From 3b69ee1e39f3b6252a357ef5f9eb62bbd5069530 Mon Sep 17 00:00:00 2001
From: opeceipeno <hbyspy2008@qq.com>
Date: Fri, 25 Feb 2022 10:22:50 +0800
Subject: [PATCH 01/28] =?UTF-8?q?=E6=B7=BB=E5=8A=A0=E6=B3=A8=E9=87=8A?=
 =?UTF-8?q?=E4=BF=A1=E6=81=AF?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/aggregators.py        | 163 +++++++------
 graphsage/layers.py             |  13 +-
 graphsage/models.py             | 397 +++++++++++++++++++++++---------
 graphsage/unsupervised_train.py |  34 ++-
 4 files changed, 411 insertions(+), 196 deletions(-)

diff --git a/graphsage/aggregators.py b/graphsage/aggregators.py
index 7dbd2523..eb73a33e 100644
--- a/graphsage/aggregators.py
+++ b/graphsage/aggregators.py
@@ -3,14 +3,15 @@
 from .layers import Layer, Dense
 from .inits import glorot, zeros
 
+
 class MeanAggregator(Layer):
     """
     Aggregates via mean followed by matmul and non-linearity.
     """
 
     def __init__(self, input_dim, output_dim, neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, 
-            name=None, concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu,
+                 name=None, concat=False, **kwargs):
         super(MeanAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -28,9 +29,9 @@ def __init__(self, input_dim, output_dim, neigh_input_dim=None,
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([neigh_input_dim, output_dim],
-                                                        name='neigh_weights')
+                                                name='neigh_weights')
             self.vars['self_weights'] = glorot([input_dim, output_dim],
-                                                        name='self_weights')
+                                               name='self_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -41,17 +42,20 @@ def __init__(self, input_dim, output_dim, neigh_input_dim=None,
         self.output_dim = output_dim
 
     def _call(self, inputs):
+
         self_vecs, neigh_vecs = inputs
 
+        #
         neigh_vecs = tf.nn.dropout(neigh_vecs, 1-self.dropout)
         self_vecs = tf.nn.dropout(self_vecs, 1-self.dropout)
         neigh_means = tf.reduce_mean(neigh_vecs, axis=1)
-       
+
         # [nodes] x [out_dim]
+        # batch * output_dim
         from_neighs = tf.matmul(neigh_means, self.vars['neigh_weights'])
 
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
-         
+
         if not self.concat:
             output = tf.add_n([from_self, from_neighs])
         else:
@@ -60,9 +64,10 @@ def _call(self, inputs):
         # bias
         if self.bias:
             output += self.vars['bias']
-       
+
         return self.act(output)
 
+
 class GCNAggregator(Layer):
     """
     Aggregates via mean followed by matmul and non-linearity.
@@ -70,7 +75,7 @@ class GCNAggregator(Layer):
     """
 
     def __init__(self, input_dim, output_dim, neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
         super(GCNAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -88,7 +93,7 @@ def __init__(self, input_dim, output_dim, neigh_input_dim=None,
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['weights'] = glorot([neigh_input_dim, output_dim],
-                                                        name='neigh_weights')
+                                          name='neigh_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -103,24 +108,25 @@ def _call(self, inputs):
 
         neigh_vecs = tf.nn.dropout(neigh_vecs, 1-self.dropout)
         self_vecs = tf.nn.dropout(self_vecs, 1-self.dropout)
-        means = tf.reduce_mean(tf.concat([neigh_vecs, 
-            tf.expand_dims(self_vecs, axis=1)], axis=1), axis=1)
-       
+        means = tf.reduce_mean(tf.concat([neigh_vecs,
+                                          tf.expand_dims(self_vecs, axis=1)], axis=1), axis=1)
+
         # [nodes] x [out_dim]
         output = tf.matmul(means, self.vars['weights'])
 
         # bias
         if self.bias:
             output += self.vars['bias']
-       
+
         return self.act(output)
 
 
 class MaxPoolingAggregator(Layer):
     """ Aggregates via max-pooling over MLP functions.
     """
+
     def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
         super(MaxPoolingAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -143,18 +149,18 @@ def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=No
 
         self.mlp_layers = []
         self.mlp_layers.append(Dense(input_dim=neigh_input_dim,
-                                 output_dim=hidden_dim,
-                                 act=tf.nn.relu,
-                                 dropout=dropout,
-                                 sparse_inputs=False,
-                                 logging=self.logging))
+                                     output_dim=hidden_dim,
+                                     act=tf.nn.relu,
+                                     dropout=dropout,
+                                     sparse_inputs=False,
+                                     logging=self.logging))
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([hidden_dim, output_dim],
-                                                        name='neigh_weights')
-           
+                                                name='neigh_weights')
+
             self.vars['self_weights'] = glorot([input_dim, output_dim],
-                                                        name='self_weights')
+                                               name='self_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -173,16 +179,18 @@ def _call(self, inputs):
         batch_size = dims[0]
         num_neighbors = dims[1]
         # [nodes * sampled neighbors] x [hidden_dim]
-        h_reshaped = tf.reshape(neigh_h, (batch_size * num_neighbors, self.neigh_input_dim))
+        h_reshaped = tf.reshape(
+            neigh_h, (batch_size * num_neighbors, self.neigh_input_dim))
 
         for l in self.mlp_layers:
             h_reshaped = l(h_reshaped)
-        neigh_h = tf.reshape(h_reshaped, (batch_size, num_neighbors, self.hidden_dim))
+        neigh_h = tf.reshape(
+            h_reshaped, (batch_size, num_neighbors, self.hidden_dim))
         neigh_h = tf.reduce_max(neigh_h, axis=1)
-        
+
         from_neighs = tf.matmul(neigh_h, self.vars['neigh_weights'])
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
-        
+
         if not self.concat:
             output = tf.add_n([from_self, from_neighs])
         else:
@@ -191,14 +199,16 @@ def _call(self, inputs):
         # bias
         if self.bias:
             output += self.vars['bias']
-       
+
         return self.act(output)
 
+
 class MeanPoolingAggregator(Layer):
     """ Aggregates via mean-pooling over MLP functions.
     """
+
     def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
         super(MeanPoolingAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -221,18 +231,18 @@ def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=No
 
         self.mlp_layers = []
         self.mlp_layers.append(Dense(input_dim=neigh_input_dim,
-                                 output_dim=hidden_dim,
-                                 act=tf.nn.relu,
-                                 dropout=dropout,
-                                 sparse_inputs=False,
-                                 logging=self.logging))
+                                     output_dim=hidden_dim,
+                                     act=tf.nn.relu,
+                                     dropout=dropout,
+                                     sparse_inputs=False,
+                                     logging=self.logging))
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([hidden_dim, output_dim],
-                                                        name='neigh_weights')
-           
+                                                name='neigh_weights')
+
             self.vars['self_weights'] = glorot([input_dim, output_dim],
-                                                        name='self_weights')
+                                               name='self_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -251,16 +261,18 @@ def _call(self, inputs):
         batch_size = dims[0]
         num_neighbors = dims[1]
         # [nodes * sampled neighbors] x [hidden_dim]
-        h_reshaped = tf.reshape(neigh_h, (batch_size * num_neighbors, self.neigh_input_dim))
+        h_reshaped = tf.reshape(
+            neigh_h, (batch_size * num_neighbors, self.neigh_input_dim))
 
         for l in self.mlp_layers:
             h_reshaped = l(h_reshaped)
-        neigh_h = tf.reshape(h_reshaped, (batch_size, num_neighbors, self.hidden_dim))
+        neigh_h = tf.reshape(
+            h_reshaped, (batch_size, num_neighbors, self.hidden_dim))
         neigh_h = tf.reduce_mean(neigh_h, axis=1)
-        
+
         from_neighs = tf.matmul(neigh_h, self.vars['neigh_weights'])
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
-        
+
         if not self.concat:
             output = tf.add_n([from_self, from_neighs])
         else:
@@ -269,15 +281,16 @@ def _call(self, inputs):
         # bias
         if self.bias:
             output += self.vars['bias']
-       
+
         return self.act(output)
 
 
 class TwoMaxLayerPoolingAggregator(Layer):
     """ Aggregates via pooling over two MLP functions.
     """
+
     def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
         super(TwoMaxLayerPoolingAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -302,25 +315,24 @@ def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=No
 
         self.mlp_layers = []
         self.mlp_layers.append(Dense(input_dim=neigh_input_dim,
-                                 output_dim=hidden_dim_1,
-                                 act=tf.nn.relu,
-                                 dropout=dropout,
-                                 sparse_inputs=False,
-                                 logging=self.logging))
+                                     output_dim=hidden_dim_1,
+                                     act=tf.nn.relu,
+                                     dropout=dropout,
+                                     sparse_inputs=False,
+                                     logging=self.logging))
         self.mlp_layers.append(Dense(input_dim=hidden_dim_1,
-                                 output_dim=hidden_dim_2,
-                                 act=tf.nn.relu,
-                                 dropout=dropout,
-                                 sparse_inputs=False,
-                                 logging=self.logging))
-
+                                     output_dim=hidden_dim_2,
+                                     act=tf.nn.relu,
+                                     dropout=dropout,
+                                     sparse_inputs=False,
+                                     logging=self.logging))
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([hidden_dim_2, output_dim],
-                                                        name='neigh_weights')
-           
+                                                name='neigh_weights')
+
             self.vars['self_weights'] = glorot([input_dim, output_dim],
-                                                        name='self_weights')
+                                               name='self_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -339,16 +351,18 @@ def _call(self, inputs):
         batch_size = dims[0]
         num_neighbors = dims[1]
         # [nodes * sampled neighbors] x [hidden_dim]
-        h_reshaped = tf.reshape(neigh_h, (batch_size * num_neighbors, self.neigh_input_dim))
+        h_reshaped = tf.reshape(
+            neigh_h, (batch_size * num_neighbors, self.neigh_input_dim))
 
         for l in self.mlp_layers:
             h_reshaped = l(h_reshaped)
-        neigh_h = tf.reshape(h_reshaped, (batch_size, num_neighbors, self.hidden_dim_2))
+        neigh_h = tf.reshape(
+            h_reshaped, (batch_size, num_neighbors, self.hidden_dim_2))
         neigh_h = tf.reduce_max(neigh_h, axis=1)
-        
+
         from_neighs = tf.matmul(neigh_h, self.vars['neigh_weights'])
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
-        
+
         if not self.concat:
             output = tf.add_n([from_self, from_neighs])
         else:
@@ -357,14 +371,16 @@ def _call(self, inputs):
         # bias
         if self.bias:
             output += self.vars['bias']
-       
+
         return self.act(output)
 
+
 class SeqAggregator(Layer):
     """ Aggregates via a standard LSTM.
     """
+
     def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, name=None,  concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None,  concat=False, **kwargs):
         super(SeqAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -387,10 +403,10 @@ def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=No
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([hidden_dim, output_dim],
-                                                        name='neigh_weights')
-           
+                                                name='neigh_weights')
+
             self.vars['self_weights'] = glorot([input_dim, output_dim],
-                                                        name='self_weights')
+                                               name='self_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -416,15 +432,15 @@ def _call(self, inputs):
         with tf.variable_scope(self.name) as scope:
             try:
                 rnn_outputs, rnn_states = tf.nn.dynamic_rnn(
-                        self.cell, neigh_vecs,
-                        initial_state=initial_state, dtype=tf.float32, time_major=False,
-                        sequence_length=length)
+                    self.cell, neigh_vecs,
+                    initial_state=initial_state, dtype=tf.float32, time_major=False,
+                    sequence_length=length)
             except ValueError:
                 scope.reuse_variables()
                 rnn_outputs, rnn_states = tf.nn.dynamic_rnn(
-                        self.cell, neigh_vecs,
-                        initial_state=initial_state, dtype=tf.float32, time_major=False,
-                        sequence_length=length)
+                    self.cell, neigh_vecs,
+                    initial_state=initial_state, dtype=tf.float32, time_major=False,
+                    sequence_length=length)
         batch_size = tf.shape(rnn_outputs)[0]
         max_len = tf.shape(rnn_outputs)[1]
         out_size = int(rnn_outputs.get_shape()[2])
@@ -434,7 +450,7 @@ def _call(self, inputs):
 
         from_neighs = tf.matmul(neigh_h, self.vars['neigh_weights'])
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
-         
+
         output = tf.add_n([from_self, from_neighs])
 
         if not self.concat:
@@ -445,6 +461,5 @@ def _call(self, inputs):
         # bias
         if self.bias:
             output += self.vars['bias']
-       
-        return self.act(output)
 
+        return self.act(output)
diff --git a/graphsage/layers.py b/graphsage/layers.py
index ca2496d9..d1b7b37e 100644
--- a/graphsage/layers.py
+++ b/graphsage/layers.py
@@ -16,6 +16,7 @@
 # global unique layer ID dictionary for layer name assignment
 _LAYER_UIDS = {}
 
+
 def get_layer_uid(layer_name=''):
     """Helper function, assigns unique layer IDs."""
     if layer_name not in _LAYER_UIDS:
@@ -25,6 +26,7 @@ def get_layer_uid(layer_name=''):
         _LAYER_UIDS[layer_name] += 1
         return _LAYER_UIDS[layer_name]
 
+
 class Layer(object):
     """Base layer class. Defines basic API for all layer objects.
     Implementation inspired by keras (http://keras.io).
@@ -72,8 +74,9 @@ def _log_vars(self):
 
 class Dense(Layer):
     """Dense layer."""
-    def __init__(self, input_dim, output_dim, dropout=0., 
-                 act=tf.nn.relu, placeholders=None, bias=True, featureless=False, 
+
+    def __init__(self, input_dim, output_dim, dropout=0.,
+                 act=tf.nn.relu, placeholders=None, bias=True, featureless=False,
                  sparse_inputs=False, **kwargs):
         super(Dense, self).__init__(**kwargs)
 
@@ -92,9 +95,9 @@ def __init__(self, input_dim, output_dim, dropout=0.,
 
         with tf.variable_scope(self.name + '_vars'):
             self.vars['weights'] = tf.get_variable('weights', shape=(input_dim, output_dim),
-                                         dtype=tf.float32, 
-                                         initializer=tf.contrib.layers.xavier_initializer(),
-                                         regularizer=tf.contrib.layers.l2_regularizer(FLAGS.weight_decay))
+                                                   dtype=tf.float32,
+                                                   initializer=tf.contrib.layers.xavier_initializer(),
+                                                   regularizer=tf.contrib.layers.l2_regularizer(FLAGS.weight_decay))
             if self.bias:
                 self.vars['bias'] = zeros([output_dim], name='bias')
 
diff --git a/graphsage/models.py b/graphsage/models.py
index b3b9db45..018fc47c 100644
--- a/graphsage/models.py
+++ b/graphsage/models.py
@@ -17,6 +17,7 @@
 # https://github.com/tkipf/gcn
 # which itself was very inspired by the keras package
 
+
 class Model(object):
     def __init__(self, **kwargs):
         allowed_kwargs = {'name', 'logging', 'model_size'}
@@ -30,8 +31,8 @@ def __init__(self, **kwargs):
         logging = kwargs.get('logging', False)
         self.logging = logging
 
-        self.vars = {}
-        self.placeholders = {}
+        self.vars = {}  # 模型参数
+        self.placeholders = {}  # 预留的位置，存放输入数据
 
         self.layers = []
         self.activations = []
@@ -53,21 +54,23 @@ def build(self):
             self._build()
 
         # Build sequential layer model
-        self.activations.append(self.inputs)
+        self.activations.append(self.inputs)    # 第一步，将输入数据加进激活层，作为第一层
         for layer in self.layers:
+            # 逐层计算中间层，并将每一层的输出都放进activations中保存
             hidden = layer(self.activations[-1])
             self.activations.append(hidden)
-        self.outputs = self.activations[-1]
+        self.outputs = self.activations[-1]   # 模型的输出即为最后一层
 
         # Store model variables for easy access
-        variables = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope=self.name)
+        variables = tf.get_collection(
+            tf.GraphKeys.GLOBAL_VARIABLES, scope=self.name)  # 获取全局的参数，并赋给vars
         self.vars = {var.name: var for var in variables}
 
         # Build metrics
-        self._loss()
+        self._loss()   # 定义损失函数
         self._accuracy()
 
-        self.opt_op = self.optimizer.minimize(self.loss)
+        self.opt_op = self.optimizer.minimize(self.loss)  # 优化策略
 
     def predict(self):
         pass
@@ -78,14 +81,14 @@ def _loss(self):
     def _accuracy(self):
         raise NotImplementedError
 
-    def save(self, sess=None):
+    def save(self, sess=None):   # 保存模型到本地文件
         if not sess:
             raise AttributeError("TensorFlow session not provided.")
         saver = tf.train.Saver(self.vars)
         save_path = saver.save(sess, "tmp/%s.ckpt" % self.name)
         print("Model saved in file: %s" % save_path)
 
-    def load(self, sess=None):
+    def load(self, sess=None):  # 从本地文件读取模型
         if not sess:
             raise AttributeError("TensorFlow session not provided.")
         saver = tf.train.Saver(self.vars)
@@ -96,6 +99,7 @@ def load(self, sess=None):
 
 class MLP(Model):
     """ A standard multi-layer perceptron """
+
     def __init__(self, placeholders, dims, categorical=True, **kwargs):
         super(MLP, self).__init__(**kwargs)
 
@@ -108,7 +112,8 @@ def __init__(self, placeholders, dims, categorical=True, **kwargs):
         self.inputs = placeholders['features']
         self.labels = placeholders['labels']
 
-        self.optimizer = tf.train.AdamOptimizer(learning_rate=FLAGS.learning_rate)
+        self.optimizer = tf.train.AdamOptimizer(
+            learning_rate=FLAGS.learning_rate)
 
         self.build()
 
@@ -120,45 +125,49 @@ def _loss(self):
         # Cross entropy error
         if self.categorical:
             self.loss += metrics.masked_softmax_cross_entropy(self.outputs, self.placeholders['labels'],
-                    self.placeholders['labels_mask'])
+                                                              self.placeholders['labels_mask'])
         # L2
         else:
             diff = self.labels - self.outputs
-            self.loss += tf.reduce_sum(tf.sqrt(tf.reduce_sum(diff * diff, axis=1)))
+            self.loss += tf.reduce_sum(
+                tf.sqrt(tf.reduce_sum(diff * diff, axis=1)))
 
     def _accuracy(self):
         if self.categorical:
             self.accuracy = metrics.masked_accuracy(self.outputs, self.placeholders['labels'],
-                    self.placeholders['labels_mask'])
+                                                    self.placeholders['labels_mask'])
 
     def _build(self):
         self.layers.append(layers.Dense(input_dim=self.input_dim,
-                                 output_dim=self.dims[1],
-                                 act=tf.nn.relu,
-                                 dropout=self.placeholders['dropout'],
-                                 sparse_inputs=False,
-                                 logging=self.logging))
+                                        output_dim=self.dims[1],
+                                        act=tf.nn.relu,
+                                        dropout=self.placeholders['dropout'],
+                                        sparse_inputs=False,
+                                        logging=self.logging))
 
         self.layers.append(layers.Dense(input_dim=self.dims[1],
-                                 output_dim=self.output_dim,
-                                 act=lambda x: x,
-                                 dropout=self.placeholders['dropout'],
-                                 logging=self.logging))
+                                        output_dim=self.output_dim,
+                                        act=lambda x: x,
+                                        dropout=self.placeholders['dropout'],
+                                        logging=self.logging))
 
     def predict(self):
         return tf.nn.softmax(self.outputs)
 
+
 class GeneralizedModel(Model):
     """
     Base class for models that aren't constructed from traditional, sequential layers.
     Subclasses must set self.outputs in _build method
 
     (Removes the layers idiom from build method of the Model class)
+
+    GeneralizedModel 这个类相比于Model类，主要是删去了中间的序列模型层，该模型需要其子类自己去定义中间层的计算逻辑以及输出
+
     """
 
     def __init__(self, **kwargs):
         super(GeneralizedModel, self).__init__(**kwargs)
-        
 
     def build(self):
         """ Wrapper for _build() """
@@ -166,7 +175,8 @@ def build(self):
             self._build()
 
         # Store model variables for easy access
-        variables = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, scope=self.name)
+        variables = tf.get_collection(
+            tf.GraphKeys.GLOBAL_VARIABLES, scope=self.name)
         self.vars = {var.name: var for var in variables}
 
         # Build metrics
@@ -175,14 +185,16 @@ def build(self):
 
         self.opt_op = self.optimizer.minimize(self.loss)
 
-# SAGEInfo is a namedtuple that specifies the parameters 
+
+# SAGEInfo is a namedtuple that specifies the parameters
 # of the recursive GraphSAGE layers
 SAGEInfo = namedtuple("SAGEInfo",
-    ['layer_name', # name of the layer (to get feature embedding etc.)
-     'neigh_sampler', # callable neigh_sampler constructor
-     'num_samples',
-     'output_dim' # the output (i.e., hidden) dimension
-    ])
+                      ['layer_name',  # name of the layer (to get feature embedding etc.)
+                       'neigh_sampler',  # callable neigh_sampler constructor
+                       'num_samples',
+                       'output_dim'  # the output (i.e., hidden) dimension
+                       ])
+
 
 class SampleAndAggregate(GeneralizedModel):
     """
@@ -190,9 +202,9 @@ class SampleAndAggregate(GeneralizedModel):
     """
 
     def __init__(self, placeholders, features, adj, degrees,
-            layer_infos, concat=True, aggregator_type="mean", 
-            model_size="small", identity_dim=0,
-            **kwargs):
+                 layer_infos, concat=True, aggregator_type="mean",
+                 model_size="small", identity_dim=0,
+                 **kwargs):
         '''
         Args:
             - placeholders: Stanford TensorFlow placeholder object.
@@ -206,7 +218,19 @@ def __init__(self, placeholders, features, adj, degrees,
             - aggregator_type: how to aggregate neighbor information
             - model_size: one of "small" and "big"
             - identity_dim: Set to positive int to use identity features (slow and cannot generalize, but better accuracy)
+
+             - features：节点特征 [num_nodes,50]
+             - adj： 图的邻接表， [num_nodes,maxdegree]
+             - degrees：列表，表示每个节点的度数[num_nodes]
+             - layer_infos：描述所有参数的 SAGEInfo 命名元组列表
+                    递归层。 参见上面的 SAGEInfo 定义。
+             - concat：是否在递归迭代期间拼接，是或者否
+             - aggregator_type：聚合方式的定义
+             - model_size：模型大小,有small 和big，  
+             - identity_dim：int，若＞0则加入额外特征（速度慢且泛化性差，但准确度更高）
         '''
+
+        # 选择聚合器类型
         super(SampleAndAggregate, self).__init__(**kwargs)
         if aggregator_type == "mean":
             self.aggregator_cls = MeanAggregator
@@ -222,61 +246,91 @@ def __init__(self, placeholders, features, adj, degrees,
             raise Exception("Unknown aggregator: ", self.aggregator_cls)
 
         # get info from placeholders...
+        # 两个输入 ，batch1和batch2 是一条边的两个顶点id，即每条边的两个顶点，分别放进batch1和batch2中，用作正样本
         self.inputs1 = placeholders["batch1"]
         self.inputs2 = placeholders["batch2"]
         self.model_size = model_size
         self.adj_info = adj
+
+        # 若identity_dim＞0，则创建额外的嵌入特征，扩充到feature的列维度上
         if identity_dim > 0:
-           self.embeds = tf.get_variable("node_embeddings", [adj.get_shape().as_list()[0], identity_dim])
+            self.embeds = tf.get_variable(
+                "node_embeddings", [adj.get_shape().as_list()[0], identity_dim])
         else:
-           self.embeds = None
-        if features is None: 
+            self.embeds = None
+        if features is None:
             if identity_dim == 0:
-                raise Exception("Must have a positive value for identity feature dimension if no input features given.")
+                raise Exception(
+                    "Must have a positive value for identity feature dimension if no input features given.")
             self.features = self.embeds
         else:
-            self.features = tf.Variable(tf.constant(features, dtype=tf.float32), trainable=False)
+            self.features = tf.Variable(tf.constant(
+                features, dtype=tf.float32), trainable=False)  # 节点特征，
             if not self.embeds is None:
+                # (feature的最终特征维度为 原始特征维度50+identity_dim)
                 self.features = tf.concat([self.embeds, self.features], axis=1)
         self.degrees = degrees
-        self.concat = concat
+        self.concat = concat  # 布尔值，是否拼接
 
-        self.dims = [(0 if features is None else features.shape[1]) + identity_dim]
-        self.dims.extend([layer_infos[i].output_dim for i in range(len(layer_infos))])
+        # dim是一个列表，代表每一层的输出维度，第一层是输入层，维度=输入特征的维度，后面的维度是从layer_info得到的
+        # 本实验中，dims = [50，128，128]
+
+        self.dims = [
+            (0 if features is None else features.shape[1]) + identity_dim]
+        self.dims.extend(
+            [layer_infos[i].output_dim for i in range(len(layer_infos))])
         self.batch_size = placeholders["batch_size"]
         self.placeholders = placeholders
         self.layer_infos = layer_infos
 
-        self.optimizer = tf.train.AdamOptimizer(learning_rate=FLAGS.learning_rate)
+        # 优化器选择
+        self.optimizer = tf.train.AdamOptimizer(
+            learning_rate=FLAGS.learning_rate)
 
+        # 构建模型
         self.build()
 
     def sample(self, inputs, layer_infos, batch_size=None):
         """ Sample neighbors to be the supportive fields for multi-layer convolutions.
 
+        对节点邻居采样，作为该点的是支持域
+        samples[0] 维度是 [batch_size]
+        samples[1] [layer_infos[1].num_samples * batch_size]
+        samples[2] [layer_infos[1].num_samples * layer_infos[0].num_samples * batch_size]
+
         Args:
             inputs: batch inputs
             batch_size: the number of inputs (different for batch inputs and negative samples).
         """
-        
+
         if batch_size is None:
             batch_size = self.batch_size
-        samples = [inputs]
+        samples = [inputs]  # samples[0] 是输入，
         # size of convolution support at each layer per node
+        # support_sizes 存的是的各层的采样数目，是一个列表
+        # support_sizes[0] = 1， 初始状态就是节点本身
+        # support_sizes[1] = layer_infos[-1].num_samples * 1， 本实验中为10
+        # support_sizes[2] = layer_infos[-1].num_samples * layer_infos[-2].num_samples * 1， 本实验中为250
+        # 以此类推，从最外层的邻居数依次往内乘
+
         support_size = 1
         support_sizes = [support_size]
-        for k in range(len(layer_infos)):
-            t = len(layer_infos) - k - 1
+
+        for k in range(len(layer_infos)):  # k为跳数，也是层数， 实验中k = 0 1
+            t = len(layer_infos) - k - 1  # t = 1 0
+            # 每一跳的邻居数目是前一跳的邻居节点数*该层的采样数
             support_size *= layer_infos[t].num_samples
-            sampler = layer_infos[t].neigh_sampler
+            sampler = layer_infos[t].neigh_sampler  # 采样器选择，实验中采样器是同一种
+
+            # 采样器的两个输入，第一个是将被采样的节点id，第二个是采样数
             node = sampler((samples[k], layer_infos[t].num_samples))
-            samples.append(tf.reshape(node, [support_size * batch_size,]))
+            # reshape并放进数组
+            samples.append(tf.reshape(node, [support_size * batch_size, ]))
             support_sizes.append(support_size)
         return samples, support_sizes
 
-
     def aggregate(self, samples, input_features, dims, num_samples, support_sizes, batch_size=None,
-            aggregators=None, name=None, concat=False, model_size="small"):
+                  aggregators=None, name=None, concat=False, model_size="small"):
         """ At each layer, aggregate hidden representations of neighbors to compute the hidden representations 
             at next layer.
         Args:
@@ -290,49 +344,100 @@ def aggregate(self, samples, input_features, dims, num_samples, support_sizes, b
             batch_size: the number of inputs (different for batch inputs and negative samples).
         Returns:
             The hidden representation at the final layer for all nodes in batch
+
+        samples: 一个列表，里面存放的是邻居节点id，
+                sample[0]是初始节点，可以理解为第0跳邻居采样 （hop）
+                sample[1]是对sample[0]中每一个节点进行邻居采样，即第1跳采样
+                sample[2]是对sample[1]中每一个节点进行邻居采样，即第2跳采样
+                以此类推
+        input_features: 矩阵，存放的是全量的节点的特征 
+        dims: 列表，代表每一层的中间表达的维度
+
+        num_samples: 列表，表示模型每一层的邻居采样数目，实验中为[25,10]
+        support_sizes: the number of nodes to gather information from for each layer.
+        batch_size: the number of inputs (different for batch inputs and negative samples).
+        Returns:
+
+
+
+
         """
 
         if batch_size is None:
             batch_size = self.batch_size
 
         # length: number of layers + 1
-        hidden = [tf.nn.embedding_lookup(input_features, node_samples) for node_samples in samples]
+        # 根据节点id，从全量的特征矩阵里获取节点特征
+        # hidden[0] [batch, num_features]
+        # hidden[1] [layer_infos[1].num_samples * batch_size, num_features]
+        # hidden[2] [layer_infos[1].num_samples * layer_infos[0].num_samples * batch_size, num_features]
+        # num_features表示的是特征维度，实验中为50
+
+        hidden = [tf.nn.embedding_lookup(
+            input_features, node_samples) for node_samples in samples]
+        
+        # 输入batch1的时候，该项为aggregators = None， 输入batch2或者neg_samples的时候，aggregators为batch1的aggregators
         new_agg = aggregators is None
+
         if new_agg:
             aggregators = []
-        for layer in range(len(num_samples)):
+        for layer in range(len(num_samples)):   # 按层数循环
             if new_agg:
-                dim_mult = 2 if concat and (layer != 0) else 1
+                dim_mult = 2 if concat and (layer != 0) else 1  # 维度系数，如果有concat=True，从第二层开始，输入维度需要乘2
                 # aggregator at current layer
+                # 判断是否是最后一层，如果是的话，会有个参数act=lambda x: x
                 if layer == len(num_samples) - 1:
-                    aggregator = self.aggregator_cls(dim_mult*dims[layer], dims[layer+1], act=lambda x : x,
-                            dropout=self.placeholders['dropout'], 
-                            name=name, concat=concat, model_size=model_size)
+                    aggregator = self.aggregator_cls(dim_mult*dims[layer], dims[layer+1], act=lambda x: x,
+                                                     dropout=self.placeholders['dropout'],
+                                                     name=name, concat=concat, model_size=model_size)
                 else:
+                    # 初始化一个聚合器类，类别是超参定义的，需要的参数是输入维度、输出维度、dropout系数等等
                     aggregator = self.aggregator_cls(dim_mult*dims[layer], dims[layer+1],
-                            dropout=self.placeholders['dropout'], 
-                            name=name, concat=concat, model_size=model_size)
+                                                     dropout=self.placeholders['dropout'],
+                                                     name=name, concat=concat, model_size=model_size)
                 aggregators.append(aggregator)
             else:
+                # 在batch2或者neg_samples输入的时候，直接使用已有的聚合器
                 aggregator = aggregators[layer]
+            
+            # 本实验中，aggregator1 的输入输出维度分别为：50，256， 参数矩阵维度为50，128 ，后面有个拼接
+            # aggregator2 的输入输出维度为：256，256，参数矩阵维度为256，128
+
+
+
             # hidden representation at current layer for all support nodes that are various hops away
             next_hidden = []
+            
+            
             # as layer increases, the number of support nodes needed decreases
-            for hop in range(len(num_samples) - layer):
+            # 随着层数增加，跳数需要减少
+            for hop in range(len(num_samples) - layer): 
                 dim_mult = 2 if concat and (layer != 0) else 1
-                neigh_dims = [batch_size * support_sizes[hop], 
-                              num_samples[len(num_samples) - hop - 1], 
+
+                # 每个节点的特征，是由自身的特征和其邻居节点的特征聚合而来的，
+                # hidden[hop+1]包含了hidden[hop]中节点的所有邻居特征
+                # 因为hidden[i]是二维的，而mean_aggregator是需要将邻居节点特征平均，
+                # 因此需要将它reshape一下，方便在后面的处理中取所有邻居的均值
+                # neigh_dims = [batch_size * 当前跳数的支持节点数，当前层的需要采样的邻居节点数，特征数]
+
+                # 
+                neigh_dims = [batch_size * support_sizes[hop],
+                              num_samples[len(num_samples) - hop - 1],
                               dim_mult*dims[layer]]
                 h = aggregator((hidden[hop],
-                                tf.reshape(hidden[hop + 1], neigh_dims)))
+                                tf.reshape(hidden[hop + 1], neigh_dims)))   
                 next_hidden.append(h)
             hidden = next_hidden
         return hidden[0], aggregators
 
     def _build(self):
+
+        # 将第batch2作为标签，即batch1和batch2是一对正样本对
         labels = tf.reshape(
-                tf.cast(self.placeholders['batch2'], dtype=tf.int64),
-                [self.batch_size, 1])
+            tf.cast(self.placeholders['batch2'], dtype=tf.int64),
+            [self.batch_size, 1])
+
+        # 获取负样本， 按照给定的概率分布进行采样
         self.neg_samples, _, _ = (tf.nn.fixed_unigram_candidate_sampler(
             true_classes=labels,
             num_true=1,
@@ -342,66 +447,132 @@ def _build(self):
             distortion=0.75,
             unigrams=self.degrees.tolist()))
 
-           
         # perform "convolution"
+
+        # 根据节点id去采样其邻居节点id
         samples1, support_sizes1 = self.sample(self.inputs1, self.layer_infos)
         samples2, support_sizes2 = self.sample(self.inputs2, self.layer_infos)
-        num_samples = [layer_info.num_samples for layer_info in self.layer_infos]
+        
+        # 每层需要的采样数 实验中是[25,10]
+        num_samples = [
+            layer_info.num_samples for layer_info in self.layer_infos]
+
+
+        # 获取batch1的特征表达，该步传入的聚合器参数为None，会构建一个聚合器返回
         self.outputs1, self.aggregators = self.aggregate(samples1, [self.features], self.dims, num_samples,
-                support_sizes1, concat=self.concat, model_size=self.model_size)
+                                                         support_sizes1, concat=self.concat, model_size=self.model_size)
+        
+        # 获取batch2的特征表达，其中聚合器是直接使用上一步生成的
         self.outputs2, _ = self.aggregate(samples2, [self.features], self.dims, num_samples,
-                support_sizes2, aggregators=self.aggregators, concat=self.concat,
-                model_size=self.model_size)
-
+                                          support_sizes2, aggregators=self.aggregators, concat=self.concat,
+                                          model_size=self.model_size)
+        
+        # 对负样本采样
         neg_samples, neg_support_sizes = self.sample(self.neg_samples, self.layer_infos,
-            FLAGS.neg_sample_size)
+                                                     FLAGS.neg_sample_size)
+
+        # 获取负样本的特征表达，聚合器也是用和之前同一个，注意batch_size参数，这里赋值的是负样本数量，和正样本的batch_size不同
         self.neg_outputs, _ = self.aggregate(neg_samples, [self.features], self.dims, num_samples,
-                neg_support_sizes, batch_size=FLAGS.neg_sample_size, aggregators=self.aggregators,
-                concat=self.concat, model_size=self.model_size)
+                                             neg_support_sizes, batch_size=FLAGS.neg_sample_size, aggregators=self.aggregators,
+                                             concat=self.concat, model_size=self.model_size)
 
         dim_mult = 2 if self.concat else 1
+
+        # 这里生成了一个预测层，注意参数bilinear_weights,这个值如果为True，则会生成一个可训练的参数矩阵，在后续的计算loss会用到
+        # 在这里设置了否，则无参数矩阵，本质上就是一个计算loss的类
         self.link_pred_layer = BipartiteEdgePredLayer(dim_mult*self.dims[-1],
-                dim_mult*self.dims[-1], self.placeholders, act=tf.nn.sigmoid, 
-                bilinear_weights=False,
-                name='edge_predict')
+                                                      dim_mult*self.dims[-1], self.placeholders, act=tf.nn.sigmoid,
+                                                      bilinear_weights=False,
+                                                      name='edge_predict')
+
 
+        # 对输出的样本执行L2规范化，dim=1，表示按行做，即按样本点做
+        # x_l2[i] = x[i]/sqrt(sum(x^2))
         self.outputs1 = tf.nn.l2_normalize(self.outputs1, 1)
         self.outputs2 = tf.nn.l2_normalize(self.outputs2, 1)
         self.neg_outputs = tf.nn.l2_normalize(self.neg_outputs, 1)
 
     def build(self):
+
+        # 构建模型的输出
         self._build()
 
         # TF graph management
+        # 构建模型的损失函数和准确度指标
         self._loss()
         self._accuracy()
+
+        # 除以batch，得到的平均loss
         self.loss = self.loss / tf.cast(self.batch_size, tf.float32)
+
+        # 计算梯度
         grads_and_vars = self.optimizer.compute_gradients(self.loss)
-        clipped_grads_and_vars = [(tf.clip_by_value(grad, -5.0, 5.0) if grad is not None else None, var) 
-                for grad, var in grads_and_vars]
+
+        # 梯度裁剪，若梯度大于5则置为5，小于-5则置为-5
+        clipped_grads_and_vars = [(tf.clip_by_value(grad, -5.0, 5.0) if grad is not None else None, var)
+                                  for grad, var in grads_and_vars]
         self.grad, _ = clipped_grads_and_vars[0]
+
+        # 更新参数
         self.opt_op = self.optimizer.apply_gradients(clipped_grads_and_vars)
 
     def _loss(self):
+        
+        # L2正则化项
         for aggregator in self.aggregators:
             for var in aggregator.vars.values():
                 self.loss += FLAGS.weight_decay * tf.nn.l2_loss(var)
-
-        self.loss += self.link_pred_layer.loss(self.outputs1, self.outputs2, self.neg_outputs) 
+        
+        # 根据之前生成的预测层，计算loss，该loss有三个选项：_xent_loss、_skipgram_loss、_hinge_loss
+        self.loss += self.link_pred_layer.loss(
+            self.outputs1, self.outputs2, self.neg_outputs)
         tf.summary.scalar('loss', self.loss)
 
     def _accuracy(self):
-        # shape: [batch_size]
+        """
+        计算性能指标
+        模型计算了三组数据：一条边上的两个顶点(batch1和batch2)和采样得到的负样本(neg_samples)的特征值
+        主体思想是希望边两端点之间的相似度>该点和所有neg_samles相似度
+        ①计算正样本对的"亲和度"
+        ②计算顶点和负样本的"亲和度"
+        ③将两组数据拼接，拼接后的数组维度[batch_size, neg_samples_size + 1],意义是每一个顶点和负样本、正样本之间的"亲和度"
+        ④计算正样本对之间的亲和度的排名，排名越靠前越好
+        mrr值，
+        """        
+        
+        # ①计算正样本对的"亲和度"
+        # aff值在本实验即是两个输入按元素点乘，再按行求和
+        # shape : [batch_size,] 表示了该batch中，每个节点和其邻居节点的“亲和度”，越大代表越相似
+        
         aff = self.link_pred_layer.affinity(self.outputs1, self.outputs2)
-        # shape : [batch_size x num_neg_samples]
-        self.neg_aff = self.link_pred_layer.neg_cost(self.outputs1, self.neg_outputs)
-        self.neg_aff = tf.reshape(self.neg_aff, [self.batch_size, FLAGS.neg_sample_size])
+
+
+        # ②计算顶点和负样本的"亲和度"
+        # 返回的是一个矩阵，维度：[batch_size，num_neg_samples]
+        # 含义是一组batch里每一个节点对每个负样本的"亲和度"
+        self.neg_aff = self.link_pred_layer.neg_cost(
+            self.outputs1, self.neg_outputs)
+
+        self.neg_aff = tf.reshape(
+            self.neg_aff, [self.batch_size, FLAGS.neg_sample_size])
+        
+        # ③将两组数据拼接，拼接后的数组维度[batch_size, neg_samples_size + 1],意义是每一个顶点和负样本、正样本之间的"亲和度"
+        # shape : [batch_size,1]
         _aff = tf.expand_dims(aff, axis=1)
+        # shape : [batch_size,num_neg_samples + 1]
         self.aff_all = tf.concat(axis=1, values=[self.neg_aff, _aff])
         size = tf.shape(self.aff_all)[1]
+
+        # ④利用top_k函数，两步计算出正样本对之间的亲和度的排名，
+        # self.ranks中表示的是每个顶点和负样本、正样本之间的亲和度排名，维度:[batch_size, neg_samples_size + 1]
         _, indices_of_ranks = tf.nn.top_k(self.aff_all, k=size)
         _, self.ranks = tf.nn.top_k(-indices_of_ranks, k=size)
-        self.mrr = tf.reduce_mean(tf.div(1.0, tf.cast(self.ranks[:, -1] + 1, tf.float32)))
+
+
+        # 取self.ranks最后一列，即正样本的排名序数，因为是从0算起的，所以要+1
+        # mrr = 1.0/rank 
+        self.mrr = tf.reduce_mean(
+            tf.div(1.0, tf.cast(self.ranks[:, -1] + 1, tf.float32)))
         tf.summary.scalar('mrr', self.mrr)
 
 
@@ -429,15 +600,15 @@ def __init__(self, placeholders, dict_size, degrees, name=None,
 
         # following the tensorflow word2vec tutorial
         self.target_embeds = tf.Variable(
-                tf.random_uniform([dict_size, nodevec_dim], -1, 1),
-                name="target_embeds")
+            tf.random_uniform([dict_size, nodevec_dim], -1, 1),
+            name="target_embeds")
         self.context_embeds = tf.Variable(
-                tf.truncated_normal([dict_size, nodevec_dim],
-                stddev=1.0 / math.sqrt(nodevec_dim)),
-                name="context_embeds")
+            tf.truncated_normal([dict_size, nodevec_dim],
+                                stddev=1.0 / math.sqrt(nodevec_dim)),
+            name="context_embeds")
         self.context_bias = tf.Variable(
-                tf.zeros([dict_size]),
-                name="context_bias")
+            tf.zeros([dict_size]),
+            name="context_bias")
 
         self.optimizer = tf.train.GradientDescentOptimizer(learning_rate=lr)
 
@@ -445,8 +616,8 @@ def __init__(self, placeholders, dict_size, degrees, name=None,
 
     def _build(self):
         labels = tf.reshape(
-                tf.cast(self.placeholders['batch2'], dtype=tf.int64),
-                [self.batch_size, 1])
+            tf.cast(self.placeholders['batch2'], dtype=tf.int64),
+            [self.batch_size, 1])
         self.neg_samples, _, _ = (tf.nn.fixed_unigram_candidate_sampler(
             true_classes=labels,
             num_true=1,
@@ -456,14 +627,19 @@ def _build(self):
             distortion=0.75,
             unigrams=self.degrees.tolist()))
 
-        self.outputs1 = tf.nn.embedding_lookup(self.target_embeds, self.inputs1)
-        self.outputs2 = tf.nn.embedding_lookup(self.context_embeds, self.inputs2)
-        self.outputs2_bias = tf.nn.embedding_lookup(self.context_bias, self.inputs2)
-        self.neg_outputs = tf.nn.embedding_lookup(self.context_embeds, self.neg_samples)
-        self.neg_outputs_bias = tf.nn.embedding_lookup(self.context_bias, self.neg_samples)
+        self.outputs1 = tf.nn.embedding_lookup(
+            self.target_embeds, self.inputs1)
+        self.outputs2 = tf.nn.embedding_lookup(
+            self.context_embeds, self.inputs2)
+        self.outputs2_bias = tf.nn.embedding_lookup(
+            self.context_bias, self.inputs2)
+        self.neg_outputs = tf.nn.embedding_lookup(
+            self.context_embeds, self.neg_samples)
+        self.neg_outputs_bias = tf.nn.embedding_lookup(
+            self.context_bias, self.neg_samples)
 
         self.link_pred_layer = BipartiteEdgePredLayer(self.hidden_dim, self.hidden_dim,
-                self.placeholders, bilinear_weights=False)
+                                                      self.placeholders, bilinear_weights=False)
 
     def build(self):
         self._build()
@@ -476,26 +652,31 @@ def _minimize(self):
         self.opt_op = self.optimizer.minimize(self.loss)
 
     def _loss(self):
-        aff = tf.reduce_sum(tf.multiply(self.outputs1, self.outputs2), 1) + self.outputs2_bias
-        neg_aff = tf.matmul(self.outputs1, tf.transpose(self.neg_outputs)) + self.neg_outputs_bias
+        aff = tf.reduce_sum(tf.multiply(
+            self.outputs1, self.outputs2), 1) + self.outputs2_bias
+        neg_aff = tf.matmul(self.outputs1, tf.transpose(
+            self.neg_outputs)) + self.neg_outputs_bias
         true_xent = tf.nn.sigmoid_cross_entropy_with_logits(
-                labels=tf.ones_like(aff), logits=aff)
+            labels=tf.ones_like(aff), logits=aff)
         negative_xent = tf.nn.sigmoid_cross_entropy_with_logits(
-                labels=tf.zeros_like(neg_aff), logits=neg_aff)
+            labels=tf.zeros_like(neg_aff), logits=neg_aff)
         loss = tf.reduce_sum(true_xent) + tf.reduce_sum(negative_xent)
         self.loss = loss / tf.cast(self.batch_size, tf.float32)
         tf.summary.scalar('loss', self.loss)
-        
+
     def _accuracy(self):
         # shape: [batch_size]
         aff = self.link_pred_layer.affinity(self.outputs1, self.outputs2)
        # shape : [batch_size x num_neg_samples]
-        self.neg_aff = self.link_pred_layer.neg_cost(self.outputs1, self.neg_outputs)
-        self.neg_aff = tf.reshape(self.neg_aff, [self.batch_size, FLAGS.neg_sample_size])
+        self.neg_aff = self.link_pred_layer.neg_cost(
+            self.outputs1, self.neg_outputs)
+        self.neg_aff = tf.reshape(
+            self.neg_aff, [self.batch_size, FLAGS.neg_sample_size])
         _aff = tf.expand_dims(aff, axis=1)
         self.aff_all = tf.concat(axis=1, values=[self.neg_aff, _aff])
         size = tf.shape(self.aff_all)[1]
         _, indices_of_ranks = tf.nn.top_k(self.aff_all, k=size)
         _, self.ranks = tf.nn.top_k(-indices_of_ranks, k=size)
-        self.mrr = tf.reduce_mean(tf.div(1.0, tf.cast(self.ranks[:, -1] + 1, tf.float32)))
+        self.mrr = tf.reduce_mean(
+            tf.div(1.0, tf.cast(self.ranks[:, -1] + 1, tf.float32)))
         tf.summary.scalar('mrr', self.mrr)
diff --git a/graphsage/unsupervised_train.py b/graphsage/unsupervised_train.py
index 44ef6091..dd4f0efd 100644
--- a/graphsage/unsupervised_train.py
+++ b/graphsage/unsupervised_train.py
@@ -118,6 +118,13 @@ def save_val_embeddings(sess, model, minibatch_iter, size, out_dir, mod=""):
 
 def construct_placeholders():
     # Define placeholders
+    '''
+    batch1和batch2分别代表一条边的两个端点，在后续的处理中用作正样本对
+    neg_samples 是负样本的数量
+    dropout 是特征传入下一层的时候的丢弃概率，有助于模型的泛化性能
+
+    '''
+
     placeholders = {
         'batch1' : tf.placeholder(tf.int32, shape=(None), name='batch1'),
         'batch2' : tf.placeholder(tf.int32, shape=(None), name='batch2'),
@@ -130,30 +137,39 @@ def construct_placeholders():
     return placeholders
 
 def train(train_data, test_data=None):
+
+    #读取图、节点特征、节点映射表
+
     G = train_data[0]
-    features = train_data[1]
+    features = train_data[1]  # shape = [num, 50]
     id_map = train_data[2]
 
     if not features is None:
-        # pad with dummy zero vector
+        # pad with dummy zero vector，features的特征加一列全0
         features = np.vstack([features, np.zeros((features.shape[1],))])
 
+
+    # 根据开关判断是否加入随机游走的路线
     context_pairs = train_data[3] if FLAGS.random_context else None
+
+    # 定义一些占位符
     placeholders = construct_placeholders()
+
+    # 创建一个边batch迭代器类，每个batch是一条边，边上的两个点的id作为模型的输入
     minibatch = EdgeMinibatchIterator(G, 
             id_map,
             placeholders, batch_size=FLAGS.batch_size,
             max_degree=FLAGS.max_degree, 
-            num_neg_samples=FLAGS.neg_sample_size,
-            context_pairs = context_pairs)
-    adj_info_ph = tf.placeholder(tf.int32, shape=minibatch.adj.shape)
+            num_neg_samples=FLAGS.neg_sample_size,   # num_neg_samples 这个形参在这里没有用到
+            context_pairs = context_pairs) 
+    adj_info_ph = tf.placeholder(tf.int32, shape=minibatch.adj.shape)   # 邻接表
     adj_info = tf.Variable(adj_info_ph, trainable=False, name="adj_info")
 
     if FLAGS.model == 'graphsage_mean':
         # Create model
         sampler = UniformNeighborSampler(adj_info)
         layer_infos = [SAGEInfo("node", sampler, FLAGS.samples_1, FLAGS.dim_1),
-                            SAGEInfo("node", sampler, FLAGS.samples_2, FLAGS.dim_2)]
+                            SAGEInfo("node", sampler, FLAGS.samples_2, FLAGS.dim_2)] # samples1 =25， sample2=10
 
         model = SampleAndAggregate(placeholders, 
                                      features,
@@ -265,15 +281,15 @@ def train(train_data, test_data=None):
         epoch_val_costs.append(0)
         while not minibatch.end():
             # Construct feed dictionary
-            feed_dict = minibatch.next_minibatch_feed_dict()
+            feed_dict = minibatch.next_minibatch_feed_dict()  # 按batch读取数据
             feed_dict.update({placeholders['dropout']: FLAGS.dropout})
 
             t = time.time()
             # Training step
             outs = sess.run([merged, model.opt_op, model.loss, model.ranks, model.aff_all, 
-                    model.mrr, model.outputs1], feed_dict=feed_dict)
+                    model.mrr, model.outputs1], feed_dict=feed_dict)     # 训练
             train_cost = outs[2]
-            train_mrr = outs[5]
+            train_mrr = outs[5]   
             if train_shadow_mrr is None:
                 train_shadow_mrr = train_mrr#
             else:

From 0bad0a0f7219b68723b72ee48f014796f988557e Mon Sep 17 00:00:00 2001
From: opeceipeno <hbyspy2008@qq.com>
Date: Fri, 25 Feb 2022 13:46:21 +0800
Subject: [PATCH 02/28] =?UTF-8?q?=E4=BB=A3=E7=A0=81=E6=B3=A8=E9=87=8A?=
 =?UTF-8?q?=E6=9B=B4=E6=96=B0=EF=BC=8Cpy?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/unsupervised_train.py | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)

diff --git a/graphsage/unsupervised_train.py b/graphsage/unsupervised_train.py
index dd4f0efd..5d0886dc 100644
--- a/graphsage/unsupervised_train.py
+++ b/graphsage/unsupervised_train.py
@@ -149,7 +149,7 @@ def train(train_data, test_data=None):
         features = np.vstack([features, np.zeros((features.shape[1],))])
 
 
-    # 根据开关判断是否加入随机游走的路线
+    # 根据开关判断是否加入随机游走的信息
     context_pairs = train_data[3] if FLAGS.random_context else None
 
     # 定义一些占位符
@@ -162,7 +162,9 @@ def train(train_data, test_data=None):
             max_degree=FLAGS.max_degree, 
             num_neg_samples=FLAGS.neg_sample_size,   # num_neg_samples 这个形参在这里没有用到
             context_pairs = context_pairs) 
-    adj_info_ph = tf.placeholder(tf.int32, shape=minibatch.adj.shape)   # 邻接表
+
+    # 根据邻接表的维度创建一个占位符
+    adj_info_ph = tf.placeholder(tf.int32, shape=minibatch.adj.shape)   
     adj_info = tf.Variable(adj_info_ph, trainable=False, name="adj_info")
 
     if FLAGS.model == 'graphsage_mean':
@@ -249,6 +251,8 @@ def train(train_data, test_data=None):
     else:
         raise Exception('Error: model name unrecognized.')
 
+
+
     config = tf.ConfigProto(log_device_placement=FLAGS.log_device_placement)
     config.gpu_options.allow_growth = True
     #config.gpu_options.per_process_gpu_memory_fraction = GPU_MEM_FRACTION
@@ -262,7 +266,7 @@ def train(train_data, test_data=None):
     # Init variables
     sess.run(tf.global_variables_initializer(), feed_dict={adj_info_ph: minibatch.adj})
     
-    # Train model
+    # Train model 训练
     
     train_shadow_mrr = None
     shadow_mrr = None
@@ -271,8 +275,10 @@ def train(train_data, test_data=None):
     avg_time = 0.0
     epoch_val_costs = []
 
+    # 从minibatch获取训练和验证数据的邻接表信息
     train_adj_info = tf.assign(adj_info, minibatch.adj)
     val_adj_info = tf.assign(adj_info, minibatch.test_adj)
+
     for epoch in range(FLAGS.epochs): 
         minibatch.shuffle() 
 
@@ -281,11 +287,14 @@ def train(train_data, test_data=None):
         epoch_val_costs.append(0)
         while not minibatch.end():
             # Construct feed dictionary
-            feed_dict = minibatch.next_minibatch_feed_dict()  # 按batch读取数据
+            # 按batch读取数据，每个batch包含边的两个端点，分别放进batch1和batch2中，
+            feed_dict = minibatch.next_minibatch_feed_dict()  
             feed_dict.update({placeholders['dropout']: FLAGS.dropout})
 
             t = time.time()
-            # Training step
+
+            # Training step 训练
+            # model.opt_op是调参 
             outs = sess.run([merged, model.opt_op, model.loss, model.ranks, model.aff_all, 
                     model.mrr, model.outputs1], feed_dict=feed_dict)     # 训练
             train_cost = outs[2]
@@ -296,9 +305,10 @@ def train(train_data, test_data=None):
                 train_shadow_mrr -= (1-0.99) * (train_shadow_mrr - train_mrr)
 
             if iter % FLAGS.validate_iter == 0:
-                # Validation
+                # Validation 验证的时候，需要用验证集的邻接表信息，因此这里跑一下tf.assign的操作
                 sess.run(val_adj_info.op)
                 val_cost, ranks, val_mrr, duration  = evaluate(sess, model, minibatch, size=FLAGS.validate_batch_size)
+                # 验证完毕再切回训练集的邻接表信息
                 sess.run(train_adj_info.op)
                 epoch_val_costs[-1] += val_cost
             if shadow_mrr is None:

From d8f85b90905695ee182a024c205bac61427b36ca Mon Sep 17 00:00:00 2001
From: 13929520142 <wangqy94@163.com>
Date: Fri, 25 Feb 2022 14:37:31 +0800
Subject: [PATCH 03/28] =?UTF-8?q?=E6=B5=8B=E8=AF=95push?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/aggregators.py | 151 ++++++++++++++++++++-------------------
 1 file changed, 79 insertions(+), 72 deletions(-)

diff --git a/graphsage/aggregators.py b/graphsage/aggregators.py
index 7dbd2523..4dcff553 100644
--- a/graphsage/aggregators.py
+++ b/graphsage/aggregators.py
@@ -3,14 +3,16 @@
 from .layers import Layer, Dense
 from .inits import glorot, zeros
 
+
 class MeanAggregator(Layer):
     """
     Aggregates via mean followed by matmul and non-linearity.
     """
 
+    # mean聚合
     def __init__(self, input_dim, output_dim, neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, 
-            name=None, concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu,
+                 name=None, concat=False, **kwargs):
         super(MeanAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -28,9 +30,9 @@ def __init__(self, input_dim, output_dim, neigh_input_dim=None,
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([neigh_input_dim, output_dim],
-                                                        name='neigh_weights')
+                                                name='neigh_weights')
             self.vars['self_weights'] = glorot([input_dim, output_dim],
-                                                        name='self_weights')
+                                               name='self_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -43,15 +45,15 @@ def __init__(self, input_dim, output_dim, neigh_input_dim=None,
     def _call(self, inputs):
         self_vecs, neigh_vecs = inputs
 
-        neigh_vecs = tf.nn.dropout(neigh_vecs, 1-self.dropout)
-        self_vecs = tf.nn.dropout(self_vecs, 1-self.dropout)
+        neigh_vecs = tf.nn.dropout(neigh_vecs, 1 - self.dropout)
+        self_vecs = tf.nn.dropout(self_vecs, 1 - self.dropout)
         neigh_means = tf.reduce_mean(neigh_vecs, axis=1)
-       
+
         # [nodes] x [out_dim]
         from_neighs = tf.matmul(neigh_means, self.vars['neigh_weights'])
 
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
-         
+
         if not self.concat:
             output = tf.add_n([from_self, from_neighs])
         else:
@@ -60,9 +62,10 @@ def _call(self, inputs):
         # bias
         if self.bias:
             output += self.vars['bias']
-       
+
         return self.act(output)
 
+
 class GCNAggregator(Layer):
     """
     Aggregates via mean followed by matmul and non-linearity.
@@ -70,7 +73,7 @@ class GCNAggregator(Layer):
     """
 
     def __init__(self, input_dim, output_dim, neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
         super(GCNAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -88,7 +91,7 @@ def __init__(self, input_dim, output_dim, neigh_input_dim=None,
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['weights'] = glorot([neigh_input_dim, output_dim],
-                                                        name='neigh_weights')
+                                          name='neigh_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -101,26 +104,27 @@ def __init__(self, input_dim, output_dim, neigh_input_dim=None,
     def _call(self, inputs):
         self_vecs, neigh_vecs = inputs
 
-        neigh_vecs = tf.nn.dropout(neigh_vecs, 1-self.dropout)
-        self_vecs = tf.nn.dropout(self_vecs, 1-self.dropout)
-        means = tf.reduce_mean(tf.concat([neigh_vecs, 
-            tf.expand_dims(self_vecs, axis=1)], axis=1), axis=1)
-       
+        neigh_vecs = tf.nn.dropout(neigh_vecs, 1 - self.dropout)
+        self_vecs = tf.nn.dropout(self_vecs, 1 - self.dropout)
+        means = tf.reduce_mean(tf.concat([neigh_vecs,
+                                          tf.expand_dims(self_vecs, axis=1)], axis=1), axis=1)
+
         # [nodes] x [out_dim]
         output = tf.matmul(means, self.vars['weights'])
 
         # bias
         if self.bias:
             output += self.vars['bias']
-       
+
         return self.act(output)
 
 
 class MaxPoolingAggregator(Layer):
     """ Aggregates via max-pooling over MLP functions.
     """
+
     def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
         super(MaxPoolingAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -143,18 +147,18 @@ def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=No
 
         self.mlp_layers = []
         self.mlp_layers.append(Dense(input_dim=neigh_input_dim,
-                                 output_dim=hidden_dim,
-                                 act=tf.nn.relu,
-                                 dropout=dropout,
-                                 sparse_inputs=False,
-                                 logging=self.logging))
+                                     output_dim=hidden_dim,
+                                     act=tf.nn.relu,
+                                     dropout=dropout,
+                                     sparse_inputs=False,
+                                     logging=self.logging))
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([hidden_dim, output_dim],
-                                                        name='neigh_weights')
-           
+                                                name='neigh_weights')
+
             self.vars['self_weights'] = glorot([input_dim, output_dim],
-                                                        name='self_weights')
+                                               name='self_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -179,10 +183,10 @@ def _call(self, inputs):
             h_reshaped = l(h_reshaped)
         neigh_h = tf.reshape(h_reshaped, (batch_size, num_neighbors, self.hidden_dim))
         neigh_h = tf.reduce_max(neigh_h, axis=1)
-        
+
         from_neighs = tf.matmul(neigh_h, self.vars['neigh_weights'])
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
-        
+
         if not self.concat:
             output = tf.add_n([from_self, from_neighs])
         else:
@@ -191,14 +195,16 @@ def _call(self, inputs):
         # bias
         if self.bias:
             output += self.vars['bias']
-       
+
         return self.act(output)
 
+
 class MeanPoolingAggregator(Layer):
     """ Aggregates via mean-pooling over MLP functions.
     """
+
     def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
         super(MeanPoolingAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -221,18 +227,18 @@ def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=No
 
         self.mlp_layers = []
         self.mlp_layers.append(Dense(input_dim=neigh_input_dim,
-                                 output_dim=hidden_dim,
-                                 act=tf.nn.relu,
-                                 dropout=dropout,
-                                 sparse_inputs=False,
-                                 logging=self.logging))
+                                     output_dim=hidden_dim,
+                                     act=tf.nn.relu,
+                                     dropout=dropout,
+                                     sparse_inputs=False,
+                                     logging=self.logging))
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([hidden_dim, output_dim],
-                                                        name='neigh_weights')
-           
+                                                name='neigh_weights')
+
             self.vars['self_weights'] = glorot([input_dim, output_dim],
-                                                        name='self_weights')
+                                               name='self_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -257,10 +263,10 @@ def _call(self, inputs):
             h_reshaped = l(h_reshaped)
         neigh_h = tf.reshape(h_reshaped, (batch_size, num_neighbors, self.hidden_dim))
         neigh_h = tf.reduce_mean(neigh_h, axis=1)
-        
+
         from_neighs = tf.matmul(neigh_h, self.vars['neigh_weights'])
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
-        
+
         if not self.concat:
             output = tf.add_n([from_self, from_neighs])
         else:
@@ -269,15 +275,16 @@ def _call(self, inputs):
         # bias
         if self.bias:
             output += self.vars['bias']
-       
+
         return self.act(output)
 
 
 class TwoMaxLayerPoolingAggregator(Layer):
     """ Aggregates via pooling over two MLP functions.
     """
+
     def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
         super(TwoMaxLayerPoolingAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -302,25 +309,24 @@ def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=No
 
         self.mlp_layers = []
         self.mlp_layers.append(Dense(input_dim=neigh_input_dim,
-                                 output_dim=hidden_dim_1,
-                                 act=tf.nn.relu,
-                                 dropout=dropout,
-                                 sparse_inputs=False,
-                                 logging=self.logging))
+                                     output_dim=hidden_dim_1,
+                                     act=tf.nn.relu,
+                                     dropout=dropout,
+                                     sparse_inputs=False,
+                                     logging=self.logging))
         self.mlp_layers.append(Dense(input_dim=hidden_dim_1,
-                                 output_dim=hidden_dim_2,
-                                 act=tf.nn.relu,
-                                 dropout=dropout,
-                                 sparse_inputs=False,
-                                 logging=self.logging))
-
+                                     output_dim=hidden_dim_2,
+                                     act=tf.nn.relu,
+                                     dropout=dropout,
+                                     sparse_inputs=False,
+                                     logging=self.logging))
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([hidden_dim_2, output_dim],
-                                                        name='neigh_weights')
-           
+                                                name='neigh_weights')
+
             self.vars['self_weights'] = glorot([input_dim, output_dim],
-                                                        name='self_weights')
+                                               name='self_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -345,10 +351,10 @@ def _call(self, inputs):
             h_reshaped = l(h_reshaped)
         neigh_h = tf.reshape(h_reshaped, (batch_size, num_neighbors, self.hidden_dim_2))
         neigh_h = tf.reduce_max(neigh_h, axis=1)
-        
+
         from_neighs = tf.matmul(neigh_h, self.vars['neigh_weights'])
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
-        
+
         if not self.concat:
             output = tf.add_n([from_self, from_neighs])
         else:
@@ -357,14 +363,16 @@ def _call(self, inputs):
         # bias
         if self.bias:
             output += self.vars['bias']
-       
+
         return self.act(output)
 
+
 class SeqAggregator(Layer):
     """ Aggregates via a standard LSTM.
     """
+
     def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=None,
-            dropout=0., bias=False, act=tf.nn.relu, name=None,  concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
         super(SeqAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout
@@ -387,10 +395,10 @@ def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=No
 
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([hidden_dim, output_dim],
-                                                        name='neigh_weights')
-           
+                                                name='neigh_weights')
+
             self.vars['self_weights'] = glorot([input_dim, output_dim],
-                                                        name='self_weights')
+                                               name='self_weights')
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
@@ -416,15 +424,15 @@ def _call(self, inputs):
         with tf.variable_scope(self.name) as scope:
             try:
                 rnn_outputs, rnn_states = tf.nn.dynamic_rnn(
-                        self.cell, neigh_vecs,
-                        initial_state=initial_state, dtype=tf.float32, time_major=False,
-                        sequence_length=length)
+                    self.cell, neigh_vecs,
+                    initial_state=initial_state, dtype=tf.float32, time_major=False,
+                    sequence_length=length)
             except ValueError:
                 scope.reuse_variables()
                 rnn_outputs, rnn_states = tf.nn.dynamic_rnn(
-                        self.cell, neigh_vecs,
-                        initial_state=initial_state, dtype=tf.float32, time_major=False,
-                        sequence_length=length)
+                    self.cell, neigh_vecs,
+                    initial_state=initial_state, dtype=tf.float32, time_major=False,
+                    sequence_length=length)
         batch_size = tf.shape(rnn_outputs)[0]
         max_len = tf.shape(rnn_outputs)[1]
         out_size = int(rnn_outputs.get_shape()[2])
@@ -434,7 +442,7 @@ def _call(self, inputs):
 
         from_neighs = tf.matmul(neigh_h, self.vars['neigh_weights'])
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
-         
+
         output = tf.add_n([from_self, from_neighs])
 
         if not self.concat:
@@ -445,6 +453,5 @@ def _call(self, inputs):
         # bias
         if self.bias:
             output += self.vars['bias']
-       
-        return self.act(output)
 
+        return self.act(output)

From 3bb58c8f4f4ec51924db34a452307b0956a57825 Mon Sep 17 00:00:00 2001
From: opeceipeno <hbyspy2008@qq.com>
Date: Fri, 25 Feb 2022 17:31:31 +0800
Subject: [PATCH 04/28] =?UTF-8?q?=E6=9B=B4=E6=96=B0=E6=B3=A8=E9=87=8A?=
 =?UTF-8?q?=EF=BC=8C=E5=8C=85=E6=8B=AC=E4=BA=86Layers.py=20Prediction.py?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/layers.py             |  26 +++++-
 graphsage/models.py             | 151 ++++++++++++++++++--------------
 graphsage/prediction.py         |  45 +++++++---
 graphsage/unsupervised_train.py |  19 +++-
 4 files changed, 159 insertions(+), 82 deletions(-)

diff --git a/graphsage/layers.py b/graphsage/layers.py
index d1b7b37e..24d2af8f 100644
--- a/graphsage/layers.py
+++ b/graphsage/layers.py
@@ -39,6 +39,11 @@ class Layer(object):
             (i.e. takes input, returns output)
         __call__(inputs): Wrapper for _call()
         _log_vars(): Log all variables
+
+        最基础的层类型
+        name：定义层的名称，字符型
+        logging：布尔型，如果开的话就可以打印训练过程中，当需要查看一个张量在训练过程中值的分布情况时，可通过tf.summary.histogram()将其分布情况以直方图的形式在TensorBoard直方图仪表板上显示．
+        这里并没有参数矩阵、激活函数等值，都是在其子类中实现
     """
 
     def __init__(self, **kwargs):
@@ -55,6 +60,7 @@ def __init__(self, **kwargs):
         self.logging = logging
         self.sparse_inputs = False
 
+    # 用于定义该层的计算逻辑的函数，该类这里没计算逻辑，需要子类去实， Dense类就有一个实现
     def _call(self, inputs):
         return inputs
 
@@ -73,7 +79,20 @@ def _log_vars(self):
 
 
 class Dense(Layer):
-    """Dense layer."""
+    """Dense layer.
+    一个基础的全连接层
+    需要的入参：
+        dropout ：0-1的数字，表示输入被丢弃的概率
+        act：激活函数的类型选取，例如sigmoid、relu等
+        featureless：意义暂时不明，没有用到
+        bias ：偏置项
+        input_dim：输入维度
+        output_dim ：输出维度
+
+    """
+
+
+
 
     def __init__(self, input_dim, output_dim, dropout=0.,
                  act=tf.nn.relu, placeholders=None, bias=True, featureless=False,
@@ -93,6 +112,8 @@ def __init__(self, input_dim, output_dim, dropout=0.,
         if sparse_inputs:
             self.num_features_nonzero = placeholders['num_features_nonzero']
 
+
+        # 根据输入输出维度生成参数矩阵
         with tf.variable_scope(self.name + '_vars'):
             self.vars['weights'] = tf.get_variable('weights', shape=(input_dim, output_dim),
                                                    dtype=tf.float32,
@@ -104,6 +125,9 @@ def __init__(self, input_dim, output_dim, dropout=0.,
         if self.logging:
             self._log_vars()
 
+
+    #调用模型的时候的计算逻辑定义
+    # 此全连接中， output = XW + b, (还有dropout和激活函数等操作)
     def _call(self, inputs):
         x = inputs
 
diff --git a/graphsage/models.py b/graphsage/models.py
index 018fc47c..e3ea8e83 100644
--- a/graphsage/models.py
+++ b/graphsage/models.py
@@ -56,7 +56,7 @@ def build(self):
         # Build sequential layer model
         self.activations.append(self.inputs)    # 第一步，将输入数据加进激活层，作为第一层
         for layer in self.layers:
-            # 逐层计算中间层，并将每一层的输出都放进activations中保存
+            # 逐层计算，并将每一层的输出都放进activations中保存
             hidden = layer(self.activations[-1])
             self.activations.append(hidden)
         self.outputs = self.activations[-1]   # 模型的输出即为最后一层
@@ -175,6 +175,8 @@ def build(self):
             self._build()
 
         # Store model variables for easy access
+        # 和Model类相比，GeneralizedModel在build的时候，并没去生成序列层
+        # self.output必须在它的子类build()函数中实现。
         variables = tf.get_collection(
             tf.GraphKeys.GLOBAL_VARIABLES, scope=self.name)
         self.vars = {var.name: var for var in variables}
@@ -219,14 +221,13 @@ def __init__(self, placeholders, features, adj, degrees,
             - model_size: one of "small" and "big"
             - identity_dim: Set to positive int to use identity features (slow and cannot generalize, but better accuracy)
 
-             - features：节点特征 [num_nodes,50]
-             - adj： 图的邻接表， [num_nodes,maxdegree]
-             - degrees：列表，表示每个节点的度数[num_nodes]
-             - layer_infos：描述所有参数的 SAGEInfo 命名元组列表
-                    递归层。 参见上面的 SAGEInfo 定义。
+             - features：节点特征 [num_nodes,num_features]
+             - adj： 图的邻接表， [num_nodes, maxdegree] maxdegree是个超参，表示对于每个节点，最多只记录其maxdegree个邻居信息
+             - degrees：列表，表示每个节点的度数长度为[num_nodes]
+             - layer_infos：一个列表，记录了每一层的信息包括，名称、邻居采样器、
              - concat：是否在递归迭代期间拼接，是或者否
              - aggregator_type：聚合方式的定义
-             - model_size：模型大小,有small 和big，  
+             - model_size：模型大小,有small 和big， 隐藏层的维度有区别
              - identity_dim：int，若＞0则加入额外特征（速度慢且泛化性差，但准确度更高）
         '''
 
@@ -246,9 +247,11 @@ def __init__(self, placeholders, features, adj, degrees,
             raise Exception("Unknown aggregator: ", self.aggregator_cls)
 
         # get info from placeholders...
-        # 两个输入 ，batch1和batch2 是一条边的两个顶点id，即每条边的两个顶点，分别放进batch1和batch2中，用作正样本
+        # batch1和batch2 是一条边的两个顶点id，即每条边的两个顶点，分别放进batch1和batch2中
+        # 他们后续会分别作为模型的输入，得到中间表达结果output1和output2，然后在会用表达结果计算性能指标
         self.inputs1 = placeholders["batch1"]
         self.inputs2 = placeholders["batch2"]
+
         self.model_size = model_size
         self.adj_info = adj
 
@@ -265,14 +268,15 @@ def __init__(self, placeholders, features, adj, degrees,
             self.features = self.embeds
         else:
             self.features = tf.Variable(tf.constant(
-                features, dtype=tf.float32), trainable=False)  # 节点特征，
+                features, dtype=tf.float32), trainable=False)  # 节点特征通过tf.Variable方式获取，不可训练
             if not self.embeds is None:
                 # (feature的最终特征维度为 原始特征维度50+identity_dim)
                 self.features = tf.concat([self.embeds, self.features], axis=1)
+
         self.degrees = degrees
-        self.concat = concat  # 布尔值，是否拼接
+        self.concat = concat  # 布尔值，表示在模型计算完batch1和batch2的特征表达之后，是否拼接
 
-        # dim是一个列表，代表每一层的输出维度，第一层是输入层，维度=输入特征的维度，后面的维度是从layer_info得到的
+        # dim是一个列表，代表aggregator每一层的输出维度，第一层是输入层，维度=输入特征的维度，后面的维度是从layer_info得到的
         # 本实验中，dims = [50，128，128]
 
         self.dims = [
@@ -283,7 +287,8 @@ def __init__(self, placeholders, features, adj, degrees,
         self.placeholders = placeholders
         self.layer_infos = layer_infos
 
-        # 优化器选择
+        # 优化器选择为adam方法，是当前最常用的梯度更新策略
+
         self.optimizer = tf.train.AdamOptimizer(
             learning_rate=FLAGS.learning_rate)
 
@@ -293,10 +298,20 @@ def __init__(self, placeholders, features, adj, degrees,
     def sample(self, inputs, layer_infos, batch_size=None):
         """ Sample neighbors to be the supportive fields for multi-layer convolutions.
 
-        对节点邻居采样，作为该点的是支持域
-        samples[0] 维度是 [batch_size]
-        samples[1] [layer_infos[1].num_samples * batch_size]
-        samples[2] [layer_infos[1].num_samples * layer_infos[0].num_samples * batch_size]
+        函数功能：对输入的每一个节点，根据采样跳数目，递归地采样邻居，作为该节点的支持域
+
+        samples是一个列表，列表的每一个元素又是一个列表，长度不一，存放的是该跳数下的所有的邻居节点id
+        示例：
+        samples[0] 维度是 [batch_size,] ，即是自身
+        samples[1] [layer_infos[1].num_samples * batch_size,]
+        samples[2] [layer_infos[1].num_samples * layer_infos[0].num_samples * batch_size,]
+        以此类推
+
+        # support_sizes 存的是的各层的采样数目，是一个列表，每个元素是一个正整数
+        # support_sizes[0] = 1， 意义是初始状态，邻居就是节点本身
+        # support_sizes[1] = layer_infos[-1].num_samples * 1， 本实验中为10
+        # support_sizes[2] = layer_infos[-1].num_samples * layer_infos[-2].num_samples * 1， 本实验中为10*15=250
+        # 以此类推，从最外层的邻居数依次往内乘
 
         Args:
             inputs: batch inputs
@@ -307,25 +322,25 @@ def sample(self, inputs, layer_infos, batch_size=None):
             batch_size = self.batch_size
         samples = [inputs]  # samples[0] 是输入，
         # size of convolution support at each layer per node
-        # support_sizes 存的是的各层的采样数目，是一个列表
-        # support_sizes[0] = 1， 初始状态就是节点本身
-        # support_sizes[1] = layer_infos[-1].num_samples * 1， 本实验中为10
-        # support_sizes[2] = layer_infos[-1].num_samples * layer_infos[-2].num_samples * 1， 本实验中为250
-        # 以此类推，从最外层的邻居数依次往内乘
 
         support_size = 1
         support_sizes = [support_size]
 
         for k in range(len(layer_infos)):  # k为跳数，也是层数， 实验中k = 0 1
             t = len(layer_infos) - k - 1  # t = 1 0
-            # 每一跳的邻居数目是前一跳的邻居节点数*该层的采样数
+
+            # 每一跳的邻居数目是前一跳的邻居节点数*该层的采样数，有个累乘的逻辑
             support_size *= layer_infos[t].num_samples
-            sampler = layer_infos[t].neigh_sampler  # 采样器选择，实验中采样器是同一种
 
-            # 采样器的两个输入，第一个是将被采样的节点id，第二个是采样数
+            sampler = layer_infos[t].neigh_sampler  # 采样器选择
+
+            # 采样器的两个输入，第一个入参是将要被采样的节点id，第二个入参是对这些节点，要采样多少个邻居
             node = sampler((samples[k], layer_infos[t].num_samples))
-            # reshape并放进数组
+
+            # reshape成一维数组，再添加进samples中
             samples.append(tf.reshape(node, [support_size * batch_size, ]))
+
+            # 同时记录好每一层的采样数
             support_sizes.append(support_size)
         return samples, support_sizes
 
@@ -351,11 +366,9 @@ def aggregate(self, samples, input_features, dims, num_samples, support_sizes, b
                 sample[2]是对sample[1]中每一个节点进行邻居采样，即第2跳采样
                 以此类推
         input_features: 矩阵，存放的是全量的节点的特征 
-        dims: 列表，代表每一层的中间表达的维度
 
         num_samples: 列表，表示模型每一层的邻居采样数目，实验中为[25,10]
-        support_sizes: the number of nodes to gather information from for each layer.
-        batch_size: the number of inputs (different for batch inputs and negative samples).
+
         Returns:
 
 
@@ -367,31 +380,36 @@ def aggregate(self, samples, input_features, dims, num_samples, support_sizes, b
             batch_size = self.batch_size
 
         # length: number of layers + 1
-        # 根据节点id，从全量的特征矩阵里获取节点特征
+        # 遍历samples列表，根据每一个元素中存放的节点id，从全量的特征矩阵里获取所需的节点特征
+     
+        hidden = [tf.nn.embedding_lookup(
+            input_features, node_samples) for node_samples in samples]
         # hidden[0] [batch, num_features]
         # hidden[1] [layer_infos[1].num_samples * batch_size, num_features]
         # hidden[2] [layer_infos[1].num_samples * layer_infos[0].num_samples * batch_size, num_features]
         # num_features表示的是特征维度，实验中为50
-
-        hidden = [tf.nn.embedding_lookup(
-            input_features, node_samples) for node_samples in samples]
         
-        # 输入batch1的时候，该项为aggregators = None， 输入batch2或者neg_samples的时候，aggregators为batch1的aggregators
+        
+        # 输入batch1的时候，该项为aggregators = None， 输入batch2或者neg_samples的时候，aggregators为batch1生成的aggregators
+        # 即他们用的是同一个聚合器
         new_agg = aggregators is None
 
         if new_agg:
             aggregators = []
         for layer in range(len(num_samples)):   # 按层数循环
             if new_agg:
-                dim_mult = 2 if concat and (layer != 0) else 1  # 维度系数，如果有concat=True，从第二层开始，输入维度需要乘2
+                dim_mult = 2 if concat and (layer != 0) else 1
                 # aggregator at current layer
+                # 根据给定的参数，初始化一个聚合器类，
+                # 其中，聚合器有多种选择，是由超参定义的，
+                # 另外需要的参数是输入维度、输出维度、dropout系数等等
+                # 注意输入维度前面有个dim_mult，该值为1或者2，如果concat=True，表示节点自身的结果和邻居的会拼接一下，则从第二层开始，输入维度需要乘2
                 # 判断是否是最后一层，如果是的话，会有个参数act=lambda x: x
                 if layer == len(num_samples) - 1:
                     aggregator = self.aggregator_cls(dim_mult*dims[layer], dims[layer+1], act=lambda x: x,
                                                      dropout=self.placeholders['dropout'],
                                                      name=name, concat=concat, model_size=model_size)
                 else:
-                    # 初始化一个聚合器类，类别是超参定义的，需要的参数是输入维度、输出维度、dropout系数等等
                     aggregator = self.aggregator_cls(dim_mult*dims[layer], dims[layer+1],
                                                      dropout=self.placeholders['dropout'],
                                                      name=name, concat=concat, model_size=model_size)
@@ -399,45 +417,44 @@ def aggregate(self, samples, input_features, dims, num_samples, support_sizes, b
             else:
                 # 在batch2或者neg_samples输入的时候，直接使用已有的聚合器
                 aggregator = aggregators[layer]
-            
+
             # 本实验中，aggregator1 的输入输出维度分别为：50，256， 参数矩阵维度为50，128 ，后面有个拼接
             # aggregator2 的输入输出维度为：256，256，参数矩阵维度为256，128
 
-
-
             # hidden representation at current layer for all support nodes that are various hops away
+            # 该变量存放的是当前层，各节点利用邻居节点的信息更新后的中间表达，
             next_hidden = []
-            
-            
+
             # as layer increases, the number of support nodes needed decreases
             # 随着层数增加，跳数需要减少
-            for hop in range(len(num_samples) - layer): 
+            for hop in range(len(num_samples) - layer):
                 dim_mult = 2 if concat and (layer != 0) else 1
 
                 # 每个节点的特征，是由自身的特征和其邻居节点的特征聚合而来的，
                 # hidden[hop+1]包含了hidden[hop]中节点的所有邻居特征
-                # 因为hidden[i]是二维的，而mean_aggregator是需要将邻居节点特征平均，
+                # 因为hidden[i]存放为二维，而mean_aggregator是需要将邻居节点特征平均，
                 # 因此需要将它reshape一下，方便在后面的处理中取所有邻居的均值
                 # neigh_dims = [batch_size * 当前跳数的支持节点数，当前层的需要采样的邻居节点数，特征数]
-
-                # 
+                #  
                 neigh_dims = [batch_size * support_sizes[hop],
                               num_samples[len(num_samples) - hop - 1],
                               dim_mult*dims[layer]]
                 h = aggregator((hidden[hop],
-                                tf.reshape(hidden[hop + 1], neigh_dims)))   
+                                tf.reshape(hidden[hop + 1], neigh_dims)))
                 next_hidden.append(h)
             hidden = next_hidden
+
+        # 输出的hidden[0]，本实验中，shape=[batch_size,128*2]
         return hidden[0], aggregators
 
     def _build(self):
 
-        # 将第batch2作为标签，即batch1和batch2是一对正样本对
+        # 将第batch2视为标签，即batch1和batch2是一对正样本对
         labels = tf.reshape(
             tf.cast(self.placeholders['batch2'], dtype=tf.int64),
             [self.batch_size, 1])
 
-        # 获取负样本， 按照给定的概率分布进行采样
+        # 获取负样本， 按照给定的概率分布unigrams进行采样
         self.neg_samples, _, _ = (tf.nn.fixed_unigram_candidate_sampler(
             true_classes=labels,
             num_true=1,
@@ -450,24 +467,24 @@ def _build(self):
         # perform "convolution"
 
         # 根据节点id去采样其邻居节点id
+        # 返回的结果：samples，support_sizes
         samples1, support_sizes1 = self.sample(self.inputs1, self.layer_infos)
         samples2, support_sizes2 = self.sample(self.inputs2, self.layer_infos)
-        
+
         # 每层需要的采样数 实验中是[25,10]
         num_samples = [
             layer_info.num_samples for layer_info in self.layer_infos]
 
-
         # 获取batch1的特征表达，该步传入的聚合器参数为None，会构建一个聚合器返回
         self.outputs1, self.aggregators = self.aggregate(samples1, [self.features], self.dims, num_samples,
                                                          support_sizes1, concat=self.concat, model_size=self.model_size)
-        
-        # 获取batch2的特征表达，其中聚合器是直接使用上一步生成的
+
+        # 获取batch2的特征表达，其中聚合器是直接使用上一步生成的：aggregators=self.aggregators
         self.outputs2, _ = self.aggregate(samples2, [self.features], self.dims, num_samples,
                                           support_sizes2, aggregators=self.aggregators, concat=self.concat,
                                           model_size=self.model_size)
-        
-        # 对负样本采样
+
+        # 对负样本做邻居节点采样，和上面的正样本是同样的处理
         neg_samples, neg_support_sizes = self.sample(self.neg_samples, self.layer_infos,
                                                      FLAGS.neg_sample_size)
 
@@ -478,15 +495,15 @@ def _build(self):
 
         dim_mult = 2 if self.concat else 1
 
+
         # 这里生成了一个预测层，注意参数bilinear_weights,这个值如果为True，则会生成一个可训练的参数矩阵，在后续的计算loss会用到
-        # 在这里设置了否，则无参数矩阵，本质上就是一个计算loss的类
+        # 但是本实验在这里设置了否，则无参数矩阵，本质上就是一个计算loss的类，完全不影响上述aggregator的输出
         self.link_pred_layer = BipartiteEdgePredLayer(dim_mult*self.dims[-1],
                                                       dim_mult*self.dims[-1], self.placeholders, act=tf.nn.sigmoid,
                                                       bilinear_weights=False,
                                                       name='edge_predict')
 
-
-        # 对输出的样本执行L2规范化，dim=1，表示按行做，即按样本点做
+        # 对输出的样本执行L2规范化，dim=0或者1，1是表示按行做
         # x_l2[i] = x[i]/sqrt(sum(x^2))
         self.outputs1 = tf.nn.l2_normalize(self.outputs1, 1)
         self.outputs2 = tf.nn.l2_normalize(self.outputs2, 1)
@@ -508,21 +525,23 @@ def build(self):
         # 计算梯度
         grads_and_vars = self.optimizer.compute_gradients(self.loss)
 
-        # 梯度裁剪，若梯度大于5则置为5，小于-5则置为-5
+        # 梯度裁剪，若梯度大于5则置为5，小于-5则置为-5，
         clipped_grads_and_vars = [(tf.clip_by_value(grad, -5.0, 5.0) if grad is not None else None, var)
                                   for grad, var in grads_and_vars]
+        
+        # clipped_grads_and_vars 是一个元组,(grad,var),表示梯度值和变量值
         self.grad, _ = clipped_grads_and_vars[0]
 
-        # 更新参数
+        # 利用裁剪后的梯度更新模型参数
         self.opt_op = self.optimizer.apply_gradients(clipped_grads_and_vars)
 
     def _loss(self):
-        
+
         # L2正则化项
         for aggregator in self.aggregators:
             for var in aggregator.vars.values():
                 self.loss += FLAGS.weight_decay * tf.nn.l2_loss(var)
-        
+
         # 根据之前生成的预测层，计算loss，该loss有三个选项：_xent_loss、_skipgram_loss、_hinge_loss
         self.loss += self.link_pred_layer.loss(
             self.outputs1, self.outputs2, self.neg_outputs)
@@ -538,14 +557,13 @@ def _accuracy(self):
         ③将两组数据拼接，拼接后的数组维度[batch_size, neg_samples_size + 1],意义是每一个顶点和负样本、正样本之间的"亲和度"
         ④计算正样本对之间的亲和度的排名，排名越靠前越好
         mrr值，
-        """        
-        
+        """
+
         # ①计算正样本对的"亲和度"
         # aff值在本实验即是两个输入按元素点乘，再按行求和
         # shape : [batch_size,] 表示了该batch中，每个节点和其邻居节点的“亲和度”，越大代表越相似
-        
-        aff = self.link_pred_layer.affinity(self.outputs1, self.outputs2)
 
+        aff = self.link_pred_layer.affinity(self.outputs1, self.outputs2)
 
         # ②计算顶点和负样本的"亲和度"
         # 返回的是一个矩阵，维度：[batch_size，num_neg_samples]
@@ -555,7 +573,7 @@ def _accuracy(self):
 
         self.neg_aff = tf.reshape(
             self.neg_aff, [self.batch_size, FLAGS.neg_sample_size])
-        
+
         # ③将两组数据拼接，拼接后的数组维度[batch_size, neg_samples_size + 1],意义是每一个顶点和负样本、正样本之间的"亲和度"
         # shape : [batch_size,1]
         _aff = tf.expand_dims(aff, axis=1)
@@ -568,9 +586,8 @@ def _accuracy(self):
         _, indices_of_ranks = tf.nn.top_k(self.aff_all, k=size)
         _, self.ranks = tf.nn.top_k(-indices_of_ranks, k=size)
 
-
         # 取self.ranks最后一列，即正样本的排名序数，因为是从0算起的，所以要+1
-        # mrr = 1.0/rank 
+        # mrr = 1.0/rank
         self.mrr = tf.reduce_mean(
             tf.div(1.0, tf.cast(self.ranks[:, -1] + 1, tf.float32)))
         tf.summary.scalar('mrr', self.mrr)
diff --git a/graphsage/prediction.py b/graphsage/prediction.py
index 0c00c68e..70f68170 100644
--- a/graphsage/prediction.py
+++ b/graphsage/prediction.py
@@ -11,8 +11,8 @@
 
 class BipartiteEdgePredLayer(Layer):
     def __init__(self, input_dim1, input_dim2, placeholders, dropout=False, act=tf.nn.sigmoid,
-            loss_fn='xent', neg_sample_weights=1.0,
-            bias=False, bilinear_weights=False, **kwargs):
+                 loss_fn='xent', neg_sample_weights=1.0,
+                 bias=False, bilinear_weights=False, **kwargs):
         """
         Basic class that applies skip-gram-like loss
         (i.e., dot product of node+target and node and negative samples)
@@ -21,6 +21,7 @@ def __init__(self, input_dim1, input_dim2, placeholders, dropout=False, act=tf.n
                 false, it is assumed that input dimensions are the same and the affinity will be 
                 based on dot product.
         """
+
         super(BipartiteEdgePredLayer, self).__init__(**kwargs)
         self.input_dim1 = input_dim1
         self.input_dim2 = input_dim2
@@ -44,13 +45,13 @@ def __init__(self, input_dim1, input_dim2, placeholders, dropout=False, act=tf.n
         with tf.variable_scope(self.name + '_vars'):
             # bilinear form
             if bilinear_weights:
-                #self.vars['weights'] = glorot([input_dim1, input_dim2],
+                # self.vars['weights'] = glorot([input_dim1, input_dim2],
                 #                              name='pred_weights')
                 self.vars['weights'] = tf.get_variable(
-                        'pred_weights', 
-                        shape=(input_dim1, input_dim2),
-                        dtype=tf.float32, 
-                        initializer=tf.contrib.layers.xavier_initializer())
+                    'pred_weights',
+                    shape=(input_dim1, input_dim2),
+                    dtype=tf.float32,
+                    initializer=tf.contrib.layers.xavier_initializer())
 
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
@@ -89,6 +90,7 @@ def neg_cost(self, inputs1, neg_samples, hard_neg_samples=None):
         if self.bilinear_weights:
             inputs1 = tf.matmul(inputs1, self.vars['weights'])
         neg_aff = tf.matmul(inputs1, tf.transpose(neg_samples))
+
         return neg_aff
 
     def loss(self, inputs1, inputs2, neg_samples):
@@ -96,17 +98,37 @@ def loss(self, inputs1, inputs2, neg_samples):
         Args:
             neg_samples: tensor of shape [num_neg_samples x input_dim2]. Negative samples for all
             inputs in batch inputs1.
+
         """
         return self.loss_fn(inputs1, inputs2, neg_samples)
 
     def _xent_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
+        """
+                对应论文的公式(1)
+        """
+        # 计算正样本对的亲和度
         aff = self.affinity(inputs1, inputs2)
+
+        # 计算顶点和各个负样本的亲和度
         neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
+
+
+        """
+        计算正样本的交叉熵损失，正样本label赋值全1
+        公式 = y * -log(sigmoid(aff)) + (1 - y) * -log(1 - sigmoid(aff))
+        正样本y=1，负样本y=0，分别可以省略一项
+        """
         true_xent = tf.nn.sigmoid_cross_entropy_with_logits(
-                labels=tf.ones_like(aff), logits=aff)
+            labels=tf.ones_like(aff), logits=aff)
+
+        # 计算负样本的交叉熵损失，负样本label赋值全0
         negative_xent = tf.nn.sigmoid_cross_entropy_with_logits(
-                labels=tf.zeros_like(neg_aff), logits=neg_aff)
-        loss = tf.reduce_sum(true_xent) + self.neg_sample_weights * tf.reduce_sum(negative_xent)
+            labels=tf.zeros_like(neg_aff), logits=neg_aff)
+
+
+        # neg_sample_weights 默认为1.0
+        loss = tf.reduce_sum(
+            true_xent) + self.neg_sample_weights * tf.reduce_sum(negative_xent)
         return loss
 
     def _skipgram_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
@@ -119,7 +141,8 @@ def _skipgram_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
     def _hinge_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
         aff = self.affinity(inputs1, inputs2)
         neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
-        diff = tf.nn.relu(tf.subtract(neg_aff, tf.expand_dims(aff, 1) - self.margin), name='diff')
+        diff = tf.nn.relu(tf.subtract(
+            neg_aff, tf.expand_dims(aff, 1) - self.margin), name='diff')
         loss = tf.reduce_sum(diff)
         self.neg_shape = tf.shape(neg_aff)
         return loss
diff --git a/graphsage/unsupervised_train.py b/graphsage/unsupervised_train.py
index 5d0886dc..ad870855 100644
--- a/graphsage/unsupervised_train.py
+++ b/graphsage/unsupervised_train.py
@@ -71,11 +71,16 @@ def log_dir():
 # Define model evaluation function
 def evaluate(sess, model, minibatch_iter, size=None):
     t_test = time.time()
+
+    # 评估阶段，这里传入的是验证集
     feed_dict_val = minibatch_iter.val_feed_dict(size)
     outs_val = sess.run([model.loss, model.ranks, model.mrr], 
                         feed_dict=feed_dict_val)
     return outs_val[0], outs_val[1], outs_val[2], (time.time() - t_test)
 
+
+
+# 这个函数在无监督中没有用到，暂时忽略
 def incremental_evaluate(sess, model, minibatch_iter, size):
     t_test = time.time()
     finished = False
@@ -141,7 +146,7 @@ def train(train_data, test_data=None):
     #读取图、节点特征、节点映射表
 
     G = train_data[0]
-    features = train_data[1]  # shape = [num, 50]
+    features = train_data[1]  # shape = [num_nodes, num_features]
     id_map = train_data[2]
 
     if not features is None:
@@ -168,7 +173,8 @@ def train(train_data, test_data=None):
     adj_info = tf.Variable(adj_info_ph, trainable=False, name="adj_info")
 
     if FLAGS.model == 'graphsage_mean':
-        # Create model
+        # Create model 
+        #  
         sampler = UniformNeighborSampler(adj_info)
         layer_infos = [SAGEInfo("node", sampler, FLAGS.samples_1, FLAGS.dim_1),
                             SAGEInfo("node", sampler, FLAGS.samples_2, FLAGS.dim_2)] # samples1 =25， sample2=10
@@ -294,11 +300,18 @@ def train(train_data, test_data=None):
             t = time.time()
 
             # Training step 训练
-            # model.opt_op是调参 
+            # merged是保存了的所有的tf.summary相关的变量，用于tensorboard绘图，
+            # model.opt_op 是优化调参的操作
+            # 其他的变量
+            # 
+            #  
             outs = sess.run([merged, model.opt_op, model.loss, model.ranks, model.aff_all, 
                     model.mrr, model.outputs1], feed_dict=feed_dict)     # 训练
             train_cost = outs[2]
             train_mrr = outs[5]   
+
+
+            # train_shadow_mrr值是一个滑动平均更新的方式
             if train_shadow_mrr is None:
                 train_shadow_mrr = train_mrr#
             else:

From f6664dc131657e6f09f761442b72ba0d61a58c6d Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Mon, 28 Feb 2022 17:45:10 +0800
Subject: [PATCH 05/28] =?UTF-8?q?=E6=9B=B4=E6=96=B0predication.py=E7=9A=84?=
 =?UTF-8?q?=E4=B8=80=E4=BA=9B=E6=B3=A8=E9=87=8A?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/prediction.py | 33 ++++++++++++++++++++++++++++-----
 1 file changed, 28 insertions(+), 5 deletions(-)

diff --git a/graphsage/prediction.py b/graphsage/prediction.py
index 70f68170..8223bc4e 100644
--- a/graphsage/prediction.py
+++ b/graphsage/prediction.py
@@ -20,6 +20,9 @@ def __init__(self, input_dim1, input_dim2, placeholders, dropout=False, act=tf.n
             bilinear_weights: use a bilinear weight for affinity calculation: u^T A v. If set to
                 false, it is assumed that input dimensions are the same and the affinity will be 
                 based on dot product.
+
+        一个基础类，使用了"skip-gram" 类型的损失函数(节点和目标的点乘以及节点和负样本的点乘)
+
         """
 
         super(BipartiteEdgePredLayer, self).__init__(**kwargs)
@@ -70,6 +73,12 @@ def affinity(self, inputs1, inputs2):
         """ Affinity score between batch of inputs1 and inputs2.
         Args:
             inputs1: tensor of shape [batch_size x feature_size].
+
+        计算正样本对之间的"亲和度"：
+        ①特征矩阵点乘(没有bilinear_weights的情况下)
+        ②求均值
+
+        返回的是样本和其对应的正样本之间的亲和度，尺寸：[batch_size，1]
         """
         # shape: [batch_size, input_dim1]
         if self.bilinear_weights:
@@ -86,6 +95,11 @@ def neg_cost(self, inputs1, neg_samples, hard_neg_samples=None):
         Returns:
             Tensor of shape [batch_size x num_neg_samples]. For each node, a list of affinities to
                 negative samples is computed.
+        计算输入样本和每一个负样本之间的"亲和度"：
+        ①inputs_features × neg_features.T
+        
+        返回的是样本和每一个负样本之间的"亲和度"，尺寸是[batch_size, num_neg_samples]
+
         """
         if self.bilinear_weights:
             inputs1 = tf.matmul(inputs1, self.vars['weights'])
@@ -104,7 +118,18 @@ def loss(self, inputs1, inputs2, neg_samples):
 
     def _xent_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
         """
-                对应论文的公式(1)
+        计算正样本的交叉熵损失，正样本label赋值全1, 负样本label赋值全0
+        公式 : y * -log(sigmoid(x)) + (1 - y) * -log(1 - sigmoid(x))
+        正样本y=1，负样本y=0，分别可以省略一项
+
+        ①计算正样本对的亲和度
+        ②计算样本和负样本的亲和度
+        ③将label全部设为1，计算正样本对产生的loss
+        ④将label全部设为0，计算所有负样本产生的loss
+        ⑤将两个loss平均一下
+
+        对应论文的公式(1)
+                
         """
         # 计算正样本对的亲和度
         aff = self.affinity(inputs1, inputs2)
@@ -114,14 +139,12 @@ def _xent_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
 
 
         """
-        计算正样本的交叉熵损失，正样本label赋值全1
-        公式 = y * -log(sigmoid(aff)) + (1 - y) * -log(1 - sigmoid(aff))
-        正样本y=1，负样本y=0，分别可以省略一项
+
         """
         true_xent = tf.nn.sigmoid_cross_entropy_with_logits(
             labels=tf.ones_like(aff), logits=aff)
 
-        # 计算负样本的交叉熵损失，负样本label赋值全0
+        # 计算负样本的交叉熵损失
         negative_xent = tf.nn.sigmoid_cross_entropy_with_logits(
             labels=tf.zeros_like(neg_aff), logits=neg_aff)
 

From 3104d644f258307dbf79737b88b0e23f584c5791 Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Mon, 28 Feb 2022 18:06:28 +0800
Subject: [PATCH 06/28] =?UTF-8?q?=E6=9B=B4=E6=96=B0=E8=8A=82=E7=82=B9?=
 =?UTF-8?q?=E7=9A=84=E6=B5=81=E7=A8=8B=E5=9B=BE=E7=89=87?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 .gitignore                                     |   1 -
 ...64\346\226\260\346\265\201\347\250\213.png" | Bin 0 -> 140657 bytes
 2 files changed, 1 deletion(-)
 create mode 100644 "graphsage/doc/\350\212\202\347\202\271\346\233\264\346\226\260\346\265\201\347\250\213.png"

diff --git a/.gitignore b/.gitignore
index 98f12e98..d3ecac84 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,6 +1,5 @@
 # Custom
 *.idea
-*.png
 *.pdf
 tmp/
 *.txt
diff --git "a/graphsage/doc/\350\212\202\347\202\271\346\233\264\346\226\260\346\265\201\347\250\213.png" "b/graphsage/doc/\350\212\202\347\202\271\346\233\264\346\226\260\346\265\201\347\250\213.png"
new file mode 100644
index 0000000000000000000000000000000000000000..807cc1f40c86d943156cf8baf9c1ca742f4b14b5
GIT binary patch
literal 140657
zcmYg&2RPOJ`~UGYgrY)5R*{`i$&po5ib4n>ij2%-Z^yTc$|i(^=Tztr60%43$S8X|
zW>)q-j{p7fe4p>{e_hXWbvd8)9`}6RulxO4T~+A>H7hj=g*tKf&P@#z>c|`lMSbMh
zU+@c=0_z3;P&jHRDWI}i*=FF6!{#^c-$0>q!e~fOkHFtlm^+UgQ7D7k$o~`!Zx!L<
ze#+gOH?-Ug=l5)^pUpkxb1^*ps<Tod{^*tXJaNv!B}zr=7k`V3{!A?0wmrjPP;+#n
z)$5oCuk_0jU*i$<nf6w6Qi)J6Q%=uwix=}D<|qNq37h&Dowx|uGNJ5F?zUc?7#oQU
z-CE<EH?`@O;tY-F-i)d9@V^mF%^39-Nlt6-y5%A{@<t|YYnSNzL?<!bBgv6gZFPIq
z-F>nR#*AvXN=1vRWRymud#~kv{ra`ER7PHY#Ld#)UR>sDaWSjw?{$wE2~?#7_jd|Z
zA!|ZH0u2pKw6f{gmGrLJU%!5}wyJtCp+-HzM`6}Uwgn~S<>ZORC=J5f^KpvA<VlAj
z4T2V3OKa;@))0>}j(wIS>NmB<{k!VD?oXdSb#%O3CV>*`=0T%SvG{=n5fKsER59Z2
z%*;$*UmqV|w{5}Px<`rzi&wHk(cp~J@h^Y*F7v`jm@^Ht3%-8+wdh1&PSDP1F}s{{
z$y`EOT3S|iD9k}Hx@l*}W9JU)$^0D*3U%AU#wOE1q&_@FOHhI}grU*^2XFTY)y(Mn
z_U+p*^Ft_;n=cNb#7@KW`Fa<R3PnXlrLkUgK1$5pky8Ee6OBgSnM9%LUL2-GjgDnz
zXV=SI@XGA66fkAUaVdG6sZHarur;WhVq7MOV&*%|hVrHPR$A)S%Q&^ioHAsa<~8{>
zPcIWat(v1d*xP%@^e>cyDtr&~)T3-TicihhOEXj3<fFrO|AI@3#_L+$N009MA409%
zL+%mZ*xVeik*qbg@5~*>)%^Hz@3?){W;ZLPZ)p^=<i55A`tY=uBRRTCRJqJBtYge?
zQS%0cuUF*3WwTtC@r35&t*Kgt@5Zl8(2S^wcw`!-XPB6ByLZjn=jW~Y4WAgBpc+IG
zNhtaZ)AQn?aU79RV<SEkjCOH#)hwd$HHd)M;f6LgTyk`A@7{fa*$X$O7hZd~1*6uS
zq3|Uk5`cS<-6Is+OixeW(a|Bm+xW$$L{>_QMfq&`T}BjYMG=|N?dIlYdkc+jGf|%;
z?F>e9iQ)c!WoH!1K>@jXhLV!fo}?9%Rv{&*`U-YPLjvWCJ9zgk1FWT`rQj`Ysh_q5
zCQEtc<w=%*p(g%=OsCVu#l@I5vSqT`=(Bc)e`IX)q+xdK>><>%q=U=-3mY}WVpRc#
z#%Lc&oN8Ls|1LSflf1mV)cy(?UCQ`7!#`SE0|Xg@ZVh2jT%w3vrqRd#-SnIiS6=wF
zu<!r-(J{SePJpH{O*d$Jds|Bg#Wiqn_m_$ad3&^9I_$eJPx92%l)^3L6peIdRJaWy
z8JfSHuF*>w+$8DR%Z+DdW@<^IxT+5xY`3@q<FL%vQ@*ic?cs5)(-d^D>jbJn4B3~v
z&Lu;y6}KPzTdEsGq-zj9&QhRQ5KTZ6*4NixR~MOR9FBBcaFM3{*grC&CUF?`RT)`N
z4g1r$(rj+IPg;WhtZ~$osIO`VKdY#yz%F%JN=$NY>xjfEhIIGzXr7?({fvkWu9HL}
z)fC<{VFLkB|IpRdrKO1C%0Pz6$H?>^nfQJ~g=IdgWnO71&hi-QSv0cBAB~5_B_(55
z!bk~maa0oV$;to!IEF`-b(Lk{S^RBK*RGSO20>(Euh%@vUMmWam}%?{bhqLKss4T0
zo7%A&xApIFiHXspsOut#BC85!4R%^z)g(t?K~FmyWYDN^Z)8=Y-k5}%nVTaQ>}{t~
zQd0hJyz5|%UN7t2m)Q{11I7Bkg$W7@`risjN=PtT=6(AX^M7;E*4BpYNKH#Sc&Yaq
zvRD6Gg#2&cu8NC4AZTeIUv$nK8XEflE|!!?!$p1u(2O_NS3dVy2h6@zP62uR-`g3+
z#_o^jus!^&vc>zqEfx^Ic=3L4^YwHzRV$dU^s}h&rw1a_J3O59?j8S`Ab}%|nV>R4
zzd+@NP`<|z3xHECGJvZVS6C5Y)XO|$`4d(s6Xk1taJ99ns;Z{uX1?D2M|b3E*({x%
zCAz5pYxQTWZEfq8m9`&82xciTvazv!)T2Q49SBuE`mjL75JoUsnSY-|@=B*eg}Wb&
zWuUkB9gKy6I#XN7|NOSNwDehOBjGd7l@z{g2PWbc+pD+rwrKlP%eA$&TJ$I`Iz(*D
zb8>T|ke^99MFxM?JqWz0>m10IwtgD2Rl9djOECIv=rcopOpfkx(?3qv2(+l{st4CC
zuvqz{=V7}|dHT-qiDhYLTmy0Vb8s!IW9A-PXi9uSf{OTE?Uo-u{NA4ls$E!E(DFcW
z$s^yL{`%$1$*2fnwO8~#$<&C9|Ie1GWcOHIVFeLL(I9}bHt?v#x0vD8k6kd((^K|8
zidwmiD2m~bt)!}W)b~drU}kZj)8Njrzfe*<h&M!)l$4}s2n3+}W?pw#=JWFLDH1I6
z<NJarxO$L}Gns&Y3qXT60fG1D*Gc<nQ8O&rW$6r4iaw2M;6vo!zi(&=HkCe*{;VYU
zrW6Bz*uyuct|YYB79=!Lq9$Sx)jALAkIu1I&*(~$abL;L_FkYd+N$cQAg{m6G<X5-
zQr8z*ZebMH^ud=!MZ`HsLlwKOM((pMk|Xy{H8)&g4M9g{=n7|LWu^W?p|<%DH~JcM
zyI{me+Fw%5dm}gb{-fqwR#y}3Wk3I|C1_(~lj49vncRSN5IgPZ>ABxpvAXziVv6v<
zsrq@o-rjtwuCU0(i*Z{=QFUQ3$|pR--kU=V1IM?r$r=0UdYOiu$`5mMbHNw*E1av7
zTwPrS>pnA1;k$euSwFW#)#<601c7hvvxhjhDs{?NqThXE2&(<Z`!m$xbW4V^dgh{)
zFJv)rfa6X~OkC?jRiY0(TdC*fpOjv+>uwlt@4dz-KKrl4PrAY}&6BfuaI5G}I;%1H
z>T&x(vIT*k<Os2br>C^HAWE$KV0}J!%nTOVbJt!~^H-`Ge;y{ko7>XT;_vT|zd7Z1
z5ZOekN}@2Ch?t4=4wblST3;R=aviy^CU0eKZob_<HfB6;uPG9%spZJrT|*fg6H_IZ
zsA>*7TXq^%sCw`b0+{I^W|f)x1sfY1zc|2^AZuQD$F6kw_8Xg{UVj{uRn!ilaIcXa
zu9ERs<2Su}=+9tLRMJrtTIfJHiX5j^;Z_;1Eiu^>D2z6ugF(wn#Ok+w$KBSubUHdf
z#<>oJ5jLp6ZbUVw#aB5cas!1zQzCCT;<h>Ds%dt4G(>Xv4)QT%qee?%K(<50(Z&gB
zwu6O;(MBHNga=}YqoYse+E?)4)ara;L6%0Fnv@6A1{P9dZ>W~;Z43x}W5FyAm#tmt
z=^~STN;7NqM<9s#^>|Ueblv${Vqus*CrT8I$P+&Z9Tdku8fL5EG%N=Em;&n5Jh&Ua
zU^0jQzm=!Uio?w5KYB#yvYZKZF>Wz)baw8_r1Leoiva-{7#ILE4PHt@nKoKE#i7W6
zpj9>g9by)ekkb~g+$FHvE?^jEjs#YXLtFs?+)U;<&p$&YvBn7$n~NiAYK}J6Ha7nH
zMup8OyFGY}zUOkL=pQksxr{DLTW{2HuvwUht5>fc)fWXBCFMd`Ij{hMr)9pvoarY<
z1`evs4{PG#EJOv_ev0YtOnq49=X6>B0jK+2Upq%vW;N3ef|9ak{1Pv`?<Q=J$ZZRg
zH6<1Q?&h_%H5jU8EbgHN$#3)c2|U>-Euke<UO7tOzcNs|`=?~)dEMi}^>rnckU!l_
z9AFOab5d10U-(WjTB=ZTK2OB>GIyRN$RvS0?{%7ZqI%#Awf~j?DAUwyE*gci(*FvL
z+2@7%d9WL9n$!j5dYM;e^+&+?fB$}C#|@Q*?Ctc|!b0YRGx!kkH}hs-O5&4}LOGfG
zRbb%=Co#7W$%9p+<CMFnb~zQ4=)%a(j$&R_X4~;TR1cUT-^;TuB_X%DuL>&6ozD~3
zurjFulUX57uXi6W5KVnJc8*vbD3HtY_o3EaNGN`XF()P^mH*H;!oI;RsTIiiTHSeZ
zsPfVLpFi4kvGIauf{=HKiFCjkZT(jF)yp)y$*42dON$1Lmu*4!_w~J|awy#Y(bAH#
z`||=s4)*E~x9VOA`p$uAt|Zvl+QulSwEMlo;Y#x=<T<;nubR99l`JSI&?Js(AWI8_
zS*km0_c4t{>Fn9F%SrUMzAQ!SaeeHgddS9KOH}2{V|$^Ta_JX=Q}sLpm)TvsY%au=
zZ;o+)D=2WC>&|2tkZ~TbJHd{r>jgb<vGvZ>ZqB3|Wd$3Aq0Y$2$T<PQyQa!mx__z`
zTjxpesHrci_}K?2riaTsVP{q4i<08v?iM6vOm$OYPJmRS3E-f5oh!66Qm)=KtGVkA
z>MCW=oZtSg!hRg=LK1bD?l}tsm>tIy&!U=a)E{&l5uQbC5jvt>hCj9$PY{!pO^l~y
zq1Sm%wGao6g8BqEqLg5J-KG;=e{}@MLP&x(6p&`83oKlxFyI8aGz44PWtrc82i~9~
zM^Urt_u{^)YJkGkgcdE^LW_Er$|C2v?%`p>{+HqqVT|8NArJ^o9~wDe=jI@EuWa@|
zqhP@S2Kq;x<b#Z^H{(bWq4|@jP_|rm7R<uZpOVGYG!{|tc7tl*4TpA64S#{$iJnSZ
zI^OcK{oUOb|KkaXi6>G{fW~PY8GV7c{l4Rcj?<yEXcGYuIYwJ=okGl3V-XCQ5*hNx
zp1Il?yGv;Q>D3?L-v8>c%-5nU<btufZ!JCbh!=OjU`~4GL(rZotuh6V1<N8I{F;}?
z5Pukav&5;DzlUt=Q)Bc0gYG7i$=#ZK$)xpv#1u2#nz(jl90{`r6`Lj<iZdr}>~|Lw
z7ssblRJwtDEE%|indwY+E$W#|X~JCew{ioY`1C2>p5L%d$zSEdpD%fNkcRlon>xa~
zVDTZ!hCOfi^T)PS%R$`9DO{hk(hcIUC4=Scc6>-YD`nAXCKT<76)l#msJ*>C)illQ
zo@xoqXN<d_wynYUY25`EPiN=#(NV5)Vh%fI*lj#yOW1@Lw}&|~vD2MKdA?E?k-SoH
zh1D_EWHL2-t-EYct#QSv)?S)v4|n$>z05ilN$L~*OLq!6SVQVoMW%Ww(jX<04Kz+%
z8ZD^UThK4C1;^=&toZN_<UJz1VscHHbWeg2;WV_M@-w@xx=+m>O*j4ITj?L?_PO<#
z*jSUrmt0pWs#El{d)m5>(PgK1SuXET`TUvo5kc$^7+kr8<<mMRv$Zp3EE~?ro}{RQ
zoWR$_Zv$g}Xd>^BYNy2&1x5vFnXWhDT{BKQ&sDZk4{WW|DV&;H@{;X75xd2i+B-!!
zD{0G|aw(>G$QDPzm&T5H-+KbW+EhIXF%j^pY)DLX1*M?ED{;(0=xQCPsb(#+C}=kr
z4cr)nEg=?73u<0t^a4fhv*tci<{DJYqH}X|%X-Swa>)>qLx{(S7+J<cPlOSbs;jC}
z*jkY989(;Ew^TCz2?-Jsj8Apr3BV~JpEmbmklUM4OPiQ#rz2I;d5xbP(KR$wz4bKM
zuY*17Q%=!(=I75J-%yC1K}N?NdiVJiR)R#hzP-fx4^(wO!2HVO`YYhIGah%oBQR~P
zrOEI84D(rw)-l}Ps+T9Xk^AInINAN`Gy7yz>rB=x7_ir?dLS!A&<5Y9-~fp_`{K+B
zjB85t^NWjVvk{Kq<Z`e#3U5wNP5p{1H-G?Zxfr%XBr_)`KVL$%pj<t+@{Ja1959W_
zFO!p#OP*K@p0w2Kt@rQW_n&3HFUA4dH#T@XMPtTtI^O1LLJrTQL*Gr6@pGXZV6auE
z(bL8{P5JrxTC%7W$QCQV_-yWW(l!+7Cbkc&c5xSI>*!PyjswVYAsFXDP!B-=hCt93
zET2ZCKKN=}%c+(??`<<PGX>`=NrDGzA{m*A{i~*jA!<b!`}@m!nK9`~a-d7Ka-wnZ
zrr$tAlVddq&xVpq&nDc(ye~|3E2K!Br3lA@4G8cz+IgA5{@&p~3ZBARd{=Tc*BkJl
zJH5rNQ|E7p*H9Ui4A}~byLZg34_7STzxNF9+2j8%At5N%H2vH`m^;b>VP_6Y6f4Dc
zPx8;pUE!rOE+zJEDn$mZW^={zxb&YEx`~2mfV5Pzu!TJj{s1|a)mzJB*U6v}Zgs7#
zt?wjEUZ{Vlj2D(-^nBoNb*VX#Tfs_-c12HD_s%4*&X|H0*qX$jIZNaaJ7a|&G2fhX
z;J#S^EgSC%etmxq)K%;5iV0X6T!{ice6Q~=YshT_<I;|qkH4Zgt*oqwgRnU%9a3k;
zz}wzOnIPu&vhn5AOCl<-mF)eRySug1d8AHx`W<n9#PJE#`B#8hF0D~f?$@uUTN5GT
zdIS;IC<}OU_@DrI2n~2CA#&m-uA1LS$AsU8e-$Fu^g~jo?7aJMg1bKkcTjb~ur21F
zv$CeLT1Xl>EV)HRae{X_tgWs6a<%yxR`P}Qu-zPrIx818zQ=)F@Hq(A$e|&~`S0j*
ziws24&Q$7((W!FOp1SKkPk*X}m^ZMXMZfYym@>Q=j4=xs<2FcNX)BfkM|L%7o(R`$
zn{IAyYF=n8?PU@+2z+FSopsmC)GnRK!#O?+?s<Pnh`zEH;GU)6r4pa55cxcLO&YHK
zF$2LH>F45l=GMn53qX2Wjfb@|y41-YWl8BbrxyPF3DCT%N^}OFB%v&Em}?(Y@i<Mz
z8m0UEF6J{@^lp&>$nENwTa3YGz;D-T@w2LiP9;N|c3$-KGYnHBgM$X2R8uqn9Qfza
zpj<bIqY#n5{ozzP0z$1flw!K81e$l#uP;wqxVuYrz4;N8TQ^}fHP7@8JX#l(Waa<V
z1#nKEMY@W;G?(kGwN&;JCM)m3jRiOjU#{75e8%-Kw5A=ryGUXcE~ub<>3%xLfs{V!
zMGR3H)>^x(i}AeCRY}R}ZPu4Fcfe$^-7)3pD}9%oOgAmGm{_rsnwqNM3;>13I;ue)
zmbiK>UU*#mgSo}6dRlfc83d1K`o~Y!e_&*vHZgqs_`Wmu0HnUv+wnNuy=l?59g6o7
zTc|onJA9v(8D;}AWVBcR<jM`);d+B4sneQan6Y01KXTE*DSg(d#Su1pr@TzW7**M7
zbRg6jLNU{TR)&gS{Xhlih~lj?XU=HUHhhDi22)=5VPB`+sI>^P1=dI#`<ddnxHu<j
z$L8M;d~+aW9%VMU9M46*<!xmp^f1^uXzrC7Nd0OocA0vxukWq|&%hX?u1IXhb~E1x
zrT;pR;ChP}MMXvG{u%^^^eDmG-@x4ZX=)R+V2BlHers?CUE$UcZXTN!vM3mU?F|3<
z0W}WbSLL$8A>KR+2ib7JP;OW}wTI_dAz)ec={oV63Evp($WzB@f|;X{zfVv`1=GMf
z)(`acQQ7!{hIRl{As1AMieaPT6Ty9jlw56$8T1Z+JC(aJA|xaP)3o9<Cv+w#CWn$)
z9@ZmC$fde*COig0CUrNge`{>--cuKstBJM$U}3*b2^$$7i$q>~;ReK02dhzJEQ#aF
zGuIgB;sfwXF=vnQnyYbGw770vxv0o2o6if46OAuP8`pyi$~6cR-IQW6AlIdLaIri_
zLl_VXl29IAsEA(8AZxmf<FqFu1b14B4(HQkPmZgpr*+)zLOm%359b!~^5x4y{)eH(
zxBR*t_%THW-wr?g6a1apLUE^K3^s}$aExWk9}nnmA|rzwb@T#^to>c%t&(e33Lrw@
z|5Bka{lGwx&-KHGAy=p+;^G1(l=i!0gQ9Dz)$Rx55OeYl<S7=)1@<Cg0RJz?+yiyA
zd8gj0<>7F_@cH<aDEs*{PkF~>Ac?RyPIV~4E$bWcCqabd3<>P$BrPqi@$@U<HtJLH
zBKEYe+Rd)M;!3N!d))ksVRl+nJXZWKT>xezcytg%{IMw@q=`T0spKWVcR@gY4QvMH
zJh;YXyVA>-FB68sSpXnm0tqm?M@(q{t?bWh2a@{LJtXl_sqhKw#%uue+>k7A=A+q<
zwY5ALhXi7@KZ0{qkUrPY$`$K6;=NhTNnR2Pl-q7p3kc{2^jlCw3NTMLkuL!BsA9;q
zt*0QifW>H^owY>2_E#XrVz*<lq<Sog54*xwv5}vhLhsczCMJudKY(yo3Va6n?N$i)
zmsi4WHye>hDdm~R6MknF3@abGkZi0e)%<c@Rmf$e7f^8to@5YM?3xnx%8-Sj`-MUG
z^0gGv6h|-TG3K)CH;-ZNBLPfep_~fj14HG%`efyiP-UvqcXx9!?M0+wPIAx0#ItMl
zQcOPA$C@m_4Z=`62J9*lZN=5R$LN#qq-_iOsR+@w(**xKR451Xh3kiz%s;~9XJ`&p
zVCC~ulGaDRztmD+Nd+6GVkL8iAxy;zku=#AfkBO5EvA)5<Yuf?;*<)+=3}HL;#^(b
zZNUqu9$yfLT$sh1`>)!E%OmSH1_uW@_kMDEPjGYIZ6Q!d+77y{{eB?G@7^J25NavI
zdpWd>2O>?)p$7csu&3yMU%r=tU01gChEy>XN1TP6@NUZCgK`t*_U`Vk{pl73yn3d#
zXy#?rt-qJ-?J-Su<S9FjfkJA!fT{a}{6$~CeLK@?P0&)FmM*CBXVveI@7tD;lyqI4
zYH`wMS^7@rv($s#>k-}VhVk#vj{r;t`IItMWj-Pe^9uCmH{KmnFmdkeI2+g&9~bvJ
zbL@$aPkGOrEZ{W~=}C<JkD($)_<KUB@YC4#iG?nwvDN3G`JD=p?~EV-Krk7}x8{3O
zc4V{MzF;QHLhi1)b*FA6?z-+hfF*JPn?RNevBs;^3%FhuK1gWUe(H@Axpz*yYd2&y
z7@Y2XB>19z=Tf~n0)t^<l-Q4j_OGoueX2#AI%Cn-f`ZN2jO<kX#@wBb6gz9{HR)PF
zoH0YGj#&IrJ8vSl)68Fmt+2ZW!2tL<6tN|+;2~{iXZjWRzR#NW>Jtd&axS};ueM4P
zoYy}~yZ%8E*~Es|s%iAu@10IfsAG>sJJ&eQLZCmxyYg9b#Ow0HvR0wNZb3O%Jetz=
z6E0Fvc;UTx^g776|IE4vgd%&qN+(%~c%ps4wp_JJ6%8LBU&mSY@~$B1&eHX6eL~3k
z=N==hx1F8n!c7i<pKCHKZtfL{+_0z%TnP;DPCs4b;Iy#xJHN+JdczDdQUKWkY`sCd
z%?sB7tvfNaQ}ytp=s8pbEFI06_z_6Q@iUSAoPi1hym^WW%{<3K2I}xGB|{*Xo5b&1
z^PQcYPZasRI9>Gh^`#s05)lx_^119B%<4IvgE11G?crRiXIv|PK0hCpBq_LR!nxZU
zB)tZEl-5Dznp>@}<FKM{)VuJbKn|5f1uo;mX;u!f0Qehc`-y}H4#va_0L&04)8fB4
z+Sx6_&{Y;P%>h3|rsx;Mcg3PQT5MLzSm|6|onoV6v&_Q)lG|`UF)f}&)}Rat1eKZZ
zJyYd3*dTTn<m5cfGDd|W2_Fj-Dz7EBuHd&~r+b@D>|kR<+W^S5wX;i$O&o?QQ=+u1
zM#kKuy=|pNf6RBkX6}ufuNv=&K;5yhvy2v9v|5^9Z8uoBytO=@*CDWO=M4p*<=Z+0
zE!st|neHCztD@jw4lMDLL1AHG8M_mslyn>)6B|kmvzZ2p@%V&a5i+lSKC_5Pvka#f
z)s-d%8Ig?psZd#OVfDq=UR0#WgF-CaoU*aDzQ^WLT@+sHEph`So*g5+jT7a2B+Z7(
z!oj`EYcK|r>D*&<+Jez}^l>`bJ$$s&x0=Yh=@55M%TkS`{feUJGeYhUhx>EZfB*c@
zN$c57;Jq+u$qr(YqeZkjuKX2H`+AlVtNeH=i`B_ItH48*4<Tt6YfUN;TUr<>kWN(X
ze{BsO{k69~q(@CEU-|0pO**(E;Cq4$dbOX3!fb;|1Q)puyYuBYCOJ>HvR}+AeMLWZ
zxX-}A%gl`TZKkfUo(jK_$g>4NBd>s)uN%>~D~r}MYxCG!?+J-)2#Su5e*OBj0v@V&
z8koov_NhrvyuD?kqh*+D;R9kC@JyCJNV#l&TaxqUpP@X;xJ{@?y^DTk?!8>-;`Rj!
zXv%jvbfwvRWsr@&So$P?6oIy3<pI%u_wJsWTH49UHxO5&H}5}DP4S{~ec5B^d2x><
z-wa=P76(f-XYA;Rh^nsTCs~!;i%O87jn~voI&7!MFxy>CfC2>mc#1}&wQ`XOOvC|+
zjU=EjvQ&zvr{j>hBOT!m3SmF<r?IioEH}<Ly7`i{^nlS>D_2*^=+`*6;m)aUT1u{6
z#D0iID^I$2#LE4%*9(h_o6~<&P3Lu4rg)jY1vIe8Ae6J;;tFK%f+wVwYejv;_bLwt
zI)UV~R@3e(#{wyZqLuq!|Cn<r*?Vpe;YGe4`_SL8`CZBDsak?@1?aHP?6lX15d=7t
zL(c1+o?Y9onC3)4v|LKqbEA!ObThg{uGT{d0{oCVe@)o2*m?x!v|i!Hp+OG9U6EBM
zw0Zw=B%Ij%JWyb#ugC8S8Z^7^ar_a$eW%IWrTB+LLh@c;@?K|}+<H*OR?rNdyUNNc
zUl033tKIr)i$|0uO!|ooc-X@7B+0ZH(6D=NA`l#LuhITJuo^PDAc><kOOk<nOMXpJ
z{%z};P&O%Te+C<$goZL@wRDYh>dx%?4kT|o$5-lLw}MZA$>`}%uq%&abv%%vi}#{5
z!H+@d;ERf?5#Y%lb^DF-+kEymgmVX2P}e&`X|4<UpQOuqb7w9)jdv(lp;~m|h8!rv
z=giD?^TyYP-s@dXFOqi%@ZZ|t)8s^e?*XEQn&Ys)4D$E^8sN;t>Y${*{<F+;^Ox#`
zQNh4afB#O{{`fvRaVTxrXLn(#u)&Ok$f7vVJ$yI$uAK-e?kx3;iMv$y<EOxbVfX38
zH8ztPWF`opTq=5HHmN|b<s~H@yf69i^tCU@Nbf?y=x&OJ)VD4sq^PlQv-K{fE-Am7
zf>{SPX3MedtUFO{6SBz5!?4kn%m9#8UPd;W)vK-8eRT@bvHsVCk)y$qu$f|YAa-K&
zj>)?QmK(Bj8w&$d*IT5iTA69mE+uxF`$yS=NY_EIfeKwNO&x_@`mBhvI*qe8MSe%B
z|Dl^x33jQxqq!tmd8o^o%q)-BBU?KU0ybJ?h}~Q7Jn7Ya)~+;SgUgNwJzfvfGoXi&
zx@DD>m4Pq(I5nC(>@k;_M_+XQJIiHo>fjaXe-(MfAalBm+!@g>SL(g%VtGnVrCB}<
zUV^NVfe>=(H55@0$jybA%OT?y?sA?FE>E~eJp_w*kO%n2W@ri4tLB~hXR)9xT35HX
zyV)V&rH!tgVx@c{j12T9hpdOwys@O2N(&QJs~fV=JzySDs<ZAPbqSJot$>mz0-L9m
z`QTC%a*2pktU_wdhswS6JH(lI(C+52C|MU_>u{||6c``eRt(@P@oXR8_eKd=ZA>Ea
zt<I!^zd{AHa@Af$8No`8#v?mK@Q<08%dSZ34{h@p50U((`YiG=Uj?#(5|A=Ls5dZB
z0@W(Z8*+0<Fe@D)dm4xP3c~~q8V!bY)^4Ocu|J~iH1)HX!+eN}qY&ywtL@XLM5wC}
zB(^{Q<8e+?sr~^OM4d;5sRDB3VlzJa^MlfJzDtTPqC{apzH-Q=o)1c4p$pu9N5tJY
zjbqD!FC7O93Kg$m8Fop#VH{M2r6VYS{(vk9Qs}9yJt&6lt~`RE?5e(!%N@+h7&2D#
z-;a=K_QT`BxtWR(M?mxf4l*QdYI?ew`^Ml?q^{??>b|>BpeeNa$o0>l)btxEBAfqZ
z#$leMLO!*mkB)s7HA>yx-D7?CM4wb5upe4TNC-P_M+$M$PTGk}-i}+keEJ_cm5h;$
zHg{PLGD*a{2Pc4OYrhAKBKoukz{QxeGBHu4*}x5Zt+sS_dv&wuB@qpPmlL?EAsGG3
z{aPH<W@enKqQi%O7x#m*=;|W4b0OPK1m|{ylX6nA>!P8@AN72_$9j55z`R(6Z!XP2
zYKVym=7`i;0sTmf7obxnCML!4%lSB3#-Q5v7t~ebH!OPV<)iDd0WmgimJSY=7QFdO
zKMvVKvIMr~W5LuFC>$PW%-BB~<KltGOy~d}_~=gB1i?+P`cY7tXi^hKbzFWNV9#^#
zyo}FK)vM`R!>%gMZW^qZ4ryn9YgIm>wH{CBGkR!$HkoYcCTJbyU;lhjk?1BOEDQkD
z$Aa=0u-8=~ru;|F&V(SYbV=qko9`e^$ouab4=t(#u>n^fT8PR@<p4I&zQ$_l<Zt$c
zZTdKf*&>v&`V-qgUkiSz*7uE!a8lb=u~1DTV8*J`U%*>H(DdV}Qp9AF<eLBk0bRs6
z=iRJ~xVXe}{fHoW*!4|CuH?4<%Yl*I<$V(nCb=cM{qX{<@e(1iapVs}MZXA%{3Kpe
zFd8tJEd9;AgP0me``vU%7jjG}tTMK>ay1uIT!+NAAqPTL^4|gNm=g;bKs)PnxRwE=
ztl``d?731K3&y~2_B(e}Q=uMR?k$}!m+U-WT7<t*0+qTj_mg_9L}3WwoU~hc^72gH
zcF;*w$rsewzsEsPM(ei!K2e19Q-SCz$yXu6+;ZHeMHIVXwcm`b*xB8mYaf3(P$58W
z^;+Ic<t3xbZO{S`U(RsX_d|lV{@kt{vkb9?QESMB!FEI$MUla#08)AT!%sUJ|J0uZ
zvY5|ld@UpE{xY{hiBMj6wmzqyBS4;o?yF{&zH)QlUs66#t6qBwRfmzUC+bPK?=26A
z{l|3f$k$Wtv!w|*fI93vzLoATPm)yf7{p<BRhNZ2{N>+po(i2{R;&G5RwnCYa<`|w
z)_pi+@<=sZCJY0Dc`kDZA1sd4Zm3xpKg{T&X|p}*KQ;KIP%K6keVdOh$HaW?7+6CF
zm)_pq-)VHj1dsuzK_xB&KBDCeU8bIa!4FTKg4|q9&wff9%dxNuO;VHkibKwT1M$Jk
zb1ZJjQQYj%@oZ9^F<3)40V>+-NdT?a^IBO12J&|MyLP)JA^bqPP~eFvS5tvm{@r_$
zPi9NHVpV!MJRo&`TZw>EFErbc<${R9r@kIS-Wsc~SC;FW{go7-H^2Ip*oZAx-x?aL
z+0~qQ+dP>QcmCR-?${TCn+eNvm&>29Cff&&`sK1P61H-|2yCtG?CLA@0<X3aQooh)
z=Y%jp6wds>*#QaA7Jix|S@U+2PWkCj%WLJa762c9!$BCuVe$$hP&a{A#)|UG+XISH
z&7TSu8Uru;+xe-{)dMv`Hd>$nG%0JNZpE`=(-UYGl)uz!QOIXc!e`|meyHVjEix$2
zs|e?OnB6ljhP(aYsmFL!33lJdGFom9DPuMm7DIGqS1@f6mkyFg&H091n8C-DU%x2N
z#(bq9!G9#I76E7coAICux<NvvIR}O9>8HNnAxHV)L$qr;0#I)*uiD~m^4tu`aDZxM
z=T-J4f||D=dw0gktQ7LAbeM`(`*FbXJ^@QCul99);+RgLMs)K?Owj&$jxr37V~@~9
z2DhhA&&BL3Uwg6cJ)>fzEKjF~<QAb&FmEUZUG8mStt}Qa?^rAcTZ&-KH{9d)$11i&
zHE7leMtfaG_v_bY*C^{vy#hYKWX2PI)o@n0r7SfMmV=x;BSK6H*qWe)YK=jmI{!0b
z-O6h<uk(9%*mgL{f&1S>jvUzRXZ<@nJG(Y>uV1Sq%xw56c;<T*zS6KD$I(yofb{0*
zl70f%(Rs&vGCG;mn0!xZHNEy3Ps2(cHy(%g)1=3iN^E2JZx{93IIc8{*EIgbq@&%^
z;d>Kxr)_zO^+fvQt(vgC-mty?>^(jHgtIg#He`lbfU}|vpF?sY;q(<ObN$$4*5>zU
zi>nDlU8v@gbFmw3<xBpeUJF)6D)r)Lje&eY$SWyHP{v)bpB)>k8DBh;`)k(>v#Tn8
zQ79(pbqI*Zn3}Ci@%jk|6AIsH@RQFv!F}OA@Oa(7m@xkw=s?Ef20z|A_>aAcZWk%b
zJ2Kk#H@3{wCUh2XQ1vfd2v^iD$Z^b1v$U)%u_^H`g3BD6=xX&|HMSK-VT57(&+Ki#
z<xJ7oO-~c?$pPh3Z{<>&V|o0Q>>wX<o0~oV9<@$CAXUQVE=Rx2&(GIzu<Uy-8`1jn
zJ{lOC)q0e~mOijh*Q$*VpZRPa{qI`5H*o6v9b{9b-VXq$7Ddu5<;gy_oJ%|gb~$-0
zalcXkuu8fSQpI_^_5#bG3oq6)bty!>A8H1I;o?Mwdd{xk>8c2L84?>jQ>|!`2OdoP
zr8tBPNL~8nOZaT1CZ50+dM_rh>I6WZeb2ADUQ1Me2lBFbkGzU6<VhIVn5b6oSD!k*
z!A2z+qyFV@@G((|?l5aHpgM$06<MuP+Q$!b<(<})aO*b{VW2Mht<uF*QYI`Sl4@yi
z0L%_QLddSgP$&TIXZmfknzJg%4^1FwM<E#dY6@Wa6&;73DLMphZ&VZ#@A3n>QhP96
zJ&%#i6c=~P6guXM-c~jK70>uU>x3Q9?10zM#dB!Px*bLmZUv6hy775}{IQGGg<e09
zQqqt0GXp<rarexw@Ia<sWwr`T9yXPC2z|EnA``~}e|G-fv4y(7P$mHAgr7qo4*Y4>
zEc_%cUbN`6IY-OrRzx#X{GXf_w(`^uQ9^rWs}D0n*aJyF7%Su$LMKc!&XC34%5w)j
zEy+Pux&b=-?9TjFc*j-j$@smT5x*n8J>A_7LuRzbxkW!MJ_<g5kH)Iok1cmkRO6rV
zhz_{EF?!8({Y){j@ZZO40e!VzNzqo3t2Nt4IEh`(jA@aLb_r6)TBg+fM^~6q^b8GY
zlLx>#s|!bF+Y~P;;m;uGthP=KEM{E*core!3<)C@b#}(6Ai$^-S+0v1ee#^Y)ZRoq
zj($0&p^j9|jAJic?ODw~-1XB2zf^)hqr-6x83a<5dUne?zk6=I^I2miv!TKCbEWY-
zZpf}@UQK8(U<8xf-BcM%7(o1qib3>Yt{`#z^o>2=q_!v3BZTPK2X#?TtRWh{bncrT
zP{w`I?#~2e>_f!!vG}n(0vjFki*dYmzKD@h!#`%#)a)b#kVM6^{_@P!H&+Vpvx^A%
z_~T!Kzy{NUO7Yfc&vkZcS$>{)^x@|*XM&a>!M5if@UNqliN_Gm40ohClKG3oO2w&^
z`e-}I`3c%Wcw_jfnK6JC5mAJt7I*S2pNZJ7b5J_5FRgxf@uoL$TfZp`RiC;O$UQCM
zezXsCYsvn*i>(fxb&!02*;V0jmY5R;Oocs=_!|5+2R?5nZ+#QJ@<{QYl2ld{7-98I
zAn`c~J3v)fL@N#IgtH;KI0Xw(Ks-NB8X~1?XE*{`p>mhXYWN<^u-rA|mCF!=6ng9x
z32I!vuy0%+f^ZtE=Xg|K0O%JLdaE^mU_m{R>!Do@Fn6{-o3-|w;v9tV;<!X{Bmsjg
z?+;kstYMH8i<sb+!i2vKvjbWE?KDngN7=I$oMxo<SN{N|W3B23Z&Zxhn6P&Il#Pqs
z{G=P7eokgLaMVzEn|$Hb8*whrK_RhQp$BC9_(vKdu_i*M#}Q^YE8wrlh+OE8Ijhy5
z-xqOQiRgX#JZC8%&&E@sU$V4|zpZm=iI4WKBTNUW0vCFKOu;*VD-3fY{|+PnPV<>f
z^I3uVK;qmqpU)T{!lJOnU6UiJ?RV_2HLCfnaFVOB?<~>E@_QZf>{+<i`-?A;_YcVL
zcNtMBf$Q`n1jPR4B)!C*l-ohz>yFnyDj=$~i=OGGLiIplcXYNkf*T>uX*&FYD#J>k
zJc+22RRGL`4<n>0$&OQvxt#mUoN5n?Z*r0<Da$u1NSlfe?r;=fNgHaQj^LMPHD{cr
z3^KZ?3z{H4woOD~S`Jnh%9P0J4i<oi;?t;J_ShZ3?%JhsN9rf<&6{hQiVV?_D~(u{
z6&(t^ca6Nrb(G`<t>lc2^=zzcVkeHKplKBNEkm*tqu0Rj8kC_o>wABA?*bApmd~19
z5QZg<gF<x!EJM(W%lLxCvg*8%*$YTyn`{|LfqbRoZc7nbsAKX+#tGCTz$ywoOumuV
zfKRd9X<0h#gH%k|BL;VVn9D^8I?j-*0?BlW|GD$$+f;1!XUdKTupqI>D+({g%K@mR
z1{N|y(rk5<G@DJfw3IM3;UtYx*8kmNklll%6X{Mx<mCZ*afVkD_Pa9`jz@Bjv9FKI
zlgDR*rwmGlXbYOShOSYJ#ve=vBA|9&mbCm&<Qe(BG}p!9JuB=&cA3i_sjFf~M;<Ck
z2zbUGgv;oe+aHBJgrtpgJB<8cpT7(9mi%&mRQ{Mw>;B^Y`aatmg_V-M@%{S!y?USd
zdIaRrs#aN(B-f1`<Ah2N_F7^yDfbK`st178(OC#X{mhVBcgtH6LbH;f4%r7`{iv52
zP}~Lv!6X8U7=+tMg*e~?gA#)X6UT{3uI-`nhomy~4@mS?tpa=|4ml5?G$uBXN?Ik)
zppJo+(6K|Dkcxt|l!WGr=#?CW5D1adiO#=()H9^0d6_+k6m&Vvpmi_@vSwoL8R`n@
zw=XALtr*B$ozQ%9v1{pc?O7El!th<%&sqgz479MyE*hKhc42`tL4>$SYz}7=$Q8Wu
zv}Z)8Wj>gWKau7fGfvltx&8hloJ}`YN*>Z+nJx43^P?@@UjI+m4YPY_S|YjfVl-7Q
zOaop0Q|ro#!#}SHZYk}wpNfD&W4fd<0<Dj;f+<DbbpQWs&v}?30^#`Hdxg~pf*c_w
z(y`W9Tx(i+huxqEz1`3o{2J+I=z=(i;K3bJ9CA-)IF7UcW-M&LHOPpzOCTEBZbZA5
z9E!k6ppj5Q!9d+hCguas(o6ca=enr#Ll=H%XGDkINs4ZsRF7@O)m;dnMMB$O;WD-~
z{v#)x<Dfe2T;lXUqx_|;jGdJaKPi87oZXm)M6I8xBN8gZ`jUVFuEeK17z`$S_Ns@t
z6LS~=lS+eSnEeW*#aWk`9myD{+p9ldfnn1gosyP@YF4LVW2OedDL=DY2)J#mp_R<S
zbeyqds-9=*y(zEo)z8jt9wL0><qUYk5LBD7aCr=I`oKR4XN0;I-un_T4f}<*89Z(L
zw5<vT?lYG7K&7_o#<7**4}z*}C11Zr3c5Tp{=H=PrsDG3*ArIyBIF?TAe97#-|k}}
zDe=>U{EF=sJ93>Jscu?+f7{B(gaQTbOB*Uo0Rh^>_0q82kmP0?%$G&Zu{;|%U})tR
zlAPZf{}wYMmErI480nWSQR2@4E(5-tXHhan4ZG8UsHl<DTI~Lck<ZfT4d}2jfI__Q
zXQTrCt`gzhmQ1LBu8J<u%N_*kCYqLL@o_v-Q9L+9JA<Gc%kNK1A6CG}Nn3l~z3(dD
zh>^cw;ST}7(CE3d%{}EC`R|5~7Xux#WU;)YpaALI=t1HK#WA+5+QC+HHtD3d&89)M
zelDJD)NeavL=65zymoxD;>W{JM{7y^&^j^o#t2(M>U-xRIjp<#j?ZW6a`N4rmQw&s
zc>zNca0<-}6r-#MaZz?pPzzZ`#H7=1b<UR_IMkRcz@A)vR}UQxHFIJQr+xud%KtA0
zJ`TIdFK1g_gBD#YspM&OhAH@FM|2$|KQBTvfC9C3(m1f`p~gTXZ$(J|r>m~-bgX!j
zi6pM~`55&8=~2{(Ojeg=Eei(WUz%nqY@Ir0`cBBlo1e3htbLL6lz$s3DBe9<BO@Lr
z8H3#)+27RETlGRpwW}?XOu+Q8p?Go+qS8xzx6f_sN%L~skUu{%GHNr)l``<e_v*;8
zuR}-Rd{;iiJP?mIBXJy+sYq;doSp;$=sTCYo#I!y26f4ZEd;NzzrfH{`UC1?5VfRV
zMg_BjWK1CpqyTN;>1D=th5}J&$`1Gjv@(ndEf5CG+%YwTs?yUz_)dCt*CVLZ>=z4?
zzN;%o_94k=8eko7rCyGEjD=h<ayQ<uI~_U=csJ18Bi_d|A!VBp&&4t=8x%7)NQ>~B
zaV2Qa66x0i8XZ3G#+iQ7qmH(51)(eK*~e*m^bHF5y`ls$DF``15?5Q*6Yw?C`XGwp
zw~@+-a<7-MM@;3SwX}x`7^}t7)z+2-NlmR)h3vA8{^~%waT<z)8p2nkk3|)aXp`RM
zM+jB<tmpIR0?|q47lnoW#_2;fHl!aS-Bz7T65N_gN9}p^mcT>=j75I7X|BbdYGP4c
zgUDUbyAJ5nb$%kdFRvJ3O!@`~y@20u2ULET96-6P2ox_wanJ?Ak`Gqotk|QZg7<q)
zdxT)8z6PLN_E=o0deF30tYyDPz$oN)#|q{VM?`yT##_er&H1l1wr+qTYkFaq9(DH^
z55W5?-u4;B&hkjc02q#xRFH!9gBBY9p{iEmg_&AZIO09$_4M_FY%;BBE2%;<dLMiO
zF1O>4J1&5(`E15GZCy&0H@=YI_#iOo;#&lfapm3LR(Fc@x3O9wJ|o)mKTYjU_DCue
zd5BjB=AKHLyo#!)iTl#%cciQdi&+4e8g`#NHD%^un&s!&(2d9J09614vTC2^!IqrN
zX^t@IXO`49C=uYPb9Ic}%c{LcTM4W+T2qiLh{y?rGDb0!U)Ib;4WSFHw&fjJLwS9^
z#j^DBc3T=)&?cA34A(@17^RKGg}N1Ju+TB`mbfJ$>I6Iis1K$^l@U_IkOd_(y}EoY
z*&s!OD!adF)<O=|3VoX4&iF^nKHPLRne(LYrI3|tYx76nu4-JT{PIk&269LH8s#lH
z#w$@fQHotO*zm+8w0;y_hQEC9^c5v<&K>UEDgvGBJGTo4MXMva!yuqY$q@PccwGoo
z13^YU>X_X{OG7*giNf-W#RMGgc-TE4bOv1nu!P71B*i*xKz()YvU!oH%;{mi(gfQ`
z)Hw9+{4+L9x{4qsYJcH-kAHHGpNywQ5}miWAG5QwPpfD5)_toY?g1=P6M7M<@8%UD
z1u)ZH@%v0<ej3LyuY#>f04IQhl*QMNAltjdI;apPh`h>h{EuV|=Gvd(vV`~XvF>0a
z5K^j8w%!(tfuUh2lRHD`?N6%c8)?Ol{r1w*(kMG{OG(7U6JP#nkOobs^)s5Ao}L~T
z9qSTWC_mM|o%8TyzaCyHv8nz!bZyk00Zd>*r(&l|jeTd{Tt0zz*>q;p^_YXWr%GN$
z1*gy9VR;Y#`u)@`X~l>pQ(Hjaa=ZsEt~m6I(ZaAUuxT3~w-W|0W>>y#*7V$P3VLAN
zNL%(ElZ=lZKMrBhK50u!48~QAn7Rcn$q|Rf1_CzQnwP}?!IDDI`xfd1t*no`b>K{J
z7gMw|;&!KGkfxU0*<1)hDdm>_)A_;+03IkK-lz0ft6Hbm{Mc^Yy48PI3F+6$6Vnxu
zVWrj<G2hi*-ZZ8FY<NXZ&J@^#-KN5J!o+Xx5Xl4G`6Fj8lgC1S53=uTO)7agIj0;O
zpEx);Fg$(_CB_OXHT2Iu+!r?ASSS`u+}+AA8Cl|}wcBokrn4&1z2$=a_PS=zC<&VK
z{k_Hg+KMeE&FbZLO8Kqtc0Q{KGq1ulfXBSYUX}lLBvi{ja`8#l)_ODBgLTAXYi`{p
z{T}o;YNNLmUXEn$Iok|m0_nw4N09rq>J0@bbsRA5CtpFWx1Q?t3(S&|tm?P@TuyM?
zP`1#bJ!sqwzYNr?y{l>#{|Ou+cPvhQI2Hd8vJbMq3__Nk9Q8sN;{8E~e8wexxY;*<
z9t&9|j}Yts=I#QvQH2x&r~{OO&zm$ZXlWE}W31qi7h$UtPpI}>s)O`PN!D&BdQca<
zCQ!15kZM|T0aLN}?#6++Q~DtP_a%cma1OB@6oOvLTYX{j=y6w;Lr62gbu#d|Rnt1`
z-UPbz@M}j=Rs0UTU7K7GK463WQpO*~DX-d6LxJF;2&BxaGjrJBQ#;q9TF$9*mV{Pr
zpiJ!vynjO<w!b-6!B%_e;XoAKfd7bEvIf%n7BU(9g;)n|^OpeGt<@XX@iFNe81Mwc
z1q6f+x?!3jh#{Z|cS7{8XgPY1Bu<hEco`nrYD$&dOE+IPWsAbFLr%3Kn0HQ5<?F~v
z>WWZsk~Gh$fPQM3-;bSM`bJju{EvdwQfs0#;Xmln#l5l$COS9t3USNmjv!N3AO<0W
zj!6gBZLh4XY@K+4Fc>95oN>Z3WRSUX$5T;k09I{)8-C8@BfrOU@7|BLHcqJJsIhCp
z$O{YI*lmRYV@7ZAEq{VCB0>Rz0_RNhbT{WAw5+W6x%v(3s+m`{4O8i}|3Huh-{p=y
zC^CRR!hCCddp_cbh2nUNUo6uf^4>1tF?#f$CMhs5dr6mWpqW6!o*G7CHmW7WwzO|p
zTnTmE`I4XiP~N~*K3-VHmdOX;`B?x2=x$xMBPZ{xEq~e9-|wzimnRqO_u$PTM9X5R
zxTZt$upirtuKHk>TLi77ax<*GBO`Q78tEiF5Rh^X%3Di7w+ZU04jL+TUxgHT+Mt;L
zsNrinYX7QN*9H@nw<2Df^sSV9WUb5x=z^INUz88@>Hb533m-lL3HG+my1K<pf1`Gj
zxr`y(z{QteuwGZMtnbom947jM)deVNW~d15=;&B460{1WltM*FK9uBespKayTzNj#
zQq(P@n6Fh(0$cpDx9(FBu;h)dSk?;oW{;g5<<DzZNVvSHIgA^GvIyilXj%d}aMoYm
zg2@L9I8%^+_TI?wPQ}NU<W5df#r|D48JXS8vBh<BfzYaRj|~h;wfkEB28{zo{~X4|
zeX(p4=?-zaQL4VR?G2W<t^4@}l2hO~J?*Wn*R&%woJ1ie<<qPAOD$W@?S~F>oE+$u
z`2snB-g@B0)DEhU;dWh#Czc;W_pJ6I1ncN%gKDA->usF6g=(PC)A>MBX(jlTU%xcg
zXV4MlAL#qYEfw369gb?Xz0Eah-r!byh;^smu=+4}m@Z#>+V%zOG1=(HIgPmv&D$)+
zmqN7o)sEXAPy4F@4f&oM5I^yad0zaa7sD1;j#9q!3s7?7c_j@ci07XTUm(Tbx)EPQ
z2AvFmbgzn5BHp`r@nS8#Bw9gG?vlJbXPX&%ksK}up%fM>8yq;Z+-)Kq7R|nsM&2>n
zpCg*{{P|K;<eq4f9a{_@01EFwR9o#WcTg%L@65r98G208_;apeUIFU_Ew`A)%W_##
zl1chrqT@Se6gS}OzRf55Z$XTHuce&ggJI6Qb4qPhVWftzkJF=sr9+-zuj!}I{|)Ur
zgu-(3^BJX!IxmTBwyf3=9q_oK-(p8FvPcd65}Wij?kl8U9R_EpX!#0?ind#&7ugcZ
zIDlr9XmY3zU?_CvL+HzXdRM#lK_EV_$9ziI*9-E;%7z1K;mT3yP&EPkY~j0U#3)7&
z8V?uCcmMokc^m4yyRm?TVo3aj-7wFwYX%_j&E2mTL)(v=s-{qKsUVyQJ^V#i>w!hk
z%I9<<5r+eFspK0#(+h#8P;6_+PihF{T!PvpG<u+~ARW-bJOc_+<G$JID_V{yD77C#
zeQ-vqH0fuNPO0w|zZ{?Qpm7~c6<f7Ss|i_2Ow>#RYNQ%9q$Q?dWld74R#g$mxVYa3
zh>Y(k4{)`XKV*euk~T&i?qUj0hJl3zH}^|oO+lViEsx7S%=e0Z6Zo(n`~M0*0bB{@
zXoxXQ_O-F_**wZv)fd%AJpPsQ3$i1k<T24y3m$|lJw7TLG<WFP8z2zeo<!^pg(PBk
z>eL>eqoQTX(jb&>PbJKFTx&&8B!-X)rTKu|!oqigRM(G!y?w$1P7cWkSTToSOm8<v
z1@B;(BmR|_<efIJsmlUPjbUcFQuYl^W2im=@_ZxEE}_9Cu8XN@Mdwem-A;#+E&L@)
z@~;RN$*z9h3@4Q|-EuI6Gb*aAA^Qds;~TUI+}Tc#vl#HE?Zqmdm|Hw(gBQj*TaNOZ
zj{ED_J)>#cqqrOP57KXch_F0T`4Aym9GP>vreyL0o}`3M)6m0#58%yk0En^id!Sjt
zp%Mz)Bb>J-#D5-~0#Jxh%nF|X&UC-8(idTd$+kpCj*3}As0hvN0Qd+TYs^I_FZcKB
z!I1(w$f*!aGeD;hU}nZZbP2W7JWBYr`cq3=8`6oN(e?JlL?0Y9@=2v+NY(k!s4^nX
zRr3oA;LL=M96I!8q#}7^H2p4yIltG=6a+Md%l(YbN9gAS9ONOLZwA60qH%;hd`pe=
z_O%|g6lU;#fBF$92PP8{=fg=z_$twlP@E5s&ccq28q4^Q(6|0-qYg`xZ2)|Lo00E3
zAS=i@6TsK9d?pCF8YIh=NB&DQd2{$EtUi#SuENd{0CNKt9yC(I*%Lujt&r|R>yM5S
zGSyL#4gor5%v|@y@wBhEUsT_@C@QKJ+<c|(!6%hZKf$>Dd+Rhp3x*YWOL-h2CilPm
z5k4xWVSg?ini$?<Cyd~n7YXY1|BblR&~qW5;pp)~NG?6%g(HxAm!WMo3C2L<OeJB7
zkmOMYkkO)W%@C((oE95}{2WR%3pZkrLHC-ZK;?tPy|3_CAp{tT|Bl1FXT~X%_b=Ff
zx?dd6f)58q2p$}%0?Y+Ctb=Lxw51k+$tNfZuffivOAv`<eo|?bT^5tX>&b%-*1~PN
z!mc3q*83wBB&gt^%-6tSy`?x_2tB42r75xv9LO5iv%$$cssA0qkC%tSk07rM0xd*h
ztm98A5grl;5Pbq8-dH^VwjHt#xJI;Bl;yF?F2q(nLU!Q@3Ta!<1)OAqD+)%?2?qn!
z5KAk>$;^lB4UoQ;f9IA4jT;(s`;U;Y%QO7pKWcQ3t(7$jIW+{*H3_KUsQxPyPQW%n
zx{~$w4>}Atkdsn^JE8DeQ&R&LuStF-TZ7=o3QYbK*1s^$VDvE8kf!hZ`xKBdUh)wH
z%2RMBxLY^@1Hi{^C9;39A1n0Ur2cz5HDm{XZrp(<z&a^=B8RlxhFT*WBLM;hpxya!
z89=@8BHtnJ`QId#`mM54T=snrj6tIe743&^>#KZqrc&vJCm(Lb&mOG=4TLjWWZjmy
zBz7nVOWkv!(@VAix$|3b@ijOd1Wpd1jx>kESrF7!@_<E)QvC=f1#}D^k%1={ZU7aG
zIUnjTuNC27mm}x48Jdjv85C|INAe7z4iq09fsEnI{y&O;ACAl@JiPZ^2{{S`sj>z^
zm%_hhrTM_CY*PN~Rc;*I3?(M863{zo3a<}T2Y;KQM}g^rrCdUed_a=@mc4^cde3lZ
zA{P1n=g%KlO}u);;u|geg#EEDD7K_CqIB+qkcrwO{-T=`@fRRNd6e7dz7rj4oC{Y6
zkRHx}6*%ESixYJT5lGR~tWOB|{m=DboS^ZHRJ5@saInIddT!r6{2d>^OZBv9{278A
z>KQ`qOR@qoH4*~j<wqg1drk+3b5y51($#&}U@nP6A5sNNB|*V;78$kyA_1AC|4Kjw
z7~FVRLgO5?xHxjq!F!-FtEl@&2Ck&VqYk_ddKlW|L(WQS2w^LMQud2uBK&Y*b6#N~
zJn=v<TH{(IB|Z=T13y8aMLh!$5D10H{_~RmyZ>;EM=S)IaP$O-3@rFxk=8-<%z>BY
zd=G`+K=i2&(f(bZ>lP%aAH$;Xd?!X8u%zP?6Ys;pCD0F`+o?VUhn9Rr&Y`LXqBoow
z0Z&vOm;g8dB(NC{ySj96sLa<=kuKnT8DIt9UX4EM8;mI77eol^_Wmmb!+^RRkl3Jg
zC9s3Vd$vo*=&_Mem^36pYpjLpYDqG2XdFi&0D(0;nhJ!mkIlqgEo0PI#9p=`rq8mE
z4(vo7!oz%t>xF-GbzjY$-`qn|%Z|I=vr0Y_O1~Tf|GyK!3Xg&94~5lI21%XbWjWPE
zZ!SuEfx-AMgrJNf4-O9xLrSFC?DJofqHlrC%Fb?UYpXx=+-M69)WB{e@0rhhlv|H$
z^Wv3b3=CiMSAvB3-m$7%QfTxmcjH3V+z$FQ&Jn+_ab?V0<H%M|BUxqq`u`c9jN+j`
zVA0vAPWvwc+z_V9x%*On>*Y*m`AQRZC7HxG^HUKo0X@u$nl4M;3cKGrW_~ABZo|0=
zZ|P!FshN<u7SCs?0>BvS!5{5ojsjvNg)dLufMzcFU!w&U{A%z^;7UWk$e4i1e!>0%
zb{FaEMG9PdyJ#RZl&Z-$QN;$5(YO(?8*Hi%`5CGtXDmyo0cFe?d6cQ$v3ryYj1y|9
z9P|XCn8G<B2rHN4*)L#cvBjN+_KgR^y==p|$>V;!o1kMm*61jxlV)YO-2zUOTioNX
z;FffKm&(YdD0d`0*+klPpcfmUGE<V48I)G*%8%d3(Pew?8UVa$z0y54K2k1p<~dZx
z)m*512{nN@?)3PUDX^rZ-;h;kv7%NG&BT8D@&$<uki!HH{9te&bn$d#gH6=S>@I&>
z>s^%wQ9|!prj{UdjuEb4cp=tUz7(p{`6*43c5!6!%NZXGa@LL%ssZVdAoU^6=~4m(
z5~Qd_i6h4*FzSiKit?4P08bj)LaNgOzZJyUqtifBq2}u*3s9A1{<SwM#a)*Eu79$>
z!H}Meo&AR*KZKZ7WTL~ReKSt5WOpsR>a>49yw7#o@2boHWADEMss7*iarhNQWE8Sj
zAySc*O;MDcQR0LsD{+jhtg=F=B;g=QMrEWz_6}u)$etOIb!2m2Pd=Z|{k`wMf4~2}
z{qr83^E$8Rx}Mi~JRX-Xd+48l#h$)CY9@o?SHR&e_$+LTGe8t`e^P9TgLFsdtkcNa
zX8=rX%^6Yz%KgVJ2A~}%qrYzfIuY<a4=hau1UcFrh7B5j;Q3-tcQ+-IYjHZbZiX59
z;D0u^oZv4OEdI>ZyLTv|YCTPxggXhA6L%X8G4}g^PNF58IOc%>6`Pe~pY*bgAPd8z
zSGk<=60ZOuHbj)09rOP9@Ic=yv~acJH_1hQQkae?R;4n#0ibiURrj3x4fp)H#L2!r
zU&UY^3&G<9v<{&AW8ly8=jKIxO{A=)yLrDGNjc-{R}TPI?Q(<N|3Zgzmemz{JK}LA
ziB}4Y%>6xX-QtTrEK~_ME2R3DY_4YO0dipIlT07B!`8j#wi6<DgN9VwC{&MB(%Jw*
zsXZ%U1{N5=*GT<kdutPD42QJLrQ(6`CdzbOp%5+`_mbGI7ni&G`mSHQ#-fq6uCBIk
zXnkc6ex-|l+U5<LpCiOq+$rWMxA0dU0#AQQmA?JYQkOl2<^)u(OD7|F_OBB=C&DZJ
zljq@>MD=erZ%u=-2Mv$^-6y{wykpTemnstxsecLkU#Al6x_kH@3<nrGEH)34f^pTk
zt(;J2GkWjZJtTHxdoHs=@d`OeQhdy<@;{#E@pGguQt!yuAc&-yYYnpP$1><cuNtk`
zCRk!-46uYZFXs0^v^|}jM8{gS<he=pCx?n+_8=rwjS1PKYy&H`W$UwBNLd64Zs^l|
zNQR7!#S<4Hv6DDU5j#pI<4g>Y#$c8oQzXAH{qW%j*BXP-G`zs4N5Wb=qAOddEDgtD
z*GIWMOaytq`HkcKWBZv5zj0|kB*kzLFGJDw-o2s?m&x8aO+vS&i_cn`uMFC+5U{8z
zp8+H9Zh&+z#A3)Df4`ZNFd)SSAAl=eU=E3j+#aP2{V#;iYdbL$s476_lP5apn;{PO
z=;{TTJ4ixrWU2TwD4vmE_(N#xrgo{(0u3xR{23H#xVBg7^xPL4PM|<-LwI|vj4E>3
zvqb-77D@nrJNKc`YbC74UJBy9!~1q;|7uFl%;%n&D&p_WBm5aN_pT}ZS!42@XHr}O
zEWuh^k`IJ~grNeHY~u?8JeAUvR=iB~%v^_sroocxZ0$<&;qBqR%qj;o(sY{6Nyfa>
zjhDa_%PYWpK|(-~@9IsUq!7ZCl23>JdMB(}kg0swrD_i*gm^-YjNKnaN@EFxZ(8&v
zaJNP02u6n{N?pAT`PRgTF7^w@U8`~h4rBPsXtbXpq5YzS@0GHL`9sj)xSLR*Br<rw
z>PW&1FJPi1Gs5Z2xB}Wr5X%hef>#u1dB>9cE0S>}2zn){eH#Ony~wK<f0C58L7n%&
zVTe^S%>VQG1%%Z;bLVkjp+h3Scrj3Mz{|?$#k~SH9nBV)TNy47#<p^#C?|>eIYU^3
zzl8J){aq2P51$7Z%U~xmFcBF<+;d={M{JA<YoDP68Lea?kVDF_-_?8?6oqO=QzP-C
zBV81zV?s2@!CZp>SapLIio`PXC?=!q^a6Hyxis1qRF2jfWLCmH_od$E(lEpvkH~}E
zRf0d^z}0)xZlBA_lAy-9BQT``WpFq^EfL>4r<5Fzb+?e@7p$a0#+<rNZuZYZ+ED42
z8F<WZ5n|QTP5_ioJ``UUaTr|4XT1+vVvd;zLgF{UZzDb~j#9$N)D&p|?x<O%+y$)`
z@~1aFHZ#Vdnw~sbfl7F^DdLdXiE#Zl{zajXnZ&4*3X?*d_LNTmfi}un<zS*W9t5YN
zBlvRYlWV0aJtc+&x*j_<tb7$>8|Gvmazcx!_Nne-paTUe9nn80Nl-u7=?Epm$%%1O
z+=EjnR{77s5EMb&v~kSI?)z_j$Os_u!kVwu$_WF-v*`sd-BVgG{`X~h+U@g|7bRxf
z;;JK#fcjiV=o5-MY3ol$lP%+>a8XEP0^TYJgzdnnndy6$n6N8FFOzu`^+k|o06|%=
zk2$<xYGZ{@YROX`v<?T~!KQoE0C&4}k(QR$o6mNMH4@1{QOjgYIW)e>k>-InCl?aC
zL5huo!(paB+|ZJ&nb`fogK<1XBECL->E657BZ<BuQAly(zwej=bO5EEyAbPry0!QY
zZptrvL9Nap^3cN!2v*hW<hK!GFkHi*8Y3VHjwfK8si`{w%)|O&m5DS1l0W*&DrQiI
zWFMWM$Uj^&?}o}k{~Z-C0)kb2(icn(vfHW&%#x^4wT0V*eH5SGcCJ~mf@}9EJ*tMx
zf^~rwF!|BvC^V^pSu)&zJ<@wXA7cUOXe7C~v;I6HJ_*N>DoD7-C0Sj^x$DU~h_GKU
zel718)qi0W|DCF8xHqheBH99uJr68p&&@_6Q8Ir$82`|fUb;)VD{jUEMohWqF5!7#
z#DVKb8Aw@$N6S)<vm}3*SL(MG2__VPM*UF!N^J`N1qRh>fMTQ1G~Jgcd;6m1@PzV(
zcV}ra+<w?@Jcq8@BOSr-zc#|C7q~kfKLJ+PkQ4+gE)!r0-qsi)aYUq^1`o7|sLiI`
z5sbKb6EQ@Ro7Q>NM1qY#YT*;fu8tbdA)}n+RT-_tL9o%atf{P`qTnPA=`e-WY-C-8
z>mD+Y=`~>JAa*KWh4aZwar?3(r{jKIpX?;5(3TF8U8~BI3n^-&960+K7vg;#Z%sgw
z3dR7o?D&obBOJ7cq<=$eUeQQFas;Bz({!+g$%kY)Z!4)Hi-F`;GlCu&`VJ?%kSy>t
z`Jp`6e~td5^Emo1u)r#Jwvaf?ZX_*g4z-*iKXhj%_{%+9<O6aD_+_Bw=obc{I5S>|
zKP_a>e7e1eq$762X;DweVR_xH3hBHA!UK0PtbiLVxe8wd9~3_`Y~rmTc|oSTvpZXn
zRzsZ}`zq4$IRCk({%K|Oz@H^u4p7>A2-ABxghQa#0zzqVT{w?-M?Qc;Ne~Ygt&a@L
zwZm6W9B^qT!Bn8GS)Lx?YG?gDqV8;FSCTc-wG3BhxcUZa0?#{<B;*MO%49;V_8yD}
zGI6^{lq4!U);iKQzbDtx-f`kB9efV*P~%4vqECYLnAgqklEY4;tcr?&xV>^7GRJ3J
zoCPByC`|R<7r4Od?@B=eMukvP&}#i+iwt8%2_7EM4rMg@>|v@`3<(LNlfuGjMvjNk
zJn7JBsk(X&6G!581RH}%caqFz0pM)1Ul&Di<6;je;te0nM5VM7Szn>T{YbsC^W?XX
zHoG$;W7maZ++1CCkMF@$kRptAjg2*g()_OFHu*gn3b+hlYq+aGuDzr70oWOpdyr0^
zWy1#P(0+Q~QV_m*45^~_NGcBR{ZeConKo>C0>l?AUnwOYEWMOwOY~W~>bn%|GZAd!
zF&2%mfk~_1td#aVm@6oB!2B8^tS}n|sHOFv-??l}aB=P8LnwZMCKT*@TApXi{4Waw
z4jCXwp{d7D;mseyNDYZ(5B3O{cYaR}%*+7UoV5oGsKmoXGoZykD>OXG$`hq2!a}vn
z{G24#0-9Md{%L&L)(_J)b#(0sU|0&0p%AUUkO3aN>y<;E8B!~abo$pm7ZtH(rB^YJ
zZ@#?^ri6hBsGw`$j%^F^N$%u`8unnPJz)ifE<w8P`!v*s+D|rp2B}&#W1*PuPEWl7
z{1&b<a=O+7NGju5fY+w?P#C&XRTQ}FwK6w^YP!#Qk&ai!K@8(TqB>6z;KJlo@+Crd
z-L=D2Z(8gJAa9w<zX%u_^S_LNZIVzWo8I`9v;{HXT5BYPnkNt)lsG2rD=ojK|6dKu
zVr9?lkPwxW`GF0|ZB30kWF>1;Z$S(bQvm=$`4~Vs9?c9j;*=&nDjCaX(UpAp64#*V
z1Fi%LlHZ+%^eTvHds@4?{zDQIdTUM-NSi{kbUu8}z4wTaK|hm80f?QB7eNKG=<c5@
zp1=H{q*5ITTNEfRSd3tkg0TPtB))ZTKIMbs_^&u;OTQlp(VRR<#V2YC>=FfHGHvR_
zSVq$JYLd3LqQUzs5M@LN9oa@a-jW?9OAQ+;42MMBVrD>eGx-VbB%|<+Llb(iArZFF
zE?lLGC~1LGIjb!oBJ2wlz47C2EOU)z8;wJcgoT8Hydk{-cF%Ln5hYNGANITvC&Pq*
z3h37e&}u}{briC#{j~^GT*nDb5UFx4%6RwaiN(DZ$XunMJen3ajuVwWNI|)i7V~RG
z4%9w4QdEdc%=mi;Z!xF;$;jK0iOR_qsL@^5>=1SJV_gW8xr9sH{f>w4@w*KE3tj?A
za90iKi%5oCYKVS#LVI7hit=o{;}W8bzPnXkr|>TqWwPo*Vt+7W1$4+g)g^RE!3F`J
z<d$FzpVu%5F`V?v2&?OY<dJT}!fr-OtNvLnRz8M)qXst0@LqB7@kLejA>aB>v9of>
zV%p=nC8FGZtc!ht9(@x%x*t5EPe$_7lwy!dK}WYilojfYV@}dOxD`vlU`G2p5`P*y
z0NJ`6(kacSeQY12v8cmF<vsy{A-R9_<&%1c`^Cg_MpPk(5FLH^ZBS%B($nBOv6EbH
zsz5Jx3vP<XM@mc_z&WO|r-T|ZkPqH=sO|Z7Ezk`BC7V0iu0ld_AKfAGKjS@Z!KGf`
z;FsU;d1A&<VQ0mFVe0n{QZp!|uu%skKS&4J+#3tngoRV24Xqscd7$hV8sW1Hs)DZ7
zAWgJ$SYhX|fFB4|NEBicopv=_0fHqj4xDr5gIrY2LEeCVYr9PoXeFBvG5vcDK{hD$
z6OjzS$GUwYsUjF&8U<?CAy6|l^<5aflxwKr@vS~!vuEaCcX!+#z0U8DZbHzZIMG3*
zetQo`2;G?`FlgM`pJ`5Ufi^XGh)bT{=H^Ges0S217I+m7yNRjKe_Dbv-D~z+NVOF#
zU-yX8&cg47)iHY~m#y~=mA^jLT@XXBK>@E(Zs-9zta=YbcNJD6gZ{g>u#pi|dQf0N
z#h*jrNvt>3EHJaz=qoP(<Rt65(LVObQWT!EqY_}I!QOM=_fn^+wsvG9dPXaRkl4r|
zB0h&|>;&|{dzumPM3b+AQgx#yX>HQ3d)vkpmJ2`D-9i&{8VXYw_cCH=GCZd$$Q)AN
z`7~1S9qmC`NDa}!+Hf}5QZ|u|x6u2bkoRl^U=E?~`sUHi-R#`eojno=n-`OI1wbLg
zV^qh{@>~nu*{XQ-BiRuNXI9`@*O;@>zy{618~4MZ0`ljiMgWjR{VTBD*(Zpk{52IR
zh6ElZh2s<n`zcjED-MnE0Mb!j7)RA8nw`kJjq{fodfF?v!Tbmw{c;r!Wt*<uJYqX`
zhXHoaV=;%4r7M(5WQZ>*JCa7WvpxsWD`_JD&9et`sN(pQQ;&7mnZ}$fAa#M%5kk+B
zA>9$1NDVaWcpvl~Nwh;5e5k(GJmlT^#?BiJ0Y&`ys{7BKbTEr0dLsgwC>m4+D|F_s
z;Uq=m+>ffi(XUI<qW2;N@Sj$L92sQr-y$ouP>v2-ESi>g_xd%;O*e*_+Bl_epoe`!
z5BrHOfa6oly9%YwHR$8N(6}`a_pU*j4C)4ArXy)<XzYY=G>utMt+NQ`7?YVjjYdF*
zMo<)yu(rIjZ^oaaIc<ZZjj=2hHhT-*Vk3nrXD(2g{(U`#MwbTB3ictrZIV6SaYugm
z-Y5ZmM-T0_YeXb2pl?wwwT~UGoYn={Ha8KXP&ZNJf7>Q^XMIM|06x3yZW}p^-ECu0
zjfOyhM!eV!4h_`zy1EP%Ad+GWuLX_ZVIWvZR^{DD`ncTKTXIj5k7ZZwm{WK>_J=#@
zGqZPrW`R4QN`w}9)ys(fl7YAaGt}59Al#uc1r85nDnZ=Vk@dzllunplOV`ny`~t~p
zX5k+$na2^E4J4DQS}&3lZ$fT!YAREl9CJd&2|^{@eURzbd7vY}&=3rv(c6wGD04zE
zXIiy<qj}C9N<>8phdE7gxL&y7XYUCVujWRrBSef%q$Bai@$h>k{ZM>@YpvzRNyf1G
zoZu05KtbWVMwaJuv>~dyp5$rsfD8kEJ2~cF2-#YrEq@B7T*G&e?eZ+?vsaXk&9UlI
zEC!z6Yq!`Jpu-BeWS<5k2k|3p4-yqXX*rsaDGxnXpj@8<k{Hij0rr2pg4iw^+Alk7
zQKpd4X3S0*`9hclTR>}6j3ThIP2ZrsLxw{fmFod+VGhO_`a4=t6T>@c>oMlrby_wX
zZrgtc!t1P@oIueGWM5rYnntZ3{InK_?1L`Fe<b6ZKNS{UyC>7831Sq!cxQe0<`*L%
ze7SM9;S?Ftf6fpX$#Uv_R?D^45)7)TBY2~U1F*zamX>St*W7Oc%r%;HSKA-|sO2?S
z+1x;`xHa$q&upnC_t|eFAtcIE==k27uG6NLaa9{}d87g(gS7V6aT6mWZ~?#rJo^tF
z1tfs!*bv{6LmR#k)$HPISx1wMAOX7fvw~syBfUm>G8Tdrw&o+#NX11#M}5J?WC{q(
zkPNZGc8Vidvms#pMsBO6{@}YQdAY=7dqPz)&M8dUcYJmu0rs2zo7>z?h8us%w*H_Y
z4If&W1F{Tq-Z)+!0&g^ozeG3V1q*aeK)Mzig<B0Mu8LtCIFDE)z&MF41CI}((YG77
z7s?*hAjBIVn<+#K>{EsVkc-LgEW@V*2TIxbIQ-YN`{7c1ptx7=!D8V)!umUBT;so6
zDym-mzV;R{Z1e}(?DOl~eV!wP2#cnpIQ@krHL3Q^Lk1{|^bk7>6>rQ>5ikLc{ln(D
z?n5WI60Z5iW`c4HrHlyzDw~T+OM5gyKM);XPm3c^v_de6l{|<XTg-(S!@&3ZG?JJ6
z%dYmI6@nn&#S!)L8VL>)0iac~eyFLs8ln(EA#A4#i0^~Dq2-D&10`G<8T*p>AT=wZ
z$($DwE*2kmA(vk(Lm!U1CLh^u2{Y>xBo(UMdFfIA5CLO>8_@Ekzh7&tsqj&n<z@Dv
z!=*;%?7J1|&Q9<GaXtVQesksCl^XG6oKUc6+K=%oiW0z#$?%MYO$DU^Woa-T%-YAO
zcdHzKPW-zQ8yBah`Hu?fvLbe%3Y|e*bpb*|O#4Yl&y2{>9>PW=V8B@!QjhsRHs_rQ
z|GP+R3jnUFelR<5!>J6uUa)0cVmc0S=tYQBvrt=?ET96THW-a^jco($hh?c@GZJW~
zAgvv8MBnu%jbeyYz#ShS1xT_MP=eJhhyd!pQhYJQ&SyNeK%D1=9~eXZC>+HYZ6Jij
z84A+ShCw;k9}E@*&VxU|C=Pa`LYDMbL`_H;R|P~DPG~%f4G>mZ0v=FM{k<?G47SSw
zLOW5cJQZ%=@=sAk8HPdCv3|~y`S!m14Kr1;AO#80TSFO$bwUploDf_+IN?nisnxj9
zTRPK0^tNd4zZVtG;^u51KoRxwbWqgtr*e!NDxny@q?#{%p=S?&XdfQR;=1oe9bixl
zQb=YO`l29e+zALD(aU;<gei~TrpQb|8Jkm1;-}Of5m4~3Z@~r-@9js^+AfmrX!H!S
zA~n_h20nEQ=N<39)``J=vnmE$lzO~clCh&;5+GXy$uJyJ^T6?-+{C`bq6^iLhf~sV
z8t;lgs5vwa{WtiZBB2efROhkNXV9C52dKDP-yiop^So3l_b=jm8VBv*<K?pYeDwcX
za+B=P>-_#8ZB3sOJDpJoEjv`hTOA<4EMT>hUdov8VD!)15!u9FL(jbqMEl^+%go<u
ze?SThEl-sOGzOsvWcNR!f0=L}Jo{AWtrCL_Yv%>RUVfDF1OP^|3gmAxf?}aA%uFfG
z$<z5tC!t&)9$3Hg)JNB>9&<Md=RwqF`{ELCydQg*mU`<<uOaR=Bp`8=X#7kgoh-af
z_0VOB0H1O!O`LyGD-{~W;sC7{SV^!L)B>mTg7v-Npk=AMWKiIskZ4$O($7t4fx@Jq
z%Wjh9S_CB~gsQ4pVW2e1+CS3H!uvBFB3V{TU3O=<kaUw5g)McFwFrWJgiteg53%)+
z<UVs}AQyo9L%(VomMl5ooZ!6SL;p_LbHGLMQ%6O_KSxz4(DSnFt|N>|9%^Xt3{SX%
zr-k)N*RLhd-c!B;&H_BKv;X-Z?#+uq2FgJF;ao@z`JLE(!kJ>Wa24B!;{&kK-(AkU
zaV(II41%S@GTTnMdzp8YgK$i_bwHcark>vji`9oFD5s;$JJTHFd<3-8KcYGO-;6;b
zK6&UmTS`Z@h1Fx!*`-Q@wJb{;^mf~=k@n|7tbhR>NFm?@C|4DMH3@PDSw?QSS>d5T
zQxh1Ky6YYKCuMee`t|?jxCUUy*6vpYquUQQmr0$q0XQLi8tTnTN8lnu6R6LR+CD+<
zQYdg|s=<D=%n3o3|48C4iw-2SU^|FgUQdFm==$pS>hq9l|IZ#HUj57&dH0F=Gbmz!
zIx)-v#M!k@Z=A+)B_^~TkWN~H^a?_V-F`i=q4j-U?oabU!qjsEsG@7Ji*Ytyt|<&Y
z$p=w3>ekbFk-b%nQ*WsH65c8I&0P@lonqDqgj)hg#OaN&*8$hH`@`V@L2lMQkYg;S
zeLba0fp`!0NIH%&REA9KAjF<Qp%AF)i~R45r<f_}$w@RU9yiU)g^H5YW56M$QjWIm
zhzlZY6f=eo*)bdezul|@g5S8?J{#BjYiR*#z2tJZ)D5_k%xR+MYsxS)+>Y6x_&(dR
zAFhkl8Cp8E#G+q@5Jd^#v&eO8j_5-iYe0b&;Lv)x4hw1_e(LwdW6$KCl>z>6ZS)*4
zq4b2u-8PIe^eM+?p$!bA`NzPY1w;79UJ4SDuH|5oFD=!FbW!tz1*!`{E=YE#`Vtg@
zVs`8}l)70*J-`rJmDgV-C!`Nh^Sk8@1*C9TmACGJGlRmBtQ0u+E!?!AS7!TW{2WZE
zGdMeFx$C20oej58CbQ}<=x|eG2pWhD#zDYH`+WfrW^t1q2h+@fcBFClYqAgP)68cc
zaG}C`L1(Z3-6k23FYOX-R*yTF7rEhD9s=AUs98b2Pw@F2$F4ZY)#As{ta{fJ-i1Yk
zJm<{PyHNFO<cO|LEIfQq5pu56&Ac2jn*gbhHY(u2keoV}^$9NV=`;#76HG>+xG~lQ
z9V>v+V=nr1jKXnvO|Jso-7?rN80!5}TL4to?A}mp20jKU?v_>XHAW_0e``_}2i)t`
zJNa7-*iA^1V_8Kf(<o+Ki^@pcU^e+U4?tnx(fl`6)?|ND2@>2m%!C!#!nzGMw*|l~
z0D)z5<(O9=l!k9{@PL<;?0W8v5@GAbCzTWt%M6y?E@SQk@HSMA5YFeF<_Ay(6(7YI
zMe2@*C%Y_M1+%JB!N(04KjgupG7o^|+EL!zh5THE?CA(Tvn8mkyCZ|lIUsl1Tg>|Q
z?YBt@65zP*&r-s92<{*F<4wAQHpRAZ`90zEY~;w#oEE;e(rf7FTI8+y<vz;Z5nmgg
zglqN<sbffGa>USeB)S*654(vPuP`&lFw-i%-4eYQ7|<LS7Z=J^0ts?8XL%}1E9B=%
zKfVKLOBzGa`bCR#7CT#u{4Wd0O4o`X0ZHJL#8;^uwwl`7b&*9)(v?uR*ye^_1&`Kp
zB;&tJU3CEtNv(V75saA>O`kiQNic?IX2WWRACegaMj|n9r6WE%?xPUsL=PHu_xD5n
zZQlBky$<*wQ?&7vNpM}*4YOegM2*cI1-r9(_6E4{NKzh4vdjZGNm#HLBROQQxE`0n
z{`IK@%0$)ig7hlRh($QbilW9K+VlGftO~Ma?uFp>`n)`-p``_?E_s?y+5nC1AJ4tv
zY+=FUO{Wc^3KS}1GD0htKN-zX1w-J{43<Mo%Dt;<{oyx^TNSqMPTwwo4nu>t7bF1)
zZ1XmlHJGd@G+=hb9~NQy^fT^&*(!x0`8IMq!5->7E|nr0sxvwoEf4T!BZoK%B5B}(
zl8-11<t*Rp-3!8u(4z-3T;3S=7>Tb9x<G1fH`+2=ejEi|<q$}La3#M-KJH}#4#ce>
zOas@_0a_%X*jTCv>ZxPOxdW<D`nq4C2l;ThekgL=(*-PouZ)T>X296saQ);_maQq%
z8w`nMuMNmy-pGeQ0-{x-ImVq2BZEKJAo&dc_$mVvjzORfh7JCFKu!DL?e4<cVGz`#
z^azICOn@K9Fz!G9_pASRn*QJ15~rOOPYjrv^k{B>jjqivW1`&NEK$_lF5MqLupUL^
zfD;Mx?|1pvMt_oAmhUo?g5tX5c8?;ZX<JI+g7NI;ADxw)&4!rLrT+HrRp<0syXj!J
zw;v_j%JyJdsMXn{>fg&-yp~*>ry_GNC{~2@z=xLK{9K!LJ@Vurwj(v|cbP@wMCHE&
zNl7JLLl<KPJ4@GQDg*;qQ~8?&|6OG5^%+wiP;1_U$w9Rl!2Gkbo0B?5j$3<2pHfF}
z%d)LchKOA=>EZCK%1Kz12j2&|yN_CQBndfVZ`qp6>YEh<rrt)%`a3Xxp1bHtgAE2E
zOS3<rn5}7K&8=*$K+nb@!dh1OYD{v5m(QOPU$};%Qi6}B<;jv_S9hNFv*l*_4cX6Z
zO|5-Gu5SrGoMQ#{?H>I%blg`5ivr)z&&;NdW0=X#fPc2un{sxR#V4{ouGpIfe??td
zCEXd$8dtS~Mn>EV$t+~+pR7j<d!MLf9Eu;mE5~y8fpgGhedQ9%GiA&C(D*IhbaiXc
zIAIYQ@}d4l*dttO$?6HywVxXe%WY+mQ5AcwtcQn)6R||GcLf|h2Pm+?eK4%E$G&BI
zl`XjZ92)!72a7>_$DiV2S5Tt-m(A|0W%3_=*Iv`Jc?B<0d9qIVEyDAzcn(XWpDdeO
zy_VpD`5|a2Hy4@m<D690B!_gpW)?a^MNF`pdSqh14mp9TIgNfKKBq_BpA*~Zc7&~-
zW_}QGN`B32Tmv7(xSd6~dn7Y?@-a+nRH9T<OVfBrIJP7Fi8YPMHb*$DeaiM(&+lLF
z<R8#d;Q2O~woy55)t1sMS>W6CBc(1D#%<;?d@&0)37IKpVrZnI!@d5KivH?4B?VoZ
z{gVFiEY6r8ni4{z&I9M}nVL2>T2`+PhlH2T4TD=Z)FV057N+C$ta>8CjVk^9*NZ=X
zs0)(fS3HmHXo6znPKkTRd9T~NO4hD7dqCgMThgb8HhTq!mhNw}t*>83h6G|eptB|z
z;+@ZqDhGT>8Wb<xEFrpTQw!Z%f}Og#)t~f+-^9}T3?b+L_ru95|L6%il|<KI=uF1(
z-Oa?#nlUk_#GUdc^o>PlNnP}r8bR<Wvl`xHtvmUUM%lj(hcx~OdRLlR#e!4WGc=`g
z&%CUIbEpmY<(~Hc{dzH-J|c-#v@4L`UX|Q-57rqB$Dn{p`H5N=8Bb&kE2|;#$4Q>z
z)xoW_?>a#yKGV2LWeP1%;#a%sG*(3k>k04S&&8kP6Csy^8fV~H&=!|}ah`BSmxXmg
zd}^=N#7IQAS&%IQ7gdrhLy?@rM!Q*E66~M3kj<Hrj8AjFZ^*_C2B=mZXfwukB*ii2
z!elu<yP+q<-g>GFf<sA@A>+(z)osV5Ouc(2%l<JX=9&E*6i8dyQnljdtg*aeffXd8
zL-&<hbJD3>j!$Z)x*YZkI@wcY|2&-k65S?doyF5xTU~=l_hX~z_G5mZBCi(#Etw{X
zLS;+Sk5gk@hgjBHU|;;8=br;B#o;xRyXl>n->f68nxBxV0_%&`jE9K+F7Dfy_4bpl
zyJIy^s%_{AZ4qBs#Xml6&rLeq*gLRB(cUs5+>d^qa*47Btbj5_)z6iG%ayX#ckl}d
z$wvcE74yoIV>ntk4rT!n|1L?(oWLs%dz^bH#z5+A^^Bsa<>7ncL|#z8FWz;i;*g=^
z0rDX*G9c7_Vt|%Uu}8mp^B_Mv<q6$A<q|0QuX0P^_x_$%-dh609wraP3bbLfH%k&*
zRTY=_ky*`vJhf_)gj9zk{O@3}9&9FuOPjL2w#_@kDarYZ7hCsJVFfg33NLz%7tEhm
zo~&wp$Zw}=b(|)fwVPjl-fR34p3HQmxI)dJ<Bm^baV1thfp*&yX4Ykf{0gB{bt-ao
zLZT8M#~0UXs92|X#L5;niPeoK(T1DrnpaZ}d3q@j;&lkQF`@o!`692CLB_?H@3Wir
zI@R<bZT#y_(C>NqOGz=w71x^Yji&@1tB`j6`a{P<i<1^GEGTtQV5U4--#P=A#69s+
zkBVm^R?KvFfA6(AeQyofI{CGC$n>uV%$V~RL62FK&@Q;BJXw3`GgF;pg$}#(^u@PF
ztkm&ZQgXI<I)cs$Q2;9!K*KoT9JKNqna`MiH&~`}oS#blwu3d8OSedQmHccRW_~in
z%Ah+wrRzr#gtn+1VF9x+huq!ie?Mp>tPCFtGE;g-BsBd_kD`jC+{f8F?lH#IJO1j-
zrZ+YygDzO6!HM=7u!6smb4HMPs&oC}F|Sds9Hv|o^4QdNjt?wL>9<Qks?_hDmWpZ4
z`>*5L8EMQs-LSzmROc99NC@9drV(~L5E*=D?g{f*!kKCoBlal|=Qb)9uS>A5!s6|}
z#y#4YlWr!N;q$|~VKsMzasBe%$`z^U`-X*l%cal)sGU|<BGWL{zc7yqdu1|Hy3fRk
z1Jp~*l*TI)$wFs_+Q{A9&k@86dM=>WzdQQ-<=4EdEVjG2kJ)WL9`k@_OsBt1;dvUN
z^}xB3<xY@EG^cm6g)DjC9sJgyPGCja^G??`dR3~wjGxh!FUyx;%O>~~rYsBLf$@yO
zf4%9w$5-JwF4G;Y2d&agza3CHMNR1-x{^8m-P&0{y*-zw(r4!F0L@*N&?u9Q&;Iv<
zP8G|^JF+U1UjxhBi8g~&N;1vyoe5PqF06$Qa(bS_!SPAF-C*2tvuHAe$#Z_Zw7msa
zvw6o_t~1PPl%-e94hFPS5-a}&(6j)Dh=}{1Hcv%g=&*bLv#xd~@<iNi+RJZf(j)&_
ztZDGya3$5uN!YGUD1}bAj+<IXJH%q-Q55g-+?emVo7aLY?@JFI%pBaH&>|K$N!_?-
zb0g?q3|cL8CYw@4Pq)rKhEER+thZUzsc<Aztpq5iKL}I9KY*EWv$G;F(I!&{)%v16
z=!{BWDNc!ShKu+2P+Wz-_^r4#HKOG?tbwjjWwF8PnVKV7>C=Nv{~Jg;LAqiUE(@9^
zFKGixF$S;VmoY6A@(x-A)AgzRHcnOK3WT8on#$NW*wIu{H)IV+oO0n`i>U9^X$`Ue
zA0L1%|N9AFx>h_6XZFDPqhw3#scF3UOSJu%)RB|W@QaR+EbUk2+PyLYv+OUdsQp0f
zBfG)0qw~jr8{kv;czy>eBZkjKrX-zkq=F?bU?m`%RrFaRylBYnR9gFF_xTJ9CRQ#g
zNA#JR-9dNn9iLdq$l;J}{1$fot6IM*)L6b7d%1i_%vn-IKaB%DM(ng;nkGI%sC%<u
ztB=@gLse4$>qk_Ia%zO-C1sKCW>!hbY5nm=2bvRI**rmP4#Qa?^Qx~EJ;<)9vFqj~
zsEqycIdNkh-LK<X)7IN@bDt**t(L+#OrCPexO|iuANb$7xh(DRU~%}+C4Nb;Whv^a
zspt!2#Nmn2tEI&E$t%a0J%bPY`aeE?<AJlC*W&QJp5x=oy_G#dGalY1rQoQTIE#YW
z=JE_SUD>en{3{)ikYP2UDRdJh%kKgR0<5?7jSLIYhZZC8ruST%22&!uJ92wkjt&UB
zy70ZrNw^_6X7>O81*spc*ikvr0_9$l?s2{qv<FJ2x@LGCI^668T{1E2mX>wGoik%G
zoM*v=pqndY;Q?m8c+b~qx*!eSWSU~VnDCmdX`$P`d;M0#_oq#t>8^+qQMcCJLy}%B
z<@W;orPi@3?ZiSmpIPl76WM}5H-%MWj)V;<yIGD{!is)S|4y9}Np~|xN9{cr`Db8e
z7T<kDT{S9Z?(03h+x*S;uaHxHmCKE6U_6z6zPKHn+W+HA^kfGeSJTA4NzwPup_!dq
zg1t2~W)<&L;NWpbQblPz(&2&6`w`b86itB)RM=O93_2}{4i$1i?I%NcDibG*PJCJ|
zQ7laN5%RBB?=sO@W`cM35ojKKo(Sa<B>et=gW~_20Dw*K|4g0)=LR^WKP`Y3SU}BM
z+-;!`iyUZb&@Ch8i7@0`P(^AmP6n_>#>P<f13t^J`2UxD#shDn0(tVzMkpwzwLxb3
zc%-zrwDdIveR*-Hr^*fj&t&B*z!%-mOg}fl3&=t5i=5otC;hCjicg*lmfy`SD(bP%
z2TUDQf1p4Dx&ousm9p?PXbs!t3`M1EXqp`fIU%#Xyi%vPpvVfP0`qQN0R~9V!rOl`
zIvy?kRwd=~`z!00wN`wfFyYM$uf+sx>ydYze}Hlvqn6720g~^F<dUjD`16T?teeIY
z7#w~G9{CwCVvxzEx>6~b#(L_m5u{?F=NR<)bp(Pc6a}E94%b@LAc!cq$cShF+W_jY
z$E{g%ue>$L&>tTEX)B=~37^3MMCMVvNwy?l>MfoD8L={29i@874FJ&(z#Hfg4n5Za
z4`h+rGXrqDkSdn2D#_z|g)Wf9)+*II$=C_Jts(djf^0vEggm*_!wkrtf_$L(P8Sl8
zX-2A?=h1V3t!QBm6{03%oM5y{?qK^b$l)>%Lm&Ip2P7~u0Aeze!`oZ(0dx^E{42a{
z0BW;%Xd!hN+kwWIu|w%Jq^d&iTA0JD7eax<v0cot+;+TDkMDt!pa>W61dfZ=CjtOM
z=c`kpU5F!0gV^K2v!(B|v$fd~fAYqqW1tEurK$b?3Z(fmi11zn(7$xD^Ytw=JXveV
z%g4t8-JH#BPHFCgS2Chw|C%ec3An;bESG*mu{w~19N^9wVAJraz#fFE=*~ujUQO2G
zfJ9h%F9G<lVAWwgAXou*ql6CC85V4<toS$5VEsSHZGOKD+y@3I^Q76rs0^TN1c(Fb
zGWD`;@>)G$fhTCyQAOfJDQHFwHuBB_M7Q;#)@~J7EsG6)H^#xu%?*I{LeIOezTE_X
z=&#h&BqNip(T0YBf$p^v{>JXSa-}(Qv{ec>nu&4BL;3TmseEO>oB##)SRExc6(5=B
zc0t`w{3+MacA?J7T$yDi6^1?PhXb*4op{<E+D$^cpsn|tn_N<MtVu4iQY#nT4T?4+
ze-yFXo{5KK|74Vzc;yoLC<w@g`HFw7*;}i+R5S!iY_A3PKr{%oAtaMOziE2TpI_U@
z=FUvrmp)qR-=bDso+Y&JvbQHTHkz5j1|H6Wa59o7R5&bTR50yFxMpm8U|SxP1zn30
zl%5o$lBPimH94F+nA`-dG<`u1vXk^P{}j9XHY+L^ek*<U_mBM6yzRB|aPAsF*N2?-
zC<ccE-co;!mFhFnk{vt`*>l$_kxe#;A3x;f(Z{szpIT~mxU(5u(KS|?pOC<1FRs9*
zn1vHUMOCNreH5K6kNKg^q2bA(?ZXRFq)jhZ$|jLQ0PHjz0aD>ocA2^!1-bJ){+Ams
z=FeRlqOTec(l!M!f9n0PgITA!wJUuu3~eLECquyK6`cB8pzVjktn9mHVnS19OQxN;
zp_{#Yzan6a?lVN<=L`@@M+0v^v_P_O6YA?5+K>NJ;hy)Eh$`6wya^zTtQd%GB^DNM
z)K{2|J<Zn|w?7HE6@X`Fc;RV=DoLDpHi8+_ZXf`P^u;VVy*e?wEm<+CU;tEg+MQx7
zbByIlOU%CmW7nND4P4NFYp|&3<TI2>;$(6A_MbV$*XE&TorHjH08q;%I9KlwPbJMX
zL$3GN>QDvm7-fE6AJ@Ap+mkmk1U-lGT0>9aXK5x-6WBu+n`-XsgA%3zEgW>g4NsPl
zcvu-KOM$DUcZ;|(JA8Pay@$T3w+M^-JLw;@0h1lp_J!Z0`%o570+TEi32F5?<6F5a
z;nT!YmfnBMzAtfGY63%`1|TRxFe!N9*NarkOHH#o^|qb>wSkB*@XGMbRH$Lhow&8b
zpUS`Euzjy)m@7!h74Q}IQmJ^pJ%xw(EpLOB2W<!`kwv6ueFP3Yl8nhU++U`f^=@Oc
zqMFwqNSl#?|KU;LS%*YPoKgUsFY5ygb5H`PVMTRxKQWMtoOd9=68<o|0t$kT^HxeW
zjbk5d_DcEVso|&Wm(KyY9KHY*o?uplZigarc?h6l<{EDnZbX~RlKP*!^gLhc?L3rz
zGD!)B1%7^BU<CFruR$Zp%E(-3Pp6#vZGsFdw}H|M;Ng8@ai-g?6-ci8-@h`Z2Eb5P
zWiMwO!;21)kSw5HBmIB$Rzw)#grv74<O5tKgl|xdUD5n-Wj(yYi<ik5&aP7Z-KwFF
zl^vv`${P6EceG7Nv*Q`)*<yqSAnD;cT$~p>D)78aJxxut!B2P0oC2O?XljoWPya-n
z4JSb=)E4l+P(sJF%%BU)9Mg!e_=`uQ)?t14O5uMQ=)*tY!!*P@aKUYVRlI!}_O4N`
zvu94;VwHdH#fZ;fPy`su4rS&bl*2w@dLa&O$zh`qpg@EC1eDj6!jaM?^&V{S)-av^
z_ut!cFG5D`?=6hRD4imzjE`2{B^`xrzu#?jS2!aqC59R!!v;r(9ctdb^}T)iV13``
z(yf$aMyM%uP?{XWNuIU;27ve?FHnDgB7M%&qZr2DKI59SMYH;(;w-P5asGO;)MHzi
z#SUr{0rdIm!}QSSu|d%Z;QpLj<FZdOp<*MP2W4@e`|mmx6<-96y4*gqJ+2djScqi8
z3_`5jlBi+oDIaKoa)DbKT9hf@1uiTU__1OQ5d1Au783S#I!#$W`T5IQ{)WQ0`oJ)<
zui=3B<Bwp897{(HTCeXmXoV#9S{d!^->pG2hNo>agM&QDLgzq7RU%;g02B3Ne#g;P
zSU0tEPM|pjtY;mj5g$mnBttQ4?f}|f73{p7#_#{JC9C|j9OJD#OgdT*qJ2ZYMq^-%
zegecI{UTf`he<KN7$`ySFH=6ln#4hM&Z(qEq6Q?^b*YC|cfJHWqM(M{gX<#r3reS+
zKYy-U0~X=daFSS(+-hpYLEg@u;o;^0R&Z6)SSuF;PmqDd-Gg6=`|#K+l=MYD^x1xo
zz5p<bs@cb3VTxkR4@NzWudRaOHTWnQL33V1p?jPx;UeoDgRVUdN=^ba$pD;z8a{?6
z0T{<n=hx`yGR2**8L#K+q`;M6yT6$!i<#RBJSuR13R;21_3YuO+RC;K%>1)5z|o@X
zhoU;Qy7Q-x-RZscOP@b2T^)wax%+zyX>F4MY~HzsRNKL+I9|odZSCZ_Xf=@jPhEq{
zn4)@h@8{234QBhF{<-6}Os*X6qBXuViP~__iR}US6-Y-!hz+8Pt^vMKUp;U?KYzPq
zXFj&JXaiXH{dZ)?v!II1xcEr`M4PB<erhs+@vs(W9g3;{{{1Ss@?+>au*(U9g}qAk
z_^P9nE-cLC#owQH5nF(Xik<@Rg#V1s;-H;<DXXnMSib)(PX`#Uf<AKO)bZ5k%njTj
z_(_BCD11B(;JXKXl0GsX?$j$V_hUCUGNN?bfcqY7y6#g0MDa@hPggtGomv)fpf}VC
z3s%JZk6g^k9~;vSVnyna|5?)fu4wU*Ke{!0Op_lY8R%YE8#UnL59K$kt|oDzeRJyM
zMwLzq(gRR-FU8kjtKcRM6Uza+F~-LNGJkBLOrY46szi#FPu6tEfoDMz#9i1Q_kH`<
zY(2C^G572385n>*aYH|oDz1eCv=OJ^1ui`x2(YG`fb!iizm_4#j()1cw(OAYM?kXe
zZUZxRB>MRBCbV{yt;ztfnRx~vGz5VSM1amDVZMbs*A_eI$D@_`5@7lQVQY@anHE_4
ziR8j@9(f94&w&TiOtR}fRaHrDX_H`sN&y36RSqv28<W@{DLH+)E90ho9UW!1VjcL}
zJg_@Bwgp50FbNG$Ycf0@0KfverZs?t7~ioF^~gdn2ZDtFw_@r8V3T$&gxNdj8MDmf
z0}dVtYOMZ<DJ4SJjiLB~4a;v=RNnX5Lu=y_u6_*-gbM_(yI?;6LS5ws2Da~4i0@h{
z^Lbj@6CaY$^Rn2?dIPpQhUrFtU@p1UZMuFtTW&`RBlRQ?@RLV;)1iT%mmN5DfEL+b
zhDY8UvUWR{oKe6)lHU8lTVPEerGa;j<Ur<rZ#VaEW0J08%gI*FNgs1u#Yt61m^pe*
zXe9E86{$T+)2O{T!>jZ``wT~o(6`a|TpOQ{iaFNZ7<D9hPaifv9k6)xM7zuDTUKiJ
zDzZPStZ&BtP&=gZ*Z!QmEwz!pTh!kBDk<kkxCN*v&z~F2IqlctwW23b-phse2@n$A
z7QZ;$EFUwVy{Xo5^z0z}aKCf&>QJw<(cdqBvJ|%$6~}Sv`K@>Ld&@6?;6fqpw-ao2
z40AC!6Q8P(HhnKN`<>jmPZ!hLa3}m^lSjxf;_+AwRw^>Be!IyUaRMB~{d<iv;(s4)
zC~&4k-}KBhdLP=9RxGRAVY~d$IDM{$-z9>E18d<g>T9xL3)}{QxZ45@Qw>M!=H1bD
zWTpyjOluxjPiyPieZm@dtcEU=e*e;=n!W5(mz}*anX|z(*o4k$=21d;rTlU)fOFd#
z^2O<eKp+lhWHjuhQHQMI5dPke@UvH2jNaGIQlociv1ngqXcYdvNGmADqkBHXg)5rS
zU@910pn!3_Od-cyzFOuvcB@p(xQ;s&o<G`Zk`vSt&2w-+!utc7u4?FF*u|gK<|1p$
zKYq~3pMOv>SXoK3ZFE3hF{<8d-V6Q2M45|B-_(=R*BTKQ59LC8SP#MI;W*4CD1sAf
zVaxCNF3q+ttfrDSA3X1yqQ6SXr_K(~uXJ~_9_vG2FUKTUsYOI^S(+IOtK=Eq6q0EB
zY@`Ha{IIhQCm9c<-C^jpmsBhU3?L$Kr?r8FFeiyvpgBGtb^ho%ZSETLCt08|j~<1Q
zLc<7^>p#y9nwR+Gi>kjG+aGNDN>r~i?hn=?f}VI4C&V;v$1v?Gs|Y~}hP~t&!n*_Q
zj8XROx2M);bTjIVj;CRzjDgB8E%swXaME6A9MkF%*z14JhYWqeF9Ni-%gPF`j7&|X
zG~hzWEbh!S>T4x6Up>iKNW{p`1s}pD>PAO-;a5%r#tZ(qywef*_G#3K_PE%wxKQ)?
zJxR0Pf}=407`8#biHjsy0<3{7{~$9J*_yN>1kRAricGJQ?IiSz=H)t;i8{P`l72;d
zb#qpiB6pjCeIJJg_L3glgSIo2L}ecc9pDRPaKnR@dT`F|lc!p&Jd1n#AGctoY~5c+
zpX`5Lc0c9i2n9yuj{lDKZI7J@YxoPlonNXum(mKXDTYe~-ARR8G`IXNzfYu#*WTw}
z;aSpjYo8Q@%xml{EjtZ4tOH=DR&StR!%6@ZZGKxY$>CU%uTJbGmWvEwv=8Ic;P`wp
z8q+0VXXG%zxJ%4Ve}$!wst1X%6_%av)tyagJ(#ceMC13NTMDVoM_dz+KDoFz&*9~X
ziE<f6xOH<4Pcl$WK6jVXvbt<682ySX>;3yy=l%AYdndbQTwx(VW_Z2|j|+JtoYXwq
zO?;n~B^a-;1$<o$TfNM$dQ<z#l|*ZCtuKB2wRPGBIDX2&JC~kw?klknT`ds6o(lyD
zy{b}g1{pW%%is}9dB}_|u38iFN=r@NcB}+q;@?K*@l7ff=DBlK_zk#;bxem2Y7R6s
zP)c~OMC^xOap8_>yu2JFBVD%k)^u3NDKDybkylG$`=#MZP(KG-u(Dh0p*P9^ql(oa
zBf0-G{k4scPvTY#Y#`Q-gIhE*bO6_5P&}l4kl%w%9<G$eG0`${zkh}kN+11uW_||`
zb$uOQyeHue=lb=Pzn~-7J2zv!HW9lady2T%b`76@243PK$lb^<wd?gN_KTkRd}`=N
zO;=ZeBAqn9r(SzmCA-yxj7*cutvh$ZwGKNr9ZYIwq9U_-Nf%jnRR)A}Du;3gX2-AE
zBy;^(ZFV4i$pT2X><g42Fvb-$awPuyU%NGM$!NNXps}akL2Nz29f(ZHQ&1T=78O$T
z)tJ+Bf<M|OChL5mmxG*-ic3<uWOV(_9C30Y3CA6oC(Yx^#`#n&6M<VCA?Mw3PKFRV
zLli20(AHNr;UX{o_F-%C>rvk<#Z1E(CSHzN`SrtnzE*I7^ixjGSxr-M*0E#LIq8!J
zf~#g?`08)cX>y?v15xDZdL3n?+``3^uC8$CI;M`_V*GvZ^4q?=(k;BKoAO8tP{7%K
zB)d)IM499oK01GN#yN<LQ;u?yleJl(4Yg8~I(6=VYss-+Gp;_&F5z(sx!2xU#{lNI
zuwshScCFik-80NZCj1SAnb^(C0t~KN<%NH;ib~h}W<&o{Qv{o;rD?~gEgxSLb$jHr
z+HdzWw_ruCOzpr@0-juk$jaeMmy)|Tfqyt4xwQs3do(VM<>zY5bN`vuLTg$lX(94*
z?rE9h?PjOVNFx4TZc9@$Uj8y!<Tw)KUUZiGar1Z~0&`)kzg=Vs!`_`$1Q`VYCbm}{
z>SH^??4o*fe0W}Ll4ujERAF>dp@n9?>iqB@B0*=uU@L*~9q0Z{UPrQEEjkflOSI!!
z+x9%7OL#SN_h^n7J5`HGCoc>+6DJg={~`Ns(Zt!fQWu%;sgOp*T3*|oxqX_sAgnmh
zM@ezq>SO6)kfmsEPPUJp7<GA=sO^1vdXQq1|2A2g;mJ@Kpd4@<sZOtDcrTJ9e35cU
z-j*pmc7Kb3a$hU4)fd!&3gxF#{%yV{I=pK}1~5)SMyBEV&A{AT5g&JQR<oxr>PMeE
z{CMW?YX#cPm@Q3tuKCCAte)pi3+P%RCcydU;s^H=XV%9PCNvmpULK*g{C+NJMrMyy
z(rvnlvuEP({ML%g^YSO%*{0RQe~0kQjcj1d{Kzg_g}@d$7L}o!F(-zN@|u$IN2`U1
zd;=ML4>=L3TtmAXH||r8EU$-k&2X*l>wIk?5UUaT1`xuIq~cQT#7}J#sk$Ly=$OBx
zhKQi(D<u$#H%|L~<ZT=TkY%WdhVU7J3<1oO1+ySsxYg;W_qWO=CJPS5{(&a9CU#PH
z(Nku}D|t~)vL)izuWNs~PQ3VT>Gg|3hFgz}KdD(VO48XG@`-XmEmXNV0-H>?htFoi
z=*T;HIRH-CX4%FRSy#ge>ht?|HokR0jrow!hX=?u?74YNlMK>SA(z%cuV=W`|M|zV
zh=}t?4Ld14DPc6Jus(0JW$9_7(<kZOp4OU+2#LUZlpoR;JPgTbYZew3P-@sXYn6WW
z+or|7DaFMq>}eE_zuw!9?kJpl?@M_&s^%}FHI5bbVtqlkf8^@EzjDFW@Ia%_Olrw7
zEx6a6(5=D#v#X1F+0ppjl@p3B_tjEA{E126j0{fN!!1oA3YufecB3jhu`7e0O()Nb
zvmQ47Xr^j4u@4s&aoBbZtPg|c05}uTQKN7q<k9;j!zKH7#6A`Xb#;Gl!a?^{t+3CB
zms^P9_UCy%M_l!S7n8%aoE8~AYS_X2g$4$D!A^niYMW|m&DJnH1J3~5j-o>2+Pb>c
zTyG#}B>du{qas@cT3VYY83_J`uCB7;O5BB!T_vd-am*q;D>)eF^6r@$pRLM8FgnjK
zELLaMT0^KN4n0pTK%S+KF02Obr}ejWIs)gn)q&DH2}By>lZ>mK*jgj_O_FZv!6(Qx
zG&JUf1*F(%e3&CnBuIE0H4nC!`E7lGw5es<*w`&c;f`}LeGe>-GWQZarDYUV@R~AL
zUfZvoCOw3m)mvEgf-8D2N&9)Oht1<DzrDS^IoN2F6lY}qY($(0|IyWBc6u%^zePLe
zOU-<{wK%bxXd~m#QyTc@$hnh@x2X@lK13OqK01Vdahs&hK>UleLa9xjgoU;BVs(hT
zvRvIz7a_5q@4Y`Prcx(cEAf)!_OqwfYw0z=<6!3fz)S~mwTp(;Pb+cbMd+ksw36O@
zwdAN7CS~ZD%1{}3y|Uch9Wx2N6jdi}DqhC>zUQM0(a9#wZ&2dmjk@%ATCe~o<l{~Y
z?nqMRS;f)dA$Oo`w-akd=ItfC+uvV@Ev?^{Hc1unqVK#iiO8A97uR1VKViN~7D1mK
zlXW!SGEYXcjVAuPx`;S~!DcZ>RaLy6Ui{1kRo11t&|*!~d#p5Z3Yg!XSMxy&x5srR
zI`pS<!^78&<+n-E{c~*bqSOXmna-cLlN}uf8uj;0*jQPIxBfogyy~NIEK0U)8id1E
zJ}eq0M;cEL3%cY^ro?EbwS}wr57@j+f5^%Q6Zt2=WkBhGIh9I`cDUD%7V?V0-(SDJ
zX<czVT5Far#SURDO?IcsWdY{s_sX)CR_}=d@*tU5X-@IkH=)w}*T#rqt;8gAvVFew
z+HiNjudVQl`}?ueXWM-lXHHX)^@;CKPJWo#>G&|c<L@=&b*GcASBgcK99k*vC+gk-
zj)@o!{G%Zb<dbRxmij3dSxL$pVTcLHGQ%`YKY516mDbdZ*PPI6%P%c$a*&qN7Y(<i
zC3QFCj{hgl&Kl_$eYPx}Z3OQ{a21LDkdmTjpMU+=UKh^#*B-XEF791<OAmOwj_8(4
zT*|J_e*-sh@3XJKz}SP4z0w!FaP*L^&n0gB6BUlk{tH)M8G60kED>MoXl+-HbT2Xa
zyCCQ;$LZrP%lc(ej6vgAn*s}03#3HD6~w&*A4o8G8WgKbv2#w_*6DMff&>e(6WRyG
z5RI+RX;f7_J-Bv7z%C;E4l5O*$WiKz|GBV*07W$hN|l(~qk2hS!0Rx^YyLR#W&h=r
zq1&gDjtM{UP4mS=YUgF$(|TTs;_BHlf>x5fxK^eoLtpsPQn1Zy-u@>tUkWlPJ#UOB
z4yUA~Y;?I+RaI?J1!iO%M;BbHHL-BA(UN24jWA~<ZRTXkE9=Y0qLRzPaO*kKAB&4M
zKE1497P9k{kS??lb^Kdvx#%k5mhvd!-b3G++Xcra+kmAD8*b{m-l+{A#lJ`PVG|Nl
zH$!KWyh~$Y=NZj#8VJ8qJ;%exD%=|X!#G$e*{Sf`K9UK#;HnCafm1+y<kG_z^=;%R
z1s6s77n5}mJN~$3XDJBAmExsv@2?b|^*%q=@~Woh!dlALdtB$@4b?&1$Fb(DFlFHH
zpO=riS>*1OPrNZ!*w@#U8=Bi;wh|`$z77O?^R9wZR<Q5M^MWUtEvxpcVDj)|PS<c!
z8(WQVS~ItlxnAK)ul8OWTo!)$J@~IvA+C4P;Gs&&#vJnbB99mJ+50dPp{N*Q-~g``
zt+pI;>aBIoz_x5=`rbIW0w$Wzy)F$@EAQL;<$NbiZu7HN!xxFJ0?p%`fo&d2-*`8p
zO;<yLxwiD4NLOCbmr02%QiU6zFX~bj)IVI<sM$8Q_SxDgCtrQrp_iGcJn+h-OqNbs
zwT9cUpZQC=j=;wcB0b1fR-J+qwc+=m!*(qAT;#%Pu{B%@F=Rc&ki!w)rK<|r_tYJ-
zrOi(~Kg(UNyeD?Zo@8_0{Y-dmnUtfsv$a5+MqIR&(}20tkK<3JT<<xbIXL03@R##&
zb@3J3+CqD{b`5|W(f;kJ`vw&?ul{Dhma5>=3jyt;PX>>Yu*ZJ49C`PRag9RpbK$iN
zeGS=8Rw>5^UvxL9S`zHeoQPw!UGprN`p)4``B+0is3A%%wcufi>g;>zqkoRf3h4_L
z#GlnZ#=qKhyr{3F^XK3uMShRHxX*dMT*EUWt>?I0oR`zN<AC@;odp>%mrDvsX=Kk|
z>qcgLAt!wV8@Q(X;3j*$U#X^s2L6glvV10p{)}=RhuCIbU_x6chF;TD#@nc4i|X{x
zpQR_&z^yKLDX%h{DgA1IE3C$yom<*bvSMmV@G9+V37Lk;_}pBIF%Q}F{bZ*sou4*X
zh@WwoCdq)lfLBJcgJF3beVqn`EHSJOB!}s)kAV?4;PAh%nrybT+Mo7eX*H3g<*l`N
zLkP>htG7kIDkbMQlSEXQmG5OQ0k-_y>hBDXddv0PfSUV9lYAdjsw}L^G3Cchocfe%
zk{!18$C8=nA445M_gBa-NiyYB#RCH_%z+PDEf;kwzFGEa@#WioQlqZf*pQQCU!bqQ
z?L~h8Yq<b<$kDoyd9h5P`zlIa9Gax>x3d3Ie7SihURply*3Fv%K7V_2+sKo-LI=L^
zMAnJ*D?wwr`$3l#r4L%$AAF{vkY(#EP8a))=lz>zNa`1~$ScpLoiabAohe64GMfMK
z<9UhD^G9*Yj}LyDG5;6-=7;1_!NdMFbgw`BvWt(VWMY1umz$fM_eQ&yL1lXo)lXW*
zS*cq;22}}Dd+mAXt#p!VXnCFdL$lDtw}}aXG^@|d;k7#s^2#>c!v9Rw<2mmZ;%I{X
zP44@qr7jlJQtg%SHdi~(>!j6)B1L-ruV$w~3aDVB<ebBsQ;?ppEzQZ{uEk^yhO1<#
zB<no9z5hdqipQjRc(2?iv8R{v>Yt_3MMOBoo|7@ZOzQn^Q0U>P1ofE$&_2V9w}1Q`
zME)0<9qqGv=af(8${t?w`LiSw!#SuU$4Sd?#1Ir5ylP4b9?OPoPwPWZb93`f&AJ{}
zdtI9DLy>hC+R_3IdX_0{*Q)>Ox3AAk3aJn+yN@yo{AfFE>ERJ;JG#7H7{nc|bYGGt
zzo)N3Pye=v>}?6=O#P<%OG46LFHBdcCQ#W+jq?9|T{km<Vb_n)Z7K|CWVj?{(!Q*C
zd7qIwKfSylhA|>tH)EP$pBGhs^Oe=Y>_?x*8P~{&h`(R1b!b5n)0bXc#ZF;t*;FWI
z*5H#E>m#F?M7TqUTHQw#c;%@r_IErqQ+8zeW%r#J4vW<oya0IO0vwONJ|)io<}<Cd
zjOhJ=O1-vUwRy8CI-yEL6SnraUGfd1fxo)@-@~aty!Fi>2_qC!$chD<=l4jre#o!?
z>Gq=jrjxC0X4Loz7Amr7)_0oi#C;@tY%js-dS`n>917OUVQ;eCmUUZutQ+B6bzIY~
zVpifhVn8WLs$Mr{d+fZ%8#_^(D~Yjj)M?I7DWO-^IrdeZB!8n2P$hSKn|wf@=f|J;
z`+G1Khd&3j@}GD5b+OE^un4-!agN(!A6s&YyFP%jrB>SH9OHITr@Wg`Oqlbjdj+6^
z1HFW@X{f2^x4BxXtM``CTR++{g!*^FYL&&kOX7o+XOmBboPQwru91pHLIZc{@wwCw
zHlh+@Qc+P_W6~L8>H1XWzK*@>d>pOBQ$j*c@HQ7sNHKN}>f1H>Evp;;nSV-aQ;g*K
z&M?hZ{XY5fmJ0?!jLOkNBGGofK4nRXu^PvE7R{}hn+u<3iL!>%1d}dn+gy2ng<k&B
zWL)V#s%lONZz5;+18J$3hs@XU$Nwhj2KC%=_|8`Od-!SV7&~Eg)gzPP_aH<B7a&>l
zS^tYe@&&MVmxnBJk%t%MrJ&u%_QY4e(D6H4;=Inr+(Kh(>>@F8@DtaZ1tq+~7y3fn
z-x#0C7^zvOo65?{K=z7%`2S(*z2m9=`~UGK@3JXzkU~fz9D8Mt?5&K9?Cd=&Wsi_#
zWrw3k_D=Rb_9kSnaP0MaoUYIJcKcm_Ue~S5>-9XZ=kxh^+}Fc8#o~zkR*H5Bi<+de
z`JFy03%sHFiw|4Pcon}aoMFow4<30EV<!ijFy>odAxZawU`gG(sgZI+mx?$#s}e8o
zT1pmCf`tnCRJVD@_{xj3<u}RA3fK%i=$)3wMYn~H{i+WeMI4U1n3nl<dn#^juJY!4
z9PTeJD5Z#WUHjC4xPAL`QWBJpdh+N}*M6p^v30==?#qRvm|bH8Va@B(pqd;AGr0?U
z`c+Mi=^y9qE7bG3%un2Fcv!0I(VY^nS-;-`S;-nh?YoGUo}Bz$(s!Q%$bu!y<x>rj
zE96cg@2G~WY26L0Ooq%-T5DYP0^<w`^=nQ}c6yu`7w$lv)!N#M!4x|tvb=v|n^0?@
z(2%A*{L#2jR`aI%0_JtTjz)$Gb0hbat`4)r<w!Cn9&__=?K>*z-^V94d;|ThuZ%I3
z1`}MlG-2OrXK5L11y1o+@DIQ8CtD2wtS*NQLn`&nl}PjT?=Mej0L)u<t`y#6^qlt+
zt^FIuLgVzI)zE#+eppVab0WCOn39UB!A(}JU#wsI+vju^v|R5lmBZ3tNfy8fsO>wC
zeUY!y5HpFm!C{fS{k*PK&pb%mZ`P+uK@Did_hV#aBsIVoI&Uh%!*>lvD3a1Uz$zjg
zEdL&}ca&Tm><*IbR+R#dv%)x)eblD3W<=4wImT=CE6aIbZ26|SXbWY+B!Ht;lTg&;
zJkn!*{}Gl1U=NfmTijfT%e9)~QjkXtEsM7JUpNdlQh=Kt@n%8yTNi#r2Y0&A<t)5B
zmr_l<{2LZv)yoPHS8iK{hK~Ce!QzDrfs7wzpD@(sS2jMm1(%8j$yf8Q*fHR@Bl`oN
zh-<Uc&IjZQHiw>{@$x^*_&bbcLnfzB^;w^M<*Kwi>Rayr7UIlL^qBn{|2OI9ws4@f
zT#P1b_yoO3{%s4YaNcJ{+AQzhsCwYY-y$P3(IA<Sy@$7QIrvMYr}%`ctdfFXmno;7
z1PBf8^agg$-&WI~%?h2GRJ?Fm<C|8~t-;!g)%YjVTKDukNqT$-&qh;_-m_#W<%?XR
zTQE;cd+MO**7D@#*FT$^apJByi}5N>H#eVZkkn02PkxCsc<R*ni(`Q9FW+Ow*$9=d
zKiNZbbIovC%yw9mMJeeKG*(&ZJcX`_tdhJI4h~V{gl)lJK$_x+m1o>3T`=oz^`(6s
zbQ_+=|3_h=*qqvtf}P&Z-cl*7$oh`*r02QXDspQE-A4djFZ70}{>igPHTerc1Wmhz
zG_fP$SEu=98_lfE?7w`Gj$QnmX~e^n%l1i!1&`weGADFBC%u<>@T%9}ix;S$LX94P
z1Qfoq{0Tz1GVubAv5$|ww=&u1tqGCy4=ja!J>{nhF&)}dsnPE>SeoW9$1i_fe!@tb
zz2fYqdC^n9izK)VP9vHLMeKA96yDJG2dT85RVw#l%;qbYEW^STRrIB$%Ux3)0O3i%
zTOnOeT_)={>{SQJInwG$?p^(7$3IqP*(`&qu6-ANPZ+d0))unAJ8mbEE1!$cptgy$
z%cfg$UwYM@P%B<9Ke)Ux^%Yb3G&8a4S|djMYYjVITwZ)_;ZyTirq9(gyJI;5jAX-v
z#mZl4ii(RH9|jeJNH2%3-0Q}L-{?lvB>|ml;%`UPCl+@cb{{Lf6Mrot6Ma<sBvzWz
z##DIG+i5uW`ADd|Yx;7>BCSTr-Dv3wZ+6Y>sy1s+{fB)mlUD!^(ZEwHz;2nB#<W|#
zz4zyXnOsy{n7?P2v#RCZD4_tH<BznYQu=bkWX6RHGIt%1uwBA~s_Nb~vJ7q3BF75W
z+9ajw;}DkD>F-+gj!OBOuP+_6D<FN~PRv6P%VikUJ){1pS7VzxJV;k~t$NLkjFpW|
z?IEBQ!`G2^rkU}ZpFUBje|A?3)Va=tn7{s}`DUn8FUEnMj>%!4*){y?1-|PX8u?nv
zx%q3`0Xn@!dTdSvVk{uFGQZxVr;(*pOcv~Bm`2)H;oqQOfVmcpt#U5Y*&TRuX=GTH
zv~vq_da8z#zI;0%s22Q$ePLV9)V(~u-l0PJgL(3RQAxL2c(p<P9Sc1mApt+JZS^9u
z6uRB=?{y{PE?db4i@2JQU?JMU|GaL<%TJ7xR`2=Ov*~-c1a&9bedMBOv|*>jN-P_O
zlM2&qAO9hUs@|CWypKso+|z1{c0<6pDbm;!9Fy|M$8_XRxcujK$FP5zLuk!%KC}Ey
zs?QHzALE-`Nbm{E+%4UO^DK{TRZT|7H-P2cNo^PwFe-ZXsV1TBZ!wEWdP%v@SuWgs
zDwYmZZ^If(Z`Y?j8}yW=iE8CXY24Bd%)B+o#z6{Iju7X2T!I>l3)1geu25G{gu^)|
z!jn$Q!8x0bP*iffDt0{9Z*Q9Pdc^kVdjQ9UpQ=HLR?v27#<}E5tf3O2XNbOY4WSVl
zdK@X#gX+6C7IyM3`<7WjIxii4p$25F@1TYxR&;bd-mm~))*Opn=k@fW2rEHR(YosD
z&biBS#ZJlx<WK2h<+*$$RC6i*y|u_00jhU|VwfcJ4m8LQ<%TiAijBD!|6p;Xv>+n|
zD1cpb;{@PL)W`*FIrfTN<DA3M?{uz3&0x}pfg^@V0t^(PZ7*PB`zadIZS8LG_+Zi#
zpgYa^<*3+@CU&d#-A7f<QH5!W@T>FpnY<K~1aj5r>GfpZzYF6c&RtGSNy#=gA^Pva
zD|%8d&!si3OB`YKC(ac?RZn5Zhzj^g$KLC8obcU|JKx>6ua=Wh5XBEPR4ciy7Wm1(
zzllnAS&Cb+7{LH>S<GTPe;*cSO5(JbrU&?r>oU10D2CJd)6@65#aZLb0$`lmx#ga;
z3dmw$R)>qdabb~0t{z)U>=nwOf)65pYrn`xS%@|+%<6nnW&=U#Zmzxt_V4I4)`S|`
z_k~*N(?z8Uu(Ptt7UpN`{QTtXP+_?gxu>tuuCBC4?54_}oE~zAjKgwtbH4-Ii{HzK
zzeN4oLSwi1mL>pN@WraiNlpx{*PVUVBeNePsq~C*kEGgXfIXi`kaBxJeZ`n4-aun@
z$^S;tQ>aDiO|E_e7y0ZN%pcNZ6PA@KctEMFV1HGN4#Y|n_-bC_=ktuxd4LP=D;xR9
zUwCD^b$mLIua%drq3!C?uxUpK5o9dG#7Jd@pK>)j=`37yPrMKIoHTq9$*L(Kc~=sc
zr29Rsj&cMtLvioc<XmZ=|4|-h>7f~PFvQ4;{50d}ZxaxZK$$eEo45TnM&Ns5`M_9}
zC<M@jQX$qY3r?iY7XUl{85O+A&dal*tv`AXqE8t#1Al)c*A2QUcQQO(1qVE;Gf498
zml<aVsyj>ETc=rY(Od0m6wO$7v#cOI_4$Li58GyE(<yi?8nd3none&E0l($B@TKEr
zjlrV0zl}{P+;8ip50r&+8;r00?|_h%QFzV>h`Zm}qMy<?0$$-UV(5!JlgA!Aw_m(a
z$qabrYC?}_PUVj$i23ar4Vp?*``yPo*xgJ2tadwqg^5l(vvxE6DSEWA<Wg6^Zhu*!
zgQzAG1-|2M<!b|N#Hh{36;Tj5hf8~49$D0#AJ_S%&*xWdI0_^?>;4*0$qf%fa`CCF
zXS9<kuRMzXp^=f#gFxHGD>VO0+v*)V*{wSpX4kK5e2Gpi+1&YcVtV1#1}UDH8E0fp
z{#vwj+ViZHx8ir9qWJvj>c`zA^?AuZA+V=&$$>04EbMqw83#X&V?|(&0Kj|yj_g=D
z&iETlGkty06r%X(3a7=!SE!c&F5+ExL{Pzv+}u)^CbX4Fh$6hoqxJz&;)M;qeSn~x
zv4gpKZC&W!uy><s({L#9$}V4Uo;*4dC@F-4YA{y6PZoRYqHl)h;f^tuUfJ%_k2@k~
zA(@qCv3bLY2bDl@<8ZFPq`p$3fi@#>0c$<tzh)jqGT{6v^Ia`-=ga!-T2=A&4vb1P
zt8KuRB0?VY6vZko_&E7AtpF40uwwX8(4^T0;~05(=uQaD?;KW^^Mkam7&2OV=e`YY
zsOnF@z%Qi2<4KQ}g2JKafq;`iEI|~|ONTE|c)JGWwy7`IzZAd)HW(&*E4*2*@Qs&S
z-CAbc$H?-BF<1QSFK!eakJp`!3w>03M1gO>DnR$m^hbO>2gScy?MMz`{J;UKXxzSB
z4HllnaH?DHsx~S^6Vt>1Ppcj3YZ|y@vI!2XDd2K5n_YcQjRtr1E16?AGC~eceZPKU
z#B*57Vy{VU4``*RDC?g68xsm2GBE!r**nC)|5dCXKpjLOjvnQE&p=3{Q&4-f0`U(X
zl3CA@2+QXUSFeuheCuN9#uQ6<z|&~0vJPIQQf=$7EpsaRk=!tYornVro;@Ec4`DVB
zaoLB<7cd__;gUqAz|v%eIv*i5-?Gu^aiDoBYx@_aY*YN?VSSLcdOK<^kFF|^L+J`q
zha5WnI<F4~Pqm;8Ha-S!YPd%b1uBxJ6{0_&9b$~kDdkG+D(}@+o123>sDa~~(5Y#d
zY2QSGa&fSi9s`!wBWQQ^+uuB!TpcbTDqYUlBQP|*cmabbkeW@j1l%@%+ffRRGtDb_
zQLS%qbTJ;2<qMkgVt>MyVmDR8vBipWKPI?JrmFV$g_`0<)FA4U>EKbAS8X*1!E+$!
z56<F)ZU!?lGP>)POrp`m9|Pa3J{sx#O%eW1z9|(-`_SC`h;!b38C=Tn$VeykAITOv
zaK;Hrz6b3T;0sU%-LR8+_>%mA2`o{l#kx~PpoX?ba5{9wY_y*$?uSB)N~`PMoYC&z
zN(CKR2ExUSTX*R`5G6I)?wcF8v$W$3*y61HxD@;{WEp-nL~I1N%sEr^mdI4&ofbF_
z+h_N;W#p}Y)w0e~n_+q_)VW|r+Ok@5Us-(F*nLEz<ma9RCl-qhvdK=EfJ8{fg}bCj
zGQc<C<atT<4=X)mBE7bTAiRK9Hr{HnKe#ujX^>95PlCSs4$fQ|8vLWybnV(~?9;V%
zbm=}H2%mx823x#Vsi?Xo<P8X8Ry@%b$RDF$>q>a9r}@2Ym;ug-<3IioS%&n^N%Z+f
z<%i;inco~k{{zb32lkJ4{tk;!f+wt?OXQ}+zlpkPyATu6GB>keQl)=uB>f9}Oj~hc
z?c+P>Z(pE$T!u+;0H>3O8uc1qcy$Ky)%2)1Ox2Rr?uVXzkyzSWB*B^emPf7FTM_%5
z%TL&Yy7A#EiA)gb#fvRoCL^<#Wls?CynKqKhGvTcsZAlp=f?Hk<#zw*?#8(-K*JHN
z_YFe~;2Yc#T!)kz0JS_aOD0=e{}a8&aMw1IlY#~MNe>qJ8=#rgyjZ3pXYT4c^!$<i
zc-r@KMz;YBWN^LklA1gP(h2zB2j>?r;H!XsYlx4yoOv#q7z@b=G5&777bT0%{^+?V
zrFmO&>L6-Sw>Jyc6JN5QX&Fh~i%~sKdeE<l!SMt*9GK`OH8H;u9>n`Amo6gW(K!L#
zQ?hu%fzo`KNH9Gd@ux8rZ^{!d`Ng#T5vfQl6;FF!sN-H;GI?+SQzICR%6Mb=u)Q!#
zJ8AFI`+AkEZIdU+3ds!BjNaC7&91h*G$C%F$Ie~O;0dQHAn(P%tXTC?49L1CfTc=N
z=?|PBxbX8QsH1u%&5NR6ez*@@)q^O<t&3=pmAh!%xC+e6f3LCi&OArj`L+Xr7|u2i
z&@!4xAK|Coucjo*mehOZ9g$j=)5d6iL?3`Zx1ER7c@iPDNa)X)t(8?s$`#Ha>BK6&
zlIT0ziVcP1lix68!&fukD==0KdJe3*^kVL3_c=3D69RqQ>4JL!eWUP)%^mwfI9gqL
zhaDy97IQ{~kH`_8fz4u=`fqi+53x^#Ep*M;&21yB|BEJm!YRjlS#_<ptg^CwaEZq=
z(j-lXweNj^G^3>VA@#)5lk&q@X9n>GCQs&=HZ)h{w$#O%Zq-^T+;!r-nGqX<`D?kS
ze?4Dn*!UZf9bc8pbmKbspqOH@D%rodbJ%$9nzsyk<ImRv8wNa?Vj}bMj}yYZ&%_)m
z>PZj!NN~roCGhs0?dEmfluQIk{@qpy>TqcEQRLD^Q60jPR|&6N3b(Zr8<e)YtS@~z
zQ{t=X#iub{+eqrN@qD3&+_%Y}yl-D~c6|8w7)H1Nd0=KN_%$8eJHKFlhV>Pemdei{
zM5#O1XL(6($Sfy25fxsQ^BAuYUs8=(M{_+4Cqu8cj7X8ThRivoWp0!^xM<auGgX(2
ztP8&ztx~M=h(%ectNBl6%yk>im*m&=uA)UiBuLQcI?@JqoDZDw_yqXgwecX%(!_nN
zu39inbG=4hD%@z5mfz@}3Hk#tc(f_g=eD5w8XU7=0oe@Bq{e_j!K9KMj_{0Qg*$a=
z@~DlHaiF=J1`Cj_$R`M|8lJ*e97B51t$=Uy5{&D98@f*82leN5<7`3%!+_O(PUXe1
zV4XFay!TiY*gQ^d=Ci6`rYRq(SllSyM%&pDGDQPp107}z%53<OHBu=gw(i%A{pLmr
z>C$PcHukhemcnVRa?y!_fh61(5$@+O+qf_?@^+hb@F_!#S;iJU%i<Dq9Cfjsw|5fN
zyLB|kC4S~xJinv$E;a4Vmydf2j9BrO&}Xs-56(kE7Wa=qCUpy8O8+rZA6t<T(b-|K
zL=AfHV73C!8aDEgh^B57f;L=|^N{Z*JIF_jbmcrnm+r#hZ1Q9M@6{?vq0b9Q>X2p2
zaN8H!xfG4tygbjGrif)P_Jd37H?f8L9oG>>z=2kX@L9fmQ7XPgR1OnvL3vmk_&!0$
zgF$Fc6?<&~RH%^w%?<aJYpVT=`|Rk&gS8?e0-}`fqe=?OdE@C~Qar1vZw|e&6cC#&
zd+#%GJAP`y+Al8yrPvtg$Qz-=7=wv<Hng@pCg=Apfn2ryue3o;7qEUAq%2S_mOqIj
zQz#D<l1#VfFyVjk>i*)j{+3l_O{A&byNumj(7R?Mz;Xt?w9>$L@A(IYGXY;-II2z3
z?TTt?(`IFaf+X5O%qwi_re~g7E<O>_j%q0lPeN2sL$AW@r>-^&XIYOvLZPJYHZ506
zvdezDFcrkON&qmZ>xj0Xf{X?r<EknbF-XIv506Imtad&1!({(;SJ$JEP^V|C+z6F|
zjb!N<EiF7w5zfSN(<Nh|EN6HVu+E^EtP~gw>?P6Dt2qQJy4R)uI}VG(kM3$&a|N3q
zeKVs@n;zWsZxB2EVwO8$%f@7(heUAq4Vv-VdWQggR|xn8=+*uZ|EEX&dqabZD`l=@
z#PyR6TZt@#c%ar%A0A30D6H^os4RyKAmmjX$8=v#UbgWUMJv*|qN-T^;zn=U;G%9I
zW)Dp?Y6J5yci=H@loCc?NltY2;?fuA;9u7G+;AFc+Q&Ln%YlfhjgpQQ;VA|r8GR3e
z*Z~X+AQqWEu9%QO<%`oAo6Ma({ylq{Na*LSqIFmj1LM`BtoTrrO;Fg~j~q%2q<XyX
zL7F?^a_QJ-|1|tH(g&~>-}5SGy?5eiX9Z(Q_g;yl)l=1a4ZkkrP?`-3D+fwy+jQeu
zXZKUZFRZckHqBk?eLwYtKulW+N@fXlgzS6Rf76U}up5JFv>~T;4Q6b2#1*gGzm9=%
z0$IvKyl$s^iOMKax<0ADV}R&Z=tZpJ{74fkZZn<bk(z=2L%=P+;<cM-BDRHsb{k-h
z39?66(C7nj(67!$ANU0Avj5=}p44+&Xbe`&bK!|cKImBydE3El{7B>3=2Q~-6*VA?
zp#oMuSpR$T9<-=Hrf(g38>@(Agks^E#k5d3laiHE_710R*@3SV5#DSd!TLt-OEhUB
zuOBp!sJB6SQFVWjgKxSC1>`G?@5K7lUIbUyhEso>yzr3i5%gIVK|V5BuWPs6g{l}G
zzT<}LnVIt{GJX!5b5=4dVw|U(%J}OZ(3Grx1G&;d`kJ(GeBB#m7x)5fov9BX3lusX
z2|R=eTGqdr3Kh{vlF`c--z5ON_^kfF8F{m1c0S6%Z3~{!K*OjT{J)lN%0rbpPw&C|
zE(5_!C5!v2{llLFmv*```?I}3$kq`u{zX>kc7ayj>?H1+_A93gjui*!@opQQ{;ar^
zD+zI3+c=sxq~~Pj=BAe>6*Y^Bv=X}>Om&!!HJ0f*Qqs`O$9N0c41&OJQj;yt()n%x
zCGwN5t_NM5*V58ly!TnX>Gsy+-|6o2_I%Z=r6OMc9+8vOD}=q%AofTag&s8fRcJRZ
zB;mF`_|6qgVN(VB0>&P$_a{yJRmdnBGLWY2yqR{i&oO;;0p_N9U$EsxA_o>rzRm=_
zsaH%4-{E6aceS!AuJTBFjy5i<YFecVPn?_k<VhdDwdEjEQ`UgJVuZy_@C)Nqp8Zc$
zi96<&gd@7TO4R?jbun>9J3bK)$N#v1iL+*PV-~hxwmqK$t>^5dwuU8%?rcL@O2+o_
z=^3caH7|LrO$h3JAGPuFx%`xsyh&_)=o?m9H^1ZIn<$o{_Uk!cvXTgXS;O!)=e9e~
zL-Ou9m67$0+}z$8+i&e=%IMV3zvfg$2O<DK=;ycOVHz)3cIGiZeWmwk1|Aa!xCV1h
z%0YKV;&O`Mqzu_?W%WNhMC74FDmSicZIL0$&>5kjJv;2h(B5gVl&n@KChsBKX2zIk
z5I_2|HxsMQdJnl!e}Zn@W&Ca0`}!l=z$;Ll%dy4oF&Kmdk~fTjDD~i#M{2!&?Z8Lv
z{m3>ji-Td_@g*UFPf|<Tcbzu?-y0NsOYV|mRA$ybqM|zHvyM?nn6y1^4SKT1b??`6
z3pNjb!pJaSW&JwGXS`%#GiO_sRF;+w)i$L%A+NNXat4k(a0@%)t*Sp!Uk++mxV&i%
zPqZiT^%dEFSH1@nI_IQR4_6lAwqM$>TV5qHwZ{P^<5~$HS!s6EEpFDB7~1KlS@;=+
zf3?q7L}z>4J754r`ikWZ#{Pw+B}VeInBlcG)*;vy7ZhP`UK2SY2?}!rp41J};47!^
zgBGc`n#oq$w`1)D6?o#F<Z7f}{ARat6v-}>0caM#ZQWt!`H`2#_U^7r{x#V*V_EvV
zrGoOeDpM2<Y47-RJwLX2KV)D;#h}TIG@s|AhH8RSkgJjQZ_AhQ$T6F)sw^|}!<&!Q
zdM$1ogE#u8Lut?c2l!@g74o<`FI2OD9ADPJSFSVjx4<g~3kxA+vD4Q9q_A+8vcTT^
zddg}b*?AHVM<#5qdl4>&4p&)n=8CSi(7Meq-WQf*jN!|Gy#&Me45rnMe^f|<pS))g
z9I3~D*B3Q{NnbX@Ys@89DP0}v?khqE`}Ea*8cltRAHj%zPc%J3OD5Q|2h=O=<wGMa
zP>VmN(kI_P(z+Zc1-;8MzuZPvCDg1y%4v+b3w3AV$~o7%9dHEVM`*2;7?Wvrp(#!6
zoxiH(g3}4oJsaY};l37HFF{?cAJ7O6^2&wIdJ4-^Sf=OdBvi=G<G&LJoI8||^V6qK
zzg>;0eD(-MOW)V`0`gR3O^PcxyaYl?0kg94Ly1X3-V*@E9HdWm;8)QHf&4gEpt(@l
zw6|I;Z?CNPZ=Z1arTt?-=i;bLzYkB%jMUg_xBj-kBUm_mH5n0bD)%)u9cqN7K?5$)
z!ewwN{QSQqf^K?xGo^oJVK#TOD#u171xxM(i{W^!PkW+k8Y1BXLz)-dH*FPLxy;E|
zi4+m(rd@R9dkO3Oc6r&dR=_r}ZGrSYd8o1GTw_&14k@#LIMKAP(pLi**4X^wD&?9>
zTvq+_PfZ)uUXO8PLX(u3XsgF(I=l(mOaGob{|=(0c6LDirw%mR1p%-%wGFqf+c$k^
zg(OYACz$qqq@&;<5qlkL2n>fQ|8U3u5-idUcqS`QfkaNPe@_i?H7(MG7Z<s8Boinc
zDPe98wPho|#1XsJSwCT93%!WUP45hl>wfPCL8{^XvWBmk!GCr<ILMfsceiC0dvdn#
zpN0X4Lz&Fa-p3oS5YU0kIbyzPkJ$Yi^PCqJaX_R6$y3t%-{6lD8ebpha{TV@u-iW_
zKm9%&UWJjE7z8Ew_q>|g>_$_kU_csILLq$Mu%NCCiVuz8*fYtqU;Sx$afvVf9`76q
z#o;p)4quO^SeYW><QtP3egwL!mTlL$xw#Vt3gRQ@g9eo)hG3^4{O)`j8is6KHbl6w
zIY{eY`HwAn$-|E?TVntNACh1mYnQ$<7KE+oTRnj98_4UvDqR11wcN%Rm*k#r{>^~p
z?!W%-SNFxbSH~7Op@=wrCc_DiU?((wH~jc_{ke27_va?AKuwqA15eNUmlUEXf+daf
z=bO1V1MNh!+lw%CzJEM;gyu->;D1;FMF|eWYxpSsqon?p@p?8N&za1-M^wjaUTaTz
zmYUhWX!scgegKGtl;6~5<l!x}z*j=uBr>f&Ys?a-$HuggLsvFchoSE&LM1&Zn{B!E
z*|NYBq}E*VFxBF>_$!9*ZU?=0{pl}Sk-63AUbcRTKP1v{;1kW4Y;M0NH;q*pl4A6;
z)Cxy!PK;~<3&5v)L7)tKf}mOAK-=48cIsOvTOeuP?dIl?L~H17cGyle+vFnxW~N^Q
zRFvy3?fpwQUV(0el+^s_AfE!8ds@XbsS*qh0-e9__+;p!ZhjrNUloDvA-29Ne6QZr
zX&G2!K+6jiZ*Cd-*}?xEZSCnqeqx0A*_yTApMyZZ^Ru4wdeKwd;7y@?PmM-sC4GPU
ztT%HPxQjt12xy}ZPam@&ZGTP8=)cB~KrXc|n0QzDeDQirON|Wfm&3tbrs@I^82di!
zT(AiU%?mc~F9*VznDO+RG|g`ixXHy0skMM60zi|K)}61P5@Wnxp4j=F-SF#UQab5W
zK7A3@*k<US@HwUevp)a8F6dpz)w=1rA3xsQ44s!cHC>~bdf58A6)T#K#H@c`>wiwe
z9PHGBPuRzsP%`}EknbX_V}vpMT@EPa0VRo*2XBf}1M|rY?b%(w=4V*RL*#uxI96mc
z;<7G}2v&!_uaD}4{Y1MtZgl<UbC=JJqoV%)m~gY+mPXoUSHT!(colmUY{9bnl@R6_
z+VD_WlI-jFU|SI)_l^c4QPv+k07PSoVRbvpdkNqn_2D7S3&cb|&QN?MjNv;rwSQmR
z`?ZGAeXdFjc7W~Zs+WAH%V3J%Ue6h&U*)qUrW1-KygkBA0aN=wMLjRyehTE#NoH`^
zYybCS)<3r&#t@tm?=M2l?%ZoIc~^mFI6j#cDa}#&wx+@BA|>_xUT&KLh5CPB49-l1
zT#`N)pX{^+xsT}xJ^Z6o(Y?dZeV{UrdF<Qzn;O?^H{J4;KWu<Yl+?O6ufI#KxM1`T
zAo2h1N4mJ^yRmrSSL(g|#BkQ{Wgw>O=ZhlUq!+Ldr^WD%Vuh#-@2TMsmAOm@Z%Yp?
z8*98mf&xqrFKt+RRX@fQytAzAmvCOfo@s|;n?gjwH3|RN(>!JKzQeU4TvL0@xBuNP
z7G(`nGcs`VOn}$UAg2Gjc+2y3uCqG&fB#N*V#x^p9cBzR9t-c*6}h|q8FqrAh8I&D
zwLvN`j9BS^m7<JybuF69KJdvbeD_yDsU{k^4uIMN9XS$s@BO`_&(W#b41#c@*%Z_?
z^oDMb!=Hcd>6y?Iw%!?{_|@nRVM3$Wku)A0=zh1ydu4+0;2eBB?OrpbUzeRpYkW^5
z7yho8vl7i|@tCzvNKHc(92?1>K+CveBAE2>@~{UfeESbaR=qC6%v|+r@Mn9Vmo^M8
z<3fb@T6{=xX2!HfkYj3(h$yKQui^Nw$7WlMM0c4TzP_9=3D|l}+gshM`fz>whJ}Z3
z>HbK@4v2qUzrqj<A0jzO&d-EO8{UDt<hcs?+YdaeCsi2}^J>4>lgIc3;77Qk`Bb@;
z_$RtgojHv+|DG)wVQmvYPa~?nJX?|ox6#$rB@yjh+`%M4pMFNemyrA|?U9gBXAj>8
z8`u8TmpJ;9%`J&WMLyKCOyB@XYG8cn0#O@ivD(T?&-2s$oyEcieYeTo>GNa1?CjW%
z7iZif;x9#Lv~Ls9S~~mK(RyL*wj|1`4wL8&XoPgsz!&;B2J1~gRnvqE5%EfBaSvn?
zUJ$+D0|$pI`cAT_|3ha1YD{tZ;%12(U#aXy<%YTa(oV+QD;yywTu>x}dWbD9V}`Ac
zm|8C@k7|h_rmGr1JVrT*PY0+1L!q%wt!2paf&ajqwUizv!fiL0Ggo`OG1nt8(+>jy
zM~lpxoMf4Az<e`VYc*?8fl4T_BJ)rEKP$@!oNkC^C#%MkSRV!W234rSCw94Pc>e=W
z;(BUKmm8t(wbS>s0aFq<GJ+Al0_bZ&H4VqPf|S&VuM3X+h_6SdzPAIqyLx|eqs1lY
z5g(!f9_7H5IblE56MF4u^hD@(cADKctvT&L{}FzY{Feawq!brCw|BC!ak8)&|2bln
zTrygH13IbHPf3uq%@dNFt9mpI;?}L4whx+T<TXM49s}_tjxhWb$TmVpk%VV&hJA@a
zQZ~VNo}6;_AZ5PCq$A~|#jgtK6m>f*dyBmooh$auTqI>NP6X%rJH<Hb{+SUZg@{PV
zQYxv;Fx2eSzP4q_5yFw?Iwu7U9Ey>BcgN0dUiZA}N#j|fR4blZw_{MHub`~F69cR9
zJQa4$b}Fw=XSZ64*gD$U@8{h+{uwmRe|F4M?Ng6G@3rSgL_}<JrRGH>)I;%&6;yS+
zLf6?%RaX=j7dsOFb2fte{cm1zrrP~K4uol-h`yDfHhRux7-`;beCUp^U{qF>{YAP+
zSAA7wnR%&=ZTP>_^m-M`IzS4Yp1!Sx;1wuNTWp^!t<3+KkMs4Oi9es~Vj7D&=$_m6
zUSMi$*$P(no(-NpyE&f$_2V}5XIKIgiux0lZyFIBUN_%~bB$&zOnmB+%XaEpEy*SP
zern=6Bwxpkyz#P*um&(XUk~b|@eP++NMzss_vUE8Jwd7+igCq7Jb=)aGw!Rt_(yO`
zD$evge%(#tyQcKr&HbDvx8x}#V)di~uMDq79_k>!@csWR$!tWY^vOzdLnwcF#u)Gr
zDyOcy*4*Ntn4rGBKvK}FLqX}Fs4U2GQ^M=HcTlRKTh%&7Z9SNu=h;p;CR-~HZY_t#
z=%voXfqi}2WYG+TuAJGLY?*V>^Ockw-<*)$vfAC(`z27l&%Rhb>6`bawi;(*TF#kk
z=rO*gq!1>RA%SXGP8BP7#J`-WO~(GDpw61-%lCb{eynO10!=QU1AIP?FWLX#-*Oi^
z5xjmvfw{`bm+-@BWC@!7LLZ~ezpwdBt)*zMg)=q%bcA@4R4!WqlF+8H<-12M;N-32
zb!Mc%-zd3V&Y73c#iLC{lA-6HHtVcJeqDNA_FxGl6c!4$Nb<%iKI$Xgbr{!b${Fwa
zivxH|?m58p;cnmikOYKH_C3dm4aK5(BGJ%$OI!GPgr757*1neb_}{FF<9w~Vyi$qX
zn`m5+wHCA$(GUu{*$T_Cw?>K?!q<3TP&r3wA<bm-uUHyMcRl1^?ICDIQrM{potm?F
z!9Flnrq$~OVgoi@=id=ttLTc+TDzv1PharT9hb%@Tg?=~JThCeq$qfnAXx=!0>L)-
z)zc>+n9OUhcIJTo&~$tI30{$Q(|B*LD&R=ej_CoDR`Vu3rtOt_&ZAniT>4?lmVu6J
zpUS_2A4=q2X+yCxJnzGCxZV5O&Mvq9#)nH17M5mKu-jtXnnVua1?RItMq$@QD6ct8
zcMrYc<GCFfD)giPw1R~Lgs1yD57#`My}Djj?HlYAoO`O_TLcCQ0~^FYv96D%>KzZ>
zFBwmCL(8%xc9lBb@6rqr^*-j;{FNoM1s(r`so|#MJ~;?Ot3dwP^5g{@L8{453s%sU
zM7q}*1d45Pg06+*R;R_pNy@o7QwsjLE-QJa9XvI%dv$l?NcaWQ<<?}cm}lR#{_nE~
z`00ehX4KCp*T>eZacair@#RD%u&_o+!>13ztR%hVI4EZ)#?pjh+U(;~FN25BTRvA8
zKcF4>T>kcTWprYQOT<I+&jfd)pt4n9Y)U(Lz0a)~okbJ{vPT{2z-~#(&76dUF(gI8
zGF4*N`0lg44T{mPxi0FlM;*NNFpf80AD-|&S|_StdYm%68C#c+b#>!M07KD|$_}08
zhh*Bo1~K}^NlhrcQ+wZkp~o8MTfzT*%(H8jy*Dd}v17HHr~p@{NMY%5jObw<Gj7>w
z7(hilyi6A<!#Q8_!Z?p6{kNb&g#J;ZBDS{a|5SfP6e8202)Lz$qr^1L4>a&#N-nRU
z|J1O>#EWFRi4r`RK@bOqoZ1jO^DhPqQ^U^L>cT{B?(Uyx{cE9hQ78P+KXEvE4T?=W
z3%K5prY|?6CRR1=1PaT``9wvP-=+Dr*4~g6UHAM^p*JpF>J#y8l(>tMGWnqp?{kPt
zQ@FAN75nKoyWAfvJ`vF1Rd~;b4|$AExTS>V7`+pBx|lH3q931-&S>BxgyC{!do^M}
zjm>6Y_~G>Tz7FfhnuS3+pGWow{~=#2M*Ma@`fD<7v}yk^wZ4-E!VVC0N8Yvn+?>4@
z8WTM?2?IIEvL&3=d}i>*OD8!flo(>h14>RK<(tq3Q|_J#p(sTd8OM6*W~C>;-;AZE
z$P1My&&4LQ>*lt$E-tZnD7k^mP)vBQ4)9p0hSc-4r@szjQjpb$3(2=cph>$xFt+O@
zZ(KC$N~5_k(SBjsUj(jNbpgF`Dz8B6D+FeH;6t&1a^TRokRn^T7^{n5aG2fVi@UjZ
zGu+m9uV|i($vv6H-{>cq{Ra+_v+$t-{PiiNK?@{EFKFy$D0_hZ(#TT9p#r%x_OaN&
zbL3DWv!-Lm`NFGmT6Zz3>+^S=#$5Z^n3R*wV6xi^C-7YFhyVU9-Ml}JwWdkT5DRWT
zc7wX_K3F|P10LcaKD`XAX(;i>;qro$pNw8&y|Y#ot3h&Q3GWd>zY#a;x3MKmTX%JP
z<o;{*WyJH?%VVzt8#pErGuF;VCJyv7;nq{*iz#aNREmj>86$cY`xkyS!(Q1LSo^BE
zTS3ILB$?{Im<(1pO_sZ{WUx-A5G4Sn)5!Z};}t@<oP3v|PX6MaD^cz5c!t;`9KL~Q
zUdz{EE#1_Z{Sdqd*?+3nf<Kzm;_DsbkcZym%PBG0q|kU&$QaSo$NK$C(2HtN#ho4e
z;zA7syNa)I-u~}OacNU|OTW+_GtQ4y-9WJyWbZQj08o1swl_MX6PwDhsY7H3ZNdM2
z$KX=)!@s^B&+TlsF7)Ec8jT1s(sT4G!KG7QQG);A?r&=&{Z{F+N*=3TR76z#*qZnI
z+Rf3ga%)N@<IcmEQ?;6xoniKDB`M$>XUj%#hzR_Y9dc5NMfPfdj;-x)SYZ2TY*She
zt%ToW&_DA(m8VBcA`yvHfsx^gIU=^-uExG^m&=-TJN|bsupyi_xCfl?yLOEnQqO0g
zaRK7E9{%~jM1}a7x5jt3@;x;A#VzBu$j5gc{S0pqWGcl??*nuvd1AxtNZKEs4pE=_
zE=2yS2<DV^pq;tSmwy&V0xUb7l25RBgh_h+!O}_mKb$OMB23=z@nG^F8M*vttX&N~
zw%OV$enW_`86%=t*T}4WXgnWZ8rT}&btURMGq74D`@&tD7O(*q7551TGnfmH4!d8}
zU4=z^jI>qlSlmJWm-#yH#xj-o{0**v2$30abC%eJcFu2agNlR|2Hk^>b#q1ntKzI+
z{l{Y;%E-+I&D&wsg%zY-%}39s(=WvUK;ps|l)Oy!W7zhM8^gysFhnNQxGCR!_a;hy
zT)?m#dj`l9gP2)=)fsR10l0gaZ;;+v#oF<vCF$<oeu$lH9q&gF@k;u06<fuaL6QHt
zp%}i_KA75;xYB2Ee017OqeTSiYF9GyzZ@8Dspc0H^mA<u{vbW&Ua}MSqdYnWn%(!X
z!*D_Ll>A5zY!Ncc><T@Zbv);cYhP9gLV=!9HtxV+Fb2<d;i|4rcj(}8b`6d+?C6+Y
zV<2x#<GEp86-^<Y*d?bMfTm;pc<1J^{e$-PF*G$>usTa@0`B0%?!+_mf1*kq;=lyQ
zeCZZf#cATB)h#Z<^km_r1catf?Nh=y>&#de0n<y+^u1DXOTf`azPWzolC!D^@k$|@
z?<0=>$y0k~$bLBH7*{P=0FlRu=ymAJEie$1)I@L{m#&u3(9lq{yMYXK45N$>lr)xm
zWoYN&wZQ8b`4X#rsuSu<cqwD2v#0A~x`<G#+Yzj5I3MdaE(DCKK3xo6>@(V&wf9OA
z4+lXYp<;ep;qYzos!7**3e97&t6z+%@STy4`T}K6TtYL%a?}omhd+-KSftXo;47n?
zo^W9~d_6W))5mK%{E}}|)|BWW`B7q?s9|`Z>>V&-004D^K+CdR=@gfPUE;qhY|?a#
z(QPgL%l$y`0|R%XWK8%5CKBY@qS-+$J?MV$g`AaC@UM*asXUD9Fk)$$Sxbt0crwv*
zN`3ysXpv?KIh182<z?V+4`mO4GNRQdoK1k1?E|Vc>m_o8E46aWc6123@V{Y9?(4Xa
z$w5b%R<h3M39+oePH2vWo-tzLNc@C1cl^AE+v*zqPCttJY)|*~kZsnA_oUTvoh#w_
zT*v5A+rFtFTpRm-P*5l1$1E_i6#23GB%1G%<&DQ;=dGUI!F60T`4}VzlK9f=3*4-S
z-k%M9I=Y)|MNXHMcj|7yP9{eO&6_yCBfc+e^@5L`E>ee8`d&6YRs7TX<qt0ZDH;`d
z<o?xoCqV&36pvZe<$De}`qsCVKv+(ru^^KfL{wd%I<t&wi>-+9>Ep%72ugA#ckz(f
zW3I`^31{-jW51dRrWJ9}eQdo<C%#^jATL}BVKgf&PA#pAv<i&0L|53h@k*rikiK(n
zZ~RAAZ@*9eP?HSSU{Fc<ood-V0yhbuT3Go$s-?v51diEUHv^4(^h@Cw;d|i7?<)_p
z9w+_0g`U0(m*(K~qTu-kujp}1SK{PS-SkPap}MfVu>Db@jCuEiA|gx$b)FE6e!2(i
zKG0+6xB3hvI!rV7|7&r?zv=R~PbRIqu3VUYUH_bKiFX!_=bU?>0}{c=m<ZA+vM53t
zHI1`_`c2y2ls)3HIZ7>8Jk#LkXgcZMz$e=qdNhhlLc#R4NLi4@qYh)}u#AXBs^I7g
zzjI92OoC;UaPd=fP1-zGkf>6#fyw(qOBam0J^m5^zI2-*RkrE8${*Bp@>$0~#1r65
z&c^1NzY)AeXjCAQ7IH8S`M4Zc6Au>h`-9?|YX-ilxfBy3)=jU}{pgWt*ajJcd=8aH
z2BiLeW^ybq9_~)9t7{cu?KHSqpx|FS_b5?r^!t^9(>uG}p;(vBN~BYe4lDzR*2of5
zh^=%Qg%gvL?z>G)z3ryo{J8Z{U0q!VnO(JXy&e%Bz&8)|zofXN0EA#bJP69(#(cCS
z_--+)*M)V2hCAw4T3nW>JgJPrPryPqmxkBGu9j?Tyb~NMzO|g_y7!2pI7dWX*BE-m
z`4%L^l#I~2P>i{az7;QkXeksMvwm<%342-gAL=sx|K!{2)6hmTFADM_03&_E7&9BG
zU=cJPsyl+bWSB4$8I7~3TZ7@UxDnrGI;JiKM|f4VwF<Vl47OZ&i}MX;AN%P(68rbI
zzu^Z~KfPZ`iqx@CZc7W8k33t7JTM(;*gT!u4BoLdHWtq`qb3ae_M+h7$@tPW*~e}q
z`V=|<mGmf2UE{mNJ+A?gE?p%8`JD?3#A6`L^Sa>pjzWqmyz8nD;UZct2|}b~dVz-c
zYPHM|0;S~r+T0QIrXL2Gc0nD3LVGqm1uK_hA2j-SYd~n+t7MBM=8b8l*FT%!ee?3^
zCqcz*D&RA=S;soX5wFnqDL&Lu2wff}z%-#a+JHvCr%=XvGmVoDOA36}VJP@K)A|B%
z^J?AHqaJn2ZgVo9%_X!CGcsRSuHENfu`10Dd224DA*kt7MTJ=|Hs|UsFNrx^AXVjI
zJzL$(s@+u6tgCW3K2b+bk?iO41Eosq>EA7r&p=J7fAbSXiJIWA3S$0X#oQcny*v|3
z*HZqsJP16;mtrMcoeQQ9h?;f<-U23`1D+=dzQ%`N5<&n}R(ZJBZnnqlBM}go7&zT{
zg}Z-+NF)7IQf7xv5gXi?EMPSEme|-g>^8lPRduq_35S4xQ)7k0(~vY+5+iWYh-syG
zg2Tt@Mz&INx~Qe|g12ZHZ5Oyo%46PDCJV;J#!7*?xvc;_*dd(2^21t%U07(&^7A!;
zqlMbZ@9>Q)8u;egm0j|7KI??zwZQLm)msFE6E0Z(+1}Jy*upCEX+|&eqWGU15RMYA
z(PCb>p9}z<#8MPV+%}5cgfBTpBdS+5e|QFvcSw$2K1j1hp<H6vmIUN)j&diCR$&yg
zewAVEK!e%$wN(yt$u(|GltH5caCrTup@qJY;d;9#FD{&4$4#cxl?Jpif+z@xu1M^o
zN;#fbRNJi1S&`YjYY}w2W?T4(83@Ulv37chm-(gs5a-ACD;?x3^aYhYUCVC)S^N}4
z2FGIT8)8)^)qhL;>K7&Ko)aqF>Wxj63w!D#Ldzs>>N?j<fC>#;e_n=9z#WK&h>7`K
zgJ!JGjie}k)x`o=em1BoRlB+%Z;X5S4O^D6RwfJfEePU2+urt)kWd|zxo@KK@aZCn
zHZVvK?A7-!wsFpUWw%Q6Rz$5-i6;xSZ*a)WK6|U6Plgl5eDSlqop-l;8H#MM=U`?&
zu1QQPgF?5D7*b$s{_H%CHeAONO^uZoXY6&RHaEu!o;WX@^bY%q=$(xE9T*;ZFI|jj
z%S_ZvDSpLvS<!gFb7w&@>Ppnu>@3|NB!UE?Xn5O9o&J0I2+gOz@hpcq=eo4XYZRAY
zM~Pr@cMKn<&iMp$#5i^KR_CE|eAk-gQ9xK2lBjNL%PvPmp3>h+%!MEGbPRA}F5~)(
z(elDyDort>i0~K?ZmNEsdxTsu&6|_pI*k=GvuEavt9@AO0{Kv5b^Q5|>>~UO%6K#3
zMGco_zdus~<F&Mk&AOKB0rOaX9kt8F>9@x%LR;Br!&9c6DSq^&Y6*d6kp31W`?lx~
z=Q9HX@@z5iyp{-RF)}7dEX^22ZgaBRfk@{m<;DC-Rp;nN2s2Hk4`h?Q9FI4)D{JWa
z8r9!zwkP1Do{QA}Z2My3E2>AeneXV|^Ha?RkU%b4oR_Ri4aM`X`66R$+udEuHp0Wn
zIYe9{A|gm+BTZbs=byB?j1{q6(J{*)s6HR|G^YIs#&o6H?u~L|<=U*Fz@5z}1UC?L
zOzd`M+2yeeo*$I3pU|a{9&^Sk&~ck!v*%|dZPcZ{FKW=LtWN9k<a#U}c!O)D=uR}*
zgY?xi<I)F(5o-ejIop@m?G%&+a|E;=3~vN&d}?en<#hfdS>UZam@bE)O<uHH%(1Fb
znkFuMTMBrqPY~F(o*XZ!|AY$O%D6#GjoD)3L5R(e*wqZ%6Ok(A6;04DtD4{u-Fud#
zlucoJ9l7vFsfW<;q$x$)B?af^A)a#RWO7U!!?221W0yLZYjItQdJM3JTPac|YQhdj
zuBTN-y}vXVNt93a3h+-)3{OSRHRoEjeb+I|{dLDxb;k>db|E`>I~K+M3BCqb%)vgo
zL%U6;9ul$WU{;8f>Al_aeS3w??utJ-s=GJtm*u!<Fg@uE7|K++&9?)}c@HuoMOIvn
z0?H@5OnR)6|Gm_RdeRC=H(Q39n*(}=C+^WF4lZ5yZ~HT+QT_OiebWBc(tFv*)d%ON
zF`?&&Zgn0<!|6TeuTgWz5$r_^eeKH#V~{K5+GvPldtIyvGT1`_bKfbL`|IF$WZ4#C
zO<er=?MO*-z_k|haY0>`s=A8WS|!?^a4KHD^Sh#jWQn$W9|ehHa`{6dqys<1xq8qi
za;v&9S6ZYnbf=sjrtm)N1FS{yQ#A>A)gZ+>@U~~rg9`my{k*1Ds6#n^>V6cF>J%~W
zmTC>?7Tr-=d9NQk`F}Fqb{OOq7wol2($n+)eEoj!n*VDHraMO!)(2u()3b(bdP$Y)
z7@73H|7H%@SrS8oKJ%{mp-ioYzUmi}9bVbkh7&!wQ0fMed<3CiRH}O1BG8W`u*<%0
z&PaMqwarB8kKpyICdZX*UO@<#{k))>Xg;@TpMAHwWBrGQ@q~!84m+4CpFia~3H?Z6
zcuhZl?$MR`Z~*dAs@floppjlKjynOrFE@fobND|+@E|KcTCyelPg|=Tvkj3-gn<&Q
zl`~uJ2ibx5q%Mg!;!X|cB>C3Wfpu^h7eeoXuhF-4u;<F>m8ABR?ebf@35UBjKn-c(
zJ>La}7%6Fm4!1oTiGigZQu#C$*-TOA-+ft6vXxWLR<!Hd_kZ1{I^Ua)Na5o+BNN?X
zifu1lt*jQic{s6nhNrn=c={mUqzv*ryG?D}hV|WQwo+&{OAMo>H%hk7SJ}@lY1;_;
z*Z(c3JmoFXNeG<wJ0Ob85s~yTaqRQBG!A*x)Y|ER)cBRS<qqzPsw|U>J6#qYzdi57
z$~W2YI2B|y%{y+H;6f)Ud1)35-&!soiz$r-RS}C$S8-6u4)0%Nwf>o3FgV5pTRT0Y
z>iuqE|1eD-TOc$g`#=Zq;DW#O!?T6<_7Dqt$s-BZ#qQ(`F+SeCme{7gP9NriRZbku
zdb2siZa<UARxsYVrR1P*8^uD(;r)CyKSkv{QF%ED)=Xg`)~8QE7a%=mat5m6!9lU2
zyIS0!f)*&BhZIn!0`VCricC4>{NkT&8mjGnKvE$~Z{;DZjYg8X%yJntZU>iI-Ms%D
zOt>!SFxQ6k`Dce3cR7uXCNgs=2PW(wUf@UrYE8bYbQL|`3tSVOw07$1<7_LO&3_n%
zv(A0bvcpl$Pbv=<mJR3HG0AP2XUz8q3Wi-lA&h=+@ety<DM+47HP4xA#Nj-Y$#CGL
zwv8xp{+<}vF+Xe2<QgbD3rO)%B5`nY=M88mGC4sQQBRn-JlPcT5c-WA`JDOGo`m+b
zox`R6#RLm~3mq&*aCPXPF)&J5==1m$OJeT*!~rB)28tJW&Y}m}`SzFn*22SVs&YCg
zyjZpRiBir6m^8!VoH<V1{GuLk{nwUx)c=7aT5cye-}W77Y}ANNd{gYC)35Q-VN*;0
zApZ<40Lc+5oq^;3+Uw>x4KkR%xfA@#DzJjLkv8Ol4+vaG;Iz!e2f>JDQ@<}9zmf9!
zxjwn&M!kDa$Eu`3*s9^^)4ADy(Q1|V)~#F1^t~<A)F?dLG}7jlGX35c_^4m0-g`_0
zsE<!7DUJ=#?rMM8%xTGrjQs?mRG%OR!5$w%@>i&`A#WxhPwnlefQ<TM8A2;#<2Qoj
zg3g~C+ZK0>J5R7WxX;mh60GflKaHG@KhjdV9i}SlvHng^w&Y7U!nWY39JRM2aq_qX
zp@b?DJ@LzMLv_EA-LN)1MwD%|d}?#1OPqA&kK?%hj%<AR`)Z2lAxeMHri~)yFu2~7
z%jLcBjTzT*<)o3~QAukce_Jj^9{O-WH;!6;b3+xf3t8%<;~a$FZ`IUwtv(8-|M)xh
z#l^W~juE3G9VhMlXTBfqy6*P1zCo)ok=>URQKcmaUCT*ZY^uQ&-x~BgwmV;cV0hRL
z;y0;r4wI5Gr6daAIYn7+l<IfP6QEj^PX}n#vTnP-S{-?aA|M<S+09i>K0Q4TM0G_Y
zNvAy~F#RcWAur7zQG9hScaCk744hult~>g^BJqJ2o7?b{gMmD#)0AQADaUP52~BiM
z**q3(;waL4@)wmZY_oBWq^2C6(3J8*K+6B0rvwoeI>mJ~$08&^NZx!zVE#KNcDmD)
z6JrD^x-r=Eb9oQIB^D2Fkl-VKXQ~)9y&&>W6X#ypc=*$67YCa|Tx8yJ2Ko<FeKc==
z-R_ak{$$G_R&2@?BX<_++%TfEGeSjs9<W08fw#GtF-Ka19riGz#elqzEZy=IY7%3F
z5@Z8EmD@neQ6uos(_+y4&;dvKxMb;^aSH;GFttbv?O2(7ovm7>k|o>7ec~~K&`dIW
zN${KJT(7j$`6KmBC?y4BWPc{%3n4XD#g&yTEFNFMeyr*Y2`C1w^Wio_l7neW=`}|e
zar|cv7gMt7W%8T$_9Fog+^pW|o<4f!7Ydh^TMQ(WbSYbzzCBVKhx5^YMVvd`^CGP4
zUC$=0iSVXaK4i2Wopx2(6_=9qq|vYiL$$LU`5h~(s-dsIo{nFhV$KsS3%dgboTQmS
zsr1o`<8P+FNS_f9n~p|%ESi$M@cgjQtmL5>ITfmwL!!gr$&!f00CNe6E>#Rb$R2m&
zc;ATOX=aBeUR;l<1#WLPLVUD9vz^#U2IA;h$&a3Fq=(8SEiW$=@(Qh~V(Xr+g)GFs
ztMwCchE}d!8flcQcQOkIhbxS?1uaNr@a;kBhpLqBYr-OqA!ym~2Sc~O<v1Z83odVz
zkmEb)Xu>1eK-D0cTFk)r)}VxgF2@Rmlq<<uN$2lr%_oM3rS3)EO#H?l@)^Rl3TC0y
zk{_~~AEW6#Tc+}SGs5A}bX<uliAF*d7=n^m;?!fl;hEgH<(b$rg;vqnz1vR3sQ%GP
za`)1OS97VtRu3UhN1r4!RBz)hDoA?Be*Gh0J&i2YM@s`BRFiR8UNmMizLo~Q^Bj3P
z6MUg%w5<*>$CR3*nW0cus!KT!eU}YPI9O1<r*<2p8R~a9LtEw`B3~ofss#H1ghj5d
zD1C&40aW)A^Y8}q#;4l5m4}tmZ!JtE(%$Fj7sT+zwtuNd3!Pvb$Y=-O{J$UD;g0mT
zG;|*7!J9CUh_CN-Eft7Pc#sBW|MBY1^_1ZZ(CN%ddC(ei?=9(OZ8}SE820!548L!W
z=sm^*+Md^?(kf)r4BcuzHZmpU<;*K|DA$S3Zk|B2w27F<%{bo91qRK<rI<Sca?|f!
zdDN#g_hSpaVgHA23kDi6IsEHo|75D~^0!l9qc?7$0Meo+OBb!KJjz)KxntRAY@RD{
z^h`4~hCvCmu$5Tdwc9P07Z|^*y0vLy+xOl|!g8f%J5#{@DuwFa(#w~m&+)NqpLfK_
zp)1;TiQ|LAjvZ=+du(hY(rXM7w^`UsK+*qfG$%vp&C(+u`XFhoy_a~*WBaGJHa7N?
zPdyVF(;~=xAnQtdSGAqOm?UX(<12_*d`Nv-#l@IHuR#HXBHRtH@Dy-tVdw^8c~BTA
z;z7Y^WR5zMBQXH+Cs#e?zkU~+l-1<;Lr_TIdcMCyib2wZ(E%!?L8H&Z`XhC(CKuwl
z#&>{xe`ZM>92}A({jH^a$-?Qzy0a2X!7OL<!2xd$9T@8i8u(R?0)lMJmR4S9THn)O
zE86qjjOH7)*%`@Tz7d8}R?kn}4h{K@hXdc51m-R_*e&!;rYf^et^jmTdD-m8Ob*Va
zKkhb#TN|KM{c!k<H80XQ2aMnSk_B*RzL8&WfYcs$+W1|wx636)QI|?KVYl`Mk`M;+
zlh82<e99qz{rHu&3#*4vR}@otW309VCDyhr74d@%E!NMQncM}C;_1(0#GrA3vjRMt
zmi5kGrwVC}lrOBnSW(ajb0AFx9DePIo)G0<*MgaQe7AEhKOR=>vq;j=M=*p#g@aNO
zwX0b!0;QpOcT!sA>T`J)!A90-Z;_tZ&3lEryYSm;Qo>r&dx_p6wX)`AJt|L&6t{`^
zG8Z6qoa19|TxtZSUA>?r@igMW*`*E-+3<iQX%nX#G~tBz+5b37{eL{2cOX~&|NfPt
z>=oHFMA=)mi0qwNcF4%yNwQaxjJLfDNwPw+ifpn8*(-Z{|DLD&^Znib-uL}J$2qU_
z8qeo-JuY9`QQ4qXca6|{(^r^MlL6x~^zg^E*}9Z(Q-KY1H(cOU*`5yO*<<OvFy;9d
zE9&Na$xG5>i#<?3JZ9LSxq6AKm<xYf0eCz^z<zeY-$M=q!;R;AyLl8a9eT?-6w1#D
zmU?qr@lpDeLA|E1gTYCoEZt6C?r!NL`tX;nfoBrB@qeOW2b?Onvbn#XDZefbKF!@n
zL{tqC4q<MyTh(tRf@a<{_tD&q1?wl=cF2IJlCH?E*vC*f_r-VNnbhBM`gjV}wj2(P
z@iAM0YQ4H;IIK}Uq4UL(yW2)v1gwn)99kiey0zaC#KXvRn(dB9{Hl2}Q{S!C(<J?m
zgRtqm{x)GhIle1F-e9=-`?SPN+db}2w3FnyFQLS4@AQdpVC=ie`+D#8ko7wd!liYs
z;)G45cCdAFxC6E#jhiKpA-q7I?RO(u7%g@%R$Z*KjU||9iPps0C8*q_C$+D}3YO^(
z`>vq!wFzsF<+?*m_RZ&jl{DnKZ&|?_oTWT3HNmB$s5DJv7Ymhh+WSb!S<v94JdALT
zK(?xreWq}WD&bg(y2+AbgGzcXieZ0lf=~+&k2kBZQ4wR>pa};j85jO51wQX@wtpgr
zU4-@<Pvvy!mD)rkPCaWR&-Sp<yZ`1=zs2b@)?u5>hQmmyvQ(Cskz1FkBM5N>x9H)~
zn-kn;A(_byy8~b0;^Jb_U*N#4x_<c~9+7{Y7G-xWef22=m^LP~9ETqrD*yoA?G{Ye
zY;yuM<??<EL`yt^4In!t;vf5?%3xi=rj~xnq95QSl-jkP&kt2A&?bw%0v2Aaz&$Ev
zX1^}6h4cOb$-EU-C5qu5$&N(I7e6!nRdwl>XDg&0|L(vW3p^1v(@a$DzhZiF0c#Rd
zUiDO{NT*oTInmk3rCi+zXk{rhzM~GV#FUhZeLBBOwxovhTp<EvLc!59b2plZAaptG
zl_OO*^pSVX0%MgM07WXe{8dsZEne;k)`D9{iu7ro8CBj~2xlXA#pv2oNWR3tKoQkZ
z<P8+)4qsMkww`-;t<Xogzx823oUznNA*_{mXT@V(RRw;>()he3R$%-!q=03kQ5Zg*
z5A)ZFB65fjWWp@pec!XlnW$_`Izn~V=$6d<HIoyW<^Ya;SdC_-Jy6eF!*=muXWJEQ
zw{eVUzEf@kY~=Y0(E-UuVhfP&Posiu*@z|`>sEG#o@zWU`sq2Dc$S3jSGu++)zr|U
z)&5{ji~xf{loy*V{;+nXpW(3x$`2*B5HMTXn7eXQi9-1N&$P4EakumKfLQg`qD{aG
zx2r$aI?}GtwMr0Zp-DSmdqOC#Nm|x+5PJx)$~01z6vezeK0?b#k=r>kc_Xitvo$}p
z^Q&CWZ$i%~w16N4o~x+wu>zp?bLUo&I+gYP3b09ia@r6}JR+)+j5kRT0Up}v-wV;v
zKa!-$s7GRBg*(*)@x_2kMv7+KzZ**VMhpuhE~Wcd(+_L)wfb+HynLd^>N=<!lyijV
z%Tm9OcZbpY-{p#5eZM|{o(Zg9{bM%Ef!{sdh~6oF{)5^8+vY=IxV}C298$|ui?dBN
z_&Tz<1B``IOo!w&__Ctgo|92yT2-)Ch$>z75UV-GzMJDXA9upOOs&O{xYWx8dg-*{
z`TmDU<p2(g+5L3$rr%IXX$XsT?HNa+YC@6sp(rY$n9wklS4c<(9?83nNN~gZPMUvC
zm~?E{YRACf>z_8&3HcT-b=oX-$NzXY@H7_fV%)f2s}Q)NogsE`*r?(D=1+7ITcZs9
zyhW@;+t9|1f&rz?y_nTTWFq|8c$iV7(0Jh6*e<x119BFhRi^;C<M|h3GR!f*{@TW*
zb_YWbV(B*h<;u;vO#8P6{h&>9O_6^G1FVG?^vl(|nbH=TsnG?pIviN@?k4s!trdng
zMt5}yvj|zJqWjs$Ecx&W3F+(IAxtkX$Ntg&3gVfuYy3ps$2pyo4V-^w>lV&L91}pG
zynL_s4o<2WbP@k#1o;KD#j!^AGB?QXPIAsIZh0+*&6>`jdNT;_(_&(L2Uk-SR}O_P
z#T9P=_1zWB=8_oJh*j7(C_jQGftc^U17dgVe)WYIC90mW5Tqx1N5DmKCgQ;T-<s8_
zh_KG~<82MVm#KYCX$r7ywGWsZGwesoE%`)wd1d7ATwbMgE&OIAmk7~2X{f8L6g+ka
zdvlNnLJf2DO)6{Gnn$7kDK%6D8{WSk3GA@U9!R}kN({p|rW|(mupi5_`}T=+wnM&N
zpk(qDhjkDoULv@(R!E1{G)7xC_H@S1`Vy_nx@XngtWr`^z9DZ_3*v~nVw`p&t@f;B
z`yrhmqc~@dx8GFX5-KRDO~25UbZB@MH?lZ|dH)KnQ$5dYp$&nSYcbz<YJ>_!g>U(F
z#L+<K?kizr^IuGg6H@)sQx^O7UMQqPW@O+D>MBSb5f%DRCyXRTw7o!dSA2aW9Eoq)
zUz`<euv5)jS+xV><jW^}HdZ!NiOYFDR>7cVC+Ef>ZjTof*c6ynflbSNc4WPfsh|KW
z^pDqF)~dxlCb;n`DXRH`9FG=L9W_<cy5}kfUEq?RbB&-{V2QgIy2J)l{@JUv8;=Mv
zGG)w7S^<QxqRjaC&r?E8_rk1&mE}&!W$M^=ISSU|I{EvQQ@jXwBl@_)2G^u(x!$op
zP)Qfd+XDOsIyo}#ZSu{pkduSXD}{yrsv<E@2LmuU0sC*#agpDZuo|_3{E@LEbB`ZA
zlCghBrS08fE&+%G%$E?-)TJp9f~7A={*FRhMcwsrWj<8H+w_A?g1{f4aEmQ}H1P{(
zqUyWKqa(wT3d6d@*bZhr?tBuuU2%-^!Sb4l5T+X@D<T;^+R1k<#zpPk6s<;~E?kfn
zo5f89TFFqWY(KETb!b<FT=(28(YyE(G7~D9sSjAz6BUmumI|u3LTi(X&A(HdV+~~f
znjpgOMBMhuHy($DqfZ2Gxe@eJj?D<8s^kCt4eIr#6zw$lW|jP;1*itm<5%?U9WC;%
zWKVc>K=qnPaoE%b`nZX&d_ryWgZZ<+W;H;(w=7m{*NJrvC@RNrnn6abv4(1g^^u<B
zJ9j;CewRWP^W_^H<C+bLeWP0KgKK4FW!m>S2@Zv7aWoR{QyA8=$3a-+J(iDff;dm2
zV=6YwVq;7AajtTNg&`tl_pxx@(8OQIy*^HWDSm-$5<ZX8bW`YL$;lIZ6ZSL__y3&J
z0tx`dp{#c$g(md9f~YA?dZkd1EoIln$vp94rh0=|qi|RsBW*hr3LkgxI8Dtaw5}Ql
zY_oxs1kc9~#52j`Bngkx3hn8l748yPJ+sZNg{X`9teHy<a}tGpeGBX<=5B7v-qG|r
zVUr*SXbgh?#()#65F9nRA%ZSQ!(*SIKvN*Pw&u0rHN5R#I;`44(2|XLagLoR<KU-|
zx1zj1y}H-YFvz>-AkUT)P2jb{tw8XR$`G{In6FByIvfbxFIhFl;$g%D%VtTd>hTmV
zyJLz+`J6fSU0hx3>{ja+ety02UO8%sR&L_bV2_nCm<H=}BwrCgQBA|7(Dj@9s(9ar
z3E;O}Q)#m;Xdc7Ib*D{SCG~W4b0ag`rX>*E@>B?xTWK0l6a*U5$YfFBb#oo*V4R-~
zqOc;U2G4NY{<f>|HVZssTi+*jyP1ZX^cdQdBg0Xso4nmG>}f2iCl1lGElDJUIEzza
z@9YMTDb1CHWRnaYII`q2eyK)vJS@cX@GRc{$<3keNs8NN>sHo1uQv8z<t^0_WDs@N
z9G;y6mKBO%j^2d!|09<q<#oG(47}<RMlW4$dIak`pk%wx#Bu!^rq7SSa~>SSMHM)O
zji|Z72U$$_I*O6im;(ImMw`nwml-_l^8yHlOE#favhVr1-s021Bexf8m6BWMOZrO`
zTdh-?a7HB-+a8<ldMV4ek$#{rLPLHz(<NlzhQ>A-Jd6VU(qpYQTA=s&j2o!VoGd_y
zfA{gJN^6{B&pe#330gx&-(en^FM^djg7!wZTbEhh$7N@M1QD`eL;X65ok->WoIh9n
zxuA#q@s_)a)9vk7FV0<dTe{TT@|+f3mZs0vf-hpwToHUQsAYxVqG<bD5e9Mmg&4!S
z4e4cSrZ;m#O<djPtMs+fuC2$i!b;dHLXniKAahz3x|02<uW7+nxDk!^oegIT*wYV8
zRBw#I$H&+76uvng5jO@K#S5RRt{<G7VW;Q^$_r!@@iT^7C)p=<o@}y7ZC%C8(7Me!
z0DoI-wiYjlYNOj?ZBHE_Xz|A@Vw0iSlG|-RwiAUVfz38lyT>KzcTxkDh~ATaQT9zY
zta298kgkRGMXtGB)7_Zak|Ml%z1~x;cHj}M=O@M#f{nn@)%TfNva`rU_gpWUiI@<B
zie+EZ)Y#9|{c=1f^amxU@pLUe<ox0ivprTInpXeb@HX<T2gJ-&>z&CCJNlT7CJ;ae
zYh~TJ1WR7vmX_1z-kSb;;K@z_RNc5F55z3^k_!XH|28&uYj6hgz%E_v*zHP0LKemJ
zFAm5btirxNxHmj)?hp*}%)=Hf4bu}}Vy@vZ1Mgfxw}P$P6bf4AIUNh}qMj(O(t1<M
z&|`XYMt@nhkhMtZnn`me1O7D+Pbi<!-k!wwIomh!J<B+9@6C&WBx1BRcjd5qHku32
zy{Y3CP@W>g>FViOo#Vm6I0Fko3Fo?TiAU^;7Lz7eocn%0En)Gj-o+=5tohzhP<^;k
zik=={W>}UkpY=yUaf}jU{bgRp^oa*=z)4nPvyF+((nRinUa-C73TEXtwnFIdl@+^_
zcY~?IPW$Gm99D`M8l-m$_nP-Y+Y<bIo(z1cC{px|_Px_=-eZkZ@t{GxU|)+XX*|IN
z+=uy7-sKcVbkJM{e}(3k$mz`<>k536wM;X12QgGC7UrT0;3Q0;8CfPOCU#@$`BdbT
zi$+m%FqiwC$<E2WB2MNQ_Lb=Yh+=VU?7bo9N6N6L!-vt!S$=2%aQ2nwOq$TVF2p%A
zKK$%SfNEeg=cNGomr`gKnK=aqmB(J=rXsT>?}QN|`_cMAvRe~rRQSpH<g=}T<xe(i
z@ay^ga{uaBHE>TGH}cNP^~&u~9cZ(dr$RjE0MEn6kKqO`h&3cen)Tby`i^mBZ=p-a
zzxeN0aPf9j8rrb2p-hrVWw`kS1v!Kq@_CXi4L?=8y6GapSfQ(+QiUnr%_@ElW_@Kd
zpRD>n_$cDRYXj5KY(M_AbIx6lrjL$MbLWh{S0-0x&7;#2IZ$UiGc_K>?XD84af_T3
zdk8~6zu+aRlq&l=rZMIfjE8LH0TDgI%m5wYy%1W9FLt=G1G{RO_a^L{0%^TNm*rji
zF)+}(O7b!~ojT_#tj&AN+*|&Hi88tz51vbUa!9%RR!d^G;zzzXYN0@%Ic<km2}ia*
z-wQu^Np;K&CcpJ24B{dhsxCS=J%YW3<*Fsm@o?(rm$`kLW}N>P<v)Z-={OPIt*z+W
zWO8zS<>ldW4yP)OpoNd#&!xr*e&p(Bd_11NpVYHB&a5@tcs>QH4AgzvMN?{MQl7s*
ztO(pD&G1_Aq}P)G_0}>tGuD-apKmN?UJmY}UM0g;0bhYfjRMZ)oRy_9simWV0k<f1
z>gbUDcYU1pPnbkaX2i@^-lqR5$%}!oA$GEC<_N>FfKsc^6q?s%(ZAY%!sx(WN&1|U
z&!(eKd6Nm$3^--MS4ALag!{#~dwK)o`Xm0mG93x7_0W+6HlizdD<n!NbzU32Hr3HU
zsBiLKP!!j`Am2;^at&QgK{{9T-aEgI^)B=B-|0o5Lv6raYPpxEwcb!I&2|Uhb~*IL
zOC0%J_ob0*XljDc)|~Yx8kdesEcwDGRJnUE$M*1Ft|d~adPBxVmNFhazD^&TcT*+*
zeb)jlb?S9`G!}kwy{VCSi5xl;?GE-8qL9fA$1*o&bFb>7=k08ash#sA;V-JkuDWeq
z>G9J0QicY6%~oU&`ls2<uIhLOI;nk$5s~7AIT?;E(LjH`-Dd7|bacT)&3dQ3oiJob
zY}HvM5;KZG2q*qm7CHirN^|R%Q(Lt3O*ALo72CDjO-VuK6-5Pwn2&=lu+>s(L*gU3
zAN{I<?u|>;t+P~I91I@Y!HOp)p^w)*GnW2-Cjr2T&f0~Tqq>mVe801CWW5Lq?SgjM
zoK|!@9QDak!w5Z0G3&(>W!fH#m{W=S9~+b6+X~+@NiLc0IjqSzqR9w4p`xOKGo7)~
z5~ba09Z)g{yr&v+{z%Ja^+K42$H&}b?IIlyT28e!;4<cY6qTqK<@@+VPc2lYan@NN
z&}2HnaUat2w2NLp)q~vNC@>pI)qVpUUL)jgZ6_Goy48RF?2$$h5TTID|Lqp}Ue=IO
zuTymF{m)$w=--Itm-^+pV}73AESQS=o^E@I_AgU7f8m2!pn^j-g<TSaW_6r~wE+}_
z@`}LXUjmPx)7R>_#=M|vkku6Oc8+`agD6rwU2qopIdq+9MGDZrMR$rx>_nF;^7-ul
zn#%Tl_@1c#1^574WzkzoBNwQg)YR0>Z>CLUEi|w#IY6Y+GCA+E!o*9C7lsv23x~PV
z<2vVKdTs?@7<1?2h^F53dfwG7G+dUYSw1lOg)sygD7Vh!PG8Vvv-J)fe~gS#>6qUJ
zQ#>}k--p$6vu`$%JIPhk0jb>~=H-QriH<SPYcm}^cY1;DovJw2+uQzw>R9;fMF3a2
zbkE>7W#Q0;Js(ISrziD%3M1fdY)~y<y@A*H5^}yBx`0cK`Ak7WUk~pD;ci+heWDqC
zAB>|i=|DS?wBk{B^t@``8JT((Kf=^5C@|!JXt2Jm=|&rqx901B=xontdWh@gNMoX;
zXKAqYCu;KWU!mt;PsYSIzqp3^x3ICY;<O)~oTg_j(8Yv6=HKFgM}=X-4|#4l$IuEh
z0$6^Ra*O*bFDtyO01kS}lNIM83*O5<AVV}w(`Fr$lZ&y&?%U)}>NIx5cosy`46q$@
zB*l=u&)UIZVsocn{qiqGjx??(9iv51+Kb!&AiA}|S=?8IH3!?Km$2_1gCSn}H8b34
z)=DX;uRtX)k1pgF>u=cJfSOS~qB6O~(e@A29C0$Dv1KsRph!Ub6K9ajZc7HOqu(S1
zl&GF(>B}ts%nad}Puz-D!IN#ytt(3_qxQx&v3fP`JTlo9Bl9x!dEVm@5i|}$bvgh3
zCvr$glKy?0V@31L{Q;^KnZBVe>k3htnEi<8&^G|582(mMgGhJU3Jr0~gI`T=4tGo7
zfRygmcHtSBf%shJsE&2rqpK$Mejtp|ogdagG_vMgaQ+(;y&dzP*Lc12Ax~6H%u10v
z)h)YS88iS84lil(B&&FyuF%C4?(9ZCV<BMg+j4Emi?KJVy9bz1wFFy(J9K6S2ATbK
zc9Sv~BHe}8uKlpOy?KAgHjV5=9U^~dC!~hN0f%s|4CWp~QzEuKfBlL$$&6n@?vDc*
zbLYvf`+tP$Ky=yrVMl{cU$}=5<4^TMo@|@pc)50g+oEtJ*ouA9QR;LBHgpoyxPu$P
z-F@Dbt4i*#e-!qH2j@FoVsUv`u)Mt7eB1pezx!E0cf9S?_V!1EAukIHIh_j_4?m2p
z{szD)`wr_^I7GEvBP6{pE<-_eY}8l1o3mr1GpBnSaok5-YC)nh4eqz7j_SxhDV>qc
zDEcnaTk!M7uMZy0uaF`F6V+80PBV5gJK<YUot_w;khM@L@5iU5Rx7AzzC*C5V`bc)
z15}d8R^SxqZC|^8Msmyv?y@bXEh1Fo?$4>o7R7bRrHNA9B{3w$%V!Qghzm+IBaquA
zc0M3=rrjsn+dVadBEp@;^E<%Dc=mZ8Uj*_MItIhZ&E5AjIj&dkk91d(u2UNzr+aXD
z{NcVc*<cY8`A)(uIiF;zdSPgciVLFqh9Z}D60KN4{d9zIONJ=D)_geW-Q{bNiZUl8
zzi6;W#l~)?6<2UlWz(6v4xIy<AO6|$GIlE_eBo1(tdhUVc}Hu2(C722X%uV1Rb1u0
zLCt>Qi?2jJBw^lm_X9DG$rmq8O9jim289(1UNu-nT`&nCcIqQv5Z}nXJUY@m+2GNd
zef(P>Lx7XG4Vb8o%6c@4T4*^_hSY+f2wi|m?OSpMAI7nzukcnD#Tfp5kiF~XVO>L7
zudOFy*TTgt#GU5Hz3m$jyV>WkyBM}#e)NnawVJy?^iirIsD2-Rxs^ZiDMYpmH2H-y
zszB?3*a3FtRQF!OWlf)MU{Hg2-pw*!tH^PZN~uI@iJaD>F8xl2@lCf$sGj8X9Tutk
zN1$ywd(I|$wi-Xnkznu3*m#ECr$zU3pOF+O{9)Yx(%b&B8Y7c>wm}DK;gWwtqF49o
z9>+<CWNH3F>}kYeV3&|5QDlC6g{nR>wZOKz@XTnY@C8%2oYk*yDIz6*hJAJDj#aGE
z!hbm$z@kj(XF7stx{R+2-U6IfWT7-leZ!1nO_2ZD%QRZ9g?s7)-G`s_-c?GGG13Iw
zOm%<Vc~Cqya~5p44B0;Vdg*_UQhq!L0$mO_nRFRJEuftaKWU64vxjhvN%_kqy-^Q5
z9ymoCDJUrHqpt-#en#6;MdSHJ%(O88rOo`xR=v5&K?bAuHB=!o%4+oVS)RuDEOvDJ
zNCjh&0AC;EoF|o^r%G<w#wo}ZfjF$lwdcGd@RU$R%KgA5u>WtX+38V>C!K!uV_(`?
zk87_~Sgnqxo_8ow*lVuQn+814%?Ut?On0{<B~@#zr`ap1-dI<$-I=TLKfkF<dqNIX
zmWYrS7lXc<K0rtxsQV$|;a3B^gi+iJ_h=1QerYYf39Q&$`r2*D7ZB*Gk53Yb0&M<M
zIHVL`mXrq^6Kg{3$ypLal_?&ui+A;LK9}^qLnlcMI5$k8r>iLOe5WN*aefiRCo=GZ
z`{J4f`dyA2B@I)<bdKbu{27dA`MqBi*0w0{Y6*`M9rT|oTm*`m&NF9`_o&JI8?tg5
zNuA&}tX*~AfwqPQzP&GnfF4C;E6EcslP`@9_Kz$`P@+<CKbY75=rhH~)mK5U8)|R=
z99}8=jBxrCEyo+6rGA5ABRG3OKTaL3;%SoV4)q{l-*A!N+b96^(Tc}sf<o(<0E@W*
z-ErStz}l(7SEK#Z4<U_+O{9tY-0sanG7pEhPYUkMu#3TBnbNd;I(%Wjd{wKx@wBt?
zrxNjK*k;b(Jw6Oue0-;tuJl3Tf(<Xgx#_$sAPFxM4y7n5bFK`mi&oH}h%Aj2V9kvo
z-zz#?sjfmFqE8#&VO7<K!IpY<ioS?V3kX{Ao|(0|02D5^*qr?*821(ncFbQ5nE3G1
z1pW2I5PlhEoO<YfZ?)&XoR6vDp;Fr}%I=+^n?X5s$P^S0EU)#1j-PiXfKkD;%s~Bi
z&4_ed?{dZJq}@r5{sr1>=5)e9NNJWG<_W-Z(_K7Ds^=5Aq(rlXQyX^P=5*RN={c>c
z*eypWL3ZVm2s0CP`ORzj_!pmH-P~l_I^njdh)FMvs@UsVKbfV~2cR3xhtM;&v9(Ql
zNuFG)Sr1|4(Yuu4^fk#BYZbAS4zsxJJq4ee?(~MjMoGnb`2E4=Wwc6kx|Yx_@E2=u
zei<3~p$cB_9q`4vP*W*E_;!pxAU7m+$=cf54rd5@Gp&IV6zFOOa;xsybPu@q?CK{R
zMAaibAuExs*pMhju$$zRGmTxJRq;Ibcs=goQaq-W>Hd%3L@X}9m!QzJWd3{SXm-ns
zlaz_G{9KB_7o+#q%5kKTq)~LDIOumrWj2)ssMx?xH2eAH)0QDpw<sSC>=fII*q*@i
z9gub{wF>EiIhylL_>CZGY4!UWWhEt*o6=W%ynM<f-2e>)-2qZ3859lG?ut8A%Ew0#
zgEJmEpY4<pxB?0n$3;hiu@7rs=sN&VXh91-3Y1f+a?je`Q0L$GQ%fK-PV(Z)fkyxv
z!x`%mj>%V#v(HV-ZU|z4AjK7sEOK!zLCo2b_$F7EG>E;4yfp)5C+@lL-aWa1;R7ID
z?S%1_7s58bwPU~N)T)mvN*R;k{Q5uU+&z}(Z5s3Jq)+x;w*UM{PNt-qP$=``_FoRk
znf?j&Xtxy<#dr;M<8;OkOG~pVE4E_dJ?R!zUCYM{=CBwP*R7OgML~;kJ+&Y(`uyE-
z4d2qHdWI~d@Y??2-pmG8pU|x)dnqu|DC=*VBU~M5S(N|52LlHYwO5~ayG`Wt;nn$2
zLMBkIvsX61e>drv0r8*Hcle?mxZie4uz)qvPl4`?YN?~NiWNE-I2#!&L2{Nx8*iH!
z@6AHy!H=JSm`S@5pS)B@<M#9>hBUfQBv%4E=+Z@s$2!ivdt{)6FA{G2xGnG7Ng`V!
z7CnFXBu03J!QDQ)>)g~QeAAKW8-MzsIQK?I)Ci@(RKBKgbp4TT%x90HTOJ!NG_^iQ
z@RFvI{SVhe&k4{q>S>232dTQT5ufx=UfyJ+i%sgZ3MiP98A424RkQwQ|C}l{`>`J6
z?<JOB0*6^{I#{8x+I0iT5tH(B-hJ+7nySJYP~t!8AQao)+~~~?K{Mo$;f-=qn`vX5
zU>j=Lf4=*`^OZ)<^o6r~tO<2$cg1Hl3F^d*)AJJW2xLgG2@v1Z<%a<Qru#;Pmv|%J
zDsUT1fPXkgQ${jForR=>IH^aN6!>@sCjJ5_$$CgeE<cnOxBV+El^fdPdgy7mWT?*k
z=>5V}PPkCU=^wTuv%2K8v|CA?0$z?~P8=^vO$mCuY*|j5JGt4o|KXh0Gi3zN;+Vu-
z)6<#tZj!h7cfpRN{n0|0v>=3g<{0i|WCB!*`^@T%HCMmc<5|M<8|ad%sPkV=y*)`~
zUu1eT^Xsp;V=YyPI07VjlKM@}>TYN@j0~K!$^Il$4Z3J>moJ6x#(5E-)go;TVkEPA
zphm$~6{5f{5HJ%;7!rU|t0&kI8Dv0c98Pp`B7A4(or706`EOz(F=7yoX_U7;?ZQfM
zc5$h_H1$*8)iy#oec*xbWO&t`@!z7gcK$A@L}NF$q}<knsgX%QT6r$)IPse-?9UR3
zxYYC)siVVPp9|9T5v@%4358P9w?~5D^|^`pJ7B+)<x5^WmZx43@5Aly^r@dRL+_a_
zW^6z{t*hq#Jq5St3g!YU-Sbb6>mC)m0`u+sqMMA<w8@P-f4^d$R!wam^#rEVkIM$S
z9(U$oX1cw8c_ZP6Tlphoqx9ouFX7Qlj!TVk-A3^jQi<sFpN+qgfthq1e0Y}BIq#b&
z`Z@?vz><sKx55pPJ?R<Du?bdN^HF4ad0bzy>Q*4G7PovSSZ-fw_CH()Sqqa0f;^`r
zq9wXNLo>_z1mFvqjHLLrwhYSa{LwT_w*^x2rk)=PpQZJ(C#ApCjrcJ*NOs^J^V6<U
z4Al+$^;~mq>SMD2FQ+=LPGt*@<0zbv=ZD|HYwqtSz>~lF`jCCuWNA@AEx+JVUp8u~
zM|NpbE-=q`vpxy^3#d&cjrJ+$-*!LErR}e@e5EF*%3V|W;Ndy-Y<zvgd4U$+>#^nF
z4?KqBV2!d@P}Q{BmuCib7y>nJjfCFL>}0RD4B^MowNmYFf|SG;1bssx=oTDLmFYN4
zE9dW~N@;+u48VFO3-KBrXh=r~QC-Gceia)UBol5oQtT5g;4?T^@W$8o!^6Yt$#wJI
z{V1hfK)qI?U#c}b3^DsHwTq-bV!c9~k%UgTs4pg)AH46Rn{4uphGUrvNRAa?AUyoV
zB8OjDKT~Zviifn#cK|8bIzJVZj_vbd$5dW6R9p&ABGXP4k)8#^=qgo8Yc4q;sAVI!
z9%$!lyNJ@$VU~DO18!sAQFn!-a}ShlMH0$FZ99b9F9@C2TXG`v5Ey<-82Ac@&Du+S
zYeA3{4czC=a&D@4&EO9^bg%6ysjdh(%|pbIzO{<Lb+EJQs6mokTi$d``lbFIuRDG}
zIUp3v&$m|_`XIKp_f9L9=~1~(-x>Kta<z)0qN4tCKfFk=rux3-I;flx^FgH^)jyRw
zrhwX9L-*8%NS`%#9f3$uvx-SID$=o+AubT#2ZNHvf5dy1#!9hRJX>2=lwk}THN2}b
zZ(!ipP8ps?{kv@jr3j4Uen2FcY(C4ec{X9|(739l?)GTTM!E-5z*q3&)EvK%5hOJc
z7mj8hL_h>78bdtHH)e^W6-1hlDj#Y0!cPF?pyjJwbCBol;PAdVT&OO))bJ!y_=e8|
zAGgI1y>^umW-m+u7D2zXJJaKIpI$3t@CuxSJ-q5}pvDBAbu8|by!fsHdrv;i?0K;F
zY9!BXiRc>z(P7o}d4E$ZeBNrle5U~8%<VNE!B<*MBkmO6?!fQr4(tu+2uCm)Q4a>A
zQc@rl668NmpFRZz5WPss*5qd*S}H0M7%`n}1BXQZ%TqFN3p%*^;gnBK4PyI@Vxyzi
zja9w%(M$Aa%vIy4A9u1<;%)E1D3P2#8N{BdNyV1HBgsj4LHKk1*kk*v3}Y~6y>^M7
z(bY6jwSXmApte%=1=hR*TOGDM?VL$t`GMnPXr+B@t<ewWk*kz6lbHkZ@A2)*;S5iR
z{$vJ}zr#Z>w$21N+NraW1S%yn^#vdqd-Pt50yI8&mby?}n>5E4-ms(O6X_L?R>(>G
z+C&2u&1bO>O7u3|Zz4mX`L=lD&kW#mwP%jUH3e~H%kZK~>Pk7Y1;lD^OJ$!yH1XDv
zbs=>cgwPk?-o5@E5sk+xUKaD<UDWP`|08|-B(-ymo15De<iX}o7UY?lAWseh%=Vu5
zh`|{)=5A2g&W@l_uDemgec(c7LdKnL;76PSay5TO!{)91h%Di=*ROv3DIJhGk@CeS
zxI{^_bhF%*bUz!6Aq>UtVq^IQ1xJ)d9$c^G<YWTyYVqMepsgCOU-jomNEm;`tR-I`
z;0am9KHoZ`_$4s3#*Xe@&{V~~j4z9Y?W4SwSPdz6Fv2be%M|xP(EyneGA$s-{lo|d
z2?_piG5YxMQHX<3CoB)hYk|61<8B`J*33|vR@H5OCQYzwDN8(>cW^tt`Mn+78cl@{
z`I4QJ`y?%$GYbm+3-lyGIf6WZK(`@$a7STyBK%U+?aRSUAM8VGG}!<0lL6;U3eO}2
zk6`!0UIbFYJYPhg{$5-AoEBzW4#6>dF8~fPT)d|phiBsMVd6Xv*Ws6LMxL>9!N6cg
zs4OFl)1Jj0&1l%9A}OF4wCrR!bsldI9%Ojg+J<F(x;!4ddj%dcwz>Ov@GB->C8m^{
zxTkxI{BHe{A_HY;bCW9S8$4y1VB}W;&XL0^328Owa!*0jQ7l)C1pUCH!9=Ph16m&p
zOHThf!kP5Y+Xzs8Qds|P-mu&tm%x(2Su+`^-{7h*;sRT)PISN}K4?@;bl(IRq$S^v
z9*ie)@a*j@EP_+KGI|PP@mrbWul$#d12-=Ad(WpPHBXv$;f$MI+_NOxtck9y{uw+Q
zLjWU4S~6|9TXD$uvy;8=IOjRLk}H!GXCizr6_}|u&lZq3#QNrZsNF*c*0kJO203s#
zqH+j!x=*sc1D{aJg3#^4d;!OABV?Yy_EFsp#l8HF?-~=JF4<s*po7vXlGn|l4-x>5
zo$yl}^2iW+EJ_CW)i)}vMd8EkS|Xsp-S=-q-s2}%GiN@)C;xDuf_R3;4^|oOuD^Sc
z#|IXx!5|h0k%2p%LgpFw7KnaBGlZHnk~17yOLKkyEnRjE=d#@_I6G}*a&LGjK(`bE
z##g~6@_~EhIyD~FhQ&?bT?Di>!1G0a`^Qg^Ie<HnfS~$9LQ1^aL?t4~?z)){{tyrN
z?I7hmAXit1Xjjj<Y4lRT=o<zPw}Gz|V70j9)ew2U4ElEj5H=>lp<GVz=G(t5CY$);
z=G!i7_;Rq3PixD40qQIrEMbH3U1Eg_5G87-h$bXKkhBY`L8a^l#;Rg)^YriEzt`a8
zK8e<0A_T-(wH`c=ld&2yUhp#uq|Mn(gpaQ;m}!GX@s!66u(iF~umKzljcV^<Phike
z<T@Z40^FD8`Vv}Z^}nxMf{suM^G^`GW*IPI>zS_57PZ{g(FuVtflBz)M~sA0Co7xp
zz^Bu*B%P)4+w5Ot_Wyo-*)63u9z1!-eS=p-S`eZgbi|JXduvgie_qAM69-(+eUAJz
zkaDXSH063IiAMon$q~qrILiS~WDQBl8CDJo_)h5Quqdf#P?@=kyhjqF@PdH?<U#*a
zT5X@4{rgpcq|R3S)aO4ezxk;FO%d}fzGty0>)^-_Q52EIw?l$bwK5`#BLs5j6iiSx
zW_b^Jt`qb<qNAjonVv@4!n{i)|0W22C5azU2>N%8J`p8BCKv_h^q(S2#Q=InEp!<!
z)q^n#@7<r2Ahake$wiQ+{35~jAP~L6A$-|B6aw2LGF*tn8zls<<H1b&?`eb5+UFI5
zqQ4V1*!Oe#XP53f-DCP*F8It#QojBD%OizH0u$h|HGG6r#}I;8lRp#IqRl|`l@*#G
zz2SX*nj1Pd4S-*D`0RNTL*t8!2+YqB`D8fa?`haPeX`zJ%o6scFFjbUG6#(v4IK7~
z-2eZ9Uqv1Y*shC88ohT|Z4?D!all(u_!9<!F=*GH)q6usyyi`f&?9zH3ix)oKUpCm
z_QC%ypd7gR8;WFF(ICbhih*YVDZwy89s9|ULN+A>8$7oR<vbF>151GJyRwWa&^mII
z7(Av2<X+(5qp6y|Je2hGi#9YJ@OL)$k~=W=14&d`9hA_ZqvqyV;)L4$dNQo_Godnu
z@NY80?MF^e{lK3p5Uz)8H6m7aj^F}G5Xb${2~fNfD6Q*MRK+?Rr%`gFM<0>E_Wy3x
z&fgzZ`x^YlRwVK(24;b8?Vw;>2w!8SXx&uceP`TK0(f(0RsDu>W^*T6;b#T+Kvn{o
zW*BSHVr@`{sME%QpyLb|gE}rN4R*4_M<RU4s918y<FbNmyyQ#qVtmPp|NlE2(<;)r
zMUE?M{V%&2FHV)OM#=wb#MV}qdVm(H>4+ZPKJ8L-com}_;(f>|D2gUwT7mgno7Ipc
zvULG8#V_Df!G$^T1NXb34*#1W@S+0_uh<^8a~D9y*3tj2-@<pXi2+qH4k0KCd`Dh&
z87gSPF+N`2rrXKSM;#<*-87c?kE@qtrT8FR_({4*Ruq~vu#lcwTe8VK%i<wF)%@^<
zr3k|k);Djn&`*EK^Z!eTYK1$A1R?T=G7x)iDBOIjlJD5{di2Pjqd{ez>EG8deV_Jz
zhEsl@NYPJ71%e0rZ&5}@fdEUg%D$}_`>vu;-SX2XW9z9{&FxB*IPctlS?`F^3;;x6
zme9QuLeXLe<j6pNe#9v!m#Vh7uy(7F?<+A7?Bdx08akF7P8!*Y1Z(PmAZ|Pq6GNH}
zIXciT0KGb(Ej}7HK9P=1=4}8Hg%t;y%o|pc=uAGse=CH417)#Vjd6Z)U#zqo%*(<S
zIMomDF{VE0t{zQ)9xjGT0Hpssr2%`|;l$!$!dVNOT2!?Esm7{iPN(3kvkGU0sI@CU
z=Ot&iBR^9Xm(zeo@?S0C@=V!x+!9Ol-We;yx8Xm!?y+(I-!ae8xyJak)~NPrAo&$!
z#V9E))gt5v!SU0=8;_xMs%0{Ef3`nXn$yFSBkYG|2h6s3y}7cFTpiZtl!WzG*Z$=4
zQ_*Urdlh@lCz67{${}WDaUCL{9jp8;Tkd8wdOe7=8kn2$yd?4oN^Fh=+ed!ElPC@v
z0&QL-$6$EX7}&?p{8iP}UAZ4LJ>+q#YBny?`DxAH=!0wB;t9y=UVB9Tk{tB)?hwQM
zwtFlMkmqP(at&ERS$lQcE%{ui^LtyDGke_=I~M-@$?yAKlREQAoyS%)-N_^zNd6vl
z?bhQhF_zBq;8eNga0f_k=-)4t>02R7?k-c2PDCWQubtUKWFHL^?7+EGD#!QLkKE>&
zObX%DE`zq}I+&*=ku1-lTjcZ_(IvT&&<T9UbXSx6nArAtwQM00X40G~Vd}JV&Khq3
zX{(6qMrGAehvOxX1RyqY%Y#7Cb(ZjJmLOHMJX4=?mcXLr{!UCWD2W=U)i+mnj*y&T
zX;|TCk@KJF$LFwohRTp2%V1c;#>z@gaZsW$vg}mCeHg#{rT*wI?8B=}aj=3JlU26)
zr7c+5L_IBp*g#LdEwE|=u7d(stuJM*{S&z~zuDn9!5=b^5(RQqpZYH@p#3m=Np#$f
z#PcG|VN94He1G5!K8i8NRsXjT#KCIQbMqv9_T9T!t=k~ELw1JOKcEgqaX=*wL*SLN
zm8;w90bw;Ze}(bxa!GjsF#!RxJWpY&a_wt0x|N2}?Q<o$%C1^JON^>~;@cmXnmSGk
z%^#c$EDL-_fk_NAi-6z#S#{ET9^kw8V0a_Tk956ixy8#b;#%Ot#tf<|#JSvhMzM>5
z_jX|UPyYQT$ZcuPyy3Pn;?Ok_-K3EWK&kYHXHl|EwiVt+t%tY{+Lslp%FBtv<qfaC
zv8H1q!*K_7jMtrp>gtj1Nyz8beK&My0&B6D5L~m~G9)VtD(2t*yZBlB56eO{?0Kl8
z6N<N9BY6xjnje*_<4Rxz%Ue)8Rw3Pjv$Tv0DKZ)128V5y9mV@s=<h*dndT$*G&L^9
zubO{$jx^W@J3a{<`AOP^L7SUEjME0PmrDgYTJW<dngV8En2ETlW^M3#zP@J7OZ`9%
zUT~lQ6}}5@pwOyDbnYj5aHE2D?3$_Qh0tCL9v<*$o0mJ+<?#p!we?_h-D8EU@kWt8
z<7cdIRZ<ddhB3h8ee;!wh-Ib`d{|8Jw%i7CFb6LO79G&Q5pBjf_inB`GBcw4<z66#
zGRGYH($dm5P$P7Q&n~#auw>%C9x~=aWrW~nrr(25HC6!9$tDFL>c@pc)V-#L)o6)#
z34)(F$s?geh*B%iy2<NEEIi!zSK*XfLLT-?Hd8KHq0p3Z{6q#{Zho*GzCQyb(`uSm
zcI$z5y-4scB&FOX!9MAh^MrG?h8%^~&yHmp!Iy(N=KZi#&`cubo>$i(+kD0MaU61p
zP@i#r>r?RbiLx%oB%9a}#3R2e521T0&JHy}GJt%M1-4*9<Q**cb)_c|4CaX<SlRPA
zqRga(D4o0JLTnWe61E5X?}a6cU}2IAy-d#(;z{ZZ-~Dt?mvb{N`*VB%u+o{*rO%dO
z`_~928)t#RtJ9UuYp_I7GB6?=fI?4Y6Fpce3?=fY6dntKVnfGzo?Pe2F3+0d@MgwY
z02=rSDW145WX&g-ex2l^`x~2QK6tScUAYB7-OW$E3A)qw#vOzl%gK;q%%k3@@Nnzn
zwugm4hnV-yD{>aeHxgxCS#mCiF^OSm1N6Y8_tH{-LS>L~k<n^oRQ9r&Tgio6_x}nT
zY|nf7rB!Qg<0fZh5Ml*Gr5Uz^UAki8tZ#O7N^s)uu`DIPgg#-6EPS`xrm5yUAo-lk
zQF!m#XOv+|gmUXc$R`#p`L!Tn2A9P6n3|rx0JBY>t$gSM59FeK7{)4$W)Hk&1*7^6
zp%NA^{C)nZ@$veS(%*6rhHkscvp5oz6LzY>g(D+wyGzjZsH1n)U0w5}^>1=v+hV#C
z;=Qy!wl+6c@iu+T|8dlsAWk{5qzEo1?E!((K0+|w-ln9?+C2LN)9N`mXv{B<t1J)?
z@hFP}scQA0@qUQ0iOKxaP3fLs5RSi5&V#QxUxVGPmr@uN)@t7L@GrnPCiHH?tm~>N
z%?~R1Wti_{x4-m11)F2hSFC1hfeu~d&x`AW^9KwoU^7bLeZ2V@=(bQj-K;;@W~&Ve
z=r`3UOTd4GLBN>N6WJ61OR(Mbybx%t5kOYi7u1RR%|7i%{7Idao9-~p)_<R5_uq;%
z`}5f@vgsjI!HubUUp|c6%;?|V78Y-I^wdsv@fCrKe(zk}d&>PFNbIMkpVEy)pL(^q
zTU6(TZxTmpBu*El(B1(KxcKH^^z|^ozVy2nz#Yp=+b7ev6F0Q+bFiqR+kV{pLz8%$
z3}?NO^NPet_5Ff@620siW-pkvmENRN!S?>MT_-(yi{ACIG=IyvHqdc9wC|t9cctLH
z#7U$%(Sp(REDBUE5KIYwJc9S8%zRR5_~glxl6s*K=eZQ&J9FWtM^iwTIxRsTZ0>~C
zTJ2T^1P9Q8o1^C{Sh1kuJwlXk%(e3llO`IeYkNZPN{6{f=5BpJ)xBv*6P=$@tjjhN
zX`qT$sI$5nBAaHmTn}NJ=(V2nTf+J!%b7aQfN#}$5&y3juwsL=vb)xs4+slfU*R~F
zJ0p--3>LosKMCUq+4tTV`bGxrw{5toG~d-V^wt7t$Kf!#<rvt}vVPU*0n9+Ty(KWX
zBmlw`Ts^b;kF$cPw?&-Y{_9o*+ac$LcpY>Ow3R0AT)T1!*<L?&+7)kyU@H93=G`Qi
zAt7%U;o!oZ2Cf4(;A|RwhThFM*Y?$8Jfk`AvwH!a+TW1M&fdfIAX@^zeSKjCxz~G)
z*YN;R%lkn;rFG5xNsgw97|Jo$9%<AZm;hGhofdzw*PPTj|B$*4g7*b&IwV4zER;s-
ztFb0jSK`E!qU-=pr;ne1ObH4NsLAO>Ig9wUj|yKr8*|YCJ;2HA&e5^!;@~-~8!&^n
zvQ?p<I6)}|+a=+q5ky_Hy_JV}uw0I}UB=AFE!|u*W(<MC<92hV5*R;%q%%dS9OoT>
zO&`DEAa>nFc$kQ8uPT$+SPa#cc{L_1#wepq6f0J~duY}3U7NOlnA`Cukz9QOzJ%a_
zoe^M|Gx^KmtDb%r5AnPYV{JV3HW!yUYxl$O+lBvPd7J(q^M?R=A^Bkgl#nqjc?AVq
zMX4_*{Lq181!X2L0I|t%DzUP<sPjLjyB7bv@5y}dYZMa6X4h@mLoqR93Jq>KT48a4
zop>m&*b}xh4CD=r*e0qW0cbC%j?6CHRLygG0Xd=ZElmwNm^wucRFE>>jd<o$lqd+=
z-4IY1oYXOTKRO82t)6K+PLP-C!ag@Dz{z+gABnd8yvqgGM#4`lrfWpS9I6E{Li<gP
zLoNkxm^ulVa!pgmW)M3m7N~P2%{_ICGM)D#zE5%DCs_g)wiV!TEReMjU^jK*_lLby
zqD@K^uf%$0#6t-%4PwMS2k%Q-G;&PEiq7g&wkHy!K#2wd^sJcWK0Om?zp^pH<u7{j
zd?kh)JnUc@7x~*<lGBk-itQ?Wp}ihi_4mjRXvd*-q;aOJ*NMj)2d3}i+E1@pxwvTh
zmA)kxAyTv94`!ZZofXf~;EC9a;?04h42FOsWFWW-;V9$~o7vfE<n(3oMqdBGg5(|=
zEZbOe`5V1=R>5Sy<%Fjwk1Zr9Vs7|gHPc7(KN$$t1MFy@vkl{oti(jNmWWoC_<J<-
z&Uq9PT#)p8n6MYPSQJd+nb6-CjK6aXj&W<}JL!++!`U9=5&9N#B_tRN!X04XbCBUc
ztzbY12k3pU`-=i9V1v_smq`z>)Hw3`(FlkLG<lld&zl~yiSuHv#m4+~6TAIOh*&mD
zU%RMX2sD29mHV-FdxK+RNfzZ5ryRG-4c3+vZ`L1|YrGq3X$LE(sA~2TXn+yr&ErU$
zT?+0((8O_w^&LXcwcH(qN@?L%o?%7DXvN;`dPx_x4?Cy$2IyX=0-l;9nXeGRbC%x%
z;zWxI%~FI<J-9*XV%rmo(qO;N@|25QJXI|tA25-yjki@~-MMnclkAt4PD;?&(l{U+
ze5}GcQ1XLP;HNlwOTea^p6he@ow3kPMsnfja;a}^t)16As^F031bb_Ur-oaR*6+RI
z{p87Ql$!RYSbu-RvUDb~<XN0g#V*aq^75v}8tk&+PWV8An~0(nke$hHdB5Tsh?WsM
zXg{PD0oKnOke;)Dn^Q!3@4AUVq_jL!mfaG<a4)P}^U>Qfh+2nWgp=42<=F38^)}=w
zSq;RI|0bv%2CdK$ZJPVAr^NWG33fra`>FDI=|bWAbM#b~*VhL0I1;n67}dtvqb!6~
z*O6hqP+EpWp<cX;i%e3$zbbkm5H(AJ={5wZp-@<T!-G--vPb;4e7P7Zg=57#bQF(Q
z!Iq;gKEct7%H&Nyhw=hT$}?t-atb)!{xSGg;zSc;jb&qJr-RfH0DUlI>+yyY9OM|i
zH88-G^IAxvp8#9Kg;a*?K%@Nd*;DT@yFIeAT#XMAY(P4|42nDjePB~*Sy`2~DAYmH
zDl|VB+)7lwd*_0{bF92sxxsH^!oGJwjbZ-Cg37|R=7L*jiXmzo&}Rc)0RaK9z1C*l
zHZU~I{x&c~Q02BC+V$(_@9N2Y4@LR?52Ignm}5i#yPt5_zBF9A2C1d+C4kGOba+$G
zU1;MvHT40IX+RRTZC`0U4<VIMrw?xW`j|T8f!|&(F9;(iWcfx=1!CbKnvAr#mC@t6
zd|@}lP!~m7At9GV#CF$`Ii(XCaeG0gB-EN+-G+&i1Vzojucrd_u?#X~&xY(OE%`>N
zE(_mCh+PCQrP%e*$frkB4!`=;=1<of&#FXia(E!NdLE;TXs&4Kr?+mcd}_Sc!8MNC
z(OG}5i7b%_=KJUYKkyPczmO>CTwGjf^JjKLn|go#hT&-CB)0<g?QE(U)2DE~YUUxK
zT24-NQ1$4TM3%N3A2&}Qb@Ric1%YODpt3z2HkTj%1)wJAREwY9NQp*(&QWDwWs(Z<
zhxoTHf9pyW7Nuok13Ipudg--Yg{)qcX$^MY>mS<@MuWnk5Hc_A!&$3u%HamNxDm%>
z!HSNX@c9AyY#{BkrgPmSvO9ta>&hiymm!E|NLl+{@BZe0SGh1$@sQ`9jZ<!APSzL4
z$=bV>VeNCqoJ-NRpOPDRQxiHiLe^%UMr-6LC{)M5r7*;SL7m~0E|wM<*+UUYSXfJ1
zANDY=Lg)$lkRClZwLWz*lzBe`$E5x}M8-)0S&8@AMU!kSovIerS*>Z=8$eIjBzEAO
z(uRX@umz9K>q*>EI68&n5x|eHSpW^aHiaer&?m=5Yba^SBr4|V#>>;nMr)r%l&5T#
zMvd`;c~d5oTX}D<B6^B>#ZUC>{zbO;zVBV=y=*7p&WTbRP~wD{yN+Z$!a{Y=RJ^Y?
zvRH%X^9lLK_K}fGLswRR!<D|OK_9GsMh6aidrZefrkmkgTcljW;jxl@;n!l$&OurI
zeLfh<>{`3w6F%ry9HLD3k)*W_`pDL5NsL-))dI(G-2tF5iQ+O<Xt2w>D<LVLOZep!
zx&mUkYex5InVGqY`v%GMp_76k4Z+sgRci-k8^0X26~kd207YwJ=)<%2pqys8_m5G?
zyvYxja66BB`!YC<Vd<)1(lMO$mfcVp8A)g49E?;j^PUDs=@a@i+dZ*%{$iT?QHs$W
z@{f!Ab_M=j<QdGu<ui|Ymbd`<vPX|~y=c%vheL~<tol9A6?0q;Z?@lI|L8tn^K~-W
z1~pblH+>fmTyf+Bm~dl!gMB+3ovfXWONp)2Wmm3KMSOm$_3xv>U=niJJK;XCZTy?l
z0tJwgkGC_Y#ia#~Gu0Wrk8bFy$(I6<So1cNj)!ri0|u2>v{@5~3*OxeTzY134=(}3
z*Us%za#Ql#>)Ia7xdb`o(t^sYHP7LxV3-Fp!l_PbT#c5gWnE`5p?yOKv(HwbC8-?&
z*mAHj>4S4z$%Xma#~^6O%C96Ccyp5BxIlvqR;Ns=nbP0sRP=DuY28Xz^s-Y>?9!o)
zP!9Pw&9ocCJj!(gQMMK@W}@IUL5@qU$VAK~ysMd|ktuZ!ND^IH`L;0ABf6m*C^gqf
z3>HPzT*;D8tmmNpjX&=Vf-gf5>TJexOBZ4bk~ewabC7)T1;|xJVsdiE;_wHEEFc+r
z(-SiS$$rK|PkIgEOFupLEGjf;$wecfiu1sA&hB{m{rW$XBvff@95G(y>}zJUj5#1%
zU{EKF#Mr^8x4$Bx@o$&Kq6YP{eMT1ia^j(p>%+)r5hL0s7xNzzZxRb%q5Kdd@r?rj
zV=iZML)rXnH1LnNULt-xFT#bSxeFDaci0U4b~JlRKMleaa{gUogP1zvAf>qv_SCU~
za-%_U0-yw`sQ9<35b54Ny5r93-;j(Xwq^kD?Q?#%BhT;!pEB@jO#40BFXp($<b?QF
zbBwv#1uw->Rz@AOkWU+$nl7A%o+7npg=)&`;rxPBN}9+Ug=-SuubY-dbC~nF0E7_Z
zw}?G#DObP3F1NTRa~{Y}SQjaZMu1pne^G)}PZ`JbBnU}tzGdbe4Eug|Y5GE|K{m1b
zSJ|Z89@+om<P)jaWp|*$>tay@qnrmrE-sv2?0CgGkR9l=Ih8f3Ygde4&z|43vaBHB
z@CKc?0F^Q2QXXhEkAu0pjCGJ;W{|fg|L?~i46@aCH-&?mL~bd0X6C<DbJRDE3t2V4
z&CMvovF&R)WFxe6utL)5J2AYm*nI(-pIJ+>u(Z=Cn0U(}IoDd#o|@6NcUqC>AsBi5
z!rl3IZ+?DmW4!X3td)Q{y$k&5BPw}XgVA`QW?JsoEyDr9#1RN>7lAZMBE?PwDck{t
zZxFxH3|*<>dnT!oTBtso&V06hHxYPJFMIhDm0geClClYb|EyywSpHG@rJ~chxqrtF
zk<ZZZdz;n($nD~02F(}E+?v-Qk-l25mtSDRmOzXPw&3vNt4{urc2F>@d5OW7gQb}v
zKLDL2qftT$qKXI}|4&k^Z1BqdJJ-}>-!z9bSY0obugJNz3$`2n-!+zWbwg!J4UA6s
zhAs}UA4~fYr|+d{BTp1yGDQlK8L{?PV6YMGJ$84mbT9e$an5d7IRp>W1NEtuX<i@=
z&MJIXMmH8~;emX0d2^4Ykgx_2OT|45`bc`Q40JvhBV?Sy%A)@bWbAn8MZWmouPVK-
zQpYD+l=@QzhG;<!@<@1Ryj7_MGvErg!j_#ws1dE74XVlEx1E26J6%AI$NpH$<}O>_
z#SkPZrjO?%7wGB?2Gf^zFGNK$2?wcGw4v!pnpkRpd>Pe2HNcL<&pK!kpn*c|sBiDn
zf1nsyD~~!vpM&9?b}h11FX=EF?$^SSFBYuM`aj)+r!pIP&QRq91kZ#t?F-f6qf^r%
ziETH!N>meW>;*ue6&W*tQSpnA0f_V?W>yzWPC}_k%Jvm5NAP%mza_=RWg#)scc-?D
zdU|A99NDwj9HS!<QeTMiDauf8(h$HOu8HTJ2t+En5OO4j@sC@V9ju+B+G1ZRReU7O
zYWl~CaVHIhk{QMobJYadrs5T$dl1l2r;EA>>jXo>k6wCjpoJkHS8l_m-LESZRnioN
zHwSO`9WPnjq;bK$#-)$RduKvr77zb!<rju-Bxn9HkAqLWu><_5$O6PZx-%#tYq4vb
zfL2j^p%}E3V5<56(9*y^$o2>V#}TRC5PPDs3;qR51=TD^NIMhL2d0fxV?4Yt&AYH_
z<wp_7XMGSA7=$Yk`o8wRvi<^SC#^YeQ)c;~?u`LvOg;K-n5Y??!t$JXZr2yIj42`J
zc4Q2Lk~WhH8hWLS&wj$q^SkWy_4kti>zM^<avRD_Yf3fQI^^FIs4jL95VI5r804T%
zLI%XMuq-dgpfBEGaC%=Rr6L#sTN{|mw{9PTeF?eq>b>(D#hP#crMmb%Dn(X;G_vYP
z0CVE>reY4iv*~7j(_acUHHVlOHC+xYD;x7?#CV2K%}ccNdfSOSbcV3{2h8!V#pl6a
z{;v@YzmLy*j2yy{4043S4F(#7l-4nX3DZP~&u7S%C|Er`J>A~sE0D8f_Q--_2*?8M
zFJsO=bs=fDLjh%?<Ge-M+-K|&glx}0o<fkc7@0tyD55TkjV_cATQB?|VP6JR<%Rei
z@Hs)sczna|#i7d&q<Zn~aNN1*Vh5Cbf!Ts^kaE8wa%)K*jf#p?R_z@RjtWsqJj2%}
z5t7Jm=vuJ3rDcWI%q$E^j8Az3%X-4wn!4Puln!s^QxTUk?!p&dAoUai#C1lAD47A-
zuS4-Z0KZm7HLE*OIZsbq3gq=7H<<{_8GT@8qiP$AS8+7F)ioFnZ2{od`CgO2Pg;HP
z!|nO$&mO;7T(d*W_=e=PKS*f|qcAL)Y}5f=%jth5!|5eQh`6~=#7jDJwh4qzkw_HJ
z34HxO@T!Wh;9vc>ZhB~-_4M?Xz9c#drg`1rN&)_1r{IMrg*rm(`_>ns%KmnvnHG|^
zp)2uxW}`0HQ8)=Y(BP(xZ+c)BXfRY3C-z)MSQm#f&6M<jmxUX~SNQj94|+s-@%|WI
z_#QhZa|UQ8rfW>Y0%^br&`fTg;bHx4J8&q%<^t(VBO~%z4|P1aYpHW`(rO-`w3>Z#
z7KPu(@PPQ8LjGj6481+hT!Lcari=0QpOH_*X7qr}3;`-;k_OpdW{|SOj=k)1q0VIv
z3;N_v-rnBeaqB=dRRoz5O%Gj3+2h+GkwjZNr-M$Ris||%zw@2Oj%4Z(=%I2Uf5}$@
zb6TVupgRn&NKaie+E0>xn8*j=A(efdpNwAtV^j;54z%h1T@5ETWM3dngS|zR1RtP%
z<`?D=9~_45+xBM8ISu&cVPWZ8iu+jQQVrA7zeeGZkJbNU?=8csYP<c>rKmihf>IVC
z0@4D~ji?|kjUXV>4T6BQC?YLV(%n)Li$+>W0qJfK7Tx(D6QBLxdtc{#J?GQ=eDJ+q
z)|~geW85QuV+@uwb0VlAw44Vw)4lSeF%eiW>HF8$0a$;eL^Vv*O)<&13WeG;TR$?b
z0sfxBjfrGxA2olL)3fO@s8%*#m>^?4^-<RRw?{LV&Ydl3P`#t5SRtd#9|{fI+fU81
ztGc?hevJmWjRyQd(wP*@M1GJnKeEhqodKg-FvRUPqq>w`(S{TgJKRQgc*uLy5@O^%
z1E08|oT>(=kJaUbaA@(otFVWd0W@F+AS&Y*_jjCr4-C`k1;oA$FvT=NnL;R87)&!r
z6;=&x=>fO-+~U(CsG#tQvw#_5${8U0_yv}3K!R=qU{cX;L#+ehB*8z&w+2&F#Tp3B
zxGWE@(SP{B71Q2kRJb>3y9v%mXmICz8M%|?1RCEiyt6>?07w}Wsz0nCNSf;BB_J0T
zO4TR<PNJRj%&&_kxPjq%_vzQEz*T=oH<2C`KP7?BfTXFvogk^-jTt4qF~G%`MqEQk
zYiFQifEG@BBiCDhj@BXs0YPvRF>NB;*qsb;Sz8PMTtbj9`DOcbJ^x(n-j+$~gh*st
z^&US<rUJEk0CaxIXyup$I(|-ws*F!vbx1IvBk(S~XK)PD@g_6}&;85%SU?fqyZ57G
z(R^9`F&kgmUPOl)?@Ho}d<Szb*Y@APXN4Gtz!?f|3s^0a@ukz%y0kzg8^rN~<l^(G
zVdurwn@EDQX$0wPEuT1muu}&1`!4ZPzlq5Cw;VN@I12Le;$w~g^YyW${^hqgEezkg
zM!oyW48*Qt*a&_x^=udiz@XeUy)Ezc^;UgtefQ)hK&gNoWP+3~b^LV=qEE+5Qta2D
zosGE|G?q^<bgiZ!v=RwV=M)IY>@k6;z0J_v1~W6WHG#g>qk<=pu>lQ*6+8kSC3R}5
z<^y{A&@W&Bqt<mD7QqJ)SusIr7bb(}P^9pl+L^qQ=9zC$@d-&7we{{rTGpEeWX#Tj
zN*qouq75HOg=iLdVGSL4NN1HK#~P;AphAMr%NMC}0T6_1FQPjQI%B`3@dut3|L`R}
zDcCex)fy3?k%O~QQ8RDwyGWt^7Sini`hUlO^{uib9QsqQWEfTUFnb~&0xOSnA0>?)
zv-mz-q3gY{vFD6nXYT%rfotFXjqYD09nbB*=Rk(22BWj!OK1kpfAyc|$v#9-kL3Dh
zcz<{^x4jF}^-O_Q{bOKz9C+lUXp$e0gkKj4gq19Vf>?i+V^h1V)_Yho7gl1$!!+8r
zvZFg?g`}61%SP}vuh;#%Qz5cx8g+xTAp=(yUP7)@r}?aU&Fx=wAA~%i=o8{<N7{WO
z{V<?nP$y|vGRh_C!c{JAD9|$4`2OA{=!`b~%<;WCNl-cE<_ZYF8I<5na`M?j9}h?h
z^uoU=brblD2rUE$ws^tz5SvQA4+bA33==|-8}~v&r_o1xO`xtBDD@(lbcq3}5t3|a
zB;wWM`(P*i!TY-=f%hZ))RMr6f_oKhy$}mQ{27huSG=xHzvux<^^^F&9Dd^Ne-DT7
zWw&2Xk9a8WRqsTVa@s2n;28k(Sd#TN5G%E{vrS)7-raU6+`0}~!G76D+=4))I$h>{
zpnULets92Gt#6(+NeHMCl4T-Ex@ZXLEGGba`!b&9v`K%<U#$HrowT&{Dto|%e;0=s
zRuBFfR{92y!XUjeX<ZQ=y@-Vc-4$+*7k=QhfgVENz;Ww^)Vp?CgNlJiNImwK3orS&
zSPw1);Fu6T$X3s8oU-m@z<Dcjn-q`2Rt3HeaQ}-JFGAbS6gvo9EpG8u6V4Cj*yte<
zw&BbF9wFtd3?PB&*#Q^}IOvWYA!2fNCXUBUX42r8v4rL|_v^aKkq;kC(+X2MRv-x<
zZhPTwd(g+p<!GyKt%rw{;x|+6a`Vi6PU%ULzT&*REB~Cc7%sVLk|0i{Q9F^hODQPX
zVGNZF>ZjJg;F|sr7zp4R>(}!4uU{I8x&7|njmPv@LHbI_-c^jafUDSMO6lGB8xSrS
z)*G)r)<Q~{nMw6gZ5V2srdcYb2MP)Z2(YoS@$<KMV*R2wA(_B4DH?<pQzk01wCRtb
zaAN6myk`?rcca>%gK||ek`%kaUWWbera!XeA{S0-4KxyfPo}p*TYJ#(IYp$vnFF@r
zWtUk|RTe|&24Nh2ukmCt?$~8{Vfo?!=`U`R8wrZw7V7_OA(5gWaAe@S%j+NA(7oF?
zzIy|LeWo7)MwWp9!*XgrHv%)U5(gNIGyw}>QINEG@UAkjZk2&7|2Ga4+86<~#2oWw
zlp)vaWd3CP@$#(QCHNAB$&EAdQFbvz4;|6ptw{dP8yn8i*blR#oueX0Pe?O(DgRx`
z*h-V=xxqdoj;;k;r@1hQ;6N_S`n6aUnwKG01$Hcv76kLvnQ@F$+k0t{EQ2^RDrYkk
z(@FxJ=zh$CNjdcoN>h=J$ga}O#LUbuuJhJk(pdiZVE}N07wVt_F9KaHdt@5aBKTYK
z{0nUB2&6G+W0fd$kwub1TbqAYaxK4E*FLli6xdwst$x#C=<31$721!>;U8Ys@BMHG
zDe%G10*UsA3^q2<F)VK6GJL!u-Q}En&JBPxFutq(_3LTia-2!s>Bjx?cB~}7biw$^
z*O~Ly5RtfcAg7}#e1l%~s>|`-+{r5b$?Aj`#c@OR+QnhPk+yluqeifnE1-wt-@aOH
z(QRR%$?#o@bz0_iH9=EGVZBPYOMoh?TijYtc~O4+96v9a4Nk~5@%g=jql+9|T2RC*
zg1hqx_-D?(MW;JJlQyYK>mS9Ey1Tn&)4B}`tlbK+e(Rg3m<o{uHbG2ykoy;=tU0;s
zTGi_L<G!o6qp$|Wv#C+G4^kA^qS}TJ88>Yn>-=TNF@U~)8C_9zoOj>p{wfl-_R>!D
zfGm2mIR*feLv}Vqe$Ba}I*q3B=Fc3>R3C;n#IoAzk@z;mP#ooqn9*eQNXl(ERwhy(
zc2kLPVHbMelmvAOK`vIF)iiT-ISC>7D<<*43<|23NRwl<p~C{ZaTM%x27=d`d5-x8
z@^ih5Te8y9{g`e3yDEI>j>0tnV9O0cqR^za(Sik+SK0p2QV-UW=b$8tCleA%EXL7E
zjM&0S+VdvOE7^Z}<cl;r0z>^7@3$;wAR!^%oWYv#G+jOW)zt^L(sTJaIXUm&7p;_B
z*0zfp!pw3H)c`s@cw7H|#J2j5x}FWR5w}13R`3u&_1btZtkO%*xl;|866=)xSYzKe
z%B_7NfdugS>Ek*MoU2M_9rHW-Om^rL8IJtG1usd%sVWSGjh^#dkFH}jw1I}tQ-`Dc
zOV7_Wz08N2jK>t`8CXx+N1^Qw6qUdWZ%YuqSdIJL{i!bwO8h<mbNtI{g|yFZ-u$~+
zV4%?C^DDE?kM@^nN2o141lTv^U|(5?f*66m#{-`IS6ogBeTM+QkJ~s|(t-A)E%|lr
zPDN0kO;rV-$A~NKWMlD+Qu@d__qrtC`W$8<&9i$q-u){~Y#@*i@xo+z6_r@f#p#2Y
zDt!Um3xSl)RA{OT4K$78SJEGv-j670>Awu%a>rPR+1-f)Sm(0CK`7K4@RT2TeU;R@
zrDuAY`lQ-Jt!wwn@)uat)v(&PZ<A69QFz9=SwVFZ<A#_1`MV@62+ndSkw}ontsE)+
z$%o=>`o$^bK^?+$B+Pdwte|wlyOuNWtTs3O2h39OR_p8Qa|)CY+`*fJ!Ph6dn%8Qs
zDvs>@@yQC~5PBaoK$_I~F@VG93}-yb>PfklKgX+T0?pSYBDl6eQ(lOLW(Oc2C@w&)
z0!~f#2=TJ2T5Cub1Vy~!@$D;W+S+qFAI?OK=!yp%L66yq37tQeoCIQz!tvl@k&PxH
zxewJzd5d>=4%JM(Umj<i-D)f^i9B=UVzohJgw2_Rvbyf1>3@8(bcAoF*B3*v%P{b)
zbgZi9D@l8lvF@iiF%M;qk5EVO#*2`OiqI>5SQ9sVX;-S+Ote|WXn$E0%6{vPY<<*x
zkLj|Qh#y<t5Z5QQC-EYU4tD37a2@a7$_V}*JZdy)@ma=>vq4zqo6J*c86ufq)O0>(
zkp`42w6Wc`TQhO@j;qmsHfkOuMZM1Y6YZ2`IkGWRI=8etVzX{{kn*11|BQPj6#Bx)
z$4~o|k~UY#Z?6><budinHU0Ig02k+0L$)<{Wo0EhJ3HVXs%g?_bBs6iC*KJiZFU2!
z<JI%UWgayHZaJYnJy+M@xmYKQgU~+#8&_6%aAaN0y5FX8d|>d|ptrih?|LiLB(!G4
zm0m|u7d8GCaI5<V2a4mLFX6kn*llkdMA8`QOk{X_hj;05OUTJ}Oiw4DPm|Pdx<VaY
z?Zy{{)o-ojP)6#n+I~`HctFJEB<)equyn>YgeU*(LGcv26B^#A7hCByt>`L(>cy<C
zsTl)_z1Oe*IMXV4bToF8{CK~=AL>xW1%4~sd{lY-{80bZdc*FUkkQeVX&LUV?d^cT
zz>bcNTlR{g_A8^6C8bsl{knWnN0DE?^y}8|FXmRqp18=FMaP!t2TE8)l3_v`pBfO5
z_M=Ct{PBcM2ZJ`Ga8Tn<0NK{j-X7rRH~l1))A}bHFE4aJ57G9NUEl1Mo0ICetC97g
z$&l(KU1H+q>lF#BV$v}fhzl90vf97dph;VDX?F%2^~(pE*{8^KM*)F*d3ix)qb^}^
zJ55oxNWCt>JTw#4so&q$<{&XJb3R;a+Q8ab;;L!Gkv6`hf4k=G(ZJdeU6t!LXVQ?w
z{S=TDw8m(ianA>ALusLARReC;)@&z*5;_FY2B{pWgiowqP)rWZgId(4{f~@f?1wL4
zO}G<ow6b?hGq&HA6F#za-Hr9V#xmQqY)nq;y0<=a*nQW@FN4hq6(iz-b?|EvEsqm%
z`cmH1W*vV3xQMdR>vb~Aqm@zQBm83nLG)0Sfq>Mg$6`zjZ8_Gi5%NVFAi`F|laU)s
z0|k4c1wR9*dhMSx9qw-}E_7lxPy(xAbc#2-x2bu4eTE&bI-FsdW0vd)0o8;wi>%3I
zg$>u>;M{rZpErfaR;YP2q>ZJWg2zRqwhzpeL{q`+8sA-VQ2jkh)r;<szoI&_ws9ou
z(cK!-SjdKhC(sY`2rWPV^K>|`hK2?#KQ*_l9@P!I9f9NBskIHPzw~wb@OOB4%_SI}
zPK1z_)%3mRC7*Jcq6Ap&2F$;<Z3j&q+<c@fsXG{Y4x9W1+A=Bo&3aY?3k|p!)m3wz
z|LhBqQNNL2RnqOuZCQO}cdQYta<^vS<=+>!zfGa#(``qQTY(B_9x1oSXMcG#Txzq2
zvvb&-dngHylhFQs{PQ_%yP_d<XA5AL(CK$iC!?h0FyZOUIgq?{b+JR}z|6H>&~0Dt
zWZ@9)gtuKT20Q3V&6Af!R_#yjcDP0Y6b4%WTR0k^8ae@c{l*u2KuF*aD#)MIA@7is
ze-U)wrmA@xLv}orJ`MDQ`mQrv>12N(_*6$5j;Q>PMR6|Be?-?8E_jS=Y1{m_>>01q
z)4Yzbgeng+0@cogoB8Co<_wpYzX$i#`$lzrvAY8glm0}@@0}9LyHeL9$13NFv7Ri|
z4HD)dHobvYKeD<<4DeY~*p#Vj-X>BVaxs+t%P&ydc|-csHDX`2OROk`-DthyociG;
zFXP~aq5*wT2qpjJ;6E9e%a&y}bFzo*fQ(GOyiT}HG$s}(eC!~U%Wd3;^Ab2-Jysfr
zmZqkR(`y5RyLRi-(|Us#s6=jzX1RB-F^Q3paSwc-g}$oVE}YstT;8TX-ubJ!fw8b5
zwKV-bc-#y~-u$7*LmEI>LxHiiwKe#B&e8Sj*S|zW09?F{rLYaC#ay8E+ZMe)n2WQs
zv&+M!poaz98<_$;6Qn!nU+Wjopq@Sh5!)m(-5kpJ#@EH?S}RL_FjXlYcM2Qkfycwp
zrLfzZ+uPe)TO$>33f#KCfB(L^YPIj=Oi}-E=w0mY-t_cz*AbRYz~N9j0r{>JG@vZ%
zziVeaYQK{VMcGm`1$lUQI5<L%W9^(s!z&g9t&Jdi=W^>A6cou^i7T_35Ov)Zv8y1x
z>=k{^5R!&m;6P1+#-CmW>l`+O_d>tzRHOg#@$nGcS7Yb_*FQe~B3or~dAU9J+l_~<
zZEY#`yaAR;lKtkAQ=y@uqE_8QLkGtPD}liJTn0TlJ3HULeGATn+3<cMVKygHl*o)T
zGzM<J<|~6I{PO50ZCq>a>uRd#>_#}rGdOtqU<kkz(PB3B+KC(0x#x7mp#(VEVr1}U
z3t{eFRP|oe-lNgv&>K2~-z};`cFWy%R#J~_HfG!UAA=|D%?{ZQ6_4Vgp(&o5fmo=g
z#4znq>%3MC%gRtRPddWA`pU+7(^skBs%6T0Jj`foZ{Hp%Kf`eRE%z!n^Udqm=NA^<
zKp>vnb$7y<#PZdyWm0DUS~x-6UaR4x#>S_bHan<8dUSN8L6rcmD?m5c2%=DQS!CkP
z0z1PtdlL^oW-TASlF@R)6dkT!Wdk~<tw~4*SaecwoZ~J?QCLsuj@u=0HM&l5yHdqf
zM<Ex&)9fUr9x3!Q<6?VBOk8}b<v!VDY-}u)xOom~SIve^#7O}r_51bpk6at>HeLPt
zrR$4i@c2_P@c%7{MeW*HS+VZy26pG3?B~vH><9)5?HkcTi7SvqxvW<f3gV8;Mjtdm
z!4N&uFUp4npr}Kq!KXLjkog0&cxxlvr`Gwa=LK`Cj~!i?=Kmd>545=qVeKq{>@+F}
zpOk2p+hGXfb{`43S?(XWovc7E>>OWFIW&yy@5%3|-tWn^z<41o+)ji~I!@SE;&r)r
zc)nO<!nWqk^Pofq1VEfD5ad5QWEFNMDz0wG%B_l|;8y_oQ%21b@+7JA39sm%pL6X;
z5__~K({YCx!dq#Rn@3Q8s(a1+>T5U~03&yVnrZ_IF(0ky<n`f0j%N8b^vWwAJKjcb
zxbo6dQBfI>D?``jIXy!ZYGwzT=%s#18VCwh8*_E*Ui0^bBO#`#Nnpa|6<zL)`<$Pj
zpG~ixnwt8Gc&KxI{gwJ0qus%T>Mmb05N~#~8S@$*>N|aA-{R1UC9b%uZ};TnWU0;E
zdY16!uf2^qP!ggj_)TC1iLioNz&1sqOD^%&hK101+FiB=Uw-k2%IJQm(_`A+&ZmPS
zLW%qoJXk8=9W^@D1-t^4Dz&O@7rJ08_Cdd(*b$V0!F-Uxe*Ac{*>5>qBLA6Grv<Fb
z@>CqSm-LL@!ZxN-x{U2M{LrY0e$YB-h25|nyfV|w3@sszN<rtmy199IBgz)xMwDV}
z-__W$D^iSv6sil@*h!i|g8EWzpTYV6cnd1-Nt<0lSuy<C+8-z|w8`t7ZqYt|HJUgH
zl62;Y6^}mjKj@TSSzab5CvSJzSvRicpu8FSGUHXMfsUCOQ=2zCE9(nyYm|F1ddjxi
z+%67ed<F>)UYfD7F?}0Up4=R^W=VlP1`b&)L1<uLkjOkTGBOkK%19h$w=i|}Kv<Ow
z^bG5n=Fn(E4qlttR?HB3gT%ZBR^ofkOR=D;D)z#&&7GaOdrsa!X(=3($FaAb$^ag{
z)5q%1Mb{#T?<AL(^PAk#&M5_rp0uFk<?z~Z2Ex59JH=ojp0E&3<Nc`7(b3^&ES1?D
zI{NxF=md|hLZQ$n&<{&nPFs5nPR;_nbuw^M!^6U0KQ>b*4jPp0_urTis+XoZPIz8>
z^|^CeEqq7lF?Li}NvBcrQN*_S(sV`niiB|ByU$!La_{Pb#;y3k1=;l}?DN<U#ZSFY
zGM8lp!HXq-(!bI9emI%oiEh!P6B{=cDt-#2$KKu^NDteTLKbU!9)A9ESgZD;D}Vcn
z(`;)5S5<yG$PeALwzjqgyw>t>`U^X~)<0TTf9Sx1$QHh;*1rC_!^OuJ`C&qdO+3)j
zx6s+)ry<#(g@Xf6=16yU!O-gqC=suI$tzu!4}+_!>_2!_3*BGuSy-4$YHfFQHD2&m
znw=Ala=jkd$OY<gbH@NtxaXFm%j?&i)%&yVAvSaEGnbN}+aRbKMv6{y#udfIB>W8i
zm&XYB+3gl$_;XVPNy{Bg(M^;e<D_L=*tk9tVABa3c8}uRTzcS9N3ZtmnK!$ty}iBh
zHPOL`H$M=L*_E>OTpi<+-)O%4T|!}7?>ZjpS3V?p!v%D76goD4*)?{O{qyHFRTc7~
zrQRa--d-W)Tw!j3!>vPxsL6d+<}$~%$%bexjyIt24y_+&7+7aiOocRd8hV>{Rzz`K
z#;dx@&uXWu)3mHP$tok`lK&~bp&{G*Y=8JJYmJ1|!fVgem&V0KB3-q$q84(*95?+7
zqg80`vW{Bl=}k{rmiowvo<mu&!Y(XYWW0TA*}K@0o#R@@dz(jOpx;5yV%f%ixuUmw
zL(6ey?Br;yqqN~DN4q+9v9s4Mx@-5E>x6fqO=DvfLAj~RQLyJo$(X{EUz2%I)U%|>
zhor&qA6X^!iy3KYX+Bt1pE+(=6@8dUKEtzcJme!;XEn)EDgkL(FLZP$HP3s($M4+}
z8X(0cm3XFoG7Y=kuqeFUugi-z`u3VuerO0}uG=1q+tKQqf$eYK*85cg2)zKODAb2#
zca4IKM;?3&)}J%yQPFyfjZ?V(*b*7tAigago=?%?$)7t4IE$J;2MooS>-(?1o%Z43
zxyXYPajmMI$|jqaoztLxn!BAS6_BNIRJm4`<Obw1F$pMVY08T8=Ja$|n&n09{`&4M
zIVEg4RuywF(-Y!KcwbO3Mnl<RP3B4wW_dYh@)A!Q;z+%m(R*YHwI1>+Dxxjku3(N}
zH?H0E_lhb%YLfr6MXJV8?zm-=1N9NW>TGe_D=1xyYL9i9UmlsP;u|Eu=Ek!ooo}lf
zF<r3Uz7ltxhu2!jhfonzz2zWPmB{8a25yBEjdY9AiU?EfXT~=d;7Sd%qTKB`goWb_
zn8-<;r&bAb3$<`t*aSgDlNAL#;^5legL*AHQq|C48XLZquEMySqRU_%KV>C;e)1O>
zZ?k+gN0S9-A$&+;IZ_ttG#$#x#T6DtoLeu_#BppFZ{qvW(*9r@WAwz*($eBON@OS^
zGScT_tkqn*DJ|<}YUyz0!LJ5ie8K(srkRz)n2ux<s6tVA<S5bky`(}8lie5Rt0@n~
z5w}dNPe|>(>jc@>6fzq@zUM&yW_;Y}Xxk0>$KekdO5SW36sBn9&J~9$J5!pwUbI!V
z|1)o3eBsVJsWa}|-9{X^ytND9ETrwu1|#>#TtHZggIVUr4S&G^@5CJq14H5Jy{q#O
zm?-Y$Ilp%GEg>W$5f~=Uzg2bN3ZsshpQ?|HQdP0Bv61hP%oXiff|4?w(E&GqT}0V_
zBpM5$7Js7udaE1wgO563ON*rm6Xh3#j?^*h)=WIF^AJP2xjivg=PG_=!yK0EkJeUc
zzJs|chX((E05Ka;3$14<4b`J=Cr8g$d~V&m=`&k@&D??<C8F_!TkdhSi~dahZ2MtG
zxWB*ng$q?TUYoXXP-feyXK4l)klb`F(>KQ`y11-78X1;-T(E2P9Y;c2DpX6n;`T_V
z;yPQ3F+sFsF4ZX0dBfD&GpNCzkoWw2;{Cc`mGM1H<lSUmJ}0MY#n?{nHvKny?1F-B
zM4U_~6_=yG?CQZt?;Z@k5E@}NwqBf>nHjECkem_)Bikk=izSlq65V;!;KS$AsiA4j
ztUGNMDMwxJ2L=X=WH~aUGoCSM*i1qMc-FCdU{d!CZF1X%cke;M8Fj+<Ep+Hp-il(_
z{H+zd6XZ~+?2l5BRaK)U#Svc|GD;V<YVQl545S+xz2ndU8JS{GP!Pw$QDBBRBGL^l
zSNwZ=dY(J+sX0SVVBy${Plbgl5pGxWi>@f=SZmB<j&>QEUMZnOZk3>U8vEoNCRX2t
zW@NC4nNPhmb8!*mj`x5pnf=F0)6>457s3ks{G2&WZc&5|qhQ3_<L{FR_4zMfkcjyG
z`?r#X&feFY9I?wzj*hSVheXwzDyJXDp-|*TIyyQV3Yf~REjg~8{+8z;HOOD*Y}Upb
z+qDBvB8gNP{FcA=B+~20dDgYJ?t}!iF~7*AylEHg?*hju^_(dvEF|JIi%O{vUdMMi
zi!yp#_JwI<lY6M8xmj0!4wPCSygF}_G>!FZB(V$}*{w7})bevxRYEoo>`HbyzDuZS
zX!r)p{-|6nX650z&J2{>?#~ogT3VW}6TN5yQ@yzlez%xgIH;!&jPH(sU}La8QL?#l
z>p-$Bvx7M!1j-(zJ98xbO8*?b`ze`o<cOr8t|9G=tSC4G*=3)^tZsiIh~Aa5(A$GU
z6uUfRcBvgQM-%12!BrZBLMaz;5=-qFc8zFtu}0^8`}TQA@v8iU6Q|qPU+L@No}Oum
zvupWa4E?|)Y}AXNk$rN&>=Sa~UC18S_$G6Y!Nz!Uc{-SUuc*$TG6A>Hv?UEPhaP4X
zo_Gw#L#ZI~RpJ1XaJPZn5y_h$6c;2u@+9w)0zs0kfSolL;INfK68eS}uw9P7S*LHJ
z&znMH-OyeLk!(p9f%Lk!xEQyz@P;MwX@U+91P0|NOxkW~(ivZum6Hq9;UQxb!nP6x
zH}#YsI~afblRe=ij~Vt}d`j0bN=uM4X6=J2_!g8Xk(A_vVbh|PwwR6P(6u#|$BfMt
z45km%Js2`Ijh+<2)O6)D1BfZk@=$i}1a!B7?D*q_etV1C;*qbU&~@s|M35i$Gc>2l
zNOm;}8x^0}rB|k?sBN|!p_;>VPv}PW{K`s4HbW)H8BwM3U!P1u5xs-hS|VEGv6nag
zndncOn#5`Czws@xLa$%?oT_0Fl^_Q?VP|yzM(G<kh2U9MEp^AlQbVoME+c_xi{#Au
zAM@$cvz*gTp&ENxPgPWf$u2KoFhyS-41JLCpmZ0{NTG~J@Lkc1G!&OU|KVk2{j{iQ
zy`WVAxqUFjr}V~LVc_cHuy0{hUmkwLJgDGJESmILzNeY_PL*y(&pmMsfO^^t%~P-8
zxD$R}-ZhVU9Rv-H5_xB!GE!BJNs)SveM8Db?ppY;0Lop`IB-BsAtLJFd7(boU+C=m
zuDs23s{|Znrr=-}F%2*fU;R7QH_k;@r`DgJXT#}Frp=9i7}?wJ^!KafIEJPKe)>6t
za`%y}4#-?)P-@fHoeDCN0?%_t8)HiL%(J9g&$~+bOw58;%@-CITiTL>>uFK4vYX$h
zY`e)p@<~`R<fp#IjT(Hx5S=YyC9Qi2rPQes@^ouJtr$v1^d<4qoM6<ISQvPDcEu9b
zdzdRL$Gl0NMHOK+On3#eb8}yti$pYIs@$&}x1^LmuW+cLFIyyo&Gzv_eS61cEL3_3
znD%X1-%NnhKJSX0+;S8|=Y=FX%KrpMF{4sdnZDJkzmYtRYj|s?PaQt3p{_2~Wle~R
zwZ*KPNY-gLH8t7PH8aaD%MWpVc}vGU%*xLnb?X8g2WZCDY~8Z&9&}Ci(!{CVKH{?h
zHYhrirGcnW%#l%1EzHTf^%Gu%p<=m8F2>0r50#BB?1s1uSGn*5uLHZZth_vl7|g*^
z$ifW0&(DAB8apQcjsgR))j*Pl6m?+7PzFz&#?YYZX}a_4mil_BDCW%X3@=Etc@C`-
zyKXna;{V8gbwXshC@2+BzXOPNvr>8CRz?@IQmEPzz^lZ;d1Ebz%#?&A)`ukY@a|4f
zP*A#_6VszmNnl4{ZfQJe`Us291jAtMb`RNy;IN08y*JoLY{?kPu7Kx6?dDElUf!pu
zbEv1B81CzOAdor}fi5-FBp0r|m6DPpV$dG>+-n`^J3TDQCr0YZV+=Ri_D2@aB|{@4
zI)<^nKGo8hPe4;$JEjo6P4@Ph9OCA_fQt9%dhuy*uY+r+kx>)$<<NUeHW#JnkB%Kc
zY<o7VqLItU{H;nH8#$b8nBVP+S1mqJ#cQO*n#x3LeQ(>o5G;=m`AQbg*1+oJzzW#g
zKI1Ma#V%Uh{o)f{XW}QAWX1<m(>R9T6=gI+qMeDK@m(f);9wL#%}xYbyeT3{k*rY!
ztJ~AyyD=9p?~Se8VUuO>+xVEkcV-L#|EPE%^W$69@lT^5HXhpNel-hf*1Hq7$=VN;
zjIPVxl5Y6Mi6id`+VI5~gylwoZ%k&khL?D@CczxO)1A+WwovYRziZ$#sN%v9mM!s-
zfo<Oyu|aPMvato-0&m8%-b6lP4ENEeqI5nL*gsBiUs(THSYEz&a#muc`ovYK2o&!>
zR1ZKz-aJA)Tk9RvF$x8{1+O;5uux{)Ks|k2h<ZPKE7vw6=IGGCS5godol9n4+$o(q
zA4+v5f=lM#1Cjx5ZoR!gzf-1Ue3St~@R=om)@i6pJySw@<9s?Afzjshp+rJcN^g64
zRgazy8@d0oFnt<GKG#1jEsh1bK6=wxorxgd{%cr`MFsHthqBj0OkUz5Ws%~K%CfS^
zWmyJ{z$5dp$TKL&TbwDW0Di@doH+yAb!r)^E6U8sSTDw-I`1qN+QkEPHzTI#W*Ln%
zA$XL!mm1b~o{!k&my}$`;L=2Ur86tCqufiD{TZVXo_G_V+x|zE#s%!Lk0j88WsW4Q
zYk2s%(kqx8LPYiAbn1HXjtC8<S%Ttno(4|0IcApt68bBxr^J4nym%3x*y6Qf4n9@&
zzL62K%dO`3-cC$R*vYh^P+e_HzP@r?_CFZaxwy!|$Ho#_P+S}qr7B6!<C`hwc~K_)
z<ui51JCWc(jEWMc^_D}0)WY}wpu?x_GP6k&s;jG~$0WP}GjY{ix~-L2SQ!fo%XqU5
zoFp2l24Ln!MMwKd#ek5FJeVeC-q?uqFIIY)(aT}^(1}Srbe8-k1x1JXDf4W83G<g}
ze+2~4tr^1*Gx<;{38VTz()7-Ia8<^>v?H-}S?tYLB4ngGi^4GEI82$O38vP9*Qdw`
zx)|A7Xn3c*9v;#q&O1J~e5@f&G7vFIh@t`9RdLbTPVmq!568u~w_cSM<o6wkofY+&
z5(0*obsqIApzC$t`nq9c`AjIIe`6g;BjBvVc?7<aV#-4=E+qE!$fcPY)b;QyI{y>@
z9L!>H7Ejjy29L?um@pXnPJ2z3VBEVTH1g(mlgxoj2~n(rN&VPCQL-%Ciy1T(b3w&p
zfa5iV%Jp1B%NFDV)L}TY%QW!DM+1YZE7>1Sw}u(`q+xu&F0xw6{cSKAn#kzDHi1sv
z?3KFU3IQ)zjyUZ=bGzu$c1=x9g?&JqiXO>@3l|0*swh$Ju(umg*M!GT1--O%1yMGX
z(?WPA$BAx_;cW=AITj{%yAHI5?OqX)fViSKzHs}~gtRnT$KBE~w`8D2Me?{&eMiSK
z3~klGP;G5(FCNHo!BnRKSZ|lPBIU6{Zr{cb7OJpZEGg*1Z5Aeho?+hHr(l35=dO$x
ziF#gmns(!(qvNQC_K98suQEL&W1u-#f+1+YXdgVnI4rs%2<K$K@tiW-0&g@$&m-o)
zZe&SD3Z$TSS}pUIV<oSS_d_dcbOCskF5}_(w4&Vc!W*Mh?XS0k_8hg9`yHgQ0{RYv
zwvntdAY)az7A-pi+lk|_(%s>*W{Xe`mslrNSD$QShIp?V4XAkJ5f8uLoq$X4O1ljD
z;g{H0+3MSpj!sVaVew!MHSWyxx|n$XxP+QlvrY)TXMwPG@Dd<)Noa08X=7WLb7R!M
zYUXbGqqfW67p&Ig-A$m8Aed+N%frU)M3XB}32teO1O5{b7#2{PhOe&$!!Z>;u(?14
zAR>rn7|OosF5CS88Ce?W0-F3G-IbKeoA{pD-aeBo3$)0+d6ykw@p*R-dhF3#!m_W0
zt63W>WohON|9N73b*<!23uw{l+Xge$3PwtME#JR$C(T+@#%PF&`vjg<l1}PWQ&YQ6
zLBT?OA90yO)Prr_pc=@vnARx!){PqxF)_r$kace@6P07IxHT8a|A&S}LlNW)N|qoz
zX~Cpez0PWOR_NO60DQm4_59DBZ^qX`E_y2W$iVY{x{<*$8kzcA;1YTn1sD!pyuf;J
z)(m36wVUsBz#UFGrjXUsWg$1z+R+iUZ~^tAyd>4fG!5(kU@gYWIpL*<+4JZ5LWUXx
zH;#PLnw9gD#CDaF+d@d9TC~UJ#UDXIxnSE!Y5dumODq;7j`N{mh!Yb=+(o~>zY4P*
zQg>Fm^qunG=fNTy9lg~l%mQKrn3$Mg9hiAJF8YwOK5E)xc(IL1BO+o?^mjoSI@`K%
zk{z6o-xty!Mdp?5FN}@~`$@{m$jmsZ4ufUPkp|)QnVAlvn;Nm7d*g7;KSf$og9!<5
zWJoM3gmTIB1C6VclX8BY-&ZULmLto-5LpajQ#82cRO6-aPS+VqHXzk<z7LDxkpT~W
zr|Qb0lOY(&Z6mviEWlUl>a+vPipB((I-`Lwj0VTYZ|Q4vS86J<<IxL2Sxi)PYBmSu
zK4#|Gyd!9vL}L(S%E`{2DJu>FxdS?hX!N>Y{WHxYTelN*qKW}a>TMw?M}~_223Mt!
zLs_oFN&gEWU}*9=E+wPUX#ccM|J0O}49U+ikcA$RO6r6E5STKlnDZlKjxK;2kZC7?
zu)F1Qf1K4bHHS=_SMT9V*-K0}wfc!^pLIE6z@2OcdSf5{vbP+Pd7vK@CV>mG*p6YN
zKw3^XjseQj^itZdaTpr1vuo2WeIleoyftaKQ1*A$<kVLKSW9n+`E72Ger$Knw%em#
zNmqaNhDpYm5JE~e=6qrtyhVa`wGufOi}O~MB6s1hO|=tCltTC5dm&(*O1inZ;Szxx
zq7<AOHgP9ES_TKxKok8unggrUwRvykz#$CHmMEJlIn+pdvwuKz1t9Lc;BX_LxCw(u
zrga+qUfh}$(E-(~?dXJx9!$09XlJ&{$ERopJ+F@M1Z{!P6Zj4$7(`WoZbT+eNL$bV
z^!(9JbBN!ahe4hy0Ein@7mc@~wg+>wLiW4T37rlO0E_@5ZELe4Tx+jAhMz=ukf8jf
zsX=uA<NPG>;yUdbdjt7<*P%IQ{D1hlcy}Pgo&p$sAt9mIvZJ3s1Y7$y2V|iIp|{P>
za|0Hr#J}&NuwOqj+^aX-<~MI0G%qd9?L@_AXS*=q_4Zy?^#0*@<bzKL;}GYSm8s|U
zVPj+4T*%9)KG6UYi~HhW?6|)-FwnumEEsgR?HTO#@i*AnLji<af+F#wW8gOkE?!>x
z@;CAQkC8QX6VT+d>!RhrS-g!oJT%l*ZNdv2iQPT{_hXkupm`Tlg9b<%2oy{{omKS;
zldyuXGZ?tIxL8;~2YIu_<;>p8moEKYe2r6k1`?>v<$kFXVVm07ap668W$qWofowt%
zWpJ%S%S&L=3o><45acN1np2ZFi3Sb93!u(s^bdFl@^9%HHs13myH(DOnF~p6XqP(~
z7P#GQZS+#AfXIfL8Y!=W(x-)f>JZ-JM}~#?)#+?d&9WX4DL_57BS7ww`?uh6_#aeA
z%sC`~A%5vUul|3(O?d?rOkaB%Nfd7!kJlY4!oKrd2K3B;_HmJ0-}Cp#;YHuU!CU(J
zY5aF1$p&&<R+A7@Q!4Xb*dNF5c9GsoqkV!LqsW^lR&+`BxqWhiBD`BjI^1`NU(MVX
z+UR7oyhPj;Om}+9d=clozU|I02aZ+kABkFgQ$9Uq+xs|pDNk={3|r5)sRWI~%fEh7
z^F)7Zz<1m9w-=_Xry24^sKQ<LZp{P-%=h902?m4v7vL{|8j@`bcxX8a;{`4s?_ysx
z^^B83sC`}!ernqGn>8If9^D-NV-YC%fB!6QBev0e>!GLfqf-)WgO3Gd`^}ClZowEl
zC!6G_gvt>-Ze?Pk>np6L)^bv!O+ye3zdG!`!_OU!b~^a8lT_hUNj}8nx|?*kn$RHo
zv9DTaw{CMkqh5H*s~m&I_MuJveBx3q>YMn;x6x3j86Wn^$`Q7&6TdlT{LvON=h_Pz
zDQ`sMu#i2;#XnvR<^F)nqCuDcz^vG1w~IVddjlqQ*e&Bracp$3uiZBDr>(St9+sB|
zz=j*4yfvlp&Dw|jVecN3uif+Pd-KBBrGrCw5tB96A#7gyIJbh$_C1v!ywHCQZ~;n#
z=O6_B%bxMzvi0`XWLpYvz(8Gq+`vlIvK=;(3D+cv49j{tEEY9W9fk83?d1pC4$y8I
z)Ar1GyQGMr_$NzY`7CCtheY_Wv#;u?a+ac4ArbGO>@}1aaHpcIN$`{WGEUTd`V+nD
zHQ__~-+6g5A*!R-n>Ch<n|5bma$`bqvXN#EO>}cWM;l`0Mx#7`v>k$WB`RLo@IflH
zM#DNXooFIt((C68(ap#{V+NYB`d$r^AJzK`Slyc@qa`0bAW5rjY|{_ioP(XIZImCD
zf`l8ZTG(ZIW5Q_o<{*C%Lfy5B+vNG{og);M_%7<vtUOw!tM!#*S$S(JZ(cMO9{J-+
zJp&$PHT>_3DvhybanJ7x=LKW1tkf+R0v0;RgMSkhOV7Mrllj=kB)}-?3b`)sgvfH6
zBgP8}?A~L1ZB~LZiSB&aHFS;V92zqXBvbCcFX_vSJG7GGIPRha3nHaN7F2s~?(N1o
z%K@#u5wT^nv2R=U62*Z515;v|wP6S9OU{lKMXQ@fD#Y-Cx=(;kppmvx&;35>CCngT
zk3L7MRgkCe%RbIwdX$=7Yj7}<;+eK0IEV07aVa=HWxg65&$;EWakDr|9^pI+2)Pe3
zUtlmR9@A7WnOKDTpydkBx=#V5!VfQa9NAfPu4{F}x28x*>>Lp~<RfXU&UgO=Q}hR>
zt8Mr`8o*$1qcGex-hYU-D23tW?`sHZ!b`$x6#9<}*sr`IrBcD|FlSTXRr6pM+K``k
zeR{Jo;xHdFfM}6(ENKQ%t<VXe8_@lIziTy$SF$GXg<1nga`OgU38OKKS^(4o{^LqO
z+C0ySBm25(e2H5@LVZk?*`heAziJhC50N1Oap%15^8uhN_l?!I8o|ML;(t8nioc96
zvC(&s!<f3betw9s;d%UboOXF$>GP)(dov_}@UtaRn305)`ZZ$4igvc_yIN1IwBG2_
z{r8&4K97_BOb^3s98Y5x)1Pjd!+y>BU3eEy4BmbH#KUH5_PI{W$cK)l31l-j7?-Yo
z<*+2t*Jk$dA&#wTA`*p%NqOO<e^cVLuVPb>6c3wIu8x@Dt8Awm>a@5#W*tgnBahpo
z69yjkcPrtu&bB*Ep3K$Tn3b6(^@3b^0Z-<n!q{3OvHeHh&2vpK=IQRFKO~4=McAU*
z@IGra#>sf4)EM_xT9@R<0++9!PKAc-7ucmg07g1cAD35;zC4Gp6}jOle^H&O>~HoT
z2}1{t#VisaE1pLO_9@+lAUe<GdOL7xnU3sPFM9%(5<%rR<0W5ex7+m5yw-MpmfKEQ
z*I-c!&rwb1-!L4+@DmJv09*mPAiC@|$-NLvcX@@&)`6$F0&i8feZtu_t-ZG-M6u=e
z9tAF3lS+v{r`RcoQ=`0Nie2_r>3}PRp!+o^=snkqEZtb_W_wrT(l@pTd=&@EAy3k3
z8do)0l$0+R0WP>$HPuM^?vP?sCo}_CYH;{RrFtB_1qYgC=>!&mIbxBX_5BD?!gbth
zDF<adUd5C(6ShlStJA2NFCX3$jn1eMMq7<vnL|#>_lC-ZcW0u*WR+-i7Wdq`=^50o
z-pN0kz=IHhl+7tIRmOIF<J$YQ`!tii#KsH+e<z{HbCIH?P~3wMVR_rU*4y_EZeCS%
zurTQqw|5(Ej%vJCoREn^nOi!~CXuZSjA`YJMl%ktxP8T4xmT0GE&;#^OYI}qiAT=v
zubd11SR55ohE}GL+(WJd+kFYfvB<UkK5Vy=vE0^LUk*1+LlnuM3T2?n`S;baXHYzE
zKm-k8Jgm?J*L-NX$kSGPPbUTt#@&g)$*83ly3TkT!_=|2#-u$7#)1x--$e?pnpz(4
zlcyeo&#=KlO07^q_HI3Wh4J%@S?u`WCsy%_yMtI@^DUJ2)Ot?3wp4w{qL7{-(`@dF
z0GKGfGv*8`PxCnm$vvWO!GpHCeBrW#skMkeyQm8jKuqg)O{uE^_w*N}2}tLOsvz^S
zKZmzwZxCIL-qDl?sZmm7ZNs~)%z|q<rx4Z-7)03#Q+$*=2IQcRZ(VO6`3PJ6USm1M
zv{#_`1m^kw{AyCk{o%gBB>o)rPzru>&z)EFQih+$MbcJ_uu<QGp48dl&Zvd6F;|33
zJp&5C=JU)2=vY8KoW%2zVn2XJPTl>l&m}3qUtX2rw8M`Ahc#ZRmhF{5rSibE0?pwp
zp~~^E#Mh^L|3ynUbl323_50<({~@-wV+|vVX<@jv@5on-<R=F2H@|)7r$RKEwWq)}
zx`WXS;|)ZJ*wWF0)3VHsV&rITtA?U#z47UU{zV$Dk8RtW;ZHYAtDGuh-O7yD>L?oI
zipfV10%?ir(m}Qwcg_V-H?UQZxxUBJDrT(c>rGjLQAo%K^GTW3Kibn*v|~vv_Vu@Q
zID+1>A*W%1XjNp3cxlnd2$L=2w^m#~$?mMt<mYN%1(p&9dL3sPPR4XofcQkE!qd<9
zEHs$@A2OeOv4mFgRDJs{bjfX1hf;DEPB0wVb@b5w$aIyY5`XpZyQQ_swO>O~1zk(l
zrpji)+D7(mLrq!CAT`MPjK<0zksWFkCp|%wt9~y`K_Tf*E0ZUhF(lXJz^P5~@E-D<
zrMu&de?mn=EN~J;?ZIp;B>((i@AoPP|CrUX(0A59>Dcp{cU(OqcN+WJa=9{-<pb<b
z<1pl}%s^RZ`n9JH<+I<0ezPNbkW=yc_oC=xapAQ)_|X^NG<kdNjM94gN41bIbo4~o
zc={)mglF2EPROWd+i$AGG6C1qUU~-Tc3W1jcnnjt^Dz`r<(%fbC*RksHG9%7A8C2|
zqnokVQ!$oGYRx`ecM%~HuZ(@Y|CIy0n-ZE&J474)6Msq1iKl1?6^F5tc8j!p$3SNm
zA<5|Us)_CC9^2rl>Ea${R_!}??!x%{+9)w&Hm<=co*nWGNlFiIw^?lUf}_gqf8233
z#Eb?18FD2ecZj<N5LHkNzrQ<rk@mP=&VV4SHcH4UcAR-=u|mTWZ8@-CPCX>vv}8@C
z43b0Uyr3Xj-C^G~OY-q4t9Qu5*i*--Sv_?a>$S8{KU@|6EBdQFGaGJLMHabstLPN;
zYJ?EQR_#RdBQRgsRD3J-9XT3L^r-5nyu-2uPW$!^j}5GYd>0lB2^KNekfnzrC<@ep
z=hOaDhgLDmAEJ~7N{0M_A=T=!mjmX{Qff6Q#61+R3lWCGt~71G6)3ON)G0W+%OhV|
z%Q%f~WO}Ob+ZDNZe4)#hD?eD?a-1l23MYG)P|MX6G)~0DujWBmp0dhEKpODrRJ)Hs
zNDAYtw_>^<g?Uyts^9b@DX$HOn&%)=-JxZ>&=gef4c#3^kjdw0@g7VNuVWQ*p00-<
zQJ>LG+z0+#415<1p|Z+1!I}ZgYxQ9>?tGX3^&s#ZqL&`K60TqM4<fM*(Vc>xCeb1u
z1w(!UPD8Qdq>3-}F3AGFfn9oPOTE@6Pyc)R96>a??cKw@Q<5r?%b1LMnl266WxCqL
zYr6`1CRmG}2Vzj{of(WgT}m9ChBG1V7H`10xhG;pW^tRe!g(^MyW^Qg!DN~E>Qp<@
z8)E+m8tOhwexoie`Dx136Dve=n3WE#MhH4%K!tSm6!Zw4ZA)CrdMuw`+r3l7cfE%O
zq0MpKAw@rUnNQ@C^)zk~U%)v-*q)0T%f(lAEsC)joqiYO%~aj_g6X~<*x?*Pkpw|a
zSQDKk%!`;G=9N?~e9vqT5x>H{kuLs|^kz3wwun+$yJ(v;@IO!Loi3J7GhzxBJ7i7G
zqjPXhBpbiZJDeHRz5lIo@OdGNj!!bLW=vCBlc7^u`9ctlF?}m;Y%N?*&M06V$c>9!
zuTe@_C4+-orJn!ZPQJeF{zNl);rA-M02+wwl0V3!!&TYI^XN1Ne9~)*{Z*w)T(s>0
zYi+KUTZvt#+ZJQm^mH6-2w@T4D&inqPJya(hSo!7^%;*ALA)?Xb0bQL%R&xazSIUu
zIio<z)5$Bq-X>H6NM4T$$ufjiMm>x=H8X9~=TEK?ynPs^8>xU^z1I{z{+?_l#iX*L
zYw(}c+5!+0W8^7<d8FEUV!Wr@$M-vkmZyn>IvF009Dk4H#iM<&y`x@nCVREss55|D
z0wa%b_T0x!r_9SXol4G1y_ID<3!aY|e{b~T|3!H~WW}PdUZ*T9?tA+r-~F7cJFi14
z(7O8#Oen9l?VgouXFagOEh`QlxOOIO!-?hY`ESRnm&C6QyXZ@UVuj)9ma*|ZH9wC=
zg;X4ZW-W&tx<zBqlFlsJmAv-S=uCOgt5GiutvO3y|GeJk>vd9O^4vc4$&VCT)B&;f
zL(nhy`&qF5RU))K5xf0uH7UAEN6}lg1%J5UhzwuRZSWnObN>J%f41zaaj70Ss@ZT!
zU&raD{)Pau4~q17B;Pwtj%ED_3<nQUONe`C>GnOEi`T+)3yVt{Es6t0-4OAmyY%uv
zo+o?q#TnA<T4NQDhm6R|@o=%M@NSGDN^n~)$;_#h;x_`jKBzpw!B~?(3=4=)zd}lS
zzY1q4_#f^Kw5SKxuN>U8XE=D-$ZqdQNIm7$RuNkf%wgJ5#%0X3py(vbEzSKL`&m0j
zw?5Wv)Dk(Y$eo|o;j`!}N1J<nIPOWmoCEv`Xb=CL@9^B)mQ=82PvW$f=P!Kbs5bDT
zTPh>0suFHMUodiNn^Q}BrcMp_&owdc<?pvrHiHV+-EGFZ-r~qkSiy}M6x-nKH4Y<z
zX}ibqtQXSUlemaaR$}|YW8<A?(X8O@nlnbX{(F>f=#{wrWL6Cs%HaVg*)R_L_HCN<
z$%K+8BMeT$Wy+D-1&CZ7RQ~Tl0~?BTCrn<Qa>A&|xVAvosS*5jta2wO4eCv>Zs(e~
zh_%?ce&FXxAiERUkuO_Nh%oWP1JK@Z!#}S81g<HPh#*3=Uh(M@!^9<rH|h>;V6mcb
z@I(&Ez?*;&TJ($VkA1b4JTJsy*XM1ArlJsOzL$SMyzDb5lxoldwy4SI{bmbYyJE{y
zd)D9b$Pt3a2W4A`S!)C31mWd}t6ZRQaaxj|wBmDkr#1N^Mp9tk*AX|YKJuEUY4p@Q
z`RFNHT9wE#b(mrrLNPI>PR@ihu3sodA5+G7=eW9#$HIVE2Pd9sOAh#|*bMb_X=-+U
ze*h~D=irW4r0d+>>?o&EXNBk$WK*#q*h{#IONg5b?-#YH=vS>RV}m|wbV*hr_FU89
zBb3!2nCdTYA<$>RU3IZ?lSI6#-UeJcrZjpt;`Y+F@^3AWBpl*%A!o>pEdM#ZuY~3%
zN4J&jZz%<G2X<9jgxoZcVHRBT@`^e4@ACce%sDV7R-F4F>Jj{bK1Omt_l^l_9ATpF
zai7LRn(>$Atd<B*ol>ZA+UDT#5?==7N8|(Jw4ep8Vp4w^O~w`FH}pzTqI}#)xcZa+
zN`PhjRud*5xk%)BD#cqG0#=hxpSV%;AbNut!pFqLJZuVaaRT^TQ1QXYw5h0Sc-6aY
zvbJ$hpHAPhQOA)ri<bO(t#UNe@ObAP(e~UnZ6j!+E})k6dx@(>8r~i3OE`xLK(08?
z{j|^H9BJ2;>6wcyv~k+=k%NuqgMX~_sCk%?|CgtII&t8)gJs^csFb`r4u=JKHROnr
z6&c4^g9AUmK%92F-lBR|K94;JfiDgnJX;HPY0$&I2fy;XJ{<u!sD?hQm+BptEF9oT
z`#11H##5bMAG${u($xOPDuv}gUh2ZIXNNx&sw^%1V*MUD={;}Ye+@QLYLSdLUbh#G
zb#sp%#lLZzjXR4XgByT4d_?K7_n?JMQh~o-UCFGkf<0;azkkztkPbe^5DFvzb8|+V
zE3M!K5k4V8%|AJvS|rswy1g{!;BnFfc<OfrI}M!p1VZR?9O4%c8`f>`K2H19{hA|o
z4)8g~*CF#>T74%(%_D9ErXYr&JwTo`oDS|>lbmc-{Xt{7HpQV%CN;|HB@82CxBGyi
zv(D1h6Rw~8(U7$vNkFG@D3&u9%KagtfedU8o7k`%!IpMe1@^X_I@DA@0%`E5Ewo;x
zfMK~;TH#txdcl0>)!?7ojF<N5<Vu0Jspbbe18a!17gZn7gpDARf`9bD$W)t84^Ot8
z9Z~a)`Q%CmS116{C%(;xUE}*hLr(^tJA+UQt!qo*NnJBKZd%4Aq!j1PjOtYX3Wj+h
zGHnu<<H6!0FLE_p?gNf+#B-Eq?VG|H*ACOodD%oa&4PV>L*d61ULKbK>M7XJsPS<V
zC)pUnz*)?8#wVd<`WJkX69L6X%Re#P8bx0`{5zcWoG+t7z?wtx#Digu%l06FRM}$#
zVQAQVVmf-=uHux$7uYJnilXROyt#`gTcV^WI`F2U@+>>+##}d)4yQj+i5FLZG~HO7
zj<fqbJG)|!Pjo@7$N^A|<h1o?+zF6PJzxhz(DOZwr21<4Co)o=8e`WW60DDQ_XW{|
z71#*70M7;D607Yr)<10Do=`$QbPGQ8D-*0L43FiSih>O_PW$9h3oF?Rg0W1-Y5Idf
ze^0F6llzK8HgR;w^ThD{>I=jgEbm+NBzFy&pXSD2<N185rA28cH}F0ml+YoG>eUSq
z-L&~p!!Q4y2lbR3_UZ)-^47M7nF3-T6U|;1`g|_!lQ!wn<IF6pi^@y>hu_~3ea=<g
zcH4P2k*#=j5+x#m-0_&B7dqzXrMt}mHO+4cqaAYT@?9vayt|q??qQPiIy5RSGcZ#X
zDdUlr6KT;BQ{tgQ-okd%`B)qVY>XtYgPO6@8)wN@vj^H}6!%d1^qj4q=!V8KiAvV}
zyUjXhuOY7>C-fuH*VSd!t5cO_a}+HpKs0htu_dcc@vfk0K6bWi@}H1m`Tn=~(LW5D
z_Pn=+#3*A!F9>pmgR{@iUQWO1FlXza6nu{$3EgGGJ)$sZ7UE`d>((&Q$h;;+^&1~-
zUfX!C2Pk~(Apa6>9yl~f3!al>6?y^3AK!8_g3h=n0l5O(2ge7Zp6S-NsQ|g>ggU(u
zrM%zIzd2xSouyO-J~TnA+Y$$ecYHSHiWfL?eaS5C`Iheh8Fb%)5!$!D{}aEpbaz12
zqu^*{YGEy;J=&5+U9<mI_V9!4??H-9@_be<_9!|mI7!>W5z*))1MZ(y1ktuTDrSf<
zZ<jWYG>>yk4Lzc8%3b|>6jrOLy6}-BoJ`!fp~L1f$S348vVG3nine`8Hm>fc-+G8d
zTsT)MN#;;KR7G1Mm>IC#ujJln;?`#aPG;Vw5gj+?0qjQUAL~iZ4PDfC9E4~o<l=8t
z9?eh8xlx3ioBK3*CAlPd>k|!TgU1;pYy;kyrAW7rLru2sU~{D=)J{YM6b1gw4>$g3
zQI90F8|%TKrTFWm?Xcxye&<GP6zLZ@$p1vy{u7LZFu;ETum6cp|0fpzAKbzJFYt_U
z4rsK&!pz*WxCMCzKY#w5kP#FV)YQ}jXv#v|+Ed6~^JRttX|8usza&7}M5Voea=}`;
z_;`4Bm@&YA4-XGRLWe6eDjo~89+YhR`Z`j103b(rx?E6sWE9|+uGb+&7jlpwrOsup
zJ!Y5vSO`)W$p1j&DS+6zwx{-&XyxVtQcD#ZH5d+(aGaXyv~J7Xd`ebUd3m`?79be|
z0s=+{0{dM7kc&ohSN}?dhT#uiq2en+%SQ$DL76c?qx+~USV4LSk?_x;)&~Gi#KXgD
z%t}#Rh7{|_NO<7K3NES;BJUuRpJ{6Yh)Ehp3KBZTRz60l09e#)h6fUfAz{XV{{m7f
z0%AQRfD6^w;e#BpsUH@Rm+|o}D>VU43b3jisANKd?1{LgqoX6EG{BmF(zmQZHkc%1
zP*WVdSIUJF**cw`LQ-dc<W_2RwJ<FsE#0WgVrnMr_jA}bz;D0^2GCjv7G`SmPb;KE
zw=2m$%oO`r9Yq2g;*M;H5wvnmjxd7n$ga0%j9j{W`L(^hxC>y4zRI6RK{7tf^ep5H
z&{sh=1^=Psgt)$*9<_`wAZU%K@lZ;L8+1Gm5aMN~8Q>DU52#zwXh<k}t_eVUWCmWL
zsrO))=rx>!han4LL{>Ss&ju>!WFiJFX)`2N`B<K&j3L?9pLL+hO5ZJ1IuWvyHGr#5
z)0?(cjWu};=l;6@jkbqe+dgtsIv(=-86dGVdIOm2D`^D_z<Ej+wL!<S%!gnBS&2Y1
zJRr9sb!Kep6eubYb|($ehlc_Ge+QWscK)|-T}2tK-D76xA_GVVJQJ^e`uCvTxXfRu
zXzENCz&odxzRp1X{o$avqGA}JjUyw1u*ReyD2RX4r+H^{6F_Zq5En~9wyGa(%c6jg
z=;O>(prt>6<%$4YZb&Tb(hE0$Y<67_gY54AL)Ci+V!i+G<9F+5Qb~)*X(*DG>~S<y
zLUth<r>yLid3P!ap|Y}NhLG&tAVu~n>n^fa_PpJ{>*>7T@6Y#l{yVwv`}KOR$MZ3+
z>w1JdR;~E``^|Mk0wgbRj~@8qMQ?w9uJwSr*Qtbq8$Jet$&s|2QYMd)yphV#OiD~N
z0+3mcyI}(OTv<L4IV=(>_bdpMYVefi>HONPOZ$LJi|6)dW4d>Nc!h%>0ln5(wQUnk
zNmJbs6xpl|lmn1)p}dofR(plXwp^=x3t;9Nf5QJfNr@;X!%sr%=YjtIe$Y|p_*K7v
zTOK8XjSGU2v>VxPnu0<?0e^aP{G+N1?nk;lmLqE;<p3XkS2__WwA|)fn?v>{9V~=?
zkJf66FXUZNUHNw@@Gud^t3qJL&w3EP*Scl*-V&cTy&w2IEERyZ2N(5MV8t|5u}A**
zXG1{mFn*9adbk<_{G)<`g39<;Qpzr1a;g(f+~^{JV*GHraPBHB<HFK7F7OHAly)yf
z&5?a|1{lvA7mD<ir7<*Q9CYO}=5E`D;?KP?GKc92S4)53nEu+joN|kV$@+eF2(?H6
zLD5&sOFkn=sJj0f51_?;{r!OQo*R{AKNsZ=kQ@kNm(<Bxs>{pETU%RCf`N~R<G;u=
z<iP&#;n@Cpxw&9QRZ1m{eDY+ki3IenRfZDrdg;A`!ox=}|6nf7%1*{KWIzJ&|CtFV
zrvjroSR*g#ghTRQTbr5!K-2OiKK?Z>j1xcHQ5AAu^@`?WbYuVX|5p6#*RNI?<*LGc
zTN42E{@-fP1E$!#w5P95vd05V9?6Zx2K@AY-&~9i_6DKLVym)zJR4YCyp#waXGm}@
zDA=GzXO9P3JBp1AP>mv;LuPjhZ54Bj#?K~z8N=bKDzE915En-%PFZUy2ZYG}D+DSQ
zGlW*Vd^~@u`TX(YBJjTdcY*-%PnO4C7TFT-mA=~^##H3VvIrUSR3>v+`Q){gkB;L!
zZ{36rr8)a3(rn~@V7>t``x)+rosY)J2?=HT<B#?4mfeFo3#?2vE)ETiCbUkm6jind
zX$}klr5<R}{GT~U7hE3)BdWJ|A8^_J`;TyPaD4rrEir?7!qs4MW|WoHu|NC2H#KE}
zj)6+IDk+KlxP%=5Rc(;eaAGVkD@#H?0mMzvPP9!a5z6G<T?IkA8GA-%CD8W^Y_X~U
z@Yp>NM^an54ql&)Itqk3R->gw46pFO$6`C^EIyNDcKVl7*hn^7csa1H29bdv75eHq
zje;$*h4uu5A@}8!hNZvngsH|kyBeIOvSGu9u8mz2>({S8eM{L9yk`r3%m+De9D!H?
z@UeV@@yA~Wx{C3?WHr=oT38&@V5YnROpo$z6Br^F`_3n+<B{`2v_;YjhcOi+7hq9n
z@wH>u{M=ke8@RCKB!_~JM~)Fk0XV-H_jJ3%2&nzrICFc7W&{>JfU4$5=Xe0i2YH(l
zqw=PvDJ~zEMTC-t^D4o`&X|y*D~jaj^kg{?WLdOY)MKtyY*;%+2vd!apNU1|@USm%
zhR{PuYwI|TZp*l;pzT$Oi3Fj%%mhe&f-(or<1Cr&(vFT250OlRv06C++^~27Je=bl
zNB!r|pCJpRZo^82tYBlv@_RPqxC`3@7v943V_^av6U9dIeU=AQH#G?zIdTNHeotT{
zOamk}XyAVpm32QU0WuepGF!0hp3Ty%+Y=QjFXJHXTr1M_V9AT=p14(q9+WTuVA_8#
zI>@=JQ19Bc-Si`zp2nHCbI-$sW9lG?GoA<53`yFV6%(s)Cn!Y1VQRQ7&&~sqF-=WP
zdU|^325Ygs?k$*iu+?bSEp&02wCvK(PSp#u+h2N8zTI)WV&OL(O3Agcee$rkb~6<c
zeLbKa8vBR;L9yCSR#Yk1an^!7K#D3U1sRjw#9l>NEiTM*w4BJn-<-Eoen^w)lkHB*
zYA*|v_lXO479}6OwDeK)>EX6Cy{hhSvDg;nT38ZavitQh?t^+K<-?q&zAqf1yyIS4
z*LNKq9aGDM^(sSfyL~1fzE3{9T08&lQM_@7>+E%)-j+t8UZ5jo*?rN=co@ils`0Hj
zEjSZ!*7*6)XxP9vxpCvhU7}#n6pJ~S@PRw|ZTCqn9i2+W2ky}F+GcU5?pJM$-0XEx
zwU9YnLqUp8efirm6`ZhAsqY5|U#oCYpmattBP1cT@Ed8(Vcq4|dduL7_$^0A@vwIN
z6JzCnUBONX`F=vgVQLVXV&hw_o(oSBF_CyvzBm*3X1XDBkl+iMqkKaF<J~4Ln5NLs
zU2x7Mp3B$xxapWMV$rr1hw|HQdtvUy%?u6>j*H%BnPPKVSgBm>>>k6HtlY=grN4+1
zMH_5R2-x(v_N_PXPiz`6KUvO`BJU6~XF@9+s%Lsvh)EPaR#OL7jLGmrFb>XE{he4^
zn|t?y_$aQ8S(lT#y!Nur0`lLri<~7e`<3BR;assRisMZU4HL}z9T@a!7PJ05r}<0_
zJ`eM3b|TWd28OGl3({Mlp;d%FKlS+VrTltzYOH!92HKVNaI-LV(UI)rLPrs)i&|*Y
zV&3m@Tu1b;5|Q6C3QjQevsah(s$w@V8^Ncr-xeRdNg@!VZe1m!Jo;9vsKd*vqNC%6
zvxBfifPil0f49&@t>Ex*Q^`rVxjktO&v--T;0%qmtf;obcH?NpVQ7qIsVUtKtLx--
zlXcgNq{^>g1nu}!_zZ*v+sr8n6aC{`)8oN{5N?ZMRSJ3LbuiPU(Xroynj6110>~>Q
zlwmt3=6iZIPLJFG(Q)#GmsY+KoU#6|#q)%&vG)C!T&q518~`%^v)7yJ;cMsKJP|rK
z=GHe8qoNDp712rfls#*gRUakqMR+tpb3ZZNvd8*9I-@&y^hTb9ROoUfuHxy!8P%%_
zt4OVoz#h+Wnzw1cPc$@&th+_p2)8p_g-bdP^Zv#Z$8e72rFvrMx9Z@bj-)JSRulW1
z3f}^X7P`pEyE}Tr`GpNQ`d1&R(0Y6I*H(Ad>)GGD=LPrU8a0yfbnXajdc2sqdnBth
zVcLcFYIpH|IEr0^Cx5{UjQY&jwcnqr)j!d3$45bs$F8T|B=VS+&jGQ?Wmd#jo<AZe
z$hPaz@bEC}i}Y;knwJ+O0%5DQm4p?x3Fv%HmG=>#&X!|NDxs(ma@}6xZ%~Xy@a3KO
zauS>p#R3=Q$?MPSecTrPVr)cAY65AJlpe?sVtFdGs&#6e$fa7Eu7f7__!{xw`O#R~
zOF_npt;~$+z1Q6Qe3U)B3-!2;%S(z*nFbbpwJ{4vhHimgsBlQZX(>^CkIBR9ub|^I
zP}Rt(yL2J`c}+Nx4p4L5gAUOax9ff_%#PzV<wE9aL*CzdF=fPpFYZEx_}vO<)10^@
zWq&L#d*K3Kq`g5zHOP+15*+cwh{56<)-4R-iuD95BKjvkqJ^(<FWy@8ujt)~-OPCn
zSD)w~H4@}pw7G2dV7}P@<ucpjJ+=^TgiWv;`@%E+F+ISlpG1~ld<Kf+St1|;N%>Ze
zD7Bx>*KDezZ9OydSr^=kh=xr5;BXo5sp|RHQnJ*YCCXN3GkzVH>%*+3OJI5&&Kvcf
z<r1{r#QUfGRrKrh3tkpXe2<eAAU=WqQdt>G?RK4>fFf7-*Vr#<Rg%G6u?4v8)u!vW
zagDUDr!=GPcJFG^>({6CcB}SgRWErJdGNM~vc)5>^z3Y5@IS@HL+mphJcjM>h+q$d
zFDxuLKw=U$?mI_IEA*pwtT1a<7M(=cr_s?^a~sn+gs|gXa;@o0xx&SI6I@?A{J6jd
z^?Z2~Q;Z-=ctR|e2p|!C(9fSWK+E-(>KuXW$gk3%iOEU%ix-KMU{gL08!;uY_YxO^
z&s$~%IMU<WqOIfi^YB>Sr0DO#yxU|#{G#n1@8VWbM7gusd)gZ^oWxfR8G<HM;-zJn
zQO?RL0FqQr{z70ydDzwRPoMUH`~9_wY6EgHsQF*2XdIPhL(LiC^zY&SIg3Yc-O<v#
zeED2yL2X>&D+ZBo*i0Gv^Y9hQHKus>OEokZ9yoD98>wI9Bc0CZWr*bUjW8PP>xqhz
zQKkNrn!kV8ix9_D@XAuA7RJdcej(CpwBV`d-77fZ_QyTry%1N3Ft>FtrTL6@6h*oQ
zf>>4EqKDY%{8koA+;})Sepi~mVS5CbiUo!^niIy^m9O08{rz#&z5vP4o52p1VG<6w
zQ?*w#pK}bUWs2X-_ju$s0=*!39|2mcr1Kuju~H7;_<1u+!_67qsH$G}l1MIm)ZgQI
z_NKkP=&lZ;pd%aHcE?XI?92NKf7)CUt;{c?;A>^&C%t&&V8Z|6<)ar}#$~x-MFfAb
z^%}~SGfcaa7x|plsae39=s(U|(_BSf)8r(GfRjgrE=s)7&UyYI_}6jB_(Asv(ev#)
z8uRCEz18QoChToHN1`G2KcuxD4{WAnD^AGO{VD0Eb=on{O6Z?`;VoZbRyEDyQbs~!
zVt9h;{BQDVB#Xn#IHgx5Y4!8NDcj+NF7W&=(ifn=sLhigpx|U|Hm(2MpKSyR7{=r!
z&#a-e@+8TyNd&gCy+XszqqeN+f!KLrVZ*|)4<4AlPNBXP(Gx)oKXy6jS9VrblWd}K
zbe=5K-<~OQscnrb93J*&1h3w)togjQ3+0_tZT=g=zrOm9yQR=yVs4$=WskS!jBA4H
zXqjN~foMql{Fx#v4~}gQ{8{>|w{WzCBgku6KrB%o3QQLv!qc!VU^m4qJpIKm+|FSp
zRc*sc(3ed({RBbbcRlv?tK&<Niy}#o-w-G2M9Kv^RT<H5m77~m#uc7WQeD_5#B#6S
zV}J#d*8eH}{a)eoL2G$LkLTE_s%Y+g1*#*&Qi5*!stAC2rj&Z^OIm_a@OkK7v3dR8
z84+6S$3$t>OA~6AY<r=9?nE5cR>*s;rl#fvOIrOvm0%ptzJm}Hd6GLh)qWYWnu4*0
zuqTp|qsGjd1}p|w`Ny+Yr2AwK0;xbGV>}jd8?>%@RS3_x!ug7Z3AY++_!lrY%~$g5
z*D<V=zo14`B?O<eqtDCPWFfz|xw$)K-$>jxQ&hgY3(@<l!t8GHTRJ)>X(kun?3c3!
zj|}3M!j53ONwvj$Odn=v_nw%h8l{DqXvRmU|DlULENPVPqOU~sIy_fKbaVu*pg7-v
zCy_h@!)d`1!5`hn4_j%mC*Dig(R>gFATGn~?f}s>Dkz|C&j>~EpAjbt*ZHKBL55YW
zKC%DC`wt5PT~X_pYTW3FJe9}UKz%eT4Et6$-VPC+iYw<0%4=&`gV}~_^pZmY0|SL^
z$ZpW{%&cx{8K`HfR_HJ!?}77_JkxUQJxRZKM*mrB@fo@#C*$_+69BVvt2HM(d%1DM
zpCH<Hyd&{_+F32L{uTPi&EuC*q-y-LHb?88M#{76tnw-Y;mf)-v0Gt_w;dox_BR-y
zN|qnZ5}-?k8~KvO(l>~*fX0IH$<VkjyLrKcjQ+WzOIy-$<e13pje1AK3PO)~U*PCR
z*<cGgHtMhItJizxa4&OH(+<jSR>-cJmvA-FbC{~PMxJ?p{}9{r;coWrGq){|Df_(@
zT(3t`KCZP!of)rQo3T`%><}d*#LaMZqK8v@63LO+tzBD{?MJyU-eiJGhoQn*URA>H
zU4%T|=jLWjj<+^zQ|jUcG?B=Wo-8@@2ae?7xl`Ko9%tzixI(8b5w4hU#su7cw*wq@
z%l*n^<F__VE2N`cxEi`PV(Von-9_sLjtX<k6OZ|`Wt3?*zQyceE;!Bfd;jWNZQr28
z9p57&Hb2{aG1}r5(Ts49kB@KVy&zCA-*WQjB!qNM@GJ^-buYW*WbT5WINwAc)g#AG
z^O-%8kwV@U=s9~X>k^2V2_hTgwXbM99@DI$(Zbo!_LaQ9@Jv?HR<(70q=>7+t?1YP
z{ff`$%!-6uI08?)kstZhtz=?wKEl27daE|X*6JKTsV`4(hlv-~A$z6A%Uv!phd9aq
z`vKzK3@jqt#Q}nudaW-52l9j-x(?%wJfgOi8iDTFeIh#Y)}2AVnyr1;)uOvzHQzke
znY5!d={?zWW6LS-wW`9#W<r1E?)e&eh)i0o_Fk~_1p$NLx3moR$Sig}LIyZD(}wIR
zY<W2xQ{#h&1t<{*+T^P-NQV`Eevz+`^ld#-2Ko7;UK&X@IU-E)CZPZC4Pu1}uvm96
zi`X+AW3ztC=>yVQ?&a$!2MoRQ*1+$s4atTkQqFbX7RX0%q@<>%oaou@cNo|E*qYZ*
zPi#;jXVy*CCr@@+vzoRRMOw+BRsQMLnB)e-95r8^pfe6PI%2vvM%UiDJ2OxdNs=lu
z5U+7MIF&RJ)G^GtN?*g=co{|PoGq)W{h0)-b<no|+hlW0fC+e022WJbX-PR<AB5Db
zr?m*GJx;0jvBfLHHYwekVr|+=dGi3b#6(^%@)fQ8uY3_|K&$G#GPb*D6(-1t`7hq9
zv9X1(1g`RvBk@@#MiC_jr`x*A^`({`OoT4N-jUo-RfO`w)?D{mPgA0;;fhhh&elX?
zYFcFycJFgj<M?O0pkiF1eL)!e3TbvXqjdm5;$qzShzLW?GqD{#a%5w}@W24VqMW0u
zf8~xQN%(J8yyd)@R`aKas785Xe2T?KGX#juT9_%gP+#9RXa?RIY~QTvyzZ(<-@7Do
zZ>@~Cx&rAA&A%1Cm!03U5q59PAZ9IsRfrbk?4YCvWM`uN5o~Gi$I}7Vh*0h33j+3#
ztwQpo^^b%-ZIri}Q}JZV_w5sw^z=(Z>PjZAaGFM?xo(IkD9Knw8HAycO^HB-W3$kf
z=qut6CW8|Xw4tHjLg3Q(+Rm^sY%SKsds+1l^!4FzM(piqUQXfMT`{mjA`m&M)~g3%
zdi6;uKrLyvj?$2j&nQF)BhTDEmi$UMruZlEf<sn}Uy1dN3^OQxib%)71JiZ)9t0@%
ze;ynf;x|^jZN$8~b5{bN*G-Q1Hk$H|kSfxDCRoF3h-6@!-<9rtvzRGSSu#@(l9G~Q
z)|i)Nei$yA(_r+IG|*z76g@89zNXDxK_mHu+VP`DQv+?qX&m=ebIH~HpFNY{rlK1Z
zcCeCvHd<vr^kr}xgzia3`C}h;b)QfH(Xknp2+3Tx?Xb92m5QQ-HGIWhldgW{zYK=b
zpqv?e>gk1m_wR)w&&2MJPe(3Rc(MTbD@G?{MryG(RxaRhmu0nqoQf=JPNIlx?vvF~
zM`c3Rl)Duf`+C#3xi%WTjibGFyu4_koWw)cG52vnb<|nIHLKu07bBNsBchp(uu<!r
zpuB>D<m@k>+0$u#1G)@*N!HNk!Ykqkxn2)(kj9pKnOn)OXMvKe!CnPZ(~&-fI|~Cv
zivu|V4Zl9Ptjmd1y<~N|GpYfG)^fFS+mc3*f|-L{gCNVaobzH2RWjU7;Z%qN1Pn%d
z5Q-=7qYNE}y+HjTAV8((Wshppce*a4i6J;0Gdw+3F3{ty{8wL81Lr#Ts!C_c=3@vL
zTFweO5#4E?fc;X_mQn##r+E%bDjhtV(qNdFn3&^FvAUO?Pr2u$rKKg8sDf>RvR$`V
z$r^Vd&n)M`)KUxEe&>M0SIgL93?aqeX(@=N2;B!@35j%%Kh9Un*;NITjaswm7DaJk
z_J+vqhkGkr5*>YnmJv7eD5CP(I+S;RA>4p%tq*{vW4GIePi{gNj8Yy_=S>+yP<pWN
z?su*k!6THBl<D|9HBC(vSsQ<Yh@QZc()T!cJ8=MF?Ic3(@!e1DP+V}cZMWlLHc3(!
zOAS7cnqh%;e=t&0YAYjrirM}^8UhuQomE_DdTsqWH1O`|2}`}+p&|KG>>WHrA9`3r
z(9OBYk55B+U;I<-b4jJd#BdYl)4Uqc6JSq5PUZ84C_@NN*@$H!C*x#jXgIDqkqVFK
zX6W|<v;H(;3F%UIzV#GgU!*a5Z$`JDnJA!98T36PUr#gbpF7vbFEjDFLw>x4F;AsV
z_BZUJDsmR2^iKWqeC2f_)97~hx{80!b(PJ`1VpkyU@J#WwG277PH(9&#eLIx$lo=S
zYM&9ujgUFi#KzVP^lnjbLN0JPWO?u3<mQyUcI(zHgcsJwBaf>jHXyOjyrZkQ4l)Iu
zF%Sl;<11CP3CTG=r<!AHU4%qmZd-%}1k6^fpxt3UouGzm(ZPXnL~|2Ez#{RLSDx*N
zBwAX`?zJc)$pgD62jp}*PGs0Q&&|}^BC_~c<(<|UUo~`>QVj2bBM^5(H*LG@^d3>}
z8n*9EZEC5b@7_s^A1x}2z5A^9U{}`nFQj9{{N-se+u@J~o5WkN6cy7-DN;=w9;&a1
zVs8J))&%$FLYuo|AJz%{;?E2uS*2UXp>tDyH{w)iu=3-HckF+}v)rrPv~34cs@U^9
zJUqcsA=PzttZic(rC4f|cC>Gyo|iiB9a|EeRa^y)D=m)CzxxwYy2Bdj$f8jZk>zhv
zyOh$@**iU1PzR?b<cVq|Vvb`K*AI!lj@;G6%u)qy?iGrNMdi%<`J$^_RA_Kh_~Huh
zrcWZ>w(HO2<CGuSL@!c9-2PCGO6}X;_F`5RvnI%}i<`zwpRvQ<uHTT{aU=vbD5{(2
z93E9IH8FCPLMw&&vd)frAykXaOuI%J;cFq7%e}t#-mVfT6m#TZ_r%CUao1jDj^FCJ
z^Jcn#dIfh}zhifxo<+)l^R~m<pS3QnGPnYrecQ$+sqZ1)zWL}(;-A}bR#CMQ&)~W(
znuJ@`N>eVb5U!SEBRePOjvXS+cynWc!-ABEBy=yBCTjCeDf0yjW94>x5_WI6PGVw_
zKM&dw<;Z#>s*$rnwYc5Lwa1h4L$xI44@!Elg!=MAg#GQ0YmuEDp1ZsA*|dLvR9(n(
z-%QA=<y`q0*4;3Cut>d;ndzR&HVt-4^&=b@gu%gL*QXsw1B|o2?K3lXo4N~Bp`T`0
zdEvrk2p5K`ZwwC$tIUX7JC$K0ObLn1{z17ng&wgsdHfxu8Ni-0BGeZ6hUt&OE8W)d
zQ1&3|gc9=Cv_+UMi*ZmC&S+agb2itatISKqM{=$=3K;;?(>bJqY#om{r<NngHx8m+
zEX;+eJC1L_>S&uL{5WFdy5ij$k8);;zFr<bi{p#QTf*1jX8a3ffUt_bm%;^QB8}LO
zOupWxUPF<DjE~EH{9SyDiISc}E|0M9bj(y-xH_AG`RFBV<&dxtd!z%78a6<38h66L
z1vZY-=ARU^*BxamD9tn?A3d)(Ek<3;ci5M31M*-l!bpPnT~`Q!=xc5B&+A09_&M(z
zOoZB&CN4_#CY#`t`iI-tjtL1h)z!5%Gzdcf8J`Ur_hmf{8#O$_P_KvZj4dvhChw&$
zvaunOt=@!0&^+l_;?+zi4LXD(CvaXI92}M^l;y;{DX1yDhM3~Bt;`6{bxBkiwf6Nc
zW@+mZwbc|W_ul+a6ccQl{=G;%4-{Zq!A+?c6`%?WP3Lsq>B^Ri8vPC5p(E)dYA+;`
zWM#WPWJXEgOFl!1cmPGwFfrT1w#~z~$Oh(bipfH>k1W`8-~Z@2hoo_+1VS<r;Z&bI
zj<TiND`u?9B?EJj<&@A{H*Rzw37IosQ<T+i&u4m?4MznHAY#sutA;qzOowGJBfQy(
z$jClwanV`sh1xSL(W~6-1As)s3fgG7buVq9goqk_B;HWckG^{gPHj|$`POaW9Efe~
zNyiGLOBX(=y5qb0E6fe>>|T6JY^s;jX`%)iKthp!VhcoZg<1GtO1v5dv6n2ZL=&2G
z?*|vxFz30K0lLomXqBYzd$#i@)Cax1sqtCUVtuPnLc`r(VXvZuGZ7(Cd}}Sm>Q&ot
z!f&HlcQoYL#l@w_r?wB7E?jvT_~K!HiB!08*<*n(9y>coR{va>bscV}3um~=-VDyk
zsq*x6+9zd4^3RUjNp;ELk|6nEW8^A(^~s$angdd|ZDsxT2DVa@8#f}=6T7y@ezB~Q
z2=E@2D5AF%g>o6yLWV|AFayGZy3Y*XVxtW~KtnptVYVy4GGN%brz$*4T)qLBM-(SF
zm|m#)%Bm>-jODjPl@!BqAs@&GWA&6K6hz0z<At$U)r9i&#z?4Y=f|O*__(k#+37H%
zH!pS8Ar(Jvd=}L&E}G${(A?I7Jete8ZDVKCU52LX&zUtpC-W_ZOySHo)MTnq;}El(
zChswRRyBNO94ij@y-Y$7xnpiV3Jva@MB5^A>^xtgn^H>}d3AylkV@5}3FO1$fF-+l
zzpkBZVRjUR-Z`Og=ryfd%#wnXmr&%}y~4(W$>f=Y)_7paUT{~EK1qoUFijsD<ddr8
z9bT;LAu?h*k}G0vVJkNWqp)T3n<Q#0Y4ooTH905kNw1EJrCsO`WNiMh9yo2hlyR56
z`T;&}G9hTuLwS?dCi)Ns?X;cdJC>SZJiTK$dNS+5KH*gP_Bl=;tpKvwdRO8_zBQD1
zY_pX8O=f;ti>{L8^z)6nOvsA8!eH$D9!#t1b`>!u%X|vUd`-aRA5N{-AoG4w5@I>)
z!}LgaUfv9h-23rr7e2t$ER_TXNdW`b*u~*3?Nxsth%?H{ai3x~pYr=LxRq9{MZh|y
z#eHObcAdctvTQ6U-sS4>5c^ofwAkR=p6lQFiwz{+i@fu&d`_m5lseu$0N!gTEqU*p
zdz?{`h@8h7ubJ3||Nd24&L-GoPyTFou-zVokA=1V08EU^&tAb2GLt1#NByn$V^7!(
z)?M=|P4?Ym?^en>syLHzkpj;(u1J<5t;aaBV`oDiZS@H#VEymc3AN!K{k!p+Xtb$R
zWg4@}cwRL4?8jr%vMMrMQqLZ2F|jlY5wlGOFUnJR1IxwhHgaagLS1@TK+zB^ATuc$
z=ZxGfLf$^ut<!2Kv}Ewbe}96Y*jLyO=k!>xj5(`+6tQ;LdyFwlT@3G`wvN}UD0UUK
zDLSg~;#EPf0^k69HQcT{u+=-CMoyMchwtFlj)hUF#nN%w)ZW5_Y}f?SC~CiZ4^^}<
zn9%9{BeqcL&)*dHch<(9e~N`-Zs5hvVVjs8z-}TtThUIc&N3u-#Y5qaU}dr{gcJ1c
zI`RSzrEFdwgG~zq$#eLMH_?2_&WQ-bsJz~p>{+#U`K{_3k2|mVE&VgPPf=v|h%$1h
zc17rf0l`B&_18b#SSnR46{Q7p-P~APTZ>Wwx;n@&!H*K5M@Y8>sDv{pFwl*k^y8@$
zekeq9JU)GFWV$!XoAY+1m$0b$la6yGqtK32bQNH}1pu6P5qjI);q2)X?nu4myfYXk
z>1H@IzqaW*9_8{jmK=3cbQ}++h5Qc=DCTz6YB|qV%+EWU%{bf0{%#6QghP1Rly~yC
z7HaYv>B+!|ENF6kHn}N*yHx|?hi8-AZlZ?LfP9GerE-8$I*)G*f2hE`2a=vQTuF2i
z`nGSjC%mdO>Wd=1UeReXQMW?t&Q5mr;LuQ%cB@hvMmvgef!f~q_t{>U`JvsSi8Vht
zkgIZvm9J?6D92-w`dC4AZt|{`{LQo((qzQ`K<1iRkziPn;^K!dIhdx1U&`4-{770^
zM=2D0QyJ@+NNxlv#FG+qM_(w<2l$6whFTo5&kOr5khW@$$a<w-L6j(^rxNF~N$itO
z6P3Y!l{V+vR5+CtxiF^Ze3;AI!FjxhPW@Dsw^&o&#P}{XoI7R%kmHOuJjFC9p6V8S
zjNWocVVj-p9e0+?j$e@$+0K;%!Lc~SYmQqwSL~odOODs+!p-0Vcx6rTRN@)gpwj*-
z69^C#)w^|{OixV}G8X57zp<U#GdKI-!GoF-L+AM3!ZLH_?(cfx74*vaA&JN6x$siN
z9{TE)h(<{G)7Dv$GN?`@%WK{ff5Z7@g)q>IqLTbj*{nfDfV+5bu0XCyXP%EnA7Rwn
zGN=p#%h|y=E(O<4Z$_$X;S6J@XbUHOc%8TIe3>poxWZsFLzda`)Fn1JX{B!agq-Uj
ztO#9Vq7(SR)p&%csyp5rf%7H#b=GC--{DMc)J>+S0u>3BMTlg|hepoGO?VIu%|9)-
z|ED*SyG5kNU}FNk`6g#0+V34Ut^XU1*Nvi2!N)+t7dLE8E34$?D_Zla1&gf^f}7UK
zF)gySQ2T(isH2GrdtTwZY!l6KsM%C<zD;*F@WzR>DvO~M`WK@Vy5B(2cuSv4C$$)P
zE{C{l+zN*fs}&RoDJROFEfhtX6GHJ@B{i82APJ=aIZ&<<hZJq`)Qe;>rncz5qNr94
z4h}Y6J&9sz&8*_j@=e&K1Mt3vhZRPSNlKb%K6SdHtu2m1{qTeK5g0AIsEP@08|&sT
z@Dc*UoSlpIm}sJ<xZ?Wtj>ik+lGL>>U;b>WYwmJXA1`e@|2TNh0Ri#$Nr!-1tzv2x
z(&09t*IvyQAA{FOv=$}LxeAPqda>9cYu$G>Q{fP-+C3)s6lF@nUeN_Dp=RiK-bBRC
zbGY=TmDL0)wS7a8;jR9AguSZJUy_98LKDsgXMQ0k37O~u1UTM*M}Oobc`~iEU#OC*
zOFRc0fVT5NBVM9#UWlQ6)J~+<VyJ0xp)pAf**lVv<TSnxl-NM=u_M1LP#0Ot6&A%m
zsst6ig9TSZE=$h*SV3&Wijsf>6u(r!4Rc@D^tf@}Rk-3sL)};Nh%j3GTsFzECOxf2
zKRx{o$Hue@*~VIk6b8*_m1-cIh1h0o^j`Cei4N3O^%-k=IK9)TXsTzOLHQ`UCe3LF
z4GAx@R05{etlHY`*45b(i-H8Qj(m={y;5E$JKH0(J6gzSpib&`G1#VwyK%xJMUj$l
z9SO}g{#sy&;LCpWV?D9d*vY*poqlJxtbW#)D_m%pB{lPaDk{bJ-Yq%w1t|f=*x-bO
z?WYiW%$zve^)Vxbo+R_ZVsn><bN5_uO3c5?jcUNkC>kT-K8H%Ei+9-?-+?BE%%Xs~
z6uD9Uax`ClX_?l*2V|!Bx(>#5z)d-LCSl^Yr;jZ<2DS1P&=?C+TqDLR4#W=?v$a@%
z6tz4)O?xyp0sfLpV=j+sH6$0#cbkvra<Wrbis^8^cau?ye#CeE6NuE5`S$bZ9&(gX
zcs-bHpN&iATHZ{@%82^N0Kdj_s3&<98hI%zRQ2}uHil~FUQTituMo1NrGE)1a@5tB
z`c;lV;3NcQQ~a+k26v*)GbWzLm}5z%_X}+O(@*69btE9KeO^c@bjJHiPR1upUhSBk
zgv3bbuTYg1xg>vv*qzzH@Y%pMWa&?imaLJEk8yDaK*;On`}hZLyfFsmk?O`L!v|9B
ziL1&o88*-~GEf$Fffzpi^{Qi%<H+aE)n8@MO*zW%t1$6V$E&RA15%GQp^~U-<owEa
z9YTIfj&%w{SqO+2vJYiB?S9!(r)@(oLqun$ab-k4@I{dUzya@SOX3LkJ=?EP8m_*>
zv;?*R?I_SQE!Lzz@Hk7s;+#|HH<d#*t|NW3QNlqnSB+lR=o5&*1}QI7=Po_>qh;~_
zM${q2*o}ICt7v3R)Lsq52^L8`n9g;%-G1PuhA#cCYIBAWSBg?s@p4K_iN=1)z6><n
zmIyR!m3zc>aYCbRNVO&-xv}?w58|e0{)U{0Rvc&julLHoBsKE!v_`c`vCfY1Df6zy
z-(|MCck?$!_(m>NNClY05ge`H;Z(#m-nO?jnouBf@&0Sa1%zubc}fyITPX(~n_i%l
z>7)Hq#SRV$hVKjCYQR$x{$7x13J#4HQzzBAw~dM(OU*d`My`?wc-x9ZKhx(@yy|#A
z@zF?@V}998?DLeNjP&&Sy75tW3#8m#SfuUB=H_@hrQ}p=-BwY8N}BdkV$uY=WEXOF
z`1v*gLs_@Ev!<dlqsN&5xTy9HMQJel-6n=>$^kftN&p<j^@=t+7)qf-4L~?ffM?JM
z1g`FZrndGIUKe7BW%QLZErzJ`>`GNA(wsV6XM@BJ+}GLBQZ6amVeN&k)f)9Wz&j7j
z-)(J$(7nqk-ORUkdbHBI<07fh)|v5I8DUjBSR!Q7`nc#<baZj;%6;$WI$EI792c6F
zJ@dCEQ$`4>9TQz^W0vEIqdtm~f!%ZGb(|&}7TztA<U;Yw0j>h=$8u#?k{OkzvbV^m
z{^K0xkDR<pS~J?{8yxzE&I9b8uIpS2AKn_Ml*HVh>sYt>p2c~mMT<(E{9w|4ER^qr
z&w!g-rQ={A$y?{e-a3o`*QwP&Qa@`JnNE?r`W8Pn^2lemObR%sZ1nZbS}3csqW|=J
zHTJ&yVLK;nuk0}qr67$&q({je#B6T>dH>S1onomJ>CH4RimzP%cz_)=S&003fofL>
z({y^Yq$O@ew*KR?A{VoOg((-Y&ZaGs2|=8HqP#YPb~j<WuWz>Ye1)D6i#^TR^(TUd
zhkLH<ITU&s0=5Di!^Z+<ynTjaEN(2jw~;*Ka{<j~`U#-fRw%G)%GJny<?p*&^N-?u
z^jDQn!<RF2E09ymIm+)5HifV&=z;6}kWlgY>CHCu_86Jd%8RuaoB*#R!DH9nm*@y4
zC>u}a>n0|L6Y7x?--njshe;JYh>YrS%2`U;pK}YiQsVQTTR!W5ylAuYXe#;&wA6hE
zX7|keSEbrqRi=THMTiYW9KOX)Zq&ne=dGS>lbRc*Em*oZMlbA3l4E1+g6C@KFqJOr
zI1X>OokT~rm$Z2N%Us-)I*Jq*c-+g#6CDMYU)ma9rR%rmNtRre0mqNUXH#hndO8sC
z%@Nb|BiVXn<ycycorz^!o+|IbtuS{;?bh<np{_Zp`BEi1{IJ%rwxL}#;+VH_N3#+=
z>2<O>yfHe|!~G*L1fvD3r4QbjoJBi7i$9eGA&G!$gGHjKS_6iooHhH-Hict<@fA%?
z_pIT$k4yfRD))}~3prDXMceuB{x}lC`q(sqbN|`zJ0EoxFPC=p7&$|^z&)uMRUXj!
z)X*Nt!6-I(fH`WRz~0|Y`#77^?WLb_!78}!j_^|Nh<U3Y8}$$jv`^2mr);_L%+T+|
zdOh#zpDR0ppDG;6Xzuzjnxsxdp3^_Lt%6>*t~<q7MpJkezJkcQ_*fK}^r7&p$U_@>
z1NUBxTh(Yy!q8H2!2RQ)hgNSk#;GTk%&U`s?P_LB2}QEy@%LFD8<juD;}0F+wM<!>
z1QAf<&MZad{NVCSYvY?lQMgEZT9%8_)QM=HxLxuCi)fTwkh4y+<e1R(=#%`>z^@y`
zp3)x)+c9@eFaFLe8awWCG5SWAOKH?Du}922*tu3jRX@_qj89{3zKCtZ(eR^`&_jir
zmI*t>H|-29Al2bR{4XUS5!}*f*RlH(=L-zdl9A?q%Zct&^Aw|b)?%}JeyurjI-R>x
zaOAhGc(&F*na@;KYwgRA=RXPc71y@Rhw>hNtbPC)S_Mh+{(F|2%_*k{ikgb7ihqlq
zHl{W^9V}Z!AspWOJ*3iZNgTP#-Jevc_+7*Q%ADK?aE$ekP&j`yx7wj5#`}~rT0y5=
zeV&-gh^CU_7_tMYhXknmn`={HN>-10u71}iSI`I;G}r;!?njWQw-^`w0(fJ30n~{r
zLaX}yPdpej@yQvK<KwQxRhECX$(GUEtUK@Haw6Kjt2u=}t6!nsYxZGj4UmW0%4~E-
zPV-s9am7QFjU59U@2Q~Uj^uEL$&`=fN@a6{eor`-tsfMQ0d!^@j?8gR&y_ZU*ZCK?
zGGfNLy(9YI!(_&y&|0N))^be_)a|4LwZ!bq?SW$A+hY*vPrRuIgX#j824TJW-9fod
zW)vzc7nPh0cnQsjJU~*C(>SFm+-TTNJqsCJSv1jVe>t?P15xJPlmXL1_0NALXp!ec
zo{^e)&!Qk)xaQwQ_;57I1pcz5X8r#?KnthF1iiEsLC-Kn?N!riZ&tf1f%&8t7!PZ?
zk;T)m!33H7d^xnKBY&pDV^oK%0Gq|re*a`z-UP7k^q8WF0E=}awvJ1)m6`KW;DRPc
z(*I4B|7``pR0zcbDE%`Ygjujo?EFq@{U07erhe(}<LD`1@>A*Co?^Sj*`5^mVcW>h
zQ|IRx3%`66yEEHHE_VhWk&@zKO5UKJLE}X%VKa!URK{`Fa<eit%#G68?35EiRi$;6
z2ZTEe=jbk!=sM<(FMJSK_%L22MWZpOOB2?Lwrk}A3~IQ2e|*u*HpUH;pPNXpK+ET;
z71Y_{xOF1u&XhMWrfL|2J&UJJ8tBgbanAkOj0X!YMZdR6PUln=Of@s|%q}Ks)xN%1
zy3=8qbYMvnHzn>+(o74Y@o{hWc~Tx#bnIZw32%<C46%Eqo|iIyrNm--_*C9{E)kom
zMZNF?WFu4OY8bn-dQ>H*wMRdzr;~z<f**Id=(329#Y;a~;Y6C{_N<QXSgLc{a)`C0
zRkb(GzP|CDt{!)Dx^(f|DQs7Zu7Pl;y+W}on|CW-M2#H2?d;$hO6cCgb<2cr9elq7
zDC14%N_O1y0TN!fD$TvCLBmo@p+st)vA_e1pxLDnKI3g6cj3oaSI7G#eUuVPzmKi%
zY53#bTDK7yio$n*WGBuycOOpBor}>eEBI7t^Yep~L=~@p?^L*d_*|zS*Msp!>TF%k
z(S1jbDv4*a@xzJT@n8k&f8_gDf-4@3S21R#7=7V~m1vWyArkYK3Y_sOk!Un;;i2#?
zMRJsjZTT}^|Dp*hRTy(pFlRW5^YDU+iI`JYXv;&cyVobZPE(aJOrz%sIQ7X<J53in
zHN>hIztNgH1?w_6h%Ptcd{^O23WZS@X9INX4`%&-EIwJUB62O8MiL?PZYtfw<v?_9
zSJ5Jsp40kf{>*Vkzb-w|PM5>GXxw(Oa$C_%38{vUtYb>gEzwk^SY^+MuKeN4f9TNZ
ztZ0DHz{(N=Gjgf%82{j2_+~Xzh$BysC)a-6Lt;4>*`EzpW53N7^GJ?1q&QDDYb@Gx
z1EP|%*lk2|#wP9b$`BDtgedv4Uhfwc967Sla$qa7zIFMB4=TGHi1V;C=w5c&tY5%m
z6TTPf4Ejn7`?B8noSr_`%z3q<J4`P|F39fJ>K?B){zB3lb!4G(q}bu=d^3&r*X{;>
zl2ECMq|JEJK6~cVQXdJ!)fjC=dGioQb-zCnC9)o(<G5hC3z@M|YJ|LqC8)YRSHoN@
z>`U(D4+_uKw|XSswya9a?fR_8`F&4ejz!l6s7y}hVkj;g(YY}LWi3mbt^)S3>B?sp
zc>HnSuPsf_y#+LozLH+pWy<cQ-Hd49bk<(N-obtbl<I^YN<A11QtD?~czAU05PKFK
zC<_LO#W-`%KL^j|j`k22;oqkeNIlt_l$@MdyAYLOy;Su!yZmpLuK&kg_OWwtJPA9$
z5V+^5JAfW;-5f(-52_L1nQ;o~h9izs`^r@~HpkAj`hqjM&T^V=OQvT?U3n~%a5<U&
zsdPN^M2{{HM+dwRD9*2S&4`Ra)uJY(Qhdp~Iz>^S0U9}RvH7M*!ih~^l{}kYg`AD5
zX=!11xVF@)Q11+JU!q78lpH&z278NVGOkO``A+4M+DUI6uT0X3H`7@3%SN8i%P7Pt
z3%^!7gyh9(Ezzl^r689tKnFD<=uBkt08OU$3cKhS$-0GZbht(UR*QOUr0+l^H;Qrq
z5ak#%y-oT0rwjAdBHQ?<k5p5wNO%{EIuop><qd8H(bS2F{3d;IZQ*P`zNNlnqO-)Q
z@`O_r{evVnx^_4CnY?UnHyYwFI7YA`LG$5ihX_{aDM2B<S@GRe6T?UzS*)?8cImDC
z=UhfvhAzECUlv#?^usbXvRHZ!DxO!!Lk!gB|9j70T5Iq}YXa>YTLE3dK55qt6u3(+
zlS1e4ef4B`D-xPVNwcp^DpUDE<z6!qiU0j)1MgB6kZm;l&n&}WpEc&SZUiRh@<8h)
z>Y1k)<&3fLg|?!3TSi0CoJOc4PRoqW%7`tf<wZrYNiZ(Z&iT!?gT0ra;P?EsJ$=cW
zD?|Q(bW%woP|T#}xD`BG4qCmM3+f=5#~Sy7xQ(L!kR_R$Yme{sR29g7I(>cy=hu9+
z<BX8<vg*{1#7+0!_97=**oZ3rg;TsFiuIJZf@G6?zbfD>e#0ka&nI32!@7I_P~G*n
zZd*hh9rrX4r}}a<@4h6NhO#X(JR7mM<k-fGHTg$8?*pat*__fVJvy{n`bxA3)!OxE
zKC+QY3<7-$V}^&3&|>UGoM@2DA{Uv~bux!vD5m~%ce=k)ua!gw2}_jQw75=YbNKhF
zvF>&qAlY0^jnWHu%0#K_>~d+Q$ELEB3x5M(Ij0F@MR<mANRhmFmA_2*{GhA5O>k7S
zH5J%6i<?uOpcUn~Iu$u}Z9?5w=^wKp4QlE}FW(~V`P!byZUh!dip_6jA6o*ikzWIE
zU9M}aI8kHV`P;`jMJcDfd8;X90)@Ml9e7qhMEXL5UQl!HO!N*j<{T5H;PY+WB$N}2
z=@LLF7qi2RKwsrFQ9P>(5%HOClxIE?@;8_X{buy(BBjr=WjaV4)c)C#%7;&-X+5?C
zia)=EJQcY*^v*cMG5kD$(Ooo~W+yw?2dV_&;3TcOcalxl6X9Jh^%wELcgEkxK0ZO|
zp*Sr<K`r}oiOD!RP((JHU$#6xwa30DN3QMHHQN;=el_q*$T-lDk+LxP$`ZhtcM0e7
zT!dwh^^^ndp!!<Vt^;Ux2m8fdq#>o|)1-o@OTv(nCvC*L|5zd2;)ba%yEU-*3|?&<
zLaE52ootfd<b{U#yUbD47MzwV*9Jg<l?@bek#Cz-N6&u_?Q>w>O52N^HWx#2W?RuT
z$*{^?Qm$OeuqIP2_a!*~a9ES&DxA7vshzkaymoZRN|EtsjXQEQ^fKYXkygy1Z{xlV
zV)<m>)=?r<J6bl~s|6G7+n|x!%?q7YDb@eXFAN#;c?XBm9W9&ELx=~T(L5q+tQB(K
zR+cdOhR<^yE?lF<Dx=E?r5*lzU)LdWfhe_{Qs|}=+*DguUxS8j1q+8}8SDIfQ3ml@
zs{`Hui+#p;9glx`fB?WGk|RM9p%M-!ZQ2MeYClC)*YU6AmqcsK<x|gD>B@0-fO3E4
z-xfI%({g;vCMOoqS6wei?A0yO^9o@1(`o?du$VDb`vY;K0}4}LTzH|`&kvIWP2-9z
zgegIjoAfKvL6Si|7o`EEDv;z5pKH%QAoWWSM@WsI6hCKWWv}17+1YNEld{Fj`8&`<
zK1x+@I^Fh7DLTHhP)&<N^=twwIUnkSZF{hA&$0tiy(`aZ91%ToD{P}4u(p%QEXV(j
zVvYU_S=tfb*+2Zz(h%LaSc>l4jE`jDVp~U}I}~ZJmM>FDrqhdd3O->llW@L0CLK)U
z7Kc3vqCkRuV2ThYJ#q=4+pl%L*n&Pc*}Y8HK^QkSioafei*iHPeN=Fb^!YOyN4oYP
zF)4EFpbwy6>_U>mYc;9;|A6TD`}gnemkYN=wrvYX45lg|Y1$RyVQ*;2o6t1bUNpND
zy!S@0?`k#`Z<uWJ)))y24Xr@F>%@DT7vYRMKW|=oM4}ARvd5l3>@sL~PCq}ma5RXY
zyK@C4#^Z^=<m@SA?x+LXVHaat$KRWki@B(=sK0d9eA}P-WN7^OGJSzj`8CpZC0%#5
zqX7P80p<2bGzTmPA<bALY2T!!U(&eY&{C5;UH(%f4avKeS)4Jv=A31=)m6KB5@A*J
zed!C>#yMg?LOb2Sdt&oYd#KY$(ra<`nIj=41}A{cnKi68E+V0?+curMKffyQeK(y(
z)mNv$UQjBn7&WdAzH6i(x8kBC7<Q$WEYwq6l$g!B{uEzLt96|g`FVK`2xdxo;`rxk
zwTc!SPU*LyCbLtOAVtP5?h`{<!duLdi>7|uk5BVLO}MuXy_AjKM7<RDOt{H;s48w|
zzvx%jK=IyRCDdQ9Y+!A=cDS)J4%S}0$RwGHNQ>mvQ0%Tyst5zxp$~h%N6-I!$N|K`
zb*lPWH^M;-Dxv<fk+v%MqsgPr-Cj?g6zg{I^u7GHKQtt(c%V2D&P=xkQ$$(E{N{Ua
z<<ed~%$BA-xuj6n*eX<`cA#gM&@FI<eCITPOx8s4%yN427lD?L29@Ej4WDqi={OBH
z7EVJ<f~T_7l6I$Eob76FndfPMbw_ElouS_w1W|tbnPxtA3!aJQkcDdZZypFglmLk^
z2@Huy!uPt_@t&nL;@7%N)YD3X+AIORpZV$yuza1V$Bp^B>O1%6?9;j8wc?WXsNWjt
zRnd8zGD+$l>qZ?zojV8vAMG`F)=0dunR$<HJCmkf`}&6H9~pEiLj|SwYFr5uT>`YG
zV2OX;3bv!in;!C38p#s?X=mnU$K{DOObY3oZ+JDakPZ76&qf+LxgS4KQAN#+>*+i+
zxLWa$fqXhKZm(`DrhZ4scZcE^MM?9T{-kKhVmnENOjoIfQa&=BG2=FmZO!}tc$G|&
z);mEOs{8FZV_8)6Tw?P<+ts7fhHXcbPab@9cPOLDV%ofJoj_|-oJ3=-c>8r&6pcjD
z+9a2sB)fCpkM%&)Vsm)@V%4H~z5^WMJ8ZFs+tG(kMl;k9{S)Kfsy9zej0ZD~XDb{s
zEu6jWAyWB0O;6xz?iT37er4UH_W+fk6K{?pq17o|P-+xwmo1au1m%yev!2T~c6hMf
zD+6CeUgIYigt(A-PHxwmY0u^17G=*l$stpq!+_iIOw4Qg;>FgcG>N6_AGc=KCoxsq
zolyvbTKlk5%}6@r;h|DB+>oSjVT;7)-If_Hr7@R}T3JA{C>%lzCa)+(z2S3$pI_oA
z%1YiW9i!ow!3%T*2YNxuSP?MKUNF6OGER*j_f~Oy2;aMw^KlVThOitSecVLL&ES=M
z#}lHq%<S&y<Qi!|kO|kDCd=)LWMU8f?P%;Q9+WZ@^+2gR!#$yEsajgHpE0Jpc*l}4
zrH|ceW>x6`1m&&@V9pa*qBV7N+IOl~!*3aZl~P-;?Uhb4sd<Z&T+z00-S4D{<OeGr
zf~m!w!X_yttEH~aDQBc$T-f#{^m4LJ!Q4chVh_XAK@^q~2odkettj_?_m!D4NnRaR
zl~bA`@=Pi)V(q=C&AO4OE25_VuE6g?B+Ewn7rd<1+NfcmA3bxWcufl?)ra?=wEOi;
z7WoTOvTFixu(e6z`Q}|?z5a1gJLp-Nf%kI}vuv|6+H$HXYw;&>d1b~dR(6fhEt-mE
z3MQCG!y`Q{zmishta7lcfL-xAK4sQzY?LHO=stdhf#S_f=B2BxWr0x?_TtbpY9yKM
z&~N%>x3z3b&dRs4t-HiFs<u{I4FGE)*F)xKoWZ4Yv~#{>u*oNUz44LpE+r3zF|Wj5
z3v($9`{timQUIQAnG9RHdRzQU+dC?6$J+5Zv&)H<eY$F|1stc*(Y>Vt;yUYZ1eF<4
zM;UXv488D~P{y?4Hs^M2QhU-tfY@ASg4GuN`yJY7wyGa!q%o3aR`Ux`V}I8)o4T=#
zqt{pGgHWB5A#-l~D*Xw@l0oPEg*Em}yW!mZ^Uelah^9ym|K;n@5)>%ezbAs})x#Nu
zRT0g@><`}T=sK;4$U<MYs0n;=J0w`M9yRieu2m10un@@qEi=63qzLM{;_29@xi+M!
z=&PpimZscJcmQb4xeFW#(-GUT#5=JpY?@~M*VZP^Iu4e#Eajcfc4^a1x#-?pi#P3x
zq$aAvXB8O_x@DyfqnIY8qu1ZQLFMI&(g^SgkdW(!GeZ*_eW`i*uA#1+dWozpP{mq4
zZF^%xQ;YSwarYCvG}mJZ2_SnlQgkY+#Gk^I@bxZr<y}nc86W8*L}z!&D0O3-o4dgz
z5zwGCzlSNXIujHXvT@g&ne*B~)6>&M@Y-vxpMyp0<l*Mrp!Hze^KTJ4DS1Is%r*3K
zr}y)M40|cjtW-wl_O8xUf`lVY_>RFjK-cS@9;rNZl;a((I#83R!2XNpj{5`laCfhu
z6f5}KXlypv?%s)vUK*97tAP4@Kx({~8e%&qL7m;kIWQ+H5Vs+>y?lJ7%ghMt)mE?u
zy>^9FgBe{+GKE~x`6?$jpfE0JKsa4o`GJ7~cbvAMZ(M++XJ1sd;0)ivakqWEQ*iL4
z027@(o*IV|15AX7l%-V11K*C|8U4^Q`}<<<LD8=c_ut;jEd7N5W-9*Gr(zU{Fl*KW
zYaB3EGikkw>PR4#Go9W&I@f%9MZ5Wjv)ImV`7@KrJqiH3hLb*R9LVy5<pqMRVnN++
z96kxK1MK2xBeOl@_ZzwAQbV^oWJIzm95Sy4U0h(~u1}{8Kx3q^I(}U1__1S~=Sf0j
ze&=qx<&l7-ztXBd(!_R88A2cFNnpf-(E!|apS-zep~&KX4^HO=Gt{ODXoVw=VDM}g
zgHB8RUecSmiqf;y3xrJ^8NKIvqOuQ)UJI7JD>@gR>iJ%!rNm_RJ&W`2fW=zfGBGC}
zfr1I;JB>e6HTaqri#Id4bgA^o8Sg-;9S-glhp*;NykmUeUTldymna@(8gwQ}S(#(<
ztKc6Ijs(%PnSkyrjkR+u_L;s<bz`17UM21`Ju|#iH(NO~%~-4uqP3b=x7wh#=gCsV
z*_nxyX5PuT!<>iTw+iLC<;J*qMrG!-Q+%HQ<D~GxYkEYRl+wPY8XPS=|4?`f^b72=
zf=AX8a3)H!jamGciTCg*D`dD#CFz{e7g=LIQRlT0Ls+^W-%zU^$nk0V1uvjQ%z-1x
zsYJ+aqwCOr4XpR)Fp`~ucL)b;q8QNs?tg|sRXut>hsdTFvRiuv#qIxB)m2yr!0gib
zxGAv>yz|!gX7+FDcoU_WG(|9aieDa(x6JfrEMA>%6ij>wkM_juG~t$>q{MIoV%C{)
zGhieTL|zt4ab^zMUKtJ>@LtUE<b}WNo`NaSA8OmaW9QDYW#`#W8>2>t)Micv#g~XV
zd(~)}7Wg?>$m3nptbJCMUiY5q_NJ#w`7@?=%WzC>WVi_E<Y#yO&b_bbJm%Ase6_p}
zk*HYFYq?dIP6A{od>Aq}MA4uvd%y1DSityJjt3~Pk(l6_Eaw+Sc2Ctppp*|I({rP1
z0dRgZPm6nasW)s!2|EQr<zobu$?}a&OiQA;9071uxov;^t2XlW=r^&#_y)rku@*&%
z$$ElSW2EADm{6_-PrTO4+;|M?GPKH?bJoRF|K#Lk%uVY#$^`;eEA8lSNX(hGDeB$U
z#V>5g^2GE;@GjPnnVgNTr`)}1;0!nqx8^M57>p?{ev$f;bIuwUww0|d`=b?0s>H(f
zH31GJr#+JNg`lgme!R-qOncb2Ib?V7&vV60eN~I!rGlkIgFTeHE=3#G9tl}u(@_)%
zuG34$YKD9fuofKU;H)3#GPctmO8ooX^(-t!qo3GQ^2f@g7(EQ}Qf@XNJ-z{f4IMhI
zRqqT4#B>uhIHtV#>i^yD&kQ9B*EsM9|MmDIxAC1jR>nJ#61Z|s5D2Uo_ISAR7A+KE
zHL@Hx6eVZ>9v&6SG5aaw{9;c=%FKT8Bk$M9a4_FAF-b;B3^-dRZZ*OV)t3X+*m%Vc
zRkEEAEN$l$8rjad(xumnM&<SQEqxE(%In0$0s1tp>fkxQ4eJ*eEqUS9pOaTA)(f(G
zpA_6~zx7JuRD#(lf#;fE%dd!s2{r7k|J0W6;xPU@^2N~&tCGJz5%SYHaZ%>zkI~rt
z^XHCd9N5Xe`HxRusyB#aGby{TKd8Jz`M?$fZoXTa-achJaD1}VVkO<n+WVu>DXQ|?
zgEBq5h29U8K1rB&hOjG{Ri#u)#O<6zd)gZ`N;c7$5f-YrJ$_AkBSD`wLXKw>I)Gd>
zyMI$A1hQeDF8q#Eh*RP91E*+U(*ELK?sou%pV+6L)spH;(dGd15+4I3k$KcIS+|Ij
z#WI=s>S0mg{e^23!hWnQeJBh7fQR31JrM029rIOrOIAnElyzD3DkV0!ORW6S0r2b(
zM?fukn1?M9#KC92xuLRQz;&p}+0l_Tg!la#ueIif?i9a7Ex5F_6lnCDi%O0^sh3{L
zqAdEqcX*(lXpL01aCX&bcEt~li)K1IX)EL8&F-H%6+%$zoR30or-|S)a&|q+I8f-y
z!u5y~n1WKy<!0!=K`~=)x9bTNep{7eKWQlbm6AU{oHr^nB6i-9b>i8Nl}x9AVjV8N
z`E>l(7a#bM*sCw7TXcKndTVNHl?S;g+hlQO^sE_+bm7<3n)Az7&1I&gjZd|h>IKXJ
zO@hIXzDu2xGrNy;!tJ`(7%0G04`z7t^{-#Qj(O2c`m_)=1C)!5_Xn>34eCo}W##))
zyZhkA2{c}N`e9P4;le2=9^oB{JI(PG@XRHTANMKY3V1g-c&GQ-hLUB<1T=eG#7Z{0
zNyouhPmiZ@_ddOgW<TzJKQ4=xAVaSXsNQpY?Z2N~bV;p%wAsbw7C2|@Mo%GGv_kDW
zEyn<B+|5<8Z4PQE=N{rj&e?*(w|&L0Z+fD)Hgo*J8~gx};2T(o?;Ii3Dc?>}4ttSL
zO&}aCS--mnDE*4t#~XJ`;In0-!j$z(f33N5lI8ZY_kfz*-0*zw`|A;hX*#Me1zyU&
z)cx;}CpvKpJy|AQi9}21b>)5dP?)&ga|z2W<K7+RVe3e#@V~pJ9tMb2mLY<iL1vHw
zde{9!u=_OKX(7CNKvG5i+qat?R=+DDh~2()%_KgJ<l-7?CHP7=H*DUIUq%FoUHzXO
z1<=d?e@9h9lZw@KQ3M{Dr|bD({JcFff8fc)tGdt1j{22%RY$2Ovi<i72V0A%QeN}A
zs`08fHq?@z7Y$X|YKJg=`#)R0qr9^9Tt>^#e`iX;P*1OCvPsvwb1`ITb$8ogs)j~8
zOFIJ~R()qrf_(w!{2%XJ;8MIhUjWa}ygGcm*(habaIkpiUd8(#BakWd_Vx}a?|N~6
z%Xi=&dm(aRILJ=fMzHrhfYk!YwUvo30&=9M;DP09m>)EnWXZ+U&dki<ZPv@U$JgjT
z)LH4^;E-xc(LO*9^5+brrjEx_JwsVEFV};s*D#*~Aq$5tvBB`GjJpA9B@=%x2u-cM
z=Rs8b0_FkA=rEJ|?`lWv-YdNRJ%se`j!th$QRRKP4cXgN4T|;&{C587wD01=gDgn-
z>@vE<v6<u1HCtODho=eIV#+T5iWe_FJbcI0v{YxASq+Zk@69Bp<rwH8I!m0fm+-@-
z2&C}ph+r?CcHbX-k9F}eecA41#C{?9Y1<(!g<LoY2}c_vmOkFzu5)sO9^9_-(`%#E
z6J7jWr?ft9-+hpL0mctmx(gh=J*1(0V}w3_zru${@smeR(p`N^{aJFT*H8=ySp3<N
zVJQ38FkD=A-fVl2Lfml?kt^8-!r!;5zU&&Z5t~|h?*y@9X44}y?^6rg)Oh_aKUn_!
z$B~OT1;LM?19wCEd1iIx;7<Yep=EnX%MjOZ-@n_@C!0DqO7(#I>LGYw{~G3x^`Vbc
zuz$`z&%Cp&^Ea(}>>Cs1IdM`>k(9i-<Ns}eE$H9J-O}7#b~{7de&WEVz41X|Va!A8
zlQtN8B{A*Zhhg?YH<Iju(&_)~fkSu#q52dGbgeRf+huycy7`+4Wi!CY_jFA2N6&J!
z_T<y2T1Jn!yhSf*1o^qSx%nfsg<vXcQ4QBZ9(8);?J0OaFJoP0w?70P9F`E?8FuO(
zAK!=SE}1S{AK84`y5cEm@tXewP$Ks|^C38Pmz&Zq+t}KE)MyM<ueLUGm?@e|EOIY(
zMuY&ZMZljsa(;-5|9|a$X*`wd*Z(b2Br1wZ=~RZ0A!W!^R5Hs@W-_LB>?re;u?)$S
z%C?a?NttCPWF{#>?9B6+A=9(2-T9sKJa7N6{x5##MW4^H@B6y1VU6FluC?6PMb|$l
zBtZ(i>$B*S$8$Jcnt9fbbkkl4Lllu9Ff6?ip^OLLRNFfT?p#Vb&7ah3PJk<#*E5#<
zhlY&&=cD`9FR!b6%!X{Oh>(8y`u6#rZ*Sk))0ds8oALI|8$@8)RR`~%kKtA4jZ^@6
zsN-x)0^1vsQ1V`QQa=@id9RN`8Mc6`--L;8nYC3Xad+<U1YKVCj!JD?+4%lFb7^T4
zx8*ldJ3kyf7rg=3ra8-2n|ArjboFlNWUr!`I6iZ`8Fxp4-twDuel|H{XFb%Mi<v9u
zFBeDd^}XK@=+RtwT^U1*&6XgY66)3rm3INBeseKwX=$m_AtyE`C`uugpLgBbZ6hkC
znP*jh*m!lsW7;h#C8t~3KyR&Mpw?rn>V={zQ9`hODCm4lGdP1w2|xf(uGL>2Rngkz
z5$#5)wv$A?9b1WtiHY&%Nu)wG7tUQ(E$TIdsx(9jnga*i9q=_xT@6c1u_A}cTBJ;D
zZ9fNGB^a#if8cQ~_{xbuO+OA-lO*-JL&jq!p4(`a*|GAn;)P~$!R$J+mQ-@q&A868
zVUf`4yis=d?uG<g5y7RaXZ>>S@7l=$03o4VqltJNKLiMeY{k|I3Q6j@y1}ISekitN
z7H_$d8C*pXz}^#zm8TYYY`Hh$49T){UwcfYTE^x*y9MUeRP#6htur|}w=7qBY*!z)
z@muDUlo(`O&o-L(-oXG4ifR2B7Sg(o1=H2PC<3^80<rR|-CbR1evLR=1gmI!N5Fwk
zzZOwcENiOesX8^aq&o3y>)E+-MlC`wh)f6JE)XgZPJZ053PngJ+<iDHC<1tT9>Gwk
zpqr^h_`iZTq1XD`?OXxc(~(j}1{)Ku@6HTn<t%pNBv}1wuRk0Ls#N7Fd!>=2o_0MQ
zqOUrepY!p4c%rpEV>6pn_3a+4dcXBW%=kXkv;f<)x|3rBoWS?gb)CJvP0x%@M{SLD
zWtTOMoMKl0`prX!+wJIgQ%;NK%kDRFl?wIU7f0KyEQfr4lMH%yB%dKQ+JD@^^#Q*{
zfo=afc3)UTWQ|A#hOy1ZMvgxcT>eT?Z}ZRS+rL7JiliGjPcdr~CD$5oI)oQ54hr29
zWd`6BEL!lLq#Yjv-wl?IxsjHyuU%XbcM)7?IH~D4+kerurBv-N!fApK@rdW~T_S9l
z8eI!fQtwApuwK@jt07B?nSa*p=;-LV7`YWt2PPKbkWsKd@KR!O`<PGWIoIlraW?Uo
z8}kS2eWQANW3{*)WV(+^yR$~`8j0_<hC0!{eOk+7uc4n?RZ*pj@@~okv&xtq*vY$~
z)$iTAH}9Xss*<G7$Ir{_Bg9Hz0}}#};|`YYL9(@a>|j>{dwBSkP)wJ|YF6&Ag~@xb
zK`wQgh~9`^t7O1AIjz=(bLwdJ?l+{#^iR+^*It3~rQ^FPuv5B5Ml-9@u~mJM3g%FP
zNH@b`;MtXgz2jz;eO-a1TT2%%wz$ap0uD3Xi3&3Yyii6qNp?cwiEhzi&EFEfxi(L@
z4<FuAT~f@h7R3aPk%*Mo=g+ruUw7OVK1vV!2{+pTroPr>)!5YJyFAS7m#WDP(h~~V
zK8C^mFg=dSYNxc{WT2kWmbsLm{5xL$t@AlT>-vawCMd9G)dCXcsp%7MbJwFV9CuPe
zQqt|avi6gG%^JW1I%2VqU{bb`za?BFEQT0->e%indl%dJwT=5kiPa+4>tQ&nv+iuV
zhuy7?n!<1f1>KluV17~2+REI(On5X~R`fcf`UqFhTd3l+S16j0ej~Fc`DqXU7#kM`
zLslbB4t*Xg4!f&tE{MbWdwh#_V+F=wc6UuaH--#wuo7!dd(VN`<$wu9DzNAIorO9S
z=4Am9QDtJ<K*I;WONfbC9fIaYrIi`fn0_LTcv&q?lRBo?Y;)FM`eWztjxD&MB{wkj
z#m9zwo%>L5-^fC!Oa#zu{7^Cp!$x>|EV1%xa5td#mUPLg``A-n!-G%CR-?-vlbXP(
zp4r+c+k%2&kjVk!)1;x`78FPz-MGf>RtogsDyse{Y)TOn^7<>x(~fBxu841&gj*e|
zs+N?!fZINQL)XR`>xxvzFoVa6EckT9uP+FMoIthXl~MajLz9yFhQmX*MdA)MBddS`
zD}Uh|O^31MLHP{Z)ip`KqY{pI!y@IFYS3F>iRx*9xSQ}49s{0O>#H7daq)N-a+y{W
zao!|U2+TLl#IR<L9y6?5Z3^b0qT)AqH%a|qtDlD3&eJqgbys>Ve(Jx$StV9iZxLE9
z(cZAIu=X{Cz(%zp1x8RVsKPu+WA~wi`MDrm!;tOyIprG%e)??bRt#=4?~_S?aGLEA
z6X?2ik9cO1+)<b97+#n}Pl@@B%xzN3CaGJAcB{l(XR)yi(SRX}Gz=c$b`RLD#-0Zr
zeR!2r58A*|`hoOGtY6<p9F`dd>?DvJ^ZfbMn^VyVP(9cGn!_zr;sPfk3)MAP`8Gir
z#j48E>L#ab{k4dy2Q2Nz`spgCwqta%EVX`ZIQ(DYsxkF|@InUHt=rjMK^d0XS^kq>
zzq&_;$k6H{LT!x#ysT3M^Ay^d&x{G+V^lcxN><tu<ODh7^`4!GP+;Hnldc`o1vbsM
zihDP$q-amws<Thzj%s-Cf?<|>tfACKKXMdyo10!xr|NS4Knm&HL2Uzdf%F$%ywF=^
zO>9ll-%M#un3_U1xTm|X%Px%N4P5uZnZH<!jje~ej$PTB!pv+hEwT^ttyEUd%>Q*Y
zUNJ9aWq8;E9J~u<l_jtuxT2@d^6@%%Bug+;Eim}4!rk%y9MS2&2J8BUWS`1JT^U&Z
zj~jt)&@Laq1TJ(9w_dDYQ$1o1>=Nx~DjrYL*4Ey6qo~W3hs}%W125m^NNWO;TK~)|
z4g}Gacb5sC;dRL*xN+afMhyzZH@&2Ev%z(z`e;`1`b292FzLUqMBwhkgDnQg>+?rU
z%8(FqVD5cH-5xVN&%4%}$I*A*WNdIRe)4w2%Dc#mkHFph4uuF<{F&OT{265Vl@eII
za1E;<q{$1pKA}6>1~A#bYgbrOQ(`Ht;Z1y=WvIy=*~vrAp+915Gt+{0s$qDTf87yS
zZ141G%=p;%DSP2#{L?2iUrc*z=Iv+8QApbNb^(ZM1Z@(IUIiqhBo#J#qaYvid<dn{
zcA0*&EoAk6db}ab7fLvHick;CXL!R6gm3Y3Bu*xmomguo-i89B44snsHQJWIOLZy-
zu=Q4EarNE0C58;&x(qd;ST$Ois0Wab%RE<5ohhxl8Hf!?n!axd-p0~X3bgagh5y-k
z-gUl3bkkyK4;E)c`mqgb?)&=nI!nPbg(Ps=j0*MXC85p0BS|w1XloxIpv{>Jh$>1p
z`(#4UEk_0l?Du|HaQ`}7*c8B*um<*<WhLRks5NCc=!sif1sjySn`2`hJ0-vO^aTC0
z=!K}$ruty(gAWaN)#Bw22pvs@A`|$~JGs(>Xp?{`WkT}u^G*N#2mluT4=3QX5CqaC
zERSwF`qPzzt5fjlV$nvsE4_MaR^e)3YZN5TbxQutPWVj#RjN9F+2Srm*5xI<{x$(G
zU9pe(_BR7Qd**)l_wiqpSo7tA63lZU>Mr}yD+35B(9JmDY$iK*vBAPd(tUk-l`AnW
zNxhKqH)uHkkFe%rJy0=bwZTFiWFFuleINlxLpe>T@4C2%>yN09EvSq{x*k$5$ST<w
zSvzv-!TFd!Jnq1P;Wx7$c<>u}F;Wl0Y?Sh*Zu~8}kOl-V)7IpVoV<xXKomoH1V2L*
zA)MiV6oh%FEVTs`R?r`RHFHWz$c|^x1UP+(7VF1V9-+nvWgtxiTr;kxqeXy_Zv`~?
zfXk<HoVB%|q5)`z{AJ5krNj&7cP2*G47iYA&VzYo(}+Ozl1+5fSKh&ypqFtXbF*@j
ze|bkVIdV@s*K|=45ZAH6&h8>V{Ibb{r2@nO?`8p2R-{`{oxdqK4Ej^XF!HKd;vTrE
zvPE;6KdJ$_+0r1Y4j>J*4!*np#Yr$b8>IvI^$Z;l4L+r4IQsF<S@&25xzGGy_JiRh
z+YsGk0gh<7)6V=uXv>n606K$rpxiL<a%GMzv}d@%9a(vf|7)^l)39EA6O^72-n#>9
zzERKtN6GBhcav3IP!MXInNX?YbzaT_J_(!0ucw@*yJ4ELpdiIApXXz*qO#f2Y9Y^t
z;ioHmsWG}sOPy^)ooW(#+WDH?FSG@fNVio%Q}L7)ti&!{l+Q{6XUKg}Gx!KJxtAtc
zQ)1M|qc7YOJ0R5jOpxB=W2qo7sK2DomkL(W(>-EZ6!!6|64woGSr+O;nhf}-=q*3=
zmL6@2Cbv28l<!Ta(Lgy<zK?((Y%+|+iimiGsw#zn!Bj>UOb3nzP6R72q+Mpk1h~3R
z2p}wL=c!~6cddb-azU-uvb=?(ncW}Ju$5D^mhEaCcGYQVl3y{&gZ^|5^c@rsCFQ{v
z*}M0E;d0*MuWHSV>$3ci13I~1_S@L~%3ln9GT^w_ucfRm4pufyPq$=s1XiiC3o~QJ
z7&1xeAbal$OP7@6?6E)@TJCPSp2dK`4)F1E-=}v-CkruJ!kI_b$F#l%^EzFh5~JL2
zErl7YeQ=@K?;!VjnokqazzzMU5OL@Ic<OIkAFOg@>%WkcI42}DKqHPBw^nYz$JDaB
z8uvlPqrI}BGIbpNP?fuS6CV#^bxBk!ZL#IfowE9oQ?gtcUsb1j*Xn-$T=uXOSKUsM
zGXd=@Kt(!0g;A%9?@j@;_QlnQP@lu^S+r+(#B3m7|2+(<HBrypyMu2~8vqK)X<!34
zoGo|H;*MYfXGu@s%{x%CFz7*`=x)yLORk7Eeay`4ioC!m%lMk>`sK_c^|X(Z$rIA+
z{#)oSi5?{;;7Ce_7x}uI)3l@2&F>NkgvB7GmvC;vB|NDeo9~b<Hf&ou1pXAh{T!rT
zKt62rwtwG?faVD%pdhJFT0!zilKQEfF4Hro?&NnvDsz`+TL0WY8QVu;trp_O(Avz_
zT_tN<lycTdk9k(?0?$-{4m{SN_f^J4ds@`Zcd^_6rRls6q-TCkPEIVzwXE`9!rWSJ
zuF=%ghK`i)*r}%B<vaUkCKpSI{NNnvfiI^!#X!p0S^?~!J$Cafhlm(5)PpgM<sII8
z8&=f{C!G;j<zcE6hPu79Hrp0LhR<O=)%nV({iQ`S8U(h+#-7{4ecyj<F1YeB))&px
ztV5E!2UMm-_`vtytDruBQfjg%0mXz+`meH@{GsBd?{~z-#ouqoFmI-5P0aXeUS3|7
zT)4X79!|YYj$669?|YeQAqzLV3*5H4(vOXci_%|idpNGI%tS?bY;^Z+9#7_%E^wIH
z+6^8~yh@TSXQ`!8LK|dvOif?1hZM;bA_3~uNsH5B0toK+<JQ~5`8*~jI8<H%?l8dx
zQ<()`4~_dkoyfA!c7cnUWn0TP44v4o@Jp}s&bTu(%<x1tkfPp|mb&A&PpmD^)GiNd
ze`dj0P?_thJiu9N>Y$iE0#vNS9syJ00-jQ=ntvvHrFt1(L2>fCw~145uFO_3RK?F`
z({*^Q<eeacO?Snfou$V7Am&uQ-LbgLunw*>gW2(ns9yjTUaCb%4Bh@l3dp&6tk?{B
zBF?XChd}B&D!SS|VGpM8ya?eQJOvp8saz>#`ucI~t(!*Xx9%x>9DCF!sMQ7dc95Pe
z8FwUf?qGA>(?>9sHw8WeR7wB2MfQs!j!1=}S9vT9v1_xWsy7EAdG5Wr{&|FmHR{?{
z2F}f%(m{)rg2gghnVKf(^M%atv*~4iGktT07u@`=lqiKH2>W(&@Wntj3s*9)n5E9c
z&afJ+d=`R31;*Xo-QZ=qdg!1a7fGpxCU&0`kKRhMKA*J95^tWLVoDe8=<oF}Xm)oN
ztI>;3(Tk2#_C-Y#c4?Z<;QC*%Bizn?X)bax<9c^*Z}3$AnGfz8&W$&0-x?GhU|4SW
z&s$fkp^6yDwF7S=6p}9DG|q{OcQ;wT?if!K{_E8vv=8Bt1c>wjzNbAv$7@B0(Lv`)
z&m2Hcp%RbfUy5m(uQj<Lm)MdJ#t4^iNLAMMfOi(=<sFa?X-MzF8A1q|gNN*Io@;JI
zpKS&3^J*JNZh)Z=6i@*vz;Z1th6k^yq#hjicX;3(9jDKi`Q{BEF)M=lw^8Oxqoysy
zdYzxxnUSXu3{$udO&D-iR%znwDe3xrpm0F3;=$IFgSt`cBI_G4m04#2LBYN0)4R@9
zJ^C)Babif-<F^K#rY!_iDntJID#}SFCTK=<fdbGz_##-(EOi8>)Ux8@;@sT*EtQ{u
zrTnF&e*rdBeQ2Z>O&>Lv=6VjE0l+c>myCXVyLA@!Z^dS*W!72r(D=viL5Z>>)x{97
zrqnehAL|}?!&4Cza}m7c)V8s<Y#L1QE>RE@xCIQ>Kaa3oNbxf93R{>4F>uKH=6?vT
zd}t2h3LjTaT8c89B~ujw-tqE1MGpMbocy$zwTHaT(8F;+o#p25AczDF3u?cdBqKWK
zs9Vr{`_b1M`Vnj#D&Gvn@j%SJ5oAJ3r-qtA7?WX=`s15*f#jGDT_EBT3J8K=)ew`i
z8*Pk2@vyHfe2RxCF>i<$VF6PRoeU8q4+X6CMo^zT6l#IYqz%3)R0usN-yFNSn4&#!
z9Pz!V11vMH91$L6*dL&jnY@#$sMX2k2M`@)2_7u|J)X||Yh9CwCa{27e$ZM$!JyWW
z6f!gFpG6ZGL0~x3=fgI|py>#+CK3ef-#@oN9Gq$3DS;~XS{Cds-ViqBLdDtWCfV57
z-~HYUCczQp=`f?|7^Co`_fx{1Z*Qxs9=}J=^P%TM5mJ<V426k6K)t`>jOdLJ1K}#J
zAd74#2pY0H93;m9u7;tRW`jV%eCNLE*;LP-7=bAG7(X2jj-)>bjV8E0rs@$;CkOZq
zY|Z0ypDoZ&yyzzih%v#$oM%dwO@ar};fn5p!&Srl`KZ_76qzSZokVMMLmx;ha#^-k
z5;wdLAw%tiV`xr`sPoiG2-=P0elrZbCf8gvQ+oIgQvjOOCOY!O5`?~vd^&Z15D`=y
z11$TP0gA8E)CN{gyuU($K01mL6Rv)N&>;lrk1{?X<oCIYf92vupH4xiu#ZgpxDzOD
zlX0ErfB&S85ZKpquOLcUb7;FITAC(WnytOPJ-xy7G6KTR@oh&#aiZQ^VW{y>tps@}
z6!0rxB5y7#?Q>P}59K&r`I~a~k&i?M9ThlemUeyfrL`e!fTn^Q2Z4lHxp#0y_(=Wh
z>ii=U5Hc04`&0<>O~SQsF+cHMQ7=8S9bgMZd7$U@$3cv$r$rbnLE0Ozs8cEg=CW6y
zpVlXsY<U`RGJrFXe5Iat*8NaCAQT~0E}#%3z6GKjVWA)bN<R&gl#X<mXAg!P;;IL!
z9k=Ih6FUb72SH<iohPvuQ@a3D?N6jv-UA4a00ex}G?SO7R4E<<>YUA-lAJ6PLo7&v
zUUzTb`q|ZFZ#9&`eHSFfQG7F4aeRe681*ZPih}4_F@6mU_ku(QHYvKOWsqod0W;B{
z&iPzYTkX2CsqVVDZKP7|#;!htopUS%3sGP`M6kTnwopIgI)vRFHwI-kV9~QbAi4lv
zALwe5#t+AV;OB;eKFry|`X+Lxtt7<7^;`t`(>1wCSuHIswY!|pkWP}aMiL0K`=JG&
z19)HCFcfFR`b7aHM1pn&!gP@Inxedip4Kw;dO8F_yR52Kw;DWF`2|0GIv5lDg5Kcs
z3N+0yFMxu0j9un7kl8!`D;=r;I4`C~$3bOP$Q!9SfMF=@0`s4^e;$H<y>0GE>S9wt
z(cwEs=qxt@o%8tJ9$V9Me8ihU>QP4FBpNjGhT#h$QOd>8z7?X^%h9{pZa!wkuExR+
zh<*ga1aygTC4&cmr`7RW#;%Y{BAD~3(0)o6*K|wtzWjjWLuykCi)8Vl*Vof^Sa)FR
z5R=55A*0!Xa?Q37Hsi4Wn)f1<1pD~$=Z>x4Mq3l0@jBz$-sR*hF2rt0EP3>y5}Ghf
za@(!@QJ>N@xdAzRe-67~f~aU0+}B+iF$<D*U74NSl}XTVaaWotU7Ln5p-D7fk$sUt
zW}1YtCWf(V$zQst`K{_)8RQ6~TABtA%T@x*R{FNqJZjU0jFEvLs|9fsMlI|q1o^N%
z;BQVh%K`jGKkaXBYddeVwPdrE=~lMZJCkJr`Cy7$w^{(p1bAr_m!?VkO<F$yfuyx)
z5j;h_vW$EWU^u=txUh<2#h%i^n6o87GelOBNmFyPMF^_W*qYZ_uLxNeLecB<{T1F9
zd70dFyKIi_^yt3BiiM#jYo)Pn;Z0>+Q4QXmtEtkzoJ$V>)UQ8#6MXkTo62goVahwT
zP^bx@znvL}Lfv{O;<#X61jZq(y)^p;qA#E8_MVBC)OiGv*rpQR4{16pHj{VHz{zFp
z0ehxqcpJuyBw7NW-iHEB?j)f#j!bdJku#@pJ<3hM4Y$*DAm4%GEI+bw+ps^O?yGtR
z2D=j$0m)+eS}Il#gz=06n*G(<Gxq@III5DBro*Afn4-@Ix4vraDgWeMBtU*ec2TPm
zYi`Oo@RrWt9P)C1SKbzmj_3blm%evNgVGzc${Cv6QSNdAf5_~LW=cCiy>tT$z*?e)
zs(KC`&t)=iBRj!)Z({7r?ZH<DD!g~Z=aObRDX@Mokio{65C#+KRH)xvGd?zvOyg^4
zD6)9`{$CXxfg$Rd-?Ft3)0+T55Qs7{uI1$CM@dS933ga+{PNS&k5J&a)k+jXD7&U&
z0aSHbjrPvkevhxu-WXHk1=eJ_+_kB({Uyz3Exst(2H7=JRoW!o48I!Yq4&9Ok!5|T
z8N?G{0s{BZ(_%of07^gzwsy2SRJn>~s;|2{aO2b?2rs=Es($$of_M-!9!1AzOPkN6
z7=n`@0MXeQdwlf%`QAMz?Gw~0Z8#J+_?rBtF*Q*%@M*AFZ?YN|Bx%n>fOJ&m9r!vf
zk+3znt^MF|1D+rfz~Msm20TGN3akp%f@=%Adr_;0M)U=simid$?kYq!xq#t%ZSn%e
zgA`I{1TDsgQc12rK^i3%AOEeIdrpGp3m1_KuJs+L>X{qUJyLw}<Z3Pg00lAQM`*$_
zC~&;f6^bSnw9K19A1PTm2qYHuK5<jjPVcDKJ*;P>-MhkX4_NhaV1%HjV2BHBgJz&X
zC+3eg<n1?twhL~Goe&|tb(<u-S30%rmRWZI^R4a2v4XSY55NRi+?nr-Cm!wkssgxi
zz7I$>zhu9_*xcgQtWqc_TT=>41eWo<)&^c+hju{GE-k~n6IzA()A!H6_m_vn5S?2i
z?2bO%;Tad0_}nLtVglsD+~0MZ7h<1Jsss6VB|_hk11`@K0qJf1^vY(lqr>1Kmpuc<
zasW*`0CrYjiTLtmfax40x{N{to=1{Okn^B^Ymi-=Duw4hC)dYApE3LkM1`6l&;1`R
z0}?oJp?D>$DTC1{JPFb{t{6Va^*__~CFKvyZA|7^ZUq08ubok*t~(&)K0hcwc{42Q
zNCEUXGffEO@815V<97AYZI~>|Gq@J2Obz<@F(pjmdWV6wHcUcY_*$xAX6+Sbqpt}v
zlLPcv)MIpvmxT`w+D=-<j4g333CPF$Rt$A+@hpIi-kN~UPZs<8>wJA?PZ5xK7OwQT
zJSPI=-dYL#-T6}y0^+E6zWqLog_gVr0^{_2?PgDLmJ<LM!?PgMS}WDlnDksvnQ6rC
z^Kjga@v)JgY!KOju5m1KZHS3p#LXqmf}7BK!Vb3P6FO7wH{N!CmA2Tt@utKGs3hMN
z-;z{5pORDnsL9Jj*Kw(TgdCL&Kr4Or#qIBHc`O5t&-=FO8^C=bU{(YAEx?5MVL#vy
zH~Znb>yaHltD}^ut%RfD;6J!(!~!*rwgGJX@1x*G-|RO?)BFi%1ATX7Lm+j{lp#Dq
z+3kZ;h0dtTyZn4+0JQ0R$=jITK}$IU*!Q1t0khIU=aZ%iwn9)aUjeZ82gbLq|B@8s
zn*^13H@xRp7f}HAg;@Zu-1z-qw*kIuyzTMK)^;o^42^u1#nGSSz~WO&att+ZU=cYK
z45V$YxYZ0AHJMpwX;Am)9R`Z>Bvb}+?>_1uH&-N~ZiN59<!9B$eLLHop_#?mU;B7{
zA$<jb1V+G!rhS;_O_WcylZJ9eV4dE3R)8tQd)@@P<uDwf0d#6ouM)DVWHi_$9QOcY
zb~TTJn)`=0pW%*#zp}A5c+;T`{GK;z3sLtYT*<gKxV!!@$LnTgga(wRpj4mH%MY)c
zTAe)<UxAM52v=OIFRwo|2M#FNhWR1SA;BEf6<sAK2S-5B39Y<cT553}+JMZ<c!T+Q
zFV?7GSQ5d-?3i%@S^8zI9;=S=3zuxjhLTQe8dMM|=$1&2JN)ngXFrq4*e$Feg8UqS
zFqLW&TtBmTLAt7^ff#3A)7aDHT(ceCRxYR63XY+`{GcTrmT(JzXu3~-z%q#(>7_)q
zEY?JPcL11RmeG%FaJY{|43h<G`EPB&BqSoj>_2u{0F(lXj_MSE$OO-`b+hucOoZf?
zRzyr0){nZhmOrXbCPg$1i7KfYLPUq7pXt`F*WOu|A03r)WdUJ4O~tCGMp0f(e)M*J
zH(<Xv%?u;_ozd^QhrjoX-8W&QO)c6<j^k)oA{BkY9>WSy6G9fqqDbemlF9Q@O4CG<
zi46WXNHj$OZ!WUkY{=W7-8wyLdgpJ6lXxz=X|df0w;BySqtE3P^~<ElH~{I_@A_%h
z(C2f-*_{?6Cm&wRFDQ6!`w|kzvM)#N`mpsS9<UjZ_@kPI`YMnH`NrwtMhbRpFPg%G
z7!@cwlk|eaXUS|%8HfSw2rzs;v~X4b;DCf%KD`o0fFA*ZEg(DAYg)nCXGx%vp}E_W
z0e;<E^0!oPb)YcZP22=~R&H+OkOpkRn3*MTu}*qr-}`v%!ZtY2@?{|2q6s3zDR2TF
z4?l>><m&Fe8<<gy(k8N-i2LfAv-&RV+2mU!8e#ZV<ZPw(T0@L<JCN_;&qj}MBWX~n
zSiinY#0eB*?<5Wq%M88ZcSW6#_D9BbZ4Cuh2kBxMzqYP!z2DqT3F&n?mwB<u|79+I
z%QiH+YA-A;9jm;ZJ5u2pAGPOMKzTe3Aqk_4QAw7#|FfGd_qqdsuXz?`_p`IXip^>2
zJsez<AQzI=V3?+fW2MLHkc;i#+0#QH5E{4n&eq07L9vY6U93-yBkX)Yi>yCjS=^ik
zt6s;R(I>P2v-CS^%oo|l^Z;{%JF?>l7bGQ4n`)4<Up5Ut;*To=loZa);3=Fu7(y2%
z8iWBd+OVjSbF`(!J|jQxNjoDKg#17=WLwgb=tvU1lgXOgr%#`LLHRkiq@>s2Ag4Y7
za%zbU130jEp{8tfuab(2aCFm3D0bmHI)R2EN?#>%VoXX*WH|fI-jVDxU|;Q<VRk<N
z!Ghkqg<fD+7<nBiErLl{;~9N4Aj9jD88x;Y8LKL{A+AEVmYtDVR6hhw$p=CARZW0Y
zcwNF(yw8w9CdSok5173mNI>^F+COQ%oeO%eXd&_p$(gz<clV7mhSO|jnLFZD9Fn4y
zN(`Z{b}yJpmOC(7=wycxzLPOl(1n8gMT2=S3XCN)fNcd*e^p0}6V?t%E5)Uy)}MD{
zOsK2*s8Lbx59r<k&-{R5sCNKw6NHX7K4s8f-Ja_U%{lg+BPI}Y)Ne%b<Vhfd(%1*t
z(lfAa<{r|J>Q*!fc~S8eu7@!-GvEab^ZWwD0RO3@4acyn2%*5RV-NoO<9|2d|6fXo
zOD+A;nQ>NEC(Uj3Y$+{<KZr!ytKaEciyrfrA5miZCP2V(sm}Wd3JLoeFiihE`hV3f
zS*2ym(mJAPKqh@CP&&&I^k=r%UK>M4Pg@eSfo4H%F!ASi@wC|VyquC(l|c|DIgwm6
zZA84!&3PuB*psLSvIX;o6jn?n@$;=C=YJ*t`xnDkzZSbgq}Q@I`NXvUdG3jZRgSY)
z*c<IS7R>lL5>27>G0qEPC7Hw-*X3U|)-4k-70(LvIe+ozZLTYI%nf(C<&H%D^L>4(
zgu;7w@;KSZQv&oH6V(?<Se<lVSZ#Kg>tF-@f%(aURAEi~e9Num{o$>-S28C>TO3N$
z-BuTWeEFWTTuM8Xz1F0b0yA`6X{@}wbEHgqLD_!ujgN;PbI1p4{{26mN3P5;|61gV
zXie~q9GMfq@J?uN-Cbksy%=9E!-So=?<=}rjFpn;Tb<r)6cMp*S|a47_|V3GuNYIT
z{S><ZpI8|${n~EBFufJYznuKlL<{RbgEYv-Rk4af_+m@EtfEJ#+)4y-p*V-IUO(9P
z3kLp~&N<x{td~6^TD23q8i3@TToz4pym%a&mdfFugz>@Eun^NH_eS+p*5c(PA8yB(
zUXma3qG{fEVe|W{_H84n;1rh^rYcA2-9D6YHwlJZ_dP#%Nv`1=gXJB`jioBp(obcM
zVf()0U&}!Xhz!!58&?G?#!j)h&tG<b+y|4E-Q_awV>|E__vYRk*@0-)8Mg;`$r#*w
zl57HDO{njqg50MY^iwlp@{a5TvDjuGRq59^-&13F9i+eAPZIlCJD#JpTk!DrHjEiB
zGOEmf&V9;Y;JC6PM{wpIX%HN1DM`yE*2g<O1LyQ0&|7PW86)9vzD0<t8LVK+sbNm&
zI*^5fz@iw*tAf-hX(O?j?ab}apeU5=&GscbazCbq9Pv1|Iu23J7cRE=doQ5a-ww`x
zLUD?JvUDx4db@@Y4}Y+Y7rvS3;JYt`0*gXDnMdVb(lpCgKI!it?x{`WmfWkfVkv<?
zvqO5~`K}HPp$Nt$F3}a<4#n7oZ5S;A^%qZ6P>m!_P&pjItzQ#zp#Lhdnqo%Z{6uW{
z;%nl|l<$+SPIrq&rc&;5V^2}-_0|w7Mnbqq91s|B`DkRnkKNK22@<WS6Q{?m^px*b
z{8la9Ax3t9f+PR4HADukN42<wDB$Ve?qOvh)KrW_KM8o~{5obSY)NAF7m3s8BPDiW
zj!c0B_6jRF!*&;=n0Kt%De8T@cRK%Y(e!E;a^uM1JSrO8`fgobMr?clUKg&QX_^p|
zmB2721nrI1>g*%5OpCbuaB)5OF({I_OcYQb<FC-scKTt5^2YJH<n6SS$J0~qu8q1l
zctmnWk|amCZAE(P{q6PAP`l&gWyube=zi9oW$vRG7NClDXMmSR<=#j?L0G+xu&x{P
z?)x>F4WiKhnX*{NdC^+j<6Z|!FZ9(4F^g+NQ`X|^l(a#OV4W)}{XI^zlb9kUtf%to
zkj4a%B{O;ZWxecd<8Mli9Bj4zDM`}|%rS>yYOdtTo(StFG`xA{J+b+ba>t^k_!Rq;
zlZ%HF;~%`QHDs?mep)%>*B6P~zZ(Um)&~RE4?dx}KHniDrWGPn<^03Fai0u!fkY<T
zpsQXQYx7pBO$ox=r0Xk;ho0NR!D&(1;SUZVF~IV_J_>yQHf7j9)^%@4Ok<kG+~OCy
zp}|v1ElRO!G6MsGSlkMTw+qr&PFB!JZB8~2>p5fn>}I)-lnE=V4!t3q{3s@R<I{1t
z{Z#ko(<)wec>Y;6m4ZWPeL(BO-d@D|UR&2kjEC%Iy(MJ8wM3UcM{5B~s;ICM1&PC#
zdAGL8pN3Q7`Rn5QhDxd%u41>93Ba!HQsdjPM07Z;RzR8{oOS8>X6BLD8aErtQQRWx
zt5Q~4R#S4eF8ml^BzULqwPe5VbaTk5#a4*)>PU~2L<Df6X>h`?;-nNLg-Is%DO*0m
zZ2BdlJhcu$xd`yVCwhmA|K{Oz9UjebmNLiFsNhGlH97f73dF(Vi^UTl0;Ni~ML(h>
zJGvOX!<UcYQuUt6K8vpP7Z?d~GkZor7Kz^U>sY91O_;6G@D&9)`LhrrY^VhTYZ~W~
z(IWjPRGhbf;`?8U)<$VGJsNYe`I-PqC}vO=l+%$O$~%BxA^=?O_u`Kml7bi6%<<{t
zenKPCqbo_iE{ny>K6YkjK$S5VjByxMuFkZ$wr5{zR{TXc87uG0FlA+z-&1hT#pd=@
zPAte$=rN3Oy#vuz_M>yk=1Ggrk7$~i3|8$z*qzn1`M6@@Ee^uzqbn5^suGH86rS6V
z9!mU~1lu71-83M&gv2SiTYVxjZduVvo+69|kD5eZ;6_^S+W+f=KS9;7c`OYXZKQ13
z4SbyhW#(MEU;rE*Bmtvi*oAR4G|U3+2uuvQ$r$$Z{C|J^?<V}063FbpIt`X0NYCqs
zGFE1U)|^d6Hn(Htr;v87(mIq;_FHv}lLzFGu4-BEj^7W!20B0^*|*A9N`;k^g>_YL
z!MSc_eAYZD=Eax+-psors>pIXaGh9d%1W9D?o~*Ll$kT9_#>D6cX}_m_V23)v2OXo
zZ=q|$qJ>7fWLZn1W-q<_E@mxP9Sl!GQsFOWj(&n*sS_~tCN3UH-8mLH#K#6m)7?<T
z<7-wpM4w;M?pQeq&tE{ku;azN?OH9PCAP*lBL*74{`)z$bcE2x*zEK=tc0CtZRW%C
zqRblEb(=5gJ`76#{_#Z@$B5xCq7Uw(=hJSNqZ!z8{IC#xHB5Ts(P-g)QtVv0nts0=
zCAI@?3EQtpmJWxyl+%)o_}_ddp7$@hB6ooWuA*Z7awEjN#m3`eiU+|!f_1O&2jN%F
zh34Fr!q4aSIa^m*wZLxuC2!Yx(H;45wEo<ML$*((zn@;UdxFGaD+-F2GW(6L?sehH
z<&Ui~U8|H{kdp?D>DYfO_UI!YumG7urY6KJQ}-|sx+?1z66WJ7BEgEues_vTXL$eN
zFU<dp(;_}!^-E2vs4uA2R?1B<bnLOC^J1b<JnPGCiv*!{p`)VFvpTEohQ^O~k0Uuh
ziqN}Mah5*vUUvpM69dkRrX;6>-R9b)_T*)@-MitridGtVeDm}+=|h+(f;~@68M}h$
z#Z8Rm&&^T*oy<p8Ne?_E%q~q*ymA#ACP-Q}2v)S#8Prp8>QBwqw<>HVu!EjQM=bDx
zzD6Y8hk1Z>0_Jrym5pJ&-`I8G^PESE9=)62geIT4m)*mm)>L|o1%)evQbC!APcFmV
zMC2sV2fM&_+_1{|5Z@^OEN@JBc%Y^9joa!MR)W%bnr5dARjNOk3<;w^;^5CM1$|Z(
zU0^R>2ERvM<{8*9oUuirlE<00?pcsDBYXI+!Jhlw@Y9%&J*<FJGpg)zE=>sJhNCUO
zqeC{(IS{_PEj!Y$>VjnG{Ad5dD-ZwKHej55@ih**FD`!5%v#N&E^x729-n2Uz^wj6
z=h2}*32L(^XjIfo-32S|DPzVFWO_;p?{^H~MoJQqb-x_AqN``~S(H~ShpHbqh44#K
zsI6G4`kK~*Z58S<^KB&m^}T!OO+MHVyfrufG|BDt$9T>SxEl@w#3{m=5DwEcua3*(
zC3L8<1Y}E+|DbEwZLf@ZBCAPMWFKLn!%mXHFns-F445`m=Ok*wj9!dt8bJc%PBf98
ztN7+`Y5NEdkxj0Tt|Q{DPl}G9;qQ|te)^5XjMSxexv{Rj66@J1m4ZVx8C~lx?4tkK
z6HI#iF&*E#t9UOLjXSkQ;j{H=CRwG3(D+bR#>zESGyKELW_w(J{|q;u)545T{$a37
z)Xd-_|DRE^YAsa^n^%@x=a{r+=PZqeGc8^BV=VII9h2PKXk{8ve{kv5&$#t!I(BgV
zgHFPHXA_9K<H=G><%*YG={)lScnQ@vC_F<U%z(K^gwV-6&Ll!JFV<Y5dvw`z7>wl}
zdBoLE{jipm@y9^vZQl+Ttp=V#feD0$y~<Yo{j{He+~6Jj7AsFGm-p3ME=Q9hNl~+f
zq;{~T!p}hv2>}LsV%$E<v5-bu_G}(xz&raJHWFt>Un#Ja!KD;(0#8nHUJP4OU%|O(
z<z-Q))!A_VdCQ$!lWDC+2b&He{9%}X7&v_P73|g(l#cPpnZqqn1MK%|jwfM&CY<ru
zn-$Xw+(b6DF5LGRjwU{IK5l+Cwr0F=XICIvkq-QGjJ9f?`~&!5M6INSH7|q&JgM<u
z&9MAyP_235W)7n;<WC!spEcdAe^}ygsOx-msV)jU-42aapCn7kn4#RM&X1vO;bn`0
zBT1Yg6Q(wf82J)&oJXW-R5Oj7X^F<TJX;pK@fC#@#=9vcVLRCJ&flcXdpF;+eh9mY
zz=_dz<mEn(OCj9wT5ugHax^wuY{Z?P;!PyVE5V9uC~NCEHo*Yp`AFeOh3G+J*NJs+
z236l1&7yXR6Nto|ex@g!?AvHJd`NtJjnmbgJGPeQ9(L>^I=cJFRhHM5sS{qGuA~30
zf}iJfvclS&e@}sFIQ|S=b)OdH*FJ&mcMXNbIfOTM0-UnSgGj1|l5@}7D=qnUVXNXh
zQZ&$r=g^3W04hn|@BLZNAg{bK(E3d^>DNzTrPxy3-ahX;rHMu+?IEq<DTMJeORg6=
zdpyRCrx^qGz*T946dNMdN|rmbvNYc7>r5eb@_3kGQ|w^UodW$*suaup(c!gRl5H&q
zIcI#1!<z3S_h$g1DE%jdbyioh`CY!UNJUYlaF#ia$@A%WwJ+?{DmXM(-mrMmNJOQ~
z&9Oz!5`kNfkqGezDYTqA$q^yz9a+jqnAM5*`7Jr8%*Ag|GZe-yD*0{QC*-1eH`H72
zd#~8yl!d;?guZBalf`4ySwsgQp&N3+BdX8)VG4XkQ!D%W>~a*J^)G&rs8bo2%(jsc
zQ|w1?k;4UCbZtQJj5Xtu3Cq?r{*|**U!@xBM*3NWia3o_&QK1)@=4xk!5s4^4Q0eb
z3V9`&XOnhy`INHh$BwSdaP6EKx6<O$smijw<?b*)0<O+Qq%u<*yrR8B_<u#1+WNmG
zo+UlcCzu^-F4;IzFQ=gDy7akt(zxe<;n=kPQH7Q^p=TPWgCUQOA9yY(6<ihRqCN?W
z8s+?+XkGETY>A5?XIw|^Mu)HKARaS3A38XX@TSe<KA9XxMJwY&#CLz#c@OJ<1v_DU
zzvS#TQ78LwF)c5<+4idLD!nUqC6TQY#PoY%oSzKO+uhzXl(bg#CtuMs)L!$E-S<fj
zyVq2lF3vn(R-H{9&T2##dn>jO27mjfe_MFl{*0LNHp}Lu5q(V3e#4N{fX5*pD1mS5
ztOm<l{|YCzF6CiO8hR%qo5|tKA2i=E+CI(Y(>oe!W8DrmA1v+T4Md29LB43|78QuI
zDVlqgV&GAea}vszca8eBxwDXJJ(d-{?5>b_KK%>}_vh_8$6(Y4tcI^pJ={vTBzkjM
zT(U0P04qO+PJ1TB_G4LdGG~S0_nDTox7|)|4z&%$JaLzr9e)(&e_bRvA=rdGkJYFx
ZHUZmV=`?@KGWZK7e_7?yTNz`o{{h%3EExa*

literal 0
HcmV?d00001


From df85c51b37fd16de87000374d989b23a5023ec6a Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Tue, 1 Mar 2022 11:40:57 +0800
Subject: [PATCH 07/28] =?UTF-8?q?=E6=9B=B4=E6=96=B0supervised=5Fmodels?=
 =?UTF-8?q?=E3=80=81unsupervised=5Ftrain=E7=AD=89=E6=96=87=E4=BB=B6?=
 =?UTF-8?q?=E7=9A=84=E6=B3=A8=E9=87=8A?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/layers.py             |  1 +
 graphsage/minibatch.py          |  5 ++---
 graphsage/models.py             |  5 ++---
 graphsage/supervised_models.py  | 31 ++++++++++++++++++++++++++++---
 graphsage/unsupervised_train.py | 21 ++++++++++++++++++++-
 5 files changed, 53 insertions(+), 10 deletions(-)

diff --git a/graphsage/layers.py b/graphsage/layers.py
index 24d2af8f..57ec291c 100644
--- a/graphsage/layers.py
+++ b/graphsage/layers.py
@@ -81,6 +81,7 @@ def _log_vars(self):
 class Dense(Layer):
     """Dense layer.
     一个基础的全连接层
+    
     需要的入参：
         dropout ：0-1的数字，表示输入被丢弃的概率
         act：激活函数的类型选取，例如sigmoid、relu等
diff --git a/graphsage/minibatch.py b/graphsage/minibatch.py
index 0a1cd963..0af16c82 100644
--- a/graphsage/minibatch.py
+++ b/graphsage/minibatch.py
@@ -151,14 +151,13 @@ def incremental_val_feed_dict(self, size, iter_num):
 
     def incremental_embed_feed_dict(self, size, iter_num):
         node_list = self.nodes
-        val_nodes = node_list[iter_num*size:min((iter_num+1)*size, 
-            len(node_list))]
+        val_nodes = node_list[iter_num*size:min((iter_num+1)*size,len(node_list))]
         val_edges = [(n,n) for n in val_nodes]
         return self.batch_feed_dict(val_edges), (iter_num+1)*size >= len(node_list), val_edges
 
     def label_val(self):
         train_edges = []
-        val_edges = []
+        val_edges = [] 
         for n1, n2 in self.G.edges():
             if (self.G.node[n1]['val'] or self.G.node[n1]['test'] 
                     or self.G.node[n2]['val'] or self.G.node[n2]['test']):
diff --git a/graphsage/models.py b/graphsage/models.py
index e3ea8e83..8abd46ed 100644
--- a/graphsage/models.py
+++ b/graphsage/models.py
@@ -537,7 +537,8 @@ def build(self):
 
     def _loss(self):
 
-        # L2正则化项
+        # 参数的L2正则化项
+        # output = sum(t ** 2) / 2
         for aggregator in self.aggregators:
             for var in aggregator.vars.values():
                 self.loss += FLAGS.weight_decay * tf.nn.l2_loss(var)
@@ -556,13 +557,11 @@ def _accuracy(self):
         ②计算顶点和负样本的"亲和度"
         ③将两组数据拼接，拼接后的数组维度[batch_size, neg_samples_size + 1],意义是每一个顶点和负样本、正样本之间的"亲和度"
         ④计算正样本对之间的亲和度的排名，排名越靠前越好
-        mrr值，
         """
 
         # ①计算正样本对的"亲和度"
         # aff值在本实验即是两个输入按元素点乘，再按行求和
         # shape : [batch_size,] 表示了该batch中，每个节点和其邻居节点的“亲和度”，越大代表越相似
-
         aff = self.link_pred_layer.affinity(self.outputs1, self.outputs2)
 
         # ②计算顶点和负样本的"亲和度"
diff --git a/graphsage/supervised_models.py b/graphsage/supervised_models.py
index 9ea123ce..ab1c7c70 100644
--- a/graphsage/supervised_models.py
+++ b/graphsage/supervised_models.py
@@ -27,6 +27,9 @@ def __init__(self, num_classes,
             - aggregator_type: how to aggregate neighbor information
             - model_size: one of "small" and "big"
             - sigmoid_loss: Set to true if nodes can belong to multiple classes
+        
+        
+        该初始化部分和Model.SampleAndAggregate类基本相同
         '''
 
         models.GeneralizedModel.__init__(self, **kwargs)
@@ -76,22 +79,41 @@ def __init__(self, num_classes,
 
 
     def build(self):
-        samples1, support_sizes1 = self.sample(self.inputs1, self.layer_infos)
+        
+        """
+        输出特征表达outputs1、计算梯度的流程和SampleAndAggregate的基本相同
+        
+        只是该有监督模型在特征表达的结果后面加了一层全连接(Dense)，用来预测
+        """
+        # 调用的父类SampleAndAggregate的采样方案
+        samples1, support_sizes1 = self.sample(self.inputs1, self.layer_infos)       
         num_samples = [layer_info.num_samples for layer_info in self.layer_infos]
+
+        # 构建聚合器并输出节点表达结果
         self.outputs1, self.aggregators = self.aggregate(samples1, [self.features], self.dims, num_samples,
                 support_sizes1, concat=self.concat, model_size=self.model_size)
-        dim_mult = 2 if self.concat else 1
 
+        dim_mult = 2 if self.concat else 1
+        # L2规范化
         self.outputs1 = tf.nn.l2_normalize(self.outputs1, 1)
-
         dim_mult = 2 if self.concat else 1
+
+        
+        """
+        全连接层
+        该层的输入维度为aggregator输出的特征表达的维度
+        输出维度为分类数
+        """
         self.node_pred = layers.Dense(dim_mult*self.dims[-1], self.num_classes, 
                 dropout=self.placeholders['dropout'],
                 act=lambda x : x)
         # TF graph management
         self.node_preds = self.node_pred(self.outputs1)
 
+        # 定义损失函数
         self._loss()
+
+        # 梯度计算过程和SampleAndAggregate一样
         grads_and_vars = self.optimizer.compute_gradients(self.loss)
         clipped_grads_and_vars = [(tf.clip_by_value(grad, -5.0, 5.0) if grad is not None else None, var) 
                 for grad, var in grads_and_vars]
@@ -101,6 +123,8 @@ def build(self):
 
     def _loss(self):
         # Weight decay loss
+
+        #L2正则化项 
         for aggregator in self.aggregators:
             for var in aggregator.vars.values():
                 self.loss += FLAGS.weight_decay * tf.nn.l2_loss(var)
@@ -108,6 +132,7 @@ def _loss(self):
             self.loss += FLAGS.weight_decay * tf.nn.l2_loss(var)
        
         # classification loss
+        # 交叉熵损失，激活函数可选sigmoid或者softmax，目的是将结果映射到[0,1]之间
         if self.sigmoid_loss:
             self.loss += tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(
                     logits=self.node_preds,
diff --git a/graphsage/unsupervised_train.py b/graphsage/unsupervised_train.py
index ad870855..373a3f31 100644
--- a/graphsage/unsupervised_train.py
+++ b/graphsage/unsupervised_train.py
@@ -97,6 +97,16 @@ def incremental_evaluate(sess, model, minibatch_iter, size):
     return np.mean(val_losses), np.mean(val_mrrs), (time.time() - t_test)
 
 def save_val_embeddings(sess, model, minibatch_iter, size, out_dir, mod=""):
+    """
+    输入：
+    model：训练好了的模型
+    minibatch_iter：迭代类
+    size： batch_size
+    out_dir： 输出目录
+
+    该函数作用是，训练好模型之后，再将所有的节点都输入到模型计算，保存其特征表达到本地
+    """
+
     val_embeddings = []
     finished = False
     seen = set([])
@@ -104,18 +114,27 @@ def save_val_embeddings(sess, model, minibatch_iter, size, out_dir, mod=""):
     iter_num = 0
     name = "val"
     while not finished:
+        
+        # 获取batch dict数据
+        # finished是一个信号，如果为真代表节点遍历完毕，则退出
         feed_dict_val, finished, edges = minibatch_iter.incremental_embed_feed_dict(size, iter_num)
         iter_num += 1
+
+        # 计算特征表达
         outs_val = sess.run([model.loss, model.mrr, model.outputs1], 
                             feed_dict=feed_dict_val)
-        #ONLY SAVE FOR embeds1 because of planetoid
+        
+        # ONLY SAVE FOR embeds1 because of planetoid
         for i, edge in enumerate(edges):
             if not edge[0] in seen:
+                # outs_val[-1]表示的是 model.outputs1这个变量，
+                # outs_val[-1][i,:]是第i个节点的特征
                 val_embeddings.append(outs_val[-1][i,:])
                 nodes.append(edge[0])
                 seen.add(edge[0])
     if not os.path.exists(out_dir):
         os.makedirs(out_dir)
+        
     val_embeddings = np.vstack(val_embeddings)
     np.save(out_dir + name + mod + ".npy",  val_embeddings)
     with open(out_dir + name + mod + ".txt", "w") as fp:

From 01accbdb844900a786716fc2144c6f559ec73772 Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Tue, 1 Mar 2022 17:00:57 +0800
Subject: [PATCH 08/28] =?UTF-8?q?=E6=9B=B4=E6=96=B0=E6=B3=A8=E9=87=8A?=
 =?UTF-8?q?=E5=92=8C=E5=9B=BE=E7=89=87?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 ...202\347\202\271\346\265\201\347\250\213.png" | Bin 0 -> 56391 bytes
 graphsage/models.py                             |  12 ++++++------
 graphsage/neigh_samplers.py                     |   6 ++++++
 3 files changed, 12 insertions(+), 6 deletions(-)
 create mode 100644 "graphsage/doc/\351\207\207\346\240\267\351\202\273\345\261\205\350\212\202\347\202\271\346\265\201\347\250\213.png"

diff --git "a/graphsage/doc/\351\207\207\346\240\267\351\202\273\345\261\205\350\212\202\347\202\271\346\265\201\347\250\213.png" "b/graphsage/doc/\351\207\207\346\240\267\351\202\273\345\261\205\350\212\202\347\202\271\346\265\201\347\250\213.png"
new file mode 100644
index 0000000000000000000000000000000000000000..4a822afd248ed0964d867904ceafcd2f0cd7cfd9
GIT binary patch
literal 56391
zcmbrm1yG&M)+S0IL4$?h?(Xgu+@0X=65J(dg1fuBJ8TH<?(XjHaNm6AoWK5iXXe(_
zOp4U5-L$XnCC_@A-}16z@UU30U|?YI65`(!!N9;xfd5|5;6MqIE@dF_0pTDdp$rWT
zy}T*60er=D6j66nvNdsZ(YH4SGqtg`Hl}qjv^O@kaWJ!WJcsDu0|O%fllUg6?3#A6
z>LP$5u?%*xND`ovAFvjhEl$t&El<5V9iE5#C>^ym$F$P4mHURHS4U1ke%vo4<db|J
zrC-(*;S@JXz{x#S`@P#2Jwj~otCu%dUHaq^X*T`Llamo<jEHUVj~vO_P<J>>m@oS7
zBmD&AlnRa?fcYIj56!>-oG-{g#Ms@LSa%xGGnMkP-tMH%RDmnhZg~rcw<O1WseCLU
zzrNi)Y3wkLm3KmKhXY!Ipikf5e<zC3{Loacp?-U$@!}Ht_4*byU&F%|VZ@m2;$LB9
zay(%~A%@yiz_$L0{ap}a*t;zqJb%?4Xzu&|MCCkF7_>$`b?H1?!2aIY)VZ+RZt|ro
zG7*{(=6W|9F;Z?eyI%JOPF7I)VyyO%mQHHf72S0=HvIaoy=Vhw;UmzC&S7bkM7-g^
z<{*?ykTDy}`r6m&>&w~%L1;qd%&NuJCAWox>TcYyGB07Kg<t9JJN@2YlH$@MgOhQB
z^uHRnE(#|4sfN`~BlR;-#+sYfU?B^g%&e1R=_R#wSA9TxDw%Q=87}VVS9`y|qpT!4
zt0b+sN-B8|<#K+i#-}|<3fo@LH@U(H(zky_Thn7D3_vpQX=`##@;^R+;j54i=Ua-+
zyuW4m`hEvre@B-hnS=0FccHA>rLTKFIc(!qVSX90WCmX8=3sRvu>3)XIB01xeD$&}
zE*PmK9EKrw=Bo9(5X(mhKv0Vp@AFAsePYOj0Y=HhK5O`I4QbEkO7)DI5X96!aBN3j
zR<V?lpUcqto0~hGlSYcBZNJCQic}&Z#((o7%k)lH(olMQ+Y$e6RK-{Pq2I1;^ORmR
zY3*p<i6f$xd0bBU8x}u*Wv5*m<8`C=*FHJMdX5MOhlnLb<vrfY-pjr+{xZg1iR|z7
zS~}WG$ZB>*)%5=U+i77C4UXj^<E3^uriG`BVv^JNDx0+RZ*&%h-UjkduL+9_OWYRa
zzf;gd>+-eU9tJuu`$h<$7~>ZsOMkRIJ$^B?yDi=nk@0i~PUA-F^{otV%GeEcv7CQm
z_{_)?xmjRjV94#b{-ag^l8~Y-c4PA(ae!F}mt<T?SIJ&5GFi4sQ6U+5?v(AzR|nXb
zDfU^_m;~l499vP!aoYX2N|P`u*+PPV1NZH@m!j`Kc9j*iQ|)kasm+VT6&NLi<?vA?
z%~dJUvrBp|K34qZR)Qq_Ht$#!LoMa{+FZSB2^=`x?fVD~^BlUvon4-8iEgJJq#VXH
z$P!fk_RK+1rdPy5_ODtPx!~-gAHcHI9`^RKla>VD{TpBI(Z~bN5Y?OZf1Ueff8Rt=
zu4=iN0JXl!_J1|H9N8gTsTSeI2n->7@ez*uO7&@zW2@FvM5{KFj4LfcHxemb{Jp5|
zefAqFO74ag0=gKqFh}ID@Ryn|FS#nz=Uc}_933t*k|?*2=DiQWC~$oul6ug#6ftab
z2^;ro!2R-^Z%0U$Ie{dh6BL2|UA!Q@h&x8XX8*k3*&hU>C_%mIX{grzTgOJIvh8xy
zDh>>Oiz9-8s(xF|go!N{O1zBcJ&IMIDVOW)D3q*UMR@jMq{{kM+Y^_`((nP4tC~_I
z1Wh=+Ti9i4{^t`Z7}$-DZm;hWy5*j>{_DFI8cfj7FDjznQ-)|GJUSKa?I(#GwHMfA
z^<m-6myac!AL3bXJF@<?OK=xHgyEmNBsE@*jb?`TXrNgFV`vw}|6N)PRsj_SJtKa>
z7&w?f<z41*LCVASJ_=Q8H+<!m@j$ZHUf3YnmNKR&{?LBnWV2P8W=hJ9@tcL|Mm!^7
z;`fBNS{f3rm!+koTNG#2aIK~veKY3FFFd!4i=Ui4_e&bC=#X-LNld!sXNGY6YeYrW
z>HZSbfU;-m>X2Y0As$&?^kB?dFtCZ^mv?CbS=oc5YS{uv>3%}I3}J1nXSp?Gg+HBk
zIQuF*%N+7Bgp91ijMVIrN}{m-5Of_)O>^2G^vRSObw(+|9;yFme<0kZ)U5Q@1M#$X
z2_q8<uR}!gIa~%cc3==PGLK(w_qie-nI2YOxwB#SFJ@Mp-QW}WQ<x}$N3L_!uKHxr
z2#NBSJu|<IXs~mB(;wqN`mX-AcVn-{`HeO6Sxv~;aB|gH^&R)uY?@o?!}4KCc?TR$
zr59*X0_(T=4R-V4<}2Dc1xsOjJ(+9#rO(4+$y9NSq=t!XZi4RE)`sh7Xg-4>8W@;6
z@@I7Ci0{rrCZXQwKAzVmhPpo?otvBYO~=IFC_P<#zJDgvkr~3*VJpq6tRhNDjEl;K
zO92ktgrU=@Ct6!^wZxwRE%a6=>#$R!W%|Wu3MTi4f{-s$U;GLej)%yAnm1a!Ca@G_
zWP-(Gv_;ktZLY7+aILlyvfsC8mUI8<$^phr53MqSn`7MecK7ouG8~%iO7%eVHOyqH
zWbWN--5HmZsz^i^9esroe)HYDQ?8Sy(FlqFIt!pgviiTgfuw<9Hd-ePAx-`oxk<k$
z7~lg5oDc>0AowWscNPNB0e@@xB0(nuA7F$?On<8i5g7e7<^ci&fOP?b{s}ZhnhD2d
zr=q47S%m`&Kmghb$S2b4o-H??$5X3)TSW#7fCCDBKL!Q{#*)ckv05mrswNaYgZozf
z?Idb1)90<as%oLp7C&(q)kznO0RQj6$rhHDYLz<Hwzl}#*wjbuU;<2kCvrtfN_u;~
z?lYMq!axh{Dff3e0e=14pFVwJOqNbU``C_MxA(pb^0;qtI{vnd3ijt-Oj=r6QBiRQ
z1m=qaSo)kZ2(T>E>)Tt<kIgQ2#3{47-cWSg)fN>|QBe{QxUVCNf(sbf1#V4kEgS+u
zLv1aJlNcBQ%E~Y>R|!$ky@_n$>gsAF`p$c(3Sdj}FnQe_<>jLq8XAhY!F}cbsQg_k
z78VvTd63;UWRKJZuKyuF7?|K{lU?`htcs3~&J6Acfu?_N<ext@z()RWZ}$I@l(K)f
zK&<2IY}Us^Inabz<m!sQeEq6N9R9mS3?#q3e^A>BE}Tr>_3E_AP3I|7w!G{uGdV-+
zA$7e9nh;W)dHo<DzCji1n6%|fdW<%Dr_0m+v~uVIpK~T#8Nfa{D%vDD%No)`4EnhL
z`ujC`4Q_8Vv)InKzuR8H1qv+tk=@81?)>!CJdf~GcMrVP$?2&zP58{SnZ|P&=m~l3
zN!RjTK>v${dk9Mp?O@TA%gh&aJC_czQy(=(7!eVP2Tt(>XGB+FXHVALNQ|6Pj49?=
zK8J5Bx?llN?~>Xp_O#!wTT~`rGSmDn7LTA0D}Pt^cg`>I%cq@CTdHc0?fohVUI*Rr
z`-M~~2)_+jd{u$!V7JD*7~)33RqdbOb{$;Tk>yrgBs!%86PlT$w`083zxJGw-}6`N
zmC;7)^2o78RPHXU5?1R$w@J+-COfLQMZ3^GHF)RU_|n406i6EWOX(##*EU}i-RT?*
zDWL(X($z@?UR+X2DPc)u^{}{|PyUvEK3v;{`)KE?c>C44tpdeq-aYi)Q_XGmFw$1U
zc)z3VnxGlSph(`+>(3!st=hpluRG0y<J*IiFsjHZD_B55_+r4yR8rO`yPJRG8Ai#3
z6YkN^c24=22&h;kk<~(-z2$Q%kvXGZ=Wj;33irpYx2GQpnM7?C`GJ!UE^wd`Bz+c(
zy5{e!giT!=yS78=R<^_{{>kL<8UjjkXM^~_;Kv_9+j)DvguRycv+2TdgBXWnEZB+&
zS;QYqSg0>BCdZ9n|GnlMNLpHJ)wpv_#jBL~E+DU_wM>c^5r-n$KQb${^8*w}snv3x
z;q&tqf0oRTtZxW>esqvfI5%9@!P2sCr=X4wVKjbOrXbQs>z@#$_7}*Gj2s(T>`(W5
zyp0=~tJsKvdxSQVTZDHV^2lABDk8>Tg)1Q<U1%F(kc`P&Q90y^pzsCJ{UH$kp#S?)
z11c6m08bXUP5=Ko>rjD^`8~GC1E`pFf22S--7i1G`{3O<jtop4%-5z;EKr(&=fdF|
zuq$9$%l`j2y7B+$y8czF|I3CcMcfn1mp|H?+H_<x?zYDAJrMD#Lxz8~g=ANp9WopL
z$s#0;zxCQ7VU%82{i~x1m1Z)?b~BrwMVTHa!Uxp6iPKJRo{-#BRQ1Q|22q0DA}Hol
znoK+O3~4{<-~TlL0n*5=4-7KRN5)1|&vg^!ieDVJ=vP>5^X8}cSyiBV^S=I7+`e>R
z_e>&(Ub1f9b}|dS8^Y9NE)+UM0VD?oVyZH(-uFTytwMxOh+trU<_U~KrUIdgV}nx)
zOi^3Zu^hK{_aLzXF)P=+sr1>|qSx4gVb^`5D4EJ?mI?<S`5@qtg}zFXlV6Dco(^N#
z-J0f)U1A05cnS=BO&*Ru7vne{mUOb>X(07bnoatOk@^V?EZaYd63fZztQ<Y_h36So
zCqBK}-ik`<;%_un2QO4$w&M9Vxfa$bAA{CLg+lun6Y{hb!qr&{*s}>p=wzZl-IqYm
zh49S8`D~xN(~>^_)2lwvE4W<j6{=AWEqvTy$;9ts1Vg~*A+SPX64}`5W~rfLOyy*r
zHC#u%`jOVd1sF+B2q6<SOIh6wlTu-W^pX%oN&7G74$!-Wz$7O)k(gSSdV|e@y-?(C
zMXF_$mA+kYVPFG7Mxl>{wNmlQbNn)6Kx+Xy(*k!#$q$FKzyu%A14iha-n@@~kbjv%
zuQM_&Nl4IE?ca;hVFr%73DBo3ZmP%iW8t8%OrSz@6UMBTF6SGs(HIaJ>Hmycbh6=l
zKbT0&E~lc-drBXj&DD1TY&<kP(5sNm6L`DU$gDmw190f@pp&%&Zf-u&VwzLU%{ahH
zSq9Dnr|9PI5YUH*a2j0<J3Y$5BW}PBkN{<V3{&VAyIld*-+{(xKx091ppv~2)7Dz8
zE6+e06d2eJu!D|WQo0k0_?!}~B8PEhQ*>s)a^10^lZ9M%6<VVHc42_f$ov6JfP%mX
zRt;Fnt;D~Idz2$(0{`BJzrAR&XZqAJs-o&V`6ZNvt1#IQTOMJzmWcq-A1TmI!Qsgd
zb8Sz1ks@8Ro=w&36Spj|{Kdb&LC>S#fa?RC9RZC0d>u}P83YP^J&~9K^N0U#Y8K(R
z_0}$$ii_9uUk3mfkVZS9-It!cz?~-V10V%O!GW(K71FwY(8ET7uZKWay|ew=DeYOL
z<YdHrugs0V$x9G`1Ah-CME1B4E;FCY=V%N}b|=5LUS)U9lL}cMA8@?5AqZnV_2u~0
zv8Y^(f!$9dbie}xi-Hv*8%&u-m&Tq`o52;N+p9vx07_y0Rl1N;1en-?DFMz@6&00X
zWWXr_I{H8uYm_6YJMkr4vYJ&PFdq=A03ZdUFZ%xol~vCQa9Iby31g*mWOOHDP+~v)
zvt}A3CJEvzK+1<jg9;2ZwN;9MO9JM5{vkkm>KBjK2VJ^7tg0D6?&81zIIDy}$v-F5
z=x3Bf#PVb^P$DRWU=kJ(28d1KpSF02LS!(m9NuhOtIGE_)eSq?HHwbJGJ$!By*X#l
zzzK2%#=Kym)RpuzF)blGin!SCG5T`^czQ@UGBdajcv_xGY}3nHGa$2gW+kujK()}m
z<zFqtS%qBoMjncaRLxE)NedO~)qX2nACTCfll_itfRk1;p=J_4m!QH*F(~n?g+5E{
z>2W{2v-5bc+~Msd9*IdX5`s`kew7xO?^jq8(}Fvu_}$Yy@}qX&jG=WOo@Ugs%#PR<
z*bfKbKj@VRRvCbmj!EAh#DOc^0}^%z#sMcJ_4*kW$C287+5=}7ppQXDMmA?7f$k8#
zpo|vbl7kGW7hQpY6P}tW^aYKS)liAU9=Kd4i~72{yu7?6+Tt}a4au}S!j2r2a-gai
zBw_9KSx;1KyWU|5JpXC)3uQ;1aOF@1T5A2<MGC7Wbb<?;TIA1vE)<yb`ymOA7&=~F
zdGa2C9&hoYaY2J{<H>^Iy`gVzd45I%S+tt*u2+OtdXr&4LZ!mC8;5M?Ql{rl*G$vb
zWVY2%O1<C?NJ~HogJ6U1v#_=9FhY*$KwdFPE+?v2&3};7Jvb5jGfDS#U87Xx6>DAF
z!h`*7?Ck6u9a%zqiLUG+{WrI^a&mH*tdCD;r*X={m6yCEw1u}XX6K3T<dD#FuiRRc
z`q8=>2+MdE7Gv9ENseT7&u#TEg0}69=FQio_{}C_odEs+79&*U0NeaJ{wMuO&uxs%
z0vAMLvGK!J#8k&UiHw+%&01a8g&S88tthP}^{TnfIlPiMy6f%vY#dAf%%Gxkzj-W3
znj<!5&jUJni$)bSR3$Grw=9ojQ70XA>krBM9ZIO+fS&RrcVX1R`xMvjUx-VwwEQ$$
zRsOE;Ql&@l5z>erPn|>Z#wo$c;ngJ@yp%NM>DAX`r=V@~nLAsl%(mBtoFI^A3~uH=
zFkLM~gyb1=Ii^TF5C-aPSobCaM%c=V7Aa|Or#~(@IN9B?>5SyRNA4FY_%$909xPq+
z)bVJP9bM3rS@@;%&1ayp>4D(X&Y+^uf2~g6WXHyp$=0?fY;zBScDV!QiLTV~_%KPT
zBW&jyCqO#-IGM1vA)&?j6bOc}fByX09gH|XIVlLe$?<kLT{tl@@%rqeU8T1b1kXQp
zxK0;%1j1)I7}V>t3XWH{&m^0F_#9{3ND`5I5Z6T-Oq_`0otIDK;MF>wVvdk!i?p|y
z<?*h<79C^KB1wvR+l%V7;|zAAfrGF*I@(cIwl$i{e!V|#8v$U8pSwT9!voPvdQ=n!
zZaSlzJ;67z3oqZOMSV5=L`0XE*!V5>-+MkBD+Q_J{IbhW<LB9z84>QA++HT(RZ+2M
z6x)K9L5s(HQe(CRQJ;VM>B~xsO2eL5fQt(*R&Uo76zn}-@1uVG`j7xDs>xyhY=4|p
zP*AO<$pH)+A~{rgXvlTGf;Te6Jw7Vx<#3vunHfb>6HZNS_;MRcT*8YSrgJg;;kGDb
z+tN{!T3(^M9WCAtup-q7TUQd!5nh;GXE8Bl0nP1K(|MkV!Kxm&e-*eDmF0~u-SEZy
z&qsnR)4_ovSOmU|q@)FBMyh!Cr<=vM*K{&jM^lC;N11gnq%SoUOy--LhY1OW=c^uk
zM!%kunN49~VIy!Jr`NnU4yW&*Z&UG5XUEo68VSb?s!>oJhKG(G`Mk&rrl#Z-HRH!`
z4cJO8#vc<=z-_<A&EUZNc(*RrMDH0k(TQkr(KSJ)JRDJu^imr@d7{9)9gIz*RJt=W
zbJb7}yS#>*na3}#<k#;DTa}Q&4_IM2AnK0{9!=wn?KFFrvazx{*=9bvIn>mipau&F
zi7S=X5D*wB*K~P*dp`dF0r~P=6#Mtz^vmP5?OGeB^^%{jFXe|1(*tI*7|=40-FA+f
zf{!WY>o%mgnL3VuPq8dVx~&~vX@y-*vb-pDL^c-6op{FU+CxQ0teEy+UL_jjP=`)P
zX~lM3>Zz)Vf~17(mY3{zT>#Hodw1s>SOax<GvoJ6JJ?xX3kr;1Jvm*KsFcE>;c?9n
z)0~9KKSKzh6A&8dfKpz@`R{c=?xghNKB6NBleyMbR%z+Drzay?+SZ?JqqTbEE*Z})
z4h@ph4-?ZWR>-slCwBXBU5*yC*7dzy9uf1dN52S2D|yXMaq_SH^4w+bT9l2H=NYCN
zNoI4~Vi-qO;o`rUJw?IHlroY-1!RB1oe(8N7A)-IkZNdU^|CjbT6u2!%8lPMK5n}=
zf*DH=rPJDgO&nDR?Wi{M@bD2EJLfrJd5Ozv>D$#8R<M?gtWar5Nmo@>K7D=th=|j&
zf)20eyBv}0*jW03ffpYvyJ&i*%^wF4P-Bd2YzVNhTik)!Sy_5O-&0c$^_H#Zcmuer
zy54!Ao-(}jC+5BR16PQ)u@UD_=DvXi8|3dBr&5u+pM9?8B`JR+jC`@!U~#8?RdP8X
zlJPPzrrj6!^=PGcyKLCFS&iF>$9s2+OA{fB)Afdtk(F)ses#yLw6L(iVln$uXI>GC
zCbOkxYOC!rk&Bl6qq!*{z`@IlH#*!ANyylk^y$exK3=Z9q_)v}B8wd5MrW_t0Wk<3
zd(Niq8+14ZW7o{g&QiS?c!PKeg7o9^-Ep}UTK|;5)t_|{zGMD^eb$qAiR)K>ZYYfy
zBzj~as0dxYIs^OQEm9$IKRRcl*#NOk@)|xh&^RS^#fA^uouzYuh54ef)5(|>`&`4`
zBY{{Xo^`AGwPzHqW<<4me%}2DM|doFW@cs?3Bnc#IfE@~OH0A~mk$CvAK{bbGI}B~
znS=B;s}1{k-L5qLph?VRE4}PeG?bB~fVfj-yXjtYI-DAUiV>2heWx8lv}oYpuA`c8
zLTQteG<8+i=OXPKAi}#4F9D5*;Dw2(+t?wXq!ngSQHk^ZR_lvv1GDFUL&e565)kmw
zFK~N1*V=kCU<n_r#n+cGcB=CkGXe(Y{F7z0G%i;!Hq!x9c&5z>n-+_Xh{(^8UQQmo
zH%&9YhFbc3&PAH|H<Dm3<8d$eaj;Ov;aBj;!$_@ie*ONi)lCWf2m0c;=mh7p(h>@4
zLC3YSX7@TIRnvCw^pm|k(W;AyY`jX{*t8l9M^ancU=(s$B+A8^8P}~IH#s>|eHR!o
zeH02K`(_zVPR^&Nr;o#_Nl9%!aujkh^9K`@-XW<9qy6HPb}_O1%Cq(fZ+8^Djcp8s
zp9z(vRMb6b=4=N|*zffj85sfO%yzxKyr{_af?;Z<bJoFbf}Zf6Gu*RJSv$#sGqY6P
z*r{1$_LS|&paSnY;wEZw^O4~4(%c9TQo2N}mZb&NSUUH-&7g2_gSi=nhexZUU?mt5
z)lYD+@Swwk10xfYFEp9amt!+^ue*2JA~yhb-4=rHvc@>2_+b4TyD>XRSk3-=T}@SC
z*5awUr{^H-YVc>F##m|<vx3AR4X#1CSLC0-g6IMR8Q#LGx`dwlu3h`ml9I2es4xj?
zDk^kzbmY|#S)spwPs(O!b+}&U?Ck8Yg@}mitXFSeR2%+e5XAC?`Uo<qsjfC-f9>zk
z>4<2kKBQ9u`{o_ugOmt5*yzse)1Xf`lgifn^5qLRUB9`7MLje6;TdS0mEvG1);1$1
zOUXJ+Mdd3AG4xJq@84L5oZSI$1nMxbtXxJ&2z+&Q6#<Xu&-P`zTf%6?a+Q8{<$0Us
z{N3U4^|C#JZ-(eZ)Zvi@E*_q2)em*`rc&yT^ulj~0fnx3h=?r-56|^I>xuBUFS-e>
zcH4cQ@cBi|=%yC0H6vC{;#Iw_<4VI-P4r2Fm*X;)z0F+oWBA$Nt_9t*FSl29+Idj%
zygNrm%r?6g`y*cMp6}|3#qLIvnHLup7+6_J885e@2v>J2@xVU7U@>`l1+R4AUp_fK
zU+WPQ2~AJ)hhvM~tm&g84N|t3MAV+#_fvk)H)V58HG6+&r3=!>O|sz{dcG+?u*JjI
z%f?Js+|ZY!aLXW$MZVAI-HvW%KmhKF?9&rgbhOxb`tpbo6~B-7`9wBILV__OLKHx6
z+nud5Q&M7zxS_$!f!kYBA_;{EIg)f~C?g|--K$<U((Hi#dOHvK{d+N6h*Nb1f3MF-
zWm5BX#m}IZ^WmT&sZgI+WkEqfsDe~?SCH(hGnlG~prV+Hte6VjmWpcYccbS4i4sc5
zLZj$rrZ2#3G`_<2PELAWZVeX7T1WhXhDFff_jz-L3#A)=Lli(K)PLKLqdfD6(&XUg
zZgB2J!*uIjbo1?a=Ww~dW5#pLr{nA8_e`(7_yh6Is5<Ojm9N;I4jUa8H}kVu9AWIz
zoRbzGe@2d(w^3R(CgY_#Z&gxjM4~+%=2{Q{E7<K5q;WWF++W>wZEYhLYhQ(gD}ftu
z<=S?0<AP3)_q+*pp3MCE{^Fpmw}GR$M5K_IzU9W@O+s^cG5fHd*y6!`R8KRp*(^RJ
z*}LnB7W?_E@b*TgJu!9-v~X9*$jvxwsPtc>F;1ys%BJq6C%I+}D)Tw)MGE?-nX@ci
zY-Y??XgfR9dSCC=G`jfIU=gXIr)`xUa}4|VyzCiXCc}5jI6Xj#V1=9L=}iFZs?<?W
zwP^HQ5m)*^XjD*s!@w83d~>?|7*t;^DQe>;<3-#l3gh=k00e`m$=|JJX5-Z=(%S8J
z9RrXOm%rsYd22>iuz0tnhKC=Peb<b@d+lEJ`c+<DUQi&aI@dp6xqY@CS<`r)n&etN
zt%Sh@N3Hge7t<`i*^wx-bD@%wl9H5!L|s!8rI?M6YiLe7_*}`KhIQRI@u+05w~ZRd
ztawqB)DdfTN&QxZYIrn3?gsT6U@e-OW)_4Saz_Es;U6x*LgRSmIyD8w-26NcsA#r(
zg0HND^TLud-~7oC=Jw~8j(o(}`Q}Lm$!%O4*vnRN?COF|J^y|9rOfE@{dRkHa33Mv
zGbZif;K}f)Ag(br33r$Yz6B$h@S6YjF3NRLTD<U~AEudmtPfJU2K#Kq+b^%Hs|(P~
zBPpz<t1&X+X(;=ZC%SAI=d~4VOM>%Xg1L`Zed1T02fn>yr;_p-<-CtV**UV1gG|7X
zgj}tU0>ewaJv7J&WegMd_Sntj_#VeAA&duYf-Cs-cbEL6`kkG(>RI--qToIRz1!US
z&9`y3TiFdjwXn1%8MYscch`>QGkz~U#730eH^FOQh9|0C7Z49XW@5yP)KE|;&m#$x
zk6c^{UQyyf?0;{G_T(nJzBd*gb%fT=-rLy$g5z{<m$C8j%F~$OhVm)-gj?s$brnnE
z__Q1<Y?S(qiycS=HNUkvv6B(y4xGjCnm2F(XdS%jKh_F^6%YHhH;2%KchAB)!&eI&
z7P}f1+<&O?p~BR4-8V+N@iAQQ?Bq%}sl(RroZ^q$c_BTck{D5k2Z|P}HmCmll=}J=
z^6)TRUfI#b1#zL}R!d+1xt6_yob@)G`pq>aO2N?`#NO$;c(U^(5sLo0GNzSxT<DC$
z=S}m&-O?f2Oj**$Av|v6i-Xs0p~RzgquD<=4YGt3oB+FGCYF1pc>l7>^DsrhzX=kn
zoKG91$gN|l2s#_RSw!EaldG)c4-eCk^Lh7jz*&gNjjmj8{Gsq~&iSjWk(b-OJY;2M
z0Zhf+{b{NG=i{T(+}yNWX1lPY@T#tf{%%;hOTDsD#K+G{ai<GsdUIqcgg>WOR<#iE
z7RF8K6t!oAdikHmG~KP&g}mT**^V$*<c=kTG7^ci(;xwCPyI@%4uU{hnKK9{!}Eu!
zx{0vE2=f0O1!J+eLyqMSgIlmZ1kH^~{$WUhl22si9{R*Z3VXi^Vjb=Vt3grv>H@Bu
zlpD0&_r`hGC9YP)`vY4WzBgrr%y5*Wwe{b&6IHJ@h6H5R2U&|qKIG-H>cP;^P%wlj
z{fFCq)=kc*CP1LE2J#60j5(D?q#`2H+ucpf!SS#L`UUc!r_pKCy=Z7@Cjaer?E~Vs
z40sHoyt9_OGaq$jPfaXT4b<hZG0AKE(QsbPkFe72ygATL>p&LC!Zsr(<Cf}bw~fRd
z+dyqsJt00oVGqC^a_?H&{pPJz%t{p4KVKs`LVIPUM`)=nea<bxetlt2eGZK@a2`N}
zEJ4df!)=OIoV+=A;0RuN-vB|eciZTE57O3ZWUr8IJAX;%*VI!EhY8NuQ9%ds&CuWO
z(CgQ<cdU+6c>CB>9B(BU78g~%C*X%O<Y#-}w~f+J$ssbeHX7Zu93*g~&-9gHH`7=j
z&Gz5hhy{#ME->leP9L=_{An)4>=5N?bKpO^<LV|u$dn&8@uCr7qN>p)bCci2s43ck
z2&F9({um&95%n9F^C~MZ?`ASr3=$I3Qwq>`9qy?28=c3cYWT1#TI{{ummAoFKs3eO
z)?*z9_r5I<)gENOGikif_RJXXHh!+nTj%1TI3g@Y+`Al%uDjeQAR|$+ie*I7OY=Ey
z-IL1BphT>1zd?K9_uwx%a<5kNqgZlIrGk^miHo`_jO~5&qDp>6U7xNn%+}Z#zNl5S
zgFrK`>B>AlcpWxH=HtvPR6aDSIsV{XcHLzOPV1FDvPh&H$|DMe64=sqtt%iI7vyeX
z&qVbK&|C@$D7}=dNzFNl$1hFdu2!!0gfLXnWCfuoZ~Mw06>2wzit}tM!H;8Pb;g*}
zShU1G6Jm9nXK7q+KPT6#l!aU&{$6zA$T@sEO*JG5YVX~isX@+}D%SOuFl6C^Cv>JQ
zOwZTLN=0^jl5=)mO`vVf;P;WutyhT$4DsQ#X@3|d;82z32sw6o%PA~RG+jPir#$sd
zW4Fa~abOIi2l{cQ90Na2EkQq3Qxqf~*CO0~_$BW#7v}PG-OTN*XYaZh?CR@h>MbJ!
z16QFxe_m<5pWmY$`GNC(bEsryR#dpZ^$Cv5X5oT3eeJ3a9L&zZ0^${M?P^^!4XK4+
zZ!?3q9WKZuP)@${XiE70?1>v!DXNz_HuWL_v!4MLRruju2sKbrkYmjtC#T7TU&fJx
zwgWHwz3i!IT#mYrKTVy1P7z$#RXDn?y5VDt-S_$lqXSs#dX(HfIt`G?7=dUgJ$oP>
zZG=2+>T^C}TlA=ixM+&(6=WNbWqEC*({g1npCU0d+;*1Y;^IQUWIFgKLZgsQ#}pYc
zP#IltQv*egM1#2~Zjn+3sax{Ojkz?)`(!GqYE6zU+p(#Umue#fpY84Gs;W*NC1k?<
zo}By}^ERwpp@4OzEWqmq8T2Y!6`UY+@M-3s&H6{e7qOY-r=H7{o)*f!pn95BtL)FW
zZ2qOc#GhkM**_LvV6)q-G+`pgCcJS>NGf(Zy&wA}CeCrk>#B1kah_6@7WZ0`%*ZIq
z&+Z@5kN+rzrHZaW7H$&}oG1re(@ehb+nXB_^@U-}hg#Eb?DHo=>G&eb6V>S*=Wn0s
z_ezTISl)cJbAUw688Q7MU#n+T{ap9_=}Fe1kj6mX$%5T?C8gSQTK@Bu*Oz88iUU}Q
zaKE*}yWSKxHT8G?0?JMX`J5pNDW&FKJ9>3VSvn0rvX2{I3#y_TKp8vikx~{(C0T7f
zFYhb<I|*I}m&&+gm2kf$dG>XDo}~?jJ+fj7)*500FBeBUbT3oF5+AqbbP86@Qeg!G
zd`;afJU#u&%V|5O8X6EPbqitVhz2bzi$l@zudg+?_V(~lP?QuEukKElLqbBJ5b;t{
zxn14q-gXm_5Q60-IdR1LZ4wSXr`hREMomJr7>Z)gduzzWUbo@iE~T?unrDR}m6TYU
z=E%1)F_rlFN1`lUUEeJJ6D%8@<$}A%|C{3DPxim0`1I^sL?X2@k?k-IPx&PDHg1dl
zPiQ-pUe)Ha<py{{(33rfTO|c0w8U(4{f4WNA>$Eiv{phzjK@dW`;yc1Zj>(jPK>^Y
zKZVE=hiR#(6xGPx!*dD?Q&Xj$q+`d9Zzv3`A#7(S+Y*cGi|5&skMEI1)|c6fGWTM8
zc%YaeOUo)w;}b{c`~B!yIlXzxAT+(7!nnMi5`1a>bV4fF4dPBvbv32P+2!c(f1X*|
zn+B|ES`tQ7@nvYv!w-CBh*fD!#~or;*=t(>?VcY{3;Ou%#Ph*(AwB&>#jbf5f~-90
zjfTj^=kMTpJbrZ45xJ>0j0|;yzO=6>OJZ9yZ7MyEK~#EguMd&fHwrpic6ubPq_~=>
zn7)kXA>lrl=k^JsUO&tunHFoUtQ}uUebi!PO<xHYhH{U%!P^3fdEO@1&v~9r=-K=G
zfa+`IqQ<K>b!<?^Fmc;lCS>r4OyvBLJMw|7Kt#XSb=L2}xQh#H030G!?7rTQrdFj1
z|IK-;l75HVRA?NT-88liW;y9IMBfoWMqa4Z;SCB`V`XK9$70c2Vwo2SC1d4}8Qm)5
zN-;I10vXhlV2u7O^RUMV+O+FT_bB&!J~}zEa&vpE#P6|Vu^+7mupmIwB_<Y(;~Foj
zqTVeu^iE=%I$8-&*diQO>6faNYn<-xitN;*)1|?~abMrBw`$HTMpyRSf2*o&l)rpk
zMsFIkxf85aV$-~7eb`QA$9|59+KW%GS?uC+)Yo`XgSlRsNMo=SBHBqhJ^KiHHL<o{
zTVhM51$pCITUwIog&G~MP;AxiV7+@Sw&O`!78Fkwtv-<x1!W7WYRJW0@!&cV>(o1N
zqY#o@i0T2@S^$F8qxmrsekPfEC|DGnt%Vg4eVD`gE91P2&YwF6+lRexW;WGak1Kb+
zZqs`&2|`y1LIM;YwT-sUY&;I9^3A6UkSRiSTO3IVe8mSVCVngQ@SzcRXQ-a(<{4+d
zpXT|VnalH!*50o@A{-BIvgPk5q?$F>w*~v@CkMtq9bH++nF|Luz9|0%6VA0X!5m$-
z_EMQIMo$yFjmlrZ5$j@Au^jRJbc~;#TJrPSe*gMq*0{4+D|U2z3|Jx>V8w2)T#KWl
z;d-0d!n{OlgM2b2ob|Geg*{RMQHHvr?lx{txc$S~uMz@44vi5!??cWc12gk9u<qoC
zp{1p|>Cq1YWNln@X3P1~I(xmWA3A!K1-w5@zzPZ%goVxM`jn?~I*0;cFm|t^q5%Xr
zxLn4{iZxo5i|g0u_}5Jb&U;~RImsU)g$k#m7=JYf&MlFS@#RvR(@u{OGzXidVK)h%
zBtHInxagsvZSdd{l$PZ7mJH_mM_N181JUDNfqS7YweHnbn^`Jd50B-Gi;INuEe?8;
zosEfXVg$U;RSh=eoEC|PJro8yUHq*ttPPp^ub4V3J;$=sXiGa5<+n7P>jJ_$|HQ=Q
zw8l6i7%-T<I5y;kyTk!)AY9or4#7MlC8h1(DEqv;+-%E0^6y({ny%7tq8!NGX}&~7
zYIOzjWV!Krv-?*h{`xuyyTP!ZOH8aZS1gj(({*Ot_9LZrbdE#rFSb5KQ}%;=BQ~0q
zOiis$(^@|pGv7c_fLDOfI5;eEu483_O=_HfAesUzfgi%XI+rI=!ODudKbCIUxRWis
z%)$}}0%>!t0y$7e-YH0er~5`pX`T+V$pt^ZI9km)1Y9no_cv`7l~0<DW0Ai(jrAu_
z(VkswzA7jrxVsiN5Gz?cMU*AE&?<R4a);D>0mi9ZM|a~q;0+t!WUF^d?E5{4WaLzk
zgmJnvcpHjNZ(%w-K5)zp1TW4sRrR#|0Oi$#$YDfBQqrI+ARB-~zKP%tkF`6!zeV$U
zBz(qS%M?BK%n{o>yrCfaB<vtY1@O@<mSlFEo&`13EKiGiv~KyPS}g1ttwq0evV+$)
zu))B<0AUoj>jhXQE&rkuHXF^|ZMw_p5}%hBydV}hm~X|rhpht~N5}$?ICZ(j!*yqH
zzRIdQ^T1!{Xz-CEf(OU(ys=qeT1YopMMG0{mO$KLyT29W=|;u#;+LE}*%eqDiTk{L
zcWNmXagxex3Sepa78a5QTfKzBCm+p$r0BPt`ylB~F=SNZ;zO7FrR3sAo7U;O*;KPb
z#+|$#^K*yce`kL->fF&2ggQ?~W^@W_K3Xa0>Jl?u9W*<nCQI+`kDudm)v;4j_DxP6
z+?{sMSDY#-DSfH(2J*o$f=ZY-X73S16}lbzhf~I1hkU#RsurCz8!I^NwhT0Le{t}+
z<mJ~#9~DIH^Ff5|g*W8M4>a+M<rl^q&`+zwoHMCshC0o}Tg%L(e>S?sNMEsV%!&sr
zQrn|p*~o2tCfTkB$3cVYRB|yM0Cb(^%@1b0BL)fq5xa{c3wjlhj6dLWKUXnfRun}(
zm6;ubMDR)dE;K}7%_nEQ^E8}M%eLl~k(X`C)n%H4)4gFa<ME8?VS4hJw(`k2)s;xA
zydFrY0=EIjX?e_i!=M5=b?8X7`R{Q^J(yKX9!(KL-riC)G+gKN@vh9uQrFd;o}9FL
zyn^W=f4Yl~<!PHa?Ky3nZd_mGD(nCiYnZn9={HVQb4tCU(`tN0^A?sq;Wx(z7?Sgq
zy0(ng^Rwx-T4kTPl&j-^UhQ9LsD3|1!amid?^T(%Gs=H3^-(=1rX`hqj?&dH>-FiD
z0+v@#U_|t73OIG$A93=$l=64DeeQsa<LT+CCLc8|t=GfF=4N+rWCk1Q+ldCQDAXCW
zBE<q2+$a!%^EV}PFwPQhJ9t4hJX!u()Z!(}2m4DI^~`NV9EquLi(<PJRuF?SPnO;*
zi*H51=Nn<yc^h6{_Sh|%b!f;uKj@RFnA1>@aDyha6$5!#M~*VFwp?2QJs_b%C+{2+
zlVjGctoD6#db(@1JO6D{e0p+1LrBQzus7l?RiObQtIXTexoJhX9f~e~+f$N|@iO$0
z3cu$x?nNyRa^O~)ZA{KT=^HZz$S=O<Ujn4Fv*a1xCsj4~Z`QwFZ9VYYE{z4Rm9@2b
zj-_#aqbgJ^b}$)lLl<UQ)@7!&O9`*G02B;YIM;&4@#e8Ax|yWJuVf4AzCslEJAl6<
z1u&3^Je2Bdf38(d^U-r*ES?cqySum4*Ryt}SmUH^Pup`XWeUEsWRmO=1+?KVFT#!V
zYH3G>{{ce^FW4$tM?;9}#eg&9Vy?yUy#qR{*b`jx4x$2tZEhHqFuYrkpZ{H59jR<?
zsgLf`%SMcwc|el-$yy%;(S^aVYUs#5v_zBo5a7k2a%6u1>+=7fzD*%68QRF4lV$ds
z#zI1Suz|r1^c%geUSgRxRvNgTgETj1NhvD>d<o7@E-C)~Qi>txI&v!5m=r1|7)+6`
zuzUKI%Y7sS%-Z>XNm8&6`3NnfZ9E1)mvd%d5;HnnuQ;goWS0ChDJk8o)MY;hzzdn6
z+nJPP!sEPgonB9MM-G)flGw2_Fo;V_190C;Xn%tVSNcI;mUjw3r9u+QKxjcO-%rig
zO9B`vhP7F#C35f}wY=@ql2>D=>VM-nRm=!~c^=w@Zc%eWx^7PheExMD5n`gEU&@a3
zdsfE+rRTkuMx|tEZsIvt3T{?ZD=qb{&>VID8`N?gpP!TwnOQ&5F<#4DX`QlDMEhI6
z2F;;j#|~Mjh)=(^wf?$l(lD4C<NI@vQ57=x?fq{^46<_uE8oEO5*F=A`=j&ASq>rS
zO;+28Q9Lv5KbAf6+^yBz8418D>KQ+n#AtZ4OD!!4B6_Y97&#XMviyUe4iV|YEuNZ^
z;(oL<2yiEysUIKiZp(wmO(V|Me<NU6QnFlL$bE{(2x`MwO8L0+PbOI0sDd6$TV}=T
zjGixKk&qvwNFz+caYIzH*C*lIW>VYrR_r7vYNl+nNhUN;^ccVkz_EDuzsFno-0EhG
z>1?;U*2Hz${-){8(iInU=5@hVTg1aVaIzLNd7NX3ToV4B-TwR1Oljo#Y6Nt;Qk8Mn
zQQ<7~W*Zsan19D_9e^GO&A~i7Zgb)N7;$?`KpXBieua5+J<ebMs-a`pU6wpOKaMqa
zKs^5f9l`LqyM3@?9uf#$6b6AyOfLrdrS`o%9ZVTDKW>L)Z9fjV49%Ag0N%VwZ_O`i
zqEo)Ve=NVKmal%}q<LSbb!TXa2?}9BMVYt5H<Y9trLz78{Xd`(U;B%;9ASliRj}KP
zMy7mHlO>&<B~5QjGvzREiog(J37QtD-|D|-UKB5Q-@FO(gELg#s9W^tG^NL)f>rdX
zz?kC^PmNLtK&q){sbyDvi%a2fxhq>*(rB-wq8zxp#33P;)#h<K=s*-Tp0O}DmynP^
zwsQY#B?5Q?qJ9Rom67m18cJA_rY9X$p94H9N4bCD0+QtvAstO7zebh;T%aqo1DCOv
zf(5n<*oT1meBD(Z%c;7>x9h~$jFq1A#D4$*<+uc0-ONnYFdvR;y_F<J!@ifQv%tS-
zfS;w5FF-b#=<u&Qes_01k{ybP$4)rwa3}hKRrwMd-@KR)w@*by^+v0umE>XdE?Vk6
zXR;)j{fzdK3Sw*3u$p`7wwx;?T&qYg))?C_+4IT+rQMwMnlz}^xDi*4wey1o6#ibl
zFs4wI?ll=YyEv{@Inv>MSa{kysAmM;xY%k~^E+)-ZU~2#7kR?{6XVxEmYFrbC_ezC
zew&rBUlLM2IA)}RPpM$MNB}X^0l4Lh8?Rf|f_OGf#{C(Da0vTqY{?i$l^XRS*#!k9
zIaX#_l?3#a`KC?R7ZPGntyI$b-+#=!tmBQnpJvY=!M*m;=~f%$&lg?xoL5d1eI<9e
zr8lfaa7WyBXp(s;WL`)v$u~}%t=3$P!2+8s@~SpjL!Eo~u_ncUlvKtvHh+soBJBCk
zvze`qpK{Fqc{U^YufX1uE!hil{nNtWw;ScJc@Y&Ds@nYuQPsH2Cer_~ny`YABDQJJ
zN?Teq0HJ^X-n<JZ#)Pr2rKPN_Two0Xxu^KoN5giaLETTgBPH@WXGOc?n`^T+C%FVd
zJffnY=cXJ}_jkPWOh_vE-a~;sT5LsT%3-vx$Y%H<{kNzb`ttp!SiJsCwx>5{`mi%o
znS_i1D<Q}>^xw~K;7q@Zi&qy$s&M1goK=3GN7EcuK#EivM~Tgw?ftmW{A&H~vXwkJ
z^rL!GN^sx=t>~+?pePPu9>B#i8A*z0_e=tf`8Qc@5jyw=_MJJ<FNVu(87Lz{tv*-Q
zJO)VzMAbF}^ije9=?T!pt!P-x>anGVkdBzrPerAvEN0@9H7Yo}p2x3Pj<+b4N!O+q
z#Gq!Z=wN!2iy<oEu^!yC8mCoXK*7~`&;-k6>XnSR)p-oFhSC=NE402-&8Z;a)-^a<
zZ&YPGgr0<Yb>EMCYSv7Tn;R@T#?nMYd@vj*(GIu<2uilW$m_ze-3&O9#LaFa?GUNz
zi-r^x*Xyd2_qt|)Iom&N_3;?z*y47iTY1+zVyNwWts5%A_Fizk-mI1IjI<Wjd0g5s
z&0ZG0F5lti%OfIKnAa?7Uq(y;BGsJrUi|g8#f8RlWKFO)s37!qsHKvc5?IAb?wb2w
zrQvq~3n(D!FCmC8Q(@giG$n-fHmytXjoJ0X!Q(|iADKe1&4i)Az>92MI#^}CU;^MK
zyNw!313aB_)NUXF9H1=&(Dn$Slh?!b@3`c?IB2FmFU1sugvFHgw3aCbOdw%f2gp0&
z2HQ%Y7r2jbLnm*v`m6PSvTltV9BykO+h|OJ9q@yK_KkW>*Q1cIEFA)!h^sCDYfN##
z5|{c#7(lfBcuuxAVAKOLy>w-8{?tyNROa<3pM`EM?ow{zi|Hp0bZ~I-=M-f%!$Mh+
z@f4?pCHIKB8$HY8#afeB_oJEDBZ~!rl2BKUgSE)xfdwo6(Y@K^QH^D|vdFZ&lpIG%
z)neaOz!}Pjh>MZ{`SAA-S#-^ox;)?GJH9+z0-3Gte&MHkeD~WUW@_q==X>4gNm#Ss
zV34e`0bcO(j5nY~g``1zvrcJkTPb_*o`11Lh4IF`P`5%`(~w&Alb)WH8m2c^hzZP-
zxY0fg6`uh_heTWMR4thyXz2az{6ClO7<=}2%$$aGGQA2Nv^WHzjA(t1smG9QT#qLW
zyAD1-nDg|>G}kXX&>@xxTnF-OmYuE7R*XA4I~#3QQ`l`YKp;?BS{exS+U9m`v)Y1+
z<7sualD^!KPLN%jB?Iu?W$I9M+elNkBK&}~2utccxG*$)!<_p=%DYGjiV`ndy)a?y
z(~iVg9Hj>f3lA+#-^1glqvOb#`Ai1GzO%eMMBHL1-?@pTP+>Y8{wBMYc&qEFlD-_%
z4w=R??sId^>(*mk+S&O#pHA|iN%+DtZ6<+?w#)DTcC@}!b^YfFFey0-GMOc-*LLSN
zx*QE7c}pb31EOFDJ~DD<1U^45Humx1;b5b0VWFwj$?$f+?YD0|&<kj0iNXG7{!KVr
zyV}-u8E^DP?%WM6=t=*TmW}4J{NIEb<yF9tx~U_1TU?J;R`<JNLo^;OW_#YYuzwK7
z=MK$G*6aP?53){a+%~4}jNdpiv3c_^tqkO~{`L1gvYs`?yak=WME*zE<k{JQb^eWo
zYQxWvx#0$%A-5N*k$m3MS6|yr@UTg16{v)Go#7U<8@R^Tci4aYG6{5IFwEPHQ<=3-
zc^{=J9i}a@`5q>ltASnGPQ=jot2?{3`7tqG1rVUg+sh*tm+oC>AH&GR#E;vfS(yL8
z(jnfM+z%1r7_x7-k_VfV<#dT#cIGV_k78t>GthNZF{=lxN7|J+QPWnh=y7wWex;*c
z4v}N;pL!9VL{aLu^W1D+7IQ@vm3tsj7zG81YO1T(IeR^_%E`&W!CeFS7wD<9DrNTU
zV|WTl7zuS^ZA}szUy|o+BBdJzcsuunvrn*0TM3Xu<7-H`si^3=OQ}pL{q~e!N4j+y
z>m>ww`b;Rfowep>W_Wzw+Wh>$*S`Qyy<E=zgG6`Af=}Ru<kiHurp^a_rKC-L^M_n0
zQr%BXFiNDnJLdo48y?{}JkR0<oy<9*rS@G?sCVhUu}KmDVrR*%oEx3-*xXNkKoMK6
zM}2lyl9=htM9UbG<TyFzbzdwWmx0did*SW~bEtE<#oOd8%KW*Ah<=q?h+6gkc5dr$
zJragnKz-A5i(XA)J`UHHh4vh%c0ROr@%_pjpGno$Gm8o=_y#Ozk{G*jzcNaX+#2bi
z2)T}?8rDKBAzWZF&WzcdL=@)p-m?23637pPg}Co-ZwpsdMUFQW%AtHRm!*g~nk4?t
z@&}7|SeUsx>KT*e5EjcG_HkY4eJ9slSh+yRHLXDNqG~?eHD09cWdm^hzjV;hn-k;v
zF)Q)5fG`CxF;zM{`y=sqm;S@S{0HV{fJ1jg*4e#2r5>LT=kjFx)8#psyxyN%CG(t9
zs)697w-oE2J40O9`X4TUVvW?$v^O)T(XppSES^K?e@F{Pl=HZDenDgnd7Qd-vBks6
z7Zo{)<0bA3Dmra>KR7%6RXB+*x|t6@itS`@H5Mu-y07fXwVUEabn(so*umz=0AfGG
z<OS>UV(hhO`S^1xN0Q%bnu=0K3oez(&Ge|6-1%yYw^7L#Y)r_L%rX&?dd6c>I;m9m
z+e^?ga<&6(gS7f<+_*Y2L7)I3GauhzUtf@(J~sCA!vl+83h{a?*Ee<%M@E)M*p`G4
zb=|!PBAo0tDn~LHndT6t;-ZS#$qXB_U*UJ9%Bt27lc|9aWi0R7qvqL_BmcmgjWj0z
z<ijs<8(!xxxAmD#$E-Z!+LM7dijm_*(}lLkCnR9tIYnX5+^4<gO7IqnwY7>=w`6A?
zH+xqc@d@MtVPu3sgaJLbeYWa)@}9gT?P!FQh2hI9JIF#TBDkcID*7s+QoT<-xneP_
z1@t&s;=tbyDE1DWcQ2hn5`zBcskjoO9qv72LEKc?)Z9Bu^ZzI7cXRp`H~H;n>1jSv
z<;mNVFZ3a%pyf#(Qi&9o8|V0ssUNtE0cM{Ie_HfLo||^H8w(LlfVqi^2L1jG)7b^E
z?J+TNlah%Xaq!%UgJ&!h^o&Z4Lyjl9mJeRkjq?1ew&+f@_6H>=0kkLQ%XJX<F$!fR
z33YZQ;rfg1XJH)#cYtr3Y)0Kc^EYVrQY$K_TGmjh)FlTz))YPdV1!#}IBwx8hR2i?
zot0*X^G*m=#_Y)Wzr5FNX1$55t|<*x5z4K@bHT=+3dPWbEt_p{iLRns08!fVSt+lS
z8MBCqx&y7B5;NPw<uyHma$qZYP^QT+w8qui#DGw-Q6ByM$~Uh9#3-~<Q^_rmt$?&P
zmsDIfToOnotfpR^j=}0OdBo}f$=Loo8ADt2OZJjls}G;%xtDW411Uh``%mPcmG?2|
zB8bS`&>H3=^QVAsSyARzR&f~_Hnz5Mc|Y9UL#nGI2u7n%6nL+y1eAoADf_x7Q^XzJ
zc)pYh6q$}(-~5;Wp{I&47!d_PemBqV=y+KkR<IqkjSlV8RMJS3OG`ew*D;a7nOce@
z0kT}+mkLmtn!z%ub0{>@6$MXE`3`$4zrt>?8M|cVA_t^uVBrd7GI&O&l}5T;PV@|I
z#e^!sR}n?#CL$(a;0V)}7XldUU^KKxB-z&&;%##hQzxixJvp#vkEl~zn+Wdh7(UIv
zR>Bo%@n7zr2cseIy<QBpN}*_;9P~>dz8wlAtY&b>vUH?~SpJPhXy6C*NF2cfr8PCL
z|FSg+nF$H!=H{fM_QZpzb5qY1D1S<{amG^+B68jd7C{^yd^4+x-&=#LEIHz^R}64w
zv9F(dx?hLHx~hq}*@+~jby;^bU~*0m-DSHpQ|9wY)}m_!7;6gA<GYl)B_%_>y`G}J
zt$!|GpABPUri6s^Nw^q5meXx+I#P*TSb6CW!Ka5=%U9%|9>%KRVy>d^z6xJ%nF8s?
zSM9o`2g|O!+wxfjGij6V$j`;HjRUmL%0&6+3~uS&h|Iz6SPz*JW$6c(>qIi2)F9Y;
z4?p(MF=P|>jJA4J&rJSN{<JDPVoD4@D9aB2|FHFyVRba!mInzET!MSh;O_2j0fGm2
zceen+-QC^YNpN>}cXywD-*4`nXXeg%&JTV-(_LM)Yp=ET+69Aps5Z3uCvj_Q3lc6k
zI2Z}XkXFLM@Gpvzs(_t{Nn-8*CydiOeNLF9<ZsS<BOY@y6H94#c+Ou68skZVr1TUW
z9yKYnVMEek|0JGI-~TqD{_lEQN#aG*f9q|P{R1VxYbm-sUY`k5t2+M2!*~0&Of}=(
zAgvtzO-~%7%by@b?fNh(Oq7j(f-x!azfetoOF@ZbIsH)srM{A*&QVeB{t{2@+x{4v
z5x#EijyE5!I1DjCQD;qFUS0z|TScqNs4xt1Du=e#-1oaQ$_O={b@a=qU%HbT9%jFg
zb>6?xv}jD5KmTa(k9P7QQ@!e}AMI0|L-WyQOo$m^_UNDd5k;&pn9iWLxzaR*f`W3E
z82-EY{@&`jH>3yBEK;UHb}%2wUEa0dc=dn)oe4#V;2wX3Jq+1BF@56|i5|Km_=?k$
zUDVE+*&#Cm0hmF4KOL^FrnGsEeRn8Nm!LPSnxs=nwP|R%6$4uQ1l84-Sl9!iS8Uv>
zE9-I1Mmt(;#z;Q)D5e#c-gI9vp)5J$rbBpj`tAls38=iyTa#9PVy^g|NWb-bz%t|+
zc=0q<;znctBrmh(=8iW35LAuFNP|}C((IDlUN97nX4LX)EtKT#3OY)o5Se3aW_*4u
z+f1Ly-d%b4WOiAWYoDSI6u-VoSC`|XK2#MA*Y0T9{Jec=$<eMg9@nHp{sZy_?{?Wl
zxe-Rt`I0wZfZFTf<oyo)2=4kv3(YPw>c{Osd5wXX8$zw2Tu44(efbxp+3j&>#2Y@F
zJ~-X0doX8(9ZErQZIe9+ZwQu3De3j9#ow%@@RDrrwY#<H8|XGN%E$H%JK7)noc_oz
zDsl+p;;v<G`?)gzQS^a_H+8D;f!s^h=pRBHVl69txFX85rf=fr4Orj4s%p2@TGhwW
z)L2k?SUxYVm_P)xe=M|jpfCreA=S^m7Je5Jo64cSDq8Sh|2SaR4Y@fWe1AF^E0WJ@
zT+vWg&MPPYTp1%lNEaVt%0gWEPkpRFJ>Gl`F8lkTWq!f$0k0QXa*01{yxiO#OEFmd
z6P@SI`Ri$zrzf=EJwNV0RVw@b3lRU6GCh5XcfBkS*l!+)I}s={%bCpV!kbfq&tw9~
zNjukiNgXesUZp2j`Z6;k9S$O$?d?s^G?+=YQ{$9BS{w!%!;^AF1{8G2n26kW5D*l^
ziz(hT03zVNY1Y(=pS}}tNrnw~tuEaG5q84Hudu++o~^8uQ&T}Yj=W`l_Io<FOXXT4
zBK~q=P7dIH7i#y&=<m4-Lz*@|Z{_Y*+Udoi7nXhJ#NaX<rwohsS$7;El<t6jSR-`s
zI{0MC{}7a3f2v);cvFQEl<J8U`54x&I2(lI16>(`{qO?5^U*5}z^MO{Ou-=<GlYaN
z3L|h$xVh6_UMASsO`ada>uPG)85t$z<s~J4BOrCf#KwBRK8hhBQT)LrQCDGQQWiS`
zl3Q(7WD@=r!ln>41*N#L((!jKquh!^)7;ztHX1McFbcstIZ;Frw(|yGXM}E>Rk;A3
zKM^$p{|s!y0_=qJ4K45F4r^Oidx&6a!H)^{Ut$G-d9lBf`qp-{PE|!F-1&OfoxLeg
z_1{kgD>}BD*Fa=Xm2Q@WO)Nhw-<R*!{Aj|Md6_>>=K8ObBXLYio)#k50py+LasUNT
zq@E5?o$9k4hj@%$A1j1!f$QP%Egv-@Q1%;I=z6uP-hR-pzt*20!qR&KfjaD=AxvRm
zLNU=bgwI^@F);>~<pvfOY<9UPePNnLMs0iA0()F6{pf418{U$z-|T0_DRsM{MoMvO
z)m4_V%Ix{n4m>M|Y0C8z=1RXw92XBAm5n^iMQgEm-ToglbqPZmgGo<FGjo!4Cigvc
z#QmB0k!5A)(`?vwsQZ^sLN|v{Kg`x$?{hc}zYR$xpHTD54%=YkXpge$Yrv+!5L;Np
zMMn#`xHR3Ju+wO_h6_DS=+dNVH#wc8qEXNR?i6`l-2jOyu85UhZm);6y}|lGzIg*^
z>==TKgNR!A&J;kFUBfjVm%FpS#cmY<T3z51<>*-x9rJl6NsQc4Q&Y1#>>s8ui^r+t
zOfs6I1p0|#<*i_!#q%r@PBp`*pjM^zSPdJi{PU<!S_yA0Txlu^H!T2XgU_no@nLYm
z1CiRg%<;ZT3JJq~n)0i|1#1|2>38$Px<;AGv3G{kcg>SVr>%8~6G$*jIfZ*<^e@QB
zkl6mdzoOd-t5t~wB*0;JU-^dc5BOTx_kFS)GIL4<6qAsg=ZZ`RPl|^nUo|0~Hw&{+
zr*(EkUWH<h{~r4*qMj?<25YxD)2i3z&(XnxkI$$+sOtaWAgs)_d{#HN;zH5g7$lWM
z1$A$P*x#6WYnnN-EpVW$DgHRU+|qG&CZ8#^dOWG}T3?3%7F|O_!}X0W1sPd?|2iP@
zNdp#xO1b@a=4%0G!M%Po`Pl?Rxhg~@<>N$lHg=1%fQ4y-d`zG@mZP^L{ejl)Oh6gr
zX{uHS^=6+UWG*!F!S$mh5oV7R`(TCd>OF-JW*XQU7ExP(pA_|p2HpS$MeU~7S#T7>
z^0!)a1h>%E^p2Wk82PWfPMn{xhY+Eb>~{Z7H7K{{CxMVHhxuZR?_Tb|SdpQDUJv7i
z)n;1aWnxMHEDP+}?Q;`C^MbH`7PY;3>Ra&&pTF$N@jAKT1??qx=9_)1XGf1HbUGN<
z4GCF-g-wYf5bg3%lgl&X=1if};rTl~oh?mRUB<%3O4ym`eRh@Bmxtflun7wKGy1da
z>jLvyg;VdJsdU~|?P`eerRk?sc5q_z-*txI$T*-6jL5+5Eklk3u=0yMnD~(D6$lw8
zFdYa_s-+(vDJqwliojRlFD0cxGSbFw&8B~OJ*23%j{HC0@i`@F;0I~-!ZOx59gKFD
zhU4q+p{4LES_n)=cYs@IcW+PgtRT(29L_^aNwKJerf1BQps_V-I!u#PSxoBhlGcr!
z{T6c5JjsE1&X+8eX+NOP3u3T?eXXs1?cs3IR#N<Gtt<ta%w7C=1#YiZF$1CN@*zQM
z>N{m!vy~|$xZh#i?<U}HR1uUA7$)!vH3U_u41Ii~?tW&&u$gC<J?9i28|}}u$38b*
zVB2gke_m(iW39fO4RJpf3B{S}5p#Yhtu_gymr(MDbv&rBH;YEOsSJT0aP{Irys2G?
za8qR<>rPM0;VExG(vmZ%S)@&b<WQtv_gYeA${1s<<Z~Y@xLIbN^|4-lkT=&dpQ^RH
zwF$#^qW<7Y#g<?~fngLp1zvYH3QP+T1q2<_ralZ2SU0E`+O^^QaHMsDhB`GqCNjS%
zAU%^yG2k_WYu;xNEneq39J3-N-&jj?Bfop-<i;ug^t1fF64P9~MCve;S=Go{_$(+p
zD`CDL?aB(`_hBIpZsY)S@|Zt^E{JF7+tZz+e_TrclT*NMa~TRHCDh^(L1*F>=}AVJ
z=z|z@2v_WN0jq3*QG)w7B||gD<RY=()DlcxO3;(7I)jk9*HuN*CWxwv(SICx#L=UP
zYfq%=E{eFL7QUOcEsXy->lp}Yy0EyN)s$bc@10RS@HSTrGem`TMbgYzk^E%P(d9t7
zCm|wi)mN|f*XB$~d0^3vk&3g)^3D*GqHa7b5*80HNEWG<$CFlAh73(Ue8l4YWOeKX
ziI5QsaHK~3{VP-(n+v_h<d`29kyBWGCQJkgA3GX(XhO9kowQ<Gz8m#Kg1@b(8SP5F
z<kHR`Ok&Fce`A;5ONudy*Q*l30W52kV1jp+;m*AY5XG^s+O&owIfx_8f|Q3I#rpiu
z+V0<tao(E(<?$+9Qd+8do7$$l^{o?Z*a_+Mu|8?y<I7NL#ElNmyXWJ%GM|<A;LJ<{
zQ1#L&yUn_Yh{z-DB2blZ@Imf@LN*x%pL^lkIhJwRqcFGB)y3;kX76wFurGQk>}58;
zQY(KJ`1CkMl_uBkK$TMHzDrI{-Bn*$pvt}W-C(VTBG9Oz;uzGNY+=p)BJ-mK`U@+<
z$~fN$eaRBr!rF4StV@38r(nGqXmx^?Ok&m8A)?q0V@qZxrt!f+h3~Ib*{DH!K;iL7
zB5fLn-S#Y1+(0?s`@=jouR<NLF0fn4QylhBkI1>*q1I1L0&d^g#}$(*haAaR>nQZ<
zQhCc+ZOC_J;c6}07GevhCkH)Mw5$fTD>z(m&zfgR%zZgQo3_bSU&)@$-bXht(1Bl4
z(om!zu9&=OfnLaJU-F<bN)FyCOnD1|j7esA$WxBgns7~2mSX?%cefh3dy8FHVs?I>
z(_vr6BiPi;Omr>W`QoCLpXYpF3c`Jo12Px)$ltIqKcC0z`FnMeyhJknA{e(X;R|co
zk1J8Ro-y3tj~C#|9@FoUS5A{l*fB$jOpNfJcvMZOaLw^DNVpk5VCxivf10AgmG5ne
zmBB-FU{H~t*|%*u&&$s}UzmYaX^8N^L|~cUYvr?jn5oTm+jS30mN2%C82$5gKv`MS
zh0633K@`8;u?)Y-vxJQYf>CASGztt#5nQa91@NI}LZEL$%)0;ZrmHC~)H|^R<5uzi
zD<dEB$8y|F)2wtr4}57M=DMo~=>V~2%b3!`prPG`klKR3DW6Sa`a9O>eo~-H#rNSp
zt(#axdfIwd-#%mU)e&A#Ua$75B~pF|9{DAXwa_lKyTQ9~AkZZamUNEAj35=sWh@Kc
z<W^x}V<W&Ka@f5MT~odD{=@>GTDx2s67l>2Ozq~^nF0^`Kr0v%26*rcGj_CV&U4nh
zXfUES-!k(H5%BdPnVqSwGiFBk*Q41qdCjO(^BgkS^APw!VD0X|65!s@d|WDZm4IQ8
z6eodv2fSc2zBCf^3M#obCM97eK_lJX$q%1voTx$y84AV@(V{}GCN5bAR~fa_l%qD%
zro-Ciit&$`)Nr^6UFW&v9Mu<9oWIq73{SprD;i|(HJO)|m(bk2Ud&~{lTzs}Q)Xx^
zDlBzN`xF{uy!4J8F#CjE+2!MD9ZOFk9}RT~U@o_X%9oHtd(3ofh73_rIXgRR5m?K+
zNRGR$RA3Mz^Fqar(q)&CV{-ND&20tWMabx_`~?dQYUfJ1r74QvSx?-Pq0{8KubY+f
zR9s%XE%@QrndqO+!uWOvZ7ur;ZD)OZz&dR18m<9eoyYD@Z#mnj9wn{FvLp7s^w$U@
zpiUK8x1`+W-u9MuGeXF6uyLSKU2<4zWE(Bs@tvD8rN{yWoKjM&m1Jcl(IaK%ZF8iM
zxdQ^RORTRM#m-{pU72#ojl{sH6pHpA*sU5BS5VSQ4!ZUvAd%7wv&?<<DeHV6pFM;e
z|Jsy`Sk=5~`?ojnq@&$yduNB=^D#OlMO|Cl>*O9EuC-{x|Lx?jKDI1=hXX_M7<6U+
zTl3AmDgDkG9JIYUks3PIt`xPaZroMSiA$Ic446S$Dwc_n)9l#3rcxGfBdegcD2f*(
zX^Qs#0V5*d^IRb4z-T1KyUbH2D8%$Jw#6XT<sxH?S)pQb!Odun&#m%#o+)XH+h9sl
z$It5fC&YZ{sB6c?#4wtSrzhSzKay+Hld;}OTF)iV4RB)geF9;#_JrW(>2$rgI%`jE
zzC)2*XnI3~MWtbwy5(1Kr#kNk0sb#7g7nwj&W;kDrsV*ETPz8tws7>A%M;5cUKA=f
zGLoc1$;7zYS+x*p+U@3ri5Pq88sEOP-SXgw@x0OPDrk#uBMP&7J6SFq70<G*>rF>z
z8`H6s5ASHFNKqCIWqz>S1ZW6em}7`F?_Qa~wLEGWHKGxl-p@AcUY?n{#FoOXF(9rp
zz;(Wgx3t>J;ZD9f5m<&g0Q2iGfHV%%H1#u__+kamZ8fit+;7O%dR+FhBD$(vqKxa>
z@H9Ke??!tPy*572UPtkZCitI_8}d&8N24x&)M{*H)0VBFndsO;#6;-T&R<DF+?sM}
zYK_*YJ0((>W-31nFK$QY`l6$vW27Q^m-S2gh5iDKZ}wo7yvq=XeusR~hS!6@n2B+P
z`e;d@HWF7{)lW%!XzQrIr0nPr4$$?Ex&Jufc_g=<AE?Pw*WIMjMgs9MR?FJO;_i?p
zwXj-Wnj5ltZ%^~YkmYxITibm}qT=m9fdzkGES-Y$PS48BT>qoz^3K`m_7KxkMv>$1
z_L~nPR%TtDbBdF!iKbQdZ?0`jB5AXx#fG^!=G$R>+fd-x8EJ0)-x;|B?`0o{c0*1o
zxYXe$lI70?(|6~Xn*&<$iQP35QRat5)Zf;6AAd=7%iptKtWc*m_Al#|4S0CCov&U&
zht0VMw$*g*`@oUQC*RF%bWP-$*_l=YFJGxwhe;j5_K*enpu6z_G-k_;fp-vKM#{j&
zpFk?P$TVU#5Sag$X(}`a30OXT2M((RB{X>eeXBzvMLk+FoY0UJHuB-{@|r<uWr5?F
zb8_U)X?em=fWq#^PUm70P3kJSimo2iifxSfJ=F4Zl>pYc(ZlWIG_^@oGLF)LY?iI-
zM&w}E#QW1r5VipnHuhv~9h34z1@`7o8cSvr7(tTz*pRrOq0H<hCc9t3AbOZ%p(`Z=
z?E>|H3RQgf#oklW_<9OEwi(KjatrxEFMKLSd^$;QU~49r=R#t7cc6w|JR<M%)YY8W
z3iJ<l>^Mg1yOaGP1zRck%R2KXk<=!|OB#O5R1&@{DBG-L^~qxD=nid6<wsO5{n@jX
zvc{_wg>b>(8r=K4S0$HvzO+0MG7r&x>=64`Xe@xvJO3;Xtl$s79zAbIDG%6GzR39I
zSs^;s^))eWx`NA-EG8w?@%>bnShy8}A>md6ge(_(S7BppZ>3+3<|Puhq$or^JmE|_
zK7+KSz=hV}5Q90@6)?lExH`SfN8d%E)nwC|S9_)fXmG8+_JT-Lz<=HuCW)u?Jt%R&
zgjB;g7K-wZ9VE-;mQq?}7hq-eseB<YaTq)GAGTQ=POSCBL)s_SmRFAK__}#x8C<B+
zMB22PRBm*e!GR(P`zpd;m*=%MB|_Bu8Mv~5s)qiJOq1>e4{b>?S|hD4LtZqjtClgr
z0TYY2+-!2Mr*(f!d^dM$DgUtNgOfBrbp@sk^+*}+GQp@r>$-rK@#7&BERV<SC$m~V
zYWA<UpT%5p3fjN<Xs-_7%R&ly3d90Uq_9A7RhMXRsn>=+Hy`-ki)LQw1Yf&TmUaC1
z1w7UUurE3jqtCE_Jke&yI0UjixMKW_6tOJlHUDG?c}e<(x$(tkz8bsz=R$^!L8Lpq
zlE`0I!Wjd_RXNqK$S^3-KsG6e6TfJC@P(x|rxQae%@x<UJt`%)d^am;2GRn50)Pkx
zXg8cv9O`nz;iaz(9czv;Hdod2l7ru#oYO%Rh#lPm$VcuN=H;g&7U`obh41b+k7G1b
zZXcT|F%aQdMoIfg^N!w>;C{$%(Ev&tybPn@;r{e_MDIK4cd+)*V6RV;qD^Q$Ki60m
z7EL}yb*1#9lV|`C)amJdp2n-C(o7&RX<9i+-u>$R1f#V?%*=>);J4QM%-v%7zU2Ep
z#Rq0`q9&i2^7rSH^)=}*Qc0`>3!ETT7L?z7V1kQKzri`McnHk<KVA+XGoh8DV^%rp
zKCU4i<N@TYg6<tMD}bqCD3qDj;Jl)(bKlxVPqyG=O(H4j$Q4<XG2U~l4@{3lemjoQ
zg_vm_<Vjn4Zf>+SGu+762Aie_kc)>0v=wc{>q3iUh`hnY0C%dQ(nQ4(%bS3*y77TO
zSX1nJCCB4rkC?|hgC-|xjWV}r+!RE3@cYH^$^)zWG361E-mVeRD|avAlt$v4);<sR
z+-vYYkn$nMKbFTn128akPf`RIV(vw2VywTwOYL|qYF*L3D0}+O6YrL;obZ?4V)ESr
zH=PajaCAOtzk3=%ga~yij7Sx<TZ+GlRJQ1L*aML2jAHB4YDw=J{l4!v2Ub_W%Td$X
ziFi>jA2p0^t}|C;XCgR3FwjU0_sH`|USCpKk0DKRIR0MFOme}Y_R#Y(GUh8vQdALi
z%cM7NhV9!>+TP9)@RmFymu|xEnXo&br4*ki@aNV|c)<N0KVLI_LPG3e8$a~fo3fxg
znRcS~8?P^R))4ksKCV|AE}|u1<FYgRAj{OpdYmaS?D6)msbwy|RgtSetYufts@>Zk
zIy}&tAN+!ESU@I*z`%)<LS-q83<If4G<1qFc&&t6*jCAZpI;l~ARa@|v!=mal))ku
z8xOfA6EN}_{+x}g;&UyKz3s8cKpK!?ZrUxzgKp5uz)>C6<(JV@u7VtfO9q+dS9S$8
zrrMCah~1$4kTs(XoiQDiBYvvAIf)ZGEKIeu0pYIW)G;q>&tVl9D|>8q3zK|Rt@k6%
zqqzm8!Om7eA~E}@#L&~stoFiB8g%@&J}XNmSvanvFvo7Y61_2*{R5Ms<ls|jg0V{x
zqp&sBmganR%7b08|BlA-SgX33QTvm^TS?7~;3dcCoYIZpvx}erQT_s$F4x}Q=bsp1
zyzXmvnvynw`VdIrH>@2=8kAgQxyUdgYX4kv({^!R{cU2yE#_G&c_`4i-<<FSXQTaP
z(K!h{W^zO7MUF|WbjPf0MX+v^KrB*3i*xE|VaZx=fm!Hb?c@s=sXPhd4Fl>kH9EN8
zCN^oxT39KG*`47*gI<*RadT+x5uz;?#-@nX#MG_b><*JpPoEYuSyOpu43)yk<KKRT
zyV+v_3y8U)v^=>1!!Yk3laRVIi@p9233q?lApHa`!DC0Q2gamU24ZU8lNm5sA=1sJ
zCa%@fX%0<F`iv+H)y-h3y%LHhTFill9KKtrWC?ZDwhub9?eCMbvRRqF1F{J99Gy%=
zTW&-7vWgQo)U5|x{(^#H_RVcmcDtq|$Q@{;DZGE$iP->nbYOh|GXfH}66Hi?WKHUM
zwUthBq8qrXo&xf+<5v#Y>4XFL4Wk2e?V_O3jmGDR5eSnHv3qrb_U+q%`qxUVaay1V
z3&2?Ect787-kUcj$mZ=z*NI5NA6O&Y5_NSV@MIx#AN)w`D58lr8%LeFQ4$igiC73J
zyF-9rTv%n$_61gB%R283pmTQqiX~*T*lF%gM|kXyW~gatrKhF6TyFLF`}+g+@9e32
zK)b&G^aVh|<u+PPMT2wEuCO@d+58O+o?eW}2FcFK9;M@wG4g3SQ(RKc{CC<`NZJGd
zk$`Z)B*EkXBWskBfR>?(=Sf-K{u^X+oUz*kPaYhkD`$)YQqWj(gN%Su7t@v`?|4fe
zgo6DNMTLu=Flc6G_EkJGKIx;1jUYuPlhHs-%-C2mV0Y3;Jaq-#3KQm<bh|WUJk6v9
zzde<%R!<PkYs{K_*XMDNSRKPp;P*fOITFmd^{t|wElUb{cuwVuBzdc0BtjRAD;{W7
zYIm5{p>94>+N$UoEVBA}kylf51#p&0N;_0aRpsY9GQOf}N}0WsXs)KNJ6R&A05KTI
zn^S7Zj!!G_6ye}VMx+t!3o^pRB7sFS0w9t+8XP0tAF605N!0OWK<vp~HseA7;m50g
zV90A`+~ZjBRmHX<zdvO1dT<ed^;G8#$iLg|J8yYZ{*81M_V-OAFvM<L{ZJIUil<fF
zkP8`!m56}X0Z(26VFgJ_IIh1GD3d>>ZMa^G9ron_t19WA|A`FyNL#mxs3SW<do0#q
zibU_R)|)<kKdn7n%GuOy85ly>(it;UV{*OKwNogYCPp5|jV~fEZ}S)}NKwgubSuF(
zr4qN&wdmm=rGD7i%s)=i$l_GfR2v!@b)Z-O($WHWVo6a$e0qRVPoO16k$!Yuh`=p*
zwBbFA`|)*-0iGI=sNKLu0;y1CqDIpSRnpE*_JHga4?F~tg_T?+d1C6_-<g>>X#=R|
z=c4w*9GDCMa|Yi1L4e;syj!>^K!2}yJVIJ&?4UZ=P!IB7UN$RJrPb}!Qc#seWazp-
zlvSa%2ewBY2=jL`NMO2!a2BkL^Q<v+1atX$dK}F#F7@!hubf}o<y&&!IJ!q06$Q91
z9usuLFiuUJu0&Yk5p#0(jgLo4OP}ItC(<9aWUIPiZyyiQQMDLp%8@-zN3dD0pE9`B
z#&z(js3oP%+=PXq#>T2vyg048u5(8>38}>FN*ASvjJ?}&G_fQ23BX{Iru@|#`Y}MI
zX|3aPO}>~QOfhnvmsK!d7zcDhasrsb3g;!P;(0cPYbm~ImM>fcy?;ck=5mfz&r!-!
z)OB^~sYk5%CCJ=Tj+ZkLu^X6yMy1XQR=fc;Hu$Kh*SkY;(@K>%^$63G9B0R^SC;z;
z!Bs&XSg^yHW_}nu-Ytwn8(M;5N(u%fke|QO>kl+)Ya({0qJOw|F+V6@N2s6B(y}x0
z279cid&p_BOlctQxZ<_3z1>zC*d3L)e6$4XrX{q@r&ChSpsaz#%4`vdl3Hf$@R@j&
ze^O~t4nZY?<QZSlxiA)5uV-5)4f2bEAlp~}lO8R{Wq=0jis>*Wo^sK-_IqFvW4|yZ
zB?S=?arH;+^=(WBtQtY)&lsXjV9x_(mzLt7qD~h$_+S~=xn6XDD85NbflvfM*@Yt4
zPWkDPLz(=eszeZ>NT;o*bry-|ZVFUn!lvbvH1Fly`->50*7p9Lh%s#XL!mAQNH$wr
zTK-S|)pa>g8(wOu!vyblH&H&4XR<t}bP%BA^t*gkZVJX5XYS-v%<!S#=a3afpAk|E
z|Gw%umm%_6T5d;EFL^<>95-RI1Dm5Q;29ayXtTm5(qs;Zk4L8)lW<2hu_+zc`$OVi
zLa@}QrCdEktPD;=$L~tki`v_P>Z~%VRYh1Qp6+wYalZcc7@n?qW2Y_}k((eri=(RY
zR~N2-Y}Nf)N+ab^wMfVjK2}DLsdubfLHsHTE(Am*9cyWLiHh6{RW6-^l*sZ0s*dx(
z%4kJ<|NGv)@qAcQhWJgZ<_YP3y#C__9nNXF&t^9m8)k!J8qNt))@P*Sji=U<Vn&zW
zvLag6>IvP$Yi8d@;HzhNU8GI>&^^cymQbdQ+`%$ib&VOx?t<wue}3z8J6AT0X2PoD
z<<WomVI5{gEP_m$B4oS=S(J}y#5nQ!v_d9t$E{@Ei@&K7xG5|sV4F`n<M;>UEQ*Y`
zb&Bc)nU<H4R#{!0_3K&pi<ojb9x4<S*{HX43+WmVt;JJ^iNZQDnKcJb`IQ}k<2h(#
zi#~QFf#X$Q1SBm)n-pEF+H(S4HMmqrr$LdGM5o)J_`CH+I=+BlsA&W%ieGzdLI(F)
zaaQ>Zu|3GT%5LyIw%-tE9<nw*ZWy3xCE>qfNAag@%qDlZ<nsklO?o|RPR|<tmQvlh
zCtt94e=+bG=HOWy>+JHG6o3xSNhADT4N5ME8F@R^Zin!CQU3wY|DL$hMlYukNJF(<
z8hHp0Jbn-<S&dQR%Kgm*`GoFWPa~X7=t_m?K#adeh*_Dl3+oMWcL)w(Eix$i!Oz9Y
zc16oBtF~(8h<<n4wdOyt6<LYX;<+?E5Jl)>yS=y5c8}lY`m>;0NZuoP2y8!!dBbk$
zJrKA4yCH8&Wq1_OTb3oG#rn6B;3B0L>&Padi`H1_yNzAISogyUw%5t>p2b#JwF9+}
zba@g>TAfiRS<nJb4RvL4BaK;-?0t<>(jNI~$QZ4Uo#oG?7KGrsKbH4FLqS+>n>I~M
zH^Zem=j;(W9Kd-rjdvI)+iHj!@yzGb9fyO-zp*da!hhUnH}dJ-jiTYK7+8z<tw(8@
zO^za;Ve3N@W?vF+8RQnRcvL%0hdto3**GNQ9UIz+&du%b_rB&nNKV`=ROxWpZTHe?
z$K(=M$&vGpP&u(JST>AEh{s5D8}0d2=a)G0{4R<>lnk6V&!ZCMeO<ur5%ND6&1iBz
z<083gyX;yXei*?VZ?$_-+|B%?Q5vdUtL_*b6Px_Svf)MP`n_-6AICu>n(Q=C9znKt
z+8>tY_yqZuJkVq&(rT^sW-{W9e`B$0^*Z=6tu5Fb4J0vc!PbbEmoMHw7eC7YMhXUs
zQ)1699U1W*yR5+on~`8a)2@Z5d`4jJqopc0zk*|ICq;n<i5yq&#WfrqL>9fxS8cd#
zY(3W-V1~l0ema%}`3BXcxj!W=wzs|)%&8=le7e(7B^~bFFaZfP-Qo+t5R#^l{22u6
z|MPZR@cUeQ%#$O@pVMQ;ty$lfnU;yKs=`nxEbpm2TX|?!C2pj%9YdASv|(*wiRNv!
zbSZ6O@)ohRjcaLT$-K0zgLY6_O3M9_#;dx`>G`SQb8R~H<UK}EE3QuS&9-8H+q<d;
zIJSbg3I>R7)s4|W;X-rc3o7cdb)Ab&)|;lAR4-x9P2(G<-g);c_1Et;&q*j<Z|+xQ
zg_Hy1(l*B24V)-FTZ}(HfCCwV2f)%0D&B-7t#GXtGn7|Bml=j^=m+t2y7(E~r;+GT
zrxQj@Vy#L<C#{Z|SKV0^b5T$euXan<*ylKheMMFNq{`o;Nl>#-QjqSLI2Q=Ja?;El
z@9e!Dhqua(vam-JPa2n5Z3fP=#Flod<uG+P<ONFN_NrbPFGT;mn5aKVa>^?c%qg%8
z8FY-g(8ABp9JZ%t9tI5P!3C|Hg-AFzO!jU;Hu5-KjfdH-llu?<Jepx)_lA<B;sHq%
zFGsAoi{`CPp7d8NWQj*lC%Vc?{D{3WtYN&`nNuq|zj6P>mWF^uFkyEzY<_h`JyYMg
zawk6BR`Y?!TkW#I6NT2)la)m{dg6qVi=Z)by7tJeNLYxu;L?%D1aJaMx`75=rT}}L
zTW+{ajrdmF_6t_jOF#N%d~)ibadfry=cB3aKxa-8ZS#t0++Kbqpp|tqW{k4Zt&7?T
zXem6nK9yfie=HA3njOYn)qZ1gHFn$HQL3#gL+ttlhDD}ckbNRy#fwi;VU+t>w4`rC
zww;N3BffU_&_Fe4O`i8`4cU!J1>Iwu7!oXYbUh+AJ-(yHoyNthO#|g{+IqD-3oUS@
zpL~1tZ_IsgRS{gZ>gA}Q>Vfs%t>@L&`e4!SR}&?^#^2;+&t{85LN`<f;bd2}IK#~k
z@>%F?wJBeg%nEo4dFuH4gS9CeQhCReO6Le-@A7kwFaD|$NLA>8fvT|(WSrM_8l<a|
z>VO92u_BVJXsqA={`1=F7SLAZKl<a?dD0Kf^5X7W+hdj0ZFQlCGxD48gw0WO8S|o#
z3UP`c)s}p7iRIa6V8kPhi;znM|JZM#omQlGyuNm<wI^Z9wcvI9RYGMj>e5xcpMYac
zn5oYq_`{^A$FttLwH<f)hN)Zkjo@?x<*Jm0_%y1nrF?k6=lr+p+xt?@nTbwqX%-%z
zXI1*AZg<@Go5w9-a9&=_P8^r3XC()cuk_mwy7cYF32RGtCn390G|>(l1PwMn<}gfZ
z%WErBMxwtggr=Y#y9Zi8wLT$y#zL-^wO<<g(pZsz0m?49T)r6bN$+?(lr9d}X7|uJ
z42n}9#67e6xk{QD_Bj%Z^q#=N<Fhfd#pJcm-^7=uU(t>s0KV3oj2HJvsqbh~_UU-)
zQP8BQi&T|lgg2ldIIZo&_vaix65SY@9&QELc|a}CFCBCdi2Di^a@rmnb>qC++0L1N
zhEs=(5(d4N+^{~qzq%V9&T0&PiGo`<uemHXIwEPU!cc{;@p(T&LX_256o1#~-W|p@
z9nhzq4RkQ(z|+kXub%K+p173sGws#`A48$kjR6VJkJ`*1C%UlYr%AE9q=f8XAAz%j
za^0q3`jh9zC~Tb&d&c<Y544=U(1xQT_P}u*egk=oZ@CM&0B*p<T9xN#;)|NQb8$9d
zj))!Rip(hR!lSe+5tWruUD5t+lY(P<#gNC*RIH%@F%Hwng%-EglMsvoE;g<2Mc4K;
zwiXubQl4o#Cx?p$>fSA<s9d*u+|F2#xtHG659O0!dI1;XlDcN<&BwfGb0<Lz2IRt+
zD>o~bk~qol2v5IGQ3bIUlb$q|1QZ}}%&9f7i(ez~#xj3tr#1WRJn^Wmx9sJ3Y?n=i
zk>uh7U<1Wp`>u+ApA3d_G6e(Jpm!=0Q=vRl$OzzX7`|1Md&J&V-UzoCs8-KW?V8?`
zCZ`)5KH-tKaX$3F2hxQex}Z!U9+pjsZ*mIGgbR_|jxPlHA+{pTgkc{Bw=n~H(-uq{
zZ;}$^7X!=uJVQA$UAIRmqC8?`7(oC+M|I2l8~ZXR>^K{_nW6Ds@5uOZ%6nlU;o<{8
zLM79`nuEz8dJaIm*fO`B{sO;|frm%rp)iHwPz3W`8r1_Eh>{BtbEsRbF^zxADLX;{
z|0eQNj5zJ^o3>Z)2!tpR@)~69Xs56bb%p-}tLR=<t9x!@C=ICB#P0X)RqcYTs}2bz
z_TfshVm!Xgtb89#dF?;tB|C~Kl&08z$iVP~@*o7-vO{ouSk@A~(V?@|zy6W;CkKg;
z+x|{diH@mfQ*evc8C^-<?tj&5#Jp&Ky{r)=%1cpBKxp_<1#4Z43D6&~v7?!OpOzS_
zeOO0RAi|J@#=5{C&r#AB!Qke*pB6vZyhnl|7d;Se5mww3I2dyOM07-77BI3-z@~zb
zvsC^vx(@<*)V6VpNB|A<?3#dY0)jDoEf+yTpWQf~pTI2-0LP*5lAi>t5lnFey_nMF
z-?mxJ?}8b<v(2EtdJ1-q)tL?k@_q<x8fMq)hRddYj`QEl;*DMbd6Jrf`vT=X8M}WQ
zR{)ADYcGD@i5EkS2biKy3_ELO`hUK&RZpws6p!d0n1$c~2n`QlT4144*BO4m#mWh4
z$eKViZ;hLEyZNh|Bp4XU{6T{@IR~Nnc0rC&$~n{l*cG7C+Nkhm(Wfxf%|ea@F+6}E
z_tUs~?y9)l_&W)qA2Jv)32CrUUJC<^dDGiL4fx}YtY3Npl&TxR!9i_c1nEJ9`vSKf
zzQ%*0fwVVzR-~S=!^FM`wDCN?F0$R?{)KG~&f_b?*XD>Pww;c{6(TI`BeXrd;i3Gq
zef*d*Tco(cDzrvR68{iIB%s()g5YAf8SorF6gooz=**sh1%*I}&)6VP92BO!f@z6z
z2s9jh_4nuVpG==@RW4hh9wjD8|3d(BCg{5~&CB_`Z5e1A04uo>`Lg-bSIY-2DU9z%
z0f4j;Ni2@`?*n~?kRp04`%{M&J)T4GjLC-qfh>qKG3V=QMg;!a=H2h$Z=B6c8>-x)
zZYJqp*RroIN1Xu^312;zkU;*~9ASD~LRL0pR!3VH5`4;lXbAmtEQ2Xd@%*2VOXngJ
zY$p)pW~~VcIT$L)DLk>uRc^ALS@q?|!sGthKx5nMQ9Kqp`^p!Lh+2MIvNvX72C53?
zfQrA?hPj0sFu!aCW2s0<%^1dTk5P$u>N3(toL4?Iz@3j!0SjxJfu=iTcxvj@q~vbf
z%1AWgClg{Hm_unTFiA=dE&amxO>^ZE<H7~0hhcsx09T@6m?VVPF_Z`kaF?qLi02VN
zW9Z_BShnyPu?Ch2XGs+<aCh*42F5>P<3XYbeqmezv_T&mbM|}cu>@+4h>IV>fq{!!
zAKaxq0=Oj@K<51m5f>C~mJOL6;EV1AhLdwYqD8$~;MW_3S>-zfB}O~zNr*gB95?}L
z4b^48OwTq0lxcC2!u`aiwLR&gUfa)zLiNFgwWEM02}sRwhAb4!2%M-_Z1M>YC@{YS
zfEdkOz3B$8B^|h$X8XkGImaD~BI5)ND`zr@?aKva^R4qW<g4&+irp;#ti+x_62%ad
zQNbRcbaxwhFc}agKj;GD4j{(>)I9(x3qbbQ=~RH}<O!#0l!2>n&Qk5Qb4!LchSCfY
zCHZGI;AjOU5j{de?|c6|Uohr|Dn#2#O7y8C-b4LvF1UjmjRoRc-n4*Glw?T%j0*CF
zIJv6!c<rK)do+$Fmly-@0V#Z(n>}+YBI%A4f2@kcZWj-op>E3Hw{7)L+eM1}b-8K<
zNClu~Nmc0&q;4G@0_FH%%<mb08kazdi>INmk0~wMdPME9<}|Itv+RKnQy9qHGC*mF
zXTXBddDF|rC*=VnN7SP5w<qcKIovG|ExHVcN7L5imB~b7`Ip$+4NjM;d6_Dyp#(1o
z3%_Og_N~e*Rr_1?lrf}c(AuDYH#s;lIuI%mu?DFj5I2>vixvDp0+YUO%)~qBq>WXn
zgs5BJxkl5Waq!6D|CtW<Iih5{O>5_-xjNT5mCYwd)Z6f2QJM^^)@rV`0_5M<%IjfW
zJ+;1t_6QOE0>#Rp<h7ultNk$zrspIlg-GT`BehtzdRO@h%e?Ok7P4av+-S$C6PkmR
z#k=h7>u1DZ1jorRS)oYdh0EFnGD#Hc%li7&MD;(a;{Q-=Fc73Exgs;r%^8rH!oII#
zBqVZ(<+m$V?*%rW?9cc_fU6nhlWNg6SX0;~&)(%J>t~V-oic@qbK<p=Cl5ne&o6^6
zi8_9N<UEs)?L9n<p3n!IjfRQWn8W-5GJHo<f1meX;aV{K;No9%8UhBswo{lkqOn^T
z6l=VdFm~0?$kF`1sLrO+#4XN@{32ik0UwY8+*+jxiFllLk{3D4V#@l2gKu!y7ukhD
zgP|Uc?lSk+YW;TNXFFjSg4+sIeA1t{8jnxV$=sNQ&|r!T0oUo-OA|`eUAG33qS|F=
zK<Sv%tgvj6+s{9?ZE#Jx-{(E|eJJH|wT&Ha@?;Q+HS{6dz8P08L@j+C=Xb|*x$yQB
z8?Sh3F3%OK@re;|{0i){eC=eWsFxfz9vx}vc(UJTeA!tNjDSo5NlSLer`QL74E{a%
zxR1q9TqNn}lk2g!TUY!nVXLcu$@c|mSrX(`Am_d4aey8pQuB(mtHb#I1u8B<^E)kQ
zr#Jj4fM!Q2>{<#^Vd<xDhV|FHC&_tk;A$0{d#t-lN8C+X<1ee*>QV~9GWBm&+2RK4
z8y4c+M^XmK>3iXqB)}nk!o%)1nwVLCgdvWC8qxmR<K{4WTu6AqD`&slHszNLAt#jg
zpUGAlL4d0HjqozhjW;&<O|t`fCdZJz{QxOo5}JRjV7^)oMtoT#d8Yj5X@Su9pX#Rc
zri6+{1FT1`V`A|*Xh6IHs)+zC@P;29v8NEITI(e5k1WhT8g&wm|NK|%s7xK|qj|?%
z)#;|GZ7JD*#NR-EUA2qzT;103ZYl{L^ppAW%~@=t)%i)|k>qrBcmwr>Y|Pju+46S`
z^t&@Pb5;5Ap7+dQV1OE52}E(2E~lyDc(e3F?pFg6<MtH>{iOg65L;PFpK%VVjQ;0X
znHt|FNLly*0h47}xyZt*uXh-wX5DCjrrzbH`$8iQN<aYG@$s>;knzxO`4q3i3fd^S
z-UkGPL75$FEPmV|ORKCrC53hMlNyNBCYHhz#LeCyGX7E7RRfM!1ia&7^q)XPlt(dk
ztYoaZhy$0gB6uL=lAc8Qd2dW_o{CPsv&BG?fGIzBR1hSB3H#l(M~zzq(TdH($i&5X
zP;oAyyd5O3Q;qMx;nDnd=C#oZNA@=U(ltH2w9^+J{VPyULKwVq-bv~^48;MT#y6<c
zuEs((yJmdRo~f;!=xO}~S{mi280}D6ckf5<L9y)y&N-?+(bdu5x4L~isym=rPFo*6
zLXj|EF%?#qvlAA!ljD4un`-G10zP6Do-_ve)O{_%;Qy%RR2abKAUPGSMiIx+D$*H=
zUh7vzRF1~oetJtv1{0J19P9jrpR3?lIZwcIUI%C<bwICNgIelr(4_NZ(rW%zr&+av
zBv~1zyUn_ldRMq3j)cG3(2X;3Mf5LF-nd>dZFqc4DEsHQL=3d7o9lIR3#!N>qFyM{
zlx=W+d=}34=9ej|+G8<)2HDeY2B606ee(<0cQ#Y?DRgD=y7Q}R^{s-Br+6itXT$dL
zP<RtENYH<wUO90pEx@O*H0hKtC2f&XETm{|mq%nJc54qrS~y_7K2K2rS@z0%^9;+E
z+A3@0Xea&PgwD`{&fz6Z&R3+p3lZ8XYl$IRV7clH{QuzX3O62oSx&ge$E~E^*EIE*
z4Ate;d*u6p$7;9&)w&{eshDyK1Qj35#%l$01UQgypXtw9#oqGd08-`fO<KV?8VaM5
zD%7Om5+2-8oZSGz;HAv23}y8Z8zTrFIW1w6klE~?EtjzOs^dD|@9Wm<BXC7pkve)*
zu;J_|FZ3OH2^;*jm)2%yVK;3Pd$r%Per3q?zP#4f`sfswq!+&3!X$=#-np7b@WEsQ
z1D#ksPV59D!QS3#J({ofok)*`fb7p#^6nAJmbYaRRpwb8uxqz97|}qBH0X@foj*@?
zZg=q+B~E$gLkUzGvcxM}oUKk25b;^K+nd<4_Qff7FSHo4=bYht7DJ#vS&;5?6tR0z
zJQ*DH<@Xove?dzJJwCg$*DZdX*V5*%$g0RGB|3Xa;(O2j7g?sKrN`9t%Ls@zA$wu$
z8ik?&FT{QMFu#NDV1rPoUSF(wZf#fSof;RAoa+Tf7_PLY(BV(_3yr6cZGTAsBiIg%
zpt45nrgcRGD&wa}o)8YQNqrlU&mY2o^7W;?YRGj)UDzHH{13DQQy<(q8cW7MKy~eB
zOf$2+ucFS-KyPKRQG-7qjl!WXOsaeTTp&N+aOI0iaItnf%zQLQp9wewSn2Ig0Eiej
z#vw~^0xEOY^b$CPNjDFWnpW`Xd{Ny2_6<;pA9=d`KXQ6uWX2h9JrU#9(mf15gfK;K
z!+swk-+n^jc9jNbB?f*jQKF@X->R;_A-~9wm;8!=s)lMx9GO!GBO>ods(cCnK<0lD
zq^^iR-U}*i0s<(zCPX;*XA*F(CC~#eL3^?_)_WtL(V_z-s0aaz)#0jNSu!Q5aoj0s
zfaQQ<7H*~5S#JTb-?4D@R&_m0Kh9^EuI#l%rRc1j`v^E&J(om;nM7VQ5&i7%;=xpi
zb4Lk`!6Z%}h#*zQ+oJs!eQw&NxKVbCT>ii<2JhzVq{am_yrK8~?fP6v23=(5`@wP6
zI6NA0I(0@3U92usRyzmN>&P@FZd$b2=z8bFp#DIUkEOl9vWw}Hsf8#r+J^>wQ{|-|
z-Z;xKW8O!S&i|+(QR&>Rlew~}>RAC2xWvT+YnP`KspqEyTHoknW$b<(Af|$ie~*li
zMF!uJv2^~jk-{Uq244sFO&3EIwwDwV6In&(?5Gjy`V8avMQ8`;th^puahGp&Sk{CP
zf3Oo;UizyGBRs9;?lz&mmhOXI%yUcmZ!2t$NgcNNzB{@!!v4vKg^88-a>`BAKaUQK
zJ3y_cB~Mt2=>@I{Wy75zZ&1gGz`h##)O}^|9hgGrQo;!PYE6h;B!0Ndu|MEKn8Fb<
z9-mMqjT62^Kv3gBBMLGIp{61C%T<0ofQeF9`^Mk@didqqqp|&4TTL7xJ86pkXB*hm
z|CHlA($rW@E_RIK89H&75OviDP1pswVHvf*M*|x?a+y0V0@X=}m9{4o5<))ivK-8a
zm7xmn5}yo24-5txn>MYbz6vjI^Pr~elED*ypzNtUc5qSK(!&E1&fzA5<4wS4)t!+Y
z{^r4-v!4#+-ffZ-T9>9tBhf(R$GyUEyk8g^5=%PzlY%KUQ5n^Xa`*~AV+HA5v)xoC
zvI#r5twv6=vh!6>d*kA6ay!oLXgt0?|J?>ruzzd{(2H|#h#u&AL4kfIiqSpPxm+RM
zo|)(|zCLP32EN!$e>0Q}pCK5pH~4a5FrCBMC^G7~GD2?pt@CL>Mx1KY(E%}!@nvF&
zO$lpA?{`%RPmmtRm*At9iHi@n<bXudpqTK1tkH|^mc3V$Wgf)M{cjOV9^%Ia$>3{6
zU1YMdLB)ow);3H7cdL!^)p2PdUyl~=94>5%R5i7(a<cbvtTO+r7C>pJFiX>in>R(y
znmSg7F%Z~J|EW#_4^2A*g3a$Z2q1M5Py*)%WZ2?n?o>;t&W-Uy%>_DLINe_vnopYB
zCS(hJ9&D{Y$708K0vyR~rQ{3(9A%6zmKkIKcH4e3H{CAjTvWVYWYo1uueTN&ekWvv
z0=uW}7rM}9P?svlkUk!SjcsTob7Ca(>qFA_Y1id`Hw>&8HcpzBSe2}sqPH-J5AKYG
zUut*oYgf|QOykB&PiZwtO}9N1m3LY8BfO`ow|B+d4~WaUS-V~2H3Y|KsuMgpHN5-I
z-%sBL6QIW6{eG=97^u|w_w2TGrqVV_n?k=fQ%XQ1fr!l5Oh!DV;j38n)wC3?)Timv
zP7O}zMr0S!(lp<<7rUn&a1l+K+U!@x#Z7NGj$2wjCy%W&EFphIdOcG$)={4uc`KZP
zgRb(Nc7Cp}JVPr#do-@)<6ExZmGQbfSI$YqfRjJRM@(9Hf6p(B!;K_WmH`&g%zz&%
zK(_R7QTpSGd0=;Wt7W{#ZZwj9<%v+zN$8@(FcFX2Z8y$K0tX?--wrmX#^R8Jbhsa(
z<%Ogc_g=|CkIA<8XWe!fN(x!dr$RA*4h48&LH_>=AsvhDVjvLSlHQ{NyG2tVbkd51
zgO^#j#Jm#O7NZ$N7jb`c=xZojr@q&54qBDjj(DmB7>L}0*$n({Y)H7|ZZviJxKs^L
zQ7nlR$lv*T6ntZNGBoFoU8$TPN?Y5@OaBpSIEbcX9EX^$gS>lW?NkC)&N5fW#W!y`
znt}84l1R5fz7qJ2p0)RFmpf(^0Ztrze_9c3$mN_C`_czI29A7&=>{jBCGekK*23X>
zzH?8B0J{equ;{OuL;w>2_fX)PVM6l1yGO&?m&{%hrYQ6nhV_BlAKw#wPmcrt-z4pI
zFKBxfV*f>=fbfmG3YL#No%$Zz;2Z3uLb~#}x0B;}bu{o%q$V7}I+FV1A`hY`>zlzL
z&Gdy}!`0xIb=DSC#f~o7rRU!kVKrc-9Qjrp=cLUw!Lemx;Z*h~42g&GtdM=ix!fEH
zK6t?Q?Sh05apP*p9*-aVU|?#m?7tJlvNOD_pC;^RJv)q`+&|@Y2M4&);GiHUt`%V4
z@Z6#hWd01W5x9Jj=sbv+0R1^(Cx*!2ov1zhF>Fd-K<6E@@XKO8Es~Je(RKm+q{=;5
z8JmM&f$e0mKy459K>Af2$ID7b|9|9xCDsi{J)llBNH$QAUUKdMq2xtortY4WlXGV9
z)1!2Hs(UCK9>0)BRNhGSl4bX$Yvl0UGpv+@6SJK?6k?><XDxGLAl(lF(tQ_;GXRzB
z%~j~$B7^4?)B?xzDPg~=un|e3_CKWtg&c*F*YRB0Q7tX06v=Yaho8O+|JEPB`dTEP
z0F5x01Zbu6Op1ZIv)gja)So_)3pdRBhPH>&XCm#qg)vQ57sjUk4;>}pBo>;-zsI-F
z5Vl$fHm*n9^ux{(MNCqa>X1uGQ$E(g(?6YN3n<pmM*wDs2yA;&hV^aT>pA0RFt(sl
zKzes*(!JyyKLqOs3`pypn8M;{|97f+pW%_S$z<}&{5;ah%`c!tcYf)AB%97QDb%rJ
z<t1WICiMF(AW$3I-tG~2Pf4Angj3W4U5Cy%IEY9Bm26IC))f<5?3_UU)2FDl7JPJY
z63S!Bn(GT58cV~v<B|15w#zJm{-dvsw;ZjY7X#15K&7&!85&J_Fk4T5|MrxRX?1dx
zT=^STK?ey;d3eXdv*zSkTDCjT*jQ9vZm!MRH(L6GeOm-s*)C$?=eD*awFPQwi5)K8
z3uF*KVpB#ZD_4G3XDr?Y+AmV7kec;E@vn9B@(y3Jtw>-2RVAtdQ>Shi$K}mk|0q_{
z*}JR!ln~YYThrJ#+i4iIFa2v3xbKatTv}@20`qqFKh9?4DWH913*!HgU#j%>>Bn~y
zYE#j_u5?d&;o|Q+=(_iO`)D^gVSL@DawHpn{{+;jyv#GVlwYT6r$r(s%1t-BFl-}~
zrQ2<8Y9F)ACG(=VmD}&fFqc5-EOpsLRQyzV{ettXce*{P=X{bY$1<A6Hz}@N^W;_1
z8$H9mup}wm(W<kx>t`KId7ULq9iNn;rAPf~$%=0IyQ|eg=i$N9gN~5VUhC40WrVf~
zO^PM)%rl2G+~n;&j!D_5u|~wRk<Q{T^`OrdR_(%`65ZZT=5L*T7LJuVo@1rOYh7Q3
zyb{OP_{aBazm<KgqU5k@Uaa3U)2q)BA`*6*G}`*&1Y@hhA`*hd?Ot-weVbP1>I(Hk
z`VRtNE;!~#N0*LhbM-3Wthw81R5N{nZ2II-{Y#fY(JXu%!Lz<Xyj~nIEv77<3S+*u
zI@+jPIu8zyxFFJ&M`?fk;lVJ#x%Ql4X>Iyw)*oKxaxMe^IoI6W-a*P+awsK@-Rt!g
z=T@8+_rlPkc$qe_;mQ5OjyR-GpC12UaH2R1CpD<5)80|kH?ovwBH$-hzZ}nC6gFP}
z?uQ_|aYa*oL~>Fbp7~^7vy$^uUrZg&y@<>dhKRPGbzvA`TuRf*AH$E|v!f0BuRv1`
z0fTvW=`@EpW~TY%kiQ3lI)(p>w6_eaYg@KOXMiLS+}$m>LvRo7?!n#NgS)$Ha3{D0
z3-0c2!QJ5w);{~(bJt$)z59KSf5{xuN0;iYTD9ucIJG8jK6Kz;kr~ik1tK?F%y?PR
zm2a@yj^ag{oU|h%5|!6-+0x`RPkfc(#WX|xN?zoYWKgebRJBu7HwSE53d}Ba6=YO2
z3})X4Ci;i#6f_}B()xdSv|I1`TX33>nes|i#coZ$8YleLj>%}1h;whV802K`F&0!`
z()%FKdT6DUB6WC@Te0jR2>|YtmA<V>g;N8T4H1W0er|Wc`4OoRqD>k?I^=hpv}6G<
zyYn;dk0p(6XYegGb7oMp$tO3AQbp5wKvl7Z>VrQaYSycFv|5(zdiLAa-B9V^H+Ozh
zbB~HwmsfN6Uay^gSw7~6nB6Km+|`tP5&<G23PNqew3i1HiO3f@_KNyw(h*r$)xE_d
zLSzD0Cd2C1rk-w^>a%o-v0e4=W$)m@Kp$?WK7OwtGXdgUyPA(b+^PS3)kZ^UR98E<
z$qi`tQkyWhL`}Kd<5r=w32zt(>l%#LY^THq_)%i+h$!Ju+sIw?d05j!To#Ar*U6Qt
z^vJ6k^j{TH%N^$)cvPp$7vtRLGeok>SYl*g;STYj??Dh!`@#w+AdsxKmW+JCrVW?Z
zv7w>i$9a=pE-HjFC0xEY_o^1yAcyxd@b=8z=<2HSb6b$ngmBoipm!o>el*EZWuUqd
z1wwby{VkM6l&wPa4NQP}oshp4z0;`sN8>USHnHLK)jg*1_lVuRtIe@+Y4Hds1YQo*
zZi2v$csy4m+vK2$khFdv72ys{jBFFS)p7g?6(xRlcmWR^HOtv1#M<=+*DFvyKH+C}
zF>tIlTD)b3ukP8DD8ND_ubX3~sh+jh7`x9CBX~lSqy*EQ&GIYsm>N|_2^)<*aR2U;
zJ3V96Ni6_-)&KBC!TLgLq(WjIvp7Y*AN}M^X@T!w65Mm1EfYrVEFJFjP~D2cRcNDt
z(x#`EOxNcf`Uux^qD%G83Mt+?+w);L?GF&ccS(D=@g#g#Kr9bYXT997;B%aoEQ6_n
z!Q$aX+$%KBu^~|r5(Kz~W2|;-W%236TICzF5bCP^98wu-wdE-Vq~ay1goP8Gflqup
z3JrK}Du)?-@1hiiuN=;{ZT~B^k)7GKPb$TiqAh4ZO-8-;vcD-0nx>gE;(aIXU+--4
zD}cO6>0HLm#A&9pvisg!@m?Ftfd8vVEI|8GGoRjA(4hl~>VQn?h(qVSe!_NsU?XN`
z<2@jp41v^a|HKexttI~FBwHdkw3vHkmaiQ>Ui3}mEZKwZa*|D9#x+6<u=8JqkN4~k
zKfM1U*RqB5PFnGWGr}POAbJ7v;Zlv>DYtTA#QzB%|4~xR6GmqJL?rx~!D^&z4bQp3
zKwOPFveDBZ9gh-eN`yZ)<PLt7!z2`r<vLtPy2qL)p~?d1IqBdT#ju6yx(P^`4(rN1
z*~$G(#Xhj+vBq2c1S61`gtaZsiMe8oGqS6qtFS`-nUaJ;X$i_5iMIv>!jhkLUuwRC
z!NU?|`<a1|xHSK?@iL0Pi#sN>z)kq=Ecfs<?Ok^97qb<5a*?s`1ia4Jo>OwhVH#bf
zSeGx_oQq;ASl!#y#{TGzDA*&^5t=8)r<=I@OJiMncjpb^|Mt@sq7HLdo${Mg16sUo
zL`4a-pt?!8qEK1D|C5y!BA7$y<OW9PwbTBil-#=V<PyVSMH~s#-Yo1NV~X`up>hJG
zGhkBOOFpgpRD(;|eWFy*u~m|4=l+*)x1grLvb=T`vVe1U{khB+UWwgZGS~Z)y2C9o
z_V@o#g;^+`C%25N;2NJ#rKHU{y)WI$VKCPiAE&?8(LY@c4|hG=wxDm>WCB>P2#QAP
zFE`tk*J)ds??qJbJr`b59X-4Lq>wY$_bo?;w&6ETW~ca2aEz*QoUHB0LUrR;aJwUT
zK4<av#b9?5TtzW(xcnotyNK5$2if_Z1o<2nShgs%J2Tf3FX5aXG|vsLcoM%kVoBWf
z<cEO(x71Nn>IWJG(lcI|zo7S+(DzVZhiIl_MaeZAW(MDQh!3FH>5bR4Qk@qWRXg0j
zGr5+Uy@AsmFOqYq%$42}-bk?{zNqiafD-9%4oE_$eyrOnA=#S)aswQv$^z)MD&?X7
zEbwJZK6wUiI1_TzJS)@Lle{ZsbW$=2AC7+>!W9KwoA%M7V1~m@$AJh;zO)o4t%Z#o
zR?}uczP!y>IAFehf@Z~jY0u)+)z1&1y}$ZH07(4tT9{W9^M2N}SCtiNL3oo>rolGc
zDI5X{KOm*Exa1fcu*bovyI5xo9gUGA72d9<&p#)&3db^QO|0FRf^MOmw6x8jOw`R@
zI34MvY?;lJj&a^d;uDia#Zkutt}TB+b5;I<j^`}<=gcc+_-Cq!pyvGYOBoi3qva#P
z-=OTAFNNvPfA*Jk34y;nWu8Qa$pqikW(1<dY+<t==chUTsS&>UdWS9`EhkNyj6O~6
zqu0&M91>rbVbZS+dL0pQfYnavY*j4VNl6_1a2UmX^K*(2$t8S9QHkV!lEPT8)*C1&
zp?c{LnYJ(ulbnx|!}I1t_>*GFpkX8PYI{2V@oF(<uASHHTe1TwTs;vVtjv2y;?1|1
zXZD3GZH$dYJ+%+PZtZm9J8g4Xgke2fCNY%u_Ekzw37m*mLU6MRnG5mq*Z$z^9A49F
z#l4E8zCCnpvmpFcTxgE<^P6><QLaS0%PlFd9}teb+v_Zia`ZHouR7OKXV~Y!6ip=O
zp(W*P`QtFIg&3E!U?$+@fv4leT6-ZyArZl2WH|pkJ+%rOL$1(Kxv}(gBEP>V)Hcq$
zhlTtuSvaO}a$cRRL^fW1+1aY4(T4<i0t-=Qe3-T<xs{#rTIoJS$w&Xgr!u~G@W~1Z
zDI;a*)}?MQlK@8MMoj39S}&%E@xY}h)USDOl0dKP+1PTbnX>YPiN^Rhp_R@c%qEGb
zkaBNYO`-{&X`0%Ulz0^NGLFV_Kfv8EV5K&5@T=dt98VtZP=i3eBy#d2LdO~jMk;F>
z*lvPmn=h!5jG$i?ZcglN%8PE3ve(YKRPaA$tJ*aa2R?v`LeivSWN{rxDxIlAg6ROG
zQ;xU*F#@o@D@n0}x4lwVEK6<U5(;Yl$7%F$spk85QEBf%fV2%i1IvjxE{neaLRHi2
ztfZh~E>SlwKR^<n#ML9VBPpn~ZK))5LB+~H6^Y>=C0eHftp|+<qe?By;TC>#yw4Ew
zK+zN0*D@@IRPk7{!_*cywF?<9tDYALYt@}wc$`)Z^~cSp6MI0%MXpPpRhKkGkyD}4
zXq26{mi_(RyCsr0Y_5$|Z8?>qseHepl?La3s)=%gN3wR5#>z}GzqT^yafI7Stv%}L
zKHaXXq>(;QGuUiIE2ykVWd!gT0}pPo_<F>dw<+NZJfF-MB0%3~Z&@3P{_H*5CN;nu
z%m*C~jCPONiL+&iSLGJ_N7~B9#%+3f0QxMl3#W?Q24{YLAKxzT%U`af_~=LN_SbVc
zU^Y%Ich?mq&s=85*ZNe?jx*aX?Yp}QP?Q&w_JOL-9;!N~3^bGUmp5?9XZI${@>{HR
z=gF=uxRN(fq9dB-tDP21x#WD>WWL};3E652WAiRVURxn|hFkmC#y5K+bnoG$6oBL(
zMwX;XA+c4UL8<%J^{3VVt^+1KWXEkyz477B4Q=WL+<QzD`(Fh6(d$NXydBMrZo*Gh
zZT1oEQxX8SJ&?+&{XI4x7eHnq435GOwk|2*OSYmY*l4sY<^`Nsv&rY)c6s<ACSRzU
zxegfkXXdq4)$%<$?$srS?L~9b%0WvQB~IQ{WPlXStmoLX?nq{76xs4}M@0Gw6e==y
zel!$CtA&r6PO{$V{$@9eta6$>f2<;61AD=Un2zY?Yvk(KJ;{5D-x%BO82K1m9hqC4
z1X>p<T$XYjvOp79VakQ@IzdPS=Hz|KdODhc>-kBT3~YZGBAL88n&lM|VQ6-c|N8J8
z#7Jfba~kaHvNwyZetu5mW(o`c><6x9S#H0A8W>?Mo)^~;*$TXUNqyE@EAn&g1wIj_
z*%9%yF{5-y|72kMnV=OA8;_SCPj1+nt?R3Ng}Kf|)}!e(%=(nGc`6wu-H$%_LxU3g
z=#f03(_+^(+LRv+5IXJCkY*u35>7&UMIm(M1|c)2so4-Ibuk+pgETUThn``6qPp)k
zKLIH#xAb|9In{Xu=qo{fA<aICn?Kx}xT3NPX-^F*I*lzMAOvvuVna-lG9|}{@4u%#
zH8@Oidfe<d>bFS+sAhlvh+;@uex(FRO`(Ur1HnVQsKIUWZo6HAQu_wGKLPm<kV4k)
z#K>}q-Ln0Rh>>r4kiA3WQ%$CRlJDE=mm?st-jV-O`_g-<o^49$!YK>LV9vO}gatFL
z3J@dTQn>`94Tv}dKv1c`2{I9r#R0Ghus9BEKOlHdVjWy@ovaiEv*qZVY(LEZ$=^<d
zs*W8k>r@8hc2C2(V;BamCr+crU+KI)K?*th0Rx-yOW&6)6C-I3$OKKh#pltZ#{dr5
z4G0J@AQgshHry8_#}*t(%;RAq{E6Q!6?Hp~is(Z^x?xe_%6V{14=<R5!Iv0^b*kU|
zdK#~@*LcYaCZ*6hDtj%jc_O64)VUbQcZC_96tWM2hs1n%777hVE~(kLI<!R2!}9{v
zKmk%lW@cs@tk$Zt-5+S?J7}pBCwd@2Dl5A@*-12a(0K<}4L}~-&~T}!_sHY$P58G`
zCI;a2b<gb(0t7H12{;$IIW_yQLs|a!i3vE|+--@+7=f4=7}Of|)c<{O!lP=)gqN~9
z-6tTIIY3!O10bZGgRf5{D7e;Cfb=^m@Pv*n7x;Ezf(KT>#%4?fKuF<uT?PnQT4n$z
zy?&;&8%3I!oD6NoeE#bQ1mygLu0YXW?h6OMA@iySRKT=M!V7je2U^jy0b#r2Cw3s2
zm6KpV$N}yaDz%A)@naOu>tSUF2M<hCU)C2#qwbCV$xy6!3!jkyVRBvoIvX_T3yj8>
zNvpLr3MolE=sV1V&nXH6DtPTkWx9|;Yarlj?Hgsrx$N#Yri&F=m^q?Q85tR|kZ^1^
zACUPmK2v^4E=p2jYxSfXz6CWGe~?nxnB-oa5`NR(*<P~q?)W1Fw*xVG2o;-@?3DOS
zpkP_=PBOUXb6iaTCLOVj4aV-FXoxG>HhkST{}n*)hNZ3HmLk_r)-0}qf9+5%1R-z*
zfTi_M6-dnIJ=^gpoSZ;e@1KP&BzurKNA3+s>m<Vf?i*)DZ9Q^)6G*xG1MWj#9;0jK
z3ec&PK&LKG#%3>czLPXkkdlgsjYUcVFZt^zkN3K%kh=T|M%D8R?-ROM^l!c$V#}<e
zT&gepN-wb}Dt9RtF(U=iCLu{uAgvCV`QyG<OTDhGdiJq6R@j2PWss0=R+&ipR}t#L
z5fR=m*PC4{AC}Ef&XlyNi_DmtAJ&(H-<HXv!`aXU!eA18CWh&RtT7hoZ}R>YH2B5G
zpXAJg_cltyb8H|EXE^FM-C0%5$j}gxOIKY!y*42vxFn9KlNt&hOaRifWKafEx9wI>
zJ!<T%*CdE|u9l<*{xe7(ng|&|*ub|Sth{%sPzEZKo&`9Cx32iJcpNQ1o>7B*Xiqae
z9ioDWs`0jp0b1+v!Ui0SB@_3|KF``SEXedj43z))F~mp=^`&4C93s1t2G!~Kfc=eh
z*<*l#_A6GJsa1ACfsn*d!r8;ch2r3oO<~5Kg1blHLJl0D#HoXNm<dG&U40NDLl9^{
zW+sQPvnq0R-|PvolBwGY+6i2wgi{`#O0gsCL!Dd8>rIM&9MSp}gLj23Gap+47IDF9
zy?f!_TH_NJ5Sf#B!Ln}xc7`4-N#JiqZZas>AJmc7x?dWAiHWI|zBd{`oaThbRpr(k
zRK0#%l~@~>R{j&>qkIwY?2#v<AFVFL=PjviC_TArViYlV!3s!$G%4H7dt*QGcw^I%
zcWv(6(7qCd)zgCvP=Mx21I^8#i6Byufjz@-_FSR^>kL|(UM^^UrA=k990iivwZ2&u
zt-J&+hTavzl?VN1#m8M6E~i_wrFFb&K=)#gOIJHn{+g#CqcE||$HMSfmk-TPZSJYS
zoq$(qvCb%}q9Q04cK{^Gf}-WB?$HrO*N~Z8=HV<_JHo;e28ORcdN0*7eVhNLqQXWj
z_}ulh9KTJ@hELMCl>8$@JFl?#;oRo~<ri}92baZ!O*Y5s0b%SYQfB*m;^h%vRioL0
z)J+eKT=XMTyg1eUy>BE30^^%d70~r7SjQu>a``vwXTT6G9O(b1<|Mu70!6bYQy-IU
zMCKSQ6f4(5UM^%dWyEPI)_Enu2kN?9w2!7jLAcmC$!#(@i95*wX=P0@^}sx;j0@nN
zstSwKKe>E)W=c^}w`VFY@>D4QhQ=Lj_?uplSk<_1uaBOKy~((*<Rb(MDr#{tB4u`g
z52x4lPAPAsLN?-KuVOmyYMKlQ<-m|9_e7m>P+Wifno0$NrPw8_b>a^?e4u{z_z(m)
z!Vy29!>nDxWfmhD?v6as6*(O-SHmdQz$m7}IE>d)#7tBC=`x3l`reFSh<bcwSV-ju
zn=u-jqhTtt!k*VdX$+o&*qrZf0H;^Y%^w4ADsY+~csP;Ji@TBdbeplP)@a92@AkpZ
zsF5M1=5d*YyM9X^9)?t?ys%mAFh`X%C$ei;kIp-630d5vl@zp73x4LOu2mJQG#!_g
zP}-P6RCa$M;V?u4jS~h-jw1^kWmCvw1g29vBzpewI_|R#Bm08hR{UtORky&~o7r}m
zTO+OK&gpDaRBNlQEv%&vpCp1_J_cr)xvOkYO+~5&31WmHLcS7ugpf)t$;|&m5JR6>
z_u^gXuD!<LWfbh`Bsa5N2b}BD<an;DZAq*mujI)mr#(6}y6zOJNX<<r^Me{=Yd#H-
z4}cI-vdI_8P0Tdl5*-MG1<39>Fe=gxJf92CvtSy;2RFi>PbTMDPP8NFOiWJJO-pNX
ziIV2xtnecd0(I@|>;!O0NdcD#`?%Z0WOqf1CdPpf)K7`=0`B&CILP+bxYKIIQY`H@
zZV8TSa@YVpdYlpMtM1o>k^J29!Z<b72cHuoKir5Hy!t<X))C^O5qJ!OtM<I3)66X_
z%8H6UPs^=31s+EAB&o0+m_a&ce~G5I7$9lDmm!69V$6-+m@@?V19r&GB)a-{vEx>3
zY-aGy&k*R2gB?s1lX!weHH4wEoTh6pbukCc&cryG)?*kjH6xyzKpi@rmz}PF8%eTm
z(O}Ft;p=Orjz_ERK-excViCgomgHqg+Eg?X;-l)#b=^Gc{Bj&e-rKkh<=Be%pat+t
z#&-fmS>E+^5zCwuq}5^pbUI4jETcatZ(nMx+s|zrf21KSx|N;Hl=j5OOQo?&rE(}}
zX!-c-3beV&!x=kIPwBtH9u#QlX!!!lR=S3@$Ol4AHJ^Q{-Bn4W<E!+OlbWj?aF75b
z5ppXNJ<rl$^DdfIT1=#!u&}U)2Uk!xp<R4?dwWAeLuBM=xD*?jl+B6WLLMBH4VOc2
zz#7rl<OAob^$FgGGb9U+#G+gwEw0XhZ|=d8pvBGzHTv%+#gso^@o{l8&dzLf;qi|x
zy+|wg=KD!>RkK?bBFF<SyKn<Mg&9)Ve%lT{l*R?B-B<uZzYo+HKz%<sV(x<|Mse&G
zjhlUaermVJRBw3!4JqB4j+sjINs_m@`GE)-lsg@7g2(z$E#nWz%V|&J0~(_(K@w5o
zzEq2r8XRI9gVqI}=nsm4`i>yriJ`)=M*9=G6YmP+K$WX911y5kI~=*p0`y&$fbkuI
zZ$rk$;o#SsLv>wvoCE$SDa!bg@>;<*Ovi{;chjnxcOW5t126&zMkwfOBcVR1w)TpY
zcA#qQ(^S&VW>JZ;D9O>GScUtGq&c#vTr@br2=GvpOfj~9YXK0Xi{Q7^!CF2tAP?vE
zk)Vd=<Yr#hu+WbyhBm*-PX>Spu-}>y2|P5X-p)fgrRw8{>V0Pg3H5t0Tg4fD9T_`r
z2b$84-QV9~e1E%AJy}NPWaN)2B_qs?TXdbCp>F<(xwXsPB9j<8*ms!d;(lg5(PqFo
z6!;u%ctesl<)`)AS9Oxbs($5(K^h%@*S-YnZIw?-7Z6>_%Y3*6C$JO0u=9%i%7z3T
zz)k;agqi2iktvBTFxO35L`(O=W=MywK%yA|l%F9fOa_uCE)zj!o=Gs3mdtoOsO*uG
zxAs6YNQ)ndDd|N6MGGLLfUf-061265PYuQDfUj|%)b~Hl3{4}7V6+Y=20ET^4)~6J
zD5SuM8zW6GY6lVfqBy`7xk~wskMZ}?EC;Dg{qiL6<o*WA|0qNT;zrr+4tACQe3Xy)
z9;<h0^~|bdy1Yq{NLNS3!3Qui`(SSu85EUs^{eK6T&eljHop0}I0;yhj?<{LT>S}v
z<q)L+rl?zmF`zX*hJhY5TT^5S$;;PPXB;?QpD}(LXk6cZi0r*5Eh)xNl&iC>(2nsq
z0o$i$NK}Sldv(|bWheU9F57O2`Oxq9uz)!+ZVvzt8<i<@D1kLYSO>Pa5~;#h%TA0w
zX!wi#T4b2-25%QpaksD?8vVGfuWt%gAdif76BE(0ouq&3Q?a)`g##Mjw%I?8xb9-q
zsgV;&q^pV$<|X)mHfGMoJx@jv6sG_~g$+8xduu(_PZ8XaZ}6s?*7X^=t;^}MeOh~2
z;%|?EetQhd;?R?_xE!=hV}icrFf*dS7&KL{f}(Z6BLmf8D!yU-Uzb9@UHT)h!o<2_
z#cu}6mwUfjOr8TWSh5RE#%2Ze8bz5Ev~CRyV=!Mb{8s4Mt&xb^xPIKH_l@yJv6X%~
z^{HLGL9Vg15Fpccz!HOmabz8!D)-W&SWb*HdMH^m`$R#v8eCkms=+vL{mp?arG(%>
z02KY{?TO!U0stVS6M^2Ghe^`w2dvLTGA!--4-b(!v9v<+E=NT)mmf>@K0qS}q!td2
zsIx&5r@^N&yC1;!G5z6ei~2rijPt5{a3uC3`N^nhgT3+=6}%c4Ha_Mqk7Sk2PyTHK
zSo=m!yvGSu<}J5B-s@*9y2v+_y<1K8WF(r7+hKB;N1*#?AIC44$X$ZoeJDFZPOZV7
zGLc6o9(tVHukf*<GUcSmmvB0^+VCg$)bAPR;H<RrmGP#>tXU|8E=8Qe$_1~ptDIgZ
zyW9%g;CDb*=IR=j<dD!MJ;*n<l)0sb!U)#gl`%3X&A2yXg{Fs>##U^4x;P$ryjIqb
zr@P`fzpncE52tHDi>XyM%0EapYMBeedGFR;P4~g$k#etEB6C~!_|BUZS711R09zLW
zb$h%(Kww|2HN_{+Vl2`dpB9N@CSdt$jirAyX*zq6Y*6|qwRI=lnmk1TQ6M;f<9vXA
zrbGkX&R-)PLlh5aArbNszjyLWiCDiSAK0=aa*cLaasEq4SxIH6vBjE4%*y1dOk7^n
z{jeG)p}*|ql;gWmZQ`!qfmPJ$g~Gdbh-{B@n9>)?xl74>=1MT#qKj6wH<=5MBVX7A
zu>vX7rMZ>o2S--9V|;YY$`IRccALc&VS^}I6gWK~Lf8!Ol6Ut0U~`!@rY+4&!Bf45
zJZGO%hG=qo?+@`XzYxJIzdQk>#U8iut^nY;CB_4ZOLEf{gShsA94!!0UJlWz*$-!?
zCe@-m#e^bhNnD16Y2Jx4<4W8-ITZGg9CQ4Knff6$7EA9}+)?63i>M~Yoy`P$iK?km
zFLA5fh2f>vaXeyd>3Q2%B~RBuxWc1ubGj1AB}b-)&EAcT^r=I51LCx%Lu1u%sombq
z*Wp^)&vuoL!<$t~=lD_Itt3@h+<?MOm$!Nqy<a`$>b<%e;1AP~RK|EpUhR*L583cF
zHF_<!9GyGr5-yobwK;9VQNx82S>7hQd?Wm&xMq8m$b0w6URep=uM8Ppw2vF@q<Ima
zDo>t`!rLM*yTX1PdvXeovGCrN68e9}DoD+5o^w2H_DF8Ui>oLsDcfRxQQ$<9XKrAe
zP?WEO#Ilwup)Izu_Kz`DL=p%V$8XRJlv`*bKERK3HxuXexIwkI3ogbHK=G8}Z3}g0
zTZnK>PyN&SGnWiUZkT83AwZao*cD6u=H-E7zpu`VlIU5M5C1!`fZ{!HU?c_)16Fbk
zisBdqiVA$FGIJkD*||HKmhx)N2Nh&0F$$iJ=1ec?lpqnh(zG*m*z7N+)$I3Z7BVJg
zJ0Xq2g?=Nv%pn*erTE+6c+Q6#Z5#Lx+5>HxpVv%P@Y%yn<)!7~;_{d*`SOFG>_T+G
z{T^(n#~SAvrz`R0C3@=G^S=JVd}xmc$Ngq|&nO<R!<B%-W(0&l$_RzMV~B^VQf1T1
zbe@M|g5p139={nP_%WpX=SzOXlCQ~yfq!&*<WQjo?SYEwolaZA8Y4f0XUgl(9>bJ#
z37X=%q{~R&{R67?FOl?&P2JJ~-IK&gMp*5)DjoUpo^BiCr!CIZ7<N}TO?$>&9K|hM
zM!oqu(+w;0Uj(opsA?c9us}p;Cqu2NztA8%n9Zrf(@2XLta{V!)4g|V1`nB-2L>0#
z4ZMWya73o2mXa5HCxp(ljcubtIG?+NB0dEwbL=arAGqIw-zmMv@u5A8Q;LC~l!jo|
z;Y^&}pp-2ql?yAtVP5dEy@a&1qdNpg0s|~-a0+(OfjK1^7rGk)^`ql)8fCEtGam;_
zJ|5To&w6g=_bf6*Q3>-ed&m-4YHB)RSG!^!Q0RLqMuql^&FoU-XN&&*wKzddOvgLC
zci6&x2m058L@LKJrN{*S(ckBW`>?m#b?~$Z`5*~DMTk$;O+J6A)(Df4yw%Y{e}T$7
zdYRe(>+dk?6PB)CLAVcgEhtzUfHUHnK2dEsvLg2_sXb(<((v`D3QEr(6L=^UcM44_
zf~F-Sro~{bH6&l{8H}2^VF}H~1C`o05N9cQIs7HFzWo?!X)%1p|5H}zy$wxycxx3l
znWpb?o21a>Tav2NAdET6(sp|}kmQ@%@DXlVKax*W*5f3tcx*du6)j;<=APKCi6)^h
z^1*Ou!GoXOx$iZSR)=*($?|M$?v{CMb-lPreBqZxw7qCgCcfacq=XU7?UHrDVps9>
z!_B<7v1|M@|G>!DnD+4425pMNMHX-EYI<DzU`Ye6-gpwdrn<8n$a4@^aNsPCfDv}2
ze4~_BdztyGQpg#4E@z_KpAkCaIGGAe($Psbux`M`RO{Jua89bA!?&8{a4(!<Vq}_|
z4<_hoR69J%$V1!C|4JW1?cZm7>QKALNqD?)TOXY*4{J3HM&LI1$*%n-u&bisgiExr
zIdjNM3s`e}4QKVUtP{OoUhgv(rgnuaxMCJ|YL3WRzYeE7^j1)Bp#1{HzGE|bhY%)<
z|9chMH13He1Ye-53h+oQ<R@BfM9xt^Jb;Z@Y0<V<JUS(ZZW)(0c#_??eGlfNLk3lU
zTi92bv%eHK_vfHgZN3FNY&d(^ZsSeYg`O`|5jqktidpqE{O<n-deKN=fDCZ2`)V7i
zcMHj;v7)MIE%K5|K9RbVCy2&&cP<(yGi_mQ(;M!H4q#r)AcQ1Yd?QJ{oGN$S$gAw`
z-((<BasamU^s0(1Na|eM8Wqcmyl+lhk)}WroiULI{pEy!s<F2EP;OSs00dm0DliDW
zQD5=_boK7-Mw7-;Q#Wm4C9N=7kxl{vP(CuSDd^?%<=wTUYD#>Ac2oqg(yb{})LQq>
zDhQ<awrX4gK6UsNL^gL3=tndf8|qO)fySBM%yAO3LSX_Og43!feHdPvB1u6k2*eHk
zc84y;r%PQS=!FRw8z!lKx0PwI(xP_+!f$B1WnwivZJ@VRnJ`nTgi=oH!@;VAJ863f
z2-NikJB}P!P%IBpoih_5fwm06{2l=#P<lhR&^bKA`;A~ZaG0A003lNknA5lSz1@^e
z2V>7VN>Q?@$b!Tx<n#5sj;%F8p8D;X0zPj)%kK089GS{iKw38Dkyw-%$_WDK`yC$u
zCPe_0Ghur#P>y?OXfAtIaN_s_!BgSiKUxek?-a-XY_;1AlI_per6p^K)di9`<=+t@
ze{V<r(~aPia1(>KTZN7Rk_Y?i|27R@2#%GIlVqL{_V_vKA~XRv=v)cNAp)Gk@9lZT
zM5lVi?jp(jRU|ObdK1<pXrbkw0WS^c&8uR1{2b?T7zeg6DM}O>cA!k<TLTL{aACq$
zrxW7mka>H4OU6l(LA6wIg7~)I`a}VRk)mhnDKwhuT?8;afbk@OdwUPspYBF19RdWS
zK%=UU&qj7Jd<4V!(xC4}Z}@qRI(!-c@km+r!T5T)rMaaQ$xoXRcO_sT2Y_P)K|x;Y
zeroXpZiE3NDk$IVta#r}nf&7GPbM}8_J9dYA`*gNL$s{x=86#VVZy5x83`8;Le`a|
z+q2(CafmsY)LA**fiCRHX)stdK>aGBLzsKJ9YyFnJMi3(yh<R^?<GVtOQQs=V!;V#
z4%4V*lFVC-Lp>@9;hm-Wdyn1;+#ADNjgPl@BnG?7>Y{^gy#ae|d>+8(Mu+3Aqy^a4
zzD}m&qAFN%*cCz5J$IOfS4$MC4e7_0f4VXU`jOB-8c}c(73NE3{R1k?$a_P9GXK3(
zy$ir^o1Ua>FEM~pQ_?{ln{U4-YWdM-Od*1QnNoYP<~?TkdjD(ykNh4+q55@mnTkgM
zl7&|bef@}q$if10sLNH|fEL}h)>+2jTQZipe`Q-!1(H&??%>I^8xEoWS>P~u?USab
zp9rk8J$B21?hg>*bTWc&`M0Nqga$CERzm{QzZy4Ay3x#;eJhH=x|n7woE%wDf;3|o
zebV+*WwTT~ar{4V*eUm3F9S=ZaL`tm`KL%28jhb2;}SrL!MzwyKyJ}&cRtueI)9+V
zOUgE3v0IU)>thP{dN9l6<a#~SHp>v!t=!8-G@V>!b`_gWv3J_585Z%7Yw_%FjXz85
z54{M{$X7%mkp21wSoFF)Fy@fvpI8YWsEE10wJmhWNfu@ec{N%OnB;Ypx_4meS8lbQ
z=^d;|tm-RD3F4A6er1A%Y#>rXQ?4iSN8z4(TCWU5$Tdkei~XZL*Jd91cSt2Abr3Yx
zt|tx0kRTYGdwNi)VtY!K78dX6)NVn+iCO`!lL?Nf=?EhEEu_+`sIpT{Pwe)A6VtGI
zO?jpvxRNr3sBn+X+1<fO%Jt6yYIl6rGY62Q8Vc_X-YJTL)b$=H+sxL+)nT=}U0l06
zs!_w4SDsfvG}1o+nXnhn%9gaNxb7OX8wCvy51;juQra{+gv+;m2P6{X@D2oHqT4=9
zWqU7}5dl@7R4DIp1bdo~B7<W@G23OpkKOQoVtW9RkR3Nkra^PEvg)kEjjF@Y7p`s#
z91A(|UPeMQ5wl}DR^6Tb^v*95>#h1SV~YPGf&iRElxC_aq!#oU+|y?eff&H*X0}tK
zm&{=RmdMAp7AOPRh-yD-`*NW0Z`iXv3b(F8(r3%e-4?+qrU(>lZJkrYG1!z=k){7R
z$zM4s-jrL0+5hEjEsDi_#oV;;;P#PvLZH*|%Vl${q{8l1NhmWB<#xF+wRAW6WR@v}
zM^*hiMxOX{G&L^ax=z$OvnR>M*2~a|qtH)@@PD8hRNTU-8LBYNI^J?=z?B5kX2K6b
ztOdsbHVGsEt%ai*=_7wiL0LD$0uS4hJIE7I8kMqlJiCJb1a}6lDtD0;#z1v!8mxj)
zckAJkz1eyCcw+ZQCw;+{Lq!%9-<3LrORt+!$H~7g)aS75vrhsmoz>6-uY@|HfbXXL
znj|hSnZf>eudBjKu(p+iu&Ci100}Q$Jt{6PGt}E+G1^Ix2i7I}uiynA=<ba{a00n6
zk-sb39SDu6#mOF`&P}cEXQ#vp+wo~;yIulyF*4JsH|?8SB!-gU*~~NArVcPuQp_!y
zoAseVSAaqJ0wb8iBbftdTQgp4zFDmUT@3@akF+|wNQV0kE*}f{>R7hLB^*K_j{!%d
z!1vYP0ORio3mD0PsX0XfNNxlpwJd-fNzNCbXsV;={@b?i!@!*_0W5mau%V~`Rj#o~
zAwO|KUi}VK9d2;2f=xVIvY@Dj9qpBtc?sNhME<1pj{pG>0|1grG(ClFg2WWrrJ#2p
zHz?qJrilHFEm(9t7IEL;A4ax`e7<F`rtcV)WBbR91WBc~T?K7XjE!2S@T#IRsV0B!
zvBvd;u^#5AF|c!qw*GYi(KThZ=AMRmouO^n%CmI`FJUZmVJuHS1UnWm-=_KDd|ocL
zY5A|F(&O(YG=gnLkxMx!U9q6X*?rlh79W(eKo`#BX!IO$V0#Q1QkFIn-NY_GdEv5G
zx{VlX)*>rn-+8P)YzmXnn%|HyBjT8bCcLU_IC%5>+i4b#bhW#E!CJLGj71Kc^&tKq
zPCvIN6{DXS55Hnk3;Th&eO!|*L>&6t7wZ~Vkctrg*i%8^H6XFzpm9~SG*?MvQmC5>
zcv#)o)oE8OJ1{B9?oof<vYE;3AkcQF)0=iqfTmlGY<oW2^!`L7(C#YBCVR?qr75_%
zXo^@^<Jw$bV082F^c1!>ws4i-GRg~#usq3d5GKtqkfLT~1V5>bn3Vz6!?$W>I(_V?
zAlb6C?W?1qJVZX=m{hjXwf6hv)a5w(zVpGr|8oIyB1{EWGkZM}qrw+3p5Ecct2rpa
zUCok`i046G5A*sGxY1*yVdl4Y7{n*tuS~}p(RMr*07mMsZI5Nke}Tv1mx)(1?XLxM
zq4{})#DL46{hAfo94~ve4;Wvo)9TOd#7eUt2HHvAHLC*>mCvTV*o^SQ7xve%cr42!
zmPZ9a31BBf#5!&ix>v%5QP0llewiZ|@o-6_=+h*pI$VmdQ<fcdeV(r00=ZEYZZJ5z
z)M2~nC1*@K4W+ToN*g<9^hRbHe6O;2a52B!v^2U+$9g**h{AWQk0W?HC;cKdzEDRH
z2p(&yC+eS}A|M&>8?}0h*@W{#O4Zhsv1D*yNdT$>`cp8deWfX$vu|#!oflK`U5tOe
zOfOD8j-u0$Ge3yXF=`popG~uKSw7M+{xYB+YNZcWuEYRCY}zui=x%h%wzUjnicw&~
zeyyr&zE?SWu5q_-M?h5B_0JSY=Hof)(QY>h1y;))%#GtjIbFGjhT@Nn?XA5-y%K-N
zm*0yzA3+4L%d5}&n$3?^4w^jYTeLjG8t0GT=hIhk0+2G&XveQYXFF3D7Ul1IUy`_1
zTVAaUPwj{fyi>;KU$P>gu$9+}J8fGN+Lkb<Ql5(t`|f)m+vvAzm^x7VY;h_>QZ^bG
ze_-}uZ?-eaU9HBAb)PHWXeXwRmbp8+m~Y@2lk};3UFbe^TQbf5V0M)M*m1|E5<8mS
zcU-xo-q?pZm5ie_-;d6`^TZJ!c+{|aVT03$jW;??)P(QS3%tNuQhhXk!Ft%_x2>jZ
z)*Ve;mulj4FdNT97oYXRF;J3913HS1GkNH7-*PiyFOXonzS947Zs_$SbUfNV;gl!9
z7dO(*Z`Rm$t+cwfBvm0PCDx+8bsgB08Vo9QW6N2M>}}I^YQ|s{;J6>c6*P1*YwR1r
zCLgv<#BB`)kHfn2ln}OBt`bLSB|ci`Obzj`E2(Fe*ZtaGD`4LEGHRHWUgQIV&SM3P
zle5En&f@Ki`ir7-G;N)X!9i{$jH=h(OZ%a~1*uR<>%$|@IzMeP&?M<{lB;mC(qd3c
zDhZGG%KT(+<uiC(?p`YX(+q+D3ao;Z{K_E*kAOkmI9-r@=2vZM3t`9^vP?=mo?dA@
zmLK>$kD=`BJ&JIIq3wlP+P!UQ7PJD8G0ZE!qpt0*G+M0#;=neNeix+%a)cU4`d~8u
zj#X4sy_eIerm%dS@VG2SE(eO>Cf7_|nO^zoKTf6I{f)=z?IU)DX#GknU_g1V<qer9
zO95$%?YctKJ?^Q5|1)V$fAU%}!7=@}x0#4Q=&mwZs(1++|5mx!4Czb;-tvIP{=OZ8
zCu+E4UMzxu;zJ`UQ~2O0uGnjp#L5P$#nDAK$wBxdF}c+(%2z=rIbCHM&&}DE6yv@F
zhq#b8%Zsk_k=RMia$)I%@39g}v6Zr$b5>$-9p!@G-yqHL3ZFsjYJULnHWS?QfR%7v
zNTGy?hn|#RQ4@0^*cd0Xx2B7W%U!;bt+=1$_Jq6hYyZUsx7!D?f@p-uJor<R%?M$&
z9nrXo`xA-u67%A)!P-!)gtjEQD#@qca$|%R2dD7txX*Y(xs($KW?L6)87S5}7l^qO
zav|Pn&hkpe^hlV3cRC@GM9!X7;z8CqTSJ6brQ_l0;j*sMrk5YRS2s|PPP%*g4PXc<
z?MI85plM8j`nrcF_0)e(d{{u))&p{){CHPGQI~sBwnQ#~^9TgA2&U5r0iY$J-^KW9
z8ZyMVm833jPPd2aeReV?w%Cmx_&ErdCyQ<BdH8T0)$e7swMX|LMuG|UHG$K^mKNDZ
z7qTr0`-=bxO!suqORW6OL4aV#>1y@2{)wf%%q*zKGg((~30#L)DNk$x1Al~$iZV1)
z&O_&BDHHs|bsEWO>w|4ONZ3AG|Cj;X7q_zsRH6MXah`{fp!_Zed-A6TAa2d~$N!PB
z;&*L`^}v_;fzM&j^(F1y=M><`9YpMrc?SoUPZLhs$-C=x-81%)I9#@^def?bI0BFq
z80rVvDyBbHM$1r8B|}b%q9_*<PThD~+2{si+KIKQ%S*l!j~jZ{P+ssahJ4>O@0;Yw
zyw3i{GO4uTXeQ?_18v>!DKBmhS1*60MnM7tb(gCN=eO3WvHxTT8FI@J<|ny)fWJy|
zdiKdlQOx>{XK?(0X<kG%6<ZThAx9wd<BlC`MLe+g^5v`Y>>N>>E0RuE?|s6;LAv)?
zm)vD48$Y@8lDd;9#`B6j>wl11=%?wh!iDTv@RVHFBGOC(kd@wUM?x_`36FG?|Kz(2
z=*xd8F)`ykJ8d~)3zP@!IdjdXw^OP<=DCNqE^7(SOa~qfAL|hkgf>?PW}l7bA<@#B
zPy*>}w_aKgk*fkKNrI3rmIB*qk=@HFHy8L%Qb8)w{XrDe0NBl!YoudRPv++XLXXMh
z;Ova9M@8kYoN08QJ8HMQMHI|!%d=riJr)37>_bIuE`;)8Ya`9!E4AtAG)2qU^K`Hg
zC-K#3aXh7;cYC|B=CzTGlpzf=b|nK<WjrgY&@s0TK~ZCa`OrrGAL%%MfDg0X8qnnL
zUBShM=PBMr(^)`QA}De>du(q>Pc<%d60UxadJCT^>7M57)}5G^;9IsvL{zW&d7GlK
ztj&x0Y@%#~_-hc%txh!0ZcjuV?I%&`v7ZJbHXYd$#6b~HF*nE638VOd-~x~exEJ<U
zM&)h#U;QB-WxnWEa~rjEsEf#KtVUc2m1ig^iLhq;)b(+k#DQiZo)b$uQ6%O%y=&$^
zSCeyp?MGMSS#F`BjEcj!A1c=5S6Z%I%Dq@4!F{>kmf)9q@&Ugc<b-RFTm_q**tsKq
z!SPWkMj#;h%yT<3eXcJM2&a|KwiC6#Q$Gz)+mM|{)lYt<`w6OB_ZDD>d^iJRU+?==
z=lZc9*Cdi&*_73M#{U>Uv33?&zj8fj1^Xn%_DgirmwBoOe;zwHSzY=UX}4GOp1l}R
zl29eMo9ZR`OQEaEw0F2$TgCS~sC(A?qYNu$ck+Dh$JP2<PfChkBkBN`=l0cT7Yp0e
zh4*cxVHH!#aa+Rb#eIURFN_v@S5d&=zjTZKwC8RllZIS<E@ZTKnBXn+tI1W&_4jA+
zP1l9NRo+n3>36$D&qqATtAg8>o7b5C&G4P8OgG>|zTtXOPtJWqFaq4wQ$r&E&CCSc
z@KLUxjLGXzlNXd;T=&~oixWh2QsV9@0_HBQ_eEwsTp9DdS>zsgbqj)5&$a}1%Vmx?
zk>myW#+Fq{fef8*jTLuYz_KYxov$>9s(04=!(9J!H(?gA+-iYrg_ZVmY<N#;0}rUw
z{Suhp=g3c=H()e=@w5@?sR&4&_K{s<y*VOq_}4&%WA|4{<n_B;C|ZcIYN|gJ0g?}P
z)Dut{5Z6h-2HN(%4Ojsx|B8J_oRu{Jj>mkyfG82D2Liy$2ZjZeW<}*Vui^V&Nr4cc
z;-`De$8y!uSU4`H>BM_KB?iBM5D4dwC+05tvoWM`-F_b6iR6$>K|p~ID}=AC_^hTk
zllYOw4vecGE62XPbWbkCJ*A)w+jU>k7HHqVPuevR0u43O<q4OF3Yh1&&nJKWs+TiA
z7=x#H3H2|RfU4}jT>@?D|I;NvaysGZb&}gI_FI~hbXGLYvRP1nF`08PPGg5(3!p35
z{Ng!s<5vhzQ;%#H7-505v129uu(+J>S;@NbX3%B;AxJUZsv@#B3?Yq969|MID<;IR
zNH@N6q%0o87f+S#120=wuBA3AIYUccY;#GcLpkfftR$;f{$kr`MyD+;n{l_4OHtd9
z-Wab9_c(JU@Q+S`P(RDn8`We61n~Vb!u2yzkSwfoOZBpU#K#LdRC5z>xwia}Yk-$#
zIJQ%P_Oeu?$s!>StwY*Q$$IjjHQOsRBKk51B3oA&4wvxGpf?!(QL;E+`PAKDRQhLK
zb0cs=0Qi_=fM`j=hpu*66@`C*?Lu7AVg3)MiIAiy^3^q+Cs6LFz_L|ylCsFF##-!q
z#cVVxy14rJf6z_eA20fow8b$zCl2&qard%~Yq@@OXSQXNxxQ}Zm?fVqH<FzUfioQy
zC)f1NBhH!uM?{oJL(|$X%4+mJTOyiQw4>&~ULExRs$T1zA@-M*cv<*C%>CpmDEPFZ
z1(uXPV9pOo$g{T?N}y8tJ(>@nv*lbQly@~KHb1*u?9fT2<kmqxIh)n+mCKmY=y7XE
zd0O3VD{0~6BKf)i2k-wP;3@oC@7Hj4zC&<_w0yLJjEPb38*9Vdx}sEQ#wNLj^LlHq
z316rEgjXy!w>oU>n-2QQ!{lItuo9Luk5@Zwfq0i^P1`C4E{Fih3aE9C0(o|(gl->N
z0-1SWX*qRP`MQ2_M^pkM0=Y3Tko%nVx!ps3p(S1PL5GZa^UG>-mF{&07J#c4S~3_i
z7NW3har$h%&b{bhnvRC8dx3wM>lxjZOPc;Z4{WTIL}|O-zO(_{gCx>IrbBJq)i+Q-
zP@S0Nk^Ew_6z)hNmC;K3Z_FR@<*%Rr-!T96oRZ1^CFJkc0DyS*{A+BV`>C+2YsWnC
z$SaqL%PaBc`HjzPCZ1@wd1D1}UH^YDzaW75yNCGCNdFP@ySxIJ|DJ<#i`#NXZi}Y1
zog*Q^orM0|YuMgiB5Z9Miqh)r!s8!=H?!u15eQ6mmCax{ut|BPD%L5Fcs0t|lKk+$
z!T#nOo(~J3uZD8s3uDlsH4p^=OdA>`QynMr&x9d@0Etw?x&AZGGuo-f#?=O#LdLJZ
z1Aki)#GHbk&mxy2h0;G`MIdztrKbufpd*iE%s1z3%}OpOLsM;p7#=MS(b0%-h6RhN
zyc_<%K>mcr{x$&eq5cl?y{)UHgOVz%TV!JMnNFh88&?!CU>Ae~)atH6G)|2&aSMqx
zA>UR$_*Zu(U{lk$7s_gZs0L&r;HLV>tp_{zj!fh-87$7Lu`8ICuaZ)v2N9ZI1bDW_
zvyPM%(_ESzddO#){Zl3EDbHF3FZ%+5f`Vk!Q45*-h49*(Z=<N`=$`aN83TYYu_OJ}
zCu{0KN|98D5$ClnMJ)*}#Z0B?qN6qsAHz+yNH;iL32*Oy#Hq8)QRqX2uSEKguZKYq
zLMIR35~9`w3X=r)ld>_{F>@}`CwIMHb%T<kevya73}6T-?xrgkqp43K2w;$HTdGa(
zT$+QvJJ65I*U(S8M88n}8i^Nu+^8Vq9Q^@ofs_OA!4!}YP(x3Jq_jifI%Z<!ZRyTW
zoxI+{lX&g)LhMIp_pK=>#@Zh%F|KaCF#bi8-c?$G>!f!We8Uhp75*5B0M$)@o%0U?
z<sQ|5FCI^5L=mhGPVD}HTVJ~Rx1-#h@wIsTd7enf|2SWn!;1wR?5<APWhX&>w}8CV
zMOfok5RUm{RqoAJCsal%2K(>Wa%0|c4*f`{PGIRC4!g^yOK(v%47)#q{C`ocEtzT*
z0?9}#a`;L$RBMi2F4o%d#GWF}dKyaaM6!n-43xJ0ej>jcnTz}Z#eT?d@)tf&_wRW<
zqsd2cwq^J4anI=GCp)>iwTg9ix#yu2q|PD%IH8)(wHcDU?A+YIeV8L3W~;=qvPxmM
zOOkf~uuH1iDRz8noa|i|YhUS;`nscNdN=n%jS1Ta)wY^`rN^m15Ril<>nxYOoN-DM
zrlWg)pJbk6ros`r4zut$<qUuWoXLBCH=ZVms6YbrjX>U{L%D-{Ce!}_;57dLz>6=V
zR(;#QCjB1}{L~U|Wn%|<hkSt@fZzsxTnUWJGgoF{Y2Vy+K6dW~Bq;gM@kO%$C}Z|D
zV+1{ZZQ;B_Us(H4`M!p@ObJHQSQx78r>m{T?5o<v?H0A+ZBrK3goUyH3wMJ2Kk&Pg
z1-?aXd(aT5w@#Mq>540+v1jvgyjEN(R7H&~`djToh2Bw=2#Q>Ho@-jtBee=Gx%0lE
zj~`_U+D16rbcaU7`Dd*W5jAL0E)WDh(VAD{!+J2A-vv()c0Z%%N3q-LmxqDHPBGD_
zF)h=XT=nY`d*^$)Tl}bf3aXq}w=+Bpt=y&MzCu!3(d2aI?|INy%W%kEg*_B1?^U^R
zR8_@YaF`5yKPzTWJsqw|ZrQ95an`zZc++DneEdwTBYSUiT3^H^t+!2AX!l;pw!<oH
z&^ylIZZD4RJ<B-|dZ#-5^2p-`O&iFM&C`tI-IGQ9g7go}_w(%zPb%fv^Zvy-lt(NP
zh%*I5fXlHsunnH_-g*u{gtx<Izq_I?X!B^ej_gt6x`ugSZ2+S3PJc6fS4-}d0jYNq
zOz-y_tj4J$AZPJiIQBE)aW71pS%oqkYi~dm$r4BUE1E7n)<3MNPaI$k<(2~Khxr)D
z3f0YcK?PZlPk%V#B=Y}&uT|?JUp>Z-OznwSGy9J`)NYWmiFYNR^;DFzsB`E33u$L$
zu>Xa$A+G>r+ozYA6hHUs+)N&$P!Dw`xAx*{B3uPXWUa;H7PO7Jw?;&ih<6WqXkWNC
zE{jI+s>*-DtQ(crI&(3Qa+~{cOHI~!z`QjvR!)1@Sw2j|WkpilT7*ouTT9Z43_c$!
z^|Ui>YOd)J8~$a);Q15U_79j}hjF(rHi0zXva}Jasd5T#_2YzKfoF(*)N7OUAD>*g
z1=RI~+Pgwng*!@5E*Ap_b$RXzslzuwkecDG>H2R{S4f$0hyRV~NkAU}Ss%z?7=e7N
zb@I<#f<Kv=KE@}b3H@&f2%y|Ifoyn8UbL_3Rq449XDYyQB<FX%I3dR;c^h53a0I&I
zdlRRA)V4&0ccM8{>bycHLC~TRI68X>-{Pgt;2y{l>!fDmI@JpRBGYd$lv|f)7~o0X
z_vqO~S2uGdT!cR>FDqSBP+m?Ygn7;g-NJsI#k|-VEJkbN(p`CW-7Pu6&I&Oi<&o^Q
zeMmk2D1)zV!A6kyDe)aZG4<n7DZm=OHyaFNgIKyfg-V5B%Ue%c<;WsBVMzWjc3tU%
zvLuN6O@%1==~f>J9|C1KcTJqa40%U=Sssj7A=ooyqwaQY_?|*~IflZzf%(r(Cq+S(
zh$8R%zwSBD=N<fPX+ik&EHoyvCZ_%(?r3haqGYUYN7(Dc*tnqlbLAG7rHWlS5T{Me
zs(`k+!9rwZ^*7w$qG3#0y1D7Rp+m=^@MyMQ?5`r7um<91<v-@;A}?+5IqVe2zmW%C
z=PUd;9!j%Y@b=<e=oWg{t$5V&<+h|0s>)-u4wb(``yHVVkCW3@6w0Mp>@6K$N#Ma~
zgt@>q8~LAuMOZjESa^6{({IVk<K2~u2wnfdrcbh6BRBxzeNak%fzV~uM2DCQ{sEu^
z@m_YoV!Ah2^DE13m|<@z(`(18+qR;KMEDPh+^!jXO2-%3D1+PU0c;f~|35K^Y=?bv
z`oz~S&X*?<=GE|9G69<jG+_P+U4M`34vqFLGm}<idMdu|%`eSa^EB3nF8uDEUg=Qf
z*{STyu|w0*7IVWu_S~4GNY(5KzuQ69XEggqMcw1JA6V|&IF=!q`5F-n9q}(xX`7F+
z>BWj;2y4R$LQd>&y>e#B*%C6=Ark&wE{E@knl|P2$kJ`|0Hy<r1aMd!H0tc~Nxs15
zuQVkq5)wWk3EovMnh=nX;k<SKcP7<c&>`Ma+54~}7($7yBP>|yVQdxEDirdcSllC?
zrJu-Gc+0jQ`IKFkFs(QeReOq=rc{pWs}oCnVYLo*znXqeGsT~ty)F7s|4>w4^%JpA
ztVxdd`7|mix%t!Rs7W_Bk69R7&6GAVxVv8m8iNwOJw8D3{n4?cM+)#itFTHdaiP=s
z#{PsY(k7Tw+wtJki|%**nvsH`r1mc7cRmet6k8NmyQx1}Ac7g~Nb#>rxDkA35j;T`
ziKa;cWY_7}O!`~>w-rSBKZSDufs-bEJJ;WHUv3GRH=vM;^!GThbd(A5x_{r26G#F7
zK|*|?5FPHh-M!D3P{caR-ODI<)SbqK?$)!ZeEhiYFz&VOAo*V^x-MLv6o@2P5ayeZ
z15n5zV1MGAUid<hCTP<WJ-m_lBZ&wZo~-=8XvgJVqp;-IrvFvgc}6w0Y;k--Z)(Jb
z0)ikQ1SKFvdNh}&gen%=1r>xqKzb7{-T;v%3eu#vs}Lz-1nD4((j?NP3N|1jMd?U+
zb5PfMZ`}_sU$WLYnKP5UXZFn5=l4JJ7$v1|vqan^1^sJbQd1{vYSF)Pb0&@;8kFX$
zr*(7l%fL6!A^K_dn$wujfJdcVu*>hahZ3;j{G{1by3vI_T2igrn=Zugr_@-`Fjy5}
zkXR+X_RK)<f3diC>eXV+k|wP2te>ZH{C1~~wUvGlu&id-OYVaYhET;7W_|p*(QYF$
zF-2(M5Ok<^9>bimp_=#YUciWsW&Eb|%xNF75VUogr)-XO!+))BW~>up;BDyedWF5a
zO3seO7uKnn`;d@G%F>8}A(2zjrA`U++I<z>u6DiGVc-s<VFO=nwO&h24PX#M6!vs+
z4V1Jb3fdN48dbrZhKO9`VChB&Z|TC_i|IY8Yi!DtqJYlePieI=HOtWk-B&8J|NbYu
za4E@dAD2hYV!F)lHO6}6aaF@{u7VGuF=AediKhR~;guEV1P%Ghoi}pxdD%45LHd(-
zh3yU3Vw(W}!2@c24U1i>SL{k#Ml*_M_MflmH-#^Rect5M)`H2q9)iv$%IQ10x@xPt
zPPZDOIkqWoGT~Iv@75-nAtz^xk+B?7(<bEq5cKmaBpqtqT)T{akWj$%Eit&gZWUzI
z1oY7o@JopnK{UmK88uz)#F-G<)saDOxAOh-tjHWTcz4gIHGwFOB8pDD=nglz+GI=V
zvFz2DbBi0s`Z?RNI#c_Gt-2GcHP036^OXl|WG;Q@h3TJ1vX@?W%Xr&0Y_nhhB=3Yb
z3`pk4$`yG1dM;n$T#QKP2@UGk`uH|Hh8&(D<e&VZ*CfI0ALmA+P>m=W8N(9X=JB|b
z>h$SzZGm`aUt?3|JPGfpMSe<Udfv=GaMHSk<*6=srS?m)pzk0i3`f>``@L~zkBnu5
z2@3S|#144x_{b36EQRb{*}jOKd>L>9C9DC5;XHfTK+9MdG3NN0aCvW=3EJ$kU!KB;
zBCg;dC}s%`5{a6H?}WxTNc;16%A?S{aJ9Yu9O??@t0M|jPDR#3diZiIm+dR5pLke?
zQKm7}C}e-++{RRjH8~%fc&4Km?PCV!C6LO5Ab93r>rIb>=n>#xh81iyhUa&Q*B|G8
z$LlBn=SKrc7tu0qE&vd`D#He{Z+7Uf4hI-V_#H8t&EmA*r#j$<uyye!&q*ZFPTyP!
zL2(Br9D-J?N8GB+uBliYz8|9a0upn9NyaUr3B>@wrN(LsRw#SP{>j-r&J6P%$2pD$
z0{~&2%sNPYtYtrqFzwOJb*mfjJOE_L$l)xLS2v^|Q@=TjM6X*BVm(eQ5Y*hxv*I-2
z+61Qwg&$G3S|v)K#>7^qQlcBu&YtSfK=F+t=BE6l-Jm@Gru_Cx$;0}`czyK1JTqbg
zLpbO#&)HDhyyaVtJ$(4kX)iqWN(E`(eYDIRg;9LZ9PV+ahx9SU+GG~E33er%d~ln&
zka*j|&69L5^FaafPULD9?;*-};*_cb+=s3%Xz+sh8{z3#a2o-n<}Ea~hl%akCiZz@
zxa@FT$<Aqe*^)<`zzV_qMnvLD(1t1vCvhe3q`@$9NRxZc7OLIQ=%g%1*P`~1<u%gz
z(_}=6iCpX|4_-mgEA;tzs*A-XJ}yr(;XZ1D3$eK6k@xU1LORT#iikL|p|DuQbsema
z+F(w&N5Q);;$i5W%Ghd4{~#@)v#i0=NL?AGX1S|<{7vW-?=6b(FfCmbFr$&n{h1^)
z$JG$<SbBA;`%+K}Zb%Fz9ElipQ_XpvI#~g`=Wv-k-9)a}Sg(Z<^c4N8oDe76069gc
zk%a*K*<sH@9qOj|sox=dx;4q+^H^tp-chw6q<&|JG1}m;y~c`Nrh%y!dr4;qd;C;A
z3Mq`dK%+gNlogh}A(A>PoGH9leXn8~+o&k~|2(@JhxIdRrHcz(8D)8OR6Y0Y_iOWi
z<fj6jBU`5<<Xg~JKTdKE8&;i-O`M}CJdHE&WH&<z<6+zRaL)ph6NPqFDyepo4X!Of
zN!fPja5nT+ZYygii@jB)g9%WZr~JY!-m8796$k=}Vreu=lF2d~$anvg^}0zZ4p|r@
zc$tk2=Yl=d>5V-p^BQ!|WnX6-N?{%S6RGIJPv&!})7on=)~>K4pC~0+JcD?V6KUoN
z79aZIJ0&<JQ$7}!`VpkWwJ>FVO__&M#V%jU>~6EG?>%^ROHT~g1PTO{>BIBH>q}JH
zV2ab+GAY8`9%)_Y4#Y-Bpnb)Pz{@JfFxw+ZT5oByjJRxY+Z(Cp5AQK^_uVGY<>g1o
z>Jw6{=K_s*f0jc@wNd;S#@6NSd||}?#J5W))q$=iVvPszhGePBTEUI;Yh?e_HcP<9
zh8#XZF!M`1S?OaEv8De+qQ;HLWvM9uSnhDv<Hja!E<o7SOpA<YpA;eJiM?k0Di>|&
zA%a1n0Me(JEn%vksAd6}uLEJII0;6zKhw3shbBN>7$Vz5f&*J6R><Ddaw&2g1rW1L
zK<QCTMtqObH$>UjiLjtI49vp6I69%3RTBVR5S9Uu%)MzdBpu!K-o5Me>4JMHTE@G9
z2}Ohf<$|%}c1|S>-%BSi9&I!tOpu#D{cv7%y>F=LJAd=#N5<r2;@!IrL<b(*M<U_|
zo~0#dd=1cLgQ-4n^N(jRw^hV3^2Yga+&;k*v;7}lY@kq_oSd#BwSkLYFA@mtg@ff&
z7anG6MfPfIe>b;yVSF3}LazWwn!(2kli{HxrbFTm-yfXatni=`iE_t}kGH2>B@zcV
z#`(fzliOVD))yVk&9A#mOHj{$#Ln!R=Z1a*A!xZbX0mULZ?#H``h##9<LgCH(VRF5
z!vZSRr|&tPKD!#t?|R_CV!EnC=_%B00ceGKRT$M<LuT5?dV8i+eP_!@f+j2XC+lnY
zuPv`Kv!95MuQ`SWoG_$<>Rl5q_f|g;yaKuJG_lKAudU5>EG#Tkk(3?inX*|~IypH3
zR>FW6DqDfCj<K~%Nt|0-+$oW0Q#rY{^>xm4KVKW0)85`47-pOb^fcJmIo2plBa0`e
zuFmI2MshcFRhC}JS5~em-~o~_OoL?#HP8Aow`SFYf=T=98g)fQ26#Mq)Y0q76HEa$
zz)J9EWy|bgS=fwe2?rBw+dzvH1`|2z*cBU_TKLfa_fLE~99)~)b43o;6;9e8D=Rb1
z&=B)|-M#F4-f7Z0k!=e~2k$Ot=3TKXSb^a3+ji~J3zY4GBJR)1dVBE_hHQR*;n^zS
zFan{?W<fkYe%hO!s^;l{C$xv}lBiT4yd@Nb?e0Cum^zyM1>ELCY)e_Q(ocuXX1Bc8
z8|_=Cl-GuRB5PNrpi?64eU4f7(LBa)^iXAac6q@Vt*}+X!Tr-t7yI+O`mB^Z%4kuR
zH7fNZfBh&U!R@!nb)OX#)_vY&J9><*Rd`V=va<RjSTkpAN-X7oNduG>n7~(xJszD}
z75z4@6mlp8%hMPar|Q&v43kn36nN3htiHsysHjLgEG%?oMY6eBNl5Kl`J)w5kCZy0
zJtpSa(D3_K_E!RZkmaYNey2bLl2xqA<(1^C=LTJy`}@N?($#N;OHv<TmE^vclv%6T
z{;eL@uhhdtSAqi%2y!pjTKZYiNmoQdbPvZ;($XktR!HDxX?Uy;a4)s}l1KE8x{Q6_
z(uu-23FxvU1y-GZUZ-_mLR6+S7$JeD)KFso6r2mT0xj<Tnm}})-4prvX`!o3STxdg
zR3H+^>PLSfP0>X(icC>StiThJHEy%HQ`J^g$f%EHKeHcZi*lVM$$y;*T$NXr)T51i
zdV4R<&Z4K3QD`M-?_Bb39dVE;sMZSKc?LM(<nrqar@|+D*a`(Q`Js3o9usDP2^07^
fgp9NQ^R>xNe%sO`rKD|+{8UfN;E$qHHh=#IM`BUv

literal 0
HcmV?d00001

diff --git a/graphsage/models.py b/graphsage/models.py
index 8abd46ed..62cf30ab 100644
--- a/graphsage/models.py
+++ b/graphsage/models.py
@@ -326,19 +326,19 @@ def sample(self, inputs, layer_infos, batch_size=None):
         support_size = 1
         support_sizes = [support_size]
 
-        for k in range(len(layer_infos)):  # k为跳数，也是层数， 实验中k = 0 1
-            t = len(layer_infos) - k - 1  # t = 1 0
+        for k in range(len(layer_infos)):  # k为跳数，实验中k = 0 1
+            t = len(layer_infos) - k - 1 # t = 1 0
 
             # 每一跳的邻居数目是前一跳的邻居节点数*该层的采样数，有个累乘的逻辑
             support_size *= layer_infos[t].num_samples
 
             sampler = layer_infos[t].neigh_sampler  # 采样器选择
 
-            # 采样器的两个输入，第一个入参是将要被采样的节点id，第二个入参是对这些节点，要采样多少个邻居
+            # 采样器的两个输入，第一个入参是将要被采样的节点id，第二个入参是采样多少个邻居
             node = sampler((samples[k], layer_infos[t].num_samples))
 
             # reshape成一维数组，再添加进samples中
-            samples.append(tf.reshape(node, [support_size * batch_size, ]))
+            samples.append(tf.reshape(node, [support_size * batch_size, ])) 
 
             # 同时记录好每一层的采样数
             support_sizes.append(support_size)
@@ -422,7 +422,7 @@ def aggregate(self, samples, input_features, dims, num_samples, support_sizes, b
             # aggregator2 的输入输出维度为：256，256，参数矩阵维度为256，128
 
             # hidden representation at current layer for all support nodes that are various hops away
-            # 该变量存放的是当前层，各节点利用邻居节点的信息更新后的中间表达，
+            # 该变量存放的是当前层，各节点利用邻居节点的信息更新后的中间表达
             next_hidden = []
 
             # as layer increases, the number of support nodes needed decreases
@@ -437,7 +437,7 @@ def aggregate(self, samples, input_features, dims, num_samples, support_sizes, b
                 # neigh_dims = [batch_size * 当前跳数的支持节点数，当前层的需要采样的邻居节点数，特征数]
                 #  
                 neigh_dims = [batch_size * support_sizes[hop],
-                              num_samples[len(num_samples) - hop - 1],
+                              num_samples[len(num_samples) - hop - 1],  # 这个维度，对应sample函数里的 t = len(layer_infos) - k - 1
                               dim_mult*dims[layer]]
                 h = aggregator((hidden[hop],
                                 tf.reshape(hidden[hop + 1], neigh_dims)))
diff --git a/graphsage/neigh_samplers.py b/graphsage/neigh_samplers.py
index 9c83d553..14292636 100644
--- a/graphsage/neigh_samplers.py
+++ b/graphsage/neigh_samplers.py
@@ -23,7 +23,13 @@ def __init__(self, adj_info, **kwargs):
 
     def _call(self, inputs):
         ids, num_samples = inputs
+
+        # 从全量的邻接表里根据ids获取各个节点的邻居节点
         adj_lists = tf.nn.embedding_lookup(self.adj_info, ids) 
+        
+        # 该步操作是 转置——乱序——转置， 目的是对列做打乱操作，如果直接打乱的话就是行操作
         adj_lists = tf.transpose(tf.random_shuffle(tf.transpose(adj_lists)))
+        
+        # 对列切片，只需要num_samples列的邻居，多的部分去掉
         adj_lists = tf.slice(adj_lists, [0,0], [-1, num_samples])
         return adj_lists

From 5d9cfdbb92aa7801f81472ee8fbda8fe06420545 Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Wed, 2 Mar 2022 17:15:53 +0800
Subject: [PATCH 09/28] =?UTF-8?q?=E5=AE=8C=E5=96=84=E6=B3=A8=E9=87=8A?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/models.py             | 22 +++++++++++++++++++---
 graphsage/unsupervised_train.py |  2 +-
 2 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/graphsage/models.py b/graphsage/models.py
index 62cf30ab..805ae12c 100644
--- a/graphsage/models.py
+++ b/graphsage/models.py
@@ -97,6 +97,8 @@ def load(self, sess=None):  # 从本地文件读取模型
         print("Model restored from file: %s" % save_path)
 
 
+
+# 多层感知机，是一个基础的深度模型
 class MLP(Model):
     """ A standard multi-layer perceptron """
 
@@ -449,12 +451,22 @@ def aggregate(self, samples, input_features, dims, num_samples, support_sizes, b
 
     def _build(self):
 
-        # 将第batch2视为标签，即batch1和batch2是一对正样本对
+        # 将batch2 reshape一下，用作下一步采样的输入
         labels = tf.reshape(
             tf.cast(self.placeholders['batch2'], dtype=tf.int64),
             [self.batch_size, 1])
+        """
+        tf.nn.fixed_unigram_candidate_sampler函数功能是从[0,range_max)中随机采样num_sampled个类
+        其中，返回的类是一个列表，每一个元素属于[0, range_max), 代表一个类别
+        每个类被采样的概率由参数unigrams决定，可以是表示概率的数组，也可以是表示count的数组(count大表示被采样的概率大)
+        range_max参数代表从[0,range_max)中采样，这里等于节点数，刚好是对应节点id
+
+        --------
+        在本实验中，就是利用这个函数，利用每个节点的度数形成概率分布，从节点集合中获取一批节点id，在后续视作负样本
+        true_classes个参数传入的是labels，但经测试，采样的结果和这个参数是无关的样子
+        返回的结果neg_samples里面是一个列表，每一个元素代表的是节点id
+        """
 
-        # 获取负样本， 按照给定的概率分布unigrams进行采样
         self.neg_samples, _, _ = (tf.nn.fixed_unigram_candidate_sampler(
             true_classes=labels,
             num_true=1,
@@ -471,7 +483,8 @@ def _build(self):
         samples1, support_sizes1 = self.sample(self.inputs1, self.layer_infos)
         samples2, support_sizes2 = self.sample(self.inputs2, self.layer_infos)
 
-        # 每层需要的采样数 实验中是[25,10]
+        # 每层需要的采样数 实验中是[25,10] 
+        
         num_samples = [
             layer_info.num_samples for layer_info in self.layer_infos]
 
@@ -505,6 +518,7 @@ def _build(self):
 
         # 对输出的样本执行L2规范化，dim=0或者1，1是表示按行做
         # x_l2[i] = x[i]/sqrt(sum(x^2))
+        # 对应论文 Algorithm 1的第7行
         self.outputs1 = tf.nn.l2_normalize(self.outputs1, 1)
         self.outputs2 = tf.nn.l2_normalize(self.outputs2, 1)
         self.neg_outputs = tf.nn.l2_normalize(self.neg_outputs, 1)
@@ -592,6 +606,8 @@ def _accuracy(self):
         tf.summary.scalar('mrr', self.mrr)
 
 
+
+#  
 class Node2VecModel(GeneralizedModel):
     def __init__(self, placeholders, dict_size, degrees, name=None,
                  nodevec_dim=50, lr=0.001, **kwargs):
diff --git a/graphsage/unsupervised_train.py b/graphsage/unsupervised_train.py
index 373a3f31..469ad95d 100644
--- a/graphsage/unsupervised_train.py
+++ b/graphsage/unsupervised_train.py
@@ -173,7 +173,7 @@ def train(train_data, test_data=None):
         features = np.vstack([features, np.zeros((features.shape[1],))])
 
 
-    # 根据开关判断是否加入随机游走的信息
+    # 根据开关判断是否使用随机游走的边，如果为真则使用随机游走的边代替图G里的边信息
     context_pairs = train_data[3] if FLAGS.random_context else None
 
     # 定义一些占位符

From 1f633e7bfdd6f8b721515d55102946b868cade38 Mon Sep 17 00:00:00 2001
From: levinxo <levinxo@gmail.com>
Date: Thu, 3 Mar 2022 11:49:13 +0800
Subject: [PATCH 10/28] utils.py annotation

---
 graphsage/utils.py | 25 ++++++++++++++++++++++++-
 1 file changed, 24 insertions(+), 1 deletion(-)

diff --git a/graphsage/utils.py b/graphsage/utils.py
index ff05072d..d80a9a49 100644
--- a/graphsage/utils.py
+++ b/graphsage/utils.py
@@ -16,7 +16,13 @@
 WALK_LEN=5
 N_WALKS=50
 
+
+"""
+加载数据集，并进行简单的预处理
+如下注释以toy-ppi数据为例
+"""
 def load_data(prefix, normalize=True, load_walks=False):
+    # DATA 1, 14755 nodes, 228431 links
     G_data = json.load(open(prefix + "-G.json"))
     G = json_graph.node_link_graph(G_data)
     if isinstance(G.nodes()[0], int):
@@ -24,14 +30,23 @@ def load_data(prefix, normalize=True, load_walks=False):
     else:
         conversion = lambda n : n
 
+    # DATA 2, numpy数组, (14755, 50) dtype(float64)
     if os.path.exists(prefix + "-feats.npy"):
         feats = np.load(prefix + "-feats.npy")
     else:
         print("No features present.. Only identity features will be used.")
         feats = None
+
+    # DATA 3, {"0": 0, "1": 1}, len: 14755
+    # node ids to int value indexing feature tensor，
+    # 用来做string id到int id的映射，其实没啥用
     id_map = json.load(open(prefix + "-id_map.json"))
     id_map = {conversion(k):int(v) for k,v in id_map.items()}
     walks = []
+
+    # DATA4, dict, len: 14755, column: 121
+    # from nodes ids to class value (int or list)
+    # 分类标签
     class_map = json.load(open(prefix + "-class_map.json"))
     if isinstance(list(class_map.values())[0], list):
         lab_conversion = lambda n : n
@@ -42,6 +57,7 @@ def load_data(prefix, normalize=True, load_walks=False):
 
     ## Remove all nodes that do not have val/test annotations
     ## (necessary because of networkx weirdness with the Reddit data)
+    # 移除损坏的节点：无val和test字段的，即验证和测试标识字段
     broken_count = 0
     for node in G.nodes():
         if not 'val' in G.node[node] or not 'test' in G.node[node]:
@@ -51,6 +67,9 @@ def load_data(prefix, normalize=True, load_walks=False):
 
     ## Make sure the graph has edge train_removed annotations
     ## (some datasets might already have this..)
+    # edge: (0, 800) 边，是个元组，表示源节点ID和目标节点ID
+    # G[0]：某节点与所有的关联节点组成的边的集合
+    # 下面这段代码的作用：标记需要在训练中移除的关联关系
     print("Loaded data.. now preprocessing..")
     for edge in G.edges():
         if (G.node[edge[0]]['val'] or G.node[edge[1]]['val'] or
@@ -59,10 +78,14 @@ def load_data(prefix, normalize=True, load_walks=False):
         else:
             G[edge[0]][edge[1]]['train_removed'] = False
 
-    if normalize and not feats is None:
+    if normalize and feats is not None:
         from sklearn.preprocessing import StandardScaler
+        # 训练集的id集合，only int, len: 9716
         train_ids = np.array([id_map[n] for n in G.nodes() if not G.node[n]['val'] and not G.node[n]['test']])
         train_feats = feats[train_ids]
+        # 特征缩放，标准化：z = (x - u) / s
+        # u is the mean of the training samples
+        # s is the standard deviation of the training samples
         scaler = StandardScaler()
         scaler.fit(train_feats)
         feats = scaler.transform(feats)

From c25041e750f3557a23a1a75d0b8eb03f19190f9c Mon Sep 17 00:00:00 2001
From: levinxo <levinxo@gmail.com>
Date: Thu, 3 Mar 2022 16:33:02 +0800
Subject: [PATCH 11/28] update minibatch.py annotation

---
 graphsage/minibatch.py        | 9 +++++++++
 graphsage/supervised_train.py | 2 +-
 2 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/graphsage/minibatch.py b/graphsage/minibatch.py
index 0af16c82..e28a6dc3 100644
--- a/graphsage/minibatch.py
+++ b/graphsage/minibatch.py
@@ -224,15 +224,23 @@ def _make_label_vec(self, node):
         return label_vec
 
     def construct_adj(self):
+        # 一个numpy 2dim的数组，用于存储各个节点的邻接点，最多为max_degree个邻接点
+        # adj shape: (14756, 128)
         adj = len(self.id2idx)*np.ones((len(self.id2idx)+1, self.max_degree))
+        # (14755,)   用于存储所有节点的degree值
         deg = np.zeros((len(self.id2idx),))
 
         for nodeid in self.G.nodes():
+            # 测试集合验证集的节点直接跳过
             if self.G.node[nodeid]['test'] or self.G.node[nodeid]['val']:
                 continue
+
+            # 获取所有训练集中节点邻居节点的id
             neighbors = np.array([self.id2idx[neighbor] 
                 for neighbor in self.G.neighbors(nodeid)
                 if (not self.G[nodeid][neighbor]['train_removed'])])
+
+            # 不足degree的邻接点补足degree，超过的随机选择degree个邻接点
             deg[self.id2idx[nodeid]] = len(neighbors)
             if len(neighbors) == 0:
                 continue
@@ -246,6 +254,7 @@ def construct_adj(self):
     def construct_test_adj(self):
         adj = len(self.id2idx)*np.ones((len(self.id2idx)+1, self.max_degree))
         for nodeid in self.G.nodes():
+            # 所有邻接点的id，这里没有限制训练或测试集
             neighbors = np.array([self.id2idx[neighbor] 
                 for neighbor in self.G.neighbors(nodeid)])
             if len(neighbors) == 0:
diff --git a/graphsage/supervised_train.py b/graphsage/supervised_train.py
index c5bff002..7da9d4e7 100644
--- a/graphsage/supervised_train.py
+++ b/graphsage/supervised_train.py
@@ -131,7 +131,7 @@ def train(train_data, test_data=None):
         num_classes = len(set(class_map.values()))
 
     if not features is None:
-        # pad with dummy zero vector
+        # pad with dummy zero vector, row wise
         features = np.vstack([features, np.zeros((features.shape[1],))])
 
     context_pairs = train_data[3] if FLAGS.random_context else None

From 9a59693d1482be5d76b77c83990f85a4b1f75f2b Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Fri, 4 Mar 2022 16:11:40 +0800
Subject: [PATCH 12/28] update

---
 .gitignore              | 216 ++++++++++++-------------
 graphsage/models.py     |   2 +-
 graphsage/prediction.py | 348 ++++++++++++++++++++--------------------
 3 files changed, 283 insertions(+), 283 deletions(-)

diff --git a/.gitignore b/.gitignore
index d3ecac84..3c79a046 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,108 +1,108 @@
-# Custom
-*.idea
-*.pdf
-tmp/
-*.txt
-*swp*
-*.sw?
-gcn_back
-.DS_STORE
-*.aux
-*.log
-*.out
-*.bbl
-*.synctex.gz
-
-# Byte-compiled / optimized / DLL files
-__pycache__/
-*.py[cod]
-*$py.class
-
-# C extensions
-*.so
-
-# Distribution / packaging
-.Python
-env/
-build/
-develop-eggs/
-dist/
-downloads/
-eggs/
-.eggs/
-lib/
-lib64/
-parts/
-sdist/
-var/
-*.egg-info/
-.installed.cfg
-*.egg
-
-# PyInstaller
-#  Usually these files are written by a python script from a template
-#  before PyInstaller builds the exe, so as to inject date/other infos into it.
-*.manifest
-*.spec
-
-# Installer logs
-pip-log.txt
-pip-delete-this-directory.txt
-
-# Unit test / coverage reports
-htmlcov/
-.tox/
-.coverage
-.coverage.*
-.cache
-nosetests.xml
-coverage.xml
-*,cover
-.hypothesis/
-
-# Translations
-*.mo
-*.pot
-
-# Django stuff:
-*.log
-local_settings.py
-
-# Flask stuff:
-instance/
-.webassets-cache
-
-# Scrapy stuff:
-.scrapy
-
-# Sphinx documentation
-docs/_build/
-
-# PyBuilder
-target/
-
-# IPython Notebook
-.ipynb_checkpoints
-
-# pyenv
-.python-version
-
-# celery beat schedule file
-celerybeat-schedule
-
-# dotenv
-.env
-
-# virtualenv
-venv/
-ENV/
-
-# Spyder project settings
-.spyderproject
-
-# Rope project settings
-.ropeproject
-
-*.pickle
-*.pkl
-
+# Custom
+*.idea
+*.pdf
+tmp/
+*.txt
+*swp*
+*.sw?
+gcn_back
+.DS_STORE
+*.aux
+*.log
+*.out
+*.bbl
+*.synctex.gz
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+env/
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+*.egg-info/
+.installed.cfg
+*.egg
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*,cover
+.hypothesis/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# IPython Notebook
+.ipynb_checkpoints
+
+# pyenv
+.python-version
+
+# celery beat schedule file
+celerybeat-schedule
+
+# dotenv
+.env
+
+# virtualenv
+venv/
+ENV/
+
+# Spyder project settings
+.spyderproject
+
+# Rope project settings
+.ropeproject
+
+*.pickle
+*.pkl
+
diff --git a/graphsage/models.py b/graphsage/models.py
index 805ae12c..73b547a3 100644
--- a/graphsage/models.py
+++ b/graphsage/models.py
@@ -557,7 +557,7 @@ def _loss(self):
             for var in aggregator.vars.values():
                 self.loss += FLAGS.weight_decay * tf.nn.l2_loss(var)
 
-        # 根据之前生成的预测层，计算loss，该loss有三个选项：_xent_loss、_skipgram_loss、_hinge_loss
+        # 根据之前生成的预测层，计算loss，该loss有三个选项：_xent_loss、_skipgram_loss、_hinge_loss，论文中使用的是第一个
         self.loss += self.link_pred_layer.loss(
             self.outputs1, self.outputs2, self.neg_outputs)
         tf.summary.scalar('loss', self.loss)
diff --git a/graphsage/prediction.py b/graphsage/prediction.py
index 8223bc4e..3e6f4c22 100644
--- a/graphsage/prediction.py
+++ b/graphsage/prediction.py
@@ -1,174 +1,174 @@
-from __future__ import division
-from __future__ import print_function
-
-from graphsage.inits import zeros
-from graphsage.layers import Layer
-import tensorflow as tf
-
-flags = tf.app.flags
-FLAGS = flags.FLAGS
-
-
-class BipartiteEdgePredLayer(Layer):
-    def __init__(self, input_dim1, input_dim2, placeholders, dropout=False, act=tf.nn.sigmoid,
-                 loss_fn='xent', neg_sample_weights=1.0,
-                 bias=False, bilinear_weights=False, **kwargs):
-        """
-        Basic class that applies skip-gram-like loss
-        (i.e., dot product of node+target and node and negative samples)
-        Args:
-            bilinear_weights: use a bilinear weight for affinity calculation: u^T A v. If set to
-                false, it is assumed that input dimensions are the same and the affinity will be 
-                based on dot product.
-
-        一个基础类，使用了"skip-gram" 类型的损失函数(节点和目标的点乘以及节点和负样本的点乘)
-
-        """
-
-        super(BipartiteEdgePredLayer, self).__init__(**kwargs)
-        self.input_dim1 = input_dim1
-        self.input_dim2 = input_dim2
-        self.act = act
-        self.bias = bias
-        self.eps = 1e-7
-
-        # Margin for hinge loss
-        self.margin = 0.1
-        self.neg_sample_weights = neg_sample_weights
-
-        self.bilinear_weights = bilinear_weights
-
-        if dropout:
-            self.dropout = placeholders['dropout']
-        else:
-            self.dropout = 0.
-
-        # output a likelihood term
-        self.output_dim = 1
-        with tf.variable_scope(self.name + '_vars'):
-            # bilinear form
-            if bilinear_weights:
-                # self.vars['weights'] = glorot([input_dim1, input_dim2],
-                #                              name='pred_weights')
-                self.vars['weights'] = tf.get_variable(
-                    'pred_weights',
-                    shape=(input_dim1, input_dim2),
-                    dtype=tf.float32,
-                    initializer=tf.contrib.layers.xavier_initializer())
-
-            if self.bias:
-                self.vars['bias'] = zeros([self.output_dim], name='bias')
-
-        if loss_fn == 'xent':
-            self.loss_fn = self._xent_loss
-        elif loss_fn == 'skipgram':
-            self.loss_fn = self._skipgram_loss
-        elif loss_fn == 'hinge':
-            self.loss_fn = self._hinge_loss
-
-        if self.logging:
-            self._log_vars()
-
-    def affinity(self, inputs1, inputs2):
-        """ Affinity score between batch of inputs1 and inputs2.
-        Args:
-            inputs1: tensor of shape [batch_size x feature_size].
-
-        计算正样本对之间的"亲和度"：
-        ①特征矩阵点乘(没有bilinear_weights的情况下)
-        ②求均值
-
-        返回的是样本和其对应的正样本之间的亲和度，尺寸：[batch_size，1]
-        """
-        # shape: [batch_size, input_dim1]
-        if self.bilinear_weights:
-            prod = tf.matmul(inputs2, tf.transpose(self.vars['weights']))
-            self.prod = prod
-            result = tf.reduce_sum(inputs1 * prod, axis=1)
-        else:
-            result = tf.reduce_sum(inputs1 * inputs2, axis=1)
-        return result
-
-    def neg_cost(self, inputs1, neg_samples, hard_neg_samples=None):
-        """ For each input in batch, compute the sum of its affinity to negative samples.
-
-        Returns:
-            Tensor of shape [batch_size x num_neg_samples]. For each node, a list of affinities to
-                negative samples is computed.
-        计算输入样本和每一个负样本之间的"亲和度"：
-        ①inputs_features × neg_features.T
-        
-        返回的是样本和每一个负样本之间的"亲和度"，尺寸是[batch_size, num_neg_samples]
-
-        """
-        if self.bilinear_weights:
-            inputs1 = tf.matmul(inputs1, self.vars['weights'])
-        neg_aff = tf.matmul(inputs1, tf.transpose(neg_samples))
-
-        return neg_aff
-
-    def loss(self, inputs1, inputs2, neg_samples):
-        """ negative sampling loss.
-        Args:
-            neg_samples: tensor of shape [num_neg_samples x input_dim2]. Negative samples for all
-            inputs in batch inputs1.
-
-        """
-        return self.loss_fn(inputs1, inputs2, neg_samples)
-
-    def _xent_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
-        """
-        计算正样本的交叉熵损失，正样本label赋值全1, 负样本label赋值全0
-        公式 : y * -log(sigmoid(x)) + (1 - y) * -log(1 - sigmoid(x))
-        正样本y=1，负样本y=0，分别可以省略一项
-
-        ①计算正样本对的亲和度
-        ②计算样本和负样本的亲和度
-        ③将label全部设为1，计算正样本对产生的loss
-        ④将label全部设为0，计算所有负样本产生的loss
-        ⑤将两个loss平均一下
-
-        对应论文的公式(1)
-                
-        """
-        # 计算正样本对的亲和度
-        aff = self.affinity(inputs1, inputs2)
-
-        # 计算顶点和各个负样本的亲和度
-        neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
-
-
-        """
-
-        """
-        true_xent = tf.nn.sigmoid_cross_entropy_with_logits(
-            labels=tf.ones_like(aff), logits=aff)
-
-        # 计算负样本的交叉熵损失
-        negative_xent = tf.nn.sigmoid_cross_entropy_with_logits(
-            labels=tf.zeros_like(neg_aff), logits=neg_aff)
-
-
-        # neg_sample_weights 默认为1.0
-        loss = tf.reduce_sum(
-            true_xent) + self.neg_sample_weights * tf.reduce_sum(negative_xent)
-        return loss
-
-    def _skipgram_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
-        aff = self.affinity(inputs1, inputs2)
-        neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
-        neg_cost = tf.log(tf.reduce_sum(tf.exp(neg_aff), axis=1))
-        loss = tf.reduce_sum(aff - neg_cost)
-        return loss
-
-    def _hinge_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
-        aff = self.affinity(inputs1, inputs2)
-        neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
-        diff = tf.nn.relu(tf.subtract(
-            neg_aff, tf.expand_dims(aff, 1) - self.margin), name='diff')
-        loss = tf.reduce_sum(diff)
-        self.neg_shape = tf.shape(neg_aff)
-        return loss
-
-    def weights_norm(self):
-        return tf.nn.l2_norm(self.vars['weights'])
+from __future__ import division
+from __future__ import print_function
+
+from graphsage.inits import zeros
+from graphsage.layers import Layer
+import tensorflow as tf
+
+flags = tf.app.flags
+FLAGS = flags.FLAGS
+
+
+class BipartiteEdgePredLayer(Layer):
+    def __init__(self, input_dim1, input_dim2, placeholders, dropout=False, act=tf.nn.sigmoid,
+                 loss_fn='xent', neg_sample_weights=1.0,
+                 bias=False, bilinear_weights=False, **kwargs):
+        """
+        Basic class that applies skip-gram-like loss
+        (i.e., dot product of node+target and node and negative samples)
+        Args:
+            bilinear_weights: use a bilinear weight for affinity calculation: u^T A v. If set to
+                false, it is assumed that input dimensions are the same and the affinity will be 
+                based on dot product.
+
+        一个基础类，使用了"skip-gram" 类型的损失函数(节点和目标的点乘以及节点和负样本的点乘)
+
+        """
+
+        super(BipartiteEdgePredLayer, self).__init__(**kwargs)
+        self.input_dim1 = input_dim1
+        self.input_dim2 = input_dim2
+        self.act = act
+        self.bias = bias
+        self.eps = 1e-7
+
+        # Margin for hinge loss
+        self.margin = 0.1
+        self.neg_sample_weights = neg_sample_weights
+
+        self.bilinear_weights = bilinear_weights
+
+        if dropout:
+            self.dropout = placeholders['dropout']
+        else:
+            self.dropout = 0.
+
+        # output a likelihood term
+        self.output_dim = 1
+        with tf.variable_scope(self.name + '_vars'):
+            # bilinear form
+            if bilinear_weights:
+                # self.vars['weights'] = glorot([input_dim1, input_dim2],
+                #                              name='pred_weights')
+                self.vars['weights'] = tf.get_variable(
+                    'pred_weights',
+                    shape=(input_dim1, input_dim2),
+                    dtype=tf.float32,
+                    initializer=tf.contrib.layers.xavier_initializer())
+
+            if self.bias:
+                self.vars['bias'] = zeros([self.output_dim], name='bias')
+
+        if loss_fn == 'xent':
+            self.loss_fn = self._xent_loss
+        elif loss_fn == 'skipgram':
+            self.loss_fn = self._skipgram_loss
+        elif loss_fn == 'hinge':
+            self.loss_fn = self._hinge_loss
+
+        if self.logging:
+            self._log_vars()
+
+    def affinity(self, inputs1, inputs2):
+        """ Affinity score between batch of inputs1 and inputs2.
+        Args:
+            inputs1: tensor of shape [batch_size x feature_size].
+
+        计算正样本对之间的"亲和度"：
+        ①特征矩阵点乘(没有bilinear_weights的情况下)
+        ②求均值
+
+        返回的是样本和其对应的正样本之间的亲和度，尺寸：[batch_size，1]
+        """
+        # shape: [batch_size, input_dim1]
+        if self.bilinear_weights:
+            prod = tf.matmul(inputs2, tf.transpose(self.vars['weights']))
+            self.prod = prod
+            result = tf.reduce_sum(inputs1 * prod, axis=1)
+        else:
+            result = tf.reduce_sum(inputs1 * inputs2, axis=1)
+        return result
+
+    def neg_cost(self, inputs1, neg_samples, hard_neg_samples=None):
+        """ For each input in batch, compute the sum of its affinity to negative samples.
+
+        Returns:
+            Tensor of shape [batch_size x num_neg_samples]. For each node, a list of affinities to
+                negative samples is computed.
+        计算输入样本和每一个负样本之间的"亲和度"：
+        ①inputs_features × neg_features.T
+        
+        返回的是样本和每一个负样本之间的"亲和度"，尺寸是[batch_size, num_neg_samples]
+
+        """
+        if self.bilinear_weights:
+            inputs1 = tf.matmul(inputs1, self.vars['weights'])
+        neg_aff = tf.matmul(inputs1, tf.transpose(neg_samples))
+
+        return neg_aff
+
+    def loss(self, inputs1, inputs2, neg_samples):
+        """ negative sampling loss.
+        Args:
+            neg_samples: tensor of shape [num_neg_samples x input_dim2]. Negative samples for all
+            inputs in batch inputs1.
+
+        """
+        return self.loss_fn(inputs1, inputs2, neg_samples)
+
+    def _xent_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
+        """
+        计算正样本的交叉熵损失，正样本label赋值全1, 负样本label赋值全0
+        公式 : y * -log(sigmoid(x)) + (1 - y) * -log(1 - sigmoid(x))
+        正样本y=1，负样本y=0，分别可以省略一项
+
+        ①计算正样本对的亲和度
+        ②计算样本和负样本的亲和度
+        ③将label全部设为1，计算正样本对产生的loss
+        ④将label全部设为0，计算所有负样本产生的loss
+        ⑤将两个loss平均一下
+
+        对应论文的公式(1)
+                
+        """
+        # 计算正样本对的亲和度
+        aff = self.affinity(inputs1, inputs2)
+
+        # 计算顶点和各个负样本的亲和度
+        neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
+
+
+        """
+
+        """
+        true_xent = tf.nn.sigmoid_cross_entropy_with_logits(
+            labels=tf.ones_like(aff), logits=aff)
+
+        # 计算负样本的交叉熵损失
+        negative_xent = tf.nn.sigmoid_cross_entropy_with_logits(
+            labels=tf.zeros_like(neg_aff), logits=neg_aff)
+
+
+        # neg_sample_weights 默认为1.0
+        loss = tf.reduce_sum(
+            true_xent) + self.neg_sample_weights * tf.reduce_sum(negative_xent)
+        return loss
+
+    def _skipgram_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
+        aff = self.affinity(inputs1, inputs2)
+        neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
+        neg_cost = tf.log(tf.reduce_sum(tf.exp(neg_aff), axis=1))
+        loss = tf.reduce_sum(aff - neg_cost)
+        return loss
+
+    def _hinge_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
+        aff = self.affinity(inputs1, inputs2)
+        neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
+        diff = tf.nn.relu(tf.subtract(
+            neg_aff, tf.expand_dims(aff, 1) - self.margin), name='diff')
+        loss = tf.reduce_sum(diff)
+        self.neg_shape = tf.shape(neg_aff)
+        return loss
+
+    def weights_norm(self):
+        return tf.nn.l2_norm(self.vars['weights'])

From 445e6b061ec166a36f47d5ce8942483eef9bbb47 Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Tue, 8 Mar 2022 14:12:54 +0800
Subject: [PATCH 13/28] =?UTF-8?q?=E6=96=B0=E5=A2=9E=E9=98=BF=E9=87=8C?=
 =?UTF-8?q?=E4=BA=91=E6=9C=BA=E5=99=A8=E4=BF=A1=E6=81=AF?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 ...72\345\231\250\344\277\241\346\201\257.md" | 55 +++++++++++++++++++
 1 file changed, 55 insertions(+)
 create mode 100644 "aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"

diff --git "a/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md" "b/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
new file mode 100644
index 00000000..5d8b80c2
--- /dev/null
+++ "b/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
@@ -0,0 +1,55 @@
+## 阿里云机器信息
+
+IP：敏感信息不放在网上
+
+mag240原数据目录：`/mnt/ogb-dataset/mag240m/data/raw`
+
+```
+├── RELEASE_v1.txt
+├── mapping    //空文件夹
+├── meta.pt
+├── processed
+│   ├── author___affiliated_with___institution
+│   │   └── edge_index.npy     //作者和机构的边，shape=[2,num_edges]
+│   ├── author___writes___paper
+│   │   └── edge_index.npy     //作者和论文的边，shape=[2,num_edges]
+│   ├── paper
+│   │   ├── node_feat.npy    //论文节点的特征，shape=[num_node,768]
+│   │   ├── node_label.npy   // 论文的标签
+│   │   └── node_year.npy   // 论文年份
+│   └── paper___cites___paper
+│       └── edge_index.npy  // 论文和城市的边shape=[2,num_edges]
+├── raw //空文件夹
+└── split_dict.pt    //切分训练集、验证集、测试集方式的文件，用torch读取是一个字典，keys=[‘train’,’valid’,’test’], value是node_index
+
+```
+
+
+
+### docker镜像
+
+#### opeceipeno/dgl:v1.4
+
+ogb代码的运行环境，想法是通过虚拟环境去激活各个方案的运行环境，当前做好了Google的mag240m运行环境
+
+[GitHub地址](https://github.com/deepmind/deepmind-research/tree/master/ogb_lsc/mag)
+
+```
+docker run --gpus all -it -v /mnt:/mnt opeceipeno/dgl:v1.4 bash
+# 启动容器后，激活Google代码的运行环境
+source /py3_venv/google_ogb_mag240m/bin/activate
+# /workspace 目录有代码
+```
+
+Google方案预处理后的数据目录：`/mnt/ogb-dataset/mag240m/data/preprocessed`，相当于执行完了`run_preprocessing.sh`脚本，下一步是可以复现实验，
+
+
+
+#### opeceipeno/graphsage:gpu
+
+graphSAGE的环境，[GitHub地址](https://github.com/qksidmx/GraphSAGE)
+
+```
+docker run --gpus all -it opeceipeno/graphsage:gpu bash
+#/notebook目录下面有代码，运行实验参考readme文档
+```
\ No newline at end of file

From 9a005b38cb5b551838d4274148aa5cfaf77c7fba Mon Sep 17 00:00:00 2001
From: 13929520142 <wangqy94@163.com>
Date: Tue, 8 Mar 2022 17:11:46 +0800
Subject: [PATCH 14/28] =?UTF-8?q?=E8=81=9A=E5=90=88=E5=87=BD=E6=95=B0?=
 =?UTF-8?q?=E6=B3=A8=E9=87=8A?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/aggregators.py | 30 +++++++++++++++++++++++++-----
 1 file changed, 25 insertions(+), 5 deletions(-)

diff --git a/graphsage/aggregators.py b/graphsage/aggregators.py
index 4dcff553..9a62e61e 100644
--- a/graphsage/aggregators.py
+++ b/graphsage/aggregators.py
@@ -27,7 +27,7 @@ def __init__(self, input_dim, output_dim, neigh_input_dim=None,
             name = '/' + name
         else:
             name = ''
-
+        # 权重矩阵设置对应伪代码中的W，这里自身节点和输入节点采用了不同的W，推测是为了防止过拟合
         with tf.variable_scope(self.name + name + '_vars'):
             self.vars['neigh_weights'] = glorot([neigh_input_dim, output_dim],
                                                 name='neigh_weights')
@@ -42,14 +42,16 @@ def __init__(self, input_dim, output_dim, neigh_input_dim=None,
         self.input_dim = input_dim
         self.output_dim = output_dim
 
+    # 输入维度[batchSize, numNeigh, numNeighDim]，依次为batch大小，每一跳节点数量，节点特征数
     def _call(self, inputs):
         self_vecs, neigh_vecs = inputs
 
         neigh_vecs = tf.nn.dropout(neigh_vecs, 1 - self.dropout)
         self_vecs = tf.nn.dropout(self_vecs, 1 - self.dropout)
+        # 均值聚合后neigh_mean shape变为[batchSize,numNeighDim],原来每个batchSize的向量均值聚合为1个
         neigh_means = tf.reduce_mean(neigh_vecs, axis=1)
 
-        # [nodes] x [out_dim]
+        # [nodes] x W，相乘后shape变为[batchSize, outputDim]
         from_neighs = tf.matmul(neigh_means, self.vars['neigh_weights'])
 
         from_self = tf.matmul(self_vecs, self.vars["self_weights"])
@@ -106,10 +108,13 @@ def _call(self, inputs):
 
         neigh_vecs = tf.nn.dropout(neigh_vecs, 1 - self.dropout)
         self_vecs = tf.nn.dropout(self_vecs, 1 - self.dropout)
+        # 这里做了两个操作，首先取并集，首先在原来shape为[batchSize,numSelfDim]增加一个维度变为[batchSize,1,numSelfDim]
+        # 然后在2维上连接邻居向量原来的shape变为[batchSize,numNeigh+numSelf,numNeighDim]
+        # 最后在2维上均值聚合后，shape变为[batchSize,numNeighDim]
         means = tf.reduce_mean(tf.concat([neigh_vecs,
                                           tf.expand_dims(self_vecs, axis=1)], axis=1), axis=1)
 
-        # [nodes] x [out_dim]
+        # 因为二者已经合为并集，故只需要一个权重矩阵即可，[nodes] x W后，shape变为[batchSize, outputDim]
         output = tf.matmul(means, self.vars['weights'])
 
         # bias
@@ -177,10 +182,14 @@ def _call(self, inputs):
         batch_size = dims[0]
         num_neighbors = dims[1]
         # [nodes * sampled neighbors] x [hidden_dim]
+        # 将邻居矩阵由3维降为2维，此时shape为[batch_size * numNeighbors,neighDim]
         h_reshaped = tf.reshape(neigh_h, (batch_size * num_neighbors, self.neigh_input_dim))
 
+        # 将降为2维后的邻居矩阵与池化层相乘，池化层shape为[neigh_input_dim,hidden_dim]
+        # 相乘后矩阵shape变为[batch_size * num_neighbors,hidden_dim]
         for l in self.mlp_layers:
             h_reshaped = l(h_reshaped)
+        # 将经过池化层的矩阵还原为[batch_size, num_neighbors, hidden_dim],然后进行最大聚合降维为[batch_size, hidden_dim]
         neigh_h = tf.reshape(h_reshaped, (batch_size, num_neighbors, self.hidden_dim))
         neigh_h = tf.reduce_max(neigh_h, axis=1)
 
@@ -198,7 +207,7 @@ def _call(self, inputs):
 
         return self.act(output)
 
-
+# 与最大池化基本相同，只是采用取均值的方式聚合向量集合
 class MeanPoolingAggregator(Layer):
     """ Aggregates via mean-pooling over MLP functions.
     """
@@ -278,7 +287,7 @@ def _call(self, inputs):
 
         return self.act(output)
 
-
+# 与最大池化基本相同，经过两次mlp layer，最后shape为[batch_size,hidden_dim_2]
 class TwoMaxLayerPoolingAggregator(Layer):
     """ Aggregates via pooling over two MLP functions.
     """
@@ -416,11 +425,17 @@ def _call(self, inputs):
         dims = tf.shape(neigh_vecs)
         batch_size = dims[0]
         initial_state = self.cell.zero_state(batch_size, tf.float32)
+        # 将neigh_vecs 将向量每个维度值取为正数，然后进行最大值降维，降维后shape为[batch_size, num_neighbors]
+        # 再通过tf.sign转化为0,1的矩阵
         used = tf.sign(tf.reduce_max(tf.abs(neigh_vecs), axis=2))
+        # 在2维上将所有数进行求和降维及获得1维向量[batch_size],这样就获得了每个batch_size的序列值
         length = tf.reduce_sum(used, axis=1)
+        # 为了防止某个batch_size序列值为0的情况，将其强行转为1，作为lstm的输入序列步长
         length = tf.maximum(length, tf.constant(1.))
         length = tf.cast(length, tf.int32)
 
+        # 进行lstm聚合rnn_outputs为结果值，rnn_states为最后一个单元的状态，这里用不上
+        # 聚合后rnn_outputs shape为[batch_size,max_len,hidden_dim]
         with tf.variable_scope(self.name) as scope:
             try:
                 rnn_outputs, rnn_states = tf.nn.dynamic_rnn(
@@ -436,8 +451,13 @@ def _call(self, inputs):
         batch_size = tf.shape(rnn_outputs)[0]
         max_len = tf.shape(rnn_outputs)[1]
         out_size = int(rnn_outputs.get_shape()[2])
+        # 生成索引，生成规则为如下：1.先生成shape为[batch_size]的1维向量，具体为[1,2,3...batch_size-2,batch_size-1]
+        # 2.每个元素乘以max_len
+        # 3.将原来的lstm序列步长减1后再相加（数组从0开始）
         index = tf.range(0, batch_size) * max_len + (length - 1)
+        # 将rnn_outputs shape变为[-1,hidden_dim]，-1代表自适应降维，应该是batch_size*max_len
         flat = tf.reshape(rnn_outputs, [-1, out_size])
+        # 根据索引将对应元素从flat取出来shape变为[index.length,hidden_dim]
         neigh_h = tf.gather(flat, index)
 
         from_neighs = tf.matmul(neigh_h, self.vars['neigh_weights'])

From 424694e9b8f0b041f1e76ac98c08c16a5f4f0ed8 Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Tue, 8 Mar 2022 17:50:34 +0800
Subject: [PATCH 15/28] =?UTF-8?q?=E6=9B=B4=E6=94=B9=E4=B8=80=E5=A4=84?=
 =?UTF-8?q?=E6=8F=8F=E8=BF=B0=E9=94=99=E8=AF=AF?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 "aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md" | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git "a/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md" "b/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
index 5d8b80c2..2a330537 100644
--- "a/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
+++ "b/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
@@ -18,7 +18,7 @@ mag240原数据目录：`/mnt/ogb-dataset/mag240m/data/raw`
 │   │   ├── node_label.npy   // 论文的标签
 │   │   └── node_year.npy   // 论文年份
 │   └── paper___cites___paper
-│       └── edge_index.npy  // 论文和城市的边shape=[2,num_edges]
+│       └── edge_index.npy  // 论文引用关系的边shape=[2,num_edges]
 ├── raw //空文件夹
 └── split_dict.pt    //切分训练集、验证集、测试集方式的文件，用torch读取是一个字典，keys=[‘train’,’valid’,’test’], value是node_index
 
@@ -52,4 +52,4 @@ graphSAGE的环境，[GitHub地址](https://github.com/qksidmx/GraphSAGE)
 ```
 docker run --gpus all -it opeceipeno/graphsage:gpu bash
 #/notebook目录下面有代码，运行实验参考readme文档
-```
\ No newline at end of file
+```

From 9fe0150ab4a56440c2b44ab9911b0088b55a303f Mon Sep 17 00:00:00 2001
From: levinxo <levinxo@gmail.com>
Date: Tue, 8 Mar 2022 17:51:58 +0800
Subject: [PATCH 16/28] add preprocess notebook for google colab

---
 graphsage/layers.py           |   8 +-
 graphsage/minibatch.py        |   5 +-
 graphsage/supervised_train.py |   2 +
 preprocess.ipynb              | 446 ++++++++++++++++++++++++++++++++++
 4 files changed, 456 insertions(+), 5 deletions(-)
 create mode 100644 preprocess.ipynb

diff --git a/graphsage/layers.py b/graphsage/layers.py
index 57ec291c..d4bcb3c9 100644
--- a/graphsage/layers.py
+++ b/graphsage/layers.py
@@ -40,10 +40,10 @@ class Layer(object):
         __call__(inputs): Wrapper for _call()
         _log_vars(): Log all variables
 
-        最基础的层类型
-        name：定义层的名称，字符型
-        logging：布尔型，如果开的话就可以打印训练过程中，当需要查看一个张量在训练过程中值的分布情况时，可通过tf.summary.histogram()将其分布情况以直方图的形式在TensorBoard直方图仪表板上显示．
-        这里并没有参数矩阵、激活函数等值，都是在其子类中实现
+    最基础的层类型
+    name：定义层的名称，字符型
+    logging：布尔型，如果开的话就可以打印训练过程中，当需要查看一个张量在训练过程中值的分布情况时，可通过tf.summary.histogram()将其分布情况以直方图的形式在TensorBoard直方图仪表板上显示．
+    这里并没有参数矩阵、激活函数等值，都是在其子类中实现
     """
 
     def __init__(self, **kwargs):
diff --git a/graphsage/minibatch.py b/graphsage/minibatch.py
index e28a6dc3..0dcb9a2d 100644
--- a/graphsage/minibatch.py
+++ b/graphsage/minibatch.py
@@ -186,6 +186,9 @@ class NodeMinibatchIterator(object):
     num_classes -- number of output classes
     batch_size -- size of the minibatches
     max_degree -- maximum size of the downsampled adjacency lists
+    以toy-ppi数据集举例：
+    label_map为输出，维度为(14755, 121)
+    num_class为label_map的第二维，即121
     """
     def __init__(self, G, id2idx, 
             placeholders, label_map, num_classes, 
@@ -225,7 +228,7 @@ def _make_label_vec(self, node):
 
     def construct_adj(self):
         # 一个numpy 2dim的数组，用于存储各个节点的邻接点，最多为max_degree个邻接点
-        # adj shape: (14756, 128)
+        # adjacency shape: (14756, 128)
         adj = len(self.id2idx)*np.ones((len(self.id2idx)+1, self.max_degree))
         # (14755,)   用于存储所有节点的degree值
         deg = np.zeros((len(self.id2idx),))
diff --git a/graphsage/supervised_train.py b/graphsage/supervised_train.py
index 7da9d4e7..98349e07 100644
--- a/graphsage/supervised_train.py
+++ b/graphsage/supervised_train.py
@@ -136,6 +136,7 @@ def train(train_data, test_data=None):
 
     context_pairs = train_data[3] if FLAGS.random_context else None
     placeholders = construct_placeholders(num_classes)
+    #实例化 NodeMinibatch 迭代器
     minibatch = NodeMinibatchIterator(G, 
             id_map,
             placeholders, 
@@ -144,6 +145,7 @@ def train(train_data, test_data=None):
             batch_size=FLAGS.batch_size,
             max_degree=FLAGS.max_degree, 
             context_pairs = context_pairs)
+    # adjacency shape: (14756, 128)  包装为placeholder
     adj_info_ph = tf.placeholder(tf.int32, shape=minibatch.adj.shape)
     adj_info = tf.Variable(adj_info_ph, trainable=False, name="adj_info")
 
diff --git a/preprocess.ipynb b/preprocess.ipynb
new file mode 100644
index 00000000..164c2615
--- /dev/null
+++ b/preprocess.ipynb
@@ -0,0 +1,446 @@
+{
+  "nbformat": 4,
+  "nbformat_minor": 0,
+  "metadata": {
+    "colab": {
+      "name": "graph.ipynb",
+      "provenance": [],
+      "collapsed_sections": []
+    },
+    "kernelspec": {
+      "name": "python3",
+      "display_name": "Python 3"
+    },
+    "language_info": {
+      "name": "python"
+    }
+  },
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "source": [
+        ""
+      ],
+      "metadata": {
+        "id": "lFBIUQovI53M"
+      }
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 319
+        },
+        "id": "FXItxCYQ5xE2",
+        "outputId": "45d18b2d-50ad-4361-fb5c-1401166b8757"
+      },
+      "outputs": [
+        {
+          "output_type": "display_data",
+          "data": {
+            "image/png": "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\n",
+            "text/plain": [
+              "<Figure size 432x288 with 1 Axes>"
+            ]
+          },
+          "metadata": {}
+        }
+      ],
+      "source": [
+        "# python -m graphsage.supervised_train --train_prefix ./example_data/toy-ppi --model graphsage_mean --sigmoid\n",
+        "# test networkx and visualization\n",
+        "import networkx as nx\n",
+        "import tensorflow as tf\n",
+        "tf.compat.v1.disable_eager_execution()\n",
+        "\n",
+        "G = nx.complete_graph(6)\n",
+        "nx.draw(G)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "# download code and data\n",
+        "!git clone https://github.com/williamleif/GraphSAGE"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "S6709xbNrBok",
+        "outputId": "51e908ea-9105-4844-9427-b57a417bf9da"
+      },
+      "execution_count": null,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Cloning into 'GraphSAGE'...\n",
+            "remote: Enumerating objects: 265, done.\u001b[K\n",
+            "remote: Counting objects: 100% (7/7), done.\u001b[K\n",
+            "remote: Compressing objects: 100% (7/7), done.\u001b[K\n",
+            "remote: Total 265 (delta 3), reused 0 (delta 0), pack-reused 258\u001b[K\n",
+            "Receiving objects: 100% (265/265), 6.43 MiB | 11.28 MiB/s, done.\n",
+            "Resolving deltas: 100% (160/160), done.\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "import json\n",
+        "from networkx.readwrite import json_graph\n",
+        "import os\n",
+        "import numpy as np\n",
+        "import sys\n",
+        "\n",
+        "CODE_ROOT = \"GraphSAGE/graphsage\"\n",
+        "sys.path.append(\"GraphSAGE\")\n",
+        "\n",
+        "def load_data():\n",
+        "    data_path = 'GraphSAGE/example_data'\n",
+        "    # DATA 1, 14755 nodes, 228431 links\n",
+        "    G_data = json.load(open(data_path + '/toy-ppi-G.json'))\n",
+        "    #G_data['nodes'] = G_data['nodes'][:100]\n",
+        "    #G_data['links'] = G_data['links'][:100]\n",
+        "    G = json_graph.node_link_graph(G_data)\n",
+        "    \n",
+        "    conversion = lambda n : n\n",
+        "    lab_conversion = lambda n : n\n",
+        "    \n",
+        "    # DATA 2, (14755, 50) dtype('float64')\n",
+        "    feats = np.load(data_path + '/toy-ppi-feats.npy')\n",
+        "    \n",
+        "    # DATA 3, {\"0\": 0, \"1\": 1}, len: 14755\n",
+        "    # node ids to integer values indexing feature tensor\n",
+        "    # 其实没什么用\n",
+        "    id_map = json.load(open(data_path + \"/toy-ppi-id_map.json\"))\n",
+        "    \n",
+        "    # DATA 4, dict, len: 14755, column 121\n",
+        "    # from node ids to class values (integer or list)\n",
+        "    # 分类标签\n",
+        "    class_map = json.load(open(data_path + \"/toy-ppi-class_map.json\"))\n",
+        "    \n",
+        "    broken_count = 0\n",
+        "    for node in G.nodes():\n",
+        "      if not 'val' in G.nodes()[node] or not 'test' in G.nodes()[node]:\n",
+        "        G.remove_node(node)\n",
+        "        broken_count += 1\n",
+        "    print(\"Removed {:d} nodes that lacked proper annotations due to networkx versioning issues\".format(broken_count))\n",
+        "    \n",
+        "    # edge: (0, 800) 边\n",
+        "    # G[0]: 某结点与所有的关联结点组成的边的集合\n",
+        "    # 标记需要在训练中移除的关联关系，即边\n",
+        "    for edge in G.edges():\n",
+        "      if (G.nodes()[edge[0]]['val'] or G.nodes()[edge[1]]['val'] or\n",
+        "          G.nodes()[edge[0]]['test'] or G.nodes()[edge[1]]['test']):\n",
+        "        G[edge[0]][edge[1]]['train_removed'] = True\n",
+        "      else:\n",
+        "        G[edge[0]][edge[1]]['train_removed'] = False\n",
+        "    \n",
+        "    from sklearn.preprocessing import StandardScaler\n",
+        "    \n",
+        "    # 训练集的id集合，result only int, len: 9716\n",
+        "    train_ids = np.array([id_map[str(n)] for n in G.nodes() \\\n",
+        "                          if not G.nodes()[n]['val'] and not G.nodes()[n]['test']])\n",
+        "    \n",
+        "    train_feats = feats[train_ids]\n",
+        "    \n",
+        "    # 特征缩放，标准化：z = (x - u) / s\n",
+        "    # u is the mean of the training samples\n",
+        "    # s is the standard deviation of the training samples\n",
+        "    scaler = StandardScaler()\n",
+        "    scaler.fit(train_feats)\n",
+        "    feats = scaler.transform(feats)\n",
+        "\n",
+        "    walks = []\n",
+        "\n",
+        "    return G, feats, id_map, walks, class_map"
+      ],
+      "metadata": {
+        "id": "jg81uw5O6Gaz"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "def construct_placeholders(num_classes):\n",
+        "  # Define placeholders\n",
+        "  placeholders = {\n",
+        "      'labels' : tf.compat.v1.placeholder(tf.float32, shape=(None, num_classes), name='labels'),\n",
+        "      'dropout': tf.compat.v1.placeholder_with_default(0., shape=(), name='dropout'),\n",
+        "      'batch' : tf.compat.v1.placeholder(tf.int32, shape=(None), name='batch1'),\n",
+        "      'batch_size' : tf.compat.v1.placeholder(tf.int32, name='batch_size'),\n",
+        "  }\n",
+        "  return placeholders"
+      ],
+      "metadata": {
+        "id": "mReVSrV9UzYO"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "train_data = load_data()"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "Rv1F4lYF_FfW",
+        "outputId": "7b9e88c6-8ba5-4128-9564-55a7d4cc835c"
+      },
+      "execution_count": null,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Removed 0 nodes that lacked proper annotations due to networkx versioning issues\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "G = train_data[0]\n",
+        "features = train_data[1]\n",
+        "id_map = train_data[2]\n",
+        "context_pairs = train_data[3]\n",
+        "class_map  = train_data[4]\n",
+        "\n",
+        "# num_classes = 121\n",
+        "num_classes = len(list(class_map.values())[0])\n",
+        "# pad with dummy zero vector, row wise\n",
+        "features = np.vstack([features, np.zeros((features.shape[1],))])\n",
+        "placeholders = construct_placeholders(num_classes)"
+      ],
+      "metadata": {
+        "id": "CjSUlil1kNZP"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "class NodeMinibatchIterator(object):\n",
+        "\n",
+        "    \"\"\"\n",
+        "    This minibatch iterator iterates over nodes for supervised learning.\n",
+        "\n",
+        "    G -- networkx graph\n",
+        "    id2idx -- dict mapping node ids to integer values indexing feature tensor\n",
+        "    placeholders -- standard tensorflow placeholders object for feeding\n",
+        "    label_map -- map from node ids to class values (integer or list)\n",
+        "    num_classes -- number of output classes\n",
+        "    batch_size -- size of the minibatches\n",
+        "    max_degree -- maximum size of the downsampled adjacency lists\n",
+        "    以toy-ppi数据集举例：\n",
+        "    label_map为输出，维度为(14755, 121)\n",
+        "    num_class为label_map的第二维，即121\n",
+        "    \"\"\"\n",
+        "    def __init__(self, G, id2idx,\n",
+        "            placeholders, label_map, num_classes,\n",
+        "            batch_size=100, max_degree=25,\n",
+        "            **kwargs):\n",
+        "\n",
+        "        self.G = G\n",
+        "        self.nodes = G.nodes()\n",
+        "        self.id2idx = id2idx\n",
+        "        self.placeholders = placeholders\n",
+        "        self.batch_size = batch_size\n",
+        "        self.max_degree = max_degree\n",
+        "        self.batch_num = 0\n",
+        "        self.label_map = label_map\n",
+        "        self.num_classes = num_classes\n",
+        "\n",
+        "        self.adj, self.deg = self.construct_adj()\n",
+        "        self.test_adj = self.construct_test_adj()\n",
+        "\n",
+        "        self.val_nodes = [n for n in self.G.nodes() if self.G.nodes()[n]['val']]\n",
+        "        self.test_nodes = [n for n in self.G.nodes() if self.G.nodes()[n]['test']]\n",
+        "\n",
+        "        # 不参与训练的结点id\n",
+        "        self.no_train_nodes_set = set(self.val_nodes + self.test_nodes)\n",
+        "        # 可训练的结点id\n",
+        "        self.train_nodes = set(G.nodes()).difference(self.no_train_nodes_set)\n",
+        "        # don't train on nodes that only have edges to test set\n",
+        "        # 只保留有邻居的结点\n",
+        "        self.train_nodes = [n for n in self.train_nodes if self.deg[id2idx[str(n)]] > 0]\n",
+        "\n",
+        "    def _make_label_vec(self, node):\n",
+        "        label = self.label_map[node]\n",
+        "        if isinstance(label, list):\n",
+        "            label_vec = np.array(label)\n",
+        "        else:\n",
+        "            label_vec = np.zeros((self.num_classes))\n",
+        "            class_ind = self.label_map[node]\n",
+        "            label_vec[class_ind] = 1\n",
+        "        return label_vec\n",
+        "\n",
+        "    def construct_adj(self):\n",
+        "        # adjacency shape: (14756, 128) ，用于存储所有节点的邻居节点id\n",
+        "        adj = len(self.id2idx) * np.ones((len(self.id2idx)+1, self.max_degree))\n",
+        "        # (14755,)  ，用于存储所有结点的degree值\n",
+        "        deg = np.zeros((len(self.id2idx),))\n",
+        "\n",
+        "        for nodeid in self.G.nodes():\n",
+        "            if self.G.nodes()[nodeid]['test'] or self.G.nodes()[nodeid]['val']:\n",
+        "                continue\n",
+        "\n",
+        "            # 获取所有训练集的邻居节点的id\n",
+        "            neighbors = np.array([self.id2idx[str(neighbor)]\n",
+        "                for neighbor in self.G.neighbors(nodeid)\n",
+        "                if (not self.G[nodeid][neighbor]['train_removed'])])\n",
+        "            \n",
+        "            deg[self.id2idx[str(nodeid)]] = len(neighbors)\n",
+        "            if len(neighbors) == 0:\n",
+        "                continue\n",
+        "            if len(neighbors) > self.max_degree:\n",
+        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=False)\n",
+        "            elif len(neighbors) < self.max_degree:\n",
+        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=True)\n",
+        "            adj[self.id2idx[str(nodeid)], :] = neighbors\n",
+        "        return adj, deg\n",
+        "\n",
+        "    def construct_test_adj(self):\n",
+        "        adj = len(self.id2idx) * np.ones((len(self.id2idx)+1, self.max_degree))\n",
+        "        for nodeid in self.G.nodes():\n",
+        "            # 所有邻居节点的id，这里没有限制训练集或测试集\n",
+        "            neighbors = np.array([self.id2idx[str(neighbor)]\n",
+        "                for neighbor in self.G.neighbors(nodeid)])\n",
+        "            if len(neighbors) == 0:\n",
+        "                continue\n",
+        "            if len(neighbors) > self.max_degree:\n",
+        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=False)\n",
+        "            elif len(neighbors) < self.max_degree:\n",
+        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=True)\n",
+        "            adj[self.id2idx[str(nodeid)], :] = neighbors\n",
+        "        return adj\n",
+        "\n",
+        "    def end(self):\n",
+        "        return self.batch_num * self.batch_size >= len(self.train_nodes)\n",
+        "\n",
+        "    def batch_feed_dict(self, batch_nodes, val=False):\n",
+        "        batch1id = batch_nodes\n",
+        "        batch1 = [self.id2idx[n] for n in batch1id]\n",
+        "\n",
+        "        labels = np.vstack([self._make_label_vec(node) for node in batch1id])\n",
+        "        feed_dict = dict()\n",
+        "        feed_dict.update({self.placeholders['batch_size'] : len(batch1)})\n",
+        "        feed_dict.update({self.placeholders['batch']: batch1})\n",
+        "        feed_dict.update({self.placeholders['labels']: labels})\n",
+        "\n",
+        "        return feed_dict, labels\n",
+        "\n",
+        "    def node_val_feed_dict(self, size=None, test=False):\n",
+        "        if test:\n",
+        "            val_nodes = self.test_nodes\n",
+        "        else:\n",
+        "            val_nodes = self.val_nodes\n",
+        "        if not size is None:\n",
+        "            val_nodes = np.random.choice(val_nodes, size, replace=True)\n",
+        "        # add a dummy neighbor\n",
+        "        ret_val = self.batch_feed_dict(val_nodes)\n",
+        "        return ret_val[0], ret_val[1]\n",
+        "\n",
+        "    def incremental_node_val_feed_dict(self, size, iter_num, test=False):\n",
+        "        if test:\n",
+        "            val_nodes = self.test_nodes\n",
+        "        else:\n",
+        "            val_nodes = self.val_nodes\n",
+        "        val_node_subset = val_nodes[iter_num*size:min((iter_num+1)*size,\n",
+        "            len(val_nodes))]\n",
+        "\n",
+        "        # add a dummy neighbor\n",
+        "        ret_val = self.batch_feed_dict(val_node_subset)\n",
+        "        return ret_val[0], ret_val[1], (iter_num+1)*size >= len(val_nodes), val_node_subset\n",
+        "\n",
+        "    def num_training_batches(self):\n",
+        "        return len(self.train_nodes) // self.batch_size + 1\n",
+        "\n",
+        "    def next_minibatch_feed_dict(self):\n",
+        "        start_idx = self.batch_num * self.batch_size\n",
+        "        self.batch_num += 1\n",
+        "        end_idx = min(start_idx + self.batch_size, len(self.train_nodes))\n",
+        "        batch_nodes = self.train_nodes[start_idx : end_idx]\n",
+        "        return self.batch_feed_dict(batch_nodes)\n",
+        "\n",
+        "    def incremental_embed_feed_dict(self, size, iter_num):\n",
+        "        node_list = self.nodes\n",
+        "        val_nodes = node_list[iter_num*size:min((iter_num+1)*size,\n",
+        "            len(node_list))]\n",
+        "        return self.batch_feed_dict(val_nodes), (iter_num+1)*size >= len(node_list), val_nodes\n",
+        "\n",
+        "    def shuffle(self):\n",
+        "        \"\"\" Re-shuffle the training set.\n",
+        "            Also reset the batch number.\n",
+        "        \"\"\"\n",
+        "        self.train_nodes = np.random.permutation(self.train_nodes)\n",
+        "        self.batch_num = 0\n"
+      ],
+      "metadata": {
+        "id": "ZYCKM4i5PmPf"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "\"\"\"\n",
+        "    This minibatch iterator iterates over nodes for supervised learning.\n",
+        "\n",
+        "    G -- networkx graph\n",
+        "    id2idx -- dict mapping node ids to integer values indexing feature tensor\n",
+        "    placeholders -- standard tensorflow placeholders object for feeding\n",
+        "    label_map -- map from node ids to class values (integer or list)\n",
+        "    num_classes -- number of output classes\n",
+        "    batch_size -- size of the minibatches\n",
+        "    max_degree -- maximum size of the downsampled adjacency lists\n",
+        "\"\"\"\n",
+        "# 实例化 NodeMinibatch 迭代器\n",
+        "minibatch = NodeMinibatchIterator(G,\n",
+        "            id_map,\n",
+        "            placeholders,\n",
+        "            class_map,\n",
+        "            num_classes,\n",
+        "            batch_size=512,\n",
+        "            max_degree=128,\n",
+        "            context_pairs = context_pairs)\n",
+        "\n",
+        "# adjacency shape: (14756, 128)  包装为placeholder\n",
+        "adj_info_ph = tf.compat.v1.placeholder(tf.int32, shape=minibatch.adj.shape)\n",
+        "adj_info = tf.Variable(adj_info_ph, trainable=False, name=\"adj_info\")\n",
+        "\n",
+        "# 接着就是构建模型了，需要改动的兼容代码过多，暂不继续了"
+      ],
+      "metadata": {
+        "id": "mUW98eVhQ5H7"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        ""
+      ],
+      "metadata": {
+        "id": "Wp3DZreLrdtF"
+      },
+      "execution_count": null,
+      "outputs": []
+    }
+  ]
+}
\ No newline at end of file

From 7a5972fca5850c4de54ea1046f069c4149fe2956 Mon Sep 17 00:00:00 2001
From: 13929520142 <wangqy94@163.com>
Date: Wed, 9 Mar 2022 10:46:41 +0800
Subject: [PATCH 17/28] =?UTF-8?q?lstm=E6=B3=A8=E9=87=8A=E5=AE=8C=E5=96=84?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/aggregators.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/graphsage/aggregators.py b/graphsage/aggregators.py
index 6f117a1b..134f6474 100644
--- a/graphsage/aggregators.py
+++ b/graphsage/aggregators.py
@@ -459,6 +459,8 @@ def _call(self, inputs):
         # 生成索引，生成规则为如下：1.先生成shape为[batch_size]的1维向量，具体为[1,2,3...batch_size-2,batch_size-1]
         # 2.每个元素乘以max_len
         # 3.将原来的lstm序列步长减1后再相加（数组从0开始）
+        # 该index的意义就是为了取每个batch的最后一个聚合结果，因为lstm为序列化的聚合，训练的结果是逐步从第一个
+        # 传递至最后一个，获得最后一个batch的结果就相当于获得了这个batch全部的lstm聚合结果
         index = tf.range(0, batch_size) * max_len + (length - 1)
         # 将rnn_outputs shape变为[-1,hidden_dim]，-1代表自适应降维，应该是batch_size*max_len
         flat = tf.reshape(rnn_outputs, [-1, out_size])

From c4eb6bedfa0cb615ab1428ac3a0328c22429dc70 Mon Sep 17 00:00:00 2001
From: 13929520142 <wangqy94@163.com>
Date: Fri, 25 Feb 2022 14:37:31 +0800
Subject: [PATCH 18/28] =?UTF-8?q?=E6=B5=8B=E8=AF=95push?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/aggregators.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/graphsage/aggregators.py b/graphsage/aggregators.py
index 134f6474..184393a5 100644
--- a/graphsage/aggregators.py
+++ b/graphsage/aggregators.py
@@ -386,7 +386,7 @@ class SeqAggregator(Layer):
     """
 
     def __init__(self, input_dim, output_dim, model_size="small", neigh_input_dim=None,
-                 dropout=0., bias=False, act=tf.nn.relu, name=None,  concat=False, **kwargs):
+                 dropout=0., bias=False, act=tf.nn.relu, name=None, concat=False, **kwargs):
         super(SeqAggregator, self).__init__(**kwargs)
 
         self.dropout = dropout

From 6f19648d15a9c29482e24cc9bde87235d08128ce Mon Sep 17 00:00:00 2001
From: 13929520142 <wangqy94@163.com>
Date: Wed, 9 Mar 2022 11:35:45 +0800
Subject: [PATCH 19/28] =?UTF-8?q?lstm=E5=9B=BE=E7=89=87=E6=B7=BB=E5=8A=A0?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/doc/lstm.png | Bin 0 -> 136469 bytes
 1 file changed, 0 insertions(+), 0 deletions(-)
 create mode 100644 graphsage/doc/lstm.png

diff --git a/graphsage/doc/lstm.png b/graphsage/doc/lstm.png
new file mode 100644
index 0000000000000000000000000000000000000000..6a019413ca8e5bfcc7e3c7a08c038b36fa230991
GIT binary patch
literal 136469
zcmdq|Wl&sE*ES5|?(W)numso8I0+Ekoe-SHJ7@wl?w+6xfe_r?9TGISTY_64xXdB<
z{XWk-zrLEPsrfOdilVyD>AlZhd+jCHx>m#+HAP%3Dl7y91YBh$IZXrvWFG_s!~hI5
z;F~EETSed>qPwOd7@>TWW*7K?@=iuo1_7Ze4*Sjm75Iz^Q8I8xKp^sX`a@i?p@$+M
zJQgU+$!J4O4@^+gr)OsF%}5Q6BU>CkYA$H*p>cDQ+LLe}JzYGT!DQYx6yDxFpDb_=
zefL*yw|=ku$XX(#Sg;V;57>X@>;_zk2M3<#j|`0e*QJrm0Lu>f-%AcLn8b(Z|GH3d
zG<TGh{`Vr`Q-ZVo-vf01|LXu_j1RY~k4%!1v>!iytgNr^D$=kznbup)H4QFgH>qdB
zBIS(KGxNtSc$KEG-0XRBy`5vV6`nBHFouQrfT*OZO6vL~wxt$teAm-c5wAOu-=kS!
zv9#iO@1w}r^}ZpQ@*1L4wEn@0B;}1L7PQa%i%y^C8z)_5GeRXID!SeuLp_-a3}D^;
z_;4u=VrA`KT%>OCI_$2r=sD3@tQb`P7B~5JI1jSU#BRJJnEcEDOVDAeGmJhUT>9~j
z*kQJ`4|p}m-#wkX`^(kPxooqiojc9^5W`B{ma(OuCFzNk)F0F1ZPuU|Rp^rMRi{v7
zP=!gQZ3nXvl<qb1lv0Rlrq8p>(zSh4!(p#o9&=N-m;e6fZ`(^9@n||pD#bXuzB-3l
zxlEddkdWRp<bdO!lap(QzhJ+<r*1Tu1@`!st;kAN6AKQRPL<72#5mN-)fgV<$sf+x
z>TXI36yK2T={QF(=<MZdEcahs1;>py=<zx(Hj^e066T#KgTWiu{mU)hg!JV2=)Q_L
z)M7|Efj6R5!nV;=I`(;neV>bTsu1^d>_hvfR}_BkxonPncFA{62`w`HCItHi3-zR4
z@41wI8ArEV<$N!Of{Mz>!t$%>uhy6Gq<ARa$;rAbh=nC0u?p4w_u5bI75Nc2QjH3*
zNo~;tm2|*3EpuQ7tU$KjVqi*k?!`WZ*T@6*;=AL4>MA3?=0$<KJLPw+T9=XnAZDY2
z*DSTp*LXInIf&}&>Kc3f+wt-7>~rStA{@?{nVBu>T<F^$%O&k~%05!PErx5P^xH00
zOWr*4(VT^rc84nK&hcAGUyO=IoBX`ATAnjW|LMZwQ%f5wgu9p*(8pSIt>7t8YCw*B
zbKZBlIgk(^{{bDJZm^pfGh!R#uDu4|_17?IrldPrepnp1q`wSMj2J_gDK_a}CVr0r
zY&h5NC4S=>!&36S&Cjf<9W!jg%Cwv+9e<ETAM#@_w;#pfQwM}DY%NZ)>*3ZFPs@EZ
z3k)u+!_CXjm$#Ta)BVl{SO_^pbm}+Gbahq(c(u;UR0UCF@+m}b$4@uuD+4x!G8cEE
z%oi8K-CoN_TZvg-uF5~-&ZnIue47)>zDOdsz{+gB>S=7c9cQxXIB}gdT4Ykns?w^9
z#BC@r(CoP%@-cPvQ;1v_T2!SLn|cIQYxUcemlLy*n}^NE7AmvvMOyD9S})|8d-(}9
z!Unx)l;lHtXyxX92DZq_5y(1ft~DG>@i@o!9PJM9-`$?tYL;WFkaB($^V~BZ9B0J9
z#Z5s~BqACkoGT9r3$xl9Oeiv|kcv!_v^6f~@bO#1F9|#B{if4yE$qzH%R{*yYpr#W
zB<b3yD)ftt$<fZ_isKtbgIdr)A>RSzez*p0|K_26*_*wT#3EBTR>0#gy-exAhb9Nn
zCqpQQh<5285gXAp^-W8o#)ey)Ll5_G<pz-}EU9yG1DQ8N|7yWlK6gYlWS*6Ak}(Cv
zlvDKWP<U;NZ0L)KkQG|S3#QUZf@X14#ulRWb}1600mJpQxMsM@ykU`UKC40;B{NQ)
zBQY<d#!2U~r<>dJi-Y-?nd5pHShe6pUe`9oj02sc(%O6HC~(m<Ziu0oV&kEp-27_p
zFVB#9KD*^m+UIUR4vxn!1VZ#zKdv4*K;~)C2Hw9LZ<D1LC|xD^#g(iG*ciy~AC&XS
zTHp=Q&Ma|n$mxcBt<xeIDVH{Q#J=w2`)pvR{ckgfDLG2cVFE^#^2?A-PdU+*m$jSR
zUF^3;TVIzz(y-Knj=A``8a3H3I%3xxLnjM9q+zLzmOotF$E$ze7F{Ti3Iv|19xOF!
zd5FI^#VRZSjwO5Un%9w(l+=zign??xDLOPI!+gsn$<=m(<?66G6obg4t3CpwN*$bz
zENZs(L4J-JC7YH~wMA~j3E57Xa^g)136Z}yVKXnGRLiC-wBUt;XErQUXGOMcL3N@<
zogNB>Dxs@_^AJADoBH`lq;i{tlBqW3m=M<n9^*S^2VjtUg)n#C2<2TAGLEv+PJq^!
zN}ZXyZ_v6+D-uqa7mQ`o4^=~ww8+%d42q_?tJ$p08Vfp2t7Xm~?{x8>(KevvwBM6*
zrd6n^t<F}~)DYhzyj@QKR2!LGZnfDHTipA&pDo8PR#-(_VZT4ea$yZsq;;1DsjNJq
zBO%1rnQ#pftn^5$bRfDUSBreja#0-EZJh&ijB~>3F4H2ka8Gj_HMufk+4ysT;h1e1
zY$mlAX<uN!KGHwJ_wIs5PC-GEyj~tl5;Fn`!}+4u-`3~yFv{&0XP#?aLuKxz+tpFH
ze6lL1<sicIe8-=IB86DhgIe}~66L>%3pWs{GSfImyiLF7X5uA=@!@zXDG)@+H7vX8
zOF-}qrypoihIP7fmVKE(B631)iSg_RYnm=ct1TAyXxbH;w4|?e`aj_1L#&&)z@^Hi
z34>Af&B5vZm*L)oZM?Jzy4LEYiOO5c3gIK3D2t}D;k1|%M7|cF^KM&l>^2N)w^1dE
z=lQmhy}z+&(y}_=aWCc@e;T$u)qRyI<!^Y7nLN_X0>Z-6-*RP)Ei0onY`f=CA+LxW
zqw)?J$^3b;Kpw@QNam>=&PdPvIm187P`r?7FVEH&#NVxSVL%e9arMTTRSkzVhb7%F
z?d9#hNYf0le>)Xyn57u*HPW0eYE@70IgU>9`IlmI@IL{DH#Um=NEbzBQAzDwRF&%b
z!=FK5P97VNJWHUA#QhnTjf*KseH8pI*CaBxecUR+n;34*Cp!cQ72^c+PVu95$+<l5
z<Qh2Yyv3Dzc}+yh|F)Z?x4)N4lbMn6^jT0JDf!&Nd>y)rpg;C6mE?h<a$<;EFt%S=
zIYV-<LHe?1HLjk#%$r8r4OOsuxMVfqOoP&viDGUR(>Y0%VtucW247|@K3f7?W11uG
zA+{6*b--H$zL}P#(GC&~%<;p+_r^DXfEYDCbi#zcs%3${ley&k?4n)Oc?M>h9);yf
zHV_`AM_0-=DDSXP!LkXUK})NbvF2YCmh3qENG!u;w)kafTo^mo#WF8LEVQ>diuArA
z0zNFZu0(J1=I?%}HtFl9hys|Pz}oSpUb%5ZC@yuLVajKURR^BJbNvs>Cht03`#7$Q
zOh^ys@=J#(Ym{|tzu*6Pj@1T1ck5Q#fl-Ih@%A*@#6?$do`cgt=SYW1>&;v^)csDf
zXOY|!rE1vE(mURq?<TlgD=SH4tX-duW=V#5<@m+i{Gx^*3nN3^mQk+^1|ktN!9!%3
z#)r}KZR8e?+n?q-<|q+w6NlQDhy^{CczJt|(1Ie0$E`TQagGr&$_Frm38DRrtxOYG
zhm}An87+7q?&0M;(Xa6aMS2u86e>-Yr%eh*H(eJMXySE_$YVLx_*qfqIf;1X+!Cr(
zA}|pH+Un%aACIzt4~)EUXFs1)Boqjjv?;P;y5?*kG!LXrtG}`MaZEf>=_grhe?wM!
zX^3f7kozCqLJ`X4;o-U3+Ejnj<VT8g?GCCRaqHby70dOSze%4Cq+DuHOkh^nh{drF
z@x)?kM1Bz=sk4JwrP%*G^Y$>I3ddUjL((ub3BKi|09QofP47la5W!)VJy5rF@tVCZ
zGHRDvXg&K#u^yQvkBJ%a4tvQPhW?CV{2c?8j&0C!hG%aT%U8l$-bj=M)sTfmSx-L9
z7Mr>C!@c7?`&~nQ5kD>oFEcjZZf>;3Cekpggm*+faBDUa_&|`lUH1_4FPK5ic>cYR
zFoE(#zqdMxBRNffMx9}<%=C?ZJ>;A19!my9MM+{YJSL{^WW7hD)Yg?Gunpz&lq1af
zFm%1UpCP208=?dre8^yhD+<I=#wVAJ69-^pOJ3;cZ(&xwE1G9kh?i_`oWVtIr;8qz
zFJ8;2KZLDnzril}=O!4;(Zdxhs04jTigk)Zc~_jHux}+S+D=eboK7$!$vGmvEt7Rx
zIgsQj?Z}_*XfI8v;JnHVqOdI8>G&EhCBNfZW-0`Z!*Q<UPTNRAT=6HR%kVq(-B&Bp
zfs>)(P<&WyaCx=;8oWV?997I57ME*NKn3kEq;{mQAz!c88+4A3-&_#RutpJG>P@xt
zohVWvZZ&OWx4|>SY%p;RHx~!zaVN@UFr7EL(1{wO6@z(=Nqyvw8otIq7%ij4UF^4s
zBPHGZczO7|ZsMWFWo>((?VrGU%=zo=&RF?j-PC<txdy?h&=^O>zs0y~jTaJkP8Er9
za1TlixfNPDV);OcKZ?n_hC9;Nw7-$m-=>1cj>z&ZO=Od&1v5>5MXFQP%NA7Y@Ig=3
zkc{F%{~{B_xa!?456r(*GxD2Ms+HEghERino`4mO+Ha5iDs84jFT{91Y^N&8B_#DD
zSA;Ak$bgWkRKbS}?IgzYX=;|dCK!@^=GEi*e9!*Ky-@4>z{wP!p|nJ3$(y_p$p}xY
zjEg>N=b969MmL;-VhcLVkOXrKYKX9avblladwtd{K)xxjfWP1;=*khB(bbEV@Y#9e
z)N<(UHml_s2^W9fu>;I>Ac(PKtxEi}Bl<it-BG5)aX8$IB-`<|2!9X?jc~{FzfBqx
zqZ{f|M+qj3x1f-%_6(7Z74)2Gm#~{@D)Wo;c_n3K{x5HU<iP+5--n2pIQU|<R<Z7n
z=Xk*(-`(A>9Cm|q+^OOv1UB8qC=SC7N0;@;5!K$sLZ*J6tX>nbAp9>(C+3{o@9EmC
z`Db2wS!>vy>l~0y4ERUnmsKR@xDf}46IJ?M2~AZjV7LhR6TchFCJoS*`Ik1MqDb6b
zH&$6`)E`jq+&HOcE+1e@ml#1ZeW~cDZYc`Pcv(WS_b=Xr9Iyp%D~*ziS`&KNg;y!}
z=#+>nRob9)lO|j&Pyr|X8xwEMICf1-u&vwVMyU(2OSfkY`Hi1ALP1noSLa7$LRwlU
zrGk(Uxd3!w#MK&K;l1hN;LE-GkYz{o5CfKYip^a@sG0<s&o>PCW#D&^kwEbkt0^12
z0Xj|vJ&Gh!t!OmRp_or4aGL_#Qrf41ABVEzPd1ffJz;bavSd)kdX6Bt5DEHx^7Xym
z#)!Sk`f}<qBLl;mFY16~=aFX94;NeT|L)HWi2s%DuHRK0|JtLU%w>v-ThbPBR)VCo
zlYnoT#N?L%&F~=d@g7QDk}uknL^<wId_@+((cC?nNhl`?C%1FQozSgq7sh-y<Ga}+
zBn|gTp`wuVskQy6J^r6)8bMeX;^o!vdt9dw!}+($<Id)44%8WFMszVI9*n;i7^De0
zFZBKt_evaT&4NDayDt;uy%ET`p}#YDr{hIX^fOWStq@oe6k9k_lFVmep*0%Jo9uAj
zXn-mT(b%f&Gd!!@)vI)G7PL>Th#?y>3OAIT_XFbI`InZKOriMHu~;m_R<;A{kok=;
z$bJT>dk`-^{|m-*BA1JAb3zk{s+_GbOIW1EZYM5enG#Pu18S(SJ~^XrDUsPQ(B<z<
z<L6Y))cb(ObnrfL^8jYrcu3nT!n!$yMt*lPam=n)%vA5V!t}Z5m6DQ#_}cN~NQUr4
zz9PPM#T#V*c6uc8YEtBYOgV4`CzURsYI>-QonoXiKO`Fn8gH?JE=*-HqMk2$R7+kC
zdMz=przwDWe5<~)orlcTu-x`&^~4uh`wF;^6HS#*A-L4pMpRZ-A_+HSKOsD*2umQ#
z%cJEY?Qfh?{+Eg(rEXvb%5H`X5*h<P!m@q;kVHVT5B%isW}B?`LE;|w9t=_r=aN#k
z<7BI?%Y@7Ulwq=5>fHlkL=s>f5z3#C$R6XRw@kA;qVd7MU*b_y{?r$H7rjiIY3xZ!
zyYfhyGOX7H<a*YtRDeH1IEcT#z1{rI+b-8flu~8&8QJgBm<6#AgSR|p^R0-esRiD*
zIjJE+7hON<KJ1jv;q?qx$2M$T^_7@asr{%pYX2yFYQJE-V~>HI)$Oi;Yx%1=mT-~A
zH-n(_#@uj+TM^R&j>gHUr0s!bJl=iC>jOdZ6qV_h71?dQG;jdk8pG3<Kt)H-x5y8*
z)ht2PjagR7<H5TRq=|E5QQFaAhBruIl<S^~G!J1L*(_MC<QC|^@WOW7rVZYY#R!Ry
z35O^#XLO!DtW1!4r?NNaLAa3r7=Q<J)^(K<tbalwMfY4;CkEf?-&8~G<`x&DJ`V(;
zaGZlcSO`$twQpKq^H3;@WwHu4Xej$$hP8msHC>^nEVXuk^CIJk{CF6ecRY(c-t6o!
z#XF1ecX(R1Y_d+H4yvQgtAU&&=p5zr)3J9e&mvTL>b+FkGZ<Y03{DURy2@)Zk)*V|
zJhVt3_s1J1;SEb*R*X<iV*!Ew2<z{NiiqSY=d;>sY&&I|Zil^eAT!%CcYPFV$*2Ws
z*kmZEX<=bXq6?H=k(v4y@b%yGBl>sHMzc;E<vGI?isUZ}(B{)3`-<~yhQzKOdSd6=
zI#X%N_rV9QTnMmde_n&flvZp4N#W?_Ry_h-Qzer$P7AdkOEO9r>z)LT)R^<o6^yRY
z3U&2cDVB^aa|IJbG(Skk$bYg-dGj~M5Mu0ySA$m3`R*8q?xy$heF`{>48D<3>A1um
zuk54jqq40zxXWMB#4(?zTh!9M-r`{+>>geExStKh*_R-$=rBat9Ec&$$0Y~-3ri6I
z5az?ta&x|Zf~LNxGx^e`u?#~*E5kks4HJB}`9YEWPKD9$&6#YgQJvs|@FSj;xO#6Z
zA~t{f8Ohy+1-UkdE;4DPc6UXNS66{kI|kWEZR0hvs8P@-)NGZbWAb7DhsGp<%kNdq
zxW|(G`qFmz!=zksW?~OatgdMz!YqKyq1vtUH8yasy5_6XpQ9%XMnsl@dgl#M1hSc{
z1m7Vo+WhR@u|pTOnpLQ#0jF>MCQq1K_0FtnU8XUaThzPG5zmH3Uw3<R2li6KhT)n+
zSUZ1I@vjUOlOV0iuJJL1TUW2aakPANxakeB#PKMRbd0cqcjJW<+eLZ}bg?v|&zvGc
zdS9R%-3qy^NRV(EWB3PBp`{%%VpO$9exx<n(UF_C{KfZY+tt>j2G?-L51U@dF8DGd
z;DfM3Ys7FLXZ|_YukoYhScrZ=s3VekkJ~MOFI|I_ipKB9x2v$j7?bCBjxJWGJ5KD^
zhRS}c>%RO1Ix<23$GmLp*B`zqx0cr*lZ-?tp$1JQqT=(qg7-sn)(X<h<aX0kER<SY
zMXJV;`SB#$z|X<T$tDW~A*)h5FirX4T={#OIl=Fd8%hib-spr|Ki=9ThFVKp64Pph
zZ{2CMN_?u0(|Jatn<@8qa{9FIYfjJVn3>o+BBE=WrQYJ%97&A^H!YlNPqaHZr+br-
z#WM!g27<SNvH}yaYf`PZ$L0ehL_R4i-Pn)sG2fbUM}0hG+nj2KpwOy)%+Zkux@1}V
zp(5<Jux3Q^zCeI#8j3xj*AyAh!B{VG7-S(O$#?-o!(I_z{njTh-rVk6YdLS8QwiTq
z-r#F<>Jf^29T1F61LrOq7b4@b+U|dPvtMB~lUsmM+pb`F$clC^LG|{Pw%mFecE0HZ
z)sEWn)o4|Vt7{qCQUl87T4@Jg=^;UK;3;l0ry*T_vug0z5e16T@{hW4E380yBPCT=
z6p^LwdePV--FnUyr@uTrCe7ALo?RdQBm465@~gFbp(4|AG_|s(XPI9y+&%}DxvY<G
zU_&ba#n&$4TPgcVRbLa!Xg%g6XIQ`lxm680QVv?pcF-5M!!On;im-Rl?#T?m9wHcE
zwh!c_R7l~FBMXRZ)GB(|W}#Fi@`Ix*nJ2pMIHi5iT3nvUwe<IU*?-o3efCr?*s?mg
z9a1d~xO+gKs#wDLc~c(Bi~Ss(4&x)bbmK?%FbBX&3rKznF(ty3G!kHeUk<;j*o+eH
zfNgfa!l+hN9DERvON_y2XhuACB21rtN6DFHajQx1pl|@|m?m{=WOr`KuyiirlP$|;
z`V*d~&M;5h^YTra(9Ome?@gZHH{nXRJ}dxuAB(pKvY)%y_;`Z5^PlSMhP4!_af?Cb
z8Acc@-n791G#{VKf6r5bGuh6za;rU)hUK<tVAQeRRvvSLy%aa5s<o~vSHjp08&Lx8
z8nH8lP@;n1I83p3@k&I_l=TL=1W>tbe3A4!st)9cN<%Lk+>%+vn$Q^6yE%T@Vr;Z0
zpqU*(#QTT_*djc8y`tb(8Lho{IvN)MhNafI-uLc=2eXYyUL{xY_Ccg&hMI6@t9Jei
zqaVGfmM#azoHz0XmwLWwnF70Bo<!)Q1X;VU6&%MZQ-LX#MzXx@1+mw<r6eYNtEj4n
ztoczWDEy<GnH!e+@risAr|LxUypf3OUE`BRTvIGLCf(i!{H&+l)3XbGFWb<$n7!gq
zk9YtQ1;7fG*S4?gd-w?#*%tF$3w{H3Q3L1B4k@!<j-mI02kCt$8IiNZuItGP%0c?>
zhfu*4xjs!nzS}XulLbm<M}_!h&k&J6qK&$72NAUMWULVq<?Gn|=l7l~8*u@vcru?h
z-d37-A&ClqeKJPU($aose}&mK@<20c3-awl+(aI;{3iqJ?5g03*AZ^|BG#mSUWy6F
z-R_VTx#UCA-o@1q17y<Y*M-uN9i~Y1!cI|E-<JqR%diG{!gUp&owzZdH@^8y(JcMx
zj+Dp;J;}W059sr(U8I8N{`z=VaHuvp&VL*N<oCH<t<SB~RJk@fxv(!)2%V_9(1b|!
zR&vn<mtifAf>$=yc9{i61FMyyWgMl#{)6tUEA*a4p!Ak?n8Y9XzMY!$cMn#T8o2Xl
z@nrEkI2M-0c}}80x0c9H<mtP)R?nThf!F=ZWc{S73D4|#@7D%&xbn2{=}g9m>-mnx
zQvmRXUw`Lx3e~CON|Vs1r=j!bMYC;UnTXab;o%^dmoVYt+xZcj#UhBB$Y<ol2hNnf
z@D6p!%UC8{RSbM!u)`8LAt+l`-t9GnDe&!#Y&fknl5$x$UY0z{JcIhADl&MWot-FD
z0wH-pFYX{ph8^v=gJpEHRuJL7gFZB|@0I1zQWL7PwFG8F5Dq;({S&HNppwc2Z%A_N
zbwY`tYFqE#F48O~J0Q!Ei6@=Mx%JRUC)tOfx`M7+8(<cGX}jz?rO796FUF+RFXY7=
zc(0k}u_Wbjs#l5X&FuVpwPD$Zdq1Hq0qz1tTIA>#$HHHl&0qe*(-7F$*jiVCNL!~o
z20KHNz+y~`wsJhG*DOkXt<8r%Z*KoBU*Q=|qhr*|5Zu|F0xFm(0%jZYewOl%psXF^
zd1#N7o&e>52h=fQU{cCg>8dYEN|Z1*g0)KVu}SxM$_+pI)>%lbZJPRiQMGf|v+JZm
zoe@qR-48l9Z2McE`AQwkloPzFPWEQC{*ml&u~5&_dd8ZaH!^(b_2N2N+nJs+fD6!x
z>Ri}i)TnpQL=1=QR*?G8x}!M}0DvA(3Jt$}t^7PI03jX3H;rR6i~p)+p?HPn1Us8C
z=O0J`zVdA4pb;hWxo@YBoMicKK9tDP=zCetI1R)<*H2Z%opE?DfzLkc;mhztZWMj9
zo;CWg^%%=vgWh%_Rt{|hj!6pjzslKPN4kRGib13Aj6kejS==EjEq|p+z`S#*he4d@
zgR6`ka-_bNwU8E)iMTnDRk~w$5OsadT^@3S@>W6?V*+`#w9dB)BY!LxVcU-if2r3h
zm=$rp@leEr)dA%DLx#NH&HUFQ{X!-@Zpnx@M%l?Cloq<PC$o7k(BOJhRxLl+;N%=P
z?Y5NGp0ekM4u)Svhe^1@9#&&M%mQBMpb0aqiTp(8XKbYrgf6W?t~2g-rc&01uNn?w
z&k{znxcqXTzU)sZR(nM`p!>dSfQRySwh8WaeRH)F9JdD4s-M6JDYT9piPc~9Sr0Fl
zi&gpBXPKb$W^7e`_ZEYN@#C9@Z5TvQ!e_sg!a2Yd-w1rLl_-1;R2bj`{`=6*RPpe7
zmMYee6%HaFm!$C$XA++PU!uYG%)T*h-yZuJOc5t6vVuE_JUaDUI+^db)o+*H9Uu6(
zcs!J%bEcMkh@lo5d@l;X5{z>4^6*z_JnWiRg3svM4bLb3*0cpNO5@63Z~S`Gag!m^
zG^&WhK)xQ&6JikE<Jp@Kb2?8JqrAMGFwVsslD!#JgHhXt*3ov_SUH1FK68yJ4GIro
zCYL&$;$0WIw7o`H2T{F0hXi8zru(JK2bTAp^I*IBWQ9$ty9A2)U9eeR{^=OcwKr*<
zE)@lUO^5&W;1d%QSHIG7G@79tj-MN`rF=Jr2wy2)wS#C(Jm|_`<6JOI6Y?%kUlOR2
zg-q}`HY42I>+f5E7Z_~;v<lrxH2lYhdVCE7h61JINIhaddjACYt~GOmaa>BnTa7SS
zgTFI-_#EO5Wbb<IjQ{baG5Lik>^3L#!{cQik<Q6|18py8;GZLRft-z0S@q{R*Y)pQ
zjJ#<~dO`LZ5*0#+DIr6582jqSPdWD^Zek!~f)ZCGjT?!$Nw!|}C)5-W`jD_r?NMR2
zy#P+%TLK@RjBdXBI8tbZn>dbxH0?gxVyC-9G=UF*|1*`jFpl1^<~eFYtan*z9VY1N
zN@haavf7ls0%hfzkNdL~R(DsY?05Ez(GONkOJ1<90U<ou0Gc4IzSWn(RG{-2AI5NO
zD2<l#&N3rh@mP8uKj)A?ebNf?TG(YR>xMpITpT?xgR8Mt3c+2htbgHS%X%w6!NV=l
zAg={H&62?9lE7vF9Jj&XTVt_97i}v}XeNAz*z)yIkzqcKAWQ%{@m}}!Jj{*cQP0c|
zlb4tGf1P`y=RTQ2?3UUH3xC-*jfe}LJPKt~xIc%8ib4X+^mO)>rDj|KWD(ALK3e7_
zQ?tOWw3DLPajcLq9u@ppPCX}j&9G9wjAm3w(+N}{EdyF#mXuh7Dxw$@gB{DUs<vBr
z$yy4<HZp8zA&x0*JPEm(QV@<|GOoDxw%ZMWArLyHqNKE?2eJ6^dn99`(o4(l&+i7v
z3Nehpm&n^oc+h5v>9n8urutq<7(-G#X`H$a&D0vb5ygbiX9#p|Yl6{M13F*f$80mH
zI$=FZcO%FVzQd#*nBVxJLXUE=ZZS&=C$~sa0>`1`jL_<jNJOU!8s+`UD1$sKy_dCr
z3UE=seEAY?r(w!vOwr-LK>1F%xo03WWCcH1KE^b7PZi=8gN}6Q99e~sOx3ms-nl*s
z@1M$!0>r=uzvREbhNpcv%*S7A=)WH?xPJNL#tC&aewA=&<hhdzWBS*0CPwmMK|j0*
z4EILN<K+aA<}IZ5z%yY-Iv3UM6SQmIZ;+by$RXs^yg@Gc@vnL=y}i8@nkIbz1&{Cr
z+nq-YCrWoO)mzrj4e5JDmlQ#OPdOXS^<2ge8PE8${ahoT+$K$~|IIFT4^CAxPsmC}
z-6~U+KApc<bPzda8$pR2je#2`z0f^Tn64Oe=aHCGT+sB^=G=>Q47?vr405Izq92!5
zvNO!;TqY7ivZ+U1FgxkZF60U(Q>q|c;Q6l#0E~=)E&1-<yVl1e^dgNyM4CZ<tjgv{
z3^LQaRczim@K|;T9e#fvRvQK5d3VV8D_}e1LFe{0<0YZH%V|mDQN*b7=&9L?w+)ZE
zXPKlzpuyj0ecywn6LLZDw3o5^6rxVA!b)XEo(rCw#dR>SSDLHZK4a7v?@}#WURrun
zr=lZJu|gwWSTi08z$N{8Oqh+K(@HyKUW&QgmzXxLUg#5O@NFvspMqNOm>42kT9cJ_
zi$VJJfXW~F(JpsI5zOWijGDNfe#S*t>n=Mvy4LdZ?=>CD?Zdd428Y41F5W${-|#S4
zR>NQ`34#Gv>zga=x{VY<&U5T&UpSTikKf_Xk_bTJrfPUu(tK>SJ;jW%m=wuQ<P)}?
z+3)MQkP9W0-)X{xf5l0Tm)R+E(+M%H!UCNu^f>5AfXk9jW}m<FyR$K(G1-qaA22#u
zA&wEvnlb-f!Btg#taf!2L8{tv2UM7{{qYW0Z)g?un!BAJFl$Ty;!({yAt|CO@)Ono
zc<F;87>v~Hc^K=~$7y9aMhaOa<FKF3b9_JPK^G1~S5cPdD|7i|%X$tqBqM?@#qlO~
z%l+hsTbn`U+6wEMc3H98-h$3I2(5jI!&1$<8_L8s*CL}Zet#o7NPhnptfc>Er<4D7
za;=F43Avp%=nu=V`9mhGF~FoAFFqcc+9`##m~vqj7-jgY#7b^p;$)dJhH*6x1lJ>s
zDHiCD&~XOuz(hg8+lSHh>T;zf&s)8F%~g2icj_iF+P-JV@@4oX=pcJ1br1o#GDV~F
z-=3jgU-?hNPW6SYX?aS$VPn}C)f}^cA?%(P{CM?kI%kf5L}Tg#HdkP^9MWT}c~3m`
z%^<F~NSsarYW_d&1_jOt-?{MB)8^2kwGtZ(8Mv^OGi2FZyZZ@kI0^%u51YGHus{|G
zEl8k*XrYG7y-g{mwx+ixL_WVJ_HPLl-~;Y6e$(e{BG1I|d5JttTk_ZI1#x(+t=qg3
zZ9N(}7W!KfUo^V5GD|t#A$$Fhh*n<y4R?zezFKdgUpqc_`7e|`+s<oNj{{Vh6<Ww>
zU9BTtM3jSwoQ!4P5%^pTfqpn+!)L!(Vw14I8*G8v#c_lz3AThdT{w$Ld`9RZw*<_l
z&M41vh%<f03A5FEASyxh(QrM|#mFOQWT|7YUXGm`MQIB*x59Xv6j{0aZBF<eV`1u%
zK(Iuw2X;pOb{_!%CJBtS>Y?iZU}{Wz_&G3!n9GE3>#J*#e(7^6J-Z;Sl4SzN5$5B<
zB(p@;-)qr~UE6gp$dY_{S>bwkWqaRLg)8FYh$G~eg{|)2P!qZHTl+Fq`Dbz$-ap5d
z7QDl3lirC!UDjReef|z*-k3P4HphVW1O3WT1h*D`XHH1YZMMVC=?F-hYn7(senI)1
z<(aJ&=D*zKWhXK1;z5l}pBw4viqWyvtfkS!kG-`&QMSuFuK*l_A3ny^cz{7XQpx0$
zG|1S(OktA}j?EQx;#C&+rr1{0V}IOCU>T=Qr5iT=apz$a0>sLMx)+PVQTs@Plr21;
z!R)Oc&m8>t8wRNERto9u)83;WuI|UO>KMU2A~f?l)HO8|8B_iNc|69AAB4(Q+1N^v
z$Z(CZAD1yLi&4mY9A7|lNqPKN#U>EBOWMPpKcV84xF?J5s8yhIB6j0U6D-Tdh_8RS
z=Xk|nmwT<jdo0D4_Or09WR~#6+p~S-aro&m#iTRN49>iQ8m)R`05!b++g5r(kNc&Q
z$F&=Ry>@W~&$3w5!=T$6Hej#tjkSIs9>!c6wnZ(FQdxq>U^^2QD8r;BK|pbM?FT>%
zID93O{Fk{}@RH!Wzu%$OdOH_gk`KUb7(br6$B1<wON6xGHIxy8&g(4$(p2*h#62Ey
zfok%`pP84hZ|o3f@K$2JXCM)p)8ZeQZgd=TJ{WF<UW`2S?zwt3R!Sm?Z=czA(&yV`
z_0P4v!`IZ(!u`fA^3U3RSOZY}JiFd8>Bs2el;ZWV<C|_5UmD{EIrI5S+Vmw~jsvo6
z;(nfCo_Y2^=?Ov+Oa~k58T{WKnH+Ok%y4vm)cGGY(q<n=b2DFBE$#&86~@P~MoBnV
zQyxLF1AI3toi-R>-@0=vcp8Q@Q2T}2>)7Y|-+JX))zDy>JQZlLi{HLVTaO_sIQ5LW
znCmzZ>2|D0Xj38)3eK=WOTq;LTjVqJvhLcI8u`_wH$=fz*1{Wy6$$xX*b{i(1U5Ms
z+q5$JLYA~PmqsADoOA~NN7}2SmLO#8QOo7=tU0OvR7d4!^J$A1jBH~L$DRl2UNoMZ
z!t&o6R{AG473Bl1pQ_IIOMWi^`Xn*iYnig)zZ0#x4j#_Z7%tNfNUOC6C~1d?;gwAh
z6x2k%^*|A9lMH3|gjKu}dWF}jvyFT=)y##*-3Tg#VpX+a-coLl-?mkypKlV;nCu0;
zye$D;Y*e<cdWrmy3P#8%v39UsT!A{$S;QTcg|2@`qdh&l4Ya(f&^kSPC;6ty*y--%
z=U8}D6!HI5>M)kR4E6V;jjTGx-jHS_p4IB%=pD6OavFQfF%Dj&riZ7aD-o!cL~l{G
zFP>DzhPCKClep*)5Ae0H_CPL!-h`BpC?eh2NlqQ5+LnJygqfyqIzbK0`aU_W>x(W>
z#W>W<<@d)_I?UqnBLa@6w+5BRA0HF5i#*3a){xZRLTSJPvU2na9ILptRk#{~;o68Q
z-NJQJ8yhsxj)9`TN+kQ~i~(yRc9vi$DeG7rB$F0y9!7mqANg^+U5P)C>=A!4kQy#0
zy^jJtj3krIMPGz45~&yZk$#==g>QyQYUEqe;?N1`3zQ`K`vbX-3B;*2BLf}aAu_9w
zaKywJ`UijO(Y3nF)555XV)^Jz%D)u_vGBkVMr_z7LsS-NnV@g^P4{8sZWFRfr_6Lr
zcFokV-u41d@wZv%X69LAD>q8lZAFq(Q|Q^g|9UG6h0Mk5kXi!0Y|!z>4Z^`E50u41
zG4tIW)t%F2bM0pa*kvpB!EL@}A%k4uWvKs<v6tp-;aTvH$Y^lQPxiJZ$6PdqTOkZw
zbW0yHUkq5wiN+(=sgsQ`p$z5r6HeO@C0JHZknnGqr2kb7p;G9=uIu1dndp_<1s!QC
z#OIT#owI!%v-l+5ZhOaNuUi^Mx9qz%oWeE950E_-x<=B0I$KcaU#eV-lpFo$_I2zy
zkqp$`Kj}|uvAee9VDXikLt#y3xW`((CPZMoB}P`J<S+^EEymGfhADp71tI<t$dq`b
z3=CCMtCo3vltXUoj+)kRW2HTy4Bh-234L6?bp5lXJa4V_lw&*mEiH3;)1pS3%xgp{
z>!UV+ngL-P&pULEA8%nX#H$?KEL?qdl{eRZVw)M;`dR@rPYD~o;H&0`qYtW!4tl}v
z$f~*Ev9%z27|gUNbY+HLUo-%nY8HzBRz<HTdbbAq96|jlK_lU-to(Cw(p%Bt)x8gU
zmEYrQV$rbk?2fjA@#F;8f}lxMr?zd<@x?+aO14umU*S>%>t{LBFM^cn86f@~ol0&|
z_Y22oL-Eyv?n+GWj_S=DnMEOv#t~tuO}SPX0qh!$Gy%IO)W$q5N(lgY7$-Qx(3N;J
zf|xbWW_qv4s7gvmy)LXl2Ps07$o0aScu**0K1r2;*@?&iB{3b_t2Hk79-qL2A?u)r
zmhl{lkOm18sFFTsz-$fHWB;M!cA|ukC`)dip>XkANWv@3VvpW^ywbyJ>YEO=&=$!b
z3F9Y+Bh}dXd!f!jzur-lQ?ESevksN{<Odk2)3R?4aY7;hVLp|;rDZGaOM+P<)8L2*
zv+|w1$QP`tuknuG1LSu&V2ZD+m2d5a%3|hR2p_P?RhC~8L<}7!m7ZwyjJ<(7%AX%D
z;&*j*p;iwSV*RVpF*Jf6mafmPb^v-#9oVV%@x=SU(1*ks-I4&h6UuI}ZKD?%oEok4
zJWoR+vx*=fBAn?R7Z@k}CE8>xj4rLYrfZAU$lCZ&Yj@5XWx+VCqlc6zk!}1m{|I0`
zO13_(po@5&H+uYdDhKKxR51(Kt}!HY7?71w`A0S}3vX2G2p|nHMpZJN+Vs`E-};bX
z622Ccd%xvzsID;h;IUz1s4m~gOirMouaU3V9h+IThCHjZV?vCo>=+0xb6LA%Rgd0(
zPO4fg$c$den#yCqlnFMl1*l+<PSu{x%}subZj`BF9rU{5sh%sk7d!WtF9+L2dgI5n
zLTKi3Ex#a>GgZ+lGfyTSMCtChmgJq>FLj51Vvi*}d!Oh++%Ap!;H9_TDvt79(YZ_9
zQ~G6@O5N#CeCdcTi^k9@2HAgT9~Ry4+`<f1*@{1U3g=YsEq8l=vq61hEf)b|{b}Bz
z*jO;B3Z5=SLJU*fB&D*N9!f+DMBHtraj1|sLIR6g!IC%ga()%gcS&fQg)y5&I460@
z&~AS!S)5V<YcQ<sEzc5ny}fXf<pcpasAp0BFVBOk7o569LBiDaUFVD?kWj_iMOCJe
z8y+U7pD9SpVE{Y1_2b~Za0qc>a(fBOiOLHuuiiQS&92Pe!8U7`po`x1lc1jze|nUr
z*o~<F?#=WJYB0-dd5ESbJ@GeZ^I!mXc&8FxWYETHpIQ?0u!`Tc-p5ABQ0TnDy7p}A
zD87yFD-(h$m@mQ+|Aof0e0fAKs-QBh^|nhiHK+5g=z#CzwR_@4nsBjzyLCnWYKTn6
z3w8W|hxiTT48tp4En6(YrTv<$f}AG;O~=r}=JL9+b{G||dzvTxy3&M`9~X{iEkWCa
zPlsRB2|gz4cGOAyY2zJ(vZAvvid_9=?Jktz$8HT)nSHIyr|!2Wf61L(w@xgUIs(lu
z3KHVk&CQL72njzvAfx4J7I3obRTDM)3nB>Bj4|7~Fj+b#5^yOjwZinnsKZ&snb6)6
zj?U6r1%6iY8m5`2d<9aB?tNlBW>C+G+0cvC?u1uY=)OGM^Q`n3(p+kN{}bW`xcne2
zx=+5CZO0OZ(D~Wn^%CJltK}2glLa!xcE#sRgU)i{rwD`igssZ0P7L#e)f^KS5-Hf;
z((C-_gLJ1*B%z3td3WlT!kljU96WHH=3k9GTf6gPX#_etx-nHQ#DAxgPakqxlm;}t
z>`d3BrT@vDp`Jc@zZu6@vTaOb7C^ZoUXf7Q5>|A-74E7m9?|wj_QePRkq2YF9R?%w
zu+x%%DHFswpX>LfL5!Dbpj=G#-#2YwT#w^P+_Ws-DAs(<Ex@Wu6rn0EE!nSk_u0<W
zQgWsghx`1>Fdw|!v``;SemvsBXn4gm@Y;^C^CxFW4L9QixuY+#OuaCnoEOXB1Kyou
z`No|BuOnRt?y}Tg^o`4uV%Z<hlOgn+&e0jQ@YbLu7X>gFj5u3-dh7hoA#P0SibnAJ
ztip(<=ZS1otp0Ozme`8sm+guoPym#9M{esK<%pr)m4|aeqV}L1;xE?iq@_f$q7bix
z!{Y?$iB-NGzbs({Xz!xM$z_qcBZig}-SPnUj+8lBnzudjsRA^HI9CJ<4YmALoB||W
zc(%qZQdBy2;T49p1LJ0F=GR3|mD#R1f!wbBpXydD**AYuBfI)#Nz<e?Sq>a@R9bJL
zVTlvb8U}{3rcmX(j`UxxXV6zV4m}5mU8GlDM}V0gHdmMUT%Mng>~nbl6V4Elzvsj~
zu?wykFF7RmvEPBVSV;S!XXEV3p^j^w0Kd!1VcFI8`mH{%L~vcw0blum-`$OYUaL-t
z=$>wAD85lc*Oe<V<l=_`h>b0({o|7#K%Xmz$eaSH8lTg!fZqjB<<_bq`>gF0?E(Zd
z79F?)8wuYl{Zl0!$+CO8NO?q>7l;RDp__A5^fdZ=p;<5DEPn0KQmnqk3UE9_UMZia
z0f-g7w<CrYN{=5QwGDwh89}f9(;@S!qH!x4b#((4b#k;=_NEH_J~=R7!d~xN7%S;Y
zU)tZm-m&BJ-eda9fr=cQ&P)39-hq*J#o%J@H~Gfg$y_>QzlM|LU$`Cp9?gz1CxX&^
zE$suMkoL||qO@+KMg8?#ywyCPzTqPMW~Aaa4eS`M2Dyw-v-N?v=8bK2pN+IQ-=xiH
zS~I~%N+408nf@*;1Dh1oDEE+`D%BvC^u6f3KHH8@P6j=d_uX*<9-DH*GIa5%g>Y1<
z1r-Y1NZ%yf#j3pJ%9SZ#g*J@pUEPi3OgtQpUe)W669r|R2K2a0)b^tn8I%(ar#*`k
zR!Ng7+nYXh7>K>v2kZqjc6%)us1<C8B2jBN<Mpw;v#_wRbo*OnHIR2NyqfoZGj*$U
zx|^DXGxzm`8@oyEv*A=IvJR)Ve3F9dsb@<Czx@>Z!D5SEboW5MJ=qkd0)P`%qt7-2
z%{)K|j49&z2lRU+D@;1z7WuesnF{DRk!jNWrBQ54nart)^)@O+rr>$Xc&?YM-`sDj
z^Sx<whnZ4XNeLIg=9B-J@FdH4bMa18<!4uSe*RVNw|LPMQZ!B-H&`Q79TCN(#rYmp
z)$ArAj^FnSU=^C}jx19Zix~9&SC~E*2QX4D-L=VrS9yGq1Gvy^r>K9YgHgL{z%nLF
zz}aWlV+b=0pO%%e0e`W<X`sGrvC@0Yltv8tCR&Y@f`a{KyYloc^D|VSX$~d$)o|1M
zs~BowImE<_4C+{VaXcexql2X?t=G-%zO&9EbK+}1c7LAy`2+UwsHMo#!&?C{GQ!d8
zv5gUWnZpPA`T$BD0y(c~Z$givdEJ=G6}wS`BWb|!!#(;_2L`_wL|Z$#lG_-A1v~C3
zWRB;1$@Rd*%nU2q{Ag@lmIVM>&Yc%Qee~oQRR~+GpN!x1zS@NX?RpYC+Kb&t@#s41
zK`izBjt*a(Cc>T;N<`Ty=y&#eWB@>mUH50Wdvt;Qh*%1CIA}zx0C);{wI*U#z+>uI
zWWERaud^i#1nyJ4`kw!1@}O5k0I4LXP&EVYxjzHcwKD~BfrsO8ooV^ZI}x<YjVXb#
ziB*dMpA4cSKlpleeZBsjw?nSRX2jgdiQAA~dFLrQz{F&3T*XOi?YLOjXE$vX>5rr5
zy;**0mfIp*`eF~zC5lX1giHgj-?m<^_=>#$^TiR12jFAtn=f_=LSNYYc-a?83d@;<
z%c5)K_5xI!=l>biLSkOYdAXSspq}<V-mgAly4lT}1>UoKR#pPYykI8b@Q4T*fC_)I
z5ygw;+pT5*@cre==oU%-3E=~}{(6EBwV-_n(A7uE+j<gI$pzFN$dTSKz4*6&0w!K2
zCL}YzQ&iLUzew}PeeeIyPE=Vb^2=2{#d07wEayLM9=n>_-!%Jd)#o@$w7%jU931qL
z@B%_2YC@b{!ljd-fMJn#2R@w9Tj({`A$SX}PjvmC*ViBt3ViUJAOQ<zzvhXkFs$RO
zcbsPm6-fn%OoMpFdq<m_!QLzH6-BIzB`wIf4M|r0FS984)ye399p566sQkIo<||*q
zT$h57<ey5!qGCSzMLDnTgQI^k;Zv9)v2FhK$MP~Qw^6;^dN=Ha$lD)G@{!@6rReBP
z>OZCdlVBZKn{))2M2G4DGvlBtzV;edN#TS68b18d;Aw>@^<2gomjm>HK;iwj7woaY
zep=-m@izw?^!f?+*bqe@OB1&FK{2tiyaX*qOMdk$?U?}+?F&H3o^>&AoPOz-K!~cR
zY)pN2u@d;`Z`|ymoZ1d>AFVO#35^F^-F~TPm*}d_7V<@ra)qu2+^VWIn4>*;{cJ19
zP+D0TxBYZ6%gxvhr8;X=8=%9RSErj#8D4xP#?k(F=1aE@%C;cp>Fi3a{<;8(h~`=j
zK=P@bszS&}NOCc><VySVB&_5Nrvf}nty*4!{4!6m>Nbh-M&C~kkK+{+oMy`*4y~bB
zO7Ea{ufxv=MnJDd0^+loMoN*(?K0E0=5pML#H{B>Jp-y`)=3aRZzL;Cntc9p7QX%z
z1Xz`K`(G=0f42Mk0)BCbH52U#yE~-`ieVdb?Ng%^t{NVNJ{NucvryG_trKy5d`J@t
z3_*IJ4Y=kFpmDQu&iZGj#0PL1k4dj_R`rPz7~D-`oZ)@6B)aK=<>Mvyis$79co%58
z9Ns0v`H#d=(sIMX!tzx!geXHfMq?%Tu;bFbfrh|&G*jT$7(60|jSx6Rf!>(cPC`bM
zUa>aj%*>3Enwn5AV2c>xAGSAW1Z)XReRd0IZaHXTZxM1NeP;;1pW*%!yX;93)cyC5
zE^3!n#+%7X!e5Deg)`}YnwztRSnNqTwL5`Lcx(3wXdp&NYreg?**I)iC4<QrX21_`
zzh8trmfX%AL3VE&Rs*;c{-(fyHasF6Xa3$Y6tANtjzX>R8}?!o4S!O1Y1q(+gkmfW
z2f7X8e}%T#TA^Qsgp7_SS7F*VUh_`9OD#(bZrbLfusNhE2sEkLFOAXgnIqIYEwZv|
z0O}jRR=A|5rbZ#;sE_8`{Y3bQDMN!A!YVo*78>Jv{&aFn>{aCC<oq*)NC5X7!44q6
zcB@Y;kX$H3AtKp$k*Ev}Smvwp)P>&1EBdX_D)y;VK&`<WO~{vF>xH^@c6OA5XH(}p
z6S5eRJp9!BmLax<-*A(%vw3L5cawo*$T!EaX-7IIWb1_LK3i^LDJ5CYrutc+3-F!K
zVHR$?Z3mZJfYiaHzQ5~({1Y*WS;NVcE7~qTv%j`n?UMkup!&GtpG)GN3$NC<D&BW|
z*_3QT>SH`P+LKeVK2<r%2&nrwxDMOUsB7!N`yjqb^8ML}uq)s%@@XeRlZ51-l*UI0
zmSF|Ri}crU3NiRNinskG5OEelCJWi0)CN!t=gA>eI&HYH*MYt$eGGt5S@rbyvj&eR
zA%g%zEP>g0@_wP#P62-TE6ogRRUS<LWI~iEEO^d=vg1=rDDaDaW;dcWgv6)mKW*a*
zJD8H92Tj1e&8J2e%>TE0Py5Q4koIqWCSFFE>Ct~$$Nw28r7f0!Mt1O(5<E@(-=h}5
zxPZRA$9&R6&>teronLtWc?rg+>E%rSpQh*ntBc10Y-fZ2I-UQS;J@#|>-}F@{C9%l
zao^KE@}~xqc&O$7_g>@wgO!$z1EJ2Rf6G<}_`+`J|K2S6e?<1*Ww$Lo87{z${y(kg
zpSYeR`<?dRqkR7>vj0l<|A$lXVYNNRkNax;Qw+9X<(;+=n~{vCkel&T4!HRTbRHQ$
zeyf7jR{CGX08WtXaXaYTlmUK8gxPlbZ#fcDA>&;C6i{yEB}SVa=$tcgoJ$2<Q3mWZ
z{3S5S!(|8b<ODHM>v*mI&>_5oEV!tVAfyk4L1YqLhR~EXl)7ly#ys~UVr%q|&om0W
z#=46mJ$w8ukkk#m%iXe!u&P@1x4O4A;)=Sb@4QQ!mnWrC<`3T)MzWP0{<ip=ZX)vP
zwB+`DrtZfcG{xX>?1HA1yE`e{+p<wSb6I^oZFF`0(UcWnHAd}BcTh{*WfD;Q{y1(c
z-D?)pdnhg(|2dm!rK;ocwqd=DtMO{}@gWRorDwW#<|H~^>aEK-4bS%r564wd`ofFU
zEkQ`({0TAH`%^vT4*e0lrF^-t!DRVsL-||Xfs!B2(mypN44i~l3kMFm4I<6{=Z1Zt
z*Y+Ui!$GW0gVDm*=PmcIU!7XlE&FBA)nLEl2=Fv}(<rQ4Y8djo;MIB`xv5+hT3waH
zst}6e6{f;K*!s-~S7TCQNXzBX#>8qIj*Cgxz~jzkyrV7E{G^fctm|JJ8q4w!W(n5#
zI})?%p_Tc>L+{YX1WLP`fSu!?MzhaLHg#<3q+8srNdvtGiWj*LwLUEe6c#|M7o6Kw
z<K|&CBpuD_E+*egWA#2wKOBrE=-kcImN+*6g7>;b#vN<ADMeSj>ij;2+S89C_iN#o
zMzf`>-_2iih?{s=E+2h}+!eDBBatw11v3jb<>m1z@#ha6n`2y9A>+ic6#fjFKa*S~
zPqCPkrkWDm_aKh+_4DQZpR?R6p=Zz^rot32X3qz3uru3ddhL3@?USetbq3fx=AfPZ
zp;4#OQ;U6M8%@yq{nj`&J!di_X_Gr_ecPAnFsHp9-==i$qjFU9dcI3Wo=3_CyGfHs
z`@t1YA<uE}Zpp31=uMY?oiC>sy>U#(sCeh}B~jj>h7}vjy(v{&Fg|uP|2;3}+%5jm
zy4oKnkZ}|oKb%}moHk^7J!a0ajeyJhbx%-CKKmG&<=${WdVS=dH<c9$KXotrv|p$h
z<U^jjmvt+LzV)i5P$QmAdj8e!J<;!9qS*rFnKU^+-Su%UqnC49Rd}VhU+Cg>x-2i|
zSr6p_r>B!zXLv2{VbyA8Z#})u1|`)gE0kM^YYl42JK98}$RKKO9N!JA#iv^#$1{{c
zy5;a5(p_QA7=p>0eoS-S%H;N<(~t4gJ*%?51^t-k3pzMu*m$||;YUDZQw5V_BY659
zp|Oj(TD{RiVT1bd<K~gpqeDZ+@pt=-g2Ug7wU0FmUmpUO{P!M9FW%OTOnSXqJipkz
zfBGiy(Cf_Kz^SGd(9DaD>vm7%@H-wp^aOlq9IZ4Qxi|~jlj*VJwIn<jxNMp_pM{iv
zSXx3{hZ~Bna}LZTb{}$8%bh7X#v~y|G3#PN3ga}ky={$XM`Ap`>hCj)`*`?iS-<if
zQGdm5*gCe4`CRD{g;a~U!jIiZ3wLpg{})?l71rh#{P{Xo++B(kcPj)+(4s|(6+&?<
zP#l8&ihGJX6xZSo!L_)%26sqs$)^9^eV)D8%Y=)(oH_5AIWzP9%-*sQugfYo8IbCO
zj-HLEZi@tx^lfZB$GTS5Z8qGm_pGVibb6U<4{##yLNSfw_x6dxhxtk2^sfV%+&$Xp
zap;q?Z`XgK)Vx-!4rFbWSfb7Wl{Tl!RPIh6UqAz1ck;p@&qyge@5D?aE(wN>BwuJI
zUiF;mwHt3OOY+$Gl175YiA_md`{$VeJofupWbHXD%se_z#Y5+BqTfYtZ?jGDiP?0M
z&si7uv2F?;3F8)em018<Or!e8HLy#g`qRlbX99H^4yj8?D^@>nk4)LJQgPQ=4@6TX
z4)&+^o2m+y-5F7iKIp<L*<8o6F-lw3KH4pVHxSXh5S^A$C77H31-vAf7uT{)rKhyU
zXhqoA+Wlb?rbNX4L;>>MH;*qic+rk~LlUQh-I15zv%j(*gb#0?5St0OVPWPv3GqDa
zGr4}cJi~J9hkES?H(V_A`ysE5mRo(WEi-UUSIYGmT|!JAs#mv4UFg3q5(iiy3x13`
zkn!B>Y;v2~`W4tIS2vL-N}AHeUDW_~eiY>AyB-U{eHLAzb_?oWWgu}qbxPG!U8qMj
zR$+sYLl#H`YN~Y*1j~0bT&d7-$XW$!_uI@})|<;8SAoyXWbIIYzuX+;N?-L=y59r*
zb~na@5<b|!&6pyWws`oN&ENbR*|Rq8zFa{j4rLD>JqhpR_hWeqrmy*6Jb%PR#@j&I
zH+&jY%=j$HE-m^c%QYRF&?{V808eYe7+R)YcK_%+^!^%;;28IwzlF!^P&ZP4l~S72
zz4uJkkjF+{SlM6}O;ZQEE3&Se$N7<GcwS*>q?^Zfh-a8Nd|TUGRLL-P%DdgodY`NS
zBmzHMM1nVllh|e|6N^ZhOJTM5)6U5bnsX!8ZQKq02h<gh74nxNf@S&np1PcuT`3`R
zPjgt7ZCb^_buv&MJ?g@h!pv^(v}Su%xX*|F(N#XG8evGsQPewL%PA{mLs2{!>2=o)
zlNsII+uzp~JOvIEkr<`-Cc{ny;t?&z=%UQsB<_XoQB(q$Vve4E91B9^{oEhup7d*r
zu)StG>O6$)4SSi!V7d+f>Zv(I_ty3!kL}72jx_-BR~)X6&Y?8NML$}uzz^#tMen!*
zMK?8iqzs}(b9nT-liP`G=0^6u*F%t4lg*kOF`HYi(8@)4k@4%1xfx`zD@^olKLxV1
z-zs_pPrD|Ql_RFXy4waOR3OD&CWLLHD=1zlRbxb{Z6G|wQAQD*wCEaB7z}5(9-z|B
zN4I4K)igSjYGcHGp^y>DsAOn0&hmUZSjDx|6-@uh5Xo$BXOPmBnlF{6dfe=(yPMqo
zBTV8y&(gkZvtkpE+g_c|xyHobU34)E_quxmL}=|H?<B;Govk@%Y10FO#QJ`rAB9v!
zAbZj;yQ>b@6+O_zHdJkLJS4MZD>;=LX`)Pb|5UjQi710+Nli&Fd*+#-youiNtcPgE
z8>{EK$0CGZA>5s9ShmK_j?mwiy{>((0Dap;wvp=vgXP}~g@(dGHO=uUi)XD`-6hjB
z6(>501+I1|J6or*+Jp2}18%DpZF%yC#M7Hq?v0e7K49x_sUfB|kg*ZZoAzcEy)h4f
zc5Q9^8>N05xRiNcWmAi!a;Z{0iKD%1iYVw2>FMm3yxH2ySkk`)nn!Per+sq;1L|o(
z(RLw;CjAq{l^<H(2$9ZqrmkiIEVWIhHavAL1j*qIkiH&bA6~G!`{XMqulu9IJk3(@
zc+tmYXws9x^a&si2wAFIQ*4dKvhS~Q>DP)hTn$MdwI^$+=C+9W^|j4`Vdds^Wqait
zKhLACPexzUFKW8?Kcb~$xb=uSjVmuZY@l|&by~g;*tiBO?u#zm^|D>@*iY%cR!M1z
z<9G@y%`myFE488&`Nal_idqcy7cH!_K=VgkX2i4JhL05*_iK|9$CHynO81B^3sO0j
z@4sGh1s?Td(6UoGu7tfiY5CDLMo}1W?8r8N69RDW9JMBN%P<HN#Br|tw);xiQ0x^f
z>8i8nwP9$+VAy_Cly=0%rVsw7!K3tTWc_)18=}P;Jq*#m8gTe&9-9HJjBv-E(%()#
z=@<3_+T30odaEmjda->dbob{xT4rd|DHEB`TK`9*9HUI0aI)Kv^c9CeVZ6WQa^D(r
z#$1;QK&0=H)<HADL!Wf*mKS{%f@XGRrrwrfSFEy|>gF`7VIyn1`LeWi?Hbbq1$&PJ
z=I>ADD=nX|dgR9AfiF{Mx>n=T@ZbZK6OdNC^A2o!_YC}aH;hbS)%EYm04f6^j0xe4
zSL0h>e0UwEtVGh6gZ2V{5U5;dk@p{+wj<@7YzV|3Gb#tc^vv9+QM?h&Cd<{C$yLT=
zv((9Gl!Z|uG*;NVjqh9@3pf0B8Gw@p_?agKJR=!<#=mzpjmdZg0Q{aUpqs+ZhhJ|B
zKorSf<17;C6r>Zy#TtKh+siiv9(0U@uO%=nH;COSBy`3Y_M!=?VZhTUcHks;H}|>t
zayb=C+UB#bg_rX`tZ;Tt$$~zW-u*f|4A>Scfc5mgYy`iP5+LLr@F)nRp}c8F`Uctx
zZyLLsi7|5?MDSuE8S_`M;<Vp(jhYTQqJ(GkVY_%SPWbB9qoIoK@(4rPo@IC;E`FSZ
zDgr;;bDf=)wfKTw6m%{Asl($PA{y(3g8{Gu+cGd#Dynr?1p%BkBoH+Dv@zMbLE}S&
zmeKI-a+Nj5Z?RE!g~tiR@-}OUS~DrCXJZOSLAB}Ko9~0|qYFrx;>?ycCXRU?8W_oS
zM|_=6C2R}W$U-Ge=JPG*y<d(f<UOs){+%gnmdOb}G_0fpE;|iyvWJJpC|7^+x474m
zvrVRNrLk_$7nyc7rAoEax9R4aIa{=2%MvMD5K3smF2&9|tR%VE4C|&S!tNhdX&q+4
z92y>Z95B}nLS22tD~+#T#B+Y-3mA`hOuKqS0bu-~c(=l$uDHwi=i`JiJ!f7t=&G*0
z`*`wYwOLWTN_%4Md8FE>DIt}6bn4QHYiXf1Be3Td(MCUdj_SH)olmQ_3Q?~D(c#rt
zchK2Ov#-kvd$B&mb1H5Lfeu6q8w%B(6J}qvT5>*L8eL?n_b5;w1o@;+E4?gK;Xax&
z@l}xaZKS<DO`-dm+lF-=x-}J22<|}M$=-kfkpI$16`nOux~H}cA-H-U=Wx%F`G>zR
zD_iUlsxdr__<PRx>F*6T5`A)$T)J=a{Y*l-DS8Z}NQf0)6OYA@!`eJzCbH*S-C6sF
zkEK-2EjgCN@b5Dkx&T+NioJDYVk;w>OL^{SBE;1g$W`dC=Jhm_Cqfk!C9;T4UxpFO
z)}DE72CerDok886qc&vB-aFW*(`X<!<`^*K6>dbE0Im#UOBeD$kLr3pYG!td*O*+?
zJOXTr=#ogT{Z3XdX5UelrUT#JOV43G1yQIP%tQ1_<-Q^OPjsyT<o%CVbaeYK7zTo6
zVv8bn^jpvjyb3YJ|Gq5$Ll@pkDAfD#uJ^Wd76IV8g(dQWB-GxDPBDKi$df{Pj6i1t
zXL4CN#7=FNEd+u`6zs%zYRR+>0>A*~)C(?{=%tmSvud}KuMJ#P$xb%|q5s=n7vE^-
zP_tacQFnOETJ^XNfOXBa=;5Ouo1E)VslEqAupJ0j1o;#S(d(mAs=ZWUU^&Q!NFCkK
z#kXy*3Xj1W+XZ**D1#i8z$Ozr7(XsWx#b>5#u}XfG+wy&z46-t12k71@iUK=#z=#7
zY>EGV%UI@DMwTbo3MSYVde3}jN4T6?(o7(ZZZtCQV*|1r(-fhm@bE==sKAfTREz)l
z6no$)f-6G3=uBDcYyi>#KvILf9VG>jb<<)(6$7_rnITzdm%cKw)Mj#dUi4ehl-?Ct
zEJoW@7|@7N1=n02ginD(-Nloc8Ie!Jtrtz|0p@ie;k>hjU_m)~q+`Ur@yyBX8M@2J
zZqOtIVkx)hIAimWK=ysLhH7f?XL@5Jw-9Y!8ln4JsRnKg7A+AF?pqvo(Y){h7*kw*
zT!78~yQ{6OkWUU?z6GACb;yyq%kr^FH&-KVWX)3M+(Y4LT+%Qb?*IKd>w>F@VT=)x
z9(2C7gGcXfuxGe={8s{pu#~=CGbV>Mh_ZE4&G@u<Cb`a$v5)gOL><e^^l0<vyz93J
zE1J##{4LTSQO_V9uCurAcgo2FQfz}!-j=B=GjQ24=WuIL<>_0|rm7MkyPuXJF2Hu!
zH2R(JmTBac$TV&yn}z_8sT-=97L(eazw7|5-25=0%GRF}B3P)AX6=h(XP{qVoI>7R
zV*OhLdk)_#Dd}%rO1X|`VOW8ODfYRUva`0a=t#lY!KGQNkZ&foG4_6)iw^@xkee@X
zChunuxlvl9#O0#$oTcK+DZC1Ldwgy?VZ*PgT{aaX>Bwp5ltxpCBSrSLB6Id`jMSd2
zbLk5bQDu7u#Zyu-%jLic(TiuZiz=j#ONsWGuTG!os1s(bb*OHREX{5F@TO(1Ad?fS
zWU+7{XT8I2lA)Hes%GCDw2l&<hgg+E-yL8<%TBCUKl}70bCl=ueb*cv)9Ru;fu1y9
z`ME94`220YGFV<tD$W#krpP2w+1(IdZ~cr+(?s61W?zXf6n2U`{4?H2xBhB*y7VCK
z6u<Fb+JpiTKAJ2Yt+DR00R5pq^<0>0*D^e}%wQmQ9H$UVzpUfU6mQ{!9k0WXuCjpy
zx>CJbh+pz9zi8V>HxK4Ong&-$J4L|C)giB=uir3;AM<KgXbM|GO?XaP<hQyD?3S5r
zG|B1RNHKKbZvZ>%t+(dADo;~hN)YPwvG@^N<I(<VWR*U_5!c6Y`}CR9RqMVk3zf^t
z&-@k^(}(IOvJVw#eca@c0fpaV(qrQt6o(qsGF!%&hV^+d`E{^+SL4_R&$4n4H@fI?
zn@Or!QqniR2ZG*MX2O#$j-LB@)Na}BOdpZ<r^HEmy!qKNTKALuG$2*zl8TCy6Wj3z
zS=>h!{GmMC$XqWQuHf49)adh^%D;|(MTu<2*n2{L7I@Sd-}>z|!W1~(`%5*+ePK-Q
zax(baB-`rkAOc!U5*^tw<_CT4t4tL*t0UhWNwy%Yn=A11Nr6Q%b+VL0PLp=-l#NyP
z1y^vRhbA9+aMJ?3SSCxUl2+=9=aOz0T~4T>r+3$?=CeTV)Ig8`2B3V##0l=0JHK&4
zY^KvVd-!+Fz;^FBB)PqE)$I*4*Fh9^3kd_`^4hoto!>kW4F|IzCwX-L)we|_0?!Dr
z@lIb%rTb#!?>wcg(nfaA{vb>l5b1Rkwc2HKZyo&JzB#M!{)Wg(`6ch)b>%dE$-JMp
z0#h;c*r~4062}5tjNF+P6v?70-BZf)!2&$k``=<nKeZ{yG^1fahZqr~`?Gt+`!zC8
zZv6nd7FkE?>-w&o48lR9{fL<sONI~hWgV9bUd(|YeJ~qNW^Laz(@P$oUOFn*m5@*|
zmztB;F&%IwHeqeoHX<}-<QpYLMSW({<X0<hD3lc)fqkpWpOGnz`?lUV%|5e<{`=9_
zVkR5W7>tN!)4e0E<SHAH(H5%022no|^KA9!y4BX0eVj*OfPS6vy38Dx(&&nO+#u`r
z!BISH)mHr)`dL|zXn6oL%Q>IQ(hRO)ZMIwJpsJwQoU)znv1kvk#i%Eh*tT4?!skGL
z)D6K=u(uc>hmWN%WyVE<vYA$(wnf!1`t%JjA112RS0^YOC?R({c0Vm%O{Jaft2^a8
z#rwA|`BgMf_L+4zJCSkJoMmW|)mvvBeDJ0(`3b#I7DXbw$HgAC*OFp&QcgoWb+^6m
zjK8FgW)piI)bXkqt9U<n6~=Zq#*C+BN$u&Kf8$_8n9zXnSnaDu^KPO{<nh6+qi#$;
zsi5;MG-vHj6s2ye3`W1Xg)~|}4K~14`~c+`X9(x6^qFo{=tZgwTA-DOUWFH&DMm1C
zx0Q}%45-!92it1fg`4QaKaywW7k<1H-#HByy^J1Vx4?tGNzKYCcK;6IEwmo6;=joB
zjy{^~_~&x0xN14_rBCNub5<X}l>i}M@q>Pi!}&?UwMO?edN~IbJh#ct^!HWY`pF_e
zXK3|{oFtmE#Ee_E+5P#*?>&aCTqaM=j>#>)=v3U_W!AA-Wb$t485VHWtsj04dcUi~
zMTwovMsM<fYE0KjfD_ZGrlmd;#O4VZv=S@Po*Ke#_K{WsY~|67^sH3bh&elg*y*JH
zJf^_CPHxA`7|lN}fu-H*6Q^FK9YxyMT3P?tQrQrtxlB&)84|D)_FWeMj=4w_cs?it
zvJ1msI<uK{5@k6_9Q!4Nj}LlqIamG0=9ZW>G5nO<q&t9@8061<I(;}w)&T6q_*{`z
z5mzDXEK)FQJV54PaZzV(m}{vh)^8o2)J5J^dS9nyi>SR;{S*MkW@ckZxGvt!wH=D@
z$3$7P*2S)=$-SiV^Ao$ZR#$4dk<t{D-!LuA8-xXfD6KfNA_znzV-447d{L)pbkA_Y
z%O7`>Qh94Cf~_k^n#LJ?HAI`mHHWgn9ZSKV{gB7ewB8Qw80({C4c}157sD67>=bZ<
zGOzSzHC8I}^HIr;Req??@X_W>=2s;s>Z6Yv;>wbgyHJIaxwJAd*^r`k_w#G@;76I`
zrW#D!u!a1e){#lcGrpX(*D00pIH(lXJ^ap>2d=;8*y+NtU|!DgVc#J_?@<1i(Kwc+
zA<p_!K+Efbc2FA$7}A5DhN0VJN>NaL6p+zRLU*<!IzjuMEyErsOk60_J9pS2Z?Du+
z75o{+g5~5F211J492}Jc&dsVju@#;A=ph-&ZN@xuTsOpoi7uTRQ};6P0za&yU!G{q
z-=6w-ldGG#qru0FmcwC_V{_eeR=ycWF2&KLr<zozG#T+Jah0f(*fA~SOE<~_v?6u;
z1r#8!NML?Sfue+5OO~xu{EkOTai8S@q49@`W;M|)eaarQU87(1r2;|QkXF_|Chfds
zo~=~c#Fz(7Dvo?AbJ+ztdSl~92}sI%%eSTDE_MZnL%-g{$Oqd_HAyh3O6w|?f_)Br
znZECM$kOLYZJE2<6$r2&lBca(;AA<RX5g(UlPu|>L(v!!@9oVqthf*~zX<yt7v>9o
z`GghTz)rvnu|07k`!mVaW!#D1Q$rg|)lCFW{0NKp2Sv7B@L(j!FpME&&yKV+{*S}f
z&G-VZcm9E(EyL;xGtwDxBBiS@E`c;+gKP_;vmf?TPJSw48FQ*Tw1cTM0qsJd%akD^
z?ff8=)oC28TPO-CDnh^3tfkw0Ky;;MN$0z@3#B(_q<o>^3;I$V%w*GjFfr-tDyFFt
zknh1C_@x-5v|!<ZO24H4>tcDIH7iIkotHzCrHldu(L3I~4v|NW@@HG>T8eeIMH4S1
zU<@^XTUoUZjm90h&b{9ux<&1nxRE$gT$9NQd);PBvh-Z4z2Gw{ZEybl><!4v6>lhI
zvHa!p*!cKqOZy-iO`mqHT-LWNF&WTxGt06SNkUknduL^nr14YpghC?;ydpmS5#oDI
zv)t<z@+G;pnp=BORm!S_Fy9a#2%amL&a<{d3o*^^7Yd!@+>RKmnq3)l#C-nSUUPFM
znAc~0K5XY<_>=CGkfumMz6+Z`hP0@fwj+5EkmN;2s~gvSr6LVwx5ny0@XUC8G*zG=
zD2<6XZzqnT@F&=~7dn-fRKY$}x!5|uP!^mb`<=uXdFfSHM%hgPyQV`M6AT!)9Uo>y
z#5*+Y7^383^Y^}T6d8QYsvA79-<&W3{^nwZ-=@<?MPM$vh^9wj{5ZUy#9{@;H?5<|
zP;(L08$FCde*dEaAc2<*uL`#M2sa71OPH4s?46-c0sjgZeR0{PpX1ivTn?BO#|Eo5
z3$-dL8DAEyIJAvw$@=qR%eZEt@|HnG;~2JKHe`0Xm%42vb`;4667Wit0j(Q~H!h1#
z=7PV>r6f*v2;i1(!}6pYGE(+^t*%l+W;6nsN{};^;OUdGTGdm6xwh^6aTt=y-K?3i
zmEftXi#mYES;go`vQ1OD=`)_q<;9<I*zk7D9V$Tm88+0?-RVgT)Zk(`$!1Z%<uS9^
zp)6rqQ~vbn)6g(~_H`)LM<<X+xPRR9dsKE7?`_IGyWH3*4z6FJUw#dL1D--7O~}5j
zyvbc3y$H>>*O^B0(`@LLy1RGtF(cZ0DKh>Ng6GBF%Gs6#&uExKQj0ONL{;r-rcre)
zb&ljBd>ZmmTY_ig)+B26#<vfPwo?XK`LG_}>+~Q&&{w%^%RF$a=w(;9NbNb<WCnet
zqM(trVUM+&owp=Eb+@soD!ft#TG{7JHX0Wej7Z@hYMp!-Q`fzuN(Z&EC<{BC@nHdM
zGv(D<B_S+`#x+yjSV}6c@0~@ShE={}3>U<k*bcXnxLlX+7ZOUof1kb0n0rl$%k>{}
z1gpzJZuOFpvgosb7(Y5d5Q#!(Oq&j)4UD&q?HGI30s#lT*<5vw*2Nz#siGlsIQk~O
z>7Ke!0iGGyj7aU_uyU8bRpi{qOENfYxceX@%i5PZWs|{Xs$;s(7G<}f%x1OI2W7ZG
zQGd~)vXb?;ovoD5-Cvt&!H&uO3H!ce^`++rhG?3BOcdP&my;b@5FOY+WhP2{Jtqpn
zzXsq2vKV#ICt))p-hA7*b8JlTFsyw6o{ez`T79W8>D!r^T<h{|=Q{<54Ls+xX#(4^
zFwOK%#4PVg<xBDEXOwY$cJjv~d}MImLzARFsp{^E;F=~~_KxdieDg6QjE!|5Uq$%g
zh+EQ#nVBYCcM}}4o<V&wC*`~PmwkW1ZE_BNOg7I~)Cs4vlqCNRw*is&7``cS`xMn5
zUr~UY*`zz3&0I#oat+(`wfu`7^;cs*qyP}2%~+%bo8)6L`3d8LK#)^y%EonC-P=1<
zwVANW5Bt1@AOmI7<g!h^yd^SirAsn>VaV6<$vVBiSa!O0`dTITb+tq~IB+&iUCtTj
z)U1G4Swa60fW7L;BAB+|EaiH}fN`(iJkhf=*UTSQD9dByd2Js*qf$~B5<h8>s_f4{
zSl+F;=swm_ODniUgC?>1Td(OEg32YMj#=YgXE9uGU;vo07Cds)$8CyQOEK$4YeS`_
z;g!nta>GR{V_Fbl_pQ3=50{D2mIpy2Diq0yli9>r;At8M8f|WDWhlz>2vl^{XR8W3
zmw{J;cP$U7xi^2qIP;|`D@5xYi;U|wYLs4T0_31LP?xQP6gt$YjdIznI5DIM8rj|K
zan0V$Mc}WBe!+wHiN~pd#Y*?x>>ZyFGhY8J3sSV<!%4603cWR~I_x31Srb<gPE-0v
zgO2ys_twbne?-WjSu$e)xbiJoEZc{;l=T2q(iZN07sqjUUQEkp1xV+0{JZbbcHQ&D
zK2_U2e1UlrA)WIdI$Ws9UuOoJv~L7_TaUv6<jf}}0oB4Z$t$JtJ@r#ZW9CexxXaVS
zU*jZAIa_TG3zdmt_or%da~1F^YkQh2H1L)s9yr$8(=_}Rf-Y7jUUZ3<iOxLg>xZ+4
zcqXuxB~nEveb}v}Lwr~L&hT|Zc9YW*$bIs&9jLFfMozOmg<SM3tH}tn9tws{=jS37
z2gSy)!LTdrvelYb(mU>*>3>YuY1}c@#c#sZ_gIHj_o@aotxrB%)}oPHgR^RIZNm=E
z6J=iW>!?DF4Y|vhO8QH?fRG)N&h$PJCp>M)!MQRNUOLPMy7ajGP>oqCf9X+h>sg2T
z0*=Zx{)?+<;w;BrO=##U4HKc=(|gh(Hfe0UiL0VEnX5A}TKE0`pvnQ0lWp^UCY07O
zjNMC5M+|G2fRp~31vl0{j^m%2&u7uqT(i?SgEI5BrnchHp>I!9voT2c#aCXEc0)`i
zMy<&{6teHwk%j+8=k88>8r!%#X4kb8q=n7B{V%s4=S-%ILxy-My9vh$R2N+kW719m
z$|eGPGLeWFsE%ca@h?^G2vFIoLK~C&c=Wat=vVswxrT{m&t?BbgiW{pUo3d#j@Gqp
zilKorwDGMcQ<u$DF;fUEe^j)Z08AuN2vkdQ_ra4HU1^JB=e3@oqY^g`L6z5$usX0f
z#7N74k;E=7Y+?v(49gBeZ{3dEbNMUApWXjL4HGb=1xW&q$-v<@DPLN59E%RYs5k|{
zJeG-(gNjj;S);`Xz~IOo#0pG`=JA`KZz&#l>6m2GZPNAstQ2$|U2=F$goIH0vbT5q
zF;v+#KW$hzHq8NyR~H@sHxf4(by+yM5do^@&sy08fqD#a4WcD}%C;1k%rBQGSAqYt
zA~;ur{f!dgSm7lmoWt%hdYTP2gQ?;anuyhIsyG`0BPU#~8n$2aUrr2R^G5)O1PrsD
zo(g3g?I39mO`lHN9qqhNX|5M(c31b0z4X0fcVh?lza$f8ld1u^Gv~qzZl64?ofx97
zWLhi2Uk<!Gy<KZWKjc5;L4gV`vojjKxL5w?^_h@(`L3163#JgMH-V+SI~O<7%8{B_
zJ#N}J{vpn;s0n^2GOR*n<GZiU?X8-UCQ`dAPzbt7=M#3c8Y`BEnmTnY(^LQoQ9f-q
zzDacTp;iANXpGgFFytzIXBw+QqgA?z2)yISQ+KbjThQ2qw^PoI({j=mzx`P0d#Y^E
zp770MDa&x{YxXYnmOr^|h{-`P{aL7(l<?fks_S?P0ICa$DGMh{X3u>6C13g*xmEao
zyEFW{hdiX+yRN;{36JYXu8aTvlYhNFufH;zblp9_Vec^5OR*v0Gm!s@4+h_7GGys;
zY#}^{ROZD^ev*KDP;^vEcXN8g`}9wH*0z}_%Jwfi49+i3?ZTO%50yiz%KGByVn;%B
zOq|=a6%$^FJ$i2-m+9RdJy}gP>342B5+KC(48`BeBn}uKO4fnMxA{Vk!P7e_|9ct2
zq`g#~|KDZEqTFG|fjiq4lS&j0kDTx9&S?BUtLc&OfV1t!Ljo5Ww*DZD1?Zg<-Dx77
z(#GzHRe{MIc3D#2YVS4O8g|syp#q;n^qUotkolJH4|?k<5W2z(+NXIXMUsy@3FA`~
z_9Wn^Fw!b`kK)8`A`6@3t`maR{jW#cp9RW3lZWOfMnoB&t0W8X^i%pB#|<RxwS|4x
z4t|yb$Jyof|6ZR>!nKxVgQXge)A;L8_+WcfVlX@fg*Y$N$xYJRvp&X%vqf~p8anok
z;lw53bQbuO{z`Var_H)Pm{`NzAl?eJM%6-5IU%pYKW37Kkt)JpvPp-Kr>W(;p9bbC
z8Vn&jw*~qCvw`~4vUHT7XQJlzo8y^N>EI3L?&Th1Uwc%!`~SO@F7~kixQ@}9ZL`Tu
zZah0>G||a7XMX0t3tMH{h{0UGySza`tM&M2g_PuKa<P)n9-gaHS1YTg<mGJ<o?Bx;
zI~ZAXvy8X{R|A=#j+>iLe%@z+jkXiX4W6ypat^c%viihnId$ydxp%CrDPq;Io}d~^
zcW%0D9sj`9A**o%YS*cy_b986O<&8iBWmkqm>h7`9J_Bj9+Rt_Ud^6@!ViTj!aCKT
zJ|zdP;t_y-);&DMfP^aw^S9(hq2dL2<NlDZ3UsJ+BHS8fWYRN|mEN82RMC2+!3-R1
zXDL}C03;@04GXw8vM6M*JS@T&{lGgOE{<w0Rovms3LcMOfRVfOlM9gwvjH6(xv5&#
z-@oCzVNv?4yEn4NheJ4b-97OiG@fgw64hA8nlh&qn8_k^1PbMIi0?qP;s@T1*oI?F
zT?T@J3sNa;?7|e?))_Qb93=+uVWd~R&4v*_j{6e))b>zkx;i5Qr}2m`dpA{T9b+|N
zQ&r>Q)2$i?RLK~P$6*S6*npQTJLY+pj%Omujc2lBK#OhPm|oQzpP6T~$mN>6%!=6&
zR?*P7IKLGLq=9shrhAjulM0jDAQbj8+;94WKo52AIp+CD;^wY<d66r?{}fSAeCsms
zF2qT5)YXe5M*stw1oFhc)Yn3hi`BhbZoNh$`9G`enODA(tA_8er$e28;v_eJe>QuP
z8e%I<_(Yt&`w8e!BDF{DVxGMV+>9=l9}>5zXJx$=hg)L;>@+=D2*7aRsGi8}0DOWC
zZd*hqiSC|?5rE9O0q2ViB5-`j<E66`Arljm^JpPTunN5Rg`z|z2_X-G4ih`P$@S&9
zsyD|uwdPdQpD+QR23y~eo>jtY6N5lan;z#gq@Pd!*vz}zyo6eETzL1muS_;9Z~bR(
zu)6~h?|mWWZTa#@CTp#(#RAn}cjNbIPZR~@u|T;M{M0Ru;Ge>USBt1Dg!35cmzAf{
zp=SX=)U!$8b#C}->;X9cW!vaco2og*a<I~})w8KREH*ZFL7f=<UEHTUL-t1wJPr#Q
zvWl+a`ZHsyof|>oKohlZ>$Q=G#8ISFo4Me|fU;1G!$flj*rkyL_eWdoN}5?L*!yDH
zlp$M^sw80b>((lZ(--k2pQ*Ts?r;z4%M^hQ&}23#HPkyFJmH)C2kxUCm?=E^rh5EP
z1?bS%&VdMy8$VP6I<R8^SP0x`EVQUl{O!U1!pw2li~Uv-@F*jK<U0*22+AAxD)}6^
zN=E<=*t5vQr*#KdwCTm(MlCbSXLM+x(m=6DngxaGBCJm)FTtQt0EiR_x!R{T4I`(j
z^JYTy{Te@viZjS%DX_h52+Yg^Msx*Eo#W;+wNKF@=9?fbwEs1t{WKx>T3d^U$W^b$
znQ%+~3IZ{TT2LM*W}A!%pG_We|2|oo`Q~gNC$)Dw=`-y+ROhU3eH0SBx#XzhNs@Cy
z5x9DhsoUap*Zv;CWqNQI!-%-s_`O7e+5p2cs5as;%n=`=1h6p^^7SP2TU9@hTVUa}
zuP~7$)7dQ3V&9r`vymq=Lhe)cb*(W0_@y9ljGK^ut*kZohy5*+u(*MVAxnCm#Ed?j
zF^6?VG4+=b@z$+&C3StdE#o1*JznY4#_{@xcL<NX(GXl(il=Ry@Nr5dU<Iz~UG=z^
zzKU>7sPvz~7Rui|(wsyYYd2NuQx<Lya0U9+fz>rQOKis!kKC`<S&AfHs77hDa0xK(
z!;h_MY>C?fY_=ITNEW!h8-w;MJwkV@*Sc+c`C8RhIP$9dBNq?R8V&*2JDmaXJxgJ)
z>gDy`#Jei*bRY`~$3qqfT6<tYY`F{Rm>J0)=JG@V)eIG%O0}@^Jf{s$^a*|(GWf^E
zC(5@<N}>b)7!Km1T9uW2W3ajTC6x_qI`UsVJPXsuCgHj;A4^Sr*mxl1X5!4%V5QZ(
z4X`BeB9D%Hm@7sHsL2#~h2kG4d*kXo><5j7qiilt1U^4MM@L8ZC@Usc76hwl?y=HF
z=mzl2ekkpdQn_Y%gBO4|L}UW_P98=`lg+b{DJEKFzlGislRLc64?*7kj&(S?VP}BI
zrjwOREZ{o!%oE*ZDVm(x(_vTyO)%c&?5F*HJDW{gu_Y-_0(OMFz{}hU;-^8YnsVe!
z_lLw7W}m(|hHUHW({B<#Q3AO+IkA0xeH9cH!5oDgx#Ne@KnH$Q8S{<fH3C()VbR|G
z`nTR7AEAdR-pOL4%Gqln9g&Fm&6kwRhKRhit~`uIa@~R|9SAZHm|tZb*p{3IC4&18
zHE-SEcX8+d9IxF`YGWM@4UM;7k%n8&*}{0v4cK(q?%|Q*D*{m+{2IG`5?l*(7cYB8
zxg^Dx-z49^cl;x#8(vB$)(%7rl1dP&yqVJ6($tMb^*w-OtTxR9*BqIe=@=HiptmKv
zfLR5DwYVh){X)EH>!z`Cub-?2nXY(oiTflj<aATnsnQ$7tH#A#_V6T-56;SILfEJ&
zJg6CFf=7ci=Q_ESRxn*<qZLvXZ?`G0QChSF=FaQ9$oxi+)v<XrUvzq}&BPx#VmRep
zTMI9Fyj@7&l0TsZNA5)$n>L$XRXd98m_)3Xsy0gb@1T%yH}_m-O9o+Mh2y3nlHspF
zh?jog=ocG)BN!p0r#-_#-8?2{EN{uj@w-DFf}a9Dm>{M>^JWdNP$vRCM+U?cm36`O
zT(DF-6-5>nCzX4vko>T23st<3npeS<VXo<Z2%M90yl9X_d3lslh6g1^e0YG5&A0nW
zZf<TmZBJK?!x^%L?Yb2rK{KRiE5h%NMw!&Da+#K1t=M`hSMPm!Akqo^1tIWW8XUkg
z(VaAxDQLv%%z2zW=)G)J_kr9jk;KUPV&1#Q8y#g?SyRb`TyV=7vXtbEiGv3H61ybT
zPeWcyGO-8{vpG`z^}1t31&S983=C%|ZH&o$rLl<fP9&Z;MDtg?&cGVbVdNPB_;JG>
zmW3+6^s2eOGj`WQsBAKS_cv2++WIePnJc-W>ZeC+Na^p`*ECqBBm*3Xz~}kqGS!p}
z3P-F<@km}q^dQ~@+TWJZ;8RFq$$%z`YwGFAh`2ggiD3Iu^4n#962)#gt#pLRy+T8S
zovgGhP-uMFQ<)+LliOa?K)&O~?}~mHp=_JW_%XfoT$2>iQ%UdlZD+4-p3edEXpXBh
z)4Zk%d`G5CK1nEz=Gv8*fweAcvaQ%mS-$Kn)dkS$(M-*3aF;Ibq<8$7!+8-+YTyqs
z?w-{n6FhQPCR=<>xeEz^1qX@LdXl?wWJNE;-J9->8}T}7`^{&~zeSdksxYXXVaNsr
zgU4<qRiG~0w&50kZB&FX-1Yr7y<$Uv+(JSms2O|R>=mv5<cB7x8D9Uex+t+<5NZ?}
z8Bi~{X<TfpE$ktnkR369iuZcr)2PWh2baQ^_EI+K-fJCvFPqL1A~2>XvT+9Uo^=@k
z3F0G_k&jhlQ1BdjJ;`y^$V70@$`;k-UqsrlRnw1hUCVVS)yXTA_DwE@&wRYu)s{oQ
z799iv<!pDr1n~Vd6tXSp)(l8Iel~DnzN9Ph$L6lpP|D|!RUsySj?;Zut@jE__a?WW
zDN_6nKVl1=FOfG#VB#B{b;ckqwXp)Wp!c#eN>DiriyvN__6n(;#)OJ|*%P(v91zC?
z|5H}|6A8ASbzuR|9?!aQ8(sEed2;kF*M?BSB?}4MAR5w#%7K={fOQ$Q?~;u;$G%ok
zDTHYBXTwGc*YGZBzN_mMz*$E(HKY|9b8-??(^8-9&~VCYHGPU=L_XJHtw=s!-a7Oo
z0&fk-ubl@v?0tGmd587J<&MD8twKuw7o|ok*M{$ZXiqo@U`Ziitam!zzBf@?l5!Ws
zs3CAA#&s+2+~*gN+vwMY<W(l^=1J`My9+P~18PN&$o$SjLIO5yaqImb8D`tvL1__6
z45?kF`*Uy43I!yPRAS+d8;8#;Jax)aXD7+8iBzL~Z;(+UfAXz{FtnBGv#sfVc~&Cm
zl_PiH$(>&C00V-r)Ejv|q&^akl1e|Fd51VIQv}S4t~LJp5P1>vI_u1M09We~xN^gk
zB`SfPO-ex#j+#ca*-L6f)P`cNK_s7aZ@9z4!Wf1n^}3TFL93pqx9-y(w)X+5#O0ye
zT09~k`_+!b%^I`G1w{WyDyL<gI1sFbp0KNTbRH7&numOBUB;=oM?Q#3i%W_T6PBXd
z#~(%IE5(;n>|zjJn2m^0qE0Y&bcH1iCrx`n{@L?JzAdL<IVNzm$a96q+R=eZ^7N($
z1#GWL0)a99iN1b}2EQfuudy7%d{;usjRg&&?2j2g3?W?glHI@-UKnQ2F6@y#1y+ka
zHGhlf3}`kPiWP0#Jcvv?DOZ4>%Y_%ROe(CeT%LFO`1y%rzLm=<rA((P<8e|TRW{-w
zqj%i>5w*&Gvh<EBe7C3VtY0Yh=p)Ahde$$PuDIX(zDp#zlziE*6EF?L+u(eB>?Q6a
z^lV@b3wPPztYhl+{%VqAH(dOxEIC@@vAzE!_i(^IPM5KB#DuOQoe#SEkW~O`z7Uf|
zYL?HuyJ!ZH6|7!b6fj(D(%eMcdvOiWXeM6wH3Rs6JkFxBZ(QP=i_mKGp-t2!qz6TF
zk%g<;N9UvEjxUM4$kn;lM$NusA#f*0qZx{OK^Y$R8ttAsJ9V+SNNd&Jgwf|~!f|*7
zf%jo-Q8odvmrf&7Cp#xp+%izpw`eoUhnTsdKY!{vJX4doqJ}fP;DJa21wD8jtERLk
zXfv&V4yb28svzPrAp26PO+3<rOCuhdChR~F?Qu6S2XrVXnxKql&n8X6a`O1H<wzJ>
zX%;Y$o1_~m9<U)0CSG9kaV79G{>;5g<SlABV&D<4`I2HWV&svqWHTb!D4K2Rh*n*p
z^ezvLHqA^YxXJ`gX4Hw{_Uob8hpJ(W%jiepqNIAujMPKI)4UA;U*TSEzghP6xY4w4
zS|G@X9`STf2K<MJClAb26e?O)&Otq8vY0!UzrX*l^z`o{ia$|vgybq{qwkN1rQ79W
zcnHn1<$@&|w9;(DnbU9^_p8}~_9CXsdlI@rj#MUN>D@jKejba9-MxJz7^<p`hqR$G
z!qx9sR5v#34MQcpFoXf(8%_f_es0!~$HMxl7MasP1<JyhvHVB6Z9v`i4C)r16%XFU
zf5$6<`O%<6`1`TWZ+OFL7ku45uhFn$O~T$|{V-a90Bq@I7d(=K@-)k5<J}qcUUZxt
z)aRMIJeDUv5R_}=5;`6rCb#eV-Y!`QZQvb7-el>%@rGdeUk?q%N%iTP&c_Dq%r;9z
zCATqQ!Kv!5>+uG0?GpyMmUI=Ccfqy;0cB<?emQ)-Z-ie0NcTPhD`;nrGcR?4kozsi
zDV0B*#??L>FY+h^Z9*SW%F)i5QXPKm7fgYmGxNA{=QWu%e#@RI_%Qs{g$A)#^O!f+
zKtL+13SR<q5VKKWP)4U@wkwwvW;=}es!C+Tf#Z6Ka0HK)`Uox#d0l#Bde(?#@|W(f
z(kSJX8bKqSzc(>SH*b6h$Jw+$hPp`1Xj?d6{j<4enAz7<J4RCZ`z<}!({u<19OKT!
zeo`<BnDEb>Pe>H@Klf7{b52TQL`1}1`?W~G(Ds{Kojp_{rwo<Iaah8;F%V+~X=KuU
z$=A>yqUD&vCWO~*`+IXFo&6BAv#%c<5i|`n{kWtlj+fcf7Ynj+5@(15;5mjwvfDv~
z$Uf)0jObYxSn*_*JpVj*Igm>he#p!9uim;^mDsg!hf-HbSh(G*Y<a^7<W9)S1^m8c
z%hL+C>BJkUBR0aa#m!<!rC&e>f!yI2JpO6&zwoz35S!YA_s+@$tGE7OYj*TJThon#
z@<Y@dJ@8Hw>pK8#J*m*Qt=4ifzHQvHm1ZLQlk&4O#D5|37R@a!mcJ=QFNZ~)chIFQ
zxe&F%wYpU%IikCn053dZ;vlc<6AWo;cR>R9Y*f!puV2^3CbNf;DArgY|5;u>>BRkC
z(I6e-JF>)`Ln>{a&dOG&K97g((A>&9Tk5Bq9e$DH#R#-jDFx@tpChfZWOs1xAIq)J
z!z9|ys@;30%K<zsmPS44&7m$?&zAEJzY3ajFso4p);xu5ka%I1SFx;<xW~?_br^D}
zb!F63Ch^vC26v77;dHy?Bg=XFv4;pzIsvXdo8L+Z-jPX}q=EKz3Ri*d$)Lm}>wV=`
zDsu*;@Hu)q0}iD}phWD0(MJARw?An&j=3%fQ<AX@uJL%P?72(sO0w&NYx1{eRhQ46
zPUEGx<JuVIwGT8#b?7UNy)KWN2-p{iJ^6`45^=>ue;n0PD=9cmxr=imRUxrvnlbYt
z{i?qI_V-H>%39jp0A$W(ZX=%eaHF{Vkh4_nTLF$ROY02lrz}qpp>`mB<~b6}@<r&A
zyt2}!^~go-nla1`YS+qDgpwI#@*PR;JBNDU^+R#j1Fxl|RxJ#KIa2zeIOruvwX^0q
zU(9at|1@jGRm8Wq6_(ZRsd}GanYD~G*FW%v2y4anH$vl1{$3DOlq#*tX*CohQ&?Ud
zJ1&pVu)GyNv60tGd_X2gTo}2tUM+2s1p#(!*rPWtb88`<%QMgAxtC1hjx`VzxBSe*
zG?8(Z(HNB3i?~WkoO&HTD7kWTec+(m4oX@~GE&K`mdO#_<a1yDV}~q!9ELF!Vlw>~
zB(mNtTdOhnIZb{I=!%C6b@-adXq-LXEn*eAhnX!jVWuwctH|hMsdp9>2O&t^E4g8-
z&FU`k4Sydm)KeOsup9PZxcS28WbJ*bZMeS`cMpoATj)R}JQc|x39&t&*O{0F%4>j}
zp5X4>-H$QDl#|YXuS2q&q4~|_liq0{vg51txK$?o4?)y;K!+~>RS}nAniHDvN1nna
zfHA1en;ZuRC+An_S9INKcG(CL;>YXR3L|bgf?fap!kKh_R3*io0PH-A+qujsJ&-NR
zrfA$4FZM1$KaYi8c+M<b?{Yt%)Fg1AkN<|mjzyZnHRSD{op4U6l@r+zlN1Lc)S$U9
z-+VdUKZyY(T^z$yvdclXrQRlzMHFNC(1DS?EA6=(&2xA5W9_Fd$?lsNmOz@M_zAZ*
z$j=6Da`HUs6R|;fzJdhcZmszlGEH<V3d^Kg!?k!4k(HFR*G~At{zEtalYkq|*Wk`6
zzux;}%3O^<PDKW}d?i|?;eWrd<vN)MC9r)#p-}nRwaeZoGAShvYbjc&3gu807&JMd
zQ1&lJvn08L)gfWH{d*~ZGslpEJ%LKV8?}>YUs%g;WMDb@z|vjxVWcn2O0=F|-Ox=g
zi`Vb>ZqeGf&&}AL5Q4>p=HlH4xYHhFkFV^V@BR`Et<TI`&3lB*a<TJ`7!Q#<^a0=f
zj;a807%;Z#fWac~rM$OFy;f)ST#dkckcLC+#jnkMY(F&lho_OkoBeC<Gw0L#z4?2k
zhv`G-hv|DPecv!zFJIa3r5ido8)~ahqHU*7tZk>~Dc{iak>yL4{%dZJ(*vi?=NG1(
zr_H`H4|vqc8Im3DeytD(KZhuX59V8Q?d>6MJr%xwa|87mtKCVp4;ORpXT98iTy7V}
z9AGfouj}174ml(4DzDa%^;OkJ-%JiqDPTUUvu&)akJY^TzKgG@+a8{&dR;RBc+ju?
z{T+pQboH*WI?7_C&mVVbH=-lZo(^gKk+-j5?QK6jwqTI2s-#u+J0jC|8Z>_-ibE2!
z=A_^i2OEFJ(sH*+9VRX~1@7}itH{q)1X$|VxBFFkn{{pGtQU@Q3woHX<dA8;<GN+-
zx5lOd_AAe@SO0`GJIu*JwJ{G{4Y6GH*~%WZRrF^CxNPf+VvN83X}Iyju7X;?5B%%&
z-zy~urhdsa7u465y%a9Zv58|Vnk?cg5i0Vxxh%!h{!>_{^GEI8$}1tbwmqhLtJ1`}
zO{FDc;lJ6#DKDKuhLY*xyh@?R3fWUbmBj*c&PB_8$v-*;=J^!F@ut>##cw!Ebw2jm
z3||!K)`_!Ce5f%=aN3mQj#?_20IMu&M|90-4;&I(e+{C*>;+#sz5iMz5T58k&GeT@
zzE{!3+guTAA)f6^X)Kgyfyntjr`r4WBIi>=CiO%twGd8!K1wX{Lil!T7M%kj9+1vs
zeN6Y(aPp;FQhmquU(q|}$%fUBSMU7oS>Zm#^;3=08kS$5@s4-=P9u$4)4P&RR(BW2
zb5asTyXVMmKw-~E1T_)RFYW!_2IHLDU;nPz0or@ga}sdw`QBXkxu&_T!M;zS1j{Nb
z<%Sk@<?h(z3UedA6cI-BZB2Nw-XbJ!=Ub&Rj2H9IIevMz$Q`Y~iQ5$MZbij;g>~Md
zX9Ze4VJCQtAGG0m%3j$H`%psd{I|Ntucmy3{#7<^pReV59?F?%mZ0BvzU;KN3%8M<
zmsB6*jGn(<w`T&^%j6#v60vH$qV+-GyB#kkl&beQ2Y7QLEGj4?t!?jN8aK~*@1bj#
z5#)Pb$Nux>le-5CZ;{SGU%YwrAICCpp~EHPFakdNCp?!ajLtc3tn7X{!Nf=usN8;P
z<V`>t?VHu#XbGDax$t`ffT`T7g({rq4q*Yhy)I4?A{+~k-Z(&eFC9fgK7@XEa~ATR
zAtepq@h<j)J2tJ8VnDF4)69DYqxrF%{kj)nlzr>tbu}<npa^|z%3pQPr#~Iz8a9o~
zikMp9-{3eG_bxZU6O{Xk!!3gxl-vk#Nc*j@0Q9Iji}2V_NIGXAml5(1r57_uw1E!~
z1}6p?FV)*Q-XW}IWL{?@D{0ls6*1!9-+xeSQ#TL^W%^r^cWnNa!|(-@Qj*LdANk-?
zL=u}Sha5+^2?Yta^inAC--NE{CTGW?TfUR~n~fOBFcEiLgZetTxznYMDIw1D6|_^6
zu?JCwONN-;V`<<Rt~E%a(dml8*Wv!PB9R|;uVQ<2PQH=!gho$nMYh7=T$t?VyjL|v
z#)EmsOjeRLjXyUfsg0`NQ#*_@`1ihVGa;(@cHZ?~zwB3x>|1X`bM=wtpB3k^(d-)j
z#hFt*bJq8^y;#{ls&x3OJ~Ow4&xLDHgp+iZ1{p~ex*5H@Ldc4*r>g3>_o##VKD5xq
z_Q1(dYGR+sH&XJ^ume%c<>tL`0e4#Q3oM!D2zIJO4w>Fb+0PG`I%oPZI%hWOA6W8{
zf(+bPr1Wj0-o!!ghf6EWr4@1E3CiAjXESrk-m=|ESncruELLw;u{On1%yZf9i|wSa
z3Lc{_b{WtGGRf**sdnoJdC`jYK!?sg)GJ|C@`Lt~@GscsP3=vSi#v8vx<aKqscoe<
z6i@Ic9jA}@xc!Cnw8$I^+VkEd5MA2Yi$BPa0iVOSp67hqIyiQU_D2{Yrd2vM7AnK3
zf>?7EM#7lT^U}K<WUK1DGZ-LlqDKGU17X6`+<;M&oZI*rbhv|R-g=FFm>+m_?yfy8
zlV60;dLZKZoC|-o(#Ro0bI-$ph~LRQ8f(<l=Sud*gw_)MsHW}AwJl+Q*ow<KYjXvv
zUXr}NRXl#*K$r8lRq_5$pLc>bYZzDX3*`VcU|YD;X|1PqTP8JY`rTV`tl@Ll;$gDY
z<pwjhCADt$p94w&3RFG9A*{eTvmih9s%&6G^oX45dNHDy{XVz<%&kbfU3|fySItx3
zEvR?3<+P36<~sAuMa4K=CvY`h@X?XprM;7fRi(&#(26esxooBgts-|wKOnN58yV#O
z!-l%!LDGn;=V20_%LZ{#mg`?w=n?akPRJM$zPBd7@fmRYDP_1NDNTe17&sq(gdrqE
zXBwYW4)TX&NWi3l&%dMsfApKX%Z&H4#43?}+Iqd(m=`bN#81c&(W3BITBsm=+*w`G
zO@Sv|gLo#ngaZ%N!wg7kVx&#F_0)bU1>R&a<#*E!sBC_L)FRndrVdqqKmcKOew=;(
zu~MIJbN~5V2a?sQFr=Qbo&9)8QHz?3F6{YvP?cEIa94`Jqw*CXpqNRxVRP~hbsU8~
zCq`k)@M%!hxB&>`I;b0qJNBo~h9pI-u`jAqM#(I1u&Dh8Q#7%H04pZ$mMta;)06U4
zvp(&HAiwl;cGvqadGB#1MIBli=@RaL-QF^+EMaXFs(l;Xqkk{^qV({3jc$EFN*r-b
zdVlIjMr*X(b`)KH+}>%+$OEEBLz+AFMKP*B@MUzz(_AjMMqN$x?-A!z=fR9PWjG>r
zu%H%Z*<Hh3{ztVWzEQ8#Hgwb*@#0>^kYk<4Ubpu#Rw0@<$X?ze4`mhRiJ7YOA_tV-
zeL9+Gn+tz$T+NIz#?Y%?t;_ns!5HzluVWzTMUIOq<>`Gj-5D@W7O9l%dsujBx9WrC
zdZ^RCce28h>mWHyEAYzF?MO!dKgpA3ug&GYsG=`zZ&7nRfafulOw~WQq&909w|zbl
z3=I=7f#e;TBWx7CctJJ;OA&FuxH^q%==#5eCo$#_lMD?^rslh4B<3rU3Rk-bX>}$1
zb`3KBi#L?`+Z|X#w4{ic$d$`ccbLDn{vu-Yy&~{t8yIxvhotWh%G@T%S6q*}$Ul<{
zntZQQCGb~=Eef;eB4J;$%6ps7pICsyC@eAIuf@09aPxf+Z0ocbM7wwhZy<j;A-fhh
zpdKo9-j+RmQCwzY!emj03KsvgeoLU(W?kibFXgEhTwbA=5RRG>eQ!LNS>z^T=v3@|
zzS-+)7n*n!=@@tR2AksB*?MQ=j}3n-n|w@Y6}wmNSh^9a=}X_EFZTVrwEnk^zJqjE
zs99d#B=u|GNxx90VlT|T!E2(^`!hi(sBUm*_D!pr#mDkfAfv(OKoEs{%o)9uX~_hP
z_jRM&uMG3CKuv#g%<asVck2HOH9^Y09x8r{rx1CZmrU+j5_!iR!8e@|a@z%A<hux8
z#o(u?-GsOmS-{Op#@e5NEa;!D-=g=ud={v8EeeIdLcWR7S1~-lYT>=Q&oUK%GAnxL
zAdHbYOm_xeCqJH+C%pFwq1SuNcX<25rl?Vi{gjM$B`e#)HGhT3zGf%(SU8p_Q3CJ0
z^A6_Cn+I1{SAzn|TjX=0Hv}^`3=w@m#EWI9=H=>hW`=6t4Erw{tJZmYW=B;+dLcP3
zZnv$Du*s!iEcn9c#SSDx9dykJVK-e6Lk2aTgkC(o{PARLS?on7xsZY2lDqSQYFf1E
z(UQ|9DAK1{ifxAf8N%c<nx&1eiAVKf11dTybH6Pfs3!3;!sz1@Y!G_PP%@xek*5m1
zP|*xmX<nf=2^GyFw4`%A+jJ7w#n;HqO?W=WoRW{>XH{#EIN!OsiaL>hUXYtF=ahQK
zXC{T?k0?1Kfw!Dsb#@Jg&ioxcNBxNLGe%(Bj&0bnV+VHa-i4hzcZ#M9%-y?p3vwSj
z9%WOdm6a8S4<9aS>A1v9&v0^S8h3VPgF9iu1cZl&8%NBGjt#@GvA?5oEpjGnlts0=
zl~Gk~rbet*w+gD)t7=$QL(K*?P^WP%)NN85^_$j3y{2_g|4q}fL9=?O->j}+gXZ<b
z_S%hVqIRuHuv)SZx161j=IMbHcQ@Q85#oNp9zVDJ1WoHyLHqaK#$W5!A&sY7@b$rc
zKR?{{^~J5rm(l%)uTif-WxUn8F`B>E2+iJYgyzI%?>0fRcbd?8Gg>wi%jWOCNnw~x
z(SnvOX!%y_W@y2DkC)8_-)h|gE#Gf}Rv)~DcRy;0_db3bt%+?uc^mJ4+!7xU+kEtv
zAg>!q`#<{p9el|A;vIbO<-7R!tJVtJw#Fyz-p8k3x4~!Mw87`?Kg5^cevB`^{Rm%u
z_pxBx4xiAHNXxbzJ{9D3?f-V)e}=Z-eTHwo{|euB_!=L4_#OuS_8a0)o~3*V#96xc
z3nq-g#3B7LWne$Fey=&I)GAM+uREsp?v2?42EcLEY9x}FPV@A_@r~<93^vA3KYWEB
zfB6O-ertyxJAH+pI(>nk{`dkNJAa9viNADdOZ*DI((<=%?eJ??W?RAEx_ymKJ-#8f
z7wp{gI}7|)uxsxQ=-&4S{Mqj(^cwgx@fY+N{F}ldCiWiut8u+AvER_&1^W-{l!@By
z1BU&90mD0E;D|05G^#5GkLpJ3iXo%B5xa}!urWR8`R;)c<9iAYAK#0Xy#<Gl>xJPH
z{=~2eJurCupJ@NfmncS}@INoUj3V?tb?^HJ#!MZAanlFVyE#bVtic$^oHGRD=MKfB
z1tT(W^1_jrvS>7>Egpkui-^2jJXW(l2Gf_0&A=JFT%y@#Sf9CUJZ3JRfY~c1V$P~b
zn73vM=B=HoEZ0oM{I$~zoQehOreVpZSy;Srrbt`6pFc*|zQ19`s%6-=U9l-_PgGj6
zDLk+f7XtCH=l}PA?2xH6*%NluxL}E=ap4J&4;(lknt`i*X%wM%2@_WL!GMLGu<z^!
zc*mb7pTL#K-#_}6-K6gqF@We#-$51&sy|>(;X9BZ6NR1ZPtcOTuT1V5q6$4R@CtoD
zc^R4LLH-Gy?>c=yX&IU9O#+aF9@pyKQ(D;YaiZl>HvUOLCi^YQKsHKE9{VZGC@z>b
zbk*TzuiVd(MIMv8mZ;-Gu+Zb?o2GAh)=glAA46gF{egl<&SUncRUd`dx$s`Q-x5WW
zSWR47a`zyk@1)}&WlQ$Aazu`wN8Z1T^Fa=ny<sTE%;}Cbo0eh6PO>RvM{;6Qc-zXA
zD}~r&e?(90vH!vatz6K=qyBQ=J|0WpX(3+^#xi^TP|Vvp3Lfz%h2zMRmNSh5K@fF2
zb~4j7`?v(h0$JekgC+)JAp@ZH9}A-wKgdwR6H3N2<Oc0ebTbG&FC@^5&tfl%1P;#$
z$A&`~4<ZlnBFO9Jr0r_jpDn8C=B2XE*CHlvrI%0$y)+hjo`!j8xLFB(gbYnfnA~(z
z9mP#;4#A?2OV2ny(RtPouS#V*-7C6}+@V^Tt1N3OBee3l^!ICN8kk3azm`f^Zpxv*
zUrY0_#`8P|`MTL)vhn1SD*pI$-thM9p6?u!^Ox>hPN^G5CUQ8shI@V_=3QWaaU(|0
z?~ZZvdgJg(JEWwhAU!=j19{@}tjGqI+rjGzJ>JH{7xQRV7cX8k4w|5j>>0<SZmu6K
z&cFNUYFrFD45yG|g3e(laGB^FavYa}kI}xP2L6?Smx7NP=^T0-PE0!P5-r^$&f&Pb
zJw{CKiFGr^<J!gZc*He!{sD-&a1PT3_ec9r+u%HX490f)16|0$^m90X6c0~PqnGOK
zfzZDkF}&~3*lM*D9+Bta5p@o3k*DDranit3a1A$*)_I@8Q*e(uZKOx^8RA)5H<AA{
zJ!8(mGxjXJm@%h~^w!WP_KcvnVafa5qff#k<`f-A;p1o<(KGfG#d(tAKSBEqq;<C_
z$`8#p4&zD3s`QBAxJ=}2-jp8ycqiDN*p9IydcotM1MYZvBhAMj$?o1rbaBV6OD>o`
zus^>4xD66cpU3TU&PY6a32B#Hao@!Qw=X+k&#L+8KcW+y{QiP>99{1fO7oQy@Mb2Q
zg3lGYK3+2E+rspYr+pgwB+z!+?|1by@eKT~oq_)~N}m}6uG5lfwvB6H1aBCY=A`X`
zH_wyvtuXi&JM;$L)=;xOh@A6a`c?(;y5YY$gL!G(Phm+^FhgiP<QARtI^}cRaje`q
z8<nb;#|tmLh_Y47!~W!U_;ZcWbtlmZLln^6HZW037X;AKpPVi~virV7Ui#5CL2`us
zi2*9fN%K#-Bp5&pr2l~w-aJkbFFDT4z~svaBF8I;#7Pjhm?1|zkk$jphVVJ~oIbal
z;7LO0M1Vbp&;Jukx6VXVTsTtGQZv~T^Z#0ML69=Ng1*q>0TWmVZrip^v<l#3v;!sx
zp%?6g$?FGT__A)WciKw#)0xB;xAtLj3m3A5EXbImfRhV4wcl0cJ7-!d-#oD-a^Wfa
z7HSQY@}1?u%vivY$h$>85sAHsBo84j!)RU9=#kJ9{rM;^J}#K@j0@>_u!#)*QMNRE
zXDl+S{ZDc%)6#z4p54BT`#FEbfHdUiDgE@6F$H)a7Vf9VJ__F}_EXd*o9bw=QHC15
zQ1<sUeq82no`G~e^;@a+3o=sm5%?Raw_nTLgw3rFlz09T3nuna*jHik#{W6LO!q}y
zX1qUH`*zCtL>NAI4w>8V%IMk3sGBa4N3eL?Sj=287=A&XqHu@beXT9AWK+1Np9fdr
z0i3wT&Rpp6e=e-<+_|$D42q|3;rTjqI|6xbkgtW_thGZhf9q&?#-B1g3@XX^2`V?`
zu}KdQ8$B*9;8FvnR*wyWr>KGAs(c=<IGPML*G@5GNtndZerAj)3336C!l^cqP#?%F
zNe@(kAYTi;tUNshk4VLaPjqgkZ6Yt|L&T$liIFu^$hc|@C$F;*h@iYv45kv7F6lgc
zzj)aGxNMk{Fe;yBSiSJ-_Ds)VHsoiYjixH}xL`sUK-#C?1E%&vlOg|dNMU$|II~Q>
z<MgOb3lK%m0pD-_K-yp2tW2-4s6URM+fNQt?x$C<wzd|XpS8x5jVX74{{H*##b{Rv
z2??Tf%slS+tI=4vVH~Ef8Hf`;d&M)XP61)$1Ikm*SCQ|O{tVZmT#w4=4CHAf0uh$v
z3EShF(7#&;1pc)j_q;qvtaxJe{2BP^tIu#??ONP%_kx$TJqES?7Bf2cz_kmP#Hf;~
zUM_H0HxDDnb%JZuNkmb;Mc?-~kerSC^v-a)3{=x;nLZY%>^FbTvZKj<nQd2XfiaIV
z(s*1}<PYovawH!3!ZrF7#!c;mt;=U4$)88t48VO7d-uKlknG`wc_WA6_pd)klDC)W
z<9MG$@_ip4-1G5*@5uugGomYY*smbL;!g2WzR^96c|bOs&Y_U(6W@Ey`IURhOL~9k
z+Y)u3Y?&avOPR>&Rr6R&!+tuKMQY>GZ0Bp|G+HL_XY#esxx{m--b+dkg%R;kyb8(Q
zJ}{8hWBLDmE*x?}XwpS&v|E5GwJP9+S6&c<njE=giwN$6PI2)xA<XE98a|?P5u+Z_
zH}h|AL_G3BIP;-rCPq>^q6EpdKOl$rAvw^*C|c)jCMw~g>AILvbZpc^+Afw9&qMk?
z&^ZKoofytbN?YhXcLb-q;7phU#w`02tM<&owOa|fo?an7J{~=L_9Q!E5_()<#8WH0
z`R1FLHf@?1UGL7FJGw)lt3q$;#z7dh>Q5YS*^baeHxgO&of9ORF9a9a7DEB2GL=Qf
zcPIndo@i{TAlY<vIwDaEMc-NBt8mRz66G-o%BU3j7ALxk!fEzfxMh;s4JDjJN*EW$
zQe3*{d=AL(xe&S}JUBm9pN+>8VByDw)hxudB9Eu^h$XwrJ_-*S$D&2G8A5>QoX4Ez
zet^~&h$DNaw_nTLgk>sWbH=<5%TttbGg)qi$$pB+FYPl&@2>eVrt34U>rtDhc}dlN
z^Li}VAwHK&6bZfUiqH#*@Xys~m|VCV5fLG#P*8;)lWX_5WdIi(aZ3W8f`J9((-e3r
z^0Cn4dGrQj-j-1$^iI%AW%`JjgdTsqGT*q|kY899c*;C_$}s;-u7TiMJ#NCuqen)Q
zA>~n-Vn~>XQ8#JHB99qE;f+F1=Ntu{BERSb=X^8?J$kuWaL_(WLXVFVr9C+iCVJ<6
zUdZ`l6nfXriYBwF(9;^gC(Sp#<1JC|cr8uNFQ?H@G~;x*v>7AkcEyB+eQ}J0-s4P|
zt`&OR509r>Xw#;Rm^+V4(b&jZ0t>wb8^$XWru*(eED5`~NA&R|=T&74Ig&Bt*s=iB
zUV|kXubG5i>^*;C06Zhlp-<l*uzTq&5_K*}x_lWcr%%H6Me}guqBGLmypinU3AdeA
zSkZqN;*Om{n!6j~E*{5-em`K9)l9hGK1WV6Io2fdjHC}ei<T&ZQ>+yVdTD#MNY9lJ
zKVAr-XJzHKndsi-8-!dqK|<LB_q}}a$kP}1ynV5L>SPT4{bwZmkO=f8f#>flre3(=
zb_pAnPDQT)KfpcY1fo*Ospg5b>G|XLgy%F%AgTjY#EE-mXkB0s)5?5vn)DdcrzJld
z9L}8TemYirZ*qK(a&nw{F0~MV*Ev1Pu_mFXPzX0c{?AUT8$$0mk<eR+Ds?L2#aI7>
ziq+o0Ul$z^d57KsijN&iF)_K84;4n!J0rxN5L{mLjMB4cM!ttaXbQ2(=hfN^%|1c?
zuTLTVh<pxJ=qcwTp-KKgXu2EDgxF*3vL0Cdm=mUJh2DSu^MV)@p;oP0#?tnrq@*lu
z6bn5s5_;1NLhqpKHqqrH?ykRql%G6AKT(K0!R(z@J)0}KL?m4%8k!%n*yDmyLye(e
z3gyim5`O%>CEta~V-0Ycc;sunb3ISwf_-ik#x<qvr>Nw76e3z9^qB0c@PGDg*gxcN
zv0l8A3b{Lpm!bbBHxFhX&sYO!2%`!&Ewypvn0e%HYwfEl^e#sp#-eRwFk|`PT+gG&
zT}8O|pM@U#C~7~vrcIk-^ytx|A0F5I{SyLDMgA3fV#4%6n7d^pJf5P^WAe)@+W#7A
zrG)U{7oLnHlPCD*3BS21BG>3KqquY}@iL<BaOs_?bVx7$f=;ER%w06E6{A>*{7?)i
zuXB@FuDRn{4^`-iW}B4POudhqIrYx_ypW$qZn~*9iQqnqqC}DQX^rHQ=9?&~q+_@i
zOO0RMW=`{C&6&}7nhU)?B=j7gSm<#r-k33CMCsXs2M-J(EJM`jt)D1_-btT5q69IX
zoY!~~b}AJ^pD|x^LQ$)iPEOtZKzaxKu*-fGhV}dr9yYu1x2G2#`1s*2G3Bxw9{L30
zkzX+GdHUg@mp|>N<4&Exni-?fwf9#z9_)yKhvckL8ca2Ms#DK-O7}o31cWoKpRQ+x
zv1H$4*`D)r_s|a)0*;{nkWN@IaTsFGohFgzO&<jx{Ld=@@h46r^w43EZ-0{jO!x4>
zO{Yt6vE7ZHUBAKl-SctvUNGWGbjMS^#gn~BATe}>?h#LptQC5yBOXH{otqd72T!Dl
zpK%NM%8t6xY1hygMm^S?nmBZ+cbwU~W;`FKmD9}U<!3eXvFbO;^ljpACkeeIXKb=v
zgsQa_p;xKu8#s2s0Z~aTg8lSKf;o;T3h7AfDK&~2sira3cur3wkVuTX8-Td<V1+Dx
z4EwZc)1z<@4RN!$Go8zqY?KduQ#^4#$Pp8l{E4-@A7AM8>eWjKy_a8p1)qQZx#)+N
z(>bjaq329OZ}3wWda4)^bN7<}!T;4}oF<Zuprty_0E;{xNs`5%*rs!?=Z$I;S<%fT
zUGY<Rotw9EYc6%PO;zx5OBOE3R%`S0;*?aV3O%M)=xOs!6?a-{<H#w6pHBCV*F|lh
z+5Vi;WLG$C8HL{T6;CGgSmf~_PJjII2b`UqpK3q6tdOsT-i!=Fk8gm&j2|@po0l8%
zL&$Bexx|Ji{Z_??+Jr(F1C?wDxKA0Es<FtUk4+qfQ9n+4@jnmb=7roWLNtF(&nVvb
zh2+vl(|_)#z#=bawRudvb3QNB3Oy#*J`2NX5G0RlD791#F`rBAoXq#q(mgUi-kceY
zr>`gkP=sC=Ojy*5gr37=2|aszdoe;)zkdBhiTl;7SB1#SNjy(e=n0Zjz%_a?=>hPH
zxqzk1Ct|>F-{9EV6}WcpEbe>z;GTyU9(t3w^9>}y=Z^>8z6d{d6zgV=$DcjE#UbZi
z@K1IY&Eujd4dy~mITz=@)&}N{9sD;LOV3{%*$^(!i==xJlH!eXK1VQMKqoAnG728H
zR(L>y`F|chcu0A8kJkT7!uh^8-LFe$aA^A)4C?t4R&SmSpLiz{c|o{xHxO6J*}i($
zkH|vL#~}0+@xt?CJ$<2<{kvkKh2wP3_oPB^i6Zn~F%x<eZw7&<QX}-lyFu@Wnm+YC
zSO1%Y9^X?RIG){uakEBY)U;vPV7*j~P9014njLHrMw3P@wL&k$Ip~_BNVG(xC_-<-
z69_$S<~@A)2+=BF$&w`q3=Bj{N=l9$yNp7wN1h2iJ+j?pd(7hraKCsK9_*uV+?h!E
z!!>*?^0;LZ*XHr_t{3m~M)prc6He{7jeHWW<>UWKHw9iQ``A|zEssR^q1?RB&mS!b
zJ&|{u&qQtb$Ms)09`;Xy$(FGQ=bBpHuEuAKS7WCvQEOlFO6WyJMTsVwKmGKR7^EdS
zI$HGS%S}9aiF_{f1{#Fkak?S&VR%YHkI64EKa|{#UE73HkX~}0GC&NtLJt=YzQWTg
z@TBQlk;l}F`?(-pyCU}7Mbik*H__pnsOF<-T`5sxMr#Z}R@6P`b3*>$7=_+75_)Ql
zULFfQ?v%~v;%1`ih*@fLmE7KsjK<U3Oz8D|Y@z4p=LcI`TjRh8xjFuy5PFL99Y7x;
zAGijez?N+bF}%-jST<oePHtWYFB>a_9X<qq60EMfcEMrMd`ubmCnirBfc@uo(TCEF
z#E3Vp(!F4TtTfS6siw)?2Iid|{I@8C9=F<{Wdz-mh=+k>NBnW*ybY#K8H92DI>UC+
z9JuY=4j)@<_#bqHyUjiv-na@2CJ)2VVV$vj^DMYWUm!aaKyk6i3&8bsf1)DvSmcSO
ziIkV3?+x8Al_rsAp2wNfnn8Gp@GAKjmFj)rb)7hLY2i5aT$#veHczjKEVyE6KT`?E
zj;b=s@h01)B=jno2|aor==n9$P#dZ%^o(gXz31vWFWI?c@-#csmySc3%4JZZTuJ=c
ztsPv#{~|F$@0o!{VaL=QYaYKEH($Fd^v(s7&|CWWLXVqf9y@jnXU?1vgE#5*!^;Yp
zh29o}&|`r|&zV>%;>+}$n653iRQwd~3ZeKZ{JrG?&A3mVnEy?Xgh?_F-a_ADCjIBR
z<Kmw1+y#ZkJT&?#im204^-ql3q{zz=Eebu#Lp|T&as3yLhYLK^(et@Q9&c0Q(^Btv
zEw%Q=T<A@E457z1gl!3r#_8qdB?Mk>{hYi;KJKG;j)dL-%-uSIgx)c_AzGoQrT)!(
z9I6ID9Uhk{hIM7$=An0a;%^dpB>F^OGzza3c}%@{pC@v&Kh=oxM@WxaBdWgq*-`IY
z&kH%9{tcn0^jkFyvXjgEk<oav&~wQXp~vAL*JwU{p-0CE2izZ#?AVdui%Ry#(F@iX
zIkG!`|Ndk2`2I5t>-;17{?rb?e)b-IYug4ZS51RY^hLxz3?lmD8aZXxDUUe~rv6~O
zRL{lvpV6t$l$K|ViY=vUCW@NOzXK8bC<HNSfe46o!p=QQ(f!vi@khH4@#l|i(fjxB
z@as2i@Y8p#F?zzEIOX9;wkME|WpZChe_SDvcTLb2S4ilwqbuGQ7AwRUt|2vwCZ_hD
zt9hI;or1>uOf+uSyC=---xU+JacH({&zFIE>D8Q597lOcVbx~8Jm5@7k{b!VWjaES
z-UV*%s?tp8F>@mHOwXye7|C$_oZ%=`q_CK;@~fXd#D$>!CZQ+pd)D7hHBMu^bUi}o
zk-%Z0H=cyvn#UA+Phw9PnS`Dj3%$!E?o7>sh0x1zub!N2gZW!3x|}3(b52)59*9MZ
z6mIZUGK?;%6VG!=$_o~5O2M<qAJJoOzIl@WjL;MD(f#8A6qUL9Na)ddg%4whPfNWx
zwA9)bbD{S*HF_$Z)VAa$^10CKPeN}P3BAAQCeQ<;acpyX6CZ~{@M&tIRBAi$svjq&
zw)9a_$v=Lf`Q<jJ*k{<ESDk4KIyv966MB}Y7oQ$EK9kV9C=4hIJ~p7-%tbZQD)r*9
z#Jm)GPlA6==#gV5#DqVh)BO;h<c-i<p74k`i?i+raN5-or(7K2<a-=`(awlWq+`?k
z$d<79V0%K&nHZ?T%;6X0yjL80rJoZQIO&CZwx~Au<E8lMKA^mK5J3C<5O#<A8IeG_
z=84OJXK~K!Bu=^=!a3jl@Qgl*;G0gO=19zgN)EQ@TSJT|2RlJ5DQ=1<^9ZX%vvep_
z1(St*H3%rvHuaup!syZPYzXS@e{3q}7yUJ9U3FZTJUVM6J<lp5DBU4Rp4e=&0#$2O
z!b?Kvl_jBPhiH0E6G$K@=rrF<Ad$!9bxZGt_Pi>?No-7*I~=dRR!9iFuRDH(^8pTG
z)I(M1>BYrV<It`v^hG4y*Dw-#XG!Rddjg^Nq{uAvHq-O!ES}{IJ_6DF`ks_*vqgXE
zR5OyF!uiY79H}-!rCvPGE%}}hjXsGErH_(3{xd>P@lzC^=#21HxCtkVKVuj}yt$4$
z4~1TSLOvFHXG!SwBcV4G9tlV2Ca};m-8_|gH!CkGUwFxXgoclf-abWqDWh&`2V>ED
zAIzy2pB_bgTA@eh;U$+C7c8NtjJkQ!XEzBwue=p{8jAcUr<j}9MW*ryvy``KzKBZk
zMc8fnASL@7NMT3}ijseE|2c12<f%f>G7Gi?r-)@<ias+uFIZY2qV9x{Fb+plawrJ}
zZW2df*{S9J)AU?w$-|}#CtJivanZfzQ6o)4k41w*t$@x*H9Z<;5PEvYS$IF5n$Tm?
zdr9YJavIciWw0J{`~^M98S+7BqBpi!uR^t&mGSb+|3T#{rE&a%4GBFDF<_0_myrjs
zVaC%vNVrcz@4iyY$4!cjO%O$1@HHtl&OB%e3raCMEdA%^g%jrvL!sARH44288lk5h
zjn5?HGBWc%lh8X8;)t;<^d586%_oJv&~wW0ZZfkHv`<Q&x<NaQgJ`Bn=TpwjBZ8aK
zk?k}3yi)!C3L7kGk=f)EH>dPUI4!6OJ>G8g9n9W?-1;d@RXDQH^S(ktk9-v=^s>gY
zA@t%|=#jDHVV=#XceC<>e6#tX6g7K9ez0@<&`3(=h&slTpFPzu^7AKz-gQOj1@d4l
z{4wWuKsdxqGHP1p(#YuvJ)Q?g3cbgp$bY&o>;&^@oZOU<#fo3TMXcR7AD)qCk&qsU
zI1+(8eM>w!%A)Ba+c`mfp=TN{pB*_9`QulxS{4}ckJkBli%KN{amN?ow}Wu_(jn|W
zXN#CrveBuOPwY(aGs)z<Fp!0wu_on#9|;|bF9Q{EqKJvid96@QkA_0rsLuz7;d`aM
zU*_)zhvj4FoH-VHhGVj%HXN@j&j;63i#(uh@rHGE^y9EQp74pkj5WKLp>mCic<H78
zpaKa!#}hlngy~_4JY|qFct;H7-<5|-O<o-L*}E4gnskbOjTE=&Q>VEmZeGkI#j1>=
z|0bdLiWn*PsR+Ggf8uecS9nt7T<BSNmhymNuDn!!N9Y{sYAqt$1v)o9m)dlgp3$e5
zo@b5b-!`TtA@kgm^F^g*yW(^BB9z{D?<+iL9QhC2$Dc=HC;vswQ=`z+9RE1qT8=YQ
zjUL$)b-(!j8mK)kJ4&JV4?}6X$wH5tZXTrvgFj;Yfl>NbsMNbzc|{(HQ|+s#mP|-W
z<|Wyvuw%x8BjFL3mXni2Mob+9g$*K?RI%{lrkhF$)&+!eQ#H*n(CSEG#t)StZu9e5
zqMp#>61jpW^v3;jLeD_e5e~ZHjP+LYQN2M$EZaI0k%?YNxa()AL80@Llde)s=$*kS
zQz!%sg|)<|(M0UuqH+%mf*_5Y`(&P@&=tS*`Wat#`~(44E{Ktj6ZpHJK9@#k)+|3n
zQ&5GbZ%1AUJ#%W_4-wWRhKyAATcV}ojO6oY48wg8uRRP@nqH>UbDd7tm_pZ>=!1wG
zp701egA=ay*tvf_29D{4a+S;C<yT%p`3j{lW%ej+*t-;aj%>hz^LyYLehQ(tNhI9m
z=9(<<c)E&U#HNwZBVi$)H$(q3_54DBk+Y|gC-@IfcE=<VdWBV?_tWDEy$qytMN%51
z&?C})V1dW|^g?cPzdY_<!q2`z=$SF&bE*^Hb4LE|DgE`#x|rnh8}~er?Qxv=Uit}N
zg~{tmfiu|}dM`tgT;Y|#Lhn5J5a;24<AUhx&;B##sd+v#_0E+Y*+){lvv40hqtG+h
z`wZV-vtzTO6ng(KGWOB?XN4ZW{4DbL;o(6k_+eGWp0PwiGFz&mK&4)o=ZWUe4*zF^
z#m^cK70l#m4R`<`Zk{Oy0;1<m45uxY^f70WF=e5rQu*&9h8bq^h-@4u$E__*)1KR$
zoEELn<Gz4G=+U`VBmKDdGXDs@C=z;O6`|+LLXT{U5Or+awNyS7de@RuYCklhw3G7x
zsLNh__;pLXR;Ccz{roXbxY*(9-5?Tr^x>raQs`x*l6Q1JxMnWR4`DaGF@NoB)M!*4
z6{?oQ;*C=fa?MEyWxjSX3MGplk#}bKQGkRV3n~^-OrCaw2mj$|J>t^?5R>Q&_rMd_
zWxE0sru9d^!M~wLuOHF3@9!8hx({Ye9ftW+M`F>0QCKo&G*(TRgte2WV#Dm2SUr0h
zmQ0_3d6P$B-0+?l*#B4b>eT_g2mXjjvj<|!z9n$>KZdAez9-z=n5hhW!+9^#l@p=o
zpSMDf`{+HdLXSlXi@bpA7vXo6oa&q8RMU5ar>`(i12f}usvqa`ME>qcp{GynH<0s#
z{Q&kWcn~1=Rk*f}2Ww-WBaHlzFnZT{MBm`sJcuFrDT#(b4atXKA4d2w246<?XOt<N
zlvZ<mA9Ctn^D$EB<sA81=*?U^0Q0tvfG1C{zz=}g2TxGvW<4qSVdICGo38PLB$~_J
zx+Jzq1CTQc1Dc*u9OU#RE~r%Uggzu}G|8}WtsNUa9z9Bo>`I0)EQyRCZ5LB|5P1Ng
zXclW}ly{t;T#J{+rMMKAAQ`<BPei16h~vV@cyb(K$%d3^h)+0I2ItqD-0V#idPW0!
zLmAD=_~GMlDs`?;ORb}3F7zfY>V@Oy|0SWffsD5*^d7QN&&W)U`eTwm7IAdlakS1;
zu!K;aPMALg#mW`I3$ML^vQ<lC<kX>vxgCmxbbrw|DuJF4eWAzeDk*N2BnFH$Y%@>C
zvq*j(MZVGV$Rh>vpeoVHJ~-ob7_B~Qf!9jChS!P~#=9TCjSHTKk-*Jg@6reTE(skW
zI1F(KvBYAI^G74}xCEMLYVycb^V*yl^FA}xecG_*VfZ}eG!LUgk$&}CM`;&iap#9f
zT1JwP3ZwU*2Uqb3Ka0Kl*JIJ*i5N4qJH`*}ig}|4W7~pRa9q0zXZP%e@8LrTKXn4}
z=g;BB#Y;%M<cgHbo=9==Mxv7kZF9wy^Oq5Q`W*ZZ9>m36+i_&w3T&S<6${4<#N@%<
zFmZSf%w0YfRwp*WDflR2Q~eQ<MD~o*VHA4?@_KecPc$*5`z>scULKfHu_<)znx>oi
zZ!ngBM(FYPgh#gy_<#1!0=lZB?ZU-sRIm_53)J21tM{wBQujikUg|=<g+g(6DHW=;
z(Be)=LWqYrB*85S-hTi7_nvd+-jkafG(k%4T6<kPGH2wS_ZeL&dp#Ws3#THH?`j}a
zy<vtiGhcd#d;9ZfuzY%ON_Cs9<Gc!X?7<>ipJ@9l<uRv`d7kklvI%F}x+A;T-P(gP
zoAjFv`z6|3BgU(d{+Eq#{S-kVflynS$v)^h1{5iM*6|vGZ04KV`Yk|y&~xVKsbSm;
zdb=X)r|5GP=<VfKOwtU3-g5-KSFoT^8>V|JJ&)W3sz}LzDpSkRmBXvv4eRtC>;_<G
zrFaTzJa?lY{+pyYQG<aDQ|g-8rbWzb-b&-CeHzWS9gHW_8^L#<HqyXRiz&yLO8QKv
z%7s2{`=KSZjS&{yx!2TE+Zn9<1%0`i(?npjJwR6G9&|uuWYv>rCppW&`#>+q^OPrR
zJf$+A_tm}wz0Lz(L&x6FVsg?b6#OltURr*KJ@|9&tn|zVR%lD|nO46{8;pytJrjo<
zeh3aZtQx8xQ4=Sgc`_!>o`}5li3aqR^4Ri0FUWl=*$(b2DL)oS%#UVR>!u_u%$tF?
zzjzsSB9B3}BN=~(*TP}7YvbLIUq#B|*@Qs?+@@5<MGEtrAcNOpyx6S;{GcbuV_Bj~
z-`B=<dD>o{uxS<SzdcbP=&9#S&*SvsWfd(zY~D2dHhDPSfB!k$eC-8z>9PCpUAIn1
z8$S*k;^OglN;>|RmVy6EUWETkOf!N}Dv*#)E#hZK|Ch+)k|cgkN<;Bn+LE7t!>0MM
z*cdw>%cf1lqyfFq;hFpKz&%&vjklh}xG5tLn>~e28lRtGa^l{v?DYb@m&yk{%4dUu
z#zAR0HpLHmT6J?zwpZBG*S-Wj*=6-)c!%{%vfpyforcvN<=rZ&ZVT$G1UdF#k$TrO
zGANBcihxS~ihAE1XtQG6xw8Jr63s{7NL#F7<HkSnA0x?6arSs5Bog`@-^JPcQ2?mk
zKy4F}Q$#<-y{;W1Ze>7kqXT-5pEAu14^<lO3FtY#iu?dS=<Uuvdi&#RKL@>bF6d2S
z0`5J~GX|6yJIURn!!ClMOs@<U?3^l<mNlMU8RNMd1@Ye^8Bg^Lt8y0XnV}E_F+b?3
z=Z?mTaaGCzQoIEiOaY&a>SBoo)fiA3O2M8zSMSdC_|9tz{=EAu+3VY513^#DgFtV0
zD!$Ujvn=TSxnF>uoS<OZ^Ge<<ilKMSpIo;P4?cT4j&5)yA&;PU=pm?fcn#DK55Z$?
z9z*hySq|v&oI<KB=moz=W!u4hCFMs!o>qKZMnGG(IThbd7=#ngYJplu)@1xu!y(lj
zopyFBeDliy6l_eT13&=c@eX7-^+h1)IlXxY9_Tq6<_n}8;L9&bLSoKTy!z%Nc=mxC
zFzBPVu;{Npkonhm%pEZd%Vy5NmgE#{O-aMHREqC^Q~yx^GfKrjNvZf>GC|P&P5=L`
zW2voa>6kfcB%ZqdI%H3oge_?)JU0Pb<L0B7(3kt?@A#?Lm$>KFOYrtb&ti7kc*eCL
zo8feOPL&pZ&^xRO^a7q`PbJ4&(@IQZv@2f&9c_FS=<S>oPqvQpD%i0HgXF7ew2?qc
z{)v1x{ienP>!YZL&g+i^13kw-p$&77u?^C)30dppt2nS|{&)W~{}b{u`M-cq5Eux1
z0zHMN3iNjVwV#7tdxGAp1ii^jz`X-{a>h0EL4c>8Jfme!OPMlMIvr&VNh}o}RmefE
zbudfD)52XT<GC9J@j#d2LL0McEW7DxbB9rCzoQI|RbLy89JhtbY2|VZsk3anqf&;o
zM8=c(j?UiV5<0gV_|D(6Xc)7YKNs`*OEh$c23gm#En^xEVT>%{O<?JTy-EeWZ3~f6
z5{(}TdM{~?-o668<u2&;YD>@?ZF=+^M<`Ic_N=8P$5pVlkf4%;j8(B1^g|ysKe;Jt
z9!^N9ML?-Z*sEEM&~hlkn}y@s-}@nJ)ocPyJeCrQ1$zQLr-xD3l&x`XOnG+D>R(`s
zG>lI@ddmn4%lJ%jSrhT}OAp}iV-7=g19^<UL#S$p;Mn@L@p#(@5T8BEI0@P^!ECrL
zpkw@(Ric4NsWezrRQgw>{An$K(nj!kTT_s~F$MD$j>9W2-G_Hxd<d~)M_}c&$=DbZ
zgZU$d<Kssk#)7fmW9huPC`wMknuYOL5f_6(eqS5E044mbBry@2346r}30TMfSH;F)
zdGtK2iI2mYgm@G!O2znLgK+!B=Obz2cr1^ZgLQGy*htX(U%~=xi;u;cc{4C~>`=Vh
zt~K6!?`cd;9EZ$ehFkj|GORw(`}Adk-r+9jT~5&ZxeE05P^w!s&&lVfctB6@W&sS<
z9s5mQuF`=Z`M-AVoJIpVWt4sczb{%l$MnP*J4>HJ+X6XT+&FM&d=gjkQLHq*M1F`7
z;Z8kwrq_<210m4*IQmbpC;OpeN`amlPZ>^4t*QdOoqHVsJ$eGYz0;#7hg=z4En?PS
z>V4r-?F*t!AgA!R88onTl;w!~!f@a_R~0T6I4|J&U|}~3;(;#p+$~x+*YpyrMNGz1
zy=*e1i`L9Fy=)q)Amir90Y1T)GYr#d{gNTfDOy0|7f0z`tz*27oGTjB%r(&ji`mH5
z^^`}~YZ#{h(rG#K!5*(C*wedpw2VP#KCTSVTkL|~*f|8f&k1^6_BH4|hslZGR0`;s
zo;*5rW(X!Lr3G0taBJ&ZaCF^c33miN0^T7t2q}UUs`_EoamPbfW9GuKSjspS=m`={
zZy-OL>ss6A{anT#>N|D7+9KvLSNd%7rEDzodavOhp~b0<P)&fWcq8Ch(5uF{Jo&5>
zF=T9Cq_3W9Ku>$+1qVGk3S~ggxnJq@4-DJBl0S{}HWT!=BqMiQG8X1dLzk~#$E#1@
ziPVWdVMFX({1dkTn`7h9<Avw(_|=!A(~IpeveW0tm@yT}lO|yJm!Dz4M<1Yn$M-OC
z*buB<xPTCth*@JsV^pUv@ZHy4(DSX=G4!*KFz=VKSe_7vafAEg#&gfacip>S&`0l~
z=i9Gg((nN&iH*iTiSgK$6pyXx@kssaM||>58@%$-GqfACk-M3o$8atosLN4%*9&@g
zs7J2~^!8L{ZXO+7Gvde{x@r8CWscp}R1@{eIK6touTiBcc_nyKo{`iT?ez`hFQl)g
zZ{dQTfieSe@;~Ua6gj?$fjasl8EZ6Ugnq`lSl3^1hidXz0i3f!tmB)g2T!s-j)gr7
zh5F6gtEX*Wl&>}QMZS3z=<WEmpD)IuIeIU<bM$Di$~ZTan^=1)WgM+2q~V!%XMYFs
zrt-7ofSz4CR;voxrR@(!c-f&CCtkqY9a#7`Ncl=FuNtyxpot(Q!zs|yFiQvY)WhZ&
zIY%;n>UpEl<++kIgqdsS8;4f!8Eg1?RkZn>R~%2!n}_t3v-m#O^z3D6VHYz*!dcA4
zz@G9fKMV2%m=-*Z{_UV=hi+Cn=qbMEBWv}14F0Ycj;M1as?|D@kV4R_;r2+muu_em
z>xUhS&wIUze5UVGUL&6sjR6z0U<j{k?Vj&5Wj<Rw=zc$dmB$L!+0;AMNT9odz_1`|
zBCfgR0vuYKaVNO*NO7lltbNRp`0wpkVZq|*%%_YC4IA7R2zva^cval`L9e{@4<5$7
zFE`y?8NlJQ{Wy6LUTk+aVtyTsEd;WE;^Xmm{6cJ8umB@J{RpkExCGtacm+Rn?~0rm
z(+tdYfBiL#?b!_<wtE&gpLaIq{qP+M=FK6zy@4|$8sd$IA4JbLU&nKI+=@4!dKB3_
zZ(P5=xS(YVJazpw==Z_fc(2`)xbZ*dB6r3VY))Q?E!5wcY1ovq0JDD?iJNb|0DsK+
zhGjhti~rU#?1{)Vp!bRaz3SDf<NDkGgPAG6R2b+TjEk|z7MtEY!kh)XKsi%Qw5&bL
zQx)u0X-8uD6DXPKNgK3Uw^Kg7pqED9LVk#R6!{_oJAqs#fi80`!ETM?$EXoHdD$GK
zu9!prMvKGgf7XWv+8lTj=;>YliH@;7eAWljP)_ya$q$mey?XLt!~+!#diq=idVBfh
zTXl2jO9VZ;>SpPR&&W-nJ(XHB{0p!+famYYo95HPrPGa4>1<ojb70R6Taas4>EM-C
z*wpHmS`0;CC%`ifj^@qLP-d<sjM2y$<7Qg22wTMqjS&<8rqZ}3v(nZuPhSrp4WkTU
z9t~s0DjG}zpL#z8cNweZ@Va!EdF>oJS-Q49YaXBG^O;^fwU*J~rpq`UbfBkIH`@{P
zzO8W3Gpj|?PRX&;b_wHVj>5&)o`>qSY7p)YCzKq@_tKyz)jX^guD<0`Oq~Bc(?Jez
z3a^*y41cR6d$+$SZ!0T#0$$EG=GUz;jDrPcpS;g{zJaKwVT3xyn;Ty~DGsTcht))b
z$Oibb&u2&~BqVCTw=K!89KbWHcQReQDh+z&q`5EdS#E;;zt=<uh|2jp3EhT$fY)Al
z0L$mkG=TR{Vlsg*5gQZYF@4A&bhzhE%o#HZE9TF~riF=EH*+rHMt*}Ix_pJMZ@h-P
z|8o)gzw;Kdr%u6u58uaqS6_zY-^XF`v}yQbKtDWi?Ug8)G7Xal48)CRorwv3`(nw=
z8JIKnTik!m<(Tu+_gI^_0P7dT;P0eF{4asf{O9j@`_a2FbaW@A&_?C`lVEzMWrN=R
z^i_i1kp#UOxbC+9U{=a+6$*N<6Z8(2?G;ReH0uFp#X&pb$giNVSGBG{m3AQ8(V^A9
z?L0i6U!X>5mgBVLgh2Wdz8<%d1k;n3wJy#yJ|``k;cHk<U+pfcV&Kj0^B8S>k1SqG
zpr_ySHzyg8Tg2map62vfU&w$Sk2Q1kiWLsp8Tl`)BP$a0yb1)ps#P~D`Q?CK8-m`8
z1icAN1SO&iQ(gs0tX-9c3n`OpD1;pb;WeJYKreaaY#Ll<Vt3Q(iVV-cNp|V%U<!8p
zzfA%pElI52xj^IT2fY|y3zmjHWUY@g#!V9FB{4s0KAk`@ef1n<66gesi<$p2`8|t9
zQ=q5enpt$R7RxB|dbzxgAW-*NvO$oSWDKP-(#9Ci=I4VA^j;_EwRb>|;qp8BUhUem
zS}9K(hm~n10Dt*|XHX~N7*ta)UbUKz1bJ2#SO_s8QT6faM{SX|it(+XZ(FwqdJ12l
z%7*7ak^#MKu~du!J`JItku(nHUVbW$IEs+$1w1Q(p5n2_AvpY)!wJ<FVBCx!nfKEP
z+Je1QN<#<*dW1XP!z;n0Kh1-O%OCgq9`w}fnY}FmeZTz-ufOsz^5#s%mW2t}mXv}m
zNlDn0xDd044aK{y??b{5-=jE=Fcuq!$^HA|jl1tgr*_Zb!)KquqyN1cU$t+CT!P-<
zPd>zp58R8@(a|W1iAC(#A8;2zuyD$BOdT{BcU*WLvi|xD>*L~(@Y}C==(?*gear}~
zT^NT=$qRY!M1p1_vj6xMuRL%ozWJ#a(rD{){#l4Zg_n-nJ08%x_LfUAJ9S)z_vjrg
zpl2GS*UdA2iq}W6@`9eEp`FHttD!u|^{OPhO@Q#pvEzZ-_vOEGrTk&m5pYNO$lvh!
z1+K4>x@?vY^aNvB>simOi9y=Rc@F5&FP5L<j8vLu%D#LW9dG;@f}Z>j^$H67bdAET
zDZHiwiUK_||IPz?f<gm&{LKe?^tr1*Z`WQ`phwAgnq`XJ)iv$pW@|iyfnL&zSv0uw
zeXv)N;W?0{%GxVii3d}#^Z#u!EoF3CtpW5IP#I4%+;YA%^pb{5hE0YvW37g5I>Q{a
zlBSHW>CtnCdTO7%JQ`zt&k*?0s4n5L0=;ZOpane|Py>4U?gc&mr&cueGM0kRy#RVI
z67=2_=*`^Mp!Xd?PnEZI^u5|Wr^q-3j3>=4_;vb7ob#X4ob43MoILg9xxIP%PHBFb
z`A%S|bwo`Zf97%cX~rmIYD4O6&Ncz1Y@N@(D13ooE*qW$NrIT%ZOmtXtLJJVV)CZq
zr4OFP;dN^{@e~a76nC|#nunnJ;ni?dgCo)5qZg1|5R0XoGf}WP4fzB;XSn9BfS&S&
zD_`7uek*nC-V)=6ag@I;*=)T3+8dAIm*IV}azQLM5VW==CmYa<`DQfUyZ=7KjUJ7)
z(R8jRPs9gpp2FuZwnO$z)@$R(p~D0Bqx(xQA!o{D4EgjUJbTxzSUW!kTN9G7fS`Bh
zr5B=L@-$2yILLrr=AVCIL;OM{5%eCp?rKaPJP<_-V(?FTGPWis5(4LA{NUbrto3z>
z%bbcVE#AO%T(Z@fo8t$)nl9*7Opjg`ZJy@n9V|mPeZ6_^{=SZXqESav+|Ok}Ptvpj
zHDm|7W|d^OHUCSS|5N(x=0{(?w0`G4E&r7)7xXkuL?egfr<lfMYKb~C=$E9eP>-HF
zbd&Fk2zlzw%PVn)SjwM}Pa;3ZS>4+KJ;_WpA>27@1Q7HHeGcd)`#?{>`?rr`J;49v
z2WjYM73l5R>i_}0KTX0`2I$!?YcsKV<qRV^y0&M}F{1iB!@OH1E}v<JZ)P$xSAm{i
zfktwB3U>a#P0C+tA=4Olv*-$;&%&MzpdFeZV<!WbO@pc)JsCC`zN~fX4Wxn8a0)HV
zqFy_Gmoe2myF~;&1A@hLs&sw+?*%;>SjR{@pl9zX11fkF=s8ARj_+Ooy)Pc3WA-E_
z>@(1NgP`{ULGSwt1wF-&(nOG#x_lmPf8;tGUB9-ooq}0qlXAh%fQtY{I@JHUcD*BU
z&$G8;R>pW<lNFw>?X`72`=am#hPiZj4m9amb7YIn=Tr>%sVAD9-q=}X(=*Jq4CpCd
z6@NTNbM}rus~Lv=+#C5D(^0rZN})56NU4X(FM(gasnl&;dcqSJ=YiS>&Ym-ekzr+?
zO{K%tXV@pW<F?B&>BrGnM$i)MZB9(UqH(|BeS+M#pMH$A3FDFY*Lb}3^yBE*?pY*H
z{R;y={Sa52)Cylc-yYeMCSlM=AK;~j?!%gS(b&8&5wYKYk9)7U6pJTJ#FPPpaM#5b
zAnmttD2|OoGOzv6HCJL<|NbbBo`=l|ao9@On=yI_UU=#r^ceCf?S;ac%ye9UoGohL
zim4RP%caAY&9XNAAX<z?j;vj2(}JC~)85b4cd@GK)hk~X@(MPlINFp*mG)D7pjXNU
z1i#(^D?X?4m$UGjd;|5|Svk-pKSl87EWRS&LVkoJftp4L$v2U|BVe>cF$HG^^vsl#
zh30rObd%R9*g{*O^H&M<{5^oi|8cKv8^NnUZ}(pZ{Tw~@=4H5Z^VIUC3?u2gdiWdz
zD&wjBYW%RL&l<j21$urtJ$S0P-OAe-O{)ZBK&>1D8vHu{CaH3jQzHnF5p*EUg*_TA
z88o$6Ip8G&N0~W#4gjhZ%ny38?n;{K9h5<KTHqx8CkV{dN}2*YUPp8IoSs1$O7-L=
znCogBy?P6+qh+8|;n~Ze_rHAwdOa&1^rQj;h2s0y*&}hujc4QJb6eqrGh5(<(~rXm
zr?o&t<51M7Ro%eJ;YS~e$mXH^uO(WY+6*nJ6V7UhOK&?L-%THg%#!(xyYlRv!WS6k
zvf(+9WI-=;V?0t;$Kd5p+M>m2P0{qEMrhJ13Qbx>qH*&G99!pT)T&hz)oWD8(G8A3
zM6-s7Y8g(2A*xjvPB^DI+Pw7$7OjqB`J&@SN5WZrrS$x~0~hr8o%LYi;u+{M;3GWt
z*sU1)^=C-`>rbqYkH@Nc^YCZiKIrs9JM?<@EhG{2ri~hbj?X@Y_u99?#~oh6mmOZg
z4_|dLbNK!oG7x<_zJsFJ1=t*)gtXr$;LDd^!pdp0k^IXzbb9dxET1(8+mcgJFmool
zb$As?KYWj(XgX#Sf5*_z-oq=8-Ho2TI$~blbYyQzHrqhtFivx}(ALl{z5Cf~II<Mz
zjpH-;0ne%m^tLbcxM=IAz+s!(@7F?|{?p!iUf9!T`7~8?^~y>RSFkY&g&R__bVC}J
zZcIkuMz7zJ$oi3Go-kP21_Zy}0WJ9|8=Td?)iWnxbLQ*O?$9@J7GQB2p#@*YPgzGl
zg#hO)sG?px`4&#Ywgo+bnS74Sb(+4y&-6ibyaBz<N#>ePkDdd17V@M($g2Xq-Fxll
zp!Xa>&%5fTr=V3(fs+HP4XZPYm{AFQa%{EVnTB%8;dNHpoJ!+43ybt09XTsOp0-)Y
zBe?j7_LRNHzghlq>=y7;c;g8qF-Ti64=GC-j|H=kR4@Zch0}~=7|aSv_S_)X{WnQ@
zOh!mc77NlWsA-;@jG+viF^V*3&gSP%ubbk~h{xNZnmKNZn;_5W)pM?8|JVCs8AoFv
zWsLpH7iXyzj@QfPz4A0HL-6Lxw6MFNXD9kgl>vIMcX<L6X6$3oOQ?9zQ$EgO`pnDx
z6Qlm<i(x<a!tkGaWB5-!F??)qy!K9e)Twt2svTMl4I9_PEAO|*u(5qHbZmDFmVWGo
zLBI6C%*=6gm}7WNhOd;p^V%1M(W}y7InZSDN)Dehd37|tpE?X*5C053Mty-E!#+pP
zAzz|vpU-jeWf$Pc+DG89BM!xdSDl3~dVhc(!@ofH;h&@Xh)>b$n=kO~#DR1y)PRwU
z{4Ggl=%%2~C&rhCZ90pw1dBgw%gdAZ*uA8!eii(kL{LpZ+VW@&81ofgf9qMi`O?!E
z-McH&CjW`t=~Iw1buu!iPQu#wIINCkeMJzQ`{NHtp7a-T32aMe&cTMnWGtT_jqGVt
zu^~PIo8l9(axUw@i4)D|^@P90)2Cx&LIOcA8EfXvMbd9SA?BxV@MGUDc<+VB@k;v#
z@Z(PdkeD+ISsS!>UK*8T7NN@7BAX-7d({O!&C$EW^yuvo=rNBRtUY>~yDC3L3&&a6
zp`4O>?*xNdBH#EjuH@fnHN{}pJ=o;m#zp=|0`k_!V@CXM81+kEjQpiPM*i9#BYx|L
zk-zuH_mhSqCT9xaBANDtuqWu#kb2nxPnq;7;9duuRBq*`FrPWTisNgjVb~Gl&iE;e
zN0vde$|n5@^PhVEm`6>+Hm{K@Ky<;(Kq6&<QT~Vk(lmDSyQZUPdW-I-=_-<dQNu<9
zWqcOekBS04pPxcif!<zzIs52odxckA&|`)v<@C666KGGRo)@*4X$-u5R}Y^yYnPO%
z<m}3ED7|bq3;({V{PzR-E9!U@Y+8V&+Y+&KYa-(Frl9k{_wn|ZFX6o|uTU>g?eTus
z=P`Oh9~M|=sa$)0uxtNYWP0_iXbfdI)%umXa;E8x(`M;1b~2*YkUFhbG&(Y##?U&g
zYmU)%K5Hz!AkVjwrwkyCopYQFb&P36Gp%i!Pb#n!{PB1hMGJ6}zE^>s)vi6OS8BNM
z^<;0<kf3BNr31XIBpsv24nSn1P}DfQ2AZAR2t$ADhy0R7<`^BDzbS<_EtzqqVViCk
zD#6z6kMidiJtyOWAfw(mGrx{;Sy+;a%%w@V<KA0v^sz@%N8z5wZb3}OG{#{n3OA*p
zU~`IzL#qOF9cR@|rvmqDPt$pC%XtPz`DTy6M3%`UI#h{>%b!cY{Tf&N_gp-1`_&lu
z>HA3j<5w)7JBy$fi&fEcv6etr6c>vP2`pW8Jl+5IpBytxHxpyJ+tr=ZZA>@AaJA`<
zY3{vxnz3PA-FbD6>8qx$uHXGUzP~^2<8l9gzh19%&a)0qLu_a>Z<tAT1g=gXp_KF7
zguKi20dlN;Z`|^xA7`r}*orz+X_7%gQ+{l5iu`U^yk&3q`R`glo#J%|9fSXw1P7Oj
zwMe$Ra9K7_&W6bkiabYmhm|#CaOJGzIaauZPxx9tI`m@>*cTs>U_^?i2*Y8XW3>xH
zX?}^EBVRWTg2vf9w!ZX6J)4$Bf|-IcYm&zEBZTUHDe6`Ij&YcJv%x7#u*`>!YdSwT
zoYf?IO#p7B)qGb1TQ3e03I>-;V6vSCzs%jn+5Qa(Sc>4jeSdB^tF{*k4#dA5J*V<}
zU}mAiv{1sSBBx6}zZ%L}H;x+bVdv;|x>;cGazuzT{h7>RfnPhfS8~A8QDy3!0wPs|
zZ-v##;KxwhnVN5o-~365MIhzlwU+@T@Qlo6<UtwB2q~v`*5K%%y~8#E{Xy5zjD_@6
zyb+KrrKJQL;Mc6IC1h6*b8NHSB*>B8#A%K`>X*ILjf%S+-OuqdsS}9$^LnEs-}6+U
zjIYw9um6{K4yn-i-bF}e{^M2CN8GR^Oykc{-impLjyGSUz#_N$dEESU{I4za!$5{e
zy1`=D_%aJy-y1apy~pB}BGD0pVBiH-_!=uZ6nw`JVx4PBQF*TGm*48oyjG(6!{<Iv
zb%$&q%u-R#O8?9$+m;@M!TDBp)gLT<@}Am!@BQk1pGM4ZzA9B(cn@a;pfWTm+t&(1
zN<z;??8McdW;n1>31ZEk^_(77Jzk+~L=;WoNW|;-Zxa4z&$|8Tq1W$xm;G!u(eUqh
z^ocAIhwDlPbadsPD82RXai8_uRtHD6=S&S1$gnsL2yBm80)%hz$Ed^!r4aCle<vMr
zhG4hEd-pcfnv+&a?)F8>Ak4ssPP^dbNDBTGXxau_8Jby|Y#TkSn200bUpR<@^#gJA
zwNyw3#Fxnl$r<by(f?e#v-XvCSEQ7w8{L9?rG$g1O_v<nQu3o*EB7>UnJnOoG*nOS
z`fuThcQwLXE$2+L$r1@-q?bhp5VvCM56#0<5zya$H@9u6KN5UAQ_me=U9euX{Qud?
z&n{=uWzKNV25*Nf(h6&6@#KN32$nKtoYg<dKP9XfGraYV0uoXi8(bPQW_Z^YWELqp
zjEm0y+#yo=4bD;R33#cCoKNolBQLLzv^cKAF)r+4B$Rf8tJX3N4S2YV93S$NyTTv2
z-><f@epJvHO6UI-c*roOk`sf}LXcciE+{}1FIhW;C8H`6#S-xcky)yJG3$4qrc@1-
zy{E?*=sPm6|GxTif@YAwjblhD%i1!+HSd8vBKX}-5^OTnG{T&iZLsjI=_bx#&0EOC
zH=43-Iyh)xF|#k5RMdPO)vRN<m`$#|nUdcxr0&Qw0~a_+gs@<QC$5)6t^-TQrY=Sd
zxzv~5z~5OTv>xBgF^%m{iNBlXp!{K@%T|C~)apo^!_xV@a{k~JpYgQ&-w_$z^4xoD
z{`=iJi5WgO5Tvyhv3N8*5IkhX-&;9SL-nmJ?EERja>$ve^r167wUHA$#}vEA^u}L~
zK__dx|KZzNwtx?lV!|Dvd7DqyJ)K0P;m3d%+|WR&uwxVn*d6m~V6R}yZd{ee{dn_c
z?BI!P+n0xv$d{+*?D!jVc*&dNNv^Pll|#7Tx!MD-*Xv#8WToY4{~>mIb!1e2K&oam
zN2F?zR~bk`hW5C{vDMzsRv5k-^Z^@E{-`L3?!qi{a{QG7@m@o4Dn%pAVx%%>Ak!3<
z9$w(ZGb#<D2Ht(ksj@yhc&Syqk9`T_Ie0Jd7Cy%UpL16K>kNENDIhg4)=WzWAZ<M)
z`p_bn=sEh($=Oi-Wb0{hU~C4K@mIseM*XdR$7(|DKoDfDA{T{WhRTD>qG>L=G6p3g
ze|63QSII56FDpM4=8s5!D(27n-K2q4Hi8cAU-5Ij!?^eyGMLp{DP1@H@rf1?wzjrE
zb)lNWECRp9@u~>-NEtDFG3V95OED^BU-&Ex@R;^zfw<YA&BnXL0JZ7AYyGWSkEvgP
zzZVc|r3VFH(wA^scXLefu-kKKShfR6j@DlOaH(q`Yf8W}u~TLa(~=XqSe@KwcN8(s
zQKMXYOWwq@hsIkb=Rcvt({iD@u*43z-s!-Ej;tt6vOllthb)ml^2^k*%AD}LL_|GI
zr#qI7PZcQQb+2`bPM&}S>fTTESr5dCtWKFge2J54O{uOWgBU+~VFM3pgf73bOtgc8
zbbbR#Zl0gG-Jf{;ZP-V%``WEZPqb+_8_DY3%oOu}>m_1&m;5l8p*Q8F!uibnX|+1l
z@9-iER_`dMC-tB#_Dkb*E>!EZw@>E5o!IVOBL78Y<NP3_l3Z+tl7@T1m5;j2zj?hL
zTei<XF|g`77rP2bPEK;vbTJWbQ1Zs<V8O|v2T+~;=TVivjhXC{c^H68I8+}4s1Iqz
z;ftqicwm1XQ+Xt6{?$Q<Ta+YiR4nE|ZcK)RA7Kzrz^uCt*(G)WOa1~e-K8IzsPk5X
zw;JUuEFcDYoAhIq%w)!beZ5ZbjI7kba5}g;^7T&z<MX&2J4WwIcm=&CPMz`>L|l~n
zf%&fyS&N(j@LN6}LkdW_XeJkYyavf*UYv6dY~~dYqJP-oY`GXow<KY;&1HkFG_*sq
zI_W8L*wfFp<C7VF=o6!G6XP=UzVA>Hgd7xEr30~?E$H?vJTdK<)aCz)kp_obba&4d
z@3Fy*RM%`1$9XU=Z!$#k@{nM%n~>Im{m)SBgC?hHQLRBtv-K)ASR>?`QszW7e2tv=
zp~n6k^}qJ#FloE`C^JF3^9>?5QXJyzgdX*@{PlUp-(Yf6CT-7G-`{jA!RDE)CO27}
z?hW!en7VY@>afQG`nxegRQU6PyX*Atvg5S&poLu$MUl3zun2PLj^}h}+sr0dXzC@x
zyEfCvWj{(7JdHmXG`;v50uN(m5s<y?#Zxj$-_53e#5tem1ue|reTYP_2q#>tGn%LY
zo)V^x`c16m##D!-rD*IRd-t}U7GV#T`^hawmUg-7*YKdLTr}WZ)fQ#=ntB**LFJ!`
ztyl5l-WQhOVisgaTC2v`_#=Lk*TF?NA1H$}iwz(wB)F{j&_nxf{avM`Eb>*}BBI<R
z?Xff)&|CI&U62j2Wq%~a-=;}`lw)L7uQM1UiM&s@c#c)O*&BJ7a4evtF?g!@TzWFV
z9gGlnHAYs2hySX=$ht+L1nZ1m*jmyK`!{sx>Lm`ZYqz|a|0oHT<MNwJ4Bu3GfqYy7
zRl$VKJw?5(C<kO53xa%^BtCPPRb@l4dPd<d4r)-?^`mfw1lGk@P$nbjYhP(ZU8!zj
zSR>|mzVk?w-qo`ByX&WyFX2#Cm@`topaDz2D5YgQyfH;$>i?9*F@Uq(-}ZbBCSLiY
zLNU;x{w<A{HOOI9`eZ*(JsIRxJ<Xe`b9dHO!=7DU@QgNNRJrKAM)hipG0l=Y`xU$e
zZl1CciW#z!iGDmI>fnMe;6`D4{>g8PM?L?7CiBfm=5!P>Y?rvceQ%U`sJ7+Wz<A#7
zN_4DBG-Sr${W@P#MrQ#K!m+x2ZgCRA2CH&4_|txusm8;2X*lTCx-hz9?^<K!(vP>j
z_>B#A;iLrF)33k4$E;u(I;?WwazFpk>WAnLc;h2Wj$EB9r|f_(+nZY?)B8R`K|r<_
z>yC}!@{P=*_|B2t%~@UQka1pU59h?`_`C47&gVzSUdH3(#0A+A#dFssebcMzOPY~z
z5ZR`~M~R){&;<^RfNv+DG*cx?*4Gl9|1VJi{ox5hC6bRFlK&NIzoSwgr-%HS`7Yga
zBYqU@TQFNomVA#ZY&$nllyj8)9Lrxp>G$;9-%jqRdnQu{6*j`k|MjkJ$G+F|?O>B7
z(J6DRAD-y`_gF^&Ug`OkK}1F@<s4%thMSX+i{y+0)awtAqrNpiUqYJmA;KRZ3sDf^
zI0%c#cT{Lx#%3+Urq-`DF7;x8SVs6b?hB|X7Esqe{{Ur=e6L{d{s4<A`Rcg~QTzS$
z9}VkA5jomNlTX0GUyNS0{7qiNzie!(O+&-1+lUYhIb>DzMcQcU`GA4YH9lJFw^~=y
zLQ$Su)h`Z9qE(1?^@kC%{7Gt}ZV4Z&-sNp!0UKmwxT6;YU_-t#APfL_{Jqk|bL>#z
zOt$`y7c*EucPxaWVVDK7Z|0VQ50kRc+ds-CfS&=HZ3sUB2NV3u9y!=s4#G=0hN*)a
zT~D)71FU!O$f!952R|ACr}-D|!8e3yq?6^HY27XAybu=T&0@KIAhBl2U*G0le5BEZ
zLhR)Oc7BpH;r|L03CMeOMg6xCVt=S;?S5qcCvQQS0M3a-Q40KN$RW*5NM(Q_#ZbDJ
z^j1!j4c9-v?JojSKPe~$94!s%reV=#gTg&X08QSg7#IQ20J?>qGFypUgzKLDIguOl
zD16XxqqohaML_m=^Y64EEOK@-TI^j0$akVs^(zR7xWa=EyE)PxbfW>|dM%%iOC-Lm
zRfOgUH)A5;bAVUZl0H<Qyvk-`XH@-`-K1@!ufH6}Fj6JI()Hix^#9jMr*_-3a(?U+
z7e3lhKJklY_yH89;0%EHItWGxS8BoTN!5_!`aUymsV_AjKV8(dfbC*B^V*@fT!VKz
z2Gj_Q4z!yaM=Blj(EPlsKl{1j08yp>_d)RBwuFc>t~-jlkVg@q{(s-YRcT~Y(_ck)
zqfD=ac%q*Lx+Z^fc9{s)0e>JPBuLuQlB7+RyybZc=$*qdMDy_jL}(}GMr{m;*ADdg
zcNTi6F(1DkP2c2y8~m(s)?U1Ke#bWA{-ez#Um?X_k(L0_v|Q%y25KNWr?6#N#iVZz
z6oJ_9G&^D*nJoczq%%jibmsrPgN@drLRAAY{!QJdV-KO9>Iq5u6IL@IOYF|5A<x*g
zAEhKE^b+T6p`Ffipue2H+W-B5(wxGd7^KDPCbi`{Rs7UHD|xC(U&N}8OP2AFf#U~C
z?$0>+K~0*J1_sJy{)w^^nU}zwA>O#Lc*=XXUATrELn$p_M~h<FY+$8U)`*8j0Ajxj
zcZBZ>|AUPq<$X6C@AN&;p{d9k@5T4a+#}5<zv5j8i{-wQ%8&_7{5@^!Xmj!!RBJ6M
z;@!oE7S42k;Yi!CTf<wu3}{_a{n>XOnj6R-hD2S0HB;Mu#^yvIDV8(ig#eM^e$^K5
z-{^l1vpx7g{!hV*lCFLmn4D`x7({-w^b=&!9XA}fz>fRxd$1S&`C7LLhW1GoFVTR4
zl?#1>%@%Pv`&%jO_KE%$Cl;m~jTAdWx5$U69jR=hQG?EuY3GI-)U(nw>OS=qB%4R4
zlnoY1J`%h{^mugJmLmh$YW?oNk`b&Kb*KUaURTAR#$bvZ=M%x)^@MxkT{QSU8*$Bw
zY_tt_awI`-nX_%|Yt}!s=;JVPdX#0u++jde6CYi6=SpF%&>A0d>VTcd-<C+yJ#3i@
zA{-65#*jIAq}7KQR@7CB0`KbrF?s{-f4i8)^u235omy}&oQh4wnyS|bw@}^^LxRTP
z+yQ|R6c;R8X;l-2#VcK*px5?#=}|3*g@?Q>{_Sn5d56k3`O#<+Ns08{#E55_p{DQr
zbWT-~P5O<gJtGpa8LezS`u~=etL)_R(X871@vFX~B2+mC7+^E}{slo*<^l6G2S@@C
z@RhNRt;F3&*?QC*!>rdOS2KD{yT};o`p}+w)4q4uvi-vg*5t{QE0ow*x&Zn*@hH{z
z?PfjYXXY>e^qEHc+N~<aqf3%9G%CbVCVFIGZRxkQcU$B+E#9u>8RrD+$Q*<2UGEO=
zp78b@a{l1Aw9dZ5N#c^I+S-?%tDaU4_dP-6VrBeggHI8wKik1=qTaNVzME4h?yNju
zu{3M)7r^DZY7x8>Dwz1ldU4j3X=FGk%8hmwqC~}M-~0+Ok|jO)jYbfQWQz28p03=&
zkGdysp}bQEAD3Z2OW;*MKlfUE_O{gNC|J((!BHEF=tyB!$<GGMRron%N1KWGyQaFQ
z^rY?y?RQTz(DUDM;jkZjxAL7b_g78|r#3rs%PwJ{tI@{$8?j#DIltqpr#9qq8VUIy
z;Cy0>kBntnVbluOFKNmxyuxUpDsu+EiKOT0%^M@%eh6vY{=TE58HtRx;ic8x3?87t
z{}baOhI?6jFy%L|U^9x@vc24Ry7CQ)Iupj*e+(q&#0s}UhawRnBO5?|z}YZdypkx>
zfsVPX+_*oH<Ti|U@Kxh$tuHG1#z3Qi4{vRsvUu67QxJ5xs|J*EKa<Y<65SVsULnCz
z3QiLM*?!gT+YDG%;W#Y*)7V;oXnXY4R=RH0t))dqhtjRsjkNno&F>2Dso>P&;D5>A
z%edeLgkkiUi0qS$PJsWT^w>$-uvbFoy9m~uQY4&l+KV8FF_y@<`Zies3h55VVF;%B
zC(#*F^`d7uQ*XOQtLLci3_f=cka#{icDRR+`#tynf<H`{-u&JY_bYzfa=6$w_uH9M
z;yrKsr)(~P)7tNpaAC1NP&$3tg$5A3TKtjU>Lxj$NfROJ*vS}>XfUN-n5MxU8TG-b
zBR}=1Tx-}!CoLg=4m+FTSzOFF2+ftY6L(S~5Jxib(%3urki<Lq-nPBZFYsXF@a((_
z|L=cNQXkP{wT5Zg`kI!e&(ouMhL0&MypP!a@#7Y@IAhZXCn5h3aQ!-i0*EhfS^3jJ
zD2N0+BknS@nDBY~;RFRBu5|L&JVD`$xb>sN*3pQCJ3d0MBiwvzoiiguQp)xBuXCd(
zU$HbC>(kxxy0-^69ua~#yj5Y*g19m>L_NF*<#nlTpScKsFmIK$am#Iy7GlLag~kj;
zVa0xrA9e2_@0@r_x}=v$_J@U9!0sNpQWY^Ori>)(uCKcuwv>e;Jyip2u3uz6r*cZ!
z)@&6k57=+`E>-6<+`UKWZnKE7p94puAX<)FBqzsJL?a`MFF?9EUI^rXV}Jq7xc+=K
zHFNa}72sOeB`-wc8!CZqkje*9Y9ZS0=PvkUK=kU%YWI8nrdE#AT<gL}!c14S?k@MX
zqTAs_SiNWVlBx`ZY_1Yzx#J3p6z1UcG$<@}-Al3eQ`X3jMn2*q8)gkFv6Yj!JbxUr
zC?8>6LQ5&d)vzSL@?{-cZZS*5plq(iRkY1d7zs_6a(7Oq{Tb-UIF%qcrD%NkslH{A
z0y9I+@fo~I_!6D}_M=#$lTczsrsy@N)zJu5+425g6eu1#z?TjR{SB36Q+$?(%|z`-
z&r9nf^&lUA^RJ~poOy?i8tW(SWb>4^FEk7Wh@2G>o%K26GrHQgQwSNgaN>?WWlgjg
zp5&p~V2;eTbiWeH^0(!cab4OKDA}DDC}=j1&7y5V^HRc-S6OvIO6%U_ZR*y)qsLg8
z{ywJnAX-G{*0HJHem$SAJY0}e+OG68Pu#3-yi({5sF$APcv~o9Yv$BY1yAVoxY}Xa
zid&iJhD+U9_0PKyyV;nzZC*HW>okwuQ~3!9Gh_HW6Y)6i;<sGoHcq8_<UcOdy-&)E
z*TC7T7xbimQqk<1pB3RS=fUm03r?7sjFr8%qGR;g3Cs*wX>{rq&DOxSNYWD~hu!}`
zOR6-dnyb|B)ffI4;aDLRunQNxxcK6m1R}K-%{Kh=?N6rr=1ZutI{#ehON#>Ivmck0
zo3fWHveh@@XhgeD2~}=wm_Cd>AEbWJ%H|137aQPuo`q=jS4W&^cl{Rd)qgGdaG;;>
zp{Pp-cMJ;KEmzn$uCuUu?1)Lp8Vei}sK4Aoc@;oOQsVy{AGa%7PrOYI=C-3urZ!D!
z3)`C}OFf}S8r1XK`}-4;8Fu~Y7W?zGRVWaV=YHHlz9W%q1^cW!$It8;4cKw^Ak&;G
zHxpPu9a5g4e=yo<_4{gS<t0@avLkLq>aD)*G@kaXXuYk@F_?YLxGg9UUoRjo`kFEL
zDDg<Y;r)9Rd7TOBRlmWV7$-ga4XcB>s^ogb!Od`Xt?rM!Qn0`EUGBvMRx%e9q=wx+
z4N{}~1Ryy^fj=&9*^0jz3v}ieW<IqSG<z0OEYv0{B9RaB-kfqj`b7#fRqS&_UjBQO
z!X%>C2aQXUBR)4w?%BV}zrI9VO+@WmWZ(k<Pmz%FR%6~>ufe)?Ilo3_>Xmw0uHOf;
zfosi-Mf}fwI?IW&%Wv%0qJSswaeyHgVh@u1S1j!}fzE+<0~g8#h?nz&O4evU{jE@c
zai|Yu-pIQcztQ>-JGh)JYIJi*9(c3*6+XY^Dqm&gsKz#$&X1VE22Wo~qXA4aDYU54
zdAT3C^Sv!_W^u(aYc!pszPL+7#$E)=45*6gqwgjL!}$H<CTjtz_We)cgS0IuU&Vao
zClX>8Q;4HoE3j|G-=(VdNX=$`!+5V!UK0yWS`Vg<c}WMQeN+rl$?Lh<5>%grq;?uk
z(ivoA{qUruMsW4bA_vQ=_Y|;V@DGorSvD9uqiG!*m)>6GntC5ukd*u=f3gyQ3~>-a
z6S?J(EbbkMxrDIIBwEXKQ|<wK1$Twi0wzN({A}LKQW4x>MJ%iZQ0(rgzawvWdK3;_
z8xY~AVe0UBm1a00fIG8HG%28}D>kqf-*7}B6}dBcL+4i9_<&dSfE&oI9n^I}&Yku`
zt3FG`aXhTQy&QN&4d3!au{y6z&_K?7=E2A7kcBpY%!xjlDY8|49(oBBC{b%J{WKj;
z=efg4M)8tKUXGeU?#)s=eQo<0YF~*K^Y)bX6#BCdt6$aDg~JpdwT6QPBjW+RK)B}U
ze(FupdYiH-$#D~+O;I5u!>02k&gc`{4&C^f0K4rE!Y{j8V+;=!zTES#vv_*-0JkOs
zx>5YroVc6Iw`JE{SsAe}CV!=IlAdaZ%isj1b=Ygq!wteLmJE$yUNoV6Z1)lY3@;Hn
zJe9<71wsUnZ%q(?WIaCU?I;cqBO=LAydxB7h3lz&$#^psU8GW_=vQ)leB*y0ZBvtL
zPmOx1=NLc|24-7yF=%Q%B!wu|`JSGqU%$0JkpITS_fa!HG+SG)flG`M$*6uH9Q4i=
zAW)Hd3e9LNv+|lo?F$xvJ=xq!g%<LV>#-RQ>grp>-*xA&{xH2aqUoj7{j_2{eCugY
zsVni$%5C|hgO8#YHPD7LJ^DaMB2(x}3N8jG8dITixAefKF-8m6ZB0)x;~^>#>x|<5
z7(yG_Xm7cbpECqH6;BJfLvh}#Fmy)|8km{7zYQpPzGctk%>G)ca&bBnL+f{z@siiJ
zo4ZLE^;|ECaY?$xy3Q<HpeV?)JyTm0BFwqNB3pp=1=tK}=C|!?1-B*B)&<p{6dPgy
zSJFG(^eM=&n7H%4-h(=bHpUr0NJ+fLg?|-@ks!4<4`dj4m?!hE_RYGuDWytF&6UE}
z`Wb?yEg>`;YJo*5R25Cc?UaNf<;y}n{*I}E`i%p1*LP}Dou%oaV)R|`;zKOvm5byE
zN8H9m#lI}hG>x<C-3l2KfZw^HK$`Lw39ZCwy(i3N&h9nLfmTI6Y|4GZ?M#vSEOUYN
zQ1Com{o59`Upj*ZLhzl(q{(RCt~R8aR1}zTgKQ`t4ZkE=)aLUWJb|wR@Hr<P;(iwn
z60TAfS=pd9N&LjN@NdSY`HfMgZ_M6GLR`LizaI5Bu4+OdJ?6JV2`zL;%whEhid)dB
zYTu5gOyO0zXE)c!fSC_38Q`KMrxKJC!vo~T9;!w{Xv6IGL^nmLa&qrm8vRAH)-jyJ
zg=WOcEZQywlAL!L?P=(fz!(8Qt&hX1Q77^K2_>LT4qBqcDl!h2Ztv4@i=}$QCTf`4
z7wI}rzZdXQo+Kj`Oi}j}=gItU@5V+=bk-4Y0)$0i9nFD&i}cH9EWN7W_v=+#r;Pah
zta}17b%k_{c2nrv)rm6n7Zs#rQ)YGQJ>6p^A5X6ildT^9THLS*;!<Jzqz7E}=w?}j
zvp(Kg30iRnj4%aGci>ud?BVq#K9c4UVspecz|ZDZ{El`4QrcyE@i4?W(8fZZL~Vlg
zGj(@g<T4J1Ej<P-9XgDDZLJ8|Xi3j2B;7hqIhZq}2~Qaq{JXK|HrNz-@F&lmaxU|^
zAgc7)RDSO#*|Ma+`1I2qTHY4vr=3{}@~3%Ihmg5Bqnl&$Y7jqFPDnzpbMtxvE%(pC
zv>e@fA$rVFvN6SlNwL4<lgkusP$YV7#odqSa-xOgJJ|CFu#s=8kRm$X;{s>WL-?)~
z{ny+wJ4_);dC23690ohuC9P2pf85@<&I^lNOzUbVOR9eLxMSP6#QIF%a{1((n!-<X
zJ60Dd@rCu+Cs>;gh_|1hJ1I&*f9_l)`+@s_9Jziw+<q;Pd}>M8=;MzGVtCGdMFM*P
z`(UkJCSQC}j&Io~nQo>#wh3B4o$0uo1pXe-VL29(H`y1)UC`_OVXUvallsdnZM}#V
zdgL>O(6u{j_QyOgp$-RIYC(biw6K+T(lZTX*GRStOn_LcKn_1ax1`p*a>9a2$+2%<
zU;WonihYotl!g84N~v(!>AL_dKJTbR%F2;`7G|N3P3phA{uue)U<1o_7KQ@-4s1EC
z<mYu8Aa8sVx*MBasPL{LW->QcXi_S|PD2h>;)&J6c*t0!5|r=l#vdr<iA5i6sk}GV
zjJ+@U0zkz5Ez=1<A+G>8eq!dK-4ep`ZaSvS(apSpRA)nUY_!$sUGWBSi}44dChX16
z^P-9Z)|V1YTm^2x2k(3oSyegFMvO>-v9(rUcMqxhtCMaNAgJHSX7<miK^Y!8Y{tST
z$!o$lNI0pzXXm(;m3>_>!PTHOrI=`RM4W)T{h9h8VxG5(b6%tbbKSW}vBu=Pj}fN;
z767yPk1s|C&BDw01+FfgOU(1N94hOn=atFR?nAA&?jzuvkFtDE#@P|2MCA~}yf1!;
zqWH7D(BPZ00DH5*<+B?f4?5z3iEc63wSieb_1hp~kbkLQaDgypj+2#LX+c$MHcCo!
zv-D&wPdSDY;$yuCht{siKOc0Um(G)Hab}|Sp@Zy<RrJQ2fgO55^5*EdE`m!uxyJP3
z)8H&a0;mH+rmM!DAmQ1SlGaAS9-ZP@55|o+kX~q#QSVU*rEcAiT@r}kRj1{{qLVy(
zPusf9Fh(tw-PQiaG(G*b$g@~A{=dM#VoclPHo)m59E1=H|33#qIB^C_oAmKJSU@~v
z0qIIifUbR%nJj6pQQ98G5H0KJzrUf<AxdfTW?sA7^w-s-l%FlHBaBO_|1I*=RCWXL
zdbteo-gU2Wsd6PxOEJukklK3;cXsbwrl=q;Wo4B(X}N2GM>X>LZ@5=;r>k2a<Usl<
zr+R7CT7=&%2Qky<@nJEHoDvT7T7ZBv49Zx683EO*>(znMhi~e46izAgaffAR(a+cq
zd<$c~inJs%nHTK}&iVfSCNce22aZ0PD^<vQ7_f)*Qm#p^T5P;{0qv6Iv}P00MpiA)
zT0M`)fd%rvp@%P3@e+<d8L}G;{Pug`@gdgiIQ5GV!j{Q?ZAIQ)V@!jT2K-cJ%b$>~
zRv>_n<yt9J3>g<hxP6sWir1wq)Smee4l-DGiK%C{^1DcHPN^56j1-_=&awrE(t;-q
zRD~vxT8ELPY3a@DW90$ejM~?`vf;)(e8XW9bVr=v#993dEZyH0?O~GEAM`z2Aa<7H
z(q<gfhT{t7KXQ~eTg%>!2fUta=tY`p$PpA@l_vu|LP0|6&G+#A92;HKqqtx%a52Mn
zs0Evhdng+4Cl=6K2=jS5@61icoqOlx?YSIi3&OFa8XbJdush7OJL$k-#78ot8`6Dp
zp|o*d@yHt(===KVlKu5vDGufXg5)moIHE(!TXFs5at-_O#{YB@;Q0q?xPeSIl)jY1
zqlfFu;fmeOna_RKrWu?SeZ>Oyq_*(iQ&CU6$7}7M^1iV?`%-Q{wbf{-EfK=?<Zi|G
zL<SBYpYo;-rjE9+z2$c6+|)4(*!PD&%~!oo043tG#x5#LFMVrzv#%R6I7rtd2R;$Y
zi6X8!$ZkxW?DwsMR=*DTR5+$O5ri_TPcKEr7liYRX=d0s^yVEu55SiCDd8SPYMeT0
zBLu-id2VX>`k|B`^J#VXTJmOZg8W>U<@Id?|8K-e%<n?n?2y@F!*m9?ewF58Rg|0)
zZTLa7KEKe-M6`Q9417n=>?;Ox4iUx-x#jxx?%m%IoV3#rY{y0Ezj;zF0{>bb_*T6y
zLp-)ZCI(w901Zx8E-o6XeNw{_&QAaE*SEUc+<L|^wi2M)X!%h~xyUMP7r+N0(}D#c
z0f=w=Q_zF5cr0Kic|ftq3g-(;ipwwyq+ml0Nt>KNSSX&makHjPFS^22FWC=G(v`lk
zm@Nf<x|o<yuKJWqD?A@Ll@qd2c3&r5OnNRZi`*=l6h+|UF+69#wNO#qaDMVS+x8LL
zwGwCUdurE@yLo$VQ8|F|FG~4HNO198jD#d1wDXp-f_*pd0UipX^eP!qIfR_|B?lVR
zfs0za12fJu%n7Z_vavF$KicMM2Np8ka>b6?iw2K~(a^b<fgE0`GIwj9mT#<<$(TIX
z#CI3XO)z{m@zk|s8JgkIV6R~U?O&ZES;IAdB!ITJ#AA+8@d}|js@tOf9fQo|RBoa*
z)9lNdn%W<PasP!m<KGwH$POA3!WEsGCKyle&4aODqGJ7$<M^9t<@>Eja+(kmbjhP-
zP$U>5XMk^kLqYxTN=j6JzvGST-E|ARb8q@@6l4B?E~oQlkE}<3;&cB;uMpo%-14&&
z>jo_#f}D&moU*M}B%NN}96&W(g=M@v>K1wRV+G`$V5u3UL*ijP#Bag|$&K93@G>%1
zvL>L{kb5`#$u3H@{&U*wR+d6>?vKwxgxJ7%2l=^hQY@eoAS>(-Smq>>n^ra&sYG7p
z34%!1;mWuRyh48=<Eq-fEBH(eeZ)q50|{yDWDim3m*&MpJS^RKJ#oW@ktko(!`Ytc
zR#LQkSLs4QG(Wc}!60}EeOxn1x-p#rhUoJU5DgPstHPr8)v=Zx<`qGlB3%al`nYp=
z6;{;Gl!8@TNb|GjD8SGfhHuU*e(2j2bSQ~9MlcI;$I!z|k>~-7BuW1TDP%IweJf(f
z!z_4>Vkj5EGJlahj^56R4yf?B(nH*x;|9*3Xz12{#%4k?Hc>;FH~LfiN0CUwN6t-2
z!I&{oyy!5|=~kci&5!GHzui(6&wDiCt%E%7UjM9uSkiJq#duw+!dq5thv3g8+-{}2
zwMpWvgo#Nwy@g6|@-h67Mr}O$BN|y4qlhzBxuw=gG+mHLk`(maa_1w*zmDTq53ASg
z;VY;Q$!Ca8q+MT4z5g!TuPFo}BOkG#a47C3|8!)$m?j_1oMGuI&5bd<HLx|@cG7jT
zzy7xEebkOZz)gh`cbF!DtL%T$i%a=4*}&6&Dx}$`MOMZg6RBH4$^f2awDymYF$`B>
z8D3oW<~~(xm*&&T<c(S`p~;^fznNFW>pf#3&V`u%Mytrd&J*6t3pj?Jp5Dd8VacEu
z){KuPTUt%Lp0ReN;zd}rBnGQ#Q!75s<FkN*7-aM0KWx}}UH&G(Jw5fhw$@yhUslJp
z*ULutZ_wK*`I%8Sc4{tNE@m<#zg=EZ4YEBCB@IpIE&_>#eehRJ$@W#p2T=qGK&xNT
ze0=BF`T+4Si{^G6H=~2+Y~%+D;|%k5=s5bO8;%J{v(z^N{DBL;Z6tEYR|2{ah5}S$
z;gR_k!x3Sd|KkU?A;B_OW>Pje0rIM~$1#B!r`1w^de3s&Q;a2AQ2-YDj{{z@V*(Kt
zJi|BpJfjffKw)dJ?4s<Td$MY~0E+bFkNU)lldyZb<EHu!PH>I|Y*lc~&J!a^20Xvb
z1;MbawCN>Sh^-PCJf?HXi&nZOj8x10C=%T^4D1CZ0d+yUBbl4sc(y4v$^u`^srIOL
zv;@{|iIKZlHGifYzd*b`8W&g>RA=zHG&Ar}#SgJCbO$F??OtJD2_#R7<4?W>9!#xF
z%A6>KuN@){clXo?#bZI(4B#9pCw$Eh^Lmkd;vMzFpns$E8bt3W&`tjnPsS(QXuQ{{
zFn9MW!88+Y_ssRd-0jE2gYWvqQS3pryB!N20jF4B3=ZbJY-=}E0eW*z^U@9%m<|hP
z^b+Si<M5Rn2pJ2E(c3{B!)fQ})*;~VSftpKi3`w5Vuwc)j*AX;9mbAm_X}5*cx9)d
z*sS%jKkMt7tt8$Hz(8Yx{ipnNRWzhysm!2Y)QBLFz&c<nUrkftSnQ|v;VGkbBk8&`
zZ5ZVo{?-M~V!q#HKj+rLn~`ABx7fRtyla(2M@~*y9QuGa2#zD(KQ8#|D&vf&<_nWp
zZd1tV2s4lHQ^S|w*QAev1PdQMxN@UryKU`3;R-}7pG;LLpWL;h!n!-t(;7~**3bIP
zeLqd-(>bJnO3?~%i`wAl@+C!lj@!hA&vCs(Y}el&jp7~9fsGWaqTIy9-biabMII1{
z?w)HjEr3J&?%2YnLmVaMT{KECoo7U9Fh3;M5pGL%m)t3>2vv|lds!w6Eo@vPA{Jzn
zhZ?ZljmR`iBMQ~P_^FNa1wvN&L9zZpc~R^vRwfT3tA3M-BLB8CW1QBqJL1p+_CETP
zDn2aZ=Q#iIaAm&vkbwa#TjhoBTvtdOn0J6_oLHabaVm{I_TSXDeLt6Wx7tu=JVaiU
zX|~N^*i0MKy22O3K@Ns7$7cx?PzDOLh@3yL5lLOI9d_+c>GprX07_Q8Q`$(kY62*u
z68+p3iJEq#xNwTDu@QgIbdbaIqWd~w?t&K{P3r(u<&O|blz*T2(Wv-C#Po|_A(y;N
zI~s!4f5?ay&xOj*?&7gnD2B|lk`zYMc}`a(9sGg(0x>01iPgzb5zz0uiKnv{$qa%a
zT(7VLB2CjYl0mONmVTuLo1YbB@^bRJY<gj!Br5hLnST5}GW#q1Sb9Pf5AL66#|q=m
zbp-CuHrhvhuKlui+zMfVOc)8TlWTQb4E!cX^Rio)z=w%?A+mqnZ+zlVGxD?}RoWU{
zy2pXH$_d-f1_|}2ceZ~Sc>2Y{x5G@q_|xhx7zL=RW}ls!4?{&&nXe>&;Il=qZQo}Q
zK)Tcoz7UJ(RjG8nZO^$jnfq{YdLVM1-TQ%Mo4oPB`YY@=7Isy1)CNbd;^mv3^(4TR
zGBxmhVT3ds4S=em(M?Yi{PNSPuSMHcgVNt3=gIDcVyQmyf)&Df3q_TLC^zO!x}LRZ
zipHrE=(Wdn2v={4mQ;WFq~zE82~jHS^<)|CF$pq6_9SysZ{9IoUkbQWXh-4xd?#BJ
zNQ!93RU*loAUzi_<8OA5B;D5RzHpa{9kMGFIU17FF4(t7{A-r@#tdqwfd>1TBtjfL
zCrPYyew$q3OMEbmQbT7lSRy0WXrzx85hW}gglc?Tj2$b8<(JT$JLg1zK#xTQkexeG
zf2u7Hn3bqzW=O&r#IUDL4x@ucK99n_ylQxFqRirapx4xT&m|GC;!1vgyXSZC%k=F0
z&o#qVvJ|#o#UdrxVa+~?`s+)#-vkX6=?-MshHxy-KOEATx_v-vrB9NhTqdDO1#5k&
zc1cKw9{;}mfao#q)~KV+qJn+${tgC!b3g|Jc+Uc74Xl(S=DKAm3$_tStm_doc#3xo
zo&O?#?x^>YYHDv9>)ox``QTXoZ9#Q>etYR8w$WMV>j#M-0j|4~C9GI6Ox)gsw-QaX
zxzit84|3yASGL_#KBmd8q`2%NRMnooEMUV%enw8+lpL}+d=Au&P>gI)kD8gp-nv$9
zf6@+M;(iCKUmN?Q-su!?srBopOZJt#s+M3Ad`XZ&?I~Y+&@tf`;4^U$-tZl}__xDy
zk-aQ5pkh)dL#Ucbp(a{LGH9fD%?ryv9$w{ORQ{x8o9lcc(Wd<ynIa*h_1B(iFu!S8
zW#4EQse)U%HJqb{ezuXBBn2T`o<M0?Mpkly)mz+M7dqIjyqx341m%e3vLnp8Fljh9
zqTIqC3pX#S$jbU9f+07exL;+QcY6giWL`C*!TF&=AauVy7!${z^gOOHrA?B^OULf-
zITg38QTiR0>uMJQQz*eDX0)!2JY?dI_O{<-I^ZM>1p4Uc!p7L#gD{kS%IsICT+;cZ
zpm*>m@Jm#HnBrnmfbVO&x9}W`-}(Y^Rf5!?BuQ&*`**`ZIS6&@N+gB{;w$+A26ThS
z4#q0;9?ZM!>D@a~kfzzNH7!)=<TX7!!z7{f6@qkzUK_&<>3Slv3yQCDYk%|8iAWZ&
zE~+<KG~bq=c+cz$pNUiujz!?LO8dgCZQ%U9t2fO90Wk3a6{-{z+fm#p+v)zh=sEiz
zx%4I7ACgB!pYDZ6sor7xW+p#9C0yAU6o3q!j248UymPnk=j+vc|BW-h9ViR0+dNju
zkLn<)3aPb_9Ri!@GyW$PWr4hs<dYe)X7fP$DDOK$(U~$v>s30x^YhWu{;g!>j$v%O
zt&<uR>}<X!I#3Xtc}$to-X!Z0iL!ITiZ6P8GrHcsZeQ6~kZ*E`p;Vfe7nC*sb1LTG
zsi<ofv=?&9<hoZy*W4TZXRXH&O;WbG5L~zy!wvIQw?{l@da@<nzC!Cf?h!dou#l7N
zRdbh=3-vPKuF*CJ027oh(O)63{84uq52(YKIjOd;qTLiaPk?`+&EWpQV_F66RWtr3
zpWD?|^1FWfmVZSTyjFkT+GrI-#r8%OUSecf_N~xa9v|H<=BC7P;^^}Ha>hqCg;pFR
z_J5S@H73e-;rqsd&s^_U{cWYW@#qK<PIT{JRoxPbFg0)KNfE!X5E2xiewEw55J<V0
z++`6ZXj~6dY#Z~RL8a1Ah3U>9ocwZeNztJ=aiZ<ac>BI}VP)wzk~sJlg&#7%XD%Sh
zy6RF2`w(ywwJ_*8sIb|OH~Ci5D>FyVyq=RJ{+aoR;2a%z4nPOC9;WjK@6w4}uIds@
ziJ)2$=nyj?wLwvBb$w$cXX;*2S!W=*yo1oSq}~?xG$^4To2wwF(i6zAG?0WWP2X2s
z*DqXaD>Z>`7~x^bO%7l32e#t}k)Y%|dRWl)3S0;)JI|(Wl@DDrjbY!x8%ye#Lt6p9
z6J}7qnNj;_pF3o%8-EsidM%RpV6S_mMt~UM&~|qjEsU=4H8-20xXgt7&|a~vaBMVR
zj;5rB1YRrrFSSg?`{{!D|E_-wich43jw?q3zc#&De-tDqvk2gl;ZA^%$wbosuw|-k
zy;N0?eJq&JWUEst-baD9C+481a&T}>C^}g&SOm6lnt<jb@JgkV;~YZMCq(e*0&fY=
z=bcc{xZ%7KIypQBq1m;SdIDwUoMx2?=uSeKKm8Ib%)Zb+Vl41=XvXIxlxA2<7i{4F
zR(j1Y#<|9@^*xSvUO7s4T9%U!Vl1JksIQinw<aD;2503&4_Ka2tT(h%woj?YNS_XI
z4_#A3hYEK9QA7TD!5XXFM#@BPSLuSUg)aj3_m(-&#BOto4KWcw-nDBIct4-n7=Lv{
zqru=7hHPI0y%J=(F#7LzOr=d7v27Y`dR@XnKB);gdKz#~2z=Xpu9Yo0{M`=s)4=Br
zqZg1!v4jH1qZv4+>R$_bG;L{ykl+K1BHXB0k#n}%hdik!fJ0NV@g1$ZFMe<V-)m#`
zy`m2b58@aNOX+7D;#$_T_9w+6l6cO$6TR75QpkqtXZMfn+r|8E;c3|z+2Rg7z$V9d
zjAF%4hdqAP1>#iT&U22nz(dwdjL6ulXe+%si(7~x@4id)&nhMEJlowY3M89Vgi=Xo
zg8MV6I&y$T0J+-}-1j{;$QZ1lkZID`{LXs+dAc4!CoUYb$6uW`l0X^>yUb+GZzAd)
zyxRWRq_2RhT7T5%d})GyVU4e_?P=ArxvMti^LI~~3wF>}kuBm%Y!R^QrcNjL=5Q14
zolF(&R@V{N4GxL3_1d*kQx%&ZV6jvmoaB8NM{3W437<5Fd(vY0VYhRJu;RhNp;FGw
zUgvxb{%!q9>K8uUebxf!e|snJ;QGQ4UwWAnmH}i=kbZ<s^Jq9oSMn=O=fd`U#g~>c
z42iH9N;$!o7Qt~frs4N=rM8dXj%f}#N;bbyg;{VN4&EBiO#L4J<FHs=%OCze9J0_N
zIU#a_o<O~VPciz_K}YQHtM@WkVB1yA%2=JA!&#ib$y#h0XUp?iuDQr@lo<;*S=zyY
z$f6u&;%Sl<E>(uZH{ZvQ2eFcIw+Go9mI~<Sj|gjTbv#JYz0WJm|L09DEa|H%VX*Lq
zqe8d$uqwOX==*lY#8W5czir<)oB4`xa?qrAA;LH~Zp~Kfr;-QWA!$u*PVc(jq(g>=
ziYbRO+XeD+`dZxc!pA+N-D9M+R@lGGe%Y8{|4&Zz&5;A{_`?Yk=j$uQdh6>lmu>-0
z8-XLBq2bfN2-h~ud7?o{56hj(N1e=XOq*M7Cx!be<J!>!TqWA*vXs$9!^4TEJ#DUU
z(g9yFjQ6IOrU&{J9f`w|6}`@)6oxF)<G{?p(ebsCWS5M|3TTucaqpdG+fm#)_V}1~
zowTSphZy$KHndR$H0`8C*rh8yW)I#ueCb*$7=%-5no=CI!H3_7L-rg{a2?H_h-di9
z=6Q&kO?Ohn<mhZ9cG?|ozLr?!-HpqFC8>y(S6;k$@r+QBm(e4@!U!-32XOXn(u_Eg
zAUUOyzz?ZNk@8zObZ%!j2n$Gq1-$=6huqyU|A|8b`9ClKV;5Re?a+K)kb&rfuX{Y%
z?oqz4@VudKxk#ltvp>34HU^+A<vCpYweQ;GkZ>esYhiuRY)MZ@sPU*PX!DoU$r<Es
zy)~PasEp#bSBGlO;q&&A94ZL(dDTfJ4SQOrHiK?!hZ<P@(fji|(~Zx{t;VjQ9t$Dd
z@(8VUOILM)IP0ws#6<C&I7AJc^w=TX>RrDuaOy>cNb|_Jy6*fNPe&KP#pCI|*;XdE
zD}32t<l0s<b$d~jFg7agHWtxu^H!{uqOrr2$@c3GiOcA_a##}toIO+a-eisKZ9q<S
z<Hn7}T?8%qNR)Ba3#Y@_{JAN(@TS^qWf1ySBQsuw@%Vm!agxgioUB!EqzGn;NadEl
zzC{Ej(sUXNsR%Wr`P$tp*_(d-MCPa)+<Smj4$#pg7TqR{`VCmive6sk{Nd<}%9eMD
znVEmJe{<KraC&!<Owa7akm*vwxx_3!#&hcucNF#)xhxm)84F0~1N{1*X^b4!_F{%`
zfUoli8Lg;^RN))5a^FfP<DERA0$*jF{{~*@l`jv>91F^xR8V(a47Bkpd5=Z7dvwq3
z6$ckvzadnB6*DAPB|E!8v2eWXtW|g)({WkHzB=U_^a{A1Gl#D!#&tStK~Xa+>h8Sz
zSnWYEQHAKxl(O&Z^(eVu;@ENvw?qYYZ2szy#yKL^fMJ363(d<K!(<H|V8zeaxg>*Y
z^fd<E6Zp(D>>-0(Lz|cSvu`=2<e^^kWDQ78c2My1@L4)~aBcer%oh~|Jf#*5fbaiF
zcyqcqj+8L0E(GHNk5_{b%sB}=HuW$5)6|x#Ga~acy`@kA6sk2kwuSOJ<^av^q(6-Q
zn~Sn?D5Z%!d!@;PG|8(8JdN8PSTfwyxKz>E8Z%1CR549EQ$}Vz2tH_tHy)E1jv|$J
zqK%l}uu%#xKE6Z{cacd10pfrhhLb2pItV6BqI)8DhziLCn9MSKjD@hg_ln|H7v!{6
z^|hAHc^Rk}ziiC-iA~gEhF4xwtt!I=Y9k^TW}H$^i%B(ULZdESP9kJ$={gC*KQ+Vt
zHycoH0l?k*h7Jsa`s>u6Vx;$&86$-$08)g-(-D39<6Qd9Ap{-zgbuBL#6+B9+2lSi
zxhk^=xJ?a>9Zu6`7Z2d_u($2H*>x-{^6*0W$Iv7ev{83LK{M%l2>8C}#m=@%eso`W
zg~;{D%vktaJF67N%EF!f!Htby*Pm{<9|^$1G&zT}mF#V439@2FYQM@D?DI^fW>Qn>
z5<8$`MSpsmt#{!YMgU*hU(Dr}YM&SshSXvPe`S+Dq>tScoRI||EnsT_JOBd^i&U(m
zd;BhrW0R<WD0l8~5UJ&@*Hg2|b1uK<Vng)icO<%bUEKT@xr%UEUJRVj*5$DXkb|B-
zvcRk*?D+w}<Ofv2BuV0EY6BWU{5qLDe~wz{F=60zBU%{;b^I(z<Elka8P1kK0FmqK
zf`yW5NBGTJqu)jAe72w+*8EsOtcIs&La+QoL3V;`EbC&fG$)gdR$eR@jfGs^py}2%
ztF?nQT%y7=dBX!JU89i-@CdDyzzDY(KM?fX!6mai=%Mo~KprrxVYioP0M1XOM>kOB
zveJVObL9hCg_A{hvw0_9LCr6x11G1?Xp_eTdHq_qMjgI5@!zox5}ub>!;boRB&MDu
z$29J68sf{|GF$U~YP<mfFs6*tD^!pc)QH6d6{D`}?AgxK<H@fc`y^x{kcD~zSy^Q5
z=Rn{4=)+dVlEBzkubDY%c^-1g^RM>X1f@9>Sa!c!gakp+(lE&L$G5!pS4XnpGKcw7
z^hjGEvD9DDX{3+pTZm=(D~vle3z+UC=&jbt8hxGtlBaI1d!#}Dn_1w{-&!(SDfZMB
zVQXBZkdYLKCdN#XZ}SV2ji?!-K|BE_if9h|Kd7%C_nq?tZucEv-|0!hfg3*BQbk;v
zl5H4wiAQl$omYd+A+Ni9L*UT5DPN+yjkdGLargB{$rtN{SV~hKzk<)XA_SegE)L&D
zEMQ|x<!agk+8?fnjIxU{%&B3G6Lzrrf5<K7e0qHOd>&3=vpg|6nDK+k-4onexz=yx
zN)9W}^%eHW87&64ZA-J*h22*y0B8aId_~J^-4r<7A_6d$sqU$${oL}bZr|~7ocTN!
z3>;IQ9ErYqJxn+WEYyxSDFh^lRd?hC3kU*-fsB2;I$r<`VoZ%U+;X%adC}3!Ytdqt
z$^8W01KceLmoepsuNeXtIS|)8<g<A7O$psE&|Z=i2N;3ABPwR(Pqj-g6o5KWisr5N
z$T+nf@qYCVb_oZR@$cF`sFZ{Ds;0s}*icGf+1#2^!$#-oS!F*yMd{ga<Ro*MuXX3S
z(sg%fNp>OJmFOKePlob&36^G<-@2~iXG>OccL1+Nn?n9+zNY+xJ&fT+5hWmL%GU3}
zn&1tj*-N{=c<;A*^g9qT=r{G{731Kg%$5PccYhns2w}W_!$--!X$>5#MsIna*S%@n
zyw5rQ(lRCMKI$kx5hpg8m;P`u1mmpO@64r<-G{1)zC18L-Qc6;3+`m39w4O&-=5a>
z=Vqgav)H*>T51}AhR!<>uTH{WSTSgcz-u5udY}Y*C`X7wRF|1ye-C`R@b~HFA#v{v
zLF8-wFiV%x5RR8)X*@c}zFM2VmkCHo-_HoX7&Z1m0(m9KL^EP8Tvolw^hm;;J@e0S
zU+rL6{Gd({ON=~sZm61D3wmlz5e!z1@qx@r`_g0eEYYL?C)rkmy~$Rr5dPmYp_;QB
zuR!<f7qrE+86ankKhZ49$lq=7iJb7(n3txapC5UTY%c+68T9L6(Gu$hTx{E#i8F6&
zSz+fH8T7S=&UTu8tnBEtD_-elCpP+W_F4x}!XY)Fl`659$zrbiR1!#p(#lUgYy3v(
z!+y12ZJ<pMvEcwF%<%866D;$&WTWNFyALW)jXCjqs20pKA5XrddvkA7h`(V9+=&}B
zM0#SNcyC~NM|OOoq$uZ1MhYW1z!BxJ8kT!mw!UB2!{MM;ZG67A;UJSgKeHoD{xlkJ
zcg@tTT3<~%y;*vJTSN)b>_ygDV|--%4CMAq+d-41+`r0$En!{GQLloiCjnvFxYY~$
z=JZ8Yguyi8lyWe);exV*g3yzrD0gmfX91kpxf^7EHKUuH0=!^DL!jGGqFNa)>g7^d
z4?iQ^q)sYrLNo01*JSJ;+W(bG=#l2<x4X(rcF}fghL#%$;smB$Ezi`qASIe9znt(A
z%cEpqM!>%l>`0j;3&7qs+~Nxf{PuckWe+mg#02L=(oY<Dk?d~m<Vv%W)PvLk0<`|S
zl6`g>f*CCo;qpUX)V`p-<(1De$<KQ;8<U(l(Myh<n`wEI-6X!|5h&LG$JAR!MftsN
zypqz0NS6rGNFz0bG)PE?64Ko<LkS3|lpvjhlnBxd0|Ek)GlX>43>`Cco#*@epR?9^
zIj?3t&)Uzu_rCAz`dn@D<-I&%l4xTT34(RPzHd|v*LJi$L+10>w5HF9CRS3g^7e&a
zw_3g1Jbn6GU*Uu1Q*W6|RBqr|jkM?T2+i8uP(r_E)l!V<=T5|DE68JwleebV#X%Pe
zI`Qulg5eD@CugX#>4V*_{FNV44TKQYl?u(#KXCnM>&r&kSLt=$#odod4hFw5R@VRX
zIiCFxfY_*4K^<LcX5`8yoojZTgo&pAEvu?ePjx+gd|D|iOAI(!%_m6qs2n%fPt|F~
ze-gWQu0DI~waOiMalmkdiELWmo@!dlD;N09G;R4QAs_bfVvp{x00;XC<2%)I3aKD+
zDNx*~5AQ+B%ZkCi;uBv%MD}Zh$eGG%+EV89Pk_Eu6u1C1hE8)Cy(S9#TMY%FH+5~N
z+oS4WjpLDEWvE(dm?)<Ig(JAni@@oL3d{396*!ORwdILDK7IP6`^uc_6`?DsV(v17
z9;-FRrJLdDN8x_eDNe*cRi#4wPp(h8(Ok@>u#5Ut$R961ulb&7EBm=XzH5g!WrC9a
zrNQ9D#k`afe9S_W8?HF^Pta!_{=M&RjlS<Ld6Grm^&XzP3d9k8+e={YG}tw946MB2
zz66;SK27a#mQv6I=<ap?!n*&rzl0?$>5?u^GJ-fF%)Qc+IlFlD21Tz$P-UsFRScpG
zIuE~$&2YCI?*^cflK`cNBsZn!0Tn0Inlgfp)FJY3v}i&B5ZCd!`jg;C{iGz^<{zF)
z$M2k<P+33ur`8$6`s@pSBL%73y6#7@5-&G{W!IWV-;4h^>9S=<LwmX3pg=d88<nx8
zb>Dr-4qoqle&VqM6Y2Zr<+NV+%!~;S;!HIGdOeh}`*b_V)N?KDi_58RbF~9R1~=IK
zFwdxNd`WD8*Z%R5J;l-1L_xx7q96GK8QyzPqrkTq+Jr&+*33<^#f5WdX|Ks#3&6`E
z_fQ_f0=Z=zyy&Zy`@qbqqsWJJNa!CxHj2)3Pm-6#oOkYh_!yD-QE-kvtwk@Mw-R<5
znRlI?D)}NI-zKTy1%2lzb7a|oTwQm5+yejD$2MA=K*@H0GPLW9_1+dTg~SgIshSc4
z9D8<d8tY&>->1pd)mE|`y>b0wO_mAx)~Ajych=LUjMqxbaB2iS9(y;KB%O7Q3o06u
zeU$o*lrZvsYs<@Im;$PSmeyvp$x<`5wI?VNiqEO60yEzJiWIMU>z%PrK0P%q(9*!E
zHGUuty}t_GVRJcIyH($R1FB=EpJlK~(h6ep6o)&@3uK;8zBel4(Dotdkj&OY`SSm+
zD3Uuqp7tmN5A{iGWz6|{?ylEWdb~gAmir2mp&oi!Y%`lC%D35K4_BZ$smKZZ20Huc
z`j#g+92(FYj|;<nh#B^>2>Njp1l$1L>4Qs%FpS*H=PwcZ8KTI=;pgA-zbC0KL2Nn4
z!2ugHJYCE~G`nLcjNFmQwKU@5q7TF0P4TMQMJ<q+?MD>lO8tqwl2XXWn19M9j!Y5e
zpr0>>B1@8*iOLlpO?t@m-B`kNXa>;FE+l~-n2IH;$S2zd!-C_apSw-FbW#h#3reuk
z>&dn<?f7!WY7l%-Cil~KTnsnvgjbWa(|5?hiWauoxXnGg``4in);wpS@PSjVxIIq>
zXp;YT6Cr{4{~HOZhhdA1+~GqStA{)_crf^wi=gWDbA!Q~7VkUKPwZ@NV3Jm{E8LOo
zqUc9|=^xJ&Re$pQ2q=Yzoeiv?gcWwj(~p0Byz`ovHl8oumNvd<LW4R2p`wR_15P$t
z7VQ)x>zNRc{TwCf^M=cT?Wrr(^V3Gad%a^t9|2wdrGVx+TG&^Q*4zG$7A0}Ho<2gl
z5C<jSzL*h#C5Yw?COYBD9x{s-_g?Qmgd#=qQW#yORpG>tdjQ_&b3QgvdreeT!^_MK
z!UW-!y-r+`owrc1dcNIa^VM92MxQ25*&z0j2(sMT1jm=@LMYjI*glgpPH6FUGi~UE
ze$kDsYb6i|O0StvS9RH4G3l2duRg3(YmO#l)nk;~l}3wJac}C4qT6Cyo2%0mbq7kk
z0lZ(S0BNid=gQXgUF=*x^XI~ye|=0kOCK&{IqBPeEb&o_!+3KzDt4SIf)RC(kL-RW
zs9e9ylW)js@KkG{Z@x~-rpT7gv(6X_a!$~5ntvbd0D->Hy^>3u76Vn_4lsh!KNl+X
zjbV3t@`dTQ6Xorrxk4_6``mvvrFabm{uKk2I;mMwCdw}RDjT2)>Yzhoeu>>*YZgbU
zZ;EmEUTvw@HN=x2VXteQLJg|}eNW;{!HEy<lG<ReMs^xJdr4{DpLEysEpfqrD0B@D
zC`z-I02T;IJ(+_<D6*<*ftv%P|5_($kO%EkjXqQ25NJ6x%Y>+_^Ym$5lOf;v<9BC2
z!aPIOA#}!MJ6BHx2pkJm`*l^@q^MVu{J6Uqx)L5^qUaFc>iK(w9BCmWtW__*qdJw2
zKzzyf%KsF;t@j)$lujo#eA&$0ba1d1n5o0n2rZ>AGNd7PuUIyAp4mG1;$*!WkS(k7
zy2q+NsK<S<8nl$t`5Uvty(mVi0&o_<^Efi)w^W!V0Jbm~k_zZ~QVDs<^fQD0vwBb7
zJ%m*JeJQbCro~Iwr9`)Ayb{0`d8pPGD2z5oWM|4?oy&q5UeW&xfSi2qW<*SId^ZR*
zXnvz)fJ?RAUQSc>hL-8AIw*|wljFjJRXxfG%Wsbyg7CVSSB2Pr=83S2K8sMfdLR7q
zj)AZA8|Mch#SrZ7=2AV)v}oENR6n07z{{B|d-AJ*kP1C`9*>XVh8dnw59_2xNee+J
zn1nO8>lOgnr2^J2U>e|uK)You0e<nUg!;jFxx5hJ1n(o{9MfeG=8PW&4q8idWqn$$
zcNfy9_EEf2RZtSQAc>XsBE;CncvxlKml}^?6yf$#E>pt0o5u<>=yih!!KU8|2+vO7
zT;NZO7e2hIEJz)8d0e&mQT;KhEI<clsFzF_X>@5IF>c(umKIc{yl+DQ@pE2p<g8M~
ztf_C&+mBHVEKE2bM&xZR1Lvmj{=ByIuch(BE-N+u=TDt3+3y6})~e*%#aU@rObIc*
z?Os1k8goSrf)Ot$nd25txJ_!OU!H%h+t%<}`QyPu{p;NaS(^?XJV*}#WOi4y-oUGG
zeP1ltt3=dqBSz*Posw)`a`GeR6gabZP+$_Cp_rWIJsvCQkP><iYei!B@N-+@!z`eI
zi#!OQqTm|Kh&*%4%Pp9TZ)~;>?}YK2S(93ByfV%8W5axbv1VnPxdyI}ABiy=<)ZQv
zMstTRAM@)ZD3$A1(64@+$kH3}yeZRS%6*l)@MlatKG@ms72q%>F0fSOdV+d$`pRW}
z^r8Eoii=L&X70sdH_|Bv<{wsYJ&ibCtxMgT%z~u)krQJc5)4=Ni{d9%nibeR2Q$x+
zjF<iXr`|XajDAk9G1+UxbvqG;=iZPAlcsGI!9|fvRFv|cT6u0+UI$Noo-5-L?EIV!
zCAnUoyV)9<TUUBQrz6D}>m<|vnif!5dqn-UUzPpqLi=2|Jn`?=>DUiDf~<NX#R<yq
zKR6sYa%X5qw4LL&RNmY$M{#f}L646M9DRcol^WTPn2cEpv~~fGh|L)E4bx;|MGS<(
z+A;W*Py0<*>|gVIt49B76;>0I0OJzX#1EY7w!B5ZZT#oB(hKF+O^oh>VfxS}XZ&!!
zXs=a24w>kQ-HeAYOH*!gLuwR1hp26elFeSu_&o2PIf6l1)vSjkayDgAXBWB#8V2%T
zNKrxp!mzsO;hytOa{0(=Df?X<(&D?&HF>EgK1tUvnc_PAT!oo90}7}H0Zd`HGqn>p
zn}bEZKquNnI%V7xoIXmvIf4vDtT()~@_}67g0Z!ua-mJ*c(B5Eh@dt6h@~gk3HLH}
z4`Q)o%3%*7*jG`poa;-f9(}Aak1)=YuiB<r*i1%jFX#l?^SQ~Trs7$+fi>#XmG=8u
z4gYGvdpq@<_7F5I`Ngs?QNC*WWiAuKrj-~IeLsmayZeuF_fPsg{TE(R$x{~{1&?P6
z0;(TM>rX%A?I+cE8C3pn0kXHjIwSKI@qeeQ@-~yDr#LVozQwEeRQ%x%Tp752G-dz`
z#26BQ?ICJ~g=4F*MwA$a7{fG67ZTSYymLP?jgo6+xQgG2?qKUs=}a08TBdk679f#G
zw?r2b<EN5t!Q?fOn_g&IuKV$&EVVVpwBGDu8`=0OvC9(bX;YF(xf=rtlO!>ycHn;A
zXXu2;)^81$n~`k$Z#n)ECGmF!TL+B#$iw$khi@`8I(I(c!4{T}d#lL;f#3z6bHfr<
z9Xj|gd5r;pNK<x8iox7OJhpS6C~WFpOfPxdv%&iy4jCs#e59Tj@JM;RWHr^9shno;
zyzdloHDE4yHJx|iHz$81;|G8f(CGTE3-FcIjnb`5$fxpxn=r1>tTj~sKybbc$ui~B
z&Yfuret4Yx;tw`Lu-_YFMZPT6wZ5Rc*cR@vXX#{$+#iItJ1L{RA0}f_iDsv%l8PM}
zsP)b4N3Rm3OiSRW@^BgwObrkr&g82;BqSF`*%#ods;gY~)etk6$nWzbg5EHcP+~ZS
z7t*1n1=%}e=(#647AJFb*YkOwrYGL3u!WsPSg=J{SoQ_+tgGqPA;DnPlW^u$RcLwx
zgjpqFKMKO?rwiL+{Zyon*^OKM%k>yFb~&}=g!?86CLg5!2LEiKnTHr0dWmk#TeNil
z6@09jDcEv?m0d1l)LK%Y{-M?x%N=$uRjOL$?zV8Z5uec+NUxxCjKpB4MYqc%cz^xA
znALasqwsdrvU&f`g%Bc^F&eX;N<R10hAH=O5f?PTU=);a+X0%j2|;qbpzGguc|Ydj
z%c&$z{t^V~pOpRru4}>qoi<B7sH`mzuDW`jhI|-d@hHe@DAqjmr~Kgx9w=(yY;Oi4
zHd90Oe3S0fk;aj=G+CJz6B3pK9;3tP?-W=|U?s{bE7d`oFXkqOxuP+fKGlw!Z;PsO
zRF(B{%l;8Fj*Y5q8_W$#@I>FkRp*G9ReO=IdmTX1D;sqTY;jOyxf@4pI~-AIA#8Hl
z5YAfTnwyIB+=*b&9E;dgGmkz8PF`W0qyEOaAqQ&z3wDo3@qNm{oViX%S#S)FJBnjd
z)b+Vl<wMLkK*Kt*AVW3$$|^)A5jA$R3cj({)65J{dfIQ4GW>|u=x5!;UH7}hMWyQU
zSo|y-8Y*iDhCcsMbW$fq{-9OTq=TPYvuRYdYs%!yqwZJVbeujPVN1oEF+#tNnqV<i
zM`(@ulj1R;;;5cKbsFG%>w~;>oVjtFjID@lwO58e%7H8qqF%O&2pb)@+A;R#3Z_VZ
zjdsqW=AtJ#MLBVmu9-URN5Y;Wfrp5HZ@Mpo;?8j@cX*b#3XmVDPtJT0*T8HK2rGnj
z(8kKchsT0OMO5Mc|3!z6v`}(F%%}88JTBS;TP!8rKvf@pv>tq`jJVE&PRm1$N#BeR
zKtxoZ7G^+I2}Wl*bT#Z|^v~!}bP^Pv<KcCH4x1d~?yUePI|Y)QsNkC*jr5nQ;m1JQ
zW!>X%Agrc^pa!~D)7@&eW3zrF>H=<`Lx5R0f!csp3Gf3fz=@|kM|a^oJ~HdV=}w%8
zUCA9y7SzA8uqP__bJOyfou&WgWdDJsaUks-`~vQmLQvx;iHXvkQH2Emfh@W=U%mz8
zYgAmtNo5SSRbo7LayTHjZvHU4D!?uvT`s{)SN`pI&jtRdJ?pzQyMn-)NyO3{M}Iwn
zxz1DQwjX?A@?qY3MMSx9NmQ&}wbI;<>*+H|w6&9}ZLDoRiri`bgWOj99{}?D5z*HV
z4gDiT;AX&Lt+l&)!>iq6=@Y87<(x(pZ>1|`@Bqbqw{cB}nBYmf=0^?Wdn<TszOIwJ
z3MYbC0*v^>H~$Ka%LMM<^9cZe&+-#_ozcn4gQ409y{>|;9kCssoKP=b2VUF=i`!~^
z0=LIsh^s)PXpuf^?%eQhezcQ}7-mcgr}2SMt|M22;0H#F{Q-v<&+WO`m%6=#foL=m
zaZ~SESL3w!_v#0w1<jjZ-A9wWzmg0DiKBcM_JUbCZv(n@59EO@)XNF^7@$ucZ;>rX
zjA|eG1Q>;@)Kr+Tnjl0{)txroyybN+0q>D%+)IAL*Ekxg0~DksInHc-N#$0ngdEw4
z0F?$*;~@`k0+y6V+(XRU7Ha`Jhdqy=XBGnaUpGh=7g*#IM@f28T7>C)I@wVFX#N{l
zpU+n$VL{XTP8!RhNP{$$!9@4W)1L-+Mj5Xyirc>;r2y-fuu<=|LKT_9;pgl}b%9d(
z4lfm|BHT&_S8ghfn=hBjNl;9XizL*MwX`kO$SRChAkPN&|E;3IphGJjpZ2xnZC1pt
z_>G8?D6yhggvH|5!sC(VJABM*ND6?j3A{Fkf>uu5=)8v#$3=m7pMTD+&~rcvZ1^=r
zrI7-mE%!CQTI8~@*0u-WSDQln|F2AO!UdPdf#7Fxoav=d(r4MKPvb(B6eCR34BP<Y
z)f4>t8NyhDFn6QSwA}ZyX-tnkIsK4w({kq)s-i2~tA0&5b*C-Z^Y}3AibILHR1B}4
zP<pv=2MUBI9lc3^nbhk0B7`cerS2rHKUa&bxD-&9=8aFwsR~q+7<K%=#mT=${*vs3
ztv)1D9zLtlJ9b9|xpGzT$RmcZJi<@|0G>@)@H{UG1<KH^>gjB+iNROgAbbo@O?A2M
zi>2S#4eyzmL+-Sd>VL#Ui#poNWvLYW=SpqASW;9o6(_{|C{1NjnjEm<vm&#->38P3
zTeIRe+o(({4JLSTb4d4N8fEfR7C_?Ai~fqOtCArU`4-v_q@Br@#3+#EAX~#%$pl}`
ze~a$<`AkpB98qVr9^kKG*(v;l&()}8ZIT^Hb?`G=)RSW|pyx*_=c0{A_|on>U|{VU
z6a_E~+^UZFb5XUcl_X`1R0u3O7#z5Tv}oF8YQF0G;z+`LDz-Bvmqr#QCgR}o3PWp4
zJuyXr@&0%8+eWO@(*Z!ARasrmNGnOA?iGO5hlyNrJAa#4xF87$c!YV6wm$;XhS`VH
zDCE+d+=#z7K4@th@8$-qA1ft6l}Zge|GZ`>8ji12_X*j3lB!hjB!jS!YF#UpGGI8m
z?|k%c$^Cb~8kFps&qQIVa-b+?@0w7N0OI}kNn-7g1%gsS;;ZpiM~B1<2IfV9bip$(
z<V!+G!<#|~X~Ys37j_r`pk71_!t>aQlXA&9gZS^|LkABw62ByuRh<c=)aixl9sT$q
zB!>|b0y0bmm^ZK3cE+}*R1NB+ns55*1cA?X>T*oRVw*y@l+L_I3c)rqFAPoqFV9Hm
z;}dWiq=z4_r=ZQCp$6|JId#RM-vQGA|M|OEJkyDct&^~fs!VFIbG<Mqrg$i{?D`?)
zF4<)WJ(xXXEW+zVe^P!wMTfJ)jR_|0=clZ8ZqR{#(f7TYK8)vSaM$Sab2_=YHF-EM
z3u^NwL0fT~*S89QD}`Jt=v<?;ym=m57O%be`iIAuBQo&G9<6Z@6UP6UE6)+n29Fvw
z<_U-X3J#pS<mvQ6-UihES8M_LoPV_CMCqfbY!=j*g_KPZm~(}jX4qj47n552@lWs|
zzK<e`lS~gx$i5Q?hC@q@&Wq=TzgF9do5=-9@4S?xK6T3375f>}8lRxxIAk;CcF}Lo
z2ZLZSA$-hs;<0<Q+4jUrynk1;cWRQMB#9SjaRv{a6?pA3u-C`Re+M0wVE1gMmPx>%
zP5xwu`_a6A8~jyU`vJIC5V*%zLorqLz9Ixe0GU(Ofu7fGq4tRpn&|K)Xp&{21vlb)
zYJ03uzit<bgtHZ+t9MuSI8bGFi&(HIENDO-F@$?^^g|v)f<|)m(8$mQ)5|5l7A%+|
zA;cs2M#fuXivmL{Fz8{h?Sn0!I9;9Jz#qGDyBt&*znhz?bwB()kmRC(fu4!RNnU{c
z!~_r`KdGH|Vs3WL-xD5cca0;*Y&_U!MunXGGnbeH*k8Zxrt@Pu*E#_Z8Iu#|E?sfb
z&67*q2pAOFL^(a{1g<h171IjYj8;!*_PP-B6`5)2dg2Moohpd}oOh8A=N4j1I7JKd
z$G<|2mn-6C*NQyr!`9zydBx9+ePZi38c55`d-;+jyS-PKi_YB1c0Ktre^o!ypEWF-
z5lvaya;ta64Ny|Zsj&Bn7k1+8FRa%d*;Z!5eP{&p4>?dZ!Kin4pQzXLb@Fax|G}v-
zaSK<o<wUU)xhV%_!RH>>uu*K7J1(pO=;zg^W5Jg82_aQRk5O)NJ!PI?1s%G;^tM6R
zIq_L1FBR@#%~sHFA@6R@@FgPn&+NuWW%BB0ZMP%0dob0NQ&_T%GYLJZ!%nT~Mg94%
zqnEt|z35|<7!-?Cu}8(a`D9D8`l-X?6LBSJDzgVD({(z6K%+S{PvT*pLZM>WX9P<?
z7{eRRkKYx(K|aDDKWO|&cErl!uk%{y>h$iQ>%Vu$R^RSL!9E<Bo7(L7rLh09b+{LS
zszw4bU6Rj_WP;YJ4chT|{)Pmoipz8riDDx19>~pzohh~UFFk<K{7<5@eolviS+GJ&
zFJ@-Jg47t5+ez{yY*>#obU^(LEx_;vpesc0(65q*Pb1W~zhi`Y*5e`;WBQx(N4&3s
z14g++a=hM(m)6}pr^Ia8pk9K;o)%T&Kt`S&P2U_=e}bkfdJeHN5`Gs=^miicJr5p{
z7HZ6xw#j^{dA-|O@idp8IwxqZ-6dbaZqaqcEpI}$yc3J#hlvpl!*jIRHZDwQ0Rjh1
z8gmjmGjq+iLZ_$^C1wSKa_n*k_zd14mfS4$Eba7*%)lBiluYE>6b-z+v~WCMOF{;|
zkqUyhqakx2?yjU?^soZY^sjI_L7bOfv;HWNvFSG&d|ynGjwKTnC2p)(8Bf~SuU2b*
z5WwrvdJy;eZY2a45*fM|t}?<lvM?+-H@oZ*>>@7^d$fTAnhK}J+*Bi5Jj$e6TmyYO
z7m_6+ma-*OK#+ewDKVqHmc*Sv7j1uFioY@<)rR-hL0-oX=l01zYk%1sQ#+&nZTi~f
z`Idyk(5^f5-;ceYmWK$lL(zb<v|)(~;d8JWO5WUPToPxp-M$tWmw3lqYT+yVDqH!x
z0)W~W7X4P}1whWMXJ+zDY8C8a!T+MTc4PmlzL7oa$+eWt9t!#qOsyMbFg0k>eiLdw
z=>e#>01i|=O?63=@5P1bAN~T5dL{%K6dPg$M_(nZI~3>M$n+N8$O@pg8A~g!40j+#
z2Y!Cq)uQwI+a34#h>5JnwRL062u+~C4hE&-12LuY9{;PZL>yFEdZQiuYMW}A6B5%e
z44n3ALW%ud1-J7$NJh_p9|ab#=XyTJBh4;Dad(vpH+8d@acXo1<(JEBOD$p3@G&Nn
z#F|yu4iky@vB^3-a}T=k7L_WioG+w#FMrtl&3tY1Jg-0baddzWV=kqk<DKZ^4Z4g8
zhv^Lb!Sv|t?F_fq#T42-=ai}|^|9LWfy7Iy0OQvL$bMn?HR={;-Fen`(pgs>82K9!
zCBwZ!hrPbQkggrfUU1|gRe4{Og3Un3QQSfj&B)7SlSo`?>jOH31sE?0;%mT9Jg{KM
z&R~KnrWJsg_qrN;!l5_&WC#!h@N=vG{!|k}2wDB%C82P*!9{XYLxc99N1M?T-Dw|h
zHd&cn$vKoAcu$0H%6qw;mCS$r?dx;)>tu^fAzH-PKZy=Iw+@1(5<LBof)Cz6_A5$7
z>zSbS0qsO#dYu;8mq8PrG>Rgl$C#p5sqfN!CVq}M2nZ_Pa;^X2;?us^1KH)DII5<(
zsrTq%|KbmlF2v@^x8u7%6n)}8B0sfby8t;SmB>PPP6`3u!JcXB&`}b;YiWk4d0_(0
z;1}4|;b)5r$*-Igo~h!^`yP(#<<_R0|DAu@hD_v~zb=jXAsqYulH{|=om_NRN<c45
z{s2Fvk+-0cHN~2vfJZypJ3uIn?LUmqGjQ*d4cGS&u25_UW>Ta&$Ehjh{L#^7y5NXt
zfT`zB$*=Rks$i$iZ+dsSk=@@LYEbAA<4?}?*75?iPB?M-m$`JUPU6kWTtbJ!`N*+G
zO|m`kiACmBvuoh%(4xru0b`Fxm;`o|AsZ@J5y7AjKtYnmr>68F<d}!;$+uo<^#NPT
zW&Ns!iIZFvE|SySzkcs-OT6pfvhgr!AFl}w>mkfOmokzvvaTir;#~u5awIuTB+q}Y
zc$7AUm3H^s+s0IMCYwkwF$Hxb92rqy7@ob6<B32RE@MpOFq64zqjs&$P=_9;kR&Up
zF^9*l#S0Wwbucx8obJqqrCzG;IsJHLjFK4p+o_%!3e}Zy^U>eUDC>4@PXt<4m0No4
za`BJYde0x?b?xIaQR?Ax3YE?w^5edILul3aW!wI}+p{O1u!~&=?NhKoX<^xEU3p$j
zMb+pmAO!IwWWU=VP?V-yj3P$-FM@wWyC2Llf_CHV#%724vK+Rcj)Y$MsA2rU7m&wo
zw$I+9Iq82S*ZpG7ul25SF=TW*urm`$X#w9`ttRJG_kt_gTBiBa0{PD&u~Q?~C3!ov
zzEZcsQ0Ec-z*{8#MDPT5(UGH0+J00&F~rnm%X;t6oy2y#3Xp&!+nqhB`h$*)BB4$|
zk_J$h^`5wG%cxF&J3e9%3}QlOcBbRE$(K>d#y@Bxs7vNtzjHWv)w8}+n3wj#-J5Sm
zRFa;8bE<!cr9n)S8R>MK_B0_XRg6o+r{QDSw6>z~b9APc^;)EgniprJ?eAGU&~>Nf
z7rCk%Wce$xJ{NM5I9uKiR(b%#b1AP=13t#kGg#-^9H7!S<`B%y)R8~Vuo&HWpnp!`
zA#rGQQ+dwZvm~FLeyE%a@eSO5TmG+UB%8ci7))tNy8EHZ5!p(TWT8CG%5-s;rSq;#
z8^==d-lJ3LRaZR}JR;FaV*IwtfaLbn1w(tB){-`Q?|f9AaM2=CwFrK6&w#;~It|6s
zX@cSOGyC9@N!LpaosI<pnZn1Hzt1k-|0TY7f5a=(h|_U!PY=NakuH{1;soY~3!u+w
zF7KY<-!EjQuGy~t35@ZWYcGEe|7JAOH?6qbFP)p$y_+xh>1F*>I1M?vF117M<8)^r
zBG|t}x-$TAO_R7D56|l^XOa0#sqLP?MqVCXm{9R~Lcsm=*`dsbrjB6zL*LLT)V{Wk
zzRqzI=dn>~e+Z-v{cVy`W{VJAd3fjxroQdCXUr@|PTU0!K3&&4ge2sU9%s<r<6MQ_
z6Xu6|31P$vD^*_eI?>88`2FSt&8-a+XBDW#J4>|t^DlV9qb}adO9u~J+;|hL#{&Vg
zs_!yiES6PWwlGYTAN~zE-mO?wGv60j@9@O?d)Ha#{Xk=$_Ybt`Y>X21?-@$W<&&Jl
z{mO^D@b(jVfd^;o`J@8Cbp&Xi`jcX89rin%mah7*u<gX)B%1*{_znuxq{{|9^*ho*
z{JZ*rRmdtz8!d=O74t+`qpV+LZ&2H`Xsz{z!=kxgSx(6A5%B{JIsMDI%p)}m#J9V+
zTWaIQ5Q<j`E^5mP)v6cHk0aK#KX=FEN<XzRCAod{klpQa_BmqWxIGaM796o;i4N<g
zYI^&ndc8Za^U4uXL>7_`$o4AmQA+bKvNL+n{?wQpi3AZZtMQm)>-EWqym6^qr-Rsl
zZ`8$9wE2DK>wW<o7Qp#x5BPVMf}rwBjI2#HIzXLBx~ioq+8uyGJ@4Rg|EFdDf_`=(
zzGj%jtVZn`KH^<(UvfE8H9N_u{|WN~=Q%VPGkc?2oh&M<u#&319WvHB%PhH^x-G{1
z_T1RSmGzw3p9)v<_$-e|@yaZS^t`n~qR2N|GHO-l9q3!;Tewxvkf57tgb^K)%R;<9
z%OmQ16d!eU^#)b{Gp_;L_T%lkPp)(kNA`1!q9oRD3~VLL<ow9L*Oe4_n)B4p<-}Er
zOufqohiIsbUB1``YnJB2f0zM{w}i>fW0cShuc)NhW7d(~g3!9wAJZ5UQAaO6_uwOq
zVeNeR$YN~%$Tw5R)hfM08KP;zDOi2Fz?<s*eBHvq3_G^9UcTcjTubFKF3!Ab>Y;33
zoiW7IA~tRX1V#IoLs+oi`&6{DHpVjJ9zcCp`c*ig*S|J)X1r3+ucMHChFohw!ux#$
zauWRoa$@}9`9lGN?fEuKEMSbm0z{K|!QY&B0eq+)h$@>n9*%Wb-$ufJi|p(4*X@tW
zXZs_eZr8C#Z`)$W5+Id}c#hW#;J~%dXwOBzQk$EKB)jofK5E{k36tJ8<`~C?z~$$G
zYndW}2eW~dpq`04>_TJ_QefjcYAYza5xG9lKiU-%kQ?6S4#Fc}2*RHO1po#t!FEq-
zKXJKyO(7Dv{+1N7(5<cR$Y6fr80F)zQu1)Lzj_%$k@bxZe~Yq7X0DwV8RCD2n-gG2
z{5D9RvOcu>`9GpxED{vwO|s#KPdb;YmRY>-gN_0h?tdSTUPp4SbWCx6_uKt8cGaQp
zh4%a9InluHHFWyR^L&wWt?QvqpG{(ZbPS6m9yutOhZ%wr{Q^Hh5C$Sp>@r8tyKl5e
z@_a|kN)PAy&)tL`zowj+$>(*};=T@nsm4q(ue~%&+t`@}f-m?E0*_wRdjzmrSV3f#
z4}$9*Wa`82WcoNKa<><voo*I1k-)1ReV?Jz$KS-9IDY_)e9;|l7P90z*}k&b+pim>
zDdCNBvH9>d4~JFLKD5tGg#_$+Ol0g1HudS9%z(OObWfihv>j~aU^~c$Zj!G;qO@^g
zo|8{kRL6w!L=QP5@^HXQ1=(b%vU9VOv%OiFAD+!8sZ7YafpmM6eQ$Uk#co0vKT3#s
zlD15$Z|Y4SQum#r<?k`RdxxtB$-xm%*zWpm+!jl&UUz)?RUQ|2xOU@khxD3vtKRUq
zT7H##4;d93fnl@Hq`-^%9Q#YX4q-Ukb3c9dQc{>C51e(y--j~X1qr+j?3r~0yKC0V
zAOq{aYs*woy=b{33BE@@n80y}RxaPmIN8rZk|-N5`4t844iD9xEF|_jefd)KhMw5l
zVP!MH&1)Qwe{8KnrQPF6f}9)?rv>9@q*S8{gyI$_jcs%Pw~P9eWev|}6TwjhOLt&R
z=qJj2?S&z7LJo%XE!yd(u~UC>ivUCMJplOycGCtvlTG6?(uwrD0VUlnTZQ8f5lPwd
zo3V7B-^@fOd4yx=yvXEKMfCp!7E!YgI2~6<@i(vCthFfg2H8;$S&fLDkBWQ?OZ_}(
zG|3ap6W5y8BG3ax-zJZtHLU|en^ygMDk3-{9|KhP%qdVfv)sezUY>ltX+QTjd|O3#
zUpOr~h=Jr-{LxEzro{QUh}Y<Ihm{3QZB^tQ%QG7$-WoUViob!CLLWZgR2T`EcM+c9
zhU8(6imbP%M7r|CrM0?p>eZX??uNXU8XQ*kcy)SBjq~CrUU+<FU~VU|DJ9*iGcWCt
zV!qp*cSy~tHweC(fR|L=rPpxzwI?xn9C#_%t@&cb*YLo=hEN)%sX%(1*3J)k9z7Uw
zpf{IM`Gqw@i`$$v$ZkC9uyceV0xA7A+;n1D)q3e^1rKfXzo^Oa0uy}+0+)Dv-P=nT
zOW0vdYE($gcPM$9K5yDdh?7o7K+l(McN0?tLpjmI_aoyJG7h3sR0-X!#XZ&D#}C+5
z1A1|qa$_tx$=HJ>&i-g@s`l{J(NzlR<F8yvYb(CRsG1~AKP*Q-d=m*;JCDFpDJR0W
z#tSpAvE?#)L6;zgS6I_%#l9S)OC$n!XigVk`qXU~S)F3m>tL3;Og=<ARiCpI_+um$
zC3!;3cdZ&b*u1SZ+n(K1D<^%elLPkBkE0*MA3}HZ)h<|=q=67`v@Jsq75sW?Wd?({
z?pt7&mMqBooV!VRJ~&g{rlcZx;SryCw*$A83=K&M{;^#2=YIYX<`2cDw--x2hL)i$
z8%vt~sH_m(43P(*4ha1{?2M7!4(=Q6E7Tq<mOe^SHR<DM`2#O-qxa#la;ZZ#j|1)0
zhOj#=nbT>qf=Q;_hwa&XfTJ@+M^<vG7iwX@7Qb3z#K-g@bqK$89C>HbyL{RaeEERS
zQu2vQI;l#ig+MTR-6s%p$v}}Sxum{Kgb#aL5+h&L|5soH|5A*+JVAy3m5|t$@kRwN
zIi}tY$eSPN&O5(ymmGI!dhNslau7-b9LQjV4^%_kVzfiHJap_6vnu*)L-_r1-7*a{
z?r^mV*<NU(F|fZm=1^S4==HE-@FD0Qsg$Z;E=Zfm>?QtWQ|M(#`M!GptY;cc9(2v?
z8)aypv7&4rXUI~o!|ifE4UW|KQrHItIsaFBcLsd+i}r)@nF0(MVYhP@%$RG*Ec5Iy
zu4$QcpTnT&m4u7>L-Y1342LYkkyP}}7?nEiI~+({kJWU&7adn`)Y3jW+jF;$_ZoC0
z!yA+nOhp_^hH`s|I+t%P=?;DxnP-5HX*J#lMlMz3(3jzadU^WPGRHUXx8B=5Um4_i
zb2uau%E|7QTu<|PiSU7nU3}6&&fpJ=OcEDf1@%l*kX?fByG4K{DW$XehNE^ofEZGI
zuuGgM7q{fHn`cJQ_4%6>objzJ!o~#Gcd6Q>tn^j3o4|6GJ<s-^8>fu?W6mvstWPLR
zr=Y;TPGNm&aNjd(B9ZCd5<K{HgVturKDFy}q)eNB#-y1M{Rw+eyo}L`Wbg5(f!vRm
zv+WMlhnbRCJ|cZKp`(@`HTmOR^UEs6bmwM#>wv7A&GgGfE!FGJRFG<|)304TBcm??
zy`{g{bD34{TiSR$Xt9eue<;Ie1!2JX@7$0cqhj~~McPN~@irO{EvIyxwK`KDT2%O*
zVx~5ftFe+L!fp5A`r#=~8P*K&tFdA~e#8Z0&AZ(6>u|{C>L^EchKuU7zhtdzMo*Z;
zTvNp=dra5WLXO~)4s?uyAy=BzZ-<wZmX3bh*ivF@3AmPZ7~<v+Ks1Ck`Hz_)-yID-
zekNYK|9iM=bl%%)cw6J%EAK2gRYm)Sv|QzT;3<LgBaMsu=zpNgx1gF(H-U<G(+s;m
zzKezE)|SX=Opz<vTUg56RNz1~?@O2j?dB;}4XB3(>oHzHt+WSH=!^qL!Y3a&3l?9K
zK(8+Atw|eNJIk>q<)f{i9!e50Odm1+v|W7qL`3_gp!UlOI|;1}h{!@nwLe&{!nDu9
zd=_Wd*suR86;>9?Wug3QN(eT54UxV6umoS5D~I=WD^HBb<`H>3McGv(_JH6^QE+Lw
zC{kWk8?;8r{!i<fb_lKbFQzZ?yR;>KB+IQ>GjF>ED{Y!HM;WHiZCbYh`h@_dzpsJF
z#`1|!yI#$8M*_UoeUX6qC1aE(#0GQj252m^2_TfO!S%<qua{)1MfB!2ggZ-rWhNf5
zb;#S+0sA*f%$*{^sb(F>OxZge4bMx3b(kk$X8HAj3+yhm6ry|@i<PVA_hr-=hkQM(
z{2Mbk`4|N3d^UjzWp(ULl{`9qOCs?BxDSX_xe-2x6b5H<QSb8KW$qq)^sU%2j8eB2
zMdE)ft;ckJqp8!_Qe<n$q0kXA)frXUYL&!om*89Vsf5N0Z<aihPqZuY)D$KXFBMGW
zd}^b`2PCvAY?(!%TWK|pMuJ>zGNC7d%U07QadOvf-P_sv+oMq_hwq0vZb$xP2eq$h
zY;%Jr<d1(q7HCRa6N4<2)u(+Ae5@k#8t(mjFyz3v-P^KQzRD90!BnvR<;%0NEf^4(
z)}6y%vp36^!7Hs_Ku3fx;mZL^=b^kq((Dcw)W_C2%|+h7eht=cOVkZ>%GoYbD)0JR
zr;BfT*l$mkJb(d_8d-E%bFPa-tbm&6i;m+x^>%wes})PBRw*Rh|Io;L)F=HRI9Y&y
z<T9C~*V&t|!4GTsG<S7M=q!*edg5kF?b!80+~FA*)0QiOBH`Q&yUO=otFCyGzXb_?
zIc>WjxxF6Tl#KOj4+w%F+@zor+t)O}gS4HWL|T|Psp@QdxP9Sd$y|OryDTHT-5n6V
zr0MZ4x+sd}`zQHYr$u1xcDV_PPlPYOa(j;JS}0z%4~OGadl_2(WD%#MATlELeERXo
zK<V%7+in|8{6}X5(Y-QFACak;p5%(kz>N+OL3a(gjRCLY4RZeK<Tf$7*)Yz>nCP{s
zwozrsD+h}vnl3io5E6N+s^@}K<eXfjoEM}DW8R^blE5j8?c%hjbrYofN_Vcl<Doe9
z`5*kBMW6Dg&o?|iioDXc_h*qg+xv$SyU|P=dr|_#!`4_+qO{a6sgRc5jG=Vb^y}eX
z^=V)KNX#m6@z@93a)f+ro~<erY~!{5)Ja3S6+)n8FklvKXf<##ujIA`YjS;xqB=I!
zAkZOMjy_$yTq5n=g}h&5VLDo!zwaH-Sqe7F8$$APK;2iu7^eJ6TQdW0#eXbM032|U
z8o>~$lP^^6n_dh}qvvs_#E7ew>#>%35nQcTK>lj#TzCD_u&=mvu{agE@SNI%cb&<P
zn)Kj(p4p-t1j4y5QQ&FeUuBO&U#!+WsbcA_j$^+`{QWcH?&7SiF>tRXn<DqPRDf(q
z->PFFDd!|ZO;v(^F*?j${Pi?qq=f&O5a@t>gZenDuTfJ<^|+vQf*36Q0F?Ms9&Y{+
zvhFPXUgoc4n!fuEvrW2To<aVxl>ItH75{9*!!;!*+Sy}~z3kxVk{Sz`ufwL=@F5-d
z*om6~K00vFYl#Kq5hyWfXLY=GdJN3*S}MOktyPtK_NGkGG)&!`zIyaKwMV?Y4B;0#
zOPQ&l#CNHrg~yR3C`r;aypOQpt5;?z8i!dpKW>~tp4AP$=;~^lt2}I-CrnJJ<K359
zYuH0)wHGxTFb+*yGq2Z3!D-IF1TdTj1Ks8mJGX>76Y8FROxAkD(lrS90`)8b!OL?L
ztiIm|<64~ZrbvCT&Xiya9aQ~gh?{m$P}q_F4%%HX%*$4~>SNmL-j^CUIvSD3ZzTnf
z*2*`0z>m5lQj~m<343Fm(d#2hG5z9hZMaPYKV$k)y9qR@-Kz%3(=pqko2K7lOT`wY
z7h$8Xm66J@Pp%=M-;Cj&FL&UTL}rFe6FlPtfy3zZmc<r!h4Ro&6-?YLe=|7R4vWzr
z)^uPVeHWPytB38;Sk`w)oc8!QhgVwupQC4`yqsu|W9$Jg&<HxRZ)fe%w({`!iQnt~
z%3b4KYN<jKhCa-=U}7Q;z*FSjL4pD!QEoJt1sSI>^j~$FhZu{{Ydv;>;QVT(4{Rx}
z&)(90VjXtnP;#>niPIHij1_6MzGCrM?Y9=N9{%l5yHrXVedZopQ14Y3BNMKy{*I8e
zkUK(mOx~6I8H$SZJHC~RR~X;uPISznv<Md}^PdEvnPkOYScV3o4gJ=Z-JjJ8V`swE
z1k|_+3ndGW#kugmgY2}oXW}a4u8|b5{4TDjGWY~)`Ia?UZjXC|-ET_-IL73ezczXH
z*3E2r{=*V}goTk2i_G1g7%ohkFP$%uBv##nK{W*j(n*0aTbZt5!j|{=N;iIm&fq*6
z<ppK{Myh;6`tW(d?s2OMp1s89vzCQVtE7ztHgAp3IbBctvo=9B!L!pQHSN4<RKcH9
zhX@*WA+&)EQY-599qX<e#b-U^3+^TsGfXbinCG^H-a%Ie-GLOS+w-})*NGMupSyoa
zK}G+|z`@x!UoStvh84UX3ddh+5+oC++!3#2Q9&?t?UMKGP(?nXS)`Clu)aF;h_$}W
z@A+!sCRD^Z(Q;mwA+$|eVW#oh6249s#z0H{e3*e@DqaxUlm)-O4S=s6H{!yEJA=JU
z^03jH3AW?WXHs^CqOD}N0G@h8(FF=FSIMB0Tl$dKSagI8N(J|RZmPsgr6<EWN{5?k
zq331thmH?1cRshXz$VHxKvur0D;pCos?iDfP%m-&N>^_gYYx84Sp{Q9WQM=AfKM6!
zRIR{-1jy?k)6(yKbMB^VvWTl%dpJX<Lkz06=q-!u^PkP`1_oi;r^!>$uHr0^+vk@7
zDf^cx0=<C?|D=Wo=hgOx;lL2evXS*|$2Z&ECmxv%$MzF{g7U3uy?F9fI=)bE=}Sq+
z5;z|$GEy!R6^?87YqU0M>A-kNU32D^BNsgaHWv<;@gSB4fZ(H($6JWnoLe%XT^=s|
zMERG(lu5O&P~BXMadM9HujWjYRL5NXIJclM-)IY2%4IlT;`H>RH&kk*^S)+tu2^VW
zUwaLY#JnoO3wfQ#^@zaqVTHL?IyF-fQuXrsxLV!-IS`%EFV99BV7yK=ga>Ns9<_Um
zY!=pa5ceZX32ky(r4>YWU_l+m<)g|(0fx-Pg!|Bn8ow>DCvG(0LWK8#>ZqS7*fAn9
zcOOV9fFeBcc6vH7J^Wo^oTy~(qXn;-mR-)E=dk$YZwFe?xi0-&eKOV}1GH2$q4dpv
zN#$X9mT|pvPdM-Cx9hT5U~t>bqF5G;9sy)0^CWWVsjHdDm#gS%vBBR%n(&-NBMMch
z6O7IC{lpwWxju1>aaHu3n9It?G}5D<mvEb&EDvyHbeMB<NpuKVVYC(Rs)v`6hm7ws
z_`>bvgSUdfbi@)9HRej%`q&YDb$1b@FK4^y+c)p=F7`+&W~Es-ntM%6wA_blhXq2g
z!)QjDo<k8X*Aar(JM{2`F;DYiQdj*tm>QGk=(F!0v1H{;V6<`C-&S?=FyJjsLS_C!
zTuot~a;TBzANV!yZR_IaCh45WBt_;|-8YHVp|kQ&_*p*8>M#HHkduoMT9)~sBPYdT
z(=YQWH43C#-^dwF8vHn5!Bgt)W?Rem3PnV+=kronFNosL!wlw*co(raJ<!KUp-)yp
z`#>3p+a5kAY4R$$^(sdqI=sp2+sNNRWZb&jQja+^S2J*7$$o{Gvh*%#q<vL#onM9o
zNEHhaW{&@w#j{iy6RuPZ`o~@)5jJGK<?#H(_PB$O7eX`5W<6r6f{YPLr3z3!lw=|M
z;J24Lz#MnZRqNH@Ao#-hsGq=t-WeDwt~1z7brZfowSbp%3Qr0kV^gU{jU6d_ry#q&
zRCxg%==7!*kdjCUNw@ePh<>jc>H-XKWmN{9|GhiFb~|{_XRq(pG0!ZKf4kMN@1PD-
zf(Vd8?DpHc>+7JU(YBW&C*Rl^m@m1VD<;Ki!TkR^6+)=aFXQFbWrM(afm*kF(Wb--
zrkw-Xs{80HByn@!=#M9;8uz1c=G#Ol)FE1y_-p|UvYYD5?*5L%EOF{WUV!qYXbtXN
zxSZ)HB*4Yg1irykAyeOZJKoq}!ON<5+wedP##H=4mWu$o`K<Fl7t)+XIP|3ww8>$f
z*JPHN!2tzu9I)dfI7G8k`He@EqT}@5kO)sJqp$hcMQ}>TOzi4@|2Q9}4voLJc$HCS
zR-r!DJNB0+E2j6#{r+2*gG4)?MU)P@N`po|#J*tVm?)n*S>2qMt@8F}(iw@7%`}uS
zU75&S^&#-wlKp8yn}_|=_xZX^Hiwrq{H>IGyD)w<o^U|nk|`aU4yU+yfk}-xl+=YW
z^44Kf;O1(NUNuppv_^}qSL=g*flEuDQq;TDc#?d1xa+H7&y>fBLj}jL-w7?ot-@az
zNr`A*N~WgT6NL-t!)h4b)0W=(Ha;lFG;aS~ug#p&uw+=?qNN<Ig||zOiwt}n2s9#H
zT(~vf-lyv5b@!-(o(utJTwag+80d5obUM-MCBhL%#H3)lh@Y%|9wPMP{I>Zbj!Wi#
zCN)F&fb@@YXM8BkY&{vxuo_3nu-+ylz$Q@2sWDX^po;E@wjx$=o}XKt89X#62m;f=
zj(@|DtJCT@!y3^Yo*C8bW8pwf^Il3pb38OX^C^NYg~CysuGyO%G1m-{X!Q$uM3ckv
z`3x7q9R`b=$@ExjW7S!`nPBs}R4$;MIuzp5-t9(GFiI(Mza+RI5hNocf^=ZcKDw!b
zAm!-m0(8vu`f8iGmD=G$L6`As6J^J&6XY(f*q+nPQ3ylu$yj_~Q%~4e`fKi~CuF~4
zJS<}8D?P$PVTV0GSanw7)9)j(LJqo4HuJ6j{^q2;OS@Rhoj>k5Y7D&I+&x;*PiK`Y
zeOd6u%04=ROnxAXm3GD)8i4;;xHKDfm*!e{yFg7`2X2ve<!Sqa2g`#ClF^1K--dWc
zr-eqI@;-bjw1xB5XVa^BwR1$!)zjtAXc&1|d>CQkg63}z0~(J`f64C-F!iVV+P{6Q
zd~mwzJ|0(xJJo|nfNdWYdD!6=t2%G+0`<HqBUT$n>RG$GB0&7ZH_MqMsQY%uo&w&l
z+F?lr&%W)-J;`b4vPgv`x^V&Vhf9wV;jJCYi1K)6+$GH_Y2vWHL<Mm-TXDA{dH5DX
zRfI*l_uYL!WZsUz>7)c2Lv<J-!)q~w+3KZuutl5yG*(<H=7@d3LH&Dva;<mC<zn2)
z@l=Sri9t-}m=K0g*pz1LC{Omj%dkwa1zOto{yNC1=D;pP*d&qXwS8!g%AjtRKxwpr
z!k3*vnWc$^{PUwL?&S5uBT0k%EQf#tmWsPKOdb#ZZRm$dwD%kQY?uAC!7t5>CvE(w
z+@kg5u4(*xP*%VXDx=Ny0nQ}LO#7RZ?UDkiQj@zM!LN&X(FAtu`rBm}X&!8ry^P2^
z2dce}HlFK>d-4BbcY=!kr?tF+mI5#oAuODnAOJwLuZE9_!y+VuTvgnEGkPs{RKzUo
z@QH;9Lf>LQ5BH78Bl+1R<{Sid=xFQAj_CIc<aG|uGqcrT;5EL{vZl&$GA9Vnt&1q7
zG>RXgLGgcolZy2>;E;8ljuG=dZnXiwUWS{G4CO@o;aXC@1aUH}#d`MF!e&}tXZ+mT
zw(^%lTKn-vfd8;idjb~75F+Y$8D|4D3ACzuMXNJKzi#t7JT3iO7RcUP@Y{o8T{2ze
zpSp-gc_AK%Q!50DEUAn4;2h=s4Q&}0DHAi`1&nLJEcgI+15`GK5<rk~zKjb2?yL0|
z<ZoU%SnAe)CTpW6r7$3z@uoR_`X?uH_l0}JXPf0d>N>|EE8CI)<^FSbW%S?H44E3X
zsY>R!)AE0cf6(Acd5_&JYgxq52r6poJZ~3BRQKp)0xYxElXmiOJdyJF@&;@>frYTc
z%;uYutqjS1w<i=*lP)G-$8f4TgQ;Ffl?wGGK$x?QTr3IzJw!J0FIGJ(6)ps)rUc5;
zM-uu(02IWth728OW{CAGS9vN@{PO))aMXbQQs2|=dHuwk565ycG2Io+fxE@!Ud#KX
ztRjUB)DGvpzbz71q8#8ogH($`?=M7}?|yz0w83V&?~Oc0H&v~78aF6ubR$hll}&NM
zxE-V}(T_2Cy<*1s(H|@5_=bk>abdy45Zq^{fg2a{ho%vWf|N8_oIlRBlq0Ka9BEhU
zXhP{P*`Fzrt4?bNSgVg3&*;iL+NG<K3q?PBAt&&-3b3-`eY}j{tw+7~Sn4ESuNF{V
zu?ZBfqEkcc#(VHmVmAIBwIrv(dy%(GiY5T>(JXHq*9X8vRQ;J?C=m8=&DcpG!iLfR
z^BnMV$YO@^Ac$Y|w%|YjohL$n0UgYOQa3ve>cHZ<#H)SS7y}G%Zc-*weQUSPanG_Y
zyVrrE?YTh=;WO;rYCY>YR+QvjRPYlw!&I%H&Y9C+wJ09~HAXJ(MJ{?&`QIupBeis$
zB6Mymr^&5Nv<Ok!Ygzu=Ws9U~a>LM<n|)<f#ryxVu|YTrn>QZCfx1#d9(AEb;XYT~
z5g@D_oY5=z0FDJh#EZwXk$B?B<S<OH`ugDKUY+yDtaqXZ%h(V@5D@Yr{C`nI|1=)L
zo(mTZA9-VFe&t5mcI$JjYU1_1iiq2OX3yaS5GzSk;>sj!5q4kEPN+@j&zG-O8-Gc)
zR50Q~raU2}Kkz^<^0|X-D;CV$eE2AAlBv<ZLAgv?p=AD{Eo&mK9yu^E4x*~)w?7{v
zy7^@hJ`+5aD}V3tz!5N+2HT&hn(uGyghI!iOyZ9>Kmi00(YzoW%*~Nu!_L-02O}Z~
z7NKzH9kHZ3Qyq~YXYD5Pe1`ImYmsm=WTn<8K{5N`XxI0Q(TgA^R)<ryinme^mJ=XY
zNv8>aGT!Xhc!R?#-ffd`FQ~bhir|_XQ(_2IvI&hOabeeo(!j>a($dS<Ih?31!7hio
zRYSDpLerHI^8jDAQzI}@nJ|BCkT+{Jp-sDYZJ~0;L?-2dI3ut3^?S@wzt^2<JfiOZ
zA?hvTn(*HLVd<7`1nKUM(Jg|6(nCr>O5#I|lJ1h0Zjmm@fpiIs4(S{@fx#I0-}iSv
zxOwd<56*U-bJaWMFd~R+@wI<!d!=K8<Bn%op_DfRZ7Pf^(Kik(LB&)dNJeMlg#P6j
z!J>6hhn<-b^P=lVevikDZJNasCBDhi*fwDI%U8f2(i2=mXFP3GdmCO!)9d`<<xmPL
zgVxDT4}SY2zic2!I9KU2_{ujgtR0GI_@u@(7hdsRR$RgDqhQ#^`$}8u8*f_})jbIA
zC`brpha@=)X9nPbT;+-mH3v-JEFy-++1F<(=Kzjm`ZysEt7vB4_E%BX9KhX!;20b?
z_oUo<IjnfmJL@+yA01Qkh^p~G-(`QiS6>oI!cY<{ZBi5@)#5$a+6B5Zi}4J81y&p8
zvdH`=LX{=Nv{LNw$I7#UA-l9z6!zh6{U;p2P*ryz4xhT^#R5G8$U{)jfsSvlsGv?y
z?}Z?SkG&B`%qdM^yNc+|{40)LRrxq-cy~{npqyJ!>qL=5G(As9`?D-ryA;vvapE7_
z_O~Y7fbbkWsu!TXK+3-2uQPYtzktLao~8ckI{r*JR7i72Wab6B+0$3lZrTNLx@XtU
z({rPk?9YjXnk(SK{al`n8j9p9_#@^Qbm>Fi^6IzlAu{LhA*H?l>R*7nH!qO?PzYJL
zF-r24-45SyD7=6$UT~mGT@P$W1-E0s!WStMDe77)tqg$rBILX9LH=SOQ$j--qRtKR
zwSKyHk(Me%*M(o%3x`#(n^S0mx$P4xd1wLEAZ+^x@rdR>U^xR59(1@**~ddzLKtx3
ze|I6XQ1UAK7W_48sUl4trEFx*^4hN3slEEE9@t-Ng7b_4dO_3r!?dUkIU0Lq86SV9
zHC^Wa5*@`YN12GN2fU|)^513`NydB>AI$jp79ft%z|}}EAXkeA418d*6Fjv6K_%*6
zg<n@k?<_IN7Fk<{ZH)Suv?iozD9+!Yp{p@LFDyT1<j;^J)&nGhb2K5mHpuIW=+h!<
zxEm+*0!Y6wVS)|6V`kkZw*Wbbt_3`=n0#0ce<#9VzL+$4*WY)06#c7xZ*BPY?g~Zj
zbT#FKctyG1MY+`xZ<HRQyuf2wKaZepmqH?lOi<p7$6%+2g&{;m>W<L8`<5wzZuKQH
zGDyf_iqr9aFNX%wDxP^&z|<W~z7cZ&lERfCqGV^|_S3ScU(>Z3CfJ%r434+ve|Us>
zT4&N+gTs*k`<m+}Pul#+`*&P9K>v@}jF*cY4K~V}{wr`wtg_OEcjAF6tiYw_fT&r{
zZZ5W`_Y8ddbENr`Np1fr4xsk?3IQ)cfX(kE;;q{XB4O8^drY}uN@g#W3lg@Wptxw*
z=g{1GDasC~pxe`kf%icTGj8nGipd|_vrvY)Zm&ql-ZP<Oyh3;JK#fO+P6TO=ZQa*D
zFS8l?6f=6<{6+kW@!@NC9}jol`n1bI4A<R_3sJ|@h8VKepG~Hrr1fjRyGFU0DGc#i
z`K{e)FRf5QngNa#@*n-%Mz~x$qaH}n@{jcvQm5#!;MLFor@d4(z>;3-UL{3m;Us`t
z_EHnJkpVda*Pts+M>$)KB?RwW`IT>C4AeI9LMF<XzD*qRUiD6InrLn&6^%$;BUVua
zeh&UZ7rOHOXsO=&t5C^T)r#)-mn-=%AomGH2-Ir+{rLoK=gP}M;swYgeqL$YD~O>r
zsg5*1JKb~*dMP?5#9R?W;_s{AcdJ~PIG~EI^^Fl%N`5A|(H8pPrqORKvM^~%BMc@1
z=uzkA8N+NBYh%9}((6dQrR_#pbO1dEEmM8*DkT=sD*)f;X^IQlYGG^B|8`f<QFMA5
z7AB?iJyKrjf`urch;%_nm)xP|2SrHy;*=QnJQ}!3U0=lMw8Cpx#z7z~ZFmD-{UZE8
z`>1&A6g5(EsMpRZJq!cJ5`cV?rTG7p6y(>;*L#ehj3ZUr^1`r0<vGQOIQbF=<HFEr
z)?nOXpTA%}-R)Y|n3tC$qq7+Sp&OsjH!{V|I?|Tr-cA20Xg`z>1uE>-`Z7paWjo@W
z+3=yD9DN{o%Y2BnkKmg3oL2k!9eF-;Vl|G9^9)zbD1C%LKCB<<pke(HHi9RLNrGXA
zB2|EQqh9lNNh;}8y9GP_9_`|8X!Vdq2md#L{c)A4erx95%zE#4w6ObdZcNT2_o{p_
zW9NS$lIFwi(E6^mDD$TYp`dlTH=_5&yC;&RyXL+&k9XT9ZFPd`#6kFaE%Ma7S%?7$
zsoqx4{S=$g_3#}fhOqxm7v^S~ko%3fNmrBLc7ju>D{1yiY<QjAtBaMnl*y(%2gPJv
zE|mNL&|ugZ<?k2ey2=P98p1%O6RSW;IIXI9O`<UCD(Ozn%}_#ViFLMk6{}mvv6S4x
zoxxrOLV4^nI(VfN5-NA5)oy+Ex}r^DZIfO_@=Z99KXe1-f70ei=UQH{ax}@iUP}FA
zc8On)Ll8yDF?4v>SHStMX+V|;#Ou)hU+*#7f|Xq11wUjiWD1T^+0%3xIr3N&WfxkY
zelPX(ZUwbkKuH25DQqK5^BTByE)6lgf6@~{FWL=?s9&Uq_QYWUke{*P@LbMdwp;{4
z0djT!7L|DDeYCcX^UwKi4$A6{KeOG0GR`f1W!zn?C=R4gWcgOuG>QYFy3wv|MlFs0
zW=)?Pp)6lQ-mm@44Mr+Xiz|OSb4oK(^u%+F$9*wH@|KeElP;a!TbB<farkc-6H74A
zWR^Oj+EVyG70(G_W4!QY5YsY+ttMFui>9@e_$cZH+~ZL&jIBVfCtFQ^*fWS`SDis{
z{@Zodtfb}X3#)UOZT2Imhvjh55?h8iDCrG|8Yw0gud40R`c0m3FC%43KQ4(N!-qwu
z69o&E47+KLnhCj*cZE$OPSh%V7pweS_N%BwTGItSJsh}ii?a<|RuOGw2&6|<ItoP2
zu4vhAY`#Y+TeXIS0=ItV5*DI#PmbEoM$@f&WQ%;U40aThX2$!xx1wcO%5owl77@z9
zZVujoS0D?2H*)x)mmqK+W8(78W-%S{Z=Tg*>yqF*i*Ox59G)T-yhUn+<9d%YMr^u1
z`^&e~1#(|<crZJZAH^w?=_&tZ{I9mtlnO3!+38FWwUxi*He7xvIb5`n{F=9xKJ$Q3
zwg2I~{c&;5V}F10-y!N9DElEP@qIypmc#f{T_GA+2@_m^39iNk6*lJy!A3v$?Z%+S
zjvi=5hY^*2?Dc8$w3Z5BwE9*BBC8<nFTiEW$1a70N=Ertu$~~ijlgU4?xzQ=TYGQw
z8lVWxz@7Z~7lrZLS8s{>_ActTc75vx&E#Pb3c$g8m_^cleCyWumqE3VuqFRs<+Xb*
z79%GW5RuZlrXxM7`n!u0B`FlAT;ZGW|6V%MR87k8iBcZ})QAE(kiaGC-n*Z^XdddP
zva|)j^ydpxHi$jE6ozOymsIX61(gEf4kr4`1RbX97uOENyO&<yOYW&_^;1fT{w{c(
z7z?B6U13SqGb%$)P)(VHPQ4B!_v9;zhQ2{6{smo0P+p}-W7I3(a4t>R++lxNw7O~O
zB}W=HN&+og0EYltl{7;^B)F&bx#&-kLQ0Reill$uY~r=g;vEtbfb$E}<K>IneS+j;
z)t-rD2pAJ&e<GxpO$f!q#j=u1dWxMkE;e#AIviIpXwSn^-cynKi|)8hGlICph{p@(
zHLU!8BI5<$tCUJHm+9r+JnTYR&53pwadK|n^4@C`MpIuVJX4fy%rElu7_JVFg_<IR
zbQx3gQdN?gaF>p1lD&+GdC$3biA91In2WnPtn9}sW}RH;Oi?zHXA=&-C{Zw2ueNJ;
zkGGzL4g!VuC{IY7gomLie~r3XLHe2fCu&>!LE1BslOQovjsPt^QiPY*K`8_61zjJ5
zHq75Vhjij=!B+f|c%-Fg{Sj}wTo&Tbr}hPc^8VTnq7J7Ln@h8}4Mo-Lf6V=AU3gQ8
z;f^%p7=W&$3j9ji55L<Xh~HtI0RQ?nRxH^N2Z~d(VNIU%_l>v>w47LhZicD<q!ZSp
zga>&1U-37T0?lX@Yq2X3EemEtE&|EC`NQf7Du;o?@wrn!Q(R<tgM#pX6y2zHpd`YW
z1)rmiTya66^c4ZF<2U(^hQUYyNI+~+R{K9g<zbSrK_O!meH|Ju?&GA^>fXWrY{Zw{
z{mF~MW|fOE*(KZ8ulny{QZrwbvIJoo+cveoH5J)Z@!$-2RME64+$?#1N`eg<dH@FL
zzXpGzjyB)Yhnb(t&>uxqep|3dE!GEJOr-IzDud$!+*~4?^<?m^e7dPII@<;ei7lhB
zuZcC<+z>3B^#pNocmHUybA(K)INj$!Ps4{3<43xXE_ASZU=;dltHlbf<fb|O;nxj8
zY<Ms;6wl(Z=<tS$%rou-VgGS|FbYscoEwiiO}4yrP)gPe`H6ADZ=SHf-`O!)(XUtc
z!*PvQ9W5^8XsR=>MF@&0YH1aT#8kNBkv*UM-D}KN$*Z#v{RiLrmb*g{d`EhIXAU3T
zKh~qvF(wL>|B*mv<b@u}l?1U73civ`Fxsd9u3seZydXFxD57--$OYUaM*4&Qt&B+S
z<jtMzrrzt0{6o_#+BIL=?w?cfxO!jPz|}!NQ2I~0I#^rH8OI<T56Rl#lo2uj2jhb@
z7JtW9N^+j+4tz`0QCSop)+%?sa{cXkTw&SfP;qK31f><We8~ZsLjp%dm%JY-61BBT
z!5;(*HVjSlMyuLV+WgY!M6Op->pje_S?XE;ebY-(YZnQQt9;2z$$<fSNy4MxNMFTn
zE|*x0P6-r^{;wEC5#%~foQ+|1#VEpdLaE_1dB9*9ReJwf_Bhu|A;m#Yfyg&*QZ7By
zzUA0XL+=GnH0T$DxKB7;iQ##m57)X3K?>I?G8axCfk$*jLxID8#4jJAl{M^qqOO!}
zpA$aRZ`h<w+3&k&f6Dt1y5Y0mEUTmlLn+`?uc2C)+~1h`EFT!4+ynIf35k*DlYeJ#
zeL}y!b|)n;!zRdi(NDorf2Ai~<Uo*<kbwb;_v=n@cn#%dr4`l*hL8tJi!8w{lcRKM
zUB!Y|jw4Gh(i2F_;W`z{A44-Q8axv`oVazSQgGKRT^?}Zd#BZ3#!<}YOiwNyw7;Qp
zOMg?lBNejI>s^V|8=UbrC=hgY-z;kwr)Vfddah_C-#8LZ%0B!4Uh4QosImyE5KJs8
zlgdy^Z)Sy<myG}6O=EOh9aS_K*Q0flIh@sphi&Le^UD^mp^lb<@+0P*uJF4IB%zHC
zhxL-7{9Nzx`E+h6oN;}f@jiVxSt4fVLe)V#^|xU9gXth9t3YpRM({I(9B5P8Zy)I_
z?5S#QN4m9GThXldg2z)y52l7)gPW+P;ztw)lu&{NJ3v3VrDiH4Dm9!S^a;7=GN>mf
z6hG{{F+EM)+Y=4jvstv5f(tbHpl3I^$KN3rx8S86fArxWGe4JQY!^Ublmjh2eq8L!
zWi**KpE2qU2cJbfyGXSKRHjJ_v#lHYlmuU<r3Lr<XGx6cDfA3G>@<rI4yi@_WpO>&
z*yMZH6uGnxjcSY`X$8G!ssXOxIDfcSn>l{sGF!<?;exV#%pjq{b3^?DyRx~5he8c!
zI@J}3{lTK8B5MW3mk@MaRc-$}AAOvNvZ@Vb2i(d92^Asc+7HG>@uV>UnNB$HyrJAb
zkhep;Ct-`&@B-8TUX2}dW@p0$T)`yL93b?xx|Km)H>BcQ5YEZe`NAsDQ?}$V@M|k?
z^mlIovx!5m)&jE;=Q=A{I(p+|mZF!xB{vnKHCER&A-P-e1DAGlArAXkVBfB86yXiR
z>b#TQY-VG<=)D-kbEs=+Qw9WN1nm-@n<2Awkb<$IMZR!h6bu$_iq0m>=Qs0CqdfH3
zQ*aMC*kalblFNDuugGyEl@`~Tuo4C+g#Ap6z%~9!43>{XP6^@lV1jbs!cFg^SMDa}
zt#!?QRBqfr_i2Rm8Ml$;ih5q=`x+OkKf~V({sg|y=TR(sL>%&t(Pn<$1RD%gDe8^T
zK?p_>VRE~b4KCPIUIcQByzX+wEvMJ-fnu9WD>6#gZ%{y^A`SDZ!#^8Ee#Ls*u_{W<
zg0KdigPKB=Aj9uw0?7BNQg5IIBXq@E-wY2v|9VmkWkTC@x$;U7`o6Za!xxA`S2dsq
z!0i;BbnNzA)QLZ7ip#Ik{^Is4vbO0Hf2X;-OZYc^{xi#N#q)zfy#WW2d1q1FGyPbM
zfxnrmntGo&Av)kpdKUwL=<GQe%dS?O@SFSkT@7gRvDL0Ip|WVf2p9X~cvjLyXh9t{
ziR8CT@+>=4J^4+4)eyr8-~P9I9qKB+yP<(S*7h}^4*sV5fYs3?-08?>jdB*r_l{0<
z-Gv~FoatPgMxea@hJ<YS4c*<*9-dZ4ja~1R*|%Kx^k_^FHq<}J3d@N{D=Gt#f<Dk$
zJ-+XS6;|N7U&AF$3F$)zxrW1`+=U$?y&ax~BfPs(|167^3W_8J&sA!Gk_;zt8pxL5
z0Zj_m*9k5xelDls*X$ULep-TqgMnd4VW!vYk6wJHs~FChfb^pda@CyT@94ZwAH?!r
zD=97uVdnS;eVGnF+;OPw)Ph`9MCUZO@ULKjmD-U7%_-lPV_yiT9FhwJvS?D0qqdPs
zbI1D_sc8$u1U8+_9G#+MX{eIg-ks0<FFh}=FfX2^vm}X`qyarmCK2nqKZxsxpoP^Q
zla_W_Mh@sT4N`d5pqlTei~ml{ctxoQJCZujOUo<pdAB-Z1BH9M>M0NzFcMe&kH0VT
zYgf_|u%n{B<LUFyAKBK6LMHvd#wZywd`TglS<vuYXG!S%i(%=3aUp^y?M1{ur#Jxb
zC%7R~(YU2iZ~*4$>R~nP@BSqD;!(`|>azI{WYQwzu-8fO7e{f51f&<N%*Pn`<mP>L
zK_Njt{!2u+O?b^550p#!D-nZ6A>vn}b|op7g)mJ#)wQAp{7^Fn7SY{Z-vIj0ilY9B
zbO=+XFCbRSy|1pon(dmu9)Qi8`En4U(EAmq0E~9ULkcI$n(q6XZ>1bQ3c>`NNu8aN
zr=$Sp;81C=mh?6Y+P|(7T1b1AY@-`5CqspMnHTVzk4ym8%%u=fC>{kSD2(wMj)~Co
zvs@(jb}WRlV!FEaRxm$AK2qfG){e{*2tv)+dSiY!ilX+v!0+PdegMYZTKZQ*r70uT
zQyDOuqBO(@@JU`}p#8~qQ&X_^>i3-Cwi%(3BL709fwjS-oiTdrGahER)-x>=OZJ+~
zROQE!bk+d(6#_8tqc#!uwdP2GlAfpH^aX$Z154p->V-<{D8X4u)WEpL3>_xNq(4((
zp=SG4NM^`MS;vvzN4jkP><XNQZ%Q!54PdC}r6c;mD$!i^{~(b69{_9bYXN?Gm7@=8
znDIl0m`1c~$&qR{+i(Bj{MYHvBK@gsCt7_<#@6o*LLHGaa6%B9?T-MYQgWn`dOAur
zuV36C+X9lN?`t-mcaqsz5GvQ2j?&UoMBQv=0$qaaqagyJ?9j{50(`k?A}WPN6Fj()
zqP?Pcpx0s9wtQ#u>=>A1^L@ivPk=7k_2#ts?S$qqbM|c^V3XIiya+x+qSFe*yM<e3
zB!vPb&nSpLZN!3;&9y!=R)Dn9r=Gg4nWvv~(X<QP=Z0K99angjhi#01WduZtE;n}Z
z4xK5Xfi|Q;sLc;i@6y!rrksIXPI2yx%0HdxH^TlmvYM0UQSdcJlRX~RdfDw7`X(yh
zT}0jJ6f7z4fpK&a7L^w^H98K9NMN^2>1gN7+`IlwrfkUg%)2EWpXI&lvEInm%jrqb
zmAiYu865i;ra-Rn_QJh>i-3F}r%in24~jUq{$=SXreKNGKfcT6!T`8bS>M>+K#)lt
znAmR-JarVCm$xxGWiNFl!|XLiJ^HboiXVx8gtSaSArBwwM*lp2f4<0OU!T0P<3E7K
z%QACZ8?GVd*|FjHcsUz#+~zDO3&`I!@C&Kkk_xEkEjL?5P|~20vi<myrq)g~`;Fo!
z8$mlZxFTII=LG^QVm@*#|2D^Z?3OOxpr^kiLRYhVcD6_TuH;<v4J<T${#;Y{Xn41i
zIj8VHHxS)G{vzI(R^cUMA#<u?eqJ9eUWV0a#<)9#nyu{J*!sVr%H-X+iD^acC%&j0
zWMRX17#g6wnX1SS8rk*l;t)|PbmGi6c^qNClHbTzDT;s9$>OV}SK-R^Q*ieOG3w))
zT@g>)+d;M~o^{kK<;<122iJJo0UGI*vdc-jYq{?UoG16^^=Rsyg)wi+CEgcyFrx74
z_d5-8$k`B1dnG~Z6_;-0e!L@z5=6FyN>|}tmF>V0%ceXx&BuOfQz-|RBO@qugL-|V
zPk2P8$oHbv;bVW8I|_ctfWl&>zAp_lvZ~fMBl=@+_lR+ozn|nuW#_@1BJv+q=pXZF
z+UMZ~gG8Pk589!ZP+F#PqgMxIYTOW8ze}g-tA+bX+wN&-S3N-yx^7V0SI#2~7(&E`
zM!Zr_1u!|vyP+Vo`6iCD*1f0`k|ckL#FhP)M3i!2Wr1}c(sgNJ?j!vrHTyUG8bw50
z`7?JPa?^v}9akFg%zJ)LwPyJ_Jrf_2Y2e=bD;=Nz+-$AY5fah9b1tA2!`l}g$;O7+
zMs@f21bfa~Sr0&LxWte+k{wK5$p<tC_@Mp}r~VF~9DZ1+P4mn6lESLyu~%p47_;Rx
zU)LXEV@2BWnz{K`)y$ofC;3YOsQg#$t{0akWRi1=KTz?(rPRlWlwo>Qpf}}^fI2_M
z@Wuv=?ckYH#ny9>4i)y|yzk$=O<`}0tAYMnr$gA65Qp+<QN4lJX3O4tt7)Pwk76rB
zkO68H;^L+<oFv8;g1r|xd}UK>GD&HYNqA_WC;2ly3gZ1&`z9;^nU}<j{d?W9fpZ27
z;S4OWLaL5jP6BO?5KO+jZ|{-+hy*@2YPEZ0;d038`~09oD*dYXG_u+3nDmWFN*){K
zpiMLR(nz-8Cgk&8P)kr`jM37`RgFG5$l_$E)aarOt>h$Tab0QYEOa8zibSW}TtLsN
zZ}U*!|5B3JZX_+WpvOP`M_IZ48;ves-Ae@Sv^h%WPJ{D4=6ll*HZqVm6OCoBL-8g0
z_`dS=7a^+2N@&<$IA*JN2n^SAh;yBO_kQ!Z7zUw&N*9qfGB=_H(=U_s61#+L;)spZ
z2j!MdZ%II98^rIf-4lB-fJ1`4ZK}zexjU6Wq6>3a=WVUtt3Dw3q{yLgwT1Xm?^OD0
zFrq#Z7E&Mn;cUWsH<%1%4ASz47)zc_$8Ua8JepzVCjSolWcU_-Y()>oSyV%@hi2k4
zdx<OHI1B2}N<@%KJq-Bq9445{y%*sokO{t*)YxQBUn+h(6JWynAkiWp%w$9iz~Omt
zRRqWsg3^HiIQ`-a1^Ca{CE1maIS3s!XskB-zPvDw9cn7jOh9DzUbQw#4JEVbA?Bjw
z@`2&)_YB2m4t7X4cWUO_n8fc?8jr*5kd(#7w^tJEkQRRrF;@$QMEV)|w{ay}hT74Y
ziMbRiw%VnnD-tE_u(=xca<wDLga6V$u->h={D&u2u^VS=Xm&mJD|qhP56gPYjo@DC
zlHKH)AZ>q#)GOa~E$GGIQQ&6CI>NU5N}mXRW=Q>bH*d6z3jdo*&$G?2@iL+ecY2<z
z#lcMI#OWqk5U5SL5I70R9q-WVt?*vYEqeYvZ1j|fc~5#j3z&8I=mm%AO*|$3Ln0(@
z>k!e?wQ9-cA*|$q!fuRsfL677tTyOH*#&~TXio$(Te(^BD9Q_m0w2A`RMYuP3H>#p
z;FT&YpJ=~SNa@#tb4qTw!MDx-X5=&qR?qZfMH9w^(WbL}pCpFF-~zixg?yX7@SY+>
z?{&bNzzW6tE8izhsL9%YS5C^w7&oGJA(=a*>^DU%ZoE#89{3~t)UjDoKE;6-2-tW)
z`bL$v4nOvGHoN#`p2zAkMhRJNQho2C_~pGeX(2`F%!QOL1T&M{r+AYgO~_QPIq^nd
z&Tx?Wb4~FOeOz|Ym1KitNt2-QrI*>b`+#o?2}^f@#=6mg@}5*u?-<2W<dk`rx(;c;
ztZa6brPY3}&i>V_U)!m@(=)*gFx**jQUp#pcU@wPxm5yWmuAKcYMNC#^F&+HE-`Hk
z#rmP_9!CM1rp{CrbgnRlm=&tdb@EvFRsHvOE{#d3z*~%G8&~(;mq|&oww{ZDv2~6N
z_EdC}_{}{dOm3oWN5+|pQ4azQmZbU^D%S!TAKcUQ+fs|fHR6IUdl5C`w^YHO<&jE<
z52{uUXd^Gv^99?CzHMn|@D2X4Nw~5$lGmAx<?%P-&dXBwqWzqt4|^TyWymD)<hmlG
z`4?#|>G^fh?oiQZQLKfX$C7p|Dyi?~+ivgqauR5mnKYNtxLSv9<qwNe&E+pE1#J35
zf^@;fjTpLGR99#vE~dB_(s2)_``p!|`@_0vuYN3+AX=J638leVDp036-Wt-)4BCAF
zVDB=fEGyyf;)=<R5XD4(lx@zg5Z`|^ETg?yvM(z)2O5^$xH6?5LW>7_)gEQ+^&108
z$ynYC+wHZP*h!MbJ6f4nTBlEu@cA&t4e6>h45rS0DS3oG?@tXhj0WSbB&ZEiUPh`f
zDN};_8lxNkcyYK`YS?s4d|9q;K{<+0x}|F*lbY;Ell@{WJ!)4GcIbmS&bPDhr1KKe
z=Qgq3cG~ncDVZFI;J?X9VjYx@@nR9+tSfsw<(gzqQfZv^qI;M=B`t{wr>?6#Saqn?
zz*HES(F*?Zw*{Bil4CY_x1s+(hDXL|`xc8m399(_0{hofw%kFHyfG6_`L!p0d|w!o
z{e5;JPiGul`c1F#qQe|Fi)jz|86zpD47tGEMh9tJk7M3FY1W52-rp16w^$5~ls_V~
zc`dl!O3M8Di&+EDPMW;WA60nwbr>y~1jHFZym!R16N9H$(zq~9aFsuz&lqK0JVCPa
zk<a^8f*L#^w*6d{hxNc9dfP;CDPXYd{D)x|A5WHGMCUc{$A7o(LQ8J1sk&w#i+0Jh
zy|~wTe|dvltK?i?seim;cF60Zb>vY=e0VIPf{GZfVK}`b1*Lu}HcpHL%8R)g5^oD@
zXhBV3rR2Ujeur^~g%4xh_GuB2T`I7oykk;+Ujaw4uU2z&i9SfjKH%JK1~ph8y8pf*
zsH`e<Wcibmm}+G6;Nmr(K7~cRNu&*Wutw~-EiX!`6ZjLRNJ`$&u<M)MSa|45a(Kuv
zLih00RawSvLlI}JQ>(9<WH^<|6-$No%uHJSWxtFnItKBHXU|pGZshQB-j@MMnN%!Q
zw}q%D3xDUI9eiR##-qo7P687amnRj*ew>zIc_qV<g_RW&w+o15u@raZ2OP`=taI_V
zt^FMFV3kznx?p{*5Sbk*Sql%>`05}$jV>s2kxgP?QD-&~Dc&me&79yZf_TV9J4c@1
z3RTSI);Ds%r7Ctj#an`?F0V^%?K2>Ab8K*d^FcW>kYN^KzbQAofGOP{t_kh<CkMiT
zqRan%6Yb>5aEwuxib3w|=_~J3lzI`(wY9Hb`Vf)Ubk0%bjb$`in%<ZOsSHX@;*NT?
zR4ZEMM`}_&F$p6+LfiioNabKRQmgmy`a;s24}QD7pIZ=_1U0B2z$6@#=oA?ft4biO
zYTmvgOUfU$(evmGteYO2`O$J3T$N(}knSaExLl1q_^u>ic+OIl&by@Qu`+utRJ5~<
z44mIgUz|fvcVLeH%&ck(9isy44Di7VT>HIytZsn^zwdc6j34qz*hqHWqhFtY5E70F
z#^Eh{7PKN(d`QIgt2d3Md(T_+E>zAD(vH2hBruRcpW5!1H6A;t&bMM!_WLd2FHcSy
zV<$_7a(a!E-Scnm9kdC{INtd+1)@r1pf!HjWhoc_6IV<Rb=>X!wq`f|CmVFE;Ro;t
zR<Hn*`qo433rsaP^>A$QcdxDFjjD~%;$1km;r2{2ZWK1L9&J!$2W6a;jE|gfs6Vf2
z8Fff>vIn|HatWzmr;O(YUnJvm68mP*H4u{wrnYlXO5XmogvX{3r;2{>-Gi-qE<vkz
zGe+@y)2~PfX-!d=ntT|lFEArKwUIEmq!7r`gW!2z!PRrJK{+Gb)BdzY_>D1dxb?m=
z1^$olF~sblFz{K-4bqv5&@%ve>u8I{PRU;AW0up$I)`ZN(&>rm*_RM<Pb0!F^Akxx
z-lw-CGwIKcog4a6di%60Ff7K8>PA~LBAIpia+bV3i{q`8uO(-gPHtpM%@$^j>gs=2
ze4wn^*!?OxCP#ygBz{%DQR!LcOdn7<o?U0kmvQ!<r?T4H1w1<U1F_G29x67f3?s|;
zRqYn!_`{TR`4i$i+bO^za0dZXR>>V!+{Y&sgB`A0v?ayk915pAo-W6@-v4ZOoC$b*
z{IpEEClIlrHkJI@;cY2Ad;Df?y<~&}=G8-dlyjgon{AQPUpoo9Wawq#DEciUcV6VD
z&|uP-#3^>%rD?{HR)?G$&mTiX{iq2Z<agCJTSNatv95#!Ou*PX57y8r-@**b?MF2B
z#vab!Pw539j;m^yHl%mf$DMxq3zc;W{&i-Vpl~K3Lg-hW4P3-hRmWtH|CRvj7#GBj
z%Nm0jWjt$a;HO?X<3uEKwDP^Vt+#qdZmuYiE-$PhiSo6?%armm=ikc7yFb=m!w<R)
zYCgVE7F5RvA7V4ZKM=AMONvZH6cli9DoQ2*x19E*S8j`Cn9(-|cGP$1n9jP|(~YnS
zzt^c#`5%#j7GIi5eWs%0W7P8g@`h3`W_(?Xve-ABk<F^Lf+I-xOfBtdpw*kNWI6Wx
zrPjW+T_NQj99hnzX<w=Tr?E0KRXwPq(a#2F$dZk+vai-Vb7^pA)r85o{QAy;V?F+q
z57jvMa4Ex*9jHJSAI9HA;xNXjvLK7u$OK|30y&ye{796-!qg(01d`Ba2Lmd%z*CGF
z)#)0={r>uG9{>5dnnp^Iu^MYW{8OayFuK_>*SFy-56QsPWWbZ9a=B<p*lmIORS^JW
zn8ruS?-~43UzCcGJ6a`rq%*<-QvE_tZ8VE8)!fyZsjlB~g`gSzd`IO)oiMagJ(BwG
zAH_;x$)hfb7R5U2%#YMrJr*VZ#Hp)QD^4wFD#>o>>p`!|TSI;+YvI7ihYxaptM@$;
zlY$bbvlinojmG?Uuk!@5m<4xlEspX8zd>Zu7z84K!epSKH&jEC>k7(yk)?J4bW6|j
z(Ph0;J9oQY<L)+>R)=49XH3^_<1sg_DpPd&&t>r1*lc#l$tu@u5XAACJihqt>497<
zaghb1vRTv_>uTc&N~$0`CPc4epPvm&ZI`Zc`q0ULG8zYw^nE66BUtZwvecitpOq$M
z-Zwsd6$R`0jZ;r3=11~va`J&XEN;z443NO`rWqW6IyFdz$Pgd}8~Grh-<gURoZaGB
zCSomv2Oh3HetlfRga;<~cE5V8J-<qTS?Cu~D4s-aIK#w;ZeB>v*c;-56H*J4DcE`$
zyd<-C(o2=09%|*YbQRr|aLSoXvEXRa`sNBhcY|(QU6o>q>B37QStK?7t$6VW%*c(%
zu9tg*wI#m_Y@F~TM6&X(Zca7f+!PYn92^oKRP*Cid?aJ<i4@S^EV+`8wXj#JK7A?H
zmhvn>4j_%acdLgBAd^(qjP2$*qDi|6!Y)i@o|V=r&H|*WL~JN^ex1IV9pggoPoqb=
zFMsLGX_t+_#WixUR|@_(q*5(I>A}h)*3v5T;ipk@xir5|Q648h?ja5j_~qH=>%&--
z*1q^&d8RSz-$UiIPTb4q%=z)edR{9l=CYUvn{mj0pn|eM?cyx)QJ&(;b=im9LM<vj
zh99|_XkAZ|1T7IbY^a5}^1?A?YW+LlyptIQP?kY!G_TydeIaHLtDJR~{n+Xs->uH`
z71%kSX$q%z5NSVU^K}rro;)lbD_nacs~(2{9B((U!v|jzX!*Q^iA6xA>`w-2)`I5u
z97k7Ez(?d0Ds0ml{1wq7uqwwf$v)z-jqXI6uqq9fiCxb}#FD(pqi(WJ9q-RJKR5Y(
zG_6J;B3*c)^B&jJjA7Y|L&^UD^(!+es0ow#551utYEuXaiy_d^e4WqH=N&bE*Ajcd
zW{|VQn8&6*CqwpuudJO<DnzwTdMT-XKXlBV+sqjm0us2DRVgUTz3Qt?F8ObtNydSt
zFy5zDt>=h_+AZ?qq?>~gJ9CEMG;c$dAyIBpXKbDE;(qB_eFk2<SGioq+2TC9ga>#b
zure#qO-^bmc5pY_R)I9bWt<mbvwMsy*TH10Id4^LYsEpYG(OPW?KO14^%dA|!=s(S
zQLJHqaoqK`VRADgf^j|E016OAKG~5DQTA?AaC-R4GyfblE>kOo&$1~Ho6&Xprd&g_
z*F=UX_wqQC3{76(%63LiI_W?ChZHV5r*RU=fD^ik&&q#?kZ+zE23{*i6ktgwgJxyM
zv#=Ka%TwrCEGbXty;F;4%<A}TtN8ezaLJ5(4$HnECjr3i?FZ0THI{dr@F~q=#6ywy
zIzx<fo!QdcPMYa|s2G&KJw)v$7!5I(PNm>Ux?XHf>lQ39p2T03>P<^b;rV5)5Yc^B
z)|VG*DVGz(j3El6#LNWP0&|2}eO_fxW=(TjBKE>V{N?I5&-hi{d=ZTLUDQNpqFMs+
zsE)&9>FtwxawKp(urmJGtLjm3kTC;gqNOAo`*M1c4er9j?7rRoW1&Vy&PQf8{)R9^
zU#z^e`%7z+L`H}9+;f$jLIri(_arC4N^=85mC}kh4N~g-imn-_*2*~wJrcjf1u2p(
z5rU~Z*l1N8uswrk*0fdX^}4Z2IO(!DR^x4ZjIHB{-SK7Ve&|q|KUg~1tS8&0TK}(4
z?5W^0<y&PzLaXa{|Mv!NKFm^XQzxhRvFRfuQ!_nEA}`ot*~zK~rO2xJ)FWS)pA=13
zlBzMz%UwQ~NJe!QRN(=H54qw2O`=u~`*PFkgi#X*8C>RdY?nQkOw04NQT+1fJEl8g
z*Qk)1elr8M0Qd^<@>>F0A6p*mBgUokiV(vdd{+hGKmX%Ng#=(Ne9js{dw$#Qoi&au
zzHt-RWo4xp>0*i-@|n}`^v2s`FJ7v%wRIu*LXNJ74|d?ZcU6PFV%0AXBt`^AGQKi=
zqeATkg}veMT}XIcJdy#YQl38USvVYtq3s~PhIz${5xQL4(Ys+SU^@i}?6Adi?I@5*
zFr<0v{1(pqaG-vp;~MZ;Y?afn%sIo{3|<BaQ1W9eb^8<nDjty}e?Fie8_khBnatqf
zwmkED$PMRsqWK;kL{Q8*KK$`HBsGsScyz_%5~E)FU8*t0eMq=j$AzyCpIgM-7L};y
z?Fx5c;ASK1TtvaaeA8v8mzlH@^zmw1!4D2i-FkjPDE#MoSA+{V(U+9&eQ27SyR=Af
z3<@#nmE+fR8di|2!Ucb4gU9WuogFM}cT9x-DirqZGMWpbyB?W043Qf+<3S?WI~@e@
zim8(t;sa(5CJ(sYwZ9lo7mUR7!eRafoGlrO>`k;~dJ{HP5}A6l<jD4C@WLBujaK@O
zQmkrEpWt8XeU#d7cF={0bbA{W$m)4D<1!>hl`&R!=RH|Covmrncg>Q7jaL=j2Ax&0
zfK;3*rV9~$auyrerSH!CAif(-=tWy3N5I03o#*S!_DPi^5%v_p=@3X!EYVGLEp;se
z3z5>bnW-TlbGGD%&`AU%%=g<R^^4^E32Vb%x=X4%VxC2=*vb2KQm?deB=dT_DI7?-
zul_W+7>tbvAOrhbDgE`Ovkv%pLkc<#DUkCI>+*%g*+)I<OOSycp5l7yeA0}9Zsr`*
zfgs0AwN3|Mh}GqlD?eOH?@mnYvs3vnA$W6iQHY4u{FU_Lu7hE$VN!gz!$N>K@7Pj8
z0Apj!j;DPwE6+5M%k{h!JuBT=8n;2tQ6%FCX$c=nTqQEl723(84EyIr3c9PyrKJRY
zS3uswdPNyZ<pk}#iYs-q3sS_%D;y#|75kb<@uj_M^vwKoct#=SFeLffH*_97*ymXJ
zMYYdev?TxMdpSYR0wnVx&bx${+HcMoQyE9@S;o+q2mp0>_J?E2$<C8ahVs|S$wSn#
z?;gR9Eqdx>hxdVzz;~Ir;J<C|wsRG4mwQ8cZ}UlUTndE1On873yt!+6cH8ObnX7Wn
z5E$h%l&$;MavG$XD*~8LwXoG1Kp7h~XC&{fv#`kZ2#4HdFV07x6X5LUaTjqGQwKXK
ztoAIBVZ~7>9$m5vgFad5cMK^>5dhAcxH>bwC5zg84jbbl2F=*C5U0VYJzNs?lw+-1
z<iAA#xq6-&b9JFX2@P*>xB&(Dp@XmZDQ9$$6(UQ9Zwjw+KWv!(vE%^ml)HhC<cx|(
z=Ku2eC#m|O+JLEh4E8rv8hsD`kefto+>RE#gq}Y)pwIWmOGgt4RT;E~y&EjkM@1~~
z5|5h?Lci|)i`;g<`<4F=dH2&s*w4LrAo4GtCLb*1@j?V-q3^TT<zk!);Vs1lOSnt^
zPEthZ5CD|MZ9m`vZkVzJt%l<$c}eK$dBa)~UjM}ffI@P@obRlRFn<Qiiqn;6nI6NE
z=OchV0SM>4rX755MzN?N#`uiWY~%Bau||q4C&x9kG>E9B-ldcJ{ZV})Y2wWopQ(^D
zs;s9>DxYOl7gd$l>y%f9&H!#|l=Wb~2Do5$T58Zzy;U?A40w2WP))dG3E(e}Vqk&a
zK(MysH6_cb5}b?#KYU1Y*DZLR?#_15%MEXVBe8Xu;okojD#wTbMhd~#@1bf}-BK&>
zq57u(#O*>I&hNf~8Nfb*O&ALVfH_q-_Pt$tV9&H3bS9`8AI$6(@Dbj>*~pouoSYAx
zuQHseF%^X$&Ea(UoRLk50mL1+BY~g)FIujf?I$Ue>$wQPKiS}?nqqk+m&LX}(R%dn
zj*72^YpbXUN(gpQed(PLy1wzto$Dl5{RT}`XUs0-a$FB*IHkUFv@}pN07VvDi?>k;
zfQlW!_3}i{#lA-#T$Wk5d0<8xDp0uzG)Pq$_LU|vMU9$C#hG^QP;F1690-k-Be@=-
zE|WWnEcn%XBxgVhQa(-1N3p7)(k(b~)PiCCk(b-~#AeLj3)l<#m71G5Dy@w^={&a1
z)$66gPz`0Wx#RRE-asSN)$#Ik&>lA&KQ5CeLAd#jndj+;SyEsf!#|oiE>~7h)N>sD
z=EJ0sndH@h(#cDr2GP;ooX8FP_t00eV60uLo567%e83GoDJiMrTm@05_Yu0exp^KS
zfGU|uK{v}?gaF+4{1dW^_;5o3iuP8X_S+1}wfOK5ix-kRy~Er^fLg`qBx&*2Iq>XP
z_e!XvQc3gPNev!=|HC%7V1kS2+yBKzVC}8!yQIiwZK&n;xYcE^TylW~SXn)CW1-qu
z&81(y+^RQt85|&azSeP7|Lc68u$cN8)&k|;9I(eI)}SC}2mgqKu<G{{lPbJpAwEr4
zWo(zL$q}igR=PkRM^jE7d%Wt{l=ThL2d1w~%y<EB_Xh;MJl3762r=@dM?R^_V{cjo
zCOfo3;>)*Eue!#j@l+0<A{40L9K~(yhRTA%!lWS&7dV#(lksHi+C{E^e`Ny#BuYw3
zW+?*WAoRt!U~)G2)Am+;5WcUmVOLGgsNion7_-(YZ}CqIBmg)ck0tyX9}mEIyQ_N2
z+!FFEPQhnJQa+yfyGRQV0f}cXuDm5Ci)bPC1t*BUCg4<~d1nfBdTmw{S-{Z(w~2>Q
zH$w=FBY`mu(pk#TAWda=1w4SZAuh$0@HK*I&bu}SR9rb7uH|?6q)PR%LLfuc(DY}&
zfgoJqxpetz?aMvOudOq^4b^Ry$ag>ONCP)_SuL2Eydu@eR#3j&5phYzMwebCLC=+y
z5;gLIt#UD53zxaNp3+~k)(H?SUwAI0Is=g}@iVx-h7A#HfdzphspRxWJ)G7h<o~IQ
zY5$E0?8fYz{=|2y60_0me_0fCn7K^;OXdw~G(iTEXM+zbltKtzzEr<HUeN#Xj(Prv
zKE`GU;?*ygm9X1ezlZB3E4^!d?#O~t9(a3vbB-Kc?RJM8>rlWA5Eo2;JBLw#ky#nK
zliEIcq4ilzTd;Y4GYeEq%{c9Kk4$+!^i^1I?ysk-yl_w5EL{tDL+_Eh`F@38mvkv-
zJl$5_ZN`?l9M53xnvLmGOa0R-&5UE=pKrEuRHZhr8J{SkR81BM94?EKN2AxJbVs(;
zi3`%LkS)hK5e0tk`toDR_s*#Fh3XNBLkolNVgrF5={)c8Yt35wYfPKJ><q>dB9zB@
z_83t0(0Du*$k$srnW3*&5YDw+{5zM4@h#M;Rwi(_k}LPnMBYi?OXpOo*=y--&y;pA
zLS;wBYl+C-@7N_>ud}!P^j58GA)T9&7$riO_Y#%!f_8W9svxB=t{=c3T_o_psBiL@
zd^6;JI^-$b80e1`NkBb}!xZ?!*w}b*Xef5okJVvZ*`1L9O!@fbYsAL;i$unIinn`J
zp_*}DBhCz%!3<)>X);<250pgHwWw^(2KT9DYb4<Z@~wdvKumAU+?PaM&70rNsmIz~
z&?Z5vqyPCFF6nFHf_G?yo^VtL&kVzYs6fejkSB(2?t#i&T8>QZf72M~kT)N?*{Zq+
zzm1)wyVBK@`CFBjt5D;5nUeW`$6)@K-2xlp^iC!yrq)#!TrN2|Fa{{7LhaR{jjwkC
z<C@CKpPp=wVQ&dM9Mr5;oS^2bEsGMhfsqY%KTk~e8sF_McGK_}A^_FznoWr)_|log
zf4={{7HOOMdjwT}FPq~780XapPy0)B$$e!$8w%S+LU#L1>i$V45SIw~9FU~fn$aPl
z!8Hpd_!ez~Wgod!J8txfJz-woSKZG-02$8y+qxvfI`o<>JWPP<XQizAjs9)JT+Qxk
z<>c8pj$;Wuk5m!<lU<8o&3lJN$V681rKUW)0y6V>;iokrl6J!fwgI?sZAh&*hyZ@o
zg1Qp;K=1$)<Cqg(<8+?9IV|u8==yJEnAGF6p@_ce|DRUhyeS^>h!e!RhchUY1bi4|
zqwV!A<TQQ#h444!4`rzR02N@0xc3T~{|AuVE|PiW>H#q97=bSh$Ee@iWiljx7XC?6
z=1u&?osGGl@KjXeauWVf*x9%5k&rZm4p5X<=O_@JJS^jZi%3~_0-;XA+M(!Js>x`Q
zl9KwN>1GL+-6^$^8)EpQpSn^Y@qbk$MK<T*-&%IA$V?wTdYt6UiUq13IX|Sl`k|B+
zlOTmP?yjC6Gggp!8iDHP_BAHbB7e_0W-MO4K1@!YF43|Q;L}}}&cU&@c5vT#>aXJU
zo$T}Qzx*;(A@E}FUu4nrmN;l*PtHR+E)It*V9gcLEn^>fbVq-h<u)5MX3x?I{J|f+
z=E(;HeMZurXW_q{O3U-)i^WodO55J96$*cn#{-w<P=joT6NK>q;N@zAa(wwmxS`8t
z&w9gw4lbZFHM`s9#Z95fo4!`*<hatxtCrUpylD}!M{U8cd^tH~2R<&d`NZ(GXzGAc
zAk`&Fou-xk@BLuF&|6NvGO1VW{BI0Q@sAge<|v(b#o}dl<s45;@`mO*c;?U|cJ0=z
zy&s_B6aq@h^%fq44EGo%b?2=5q@Q2jn%1OlX(t*#=_A>UG#){stwD89lV!G0zeq>Z
z#XZ4-e=j)X_?2*{tcY2P5wTj(pp`4z2n%^l|1yy>&po_dB)cmcJl*-ve4y?43ytP|
zeR;vu-<_&XJZwtCPs;2#O>(rtn+kCa<&Tb*f5J5*29Y1P2IaZlHNCi`8jU`vrMLMT
zOl(nV+YFgA!@uVDXPH)ITvlWH>NIbrlyi5zh}4TWlm3UiYV$imbQJ4Xclxm_V6m`Y
z?KT&7ZYeH*vhY_#@UDF4F#lCTVz(;Z)DB#n){{aPJRPQ1@ag0A2a;m+K0pwn5taZh
zOhy9a8}E9HihxO-Qg6Tfm8e3>g0@WFy`=n&2#@%t577Pn-wDB4Jf>u|X1@jpW64R$
z$yu?D8=_d?=vq*sp_uYs8)pblUe_5XHh)vak<1U+L=8(k7Zr#Be);X$p|}!EZ@><&
zpJS=4?~OJneR|qD4Bvcwkhi}WQ--a*VtlRMfoO_{%w0|#j$PpOBK!tjsd&0|-3NM`
z?>cGq3l+$Z)@$G``aB=ik{RD8n4RWo^VQ_5{I89j+v_&S`;&3Msu)%2?z!I~vMG9%
zW6OKFPR$e%ACS*W#!`=jT<@B2U5oh9y@lZhFU2A3*^;c*xGQ{eo9wnk2>?L}2DN75
zftO>P%N<^vZX4am{z2d`N|2Jih8lJC%9IMJXGSEW^c#@Y6%Hb^EcaSE#hvD5*@|gn
z(as{ECv<3FNb)nskF#9s*$!u7+_rYEb;6D}gZiMaCV{hWf6C|whmKr!TZCE2KPb#>
zp-tn*MXT-2JlH@kYj6S8Q~+05eYteOuVz_%9y(`|`x$D9vy+NWV4I6UCVMYaXP=eG
zJYS#lJdK{4K&z&IyPo@56q|+pIiHsl%MUE(N*c<8I+VkCC0g)ITLnsTSeXjXQ#Clb
z))igg|NZ{q-)l2iQ?MNT5#xNbH#wS^DUMXkKZg;Bnl!MgCp3_N=&tt0So55(4km>I
zZ@)0ecxy@-zC57egHuT*_wCuMeBlD|0;VZi;=Wpuqa8o?y|hXhPYGY`{SoMw>qq+I
z@8F|z#Vq?f=X`^6X^+hdpJ250<fy;u0Y-KDr2Yg4YU>9J4WL6$vKMlZXc?3l&uXBk
z9J!?&WiomWp1%T~Z?M680wjX^ighV2?|~K1((ewbh;Il$mlF1}fG)%s=URM!m6Sr@
zEmD@tpoBy_T_9=`CnWJ-d>83Gq_8fRN#Jo`Z=@n%i*(a`&)Eq0$?7&`@UIH~R!kOS
z3K5c{K=Q2Qb-n0SlgYQqN5#9hJBhzWzQN+adMw{YOGSYt?1i~GExM=ls{TZMD-nu;
z>z1xFyoMB3kBdZoSxNe|g>2t?SwF9Tv8^5bd)@sIqj_SOQjg=>3rA?}aNj*stdr}k
zCjSyvd&9XYsagQqd8qxU3cMcJ_+3i_@RIs0*3!p|g<t$}JM6Y<$v39kkI=*D(>Pe|
z9^XJlHBHL?_~GZQ<)Xo!@U=X%AqDaPp~_ZblU({dT40o^mQwkSIlap0B`kQBl>9K6
zgA^K}vYksx<Mo2nnEQq9OExqWn|mKFn`~cYBkW!D)GP;g>>ayrOv?2P%D-2Fmi?9~
zF@_%_cv{C}$7bU*_`|e9lF01;+ZMRDB6Ie(6kOXE3*#dCB!oh5rX+<9z8F_|p*wRA
zQ@s70P;gC7C9Bh)QBD1CYa1g?L2II3&hMXM%OdjbF78D_WI?MQw@Jfmo)0xK$qG+2
zy1KeM_W-Iuv#+11Km*>W%(zYse{e24{E%>WXTol&arP?PNq)zd25^-wpGArTUTf-O
z{rQ<66xQEnSFn2j9rnubsrHBK+_`X)((UqEC#vDO`tL@<n1%;Yz;e?hAwVi`3J3;q
z-;}Z!a>mQLvG4jR#-p{ghdK!$#76DyNy5u=J8eaz-5DysFbsKc_2x<^GvJsp{dIf}
z3D2WW(vYfOyBkF=?oe=y2rq(n2Ol$8f*Uw`N2KZqDK9WLjkhlXL%1pNsE6*uX+M`N
zN?nNoi)=*&X9J#QH|_M6Ad;WxKS>b|+5JU}k)lmNjbw1G`u0f03y0G<^>$F>5sgNh
zA%k1;B>gxFZkA!1EvYEmbn`a1SVoL#@)dI^|4Dp~t_q2_eOvq2@1+pIj+3x<)0Sum
zs4De&={EhphlYROm;~x#x4xQTE<B{xQNe#omg7QdGFRPMo8HZj68|h%Tg{;BIQVuo
z5Gtp@ru}w>nT#{q=~a0(>vPDqyaM6cy53`<o8QT<J(l6Y!o#}@pEC@HDZvn{;i>c`
zX9;cpoQWr7YI0>JT1N4gJh^tf{n#rGv-){wzCo^a;>9Um*_7$Xi4?W$Zw~C6IzAPA
zf7<_agoVM#e*yj<SzjF%RoC^cA|fH(Dc#-O-6|m+(j9`d#Lyv~GLi$*4H6?=($d{s
z!_YC`dG7mtzW2MXk3YEP;&A4i6?^Zs*ZQsBR>YhLSrCb_`ilL<;F^HaPWQDq3l<0C
z3d0fK!4!`H<gtT74WmdZOp*VxH%oSy+i@lVV3$^n&E<QE8DAEO>FVnG`=?Hs+7?b(
z27vo3US-v*#LwZi!IXG7lU?Vk_+FBGqy{8=5`gfRMnuFwXiv3fm)qf1DfYV{@gZjO
zdu%`a=lBD!dX6Fxw*hCOt&1W3V<$Lp$tFePCX@~vFFfzg-7U<{0>;L5;&Y`fUGlu&
zH(GlN<QR)#Nakd2Z#|K&BFg($eq7EGRU`8fsY}nRSS3QFC_hxZ$3#D3Tz~7dcnl>J
zeH1i{KrkF(%*9gl3*rizXP>_6tdKaYnf4YJpfZk8SJbt9l;#AXv{eT;2SOQ<cPRB0
zjiBkx{S=28e#<ppV9^<_>tE(0lzRDOx~J~ciONgcXx_MBm$oV0$RwIxw9U{;7jw{}
z;}>UBjD+toYYLvLEf~&=dR+k#6iZEbO%-7%6Ulq+PM{yVX;wW1XLEm`QT6b-Hgb~#
z7pp;m;kBC2ByEk8PljKDMDLsw?J7ccF;NmtrAI}#eU>F`9N$jVxK*LI35w=Fwz}lI
zMp-Z!{bX7}Y>0n3_Ut|RJu*kuMuH;K_%73hK3Ha!)`HAm^C-lW&$uetZw;@VNCI-T
zF*n$LCf<P60lDrzg-ei2)?q!_V2$$^->*GL|6;_X-s{qSX?i6+U}cJ%B=MRdkI;WT
za;^_9T1TW<&^X0kbm%dTr?H~S6?pgB|G~%RLx};U4?2O|QwV6HM4tAGqx<jeMf`y!
zMz`THZgf#<LjKyl&C4`CO=WXs2k<xZgRqrUzlk~D0d$wo_KGy+7%)PnNHJveAGZ4&
zwv#`pe+4d+h@(7yrMC35ybGzh@wolKrY}qpb8OQ!V@{YRGTO<rJ0H@B$;-9qCDu1a
z9^Ppg*xv;8kDhViSByfFKU0xFbA@e7&qo(cc<-I+f087M(HrT)MNzSt#;7@^@l#0;
zym+@M(`pBd4Z}~0J|DSJH*GNFnyOa%5}^LU@-e%m5@7++1^PV4$McYKnxL)@xsN>a
zU^deZBTr<`_u-?fKVV9<_t^#w_Up5zK?I?!u469<{SglrFU=tj+-}=%G3p+XvyqNm
z!yk6&eNjwj22xq^3+;o2pr(^}bM866PuYiy8oz^b-tBwD%cYp?xy9s0qL+uPK(zfJ
zhVSQ{+eShjOp1g+e$Lh79}GxRPhSa}E{(*8x~&vzN2y~>4q;4h+JqH9GL6ybww%ze
z6fEGQw;59{>Zb)AsF4H}g4-miREKGye!`{;ez}8g%)7n5t4B0&Au6NFB@q^hC~eka
z<m6^Q?xx>vL?7ujxO-Xbk~UB`?0%&;r)H3Rj!TO&T36@(Qz4~ovZV7@?pI}>U8YYs
zEbsg%U;(`GKx-plp7m&k0iHneAXE=MPU_Wnbcm6UAv8bV=(QN!x{)#DA!5^i>-((U
zVOGQc!eL954@Qhs)qa&VhX8VTGzvZ@?Edq5d-9aMDOQhM%jR4UoJZKw<z8;=K5WRh
z))u5rQzi2mM@+w`2M&e4?U1Dm2E={*;D(+xr*}PGq(M(SJGGJVYm@tSM>CEs4WG>P
zeB5FAI%COLmR3}!a{YD8)R(Z5SpVMLZ_`@~q{q}VVDw$REyFDxlC7|dKM052?k>%s
z__1Ue$9I{m+g2Q++d=YdZXNT=)yX53K=SGN3v?c*GvY^|Gm&jdos#ALeQHQv+)ghB
z<gjdcIIH>(@$KA)li9r%bqpz`BkAU2ldpkzP?`Hmb**xL^Q{puk<Z080slVvr=oPb
zD6vjGCZXYM6i^I)U-t28Pk0;|HX%+dm9MvKo#4Uxh(^^$=$Ponv8IV)L}k^+2-f0S
z1D1io+4QyHkK@*n@q_Puw)1(^S?xMh>Z|1u3@(4=ll2@5v|F=m$A7&#XetjE8Tt9U
zSlS2IwMe_h+AXD2o0z@~U#dpsR;y6P1fNZI_hT%~c#YtJ5eucG>W9{#7?V!TvuHiG
zuzUja0;S%3CZmqmXsN7fS0Jsg8BkB@EmmVGtH3+xGgvyy*BA92W_GoCtGB9?cN@1o
zO}b0$d@j^11u{PPgrl^tf+l!1wGFBg8=*4Wa4P4KAMPW`l33fxFoPUN)o)I23*a=L
z+aQCnKZ<r|HL$4;<P6W82V@`R!wh5sLbFK(@?i1X<76-_E#ad7_LF-clXxN)OSV{!
z1chRbCaMG{t{VNtgo{e{EYYB%Od&P^@ss;y6TMU5KA>o26gku?G&IW=tkR&GXVA`5
zZKqFpnWZb;oyI?2=tRNiS0Q0g?;<ywSh>BZK3;6;QSa9TYVND@wr9VwP8R?c)7i68
zMwgA{v6U`Au=j62@Mo!hXP3zU*eenL@+C%f(=ye4J*nHDy;|ijjj1{kru}%SmdB9M
ziKIX*+gw4nYB}-feGO0z>fo1@O>b?d#PxxQ-zH~p)pH+uK56?YoBANk<meHv`5Sro
za^T^_c`_!Gb`8D>v;MaeB$ib%`Tg5MkAlOjs%NQ^$)|6A>A}bHWvl`MaXtK(FD8?~
z?Ru~dnwKUWA^tAR7Mw26I|TPkig<>{3Sk!Rq`o_wwQ6x)HskH7;M@usINNx;)?2D&
z-dnWsj$6BI1z+o2nMYybJ8V|>>#~R2ivpQj7_q2z?CViVNuSssOs260ODLg)p8X==
zqOA+e%_Vg-8Xb4l-AuCQ`8uJS#OIU{@1&>icIpA;KOpnrmd_=re`JuOgZimt_jF+G
zsQAwqag%S_(|rvXY7-d27453<zcVr3fjQ6R$$wyAT3kJ<u;Kb%Vmoa<3Ug>%-~?jX
zV#JBd3BWsAej(<SPuWIeBu*J%qLKvefejbC7wev1#M5KGoZ{RHrr%Ck5O!3mb@YwN
z;OtJi)w|o+yAf7x-iL4EYg2zW22mznbiatdVBNrPDCdu!lWD(JgYw%SvJJQiqMJDP
zZaT<YC2lEhodSs*VseLfEr}Ig%RLR70he}UTpJ>ZX^Hi*zoZLct7I*&l9hrU#2#q>
z+J1;yG2GHb+iE%^nb~R%8-&j55#M4v)QaDy#8Ze`EdTnP59z#+x&nF<?ooA)J-k?r
zvE3>0#j=z}W%Fu6hnOl{WF8SSi&L^I5%)}WczlTnJ?40ab;oy*{qSk+`G_D^L8y#r
z?RI74k||h`Ze#V~nSQQjn)XP#M9_k-baI73;DT?A6K6=;Y*wk0CF4kiu2ES(jb>RQ
zue-^7uz?fT?3EoY=-ej~#%Mzy^72>&yJGC#uD(N(H#waC7Nq%Hy~{1kXaQ@M;zB2H
z7GHk&-S1ok0X5IfJ%v{+DSw2W+(_Smok8+38X-&M-Q}U>=u*qsxo9+&q#p|X5P|Hv
z1xO3PLt2>QTTdcDe1x1~UG%ob%D5(faOpbs5Fx$FMg+UsXcq-$+!opvi4SyM_=LsJ
zpk~s(BwA8A=`Xg#<lN%v4w`i~C4Hjkq~xE3&e)Z9v%qO|PbPV*ArFRgc+6T)32)~w
zp44g)Sdh}<;lmU{JcFcV))WpbU+Q>SD6&tG_@&)>lGCk_LSvRVX9GNAfxEoaY4(aX
z+GBQd98#(c=6Ih&lsKw=tA41~&#28(lKT9+DEmd?+@~GE#Fu(`{!DTkT1Jq3Pp4>R
zS~Ws8-EWoLiE;+ixXB<D*d@c{=Ut9+e@$ZTIqchoI@=M=Sys;N1duZIud6?3sjBPU
z&J=5V?XPV%USNBlzO0NwT~;=Wve73n5Xno=D&fs<kQ`v^<))gfIaQ?IQDE}#_O>iO
zrt^H<k>zuodD*L)#*k>Dlw<Q8cf)7D<B#m@*Hg~J!;!<=C%o&otM9Yku;1QA-lsef
zT{AoiEJJ)@SMa3D0LqT1dMa<!?OicblS;OY=casAoli6XeWavi7E(R@j-YWj>&69o
ztKivscYnTmh9wzI?GYr8kc5U_uY>5!uBb&4L$suEKfy62I#W$HC`O(>o==o=vqQ@q
z1yMS*J;?)S6~FzCiPTF=G*~MuzT|VyVq;*lAfV`OhH1WLF)1#b8<iQRduSfc?Rt$1
z#yQ77iYbCgtPR;F$DMq?aM4OBC8f%kW$u#u*suEeJN3w^^3>&2zp!SWaO3+~yV-6L
zlRSj5F=5EV%I<ZLS7q$XfEzygJIcclOM(-3`^u6Z_vyW<UI^`v#Ri#X8|JA%i_g`r
zUo@*wiVEg-U~RNz2>`z5{T$B7V`plOo9aD@bA)!$bweL;HXNCtT@>i!3*MFO2tob^
z?xECEldNzzxL{Ox>dtcP2Wn?m%@NOoMqy&#Mp~V(y5TyZZbyHtCrCfdCqaaTRd;RG
zow9pFW%+8SGM;o>|CTn8!`_VWORGJqJIhJ*$0nOFFi`3#zmWPCbm3b>e<nZfxw8L5
zyG7gu=Jt`?Z6tHXG^e%Kw!wllj$F^H0NejMq{qkA8?yqdO_z7z)9%jzKQpDr#y@l7
zPKd3|o@Y57IBJ-;uf?wCEzass&)u%CY;UdK*Po74x^zray0l-WnMGhysHw)eKpz{C
z=cyOw<_h<fd{W9(Qd!p9uUGJlWjRgSA|p$<Jnvz~DNM352`{cUid25+M<s;-4ki=z
z&PppuKR>oB$!f@;8oz>7rx-t)kW=M3gDCtV1q<}hsufJbWKbvs)QOetQWf@Vx3wGn
z<x{FYWa^`Cv*^4{lW&Q(QpQeE>^&gRE>)5LQ2vfG^=E1M&uaGuGzQ&LZRP5p>hvGV
zK7OqB9bpOKx7Q`8H=p?#oz?Jj;?%z6V|Lo-$Ob!OV}pZ_P>~u^`@gi4rw5;gp-vfX
z6&j!amcI|3ERnvSs&#(z7hn>1DX2DA|LbULIjDf<z+V|<5ceVJd$rS#qwd4GM#4py
z#2cmnmD%uktM`V&m2W7@Ua>ojBl!1>GN$*JGNv7WM)H&xl+_adEq-^2_Q9)Xh(Igv
zo$!Yeea!7r$6*~dS++QL0+l?r8nxRbi-X7{s}E_qAB%-lQrN?Bw`20?q7X?|F~F!V
zCEF88YPKO{)ljuowy>iC)}<k>Uj#qXWVa7ewW{3M>z!6H5Rcoue+<N4p}d$~_pgQC
zygFalwCzc*#-mYpQdgyh?|p16ZUB<#TPQrT^zBMU0X+p(3Q>00qx8sKh`$o$E9<d?
zK~KTpyP%jcU(ZfI(|X11>#RKa2nE}j4xFrxm1l9L6$#-^krf98(lNfab4h`CWaEgq
z1wN^Q#!pj~&$Bvaut!W?A{mYk@KO=kZGEaV6*dBSpVduW)DD@P#O=NZX2T54F9b>R
zdb8suetf3*P^t60f$Xnb>+C|*$%0yMLYpjfKB3LBvuX9D<lQ+lyI~{#a=DO_>PVIA
zS>KB)=jC6-JfB{_c=4j=8mjr^6T4}y$n|A(j1JzxnAh~$oGx0!&4DZt`cX{zd`KDB
zpK<T81b}6DQzx%#-eqj<*Pt!3awkTH{$0UCYf;ViDo;6;62I)9-N0wp<?dx$f=M^}
zCgwwIN!H=b`PzjA^#TTcbiVFo1nCj)819If7L%&5z>wz+Fz(dE>B}J|VfFZ;=0jTO
za4!UpF0C2Y;0EP#u`5TW$lZVh^u6^R%Q7pH+_k!dVmw(2k3f0>_idoy;j?WM72261
zea#;F(!}_n5aNLMw8tL9SoireW<$}>_XO7+JntLPSAb3KFMl|-$mf%=(eI+@UI%}b
zzbQ`rojG@jH~gtv!SXUEk<+{uygsHnce3aL?FCl)9^0ISFE2jEIkLlx*-s-#Fhg-%
z?NFBS1BFJUkJwLMbas|cT1mfbz~8bDuHe75fXNPFeOs}EssAu`Y~1wZv+s1b9=wl<
zxQK7Ts#T~kt@f%rv~#MOhsjnxXtZDjZTX*IzTb!b3D_K4HD`I;8e84|oaUodQkX9A
zKDT?ycIe`LX<@~d=`48Rs}Wvd0Hbxiv8Jk_rQc}5&C2<6ERH-p@2yxkpS%8iA`LVp
z=(xc0I6K-iK<}KKgmz$+RBMG&F51STo7*q8@T=`W3Z<)I6B-eR33h1N-XN{6U7P#1
z3R*Y_WWM$rm?e1KbB(6@sH88xvWeB)?N33}M|pQPNZ>~ZdV}_lSl=c6&{rRp-mE^r
zg^~A6H||ZLGS{me@(CuZ>kx_6&GZh*wb{VsiMrK<`mux?r{9`;qJ@mUZ*B;TCGUxm
z(FlNI1X}VMYvc<MdcI?CQi<ks#NsJ5sw(J)9D4PLiWGj`U0ox~Wqn{9n=SEb)A~Vr
zc!F3p=VABdA0x3hyoBlu@%?OzH9g!J>N$0v$agr%5`2akE!KwyfzAD!7+v*#-2P1c
z9WiuPBb)9~Ck|boKcS(!{zAU&U+ZRNg#0+eh7WpOjq6@IJ478iuK%3$>BdFk08NEk
zRYQ(79{e70n;nhNn$%g>I|eCpv-LVHdD)xhy8X(KbX`1p4vyo`7S>-TeAjSYm0xmF
zP7ohGTRNoKYnYvut8vD1YvL9267F(}Kw4_53R%?kYj$i5J!BAVcGO4Ry<}Z$a~95-
zVGG?nC~`fk?PaOzt-dgdBybM0VEzuRn8IK);zz1WQ|l?0Pv&x35mL(%C&jal?YXh*
z<_<jIiv7;>vioT-l)Avra(rNQ+-7fVPkf8<K_aqpTdE^>`P>0Pb*2ghkX)-S=ayAf
z=iM%zRPcwKlNinO=>+5R?F1^|JPKoE=*c<iGAo)N=EHN@{_E~AVcMGC<_x;;Ki+^B
z5?HVEv9FnMYvAN2KKS9y{Wun+uW)kQ^pthBKB9{k>}DA*@kwCiK{T@s$?$UJsL#Q+
zbP~eKyHXnoD1|Ll&ndrx{32hPNW6w-1e^)C5-Dn)_9Z5_x}No59wjV=BPA|Ii4rJG
z_--Wbi3_Q~g&W8^AYSWpQQw7pL(&+p*XEcv>=v9iB+w~E+gHEVKm9Sed3_SR+R~mf
z%26)97xo-4unP{d#KEP^=z<h(YkqmGG;UQ3&TJUB42I~CK3+o+y~XO;J#RS{Y&(I;
zEadk<0dd2g_rzz9_ER1?W6>}Vd<fSVtVtHXEQcNeNnpKc%3Y!Whkj(2Sl0(3Y-);-
z3yYh>Hz;tBwC`->#0od#q?QqO{oMYJfsBUG{>!avdX#;Iz}&$__y-Hjc=7U*KBhD^
zmp^BxQ{;O3?^^<kh2YHRwgy`cU3Ss#kXn3j()hixSlyATXMJFC9dCn}M_ZR$w1n=u
zUu3!p#s$8_+M_<7y90F*{eb1?y}m<iHIgy$II)`{dFi1k(-g2`o{*O_y0vJ+?=aBg
z8j=5p$m+1Lmv~jit=2R_Rb`C*OsfIc-`CC9j!fsqQaFrBPdn~8(%3$R6c!eydvm3&
z;VE)CA!GINyvCWmno=LrV4Ul6W1@mjf~Z*p<LG_FCHao%KuGEz?u4=>8mEuTmi^TV
zvZ?Ri2VERyM3EunwM}<8qZexUcWKEJ?}9Xnm8;PvAfl$rcSM6RYAkvT-g=uv+YF|P
z7~p9n%iYI#;|RgXQ^UE=8f;b0`9WX?ozdE>LE1K-^ywLijC{S+;(?->E2b@uhMn$;
zgRYynmvztDcsEB&<HecJ{3AM^5&F&bCF6(OHZOcgB}F%hG>j;!-&J6rc+~V-aUheH
zSil+EWDtx}h0hjCQxv^6?M>Egw=8p`HBR=qSMkra7X$Ze2p*DKcp$hHC)z`>+f^Vl
z17D-?S(eXtl5mefN7>Go$kw~gT3-jCF5jT7(oY)S{3$LNklzV1KWiq5%rI2D3y7J)
z;as9I^-A{7?K4y}P!u3?PX7LO@tuf;c+}aVd5>!jR9-c|7K?Kl;nR8puH3A64Qisb
zrJ~&Dd|?x{`yqPO8@R%(H2a--rc^a9=T%~QPHvQEEP4;INO9kVpdGW4kM6JxGZ7kY
zY1`<^52>t7HE3zu7(@7EL%zE@%_Pk69g~JBdZB|#r+Yq@2EDeHyx4c8U|54$PBfMr
zeEL~V<fR_*uHtnV)-p=}fJu$G&#_ScX38<+;FBj!yJ3>-UY2&?$=Q84nE#6NA&!&%
zHWQyl<APf||B@{&gE9}Ydx7mcgI^J;R%UtjZSw&Wr=TlI(%fV%@sx7>QVG<iUNEQF
zmSE=te%k%q7B|I}81ZQr$Ercw=CWYqXCY5(4|~3tDcd$kF;Lu_^I89>bLWkGK=q#h
zh7CMRkR~^btTgcvNwexVUXdZXcm+X%A<FGNNp3;1BJ8gwYIkT&SguHXl-DEopGK5M
zWYMHKk?@!Y=C4shw)Nx=uc2zUvrv@GB~jV3Z`QQ~n?Jy;pX9O2y=JFu-QvnuZ$tc-
z*%Hpn|5ClzwEAR0Yp^rIcgI;JSU*<0bstw{C}NkP@M_>XLMh@D@6u#q$N*2xPCF=|
z7U}X9cegZ`tl#ABLY0b{JsI&=+Z*5>Es~#tqb<~P*KcVH`uKuMRZ-SG|DfVP&ZL<4
zsr9%XVZ)q+Q><L`O|E0XzLy74(uSis!n7=LU%;^%ys6rsEc3kar*yBn)zYQu#xcj$
zI{_ryvmxxVhgz*C&D$S#TSOQr84nb?J^V%Cu^_5deG4??j`1#%Ocyb?i&DX_p9Jzc
z`G?pp_lsw7wq72LX(KT9W}!-14ud6ssFIso$iBwW!NW)Ia4d-)!o~Xq)`K!L6cZev
zeF;>Y1}A*GgZwAtCFkGz4ri-~_%DMVFZH79H}N>lpMEB0w79>Gm#fP2f2BQ69HNi*
z#}dKtPNKJGO_K&2L5F+3L?f#q8Q<8i{ftVn!=`_N|I6jEfQ44-i(d4NP)-pG$SNr=
zZD+G*!sL&yJ1p&<8<dK8nrJ3Pt3zCfdjC)In0Nb6JE)yiXKnssaY))96**ZMkbi}2
z1esKA3Gz1jSJ^hI947HlbWSU(sdCf2M15ag@PJNcReGG>umxLl*A6^vvg;QFQ-M*y
zqTp<WC-9j1a8nW?f^Zq%M|7TfTUu2kbS(9y!jUYy6AKYFgUo)j*xGZ>!Q9!J?*@b_
z9ejy>+e0rjX;IKs&VB?W_p7PzWg@-dy&SuW!S+RPlp3jblB7;0|5cj(maEPkK`jmu
zy9`Hwjx*{DIdy$U72dhagzrvlzGda(`{0Q(8%C)|Q}cV1<Hr<4Q&CIn8?nH5P#%8|
z91Zfw-5dB;f$f$=R3@kMV$aw|={hdglPdF};L8mA@Yhhk3Gr_h_$Gk2rtuMeX<Igu
zV7B|Ndj&c7`J_h*mObE8p>6*=ISy6z$R6kI^>{TVeU}0A2P<sy{o%Ve9!kdc#bL-3
zbehM`X~AZHkn`<4M@bkhLlv*&+ORVq6?Be$dc3zhU1?X+&>&1MhBAF}$+MXShpm{&
zF(i&}rb8+Rb;~H;_2lyN=RwrwM&grOWn76~SL(F>>`+q^d=8bm-Pt7Lob6Ea5Sj)k
zDgI6lN74@9{7uIa)(*tvv$=Em2FiVl(y7_yc4G2Y;7-_l&ymm<YI#~S5+3~Cz${PD
zyz3=3Tz@T<RZqj;#?rxo^_Sg&;aPd7j7#1WRff4Bv7$x=#NYjG@06X^+rD5)ba5Od
zcd5ZAY1B%Pe!dj@l{Eei8#o%KGqjod!bfYg7=!TFA8g~LO82*8f%37HCGS2k6v=l4
z;{S<?`XQS_AN8t^%ksc!A7N9g^D?oBT^vOh{}4p{sdiPTB=YSABC|;NSWQb*&`rYA
zIiwT}ZjNVLL!cHM{+u(m#(_lh;HUuyxQ<~X#;%daC<a!0&w6?shXo9N`TMhD`HRtG
zrDx_aRNzU6R!_jxufsL=+^nR0!zg1WmN?;m{@!n=Y#r0h+zM~Q0*FJt#)yK(>n^fF
zdjL4?iG42sWnUB)Dl={s2t+&ryu_|j{t?%FOc#5J<A(gNX#o%!*5}R^(f)lBbFeuB
z!UeyX_crLnys%-*ZC3IL>@>xJckDAD%qJ2N6sq9kr)&N3R~J#~b@WL~lSIT{+eBRf
zv*@;SVewGODM&hD+|}&Vk#}hufcAoQ3i+LXDh@LlOrr0l+ZV&KnXNQlKR#w}6hEC2
zrTw((0@on##%~SII1wQR{K@9zm09tDSOrjGcj(byvRfALp%g1XXTOC_3{V5DN}jc}
zmt(;bS47fv7ljCAvG1*#GN;XW;UoLOZs<Y%fvXiH?zyCVdHwvK$iKx=ij#74bMGH)
zj^Y(_dOjgd>0x#yCLc0^$$zlE7)#nAYNUaeuni)5meAy@@}T)jQ5FO5g(S1QZ6hlJ
zDXm2f3|4tH85b|kmE@9$V*+r?pR5lqdc8F3tPY%not{SQ3j1dwe>n{9R<%NELkt!&
zX-})r%KNvbs8Cp;$ow_^pG1$1WR<r-uVF6j6*O=HL-_~oZ;wQ%ZBV4%)4!8Cfe!uC
zC3CjVLz%~PWjN#I060O4uJq$Y{;>@oQw(!^dXh?}N3SaRa&o#7#;xn@NN;=fqytBi
zv(a$1=0upuNEW|#qvJ<^_=MtIvs(f2j+pp!67=@hK##eeTg(dQ9N_^&2j1>UQH=-;
zQG5>$^<MUIlSgV=J0D0F!$t{8u4>{Alo<I7NyW4E#6-M`gQbziU1`QC+}j|>#is(F
z5rUc1KL2af{rL}%jaaDd^GlzT!&X{U8F}eO`SSES{C(=}rd|O!o$kxSSzeu*?>Ym0
z?7k8i_JXf@m%`t|?xeT#$I+8ADf9XPlZH`I4l0s`fqKDDFI1me`Fxq)1H2exG6k0Z
zpNYX?n@8<Zj%T0oWn8*8B+lLle=E&muQWHUu|N*Tq+qdMy4|!nd6Ae~kDn*_YD%|y
z+)y*P9=9zUl%TNvw0430B7UHQmgb5{1W0UT=!<Zowws0rf_|v_$&u<Y-^3OX$q&^}
z<}-A}z9g9v33`CqyDT_+VrG3`(bX&nV6($zHyTXq{W}>bP+L72BUWtkjl1EZMl7hk
zVkW=rQP<#1|A?7SL$BazPR%O7Kt^*l-{!H+_Fg&9MtwW?Qy!5flym9L?@BJs*sSXq
zVxzsQWkZrHW;Xa}u6PWkxbN@zdEst6A_p0lg`d*I?S+h9pS1SQU7XuDO%&PSUj_*#
zfu_A6(rtEr(qk}SjxjV+3Rd7V!iA^oqF9rYo%Fa0GJiT*S9<3JzF>nZm=!O-Hj0nY
z24BI0Sqp4NbE0l9cDq+ssRW&uun8tIqr7X;=Joe(EwUc9)VxVTTmc@2m@XMjPtf{m
zY4UeE6ze0`mT)z)P8Brq!Ss-YI<xv?<cT9OWjbwFuY)VMFO>QWVJj&`NI0YcMgfz&
zuP8Bs9sX3Bw6k<{55ER%K=?lWO;2Y@Ly9Km9Y3{HaQ$Ab@?R(u6SfUZurm<iy&e}|
zwUH|3qg|~}{l|!#l(CAc$P*qMWL*;d68DpJ=MN%6gYkY<YlB06`;c(E_PA0&@CMI1
zT498L;7q-Q<cKa_&q5<)-u-$Q4smeoU%a#$JY~leDKBkOR-RBPpR~On?svs)dsuc9
z;)ot9&>?Q$6a3<23PMRZu@TC!@T08X2VjCExL-*1_vCX?d*C|24u1Z0j_`;^g2-GN
zLIkuwkla``Za&wm7Xq;s`wWq{k0+<3_y_X!vsmXfA$kndj8>>Fa{6f{m~{gdrcmOw
zzlbniv2wyh+aT0_=;9GL85?3R^P*7Pl!cUwu6Wgx*keQhU^vQ9G7ZcVZE5J*1?6{A
z5+~Bg=zAPrG1SuK8DV0PRut>Sgx`Y~>j%kC8?MK9I&7c8AyF~+=6+vUkv}d?_R=Jv
zm`6!1B{NB@=0WP)jGC$9+dB`(?=A`H;IIPQ(hdD}@$3*0QxQ}4<}tQQ8EH+YRzkR)
zDzIf<%qT4?Gp(Ap-{0mtxIpkpXfD<t@skaf1BVXxi$_a|J;vu<w;txo-&neFIxZb5
zOHZ$Iw#_^QA?{IZI26n)FNkAJV4^eRhRhu`Kqm1Wj<aLGF+XU$KnP?Aze&a;fhehT
z5NBBu)Qv9^9)(L1KE@ZB@FsqtH3?Yo;Ha$_j}fAXo!=7sxCf%69LN}_twq`<1lpQN
z5?z)2%NPyebo8beBcsVmGAu&QRD_kebPCHxTRhLxD|6q2;Y1OJ-bpoH5#oI#r!fsE
zdK(*|KR$$+qyPYZexR|F&*CK9yxW|S&uRI!4-8*skf&(9CIESK^&E0B9aFL^ZTzhS
z^T=FT{fK*hhUtwszoGJAj=cIF{R2hRl%T(PYapot`vXO5yz`x2NAN8aaZgpbs@5~q
z;K6M%!Rb0Dtf%$_z#%=-4M&?MZudYYLcPEc{_<Oh)=*h_upA<+i`q|*>nj$yH1k$3
zw&D2A?nq3czo_zmMQ`^)Wt|=AuI?4DLHza2Y3(o2VqV=MF=lh9rO>n(98LBy_**;p
zR`DH7UW*fcLnaPI7v);C;!}9MMHg#g>y9(zT=9|c_oV6%@v`g+dD5>5WVn4<#6N1V
z*DF3#tf*D2)O>e?Kbk!5vG9xdbJky(+%}MQp1b@Js=lK1$8RObYRtv44<GDl*w_u1
zy=3zg)f^<12aTlE_l%3Yea5$!GivRbkUu?QWBVML-o6S?xILRdDxW`&O{9ZcH)G0<
zPY0}NC-~5<2<6`hOy%?R0i97N-^Nwx^7>czYLrrz1?cHbi7#u=t|deQ`l&el2saYs
z+pG7#qMWQiD<qDEiRN|p+}0~s55qeze1C7Ay_1VF;OXTqo~{UX*z{p?bfPOey9u>O
z^4ymKz6o#yUUDL;2s4{z6b5#(x5c_blwQw*9v`&=CQZq6A1;Ww(={#nCf73#;VAJr
zD{{BI*uEibSNNUYsXZ$#t<=@4LRbx?vvMo`Anh7^`Xj9Quo$dRV5})@=M$Ne)HGZf
zlJgnxir*`%>^R!XoNg>j9mSpXMIwvW{B}a)uyXw!R{jF$y6f=aGDGR(I0hxg1ob`b
zHRRu0k&CB?(80b?Di&N!gi$7-jk=))S<tjPsi<@a9e`GleO>O1m!f(d7%X7Xov$VK
z6>ElQFYU%!1{!5fPy2&9BVKgnq5rb%%yJV7v?D^+87k4)$Mvn}Lv0U}%_oE4`vUVq
zc-H0?Y3)VAi!~oxoba(nRF5OY4g0QM3@qPo1ZzKzRzh3#*M-R}iLSz@;#-;rR`*Mj
zx%$QPeyC|N-Y6Xz2jIrld54-eJ!t}k3)(P~BKuS~F3$^ed|%Qy$4p<c$3TU~q$J^V
zJs^Nmkx&>rb;QU8n5!sESaT19@N78cH#O6_iNz2oU&tGKqqb<bL<b;Rfth-rZ)X7#
zn@S1=pa;*2?Rl=*@^9UD?|2UF;&o`4bW!_ur2N$CsJ`CD<H<FRARbqtg6uN7)Qdm}
zdLJmfF190t)%mIQY3}GWc8YGM$J>hp$CnNUxI6QIM!y*66Tu58t0DKS$4dpy8-hbG
z;q|NKY=nVeS<t_!F(77t$(s}Y8v)axp3x`nMXiCLO4QA54o;1&pmz;@_AsCNAp+as
zjAfZFThp2&WX-#7dr&plV6Qlq>vqlsB#wly?0Vn#Dw@yXY{W;g-fKCIqj2kKY@&+K
zA{b62nnRBjXUz8=C^Oc1@D)eP`-@Pd*X;E0NhQ;>#y5lYWvVE#Em0$-H!&RP>FL2(
z@(wHA*kKGO{a}+&uFvvRj`N#QP%1=)1YjJ%;kg0l2aTW@bNTd2u`LR{d7ge9iIqBq
z>&OgRG5<$qu9n4}miy<f;?>WB=APm+y$X7_3KGB%exu2TxGqZ?A;ZP=#H}~pLMIQ&
zA?2Z4WHLI5=}Aiq{TI(F-{$ER1oBhEe=dCJO^gH({R33(T6yV&?*~HxClbyCH2XNx
zwRh8NtY7W>qK{)gE)K=W6|qinx&L&1Hime@Syr?Wj_)R3Tyb6EMq$LoT0^@<p@#Q(
zl#2K7*!(rv`~GX45hX4;#lLe)e|wAFv|O|U*D~$lwP6!#v+Fhnz@6BBldCv}G;gpc
z?gJy<<pg)Vn2pLf;lenKuaNPq#o`dFrc71w$Vc<hU^5w$Nsy-$)&56ra!zqHys^?T
zkP|TRa51oveRYbMNgTMp-c4PfxRohD;+lHGQqz~fCi6dX<GIba(KtQWbnQWOU+mj0
z3+v>0&$7W`iQ{F~@82lWHcj~yGr*f`2eZ<wJ{2djzhA7p6*v~sv}BRcz}U#2v~4Z?
z)R^Z5%|5kq-nIzp3QaxZW8X}>>V0uN7}@T1Zt5~J_50%H_<TNoP&LB*Ai8sI%aW1d
zGZ9{!<hr(hgUy`3$-6yem8W5Ud}{yUJYF)cDW}&AReNy(9oaoiJ_v=eg56-C;8I_y
z>*p_E*#~4XEClgU@@HaDm0PGp1)H)=U+<VHx|LYx^}lCR%41}{^)C^NNc)u6&wdu<
z8{Z#!E56=5NT@7`R|neM5?c!Tf)#;;X$BAJ*wTAt+08B5vxHvQ#`+{{iI%6QN#M(0
zwyD2`DWVCJRRAK+5Nit-8rJ1TYpQgSqJ-o>^Yz$rMt1l5nNu+Q_A&8mPJ$_BHB<f<
z{j``dr#7{!Tz%l|rPd&#O0qP|2N@Xa@cpLNUxS?1ZAO-Ff5u8~jR^9H-hy%qw>*lk
z*mEZYz!!^LK_kSmAFuFo1li?`HmrQ}Jt77yKRM*}-vq(a{aH~DuYZ|M!aDVveP3Jw
z=NrS-sf^YZJl%8B5yfVQoV>Tlo%!*M9H}yz?_js@Jbt7j;(O>fkbPevS$JCx_SpeB
zZ^1cVjqbsW0mq3fNH9{5p|`WerU@t7K=`{gA$$j)+RerOeKGvRe(6lx>3c@p-4o=*
z@oZU{Di+wbWu6{Vf4&!!1Y^6*SbnO8;>4u26DKHlYgnGpLl0E4iP$$O-i-!V#K2=g
zeN<mjqS{c0Us+@=zb406it@vHdgRFVtzO%zTnB=Uh|OOh;-)jfKVFpK?Kmbv-hy9q
z?IbvlTXuF+HdC}}0J1avCF%b-F_>tuD#L=n7ivng?D{1AEIwyBlKzmvd?Pwgzr&i8
zqbHXzRC<TulV4u{+7C8{D-7N8+_;*UdkdVHc3T&@WueH7h$lGBYr-}-Q4kaW0+?Xe
zi+!F-4_l#@qh|qd0VEhMUq1|RQYtGegN*8&8IW1jaAl78#v}`EzJ+Ae;TPT*t;Ex3
zL(KAxP5-zOhnLjleYu%My0jaWSUa^(a;CFb=Fn3!k$8rudlhjXylvrr?Q#9!x-Nli
zv!%N6BpjUKm#5$8)i{~2C{v|!X)i?Hk!nIp2haUBW?lnHjaoo6m*Zf?G*RC&OYo1&
z)D}r<k7|o^Y6u_*w#$r9OS7(e=aNqPeruEYzSv&%5FEi?)zrBC2##St2QA=&=LJ$P
z{G1zK?WU6$x*69$rr0oEft7k5IaOlA{Ps_Qw9zgS7(Kh%eb`gYg!^@JI%Z#`<)Pb!
zhI^ykVBM=AclQ&RGTU`lyd>dMn9oT;HERz#V^aL)BvLf%-uw(p5`JF)6f3!`ej5$E
zIs0H6`h~2Y)fXnA;1=D25|UZk%Q-eY!$d_=V`YtESRiJDVrcDEt<F9+KRk2SPBv@|
zK}6Af|1A-zIE&FUC>r6ml_U7%Nfn|z?AJ9D@S>N6x<zA)+dXXDvihQp*e^b3*C^I-
z32RFCO^*U%JF?^R*Pa!(cROrb<zXvts8sXPgR^)Z)oGcTl`)fZiZby%Gn9f!&*96P
zb3yj(JAMh>10YGu#wn(k3KSz(@fUp+hjZ?~4b{GhSahCd?23~bh90}Z`lO3uZVK>(
z{6sN^0lw0^Y0!9?ei7~!veR2I9uR>_!#In1YB)F<P?{d{YiMrCcrRr%IdA?RTTl`D
z3^fHaPnj96DBiEW9T&)FUTfT_TbM)_dTi{mX;k0#k8^y_n(b(=dKh<PTIdvBX{_KF
zwIIMRxnxh&oho0ymFUbl?m>VHCCYwr#I^YOwF3KIuyC2ehSC}bx3rHJN|4|eVTo6)
zZ4h(v+thup?zP5Y!eFB7wTI=mFlKzq@wy@Q>)q4vq}OojJZ~jN_@6%{GGha2cAv5h
zay_!xs|0Ju-ZcKPY1rO;6m3*EHLaBUb*A@PW93p>X~Vhv{9{PzRRFjBN>ELg11^mQ
z1<~rM%`94>jdVPYuhGrYC9A+Xu90Bn?=5;HZY$tekU1fo2o2xOzas_Bu9Dw+$dOI9
z$Unl*pP}QRq7(u9wLdQmL^ROGnY7;jl6={TfME)qW=5e72X*b;LM^5jZiNW&GHE4j
zchcGyubrvs?t}<`Y4$W1LCrRzO{;%fzVv$K`<7;u2<YCgcHi{>>!^<e&<oM$={>&>
z5nQ7|E=*V2Vz$uagbyU>rx2lX`I3Bm*-FYD;-X$s1UMw_FdeXczvpc@HXvO%{Sa@b
zF8loVVeWLAozrlAt=~zbf=^?u-%s|6x<-PlmL7rW+AoBL#i`#1a~q?IlBjZBlI2%Q
z7du`z*P^dPq*=r-Dyiwukm04nL#D!SBPAeki~&^I^_nru8gKSnmPX36uKJ=-E_)5{
zKW@ZmRpP2^e%68(&Qm_CF<mW&d68;sgyQOZ6Lll2PsOTx6Tk4Eb51dZE3->kOT@5G
z4a8^S07Haqo}1S_uw_2&maS2ZR&?j0mS04YR@7QysLRpecEt*T%OP<Ytt1lpreDvZ
zvR^CKbOC)z52HJLd9)cfNJ8&TheaGYEu8NG;8clhC5%~C6zH6>LM!H^iuB>2UF&e5
zyaM^W<?|AZ>ae6cdcgKM#+Io%AqrU4Tsb44JuJKy6|Z!vYij|5L1&YEn~1?>DvcGG
z^zUusIlTQh38AWIkbA0(+I&uH57m20MfV({u+2ZrYio&yl$I8nmKV~3f4nI}m1Fk(
zk0LBfcRvX^=SG;<`R2J-pbcbmN)G9<#Ed8!;gCYH@}gL(D6lb9#gtTGK-s!}TRM!6
zT&u;@Tr7D8y>`Uh;OExr9b#T#S&+{hTnl|$fw!1pKpGL5LS;VIsN2@}_sYiJr#k<0
z;AKC6?4!+cy7bD?tgAq-a7^_lU1VA0Gd%WIF-*ilH>63gN_8^0C74yVshZ=Nu2;Ka
zzA$U#ABgz1?CF1=?ey7;!B15b?~UV|3sy&AEK+_y)oe&ANZ-cE#;gqL#SC*aFI!H?
zIu5)sU5S&02wEzZpv>IPnfIoVWE_j&CZ_@9Guxrne=S)sQJfWJCR53`HHpdjV9(f_
zPL`z1dT@k*RYw?KU!#U1>{bJBy?q$-URGeuwa1z~dhY9{O%pAfm!yXAb*FH&X@<=y
zdH;MXDbijLuMakniT6n^4wBo6^S7v_Xr@5PI`(rCM^Jm>_aBMu+w5-)N1l2YNmC5N
z9W%!|b>}3@{&P8145CYzSvP$IwpSGkk|f?eL5Vs&?#4^_&1j8sOT^iHYDyz<pH<%m
z&R`sK9^oT@vlMqIh$o-WxfDyz2q!+0I+bjWDB(3fTWWg|w2;tMPmjdqJ?%Q@lo9<o
zBHi33QuOK_w(nYxFS%F?P6ziQ4ypIE^;t|>nnWKtV^Ut8{m3Y7WyR0Xk;Ax>Rp%D)
zxeKlJWuU^JBksxW*uw8PhT`Dez>kUHIHo=S{H4OQABl!syoo@xyRCX>0B8pm$Kn01
zV0MIyHgpu;{dvmbr95TlvY(***JQlTj_s94Q6UOEP{n<_`5_PLd&28JDi6Q4;g!<@
zwQbyiWI;}#2{0NJ{gwIog0lWEE7COQ+-R$82%gPrj%7Z)VSy1I_J-BQ>|%5dYL4X+
z*FlBaRiWfGUOQzkG8=S`yj4|!k?;!r{;gw2&HhC`osZFo%zDY9vA>HPI!^o?6H2!P
zs_#ivDEYE3<QC;}AyKO;S!s>d_KgO)Y*US5ujVJb8fTC24Hh$}rU3*u5A^5b`!SdB
zfzZp#@x^^7VfOZlca%GphsUX6`e3ltFEn9edUaL9V}H169Xq6AD>%Jvg$K(bh+Ulq
zF#~bXcyD-cQ8rf`7`WrfZwRs`b68BV>a^3qD}ZS<QiY!bXA#0>d`BaBb>qG+FSN|<
zwvwc8j>ZsrQ>L&phVEj<X|{z}2K%km`#!g3BZ+tOxYt_Z!LQSc@A2>FcAeuwr)(uO
ze>C@VyzuU8d}ibL#<Mf+tpyjSag;NIv(+m+StA`7&@{iY<GdkY9T~BG1(}Drpc^hr
z*ex8~d@wAlf8*GsO>SdXmN?owG?SCIk`@F}6SQh6Rh`zDo8-nf%wJKsuV~V-vnx|I
zTrB3f^vrBqfjzuyAlNl;T&|qYj(2gYDlS?mvMVcgYTS0p`aZfc<mJcp*w{GnWMi1h
zU^sO-K7Y;wVyEg~&@HkD0QjS`&nx|$YWy$Q9`+?LH`6h@GC%>`K7()B+%sd_CCFTb
zpU8pyzDJvjGM0Pc_|)%5yB6qEsRxcebQ$$RH?A`b)XwE>t{Izq%?p$*;@9i9P}LxF
zSbav<6B(!cy8g1C7!YIR1@AsW%`S!5KD2yA2)Y5yw?B=Vear2roPq*G?pF7$?Coq3
zhqHDFvI*bT#sr-Ai~W%^SEbwy0kO9230Kg@(K`g3r-840cK=#ArnD&UvC0(k=puJ`
zyda^V=0rtnum}NWF&qRQBl}?lKt0m)8dqG_=20(}MbKfJNvN_x-u18BQior1t}@z?
z`<H7JuLm*T+`UPk-B%hps6EBv9K?KAf**dGQ2M~Oq$iR<6Mj8d#C1P7pY)edg66Ke
zYz<AWI``-UxP}_RYqq$Z7R*nlF^O9-<-*ZdL`HJ=90#g&i@o*Ws0YboRRLOM&-i@H
zID-e##A}@rN8aRPMQsy>D0oo6ZR&$xAv-%VGsS4=FK8Ny`1`HYQqk}vQ<`1P#mWV=
z;xvIOJ*WBrA?VD$edE4r`cbcMZ(k0_1eY-l#7e!)6D0l!rcp2EFn%vFnOMn8^=dyF
zG3|-m;2E{h!n|*%8tOTpJ2f%EW85oI;h9=rF%&li@Qc!(@P~R<RMjc@;kZ#d(P|6M
zG)ms4=4HsAo@|?3RBsMg8@2gTCOic)<wBQu%(+7!X~0v<eoby0o=h{&9(@)OgT#^$
ztXJwcl<iM9FG3lD5)$izde4L<Et%W~j5ffMmR!Bu)=zXZs4&C@|C7vOWE)?iW2Ef<
zHTGo?-u#S3P_J%*kNBe9zTIC)Srfek{c_4J7X)Inu%C53sPxM~f^Cuzka#CRXv9V}
zy|hU$EhpFfSFu<|W{u~2W)!!;1q9lgDJuzn<Ms--Uut#jX_(1DhX_q};|<=zvN`Pi
zd148$=kJgeV(1s_ojLWM11FBx;y7Oon_MF87H=rPQ<kZ48bx&RSEfuS?!C*j793z1
zvmWy!u{P59!yU`D6soh$$b|F5j4>`-UvD)d9xZ=-@#R1v_RI%ACiaN&cVl@O3WvoN
zbPMOF<}9~CJ7ybikzl^X){0wvaw7X?3c$i4>Qr#~5h$++8h{N}c(cMUe?ck)H>MnY
z41K329eM3t>PNE1I@7=jB%TPwKG!Q&F`M6?yJ;L(b^+k$y<2W>%b_6d`Ce5-{sw(h
zu;Rnx{W&e&s#@2=8;9`{W!kNgbRsL(l;B-by_aG4Z<&UOq#6Bju(xT6zl4-NRT<uu
zEv6!L6e!-$G7n3ssMev1tt+ZtAkgMD>U3=HR#Q;e)d8Hl@3|q=xK1xH##7$lKp?sO
zYSGbUG&C`8USsZE*3JXzydB@7VPUCgqffTGzX#=Cc=_ZEH~~~)4N)~LsE+x2lp1ik
zZ?Khsm|cuomFpH`ZM60?ews5mDoC99xGU{a^;9)~FjYTit)ka7!eV@QF@X@+73BB3
z5HNRGP^g!nn$A7`k|Ii}eQ&>EH32#R(tlrIW1B`4U)@|H`V$a+$3Rhe9rlsUfExt>
z=z<ig^!jI^pFx{hMf&Z_s3wLK>owEQDHk)#@K4r?fM(9c?7bTGWd*O3#ep=$s=^_t
z-D0zNa{1G+CqAOpaNho^OSf*ei1q06essW>@61ldzZ|QEtCA{LbXE0+BkWaonM<0s
z#Y(Q)RszTtap&V39d5NIxdB2kU8m{#R5(71m4(J^s8hM4`aT+2z%LO)Hwl%jVqgXN
zoZZ0&zf$q}xL}at7A=;2@ng=xj^*mkQy6RLoiTW!0q;t8zqpz4gQ1&RnWtT={DQ9`
zzp=JS*NUUVca&{|jEl}$j%l&7+4&TXiI_&0rNemHC(*s-qtN7CT36wd{0xo#3DX4p
z;A`Cajs3(;P%fqad2<!JW;i)n5GwS|6VBzrv{d)3{l=`cq}gsB`x&Z4q-7^c&tyc}
z!@d4l;7wS@AR=3NG4zj@^@%FTHyeY8iQ+K!NpL}PtbUTqAYC_0L2XRJZvM=xJ?;_#
z5KK9@b5!ufX5{km_>!;WP6TA^*9<aE`5A4TS9c5faO99%V?^aTG<zc&U+bErk=8(;
z?$G_NPvq-%MiQY?&p%Wepgui+vJh{wa93|HEc|pgNFYn@+5YFt%^9`sLX}+J2Iw2)
zjP+(O8Osd@N5x5T$3iv9@}YfG?Ckfx$j8{L+G+>W(E6aU&(5%NSj15E{>W_D->p%s
znlvb|^NmzXUWB&pTRl{9@&9lvm<HYx&<~}^MZKdxo6?k|rlyu&W&wUD23G%Aga%1w
z@jHaK`CR1t_rlJuuls@i@ek;zsHjn~!JypsM;3e*BYa@K@9w7x3PC607+CutBSSO%
ztqf!`mfNnCFGpfw_i(eR8JUov*QBDPq*O@72FIini`-$)qSfpWV=^rz2ME~CZrn~N
zD^z7<Hg4A^3KW43Y`Q$Xmp&KxQnLy8dDpc~p1{4IDBK&y?;o#Mm_7o8WRef##%SWe
zi>L*IR@S>CCi3Ocso_I9Y!R0I@ydCOfCP5crf%-;G63mR?^{i(Ksb(tg+<|b^0#jT
zJX`j8vk*;FfJuq8HzxfzP}DL7GB|W}JRm6IiA*t`TzWW-J(1A`h+0f_AI1*p|8otA
z(3gXQgUP5FfbZR-iFohFUhIyk{_`lSfoYj4fFz6#t^yF<#k+LezH)POJJ9`)rr@E-
z6isUl_R}h_+j*>p*T(Z?f#e||59yKQ($bRB9zh?g4{R+|gK7=H7XEvQ#JytN?`-Y)
zD-7bVlSM7^RYDzi2dw|TA~NQHzrX+Zug?w>z0r?D!AK}_acyB?VRF{L1ADu>WpR9~
z+Cr{ZgQ#rAa)AWf|JY+?tus_!O1E5}+yA_msM+`O=Pw-@;M(Kb*H{!&yQyM;YJBPa
zSBT-ae{QK;bMbF)Yg3asUG`q>&E&QDU&RS}?rEhkzgKwOJ|=Mwph^C<0i`BnywS-p
zd3>?muZfr+c&rQ!@7n%H-71qf8u{qgd9yW^ia8>mfhsa^1Dk9mp)bJbAvIrX1sXq`
zsmR9n#l^+Vd@<XG^77>m<5n-3y;Qc3xT4-CDgV6Blv!Sr^$2P5xZmz~?|<Lr9biO3
z`@c-o@MuEzXaGk`BOsvrU*8=^M?>o>tLk8RQ427pncZI<|Bueu;4>A5j4TB$FG^)2
zFok?zh5tPHW4IAJyf2m{)O~vhHSc;of=cBd!!lQo1Mc4GP}R;5Mu~M+|M^f?%>PEK
z;T!NXK&-!4VG1?5!6WoNUTo@*!ejh@wB)=blsta8)WTb7)ROO<3jDv4r2ZZV1X>5^
z#)tqk%bK7Iz%-u=7_RtyOcG(X*%z13DumuHxa$A@UnR){O2T7Mle&d%%OkF_ByObs
zt8J4nUxZe^Y`y{1U2PtE#tY~UH#3+4{>7D%(D%rchL#pcc>Lq&1_1Kq7#Otxeb@Dh
zpXPtRfD*;bXmeC1!4LJUiT_{K0LK(H+dR`m0<b}T7_MW_|M9k3v=)Os_3t7C1S}u0
zWfCITJN$2J|LZv}cFn2;L9*R42?1ycY2JvZfN8+&GBlQev8;v(xowSB+I)J=G#}Wj
z|5Y*wP_i|EKmZzfq6d|H`2NSEfq^APx?<?kP9J8HCj(mB-v_wZ{-YNFnYGyHl)sCE
zE}q*#&&8$nzs}>aJ1SqKk_wzA(Q%s2Vf^m@&Mq2YiMXcKKr+PbSddn=nRIhY%lb;Y
z-%rqghDhHT6!EC}ryqb(u)!DV?T{oTCI6KZ1CWzlzxo*f4kK(hU83LE=#BaB_yw)Z
z*V*)j^|8Ux@R^=5GBPf;`|15(7dT#NC)4@)0Y|1vr%YGK7hdvThhqId?Y(6~RBabF
zEZyDR-AFe>cS%TxNJ)1$NJ}HlAR&#?J=73VDkU8P(j(o^=DP0tz26`3e0=7EUvT#9
zv({SYTE}s$b3b|;K&@~?=JlSt#>anmIGk@I06dXc4Zbp?JO8^og%cPU2poH*|2=M{
zlVr9^Uk;e)X~3pEOWgc#)Z^JaO28i#gsZc-Og8(YG5)_-q`t$U?lvR{;4rB9_XPj2
zl=WTjg9Ehg<WRfcPUKpx^#uRdxHM01k^9!M*_1%BU;Xp&Z(-5Yp<7Fz^;puUO**{T
zT-JJOjRAA5{05VAuMMf*pw<|)C+P4U@mZ;oEBJZ4y$%N!lah)`L4zz*>qncr&EVi*
zrOC6Q{Eh>xCypJkDub1IfHDu4aTK5}@|pPgTVx0r-3&^}#hL|T{S^X^Ka~^v|6OYX
zx_R{OXIBWA*vi*m%W(s+$OZb^J$HZBK7a2I3W|ASW8?MpRtli5_Up?3Z0A2l{=WC=
z>dEbJf$wLvA)s!atmg;s{lBg2@8?(SbMy<e*U0dDV`j1O*$2Hrr1Tpa8rn$Ll9B}!
zYon&~05L^CNa^!`MogJ<eR~VG8i)Zz_wv8Etcd+@hxi8umI8#M{&0`^_j`Nx|NMa;
z6T1s|!_!5(qK{FlfSjlM)-cw84#K+M_BCv4^9O$2haC}HKz{IlD`3*;Q~OyBG4zm<
zh=@t)YZ`D`c|}fD)zt^HEkf9KI;~5D{r|o(?F}*ecWa0M45#TP2R*uT5zUQ4<$oVU
z1o(+NK+mPoeM@m>w;;MODG958VBp`w|NJ;r)qX(MflHp+^ZptJx(D2j@*83Z1Y-Rt
z0q9*-oG!rs_d(6mOJ8ggNQdGhzwFOeGcYpBi)jOv8Z(mPeOh`t3~&R$==Nqx-H(0=
z{yVy@^|nvImVch5&GmKKEy8~dJ5c~2%{g2s8F>23slMdh;ukPZkdaZ2SMl@mBV$NX
zTb%*tQ?A|PHzG8Ng2ng-*fNa#2_qxQU1yWxmD(i{q!|h=|9D6yV0c^p^?n%>(ogpU
z&k~rxMvksFtRJh(`{!vmxqY6V0RaK=SzV1bP-|#1Rh#cgp(7B#(51Sc+V@}(bN1!%
z+3<SqsKX`-DFrkE>s0-LBbr)V1&~TSStVll4U>n$U@#!KD*5Mtk|Ey_*4ggr2>DA%
zGHccLZKlkR_o)M4W3ViOz(yibaMaz=c}|ww8G_>hgOWvOh$>0!eaPu=8THa`l9rYh
zH}LXs==OqQ{rDKp&`=l<X*BrRM<c!yNx%YT)9z%aWcP(-U_bj_oJ%xgTfF(UA_;$8
z%*Y<kUCkPkR%&YM!ia-;=RR&@;5s{*FU1Ale0Oz{5L{-F`r(76Jvf;o&;QdZ5Npv2
z2oR!RQ}zLTk@chBOq+zqe0`B4fFsg+Cr#sNzHk=uMZ5xd#kU`9fbnYj1eunG{Q1sA
zx5=!L8^zOeOJ2LwBIliF%&K3`shYu_w8^9jX5`|~uZv*T$Scv?#o>Fipxl;<y3Iz?
zb~g*SkT;ghbg(oz0zb~VXRzAr>ME(AHKNqxo{fO(D*mkFLa};#Jga_(glaZ>tcNXo
zUwJN;akG8&l=L0^7BmT(HIb)43l!*qZIVr5oZsIae|fT&vgo<99(%k3yUlYg_F3rw
zq{<~!JrZ@h7*%J!m@a;))T<C$wSN)|xo#AHJ6XUGMH3o8ABg+Bj|oK$_F$E{raxTl
z>?(prOdIi4^<Eg%sbanQJWuGo)C8?A=qwpvSzAAHSz9mA>XkjaJb{#3sby3fla#T~
zffmP_XPRDf-1EKV)XW#5gMSG?O&HBJPrpu`xfUiS4}T+j2^_eW8THy|wqu5h?PEff
za}7wjy%%t+*LxUUvlmWYv^_VF3j1F>&aY}`qu%X*a2EVkfo>lBAjoOnrThvsIwuK?
z700*?56ZT>S`CAs6zG(x>Sxqx!`W8p%MR@eRnAnR<@P)Zt?r+btp@c{{OVmJa_zc<
z+ZS+59Gi72epeR%#sUv;O0v6?1w%4ioTTk6Y0sW4;6rH?u&Q+vo4-7Kg@uEjR(ye8
zZ*W!f4)3#EL&CH|s3TVOvTYI{nAKrYVbZO)RvVKtoWF?FF(JzuP(FwQ3Gz2@C<#LO
z=4?uc%mi~iIAB7|?<o?8*Z0_zkDnz$`G%_~YH%Fk_?Aio-gTy50en&CvT|E#{1xt3
z_Y>1iJlme+5|WLXkLBfC8qNhgnGh!B>nx<D2R_dSt3>bRh@P*MFo&Wz`<vtWO6^uW
zM`8P@sor~Xj-$h8K|c9Uz#xuMpt#HM(DWieZ)VE==O?l;gSt56+NG_?lMsaz*h}7k
zGL-TF7ij(kN~8VPB2^mBWiQm1E3KA^wFCi4MRD9~PZ2TBYsDfJ)Z{{rQ764m)GmZi
zZj_9H^nj)N6b;lKjQreNk&ZP5-z~Ta*#9Ig8#4f1ymj5|jV!lJo`(~mlESH<e$y6h
z7VGqef4S7PQaxrjj^Z@AD^74jxPe&!jT&0t=VQXFIO!?h*}AH*h@PoEP);m$^NAvv
zTKw%TtubI4t|B5CAF)$<d+#=4k1HMZP=2$|%ar<1&G>Fwb1qXNqVyuRDK1AJ#EB15
zxQV-)G2P@;9-Pt!;CD0B4a}U-@4UN&hCbya<<|EkeAjm>&DC%|KahR_XdZjWu%gA8
zVuuX-51#OC4VP$k2UKd8kvj@FCQV8w^%3rx8FdI#Qd=2r5EY=#pZc}M4C#IIz&e&s
zIdwW^tV>qzBL~|}+b+;ZgsX5b!8A@vD)sB>fQ)oP#|4Gu^?N*Pb@=3jPg;O!QN|Kg
z+;eyKvo|8`x!%VB_9)qJCt-+EZ_;pAQn;HWHO>_+dQ1OFUphSh)q4}Sgf=)5^sfAY
z^mY6!)YfUIZISdwO@=rrE&QPBNb8M8Y0?Ll$GG7mv0`PIuKi*M)?B>>54;R;_?OX$
z{Y}q${LkKYU9F;wSX2_S3kDFZrQ~4Utymcmcwo6eM%~cckx<^+^HYvyi{KRuuql_8
zaV$kK$Ez}G{S~dVI&`fYh`yi5*9f_8dh*No>JSBH8V4^`6m2au+HB?kzM&n*RP^xV
zB;{B#1T>>-s{_clc|Dd0my0;Z2CjRzrAxarNN-YyL|SJYFrKF&uH<ZsFUL`D%x@_4
z3`!<CJE*`C1FxO^t}TWorX;VocV$@(OQlq^Yr|Ytq==w4?reaIv5KMu_ypUq?cI}}
zE9$XVHHo*bT`A_C`3-JsWNjbL&QZG+=uZGh`Ky2i8%!1yG9Nt=gOmn}Q%2|?B27k?
zeJXzh>QqM50oQ&eMm!&-Nz2>=gRZ;T6jrlF^xS6qIHr~s(BbGWeqhd!i2auxPYw?g
zh3b34(FMuD&e_K~djTJUAV}yV_#-I|YufNH%bie>Y`<xvkNxyMu5yLZN^ju}KgI;9
zE;ysFx(u+<!i`@Q3|2B#&+Hv(>3lX|=z5qlv0i7DYI#13aG_hhp7WYshYNU_jz8s(
zT3k!kim;6uSkO6wx7P832nr#>>qH>TvbAmDsC9;pFAb$-ibF*W6B>T0S{v8Hf``<U
z@UGU2t9qoOQ)m-x*z9^G{HyduUxJr}L@NMK6h-$12s|wlrQc#>;6M%6g5>B^5EoF2
ztoFVL8o8fKptZm82s=cafS<~xzeb;>WI?j+!V%TyAr8=KT_^Xan8c&ZVH)`DhneLp
z8ISY0L_kzED|^u5YPr&Z)gP|f$U>JI*avc&TJ5%9ITn0Xvq<V1GDGpTE4g3%y~!h6
z-=?J+QNVG@JkknKsNH^Iqt*M2WJ}F>`i-$yeyJ;hVB=3HnVzZ7#zmcUanNt_y?ZM|
zN+}Qdxqda}ZTwT=p78K+>pnv2H)Z~@IJ|G$#q%F{`l9dSpm8XmJ^#jOhJe@)DRhmy
z)FFDWggVo(p5`h^0#>W0J#l(D_B0NbgSoBm$FS@#HH9-68|Oc(kx)wbX={=&0ig;!
z0GNWs4&TdqUeScgXOcHhx5<&H!&X!D9}P!E&0`H01e;sNjkftj*ZGmG3A@~1KtWUD
zX^0ktqKhLYKydq#c!e3B$L7zZOxM*~2&-Nhoq6yU)%6V($-_Djl6y$C{e9o5?elob
zK*(lDMb%63URoH@n(HP$$!fka=05C{0Q`1^^YR;g$`jCL;?RyTd{^PXApODdn4CI|
zb&n_~?Dg^8%>D|GPM4S9{q>J}5%2vb%rW4}B5j?_p`mvJIMT&#sbKp@7ejTARn)2C
zgwuG{7>*k}Lt@wEG&|aJfkr$00OKyu&!|E}k<@H?*S{~jyrc@l9I8{me*%lZ>cE6w
z-jHnTX^!HF0<X*oNCTVe>ZavT)O%!B*<bIVke+-YNPaYo8u4TaKrgr0YwXj@;RK#W
z*t+lT{EG`epjG+2vGg~$@qB}u399)M7$-ifHp!(ZL?Ky1cOnzm;+jaU-=M?8V7sX`
z?G96Y^s%0S<>nsH`&y>(mSiOIILTU&GC{yjmZVxL&}B<it&j!9I$u+{Yu)iLQ{lZC
zHj?jTk0IX#&}v!WDD1DkV`=D9_!p<PD0ysT=}w7~8z8fzQ(~<k;53W9bvQQwdEy@;
zAvAQ8fPj)aePgx1rN&R&IJ(Cd29&UDemw0Q3>>^;L_LA9g;_JUuGlX;%Dt-8Md%K=
z0$E#wYh-{M0q-+Z@{#*wVW!l6riQHQr%PB7lxO%$Y2)^?$et=bXiCG&SG!ZSu_3{d
zW#mn&uc3D$<M~1yzToo!&c)Lr1T|xO=NUSbmWgSkJJS%TcH3~U#Dbz7n<;E&Hkp86
zxhGc~Fv<DI*dczjgi1<E0xrMcXUThGbbLD3_es>?c~W!a=~H^+=)O$7Z$l9N%i}`%
zA=QsS4Sjl9&Tl$1&y9I)bL=r1fp0bTU54>ho)HZ`OPiR8&O&FD?z9qI;W(_C^tuA$
zU2a#5qDC$dp*wFbQ9u78$|Vbhu!~IfT&;Ged9ZqZD`6d!2a;NJGa~Lr|FjNz?91ec
zGTL+4eu`TTV3998Wq2D7An|j7<qzMbehdrxj_{>d>p=N5Rm(#pfI8StyubX0qEhFF
zA~qwC{j&;x$8nCoxy9#r?Uuo6N=nuHC-TNbt55WloALqQ>V!Zoe9pV=1M5~e&a!@h
zq7-9n07`QB{D$8>c|HREhN7mvh=xy=1iyI(h-U_%w$Rj`(U<J1&`&X>Au@>C5hprD
zuogrb&e(imP&KGZ(2}~ruh1+^6ST$07bRRK%4gvXZQbgu)UC$rA>$*d3txV?6Elcq
z!O!H*qo$kRAkV-Qe=;%J&r8&Y6W-V>!p#2A>4w8>L7+~bazF8xz}HIav42q-?if00
z)BI%XhjMnh=et1_K;;BVv;W9BSfa*?Q{ETBho{-I3wum?06@)H=X*0|>S3HF7r?P3
zZo?3f?z9MkkyYu}C%ddYyr=+lA5Pn&^dILfY2-s9H}UN1w&`S<k)tt~XbAFOv=;j<
z+@;Ynu5C~%AGAfhzD8-Tco8REo0$^G<zALPmsW8~eN!uLq|xFOxz+eiOF-&G8ciWS
zbI`6RIHt0a^^YEKi$nZUHf_V#?EVnYWWP)*H0xyfExy*M<BGq$?bd1=?48r#$Z?pQ
zA8kWApS41dQuSe|(D2%sbAdb}wWXTRV=;hpjRecqi^JnnDq*F<3SC-p0{yPa`6ro)
zgnU+>Ay=?4TDLdkO8IbQVAITQx+&Q3XM@9g6#|w7OC2B%+uhii$ZIfZhbQ7N0ByBO
zL(aV)33<&bXvfugZr229;V@d1Z2C59NtN?E`7jF^?@d820~)B^CT{c=Uk?YVIHi2B
z)2W244}(zDn=MlGQM4dQ@Vvg@y|K{{Hu*Ch30*sQB<DC%+g4&BE(?W458KAzJD(e@
z2EYOg94(y(1qKv5W}@5A)p5mqGmn597q+~%*7{~1jHnqd+`xVg7uOx=hTe^37q8zI
z5s;~cS?+fX;Wg1pG;Stnxtom75>FrbfguS7&_eV0p?OMt@$1et<>zxTOg3`L@XHAJ
ze2ca;@5q}!-M@l-XHwg4ea@M}FY4|AVqO>sD3~&t?E0`jzLJezN{EM77BW*aYTBjd
zPCwuMB9<KfHjLhDB!c9rLFSvMI-Et&YT3C+LnZY@l<6p>c#vFV-7+ub2ct#bCN*QA
z!95{(eTe7qM|N{cuZ$#Bq0sl*tOl(!p*a<2gE%K~H9ZTgT089bFguS6#PU>?zyu%|
zQp2|Ufj7}J{;oI?y^=akDDQ6Q(!HQY%oX4{k5W8z_oPtvtM>G5_DO^NgH9XraJ>Ho
zkp6B@^D`lHBkQ(X00(prSFScN1w8Vh9QTlHkPogMeKs45<2(oW-V#LW@JN9wM&5^*
zJrq(%Hwl@8JiV8MTuHzjh@u>7_1xZaykac5T3Of0>4uS-KZT%H3E4-#b@d)1g$8dk
zxIp+4hgQu*qShkiwCQ6O>S8p`E-M!rh&nFv;5+3=)`{hN=mt5@zpV(jFO`Fi`$yLf
z9Nv_)#+ckk)F+rDXe#l5@77gV&Fc7j9(GJj+dULo_Mg-ES|udp<T1W$PcJf^W}yIq
z$W2TUZV@KWX?Bp(y847!HfBh#w1_R26<|}QRkK%dOyozT*%>+041JOD*%+s+dKT&$
z`Ymc;)y*$$+OsTd{3B&82Ab$RD!%`wi3IxK`}h$(<IAqNW%iL~yJhqdv$h>)boudQ
zniAureIp=?lwYq1qK^&du)925Oc3+l2m74O^N#xCEE@&IiJmGiqd_V#`{)&$#694;
zhq+~<_b;^TRH=e4^+FepaKa+jnEs-*Zg@zb*&GR%Y#dPOG<hL4(e(<<d(ucGL*x4r
zb}xS>Na8i5PM<NB?1G?Pn5un*UwHJtqUE-}O9;NZ{45G&w?~fGUGf4ik7c`KF06ih
zB2!$dg?OxulQwxFR0-He;{SG@zTBcVY7it$evC6BQU-AzV()w)eFuBoW!0^cAa6Pd
zgI$`eZwC?`6WTRQ6Fl=tSV1Wv7^x&7k@BlpbbqfyTX`RSPN+>0(s{?vUJPj@O@nPY
zX7s|85P{$KYft5hE=*5<NiO7&9B{N6eRQ1eCQ<m}N(SG&XP(ddH}g-2OZNEG1*gRu
zg8OzRL>D+TH}N3yhv6v&g}j)iTIAIeK1>eZSjbD><a`$2j~z(!+5lK^&x62>ZscQp
z_-ob%2KQF6O=1}HTvYlQqtZUHyBQf@0_913X*O)5{z1D$(H4_N$~4jOIi&M+yE>NS
zG;JasQqAb^T*Dds^02#(KW4^R8%6D#30`*)jO)`6)Y>dPj@0=Qfi?E3E_tndqCX8m
zr{PMa4_wPK4ajEdG`+)m<#5ZaSigW6Ym>cB<WX>`fzPn-Uuh;Xu_E$^v}Ar=9WGaG
z+l+I;6MaJLDNJ?tz<M;TjGs0A!wB6vT*f<Rl2ZAPy7p%dzTZ}fqx2zhZkX?(*>ZLz
z@|no2#^wA0nB)nzOOcIZJk)2tMKZBGou#ja##k-7nUN>VkUHgd<$2(2Kd}5IvmVWq
zCm(esMeP;@x@{l2mo<3~TDT>x>xi86Eed112H#^<E&dz?vTbip7yt)<$aWrZ^(Vox
znY?%_^e%O)i@QUmJyx1A^->tg%J=j~lme^aS32}8?u6D*%D5W}1U{&(xTD~^@ctpy
z1lRrYdP(f;Fh(;jGEt40;Qg%Yw^J|a;}Ff^`Qi{)oomWoxSCyi*I=mi<!a-{Zn!gE
z;tw^9Xf|2YJG2mfSS;47&}?(th{)A@4mH9!46R{KQ)@a6_A2=Ad{ni<l(Cavl_>q(
znItf)7}07&WVV8}ST^15iEs1hAPaHjZl(}bH`*%+blmf!DU8p8Zj8BLYmEsF-+729
zmyhBjd8mnU;Q8wue6hA{mi6I<^!MU)uJ;xaaLnYh;a;+yeuaFXlN(a@_h?`NW+DX;
zJUK5BDYKlxzYd9~BBtt*aR+^=R5ovuMm~BHM`aq*SG;hX<#XK8T{yxV!eMw@EFbO2
zYSxb2vnL6U4dw2SwkBAO4WxwaGLcV6!V#8KOq4>?%S=!ujd%O?xT4mG;Acj8fYUvt
z)GU@XT{a%ODGpi;)&p(+R%{^*IPnT1$#fFx4^^5V7)U##<&gL$>f7lR2Y%t;Dj;<q
z%!r;@8~?p|X=BCpu5@;Zrc`DFT)FazO7Z>={M8oh6kEST$S>wpIB1)76fd3$$=f=k
z`@LVIsa&!bS6MH&1la*YJ<yLkcO;Ve^i4%_#^he6iFR#>lYql_^<;Y{y3_$&V&a%9
zac|IPReio0Z(&M=`c$nMtgvE&pfOYbZ;T>Ty-{@IbHxD%zS>|fm8OUFmqGVSA{n1&
z(1$qYAa|0;Hjl?{=yUj^(o6$My|As%7HvzyNEYhC*!?1w1mTO`?wr^=hit3|ot9jx
zhE>sk6rE;r4;x*oTpvMt==BVgRksd3M*J{(1;(h*6umq=7M4eP67ZsS^@h|oo<!-a
z<p>shvQhb?QZzoRK1s+d(E2jQl8^eu9ojBxymM5qrDOLCpjh`kt>V3e#vY&lJNVXO
z`FMAGGB+%0I5g6BugDy+7#pIxj2)M9z#;MQ0Fh9&ezg$$tV0akndG`EcCr&xl1=6>
z*m6wpuh8NB7O!0?O|I`Fd9<p5EX3MYr^(MDqsEA$SF0g<EBvdG(p?QxofV1-C`D<)
zbC<ZM_`rJPOfX|Ex4~)f;EiuABF+5!m9WSseNIUa;_76JC?|XPW%|)@#Z`#F2}9*l
z9Ra&x9lC{u#o~l2h#qBQ9U^DHSdLXS@AQ61JaTW5nNLvwO%klKhsKu$T<|_d5h-bw
zLZhNt%}VIwvufUhY@I*UTe=JEPh;tIUk<P_7Q#2n|0R%UWtoDhfO&YST_D@8QBU9D
zTl2taD(9@l#mOu>MLokLwpDWqyWBLJhl^jo*DEV;zqr1>UT+p&Dz#pDo6{~znJg>3
zp$LmKY50Ee9k$1m2y`3U7j!I9Z%1d<>7Ya>VUKd0Snb~<5+AwbHc*n%-Z@NKH|TyN
zf}Fm#4k-iy@w4hl=@R4LS0xoKI}V_!^rMHm_~W9>Setc4_f%g?__uyE3Ip>%Gwfz^
zRCIKF&j-LscXf?qf$HVs@GIxqu!d`Xfl(!0J~GTS3khUJROY#1tHeZ3DX2GDM8*qU
z=rD%DWpEn6yN0e}w1|T@RFboh3Kd<ySopIdr*|b=&(2^NsU^?m?Y5ya&nP7?if2$d
z>^UzS$0^p|9ACSQc-;#Oz>VV+GAgNuzdo1zwniRBd=ebKIiwaJPOvS2@f@z?wZ4<x
zdOI=Gpl?#MK%ohFg^y&yIL0ah^!Ly>1X2m9#D!6UGs2BKgWd~A7;>@PMPRo^-I2Wd
z7h?uB5$@yjI+i3TcGN2AKsOZ0E7rQj-Uv}AY^$M3+?omj;9@2))tOPqez|8<eG`hl
z{kTjS;Q^$K29{3V8z{ZL<EBe}0LmrUP4+9q&Bk5WOToK;KK3#M#jy+#9DvgQwD+Rn
ztc_ov06wk7)PH*_52Z{kYjB+{VclaJKCVmYbd&qzwo?}a{syHV(%05#%i~Mlzia?4
zP?;CmJXQ(UcczOWfXRK{wT^W&curw;Fdo5<&wi0ONXi<ohpHDJ!&iQ~vEF8M616Gh
z#Pd~^DH2fOfDXjs%$lkJYn2GB6;J46MCIzaccS+KvLSj*26-Lh40sn`wzt&9W_Cl5
zRLluY)vzZ1-v2MyE!EooX>`h{?I?<rXew`z+7nAurNYLL@0SCPap)>Zx+G2?!)X<1
z-w?-pdGk|AkIAcbjoym36S-aUQrIuy-i@Raln+~&5fq~ODFj!IVfk*=nii#~RCWL^
z;Y*e}aC)pRas}<DEBSI;{IPr(&zE0s#}P|<UTJ)t-a5_(F7+L;!!|d^v;rX)&V;P0
zpn4g+3<jWlH~UuP+WZxgIsY=aZa|b1n_HG|Z_bkyS>5!-6a&k~wl9(<g!9rb&E{BY
zviYy?U?0X=YZN#L2&}(8n4n@}K0_bqQJ#ST_RBR$j)H)m;gdwZza`6GCeCvM6y0~S
zUSjx>niltAz~(Ae4i5rEGEs*on`6&PNb&t~l-t~c(!#tq+>HnVkk%4&6Y3SeMGrIr
ze?kLarRGWB6w4^ccwDJ#)d`?$4Bl3n<a`w+HH+nuzq4Qg|H+Qz^G@QKOFj7>Wj>^u
zIW&@oLoK20$wfy8<EJ-4?*>ID^p`Ilke)Q?zn9>%eoGo!>Z_D!uuF=Y;J}?XT_vYP
z$e~X?G-e<R^U&AnQItC`?9lbfJ1r0H0v-2pTw~`4Z^G7K<JxCQb=Pj|jOXUt(L}L4
z1CcGY3HNz}tHJoez76{$Oy8E*BVCy}L7g<Q*2gWDV%$jKGdO&qCalIGr$hkqBe_7_
zcWkBxxbxxy$c~hlq&(CTyIvMb>XPN!GzF>zaca6BoPV2KHVftF*Znw`Xri_5Qsid%
ziL}9KSgHoK2@ykM#b()0);qr#wi)nMaML8wqynGRa=u*h=KAg>UiIuHo9n&fvn{}(
zjKpuu4Vrf(_Lr}lu<I3abz)`z=@Q?%8iE3pRG$l}+b`a&c9Ci1waOp8t+#^poVkIo
zqflII8h7#BusE&?`%lsh!S8+>1EvmsOus4yL?|v0QcJ)r`Ytxmjem$mxIv7O+9`s^
z=64}()8lnu({-hINa61&KJL4XAPL!X&A*uH{rsv^umz^uqI-i8LHr1q+hr$x03DWj
z3sP+lRSa4kN`sgPUP5b3u|8BwAz>8&#yq1^{6>NJ*u4F3ruJ7W4t29;SrJXTOFu;M
zIwXwVBqWmu*k$%wiTx;;)uNNE<57t6!u;t)v}%@O!HzSpOI+Sw?=iVw-r%yN#|N})
z?_31Ori1tOit3;urs~G`gT)BG&F|Y1noE2)eD0Cw6eYr(v?)V!55f#XsiH;2L76z8
zN~x_UcU}k_Enoqt4M>Sx?3;NFv^dJBqb1+s*9lqn3tfT+$C%-UtZ|}+x^&CSY0&hF
z1P9`NqLN(&IHK0!iz3^Ucp#S4j??wqI-{9!?n5N3p2Hm)*B@r-R<ao_Im<AQDBgk@
zshb?c61#>Z`u5!7JrIjqe~+=*w%BY)j_4GmyHDcv9iT5F2%`8ZbX{sXq3CzjeT7Fx
z_6$Lg$L<4a*3%st;tN1^1xW&lQ1Zv*ws=lmk2`7S8-^kJlVZ0Sbo&bf;UHl*bk%u>
z@c!S`8wIW*Ltj=PMpOY<=8T(N_XL)`Eby+ub5ylbx7ghz0aj^OIyDkLWYi_2Zm{<E
zA={B}YbD{YiV@q8hSKZlo(QuL=fgx2J(iPN)-d}8QV@*P9>4jh(HeLEAj+%;c)Scs
zr+YIDg7eQo1p^RNIDM~Sas={#crUZ!xG|{dd`C7U!^Go+Dh75>ZJ%aXbw2aQXK*L+
z#Arb`slrKnmk}}#1Y}^kOw&DT!koW6C^>>RHz-v>Cw}zIB8P(qMLtlk`6M^Aw#3Xw
zccXx7P6KKYw0SwZ2B8m8<_p0hX+IvUS|p91BrspLBK3b9EcRS5Q}<X)F?JRQ2PEa9
zqN09?WCq+k?ddvNea|I%RjS3u-W}Ag2!CTDm)Bdw5(?XKYfY1pLo|lGNM=w*Xo3B`
zqY|L4ikW2YoTB2g)CjbX6T2h@r0`TrK`YFA($KhwPEx-ZxA9eHkvlWF`4i;Qs|+MJ
zR0);GE%rGivw@gmdAjp1Q2P?|4Sz3VV#`|0cm0Fn1D2IRRR8CY<4vfW@Ik9sR}%lM
zqs`xa8zY6O7EeF?e1<x!jR1?LlY2>YLiVJNq}Bc?+oi@gQ=&l<@3`X)YGQOVtJd=#
z2i8OG=?zfU!hQRk{U|#lTu~B-NaUEUe$Twz&Z`^9{Jwp1k)KjHgd@jdo&SK`hBT~0
znepP}^x%ygI8A=-RKIg3CTjK7l$^EZ`sYFvo!*sB0MOD>DUsHl!pLau@<y%XN~M)Y
zyIGOyemXQ?WzQX}?QBK_v+yWvKPeY1|9i}DtGCzkaniEw#v*K>hxVS|1Aol0HjY2j
zar4X~*6IfG(haj)VUr6@kLs$#d!9OHr$_BufpB0TX$Mw898O?Tzt+|c<YAA=qr=zg
zBt7cntw`xbN+e(4R~*srPHt#D;WHe5Bb#&fOUL^8Qo+WPzCer~>GpS9CFu5#9ji_c
zK6BkNXAB()0-bi>Czb?BVkhiyVLlsALSyeo9@VVsME2%=n{5z^$n8)cA20FOIck~G
z`8Du4JJ@G@o^RBFgE9p=z_m$@WHS$=1Nug$lR%i(IhPsm8?us{g`x%-eUH$)wZNxR
zhuG%wQ*7fp9lBWe=Vw!GArBq-=CXj(qh)4Bwg`S8LeFN6E>Y{UJn>E4j1p<PE`UIL
zF+uHBSlZ<vv-83(!tpEU2kG`C=_~otBbK$9GjqJz{kh>jq&hbVF95K@)plz-Fpt2p
zRc)oWu^vclPKTD$Do!pvc{#Ytuv@AF)tU~sPU9Ko8o=#5a-)Onaz&ZL$A}#&9Jt0>
z$M53tWscJ>VO1_y7QENr+}wQgJ2h%hE%_bW`EW{6l(9EOsaZw+M$kR^a)e9&5DT{F
zGIlYn`lLj@9li2t*o3t6k~W`p?=N~ntX&_{!V&hMiN#Y)TINBqmjcDi3b3UTwlV8%
z%9pGJj&2zgMV%wev^jTZq$1I(udE~P-lG5V8JM_2M+y~g+|;J*$=iJ2vUUk6I>6~u
z!(7Hsp$`3~@Y@*cbV7LT8T&Z#`MJg+V{J_zgGm)1p%7nsH-h9Q>hn>a1n2bAwC>?b
zKR>?J3i4ANe&&9OX_De~)d{J{O<{!$y9!zA%RPF<Zzjpc8pg%;pTF2Y4e?m_F|$1R
zwb7&uDM`j`R?K*=dtEawv)ONnHgo}8)YHJGCkDS*iLi&&uY1cjv}5{pno_P2q*Pi4
zv?{<MTyAqm2u-{}h*rz)BZ*YBxd~UUk?_Npd&{L1{VPeQO4ZEkrjF(?B|idV${mWp
zSJ0I*>asN5^dQb^+(DSz^+5Nw<1qSQ22tc@WzbO0jtz8TPq-Xe7tg)HP=HfBC4tlQ
zaNT?DHBETmVu{6E2V;^ZKq=h)alD{_`ML+mS?VN-jk|M0^>Ozr+z*|0&6hz}4_BLa
z63?0He@-EQQ8Z|JcP8f8vKO|P*S|qN@-QhkbCcJ@`l^PoO5k0@s%Q29Y*2&GH4`Yn
zq5o3GMIN+H+m&kCKNkh`$>U<`ds6<?-EC0iCnA~1P1pR$-jlMTW_CUuQbq5f9Kc`4
z4<S09K|77_($EyewxPrZJFSh{-)yv!N`(?pU5OYRNyjByJjJ8ySKI7-`v}P58F&KF
z>r{e1`mxbbv>~+&#RHtsp?)qaa5canRbIz4Odm1P08MhRVs9O_76;W5vzoRruNKof
zLM~DH)+3y@soG*$(u3iblNC#$4RNfmA;?1<vx-xEp6dz+UC9Ogz)1j`)Vt^W33;J*
zRT!m)i#9lN*YsrJh7}3|ok$VaUnxZzuE}2HHm~pRvm*e7XG(Z#@1@xZB3q6ur2|bf
z-lagC^0s4z+MOevC_vm8rx&uEq_j@$K~%UgzfLAA#eB<+)~2@i(iv}TwkDyUNp0&Z
zwVYM!0p)T8P4N_`aM#m;TcYo36jPv9UuH#Ed>0bVliIcIIp(+X<%DJymjRB=@;&pO
zH2GDLzdS3@Z*m7XJ@TRP&gG)c3IoC&edR6;oJI}WA!i)GOi4zYe~ZT0yIkDe#>`u<
zP`Nbfx}s$6q?XU7LEcK5EZVqI!Du+2RuPF}N+?c1(OPe>&yT2s@X=D+7}P1m)>&*!
zXob*K?8Dm?8Hu2n4xJ)t`km}rU0e2Xa*7gjFoS%VmG*JH5q!eu`5@yot`|9B!oAdC
zK1B}^hQ0eqCs3XNzgViur~SZfiG$74jR6mHa$w4kk9)RvcVhvZ71+4jhqO6JvP<PB
z+mD82VuVjM#MV4`E#h*pp>+rQ3kZqd!@*-t$1qOvn_bcynZ(-z$l9d>V=90kS|W;Z
zd0+vX&C{IzaUE#%TM4PTNP%ol^jv!^n0B3wi(ndcq51SQf+^aaeloO0OGl7UdHw{?
zTFa0W%B8RfWdTW9vUTn`XbO$eL5s6?y8+kvM**5ADn1%6ac^j~#d&{UnN!2xYb3uI
z^GIJKiLU;wq5LlD5;UV0Y5W!FW1QjI11?qXp#p5kO5}PClD1MY9sHfYYd%mS<l~lX
zqdI&(yua=_2WiVR!kG{i$&dbqX;1V-XvE6Qo`k*7>U}wn`Q3FO-vhx{`%w@3wUS6m
zH5mZ+MhkSzCDwhGrzFbzWe<Ba%;^+G9LLTd*gUc_YFec6iJk0^mZ#DxoGjh($CLci
z8hFhmi&8WWB#({0=``Q>`L})edR~0Nfn6}rZ>9r2>D1HaU-naeL#_WZjnoH)BgK3u
z;%FwPl?l^R82P(<B%h5OuGj@-!kBkn34jry2PpY-X+W?pW7-o$)*W~O8X_0E-fInS
zkPa|bukH7A`0K^Fx~!|TIW==dpLH15e?Jl4u;PneoeYI!-p8sT-e43$)8V<a(0_H&
zv#q1QANY<?G3pv+z1U+fa%YZj1}aFlfTSVJ7&UNEj*p&*Ael2*43wg*#RTp-&w#Gy
z47+uj?9tQ<`4z5-o)eS^k5Q4Gk=eD{cqgHk3~xJ+hYl}FjG8#eeW2WZOi_jYh1D+)
z#fE51*=jM*w+V5zVBcPIF5m;19Qua8(@Y{Lt5M@H#09Fva3Jg5^dLd`P!AEjMm8Pw
ztr1fcTIQ07`o5tP!M5uDSBv(^iyVV(?ql;Dip`Nw>0+Qo@vC-pQUUwo>m9>O{AXBu
z7AOVMd_uJvLJ_TM2&yN$Y>>uBaP@FgpW-F<Yt@}**i0VlJWcFU3SW>k_P~<xZ3Vv=
z&!zO5(OrTj>`i=nm^%%DsN@{FqyDC7ReE5q1)z}JC~ClyOWH!8q<jt+!pWPRwX3~N
z61|OZctY|0WEsyvE;fQUM6T$AIi~)DYy0jpLuWRypxiF6Ef+n;OFu8osgb2Y@$xbh
zG}*c!UB5_WS#uGLF2+>sE3s2l&8Sif$^gyR;bVdnB`RCYs90D;W)Pqt>4ANH*9zc`
z!66UGmGhst6u)_d*~z4B3iqd}Y%@Wa1YX0an0x!Sb0Vz?1$Wj(1tr;tVd(lJp1b~0
zAD&R@Oy05{(FCZe3l8-P{=F|UA8g)EH%+mCG$^dH8g`2A3%`4>K_NVb?f?)yo(s;$
ztKA7e@e!;HOS3->U{Uv6;&L3D*Z%NZ<BgpokS@@RrfOZ_Yab^}&o5Wp5aEiVJ{ov>
z>9FKz?ApnIX%<2jZe7M+ynE7(M?+J;^Nd!l2<LDB(AY^90T=RR#zOY<7Z@Vc5;*m=
z_`X2Tk!f%lv3+~A_6Opig8LzFyoNZm)YlDY9<^c^vd0R=?`4zvV}>jlxfs&_wCW=C
z=0@}i%A)1;G9lKkJyB*>&Gmg!NgwRPelDS5EE^BIk2P2g3^u7buioNbE=6vrOtt0I
zMI+{{3bO*pkP1@yBxwniZGWZZ(zYF`AxfYit@v~n@C9Z`p}+hwx{&GhVQrbsqRXY+
zomPm<20asR&^4;4Wj+A?L3@__at7L^RZO?5kz~|cM@~v(LHY~u2*a3CZMvM5UAli5
zpSJm!eYUcL4QpOxxn+@SfHir}U%8KXiBYHYyrj}@G_Vn{UmyiEg0q{Cfj6cEWVj(4
zzy;|o^pe%Al)Z;Wl7{25kgm~}zzK95=aO$ux2+^^Sr#uz4+uCDo6-4_2xCY(bpW{=
zX5U5Kc@(C8Sf7K}U9bTA;!n9d;UxwM3Da95z<{G)vfKwdXKNN{_kaYw9!jJsvE>K2
zuE}puSH9LW0J5-=)2^2^%V0ZVf8TY5#`sMeztKi==l1v)F9>f`<Yix>A5{!-5O*=L
zgAZ|v1`HpDi#?Kl+0TB-9GS8ReJI<}2B=Q9N<i1%L~-8{xH@w#WbMjdm0Uek-7kDl
zv=WZBmaBF(HWQHwv*WfI;|f9Bsp*XDKrs|TCO(e~W<MXi;?JFCup2+KYmZ(g4Bm?6
zUWyQ__mM3do8@!4_j@i*{s-|g-@n%7Pw;hjQn6OiTc;Uz09b2^$ygpM4PGP@2#)gJ
zie9w&7u*0B0F*Gc_6zs*x`Xw}XMmn^`msvHtF$SLMkK+4g<GMW{9?LEL9NE95%gYS
zk;)WAZx`_}+8uahsWjzL8k3}5o`w}de#T;OC=EB&AU}kHT2y?EzQJSE)@Xsc8@`<*
zoEh)<Yxo@eB&ov4-lLeHf6NL)pOG{Bdlkr$VhT!$8`tTU<q&XLMjU_VtWb}1;|x$h
zqGA9QC6O1KA*W}L_XOVL0t(T;G-g6`xqt;>mn&7$ks0pS6j6Tci9va_epAJYOR)C6
za@%{`w0opIApOG`eIpcsp{?=_%b^PPEZ1I#{V2tO#|td;Cq$yEpF>m94~+7Kl-)pr
z91tzJIU|+uLXFR4jbyG54IAYU?tj>`^+VL<B?g5P-k{SRg3{%A;ALS?Soq<T@A(#8
zS;14fdphJz6A3{R`|EG;Q??C*>cMN2cCTD7$=P33ah5xzA~XPZW9&ib(8iRgr*?<e
z>sPf)q`}OS(etn~w+ayjhlgGbzLDT~*<)6B)SJxNlDH-*70knJ6?b$f@7TJB#EKd|
ztix!R+?|9Q?`ngzrU}slNf^pLjx%i8*Wb))(MrUwUmvYgs+ge4NmdQOqW4m3qp2f@
zs8QsBKIID0WBS^kedO+?9<?~vC}hJ)Q)J3e1<Qujoj-16Nj!2ZGKuTJsAxo0m|G9q
z)3&_hyRfoH#80K%ajHkjIw=z%*O__RrN8f3^&rSAQa6M;Iy&Gvv$+p|_oS8xxi@;>
z?b`cDPQ+oTFi~D_HO+~#KqI0CA;SUpug#u@p{tkKPQHwS|AK3-%d2ReG78pm+IX4%
z@XgtFKPC()`-nsqZe@d2ALdw1O5l;PDc=LMo>2%sV7Sq)Unes(3cne_eN}6W1XR``
zz6eS-vr>iiFJ)g*`y)Le)qg;Vccty!q*A2krAIr~nV)s_|Hydk=VFN;C_$Sc9w)Xt
z66=<DZ=ycY-<>k2En(C8gb3IOoZtJca-$xZyl&H$rX5}?m)OtTtnha{=cLOxkK3c^
zU<})%TP&SY2XU`Qf7bGjBM3Y7um~$!-Ew9)C0K3Hf^qeDK!XBNwrPKbes>h4{ip5^
z$P2p@WA}Za8-2z24<61%>X<^Cof7lvR_%6U=lkhy6l^Ln&f5;PQ~;!S*Y~3h@%nBn
z+adr!-3>Q5+#~nU2w{Y}NMOfq8Ls#LVEkb338211v=E(^8Du`py(DT>>Q4zt$dc5*
z_D5QaI%qWFT%?Sc3P=V`+;Yvdxxd6D6GNPtnejW{OU(uNYH`ai>Tq}5krb22qk;a6
zrTB5(I6XIH1XL}3kJAb@*g>=%F@Z6wdRreT2){B8Bt;dPeEN9YG%5KvD`$5;)0%Ae
zy^<TaEyjj8yASc(sYDpJG<KLq=^Owa#*4wGNd~wYH)hYrlkQm#$t~$MaOd2a%u7K|
zGARzaxV1l5v^y{8!8{YXwH{s71E9h<6(dl$Ph}>EqUt(w<;{mWxWp<D60I6)Kdi~e
z`Sopzl;43<mOsBF`EC8M&@S_N`KsC_(p)l-{&o=Uj|dZ^hiDt}{kM-VNri$KtLAGG
z&}S(&n_X7;tUn7AA7Mhi*koIgs9W=F(xG*B8j*yc91NUT2Qxk}e*w^!n#1Ieli7%C
z?7bo)xQ2(~2^*d(uwQo;09#?K5+GcZs~2*1gpX%Q0Mmv+T@VQAN2*+FfCHfD=y}9P
zGftcQHOr)M*$*Te6Q0*;YeqzAh}k0)jOWUy3XGrMG%CSV&&}cuT?R4_^H~iVS)`l%
zltcoi8kDxMivSquzWuKzvbR&Y0R0ScMotz8fBR7Zpfz8cZIaVQw_&|(1&%am2EPnw
zwQ1XQc#U_f{yIWoNp6EQhx4|n^XmYn7Jbz?FEtVe@oEW4BQq_sX6L{uGziD*ww!0|
zmw+OKzHIu7EroiKl$;k)=EE|;V8Cv$*l;ma%i>nZ^QkWZF7vF$jbi3c=Yg2Ud*L^o
z;Ell*UO5+Q4c?SVe}d{86Ec^@QW<rp*{wo244(q|Uf|AMuI8IjqfNlXRx4}~N=1Eq
zY)y`|C@uro$w(A=fU}7KbRohUGpLH7XW{_sd~Kca9i&{SlR^|OI%C}2Tqn{1@$i#j
zL)^iFQ<3sTc7!)M1cw%N{lAq3cKi<IuQt+k>dll0i@kvSefXrA)L`yx^$;oY_X@y%
zNnnYp?(r>rZY20!mp`7x12v%%D|SrEWi{-^b`%IkaF)KqD#FZ(c=V(KpqMD-?I(k;
zj5WPi;`zq&VP|_P*Xsw=_P35oKu33+%<7+@uWz|{1I}rz(=LI)0V9>ApyHX;P=~i0
zjtMLVulpurlk*a}Er)@kLLPGEjU!3Z<DqYmxEHn}&-w71JFcAlHZ!+~mA9Dr2l4f>
z-$3J90bt_;ZdSu;EO!0c=+<ApeE>gt$hYV4>l?K~;Ms5?bN&L8fXB50n0nU@dL*x^
zO{p?D=0O($Tr=kdDX<M!IoR3u5NSgnQmH}RK#o2I-<%b}@brlVbP5WU#*qia^T3Wr
ziXcvduZ<A#6R*LsOb++ka|b^5_Iu`MLNC|0NJ&LF6&s3HPft${zb(7epG!95rT`)#
z?>wMmDRQN81>b(NpMHbSkGTjz3PFWtu?wA2AG2q0=SlY(StJS;WJe=U{t893C=+;P
zjJjp3-u@m-K#JX(ej7(D_`Tw`cvYI#XWeQ#B!>y@PIsrD4!@AS^O_ZEhikQ)UGz3Q
z9Y35ns6OtLrbt8E+_&8Krs)`2SjH{YL7PlS(sz^<QHUru$A_ZVx7DuG<wpmVuUqk|
z7Je!*{bida7l_nIH;51%o&V{$j1Pd#OOCt&5U5;};|ehVSwsRzUPz&#p>(Hj%2I)p
zRzRu^K-<4MStBW3a0bdN0Kt0*;F~ALU{Qd9NCzNf^8&ogurww$Fu+sXTWDmsy*xl6
zAt8y1iP;2dLSHU`K)N5EvbfP=Ll#>?LW0+RnzrlixU&+FzgNxX<_c1UL?AuOrUMj0
zoB}a#{7Bv!fY*yT(q~rnsSn`MTDm1dBFA7OIRLi{4}1jhiWLv$fE%?35g~^GCZMOS
z1klfQKI<|y+0VdRgxoNdYZX1$rQ#p9UuXis^+(w{+-I!b|2}U4(Z^?&+J8Q&<AA;*
zU{C-1k>>^^JOB5;0XhTbKRi9)eEj>_=Kp^W`0pH~%DD$rNFMqY8a%0CjGkV!#I{;|
zx$VWC58iM-!CE`mf11KlrT$)Ir^a~aPjMJXzI5SFviqfAj(nYMsf`nBeE0=D&1#RA
z&}+C!=}Z+z{}4bBB<=C6RdFs$N$fP^?jm|u0k9SH6mrjGS>S#UkBQb@@t)EKt}K>W
zk-t9!Xxhcsq)`w6v&x?$;X2RwvhdinsVe-_qs$Y3o*yeb?c-lHlk>_=lLb42ckH{V
zJ~y^IQuM}VXFpF2f17^q5CYdEC7%(#H9!RF-^%xS|1fwr%OX(;i-=!Ht0<mBl1|-T
zAxNSqwbfiGNVtl3ODf=2RuJQ|Lq0KF7LYuVZ~5TP{SDC~di|Lp7^Fora_~8>^k++j
zu$z{h+Pa>x)pv~Gup*b4Ql27C+yF8>6D<v`?V<0?T#ebF_4k1@K~ij!;jI;P=NWge
zRJQ(H=`kzhT8{B+xpV&sH-5|ZeSVU!`N-gE<;T=ZCf$r`V4KQghySxfd8p(+d~|$~
zkNGrv>9$B7GYB2aeUem{Kc}|%H2qEskB5@7qpiI!#EJ3e8(rGR%2@pO<#=A~5I<xG
z)gP6aaxBwEn4d&?zNFwUOR0s`Qwz_&xMve+lN#RaQ+*e2Q}&(Twwa4-CJSt<ESb_|
za+blbHlf*Gs9IQiqQImzeDx~+tQFsb4N-+0#Pho`Bd8ZuC!G!U#>?4zcf^6*;^p{X
z!rw*1FDsPF|NbLuf%VTp8&pF%SpNl2&$}Ek^7H3MYGm0@$av~?W(m!gPIz7>W}l%?
z1yCtFoOg?kL|qiL%`z{W!)A4H0qD=`I7WhV(20YxbExDusFrV-Gp(gADxaR1fj=)k
zf^~s6dDcw}r%frp-khX{$j1`Yazb$AR6>fgJ~A6(K3i|rZYm(kz@}H~RIx27*&p+R
zR}9*8r9ZBlzolni;9KfcU0~z5>v)<-zN+QGEBX7`K%}nx=cQvS(7W><2}#AQuiBqJ
za}>fy(&|Q{GW#mhdsItGB*$wO5&Ka#M@L!WPMV;cAS>hbRcj}wdaQY2NJ^t8`?^3c
z32m(!a8SM@!?`)C3-%(n#HcfA&F$K@)hix+;loE~=NuRLeN$238{9Dm(wcB`yfSsv
z9yc8~oI{nYUG};mg3hc#RUea%xv=PYkZnH(Jv7aF^G5_ckx2-X>*jN8v4N(tCxgXT
z#@UX=BZGAVU3BH*JpgznfJXe!rXJ9Xrlw166>qcLn)#s|0<IGP<$_WoMz2dx-Bx4I
zo+?~$+mec#$F2fn;A)9<F$DB5jJSTNvl-iKCCG@|iXq>T#v^ZqNtWJdz?Is>eVIF{
z-{xz6A+Z9I?Yj-CeH1pcbAqd&SLpb>z}Ix<$DVY;cHsHtQCgwhB-<vQ|5Ykzj}3Qz
zT=^T!T|o+KOgA)S4m$Gz+@(vuGC#FRi8Z!jS>&}A`)qtgDgQCLU0%4d+j>#dTdPbK
z?_kAYACQWG?s;Ev9ogk^9Vj#MB2|dJFBkC34lu?xr?x6dmxp+Zn?dm}%oQBrs%>-Q
z?HdX4)q{I#71Cw<I0wUY#D@Q!`lb%4AkEd)fbi5`{(tPvgg&Du3Pcs6{UHq43@v^5
z5s?<+colikZ5?d!sZHygo|TA?<+RD!U>9sE_uEvDGKID!!KP{+x+nS-*KU|s)nr^P
z{-8!nZDDUxA>Tx`q~cR;manloL;dqm7K#2d6y=m@y;JGs&h>;O7sA8ZO9gQA!dGl|
zZ(O)~;*6p-6lk+7<IT@dK)%<x;jcwNbfMLYTY=o~0{C2%>&CyA=of@>1!5rWvSNq>
zv@3{AM?YIB!mk#wEoy7pmb=@>9rpi_W>vSDt4{N@9E@`a8s~pa^Ofr)1WjG{C{rQZ
zx*#QZ@{1_OduYWL<2STjer6c*EalLGgY`aACL(zToZJx3&)xcAMtTxXCA4Vfn>aqp
ze}L$H|1rIB+1PCGb4{nb<VyudjZFQHQCSs;=kyB#NMB(h`x|hu&#PO?@7(!(e}D6u
zR%?)mA<J?o*W1JU2IX7ehJUN^^OMGaj4=_s;w75{_T){Fh-B#*|Egh`_aZ48QjCv_
zgDFaxX(9J8DSeS5l<vJ#`P7zPPL+)AIG+NUNhVL|*Q`v?6zXlD_Pr()_T2<|csILc
zx-@g=PI$=22KGz|Or@ZEcI1G4!Q-`)-}?+H@eP|j{YZbRo-)l)OWKIU9XPe+vu>OK
zdCfVbHf$M;Y4gcq?_=tm57wV%y(vA!(`BTW7eO5Kef$$Y7>yQgKik@po|-^Fs18ux
z1YxPmA4p+~bNFlu71rrrg04wp=ojFJt>s0Zj_7e5DrA+lKJpgI?!7JyeEXvz26j_n
zvG?bLb!UKvrb+x}(?|VXX5bxRoBw}vk}BdSKyK(#UF-XFMH;uk_?@zA!<A~!R%S2m
z523`I^^xQFX9@s#pyqBK-^Yg?rdQh1%Z!K#?=_)G;#EPY^EPK(m52K~?NVN{?aEkr
ze)ektwx^21*rpZ}h)j+|=d{`h1kE9i77H9y=pikEB?imqhJECIQ!uRttDz_>p`Pkb
zJ|0!ijD2=0LJ|Ksm8ahBz)=U%81RdZCqeDc1k5$X)-;(5hzvrwxr1mfhe4H9-z6qf
zS3jvHg+}@k#XX9fon`0#=;|2$d$62uo-(n)6#VF6pX^$(bN~9LJ(?c(p)$JR(s^~`
z;nr@RhUmQ9$;|Qa?%|`*fFMV>yR7h)wPv=l)353S4Y-=ilM6dc-{PR2@8>CR;YSjW
zUDb!<2+w0)QqFySSgrZkZuq*gt7=iO{Oh;NnuP&@9j|)nU&X#R#}EA+eOqV&FT^kM
zihWl`%!R+p?TtDGh19jL8a;>v<8nyX?}3yq4|UT+W@3~67-z|;Jr!iir0wVk-2S-X
zG7wFnqjdiKhkAC>q4}8lT-VPLTDRK0Q8Ny6$qhAhsQI<j=5xG7{<S?RQ@S_gq`KJ_
z2d?l#F-n8pQiM6B&GdM`og{BvbZYO9!bPdIJ*>b0IohDC;-7VAF?Z9)ox*{A)WF-l
jcmYBEzyEKEz=Uvs2kxo#`vElICtj#1Xv)`tEF=C8K7RG4

literal 0
HcmV?d00001


From 81bc968c14322e12e6ba57ee61264c6e319f40d9 Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Wed, 9 Mar 2022 14:59:01 +0800
Subject: [PATCH 20/28] =?UTF-8?q?=E6=9B=B4=E6=96=B0=E5=AE=9E=E9=AA=8C?=
 =?UTF-8?q?=E7=BB=93=E6=9E=9C=E8=AE=B0=E5=BD=95?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 ...36\351\252\214\347\273\223\346\236\234.md" | 98 +++++++++++++++++++
 1 file changed, 98 insertions(+)
 create mode 100644 "graphsage/doc/\345\256\236\351\252\214\347\273\223\346\236\234.md"

diff --git "a/graphsage/doc/\345\256\236\351\252\214\347\273\223\346\236\234.md" "b/graphsage/doc/\345\256\236\351\252\214\347\273\223\346\236\234.md"
new file mode 100644
index 00000000..6d12a507
--- /dev/null
+++ "b/graphsage/doc/\345\256\236\351\252\214\347\273\223\346\236\234.md"
@@ -0,0 +1,98 @@
+### 有监督
+
+```bash
+python -m graphsage.supervised_train \
+--train_prefix ./ppi/ppi \
+--model graphsage_maxpool \
+--learning_rate 0.001 \
+--epochs 200 \
+--batch_size 1024  \
+--validate_iter 10 \
+--random_context False \
+--sigmoid \
+--gpu 0
+
+Full validation stats: loss= 0.35714 f1_micro= 0.71096 f1_macro= 0.63525 time= 0.30733
+```
+
+
+
+```bash
+python -m graphsage.supervised_train \
+--train_prefix ./ppi/ppi \
+--model gcn \
+--print_every 50 \
+--epochs 200 \
+--batch_size 1024 \
+--learning_rate 0.001 \
+--validate_iter 10 \
+--random_context False \
+--sigmoid \
+--gpu 0
+
+
+Full validation stats: loss= 0.49029 f1_micro= 0.52777 f1_macro= 0.34071 time= 0.25776
+```
+
+
+
+```bash
+python -m graphsage.supervised_train \
+--train_prefix ./ppi/ppi \
+--model graphsage_seq \
+--learning_rate 0.001 \
+--epochs 200 \
+--batch_size 1024 \
+--validate_iter 10 \
+--random_context False \
+--sigmoid \
+--gpu 0
+
+Full validation stats: loss= 0.34045 f1_micro= 0.72752 f1_macro= 0.65601 time= 0.41571
+```
+
+
+
+```bash
+python -m graphsage.supervised_train \
+--train_prefix ./ppi/ppi \
+--model graphsage_mean \
+--learning_rate 0.001 \
+--epochs 200 \
+--batch_size 1024 \
+--validate_iter 10 \
+--random_context False \
+--sigmoid \
+--gpu 0
+
+Full validation stats: loss= 0.42566 f1_micro= 0.60157 f1_macro= 0.47732 time= 0.30706
+```
+
+
+
+### 无监督
+
+
+
+```bash
+python -m graphsage.unsupervised_train \
+--train_prefix ./ppi/ppi \
+--model graphsage_mean \
+--model_size big \
+--print_every 50 \
+--epoch 50 \
+--batch_size 1024 \
+--dropout 0.1 \
+--learning_rate 0.0001 \
+--validate_iter 10 \
+--random_context False
+--gpu 0
+
+
+python eval_scripts/ppi_eval.py  ./ppi ./unsup-ppi/graphsage_mean_big_0.000100 test
+
+F1-micro 0.761944872
+```
+
+
+

From 3bc371ccba0cab1a18b5b3927f57a9d9a5173c21 Mon Sep 17 00:00:00 2001
From: yxy <1016545097>
Date: Thu, 10 Mar 2022 15:02:24 +0800
Subject: [PATCH 21/28] =?UTF-8?q?=E6=B7=BB=E5=8A=A0=E6=B3=A8=E9=87=8A?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/inits.py      |  9 +++++++++
 graphsage/prediction.py | 20 +++++++++++++++++---
 2 files changed, 26 insertions(+), 3 deletions(-)

diff --git a/graphsage/inits.py b/graphsage/inits.py
index c3351494..a38880d5 100644
--- a/graphsage/inits.py
+++ b/graphsage/inits.py
@@ -8,23 +8,32 @@
 
 def uniform(shape, scale=0.05, name=None):
     """Uniform init."""
+    #根据shpae参数生成一个均匀分布值为minval和maxval之间的矩阵
     initial = tf.random_uniform(shape, minval=-scale, maxval=scale, dtype=tf.float32)
+    #创建一个tf的节点变量
     return tf.Variable(initial, name=name)
 
 
 def glorot(shape, name=None):
     """Glorot & Bengio (AISTATS 2010) init."""
+    #获取矩阵元素的平方根
     init_range = np.sqrt(6.0/(shape[0]+shape[1]))
+    #根据shpae参数生成一个均匀分布值为minval和maxval之间的矩阵，值为上面算出来的平方根
     initial = tf.random_uniform(shape, minval=-init_range, maxval=init_range, dtype=tf.float32)
+     #创建一个tf的节点变量
     return tf.Variable(initial, name=name)
 
 
 def zeros(shape, name=None):
     """All zeros."""
+    #生成一个全是0的矩阵
     initial = tf.zeros(shape, dtype=tf.float32)
+    #创建一个tf的节点变量
     return tf.Variable(initial, name=name)
 
 def ones(shape, name=None):
     """All ones."""
+    #生成一个全是1的矩阵
     initial = tf.ones(shape, dtype=tf.float32)
+    #创建一个tf的节点变量
     return tf.Variable(initial, name=name)
diff --git a/graphsage/prediction.py b/graphsage/prediction.py
index 8223bc4e..a64be482 100644
--- a/graphsage/prediction.py
+++ b/graphsage/prediction.py
@@ -45,20 +45,25 @@ def __init__(self, input_dim1, input_dim2, placeholders, dropout=False, act=tf.n
 
         # output a likelihood term
         self.output_dim = 1
+        #生成命名空间
         with tf.variable_scope(self.name + '_vars'):
             # bilinear form
             if bilinear_weights:
                 # self.vars['weights'] = glorot([input_dim1, input_dim2],
                 #                              name='pred_weights')
+                #获取已存在的变量，如果不存在就创建一个，pred_weights为变量名，shape为维度，initializer为初始化的方式
                 self.vars['weights'] = tf.get_variable(
                     'pred_weights',
                     shape=(input_dim1, input_dim2),
                     dtype=tf.float32,
+                    #该函数返回一个用于初始化权重的初始化程序“Xavier”。这个初始化器是用来保持每一层的阶梯大小都差不多相同。
                     initializer=tf.contrib.layers.xavier_initializer())
 
+            #如果bias=False就不生成矩阵，初始化时定义的默认值为False，如果为true就生成为0的矩阵
             if self.bias:
                 self.vars['bias'] = zeros([self.output_dim], name='bias')
 
+        #根据loss_fn传入的参数，把参数同名的方法赋值给loss_fn
         if loss_fn == 'xent':
             self.loss_fn = self._xent_loss
         elif loss_fn == 'skipgram':
@@ -82,10 +87,13 @@ def affinity(self, inputs1, inputs2):
         """
         # shape: [batch_size, input_dim1]
         if self.bilinear_weights:
+            #inputs2矩阵和权重矩阵相乘，权重矩阵做了参数转置处理
             prod = tf.matmul(inputs2, tf.transpose(self.vars['weights']))
             self.prod = prod
+            #向量prod * inputs1形状张量得到一个一维矩阵
             result = tf.reduce_sum(inputs1 * prod, axis=1)
         else:
+            #向量inputs1 * inputs2形状张量得到一个一维矩阵
             result = tf.reduce_sum(inputs1 * inputs2, axis=1)
         return result
 
@@ -102,11 +110,14 @@ def neg_cost(self, inputs1, neg_samples, hard_neg_samples=None):
 
         """
         if self.bilinear_weights:
+            #inputs1矩阵和权重相乘
             inputs1 = tf.matmul(inputs1, self.vars['weights'])
+        #inputs1矩阵和neg_samples矩阵相乘，neg_samples参数做了转置处理
         neg_aff = tf.matmul(inputs1, tf.transpose(neg_samples))
 
         return neg_aff
 
+    #调用loss_fn函数
     def loss(self, inputs1, inputs2, neg_samples):
         """ negative sampling loss.
         Args:
@@ -138,9 +149,7 @@ def _xent_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
         neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
 
 
-        """
-
-        """
+        #计算损失，logits和labels必须有相同的类型和大小
         true_xent = tf.nn.sigmoid_cross_entropy_with_logits(
             labels=tf.ones_like(aff), logits=aff)
 
@@ -158,15 +167,20 @@ def _skipgram_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
         aff = self.affinity(inputs1, inputs2)
         neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
         neg_cost = tf.log(tf.reduce_sum(tf.exp(neg_aff), axis=1))
+        #off矩阵 减去 neg_cost矩阵，再求和得出损失
         loss = tf.reduce_sum(aff - neg_cost)
         return loss
 
     def _hinge_loss(self, inputs1, inputs2, neg_samples, hard_neg_samples=None):
         aff = self.affinity(inputs1, inputs2)
         neg_aff = self.neg_cost(inputs1, neg_samples, hard_neg_samples)
+        #tf.nn.relu：将输入小于0的值赋值为0，输入大于0的值不变
+        #tf.subtract：返回的数据类型与neg_aff相同，且第一个参数减去第二个参数的操作是元素级别的
+        #tf.expand_dims：用于给函数增加维度
         diff = tf.nn.relu(tf.subtract(
             neg_aff, tf.expand_dims(aff, 1) - self.margin), name='diff')
         loss = tf.reduce_sum(diff)
+        #得到neg_aff矩阵的shape
         self.neg_shape = tf.shape(neg_aff)
         return loss
 

From d2ad9f0459e424abf0539acc6b90f2c3b3c35c52 Mon Sep 17 00:00:00 2001
From: kashmir <wangqy94@163.com>
Date: Thu, 10 Mar 2022 19:02:29 +0800
Subject: [PATCH 22/28] =?UTF-8?q?lstm=E5=87=BD=E6=95=B0=E6=B3=A8=E9=87=8A?=
 =?UTF-8?q?=E5=AE=8C=E5=96=84?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/aggregators.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/graphsage/aggregators.py b/graphsage/aggregators.py
index 184393a5..fdf62f63 100644
--- a/graphsage/aggregators.py
+++ b/graphsage/aggregators.py
@@ -441,6 +441,11 @@ def _call(self, inputs):
 
         # 进行lstm聚合rnn_outputs为结果值，rnn_states为最后一个单元的状态，这里用不上
         # 聚合后rnn_outputs shape为[batch_size,max_len,hidden_dim]
+        # lstm每个cell的输入为上个cell的状态c和上个cell的结果h与这一层的输入x相连的向量v
+        # lstm通过3个门控制每个cell的输出,3个门实际上就是3个sigmoid函数，每个门通过各自的权重矩阵控制输出的结果.
+        # 第一个门是遗忘门g1用来控制传入v中那些信息会被保留，第二个门g2输入门用来控制v哪些信息会被输入到下一个状态，第三个门g3用来控制v哪些信息会被输出
+        # 每一层的c计算方式为 (c * g1) contact g2 ,每一层的输出h计算方式为 tanh((c * g1) contact g2) * g3
+        # 传递至最后一个cell的输出即为整个输出的结果
         with tf.variable_scope(self.name) as scope:
             try:
                 rnn_outputs, rnn_states = tf.nn.dynamic_rnn(
@@ -464,7 +469,7 @@ def _call(self, inputs):
         index = tf.range(0, batch_size) * max_len + (length - 1)
         # 将rnn_outputs shape变为[-1,hidden_dim]，-1代表自适应降维，应该是batch_size*max_len
         flat = tf.reshape(rnn_outputs, [-1, out_size])
-        # 根据索引将对应元素从flat取出来shape变为[index.length,hidden_dim]
+        # 根据索引将对应元素从flat取出来shape变为[index.length,hidden_dim]， index.length = batch_size
         neigh_h = tf.gather(flat, index)
 
         from_neighs = tf.matmul(neigh_h, self.vars['neigh_weights'])

From 197eb3df9391cd9a0843a8f5843a3b588451c810 Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Fri, 11 Mar 2022 15:15:57 +0800
Subject: [PATCH 23/28] =?UTF-8?q?=E5=A2=9E=E5=8A=A0=E8=B4=9F=E6=A0=B7?=
 =?UTF-8?q?=E6=9C=AC=E9=87=87=E6=A0=B7=E7=9A=84=E4=B8=80=E4=B8=AA=E7=96=91?=
 =?UTF-8?q?=E9=97=AE?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 graphsage/models.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/graphsage/models.py b/graphsage/models.py
index 73b547a3..746b6c0a 100644
--- a/graphsage/models.py
+++ b/graphsage/models.py
@@ -463,8 +463,11 @@ def _build(self):
 
         --------
         在本实验中，就是利用这个函数，利用每个节点的度数形成概率分布，从节点集合中获取一批节点id，在后续视作负样本
-        true_classes个参数传入的是labels，但经测试，采样的结果和这个参数是无关的样子
+        true_classes个参数传入的是labels，但经测试，采样的结果和这个参数是无关的样子，而且实际是有可能会采样到负样本的
         返回的结果neg_samples里面是一个列表，每一个元素代表的是节点id
+        
+        参考 https://github.com/williamleif/GraphSAGE/issues/76， 这个里面作者说了是可能会采样到正样本
+        只是他们假设，当整个图数据集远大于邻域计算图时，采样到正样本的概率很小。
         """
 
         self.neg_samples, _, _ = (tf.nn.fixed_unigram_candidate_sampler(

From 29cffe796b6d1ef043ca19b7d40dfe10f808bffb Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Mon, 14 Mar 2022 16:44:00 +0800
Subject: [PATCH 24/28] =?UTF-8?q?=E6=9B=B4=E6=96=B0=20dgl=E7=BC=96?=
 =?UTF-8?q?=E8=AF=91=E6=B5=81=E7=A8=8B`?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 ...26\350\257\221\346\265\201\347\250\213.md" | 38 +++++++++++++++++++
 1 file changed, 38 insertions(+)
 create mode 100644 "graphsage/doc/dgl-\345\274\200\345\217\221\347\216\257\345\242\203-\347\274\226\350\257\221\346\265\201\347\250\213.md"

diff --git "a/graphsage/doc/dgl-\345\274\200\345\217\221\347\216\257\345\242\203-\347\274\226\350\257\221\346\265\201\347\250\213.md" "b/graphsage/doc/dgl-\345\274\200\345\217\221\347\216\257\345\242\203-\347\274\226\350\257\221\346\265\201\347\250\213.md"
new file mode 100644
index 00000000..214dffe9
--- /dev/null
+++ "b/graphsage/doc/dgl-\345\274\200\345\217\221\347\216\257\345\242\203-\347\274\226\350\257\221\346\265\201\347\250\213.md"
@@ -0,0 +1,38 @@
+
+
+
+
+### 镜像名：opeceipeno/dgl:devel-gpu
+
+
+
+#### 启动镜像
+
+```bash
+docker run --gpus all -ti opeceipeno/dgl:devel-gpu bash
+```
+
+#### 编译安装dgl
+
+```bash
+# 查看环境列表
+conda env list 
+# 激活pytorch的编译环境
+conda activate pytorch-ci
+
+# 源码在/workspace/dgl 编译
+cd /workspace/dgl/build
+cmake -DUSE_CUDA=ON -DBUILD_TORCH=ON ..
+make -j4
+
+# 安装pip包
+cd ../python
+python setup.py install
+
+# 测试
+python -c "import dgl; print(dgl.__version__);import torch; print(torch.cuda.is_available())"
+
+# 一个官方的示例脚本
+cd /workspace && python dgl_introduction-gpu.py
+```
+

From 570e0979a8368350130cc57aff755950f63fee6e Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Tue, 15 Mar 2022 16:41:06 +0800
Subject: [PATCH 25/28] =?UTF-8?q?=E6=95=B4=E7=90=86=E7=9B=AE=E5=BD=95?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 ...72\345\231\250\344\277\241\346\201\257.md" |  55 +++
 graphsage/doc/preprocess.ipynb                | 446 ++++++++++++++++++
 2 files changed, 501 insertions(+)
 create mode 100644 "graphsage/doc/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
 create mode 100644 graphsage/doc/preprocess.ipynb

diff --git "a/graphsage/doc/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md" "b/graphsage/doc/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
new file mode 100644
index 00000000..2a330537
--- /dev/null
+++ "b/graphsage/doc/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
@@ -0,0 +1,55 @@
+## 阿里云机器信息
+
+IP：敏感信息不放在网上
+
+mag240原数据目录：`/mnt/ogb-dataset/mag240m/data/raw`
+
+```
+├── RELEASE_v1.txt
+├── mapping    //空文件夹
+├── meta.pt
+├── processed
+│   ├── author___affiliated_with___institution
+│   │   └── edge_index.npy     //作者和机构的边，shape=[2,num_edges]
+│   ├── author___writes___paper
+│   │   └── edge_index.npy     //作者和论文的边，shape=[2,num_edges]
+│   ├── paper
+│   │   ├── node_feat.npy    //论文节点的特征，shape=[num_node,768]
+│   │   ├── node_label.npy   // 论文的标签
+│   │   └── node_year.npy   // 论文年份
+│   └── paper___cites___paper
+│       └── edge_index.npy  // 论文引用关系的边shape=[2,num_edges]
+├── raw //空文件夹
+└── split_dict.pt    //切分训练集、验证集、测试集方式的文件，用torch读取是一个字典，keys=[‘train’,’valid’,’test’], value是node_index
+
+```
+
+
+
+### docker镜像
+
+#### opeceipeno/dgl:v1.4
+
+ogb代码的运行环境，想法是通过虚拟环境去激活各个方案的运行环境，当前做好了Google的mag240m运行环境
+
+[GitHub地址](https://github.com/deepmind/deepmind-research/tree/master/ogb_lsc/mag)
+
+```
+docker run --gpus all -it -v /mnt:/mnt opeceipeno/dgl:v1.4 bash
+# 启动容器后，激活Google代码的运行环境
+source /py3_venv/google_ogb_mag240m/bin/activate
+# /workspace 目录有代码
+```
+
+Google方案预处理后的数据目录：`/mnt/ogb-dataset/mag240m/data/preprocessed`，相当于执行完了`run_preprocessing.sh`脚本，下一步是可以复现实验，
+
+
+
+#### opeceipeno/graphsage:gpu
+
+graphSAGE的环境，[GitHub地址](https://github.com/qksidmx/GraphSAGE)
+
+```
+docker run --gpus all -it opeceipeno/graphsage:gpu bash
+#/notebook目录下面有代码，运行实验参考readme文档
+```
diff --git a/graphsage/doc/preprocess.ipynb b/graphsage/doc/preprocess.ipynb
new file mode 100644
index 00000000..164c2615
--- /dev/null
+++ b/graphsage/doc/preprocess.ipynb
@@ -0,0 +1,446 @@
+{
+  "nbformat": 4,
+  "nbformat_minor": 0,
+  "metadata": {
+    "colab": {
+      "name": "graph.ipynb",
+      "provenance": [],
+      "collapsed_sections": []
+    },
+    "kernelspec": {
+      "name": "python3",
+      "display_name": "Python 3"
+    },
+    "language_info": {
+      "name": "python"
+    }
+  },
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "source": [
+        ""
+      ],
+      "metadata": {
+        "id": "lFBIUQovI53M"
+      }
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 319
+        },
+        "id": "FXItxCYQ5xE2",
+        "outputId": "45d18b2d-50ad-4361-fb5c-1401166b8757"
+      },
+      "outputs": [
+        {
+          "output_type": "display_data",
+          "data": {
+            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAb4AAAEuCAYAAADx63eqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdd1iUR/c38C8iscSC0hYBUTAWCDbsihoBNUrswY7BghW7YqGDCjbQKEoQNYpd4xMLFiyoURBFxYIFK0W6FJG2y33eP/LqTyMgZXfvLfO5rud6QPaeORuWPTtzz5xRISICwzAMwyiJGnwHwDAMwzDSxBIfwzAMo1RY4mMYhmGUCkt8DMMwjFJhiY9hGIZRKizxMQzDMEqFJT6GYRhGqbDExzAMwygVlvgYhmEYpcISH8MwDKNUWOJjGIZhlApLfAzDMIxSYYmPYRiGUSos8TEMwzBKhSU+hmEYRqmwxMcwDMMoFZb4GIZhGKXCEh/DMAyjVFjiYxiGYZQKS3wMwzCMUmGJj2EYhlEqLPExDMMwSqUm3wEwkpGRV4Sj0Yl4kpKL3EIRGtSuidaCBvjVXB8a9WrxHR7DMAxvVIiI+A6CEZ+YhGxsDX+OK8/SAQBFIu7Tz2rXrAEC0LeVFmb1aYF2Buo8RckwDMMflvgUSEjka6wKfYJCUQnK+62qqAC1a6pi5aDWmNCtmdTiYxiGkQVsqlNB/Jv0HqNAyH3zsURAgbAEq0IfAwBLfgzDKBW2uEUBxCRkY1Xokwolvc8VCDmsCn2C+4nZEoqMYRhG9rDEpwC2hj9HoaikStcWikoQEP5czBExDMPILpb45FxGXhGuPEsv955eeYiAy0/TkZlXJN7AGIZhZBRLfHLuaHRitdtQAXD0TvXbYRiGkQcs8cm5Jym5X2xZqIpCEYcnye/FFBHDMIxsY4lPzuUWisTSzr3Ypzhx4gQePnyI/Px8sbTJMAwji9h2BjnXoLZ4foXCDzkIDNyDly9f4vXr11BXV4exsTGMjIxgZGT0xdcCgQAqKipi6ZdhGEbaWOKTc60FDVCrZkq1pjtr16yBiT9bYvp6BwAAx3F4+/YtXrx4gZcvX+Lly5c4c+bMp68/fPiA5s2bl5oYmzVrhtq1a4vr6TEMw4gdq9wi5zLyitDT91K1Ep+qCuH60p8gUP++Qo/Pzc3Fq1ev8PLly0/J8eP/x8fHQ1tb+6uE+PH/NTU12WiRYRhescSnABz23kbY49QqbWlQAVD3XRwKL/wOV1dXjB8/HjVrVn0iQCQSITEx8dPo8PNR44sXLyASiT4lxf9OoRoaGuK7776rct8MwzAVwRKfAohJyMaYoEgUCCu/ib2OmioOOXRD1osYuLi4IC0tDe7u7rC1tUWNGuJf+5SVlfUpEf43MSYlJUFXV7fUkaKRkREaN24s9ngYhlE+LPEpiJDI13D7+z5KoFrha+qo1cDKQW0+1eokIly4cAHOzs4oKCiAh4cHhg0bJrWpSaFQiPj4+FJHii9evICqqmqpI0VjY2MYGBhUa6TKMIzyYIlPQVy6dAl2nkGoZ2GHohKu/GlPjoOaqgrchvxYaoFqIsLp06fh4uICVVVVeHp64ueff+b13hwRITMzs8wp1NTUVBgYGJSZGBs0aMBb7AzDyBaW+BRARkYG2rdvj507d0Jg0gV+52Nx6XEq6tSujcJSzuNrq6WKG0FueHTtDNTVyz6Tj+M4HD9+HK6urmjYsCG8vb3Rr18/KTyjyisqKsKbN2++Wmzz8X+1a9cucwpVT08PqqoVHykzDCPfWOKTc0SEIUOGoE2bNli7di0A4MCBAwg58j8Mmb8aT5LfI7dQiAa11dBatz5Gdfz3BPbp06dDTU0NW7Zs+WYfJSUlOHjwINzd3aGvrw8vLy/06tVL0k9NbIgIaWlppY4UX758iczMTBgaGpY6UmzevDnq1avH91NgGEaMWOKTc5s3b0ZISAj++eefTysiJ0yYgN69e8PBwaHM67KysmBiYoITJ06gc+fOFepLJBJhz5498PT0ROvWreHl5VXha2VZQUHBV9szPn796tUrNGjQoNSRorGxMQQCgUQWATEMIzks8cmxu3fvon///oiMjISxsTGAf0dnOjo6uHv3LgwMDMq9PiQkBBs3bkRUVFSlFoYUFxcjODgYq1atgrm5OTw9PdGuXbtqPRdZxXEcUlJSSh0pvnjxArm5uV9s5v88MTZv3hx16tTh+ykwDPMfLPHJqby8PJibm8PNzQ3jxo379O+RkZFwcHDA/fv3v9kGEcHS0hJDhw7FvHnzKh1DQUEBAgMD4evrCwsLC3h4eKBNmzaVbkee5eXl4dWrV6Umxjdv3kBDQ6PUkaKRkRG0tbXZZn6G4QFLfHJqypQpKCkpwe7du7/4dxcXF4hEIqxZs6ZC7Tx9+hQ9e/ZETEwM9PT0qhTLhw8fsGXLFmzYsAEDBgyAm5sbWrRoUaW2FElJSQmSkpLKvLdYWFhY6mZ+Y2NjGBoaolatWnw/BYZRSCzxyaGDBw/Czc0N0dHRXy28MDc3h7+/PywsLCrcnpubG2JjY3HkyJFqxZWbmwt/f39s3rwZw4cPh7OzMwwNDavVpiLLyckpdTP/ixcvkJiYCB0dnVJHisbGxmjcuDEbLTJMFbHEJ2devnyJrl274ty5c+jYseMXP0tOToapqSnS0tIqdc+usLAQZmZm2LRpEwYNGlTtGN+9e4cNGzZg+/btGDt2LFasWIEmTZpUu11lIhKJkJCQUOa9RSIqcwq1adOmUFNT4/spMIzMYolPjgiFQlhYWGD06NFYsGDBVz/ftWsXzp49i0OHDlW67bCwMDg4OODRo0eoW7euOMJFWloafH19sWvXLtjb28PJyQna2tpiaVvZvXv3rswp1OTkZOjp6ZWZGMvbu8kwyoAlPjmyfPly3L9/H6dOnSp1mmvUqFEYMmQI7OzsqtT+uHHj0KxZM6xevbq6oX7h7du3WL16NQ4cOIDp06dj8eLFrO6mBBUXF+PNmzeljhRfvnwJNTW1Ms9aNDAwYJv5GYXHEp+cuHDhAiZNmoS7d++WOmoSCoXQ0tLCs2fPqjyqSklJgZmZGcLDw2FqalrdkL8SHx8Pb29v/PXXX3B0dMSCBQtYKTEpIyJkZGSUOlJ8+fIl0tLS0LRp0zITY/369fl+CgxTbSzxyYG0tDR06NABe/bsgaWlZamPCQ8Px9KlSxEVFVWtvgICAnDgwAFcuXJFYhuznz9/Dk9PT5w9exYLFy6Eo6Mjvv++YmcBMpJVWFiI169flzpSfPnyJerVq1fmWYtNmjRRmM38GXlFOBqdiCcpucgtFKFB7ZpoLWiAX83/rXzEyDeW+GQcx3GwsbFBu3btyt2isGTJEtSrVw9ubm7V6q+kpATdu3fHzJkzYW9vX622vuXx48dwd3fHlStX4OTkhBkzZrAN3zKMiJCSklJmofDs7Gw0a9as1JGikZGR2O4dS1JMQja2hj/HlWfpAPDFAc8fa932baWFWX1aoJ0Bu1cqr1jik3F+fn44dOgQrl27Vu5KPVNTU+zevVssJcTu3r2LgQMH4tGjR9DU1Kx2e98SExMDNzc33L59GytWrMDUqVPZgbRy6MOHD59Kv/03Mb5+/Rrq6uplTqEKBALet2eERL7GqtAnKBSVlHu6iYoKULumKlYOal3q6SaM7GOJT4bduXMHAwYMQFRUFJo3b17m416/fo2uXbsiOTlZbFNNCxYsQE5ODnbu3CmW9iri1q1bcHV1xePHj+Hq6go7Ozt2xp6C4DgOb9++LXMK9cOHD2jevHmpK1GbNWuG2rVrSzS+f5PeYxQIuW8/+P/773mWjPxgiU9GvX///lMdzDFjxpT72ICAAERFRX1VxaW6/ZuYmGDfvn3o3bu32NqtiOvXr8PFxQUJCQlwd3fHmDFj2EpDBZebm1tmofD4+Hhoa2uXedaipqZmtUaLMQnZGBMUiQJhSaWvraOmikMO3dBWn017yhOW+GTUb7/9BlVVVQQHB3/zsTY2NrCzs4Otra1YYzh27BhcXV1x9+5dXqYeL126BBcXF2RnZ8PDwwMjRoxQmMUTTMWVlJQgMTGx1JHiixcvIBQKy5xCNTQ0/OZr12HvbYQ9Ti3/8OYyqKgAA0x0sH1Cpyo+O4YPLPHJoH379sHLywvR0dHfXO1YUFAAHR0dvHnzBo0aNRJrHESEX375BT179sTy5cvF2nZlYjh79uynGqReXl6wsbHh/X4QIzuysrK+KBT+eWJMSkqCrq5umZv5ue++R0/fS18sYqmsWjVr4IZTP7baU46wxCdjXrx4gW7duiEsLAzt27f/5uPPnj2L1atX4+rVqxKJ59WrV+jcuTOioqJgZGQkkT4qgojw999/w9XVFXXq1IGXlxesra1ZAmTKJRQKER8fX+pI8cWLF6jTcQjqdBkFqFa9xFvtmjWwwLolpvc2FmPkjCSxxCdDiouL0atXL0yYMAFz586t0DWOjo7Q19eHk5OTxOLy8fHB1atXcfr0ad4TDcdxOHLkCNzc3KCtrQ0vLy/06dOH15gY+UREmLX3Js48zqx2W8Pb68Fv9Lc/qDKygd0wkSHOzs7Q0dGBo6NjhR5PRAgNDRVLYenyLFq0CPHx8fjrr78k2k9F1KhRA6NHj8bDhw8xdepU2Nvbw9raGpGRkXyHxsgZFRUVFJF4Fk3lFgrF0g4jHSzxyYjz589j//792LVrV4VHVc+ePUNxcTF+/PFHicampqaGbdu2Yf78+cjNzZVoXxVVs2ZN2NnZ4enTp7C1tYWtrS1sbGxw584dvkNj5IBIJEJERARexz0WS3sNarPTMOQJS3wyIDU1Fb/99hv27t1bqQ3jp0+fxuDBg6Uy/WhhYYH+/fvD1dVV4n1VhpqaGqZNm4a4uDgMHDgQNjY2GDlyJB4+fMh3aIwMISI8f/4c27Ztw4gRI6ClpYUZM2bg++IsqFXzz6d2zRporctqmMoTlvh4xnEcJk2ahMmTJ+Onn36q1LXSmOb8nK+vLw4ePCiTo6patWphzpw5eP78OXr06AFLS0uMGzcOz5494zs0hifv3r3D0aNH4eDgACMjI/Tu3RuRkZEYMWIEHj9+jJiYGOxynYEaqtV7GyQAozrqiydoRirY4haebdiwAceOHcOVK1cqdXjo+/fvoaenh7dv3351Crsk7d69GwEBAYiIiJDpTeXv37/H5s2b4e/vDxsbG7i6upZb/YaRf8XFxbhx4wbCwsIQFhaGJ0+eoFevXujfvz+sra1hYmLy1exIfn4++rkeQLKqNlSqsEeU7eOTT2zEx6Pbt2/D19cX+/fvr/SJ2RcuXED37t2lmvQAYNKkSahTpw62b98u1X4rq379+li5ciXi4uJgYGCATp06YcaMGUhMTOQ7NEZMiAiPHj2Cv78/Bg8eDE1NTSxduhQAsHbtWqSnpyM0NBTz58+HqanpV0nv7t276NSpE7Qy7qLOd1UrjVe7pipm9W1R7efCSBcb8fHk/fv36NixI1avXo1ff/210tdPmzYNP/74I+bNmyeB6MoXGxuLPn364P79+9DV1ZV6/1WRkZGBdevWISgoCBMnTsTy5cshEAj4DouppNTUVFy4cAHnz5/HhQsX8N1338Ha2hrW1tbo168fNDQ0vtlGSUkJ1q9fjw0bNsDf3x/jxo1jtTqVDEt8PJk4cSLq1KmDP/74o9LXEhH09fURHh6OH374QQLRfduKFSvw6tUrHDhwgJf+qyolJQU+Pj7Ys2cPpk2bhiVLlkjlBAqmavLz83Ht2rVP05fx8fHo27fvp+lLY2PjSi3uio+Ph52dHYgIe/bsgaGh4aefVfR0BhAHVRWCx9C2LOnJKTbVyYO9e/ciOjoa/v7+Vbo+JiYGdevW5S3pAf/uObx58ybOnz/PWwxVIRAI4O/vj/v37yM3NxetWrWCq6srsrOz+Q6Nwb+Lve7cuQNfX19YWVlBR0cH3t7eaNCgAbZv34709HQcP34cM2fORIsWLSqV9Pbv349OnTrh559/xqVLl75IegAwoVszHHLohn4tNUCiYtSq+eXbY+2aNVCrZg30NlbH++Oe6NK4WCzPmZE+NuKTsri4OPTo0QMXL15E27Ztq9TG6tWrkZaWVuXEKS6hoaGYO3cuHjx4ILcHyL569QpeXl44efIk5s+fj7lz56J+fbY0XZoSEhIQFhaG8+fP4+LFi9DQ0IC1tTX69++PPn36oEGDBtVqPzs7G7Nnz8adO3ewb98+dOzYsdzHHz9+HL8H7cboZRvxJPk9cguFaFBbDa1162NUx39PYPf398eJEydw8eJF3qsZMVVAjNQUFRVRx44dacuWLdVqp2fPnnTu3DkxRVU9o0aNIhcXF77DqLanT5/SuHHjSFtbm9atW0cfPnzgOySFlZOTQ3///TfNmTOHWrVqRZqamjR69GgKDg6mN2/eiLWv8PBwMjQ0pNmzZ1f4dzp16lTy8/Mr9zFCoZA6duxIf/75pzjCZKSMJT4pWrRoEQ0dOpQ4jqtyG5mZmdSgQQMqKCgQY2RVl5iYSJqamvT48WO+QxGLBw8e0MiRI0lXV5c2b95MhYWFfIck94RCId24cYM8PDyoV69eVK9ePbK0tCQfHx+Kjo6mkpISsfdZVFRETk5OpKurS6dPn67wdRzHUZMmTejp06fffOzt27dJR0eH0tPTqxMqwwOW+KQkNDSU9PX1KSMjo1rt7N+/n4YMGSKmqMRj06ZN1Ldv32oldFkTHR1NgwcPJgMDAwoMDKTi4mK+Q5IbHMdRXFwcBQQE0PDhw0ldXZ3atm1LixYtorNnz0p8NB0bG0sdOnSgX375hVJTUyt17b1798jY2LjCr+V58+aRvb19VcJkeMQSnxQkJyeTQCCg8PDwarc1YcIE2r59uxiiEp+P0z579uzhOxSxi4iIICsrKzIyMqLdu3eTUCjkOySZlJmZSYcPH6Zp06ZRs2bNSFdXlyZNmkQhISGUkpIilRg4jqMtW7aQpqYmBQYGVumD2OrVq2nu3LkVfnxubi7p6+vT5cuXK90Xwx+W+CSspKSErKysyNXVtdptiUQi0tTUFPt9EHGIiooigUBAmZmZfIciEeHh4WRhYUGtWrWiAwcOSGR6Tp4UFRXR5cuXacWKFdS5c2eqX78+DRo0iPz8/Ojhw4dSH/0nJyfTzz//TJ07d67QNGVZevbsSWfPnq3UNcePH6dWrVqxaXE5whKfhPn6+lKvXr3EMlKIiIggMzMzMUQlGXPmzCEHBwe+w5AYjuPo/Pnz1KVLFzIzM6Pjx48r1PRueTiOo4cPH5Kfnx8NGjSI6tevT507d6YVK1bQ5cuXeX3T//vvv0kgEJCzs3O1pqQzMzOpfv36Vbp/PnToUPLw8Khy34x0scQnQZGRkaStrS22EZqLiws5OTmJpS1JyM7OpiZNmtD169f5DkWiOI6jkydPUvv27cnc3JxOnz6tkAkwOTmZ9u7dS3Z2dtSkSRNq1qwZOTg40JEjR2RiZJ+Xl0cODg7UvHlz+ueff6rd3oEDB8jGxqZK18bHx5OGhka1RpuM9LDEJyHZ2dlkZGRER48eFVub5ubmdOXKFbG1JwkHDx4kMzMzpVgMUlJSQkePHiUTExPq3r07Xbx4ke+QquXDhw909uxZWrRoEbVt25bU1dVp+PDhFBAQQHFxcTKV3KOiouiHH36gSZMmUU5OjljanDhxIgUEBFT5+o0bN1K/fv1k6r8TUzqW+CSA4zgaO3YsTZ8+XWxtvn37lho1aiTzCYXjOBowYACtW7eO71CkRiQSUUhICLVo0YL69u1L165d4zukCikpKaHo6Gjy8fGhfv36Ub169ahXr17k4eFBN27ckMmFPEKhkLy8vEhbW5sOHz4stnZLSkpIS0uLXr9+Xa3YOnTooJCLvBQNS3wSsGvXLjI1NaX8/Hyxtblz506ytbUVW3uS9Pz5c9LQ0KjWm4g8EgqFFBwcTIaGhjRgwACKioriO6SvxMfHU3BwMI0ePZo0NTWpVatWNGfOHDpx4oTYRk6S8vLlS+rZsydZWlpSQkKCWNuOjIwkU1PTardz69Yt0tHRqfa2JUayWOITsydPnpCmpiY9ePBArO2OGjWKdu/eLdY2JcnLy0vm9htKS1FREQUEBJCenh4NGTKE7t27x1sspVVJGTNmDAUHB1N8fDxvcVUGx3G0e/du0tTUpI0bN0pkRa2rqystWbJELG3NnTuXJk+eLJa2GMlgiU+MCgsLqUOHDrRt2zaxtltcXEzq6upS2w8lDoWFhdSqVSv63//+x3covMnPzyc/Pz/S0dEhW1tbio2NlXifn1dJ6dmzp1SqpEhSZmYm/frrr/Tjjz9STEyMxPrp1KmT2Pbi5eTkkL6+vlj27TKSwRKfGM2fP5+GDx8u9pvbly9fps6dO4u1TWm4fPkyNW3alN6/f893KLzKy8sjHx8f0tLSookTJ9Lz58/F1vbHKilbt26lYcOGfVEl5dy5c3JdczQsLIz09fVp/vz5Ei3Rl5KSQurq6mK9f/7XX39R69at2d4+GcUSn5icOnWKmjZtKpFl3kuWLCE3NzextysNdnZ2tGjRIr7DkAk5OTnk4eFBGhoaNHXq1CrfA/1vlZQmTZpIvUqKJBUUFNDChQtJT0+Pzp8/L/H+du/eTaNGjRJrmxzH0ZAhQ8jT01Os7TLiwRKfGCQlJZFAIKCrV69KpH1TU1O6efOmRNqWtNTUVNLS0uL1PpesyczMpBUrVlDjxo1p9uzZlJSUVO7jCwsLZapKiiTdv3+fzMzMaOTIkVJbIPLrr7/Szp07xd7umzdv2N4+GcUSXzWJRCLq168fubu7S6T9169fk7a2ttzdm/ncH3/8Qd26dZPr5yAJqamptHDhQmrUqBEtXLjwU0Hlj1VSNm7cSD///PMXVVLCw8OpqKiI58jFr6SkhPz8/EhTU5N27doltWT+8f7527dvJdI+29snm1jiq6bVq1dT7969SSQSSaT9gIAAsrOzk0jb0lJSUkI9evSQueLasiIpKYl+++03+v777+nHH38kgUAgc1VSJCkpKYmsra2pe/fuYr3/WRFXrlyhjh07Sqx9trdPNrHEVw0RERGkra0t0WXhNjY2dPDgQYm1Ly33798nLS0thbgHJQ6lVUkZMGAA9ejRg9TV1cnNzU3m99WJw9GjR0lbW5s8PT152TDv5OREzs7OEu0jKiqK7e2TMSzxVVFWVhY1b96cjh8/LrE+8vPzqX79+vTu3TuJ9SFNS5YsoQkTJvAdBi/Kq5ISERHxxZt+XFwcTZw4kbS0tMjHx4fy8vJ4jFwycnNzyd7enlq0aEGRkZG8xWFmZkYRERES78fR0ZGmTJki8X6YimGJrwo4jqPRo0fTrFmzJNrPmTNnyMLCQqJ9SFNeXh4ZGhrKfU3Linrz5g3t2LHjiyopjo6OdOLECcrNzf3m9bGxsWRra0sCgYA2btwo1kpAfLp+/ToZGRnR1KlTed3q8ubNG9LU1JTYbYrP5eTkkJ6enszX2lUWLPFVQXBwMJmZmUn8jcjR0ZHWrFkj0T6k7cSJE9SyZUuF3N8kqSop9+7do6FDh5Kenh5t3bpVbhe3FBcXk6urK+no6Eh0pqSitm/fTuPHj5daf8eOHWN7+2QES3yVFBsbS5qamvTo0SOJ9sNxHBkZGUm0WgVfhg0bphBnl5VXJeXOnTtiX8UaFRVFAwcOJENDQwoODpbJItJlefbsGXXt2pUGDhwosRWUlTVkyBDat2+f1PrjOI5++eUX8vLyklqfTOlY4quEgoICateuHQUGBkq8rydPnpC+vr5CLoP+eHbZs2fP+A6lUjiOo2fPnn1VJWXx4sVSrZLyzz//UN++falFixYUEhIilam6quI4joKCgkhTU5N+//13mXk9FxYWUoMGDaS+4OTj3j55e+0rGpb4KsHR0ZFGjRollT/ejRs3KvRp5hs2bCBra2uZeSMsiyxXSbl48SL16NGDTExM6MiRIzK3TzI9PZ2GDRtG7dq1o4cPH/IdzhfOnz9PPXr04KXvDRs2kKWlpcy/9hUZS3wVdOLECTI0NJTaCksrKyuFLvAsFAqpXbt2tH//fr5D+cLnVVI6deok81VSOI6j0NBQMjc3p/bt29OJEydkIsYzZ85QkyZNaOnSpTJ5T2vevHnk7e3NS99CoZDat29Pe/fu5aV/hiW+CklMTCQdHR26fv26VPrLzc2l+vXrK3xx54iICNLV1aWsrCzeYuA4jh48ePBVlZSVK1fKVZUUjuPo+PHjZGZmRl26dKFz587xkgDz8/Npzpw5ZGBgQJcuXZJ6/xX1ww8/0J07d3jrn+3t45cKERGYMpWUlMDKygqWlpZwdnaWSp//+9//EBAQgPPnz0ulPz7NmDEDqqqq2Lp1q9T6TElJwYULFxAWFoawsDDUqlUL/fv3h7W1Nfr164fGjRtLLRZx4zgOR44cgZubG7S1teHl5YU+ffpIpe979+5h3LhxaNeuHQICAtCoUSOp9FtZcXFx6NOnD5KSkqCiosJbHHPnzkV+fj527NjBWwxKi+/MK+u8vb2pb9++Ul1AMG3aNPLz85Naf3x69+4dCQQCiRbh/rxKipmZGamrq9Pw4cMpICCA4uLiZGJqUNyEQiH9+eef1Lx5c7KyspLoJu2SkhJau3YtaWlpUUhIiMz/99y0aZNMHBT7cW+fpIrbM2Vjia8c169fJx0dHUpMTJRanxzHkZ6enlJVdA8JCaH27duLbXn+xyopa9as+aJKiqen51dVUhRdcXEx/fHHH2RgYECDBw+m6OhosbYfHx9Pffv2JQsLiyofsyRtAwYMoGPHjvEdBhH9W7KtTZs2cjOlrihY4itDVlYWGRoa0okTJ6Ta771796hFixZS7ZNvHMeRpaVltUa5/62S0rp160pVSVF0BQUFtHnzZtLV1aURI0bQgwcPqt3mgQMHSEtLi9asWSPTWyo+l5eXR/Xq1ZOZOqgcx5GNjQ1vC22UFUt8peA4jkaNGkWOjo5S73v16tU0d+5cqffLt6dPn5KGhgYlJCRU6PGfV0lp2bKl2KqkKLoPHz7QunXrSFtbm8aOHVulmYXs7GwaP348tWrViplLNmsAACAASURBVG7fvi2BKCXnxIkT9NNPP/Edxhdev35NGhoaFBcXx3coSqMG3/cYZdGOHTsQFxeHtWvXSr3v06dPY9CgQVLvl28tW7bEnDlzMG/evFJ/LhKJEBERAQ8PD/Tq1Qt6enrYvHkzDAwMcPDgQaSmpuLAgQOYPHkyDAwMpBy9/Khbty4WL16M58+fw9TUFD179oS9vT1evXpVoeuvXr2Kdu3aoWHDhrhz5w7Mzc0lHLF4hYaGytzfl6GhIZYvX46ZM2eC2FpD6eA788qaR48ekaamJj1+/FjqfWdmZlL9+vWpoKBA6n3LgoKCAmrRogWdPHmy1Cop7dq1+1QlRVEKNvMtKyuLXFxcqHHjxjR9+vQyR9xFRUW0bNky0tXVpVOnTkk5SvHgOI6aNm0q8XKDVfFxX2tISAjfoSgFlvg+k5+fT2ZmZrRjxw5e+j9w4AD98ssvvPQtCzIzM8nZ2Znq1atHhoaGMlUlRdGlp6fT0qVLqVGjRjR37lxKTk7+9LPHjx9Tx44dycbG5tMp8fLo4cOH1KxZM5lddXrz5k0SCAQKf/CwLGD7+D4zZ84cZGRk4MCBA7zs77Gzs0OPHj0wY8YMqffNh6KiIkREROD8+fMICwvD06dPYWFhgZSUFHTo0AFBQUG87rNSRikpKfDx8cGePXswdepUaGlpYe3atfD29oaDg4Nc/z7Wrl2LN2/eSHXPaGU5OjqisLAQQUFBfIei2PjOvLLi+PHj1Lx5c8rOzualf5FIRFpaWvTmzRte+peG0qqkdOnS5asqKcnJyaSpqSmWlYdM1dy5c4eaNm1KqqqqNHPmTF6r64hLnz59ZH6aNjs7m+3tkwKW+OjfvUja2tpSOYm5LJGRkfTjjz/y1r+kJCcn0969e8nOzo50dXWpefPm5ODgQEeOHCl3SicgIIB69eolc4WXlcGJEydIIBDQypUr6enTp2Rvb0+amprk7e0tt1tDsrOzqV69elI7QaM62N4+yVP6xCcSiah37960evVqXuNwdXWlpUuX8hqDOHyskrJw4cKvqqQ8f/68wu2IRCLq0qULBQcHSzBa5nN5eXk0ffp0atas2VcjjqdPn9K4ceNIW1ub1q1bJxcJ5HNHjhyhgQMH8h1GhbC9fZKn9InPw8OD+vXrx/sGXHNzcwoPD+c1hqooKSmh27dvf1ElxcLCQixVUu7evUva2tqUlpYmxoiZ0ty6dYtatmxJdnZ25U73P3jwgEaOHEm6urq0efNmmTx5oTT29vb0+++/8x1GhbG9fZKl1Inv6tWrpKOjQ0lJSbzGkZycTOrq6lRcXMxrHBUlzSopCxYsoN9++02sbTL/RyQSkbe3N2lpadGhQ4cqfF10dDQNHjyYDAwMKDAwUKZfuyUlJaSjo0MvXrzgO5RKWbdunVycWSmPlDbxZWZmUtOmTWXiZveuXbvo119/5TuMMn2skjJ79uwvqqTs3LlT4lVScnNzycDAQC5Hw7Lu5cuX1LNnT+rXr1+FK+b8V0REBFlZWZGRkRHt3r1bJuug3r59m1q1asV3GJVWXFxM7dq1o3379vEdisJRysTHcRyNGDGC5s+fz3coREQ0atQo2rVrF99hfCIUCun69evk7u5OPXv2pHr16pGlpSX5+vrSnTt3pL7g5K+//mI3+8WI4zjas2cPaWlp0fr168Xy+wwPDycLCwtq1aoVHThwQKYWJXl6etKCBQv4DqNKIiMj2d4+CVDKxLdt2zbq0KGDTNyfKC4uJnV1dV43aMt6lZSPN/tXrVrFaxyKIDMzk2xtbcnU1JTu3bsn1rY5jqNz585Rly5dyMzMjI4fPy4T03TdunWjsLAwvsOostmzZ9O0adP4DkOhKF3ie/DgAWlqasrMsT/h4eHUqVMnqfebmZlJhw8fpmnTpn1RJWXfvn0yWSXl1atXpKGhIXf3aWTJxYsXycDAgObNmyfRDzMcx9HJkyepffv2ZG5uTqGhobwlwPT0dGrQoIFMfMitqo97+65du8Z3KApDqRLfhw8fyNTUVKamFZcuXUqurq4S76ewsJAuXbpEy5cvp06dOlH9+vVp8ODB5O/vT48ePZKJT+bf4uPjQwMHDpSLWGVJYWEhLVq0iPT09OjcuXNS67ekpISOHj1KJiYm1L17d7p48aLU+v5o7969NGzYMKn3K25HjhwhExMTNt0vJkqV+GbMmEFjx46VqTdOU1NTioyMFHu7Fa2SIk+Ki4vJ1NSUDh8+zHcocuPBgwfUtm1bGj58OKWnp/MSg0gkopCQEGrRogX17dtXqiOXsWPH0h9//CG1/iSF4zgaPHgwm+4XE6VJfMeOHSMjIyOZOYCS6N+9OlpaWmJbCFBWlZSjR4/Su3fvxNIH365du0Z6enoy9XuURSUlJeTv70+ampoUHBwsEx/2hEIhBQcHk6GhIQ0YMICioqIk2p9IJKLGjRtXecWqrPk43c/29lWfUiS+N2/ekLa2Nt28eZPvUL6wbds2mjhxYpWvL61KyogRI2jbtm2VqpIib6ZMmaKUh/VWVFJSEvXv35+6desmk6+DoqIiCggIID09PRoyZIjYF9l8dP36dWrbtq1E2uYL29snHgqf+IRCIfXq1Yt8fX35DuUrNjY2dODAgQo/XpJVUuRJRkYG6ejoyN3p39Jw7Ngx0tHRIXd3d5l/PeTn55Ofnx/p6OiQra0txcbGirX9lStX0vLly8XaJt/Y3j7xUPjE5+rqStbW1jK1r4jo30NX69ev/80pyI9VUmxtbUlDQ0OiVVLkye7du6lTp068l5qTFbm5uTR58mQyNjbmtdh6VeTl5ZGPjw9paWnRxIkTxTZKbd++vUKuhPy4t09Rbl/wQaETX3h4OAkEgi8O1ZQVZ8+epV69en317/+tkqKlpUVjx46VSpUUecJxHPXp00eu6i9KSkREBBkbG9OUKVPk+sNQTk4OeXh4kIaGBk2dOrVaR3QlJiZSo0aNZH7UW1WzZs0iBwcHvsOQWwqb+DIyMsjAwIBCQ0P5DqVUjo6OtHr16lKrpFhZWfFWJUWexMbGkqamJu+1VvkiFArJzc2NdHR06K+//uI7HLHJzMykFStWUOPGjWn27NlV+v1+rCWrqLKzs6lJkyb0zz//8B2KXFLIxMdxHA0dOpQWLlzIdyhf4TiOnj59ShoaGvTTTz/JXJUUebNy5UqytbXlOwypi4uLo65du1L//v0VNvGnpqbSwoULqVGjRrRw4UJKTU2t8LUjRoygP//8U4LR8e/w4cNkamoql1uT+KaQiW/r1q1kbm4uMy+IjIwMOnToEE2dOpUMDQ1JW1ub6tatSyEhITJZJUWe5Ofnk5GREZ05c4bvUKSC4zjasWMHaWpq0qZNm5RiRiApKYlmz55NjRs3puXLl3+zbmVRURE1bNiwUolSHnEcR4MGDeL9LFF5pHCJLyYmhjQ1NenZs2e8xfDfKikNGjT4okrKxo0bWe09MTpz5gwZGRkp/Gg5PT2dhg8fTm3btqWHDx/yHY7UvX79mqZOnUoaGhrk7u5e5l7OixcvUpcuXaQcHT8+7u2TxW0rsqwGFEh+fj7GjBmDjRs34ocffpBav0SEhw8fws/PD4MGDYKWlhaWLVuGGjVqYP369UhPT8epU6cwb948mJiYIDQ0FIMGDZJafIpu4MCBMDc3x+rVq/kORWLOnTuH9u3bw9jYGFFRUTA1NeU7JKkzNDREUFAQIiMj8eLFC7Ro0QK+vr748OHDF49Tpr+vZs2awcnJCbNmzQIR8R2O/OA784rTtGnTaMKECVLp62OVlIkTJ1aqSsr79++pXr169P79e6nEqSySkpJIU1NT7HvB+Jafn09z584lAwMDXmpdyrLY2FiytbUlgUBAfn5+VFBQQEREbdq0kXhVGFlSXFxMbdu2pf379/MditxQmMR3+PBhatGihcSWc3/48IHOnDnzqUpKo0aNqlQl5X//+x9ZWVlJJEZlt3nzZurTp4/CVLW4d+8emZqakq2tLduzVY579+7R0KFDSU9Pjzw9PUlbW1sp7n1+LiIigu3tqwQVIvkYH2fkFeFodCKepOQit1CEBrVrorWgAX4118f7jGR06dIFoaGh6NSpk1j64zgOd+/eRVhYGMLCwhAVFYUOHTrA2toa1tbW6NSpE2rWrFnpdh0cHNCmTRssWLBALHEy/6ekpARdu3bF3LlzYWdnx3c4VcZxHDZu3Ii1a9di48aNGD9+PFRUVPgOS+bdunULkyZNQnx8PDZv3gw7O7sq/Y3Kq9mzZ0MkEiEwMJDvUGSezCe+mIRsbA1/jivP0gEARSLu089q16wBAqCSHIshLWpj7bLZ1eorPj4eYWFhOH/+PC5evAgtLS1YW1ujf//+6NOnD+rXr1+t9okIBgYGuHTpElq2bFmttpjSRUdHY/DgwYiNjUXjxo35DqfSEhISMGnSJAiFQuzduxfNmjXjOyS5MnjwYHTt2hWXL19GYmIi3N3dMWbMGKiqqvIdmsTl5OTAxMQEhw8fRs+ePfkOR6bJdOILiXyNVaFPUCgqQblREofa39WE86A2mNCtWYXbz83NxeXLlz+N6rKysmBlZQVra2tYWVnBwMCg2s/hc/fv38eIESMQFxfHPsFL0Ny5c1FQUICgoCC+Q6mUQ4cOwdHREfPnz4eTk5NSvFmLU0FBAbS1tREfH49GjRrh0qVLcHZ2Rk5ODjw8PDBixAjUqKFQ6/m+cvjwYXh5eeHOnTtQU1PjOxyZJbOJ79+k9xgFQu7bD/7/6qjVwMpykp9IJEJUVNSnRBcTE4Nu3bp9mr5s166dRP8w1qxZg+TkZGzevFlifTDy98k3JycHjo6OuHnzJvbt2ye26Xplc+bMGaxZswZXr1799G9EhLNnz8LFxQUlJSXw9PSEjY2Nwn7wJCIMHjwYvXv3xrJly/gOR2bJZOKLScjGmKBIFAhLKn1tHTVVHHLohrb66iAiPH/+/NP0ZXh4OJo1a/Zp+rJXr16oU6eOBJ5B6SwsLODs7IwBAwZIrU9lJS+ffK9duwY7OzsMHDgQ69evx/fff893SHLL0dERenp6pb7hExH+/vtvuLq6ok6dOvDy8oK1tbVCJsBXr16hc+fOiIqKgpGREd/hyCSZTHwOe28j7HFq+dObZVABYNpQBN0XpxAWFgaRSPRpRGdpaQkdHR2xx1sRWVlZMDQ0RFpaGmrXrs1LDMqEiDBo0CD89NNPWLp0Kd/hfKW4uBgeHh7YuXMngoKCYGNjw3dIco2IYGxsjL///htmZmZlPo7jOBw5cgRubm7Q1taGt7c3evfuLcVIpWPt2rW4dOkSzpw5o5DJvbpkLvFl5BWhp++lLxaxVJYKJ4KDIB5DB1iiTZs2MvGLP3ToEEJCQnDy5Em+Q1EaL168QNeuXXH79m2ZWiTy9OlTjB8/HgKBAMHBwbx9GFMkT548gbW1NeLj4yv09y4SibB//364u7vD2NgYXl5e6NatmxQilQ6hUAhzc3OsWLECY8aM4TscmSNzd3qPRidWu41a332HxuaDYGJiIhNJDwBOnz6tNNUkZIWxsTEWLFgAR0dHmahqQUTYvn07evXqhalTp+LkyZMs6YnJx2otFf17r1mzJuzs7PD06VPY2trC1tYWNjY2uHPnjoQjlQ41NTUEBgZi4cKFyMrK4jscmSNzie9JSm61RnsAUCji8CT5vZgiqj6O43D27FmW+HiwePFiPH/+HH///TevcaSlpWHIkCEICgrCtWvXMGPGDJn5UKYIqlqmTE1NDdOmTcOzZ88wYMAA2NjYYOTIkXj48KEEopSu7t27Y9iwYVi+fDnfocgcmZvqnPznLVx6klbtdlTePkTzhHPQ19eHgYEB9PX1v/haQ0NDam88UVFRmDx5skL8McmjK1euYOLEiXj06FG192JWxalTpzBt2jTY29vD3d0d3333ndRjUGTv379HkyZNkJycjHr16lWrrfz8fAQEBGDdunWwtLSEu7u7XO+5zc7OhqmpKY4cOYIePXrwHY7MkLnEN//QXfzv3ttqt2NpVA9jmwuRkJCAxMREJCYmfvF1QUHBp2T436T48WtNTU2xJEc3NzcUFBRg7dq11W6LqZrffvsNGhoa2LBhg9T6zM/Px+LFixEaGoq9e/fCwsJCan0rk+PHj2Pbtm04f/682Np8//49Nm/eDH9/f9jY2MDV1RXNmzcXW/vSJC8rnKVJ5hLf9isv4HfhWbWmO2vXrIEF1i0xvbdxmY/Jy8tDUlLSVwnx8+8/fPgAPT29UkeMnyfHb+3969y5M9avX48+ffpU+Tkx1ZOeno4ff/zx0ykHkhYdHY3x48ejc+fO2LJlCxo2bCjxPpXVtGnTYGpqivnz54u97ezsbGzcuBFbt26Fra0tVq5cCX19fbH3I0lsb9/XZC7xiWNVZ62aNXDDqR806tWqViz5+fmlJsTPv37//j309PTKnFKtVasWLCwskJ6ezj5t8Sw4OBhBQUG4ceOGxAoVlJSUYO3atfDz88PmzZvZijoJIyLo6+vj8uXLEp2SzMjIwLp16xAUFAQ7OzssW7YMAoFAYv2JG9vb9yWZS3xA9fbxgeNQ510cDjlaoW3btmKP7b8KCgqQlJRU5pRqXFwc8vPz0bRp0zKTo76+PnR0dBS+nBLfOI5D7969MWHCBMyYMUPs7b9+/Rp2dnZQVVXFnj17xF7yjvlaTEwMRo4cKbUygCkpKfDx8cGePXswbdo0LFmyBJqamhLvVxx8fX1x+fJltrcPMpr4qle5pQZsNd5i26rlGDt2LDw8PKCuri6BKCvG1tYWVlZWsLS0LHNKNTExEdnZ2dDV1S1zSvVjcmT1G6vnwYMH6NevHx4+fCi2rQREhH379mHhwoVYunQpFi5cyD7ESAlfZQATExOxatUqHD58GLNnz8bChQt5fZ+pCKFQiI4dO8LZ2RmjR4/mOxxeyWTiA6pfqzMjIwMrVqzAyZMnsWbNGtjZ2Un9zUgoFEJbWxuPHz/+5rRIYWEh3r59W+aUamJiIt69e/fN5CgQCFhy/AYnJyckJSUhJCSk2m1lZWVh5syZePDgAfbt2yeV+4fM/+nVqxdcXFx4KwP46tUreHp64tSpU5g/fz7mzp3Ly8rhioqIiMDIkSMRGxsr84lakmQ28QEVP51BRQWoXVMVKwe1/qpA9a1btzB79mzUrFkTW7ZsQceOHSUb9GeuXLmCRYsW4fbt22Jpr6ioqNTk+Pn3mZmZEAgE5a5WFQgESnVO2X99+PABpqam2LFjB6ysrKrczuXLlzFp0iQMGzYMvr6+Uq37ygDv3r1Ds2bNZKIM4NOnT+Hh4YGLFy9iyZIlmDVrFurWrctrTGWZOXMmAGDbtm08R8IfmU58AHA/MRsB4c9x+Wk6VPDv5vSPPp7H91MrLczq2wJt9Uv/BMNxHHbu3ImVK1di5MiR8Pb2lspZbU5OTqhVqxY8PT0l3tdHxcXFn5JjWaPHjIwMaGtrl7taVVdXV6GT46lTp7Bw4ULcv3+/0m+aRUVFcHFxwb59+xAcHIyBAwdKKEqmPAcPHsS+fftkqgzgw4cP4e7ujhs3bmD58uVwcHBArVrVW2QnbtnZ2TAxMcGxY8fQvXt3vsPhhcwnvo8y84pw9E4iniS/x/nwa/ixpRH6mbfGqI76FV69+e7dO7i4uODo0aPw9vbGlClTJDr9aWZmhh07dqBr164S66MqhELhF8mxtASZnp4OLS2tLxLifxNkkyZN5Hql6ogRI9CuXTu4ublV+JpHjx5h/PjxaN68Of744w9oaWlJMEKmPHZ2dujevfunEYwsuXPnDlxdXXH//n04OzvD3t5epv5WDh06BG9vb6Xd2yc3ie9zU6ZMQffu3TF16tQqXX/37l3Mnj0bIpEIW7ZsQZcuXcQc4b+nuZubmyMlJUUu77kJhUKkpKSUeb8xISEBaWlp0NTULHe1apMmTWS2UkliYiLat2+PiIgI/PDDD+U+loiwZcsWeHp6wsfHB5MnT1b6lXF84jgOAoEAt27dgqGhId/hlCkyMhIuLi54+fIlXF1dMX78eJmYSfl4eknfvn3h5OTEdzhSJ5eJz83NDSoqKnB3d69yGxzHYe/evVi2bBlsbGywZs0asS5L3r59O65fv469e/eKrU1ZIxKJkJKSUu6CnJSUFGhoaHyVED//vkmTJrxNB/n5+eH06dMICwsrM5ElJyfD3t4eWVlZCAkJ+WaSZCTv5s2bmDJlityUAbxy5QpcXFyQlpYGd3d32Nra8r7yV5n39sll4gsKCkJkZCSCg4Or3VZ2djbc3d0/HVEyffp0sYzQhgwZgnHjxin9BmaRSITU1NRyk2NycjIaN25c7mpVPT09iSRHkUiETp06YenSpRg3btxXPz9+/DhmzpyJGTNmYOXKlUo5LSSL5LEMIBEhLCwMLi4uKCgogKenJ4YOHcrrzIGy7u2Ty8R35swZ+Pv749y5c2Jr8/79+3B0dMT79++xdevWat30LSwshLa2Nl6/fi2VRTTyrqSk5FNyLCtBvn37Furq6uUuyNHT06vS6r6bN29i+PDhePToERo1agTg35J28+fPx+XLlxESEqK0iwBkVefOnbFu3Tr07duX71AqjYhw6tQpuLi4oGbNmvDy8sLAgQN5STzKurdPLhPfw4cPMXr0aDx69Eis7RIRDhw4gCVLlsDa2hq+vr5V2uR87tw5eHl54Z9//hFrfMqM47gvkmNpCfLt27do0KDBN5NjadsOZs2aBSLCtm3bcPPmTUyYMAEWFhbYtGmTTO/LUkapqalo1aqV3JcB5DgOf/31F9zc3NCwYUN4e3ujX79+Uo/jxo0bGDVqlFLt7ZPLxJednQ1DQ0Pk5ORIpP3c3Fx4enrizz//hLOz86d9gBU1b9486OjoYMWKFRKJjykdx3FIT08vd0FOUlIS6tev/1VS1NDQwIoVK2BlZYXw8HBs27YNI0eO5PspMaX4888/cfLkSRw9epTvUMSipKQEBw8ehLu7OwwMDODl5YWePXtKNYYZM2agRo0aCAgIkGq/fJHLxEdEqF+//qdP+JISGxsLR0dHpKenY8uWLejdu3eFrvvhhx9w9OhRtGvXTmKxMVXDcRwyMjK+KgAQGxuL8+fPo6ioCGpqaqhXr165q1X19fXx/fff8/10lJKtrS1+/vln2Nvb8x2KWIlEIuzZsweenp5o06YNPD090blzZ6n0nZWVBVNTU6XZ2yeXiQ8AWrdujePHj6NNmzYS7YeIcPToUSxatAi9e/fG2rVr0aRJkzIf/+zZM/z0009ITExUqpvF8oqIsHv3bixduhQrV67EqVOnMGjQIEycOLHcBTmJiYmoU6dOuQty9PX1q30wKvOlj2UAY2Njoaury3c4ElFcXIzg4GCsWrUK5ubm8PT0lMqH6IMHD2L16tWIjo6W6ynkipDbxGdlZYWlS5eif//+UukvLy8Pq1atQlBQEJYvX465c+eW+uLw9/fHo0ePEBQUJJW4mKrLzMyEg4MD4uLisG/fPpiZmSEuLg7du3fH3bt3yz1dgYiQmZlZ7oKcxMRE1KpV65vJkd1DrLirV69iwYIFiI6O5jsUiSsoKEBgYCB8fHzQp08fuLu7S/SDPhHh559/Rr9+/bB06VKJ9SML5Dbx2dvbw8LCApMnT5Zqv8+ePcPcuXMRHx+PLVu2fHUzun///pg1axaGDRsm1biYyjl//jwmT56MMWPGwNvb+4vVoB4eHoiJicFff/1VrT6ICO/evSt3QU5CQgLU1NTKXZCjr68v0Sl9ebJs2TKoqanBy8uL71Ck5sOHD9iyZQs2bNiAgQMHws3NDcbGZR+yXR0vX75Ely5dcOvWLbk9cb4i5Dbxubi4QE1NDa6urlLvm4jw999/Y/78+ejatSvWr18PAwMD5OXlQVdXF2/fvmWf4mVUYWEhli1bhmPHjmH37t2wtLQs9TFt27bFhg0b8Msvv0g0HiJCdnZ2uQtyEhISoKqqWqHkqOjT623btkVgYKBS3If6r5ycHPj7++P333/H8OHD4eLigqZNm4q9Hx8fH1y5cgWhoaEK+3qS28QXGBiI6Oho/PHHH7zFkJ+fDx8fH2zduhVLlixBixYtsH37dly4cIG3mJiy3b9/H+PGjYOJiQm2b99e7h7LixcvYsqUKXj06BHvi1iICDk5OeWe55iQkAAA5U6pGhgYoGHDhnL7ZibvZQDF5d27d9iwYQO2b9+OsWPHYsWKFeWuO6isj3v7XFxcYGtrK7Z2ZYncJr7Tp09j69atCA0N5TsUvHjxAvPnz8e1a9cwevRoBAYG8h0S8xmO4+Dv7481a9Zgw4YNmDhxYoXe/CdMmAA9PT34+vpKIcrqISLk5uaWe78xISEBHMd9Mzmqq6vLZHIMDAzEtWvXxHKOoiJIS0uDr68vdu3aBXt7ezg5OUFbW1ssbd+4cQO//vorHj16pJB7++Q28cXExGDChAl48OAB36EA+PeNR0tLC3Xr1oW5uTn8/PzQrFkzvsNSeomJiZg0aRKKioqwd+/eSt23SE1NxY8//ohLly7BzMxMglFKz8fkWF6CFAqF31yQ07hxY6knx6FDh2L06NGllpZTZm/fvsXq1atx4MABTJ8+HYsXLxZLxShF3tsnt4nv3bt3MDY2RlZWFt+hAPh3Gm348OF4+PAh1q9fD39/f8yfPx9Llizh/ZBMZXXkyBHMmTMHc+fOhZOTU5Wq4m/fvh0hISG4evUq70WFpeX9+/dlbuH4+H1RUdE3k6OGhobYkmNRURG0tLTw6tUraGhoiKVNRfPmzRt4e3vj+PHjcHR0xIIFC6q1KEqR9/bJbeIjInz//fdIS0uTib1SPj4+SEpKwu+//w4AeP36NRYuXIiYmBhs2rQJNjY2PEeoPHJzc+Ho6IiIiAjs27evWpuAOY5Dz549MWXKlCofg6WI8vLyyr3fmJiYiIKCgjLPcvz46qOpHgAAIABJREFUtaamZoWSY1hYGNzc3HDjxg0pPDv59vz5c3h6euLs2bNYtGgR5syZU+X71AcOHMCaNWsUbm+f3CY+AGjZsiVOnjyJVq1a8R0KevfujRUrVnx1Gve5c+fg6OiIVq1awd/fX2LLkJl//fPPP7Czs0P//v2xYcMGsSxMiYmJgbW1NR49esQOnq2EDx8+lJkcP36fn58PPT29bybHhQsXQktLCytXruT7acmNx48fw93dHVevXoWTkxNmzJhR6dknIsLAgQNhZWWFJUuWSChS6ZPrxNevXz+sXLmy1CXp0pSVlQVDQ0OkpqaWWgC5qKgIfn5+WL9+PWbNmoVly5ahbt26PESquIRCITw8PBAcHIzAwEAMGTJErO0vWrQImZmZ2L17t1jbVXb5+flISkoqd0FOXl4eOI6DmZkZWrduXWqC1NLSUpqp6MqKiYmBm5sbbt++jZUrV2LKlCmVOhz6xYsX6Nq1K27fvq0w6xbkOvHZ2dmhX79++O2333iN49ChQ9i7dy9OnTpV7uMSEhKwePFiREVFwc/Pj/ezuBTFs2fPMH78eGhrayM4OBgCgUDsfeTl5cHExAR79uyRy6Nw5NmDBw9gaWmJQ4cOlTmCzMnJQZMmTcqtraqjo6PUyfHWrVtwdXXF48eP4erqCjs7uwrf916zZg3++ecfnDp1SiHes+Q68a1YsQJ169aFs7Mzr3FMmjQJ3bp1w8yZMyv0+IsXL8LR0RFNmzbF5s2b0bJlSwlHqJiICEFBQVi5ciU8PDwwc+ZMif5R/u9//8Py5ctx79493k6MV0abN29GTExMuQdPFxYWIikpqdzVqtnZ2dDV1S13QY6Ojo7C7xG8fv06nJ2dkZiYCHd3d4wZM+abz7m4uBgdO3aEm5sbfv31VylFKjlynfi2bduGmJgYbN++nbcYOI6DQCBAVFRUpaYBiouL8fvvv2PNmjWYNm0anJ2ded8oLU/S0tIwdepUJCUlISQkROLFyoF/E+3QoUPRtWtXdq9JigYOHAgHBweMGDGiWu0UFhbi7du35S7Ieffu3TeTo0AgUIjkeOnSJTg7OyMnJwceHh4YMWJEuSPi69evw9bWFrGxsWjYsCEy8opwNDoRT1JykVsoQoPaNdFa0AC/mutDo55sfzCU68R38uRJBAYGfnOKUZKioqJgb29f5UNx3759iyVLluDatWvYsGEDRo0apRBTCZIUGhqKqVOnYtKkSfDw8KjU/YrqevPmDczNzXHz5k22UEkKPnz4AIFAgKSkJKnUKy0uLv40cixrQU5mZiYEAkG5C3IEAkGVts9IGxHh7NmzcHZ2Bsdx8PT0hI2NTZnvQdOnT0fedxr4vssIXHmWDgAoEnGffl67Zg0QgL6ttDCrTwu0M5DNze9ynfju3r0Le3t73Lt3j7cY3N3dkZ+fj7Vr11arnatXr2LOnDnQ1tbG77//LpURjLzJz8/HkiVLcOrUKezZswd9+vThJY5169bh4sWLOHPmDPuQImEnT56En58fLl26xHconxQXFyM5ObncBTkZGRnQ0dEpNznq6urKTHL8WH/YxcUFdevWhZeXF6ytrb96fQdeeozVZx6jRs1aKC9xqKgAtWuqYuWg1pjQrZlEY68KuU586enpaNOmDTIyMniLoUuXLli7dq1YFjyIRCIEBATAy8sLv/32G1xdXVmx6//vzp07GD9+PMzNzbFlyxZeyygpQy1DWTFz5kwYGxtj8eLFfIdSKUKhEMnJyeXec0xLS4OWlla55eN0dXWlun+O4zgcPnwY7u7u0NbWhre396cDuEMiX2NV6GMUCLlvtPJ/6qjVwMpBbWQu+cl14iMi1KlTB+/eveNle0BqaipatWqF9PR0sb44U1NT4eTkhAsXLmDt2rUYO3as0o4sSkpKsG7dOmzcuBGbNm3C2LFj+Q4JwNf3OxjxIyI0a9YMZ86cgYmJCd/hiJ1QKERKSkq5yTE1NRWamprlrlZt0qSJ2Kf7RSIR9u/fD3d3dxgbG+O3hW7wjshDgbCk0m3VUVPFIYduaKsvO9Oecp34AKBFixY4c+YMfvjhB6n3/eeff+LkyZM4evSoRNq/ceMGZs+ejQYNGmDLli0KUy+yot68eQM7OzuoqKhgz549EjmCpTqmTZuGOnXqYPPmzXyHopAePXoEGxsbvHz5Umk/+IlEok/JsawEmZKSAg0NjXIX5Ojp6VUpOQqFQuzevRve4alQMWgHqFR+O4iKCjDARAfbJ3Sq9LWSIhsTzNWgr6+PxMREXhJfaGgoBg8eLLH2e/Togdu3byMwMBCWlpYYN24cPDw8lGKEsW/fPixYsACLFy/GokWLZHIVna+vL0xMTGBnZ4dOnWTnj1pRnD59GoMGDVLapAcANWvW/JS8ylJSUlJqcrxz586n75OTk9G4ceNvJsf/btNRU1PD8LF2WBd/CcWiik9xfo4IuPw0HZl5RTKz2lPuR3wTJkzAgAEDMHHiRKn2KxQKoa2tjdjYWOjq6kq8v/T0dKxYsQKnTp2Cj48PJk6cqJCbcbOzszFr1izcu3cP+/btQ4cOHfgOqVx79uzBpk2bEBUVJZPJWZ717dsXS5YskeiHS2VRUlKCtLS0chfkJCcnQ11d/auk+LKWMS5n1EUlbu19pXbNGlhg3RLTe8vGSmi5T3zLli1Dw4YNsXz5cqn2e/XqVSxYsADR0dFS7TcqKgqzZ8/Gd999hy1btsh8YqiM8PBwTJo0CUOGDIGvr69clHUjIvz0008YOXIkHB0d+Q5HYeTk5EBfXx+pqaly8TpQBBzHfZUcExMTcTHfABn1Kn6cV1mGt9eD3+j2Yoi0+hRiqjM2Nlbq/Up6mrMsXbp0QWRkJHbu3ImBAwdi1KhR8PLyEsv5W3wpKiqCq6sr9u7di+DgYPz88898h1RhKioq2L59OywsLDBy5EixnoStzMLCwtCrVy+W9KSoRo0aEAgEEAgEX5xoMvnPW7j0JK3a7ecWCqvdhrjI/VzZx3t80hYaGopBgwZJvV8AUFVVxbRp0/D48WMQEdq0aYMdO3aA46oxF8GT2NhYdOvWDU+f/r/27jyuxvT9A/gnI5Nk7VRKkiVK2lOnwZStQYyZ0RijspShksqWEWNMIvuSw8iWjDBjmRnzo8GMvbSIyhalLEmRrdJ6zrl+f3jVV4O0nHOec073+/WaF6Nz7vuK6nqe67nv676F1NRUhUp6VYyNjeHt7Y3AwECuQ1EaXF1YMq+VlpYiNjYWa9asQWriRYmM2UZNfo41UvjE17lzZ5knvvv37+PRo0eNOudNEjp06IDNmzcjJiYGO3bsgIODAy5dusRpTHVFRBAIBHB0dISvry9+//13hT7yJzg4GMnJyYiJieE6FIUnFos5vbBsaogImZmZ2LNnD/z8/GBrawsej4fAwEDcvXsXtkZ6aPFR4xYYqTVvBmNd+dmTrBSlTlknvpiYGAwbNkxuFjNYW1sjNjYWu3fvxqhRozBq1CgsW7YMPB6P69DeKS8vD5MnT8bTp08RFxfHyYpcSWvZsiU2b94MX19fXLt27Z3HUzF1c+XKFbRr1w7dunXjOhSl9PLlSyQlJSE+Pr76P3V1dfD5fNjb2+Pbb7+FtbV19ddwQXE5+q04BdTaq6V2BMDV+v0rU2VN4e/4tLS08PLlS5SVlclsTnm8Gm3WrBkmTZqEmzdvomXLlujduzd+/vlniET133AqTX/++ScsLS3Rt29fxMbGKkXSq/LZZ5+hb9++CA0N5ToUhSaP31+KSiQS4erVq9i2bRu8vLxgamqKTp06ISQkBIWFhfD09ERqairu37+P3377DbNnz0a/fv1qXLi1U/sIOuKnoAY+SlFRAQb20pKbrQyAEqzqBICuXbvi33//lckVYllZGXR0dJCVlQVNTU2pz9dQaWlp8PPzw6tXryAQCODg4MBpPMXFxZg5cyb+/fdf7NmzB5988gmn8UhLbm4uLCwscPbsWaXsNiILDg4OWLJkCYYMGcJ1KArn8ePHiI+PR0JCAuLj45GUlARdXV3w+fzqOzozM7M6d5rKzMyEu7s7WnQ0Qr7ZtygX1j9dyGPnFoW/4wNelzsfPHggk7nOnTuHPn36yHXSAwBzc3OcPXsWs2bNgqurKzw9PfH4ceNXZjVEYmIirK2tIRQKkZKSorRJDwD09PTw448/wsfHB0pwTSlzBQUFuHHjBgYMGMB1KHKvoqICiYmJCA8Px/jx49GtWzf06tULmzdvRvPmzTFnzhxkZ2fj1q1biIqKgo+PD6ytreuU9IgI27dvh4ODA8aPH48zh6Pwg0tvtFStX8p43avTWK6SHqAEz/gA2S5wUaTVZioqKnBzc8OoUaPw008/wdTUFIsWLYKPj49MusILhUKEhYVBIBBg06ZNcHV1lfqc8sDHxwdRUVHYvXs3Jk6cyHU4CuXvv//GoEGD2EG//0FEuH//fo3ncmlpaTAyMgKfz4ezszMWLVqEnj17NrqxxZMnTzBlyhTcv38fZ86cgampKQBUN5peeiwdZUIRaruuY6czyEBQUBA0NTUxb948qc9lZGSEAwcOwNJSPjZi1sf169cxY8YMPH36FAKBQKpX1VlZWXB3d0erVq2wa9cudOrUSWpzyaPk5GS4uLjg+vXrcl8dkCfjx4/HwIED8d1333EdCqeKi4tx6dKlGmVLIoKDg0N1ydLW1hYaGhoSnbfqrEsPDw+EhIS88wIkLecFNp/JxOlbT6ACoOwd5/EN7KUFX6cecnenV0UpEl94eDgyMjKwceNGqc6TkZEBJycn5OTkKGz/QCLCgQMHMHv2bDg5OWHlypUSbblGRIiKisLcuXMRHByMgIAApWytVhcBAQF49eoVtm/fznUoCkEkEkFbWxupqam19qZUNmKxGLdv365xN5eRkQFzc/PqZ3N8Ph8GBgZS+7nTkLMunxaX4+DlHKQ/KkJhWSXaqKnCWLc1XK3l/wR2pSh16uvr4/Tp01Kfp2q1maImPeB1+XPs2LEYMWIEQkNDYWZmhuDgYMyYMaPRRys9ffoU3t7eSE9Px6lTp5rcaRL/tWTJEvTu3RsXLlxA//79uQ5H7iUkJHywIbMyePbsWfVdXHx8PBITE9G+ffvqBOfp6QkLCwuZlXsvXboEd3d32NraIjU1tc5nXWpqfCw3vTfrjZRAQkIC2djYSH2eoUOH0uHDh6U+jyylp6eTs7Mz9e7dm06dOtXgcU6cOEGdOnWimTNnUmlpqQQjVGwHDhwgU1NTKi8v5zoUubdgwQKaP38+12FIVGVlJV2+fJk2b95MEyZMoJ49e1Lr1q1p0KBBFBwcTEeOHKH8/HxOYhMKhRQaGkpaWlq0b98+TmLgilKUOnNzc2FtbY28vDypzVFcXAxdXV08fPgQbdq0kdo8XCAi/P7775g5cyYcHBywevXqOl91l5WVYf78+Th48CAiIyPZEvT/ICK4uLjA0dFRJs+gFZmVlRU2btyo0HfHubm5NUqWly9fRpcuXWqULHv37s1584vs7Gx4eHigRYsWiIqKQufOnTmNR+a4zbuSIRQKSVVVVapX1X/++ScNGjRIauPLg1evXtHChQtJU1OTli9f/sG/z7S0NDIzM6MxY8ZQQUGBjKJUPHfu3CFNTU3Kzs7mOhS5lZOTQ+3bt6fKykquQ6mzkpISunDhAq1evZq+/vpr6ty5M2lqapKLiwstWbKETp48SS9evOA6zBrEYjHt2rWLeDwerV69mkQiEdchcUIpEh8RkYGBgVR/sEybNo3WrFkjtfHlSUZGBrm4uFDPnj3p+PHjb31cJBLR2rVricfjUWRkJInFYg6iVCxLly4lFxcX9nf1Htu3b6dvvvmG6zDeSywWU2ZmJu3Zs4f8/PzI1taW1NXVydbWlqZPn06//PILZWRkyPW/b0FBAY0ZM4b69OlDqampXIfDKaVJfJ988gmdP39eKmOLxWLS19enmzdvSmV8eXXkyBHq2rUrffXVV3T37l0ien1lPnjwYPrkk0/ozp07HEeoOMrLy8nExIQOHTrEdShy6auvvqKoqCiuw6j24sULOnnyJC1ZsoRcXFyIx+ORvr4+ubq60urVq+nChQtUUlLCdZh1dvz4cfYM/g1Kk/jGjh0rtQe0aWlp1LVrV7m+mpOWkpIS+umnn6hDhw40btw40tLSopCQEIUqScmLs2fPkr6+PhUWFnIdilwpLy+ntm3bcrrIIy0tjbZu3UpeXl5kampKrVq1ogEDBtDcuXPp0KFDlJOTw0lsjVVSUkL+/v6kr69PJ0+e5DocuaEU2xkA6bYtq+rWosjbGBqqZcuWCAwMRFpaGv766y+0b98eNjY2Mun8omw+/fRTDBkyBD/++CPWrl3LdThy48KFC+jVqxe0tbVlMt/jx49rbCeo6mdpb28PPp8PX1/fevWzlFcpKSlwc3NDnz59kJqaqtCHVUua0vz06ty5M7Kzs6Uy9tGjRzF//nypjC3vYmNj4eHhgSFDhiAvLw8XLlzAjBkzsGXLFqxfv54dHVNPq1atgqmpKTw8PGBlZcV1OHJBmqcxVFRUICUlpTrJJSQk4OnTp9VJbs6cObCzs1Oq7joikQhr167FypUrsW7dOri5uTXJi/ZacX3LKSkHDhygr776SuLjPnv2jFq3bq1Q9XxJqKiooIULF5KOjg798ccfNT5WVlZGy5Ytow4dOtCiRYua3N9NY+3YsYPs7OxIKBRyHYpcMDExocTExEaPIxaL6e7du7R//36aOXMmOTg4kLq6OllYWNC0adMoMjKSbty4odQrGe/du0eOjo40YMAAtoq4FkqT+C5evEh2dnYSH/fXX38lFxcXiY8rz27fvk12dnY0bNgwevTo0Xtfd+/ePXJ1dSVDQ0P6448/muQz0IYQiUTUv39/2rx5M9ehcC4rK4u0tbUblIyKi4vpzJkztHz5cvriiy+oY8eOpKOjQ6NHj6awsDA6ffo0FRUVSSFq+RQdHU1aWloUFhbGLqo+QGkS34MHD0hPT0/i406YMIE2bdok8XHlkVgspq1btxKPx6ONGzfWOZGdPHmSjI2NadiwYXT79m0pR6kcrl27Rjwer9YLi6ZAIBDQxIkTP/g6kUhEN2/epMjISJo2bRpZWlqSuro68fl8CgwMpP3799Pdu3eb5MXXs2fPaNy4cWRsbEzJyclch6MQlKJzC/D6CBx1dXW8evVKYg+lxWIxOnbsiISEBHTt2lUiY8qrN48iiY6OrvchqhUVFQgPD8fy5csxbdo0BAcHo1WrVlKKVjnMnz8f9+7dw969e7kOhTMuLi6YOHEixo4dW+PPnz17hsTExBrP5t7sZ2lvbw9LS8smf3zR6dOnMXHiRIwePRorVqyAuro61yEpBq4zryTp6+vTvXv3JDZeYmIimZiYSGw8eXXs2DHS1dWloKCgRne/ycnJoW+//ZYMDAzowIEDTfIKvK5evXpFhoaGdOLECa5D4URJSQlpaGjQ48eP39nPcuDAgTR//nz6888/KS8vj+tw5UpZWRnNnj2b9PT0KCYmhutwFI7SrOoEXm9pyMnJgYGBgUTGU6RDZxuipKQEQUFB+Ouvv7B37144OTk1esxOnTph7969OHPmDPz8/BAREYHw8HCYmJg0PmAlo66uDoFAAF9fX1y9ehVqampchyQTVf0s9+7di2bNmqFbt24wMDAAn89H//79MWfOHLnoZymvrl69Cnd3d3Tr1g2pqang8Xhch6RwlOqgtKrEJylHjx6V2jJrrl25cgW2trZ49uwZUlNTJZL03uTk5IQrV67AxcUFAwYMQFBQEIqKiiQ6hzJwcXGBubk5li9fznUoUlFaWoq4uDisXbsWY8eOhYGBAczNzbFz507k5OTA1dUVDx48wPXr17Fjxw589913MDMzY0nvHcRiMdavX49BgwYhICAAhw8fZkmvobi+5ZSkwMBAifXTzMvLo7Zt2yrdcTJCoZCWL19OWlpaFB0dLZM5Hz16RBMmTKBOnTrR3r17WfnzPx48eECampqUnp7OdSiN8q5+li1btiQbG5u3+lmKxWLq2rUrpaWlcR22Qnjw4AENHjyYHBwcKDMzk+twFJ7SlTol1b3l+PHjGDJkCFq0aCGR8eTB/fv3MWHCBBARkpKS0KVLF5nM27FjR0RFRSE2Nra6/CkQCNCnTx+ZzC/v9PX1sXDhQvj6+uKff/5RmM3GhYWFSExMrNEFRU1NrXoByrhx42BtbY2WLVu+9d709HRUVlayr4E6+O233zBjxgz4+flh/vz5rGuSJHCdeSVp//795OrqKpGxxo4dSzt27JDIWPJAXvb4CIVCEggExOPxKDAwUO6ObeFKZWUlWVpa0p49e7gO5Z2EQiFdvXqVtm3b9lY/yzlz5tDBgwfpwYMHdR5vzZo1NHXqVClGrPhevHhBHh4eZGRkRAkJCVyHo1SUKvHFxsYSn89v9DiVlZXUvn17ys3NlUBU3Hr+/DmNHz9e7vb4PH78mLy8vEhXV5eioqJY+ZOIEhISqGPHjvTs2TOuQ6H8/Hw6cuQIBQcH0+DBg6lNmzZkZGREHh4etGnTJkpOTqaKiooGjz948OC3OgIx/3Pu3DkyNDSkadOmUXFxMdfhKB2l2ccHvC7l9evXr9HlzvPnzyMwMBDJyckSiowbZ8+excSJE+Hi4oJVq1bJ5R6fhIQETJ8+HWpqahAIBLC0tOQ6JE5Nnz4dIpEIW7ZskdmcVf0s3yxZvtnP0t7eHvb29hLrZ1lUVAQ9PT08evQIGhoaEhlTWVRUVODHH3/Erl27sHXrVowaNYrrkJQT15lXkioqKkhVVbXRR+bMmzePFi5cKKGoZK+8vJzmzZtHurq6dPToUa7D+SChUEgRERGkra1N06dPl4s7Hq48f/6cdHV1KS4uTirji8ViunfvHv36669v9bOcOnUq7dy5U+r9LA8fPkxDhw6V2viK6saNG2RlZUUjR45k+xalTKkSHxGRrq5uo8/OMjMzk9oPHmmr+uYZNWoUZ+ebNVRBQQF5e3uTjo4O7dixQ6mbCddm7969ZG5uLpEzD//bz1JXV5fzfpZTpkyhdevWyXROeSYWi0kgEJCmpiZt2bKFlf1lQOkSX9++fenixYsNfv/9+/dJU1NT4Zq8Vn3z8Hg8ioiIUOhvnkuXLpG9vT3Z29tTUlIS1+HInFgspqFDh9Lq1avr9T6RSETp6em0a9cu8vb2fquf5b59+yg7O5vTrw2xWEx6enp069YtzmKQJ7m5uTRs2DDq27cv+zuRIaVbF9u5c+dGbWKPiYnBsGHDFGoDbV5eHjw9PfHkyRPExsaiZ8+eXIfUKDY2NoiLi0NUVBRGjhyJL774AkuXLlWqM9Nqo6Kigk2bNsHBwQFff/31ezsRvaufZbt27aq3E0yaNEnu+lmmpaWhZcuWMDIy4joUzv3+++/w8fHB1KlT8cMPPyj8wbeKRKk6twCN796iaN1ajhw5AktLS1hbWyMuLk7hk16VZs2aYfLkybh58yZUVVXRu3dvREREQCQScR2aTBgZGcHf3x8BAQEAXjdhv3LlCn7++WdMmjQJxsbGMDQ0xMqVK1FRUQFfX1+kp6cjOzsb+/btQ0BAAOzt7eUq6QH/O3RWUfYqSkNRURG8vLwwZ84cHD58GCEhISzpyRrXt5yStnLlSpo9e3aD3ltWVkZt2rShgoICCUclecXFxTR16lQyNDSk8+fPcx2O1KWkpFD//v3JxsamUaVsRfHw4UP69ddfqX379mRqakoaGhrUu3dv8vT0pIiICEpNTVW4cjwRUb9+/ejvv//mOgzOxMXFUffu3cnT05MKCwu5DqfJUrpSp76+foO3IZw7dw59+vSR+5JaUlIS3Nzc4ODggNTUVLRp04brkKTOwsIC586dQ3R0NL766isMHz4cYWFh0NbW5jq0RisrK8Ply5erS5bx8fF49eoV+Hw+Ro8ejWPHjiE9PR2dOnXiOtRGefbsGdLS0uDo6Mh1KDJXWVmJJUuWYOvWrfj555/x5Zdfch1Sk6aUpc6G7uOT9zKnUChEaGgoRo4cidDQUERFRTWJpFdFRUUF7u7uSE9PR9u2bWFqagqBQAChUMh1aHVGRLhz5w727t0Lf39/2NnZQVNTE/7+/sjKysLnn3+OU6dOoaCgAEePHkVkZCScnZ2xYcMGrkNvtBMnTsDR0bHJnEJR5fbt2+jXrx8SExNx5coVlvTkAde3nJKWlZVFBgYGDXqvkZERXb58WcIRSUZWVhb169ePBg8eXK/WUMrs2rVr5OTkRBYWFnJb7n358iX9888/FBoaSiNHjiQej0f6+vo0ZswYWrVqFZ0/f55evXpV6xh5eXmkpaWl8A2dPTw8aPPmzVyHITNisZgiIiKIx+PRxo0bFXqltbJRusRXXl5Oqqqq9X7+cfv2bdLV1ZW7L06xWEy7du0iHo9Ha9asabJ7295HLBbTvn37qFOnTuTh4cFpm7n/9rPs06cPtWrVivr379+gfpZvioiIIAcHB4X99xeJRKSlpUV3797lOhSZyM/Pp5EjR5KVlRXduHGD63CY/1C6xEdEpKOjU+8fgOvXrycvLy8pRdQwT58+pa+//ppMTU0pJSWF63DkWlFREQUFBZGmpiatXbu2UX0k6+rx48f0119/0YIFC97qZykQCOjSpUsSi0MkEhGfz6etW7dKZDxZi4+PJ1NTU67DkIm//vqLOnbsSN9//73SHWumLJQy8dnY2FBiYmK93uPs7EyHDh2SUkT1d/LkSdLX16eAgAAqLS3lOhyFcfPmTRo6dCiZmprS6dOnJTZueXk5JSYmUnh4OI0fP566detGbdu2paFDh9IPP/xAR48epSdPnkhsvndJSUkhLS0thevIQ0S0aNEimjt3LtdhSFVxcTFNmzaNunTpQmfPnuU6HKYWSpn4Ro8eTYcPH67z64uLi0lDQ4NevnwpxajqprS0lGbFjSRXAAAY20lEQVTNmkWdOnWi48ePcx2OQhKLxXTw4EEyMDCgcePG1buF3X/7WX7yySekrq5O5ubmMutn+T6zZ8+mCRMmyHzexrK1tZXohYi8SUxMpJ49e5KHhwc7aksBKN12BqD+KztPnToFOzs7zldIXrt2DePHj4eRkRFSU1PlfluFvFJRUcGYMWMwbNgwhIWFwcLCAkFBQQgMDHznwcKvXr1CcnJyje0EIpEIDg4O4PP5CA0Nha2tLVq3bs3BZ1PT4sWLYWpqitOnT2PgwIFch1Mn+fn5yMjIQL9+/bgOReKEQiHCwsKwceNGCAQCjB07luuQmDpQysRX37ZlXG9jEIvFCA8Px9KlS7Fy5UpMmjSpSXe2kJRWrVohNDQUEydOREBAACIjI7FhwwZ06dKlRpK7ffs2zMzMwOfzMXbsWKxduxZdunSRy38DDQ0NhIeHw8fHB6mpqXLXmeVd/v77bwwZMkTpupPcuXMHHh4eUFdXx+XLl6Gvr891SEwdKWXi09fXR2pqap1eS0Q4duwYjh8/LuWo3i03NxeTJk1CcXEx4uPj0b17d07iUFbPnj3DnTt3YGdnh/z8fAwfPhxqamoYMmQIBg0aJJf9LD9k9OjR2LlzJ1atWoWFCxdyHc4HHT16FC4uLlyHITFEhMjISMybNw/BwcEICAhAs2ZKtyVaqSnlv1Z9+nVev34dzZs3h7GxsZSjetuhQ4dgZWWF/v3749y5cyzpNZJQKERKSgq2bNnyVj/L8vJyLFq0CJmZmQgKCsKFCxdQXFyscEmvSnh4ONavX4/MzEyuQ6lVZWUlTp48iWHDhnEdikQUFBRgzJgxWL9+PU6dOoWZM2eypKeIuH7IKA2ZmZnUtWvXOr12+fLlNH36dClHVFNhYSFNnjyZunfv3iT6TkpLbm4uHT58mIKCgujTTz8lDQ0NMjExocmTJ3+wn2VWVhZ9/vnn1KNHDzp27JiMI5eMVatWkbOzs9ztPX3T2bNnydramuswJOLvv/8mPT09mjNnDpWVlXEdDtMISpn4SktLqUWLFnVadffpp5/K9JTy2NhY6tatG3l5ecn8AFBFVlpaSrGxsbR27VoaO3YsGRgYUIcOHWjEiBEUEhJCJ06coOfPn9d73KNHj1KPHj1o9OjRlJWVJYXIpaeiooLMzMxo//79XIfyXvPmzaOFCxdyHUajlJSUkJ+fH3Xu3Jn+/fdfrsNhJEApEx8RkZaWFuXl5dX6mufPn1Pr1q2ppKRE6vFUVFTQokWLSEdHp15bLZoisVhMd+7coejoaJoxYwb17duX1NXVycbGhnx9fWn37t10+/Ztid3plJaW0tKlS0lTU5MWL14sk68HSYmNjSU9PT25XUJvZmZGcXFxXIfRYMnJyWRiYkLjxo2jZ8+ecR0OIyEqRERcl1ulwdraGtu2bYONjc17X/Pbb78hKioKR48elWosGRkZ8PDwQLt27RAZGQldXV2pzqdoCgsLkZSUVH2Yanx8PFRVVau3E/D5fFhbW0NdXV2qcdy/fx+zZs3C5cuXsX79eowaNUouV3b+17Rp06CqqgqBQMB1KDXcv38fNjY2yMvLU6iDnQFAJBJh1apVWLNmDTZs2IDx48dzHRIjSVxnXmkZNWoU/fHHH7W+ZuLEiSQQCKQWg1gspm3bthGPx6Pw8HCF7bMoSVX9LLdv305Tpkx5q5/lgQMHOG/CfeLECerVqxeNGDGCMjIyOI2lLp4+fUodO3asd7ciaduyZQu5ublxHUa9ZWdn04ABA8jR0bHJ9BZtapQ28fn4+NSa1EQiEWlra0vtuc6TJ0/oiy++IAsLC7p27ZpU5lAE7+tn6e7uLvF+lpJUXl5OK1asIE1NTVqwYMEHT1Dg2u7du8nKyooqKyu5DqXa559/TtHR0VyHUWdisZh2795NPB6PVqxYoZAH/TJ1o7SlzmXLlqGwsBDLly9/58eTkpIwceJE3LhxQ+Jz//333/Dy8oKbmxuWLFmikMvlG6KiogKpqak1SpYFBQWws7OrLlna2dmBx+NxHWqdPXz4EHPmzEFcXBzWrl2Lr776Si7Ln0SEwYMH44svvoC/vz/X4aC8vBxaWlrIzs5WiA5Ez549g4+PD65du4bo6GhYWlpyHRIjTRwnXqmJioqqtcyyePFimj17tkTnLCkpoRkzZlDnzp3p1KlTEh1b3ojFYrp//z799ttvNGvWrLf6We7YsYOuX7+uNOXd06dPk6mpKQ0dOpTS09O5Dued0tPTicfj1bs3qTScOHGCHBwcuA6jTv755x/S19cnf39/hVrYxDScUnZuAT7ctuzYsWMICwuT2HwpKSkYP348zM3NkZqaivbt20tsbHnwvn6WVXdy8tTPUhqcnJxw5coVCAQC9OvXD15eXvjhhx+goaHBdWjVevXqBR8fHwQGBuLAgQOcxqII3VrKysoQHByM3377DTt37oSzszPXITGywnXmlZbbt29T9+7d3/mx/Px8atu2rUTOyhKJRLRy5UrS0tKiPXv2yPVm4roSi8WUnp5OUVFR5OPjQ1ZWVqSurk729vYUEBBA+/bto6ysLKX4XBvi0aNH5OHhQfr6+rRv3z65+nsoKSmh7t27y3Rv6rsYGRnR5cuXOY2hNqmpqdSnTx8aM2YMFRQUcB0OI2NK+4yvpKQEmpqaKCkpeeuZzO7du/Hnn3/i0KFDjZrjwYMHmDBhAkQiEXbv3g1DQ8NGjceV58+fIzExsfpOLiEhAW3btgWfz4e9vT34fD4sLS2hpqbGdahy5cKFC5g+fTo0NTWxceNGmJqach0SAODEiRPw9vbGtWvXpL4F5F0yMjLg6OiIhw8fyt3zULFYjPXr1yMsLAyrV6/GhAkT5C5GRvqUttSprq4OdXV1PH369K3FFMeOHWv0aQz79++Hv78/Zs6ciaCgIIXZpyQUCnHt2rUaJcuHDx/C1tYWfD4f3t7eiIyMRMeOHbkOVe71798fycnJ2LJlC5ycnODh4YHFixdzfryVs7Mz7O3tERoaimXLlsl8/piYGAwfPlzuEsqDBw8wadIklJeXIyEhAd26deM6JIYrXN9ySpO5uflb5ZbKykpq3749PXz4sEFjvnjxgtzc3KhXr1506dIlSYQpVbm5ufT777/TvHnzyNHR8Z39LOVpCbyiys/PJ09PT9LT06NffvmF8/Jnbm4u8Xg8TrbSfPbZZ3Tw4EGZz1ubffv2kZaWFi1dupRtU2CUdx8fEZGLiwsdOXKkxp+dO3eOrKysGjTe2bNnqUuXLuTj4yOX+7pKS0spLi7urX6Ww4cPp59++omOHz/eoH6WTN1dvHiRrK2tqX///pSSksJpLAKBgAYMGCDTJFxcXEwaGhr08uVLmc1Zm+fPn5Obmxv17NmTkpKSuA6HkRNKfZ7Gu44nakiZs6KiAsHBwfjmm28gEAiwefNmTp6dvImIkJWVhb179yIgIAD29vbQ1NSEn58fMjMzMXLkSPzzzz8oKCjAsWPHsGjRIjg7O6Ndu3acxq3s+Hw+EhMT4e7uDmdnZ/j7++PFixecxOLt7Y2ysjLs2rVLZnOeOnUKffv25bzcCwBnz56FpaUl2rRpg8uXL8PW1pbrkBg5obTP+ID3J76ff/65zmOkp6fDzc0Nenp6SElJgY6OjqTDrJOioqLqfpZV/73Zz9LV1RU2NjacJ2QG+OijjzBt2jS4uroiODgYJiYmWLZsGSZOnCjTs9s++ugjbNmyBcOHD8eoUaNk0jhAEs/PG6vq7MVffvkF27Ztk/ttFYzsKe2qTgDYtWsXTp8+jaioKACvH25bWVkhPz//g4tRiAhbtmzBokWLEBoaiqlTp8rsYb1YLMbNmzdrJLns7GxYWlpW75vj8/nQ19eXSTxM41y6dAl+fn5QUVGBQCCotXG6NAQGBqKoqAg7duyQ6jxEBENDQ8TExKB3795Snet9rl+/Dnd3dxgYGGDbtm3Q1tbmJA5Gvin9Hd+DBw+q/z8mJgafffbZB5Nefn4+vLy8kJeXhwsXLqBXr15SjfPJkyfVLb4SEhKQmJgIbW3t6gTn7e0Nc3NzqKqqSjUORjpsbW0RFxeHXbt2wcXFBV9++SWWLl2KDh06yGT+kJAQmJqa4vz58xgwYIDU5rlx4wZUVFRgYmIitTneRywWQyAQICQkBGFhYZgyZYrcrSpl5IfSJr6C4nLEPW+F7I6O8IxKQhu15kiIz8Fk59rLHn/99RemTp0KT09PHD58GC1atJBoXBUVFUhLS6txN1fVz9Le3h6BgYGwt7dXqH6WzIc1a9YMnp6e+PLLL7Fo0SKYmJhgyZIl8PLykvpWmDZt2mD9+vXw9vbGlStXJP41XaWqW4usE05ubi4mT56Mly9f4uLFizAyMpLp/IziUbpSZ+qDF9h0JhNnbz8BQCgX/u/To8pyfKymhoHG2vB17AGLzv9b6PHq1SvMnj0bx48fx+7duyVyZUxEyMnJqZHkUlNT0a1btxolS2NjY5k++2G4l5KSAj8/P5SXl0MgEMDe3l6q8xERRo4ciQEDBuD777+XyhxOTk6YO3euTJ+pHTp0CL6+vvD19cWCBQvQvLnSXsszEqRUiW9P/F0sPZaOMqEItX1WKiqAWvOPsGCEMdz5hrh06RLc3Nxgb2+PjRs3om3btg2av6Sk5K1+lkKhsEaSU+Z+lkz9EBF++eUXfP/99xgxYgTCwsKgpaUltfmys7PRt29fJCUloWvXrhId++XLl9DX10d+fr5MFlgVFhYiICAA58+fx549e8Dn86U+J6M8lCbxvU56N1FaKa7ze1qqNoO1yj2c/HkRNm7ciG+++abO7yUiZGRk1Ehyt27dgpmZWY1WX4aGhuxZA1Orly9fYvHixYiOjsaPP/4Ib29vqZU/w8LCcOHCBfzf//2fRL8uDx48iB07diAmJkZiY75PbGwsPDw8MHjwYKxbt06uGoUzikEpEl/qgxcYty0epZWier9XRVSJrd+YYKhN7QtY/tvPMjExEa1bt65xN8f6WTKNce3aNfj5+eHly5fVp0BIWkVFBaysrBASEoIxY8ZIbFxPT09YWVlhxowZEhvzvyorK/HTTz9h+/btiIiIwOjRo6U2F6PclCLxTf3lEk7ezK+1vPk+KirAZ711sMX9f5tbhUIhrl+/XuNuLicnp7qfZdUdHetnyUgaEWH//v2YO3cuBg8ejJUrV0p87+i5c+fg5uaGGzduSKTsLhaLoaenh7i4OKn1v7x16xbc3d2hpaWFnTt3su89plEUPvEVFJej34pTKBfWvcT5Xy0+UsEiiwpcS36d5JKTk6Gvr1/jbs7U1JQ9OGdkpqioCEuWLEFkZCQWLlyI6dOnS/Trz9PTE23btsW6desaPVZycjLc3NyQnp4ugchqqtpP+8MPPyAkJAQ+Pj7s0QHTaAqf+LacvYN1/9xuVOIjYTl0Hl/CyO5q4PP56Nu3r9IdJMsopvT0dMyYMQN5eXkQCARwdHSUyLgFBQUwNTVFTEwMrK2tGzXWkiVL8Pz5c6xdu1YisVV5cz/tnj17YGxsLNHxmaZL4dfQp+cVNirpAYBK84/Rb8TY6n6WLOkx8sLY2BgnTpzAjz/+CA8PD4wfPx65ubmNHpfH42H58uXw9vaGSFT/Z+NvkkabsiNHjsDS0hKWlpaIi4tjSY+RKIVPfIVlQomMcyntOiIiIvDHH3/g4sWLuHPnDoqLi6HgN8SMElBRUYGrqytu3ryJrl27wtzcHKtXr0ZFRUWjxp04cSLU1NQQERHR4DEKCgpw48YNiXWEKS4uxtSpUxEQEIADBw4gNDRUahvumaZL4Uudgb9ewR8pjb8C7oon6FEQi/z8fDx+/Bj5+fnIz88HAOjo6EBHRwfa2trv/LXq9x06dGAb0Rmpy8jIgL+/P+7evQuBQIDBgwc3eKzr16/DyckJaWlp0NXVrff79+zZg0OHDuH3339vcAxVEhIS4O7ujn79+iE8PFwuTnhglJPCJz5JPONTa94MM4f2xLRPu7/1seLi4hqJsOr37/qzoqIi8Hi8OiVKLS0tdiXLNBgR4ciRIwgMDETfvn2xZs0adO7cuUFjBQcHIzs7G/v27av3e8ePH4+BAwfiu+++a9DcwOtV1EuXLsXmzZuxadMmuLq6NngshqkLhU98kljV+XHzZoibNwiaGh83KpaKigo8efLknQnyv4nyyZMnaN269Vt3je+6k9TR0UGrVq3YajbmLaWlpVixYgUEAgFmz56NWbNm4eOP6/d1XFJSAlNTU0RERMDZ2bnO7xOJRNDW1kZqamqDTwrJzMyEu7s72rRpg8jISHTq1KlB4zBMfSh84gMkv49PFsRiMZ4/f16nO8k3S651SZSs5Nr0ZGVlITAwEOnp6QgPD8ewYcPq9f5jx47B398fV69eRcuWLev0nri4OPj4+CA1NbXe8RIRduzYgfnz52PhwoWYMWMG+5plZEYpEl9jOre0VP0Iv07lw1xfvk8mZyVXpi6OHj2KgIAAmJmZYd26dTA0NKzze7/++muYmJggJCSkTq9fuHAhRCIRwsLC6hXjkydP8N133+Hu3buIjo6Gqalpvd7PMI2lFIkPaHivzgUjTODON5ReYByoKrnWJVHWpeT65sdYyVX+lZWVYc2aNVi3bh38/f0RFBRUp1Z6Dx8+hIWFBS5cuFCn7QNWVlbYuHEj+vfvX+fYYmJi4OXlBQ8PD4SEhNS7LMswkqA0iQ9o+OkMTdmbJdcPJcq6llyrfmUlV27du3cPs2bNQkpKCtavX49Ro0Z98D0bNmzAn3/+iX///bfWC5yHDx/CzMwMjx8/rlNHmZKSEsydOxf/93//h6ioKDg5OdXnU2EYiVKqxAcAaTkvsPlMJk7fegIVAGVvLHpRa94MBGBgLy34OvWQ+/KmPPpQyfXNj71Zcv3Qs0lWcpWeEydOwN/fHz169MCGDRvQvfvbq5erCIVC2NnZYdasWXB3d3/v63bs2IGTJ09i//79H5y/qqWZjY0NNm3ahHbt2Pcdwy2lS3xVnhaX4+DlHKQ/KkJhWSXaqKnCWLc1XK31G716k6mbupRcq379b8n1Q4mSlVzrp6KiAuvWrcOqVavg4+OD+fPnv/fcvKSkJHz++ee4fv06OnTo8M7XjBkzBqNHj8aECRPeO6dIJMKKFSuwfv16bNiwAd9++61EPheGaSylTXyMYmlsybW2RMlKrv+Tk5ODOXPmID4+HuvWrcMXX3zxzgsIPz8/VFZWIiIiAgXF5TiYnIP0vEIUlgmh0aIZfo1Yi3O7VqKX4bu3H2RnZ8PDwwOqqqqIioqCgYGBtD81hqkzlvgYhfRmyfVDibKuJVdtbW1oa2s3iZLrqVOnMGPGDOjr6yM8PBy9etU8j/Lly5fo3X84HKYsxtWC16ula+yVFVXi448/hlMvLfg69oBF59flSyLC7t27MWfOHHz//feYOXMmu+hg5A5LfIzS+2/JtbZEWZ+Sq7a2NjQ0NBS25FpZWYmNGzdi2bJlmDJlChYuXFh9mvme+Lv46cg1VIoIqCVxvblQbLhRa3h7e+PmzZuIjo6GhYWFrD4VhqkXlvgY5g3vKrnWliiJSOFLro8ePUJQUBDOnDmDNWvWoLyzHZbF1G9rUItmQGlcNL626ohly5bVafsEw3CFJT6GaYS6llwfP36MwsLCt0qu70uUXJRcz58/D+8FYSjhTwV9pFrv97doBhz06cdWSzNyjyU+hpERSZRc35coJVVy/W53Ek7ezAdQ/7G4av/HMPXFEh/DyCEuSq7y1PCdYaSJJT6GUQLvKrm+L1G+r+T6sE1vXCrXgZAafudY2xFfDCMvPtxriGEYuaehoQENDQ1069btg699X8k1qQAQNmtcubRMKEb6o6JGjcEw0sYSH8M0MS1atECnTp3eOvvuZlQS7qc/bvT4hWWVjR6DYaRJ/tZWMwzDiTZqkrkObqNW/xWhDCNLLPExDAMAMO7YBh83b9yPBLXmzWCs21pCETGMdLDExzAMAMDVRr/RYxAAV+vGj8Mw0sQSH8MwAACexsdw7KmFhm4HVFF5feQX28rAyDuW+BiGqTbdqQfUmn/UoPeqNf8Ivk49JBwRw0geS3wMw1Sz6NwOC0YYo6Vq/X40tFRthgUjjFm7MkYhsO0MDMPU4M43BAAsPZaOMqEItbW4ePN0hqr3MYy8Y51bGIZ5p7ScF9h8JhOnbz2BCl5vTq+i1rwZCK+f6fk69WB3eoxCYYmPYZhaPS0ux8HLOUh/VITCskq0UVOFsW5ruFrrs4UsjEJiiY9hGIZpUtjiFoZhGKZJYYmPYRiGaVJY4mMYhmGaFJb4GIZhmCaFJT6GYRimSWGJj2EYhmlSWOJjGIZhmhSW+BiGYZgmhSU+hmEYpklhiY9hGIZpUljiYxiGYZoUlvgYhmGYJoUlPoZhGKZJYYmPYRiGaVJY4mMYhmGaFJb4GIZhmCaFJT6GYRimSWGJj2EYhmlSWOJjGIZhmhSW+BiGYZgmhSU+hmEYpkn5f+++Zt8tkon8AAAAAElFTkSuQmCC\n",
+            "text/plain": [
+              "<Figure size 432x288 with 1 Axes>"
+            ]
+          },
+          "metadata": {}
+        }
+      ],
+      "source": [
+        "# python -m graphsage.supervised_train --train_prefix ./example_data/toy-ppi --model graphsage_mean --sigmoid\n",
+        "# test networkx and visualization\n",
+        "import networkx as nx\n",
+        "import tensorflow as tf\n",
+        "tf.compat.v1.disable_eager_execution()\n",
+        "\n",
+        "G = nx.complete_graph(6)\n",
+        "nx.draw(G)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "# download code and data\n",
+        "!git clone https://github.com/williamleif/GraphSAGE"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "S6709xbNrBok",
+        "outputId": "51e908ea-9105-4844-9427-b57a417bf9da"
+      },
+      "execution_count": null,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Cloning into 'GraphSAGE'...\n",
+            "remote: Enumerating objects: 265, done.\u001b[K\n",
+            "remote: Counting objects: 100% (7/7), done.\u001b[K\n",
+            "remote: Compressing objects: 100% (7/7), done.\u001b[K\n",
+            "remote: Total 265 (delta 3), reused 0 (delta 0), pack-reused 258\u001b[K\n",
+            "Receiving objects: 100% (265/265), 6.43 MiB | 11.28 MiB/s, done.\n",
+            "Resolving deltas: 100% (160/160), done.\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "import json\n",
+        "from networkx.readwrite import json_graph\n",
+        "import os\n",
+        "import numpy as np\n",
+        "import sys\n",
+        "\n",
+        "CODE_ROOT = \"GraphSAGE/graphsage\"\n",
+        "sys.path.append(\"GraphSAGE\")\n",
+        "\n",
+        "def load_data():\n",
+        "    data_path = 'GraphSAGE/example_data'\n",
+        "    # DATA 1, 14755 nodes, 228431 links\n",
+        "    G_data = json.load(open(data_path + '/toy-ppi-G.json'))\n",
+        "    #G_data['nodes'] = G_data['nodes'][:100]\n",
+        "    #G_data['links'] = G_data['links'][:100]\n",
+        "    G = json_graph.node_link_graph(G_data)\n",
+        "    \n",
+        "    conversion = lambda n : n\n",
+        "    lab_conversion = lambda n : n\n",
+        "    \n",
+        "    # DATA 2, (14755, 50) dtype('float64')\n",
+        "    feats = np.load(data_path + '/toy-ppi-feats.npy')\n",
+        "    \n",
+        "    # DATA 3, {\"0\": 0, \"1\": 1}, len: 14755\n",
+        "    # node ids to integer values indexing feature tensor\n",
+        "    # 其实没什么用\n",
+        "    id_map = json.load(open(data_path + \"/toy-ppi-id_map.json\"))\n",
+        "    \n",
+        "    # DATA 4, dict, len: 14755, column 121\n",
+        "    # from node ids to class values (integer or list)\n",
+        "    # 分类标签\n",
+        "    class_map = json.load(open(data_path + \"/toy-ppi-class_map.json\"))\n",
+        "    \n",
+        "    broken_count = 0\n",
+        "    for node in G.nodes():\n",
+        "      if not 'val' in G.nodes()[node] or not 'test' in G.nodes()[node]:\n",
+        "        G.remove_node(node)\n",
+        "        broken_count += 1\n",
+        "    print(\"Removed {:d} nodes that lacked proper annotations due to networkx versioning issues\".format(broken_count))\n",
+        "    \n",
+        "    # edge: (0, 800) 边\n",
+        "    # G[0]: 某结点与所有的关联结点组成的边的集合\n",
+        "    # 标记需要在训练中移除的关联关系，即边\n",
+        "    for edge in G.edges():\n",
+        "      if (G.nodes()[edge[0]]['val'] or G.nodes()[edge[1]]['val'] or\n",
+        "          G.nodes()[edge[0]]['test'] or G.nodes()[edge[1]]['test']):\n",
+        "        G[edge[0]][edge[1]]['train_removed'] = True\n",
+        "      else:\n",
+        "        G[edge[0]][edge[1]]['train_removed'] = False\n",
+        "    \n",
+        "    from sklearn.preprocessing import StandardScaler\n",
+        "    \n",
+        "    # 训练集的id集合，result only int, len: 9716\n",
+        "    train_ids = np.array([id_map[str(n)] for n in G.nodes() \\\n",
+        "                          if not G.nodes()[n]['val'] and not G.nodes()[n]['test']])\n",
+        "    \n",
+        "    train_feats = feats[train_ids]\n",
+        "    \n",
+        "    # 特征缩放，标准化：z = (x - u) / s\n",
+        "    # u is the mean of the training samples\n",
+        "    # s is the standard deviation of the training samples\n",
+        "    scaler = StandardScaler()\n",
+        "    scaler.fit(train_feats)\n",
+        "    feats = scaler.transform(feats)\n",
+        "\n",
+        "    walks = []\n",
+        "\n",
+        "    return G, feats, id_map, walks, class_map"
+      ],
+      "metadata": {
+        "id": "jg81uw5O6Gaz"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "def construct_placeholders(num_classes):\n",
+        "  # Define placeholders\n",
+        "  placeholders = {\n",
+        "      'labels' : tf.compat.v1.placeholder(tf.float32, shape=(None, num_classes), name='labels'),\n",
+        "      'dropout': tf.compat.v1.placeholder_with_default(0., shape=(), name='dropout'),\n",
+        "      'batch' : tf.compat.v1.placeholder(tf.int32, shape=(None), name='batch1'),\n",
+        "      'batch_size' : tf.compat.v1.placeholder(tf.int32, name='batch_size'),\n",
+        "  }\n",
+        "  return placeholders"
+      ],
+      "metadata": {
+        "id": "mReVSrV9UzYO"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "train_data = load_data()"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "Rv1F4lYF_FfW",
+        "outputId": "7b9e88c6-8ba5-4128-9564-55a7d4cc835c"
+      },
+      "execution_count": null,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Removed 0 nodes that lacked proper annotations due to networkx versioning issues\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "G = train_data[0]\n",
+        "features = train_data[1]\n",
+        "id_map = train_data[2]\n",
+        "context_pairs = train_data[3]\n",
+        "class_map  = train_data[4]\n",
+        "\n",
+        "# num_classes = 121\n",
+        "num_classes = len(list(class_map.values())[0])\n",
+        "# pad with dummy zero vector, row wise\n",
+        "features = np.vstack([features, np.zeros((features.shape[1],))])\n",
+        "placeholders = construct_placeholders(num_classes)"
+      ],
+      "metadata": {
+        "id": "CjSUlil1kNZP"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "class NodeMinibatchIterator(object):\n",
+        "\n",
+        "    \"\"\"\n",
+        "    This minibatch iterator iterates over nodes for supervised learning.\n",
+        "\n",
+        "    G -- networkx graph\n",
+        "    id2idx -- dict mapping node ids to integer values indexing feature tensor\n",
+        "    placeholders -- standard tensorflow placeholders object for feeding\n",
+        "    label_map -- map from node ids to class values (integer or list)\n",
+        "    num_classes -- number of output classes\n",
+        "    batch_size -- size of the minibatches\n",
+        "    max_degree -- maximum size of the downsampled adjacency lists\n",
+        "    以toy-ppi数据集举例：\n",
+        "    label_map为输出，维度为(14755, 121)\n",
+        "    num_class为label_map的第二维，即121\n",
+        "    \"\"\"\n",
+        "    def __init__(self, G, id2idx,\n",
+        "            placeholders, label_map, num_classes,\n",
+        "            batch_size=100, max_degree=25,\n",
+        "            **kwargs):\n",
+        "\n",
+        "        self.G = G\n",
+        "        self.nodes = G.nodes()\n",
+        "        self.id2idx = id2idx\n",
+        "        self.placeholders = placeholders\n",
+        "        self.batch_size = batch_size\n",
+        "        self.max_degree = max_degree\n",
+        "        self.batch_num = 0\n",
+        "        self.label_map = label_map\n",
+        "        self.num_classes = num_classes\n",
+        "\n",
+        "        self.adj, self.deg = self.construct_adj()\n",
+        "        self.test_adj = self.construct_test_adj()\n",
+        "\n",
+        "        self.val_nodes = [n for n in self.G.nodes() if self.G.nodes()[n]['val']]\n",
+        "        self.test_nodes = [n for n in self.G.nodes() if self.G.nodes()[n]['test']]\n",
+        "\n",
+        "        # 不参与训练的结点id\n",
+        "        self.no_train_nodes_set = set(self.val_nodes + self.test_nodes)\n",
+        "        # 可训练的结点id\n",
+        "        self.train_nodes = set(G.nodes()).difference(self.no_train_nodes_set)\n",
+        "        # don't train on nodes that only have edges to test set\n",
+        "        # 只保留有邻居的结点\n",
+        "        self.train_nodes = [n for n in self.train_nodes if self.deg[id2idx[str(n)]] > 0]\n",
+        "\n",
+        "    def _make_label_vec(self, node):\n",
+        "        label = self.label_map[node]\n",
+        "        if isinstance(label, list):\n",
+        "            label_vec = np.array(label)\n",
+        "        else:\n",
+        "            label_vec = np.zeros((self.num_classes))\n",
+        "            class_ind = self.label_map[node]\n",
+        "            label_vec[class_ind] = 1\n",
+        "        return label_vec\n",
+        "\n",
+        "    def construct_adj(self):\n",
+        "        # adjacency shape: (14756, 128) ，用于存储所有节点的邻居节点id\n",
+        "        adj = len(self.id2idx) * np.ones((len(self.id2idx)+1, self.max_degree))\n",
+        "        # (14755,)  ，用于存储所有结点的degree值\n",
+        "        deg = np.zeros((len(self.id2idx),))\n",
+        "\n",
+        "        for nodeid in self.G.nodes():\n",
+        "            if self.G.nodes()[nodeid]['test'] or self.G.nodes()[nodeid]['val']:\n",
+        "                continue\n",
+        "\n",
+        "            # 获取所有训练集的邻居节点的id\n",
+        "            neighbors = np.array([self.id2idx[str(neighbor)]\n",
+        "                for neighbor in self.G.neighbors(nodeid)\n",
+        "                if (not self.G[nodeid][neighbor]['train_removed'])])\n",
+        "            \n",
+        "            deg[self.id2idx[str(nodeid)]] = len(neighbors)\n",
+        "            if len(neighbors) == 0:\n",
+        "                continue\n",
+        "            if len(neighbors) > self.max_degree:\n",
+        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=False)\n",
+        "            elif len(neighbors) < self.max_degree:\n",
+        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=True)\n",
+        "            adj[self.id2idx[str(nodeid)], :] = neighbors\n",
+        "        return adj, deg\n",
+        "\n",
+        "    def construct_test_adj(self):\n",
+        "        adj = len(self.id2idx) * np.ones((len(self.id2idx)+1, self.max_degree))\n",
+        "        for nodeid in self.G.nodes():\n",
+        "            # 所有邻居节点的id，这里没有限制训练集或测试集\n",
+        "            neighbors = np.array([self.id2idx[str(neighbor)]\n",
+        "                for neighbor in self.G.neighbors(nodeid)])\n",
+        "            if len(neighbors) == 0:\n",
+        "                continue\n",
+        "            if len(neighbors) > self.max_degree:\n",
+        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=False)\n",
+        "            elif len(neighbors) < self.max_degree:\n",
+        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=True)\n",
+        "            adj[self.id2idx[str(nodeid)], :] = neighbors\n",
+        "        return adj\n",
+        "\n",
+        "    def end(self):\n",
+        "        return self.batch_num * self.batch_size >= len(self.train_nodes)\n",
+        "\n",
+        "    def batch_feed_dict(self, batch_nodes, val=False):\n",
+        "        batch1id = batch_nodes\n",
+        "        batch1 = [self.id2idx[n] for n in batch1id]\n",
+        "\n",
+        "        labels = np.vstack([self._make_label_vec(node) for node in batch1id])\n",
+        "        feed_dict = dict()\n",
+        "        feed_dict.update({self.placeholders['batch_size'] : len(batch1)})\n",
+        "        feed_dict.update({self.placeholders['batch']: batch1})\n",
+        "        feed_dict.update({self.placeholders['labels']: labels})\n",
+        "\n",
+        "        return feed_dict, labels\n",
+        "\n",
+        "    def node_val_feed_dict(self, size=None, test=False):\n",
+        "        if test:\n",
+        "            val_nodes = self.test_nodes\n",
+        "        else:\n",
+        "            val_nodes = self.val_nodes\n",
+        "        if not size is None:\n",
+        "            val_nodes = np.random.choice(val_nodes, size, replace=True)\n",
+        "        # add a dummy neighbor\n",
+        "        ret_val = self.batch_feed_dict(val_nodes)\n",
+        "        return ret_val[0], ret_val[1]\n",
+        "\n",
+        "    def incremental_node_val_feed_dict(self, size, iter_num, test=False):\n",
+        "        if test:\n",
+        "            val_nodes = self.test_nodes\n",
+        "        else:\n",
+        "            val_nodes = self.val_nodes\n",
+        "        val_node_subset = val_nodes[iter_num*size:min((iter_num+1)*size,\n",
+        "            len(val_nodes))]\n",
+        "\n",
+        "        # add a dummy neighbor\n",
+        "        ret_val = self.batch_feed_dict(val_node_subset)\n",
+        "        return ret_val[0], ret_val[1], (iter_num+1)*size >= len(val_nodes), val_node_subset\n",
+        "\n",
+        "    def num_training_batches(self):\n",
+        "        return len(self.train_nodes) // self.batch_size + 1\n",
+        "\n",
+        "    def next_minibatch_feed_dict(self):\n",
+        "        start_idx = self.batch_num * self.batch_size\n",
+        "        self.batch_num += 1\n",
+        "        end_idx = min(start_idx + self.batch_size, len(self.train_nodes))\n",
+        "        batch_nodes = self.train_nodes[start_idx : end_idx]\n",
+        "        return self.batch_feed_dict(batch_nodes)\n",
+        "\n",
+        "    def incremental_embed_feed_dict(self, size, iter_num):\n",
+        "        node_list = self.nodes\n",
+        "        val_nodes = node_list[iter_num*size:min((iter_num+1)*size,\n",
+        "            len(node_list))]\n",
+        "        return self.batch_feed_dict(val_nodes), (iter_num+1)*size >= len(node_list), val_nodes\n",
+        "\n",
+        "    def shuffle(self):\n",
+        "        \"\"\" Re-shuffle the training set.\n",
+        "            Also reset the batch number.\n",
+        "        \"\"\"\n",
+        "        self.train_nodes = np.random.permutation(self.train_nodes)\n",
+        "        self.batch_num = 0\n"
+      ],
+      "metadata": {
+        "id": "ZYCKM4i5PmPf"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "\"\"\"\n",
+        "    This minibatch iterator iterates over nodes for supervised learning.\n",
+        "\n",
+        "    G -- networkx graph\n",
+        "    id2idx -- dict mapping node ids to integer values indexing feature tensor\n",
+        "    placeholders -- standard tensorflow placeholders object for feeding\n",
+        "    label_map -- map from node ids to class values (integer or list)\n",
+        "    num_classes -- number of output classes\n",
+        "    batch_size -- size of the minibatches\n",
+        "    max_degree -- maximum size of the downsampled adjacency lists\n",
+        "\"\"\"\n",
+        "# 实例化 NodeMinibatch 迭代器\n",
+        "minibatch = NodeMinibatchIterator(G,\n",
+        "            id_map,\n",
+        "            placeholders,\n",
+        "            class_map,\n",
+        "            num_classes,\n",
+        "            batch_size=512,\n",
+        "            max_degree=128,\n",
+        "            context_pairs = context_pairs)\n",
+        "\n",
+        "# adjacency shape: (14756, 128)  包装为placeholder\n",
+        "adj_info_ph = tf.compat.v1.placeholder(tf.int32, shape=minibatch.adj.shape)\n",
+        "adj_info = tf.Variable(adj_info_ph, trainable=False, name=\"adj_info\")\n",
+        "\n",
+        "# 接着就是构建模型了，需要改动的兼容代码过多，暂不继续了"
+      ],
+      "metadata": {
+        "id": "mUW98eVhQ5H7"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        ""
+      ],
+      "metadata": {
+        "id": "Wp3DZreLrdtF"
+      },
+      "execution_count": null,
+      "outputs": []
+    }
+  ]
+}
\ No newline at end of file

From 3fab4e3b4d7ff21bcfd2b5386c8255d93ae46438 Mon Sep 17 00:00:00 2001
From: java002 <dingqin@163.com>
Date: Wed, 16 Mar 2022 15:38:50 +0800
Subject: [PATCH 26/28] minibatch cimmit

---
 graphsage/minibatch.py | 92 ++++++++++++++++++++++++++++--------------
 1 file changed, 62 insertions(+), 30 deletions(-)

diff --git a/graphsage/minibatch.py b/graphsage/minibatch.py
index 0dcb9a2d..e83418f5 100644
--- a/graphsage/minibatch.py
+++ b/graphsage/minibatch.py
@@ -5,19 +5,20 @@
 
 np.random.seed(123)
 
+#边迭代器，在无监督训练中使用
 class EdgeMinibatchIterator(object):
     
     """ This minibatch iterator iterates over batches of sampled edges or
     random pairs of co-occuring edges.
 
-    G -- networkx graph
-    id2idx -- dict mapping node ids to index in feature tensor
-    placeholders -- tensorflow placeholders object
-    context_pairs -- if not none, then a list of co-occuring node pairs (from random walks)
-    batch_size -- size of the minibatches
-    max_degree -- maximum size of the downsampled adjacency lists
-    n2v_retrain -- signals that the iterator is being used to add new embeddings to a n2v model
-    fixed_n2v -- signals that the iterator is being used to retrain n2v with only existing nodes as context
+    G -- networkx graph  拓扑图
+    id2idx -- dict mapping node ids to index in feature tensor   节点索引，映射toy-ppi-id_map.json文件
+    placeholders -- tensorflow placeholders object   tf的占位符
+    context_pairs -- if not none, then a list of co-occuring node pairs (from random walks)  随机步数采用结果，采样流程见utils
+    batch_size -- size of the minibatches      批次大小
+    max_degree -- maximum size of the downsampled adjacency lists   最大的度（邻居节点的数量）
+    n2v_retrain -- signals that the iterator is being used to add new embeddings to a n2v model   重新训练n2v模型的标识
+    fixed_n2v -- signals that the iterator is being used to retrain n2v with only existing nodes as context  只用外部节点训练n2v模型标识
     """
     def __init__(self, G, id2idx, 
             placeholders, context_pairs=None, batch_size=100, max_degree=25,
@@ -30,18 +31,20 @@ def __init__(self, G, id2idx,
         self.placeholders = placeholders
         self.batch_size = batch_size
         self.max_degree = max_degree
-        self.batch_num = 0
+        self.batch_num = 0      #批次数，初始化为0，训练过程中递增
 
-        self.nodes = np.random.permutation(G.nodes())
-        self.adj, self.deg = self.construct_adj()
-        self.test_adj = self.construct_test_adj()
-        if context_pairs is None:
+        self.nodes = np.random.permutation(G.nodes())  #节点乱序
+        self.adj, self.deg = self.construct_adj()  #adj矩阵：所有训练节点的取max_degree个邻居节点，deg矩阵：所有训练节点的有效邻居数
+        self.test_adj = self.construct_test_adj() #所有节点（包含测试和验证节点）的adj矩阵
+        if context_pairs is None: #若入参context_pairs不为空，取context_pairs作为训练边，否则取G图中所有边
             edges = G.edges()
         else:
             edges = context_pairs
         self.train_edges = self.edges = np.random.permutation(edges)
         if not n2v_retrain:
+            #训练边剔除顶点不存在拓扑图，顶点有效邻居个数，顶点为test或val的边
             self.train_edges = self._remove_isolated(self.train_edges)
+            #验证边取顶点为test或val的边
             self.val_edges = [e for e in G.edges() if G[e[0]][e[1]]['train_removed']]
         else:
             if fixed_n2v:
@@ -49,19 +52,22 @@ def __init__(self, G, id2idx,
             else:
                 self.train_edges = self.val_edges = self.edges
 
+        #打印训练节点和测试节点的数量
         print(len([n for n in G.nodes() if not G.node[n]['test'] and not G.node[n]['val']]), 'train nodes')
         print(len([n for n in G.nodes() if G.node[n]['test'] or G.node[n]['val']]), 'test nodes')
         self.val_set_size = len(self.val_edges)
 
+    #剔除顶点1为测试或训练节点的边
     def _n2v_prune(self, edges):
         is_val = lambda n : self.G.node[n]["val"] or self.G.node[n]["test"]
         return [e for e in edges if not is_val(e[1])]
 
+    #剔除顶点不在G图中，顶点的有效邻居数为0且顶点不为test
     def _remove_isolated(self, edge_list):
         new_edge_list = []
         missing = 0
         for n1, n2 in edge_list:
-            if not n1 in self.G.node or not n2 in self.G.node:
+            if not n1 in self.G.node or not n2 in self.G.node: #顶点1或顶点2不在G图中
                 missing += 1
                 continue
             if (self.deg[self.id2idx[n1]] == 0 or self.deg[self.id2idx[n2]] == 0) \
@@ -73,26 +79,32 @@ def _remove_isolated(self, edge_list):
         print("Unexpected missing:", missing)
         return new_edge_list
 
+    #获取adj和deg两个矩阵，adj矩阵每行为当年节点的的指定数量邻居节点id,按ididx索引排列
+    #deg为每个节点的训练邻居节点的个数
     def construct_adj(self):
+        #adj初始化:一个节点总数+1行，max_degree列，初始化值全部为节点总数的二维矩阵
+        #deg初始化：一个节点总数行的一维矩阵
         adj = len(self.id2idx)*np.ones((len(self.id2idx)+1, self.max_degree))
         deg = np.zeros((len(self.id2idx),))
 
-        for nodeid in self.G.nodes():
-            if self.G.node[nodeid]['test'] or self.G.node[nodeid]['val']:
+        for nodeid in self.G.nodes():#对全部节点循环
+            if self.G.node[nodeid]['test'] or self.G.node[nodeid]['val']: #测试或验证节点直接跳过
                 continue
             neighbors = np.array([self.id2idx[neighbor] 
                 for neighbor in self.G.neighbors(nodeid)
-                if (not self.G[nodeid][neighbor]['train_removed'])])
-            deg[self.id2idx[nodeid]] = len(neighbors)
+                if (not self.G[nodeid][neighbor]['train_removed'])])#取所有不为test和val的邻居节点
+            deg[self.id2idx[nodeid]] = len(neighbors)#deg赋值为邻居个数
             if len(neighbors) == 0:
                 continue
-            if len(neighbors) > self.max_degree:
+            #取max_degree个邻居节点
+            if len(neighbors) > self.max_degree: #邻居节点大于max_degree，无重复采样
                 neighbors = np.random.choice(neighbors, self.max_degree, replace=False)
-            elif len(neighbors) < self.max_degree:
+            elif len(neighbors) < self.max_degree:#邻居节点小于max_degree，有重复采样
                 neighbors = np.random.choice(neighbors, self.max_degree, replace=True)
             adj[self.id2idx[nodeid], :] = neighbors
         return adj, deg
 
+    #取所有节点的adj矩阵，方式与construct_adj相同
     def construct_test_adj(self):
         adj = len(self.id2idx)*np.ones((len(self.id2idx)+1, self.max_degree))
         for nodeid in self.G.nodes():
@@ -106,10 +118,11 @@ def construct_test_adj(self):
                 neighbors = np.random.choice(neighbors, self.max_degree, replace=True)
             adj[self.id2idx[nodeid], :] = neighbors
         return adj
-
+    #判断当前epoch是否结束
     def end(self):
         return self.batch_num * self.batch_size >= len(self.train_edges)
 
+    #将边集合转为feed_dict,batch_size:集合总数,batch1：节点1集合,batch2：节点2集合
     def batch_feed_dict(self, batch_edges):
         batch1 = []
         batch2 = []
@@ -124,6 +137,7 @@ def batch_feed_dict(self, batch_edges):
 
         return feed_dict
 
+    #下一批次feed_dict
     def next_minibatch_feed_dict(self):
         start_idx = self.batch_num * self.batch_size
         self.batch_num += 1
@@ -131,9 +145,11 @@ def next_minibatch_feed_dict(self):
         batch_edges = self.train_edges[start_idx : end_idx]
         return self.batch_feed_dict(batch_edges)
 
+    #取当前是第几批次
     def num_training_batches(self):
         return len(self.train_edges) // self.batch_size + 1
 
+    #取指定大小的训练边的feed_dict，首次取数
     def val_feed_dict(self, size=None):
         edge_list = self.val_edges
         if size is None:
@@ -143,18 +159,19 @@ def val_feed_dict(self, size=None):
             val_edges = [edge_list[i] for i in ind[:min(size, len(ind))]]
             return self.batch_feed_dict(val_edges)
 
+    #下一批次的训练边的feed_dict
     def incremental_val_feed_dict(self, size, iter_num):
         edge_list = self.val_edges
         val_edges = edge_list[iter_num*size:min((iter_num+1)*size, 
             len(edge_list))]
         return self.batch_feed_dict(val_edges), (iter_num+1)*size >= len(self.val_edges), val_edges
-
+    #去下一批次的节点自己到自己组成的边，并转为feed_dict
     def incremental_embed_feed_dict(self, size, iter_num):
         node_list = self.nodes
         val_nodes = node_list[iter_num*size:min((iter_num+1)*size,len(node_list))]
         val_edges = [(n,n) for n in val_nodes]
         return self.batch_feed_dict(val_edges), (iter_num+1)*size >= len(node_list), val_edges
-
+    #将全量边分为训练边和验证边
     def label_val(self):
         train_edges = []
         val_edges = [] 
@@ -166,6 +183,7 @@ def label_val(self):
                 train_edges.append((n1,n2))
         return train_edges, val_edges
 
+    #洗牌
     def shuffle(self):
         """ Re-shuffle the training set.
             Also reset the batch number.
@@ -174,6 +192,7 @@ def shuffle(self):
         self.nodes = np.random.permutation(self.nodes)
         self.batch_num = 0
 
+#节点迭代器，在有监督训练中使用
 class NodeMinibatchIterator(object):
     
     """ 
@@ -182,8 +201,8 @@ class NodeMinibatchIterator(object):
     G -- networkx graph
     id2idx -- dict mapping node ids to integer values indexing feature tensor
     placeholders -- standard tensorflow placeholders object for feeding
-    label_map -- map from node ids to class values (integer or list)
-    num_classes -- number of output classes
+    label_map -- map from node ids to class values (integer or list)   所有节点的类标数据，映射文件toy-ppi-map
+    num_classes -- number of output classes   每个类标数据的维度
     batch_size -- size of the minibatches
     max_degree -- maximum size of the downsampled adjacency lists
     以toy-ppi数据集举例：
@@ -205,17 +224,20 @@ def __init__(self, G, id2idx,
         self.label_map = label_map
         self.num_classes = num_classes
 
+        #adj:采样邻居矩阵，deg:训练邻居节点个数矩阵
         self.adj, self.deg = self.construct_adj()
         self.test_adj = self.construct_test_adj()
-
+        #验证节点集合，测试节点集合
         self.val_nodes = [n for n in self.G.nodes() if self.G.node[n]['val']]
         self.test_nodes = [n for n in self.G.nodes() if self.G.node[n]['test']]
-
+        #非训练节点集合，训练节点集合
         self.no_train_nodes_set = set(self.val_nodes + self.test_nodes)
         self.train_nodes = set(G.nodes()).difference(self.no_train_nodes_set)
         # don't train on nodes that only have edges to test set
+        #剔除有效边为0的节点
         self.train_nodes = [n for n in self.train_nodes if self.deg[id2idx[n]] > 0]
 
+    #若类标数据为list，转为一维矩阵；若为单数值，则创建一个全零矩阵，并将该数据位置位1
     def _make_label_vec(self, node):
         label = self.label_map[node]
         if isinstance(label, list):
@@ -254,6 +276,7 @@ def construct_adj(self):
             adj[self.id2idx[nodeid], :] = neighbors
         return adj, deg
 
+    #所有节点的adj矩阵，包含test和val节点
     def construct_test_adj(self):
         adj = len(self.id2idx)*np.ones((len(self.id2idx)+1, self.max_degree))
         for nodeid in self.G.nodes():
@@ -269,13 +292,16 @@ def construct_test_adj(self):
             adj[self.id2idx[nodeid], :] = neighbors
         return adj
 
+    #判断是否结束
     def end(self):
         return self.batch_num * self.batch_size >= len(self.train_nodes)
 
+    #节点集合的feed_dict，batch_size：集合大小，batch：节点的index信息，labels：集合的类标数据
     def batch_feed_dict(self, batch_nodes, val=False):
         batch1id = batch_nodes
         batch1 = [self.id2idx[n] for n in batch1id]
-              
+
+        #按照batch_nodes的顺序，将batch_nodes的类标数据堆叠成2维数组
         labels = np.vstack([self._make_label_vec(node) for node in batch1id])
         feed_dict = dict()
         feed_dict.update({self.placeholders['batch_size'] : len(batch1)})
@@ -284,6 +310,7 @@ def batch_feed_dict(self, batch_nodes, val=False):
 
         return feed_dict, labels
 
+    #从测试节点或验证节点中取size个节点，获取节点list的feed_dict
     def node_val_feed_dict(self, size=None, test=False):
         if test:
             val_nodes = self.test_nodes
@@ -295,21 +322,24 @@ def node_val_feed_dict(self, size=None, test=False):
         ret_val = self.batch_feed_dict(val_nodes)
         return ret_val[0], ret_val[1]
 
+    #取训练或验证节点的下一批次节点的feed_dict
     def incremental_node_val_feed_dict(self, size, iter_num, test=False):
         if test:
             val_nodes = self.test_nodes
         else:
             val_nodes = self.val_nodes
-        val_node_subset = val_nodes[iter_num*size:min((iter_num+1)*size, 
+        val_node_subset = val_nodes[iter_num*size:min((iter_num+1)*size,
             len(val_nodes))]
 
         # add a dummy neighbor
         ret_val = self.batch_feed_dict(val_node_subset)
         return ret_val[0], ret_val[1], (iter_num+1)*size >= len(val_nodes), val_node_subset
 
+    #当前是第几批次
     def num_training_batches(self):
         return len(self.train_nodes) // self.batch_size + 1
 
+    #训练节点的下一批次的feed_dict
     def next_minibatch_feed_dict(self):
         start_idx = self.batch_num * self.batch_size
         self.batch_num += 1
@@ -317,12 +347,14 @@ def next_minibatch_feed_dict(self):
         batch_nodes = self.train_nodes[start_idx : end_idx]
         return self.batch_feed_dict(batch_nodes)
 
+    #全量节点的下一批次节点feed_dict
     def incremental_embed_feed_dict(self, size, iter_num):
         node_list = self.nodes
-        val_nodes = node_list[iter_num*size:min((iter_num+1)*size, 
+        val_nodes = node_list[iter_num*size:min((iter_num+1)*size,
             len(node_list))]
         return self.batch_feed_dict(val_nodes), (iter_num+1)*size >= len(node_list), val_nodes
 
+    #洗牌
     def shuffle(self):
         """ Re-shuffle the training set.
             Also reset the batch number.

From 3515aa79c7b09a9d5eea69fbb6f5c0a641a12c86 Mon Sep 17 00:00:00 2001
From: pengyi <hbyspy2008@qq.com>
Date: Wed, 16 Mar 2022 15:42:10 +0800
Subject: [PATCH 27/28] =?UTF-8?q?=E6=9B=B4=E6=96=B0=E6=B3=A8=E9=87=8A?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 ...72\345\231\250\344\277\241\346\201\257.md" |  55 ---
 graphsage/models.py                           |  62 ++-
 graphsage/supervised_models.py                |   4 +-
 graphsage/unsupervised_train.py               |  24 +-
 preprocess.ipynb                              | 446 ------------------
 5 files changed, 59 insertions(+), 532 deletions(-)
 delete mode 100644 "aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
 delete mode 100644 preprocess.ipynb

diff --git "a/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md" "b/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
deleted file mode 100644
index 2a330537..00000000
--- "a/aliyun\346\234\272\345\231\250\344\277\241\346\201\257.md"
+++ /dev/null
@@ -1,55 +0,0 @@
-## 阿里云机器信息
-
-IP：敏感信息不放在网上
-
-mag240原数据目录：`/mnt/ogb-dataset/mag240m/data/raw`
-
-```
-├── RELEASE_v1.txt
-├── mapping    //空文件夹
-├── meta.pt
-├── processed
-│   ├── author___affiliated_with___institution
-│   │   └── edge_index.npy     //作者和机构的边，shape=[2,num_edges]
-│   ├── author___writes___paper
-│   │   └── edge_index.npy     //作者和论文的边，shape=[2,num_edges]
-│   ├── paper
-│   │   ├── node_feat.npy    //论文节点的特征，shape=[num_node,768]
-│   │   ├── node_label.npy   // 论文的标签
-│   │   └── node_year.npy   // 论文年份
-│   └── paper___cites___paper
-│       └── edge_index.npy  // 论文引用关系的边shape=[2,num_edges]
-├── raw //空文件夹
-└── split_dict.pt    //切分训练集、验证集、测试集方式的文件，用torch读取是一个字典，keys=[‘train’,’valid’,’test’], value是node_index
-
-```
-
-
-
-### docker镜像
-
-#### opeceipeno/dgl:v1.4
-
-ogb代码的运行环境，想法是通过虚拟环境去激活各个方案的运行环境，当前做好了Google的mag240m运行环境
-
-[GitHub地址](https://github.com/deepmind/deepmind-research/tree/master/ogb_lsc/mag)
-
-```
-docker run --gpus all -it -v /mnt:/mnt opeceipeno/dgl:v1.4 bash
-# 启动容器后，激活Google代码的运行环境
-source /py3_venv/google_ogb_mag240m/bin/activate
-# /workspace 目录有代码
-```
-
-Google方案预处理后的数据目录：`/mnt/ogb-dataset/mag240m/data/preprocessed`，相当于执行完了`run_preprocessing.sh`脚本，下一步是可以复现实验，
-
-
-
-#### opeceipeno/graphsage:gpu
-
-graphSAGE的环境，[GitHub地址](https://github.com/qksidmx/GraphSAGE)
-
-```
-docker run --gpus all -it opeceipeno/graphsage:gpu bash
-#/notebook目录下面有代码，运行实验参考readme文档
-```
diff --git a/graphsage/models.py b/graphsage/models.py
index 746b6c0a..7d5d84c0 100644
--- a/graphsage/models.py
+++ b/graphsage/models.py
@@ -97,7 +97,6 @@ def load(self, sess=None):  # 从本地文件读取模型
         print("Model restored from file: %s" % save_path)
 
 
-
 # 多层感知机，是一个基础的深度模型
 class MLP(Model):
     """ A standard multi-layer perceptron """
@@ -301,7 +300,10 @@ def sample(self, inputs, layer_infos, batch_size=None):
         """ Sample neighbors to be the supportive fields for multi-layer convolutions.
 
         函数功能：对输入的每一个节点，根据采样跳数目，递归地采样邻居，作为该节点的支持域
+        输入：
+        inputs：一批次的节点id
 
+        输出：
         samples是一个列表，列表的每一个元素又是一个列表，长度不一，存放的是该跳数下的所有的邻居节点id
         示例：
         samples[0] 维度是 [batch_size,] ，即是自身
@@ -309,11 +311,11 @@ def sample(self, inputs, layer_infos, batch_size=None):
         samples[2] [layer_infos[1].num_samples * layer_infos[0].num_samples * batch_size,]
         以此类推
 
-        # support_sizes 存的是的各层的采样数目，是一个列表，每个元素是一个正整数
-        # support_sizes[0] = 1， 意义是初始状态，邻居就是节点本身
-        # support_sizes[1] = layer_infos[-1].num_samples * 1， 本实验中为10
-        # support_sizes[2] = layer_infos[-1].num_samples * layer_infos[-2].num_samples * 1， 本实验中为10*15=250
-        # 以此类推，从最外层的邻居数依次往内乘
+        support_sizes 存的是的各层的采样数目，是一个列表，每个元素是一个正整数
+        support_sizes[0] = 1， 意义是初始状态，邻居就是节点本身
+        support_sizes[1] = layer_infos[-1].num_samples * 1， 本实验中为10
+        support_sizes[2] = layer_infos[-1].num_samples * layer_infos[-2].num_samples * 1， 本实验中为10*15=250
+        以此类推，从最外层的邻居数依次往内乘
 
         Args:
             inputs: batch inputs
@@ -329,7 +331,7 @@ def sample(self, inputs, layer_infos, batch_size=None):
         support_sizes = [support_size]
 
         for k in range(len(layer_infos)):  # k为跳数，实验中k = 0 1
-            t = len(layer_infos) - k - 1 # t = 1 0
+            t = len(layer_infos) - k - 1  # t = 1 0
 
             # 每一跳的邻居数目是前一跳的邻居节点数*该层的采样数，有个累乘的逻辑
             support_size *= layer_infos[t].num_samples
@@ -340,7 +342,7 @@ def sample(self, inputs, layer_infos, batch_size=None):
             node = sampler((samples[k], layer_infos[t].num_samples))
 
             # reshape成一维数组，再添加进samples中
-            samples.append(tf.reshape(node, [support_size * batch_size, ])) 
+            samples.append(tf.reshape(node, [support_size * batch_size, ]))
 
             # 同时记录好每一层的采样数
             support_sizes.append(support_size)
@@ -383,15 +385,14 @@ def aggregate(self, samples, input_features, dims, num_samples, support_sizes, b
 
         # length: number of layers + 1
         # 遍历samples列表，根据每一个元素中存放的节点id，从全量的特征矩阵里获取所需的节点特征
-     
+
         hidden = [tf.nn.embedding_lookup(
             input_features, node_samples) for node_samples in samples]
         # hidden[0] [batch, num_features]
         # hidden[1] [layer_infos[1].num_samples * batch_size, num_features]
         # hidden[2] [layer_infos[1].num_samples * layer_infos[0].num_samples * batch_size, num_features]
         # num_features表示的是特征维度，实验中为50
-        
-        
+
         # 输入batch1的时候，该项为aggregators = None， 输入batch2或者neg_samples的时候，aggregators为batch1生成的aggregators
         # 即他们用的是同一个聚合器
         new_agg = aggregators is None
@@ -437,9 +438,10 @@ def aggregate(self, samples, input_features, dims, num_samples, support_sizes, b
                 # 因为hidden[i]存放为二维，而mean_aggregator是需要将邻居节点特征平均，
                 # 因此需要将它reshape一下，方便在后面的处理中取所有邻居的均值
                 # neigh_dims = [batch_size * 当前跳数的支持节点数，当前层的需要采样的邻居节点数，特征数]
-                #  
+                #
                 neigh_dims = [batch_size * support_sizes[hop],
-                              num_samples[len(num_samples) - hop - 1],  # 这个维度，对应sample函数里的 t = len(layer_infos) - k - 1
+                              # 这个维度，对应sample函数里的 t = len(layer_infos) - k - 1
+                              num_samples[len(num_samples) - hop - 1],
                               dim_mult*dims[layer]]
                 h = aggregator((hidden[hop],
                                 tf.reshape(hidden[hop + 1], neigh_dims)))
@@ -463,11 +465,10 @@ def _build(self):
 
         --------
         在本实验中，就是利用这个函数，利用每个节点的度数形成概率分布，从节点集合中获取一批节点id，在后续视作负样本
-        true_classes个参数传入的是labels，但经测试，采样的结果和这个参数是无关的样子，而且实际是有可能会采样到负样本的
+        true_classes个参数传入的是labels，但经测试，采样的结果和这个参数是无关的样子，而且实际是有可能会采样到正样本的
         返回的结果neg_samples里面是一个列表，每一个元素代表的是节点id
         
-        参考 https://github.com/williamleif/GraphSAGE/issues/76， 这个里面作者说了是可能会采样到正样本
-        只是他们假设，当整个图数据集远大于邻域计算图时，采样到正样本的概率很小。
+        参考 https://github.com/williamleif/GraphSAGE/issues/76， 这个里面作者说了是可能会采样到正样本，只是他们假设，当整个图数据集远大于邻域计算图时，采样到正样本的概率很小。
         """
 
         self.neg_samples, _, _ = (tf.nn.fixed_unigram_candidate_sampler(
@@ -486,8 +487,8 @@ def _build(self):
         samples1, support_sizes1 = self.sample(self.inputs1, self.layer_infos)
         samples2, support_sizes2 = self.sample(self.inputs2, self.layer_infos)
 
-        # 每层需要的采样数 实验中是[25,10] 
-        
+        # 每层需要的采样数 实验中是[25,10]
+
         num_samples = [
             layer_info.num_samples for layer_info in self.layer_infos]
 
@@ -511,7 +512,6 @@ def _build(self):
 
         dim_mult = 2 if self.concat else 1
 
-
         # 这里生成了一个预测层，注意参数bilinear_weights,这个值如果为True，则会生成一个可训练的参数矩阵，在后续的计算loss会用到
         # 但是本实验在这里设置了否，则无参数矩阵，本质上就是一个计算loss的类，完全不影响上述aggregator的输出
         self.link_pred_layer = BipartiteEdgePredLayer(dim_mult*self.dims[-1],
@@ -545,8 +545,8 @@ def build(self):
         # 梯度裁剪，若梯度大于5则置为5，小于-5则置为-5，
         clipped_grads_and_vars = [(tf.clip_by_value(grad, -5.0, 5.0) if grad is not None else None, var)
                                   for grad, var in grads_and_vars]
-        
-        # clipped_grads_and_vars 是一个元组,(grad,var),表示梯度值和变量值
+
+        # clipped_grads_and_vars 是一个元组,(grad,var),表示梯度值和变量值，这里只取了grad
         self.grad, _ = clipped_grads_and_vars[0]
 
         # 利用裁剪后的梯度更新模型参数
@@ -597,20 +597,30 @@ def _accuracy(self):
         self.aff_all = tf.concat(axis=1, values=[self.neg_aff, _aff])
         size = tf.shape(self.aff_all)[1]
 
-        # ④利用top_k函数，两步计算出正样本对之间的亲和度的排名，
+        # ④利用top_k函数，两步计算出正样本对之间的亲和度的排名
+        #  tf.nn.top_k函数是根据输入的数组，返回最大的k个数和他们的序号
+        # 这里两次利用tok_k函数，得到rank值
         # self.ranks中表示的是每个顶点和负样本、正样本之间的亲和度排名，维度:[batch_size, neg_samples_size + 1]
+        # 示例：
+        # result = tf.constant([0.5, 0.9, 0.3, 0.4, 0.6, 0.8, 0.7, 0.1])
+        # _, indices_of_ranks = tf.nn.top_k(result, k=len(result))
+        # indices_of_ranks.numpy() : [1, 5, 6, 4, 0, 3, 2, 7]   按顺序看，表示最大数序号为1，第二大的数序号为5 ...
+        # _, ranks = tf.nn.top_k(-indices_of_ranks, k=len(result))
+        # ranks.numpy() : [4, 0, 6, 5, 3, 1, 2, 7]  这里得到的就是result中每个元素的排名序号，0表示最大，以此类推
+        #
         _, indices_of_ranks = tf.nn.top_k(self.aff_all, k=size)
         _, self.ranks = tf.nn.top_k(-indices_of_ranks, k=size)
 
-        # 取self.ranks最后一列，即正样本的排名序数，因为是从0算起的，所以要+1
-        # mrr = 1.0/rank
+        # 取self.ranks最后一列，即正样本的排名序数，因为是从0算起的，所以要+1 
+        # Mean Reciprocal Rank(MRR) = 1.0/rank
+        # 通过正确的检索结果值在所有检索结果中的排名来评估排序性能， rank越大，mrr值越小
+
         self.mrr = tf.reduce_mean(
             tf.div(1.0, tf.cast(self.ranks[:, -1] + 1, tf.float32)))
         tf.summary.scalar('mrr', self.mrr)
 
 
-
-#  
+#
 class Node2VecModel(GeneralizedModel):
     def __init__(self, placeholders, dict_size, degrees, name=None,
                  nodevec_dim=50, lr=0.001, **kwargs):
diff --git a/graphsage/supervised_models.py b/graphsage/supervised_models.py
index ab1c7c70..84a6c1ad 100644
--- a/graphsage/supervised_models.py
+++ b/graphsage/supervised_models.py
@@ -118,13 +118,15 @@ def build(self):
         clipped_grads_and_vars = [(tf.clip_by_value(grad, -5.0, 5.0) if grad is not None else None, var) 
                 for grad, var in grads_and_vars]
         self.grad, _ = clipped_grads_and_vars[0]
+
         self.opt_op = self.optimizer.apply_gradients(clipped_grads_and_vars)
+
         self.preds = self.predict()
 
     def _loss(self):
         # Weight decay loss
 
-        #L2正则化项 
+        # L2正则化项 
         for aggregator in self.aggregators:
             for var in aggregator.vars.values():
                 self.loss += FLAGS.weight_decay * tf.nn.l2_loss(var)
diff --git a/graphsage/unsupervised_train.py b/graphsage/unsupervised_train.py
index 469ad95d..c7bd8df2 100644
--- a/graphsage/unsupervised_train.py
+++ b/graphsage/unsupervised_train.py
@@ -70,10 +70,20 @@ def log_dir():
 
 # Define model evaluation function
 def evaluate(sess, model, minibatch_iter, size=None):
+    """
+    评估函数
+    输入：
+    model：训练好了的模型
+    minibatch_iter：迭代类
+    size： batch_size
+    """
+
     t_test = time.time()
 
     # 评估阶段，这里传入的是验证集
     feed_dict_val = minibatch_iter.val_feed_dict(size)
+
+    # 这里只跑模型的loss等指标
     outs_val = sess.run([model.loss, model.ranks, model.mrr], 
                         feed_dict=feed_dict_val)
     return outs_val[0], outs_val[1], outs_val[2], (time.time() - t_test)
@@ -288,7 +298,7 @@ def train(train_data, test_data=None):
     merged = tf.summary.merge_all()
     summary_writer = tf.summary.FileWriter(log_dir(), sess.graph)
      
-    # Init variables
+    # Init variables 初始化参数
     sess.run(tf.global_variables_initializer(), feed_dict={adj_info_ph: minibatch.adj})
     
     # Train model 训练
@@ -300,7 +310,9 @@ def train(train_data, test_data=None):
     avg_time = 0.0
     epoch_val_costs = []
 
-    # 从minibatch获取训练和验证数据的邻接表信息
+    # 利用tf.assgin函数，将邻接表数据赋值给adj_info
+    # 训练的时候，用训练的邻接表，验证的时候用验证的邻接表
+    # 此时操作并不会执行，在之后sess.run的时候才会真正执行
     train_adj_info = tf.assign(adj_info, minibatch.adj)
     val_adj_info = tf.assign(adj_info, minibatch.test_adj)
 
@@ -321,7 +333,6 @@ def train(train_data, test_data=None):
             # Training step 训练
             # merged是保存了的所有的tf.summary相关的变量，用于tensorboard绘图，
             # model.opt_op 是优化调参的操作
-            # 其他的变量
             # 
             #  
             outs = sess.run([merged, model.opt_op, model.loss, model.ranks, model.aff_all, 
@@ -331,16 +342,21 @@ def train(train_data, test_data=None):
 
 
             # train_shadow_mrr值是一个滑动平均更新的方式
+            # 在得到该批次数据的mrr的时候，不直接更新，而是将历史的mrr值也引入进来计算，避免评估结果抖动较大
+            # 参考 https://www.pianshen.com/article/11191400472/ 
             if train_shadow_mrr is None:
                 train_shadow_mrr = train_mrr#
             else:
+                # 1-0.99中，0.99为衰减率 
                 train_shadow_mrr -= (1-0.99) * (train_shadow_mrr - train_mrr)
 
             if iter % FLAGS.validate_iter == 0:
-                # Validation 验证的时候，需要用验证集的邻接表信息，因此这里跑一下tf.assign的操作
+                # Validation 验证的时候，需要用验证集的邻接表信息，因此这里跑一下tf.assign的操作, 将验证的邻接表赋给adj_info
+                  
                 sess.run(val_adj_info.op)
                 val_cost, ranks, val_mrr, duration  = evaluate(sess, model, minibatch, size=FLAGS.validate_batch_size)
                 # 验证完毕再切回训练集的邻接表信息
+
                 sess.run(train_adj_info.op)
                 epoch_val_costs[-1] += val_cost
             if shadow_mrr is None:
diff --git a/preprocess.ipynb b/preprocess.ipynb
deleted file mode 100644
index 164c2615..00000000
--- a/preprocess.ipynb
+++ /dev/null
@@ -1,446 +0,0 @@
-{
-  "nbformat": 4,
-  "nbformat_minor": 0,
-  "metadata": {
-    "colab": {
-      "name": "graph.ipynb",
-      "provenance": [],
-      "collapsed_sections": []
-    },
-    "kernelspec": {
-      "name": "python3",
-      "display_name": "Python 3"
-    },
-    "language_info": {
-      "name": "python"
-    }
-  },
-  "cells": [
-    {
-      "cell_type": "markdown",
-      "source": [
-        ""
-      ],
-      "metadata": {
-        "id": "lFBIUQovI53M"
-      }
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "metadata": {
-        "colab": {
-          "base_uri": "https://localhost:8080/",
-          "height": 319
-        },
-        "id": "FXItxCYQ5xE2",
-        "outputId": "45d18b2d-50ad-4361-fb5c-1401166b8757"
-      },
-      "outputs": [
-        {
-          "output_type": "display_data",
-          "data": {
-            "image/png": "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\n",
-            "text/plain": [
-              "<Figure size 432x288 with 1 Axes>"
-            ]
-          },
-          "metadata": {}
-        }
-      ],
-      "source": [
-        "# python -m graphsage.supervised_train --train_prefix ./example_data/toy-ppi --model graphsage_mean --sigmoid\n",
-        "# test networkx and visualization\n",
-        "import networkx as nx\n",
-        "import tensorflow as tf\n",
-        "tf.compat.v1.disable_eager_execution()\n",
-        "\n",
-        "G = nx.complete_graph(6)\n",
-        "nx.draw(G)"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "source": [
-        "# download code and data\n",
-        "!git clone https://github.com/williamleif/GraphSAGE"
-      ],
-      "metadata": {
-        "colab": {
-          "base_uri": "https://localhost:8080/"
-        },
-        "id": "S6709xbNrBok",
-        "outputId": "51e908ea-9105-4844-9427-b57a417bf9da"
-      },
-      "execution_count": null,
-      "outputs": [
-        {
-          "output_type": "stream",
-          "name": "stdout",
-          "text": [
-            "Cloning into 'GraphSAGE'...\n",
-            "remote: Enumerating objects: 265, done.\u001b[K\n",
-            "remote: Counting objects: 100% (7/7), done.\u001b[K\n",
-            "remote: Compressing objects: 100% (7/7), done.\u001b[K\n",
-            "remote: Total 265 (delta 3), reused 0 (delta 0), pack-reused 258\u001b[K\n",
-            "Receiving objects: 100% (265/265), 6.43 MiB | 11.28 MiB/s, done.\n",
-            "Resolving deltas: 100% (160/160), done.\n"
-          ]
-        }
-      ]
-    },
-    {
-      "cell_type": "code",
-      "source": [
-        "import json\n",
-        "from networkx.readwrite import json_graph\n",
-        "import os\n",
-        "import numpy as np\n",
-        "import sys\n",
-        "\n",
-        "CODE_ROOT = \"GraphSAGE/graphsage\"\n",
-        "sys.path.append(\"GraphSAGE\")\n",
-        "\n",
-        "def load_data():\n",
-        "    data_path = 'GraphSAGE/example_data'\n",
-        "    # DATA 1, 14755 nodes, 228431 links\n",
-        "    G_data = json.load(open(data_path + '/toy-ppi-G.json'))\n",
-        "    #G_data['nodes'] = G_data['nodes'][:100]\n",
-        "    #G_data['links'] = G_data['links'][:100]\n",
-        "    G = json_graph.node_link_graph(G_data)\n",
-        "    \n",
-        "    conversion = lambda n : n\n",
-        "    lab_conversion = lambda n : n\n",
-        "    \n",
-        "    # DATA 2, (14755, 50) dtype('float64')\n",
-        "    feats = np.load(data_path + '/toy-ppi-feats.npy')\n",
-        "    \n",
-        "    # DATA 3, {\"0\": 0, \"1\": 1}, len: 14755\n",
-        "    # node ids to integer values indexing feature tensor\n",
-        "    # 其实没什么用\n",
-        "    id_map = json.load(open(data_path + \"/toy-ppi-id_map.json\"))\n",
-        "    \n",
-        "    # DATA 4, dict, len: 14755, column 121\n",
-        "    # from node ids to class values (integer or list)\n",
-        "    # 分类标签\n",
-        "    class_map = json.load(open(data_path + \"/toy-ppi-class_map.json\"))\n",
-        "    \n",
-        "    broken_count = 0\n",
-        "    for node in G.nodes():\n",
-        "      if not 'val' in G.nodes()[node] or not 'test' in G.nodes()[node]:\n",
-        "        G.remove_node(node)\n",
-        "        broken_count += 1\n",
-        "    print(\"Removed {:d} nodes that lacked proper annotations due to networkx versioning issues\".format(broken_count))\n",
-        "    \n",
-        "    # edge: (0, 800) 边\n",
-        "    # G[0]: 某结点与所有的关联结点组成的边的集合\n",
-        "    # 标记需要在训练中移除的关联关系，即边\n",
-        "    for edge in G.edges():\n",
-        "      if (G.nodes()[edge[0]]['val'] or G.nodes()[edge[1]]['val'] or\n",
-        "          G.nodes()[edge[0]]['test'] or G.nodes()[edge[1]]['test']):\n",
-        "        G[edge[0]][edge[1]]['train_removed'] = True\n",
-        "      else:\n",
-        "        G[edge[0]][edge[1]]['train_removed'] = False\n",
-        "    \n",
-        "    from sklearn.preprocessing import StandardScaler\n",
-        "    \n",
-        "    # 训练集的id集合，result only int, len: 9716\n",
-        "    train_ids = np.array([id_map[str(n)] for n in G.nodes() \\\n",
-        "                          if not G.nodes()[n]['val'] and not G.nodes()[n]['test']])\n",
-        "    \n",
-        "    train_feats = feats[train_ids]\n",
-        "    \n",
-        "    # 特征缩放，标准化：z = (x - u) / s\n",
-        "    # u is the mean of the training samples\n",
-        "    # s is the standard deviation of the training samples\n",
-        "    scaler = StandardScaler()\n",
-        "    scaler.fit(train_feats)\n",
-        "    feats = scaler.transform(feats)\n",
-        "\n",
-        "    walks = []\n",
-        "\n",
-        "    return G, feats, id_map, walks, class_map"
-      ],
-      "metadata": {
-        "id": "jg81uw5O6Gaz"
-      },
-      "execution_count": null,
-      "outputs": []
-    },
-    {
-      "cell_type": "code",
-      "source": [
-        "def construct_placeholders(num_classes):\n",
-        "  # Define placeholders\n",
-        "  placeholders = {\n",
-        "      'labels' : tf.compat.v1.placeholder(tf.float32, shape=(None, num_classes), name='labels'),\n",
-        "      'dropout': tf.compat.v1.placeholder_with_default(0., shape=(), name='dropout'),\n",
-        "      'batch' : tf.compat.v1.placeholder(tf.int32, shape=(None), name='batch1'),\n",
-        "      'batch_size' : tf.compat.v1.placeholder(tf.int32, name='batch_size'),\n",
-        "  }\n",
-        "  return placeholders"
-      ],
-      "metadata": {
-        "id": "mReVSrV9UzYO"
-      },
-      "execution_count": null,
-      "outputs": []
-    },
-    {
-      "cell_type": "code",
-      "source": [
-        "train_data = load_data()"
-      ],
-      "metadata": {
-        "colab": {
-          "base_uri": "https://localhost:8080/"
-        },
-        "id": "Rv1F4lYF_FfW",
-        "outputId": "7b9e88c6-8ba5-4128-9564-55a7d4cc835c"
-      },
-      "execution_count": null,
-      "outputs": [
-        {
-          "output_type": "stream",
-          "name": "stdout",
-          "text": [
-            "Removed 0 nodes that lacked proper annotations due to networkx versioning issues\n"
-          ]
-        }
-      ]
-    },
-    {
-      "cell_type": "code",
-      "source": [
-        "G = train_data[0]\n",
-        "features = train_data[1]\n",
-        "id_map = train_data[2]\n",
-        "context_pairs = train_data[3]\n",
-        "class_map  = train_data[4]\n",
-        "\n",
-        "# num_classes = 121\n",
-        "num_classes = len(list(class_map.values())[0])\n",
-        "# pad with dummy zero vector, row wise\n",
-        "features = np.vstack([features, np.zeros((features.shape[1],))])\n",
-        "placeholders = construct_placeholders(num_classes)"
-      ],
-      "metadata": {
-        "id": "CjSUlil1kNZP"
-      },
-      "execution_count": null,
-      "outputs": []
-    },
-    {
-      "cell_type": "code",
-      "source": [
-        "class NodeMinibatchIterator(object):\n",
-        "\n",
-        "    \"\"\"\n",
-        "    This minibatch iterator iterates over nodes for supervised learning.\n",
-        "\n",
-        "    G -- networkx graph\n",
-        "    id2idx -- dict mapping node ids to integer values indexing feature tensor\n",
-        "    placeholders -- standard tensorflow placeholders object for feeding\n",
-        "    label_map -- map from node ids to class values (integer or list)\n",
-        "    num_classes -- number of output classes\n",
-        "    batch_size -- size of the minibatches\n",
-        "    max_degree -- maximum size of the downsampled adjacency lists\n",
-        "    以toy-ppi数据集举例：\n",
-        "    label_map为输出，维度为(14755, 121)\n",
-        "    num_class为label_map的第二维，即121\n",
-        "    \"\"\"\n",
-        "    def __init__(self, G, id2idx,\n",
-        "            placeholders, label_map, num_classes,\n",
-        "            batch_size=100, max_degree=25,\n",
-        "            **kwargs):\n",
-        "\n",
-        "        self.G = G\n",
-        "        self.nodes = G.nodes()\n",
-        "        self.id2idx = id2idx\n",
-        "        self.placeholders = placeholders\n",
-        "        self.batch_size = batch_size\n",
-        "        self.max_degree = max_degree\n",
-        "        self.batch_num = 0\n",
-        "        self.label_map = label_map\n",
-        "        self.num_classes = num_classes\n",
-        "\n",
-        "        self.adj, self.deg = self.construct_adj()\n",
-        "        self.test_adj = self.construct_test_adj()\n",
-        "\n",
-        "        self.val_nodes = [n for n in self.G.nodes() if self.G.nodes()[n]['val']]\n",
-        "        self.test_nodes = [n for n in self.G.nodes() if self.G.nodes()[n]['test']]\n",
-        "\n",
-        "        # 不参与训练的结点id\n",
-        "        self.no_train_nodes_set = set(self.val_nodes + self.test_nodes)\n",
-        "        # 可训练的结点id\n",
-        "        self.train_nodes = set(G.nodes()).difference(self.no_train_nodes_set)\n",
-        "        # don't train on nodes that only have edges to test set\n",
-        "        # 只保留有邻居的结点\n",
-        "        self.train_nodes = [n for n in self.train_nodes if self.deg[id2idx[str(n)]] > 0]\n",
-        "\n",
-        "    def _make_label_vec(self, node):\n",
-        "        label = self.label_map[node]\n",
-        "        if isinstance(label, list):\n",
-        "            label_vec = np.array(label)\n",
-        "        else:\n",
-        "            label_vec = np.zeros((self.num_classes))\n",
-        "            class_ind = self.label_map[node]\n",
-        "            label_vec[class_ind] = 1\n",
-        "        return label_vec\n",
-        "\n",
-        "    def construct_adj(self):\n",
-        "        # adjacency shape: (14756, 128) ，用于存储所有节点的邻居节点id\n",
-        "        adj = len(self.id2idx) * np.ones((len(self.id2idx)+1, self.max_degree))\n",
-        "        # (14755,)  ，用于存储所有结点的degree值\n",
-        "        deg = np.zeros((len(self.id2idx),))\n",
-        "\n",
-        "        for nodeid in self.G.nodes():\n",
-        "            if self.G.nodes()[nodeid]['test'] or self.G.nodes()[nodeid]['val']:\n",
-        "                continue\n",
-        "\n",
-        "            # 获取所有训练集的邻居节点的id\n",
-        "            neighbors = np.array([self.id2idx[str(neighbor)]\n",
-        "                for neighbor in self.G.neighbors(nodeid)\n",
-        "                if (not self.G[nodeid][neighbor]['train_removed'])])\n",
-        "            \n",
-        "            deg[self.id2idx[str(nodeid)]] = len(neighbors)\n",
-        "            if len(neighbors) == 0:\n",
-        "                continue\n",
-        "            if len(neighbors) > self.max_degree:\n",
-        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=False)\n",
-        "            elif len(neighbors) < self.max_degree:\n",
-        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=True)\n",
-        "            adj[self.id2idx[str(nodeid)], :] = neighbors\n",
-        "        return adj, deg\n",
-        "\n",
-        "    def construct_test_adj(self):\n",
-        "        adj = len(self.id2idx) * np.ones((len(self.id2idx)+1, self.max_degree))\n",
-        "        for nodeid in self.G.nodes():\n",
-        "            # 所有邻居节点的id，这里没有限制训练集或测试集\n",
-        "            neighbors = np.array([self.id2idx[str(neighbor)]\n",
-        "                for neighbor in self.G.neighbors(nodeid)])\n",
-        "            if len(neighbors) == 0:\n",
-        "                continue\n",
-        "            if len(neighbors) > self.max_degree:\n",
-        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=False)\n",
-        "            elif len(neighbors) < self.max_degree:\n",
-        "                neighbors = np.random.choice(neighbors, self.max_degree, replace=True)\n",
-        "            adj[self.id2idx[str(nodeid)], :] = neighbors\n",
-        "        return adj\n",
-        "\n",
-        "    def end(self):\n",
-        "        return self.batch_num * self.batch_size >= len(self.train_nodes)\n",
-        "\n",
-        "    def batch_feed_dict(self, batch_nodes, val=False):\n",
-        "        batch1id = batch_nodes\n",
-        "        batch1 = [self.id2idx[n] for n in batch1id]\n",
-        "\n",
-        "        labels = np.vstack([self._make_label_vec(node) for node in batch1id])\n",
-        "        feed_dict = dict()\n",
-        "        feed_dict.update({self.placeholders['batch_size'] : len(batch1)})\n",
-        "        feed_dict.update({self.placeholders['batch']: batch1})\n",
-        "        feed_dict.update({self.placeholders['labels']: labels})\n",
-        "\n",
-        "        return feed_dict, labels\n",
-        "\n",
-        "    def node_val_feed_dict(self, size=None, test=False):\n",
-        "        if test:\n",
-        "            val_nodes = self.test_nodes\n",
-        "        else:\n",
-        "            val_nodes = self.val_nodes\n",
-        "        if not size is None:\n",
-        "            val_nodes = np.random.choice(val_nodes, size, replace=True)\n",
-        "        # add a dummy neighbor\n",
-        "        ret_val = self.batch_feed_dict(val_nodes)\n",
-        "        return ret_val[0], ret_val[1]\n",
-        "\n",
-        "    def incremental_node_val_feed_dict(self, size, iter_num, test=False):\n",
-        "        if test:\n",
-        "            val_nodes = self.test_nodes\n",
-        "        else:\n",
-        "            val_nodes = self.val_nodes\n",
-        "        val_node_subset = val_nodes[iter_num*size:min((iter_num+1)*size,\n",
-        "            len(val_nodes))]\n",
-        "\n",
-        "        # add a dummy neighbor\n",
-        "        ret_val = self.batch_feed_dict(val_node_subset)\n",
-        "        return ret_val[0], ret_val[1], (iter_num+1)*size >= len(val_nodes), val_node_subset\n",
-        "\n",
-        "    def num_training_batches(self):\n",
-        "        return len(self.train_nodes) // self.batch_size + 1\n",
-        "\n",
-        "    def next_minibatch_feed_dict(self):\n",
-        "        start_idx = self.batch_num * self.batch_size\n",
-        "        self.batch_num += 1\n",
-        "        end_idx = min(start_idx + self.batch_size, len(self.train_nodes))\n",
-        "        batch_nodes = self.train_nodes[start_idx : end_idx]\n",
-        "        return self.batch_feed_dict(batch_nodes)\n",
-        "\n",
-        "    def incremental_embed_feed_dict(self, size, iter_num):\n",
-        "        node_list = self.nodes\n",
-        "        val_nodes = node_list[iter_num*size:min((iter_num+1)*size,\n",
-        "            len(node_list))]\n",
-        "        return self.batch_feed_dict(val_nodes), (iter_num+1)*size >= len(node_list), val_nodes\n",
-        "\n",
-        "    def shuffle(self):\n",
-        "        \"\"\" Re-shuffle the training set.\n",
-        "            Also reset the batch number.\n",
-        "        \"\"\"\n",
-        "        self.train_nodes = np.random.permutation(self.train_nodes)\n",
-        "        self.batch_num = 0\n"
-      ],
-      "metadata": {
-        "id": "ZYCKM4i5PmPf"
-      },
-      "execution_count": null,
-      "outputs": []
-    },
-    {
-      "cell_type": "code",
-      "source": [
-        "\"\"\"\n",
-        "    This minibatch iterator iterates over nodes for supervised learning.\n",
-        "\n",
-        "    G -- networkx graph\n",
-        "    id2idx -- dict mapping node ids to integer values indexing feature tensor\n",
-        "    placeholders -- standard tensorflow placeholders object for feeding\n",
-        "    label_map -- map from node ids to class values (integer or list)\n",
-        "    num_classes -- number of output classes\n",
-        "    batch_size -- size of the minibatches\n",
-        "    max_degree -- maximum size of the downsampled adjacency lists\n",
-        "\"\"\"\n",
-        "# 实例化 NodeMinibatch 迭代器\n",
-        "minibatch = NodeMinibatchIterator(G,\n",
-        "            id_map,\n",
-        "            placeholders,\n",
-        "            class_map,\n",
-        "            num_classes,\n",
-        "            batch_size=512,\n",
-        "            max_degree=128,\n",
-        "            context_pairs = context_pairs)\n",
-        "\n",
-        "# adjacency shape: (14756, 128)  包装为placeholder\n",
-        "adj_info_ph = tf.compat.v1.placeholder(tf.int32, shape=minibatch.adj.shape)\n",
-        "adj_info = tf.Variable(adj_info_ph, trainable=False, name=\"adj_info\")\n",
-        "\n",
-        "# 接着就是构建模型了，需要改动的兼容代码过多，暂不继续了"
-      ],
-      "metadata": {
-        "id": "mUW98eVhQ5H7"
-      },
-      "execution_count": null,
-      "outputs": []
-    },
-    {
-      "cell_type": "code",
-      "source": [
-        ""
-      ],
-      "metadata": {
-        "id": "Wp3DZreLrdtF"
-      },
-      "execution_count": null,
-      "outputs": []
-    }
-  ]
-}
\ No newline at end of file

From 091912aa7066a720a9d8ab2604bd6db14b8917aa Mon Sep 17 00:00:00 2001
From: java002 <dingqin@163.com>
Date: Wed, 16 Mar 2022 17:03:55 +0800
Subject: [PATCH 28/28] minibach annotation commit

---
 graphsage/minibatch.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/graphsage/minibatch.py b/graphsage/minibatch.py
index e83418f5..e6fb8b86 100644
--- a/graphsage/minibatch.py
+++ b/graphsage/minibatch.py
@@ -1,4 +1,4 @@
-from __future__ import division
+from __future__ import division
 from __future__ import print_function
 
 import numpy as np
@@ -335,7 +335,7 @@ def incremental_node_val_feed_dict(self, size, iter_num, test=False):
         ret_val = self.batch_feed_dict(val_node_subset)
         return ret_val[0], ret_val[1], (iter_num+1)*size >= len(val_nodes), val_node_subset
 
-    #当前是第几批次
+    #返回当前是第几批次
     def num_training_batches(self):
         return len(self.train_nodes) // self.batch_size + 1