 7 - Pearson and spearman correlations","text":"Pearson correlation coefficient Problem Given two n-element data sets, \\(X\\) and \\(Y\\) , calculate the value of the Pearson correlation coefficient. Python implementation Using the formula $$\\rho_{X,Y}=\\frac{cov(X,Y)}{\\sigma_X\\sigma_Y}$$ where $$cov(X,Y)=\\frac{1}{n}\\sum_{i=1}&#94;{n}(x_i-\\bar{x})\\cdot(y_i-\\bar{y})$$ n = 10 X = [ 10 , 9.8 , 8 , 7.8 , 7.7 , 7 , 6 , 5 , 4 , 2 ] Y = [ 200 , 44 , 32 , 24 , 22 , 17 , 15 , 12 , 8 , 4 ] def cov ( X , Y , n ): x_mean = 1 / n * sum ( X ) y_mean = 1 / n * sum ( Y ) return 1 / n * sum ([( X [ i ] - x_mean ) * ( Y [ i ] - y_mean ) for i in range ( n )]) def stdv ( X , mu_x , n ): return ( sum ([( x - mu_x ) ** 2 for x in X ]) / n ) ** 0.5 def pearson_1 ( X , Y , n ): std_x = stdv ( X , 1 / n * sum ( X ), n ) std_y = stdv ( Y , 1 / n * sum ( Y ), n ) return cov ( X , Y , n ) / ( std_x * std_y ) pearson_1 ( X , Y , n ) 0.6124721937208479 Python implementation Using the formula $$\\rho_{X,Y}=\\frac{\\sum_{i}(x_i-\\bar{x})(y_i-\\bar{y})}{n\\sigma_X\\sigma_Y}$$ def pearson_2 ( X , Y , n ): std_x = stdv ( X , 1 / n * sum ( X ), n ) std_y = stdv ( Y , 1 / n * sum ( Y ), n ) x_mean = 1 / n * sum ( X ) y_mean = 1 / n * sum ( Y ) return sum ([( X [ i ] - x_mean ) * ( Y [ i ] - y_mean ) for i in range ( n )]) / ( n * std_x * std_y ) pearson_2 ( X , Y , n ) 0.6124721937208479 Spearman's rank correlation coefficient Problem Given two \\(n\\) -element data sets, \\(X\\) and \\(Y\\) , calculate the value of Spearman's rank correlation coefficient. Python implementation We knwo that in this case, the values in each dataset are unique. Hence we can use the formula : $$r_s=1-\\frac{6\\sum d_i&#94;2}{n(n&#94;2-1)}$$ n = 10 X = [ 10 , 9.8 , 8 , 7.8 , 7.7 , 1.7 , 6 , 5 , 1.4 , 2 ] Y = [ 200 , 44 , 32 , 24 , 22 , 17 , 15 , 12 , 8 , 4 ] def spearman_rank ( X , Y , n ): rank_X = [ sorted ( X ) . index ( v ) + 1 for v in X ] rank_Y = [ sorted ( Y ) . index ( v ) + 1 for v in Y ] d = [( rank_X [ i ] - rank_Y [ i ]) ** 2 for i in range ( n )] return 1 - ( 6 * sum ( d )) / ( n * ( n * n - 1 )) spearman_rank ( X , Y , n ) 0.9030303030303031 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day7.html","loc":"redoules.github.io/blog/Statistics_10days-day7.html"},{"title":"Day 6 - The Central Limit Theorem","text":"Problem 1 A large elevator can transport a maximum of \\(9800\\) kg. Suppose a load of cargo containing \\(49\\) boxes must be transported via the elevator. The box weight of this type of cargo follows a distribution with a mean of \\(\\mu=205\\) kg and a standard deviation of \\(\\sigma=15\\) kg. Based on this information, what is the probability that all boxes can be safely loaded into the freight elevator and transported? Mathematical explanation This problem can be tackled with the central limit theorem. Since the number of boxes is \"large\", the sum of the weight approaches normal distribution with : \\(\\mu' = n\\mu\\) \\(\\sigma'=\\sigma\\sqrt{n}\\) If we want to know the probability of the sum of the mass of all boxes to be under a certain weight, we can compute the cumulative density function : $$P(x<9800) = F_X(9800)$$ max_load = 9800 n = 49 mu = 205 st_dev = 15 import math def cumulative ( x , mean , sd ): return 0.5 * ( 1 + math . erf (( x - mean ) / ( sd * math . sqrt ( 2 )))) mu_group = n * mu st_dev_group = st_dev * math . sqrt ( n ) print ( f \"Probability that all the boxes can be lifted by the elevator : {cumulative(max_load, mu_group, st_dev_group)}\" ) Probability that all the boxes can be lifted by the elevator : 0.009815328628645315 Problem 2 The number of tickets purchased by each student for the University X vs. University Y football game follows a distribution that has a mean of \\(\\mu=2.4\\) and a standard deviation of \\(\\sigma=2.0\\) . A few hours before the game starts, \\(100\\) eager students line up to purchase last-minute tickets. If there are only \\(250\\) tickets left, what is the probability that all students will be able to purchase tickets? Mathematical explanation We want to know if the sum of all the purchases will exceed the total supply of tickets. Since each student buy follows a Normal distribution and that the number of students is relatively high, the probability that all the students will be able to buy a ticket can be computed by applying the central limit theorem. The total number of tickets bought follows a normal distribution of mean \\(\\mu'=n*\\mu\\) and of standard deviation \\(\\sigma'=\\sigma\\sqrt{n}\\) ticket_supply = 250 n_students = 100 mu = 2.4 st_dev = 2 mu_group = n_students * mu st_dev_group = st_dev * math . sqrt ( n_students ) print ( f \"Probability that all the students can purchase tickets : {cumulative(ticket_supply, mu_group, st_dev_group)}\" ) Probability that all the students can purchase tickets : 0.691462461274013 Problem 3 You have a sample of \\(100\\) values from a population with mean \\(\\mu=500\\) and with standard deviation \\(\\sigma=80\\) . Compute the interval that covers the middle \\(95%\\) of the distribution of the sample mean; in other words, compute \\(A\\) and \\(B\\) such that \\(P(A<x<B)=0.95\\) . Use the value of \\(z=1.96\\) . Note that is the z-score. Mathematical explanation The margin of error can be computed with : $$MoE = \\frac{z-\\sigma}{\\sqrt{n}}$$ Knowing this, we can figure out what values for x paint the exact middle of the distribution for 0.95 probability, that means theres a 0.0025 leftover on both sides to make the total of 1. zScore = 1.96 std = 80 n = 100 mean = 500 marginOfError = zScore * std / math . sqrt ( n ); print ( \"A =\" , mean - marginOfError ) print ( \"B =\" , mean + marginOfError ) A = 484.32 B = 515.68 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day6.html","loc":"redoules.github.io/blog/Statistics_10days-day6.html"},{"title":"Day 5 - Poisson and Normal distributions","text":"Poisson Distribution Problem 1 A random variable, \\(X\\) , follows Poisson distribution with mean of 2.5. Find the probability with which the random variable \\(X\\) is equal to 5. Mathematical explanation In this case, the answer is straightforward, we just need to compute the value of the Poisson distribution of mean 2.5 at 5: $$P(\\lambda = 2.5, x=5)=\\frac{\\lambda&#94;ke&#94;{-\\lambda}}{k!}$$ $$P(\\lambda = 2.5, x=5)=\\frac{2.5&#94;5e&#94;{-2.5}}{5!}$$ def factorial ( k ): return 1 if k == 1 else k * factorial ( k - 1 ) from math import exp def poisson ( l , k ): return ( l ** k * exp ( - l )) / factorial ( k ) l = 2.5 k = 5 print ( f 'Probability that a random variable X following a Poisson distribution of mean {l} equals {k} : {round(poisson(l,k),3)}' ) Probability that a random variable X following a Poisson distribution of mean 2.5 equals 5 : 0.067 Problem 2 The manager of a industrial plant is planning to buy a machine of either type \\(A\\) or type \\(B\\) . For each day's operation: The number of repairs, \\(X\\) , that machine \\(A\\) needs is a Poisson random variable with mean 0.88. The daily cost of operating \\(A\\) is \\(C_A=160+40X&#94;2\\) . The number of repairs, \\(Y\\) , that machine \\(B\\) needs is a Poisson random variable with mean 1.55. The daily cost of operating \\(B\\) is \\(C_B=128+40Y&#94;2\\) . Assume that the repairs take a negligible amount of time and the machines are maintained nightly to ensure that they operate like new at the start of each day. What is the expected daily cost for each machine. Mathematical explanation The cost for each machine follows a law that is the square of a Poisson distribution. $$C_Z = a + b*Z&#94;2$$ Since the expectation is a linear operator : $$E[C_Z] = aE[1] + bE[Z&#94;2]$$ Knowing that \\(Z\\) follows a Poisson distribution of mean \\(\\lambda\\) we have : $$E[C_Z] = a+ b(\\lambda + \\lambda&#94;2)$$ averageX = 0.88 averageY = 1.55 CostX = 160 + 40 * ( averageX + averageX ** 2 ) CostY = 128 + 40 * ( averageY + averageY ** 2 ) print ( f 'Expected cost to run machine A : {round(CostX, 3)}' ) print ( f 'Expected cost to run machine A : {round(CostY, 3)}' ) Expected cost to run machine A : 226.176 Expected cost to run machine A : 286.1 Normal Distribution Problem 1 In a certain plant, the time taken to assemble a car is a random variable, \\(X\\) , having a normal distribution with a mean of 20 hours and a standard deviation of 2 hours. What is the probability that a car can be assembled at this plant in: Less than 19.5 hours? Between 20 and 22 hours? Mathematical explanation \\(X\\) is a real-valued random variable following a normal distribution : the probability of assembly the car in less than 19.5 hours is the cumulative distribution function of X evaluated at 19.5: $$P(X\\leq 19.5)=F_X(19.5)$$ For a normal distribution, the cumulative distribution function is : $$\\Phi(x)=\\frac{1}{2}\\left(1+erf\\left(\\frac{x-\\mu}{\\sigma\\sqrt{2}}\\right)\\right)$$ import math def cumulative ( x , mean , sd ): return 0.5 * ( 1 + math . erf (( x - mean ) / ( sd * math . sqrt ( 2 )))) mean = 20 sd = 2 print ( f 'Probability that the car is built in less than 19.5 hours : {round(cumulative(19.5,mean,sd),3)}' ) Probability that the car is built in less than 19.5 hours : 0.401 Similarly, the probability that a car is built between 20 and 22hours can be computed thanks to the cumulative density function: $$P(20\\leq x\\leq 22) = F_X(22)-F_X(20)$$ print ( f 'Probability that the car is built between 20 and 22 hours : {round(cumulative(22,mean,sd)-cumulative(20,mean,sd),3)}' ) Probability that the car is built between 20 and 22 hours : 0.341 Problem 2 The final grades for a Physics exam taken by a large group of students have a mean of \\(\\mu=70\\) and a standard deviation of \\(\\sigma=10\\) . If we can approximate the distribution of these grades by a normal distribution, what percentage of the students: Scored higher than 80 (i.e., have a \\(grade \\gt 80\\) ))? Passed the test (i.e., have a \\(grade \\gt 60\\) )? * Failed the test (i.e., have a \\(grade \\lt 60\\) )? Mathematical explanation Here again, we need to appy the cumulative density function to get the probabilities : Probability that they scored higher than 80 : $$P(X\\gt80) = 1- P(X\\lt80)$$ $$P(X\\gt80) = 1- F_X(80)$$ mean = 70 sd = 10 print ( f 'Probability that the the student scored higher than 80 : {round(1- cumulative(80,mean,sd),3)}' ) Probability that the the student scored higher than 80 : 0.159 Probability that they passed the test : $$P(X\\gt60) = 1- P(X\\lt60)$$ $$P(X\\gt80) = 1- F_X(60)$$ print ( f 'Probability that the the student passed the test : {round(1- cumulative(60,mean,sd),3)}' ) Probability that the the student passed the test : 0.841 Probability that they failed : $$P(X\\lt60) = F_X(60)$$ print ( f 'Probability that the student failed the test: {round(cumulative(60,mean,sd),3)}' ) Probability that the student failed the test: 0.159 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day5.html","loc":"redoules.github.io/blog/Statistics_10days-day5.html"},{"title":"Normal Distribution","text":"Normal Distribution The probability density of normal distribution is: $$\\mathcal{N}(\\mu,\\sigma&#94;2)=\\frac{1}{\\sigma\\sqrt{2\\pi}}e&#94;{-\\frac{(x-\\mu)&#94;2}{2\\sigma&#94;2}}$$ where, \\(\\mu\\) is the mean (or expectation) of the distribution. It is also equal to median and mode of the distribution. \\(\\sigma&#94;2\\) is the variance. * \\(\\sigma\\) is the standard deviation. Standard Normal Distribution If \\(\\mu=0\\) and \\(\\sigma=1\\) , then the normal distribution is known as standard normal distribution: $$\\phi(x)=\\frac{e&#94;{-\\frac{x&#94;2}{2}}}{\\sigma\\sqrt{2\\pi}}$$ Every normal distribution can be represented as standard normal distribution: $$\\mathcal{N}(\\mu,\\sigma&#94;2)=\\frac{1}{\\sigma}\\phi(\\frac{x-\\mu}{\\sigma})$$ Cumulative Probability Consider a real-valued random variable, \\(X\\) . The cumulative distribution function of \\(X\\) (or just the distribution function of \\(X\\) ) evaluated at \\(x\\) is the probability that \\(X\\) will take a value less than or equal to \\(x\\) : $$F_X(x)=P(X\\leq x)$$ also, $$P(a\\leq X\\leq b)=P(a\\lt X\\lt b)=F_X(b)-F_X(a)$$ the cumulative distribution function for a function with normal distribution is: $$\\Phi(x)=\\frac{1}{2}\\left(1+erf\\left(\\frac{x-\\mu}{\\sigma\\sqrt{2}}\\right)\\right)$$ where \\(erf\\) is the error function: $$erf(z)=\\frac{2}{\\sqrt{\\pi}}\\int_0&#94;ze&#94;{-x&#94;2}dx$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/normal.html","loc":"redoules.github.io/mathematics/normal.html"},{"title":"Poisson Distribution","text":"Poisson Experiment Poisson experiment is a statistical experiment that has the following properties: The outcome of each trial is either success or failure. The average number of successes ( \\(\\lambda\\) ) that occurs in a specified region is known. The probability that a success will occur is proportional to the size of the region. The probability that a success will occur in an extremely small region is virtually zero. Poisson Distribution A Poisson random variable is the number of successes that result from a Poisson experiment. The probability distribution of a Poisson random variable is called a Poisson distribution: $$P(k,\\lambda)=\\frac{\\lambda&#94;ke&#94;{-\\lambda}}{k!}$$ where : \\(\\lambda\\) is the average number of successes that occur in a specified region. \\(k\\) is the actual number of successes that occur in a specified region. * \\(P(k,\\lambda)\\) is the Poisson probability, which is the probability of getting exactly \\(k\\) successes when the average number of successes is \\(\\lambda\\) . Example The average number of goals in the soccer world cup is 2.5. The probability that 4 goals are scored is then: $$p(\\lambda=2.5,k=4)=\\frac{2.5&#94;4e&#94;{-2.5}}{4!}=0.133$$ Expectation for the Poisson distribution Consider some Poisson random variable, \\(X\\) . Let \\(E[X]\\) be the expectation of \\(X\\) . Find the value of \\(E[X&#94;2]\\) . Let \\(Var(X)\\) be the variance of \\(X\\) . Recall that if a random variable has a Poisson distribution, then: \\(E[X]=\\lambda\\) \\(Var[X]=\\lambda\\) Now, we'll use the following property of expectation and variance for any random variable, \\(X\\) : $$Var(X)=E[X&#94;2]-(E[X])&#94;2$$ $$E[X&#94;2]=Var(X)+(E[X])&#94;2$$ So, for any random variable having a Poisson distribution, the above result can be rewritten as: $$E[X&#94;2]=\\lambda + \\lambda&#94;2$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/poisson.html","loc":"redoules.github.io/mathematics/poisson.html"},{"title":"Geometric distribution","text":"Negative Binomial Experiment A negative binomial experiment is a statistical experiment that has the following properties: The experiment consists of n repeated trials. The trials are independent. The outcome of each trial is either success (s) or failure (f). \\(P(s)\\) is the same for every trial. The experiment continues until x successes are observed If \\(X\\) is the number of experiments until the \\(x&#94;{th}\\) success occures, then \\(X\\) is a discrete random variable called a negative binomial Negative Binomial Distribution Consider the following probability mass function: $$b&#94;*(x,n,p) = {\\binom{n-1}{x-1}}p&#94;xq&#94;{n-x}$$ The function above is negative binomial and has the following properties: The number of successes to be observed is \\(x\\) . The total number of trials is \\(n\\) . The probability of success of 1 trial is \\(p\\) . The probability of failure of 1 trial \\(q\\) , where \\(q=1-p\\) . \\(b&#94;*(x,n,p)\\) is the negative binomial probability , meaning the probability of having exactly \\(x-1\\) successes out of \\(n-1\\) trials and having \\(x\\) successes after \\(n\\) trials. Geometric Distribution The geometric distribution is a special case of the negative binomial distribution that deals with the number of Bernoulli trials required to get a success (i.e., counting the number of failures before the first success). Recall that \\(X\\) is the number of successes in \\(n\\) independent Bernoulli trials, so for each \\(i\\) (where $1\\leq i\\leq n): $ X_i = \\begin{cases} 1 if the i&#94;{th} trial is a success \\\\ 0 otherwise x \\end{cases} $ The geometric distribution is a negative binomial distribution where the number of successes is 1. We express this with the following formula: $$g(n,p)=q&#94;{n-1}p$$ Example Bob is a high school basketball player. He is a 70% free throw shooter, meaning his probability of making a free throw is 0.7. What is the probability that Bob makes his first free throw on his fifth shot? For this experiment n=5, p=0.7 and q=0.3 So : $$g(n=5, p=0.7)=0.3&#94;4 0.7=0.00567$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/Geometric.html","loc":"redoules.github.io/mathematics/Geometric.html"},{"title":"Binomial distribution","text":"Binomial Experiment A binomial experiment (or Bernoulli trial) is a statistical experiment that has the following properties: The experiment consists of n repeated trials. The trials are independent. The outcome of each trial is either success (s) or failure (f). Binomial Distribution We define a binomial process to be a binomial experiment meeting the following conditions: The number of successes is \\(x\\) . The total number of trials is \\(n\\) . The probability of success of 1 trial is \\(p\\) . The probability of failure of 1 trial \\(q\\) , where \\(q=1-p\\) . \\(b(x,n,p)\\) is the binomial probability , meaning the probability of having exactly \\(x\\) successes out of \\(n\\) trials. The binomial random variable is the number of successes, \\(x\\) , out of \\(n\\) trials. The binomial distribution is the probability distribution for the binomial random variable, given by the following probability mass function: $$b(x,n,p) = \\frac{n!}{x!(n-x)!}p&#94;xq&#94;{n-x}$$ Python code for the Binomial distribution import math def bi_dist ( x , n , p ): b = ( math . factorial ( n ) / ( math . factorial ( x ) * math . factorial ( n - x ))) * ( p ** x ) * (( 1 - p ) ** ( n - x )) return ( b ) Using numpy import numpy as np n = 10 #number of coin toss p = 0.5 #probability samples = 1000 #number of samples s = np . random . binomial ( n , p , samples ) Using the stats module from scipy from scipy.stats import binom n = 10 #number of coin toss p = 0.5 #probability samples = 1000 #number of samples s = binom . rvs ( n , p , size = samples ) Cumulative probabilities A cumulative probability refers to the probability that the value of a random variable falls within a specified range. Frequently, cumulative probabilities refer to the probability that a random variable is less than or equal to a specified value. A fair coin is tossed 10 times. Probability of getting 5 heads The probability of getting heads is: $$b(x=5, n=10, p=0.5)=0.246$$ Probability of getting at least 5 heads The probability of getting at least heads is: $$b(x\\geq 5, n=10, p=0.5)= \\sum_{r=5}&#94;{10} b(x=r, n=10, p=0.5)$$ $$b(x\\geq 5, n=10, p=0.5)= 0.623$$ Probability of getting at most 5 heads The probability of getting at most heads is: $$b(x\\leq 5, n=10, p=0.5)= \\sum_{r=0}&#94;{5} b(x=r, n=10, p=0.5)$$ $$b(x\\leq 5, n=10, p=0.5)= 0.623$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/Binomial.html","loc":"redoules.github.io/mathematics/Binomial.html"},{"title":"Day 4 - Binomial and geometric distributions","text":"Binomial distribution Problem 1 The ratio of boys to girls for babies born in Russia is \\(r=\\frac{N_b}{N_g}=1.09\\) . If there is 1 child born per birth, what proportion of Russian families with exactly 6 children will have at least 3 boys? Mathematical explanation Let's first compute the probability of having a boy : $$p_b=\\frac{N_b}{N_b+N_g}$$ where: \\(N_b\\) is the number of boys \\(N_g\\) is the number of girls * \\(r=\\frac{N_b}{N_g}\\) $$p_b=\\frac{1}{1+\\frac{1}{r}}$$ $$p_b=\\frac{r}{r+1}$$ r = 1.09 p_b = r / ( r + 1 ) print ( f \"The probability of having a boy is p={p_b:3f}\" ) The probability of having a boy is p=0.521531 The probability of getting 3 boys in 6 children is given by : $$b(x=3, n=6, p=p_b)$$ In order to compute the proportion of Russian families with exactly 6 children will have at 3 least boys we need to compute the cumulative probability distribution $$b(x\\geq 3, n=6, p=p_b) = \\sum_{i=3}&#94;{6} b(x\\geq i, n=6, p=p_b)$$ Let's code it ! import math def bi_dist ( x , n , p ): b = ( math . factorial ( n ) / ( math . factorial ( x ) * math . factorial ( n - x ))) * ( p ** x ) * (( 1 - p ) ** ( n - x )) return ( b ) b , p , n = 0 , p_b , 6 for i in range ( 3 , 7 ): b += bi_dist ( i , n , p ) print ( f \"probability of getting at least 3 boys in a family with exactly 6 children : {b:.3f}\" ) probability of getting at least 3 boys in a family with exactly 6 children : 0.696 Problem 2 A manufacturer of metal pistons finds that, 12% on average, of the pistons they manufacture are rejected because they are incorrectly sized. What is the probability that a batch of 10 pistons will contain: No more than 2 rejects? At least 2 rejects? Mathematical explanation On average 12% of the pistons are rejected, this means that a piston has a probability of \\(p_{rejected}=0.12\\) to be rejected. The probability of getting less than 2 faulty pistons in a batch is : $$p(rejet<2) = b(x\\leq 2, n= 10, p=p_{rejected})$$ $$p(rejet<2) = \\sum_{i=0}&#94;{2} b(x\\leq i, n=10, p=p_{rejected})$$ b , p , n = 0 , 12 / 100 , 10 for i in range ( 0 , 3 ): b += bi_dist ( i , n , p ) print ( f \"The probability of getting less than 2 faulty pistons in a batch is : {b:.3f}\" ) The probability of getting less than 2 faulty pistons in a batch is : 0.891 The probability that a batch of 10 pistons will contain at least 2 rejects : $$p(rejet<2) = b(x\\geq 2, n= 10, p=p_{rejected})$$ $$p(rejet<2) = \\sum_{i=2}&#94;{10} b(x\\geq i, n=10, p=p_{rejected})$$ b , p , n = 0 , 12 / 100 , 10 for i in range ( 2 , 11 ): b += bi_dist ( i , n , p ) print ( f \"The probability of getting at least 2 faulty pistons in a batch is : {b:.3f}\" ) The probability of getting at least 2 faulty pistons in a batch is : 0.342 Geometric distribution Problem 1 The probability that a machine produces a defective product is \\(\\frac{1}{3}\\) . What is the probability that the first defect is found during the fith inspection? Mathematical explanation In this case, we will use a geometric distribution to evaluate the probability : \\(n=5\\) \\(p=\\frac{1}{3}\\) Hence, the probability that the first defect is found during the fith inspection is \\(g(n=5,p=1/3)\\) print ( f \"The probability that the first defect is found during the fith inspection is {round(((1-p)**(n-1)) * p, 3)}\" ) The probability that the first defect is found during the fith inspection is 0.038 Problem 2 The probability that a machine produces a defective product is \\(\\frac{1}{3}\\) . What is the probability that the first defect is found during the first 5 inspections? Mathematical explanation In this problem, we need to compute the cumulative distribution function $$p(x \\leq5) = \\sum_{i=1}&#94;{5} g(n=i,p=1/3)$$ p_x5 = 0 p = 1 / 3 n = 5 for i in range ( 1 , n + 1 ): p_x5 += ( 1 - p ) ** ( i - 1 ) * p print ( f \"The probability that the first defect is found during the first 5 inspection is {round(p_x5, 3)}\" ) The probability that the first defect is found during the first 5 inspection is 0.868 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day4.html","loc":"redoules.github.io/blog/Statistics_10days-day4.html"},{"title":"Premutations and combinations","text":"Finding patterns in the possible ways events can occur is very useful in helping us count the number of desirable events in our sample space. Two of the easiest methods for doing this are with permutations (when order matters) and combinations (when order doesn't matter). Permutations An ordered arrangement r of objects from a set, A, of n objects (where \\(0\\lt r \\leq n\\) ) is called an r-element permutation of A. You can also think of this as a permutation of A's elements taken r at a time. The number of r-element permutations of an n-object set is denoted by the following formula: $$_{n}P_{r}=\\frac{n!}{(n-r)!}$$ Combinations An unordered arrangement of r objects from a set, A, of n objects (where \\(0\\lt r \\leq n\\) ) is called an r-element combination of A. You can also think of this as a combination of A's elements taken r at a time. Because the only difference between permutations and combinations is that combinations are unordered, we can easily find the number of r-element combinations by dividing out the permutations (r!): $$_{n}C_{r}=\\frac{_{n}P_{r}}{r!}=\\frac{n!}{r!(n-r)!}$$ When we talk about combinations, we're talking about the number of subsets of size r that can be made from a set of size n. In fact, \\(_{n}C_{r}\\) is often referred to as \"n choose r\", because it's counting the number of r-element combinations that can be chosen from a set of n elements. if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/Premutations_and_combinations.html","loc":"redoules.github.io/mathematics/Premutations_and_combinations.html"},{"title":"Conditional Probability","text":"This is defined as the probability of an event occurring, assuming that one or more other events have already occurred. Two events, A and B are considered to be independent if event A has no effect on the probability of event B (i.e. P(B|A)=P(A)). If events A and B are not independent, then we must consider the probability that both events occur. This can be referred to as the intersection of events A and B, defined as P(A∩B) = P(B|A)P(A). We can then use this definition to find the conditional probability by dividing the probability of the intersection of the two events (A∩B) by the probability of the event that is assumed to have already occurred (event A): $$ P(B|A)=\\frac{P(A\\cap B)}{P(A)}$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/cond_prob.html","loc":"redoules.github.io/mathematics/cond_prob.html"},{"title":"Bayes' Theorem","text":"Let A and B be two events such that P(A|B) denotes the probability of the occurrence of A given that B has occurred and denotes the probability of the occurrence B of given that A has occurred, then: $$ P(A|B)=\\frac{P(B|A)P(A)}{P(B)}$$ $$ P(A|B)=\\frac{P(B|A)P(A)}{P(B|A)P(A)+P(B|A&#94;c)P(A&#94;c)}$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/Bayes.html","loc":"redoules.github.io/mathematics/Bayes.html"},{"title":"Day 3 - Conditionnal probability","text":"Conditionnal probability Problem Suppose a family has 2 children, one of which is a boy. What is the probability that both children are boys? Mathematical explanation Let's look at the possible outcomes : .dataframe tbody tr th:only-of-type { vertical-align: middle; }</p> <div class=\"highlight\"><pre><span></span><span class=\"na\">.dataframe</span> <span class=\"no\">tbody</span> <span class=\"no\">tr</span> <span class=\"no\">th</span> <span class=\"err\">{</span> <span class=\"nl\">vertical-align:</span> <span class=\"nf\">top</span><span class=\"c\">;</span> <span class=\"err\">}</span> <span class=\"na\">.dataframe</span> <span class=\"no\">thead</span> <span class=\"no\">th</span> <span class=\"err\">{</span> <span class=\"nl\">text-align:</span> <span class=\"nf\">right</span><span class=\"c\">;</span> <span class=\"err\">}</span> </pre></div> <p> B G B BB BG G GB GG We know that at least one of the children is a boy, so only \"GG\" is not possible. The event where the family has a new boy is then \"BB\". Hence the probability is : $$\\frac{BB}{BB+GB+BG}=\\frac{1}{3}$$ Draw 2 cards from a deck Problem You 2 draw cards from a standard 52-card deck without replacing them. What is the probability that both cards are of the same suit? Mathematical explanation There are 13 cards of each suit. Draw one card. It can be anything with probability of 1. Now there are 51 cards left and 12 of them are the same suit as the first card you drew. So the chance the second card matches the 1st is \\(\\frac{12}{51}\\) . Drawing marbles Problem A bag contains 3 red marbles and 4 blue marbles. Then, 2 marbles are drawn from the bag, at random, without replacement. If the first marble drawn is red, what is the probability that the second marble is blue? Mathematical explanation On the first draw, the probabilities are the following : we call B the event \"a blue ball is drawn\" and R the event \"a red ball is drawn\" \\(P(B)=\\frac{4}{7}\\) \\(P(R)=\\frac{3}{7}\\) On the second draw, if a red ball has been drawn at first, the probabilities are : \\(P(B|R)=\\frac{4}{6}\\) \\(P(R|R)=\\frac{2}{6}\\) Hence, the probability of drawing a blue ball if the first ball drawn was red is \\(\\frac{1}{3}\\) if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day3.html","loc":"redoules.github.io/blog/Statistics_10days-day3.html"},{"title":"Day 2 - Probability, Compound Event Probability","text":"Basic probability with dices Problem In this challenge, we practice calculating probability. In a single toss of 2 fair (evenly-weighted) six-sided dice, find the probability that their sum will be at most 9. Mathematical explanation A nice way to think about sums-of-two-dice problems is to lay out the sums in a 6-by-6 grid in the obvious manner. .dataframe tbody tr th:only-of-type { vertical-align: middle; }</p> <div class=\"highlight\"><pre><span></span><span class=\"na\">.dataframe</span> <span class=\"no\">tbody</span> <span class=\"no\">tr</span> <span class=\"no\">th</span> <span class=\"err\">{</span> <span class=\"nl\">vertical-align:</span> <span class=\"nf\">top</span><span class=\"c\">;</span> <span class=\"err\">}</span> <span class=\"na\">.dataframe</span> <span class=\"no\">thead</span> <span class=\"no\">th</span> <span class=\"err\">{</span> <span class=\"nl\">text-align:</span> <span class=\"nf\">right</span><span class=\"c\">;</span> <span class=\"err\">}</span> </pre></div> <p> 1 2 3 4 5 6 1 2 3 4 5 6 7 2 3 4 5 6 7 8 3 4 5 6 7 8 9 4 5 6 7 8 9 10 5 6 7 8 9 10 11 6 7 8 9 10 11 12 We see that the identic values are on the same diagonal. The number of elements on the diagonal varies from 1 to 6 and then back to 1. let's call A < x the event : the sum all the 2 tosses is at most x. $$P(A\\leq9)=\\sum_{i=2}&#94;{9} P(A = i)$$ $$P(A\\leq9)=1-P(A\\gt9)$$ $$P(A\\leq9)=1-\\sum_{i=10}&#94;{12} P(A = i)$$ The value of \\(P(A = i) = \\frac{i-1}{36}\\) if \\(i \\leq 7\\) and \\(P(A = i) = \\frac{13-i}{36}\\) hence $$P(A\\leq9)=1-\\sum_{i=10}&#94;{12} \\frac{13-i}{36}$$ $$P(A\\leq9)= 1-\\frac{6}{36}$$ $$P(A\\leq9)= \\frac{5}{6}$$ Let's program it sum ([ 1 for d1 in range ( 1 , 7 ) for d2 in range ( 1 , 7 ) if d1 + d2 <= 9 ]) / 36 0.8333333333333334 More dices Problem In a single toss of 2 fair (evenly-weighted) six-sided dice, find the probability that the values rolled by each die will be different and the two dice have a sum of 6. Mathematical explanation Let's consider 2 events : A and B. A compound event is a combination of 2 or more simple events. If A and B are simple events, then A∪B denotes the occurence of either A or B. A∩B denotes the occurence of A and B together. We denote A the event \"the values of each dice is different\". The opposit event is A' \"the values of each dice is the same\". $$P(A) = 1-P(A')$$ $$P(A)=1-\\frac{6}{36}$$ $$P(A)=\\frac{5}{6}$$ We denote B the event \"the two dice have a sum of 6\", this probability has been computed on the first part of the article : $$P(B)=\\frac{5}{36}$$ The probability of having 2 dice different of sum 6 is : $$P(A|B) = 4/5$$ The probability that both A and B occure is equal to P(A∩B). Since \\(P(A|B)=\\frac{P(A∩B)}{P(B)}\\) $$P(A∩B)=P(B)*P(A|B)$$ $$P(A∩B)=5/36*4/5$$ $$P(A∩B)=1/9$$ Let's program it sum ([ 1 for d1 in range ( 1 , 7 ) for d2 in range ( 1 , 7 ) if ( d1 + d2 == 6 ) and ( d1 != d2 )]) / 36 0.1111111111111111 Compound Event Probability Problem There are 3 urns labeled X, Y, and Z. Urn X contains 4 red balls and 3 black balls. Urn Y contains 5 red balls and 4 black balls. Urn Z contains 4 red balls and 4 black balls. One ball is drawn from each of the urns. What is the probability that, of the 3 balls drawn, are 2 red and is 1 black? Mathematical explanation Let's write the different probabilities: .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } Red ball Black ball Urne X $$\\frac{4}{7}$$ $$\\frac{3}{7}$$ Urne Y $$\\frac{5}{9}$$ $$\\frac{4}{9}$$ Urne Z $$\\frac{1}{2}$$ $$\\frac{1}{2}$$ Addition rule A and B are said to be mutually exclusive or disjoint if they have no events in common (i.e., and A∩B=∅ and P(A∩B)=0. The probability of any of 2 or more events occurring is the union (∪) of events. Because disjoint probabilities have no common events, the probability of the union of disjoint events is the sum of the events' individual probabilities. A and B are said to be collectively exhaustive if their union covers all events in the sample space (i.e., A∪B=S and P(A∪B)=1). This brings us to our next fundamental rule of probability: if 2 events, A and B, are disjoint, then the probability of either event is the sum of the probabilities of the 2 events (i.e., P(A or B) = P(A)+P(B)) Mutliplication rule If the outcome of the first event (A) has no impact on the second event (B), then they are considered to be independent (e.g., tossing a fair coin). This brings us to the next fundamental rule of probability: the multiplication rule. It states that if two events, A and B, are independent, then the probability of both events is the product of the probabilities for each event (i.e., P(A and B)= P(A)xP(B)). The chance of all events occurring in a sequence of events is called the intersection (∩) of those events. The balls drawn from the urns are independant hence : p = P(2 red (R) and 1 back (B)) $$p = P(RRB) + P(RBR) + P(BRR)$$ Each of those 3 probability if equal to the product of the probability of drawing each ball \\(P(RRB) = P(R|X) * P(R|Y) * P(B|Z) = 4/7*5/9*1/2\\) \\(P(RRB) = 20/126\\) \\(P(RBR) = 16/126\\) \\(P(BRR) = 15/126\\) this leads to \\(p = 51/126\\) and finally $$p = \\frac{17}{42}$$ Let's program it X = 3 * [ \"B\" ] + 4 * [ \"R\" ] Y = 4 * [ \"B\" ] + 5 * [ \"R\" ] Z = 4 * [ \"B\" ] + 4 * [ \"R\" ] target = [ \"BRR\" , \"RRB\" , \"RBR\" ] sum ([ 1 for x in X for y in Y for z in Z if x + y + z in target ]) / sum ([ 1 for x in X for y in Y for z in Z ]) 0.40476190476190477 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day2.html","loc":"redoules.github.io/blog/Statistics_10days-day2.html"},{"title":"Day 1 - Quartiles, Interquartile Range and standard deviation","text":"Quartile Definition A quartile is a type of quantile. The first quartile (Q1) is defined as the middle number between the smallest number and the median of the data set. The second quartile (Q2) is the median of the data. The third quartile (Q3) is the middle value between the median and the highest value of the data set. Implementation in python without using the scientific libraries def median ( l ): l = sorted ( l ) if len ( l ) % 2 == 0 : return ( l [ len ( l ) // 2 ] + l [( len ( l ) // 2 - 1 )]) / 2 else : return l [ len ( l ) // 2 ] def quartiles ( l ): # check the input is not empty if not l : raise StatsError ( 'no data points passed' ) # 1. order the data set l = sorted ( l ) # 2. divide the data set in two halves mid = int ( len ( l ) / 2 ) Q2 = median ( l ) if ( len ( l ) % 2 == 0 ): # even Q1 = median ( l [: mid ]) Q3 = median ( l [ mid :]) else : # odd Q1 = median ( l [: mid ]) # same as even Q3 = median ( l [ mid + 1 :]) return ( Q1 , Q2 , Q3 ) L = [ 3 , 7 , 8 , 5 , 12 , 14 , 21 , 13 , 18 ] Q1 , Q2 , Q3 = quartiles ( L ) print ( f \"Sample : {L} \\n Q1 : {Q1}, Q2 : {Q2}, Q3 : {Q3}\" ) Sample : [3, 7, 8, 5, 12, 14, 21, 13, 18] Q1 : 6.0, Q2 : 12, Q3 : 16.0 Interquartile Range Definition The interquartile range of an array is the difference between its first (Q1) and third (Q3) quartiles. Hence the interquartile range is Q3-Q1 Implementation in python without using the scientific libraries print ( f \"Interquatile range : {Q3-Q1}\" ) Interquatile range : 10.0 Standard deviation Definition The standard deviation (σ) is a measure that is used to quantify the amount of variation or dispersion of a set of data values. A low standard deviation indicates that the data points tend to be close to the mean (also called the expected value) of the set, while a high standard deviation indicates that the data points are spread out over a wider range of values. The standard deviation can be computed with the formula: where µ is the mean : Implementation in python without using the scientific libraries import math X = [ 10 , 40 , 30 , 50 , 20 ] mean = sum ( X ) / len ( X ) X = [( x - mean ) ** 2 for x in X ] std = math . sqrt ( sum ( X ) / len ( X ) ) print ( f \"The distribution {X} has a standard deviation of {std}\" ) The distribution [400.0, 100.0, 0.0, 400.0, 100.0] has a standard deviation of 14.142135623730951","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day1.html","loc":"redoules.github.io/blog/Statistics_10days-day1.html"},{"title":"Counting values in an array","text":"Using lists If you want to count the number of occurences of an element in a list you can use the .count() function of the list object arr = [ 1 , 2 , 3 , 3 , 4 , 5 , 3 , 6 , 7 , 7 ] print ( f 'Array : {arr} \\n ' ) print ( f 'The number 3 appears {arr.count(3)} times in the list' ) print ( f 'The number 7 appears {arr.count(7)} times in the list' ) print ( f 'The number 4 appears {arr.count(4)} times in the list' ) Array : [1, 2, 3, 3, 4, 5, 3, 6, 7, 7] The number 3 appears 3 times in the list The number 7 appears 2 times in the list The number 4 appears 1 times in the list Using collections you can get a dictonnary of the number of occurences of each elements in a list thanks to the collections object like this import collections collections . Counter ( arr ) Counter({1: 1, 2: 1, 3: 3, 4: 1, 5: 1, 6: 1, 7: 2}) Using numpy You can have a simular result with numpy by hacking the unique function import numpy as np arr = np . array ( arr ) unique , counts = np . unique ( arr , return_counts = True ) dict ( zip ( unique , counts )) {1: 1, 2: 1, 3: 3, 4: 1, 5: 1, 6: 1, 7: 2}","tags":"Python","url":"redoules.github.io/python/counting.html","loc":"redoules.github.io/python/counting.html"},{"title":"Building a dictonnary using comprehension","text":"An easy way to create a dictionnary in python is to use the comprehension syntaxe. It can be more expressive hence easier to read. d = { key : value for ( key , value ) in iterable } In the example bellow we use the dictionnary comprehension to build a dictonnary from a source list. iterable = list ( range ( 10 )) d = { str ( value ): value ** 2 for value in iterable } # create a dictionnary linking the string value of a number with the square value of this number print ( d ) {'0': 0, '1': 1, '2': 4, '3': 9, '4': 16, '5': 25, '6': 36, '7': 49, '8': 64, '9': 81} of course, you can use an other iterable an repack it with the comprehension syntaxe. In the following example, we convert a list of tuples in a dictonnary. iterable = [( \"France\" , 67.12e6 ), ( \"UK\" , 66.02e6 ), ( \"USA\" , 325.7e6 ), ( \"China\" , 1386e6 ), ( \"Germany\" , 82.79e6 )] population = { key : value for ( key , value ) in iterable } print ( population ) {'France': 67120000.0, 'UK': 66020000.0, 'USA': 325700000.0, 'China': 1386000000.0, 'Germany': 82790000.0}","tags":"Python","url":"redoules.github.io/python/dict_comprehension.html","loc":"redoules.github.io/python/dict_comprehension.html"},{"title":"Extracting unique values from a list or an array","text":"Using lists An easy way to extract the unique values of a list in python is to convert the list to a set. A set is an unordered collection of items. Every element is unique (no duplicates) and must be immutable. my_list = [ 10 , 20 , 30 , 40 , 20 , 50 , 60 , 40 ] print ( f \"Original List : {my_list}\" ) my_set = set ( my_list ) my_new_list = list ( my_set ) # the set is converted back to a list with the list() function print ( f \"List of unique numbers : {my_new_list}\" ) Original List : [10, 20, 30, 40, 20, 50, 60, 40] List of unique numbers : [40, 10, 50, 20, 60, 30] Using numpy If you are using numpy you can extract the unique values of an array with the unique function builtin numpy: import numpy as np arr = np . array ( my_list ) print ( f 'Initial numpy array : {arr} \\n ' ) unique_arr = np . unique ( arr ) print ( f 'Numpy array with unique values : {unique_arr}' ) Initial numpy array : [10 20 30 40 20 50 60 40] Numpy array with unique values : [10 20 30 40 50 60]","tags":"Python","url":"redoules.github.io/python/unique.html","loc":"redoules.github.io/python/unique.html"},{"title":"Sorting an array","text":"Using lists Python provides an iterator to sort an array sorted() you can use it this way : import random # Random lists from [0-999] interval arr = [ random . randint ( 0 , 1000 ) for r in range ( 10 )] print ( f 'Initial random list : {arr} \\n ' ) reversed_arr = list ( sorted ( arr )) print ( f 'Sorted list : {reversed_arr}' ) Initial random list : [277, 347, 976, 367, 604, 878, 148, 670, 229, 432] Sorted list : [148, 229, 277, 347, 367, 432, 604, 670, 878, 976] it is also possible to use the sort function from the list object # Random lists from [0-999] interval arr = [ random . randint ( 0 , 1000 ) for r in range ( 10 )] print ( f 'Initial random list : {arr} \\n ' ) arr . sort () print ( f 'Sorted list : {arr}' ) Initial random list : [727, 759, 68, 103, 23, 90, 258, 737, 791, 567] Sorted list : [23, 68, 90, 103, 258, 567, 727, 737, 759, 791] Using numpy If you are using numpy you can sort an array by creating a view on the array: import numpy as np arr = np . random . random ( 5 ) print ( f 'Initial random array : {arr} \\n ' ) sorted_arr = np . sort ( arr ) print ( f 'Sorted array : {sorted_arr}' ) Initial random array : [0.40021786 0.13876208 0.19939047 0.46015169 0.43734158] Sorted array : [0.13876208 0.19939047 0.40021786 0.43734158 0.46015169]","tags":"Python","url":"redoules.github.io/python/sorting.html","loc":"redoules.github.io/python/sorting.html"},{"title":"Day 0 - Median, mean, mode and weighted mean","text":"A reminder The median The median is the value separating the higher half from the lower half of a data sample. For a data set, it may be thought of as the middle value. For a continuous probability distribution, the median is the value such that a number is equally likely to fall above or below it. The mean The arithmetic mean (or simply mean) of a sample is the sum of the sampled values divided by the number of items. The mode The mode of a set of data values is the value that appears most often. It is the value x at which its probability mass function takes its maximum value. In other words, it is the value that is most likely to be sampled. Implementation in python without using the scientific libraries def median ( l ): l = sorted ( l ) if len ( l ) % 2 == 0 : return ( l [ len ( l ) // 2 ] + l [( len ( l ) // 2 - 1 )]) / 2 else : return l [ len ( l ) // 2 ] def mean ( l ): return sum ( l ) / len ( l ) def mode ( data ): dico = { x : data . count ( x ) for x in list ( set ( data ))} return sorted ( sorted ( dico . items ()), key = lambda x : x [ 1 ], reverse = True )[ 0 ][ 0 ] L = [ 64630 , 11735 , 14216 , 99233 , 14470 , 4978 , 73429 , 38120 , 51135 , 67060 , 4978 , 73429 ] print ( f \"Sample : {L} \\n Mean : {mean(L)}, Median : {median(L)}, Mode : {mode(L)}\" ) Sample : [64630, 11735, 14216, 99233, 14470, 4978, 73429, 38120, 51135, 67060, 4978, 73429] Mean : 43117.75, Median : 44627.5, Mode : 4978 The weighted average The weighted arithmetic mean is similar to an ordinary arithmetic mean (the most common type of average), except that instead of each of the data points contributing equally to the final average, some data points contribute more than others. data = [ 10 , 40 , 30 , 50 , 20 ] weights = [ 1 , 2 , 3 , 4 , 5 ] sum_X = sum ([ x * w for x , w in zip ( data , weights )]) print ( round (( sum_X / sum ( weights )), 1 )) 32.0","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day0.html","loc":"redoules.github.io/blog/Statistics_10days-day0.html"},{"title":"Create a simple bash function","text":"A basic function The synthaxe to define a function is : #!/bin/bash # Basic function my_function () { echo Text displayed by my_function } #once defined, you can use it like so : my_function and it should return user@bash : ./my_function.sh Text displayed by my_function Function with arguments When used, the arguments are specified directly after the function name. Whithin the function they are accessible this the $ symbol followed by the number of the arguement. Hence $1 will take the value of the first arguement, $2 will take the value of the second arguement and so on. #!/bin/bash # Passing arguments to a function say_hello () { echo Hello $1 } say_hello Guillaume and it should return user@bash : ./function_arguements.sh Hello Guillaume Overriding Commands Using the previous example, let's override the echo function in order to make it say hello. To do so, you just need to name the function with the same name as the command you want to replace. When you are calling the original function, make sure you are using the builtin keyword #!/bin/bash # Overriding a function echo () { builtin echo Hello $1 } echo Guillaume user@bash : ./function_arguements.sh Hello Guillaume Returning values Use the keyword return to send back a value to the main program. The returned value will be stored in the $? variable #!/bin/bash # Retruning a value secret_number () { return 126 } secret_number echo The secret number is $? This code should return user@bash : ./retrun_value.sh The secret number is 126","tags":"Linux","url":"redoules.github.io/linux/simple_bash_function.html","loc":"redoules.github.io/linux/simple_bash_function.html"},{"title":"Number of edges in a Complete graph","text":"A complete graph contains \\(\\frac{n(n-1)}{2}\\) edges where \\(n\\) is the number of vertices (or nodes). if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/Number_edges_Complete_graph.html","loc":"redoules.github.io/mathematics/Number_edges_Complete_graph.html"},{"title":"Reverse an array","text":"Using lists Python provides an iterator to reverse an array reversed() you can use it this way : arr = list ( range ( 5 )) print ( f 'Initial array : {arr} \\n ' ) reversed_arr = list ( reversed ( arr )) print ( f 'Reversed array : {reversed_arr}' ) Initial array : [0, 1, 2, 3, 4] Reversed array : [4, 3, 2, 1, 0] Using numpy If you are using numpy you can reverse an array by creating a view on the array: import numpy as np arr = np . arange ( 5 ) print ( f 'Initial array : {arr} \\n ' ) reversed_arr = arr [:: - 1 ] print ( f 'Reversed array : {reversed_arr}' ) Initial array : [0 1 2 3 4] Reversed array : [4 3 2 1 0]","tags":"Python","url":"redoules.github.io/python/reverse.html","loc":"redoules.github.io/python/reverse.html"},{"title":"Advice for designing your own libraries","text":"Advice for designing your own libraries When designing your own library make sure to think of the following things. I will add new paragraphs to this article as I dicover new good practices. Use standard python objects Try to use standard python objects as much as possible. That way, your library becomes compatible with all the other python libaries. For instance, when I created SAMpy : a library for reading and writing SAMCEF results, it returned dictonnaries, lists and pandas dataframes. Hence the results extracted from SAMCEF where compatible with all the scientific stack of python. Limit the number of functionnalities Following the same logic as before, the objects should do only one thing but do it well. Indeed, having a simple interface will reduce the complexity of your code and make it easier to use your library. Again, with SAMpy, I decided to strictly limit the functionnalities to reading and writing SAMCEF files. Define an exception class for your library You should define your own exceptions in order to make it easier for your users to debug their code thanks to clearer messages that convey more meaning. That way, the user will know if the error comes from your library or something else. Bonus if you group similar exceptions in a hierachy of inerited Exception classes. Example : let's create a Exception related to the age of a person : def check_age ( age ): if age < 0 and age > 130 : raise ValueError If the user inputed an invalid age, the ValueError exception would be thrown. That's fine but imagine you wan't to provide more feedback to your users that don't know the internal of your library. Let's now create a selfexplanatory Exception class AgeInvalidError ( ValueError ): pass def check_age ( age ): if age < 0 and age > 130 : raise AgeInvalidError ( age ) You can also add some helpful text to guide your users along the way: class AgeInvalidError ( ValueError ): print ( \"Age invalid, must be between 0 and 130\" ) pass def check_age ( age ): if age < 0 and age > 130 : raise AgeInvalidError ( age ) If you want to group all the logically linked exceptions, you can create a base class and inherit from it : class BaseAgeInvalidError ( ValueError ): pass class TooYoungError ( BaseAgeInvalidError ): pass class TooOldError ( BaseAgeInvalidError ): pass def check_age ( age ): if age < 0 : raise TooYoungError ( age ) elif age > 130 : raise TooOldError ( age ) Structure your repository You should have a file structure in your repository. It will help other contributers especially future contributers. A nice directory structure for your project should look like this: README.md LICENSE setup.py requirements.txt ./MyPackage ./docs ./tests Some prefer to use reStructured Text, I personnaly prefer Markdown choosealicense.com will help you pick the license to use for your project. For package and distribution management, create a setup.py file a the root of the directory The list of dependencies required to test, build and generate the doc are listed in a pip requirement file placed a the root of the directory and named requirements.txt Put the documentation of your library in the docs directory. Put your tests in the tests directory. Since your tests will need to import your library, I recommend modifying the path to resolve your package property. In order to do so, you can create a context.py file located in the tests directory : import os import sys sys . path . insert ( 0 , os . path . abspath ( os . path . join ( os . path . dirname ( __file__ ), '..' ))) import MyPackage Then within your individual test files you can import your package like so : from .context import MyPackage Finally, your code will go into the MyPackage directory Test your code Once your library is in production, you have to guaranty some level of forward compatibility. Once your interface is defined, write some tests. In the future, when your code is modified, having those tests will make sure that the behaviour of your functions and objects won't be altered. Document your code Of course, you should have a documentation to go along with your library. Make sure to add a lot of commun examples as most users tend to learn from examples. I recommend writing your documentation using Sphinx.","tags":"Python","url":"redoules.github.io/python/design_own_libs.html","loc":"redoules.github.io/python/design_own_libs.html"},{"title":"Safely creating a folder if it doesn't exist","text":"Safely creating a folder if it doesn't exist When you are writing to files in python, if the file doesn't exist it will be created. However, if you are trying to write a file in a directory that doesn't exist, an exception will be returned FileNotFoundError : [ Errno 2 ] No such file or directory : \"directory\" This article will teach you how to make sure the target directory exists. If it doesn't, the function will create that directory. First, let's import os and make sure that the \"test_directory\" doesn't exist import os os . path . exists ( \". \\\\ test_directory\" ) False copy the ensure_dir function in your code. This function will handle the creation of the directory. Credit goes to Parand posted on StackOverflow def ensure_dir ( file_path ): directory = os . path . dirname ( file_path ) if not os . path . exists ( directory ): os . makedirs ( directory ) Let's now use the function and create a folder named \"test_directory\" ensure_dir ( \". \\\\ test_directory\" ) If we test for the existence of the directory, the exists function will now return True os . path . exists ( \". \\\\ test_directory\" ) True","tags":"Python","url":"redoules.github.io/python/ensure_dir.html","loc":"redoules.github.io/python/ensure_dir.html"},{"title":"List all files in a directory","text":"Listing all the files in a directory Let's start with the basics, the most staigthforward way to list all the files in a direcoty is to use a combinaison of the listdir function and isfile form os.path. You can use a list comprehension to store all the results in a list. mypath = \"./test_directory/\" from os import listdir from os.path import isfile , join [ f for f in listdir ( mypath ) if isfile ( join ( mypath , f ))] ['logfile.log', 'myfile.txt', 'super_music.mp3', 'textfile.txt'] Listing all the files of a certain type in a directory similarly, if you want to filter only a certain kind of file based on its extension you can use the endswith method. In the following example, we will filter all the \"txt\" files contained in the directory [ f for f in listdir ( mypath ) if f . endswith ( '.' + \"txt\" )] ['myfile.txt', 'textfile.txt'] Listing all the files matching a pattern in a directory The glob module finds all the pathnames matching a specified pattern according to the rules used by the Unix shell. You can use the *, ?, and character ranges expressed with [] wildcards import glob glob . glob ( \"*.txt\" ) ['myfile.txt'] Listing files recusively If you want to list all files recursively you can select all the sub-directories using the \"**\" wildcard import glob glob . glob ( mypath + '/**/*.txt' , recursive = True ) ['./test_directory\\\\myfile.txt', './test_directory\\\\textfile.txt', './test_directory\\\\subdir1\\\\file_hidden_in_a_sub_direcotry.txt'] Using a regular expression If you'd rather use a regular expression to select the files, the pathlib library provides the rglob function. from pathlib import Path list ( Path ( \"./test_directory/\" ) . rglob ( \"*.[tT][xX][tT]\" )) [WindowsPath('test_directory/myfile.txt'), WindowsPath('test_directory/textfile.txt'), WindowsPath('test_directory/subdir1/file_hidden_in_a_sub_direcotry.txt')] Using regular expressions you can for example select multiple types of files. In the following example, we list all the files that finish either with \"txt\" or with \"log\". list ( Path ( \"./test_directory/\" ) . rglob ( \"*.[tl][xo][tg]\" )) [WindowsPath('test_directory/logfile.log'), WindowsPath('test_directory/myfile.txt'), WindowsPath('test_directory/textfile.txt'), WindowsPath('test_directory/subdir1/file_hidden_in_a_sub_direcotry.txt')]","tags":"Python","url":"redoules.github.io/python/list_files_directory.html","loc":"redoules.github.io/python/list_files_directory.html"},{"title":"Using Dask on infiniband","text":"InfiniBand (abbreviated IB) is a computer-networking communications standard used in high-performance computing that features very high throughput and very low latency. It is used for data interconnect both among and within computers. InfiniBand is also used as either a direct or switched interconnect between servers and storage systems, as well as an interconnect between storage systems. (source Wikipedia). If you want to leverage this high speed network instead of the regular ethernet network, you have to specify to the scheduler that you want to used infiniband as your interface. Assuming that you Infiniband interface is ib0 , you would call the scheduler like this : dask-scheduler --interface ib0 --scheduler-file ./cluster.yaml you would have to call the worker using the same interface : dask-worker --interface ib0 --scheduler-file ./cluster.yaml","tags":"Python","url":"redoules.github.io/python/dask_infiniband.html","loc":"redoules.github.io/python/dask_infiniband.html"},{"title":"Clearing the current cell in the notebook","text":"In python, you can clear the output of a cell by importing the IPython.display module and using the clear_output function from IPython.display import clear_output print ( \"text to be cleared\" ) clear_output () As you can see, the text \"text to be cleared\" is not displayed because the function clear_output has been called afterward","tags":"Jupyter","url":"redoules.github.io/jupyter/clear_cell.html","loc":"redoules.github.io/jupyter/clear_cell.html"},{"title":"What's inside my .bashrc ?","text":"############ # Anaconda # ############ export PATH = \"/station/guillaume/anaconda3/bin: $PATH \" alias python = '/station/guillaume/anaconda3/bin/python' ######### # Alias # ######### alias ding = 'echo -e \"\\a\"' alias calc = 'python -ic \"from __future__ import division; from math import *\"' alias h = \"history|grep \" alias f = \"find . |grep \" alias p = \"ps aux |grep \" alias cdl = \"cd /data/guillaume\" alias cp = \"rsync -avz --progress\" alias grep = \"grep --color=auto\" alias ls = \"ls -hN --color=auto --group-directories-first\" alias ll = \"ls -hal\" alias sv = \"ssh compute_cluster\" alias ms = \"ls\" alias jl = \"jupyter lab\" alias lst = \"jupyter notebook list\" ########################## # bashrc personnalisation# ########################## force_color_prompt = yes export EDITOR = nano export BROWSER = \"firefox '%s' &\" if [ -n \" $force_color_prompt \" ] ; then if [ -x /usr/bin/tput ] && tput setaf 1 > & /dev/null ; then # We have color support; assume it's compliant with Ecma-48 # (ISO/IEC-6429). (Lack of such support is extremely rare, and such # a case would tend to support setf rather than setaf.) color_prompt = yes else color_prompt = fi fi if [ \" $color_prompt \" = yes ] ; then #\\h : hostname #\\u : user #\\w : current working directory #\\d : date #\\t : time yellow = 226 green = 83 pink = 198 blue = 34 PS1 = \"\\[\\033[38;5;22m\\]\\u\\[ $( tput sgr0 ) \\]\\[\\033[38;5;163m\\]@\\[ $( tput sgr0 ) \\]\\[\\033[38;5;22m\\]\\h\\[ $( tput sgr0 ) \\]\\[\\033[38;5;162m\\]:\\[ $( tput sgr0 ) \\]\\[\\033[38;5;172m\\]{\\[ $( tput sgr0 ) \\]\\[\\033[38;5;39m\\]\\w\\[ $( tput sgr0 ) \\]\\[\\033[38;5;172m\\]}\\[ $( tput sgr0 ) \\]\\[\\033[38;5;162m\\]>\\[ $( tput sgr0 ) \\]\"","tags":"Linux","url":"redoules.github.io/linux/bashrc.html","loc":"redoules.github.io/linux/bashrc.html"},{"title":"Efficient extraction of eigenvalues from a list of tensors","text":"When you manipulate FEM results you generally have either a: scalar field, vector field, * tensor field. With tensorial results, it is often useful to extract the eigenvalues in order to find the principal values. I have found that it is easier to store the components of the tensors in a 6 column pandas dataframe (because of the symmetric property of stress and strain tensors) import pandas as pd node = [ 1001 , 1002 , 1003 , 1004 ] #when dealing with FEM results you should remember at which element/node the result is computed (in the example, let's assume that we look at node from 1001 to 1004) tensor1 = [ 1 , 1 , 1 , 0 , 0 , 0 ] #eigen : 1 tensor2 = [ 4 , - 1 , 0 , 2 , 2 , 1 ] #eigen : 5.58443, -1.77931, -0.805118 tensor3 = [ 1 , 6 , 5 , 3 , 3 , 1 ] #eigen : 8.85036, 4.46542, -1.31577 tensor4 = [ 1 , 2 , 3 , 0 , 0 , 0 ] #eigen : 1, 2, 3 df = pd . DataFrame ([ tensor1 , tensor2 , tensor3 , tensor4 ], columns = [ \"XX\" , \"YY\" , \"ZZ\" , \"XY\" , \"XZ\" , \"YZ\" ]) df . index = node df .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } XX YY ZZ XY XZ YZ 1001 1 1 1 0 0 0 1002 4 -1 0 2 2 1 1003 1 6 5 3 3 1 1004 1 2 3 0 0 0 If you want to extract the eigenvalues of a tensor with numpy you have to pass a n by n ndarray to the eigenvalue function. In order to avoid having to loop over each node, this oneliner is highly optimized and will help you invert a large number of tensors efficiently. The steps are basically, create a list of n by n values (here n=3) in the right order => reshape it to a list of tensors => pass it to the eigenvals function import numpy as np from numpy import linalg as LA eigenvals = LA . eigvals ( df [[ \"XX\" , \"XY\" , \"XZ\" , \"XY\" , \"YY\" , \"YZ\" , \"XZ\" , \"YZ\" , \"ZZ\" ]] . values . reshape ( len ( df ), 3 , 3 )) eigenvals array([[ 1. , 1. , 1. ], [ 5.58442834, -0.80511809, -1.77931025], [-1.31577211, 8.85035616, 4.46541595], [ 1. , 2. , 3. ]])","tags":"Python","url":"redoules.github.io/python/Efficient_extraction_of_eigenvalues_from_a_list_of_tensors.html","loc":"redoules.github.io/python/Efficient_extraction_of_eigenvalues_from_a_list_of_tensors.html"},{"title":"Optimized numpy random number generation on Intel CPU","text":"Python Intel distribution Make sure you have a python intel distribution. When you startup python you should see somethine like : Python 3.6.2 |Intel Corporation| (default, Aug 15 2017, 11:34:02) [MSC v.1900 64 bit (AMD64)] Type 'copyright', 'credits' or 'license' for more information IPython 6.1.0 -- An enhanced Interactive Python. Type '?' for help. If not, you can force the installation of the intel optimized python with : conda update --all conda config --add channels intel conda install numpy --channel intel --override-channels oh and by the way, make sure you a running an Intel CPU ;) Comparing numpy.random with numpy.random_intel Let's now test both the rand function with and without the Intel optimization import numpy as np from numpy import random , random_intel % timeit np . random . rand ( 10 ** 5 ) 1.06 ms ± 91.5 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each) % timeit np . random_intel . rand ( 10 ** 5 ) 225 µs ± 3.46 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)","tags":"Python","url":"redoules.github.io/python/Optimized_numpy_random_intel.html","loc":"redoules.github.io/python/Optimized_numpy_random_intel.html"},{"title":"How to check Linux process information?","text":"How to check Linux process information (CPU usage, memory, user information, etc.)? You need to use the ps command combined with the grep command. In the example, we want to check the information on the nginx process : ps aux | grep nginx It would return the output : root 9976 0.0 0.0 12272 108 ? S<s Aug12 0:00 nginx: master process /usr/bin/nginx -g pid /run/nginx.pid; daemon on; master_process on; http 16780 0.0 0.0 12384 684 ? S< Aug12 4:11 nginx: worker process http 16781 0.0 0.0 12556 708 ? S< Aug12 0:24 nginx: worker process http 16782 0.0 0.1 12292 744 ? S< Aug12 2:43 nginx: worker process http 16783 0.0 0.1 12276 872 ? S< Aug12 0:24 nginx: worker process admin 17612 0.0 0.1 5120 864 pts/4 S+ 11:22 0:00 grep --color=auto nginx The columns have the following order : USER;PID;%CPU;%MEM;VSZ;RSS;TTY;STAT;START;TIME;COMMAND USER = user owning the process PID = process ID of the process %CPU = It is the CPU time used divided by the time the process has been running. %MEM = ratio of the process's resident set size to the physical memory on the machine VSZ = virtual memory usage of entire process (in KiB) RSS = resident set size, the non-swapped physical memory that a task has used (in KiB) TTY = controlling tty (terminal) STAT = multi-character process state START = starting time or date of the process TIME = cumulative CPU time COMMAND = command with all its arguments Interactive display If you want an interactive display showing in real time the statistics of the running process you can use the top command. If htop is available on you system, use this instead.","tags":"Linux","url":"redoules.github.io/linux/linux_process_information.html","loc":"redoules.github.io/linux/linux_process_information.html"},{"title":"Check the size of a directory","text":"How do you Check the size of a directory in linux? The du command will come handy for this task. Let's say we want to know the size of the directory named recommandations , we would run the following command du -sh recommendations It would return the output : 9.9M recommendations","tags":"Linux","url":"redoules.github.io/linux/directory_size.html","loc":"redoules.github.io/linux/directory_size.html"},{"title":"Check for free disk space","text":"How do you check for free disk space on linux? The df command will come handy for this task. Run the command with the following arguments df -ah Will return in a human readable format for all drives a readout of all your filesystems Filesystem Size Used Avail Use% Mounted on /dev/md0 2.4G 1.3G 1.1G 54% / none 348M 4.0K 348M 1% /dev none 0 0 0 - /dev/pts none 0 0 0 - /proc none 0 0 0 - /sys /tmp 350M 1.3M 349M 1% /tmp /run 350M 3.2M 347M 1% /run /dev/shm 350M 12K 350M 1% /dev/shm /proc/bus/usb 0 0 0 - /proc/bus/usb securityfs 0 0 0 - /sys/kernel/security /dev/md3 1.8T 372G 1.5T 21% /volume2 /dev/vg1000/lv 1.8T 1.5T 340G 82% /volume1 /dev/sdq1 7.4G 3.8G 3.5G 52% /volumeUSB3/usbshare3-1 /dev/sdr 294G 146G 134G 53% /volumeUSB2/usbshare none 0 0 0 - /proc/fs/nfsd none 0 0 0 - /config The free disk space can be read in the Avail column","tags":"Linux","url":"redoules.github.io/linux/free_disk_space_linux.html","loc":"redoules.github.io/linux/free_disk_space_linux.html"},{"title":"Check your current ip address","text":"Check your current ip address Run the command ip addr show and that will give you every information available 4: eth0: <> mtu 1500 group default qlen 1 link/ether 5c:51:4f:41:7a:b1 inet 169.254.33.33/16 brd 169.254.255.255 scope global dynamic valid_lft forever preferred_lft forever inet6 fe80::390a:f69e:1ba2:2121/64 scope global dynamic valid_lft forever preferred_lft forever 3: eth1: <BROADCAST,MULTICAST,UP> mtu 1500 group default qlen 1 link/ether 0a:00:27:00:00:03 inet 192.168.56.1/24 brd 192.168.56.255 scope global dynamic valid_lft forever preferred_lft forever inet6 fe80::84cd:374e:843f:f82f/64 scope global dynamic valid_lft forever preferred_lft forever 15: eth2: <> mtu 1500 group default qlen 1 link/ether 00:ff:d2:8a:19:c3 inet 169.254.40.62/16 brd 169.254.255.255 scope global dynamic valid_lft forever preferred_lft forever inet6 fe80::f199:2ca0:7aff:283e/64 scope global dynamic valid_lft forever preferred_lft forever 1: lo: <LOOPBACK,UP> mtu 1500 group default qlen 1 link/loopback 00:00:00:00:00:00 inet 127.0.0.1/8 brd 127.255.255.255 scope global dynamic valid_lft forever preferred_lft forever inet6 ::1/128 scope global dynamic valid_lft forever preferred_lft forever 5: wifi0: <BROADCAST,MULTICAST,UP> mtu 1500 group default qlen 1 link/ieee802.11 5c:51:4f:41:7a:ad inet 192.168.1.1/24 brd 192.168.1.255 scope global dynamic valid_lft 42720sec preferred_lft 42720sec inet6 fe80::395f:3594:1dc2:57e3/64 scope global dynamic valid_lft forever preferred_lft forever 21: wifi1: <> mtu 1500 group default qlen 1 link/ieee802.11 5c:51:4f:41:7a:ae inet 169.254.12.77/16 brd 169.254.255.255 scope global dynamic valid_lft forever preferred_lft forever inet6 fe80::58d5:630:cbbd:c4d/64 scope global dynamic valid_lft forever preferred_lft forever 12: eth3: <> mtu 1472 group default qlen 1 link/ether 00:00:00:00:00:00:00:e0:00:00:00:00:00:00:00:00 inet6 fe80::100:7f:fffe/64 scope global dynamic valid_lft forever preferred_lft forever 10: eth4: <BROADCAST,MULTICAST,UP> mtu 1500 group default qlen 1 link/ether 22:b7:57:52:5f:ff inet 192.168.42.106/24 brd 192.168.42.255 scope global dynamic valid_lft 6659sec preferred_lft 6659sec inet6 fe80::5110:eb6f:deb0:45c4/64 scope global dynamic valid_lft forever preferred_lft forever You can select only one interface ip addr show eth0 and only the relevant information will be displayed 4: eth0: <> mtu 1500 group default qlen 1 link/ether 5c:51:4f:41:7a:b1 inet 169.254.33.33/16 brd 169.254.255.255 scope global dynamic valid_lft forever preferred_lft forever inet6 fe80::390a:f69e:1ba2:2121/64 scope global dynamic valid_lft forever preferred_lft forever Check your current ip address the old way The ifconfig command will return information regarding your network interfaces. Let's try it: ifconfig eth1 Link encap:Ethernet HWaddr 0a:00:27:00:00:03 inet adr:192.168.56.1 Bcast:192.168.56.255 Masque:255.255.255.0 adr inet6: fe80::84cd:374e:843f:f82f/64 Scope:Global UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 Packets reçus:0 erreurs:0 :0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 Octets reçus:0 (0.0 B) Octets transmis:0 (0.0 B) eth4 Link encap:Ethernet HWaddr 22:b7:57:52:5f:ff inet adr:192.168.42.106 Bcast:192.168.42.255 Masque:255.255.255.0 adr inet6: fe80::5110:eb6f:deb0:45c4/64 Scope:Global UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 Packets reçus:0 erreurs:0 :0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 Octets reçus:0 (0.0 B) Octets transmis:0 (0.0 B) lo Link encap:Boucle locale inet adr:127.0.0.1 Masque:255.0.0.0 adr inet6: ::1/128 Scope:Global UP LOOPBACK RUNNING MTU:1500 Metric:1 Packets reçus:0 erreurs:0 :0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 Octets reçus:0 (0.0 B) Octets transmis:0 (0.0 B) wifi0 Link encap:UNSPEC HWaddr 5C-51-4F-41-7A-AD-00-00-00-00-00-00-00-00-00-00 inet adr:192.168.1.1 Bcast:192.168.1.255 Masque:255.255.255.0 adr inet6: fe80::395f:3594:1dc2:57e3/64 Scope:Global UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 Packets reçus:0 erreurs:0 :0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 Octets reçus:0 (0.0 B) Octets transmis:0 (0.0 B) On the left column you have the list of network adapters. The lo is the local loopback, it is an interface that points to the localhost. Interfaces starting with eth refer to wired connections over ethernet (or sometimes USB in the case of a phone acting as an access point over USB). Interfaces starting with wlan or wifi refer to wireless connections. On the right column you have some information corresponding to the interface such as the IPv4, the IPv6, the mask, some statistics about the interface and so on.","tags":"Linux","url":"redoules.github.io/linux/get_ip_linux.html","loc":"redoules.github.io/linux/get_ip_linux.html"},{"title":"Check the version of the kernel currently running","text":"Check the version of the kernel currently running The uname command will give you the version of the kernel. In order to get a more useful output, type uname -a This will return : the hostname os name kernel release/version architecture * etc. variations If you only want the kernel version you can type uname -v if you only want the kernel release you can type uname -r","tags":"Linux","url":"redoules.github.io/linux/version_kernel.html","loc":"redoules.github.io/linux/version_kernel.html"},{"title":"Running the notebook on a remote server","text":"Jupyter hub With JupyterHub you can create a multi-user Hub which spawns, manages, and proxies multiple instances of the single-user Jupyter notebook server. Project Jupyter created JupyterHub to support many users. The Hub can offer notebook servers to a class of students, a corporate data science workgroup, a scientific research project, or a high performance computing group. However, if you are the only one using the server and you just want a simple way to run the notebook on your server and access it through the web interface on a light client without having to install and configure the jupyter hub, you can do the following. Problem with jupyter notebook On your server, run the command jupyter-notebook you should get something like : [I 11:18:44.514 NotebookApp] Serving notebooks from local directory: /volume2/homes/admin [I 11:18:44.515 NotebookApp] 0 active kernels [I 11:18:44.516 NotebookApp] The Jupyter Notebook is running at: http://localhost:8888/?token=357587e3269b0f20f2b7e1918492890ae7573ac7ef1d2023 [I 11:18:44.516 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation). [W 11:18:44.519 NotebookApp] No web browser found: could not locate runnable browser. [C 11:18:44.520 NotebookApp] Copy/paste this URL into your browser when you connect for the first time, to login with a token: http://localhost:8888/?token=357587e3269b0f20f2b7e1918492890ae7573ac7ef1d2023 and if you try to connect to your server ip (in my example : http://192.168.1.2:8888/?token=357587e3269b0f20f2b7e1918492890ae7573ac7ef1d2023) you will get an \"ERR_CONNECTION_REFUSED\" error. This is because, by default, Jupyter Notebook only accepts connections from localhost. Allowing connexions from other sources From any IP The simplest way to avoid the connection error is to allow the notebook to accept connections from any ip jupyter-notebook --ip = * you will get something like [W 11:26:45.285 NotebookApp] WARNING: The notebook server is listening on all IP addresses and not using encryption. This is not recommended. [I 11:26:45.342 NotebookApp] Serving notebooks from local directory: /volume2/homes/admin [I 11:26:45.342 NotebookApp] 0 active kernels [I 11:26:45.343 NotebookApp] The Jupyter Notebook is running at: http://[all ip addresses on your system]:8888/?token=52af33d628881824968b4031967e8541a27cc28b1720c199 [I 11:26:45.343 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation). [W 11:26:45.346 NotebookApp] No web browser found: could not locate runnable browser. [C 11:26:45.347 NotebookApp] Copy/paste this URL into your browser when you connect for the first time, to login with a token: http://localhost:8888/?token=52af33d628881824968b4031967e8541a27cc28b1720c199 and if you connect form a remote client (192.168.1.1 in my example), the following line will be added to the output : [I 11:26:54.798 NotebookApp] 302 GET /?token=52af33d628881824968b4031967e8541a27cc28b1720c199 (192.168.1.1) 111.17ms note that you should only do that if you are the only one using the server because the connection is not encypted. From a specific IP You can also, explicitly specify the ip of the client jupyter-notebook --ip = 192 .168.1.1 [I 11:44:58.104 NotebookApp] JupyterLab extension loaded from C:\\Users\\Guillaume\\Miniconda3\\lib\\site-packages\\jupyterlab [I 11:44:58.104 NotebookApp] JupyterLab application directory is C:\\Users\\Guillaume\\Miniconda3\\share\\jupyter\\lab [I 11:44:58.244 NotebookApp] Serving notebooks from local directory: C:\\Users\\Guillaume [I 11:44:58.245 NotebookApp] 0 active kernels [I 11:44:58.245 NotebookApp] The Jupyter Notebook is running at: http://192.168.1.1:8888/?token=503576dd8fa87d1f2c416df307e9b900e520b4942e317b32 [I 11:44:58.245 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation). [C 11:44:58.258 NotebookApp] Copy/paste this URL into your browser when you connect for the first time, to login with a token: http://192.168.1.1:8888/?token=503576dd8fa87d1f2c416df307e9b900e520b4942e317b32 [I 11:44:59.083 NotebookApp] Accepting one-time-token-authenticated connection from 192.168.1.1","tags":"Jupyter","url":"redoules.github.io/jupyter/remote_run_notebook.html","loc":"redoules.github.io/jupyter/remote_run_notebook.html"},{"title":"Running multiple calls to a function in parallel with Dask","text":"Dask.distributed is a lightweight library for distributed computing in Python. It allows to create a compute graph. Dask distributed is architectured around 3 parts : the dask-scheduler the dask-worker(s) the dask client Dask architecture The Dask scheduler is a centrally managed, distributed, dynamic task scheduler. It recieves tasks from a/multiple client(s) and spread them across one or multiple dask-worker(s). Dask-scheduler is an event based asynchronous dynamic scheduler, meaning that mutliple clients can submit a list of task to be executed on multiple workers. Internally, the task are represented as a directed acyclic graph. Both new clients and new workers can be connected or disconnected during the execution of the task graph. Tasks can be submited with the function client . submit ( function , * args , ** kwargs ) or by using objects from the dask library such as dask.dataframe, dask.bag or dask.array Setup In this example, we will use a distributed scheduler on a single machine with multiple workers and a single client. We will use the client to submit some tasks to the scheduler. The scheduler will then dispatch those tasks to the workers. The process can be monitored in real time through a web application. For this example, all the computations will be run on a local computer. However dask can scale to a large HPC cluster. First we have to launch the dask-scheduler; from the command line, input dask-scheduler Next, you can load the web dashboard. In order to do so, the scheduler returns the number of the port you have to connect to in the line starting with \"bokeh at :\". The default port is 8787. Since we are running all the programs on the same computer, we just have to login to http://127.0.0.1:8787/status Finally, we have to launch the dask-worker(s). If you want to run the worker(s) on the same computer as the scheduler the type : dask-worker 127 .0.0.1:8786 otherwise, make sure you are inputing the ip address of the computer hosting the dask-scheduler. You can launch as many workers as you want. In this example, we will run 3 workers on the local machine. Use the dask workers within your python code We will now see how to submit multiple calls to a fucntion in parallel on the dask-workers. Import the required libraries and define the function to be executed. import numpy as np import pandas as pd from distributed import Client #function used to do parallel computing on def compute_pi_MonteCarlo ( Nb_Data ): \"\"\" computes the value of pi using the monte carlo method \"\"\" Radius = 1 Nb_Data = int ( round ( Nb_Data )) x = np . random . uniform ( - Radius , Radius , Nb_Data ) y = np . random . uniform ( - Radius , Radius , Nb_Data ) pi_mc = 4 * np . sum ( np . power ( x , 2 ) + np . power ( y , 2 ) < Radius ** 2 ) / Nb_Data err = 100 * np . abs ( pi_mc - np . pi ) / np . pi return [ Nb_Data , pi_mc , err ] In order to connect to the scheduler, we create a client. client = Client ( '127.0.0.1:8786' ) client Client Scheduler: tcp://127.0.0.1:8786 Dashboard: http://127.0.0.1:8787/status Cluster Workers: 3 Cores: 12 Memory: 25.48 GB We submit tasks using the submit method data = [ client . submit ( compute_pi_MonteCarlo , Nb_Data ) for Nb_Data in np . logspace ( 3 , 7 , num = 1200 , dtype = int )] If you look at http://127.0.0.1:8787/status you will see the tasks beeing completed. Once competed, gather the data: data = client . gather ( data ) df = pd . DataFrame ( data ) df . columns = [ \"number of points for MonteCarlo\" , \"value of pi\" , \"error (%)\" ] df . tail () .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } number of points for MonteCarlo value of pi error (%) 1195 9697405 3.141296 0.009454 1196 9772184 3.141058 0.017008 1197 9847540 3.141616 0.000739 1198 9923477 3.141009 0.018574 1199 10000000 3.141032 0.017833 There, we have completed a simple example on how to use dask to run multiple functions in parallel. Full source code: import numpy as np import pandas as pd from distributed import Client #function used to do parallel computing on def compute_pi_MonteCarlo ( Nb_Data ): \"\"\" computes the value of pi using the monte carlo method \"\"\" Radius = 1 Nb_Data = int ( round ( Nb_Data )) x = np . random . uniform ( - Radius , Radius , Nb_Data ) y = np . random . uniform ( - Radius , Radius , Nb_Data ) pi_mc = 4 * np . sum ( np . power ( x , 2 ) + np . power ( y , 2 ) < Radius ** 2 ) / Nb_Data err = 100 * np . abs ( pi_mc - np . pi ) / np . pi return [ Nb_Data , pi_mc , err ] #connect to the scheduler client = Client ( '127.0.0.1:8786' ) #submit tasks data = [ client . submit ( compute_pi_MonteCarlo , Nb_Data ) for Nb_Data in np . logspace ( 3 , 7 , num = 1200 , dtype = int )] #gather the results data = client . gather ( data ) df = pd . DataFrame ( data ) df . columns = [ \"number of points for MonteCarlo\" , \"value of pi\" , \"error (%)\" ] df . tail () A word on the environement variables On Windows, to make sure that you can run dask-scheduler and dask-worker from the command line, you have to add the location of the executable to your path. On linux, you can append the location of the dask-worker and scheduler to the path variable with the command export PATH = $PATH :/path/to/dask","tags":"Python","url":"redoules.github.io/python/dask_distributed_parallelism.html","loc":"redoules.github.io/python/dask_distributed_parallelism.html"},{"title":"Plotting data using log axis","text":"Plotting in log axis with matplotlib import matplotlib.pyplot as plt % matplotlib inline import numpy as np x = np . linspace ( 0.1 , 20 ) y = 20 * np . exp ( - x / 10.0 ) Plotting using the standard function then specifying the axis scale One of the easiest way to plot in a log plot is to specify the plot normally and then specify which axis is to be plotted with a log scale. This can be specified by the function set_xscale or set_yscale # Normal plot fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . plot ( x , y ) ax . grid () plt . show () # Log x axis plot fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . plot ( x , y ) ax . set_xscale ( 'log' ) ax . grid () plt . show () # Log x axis plot fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . plot ( x , y ) ax . set_yscale ( 'log' ) ax . grid () plt . show () # Log x axis plot fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . plot ( x , y ) ax . set_xscale ( 'log' ) ax . set_yscale ( 'log' ) ax . grid () plt . show () Plotting using the matplotlib defined function Matplotlib has the function : semilogx, semilogy and loglog that can help you avoid having to specify the axis scale. # Plot using semilogx fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . semilogx ( x , y ) ax . grid () plt . show () # Plot using semilogy fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . semilogy ( x , y ) ax . grid () plt . show () # Plot using loglog fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . loglog ( x , y ) ax . grid () plt . show ()","tags":"Python","url":"redoules.github.io/python/logplot.html","loc":"redoules.github.io/python/logplot.html"},{"title":"Downloading a static webpage with python","text":"If you are using python legacy (aka python 2) first of all, stop ! Furthermore, this method won't work in python legacy # Import modules from urllib.request import urlopen The webpage source code can be downloaded with the command urlopen url = \"http://example.com/\" #create a HTTP request in order to read the page page = urlopen ( url ) . read () The source code will be stored in the variable page as a string print ( page ) b '&lt;!doctype html&gt;\\n&lt;html&gt;\\n&lt;head&gt;\\n &lt;title&gt;Example Domain&lt;/title&gt;\\n\\n &lt;meta charset=\"utf-8\" /&gt;\\n &lt;meta http-equiv=\"Content-type\" content=\"text/html; charset=utf-8\" /&gt;\\n &lt;meta name=\"viewport\" content=\"width=device-width, initial-scale=1\" /&gt;\\n &lt;style type=\"text/css\"&gt;\\n body {\\n background-color: #f0f0f2;\\n margin: 0;\\n padding: 0;\\n font-family: \"Open Sans\", \"Helvetica Neue\", Helvetica, Arial, sans-serif;\\n \\n }\\n div {\\n width: 600px;\\n margin: 5em auto;\\n padding: 50px;\\n background-color: #fff;\\n border-radius: 1em;\\n }\\n a:link, a:visited {\\n color: #38488f;\\n text-decoration: none;\\n }\\n @media (max-width: 700px) {\\n body {\\n background-color: #fff;\\n }\\n div {\\n width: auto;\\n margin: 0 auto;\\n border-radius: 0;\\n padding: 1em;\\n }\\n }\\n &lt;/style&gt; \\n&lt;/head&gt;\\n\\n&lt;body&gt;\\n&lt;div&gt;\\n &lt;h1&gt;Example Domain&lt;/h1&gt;\\n &lt;p&gt;This domain is established to be used for illustrative examples in documents. You may use this\\n domain in examples without prior coordination or asking for permission.&lt;/p&gt;\\n &lt;p&gt;&lt;a href=\"http://www.iana.org/domains/example\"&gt;More information...&lt;/a&gt;&lt;/p&gt;\\n&lt;/div&gt;\\n&lt;/body&gt;\\n&lt;/html&gt;\\n' Additionally, you can beautifulsoup in order to make it easier to work with html from bs4 import BeautifulSoup soup = BeautifulSoup ( page , 'lxml' ) soup . prettify () print ( soup ) & lt ;! DOCTYPE html & gt ; & lt ; html & gt ; & lt ; head & gt ; & lt ; title & gt ; Example Domain & lt ;/ title & gt ; & lt ; meta charset = \"utf-8\" /& gt ; & lt ; meta content = \"text/html; charset=utf-8\" http-equiv = \"Content-type\" /& gt ; & lt ; meta content = \"width=device-width, initial-scale=1\" name = \"viewport\" /& gt ; & lt ; style type = \"text/css\" & gt ; body { background-color : #f0f0f2 ; margin : 0 ; padding : 0 ; font-family : \"Open Sans\" , \"Helvetica Neue\" , Helvetica , Arial , sans-serif ; } div { width : 600 px ; margin : 5 em auto ; padding : 50 px ; background-color : #fff ; border-radius : 1 em ; } a : link , a : visited { color : #38488f ; text-decoration : none ; } @ media ( max-width : 700px ) { body { background-color : #fff ; } div { width : auto ; margin : 0 auto ; border-radius : 0 ; padding : 1 em ; } } & lt ;/ style & gt ; & lt ;/ head & gt ; & lt ; body & gt ; & lt ; div & gt ; & lt ; h1 & gt ; Example Domain & lt ;/ h1 & gt ; & lt ; p & gt ; This domain is established to be used for illustrative examples in documents . You may use this domain in examples without prior coordination or asking for permission .& lt ;/ p & gt ; & lt ; p & gt ;& lt ; a href = \"http://www.iana.org/domains/example\" & gt ; More information ...& lt ;/ a & gt ;& lt ;/ p & gt ; & lt ;/ div & gt ; & lt ;/ body & gt ; & lt ;/ html & gt ;","tags":"Python","url":"redoules.github.io/python/download_page.html","loc":"redoules.github.io/python/download_page.html"},{"title":"Getting stock market data","text":"Start by importing the packages. We will need pandas and the pandas_datareader. # Import modules import pandas as pd from pandas_datareader import data Datareader allows you to import data from the internet. I have found that Quandl and robinhood works the best as a source for stockmarket data. Note that if you want an other type of data (e.g. GDP, inflation, etc.) other sources exist. #import stock from robinhood aapl_robinhood = data . DataReader ( 'AAPL' , 'robinhood' , '1980-01-01' ) aapl_robinhood . head () .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } close_price high_price interpolated low_price open_price session volume symbol begins_at AAPL 2017-08-04 153.996200 154.990700 False 153.306900 153.681100 reg 20559852 2017-08-07 156.379100 156.487400 False 154.272000 154.655900 reg 21870321 2017-08-08 157.629700 159.352900 False 155.847400 156.172300 reg 36205896 2017-08-09 158.594700 158.801500 False 156.674500 156.822200 reg 26131530 2017-08-10 153.543100 158.169600 False 152.861000 158.070700 reg 40804273 #import stock from quandl aapl_quandl = data . DataReader ( 'AAPL' , 'quandl' , '1980-01-01' ) aapl_quandl . head () .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } Open High Low Close Volume ExDividend SplitRatio AdjOpen AdjHigh AdjLow AdjClose AdjVolume Date 2018-03-27 173.68 175.15 166.92 168.340 38962839.0 0.0 1.0 173.68 175.15 166.92 168.340 38962839.0 2018-03-26 168.07 173.10 166.44 172.770 36272617.0 0.0 1.0 168.07 173.10 166.44 172.770 36272617.0 2018-03-23 168.39 169.92 164.94 164.940 40248954.0 0.0 1.0 168.39 169.92 164.94 164.940 40248954.0 2018-03-22 170.00 172.68 168.60 168.845 41051076.0 0.0 1.0 170.00 172.68 168.60 168.845 41051076.0 2018-03-21 175.04 175.09 171.26 171.270 35247358.0 0.0 1.0 175.04 175.09 171.26 171.270 35247358.0","tags":"Python","url":"redoules.github.io/python/stock_pandas.html","loc":"redoules.github.io/python/stock_pandas.html"},{"title":"Moving average with pandas","text":"# Import modules import pandas as pd from pandas_datareader import data , wb #import packages from pandas_datareader import data aapl = data . DataReader ( 'AAPL' , 'quandl' , '1980-01-01' ) aapl . head () .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } Open High Low Close Volume ExDividend SplitRatio AdjOpen AdjHigh AdjLow AdjClose AdjVolume Date 2018-03-27 173.68 175.15 166.92 168.340 38962839.0 0.0 1.0 173.68 175.15 166.92 168.340 38962839.0 2018-03-26 168.07 173.10 166.44 172.770 36272617.0 0.0 1.0 168.07 173.10 166.44 172.770 36272617.0 2018-03-23 168.39 169.92 164.94 164.940 40248954.0 0.0 1.0 168.39 169.92 164.94 164.940 40248954.0 2018-03-22 170.00 172.68 168.60 168.845 41051076.0 0.0 1.0 170.00 172.68 168.60 168.845 41051076.0 2018-03-21 175.04 175.09 171.26 171.270 35247358.0 0.0 1.0 175.04 175.09 171.26 171.270 35247358.0 In order to computer the moving average, we will use the rolling function. #120 days moving average moving_averages = aapl [[ \"Open\" , \"High\" , \"Low\" , \"Close\" , \"Volume\" ]] . rolling ( window = 120 ) . mean () moving_averages . tail () .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } Open High Low Close Volume Date 1980-12-18 28.457667 28.551917 28.385000 28.385000 139495.000000 1980-12-17 28.410750 28.502917 28.338083 28.338083 141772.500000 1980-12-16 28.362833 28.453917 28.289167 28.289167 141256.666667 1980-12-15 28.335750 28.426833 28.262083 28.262083 144321.666667 1980-12-12 28.310750 28.402833 28.238167 28.238167 159625.000000 % matplotlib inline import matplotlib.pyplot as plt plt . plot ( aapl . index , aapl . Open , label = 'Open price' ) plt . plot ( moving_averages . index , moving_averages . Open , label = \"120 MA Open price\" ) plt . legend () plt . show ()","tags":"Python","url":"redoules.github.io/python/Moving_average_pandas.html","loc":"redoules.github.io/python/Moving_average_pandas.html"},{"title":"Keywords to use with WHERE","text":"Keywords to use with WHERE #load the extension % load_ext sql #connect to the database % sql sqlite : /// mydatabase . db 'Connected: @mydatabase.db' Assignment operator The assignment operator is =. % sql SELECT * FROM tutyfrutty WHERE color = \"red\" * sqlite:///mydatabase.db Done. index fruit color kcal 2 Apple red 52 7 Cranberry red 308 Comparison operators Comparison operation can be done in a SQL querry. They are the following : Equality : = Greater than : > greater than or equal to : >= less than : < less than or equal to : <= not equal to : <>, != not greater than : !> not less than : !< % sql SELECT * FROM tutyfrutty WHERE kcal = 47 * sqlite:///mydatabase.db Done. index fruit color kcal 1 Orange orange 47 % sql SELECT * FROM tutyfrutty WHERE kcal > 47 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 2 Apple red 52 7 Cranberry red 308 % sql SELECT * FROM tutyfrutty WHERE kcal >= 47 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 7 Cranberry red 308 % sql SELECT * FROM tutyfrutty WHERE kcal < 47 * sqlite:///mydatabase.db Done. index fruit color kcal 3 lemon yellow 15 4 lime green 30 5 plum purple 28 % sql SELECT * FROM tutyfrutty WHERE kcal <= 47 * sqlite:///mydatabase.db Done. index fruit color kcal 1 Orange orange 47 3 lemon yellow 15 4 lime green 30 5 plum purple 28 % sql SELECT * FROM tutyfrutty WHERE kcal <> 47 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 7 Cranberry red 308 Logical operators Logical operators test a condition and return a boolean. The logicial operators in SQL are : ALL : true if all the condtions are true AND : true is both conditions are true ANY : true if any one of the conditions are true BETWEEN : true if the operand in withing a range of values EXISTS : true if the subquery contains any rows IN : true if the condition is present in a row LIKE : true if a pattern is matched NOT : True if the operand is false, false otherwise OR : True is either condition is true SOME : true is any of the conditions is true % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" AND kcal < 100 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 3 lemon yellow 15 % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" OR kcal > 300 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 3 lemon yellow 15 7 Cranberry red 308 % sql SELECT * FROM tutyfrutty WHERE fruit LIKE 'l%' * sqlite:///mydatabase.db Done. index fruit color kcal 3 lemon yellow 15 4 lime green 30 % sql SELECT * FROM tutyfrutty WHERE NOT color = \"yellow\" * sqlite:///mydatabase.db Done. index fruit color kcal 1 Orange orange 47 2 Apple red 52 4 lime green 30 5 plum purple 28 7 Cranberry red 308 % sql SELECT * FROM tutyfrutty WHERE kcal BETWEEN 40 AND 100 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 Bitwise operators Some bitwise operators exist in SQL. They will not be demonstrated here. They are the following : AND : & OR : | XOR : &#94; NOT : ~","tags":"SQL","url":"redoules.github.io/sql/WHERE_SQL_keywords.html","loc":"redoules.github.io/sql/WHERE_SQL_keywords.html"},{"title":"Sorting results","text":"Sorting results in SQL Sorting results can be achieved by using a modifier command at the end of the SQL querry #load the extension % load_ext sql #connect to the database % sql sqlite : /// mydatabase . db 'Connected: @mydatabase.db' The results can be sorted with the command ORDER BY SELECT column-list FROM table_name [WHERE condition] [ORDER BY column1, column2, .. columnN] [ASC | DESC] Let's show an example where we extract the fruits that are either yellow or red % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" OR color = \"red\" * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 2 Apple red 52 3 lemon yellow 15 7 Cranberry red 308 Ascending sort % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" OR color = \"red\" ORDER BY kcal ASC * sqlite:///mydatabase.db Done. index fruit color kcal 3 lemon yellow 15 2 Apple red 52 0 Banana yellow 89 7 Cranberry red 308 descending sort % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" OR color = \"red\" ORDER BY kcal DESC * sqlite:///mydatabase.db Done. index fruit color kcal 7 Cranberry red 308 0 Banana yellow 89 2 Apple red 52 3 lemon yellow 15 Sort by multiple columns You can sort by more than one column. Just specify multiple columns in the ORDER BY keyword. In the example, we will sort alphabetically on the color column first and sort alphabetically on the fruit column % sql SELECT * FROM tutyfrutty ORDER BY color , fruit ASC * sqlite:///mydatabase.db Done. index fruit color kcal 4 lime green 30 1 Orange orange 47 5 plum purple 28 2 Apple red 52 7 Cranberry red 308 0 Banana yellow 89 3 lemon yellow 15","tags":"SQL","url":"redoules.github.io/sql/Sorting_results.html","loc":"redoules.github.io/sql/Sorting_results.html"},{"title":"Filter content of a TABLE","text":"Filter content of a TABLE in SQL In this example, we will display the content of a table but we will filter out the results. Since we are working in the notebook, we will load the sql extension in order to manipulate the database. The database mydatabase.db is a SQLite database already created before the example. #load the extension % load_ext sql #connect to the database % sql sqlite : /// mydatabase . db 'Connected: @mydatabase.db' Filter content matching exactly a condition We want to extract all the entries in a dataframe that match a certain condition, in order to do so, we will use the following command : SELECT * FROM TABLE WHERE column=\"condition\" In our example, we will filter all the entries in the tutyfrutty table whose color is yellow % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 3 lemon yellow 15 Complex conditions You can build more complex conditions by using the keywords OR and AND In the following example, we will filter all entries that are either yellow or red % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" OR color = \"red\" * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 2 Apple red 52 3 lemon yellow 15 7 Cranberry red 308 Note : when combining multiple conditions with AND and OR, be careful to use parentesis where needed Conditions matching a pattern You can also use the LIKE keyword in order to find all entries that match a certain pattern. In our example, we want to find all fruits begining with a \"l\". In order to do so, we will use the LIKE keyword and the wildcard \"%\" meaning any string % sql SELECT * FROM tutyfrutty WHERE fruit LIKE \"l%\" * sqlite:///mydatabase.db Done. index fruit color kcal 3 lemon yellow 15 4 lime green 30 Numerical conditions When we are working with numerical data, we can use the GREATER THAN > and SMALLER THAN < operators % sql SELECT * FROM tutyfrutty WHERE kcal < 47 * sqlite:///mydatabase.db Done. index fruit color kcal 3 lemon yellow 15 4 lime green 30 5 plum purple 28 If we want the condition to be inclusive we can use the operator <= (alternatively >=) % sql SELECT * FROM tutyfrutty WHERE kcal <= 47 * sqlite:///mydatabase.db Done. index fruit color kcal 1 Orange orange 47 3 lemon yellow 15 4 lime green 30 5 plum purple 28","tags":"SQL","url":"redoules.github.io/sql/display_table_filter.html","loc":"redoules.github.io/sql/display_table_filter.html"},{"title":"Displaying the content of a TABLE","text":"Displaying the content of a TABLE in SQL In this very simple example we will see how to display the content of a table. Since we are working in the notebook, we will load the sql extension in order to manipulate the database. The database mydatabase.db is a SQLite database already created before the example. #load the extension % load_ext sql #connect to the database % sql sqlite : /// mydatabase . db 'Connected: @mydatabase.db' In order to extract all the values from a table, we will use the following command : SELECT * FROM TABLE In our example, we want to display the data contained in the table named tutyfrutty % sql SELECT * FROM tutyfrutty * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 7 Cranberry red 308","tags":"SQL","url":"redoules.github.io/sql/display_table.html","loc":"redoules.github.io/sql/display_table.html"},{"title":"Opening a file with python","text":"This short article show you how to open a file using python. We will use the with keyword in order to avoid having to close the file. There is no need to import anything in order to open a file. All the function related to file manipulation are part of the python standard library In order to open a file, we will use the function open. This function takes two arguments : the path of the file the mode you want to open the file The mode can be : 'r' : read 'w' : write 'a' : append (writes at the end of the file) 'b' : binary mode 'x' : exclusive creation 't' : text mode (by default) Note that if the file does not exit it will be created if you use the following options \"w\", \"a\", \"x\". If you try to open a non existing file in read mode 'r', a FileNotFoundError will be returned. It is possible to combine multiple options together. For instance, you can open a file in binary mode for writing using the 'wb' option. Python distinguishes between binary and text I/O. Files opened in binary mode return contents as bytes objects without any decoding. In text mode , the contents of the file are returned as str, the bytes having been first decoded using a platform-dependent encoding or using the specified encoding if given. Writing to a file Let's first open (create) a text file a write a string to it. filepath = \". \\\\ myfile.txt\" with open ( filepath , 'w' ) as f : f . write ( \"Hello world !\" ) Reading a file we can now see how to read the content of a file. To do so, we will use the 'r' option with open ( filepath , \"r\" ) as f : content = f . read () print ( content ) Hello world ! A word on the with keyword In python the with keyword is used when working with unmanaged resources (like file streams). The python documentation tells us that : The with statement clarifies code that previously would use try...finally blocks to ensure that clean-up code is executed. In this section, I'll discuss the statement as it will commonly be used. In the next section, I'll examine the implementation details and show how to write objects for use with this statement. The with statement is a control-flow structure whose basic structure is: with expression [ as variable ]: with - block The expression is evaluated, and it should result in an object that supports the context management protocol (that is, has enter () and exit () methods).","tags":"Python","url":"redoules.github.io/python/Opening_file.html","loc":"redoules.github.io/python/Opening_file.html"},{"title":"Opening a SQLite database with python","text":"This short article show you how to connect to a SQLite database using python. We will use the with keyword in order to avoid having to close the database. In order to connect to the database, we will have to import sqlite3 import sqlite3 from sqlite3 import Error In python the with keyword is used when working with unmanaged resources (like file streams). The python documentation tells us that : The with statement clarifies code that previously would use try...finally blocks to ensure that clean-up code is executed. In this section, I'll discuss the statement as it will commonly be used. In the next section, I'll examine the implementation details and show how to write objects for use with this statement. The with statement is a control-flow structure whose basic structure is: with expression [ as variable ]: with - block The expression is evaluated, and it should result in an object that supports the context management protocol (that is, has enter () and exit () methods). db_file = \". \\\\ mydatabase.db\" try : with sqlite3 . connect ( db_file ) as conn : print ( \"Connected to the database\" ) #your code here except Error as e : print ( e ) Connected to the database","tags":"Python","url":"redoules.github.io/python/Opening_SQLite_database.html","loc":"redoules.github.io/python/Opening_SQLite_database.html"},{"title":"Reading data from a sql database with pandas","text":"When manipulating you data using pandas, it is sometimes useful to pull data from a database. In this tutorial, we will see how to querry a dataframe from a sqlite table. Note than it would also work with any other sql database a long as you change the connxion to the one that suits your needs. First let's import pandas and sqlite3 import pandas as pd import sqlite3 from sqlite3 import Error We want to store the table tutyfrutty in our dataframe. To do so, we will query all the elements present in the tutyfrutty TABLE with the command : SELECT * FROM tutyfrutty db_file = \". \\\\ mydatabase.db\" try : with sqlite3 . connect ( db_file ) as conn : df = pd . read_sql ( \"SELECT * FROM tutyfrutty\" , conn ) del df [ \"index\" ] #juste delete the index column that was stored in the table except Error as e : print ( e ) df .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 6 Cranberry red 308 7 Cranberry red 308","tags":"Python","url":"redoules.github.io/python/Reading_data_from_a_sql_database_with_pandas.html","loc":"redoules.github.io/python/Reading_data_from_a_sql_database_with_pandas.html"},{"title":"Writing data to a sql database with pandas","text":"When manipulating you data using pandas, it is sometimes useful to store a dataframe. Pandas provides multiple ways to export dataframes. The most common consist in exporting to a csv, a pickle, to hdf or to excel. However, exporting to a sql database can prove very useful. Indeed, having a well structured database is a great for storing all the data related to your analysis in one place. In this tutorial, we will see how to store a dataframe in a new table of a sqlite dataframe. Note than it would also work with any other sql database a long as you change the connxion to the one that suits your needs. First let's import pandas and sqlite3 import pandas as pd import sqlite3 from sqlite3 import Error # Example dataframe raw_data = { 'fruit' : [ 'Banana' , 'Orange' , 'Apple' , 'lemon' , \"lime\" , \"plum\" ], 'color' : [ 'yellow' , 'orange' , 'red' , 'yellow' , \"green\" , \"purple\" ], 'kcal' : [ 89 , 47 , 52 , 15 , 30 , 28 ] } df = pd . DataFrame ( raw_data , columns = [ 'fruit' , 'color' , 'kcal' ]) df .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 Now that the DataFrame has been created, let's push it to the sqlite database called mydatabase.db in a new table called tutyfrutty db_file = \". \\\\ mydatabase.db\" try : with sqlite3 . connect ( db_file ) as conn : df . to_sql ( \"tutyfrutty\" , conn ) except Error as e : print ( e ) except ValueError : print ( \"The TABLE tutyfrutty already exists, read below to understand how to handle this case\" ) Note that if the table tutyfrutty was already existing, the to_sql function will return a ValueError. This is where, the if_exists option comes into play. Let's look at the docstring of this function : \"\"\" if_exists : {'fail', 'replace', 'append'}, default 'fail' - fail: If table exists, do nothing. - replace: If table exists, drop it, recreate it, and insert data. - append: If table exists, insert data. Create if does not exist. \"\"\" Let's say, I want to update my dataframe with some new rows df . loc [ len ( df ) + 1 ] = [ 'Cranberry' , 'red' , 308 ] df .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 7 Cranberry red 308 8 Cranberry red 308 I can now replace the table with the new values using the \"replace\" option db_file = \". \\\\ mydatabase.db\" try : with sqlite3 . connect ( db_file ) as conn : df . to_sql ( \"tutyfrutty\" , conn , if_exists = \"replace\" ) except Error as e : print ( e )","tags":"Python","url":"redoules.github.io/python/Writing_data_to_a_sql_database_with_pandas.html","loc":"redoules.github.io/python/Writing_data_to_a_sql_database_with_pandas.html"},{"title":"Creating a sqlite database","text":"When you want to start with using databases SQlite is a great tool. It provides an easy onramp to learn and prototype you database with a SQL compatible database. First, let's import the libraries we need import sqlite3 from sqlite3 import Error SQlite doesn't need a database server, however, you have to start by creating an empty database file import os def check_for_db_file (): if os . path . exists ( \"mydatabase.db\" ): print ( \"the database is ready\" ) else : print ( \"no database found\" ) check_for_db_file () no database found Let's then create a function that will connect to a database, print the verison of sqlite and then close the connexion to the database. def create_database ( db_file ): \"\"\" create a database connection to a SQLite database \"\"\" try : with sqlite3 . connect ( db_file ) as conn : print ( \"database created with sqlite3 version {0}\" . format ( sqlite3 . version )) except Error as e : print ( e ) create_database ( \".\\mydatabase.db\" ) database created with sqlite3 version 2.6.0 check_for_db_file () the database is ready You're all set. From now on, you can open the database and write sql querries into it.","tags":"Python","url":"redoules.github.io/python/Creating_a_sqlite_database.html","loc":"redoules.github.io/python/Creating_a_sqlite_database.html"},{"title":"Setting up the notebook for plotting with matplotlib","text":"Importing Matplotlib First we need to import pyplot, a collection of command style functions that make matplotlib work like MATLAB. Let's, as well, use the magic command %matplotlib inline in order to display the figures in the notebook import matplotlib.pyplot as plt % matplotlib inline # this doubles image size, but we'll do it manually below # %config InlineBackend.figure_format = 'retina' The following parameters are recommended for matplotlib, they will make matplotlib output a better quality image # %load snippets/matplot_setup.py plt . rcParams [ 'savefig.dpi' ] = 300 plt . rcParams [ 'figure.dpi' ] = 163 plt . rcParams [ 'figure.autolayout' ] = False plt . rcParams [ 'figure.figsize' ] = 20 , 12 plt . rcParams [ 'axes.labelsize' ] = 18 plt . rcParams [ 'axes.titlesize' ] = 20 plt . rcParams [ 'font.size' ] = 16 plt . rcParams [ 'lines.linewidth' ] = 2.0 plt . rcParams [ 'lines.markersize' ] = 8 plt . rcParams [ 'legend.fontsize' ] = 14 plt . rcParams [ 'text.usetex' ] = False # True activates latex output in fonts! plt . rcParams [ 'font.family' ] = \"serif\" plt . rcParams [ 'font.serif' ] = \"cm\" plt . rcParams [ 'text.latex.preamble' ] = \" \\\\ usepackage{subdepth}, \\\\ usepackage{type1cm}\" You can change the second line in order to fit your display. 163 dpi corresponds to a Dell Ultra HD 4k P2715Q. You can check your screen's dpi count at http://dpi.lv/","tags":"Python","url":"redoules.github.io/python/Setting_up_the_notebook_for_plotting_with_matplotlib.html","loc":"redoules.github.io/python/Setting_up_the_notebook_for_plotting_with_matplotlib.html"},{"title":"Why using a blockchain is a bad idea for your business","text":"What having a blockchain implies? storage costs : everyone maintaining the ledger needs to store every transaction bandwith costs : everyone has to broadcast every transaction computational costs : every node has to validate the blockchain control : the creator does not control the blockchain, everyone collectively controls it developpement costs : developping on a blockchain is way harder than on a traditionnal database What to ask a business when they tell you that they are using a blockchain? When a business is telling you about their innovative technology leveraging the power of the blockchain this should immedialty spake some questions : What is the consensus algorithm? who is responsible for validating the consensus rules? what is the nature of the participation ? is it open to access? is it open to innovation? is it a public ledger? is it transparent? does it improves acountability? is it cross borders? how is it validated?","tags":"Cryptocurrencies","url":"redoules.github.io/cryptocurrencies/blockchain_bad.html","loc":"redoules.github.io/cryptocurrencies/blockchain_bad.html"},{"title":"Synology NFS share","text":"Setting up a NFS share login to your DSM admin account, open the \"Control Panel\" and go to \"File Services\" Make sure NFS is enabled Back in the control panel, go to \"Shared Folder\" Select the folder you want to share and clic \"Edit\" Go to the \"NFS Permissions tab and clic \"Create\", add the ip of the device you want to mount the mapped drive on. Make sure you copy the \"Mount path\"","tags":"Linux","url":"redoules.github.io/linux/share_nfs_share.html","loc":"redoules.github.io/linux/share_nfs_share.html"},{"title":"Mount a NFS share using fstab","text":"Mount nfs using fstab The fstab file, generally located at /etc/fstab lists the differents partitions and where to load them on the filesystem. You can edit this file as root by using the following command sudo nano /etc/fstab in the following example, we want to mount a NFS v3 share from : server : 192.168.1.2 mountpoint (on the server) : /volumeUSB2/usbshare * mountlocation (on the client) : /mnt we specify 192 .168.1.2:/volumeUSB2/usbshare /mnt nfs nfsvers = 3 ,users 0 0 the client will then automatically mount the share ont /mnt at startup. Related you can reload the fstab file using this method : https://redoules.github.io/linux/Reloading_fstab.html You can create a NFS share on a Synology using the method : https://redoules.github.io/linux/share_nfs_share.html","tags":"Linux","url":"redoules.github.io/linux/mount_nfs_share_fstab.html","loc":"redoules.github.io/linux/mount_nfs_share_fstab.html"},{"title":"Installing bitcoind on raspberry pi","text":"Installing bitcoind on linux Running a full bitcoin node helps the bitcoin network to accept, validate and relay transactions. If you want to volunteer some spare computing and bandwidth resources to run a full node and allow Bitcoin to continue to grow you can grab an inexpensive and power efficient raspberry pi and turn it into a full node. There are plenty of tutorials on the Internet explaining how to install a bitcoin full node; this tutorial won't go over setting up a raspberry pi and using ssh. In order to store the full blockchain we will mount a network drive and tell bitcoind to use this mapped drive as the data directory. Download the bitcoin client Go to https://bitcoin.org/en/download Copy the URL for the ARM 32 bit version and download it onto your raspberry pi. wget https://bitcoin.org/bin/bitcoin-core-0.15.1/bitcoin-0.15.1-arm-linux-gnueabihf.tar.gz Locate the downloaded file and extract it using the arguement xzf tar xzf bitcoin-0.15.1-arm-linux-gnueabihf.tar.gz a new directory bitcoin-0.15.1 will be created, it contrains the files we need to install the software Install the bitcoin client We will install the content by copying the binaries located in the bin folder into /usr/local/bin by using the install command. You must use sudo because it will write data to a system directory sudo install -m 0755 -o root -g root -t /usr/local/bin bitcoin-0.15.1/bin/* Launch the bitcoin core client by running bitcoind -daemon Configuration of the node Start your node at boot Starting you node automatically at boot time is a good idea because it doesn't require a manual action from the user. The simplest way to achive this is to create a cronjob. Run the following command crontab -e Select the text editor of your choice, then add the following line at the end of the file @reboot bitcoind -daemon Save the file and exit; the updated crontab file will be installed for you. Full Node If you can afford to download and store all the blockchain, you can run a full node. At the time of writing, the blockchain is 150Go ( https://blockchain.info/fr/charts/blocks-size ). Tree ways to store this are : use a microSD with 256Go or more add a thumbdrive or an external drive to your raspberry pi * mount a network drive from a NAS If you have purchased a big SD card then you can leave the default location for the blockchain data (~/.bitcoin/). Otherwise, you will have to change the datadir location to where your drive is mounted (in my case I have mounted it to /mnt) In order to configure your bitcoin client, edit/create the file bitcoin.conf located in ~/.bitcoin/ nano ~/.bitcoin/bitcoin.conf copy the following text # From redoules.github.io # This config should be placed in following path: # ~/.bitcoin/bitcoin.conf # [core] # Specify a non-default location to store blockchain and other data. datadir=/mnt # Set database cache size in megabytes; machines sync faster with a larger cache. Recommend setting as high as possible based upon mach$ dbcache=100 # Keep at most <n> unconnectable transactions in memory. maxorphantx=10 # Keep the transaction memory pool below <n> megabytess. maxmempool=50 # [network] # Maintain at most N connections to peers. maxconnections=40 # Tries to keep outbound traffic under the given target (in MiB per 24h), 0 = no limit. maxuploadtarget=5000 Check https://jlopp.github.io/bitcoin-core-config-generator it is a handy site to edit the bitcoin.conf file Pruning node If you don't want to store the entire blockchain you can run a pruning node which reduces storage requirements by enabling pruning (deleting) of old blocks. Let's say you want to allocated at most 5Go to the blockchain, then specify prune=5000 into your bitcoin.conf file. Edit/create the file bitcoin.conf located in ~/.bitcoin/ nano ~/.bitcoin/bitcoin.conf copy the following text # From redoules.github.io # This config should be placed in following path: # ~/.bitcoin/bitcoin.conf # [core] # Set database cache size in megabytes; machines sync faster with a larger cache. Recommend setting as high as possible based upon mach$ dbcache=100 # Keep at most <n> unconnectable transactions in memory. maxorphantx=10 # Keep the transaction memory pool below <n> megabytess. maxmempool=50 # Reduce storage requirements by only storing most recent N MiB of block. This mode is incompatible with -txindex and -rescan. WARNING: Reverting this setting requires re-downloading the entire blockchain. (default: 0 = disable pruning blocks, 1 = allow manual pruning via RPC, greater than 550 = automatically prune blocks to stay under target size in MiB). prune=5000 # [network] # Maintain at most N connections to peers. maxconnections=40 # Tries to keep outbound traffic under the given target (in MiB per 24h), 0 = no limit. maxuploadtarget=5000 Checking if your node is public one of the best way to help the bitcoin network is to allow your node to be visible and to propagate block to other nodes. The bitcoin protocole uses port 8333, other clients should be able to share information with your client. Run ifconfig and check if you have an ipv6 adresse (look for adr inet6:) IPV6 Get the global ipv6 adresse of your raspberry pi Link encap:Ethernet HWaddr xx:xx:xx:xx:xx:xx inet adr:192.168.1.x Bcast:192.168.1.255 Masque:255.255.255.0 adr inet6: xxxx::xxxx:xxxx:xxxx:xxxx/64 Scope:Lien adr inet6: xxxx:xxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx/64 Scope:Global UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:42681744 errors:0 dropped:0 overruns:0 frame:0 TX packets:38447218 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 lg file transmission:1000 RX bytes:3044414780 (2.8 GiB) TX bytes:2599878680 (2.4 GiB) it is located between adr inet4 and Scope:Global adr inet6: xxxx:xxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx/64 Scope:Global Copy this adresse and past it into the search field on https://bitnodes.earn.com/ If your node is visible, it will appear on the website IPV4 If you don't have an ipv6 adresse, you will have to open port 8333 on your router and redirect it to the internal IP of your raspberry pi. It is not detailed here because the configuration depends on your router.","tags":"Cryptocurrencies","url":"redoules.github.io/cryptocurrencies/Installing_bitcoind_on_raspberry_pi.html","loc":"redoules.github.io/cryptocurrencies/Installing_bitcoind_on_raspberry_pi.html"},{"title":"Reloading .bashrc","text":"Reload .bashrc The .bashrc file, located at ~/.bashrc allows a user to personalize its bash shell. If you edit this file, the changes won't be loaded without login out and back in. However, you can use the following command to do it source ~/.bashrc","tags":"Linux","url":"redoules.github.io/linux/Reloading_.bashrc.html","loc":"redoules.github.io/linux/Reloading_.bashrc.html"},{"title":"Reloading fstab","text":"Reload fstab The fstab file, generally located at /etc/fstab lists the differents partitions and where to load them on the filesystem. If you edit this file, the changes won't be automounted. You either have to reboot your system of use the following command as root mount -a","tags":"Linux","url":"redoules.github.io/linux/Reloading_fstab.html","loc":"redoules.github.io/linux/Reloading_fstab.html"},{"title":"Updating all python package with anaconda","text":"Updating anaconda packages All packages managed by conda can be updated with the following command : conda update --all Updating other packages with pip For the other packages, the pip package manager can be used. Unfortunately pip hasn't the same update all fonctionnality. import pip from subprocess import call for dist in pip . get_installed_distributions (): print ( \"updating {0}\" . format ( dist )) call ( \"pip install --upgrade \" + dist . project_name , shell = True )","tags":"Python","url":"redoules.github.io/python/updating_all_python_package_with_anaconda.html","loc":"redoules.github.io/python/updating_all_python_package_with_anaconda.html"},{"title":"Saving a matplotlib figure with a high resolution","text":"creating a matplotlib figure #Importing matplotlib % matplotlib inline import matplotlib.pyplot as plt import numpy as np Drawing a figure # Fixing random state for reproducibility np . random . seed ( 19680801 ) mu , sigma = 100 , 15 x = mu + sigma * np . random . randn ( 10000 ) # the histogram of the data n , bins , patches = plt . hist ( x , 50 , normed = 1 , facecolor = 'g' , alpha = 0.75 ) plt . xlabel ( 'Smarts' ) plt . ylabel ( 'Probability' ) plt . title ( 'Histogram of IQ' ) plt . text ( 60 , . 025 , r '$\\mu=100,\\ \\sigma=15$' ) plt . axis ([ 40 , 160 , 0 , 0.03 ]) plt . grid ( True ) plt . show () Saving the figure normally, one would use the following code plt . savefig ( 'filename.png' ) &lt;matplotlib.figure.Figure at 0x2e45e92f400&gt; The figure in then exported to the file \"filename.png\" with a standard resolution. In adittion, you can specify the dpi arg to some scalar value, for example: plt . savefig ( 'filename_hi_dpi.png' , dpi = 300 ) &lt;matplotlib.figure.Figure at 0x2e462164898&gt;","tags":"Python","url":"redoules.github.io/python/Saving_a_matplotlib_figure_with_a_high_resolution.html","loc":"redoules.github.io/python/Saving_a_matplotlib_figure_with_a_high_resolution.html"},{"title":"Iterating over a DataFrame","text":"Create a sample dataframe # Import modules import pandas as pd # Example dataframe raw_data = { 'fruit' : [ 'Banana' , 'Orange' , 'Apple' , 'lemon' , \"lime\" , \"plum\" ], 'color' : [ 'yellow' , 'orange' , 'red' , 'yellow' , \"green\" , \"purple\" ], 'kcal' : [ 89 , 47 , 52 , 15 , 30 , 28 ] } df = pd . DataFrame ( raw_data , columns = [ 'fruit' , 'color' , 'kcal' ]) df .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 Using the iterrows method Pandas DataFrames can return a generator with the iterrrows method. It can then be used to loop over the rows of the DataFrame for index , row in df . iterrows (): print ( \"At line {0} there is a {1} which is {2} and contains {3} kcal\" . format ( index , row [ \"fruit\" ], row [ \"color\" ], row [ \"kcal\" ])) At line 0 there is a Banana which is yellow and contains 89 kcal At line 1 there is a Orange which is orange and contains 47 kcal At line 2 there is a Apple which is red and contains 52 kcal At line 3 there is a lemon which is yellow and contains 15 kcal At line 4 there is a lime which is green and contains 30 kcal At line 5 there is a plum which is purple and contains 28 kcal","tags":"Python","url":"redoules.github.io/python/Iterating_over_a_dataframe.html","loc":"redoules.github.io/python/Iterating_over_a_dataframe.html"},{"title":"Article Recommander","text":"import pandas as pd import numpy as np % matplotlib inline Loading data and preprocessing we first learn the pickled article database. We will be cleaning it and separating the interesting articles from the uninteresting ones. df = pd . read_pickle ( './article.pkl' ) del df [ \"html\" ] del df [ \"image\" ] del df [ \"URL\" ] del df [ \"hash\" ] del df [ \"source\" ] df [ \"label\" ] = df [ \"note\" ] . apply ( lambda x : 0 if x <= 0 else 1 ) df . head ( 5 ) .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } authors note resume texte titre label 0 [Danny Bradbury, Marco Santori, Adam Draper, M... -10.0 Black Market Reloaded, a black market site tha... Black Market Reloaded, a black market site tha... Black Market Reloaded back online after source... 0 1 [Emily Spaven, Stan Higgins, Emilyspaven] 1.0 The UK Home Office believes the government sho... The UK Home Office believes the government sho... Home Office: UK Should Create a Crime-Fighting... 1 2 [Pete Rizzo, Alex Batlin, Yessi Bello Perez, P... -10.0 Though lofty in its ideals, lead developer Dan... A new social messaging app is aiming to disrup... Gems Bitcoin App Lets Users Earn Money From So... 0 3 [Nermin Hajdarbegovic, Stan Higgins, Pete Rizz... 3.0 US satellite service provider DISH Network has... US satellite service provider DISH Network has... DISH Becomes World's Largest Company to Accept... 1 4 [Stan Higgins, Bailey Reutzel, Garrett Keirns,... -10.0 An unidentified 28-year-old man was robbed of ... An unidentified 28-year-old man was robbed of ... Bitcoin Stolen at Gunpoint in New York City Ro... 0 Basic statistics on the dataset let's explore the dataset and extract some numbers : * the number of article liked/disliked df [ \"label\" ] . value_counts () 0 879 1 324 Name: label, dtype: int64 Create the full content column df [ 'full_content' ] = df . titre + ' ' + df . resume #exclude the full texte of the article for the moment df . head ( 1 ) .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } authors note resume texte titre label full_content 0 [Danny Bradbury, Marco Santori, Adam Draper, M... -10.0 Black Market Reloaded, a black market site tha... Black Market Reloaded, a black market site tha... Black Market Reloaded back online after source... 0 Black Market Reloaded back online after source... from sklearn.model_selection import train_test_split training , testing = train_test_split ( df , # The dataset we want to split train_size = 0.75 , # The proportional size of our training set stratify = df . label , # The labels are used for stratification random_state = 400 # Use the same random state for reproducibility ) training . head ( 5 ) .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } authors note resume texte titre label full_content 748 [Jon Brodkin] -10.0 Amazon, Reddit, Mozilla, and other Internet co... Amazon, Reddit, Mozilla, and other Internet co... Amazon and Reddit try to save net neutrality r... 0 Amazon and Reddit try to save net neutrality r... 1183 [Jon Brodkin] -10.0 (The Time Warner involved in this transaction ... A group of mostly Democratic senators led by A... Democrats urge Trump administration to block A... 0 Democrats urge Trump administration to block A... 769 [Joseph Brogan] -10.0 On Twitter, bad news comes at all hours, with ... On Twitter, bad news comes at all hours, with ... Some of the best art on Twitter comes from the... 0 Some of the best art on Twitter comes from the... 57 [Michael Del Castillo, Pete Rizzo, Trond Vidar... -10.0 Publicly traded online travel service Webjet i... Publicly traded online travel service Webjet i... Webjet Ethereum Pilot Targets Hotel Industry's... 0 Webjet Ethereum Pilot Targets Hotel Industry's... 892 [Andrew Cunningham] 10.0 What has changed on the 2017 MacBook, then?\\nI... Andrew Cunningham\\n\\nAndrew Cunningham\\n\\nAndr... Mini-review: The 2017 MacBook could actually b... 1 Mini-review: The 2017 MacBook could actually b... from sklearn.feature_extraction.text import TfidfVectorizer , CountVectorizer from sklearn.svm import LinearSVC , SVC from sklearn.pipeline import Pipeline from sklearn.model_selection import cross_val_predict from utils.plotting import pipeline_performance steps = ( ( 'vectorizer' , TfidfVectorizer ()), ( 'classifier' , LinearSVC ()) ) pipeline = Pipeline ( steps ) predicted_labels = cross_val_predict ( pipeline , training . full_content , training . label ) pipeline_performance ( training . label , predicted_labels ) pipeline = pipeline . fit ( training . titre , training . label ) Accuracy = 80.6% Confusion matrix, without normalization [[624 35] [140 103]] import re from utils.plotting import print_top_features from sklearn.model_selection import GridSearchCV def mask_integers ( s ): return re . sub ( r '\\d+' , 'INTMASK' , s ) steps = ( ( 'vectorizer' , TfidfVectorizer ()), ( 'classifier' , LinearSVC ()) ) pipeline = Pipeline ( steps ) gs_params = { #'vectorizer__use_idf': (True, False), 'vectorizer__lowercase' : [ True , False ], 'vectorizer__stop_words' : [ 'english' , None ], 'vectorizer__ngram_range' : [( 1 , 1 ), ( 1 , 2 ), ( 2 , 2 )], 'vectorizer__preprocessor' : [ mask_integers , None ], 'classifier__C' : np . linspace ( 5 , 20 , 25 ) } gs = GridSearchCV ( pipeline , gs_params , n_jobs = 1 ) gs . fit ( training . full_content , training . label ) print ( gs . best_params_ ) print ( gs . best_score_ ) pipeline1 = gs . best_estimator_ predicted_labels = pipeline1 . predict ( testing . full_content ) pipeline_performance ( testing . label , predicted_labels ) print_top_features ( pipeline1 , n_features = 10 ) aaa = gs . predict ( testing . full_content ) == testing . label aaa = aaa [ testing . label == 1 ] testing [ \"titre\" ] . iloc [ ~ aaa . values ] #pipeline1.predict([\"windows xbox bitcoin\"]) from sklearn.externals import joblib joblib . dump ( pipeline1 , 'classifier.pkl' ) gs . predict ([ 'Google' ]) array([1], dtype=int64) steps = ( ( 'vectorizer' , TfidfVectorizer ()), ( 'classifier' , SVC ()) ) pipeline = Pipeline ( steps ) gs_params = { #'vectorizer__use_idf': (True, False), 'vectorizer__stop_words' : [ 'english' , None ], 'vectorizer__ngram_range' : [( 1 , 1 ), ( 1 , 2 ), ( 2 , 2 )], 'vectorizer__preprocessor' : [ mask_integers , None ], 'classifier__C' : np . linspace ( 5 , 20 , 25 ) } gs = GridSearchCV ( pipeline , gs_params , n_jobs = 1 ) gs . fit ( training . full_content , training . label ) print ( gs . best_params_ ) print ( gs . best_score_ ) pipeline1 = gs . best_estimator_ predicted_labels = pipeline1 . predict ( testing . full_content ) pipeline_performance ( testing . label , predicted_labels ) print_top_features ( pipeline1 , n_features = 10 ) {'classifier__C': 5.0, 'vectorizer__ngram_range': (1, 1), 'vectorizer__preprocessor': &lt;function mask_integers at 0x00000237491B67B8&gt;, 'vectorizer__stop_words': 'english'} 0.711180124224 Accuracy = 71.2% Confusion matrix, without normalization [[153 0] [ 62 0]] --------------------------------------------------------------------------- ValueError Traceback (most recent call last) &lt;ipython-input-9-3e0781e307fb&gt; in &lt;module&gt;() 25 pipeline_performance(testing.label, predicted_labels) 26 ---&gt; 27 print_top_features(pipeline1, n_features=10) C:\\Users\\Guillaume\\Documents\\Code\\recommandation\\utils\\plotting.py in print_top_features(pipeline, vectorizer_name, classifier_name, n_features) 81 def print_top_features(pipeline, vectorizer_name='vectorizer', classifier_name='classifier', n_features=7): 82 vocabulary = np.array(pipeline.named_steps[vectorizer_name].get_feature_names()) ---&gt; 83 coefs = pipeline.named_steps[classifier_name].coef_[0] 84 top_feature_idx = np.argsort(coefs) 85 top_features = vocabulary[top_feature_idx] C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\svm\\base.py in coef_(self) 483 def coef_(self): 484 if self.kernel != 'linear': --&gt; 485 raise ValueError('coef_ is only available when using a ' 486 'linear kernel') 487 ValueError: coef_ is only available when using a linear kernel from sklearn.naive_bayes import BernoulliNB steps = ( ( 'vectorizer' , TfidfVectorizer ()), ( 'classifier' , BernoulliNB ()) ) pipeline2 = Pipeline ( steps ) gs_params = { 'vectorizer__stop_words' : [ 'english' , None ], 'vectorizer__ngram_range' : [( 1 , 1 ), ( 1 , 2 ), ( 2 , 2 )], 'vectorizer__preprocessor' : [ mask_integers , None ], 'classifier__alpha' : np . linspace ( 0 , 1 , 5 ), 'classifier__fit_prior' : [ True , False ] } gs = GridSearchCV ( pipeline2 , gs_params , n_jobs = 1 ) gs . fit ( training . full_content , training . label ) print ( gs . best_params_ ) print ( gs . best_score_ ) pipeline2 = gs . best_estimator_ predicted_labels = pipeline2 . predict ( testing . full_content ) pipeline_performance ( testing . label , predicted_labels ) print_top_features ( pipeline2 , n_features = 10 ) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - {'classifier__alpha': 0.25, 'classifier__fit_prior': True, 'vectorizer__ngram_range': (1, 1), 'vectorizer__preprocessor': &lt;function mask_integers at 0x00000237491B67B8&gt;, 'vectorizer__stop_words': 'english'} 0.805900621118 Accuracy = 78.1% Confusion matrix, without normalization [[140 13] [ 34 28]] Top like features: ['use' 'just' 'year' 'price' 'time' 'Bitcoin' 'bitcoin' 'new' 'The' 'INTMASK'] --- Top dislike features: ['ABBA' 'cable' 'cab' 'byte' 'publication' 'bye' 'publications' 'publicity' 'buyer' 'publicizing'] from sklearn.naive_bayes import MultinomialNB steps = ( ( 'vectorizer' , TfidfVectorizer ()), ( 'classifier' , MultinomialNB ()) ) pipeline3 = Pipeline ( steps ) gs_params = { 'vectorizer__stop_words' : [ 'english' , None ], 'vectorizer__ngram_range' : [( 1 , 1 ), ( 1 , 2 ), ( 2 , 2 )], 'vectorizer__preprocessor' : [ mask_integers , None ], 'classifier__alpha' : np . linspace ( 0 , 1 , 5 ), 'classifier__fit_prior' : [ True , False ] } gs = GridSearchCV ( pipeline3 , gs_params , n_jobs = 1 ) gs . fit ( training . full_content , training . label ) print ( gs . best_params_ ) print ( gs . best_score_ ) pipeline3 = gs . best_estimator_ predicted_labels = pipeline3 . predict ( testing . full_content ) pipeline_performance ( testing . label , predicted_labels ) print_top_features ( pipeline3 , n_features = 10 ) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - {'classifier__alpha': 0.5, 'classifier__fit_prior': False, 'vectorizer__ngram_range': (1, 1), 'vectorizer__preprocessor': &lt;function mask_integers at 0x00000237491B67B8&gt;, 'vectorizer__stop_words': 'english'} 0.80900621118 Accuracy = 79.1% Confusion matrix, without normalization [[141 12] [ 33 29]] Top like features: ['time' 'Google' 'Pro' 'Apple' 'new' 'The' 'Bitcoin' 'price' 'bitcoin' 'INTMASK'] --- Top dislike features: ['ABBA' 'categories' 'catching' 'catalyst' 'catalog' 'casually' 'casts' 'cast' 'cashier' 'ran']","tags":"Machine Learning","url":"redoules.github.io/machine-learning/Source code for the recommandation engine for articles.html","loc":"redoules.github.io/machine-learning/Source code for the recommandation engine for articles.html"}]}
\ No newline at end of file
+{"pages":[{"title":"About Guillaume Redoulès","text":"I am a data scientist and a mechanical engineer working on numerical methods for stress computations in the field of rocket propulsion. Prior to that, I've got a MSc in Computational Fluid Dynamics and aerodynamics from Imperial College London. Email: guillaume.redoules@gadz.org Linkedin: Guillaume Redoulès Curriculum Vitae Experience Thermomecanical method and tools engineer , Ariane Group , 2015 - Present In charge of tools and methods related to thermomecanical computations. Focal point for machine learning. Education MSc Advanced Computational Methods for Aeronautics, Flow Management and Fluid-Structure Interaction , Imperial College London, London. 2013 Dissertation: \"Estimator design for fluid flows\" Fields: Aeronautics, aerodynamics, computational fluid dynamics, numerical methods Arts et Métiers Paristech , France, 2011 Generalist engineering degree Fields: Mechanics, electrical engineering, casting, machining, project management, finance, IT, etc.","tags":"pages","url":"redoules.github.io/pages/about.html","loc":"redoules.github.io/pages/about.html"},{"title":"Computing the Mayer multiple","text":"I've learnt about the Mayer mutliple from The Inverstor Podcast . The Mayer multiple is the ratio of the bitcoin price divided by the 200-day moving average. It is designed to understand the price of bitcoin without taking in account the short term volatility. It helps investors filter out their emotions during a bull run. Let's see how to compute the Mayer mutliple in python. First, we need to import the data, we will use Quandl to download data from coinbase import quandl btc_price_data = quandl . get ( \"BCHARTS/COINBASEEUR\" ) btc_price_data . tail () .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } Open High Low Close Volume (BTC) Volume (Currency) Weighted Price Date 2018-12-12 2966.00 3076.71 2952.05 3026.00 1447.627465 4.372890e+06 3020.728514 2018-12-13 3025.19 3028.06 2861.15 2886.91 2125.242928 6.261750e+06 2946.369017 2018-12-14 2886.91 2919.00 2800.32 2835.50 2527.558347 7.256959e+06 2871.134083 2018-12-15 2835.49 2865.00 2781.47 2830.45 1267.004758 3.568614e+06 2816.575409 2018-12-16 2830.45 2830.45 2830.44 2830.45 0.144249 4.082886e+02 2830.447385 Next we need to compute the 200 days moving average of the price of bitcoin moving_averages = btc_price_data [[ \"Open\" , \"High\" , \"Low\" , \"Close\" ]] . rolling ( window = 200 ) . mean () moving_averages . tail () .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } Open High Low Close Date 2018-12-12 5507.88295 5611.72570 5380.60150 5491.43610 2018-12-13 5491.44155 5595.14440 5363.77840 5474.38560 2018-12-14 5474.39910 5577.89585 5347.27500 5457.99125 2018-12-15 5457.94930 5559.49145 5330.78235 5439.75570 2018-12-16 5439.71660 5540.89425 5313.62955 5422.17700 Finally, we can compute the ratio and plot it. % matplotlib inline import matplotlib.pyplot as plt plt . rcParams [ 'savefig.dpi' ] = 300 plt . rcParams [ 'figure.dpi' ] = 163 plt . rcParams [ 'figure.autolayout' ] = False plt . rcParams [ 'figure.figsize' ] = 20 , 12 plt . rcParams [ 'font.size' ] = 26 mayer_multiple = btc_price_data / moving_averages mayer_multiple [ \"High\" ] . plot () plt . title ( \"Mayer Mutliple over time\" ) plt . ylabel ( \"Mayer Mutliple\" ) plt . xlabel ( \"Time\" ) print ( f \"Mayer multiple {mayer_multiple.iloc[-1]['High']}\" ) print ( f \"Mayer multiple average {mayer_multiple.mean()['High']}\" ) Mayer multiple 0.5108290958630005 Mayer multiple average 1.3789102045356179 Lastly, I wanted to plot the distribution of the Mayer multiple import numpy as np x = mayer_multiple [ \"High\" ] . values x = x [ ~ np . isnan ( x )] n , a , patches = plt . hist ( x , 100 , facecolor = 'green' , alpha = 0.75 , density = True ) plt . axvline ( x = 2.4 , color = \"red\" ) plt . annotate ( 'We are here today' , xy = ( mayer_multiple . iloc [ - 1 ][ \"High\" ], n [( np . abs ( bins - mayer_multiple . iloc [ - 1 ][ \"High\" ])) . argmin ()]), xytext = ( mayer_multiple . iloc [ - 1 ][ \"High\" ] * 3 , n . max () / 2 ), arrowprops = dict ( facecolor = 'black' , shrink = 0.05 ), ) plt . title ( \"Distribution of the Mayer mutliple\" ) plt . plot () []","tags":"Cryptocurrencies","url":"redoules.github.io/cryptocurrencies/mayer_multiple.html","loc":"redoules.github.io/cryptocurrencies/mayer_multiple.html"},{"title":"Load a saved model","text":"A keras model can be loaded from a hdf5 file. Becareful, a keras generated hdf5 can contain either : the model weights (obtained by the .save_weights() method) the model weights and the model architecture (obtained by the .save() method) In our case, we will load both the model and its weights with the load_model function from the keras.models module from keras.models import load_model model = load_model ( \"my_model.h5\" ) type ( model ) keras.engine.sequential.Sequential","tags":"DL","url":"redoules.github.io/dl/keras_load.html","loc":"redoules.github.io/dl/keras_load.html"},{"title":"Get model input shape","text":"Information on the model can be accessed by the .input_shape parameter of the model object from keras.models import load_model model = load_model ( \"my_model.h5\" ) model . input_shape (None, 28, 28)","tags":"DL","url":"redoules.github.io/dl/keras_input.html","loc":"redoules.github.io/dl/keras_input.html"},{"title":"Get model info and number of parameters","text":"Information on the model can be accessed by the .summary() method from keras.models import load_model model = load_model ( \"my_model.h5\" ) model . summary () _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= flatten_2 (Flatten) (None, 784) 0 _________________________________________________________________ dense_3 (Dense) (None, 128) 100480 _________________________________________________________________ dense_4 (Dense) (None, 10) 1290 ================================================================= Total params: 101,770 Trainable params: 101,770 Non-trainable params: 0 _________________________________________________________________","tags":"DL","url":"redoules.github.io/dl/keras_info.html","loc":"redoules.github.io/dl/keras_info.html"},{"title":"Saving the models weights after each epoch","text":"Let's see how we can save the model weights after every epoch. Let's first import some libraries import keras import numpy as np In this example, we will be using the fashion MNIST dataset to do some basic computer vision, where we will train a Keras neural network to classify items of clothing. In order to import the data we will be using the built in function in Keras : keras . datasets . fashion_mnist . load_data () The model is a very simple neural network consisting in 2 fully connected layers. The model loss function is chosen in order to have a multiclass classifier : \"sparse_categorical_crossentropy\" Let's define a simple feedforward network. ##get and preprocess the data fashion_mnist = keras . datasets . fashion_mnist ( train_images , train_labels ), ( test_images , test_labels ) = fashion_mnist . load_data () train_images = train_images / 255.0 test_images = test_images / 255.0 ## define the model model = keras . Sequential ([ keras . layers . Flatten ( input_shape = ( 28 , 28 )), keras . layers . Dense ( 128 , activation = \"relu\" ), keras . layers . Dense ( 10 , activation = \"softmax\" ) ]) model . compile ( optimizer = \"adam\" , loss = \"sparse_categorical_crossentropy\" , metrics = [ \"acc\" ]) In order to automatically save the model weights to a hdf5 format after every epoch we need to import the ModelCheckpoint callback located in keras . callbacks from keras.callbacks import ModelCheckpoint We now need to define the ModelCheckPoint callback, it takes 7 arguments : filepath: string, path to save the model file. monitor: quantity to monitor verbose: verbosity mode, 0 or 1 save_best_only: if save_best_only=True , the latest best model according to the quantity monitored will not be overwritten. mode: one of {auto, min, max} If save_best_only=True , the decision to overwrite the current save file is made based on either the maximization or the minimization of the monitored quantity. save_weights_only: if True, then only the model's weights will be saved ( model.save_weights(filepath) ), else the full model is saved ( model.save(filepath) ). period: Interval (number of epochs) between checkpoints. The callback has to be added to the callbacks list in the fit method. save_to_hdf5 = ModelCheckpoint ( filepath = \"my_model.h5\" , monitor = 'acc' , verbose = 0 , save_best_only = True , save_weights_only = False , mode = 'auto' , period = 1 ) model . fit ( train_images , train_labels , epochs = 5 , callbacks = [ save_to_hdf5 ]) Epoch 1/5 60000/60000 [==============================] - 29s 478us/step - loss: 0.3975 - acc: 0.8575 Epoch 2/5 60000/60000 [==============================] - 98s 2ms/step - loss: 0.3498 - acc: 0.8721 Epoch 3/5 60000/60000 [==============================] - 95s 2ms/step - loss: 0.3213 - acc: 0.8825 Epoch 4/5 60000/60000 [==============================] - 64s 1ms/step - loss: 0.3021 - acc: 0.8887: 6s - loss: Epoch 5/5 60000/60000 [==============================] - 61s 1ms/step - loss: 0.2855 - acc: 0.8953: 4s - loss: 0.284 - ETA: 3s - loss: - ETA: &lt;keras.callbacks.History at 0x1d70aa62f28&gt; import os if \"my_model.h5\" in os . listdir () : print ( \"The model is saved to my_model.h5\" ) else : print ( 'no model saved to disk' ) The model is saved to my_model.h5","tags":"DL","url":"redoules.github.io/dl/model_checkpoint_keras.html","loc":"redoules.github.io/dl/model_checkpoint_keras.html"},{"title":"Logging the training progress in a CSV","text":"Let's see how we can log the progress and various metrics during the training process to a csv file. Let's first import some libraries import keras import numpy as np In this example, we will be using the fashion MNIST dataset to do some basic computer vision, where we will train a Keras neural network to classify items of clothing. In order to import the data we will be using the built in function in Keras : keras . datasets . fashion_mnist . load_data () The model is a very simple neural network consisting in 2 fully connected layers. The model loss function is chosen in order to have a multiclass classifier : \"sparse_categorical_crossentropy\" Let's define a simple feedforward network. ##get and preprocess the data fashion_mnist = keras . datasets . fashion_mnist ( train_images , train_labels ), ( test_images , test_labels ) = fashion_mnist . load_data () train_images = train_images / 255.0 test_images = test_images / 255.0 ## define the model model = keras . Sequential ([ keras . layers . Flatten ( input_shape = ( 28 , 28 )), keras . layers . Dense ( 128 , activation = \"relu\" ), keras . layers . Dense ( 10 , activation = \"softmax\" ) ]) model . compile ( optimizer = \"adam\" , loss = \"sparse_categorical_crossentropy\" , metrics = [ \"accuracy\" , 'mae' ]) In order to stream to a csv file the epoch results and metrics, we define a CSV logger. It is a callback located in keras . callbacks Let's first import it from keras.callbacks import CSVLogger We now need to define the callback by specifiying a file to be written to, the separator and whether to append to the file or erase it every time. The callback has to be added to the callbacks list in the fit method. csv_logger = CSVLogger ( filename = \"my_csv.csv\" , separator = ';' , append = False ) model . fit ( train_images , train_labels , epochs = 5 , callbacks = [ csv_logger ]) Epoch 1/5 60000/60000 [==============================] - 9s 148us/step - loss: 0.5020 - acc: 0.8234 - mean_absolute_error: 4.4200 Epoch 2/5 60000/60000 [==============================] - 8s 138us/step - loss: 0.3765 - acc: 0.8630 - mean_absolute_error: 4.4200 Epoch 3/5 60000/60000 [==============================] - 8s 129us/step - loss: 0.3371 - acc: 0.8789 - mean_absolute_error: 4.4200 Epoch 4/5 60000/60000 [==============================] - 8s 133us/step - loss: 0.3129 - acc: 0.8843 - mean_absolute_error: 4.4200 Epoch 5/5 60000/60000 [==============================] - 9s 151us/step - loss: 0.2952 - acc: 0.8916 - mean_absolute_error: 4.4200 &lt;keras.callbacks.History at 0x1582adc6780&gt; The results are stored in the my_csv.csv file and contain the epoch results import pandas as pd pd . read_csv ( \"my_csv.csv\" , sep = \";\" ) .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } epoch acc loss mean_absolute_error 0 0 0.823400 0.502013 4.42 1 1 0.863050 0.376516 4.42 2 2 0.878867 0.337097 4.42 3 3 0.884317 0.312893 4.42 4 4 0.891583 0.295157 4.42","tags":"DL","url":"redoules.github.io/dl/csv_logger_keras.html","loc":"redoules.github.io/dl/csv_logger_keras.html"},{"title":"Random integer","text":"The random module comes with the randint function that returns a pseudo random number between 2 values : # Return a random integer N such that a <= N <= b. random . randint ( a , b ) #### import random min_int = 126 max_int = 211 print ( f \"My pseudo random number between {min_int} and {max_int} : {random.randint(min_int, max_int)}\" ) My pseudo random number between 126 and 211 : 206","tags":"Python","url":"redoules.github.io/python/randint.html","loc":"redoules.github.io/python/randint.html"},{"title":"Using tensorboard with Keras","text":"general workflow Let's see how we can get tensorboard to work with a Keras-based Tensorflow code. import tensorflow as tf import keras import numpy as np In this example, we will be using the fashion MNIST dataset to do some basic computer vision, where we will train a Keras neural network to classify items of clothing. In order to import the data we will be using the built in function in Keras : keras . datasets . fashion_mnist . load_data () The model is a very simple neural network consisting in 2 fully connected layers. The model loss function is chosen in order to have a multiclass classifier : \"sparse_categorical_crossentropy\" Finally, let's train the model for 5 epochs ##get and preprocess the data fashion_mnist = keras . datasets . fashion_mnist ( train_images , train_labels ), ( test_images , test_labels ) = fashion_mnist . load_data () train_images = train_images / 255.0 test_images = test_images / 255.0 ## define the model model = keras . Sequential ([ keras . layers . Flatten ( input_shape = ( 28 , 28 )), keras . layers . Dense ( 128 , activation = \"relu\" ), keras . layers . Dense ( 10 , activation = \"softmax\" ) ]) model . compile ( optimizer = \"adam\" , loss = \"sparse_categorical_crossentropy\" , metrics = [ \"accuracy\" ]) model . fit ( train_images , train_labels , epochs = 5 ) Epoch 1/5 60000/60000 [==============================] - 9s 143us/step - loss: 0.4939 - acc: 0.8254 Epoch 2/5 60000/60000 [==============================] - 11s 182us/step - loss: 0.3688 - acc: 0.8661 Epoch 3/5 60000/60000 [==============================] - 10s 169us/step - loss: 0.3305 - acc: 0.8798 Epoch 4/5 60000/60000 [==============================] - 21s 350us/step - loss: 0.3079 - acc: 0.8874 Epoch 5/5 60000/60000 [==============================] - 18s 302us/step - loss: 0.2889 - acc: 0.8927 &lt;keras.callbacks.History at 0x235c1bc1be0&gt; During the training we can see the process, including the loss and the accuracy in the output. test_loss , test_acc = model . evaluate ( test_images , test_labels ) print ( f \"Test accuracy : {test_acc}\" ) 10000/10000 [==============================] - 1s 67us/step Test accuracy : 0.8763 When the model finishes training, we get an accuracy of about 87%, and we output some sample predictions predictions = model . predict ( test_images ) print ( predictions [ 0 ]) [1.8075149e-05 3.6810281e-08 6.3094416e-07 5.1111499e-07 1.6264809e-06 3.5973577e-04 1.0840570e-06 3.1453002e-02 1.7062060e-06 9.6816361e-01] This kind of process only gives us minimal information during the training process. Setting up tensorboard To make it easier to understand, debug, and optimize TensorFlow programs, a suite of visualization tools called TensorBoard is included. You can use TensorBoard to visualize your TensorFlow graph, plot quantitative metrics about the execution of your graph, and show additional data like images that pass through it. When TensorBoard is fully configured, it looks like this: Let's start by importing the time library and tensorboard itself. It can be found in tensorflow.python.keras.callbacks. from time import time from tensorflow.python.keras.callbacks import TensorBoard After having imported our data and defined our model, we specify a log directory where the training information will get written to. #keep in mind that we already imported the data and defined the model. tensorboard = TensorBoard ( log_dir = f \"logs/{time()}\" ) Finally, to tell Keras to call back to TensorBoard we refer to the instant of TensorBoard we created. model . compile ( optimizer = \"adam\" , loss = \"sparse_categorical_crossentropy\" , metrics = [ \"accuracy\" ]) Now, we need to execute the TensorBoard command pointing at the log directory previously specified. tensorboard --logdir = logs/ TensorBoard will return a http address TensorBoard 1.12.0 at http://localhost:6006 (Press CTRL+C to quit) Now, if we retrain again, we can take a look in TensorBoard and start investigating the loss and accuracy model . fit ( train_images , train_labels , epochs = 5 , callbacks = [ tensorboard ]) Epoch 1/5 60000/60000 [==============================] - 41s 684us/step - loss: 0.4990 - acc: 0.8241 Epoch 2/5 60000/60000 [==============================] - 49s 812us/step - loss: 0.3765 - acc: 0.8648 Epoch 3/5 60000/60000 [==============================] - 46s 765us/step - loss: 0.3392 - acc: 0.8766 Epoch 4/5 60000/60000 [==============================] - 48s 794us/step - loss: 0.3135 - acc: 0.8836 Epoch 5/5 60000/60000 [==============================] - 49s 813us/step - loss: 0.2971 - acc: 0.8897 &lt;keras.callbacks.History at 0x235be1c76d8&gt; TensorBoard also give access to a dynamic visualization of the graph","tags":"DL","url":"redoules.github.io/dl/tensorboard_keras.html","loc":"redoules.github.io/dl/tensorboard_keras.html"},{"title":"Install keras using conda","text":"Keras with the Tensorflow backend can be installed by running the following conda command conda install -c conda-forge keras tensorflow If you want a Intel CPU optimized version, install tensorflow-mkl conda install -c conda-forge keras tensorflow-mkl A GPU compatible version is also available conda install -c conda-forge keras tensorflow-gpu","tags":"DL","url":"redoules.github.io/dl/keras_install.html","loc":"redoules.github.io/dl/keras_install.html"},{"title":"Day 9 - Multiple Linear Regression","text":"Problem Here is a simple equation: $$Y=a+b_1\\cdot f_1++b_2\\cdot f_2+...++b_m\\cdot f_m$$ $$Y=a+\\sum_{i=1}&#94;m b_i\\cdot f_i$$ for \\((m+1)\\) read constants \\((a,f_1, f_2, ..., f_m)\\) . We can say that the value of \\(Y\\) depends on \\(m\\) features. We study this equation for \\(n\\) different feature sets \\((f_1, f_2, ..., f_m)\\) and records each respective value of \\(Y\\) . If we have \\(q\\) new feature sets, and without accounting for bias and variance trade-offs,what is the value of \\(Y\\) for each of the sets? Python implementation import numpy as np m = 2 n = 7 x_1 = [ 0.18 , 0.89 ] y_1 = 109.85 x_2 = [ 1.0 , 0.26 ] y_2 = 155.72 x_3 = [ 0.92 , 0.11 ] y_3 = 137.66 x_4 = [ 0.07 , 0.37 ] y_4 = 76.17 x_5 = [ 0.85 , 0.16 ] y_5 = 139.75 x_6 = [ 0.99 , 0.41 ] y_6 = 162.6 x_7 = [ 0.87 , 0.47 ] y_7 = 151.77 q_1 = [ 0.49 , 0.18 ] q_2 = [ 0.57 , 0.83 ] q_3 = [ 0.56 , 0.64 ] q_4 = [ 0.76 , 0.18 ] With scikit learn X = np . array ([ x_1 , x_2 , x_3 , x_4 , x_5 , x_6 , x_7 ]) Y = np . array ([ y_1 , y_2 , y_3 , y_4 , y_5 , y_6 , y_7 ]) X_q = np . array ([ q_1 , q_2 , q_3 , q_4 ]) from sklearn import linear_model lm = linear_model . LinearRegression () lm . fit ( X , Y ) lm . predict ( X_q ) array([105.21455835, 142.67095131, 132.93605469, 129.70175405]) without scikit learn (but with numpy) from numpy.linalg import inv #center X_R = X - np . mean ( X , axis = 0 ) a = np . mean ( Y ) Y_R = Y - a #calculate b B = inv ( X_R . T @X_R ) @X_R.T@Y_R #predict X_new_R = X_q - np . mean ( X , axis = 0 ) Y_new_R = X_new_R @B Y_new = Y_new_R + a Y_new array([105.21455835, 142.67095131, 132.93605469, 129.70175405]) if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day9.html","loc":"redoules.github.io/blog/Statistics_10days-day9.html"},{"title":"Multiple Linear Regression","text":"If \\(Y\\) is linearly dependent only on \\(X\\) , then we can use the ordinary least square regression line, \\(\\hat{Y}=a+bX\\) . However, if \\(Y\\) shows linear dependency on \\(m\\) variables \\(X_1\\) , \\(X_2\\) , ..., \\(X_m\\) , then we need to find the values of \\(a\\) and \\(m\\) other constants ( \\(b_1\\) , \\(b_2\\) , ..., \\(b_m\\) ). We can then write the regression equation as: $$\\hat{Y}=a+\\sum_{i=1}&#94;{m}b_iX_i$$ Matrix Form of the Regression Equation Let's consider that \\(Y\\) depends on two variables, \\(X_1\\) and \\(X_2\\) . We write the regression relation as \\(\\hat{Y}=a+b_1X_1+b_2X_2\\) . Consider the following matrix operation: $$\\begin{bmatrix} 1 & X_1 & X_2\\\\ \\end{bmatrix}\\cdot\\begin{bmatrix} a \\\\ b_1\\\\ b_2\\\\ \\end{bmatrix}=a+b_1X_1+b_2X_2$$ We define two matrices, \\(X\\) and \\(B\\) as: $$X=\\begin{bmatrix}1 & X_1 & X_2\\\\\\end{bmatrix}$$ $$B=\\begin{bmatrix}a \\\\b_1\\\\b_2\\\\\\end{bmatrix}$$ Now, we rewrite the regression relation as \\(\\hat{Y}=X\\cdot B\\) . This transforms the regression relation into matrix form. Generalized Matrix Form We will consider that \\(Y\\) shows a linear relationship with \\(m\\) variables, \\(X_1\\) , \\(X_2\\) , ..., \\(X_m\\) . Let's say that we made \\(n\\) observations on different tuples \\((x_1, x_2, ..., x_m)\\) : \\(y_1=a+b_1\\cdot x_{1,1} + b_2\\cdot x_{2,1} + ... + b_m\\cdot x_{m,1}\\) \\(y_2=a+b_2\\cdot x_{1,2} + b_2\\cdot x_{2,2} + ... + b_m\\cdot x_{m,2}\\) \\(...\\) \\(y_n=a+b_n\\cdot x_{1,n} + b_2\\cdot x_{2,n} + ... + b_m\\cdot x_{m,n}\\) Now, we can find the matrices: $$X=\\begin{bmatrix}1 & x_{1,1} & x_{2,1} & x_{3,1} & ... & x_{m,1} \\\\1 & x_{1,2} & x_{2,2} & x_{3,2} & ... & x_{m,2} \\\\1 & x_{1,3} & x_{2,3} & x_{3,3} & ... & x_{m,3} \\\\... & ... & ... & ... & ... & ... \\\\1 & x_{1,n} & x_{2,n} & x_{3,n} & ... & x_{m,n} \\\\\\end{bmatrix}$$ $$Y=\\begin{bmatrix}y_1 \\\\y_2\\\\y_3\\\\...\\\\y_n\\\\\\end{bmatrix}$$ Finding the Matrix B We know that \\(Y=X\\cdot B\\) $$\\Rightarrow X&#94;T\\cdot Y=X&#94;T\\cdot X \\cdot B$$ $$\\Rightarrow (X&#94;T\\cdot X)&#94;{-1}\\cdot X&#94;T \\cdot Y=I\\cdot B$$ $$\\Rightarrow B= (X&#94;T\\cdot X)&#94;{-1}\\cdot X&#94;T \\cdot Y$$ Finding the Value of Y Suppose we want to find the value of for some tuple \\(Y\\) , then \\((x_1, x_2, ..., x_m)\\) , $$Y=\\begin{bmatrix} 1 & x_1 & x_2 & ... & x_m\\\\ \\end{bmatrix}\\cdot B$$ Multiple Regression in Python We can use the fit function in the sklearn.linear_model.LinearRegression class. from sklearn import linear_model x = [[ 5 , 7 ], [ 6 , 6 ], [ 7 , 4 ], [ 8 , 5 ], [ 9 , 6 ]] y = [ 10 , 20 , 60 , 40 , 50 ] lm = linear_model . LinearRegression () lm . fit ( x , y ) a = lm . intercept_ b = lm . coef_ print ( f \"Linear regression coefficients between Y and X : a={a}, b_0={b[0]}, b_1={b[1]}\" ) Linear regression coefficients between Y and X : a=51.953488372092984, b_0=6.65116279069768, b_1=-11.162790697674419 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Machine Learning","url":"redoules.github.io/machine-learning/Multiple_Linear_Regression.html","loc":"redoules.github.io/machine-learning/Multiple_Linear_Regression.html"},{"title":"Least Square Regression Line","text":"Linear Regression If our data shows a linear relationship between \\(X\\) and \\(Y\\) , then the straight line which best describes the relationship is the regression line. The regression line is given by \\(\\hat{Y}\\) =a+bX$. Finding the value of b The value of \\(b\\) can be calculated using either of the following formulae: \\(b=\\frac{n\\sum(x_iy_i)-(\\sum x_i)(\\sum y_i)}{n\\sum(x_i&#94;2)-(\\sum x_i)&#94;2}\\) \\(b=\\rho\\frac{\\sigma_Y}{\\sigma_X}\\) , where \\(\\rho\\) is the Pearson correlation coefficient, \\(\\sigma_X\\) Finding the value of a \\(a=\\bar{y}-b\\cdot\\bar{x}\\) , where \\(\\bar{x}\\) is the mean of \\(X\\) and \\(\\bar{y}\\) is the mean of \\(Y\\) . Coefficient of determination ( \\(R&#94;2\\) ) The coefficient of determination can be computer with : \\(R&#94;2 = \\frac{SSR}{SST}=1-\\frac{SSE}{SST}\\) Where : \\(SST\\) is the total Sum of Squares : \\(SST=\\sum (y_i-\\bar{y})&#94;2\\) \\(SSR\\) is the regression Sum of Squares : \\(SSR=\\sum (\\hat{y_i}-\\bar{y})&#94;2\\) \\(SSE\\) is the error Sum of Squares : \\(SSE=\\sum (\\hat{y_i}-y)&#94;2\\) If \\(SSE\\) is small, we can assume that our fit is good. Linear Regression in Python We can use the fit function in the sklearn.linear_model.LinearRegression class. from sklearn import linear_model import numpy as np xl = [ 1 , 2 , 3 , 4 , 5 ] x = np . asarray ( xl ) . reshape ( - 1 , 1 ) y = [ 2 , 1 , 4 , 3 , 5 ] lm = linear_model . LinearRegression () lm . fit ( x , y ) print ( f 'a = {lm.intercept_}' ) print ( f 'b = {lm.coef_[0]}' ) print ( \"Where Y=a+b*X\" ) a = 0.5999999999999996 b = 0.8000000000000002 Where Y=a+b*X if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Machine Learning","url":"redoules.github.io/machine-learning/LeastSquareRegressionLine.html","loc":"redoules.github.io/machine-learning/LeastSquareRegressionLine.html"},{"title":"Day 8 - Least Square Regression Line","text":"Least Square Regression Line Problem A group of five students enrolls in Statistics immediately after taking a Math aptitude test. Each student's Math aptitude test score, \\(x\\) , and Statistics course grade, \\(y\\) , can be expressed as the following list \\((x,y)\\) of points: \\((95, 85)\\) \\((85, 95)\\) \\((80, 70)\\) \\((70, 65)\\) \\((60, 70)\\) If a student scored an 80 on the Math aptitude test, what grade would we expect them to achieve in Statistics? Determine the equation of the best-fit line using the least squares method, then compute and print the value of \\(y\\) when \\(x=80\\) . X = [ 95 , 85 , 80 , 70 , 60 ] Y = [ 85 , 95 , 70 , 65 , 70 ] n = len ( X ) def cov ( X , Y , n ): x_mean = 1 / n * sum ( X ) y_mean = 1 / n * sum ( Y ) return 1 / n * sum ([( X [ i ] - x_mean ) * ( Y [ i ] - y_mean ) for i in range ( n )]) def stdv ( X , mu_x , n ): return ( sum ([( x - mu_x ) ** 2 for x in X ]) / n ) ** 0.5 def pearson_1 ( X , Y , n ): std_x = stdv ( X , 1 / n * sum ( X ), n ) std_y = stdv ( Y , 1 / n * sum ( Y ), n ) return cov ( X , Y , n ) / ( std_x * std_y ) b = pearson_1 ( X , Y , n ) * stdv ( Y , sum ( Y ) / n , n ) / stdv ( X , sum ( X ) / n , n ) a = sum ( Y ) / n - b * sum ( X ) / n print ( f \"If a student scored 80 on the math test, he would most likely score a {round(a+80*b,3)} in statistics\" ) If a student scored 80 on the math test, he would most likely score a 78.288 in statistics Pearson correlation coefficient Problem The regression line of \\(y\\) on \\(x\\) is \\(3x+4y+8=0\\) , and the regression line of \\(x\\) on \\(y\\) is \\(4x+3y+7=0\\) . What is the value of the Pearson correlation coefficient? Mathematical explanation The initial equation system is : $$ \\left\\{\\begin{array}{ r @{{}={}} r >{{}}c<{{}} r >{{}}c<{{}} r } 3x+4y+8=0 & (1)\\\\ 4x+3y+7=0 & (2)\\\\ \\end{array} \\right. $$ So we can rewrite the 2 lines this way : $$ \\left\\{\\begin{array}{ r @{{}={}} r >{{}}c<{{}} r >{{}}c<{{}} r } y=-2+(\\frac{-3}{4})x & (1)\\\\ x=-\\frac{7}{4}+(-\\frac{3}{4})y & (2)\\\\ \\end{array} \\right. $$ so \\(b_1=-\\frac{3}{4}\\) and \\(b_2=-\\frac{3}{4}\\) When we apply the Pearson's coefficient formula : let \\(p\\) be the pearson coefficient let \\(\\sigma_X\\) be the standard deviation of \\(x\\) let \\(\\sigma_Y\\) be the standard deviation of \\(y\\) We hence have $$ \\left\\{\\begin{array}{ r @{{}={}} r >{{}}c<{{}} r >{{}}c<{{}} r } p=b_1\\left(\\frac{\\sigma_X}{\\sigma_Y}\\right) & (1)\\\\ p=b_2\\left(\\frac{\\sigma_Y}{\\sigma_X}\\right) & (2)\\\\ \\end{array} \\right. $$ by multiplying theses 2 equations together we get $$p&#94;2=b_1\\cdot b_2$$ $$p&#94;2=\\left(-\\frac{3}{4}\\right)\\left(-\\frac{3}{4}\\right)$$ $$p&#94;2=\\left(-\\frac{9}{16}\\right)$$ finally we get \\(p=\\left(-\\frac{3}{4}\\right)\\) or \\(p=\\left(\\frac{3}{4}\\right)\\) Since \\(X\\) and \\(Y\\) are negatively correlated we have \\(p=\\left(-\\frac{3}{4}\\right)\\) if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day8.html","loc":"redoules.github.io/blog/Statistics_10days-day8.html"},{"title":"Spearman's Rank Correlation Coefficient","text":"A rank correlation is any of several statistics that measure an ordinal association—the relationship between rankings of different ordinal variables or different rankings of the same variable, where a \"ranking\" is the assignment of the ordering labels \"first\", \"second\", \"third\", etc. to different observations of a particular variable. A rank correlation coefficient measures the degree of similarity between two rankings, and can be used to assess the significance of the relation between them. We have two random variables \\(X\\) and \\(Y\\) : \\(X=\\{x_i, x_2, x_3, ..., x_n\\}\\) \\(Y=\\{y_i, y_2, y_3, ..., y_n\\}\\) if \\(Rank_X\\) and \\(Rank_Y\\) denote the respective ranks of each data point, then the Spearman's rank correlation coefficient, \\(r_s\\) , is the Pearson correlation coefficient of \\(Rank_X\\) and \\(Rank_Y\\) . What does it means? The Spearman's rank correlation coefficientis is a nonparametric measure of rank correlation (statistical dependence between the rankings of two variables). It assesses how well the relationship between two variables can be described using a monotonic function. The Spearman correlation between two variables is equal to the Pearson correlation between the rank values of those two variables; while Pearson's correlation assesses linear relationships, Spearman's correlation assesses monotonic relationships (whether linear or not). If there are no repeated data values, a perfect Spearman correlation of +1 or −1 occurs when each of the variables is a perfect monotone function of the other. A Spearman correlation of 1 results when the two variables being compared are monotonically related, even if their relationship is not linear. This means that all data-points with greater x-values than that of a given data-point will have greater y-values as well. In contrast, this does not give a perfect Pearson correlation. Example \\(X=\\{0.2, 1.3, 0.2, 1.1, 1.4, 1.5\\}\\) \\(Y=\\{1.9, 2.2, 3.1, 1.2, 2.2, 2.2\\}\\) $$ Rank_X \\quad \\begin{bmatrix} X: & 0.2 & 1.3 & 0.2 & 1.1 & 1.4 & 1.5 \\\\ Rank: & 1 & 3 & 1 & 2 & 4 & 5 \\end{bmatrix} \\quad $$ so, \\(Rank_X = \\{1, 3, 1, 2, 4, 5\\}\\) similarly, \\(Rank_Y=\\{2,3,4,1,3,3\\}\\) \\(r_s\\) equals the Pearson correlation coefficient of \\(Rank_X\\) and \\(Rank_Y\\) , meaning that \\(r=0.158114\\) Special case : \\(X\\) and \\(Y\\) don't contain duplicates $$r_s=1-\\frac{6\\sum d_i&#94;2}{n(n&#94;2-1)}$$ Where, \\(d_i\\) is the difference between the respective values of \\(Rank_X\\) and \\(Rank_Y\\) . if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/spearman.html","loc":"redoules.github.io/mathematics/spearman.html"},{"title":"Pearson correlation coefficient","text":"Covariance This is a measure of how two random variables change together, or the strength of their correlation. Consider two random variables, \\(X\\) and \\(Y\\) , each with \\(n\\) values (i.e., \\(x_1\\) , \\(x_2\\) , \\(...\\) , \\(x_n\\) and \\(y_1\\) , \\(y_2\\) , \\(...\\) , \\(y_n\\) ). The covariance of \\(X\\) and \\(Y\\) can be found using either of the following equivalent formulas: $$cov(X,Y)=\\frac{1}{n}\\sum_{i=1}&#94;{n}(x_i-\\bar{x})\\cdot(y_i-\\bar{y})$$ or $$cov(X,Y)=\\frac{1}{n&#94;2}\\sum_{i=1}&#94;{n}\\sum_{j=1}&#94;{n}\\frac{1}{2}(x_i-x_j)\\cdot(y_i-y_j))$$ $$cov(X,Y)=\\frac{1}{n&#94;2}\\sum_{i}\\sum_{j\\gt i}&#94;{n}(x_i-x_j)\\cdot(y_i-y_j)$$ where, \\(\\bar{x}\\) is the mean of \\(X\\) (or \\(\\mu_X\\) ) and \\(\\bar{y}\\) is the mean of \\(Y\\) (or \\(\\mu_Y\\) ) Pearson correlation coefficient The pearson correlation coefficient, \\(\\rho_{X,Y}\\) , is given by : $$\\rho_{X,Y}=\\frac{cov(X,Y)}{\\sigma_X\\sigma_Y}=\\frac{\\sum_{i}(x_i-\\bar{x})(y_i-\\bar{y})}{n\\sigma_X\\sigma_Y}$$ Here, \\(\\sigma_X\\) is the standard deviation of \\(X\\) and \\(\\sigma_Y\\) is the standard deviation of \\(Y\\) . You may also see \\(\\rho_{X,Y}\\) written as \\(r_{X,Y}\\) . The pearson correlation coefficient is a measure of the linear correlation between two variables X and Y. if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/pearson.html","loc":"redoules.github.io/mathematics/pearson.html"},{"title":"Day 7 - Pearson and spearman correlations","text":"Pearson correlation coefficient Problem Given two n-element data sets, \\(X\\) and \\(Y\\) , calculate the value of the Pearson correlation coefficient. Python implementation Using the formula $$\\rho_{X,Y}=\\frac{cov(X,Y)}{\\sigma_X\\sigma_Y}$$ where $$cov(X,Y)=\\frac{1}{n}\\sum_{i=1}&#94;{n}(x_i-\\bar{x})\\cdot(y_i-\\bar{y})$$ n = 10 X = [ 10 , 9.8 , 8 , 7.8 , 7.7 , 7 , 6 , 5 , 4 , 2 ] Y = [ 200 , 44 , 32 , 24 , 22 , 17 , 15 , 12 , 8 , 4 ] def cov ( X , Y , n ): x_mean = 1 / n * sum ( X ) y_mean = 1 / n * sum ( Y ) return 1 / n * sum ([( X [ i ] - x_mean ) * ( Y [ i ] - y_mean ) for i in range ( n )]) def stdv ( X , mu_x , n ): return ( sum ([( x - mu_x ) ** 2 for x in X ]) / n ) ** 0.5 def pearson_1 ( X , Y , n ): std_x = stdv ( X , 1 / n * sum ( X ), n ) std_y = stdv ( Y , 1 / n * sum ( Y ), n ) return cov ( X , Y , n ) / ( std_x * std_y ) pearson_1 ( X , Y , n ) 0.6124721937208479 Python implementation Using the formula $$\\rho_{X,Y}=\\frac{\\sum_{i}(x_i-\\bar{x})(y_i-\\bar{y})}{n\\sigma_X\\sigma_Y}$$ def pearson_2 ( X , Y , n ): std_x = stdv ( X , 1 / n * sum ( X ), n ) std_y = stdv ( Y , 1 / n * sum ( Y ), n ) x_mean = 1 / n * sum ( X ) y_mean = 1 / n * sum ( Y ) return sum ([( X [ i ] - x_mean ) * ( Y [ i ] - y_mean ) for i in range ( n )]) / ( n * std_x * std_y ) pearson_2 ( X , Y , n ) 0.6124721937208479 Spearman's rank correlation coefficient Problem Given two \\(n\\) -element data sets, \\(X\\) and \\(Y\\) , calculate the value of Spearman's rank correlation coefficient. Python implementation We knwo that in this case, the values in each dataset are unique. Hence we can use the formula : $$r_s=1-\\frac{6\\sum d_i&#94;2}{n(n&#94;2-1)}$$ n = 10 X = [ 10 , 9.8 , 8 , 7.8 , 7.7 , 1.7 , 6 , 5 , 1.4 , 2 ] Y = [ 200 , 44 , 32 , 24 , 22 , 17 , 15 , 12 , 8 , 4 ] def spearman_rank ( X , Y , n ): rank_X = [ sorted ( X ) . index ( v ) + 1 for v in X ] rank_Y = [ sorted ( Y ) . index ( v ) + 1 for v in Y ] d = [( rank_X [ i ] - rank_Y [ i ]) ** 2 for i in range ( n )] return 1 - ( 6 * sum ( d )) / ( n * ( n * n - 1 )) spearman_rank ( X , Y , n ) 0.9030303030303031 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day7.html","loc":"redoules.github.io/blog/Statistics_10days-day7.html"},{"title":"Day 6 - The Central Limit Theorem","text":"Problem 1 A large elevator can transport a maximum of \\(9800\\) kg. Suppose a load of cargo containing \\(49\\) boxes must be transported via the elevator. The box weight of this type of cargo follows a distribution with a mean of \\(\\mu=205\\) kg and a standard deviation of \\(\\sigma=15\\) kg. Based on this information, what is the probability that all boxes can be safely loaded into the freight elevator and transported? Mathematical explanation This problem can be tackled with the central limit theorem. Since the number of boxes is \"large\", the sum of the weight approaches normal distribution with : \\(\\mu' = n\\mu\\) \\(\\sigma'=\\sigma\\sqrt{n}\\) If we want to know the probability of the sum of the mass of all boxes to be under a certain weight, we can compute the cumulative density function : $$P(x<9800) = F_X(9800)$$ max_load = 9800 n = 49 mu = 205 st_dev = 15 import math def cumulative ( x , mean , sd ): return 0.5 * ( 1 + math . erf (( x - mean ) / ( sd * math . sqrt ( 2 )))) mu_group = n * mu st_dev_group = st_dev * math . sqrt ( n ) print ( f \"Probability that all the boxes can be lifted by the elevator : {cumulative(max_load, mu_group, st_dev_group)}\" ) Probability that all the boxes can be lifted by the elevator : 0.009815328628645315 Problem 2 The number of tickets purchased by each student for the University X vs. University Y football game follows a distribution that has a mean of \\(\\mu=2.4\\) and a standard deviation of \\(\\sigma=2.0\\) . A few hours before the game starts, \\(100\\) eager students line up to purchase last-minute tickets. If there are only \\(250\\) tickets left, what is the probability that all students will be able to purchase tickets? Mathematical explanation We want to know if the sum of all the purchases will exceed the total supply of tickets. Since each student buy follows a Normal distribution and that the number of students is relatively high, the probability that all the students will be able to buy a ticket can be computed by applying the central limit theorem. The total number of tickets bought follows a normal distribution of mean \\(\\mu'=n*\\mu\\) and of standard deviation \\(\\sigma'=\\sigma\\sqrt{n}\\) ticket_supply = 250 n_students = 100 mu = 2.4 st_dev = 2 mu_group = n_students * mu st_dev_group = st_dev * math . sqrt ( n_students ) print ( f \"Probability that all the students can purchase tickets : {cumulative(ticket_supply, mu_group, st_dev_group)}\" ) Probability that all the students can purchase tickets : 0.691462461274013 Problem 3 You have a sample of \\(100\\) values from a population with mean \\(\\mu=500\\) and with standard deviation \\(\\sigma=80\\) . Compute the interval that covers the middle \\(95%\\) of the distribution of the sample mean; in other words, compute \\(A\\) and \\(B\\) such that \\(P(A<x<B)=0.95\\) . Use the value of \\(z=1.96\\) . Note that is the z-score. Mathematical explanation The margin of error can be computed with : $$MoE = \\frac{z-\\sigma}{\\sqrt{n}}$$ Knowing this, we can figure out what values for x paint the exact middle of the distribution for 0.95 probability, that means theres a 0.0025 leftover on both sides to make the total of 1. zScore = 1.96 std = 80 n = 100 mean = 500 marginOfError = zScore * std / math . sqrt ( n ); print ( \"A =\" , mean - marginOfError ) print ( \"B =\" , mean + marginOfError ) A = 484.32 B = 515.68 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day6.html","loc":"redoules.github.io/blog/Statistics_10days-day6.html"},{"title":"Day 5 - Poisson and Normal distributions","text":"Poisson Distribution Problem 1 A random variable, \\(X\\) , follows Poisson distribution with mean of 2.5. Find the probability with which the random variable \\(X\\) is equal to 5. Mathematical explanation In this case, the answer is straightforward, we just need to compute the value of the Poisson distribution of mean 2.5 at 5: $$P(\\lambda = 2.5, x=5)=\\frac{\\lambda&#94;ke&#94;{-\\lambda}}{k!}$$ $$P(\\lambda = 2.5, x=5)=\\frac{2.5&#94;5e&#94;{-2.5}}{5!}$$ def factorial ( k ): return 1 if k == 1 else k * factorial ( k - 1 ) from math import exp def poisson ( l , k ): return ( l ** k * exp ( - l )) / factorial ( k ) l = 2.5 k = 5 print ( f 'Probability that a random variable X following a Poisson distribution of mean {l} equals {k} : {round(poisson(l,k),3)}' ) Probability that a random variable X following a Poisson distribution of mean 2.5 equals 5 : 0.067 Problem 2 The manager of a industrial plant is planning to buy a machine of either type \\(A\\) or type \\(B\\) . For each day's operation: The number of repairs, \\(X\\) , that machine \\(A\\) needs is a Poisson random variable with mean 0.88. The daily cost of operating \\(A\\) is \\(C_A=160+40X&#94;2\\) . The number of repairs, \\(Y\\) , that machine \\(B\\) needs is a Poisson random variable with mean 1.55. The daily cost of operating \\(B\\) is \\(C_B=128+40Y&#94;2\\) . Assume that the repairs take a negligible amount of time and the machines are maintained nightly to ensure that they operate like new at the start of each day. What is the expected daily cost for each machine. Mathematical explanation The cost for each machine follows a law that is the square of a Poisson distribution. $$C_Z = a + b*Z&#94;2$$ Since the expectation is a linear operator : $$E[C_Z] = aE[1] + bE[Z&#94;2]$$ Knowing that \\(Z\\) follows a Poisson distribution of mean \\(\\lambda\\) we have : $$E[C_Z] = a+ b(\\lambda + \\lambda&#94;2)$$ averageX = 0.88 averageY = 1.55 CostX = 160 + 40 * ( averageX + averageX ** 2 ) CostY = 128 + 40 * ( averageY + averageY ** 2 ) print ( f 'Expected cost to run machine A : {round(CostX, 3)}' ) print ( f 'Expected cost to run machine A : {round(CostY, 3)}' ) Expected cost to run machine A : 226.176 Expected cost to run machine A : 286.1 Normal Distribution Problem 1 In a certain plant, the time taken to assemble a car is a random variable, \\(X\\) , having a normal distribution with a mean of 20 hours and a standard deviation of 2 hours. What is the probability that a car can be assembled at this plant in: Less than 19.5 hours? Between 20 and 22 hours? Mathematical explanation \\(X\\) is a real-valued random variable following a normal distribution : the probability of assembly the car in less than 19.5 hours is the cumulative distribution function of X evaluated at 19.5: $$P(X\\leq 19.5)=F_X(19.5)$$ For a normal distribution, the cumulative distribution function is : $$\\Phi(x)=\\frac{1}{2}\\left(1+erf\\left(\\frac{x-\\mu}{\\sigma\\sqrt{2}}\\right)\\right)$$ import math def cumulative ( x , mean , sd ): return 0.5 * ( 1 + math . erf (( x - mean ) / ( sd * math . sqrt ( 2 )))) mean = 20 sd = 2 print ( f 'Probability that the car is built in less than 19.5 hours : {round(cumulative(19.5,mean,sd),3)}' ) Probability that the car is built in less than 19.5 hours : 0.401 Similarly, the probability that a car is built between 20 and 22hours can be computed thanks to the cumulative density function: $$P(20\\leq x\\leq 22) = F_X(22)-F_X(20)$$ print ( f 'Probability that the car is built between 20 and 22 hours : {round(cumulative(22,mean,sd)-cumulative(20,mean,sd),3)}' ) Probability that the car is built between 20 and 22 hours : 0.341 Problem 2 The final grades for a Physics exam taken by a large group of students have a mean of \\(\\mu=70\\) and a standard deviation of \\(\\sigma=10\\) . If we can approximate the distribution of these grades by a normal distribution, what percentage of the students: Scored higher than 80 (i.e., have a \\(grade \\gt 80\\) ))? Passed the test (i.e., have a \\(grade \\gt 60\\) )? * Failed the test (i.e., have a \\(grade \\lt 60\\) )? Mathematical explanation Here again, we need to appy the cumulative density function to get the probabilities : Probability that they scored higher than 80 : $$P(X\\gt80) = 1- P(X\\lt80)$$ $$P(X\\gt80) = 1- F_X(80)$$ mean = 70 sd = 10 print ( f 'Probability that the the student scored higher than 80 : {round(1- cumulative(80,mean,sd),3)}' ) Probability that the the student scored higher than 80 : 0.159 Probability that they passed the test : $$P(X\\gt60) = 1- P(X\\lt60)$$ $$P(X\\gt80) = 1- F_X(60)$$ print ( f 'Probability that the the student passed the test : {round(1- cumulative(60,mean,sd),3)}' ) Probability that the the student passed the test : 0.841 Probability that they failed : $$P(X\\lt60) = F_X(60)$$ print ( f 'Probability that the student failed the test: {round(cumulative(60,mean,sd),3)}' ) Probability that the student failed the test: 0.159 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day5.html","loc":"redoules.github.io/blog/Statistics_10days-day5.html"},{"title":"Normal Distribution","text":"Normal Distribution The probability density of normal distribution is: $$\\mathcal{N}(\\mu,\\sigma&#94;2)=\\frac{1}{\\sigma\\sqrt{2\\pi}}e&#94;{-\\frac{(x-\\mu)&#94;2}{2\\sigma&#94;2}}$$ where, \\(\\mu\\) is the mean (or expectation) of the distribution. It is also equal to median and mode of the distribution. \\(\\sigma&#94;2\\) is the variance. * \\(\\sigma\\) is the standard deviation. Standard Normal Distribution If \\(\\mu=0\\) and \\(\\sigma=1\\) , then the normal distribution is known as standard normal distribution: $$\\phi(x)=\\frac{e&#94;{-\\frac{x&#94;2}{2}}}{\\sigma\\sqrt{2\\pi}}$$ Every normal distribution can be represented as standard normal distribution: $$\\mathcal{N}(\\mu,\\sigma&#94;2)=\\frac{1}{\\sigma}\\phi(\\frac{x-\\mu}{\\sigma})$$ Cumulative Probability Consider a real-valued random variable, \\(X\\) . The cumulative distribution function of \\(X\\) (or just the distribution function of \\(X\\) ) evaluated at \\(x\\) is the probability that \\(X\\) will take a value less than or equal to \\(x\\) : $$F_X(x)=P(X\\leq x)$$ also, $$P(a\\leq X\\leq b)=P(a\\lt X\\lt b)=F_X(b)-F_X(a)$$ the cumulative distribution function for a function with normal distribution is: $$\\Phi(x)=\\frac{1}{2}\\left(1+erf\\left(\\frac{x-\\mu}{\\sigma\\sqrt{2}}\\right)\\right)$$ where \\(erf\\) is the error function: $$erf(z)=\\frac{2}{\\sqrt{\\pi}}\\int_0&#94;ze&#94;{-x&#94;2}dx$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/normal.html","loc":"redoules.github.io/mathematics/normal.html"},{"title":"Poisson Distribution","text":"Poisson Experiment Poisson experiment is a statistical experiment that has the following properties: The outcome of each trial is either success or failure. The average number of successes ( \\(\\lambda\\) ) that occurs in a specified region is known. The probability that a success will occur is proportional to the size of the region. The probability that a success will occur in an extremely small region is virtually zero. Poisson Distribution A Poisson random variable is the number of successes that result from a Poisson experiment. The probability distribution of a Poisson random variable is called a Poisson distribution: $$P(k,\\lambda)=\\frac{\\lambda&#94;ke&#94;{-\\lambda}}{k!}$$ where : \\(\\lambda\\) is the average number of successes that occur in a specified region. \\(k\\) is the actual number of successes that occur in a specified region. * \\(P(k,\\lambda)\\) is the Poisson probability, which is the probability of getting exactly \\(k\\) successes when the average number of successes is \\(\\lambda\\) . Example The average number of goals in the soccer world cup is 2.5. The probability that 4 goals are scored is then: $$p(\\lambda=2.5,k=4)=\\frac{2.5&#94;4e&#94;{-2.5}}{4!}=0.133$$ Expectation for the Poisson distribution Consider some Poisson random variable, \\(X\\) . Let \\(E[X]\\) be the expectation of \\(X\\) . Find the value of \\(E[X&#94;2]\\) . Let \\(Var(X)\\) be the variance of \\(X\\) . Recall that if a random variable has a Poisson distribution, then: \\(E[X]=\\lambda\\) \\(Var[X]=\\lambda\\) Now, we'll use the following property of expectation and variance for any random variable, \\(X\\) : $$Var(X)=E[X&#94;2]-(E[X])&#94;2$$ $$E[X&#94;2]=Var(X)+(E[X])&#94;2$$ So, for any random variable having a Poisson distribution, the above result can be rewritten as: $$E[X&#94;2]=\\lambda + \\lambda&#94;2$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/poisson.html","loc":"redoules.github.io/mathematics/poisson.html"},{"title":"Geometric distribution","text":"Negative Binomial Experiment A negative binomial experiment is a statistical experiment that has the following properties: The experiment consists of n repeated trials. The trials are independent. The outcome of each trial is either success (s) or failure (f). \\(P(s)\\) is the same for every trial. The experiment continues until x successes are observed If \\(X\\) is the number of experiments until the \\(x&#94;{th}\\) success occures, then \\(X\\) is a discrete random variable called a negative binomial Negative Binomial Distribution Consider the following probability mass function: $$b&#94;*(x,n,p) = {\\binom{n-1}{x-1}}p&#94;xq&#94;{n-x}$$ The function above is negative binomial and has the following properties: The number of successes to be observed is \\(x\\) . The total number of trials is \\(n\\) . The probability of success of 1 trial is \\(p\\) . The probability of failure of 1 trial \\(q\\) , where \\(q=1-p\\) . \\(b&#94;*(x,n,p)\\) is the negative binomial probability , meaning the probability of having exactly \\(x-1\\) successes out of \\(n-1\\) trials and having \\(x\\) successes after \\(n\\) trials. Geometric Distribution The geometric distribution is a special case of the negative binomial distribution that deals with the number of Bernoulli trials required to get a success (i.e., counting the number of failures before the first success). Recall that \\(X\\) is the number of successes in \\(n\\) independent Bernoulli trials, so for each \\(i\\) (where $1\\leq i\\leq n): $ X_i = \\begin{cases} 1 if the i&#94;{th} trial is a success \\\\ 0 otherwise x \\end{cases} $ The geometric distribution is a negative binomial distribution where the number of successes is 1. We express this with the following formula: $$g(n,p)=q&#94;{n-1}p$$ Example Bob is a high school basketball player. He is a 70% free throw shooter, meaning his probability of making a free throw is 0.7. What is the probability that Bob makes his first free throw on his fifth shot? For this experiment n=5, p=0.7 and q=0.3 So : $$g(n=5, p=0.7)=0.3&#94;4 0.7=0.00567$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/Geometric.html","loc":"redoules.github.io/mathematics/Geometric.html"},{"title":"Binomial distribution","text":"Binomial Experiment A binomial experiment (or Bernoulli trial) is a statistical experiment that has the following properties: The experiment consists of n repeated trials. The trials are independent. The outcome of each trial is either success (s) or failure (f). Binomial Distribution We define a binomial process to be a binomial experiment meeting the following conditions: The number of successes is \\(x\\) . The total number of trials is \\(n\\) . The probability of success of 1 trial is \\(p\\) . The probability of failure of 1 trial \\(q\\) , where \\(q=1-p\\) . \\(b(x,n,p)\\) is the binomial probability , meaning the probability of having exactly \\(x\\) successes out of \\(n\\) trials. The binomial random variable is the number of successes, \\(x\\) , out of \\(n\\) trials. The binomial distribution is the probability distribution for the binomial random variable, given by the following probability mass function: $$b(x,n,p) = \\frac{n!}{x!(n-x)!}p&#94;xq&#94;{n-x}$$ Python code for the Binomial distribution import math def bi_dist ( x , n , p ): b = ( math . factorial ( n ) / ( math . factorial ( x ) * math . factorial ( n - x ))) * ( p ** x ) * (( 1 - p ) ** ( n - x )) return ( b ) Using numpy import numpy as np n = 10 #number of coin toss p = 0.5 #probability samples = 1000 #number of samples s = np . random . binomial ( n , p , samples ) Using the stats module from scipy from scipy.stats import binom n = 10 #number of coin toss p = 0.5 #probability samples = 1000 #number of samples s = binom . rvs ( n , p , size = samples ) Cumulative probabilities A cumulative probability refers to the probability that the value of a random variable falls within a specified range. Frequently, cumulative probabilities refer to the probability that a random variable is less than or equal to a specified value. A fair coin is tossed 10 times. Probability of getting 5 heads The probability of getting heads is: $$b(x=5, n=10, p=0.5)=0.246$$ Probability of getting at least 5 heads The probability of getting at least heads is: $$b(x\\geq 5, n=10, p=0.5)= \\sum_{r=5}&#94;{10} b(x=r, n=10, p=0.5)$$ $$b(x\\geq 5, n=10, p=0.5)= 0.623$$ Probability of getting at most 5 heads The probability of getting at most heads is: $$b(x\\leq 5, n=10, p=0.5)= \\sum_{r=0}&#94;{5} b(x=r, n=10, p=0.5)$$ $$b(x\\leq 5, n=10, p=0.5)= 0.623$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/Binomial.html","loc":"redoules.github.io/mathematics/Binomial.html"},{"title":"Day 4 - Binomial and geometric distributions","text":"Binomial distribution Problem 1 The ratio of boys to girls for babies born in Russia is \\(r=\\frac{N_b}{N_g}=1.09\\) . If there is 1 child born per birth, what proportion of Russian families with exactly 6 children will have at least 3 boys? Mathematical explanation Let's first compute the probability of having a boy : $$p_b=\\frac{N_b}{N_b+N_g}$$ where: \\(N_b\\) is the number of boys \\(N_g\\) is the number of girls * \\(r=\\frac{N_b}{N_g}\\) $$p_b=\\frac{1}{1+\\frac{1}{r}}$$ $$p_b=\\frac{r}{r+1}$$ r = 1.09 p_b = r / ( r + 1 ) print ( f \"The probability of having a boy is p={p_b:3f}\" ) The probability of having a boy is p=0.521531 The probability of getting 3 boys in 6 children is given by : $$b(x=3, n=6, p=p_b)$$ In order to compute the proportion of Russian families with exactly 6 children will have at 3 least boys we need to compute the cumulative probability distribution $$b(x\\geq 3, n=6, p=p_b) = \\sum_{i=3}&#94;{6} b(x\\geq i, n=6, p=p_b)$$ Let's code it ! import math def bi_dist ( x , n , p ): b = ( math . factorial ( n ) / ( math . factorial ( x ) * math . factorial ( n - x ))) * ( p ** x ) * (( 1 - p ) ** ( n - x )) return ( b ) b , p , n = 0 , p_b , 6 for i in range ( 3 , 7 ): b += bi_dist ( i , n , p ) print ( f \"probability of getting at least 3 boys in a family with exactly 6 children : {b:.3f}\" ) probability of getting at least 3 boys in a family with exactly 6 children : 0.696 Problem 2 A manufacturer of metal pistons finds that, 12% on average, of the pistons they manufacture are rejected because they are incorrectly sized. What is the probability that a batch of 10 pistons will contain: No more than 2 rejects? At least 2 rejects? Mathematical explanation On average 12% of the pistons are rejected, this means that a piston has a probability of \\(p_{rejected}=0.12\\) to be rejected. The probability of getting less than 2 faulty pistons in a batch is : $$p(rejet<2) = b(x\\leq 2, n= 10, p=p_{rejected})$$ $$p(rejet<2) = \\sum_{i=0}&#94;{2} b(x\\leq i, n=10, p=p_{rejected})$$ b , p , n = 0 , 12 / 100 , 10 for i in range ( 0 , 3 ): b += bi_dist ( i , n , p ) print ( f \"The probability of getting less than 2 faulty pistons in a batch is : {b:.3f}\" ) The probability of getting less than 2 faulty pistons in a batch is : 0.891 The probability that a batch of 10 pistons will contain at least 2 rejects : $$p(rejet<2) = b(x\\geq 2, n= 10, p=p_{rejected})$$ $$p(rejet<2) = \\sum_{i=2}&#94;{10} b(x\\geq i, n=10, p=p_{rejected})$$ b , p , n = 0 , 12 / 100 , 10 for i in range ( 2 , 11 ): b += bi_dist ( i , n , p ) print ( f \"The probability of getting at least 2 faulty pistons in a batch is : {b:.3f}\" ) The probability of getting at least 2 faulty pistons in a batch is : 0.342 Geometric distribution Problem 1 The probability that a machine produces a defective product is \\(\\frac{1}{3}\\) . What is the probability that the first defect is found during the fith inspection? Mathematical explanation In this case, we will use a geometric distribution to evaluate the probability : \\(n=5\\) \\(p=\\frac{1}{3}\\) Hence, the probability that the first defect is found during the fith inspection is \\(g(n=5,p=1/3)\\) print ( f \"The probability that the first defect is found during the fith inspection is {round(((1-p)**(n-1)) * p, 3)}\" ) The probability that the first defect is found during the fith inspection is 0.038 Problem 2 The probability that a machine produces a defective product is \\(\\frac{1}{3}\\) . What is the probability that the first defect is found during the first 5 inspections? Mathematical explanation In this problem, we need to compute the cumulative distribution function $$p(x \\leq5) = \\sum_{i=1}&#94;{5} g(n=i,p=1/3)$$ p_x5 = 0 p = 1 / 3 n = 5 for i in range ( 1 , n + 1 ): p_x5 += ( 1 - p ) ** ( i - 1 ) * p print ( f \"The probability that the first defect is found during the first 5 inspection is {round(p_x5, 3)}\" ) The probability that the first defect is found during the first 5 inspection is 0.868 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day4.html","loc":"redoules.github.io/blog/Statistics_10days-day4.html"},{"title":"Premutations and combinations","text":"Finding patterns in the possible ways events can occur is very useful in helping us count the number of desirable events in our sample space. Two of the easiest methods for doing this are with permutations (when order matters) and combinations (when order doesn't matter). Permutations An ordered arrangement r of objects from a set, A, of n objects (where \\(0\\lt r \\leq n\\) ) is called an r-element permutation of A. You can also think of this as a permutation of A's elements taken r at a time. The number of r-element permutations of an n-object set is denoted by the following formula: $$_{n}P_{r}=\\frac{n!}{(n-r)!}$$ Combinations An unordered arrangement of r objects from a set, A, of n objects (where \\(0\\lt r \\leq n\\) ) is called an r-element combination of A. You can also think of this as a combination of A's elements taken r at a time. Because the only difference between permutations and combinations is that combinations are unordered, we can easily find the number of r-element combinations by dividing out the permutations (r!): $$_{n}C_{r}=\\frac{_{n}P_{r}}{r!}=\\frac{n!}{r!(n-r)!}$$ When we talk about combinations, we're talking about the number of subsets of size r that can be made from a set of size n. In fact, \\(_{n}C_{r}\\) is often referred to as \"n choose r\", because it's counting the number of r-element combinations that can be chosen from a set of n elements. if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/Premutations_and_combinations.html","loc":"redoules.github.io/mathematics/Premutations_and_combinations.html"},{"title":"Conditional Probability","text":"This is defined as the probability of an event occurring, assuming that one or more other events have already occurred. Two events, A and B are considered to be independent if event A has no effect on the probability of event B (i.e. P(B|A)=P(A)). If events A and B are not independent, then we must consider the probability that both events occur. This can be referred to as the intersection of events A and B, defined as P(A∩B) = P(B|A)P(A). We can then use this definition to find the conditional probability by dividing the probability of the intersection of the two events (A∩B) by the probability of the event that is assumed to have already occurred (event A): $$ P(B|A)=\\frac{P(A\\cap B)}{P(A)}$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/cond_prob.html","loc":"redoules.github.io/mathematics/cond_prob.html"},{"title":"Bayes' Theorem","text":"Let A and B be two events such that P(A|B) denotes the probability of the occurrence of A given that B has occurred and denotes the probability of the occurrence B of given that A has occurred, then: $$ P(A|B)=\\frac{P(B|A)P(A)}{P(B)}$$ $$ P(A|B)=\\frac{P(B|A)P(A)}{P(B|A)P(A)+P(B|A&#94;c)P(A&#94;c)}$$ if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/Bayes.html","loc":"redoules.github.io/mathematics/Bayes.html"},{"title":"Day 3 - Conditionnal probability","text":"Conditionnal probability Problem Suppose a family has 2 children, one of which is a boy. What is the probability that both children are boys? Mathematical explanation Let's look at the possible outcomes : .dataframe tbody tr th:only-of-type { vertical-align: middle; }</p> <div class=\"highlight\"><pre><span></span><span class=\"na\">.dataframe</span> <span class=\"no\">tbody</span> <span class=\"no\">tr</span> <span class=\"no\">th</span> <span class=\"err\">{</span> <span class=\"nl\">vertical-align:</span> <span class=\"nf\">top</span><span class=\"c\">;</span> <span class=\"err\">}</span> <span class=\"na\">.dataframe</span> <span class=\"no\">thead</span> <span class=\"no\">th</span> <span class=\"err\">{</span> <span class=\"nl\">text-align:</span> <span class=\"nf\">right</span><span class=\"c\">;</span> <span class=\"err\">}</span> </pre></div> <p> B G B BB BG G GB GG We know that at least one of the children is a boy, so only \"GG\" is not possible. The event where the family has a new boy is then \"BB\". Hence the probability is : $$\\frac{BB}{BB+GB+BG}=\\frac{1}{3}$$ Draw 2 cards from a deck Problem You 2 draw cards from a standard 52-card deck without replacing them. What is the probability that both cards are of the same suit? Mathematical explanation There are 13 cards of each suit. Draw one card. It can be anything with probability of 1. Now there are 51 cards left and 12 of them are the same suit as the first card you drew. So the chance the second card matches the 1st is \\(\\frac{12}{51}\\) . Drawing marbles Problem A bag contains 3 red marbles and 4 blue marbles. Then, 2 marbles are drawn from the bag, at random, without replacement. If the first marble drawn is red, what is the probability that the second marble is blue? Mathematical explanation On the first draw, the probabilities are the following : we call B the event \"a blue ball is drawn\" and R the event \"a red ball is drawn\" \\(P(B)=\\frac{4}{7}\\) \\(P(R)=\\frac{3}{7}\\) On the second draw, if a red ball has been drawn at first, the probabilities are : \\(P(B|R)=\\frac{4}{6}\\) \\(P(R|R)=\\frac{2}{6}\\) Hence, the probability of drawing a blue ball if the first ball drawn was red is \\(\\frac{1}{3}\\) if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day3.html","loc":"redoules.github.io/blog/Statistics_10days-day3.html"},{"title":"Day 2 - Probability, Compound Event Probability","text":"Basic probability with dices Problem In this challenge, we practice calculating probability. In a single toss of 2 fair (evenly-weighted) six-sided dice, find the probability that their sum will be at most 9. Mathematical explanation A nice way to think about sums-of-two-dice problems is to lay out the sums in a 6-by-6 grid in the obvious manner. .dataframe tbody tr th:only-of-type { vertical-align: middle; }</p> <div class=\"highlight\"><pre><span></span><span class=\"na\">.dataframe</span> <span class=\"no\">tbody</span> <span class=\"no\">tr</span> <span class=\"no\">th</span> <span class=\"err\">{</span> <span class=\"nl\">vertical-align:</span> <span class=\"nf\">top</span><span class=\"c\">;</span> <span class=\"err\">}</span> <span class=\"na\">.dataframe</span> <span class=\"no\">thead</span> <span class=\"no\">th</span> <span class=\"err\">{</span> <span class=\"nl\">text-align:</span> <span class=\"nf\">right</span><span class=\"c\">;</span> <span class=\"err\">}</span> </pre></div> <p> 1 2 3 4 5 6 1 2 3 4 5 6 7 2 3 4 5 6 7 8 3 4 5 6 7 8 9 4 5 6 7 8 9 10 5 6 7 8 9 10 11 6 7 8 9 10 11 12 We see that the identic values are on the same diagonal. The number of elements on the diagonal varies from 1 to 6 and then back to 1. let's call A < x the event : the sum all the 2 tosses is at most x. $$P(A\\leq9)=\\sum_{i=2}&#94;{9} P(A = i)$$ $$P(A\\leq9)=1-P(A\\gt9)$$ $$P(A\\leq9)=1-\\sum_{i=10}&#94;{12} P(A = i)$$ The value of \\(P(A = i) = \\frac{i-1}{36}\\) if \\(i \\leq 7\\) and \\(P(A = i) = \\frac{13-i}{36}\\) hence $$P(A\\leq9)=1-\\sum_{i=10}&#94;{12} \\frac{13-i}{36}$$ $$P(A\\leq9)= 1-\\frac{6}{36}$$ $$P(A\\leq9)= \\frac{5}{6}$$ Let's program it sum ([ 1 for d1 in range ( 1 , 7 ) for d2 in range ( 1 , 7 ) if d1 + d2 <= 9 ]) / 36 0.8333333333333334 More dices Problem In a single toss of 2 fair (evenly-weighted) six-sided dice, find the probability that the values rolled by each die will be different and the two dice have a sum of 6. Mathematical explanation Let's consider 2 events : A and B. A compound event is a combination of 2 or more simple events. If A and B are simple events, then A∪B denotes the occurence of either A or B. A∩B denotes the occurence of A and B together. We denote A the event \"the values of each dice is different\". The opposit event is A' \"the values of each dice is the same\". $$P(A) = 1-P(A')$$ $$P(A)=1-\\frac{6}{36}$$ $$P(A)=\\frac{5}{6}$$ We denote B the event \"the two dice have a sum of 6\", this probability has been computed on the first part of the article : $$P(B)=\\frac{5}{36}$$ The probability of having 2 dice different of sum 6 is : $$P(A|B) = 4/5$$ The probability that both A and B occure is equal to P(A∩B). Since \\(P(A|B)=\\frac{P(A∩B)}{P(B)}\\) $$P(A∩B)=P(B)*P(A|B)$$ $$P(A∩B)=5/36*4/5$$ $$P(A∩B)=1/9$$ Let's program it sum ([ 1 for d1 in range ( 1 , 7 ) for d2 in range ( 1 , 7 ) if ( d1 + d2 == 6 ) and ( d1 != d2 )]) / 36 0.1111111111111111 Compound Event Probability Problem There are 3 urns labeled X, Y, and Z. Urn X contains 4 red balls and 3 black balls. Urn Y contains 5 red balls and 4 black balls. Urn Z contains 4 red balls and 4 black balls. One ball is drawn from each of the urns. What is the probability that, of the 3 balls drawn, are 2 red and is 1 black? Mathematical explanation Let's write the different probabilities: .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } Red ball Black ball Urne X $$\\frac{4}{7}$$ $$\\frac{3}{7}$$ Urne Y $$\\frac{5}{9}$$ $$\\frac{4}{9}$$ Urne Z $$\\frac{1}{2}$$ $$\\frac{1}{2}$$ Addition rule A and B are said to be mutually exclusive or disjoint if they have no events in common (i.e., and A∩B=∅ and P(A∩B)=0. The probability of any of 2 or more events occurring is the union (∪) of events. Because disjoint probabilities have no common events, the probability of the union of disjoint events is the sum of the events' individual probabilities. A and B are said to be collectively exhaustive if their union covers all events in the sample space (i.e., A∪B=S and P(A∪B)=1). This brings us to our next fundamental rule of probability: if 2 events, A and B, are disjoint, then the probability of either event is the sum of the probabilities of the 2 events (i.e., P(A or B) = P(A)+P(B)) Mutliplication rule If the outcome of the first event (A) has no impact on the second event (B), then they are considered to be independent (e.g., tossing a fair coin). This brings us to the next fundamental rule of probability: the multiplication rule. It states that if two events, A and B, are independent, then the probability of both events is the product of the probabilities for each event (i.e., P(A and B)= P(A)xP(B)). The chance of all events occurring in a sequence of events is called the intersection (∩) of those events. The balls drawn from the urns are independant hence : p = P(2 red (R) and 1 back (B)) $$p = P(RRB) + P(RBR) + P(BRR)$$ Each of those 3 probability if equal to the product of the probability of drawing each ball \\(P(RRB) = P(R|X) * P(R|Y) * P(B|Z) = 4/7*5/9*1/2\\) \\(P(RRB) = 20/126\\) \\(P(RBR) = 16/126\\) \\(P(BRR) = 15/126\\) this leads to \\(p = 51/126\\) and finally $$p = \\frac{17}{42}$$ Let's program it X = 3 * [ \"B\" ] + 4 * [ \"R\" ] Y = 4 * [ \"B\" ] + 5 * [ \"R\" ] Z = 4 * [ \"B\" ] + 4 * [ \"R\" ] target = [ \"BRR\" , \"RRB\" , \"RBR\" ] sum ([ 1 for x in X for y in Y for z in Z if x + y + z in target ]) / sum ([ 1 for x in X for y in Y for z in Z ]) 0.40476190476190477 if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day2.html","loc":"redoules.github.io/blog/Statistics_10days-day2.html"},{"title":"Day 1 - Quartiles, Interquartile Range and standard deviation","text":"Quartile Definition A quartile is a type of quantile. The first quartile (Q1) is defined as the middle number between the smallest number and the median of the data set. The second quartile (Q2) is the median of the data. The third quartile (Q3) is the middle value between the median and the highest value of the data set. Implementation in python without using the scientific libraries def median ( l ): l = sorted ( l ) if len ( l ) % 2 == 0 : return ( l [ len ( l ) // 2 ] + l [( len ( l ) // 2 - 1 )]) / 2 else : return l [ len ( l ) // 2 ] def quartiles ( l ): # check the input is not empty if not l : raise StatsError ( 'no data points passed' ) # 1. order the data set l = sorted ( l ) # 2. divide the data set in two halves mid = int ( len ( l ) / 2 ) Q2 = median ( l ) if ( len ( l ) % 2 == 0 ): # even Q1 = median ( l [: mid ]) Q3 = median ( l [ mid :]) else : # odd Q1 = median ( l [: mid ]) # same as even Q3 = median ( l [ mid + 1 :]) return ( Q1 , Q2 , Q3 ) L = [ 3 , 7 , 8 , 5 , 12 , 14 , 21 , 13 , 18 ] Q1 , Q2 , Q3 = quartiles ( L ) print ( f \"Sample : {L} \\n Q1 : {Q1}, Q2 : {Q2}, Q3 : {Q3}\" ) Sample : [3, 7, 8, 5, 12, 14, 21, 13, 18] Q1 : 6.0, Q2 : 12, Q3 : 16.0 Interquartile Range Definition The interquartile range of an array is the difference between its first (Q1) and third (Q3) quartiles. Hence the interquartile range is Q3-Q1 Implementation in python without using the scientific libraries print ( f \"Interquatile range : {Q3-Q1}\" ) Interquatile range : 10.0 Standard deviation Definition The standard deviation (σ) is a measure that is used to quantify the amount of variation or dispersion of a set of data values. A low standard deviation indicates that the data points tend to be close to the mean (also called the expected value) of the set, while a high standard deviation indicates that the data points are spread out over a wider range of values. The standard deviation can be computed with the formula: where µ is the mean : Implementation in python without using the scientific libraries import math X = [ 10 , 40 , 30 , 50 , 20 ] mean = sum ( X ) / len ( X ) X = [( x - mean ) ** 2 for x in X ] std = math . sqrt ( sum ( X ) / len ( X ) ) print ( f \"The distribution {X} has a standard deviation of {std}\" ) The distribution [400.0, 100.0, 0.0, 400.0, 100.0] has a standard deviation of 14.142135623730951","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day1.html","loc":"redoules.github.io/blog/Statistics_10days-day1.html"},{"title":"Counting values in an array","text":"Using lists If you want to count the number of occurences of an element in a list you can use the .count() function of the list object arr = [ 1 , 2 , 3 , 3 , 4 , 5 , 3 , 6 , 7 , 7 ] print ( f 'Array : {arr} \\n ' ) print ( f 'The number 3 appears {arr.count(3)} times in the list' ) print ( f 'The number 7 appears {arr.count(7)} times in the list' ) print ( f 'The number 4 appears {arr.count(4)} times in the list' ) Array : [1, 2, 3, 3, 4, 5, 3, 6, 7, 7] The number 3 appears 3 times in the list The number 7 appears 2 times in the list The number 4 appears 1 times in the list Using collections you can get a dictonnary of the number of occurences of each elements in a list thanks to the collections object like this import collections collections . Counter ( arr ) Counter({1: 1, 2: 1, 3: 3, 4: 1, 5: 1, 6: 1, 7: 2}) Using numpy You can have a simular result with numpy by hacking the unique function import numpy as np arr = np . array ( arr ) unique , counts = np . unique ( arr , return_counts = True ) dict ( zip ( unique , counts )) {1: 1, 2: 1, 3: 3, 4: 1, 5: 1, 6: 1, 7: 2}","tags":"Python","url":"redoules.github.io/python/counting.html","loc":"redoules.github.io/python/counting.html"},{"title":"Building a dictonnary using comprehension","text":"An easy way to create a dictionnary in python is to use the comprehension syntaxe. It can be more expressive hence easier to read. d = { key : value for ( key , value ) in iterable } In the example bellow we use the dictionnary comprehension to build a dictonnary from a source list. iterable = list ( range ( 10 )) d = { str ( value ): value ** 2 for value in iterable } # create a dictionnary linking the string value of a number with the square value of this number print ( d ) {'0': 0, '1': 1, '2': 4, '3': 9, '4': 16, '5': 25, '6': 36, '7': 49, '8': 64, '9': 81} of course, you can use an other iterable an repack it with the comprehension syntaxe. In the following example, we convert a list of tuples in a dictonnary. iterable = [( \"France\" , 67.12e6 ), ( \"UK\" , 66.02e6 ), ( \"USA\" , 325.7e6 ), ( \"China\" , 1386e6 ), ( \"Germany\" , 82.79e6 )] population = { key : value for ( key , value ) in iterable } print ( population ) {'France': 67120000.0, 'UK': 66020000.0, 'USA': 325700000.0, 'China': 1386000000.0, 'Germany': 82790000.0}","tags":"Python","url":"redoules.github.io/python/dict_comprehension.html","loc":"redoules.github.io/python/dict_comprehension.html"},{"title":"Extracting unique values from a list or an array","text":"Using lists An easy way to extract the unique values of a list in python is to convert the list to a set. A set is an unordered collection of items. Every element is unique (no duplicates) and must be immutable. my_list = [ 10 , 20 , 30 , 40 , 20 , 50 , 60 , 40 ] print ( f \"Original List : {my_list}\" ) my_set = set ( my_list ) my_new_list = list ( my_set ) # the set is converted back to a list with the list() function print ( f \"List of unique numbers : {my_new_list}\" ) Original List : [10, 20, 30, 40, 20, 50, 60, 40] List of unique numbers : [40, 10, 50, 20, 60, 30] Using numpy If you are using numpy you can extract the unique values of an array with the unique function builtin numpy: import numpy as np arr = np . array ( my_list ) print ( f 'Initial numpy array : {arr} \\n ' ) unique_arr = np . unique ( arr ) print ( f 'Numpy array with unique values : {unique_arr}' ) Initial numpy array : [10 20 30 40 20 50 60 40] Numpy array with unique values : [10 20 30 40 50 60]","tags":"Python","url":"redoules.github.io/python/unique.html","loc":"redoules.github.io/python/unique.html"},{"title":"Sorting an array","text":"Using lists Python provides an iterator to sort an array sorted() you can use it this way : import random # Random lists from [0-999] interval arr = [ random . randint ( 0 , 1000 ) for r in range ( 10 )] print ( f 'Initial random list : {arr} \\n ' ) reversed_arr = list ( sorted ( arr )) print ( f 'Sorted list : {reversed_arr}' ) Initial random list : [277, 347, 976, 367, 604, 878, 148, 670, 229, 432] Sorted list : [148, 229, 277, 347, 367, 432, 604, 670, 878, 976] it is also possible to use the sort function from the list object # Random lists from [0-999] interval arr = [ random . randint ( 0 , 1000 ) for r in range ( 10 )] print ( f 'Initial random list : {arr} \\n ' ) arr . sort () print ( f 'Sorted list : {arr}' ) Initial random list : [727, 759, 68, 103, 23, 90, 258, 737, 791, 567] Sorted list : [23, 68, 90, 103, 258, 567, 727, 737, 759, 791] Using numpy If you are using numpy you can sort an array by creating a view on the array: import numpy as np arr = np . random . random ( 5 ) print ( f 'Initial random array : {arr} \\n ' ) sorted_arr = np . sort ( arr ) print ( f 'Sorted array : {sorted_arr}' ) Initial random array : [0.40021786 0.13876208 0.19939047 0.46015169 0.43734158] Sorted array : [0.13876208 0.19939047 0.40021786 0.43734158 0.46015169]","tags":"Python","url":"redoules.github.io/python/sorting.html","loc":"redoules.github.io/python/sorting.html"},{"title":"Day 0 - Median, mean, mode and weighted mean","text":"A reminder The median The median is the value separating the higher half from the lower half of a data sample. For a data set, it may be thought of as the middle value. For a continuous probability distribution, the median is the value such that a number is equally likely to fall above or below it. The mean The arithmetic mean (or simply mean) of a sample is the sum of the sampled values divided by the number of items. The mode The mode of a set of data values is the value that appears most often. It is the value x at which its probability mass function takes its maximum value. In other words, it is the value that is most likely to be sampled. Implementation in python without using the scientific libraries def median ( l ): l = sorted ( l ) if len ( l ) % 2 == 0 : return ( l [ len ( l ) // 2 ] + l [( len ( l ) // 2 - 1 )]) / 2 else : return l [ len ( l ) // 2 ] def mean ( l ): return sum ( l ) / len ( l ) def mode ( data ): dico = { x : data . count ( x ) for x in list ( set ( data ))} return sorted ( sorted ( dico . items ()), key = lambda x : x [ 1 ], reverse = True )[ 0 ][ 0 ] L = [ 64630 , 11735 , 14216 , 99233 , 14470 , 4978 , 73429 , 38120 , 51135 , 67060 , 4978 , 73429 ] print ( f \"Sample : {L} \\n Mean : {mean(L)}, Median : {median(L)}, Mode : {mode(L)}\" ) Sample : [64630, 11735, 14216, 99233, 14470, 4978, 73429, 38120, 51135, 67060, 4978, 73429] Mean : 43117.75, Median : 44627.5, Mode : 4978 The weighted average The weighted arithmetic mean is similar to an ordinary arithmetic mean (the most common type of average), except that instead of each of the data points contributing equally to the final average, some data points contribute more than others. data = [ 10 , 40 , 30 , 50 , 20 ] weights = [ 1 , 2 , 3 , 4 , 5 ] sum_X = sum ([ x * w for x , w in zip ( data , weights )]) print ( round (( sum_X / sum ( weights )), 1 )) 32.0","tags":"Blog","url":"redoules.github.io/blog/Statistics_10days-day0.html","loc":"redoules.github.io/blog/Statistics_10days-day0.html"},{"title":"Create a simple bash function","text":"A basic function The synthaxe to define a function is : #!/bin/bash # Basic function my_function () { echo Text displayed by my_function } #once defined, you can use it like so : my_function and it should return user@bash : ./my_function.sh Text displayed by my_function Function with arguments When used, the arguments are specified directly after the function name. Whithin the function they are accessible this the $ symbol followed by the number of the arguement. Hence $1 will take the value of the first arguement, $2 will take the value of the second arguement and so on. #!/bin/bash # Passing arguments to a function say_hello () { echo Hello $1 } say_hello Guillaume and it should return user@bash : ./function_arguements.sh Hello Guillaume Overriding Commands Using the previous example, let's override the echo function in order to make it say hello. To do so, you just need to name the function with the same name as the command you want to replace. When you are calling the original function, make sure you are using the builtin keyword #!/bin/bash # Overriding a function echo () { builtin echo Hello $1 } echo Guillaume user@bash : ./function_arguements.sh Hello Guillaume Returning values Use the keyword return to send back a value to the main program. The returned value will be stored in the $? variable #!/bin/bash # Retruning a value secret_number () { return 126 } secret_number echo The secret number is $? This code should return user@bash : ./retrun_value.sh The secret number is 126","tags":"Linux","url":"redoules.github.io/linux/simple_bash_function.html","loc":"redoules.github.io/linux/simple_bash_function.html"},{"title":"Number of edges in a Complete graph","text":"A complete graph contains \\(\\frac{n(n-1)}{2}\\) edges where \\(n\\) is the number of vertices (or nodes). if (!document.getElementById('mathjaxscript_pelican_#%@#$@#')) { var align = \"center\", indent = \"0em\", linebreak = \"false\"; if (false) { align = (screen.width < 768) ? \"left\" : align; indent = (screen.width < 768) ? \"0em\" : indent; linebreak = (screen.width < 768) ? 'true' : linebreak; } var mathjaxscript = document.createElement('script'); mathjaxscript.id = 'mathjaxscript_pelican_#%@#$@#'; mathjaxscript.type = 'text/javascript'; mathjaxscript.src = 'https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.3/latest.js?config=TeX-AMS-MML_HTMLorMML'; mathjaxscript[(window.opera ? \"innerHTML\" : \"text\")] = \"MathJax.Hub.Config({\" + \" config: ['MMLorHTML.js'],\" + \" TeX: { extensions: ['AMSmath.js','AMSsymbols.js','noErrors.js','noUndefined.js'], equationNumbers: { autoNumber: 'AMS' } },\" + \" jax: ['input/TeX','input/MathML','output/HTML-CSS'],\" + \" extensions: ['tex2jax.js','mml2jax.js','MathMenu.js','MathZoom.js'],\" + \" displayAlign: '\"+ align +\"',\" + \" displayIndent: '\"+ indent +\"',\" + \" showMathMenu: true,\" + \" messageStyle: 'normal',\" + \" tex2jax: { \" + \" inlineMath: [ ['\\\\\\\\(','\\\\\\\\)'] ], \" + \" displayMath: [ ['$$','$$'] ],\" + \" processEscapes: true,\" + \" preview: 'TeX',\" + \" }, \" + \" 'HTML-CSS': { \" + \" styles: { '.MathJax_Display, .MathJax .mo, .MathJax .mi, .MathJax .mn': {color: 'inherit ! important'} },\" + \" linebreaks: { automatic: \"+ linebreak +\", width: '90% container' },\" + \" }, \" + \"}); \" + \"if ('default' !== 'default') {\" + \"MathJax.Hub.Register.StartupHook('HTML-CSS Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax['HTML-CSS'].FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"MathJax.Hub.Register.StartupHook('SVG Jax Ready',function () {\" + \"var VARIANT = MathJax.OutputJax.SVG.FONTDATA.VARIANT;\" + \"VARIANT['normal'].fonts.unshift('MathJax_default');\" + \"VARIANT['bold'].fonts.unshift('MathJax_default-bold');\" + \"VARIANT['italic'].fonts.unshift('MathJax_default-italic');\" + \"VARIANT['-tex-mathit'].fonts.unshift('MathJax_default-italic');\" + \"});\" + \"}\"; (document.body || document.getElementsByTagName('head')[0]).appendChild(mathjaxscript); }","tags":"Mathematics","url":"redoules.github.io/mathematics/Number_edges_Complete_graph.html","loc":"redoules.github.io/mathematics/Number_edges_Complete_graph.html"},{"title":"Reverse an array","text":"Using lists Python provides an iterator to reverse an array reversed() you can use it this way : arr = list ( range ( 5 )) print ( f 'Initial array : {arr} \\n ' ) reversed_arr = list ( reversed ( arr )) print ( f 'Reversed array : {reversed_arr}' ) Initial array : [0, 1, 2, 3, 4] Reversed array : [4, 3, 2, 1, 0] Using numpy If you are using numpy you can reverse an array by creating a view on the array: import numpy as np arr = np . arange ( 5 ) print ( f 'Initial array : {arr} \\n ' ) reversed_arr = arr [:: - 1 ] print ( f 'Reversed array : {reversed_arr}' ) Initial array : [0 1 2 3 4] Reversed array : [4 3 2 1 0]","tags":"Python","url":"redoules.github.io/python/reverse.html","loc":"redoules.github.io/python/reverse.html"},{"title":"Advice for designing your own libraries","text":"Advice for designing your own libraries When designing your own library make sure to think of the following things. I will add new paragraphs to this article as I dicover new good practices. Use standard python objects Try to use standard python objects as much as possible. That way, your library becomes compatible with all the other python libaries. For instance, when I created SAMpy : a library for reading and writing SAMCEF results, it returned dictonnaries, lists and pandas dataframes. Hence the results extracted from SAMCEF where compatible with all the scientific stack of python. Limit the number of functionnalities Following the same logic as before, the objects should do only one thing but do it well. Indeed, having a simple interface will reduce the complexity of your code and make it easier to use your library. Again, with SAMpy, I decided to strictly limit the functionnalities to reading and writing SAMCEF files. Define an exception class for your library You should define your own exceptions in order to make it easier for your users to debug their code thanks to clearer messages that convey more meaning. That way, the user will know if the error comes from your library or something else. Bonus if you group similar exceptions in a hierachy of inerited Exception classes. Example : let's create a Exception related to the age of a person : def check_age ( age ): if age < 0 and age > 130 : raise ValueError If the user inputed an invalid age, the ValueError exception would be thrown. That's fine but imagine you wan't to provide more feedback to your users that don't know the internal of your library. Let's now create a selfexplanatory Exception class AgeInvalidError ( ValueError ): pass def check_age ( age ): if age < 0 and age > 130 : raise AgeInvalidError ( age ) You can also add some helpful text to guide your users along the way: class AgeInvalidError ( ValueError ): print ( \"Age invalid, must be between 0 and 130\" ) pass def check_age ( age ): if age < 0 and age > 130 : raise AgeInvalidError ( age ) If you want to group all the logically linked exceptions, you can create a base class and inherit from it : class BaseAgeInvalidError ( ValueError ): pass class TooYoungError ( BaseAgeInvalidError ): pass class TooOldError ( BaseAgeInvalidError ): pass def check_age ( age ): if age < 0 : raise TooYoungError ( age ) elif age > 130 : raise TooOldError ( age ) Structure your repository You should have a file structure in your repository. It will help other contributers especially future contributers. A nice directory structure for your project should look like this: README.md LICENSE setup.py requirements.txt ./MyPackage ./docs ./tests Some prefer to use reStructured Text, I personnaly prefer Markdown choosealicense.com will help you pick the license to use for your project. For package and distribution management, create a setup.py file a the root of the directory The list of dependencies required to test, build and generate the doc are listed in a pip requirement file placed a the root of the directory and named requirements.txt Put the documentation of your library in the docs directory. Put your tests in the tests directory. Since your tests will need to import your library, I recommend modifying the path to resolve your package property. In order to do so, you can create a context.py file located in the tests directory : import os import sys sys . path . insert ( 0 , os . path . abspath ( os . path . join ( os . path . dirname ( __file__ ), '..' ))) import MyPackage Then within your individual test files you can import your package like so : from .context import MyPackage Finally, your code will go into the MyPackage directory Test your code Once your library is in production, you have to guaranty some level of forward compatibility. Once your interface is defined, write some tests. In the future, when your code is modified, having those tests will make sure that the behaviour of your functions and objects won't be altered. Document your code Of course, you should have a documentation to go along with your library. Make sure to add a lot of commun examples as most users tend to learn from examples. I recommend writing your documentation using Sphinx.","tags":"Python","url":"redoules.github.io/python/design_own_libs.html","loc":"redoules.github.io/python/design_own_libs.html"},{"title":"Safely creating a folder if it doesn't exist","text":"Safely creating a folder if it doesn't exist When you are writing to files in python, if the file doesn't exist it will be created. However, if you are trying to write a file in a directory that doesn't exist, an exception will be returned FileNotFoundError : [ Errno 2 ] No such file or directory : \"directory\" This article will teach you how to make sure the target directory exists. If it doesn't, the function will create that directory. First, let's import os and make sure that the \"test_directory\" doesn't exist import os os . path . exists ( \". \\\\ test_directory\" ) False copy the ensure_dir function in your code. This function will handle the creation of the directory. Credit goes to Parand posted on StackOverflow def ensure_dir ( file_path ): directory = os . path . dirname ( file_path ) if not os . path . exists ( directory ): os . makedirs ( directory ) Let's now use the function and create a folder named \"test_directory\" ensure_dir ( \". \\\\ test_directory\" ) If we test for the existence of the directory, the exists function will now return True os . path . exists ( \". \\\\ test_directory\" ) True","tags":"Python","url":"redoules.github.io/python/ensure_dir.html","loc":"redoules.github.io/python/ensure_dir.html"},{"title":"List all files in a directory","text":"Listing all the files in a directory Let's start with the basics, the most staigthforward way to list all the files in a direcoty is to use a combinaison of the listdir function and isfile form os.path. You can use a list comprehension to store all the results in a list. mypath = \"./test_directory/\" from os import listdir from os.path import isfile , join [ f for f in listdir ( mypath ) if isfile ( join ( mypath , f ))] ['logfile.log', 'myfile.txt', 'super_music.mp3', 'textfile.txt'] Listing all the files of a certain type in a directory similarly, if you want to filter only a certain kind of file based on its extension you can use the endswith method. In the following example, we will filter all the \"txt\" files contained in the directory [ f for f in listdir ( mypath ) if f . endswith ( '.' + \"txt\" )] ['myfile.txt', 'textfile.txt'] Listing all the files matching a pattern in a directory The glob module finds all the pathnames matching a specified pattern according to the rules used by the Unix shell. You can use the *, ?, and character ranges expressed with [] wildcards import glob glob . glob ( \"*.txt\" ) ['myfile.txt'] Listing files recusively If you want to list all files recursively you can select all the sub-directories using the \"**\" wildcard import glob glob . glob ( mypath + '/**/*.txt' , recursive = True ) ['./test_directory\\\\myfile.txt', './test_directory\\\\textfile.txt', './test_directory\\\\subdir1\\\\file_hidden_in_a_sub_direcotry.txt'] Using a regular expression If you'd rather use a regular expression to select the files, the pathlib library provides the rglob function. from pathlib import Path list ( Path ( \"./test_directory/\" ) . rglob ( \"*.[tT][xX][tT]\" )) [WindowsPath('test_directory/myfile.txt'), WindowsPath('test_directory/textfile.txt'), WindowsPath('test_directory/subdir1/file_hidden_in_a_sub_direcotry.txt')] Using regular expressions you can for example select multiple types of files. In the following example, we list all the files that finish either with \"txt\" or with \"log\". list ( Path ( \"./test_directory/\" ) . rglob ( \"*.[tl][xo][tg]\" )) [WindowsPath('test_directory/logfile.log'), WindowsPath('test_directory/myfile.txt'), WindowsPath('test_directory/textfile.txt'), WindowsPath('test_directory/subdir1/file_hidden_in_a_sub_direcotry.txt')]","tags":"Python","url":"redoules.github.io/python/list_files_directory.html","loc":"redoules.github.io/python/list_files_directory.html"},{"title":"Using Dask on infiniband","text":"InfiniBand (abbreviated IB) is a computer-networking communications standard used in high-performance computing that features very high throughput and very low latency. It is used for data interconnect both among and within computers. InfiniBand is also used as either a direct or switched interconnect between servers and storage systems, as well as an interconnect between storage systems. (source Wikipedia). If you want to leverage this high speed network instead of the regular ethernet network, you have to specify to the scheduler that you want to used infiniband as your interface. Assuming that you Infiniband interface is ib0 , you would call the scheduler like this : dask-scheduler --interface ib0 --scheduler-file ./cluster.yaml you would have to call the worker using the same interface : dask-worker --interface ib0 --scheduler-file ./cluster.yaml","tags":"Python","url":"redoules.github.io/python/dask_infiniband.html","loc":"redoules.github.io/python/dask_infiniband.html"},{"title":"Clearing the current cell in the notebook","text":"In python, you can clear the output of a cell by importing the IPython.display module and using the clear_output function from IPython.display import clear_output print ( \"text to be cleared\" ) clear_output () As you can see, the text \"text to be cleared\" is not displayed because the function clear_output has been called afterward","tags":"Jupyter","url":"redoules.github.io/jupyter/clear_cell.html","loc":"redoules.github.io/jupyter/clear_cell.html"},{"title":"What's inside my .bashrc ?","text":"############ # Anaconda # ############ export PATH = \"/station/guillaume/anaconda3/bin: $PATH \" alias python = '/station/guillaume/anaconda3/bin/python' ######### # Alias # ######### alias ding = 'echo -e \"\\a\"' alias calc = 'python -ic \"from __future__ import division; from math import *\"' alias h = \"history|grep \" alias f = \"find . |grep \" alias p = \"ps aux |grep \" alias cdl = \"cd /data/guillaume\" alias cp = \"rsync -avz --progress\" alias grep = \"grep --color=auto\" alias ls = \"ls -hN --color=auto --group-directories-first\" alias ll = \"ls -hal\" alias sv = \"ssh compute_cluster\" alias ms = \"ls\" alias jl = \"jupyter lab\" alias lst = \"jupyter notebook list\" ########################## # bashrc personnalisation# ########################## force_color_prompt = yes export EDITOR = nano export BROWSER = \"firefox '%s' &\" if [ -n \" $force_color_prompt \" ] ; then if [ -x /usr/bin/tput ] && tput setaf 1 > & /dev/null ; then # We have color support; assume it's compliant with Ecma-48 # (ISO/IEC-6429). (Lack of such support is extremely rare, and such # a case would tend to support setf rather than setaf.) color_prompt = yes else color_prompt = fi fi if [ \" $color_prompt \" = yes ] ; then #\\h : hostname #\\u : user #\\w : current working directory #\\d : date #\\t : time yellow = 226 green = 83 pink = 198 blue = 34 PS1 = \"\\[\\033[38;5;22m\\]\\u\\[ $( tput sgr0 ) \\]\\[\\033[38;5;163m\\]@\\[ $( tput sgr0 ) \\]\\[\\033[38;5;22m\\]\\h\\[ $( tput sgr0 ) \\]\\[\\033[38;5;162m\\]:\\[ $( tput sgr0 ) \\]\\[\\033[38;5;172m\\]{\\[ $( tput sgr0 ) \\]\\[\\033[38;5;39m\\]\\w\\[ $( tput sgr0 ) \\]\\[\\033[38;5;172m\\]}\\[ $( tput sgr0 ) \\]\\[\\033[38;5;162m\\]>\\[ $( tput sgr0 ) \\]\"","tags":"Linux","url":"redoules.github.io/linux/bashrc.html","loc":"redoules.github.io/linux/bashrc.html"},{"title":"Efficient extraction of eigenvalues from a list of tensors","text":"When you manipulate FEM results you generally have either a: scalar field, vector field, * tensor field. With tensorial results, it is often useful to extract the eigenvalues in order to find the principal values. I have found that it is easier to store the components of the tensors in a 6 column pandas dataframe (because of the symmetric property of stress and strain tensors) import pandas as pd node = [ 1001 , 1002 , 1003 , 1004 ] #when dealing with FEM results you should remember at which element/node the result is computed (in the example, let's assume that we look at node from 1001 to 1004) tensor1 = [ 1 , 1 , 1 , 0 , 0 , 0 ] #eigen : 1 tensor2 = [ 4 , - 1 , 0 , 2 , 2 , 1 ] #eigen : 5.58443, -1.77931, -0.805118 tensor3 = [ 1 , 6 , 5 , 3 , 3 , 1 ] #eigen : 8.85036, 4.46542, -1.31577 tensor4 = [ 1 , 2 , 3 , 0 , 0 , 0 ] #eigen : 1, 2, 3 df = pd . DataFrame ([ tensor1 , tensor2 , tensor3 , tensor4 ], columns = [ \"XX\" , \"YY\" , \"ZZ\" , \"XY\" , \"XZ\" , \"YZ\" ]) df . index = node df .dataframe tbody tr th:only-of-type { vertical-align: middle; } .dataframe tbody tr th { vertical-align: top; } .dataframe thead th { text-align: right; } XX YY ZZ XY XZ YZ 1001 1 1 1 0 0 0 1002 4 -1 0 2 2 1 1003 1 6 5 3 3 1 1004 1 2 3 0 0 0 If you want to extract the eigenvalues of a tensor with numpy you have to pass a n by n ndarray to the eigenvalue function. In order to avoid having to loop over each node, this oneliner is highly optimized and will help you invert a large number of tensors efficiently. The steps are basically, create a list of n by n values (here n=3) in the right order => reshape it to a list of tensors => pass it to the eigenvals function import numpy as np from numpy import linalg as LA eigenvals = LA . eigvals ( df [[ \"XX\" , \"XY\" , \"XZ\" , \"XY\" , \"YY\" , \"YZ\" , \"XZ\" , \"YZ\" , \"ZZ\" ]] . values . reshape ( len ( df ), 3 , 3 )) eigenvals array([[ 1. , 1. , 1. ], [ 5.58442834, -0.80511809, -1.77931025], [-1.31577211, 8.85035616, 4.46541595], [ 1. , 2. , 3. ]])","tags":"Python","url":"redoules.github.io/python/Efficient_extraction_of_eigenvalues_from_a_list_of_tensors.html","loc":"redoules.github.io/python/Efficient_extraction_of_eigenvalues_from_a_list_of_tensors.html"},{"title":"Optimized numpy random number generation on Intel CPU","text":"Python Intel distribution Make sure you have a python intel distribution. When you startup python you should see somethine like : Python 3.6.2 |Intel Corporation| (default, Aug 15 2017, 11:34:02) [MSC v.1900 64 bit (AMD64)] Type 'copyright', 'credits' or 'license' for more information IPython 6.1.0 -- An enhanced Interactive Python. Type '?' for help. If not, you can force the installation of the intel optimized python with : conda update --all conda config --add channels intel conda install numpy --channel intel --override-channels oh and by the way, make sure you a running an Intel CPU ;) Comparing numpy.random with numpy.random_intel Let's now test both the rand function with and without the Intel optimization import numpy as np from numpy import random , random_intel % timeit np . random . rand ( 10 ** 5 ) 1.06 ms ± 91.5 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each) % timeit np . random_intel . rand ( 10 ** 5 ) 225 µs ± 3.46 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)","tags":"Python","url":"redoules.github.io/python/Optimized_numpy_random_intel.html","loc":"redoules.github.io/python/Optimized_numpy_random_intel.html"},{"title":"How to check Linux process information?","text":"How to check Linux process information (CPU usage, memory, user information, etc.)? You need to use the ps command combined with the grep command. In the example, we want to check the information on the nginx process : ps aux | grep nginx It would return the output : root 9976 0.0 0.0 12272 108 ? S<s Aug12 0:00 nginx: master process /usr/bin/nginx -g pid /run/nginx.pid; daemon on; master_process on; http 16780 0.0 0.0 12384 684 ? S< Aug12 4:11 nginx: worker process http 16781 0.0 0.0 12556 708 ? S< Aug12 0:24 nginx: worker process http 16782 0.0 0.1 12292 744 ? S< Aug12 2:43 nginx: worker process http 16783 0.0 0.1 12276 872 ? S< Aug12 0:24 nginx: worker process admin 17612 0.0 0.1 5120 864 pts/4 S+ 11:22 0:00 grep --color=auto nginx The columns have the following order : USER;PID;%CPU;%MEM;VSZ;RSS;TTY;STAT;START;TIME;COMMAND USER = user owning the process PID = process ID of the process %CPU = It is the CPU time used divided by the time the process has been running. %MEM = ratio of the process's resident set size to the physical memory on the machine VSZ = virtual memory usage of entire process (in KiB) RSS = resident set size, the non-swapped physical memory that a task has used (in KiB) TTY = controlling tty (terminal) STAT = multi-character process state START = starting time or date of the process TIME = cumulative CPU time COMMAND = command with all its arguments Interactive display If you want an interactive display showing in real time the statistics of the running process you can use the top command. If htop is available on you system, use this instead.","tags":"Linux","url":"redoules.github.io/linux/linux_process_information.html","loc":"redoules.github.io/linux/linux_process_information.html"},{"title":"Check the size of a directory","text":"How do you Check the size of a directory in linux? The du command will come handy for this task. Let's say we want to know the size of the directory named recommandations , we would run the following command du -sh recommendations It would return the output : 9.9M recommendations","tags":"Linux","url":"redoules.github.io/linux/directory_size.html","loc":"redoules.github.io/linux/directory_size.html"},{"title":"Check for free disk space","text":"How do you check for free disk space on linux? The df command will come handy for this task. Run the command with the following arguments df -ah Will return in a human readable format for all drives a readout of all your filesystems Filesystem Size Used Avail Use% Mounted on /dev/md0 2.4G 1.3G 1.1G 54% / none 348M 4.0K 348M 1% /dev none 0 0 0 - /dev/pts none 0 0 0 - /proc none 0 0 0 - /sys /tmp 350M 1.3M 349M 1% /tmp /run 350M 3.2M 347M 1% /run /dev/shm 350M 12K 350M 1% /dev/shm /proc/bus/usb 0 0 0 - /proc/bus/usb securityfs 0 0 0 - /sys/kernel/security /dev/md3 1.8T 372G 1.5T 21% /volume2 /dev/vg1000/lv 1.8T 1.5T 340G 82% /volume1 /dev/sdq1 7.4G 3.8G 3.5G 52% /volumeUSB3/usbshare3-1 /dev/sdr 294G 146G 134G 53% /volumeUSB2/usbshare none 0 0 0 - /proc/fs/nfsd none 0 0 0 - /config The free disk space can be read in the Avail column","tags":"Linux","url":"redoules.github.io/linux/free_disk_space_linux.html","loc":"redoules.github.io/linux/free_disk_space_linux.html"},{"title":"Check your current ip address","text":"Check your current ip address Run the command ip addr show and that will give you every information available 4: eth0: <> mtu 1500 group default qlen 1 link/ether 5c:51:4f:41:7a:b1 inet 169.254.33.33/16 brd 169.254.255.255 scope global dynamic valid_lft forever preferred_lft forever inet6 fe80::390a:f69e:1ba2:2121/64 scope global dynamic valid_lft forever preferred_lft forever 3: eth1: <BROADCAST,MULTICAST,UP> mtu 1500 group default qlen 1 link/ether 0a:00:27:00:00:03 inet 192.168.56.1/24 brd 192.168.56.255 scope global dynamic valid_lft forever preferred_lft forever inet6 fe80::84cd:374e:843f:f82f/64 scope global dynamic valid_lft forever preferred_lft forever 15: eth2: <> mtu 1500 group default qlen 1 link/ether 00:ff:d2:8a:19:c3 inet 169.254.40.62/16 brd 169.254.255.255 scope global dynamic valid_lft forever preferred_lft forever inet6 fe80::f199:2ca0:7aff:283e/64 scope global dynamic valid_lft forever preferred_lft forever 1: lo: <LOOPBACK,UP> mtu 1500 group default qlen 1 link/loopback 00:00:00:00:00:00 inet 127.0.0.1/8 brd 127.255.255.255 scope global dynamic valid_lft forever preferred_lft forever inet6 ::1/128 scope global dynamic valid_lft forever preferred_lft forever 5: wifi0: <BROADCAST,MULTICAST,UP> mtu 1500 group default qlen 1 link/ieee802.11 5c:51:4f:41:7a:ad inet 192.168.1.1/24 brd 192.168.1.255 scope global dynamic valid_lft 42720sec preferred_lft 42720sec inet6 fe80::395f:3594:1dc2:57e3/64 scope global dynamic valid_lft forever preferred_lft forever 21: wifi1: <> mtu 1500 group default qlen 1 link/ieee802.11 5c:51:4f:41:7a:ae inet 169.254.12.77/16 brd 169.254.255.255 scope global dynamic valid_lft forever preferred_lft forever inet6 fe80::58d5:630:cbbd:c4d/64 scope global dynamic valid_lft forever preferred_lft forever 12: eth3: <> mtu 1472 group default qlen 1 link/ether 00:00:00:00:00:00:00:e0:00:00:00:00:00:00:00:00 inet6 fe80::100:7f:fffe/64 scope global dynamic valid_lft forever preferred_lft forever 10: eth4: <BROADCAST,MULTICAST,UP> mtu 1500 group default qlen 1 link/ether 22:b7:57:52:5f:ff inet 192.168.42.106/24 brd 192.168.42.255 scope global dynamic valid_lft 6659sec preferred_lft 6659sec inet6 fe80::5110:eb6f:deb0:45c4/64 scope global dynamic valid_lft forever preferred_lft forever You can select only one interface ip addr show eth0 and only the relevant information will be displayed 4: eth0: <> mtu 1500 group default qlen 1 link/ether 5c:51:4f:41:7a:b1 inet 169.254.33.33/16 brd 169.254.255.255 scope global dynamic valid_lft forever preferred_lft forever inet6 fe80::390a:f69e:1ba2:2121/64 scope global dynamic valid_lft forever preferred_lft forever Check your current ip address the old way The ifconfig command will return information regarding your network interfaces. Let's try it: ifconfig eth1 Link encap:Ethernet HWaddr 0a:00:27:00:00:03 inet adr:192.168.56.1 Bcast:192.168.56.255 Masque:255.255.255.0 adr inet6: fe80::84cd:374e:843f:f82f/64 Scope:Global UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 Packets reçus:0 erreurs:0 :0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 Octets reçus:0 (0.0 B) Octets transmis:0 (0.0 B) eth4 Link encap:Ethernet HWaddr 22:b7:57:52:5f:ff inet adr:192.168.42.106 Bcast:192.168.42.255 Masque:255.255.255.0 adr inet6: fe80::5110:eb6f:deb0:45c4/64 Scope:Global UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 Packets reçus:0 erreurs:0 :0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 Octets reçus:0 (0.0 B) Octets transmis:0 (0.0 B) lo Link encap:Boucle locale inet adr:127.0.0.1 Masque:255.0.0.0 adr inet6: ::1/128 Scope:Global UP LOOPBACK RUNNING MTU:1500 Metric:1 Packets reçus:0 erreurs:0 :0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 Octets reçus:0 (0.0 B) Octets transmis:0 (0.0 B) wifi0 Link encap:UNSPEC HWaddr 5C-51-4F-41-7A-AD-00-00-00-00-00-00-00-00-00-00 inet adr:192.168.1.1 Bcast:192.168.1.255 Masque:255.255.255.0 adr inet6: fe80::395f:3594:1dc2:57e3/64 Scope:Global UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 Packets reçus:0 erreurs:0 :0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 Octets reçus:0 (0.0 B) Octets transmis:0 (0.0 B) On the left column you have the list of network adapters. The lo is the local loopback, it is an interface that points to the localhost. Interfaces starting with eth refer to wired connections over ethernet (or sometimes USB in the case of a phone acting as an access point over USB). Interfaces starting with wlan or wifi refer to wireless connections. On the right column you have some information corresponding to the interface such as the IPv4, the IPv6, the mask, some statistics about the interface and so on.","tags":"Linux","url":"redoules.github.io/linux/get_ip_linux.html","loc":"redoules.github.io/linux/get_ip_linux.html"},{"title":"Check the version of the kernel currently running","text":"Check the version of the kernel currently running The uname command will give you the version of the kernel. In order to get a more useful output, type uname -a This will return : the hostname os name kernel release/version architecture * etc. variations If you only want the kernel version you can type uname -v if you only want the kernel release you can type uname -r","tags":"Linux","url":"redoules.github.io/linux/version_kernel.html","loc":"redoules.github.io/linux/version_kernel.html"},{"title":"Running the notebook on a remote server","text":"Jupyter hub With JupyterHub you can create a multi-user Hub which spawns, manages, and proxies multiple instances of the single-user Jupyter notebook server. Project Jupyter created JupyterHub to support many users. The Hub can offer notebook servers to a class of students, a corporate data science workgroup, a scientific research project, or a high performance computing group. However, if you are the only one using the server and you just want a simple way to run the notebook on your server and access it through the web interface on a light client without having to install and configure the jupyter hub, you can do the following. Problem with jupyter notebook On your server, run the command jupyter-notebook you should get something like : [I 11:18:44.514 NotebookApp] Serving notebooks from local directory: /volume2/homes/admin [I 11:18:44.515 NotebookApp] 0 active kernels [I 11:18:44.516 NotebookApp] The Jupyter Notebook is running at: http://localhost:8888/?token=357587e3269b0f20f2b7e1918492890ae7573ac7ef1d2023 [I 11:18:44.516 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation). [W 11:18:44.519 NotebookApp] No web browser found: could not locate runnable browser. [C 11:18:44.520 NotebookApp] Copy/paste this URL into your browser when you connect for the first time, to login with a token: http://localhost:8888/?token=357587e3269b0f20f2b7e1918492890ae7573ac7ef1d2023 and if you try to connect to your server ip (in my example : http://192.168.1.2:8888/?token=357587e3269b0f20f2b7e1918492890ae7573ac7ef1d2023) you will get an \"ERR_CONNECTION_REFUSED\" error. This is because, by default, Jupyter Notebook only accepts connections from localhost. Allowing connexions from other sources From any IP The simplest way to avoid the connection error is to allow the notebook to accept connections from any ip jupyter-notebook --ip = * you will get something like [W 11:26:45.285 NotebookApp] WARNING: The notebook server is listening on all IP addresses and not using encryption. This is not recommended. [I 11:26:45.342 NotebookApp] Serving notebooks from local directory: /volume2/homes/admin [I 11:26:45.342 NotebookApp] 0 active kernels [I 11:26:45.343 NotebookApp] The Jupyter Notebook is running at: http://[all ip addresses on your system]:8888/?token=52af33d628881824968b4031967e8541a27cc28b1720c199 [I 11:26:45.343 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation). [W 11:26:45.346 NotebookApp] No web browser found: could not locate runnable browser. [C 11:26:45.347 NotebookApp] Copy/paste this URL into your browser when you connect for the first time, to login with a token: http://localhost:8888/?token=52af33d628881824968b4031967e8541a27cc28b1720c199 and if you connect form a remote client (192.168.1.1 in my example), the following line will be added to the output : [I 11:26:54.798 NotebookApp] 302 GET /?token=52af33d628881824968b4031967e8541a27cc28b1720c199 (192.168.1.1) 111.17ms note that you should only do that if you are the only one using the server because the connection is not encypted. From a specific IP You can also, explicitly specify the ip of the client jupyter-notebook --ip = 192 .168.1.1 [I 11:44:58.104 NotebookApp] JupyterLab extension loaded from C:\\Users\\Guillaume\\Miniconda3\\lib\\site-packages\\jupyterlab [I 11:44:58.104 NotebookApp] JupyterLab application directory is C:\\Users\\Guillaume\\Miniconda3\\share\\jupyter\\lab [I 11:44:58.244 NotebookApp] Serving notebooks from local directory: C:\\Users\\Guillaume [I 11:44:58.245 NotebookApp] 0 active kernels [I 11:44:58.245 NotebookApp] The Jupyter Notebook is running at: http://192.168.1.1:8888/?token=503576dd8fa87d1f2c416df307e9b900e520b4942e317b32 [I 11:44:58.245 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation). [C 11:44:58.258 NotebookApp] Copy/paste this URL into your browser when you connect for the first time, to login with a token: http://192.168.1.1:8888/?token=503576dd8fa87d1f2c416df307e9b900e520b4942e317b32 [I 11:44:59.083 NotebookApp] Accepting one-time-token-authenticated connection from 192.168.1.1","tags":"Jupyter","url":"redoules.github.io/jupyter/remote_run_notebook.html","loc":"redoules.github.io/jupyter/remote_run_notebook.html"},{"title":"Running multiple calls to a function in parallel with Dask","text":"Dask.distributed is a lightweight library for distributed computing in Python. It allows to create a compute graph. Dask distributed is architectured around 3 parts : the dask-scheduler the dask-worker(s) the dask client Dask architecture The Dask scheduler is a centrally managed, distributed, dynamic task scheduler. It recieves tasks from a/multiple client(s) and spread them across one or multiple dask-worker(s). Dask-scheduler is an event based asynchronous dynamic scheduler, meaning that mutliple clients can submit a list of task to be executed on multiple workers. Internally, the task are represented as a directed acyclic graph. Both new clients and new workers can be connected or disconnected during the execution of the task graph. Tasks can be submited with the function client . submit ( function , * args , ** kwargs ) or by using objects from the dask library such as dask.dataframe, dask.bag or dask.array Setup In this example, we will use a distributed scheduler on a single machine with multiple workers and a single client. We will use the client to submit some tasks to the scheduler. The scheduler will then dispatch those tasks to the workers. The process can be monitored in real time through a web application. For this example, all the computations will be run on a local computer. However dask can scale to a large HPC cluster. First we have to launch the dask-scheduler; from the command line, input dask-scheduler Next, you can load the web dashboard. In order to do so, the scheduler returns the number of the port you have to connect to in the line starting with \"bokeh at :\". The default port is 8787. Since we are running all the programs on the same computer, we just have to login to http://127.0.0.1:8787/status Finally, we have to launch the dask-worker(s). If you want to run the worker(s) on the same computer as the scheduler the type : dask-worker 127 .0.0.1:8786 otherwise, make sure you are inputing the ip address of the computer hosting the dask-scheduler. You can launch as many workers as you want. In this example, we will run 3 workers on the local machine. Use the dask workers within your python code We will now see how to submit multiple calls to a fucntion in parallel on the dask-workers. Import the required libraries and define the function to be executed. import numpy as np import pandas as pd from distributed import Client #function used to do parallel computing on def compute_pi_MonteCarlo ( Nb_Data ): \"\"\" computes the value of pi using the monte carlo method \"\"\" Radius = 1 Nb_Data = int ( round ( Nb_Data )) x = np . random . uniform ( - Radius , Radius , Nb_Data ) y = np . random . uniform ( - Radius , Radius , Nb_Data ) pi_mc = 4 * np . sum ( np . power ( x , 2 ) + np . power ( y , 2 ) < Radius ** 2 ) / Nb_Data err = 100 * np . abs ( pi_mc - np . pi ) / np . pi return [ Nb_Data , pi_mc , err ] In order to connect to the scheduler, we create a client. client = Client ( '127.0.0.1:8786' ) client Client Scheduler: tcp://127.0.0.1:8786 Dashboard: http://127.0.0.1:8787/status Cluster Workers: 3 Cores: 12 Memory: 25.48 GB We submit tasks using the submit method data = [ client . submit ( compute_pi_MonteCarlo , Nb_Data ) for Nb_Data in np . logspace ( 3 , 7 , num = 1200 , dtype = int )] If you look at http://127.0.0.1:8787/status you will see the tasks beeing completed. Once competed, gather the data: data = client . gather ( data ) df = pd . DataFrame ( data ) df . columns = [ \"number of points for MonteCarlo\" , \"value of pi\" , \"error (%)\" ] df . tail () .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } number of points for MonteCarlo value of pi error (%) 1195 9697405 3.141296 0.009454 1196 9772184 3.141058 0.017008 1197 9847540 3.141616 0.000739 1198 9923477 3.141009 0.018574 1199 10000000 3.141032 0.017833 There, we have completed a simple example on how to use dask to run multiple functions in parallel. Full source code: import numpy as np import pandas as pd from distributed import Client #function used to do parallel computing on def compute_pi_MonteCarlo ( Nb_Data ): \"\"\" computes the value of pi using the monte carlo method \"\"\" Radius = 1 Nb_Data = int ( round ( Nb_Data )) x = np . random . uniform ( - Radius , Radius , Nb_Data ) y = np . random . uniform ( - Radius , Radius , Nb_Data ) pi_mc = 4 * np . sum ( np . power ( x , 2 ) + np . power ( y , 2 ) < Radius ** 2 ) / Nb_Data err = 100 * np . abs ( pi_mc - np . pi ) / np . pi return [ Nb_Data , pi_mc , err ] #connect to the scheduler client = Client ( '127.0.0.1:8786' ) #submit tasks data = [ client . submit ( compute_pi_MonteCarlo , Nb_Data ) for Nb_Data in np . logspace ( 3 , 7 , num = 1200 , dtype = int )] #gather the results data = client . gather ( data ) df = pd . DataFrame ( data ) df . columns = [ \"number of points for MonteCarlo\" , \"value of pi\" , \"error (%)\" ] df . tail () A word on the environement variables On Windows, to make sure that you can run dask-scheduler and dask-worker from the command line, you have to add the location of the executable to your path. On linux, you can append the location of the dask-worker and scheduler to the path variable with the command export PATH = $PATH :/path/to/dask","tags":"Python","url":"redoules.github.io/python/dask_distributed_parallelism.html","loc":"redoules.github.io/python/dask_distributed_parallelism.html"},{"title":"Plotting data using log axis","text":"Plotting in log axis with matplotlib import matplotlib.pyplot as plt % matplotlib inline import numpy as np x = np . linspace ( 0.1 , 20 ) y = 20 * np . exp ( - x / 10.0 ) Plotting using the standard function then specifying the axis scale One of the easiest way to plot in a log plot is to specify the plot normally and then specify which axis is to be plotted with a log scale. This can be specified by the function set_xscale or set_yscale # Normal plot fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . plot ( x , y ) ax . grid () plt . show () # Log x axis plot fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . plot ( x , y ) ax . set_xscale ( 'log' ) ax . grid () plt . show () # Log x axis plot fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . plot ( x , y ) ax . set_yscale ( 'log' ) ax . grid () plt . show () # Log x axis plot fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . plot ( x , y ) ax . set_xscale ( 'log' ) ax . set_yscale ( 'log' ) ax . grid () plt . show () Plotting using the matplotlib defined function Matplotlib has the function : semilogx, semilogy and loglog that can help you avoid having to specify the axis scale. # Plot using semilogx fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . semilogx ( x , y ) ax . grid () plt . show () # Plot using semilogy fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . semilogy ( x , y ) ax . grid () plt . show () # Plot using loglog fig = plt . figure () ax = fig . add_subplot ( 1 , 1 , 1 ) ax . loglog ( x , y ) ax . grid () plt . show ()","tags":"Python","url":"redoules.github.io/python/logplot.html","loc":"redoules.github.io/python/logplot.html"},{"title":"Downloading a static webpage with python","text":"If you are using python legacy (aka python 2) first of all, stop ! Furthermore, this method won't work in python legacy # Import modules from urllib.request import urlopen The webpage source code can be downloaded with the command urlopen url = \"http://example.com/\" #create a HTTP request in order to read the page page = urlopen ( url ) . read () The source code will be stored in the variable page as a string print ( page ) b '&lt;!doctype html&gt;\\n&lt;html&gt;\\n&lt;head&gt;\\n &lt;title&gt;Example Domain&lt;/title&gt;\\n\\n &lt;meta charset=\"utf-8\" /&gt;\\n &lt;meta http-equiv=\"Content-type\" content=\"text/html; charset=utf-8\" /&gt;\\n &lt;meta name=\"viewport\" content=\"width=device-width, initial-scale=1\" /&gt;\\n &lt;style type=\"text/css\"&gt;\\n body {\\n background-color: #f0f0f2;\\n margin: 0;\\n padding: 0;\\n font-family: \"Open Sans\", \"Helvetica Neue\", Helvetica, Arial, sans-serif;\\n \\n }\\n div {\\n width: 600px;\\n margin: 5em auto;\\n padding: 50px;\\n background-color: #fff;\\n border-radius: 1em;\\n }\\n a:link, a:visited {\\n color: #38488f;\\n text-decoration: none;\\n }\\n @media (max-width: 700px) {\\n body {\\n background-color: #fff;\\n }\\n div {\\n width: auto;\\n margin: 0 auto;\\n border-radius: 0;\\n padding: 1em;\\n }\\n }\\n &lt;/style&gt; \\n&lt;/head&gt;\\n\\n&lt;body&gt;\\n&lt;div&gt;\\n &lt;h1&gt;Example Domain&lt;/h1&gt;\\n &lt;p&gt;This domain is established to be used for illustrative examples in documents. You may use this\\n domain in examples without prior coordination or asking for permission.&lt;/p&gt;\\n &lt;p&gt;&lt;a href=\"http://www.iana.org/domains/example\"&gt;More information...&lt;/a&gt;&lt;/p&gt;\\n&lt;/div&gt;\\n&lt;/body&gt;\\n&lt;/html&gt;\\n' Additionally, you can beautifulsoup in order to make it easier to work with html from bs4 import BeautifulSoup soup = BeautifulSoup ( page , 'lxml' ) soup . prettify () print ( soup ) & lt ;! DOCTYPE html & gt ; & lt ; html & gt ; & lt ; head & gt ; & lt ; title & gt ; Example Domain & lt ;/ title & gt ; & lt ; meta charset = \"utf-8\" /& gt ; & lt ; meta content = \"text/html; charset=utf-8\" http-equiv = \"Content-type\" /& gt ; & lt ; meta content = \"width=device-width, initial-scale=1\" name = \"viewport\" /& gt ; & lt ; style type = \"text/css\" & gt ; body { background-color : #f0f0f2 ; margin : 0 ; padding : 0 ; font-family : \"Open Sans\" , \"Helvetica Neue\" , Helvetica , Arial , sans-serif ; } div { width : 600 px ; margin : 5 em auto ; padding : 50 px ; background-color : #fff ; border-radius : 1 em ; } a : link , a : visited { color : #38488f ; text-decoration : none ; } @ media ( max-width : 700px ) { body { background-color : #fff ; } div { width : auto ; margin : 0 auto ; border-radius : 0 ; padding : 1 em ; } } & lt ;/ style & gt ; & lt ;/ head & gt ; & lt ; body & gt ; & lt ; div & gt ; & lt ; h1 & gt ; Example Domain & lt ;/ h1 & gt ; & lt ; p & gt ; This domain is established to be used for illustrative examples in documents . You may use this domain in examples without prior coordination or asking for permission .& lt ;/ p & gt ; & lt ; p & gt ;& lt ; a href = \"http://www.iana.org/domains/example\" & gt ; More information ...& lt ;/ a & gt ;& lt ;/ p & gt ; & lt ;/ div & gt ; & lt ;/ body & gt ; & lt ;/ html & gt ;","tags":"Python","url":"redoules.github.io/python/download_page.html","loc":"redoules.github.io/python/download_page.html"},{"title":"Getting stock market data","text":"Start by importing the packages. We will need pandas and the pandas_datareader. # Import modules import pandas as pd from pandas_datareader import data Datareader allows you to import data from the internet. I have found that Quandl and robinhood works the best as a source for stockmarket data. Note that if you want an other type of data (e.g. GDP, inflation, etc.) other sources exist. #import stock from robinhood aapl_robinhood = data . DataReader ( 'AAPL' , 'robinhood' , '1980-01-01' ) aapl_robinhood . head () .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } close_price high_price interpolated low_price open_price session volume symbol begins_at AAPL 2017-08-04 153.996200 154.990700 False 153.306900 153.681100 reg 20559852 2017-08-07 156.379100 156.487400 False 154.272000 154.655900 reg 21870321 2017-08-08 157.629700 159.352900 False 155.847400 156.172300 reg 36205896 2017-08-09 158.594700 158.801500 False 156.674500 156.822200 reg 26131530 2017-08-10 153.543100 158.169600 False 152.861000 158.070700 reg 40804273 #import stock from quandl aapl_quandl = data . DataReader ( 'AAPL' , 'quandl' , '1980-01-01' ) aapl_quandl . head () .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } Open High Low Close Volume ExDividend SplitRatio AdjOpen AdjHigh AdjLow AdjClose AdjVolume Date 2018-03-27 173.68 175.15 166.92 168.340 38962839.0 0.0 1.0 173.68 175.15 166.92 168.340 38962839.0 2018-03-26 168.07 173.10 166.44 172.770 36272617.0 0.0 1.0 168.07 173.10 166.44 172.770 36272617.0 2018-03-23 168.39 169.92 164.94 164.940 40248954.0 0.0 1.0 168.39 169.92 164.94 164.940 40248954.0 2018-03-22 170.00 172.68 168.60 168.845 41051076.0 0.0 1.0 170.00 172.68 168.60 168.845 41051076.0 2018-03-21 175.04 175.09 171.26 171.270 35247358.0 0.0 1.0 175.04 175.09 171.26 171.270 35247358.0","tags":"Python","url":"redoules.github.io/python/stock_pandas.html","loc":"redoules.github.io/python/stock_pandas.html"},{"title":"Moving average with pandas","text":"# Import modules import pandas as pd from pandas_datareader import data , wb #import packages from pandas_datareader import data aapl = data . DataReader ( 'AAPL' , 'quandl' , '1980-01-01' ) aapl . head () .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } Open High Low Close Volume ExDividend SplitRatio AdjOpen AdjHigh AdjLow AdjClose AdjVolume Date 2018-03-27 173.68 175.15 166.92 168.340 38962839.0 0.0 1.0 173.68 175.15 166.92 168.340 38962839.0 2018-03-26 168.07 173.10 166.44 172.770 36272617.0 0.0 1.0 168.07 173.10 166.44 172.770 36272617.0 2018-03-23 168.39 169.92 164.94 164.940 40248954.0 0.0 1.0 168.39 169.92 164.94 164.940 40248954.0 2018-03-22 170.00 172.68 168.60 168.845 41051076.0 0.0 1.0 170.00 172.68 168.60 168.845 41051076.0 2018-03-21 175.04 175.09 171.26 171.270 35247358.0 0.0 1.0 175.04 175.09 171.26 171.270 35247358.0 In order to computer the moving average, we will use the rolling function. #120 days moving average moving_averages = aapl [[ \"Open\" , \"High\" , \"Low\" , \"Close\" , \"Volume\" ]] . rolling ( window = 120 ) . mean () moving_averages . tail () .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } Open High Low Close Volume Date 1980-12-18 28.457667 28.551917 28.385000 28.385000 139495.000000 1980-12-17 28.410750 28.502917 28.338083 28.338083 141772.500000 1980-12-16 28.362833 28.453917 28.289167 28.289167 141256.666667 1980-12-15 28.335750 28.426833 28.262083 28.262083 144321.666667 1980-12-12 28.310750 28.402833 28.238167 28.238167 159625.000000 % matplotlib inline import matplotlib.pyplot as plt plt . plot ( aapl . index , aapl . Open , label = 'Open price' ) plt . plot ( moving_averages . index , moving_averages . Open , label = \"120 MA Open price\" ) plt . legend () plt . show ()","tags":"Python","url":"redoules.github.io/python/Moving_average_pandas.html","loc":"redoules.github.io/python/Moving_average_pandas.html"},{"title":"Keywords to use with WHERE","text":"Keywords to use with WHERE #load the extension % load_ext sql #connect to the database % sql sqlite : /// mydatabase . db 'Connected: @mydatabase.db' Assignment operator The assignment operator is =. % sql SELECT * FROM tutyfrutty WHERE color = \"red\" * sqlite:///mydatabase.db Done. index fruit color kcal 2 Apple red 52 7 Cranberry red 308 Comparison operators Comparison operation can be done in a SQL querry. They are the following : Equality : = Greater than : > greater than or equal to : >= less than : < less than or equal to : <= not equal to : <>, != not greater than : !> not less than : !< % sql SELECT * FROM tutyfrutty WHERE kcal = 47 * sqlite:///mydatabase.db Done. index fruit color kcal 1 Orange orange 47 % sql SELECT * FROM tutyfrutty WHERE kcal > 47 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 2 Apple red 52 7 Cranberry red 308 % sql SELECT * FROM tutyfrutty WHERE kcal >= 47 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 7 Cranberry red 308 % sql SELECT * FROM tutyfrutty WHERE kcal < 47 * sqlite:///mydatabase.db Done. index fruit color kcal 3 lemon yellow 15 4 lime green 30 5 plum purple 28 % sql SELECT * FROM tutyfrutty WHERE kcal <= 47 * sqlite:///mydatabase.db Done. index fruit color kcal 1 Orange orange 47 3 lemon yellow 15 4 lime green 30 5 plum purple 28 % sql SELECT * FROM tutyfrutty WHERE kcal <> 47 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 7 Cranberry red 308 Logical operators Logical operators test a condition and return a boolean. The logicial operators in SQL are : ALL : true if all the condtions are true AND : true is both conditions are true ANY : true if any one of the conditions are true BETWEEN : true if the operand in withing a range of values EXISTS : true if the subquery contains any rows IN : true if the condition is present in a row LIKE : true if a pattern is matched NOT : True if the operand is false, false otherwise OR : True is either condition is true SOME : true is any of the conditions is true % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" AND kcal < 100 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 3 lemon yellow 15 % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" OR kcal > 300 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 3 lemon yellow 15 7 Cranberry red 308 % sql SELECT * FROM tutyfrutty WHERE fruit LIKE 'l%' * sqlite:///mydatabase.db Done. index fruit color kcal 3 lemon yellow 15 4 lime green 30 % sql SELECT * FROM tutyfrutty WHERE NOT color = \"yellow\" * sqlite:///mydatabase.db Done. index fruit color kcal 1 Orange orange 47 2 Apple red 52 4 lime green 30 5 plum purple 28 7 Cranberry red 308 % sql SELECT * FROM tutyfrutty WHERE kcal BETWEEN 40 AND 100 * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 Bitwise operators Some bitwise operators exist in SQL. They will not be demonstrated here. They are the following : AND : & OR : | XOR : &#94; NOT : ~","tags":"SQL","url":"redoules.github.io/sql/WHERE_SQL_keywords.html","loc":"redoules.github.io/sql/WHERE_SQL_keywords.html"},{"title":"Sorting results","text":"Sorting results in SQL Sorting results can be achieved by using a modifier command at the end of the SQL querry #load the extension % load_ext sql #connect to the database % sql sqlite : /// mydatabase . db 'Connected: @mydatabase.db' The results can be sorted with the command ORDER BY SELECT column-list FROM table_name [WHERE condition] [ORDER BY column1, column2, .. columnN] [ASC | DESC] Let's show an example where we extract the fruits that are either yellow or red % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" OR color = \"red\" * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 2 Apple red 52 3 lemon yellow 15 7 Cranberry red 308 Ascending sort % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" OR color = \"red\" ORDER BY kcal ASC * sqlite:///mydatabase.db Done. index fruit color kcal 3 lemon yellow 15 2 Apple red 52 0 Banana yellow 89 7 Cranberry red 308 descending sort % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" OR color = \"red\" ORDER BY kcal DESC * sqlite:///mydatabase.db Done. index fruit color kcal 7 Cranberry red 308 0 Banana yellow 89 2 Apple red 52 3 lemon yellow 15 Sort by multiple columns You can sort by more than one column. Just specify multiple columns in the ORDER BY keyword. In the example, we will sort alphabetically on the color column first and sort alphabetically on the fruit column % sql SELECT * FROM tutyfrutty ORDER BY color , fruit ASC * sqlite:///mydatabase.db Done. index fruit color kcal 4 lime green 30 1 Orange orange 47 5 plum purple 28 2 Apple red 52 7 Cranberry red 308 0 Banana yellow 89 3 lemon yellow 15","tags":"SQL","url":"redoules.github.io/sql/Sorting_results.html","loc":"redoules.github.io/sql/Sorting_results.html"},{"title":"Filter content of a TABLE","text":"Filter content of a TABLE in SQL In this example, we will display the content of a table but we will filter out the results. Since we are working in the notebook, we will load the sql extension in order to manipulate the database. The database mydatabase.db is a SQLite database already created before the example. #load the extension % load_ext sql #connect to the database % sql sqlite : /// mydatabase . db 'Connected: @mydatabase.db' Filter content matching exactly a condition We want to extract all the entries in a dataframe that match a certain condition, in order to do so, we will use the following command : SELECT * FROM TABLE WHERE column=\"condition\" In our example, we will filter all the entries in the tutyfrutty table whose color is yellow % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 3 lemon yellow 15 Complex conditions You can build more complex conditions by using the keywords OR and AND In the following example, we will filter all entries that are either yellow or red % sql SELECT * FROM tutyfrutty WHERE color = \"yellow\" OR color = \"red\" * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 2 Apple red 52 3 lemon yellow 15 7 Cranberry red 308 Note : when combining multiple conditions with AND and OR, be careful to use parentesis where needed Conditions matching a pattern You can also use the LIKE keyword in order to find all entries that match a certain pattern. In our example, we want to find all fruits begining with a \"l\". In order to do so, we will use the LIKE keyword and the wildcard \"%\" meaning any string % sql SELECT * FROM tutyfrutty WHERE fruit LIKE \"l%\" * sqlite:///mydatabase.db Done. index fruit color kcal 3 lemon yellow 15 4 lime green 30 Numerical conditions When we are working with numerical data, we can use the GREATER THAN > and SMALLER THAN < operators % sql SELECT * FROM tutyfrutty WHERE kcal < 47 * sqlite:///mydatabase.db Done. index fruit color kcal 3 lemon yellow 15 4 lime green 30 5 plum purple 28 If we want the condition to be inclusive we can use the operator <= (alternatively >=) % sql SELECT * FROM tutyfrutty WHERE kcal <= 47 * sqlite:///mydatabase.db Done. index fruit color kcal 1 Orange orange 47 3 lemon yellow 15 4 lime green 30 5 plum purple 28","tags":"SQL","url":"redoules.github.io/sql/display_table_filter.html","loc":"redoules.github.io/sql/display_table_filter.html"},{"title":"Displaying the content of a TABLE","text":"Displaying the content of a TABLE in SQL In this very simple example we will see how to display the content of a table. Since we are working in the notebook, we will load the sql extension in order to manipulate the database. The database mydatabase.db is a SQLite database already created before the example. #load the extension % load_ext sql #connect to the database % sql sqlite : /// mydatabase . db 'Connected: @mydatabase.db' In order to extract all the values from a table, we will use the following command : SELECT * FROM TABLE In our example, we want to display the data contained in the table named tutyfrutty % sql SELECT * FROM tutyfrutty * sqlite:///mydatabase.db Done. index fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 7 Cranberry red 308","tags":"SQL","url":"redoules.github.io/sql/display_table.html","loc":"redoules.github.io/sql/display_table.html"},{"title":"Opening a file with python","text":"This short article show you how to open a file using python. We will use the with keyword in order to avoid having to close the file. There is no need to import anything in order to open a file. All the function related to file manipulation are part of the python standard library In order to open a file, we will use the function open. This function takes two arguments : the path of the file the mode you want to open the file The mode can be : 'r' : read 'w' : write 'a' : append (writes at the end of the file) 'b' : binary mode 'x' : exclusive creation 't' : text mode (by default) Note that if the file does not exit it will be created if you use the following options \"w\", \"a\", \"x\". If you try to open a non existing file in read mode 'r', a FileNotFoundError will be returned. It is possible to combine multiple options together. For instance, you can open a file in binary mode for writing using the 'wb' option. Python distinguishes between binary and text I/O. Files opened in binary mode return contents as bytes objects without any decoding. In text mode , the contents of the file are returned as str, the bytes having been first decoded using a platform-dependent encoding or using the specified encoding if given. Writing to a file Let's first open (create) a text file a write a string to it. filepath = \". \\\\ myfile.txt\" with open ( filepath , 'w' ) as f : f . write ( \"Hello world !\" ) Reading a file we can now see how to read the content of a file. To do so, we will use the 'r' option with open ( filepath , \"r\" ) as f : content = f . read () print ( content ) Hello world ! A word on the with keyword In python the with keyword is used when working with unmanaged resources (like file streams). The python documentation tells us that : The with statement clarifies code that previously would use try...finally blocks to ensure that clean-up code is executed. In this section, I'll discuss the statement as it will commonly be used. In the next section, I'll examine the implementation details and show how to write objects for use with this statement. The with statement is a control-flow structure whose basic structure is: with expression [ as variable ]: with - block The expression is evaluated, and it should result in an object that supports the context management protocol (that is, has enter () and exit () methods).","tags":"Python","url":"redoules.github.io/python/Opening_file.html","loc":"redoules.github.io/python/Opening_file.html"},{"title":"Opening a SQLite database with python","text":"This short article show you how to connect to a SQLite database using python. We will use the with keyword in order to avoid having to close the database. In order to connect to the database, we will have to import sqlite3 import sqlite3 from sqlite3 import Error In python the with keyword is used when working with unmanaged resources (like file streams). The python documentation tells us that : The with statement clarifies code that previously would use try...finally blocks to ensure that clean-up code is executed. In this section, I'll discuss the statement as it will commonly be used. In the next section, I'll examine the implementation details and show how to write objects for use with this statement. The with statement is a control-flow structure whose basic structure is: with expression [ as variable ]: with - block The expression is evaluated, and it should result in an object that supports the context management protocol (that is, has enter () and exit () methods). db_file = \". \\\\ mydatabase.db\" try : with sqlite3 . connect ( db_file ) as conn : print ( \"Connected to the database\" ) #your code here except Error as e : print ( e ) Connected to the database","tags":"Python","url":"redoules.github.io/python/Opening_SQLite_database.html","loc":"redoules.github.io/python/Opening_SQLite_database.html"},{"title":"Reading data from a sql database with pandas","text":"When manipulating you data using pandas, it is sometimes useful to pull data from a database. In this tutorial, we will see how to querry a dataframe from a sqlite table. Note than it would also work with any other sql database a long as you change the connxion to the one that suits your needs. First let's import pandas and sqlite3 import pandas as pd import sqlite3 from sqlite3 import Error We want to store the table tutyfrutty in our dataframe. To do so, we will query all the elements present in the tutyfrutty TABLE with the command : SELECT * FROM tutyfrutty db_file = \". \\\\ mydatabase.db\" try : with sqlite3 . connect ( db_file ) as conn : df = pd . read_sql ( \"SELECT * FROM tutyfrutty\" , conn ) del df [ \"index\" ] #juste delete the index column that was stored in the table except Error as e : print ( e ) df .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 6 Cranberry red 308 7 Cranberry red 308","tags":"Python","url":"redoules.github.io/python/Reading_data_from_a_sql_database_with_pandas.html","loc":"redoules.github.io/python/Reading_data_from_a_sql_database_with_pandas.html"},{"title":"Writing data to a sql database with pandas","text":"When manipulating you data using pandas, it is sometimes useful to store a dataframe. Pandas provides multiple ways to export dataframes. The most common consist in exporting to a csv, a pickle, to hdf or to excel. However, exporting to a sql database can prove very useful. Indeed, having a well structured database is a great for storing all the data related to your analysis in one place. In this tutorial, we will see how to store a dataframe in a new table of a sqlite dataframe. Note than it would also work with any other sql database a long as you change the connxion to the one that suits your needs. First let's import pandas and sqlite3 import pandas as pd import sqlite3 from sqlite3 import Error # Example dataframe raw_data = { 'fruit' : [ 'Banana' , 'Orange' , 'Apple' , 'lemon' , \"lime\" , \"plum\" ], 'color' : [ 'yellow' , 'orange' , 'red' , 'yellow' , \"green\" , \"purple\" ], 'kcal' : [ 89 , 47 , 52 , 15 , 30 , 28 ] } df = pd . DataFrame ( raw_data , columns = [ 'fruit' , 'color' , 'kcal' ]) df .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 Now that the DataFrame has been created, let's push it to the sqlite database called mydatabase.db in a new table called tutyfrutty db_file = \". \\\\ mydatabase.db\" try : with sqlite3 . connect ( db_file ) as conn : df . to_sql ( \"tutyfrutty\" , conn ) except Error as e : print ( e ) except ValueError : print ( \"The TABLE tutyfrutty already exists, read below to understand how to handle this case\" ) Note that if the table tutyfrutty was already existing, the to_sql function will return a ValueError. This is where, the if_exists option comes into play. Let's look at the docstring of this function : \"\"\" if_exists : {'fail', 'replace', 'append'}, default 'fail' - fail: If table exists, do nothing. - replace: If table exists, drop it, recreate it, and insert data. - append: If table exists, insert data. Create if does not exist. \"\"\" Let's say, I want to update my dataframe with some new rows df . loc [ len ( df ) + 1 ] = [ 'Cranberry' , 'red' , 308 ] df .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 7 Cranberry red 308 8 Cranberry red 308 I can now replace the table with the new values using the \"replace\" option db_file = \". \\\\ mydatabase.db\" try : with sqlite3 . connect ( db_file ) as conn : df . to_sql ( \"tutyfrutty\" , conn , if_exists = \"replace\" ) except Error as e : print ( e )","tags":"Python","url":"redoules.github.io/python/Writing_data_to_a_sql_database_with_pandas.html","loc":"redoules.github.io/python/Writing_data_to_a_sql_database_with_pandas.html"},{"title":"Creating a sqlite database","text":"When you want to start with using databases SQlite is a great tool. It provides an easy onramp to learn and prototype you database with a SQL compatible database. First, let's import the libraries we need import sqlite3 from sqlite3 import Error SQlite doesn't need a database server, however, you have to start by creating an empty database file import os def check_for_db_file (): if os . path . exists ( \"mydatabase.db\" ): print ( \"the database is ready\" ) else : print ( \"no database found\" ) check_for_db_file () no database found Let's then create a function that will connect to a database, print the verison of sqlite and then close the connexion to the database. def create_database ( db_file ): \"\"\" create a database connection to a SQLite database \"\"\" try : with sqlite3 . connect ( db_file ) as conn : print ( \"database created with sqlite3 version {0}\" . format ( sqlite3 . version )) except Error as e : print ( e ) create_database ( \".\\mydatabase.db\" ) database created with sqlite3 version 2.6.0 check_for_db_file () the database is ready You're all set. From now on, you can open the database and write sql querries into it.","tags":"Python","url":"redoules.github.io/python/Creating_a_sqlite_database.html","loc":"redoules.github.io/python/Creating_a_sqlite_database.html"},{"title":"Setting up the notebook for plotting with matplotlib","text":"Importing Matplotlib First we need to import pyplot, a collection of command style functions that make matplotlib work like MATLAB. Let's, as well, use the magic command %matplotlib inline in order to display the figures in the notebook import matplotlib.pyplot as plt % matplotlib inline # this doubles image size, but we'll do it manually below # %config InlineBackend.figure_format = 'retina' The following parameters are recommended for matplotlib, they will make matplotlib output a better quality image # %load snippets/matplot_setup.py plt . rcParams [ 'savefig.dpi' ] = 300 plt . rcParams [ 'figure.dpi' ] = 163 plt . rcParams [ 'figure.autolayout' ] = False plt . rcParams [ 'figure.figsize' ] = 20 , 12 plt . rcParams [ 'axes.labelsize' ] = 18 plt . rcParams [ 'axes.titlesize' ] = 20 plt . rcParams [ 'font.size' ] = 16 plt . rcParams [ 'lines.linewidth' ] = 2.0 plt . rcParams [ 'lines.markersize' ] = 8 plt . rcParams [ 'legend.fontsize' ] = 14 plt . rcParams [ 'text.usetex' ] = False # True activates latex output in fonts! plt . rcParams [ 'font.family' ] = \"serif\" plt . rcParams [ 'font.serif' ] = \"cm\" plt . rcParams [ 'text.latex.preamble' ] = \" \\\\ usepackage{subdepth}, \\\\ usepackage{type1cm}\" You can change the second line in order to fit your display. 163 dpi corresponds to a Dell Ultra HD 4k P2715Q. You can check your screen's dpi count at http://dpi.lv/","tags":"Python","url":"redoules.github.io/python/Setting_up_the_notebook_for_plotting_with_matplotlib.html","loc":"redoules.github.io/python/Setting_up_the_notebook_for_plotting_with_matplotlib.html"},{"title":"Why using a blockchain is a bad idea for your business","text":"What having a blockchain implies? storage costs : everyone maintaining the ledger needs to store every transaction bandwith costs : everyone has to broadcast every transaction computational costs : every node has to validate the blockchain control : the creator does not control the blockchain, everyone collectively controls it developpement costs : developping on a blockchain is way harder than on a traditionnal database What to ask a business when they tell you that they are using a blockchain? When a business is telling you about their innovative technology leveraging the power of the blockchain this should immedialty spake some questions : What is the consensus algorithm? who is responsible for validating the consensus rules? what is the nature of the participation ? is it open to access? is it open to innovation? is it a public ledger? is it transparent? does it improves acountability? is it cross borders? how is it validated?","tags":"Cryptocurrencies","url":"redoules.github.io/cryptocurrencies/blockchain_bad.html","loc":"redoules.github.io/cryptocurrencies/blockchain_bad.html"},{"title":"Synology NFS share","text":"Setting up a NFS share login to your DSM admin account, open the \"Control Panel\" and go to \"File Services\" Make sure NFS is enabled Back in the control panel, go to \"Shared Folder\" Select the folder you want to share and clic \"Edit\" Go to the \"NFS Permissions tab and clic \"Create\", add the ip of the device you want to mount the mapped drive on. Make sure you copy the \"Mount path\"","tags":"Linux","url":"redoules.github.io/linux/share_nfs_share.html","loc":"redoules.github.io/linux/share_nfs_share.html"},{"title":"Mount a NFS share using fstab","text":"Mount nfs using fstab The fstab file, generally located at /etc/fstab lists the differents partitions and where to load them on the filesystem. You can edit this file as root by using the following command sudo nano /etc/fstab in the following example, we want to mount a NFS v3 share from : server : 192.168.1.2 mountpoint (on the server) : /volumeUSB2/usbshare * mountlocation (on the client) : /mnt we specify 192 .168.1.2:/volumeUSB2/usbshare /mnt nfs nfsvers = 3 ,users 0 0 the client will then automatically mount the share ont /mnt at startup. Related you can reload the fstab file using this method : https://redoules.github.io/linux/Reloading_fstab.html You can create a NFS share on a Synology using the method : https://redoules.github.io/linux/share_nfs_share.html","tags":"Linux","url":"redoules.github.io/linux/mount_nfs_share_fstab.html","loc":"redoules.github.io/linux/mount_nfs_share_fstab.html"},{"title":"Installing bitcoind on raspberry pi","text":"Installing bitcoind on linux Running a full bitcoin node helps the bitcoin network to accept, validate and relay transactions. If you want to volunteer some spare computing and bandwidth resources to run a full node and allow Bitcoin to continue to grow you can grab an inexpensive and power efficient raspberry pi and turn it into a full node. There are plenty of tutorials on the Internet explaining how to install a bitcoin full node; this tutorial won't go over setting up a raspberry pi and using ssh. In order to store the full blockchain we will mount a network drive and tell bitcoind to use this mapped drive as the data directory. Download the bitcoin client Go to https://bitcoin.org/en/download Copy the URL for the ARM 32 bit version and download it onto your raspberry pi. wget https://bitcoin.org/bin/bitcoin-core-0.15.1/bitcoin-0.15.1-arm-linux-gnueabihf.tar.gz Locate the downloaded file and extract it using the arguement xzf tar xzf bitcoin-0.15.1-arm-linux-gnueabihf.tar.gz a new directory bitcoin-0.15.1 will be created, it contrains the files we need to install the software Install the bitcoin client We will install the content by copying the binaries located in the bin folder into /usr/local/bin by using the install command. You must use sudo because it will write data to a system directory sudo install -m 0755 -o root -g root -t /usr/local/bin bitcoin-0.15.1/bin/* Launch the bitcoin core client by running bitcoind -daemon Configuration of the node Start your node at boot Starting you node automatically at boot time is a good idea because it doesn't require a manual action from the user. The simplest way to achive this is to create a cronjob. Run the following command crontab -e Select the text editor of your choice, then add the following line at the end of the file @reboot bitcoind -daemon Save the file and exit; the updated crontab file will be installed for you. Full Node If you can afford to download and store all the blockchain, you can run a full node. At the time of writing, the blockchain is 150Go ( https://blockchain.info/fr/charts/blocks-size ). Tree ways to store this are : use a microSD with 256Go or more add a thumbdrive or an external drive to your raspberry pi * mount a network drive from a NAS If you have purchased a big SD card then you can leave the default location for the blockchain data (~/.bitcoin/). Otherwise, you will have to change the datadir location to where your drive is mounted (in my case I have mounted it to /mnt) In order to configure your bitcoin client, edit/create the file bitcoin.conf located in ~/.bitcoin/ nano ~/.bitcoin/bitcoin.conf copy the following text # From redoules.github.io # This config should be placed in following path: # ~/.bitcoin/bitcoin.conf # [core] # Specify a non-default location to store blockchain and other data. datadir=/mnt # Set database cache size in megabytes; machines sync faster with a larger cache. Recommend setting as high as possible based upon mach$ dbcache=100 # Keep at most <n> unconnectable transactions in memory. maxorphantx=10 # Keep the transaction memory pool below <n> megabytess. maxmempool=50 # [network] # Maintain at most N connections to peers. maxconnections=40 # Tries to keep outbound traffic under the given target (in MiB per 24h), 0 = no limit. maxuploadtarget=5000 Check https://jlopp.github.io/bitcoin-core-config-generator it is a handy site to edit the bitcoin.conf file Pruning node If you don't want to store the entire blockchain you can run a pruning node which reduces storage requirements by enabling pruning (deleting) of old blocks. Let's say you want to allocated at most 5Go to the blockchain, then specify prune=5000 into your bitcoin.conf file. Edit/create the file bitcoin.conf located in ~/.bitcoin/ nano ~/.bitcoin/bitcoin.conf copy the following text # From redoules.github.io # This config should be placed in following path: # ~/.bitcoin/bitcoin.conf # [core] # Set database cache size in megabytes; machines sync faster with a larger cache. Recommend setting as high as possible based upon mach$ dbcache=100 # Keep at most <n> unconnectable transactions in memory. maxorphantx=10 # Keep the transaction memory pool below <n> megabytess. maxmempool=50 # Reduce storage requirements by only storing most recent N MiB of block. This mode is incompatible with -txindex and -rescan. WARNING: Reverting this setting requires re-downloading the entire blockchain. (default: 0 = disable pruning blocks, 1 = allow manual pruning via RPC, greater than 550 = automatically prune blocks to stay under target size in MiB). prune=5000 # [network] # Maintain at most N connections to peers. maxconnections=40 # Tries to keep outbound traffic under the given target (in MiB per 24h), 0 = no limit. maxuploadtarget=5000 Checking if your node is public one of the best way to help the bitcoin network is to allow your node to be visible and to propagate block to other nodes. The bitcoin protocole uses port 8333, other clients should be able to share information with your client. Run ifconfig and check if you have an ipv6 adresse (look for adr inet6:) IPV6 Get the global ipv6 adresse of your raspberry pi Link encap:Ethernet HWaddr xx:xx:xx:xx:xx:xx inet adr:192.168.1.x Bcast:192.168.1.255 Masque:255.255.255.0 adr inet6: xxxx::xxxx:xxxx:xxxx:xxxx/64 Scope:Lien adr inet6: xxxx:xxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx/64 Scope:Global UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:42681744 errors:0 dropped:0 overruns:0 frame:0 TX packets:38447218 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 lg file transmission:1000 RX bytes:3044414780 (2.8 GiB) TX bytes:2599878680 (2.4 GiB) it is located between adr inet4 and Scope:Global adr inet6: xxxx:xxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx/64 Scope:Global Copy this adresse and past it into the search field on https://bitnodes.earn.com/ If your node is visible, it will appear on the website IPV4 If you don't have an ipv6 adresse, you will have to open port 8333 on your router and redirect it to the internal IP of your raspberry pi. It is not detailed here because the configuration depends on your router.","tags":"Cryptocurrencies","url":"redoules.github.io/cryptocurrencies/Installing_bitcoind_on_raspberry_pi.html","loc":"redoules.github.io/cryptocurrencies/Installing_bitcoind_on_raspberry_pi.html"},{"title":"Reloading .bashrc","text":"Reload .bashrc The .bashrc file, located at ~/.bashrc allows a user to personalize its bash shell. If you edit this file, the changes won't be loaded without login out and back in. However, you can use the following command to do it source ~/.bashrc","tags":"Linux","url":"redoules.github.io/linux/Reloading_.bashrc.html","loc":"redoules.github.io/linux/Reloading_.bashrc.html"},{"title":"Reloading fstab","text":"Reload fstab The fstab file, generally located at /etc/fstab lists the differents partitions and where to load them on the filesystem. If you edit this file, the changes won't be automounted. You either have to reboot your system of use the following command as root mount -a","tags":"Linux","url":"redoules.github.io/linux/Reloading_fstab.html","loc":"redoules.github.io/linux/Reloading_fstab.html"},{"title":"Updating all python package with anaconda","text":"Updating anaconda packages All packages managed by conda can be updated with the following command : conda update --all Updating other packages with pip For the other packages, the pip package manager can be used. Unfortunately pip hasn't the same update all fonctionnality. import pip from subprocess import call for dist in pip . get_installed_distributions (): print ( \"updating {0}\" . format ( dist )) call ( \"pip install --upgrade \" + dist . project_name , shell = True )","tags":"Python","url":"redoules.github.io/python/updating_all_python_package_with_anaconda.html","loc":"redoules.github.io/python/updating_all_python_package_with_anaconda.html"},{"title":"Saving a matplotlib figure with a high resolution","text":"creating a matplotlib figure #Importing matplotlib % matplotlib inline import matplotlib.pyplot as plt import numpy as np Drawing a figure # Fixing random state for reproducibility np . random . seed ( 19680801 ) mu , sigma = 100 , 15 x = mu + sigma * np . random . randn ( 10000 ) # the histogram of the data n , bins , patches = plt . hist ( x , 50 , normed = 1 , facecolor = 'g' , alpha = 0.75 ) plt . xlabel ( 'Smarts' ) plt . ylabel ( 'Probability' ) plt . title ( 'Histogram of IQ' ) plt . text ( 60 , . 025 , r '$\\mu=100,\\ \\sigma=15$' ) plt . axis ([ 40 , 160 , 0 , 0.03 ]) plt . grid ( True ) plt . show () Saving the figure normally, one would use the following code plt . savefig ( 'filename.png' ) &lt;matplotlib.figure.Figure at 0x2e45e92f400&gt; The figure in then exported to the file \"filename.png\" with a standard resolution. In adittion, you can specify the dpi arg to some scalar value, for example: plt . savefig ( 'filename_hi_dpi.png' , dpi = 300 ) &lt;matplotlib.figure.Figure at 0x2e462164898&gt;","tags":"Python","url":"redoules.github.io/python/Saving_a_matplotlib_figure_with_a_high_resolution.html","loc":"redoules.github.io/python/Saving_a_matplotlib_figure_with_a_high_resolution.html"},{"title":"Iterating over a DataFrame","text":"Create a sample dataframe # Import modules import pandas as pd # Example dataframe raw_data = { 'fruit' : [ 'Banana' , 'Orange' , 'Apple' , 'lemon' , \"lime\" , \"plum\" ], 'color' : [ 'yellow' , 'orange' , 'red' , 'yellow' , \"green\" , \"purple\" ], 'kcal' : [ 89 , 47 , 52 , 15 , 30 , 28 ] } df = pd . DataFrame ( raw_data , columns = [ 'fruit' , 'color' , 'kcal' ]) df .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } fruit color kcal 0 Banana yellow 89 1 Orange orange 47 2 Apple red 52 3 lemon yellow 15 4 lime green 30 5 plum purple 28 Using the iterrows method Pandas DataFrames can return a generator with the iterrrows method. It can then be used to loop over the rows of the DataFrame for index , row in df . iterrows (): print ( \"At line {0} there is a {1} which is {2} and contains {3} kcal\" . format ( index , row [ \"fruit\" ], row [ \"color\" ], row [ \"kcal\" ])) At line 0 there is a Banana which is yellow and contains 89 kcal At line 1 there is a Orange which is orange and contains 47 kcal At line 2 there is a Apple which is red and contains 52 kcal At line 3 there is a lemon which is yellow and contains 15 kcal At line 4 there is a lime which is green and contains 30 kcal At line 5 there is a plum which is purple and contains 28 kcal","tags":"Python","url":"redoules.github.io/python/Iterating_over_a_dataframe.html","loc":"redoules.github.io/python/Iterating_over_a_dataframe.html"},{"title":"Article Recommander","text":"import pandas as pd import numpy as np % matplotlib inline Loading data and preprocessing we first learn the pickled article database. We will be cleaning it and separating the interesting articles from the uninteresting ones. df = pd . read_pickle ( './article.pkl' ) del df [ \"html\" ] del df [ \"image\" ] del df [ \"URL\" ] del df [ \"hash\" ] del df [ \"source\" ] df [ \"label\" ] = df [ \"note\" ] . apply ( lambda x : 0 if x <= 0 else 1 ) df . head ( 5 ) .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } authors note resume texte titre label 0 [Danny Bradbury, Marco Santori, Adam Draper, M... -10.0 Black Market Reloaded, a black market site tha... Black Market Reloaded, a black market site tha... Black Market Reloaded back online after source... 0 1 [Emily Spaven, Stan Higgins, Emilyspaven] 1.0 The UK Home Office believes the government sho... The UK Home Office believes the government sho... Home Office: UK Should Create a Crime-Fighting... 1 2 [Pete Rizzo, Alex Batlin, Yessi Bello Perez, P... -10.0 Though lofty in its ideals, lead developer Dan... A new social messaging app is aiming to disrup... Gems Bitcoin App Lets Users Earn Money From So... 0 3 [Nermin Hajdarbegovic, Stan Higgins, Pete Rizz... 3.0 US satellite service provider DISH Network has... US satellite service provider DISH Network has... DISH Becomes World's Largest Company to Accept... 1 4 [Stan Higgins, Bailey Reutzel, Garrett Keirns,... -10.0 An unidentified 28-year-old man was robbed of ... An unidentified 28-year-old man was robbed of ... Bitcoin Stolen at Gunpoint in New York City Ro... 0 Basic statistics on the dataset let's explore the dataset and extract some numbers : * the number of article liked/disliked df [ \"label\" ] . value_counts () 0 879 1 324 Name: label, dtype: int64 Create the full content column df [ 'full_content' ] = df . titre + ' ' + df . resume #exclude the full texte of the article for the moment df . head ( 1 ) .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } authors note resume texte titre label full_content 0 [Danny Bradbury, Marco Santori, Adam Draper, M... -10.0 Black Market Reloaded, a black market site tha... Black Market Reloaded, a black market site tha... Black Market Reloaded back online after source... 0 Black Market Reloaded back online after source... from sklearn.model_selection import train_test_split training , testing = train_test_split ( df , # The dataset we want to split train_size = 0.75 , # The proportional size of our training set stratify = df . label , # The labels are used for stratification random_state = 400 # Use the same random state for reproducibility ) training . head ( 5 ) .dataframe thead tr:only-child th { text-align: right; } .dataframe thead th { text-align: left; } .dataframe tbody tr th { vertical-align: top; } authors note resume texte titre label full_content 748 [Jon Brodkin] -10.0 Amazon, Reddit, Mozilla, and other Internet co... Amazon, Reddit, Mozilla, and other Internet co... Amazon and Reddit try to save net neutrality r... 0 Amazon and Reddit try to save net neutrality r... 1183 [Jon Brodkin] -10.0 (The Time Warner involved in this transaction ... A group of mostly Democratic senators led by A... Democrats urge Trump administration to block A... 0 Democrats urge Trump administration to block A... 769 [Joseph Brogan] -10.0 On Twitter, bad news comes at all hours, with ... On Twitter, bad news comes at all hours, with ... Some of the best art on Twitter comes from the... 0 Some of the best art on Twitter comes from the... 57 [Michael Del Castillo, Pete Rizzo, Trond Vidar... -10.0 Publicly traded online travel service Webjet i... Publicly traded online travel service Webjet i... Webjet Ethereum Pilot Targets Hotel Industry's... 0 Webjet Ethereum Pilot Targets Hotel Industry's... 892 [Andrew Cunningham] 10.0 What has changed on the 2017 MacBook, then?\\nI... Andrew Cunningham\\n\\nAndrew Cunningham\\n\\nAndr... Mini-review: The 2017 MacBook could actually b... 1 Mini-review: The 2017 MacBook could actually b... from sklearn.feature_extraction.text import TfidfVectorizer , CountVectorizer from sklearn.svm import LinearSVC , SVC from sklearn.pipeline import Pipeline from sklearn.model_selection import cross_val_predict from utils.plotting import pipeline_performance steps = ( ( 'vectorizer' , TfidfVectorizer ()), ( 'classifier' , LinearSVC ()) ) pipeline = Pipeline ( steps ) predicted_labels = cross_val_predict ( pipeline , training . full_content , training . label ) pipeline_performance ( training . label , predicted_labels ) pipeline = pipeline . fit ( training . titre , training . label ) Accuracy = 80.6% Confusion matrix, without normalization [[624 35] [140 103]] import re from utils.plotting import print_top_features from sklearn.model_selection import GridSearchCV def mask_integers ( s ): return re . sub ( r '\\d+' , 'INTMASK' , s ) steps = ( ( 'vectorizer' , TfidfVectorizer ()), ( 'classifier' , LinearSVC ()) ) pipeline = Pipeline ( steps ) gs_params = { #'vectorizer__use_idf': (True, False), 'vectorizer__lowercase' : [ True , False ], 'vectorizer__stop_words' : [ 'english' , None ], 'vectorizer__ngram_range' : [( 1 , 1 ), ( 1 , 2 ), ( 2 , 2 )], 'vectorizer__preprocessor' : [ mask_integers , None ], 'classifier__C' : np . linspace ( 5 , 20 , 25 ) } gs = GridSearchCV ( pipeline , gs_params , n_jobs = 1 ) gs . fit ( training . full_content , training . label ) print ( gs . best_params_ ) print ( gs . best_score_ ) pipeline1 = gs . best_estimator_ predicted_labels = pipeline1 . predict ( testing . full_content ) pipeline_performance ( testing . label , predicted_labels ) print_top_features ( pipeline1 , n_features = 10 ) aaa = gs . predict ( testing . full_content ) == testing . label aaa = aaa [ testing . label == 1 ] testing [ \"titre\" ] . iloc [ ~ aaa . values ] #pipeline1.predict([\"windows xbox bitcoin\"]) from sklearn.externals import joblib joblib . dump ( pipeline1 , 'classifier.pkl' ) gs . predict ([ 'Google' ]) array([1], dtype=int64) steps = ( ( 'vectorizer' , TfidfVectorizer ()), ( 'classifier' , SVC ()) ) pipeline = Pipeline ( steps ) gs_params = { #'vectorizer__use_idf': (True, False), 'vectorizer__stop_words' : [ 'english' , None ], 'vectorizer__ngram_range' : [( 1 , 1 ), ( 1 , 2 ), ( 2 , 2 )], 'vectorizer__preprocessor' : [ mask_integers , None ], 'classifier__C' : np . linspace ( 5 , 20 , 25 ) } gs = GridSearchCV ( pipeline , gs_params , n_jobs = 1 ) gs . fit ( training . full_content , training . label ) print ( gs . best_params_ ) print ( gs . best_score_ ) pipeline1 = gs . best_estimator_ predicted_labels = pipeline1 . predict ( testing . full_content ) pipeline_performance ( testing . label , predicted_labels ) print_top_features ( pipeline1 , n_features = 10 ) {'classifier__C': 5.0, 'vectorizer__ngram_range': (1, 1), 'vectorizer__preprocessor': &lt;function mask_integers at 0x00000237491B67B8&gt;, 'vectorizer__stop_words': 'english'} 0.711180124224 Accuracy = 71.2% Confusion matrix, without normalization [[153 0] [ 62 0]] --------------------------------------------------------------------------- ValueError Traceback (most recent call last) &lt;ipython-input-9-3e0781e307fb&gt; in &lt;module&gt;() 25 pipeline_performance(testing.label, predicted_labels) 26 ---&gt; 27 print_top_features(pipeline1, n_features=10) C:\\Users\\Guillaume\\Documents\\Code\\recommandation\\utils\\plotting.py in print_top_features(pipeline, vectorizer_name, classifier_name, n_features) 81 def print_top_features(pipeline, vectorizer_name='vectorizer', classifier_name='classifier', n_features=7): 82 vocabulary = np.array(pipeline.named_steps[vectorizer_name].get_feature_names()) ---&gt; 83 coefs = pipeline.named_steps[classifier_name].coef_[0] 84 top_feature_idx = np.argsort(coefs) 85 top_features = vocabulary[top_feature_idx] C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\svm\\base.py in coef_(self) 483 def coef_(self): 484 if self.kernel != 'linear': --&gt; 485 raise ValueError('coef_ is only available when using a ' 486 'linear kernel') 487 ValueError: coef_ is only available when using a linear kernel from sklearn.naive_bayes import BernoulliNB steps = ( ( 'vectorizer' , TfidfVectorizer ()), ( 'classifier' , BernoulliNB ()) ) pipeline2 = Pipeline ( steps ) gs_params = { 'vectorizer__stop_words' : [ 'english' , None ], 'vectorizer__ngram_range' : [( 1 , 1 ), ( 1 , 2 ), ( 2 , 2 )], 'vectorizer__preprocessor' : [ mask_integers , None ], 'classifier__alpha' : np . linspace ( 0 , 1 , 5 ), 'classifier__fit_prior' : [ True , False ] } gs = GridSearchCV ( pipeline2 , gs_params , n_jobs = 1 ) gs . fit ( training . full_content , training . label ) print ( gs . best_params_ ) print ( gs . best_score_ ) pipeline2 = gs . best_estimator_ predicted_labels = pipeline2 . predict ( testing . full_content ) pipeline_performance ( testing . label , predicted_labels ) print_top_features ( pipeline2 , n_features = 10 ) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:820: RuntimeWarning: divide by zero encountered in log neg_prob = np.log(1 - np.exp(self.feature_log_prob_)) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:823: RuntimeWarning: invalid value encountered in add jll += self.class_log_prior_ + neg_prob.sum(axis=1) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:801: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - {'classifier__alpha': 0.25, 'classifier__fit_prior': True, 'vectorizer__ngram_range': (1, 1), 'vectorizer__preprocessor': &lt;function mask_integers at 0x00000237491B67B8&gt;, 'vectorizer__stop_words': 'english'} 0.805900621118 Accuracy = 78.1% Confusion matrix, without normalization [[140 13] [ 34 28]] Top like features: ['use' 'just' 'year' 'price' 'time' 'Bitcoin' 'bitcoin' 'new' 'The' 'INTMASK'] --- Top dislike features: ['ABBA' 'cable' 'cab' 'byte' 'publication' 'bye' 'publications' 'publicity' 'buyer' 'publicizing'] from sklearn.naive_bayes import MultinomialNB steps = ( ( 'vectorizer' , TfidfVectorizer ()), ( 'classifier' , MultinomialNB ()) ) pipeline3 = Pipeline ( steps ) gs_params = { 'vectorizer__stop_words' : [ 'english' , None ], 'vectorizer__ngram_range' : [( 1 , 1 ), ( 1 , 2 ), ( 2 , 2 )], 'vectorizer__preprocessor' : [ mask_integers , None ], 'classifier__alpha' : np . linspace ( 0 , 1 , 5 ), 'classifier__fit_prior' : [ True , False ] } gs = GridSearchCV ( pipeline3 , gs_params , n_jobs = 1 ) gs . fit ( training . full_content , training . label ) print ( gs . best_params_ ) print ( gs . best_score_ ) pipeline3 = gs . best_estimator_ predicted_labels = pipeline3 . predict ( testing . full_content ) pipeline_performance ( testing . label , predicted_labels ) print_top_features ( pipeline3 , n_features = 10 ) C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - C:\\Users\\Guillaume\\Anaconda3\\lib\\site-packages\\sklearn\\naive_bayes.py:699: RuntimeWarning: divide by zero encountered in log self.feature_log_prob_ = (np.log(smoothed_fc) - {'classifier__alpha': 0.5, 'classifier__fit_prior': False, 'vectorizer__ngram_range': (1, 1), 'vectorizer__preprocessor': &lt;function mask_integers at 0x00000237491B67B8&gt;, 'vectorizer__stop_words': 'english'} 0.80900621118 Accuracy = 79.1% Confusion matrix, without normalization [[141 12] [ 33 29]] Top like features: ['time' 'Google' 'Pro' 'Apple' 'new' 'The' 'Bitcoin' 'price' 'bitcoin' 'INTMASK'] --- Top dislike features: ['ABBA' 'categories' 'catching' 'catalyst' 'catalog' 'casually' 'casts' 'cast' 'cashier' 'ran']","tags":"Machine Learning","url":"redoules.github.io/machine-learning/Source code for the recommandation engine for articles.html","loc":"redoules.github.io/machine-learning/Source code for the recommandation engine for articles.html"}]}
\ No newline at end of file