From 6aa7d41bcd3069db94184a5d8ffc226fa657431e Mon Sep 17 00:00:00 2001
From: "R.SABARISH RAJA" <80580833+sabarishraja@users.noreply.github.com>
Date: Thu, 24 Apr 2025 13:13:27 -0800
Subject: [PATCH 01/15] Added .py file
---
.idea/.gitignore | 3 +
.idea/ML-Project2.iml | 12 ++
.idea/inspectionProfiles/Project_Default.xml | 12 ++
.../inspectionProfiles/profiles_settings.xml | 6 +
.idea/misc.xml | 7 +
.idea/modules.xml | 8 +
GBTesting.py | 112 ++++++++++++++
Gradient_boosting.py | 137 ++++++++++++++++++
README.md | 18 ---
9 files changed, 297 insertions(+), 18 deletions(-)
create mode 100644 .idea/.gitignore
create mode 100644 .idea/ML-Project2.iml
create mode 100644 .idea/inspectionProfiles/Project_Default.xml
create mode 100644 .idea/inspectionProfiles/profiles_settings.xml
create mode 100644 .idea/misc.xml
create mode 100644 .idea/modules.xml
create mode 100644 GBTesting.py
create mode 100644 Gradient_boosting.py
delete mode 100644 README.md
diff --git a/.idea/.gitignore b/.idea/.gitignore
new file mode 100644
index 0000000..26d3352
--- /dev/null
+++ b/.idea/.gitignore
@@ -0,0 +1,3 @@
+# Default ignored files
+/shelf/
+/workspace.xml
diff --git a/.idea/ML-Project2.iml b/.idea/ML-Project2.iml
new file mode 100644
index 0000000..58282ee
--- /dev/null
+++ b/.idea/ML-Project2.iml
@@ -0,0 +1,12 @@
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/.idea/inspectionProfiles/Project_Default.xml b/.idea/inspectionProfiles/Project_Default.xml
new file mode 100644
index 0000000..54264f1
--- /dev/null
+++ b/.idea/inspectionProfiles/Project_Default.xml
@@ -0,0 +1,12 @@
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/.idea/inspectionProfiles/profiles_settings.xml b/.idea/inspectionProfiles/profiles_settings.xml
new file mode 100644
index 0000000..105ce2d
--- /dev/null
+++ b/.idea/inspectionProfiles/profiles_settings.xml
@@ -0,0 +1,6 @@
+
+
+
+
+
+
\ No newline at end of file
diff --git a/.idea/misc.xml b/.idea/misc.xml
new file mode 100644
index 0000000..e842e29
--- /dev/null
+++ b/.idea/misc.xml
@@ -0,0 +1,7 @@
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/.idea/modules.xml b/.idea/modules.xml
new file mode 100644
index 0000000..9c12a60
--- /dev/null
+++ b/.idea/modules.xml
@@ -0,0 +1,8 @@
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/GBTesting.py b/GBTesting.py
new file mode 100644
index 0000000..d54ae2f
--- /dev/null
+++ b/GBTesting.py
@@ -0,0 +1,112 @@
+import numpy as np
+import unittest
+import matplotlib.pyplot as plt
+from sklearn.datasets import make_classification, make_blobs
+from sklearn.model_selection import train_test_split
+from sklearn.metrics import accuracy_score
+from sklearn.ensemble import GradientBoostingClassifier as SklearnGBC
+from Gradient_boosting import CustomGradientBoostingClassifier # Update with actual path
+
+
+class TestGradientBoostingClassifier(unittest.TestCase):
+ def setUp(self):
+ self.seed = 42
+ self.default_params = {
+ 'n_estimators': 100,
+ 'learning_rate': 0.1,
+ 'max_depth': 3
+ }
+
+ def _train_and_validate(self, X, y, params=None, min_accuracy=0.7, title="Test"):
+ params = params or self.default_params
+ X_train, X_test, y_train, y_test = train_test_split(
+ X, y, test_size=0.2, random_state=self.seed
+ )
+
+ model = CustomGradientBoostingClassifier(**params)
+ model.fit(X_train, y_train)
+ y_pred = model.predict(X_test)
+ acc_custom = accuracy_score(y_test, y_pred)
+
+ sk_model = SklearnGBC(**params)
+ sk_model.fit(X_train, y_train)
+ sk_pred = sk_model.predict(X_test)
+ acc_sklearn = accuracy_score(y_test, sk_pred)
+
+ print(f"\n{title}")
+ print(f"Custom Accuracy: {acc_custom:.4f}")
+ print(f"Sklearn Accuracy: {acc_sklearn:.4f}")
+ self.assertGreaterEqual(acc_custom, min_accuracy)
+
+ self._plot_results(X_test, y_test, y_pred, sk_pred, title)
+
+ return acc_custom, acc_sklearn
+
+ def _plot_results(self, X, y_true, y_custom, y_sklearn, title):
+ if X.shape[1] > 2:
+ return
+
+ fig, axs = plt.subplots(1, 3, figsize=(15, 4))
+ axs[0].scatter(X[:, 0], X[:, 1], c=y_true, cmap="coolwarm", s=10)
+ axs[0].set_title("Ground Truth")
+
+ axs[1].scatter(X[:, 0], X[:, 1], c=y_custom, cmap="coolwarm", s=10)
+ axs[1].set_title("Custom Model")
+
+ axs[2].scatter(X[:, 0], X[:, 1], c=y_sklearn, cmap="coolwarm", s=10)
+ axs[2].set_title("Sklearn Model")
+
+ fig.suptitle(title)
+ plt.tight_layout()
+ plt.show()
+
+ def test_basic_functionality(self):
+ X, y = make_classification(n_samples=1000, n_features=2,
+ n_informative=2, n_redundant=0, random_state=self.seed)
+ self._train_and_validate(X, y, title="Basic Functionality")
+
+ def test_high_learning_rate(self):
+ X, y = make_classification(n_samples=800, n_features=2,
+ n_redundant=0, random_state=self.seed)
+ params = {'learning_rate': 0.9, 'n_estimators': 50, 'max_depth': 3}
+ self._train_and_validate(X, y, params, min_accuracy=0.65, title="High Learning Rate")
+
+ def test_low_learning_rate(self):
+ X, y = make_classification(
+ n_samples=1200, n_features=2, n_informative=2,
+ n_redundant=0, n_repeated=0, n_clusters_per_class=1,
+ random_state=self.seed
+ )
+ params = {'learning_rate': 0.05, 'n_estimators': 200, 'max_depth': 3}
+ self._train_and_validate(X, y, params, title="Low Learning Rate")
+
+ def test_shallow_trees(self):
+ X, y = make_classification(n_samples=1000, n_features=2,
+ n_informative=2, n_redundant=0, random_state=self.seed)
+ params = {'max_depth': 1, 'n_estimators': 150, 'learning_rate': 0.1}
+ self._train_and_validate(X, y, params, title="Shallow Trees")
+
+ def test_noisy_data(self):
+ X, y = make_classification(
+ n_samples=1000, n_features=2, n_informative=2,
+ n_redundant=0, n_repeated=0, flip_y=0.3,
+ random_state=self.seed
+ )
+ self._train_and_validate(X, y, min_accuracy=0.6, title="Noisy Data")
+
+ def test_imbalanced_classes(self):
+ X, y = make_classification(
+ n_samples=1000, n_features=2, n_informative=2,
+ n_redundant=0, n_repeated=0, weights=[0.9, 0.1],
+ flip_y=0.01, random_state=self.seed
+ )
+ self._train_and_validate(X, y, min_accuracy=0.7, title="Imbalanced Classes")
+
+ def test_non_linear_boundary(self):
+ X, y = make_blobs(n_samples=1000, centers=2,
+ cluster_std=3.0, random_state=self.seed)
+ self._train_and_validate(X, y, title="Non-linear Boundary")
+
+
+if __name__ == '__main__':
+ unittest.main()
\ No newline at end of file
diff --git a/Gradient_boosting.py b/Gradient_boosting.py
new file mode 100644
index 0000000..50dc12f
--- /dev/null
+++ b/Gradient_boosting.py
@@ -0,0 +1,137 @@
+import numpy as np
+import time
+from sklearn.datasets import make_classification
+from sklearn.model_selection import train_test_split
+from sklearn.metrics import accuracy_score
+from sklearn.ensemble import GradientBoostingClassifier
+
+class Node:
+ def __init__(self, feature_index=None, threshold=None, left=None, right=None, value=None):
+ self.feature_index = feature_index
+ self.threshold = threshold
+ self.left = left
+ self.right = right
+ self.value = value
+
+class DecisionTreeRegressor:
+ def __init__(self, max_depth=3):
+ self.max_depth = max_depth
+ self.root = None
+
+ def _mse(self, y):
+ if len(y) == 0:
+ return 0
+ return np.mean((y - np.mean(y)) ** 2)
+
+ def _best_split(self, X, y):
+ best_mse = float('inf')
+ best_feature = None
+ best_threshold = None
+
+ for feature in range(X.shape[1]):
+ thresholds = np.unique(X[:, feature])
+ for threshold in thresholds:
+ left_mask = X[:, feature] <= threshold
+ right_mask = ~left_mask
+
+ left_mse = self._mse(y[left_mask])
+ right_mse = self._mse(y[right_mask])
+ total_mse = (left_mse * np.sum(left_mask) + right_mse * np.sum(right_mask)) / len(y)
+
+ if total_mse < best_mse:
+ best_mse = total_mse
+ best_feature = feature
+ best_threshold = threshold
+
+ return best_feature, best_threshold
+
+ def _build_tree(self, X, y, depth=0):
+ if depth >= self.max_depth or len(np.unique(y)) == 1:
+ return Node(value=np.mean(y))
+
+ feature, threshold = self._best_split(X, y)
+ if feature is None:
+ return Node(value=np.mean(y))
+
+ left_mask = X[:, feature] <= threshold
+ right_mask = ~left_mask
+
+ left_node = self._build_tree(X[left_mask], y[left_mask], depth + 1)
+ right_node = self._build_tree(X[right_mask], y[right_mask], depth + 1)
+
+ return Node(feature_index=feature, threshold=threshold, left=left_node, right=right_node)
+
+ def fit(self, X, y):
+ self.root = self._build_tree(X, y)
+
+ def _predict_one(self, x, node):
+ if node.value is not None:
+ return node.value
+ if x[node.feature_index] <= node.threshold:
+ return self._predict_one(x, node.left)
+ else:
+ return self._predict_one(x, node.right)
+
+ def predict(self, X):
+ return np.array([self._predict_one(x, self.root) for x in X])
+
+class CustomGradientBoostingClassifier:
+ def __init__(self, n_estimators=100, learning_rate=0.1, max_depth=3):
+ self.n_estimators = n_estimators
+ self.learning_rate = learning_rate
+ self.max_depth = max_depth
+ self.trees = []
+ self.initial_prediction = 0
+
+ def _sigmoid(self, x):
+ return 1 / (1 + np.exp(-x))
+
+ def _log_odds(self, y):
+ p = np.clip(np.mean(y), 1e-10, 1 - 1e-10)
+ return np.log(p / (1 - p))
+
+ def fit(self, X, y):
+ self.initial_prediction = self._log_odds(y)
+ current_pred = np.full(y.shape, self.initial_prediction)
+
+ for _ in range(self.n_estimators):
+ residuals = y - self._sigmoid(current_pred)
+
+ tree = DecisionTreeRegressor(max_depth=self.max_depth)
+ tree.fit(X, residuals)
+
+ update = tree.predict(X)
+ current_pred += self.learning_rate * update
+ self.trees.append(tree)
+
+ def predict_proba(self, X):
+ pred = np.full(X.shape[0], self.initial_prediction)
+ for tree in self.trees:
+ pred += self.learning_rate * tree.predict(X)
+ return self._sigmoid(pred)
+
+ def predict(self, X):
+ return (self.predict_proba(X) >= 0.5).astype(int)
+
+def main():
+ X, y = make_classification(n_samples=1000, n_features=10, n_informative=5, random_state=42)
+ X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
+
+ start = time.time()
+ model = CustomGradientBoostingClassifier(n_estimators=100, learning_rate=0.1, max_depth=3)
+ model.fit(X_train, y_train)
+ custom_time = time.time() - start
+ y_pred = model.predict(X_test)
+ acc = accuracy_score(y_test, y_pred)
+ print(f"Custom Model Accuracy: {acc:.4f}, Time: {custom_time:.2f}s")
+
+ start = time.time()
+ gb = GradientBoostingClassifier(n_estimators=100, learning_rate=0.1, max_depth=3)
+ gb.fit(X_train, y_train)
+ skl_time = time.time() - start
+ skl_pred = gb.predict(X_test)
+ skl_acc = accuracy_score(y_test, skl_pred)
+ print(f"Sklearn Model Accuracy: {skl_acc:.4f}, Time: {skl_time:.2f}s")
+
+if __name__ == "__main__":
+ main()
diff --git a/README.md b/README.md
deleted file mode 100644
index 3a7d465..0000000
--- a/README.md
+++ /dev/null
@@ -1,18 +0,0 @@
-# Project 2
-
-## Boosting Trees
-
-Implement again from first principles the gradient-boosting tree classification algorithm (with the usual fit-predict interface as in Project 1) as described in Sections 10.9-10.10 of Elements of Statistical Learning (2nd Edition). Answer the questions below as you did for Project 1. In this assignment, you'll be responsible for developing your own test data to ensure that your implementation is satisfactory. (Hint: Use the same directory structure as in Project 1.)
-
-The same "from first principals" rules apply; please don't use SKLearn or any other implementation. Please provide examples in your README that will allow the TAs to run your model code and whatever tests you include. As usual, extra credit may be given for an "above and beyond" effort.
-
-As before, please clone this repo, work on your solution as a fork, and then open a pull request to submit your assignment. *A pull request is required to submit and your project will not be graded without a PR.*
-
-Put your README below. Answer the following questions.
-
-* What does the model you have implemented do and when should it be used?
-* How did you test your model to determine if it is working reasonably correctly?
-* What parameters have you exposed to users of your implementation in order to tune performance? (Also perhaps provide some basic usage examples.)
-* Are there specific inputs that your implementation has trouble with? Given more time, could you work around these or is it fundamental?
-
-
From 2611f864cf12ee76c1cf2c2c99146fc35e1a7ba7 Mon Sep 17 00:00:00 2001
From: "R.SABARISH RAJA" <80580833+sabarishraja@users.noreply.github.com>
Date: Thu, 24 Apr 2025 13:41:41 -0800
Subject: [PATCH 02/15] Added output images and test case
---
Output images/Basic functionality.jpeg | Bin 0 -> 88915 bytes
Output images/High learning rate.jpeg | Bin 0 -> 87192 bytes
Output images/Imbalanced classess.jpeg | Bin 0 -> 80439 bytes
Output images/Low learning rate.jpeg | Bin 0 -> 95072 bytes
Output images/Shallow trees.jpeg | Bin 0 -> 88324 bytes
Output images/non-linear boundary.jpeg | Bin 0 -> 94836 bytes
GBTesting.py => test_case/GBTesting.py | 0
7 files changed, 0 insertions(+), 0 deletions(-)
create mode 100644 Output images/Basic functionality.jpeg
create mode 100644 Output images/High learning rate.jpeg
create mode 100644 Output images/Imbalanced classess.jpeg
create mode 100644 Output images/Low learning rate.jpeg
create mode 100644 Output images/Shallow trees.jpeg
create mode 100644 Output images/non-linear boundary.jpeg
rename GBTesting.py => test_case/GBTesting.py (100%)
diff --git a/Output images/Basic functionality.jpeg b/Output images/Basic functionality.jpeg
new file mode 100644
index 0000000000000000000000000000000000000000..244f25d4a0da2bcc1ed0ef263637ad6efe60d22d
GIT binary patch
literal 88915
zcmeFZ2UJsCmp>c?r1#!Y5Tr|2DG`(|RRIAZDk3F9qz4ED=^X(91u0Slq=b$XsR0xO
zq&FcDno=a8ga{%0Y6r`(z
zN&g2>uu!sIkk>lTX8w@sq9?n;^OU#LBHA@w92TP(QN>4IVKlUyT--dom&C*`OGqjy
ztEgVRcKzlp9bG+r1H(IaEv>BY+1NThc6#FM;_Bw@VJkiH%E5dzqe*
z`Ra96L19txyOPqf_qBEP4UJ77np?VikiC8VpFR(ajZaKYP0!5EEu&Xf*VZ?_|JcOt
z?H?TCj_|*Zf73+by#t^zGA&eFI-yr^YBP
z8?9;++xk)(2cLhER(Uf$2BgsxhWV!9?D3xHK1^;NR=nS~FVD@F9DCgcrCi*P7%r!QqUeU-T}6ASL{Tw`w*0YwqT^EWTTqSx2F$hO$62l5z!E+8g6aEYr+fN5#BE55Z7jAl&9qs2=Kc>DNS
zWngKd!^b>s14=O;d9MY$ShfaRtbO<`!VcJ40PwC4kiul%MTM&>2&L|~Dk
zHVcn)(qsMMF`GV~h
z8`!9Ex@jO?yVAhjXH8!}^#S6z-qcEoxSuH_zT3P6v66v~P1
zg|8t!rcNx)H&+Rj#OxGCKCvsRiL(yzohYF;XE@e&dfBu$IlLCnxNkpN)iA@hZ3%y7KU7cIkZ+|^?Eu&6&4AK6Af}T
zbI{y!Q2o?fH5fBqkfUfy;a!
zu&}tar(uVYFPjwX^w-r3N7i5HKfXI3{s9oYqZ42q)~>L1jOYzY4$sxZ$TUSYtYL~X
zI;vWIXD8
z0?Iora{@p6DuYvz%cyPh;xNr#qxBb<0SHFxu$z(n)&AGUfNDUK12V_8ghC<`{I4@@
zk5oVhO|d2M9Zf!oy7S0mjhxnA9Nj%Y%99Ul;v?kFN4Hk;m3u#*ND=HX{oDJP%+9Qi
zh;8BpTRa4LJ-1;+?%WJms$_1{+@WEj#;#YoBIQ#?-F=Y&3x%RcrfS58(tM^wO)>&A
zJ{Hpo!tbEt;j_uzS)B1J#^ODrE`l@-lNH3z(CSOu{(jS}lJ#H-o5?rzA?6AZR?jL9
z%P?$99Z3W^tg{&w7}ZX1PvFIfF7mb3wqxpZ#iPx`?@m6>P>Dj**#$gqYG5fCpexf}
ze830zjHvvc&y)b;!bq*;^)N-l>qe5*HI{)N1z$y{|dq
z+nmp)Bg#@u8FoFBud6p$4}(IsU&UHQYL=~_qQ;>Y5A5r$j-MV^ek2Teuux~{HOa8b
zKU#Uzx+HwmH1YNjjN<>|RQhlU99o
zB~90#gVz56s}?yM8WGhl4b{U%ekEQfsNg-%05oek{Sj5L3+1k6Zn(MNp|X?5{$}gW{2KQ@C%V|tgfAuSq=Ts
zePZ-_LCoJwThINNh9}?e>qaf?
z7FOH&&fHUj?1`6<(Yx#s^6MttuV70tT|$
zbh*l=Z3twt&Rsh(9DY&Fz*;DE{n?U=ZbvtSfnbHV!w#ZN()=;q8+S2qq!33_><-M@
zwcMscMe^8(MM-h)^TltRtn=>Ym7c*mO#oMqIpYBMx-KA@<~7(=y#0wLbVLuM(^$3w
zityaBvh9dipLAae8Wgx)J0RTh?JLcTe#phB2NkLkPRfv8I0f-4%p3#RVZhQML&(E#
zVy%$~s-9(2dZ@847#zFQ)EE_9>Fre^LzsQ{!7+5r+)~RiRd1FwqLDCTyz2^t*%U*0
z5X?0KqQKXV-wFFE2e>#pd~|F5#Ts7Z6v6lTaK{b~HrxyEUp#
zH+luZ2hB8_g7`4Sv~zBB&oRPpk!A!HX1Yg?
z|6a&wJ1ur?J$CK8tzNfX9{r;I={vSHHwgTpWLLE&Za_<7k-cOvaaLGJ6m!%$l1QY&
zvz(}XyGKwbPGdpc?jbAYQ|d}t4MmE^5Ovk+rnF~!!EajPV^HUBP=ER)!hy?SC&TXU
zG?R2LH%t@ni;eHJHNcD@yGm6fc9hEJON(AH1TItx&a}~g7ELk5fkVw-dq3?LebSzT
zcU{upTZZ(EFc7X{_T0zY6pQCw{Wd)4D&!*;Q9JUKu|ExzZ`$LpVK$Iw02Uum+@dh%
z1clG+dGYJ|-F?)3<&P!DBa4?1uW_ETVlM&~L9e6wwgFVGHz?W7|7w4Y!T+udz|xlM
zDv4vcqWHp_o-B_S%J5m@o*XlN~B;V7+kNTemR9RS1aUyq-H{1?7{xBfrhv
zRrOcal8TAQLmX;OZutl)bB8oR8A9X<(xWo!T4<)ItcJCo@2aY0&Zi|0gIsXa^1X_L
zScND(<5h_{EfMw5M`wT@o6Q!vh<2%tWFp(p@(4fSGG-s2wcAr-W0dV~g6xrgG2{-b
zVfeH|ky0anF5$}EgQ~4Q6d!zRyOGnr7=@>&$FZ4Js3F>Pr&
zv8>fS17LsBM@u4a-yTIPaLfRF=8|_jv65R>_`HWpLbKf)$czqJ-|R%7bsL23=95fj
z9e4hdoylb}qcDWngXJ4{HC+`b1%PpIH#Q|8V?8Py$W*27j|RqO$q_{79P(s~vL36n
z+=vqgx3)$ZsK0%U)-g-4nMsh6dnUc)Su>`-zxygkh!Kvrb#Sbo);dkAl|9eQW0
z#h)jYC#RruaVHJT&@6n8W*<%i6U1*|f{gKOShemm0HY&THAejsW(pWhl*Lp}$+%Zp
zsMy&=G)}VuXBP@`XUfUNctPP2Gt9m(s!UWo
zy^z5js>@kDP(!~WvqwOt1}J|G`O;2@?<$1p;)h7{D?XQ{4ntVHKCLvQcROs{JPu1)
zZ~OF0%;1G#JdXNXy4i5T=VscPtqwN?1wIdLy1jzs>U5wR2iE&81oa$QT@NnF;n(!}
z`N|R{A}MR?_dw%2Ir3nHV@rrEL<$dHhWG3;5Oncx(Le#b^5Lg!WFAY^&!;=KGzCm;
z4guvJ5&Fn4ub)OFh(JCa0i%Stl|cf~lRgw}*KT
zmq0D@Wa%D0QH>#=rx5+(Uo>u@pl@YBit7^
z?$8UrF1kV`5ybzV_lA7Bs4B-+t^g(v31JSI$fD}|=-OAtUhJ?
zGOMy}>uEO)C}6L^Z!x=Yg|Ds`;Fou<{fYrLt^4?0VHwn9rHA(@>Tqy`VQWsk$-qY%
zu``f)HwR#olB%C`|MDl#PJ|vp=ror=i5?-}0&#l~?i8KI$#KB+KTaQH?4#3mR9Ibc
zETSKDFDYOQRobV1dFfJ%&-`DFVUcez#RqOAwo8PJ@#<}lF6Z%Y2{K?#qa)_~LM~4n
zT(e>@JQUoXo15u+rlqoJBbutG)%N~H@FN>-u=UA9KlmS*K
zQkj9X+g{JUX+o~3y*aYbhF47it4DiYl!(oV6_O0~BAdj|uRsVFrhv4&@MWG8XM*$y
z>pf@?WFFSBO%lnOZwmpldn4IR+w&@f*QhrM5B6+@b~vzd{UJ(N&P+`I(rw$LNP+~~
zl_2vT%%v&aRNH!D+4}x|o0Tz}M)bXhL;E7XfeIY_{En7lwE)tF+)lVVRaRczKuYc={R#e-a_&mGIAA>+1G
zoyPt+xv!c$u9>!s#jV=HWplBv%C-HoF#MaPgM7DWSJaDd+~OM>bv>LmSaPqPv~GErvr
zG1E{K$2?93--StsNt^*VKykhhj&DD%!{;kI#vI!w#Ik1FP>O@b*p@eH68D#@q4`Wz
zD?aI*_huBX(pNBQQN2Zk!yS{8h?gXm{pii|ZxWViFRsQKF
zC=i!|-5`j!vtxjLdC#puOz>sRB+up6QN+!%q9%E5XRum>{`hv?tMu2_vK5vuRdhTx
z)lXj&g0W&7xWZAXwvz02k&35rZeHhKb*6cAF#?)176K>Tzb+`3Zo3~5d@rOADBjE*21(g#Adl@pCOaa
zM>cbtoL)ozN-5hBh9Xg>3x(I_c|-5z6mGczE*Hf8Uu*cf6*$uVd=E*im`LO3vYVKz
zh6gQ`9xNylnQeTGb+`1!8?HUpQyu#TvO^3hc|~aPr%Fl!{Bn>}sD2$%{4|OnbfRqP
zik02yZ4qedtyI|qvbnVssk(c+Jyo`$sI?cqw|eVly}{`OMXHzSdx8c6;jpz(xvH``j-akoKaUSrq%TnH6O1=&_!uDodNpogSrQo
zMo5D}AHtu^2knN_*lbN_9FbBW#u+&!FD@k6RN0HFt)|#pC?&@3eap#zym#E+lQ%~n
zGLpv$e6j%2gLB==W$qXwFrX2Qb2nux>8D26IaO=gVjNpz(r;>x(#J*ID|lsWUZIFvCsa(aLa|`(VzmeD>NZ37?6ty!pB>ri+`c6ZdekKLvrFakXg61D5m!9?INSDUFu#_
zeJMp7b_S3?m)W^HyW1zoP8$B?7+&k`A8%ojnA|$drakS5(YP<2#5?z!JJ0uhS5;8T
z({Hu05x2f%$C{BLN8kpt+H*h}gYF!z*0=ysV?4j@Fy#su4F>kMk8a=gn
zIQYfN576`y=tEv^T>!Pl#bMK@d*$c}n)qal#Zgy=gL=cq516V#*#?HyCOLQJT%({?
z=-1#>i16!k(Jx>Xt+fmReN7l4bR~+w_6W7a&=g4y#i|QA7B{BhwjRcpuEpl4uZX90
zTB$YdTwR*3B+YZKcv5I}WC@bsV=TUN9T?t_=(T0rypJ}m?e#3YKi@si6R!EANT>U}
z$hE70EI{9BOT#WR-Wg|s{n@k(V&2T+EQe60&TsPCf6HjHI@zrpP4Mcwc|r1v0+Y5*
z0AmBz5v`2cZYm-0hxhRo0Xf#Nw^4Km-&eBs5VQ*Y-ooai8~(fQvP|`{3XH$LZiwf3
zomLZtpdIME?d26AGANTiabnu{-uBek_zsKKI`{5_6k>pl*+Z6O>FW=?pIiC@XTy6O
zSE6I!E_^a3M0dNpUp=HRjJ_(ZOCsL8*c7z({niKsLQj#a625gq&=Zq}i^l>ZJDwmy
z1ATbUBd{W9Yh)s;e97#B+oI6|w+yn$KhwXCN`l{Qu=$hU6CW}Ibw59$hMy^gP(`#u
z9Tw8r5*<#^!{o@UP9Jbrjhic88e8w(-nbEe{hR8~amk3Ql3kuz`^mHf20RM0vOGc}
zBayrlcif+#$@-lukh5d?nzCI`fJY_59c>W7rIUpkjx&P3I=3ufl(-THj{!c}jXwkM
ztIK11dWlM7BmhGJEktWHM!GInGTI${t8^Lo)PMCuNe$IVMhm
zA<4NLnUlgn#@ZzatRFBO6U2*?hkWFM&5Y&rH4Kbc2MGtIoAlowC%o*Dp!aKaf9fA{
z@w7_ahF((=8yy-#xd1I+J_>`#Jw0@XLgVJfHkW1n8$Qh<+;ptI+*MA_S;nXY(a90q&$p+U
zC3Zh-W!B5Qy}ehBuH6b7lely>j^<6y2Qo(E0MIs*KiaNi7Thg_Np;EKg}jt?tUVFS
zpN6|M?thDYyucoEXIDYCtL-626Rg)NIho{K<}tvY;zcUKKpp}Jovnrt1Qpqul)D$}
zf!*wSvIeOVE1xdEN(bblL>MZ(=pKerxsrlnMwn@0BqFW@@788~2KdAd@39~#p8?*2
z(Fh_hf(^b$8F&8UA=p;#)z=k}M+KQDFwx2(U$%9)PWZw1I!{ruU^jZfuvU)GV{
zGZ2}H)Ul+Gm!;L!$y{CO8?QQc3t&Y*Qzk
zt|3RHXHlGlprcmqOv_Xuq$`y`3~twvBb0U8b^4#3C@SJAy{RvS^1Pre!|=2jy(f@i
zZ1GXU3{-csdopSqO+IpkQS52=NE(PWQgY{n*Y=
zWl|y@-x7u@w=~7h*2GO
zfVB`-6}{Vo2+z{VXvK;kYg}-Fi3D4xLhegB8_j8IERL}@BiaUi?C0)LS;?ifKeNIXug=>w_K@~x9-r_E{X+yR!K1x2b&qcNnY@
z-BO+{#0pL80y_njnJ&xt5&MAk_g%(&pvS(S%^ouwc#|cWRW{%J44}Vuxo8Qx(X~ql
z4q-Vh(Y*AdKajk~8dpFI`EGx?uK%C#+&LiHr^O^tAKrUiF8pBwwlSv+AGdC#p#*lIoH{6#6*mg-;FZ@{
zDDw2!%TLGHiC*Y8JT%;6N<{5hw4Z~|-e^4oOgP)+GlBkgkq1H;zi&G2%;>o5Ok@an(2{-$4(99Y5TQ_44`rx*QsEMc&sq?N0PM4;nHG$}Q8
z6s0MTPrHMkeh-#g{^9=4D7%W{lovd|GS$0OtQngaTF)F}
zczSPl(8P4>z7Z`K%}#Ai>_|UA?FFA}%fyjNDAO$Q{iJ+|;KwsS5u0_Tbt-WhxaZcv
zeFoURIo)v(=Js`u$+F=Lkn95A$~pr`W)%GGmr&Zv1);QiOl)U>&vUSAXMkGx3e)LT
zXvoKqws7PbV9zg)5x$*=C#A$q1cm;Nz5l|h4;^u50B%ho0vo0i3qjgHbDd%}X{_Ff
zEzWR##md4$CjG3!^Z}XZR7q%C&@wP~k*ylr5vB}1lo7Jf#+bwncvV=Yr-wQ@J9r%r^SQ|O!2_wzy+{~Npq^u_Po2J`N@4E
zRBruKbWplBE%`9_9u>v(l#-Q4miL|Zenix;Y#y;?q%Ysy-xs+{O$s~Mu{F^)f${~6GiD1dmXmAyfxe&K#u=YYswf$4q1#7R>JhSu0
z{=A&cMbK=W8Z77Y(wo8WHvO;*w)eZe(r#_?+zVpXxg@%fHu0k@969!u|Mrgt6KnG{
zli~Te2^2ZpeZu*09Yh%=jDY1qG{=}%f03`!#CHQE@^^;eiVbn=S>z)A_J)Uc7I8L$bj4V_9W1%
zSoxbjFpo1pYC49EI6#Ra7G7U)3HVneYfeJ5<~m-vno}X7KR9u4OeKM)5TCK`?CI6=
zu6^#V%6CB5V#YN6>YD@#fMePX0sBAS-T-5Vegou&F{|He;cw9xYA=QtqHG$v?-%*|
z`778Bt(oLWVIv;DS(Dwy6s28bw?tW9bUxFfcTQj6VfIqV|5ExAkpo7*r=L9&9tCOSHJ0LPPsSppGDoiya|Ph
zf1nZo^L6|3f><>x?|j+&@-8xPH=D`|_U+A_LGSE>jO7lbKS+?}gdt=U#0qx-GC;HT
z(SnJtJfj{v==zDLJ9FGsZM|OD2dQ-ZH8jEp>0WAiC6ZD>;2`}7m1;fu2dS*ewE4Mh
zv>8<_o^clb&hgnRSMUV)+q#@_#Wd~|KQm$>POZ}(Z*|x$bD#YyD}S@|pJ_LPd^r-|
z9F0fG&m+e^SPr_I8Fc=H=DVtF$Ce^8I-bDilX`H`__Q(tX|hI~3ue}Hl_BPM>lr}oMzhXm^Y!)*AouJH*I6?Ch82ajShJwJTZtluDZ^jR_oUisL^1YUl1obqVz`r
zfYjf|+X5#8ZX(331p}jV!3zI?MSuh
zdP!VIt$IQ8fbfZQSJG(C@M!6X^fh7fj|gEbkdRZ-lXoIcymAH@kB;7Iz*}SKZsTh(
zs2(7jlATnmnh@I$zvCj#BR5W$)VuT?)%v2&rJ{oQFLXQ|NU^oFdOYbJACn3Zg^r(L
zP64ER#{aYWUhv_GR>(*@<7ojw0J1E{xqzAtj@yAe-IdO{RXxC7w*1`6qC$%zkXcu1
zSo8A7>FqlwQm2_q!--<=j@7X}xSSh2ghKVm;?`GUUqc-0$s#G!O1afRJDw+xNUJ|M
zu;Di`a^dP+C5x`(b6Wl)4t9g0EolZK5^?8s$vm4pZa988=%?Dgde322cUq-b9sfL)
z2XM)s(K7nx^!e%|NyGasM&P`MmtL*TNV8*cWf5;a7P;=b?0^^N%1hQ80v*0%Av3!x
za*4$2FoER0c8$>oH4}HtbNjAUqFvs$M{-c-u%|t5UQ`Uz{hrPJU8DOoySWx`4}~kg
zi^_Xpz~uvgPsvoZiM&B;p6V;~Wi-}T>g&6*ZsP{_p9QV=Nz5-y_(KBzxpOlrCRe!z
z@3~|Q?{7&jIjW_MAxJ4+R!!y~S-Xj7yk+4gkp?pq?q-`*K$PP2-=P)z)jr}fHN>Bg&-{MB?|vwi
zs2LT)`UTXHXflpgJxx0Ukc{k)9wf)+aqzZndSqkaxlc9r+Gz!!BKf5*3oGPmb*&1f
zuL+W16iHSS|3y~2&H%^Hu$2Fauu7fih79DMJ1v^l$?mnyEiYdD{zNBi%hZP+Yf#_z
zw5o!7AdJb(dxGDZ{KOJwjREsUT4KDR8r9Cx8wRhUP#>sNMi
zI{q{Fx_ZEg6Tt>Evg7ZJ(KErwBrFQKTMXWI$)z#g`QjZR+zu+pK(<@XBAyH!M5ch60p
zUtEdKGEiLgw0!8$wmB9p`gq>lw)EZ4bV4Huth^5W=hWvPDmMVtznj1bKNrI3Mq)8S
z$+ZEry(@Y3=}Ow*zEg%yBDp4G@8IJ%f>eMvv=(lX+bFuI@7N4u_ubAJka?2>}G2{>hU6TXgGpAL}fFXP7bQMtJx1
z82})Lugu>p?x^=!{V;C%&}B{~qRo0=hd;D&CH>LeK1N1e4aqpb9%MGTGenM5&i%Pf
z4L#3YUv&NB>0MDk+I7k~lasZV8uM!&H%COv9BDoLXf4SUB6)v>Gc73R!~IsS
z#A1s^!R(i=<)4d-to@IG%{YzjBh%`Mf_uJvd_kq7$hb6Q)!S%YKvLmU1xX3?`%RL-
zeZJP;;TN+1^gC%89Usp1v?CH(YeD_^brXM^wsVj_(RL|Y(SL}6!~mVjG^M{k
zVezM^gctWR6#v*~qi;&<7rk|X!id=q83CZ)m?bTF)xb+Fjd0@~m-3=u3|K|GU?Wi7
z80VlxOH*5VwkR`;ho%ir3T=L7^rLLNSlhTR-fkHYmDi9NMiuQV$3ep2E_m|ZCy*T!
zhg;wc>2
zB%?l7{>bI|dML!Q<*z+&GFB6J;o^$FSmia%<_g7J*`7jS{my13UR2~(i{V=F+
zQYtYlwmZ4n4J
znu9E_!~;j~#6w=!vOTUSA_r8+
z8Xu1Su@-GG`j*Y)k@24V6Y7}wKAW6$U45R1sZ`3J2!PW<7}<&LciyPxrI-dO>#-`%
z#k~>pNRh0Ioi8CNuMMji!uWs^pE%U97HjP#PZhT9Z;RG^eYbGA^_tx+lo4Tiv9)jv
zWm=tjs4QO71pe@%FZ28*1RPUggqEStuu
zOD)+upGY}O9Q=^k_8{~nKaR%xBP2dJ`7`QnLG6R+xKB-B)5VGWwBbxI*7$Xg@fYFo@>8zL7*FM-oJ7Ax6|nI2R`_Zvv-y=5*~ewij!+&E5uLV
z`iMQB#l7g7afCJS1z3IuV{oWgC@7q!>omNi6M$Z{JXUtE1p3p>1gH=hB0LJSSYh9NjWyz>D-`
z@QS_Wt1p-vX5r<6wDk)GBvFZrKmTEuz9-&;m@l-%_U~Nt#9W6*$AXL63xkU17Al54
zR#Ip5Z>N^=`8Bd*b&4#L(e#FNDTmiLQK0-=ku?f#J4&MCvtJ7J=Jl
z+iH_mcZY2Z)M~!69a`RU({(TXm~1MPw&FG51Ag>&khf+%yv5)CAkd@UFOU@B>B*bb
zTgvN32%!iD80RR_`YE|v<$@;}Pu|WUMOV)rCQpJQShgh~9{|wAeoX(_J
zl7rt-*;RDT!anaA$rVpB`(q&f{ag#PMt3klg?h50y)_i-evIg_m(y63!gb>
zAPxY~h=!7hHRnZu_%(T9-lJ8ABTnl=@O3F+m6=Py
zG~+l&z6&j`eXn%fo}U?^Hu#l&T$;XKzHhAXnAGdvTcXq82%q)x{C3_pJ)>^Q9lJ06fq
z4xN${IvP3>fi42xJwQ$$sTr6WyR~ruL^HJ!<7y#sI@WXL21uNi%=ueU(^VPJJ%Vyx
zpDK_Ers|2MrC!FWXW7@eVe`#6xBJ>%x(WF|hqbY9R^ooR8p^sGsNMVCwerL1Wz9Xq
z`iGW7US%Hxcp;17`@f#H*eg`K!-AH;yX^O&7YF-nFZIV%vVHk5_GscF_){ie?JdJc
zU265Dl?kguSmF|BhP0@Nk|nfH9v{;+MY|H
zedi4*)FusQ$(aPc1I^}wIH{YT{w2ivN2v90&r)G{
zY#~8=n#c?-(ZgGoDrd%3RT#RCOLvPeG#kCw_pF&Ef8VG(^}tI$WffG7c;4Z{L=Ke$
z)#EpOOCy`xV-Mmrbn*80Z@je|e$p}dcXoWGtQr&sSVa=Qsw}bB
zzIhl$?$kUycRU>DUBVkTrZC|B#zQaWcbfsq+
z#;}yf?sh!{fre0h&uYM>5}hK4aCz)|b1GU3Y%QtL!u)0Yp+Qz_?B&kNkGex#2&0dJg`^Y*`Pq
z^PRT(Aef0`bp%R5KugqTU%MPefqM(#Quf+5MoN5F6I!lN1qeTk*`%7ZWG>^)T+@&q
z5${{wi;%Bg^NvfMRO{V7@K{W?s46f1alD=xP_T*fT$tJ9+_PP<8&F6^Zd#5;KfC)P
zq+vbpJn#u90!}99Yj3o?80!h~FMrOzzhJNa^F{cFTaoRX&Uts9r#_S4tiR)FyijQp
zcyHU7maWI4bn@DKmPD(XyYEyN$vE)8{gV^pg<=g6!%Hm-Mf`+;S594(h9qY*6ffwe
zEb3mpd&|sbX1`N_OdP<CqR45j5+Y%dzaD?#A(}@0
zD()+vWAVLTxUq#_(Om#{mkr85xdRP`^@$Ho%aL%llx%4i_;l0U7)hD~EP9;W1
zXMnJgN`hxFT1K%ekXXFTa82%T-dJ`sSay`S(>sUk;~o)-dRADvHDNu7s_yTF&sn2Y
z>d=FbL_gbz7Ryv!T$BHt#Y~&3&L^-BZ4AHrXW_8B@Ap(6oI;*@Fx2brm^)u?ce$C)
zl;pXSmGLrv?UK~Cx~#JvT__?fhkDpj#p;ym>fsWw_BWM
zoB0S!nxmOGf4GnZ$C+^XU
zfj~4d<#CA57BpUZ>FxfLpyK+I-LIsXM?#Roe;~*|oe}J(2+~GD;k{=7N=Q8<%)jXh
zxYC~@lV#4H4ji8?-})s^g2ukWKo@vrpj_@+m+8z&H;Nl3gs~me#m@(Mt9)om`wZb&
z_}cuTu}%&4q38wYx2}kU^DB`~4)V|Q)HYU|z8sSB06*ZQg-R|b(0%W{6SWgu!>`JD
z%zk@xqxI*&-R;YBBRg7D{co`}=y5Kars>u3_AkaGa~GUESqEDMZG8AwE|`41t{m*D
zb8S4)$lFImdf2mB>Aex9ruEeigSUR>o|HiByW=jGtJg7ZbuQ9;#42|r%E5#ROwn|=
z_9_oueehX(>|J~ok6I@l6iN9r
zpyk*)T>YEm@neLrr~)R}CRx9_>|oc@YsdRVo{iyIo{S$KJj&~ksLG41XRynpc!
zMDXp^=GNSGZ>8}M-1c9BE`I_5&KqnRXh!O2-bE=lTw6{;_Ap5EyD1x*{8WWiR#m#g
z_w2*sF2BE~{U=botSJkF;K3MGordXlQGBRt|+!uNA47k)Q<49
zcsF@XO~PicvB_Y^+aXR0oh*n?!+j0MeX1U7Q!W8lF5N7ht+>c1Gk4Q9LVQlp)o+K@
zaY)>oiXpV4lm2em_LyJuLfRVZu($;PwSLU`VmTjd335uTUE5tTV?G!amP}tfZ
zF2v0l`|{lWL@Qb9q}$2JmHI`&4||Z#epOXjiV&9Jteb0dN6zQshJ=J<@=Bc5k#$-V
z`e^J+#=WJ)^(>vFs}pst{FLOW1{#!(&^!H_bl8QWqeU95Um{}O&hSB&yTY+lQPOQ^
zF!xNT=P;8S0xRB?0@K06(Q;%Xr9%&Lg~w%;)>~qn-tL=KEh(rvswezDrlUjpJ8DhJ
zPLw{Q?e*?s6L^Of+k_)8)<+gbT`iEksq)k2v-dnL+3Lpg0YleI`46jK2j6aKyx0(8
ziv_3hSMH0qb{K`^I*--==h3J?_L_qYo%rDSR?YxP>TLOQF#47W{l@l`!yt|8c-wC#
z#nzu6^k#ZpU`k(@e6h6i>98ZT1&GujCx!AN>mhM-;STusQ%JEw#V=yP^D_l);c5$x
zmn&w~UK@NKHgR(w>O(0`SEi2_XWaS10c|!hks|K_l8?_T_)w7PFaF;)Sp%w1Mit4yy#O%}C|$Kt$7c5^s&mD
zbei|3lBFlpTBcjZ{hNLx=Mm|idW>qA4=Kp|4L{t|b$qEQM()mF^W>v(w9)6x{*nVa(74g2qpOEG1V_9
zNW~{wtdE=2CxBSZ`8YL#)gd;XdQO{DH-O9+;ZMGbhvQq(&b0-ZE5zqR!#Mj#+c6sQDs_VtXL1Akr3w>^BveoGpsFW!-DKr13(+Yyw<#?)`
zPvxXiTrcf*qx{CIWtdTXlm!*VuV~Hi;s8|NrvlU1JTjY$I*!Cd^w$F`55Y5l22l@Ts+^$Eyw={wFkt8~ws<`1lIITO`gXaXRvm<{}g6+
zl?`2N<@ufWU9bJ${q(zPw)KWk)t1}vQcAn1`O!#_((=>sOWplkNhP0oj53DG7pEuv
z{Kkp29sZWPhja{*1HndjKQ`6XHzo{f^djT=J%xFszaS|AzG4;DnZz+TwHzspC_ELy
z?;5!GgDZ20;;X5!Bb&N~^m9?V;lBBVgr%jW6)FRjG?%?AcT1;{cSP?r^hotz{Auy&dyy~CeP&%9ue`d<
z?V%;aY@N=e&j{_NZBns!^-2GYS%sSC{?6LFCsw!>7dfPsB{4V=_N`phF*e{AP7j8
zt{_CE*GLJy_l|T3QbO+}AQ0*|+jH-kIdi@{^Pl@)EY{wz7GV+I{k-q(%YHSQs+WE(
z(aJ+wtO)aZkmcc8dYysE$=jXd9pESfU^7PltP+2_TKM8-mwiG5K`>$n7*;-iutinb
zZ3A_(|Jv{7ii!Lx@FX7{+mSe;^~ja#qrC!F2QkyFUonzBjuM2Z*H5r^yFulmB*%t!
zzvaf?(9O%|Fg+9Xanuiw}Nf)
zv(h{|!?eI}U%Lqrw!YT~**zof5IWUXWE54d+#*GeslP1Z>)V+9P~$}3uK!JiR#GFw
zTvDdUVonzx1OUechktK*{0bOW^ci7y}tY=)WnMScsy$2J9>wqZA&~u10il#7pvm=rX!fC
zp_=oRlo05s5GNd?Kybz7uZ0f*K*J3Mx(kKejZB7Uvssvn88#bTlAV_C&7Ut7c{S_T
z6>7lg+JBq)qyMTp$VcI(tl^Tm)8a{JgA-2Sc}P2Ybv=AnOwfDckj`7AlI$>y(t}B_
z{%))xA#ImapC}>}
zMDM=w6x%l;YO05RH=c!My(}9?8DVbv`q)@cx@vQKtFjrJh$+lHl9$u;g-6_rQ~=nB
zsZ^HSKrp^tMp@)95G@tePm&)a|DjO1){DNlkGHc~j%eaCf!$UP@01{G)6^HFGZQy-
z65ny{gb9wv-uP#=F0v@`&M=lTHa4H{eUp491hlAnkN#jEvLwe7lY!FqaMGoEJQcJ+
ztsq04a~z^kEi$iSS9W0e_$IuVI*jioNLw*M8ZzUgi$@jt)BY505}P$^U-F?j4c?=s
z=zq*9ukk#HqMWWv5wuJcw&XNvUwkTe<#!Qh;ahBz+%0(gKvO#%ZMyK*G`VSEB{&kW
zT-5@B$Y7$2YlZ$wI3KjB*64%?edX)5G!@qfmg|puZ3#Goq&S24W_1kh(w~Gr*Ejl^
z#pa-voiJDZUByZ%Fwh@x?gW3?%2a+Afc0sm7A#)V2Es;2nbQZA!8K96Ny5MQIuGEJ
zJT0xmbMthQ&ax_)0C0!ITWj>2vz*YzohlD_ag{HJ;X?_6HQiZlf_US$Nt)8#0$K*MhQJ^w@F}xusvwszm8EMeNh-06sPP}GkJ
zzDKXDd{VGecwgO2a7FbezD3Z}@x;VFO|A46HQT9;1DP_F3|$kk+ZQ_mNw!oyrrdG#
z_(qr2woaZNU;C6tJ8xATuc;Ehb1ABNi%y>cZUj6T3^~YO?V>+7QxYxISo!NzjDFVY
z-tI>z6O4?+3mjw_l|AhZUPgD{fBE{5@0JP)MN3P6
zvTB#Qu(vYp{jTL7$z|hr-+Gn|>{T|QC43I#bN+E!{99}N&kPhK`(+l~A&xk3FpVv_
z5SFJo?!}+s=*wPSQBKC(@%nk}zzUK6d`K-OK%u^4@dMPbWkd
zM;a!nG#JA9!mPh-R`rrF{exNQN-tQZPS!7Xo2BU^b7{#}AyR#4@y5V6n-{7?Spvi=vXOi1VKN`A
zt3Ugw@Fl(eVoeT%n6E~jo!92PkOie_v5JMbQq-CXeomjdNXM)3Vd0()vckqHk
z@;wqT55byc%)3xsz0fp6bIk+UW}yKq-B`YgJmZ5xNZ-mFLZ+hJ#Cz)qg_sjHTyNe9
zKYxBT!S)D0ljARqe_x$86emD?M`brn)_csT>zT8{+GF43Xuqk28kuANI+}8;AJKFBN$8$
zleA|rWI$Ja?JHo&+!;ar8Z~>)tQe-+M>0hEMEj@L;*|stKM0(oZ!2J2?BPre{CAo%
z?5i?aM~?Q1v-$Ry;@;fQMt7#~G5=yx2#%>sSaTg!4NEOjsx!+L&#|+2(2GebR#Dm+!6D;h}
z(Gnjyw*!zO$`sQaQrR`D?#P;mbdm`%csae2Q$enE|NwBP~OP9>DXVB+w2
zwqHXjR_z^h;08gE=^FbhnrpI=+dSVB)ZRFmG@3oD-B#;8ao*Rg2L30X{~#*Y4)S(q
zmxR%#ies{Tf!9s8OKd~|L64Khw;9yn&!jg8^ua~aU5bexAE;i7nh9K
zIX7{c7PIlDK7V&)KcVmaE9rJQ<0{1#3223qzxaE_i^^Zf+U+F~qoqe%Qg5?J>!3}Z
zjhm3HK*;wUIvoo;-1+i;2;2@%th5YfGKI{Q%%P{zCPv>YD+ct_i*6Nu(~sahZs}#$
z?^Ea7S!{dFzg*N!5f+uSvXd}ZF~)-oY5?9BA|9OBen%X|wErV~0R2Mf(L={_#e7BX
zz;J~*~6k+%eD^;$shL2_80IfOGMU7mA?_9x*Zcr9^hi5
zIe&oK&8o_D(dKBn{0=0b5MGVe5hv%u5PIXHr@wvYm%5HyKbA*akGNHIy||d3%91Ra
z)tYSq6W5NU^1qK|h{5i+6)7CsMqCSH$T;b}YisA@4!Zq$=F{S}#1?|<2<%krAoc{a
zA9g`@mR*-<&thKn#d*cNYtx0c?Pmfq5oP@+IT$9CFJOE%%gj)kjFXq@?N
zrc9lY-n*mw^$KhaGTvPw9HWsP(lP2%3t&}>aq?we5kxsXcd*D0+ny{%pR
zeQ;|q`uqBE+GvCF8REU((`M|-WIuPjRLYIzFRCvJUXi}CbZyTpR9p#?j@NzRS2ogH
zhQiW|NV_(&KzU_KZlfw2&q_12zuA9}tmywgvnT*pV~?06!0hiVmu}48#)-!jZPkU9
zkj%316UAM!&H5^GFxn8N7+zC3(NbL&54?Rfqj0T>9~k2hE>D5u`gK;)p~;msXv6tn
zj$U&wz5?XK7beM0P`Q{Xf-vF-Xel@{S!+iZ8IP?yT{!#d63>dJ{)V@ay**5HO-2XW%0bpjAk
zub3D&k4#RUUd6q>@(Xkl_K4rz(uSwdVUr}_SB{;NuvUuJvBF~>WJL4pI5hY;5?{2u
z>(K_Mf0eE~>NzLccXH_9;J}qJTF2MXC91?V@g@0n{Yp;@iNc!+z5K%N&#cC48jx^Y
z98pA)4QFD5`D9&b#N`ICoBqVL{S8U2ZMvd={lG6-?}LI{vzGEZ&;QcQUXc&|Dh
z5QN!W-~S7w^9M5cM^XRh3;e(T5Pt?fGE+FQMV=!La}m!Sk*;3H@>H$bzTqIfkK3pJ
z=hZt#aoo)DPM+_(gv75cpn%xg{Hq}dh2Vu%Hq2Sh@(kEr;xwz(=2A$F
zvl5;hKv$xbAA_P3ry|k7z>P=0E`0H7ITuQeKjGyhXocQ
z^(D*_jEvy0S^N1Wr-5oEI)nAT|4F6&kv+ry0__^({Mr!2fs|X)`%xEs{u(w#J)%mZ
z&oj+@Vq~3-IsNZMtY*Len2iOot~CbNB0nG~W$AIIZBqb%c@;`ol5PH|zGbIJe|**a
z4ypC)t!!2rX&OG;^eCNQ`+ZxOd@Be2E~UP-1a}>8DEG*m)1gI?$9mkf=~MiTWvWn?
z=Ew#jY_WarU!a#z1*3}bz1UpxANSN)vJ>Lc9#j-jC2K9s3=x0}<7q^VmR}TpC3^7y
zbSKmUOwq&yW5h6qe#g4B4_+7KP>)nW^r`r=sF~i*_Hj3!UL-G*<$D(dx}&Mdga*eTAs^mIvTf0U
zv0jGqzKj+e%o+UgnL+qr25d1Q?faMPF*Wzn!sEshbL~=|Vjv_e>Io3QK(;okdyL
za?ACxB>V4srCZ0cs3%3)+xbn;cKu`^sN?4NpFj&a5d4ql7jJdk0)#JfTrlBBpo`$}
zk&OhSO@@>y=`^$Ub-9kwiz|(fiqXC{q_bhChhJV9b|a`29EPJZAK~McAT0KXTe3Pi
zCfg#va!^UqU&=~QA+)QlC?=j+wx6o3A59Ae-CuS+Nnf2qWLsx4+kS2&wak$m%ac4*
zYJ?;>K)sN9M^VS&qgQlW3Z(vu3SAkZ7BEB2kliUB<`jKTh585O$**}Wsa=&dc4cWH
znE=<7y0eSrKaRquFY)K2v}ug&O*kV0NWz6U;s#RY7RcgY#TtC^s30YGX4g}~sSFs^
zvcxVkj#deUr*|lHViU&h6?vwdn3zp&dvO+UMm|XoQ%>PB(A=7Bd2&PcI{)weWBBFg
zK-Rz^Y41F{MUMjWIVT#9dB~O?XJStTLg--`wLYTrN{g$EGy`jw{PGWhbt+mox0nvdCvfcB8>=U<>Hu`7K+rGNe6ADLYAD{O!F
zTl^6&NBjj^*_CGn^xq1&@T)Jf3V$p_^lOmHFJQX8Kr(z9IE3bTrhq-iIdipZ#m=GL
z5H;8`WpV7`$&SGr(RFS~!-eOY*foXH(-e+cty
z6Gu0$f3U8#5?ropC3p4pOT7_1G;iO{4x%!fD!<^tP2&;R*>?CX)!Im+XU|GS+A^LE!sA4AFE)&onX2Pj)qYM$9;u@8b6-6U-SN4^}!N1o}{#6>O;U9nt1N1(r{A}m!Il>&P+9~2+o8K;6sfwZ90$ipp%1jjQQxo#suf?mXtju
z(oj5~3azg9)g0z1@gflM)-$?Ti|K&
z$@q>ayP}eWX^RM>hJykx0kTinJ|+xl%d+Ro^!Lum!={}54z|;%uo4SVE8Ek&swn4l
zV84$)hSKY4o6ef^&4{UW4z7rA2yw+c;QGij-`7~~{`dr$1b{4v*|fhv%L|>`e}Tdc
z5mc#a6$$92eXW{UOm>Rshq#nQ>gZ%SLy$Nli4vLN@kkw7Z*rx(bGc9k?b{k6U1
z#AZK3#p$rIE7e$C7bV?Zqi*-67Epl-4h)Z1FInrAvS<-D$HL8
zv%tJz5XgHAnZbNilM=QvbT8ct??tm0I7+kInC!zV7j!aKdq81`>S^
z=@|!{lb`>w>(QT{;8L&1C~yRrsomR!FjKX%QQ-zg%%Gy@cK@?D{}V+JBl-Q|{(DVW
z(5d2MCN`(#bimXLzwM=wVY#=ygoxB|(q3CAew*Es;C~*3E(v|%3VH!zrIaj{{xq#P
z!;i?bVPmlU9K$nNA~_bF#Vj~?S|?g!n}VDWI=i1T4bm$hNSBtRMj$>`7zTuS~$=R{Y7aMCL-$FU2ubjTzU(e}0jYWMxuEay$~
z{_|<;6n)#nmQz(Sj?<5OL|Rm-7koHGw>CQbB^KEdciYiqDx!h*xW}?S8JZkk+x>t(iq%kwsYPr7GM-S82
zzxEPD@;u>r+jLrFP9@~?T$)z(^_7W7b`gmkFMZ5Pu`l;aV=M%>{TydaD&rIZnODul
z|J5_~-_rj6!9tMi*lxfE*Wm)#-~jkzEc^pMa8Q_A+d%LXEPQc4)LFp+
zPB!fnC_I{zW-5j1mvr7Qxpyc+D^g=wzR1fh^Mlc6fWi-@ByZ;{k+nH}B+_g1`?WSf
zXl}B)mQIEeH_!N#azpb_D{!pM8~)#pwf=7B@C!G*A;!KEkW$?WWYk_3AGgx94;`P?
z7U#LMq;@Y^XSzqduilk-zIKRz=j|Z^aFSOH1I=R06bp2{qnSjh$jres1#MH4Hvjnj
zO4lH5fpyQ&*vA`^U0NPR%vuyG9&?xmZxW$_f~5-5-m#d48g%k*mZoGQIPH?fpAPHN
z)~JAq3lY9zgFTt=Rt^xlqoK<9AVTPC>?UDHWwi18m^GUZR>OB5TEyx9-j#1x9owxe
z5gNL!!Ox>RkDjNy!TRuC=res*BEa^vL1a`ESzflfgrn&+$S@Y)o1K{3+ArEVKMkeR
zdA|0i`MujvQ8^Cf9y_{e!z1T&xPAC8S^G>;%;9nLDcx$@@^+9+en0cn1*}nr!pFA!dK_ra^CB@&SaZr
zM+a}pQydIDw+&q8;YVcbklT@(V*?vF|^uarX(d~iYKD{9m9YMEh`Pia0v8LxFJ
zF$1oK@P+3!2hJ0bd$yKZU6GIUKe$fYg03~BEx{KN@Cryf!>H&mnqdJL
zSyx@GA~=Z3XzVORE9LcEi+ujQ{uGHycXF9UUF~7wm;M3~k1Y-FwlB+p0oIOXA2n#&
z_EstT@P37Z9Z^G?NAYWS?t{pl(Ykx(O-ouvRt{okI@n=Mb7&wCMf4mGU#yp?8m-Ll
zd&U`XpgPc*sp7KaZ7ZNn@C8`gbO}g3^sIK*@^G0LOoP)*?kGa`S;l%lf2*GV{8n9q
zOU|hJ+5BeF5wH-w5y}7CWBzZ(U828P;{Ge|{J(lA4I{6n-*R29@8FgfXgB3Wafaby
z;2U^TwcdEY!ZYYV(!k+PnbF{-e6sEd(1Xrp@^%$Qa5bTKIdVCpG9wb
zZCp`i`XNa-QYA!WD>QOvWfNo=?5t=K)MOL1-(t1F_>)d~F8mBuMQJjS9M;@6>D?sT
z3`0Q;*3}axH<00C$v0Tg^O_F?Zn!mx-F~?e*7lJqd*h!Z!+&w*&%}o>Ybs!d-+Aq`
zIaZDRGO=gzKYRb9CRQwgjdJP+9Oh1&RwF5&0xeVBOZ3uoIlAbY?m+#fS?^|VtE)aR
zY67t2etWOT;rTI6YeZK0nc>5l
z_WR1N0iAZr(D_F`P98H${-sO~V*?M80#aD&8YOEzWH^xlriRXE)r}2FfDBzAaHP39
zFF5c9EXoA$l@0TIG1*|8XHr1fQ}mi*-~*CHeYX7vS>NCfpPMzT$+hD#V!k3hedW>4
zJ|E7ih||C921xVgc%KRC9Ldoi#_i~ztKz4QBe8KG#I#xQ-d%0$Pa9+Bgaz&;$T%f?
z3cTd6xV8DNer@xftIT=ldgO(K73Lf;Y-*RxWN`?f;>+L@bsd3}
zv*i4Dt3g!@(y}7gKeMxbYI0
z!V9mG=kI|xFTM~NKQjK^bEFiJ)euVX$tF@Mp6MIA^qyjJPtOpkBuC00<*My0cqXsa
z=ZZPJ`Jc@Mp?gInRze-`KQ3M2)b@{gGaW;P0r-Q3r!^87{op|XVKJ;NuD
zQbl4?keV(|zfCOAMD*)#6?Si8AsuW26BiWUFP@AEi_MWqP@|2;W!MxP~vFQ_SIy~Yj77})(z<^IVee=8I-lRr<
z?Tm)6+r5Nt0?sd#Ai~5TsZxMn+vUZ!ABR;>*x=zlf!re*411GTzn(tI+M8j&_Sx9f
z3h>QJ5pc+zM9B(%!!NW;=e?KJuon7VVmYZHQ=>t@KwC;m=uRa|-6MIgS^A0-@3hps
zV~TU^2C_mbghETr90EDRCLNqzMDa(wV!kOVU~8d!*F~RSP9H1mONf9IoxL2@({<0~
z{4%>Kt*2SfLf>%9#0tZtL<d9Ar9|sJX9&aG4nB^kFu2axa>2L>cEMrra*`J9(5EJBY*=fbn9`QxRLw#KbPvk)
z>Erb7zV|d#-pg*1Ovn5xH7TRaLYP}axFQWEXe(>1p&S-AsHr4P9Rl5Imt1CT7s=74
zP^){1A#G229O{Xc#F%`iH#t-sQY
z8oo$Fb}5|SI@-TvPYaSWn7Wet_7|ws>7vji@?Y+O=@m@@mt~cWhT7Qs892r#xT&a0
z&8N?BjA6U?sz<#Fc6Nnmw$00eT`HBRAu$^UA3R=bf{iq|EJ2t7i3X|sJ-pfyk}x2#g^m9}x!x^T
zF}0pvyE;ha%~
z=%!EvQb{N#&MfU_ZRl
z3$e?;=(Iqug8A&pKltdfSZG(X8^H{X+}gZ9(Zkoyp3fdK{t97>45QkWD7c`x%oPCg
zN5a7*qJsRbQN`%ZQPbI3_39}avkxccMPX=HR_a=#BrPLr=y~S%=+lYxdee*x&7_8A
zYE~sai=h~sVJ+ETgG##r7R?gqy;dc99K3r0&$AxZK#~sw=4n&NN_Z6D)0G>b3cKj{
zj~&{Keicvae^fnJ<&)pw=H|RsxAiiH%U2|dyQl2*2wt?vdnnfYEYZ8(4o&yRt&d<8
z+m8hkv*V+d%~__D>SMQ-CVlTw+$r8xQM2zUYEj!l8@JmYXe;bgk|ETv}=yN%72T
z6PJRK5eq-8(=sCHN_r#p?ZUe~p_Z|=TjuEbC-*WLIQWbeU+D))vS+^cA~gJF@F`l(
zOi)F+(#*PA+avaop4o_xBGBw`5xv?7yoQws^NM-e!N!Vpj>y?ct6*quEFS~>;or5E
ztUX^0Ck}HS4x9X=;)~gl)^t{8UIjJ|gD>-d(Sv?eXpI(zE6Q3NC;HA{=$WHwl0H6033kzMePJot}x#I2goLv7vvKSB1^+`FuQJpx)nC^ii8uruwK>yj^~FgH9G8^M3mXNs{Q1X
z5;%5-R5sL8KM)L6yjcAoyP-*Eu;NGr>M4QVwAk85!8KSmpTp_7Ug;Y^VE@iKB1M1KKn>9PF}4jUJIZ7`=IsxYo*n$?(>jVSLXAd88fqOE0FHF@=Vb`!i$5Z*h<=VYEgkBZ4iEe(R$HK*l>-Y+qYU
z=|kV1nfbxt7Srt+{eqwEUHOy46XD=yYe-Na304`ocoR&}3eVlfUyUWc@DLm9A*9pe
zX=i`(HD9yd$QdlI>BmK<+uoM8$Gp#CD>b3dqc$3>qfVVv>v9KCa=dFZVYb@Ddhq}Y
zZquefz6a8toEMZvQw+Ye(>6pOR9*A&WzLudkyp@$5`4X0m&Q|)pk@r_H9wH7O4b#{8x7ou&-RQlm7-<_T-+-dy$!cEop3ogeTlm%zeX_vy9+
zh5iJwZkYh%LL1?&snmFmwg-Y?D6!~@p5SwfdoSLUa>us*bb0IgckuB@qWSk9F55&zNHt{?;ap4=+)%)Dkv5IrSGSB>5EkxIGYFbjN_
z==|Z;=+%R9`D^p&xLX=@32NkTW8%)ZRz6zSo2q5ZCXw9zk)8jBM#OkLIXMxi^OI`t
z_1(b6$S3;tx+jG^3#zB6>iJ4-{n`yUsOh1RKMZf4WPLRBG=;Z#^`qL#AZ?Os
zDQLcync`jR3PHR@1fV*+aakg(>3a@D#I;Lo!pTIucOD^RilpJw5`Rk+3-KnJjgvSGQM(g{hrPl
zK?fCxroY8tD+lm+Ko}eRiVU9j)ENzLr6N;v^QB+r*kUj6k!`FKBueIB
z1|<{@COmFSOxyUToO@p1D3uL>h=j8U$Zdo)-$1Z^eatXZ4N2aCdH$0-wsC!u!m`-
zj^(m}Thx2giaAej^!o4u%4a9l>Rd|^vkAwDq_@|Fhoik>A!V0^ysy)3j~?znVPYB&
zqu3&qB7D2v4!OR$K;A&`6;Gg6m$^2kSQgnQwazh+YQH+mdR+m6!2Lrhb}vUS)bViq1uXv{0O=&x`BE%+6mR
zkvX7mrL;=crVdX%*CbvS6HnsO^o%o@ojl%$%@J61>GGyIl?@#YXpRKKPWziP>w
zQXvLY#>kES+KaD`f3?0KXeFa0JuTY137+9UhF9RB;I@cBwwbvNepg}eP<-cjHHSiz
z(|dlQi-%@mAYML@1^?7Wo;DSk8>-YM6PIb1VL5;IWwPX)7*n4?l=_sZ)8rY`)h#SM
z%%1>L4K>`bxb#Vz)i?4c;rzN``?&z8PG*PY+VkX5AUd~%;@$*g76^kPJTUFbxa
zNndZmEyh><)>`<}O}q+G>YDKvSKmng3Qgrf<*|*~0EzvQNf4$p$0__aI%Me@trKGKHk^#ANO)*29fn*yQ43_%YM9K!o;v5m8p`uN$s3xHopyZ
zvV*L@*-r;ebHp}8epVJ4RS6_uj_J=4&?0jWL4C>c%Gy{!Y_6C#{rqjNwKaOKJ1PJS
zLK*cNgN)#9yhik#OAMI|97WLYmp>)BL>+#^(YT)VSu3ZAerTCpLHv7B+*4
zM+|!fFR2Dm%a#$K`a*SjTFD5m6Z22n7bLx`Optc{^h=A_8DXB2Zgn%j~aiYrZ8Ct`$%$*(2gZ5J!;>4)Ief|Do0%*x%V`%KfuMvd
z>h=<$xML3}_bf$imrKLo^latnf{#ne!>M&->4gYJm<-Ra!NMKVAeOUX0`pEXaj0^e
zecD)8)ziW($t*cc|Ls^t8x>H$^W!p`J9Yid?PEI`AFnPy!(l7jr-q?IWMsjhoX}44
z?r2}n*<0pV?T}7OE)~KI=FWU&U1)2)-;@t@e!s3U@&5P7;Ir)=aS6s;=KJTe+QO+c
zs|)trj7
zm09@Cw{s{g{rEi|Q4C|Wz;dOeI?zX>V`4hLB*-YTy5BqXNWnzVP-Q%cj)tDroEDCC
zAZ{sG?xXtvl>Qi%za8(0`(**RQtaj;Ao>*K#|ak+e_O4
zelI3$QH%*nIVRma)iD(n0}h;zZ$c$+*1QXLB^_F5bwT@Kn!El2U3apO{|T=Y?boJc
zrYs`PqA9NGiT^A+9RJx^o&T2NaG&D!hFKsfECHB=@WTGBPY3+EqxhLm{N87_hWK_7>jYM@n=7>SzZh7jnWHmV<*}<45-GOE8hzK*mtQ>(@<`<%`R#{E
z-8sh^Vhm9EuzC#g-A-ltrL(N22VHsZdG4$8dz1tBp168{)oH_pGM%n3u3ylqU>4px
z=Z!b3iev
zXvysl4I6Lx
zCmCC*uQ5q?U-*(Uxmh5TKsGRlQayI>oz3?Je5yWCvj{bP!|1S3VSFYthLstEd9R6&94fqr87Rep4clr0fpFJ2KP&?KO
zd|&i5W){~#H&=bTdFHfe!{p-lvPPC@NjMbD1dLw_S%kOumKC2V9Weuz92pnS^G*08F@r*n!hY-4sNkn9{Lh;A0e2-}|4>ZTcLX;)>?{ti%h
zVqXjfZLx7qG*Do64ZBP%H_=CHsZR0*wprKbcfaciyrOYK^^;pK-_dk2=$$h=eVa7L
zTO%KfcB}*>=t_^SHz;dQA1+oRZj*F}jiJE17?_xF`DK2Na$)$+b4$LV_TU$1fOfS6byzjoG(sjH;&=JUnvrVR~$3T6wCq-wY
zPbE?q)!Ya_Ny?bt&#D(XXM)5Q$(4yMoW#z8Zh
z5kr}W(_Y*KdOQTNBSV_N-cPv(o96~YWi>8bZB)sWA}x}4VaQ={RC5KmbNF2y=i8HL
z`F6MOztrwTAhenu-9MpVk4a880*xdl)cafuN#Ng`hQFZ-OZ0!^60BiW7&Xxhkb1|`VHW~gMpZFZ1grkhNDcC$=7nrR6DDboATF@Y76}OpRO+8xeY^pH}Yc)kjpX-rGNlt
zJaEvn{HswsG{iR$qVVAIaJ6PV+P_xzWakhF>V1vRJ6GR=lQrlpM)Awhn_QYOL9;b7
z*iFOGcJ3YG#s=bj$l-g-V9(2_Z2L2n@ZZy-P!f&Gv0_0(Ylfws1#+EsyQ!(0EAPw^SO|j~dvA{+IBN>a2
zmM-G$8cFnCd3)aLwLj)On5d)V0j9mM4`jWN#&O3Ps;BBo6MuHgzo|Ow@Tn`bYMsP>
zIWgp2Vn`&$xo-!Pr4EaX&(XBdbr3fEE?aRt0?g+I@;bpR+^sURAC8p9zong_(O8o~
zUt8*S<#G`o%)pzMKYH}0_A>To4BOyz4fI&;XTTKdJ5NlJw_3(T1g4&>nZj-+vI_HX
zO6hEzuQmE>HH3u
z*}ncRHqh}hU7mA$p`2>>CemL5-|Lfa26qx`4E=3j_+*XLdCh($64$Gsu($BVf7V9o
zb}H0e`iGvr{?6|M1@)Ydw)*7*T2J-j4cbw9@ybW{(#0bjpkT~$({kXE|G&BuxK6swF0FS&v^IC%+j-APMx6zsJDj{>!V9fi*_
zQR{eO8#Hfj*3n>?Nu#HiO|J!H*<{7|B3pbRLrX;x2btAAz8AOjtS$}!fI~5_aGb6h
zUPTVsGBJkzAD|D2NZzd6pQ~w4pd#UFNIX^?LoJ{g7xOy6GgfBfP*JFSn@qq{t@=np
z(|n=8iM20akMaGwRn;Uoo`d@>h=Fw#*g%h&1f6Z3D?Nx+b|8_5Cqa+&{Q^;8U4Q_k
zLV&+GUdJO3me*AcU23M=Ol^juYT8>JocrVW-^j!s8{7_gxTON-K^8eZ7vp^*e`Y9K
zDm9G#oGA{O@hViOQPbMZ@9oAoO53M;H8s5fG*8+1|0{>{uLQ?`*Ur@tx3s?lB+uOR
z>p)XQI5$1Ic#NI8;E;&ldTSmVc4j$iy1=Q|evJ5?&YNxb(5&x0BX2P23uUJ%p08wQ
z0mUiO0jQ*@DaLZsws^V0u3>Ta!fWxze5Gq$S`#`6uArrctbZSFfQ0;Aa>escu8MwB_%X#c{;_KB?2?pQfIx+G?c_{(&=u|e_#f!6K9wr?g4uih<-_d
z0$4@E!3$(p-1GiM*xZc#T%@a=I9A(}ZL(#6^kd7rr{qPCE(WH_A9uc8_|Tw@X(n}W
zcghsN5vPCO{E=)dEOxVI7xP35B(%cAiIrL_aX_bJ6dRdt)oETf?k4{Np5(#GB2L||
z_Pg=pA6sJIv%ak9(u8>WV9GxIq9==dzslmpC{W^=T`zxrIb3ls{OJf3rFPUGaIa5L
zd6m#(o~usa;AvfTEYaq<%B~j4sN8)P)k;g22W-l|)(m(JQ1~jyJPqMR^XVmVG
zdZUkjVUg&uMIdxu3P!8(Qwa4Z7s2aP1>yc!BR2?Qfw}4W
zg(jYEGNnn*)K}oh&a)!358F>ZcpWLp{%DtGkyu|+F!mbqCpZ*RMbfSv1*Z`Typ0d+
z)(>737_{Zip6oJx3u
zFJ@3&M)1*3i*~=fSv4|KRU1-4gvSG}ceZ99;hv|%E~U_tlH%Q^ii+x3Qg)?So~+0D
zABeB#vw}!KWEy@l8c*+5qQbyZ0|~zNjDDHLZsX@u-L{eu%ecf&31{26MsEnm=A#{3
zDw0%#)`|9l@w7OBeY=}3Hj`IHLo4Gt7B=Uq{uT{*W^9NFK0eQ$!?12_;Gv@Pjsg2Y
zUs?Fa*eASiK0eGOR(6+K<)Zg?Tg0N=ra!#4=O2P938I|)J$m8J3t{j|?v4Q3wx1Wb
zY8z9~eQOw$=}keUH>=6h-0{4&o|FX<#1gJ%tZYU5U{cRG7JD$c?8EOD&&Gk|KEUCU
z+fJDS7$`#5ZVo~4)2=t+)PyHUMAIFABYgb1O<=!;c=$6^H5L(-rY8D9sKh2g#$j4O
z_hYmg>;zB(x=+v5%PztYJIuID?T`g(FXT-35?x(VYF%cP=eby{e%hFJbNGN-qSPiw
zP<5oXiPI4;qrlcsj*>}
z;dow>7C131&H&ryz{WqZ=NXk-k>JNKNz_mpt$$IKU}JfW&^n)_N=0le@aF=<)h!?+
zNpoD*W$AW|0h|oq*2V89R<5McooO)ONRUz3gG-pXiL^RaU@
z3u)-m7*ybAco`e&{E=h3=_TSQfWPDI#CnH*^slNiCpz(jbo~TXOkD5hOonUY=sdFm^$KRQNYA4a{)(-I3pAB4(m$96h7z(?s)X|Dcx2Bth4LK*+n7nF^MW
zP@tO?5*?;j<(l%hvmyL|uIxLlY
zx$;^4h2zP*#{Hi+c}t{{{M_?%JTub)<1znN{EvU#r~YsE|Lt7S$gGm>I_@yf^b)_0
zw9eS%T%#_Vy2hp&RrYrz2EMGJI(w>LZ%+I@E!`p+NR)t?cA-VtIIwmRo+WU#+TnP3
zu7!%1ix4H{3mnbu}hTeEvMFIC-{sG
z++k3py})Ui;41|0?G!@W`^GNJK9BnJUDBhEEgkP~!5c_dsJ9R|uGnTSc<^oj{vB?>
z5`BW*FmQ4d%Dru3+bb}u>>b>Jad&<2$SrM9>acP$90{RFeJYq^k!*{j)YM`a_WTl=aDVjZ
zkw_X2KD~adGYPkig43D-p50?~=lTb>IYM!_KmL%@P@~jVJ1gED5JAB(SWV}}W@oSF
zUYu^2M|AruzSX2vM|~D|0~3^!Q{7BJ>whM^=u93B-flP1AF5rtV=$&URy^i^VyGve
z^Q^YZ_*0M+dw^;lQ>_?NQPpi$;izUi_t{oD51WAF&P7%vV2$XhElgt)`{>}Z#(gx~pvTuV`0PVZn?-l(+c3Ami+tZ|k
z(CRzOK}-SZm|Q{0VUZ{~61
zQYD%i-*v(AdmI*9cZkQ@RQ;|Zgvei_$C98nO#>JQc(*^AHX)jpD}gGzCT&~S=wpzD
z-Veg4j9VWAy)NIh)5CwD=H{?^In()dcKV||HM@u}1WA_S9EpW;-h-{P`<
z-$<)Cv10c*RRp^Xk^w4cNfQL#Z2Aj`SONzleyT61)OY*youZ*ui
z@|P@*>M+BEe#(kq^p9}aMZpHKH-f`%6WfH~MwdBAmlA)k&AUt>n?NSMgK=oe46-~k
z$my8;*n!27bTIc(2b}cm;ln0^07MAkUH6^fjO1e%|}$q
zX`2eg)Smdd#1wAV;Nmhf)pouC&KPSj<%p5gKuN70Hr860vsV@0v@$`KB1e@UT9jei
zvxQ@>Bv7)uOG5dx%2rD(lhMg5q9p<&KWGShZEL@J7gp+KLlYXna}pyw`?+@cwZeez
za}r8teEYeB@ql*lecLm;?-ou*y_$_woTH9T@icCqZ8%jQyO$Bb@h3+k)^lqIk7xW4
zGJ`jKu;}mp85+V1Mss|S-h4dj5#F)0g0e9!eyZiJ^&0ty+dGph;-}H|B;<~@4y#Xo
z{=T)+`V_%lc~NUFzfB*m)2$aVsRpsPvoA4KcwGWj0D)X+`-vV7JQAH!x>uG2n4)le
z)siKz3xAujo*h#?s;J}4Ai>c}ZC5XyZNL6DbLBk$^#o#gZab@^=n4j)@PEta_&*>2
zKN#1H@3;_|?AHrttcTl6(++Ak}ED2iI@g65v9jeN`LmHc_X|6dr{EK7@=
zpoZA}r84U6NP~Z0uA2V521*e553G^NNAs%zY5KG7F+cKFqpE?~$==O7j
z8WBhPg2u+hpZy3IgYs;{>;fnXt`|VNMz~_QI|mWDh2A_YOT9OY+20n2xt0zZJ}mZ4
zH|*Ia-g{=V3@Gy^cwIGwN)#3bHfj%zxd9E5J?C_1eY~O0g1U6$HM)WCKOD0_R+0*u
zK6h)dN6KuYzqtK2#mgVQEayICDafXai_emzMsJf8jf+byDpPB9eDyteLWcOH?QuCW
zL+UL`XSQ&*iV@o`oYG;MKP74#LCqdR10+r4y5tuu4(8^T9uA5PDDYf~OV9NatN7DN&A>4D;lYhicuW6f{8D5jVNM(&qiXlMS9jTac3m}o
zBe#4qN3Oc0I^
zKm)>fWuVS%d1-_#PM`I0dM0zk;j@;8l)|GMmx078$BdT9NGSClq$ScW&fvGLwH%X<
z<%v~}eviJ#fC99?=zfv!tR9pP81rVg(qFhAI>Eu!K@!(FHYh=~XYST%??Ibczy_jl
zWzp%8;{vJD&SpwFY43oRC>Jv}qa1Ep|GPPR2Wf_%l3Yc-c}RQ3n(6u@@U>xs;SrH)
zT3Vzw#xPs1E=Z{@%SBD7hAfAc{)3xh;NrWldb$a^X{`jkh|YMh)Vdbl;VM7EZYV-!
z+J&vTErh%ItZN&GzD5e;M1`W+k{AU-GHvGeX@!a0Q6e?y%$=yHehA%uQK=aYc+(o+
zm(WLadiE(Hp^Se|R5C3s!A!a2ToMtb$O?T#KAIp>Nc-K+7xVw
zz$Wz<>~0AOM{^Zvakuz>F79dAG@wq}oP2#rBT7wB2;)8-HmUq@D({
z2Bq!a!bj-uHc7g&-wbvW}2N37)Z~}S3Ws8o)Em8ysbcMp+r~w-IU;+@0}sfeD?Qp0>55C+C=H}F3ydlnq_u&^43NqG*FOJE_u5{Y
z4T20P{(472ph+g=UhgVR_@xyK)WSwDe3&B*6=P>dwA%XxV;@LW`Da7h#ZXa_|Fnax
z32TZ1Yw3`605<8YQ3~=Ov$g&)bOLCBctQvJWuoqw5O&dFLX-6S8H|r-d%kK21E#7?
zmqGmZ9Aa+jqWNiX?BC~NQOYQEEQ?QyPU`(^t@EpRFSy;db5
zxUB+aeLU)yDR9c{jdrR&gu~U1<~P0#l$xN~NpJCz9L{H(oMmJsGp#MJ7?wq;TJ3b?8ka2vr}d)=NX!l(5Ar{h6k`
zMAOGM4|^$yGT(*W*CU4!@cdExV_D6EeE@pQCDbhzLKyhIu_@JrEoxO2Q!GiC-A4D<
z#MBJ4HNDZ`JCS}SXu+lRtdJE2Sbqi*WI1Y*Qt>*%baw2g(7N@Vx1zc=wh^^%nO{PtwE@R7A
zHw@ZfOEQXWXt=Lm`{;o_<+
zS&HbjK*nyhYEkdMvuliS`RyK?ckqRd&i@6(aHZwpSNUd}=Fjpw&0XsxGu5He7`M5>
zPk9qjnncWuZ#TMAjY~uX+7oGV8ge9(Qh`874|%5?|t62|RT5
zcc>6@<401uL3(qot+_Am{$zY+bOOBwK*@?(0j8mtNUPr0cu?~0lx0Pg0@`MO+^RWe
zM|1s2lqaR$CRhIUgN&_2?qbKe)xw1qO`_AOML3;*e%$QRAQ5MO?Z)TW52>R2O_B}%
zHzWq|%t#@a(xB0Tq+sEp$c@=`hKP}`&bUyyEyI+=Gg~pNbk-t}k!}HMB0%EuOh%@<
zuO>=|s03IfOZiq?*Pt(Ev<&>AD@fay#DSY}*Px&7+s{bxE
z3T_2$MafJerGO2GND2V+sz!6SOQ5`9qA2OLn$@7RiIuZsl;1BzYvP?n;GE900K>+6e^Bi`NF>la$BSO<+@QN#m=e6?
z4$4H?pu~b#8r?>Bol#7mll+2CzY3*n-EwTPXAVq;i%3f%7Tya5JbVjR
zn&kzHReXGIt{Ngfbnm_{wra-XFm2)7VEEvBG|zfAEy)9a!Cq6tu)-yK0N#DP1!;`u
z;BfW`El(TnsuucB{_wxQXn*%za&W5zMh0KI448faGuy-&N4S{UM~zGmU~->3V+?gs
z8W6N*vi(5y(u@am!e~HSlTfDhkZ7TSy4rn#$+h+SZvpVPI6LS>IUPi1n3ij@QgroGw<
z?Z`iG&Fsu6P8G?oK{fI>n^MNyY#?*
z=}FvwZL$5E_-s3iWIl+=Xhu@5aPg>x=h!HN_C3BA$b)o5o$R}vbr2r&f<2wxn*Ly$
zeCy?x*5cHlw8ev82_4l^(w2i#2pOQ^)A7Jx(~TA^H=n67y_wEzbOu$a9fMG5=SF)a)PQ7-BvLmCE9~w=^6h6<|
zqvmM3E$d(c^B|+)e!383XSG(|KCR@`VTpcN4SV4bt&p8XW!sPRg2nobwXk#Rg&nHc
zHu4tSSeE%a@)h;1WIW+-*3&0!Guvrn9ub$Ia$i|Pd1{Y(T->Mr62Vwxrrs+Be_>$X
zhJ-a)J)LwY`_(AzuX_IEO9{IR>qMoal~vrOmbz1rA41`dfDb`AV1)@{dT_^JDp!b1
z)}T7?Egj@OTY1%}yEN*C9i#u+^br;gw_hoz^a1K{U{k%7F0i>W2dbK{N=qaC+cs1!
z*CQ*KVjL{}^H@0)%+`_|Ucd6_a8kYgqoaq}(`5CmzG9Vp>T6*Eu0cs76SJ9wx#jE=
za8zLOXbmShKed7SVsr6zb5>d~Y%TT3FGS%pG97;28$y95E=NltyaiQ@K-w|(XSRzh
z6h7YG&JSrUD5B(ObUpwSXQsNV#X;HG7cw~JPDZIt3xpnl0pHb5WF&#h1L#xJN--^rUbrh|VGe=Zbd-sG165{;dkFrbJqx
zOo6Up^VU&kbO)u2fm@!5Eu^WrQ9u0J1~tG!-}vgsgEAGyHlz9lO+MV1uGvfDPog6W
zb;m=0!W1G}2)Lq8nYjl4lA~nuPhbc8m(=^ekYRa7zRB4M{)l@{{DZb6o-Lr?}q-uh{{2-
z0&k4u#gHGt-J2mbL3M76e_=GB%Vp|rPcz3+B1@NkgP-HBpaTf22r$#uI0IT@E0@6G
zj%B*YC*{}&L|p}2%si<{GO%Ow72OP#1YK$3sumgYz^m8
zyO9(hU)(!|y~MUIw0(y4C9l9Sm(qI;I@-+yuLnDBSb&EC*u!l!im*46Rx2%0LH=BF
zQvC90Y(kHEu@PkNyV1DM!=*Z1&mg>{^z-+qhJ;pa?CXa5`pDN&c(c|lETLuJyfNOi
zRRjMrAZ`9Z2#_}4`zLLl>0hPI|F4cgticYE;JI8ppxnN-R|N`+d<>8rbnguc#aK}rwn$lIY
ze)C!E+D;#42KOygf<153?jGI!G+j;GnoZfpPy`v!4&!WsanYk`$l!wAWIUVF57s*
zq|yAH*kqAoqCTx!pg4u#euq55saZC
zyKSY?ZD6l|7Xml~_uOJvmhAk3NPjaOD{h7mHFi}KB89+;#SNVx;+?sH48DTEru@KR
z3x@oO9o?h^(E$dncQ=fk+%j)->Cw#akO5+u&FH?6n2=voTtVKCXIN`86!|3WB!{19
z8$4$8^4a5>#DD}B7Zk?SDs`uaa&`gp1FpYUAY=|l-(n_yYc+N!Q=AMnFhjhyr
zuTM}8dx7aanyUPsdU+{X!o(bo3#5e9rh~5n%bkMK%7a&PA63aXSP=jjd`u*jlaZ3L
zFQTJX520Exx=(z0aaLj$W=^n2@O!m)u-bs@;Jq@(f&+m`Hy_4Khcr{gZL^^tPaSLq
zgdYzWc>MA(;wsLvqdS+##mRb0P48JiPf4S*omg~g_4(=@Pe|2ZF4`T*zJ8j#dOrjc
zIamD#LM5=ev}`i;2o(69sO-@Ad2afO3Uh+MzWd8goDO1tg>~gH9yI29C!{?~Pt}X3
ze>rTgDI3wE%z1B*;2Q=8rt<8Sw}Gef*Dug4E2wd2S&#x$#wOU+?BEZ<$$(x?ks^1K
z*iRka5K-#?)x3og>YCyxbg|v{h$Wnj7Zx_$Q!2{%k%D@}Qzx$VZ6pP-Bf|&*F-6JL
zzs;*~lqP7L=dP}|H7RVIg>5Akhc--cQm=GltUpv?b>D2yFMh6P6ff@ai%8SNVxz|0
z{mY8f#T*XBRwu-7Aon2{8z~pCbxMc~CN|PG8{*EV;)J=m?UJDu|V+THx>yv@mZC
zppp?mv7%(|C|nvZ`7S#%FGKiH@~0>xx36b=_P-=2FNN*nZaAAysnHUE?D@g!{NyHX
z7+7l4fxGA&mOuYuh))+kq$AwGy7ENq8wu;}M{9)7!g2KaXvhPn45)AZjd`zp%UnfK
zTND>l{pclSDk--XOHk%;*Q3V+V&Csbi~)B)FmsuVbw{fqgu|*1qq`YN9jd%7C|!7Q
zB4E(VSpjt)&;DLqc61k>9i0-s^eqrb|44xdMDOukUy<5Crwijjcw6#6N^<@SZ-D;b
z$H5DoIKk9t?K-6TX62hpskhr*b)ytjDA1EnSk&2P1l!LWM=C?RvIr1X(5>}knAsM9
z00Npsd_6p_uorZJf&Q?f2xBvq-{p~>wizm;_pXOrhN@KPRMxSM*GOglC~C;1pP8V|
z5Sm}VchtE46zY3ufeq6v(zTiz$!iSv8-6NYXyUtew#hycg@8w=RK?xb&~iI
z)^VehvbZu$$%!=rjR16IZ~==6C7Vf>L?q@XqWhxTTaaL%kZwh^PSN3C7?dxxNF1+p
zCWT`QF#~o+s1>+n7+uULmsUp=1hRN^|PyBvcXHm@^5sk9M;Xx2vc{KHJcmc;DY
zxN$cR{zRhEHg~!)Rq0pT58jbl1$1c_pIbV11}HXFohrID*R9t*xlPKpELjPuYm7w;
zmV-IqIbns!c#V$PdKr>~hL-H2_TOIPcKnTs4!Z_y1@{f9nXOA1%4f>YOD?JBJIIlA
zgFivEdwojLmI$7&j+F1@>QZbHx3g6qO7abqIEaC|;*12YuB%T@>!ne5J17w$(x@
ziirr`p_~8&enA8XJl)?f=ka+y2UT#C^klTOo3=MdS0lYpZ^1?gKbT7ZTu%?qp1{aN
zn|PU3!`@aN={sig^J}Scdx$l4sBwwb0SjEz#9S
z;#s{(;g^wf&GbjcePJ(S6w03&{(6td`qq6&;sL+v3T6aSvfEj^EXl)Fyelb}&6Z-V
z0)Gs@^>sZSGfBdr;lOynm7iZK&uKnCyHoo4&Edx5uxzCRxrTxuvOvgHXZzLb|G8?9
zfz^?*8`bac?$6yl_U{OIGYwGpO~wRHQ%1$}S$Ji+Ck_kPh<*%XJxK#*GiooK-f$`b
z9{qn0S5Til*stb6VxbiMaW50#u*=Rr*}sM}yf~1ZoB{%6N$%dKYpl=HtsmgIZHV1#
zSBn`|wV>VN_kmOFMD=wKc&teM)Ugr|h61jSgb;fsynXSKaI}!k_f?SJw3*OE9(}Tl
z^Y^fZ^u!OD;OtKR4f^$kVqoNxTne^8$^pST(1B(7Q(RBiP|lt3b;eJE>Fvct
zu#u&f9%rHXvUov5m3n+$#>{weW^GoMUQ7GjjoZ}is?#m7bp2O~{eR>K{joCBb#51d
zu>r(rnBQL*-zOUXFfsp@z4a%?4s+VlEWI`rZEUw8t3Tl|D-`GB!TG~)I
zoiL}qlND^nXLiM^3=|W#Ql=A6?HA+Jj(62Dsyhf`NLx5LXz6J;(NmRQa4W60NBQ_y
z4PJ(}B7-)Q?=PAm9eqls9==Q^MEQ~hs0*epS#hL>DdUE#T(?d_WiBV+GTmQc>RlOk
z2jKuv+Rj8~R%QH>!ydE~Gymhq8Gk11+-fSX`u!&4ycC4i5Yz#c?w}
zQD08F$;BoECi%Tl0YXn-12bX)2i^Q=!f
z%}yfkcC2qd&s>F?-=nfPaDMxJ{UqGR_j#H2K9$iuvvHA6)_5SPwHj8z9WKA8a=wmo
z?JNe-uePgsm567xC3b+g8pqPVXfBl$&6n-N?q2*tV;(>#c<=<{!`rWI
zf+64hZFDI{nsa2+eD3mdRxMptV=?oRuLpPp<4#>hIdm|%M2=5fuwt!;a~<%WlrnPNk5B(^EO
z&UZ`!u%1SBcr$@PbJ!O)zaIDtopbbPEy`>rhjludt;O9eBsibE(DMrN*j1KH^!d(x
zxM7oyhhg1PXBRUDN`Nbu+yS!2x