From c5cfe919c723b222d7bff5b1a95df2a31c2ed566 Mon Sep 17 00:00:00 2001 From: mhjensen Date: Wed, 26 Sep 2018 08:27:07 +0200 Subject: [PATCH] Updated log reg --- doc/pub/LogReg/html/._LogReg-bs000.html | 40 ++++--- doc/pub/LogReg/html/._LogReg-bs001.html | 38 +++--- doc/pub/LogReg/html/._LogReg-bs002.html | 40 ++++--- doc/pub/LogReg/html/._LogReg-bs003.html | 48 ++++---- doc/pub/LogReg/html/._LogReg-bs004.html | 42 ++++--- doc/pub/LogReg/html/._LogReg-bs005.html | 45 ++++--- doc/pub/LogReg/html/._LogReg-bs006.html | 54 ++++++--- doc/pub/LogReg/html/._LogReg-bs007.html | 64 +++++----- doc/pub/LogReg/html/._LogReg-bs008.html | 58 +++++---- doc/pub/LogReg/html/._LogReg-bs009.html | 61 +++++----- doc/pub/LogReg/html/._LogReg-bs010.html | 55 ++++++--- doc/pub/LogReg/html/._LogReg-bs011.html | 59 +++++---- doc/pub/LogReg/html/._LogReg-bs012.html | 56 ++++----- doc/pub/LogReg/html/._LogReg-bs013.html | 57 ++++----- doc/pub/LogReg/html/._LogReg-bs014.html | 60 ++++------ doc/pub/LogReg/html/._LogReg-bs015.html | 101 ++++++---------- doc/pub/LogReg/html/LogReg-bs.html | 40 ++++--- doc/pub/LogReg/html/LogReg-reveal.html | 102 +++++++++------- doc/pub/LogReg/html/LogReg-solarized.html | 112 +++++++++++------- doc/pub/LogReg/html/LogReg.html | 112 +++++++++++------- doc/pub/LogReg/ipynb/LogReg.ipynb | 78 +++++++----- doc/pub/LogReg/ipynb/ipynb-LogReg-src.tar.gz | Bin 210 -> 209 bytes .../LogReg/pdf/LogReg-beamer-handouts2x3.pdf | Bin 255477 -> 256334 bytes doc/pub/LogReg/pdf/LogReg-beamer.pdf | Bin 245185 -> 246755 bytes doc/pub/LogReg/pdf/LogReg-minted.pdf | Bin 287260 -> 288128 bytes doc/src/LogisticRegression/LogReg.do.txt | 70 +++++++---- 26 files changed, 771 insertions(+), 621 deletions(-) diff --git a/doc/pub/LogReg/html/._LogReg-bs000.html b/doc/pub/LogReg/html/._LogReg-bs000.html index ae97688d0..b583d2d04 100644 --- a/doc/pub/LogReg/html/._LogReg-bs000.html +++ b/doc/pub/LogReg/html/._LogReg-bs000.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -139,7 +143,7 @@ MathJax.Hub.Config({
    [2] Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University

    -

    Sep 25, 2018

    +

    Sep 26, 2018


    @@ -156,7 +160,7 @@ MathJax.Hub.Config({

  • 9
  • 10
  • ...
    • -
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs001.html b/doc/pub/LogReg/html/._LogReg-bs001.html index 85b316868..d07fe970b 100644 --- a/doc/pub/LogReg/html/._LogReg-bs001.html +++ b/doc/pub/LogReg/html/._LogReg-bs001.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -172,7 +176,7 @@ failure etc.
  • 10
  • 11
  • ...
    • -
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs002.html b/doc/pub/LogReg/html/._LogReg-bs002.html index bbc4c96a9..a90ee5fb8 100644 --- a/doc/pub/LogReg/html/._LogReg-bs002.html +++ b/doc/pub/LogReg/html/._LogReg-bs002.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -123,7 +127,7 @@ MathJax.Hub.Config({

    Optimization and Deep learning

    -Logistic regression will also serve as our stepping towards neural +Logistic regression will also serve as our stepping stone towards neural network algorithms and supervised deep learning. For logistic learning, the minimization of the cost function leads to a non-linear equation in the parameters \( \hat{\beta} \). The optmization of the problem calls therefore for minimization algorithms. This forms the bottle neck of all machine learning algorithms, namely how to find reliable minima of a multi-variable function. This leads us to the family of gradient descent methods. The latter are the working horses of basically all modern machine learning algorithms. @@ -151,7 +155,7 @@ models, as we will see later.

  • 11
  • 12
  • ...
    • -
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs003.html b/doc/pub/LogReg/html/._LogReg-bs003.html index ae24aff62..68633dc0d 100644 --- a/doc/pub/LogReg/html/._LogReg-bs003.html +++ b/doc/pub/LogReg/html/._LogReg-bs003.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -124,20 +128,20 @@ MathJax.Hub.Config({

    We consider the case where the dependent variables, also called the -responses or the outcomes) \( y_i \) are discrete and only take values +responses or the outcomes, \( y_i \) are discrete and only take values from \( k=0,\dots,K-1 \) (i.e. \( K \) classes).

    The goal is to predict the -output classes from the design matrix \( X\in\mathbb{R}^{n\times p} \) -made of \( n \) samples, each of which carries \( p \) features. The +output classes from the design matrix \( \hat{X}\in\mathbb{R}^{n\times p} \) +made of \( n \) samples, each of which carries \( p \) features or predictors. The primary goal is to identify the classes to which new unseen samples belong.

    -Let us specialize to the case of two classes only, with outputs \( y_i=0 \) and \( y_i=1 \). Our outcomes could represent the status of a credit card use who could default or not on her/his debt. That is +Let us specialize to the case of two classes only, with outputs \( y_i=0 \) and \( y_i=1 \). Our outcomes could represent the status of a credit card user who could default or not on her/his credit card debt. That is $$ -y_i = \begin{bmatrix} 0 & \mathrm{no default}\\ 1 & \mathrm{default} \end{bmatrix}. +y_i = \begin{bmatrix} 0 & \mathrm{no}\\ 1 & \mathrm{yes} \end{bmatrix}. $$

    @@ -159,7 +163,7 @@ $$

  • 12
  • 13
  • ...
    • -
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs004.html b/doc/pub/LogReg/html/._LogReg-bs004.html index ec1fa7a21..63a7b3f19 100644 --- a/doc/pub/LogReg/html/._LogReg-bs004.html +++ b/doc/pub/LogReg/html/._LogReg-bs004.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -130,13 +134,13 @@ We would then have our weighted linear combination, namely $$ \begin{equation} -\hat{y} = \hat{X}_i^T\hat{\beta} + \hat{\epsilon}, +\hat{y} = \hat{X}^T\hat{\beta} + \hat{\epsilon}, \tag{1} \end{equation} $$ where \( \hat{y} \) is a vector representing the possible outcomes, \( \hat{X} \) is our -\( n\times p \) design matrix and \( \hat{\beta} \) represents our estimators. +\( n\times p \) design matrix and \( \hat{\beta} \) represents our estimators/predictors.

    @@ -158,7 +162,7 @@ where \( \hat{y} \) is a vector representing the possible outcomes, \( \hat{X} \

  • 13
  • 14
  • ...
    • -
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs005.html b/doc/pub/LogReg/html/._LogReg-bs005.html index 20b83aa95..c5b864af6 100644 --- a/doc/pub/LogReg/html/._LogReg-bs005.html +++ b/doc/pub/LogReg/html/._LogReg-bs005.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -137,6 +141,13 @@ literature. This model is extremely simple. However, in many cases it is more favorable to use a ``soft" classifier that outputs the probability of a given category. This leads us to the logistic function. +

    +The code for plotting the perceptron can be seen here. This si nothing but the standard Heaviside step function. +

    + + +

    
+

    @@ -158,7 +169,7 @@ the probability of a given category. This leads us to the logistic function.

  • 14
  • 15
  • ...
    • -
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs006.html b/doc/pub/LogReg/html/._LogReg-bs006.html index e6f6e2210..b395f7c6d 100644 --- a/doc/pub/LogReg/html/._LogReg-bs006.html +++ b/doc/pub/LogReg/html/._LogReg-bs006.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -123,13 +127,27 @@ MathJax.Hub.Config({

    The logistic function

    -A widely used model is the so-called logistic (or sigmoid) function which ranges from \( 0 \) to \( 1 \), +The perceptron is an example of a ``hard classification" model. We +will encounter this model when we discuss neural networks as +well. Each datapoint is deterministically assigned to a category (i.e +\( y_i=0 \) or \( y_i=1 \)). In many cases, it is favorable to have a "soft" +classifier that outputs the probability of a given category rather +than a single value. For example, given \( x_i \), the classifier +outputs the probability of being in a category \( k \). Logistic regression +is the most common example of a so-called soft classifier. In logistic +regression, the probability that a data point \( x_i \) +belongs to a category \( y_i=\{0,1\} \) is given by the so-called logit function (or Sigmoid) which is meant to represent the likelihood for a given event, $$ p(t) = \frac{1}{1+\mathrm \exp{-t}}=\frac{\exp{t}}{1+\mathrm \exp{t}}. $$ Note that \( 1-p(t)= p(-t) \). +The following code plots the logistic function. +

    + +

    
+

    @@ -151,6 +169,8 @@ Note that \( 1-p(t)= p(-t) \).

  • 14
  • 15
  • 16
    • +
  • ...
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs007.html b/doc/pub/LogReg/html/._LogReg-bs007.html index 7eb3a129b..eab7581ad 100644 --- a/doc/pub/LogReg/html/._LogReg-bs007.html +++ b/doc/pub/LogReg/html/._LogReg-bs007.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -120,19 +124,24 @@ MathJax.Hub.Config({ -

    Probabilistic classifiers

    +

    Two parameters

    -The perceptron is an example of a ``hard classification" model. We -will encounter this model when we discuss neural networks as -well. Each datapoint is deterministically assigned to a category (i.e -\( y_i=0 \) or \( y_i=1 \)). In many cases, it is favorable to have a "soft" -classifier that outputs the probability of a given category rather -than a single value. For example, given \( x_i \), the classifier -outputs the probability of being in a category \( k \). Logistic regression -is the most common example of a so-called soft classifier. In logistic -regression, the probability that a data point \( x_i \) -belongs to a category \( y_i=\{0,1\} \) is given by the so-called logit function (or Sigmoid) +We assume now that we have two classes with \( y_i \) either \( 0 \) or \( 1 \). Furthermore we assume also that we have only two parameters \( \beta \) in our fitting of the Sigmoid function, that is we define probabilities +$$ +\begin{align*} +p(y_i=1|x_i,\hat{\beta}) &= \frac{\exp{(\beta_0+\beta_1x_i)}}{1+\exp{(\beta_0+\beta_1x_i)}},\nonumber\\ +p(y_i=0|x_i,\hat{\beta}) &= 1 - p(y_i=1|x_i,\hat{\beta}), +\end{align*} +$$ + +where \( \hat{\beta} \) are the weights we wish to extract from data, in our case \( \beta_0 \) and \( \beta_1 \). + +

    +Note that we used +$$ +p(y_i=0\vert x_i, \hat{\beta}) = 1-p(y_i=1\vert x_i, \hat{\beta}). +$$

    @@ -155,6 +164,7 @@ belongs to a category \( y_i=\{0,1\} \) is given by the so-called logit function

  • 14
  • 15
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs008.html b/doc/pub/LogReg/html/._LogReg-bs008.html index 37f791679..845f338c3 100644 --- a/doc/pub/LogReg/html/._LogReg-bs008.html +++ b/doc/pub/LogReg/html/._LogReg-bs008.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -118,25 +122,26 @@ MathJax.Hub.Config({

     

     

     

    - + -

    Two parameters

    +

    Maximum likelihood

    -We assume now that we have two classes with \( y_i \) either \( 0 \) or \( 1 \). Furthermore we assume also that we have only two parameters \( \beta \) in our fitting of the Sigmoid function, that is we define probabilities +In order to define the total likelihood for all possible outcomes from a +dataset \( \mathcal{D}=\{(y_i,x_i)\} \), with the binary labels +\( y_i\in\{0,1\} \) and where the data points are drawn independently, we use the so-called Maximum Likelihood Estimation (MLE) principle. +We aim thus at maximizing +the probability of seeing the observed data. We can then approximate the +likelihood in terms of the product of the individual probabilities of a specific outcome \( y_i \), that is $$ \begin{align*} -p(y_i=1|x_i,\hat{\beta}) &= \frac{\exp{(\beta_0+\beta_1x_i)}}{1+\exp{(\beta_0+\beta_1x_i)}},\nonumber\\ -p(y_i=0|x_i,\hat{\beta}) &= 1 - p(y_i=1|x_i,\hat{\beta}), +P(\mathcal{D}|\hat{\beta})& = \prod_{i=1}^n \left[p(y_i=1|x_i,\hat{\beta})\right]^{y_i}\left[1-p(y_i=1|x_i,\hat{\beta}))\right]^{1-y_i}\nonumber \\ \end{align*} $$ -where \( \hat{\beta} \) are the weights we wish to extract from data, in our case \( \beta_0 \) and \( \beta_1 \). - -

    -Note that we used +from which we obtain the log-likelihood and our cost/loss function $$ -p(y_i=1) = 1-p(y_i=0). +\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left( y_i\log{p(y_i=1|x_i,\hat{\beta})} + (1-y_i)\log\left[1-p(y_i=1|x_i,\hat{\beta}))\right]\right). $$

    @@ -160,6 +165,7 @@ $$

  • 14
  • 15
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs009.html b/doc/pub/LogReg/html/._LogReg-bs009.html index f2d6942b1..c7d9b85b1 100644 --- a/doc/pub/LogReg/html/._LogReg-bs009.html +++ b/doc/pub/LogReg/html/._LogReg-bs009.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -118,28 +122,26 @@ MathJax.Hub.Config({

     

     

     

    - + -

    Maximum likelihood

    +

    The cost function rewritten

    -In order to define the total likelihood for all possible outcomes from a -dataset \( \mathcal{D}=\{(y_i,x_i)\} \), with the binary labels -\( y_i\in\{0,1\} \) and where the data points are drawn independently, we use the so-called Maximum Likelihood Estimation (MLE) principle. -We aim thus at maximizing -the probability of seeing the observed data. We can then approximate the -likelihood in terms of the product of the individual probabilities of a specific outcome \( y_i \), that is +Reordering the logarithms, we can rewrite the cost/loss function as $$ -\begin{align*} -P(\mathcal{D}|\hat{\beta})& = \prod_{i=1}^n \left[p(y_i=1|x_i,\hat{\beta})\right]^{y_i}\left[1-p(y_i=1|x_i,\hat{\beta}))\right]^{1-y_i}\nonumber \\ -\end{align*} +\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{(1+\exp{(\beta_0+\beta_1x_i)})}\right). $$ -from which we obtain the log-likelihood and our cost/loss function +

    +The maximum likelihood estimator is defined as the set of parameters that maximize the log-likelihood where we maximize with respect to \( \beta \). +Since the cost (error) function is just the negative log-likelihood, for logistic regression we have that $$ -\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n y_i\log{p(y_i=1|x_i,\hat{\beta})} + (1-y_i)\log\left[1-p(y_i=1|x_i,\hat{\beta}))\right]. +\mathcal{C}(\hat{\beta})=-\sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{(1+\exp{(\beta_0+\beta_1x_i)})}\right). $$ +This equation is known in statistics as the cross entropy. Finally, we note that just as in linear regression, +in practice we often supplement the cross-entropy with additional regularization terms, usually \( L_1 \) and \( L_2 \) regularization as we did for Ridge and Lasso regression. +

    @@ -161,6 +163,7 @@ $$

  • 14
  • 15
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs010.html b/doc/pub/LogReg/html/._LogReg-bs010.html index d0ee85632..d8b894fe4 100644 --- a/doc/pub/LogReg/html/._LogReg-bs010.html +++ b/doc/pub/LogReg/html/._LogReg-bs010.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -120,14 +124,26 @@ MathJax.Hub.Config({ -

    The cost function rewritten

    +

    Minimizing the cross entropy

    -Reordering the logarithms, we can rewrite the cost/loss function as +The cross entropy is a convex function of the weights \( \hat{\beta} \) and, +therefore, any local minimizer is a global minimizer. + +

    +Minimizing this +cost function with respect to the two parameters \( \beta_0 \) and \( \beta_1 \) we obtain + $$ -\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{1+\exp{(\beta_0+\beta_1x_i)}}\right). +\frac{\partial \mathcal{C}(\hat{\beta})}{\partial \beta_0} = -\sum_{i=1}^n \left(y_i -\frac{\exp{(\beta_0+\beta_1x_i)}}{1+\exp{(\beta_0+\beta_1x_i)}}\right), $$ +and +$$ +\frac{\partial \mathcal{C}(\hat{\beta})}{\partial \beta_1} = -\sum_{i=1}^n \left(y_ix_i -x_i\frac{\exp{(\beta_0+\beta_1x_i)}}{1+\exp{(\beta_0+\beta_1x_i)}}\right). +$$ + +

    diff --git a/doc/pub/LogReg/html/._LogReg-bs011.html b/doc/pub/LogReg/html/._LogReg-bs011.html index 0a94cd42d..26ff89a72 100644 --- a/doc/pub/LogReg/html/._LogReg-bs011.html +++ b/doc/pub/LogReg/html/._LogReg-bs011.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -119,14 +123,26 @@ MathJax.Hub.Config({ -The maximum likelihood estimator is defined as the set of parameters that maximize the log-likelihood where we maximize with respect to \( \beta \). -Since the cost (error) function is just the negative log-likelihood, for logistic regression we have that + +

    A more compact expression

    + +

    +Let us now define a vector \( \hat{y} \) with \( n \) elements \( y_i \), an +\( n\times p \) matrix \( \hat{X} \) which contains the \( x_i \) values and a +vector \( \hat{p} \) of fitted probabilities \( p(y_i\vert x_i,\hat{\beta}) \). We can rewrite in a more compact form the first +derivative of cost function as + $$ -\mathcal{C}(\hat{\beta})=-\sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{1+\exp{(\beta_0+\beta_1x_i)}}\right). +\frac{\partial \mathcal{C}(\hat{\beta})}{\partial \hat{\beta}} = -\hat{X}^T\left(\hat{y}-\hat{p}\right). $$ -This equation is known in statistics as the cross entropy. Finally, we note that just as in linear regression, -in practice we often supplement the cross-entropy with additional regularization terms, usually \( L_1 \) and \( L_2 \) regularization as we did for Ridge and Lasso regression. +

    +If we in addition define a diagonal matrix \( \hat{W} \) with elements +\( p(y_i\vert x_i,\hat{\beta})(1-p(y_i\vert x_i,\hat{\beta}) \), we can obtain a compact expression of the second derivative as + +$$ +\frac{\partial^2 \mathcal{C}(\hat{\beta})}{\partial \hat{\beta}\partial \hat{\beta}^T} = \hat{X}^T\hat{W}\hat{X}. +$$

    @@ -148,6 +164,7 @@ in practice we often supplement the cross-entropy with additional regularization

  • 14
  • 15
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs012.html b/doc/pub/LogReg/html/._LogReg-bs012.html index 8747675c5..8777c8869 100644 --- a/doc/pub/LogReg/html/._LogReg-bs012.html +++ b/doc/pub/LogReg/html/._LogReg-bs012.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -120,24 +124,7 @@ MathJax.Hub.Config({ -

    Minimizing the cross entropy

    - -

    -The cross entropy is a convex function of the weights \( \hat{\beta} \) and, -therefore, any local minimizer is a global minimizer. - -

    -Minimizing this -cost function with respect to the two parameters \( \beta_0 \) and \( \beta_1 \) we obtain - -$$ -\frac{\partial \mathcal{C}(\hat{\beta})}{\partial \beta_0} = -\sum_{i=1}^n \left(y_i -\frac{\exp{(\beta_0+\beta_1x_i)}}{1+\exp{(\beta_0+\beta_1x_i)}}\right), -$$ - -and -$$ -\frac{\partial \mathcal{C}(\hat{\beta})}{\partial \beta_1} = -\sum_{i=1}^n \left(y_ix_i -x_i\frac{\exp{(\beta_0+\beta_1x_i)}}{1+\exp{(\beta_0+\beta_1x_i)}}\right). -$$ +

    Extending to more predictors

    @@ -158,6 +145,7 @@ $$

  • 14
  • 15
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs013.html b/doc/pub/LogReg/html/._LogReg-bs013.html index 61ea427f6..49249443a 100644 --- a/doc/pub/LogReg/html/._LogReg-bs013.html +++ b/doc/pub/LogReg/html/._LogReg-bs013.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -120,25 +124,7 @@ MathJax.Hub.Config({ -

    A more compact expression

    - -

    -Let us now define a vector \( \hat{y} \) with \( n \) elements \( y_i \), an -\( n\times p \) matrix \( \hat{X} \) which contains the \( x_i \) values and a -vector \( \hat{p} \) of fitted probabilities \( p(y_i\vert x_i,\hat{\beta}) \). We can rewrite in a more compact form the first -derivative of cost function as - -$$ -\frac{\partial \mathcal{C}(\hat{\beta})}{\partial \hat{\beta}} = -\hat{X}^T\left(\hat{y}-\hat{p}\right). -$$ - -

    -If we in addition define a diagonal matrix \( \hat{W} \) with elements -\( p(y_i\vert x_i,\hat{\beta})(1-p(y_i\vert x_i,\hat{\beta}) \), we can obtain a compact expression of the second derivative as - -$$ -\frac{\partial^2 \mathcal{C}(\hat{\beta})}{\partial \hat{\beta}\partial \hat{\beta}^T} = \hat{X}^T\hat{W}\hat{X}. -$$ +

    Including more classes

    @@ -158,6 +144,7 @@ $$

  • 14
  • 15
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs014.html b/doc/pub/LogReg/html/._LogReg-bs014.html index b7a8dcb07..83da3d97b 100644 --- a/doc/pub/LogReg/html/._LogReg-bs014.html +++ b/doc/pub/LogReg/html/._LogReg-bs014.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -120,30 +124,11 @@ MathJax.Hub.Config({ -

    A scikit-learn example

    +

    Optimizing the cost function

    +Newton's method and gradient descent methods - -

    import numpy as np
-import matplotlib.pyplot as plt
-from sklearn import datasets
-iris = datasets.load_iris()
-list(iris.keys())
-['data', 'target_names', 'feature_names', 'target', 'DESCR']
-X = iris["data"][:, 3:] # petal width
-y = (iris["target"] == 2).astype(np.int) # 1 if Iris-Virginica, else 0
-
-from sklearn.linear_model import LogisticRegression
-log_reg = LogisticRegression()
-log_reg.fit(X, y)
-
-X_new = np.linspace(0, 3, 1000).reshape(-1, 1)
-y_proba = log_reg.predict_proba(X_new)
-plt.plot(X_new, y_proba[:, 1], "g-", label="Iris-Virginica")
-plt.plot(X_new, y_proba[:, 0], "b--", label="Not Iris-Virginica")
-plt.show()
-

    @@ -161,6 +146,7 @@ plt.show()

  • 14
  • 15
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/._LogReg-bs015.html b/doc/pub/LogReg/html/._LogReg-bs015.html index b9cf8ae1f..edca491af 100644 --- a/doc/pub/LogReg/html/._LogReg-bs015.html +++ b/doc/pub/LogReg/html/._LogReg-bs015.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -120,62 +124,31 @@ MathJax.Hub.Config({ -

    A simple classification problem

    +

    A scikit-learn example

    + 

    import numpy as np
-from sklearn import datasets, linear_model
 import matplotlib.pyplot as plt
+from sklearn import datasets
+iris = datasets.load_iris()
+list(iris.keys())
+['data', 'target_names', 'feature_names', 'target', 'DESCR']
+X = iris["data"][:, 3:] # petal width
+y = (iris["target"] == 2).astype(np.int) # 1 if Iris-Virginica, else 0
 
+from sklearn.linear_model import LogisticRegression
+log_reg = LogisticRegression()
+log_reg.fit(X, y)
 
-def generate_data():
-    np.random.seed(0)
-    X, y = datasets.make_moons(200, noise=0.20)
-    return X, y
-
-
-def visualize(X, y, clf):
-    # plt.scatter(X[:, 0], X[:, 1], s=40, c=y, cmap=plt.cm.Spectral)
-    # plt.show()
-    plot_decision_boundary(lambda x: clf.predict(x), X, y)
-    plt.title("Logistic Regression")
-
-
-def plot_decision_boundary(pred_func, X, y):
-    # Set min and max values and give it some padding
-    x_min, x_max = X[:, 0].min() - .5, X[:, 0].max() + .5
-    y_min, y_max = X[:, 1].min() - .5, X[:, 1].max() + .5
-    h = 0.01
-    # Generate a grid of points with distance h between them
-    xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))
-    # Predict the function value for the whole gid
-    Z = pred_func(np.c_[xx.ravel(), yy.ravel()])
-    Z = Z.reshape(xx.shape)
-    # Plot the contour and training examples
-    plt.contourf(xx, yy, Z, cmap=plt.cm.Spectral)
-    plt.scatter(X[:, 0], X[:, 1], c=y, cmap=plt.cm.Spectral)
-    plt.show()
-
-
-def classify(X, y):
-    clf = linear_model.LogisticRegressionCV()
-    clf.fit(X, y)
-    return clf
-
-
-def main():
-    X, y = generate_data()
-    # visualize(X, y)
-    clf = classify(X, y)
-    visualize(X, y, clf)
-
-
-if __name__ == "__main__":
-    main()
+X_new = np.linspace(0, 3, 1000).reshape(-1, 1)
+y_proba = log_reg.predict_proba(X_new)
+plt.plot(X_new, y_proba[:, 1], "g-", label="Iris-Virginica")
+plt.plot(X_new, y_proba[:, 0], "b--", label="Not Iris-Virginica")
+plt.show()

    -

    diff --git a/doc/pub/LogReg/html/LogReg-bs.html b/doc/pub/LogReg/html/LogReg-bs.html index ae97688d0..b583d2d04 100644 --- a/doc/pub/LogReg/html/LogReg-bs.html +++ b/doc/pub/LogReg/html/LogReg-bs.html @@ -46,14 +46,16 @@ Automatically generated HTML file from DocOnce source ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -97,14 +99,16 @@ MathJax.Hub.Config({
  • Linear classifier
  • Some selected properties
  • The logistic function
    • -
  • Probabilistic classifiers
    • -
  • Two parameters
    • -
  • Maximum likelihood
    • -
  • The cost function rewritten
    • -
  • Minimizing the cross entropy
    • -
  • A more compact expression
    • -
  • A scikit-learn example
    • -
  • A simple classification problem
    • +
  • Two parameters
    • +
  • Maximum likelihood
    • +
  • The cost function rewritten
    • +
  • Minimizing the cross entropy
    • +
  • A more compact expression
    • +
  • Extending to more predictors
    • +
  • Including more classes
    • +
  • Optimizing the cost function
    • +
  • A scikit-learn example
    • +
  • A simple classification problem
    • @@ -139,7 +143,7 @@ MathJax.Hub.Config({
    [2] Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University

    -

    Sep 25, 2018

    +

    Sep 26, 2018


    @@ -156,7 +160,7 @@ MathJax.Hub.Config({

  • 9
  • 10
  • ...
    • -
  • 16
    • +
  • 17
  • »
    • diff --git a/doc/pub/LogReg/html/LogReg-reveal.html b/doc/pub/LogReg/html/LogReg-reveal.html index 71357da53..2615d7966 100644 --- a/doc/pub/LogReg/html/LogReg-reveal.html +++ b/doc/pub/LogReg/html/LogReg-reveal.html @@ -148,7 +148,7 @@ MathJax.Hub.Config({
    [2] Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University

     
    -

    Sep 25, 2018

    +

    Sep 26, 2018


    @@ -197,7 +197,7 @@ failure etc.

    Optimization and Deep learning

    -Logistic regression will also serve as our stepping towards neural +Logistic regression will also serve as our stepping stone towards neural network algorithms and supervised deep learning. For logistic learning, the minimization of the cost function leads to a non-linear equation in the parameters \( \hat{\beta} \). The optmization of the problem calls therefore for minimization algorithms. This forms the bottle neck of all machine learning algorithms, namely how to find reliable minima of a multi-variable function. This leads us to the family of gradient descent methods. The latter are the working horses of basically all modern machine learning algorithms. @@ -214,21 +214,21 @@ models, as we will see later.

    We consider the case where the dependent variables, also called the -responses or the outcomes) \( y_i \) are discrete and only take values +responses or the outcomes, \( y_i \) are discrete and only take values from \( k=0,\dots,K-1 \) (i.e. \( K \) classes).

    The goal is to predict the -output classes from the design matrix \( X\in\mathbb{R}^{n\times p} \) -made of \( n \) samples, each of which carries \( p \) features. The +output classes from the design matrix \( \hat{X}\in\mathbb{R}^{n\times p} \) +made of \( n \) samples, each of which carries \( p \) features or predictors. The primary goal is to identify the classes to which new unseen samples belong.

    -Let us specialize to the case of two classes only, with outputs \( y_i=0 \) and \( y_i=1 \). Our outcomes could represent the status of a credit card use who could default or not on her/his debt. That is +Let us specialize to the case of two classes only, with outputs \( y_i=0 \) and \( y_i=1 \). Our outcomes could represent the status of a credit card user who could default or not on her/his credit card debt. That is

     
    $$ -y_i = \begin{bmatrix} 0 & \mathrm{no default}\\ 1 & \mathrm{default} \end{bmatrix}. +y_i = \begin{bmatrix} 0 & \mathrm{no}\\ 1 & \mathrm{yes} \end{bmatrix}. $$

     
    @@ -246,14 +246,14 @@ weighted linear combination, namely

     
    $$ \begin{equation} -\hat{y} = \hat{X}_i^T\hat{\beta} + \hat{\epsilon}, +\hat{y} = \hat{X}^T\hat{\beta} + \hat{\epsilon}, \tag{1} \end{equation} $$

     
    where \( \hat{y} \) is a vector representing the possible outcomes, \( \hat{X} \) is our -\( n\times p \) design matrix and \( \hat{\beta} \) represents our estimators. +\( n\times p \) design matrix and \( \hat{\beta} \) represents our estimators/predictors. @@ -274,27 +274,20 @@ We will encounter this model in our first demonstration of neural networks. Hist literature. This model is extremely simple. However, in many cases it is more favorable to use a ``soft" classifier that outputs the probability of a given category. This leads us to the logistic function. + +

    +The code for plotting the perceptron can be seen here. This si nothing but the standard Heaviside step function. +

    + + +

    
+

    The logistic function

    -

    -A widely used model is the so-called logistic (or sigmoid) function which ranges from \( 0 \) to \( 1 \), -

     
    -$$ -p(t) = \frac{1}{1+\mathrm \exp{-t}}=\frac{\exp{t}}{1+\mathrm \exp{t}}. -$$ -

     
    - -Note that \( 1-p(t)= p(-t) \). -

    - - -
    -

    Probabilistic classifiers

    -

    The perceptron is an example of a ``hard classification" model. We will encounter this model when we discuss neural networks as @@ -305,12 +298,25 @@ than a single value. For example, given \( x_i \), the classifier outputs the probability of being in a category \( k \). Logistic regression is the most common example of a so-called soft classifier. In logistic regression, the probability that a data point \( x_i \) -belongs to a category \( y_i=\{0,1\} \) is given by the so-called logit function (or Sigmoid) +belongs to a category \( y_i=\{0,1\} \) is given by the so-called logit function (or Sigmoid) which is meant to represent the likelihood for a given event, +

     
    +$$ +p(t) = \frac{1}{1+\mathrm \exp{-t}}=\frac{\exp{t}}{1+\mathrm \exp{t}}. +$$ +

     
    + +Note that \( 1-p(t)= p(-t) \). +The following code plots the logistic function. +

    + + +

    
+
    -

    Two parameters

    +

    Two parameters

    We assume now that we have two classes with \( y_i \) either \( 0 \) or \( 1 \). Furthermore we assume also that we have only two parameters \( \beta \) in our fitting of the Sigmoid function, that is we define probabilities @@ -329,19 +335,19 @@ where \( \hat{\beta} \) are the weights we wish to extract from data, in our cas Note that we used

     
    $$ -p(y_i=1) = 1-p(y_i=0). +p(y_i=0\vert x_i, \hat{\beta}) = 1-p(y_i=1\vert x_i, \hat{\beta}). $$

     

    -

    Maximum likelihood

    +

    Maximum likelihood

    In order to define the total likelihood for all possible outcomes from a dataset \( \mathcal{D}=\{(y_i,x_i)\} \), with the binary labels -\( y_i\in\{0,1\} \) and where the data points are drawn independently, we use the so-called Maximum Likelihood Estimation (MLE) principle. +\( y_i\in\{0,1\} \) and where the data points are drawn independently, we use the so-called Maximum Likelihood Estimation (MLE) principle. We aim thus at maximizing the probability of seeing the observed data. We can then approximate the likelihood in terms of the product of the individual probabilities of a specific outcome \( y_i \), that is @@ -356,31 +362,29 @@ $$ from which we obtain the log-likelihood and our cost/loss function

     
    $$ -\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n y_i\log{p(y_i=1|x_i,\hat{\beta})} + (1-y_i)\log\left[1-p(y_i=1|x_i,\hat{\beta}))\right]. +\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left( y_i\log{p(y_i=1|x_i,\hat{\beta})} + (1-y_i)\log\left[1-p(y_i=1|x_i,\hat{\beta}))\right]\right). $$

     

    -

    The cost function rewritten

    +

    The cost function rewritten

    Reordering the logarithms, we can rewrite the cost/loss function as

     
    $$ -\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{1+\exp{(\beta_0+\beta_1x_i)}}\right). +\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{(1+\exp{(\beta_0+\beta_1x_i)})}\right). $$

     
    -

    - -
    +

    The maximum likelihood estimator is defined as the set of parameters that maximize the log-likelihood where we maximize with respect to \( \beta \). Since the cost (error) function is just the negative log-likelihood, for logistic regression we have that

     
    $$ -\mathcal{C}(\hat{\beta})=-\sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{1+\exp{(\beta_0+\beta_1x_i)}}\right). +\mathcal{C}(\hat{\beta})=-\sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{(1+\exp{(\beta_0+\beta_1x_i)})}\right). $$

     
    @@ -390,7 +394,7 @@ in practice we often supplement the cross-entropy with additional regularization

    -

    Minimizing the cross entropy

    +

    Minimizing the cross entropy

    The cross entropy is a convex function of the weights \( \hat{\beta} \) and, @@ -416,7 +420,7 @@ $$

    -

    A more compact expression

    +

    A more compact expression

    Let us now define a vector \( \hat{y} \) with \( n \) elements \( y_i \), an @@ -443,7 +447,25 @@ $$

    -

    A scikit-learn example

    +

    Extending to more predictors

    +
    + + +
    +

    Including more classes

    +
    + + +
    +

    Optimizing the cost function

    + +

    +Newton's method and gradient descent methods +

    + + +
    +

    A scikit-learn example

    @@ -471,7 +493,7 @@ plt.show()

    -

    A simple classification problem

    +

    A simple classification problem

    diff --git a/doc/pub/LogReg/html/LogReg-solarized.html b/doc/pub/LogReg/html/LogReg-solarized.html index f37054fab..6211d4f48 100644 --- a/doc/pub/LogReg/html/LogReg-solarized.html +++ b/doc/pub/LogReg/html/LogReg-solarized.html @@ -40,14 +40,16 @@ div { text-align: justify; text-justify: inter-word; } ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -89,7 +91,7 @@ MathJax.Hub.Config({

    [2] Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University

    -

    Sep 25, 2018

    +

    Sep 26, 2018


    @@ -132,7 +134,7 @@ failure etc.

    Optimization and Deep learning

    -Logistic regression will also serve as our stepping towards neural +Logistic regression will also serve as our stepping stone towards neural network algorithms and supervised deep learning. For logistic learning, the minimization of the cost function leads to a non-linear equation in the parameters \( \hat{\beta} \). The optmization of the problem calls therefore for minimization algorithms. This forms the bottle neck of all machine learning algorithms, namely how to find reliable minima of a multi-variable function. This leads us to the family of gradient descent methods. The latter are the working horses of basically all modern machine learning algorithms. @@ -149,20 +151,20 @@ models, as we will see later.

    We consider the case where the dependent variables, also called the -responses or the outcomes) \( y_i \) are discrete and only take values +responses or the outcomes, \( y_i \) are discrete and only take values from \( k=0,\dots,K-1 \) (i.e. \( K \) classes).

    The goal is to predict the -output classes from the design matrix \( X\in\mathbb{R}^{n\times p} \) -made of \( n \) samples, each of which carries \( p \) features. The +output classes from the design matrix \( \hat{X}\in\mathbb{R}^{n\times p} \) +made of \( n \) samples, each of which carries \( p \) features or predictors. The primary goal is to identify the classes to which new unseen samples belong.

    -Let us specialize to the case of two classes only, with outputs \( y_i=0 \) and \( y_i=1 \). Our outcomes could represent the status of a credit card use who could default or not on her/his debt. That is +Let us specialize to the case of two classes only, with outputs \( y_i=0 \) and \( y_i=1 \). Our outcomes could represent the status of a credit card user who could default or not on her/his credit card debt. That is $$ -y_i = \begin{bmatrix} 0 & \mathrm{no default}\\ 1 & \mathrm{default} \end{bmatrix}. +y_i = \begin{bmatrix} 0 & \mathrm{no}\\ 1 & \mathrm{yes} \end{bmatrix}. $$

    @@ -178,13 +180,13 @@ We would then have our weighted linear combination, namely $$ \begin{equation} -\hat{y} = \hat{X}_i^T\hat{\beta} + \hat{\epsilon}, +\hat{y} = \hat{X}^T\hat{\beta} + \hat{\epsilon}, \label{_auto1} \end{equation} $$ where \( \hat{y} \) is a vector representing the possible outcomes, \( \hat{X} \) is our -\( n\times p \) design matrix and \( \hat{\beta} \) represents our estimators. +\( n\times p \) design matrix and \( \hat{\beta} \) represents our estimators/predictors.











    @@ -206,24 +208,18 @@ literature. This model is extremely simple. However, in many cases it is more favorable to use a ``soft" classifier that outputs the probability of a given category. This leads us to the logistic function. +

    +The code for plotting the perceptron can be seen here. This si nothing but the standard Heaviside step function. +

    + + +

    
+











    The logistic function

    -

    -A widely used model is the so-called logistic (or sigmoid) function which ranges from \( 0 \) to \( 1 \), -$$ -p(t) = \frac{1}{1+\mathrm \exp{-t}}=\frac{\exp{t}}{1+\mathrm \exp{t}}. -$$ - -Note that \( 1-p(t)= p(-t) \). - -

    -









    - -

    Probabilistic classifiers

    -

    The perceptron is an example of a ``hard classification" model. We will encounter this model when we discuss neural networks as @@ -234,12 +230,22 @@ than a single value. For example, given \( x_i \), the classifier outputs the probability of being in a category \( k \). Logistic regression is the most common example of a so-called soft classifier. In logistic regression, the probability that a data point \( x_i \) -belongs to a category \( y_i=\{0,1\} \) is given by the so-called logit function (or Sigmoid) +belongs to a category \( y_i=\{0,1\} \) is given by the so-called logit function (or Sigmoid) which is meant to represent the likelihood for a given event, +$$ +p(t) = \frac{1}{1+\mathrm \exp{-t}}=\frac{\exp{t}}{1+\mathrm \exp{t}}. +$$ +Note that \( 1-p(t)= p(-t) \). +The following code plots the logistic function. +

    + + +

    
+











    -

    Two parameters

    +

    Two parameters

    We assume now that we have two classes with \( y_i \) either \( 0 \) or \( 1 \). Furthermore we assume also that we have only two parameters \( \beta \) in our fitting of the Sigmoid function, that is we define probabilities @@ -255,18 +261,18 @@ where \( \hat{\beta} \) are the weights we wish to extract from data, in our cas

    Note that we used $$ -p(y_i=1) = 1-p(y_i=0). +p(y_i=0\vert x_i, \hat{\beta}) = 1-p(y_i=1\vert x_i, \hat{\beta}). $$

    -

    Maximum likelihood

    +

    Maximum likelihood

    In order to define the total likelihood for all possible outcomes from a dataset \( \mathcal{D}=\{(y_i,x_i)\} \), with the binary labels -\( y_i\in\{0,1\} \) and where the data points are drawn independently, we use the so-called Maximum Likelihood Estimation (MLE) principle. +\( y_i\in\{0,1\} \) and where the data points are drawn independently, we use the so-called Maximum Likelihood Estimation (MLE) principle. We aim thus at maximizing the probability of seeing the observed data. We can then approximate the likelihood in terms of the product of the individual probabilities of a specific outcome \( y_i \), that is @@ -278,25 +284,25 @@ $$ from which we obtain the log-likelihood and our cost/loss function $$ -\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n y_i\log{p(y_i=1|x_i,\hat{\beta})} + (1-y_i)\log\left[1-p(y_i=1|x_i,\hat{\beta}))\right]. +\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left( y_i\log{p(y_i=1|x_i,\hat{\beta})} + (1-y_i)\log\left[1-p(y_i=1|x_i,\hat{\beta}))\right]\right). $$











    -

    The cost function rewritten

    +

    The cost function rewritten

    Reordering the logarithms, we can rewrite the cost/loss function as $$ -\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{1+\exp{(\beta_0+\beta_1x_i)}}\right). +\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{(1+\exp{(\beta_0+\beta_1x_i)})}\right). $$ -









    +

    The maximum likelihood estimator is defined as the set of parameters that maximize the log-likelihood where we maximize with respect to \( \beta \). Since the cost (error) function is just the negative log-likelihood, for logistic regression we have that $$ -\mathcal{C}(\hat{\beta})=-\sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{1+\exp{(\beta_0+\beta_1x_i)}}\right). +\mathcal{C}(\hat{\beta})=-\sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{(1+\exp{(\beta_0+\beta_1x_i)})}\right). $$ This equation is known in statistics as the cross entropy. Finally, we note that just as in linear regression, @@ -305,7 +311,7 @@ in practice we often supplement the cross-entropy with additional regularization











    -

    Minimizing the cross entropy

    +

    Minimizing the cross entropy

    The cross entropy is a convex function of the weights \( \hat{\beta} \) and, @@ -327,7 +333,7 @@ $$











    -

    A more compact expression

    +

    A more compact expression

    Let us now define a vector \( \hat{y} \) with \( n \) elements \( y_i \), an @@ -350,7 +356,25 @@ $$











    -

    A scikit-learn example

    +

    Extending to more predictors

    + +

    +









    + +

    Including more classes

    + +

    +









    + +

    Optimizing the cost function

    + +

    +Newton's method and gradient descent methods + +

    +









    + +

    A scikit-learn example

    @@ -377,7 +401,7 @@ plt.show()











    -

    A simple classification problem

    +

    A simple classification problem

    diff --git a/doc/pub/LogReg/html/LogReg.html b/doc/pub/LogReg/html/LogReg.html index 1dab7dca5..02994c5c2 100644 --- a/doc/pub/LogReg/html/LogReg.html +++ b/doc/pub/LogReg/html/LogReg.html @@ -45,14 +45,16 @@ div { text-align: justify; text-justify: inter-word; } ('Linear classifier', 2, None, '___sec3'), ('Some selected properties', 2, None, '___sec4'), ('The logistic function', 2, None, '___sec5'), - ('Probabilistic classifiers', 2, None, '___sec6'), - ('Two parameters', 2, None, '___sec7'), - ('Maximum likelihood', 2, None, '___sec8'), - ('The cost function rewritten', 2, None, '___sec9'), - ('Minimizing the cross entropy', 2, None, '___sec10'), - ('A more compact expression', 2, None, '___sec11'), - ('A _scikit-learn_ example', 2, None, '___sec12'), - ('A simple classification problem', 2, None, '___sec13')]} + ('Two parameters', 2, None, '___sec6'), + ('Maximum likelihood', 2, None, '___sec7'), + ('The cost function rewritten', 2, None, '___sec8'), + ('Minimizing the cross entropy', 2, None, '___sec9'), + ('A more compact expression', 2, None, '___sec10'), + ('Extending to more predictors', 2, None, '___sec11'), + ('Including more classes', 2, None, '___sec12'), + ('Optimizing the cost function', 2, None, '___sec13'), + ('A _scikit-learn_ example', 2, None, '___sec14'), + ('A simple classification problem', 2, None, '___sec15')]} end of tocinfo --> @@ -94,7 +96,7 @@ MathJax.Hub.Config({

    [2] Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University

    -

    Sep 25, 2018

    +

    Sep 26, 2018


    @@ -137,7 +139,7 @@ failure etc.

    Optimization and Deep learning

    -Logistic regression will also serve as our stepping towards neural +Logistic regression will also serve as our stepping stone towards neural network algorithms and supervised deep learning. For logistic learning, the minimization of the cost function leads to a non-linear equation in the parameters \( \hat{\beta} \). The optmization of the problem calls therefore for minimization algorithms. This forms the bottle neck of all machine learning algorithms, namely how to find reliable minima of a multi-variable function. This leads us to the family of gradient descent methods. The latter are the working horses of basically all modern machine learning algorithms. @@ -154,20 +156,20 @@ models, as we will see later.

    We consider the case where the dependent variables, also called the -responses or the outcomes) \( y_i \) are discrete and only take values +responses or the outcomes, \( y_i \) are discrete and only take values from \( k=0,\dots,K-1 \) (i.e. \( K \) classes).

    The goal is to predict the -output classes from the design matrix \( X\in\mathbb{R}^{n\times p} \) -made of \( n \) samples, each of which carries \( p \) features. The +output classes from the design matrix \( \hat{X}\in\mathbb{R}^{n\times p} \) +made of \( n \) samples, each of which carries \( p \) features or predictors. The primary goal is to identify the classes to which new unseen samples belong.

    -Let us specialize to the case of two classes only, with outputs \( y_i=0 \) and \( y_i=1 \). Our outcomes could represent the status of a credit card use who could default or not on her/his debt. That is +Let us specialize to the case of two classes only, with outputs \( y_i=0 \) and \( y_i=1 \). Our outcomes could represent the status of a credit card user who could default or not on her/his credit card debt. That is $$ -y_i = \begin{bmatrix} 0 & \mathrm{no default}\\ 1 & \mathrm{default} \end{bmatrix}. +y_i = \begin{bmatrix} 0 & \mathrm{no}\\ 1 & \mathrm{yes} \end{bmatrix}. $$

    @@ -183,13 +185,13 @@ We would then have our weighted linear combination, namely $$ \begin{equation} -\hat{y} = \hat{X}_i^T\hat{\beta} + \hat{\epsilon}, +\hat{y} = \hat{X}^T\hat{\beta} + \hat{\epsilon}, \label{_auto1} \end{equation} $$ where \( \hat{y} \) is a vector representing the possible outcomes, \( \hat{X} \) is our -\( n\times p \) design matrix and \( \hat{\beta} \) represents our estimators. +\( n\times p \) design matrix and \( \hat{\beta} \) represents our estimators/predictors.











    @@ -211,24 +213,18 @@ literature. This model is extremely simple. However, in many cases it is more favorable to use a ``soft" classifier that outputs the probability of a given category. This leads us to the logistic function. +

    +The code for plotting the perceptron can be seen here. This si nothing but the standard Heaviside step function. +

    + + +

    
+











    The logistic function

    -

    -A widely used model is the so-called logistic (or sigmoid) function which ranges from \( 0 \) to \( 1 \), -$$ -p(t) = \frac{1}{1+\mathrm \exp{-t}}=\frac{\exp{t}}{1+\mathrm \exp{t}}. -$$ - -Note that \( 1-p(t)= p(-t) \). - -

    -









    - -

    Probabilistic classifiers

    -

    The perceptron is an example of a ``hard classification" model. We will encounter this model when we discuss neural networks as @@ -239,12 +235,22 @@ than a single value. For example, given \( x_i \), the classifier outputs the probability of being in a category \( k \). Logistic regression is the most common example of a so-called soft classifier. In logistic regression, the probability that a data point \( x_i \) -belongs to a category \( y_i=\{0,1\} \) is given by the so-called logit function (or Sigmoid) +belongs to a category \( y_i=\{0,1\} \) is given by the so-called logit function (or Sigmoid) which is meant to represent the likelihood for a given event, +$$ +p(t) = \frac{1}{1+\mathrm \exp{-t}}=\frac{\exp{t}}{1+\mathrm \exp{t}}. +$$ +Note that \( 1-p(t)= p(-t) \). +The following code plots the logistic function. +

    + + +

    
+











    -

    Two parameters

    +

    Two parameters

    We assume now that we have two classes with \( y_i \) either \( 0 \) or \( 1 \). Furthermore we assume also that we have only two parameters \( \beta \) in our fitting of the Sigmoid function, that is we define probabilities @@ -260,18 +266,18 @@ where \( \hat{\beta} \) are the weights we wish to extract from data, in our cas

    Note that we used $$ -p(y_i=1) = 1-p(y_i=0). +p(y_i=0\vert x_i, \hat{\beta}) = 1-p(y_i=1\vert x_i, \hat{\beta}). $$

    -

    Maximum likelihood

    +

    Maximum likelihood

    In order to define the total likelihood for all possible outcomes from a dataset \( \mathcal{D}=\{(y_i,x_i)\} \), with the binary labels -\( y_i\in\{0,1\} \) and where the data points are drawn independently, we use the so-called Maximum Likelihood Estimation (MLE) principle. +\( y_i\in\{0,1\} \) and where the data points are drawn independently, we use the so-called Maximum Likelihood Estimation (MLE) principle. We aim thus at maximizing the probability of seeing the observed data. We can then approximate the likelihood in terms of the product of the individual probabilities of a specific outcome \( y_i \), that is @@ -283,25 +289,25 @@ $$ from which we obtain the log-likelihood and our cost/loss function $$ -\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n y_i\log{p(y_i=1|x_i,\hat{\beta})} + (1-y_i)\log\left[1-p(y_i=1|x_i,\hat{\beta}))\right]. +\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left( y_i\log{p(y_i=1|x_i,\hat{\beta})} + (1-y_i)\log\left[1-p(y_i=1|x_i,\hat{\beta}))\right]\right). $$











    -

    The cost function rewritten

    +

    The cost function rewritten

    Reordering the logarithms, we can rewrite the cost/loss function as $$ -\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{1+\exp{(\beta_0+\beta_1x_i)}}\right). +\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{(1+\exp{(\beta_0+\beta_1x_i)})}\right). $$ -









    +

    The maximum likelihood estimator is defined as the set of parameters that maximize the log-likelihood where we maximize with respect to \( \beta \). Since the cost (error) function is just the negative log-likelihood, for logistic regression we have that $$ -\mathcal{C}(\hat{\beta})=-\sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{1+\exp{(\beta_0+\beta_1x_i)}}\right). +\mathcal{C}(\hat{\beta})=-\sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{(1+\exp{(\beta_0+\beta_1x_i)})}\right). $$ This equation is known in statistics as the cross entropy. Finally, we note that just as in linear regression, @@ -310,7 +316,7 @@ in practice we often supplement the cross-entropy with additional regularization











    -

    Minimizing the cross entropy

    +

    Minimizing the cross entropy

    The cross entropy is a convex function of the weights \( \hat{\beta} \) and, @@ -332,7 +338,7 @@ $$











    -

    A more compact expression

    +

    A more compact expression

    Let us now define a vector \( \hat{y} \) with \( n \) elements \( y_i \), an @@ -355,7 +361,25 @@ $$











    -

    A scikit-learn example

    +

    Extending to more predictors

    + +

    +









    + +

    Including more classes

    + +

    +









    + +

    Optimizing the cost function

    + +

    +Newton's method and gradient descent methods + +

    +









    + +

    A scikit-learn example

    @@ -382,7 +406,7 @@ plt.show()











    -

    A simple classification problem

    +

    A simple classification problem

    diff --git a/doc/pub/LogReg/ipynb/LogReg.ipynb b/doc/pub/LogReg/ipynb/LogReg.ipynb index b6532eb40..43615a1d1 100644 --- a/doc/pub/LogReg/ipynb/LogReg.ipynb +++ b/doc/pub/LogReg/ipynb/LogReg.ipynb @@ -10,7 +10,7 @@ " \n", "**Morten Hjorth-Jensen**, Department of Physics, University of Oslo and Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University\n", "\n", - "Date: **Sep 25, 2018**\n", + "Date: **Sep 26, 2018**\n", "\n", "Copyright 1999-2018, Morten Hjorth-Jensen. Released under CC Attribution-NonCommercial 4.0 license\n", "\n", @@ -50,7 +50,7 @@ "\n", "## Optimization and Deep learning\n", "\n", - "Logistic regression will also serve as our stepping towards neural\n", + "Logistic regression will also serve as our stepping stone towards neural\n", "network algorithms and supervised deep learning. For logistic\n", "learning, the minimization of the cost function leads to a non-linear\n", "equation in the parameters $\\hat{\\beta}$. The optmization of the problem calls therefore for minimization algorithms. This forms the bottle neck of all machine learning algorithms, namely how to find reliable minima of a multi-variable function. This leads us to the family of gradient descent methods. The latter are the working horses of basically all modern machine learning algorithms. \n", @@ -64,16 +64,16 @@ "## Basics\n", "\n", "We consider the case where the dependent variables, also called the\n", - "responses or the outcomes) $y_i$ are discrete and only take values\n", + "responses or the outcomes, $y_i$ are discrete and only take values\n", "from $k=0,\\dots,K-1$ (i.e. $K$ classes).\n", "\n", "The goal is to predict the\n", - "output classes from the design matrix $X\\in\\mathbb{R}^{n\\times p}$\n", - "made of $n$ samples, each of which carries $p$ features. The\n", + "output classes from the design matrix $\\hat{X}\\in\\mathbb{R}^{n\\times p}$\n", + "made of $n$ samples, each of which carries $p$ features or predictors. The\n", "primary goal is to identify the classes to which new unseen samples\n", "belong.\n", "\n", - "Let us specialize to the case of two classes only, with outputs $y_i=0$ and $y_i=1$. Our outcomes could represent the status of a credit card use who could default or not on her/his debt. That is" + "Let us specialize to the case of two classes only, with outputs $y_i=0$ and $y_i=1$. Our outcomes could represent the status of a credit card user who could default or not on her/his credit card debt. That is" ] }, { @@ -81,7 +81,7 @@ "metadata": {}, "source": [ "$$\n", - "y_i = \\begin{bmatrix} 0 & \\mathrm{no default}\\\\ 1 & \\mathrm{default} \\end{bmatrix}.\n", + "y_i = \\begin{bmatrix} 0 & \\mathrm{no}\\\\ 1 & \\mathrm{yes} \\end{bmatrix}.\n", "$$" ] }, @@ -106,7 +106,7 @@ "\n", "$$\n", "\\begin{equation}\n", - "\\hat{y} = \\hat{X}_i^T\\hat{\\beta} + \\hat{\\epsilon},\n", + "\\hat{y} = \\hat{X}^T\\hat{\\beta} + \\hat{\\epsilon},\n", "\\label{_auto1} \\tag{1}\n", "\\end{equation}\n", "$$" @@ -117,7 +117,7 @@ "metadata": {}, "source": [ "where $\\hat{y}$ is a vector representing the possible outcomes, $\\hat{X}$ is our\n", - "$n\\times p$ design matrix and $\\hat{\\beta}$ represents our estimators.\n", + "$n\\times p$ design matrix and $\\hat{\\beta}$ represents our estimators/predictors.\n", "\n", "## Some selected properties\n", "\n", @@ -134,9 +134,25 @@ "favorable to use a ``soft\" classifier that outputs\n", "the probability of a given category. This leads us to the logistic function.\n", "\n", + "The code for plotting the perceptron can be seen here. This si nothing but the standard [Heaviside step function](https://en.wikipedia.org/wiki/Heaviside_step_function)." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ "## The logistic function\n", "\n", - "A widely used model is the so-called logistic (or sigmoid) function which ranges from $0$ to $1$," + "The perceptron is an example of a ``hard classification\" model. We\n", + "will encounter this model when we discuss neural networks as\n", + "well. Each datapoint is deterministically assigned to a category (i.e\n", + "$y_i=0$ or $y_i=1$). In many cases, it is favorable to have a \"soft\"\n", + "classifier that outputs the probability of a given category rather\n", + "than a single value. For example, given $x_i$, the classifier\n", + "outputs the probability of being in a category $k$. Logistic regression\n", + "is the most common example of a so-called soft classifier. In logistic\n", + "regression, the probability that a data point $x_i$\n", + "belongs to a category $y_i=\\{0,1\\}$ is given by the so-called logit function (or Sigmoid) which is meant to represent the likelihood for a given event," ] }, { @@ -153,21 +169,13 @@ "metadata": {}, "source": [ "Note that $1-p(t)= p(-t)$.\n", - "\n", - "## Probabilistic classifiers\n", - "\n", - "The perceptron is an example of a ``hard classification\" model. We\n", - "will encounter this model when we discuss neural networks as\n", - "well. Each datapoint is deterministically assigned to a category (i.e\n", - "$y_i=0$ or $y_i=1$). In many cases, it is favorable to have a \"soft\"\n", - "classifier that outputs the probability of a given category rather\n", - "than a single value. For example, given $x_i$, the classifier\n", - "outputs the probability of being in a category $k$. Logistic regression\n", - "is the most common example of a so-called soft classifier. In logistic\n", - "regression, the probability that a data point $x_i$\n", - "belongs to a category $y_i=\\{0,1\\}$ is given by the so-called logit function (or Sigmoid)\n", - "\n", - "\n", + "The following code plots the logistic function." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ "## Two parameters\n", "\n", "We assume now that we have two classes with $y_i$ either $0$ or $1$. Furthermore we assume also that we have only two parameters $\\beta$ in our fitting of the Sigmoid function, that is we define probabilities" @@ -199,7 +207,7 @@ "metadata": {}, "source": [ "$$\n", - "p(y_i=1) = 1-p(y_i=0).\n", + "p(y_i=0\\vert x_i, \\hat{\\beta}) = 1-p(y_i=1\\vert x_i, \\hat{\\beta}).\n", "$$" ] }, @@ -212,7 +220,7 @@ "\n", "In order to define the total likelihood for all possible outcomes from a \n", "dataset $\\mathcal{D}=\\{(y_i,x_i)\\}$, with the binary labels\n", - "$y_i\\in\\{0,1\\}$ and where the data points are drawn independently, we use the so-called Maximum Likelihood Estimation (MLE) principle. \n", + "$y_i\\in\\{0,1\\}$ and where the data points are drawn independently, we use the so-called [Maximum Likelihood Estimation](https://en.wikipedia.org/wiki/Maximum_likelihood_estimation) (MLE) principle. \n", "We aim thus at maximizing \n", "the probability of seeing the observed data. We can then approximate the \n", "likelihood in terms of the product of the individual probabilities of a specific outcome $y_i$, that is" @@ -241,7 +249,7 @@ "metadata": {}, "source": [ "$$\n", - "\\mathcal{C}(\\hat{\\beta}) = \\sum_{i=1}^n y_i\\log{p(y_i=1|x_i,\\hat{\\beta})} + (1-y_i)\\log\\left[1-p(y_i=1|x_i,\\hat{\\beta}))\\right].\n", + "\\mathcal{C}(\\hat{\\beta}) = \\sum_{i=1}^n \\left( y_i\\log{p(y_i=1|x_i,\\hat{\\beta})} + (1-y_i)\\log\\left[1-p(y_i=1|x_i,\\hat{\\beta}))\\right]\\right).\n", "$$" ] }, @@ -259,7 +267,7 @@ "metadata": {}, "source": [ "$$\n", - "\\mathcal{C}(\\hat{\\beta}) = \\sum_{i=1}^n \\left(y_i(\\beta_0+\\beta_1x_i) -\\log{1+\\exp{(\\beta_0+\\beta_1x_i)}}\\right).\n", + "\\mathcal{C}(\\hat{\\beta}) = \\sum_{i=1}^n \\left(y_i(\\beta_0+\\beta_1x_i) -\\log{(1+\\exp{(\\beta_0+\\beta_1x_i)})}\\right).\n", "$$" ] }, @@ -276,7 +284,7 @@ "metadata": {}, "source": [ "$$\n", - "\\mathcal{C}(\\hat{\\beta})=-\\sum_{i=1}^n \\left(y_i(\\beta_0+\\beta_1x_i) -\\log{1+\\exp{(\\beta_0+\\beta_1x_i)}}\\right).\n", + "\\mathcal{C}(\\hat{\\beta})=-\\sum_{i=1}^n \\left(y_i(\\beta_0+\\beta_1x_i) -\\log{(1+\\exp{(\\beta_0+\\beta_1x_i)})}\\right).\n", "$$" ] }, @@ -364,6 +372,16 @@ "cell_type": "markdown", "metadata": {}, "source": [ + "## Extending to more predictors\n", + "\n", + "## Including more classes\n", + "\n", + "## Optimizing the cost function\n", + "\n", + "Newton's method and gradient descent methods\n", + "\n", + "\n", + "\n", "## A **scikit-learn** example" ] }, diff --git a/doc/pub/LogReg/ipynb/ipynb-LogReg-src.tar.gz b/doc/pub/LogReg/ipynb/ipynb-LogReg-src.tar.gz index d76ecdcb2c2fd008efc3712064c19e0aa0fd215c..c730247afabe827342b26919275fd30dde2be4ee 100644 GIT binary patch literal 209 zcmb2|=3tniwmO=D`R#e%EG9>Rw#4i67Wo-``I9Ew(e1!8F=j#Pt(o7JNAHd<$eSOi z7QH_!I`|Y0| z?q2%Rs9NHVS5*F^rK$&I{LLgT<_X;0W)}JWOI{VnKJF^{%`Z34XJkME4fRjBwH^mE JGH5U`000#)UuOUS literal 210 zcmb2|=3t1Nu_~H@`R)1hSxklk#|zKf9kn?)X-8zW-j(JA$0On$Id^>jtC)2@*td7J zQ*~B;kk|hCr~4Dz64N$U)KtZVJPlDRUp!@&wDRVkEmhvNZg%A2-gR47 zY&1wydR3=hYxykca)sCbM^0rAZ|^M3+5bQB{9DnVXY7-1|JFJEHQ<+o%8^H9hWk!w zl}TR@{b$fJcgw}6-rkdbTl{)*q|Gts_OD0D8}ctZcg1dczjM~VZ^jJB;6wi$#yHKd Jy$l)*3;->mXCeRq diff --git a/doc/pub/LogReg/pdf/LogReg-beamer-handouts2x3.pdf b/doc/pub/LogReg/pdf/LogReg-beamer-handouts2x3.pdf index 5bdf2d6a0a635f6a2ea749bb31715dd5779a2d4d..dfca580013e7ec197c08546da049004a3abfdf64 100644 GIT binary patch delta 64047 zcmZ^}RZwNkwylf1ySuwf;ZRVxyA|&4QZRu+3U_yRcXxMZ;uP-gOgR5u>)d_Tdbly3 zGDc*^7#TgkY|(mqoJ5^jK&=Y|htYDdG?rCJ43r>mECYXo0+4wxvbM2tT9@}J%lv&I zPnUo5n0oif(uUYyQBs|{uKwH=pABw$u`ds6!Qyi-pS(2EcCw3iF^a^_VV2H-)YF#d zhBG@U8Cy3Y?lP=!pN9&Ij3k@Q)>}2^qc#Dg1wqMKH0zA^F*O z5^uvO0NI)bwfi(^ezUc^^YH{WY#v#wHl2(bj2e29>tX)V`k7Q#_G8>$wd+530t1I; zEHug&x&x6QU5}=WrDbM{s7QGZ2gi)e*m{DBsWb+y^_GLL`OUk?6DdlkOC5PYJM6Hg zR8WYyT{UHEyhVh(wNMwN(r)WXP2ZZhy%9NY0LA$0VrYg7Qcmdo_QHwL8Zjwil1zX#e&WKAy>o+#*NKaXVc`my4Lj`tx`OTH9){%2kxB;yf8e!fVa z(vn()(uF;%WBO`u_imUa=h*udDkaC(F6QkPA`Fzqa!>L`QHnYBNXPl}vMcNY>IMKabO@`?hng%&RN7(Wk2sESJdN^%lHViAEy7iwuv6 z5+cP*(Io~BA_5M}x-uwE$s#T25!;5Es`Lg+o)nQ%TNRZ^ zP#RTy2L8Z>RiYQ#_6)DpF!(@0v7NgF_mry)G5=1xiuPnASW4l@x7qHx8E~*;#YhZ4 zt2(W=4fuT+h{B{g7t#5NAzd&RS#QOLUP<2I1Oqtp$oTBMVQzIyFz`HdeTR{6*< zh=8Jh=gVAd184h{#y<`|3YBrt@e7m*wY>uXw>nZ@NPNV#{Lix*gl7#7$%4irb5#b42JHr+`Z_IdpczHkiQwO8JS zNj+v_@rVQ*#C(^B+`nlRPO^zo=LJxmh8u#LfyOj!pN4I+ju#;?z5=wfDt$hJ#7ih0 z9JaH#-g+SSx5=MZ-*ynKTIVhvNp_fXxyH zKw-h?(Qthv=Vzy;?rtAFVCUt`trg=&OyE%KgvH6G9`Z-Vh$b$N zQT{#JizotS)7d3jTac886Oz@6caAY>k!yIxq^UxD>Jy?ITQy%z7NGo0L3d2{re zVyl@I*cJThx|#dt$FLrO;tOMlV|B%&UojC!a)0{eZ{cqZCa`4dyV%IV2Zk^x>hmnb zyVkF)Y8p}<#s>>Z4wpG~IoQu{)Ufhu+;+`kCVz-atE%;8D|7K(Qv%|PMsP~0T!Zq* ztDt|@n`x)vtMUwkokBCw$LyggALxsP)`8ezu38u~>MFy~Uh5HOJ;{kHI*4*< zPPX_l3cGH0e|HV3(iP|T*bla=u|GZLmZNb_A;D_EuS7v3e@T0r;szoy-t>%u-sIV^ z-11K_adZx~;&;tZP66-ViN$ajfAjIE!YUI=o-++=ld&>!i`Uh)2{ zddnXR))=7HqQ+{;bjfbQcixEGlBybfAWd(yfMfU(kO+qm7M`x_2@>mTZqD*(PVtlc zX$@pPFDUZbSqP6K!p0XL>O{v9Y$PH%s8*rXV}v zk-vwz!$rn@${JC>tiKk^fo@g*^fhPPjXbHPcqpltUF7ren%d1p))!ok&~pyRrFlBU zsKx6?E-LC%Tr$2}M_sLvMmpCtsRII34M7%w#2( z>o}Bk%Er>zem*3|T=1AwI`72=R8x|WAuAxG+7(%HCfhwsbMHkFgc>~{1J>usHt8mi zy>?bTHdw@Hvn6*R5<2{tGP5)9&8!B!96?1kp5`J+Az!s4t8K2Pm8*Nr#2*K9zNQk- zM0(#I`c-{T269|4@X8lau~iqmJhHdkahk0%%O}@rwCvu0{oM$2g2CNL^lkFb7ebQM z+sI+jQoTD^X-CP5cv80i0B~7x{22G2KKDCp0i$C1!D6v1a}{>Rk;-O_6v%${knwk5 z#Nl3o`(>HpdeEiaeWunSIGkgQ6G4@{G=8#iR9cF|hJ!~O%NzoKF-iuNK?v@kI6ny? zFxg3}`?0Leh_5svx2v9CB0`oA;xnz0oG2GeS**f>6FuMhN4XXmH$cP7gJO6Vp`D69 z#3D%r|4{WSi4Eu((r!uyh~!z4%j5Py{jJJYA^PIkT0phd*ZK_?W5g^JbKkgX?3B9kK<@_~Mq_-#!<}5EkvFgfIhe*h+DT(tQ#TgD( zY<6%TRpBR_5 zPo|P<3OYooBqN5;m$kwb^@)pazv57E3qxW>J`IOZOTNH?JpLbvbCmyea1E&V|4|Q| z|5p0dT5^tryl6e=T4+MmV3FkQo465%5j-Bg^a%)Kd%T^5Bk~|APCO)QG!Ym>9??DEWL-lixg@HL(i(`Y!=EZWSy4bI1lcNF ze1yU}DJF&oWe!@ga8Hi$YP#rIn&zQ7Fz^63=U`3efAjAmND57qT{KhdqoTV%EvS%rNi@~U*{Ki-?@}%P`!YaS)<@dQ5Kwt# z_1PAD3f;cv5fwT9cm_d+p#-X0YTU4RGlPWe62 z`U(yc5CR5rgWV$Thv)M%pTl)zPuEzg`DP-~Z4>+H5j$lgbw$ zVB&?cbOQy8emPW<;X39@OQis&oTsFuE%HkN~v;rIQfd4!Xr3d?Z;U| z1k?>#;%y6{EAu2H9(+xtd$A!w`0P`WhbDBRLGPIj0P4*s9Fe4BZ`eaVg<7@-J9HxG z*ul_wOJ2Hs^1a+%x_E6rrn?14_BVfBRVfDrQ``8DHvyX#e>5wWb1j)ECdNIepF)jjNymeo zjuDjvbl+h%9{C~8?h!brXhAY%%Nhhcr{LHsek{TI*~$sgZUYauzhQ{KO7OJ6XE*2Cpg2i4S0Ws63qUB-F2fZ zLojzRb9HkuH%|Okg3!21@A@AW3R$8PV^5+eK7L{kn=uqSC)d9ad)flEM zGVvJ81+0?1kvNipmoRH#)53Hd_K5S;^#@+}&&1nRawlG$k^w^nW%D;TPDfK~Q1<9z zq&Q|C#ZJXee!=)be}j$j5rE-4O^QZsrN2dBeoZ%KqvzCaY|6y0cR9}kx0d-Fr z$F7S;WT~202p|b$ewyQm!Rv>&+zinBB^3#iu9cnsNg5luF6d8GT~3R8U1J;W?`Z5w z8vkiKb!^l`K>mISGlegiw4X+cv}K1D%`QXNm#Q5BwN*8rdSX$UQkgbUrn^wK#iGT< zV7&BbZhXueR&p}gqT}EF>U#TW0TFSFM2qh}*A~Ka444JW$5094gctz}mWhV1=j(T7 z8WlAm5Mt7jm$La;7Uqw7o-y9gxVpJj>F5noPUqELa@dscZKsa9xTEF+)>ezD92#hC zC#2!FZzF2Y9Y&~;EHhW7NZD!_r|mMmRqd$xG-fCn$=^&t2&QNp;y6Y;%{|ySI=*L1 zj!JD`0222Vxu3@8JGdLq8yHctH+Mr_tpL}pmo-0xDjC2C!Cr(} zJsd*1W|^byfe&ghs|W$B=4xk@GnHq++PUNnDgJ>sj{&;6l9p=Y2Oz;FhM26Kmyaj0 z<^EuHzE2BJRvwlmJv$9VVpnwgJXXg%hV(DdM?1P08*lqmSNKvW) zq{AqlL4*y!^zHZ)^$h$CTz-^`BII>Cov^TdLf9|u`bMXqpC!2};jps11fFHBC?4xr5O2fozfx?UVz`q#>s}jyFQbJn2doqQUdE5e!k6AkVg6O5_XQUx7E@uu3|ggh>o0RA8kykm?7&%C_m!sT4cXoxEDTnZTN*cA>y#;*N58@PJ*KWrvTesLyINTK|G?K)OFdy00uhQ``MCHxzMwqh%2#>Y9 z4_M+%5q@B}^w%HUQFV|Lnc{B5MIz}ICnAz=Qkr)XQXkO=Tl81#*Ll8OcG8v31!?oQ zKIcx@Z>8885Zg_N68`uA^7#o~$^P6H%F;bl5Mf^PH9wSapTPAGi;>jucI=3&*lcd$ zuiJ{?-pHcjE;~Sw0%_1CP=(|J0=Cuu2FuhM_rsCXJ}x5#XWiD{Jdm>zl{zK;CamBk zA6!lyWIIBe>=`d&?bM!twg?xc57Wo)NCw>{=Wz={miG~vyjd~X4)voh)@qz!KC}n+ zth~#DnK9Aq@os%b0QZTw%?*nFjqii`cuAh705K|#WEMwOis4Fcc&PptfTUZ1b@5lr z2md6?is6S{j`Fk_E3|;Hj*~1uWjeiYDt+u;rfgQ1RQvOzN78I1t#Z!PTGpi62#wt* zk*@u4fRYpy5=WMcjl!3MFL&)XWAH@;8dhSEo1A51g1?e(L!>qRY{_94=3sdeu1Yr) zcd=0cwg|}wD>SvXiXqQ;$x_Fv%$AC%EK(&Q9b;^dOW(+f%g%P94HoDgJKP#26JJ`r|K$bykI-OoZXZ0g#IlK=L1e)3tPI8q0O25IKJOg`sW~Ob>0-WH$9d zbYe(G^Tl58Pt8FYPzL41j?HHeWDkIFEa3h|7!OjLWEf9i>1tUC`?%lD_jX?=shA}L z854;RtLNOfJ0#h@i1_o-`@7gM>@VTWZfBar$)xHqh z#3xOOg{4}Izdx<|+qK3%xvzZCCSX%Q!r;zQ=-6q9!*gaabfe!@*mjI4u`{OTHtaZ7 zBwL6%vzzy`)DpLno{SA@DtdH1_n#AS3q@$->E*GsvAOW5y2M6RT+hnIl6Hdv*vc$N zbc66Tv|TQjd+|9FHS*FM0aX_r(-YCwD@{+sDkq z+LdPmyzaeq4uai(CCyO2hY~=pe3KLQ#TSbk387K^$0<@#tfv`3TsEY8lG_U!n1C%u z;L@~pM7V8TfnJWKZ>a>+mN!NS90SLtNBFu-b~kNoL9F`tu`-?w065s8#0R~)0Gs`> z1`f>Kakj(QbOJIDkIC?%kw!}MAU(e$<-p86vglsJVV7(7#i=zXR%Gs48?)ta`Z9nz zJunAuW1Qxn-Cc<(xS#99pgxAq-+Y!=IKS9-xE6aHGJl#_%vD_};&W79tSfkRIJ2~n zPR2eqT%A|Q#f#HV0iYUjOq=&l6r9wD^ERrRVneCQmxJ{#kov6>jDJk<9kd26IgYQX zF`?F%f=dcv#cP9U{2tn&Pi$|bliu7iRT^^JOT*-DdeK>D1&?H}(t&#nThh&whp@d1 z+zjylfoPKtQ#R*3( zA}JuRguyWD%w5EgP+2BG@oy8L%BM`W75M9Oi~PX$!Ph#F)JI31Om_8t(qTy8-(M=8 zBO-8^5%&R#hZ#_F;)KYn|0*P^zBqChCdg?392@o&RTEM@=(1Qj}wR z12eD*HI0}d+$0*oBdUik?C6^hO15fu_9R>A7Nc>wHrTMFzw7V|uEy^0yTEmbK?&u? z&C41W*Y*6Sg>oLRJK{HqFrJ`VS3mK&yE9spSlz~xTuJM0$TH2R%BZEiMiT_YWX%N| z8J-XUwtxb0!`-$aqESZlX-+c=hJ`ZGn=AjIC+zvY`Tf2#P~qL_?%=&yYL)=*in>9h zx$`Q-0&=#e8|>fc8J`>GE+E2`>Gz9vXPF;{=pHvrBAxQ-%$ea^22Omin=@2uS^(uO zRpGVOUCS2PG)>@%99H)(KGQzHj_^?ddufywsEI`Gm>1S51V@*3dpfUFNrR`EmS zVAZN2#RH~d=i3=N1Eok>rsD1%_lYUao94G>i#~l;#eI#aR_4YG(a|dAZjyn)vU}J3d5CzIz`9q%wV|C$SOvo#;1}>gYHl2 zBZQLLw*ms{#Bb{HQ`X?}4QeVYG2{XMMI+)bix?*DdYYLpyKjm6t|T*g<((^E3IU0V-I81~1OX&CMDH z!EE^chSlGqvWZlA4(*qV%dhVCTU>yQw-HOzR2#?qc81!i3AdH^fUBeRT6>KUgWnL` zTK3J}KbgBY^jMCGngKInAN09L=G|%-R}&k0S_-3Qq;?J!^YL;EEoYcE;#-|IjSi-@ zVYmE<4jUfZ$ir*(Z;)|QzKv@smC`cO3FT5F5Ie1{IVeXX9);dZ?5a@U2LY1(id1mu zuK9|t`G2IV^qwAhWg;@>gN$jYGn|hyYBoMxBOI_-80Mu>$^j{tof*?38)lA;D1%eK zu`fIadP}_!{>}u_SU|SRiA%~B23yd>Ua8E+S%znE&^1ltj3$KrIgca4(81MKfiCZt zcV>OI6({b;k}AfxN)8&OnhDjcr}0^u&Og0?rkZIca(1|g3-%~5<{Q+9&0- zG~#)W6F{W?0Z&CsAj!S3XMG%l6Gra4;{rRTRZ{$M>kIh%>ldm2ed5h~+xK5cgKTgE zl}-?Y=!_&LZAuj+(qhs4B2el{Jd?F&Xt1ec?>=hk{s5>sZ{HuMkP+s zZ7y>Xg2d$LfJcRg=(+v~<0&%jj>vZc2A`X=Pk>6+Px()X5Aq(we_>u<;6E@=o(L3Z zfZ_Q!<3ZA-iE{`DiAPGbiM#&@Oa&Q)#J=zHQ0yE$9RJdo0BwW+>0J8#Pvm4IGM@(@t z9hId||Ic%6986Y&$MaA362C+~f4yJ>*h!n%^YD$YaE#w|=b!FnllbK21xf!4;S&Bd z+);Y`$u8ik4+nLQBM`4&mTf*T6BJ=VeOB^hamCsF)1hJ$oyipXY(r4rb#jKu7KewapwaD&XFnm>#H%3)gZ$Z@erILxA17A%&5t`b3C^j6P+(#TNb`Ys*_M$ax*eiKi1(y(0rY0$5&4* z4km4gR?0{$mADLij5*(paWjIXqwlVu8y+6K?r6?raWi?=B_e@unM6!p8v5{mP z=fpRpuNwa9Mvrg6KsU$PZ9a4h=JjaJ_hgJ40{OW3mp;o8>ix5NTG>v%7COqZ>!X zj_$3wK;A28>Nej-^`+TovbH<_QL%B=28$48Di<+T2s|W)KhoP+@3>W4ci&bd!y&b= zqj4uB16lTkHCOxYOIUg|7($JMmjPDvmRo@CkwEiUNQ&4bQHjIA>zpQ4?T~<&8$s+K z>RznFJ1)A>iO-o!Y`s%>f@kHkE+BNs6qiuPA;_qYHNm8@5cmJxH1WGLjT>+MYgw>c& zYE=-)J#6ZDj!unoVxkRmF>d4wU9I1D=SfBH5W;Kfluds6X%n){h(GUDYZew!~;< zlzSM>f+C9XM}gXF=9m8cC4jk{aORDU^=aeqQ94LQ`69mj-O7R~-ptYVz44Gz1scvQ z^KZM;^sUJn0*5V_PsG8os3?4mboTxSbM&@`!Jz`I0Rii&=)h3@?3*Quj*h`A2ecf0-dJ%*Wxzzg8!qk z#{bv)v`0u3EP+d;miP~TE+)QdApH+C8)uY2P>IVw{`VcbbZ26o5>+FNW*amvHyIn5 zqlpcIpdbRPyt#v=n-v*5A0JsNn;n~S`;>cKkkl!?gkE-{5g(|URC z-Qv1X(X#%zIn`Cde1jpJCGnh@aG)~u}E<5g=H_V?FH6Vf!l5TdOzMBYmF*J(wCpII-jF>+W4K_{lj(=mJ;T`XKgS_ei+n z|NiSgF}g$;rMkpW6M0BZcK&}S(h?Np21U_AuiqJuj*jNsdmGy$bZ_$1AxMhC5{m3m zfNK`TyS1Jki8gxZV$v2CJ6(f<>A{0Ad{gkfV-SsfT@oR`8xr`?4@Ii)Y=cf|SBOoVmRu(Hd9YFs9ge`po{`J$< zGVtjVNXM6S@!R2VySzQbL1WGR+iLA0?x9v-fPj`kZ%>_Mix%O`h_=IS$&|$1wNTnR z=W+StoEIn`o)v13iTI<^AI`Md|5;-U%HB7Zc@ow|D<@vYfKBgDxKiw|R5^LBfxz9W zT`MVqQwqF0{E}#VYm!U}krU8z!0+Td1I!jbp2V1y!n!-PdSLBY%dN5C^0cx7?CWyU zqK|Z2um8^@B8QCf>;671OCe{J!k`hRV)v_?tdu0AvY-#^u)O2#V1 z!$-!>$^L%_ylnpr62Y-l8&wQM!NFM?6ATL=pt!MJIsdg(x88p(^-aG7ZOzW{*^$aN*rMNBVCqCiYPAc&Vxi zWlQ3Ga1;=(WXC#1k?5oT99I4g5Em;W%*23mmsk?-14k9i0v8nnQ~nM?$Iw5Eine=< z`<3}eR2D2rG=~jNG!pj)3_Pr3zmb7L-!G2M7j{V@r}qnDt)B=iudR*lnbk9(3lR$* z^7k$nGuj0BnM)!%k_!wTSsxqS#BnUzg(?bNQY92=ZfYtn>)3EMZeUX?U~d=3t6v!f z%&+X9&;)!4n12e=989|?Aax{i7JL4<7Q#(G|E6~=1AyKJ-kS*tb_el=dDZ$nE6Vo5YKevy<>UP6(b6iPz8pV#ks?;X^p=oE~*d3~KojD%AEt^+ePm^=%I z;(8xY!3a9qOuPB22TG`b8T&hftDvKUEg^$F!U)`e%I!z5dDq)Z>Di?zN(&ovV4i8e?{H(GJtv=i~jCid>~qWMc#-w#UV}sk9qf{I!MYg z+2_PJksv>M%W39%Ewm630=?Tlcq?KcwxWIuZf?JdL1K^ILyuM0D2iAZl8iTIG6ww#@BHs#A^uHuNG;vHF%#!m`h~@ zw)foNpJ@QFFVZ;XsZUa*G1d>#n^nrMJ-83d_eC|RzGI(8@7=F#6uKwYT_xQ4ow2Vm zD6B8uyRgJHCAjbauK*Raw9ZP-v18H`n#9hq$jRF3Psf|sl^~%w-awV!1Hj^Q?`$s) z0oEBZ^uCf3Szp3l4EG+hNX5ZvmzQQ(kw|?_MiIdE=0$;0PiW?kvns!#c^C_s(t+B$ zfQ&_IGBEbg0*@8cLU8U>kdbr3o%{Y%TV#dQzZrhU?hF5>(dPi!wRk7VHRGcgFU9sw{x*KDjtc?ydhUti9iGJr2^_Xry z!v{d--rQN(F61uE zALXNYXY<)nG8Vc7e=PdxJpZd7Pc10>yt$(4HQX3(Vw^HlAx$<+85it>$7)8ee6BXG zqdH($37_iCD5`HuD;bIYq-s*B*)Py)&>h7dzaTHC9YI7gWu}v|dEe zb`urHs6N+KN=Ko_rC3KBCi}N1?1@t@1iK4XSv6=NpK!)fiplwcd%Hd4 zcGWa`)v7vBsFk0+WA!st(MsxjtFF9acIV;}PlZLzzS`wCrcsV^e1L&rYpg^L(PB7x4*UkxAD-69i;Y_cNBbAoEN^1)Bm&?tE5w+o_(RAU~Rd6|!C2 zXb-EhFWP6%uh z_C_{MEqBJRLz};4#Hv&*M=cZlu%Z@NkHAJggHnJlh?&OHvr!>n&-u`Sj0$ZX#xCeifDjFgDFwVpvWI4Y`#uGgpczS2ws-#V-f zdu$v>#n}6qFtYY%JLNcxdpJB-bR8oBt0dD;2kEazI(1{t##Y*SLoz#6HTZF>yB@7) za<`Ux*D~qXv=1ah;MFq8&j4Pk(3YZX=283>PAPPZ+MlbE_Bae8;ne|t{_UnYbdJe9 zzr4k1Ybc$k*90#9ajU4*^-6L#S9x^`3Rugi!WfG;-f4#&-! z{*&|EX`g%FVxCgLIv|(d$4hyXC`8kI;_O1Raa{MP6&H+DZ^sUQ9XqqtzIn0xIeX(~ zGM8F3fld)ly(N<*sF&^u>&Y1VF^N+>FL%A#P&L=5Crk_PYMqc_O-T>P9dfRWm63$` zJ+2AfW(o+X4)MOIRT2uEn9I(x;gzCNDETB~1oXdbRparQ1Nz?m|MF{@AbxfJ5L1=g z`KinE>t?Dt%c-CJy+7dG>wd z4%^82I6o-n}z9d5MVfPAjJ`~Y!w z!4(@-{4cHM9)o^!mZT!5v2YLGX5v)^I%QR^UQKwiD0E|rNQay5*Uus&o7-#Jvr z!F(a^G-CeUI>syt+u;c|N#Gu>+ugDKHLWMIqk|jH--~K`Kg$q?fR>u;R;{G7(5g(T5?#(pzvP&Q)9{3FWmcCu zJMC_q0yBQyUkv?PJ5!Z;-#Tf=){bh`Y^hb(W);Cl>yj|7A(x(@Z>6+j0o{Yr?`esB zhKmFPuK!mqmro(6{W$R*Cr^;60!60E9?(J?8GA8tkkY^*SlW{`s-JGq<3cs3Syqq4 zSg>~d=nW~rgMRUC6!R%E#@90zKUzP*!F1tapn<8MCKnz?@TRk@oCmQsXy-}fJ0mJD z&+gMNYL1QjkXFA3}NjzJ;jmLRgw)=gn6%Atovuqa0Uw4m!2}4DRl97|SyJzN@#FBm@^{AwQUspI$ zSY#C~V&q%Fjz-FL0!tN1f>&)ZC(IdW%>1Kn=htfp^D|L5qE$vCdnk>!^7oH&9r@%# zQDav*x3n)t!q7lt9fS()q~pMeH-KwxIe8J-k83!geacYy9_$=*B;r{c<0z5^6#m@2 zgf(U70CUXAvB&E1dC$_Y)-Fg&(A-nn%{K1KlU#Z}F(rnVN{b(Yhl1zRBLNPiE9|#i zt$Kf^)!3(2GW7s!C!5AASciGc)K?uqm!m+oZCFxAXV7=l?jNuXi%LEVs{lVu&flcJ zK}@X%L;*^D18FxUw{ZM_?HaJ2bnuBcPI&3}OdA@jLQRL0gPF=e zt(=UsP$*Wban%iGX|13{E5DaH9t%|7=q(yFnjVW?*1!R9`E&eGb$~*2^ZKJ#+r)S= z{65r34Ze(?1WVWtK}IjR>wC4rzo(jI$%SVle;I}nN6IT6Us5su$(pyl=5Us(XHLcN zZM2}_9K{1q7b6al{>&4Mj;j*b=tvXjcvjg6BcB0joSMGN|5#fTxt21yMgbf!J$%h? z|NWC{ohyz7RAlgRSc|L=+;kC~qZK-$oB`Fzj>@l1Ky03UIumzKilZ6<6- z6P*14^khMFu%Xc6l&fF8G?*o1?CP6Ynec2WW!dh*N6?hC%ooTedRO0?2a2}n#M9eb zGsuplXR`$&Edn|r*wRm?{SkgV5vAs=?dS$ug>!a+pn3;^ReF$Rww;*)Io)`son&~1 zE%|c?(cXlz;$9T6OP+JiLvQF}A%xC0w?cl({je93Syyt&5}N|MaAbiitAMv69{S9s zMlsE>RGXv-Lg~GZTQUBJtINk^+@!8UMKQlKB?jDnaln%*CgJoFV=gvNrs8x9)A~F% z9sUGGFtNC69CCcCbu!}JAU(O}fskQwHv7q9-Pw!=irde@V!hqY9rYm4276!6Y9P>B*lJPiYxcKLL@1Y-s>2+ z66{JiJ^)xj7+ml5qkGqM*Tws&oS6s5&TN=BM1^2u_Lv~Csa0EP`)*9l)cIM@q z`%KFPLy5`niXuzIWdU)GE}`$rkS6*~1d}G^XBlP)di=epppd($Rxjj&@V85|zi5t1 zT!=oo;ev;m@9Q}W%Mvw(n()a#8KMR*1PDyX;{i^a2E(6pOjYoSt zB(+QE*5z-(DK0cXseFRB?>xng#Gq6HNlWy|X`A(qH3zfU1ZmQ> zSb!0tAJRHdo9RZtJmx*3CfYZE^V!JJBEoZj7Y%XW$~jmC+t~lF#FY*xO-pDrFNCpN zDJyuvd8|wka{$}mYiNeD_|Q+qGcR(x(MBMY#d8_5ntTI2M!`=lu-GA8mK1!9(t3*5 zSfXiKs#Oh@EmzF5rG%x2yC7V0!dcdN0H~ z?U!%aIt2tU@^jllDnWk9UsOX1iF(42I<{j3S%F`eZ<5FE5P zG)-g(_ng1P&KNhR(@v1cEh|lr?32DkbHZ321jU@3VAO*o{kM^3;;#;o%7m%J65H0Y z&b(j zz^EawU#a)|!7Z4TacVC`tnloMzQP9~Dw~K8x450+L_#RvY7?8Cn#i_(Q$^d3i)mp7 z>xt&?f9nZ5jy3H3U19Z-?#X^Q*)nnd9+N*7RUB$|421-JxxxNACRm>}-wt3{tPPYV zz10p|BLOPZ4Z1lHEb8!4;^hut-P`I1D-y|^A9Fg918?mDHZFOLtGY_9agZ&oFBF9L z<<-MeS*v>F9gXX3d+f--5D*mwe577U_=kRV3G|Tn+U7(cKfo^>je3JAwH&2>y9i09ch_iZsM(xk1T)wQv2x)@5?I}}7 z?A^MF0G;E&YAW5D@?c@lWTF3tkW?{2!Jh?dfE)Qz8(87p5r#($ZZfX-53x@YmEFDZevZl)L?H>;Y7=VmV70)5Rc=F$$ z{8|wagY}%%g6b(#NCL1m=})i{qC8Oje|gCdO3hOfmre};LmW-)bvVLW4)AUHG z^_8@&V9l9bCHy2*vmlqr%q{pqV_eWTb)gbUr}78#N&xWi6wV>HKik*wPt^3%F7UXB zQuacE=E&d|7M;c~Oi=`1q4^d^|C|xjDJ{&~bK;`nAPf2W2aX@`8LXzb@-$*Y46M|Q z65ToXv$X*mwyURyixsLC60NY7+%o;-(-UOlhYaHM2T6h0G|n&<@Shh@e>O^Z;6W<9 z);>jE`NIGY2}}A{-IZ+ihmR3$>)AZ=97j7sSq7@7mW7$7>lyc}nmupi7g-uv>x(Wg zs}K5oJIg7S6xpF0Mj93_=$m4>LvdzTA`GkRgIwn0<)u)ch7OZ#Ei7~XS-*Bdl1 z^liO|TYVYXyy0hO4{DR?&**Se*7!SqI_#V5lL7$evy7z`C%<#Ev}+kp%MLK1sSMaM zue`PTt>OZrb0998`VJz34^iv8ZqZJfoz*>5tWCEfS09u{i_as6;nLxW+4_z&y@fF4 z!SJZ4cAhVA-obB$l0ELQL1EVxzSQ^q{{T5a#=kjH%|1yjT&O6EP%4l%2tLM&Cz-40 zFFwAeW}&I*fB98?J;5n*H;mW2Z94{GNpp`_8=j{);o{Gx$@W!~!!4E#Z55;S>--Uz zG2VBfQq-o5Es*BBwMrnb* zJHIDDdkDVf*_G`8(>jq6FX@V8v;>LSFkTM}LG{pBe{x?LA$j1#-%8hA64l3kR!_Jp6dS*Zs|78n1Vp2S1)Z zOYVHXe`lKc#NW*5YO{UTpS#wMSl(pBf7p1 zU0!u(aDSy-*t;qS`jE<*aizPN<(a}+c;u``m_+$LUFBEn~?hQO0 ze|ss#6qB&iu)NqOog%pyMq>~N&Kc5?UzvNjP;suR^Y&mcJ-`_AWs@{1nvQB4z+)g^03rq;fdeY z>SfWD=fFjaK=<>b*#rlHPv5$PpA6%D=AC`AnGm6;i1t!qxQU3@1#uq)92YbSE9yQ9My<98;0PMyHSu~vH?1j#v8KR#NsEqm+-ZwCm@C4%unjNf2D%B z*$QXf)TPB9>1?J0y#1Cl&th}nU zu#m?)&^lgbNHmlt&^XT6+eviT!D+HpQo?C_iY%)rnZZJ5-|hGR>i! ziqopGalFtqCEvAiEjOBK>DqPaczO3L>F%Z!MeiiD0+9&*gvCq+)B8)bf1Q&gGNXgy zLsP4yRjFa)`u0E)L5OMcdc~mFX;@ENP?)U+9Y`_l!=v{|Qq+yTO(ZX^_s z0$&ibXY6KvQ{xH6eoq=ZfB#AjrsO~PVQhM8(@wI=l$ZTOCkw7qyX%2NHMYtJv!bRQ zg6ft+_TaO~sh9a#L@6>om!>R{!3tyJeARQ_6U*q(T32U+d(|b}S?AQ7JhQjwDNhrs z*9jz;$jW@@n4f1PV{rM}+S$*O%; z$MfLOe&(pNTV8S`XW?kVd(OTOZs%O~f;u`Jf>rh5Yd=<8VTO+5r%Qyv8%EV(s7C_` z&9{m`_G&=}l{~!mo$SIR> zB{%r$8^{LY#~hC~e=f<|{9+@)d~Bi&nw)B?&5^&r%?9ANA=#xKCumw04sZJ0>cGl* zbKsGwia(XXB5E94)Xx0I)B~1L?`mKZg%CWn*+~|z>0}73 zSBb+v25*m1|AEyysEib*zL2R_$emARqVK8AD+into(ki~e{@AyV>krG!j%Tq#=r|a z2JMXco$k8PIcrv!2?Tgw)|Jv)Tk4KemmVao@Op6B7c}B%xs>AOH`CP*^E3ui)rb8~ndcerzsSm0l=Zmv7 zv0+;7MqJ!xAZxCQ*XKiHh7=Xd5hlxlgY5$8K1{7Sx1a@8Optv0bap-Xj{;7y`SYzR zJ_81Q&O0B-++FpzDi1lQ2}nO;(^pzD_kN8EL9Il^o@B$zOZayxz|7C zqcL<%f~Y3vTr=-lCOBAfWCg_g`Xt`L%h;>o5WE9!zVI*4_|WtJ0DcehZ<8^m6qCPb z6|=&JgC3KuhZVD0ral{!MWz+At-<;nlbFF3v%KCg9FwQs6|({Rz#^0E`W3g<9RVC3 zm%$wY6}JaG0bU`O@jC$)w|iXyavPUsT>%!i19<_@8<*{P0T#FDlmX@lm(r906u07? z0oGZU&Yb}jm(BM94FoeXHaC|+^a2$GGdMCemr*7ID1Ul$aAxn)ZIX#?5Odaga ztp10S3&Xz%*t-LrUH>6w?(AR(kQY}ImQYks1AmB%GOCFJjP1<;a#H`4+q=5({DTIX zI@|qEcXR;P|5~=j|Fum1YyCIt?D=nt0hXB=U}k0N3NQg$SlPoe{*P|b_T~-%j(>;E z+#LUt`iGFqKl}hx|7b`JFaw(ZgLbpERWP;#0;ohC>>S-(fzAMV2Q#3vJwV05&e;Ax zD1T!+D_gJsFYf;$q+N~w(LvbW;veOh{vEP%k+AXvnkiYi{v)-yvF$$+{VT2s{P$es zfo4{2cK@yi{-eh~6E$yP<8!B zOJirV{|5e*Rx-Bw_xAjU%zw5C!2JJwd4FS9XDd&DE)&E5tOwKoJpZxu{~t<3#KF^t zo|%gaK+np;1Yl-kV+U}vvH1Njx~6W<&Om$De+m9iFaPoXwMamqC(smjeZ|3)H`F?# zHN4VaykxElf_Cn=kvi=PGwA9}@3k(9)v%?R;n5}Tof>YT2Ry-SkO}|Z#jf;S3xA1m z0ucUWYC4uUQhx>arHr)7!oP0&O}Xf`A=2OqTm#3>Q~~@t>ee-DV?9`jW|}Q0|GZAd z=_fymBy+4$Mdttn@=oH~;)A^;wLT?kw$2wk@PM`!dCoGd_{v1^Wq%zkqA+9+{+b-e z(6YEP$nhvb`<|VuoqKnwgZ@T}EPo}89u}aw3gXPj#W^n1PJ$_ulw&SfCKSdz4Juss)LZq9i0v&UHvP8`g`hP%OmWqQm zt1bUSgbIhGLjt_2^YOW|{b0S6J%Q_vi{VvGHY^V&?oRZF+_2m4oj$~ESfdFjRG%hB zzp!!|u`@U!jrHcnZysq+BdSCF9prBM!h7s*24m)$<&g}=r5}pa-=9KEj-Y(@^q~Z$ z;pQvA!s&K?uXAS+R6f;q+<#z7ES-ynSd{Fp_te;n3ILs@ri&kJQc>Q4xgW%7k(E?-cn8YUD5k2YkzVRvNhdg%784A zh^2#-F=T7p>_6j@v$>yP#b&&}DLryx`swRvgZKyjFva8lA_2~Kp z5hzQGL`&C{$8us+CkR3lO9;T=vL#p;;N5q(m z&y-c%VIXlTo$lsWvQJwGPMhSU)g6)W)J_ps$VlANqT&aEQB1GFK4A8HRTs)V^b#v~ z%wmIju4pO(8zBv&zkaD?sVb1kA1R{k8HCWcUOAJU*qI2gqJQmSh}gaz{HE!q#m07v zPcTP4JDA^s_gxZ8B@CMA-gh)LPGy%=9pp3pQ4{KSg#;w(O}idH4WnhgRUS<26>6u9 zuiKiwT>;~4}8C~J`Fd^`gwInQkA1cop?*_r`T)`u!`@u65U#UR;G7;CHmQO}RS`ik%fq(4tIXCx%TEOkKwpoa*e>B2^ z$e)PlZy^I6(2GPeB{14n1d+jT7ngcKi6^~cXm%ZspC{rTdq|nL4YwmUoDGhLhhwFr z*`H-g6*2DiJOPoG%8^O&Efgn<;c<#ROBlniO%T<4fJ86wUd)Y=l)3ix_^WSPWAHQU-m@*W z31gac_XhK2LBI0u#c@fN_gGZs!w6bi-l|F!UVnIw%ICvIwEHN-KD<6V_4s0&0(=R> zvGIDBX_^vp1NGmsHk}g$h<}>|`Fo&5^9C!W>eQ)jhRHNV?e6PJ|E#VT!(9XBBl;J#=`q z7=LnJ2q_~FYrRBcLg?}STY1cm`xG}JEMHsM1CgGY7xLU9&=E#&IQct1cB1L!3JDgK z-Nd1rdGrT*=@ccGaj8e4`c0?y;1fYqRWYRxEdo-m^kTIJawd>q@Gz=t(bMp4&<1#F z*$X_Y-?K|yud0iU3f6}nZ1VsG24x|20Mp(~86A3&p4%cyr z4$k;2EF8}$BAo3mmSV(rp*(_x{X8rjSg{09+QOQgjM1OOw^h6Mb0=QCg8aP!&ai6> z4J@Yce3#+e;D2YbbC3d_#g7^obkaI4CM@da9gytWlI^unY71Z74Y1{1 zLbl%}ayWt9OAFGW9^+&TSATOjYG8EKa%5wVrX=!Zr`{v{>x)L}x4-}Tb%b4hEQJj) zOz(1ZU3vKEVQS#Gh8&cPHcaQE+;x`DZ%6JJT*JtBqAE`zqfP?3vGk3n(jOoH07sH` zC{inRt{$+??+G!++@`gBG#L=c9X- zgoMo=J~q&UgB^8ZEeUx$*=3BU9i53Dq*4 zsiR@2NnJTOpOrbYP*KU~1|9*``I}`v+3oDd!*r?G+h>>@YYOg7GDKNghQu;RJzwPG z1Laa;!!67n3w|Ca`E$jTQ)6lKxw^D7Tu4`!uXVdVZd`1LZK&iRA`&9_wW2XuejQ9ASk3o>o z7S!iv@U3G((0^vR80@l7Y9-Y%%R)}sA0w^`s$7T|6zNMqk`D^y9#eh4y?G8yoI0V~ zL`h=#+a(7IJs3`wB|G^XG4Q)|?Ihh=%sO5nkKHJ@g!B!SOP5@Yg(e@-8Je>#j0`{X zbuZ0>JINTesTDk27{)PiyNjbiL2b}A{LxqmI0J%|bR{C~RBk)JI>i}(wnYQ{k!-n~Xb4%(8&}WfII~o0f|8+-!Tc{_ z;noD~4x7+8z0^vHz}D-Wgza~_CJ7HLI}Rg{ZgBGXxJ^OFBbVnU;R*q3NOfO$8%he`^&2*gW%+Hwv|1lcj3Asp!JB&sT7h@js76@P!U&B&i7#Ze}Z+iTYKXqfu2%C^_C zX|3a|M-kfEkWpIe$Q~jNIk<7uA6q}DCf2b+I0vCTUGP9APCY_m?cw}@?O%@;A>-PTsO}ztYz7sX07k}ev||Q+Aq6)S z2Y)=41nJ>N?$CB*-Cc&nl@5rQYB00oCn_|G@Vaz;P9da;zp=San9`rExjTYk>#;{%!79)-NL);`=-=ZT-H5!%PkRDk>qd)a(1R2tcb>5FT7SJ0 zTp6{m3OF@OjQMSJTR);vB6F0mN{1y_AsEWmgBUL+%`i1M}^{pXXV;6`WPTR|BHr)1?j_~faEHr;AZd`faZ^u!zNC4Fhb~N6CaJ48Xr+h zj$Ni0gg3gnBHK6#btv>`0lWX*ev-P?__pm%nuTT|v7?V8AW({an(;qK|}aH$DE(_(7g7%aXCQiujayjy;OJi_flK(QI8) z<^A~#2|cp1wNxzIkg;d8O7>#0YUU?Wz1TZ(Rfn_wBVkkP2&=u|lYiuDxFW=|UGPTY z7HIZnP&y$ZDDaL?Z=JhAd_Af98|vb493OMnPxE9o1~`RcM%iw^^6cSC7=pO}t=?vx|ZPfSwq}MI9ExsM8D?WFLTw0*NqY( zJnLNL^a1nz#-dmlp&#?ErF2RycccBRW@FQ}Lo-RaYC9Wtk$(@R9=XbrQi28&je-nc z)3bC44_Q|17HuSSvuAgdEnVsAA$K|!;CMS!TT{*-LG-}gdX!>|_}uxzv>tYcwg%RG z&}T-+UUF1yp7aq9aJc~?mr!4!(wHl%Ixr{zEPREbI_I~$!ClH*@{pa~eqM5dm`;fCPA&fRVz2kq1%gz5R(~oeL{bLBMnqRp=X#ZhA$OSr zNTvtck!iuB&u>WG>4i`!gonVE>D=rRoUA}BiYY^g3q2YjpP2;%CA@?NgspOnq0dCV z2j>ygocchvKj94N&M>0cwwD7POTS#4c7SaZb8MQXZdj3TTh{2Zm{Y0~vf|G6Ooug{ zmsV4rynoMr7O*l@w{QM8*wjuzs=DijhG`C*3%0**=S_}heGBr6#a7rH$;5V_bQ(8B z5s4~k3uCU9!_rM4$mo3rm<(uC1tzs%%9E>2SzXzkI%Ta_KQjxFc0a0Ln5d2kdD`zS zX?3Xrg=sj<0@RTSlIx(-9=*(SWbB&+7sFfGw}0v_X!`)lO#y}WMa^loI~VS}^%(_B zK1_QwlAU?Fs>6J$T%1`E9y~?&c{sFKn59Korb)UEWAL^Otxd@6Kl}+o!P;{1>xI`3 zy;zo9RD^ofk6s^1$dLhwRc}8*dI+4-13?ow3WOY($}ns~#BQf|LQAb1rvqTD_)?nG z!hejQ8DFFkLgb6wN*{&4bAh&}?uOoCkTy%~i07N9x}SG2c-gZ3tpui@0$GF>XQ0t` zib5oH#RFY#weEr-mc0qJ%}=q?62o$0RXAD~X0IS^d9ieYryb&@fesSJ^_BeQ)%bH% zOHQ_s@Gp$gGA#X5S@ln8t+F?6goS3onSY-g|J=?ep)|NqWF8PgCr2ehX9j{i2wxy^ z>jcn&b&j?y&jp|`1$8rvdUS^v7@WXXS(e5ph8=ouUp%fM>qVSEO+SY?c1Qp z+E#108Y{ya*Edxa%zA=Mz?)>2xBKL1uGN9533RL&98czybjnpi{#&Br>NmYg1#W`g ztu6Uv5nxwX_We{#N><#L!36kmxPQ7#Hio&;V4=RFjA0BCG%3SB9^!F8lm$V3DBkNW zR+eK=yqP%k=IAEEf;s8~W1xiIc3~{utEh;!Z!CFeCPqj+d9l^rM zP^x;U5AlEGRXoFc@6Bd}4HEL`^pz&nU(KI1u)dO`>EmMcHg&CNp6Q@9NUfOmyQLuq zmG;D?kOLi=8pWh*&>vEcv43)Q?tBJU^8_;BAv#AtOfaG~y~+a_=%)65`bd9tstlk| zhI3|+V$Y}kKo_R<=0cwXwL2T>#r3yo4)*yNGo7{1gh5>vvaxLDbZ4>HyvAHRh3$aS z9kL^`C0N*hA;Cl)6d?y!DFo+ABiOCri`whUs*dIqJUU!!314Pd`hQ&%r!m$YAFag3 zBD=FGg@rkvv^q$%J?r0I{eC#ctEcie?914R@xkx9O+?;;Or?}RR#^Oj51P3IHISbc zfl;+{upw)w48^V=R+TrO6+agm08Qzgyl2ZmR?W#wAe|yfk^X z$4Qm4F0gPC1B5eqD}Ox6UaCQ@&!r@uT@VsZZ;-KiD*H{Igo=S|oL2!Dq{txfBDjqc z)xcRZtL$J7ljxbF=a>Ph&2I|CL^gTlqMBNrwz-P(Cb8`{S8-Tsc6Ev}{hKB@BJ$0$ zOBRkKH>!7+zs-&?&9}j`!j9PSq%+)8h*VUPw0TDc!M<|1c7J98_w#&!UC%p?p`nFa zxON7&7`SeO_IFL`;p=Vi7g3{!aCAW5U?Tlyrs#eTgoUa9{M`*!*&JfIdmENemCPEE zfoBb3aNDDMJz1f%_)LTlXfBjYGr6G&PNf|bp|xb7+ky5Oa_hiI{jBU{Sq1Uzr~KrC z;F?`rYMzVeLVw&JCh0Ly5Qbr^R*F+Rw^ZWU<{YB(Oier#z*~FP+z+WlLq6ZnekWyT zLB2zlgGGYA3~t?1!qC{Uwg$!nssjtC#a>#gN_~lY`hEZBM-A6bH%FrBi+g_22VVyj zfhxdGkY`*1vTYA*`I2FmHdT&Y4D!ze6VxKSWmK3lgnvOEl~7Uo=RI~`%g!`3gm*Cw zRI6(aa8VqGnaHK3BF2*_GffXB8W4(iHr+yx8#|=%TIbe)P;GdcbY=dlqUPlJNmWk- zUiRgS0XCuZ(&EXlc@9fb-%ga0JgQ?GR-@}Dvy?Jx#noQakmRk(!Rc`L0ema!39f0x z{k26*_kTkyE7Stfxj!Ey0fJ2ki*zJjT#pj<MMu2~m%`*n)UkNi7JQw^5_ zTtX4Lb4ro`ZydUZS~AzXan{j%~VXw-0E;>j`vyZK-* z+e19cDHReTl^9rl`UekrSY&PdZwb$gj~k~5I73c6Uxc4K)S&06WPK;Q+0e+%-gO4< zrGH+?39e3G)HT+f_IFh3uy2Dw7r!pTkk1HD8PD&P0(cRjq2N14R??B4POwZd&gr1v zy~}sVtT^d0>6zK{;R}sfRX?9QK@N8oc7nZJ>ewsXadn&7li6TUEDRc}jye9gbUL)N zq#q_(FqM&&(ACZ%ws;adj~xjF9{Do7W`E9)X2=e4YXo=Mx~p7dp=X&6=m!;Ahq%e?Bh=lIBsRhUU-xu#9rU9b2=$Z;kgY!~NUtBiQ z7qbW|A?T%C+qUZ6gKW5iApE6ghcsTA`{k3T*rsKA_jM4*={eV<@dppw%`sGesed&q zf1kM&hK%;fmSa>+dOzr-Mgf;eMK9=fslrw+g0Ka;y9?TC3Sw|J&1pt6TVX=4X<2IG zGde`9f+Um7N;t1>7w7r@;)_BJ=p9Z0$tToW_*tC~@lQ1>)I~eB$&bYWlFH1ArWNrN zNHAYpO|<&OX~kpX<&8i0cz<-Ar4h9aJY}_tdg1zsc3dT+X%20b$R2X=q3#h# z7P)_+k3rQA4`*^3Bp;>)H>7MtILpRUkhi|QYsW!N{b=s=N0h`s#@oj-s8fTjQ>XSh zvq(_uly?(DD+lhSiPZzgPZVTUNRy?s4i=;RWOwjJHq(yGspdBtW%XRLh<}kcMks>S zHwk9PG9JXXiJ;EbQP_IqKV(Q`G@+B}ePW2a+LAfh(ij$fr;S#E6`R&;jEK|`qAAXK zyd|5{o)d!l1ZTx}b|ZLLfzDPi-RGaz8Y|pM!f~{2h{}Sg zajl9-iAH;^vV2-tlp6}P;(v0&yJ9tOYub9Bfg^n2wh60KyrN;Km&4|D|JAbYRabHQ-77R_l9STku}#5bToeYfQ~s(-O;R<7?qp|`w& z^5LIwxvd&2Cx}T%%Ix<3Qrbns<#n60zWU>yE-V&SIw+BjZGF`46_w(l`W$Qh+$0ny zuQu8PfoBd|a{H=QC&!Q${}`EK>XK0>IfEx(Qf-_|l%*Tl2qvRm3O&(7u=*%}XX}~n z%30KZeEm^wHT+21bbq-tSZi6d!-dK5E(tmG;b7{rl)2*y&bwoump$aC1p(}h4R{$` zX1mn=acaytw&NBZu39AkOnCNj14fxgY*R^8M%3uMfo_EM?~#X?FZ~_2o7TMf=9PqJ zga%lAPb5-?Qi9Otj|u(N^gPUU#UEmz4)zF;IHKUa)dkM2#7IK}yK5{73*EH!azdJz2`#pgX zP-0J`6^el?q)#Aa1AD8@&ks!Q4^S@vwL*XQ_O|H6DFg7JK|y(=QS;u-7&}fw zUVp9MdBq5V3x5+07m*_2D;Py?PT{Zl-!a@YVMN65D`Q+$;N@Siah0R2bv%iTW-DRc z!sPf}a6L5Kkcb5T^2p2>azr7?U{TkCYmrU=b#f0P`e1mQ&YAryEaMV0D)$ce z=xy1q_hO}Tw-W*;MwHG=cpAZtXWoK=)?>Jp?kH-{`+vNcw1-0ldKDwJYv~R&bl*t3 z*38fX1-<@#>W{udrRGdUW&yW)xPm0~@$h!B&Kl|YdxbsLJZNVLIOC&$-Fz*)#P^Q_ z+t`9w8-MaN$b((pu)BPv^iVPC9$o_4kcy=GZkwjTdf(;QA)tDYJ z1p!?UCY(9eL+e;RK+A9Y0#{r8t6#m`ZWJigk16@D_+wx0_@iF5U zVO);iLPdga^Yb?WHQgt4N#R-N03LOCCA&cS+oorTw&7;|$K{dSSZ8tdYIsEy0wI+@ zh<}=&v1<|oIouB%6nWVt=aPb2lbpIJ z2Pqk@jn^&y=?9AsQ$UzV1VK@Rt^E4-7@NrMVC!J3#lDPj(mA$vfL8O-&zFA4;Epa~ z+3n*F{aJ63cgFp6Lz(?$!Dsk5uj=-3=zl`cUrMqu*mS+rk<}D)sD@f=OCamSE@cWS zWpmrZYQ8X%y>&tq0>Bnz(i5;ivm-qB1670X1JO+U-vyPNfNeh>o%5~~%C&lI-*KO} z+K-=GgbWcoVyw`A(UQ9ON9fnE0gpeF@g~_}ZWHTVJB}bi#+7;&DRo1m^?=YkK!37* zjbIo>&MB%Yn3&;Ck{jww-QK<$%ObFxl`-vvgJMk&*Y4l+=TE9@)7T28yzcz)B(?;v z9M6&4kNsYHU^H%?)>-Y@7u8BGV@LNmg>3%A46`1Xg}bc}jwA(A-UGKFkU>eb7~AGkB|%{6U^IQ&^oYtUxo%$I zQIJm|?(@Z)TR5>$?RGH}UfKx&AM@Ty@MY>f2oQf(SYE9z#P#MWOl^kpeGAyGR*X?n zRivB;p&bYyz#RqTNlroNK3Cex&F&9$1zrKwgNy;;ri#toVe?z~srwuaI zUi~fy%MV3~DnF!K6ECZpF2{V#L=(ShY+m!o1U#ojhLOALqcF;mJh~&%BRRz~hfRQkwiS zRKdf{4n|~E3X0&r)OP&UJwV?}_bCtKj1KMHkNLm9VnWZ5uC*Y)WwAEQ{ajq~@IztXY7OZ?9Ag!qpon zN>%bB`)Ym}8AKJcG^+AYCYzx}&F``?Hg`A{l|0Y8ufNl{yrs+hYHt^`yGl{k>531k zj@(e}-@aQ9CLZgs-Fdp!h7K|W9dOK7Iva`Vx}AE7VnSLwiGL`PdzBIr1LO4nNQN6R zT_uBkFw{jBl?MQxZhtar~ zn6#~50>yq_Gxfy|KSM_wj8S10^SHQV9>`X*o#iE4yTZBFc!PN7?GQY7&ne^)`1R5~ z3JfWui4s4MfE$OGUE=Wf{H^y`9S=3I*;&g<91i2IPk)vj?K3Eco~dvy>Zf3a4-Zk) zw!)V$-5Z*AhgxKEwBI8B(BWkXK`HC9=YQ zeLP0;&tM;JXnV(e1T(SVZCUEr9cJniW`2f~{Nj%liY(oQI$@T~Br1#+ilk;RPrwi9 zD3-={fqzFDawS)l%TD-Q#I3Id#l#JPcc5Z{a6Wi9S2o?bLtA(b`YHDs z$+8?}jgt7MJP;gtK}j$URgApX-|Er|-+HJ(=6}^rnDsE;ls_kgBFtu+kBYqCN=?@s zV^HqQdJ`n}>qB|;DC8JnD`c9U+f(*BI9j>61FnIh+VGeQ^|s>&8RA`W=F$z7Y}oH^ zfFd~wh|w46{-!kN(W_qB8|XE34cpqiLT7>PQOxLlEacG2`CKSyT*JvvbWXu}1-Q2E zfPbJSaIGX-p~YoLOKmw_4R!Sy^I=Y|7ZC`s6Aq_YhsvlOy{8sNBowl>rFZ$T@FXIa zcOQjEVo5+;L0ra9IOd;1DN((e-@C2R!skJ4=I#Z0JI7X%!4Lk=SfRS}=J>z5eM8l_ znQ#CP2pr>u+Gek2SB%r-3to89_RG4`V1ENYx4<3SJ_=uu4gr> z>Y-*xhf3jGYx;ISYZ`ERZyiQ*9PgeAj8FDYPj#+)%<2PT~Y>$R&$9>U1!eG>@zQV`AOZ>xXv=!SHh-}Wl zFBUPqoVfE!JJcdwz2t_hhOq^WVLgW1YOe+9jT(OaEJnW5(kOe%V&_cCi-lt^sKIPK zgv9s472&fk6+o+cj)W^~0`sKsjep)5Viu{c>mBCc-=N+M+Leh}R@Rd&0n0DPTe0Cp zZ@&~~p&MEFlMvteZRZC^u114WlPpXteMye5j`*rz#2-nFYmo2*HrGJSDDwB=R6*t+ zGX_K1gd(A6u=h@*!7?P{hcS?*SNPN4p+~$SEk&59%zwQbn#JPpr`KMY&B6*Ocu&z^$CLGspKBBg1O?H(f{IBf?c{S z4J~z9%%jq0b^Jo+lTJOEc8Tjn`XOlJ`T{4;e ztofK8E|^jzhJ_ea*jx1T7i@7J(=+9EwZRyqL>zZ+#~sQziMu-pOD7>%M!r&@X7g(8G8BOX0sH$zbhB3U4D>!7}UXwk(=Id3ZF`W>Z&g&+nD!r%=7Tj7(q7rei%|p1`#lzy44-60rsf?P*fyxH?b1_3j%DJu=~glbp1v zC}Am~DJrS7O!!qnbh`@6wMzUF_oQ!-Y6~nK(h_-e4&duhvd#uXhinSa1yqxvN=HX_ zySeKoUgp7bdVduyfGn_vF$ZWJlVIhsM4b}L$?7{`@Id)2r-CVpqGn!^_17|(WvfiFIx zz7}}M#t@&g-xJDa72hp2QOvZ2({KL$RJftYeYN8_)#f*4#V@c_GIgqOSji*cK@tcQz!wR*B(7 z!Z?&J-BKrTdZ#z+KK;r+7X8^vE7nJv{9|m0uSKRmA~|uF);;y6N`OnBaFBq22c{JX zdTgcW(#rY{u>nSk6jL2d4SXiE_QOd*`+wTWDv|_cdERXJ*AY!9LP42Nne=-Tmpm<- zA5*j|dw332Q@#=+8a&3>!_wfwG32;Hls!>t=C5444-4Rl*i#Dw4h7JhsK-v`ylYhJ ze)ihv$48`*J6}MIxSmU4wiFX6UHGYQ#|ploK{KvOfk7JP#hOZy@O%+Em%mob=zr;5 zjaYsg6;ede2@7h;@C~YANd7$%7T23zs>=)pS=KMZ=MTij^Lo{kH{^3!2XRQ3O$<$# z$iF5~Px5|`LWSX8A2j*!NM;_Hk1GKjxGp03yZG>aI?tGeZZOtTC1d05)kx`4Ea}=q zH9uyHS1v@veyIAz^Sx^-GKB0oJ%2l!#t8=9N8)dFIX}G`C?d3y>ti=*Z-28FaBdF* zbfuMJ1Pch5I=UFDVmhhe=7{|l*feVRX7V^&f99YrW!ME!FHuG=hBb;jYb z@5HLl?kd%zE~=ARds`XfFq1||TbsoJ@#HW73jT0>h%H$gmq-JiG2iXSYk&1tBWh7( zA5K-147dEubkIZflN<9~)*xruZczqFSh3LM_#rFV_lj^DRybvY4!cN4I^l}Cu@-L( z%?A3RAs*&M>61YW#h`IX=TZC6Jy`^Yv(Dg9Xy)><&IvNiGiPr8K?_tm>xBpDN`F4% zX`Rn;#!M<#M2ksKeQwW6Dt|PE{O$C=ugePlVbGtpV6$|^c*CmTKVQCK&n1+d=2#IYqKCNA>@n5=^WsP)T zH~acuv;6F{gn=GXnwYBEy(S`h*+TSLfbd0K*rhC~Xd0w}o$UC2RJ1$XinY->~xER9eOl57ob%-_{C#}30 zeY9)~j@Z!FvV9?^-9*~D<=s3W+Tpa`kMs9=8%3S#r{NDakb;Eu{7hbRbP`|8Qa6Ei z6#Jsiey5w3A)EPIRxH{?zW07ZaazZrn_|NWM33qs7tDl(WPcW9*28IWTD-w}ZYR-r zev}}7eMD)7@PMhwsn$TTTUXa7!IWnQcs+dKr6W2KyHCfW(fe_li7b?WXHobXDlK>7 z^?0Yke1RJ#hJvjNRCcsTk*sAp02nG4f=v8+Bk!%7Y88jT-{+Auk~LBEL~4tMZY5rk z-PjWmjCzQK*ncs^MxCuW4b2W&ri;!|Y4MFOAIp45$5wtqI~#kCOqR{%c~(wJ>5ujj zu2gnNic|r>Zrcy7(gJy<)IXN4l6j3EGvDU3L63JtjknqBkF3-FQPs%suxb_zS#QBJ zRT|cBlli*n-hVH{_0Yc5 z&v4AxxI%jRuT2YE;i#YaZr?~;#eCWm0ZW3t{jJrI<_x9)b2 zjrfywO7sS|X@THUY|Th1`Y71n=vmwCtJ=O@e1GSB%&RM!)-*xPGe_MD4$cuZK{WG* zs}vmsi7uFbahalzphULk4EPQTCJ+&!HW|3?;r5u0sW+4K^>K?V9N9E+z<(%W)VKvl zf!HTUz|gK&f0Hp-Nf`_6z>+eV?rpV}g=S2hlVTc3utrNRN4pJdNEZ_uX2W=rl5MVy z#DCV5(>KVo_IST#!X+n^7VX90bLVd*Z}MVh+X(Nb2^)f*4>ksSY(3*ORXiWUM`2$4 zXfg=Wnv8AX2&q$y>}VUWB~Eg_*fCYrOmC9OiqH@+&dWdoRWmY^TfPM}r+XJ8saiN# zd21m++4i)8=Iu{;uhXEZK`T8l{&kGbc=WGLBvtJmaV8_| zRPKYT-GU~9AzUI(FW&zFhKDFSdP0yi+1 zFgyV$f9+jacN;gde%G()w@FT}g?+mxXU>sq%dr#NiX|@@j~{5NB{88yg^Od)ub;09 z%_h1`h2B3ffI!afvQoc&5QmL}qR$5oi#Y(%%Ctc}SMeQr2s^n5- zY^7u}m5G(oNVX^&qS>>^Xr-(yibs!MPrs-ye^y3ESviT{gkMz|XH(@gib^~f6rt*= zXh@X~Xj|Jokl|dAl~$Ga%2g^>!J}U1su-lJKwpd|RbEvw1ygxjB_)03LzOh8Cy=Xb z6+m{dat~RQl-5)MbxDKjg9qYJpQl$LqAYlkF{1K7Oa_nwZWWgRZ9tO0f^d|w0ipmx z3?Um5lz-Wakc4uU+6Ur81OY%Ih)yW3tpZ7&5U`m@Y+#g#0hS;U{HF{m2VWH`biRrx zW{q=u5j9$iSEv+-1EG}SB9RV?tmih74AD`eNViE7t*FaM2tz@qz=&CrB1BK*BCmJW z?15dvVj`-51*%_DTBJPCx$?;ofp{Avv1(9DAb-s(gW7|K02Sm7eu)C8Q^lYGp-`2A zk9!I#)iM&5NY^Stb+9$95{iKnAP{pPv(u21haqXmqp}W&gZ-c_AV?4NDnL1~Lm)x# zLClplQWEcqFi5~tl7@I7E+DXCG9zPQk&rDAm`tKqo~jIl0Kz0;Hn3u&32}Ze;T6;Y zm46#T$dEvhH83Ek9|$SV66PXFh&q5vq*+j6Noo))5GVoq)=m*&iIC{_zW;ttoPN5g zD{(xysQ1ME$@RRxp3hJZdgF^d@uHqh7SmxpBb9T?)B0>QIGlW}-by4ytM*a9OKx8W z|85WUOi=+e%(?&hPj2F1F~6Klt8czVYk#BpxMu8R@&4C(IA{F+v>wbU$4`UP`c3u6 zXnt88)`P2hS`Ej8*{u3dPiLdab!GOANB-mK-Y7(`gSSAtfwPrh-EkDM@m$C;%GFT&F^0hrWoz^#7~3f0Yl@Scs)8p zwKP(){-}Zn^=vpD-H^o5Lpk|yFn_D*RfXQb2@kJ_le5wF1r2$HDz~FT`3+Q9xt0no zdsJA}qe81Mr^4C}6>3F=UQdPTQ(^j4*e|ETrGg52SQ25^6XE+rxIPi#Yek{LiVDA; ziqxkf^r$fM4rXCWD$I(A^|onrpNPI~8nYzgEv9t2?;99`+cyog!S7p;>worDg9l8A zD-5#;a%pe(LuG>8cMWWEV1QH0w<@6SjOVvs19LqY=2l8*F>rf>Q5DdlRf2FQWZaLK zS^}W+0Vxi--7+WU1k?v*Cn|vn#$gaa^(QR_}8q<9;EV?L(b+-6Lv z(>|mdv#e6V1>rfpl$CW_`h zqe;WYGqjfK%x_I3X*o$WFy27$tKPv_j8Uk(gqVg&2n;b>?vr;w#E?U%_Ibu;7WGLArVqkSMAB^Ypm88flRMYXZu1efUHYqNj(@RZS@~Ze5p7*L` zkvl^J&ag^|l8lb2YkzV&GF2Z)HW_aqSp^xIPBiXdtwAbhoDAmcV6G=acqwSQgiaFM z@RCVL1tei(r163KbYL;MWe$uOaYaw|UPgFQB4kSEzYpjC66AOEY4u4u?_C9YTEX%Da|k`l28d9Oa;R; zcnNHingxv%M3^5~6(Wet zQz`|~B(=ZtR?4c-WAZ>?UPNetoT7%>X%^EUrr%dDnQUbyjj# z^xvXGeMcQi{`db!PfPLrjT%~4>z(w(D%wg>c@eZ+oPgfZ`FTC9uW9?_tukUbnhwYH zaB}l$I^VL87r!qi^ZM-lcz;}<&)bLT=;CsY2~V7lFolY9`q5%W79i@6!|~v%%_^=J zSMO;<;o@4HO~&KFRNU0lVSPOpgDYUO!Sz|r%$158+Q8w)7|V%Rv*GpZH*qsw%*1eV zbu|!|pKdPeYpQHCITN$-V0J0~Q%@&ia$Sr09}_XZoYu8CpDd;z^M6AvW}}Z{R)0X5 znk&5?A>$CHdyV9)QO3sgY$od87lX05pv{|l3gYS+nSmJ0#ep~!_r(M8P#lRz;s^0q zJP|*NpTtw~OgtCI;%D(foQPBLQoItc#T)U9_?LJ;9Snb~=giZ4pqbPrb;;jwG$QpZ zt_17@QW;LBbruVk8-IxRVkpi8_z@T4QjEl};x{oCSK?Ys#Etk}OvOyh#X@`#e~6Fb zllTv}P&chk^Hb5N+3S9|g-&Tn&a(Nap<%HU?9l4~8txvk^!}?Uj~VOi8Z8`V8$4kl@7S z;`(ARU0jU^i+{P8T)=jJ6T<=3Q(uuNXwO$1JfzjAz`%J9LnXVVEnV?bwxs9H&qK;^ z1?4WroWm+HpdUxXkX?`P5XCrWpEN(UfQYZa@%!;?yFtOF+yi$*$ZiPP4I#TBzVG2!EGgx-N+D1rcT~5xyTx54KN45lr`qq6-vU z4pj^}Ohhp}+THgYN^Nk<5kluEW*9kg7%3XyJp~<-wH#9=yn%`XSdnUQQkGZ<9M2RwNRb5! z2?dI7HNh;C<6Et47*Z_DwDGMHD}#0v-?AKU@*LCB9h~A@VHsWKbc$y=$jVT3i)!I# zif_dxN12zgT>=?@RY@pRVho8S<1EZA|4OwQMP~(at(nyfI|hM>+p8s=mC2={=cBn#YnHH8wM-2iWS>S^UOtyi(P>iF*j%~|J&3SB-o6eY`CXRpQ zG_nbff+ix_v%V;z2Ld&h1%dRIQlgG*RiTM$(jg8Gzsf#;DAn;LYV4~rv^A_SXXJi* z_m_n%R6kh3ycvj0-YvX_fgVzkZGeFOT^C&2e#8H7VH1q*r|8sOnEcbY%HI&z(BgeP zA5_YT8h#brYKqn_J$3fpVX;g98aFA*woYsra62QD17 z?dpzJoYDb*Y%!uE*5aB@C5sz!EySnoY6y-N1yz*d7-J7D1W=zQ*L_*94C1$L?GP3m zc(fwPeg8sFtaLt<^zFo-)}Lw&{S%(4f2uLu;e=-_y^pv$;F;E+YK;8@o~eJTG2Y>T zXDNw-g!M-nW8Wc7{UeRhe(gzD=~h&%Khqfd4r%ItpJ|MDa7dL@8~kEA;J=oRzE4W$ z#k+(gx$+0YpN##Xc=LJ4)6jBC*}m8ol`>eLx@@mE8}_s1=|01?Sv6nVpqz!15ZNin zkVKZ zBb^I>!gxXl)S{+gKP2{e5|I(}86eBPPIjLypX$`?TqtrHQp`=Ud*a!DVn=k1cXrD| z%CK>QZo#GWOaGEhDZABtUgs?z0JFuZHp@O%cK#S<7cRprmj<#&bsGjJJAXx1q|NqWOs|hz8OXjo|)&On|p+=<~42-_+fCh5vDW1jX&q42MkD14+P3U6VFVxnK7sbWvM>MWltT0d7k zZu?g(mArMwQpUSqvfOT^;LC-AFV+ca{WiW@-uqlF6jXmBpDyx#`5t=Mv|>?d10}Xk ziI%r0ajP}U%87K6XY{*6BR&5`cF;@cv}D(mT}mT)*O8CSOWK?biLb)&cAVFLe6~L4 z476$3Y-B6iybGQ@cZaok@`~@SQI04R*QE+P^1Us(v%%%uvRx&O=jLS#ec&_s#}pTs=2#BPAQ$ja#?RmA&O|l=sgKUq3V)pml+_FTMhQ=6W)W>_wE& zS(==Kcf9?_n`53g4GjevW1jpY-Q)E6!XW3QeL=b?n1j!Tw;dC28|7_KMFn!EXn6}U zUjxs$mAn&1yO$h<<=fG@armu1-zH>A7^17u^+2=a$L1ipZa8S<#5c6*_Go+EHsjjx zMcY=AoT>u{?Ij1ZUEW21q?9BR`%YZ4$e$ry)6|u<#5LXj=3PhH$Vof6gC*P<#tz;x ziQU3vw=mf)Om+*C-C6qGps*Vhc7wuhP}mI$yFp<$D15=7(A$)x6D@5dk-m_dzir7+ z{5j*#?lp>|r%w(~fBEkI(}!=A?CpT9_OJ4O*0aGCz3G$?N8`bN#jLV9$06@P-qW_a zrxSg7*it|#=y!YKU`7(BhdptBaPtHA^ZZEd&^s=(CmzoSyGN#9sj9pK}_mo`>j+-{rjx;fjX`2QOd0VNT*!tal1*HZS6wg5G}-=dt9RneH;wZq16*gO@L#9fQ7; zlP#6mEwPp*uQE$1B~YD8j#8JS^f`{?27XqK zc+&}u`x^CsmAs~s*HiFE-1Ycs<3{4?t5*jHkBG`&w$x#_#QL!%Rrh%2qVLX!;L4fH zb!RSDuyt$VD#jMuQ6YJ#t9980jOzknIu3X|uF2^sP9X@6*{dQt0 z3#^q22Ha%nJNfiA&gcuFMBfn#J+6de<3^!4-5;QTmfte#vXxr4UOj(!^yoQLxt$ev z3ap<(VZ({O^{*lwkuUpe==Rr8_E!|ytZZNM^%;e^HtZ;VesKEw{-f`no*X{i(vZ6) z*763n?pe+MlK@jCT}+aAD0$<@mN%<6dSD$he$i!S+epE9-qo2w_~SY zX5)N+VO*jRtD$=fD~2L{k9@@tKeh}R8frHgVTYwhSTP=Bh4>1?E3N4|Yr}FyIBZs? z**BtvU+$jgc|l8Sz-9qFYLpy-17auEbX!*0ZMvhd;c8^j^Ri zlG|=J>y=qQ#>WWxX+ADb|KE?ojmJaBhHU9+#+!{c)D?Km3*=dEgm)`f>-E8_#}8kC z%99i6x3r~xnGGJK0h#o_dkGP}!-Rq}6FQvHu`sYOGV=gbBve#bn7P>j z%#3Vs6cnOPW=1YRdpj{B7c(A!hM6fq#moV~!U|wzW@dq-0EpThPDVBWMOR}Rpb0<@XkupPY(@hxw|4^A{O15Lv9~h?{!5cH%mv7Q`=Jo)N|G=iM4*yO4Bgpw5 zeE{lzJfs1bnwkH@cD1olFtRlRP>b5zI=H%+IRWJDP0gI_04hdy&i^=n;siARFV)Bv zXyf_+Oa8wMX&0k^%n-J-_{Tox|3E-z3806WsUpzjAHU6wZ2s}--?)a^e}*G(W(st* z{ZGBwKc@UMR#SUB8_)lH) z#>xU<`N7Ev;Ns@;{r|`{admPsvvc{k>;Lrf-}Aq>$;`~d%mi*@)!u|R#45cttin&c zXucAXZhqEKjc%0%bZw^hMh6u*Y-wt6e1&(fikI(>KsXm@%)ftsxhH+!LTVImhHx@9 z9YYeKw~F^#LRM+vSG)7BRPfdip??jbj%#b80C5v}=aRX(5u`{v{o_~ed993N89%Bd zON?Q8=Kv(iZo>M~qn#v;9u-=a_BVprAzcm1yk%(NwXxvq!3J1(e(*lR4F#@&Wno32 z!*RIQ13Ps)_qywUiIy69Ed{)6cjarariict?8}UWcF&vDTOcaD$zV{_QF;|zg%2V} zGmF(4M)+T~o+Ya;5hLCj!CI)4Vb^uLTxbM$?D0DaY}s4VGdbTJ?y4}st($#9HWeh}2t<_STR z-rUSEZEm*o*;J`*z*5YaL=bglANmOv8YAg#t+Jp{J{|L^UP)Q1 z&BMk`+kWo40IHw4G~`=0afZ}i#-u;?7Lro_60r-Auny?q6=R1n2XMO z!VEQVz#)Qe;iprzqf_UjFz1ag0szv5BF=u2eg0^F`m3wDwq}HgGLXxnrre(0l}erM zgM_5(zgA8%re3m<244zuGUahub`>$Q>oAQA>r7P~&ua*fmQaPQSRu!wboFExjF-pW zj+0vLkn4UxZO5>=3BV@31>GGAGa#cAOsWI3a4$R04M7O3T}s8E5!|jS^8l>UuAx%_ zY?$tU{p?#BCA&1tLdS?d%Ag2u1**~Oe&nggPh&pXq_6!`UfIwJ zTIY4dhe`J%ElZtqHqTiF&--n?YV7n!J=eN_`gN~cD5Hs6%yy6PwtDyV<2tOtWJk>*HR>D-O*zEur4=6BgDCGPaj{ zmKU#{CZhdJVD1suXP^E0VRH}0brtR8X^b;f?ib}IkPAf?sxOfrlLt8ONk0Ql_3 z^-Oxbl%;cr|?5NL1}r4?OE9+azl zCB^Z+zP>q+%RR1upTl4;mAEd*{UN;!{)S*Ut-bW(rlM_lCGwe#K7tFnQsB*!?(i zDKX@5910dGf=YCV;}QYS*kPECv0R-}gI0F8TMB5dhr-nsWPffR<0e}+NlWCKXpMOA z3RCq17JvR#N1S(QDo#_D*b47kJ>(V%xOWc6>4=X|Vh;+&8xELMsyunA{gpwrc~wL9 z=ts=a@lzp*9`*acF1(c3ERbt|NC{9Q2@Hs&K@zcFE}_~pD_{P(bcb;^vOS4f{XAFv z@iNpwWW5Q*n@h;SmNM3lU((CW&W2}04}PnyFNKHj!@IjoqhiBwIU3}9YK!2t#}=5j zz;>8AXHvz99_MMS(-M(^h4mjwwR|+QKbOND6J$2kI8eyp^AU109-g*;HZVX4cD?Ks zYtqkUgfHSePEL*%?Xv407&$Kow7MW;@-@%3U)2E!l_If2@I?zo)D#J^G#|RC1U?P( zGS@L}n&tw6QPgPF`t&BcaeA9)g#7ed;07{D#7}|T_=?&Tib)VW>NUf0M_CXVIzN9O ztHjQH7f>iRt`f2ie8{zbnA1qE2nl4E%3g++NN-`?^)0$bl;EA$Z@Q3C4d!p-Tn8yt z#{-xX&mk~|*zdx{7`-I*PB)YsT5>5m?B}ITE2&_4 z7uo#SsODua#DRkZaJq-V%A@jq-VaJc0fqAf@H0OyDiN zb?ti-Pt_gK<2?CS!6NSR*N^L(v5wW95j~=AI%;@xWBmqj*A^1z6BxO?}^_{o$5d!AH>^ZZa!m4wS!4g}Ho+cV>QJR3<$C2R23FsHEVB@q5E$`ebXzeZu{3WNyKST)@|bWk%& z%Bk0`3#C5;s8ssCe@Aw1`51trQ+31PoeR;c9tUjFQ zw3HeTy_@3@L3ye?kbajvn-*70Y>G8&T}jM@q+}RfyFGok3bC(wueXCy)9cLxeFq)! zIGL*jS{k(TwITRXT(wz)4gKS|OcKRHv8m2p?n~W&GLb$&G2+i(&_;UDzAcB8a0e&6 zT`EVeB?gTe>RLfjkwblnx$wHL#}ZgX?OmLNDd~hLP^L;1cDC}4qUUeRu3b$q(FW9B zw@aw&b&FP{Eg#dDJ(C};>?%z)C%9;cw5hJouMMKsX6^7C{u@5y-SQ#cP5u+$q1?%k z4eNt{1DF2KUB~1Ys|WBO6P5mlw2rlR)e3N7f}Y4nz2Blg?q~IY#$xcM_ea1{yODQ52V9) zRpYZbqFvKo+Mf-Z5`ghaM$MBmOY(a`E0|1wo(-%!TP1YGilOejJK z|CMu(54C=+svbLfHuBARI_BP_fk>z)7A1)l*Ol?CeWKF)jtqIN*Y=hC#~O3vJ^03U ze$RZ!8jBO5zp(AGzot^AEz0O);9vI(p@f_aiVvAoOKQ0;STq7*F^#*P$uECCCF< zi9=%Wlzb?(g&ZC~{>;y29HW8459m;Z?afd#Zp_ifflWvw9THf}cTFqGy?&t3y zFNGS?Om77u3%x^<1SJA^KKmVRev40kagfVEqt1Iwb@2q9hrs@MCt91SI&UZ^Sv=u0 zWa_uS-!cv*4}E^OCt*+$p3WDVc;!KoVjNvwJQ&JiLT!~6Jc;pU1G((bwGW->MW^NF7;V^wxp(+0QD%)R2us^0W!mK{Hvu5@{o@?p7Vw21ka6nDK+0Cylp~T$qo?B zqe+O31V{0Tr4N-2Q<|D^{B(E!qrGZ~h zU9l{$ML1+=ItPIDPU@xiZg z2PMZ)sMm!&P7NW$m%2=SFw{4CHLb{buAA6neV+Eq@XKfaY-X5Rw&@~v;lLHBT%NRx z(>9-%x%xzkqlFEf9o-+fK`9|rGX#@J%h#7a&qo?Ji?OA|2yP>PbIW+{q!!=19~cH& z6f=e*3+mrrc`E$$$o>{2`X=(LiaT`BebrwB``ApGOojV?69A0(#? zo~2AEIkH#d?_O(vl<@V&BAb_-Zh=iR6|GY)aVu|ZvbX#NYu(x<`#n!ynf7n^Su`#M z#F}u*uC?JMK2#+VImXJdBQ|1azZW#|)q(+E!R>8r-=xFawxRi4wVpIcPkpWu~#3f)g*Ks0();bf04b~1{u{+t_`HCwK2;P z*p1Wi#y#H zhnNkB&sX7pNC-=uB6a&#$U+YZstL=Sb&17B))TI>-XGl>KyF~uLk*0;z#{Vp!*G8( zAYKhZ!g(M`f!Jv*C7d7>=tz$|2r8{C{hgsR7)>7@ZCI7+<5u=*u+vR;Y_f_*TxUYj zhiZER=}18w4W`4bFW*}&lO}@fAN}-}uPJkxHXWjWeh!CW(WPce>hanHaV&1B;P_?A zV9?AN4G|THMfsZ>TjS&gED*e>WiQuPQ08Yi>hKC97~1Uf5amI}I3sPIoY!_hJ9JAh zT)A3#dSmRT2OZ06;gg$3F*3iX1}mBjY2Bv5nP*~7uy5l!2^y92;{ZJ8gCBK*+HYR+ z?=XS^7Q7WVBHa;U;jRDbe0Z_F~--zot;IV2mkF(`*2^C&v=T5U-K z^4I(a*?g^|ki5+fmsvGeys9Mf+jC2z?F}`*p?2w=LTgt=d^!E4K)4VXl$gMCZTI!s zu^`T!W0Dd;Ui&ALo@HyUL!2u8M9;J9 zl+p~Cuk907k~88q+3*rwhTi`^YP7-!?F@75>u$3AnSV~;k5tl_Vy7MeK^d-McDgiu zB-+u0R(z?%;Pl3gzDzRRtll>tu0uq zAZXp7BYvG%rcc9U+Ssu>=l?1gNSuHlM)61E_9=QhR4@jXidXgvgjRM4@M~S*A(4=L zx*ledcm=-W#A(DkJrzSl-UgEyN$I%4A`a}*iNuE4Flw8}OZ}NoZIan-iP1LplxdV5 zhlaCa?nzaKkb`Y_ZB;@tEQhL{;ScnGh^XC{nXTWL7*^IRjw}#j!@FnUF;5SN$228B zXE(gFZ#1vU%X?S9Im&P7N<0N-b(gAM@Z;i~e#w6q-Y0mcRJBGm2e7vnF1k_o(Ylg^%Ic@c!oZj`zu#cUuCtf$DvaS|#UiCC0 z(8&+>&gvG+_|EGc*mftKe=KP*^50yZeO zdrsv;EtQ_Wzoy4s8J!{j<3Y{+lFkh!UB1)1kS=lWGMD{S5_c~NMe8N0UDw7%l!!sit`f~zxx^gz*? zaDuO9AposW(ltPq+~L;Li%7~SSFniNQ6r%PgxECaZ;E}6vJ866k6~8z z@^oNP*@%Gab*V+1d3Z@z5t!9a^THg_53^LpG*Hojp(NT4?Hy--g+9bf$7a`q5{(l2 zr$w7RxbtOszDesSnx1C-5rF7?qQ*sB zrk;}vB4oAAzUU$KsEV`6AKQbeyvNv=Td4(+2uL&w5Oj0M^jWK|uU##Qvn$K{SY9s} zjzHp(&E&gikPu^k_0QY=$i#EFdK9M=Qr$L3e6*PD8at?Jmmp*(cHL=8h)L3wm?itIVI=u2)k7R2} zKDKP=#6GsFuj-ZbuMfqJ_AETT+<4Gf;`^71*U__o5aupt_^;|4}&?~ep zV=s6vNdaVkdwD1N29=|(Z`4Cv@dPae@uAx!u$T4KQGjTeKZ9EAt+A^1QBr0%OQ;i{IWTxJn zoUZ&HPiIm5SHTE%_V8R2KDd8G<+nORUpd_9uc}FZPXshXj#lt)u6#dTJ^jq$I?~;I z9HW!pPDaC2yhDUE9?X2@wGs(C-irre+wq1A?EI|m*zE-|cteSGl*_q0{gEY)k11Cel zY9$d!E2u z3_{%T*G7F^>e2eBK!v$!lmiRTiUTeFU|oa@?${~qW-P>tcj7A@;DdP=8_B2HE74(? z;r-@P0c+zq0Yd_tU?OO9#f0L{c{keH_C%0_Q}`9t zyCTSDs@j#puTQ4n_)o}AKudWE%3 ztE4jVg)Tno0TKR1*45Ta9$BwJgP{=vK`h&J<=+M$HH{S~sYj<1rKIzJZ;=_kXsTZh z*-MriQEIwlN{OjpxA9}XWSN6eLvETwkx226yM)?uU#%I}JW$bH>My#g`HGHe44fNj z-bd;sPryXmqzp1Y`wgUg8`*MyDEf{z9A5Myf}gVGls;eeQY&!w2_VozsIRgr;P#jO zIx4xZbbI98AfB~`1#PTn9kpmol@-O9&oJ4+Jc>)T%*}JUF(bGsv!KNWm9AWJN0``U za%eTKcM9dHthISL_@bLLUHQ-V_XHoG@3q?hygGi5ilPLaB=oP5pegi!4K+Oe)j^9< zbxGxp`H+><)f-{rBra9YYhSFyH^!K1+%uLW2{e=5-SZ%nNvi+V=esgf>y`gU_AGuh z3k{vx*^{;`vvF6k$aTu+2X#KxhwPo>edF7~@o~%FFTV9~B6m;=@9iCQ{3KCM$N9eS z9PBiWD1nLX+~>Q4njp7-ILrc;D(va67L~haAO4<;J^H)rcts1V7SUBc1 zuROWN$K(!>r@DM;V#Lt!r^Ti1Bt|zJNpR1vVllqx6uJYPD$rSf%>t9deODgy?K4HO zHL$GW4azV^8t2uCw576}6l!~Pi8s0*zj`k$Ad#$HAx0>1Nf6B;-TB z!QRj#frXUE_~``csVMWY>3F|_j1}rgotqD5O+g1+qU4CT`{P$nvZ9oQVR`>l3xI4g zYkT@Nc;oUra_y6UcCTi?{;qJJ>yZL-=%DOP>zYn^agU??MgorsgYnT5sQIIW zFl9Q7^q(EjQ`uX88b$lwY@s5p-{?Q9kRSn2v=)6n-L7>sOA?iU;O^U&^m1z)L% z`P4$2zK2uBgykTl_HPv`A)*W~`|LiJovG-EA7bj5ca~fE9qNeAQ3N$h@X#EL!~RNT zBUGFjFa!!B?lMT}Q46OeABuh!)a1y~ga|BI5|+V#`D>v4+ykiXB*P4^_il7u5AGQc ze=iu1_y?+~E2ND*hIZcv6z5iG067j@p5fy#eF&@#y$7nv&v0b^I6I28+8&c@1QIWg zfgizWZbQ2iUp`iy6MafSQ;aEuFOIpl{3K|6@QGKI9K-4ffBrdyX(%8Fe%9Vs8I_%;vUsT;Zn5*qmcO?W zD7x*}$}Tr`USb9*TF)D9+-WyPpvhW9c{%HUZ?ma{t?`OIO)NbrAc6*+SI;TjI{m9b zV^yy}KT9Q{Mh$|%-TByV%F>D0U7uB2CTNEqzY!$Pby!?HsrSjPo*X_ykBdbg+iD=& zYu(ffv)NvYVRh8hiZM5ULyUhQOXrw*msdp|%kBLYjqrA0BB?ri;t8J_niMRWKlMm| z@I>z7fgk|fysk5wRm6|?ULm}_F!5(qzG64hp9G^SaB%o(6y3th0e zN;;0>sW|QKi`M%>W%L>nAT45q(;?u0$Q2OzaV;~Ylb$U!qI`TkTg-l~j@{nVm*HZO zAoul0%04iwM?&iR`cx(O2zgy&uy+8}R$L@PXOUoAMdImv1F*zAwHhGQx!qM3UBHXe zU7M??$i$$5(%Ei8jy*Bu*kGQN9bo18D)@xpq!d@d|M z(jL!Q{zzbBL3`~wsRZtG7d~WVj6uG1+Hc|-P4rae(EHJngoS3+SaWEdD*d2-v2#0d zEZ_*XWIGA9aG9x;F@xwX7U>>;>rFQ1PH>lx-ZhCjj`YYhok%CrVW~r;tLh&5ZkE}t zufs2bN^G}VJ(_@%nWw*Z9WPP~6-yQ)mXu9vU31Abf1O!tkkFs4@Aq6{&fd&ZAA9&o zOp@)5e#my<>&CdSXy2r{iuxAjW3@o~_Wz<6kFWi?Z+C>g>E!%o5>F0)$w6Vj4Sd0eQH_-dQ;lDs_OH7ri?)@rOtnvZsaV@g_Zy(fW7ucaX*htdbz z>NegZQus!&_ZO19+u>LmU+;b{IUHMy;$z7#7YQziDI^vnp7kd9Yda=G7OIYDv(H$W z1$%>r&Mj7g`7`;Gco3F$Sg!NThIaQ%qb$u z@7{s>N#7btoI-Se+1_)eH?B9cfdl0CmMMB8gUL)g-9tfRq_Tt7R2$d&iqerWX|kBE zfo=0iN2FE?wmtUO0JS)521RS{yiB>{t;}as{j|UdfP!JAo9o-bzKyb=G=+#U3Aqp@ zY5@Nj(o1)H|M`Qw`cnW}cdcAkM=e8w>>(oqv(H@w`!8L8oBZH3r1p;h=0gg-nw6Q${Mb-3n2OkRkVbZ&BEt5uP@l=kf)*z!=M)V299Q{=s{ z9XiHnmAxp-rD3cX376woJ>8k5olQXw=B;0dlk=cn%~#t%$rntg15TzC(F$27N+)3% zT#B*mA***$F}efN(LgH75ro(T(*kLf6*v0}PWT*upXjoE-)YP2yvnV=ti60kaop$4 zx!UylFKaH8_gV0@X1nc*W~h)un_}*3-Gc$BuISva0yyZs$Sv`TY6F$2G#Y%6isl%N z5=cqcQN_@~8<`DJAXl>R?1WlWUHRYxIA@(%{nH|Rer}{_vIf?XTQiH#${#gP36&un zlawca0)bAx`}@D`MBm!}Bt$6LB<5mOG zBkv&-g53JW(HwCYjXQZidRN$=sW(|U!^<1FN600e${|28Cy6X@%Cw;0M9>!S)#znT zyC#kC?7_aZ(p7Tfnm9g$kt4lVH|oRT^9whBZ77n%-~I#+xKSzo0Kjm+KVfG6^mpKd zW;m%f-<8nTtr>g+^Fj<;| zW~4Giqq5?De^sc|`8Fo%wSD2t!#Gb2pe2Nc3h^aW80+d@?}8HWF;@I8Aq@5a+e=`7 zP28G!GYW7SRcRuusuy0_6L*f(%W$9C*=Xfgg}cvyMpm_zJ-d^@_ihz}-(=6{L zI}OU(l2OeDvlD^0^IBST(Q@aj>)vrhXUG5bMZWrahEcS0A|mwyhw=TL~i_aoVc=?9#*lRg=LWmpyIrGf*HIY@({K{@Ja^QQzch%I%?7_J~` z_{6_TSv9@s-lI~{PP*D`O<_n3!3Gibt{n+vUj|5Z6Xfx+D`~vcwbC2SeUguL`y?NO zQarf;XYd#3CHaBkiEC)P;g6*7Y76_8L@wEQW)PWHhLsK)`G-Xglr`G8zt?bo>kCB$ z=?oB#+!5m<151w9t$EPesoGa^vQ;e4CMAcvZ?|uC*;z1bavLhlwl*#M4nc{&74*x% zvlyw?X+iQ(BoAXbYtYllA)XRo`W2Hgv~(kAaXtv`(JN3~v_ zvo};}TeuuX@v#IDealORD?lkJT@$N^GZ|*SpcKjY8|Tr{?C{eX5AQ{lH4Fm6U=0IAo9drL^eo zdco+Sz>Vfj1KY8I;5-w5cU!|m0we7s@in84_X^RE_U-hBbua0M2;Kpixex$>_A0fp zZz7P|lJZO{UX$JJ1?|6ZnOS~+AuSI#EHCv0S=ABUk^AG0-4#pjmU*Cq{|u8(!1NOm z1;el9_Uj}&F%j%f>_4v&eitCnMWovSnQbis8KiOryv1QlaJ8_1Y6N8h557h?lKvW) zpfP4TRMpe6yT_#-;%)T8#YL7MJ@I%CV?Et-{9X>%3hQ_0XGU_&2bDjSfq)JeOgBC= z>zNY5-bC~v^quzQjkf}u!K9lU3P#z&sZb)Br>c0%pO3o}5jR6!g|a97MZ%V7S+%Em z2$tPA{!Q}^BR>sSaM1LZ(nt_WoVbNv_0nrsA3X4;PvgGfK&FRZ_(?-(z83`y&B1=gwItb5w zT(9Y~FZ@z}Y)K^AI*i`#DKoq1t|XJymW@LX03+Kd;WIOkc2LygZD%s(eu>%fTUvZH zN4Ve2WvKC~rD5FApLH@u&v?0*hS}@SMP3F{AJ<7uXKJ-CTDbGL_IeXYot&m`fj&v& z#cJw-e)RA{Yh?|H9~FXlrQ^++?vm>z&h8DvO>+`|wBs@W^W+&A#CNZ6pX9HHTykH_ zfoT%kan0m9T7G74gw~&_vA3C`>+JNyUht0h>sy=Q;L@5*a$jp8c-W{eTCIG2krE~l4j!H#1%DpAkt=Vxai?g%VFut#6 zZ00R39Md-6D0$hM=m;$$n}8x@!9E?Vq2_3RwL-MNQQu$hw*l>kGz*WDnD3WC<}$_V zF@>E&bS6-jMPu8xZQD*NwrxB4V%tVlY}>YNTNOK9{dcdP_VB*-rg!q*+552TZk+s4 zq8*yTV95gZZ&H3Kw=#uN8FtN*;uO;p-2$l>Ko&Xe)K5wl!$dPcXq*i8@5&mGAwtO9ZeOjTwtc zeUTc0Aix;>^}gUJBS-}YU4Y55ajDPoHl(Gu-8Sf>EbE}pQFz0&DEf!-+~!J6JM0rV z8uqDDqMffW*2LAW;dbEu65tk0UjIbro!|bbc3WNRw?1 zOe|XEeiW;Bu_#Cwbl@rT`*1)WneWGR*w4B(I9>S7`dB^GmVu>IN~qg)spNPAm96pY zGI?J=-ko!h~XfVdJt-%aL+A3)(Hl$nS4$L%Eu z9Sjew_eLRAF|P`QXZl}?iuQ6L9?cTL(D*!pWw+A zr%LhI=`uU6m_IZyk~ZF1l#0mSWhRW4KsWieNzq zxw@D-D(_?w2AKL{-WDmmxV&l&%bSwExvw)|^xh0suPi9)fJHeVy7`w77gK zr>k#~1wE@0G`W_Z`|$3%dU=?b8W0tkgX*m+cq@Dvt3hzk0F3Ea^o2IS0QcDrgTdby zdT$vVdBqoR-)EMsb7&lxmi!@UzM7YYRfPRLGQi!(;{}H&a;<+G=-6J*P?Atoymp5) zauvZgn8!MR51@yqKmoJknf^Wth_FK+Z;&vW_N<< z(cT6Efn(rfy+R6=LR~+Ak%BnYtpd~VL2^14C;kX={2K~YcxbNtI1+!T?fv*IS8qbb zA1-Z1#ImN4iwk4_d!%Y{+5inU(RhmMfe^0O6neno4iLHy+H{A}suiY$ydk7OBcXt@C3Z-?y5r3rlJgy;q*n8reTXmz(Zs^wT1oxb`oW}k3ms;(gxt48CVuu1+LU zITA!k1t^U^wY0>I6)E0yyRgcXhC{#(7S6SZaCC!aiN(c6M_uWsEhM&doq}+V5WC!) zo?Lc5Bsvy6ZjRN8LSb@YyIbly1e31+noKyuu9O=|%8B5QTP1XhR?CK90bFq8XFWd1xY`Lgl@cg`s2Js4@?j<$A1R26g??&h*r65Q^0>c&9960&;nw z-_p6P+)MS2b$npcOL)jl@QXp#=SX5&;lRenzoeB4^S`&tapev@R_mn6P_rBCoBZ9+ z2`D@|)itcSC-qGa`*WC3?4JS6V&Q3RNNf$XljuQTXp2G!0tyJQQKHxAZ!C&md?K9M zN)K}IRR^0(+?oV`GjLlwN(*zHSh@SABh^9)6-*a0-n?) zI(8BnnAgr60~%qMg;?8@-AWJ+12<w3EbkGE!xhyBhN$mG# zF*Xzrj93fTJiu3*Q43RqafUZqpy@~m&ovKYilVNr$s*Cig&|Jw5YT6dMj5>i9c#J; zN^oms6;raBdrziCsX7z&bqo&M0ZAC`Em5V7N8?Gq>TLM%r?I)D{kM!I^QhY;p6*q( zD=EuDLkHb&t7YG4W69UcZ20U9ma7)_gya2L)2Iw4<;|mvhtaB3t_hg29LdbOrnFJz zk@%ykgNaxPP5*Fk7J3ZyhX29`tUgk~45W&Q@L*BqmkC9TS|dTq(-}-41Fn2R<<4;I z=}{&Tmez|>Yg$za2I-aFXZozQoXL`f`Z&9&HIuY}j*ZT~V9 zDs#aZc>8^s;X81|$$eA85il<-t3Gx0vTeRHxQ95 zOO}XOWlr(zeXw$W4aq_qv9x4Pi1!^u4`Q&_Kyqs&0d}Z$=OzoPio)0<8I=@x>3doMzJ80@!9`L#s#URkskwT)={VyDhY6-^+49jVbdte$OU6ou!YHBjz|`2l=U ziG_awJ0b`2LHD5Zu7w(^Ct_PR-_1n?Lf&_b0(<=(1c2mL^F4bq{U{W`tLt~;x_pf> zepmvD7yGpY1F(@7@^Lb=jn22nh~&7h5Zb7aOy!3f4`<5_Yx1G zTM5^7uSumQ?~B zkmd&rgfkA1$!>4S?!>X5!^-jE zyVQy;8M`TD%3YVN!S6p#k#-U+N}U4JhrAu`+^}`_t-m8<8_pDQnmMrNwgps8|GpHm zB{5z})LS!!T=s@1bPiFkqHv6K$=9{(w{k>N1A0-8d)DScal0~IZZ>|o2cb(79a2;60*c*J-AXD86OMBAVGCu**tMB?eW(uH3+oyjr1Ek^< zc8gve12Xz!dV~Uoe&IbCF&90;Aj3&OllwN)=SL%z1$?#=H^>=&pKoEYLl*8F;TTh|-M zftX?Tf}t-(W4c&XpyYqsX<(Y`Ed_SM{2Gs>TrH9erM9YO_Q9!dXivbh$hk@4x=I*_ zii|=YE7!y6kQhZLms9M+vs2iK9$IhLh51lP?t|z9CSB_rkz#L;HGuL3Xz-fyU)%d7 z&V7=gkcrOu=lGCCNt;Bz!WGV@{;Y)P+ucsbl!AgAxu~@T>>$@`VLtBX%+BA`;vWd9x$I*Vx93G&BKOon29wAhMhr>Qzw=MrS=ojE zS(_|DF1Kf&TF1SEPJ@kkP7x*0(d+hVCw6p=e%+ZpBpC#rIg2+G0v@f}U)i}A-q6{l z4P@iR-VK|QLx9|^n2dCF1>Z**V?S?dv;HA~BA?@UQzC*%bZ{Z7+?*U~BDP$g3EGJ= z7~(Ya`(L7oBKwk9Tha?9Pq=eloD47!vkeqbt}z49OMqUC`5wLzKI|7e+JoEJ3Pc@w z>zXb?^SZ55BPMot0I=wnjwC?}57TCO-ePUa3FmeXmPSo;>AiX;NOTc%%0uWRepUkScr+m_Nd0~Q^~2(@4c2a2^%fCTHg8)4|l^6phX^*;5U_S8;u)G6sElgIGm^-k$8 zj)0cUsTJfJsY0&;rK))TS%AA4_&&Z7*F9$;ud?ind4G_EG3t(A8tcGFn|v?Tdh~PA zwl7jAx?o}+U^zVaPn&WNQ{q-ELlp6_V;sjLR+56KP--BX2N;>n5R zve#d{HC?Bk17_UZwioe&}<%%xtOBTk`g%&byd+F1jrC#{M1`sY1vhrAM7IS+neS0iz zouUasjvmkq#8P*AH#hdEIepXIr+F@yYTfVOuC@AlK2IBZodP`8*piSDCF!UU$*IY) zqDVj~z#!aSo?Jk9wp*62uF4EVDo|_l`@z&;WKeU!T0j9rlt7f20nx$1sIdq{$7g_} zW001HFd(ey%x?(b4fhfZL!UZWq0RZ3d3;-YcqDfZ4@?gaHxm~pRI`soFt9)%BpakY z6kA9Tsq%{R@yam}EHkwxpavqygy?d7q3P)<1X~0pknCVQU12#OSxL-*Z9(Y1t@(bC z&vXinmR@kKwqRW%K!{cc*9K@xF*|_VR8@@)ctQCpOBqN75;h=dE8iV5qH>RrV{B{r zsrFnGDB*{Mckzd$@#ptX1w-DCn5|__4iIb|H<%$r0|;+H{m*fBS8y?DzzgBp!1Vd< z0-$8!v<4oK_UsB?Y2JYypmC`LK%S-~DD0aj>z9^~i%2q=rE?Hk}U1aAAr zxV%JQd;0W4{=+CrVEjG`2Mt*%`0Vrv8mMR1mq!H4&U6&Nn^xlf&(nyz3jF25JwXIp zm;EL9RqziPZ&!zvgygOL;lH1y0}zhY6jT;5R6gJI(!X}nT^rt50=w6N01*S=*xl-1 z5nns2%wfJQLch@;Ze@VEegKRb9HDIi{J`&cw*P?fQG?yy#QmRoGID%=KTYnpH=wF5 zT|6KKdtd}0LBZ19AJWyW>0$6AptsE@!N$+*pKf`g1j;cyJFA^}U8*IDYVNKrgAT_A zkiXkUZILUTUJ7fW^)s-3x{mqHp?`0%*PAJ{@*p9O#S${dPR_$V-2kYuRv=I&08_R! zo(0FetDcOOfr(!8>tFmog$xU4c*5CBJG$WeBo{DAJzZ#3BYYFdnJ&72_H=22XeY64 zl%KC5NoL+lhcgWj8sx&@Jh2H4n1nWOE1W2*2VrKWhi+Qa_DJqUIG7pEq++XKCxUa_ zNd?lH0!2oe-hu!hrhtyxxmf}ir!+&Xr1DHe7S`rm%s5i;6c+=tI?g$gjYW!qA+7hJ zjK+6C3SBKFf1L^Eo(PTZa5vC>-5tGUb#nu8NeTnrlX-m@OGx%HRXj88ZB?}hhrmw?paPwXs~yI-U`zvyzy z@ms|EXH&vZiU*Sj!o5@!RYRh&t(6Jc-Zfe&YRnN8c5LZxak=>Z!jyjqn#CQ7D0vCs zf7ZoPV)GOhl=mI3pT<>G50e)a1mYQr?5i}r4pLt>6tw;^kPenkcT2=T+u@(?Q4HM| z15sWYOA(-Fa|CeB-me=7Ju?}4GGEY(n#>Lc&s5;PvPha0m1v~*f`TwAb-E~Js!4=^ zm=PBJy&mS;h@cqZ*xq;X_!Dm58nY=IKhw{jC6}36iA~Itaz)E0DnT$aI#VWSlSe0j z=vPogUNL-OB*i=}^}RCoQ*Btrfn+9Ug+u+ERcqnMRu4GyhMf(6PRHSF8@B|e8{-pi z+1~zpxjxM}7HR2ED|fE%p7iGKe(qt^DU|BC?d=F@)W2caU!KRJ0i-Q@x9Zsr$~yH_}XYpC8gk&7iWZ zwdC;jmmeU-FNV1^lk+yBDie7p#hH}Tbk6xTFA2#B=1h-udKV#BS(u^{A-))MnT1Co6*FYL zad;u7%@e85I{iuz7jkxf7wKyjokO-fsh|V{gv~#gS1jnFv!yElJ-D~j%O?S`ni8aR zFQGL2BFX7Zt^HV$Jff`i<%qHVsY7n<}+PPR${!}y*x zj|pJM!DPT_u)82;I7mWf8GGN6x7O=s@4Yg+b*xv!G=w~}drk#E2rLQ?F5r95ylpM@Vq*xF=WB`rO#?Ej3W*fJ){Pf}JfV%j{%;DIvIdq5p@_ zqfKjsy=g;DNCmzlyfkt+7dV#~1<4=JstO=g4Th0yHz3x|2W^O*&w_Zp@{;;p(ZaSV zRO^8LcS5wPWkZ(!!}kKFVD>^@a#Q!+FX~9#M7Ii^iJzGf+lZjNTUR?cM)|13SvQ`t0PbJ>Z}I@zg)y&&_01Z~myxloX%{ z4g?O8usHeB+*Pyq_$4(vM{rXwnMIPO`uirFX*NZYWXe?x?kEPlB3(>=>(Gv?#5SF5DZ1sH|M3cK zVU#3T>q{-KSD>OXxOo*v-d0H9#u7kALXfM?e!>sYut_h+iYZpCtR7m2pF2@AIl2MS zeyMxV$ds3aYDUE3pedF1&JgYE!36c(Nw+L$!0e`F)fyJ(D@69^25$m;}NT-&X z3r42Y<8+_gY<{1_KC9b5N?dkV9=#cR1n@9FA#0F%ZWnO`H@c-0la_sZyaqs{;TK%^ zE#kujW}@zPsmU;4oW6J^thl_|T8m~@ed*12XxrNZm?Q3ezv0_;#y)_kj5#Cp_Cae5 z_$4cMqGPJqC8xw7x@irMx@vC0BB{grTSw9917E86m$NJG5Ly|meN^xr@ zzfa2BA=|IEp42M-8w zyv$UE#Pt}N^eHK;{mIZ(;voM!S`cY;oz{I@+3h9}95c%Yv8AGx!(Zb~W>CM=i1QXv z3J+X>^Xo48WooQy@+3gF*UgyA3|D(Z-gul-L3M#S*dRKMOJjZ+uT4E#_ajN{skmF! z*K-A(uW9E(b>-DoQqd-|hdKxdGb1Fb!X;)2ps-e-%Oc~9PvW5t?8Mkn|M(#0C`T?? z<*=270j$-iZw!2~$)rai`wwAbjEOi54-I1 zqeT)VVZ40wSy1p~)(c??1z7WKE z36*ma?qdWGIDMJ)w_;@@;V-uI^$-Pd*c-gWldiD@ut0z|W+Vh$FrM`yl}3|fn&if} zYO9!29P{y} z`8Qz5&i-NVTNteRr4z;+{}wv6p`c_Ye9{>7vVST!Y_hAb&QLwtuwRL>gRDVgkA&M( zz7iljNYkVzAg!&}%&dEbZr+#3fT zWJ1M=ar~5Jo|5dJNnoqGRr{K|UQ#WhVhz}Z`LeTStI$*&T&-|>Y6^{z56xIFX!G>p zL2lKJ!{;P`!bA{*OPnM#xm#435$D`WwgL~bfKKbRDsW4+BZA~cjC0k-Q}lY0n>?|t zS6y#QVP@A)>fZVzV9(aShpNbZCT02s;Y>}&hK}1S6Q2j}H`~>S zd=voM&OmP~8?xNL}VY*4hw@Kyw>rC=W5w(tXB#OtSNax@;mKO zT|93SV-UKTQ;SRxtQ^*e`FI(Qa507C87i_*zt`;p2^N0r0dIYW?Nl@;)iH?2W7O0V z(g16R?=;%`+H!HZiEO6as*T34)mc=guuwiZ1^mrdOqVb-hHPG+r3n z9W`vQb(aVFF*tDnNc4DoDo#>M+l(F^5(k0JRsRcbUx9DGG^5A{s z#sa*0SHY>h6qkeTfIr@Z=N0_-R8&55-t>12U-|OY&L5G$G2YS}4Hm3868@4yaLbw-V+; zlwoHNQ_iak zg32z>A=ljTFnroVJE6j_^re;pY5h*04(!>LdIpK#$kW#eG${wEcqS8lU^mBk@(i#W-wQ^gJhaus6vK z)D$`^1~ELf3s~acYtJCJ@(;Dd9K6$bTmz$fXhH4*nkrob6$3bsd(%4l(u}RCM#0PmsS*V2LqAf zQ`=$kh47~8`h&nA&2om(n4Rpyr2ZC~YC-IpA!pVUr!=kAiN+sAr4 zbCp|D<*nd6-s;-+PRbGHXvNdH05;3}`%lbWNHIWHmcGGXt_y2Ka?GBF{+*L-$x3O{ zXdIY{^5Ty=I(VXH8yeaiDU*a{DNlu}v*w%_T>}(X z6}oGYDtTsv4r*cB<$vlHDI^kV_E@d`!DJedqgHO)91jJbl`C0A(c0Kdsn&}T zfP<_xKOWd~wdm01yBY()lSu3A1;(Xxg07o6fpUXi)i zw8DVpQ&A^i^b{qSO^tqy>NX0l(O_5wts7oCGi&2w-k_vpi11E$9}Bqv;$C_bEd^h;&~F#GsO)cBco&B z@Yu;3gsl|g9Gq^Uc_DwDuVnD4_gYf_ehl1&6D^lt+DTXomN8|3`eUg&+`ByvG?2;@ zX>y12YTX?=Gp>86H70m3dTB!OKE7%u{4Y{H(gSH_be3eP({=DY&P+3LPXid%ZLa`& zu|Uzu%pNvzAwRd@JB8KSQlh+iPV>KC@yeFz?qQnG-pfy#RIoJ#59V0=ok!L<;9uNY zpxFX8GQ*LYInw6e5jOdr<$2}xiX(E7s1cop{tsN*H%&v^Q%lGe*;Pz`^_XP4I63c< zSxH5p1ZV8pj@j*@@iidOfG-L8&YC@r;1dXqbE2kFOB||G+90d>`b9f7Gj>TDazw zXhHp1WxpQ*A{JmAP$Y%buR^oHAU%@pDkCKZ(8AJx!iv-plkD?=@B{w*)hR3YaB!6+ zmi`r83{mQV(AcSkhMTqnd1FJ@$bUJqetazol14DzD0N@dMpq{9M=4X}9SU<6{PqU~ z-bnGm5skEWyf5C9F8uE1U5suE)a-^TDO$pX40Jm4NX1qVR-)M0q%hI`3LXBoc0cMh z5^KSyFlnBuln^Xhyb$n@9dQXuaKuegKJ!}>W(aE_SU5~qjO%Amv}N|Xp6`(&LtVp_ zImEB?7bT){+1!z#fipvn>U2cLdV>eX&UHyBYVfqvR9J!D$JdWS9q!!gJAW~28}DxR zIFGZYZ35Qy3wcTf3#4@N^C`AJUG7ebVBm@jwX6`=0xzeh49H7>y zuvmqK&p{vLGufV!-YCv&dZL}pitpN`ayW3!6`F498JwIm1LP(iK%8m+1NoAfK?)WFfPIm>O4T!jXKV^|D*^SW_A1UXZ~VX?4w`tGa6=^X?;oNaVBz$g3{Y-eT{ zs6|+;`b8qj)XK^$rLX}rywAa(wM36Z9%jPEbPamg<9Rh66i9r8L2auki*jKqPB&DB zGf_h>DDhA`}%Q?%k5+?90&}kDAKvt zoc|^iNo^G$8ZhZbST*vDYaBPJ~d=? zX9|{ErD53HFi*$4$%>K$frG93QLtLKB^^R8jz0@1>lh5+EQ5tceQB2KDMw@}D+%1YQtw*(KwZ5jbMU%5x0E*mYRja^NEI(r~5IcXB8bqckq)T#yl#9wWHttVYaU1h1LIH zs3hF%09*IIdYgr2Cs@cpM(dmsQk{PXkABYb-r33G|dFhpe3Bv16f>5 zf!tq8S%p~tUXI8ZBO8cMXsbX1FCk$J{sNRU%|jg$iAl9vyB2e%<0TfW-d4*39mcg@ zWg$#VOEuXlqI{#U7+r8}i{~yMk1h#Nx;j0t1bW}P`@BogO1(P*AD%ciW)vlE=qi^z z)@xr!KkWXJU143^4E4x&7)%DW@@1L^-G?YwC#0Xi{cDqwvGG2ni_MyrBqvRtX9G~7 zHUl+;*cS@&0Br{2LEacjd!6XKX4m@c^lXZ~z0<()lOLiO8b^Jr(w~4(PEEAO}ul!loln-TReswv3#vejDSP4d}!#ZkU7 zRU|3M!|}fI?9GWLL$?-dL=c%w8v;Mb-3#^yR$PBa0<@*IfppPdr@WqsXAbb_UEgAU zgq6-v8E=jUZoG|sH#Jb^>}IY}`r9iZT3Hm`XN3lW@b_-vRG#Jrxkz`%RU{JG6e-hU zEv|F6tL?ugJb2dvam~x1H3rLB4@Uc+GYnU++-G61)`qp z3p9>Z>TR3CZN828r}OU(B-9jnHm%Ciq{bAOQJRS~tS0dxr|G3}*VM(X$iUgmzJA`jwP7 zljRR+Dob50d8c7F7-`U|Is1lHfA4(asQ=T84WqUzL>`qYA2<2Ew!i=qU5r;pl7ygM ztIzCL82Q+$4EC#)4V}J|G8vMv@R8&i7GX-|j3vIwCUok4BFjHj`?CO70%D03?@j(~OX^8( zA3G@CkgZLBHqLP~;e9|y9)r2u8}AW*$y`=|}a?+CVGP9%a!utO?K zM!Oe7iguSAnV5(_2K0-xo`VW2XMY?IhNa*X*vXvi3SoQ&%l$Z);&`9G{Q4bz&V<2rNiLxkFm|SPLCHG?mn3U0=NnnD{WmLm_ zWa$DFjUb1>22kkXwS7(no<=SErTPUmkjmzJtn7qQ}KH+sfdhCNKI1>yYPI z_sE!+%YwER7hOquD-M1+(|PYLjyeg~iFbwS<$*o1lkI@+g72}(OA8fXc!L$kF)m!v zT`|kRcr{&uhiz%mE&c9a`ZU0aLS)vK0o8RICwtN*L)H`^O~GDA!DOH@v&4kQyp(2P zxbZlaT7)r4s1b z^p;?)rvw0d5JIu^!;s z*6b*AJ>Ja?*XzLbXC&;R%I25yCj$rQNt>CYe5gvSDfIEqz9>5_!sT(i0>AQapXbW+ zD^jimu+zJo^4Y2~Igh@)3+iGcjqVDm;l{OY46OkB{-FK=DeWRwy^wjfW#+;`nW7U; z67!8k{mZbsm?=jId3 zUYKx*!ry`xnkzGbdFO2JL{gkDEH}>9-Y|t7v!Oek1hmmdk@xTU8@dy)@3GpNEJ{S7uaLA-{*)Y}TD#svz=r_tLs!Q~Siq;IzSM=x+V&~7Cvyw__x z_2nX`ZdjK=P&9B16OkSr+gOc=WX20RhEdMU-on+Ah?9$%=zrQmbJc7dvYF7nW@_(7 zv4SH=*Q4H6ID8%>muWUYYT>K0v<|BZ_jMKr1hF;4B;U!+`19VSkt|5&RsGNDor?59 zzvi+_Lylc|@c&BZA;4{2_NA0I*=Dyb`K}b?zVl6({uW-1R`g;ruvKnP#!Dv0$uZN% zdcC$Ja1J`BI-G>AXS!hMG96;Q2j7?rCD82yV%UeNF#>A^?DfOG?nNIj1XDX^^y6V5 z2|>w52n9rN4J8P{V4D(xsNW&jypE&(Wf$}j?1L{iD+Ck0K9dUGe4ZY*{9T~D-HbD@ zQoF8*C=;$(jy6UScX@TzUnCbVBl|cP_UMWxJUSPy!WH~LuO?b#4d>CDSnMu3VcrO| zbel{2j&bP;=qEM(3)!`j5z`$aNt|!yl~RUooCjtM<8ar4f}P(JvMDr8m*m?paj!Iy zbZ@p7tt6HJk)IHz-$5n)1e~v2!ymM+(?4^|Xs=M?a*u7ZGW(R(9rUY9z)LW@MgcE8 z4a~(-oGJ5LQqY6rOIJ!2itK!>6WeSY*PBvoex@=AkW0fnla}G`@XDn`bc^fZ+-(k^ z{L_w?L4gys-Eha!crR|T96d8Jlp#TC>>EOmR)?WiUqj;7Dy2w3ff&F2xg@Vspx;PH z{Z4%m`FzeKp}S2_yu1ygoSCLwunbjRT4#T|Ss7G)R4rt8u2a^X-T%+DKS~hq)>`PZ zYV*AZ*x|{;l`byXCV^^Qx~?LrknvEWZ$)0(%9^R$WmIcb+#?`R8|@{_`? zztE|uDvUPz*X0>xvO;$Y91}@QT$P~mURnJv5K%}2+?Ot*>b1|%Sj5PPuDXd72OfGj zE$tJQ0)yk(+9Eizz93Dry7!hX|8r()E7Rpxm-Qhklbi^ovdwE|OSx9;B-(N&;#qAT zJ9=j|jVU@@sCYS>GU_WP)r0w@$XXeIfs%r3g_tlw7l~U4V=n|6GMYvN%FcvnCDlPb z>SkKqUpgm2{}HOEOr{X+kuY$j(1uXhJgNlgK^*422H#qUnpFT5UVvRv2&%n@gBQ|m zf8K>7qcM&)fzwx?i~27lt$I5LsVqeBYU2jncS&$6~Qpq%*syv=j1+YnW0YNx~58o(Z%x zv)U=Uc0oP*cCazoaiux~rWd9aU9pdQ!tvLkBLP958CEbX-G0eoG}#Sj4_LJQV{|6bxAhxa9ox2T+qP}{2|DOFd1BigcWm3XZFYS7f8KlEd(Iel)TdP+_NY-+ z^S9QVdr!|JKTROl1%pCqTgtkqaso|sNE*vQ-yi_OJ!e?@=^0Jx#^vNL9|`i50er^+ ze2VnJE*DfZ$1j=y=K?EH-H+x)F+I5K_QfMlW_s>sX`Z?vxEW0HsbB{BQd|gDcQtc| zdL%=prFQF(fgy>6lj+|VbvT*#3(&5seWao$=G4+3-u>Z3bLPG}y=jP!6y0DW1 zax^!hHkeTYX0%S{f+hbP+8IuGVvUR1z&;jW&GO2UE~hr8cFMg!5y%t`PFvETqsrmL zvs+E$$0j8S$c$}~c46bhAPiI06_eBP{B8`njZzzyW0h{u^Y{643K+qT`GGYl;FCDl zNh`C!kPEYV6X~T}%&xZr>Ept7Tk>Naz)QtnVlnN8?06ST&weL2Hx*F!Q2J1tN6C`o za`)9-HKu?(7C>@c^!aevlu7(v2kJEqGhvgOmTm9((}JA=r?=L|AlbZ;Vh1=QRmJ-x zrM+}-X{Y=_w4ijxUlK&*+D zI`08D4HCJ6gqNT;muvw#8_=UQTnR;LMUQhBm@o|CeETl(=XchfcvC4dCs~9QPHF$o zdLn3ltE0P9<#&!|qkCQ-7i8&j*>Hw8J07G$j|v{l;4)GpXl`9yeKm@}r`yPa^?;if z<>%Qw&R||FPl)4QoCF%OxmHAfK>y@}4y53jYsv|In*g+M-sk{^a$o?#a?ijblC((K zV^e-yi~+V1qk@yY_8&YCH_x9AWF(kv7^6zHTEFLE-f_ua*@qMOmC)lz$mQce>rRVP zLS68JFvlw}cMs>A%3;eeYWs70LWpL{0_h~93qdOKJ_C#CM;NT`X>B0S0kWaM10Fr9 ztjKbqby$Vzx{kq1;1!YX8l)IFDrhRdf0{rM@rSV3kC5JVk!w+$EyikT&$7b)_~`vQ zd<*rW82(XIVOlo5E68C|&N`u>PGP3yX6xAo9|XaKk(9ScXGdY8LLOlXBX}3ui%969 zk^;x8n_cH#Hugl=2sm>ez~UFN9tGig1^y6+xJoK@y`%kaT>)r*AY!5TC(mADW}EH^ z{8Hh($WEc!5AT>3qCrg=hgy?$*@adu(8Q*MYQAB;&@y*4F2^uPk`V~gwz7hNw7xhO zTjAi&K54d*N8%VT56JROtP8kyq`6U%9*v1NeYr1Ly;h1%3CIyg!248})L?r{eQd68 zl|bZZ_$(VK6~oB0)RM8Jo<-(dwr?&U@(M4HP(9~ZW29>ATvqpmF9AIFehq&URz$_i z2r+7Z{f*{Bck~N(XLnD2WDbpk$i+$=V(HO)Pxg+;_3^Jwc;!&4zgInIO*MP2OrN0$ z#@RUCU+?iS`t1NvhyetJ_#|wr|EbSdY#UH;W^V5P(r1a5UVIiedf-g$Em>kFTeqq8 zQkrc`il>(DpQy!@=J(zpHW7*h(D}>D+8r}{H0kta+Y%i6$&;JScmi2M7>XOsM_Y4l zR(v9tFw`Gz(mzS{-HiDuf0J}9y=N&1@j{uzXq!BbrGQkJB@t=HL#*s1f;JTr&x}d8lTA{YdlPG4S(dBH~ zogk=ci(-c*lVX?L4#JV+E0ofDs0!Bjpk_Qt-U8Uiv)bw47us6_iBcP!Y!cSJu=Up&eGys_9$Lie8XB1}7O-ExX5IaIof3 z7^|>3x3>Wqy%)3JI(gEs(KIGex4JOS7R`q_ED5E)f?vi=$SH7D6`&)*Xq9=bX$i{R z^8sx)rqVd9M#|ITwUR2HK6btbO%vz+Rr*>=hI^L}nHQtr-R4VU2nByE%Y-?yHhS!` z29kL6MJLtkjvKYub468mbs+{XDAcgQJxB_W`j5~Hh2sSM{%qH2OEF!cbcjUkEwRbo zPdx7k^mMC=C=(FpQI|wVAuR-%PJ3b7=mJ85DaJy;JTuyM>12^u2}pcWJaXg+Z|4%U z@kGtKRF(C)S3PeIo@dQdvQl#3{dZbliYCGfC*L;^1adzL(@qS-v6bo(>?-JpY+G&; zo*Z6pl2W=glhv+ktv;#_SRghV2eqa#+~t#z;lm0@=4gH-gd3{(sfPqGk^_v0 zqeLjo>f+^0`kJN88Zt(C&BMH88}(elhJU$WN?%aSqZi{NUN#J{u#;hC(C@@whIY)n zv(g{cG;PeMoGQrmg)Bt@mIyCr(nvQms)bCX9%zOsbm?l~0_E$A9MBwmoO$^J=x$^; z+$XjcG)SjVr%bFUmya(gsV-|v4FR@&+a^e!H@f~yi^wRw3vjOke-VUU5xWDjIwvdd zh1fLARq?*eaR!3&s@JS8LM{7}3i2pPefJP)-%*ANM%!!E<(Mzp(1I8ULGM0bP?~=- zmfxJ$9Q>+Vq(PGg67gmQ>ay5-UMRaAy~9UrklqM`+vWD`rmT+yEB%FMQ>% z90~g;OWIQNOVas-S_m3?)NvXxNu2l(e0DaimD3G8D}8sAs8)j7&%{hyGWV|#^k$%x66zs8%`3~pB4K>der%Qrlk>YPGp2=<)gA3tBnvA~ z6Q$Q@{?IM0zgaw$ad)}h29vxPc1nl*#lYeQ<;{aQ$u{1mKnfZgkZ}-ZR9vKBdzMrn zvKQvf#_@gRycO`gjSO1}a(7G2z&MR5=>LudZdm*c-5vl^0rVsMUx|2DayYK*5n5Agd}E~1QRn`<(PF0y<=qsy7NP#O+r5? zEuyL?Hcpw`^;&-t;KgF%7jt9^yy5`X`e8FQ5O@yZmS7xR1~3uvS|r)NJ93?n4lW|9 z(gR)Y-md!UcJySg1s(uXe><^%*Ccf50D0gj(WO(!^ls7;3htUCeu@QYg zHH`|srg;*=gOTYmzTxE%#TC4=WKYzm0M@oj4nD6+rRh^Z;ZG@OACr~^JAT;U z^;8ix-pa1YW9iG$c>}AALH6sm&2~H(Nt=H_!N=L-Q0Y)+_|zt`TROiDc2;l-!)&N; zcDo6?f=!N9@Z{?voi_!rnV|O*tmh0L^tY5t{kvc3CKTMSNwQd0UFbkaYK6ynmVJrB zP?-kzSnDG>YBYMAu(2%4Dm(fX{qonJ5-#H}RJO3%Mh<|{2X*!)q@uutUqjD%=+g-RR7FYPJbDUOQi(iU;H6@7&K zZd#lX9zX@i^-0BaG+XEIi>l8vU;2oeH`IzUO1^=Vx#Uj53fk^6dxCOWL+XuAZd$8e z8Jl+*G1tdCVeA2-qxE=#n+u~r#uoWMn}D;~vN* zHVEFh-pYO-`JyfGvo`21NT=p(w#6=FS|mEI;KA*Omn1IcBw2p$JkKoMY~b(?!)IeI znjHd=4gyFSNHfV~okfIkQ0o0bpx4Hnkw)nYuUtUT7s@d!U6Q2k^~cs7^wquGRX6OY zF5gQp)$Nq(q;Xm{2qaF(k5&}@@vCKOoiR0t4x3>k4p9yhl%;k0;i2U@bLI)NQVU&^ z{uUz`I&oXg5dj}GslyVHlsdflH2joH)hqyYuCw#l!k>qnaaTcC4VFISGpa%XKa?nP z3+VH8)-pSN39!7{rdYYnJZcp~(USK_$uvp&$f2US$u?>?02Ea={+gGY(GGY&ee{NCQ#ZroS#furbk5H=3xOX zlL5^EUgu)5v3z@qqP#<|$EUNC?}PLu?)GWA*_K1*Y*Gc;geoPtH1_e+=j-eClc9H? zmS|HAI$fsWKDiN5S7}j+#<}Doq26>Kir@`b9D#-neJ;~ZxbCp9tHvUGw=hu1xJq(U z(DLfWVCLL%PLVV@coU;jThK`f6n+5w#vec(Ys)yBY0M6vWBOqhnX)YeB$~}a`uISd zd$~Q-qZU2$4(mQz>qPpmRP6oQAz^?x!h`J^_I>^Sb>$_|&#c+Yj(dKm0`!qP0?AbR zkfL3e`T1*I4n8WBnd7haKh8K$5;!)nX_7Y#z=%X;AIVgOwPyvYKV-z?s0aXUdos$s z%!pUzmQ+!%`8-6l6HZORgmk+FpJpHjid2~dHai&_d&7zeIlHiS@Gt?i|D(OP6ql0; zCO4!;RAu{{wUs@md}3u%N5>r4D>;?acYAV~V#}105mE)-iLDPrzb-z_Jd1ZxW+=EC zjdby;FM(@!KsH|wlaPn*GArQc0rz%ZU+%o zhm)jIkr4mmg4b%!3O3yK@FMg-Zn;1HK9r?$69nZ$!_X{R- zZ-g&omDk1IMh)8V|C>?l(QN~@a1enA?C?M%3OL{uy*xN4=l>zc9}NYk4Q7O{cdgq~ zV|Y4MqEzV1IupBK0w{OHIsL=(vqa#vtDs>=` zSig9+Wi*qX4Ef=j%pTXltYQX{INe(17~_$;JhmvvV8;4^za@b@>8WD^k>6mr%7w}& z+x=#B5Yc?LZJ?TfQ-rK={d=?|4t6O$?Z2K^Hf#1fUanA9uAYXze^q5%(oh61j$a{g zz0w16VX|RI^Eu;8T%n6qcu?O6yF8dP`aF1brlrgj1Px-Te`6N0l!R$-w5LmcZw>hB zVlDZ)v%RMJ=R8W+kIfw|so;MpiQ6k*vT9YQ2XFmd%IbD79sm@GraD2@?EjsZ? ziKYn5zsi=>qi#Z~H(B4wBXHH1ite5zGz5+{;b{&49^#1SV?UE$<8u8pKmC(oryo(D zC9G14ECORsb)j@S(!_V5-vNb0$ynlE(7p|4qV&g`3l9Uz%`DM{)t%TpQO1b(ei|*I zModl@p7-ItP7Js>t!Q00T922;^LCU#CB;&JCo8hH&t#|2*m=6TQONei{#DEO-yGIQ z*0W>${$oX<_m`*wPWLHr&izZC!0a!xRGU}m!P0REU?!{0|M0^=8n$uZ1id3T4;S-) zO{{-34gO<^1OCerInf7(Ul5CF`O@KaSqZ@5grQ&4Upc zMZPwlfEAd>WnmYhgwKU*FNiD3$?Z$fY3f4!5q=A6fGGPgd5N%*abmGGKs>KVNY5=H zqOzvFo3^RCrq305?)44Iz%A9+^UKaf0B__3@?L(bd))}A(g7BJA zr;akklOXT^^{FGWoGyvHL4cNM3;#6RBKba6`A~%La-L;RX!`;cJxRm=I3QqhNZ*1| zEa2O#Elu)1ZY?}}6{kkX0o7nJr?T{K5{?#Jdr4+k9)}U7;D3Rr)i8)SY-h&Ws#@N8 zwwj4kD14<_X`KHWTZ-^#9063)F&a=ySIN=YA`<+>^r(B=F5n`Y?(Oxau;8AF3?)w=L^uA|C5vd|3(we9qYkhsB^PT!IoQxjxB&&Q#U3d8IDW6o zrz}xEsK~cztDpF;=;lRz);*9^T&hCIv|g2d7@Bs?w)DJXRJA?)i+z#vkT!??8bl)LRT;CNK!(e#a;z*dX&3% zwA1%^TBAg)i+s}Ae8i$5t#K0ymZhZ3-thiryE2J$|Ehi@(YgOEBI+S5;apYL(y#7P zF)%5WRAgR%gDU2sn4M>7=1Zbd{As9;gi3o1^fvoht`7Fn3xHUP4ARJ5;dnoL>}z)h zvc~r0W=77HYB$T;h8pvk`oU0YrIGQQ?Uq+jPh-Wk>1@Cz_Pko-l?Q(n>yG(Ka`*uY zk;ey7oScgzm8ANuu!H;B)JC6>dtY2BdUW$O|9i0Ko}RC=Ma>-xl_|bt|L|7=_?=td9842aGO)FNg~+b7pYWshLhlQ zjQlJE>FLqW>`j_~*NsUyaSb-7K4~dVseFM$SSoPEis^IhJwJxDYSJ*{P}@4}IBQHG zLfLsg0URcPGZCrQh$m+!S^Um0yLWWj4!##2Vw zL3cIKyr~}W=c<&*vX)Y4hEy$`x{lVGS=u{PQN``NLL_-NB#61(YFIY}TSMFRYN;oO z6{wM)(Fmx%Y~LD=wOMZRG`S&q(^&IM{1nNyth#?;KL0)BSn-7Rz;UVJv6s_sq{EfF zb4$t!olr0w^k+J0#!wQa*rnhXT4K6$vRd`8ThTvQQJ<=<`E?H}m0+<;%UPlyzdiEb zScW{YU@+Cv!BBVxC8GL!{LHpi9j>78zLPv$Z)X6pQ$J|ZnN|!gq(CW|sv7GEjT=B6 zm-9l+A+Td+LXkFqxZJWs7$94BQK=W#1(B55CRAqa+L$mGZ2HlD>UO|8cT(v;25fCY z6{iZ?2LP~fBL~6C zamfSaGA45*ME3W7Sdo^(k7SqE4>8RL!mh@cOo1&Dsrd!7!d8z-xstu**q4XlN`xu8 zl`7ye!mw!VVsq^+3X+2>aFpt1OYt_dL_GrLcxfa@`)pPiSD`HvlSOB#9O{r&XH?-n zZ|87rGD~qy1(ST`H_Z-a_OC;nHO0%HE8;~1k|$SH@2|l{;I;iBF_eXkpB3FhiEpsY z?u?4Kl&w%bj_NYZ6a3{5PidMYZ$TTaj7E)yg1?N+AbP@od_?`#mx6h^g1JL{MNtK$ zbqd4ewVTUFs=bqN^vNu0V-s4$tTtuao)!nO@Stg<5B2bf{nN|i3L86zHW6Nvxkw|G zZ*sN_>*(E_w`qg^2*%Tk#)F)C!HSE7;Sk1MlYU#Wl~HlRBu*J)*zv5h8|M53A@3z^ zMm@E8;Xc!w%QE_eSibB=fnybraRu=5`{}Ck1_{mA+Wf_%1rE{aZ+`{hn*$HBk;H^U z2P}cpT3P`|`1Z!jf>@paQwQx?Br#I|Dchl|Siyc229LeaBLuEY$`q+iBb+`=Kmc0N z{<`D51f6tPEj>xW6dDQ@SdL`mCzFFwPM~ni-L}UPsYf;(ludsMq0FRE#UqD)hT@?N z?=P;X`7rE@O4aJf=}548o_aAqPjk||=(@tH(0&{xkh0>=S8%Md99@anVjcpB=EVme0DNmj& zkE@29(spJT8OcKb(TCoeSIzr+Qzlt0rltxcb=Z> zq_wmRxYYBbdeC?Zcb-$#6D7^|0$&e?zt)Sl;B_S+8N! z@>p%(dYbT5^DtR!-qN0NFeY(tqwf4uf@PQzu%^+K5+qzzA5S`>j!_4g9us;YRgaUP z9+JJ3$Sv^%P0T(H9rdtH!;jWSKy{K<>_lNAwPQx}M|nG?fDnPhiu!2)1C3QrNskUx9Q{C)ejo4WoDzfO7huf``2C)JO=?~O(lcZ`lY=k^wZq=?F))W=u!9DgU z-kIxU<0`D>2MkFp4Eeyz7HYf|lKfBU*s$9rRR1sN zIAKLtm&A2W_qTpC-Ivs}S>j9utFMWTnW@yVO&AUmD3+U zyBtea=zW+~{-UkcFy+VRQua*O;KX$TqD|l;$V1U--~0pslst%T`6BjbBwd|gFe(Tr zVuX2E zjp%9Pp?Cv;U1FO{@ih0MQ{g{7<7GBHd8mv#D)MA!kO?$PfHnb5`c~W`+XVQy01BEs zZ<^VQUmUc_Cj{SL9ftpN#7{v(vPeeaVe=rl4FlqeM zAUN2K5sg$s5St>5flrp*GqtWWq(iQxJ{-ai2>2)KvAhu#O}7Z{wdC?}t6$%(GA-B^~(;oiTY=U{{p(2bORVCH}GAg8Jv^jKO`pw{MOiT{FiJ5ylCCdNm^+Z z*AUjF8R!~2F@V|kwu4HXH-qB(DX=!ZzpHc)dL9>UDHAuNpGTe~pff;nNYlztQVv4%)v?$~qS-WazjUW;6o%?6={wZmW%?#tJ$ z%F!XziCV1!*!Po|64YnyuZ-_-X3{P!UjqkmV#n^rczZ{5L-<1fR%O0`j7yDgGeJy4 zm@DIr)U`s*m*fBX2J7CY*Kn+4epBy9sz>FwI8TCORZB6trT*iSv(rDAs z;7T+q>(H!JygARBhzX#zbzIrtnme73h`a%rn5fKYN}4d~85I3eMJZ<=ONU;$xHi2WeVAB)^>;7DXZgnOi#BP zbIaAL{TLFDMG@g-X6ME=`V;i4XzibeRQz8*Zi-?iBnvk&D=SbRhX{C11P5$}Ljr2? zQULWt;ZT?)Isa=!A!a64B^DHf12UtNgK_+K`l0Zj%)jfVc6UxxQ$O%mG2CHNzFz)l zNd+=wI|K1*)9=Ale3>|}hYLP*ij3s8F-mEsQVD_S5kIjhlqLy?PfsyRy0Db~H&{=Y z{Unnqwn6lPWSK?mrnc9E^Vydgr&KPl7xjOcx}s}TD&8Rbu_);<_jH@&1jn0S{U#uT z{3x3dX~@inn_zRG`y-wC5A(aL78T7$V~=mWr<1})NZp(ONCwo%Jc03`6pZi97lpSEJuiQ z5?mi9Aa%($cd^nD+6kGQtked7<5qa<`B?pTdqa)rmr(Yw*>chpodz!au)b%;gIxhz-(Q z7lAjB#7Vgxtp(@IZK7_R!&EzX?rW-t#Y7eEmlgGsb1tuIP2A(A@J%!(n{Z1j8AW%vVDy?b9#83W?$^1az8NSa9XEXYtq z=&TrPDE^fEL4z5&Ane0G`xgO@!u-7C1f(+o`MvVDbBk)^#|UVS9LnS|+>awHRCO*Q z1|MOzx`4~K)T#NRDBV1?v%?Jh0!IGDs&uiJmkxn>Zk;H#or41 z{NwoO(yCI&D^S?0vh&Qh^q^m5);G!N=BCU1HxresH+iU2SS>Ys_4Fe?jG=KSUL}>+ zA2sY3SqG#0NuiEfnx%mB_Pm`NM={K}fKbMKn%YSZx@xYR3e!+YKVHI@V~~m&Dv${| zUS+V-JgA}tW7X=FlBM!hrNo?P2_lKtAvI*l3Nb&thX7k80k6dMu!j7!>t{w+=!UOc zaYhm9%xk=k2--@INfO98M&R58e!~eq+|J%;G?KRoa2_keIFOHF<+_CHZg%M1wTl3R z^NsMf#(HBNBinK|PDlExM7|DEK&ZxSf;IIM+18kC)%*KOIczIvp?g&Za%ym(*v{R}o`sbhWH?Msq6z;5<%p{;YxbuiVBWhlPVeaEn>=uDrq zt1z#BB|#(&-%0tl(2`QMqYiOvWx9}b~WN@)-hm~5!=|J^46 zq8ODDbFmP!u=4yD`~HEDhnR(p_5UrnnEw?V|BRbTV}`CUC@5j$x?Uj|gr1EX=YQTU zC>szIl?GTxNB2LL1my(&Fe*j;|6NA45zkl#6co16!Xyd;B96S-2Pha$^Miw%<-flT zD={Y*7fTw8J}5c_7Z(pVFd>`<;0~#deu>JtMZ8rXBt0^}MKaGY)j(f2$Ve~8xP?es z;!dL)0tPovLkuI!g*p!=EAx4J;&t0Eb?cSpz5dqr~Y zjGaV8O-zocJg4FZgAN7*`wIpJ78cFf8CxvTF`?%Q0>2dHCl)#q@C!K{00#ypy1lM0 zGOAaIgb(Gtdj$m&1roGBFAQr#LK0M$kg&KT&Mk_Y4|gBT3*rs|g1pNdi3J(~w>dxr z>ftK6z5VcF1cH^02a=qWwEKRAr0g0&oNtVRkPj}(E4Jz61QEIkBI(SG9`g7t1;eOg zMLsX4t3121q6KnzO#vbW;2v{(0QE1BwhAH{gb%q5<^~B&L70VX6Z=Z06z&CO_6+y^ zgv9P6NbDwtfN-{R4hF=pP}0*82r=+6d;Q|5vZRx;27Ig2<14>CsTi6rI%F#Y{4Gztyj0lv0_ z8d!z|R!)ZlBHV)URWSn}6XZ+W1<{%B*kgD0oa08wm|#~I-_;?q3wuK^vj zV;E@Z&y*X7qOa+-lX#fcFL@ZQ-Mo7Dy_Wefkk+r??sOz9B)3;sx8Hgy6(X!>UF~L-fnp_pE*)+6aLMv1LSY8sOWdt783UA zlV_;+*7!XaL|aZCz_0=ee1ISjF|#h68|0^O&|tlrX3h)ja6_i{2=R2q-qb z&jna30~5?Hj1;$4f74!9NFWI3kn_e9qaFV z;`biR2i3EZl1xzk$klQ1r;~hz(C9a>%vT7CGqe-)Jm~q+9@PCempk6f#KzX=OVpRr zEl~iNADJUK8Nlwf`4=D$;6w#8hzv%Vp7Ar(g4DoJqB(nsmgVn#ZTx7!u{PJl@v+nt zv2GUtkhE**+%t)2B$`caf#9AvU0dlwKsVi>U*MT}?qZfngxA>t)7cbY&fWYKXY8BY zeq(kPH-BMB=`R^*HWb@SHzT~jtTjpb_zaC7mb;xq{x*be)h&Pf`Srn@`^eUMh6FH*AO}P<&h& zbE1Fj1Pv5kW4Patz}@SN-a3OheoQE4ON_?JQZPS5G$oB8HKvi0&v(c4Oql2H9o*A>)*-%nt~wL7DcR z+*!F~DQnR+A$L%l@shs$M)mmcVJ^!JsQ%$&9#t}N+R8NsX)b}=io-Ec;CIm%Cqxt} zc`dD&SL8px zMtdw7O|Q#Tn0D!-cvWIuaDs%bHDaR@ab=qer&6hVkFLQ|%NO}X=5C@wKy)e(07mL^ zTpkepR;GzgWERqB=xnPRdsXsnPsQ{MK+{~A=(0Zfts`2R%o2V9v2)Zb5N=UZR_CTa z?`n6^Q~jefF2L@4(=%+2yogr#t4$Tj=?tLkp0RE5`u@j1z2z?Wp9Z`2I_ z=&4NKDr+`{bwvLwwGK)yz&}PP@iEYBi24SVWp=@5{vTVZ|NZ=`a!JJGv_OHuI@#@5&?J~Ei=z} zpI>biPl-6oDfLtqDWWpFO1tOr>!EpFK%;a@%3og0^wXOs<8yUfMw{#aC=KtM6c)~z|mF@ z+_Nf9Z#tlfg8 zd}g{(0L`6>LDVI$Zj0k@UU5r(=zmK|O?TM%^BUiaN$5VkW3f~fpa7h(6jT_wpGMtP zyVxKw8*!az&hZQ{2nu6R-{U6%uzo1>cgx%x1U&k|dz+KlUN zQ)2I|Ig68Q^z3^p0A(#@;bb|M$p7Asa?6E5@nr%II?~m!L4Wj5W(IH8PMei?Y^5az zQbv~3+TrG$L?j#HV#z8MMtkbI5k^ktFm4;$e` z{8MX&;_&rc{nhD5>mbaSjTGjqIp~9kF#7Qn6}u9NK_uV>;5T)^P_KM@R=7VljDJ|ec#B9=f(_mh)k*}u* zo<4zcD}(gH;;s;0-NPo{%QqCmU&kt~`EZJQiB}1!=syP4sS|ct%POAA3NSZY0Xq`W z1U_YNCjtX6F5~5hLGy)-9STQ-=k-=~ly?CHJHhd%r~OByLolRm1dV|HEh%gPH^nXA;+(|<#) zuPoMQgm~szPnFTW;McIvt=-e?=OH*2geP7vuAO|m=zp024OJ38Fzf=CJc1{SF7=u3 za~(=$0XJJ8OZ*CUOJ|qe<8aga$)i$aJBsQYrZLXG2-ksdPr_E^>5=v$7-?OvzM6P3 ze!ZpbOV{j)$i@1l;qhj$_p*VtuzHDkTr+lvBl&>z3B3mw_m1YbLp60heZ8_KGKC{C z)|}~AgQP-d#`}{#>1^Yk!Q3gLUDs{6j`v+DfSq8neOqTydNrP6)mXZ`V{sEJeoYpc z!r%-a zjuz=5n0B7697HetE3d*|UPVu-;nC=CR;(HL^yo?i+QD#j6?_(TUOYrhBRrb4HHZ9n zfQ&Cy41j~n(|nk+)@zD~IL=XU(6`=ceTMW4WH&Y~=iR3PCp#QY_-~pAxr=f04pB|& zTNIDi%4yODOi!v?K3$FCiXVA1a(A!WOkkhHFCD>E_{nVdW_jJ0E_e~;O@JwUFAF*! z@_x}aI6%p$nKkQB8b1wT2jxUB<2pqYz$)#uY~5=%6e&$JOMIH8$ena(t;DRM&@NZ^ zJAH+mcDAtPn67{nMEy$b;?u6!C3w_5I=IG_Y@uy2+{fuNVa=cmt;O$gMeO zaX?Rl8QW*D$bqZ#m$8-$!$=wJz^X7EsYN+vJmZMSEfSgx#rX4rzy;`908G>~^{_=j2&Kr;lCNNqVaig}eN~IK;1AYoEygv~ z*?gx6!^@3>F)thXqmAWMV~5B^kHcF}pgdcs*9(+LJ-J?d@x-qPL2+mgtm}Z{!5@qX zUA{RPP-l65xP-j;X@zZ=9FiZVK2fwRE?c+}HmS|cKxDPg_X_u`&%(1NfDD(L>C@!9 zB+0wF^6*Ri=>gwqF&jJelomi=f|JOmQxbVqELto>Cm#otVZ|5sj;J98D1keJq{{J; z>}d_q8-cT#A4ASt_v86HFNHUeev3okBiVtmrfW@`(?h(z{PE(THY3xj69oBJu1*`} z;uy{B{J3n$uxtL8sJ)d>CK8dCNN`GYAeAr4MkPtcXsj26A+C@1eZ|)K_3tcM12k_YZrB*HP~wY;Dbek6hZ4Ix6cxell&>BWfBsd&C5qnrX?E^-@JBaI zLvc#s2YOfP)Jd4kazf1G8<$+D*ucz%>OxG7HIl>JNjCvb=(15kH#*SWh#Wik?YEZ4 zCDK|uQX`@m0h|UZV2|R@c(6{|M#j!Cbbo+ea%b1hL{W^N#&3#DCr>COd`(Z4_?fCZ z!MWd=Bo%R_?)=n+@CA(~`sfm6otpmagM^sNNlxn>(OH!NeU8nxj~xqX1Wr<-9L%dY z(`Ca#D;}O4NlscqDUFQ=jd# z9J{_-{0Mn6503y*y+^sX2FSzm zPsO*F{I5VC?xBesJ}`n$LbJ%}g9sR{lE>)YCR9f00K6~b0)ieQ5r6DU|GH2%zoS0F z88C{sm5S_9W%NVfIYRh8IX>&3A14jqnhT|!Ka1Nz`u-x3B=@$it)8&BFf377rz;dh zNT#jNhsPd5Mdlu->gQ8zl6#99Y=pq-ZZ7kBQ9a$N4W|B`JZ${*jcr1e;<;u-Rh~vn ze}UrE3K+2NUS_d1J0190zN6Y(Q#L1%#r|2$!-)oB?|p0qEURteV0CFk*Z<>^TAMNw zauqAj*6w3&^`pMp#CpU&dyl-quSd<=$b`88t1%FP7#7}R6qTH={5KxQV?j6wzkK8?_2yagiI5PZE&p%!W z+QNTVc3a^`o}Xd0*q^n|Bq4jmYN}d$)r+@ryE@0$Wl>&DwYrsGKAqUMA$Oo`H5i!e zw=t{lT_V*zY_v+Eh;-pF2(!fpkgDd;g`olZ+6#i)hfhIlL!Ur3;Du|y>sN=WN9C~}K zu6utQu6&C$w~=u$?k5R88AoW_6v^iwE#S~dtU%@{R9r(IjB-+=Hn_PPRz^hhHOb+~ zOGmi+Zi$DsbYu$b?O_?Vfu#$3RXfv321sB0Hi)N7Coj|b$*<&q2AiNF6{JZkU7rhG z0=JHOa?2E7iRs*LRL^?;8^@UvVc8}=o7xhsD;?9A;|@;%t{a@iJ&`RDk4EijvwXp~ zIoXqGN6+>G%x;y1m4QiIfFLOkqNRDz2YNE;x|x-OMz;by597 zsQ|4tQec(dWJAVq=$m8>P3^1C^!JwZIu@%uRWs6oNUp-Qtg_%6JH)$TKK(c~rlQ)9 zsvAXxID@NM+3t%2dygq)H+#0K4uI&HJ2qqfi8BU9na@7*mWf@GCcV!4f~GGU$WL_W zQ>K8>K}@$$d1pnb)%XoT=P|y-mc_)zm#)7)g9jUIO3KD4$O0;qeIUf5XfNZ?B~$xr zXVf$yy||8BlSQac#7N1PVEs{~)~!u6GyX|oJq6YxdE6jZ$4V}?juFXy4p6MnbxP!L z*;(=x&o;Csd4!1gI##>}O=#=Zly#zpQHE4L%%iUPpu@PYuer21SdN9~gq0dT#EU2Z z5fQ)qr$shxzOUrXjI~<%J#h(|i(wjHP<13*+IABX&C+7L;f*zPJI3#a(2^6s@5sdQ z@GJQg>ate@u9D|smS82;Pk@Ec^)45+2Ads@hK3&oGv#s045M*O5ga)OdX#=-Z6&a~ zwF6}kFL-yYl2ZkYXxY_zZ%ufpu)Ty=15D|627u7tU?Beo!@Y3( zIs?1Cv5s)yIS$pr5I`tqDfv%;+vL-BEJ2cQWyaJr432!QG@P_J-kh+*oJPE z2FU`|Yr{>3Dr4fxH9%lSj;__p$E~3Pz8N$KT$gMgH#OW7%k$$iJr&rX$EXsUQy(VyTkNvi z7wdx9F`uR%dGJPDpmx$B8^F2S1cA~A$Wp^)+=uGwplq4u0zeha6NT2m9#t$F*@gd- z(v`jzH);b%*Of>bbCun|8n%-zbi0K8zKP9Y=RXb^JXlGN2{A=MOcA!LTsV{@ENXid zjvJ`L@}A>2vYI*^@PjR^v>eety1un_-)ZWvU2Tp~S!0qR1!nRM}#=H;~1vJdB zyCZ1}u@JUp0qFGQIV=C|Ia#hTNdH{PG3L&@^%e2>N?@8)hNkKPw`VH|olfN2hZ(^pX06 zJcSqsq7GWU@M4hFM9}l;RxaAM)K`U=l0^jK7uC^-0H&$Te||^0$!X*)ju}qJUVnxW z9zb>97zfEmRWENhMsPwD^^qI>KLAWXv%l`8nBPAa>}O=IWWA2R?(h5r4{ zs}M_;;a+Y+&7_aYDf}2p_w18V%3d7;TiEl9W-;AesQY<~{=&heEff3Ct3P(m3rwDS zq_#EQpr3pceB9H2l`YH!wdDt@9IABgcB`u;`1A*jH1vT&9wqL-5g#xmAK5;^r{1iW z2r@%fu3tY-l5qcVG81T`8o&4??AfPaXs}o1kNt+6I7yK8s$K9KY1g|4c`@(&Q(JhE zF;?N!AYBk_oC9AbUnNj#V&lG*wu;YJ&CO)j==}%*zqZ|fc(^s~T@qbbfy$(NAg4Cx zS8Z<31WvSd%+{}qKfq-9^Ymf{?`E*kXlNT;P!;}af;;SrW5EjrWN`B63oU94h!O*E zx$rw})03F`go0S44(QZvh!E`|_=bO9p%YB!N=~w(FaD$@MADAs=A$UY0F5L6{oC`R zHM{B2cm2J8rV0W^2D^s`jVv(~XyqH}wd&xfvQE+44<^1;o6;iAQpP@6JD<+3EFqV? z^hHkLcF(Wwh?gviq@eS!SF{qJy=D<=ogR!?ln`yvzSjAcwM+-(k>?3h)o2h4y-)h^ z)4e{gw~uIj-}4-vWfi4$4Xw#AY_EB7RVT6QE6=ol0i5A_NRIdiDet=RyJ@}0&`wAe z1*9(ze`LnaMMMza+=HyHdog>xR4eUU7Xf|9;CXVbzn$Zg&Qs=UjVE@3rg|a1k)%-N zw;{5On&Wsg`cmzCc>~n+rlau2rSEh3%cke)`Xv#}x_*{VPYAz_zJUD*A=96l1v0Ul z;g0%$l;tQ%F1CvoIRrd@o2R{Zh}77(7t_DY^W_uBYfWKydWTaPm3Td>EpJPG68Cw%-ca_8K28Z7t0-SFB8G z48jzOM@d~O?~C6Z%o=|#%Nn1&*SNaeFPAQVvm8Nd6b#NA(NkJmc(_!3p`q7tIGi11 ziuJr%jttG!M}9OxzH*QnCMrkG9$4;t6@0#!A$nwpV=MRa)A|Q|@)I#e=m#e%m3d!m z=v}<&xw-j#$Hr_BB6&GoW6UXkzV^_095up1|T14{WeR@hW)pB)}z*c7E*DAov^=x?VD2aF z>nzJh^7$>(Z{_iJDiX)5CF=VU)wNmQM-y7efufcG26sAzi0@)*wJ(awFlFn1?&LfT zvD9G5emCQ?KY&pB3cHqg1R%8AZK4==)my{Tifj_`TdEIf5q(qk`NMYcl&+2mXiP&J zyVN|rc#hMrCkWQR@TBR^*pB&OJ;N;ely5YO@qGq{x~oWFeO!1Kqn;v;LmG9*=Lmg|Mm7_qc}1LFb)6WFw#fF7&1`|Oy+M0HRRm9*If`~%$z)-4~E z-W%aeW1$|~5(dNom+s{=jOlJSS2IKgE`1I{b@HQpWz?}rQOG@$Ue#ur!c?O)ff@)$s17x9HezkqLY}*+;sHuI zt=IT5`Co0x+NV4f(X(IDM`y1FRt>$OC(P;blWBS&j$5B*7j{208qHuxj*Twe$~4W^ z$<7d4aKH=Q$gch^>*ZiEs?`Eda^?>G=uBl?!?V~)u>Ns$jZGVW+hilR;_1h|>4)f# zcLfSpr*ZPj-B_MUw>8c^<7;xzY%U8tYEu)57T{>%G2#cn2OQWa?z)c=Yku>X^_k98 zKuTTTm{Fh6ID~RpyNa6VP25N~<<_nS#+RUT?Sy(-qSMgV(;SAtu}Z}jMb^~ zcSU!)!`z%JRbU~jE~EQ4pmv1s1ZcQStNDKb+0htwlh&juvuB2b9+N%56|?`OJ{yxv z-4(M^!1^4Mk@*#~Vcjqsmu4IR6|<)Kz#^BIIsp~8dmI5A9+wMT0Ts8bIsslGmyve? z7q=l?0dX6bcas4Yx2Sgk&l{I>oBV8YdslTYM-VF`D-#PJP)$np z3o8pRCy<4S9iEa>+!pt75(t+g3Y!P*RD?*gI)S~xfZZU5~6%^d8_t^dWz zh3Ovx_U<5O*T2LpoE_|dUnJE%UoN&wpD?@T{yrb89nKpee}G+8&kiHGw_RpImp=_sOn&6Vt@Z1l!=|S zt=In-_df_(SChXwh}v8JRgUG~A!`>YYfq55vbF19sVz)w|4Q^vTnqH?xqJbcTf5o) zyB_pckG~T&cd)nh`ky2Jw(-w_%(`L{>N2AA|6PE8(vtRO4(8VOmOwSvzgn6&oBucP zPg>c;`rq60A2R>hCLrto^?$#ZxH?;V0`*v!{G_3(OdY0}7B50Ss7Inrdtmwj2_-SVaJfa|E8}X77&&aB%GRSd1 zLg$u~rk!_dp@Z>ChJPYCj1e7*&kM5G9#cLgG985skv1yU!1>b!rfY~Nie;f&|G;bD4Kmk-N@i@gztP#AQZ+U!N%fHxk6LHB85_6sYclQ>2Y z(OhZ%{>dlnX-uvpAH_ya1s@`}r=y;s_>S&lpNn5^nJuR4~yl z=yl>OhAyDij(;CajiYNKfayhZULgPf>ehf9qqcqYo17KV*p`tS19aEYs{kgy zs7+d|q+#5=4Mk1myS2;Q%ML&O)SvjezQx=F+^}@;Zu3xr(Y7}4^o^fS180Lz+$2h<`Bl2=V0tY&WC#GPtm`H=F z;~LNSnWR)qt0(%9=dV|P1ls}Vz@0bTgx$UA^51u&$T(DM)%35vhyL~~^!NB>)q?Ux zWq-K5Z)4@O04ibfOA_&37kTVz$?`@5H~-C}8eHn{Q~5`4L$*`DoclzCkV@xEV9fNz zOeL+OIq-d}D;MwKLI;RZFuNqxW|G;F^}dVDYh+sh#;a6jNe(q}-=UNnv^Ej77_YXB zdL$Ivm4J=-m^>n9*pEc99sxkFErBbCs((7MnFob=8>fvxEQ&yl1{mHDyilViV{Qn4 zFFt7p_)T%T?)a|&Tml$vH|`=)-7ONLuV@r-o|)Ll!Un&WDX4}?oZBkHb83kPnh6d_ zphpPtaZ_;t*;(Ec-N`OSgOaPJg0~t>nV)R0jAYow+uFKUYiYrTqQXv{-!0%wR)0=l z+m$tWTFSx`E86n)jrUW1p8%Z3VwO=G_wKHw`CYZ>r@~c4=XwdUrkqTWNjKRH zf3Bh+tE)e0C=63*&Apv-BD%Ri%76A3FNr4j8ESOlHpCt6?^J*?8YQ|N`Yg~l6RnF} zh1&d8`&6x?{kt8yP8~C3KEY+_*ECes4ALJfvc5dKRx@%CIu=#>rl0|FVze%t+}oe= zo>a5#1rr8MFOmIAd08fPAL4rFW*% zUV1jP$NuPC@W^9DOUtce?Q^q-;D8?|(hlIeVVjd^D!2t-!asb}mg$DyJ(+*n?V;~4 zG&vnKd67$9p+2|qCtRYeE`N7#W({;ZZW>fNqv0(Gu=v7sM%k2fofpWs+RE>W{g`~B z%qavLV)llmn6J@CP>0)wA))n*ucx@iodHR~;M;|rju}Z%((A_(JJo`7sT)4wZ?oI< zp}*6TrNhb|Z-5=jUj@)AJM?42oS4JFbS25?KY)=zTk(o* z&}QNFXa1)@^a7D z+W>;*CaD%g0hw8C0zNp($XXWEkSl$Eu3Lj5dt zDKcEX&d}Rj>b^`RBCsDF8wofXep#09f%H}9K7 z1Bp!W>=6cT(1jI|}5uc#a?_Lycj;9R>h6 z(Rj`z3v;M-Nw(q7!r+EE@O`X;?5ywP0X2b{Dz5k8%ICjd5V}dob*jLB3fGa$3mwYN zb}Y4s)rw?};0wZTsn08Ht!jhkobEss(If7#C1TJW#DA0F523{)rfEJ>fzP-_lqsrY z?`~_$pO#vb%3R-HxY9i}25%{%1bHH)gSp6@(}l^g4#UXT+u{ivL;q2d^0bU`C0B?U zIF4S;KeUadA~h^Bi?=7vgj&d8c$jMg?QqCq2;G_dnE!)O1vUjI{$Y4lMI?3uKeQC1 zdSh^L$$#&&-WB_&+jN%~{l2Z#mqXO(ry0z039IWEn}${?3ahfmly=qyJoAx`9EeDz2wL#-tZn6;?4#9|`a zEdc-IG54G54q1kIXvMeB7}1t++6wpt<<(wY$Dy7d~4tBFUht3jCd& zM}O*{2-rLvZM7OZV;#u2{DEinChv_eY~MtmWk-5tcw^}2X?4XkLJrIu*#s653H{{w zrHtDSIO)NP<_~eBLy0F@f`SSQ-i$H$m&3R2kQZ@MD^jhSrUZR)DL&bSat9&&XEq`-? z{zU__t$?#=KRpjJ1yo+RK2nwz=vQg(F4Ft6U$#bbzbvbll9)6nBTR+p^oq6siiCOF zP&e|=N=jg*pd}7jp#n5Tk@6es0708bO-w>vyF^ubtUPS@ErSRxSNt6f; zm;Iu_tf-9wSBpaDn-tr95%mLRH-CCzc7~z3!}z$)Pdh&ZN_9G{?MV=?2h>ZVgAvK) zkDAtD)4&;*sE8*(TBtOcbAo#ifwFSgpN}c`iWc0IWKHge#eMSz*%TU0N3#S{w1z*2 zlkcv!vLQ*y=_19q!BYIk(A>I#!@0%yP1dbgOVG;()zpzHA!~hz?}Cf)b$@%B*gS@-FGa49@4c*w$!nHs;}7LoZwB>_#PNrOPLn_HelatSkTm-)c~ghK~%Sapk6(+LV8+kdk>5C<&B5R(Q7iyLNEjAc@@Q&SGTw{GM|eviiNr4* zVC07xYlB=?fTDd5n6rWtmw%ljy2XZ$Eer8>YbtQ5ch3>}_!fM{0wGXNRuYgGB*D^n zL=sh%h0RQfLJsSr^c4*Zb|vX`mOWBhAs;r9l$(RM1w$uN1ht>lD7(&#!$D@tz?ZFW z7;4{?@*Lyvn?eTv0vYn|oB$ljlXf4F{KEl`0hNT9<>Fv1f;zBH-G8q%$y&ui)D$t8 z)eYNPU;19q2--5MIl&@ffxy!jSgcbpE$V%FM|I{~G3nLQT~+ept=@bx6&yz$EU;t2 z4I3+hs*5P_B5?>Y%yf_?W>eQC{&f`w8@cxozG2c>wWQIob!XBN={xO`No+JhXujsaY2};$=Wc|52}+XDh<}yHo6UbJ&gYWw3CZ%# zNOmUw<(EFu=VV}fG@D~yK$-`tF(Jd$|K!blBAJrpmFWGDn zI_-Ai6wYe(AAjQd+6SVDyg-o#Cj$B2`W!F+NDe&WcnmqHqukGZZ^}qoI3~1kjm|0Q zCY58Yc7>4P42+!}1Z=u!w~7At zx_5ocEHiF(2LX1C-}=cEV$%J!Vy3kwil@eJaFG)8VN20*ZNh3t@}!MppzyJa1TWPfuqhie*5GDb($o}%6{zpWY?4x49g z%IIQ#*VWXnSb%CMlNQ5-!QPuK0@sG8tY7iI4+{w+Q%Oac9BLvda~K)wkdV}k&^eQ? zPs%QFvQ_3IMZg24Xf|m>7Di$~cb53a!02t+QsAl`Wt#(H7F^0 zA%6}q$l}2);Y^-6==ZcZBy-uf8(1KJZeZo)4$TWcHhx@;@VQCqs7gRXUO)PV z%YKpL^q9MBbPb%LX1P4NGJg?CLGc3_ciWQ##f`AF-n33e20_Ju5N~wd zA8R+T_ricqQ1BILWv>jv&CwMEO+<{jOB=)C96L4ZUDM;G9va$Y-tk$5Lj31MVI|z>6yb~LzAv_AO%vb%2GMRNfm%@!Onr!Ergm%% zA67|MfG_2N8*1j~=il_^x9)G^H8Q`|@AjV+fBCE$JBC2f-9vwivFalaFV;WdU z*D>t(yB>~WN0ALbka=DYM)k2}&)q>tlx(HMn9f7h>Ew?L+TNJKl~-A$xTjjbf_oj6!by#4yJbbe`aoqK+dL(4Fxhkkg3AC*ca68KpA(a%b_igw<>KM8T7MI6G5c0RdnLZB zp6q;$TO~uv-MV3tD9AcHVdCNP*=L5f=}06YgJUu?AwL(d$|c~Ja@sqHQVOnj!aOji z0kvPY+yFK<#O|~)6DXD_HyXpXAFGA3yF+HoB@K2S%!U>H<_T-379kLIzyMGu##NkAlZTV;`vp+BgV= znc%aRK|&*03I)>$p97P+4`fHL!x@)}0)0J>HV(WIZed*YnSaAC=EN{3Km4aCC?6Q$ z5*#c=)n0S0mwK8h3ErgxFhuQZMPLu%E?m%inh|ik42X`J9lxPFWfzbVR7IxPn#678SeDWbP^`nGr=td&6cdzx=L=_jqxRxAKxjA{_DbneFRZXfBTf zat>_+a&CpUGJi|`F+S-jqOIy=l%1`|Y;S;U;5XrJv_h2>?nt=aHpS8O7vs=aYG}0- z;`+MBz6AnqBfK00KXRL}XaLNSjpCg2oP>VrJj5@m=t~!a=amusw&BKXKVjdJI7X(% zLryiG+REm}=aDx^YxQ z9R;oEy{_yG_Q+N~K;zazve|x$tol zy(deEA9bMAk4>}O;LZs3+||XsXSWQ@*UWp%-_d|=46PBa%;YI1OYJtgCK#sBA9PtJRGy}+vOLS~^H>IP`xdAJowXGxnoXyOTi>Y);V+zp zJa4DH@%XOwA2F>XyB-3b_%ux>fe~0?!#0TC!34Xlmt~Vs;{Iy>8Ho=IlYP(S^?wof zYFi}8zk2**uKH-Nue)nj>st&`Ujm-9n^@(4_hO?|MFTwetQo%ygid0uV2u9CbH;O? z&gki?YcKywra0_o>A^*ELux#z=UC)e*^sUA^s-zt-!YmE#-oMEU+xOddRY*k)AeT` z&81+Z`<`X9t-2eH1y4jTr}<|J|9@Gn4gATid7VPwA7kIkpPFWUGK923egocE9Pw%{ zR%u;{<#5ktbbY26NG60a$I#eAMN;MESM;dPFnJe&=_RN|iTd0Z515UaL>`(xn6L}0 z#5^(ks$jM`(zA($D5zTp${vCjhz*!Hq4FunTcgy&iGi0MNC=%j;X5u04}VU7;pZ-w z2HNv?yt$3{U|}|-lB<1pK-`!sGiSuAD`HwkurDlE$jWy7eO*KOs3>4__UFaY*uUiO#xdI^&}OrA5C5DIccefbj+Hgnsp#` z?n!K|A~}B`%a0E=j_lZmVEz!(4+dsttI~psdUX{l*o_Qf84n9$mu<6yB$Zu!am{CR z1sEFUU)UTUwnN)9RK;HFEk&AM5krhde>^G2lMkB&Ac>g$HT^*1>3>9v1nxfB)VPUa zm~=qAv?YxhC_1O>$+_Wh9<1j5w)7^XhqyKMX7vsmr0=;vGf))!1p-e{w~~KE%fbYb zn2GR8Ea3#O0PqZrS!q@X&^@ByMB|Tz&%g6f!$F_c?Nj5Q#=XAQ>;oamAp|AM-F|(u z{9alZ?CE!-d)q)w0Dp+6+Tqeo8qCpf`?)-Mb?Bl5dyUZKp*9aQOG9bBcP|KIPaHZy-cHT(TTxhyPp5EU~xod z>p$A?eUQFKyHJx|%u`rr)@s$5bzkx+YE)eROX1=SXnX$EhZ%xS*58NLzX(N8BzU#+ zcefgMISS=14}WFKuY&cCZTULWI?d;`q+dnQeK)(M$!p%_CZQ3H^DCs?S3LSHBvt90 zhcI0O8r`O8&8=dM8p1{9jBlHn1!|;1vkt6x+Bwx~n5&?J75!=%fdu6(`o^sWA|4e6 zr!3hVROx=-btkYOat%Ui^(z_}+q=Q?&qR|$Q{Sie>3_ZrG7vVxDci9LizHhv#UD|= z`fssb&QQl@XomCLbQgEm%%ZALOD%a25zi+qSSkokzLewcBEo6j_bs5Nlr-iA=T~30 zvc0{Fde#-(<@eR@NAzZ3_(c?ea@v0ze2E;q{aMVMHNH=a8KblWPvB5$-Pe-NHjf~Y zxhWF%Lw})ms>-VB&28cP^beHgO?H$xUiX+L?JBVHuRmD&@H^xo`m|?HN2fGR5e@Ci?eJsQ`^`xblBZ8^3*E6N5G5&AM6VlwYTNT~0WyWy`> zu7CZ*m1G}t+?w~A|DaNdS_rP$(J zz>ggU0SIuv3x~QBk(QD~=LD@+V*T~&m>i2r_v(DXxoA;fOqU?8>2p}P+OW;_3DgH} z>O7njp*Ws%`n4C6^ey^&?aoS!u|fG0Nq-6qjVEeX#}?xN+Jxfw(;U2{KmA%cjafK1 zL2T`vHpStj7h;hcR;lCAPtQ0_np!g=!VKLRdW@~Wal=sK#-cYL>>kI@rW)W{VLCDI zyE3LS168VQEmyShq{Uuk2acQktHW?}1F!s#b&@eG%%+R@OE1&6Pb0hN>zKaxJ%7yZ zIM|=focZ*^E}GT50L`WiG5)jflD<2%4uyDQ;?t_V^+F1Wg z*vk<(8{Yz`o`U*W9xbf(+(j_M-itJ=F4pc*xlP<_-EV%o9jn6HyP~)%SY#dR70M_d z8d55*#0%vA9@_65kfNkxzpmqMA%8nFfyRpyt| zJj0khlv%8AnFm{!fGCJKXUo)C+@QZEE9@1B5_{u`-E&z(o>=IHYB)%(2(X7DcL6tzU;t2 zs#@tts5kqJ87M|8nPs5eYhs3+o+G(hj$O>xUWk&f^Zp((W&H<&f zlEzRA8P%0QUd~)z9jnp;pX5sTa>IGP>x5hMpf&Gn1J;w$ zYGP0H#SE#X40>K%(ca^`;!93+tY*Wyee!uO1)1>Vnh=m5rN1r=Cx54r<@g5YJl#Yh z-S>8uEZgcG`R&u4225BlL%t!km$*~Zgq$qe>oKO1l`-B(Md(3Po#)_o_F%^@z)g75 z11SpU%dRVKd_#fKd@rKfeJp~clrWSKUZK4%0cK&kYBPN>7hHABls@NiAq+>6%p;84 z^fOXl^Ko1yO)&|>rhn60z^KcHV=Fa%<|xE$j4A#YMrFKt{DV<}Yb_<<2ZT#n^ovXS z44JQO=_Y2q9t^1&AJ@J-!Pu+Fj!e?;KoLNojh5~z)6KAA;jBF1yz>q3@uCLqfB)K8 z*RQU9i>{>$X<)7qP#RQL{uF7viL_H39_6m}TN1dxC^dE+EPukXyHtO5F)ii71&4YF(XCL3j`t*Z{~`(DmF~NqLl*vAl$K1LrTU!A*1rR$|3x$~nFn9AP*_!~vm=F@LK-pQuoKbjltfv6EsvP-OVd zxoksIdZJdI1u$DY=O5WU*_yc}G&ayrlGw`@)=ED`R0xPsDpdx2dySB?@f7!ft6#TD zOmHTOuDk>w(N}s`lP<)XK&47#qGYHdwkq9vAHEf=4Ui~E1q8Hnguk>MDrY`*|NK=s zl=j&Cn}5_|zMlvwX(`o|*{NdAvk+&^r&Uyo zh&MmtI^Lzi_k%f-;8l)@5#N2U?)}=feM3FScso9t_oyW=vqcI7WdtZGdFViP1&McK z*q+xYN4OsAdkVdo1F+gXS)mz|OL7nIRlLJth=2OcFnpS>T`tlUJ!nx_ zX@IbsLl}Ge_QJ`~6T9T7F@3)dB9`^+CXGGMu#Q5@kZlpOhpa{F`DhHqkRa%h{Sw&l z8VAP(oLnv|Q>Y!j!>9*%_J!l+aNie+=zr}gI@qz%-rLE^Ye=#z?~zt{)TU9RAX0>k zRn;Pq3yI#IMVKx6y9S z!++z7pyszI{S$`y5v#0BUVUpQ$PD^Dp{vZg$9j}@KX7HNrnx@(l6`_=O4=&HkfI0;}>&l1NhXQC_Jyb%JyG+RxDIzDp`TQFwTRIN^cGy>W)x*5OpyP3x$& zn4$w^O|W8zHY zgUEY)S^i%GcYj}Zn^I3beQ*XjbqS9~=!rSC0N_D`$o;*iHt2#l;jqm9 zVRfS3f+$za_X-3tD%)THe&SeSWYfq9TgHXMTtw&U9YRP>AQZGVr4&Z`jGxnBZ7`)_?Nk{Ov6&Zxb{(s8r}R9Wl-MQFCt zq1deOw%bpNRzMEqh|g7RAz!uH^$|~sL~pr^y%FzV8C=R(i}TIK{`a%erdf%&*}7#~N`Ipf`5m5FdDV=v zECS6uezHz4I*|WcqmU*|uN(0bG=7~B25E`0?Dbu8CULHBPeLm6q~*Cpum;J)Pm*;?)Y;Kl*WCIJ zj1Q;a#??fwFO5FTDtmcszA;w?(8o~aEy5+dLEg>Vl-!3OGLKqXWOY7Z&GC^xJ8!c4cIqfyJSgLW#tU-lbALCx?xT@={B?-Dq&e`yhu~IQ zq2j^pPvlSSV2}^p_1ZHUC&;?R?#}%ls%ACSb-F`?bVJ zv)?ZWp-02q_BSN>5woqDJdffmK7?0e7*; z!N^YxyQZi+dwR9DM*=S{PA;8<4HB`m`tB#(%^eBAtq>SH$oze^m=ieszK)He4wQ*i zN@P5P=vQ2>V^_fGQs3038>X_r4Vz9!bom%B0fy9oCwG2k?DwS%@WI#=6-%BAZTl|^ zGJl|UZ|5_}S2HlvvMyeVMIFx#ffsQzmxNyI?SN(5)T2=O4(SjDHFc}(2LwFyAD3JpZ8Q$ib{}62u6fZ z4wTJYQQWNgdQpnpw>jaw4x-Rbc1}QB1Al&PYDmf3f>*=@I#rX>;5foYfa_D$ClFe4Lx5#cUX{Js$xlZvGs+R6SCWbMbDSJMFV9Bt{vW=Bv^pxs34f!n zA3w_6Mp$XmiLsJnU~KV4+xn(cuV>z;I|HBq>tRCCSU7tGnRau(2-Zv&GQ2XX?J0jN zZ7;@GYG0z?$wKBHE@Y#pT8bMvC6#fFcqW>RaXDF0mqb{83f1D9Eo=ymLgUJac{&U; z?;Jx8M4~i%YcXlG%ar&Ss1E;7Gk=HwWf)(#?|Xqcx?SeYNsi>EK(ovgFjn zKj^cFhRo)+T2Uw5PFbQN@?ipm@`sGns~t`8TlM<3`CEbE2%<1MlWN=lWIZZ->*<2u z`)a9+Mky8a*}V6_NTM@>K)VpUn#FjGsEjL;!ox38>3P%&?Tng{c%>qrJ%6vX@}P7+ z`yu9`#2H^U8|$nZ{$dWebAlAiHYEX&HkhcWHD7c}lUa{x^NC_<7}QrhKpSO1dK73G z`Rao2Qv1%rnLZF4%{!ocaOFX-tFo15iCtfJq}=C$p0l@sET(*bCCV}>e|75Rx85x5 z!plL$Qd27H%f%Y}?9JoZNfnNYOL#f&?SCCWaRoyfI z1kzyv<$kH1UUoNP@9@cQjm0ev;}Q2T2KTu!%Hj{#Nqm(nZAeKI1 z`h1+qJfB%I_;o#7;(ugfarB3TX)qU0EBoARD?kUftuOnl7$4i#VOi@}+O^M~hZf(2 zLS8F)bnX)~7gJjHoI^2E$4=3Oh7lU$5NLio-&PZ=k02oen(t5BhM>?b`>s1>M_;QcuO4Th8eoP;x?O%CWFFEDVYc-4!i&ThH7rbXL zVMNP&N9`-4&A-Sb`TJP@7fAIehi#GG>&m?newNwz==G${#=yK{u=TI z8j)@efxvHG(PRV?%I(T#2hPT{%Z*(o*f9<2@YTPwpbCfK8sq*CAUtgamnCWfHMjRH z0&pk-HaL@jDk*&gTagso8?PTm)JS4XkqR%4 zy}y3HZVZPrBhsPe%q*SO>SZxcf3+G~?UEz~1bS$Fw zg;7OvsW7%sGMU1}LTMyhlnv4BS!A?Q78b>$$FHYf)E9pXqoXXG#Bai{DvYzKa2iD= z9t?_5^;9&ZLI6*3Ce%$MMy$9OYH;kA%XxP5sfDl*Op}K34uTpiOpJS8jWdc142}$HZ=hG8qJSL&QE3li z29_*qhcM{UBMHLug)@Yp5SfuNFj^#PW5IKxSDva2gn%wop$&-+Ji~qZl z@~YYs4=3Y!HJ;B<4|?Oto;ayylf`sU%}C{(^1QkjmWPv%#aoGlXw^RIcggMR;NR_` zo+&DThB@~?|H(}pEaumfY4OdsXl*zjRg8a~F3x|e26M(APOEZGIese7syD?S!})b_ zSd}-`v>1%a*{t|bO=rW&xG?+1BmePqa5%E4=MkFe9tErf0pHJTHfAP7lk+~M>B}(`0VkU{iA=y zXmnnV$3)55^>9|;|CyryDPA=sLp?5O?;hxVzofMI_cz4pPs0xt%u>03e_Oo2E$7#> zs%)6``+8q}yA)#9(-AbpvYYcGC8|AfG@Q=n53kE9M!P-nQ&~S?XxtO8hZm@pMoQKn zRq&{q4W`3ek~n%OCm)uxie42Mv^am^@pv$~7>=)K$ZM%`J1UglK!ufSsnD`Vg;hN& zwEA)?tnE;tR#fQqRG2;$rcZ_aaw=SEQ9%z&BJ6r1e4hx{Cn9{UC{$Qc;n!1<`c#A- z6-M5}EKHjUvm#=>Z5rJtqHmkVEQxrFDP8XS8iwHZO$}}E`xfN7eO2Qj*%p5mv_>xN z?S3dsko&HNbq)-0YWY?L)SdDC_G@6SC&S!I8(MfO-e6P(v}lze+zA=?Bc_%B=zKtm zLvFXs8EBS5{Wehv_#_VFa5RR>;9Za;8>+)DaYd~=osi;f0FGIha&wz8rB3^hZp^Yu zEiMSp>1|nAN3~(v*(YbtkUW1|xK66kMoz~>ing>fXa)WR)yyfu`Bm9g{tS&K4I9tU zTB|=*OMXEL}GqMlSS5)QGtw*NVl3 z=@Wr!AP%^^nH9ovS|+|G0kTDP5h8kPr4y&eZ|LW2VvG_pyZRlVs z7@omPV4Kt|Xrwq2afGap4zJC}RlcaHJ|2@C)rz$Z_agY_v<~ofmI%Gk%9E07+|!6p zV|PU3$ibZ@H;bhV59&2H4Ga~TjIoi2{7_>#HB(0ZAwr zum31XAxY6cJiI6r`8eY7N36~+@M1n)RK+DNMBdR0dWR*^;Ix_-Z^iMWqjyWcf9$Ui zL1dm%TM%_p^DFP9tO{)wgxS!{G847*C(>)twn2v7q%6s=ueGZgAcvF1xNRfvYA?0Q zO3sS@TXd-JsYA*C{(tne6yM*dp>?(1OHZt#trV3PLCeJn=p9~OR?})s+b3_85rg4$ zFscTV+fUQ^mW90deKDC=7w4lp3PoaXG{kDlX|qiy2vfs6GxxoS7>Xx3q!7jWL!Jv1Y^T*?Vz2TFk^? za&uFP>rc1W)tD+9PAuFVq%gJI2GCx#ee>VImX4MCj zskqYd5E%zB-7%7Hh8Y`Gvze%VUz8(pMVmL(6vS0CG6PZ0#ep~!55*(#SR9Eb;s^0m zJQF{PpTu+VLcA2m;%9LpPQ{t{MZ6NP#T)T2@vAtWmV@`zoOwD2nn|rwm;4QeLsHM; zM!-HGmBD0MWwCI%Qk;u{f4C6fM_h?(F%-Xv_hKY^zkt@vF`#Z1h_LVOT^h>zlv z_z$;G)vZqRQ>(T4a0H`MUeyY_nEWvotQL~>2(}uVDcoKi}LCUlFvV6e?(c!L>_GOqEJzj zd6ZYv@)n|8pw@Ydl5;HB8)7;SxxI$8{8)`f!`s<#My<@RCkrykLM7E`o@9dr1CuCW zNabKLuS8bY<*527SeD|Z984jZ^C@f$ramtREY6D|NJj02mRd|nu7m0V?UzV!;(9T@ zDyNH^QMs6l$rWt(f4vx#R8Mt7qM$uraqyT{qXGlxB@C797K;Wge#(~gvi^BY8E&B5 z#fWoQB})2nL=4&WkTzt+C}*G4Kb3%pZ@}^SXtv#;;8N~^yCGyZgzScp-4Jq*A!Iiw z>;{G1ps*Vhc7wuhP}mI$yFp=#*llndg6Tf24c@wbFx|&qfAH4bBbXjHh4g(|PenhP z9%P@2`$W@ai;83$B7#~^L_e4wbdLyspJ2Lf5#d`zn6*UselR`QJ`t^8x=$2cpx|;S zV#r}4is8}jzUNSCja!ZoI!7_X$dSWH(E#r$=#Z@Cm@45lR2;yHRD+YUjfKGRJc2gn z$QzZgTt^$He^9>XGC5Tv{1vOpoB zK+&x_m}PQ&tC0;uie;H5zSYLsK|6|XS&lb(j%n!*PVudz!o7)XX$ftp2=C`9M@oI!q@MXg)){GXdEiOYGmLlqj!w6IJ%Y7qFy0v}vpvJLcw zVjLxQY+GJx&SUG`bjB1lar`T%kxg(EG!e<3^+gdq5U9B<2&A`^5_M#&3UySI4sme! zRrWz!e;r?<#=a^;Q^N{#M((F~e_6;v^@A16n}NvW{laS)=phx^1PJKgy5QRO8~%R_ zn_zT5MW^n={)V`Q7U$KxER+)!{3^KB6s=u)>g>J4Vwe6kdLa37^I(g zV-GC^P@g8(eOa#z;(fPv2n!B8T9M?wf1xMVc0QE!&BUM9pK1*K6P~GmsxjQ-glBAf zA8~cSGp#?>82bl2Q~y+ByvG60wj^34tUuBi`wnU9A8CyDa7b0>ysdOAD%PKAjD3eR zfA!Ba#(OxVDQr02N$305lF@fb>9}~8jwDy^VEB`vKeXO_9`Y=-oKdzfwne23)~7Dp z>rICJYBte%SVXp6J$st%icxpWB8@>nSGfWwusEFg~Z3<93MN=rcpm6c6btz5%Uos%dSrLo-H5h z)a+X*@)=UhO|f_4*?nR^bd7U%$3e=l_JQuerS#kGC7V+As`6eC7;;NIL53jCTe;t({TlU_5ZZcp35VHlkFpG9LyB(FE#4TTbo7ayt zxwY##9CoY0^LDgK0eN9HrNzx+G#}oMwp{$m6C@;>Kk1)%yH^mIY@%lF2^ zh8=pk0=%9Q+owd!JCwN9Vx;j7os&F4-ya(3`7g5P-G)xnW>GNgHyX*iiF_Ph(xz)j ze8q*g*}Ud6^Eqdr&A@tte_GL|T=3+QJFK~rS9~XpazvT9F4e*#-^Zf67+lUR+f~wd zZeBLf2R>_WEPF4^2;(_DGeB&SvrN~GUA@ywnB$#CWB9g97KNu96!6J&9%h7|V^Fz7 zlZ*aLu(UVMm*z|g%QsmpdHr*kjsghzC|H* z^F(nE(>&4e*-Zb6b^;m?fW8j9ADoJ{~PCc+X=y|)y^+{4P))ps{4Dt z%Jerq>AM(TOo_1@?)R%rpBXUn4kdR!G+^$r1#H|VN;Z6IK$-qNu)6cH0s4MO<8Bzf zR%kfz>H_akdtzU@xC5kbM?F%XefXf^W-1tex}bC0XZja_|XNx z98@;EpO|=WDDQ77Dv&cp%X@?QdUwXHH1e~6<|c{M9+&T+{5kq@*j9_U2h7q%2$3i=%ynvwkJVNX0PZ-3xkpC73m zddFq<#M5~>8V(M~hUw(x>Abpm#o<2Mdm{$E*${_b8{Y26saN6gO~LYR-1|;k{M*dx z(GcV*+e9wKi-TuJKORCZ&!287$!>|Yk~A7p$p_5ne}jIqB}gYbf~3clAYD(650qU> zk`=I$WHm4AA$)nPx*2L)2dL{V-{rlz$0Hse9Q^Y74RaE=V!cyfvkejF6zBenIFD`4 zndvS=&2FqXJNV_77ssIQ^mI#Qc1x^f$*as#N(oe_Hb<$;QTiN5asxlB5irk~Yv*p7 zvb#(-e*pq<{OjYx(*tH z+40loOx@N!vRh;=ODQZ_3gK(i_<%0f)lA7YQ{=%Mv%f;##Soqa{3`Y*;7N3uZH{n@ zr$0YDd+{h!zMb;ACDyV^r@HG-Xx!JRugz;}fAe|<{)oFCUrne;Jb(4-;NS^S`L8W? z*e$VsY)Qofo}TCv@*%kP^yIqJlWVbcXLf3hEw-aVa!MOFuXl`SdEBUT45@WGiO()I zCzrZ%Nh`S=pS(DH%3S*G#L_OXRw@{96MyoaKCBsiODNHIghG!iq1d=lC{FhWsO4wL ze>!Zr<_3Ux_44u2lb1~8c2?Xeuzm`K4JSwc`GQEt-P`>&bo*;)_g56ztZd)r>oW>- zO{7u${OIiU!zbT8KRtZDr6G4qtmO@CJ+PX-_m9aq>DDBPhc<8g*zjicMmMWt#wT57 zR=&>Z*~=e)dHvn#(bF@TC@sXhWN2z$k0%`$p||v zJ;5UJ7>mMJ7+z@^*I5p>7lDI%F`9iNTI=QRd70O+wCtvi&o*T!JD%i+r9dVTQf>EjbndwMGUmNwNdv%!-zAeaC19TC06gcgS;bU38P z6^ABn#o+;l`ya$J-gP*s_Xb*4lh@Q_GcAkxZD}*K>rq=v+bn3uWUwmlacupw@qpS1TL3o)&d~~F*G?iF_$r10Tcx>H8nUl zlYuHHe{H*CaAxfiH5}WTICpH@b~3ST+xEn^ZQHhe$L7RNCYtv-=e*~<^;Lbf|Mazb zbzQ4hch#=_W0Mn0IM}jLR8*LmxY+?r46JbE?(XhF?k;W&&Taw>{}id1nE_la%>d?Bwq^iP1w}3CpOOG7$)9QfNi%yhXCqsH zqMNa;l?g!3%EZjx#f%zY?%)it{m%hl;$UxT^)F2>4F48j?{4Po`j0SkX9qigytt~c ze}uwMRe-oCqpB#t$leqnC-qOcy{ik)KWsA-7ngr^M+b2IUu0|azsUH1(f_i}p8ts$ z;Fy^KrdB4d0An)?D|shRmdY&Tom zpGJ0O04h-jJ4ZKHGiQLjgQ=ObJwV0Cf8OOEN1Uz9|4TKpv$FO2|0VxlhP12EKV}Hq zTl`}m(|;f<7YQp*GgCz?*MIyrH?sZ5r+?!bX8#$EyqT$$o85ov&HgdvpRt-c*xP#j z-z)#=<=+=FN@~d}3k%c!ZwvmN7PmKXFtxI`0I0bBW2lj{>HmO#rxlH?{#vTzq7oNtFx6SK!=IpUpvC|ujjv!-v48Xh&Xuq(6ju5)3dTL1DM%3IRRYUTz>x_ zxh8JT&Sv(m|91VKUjBRj_cocCd77EPZ2%oictfqz+kRF0ix)3cLDDYF8LH6&nL*cP z`)+kmtwt{3Fr)Oe_BlLi{Z>6MF7XEcR z?@EP#8YA>?Ak=Z}OnyS#M&7$-Zf*oC(#)`B=UvpvIF<9GN;1b7R&)(QqUtsAqAXa372OyMz8!3Ug%^bEBixeX7+4ln200#wYdx}4e|2!LyPayO zk=2pI%l1^gg=mTh|Ac*=wb1VUWBn%x72aehIO-_98m`h8k)ws#dJQA|uUhZ2b+?ES zZ>?Y*RLY3kx_urrf`?BO5>p{QE#se0lleb;@4|iX>FvVmzOk?}>Qj4k;Am@gE?;vt&_O($VN*Rs=lwf}g}AF=hq9P|usE_)&WEvm+FDCzemlu#Cl8 zM>s9+kOKXOuSgkw2}F!TnLb1=5(LZ9waF3vN7?>_S=Iwa4eelzf5uJVk^NMS&fBu( z2sQ2Qc62$EsqI!}n6rr>>dL-!lg!jc(%V|)!C`zl=F@$WvXq;LO`CQD-1UK!3VGCI zTefkAR9|H5V&Nc&q$umLB0cQ8U%&SSJ`EwS?Bb#eU-g)Z&wGCvYG8vy1mD5Wq-sZ} zE<|B27+(egq>DscfBYr;1JLw=KwUdC0z?_e6;V@eubwKUu8tuBlJ)G>(~RlY9HgPw zqTEb*9Ohj`jGTH*Ux4~6NG(ebC$t+H^gx-JYs@NHbn#Gnz}udDI_EYfaaf75}im>vV{TNhUv}Pqrz{09x?ec@7+DFXJbP-=lGt>c%*gc7aFlv(-L?YV?{S z)+cuth3wpqK!$sJUHz6be1>{iUz=-KKMJ_c(@}gG(3#Z@t>ATDCp?&Rf0FXlc^C6s zAb9?X4ba&6f46#`P0ib0!`z59tSujT#hv<~bYCZJ`i*+6UZb{4+qgT)7T%a$kJN$h zaB-*VD(4R(SMCWtftMy;NyGSZ$l^&cYEm%F|VycY&jO|qZ+aU_;nb~9*6&Br0FahwnkDJ-_1}RmbO1v=`Zq>7= zPlgSX3<~9UfpaHzsQDNkA5sh4SJW^DL|rCFOBE1Dxd#5h!826swUEqPvW&=4u8KFK zjFu!ke-y}0n`-+B9beY<;J_4@e|Ul;2?7nSqO_`O$%AtJprkm_-`_t^d`$h!Fi|`^ zNNf{*R3^0Wt4D&vtB4Wm316F*8JwJ%Q3>HOWF8}u{iY%U?I)Lb z+6T1>o`XwbPsy37%iHHA856mx(tBfEEWcuxfBY|b;n@8+GAS|SaBOmBDf}vQh~rWL zuhymYJYV|Z9&zLJ^BfAY~oBvqF4PPs2evW ze>Mx`22ujlNCE>QX^2=X+civkcJ(VDkM=OmR<<`$Yk=pPKVF6^n6xi}Xmc4k#8SrQ z=}UTr$;I$|_|YHO_F8lZKeD^aI3_j%m#aaxr?v!McWi-aXVn2y??R$D+3PZcbyg}e zxVX-yRL4g>r?3+4lpwRI#(_cxUx1L8fARRdwSfVGzw7OwSet$!BYYX>d3t)ZWS>+2 z$iR6usMQS_Q=oaF{iY5;s1k`ChA&<;q9RX-rT)-G#rJJokhzIz*EAOpjG{uT(Wf)f zjnmsaC*Y^s0ymIJB6<$u##7WLS4@K7QLi0|JIaE{&`~%!R*9YcE+kiM0urzce}2fd zno~=z3JGMG%3g()N^fD^_b+)wl;U1AY`T(C4i#);-vlew!~>WT(1(gMf16`OGNjVX zu@5m?&qh?D5+ri6x*_G@`vIfu$D&OP}PA! z1Kr+XyJCx%ZJi zLclzjyI}JF33Lf|S~cH=lcRa0s+kf!-_YgVGa7?30R{eSqb?Bg=Lm16e?8&~^_o&7w%Y){o(L2L8!RhSAfgl=xcWzvO zYb!~iWCPyz%lX&MG6??{#i=FXU!z~?KM4?|v1)sGXrX46lv8iq7RwX@DOLKvPollM z^$>D@ai*c)jj6}K%cl+Qe@lblcpc)BgHMxFO{<8_LAC+2KHn*6N$Jv%gTLj`(2?Yp1d5(j38Ni(SRT7R_EPgW0cVE>E zR8W=pG=%zDg5^$d-s%u9?~)M9oTR(EPcx(UGP+1f8E?Dmt%RktOQDI zd!AuuxnWmn39M4YUi0&V#FX$nfF}l5HF;tAT-~jesE_13FQ>*s@8&v2P@E|brr+nx zrNxyHnPSb^R1q;EDH%rBZO_~TA@(&N^mZ_6`+Rty@1Y}}rt-8v%Yt{lHUvLPsyAzY zK(if}OQKjPHrG4Ie|@Q2Cej5eM*RK@+C&H1zvY+`?&yrWOX2ZqbIcJ)+lOY)&b8Eu;Dw=BOmJ1954wU#+?k=f4Dw0cop!{eN2V{Jb?e0 ztO_`!ajLto`3d(+&jkFcRvSSxI8mZFM4RpRnwqGechMtz)&|sn*U$!*5Xv?FKkC__H@dO zMLAsz{M&v}9CB_U&Gzxj)GDUA&sgL%<2bc`>r^SDf710_Pho~s&<1u$#%uNd07mcw zy3o5emgMUPKXPB>egtMHguYB^kW@Ms!zrKUIy5=B1X&;pQD_XVk}tWokmJ+G?}fRH zV>D3sK^@A_+)n;#?y#qF^m3E{lFbT3%e!7w>v;=;{em6jVoHVH&5_IDD3Zd!nN7zi^dA#rBgmb#sP3YVERrdDb5&7K|5NJ)2>LHZ|Sj>wrKe?G2wT}YxNhZ!Z(YH-g?EF?ia;5r$_ zIFWOEwGrb0R~nLGKiHY38ImlNVRp#5xk22i>%KeZU00y0I}W6Uruf^N>_9#K{nvn!-J|TEU-`4$N~kZm*TU1S!9$(i!`g}dxUL_YZD=AsB$wEe;*kpS$%OVHT-(&s%1qT!XbSVM+D)OM%G3> z?&dI=&T8++(^g9Se&>)){vY6--gNaPm4I(qnRamVos~U=5 zYS`NOOJRh3rZIAcHl|x}@eHT#b=~^SJ>&y?+u2=I)*uxdrO-fTn0SOXshy^Mf6;AK z`q87ELwvOcVfYex>pF;ib;TKAKm2DRXQk#W|N__8OaKy@@gdq%BQ2*iD>!*Te&bJ`p zH=%zOS64$O2ef1ITG$i6-=JDg0rHhcOR_V+c^xV=gu*B!NWRQx|7(7b-ob5O2wFg^ z5gP1rE*F=~6B#heSzgqqO6vP}F_u(e_{X<9ZCFMCNNzbibGcA*WS_?0f4#OS;hT*m zR&P1oLfaNf8s|KsHs07|ANfm``n4<3oGI!Rk96;~<2#Px*#{Rv*ukeVV3Ekz7s&z^ zyDV-l9v3gM*B`(dV%koDy+7E0k-gi87}Qa245VwcF)I+*jp1{X9(j!DcxD1z{%(^! z-4wE9FSL2e5&+~o0PNt#N?Tav7F!vas_TRzlGa5fT^>urOa??3e?T~5f>P&5-Tqb5 zutWSBf^rvKBC*l+gzK#LC-+8>TiEn410yi7$bz9?I0}x4*F%tSo=8$4_8Q9xrwD~Q z(xZ=pN~_C%XK4+_(nrP`fl~e4%D#>Ey2(z>*3pRTj41k0?SDWzQxL~OXmJ`U_JHNm zgpdPcpFZ-n<*qZPf5TKS;V{g)RE$Zz-kTs!C9Rbl*{1XcEu7I1Q9)P~C)__YPH(}2 zzT$mV@8%VAGc0(*0bH|k72vrfFgdJ0a74b2=q#f;CXiYdL3907cMbL2_SC+Q%NtfwKt*8 zRsNzEIrVZue~5qM=&O~VLh@HG zWe5Flf9O$B`!6#)|8X&_tT${~D~JuB-o>YUJ!~G+l!Dyd@UH%`{AzEXUHz6Q|KV%# z6d2$xWrN_yf4onw3^Fz?o;^lzEo&rGKTe_&RobO1)3Mf5iTjaesu7>XWE8m);QgtUPQs@VA&~ z=-0d7DgzrvcjlEf_6eG1`Qj^qA+DGNEaA^nbdzCx{-xIBpjtgyC2o+6&15J9B+#KU zq-Mq5%T~8>5WWkVnnO9go7WLvdp|GS9tve$E#&-~8AK~*f7pBLJ1paS?{};Ady&SX ze-^fcT8EDQ%P?BSG zMz;&ckd`!CD0gc+`WnpGwY6sS6+ah0uh|jYTp*+ei`Rq`{4|RIXibuCfwE+dcc$Ka z!e4xRwE28F0-b7xaj?K05q6=TdmNw#JZ|&inRkuYJfHI*!$SX$wnNY)K|hn>f1Se^ z4?lHdg-BiOp_8|(XVbUN#}5jlq!J60dTPh&k_{SOg<=mNr~B^<`y(@R2RwaxzMVfCoW^C`CN zp;X@EA6Hwcg^>tI)Qb?b^T>2rz_z#UR>is1m3=JlR}3dBqS4Ld`)H6*e`EE}yZx!p zt_P812c#ObuTZ|olJDNvXl8~1w}y-Zsb6HPdw7V+%Jt%IIi>PLzEMpsEvdJpCEowc zYe)50`uPp`_Hj{_tH|@Wn0lilXm>{Kobj(Y&S&bq6f<8h_dez9G*xe!=z)hsrkvhJ z6{a@Oyrrvo)R{cG@a5FRe^OIp!717oEQ6ZxsvAAxt!4Sx^5Ij5*y{f3H$mxRTyDkKxblq(4PSAd-$O-par z0#1d7Mx7JKu-(q&+1L-`DR|M=`K%*RfvegY3@IZs^_Jvx<@b16i;}-TjZo(fFErsp z21Zp*)am;x;Ks76f2X_<&=5J=ziR&Sac;ggbb;lzGX;JcH zN4yF&hg}(|kDE>OC6{6`UfrbwhI*dGGlseR@xBH97|0OG1mn;m1Xb@S?>w`O|kbf$*`Ej5(W!L{03C;Rdcf0{A~QRiP<^>wKyo1;P%rsgpY zEL>|2wD^N{5iYo6=d|1LP;1`FuXKPf=6!4=pJty#r(uTAAJ7EVKlMFN)Tv4028HSyLz_!$QU3OX zozy0UMWi`aYfq=DIE%|TPRzL7>_@%%v4cctR?%{s|JpGat8?TINPqJ-VXi}{je z3PBCMZ3#mn!8`62YR`MKVOaA-MR#qu?5^P}KB_fvX`+4~ZIC<#6K$6=$ow2Ikn(F{ zf6XQDKiY77)r$yu&XH64eA7#<#NNk;Ko6z5&Z>mlU-9p(;=b1Hm3N1D(Har7wV89$ zqBd1l6k|HaWC!ysDbq4H&+WmC;HJoe78_E!cFh}QWS7aM(YViGTU^gSku5`3C4uttpbf3ttM@#(J)T8ye|DtF9>temdiC?h9PnRTI2N!Ef2~_^~WBbSf7wn)1x1UBzOzXchK>YL^+)n`onX7q-jJ6Om64B+#l2if4d)- zmI6+tV}@-SATK`YeIgV4l8*268G0nT#7UcFHu3bB5B(04Mv!;k7nk)Mia_>xE40JB z*n1hIPsiq}@thCy7uVsh=&aD>FdF=u$?$uvHucC2`6DwZcl4Gk46 zuI#2Tdf!PBLtlmx2;)5s44jSiZLQ+`wnU$TC zWNvrAt_1Xkhzas8Hn>EOe>#q3Hpz0wWhX>;QcUBi`&1feE}9g?0+e-z`fy{~J?X*o zc#^5A8#|-HYHnC2TIE5vA0_`1E=A??aCxonyg{vaJoMz<6e`6Uxfs;_ZzPQmX zex}{>7PYrs1Yo7>i_JhkjAAgbI(uAhI4nSW@D+ZgA{I~yZTcNf8xvH3kU0FQR0$QO zf8A&Iwd_hoNBj^|$Go@PD(F;4bcw>RU51C|U>FHdDj%if%z(lFDdHi6lpeKsM*N}Z zZ$U+d98G}0oF!oye^RgpI>0@M+Ce-*|MuWc+x_T~@%Z-6g| z4|M6d-XA-nTD!x3f{7ysbsb_<7;{k~8RYmqX#$zJoyneiS84>LoTol3YMwb8?(hhiyk;;YR!-r$RWl* zn5A>fw9Bg^kLCXUhDLBVIGI$FGx?0i1Wf`K&7XQCe|Rc)`G_9~ZeHJ&!y@9(`|vZo zqbO0Kx34!rHMqEHJp(QA1f?QpZ2IFfXKV5>DDA}4b&nF#%e$C&XS7p0{@E@L zWAY7Yfz;I;GvW^7drRU9`izyJ00EDznqQIeRQ3mX2zB+WVl4TX@3pq?7*G6ofRjr6 z4}F;Mf9-9i4Lo;OhWUA~N8>pBqeq$-dly$ZBu-W9x7yqXRb!bq4*V}vah`_BuR~*6 zdQ~~Q5HWSvTcWVlZU3$DU42W=qrirPmn3%KprwbNsvh7Lw%#P&&=L-@%$7?Y8k#h{5zepn- zmRht_=g69;2Z`=_Z{e!fFjtoP#x#L7p1IMk+ydura+rcE=eqz#s=Z7w}0FuvtSuNf>lpDWDfXil*yGuSDPoa zFuM1?%Xse?>U9nujvmB*0WzXE6)eVXe?$()-g(pmw@%g3xjt1c->WwvaOOV-!U5bVB*qCUK1AvC^1Kv&u40Re8dB z1stUmixUa^>`L@Yy5p29H*Ny{uk}f%kkP`j=#N`3W_B-Z69gK zyoN$kbc`f$sDBqsH5s?&5j2|HR!%lnli}fSGkm$02-2@`*JkZPbSs%o7n!r|p(m#t z;RB1N=wqBKH&wZp2=p6bf3PhErbHQ> zS@Ub11siFvhkY3GYyE?Le;fUER9ZcI0@jSZ1Cfb9GOHe@ls!U2Vdtz*5VytpwqDY( zm_Bskm|?NjUL;GosPO3Hc9M`chI%`8w2_if2Z~l_G7@_X`xLeDVqWNu*Trbh10Oq! zmgfqR<+HZ(qKcc7gVuIhe|I#jlWf0NHK_GW^f104T~5WUUmDu&1|kZHDuC?`o&7aj z&Ca=EhFyB_=l~pn;eyE{+&QN>k;GYnaiq^;3G-%Dl0bj`0VU0KLV(tJUc?GoZG1s&{4Ke?PzF-bsCJ4kr0~ zfg900FFY$&$q3JTvXH|2o28t#4;8No?MIxPRD*GlQ!&lO{cPeyS%WIhK4**4IM>x?@+{C-2u2*pI^hJ zTAd^D0eXVY`>%MG8aqXR&k*Z2zjN(BJ`=jc&PFPv0lU7vB!7vWOx zq4gsYD@uoEz3;hL$nO?89LbkiF5e;GthPPJZQN{v_h)GAe-83*_p2Z6qB+w+fDsDn z@6D{Q6@0{cHBigcf4-ya_h{lxqjv~J|{o_5Oq;@ ztUN*wW!ZRsZgE`4suS2JVXK&~JJ^{BxCiI2XjP9~vsI>H?z-wNSg`PgOAHQLu>?Zh zosIA9pZ)Ot+ZhhROD5K_X9 zswp?U=mE{urkPp4xZkIBhdS`tyNdTtm$hH7r`kNDh9!6G{HSRnni!=g-@|LKn5`lJ zqewGYbH8h?gK1={81ANT-bh+2j=jI++8$qwQsUPNvyazp`V~d(>NQCn))G;p>=)W2 zE>btme}dt0+#MPoKJn53Tx3}+K7&Fi{R1GnVhfI^L0?iB{B5n{conqDQ7@mdDe~9< ziz-vxn2Rzg2i?J5_vA8MZPEW7*ZFMb85XMJc;G3vbxBKQb>)L+h3s%RX?H1f(b>Ws zioKheU}&3z5PDaVD9C3&(xH{GWEckb+Y-;Ff8FF(W+c1>r>%r6LWz)~+-mWYBUppVvP(4+iD8*|IRkzA( z1tH0Z?_xQqGz{gd=BaZOF=Z|QE}CAdiWL(3|a4IHha^!LufeSKI) ze?|LzzZpYph1aQk0BtQs*r``A%$<@wUu4!d9rLQg{SX9LN#O*hO#=OJm+;SDm zz<{h@%4a@NM*>UBH7X9XYYHZGE#==+cqCAAd{>gnxz3afYkNd%!Y}bdOUGZKhq{%9 z(b${{lY%&emUJ@4C2jmd0(PF5{g*rPe;uz8>Tbjklrt#1U)7WS8**n zs?FjI?g=08SE{_X!y(q_ZAst0ZYxm9jx4MV2kAFKwDp6)&ii6@HZzsLzc&kzME6aI z;NxrpwK4$j4MR$kSa5Q3&>{h|;8a$8sqnQI0ELnvtejnyrk&65fE;R0& z^1JX~==y1g(f8K$$ME7BL&>-!3?w9Ifn{Cp#qfC~Hw2w$Vlk*{pW3Eu%P*aV9Oa@+9@z})-if2g#%k-GZ42=I$PXfJf+z&%0Djn$?}eMd7BB!d(y ziep|O5!e#!jx$9UAV6Z5X5uQ`!C{=S0Z%b3LQsK6wK88VQ@ zAdkys)4#-a9*}T8l4vAsZA8LD&&e<5bm_X$Dvs<`3C<`>X<(o62A-&me>nvRIc(w- z6z64SKlRS3tbJ+FZ%;W<{B#}|+c#>*)y81EvgYD!Cz|_z^r86!xpjmro6+69uw2GB zTbnqO6s4eEv{1h6WBLZhEybvRg1ZZd$L$(CZkc2{%KU_x@Oke`nn3p);5jF+YM4Wz!v&<>43@-_CV-R44WLlBR)rp|Q-})f8)t zJ*lox_(YOmB3nD{)oc%b^%OFEc}bczf$E^GdAgviGB`1Gbdi!M+^X-!t%BN3xic6` zei` zyH)r*15xE*qxLB7TswCVbT4~yiQZi>U$pDf-(jn@Ho+Nvf^CY{oin07_u-NsHfFSK zWgg#fn4QV5r4KnTnKesVbu5j97V>u5ujAx`NxG%)f(3F$#rM3RgVrJ>iTE|8I=#l+ z*xsak)HU_Qe{W0qg?;mGD4E+~FF#2V!RMNdq`1OIRtTH9JIqFYlwW2{OD+a7PAwIb z8-hLW(_zN~g+;TVEzo9P_q0BS7K*QpD$RfQ|Iwp($LYIBo_i&mg^O=)?L4a|ZN8k3Ut!Z(*LmBr&IB_%AYp^-EDz5~MEeZPWz+nqL>xGDFVqWR_ z5KZjDQD%j^D_Q#o{~%K~k+}7&Xl7aBU;39%PLy(Lr!j9Ur>QcVhjgAfS1OFP z@J|!6fAW&6@^Yjri_2(GdTHS81@Abkyo&n)E@f!HcK=WJFPf-cG?XOb(dVfc7wU_c z7Lb`8!9rb+>_!~t5OCj#GUG^kagdl-GcS}M_;O= z&vB!DtCwMmfU=1_=~~%sJ@P2)1Rcu0$9DA_f08-62lC?*;M4 z6V8m~TABM1>5V^`g}Vxm1Dgvp$(w#|QZE2kagd7TzG4adsyk^CpCgy$^TAgUUd1(4 z@{ArP(N}mSB>cn%l{iu2gkw#6GFKAMog%gKylQ_ye1lQWgS7_y@NoEoxNn;#xjnya zf6k^kD816cTxLSfMS!)jNihTzGm3GWOems&pG{qu?FZQ#0lCO2?K-rc2OTBB-eHa) zQi^^J_wXDiw&-nmnOIfeYgx5q{YV`gUk|X)^-S~PZy`t~-LR}O8kw4RxX2Gv4yY?@ z&zzgE=kKc=h#Hy0QLsnyIHl24PlGlYe_gTr1r_JRV?npYf-5Ll_}ozVk3w5i)v>MM z+9O240e+X{X6QZxgdk+%&MzwG3aSP}f95<3I@q^WMH1fX>m3v6PcHCbaVh=?g5g>d z@fi+%Y#~mCN*8N+WBs(_hoT$l&xt9{txkQM>h7RZ%ni+0^|)qy+EhvChH$7!e-gq- z47bI=yT6T$#n7zSeY0Hh_w(Tz-;?)EN}fB!nT}7E1EeC;v}*(#*hvOq9^~-%Q*g`Q zjr$c8^Rt#(bUHBpj~{%a2r)u;j7Egp4lCVI1m@9P-*xcgX3?^eCS9(l#2JN6r$>-w z_(MPyp8-gjppE5xQ7_#U@f*rmf76|Hir>2^1&G`vJ6ZIwK=BMol~nKuPvQ?QMv_%V z*{S6Ij!e6Kg_4SqUFW8ah*;-{?NZ((RrJ3%hI0k))L9ITo;rvdum}k+#Bsem*upfo zvWn}^NlR_Uo9EN``FOZX@zy@NvG$jWKSbWl@$U!~cxFv#eocgPJ$|o(;?S?vrr4e>JSF@R_R@g9%&`Z1>Idy&Drf(Q7v4jJ8kZg|Uq!Vv!2# zanFNx*1}JW1g7$z5H<4>2}2x8pY;b-ETX~Cb~KWzPCiM)LghARF9}UGB+Xk4|MK1Z zUEJ)+9Sd_wP#9yp;O>BrcrY}_UOLGKCND{RK5FQy%0bsLIc#Ebe|tJfsqP9a(>fq! z;fv9|s9Zk5seEgA{k~Id7jS0sWIJOxSZ{fST~l!1(U*^HG`4Mz`*f7zYc$8+Ys+?jLlnK|bJ^CMU~C-0EN+_i(Gty0j#PLT<7JC4@+ zMA`4n!tCd0voF-v@stp(92X1Qmh&uwCv&n8)AK!dr3{%3S2;QkUBgIb$cStYXzRJ~ z)W$cWizTl%{N+|ZD8beY#Mj*2?_tl47Cn09YaA)v9;D>&50zs0x*T(65Oex! zYTX;qssBLV1#47BlW=OT_(2pJ%nO@50$8erTZb|Q9Bc@|7(ajiyD^_L7 z7kz=CTUOA)7U=*}GG3Z*&!axXYn_;_%;GlNt9ByXgAD>wOeL+UoP*NYvGD1iP9=u# zCYwX!PTACNyZ~BAd0FSTC&XPW{s{-G8zJ@OT@W{4Q)wJ^QsvG^U=JjkK!k>HaK~q?cpKZfe)E@_{Gy0r>E#4Ie>)jEV+M@Vbef|wobChc)yg)n;2ATr<7QN7 zkZu6VZB#Y~>8Sb%7X7{B6QY08V{({H8Y0Q$>i z_kq2^55-l~^c?u+MSSLN?LzRYtxm&>GN~~`zAIrW076~>^cHz9SeW>u@9TN zzJEVVLk!)XNjnNOlo8bux^~vOXBF70j@|HEj#gXj?F;J(&ST?&SCk}yihzr`i3_oM zJD}>#ckHw;(yFvJ6?q)aJJE9Z{8s$z+X)dhwFp&f894T5@zkn6#UHYUQDOY$K#(g6 zkyyR6WK>FO;xxqS5$4I`=K$qp`t1x|bFTW$_qbW65^fxNV~Nx!KIDuHMGvny^$Z@@ zt+xi@-y3bB`l#xClT#HH>g3Ei2ZfyMHUJ;Qri-y-9c}iE^y*isd?B5eD+%F=lhe4e zgE`O;Z%EA)#b}MV(fwv+@0!5Q4nu8wa>Rpe#1HPWsho5N*gLK0YtFcfF8|AddV=@qA*G;&t9ygfjfQr)o4H4Fga`ew++VX<09{?Di z{ys6>b_jf`@l*$8KZ!@zvU#lc^e%$l97YUPmG|Gcd-B2Rd~WvOFVkHbp$R zF)AYJ3!d2Js?pK?3aL+*oY!+&0SuH}m)<^XLi`F+{!TYE&|{Sb`U|uCWCrQI2lKQa zYj#&BI^Z5#RpAhmJC~4dB-@b5ESWM#o-9GFc-z`8ojCsn3Z@^8WkmR+Ux9nNr)Ls5 z(<#IU|4%iHN}0Tw$iQ|PFHD*5Ec!MbxPrjVl>xhOgduePgsWj>y<($#6q)dGcCz&G zlFK)1H!^iyg<$P&6cxRre+mbH#DFs{mynqD?BJ3*QUc`B)aPJmzntJehi;_y&xYNT zTx-8pGbKA}B286;^mKwfMc~Yv{mDb@dD(-T7(qOZVfKO#D|h>kOs0;8bN(Lb=}5q& zf0v0>pv!@Vx2Ey8Dt*3P`!ec0(~XY-ZyzD>o|TY>r9*z*E{7j7KEUu%sbBI#1p-(lmE;M9B?7co|1B$!3I zxQ{2aHrtwJ{e7TTle?8DkfOZ5h3(CT*80VLWitjzm5X>7U#OAqS_N08Lm9!&s} z#(GzD;7y|S*Y2!C#P*A6-$D41>o9b~DaPVQroEu7P;m39nl^z+#FW3b!A60)Hw0~1 zvH4!<$F4hA6e*B!#3n*z7=9Hbn%$-v%bJ&ZUL7iZ(CWBs!%eIXFLp^5N%f*j3~Sr& z^p22D*KiC+&9Kx_@@v2h>r2zLqc~RgoX);+rGIG^;VGEr)u}hkv3FnCy7ZhG7NmE! z$LX1jRj&#E0Izb*a!FT5X9k9cAf;b&=KIqsRTtK4% zHl>MG3dJ7wJ2Q2C7t!Q!-|S7J-6f~9Sn%EPX%y=_zb7b(>fiR4dVv2;gug(k4#)Q# zM&+VoZ46LRh!+Vpf=a@(Y)!@Z8T>`Bh4Ly*x94FAp?+Q!w~mgYyj88y3ExC4Tp{FE zPjFGQ(gSz2&hA;-VI0Yp(|5sGp+87_GEWMGGmmyR*07Wk@7mE!PRnfkZ8fP6F*~8jhTZo3bleT(Z+x(fj9sVF`5f!euo@A*m%t9Ig&v^B}kG?Y?XZD8+ z9e_cYPME-~*%LfxDvwtMgZLd3jRZO|EnQXDW1$(E~Vq)KSCAEd~*L|uM`4fuhIZsvg2=J3M;G&=wr%X-h;pi7n2v7}q#UWk(JIIvOluJV0q zGMIxFSlUA&|K^(Z+xNb>T7aR--LM#BpLddo;GIdpu9q5U$UWag1oq= zzMi^8`H5-$d6a5g^$~j0Uo&TvF%h8#`sv`o#qHjfgMS20wqwJxE7ul0wdX8$SHRf5 z(#}t~fTP)p!6Rkn@9Vq_1=c7-oW87@iH0?p#UpBBBo4;z^yUzSB=sC&o@XZZA+4XB zq*a&ZrI}xCMf!C6M&maN@wy#I83OEhA3~}Mq$MBOrxJD{V+FRj=P&xyI9xGjI%_Ib zpZ%?WA^q_bZ9w|Z5Og3VI|w959$wC5*G+6FUQSNdf4Y*ifTO#m3ZYI77nE1C31mn+ zCZniJ1kx2G5^JxiiV$7MNJDJRB5i6=W~G6HEKMr?J1L@utXTvGTHzztgg1 zL9&-ysI9Mngg4*!R&Pmg((h~)`Xmz6JzpyN=pu=F6U|)}27f7{Kydr=%GC<+=zCueW?rH0PveCeE5GeuA^SP3c-Mmzo7CU(<&8~Ky=jQe}wD%I=iK&p^H7Skr!|;cBuW7)|8|7wB z_`J=LUMbUsEg7C*~t}lK&4xxg2B2+hAc_l&d$?oI2X*X>6OA-%3GzW&wG%};`zoY?PU9{BTFk#T z1eu;u|AUo$we(9o=kDdKjEG#nj0mY3Vd(peqL!r0FPr&GE@+KrDwbmu|84{QH7?H% z9E^E{mTBVg5K}nH%f~gU1k}HIdDWy?D~It_Dh@{=GPg;P^6vu=?Rc%OVhn zA6$$ESBQjdWnp+A7|ZGK2O;J6HCY;(C8317CiC!W?g@Um=lERzwTji$aot>>g$)tt za77_Q6a!XiyQ|7A)NwUP*>jio^!Xm$((DW)e`dleMDc8uEa_MQoGStVUzu5WoAd$t z&(c8zcgB>2)M}BD)06vWfXg$gCw103oOY+`%akWq{fo{5nHrT*H~yTIS_`{|w@uCv z&JVm*Zr3tb^-9fLd)bR;dvH;_7HLiwYB5;<$?gQzf>zLPl=%$$2q7GsRc9)F_Ok_lkw42|LEmA-&mxj zB-;BfHUD>aIZLI`C>RM8UpDRKUfBF~Yo(G?mm|HvVviJz12$h>*;B)*CXKvVb%)Wf3 z^G95D?Wt80%Z&N+9_6kn4Fj^dKx<`F-b>=StZ-uAv)BZQ+*i*V{rDopoWE+^gQQQREeZ?R@RF9?jK=Y;5Ryv@^dE|#{n5B zY-F~HV6LR^K+jh^|GK^{J(V&z5L*D*xt~pUHGy2*`PqOLSxcIp;HgmYZJF%uo!7Ng zv}`ov&86rkcRYT+_lpnS_KlktnLpP7Or&t!4(tJ3So(kqP*FY--U(N= zO16IEDFK{b+u28RRfKcHU;CxlWn?|nJx4f!6=hbFurA|}psIGBP70qwDJf5NMmEIE z8PZpZNwByIvaicfm|^*eKcYdd>kLo7fguk`*pR!!p>?w!GZ~s>mbdtJvK%Bm6Rsk= z&jD|EJJ+K?AZFMvshG=?^0)I2Xhc;H*%~(@8vIz~fsomlkp!t_fn@AyRaCo!I1KjB zK#P*=*B>*I$YxeUlBan|>2{68a7&X4lhn;sNZvf@h)P8YG9abJ22jLx-G(!&*z#f+@vufP zN_@d&G)^ivi$d^l|=jiJ3A!A>(m zdY)SX)_laNKE}?{?m8q*HWzgPLc_BE7UAKL2VHL;XFtm-92Z4*2$rvvbcM$1l|Zda z^D17FTq%Jy<)16xFZk;~0fQ(O@&ku=J>2@CFP^^lx+at90g#KDG(V0UNB?ND+9wo+ z#y1;Iwc?pt;2YdPZ*%VxA!5-~UZHb<_q*Q%`o?NLGfnmqjShPi^(7~$)?Xc2rL5+6 zWVn8-+FQhvgC-16<`L2caM-$8B1no#!qOrF=ew~P9juq#kAWF2tby*1ogVlw&jABv zT`Y)^^gyGL=MQFC6%EMnFWvQ}!Ph}Y8?2XBTh15zFHplp%!r7KsT#>&82+>ob{p*h z9mKrlB7w4moJwQovA(?(!;CtSXgySK4Uc%U((j3B{h4BK`6W~oBkR)5?2Ij>O&U_d z(2d1jK0ai)LJ^s4We)g7L^cO>!{#CzOrtIN((f@PlEYunzNNsiLal>O)ZVq{ra$u` zg}M^)iUrhL#AM`dH)lwDLsY3Fj}4UpbcfS;nLklm!$ELrya*FXnvaYLT4n|Di;s=4 z(_Hyl?CLh`0&<1Y)*Ctf<&IO9il|QtqV^RY@CP_zbhx-auqBYgl!h6$`v=FlQD3oH zpbWz1veeQP2bVE z*Rpe)2(tlR{F1~w(DxzR=_smwS7`>AO459i&cBv`so|~fU75wm_x#QS@lZT^?z{W0k z41FZ9Y~zh$n);hyhd2f>k;Y{mIAgpaCnKYxYLLFP{=y!yy&GwMx+aB|OP3yib= zr4@_pxNz~e62_|P(I;z!Lx2o#)gXw8J(b-FTT-uCpc;Lc7>j$NZcERSX$q)<+ZlWue9cvP7s4#03F~R=2FcW= z4^BpON2@iTg+ce{^O=B8zBQ@ep&_N%n!1*I!TnmbG7b%HexD0HC$;OXpd1=3526uo zQN*2rR}MDMAHp`3*Ex>I5sXL2K?2@0W>%#Fv{Kd|=zoU?lSX<5k>7t}I7jp|1o`eG zz$a4wnDF88Vy>QGDfu+2d1Y%%9;YQb-3^7fhn+dCPlw><;qL*CzEei>#r)m?%MFUQ z&wr&&H-Fo^x%6JdlzY|Fh+~Q0XzqLceL_u!HBN45fk@H7e8wWLm9;)_>$mGm_{w}5 zqe)Fyl&XL?%OJ%!CfjPe_T2f-04sdQHP8-~ADNnb!;We3^@V2>F=aZtPXon2Js`HY zr_O)#sMx?2$DkJQ(TYmud1aD-2~Y5VYEETpgOb(w#@xevLp9pFh}OhHEiayPgQirD zHp}WflmeR0ftEXon(Fz`&!|Bu#NuO$ErJ`h4i%#r610I=TJxncN~WtH1doOz;F#$j z9gZGZ<8^X8rjXFytJwYEeK`L1=4w<#MO|`Q44$ggmZ%7;U@XVHfomZ z;oT>mkJWPY-om5AU|f9{)W3pY74}P~M9BWWq%!w8octT-c?ZFw8e;>@CBZS(5Dm2u ztjj>UL-hM&{LMLi-{9BbwaG`7&x;Z5m(S-WbYW|pc-|y+R~PlSLimHN<)&6 zG(?4KcA9(OPTXk~CZexq2Qh)u5oQ7gHm>ATC`|r%R63eX)O2mOI1*>Dh@7YTu}>z) zdw@{N-RU0*isSt)i*_3E$NJt!k^4(*vCl83UQ4C%hEyXYj_kT6Jv)NsmxMJ>+N)>M zFwb5mfreH$=*q$LYnwJBjc&hr;!M3AOu~?FECOD@MTkd5-Nu&Vn?O=0Y=zbGseU%q z*((fn3f7`}i=0gnYYD^cx~YJX&<=y;AK9B0B)mWpDUXf1ga|0 zWPO(fUt~9{Rkg0-MPd}Wp4BJoRMwUU4~yKbT?XU0;opve9G`Oe>?BCvZsjGk)bRX) zZM9rr4OTH;>i7jNMj)C(Pv`z;7z`;rt|dL+>+39S3EdUt4qA@*AWC^?#1hx8y+^$@ zQ}qaM`}#-B#)g)?Tb}N{D zM?h{1e~Rb>gZ93_K;k1ML5nk|p8{ob>x>-GV4>xLEI%8d+&C0|GSngNAWGaAU|a5c zGx!|h!-?yn;2RdI$(5eqX}bu)>nyY((@++!lv%ZTBCcW*`7QS=m8CxOy^$1p+gCm; ziotsxO2hIZVN}1>(bkzzllh6%&O&T=HpggPS- z+K|7dIqn}%O$jpF(G~~SfgeTDhep7{)*h1h_G}GDFWi|zFZ+htktX$p;Y>k)bn599 z!vGaItmi;qpbN=37JMtB7)2nwHC%M)gG}gPz^H4y#NuAaeBI+E4NT&xHd1V#zLDwv z9n(Gk3)gNGykkRTzco}&PR*TorQ17jqFqO$u7v+8{dTJJ2R!3;4&4(rB3<7=5X{^O z3*Lj3u2y#|3pNI%NdC5*(+oLR&NYr%Hb;NxrjEdw)>r2s?y<0dv!|0cjJJvJ?d}BG zI^mp;E+CuCKjPWNDw{fi(owMAeq>E*IZE)*n)%57qIr7|9vgo?N4 zvP@B%qHt?)#uKE>#(6B{yD)8a7z8;eG*e(?=xuJ+4X%RwQ`4DwGu4&+v~H zJy0ep%>&*xiT!N7?Gh{bF$z@|UVAi*KXe29`7mUlip3Uyyrdv{vx~RYG}C82UFX;D zEzyETJpKp1INtlq8MBQ1&v(Fi`6%Tk+Lw+EPy8erUF%o>%d4i&+II&6^AH%dN;u8P zGak>AgN%`LL?-xt`;)qH#FJXm!z0Jln)&v&Q>>1QyeYQZepa3i9J2%gQ(hMnorMbx zP1^-H@3L`cj+t@${dL z(^2z-c0ZFQ&LyICvC^XZE~N#m+!)t)R~q=4-;VC6FwT_`83rlBKq=MNgh7ja~RaVo~-R{nT6GRwFQ{8m0&ugj9YDI3xlR- zeyVs5ejZuyNe%FzT`Vg%=t5fMfyUmw3E6d#wD;ug(i%QXvz{RPuz1kb+%uoP?O4Z> z@0fdZYX*tJFusUS*G5b`iy`x*(05rd z+)Ex`tW1e~GX<)6G#%N(!-p6R_i3rt4vWc~tO1i!qtBLdUvW5H1w-{kc)9R+mk%EV z=6q4owYc@)pna^a5=JCpPwUl^K;4eyn&0`F{f<0Mj0L|nGlhR??kwGyEHd(8>iXjm z*sf~!$3_J`SDhnKFc;S){E0bG%=>3W`ly8R)R=ZlV-36}!ec_UYX8E-yzt4l12aI7 z`zoQ_croclMsk}LUjAUR=95|CynigFC5@h%3V|&<@kkrw9Aro?*~&O+;NbG=jUZ{! zM!13$PaBU&YaTuyxmvN5q)#aA{;0T`F@h3`_c(vRo{E-Ub+}!TA*gXAvXf-Sf9t{H z1N#{_#sCz+%~L9jqOh3%W$yMBN_lL<`nM9yPGS&op|B5Yxm>_)5)|mxoiu=D*`&u^ zQ)i15CB0o?9^3D+ZQqR-)@ax!Fbr+G)YgrMA(5ZIQ-K@J%u7~_xHXw1KWKt0|3Pk@ zt6r#46Te9~bgy|JD*zefY#&j=2gbC^sS#4Bb_874FoN6a8r->_qv{E<-S)vk)!qw) zGm}+f2Y{71TtnD7Ffd}nR?!<+L&`FOT$*6UX(Q)Xzv2-YG;9$W;D^n;48tnVPTQ7> zDHsFP61tJrugrDuiJWN2-jAf(DFcCZPg%xJop*xV_wU&G8A__H6bll6Zge|hG8*p( zgn^{*qWwKylews4!C&2e>w;rQ+bU%1gi+*gSqRIt?z-xmIijIn`eILp*B6bV!5pKWM2YSk2vep6c`=z~4oK zE8qr4Bw`z7um?yhM3~txOj@t7Y@TYgxI}r@-#I2kt!a<7e%8o=3LSHsOK5>7A%ay` zt(@^TC@->Bk<2lY#LDTdRAv&xuN}&gR&&jm9MEsA`+TMt0~Ku2zUyeyC;D%K7J!UP zs)dfd8EwWOPNJXohX3uC;=Oxw8mWYAJLR~z^P7(`!=r1anPthM<{=T^J$=142A)>q zaWY4I)kCPdvKRwU_xV|zsma=~mZBdOu0JttQ)Hwt{bCWq^~DS$JH{%CB$40!(`&&S zT1O+QAkXZYjBQ<@eUC@FB*Z|99Z))RsXer&cly#QYJ!n5U0 z`+Xne>B=Bj>$v-~rc#rhFPajQtmN^hLc~O}a@utOtX!Qf%)YLxxGX3{z2)j@Rk;>k z^@(5!+rE(E=f*2FCC=DA%H>G8m;PgV2q#U}d_7uZC~EkD{sO{0xe<7;aFwx__YI$< ztwN5R7Q3nPaBe0R#0r+5FJSWdBk@i zj*>%K{wpsJrPZ`U6px1)Aaa?rYdkJXd2!Ln^t@kXInnAf({h-?1}NP|GhC&X;Xo|y z;-JOl&vGyDDFct~e3nmz=YNV^jo%HxfS7LD~8KKhVPg<7dz`JqZA`_ z7R{w~6^i6d^TZ)OX2cyvNSL=`i`nATHW5;Pyf(c5F{7-=QGt4rv^kgo*Ievf9#FASxUJy#793`l zK2py~)$SHNm2pQFn`K|I%XY)y)q=$ImzR%xucmbJ3V;ukJ7xGb?!3|+HXLGc1 zN&(@{CV(|>h8oHlSEPTJOMORLU_n$17dn?)2dZS@%8rsK1R}K!IR!TEAag3Oi>nDis9nPxGoP ztozT(WEEv<$<4j!LY>71AyOawS@7g;0vZaNxQe+Bb1ePaaj(+JQtI z?wcNiYkV)YY>6HTWW4L!oF9)(nm3WHjM(ntv5v#UsQ3fEe{{t`85bW)kaT5h`(H7J zUfJZcGlVxwgUcXIZ?T1vb)O*{_mzm5hhAy@09bp_HrOxD6xR7T2){S_%*Nu&H)MJF9*U$_%i`6CH z-IF749?dCBotnpDey2~-s~T5{r;G4Mh+lSgk5q4$t`2+_P(K=m4uG<;8QC{C0oHQB z1EJ6>lR?mm?{_^R%X8(uUB^BK@OH5d$dvK@$-7`|FN&eSP_1J%N)}6(@bXA*2-qef z##xS$bSi(h+^VEe|M@tOxFF@q@hqx85>1vo;~W`Q<9t85_~HG~*<~YcHR%@{SlKm_ za5y4PvJX#}w$`P)7J_W*Fw@cm?9HuLyO=xUo4q35X2Y0+IV9r{J6CqAD{9Qr{3Xdr`^i24|TZwsGPj9YEa%Wd_QxJw*vQyU`ZTI=KMsxP7 z$zzJkA%@FQBKPuhMpkeQ_IGnhOlccLX?BF%lKLoqQQ?6fRts~)g*?hN;M)^odQqe| zAXmhRLZXV{A5?w$=%(FyOgGh@bR^PSSyb{nESR}GUjmn*gSMJSYc&y3-mvh~Ybh;ohk77XQcA0cJ#%a*Vr_VSDt*~du43cJy z*i{eJQjbU=g-ARDYwNmu`fK155*8Y^bSA?_h)g25_9k)g)^em7@DoMHWuko1FN{HG zucK}}NHfEcqi>@2z_CzpOquNwzt+9=|MSoZ<^v|M6-!KSLePtvJ=5a+!p$K0oC_>8#9T{GN2``si1oI+X! zy2rjqhKj0!2$8uaNB#&NL17fVNq%p}$t0TFsuZ67>rtDU0B{D2yRH#O&M@%Bp)=n} zB^hepdT&2k;+C(gI(?4DnmmN$jh281*)rZ)XLt$_;WM|(N$Orjr@pq;vKO`E+IHf> zNKhKSU9jSJy@2HL)-?NN14~sYOgL6z0t5UMe#Zxb`v6m6HYoM?RuT479$q{I2_4bc9vg;CV(j zJu-LuZOJ~0q`S$tpf9~EOXyd*cGShh-dM-i$HBSD?mz88r!YF~Vr^==T$ftS*A*Uq-P`q$e|8*TAVxs+ z6iv7F7kuIoLjJUkr9WSVq9mZrXQ z1}j#;vG$AWyarK*evO9aSfTDblVh62siJ?*fUFL@3bSLqakqIgVvU1Ya>7lo@XGcf zVFpaDBoCZDdNTT^EqWG4$ghE&f9zvA zQm#8Xp}YIM=_dQ^PWbo+8xzbx#BnzD=V&N@id}Irz5-7z=}dfI1&tcZ?LMO%C$@2` zeP-#gsE7-QjxCxiOX)$0Sjm0ke4s~yy?l{86E7aWL0viYpP=PQ+ilnlsN9j(E9o1% zREXFSOxH3BD2NnSiM{qWx+K(QYKDTOw6G{a+e}C`YbZlYIJ&})K+6pPMeIH6bJ<4- z#b^T0a6=Zz6^bT?h46zc2#bAWi+p4koBKwca&~y{!k>i4A}tmnrm>y_S$NwYuX!DT zK7YebQd#JjS13ic0?9+nhwzHe+YfsIEn7qewJB_XB*Ii`twA-IVl;E}Pr}iCN;yETPd3RXjTzfbay_=!Sz8 zt*L=j$XKfLZiFRCRuGg?CCA>`+n#dvLubmUVyXn-YhnJuGiha1bxepIZHoTIh?|!x zu3yfKuYx=c>2LpdRwbw~88OK7PKF(UC7(7d3zhw1BJ=vCRIML)UTE;$f8Ukq>u|i- z_nC4fIVW_*=Wn7~?^EL*w{+4P-(XZXKZz~5Me|YVYr7vT3SKBGjLO>F^omyJ;S&3I zdA0t;D6@PIg=Cj8c^SHD(p1w?C{K!ls=l`BCTnQ@QT?#7Cl%lA@cq(Z{!FF(())?H_b zJPTFMv{_mSO6p2@$^ct+*M#CmB@dggD*;-OXzSD|5#JIEj8cAdKT_y#pMFa)KVX)M zf|C4a5(L`B0e&aJ7HRDrPDVF7jXs-mHJ+1a&8ji{A!8xFx!eZ8n_di|{|`$EZZ1&9 z5hOXlZHU04ZtG)7%Erx&z@km6&qm5Y%Jwg*=IreLFUdj5N~(syBI{)3{4e7CpGcBa zUr>ydOOk_)Rf0#FTZ)52nv;{8RZ@(D{hu2JI}aBtE1NLs|9`{~@js82lM|%64@nOY z&9y6Qh+z%ie!HSaq{sPUoKq((nh=q9w*YO-Ku(H^j%J{$f{82?nN|1)+^o6q%bLDy zOn7Lh4z2_vS+5cSVqWCr+q=^J==DUFz_+U`HV&3_T`2gP;0&G&{ZS6pT1NV0av=sg zZmIn=DJxfkk_PsBn~K!h59XoFhO7nvm3$1!mL=g%A@aG08#j0p&BzxHTLo%9A-F)l zPFw#A0b>|8G|41)hNkjlIGK~p}UJf{bZ;J72 zhncgVh!=i&7i*%eaiaY8sbS#~e|7B2i?)yn7Gj7;7 zK_)i><3+VrR5K@1#u->4Cke#eE*Fbit`<=|rE(nRKK~X3&>*u!FQKI*oe8&gfX5RRNc0mez3Tu(@ysE7>gV(2S}&(%PuW`Y_)Ia?w zi5)wUKo1|;G$H0b_UzY&gItzTk%3Ni;~CM`g^}$b5r-g!{CYKIr=L(Hc{Waazty4( z3Z-k2J*2miC+8MunP^O&jn4)GUx zg7m_89yb0!JKjHLu5JH{Y_=aP?^FGwGKwW;ZgH_3a zjiy2#A%HdO30gl&6BP?HI#DPYE781h#k?1654k`y0?h>TT^5n`@iu@@&lz8v$Zf_w zF#({cp=Am@pB<{%kjN2uF>##FpiDX zuv>k`|KO|QBwvdLv6BHyFB0Pv`5~FZAB=i`&HJ0`-AJGHQu`^&UT4#(UO@~*x-qD1 z8%7^=*!KN>b7fFU(Q)dWRKN90KPi*X4rhdxE7c|}t}{MGV?uXzR9Y(%NIS95Zn@yN zZe6>=KftKNZ?-vCGruu1{AuFx9q%pTnmRnWs^7cc!4aIK>;Z+CXEYpW+y;Y&0%zs= z@1z9gXeh^TP^0+o820@o%PA*mIq1~5%5|%7v%x4w?&6DVmJV7DTimT#H{1yftjhWQ zQY0u6hI5?dD2Plm8VNX@>Dae0uVdBf2`VPu$Xu|{MeROFxyy`6!oLyhhIzkA3Kk!fuQs}$-tCb&&tq`8PMsv`VnO zX_z;LSK*(Zdx$Nck2zfXkOW@Ni_9+Fs zW2Gvu)1!beOmR^90*icaNuHFRe9)j@UW0<~wpytKwfdn%TwBKGi^}M~0ZPlj0F^Zf zHzjP@-QI7MNAv89zakUQsDAkg+QGCCl(*<$ei6@A^Uy3yYlsjb{n|NCWbmYPPGF}&&~ zSvll>S|u#eqXKQZ)36B1xOdGiyIh|!Ln+&%@tJRa(&gdeNNoxZFd!JaJg@VHNtoi#57qA1-Zhnt zBZpryKLRxv8_$379oVd?>$t%Q@SmyO{fn@OP)oX0$OM@Qwh>?)7>KB)8<2RaRYvcX zv`PQ*#0Q|!R!MiGo28Ojyz_9_X;07h5uXjGYv85H22A(th7sqoR<(=EvB+sh=|n$k z_6WSqdE{x41>x|wI)DT_lMlbTz#{=$_d1)?NFSOTu#9xd;u-ORV` z1+V&^|e=*d!EAY*4_5b z_Kb8p-gQ)XmJH$2UKhBuADO4Z`FePP{8ToV!4hJHo1vM5py0q?;?J7LgWQ6nLzBqO=9806LO z*9ppn7Ck$?hGi%z@%)kC7RR9!N}7DhL&O;f2!GL>o4zDUshC8YX7T_;|EY8anQ56N zOBrM!twJUM!tG$de3fJwuaQU`Ut-HYRoC7IUt)nmUYr2)ZK0!+GknHi!5>y~FGksd zq<{C_z~>&m%x;FCxZ-rwC1LAS^+Ye8M|gO zXZA%g0fV$+_?D~?)kqbOa#F&Y$&FQwFiKIyx#`YPM@pvh`34sM;!0pp?VjtUgjHxB zH5=;eU^9)K_v|d=O6wix1~WgI3y|BrxKj?)f!80xy6@ip^8ub=sb);?$$O6~4L-K- zu6!4saJ<{X@|3-I1?p2}JO?>xSGN{)5Cl9ELE>JI?tQc6A@sWg4oSAc};XjT=MH7kr`lMzClyX%MHrZ#6 zh#f?x*aV&JaZIBMa)g1;^EB>$4UYT3-Ti21# zZ()~9}%xyqGDR*R1oXjK#UIz<$J-w`zAv2OO(G(`U7%f&n9ihB@jf07> zE>j~6Al8JZ@-=j*=CY#e)%1T&FGPP`3Ao}@2 z63bpKu3nV(frTI%@!0Y6=p7AXzBlieBpXRQc!k!)jzrJr{R06|LV_(v)K=^kT6;WB zwCShb?HmsMo;i{=oW@FigLLYLjFkX0tEbGAp@fl=R8_#FyJvgmOv&$Lzk}@Nr-Rvr?z`LyzpEGgfuUWL z=a>;SxcC~*taYPTbsj;N(-mN1lmF!%r(pK-u)8CWwpHnB!@QXTF=sktcP>?6-Bd^8 z!{n$WHyMTeA;I1wbPEh+t!Xtq@K*+2)WTyQ42$Pr+k$fxJYD1Tv&4l)n)X53L$iRd zKWb?y4nz$#Ok_GQef+c}bD>p*mL(}oG%AaFS>-gfha|BA0R?*;x(m?cX-qp7Vxau+ z%bNaBN(B@;_=SY1e&_(PsI)>aR*JeZ?i>e(9wrq5M~Kxd*$fAcnGEAdv?mh#iIimp zVGy_%#tX-eSt=8bpr_5@Ey2g~A|nCkXTNFTah&gLsFzbc#?@M_EkMIn`*tO7V8Nh~ z&a}_Ek=g$+k}NW4WDSh8|5Imc4SnT7jl*Uq+vbrTQ_u+by*ct-4^Ke0e2Q7tpVGOD zPhEsLGNGWKHv1=WCbR`)Pg zRsJ}PuW=B{BSq8R{o2J^ZtdL%GVWyz=}lQVwECljdEna@6Ar{vG_drCI1#ly0IME* z#AJf%ls&PG(J*uVibhO8Ts46T#}vhdYx0M2(yh?jjc(yEYk(uY`S;KW|sY`__QXF(A~%AuwK53kWrfF0=SbXb&f%yOYc3OPB`uR zu)xkAQ#KXFkxV zWZyz>$C$;SD5fMCwEfxj^CiF!e~27UkBrS*lrv%4>rNi;Rd@6niy!gTK7dkRht|-N z^n05>O=&F{%X(VFM#ntVR2NH<^9zws#$Its8=Fr0g=1gV5;419J*66$mbGpe#KfNPRme0kp8 zbXzsx=U}6%WP8X+_UVfn|lc}g`Tx-MySHunmX6FVVOD0;mWc{M&30)!D7r~Xn&m1 z7$MG{!^y^L_8QT^|7ktNG{BXD9AD;Gh*4Q4(RkW3h~h^F^$Z-WORv{Qc++zFP>FL4 z$*Q~TxaYsgPZ0;sfsrX{^szj0pILoWoCjl+;T>N$bG%|MMK8|I=wf~^A-l}L6yuF5 zVae;_DY<_$=b{WYcpm2xRg@4FKs%)g93V8^XGqr156knFIXPid@Xy60=Z9%{$&@)8 zy|Fl0_F{1%VjUdAw)fJGil!V*h4N)t5Jaa?;A1WZ!4?Ct_`cOavHoVhhrfuilGS2b z;%il=fVl^)Gze-l7Z1w8EzssYer>e7p4=``m-U95T6T7PHOV+N7=WurzH{D}<<7rK z-Kw+Aqat}YUlEkW1(pP-g;s}+O&wlnRX4ue3Qn?-2l}^r`wNh*RIwyB?P1l!;ewui&)Iw^^_<+qc6o%C z-;q$%ksntJk!?OudGEH0#5F3iM0uP}^%h~FbD6Sr@U-4kjlnA&FG!IQMO#T(>wX^I zw|g5KA50$|Kv2X#(4+q+LO5B8nTh|4CH(wwOmgN9mTp$WEUZcVw7_Pq|G~*E^*GZngFlbKZSB5nq+4X4C6tihsG1|FHM^9lN|*xV5RVSdp=G9b!t`D4QS}BUyk&-I z!DQ{wFu`GUtS^sLsK)wTB60F?I$P{yBULmwDlAuOZyjAdum1Z|J6|md<)~0AoraHg&fht)+8)85(0n2<3$Tw!g`K;@Wje8^m%Dnio1`ptyraJ1&?9Y1cC+B zkd)dJ*~kbkKpRY}&zRN=eVaQc%Rk^&@rqtDx(*f*R*gy~45QkCBAa|wB8#g7o*5Y7 z1q%LZb_85Hm(PIfj{m#bq?7>Lif1_0Of&y7nE22~GIO|S)}W!2duMUQ&k^mUF5j0c zY4u-^eT*w@n;?9tXH?0`{IgQ8KBNIO^OJgjpdyQ_{gH*mI!QbYof~(~FRS$9OAB^1 z>&uh`OoV*>r}XZ;1SWlbF!E=7Q(P}XnCP_}Z=+`*!_P#K3NI#1&rNPz6OGUi8NFQ; zQ?8s<(!oP9&6fBtv;ALOlU1!msxa)tqM5*MVx`DkUaa!}y>1m{9 zN`{DmGdOv)iYE{K&Ipz~{|da(=Xd;hHpB2o@nv|82d2Uid?@gfflE}T+N8jm5%sS|Fux$4?aBFQ{~KQsXNsg5815Yj1OS27cd~~Y zb8AuM!|w-_tlVa4m9pzPEkUhQya;^-jB&=P-u6 z`Rf@L4f{UnjpEP!=*Z`?-c?KLw&FNxn(W?GDWN(f6=fbN_@Z zp^!7+dwYwU)}Z+M(_p4Kp_0omaYPgj`Bv{S^!B+k1f8B!xaFq2DLPa#J($)t^ z&L|EjwKT(ZT!cp#UauG9Z?HfL*0wq*nauvC$&(NEJLr9m{C-Z0q~) z_U;`BtCaf#y2At63WxrGs4Y(JWF-Uu7#r(sJ%T!`}XGtlEgZuxkdGqd$hw`GPMnRyrbY0g2XHX>FEg_Yn#e{ z;v19?!g;7}4~=Dt*&J>OQpe!S`+1(ruwDd$O>3pjxzp=oHkrs$bA>C0l$IGQs$dNb zh(jMKEQe-ClgTg}ZpS&txuKXoT``uTn!I&%T{6ujp}RPBxNtcoHVo}xu%i@AZAGQ< zgQgrEYh*Rf5kT5G^nz~D02br?=SaSB!_79+p6tnC`@0td+d$a2C4X2H1tp zR8)b~qPMw$yGyGWn5(66*O^UW38h&v3NWRq0=^7~)>q~P#RA75y&`ILv(Q!Q@@lij zYr*6UG%WYL0>+NqI=<1!$$E|aaxICYBW77~=_-+}{K(ac>H;WA&xzF?#EO=IMxtK$ z*3IyqzT|pM_U&DWN(_G6AYeUrHJw{(NBukW9vfb|8XGqzHnUZB(=-j#+1K`M9Tz55 zf+u+szt{dqB%6e)r1K=PMq_YoFLa=kM>Q${1o}v%WwK6F^o_9PRqTvn#CXo6zY-ww zOuWGI!r0HdBL>0RaLN4@+b8*e@j-)MWP(xYO!%&a?}R-HBNW0o1Gt@+MgFp9V5TZE zqhr!FBRzql3~EetZmj0(MSDR2SI9tudA&_4bol~3Yv!j*l z2P!EQsJ>jcF+rIw8Jx$7U%ExsnKq+disvj|@i{^hK4}mu-2Ry&9&VFB(pj4F=jM2s z=;mds?)LZF$b3kCUvL760J0ASx8VF+?1NjwzaF#V9g>{8Um_WGTC@a^ydldl{=Jt= znsDg?^FJh#w=6NcWcLRPf0Z-(uE5Ov~>{0imc{9?2Ymn_A&D*#@ACw^%J9VzTc+>sh!K4t6|)cMVKfhrludV93|PO zYuK{=fMLHh!8E~EdiG$a(~G3o5|t%T3s^T&*6uIUg2ku|6Q`LAoq7VHJCJV(-;cAR z8`B)A$6$WN$Yuid>Dgi-j&U%skH|Z;%xFml%qe zM-&#Fi~|emXG9+Inrhz}59&9*OXxoydyr|dh-L>u0Cga@>p`%MuB$ra(o{!tg;UP@guR}GgDsQYxx^WA zij$_tQfawGccGih#rsa)Mhkd!Fx0i5P1YK|RX&0ml%eS$L1L(Ge*I#b@O zD)>ag*hLH5MIoC!!J_v!e3YF>ch0o%m@V{?un4A8Np4nFh4=!_iQD}JaYBEU>bFOw zeIBr8+HhrA{vp1+){oP@{24Q|a(`T{q;VVClS>w8Kf#)f3n**1nV1@J*-Kq>;W>%; zjkK@$#?Ew}ljNkix!GpJtI^ZgPV-A3t;l|fYBA)W$C1U>umSRcE9PeXRVQB5obwlW zAb8y5uf|nO^&+yv(G)V>z?%&`MQB&k@HrqqE;daNh>I|PDHW>trX952YoVg3p?`8w z8O|6@p^62YIhicJS9D*FBNd_oL z>de~|#lHO48cEEO31lMf57~4qK@zh9yPau7Xq-;54O~`eT|!_N>!lsnS%|T8oHl~O zU@1Ow9ZcTI*qHeW&fDrLNk`79*tq#IzqeTzw_h60$dJfdV?TEwc+c!A*a1JBF`#q+ z6q~|ZPZ(51=AGNsPSbu6BGN5Z9t^1H93U6dd50?j@icltAd87w7J^&`aq#$T=IB#y z8FuRo^Jj!a3%`qd3w}3w`9KUL`Y*_5C%-RIgfa3WDdc_x09j;{N*%cjE!j8-PA zSo3#8Fc~iiBg-Uv`ugpLJtd0?uji3J%=-=^9UHO`zQY$S;JLVl`N@i-JQH}K{j@u5 zg>_OVY1hfcYCu9Qf_;-|wB58d0q0*wndU7Ow)wQ>qC4{;?-rxnT5_V&)=aEvO!NMn z+dT^{0Hdwm?oy}nt6xT7%kzYp>=pGupIF+QeH$oS7W}y`{t11c-`A_*(0S2`oI=SX z&-*7b;<+*1esaBq-iJomJPDZ4$5Ngfmavv4IjY#B#BY31rhK5p`!;U)%n`K!RaH1P zzyzXzyhMJx|+3dfKBDZZfE zh0N=l;b@3P{K1PT{ln?^y1RAogXHPw{LBOxE~xw85KZvE{4;Kr{|lp8k``%!S6UMe z2b_Sd6S}q(7M207ay)-^|?e~2_!^T%Thf>Jb8Tb+q*$jk{nOWRfYnz{v>Il zxc+_a2k}fjj1EbVe>Vx0CY&bR$_uk7iY5XcBkBTzYKfBh-2L$iRIoXqN^ z|5PvsbKS>AYGMyxp8z2pn_(s;#-pC>#4pYQNV4ZFr+S;Yn-&FajcxVpB| zT&!=z6~M(dqjg!Qqvo1zA(RZ@s1B=Csj?p-l&Z-zYyr&XHl21cPr`w_o z;A+_Kd@k{zl2HL?t`n6|UkXFvHfP6Dtdo6j>>)*PjnMM210f-W@abwkL;ACv zS??L^gRXO3CfQ%ydwv1`{7t70NBx7IqrD9|AX>YcK!Qv_PXh)XT8u-QUysdE;sA1+ z?#v7NR`v+S#sc`}3DP#{*LUYgBpWmZ;=6)@|Em1kuTL{Z z$09leUdhE0{{5URFz?yGURp3x%9P2ByCDEaR~&x}8)-l8oZtlH9@JBb^lD3qVdS8l zrRy>ubOPyYM;H;jDs;;C*Xay-s5Y#*PWxuU4ewPM6^y$vF%L3TG(J|4N3!OmED}^m zwtr4>e9`@>8{6UOuFvw-|M$S^{3G9R>uA4qxKTx;D|0XAbyW-hA=O50XPNPPAa{&g z)@I{1(PfSB1;Pc`2kL!3ciHdKbTu#F~Y_|fc4^nvn z$d}D|_3TI70)+ez<6j{;qCS)@RuwK>dqqOA3e;N0q(`vOJ0}t?Yy64`iVgyftQq_4 zOS#;9IG8TF%A6#vIkS$nzP?Vq8D7SXfns9IodkHm>rOB*hrD0pvh8+us0(yY{;eCU z3?crstbowy>bIyyVQTkQt(=h7--a~&@9`<<3LBxxSN+vRb?y&X<&1Okd2+NL5GCB36c zDr~baxhyQZFNI49?RWQiaJn;&2%|8jNYjR}p_`?yOTQcv5G7j-yH1Eu8wyqd>4>5O z&vbdl@E-j0XbhyLiwqm`i<~syiFKW<5ej2`Z^D^?)224um|`fRdmX|>G&%>AE)&-6 zNusJz+tesoCj5Fq9dY6+6^*>y-530wXv_)S;lw%6)VLj*PIYm7%|Cl&HFlYe3iO+z zEO7BNS;qPCsr&j6y587_$E5pB^Zz@gWI}FeP}XEhZc*SmGr<2uQ|q!ogtX>rGhc7#>3?=c_3!#ToZKC?%u zNttCkPp#M8aRNAnGEfPCCXV z>~)~EfNCB(kFB#_94rn3T-iNUzyK$ZCynbF29u@5o~?Mv9#0oj;-9sNWE94jJ+`>C zfwjOU6!2#Xoa4q&{2R<mUSC z#qgLi)_b#)Cs?EeM@3i4+%Q|MOsv}{(&lrldYDhEz>=yEs)K)QO#oHrA$jEbOMmiiM?TnVc8e7Dt6;S9s;{uv#>#>UX-R@Z8v|kHN;%NVfU+1f zf(a66f@qbnk>j$<6h|npEp*tI`-xxe3fDkd-=|znQT*YL8g!<#weo%Iil>LNbD~%% zimV#j)R)VQOR1cK+6oqa1@{6e0u}e;Kf-8WbBaip%E%ckr}i1m2?cM`4p7=QK&Z9H zJXV)tCQaY;9e@G-Jze4E0uvla0+SO~q7&aJx$DfExaX?8Erm)>8;%8D#xQm}!SP-lU(q=La@Vr%jAqRVL< zviWtJp`TjoUIDE#S`U`2{!b-&D=ly$lbuH4lKiEX+;F7auo82zFmt5c0zfgqSlE+6|D)L4Ak_eN z3!L)|1f+EVQX}(QG+XoZ9q{z^@DuY}QqckI)cH+K^YcL}GO{vK;UV)v@2{OR2OnQ2 zzB_BK?k~9)nHfIE-4=w}e1Bp$1sboM*4sbGc3JWZ7dcng&=7xxZ*ffL2 zF0%!5(<`{`8O;Quwt@C!frH$G{h(jBzRUQeyCFVY)LOr~UYXjavdj%Oy6K~-4*H#9D(B6B4%*u!MT)0A$8P`yI7v;I5`LQD__3~-IPLziS}{%z3hE_wJA7-U~S&qBoiPY)qiM14GyHvz#zFk z1XR#{9dD-J{?xCDDT5gMJA*2tqJk_Uf;>U--G0d#ogTn_K?1>mnT|Fvx2_UxV3!W# z0Z_e{`CLN(@r8*MsX;;?1E5~N@LnH6kRiZ`nVI2?;G4ty?tUWuO~cl|es_BT$tUrQ zrh#Na!~s7iCqh5;jsPf;-WxUdqv9@%f?viyKssMk9f7*J1$L5WvT zVG?7KfU+_WV&LwK@IT_!o9NF8Z32gIkcgk(J6V!newH&kq((miB=`YBrq*N*MevYy zKY!ie{f+%^r?75+s$L8ke}poAV(xorzj_>t_V+G7;VplnZ-tx^;3wBkxDF)$5te1K z%!+OzfPeOsQ_l8SXu`tSv~1MXb4<(VtNW zzz^4bc=D1a!T9l_%2`n2vB(;W01|XCziURQ=$S zwn$4|i$Ai!X8LNuy>Keb%su7E`!LiJS|-xBAI`%qI45A-Osc&q80nZWrEWb2o*g^l z?}yBJR(^rnc4viS*I&Ed6B%q(Y&Q!O=Lv}`RGn9Mq0JGuQGbo1I|3|G?_FY^GVG_h zfrxCJQQ?GX$#y#*&#QN1R1KYbw!_fz?YH192awU9lbDcb3L}0eD|&yMWA^-PHmO&(5oqf1kUxgN27Y}^e@iF4>e|w78A_|Rnyb27vZ+u z#w5_G&32Yjld7^S)KP`Y?01Jhv&n{HbV4hs1oaot%{(Z$H!#he79!6sMyk8L)@4s} z1Lh35L;3F*CUyY~-luzKscZ6p7Z=|9@UO4PRhPff`f!ZJaOiJ^xkI7#I$;Fwz|?N% zEquo@Fb!EVq;kEzeTM9d5dL|~3hwgM=sv#6KzB>zkp9-e5``~*RkfzL+#i_tyCV)) z4Wl>Bszdo&xtV(=KT~DRl-~F1%4>$-?Chdx(8xL0yFA8}N->U)FyJ()E(d>4ti7jp zvWWF^(yCWIp2=y1yh9nKGxW~+f!P#&bR{l`oKp}XrofVW?`5ub@)rp3&!!Oh92YmL zqpFbHX?;Jme=ZmA6G@e4jy1x@i`Edyd4=r=Tz~)fe?*5cNmAjkM(r_CJby z!<(j-J7YIt&AgfM$`wm7OSr$RDEQW+Fc8lno3jTWDf)iif7S-w zHjX1AEPeE7*#|QnvaH74tRBlcj?uss!l~!OjJIR$x>09iE3N!N>7A+?oP?EK_ttaS zJ4@Xg>5LnyN5VnSYU$J$AQyR9OHmHP2u=%|Br001;fjPkCaq9pb%39LyJ;@9V=AYw zmndM%@;l|`xuW4a|F8r6E7o%iU~+7L`kb4jgUG8bQ6hpP?E@Xx;w-AJbU>KPtydf3 zd|j37X;%uVKdbO8M~gF&^!db_Ik z>^ytc`@44`U$Jl_U)Lr8Uy-)&Vrl5-_ zG{M?Tt^un!sbFyu?B+2DVzKQ98eXm>9*xcs8-VgtM9!(SYpN?N5DzZC< zI-I_@lhxTyeJmf{x+yz~=}%F2djFONSNk5H{l4gW;3xz0XBrtU>~zw_R`FMMgGvr=NlC$LG`5 zYh&FQS%#aTjNy0T|9UOqHnoNQ*<`&39$uX#K;g=OMNOMjb6&4&bf~_-_09w5Fl`6( z`Gn8+Mx~lSi)GF5Qk%%KG)D|t}(#Es)Uc>`VO-9yhd({`(?O3_FGGxAuaT-H-bY?f*AbKE5}tjdRm zCO-n`seaP77 zAZ@T77dR_HpQYwB*Nto#!97CIDgA~+Wt}6S{KubMM9HmI1=Y*;w2th*4(05r+qxR# z&-^T(ceC4(Nj528tYcAL^F^o^eQC7%x`Uj zNOWR<%q-+xb;Raw_$QAm@CnzdD!AXU=bp=|#ros^U6Pb*j9hZwH7dC`qyJb0$XCo3TXk$!;Jh`j@X{OdXmER%OaINIx|8`mFS1Ot z6QaTN`R1{E|NUw|N&djh=clhglCH7`wopaKUydLCX<+3q?M@ld%h2z3A)nPKt4E+K zTs?X60_WpIz2qH1e~ymx@rcKX)k|_Poqz0apzov1gN5S1?I*e?2*!=DJzoEu`zUYPoEH4(u}7amo)I`2qp^( zt)N5=^A_%?|G7zGtRhVEtSx4PI{z9q|E$~beG6rH!S8~%%4}o_qx4b||18&*OFa@c zc9nHY|DnSR3pCb-tI$e02^@b1y4IGH6oLG@fe|<$3xn;($TdeGn6WXAA)H6z$;(e# zRdNn6N1qsdsvev3C=GAz1Sk8NeNMmK#(I60%_txsL-SN_@q=@h_jrE7#RPYS=FQWr z_orWpe{LmK3$S*wX}pGZm_tu{(*|}r^5xiur~K;(5=ZX(1>LZq=smvz^wVGqCJO$d zZ`H>SQ0(nbzb(0g;n}xqz>r&-H|tl7{!N3U_d8dYt`f6W{%hT_-q8|qAdstsUcZ0z9qZ!#X>+hWHi^G*X9pyTgsuu>mcHk4L>H5^V4rz-rb9-<736J z2N1(GIMTXejN!ld={#j`9#sGCpJ|k({yiVwryWclF0XicO+&{X1#5fDWh+(7nvCP# zXhFd|PFy=%h&n=un8O_zQ^B>-mcrHcsIn16JYS=9Y7&?GwYng5BWZGj1U#gF{F&R{ zk4Ur56U8RYfIcvJ0oLo1@BC&~6(tkj2oGT|;kdhDyQ$(p_XZpaw-I$*OanHYS8v88 zLJKR;tzHRB_n~X6y6gwpJRO?zTs~PcvXp_FjMg1{GVKebmG|#Bml_z2mXUJ{FG*;X zKF8kB?|B+SrPsU#BI~4zUE3u{TK=J4iHfuSs9o@j_FYvLz`!C~F^lt^o>T@bM20ez zTcCG+Mhn+6(fJbT&EB2D@J!vWF_oq>?@Az1L(|pEVH4O3*;NB<)V*%{A385HK|9Kj zoX<56MtBDsQcX75`jspF89chq-s$CWkCsx#?QR@64GGHvz8w4ywXL}|;TG*g8hdM6 znc<8a=0Jo6UO}stO-dA=)ZV{45uV5EtEW_~l+Ggs5x;XqTC6@%;Ij%k-qa#p9tLNY!c+_W`Wyx| z&NyiZfv9T&Vq&XxD*XKb4T;8~fI)E%%jrVh`wC;gNfP%6SP1)MgFx2nAI{|^H3|z5 zTznS&FkMVMv0nyfYxpD{1B2Q_57Onxjl_FUq#~wG$Oq=&qz7$(Tsc1Eqlpl7-$p?d zp_e0ZfS|gUNdA%>9*mOt98HYoe4Dhs38`-H=`7wB9CLgPIdLIt!=MFJ)?lx;)33b0 zr&}&*OH6_*iY(!m_(at^1;mxWP4t>@CrrxEGtJ<1d3un)LhoZ*JrN5d->=N}Q5+T7 z;l1-B`H!+b)^q2V#A^O(z@{3~#`ItE;hK;n0-ZMXhicI&sJ-b_fFCty;~S`Seu~~^ zTVeDX1kKO!um|zye;2xY-?6qec>dkvi5;xK2Oil9xp25(*YhilL9N8wPxfP~KiOj= zs9r&~F7bx^aiLt3EWmvi=PYg{QQ*<5ojA5mJC>Fr_)5c-utbfXvRVJP>R=Y1Bt^6u z4>W@JLs(zaqQB)ckNb$KiS-F!dogmfi1IktMS(xCat=|(F!tXUyVm|n*Ay7Z52Y(t z%nq4%9xYQq@5j*p8Js37KJruc$dBG`wBZY5^jHF~CfPuZllM~%EOy9{Ap#vGv!3KK z7HgW4Y*j^O&J*!yDPip9D2$XCcb0J;2QJrEu@F#E-Z^|6YM=9xE7>Z>q{M=a3AI6b zTve0Ek6n4#Ry3iVKYn!I4xuI6yYCWd3ti^n>R3B$hdT;#Qz&cl;;uTp5V0myz2y7&hXL}F3i#X0b1Ip zOyL{Ayc8_5&?U@jx8o*r$Vky3dS@(Bp3;>EK`^8y8FeE_?Ke`6?`z{LkDH1uGH)$u z&zVLOlr7`)CdyLLXY8A^-V?rIdg4{wP}`FWs1+TyXjI>ski7ZVguf2+9OvsUCnTi| zg5`g!)1M`k*wfV-P4xX%;`Tg(14jQI-e|D9Pb9Z6m-xj6wI}%dg4U{(M1O~hHU6}5 zEoRWuH|^mdqy@b)LG`tW36^ESNANIIc@zHe4y!{HUjT`>Ho4iUiFhlRJl1whL=!zk zSGeGSw>$hK-k>A6!s<1{gXL(VW&GkJu3$8#ILzz>0s-=BgXL|Mzdn1e9Z0)S8z@C| zrxm_RxF%mW;O2n4pv_H&o!1X|u+MXU!M6|TNlovdZ zQ;SSvs_K^e&zByBy_ZVQJeia>Wa}IMvxJwqdCIu!z|?1dg4M?ha$f=2(p5EDXcNp`cbOV| z&(>|!*99iDhT@$mCjetQ8#NeALRml2K=J&mjxE)8CAYqC#&pw!5?EiUtB9x2qJ|r; zelI$o*LgHYr^acGqpVVMM}XlzW4^~Rr~#IA`Vc-dl%gq{y^{sM3n%8{pq5?)ab%HW z1UaLLyVhckKMF5aY56?$!m@C__Yfri9b-_pc@ z+q09$3tmKav6E%U6r2n%}7f@?vB#a zu|J(oOXl`9M@z&O0Lm6}ga^4DgpVGf63sIIoaMstBO>t&^Zge-*_tM10EO;Czhbqn z^3Ku#oIX@8;X~5Xn&*Q5K_)4(AL*_IW#Czx-j@lud7;FD4zO66c8%`2;w99Vv*6>| zuN59XMAun0sGjT(VE~2(%_$%$#vR$;*Hdmla*l$abh3Xf)X~IFZZ)&EyR4F`MFQQo z$v2XOtpyu`3?Hko{&(L6IfGkZ4s^8zQd8{x`TAn|Epl7~EWYG@t^8ymqF1gD&x4Mr zX%3sDXt4XKBJkwYC!JP4Pw;6f)5sr)Q_`?oB$PhKEZ8)zpJ;v1G>Ds3b8QU&FF;rC zY)0H`R@myKJ_3qDsj;E9vPIqyn>;&5=zz`ZRIa!&QA*ExZBO$uJ>;il{o2~5>4{A9 zCuK?Ank}bF@L8Z{UN(z?ga7{lZ9tO0{3cl0SDdd!a6((^8W(of61c=+F zHVOSvH)(pLc%j9Ac0*?=`C8VUG4%Y#SZU9oA)aq+8%io9_O{Z_wC~P!?yli^4fskX zN@?u8cxZ(`5=i{zexj+`DPP}RhBBtBn5$k7mErBu<<47&w`+pT83%hf2-M?;6%u*S zhx2~nfDP0BxxDneOWQwG0QuDaq61X)^L66WSH+scW1f) zJ=w#a8 zNGWDMBjm-$Kv@?3)o9V$-;T+BbDY0QYKi_MEhX@R~w zzb8O@2)^domF)o2I*}1C>562u1c})&UJnaF_0U*{UqD5=&92dj}h{CL0D{mo+`5ZJgj+Z%d;!c}KQuTtiuiEKA~H{HiZ3R?@)Oy$br8v-!+WBkM~6m~`K4Llux zdnv^fld#jUyx1q5BDokwV-N_=8PbtonR~cUajvTK_FynQz!>volQb!slc&r`tW4Pe z6=tBUgwT679##>+e_6ECuZVOhu_4Z*CD~61bH0cHowSSTFu+(PZiQm@h zWzm)Az(tEd_w%FK1P6gn-@1gK4C8&~oqe*I5TU1t_EKWFiHO$)a{tVM+VYb_xwVu` ziBe(H=h-?}qQ+Sq)rdJ8hTxOCQIKJ>0Xme%8?xKP;wQS7@U^BVAcf(~PwKjVrGmKG z3TNHarNtiUY^DRe{gyN3U>ht-yj2?-?MHfF&K>WM0W+DhUOyH*V5#v8-%O)xI1Y6+ z8Pah)8dm4E*I*W1YkT!3Ba1I{Gb5P&PV?n>(@rVXva+-D_~19U4SA#WAKajj*4<1+ z3pwQ=f+sH4hdNgny&2S^RaG>94n2jL=xGmYPB+fw9V!n-0_F6LdETA{Sgkj#ysEUY zkjFdFI$mZ-G?XULIL_GHNp#r3X|h#P!fAVoEUPHv^?180Cj91OI8%u`RHFDY&7qu% z)2gv?ywEiz-?ecqH=1hc+I8u8dG{;n?xqw)?&y^sbS;#_COLrh-vbA#h}<}SWjC}n5_jJNHOihqxVQtmH5|$dt#*MxbhMNyn&3n zAU}gO7@7|XtFt74HN}Gyb;|axnPZ~Ra|4I&~w!Zxw#o;yqNW{! z>Xt(G;Iqi7m-$&lDKb5mrYw=c3S;AZ)pOnx%jnQrS7(BI)g|0n=hT}#v$y9dPZO%w z3^w1G<>C%B1k-1}@fs-&%CIvIgi<#9b`Q(XRL-rAM)wP5YOvRToojQYzTVi$s(n_+ z^We~a=BTqON7B2M%7`cM*|4W zw~9dINMJ7(GASq%WR*k^FSSB<-49v|^P(+PSWm1wdT7IX5Ttc9I|yR`AP2R`DU)y| zH~8us$Ohua9FI1CF3H;bVk5zPY@!UBoNB7gk-xyr2H>|L*`*#QXj&EyZ~ENoz{+`Z z;E}0{Kb65EY8+eC&iuyI1C~+mYG4zE5InTmNfxf@WC-#}zNU-99^83B`*wj1;E6kf~P4olj+=@2Slz2b>e03ggFrbVXQWI0VJQl?K(uzzaME z?Tq@J?z+)AYgU*E1bAQ8mC{;U>W)*F9we>sdT`knG~#Hvl;Y+$)7202GzjOOszS$paoS-kbL`ec0Kry0#32{^Q|gA z0|tH0J0Hm0UG=vr4>_j^NIzoJS6VXnevJx52YeQP8ESM&-W30? zF?3CWs3zxJGw)g^I9PII1;qRMB;LWx*sI|XyaR5&@GsB!(DVNQeh>0*lW|}a0XCCy zniaE~V1XV2GdPoRniR7#e?A)lFtc%*4jln9lX039vy`_m904?wahepf+0(!x0XCCy zniaFJ_8J}nG&qxSniRL+6ail$0WgzsniRKFHUV-Q0Wy6af?kF*7(aHItzbCx2~v zb8u$w(ruE7ZR3q?+qUhzv2EM7HSxr@IkC-2GO?Ze`_8%N-1@3+?SC4p>sh_JckOzr zHaW3`gT1Somm`pwo|%D(2cRmUBG1gk%?@BtO~Gk_}4 z5x~p>U}0inh9w7xIyibcTUl7T0)MDXss9xOXxJK?S=m`R12i0L9o(%K+eh( zXzv1~2ADfI18o2808Ab1&8+^1lMBPY2-v#=on8MSX721@2ap$66P8d^P=5o6i!!Q- z0*vj=0CH0Ql-s+y@ce@YnmXJ4Pj_?x*Z*3!#{acU{%id=>+Jb&ivgCI8DM5*>IyIc zT3Fe`GX9Tl()Q*K0FHl$&DU}j}+0Z?`QM@wU8v;PMEl~yvg`uFzyhs=Mr3BdgSe1CakS7$3vfG!im z|Eveo|2+S(^#319M8v_3qa4x!USMuV`B$!v$6R7FS@2~&dxx4*MABAPcQ%R z|FuX!peN82c74UclsD8mqcyzJU%X_l3W9d-w~;#S3Nz^HOz*WWiq){Cnc>kT?wuNL zp$9y{Y>)~6-o>u;U4ILSaRLzjWNJE=I8uKF_oa-q%EG^H`%Ss%wIR~r3S0xn&Qt;X zI_lOnYhyiFiDsHDC;z-o#_1B{T<|P`oeqcZw6!Ln&puU z#-$&M)Zd>%Opc&@_Vl3yrQzl)z{2Tvey?+95mY|ac7NPpN-Uj=hUA;!{q$v>zLqkt zu~o4ht;3Sy-~~)S%0hXa^F8p>!zfeVSbBE4E_CyU)k?Skf^t^wmGkv-wl(PT$~-u) zYW5kzQ`(-!034n{gl7vsaJ(FNBkE->O5RdU z_+8QaEPrcq6S6hkWXganl8B{)l`&*%-0VN&lC!y=VZ~;=zbQR(V*2UpXoL6%{xHSk z{~`g-tUmJ@`;4J%bAZ??UcYiIU0Hqy4p(z`16gH1a$5_*vq!0%V zE_TQS2p*=WAw#P6V@Jf8i_er*++iSbDxL1;SF%rA2u_>iq}3gf@YGHbSI9`*)1u-B zfl*Aa!9HO2dsP?8J@gVQcg$jgdah_H0vjO>qrZNsWT`5U$sZ}A?HPp7w_Z7uo!FTO zuYaQLVTjni9sH*0rp3l~i%&2|Jv*4+g7;k#OC=1N=-ziUHBM!hR2}3q{ZSL@cZCEb z>P@>IKMkX0zEvJf>=kOKjIZ08zFh&|)4J2&Mz8I4z8{4Ip2zyZSc{JeB0MUy$*5lI zhp;V*W4cezM_x82OHITcYn}{fD%vjHX@B+G{6NL@>@K9AkjUeL7X?R%-9ZuIYqO!y zDbO4VrX_9kDn*v z9(zccw+*)=Hk=KPhlgXOq}iWkOBFHh_B;WRmdcSy@huc5i{WvKJxdtFuT2oudw@hQ z@LtS~k(9ai_4uoAT4VNSbgD|R0DrkQ7UioGV$6VwYG#`m3dV5GWF9X-g7kX#y4%{i z;O%eA@D^GVN0`naxiwMdrDEP(nQeI~)thSUa5GVn*>@0RdxMxz*Qr1tGPO(ym)q;D zz2#u{{M`DzCCTCBINq}@wFzUIboU1HWkJ94?!|FQmiJgx=EDeDTi&Wl6@OlMj>_l5 zMzs4V!#=z|JN5Wtn*w|Z!?E#tmuZ?3as&0>vNoL)4X#5T3~;)*L-g@u8IXoqM#=qD z4lvwp7%1eD$WtL~sX3&k$@jy|BpN}Ihb9!>o$I%8KI~*3!p)Jas=^#nlhr-7NJzTa z)J}vD2Vsi7US}0`Ry}lhw0{_KUI-~85No|eV?yZh{#$v>j{6ihAuL~8*#nWDnHTch zBG3^=Z#eloK6aw%Hi}c7NRUz%?0;$4Tamjg7>Vgp z|9d`{EEjbSy_HIp%>E;*DLyW=PT1|#5bz;*=J3TwF`O^$VNn8;fG4B6*#Xyeaw0inOj``I>&QP^3xSv62!k=Yq6$CbKY#~bwn<9$c?w5~=-Gkr&Hd+zhtSlj ziFJ!KH3e&`?g6J|`+tTl6TUMoZ8DTX$i~7~y>BWP#{@+D zybh!OPexeJL=y=-Fb>yohz`#9EG!()C?cHgE|y}%ccDCjh5bA%99XdgP};(poQ%<* z#J5$u_j4y+y@LF`0nV^%3k@u$@O+oy+~9v_vU89Ep2d$CZ+{Un(~k8H^aM5-|5!h` zJ`1WdhL<@7^3HdvFf94UV^bqCpGjfVbgT~t1j0aF0~bBgo_NU^t$TI$UdG|91iNv0 z$bB4WFm%#7Eha4L<{gqxV67V#QqVJL%D#~azCCVoCnv@wQGmrkhv(?9Nw>3@D_BcH zsKGBj0cfU}Ie%(-h6idQVHTrwqQoR4w0(W6BL__?hsz>ywB{a*5!BNQW(yE6>qiPF z?O{zs#y#`fELU$~NjkOKcRw_H&QVr*chTlzKD-jS<|i*NTH%GW+z2Imi}~2c5VRC{ zYa6Mgwvz3&P-+Wb+zqhhT|&0sC2}}{+)E46p&sL841ZU1IBH;Y)N*8FkESH@WvAXF z{OgNG>bJlD`gMd|ek_FzFih`qbX|G)=wWK$xP}~*j5bW?quh0t&TmKV7+k~1ccLm! zA)`(Lxv}((r_vuE{{Tmlb|_LSb*<9Qv>)bfX>$#sSIojTn4(smQoOFc9C71(_TM9H zd)%Dv{C~sgBZC&P$>*bcm4t-N9bm>^SO6#7$o`8YG!vT!({n^OGL!*}HZ=Ihu}urJ zbMvG=f^%?tAZanx~n}0NWO4z(s&)%Fm4qZNafa2u^|JJS7 zA{63RF0MChJ`C6!S_#!MovEW?s7YNpIG>d{vrtjV=ms7E)%lxcKH2T;$HR1~*xP5A z9BT^hO)^ATT86|jNIhTV;{)YVV#6)W9t(aRC;4;5lv87A^SQdTGh9elm#=lZK5vz# zd4Dx~oyRegRO-6O8xxo{vG0&=%C^X7H_JL4VL@xftxSPiiIAG0Q?u*&idW3aVU)7!>JC zK#~s%wW$?6To}eNal4D7K|yWMHU6|8^3q$RIDh@q z85E<^burKl9Na?z4(DZ0>vqfKgz6?c<2VC?lyoH`?Nn|%zdFSkf3`&f{E=+AoM;GH z&Kp<*jIIK9+LiNMzDoP_Onx+Vz^EISS(k8W`C`M6C% z$0L{LCgBPJYe;opcpFV%Szxy{>wj4hv%lTp6~U`%B|Jcv^K@t%FUGuI(;D>h=sP%6 z#89<7|H(jRcAuHwlaIDPOrH?c>cykrSNow*CzXNd8GdHh>I-Wr*feX>Ell>APJe3^ z0@ytBk~a#qUz4SR467ibk!iV;Bq(VTm~%HOT0N`>5+NMu>?EoxV~C*M{(lvJw9Uw$ zCdE-EklSn4^=O#-u*$aAvT3d3tVa>r+K^FN>&PA=4mr4S)E`?vs3z92LO2JZJYDse z*DE2;?>rTP%Qquf+~p*0-nOyNE>Oio*5UFB1xitW?oK^IV(sDlfbCz879r!>lBn(; zfouj9nE*z{aI|9veIW%m6n_UimIUeHNAA#eWZhkc#FY++m})Sy<0mRKitxI0eNG{y ziNCSAb9$?;Bhq9mG}}8Ch%`BEqMiA4n>;8W9Qa!uL?LdON{w#bXz~7Q6h7cuu6v|SRok7)`J)?Ce1KS zH!Z{Ib~`gFruPKL&f|X3xTl88THrY`mp`W`6R6j{vd|QI`nbC=1X9`!pZg_NGk6S@ zKBilc-uQq=^Di*xMl3>Nz~cuBaIzXhiyV2_JP=fc#Mf%6(|;(Wi#^O>QeREZq&ZOr z;`74Jju^90sQPB1|?ni~+33Vv+XaT$b-F}j~)%bf*q1b=XW6GE& zgc~^ll4A-=zJGO*3gu`yp&dqR*}ifvVT|A7afgs$;SRi>>qT`sd}E|==7%}L(cCYw zW)y`?oEM4hFgoqL2&lUK#rF{a6Tt{Pqb_o7qRqPtg~VZ91(&~c`&~i1M_|A$^!;4Y z#iEadZRBG}i>IDLUAD7Y#%2!cD2Q&3W&sBgVog{OE`J+VEo3>(q)5Tn2o3E_;g$`I z)nd|HFwEAYqVwK+)-~)GnutHb&(}Z=Wd2C)5Bdjx3G91#-Ea1nqBlMM&-g)}F3Xa! zw2JtYd5%4byo=ASU(sw`Q|0~n3<*85vb9t!+mNwmvr6`2v1;ZgQoYzaaaD)2{v%;i z>j7jIA}H{VPj8*OL3}-_`Wx!va2y|V*H80gH3m3^ zVn*3+zw+$iN*IE;|Kzg_aN`xK9)X@I*-$7l@#sf$THfI?rM&278QU$#3T?I1L0!4V9f^<7pa>s(yD2w$DGSPu=$ z?qM{d6_|^BC8JNk$0g$&ar+bzzi$f9#>|r8b7>r`+#uO*Via@IF6(B7wYrwz_E|&R zM>tnWvP8e;-!F63hu4h~Aw26`<@5pb{l=nL7@;5YuBCKJEq9~+tY%}=wL>#WxoSHb zc7Kr%r5?G;l2U>O5{-fkU(>U62oG6S>=tb#bhBr7lr3H9>LGVJ7T|b0R9jQdA3^lM z-FlQ_i}>96!n7WChqeaRe9&h`$6j(&Y@YNH4{*5wA(v2Jq0*QusyZ+z04#ikp*pYC znO!}bV=fX?MkBu@+f@Zemi#1+s=ibk|AqR}`6G(B)46l5#4kz933sfRwagLa4ltkJA2e`$kseBx3W--V5T z-kU-PYrk1rr4F&_8j~nm?Ce29B*9(ETk?>d-F{wjf|yQ-@lGxN^mg(H=5}d3+EQ%>ZhzmU$ zAfK5910}qK286A0jG@m&z6a+K)SUW2wm;zv>CP~s*|wJh9ZSDloOXb16mx8vrfyh~ zZ(G*rvY1n<6SCsY_DqL0otIWqo`1a0eipDYRJU*bH`vrpL8`jzhK6YloC~(UZs$#o zXMGFuip5sg9LdCXpL7~GMG=WAX$xbnmc!CbAjs%_2AB+JR0SrrV9JxLO<7&pojPT$ zS3ffgk#;|-Uzn(l33=M@EopVB0)=Tf%mUPr36kre(jL9cb7btB1Q)|w*?+g{Eol1y z%1r@<_C?KUwL2H?y!9CcOg>C|G?JZpx~ju`s$85|5gt56_jx$9SeT_nS*A(44rB1P z4XsVc>_7YoLc!W{@#}@x54~8HTvUX5)sJ2uNyw1_iB)euL3#+B(gQ&gI0}Run94A0 zLd0&TcS1|88>a(ctoTxz)PKT^pc!AJ5klmP+)5vXzjJ}Mr|yQ{VvsgV?1<-^r@Eha zFnHOr{jCJ1p8{Eg7H6Q*c8WqIb;Sc+Znf@$AeOxewarhl(h|dRVpTX=7iO;@ZF#YD zf~OtgrGX9-#`Tr_=GFLfR7*~_knk^z(lRXlQ(5&-X|1w1ZiIzq!GD>b9RJ+TC!sXB zP-GquLMKNhLT3hoJP2PPaq9%oh)a#6Zy!?fLH&$8Eid#3#-~PD@-lvHviZh0U0$l6wGxEd?N8`n2g70h~qOu(CDmbd%lXs*?PsR?wf7#vUL zlyu5fLjGH#;_5fON(F9$-mNY9WD#IjSoZx?OG;MUm%#-1aeuhFOg4tO(O{vzql{q; z5;Q5pKOW+7K$HbReJI}REmoFePI}|b4oW|8`nQbVFO@v@XXZGYf@>vL6F}8`c`69S>+P0X9t!|Cbra{$}K85KEYZ_rBAg>&%d+x zNnugZM@a{zoI^&2!5zWE$xy0#s1Nafp%bgB%XP=<46kYdlL{y-O|_2xpK1GPIF>BaT8X%6=J7&D!<&xAo; z7P7Hy=5%MV*u2JEJB96l(jBrRvL#s9e<8s{9TXu4S1APNN+Z~<;EUSp%c_p%6g)ay zYYAUwSbzFm6sIxP9Ura4#v;43DTReOpR_tiv_0$JUj2SJ#;d3DH|)#UiSfbjx=lph zf=s29KUP@$fe)Iw1T~PK7J*TEY{b@E1{|hj4U2-(VvBW~S(V4}^uO z|NPwzR@oe4xqBOyP?gLYk%4CoVsP7|dp%j9v-nJe5NIxxOf$Km2~MRQ6rr_bpxc4= z8FK5uNd2trWLX9A?5F(Xg5a86Txy<+=zl`oA13KBP!NV;t5%9rJhxQh+2$Og@=Q%U z6u?`1*4z)NL_5F@Q(Fb1#6@eItrC#Qn8JOn>)7EGyIk(YZe#Bmsg=2#a(iUR;k7_2Zt+%J0{uoGHwwte>RXT6$Uv-DS9^csAY$x_AoXLTP9+5S_%ZdX8s}+fOUvu z718OALiV@Q=W}x2My~oR=xj9)-t4@$ww!ux|FuedTjqvu`66W--GY}DsDD)g8_XJq zQm$DScl&jU%#Zv#OH&P(0$f57x^qgB0B;<+hgvzqAffllFKzsIwK)-f8REg`z9C$E z;r+7mCTP@fVB*O#1-toRFWW;r$|)5RB9$0ee)acHvK^MO+!jR7hPZ`hel>&GX zp`qYAMpn|1o=&h#G0y3r-@VIs$gDW&G3lAv^Wh7PSyexuJ3$V27IuQYT+;MfA z*^}8|P%I1@tByJTxO6(Sv!ov;SumB6mC)7BA+~rDJC7X+1RnV^ynklSk7mdYaccy3 z*}AJ-WT9u74&%T7WDKIiZJ_WRNfOu^ghTnPzp;FJ_Lrs78>BFhP>6)?xv2%m{NESv zBBlYNx9FM*OM~-LOM%_kVbFouv`A4LoJFihAMt ziFRBiqiGIpmB=1)@S*MzNEW$&p^rh;4i9H?8YCa41vjK@L^#XFQ;@g5y=%uoP5o%@ z^hcD$K*rn0GN@C7ty8D=IkQMm>y&pBLn{aFrHR!8$4?YwR!Ebjv3w2|yV{aD+0qym zeW#69f)$(AYmA805uz#1dAuc?)1E@wb7p;hWwK7V*!7)FiR;{WLj-5VcXlIqSb@$~ zFx}^$*BUF_Ny2foZivc)sd24}NQp*!t+IStSd<$Iw147q!nebiYLcu`!kWk?oXc?xO`&S7Tr zlFKU}z(-1FtY^x%O8c~|D2q}~bIfwieB45m+;6Uc@(` zKz+C6Ie)6LY*w!CKcTn0f%4&>aJj7-D<_CaNXqQ?{!-dS!{v3Gv%dP{o-QmFRyrt= zj%|I^?iH2dq52$a{oEuJC$Bcz1A%7_TXOrVRwu`h7XKKTV(OAnCpm*BUs7$HOO&M> z*$5`1UJ5ZZ-T!+kbSqG+1j{w8MqT@h%BD^x08DuHaRWw~M{H9`R7TY3yn$|n z_V1C0m@oYux0}|y`R0{`XM_e=d`~1&hEjsi=8p;e)$}~fb;TcIpbsDwY%mo5frLe- z34dPgODe`KSSeBlXUyBvn{S5c^nw&))YmlWF26fK#``^i5l~`JqZNvQE2K{#W#s-4tOI+i&Cd@^?hjBe0JTDY_x85v z#3=*tpg}=-qfztT%@{jQLtcNa-+9Fdf`1DW4HuCj;VT$LZcgE^`QI_zG+{)nR_|MJMp7;;1*$zV~}f@_gY|8;T?BKlx> zo6ed2D=g@_%20|s=_>aQ_vmfeuJ>Z4a<>x#CPtLbOL!W=jc4A1f!1TVmF_5N&wu;8 zn6!sO1$q@DwQK1PG<4rcyVlIm0tLPPed>?CL#5_SMP>oFdboll^zra^vCbOl`Fn*u z);wru2{_}UfZcp8yu|m91KZewSQ~%xG{}Qp-mtrTrSwoS>KF$Do#5GI^C){bu&9F5rGO>>MgloqN=mFjN|yU+tpt5 ziyu2=b8{lkM+vOS>&_ZvM^U@tnzGL4M}f_t9}WyV3(*I6vnJU!*ikuHrhhx~^CWBn zVOD}wYlX%t#8+Ng`^&gfOImm=159FYT9XiX4;r+(wAS1_t?JpW_S!23^Tz>;I28Wf z++;_IqXS()1l#gUit#by7hznE-$F%#Z}amv0X5wxbV=b^=KvmccqO|)`rD>wh_>No z{m12z+*oIE^=f!U6apcYKYxgtpRsEa13BCyo`*^k%x#b`WHnPZVfhwPNg;eA09+L@ z!Tq5&YZQ6eCFhcYT9cf*Cj+mHQTdSEnep4M6I*%#GHE@MacIE8Hf!wjxv>4mw zQzb!Q>0mT{+VqIZE4gl7;8BoIA@1|Vn_D=sQ0;ax6JFX003Y+-O7LasJqQqgR#;xG zF2wcbDNJpK@_h@~u2zguQdOpJ_b~WrQF@HW6(r&tt>e%!8-sfyHY0_Q#T0nR@^iE1 zwJnhxGz4E=9Dl%ZeYBTg^j4VjMAixS6&a7Q#;p|TqL2YPNcMXfJXlwM9*qe;(=*aH zk$7ULp%NHMQTO>!n|;CJ$5~HA5sJJXtSQ{5wZ@X-S0Q0{^bO8}uFs~>j0J{my@=vt z$Ntz4oWFh=2Z_Y7_}M?deFz?ptwM+tu&n zjG#zN#`w-fgN1~F_0Rcfr1I{nciwieZm1~(jmhKBYQ+a{+LSVmX$i`G@G%!syT87T z9!bhGn2S40w|_&wv!@L*)n5HB2g?sdi7G#&TN5v4JwHsFhY5~%X( zLJCnge}B~e0WSh8x6+zUKTVtL4yI8-aLC1P7W*n`fPL8G$eN13^}@WMB+PE70EJAZ$V!AKWe-fudadspn72plil^56L{ zT;=I0M66kWk#Db548qkLC`wiGBl~K885u+svoxymP$rw9Ma}QBF*bKN7L`2DyRW~~ zxV)vy{AzC(w7W`C*6E55s*c=H?BBjy4<;V#u-$pO)`ku;1RZe9S2`Ps>bjkJiDE)p zJAa8Nl6#dB5(DG(|44=#F=Os# z8>v4lk3FBAz*o6dThf&7<8ab$A=Sj~lz#>L;^0R2F^15Y0jGua*Tmh9Z1Dj*|Bu@; zE)%)Z8n<#1^6f}|ZimshmzcDzUjoH`UNiN@4nIRj8;nt57W25cWFE*?vYq86Tf4%! z)_8+>=j{+Ych4#05%~4eJqipdqlpqfkboP9mtErU_x!E*SRD^Fu-RG5N*oU3u76LK z9qltHhn}f$F6yUXh7S)>)V9KxFWnoOc86MIaZAa3!+Be|fr#uiGc|l1q4pof2*Wc>W3g3FDK!4`dPMGyD z-jqKlgd)sln~#dT-%3r_9Ai-K%z6_f_Ul7=^eE&QVJl>sp4(IQIyhRnxdX0&qT2A7 z3-z|+2N~jBapuwum2BAWZh#^=35d}b>HelP=h3TP*&FCJbPe0uy+UV!?NQ9=eJteA z%K2O<{U?RIX<=s_LO;NQX+{TxYz&nu4KV%r4413#x0+ZS!e;fxqfNto-F5P0rG*$Lh+ZH#w0k8MPL!IP zd#MQM^Mlr^?4`1G>3;~WdpXI!omCVaJwr^{r>I~yUXjU_rQ5UqG{VYbl z)6yt=%3|kC%8P|#FQ~z6J%q&f!WH4OE)_tldX9uEYXbA6@PCco8DbWxuInA<;NPI$ z4BC~6SytAQECI_e$6K-CL~p+oW}zEd_>&Od`EBP1N3KSLQj;u9D}70hua5YtV8kCu zjBAka12)${%_#Es;Z#B9A2S9+*@PmYXt4KAqroyH4On*zzn0d$sl3WXpXny-tu$pJT%zj-NzEZU-lBrA{nE1Pi$c3!f5|e#m z>M=%0_+ha{P$j!?Yt`g_a^*8qFM6jL211NHiDx9|%)bOOxna@K==M)0@P>_;P`)R_ zigqXWMHtkokN`WYy|Rn&Ea3`Kt<9XTlSz_@Nmm&>5r3&uv~+sYiZ6{k@xv1A?KvB6 z6+2t19xO=fj3?FOz>pbD2Nahc?CU_B%vYw|e;#O&S9I}FVW6k(-fT5dQcM=gW%UVy zgqx9xj+tB!-0;RoGkf^A~J!9@8`BcD2D6q(mHdZpR(U zIElME2umj+SVq25pyS8|8=Km9GWhi;QFaYP9M+dDs(YP$f17PR@UbJ8DlM%MNbG|kOgk%A$}qS{9dbe9de@?#NeKOx z?S2?iN(K=ypt{u^tpcrCblVXhR7yI^xP13k=YSaebg{^Sd94EW*~VT9posUOyHmM* z%zxOL)NN!fvAo$ew)^^_i^LM{l5+c+^Zz9IoSsj0Oe-igrPE3FzWRA^rNV2&bnj1l_P9gdM1nH@Bo@C} z8iS4ziD6~b>r5w28z}L;8jO6Yh=0eq%Z88hR9V~2EhC0K&mvvCN%(EPeVKUQS$oyE zJtlrcSW)S{p5%fgx<~-mNP8D zx$0U7>1#wg$M{hxd=AKn?7`F#?a)3WNS;h2HviPkN>Qyzx*YZAd+o4bV1M=M!~Y7Y zp?72$--dA7Ks!j42uPF7#Qr+l{KVls#zKZ+kjAa?O1Vh*MCXae_(4RCei=yKFIR-qM)!FA7JTUwSOLT*~Hr)IC&w* z9ip!Ph1eD;A9pq+y;h0gMZ!3gF5OZmaC)aV>^}XWIpKy?XfCr`(33_a$=+esi4zU47iWE~FO$~e|v-ZPDL4W(&$tscrWqICg z_}39lC_+J*Pnq<46PG+Kn;%oOD|>hjRa3qaA{so#*u&D`!ZGByLXwfmy=*LH-kvm^NjJTdlVYU<#C|&reZ^sJ0p+PgQN`XNd z=Ea&yk??#GI+wpz%zxMS(FqG`$?y%TVMzWx5*F8+UaHFs23giG!{-mg z#`AjBlsDvaSqE`QmrV>!n8?2-P*3uHk3xmvULQ31@JMDJnU5<09Jnqb`Mdb=emc*X zg>Ep`QYB;K?bS%>Q7q}&Lp48Ui&rj0#D1vy#q+&uDl&xZIe$Gno5l$S-ACeYbvZx1 z8Ym*PlIvqPX>WhC7jSM50(7O7V+0Ebm^!)`s$x2+;pT|_7uYmv_~jz9tFS*@Wn)%R z{T)RpUPS$K2CmyIRCUJTvG2sH&+aPKqb{nGS$kU<<1mv(NL!o50rBK801Ezae26Vs z8<$7}o-yC;$A4?}RwHUrWFJmdlMJ`~%yiI0^^+U(T-G3G*=|t=Nm#Ma<@g~h+4qWY z8df-EgATh$M>^q(y0I2-49y1mp&=gTMd_1448@>vN#{}f&^=iMhqKP$P-y1zvCauH z%rj?h{y_^=I_rf8=}Lb-<7u7GamGw4S44|RPs_Y1Un{#Df4CqiwE@Why9c_F&gu5=_!SHLwAUqzHeqQG6PSjH zp$Lg>H9oCftnpvEmt~D~U^n~vU$gw|vxI>jQks~m+Px+sdf7tsS%C0GUD%~8sc0Ic zcc>j32!A?mbGqc@_T~(>SdzLaNbX1wU^=wS46}pW3_n2 zho7`<4Zhn)L#CIl(VtOB)z|;7z#iv~1&*B-9Ve~48GW>D3y#>(*0Oyer`<%_y5-$GAll)y-jDP5c^gHY>!;xl zHjsjZ_54g;b9544%u+Xjb`<-f&VHwxmLZ$@TUIREM85ZaLvdQip_^jE2}F1etkq~hVX!?$*I;rv0GQyC&83w2Y5Yv;iV%w5xY;v zqS5gn!sE#73R1IStJYS*DB5QEBmwFCWW%NXJ%w zLOUCKk4%=$<#|?4O6iaG60THsNQzVez;4?QtyNC{{!!J)@UUtY3|Vi%GgTSa2>Rrtu;Ag`C69L*q&v&*%d2CB>ucj%Ot;#2Vt;*E z;2e)&ZJX5Wa+L-jb@K(AMX=I=Ap{^|W+t*-W;fBcbQ-;(z8`d-nt-O@BI1~_iC-^L zbQ`P-;*@> zXmz9)vxIfPvZj*QCb7FD3lI<&+JEHlFQ(+3(`Hrrl6#JWf<Bj;S}3 z^!0IzEF9T1aln5lV$`?=M}gQUN5IgoSAUZ+SVUneJ`1mW5_aos(i3NU%mr zE=Ri!Y)BUq8)n0Jl9Fw%jeo?}l+!oJv-WtuWx^#Vlosv9;B)71C2#U#X4?qwrU@H@ zo)0z#du%=9HB~$x!bf3V{Ae-=(wdBI;RvZyjO=I|uO&`$zSuEU)l6@a$%@brFwVHk@783-l96aZa7qQ4j$`9~=h)G11y_pm1gR9+wCW0YcB2F*f{{iGVd8CsO zI1`tpb^$33GYT(EWo~D5Xdp8DHboEh$}wGv8bBdihH;}4f0 zbVf4@W0TNUn52aX!a4LL;R7yN1g(VeB6!=gUL>7`jUroIlE?}4oX`pg4%9WKIDZf} z30Dh}RtaEg8?Epd_c;q%IO^Q<7>~gShw-F5mIF5lcup8?jPMC;DO@wI4&$EKOyCFX zgdhU=4GCA^N0WO2Y{qB=bB)CUI8;#y{8YbxN5MTyog@d5S@q^kI)53 z*cenV5$ib#i0HJ&)#8f9V}^*z34cs=4)T$x9_xWM=d%^b2^VY-SqYF=B9n+w^)jgd zd?ctqfe&cj8}N~%@D|KrDeqJk+2Kk`QW`RYd#x}tWa`0C1Zk}l8cFpu@O_5zwbmji z*iFO2L1hgbQWY$Q0Yf^$g+!wZp*bj!`~bg@95qNaN%u7AAb~Io4HKY|seb_%F%Z*w zYG}xz$e#=vh8iT9p&G%(aeWWDktQ9LZ-h@j)tr8L2N51`Mw~X1G8q zC!cf)#?YR&h=R?vkF;61g?|}7xZL~Z8zGN{d^|as2>C#~22)Br7xK|?@kadXUl{Mn zli#O}kk1AeO;6sRj2F##F$Zx%?8)cNd~!8AZRWJ=xaX)j8x9U9AH-|grZ$*96itrZ zhp2CQpf?+g=c2CsoK^wx8K??Vh}LyG?7aHr-=NSZXE)IjAu+xhjep+U0cxtKW&>3~ z+Rt|28%^A8;eTe;fKZJG!NYKnt8C)otp;=+g0Z~ zqJAcwQ;ctQ?kI}3IDdhXNY%MRG~Gw%HeA=ab$73G>vz`qj;NnW=Xa`Gr#Ez)?_Re~ z?X24!QNIX(MjM&kwR2Gx`L-XKKD7PF^0Br9DuT3+z5Ir~+-IdHz>mxZ@kUr!oIUrm?Q; zbX{t1Hwxyu8-E40w;Kh2Z5s!@v;KEP{VT)njkfqy{ihB6htHt@xV!%2mmWs9* z(eT}5a`tO8Ukon?o3S<8oqcOZ)X(HwH>cm-e6CwM;eQMl<4s4sPN@Y8#l6a_v{7Eg zeD`r7?OtA`zqGts>2tdsyY}kZaTnXIeBe{7fF+}FZKM^6>^ zfQ3$hDcinZgWegy7<@kg#y(2Y5U9sm@Q|$PTYnLf?$hi2+>aGA8AUV=Kk;;k&N#pk z{h(t%@jGTrC`idECg-T9v{8cvn`o&Rh$I*Zb#dgMfE8O#7%KX-A)&X*!)@v<5>!2T z1yVjMtnak~!P+(?hxJQMa<7RR*j8DZ;}bASPga@iP-b@U3$$5-ALmGFr(HkM9`-!G z2Y=K#qP(HfTBlC@a+;9Z}-C}0&9%otfVPddj zWJymOFnYE2)5?)@>Lq5)JBVO+7VWQm15E4|n^2p`JGWdC_OEVX^!y8f_UgxBN`Jh@ z4zS`^+RHU~#R8KhIke^*hq{wYz^s-o?aSq3cZw%u)>I);sBVMW72OsstS|SGOTJ;2 zqHl5vxy_!!3&Yks=~F3^x_W9fWh^qyLbNE5O9R;Jzaza{g>}fO@o*M#>>KbY_&*xKS&1l!$fPv|32H)|nU4G^d zHj+nDE49~3SH9zakrNzAfBW^jYg4u2D%DOHxXHpCu#4l?1j6mAd9EjwyK#B6Ul5+O@THHtQ_0wc-g9w-a}a$3OOYog?(yYF z!6y+!jF47%$q+p*AAR8RC2=&-FCONF3*u->0eze7u~D2$$Dvx(r*lVpl1&XWY^LY zeOS%y;1_6(=j_kWQL?C`<+%YrSS#%cHK&=?YbAQE@}QZY&W6*)WL93YJRMxp$kBI) zC%=4s|LEZ>tuQbeT+D?l7>E3J=^padJ(KzWK!50QQ$%SY59X)zBFSOO{lWCxW_a;# zA#g*B(7?T*aQ9>}7!6Mk#uuZ829FobiS1ft@h+mQ5^p^9E1W#K9XPMXKNq@JI=j7nUi>J>Z&++j#OCl1nKsh5#`Mg~o zy4^01x=|YE)}$HEPOmP{N6iP>{O4*gT6ufNx8-znHD43C{8PcMB0u~0!^7hP7I}*a zgKvE|(8~b5x5z1O6`AE+m&V*`D}MLgvnNL^&DOn`V=GUL`aHcN@rR*AlsKYz$) z@@M%Q({e=;h0UgU2M*FIzKJ7WX(&@qcJWzY+@J< zRkj{FEExBd9q7bwPcIagM$PGD`ul9LCY8Tc;&<}h@6&h97(oj=6@hDf zb@{fL&4(9bIgyibgGhFBzE0($NzOQiU&-I(e<_m7|H=O~v&p))Zx#BU9vvKh`vTg2 z4zJ06OGr-%>gIppc3x!HDbv|gwm8m5lJQ5B-AGc!{8~Ww%h2_I>(c;hmoYg46ag@k zahemi{15_3D3frS6qo!E0u8rjIRf`B0X36xniH4&5CRPaF*PwaFqiNY0Tcx>H8(dk zlc5kNe{H&BaAxfiH5}V^?%1B#b|$uc$F^;IV%xTD+mmEsJDK-6=e*~<^;Lbf|Mazb zbzQ4hckTUWQ;lvU}a`zfujJ3+BNZBEKwF>_K;7QP-VJDC z3E*>ca}#oNc4c&O6=3|QNX5(y;9_Y8FbCS00YrZ)YDp_d0;nYw)BuuZc4kgSHULFe zV;i6eKn`ePX6I~112DID0@(cL05Gw)GX?%jlQZML1=zWnIl25J%-qS|79cOKDlG9+ ze?b)>F3O}T3NW%W1;|PLQ*P(t%<~W1%*5IGU)|9IT>cl?82v9Y{$KRJtdqxoB1SkC z7JwVG_>0hpSZ|HF2*u~9Ix zH3LwK+S@v~x|lfumaGKxYY{hncA&(B&V$&5dmS@#){VhS`6HBX4F3bhZ6Yz1cse{4-WldpjG?|9j;> zz5M$^CN*ItITdxf|82p))8cj}_NG8P3xJBtKZY7Pnf?#>cUsX1_@5Q}ubBU?e-wb_ z|DENHT%3R&03Bw=f9(kKzn=d_djF3lB4Y31&A|E(&cMdX0$};U$qC@%=JNgj$Te|w zax$}X`M2x;^zz^HzqiTE%)`tCZe!Kngg3-0y)~@DPrPWp5|VCy)=-UZl?8NdruRk% z6*z2ZYH)mocdv?_gX?$Y2jD9^R86z z))1k84WW)}YoY*g6M5&7xw#RfNIU)GSMGVOjAI!;sw7K{VR`2OB+72W`qHDFB#j;w zT9)=Vg4rQm4a&S_XyLW7;OoH#Sa^Q$KEe$Ju7PD?MWDlRxYh$ZbvyUEf9r{s8hI@R zyli*nYp|w>umbGMjD>d3o7Gz&D!j>HP}EU+61b6Q!B<2DFI;OWzlexFOci~?6^fqC2pIBHK^~pVYaJ02r=P!IX)*$U1 zYFX4pimj~&YtlA~Uu2dzf3YdY@w-qRZU&H0MOwJb4&WcjF1oD{m(Ri?0ziyLCNZw| zOr=n>I{${a8S1cE@AR!Nd;3vs&j9GpV)YIk(L6qLy%rNEcZ_7s(Sf(PaBiv>; zNP)h?7o-f|1Y)MaOmAXm3Bsl5n&gPSqhEeRSyuf<^=)8G#*N^Se|^*qPTR6&2-R(F zw)ELlscpbg%$Y20mDpin*?^Qm4*S*p#$#!cIP?z#Y~pSd*TTQ+fq z)L-Q6V&Nc&WGL&gBHiq}Uw`%mJ`Eu+ZR4T~Ui6rY&U(TOHE_Tof^OlbQ?;X0=c6#^ zjV}TK(uE?-p52v7o$Z5!r0c&{PBNxmvXKT~3Ue~$ zaand1F|zA0jSK5cRUFT22#}Ugg{@d2$D?%hWEYH=$KH;UTJ4bQen4%SY^Dk)75W2C19+R^|8K!^`(UCI2j?30i9mi&{f=YSR*>Z+R2sR__XitJNMkk|Ee-pkfLoEBTIOm5s;%NmO!Zq_V zsY*BQqZX_lauIdLe#Ukx|IHw!)$~j*h>Mqmt-ligIpzVMkyU> zc*rU{e_g80Cv<#i=e<31T;AbtTuBgUa22H$T}vL6t9vEI@xH#kIg(?VC&uxjnE?{( z=%Z4h`LJ#Y4$nd+s7C^AIu>|J4Fs%_p7vAuRzgD-8$>09hu}GkNcQXU2s8yQ@w5*b z6MTE;#O~r#Q|H&u3vy-(Ri*caxLAJ0PWdo-f8p5uIC3d55S^#c}v{#8eucWEk4Q!5ax)&Df3`L-KnQld>=kR$&t-%!;yg}Hju!2*>mC?6 zF9)=`AY<}1&$VCG0SJ{Mu|x1h3r5rw39&REx~K#`4f8VBF>RXW0)kQ0Xw~}kCc1HY zn`eal^jqKtGD*Zwf!z3t+7yaO5IpKN!*NGh5E(i@e;=#F&U_b8C^oJVvJQO6f3=v? zNUjJ8WSGichL%WgVcqpDx<{1Yo!4)=kWmfhZ{u7CDOJY#iN2<;1%n2_ zA(R`a@dB@vR)3Id(7fIfdE$-oe|bEC{vsfy<*7{IExdK@dlOI99ns@F`B%Xr?()}< z>zc8S)twPNqHa2BcynX@25{FF66X^dOOoyMLQ)_v`POBEwiDid{^IL~e_{2VxAYML z=E3Yav!BAMbCBbT`7WFs?E`i7r0Cg(F7KYvD3l54szNJGzK|bBcoW?be^-dd^o}M* z4OiUGZ)b<$&bgx?Z);G6TH#F%^^;Gr5=LFu=|vkJG*^wDX})nzCsz&x(fHdl<9s|D zNlGPa@YXP=u=t^2+oY3fX0C_}v;D<|Y-;&A zF6#T+H$%A{lqr2gYH`WjVq6|~TOMs;6|YjrUjWKu47}3-=ES#xXw+iilX|Z}AhGRn zij(DvQ>i7eLK%C-&kqt)%ySQ(7*yHliRFEHyH>0|oaeNZ8V|jj;}Ai4syvW>mpz*n zS4?b*HEUf-%!H(57+t$PeYXm+uX(SxgHhA#%>#W09q~Ass|8vbwDYwg_)%Q7S%VGz zXK0q#!C%j!MN3JCXjT-7&K~a%IeTlj7 zy06C)SVZkzoP;UqgeXv^N)>jt@{XeCZ_BP-O)$|0)LyqssOxo$R-`Q-)0aJyAFb>v zO*JRDXo$3_uFtOxqSj{Z@EradKI7f;A>K{?6X2oT$&d}}e}e;;{?A><7JYwi_GZIT{Yt+wwxvrJoY~SVeM= z-l}GOlEg>oR%kHi+03snxg*>)y8}mDCpP{%5dqFFhe@vbYtUFsRT;!*)HCR?yHoF8|@`g+(LJ0qrbB_mHfvVbK^bu#&&+se8?J$6QRGb?Xkb6Ql>4+ z=wskt_Y32Ya|&p;kDn)3FwMP3Bd3_gX!KhqOPG|de`dQ2GNb}GaDp>ls`mRag6`3U z-nFqLUq1Medn5NFFhd~pWl932(zzH<_%zp{DZnMj16YYeV(^rFD71wf9zXuf&t@E> zfx-{yP=(}l@KgT196`Z#$SQLm&#Y+sJEFv0yP;phAL)HF)hQPMI%kq zXDTUoo?c^akv8A#-T{r2baNhHcr@mSoIc{?e~QAhl;%ocD!O~BdBuPx+uT0t8iDh)03`i>~N=lJIQLj!T%xed2_mK_)} ze@t3xIZ&v|B)lov!LuX#HE3L2u;*GMs6b`UGF$EXQlZXT{w#G9Lnlg->utM>AGg=Y z-AX+>p!?(}xpPv)2Zz~7Zef_5DW6xc%MQ0Xn&rt35Y3}Wh>he&-vP-UE?z%FE-Dy# zTRmy(LJJD$m#+7Q1}9@{Bti{QZldNRf5#-RD~hFoUr$}JEU!g4WN74wAiC7Z+Ni_Z z93t0Q>G^ovN{Qd^7_`oNTfNnrs=A;S@F^|T4r-dX%n6Vwet`=C=xvZ#r7d(;LGeos zSvh_E9Hy9Vh@7U2=@MKx#jSl=w|aF8{s7;0auby`NX0=ZFpwD{8Kz5WqitJoe_fG& zaBt%fU#UhIx_5G&^OR6CJ>+9mrvPhOe!Z}?d>E(OGzaLTT=;Uzv)B@#Kt%CRFh zVrah?H1XAf0bjxGZEfGA!`rr@`CPT0G)PZwNk}CcGD>f4-B2NUsCV6!r24`uCc#9S8AW2j_t}K_@a`k;qr)$pRL; ztgg=P=g+ZMAFI_QbR7bFZ#aLEz1jvD)lseuq^q?t%MsX(;d7E6c#PuCm@A-5Nk{VADeljKIJm^9RFle>xyu4MM_sAW4DPX)Gn2AQb3G zk30w}tt|bWp)(jwA0BO3mFnYG_Gz%wO?GUuibh;#LeYn6djsi6K^zUH!>uphTP>3& zg6tpt^p>wFbD1_Bf1-X4hhfpBW=iVu+5~YdZmHn-Wy)aC%oz<46^KRon;To>EZJwOhc0fCHOE6ryT6ub7?577E%WL72n@2G+ zzo-T)nha^(roowKVotDc<2nf%mGk2OJm-TSb%NS&Uh?lNf9i<-&+VH}ZAWLk7>E}3 z?G|#V!cSCx@;PtJGT7fL0X;b+8?`YghavMQI`dj>NdxlN{0G^5t)q~<%?_7YHCMc< zB=XyHOQP)!HNT;D>77DrS4Dg|{iQ&-5Ezu0z;kW)_1du@&YfeD5v=x!UeO4=(A27qUe+TAwi0cx{-<$!UpSSf3_fK-Jm0WomZw$!(`gnu{-Dg zDi}zdfFDNjN8|P>dOK7w29}Cf_6vkob_nomUEm>+kbJrxW|DXXzT?Dc#5+9|Lqy&N zlNm|rxWXb1?9z$EhS@M`o5xH2nNMw!*=>o@HujWhlpTkLvtsT^Rfdp*ZFp@}LNY9e zs-58vfAol`-ItlI-%I5Msl-XW=nV4~NGzB|m33yt8jKugc4NSHC&RZ|F)q z1!i@Zs$THp;+%fTe;3{-c&AjgMl}bpw-+wDQTNfkZ~)m9*P`G0$uTnx7WW|-bTtH! z!npAl%}T58=ghTm%sfr`(!W$$cp1Gpq1mTce`J5mxI4l~^-ffqO|K7lQXVqv|64>n z820YF!uSKDEA!F{=Qo;W*}@BI+!7II$oG$PQ+5BARL7R&g~>mAs3C(=;Z zf2@Q-;k}DJ0%-j?Za;alQ|`^$;{k(AQ*Wf9Ej9 z-B;cCXQVF9;P2PVC)3xChYw1lq+$z`IvUZ)qu{mb>kL9>{*U8!f7597JrS)~62aWO zi4I7^nNI@dlUCf z1O`*rBYqUTTLHT*|6!Li|1Eo7^}65jefJT7=zOBaMO>zylM5nbwavchA@!(=v&kRZ zgQ>j7*q2+W1(66yGz$=PbI9~rtF5nHEsC=%%llYfFBpzM;*rhdyJ(OQe`EE}+x^MU z&U=w$d!%Z#uMob-;_sf9XcoqP*ZPbDsW9@DJ$%Gu=OAwpQuLX=F}UqVy}PZ zwWAuAes&GMeOy@SBJ%VtrrzKH+Lci=YaBMq`9!mqV&>!N)~oz0P1TDwy8k|rIlHGp zg}D_pckyx#bvl-8%Ye_z~Z0N*3wyLk{mGrL<#g6tY zOHJTHv*o&)$p}*(T;|CqU;Zz9;x+QEPv1|v@K&VcrHl+e`I@kC;A4Jqpok% zLtOC$IYi|2RgL^2nQEco+`y~aPIa;lrL~B!es!+hQ^F#w63N6c| zX(G7ATAF;U?7yf}kXuJ%LLh=3fz0TkX2L>0oA1RWWY?YWNa`Ssc;ggbbjKnE=}_`w zhduK(hgy6&})0ZD~27_2+{VDLPH7>UEzogM3&tx)SXg_gi3>%lJW%oAL&0h#5lG|iD&l2=7D46eAxa_C-LfjVN?TAGnD1%Jyy_3Wg2BC${FwWEfwM%!J z17R~dV4?sPGrWblu69TV0l$Avg;GR}(1bYlbbayoJ!k#WC65Y}Y*$iNT02sWhgS># z=X;z8J5PUN)wFWxrJ0^OmFaFzM+4<}aHaOz!M-$zf2IsV-0{~&eO>C&`lvvKxoMOG z3(txJE&gC#gbVK2DeYz~#EN&~D;?m2c^4bWr`apfVVL3l=28J`<2eCC0-InWXmiDc z;?8+D+S>L+kb_hASJnRv8LYafD%m#%>RE>fhoxzBzeK%VU|EGxM!~@1{NSd+IKRND zGpjotfA6c^&9E*xk-*i9B&9N*Atm&N>v@Em2}+ZVLJvRT!tBC)3ioKLK<8BVcWZ; zI_Y57C6kXn&z_Q{6u(luz$VSrH(cNU(Uf6{t|wN0y}GVz5jKI#Dx{zcZ+)=M5) zuR(*M5d%Ri+jQmM1|Kzz6(^}jrxT^5^M7xV8NO($Uk=$zmK#xOx?@U-sbIJ9W4>gW zgHc0nnnRID@sGQN+Hzm58P_~e(Ov2LpLWMBAhcGC%tbq8A0Pz&$v9d!I8QBKGCzVICEG>s^MiS69yyMvk_f4Ae3 z62OUc%#d|GbRF!JvE!jhhSA;>;&xpt^0 zdk>>@d?-Up(gD(t*C9l1(b`Jd5Qy_vOpdJ@u8>$b<}|N7xyHxj4v?q1d}(6D(D0|l zrR^j}HylZD&#z)JzUUOX1Dq<*e_71}lf!*i9`o%pMX@!otl|yIFh&~Z)rqvFvYQlY zdvu96x*xxKFD&HDip()e1OMENKS9UAr_SkS)rutKL%zY@&?A9`l*jn#1nH?L^Rel8 zzk-Yv>PVfN4`)q52V0`#h`0OWS5LB{l!ak=|5OWrY%*(m`Zajt@;h?vf0K5vX21Tf zaG&du0&?h}>`m*MPI+;Uqx?rC_QdMy2d;zh(G#foql7SJI*jz69ne$RTYnlw``&D! zBCX%&?$ngq)9U5kP~=Y@%A8+m&hx_7H9S0>ID2gFRa32+iuWF@J7}CE2}z+{r$AdN z$(*h}T?yz75fkKH9B_$le|22TU!+Uz7ab5?Nihv4Zj))CIcQQ4^H5e9>O&1_cVq`o zV@al}uIx+(D>X>(yTlpR8h|W<2HB0c&9E`*MN@XKdoEb0#3L@?@Na;}vrz9VW zeiqc^$kBudELjqkf5G`{p#9tfsO==f46pZYbX^bb84rIi7?1b|s;MiajXj2T-v<=u zR%ie@4qKk#<1l>)tPQ;fs>#oAWd1liinQ7ulWPPLFOPvA!DwznyA)qOR-F@lNIr}TIfQ8_AP9ce-d7owou;yQsUB{z^U9XLw-PA2?bpgKH+5cO1}R$48*bcb zH%6ezT10s{f9r3vsf4ZZiakv%Jt-i92Ax;WDcm~!t3hK`uRuRbC80(Qg23JR*lxi@I(y;?pBb7IESf*{e@O5|?&5(U0NlK;Gn-Y!kM~|7 zyuC2-XH~vpH`1R3qbhK5)jCExl5r|U&e-&aC(f4SAyB&U$E$87q~~`rua0QPHiFY# z9HYsS_#l3c`0V?3cM*g&0eBe7uJU|}u;|PgpkNDKu(?V)j^e2}?e2@#`$A>(8WJEa zVuaHnf8fX!5czQ}Go_QBEi|Hhd_7yteyxt(-qV-iVv!*C^+(D+FsnyG>ihasCHM$= zU1P9!0M%ApBtmDAU|U7v>3jpQ#5}bcAk?|tRTf>qi__%E!qxQ(cBdAeL!FC7nb^w3 zzNArdBby$*-F2)1N93AJ#UsgyB;=q?uN ze;(^iHs(%nmyg~xi8_w-$TXcuC(>c5L!_(f9{O&U*{!d`FM>*Jw_81$fRmZ0zjqxk zQVSJJ79*CFO>13q$u)nSS!1N%u^qG_(@EX?TvoOcHryAxUgv7q`8Xv z7UpBMK>GIoq8E>^{kd;#A%K+Gfi@ixfiB zSX;gAyJW`?fB<8A*HRUyn7jBefAAu^wHb`Z_Bf(ze~^PU6RKw?9{kw7bwh0W;Cj|K zeY`yx$cz8)6oytLVc8U+>hNbpVL^E5^0qo2vBFgXHf%-|`8cCc&e{|X2bEY@0 zH?x5QTiRyR{*gkvu}c&0yL z94BO^rkZ+V4PJYy+o8W>4q>Hn&;FGvk~6{_+dNm>CAq|@0o@F~f5Lx<=N-dGwTcs^ z>2r0s=0Z$fh?aD2a$~Dik-C)j?I76lP^8qg`7cxCy|5iR#%PtjD9fc`tQZNG<5)f2 znWddgK@R4vUx<_QpkB>a+d#<|Os4}*rWDZ%Stm*-VHsSCvFstMcTq9A1JcnzD#{Ur z*aXu8X_OT=`wLF^e;l9avVGra%j~?$t-q|jd`EHI=gqm=^!hJrE|m9K@U>>U?TTio zkVBhd?rYtH0jRF%+^+&S=)K4-@rr5#m8moue2|Lf7>yE0N!L-u(7_v-4N)LhvheJL zT2x*6-~%{komu_UB7A;sq-e4R){$E?i_gj*HBSkZAsmyGe4Xs z%WnAgdMNwMEGhDHB-lnZ_({3$6IkmbowKcY?V;mV1JWb!Arpe!`o+;4aTtv|c|Uqr z*q^C4SvkYY8@WfwC7sG4KrttYEO5%Spx;E$7Vy>RWlp;$jq&WkzO~X-a^spfK7^4Y zy;nEt!{PG_e>ZI?lEdHr1P-`SDgFS!aKArcX8!bd;Dly4tKZ98XMY+B&zFt8ocS4q z+aq^JPVKD4MY9mCMOvs|mivg1QHSGlKBA|q0QxXlnuBJfGDM@Y;(mWssMPs3ChE0) z;mpH0PYj?XgoX<7B~%#e>R#`H67Vrr{4OC3_5j;Ue_&1Ant3w{a2Zu;BCM(xUfC0O zj?~L=pW4}I@v~waQx5~F1LXb0$nUxVM7RM7)xz(3{4!`G6h`{$F*@o!{ zoVSxce;H+173rmd1CTjLgP}n=>S^<*1Tly$b;}s8AZhr-ze-s(z3JYgQqfMj+H6f> zNDRRS5%#Vf31nXeNOcqB@v$puywtVQ8_j)^k9GSbAA?dnxd3PI7w9GVf#Qj4XuIK$ zr0{AB`<6s5*?49UnO26C4jTD~MGllT+PJ^he{kyyMFi;#5RTjt<01n~j@GSt(Aufm zS8}pdEYBt-hr4gLZ*|#OFl=%gD$KSvE&C2ZiM|!|%fPc3sn%&h@=zoXV>xTk)5#&8 z5@7lrbf8mP4SOX>Lft%K)yHZ#tlB=cw^Wz1UNYCmu|G&?<583(^ERJLh9B|8Qhq1K ze**W87favuoTUUbS!+9$*TZ8;VV8A`r7z#$4MdSCi1N?+VYGh*k<^WFek-1`CCV#( zKZo~6fQsE+wyDi->gL%kzu5-e3%b%K?>MbLjP6IZUY@fzRB2nd97gf61Q35D`+%IR z-wv1tmtGVkG|9IIah2fZ(q7c5Nji~~f6YsKgQ7;XpWY%Uf%*y(uv-%wht`;eO6{Cd z-}+(p4hwm6(P^l^IDsNRC8ft$eYEvO74knEqLthK*rluo;Gl2{`MOK0`{$3$a!n=G%R$h`#;6O0bobf&MsTkNTyw=4ym30hzfF0D<-@wXts^klK>+Oe$WJ-R%YKzi^pZ zet#h?4>v3?^#obf5#5pd`&}JuMvJ1 zAkamm+X0zvEdm*&as|A_VM}nef3Ru>ZK9 z9d#~(S;`%jsxUH?x)9YmltDOe-~BVQg0zI&**n~dxh&v4Z^3Ikf=9a5Ve^P8oB-%QR-tH+g zyXdYYlhu}uLk|EW+bH2PGmv&r)Z=YuGUk4X+3{Oid^AV6-^^vG@u{U@+|ZwOGDgpM zxtNC8>(51A22vl_Nlj;JwJ%z@^SJhU6G)w$rf-2hN#n(8>Vbar@Iq^44Tv8Vf_SCl z&6)0!>m|os82wNA`ZDnQbH|M(%u+j0xJk8WSgDF4wn?G zYB#r+2}AcCD>5dM$MOMXC`#EYV{-HBfTX$ntYOXsg{34pe-xHNOXNy(@i?CMivU~0 zOjMouWhq1%cy#j}iNuvNdl6>S3bP}C4)L?tuA z@P#KlbDJgrsa~aX`|gfPML5d6CDpCjaVv|nv(_-auV-xLEiD|=Hs2_D*_!AGEh3wM zB4oin9ju||e`vKrw7*f`U+=d8?T0iAkCT}1mqF$-#p*FNT-xTNP%FgD7BZRveza2e zscB>gRmMtGTPo@h{3~bEw)~*W-BV=;zDl ziBfNhbv#~Ln(^LAsX`Dg`Y(_@t&GNpIyyuj&9eqGf1kh9Jj?XTBTKKAtoX`{#0`EZ zw3EXI3OvDp>r4M~;<%@BDQsV~UXWuN){5f~J59f%-ynZy$Bf8~mG?--1|Al}QTozw zN#>1*s&T88`)iwpIuM!7tpjbn6e zvhTw3f9rLFC9(A^dY`%URqv^54$YmyMVj>n_oQCGULqeD|T zg5r2eqT|9%5j_87#eHhGgO@@1YUNgiY%VgVNI<=G)TZUy6Ew5v>yyLDQQ2eb??j8+ zV%v&yfm7G!7P+vzRO~Y=ZFS@?Q5B0XYV5Lhx zB!}-UEou;4VYDk(bF;dC8otE)lU(@mMXocZZ2C9(8er3?rw(Ty2iMb;&pt_MXd;^= zMSYY+f+(vs@h+2rLoBEp$HDk&OJ*ZXe+C25c7zdTTJ4p$q|?>sOk7@PnhDQSg3H*K zo%8y1*X!evm4J$goD+IfKkq^7jaMGxvH=%xr2Q1aE)4v+3Kl)Uh4N^FiR?U&Zop-m zTjOGZmXrHOX4jsU7^E5g#=<$kI(ZBnLhWAffpEF8E~b#ArP%Ng#_WhR!=5w3e;q9i zCK3c~I?ev)$FLh4y?jZ5*;c~r=gVJvrAMqUATroaof`^n zqQ%FV1x~i|bdaq}#9nauf_JztsVzqrCG$2)Z=%KMO zZn#M4vcru{wj>-1zOP`qQIxX-JWD(-HahCuFl{!mvHb{)Ymmh4*8K3S?Jm*TsKg4r zxzd5k#Fv`RA_QGcN5o*TMzq|1gTA$EXtnLqg{O(ddI>rX;^1r zkywqy*5I093RAgUsDhy-MP#HW{>?V8!DK{Ilnj>UMt&1Tru5_Mwmmq#W03>LJKdrE z%n_;*LsSv68y}fMhKBb2xT=(U zp^K$hDYfHwFj}aX-4==WXG^0t8dI>0G9AFEP`e;x!Eq+HXdl-_S3nVnsthJ(BQpQR z$yMOd+ZC=t04&{8!}X?$i}K=7<&_FY4B4VUE6^K_0;uOcq4LMy92wBYkmgs4QmdNO z3Hun7Unje5bzI4lg}b@hX|$7cKn_g5L$zb5wQpdwX*m~;BCL+9!1&2YRNolamQ;)%(M1;WpJ&dxf^NJnFn$IO zy8Kv^1e|9?66f+dI?0h<+f=`~F!X@y*68uTVYI-A*<;W9}D=aA=y>}OH zFQ8fJBIf5UiSR$87{H9S>qxImBw_ZD<;b>dR&jm7YO!D}|2S(_zeN-1Hasu0p*ngV z|6aHH%P9SA#d}(Qxt?kVDS zX!%bnW9`?w+r_L1lYX6V7cWD2B;|7~oTbQgEP}}LJP{_muk#x`I zlABtV0pnX{ODFxVa;`6xzJhWic6rSQ4r3EpA<|DaBePT`>gvY?>%0Z(6>)C43{vT6 zfMe<_$#|nwlTL!2p<6r!kE{2~p0&*4lZrs)4?giE3$fhds)BVKt#nM5bROPPd$qcV z*G~&mx3|~CSORaJx}9NC!rJ#coT(sVGm`HZgu#O|P%T9)UWmyEwq2ECp0&G58Ihe^ zl&BgsGeNQpIaZRZ-RLQskjQTg5uhWa0R42To3_X3WJ5M;Y)Aw4q}Wmu558H{! zw;XD$nTTuMa6KIn2z}c+1nPCu2Y~&o5&F>JC;_mShfar^Jbo+ktnpa4^~|#4fwtUbB${-BPX7Lq z{DUO+ePAJEA*O5`S<4c9DM)c{#~@jE*K_wceX~cRBo5KpUhTRLw;U!KV1lT7LkVKP zW%Hfiep*AA5h{^}>zPn3r0P|Gotz1}-R0cc3s{Aytq^}U9dNH)-J(%y%-BjHS8YFI z3x54_iFA-;Rc;fU*yC&c!2@68*i;x9TX(GVt$`DFdP7j%{N|~EJ&EaDveuS4YBK!rhc77#`}OWtKOyUXPXAooBmQH7Z%85*CwuyG%0;S0C+@|cM9xNk zq9iFIMWJM*FN|;n#7_%4&b(q^GEu3`XXl4KZv!k6Ofo zY?x#&mwtrZD}Wj(dw;hg_`!IpNR0&+;$OraxxMD>TOG`kg^5BG*p=hCI%_t z7-nCSk$+-F7-hH!YV%xW`h01nvO-SPLvyxzL$|k1j+Y^}RxZb~G)viNdgohZ$0 zpj~YnYY;>O>XU1(d8BBs-|xbDHtlENN&EuUf)hjB$Qz!|u4=mCIFU0fp0Et1Y0c*< z@|FFs+KkM9dP_qbvOLG5s8)(5!)PpNSiEr=8aooQ&T*}gx-SvMp`)Tv#L9PaIVDEX z%jf*|;oU4~!wju;Xve-Qr|?1cfsm>8jYx5H#OXx?KI**2{g<~@BzO)J6f-foejV(w zD(jLe{BcLHt356z{`}#fXHH4UgId(o2(kOqYj!607b1eMrm5b_>!U*Mcu8qjC~;LP z+5{QzZPa0Bj-<^=5ww4L`ZRtwml@Zo#DQcK?(Eb^S%=R%A>)^N-9aY4s!= z8p%*W07M-so{i&Vd7zzu%s}pjgY(e6$yT(q2O#Keo} z#qL*O3XAI}jl>&L>Q9d0KUXJ3?V`6d!bW4L9wv-Q6NPz)KiD`#nzT8r!Ok|O9-2nH zgN}ktd5@4KF)?elt4BBWO;oNe?vjiGk6k6|fP#CQmS+y0*%wR>86&xP@mJ%&$sr&= zY?w{-^@U!Cm}1{AtF!(gfTHgs_~W8NN%ROIOFUehX`=Sr?+LnzvRD$d3_BlUiK07F zIO{UA#SeJXUR;cDP*Zi3FzzwEu=79{)=Ve=AV2Pt1KsXb>>p%31>352BFmcfBNJv0 z;0LJKhn^H+3NQ0oS?*kQ$|2WACyrKCL&>dXCRlV4OUh};X4&TU2m|d*MpPAN7&$X# z{s(96#swErqygL>9O_s9#0I_Dr>A=_0x-~xTcAxyManuN#v}K^T}VIJ*z{bNi4kw2 zQLQu_eIv<#VBc8sV#KNk9ib5n)tm3R0_^= za3%vQ_Wq$p-`_-C5N(tnsKTX&o^Mlc>5d$pb ze>JOiGAFLbGDeXMILC3`<0L6MT;{?#J=X&2wDtRQD`HTqGd+qg5A!qLFVu#Pd}dpv zmTc#yYfL@4rmmF)s=7HtYqYo~mXI=ghP3d~IoNyzS z_Fa&7Bc*PXezfde*vk?F^6B+J+i3y5Y@p}Vi_!F+jF5TjE;u&}=C3$DvrU$NW6dhu zWG@Nsp>QpHa)&YB7*GzP8RWtlR2o@rG7ri6H6hes?eb|GB-%)4guB6KTN_ zxzY^xSNzZGOj&I$H7g2U!3RN_@1Kxoj?Ok1yPCnB{MXAJ4|Cb+`0ZPO-qDhB|5PrB zrMQyvyfOtZO-0c)6j(oU{P;_5N$Gi{>cz4#e>#u`GmN0Cyae?YNhrH~K_3){B>pd1&h=k>e7pz@ z37m_K^FQ!7CKwxAat$F3;PAsjIY+OC;qo#UGzcnuThZBSdYiR-o3*)3EJ4`W1D26Q z`g;5F(h)tUdxGaE*X>NB3yr@Zb;QsK(Pad}6BFY|_DITL*}?ewBJv<|QrH0-La^QIGd*DM>6BWHU69<( z!TQ8Nh)yf_DtO9ofP=zZU4tETR`onf6+{ghJ|Jm5?+0{5`8E;f@cP0-^@%n}!WSvu z+!ty6SK-$m#@sJ)d+VGW5cnD%2xF)^FuwfSuY>IN;NN5cPejYT6DM1{cz(MC5Erb{t)`e{|An@ zXu?ZD^HqNd+)mL0M8j42Pq+(qkA4Jg+6bVE~ zsATJlY-xRB0CFGXW$i(z{yqDvLxDJfY8cHpf+rwyVT!?ja=ynrU1d@UKs zG(xJA4@2<8B{E_bUc36^LRr}dH#yON*_5_TdMnDw!gwqlTM0iJob!WBFzs)k=-}U% zAm9x!Z>^r5B6M>}Gsa0O%S2{nYxs#BM+TYVW@J&rHBGuYN7>u2^V*+L|0+bOucPd* zH|p9Mq16%o9ehWB(_lf<(nvyz(unVH#t_aLnqycU-$HjoT_XZ&#yN`a)?FLQ_?2E- zTy=w&;%2`_NW%-s1D<7)LdP_z7lde(+2kHLOC5N}&0@V)A=^};|5--RDA6;O5{CA> zFNrYROHD~VBpTONm5BXStBJD85?OK6p8g7toBsx`>`llbZeLW{OOW8bCYB19w;;c) zdvE0^?oZ_aMNxhrzOm?z+TZ6snzOq6re8)f!7}OJ6R|Kh1tvO`LU+W$ROg3N1R2-? zXYQ%n6(iwCW@As5QwA}!slMRJKX}iqQWizU|J>NX!I+fW+>|plBtyU~h>C752Dn!v zC5_HmI~SB(hDN{88|hO zW|@%wTpa$YG%n>tv5>daH{i$w==X!&#G}xSuU#2Jg^Dp1`XS(g1#A z@UAb}kZ<&(9T}RXKN7OwW-@7)a=&6!-9~@ScU=4UMq0@22tfO27?qbc74KcC0DdWc zF)SsSTvrhlnW&p7u4G*1)2`3CNhmIG#|CT@TS&pGB7{xqekWA%?Ww#ss`i_Lw@eYW z@r}PP7;m9UdC;EL8a$}EhR!rYw(zJ_Z-|HvTfecZ#+z9G33H#}g<}lgecz*BxD@r7 z9yhdBJ0q@#Ix`vBylkXz7vYiO0?M7yj%eWU&Txc`3B*}Stvm{-S)gl8!wazO9>{!F z7#4%LQM2>f$)2<5ow60kgd`cE?EWFW;z6fvjqL%L!CfU@J_*Q`RA43B2_@mDNiHuM zEeA>z5v5H}`%JYDt@6uzJ2O}_lAa@Ti9@5QP01A`*we$>gW6C#8(|GCFRSyE0SAbaJ;4+-!L{78fIINujk`A+VAidc|A#VU2O~SdlLlpHe^nG_tP=S36 zuF5gI`JTe%Pn{}At?9q|MeR$N=~rMf3$QTZnh=(C=<5c@sO~yjwxAW90N>8YB} zD^ST4(z1dxcReI<6_}Tm6yk1n9Q8vs{%eq9!yGGKS_`Wuz>}z*99@U(IN#A{V$MfO zJt=Cn+nlEp_3-2HBM9-1Cws|aTeiD0-3iPfQW`^7f29_Kl~NUxfP8&0;w<7kJ% zVrGZbF{{HrN`mU(ySJ{a`@Dsu9Ndge;) z^2&^2Yqp5!OO{rwQhIEKdP**D1k%4%T;1{dF-j?SFD$aZJH%iox;_)e*`;45qQ+#Z zzTq(tE?E1t14nAM+f{OY58)v^R3bgNRL}})vrMtzLglp&dZeuf%SDIv$4k=^yrI5) zN^>RDURuyLCBR6Q&Xj{S(|CK?a6DOXopD&e+OMnPaX66v`@gv`O6!w(-O}Dp*?!f< zw^ZQr%OqI}IeS-DR6o;85fVb+?I1=Ss!2qKHF(Yb9dy;=i7U}1eDu~yP?x-ng20rxwKIB684^uPmLpSLK^ zQe)MVfieBA@21=qc)EiMrXyU6>a#4tM$u{9S~Cmy&6?5rZ%N`0zdO`@Jr^x-hlY-rtEk%Tq{IIIU-4f$G#7 znu4CLF&j|Iy@kTQ!62wtS|xE(S5oxt^w$d10%F*wk6c%6u2@cpVQ1ZbbSOfkOlS8# zv$I=nL;5FFkjEx~uzzLEHo9|BwvbNvCP8psuHoqumAJAKJCXM@35UkM766fBmwY=ZI8Ua<_lW-s@WpX{=xZNaKfB;0|4k8ZnYkUI1@Rb4~|kjQw8S?C&XLtuGntSh$#X zyKw>EpG54Lc4fRCd7_r!I=7YGH%1YBtz}TTTsK;SRFfnX3P0U>U6Zull5>oS5+W(P zp-ol}h_~iG*i`gzi!d0Z@V!;#ey4$gsFKZls$Vh*@MrgS&;#g(FJfbACChl4HqA8n<+r9a;v{)R>RA6J=*Zr)jjqIVEN%V)c~!ZDAaO_Q#!kaz416g&8KHcGfRt^Q z3D!0t?6p;-thMRTE}MH9Q{?mHawXN(Fh6P(e(;L&&mxAKjL(Jin`*B|-46h7VPvqE z3)%1~EwVn^GUrgzI~HeirYUtIZ>oz$qP+?;R)899?6|6v$Yn~`FpA0*Uwka$zLD|z zVK0bphMPh4)@g-Z1Xv{Q*w$Yh3BTjkp`#e7YD=Ax7N*^6O60C#^}#l9 z>5vOSl);;@94sIZ&83h&!bEl(c711dzKChUY3nVwtD#V0NP&MqSPL;>T|#O-25?UTiD2X}zE|+gk(ff}jCo-Uf7!zM<~uolHGu?NW4?o!acbZP_T;F68(;9SD@k(% zjmoQG5X|4sfuc1&Bj1vOs)>A*#-U>3nGpCe8p(F9J|Ih@T%ZdP)+t2(#p`2c5BG*as)H+M^P7bOurd1 ztGWGJ3#gCLuw9vQtRDI|z5StE9uyp1*s@;GkW2}QZhc6p#q~NdXG_ngZzHs|Ku|Ne zfZI3Ek&8e}m+WF>u9kts98G|2ZD98_?!EGJlM@B@oER1!riaPFV^U;n)QL_n7qDrsYqZ7{7!Nr`4Ml2i&43J2@iHiA8i%v+3$Fr!-u{~^5fV)sp=bsvs(HUw`0sedEk#xzuIoJ6qVR^e}zx&UT zFao*)2jR1!zP`(G-LAIt;d)ML8A^nzS$Q&X`8Z8M?(OgZ_9T3}*;cug3i4z<+&mi% z#y{(WEAL@FgnzI_sx0HacaNbKf<4>A*#SJtaXHbd{HuH!p%3lr+Nr()mw%`oIm5?% zMH=!J*uu_OY4!=CfKIQ=)7y%08OK@eW&R&u_ax`}rz7xb(KB4Q!QP~oFyokPSR{xv zZV-t#7oI`i%ic5+bMTMi@r+DvVTE||X)E-N)Qk|oZq0{ur$;9FF3<|8Pr0~#$N>=q z)e{fZA!HjMcVnT2VvNnrmUwZ4 zUuAVm8`U67w9?UZ0K4_=?K}1sG$7udWoUH6eQJwLf!$fxvw4^;RW4&5{S9uk?Dty@ zJtA>~9W7msv{}NIaO|p;XP#v1>O?+$+0Sa!R4~a_X-~z9N&qLn}=sgd^|e>6+^Y{9UJS_pOli>GwWG zW9q$Eb69WWa#zgZ19^3j?%%s=$;PSHfn7Trg&rKzQN5V0S{=w;CbxS!4NQE8OJE-hD92Y`dnXeq9`v%Y4C zQd3y~bsdwiTVGh@e3jZ_?zz|!bgw@Yr3b;25Mw8lTb}$eZj;0(dCxo{OGvKJ5}z2h z+lno`w7b})hAyKe%oULSnuMI0S~+VRsJTJ>itf2|d(qNFGi}RDGp|1oC`7c(Q82P| zL{)MN-?QGwns^273zXy{sgrc&avbQpYywKtu5|cGybpo~=6C_*ud7gVce-k}T)QV2S{D&{?n5SWEYb0cgZYiZddyKK49-g$zE~X+;K{B;( z#B6oa`WhFD0Trj3E8c4NDTY<|4I>*LQ(bD!`a`cefyD%m3MP`KmZmtSCrCQ5hC7-U@| z-SM9E;n$b1;`Hm_7MIjX(UNZD;1ikqYW70#lD|#O3KAX9F%hq-ccPvnab|rAl4ht& zh#;aRfP#MniCb8LGhUMFvEQ-?V^|&G>|VN3To0p?J&V`HOs6zC`ZAvE9zl)27%{cm z+NLZmf(2?+n=>lT3nD1)Pq&1kI!^~}#YNa10>dcufwt}LlP8Pjk&XtBlQ>%{PMRI# zdBQ^xcM7k=aG0eg`yIA99HXk)5)23s(}3kQfM~!vbL#=YoldiL7rjc2)g~-_8g@62 z+5U*^QfYF{6XSSFV#^_w(}{b!zvx`tvvu#fHle8N@W+cMk1t-@k8PZQC` z7Z;zYM2uJveRh8>Cwe6EvJlm$Ycaqd%xLhUK@%YL>Dr82m5ESueMe_J23Jc1XoUU= zN^s0YJx>G32p+5~e=?tINlMEH=~-F@yWSK(CkMI<<;ThhsuH?XsdqTI2ZXnrK;A;B z82HQ=Yz_w})^pTSobiL@Kz(UfnrgbF+dgUS7{)m-w2-@TA~B+)$^5+V?C;h#F=%J< zN|x|6;DL#&LMvaJB{P!kc5L4S8nC-0_sZVy!?pC&VdP?@HBue;HBd2KDOs|NxG#i=c5 z1ry$ie$Vr7teP!28LdlsxE*GGRMS)X0eRTZluOOHykFQ8X75f>Q27suO2W$yu>a9r zYnRcko4u?j5B+11(d*B(Pr{v7R9;(Bw7xTqCV#aLhIWEy(EPWiT{%2#!JjJ1Sp_)% zevZhQBI`&F=_B|>9R;r(e-W+bo z&v8yK`#Tj{jmCnS_%qFeZbMWn6VeY6lH@mJ?Y#HsW3witD9Dm$0CsBh29UZC#{wZ< zkhNfZ=u2Z6uS31(>}sFQ&NcCuS6T!Ciaj)A)2L5%hC?vQoWdK3J<2HuiF5bAaSIZb zwZ&U~J;FCHMKtW7Li-m_yBBP4&IjPoX)gQ0gNkX_=)$C;z0>sFrkaGTX41!!`BMrS zOlhw9#;uubhCU}7Kw}*;?@se&T86)Eid8iluLP`LU4-eWBvWnP+Fnl+a#@Z5=EnKv zl6$T*Oh&qN$p8~BnWd1Ng7XeY8g`akM%3Mqwh;f*dQ+7*>3iEbXW8m_k(3ZG=j-C5 zHy4I1{c@}cVPrC02x1>k7sM}k3ByfE@W!S(vN?afvRYzZVEW#>w$XAQC!Mi8-Vz_u zbOZN#ytmZ#yQNl1p;tn*su-rvA}thA;a0(TuJ$U0Xh-XLBntUB8S`8LWBN=~!DVxN zjk?6c#eEkIIGUTa=P!bm?N^c0XKFW-vB!~#_#Il^;ADfB%qpiy+?jof!MQ}UVOOxh zzZ(B=^0|tFo}$36Q(lr(p8_{TJDP_2qEdYEv=$MiD#!DiEgK4TlV%SYX?k@!^X=H^ zb9cuj$EwBZW840k2G0~_hBd9!E7Lnj>5NP&ikHI)Fmj4p2)@x4{hP~@SBQ?I#B=(g zrxeNLbM98+P`v(bqhIxSi>zo=zh&IjUtnW2*-e9)ozN`2g+|+>th$!0uuEHB;%?11 z0l&&bi&4qdJ)mB=`TkAwpUKz&dW&M@eu?VAU%%%@I3Ur@bZIb280NY1*m04GpS{9p zr{XUzpj3g8++&SpQ&9wme&S3&>Gm3|tGVw&a^1-hO)P(K8uS+owmlhn;I9x#Vw{;O z&6k0W1#T0WBQxe~46=w@ePw3?s#jjv@f`$7!)dpr8nTJproTZGW z@tnGBh7>$vB!!k$m~t6Yv2U^&y{4b&!dJx(Fy&52BDv_jCa_^mGsffN0K*ruzUKek zErUy1{Zg6xynu*^im=1Si9B#VK$Rj0+~B}5YSsBOf_;Dsg>VdFkJ_5a;mMe?#VtoR zCgPV7!yKJwpW@>2FX!C>X#_EuXwh-QnJ0>RG^)4G2K8-4Ap zz|I;KwxR_r$_RDToyYfKux7JFIkXCF?qj^-+u?j{s%Y>e2%*PP>H&W8bV15`uszTb za2dF{l8a*Kbs=n-{lKjAqw?$z4}*H&KGg@YgWq-=s*powPliS-&?h*zsMu!l^Q6b(W3J(Hy+uRxOHH>hCbol zhK$&{VTXz#Eoh=Jxvh0TWzFj0woGxqEhX6BV6XjPa`2cb5~2e>DvL0@_;1!aL@`Me zmKo8D5A<#}=!+@|3t~=Hv|@(Sf`nLRvs_bGPW;cKYj4J`BQ+!P(|>6V=HYF}0l{r3 z;aG+N=UhM~17R-D{5D`h4ah$rNMKG?fB3x)L z+HysoJ=?lYYr5D~kByBQUb@WuGtY&f*;f(wT#C={~N zUYrcfJz;+(mgah5y>zwphAU{D3f*iYq>J8|JNS~xCwG{E9#FyGSWUAVF}PgXBB5EEH*G>fV8@hEahuG+MH&fthVY7b z##i~heL_Y;!%smajS}gD#xjoT!B!382?{A8fj|oZM~;JV78AVBRKhbBcLv1*!$FaMgN=hH zmI|nZ@!d#}rUo4Z5A_AZkGM}z5+Nalz2yl&PZ&Mg2EHZ!dkW@@kcUeW<6UiXIDb9y zi$3#bT9l~p>h|B!>#)mr;VJG+__-|$w)w2_T|Mw`-EA+BoXP{cn&bd`k`g50cBEHVhq?xf8u?fHXlIPcF zb}mCFZtZBU?Fm}dBBBSw z96w}HBTPza_8N!lNizL1X3<1So7mIv)a;p2U8WC9SQg)eFi9O8OJ$W%3Hq0?n!D+D ziK9aJ!<<>86-m|L4Z=y+WsBpg>UHuNF#qY!pg8vtNuqac}wv-Ri7qf~Na26bVx>0*f z>J{-D%e2MzZSwN^tcAS(`Gdh%Ij{t&eTu-nQ|MKvo~IdLKCXk(Pwn}hX&8Jq376B+ z`{3`!e7p(LEXboV!$b4y>`#ZJR@Ip)aGyP8RcSbvztlXbgS$>ub!iTcv{cy9!`Ek0 z?^Vzvw9smZx=FX6bJaWv7zX;gEjctGwi-cVV{SYYSOrhwUu!iBwhQugHFIurw8A%z zxt$6pe-!w9oW%$GEaR<}chtrh>Nv<_$fu`sUBs;r+96Z6yE4M{H#W~01MiO)GF?8lH)oGWvcJ>xmiQi? z!n*`6w(|7bFYW=|pS)KR13o8&V?}K$GqkS?+;OVSWE66Kd$z(r!IjqQQo-Q)(6h`I z&UpIptkfehoE7mM%1dAS+v~Tp#t5>vaxilyR%zY(SyP%TrNtG_vr*lE4qD)kP$sGA zL#4nlsSuUQyXuy!$z=ANx%>K9uf&fKdmUk?)}44yd`rNEm5iC1Hh2>iqL>xV&0+yW zK=co9dXKg7$#%KtH!BLtffmfkNj)S40ZmX8W_f8ZoZ;oW;x2p zM%F+}cDNO$aU;ufwL>Mjqri(vTUp{(@OU7E=^ZD9DM(>5S!a{``f56D`J$QY$Skh8iuw zj#Fzb-scB-bo0QU#UaHzM1rlA^o~q6`Lg(*EL<>d&h({La1wx<%K(l^&Boh;n1!1Q zj!BDHkA;|xnB`wn)yc{IUz44fnOGH$NygFA>0ib1|0;1}J$`NxF>YoKZV?tvR&GvF z77k`k4t7=+HdZ!PR#9$d2~I)c|Nj=Ge+S81I9j<|6SH$O6DQXaQUesTGHx*{7=t%= z>rEy5l9WH)$rBS`9AJ2lLx@?XSR;s8LOJL((gZ_e4-8aBb=7P0Qy~AO&~Ek3N=AO> z+511V`%U^?bvO~Rja5R{b0bo17iAtSeZ$Cyp|A$Kvs9xNwO zLoAO&G>=QFVd3ho*8n5s5VQq_X?WCjCvf&LS)uk!QTsq7>(mB%uL8tXox!Mi==Zeu ziOumD(~^phpViyGiOGX-HBRSgsaEZ);P})?!g&=(9oX-1*J|%8P3Iv7HOP;{l4FKAq5s@cAAvDjGQ-jyoR~flosK^E z@ccMQ-)rx6h1dgb4m6b-nwZYSr6`dk5(Va z>xr#c0voBPMkWFq$%jY$8|j2g0+rHAL;L~>{*ON1rOKJa6l#sH1uRc$f=@DV@tdA- zaMeOSqZpmQU5eWC(}RZyPM)}@QPySux)ySuwPe7M7lLx4bV4;mbT1$TFMCpZLmKYMpq*Qx5iYu&7M zKgL{hJag_xqxQX_{GbNsO_@cbPf|q%?5Wo?qVhWB#+E_$e>1k^C-?)29_`mS(O;A;+Y5ymoOg5m)fV(#HWV z=Pw%-oj8#)*0FYQjYpkAkjnC7$|cL&48C5=?2mOL|J(SEu4LMwfapNWML&!o?y=&JCGAtQM1 zr5m)47G%QZ=F7wk9fqkQlgsZ=r&DjlPnV|1KfR8`d<8=a7bEd`Gh(NAlplEV!Sb+I zKNpXXVUaPa&lW}RYU89!L(7#2bpJ{i#p1f z-Ku-9KW!6sc$+?6ghx64Wc?+g3(j2*ma)XIokfw@GW3^W`oUn z#c>_fxWPZ(%ErlfoZQUgQlV~ZY+=3xK$Lrd4)15vs3)?Qk ziVU%9w!5P#OBxwitYX!wX9wy!3_a34Dj9hdw=Am|94yT2;-AGf4-Z4XzD(S{StyzB z4D`Lsd>EQ;U+M6+q>?57rJAB&=Pn=A&!xy70(=!`Yh`wdS!pQ^Zldw$DjsOar3bZ~ zd9Ck5?@~78yJ^r37lh8z)S@3|wvJoSjLm!6sHrK?IGV)?rJ7rn=+{g5rx)h?TF+4= z2-U?qBcy~y?~NQ`;Xg$Y=`Rvwl+LE}?6*bTTCCS=wr1ZzQTZx3C%z+|js`L_kFf4^ zKvDPOV=St03J+m#8#(;+d`)<9>y1yWUFVkQv)b+KcaaGG=I!s4lTpMWtZXn=5*~n_ zAdWe08zP8)@LV}N`Q#C!cLRz9N9_BPqR&^J>ct7YU1WCpT=HD&gOi5WdaL17L1i3y zD{=(h4D;Q)rvqi=ZCwW0%?No!i9OXja3;}zbO6?vPPpxO4%YNzrb*0$$d*s#AZ?#w3jD;*%1@uaDd7`Y*-D}iLY|;$ z$Af;=MCEWa9ShUylzJIuK4vOqfYS|)N}wB)ATz7Vt5u^$SKuV)4+R+N4T^9Tw5|=j z=9HJUD8JOESi>X!%6iYdu`5#+08SnaCGn@MaD-)FaJUVe|JHx%foS2UJ@}cs+NoYO zv}cX%K){0m^MoSvDWW}d>$^=)z+fe3;rE{bRs0RTFq)hcw9F%(wCXSM4uw_X`2jzJ zP!#iWf9e}1JH(iHId4>`_4Wn)s_oKK+ImKSAYXjive@?dxDT80(Zxzai6^ovyyb~#!Y62Jil=3*VDmQtKkRqjUtCYiX0Lh1RF=n z92y-^qcf4P!GpCmqw7i!$sP?-fNzJNrJu#26JVQ#P{t+tMuM3{t(&^9($lR=L9N&< z1fN_jr=D@#Jwq)%6H}0XIXa|Z{;Td2k`~Hs$aS3BBHMf%%YkWq`y<%vv`ROZA`Jf@ zAs?P7iRs&2u!qlAFI^FYM{4{mSSE>YVsEyPBH$ecD=g^ch+j z0|9qx)*NHQhmBOz)*lG=X*o1MSi78SEpMW`jol1ydf~#yBXX%CxUm}76DeH1Oe~|l zT8V6%uN1w`cc*3W1vAYjuXK8T2}1C-wvXdP;)6b_r619NDJD3Fq{Fic^2w+1Ld6Fj z#rbXJgPiXCK_9Rb_R$_+y^UETN|ML^VCi*vMUqb5bnq{ri1Lv0Vx)XChp8*9L z8N%Ri2=|f5xhLWWkmOc#^TC30b7(?x+-|FXsi%kx#Et2?L#*_yiseiQt|T9JGf~6~ zNbZG*bBvWXJ)FnuJVdOzYX+XkQziFqAKRr+M||-fA`?Z%H1!d>4z@h7FE0S)jaDF| zlgi`#vlHE{X1LuJUB*}{BZXZ#5wWJwt&(%lji$Jgp;CcX0-{wxg<}jS1S|bC^I|~5 zL?agideU?G53h8*LYHwY1XV1AV{(?Y$1GBu2eE&nD3X&DrsM5a%=XAzlgEHMNp^=@ zq5N+-XagO^$4XIronxL&g9U&jbiv{~JNUM{fAKI1u{xIvsgv)E(Oz4E*Vf}9FmwXz`58Vd+T(8?ooDuO+0W)S&R=gcDi@jeGPz8a zIJ>&X4;6^wUAuCs<@BoCQh<>lRXvE~3wq1l5o&e_zu<^7#D~~3{S~0_CzkEU&@AOs z>~Y`=YBU$~btKoj-KfWxTuq$EdwkU!re9uFOJ0tC5n<-sk+6RALr^YvXRgBuBzf<& z9DUrh3Nb+P!bDf?cpVSxcbTepqRdy>2G`y~i9`?Xe>(_1es{g7xacaGHKsA~nP67t z!|4@r%p}WF>QSs1jRfLd-P6!pY1viR#0r#{s8u2Md$q!%&3SS*6sExhOOm`9VyGf~ zW40mA1WYoizp;>W#%2*Idcp8NeJj(0WajiWxpTUgXX)bLN;i-uEP4YCAU++P3qURp zRR-KvUccs%2nBxzeeNG!0!3~IeF)YrRvw=2*5=Oty@4!JTF}%0EEG21g-12}Jgy6C z?GhCO-Z5o4i?Mpnjnl-hU|BMif2|dZ@&WKrq$55%K|Z~s$!xdQgZ`8wN&C;Rx_Id9 zYLr;u$f=S_42n3#SF}0XLK>2I(7C#;v!8AJPlpZy?VT_xttAq}Vx@={H) z0<@yVAbs-%_LFjeKoyF2j!E@gpu4D34-lyGI7j&#V zt}SyW7E0S-_{Sd+*GK*j9#?B@u(;}9Ljzv0{(a)V*9R(q`KMv07hIWJy>NT}JZtKP zjZUqz@(3oqpXYf3N~X;rPyUZ~T0!;vJ2}y}rdI-so39@&gv0C80i4^76oYpN$2W$*p3xny4t*& zsr0JhY1MINyWpUFZ2NF*oY+>m6&}?0n^sJAK7$Bg0>iro?BYp{gJw0|6~v2gYzu$Z z`u$^{#WQ}DO)|uR?Kl@Qa!l* z)R?)QTJDFdEL>APIy0@(fZf5*8jnubyH~=CJI{OELM@Ad|JzUw2;4kmY-At{2LvG@ z1Xd+$7h6v|GJfuq88n9fe765-Re8|5u6~pGCx>unl!$yU%7EA^?ZQ$Kk|0KFfPB(_nl7dtpWJok^92(AZr7g^T z4@B1TnP6vkq|Ph-byz0NxkjO8S0=kSs^SWW+5~A^a;J|yyAdKz#Y*>JL@hKR5ersx zQx4E86}ON9+lu~{;7gAt-g_z$y#SjmW!)Cs|kR(n;lhZ|cz5+GdWS z^#Hc8jAT9-zT?4X$N-5_Bv^$v@l5Rs4|&+@=xEmXFigBpLMM_{GALV!03Y47wvNaT zs-J_5}4k zNjN|`EkFP3N*bSjO*`rK^r%XM0dZ)Z4z)BN%$1d{Alvr{3HOb2wY#m4`In%ZF^9R> z9|`-u!0T3J@0bg+(nQBZNwH1s-<0kahh$1JpI{Qm6YfZ)S zu4sP?*Cb|4P#EiSQ8IIiqHqy_B(;XD*fz-*GjRq7KOE!H@!0jgo8R&pAP5d z=lEY~^v{3N=zATwhCEzDZXRY$Vg83ej<5 zw4oqrW1rn5l$9Fc$Peb;Vr`gJ1||3gd?A@Nz(FLQ`^CF(o45l24E$X`D}<~s|) z!13JGq!0%bESC0xl#?- z-VsZ4rX*RhmZpOXWB$a~STV56w-1u~+DM0A#>mqdITZQda>9!B)6o*-i@EK0z4Q(j z0PT;aztBoFez}H1=_IDr zXUDQuEGUf0!SQRX&|mw(uCuG6a%9ng+`1E`+(bC*2gc6q-DG86B1T`9PJi=BwMNq< zE4pz&o82b!+)UqR{z={~!<@0KDvxr^P^zJS_xS`)Bi;CTT#|*wnc$hD`f{5U8Z%NZRpE3Qt4R?X$|6N-^@dfSp1)eZ{oe(Qg#o*?#|F9%@i&F$KD~ zR?QlE^wm4N5e0nad~whg^aH8X zKMz}`^4kEmff&zDb1yZMdv)TR9$_nd5M?msM227lm`@y9TatU zn{5*l7VsD1ANrKrFup+?)QI0-Nf#6Zh#{_a;~IUOic+!~zpNYiF#fu8bCusG2Z;kkm6nfWPt$Uh``hbNmci%b=yL{l%je& z)+?b8cSC6G2Wo-Qj#(JWmlhk4V3J@~bqW@zm?%0h(V7>4KY_NC`|n~bYS#Ni5teKc z`Wz0W*Y-m9Oi)|ZD+OjtO%CjxTk$ON`jT;2P_(Y8ou$KS??pw2_AoQ3U4&IrsoQXL!?zqcnhF>szEtstK}Xx776v$*_`g* zx%Q-1g}h8gyItm8XF;DM|7303z4l4Ui@wB2sE>0WlTMI#_nvbM;96@g?iNzCE7hiz zBs_YmVENakyLosz(zXLE9?qTQbr8QNE*19#sIi3+v*kYL*WUC+?|iYDII{$U4G-cJ z=={&FkC!7w9SaMBgXe!yl`TER|2BHvSHE}FWh_{B5}~@VPGCmK#V{d-sptzm^MxC&I<1$8N&Lvg&wIz3rf6fcjN*%&`lyd zr#|>F-^6X{nsYn*RCC%}-4p8t1E_2ClTM{k`;$N3;MnS^;uiCc`-CC;T6L{9_jrYO z^Kr`@44Qkl=LtB7%Rq;+cj7*kqM{*EreCINHS{m!V}X7SJxb96$BA3-ka#aXbG^5W z&BNicRx%&w#@p3HIREN$GymsMX0{KwiU6_ z8iQlQ%Rj~+rJ!hvA09u0G$4h;re`u6tq_n)+d!+}&IHs*WDPDLNJ!yuoef<;kXnlQ zydk}+os1w}?Y3*yi7oH#^rc5XCFJb1DSm03$b~LskdJZvQrWv^8_Pz6okq190;jBo zP5s~D0DoKxZMWrbMYMcW^7fc{4Twm^^0CPAZgjN56{g9%noO#@3qNGGaFg`Fbi)Tt z=`;L})qx+az+h3#OFuhp<<-rErx8Rddv)K0%GpdR3meU^h}_%*|93*VMTv(730lok zWdasCc3rep@~NR9jybIHG@RqCgrInipj(w#68E}IuM8(>Z$a^74!Fk>#Ygos{qX`) zk#!2?1ue)N~+$ zqBkjswq0D!H{R{}_nhFWRL6Ix-nVcL^Li}-1W&jc^!Jdp#@|?M0VQ9O@w=9IeX<85 zWwNGZ5fZ``SCmp#*AS%l1AztmaO9d1c(16x;B|_I`>)F_25XfoX!O`2m&DA>FJ_Di z#{gCB!l_XN=U}7AQkZki==9Cm&Dcw9+his8POhJsO#7`xpIuw;7}_U%wWI1r=&BQG zie8V_2xIj7O^k(_C$0R5ht{af-bvMZSPV%Y(|ptI7Jf=vmja9Jc%E(DAZC!Rvq$c1 zw2ZeYPu1*o6s3yW&hm>8QCzy$A9!NWHgHtP9d)$vv=A*4@RL`X15I22uK4nOj#&w- zE>yDggcs|@J883#saXPW#KhbXa7z7$sQ)$eOd3Z6!feD~4A%PnT>h1-+&g~0jw#p22WS#wcq4eDuMk>3%Vdfv zDH&|(h;b)Pn~6A56qLLmX@P;5{tpSxmYMsVIrqS?s-mKa*S-nYLPH zsVV;)l0G>6423cuyoK}Fi)hncf)tiYswXS6rgx`<$Hy#Q&zD*f+t}P64+)DnA$aQk zG)&4RaFwma%Tcx7d-qiIQBj(Oo{ib4yS%T|*6RxCr>|P2}g$k`bt;eXc~5@`KT13GRK)0v}{e7;GF#j8K*xJJu2dQ8pp(2-Ba4o zTL%oIDvI1aG-iA+Gv#YM4h`Qrlx)a{B7>fpj3VYlI-Q;w!mTghnqAg~w=zj797dNL)JxTK z2(9qJPVA5+qt*gpmmxsk%$w|LE68dUTWLfyB=b+9rwd|BU)V^%EQnK12Q`AVZY4fd zxnuIyf{63)XTyQgOM@@XJ%K+N8?CXMD<-Rpar2~fd$pI9iJDV^(x(U84?UtS&agPJ z$eVAAJdOPEq3(OX9yL+B12eMy&69tOoi}ry7jY`%=`|m3tJu2!`vYl*$AlwO`ZK4a zQfdMJ*z|GXU#=2hnmYNh{i3CX2a$I0V@F4SM+N zY>Y5_vh4vT_wpwbpyY5{8TZ)o)U|l^j`5GZD@{o)@*0fuJ=c)=kJ^Sc-~3IzW%B|= zIrWQ3Kv9`A#!yV&2C~gj$ycE`0-Fx@5?%C(1(W}mnr#kq6qV+u7;$nqp86M z2Y54H=BlxPxn8p7@<-Nl>i!?By*rM!L?-zpRc6cvOMxp*7S7{2%#yvpD2{}<@CMsR zHwf6DrK=G8LI~zqWdq=N)c!_d;2H|Ad>&5lToz$c?ZY(YDsCZ4@ty4a6k+}c4++$9 zaK%6bsto+Fy^0kmpu#rl+8!>wj9f2QDLMh-nQ1-%&dd!XB;N;x6q3D8kXeH)A?h!A z%v;G>DArGD`be~)-z`BB1Qa3xB;C?4!g@zb{J z_(qRPjJFq5Ig_cgFYlz;l>t#g9nItlUm-I0gF4@P{8P}c|Ly@&J!NKB6D5{QAqQ}0 z`ghYLuav(&Nig>D_WDBjK<4qLjQ-|~zM8YKk4L(zC%bV#CrSP#A9X7_7Ce0PvOz@6 z!%Y4UPb>fJ>rdk6p^NEjk~0O>2e3T=d-d~k;qzf>$W&W5A?kmYUOuk>JGMjHk#Y0> z?@c+`zgJs;PAR z3=+JW$}wiG{4Vie9u}TtMTHN@L8W;ki5m|63|*}VDqK!c1)N{6(8DCmvaYCIUo%Bp zZoS#FrZchdXNT(zs}SMtogv>}eBwj!Aw8fmK=K)2ds|UE&rlWS!S@!oZ2gHHc1XxN~>y5Tl@P=mHKx zzP1Uowd4Z&Jc6q}K30=B)$j}pNNu3;fE6kq<8nyilgu|5=4CSLzGD-1Ulg}PM@dV6 zE~sS#b4NE`Ih)*J!j-~8E=$s6UUWw_zqxO71dve?vf&FNE_9SimWf+oDf@Z$X1^&f z@8kp^JpahZ+bTsk=UB8(Ab=03n*o#)*(klC$Ylt2)iS73-OF3q=oeUre!mNQqY^8E zzuPs3WIF^1KTns zt>wiL=Vsa6t265gsNHN}=DdaQv18%R9GO5h_-7!b#|^I0$7bvEttaGNhUwf> z+!!dxtP}tLAvga2TW)L=T1 ze1gy2xgZ90R%+WAjT~#KwD9yufaDZflQh(ax15%1`X^1nzu|qX z_QyV#-d*Oyus&JH;&r2ZIAG;f<9mL+}v3 zh}d-mr)vZJS*$xO_U<|Qpx1@jL>1j-t`hf z0>%!-y2ITt1~bj!eMArOL?BU~bSH^u8YZA7;rGnEW4`;bJs13_7>(NH+ykeYzK1Gc zr@<=S0$>J~L+Sb8zkr=GPd4a{g{Qf*#S{mLFxgseszGw>){c|k9t8k%yd@`-^7C*mj;jdeW) z9`=VIIp`=fu8!(D6@ zH36lqnJb?ba>S8taopKd-J^DFot)*=dhud@;=}-tki}z6u@OZ9HHgxDnBoO9jp~)= zh4NL^q`XII(#Vg$6|lvNg>BzoM7Ya-cF0{w=+7$p1mLzo>I1428Dv<>Pv5sBu@41~ zeuUX&jGb7cta+eBUb!Af!*xS)Q%r13@F2irK z#L)<#$J`A4NxEV#A_g^p97?J+U{iPVxe@;Pdm`rxY|89m_kVhwpQ{mv85IhWosBK! z1_HBDfPLk^5AOd`yA%grRv?EBD|qI2T|+bmeJjsOi3X|nd6Rv16H_*WwfX@S${&Rg zaP17BCp?r(Yz4==t5|FdYF1^5=Z4ZLI zJI$CNY`MkbuHOfX38c{f7Lar%Vhg!2`aA-)y5)u7%0zhlR8SVWeAjif-YaSk>$yoLF&|3t=;k zkW=wi{n*SpHLW3mB4Dr}bCwHrn#H)ulHO%)e-I5>`Kw#+uN^Ef?`=|Nx)(XBYS_^? zwpI6Sz}fKifFw)^^_u65RJ{Pqq%oCRC@r{R0w=ED_=ot(DcLs$N>y)@{9Uk@M=smH z_N`IBUdRnuIjVxub9xhT?8X(IDJlddC)lucH=C}1UOf({!2rvxw$P!7-0%y3$>13% z`TL}4W~B41%fZN@TjlhT5PXGPl*-ObU!Hb%wD0+Hz_(feVplOk_3F)oQ{Lcbe#f}0 zI!YOvN2(YL@k~u;GsAp6Q-n+m@fdmZT$Jv6-hB}G&jQ~D5n^aIpavs~2c9>DOu~pf!pq+PpG4i`we!36V^!J;XF(FI0Miv3} zJ^T#XzN}uaePxB72P&XX#o~SiOCZJE1I2^9zB3ukbP9jHDdpX9MRZKLRD;0?2}2I7 zNPpxKeYu+5-qJQDE^M31J?q;3|31xl>xG1eVCVc_=42Cu1P7;Xbn@Ev$bSJX>U2^&xU2 z^IPQe%u@|ab%QKSiY!~m6u&&_HNqee=IP1c6nHV_Ar$04PL6%9J}y6IUcMTS|6K~s zaNbQ%zD>H;oe|zJ^7melpSj{ClhTn$OjR_u0#32 zg42*@q1z-sQ)$F{!P&f{{6C=a`iPUd$)KQI9bLo2$k~1y;({_Ep}cmdYu8qrZ$C@;vh9gN^I6o_(0d@!*V?9YlB#P|?@vM#930+${Ktat3A<|zj@ zSnSpgL8z#Eis!U^1}t#!oyRTH>zV!b{@rf2#T{B%y2-v9U3`>nk}w*!@-kj2%?H%u zWvOdW4ki>l8fG#^IvOslm=HLcXkFv(Z^ZZyVSubhe3$jk8CjeTwH3DarWr{NZ4ur5 zO)}&b;zbNhE`pZ|?jgTdR252$tQQ<6J+93shfYd|S8b^gg zdwqCW5j{7IjR!&!iKR z03`6x?q|ea$xia6J_qzMkKkbCKGLq7i$AB=juYWpKb7Ekck}B#_gdy7!CF6SJQ>N^ z$*(VPufGhhdW=8+vk--S_*HTLH7&QbF?GQ^{eZg%xc(4aoccoD*J;QWXm5}&~+(N-yeee$V-5S5;g=)*o2TUqpAP0zpk+bVE`M@H?Lf+wB@?RdQ zV8p;dwDaJ9pA1LA#KG~1{LUd-nOWg>;bhU?{WS}I(232GfJ1ahxU3XVYT%#U198`c zm^iPnR;XUDyf9c8Z@ABINnd+#@3fDqs`4QPBNs=ZA0VX)(a|pf`Oh#kS6C3+Jowq+ z9?b0*uP5Qm#KzXgQ|zbeHE9q;0Hq5gCE)nd{0+zlK(ug!s1P)nSrOsZ6vigf&ACeq z?9aEq6GwxN^mr$Zj%26E47vaydDq6ZXA;>|BA3n@$unvCccmLK<8*^jp?CI~n`JgB zVP^+iXH$?BU-M^znSW~gmE~!|{J9BDpiHpk(Dz=(8L!6Eozys>qw+AG@)tR~-fXq1xPSQ4lY zWxy>hHX$5`h0^9;n9&2U#OhGRu73k7fR*|~K-C*Fw`#NfY>N^`DG%(YRDYb^7I@l<|3!>BwL< zFzq#yKtn)~^cEgnd;eUkH0psshBcbFI%s^q`1R#P%lZgl>a0w3*&WyDOO&Rv zN1a3M9QF!FS=W};c^J*RJD&Gcqn9QGIeu+=XDkaK+Klsl-lc}I7cC9B=bn95-K>nFYJrtzqWU4)L z{<3FUK1?goz~g9{Ro2`5YO7RQ^l4tHx0-k{t>s0=Eq_1{{mTL-4Jf%_c`-Y{aGp}o z-DMebvI9^Z-kk?+FOE26R0^w z;0^Qnbg$x843H2Qst0a4RHiqbF{KC!N}h{O$hv0Q|(bACoxObv}E_452k zp`CM^;+U^NlI*GdPbQ1}Y_|UG;j9^}y4>=)rU18#_wTBFF)r3;%kL(Z?1wXG(ha@K z-NA5DX`%WS#2-P)QjrcU?cUWv_ZCDZ)QN?H#1fmRjQbO(XgG<1vyjS-hbKCGg*nH`Iz z8h~=jgpG*u$n7-7uIBj$vE_*SM01{ZRAERYv(}yvd6<^9Mr-NTyWn%20?1y}TOWk& z!9Coz|B5};HiK5Z?z=Olw`M;)T+gqO*sxd~%NE*5;zA zb*43p@3+L6t~ukr+mzHhYsC&yh?{+D1<=;g7LJ#{lLp>8(rmdgt2|90z{a?nG#HKk z$<7kU*=e)&{a$HQgG@kF51~J zuvUnt2>Y+t0Nq|JiRi?Rxg=`+r_t*4y+KbnOPxD|*ir`ISln{bZsZON zIX#Ikum%BbAf7|BmrWqZ2c$q%9O!g2s}T@%)p>i+8-BF#dDCEiiK)N07oiccS}U{M!s4zdVcp#(;nNok^Rr8p?tBz=z4VK;Y}_B?>huZ6 zoMm-yHD$P~t)Lz0IAXuD*JI&mEm#{+l`~OSz+>#cdJq>W_38jPM|4 zTb>!?G=iPc_2REX7$4AE+P-wjm4sSiR2r3N34f~){2SgdDW7-75qYEl$eb{|bMx$I zem&6CGBh$QYob&>l;p^pelboia%H(a?vu+k>lyqxMY`+0jnMJ7D+@S^q&l^A7H3uy zs#J|-D!G(2aS+w!P%01p+xw;((!Pci--wJl~^wnp{_#E zy3U86ly!t(hoSaB=myC8)W8Ou-5%y6)pTFdyrl3CLqom{PwKPeo}jz&7_N-DheL;B_gCQ!pr4LK1^Bc zoT5jt`RvePDgs1zCA$P)45dF+^YCm~^~;KY(YOd#`VcS9w-e;o#nrdFtCQfqIl3lL z^|K^3IAjBk-N>~Scu7!CgC*xjsQA9S>!+En8}mpR!@#N-BZYN2P9n>Q_%#ZqJoWhF zg78~bplk%wJ^=*fI}7CK9NU8%Ff*gu2zA3J{c@oJ20%!7rysN^i=q{~*a#LbGEdn^ zc&_Sl6#l_|`Hg)^dph4K&ir)cY$m{ob#HGo)z~3^-sAk*6RgA;?(+mAUQcD1STgY~ zT0{!g3->apWDuR@N0)zI7R+gW00D^rQASZ44!6v^gC?0N%N4}cgmoyO2$cM=@;OGGq3m) z+>kb;rAQOZplEQvr+V7~h9d~}^JA#_>jC`F^Rk2!nb-Kleli`{YX)`Q=%x2HmuIYd&KEq^;yoFw| zdRzsmYIV9oUO_Dd# zaD{Umcft`c+t~1xJupr_?tL0n1e!OKHtbE=X^12xRT+2r!pS_Hi;LlR%2$ubKAtrR zNaOY*EYDo_{}^Owt4t}QV|ArZ9Y@M9|A@bT{0(257p1u$l4i(?GG|c?d;l_D2@-%uA$Bb zdBdO(>3FLr&Q#rq%++L*S0qq)3egoI7B-q2VM&*D>I8BPlHe>SgVsCZaw>!RT$&y3 zI~FpCL9!CuY^(UwWy7K?Uf$fvAYIY4MxG3^<1yu*Qc`$Eka|-{N20P6{)vo$O73h1 zv1du9pvb_9-}Y$_#9|9OjpYH$$Oh{-DIFJj-Z%5D&>t9ibP&jS4@Pw}Mddd|(={rM z(P81x2gWl&~d__eB-0v=lLt zSi-Dd4-=~~g$FPxNllc_b|{-U+ZxJ#^h1cnF;>jPyZqt49pDwr5R7M+h?T7T5q}hr zAX$>+JIc2;KowbVBDK8~cmaNQ3rp(!jvaa&oz`qp+3q82t17oi+G7g z|8XjPcB5&2!?;HW3n9f%1QFc{DmZN**Av63sY!lyZdj|3)rDv+2P`aAbVY7!X zJagJK?=@4}@(ePjb2LyZFksic%tj^=#N|a@3fJyi|pjZ#2B5OG(U3Su=?2)%p z_aqnVq|`}j*I)7$B;D_q-L)bKw0=emYmjB?M02X-0frOQE-{@9E)b1%2 zL?`0lk>U4xq46T{7NMK6>xuxXf-Kv`{+xAIX@x@$3yt3wy@VUrt8;=~*5%c-t6K%- z(@AX`iu-D|gTbi*8?#2fzht|IO;^cPP|lqP;kE=*1o5Nwq^CD3~sNZ)UHwHHnR4|17sj46G$AIVgv)@gq<77lqp?AOKPb?(T;01 z2RC;k%ScH-C%Ih&7)e&&YzQ%z4lPo8d)SBV;Ta=eG*9(Y0lD)U<3z?xsxsXOAysEg z_#f)BAvz3l^*>>MA*^E@U$Z7w;<)yk)^nWI;JeZwE!!vN(%E2kW#X7|-w+BTbVIUx zCUGVa(rZ3!mM{1>r+TyQ7&@FoIIeQqN@?r(!_QuGggQxYbZd*19A48$-o2OxX@9GkKFz#oI4Dc~A;&Bptc zA-68@glVJ@EJS+XSEh{ALB_aHd1FhX+xP{==rz8?nZwE{m}#^=LkJ&gLBYu)!VWH) zyD!SF;w10fCI9=|&ZxzY%#u1j9d^+^aZ^=);`Mv+-yUre*@+Lz>uK=TspG~!^=%cu z*Rh~@&H*LLT_>c@|2lttCUOp~$s8ghzl@cv!IC(5H02y?Vwa(m5A$p3-08FYHPTsH z94yBr1mUJf4GAC%Lq#Vp|7lUknD6`bYROTp_Lj5+%ga1XB%(2rE9bBYjcH>&-tfv1 zz8xQcF1iE~@*kNv8h)Xg!dUicAW-#Q%n_-?jR34gFL!zAv^gE|wY3AV*=UZ^W?0PP zixH@}v0{y4epjY+w|1Zn5{B-sRq`mJj?M7*p%_=Q*MnyMH0Urs-Si{>)0y>->|sz| zuA(RIiF&M}9U6MGM)Hs!h6dA}t{|dl2NYia{zW{PBVCPdI9ztO+glSGDr)~Fpbeo~ zlLe3j8V?kpGvA7}-w*qUYRJz7lctcQvnkDLDvQ?kL9E3a;#xtnSw_~As{duBZwM>* z)GNA3Ts(_FsS@;H6WPnA>uiFkUd_n}cG>#IzZ6gBMLH>fLw412S1kB9uM`eo`e#r6 z{jkB{$qCWERC2?hN}GIv_NCz}OM@k8okZ&LHgQi}^&+u9dc&t@ zowJhI&ShHG>@{B5^^;xU_c6aF^n659UT{abu#J@e4`@J_ztQlJJ*M*ore4+>^Q^3Z zoJqE!f2#=fC0E1YCyf&$>wMKu?Oazl60$YD?Ln6?t82sA7;jn!x|%^tc@>usfw)L3_-%3V?kI_ey=GGcAxJOb0#x8*GKa{9Tx@m5^f?cB#F zXci+5ewsVR93$c2jQ1X+eX`QpbPiZQF&?>_e=A%}Hmb@sVrJPbNQu-SFaxA>H4yIx z6c%WmVZEG8+Uz8&5V2rQV3Pllto=M;$!rQlqW)Z_Cfv7JXmJhP{(6&1DYG_xK74n0 zUYiUY_D&5Xvsz0eiD*Ld~R)$|?L9O84xOQOaH&0$bSgi)Jz1U8wtci~hpF zq%9Ns&#OOn&I?SQd!)8C-k_g+6nxy%f0ZrF1hwS{svN3x?sluICHV9QjWqOuLLMdV zzY!lWB_G*7!KdD=mk2ULR<2(^Pm*x|aWWHVq8h*WB<$IzU}&&c<&XV_oH$94^{QR) z8)?_O2YE5?{8L+akug@`)F53DY@7pMCSN5`YGUKQm9~n{SIy02*XaES0l&80e|Wex z?OhUGSb@rGMYJxlLieteG z1!QpY=nE}s42Ti~aJle1ZPSyO`hrRZHN%sR(!~l&W|NYzZqBXne(Rcm5f2IloMh3fw2aPN-6lmoe>9y+Mr?O7b+YctbRGZQw z&QiucSv#N3t}G#!y!1s*;dalj?ueHxilm_PuUE7ZpS@-gYMma8S(FfM(Z1ICmbFX= znqQ+e*v7~dPt7= z2PyBm@w;ig$Iwnl76qg)4}WCF&P7BJ-`s<&u6r?iy;LjhTNeR+$l!T$t-qb)lg?A- zYKG~xR%({M-PfrNH zjlO{W2qDv-ngueko8gZ7f0X4YNiMdF7dZqxew(MgcZk&3w-?jD%=6_F$ZJhucY23Y z8I^cFsxCd#v5$EmdL`{mp=^4p2x8P*&k!yPcnV|fB((E*=Va^)iqx2wxK3Y?BevfP z*!CJ9Ep08$G*_%lY7D{@ibqLZD({Qm9n2bkF3TF9yw|w8+%K0df3qAxYZMI58_`o* zTX?useW9V(aX6eEWQz5?S&j_N)kl6bLB4X38YU`7%^q0pd=-4Ym?3&(h+`}F^3(bU zeDV`9M(778DwTO(Z0KFQ>AAW2hF&bNAV7k7B4o;W;DOLrZv1UgGYKj+!-Re%igixeT}YhfiN1OoW4DG`&oND*=M!{ z(XE&zCJ)}=*{)AI$n|Q>GhPn6R;a6e*;SSHdfeFYo8q+ZbDeF zg~`XBM3BuuvxObG&6t7}y+h}{=`uE(hX}DsL@0U{+OifW-l#Ah?UaaOQB!}wmFu1+ zRxTGw__RE;{9x`U?dvSdNAmeC({JVRb}ACbt0n6D64kX?-$xT#$bq7k00wtDg^2HB zYPBzl$}nZ?f9~Ww4YAZ<$bL8DvOj=O`U<<2cmyD{+ijv4chy_N(u!;n@ms17X%T%> z_W8qh@szHP320118@tpzy?Bn(uO|rBzwo5#&e)FmVLihv`jl@pit&90hPtarV0~P8 z7o(mcjzb!C$L9+}XVWhjl?;|ur?hMM+?MNw(ipL{e*@zJ1{2t{o`4>!w)^au#6)#a z5tX#r1^ffu3f3(jmEIfSOk<%Q+Y$!E0GICNGmPnOH&-)61}=RLLUr<^d}Y+JNm0l> zlU~(qL;3?@3j-QlV{yX@h_9l;Bd0LG)zugg@5vic1stSxO`?;S%iMJJ0%fL177rkd zBoaTaf3=jS6T>kOfI9LY_SJf{Z5NM8kMr{_ExA#gM<>oi&^LMCq%!ivXdT>E1ARP8 z6^4=QxXpD}CX=r#9|&&*4L8g0Oa|+Fjx@FE>H2%I*K|swbHCLt{qcguj^?;*`~3%Q z3?;aaF|3tl^1WRsctl0z$ZWBURSv^4%~E$iiAF{;%9PjcoC{pd_(T*I^2NwEHLbd60L zf7@gux8mu?z3GSOk9P$MSEq6E%iUO>Nw+o5J>zR~&}=RXJZe)Di5B2!;W6R|zy}=I zDDJwC5o><)nDv>?R6t5y-o`e;>jse3Euf845>=uNYRsGh84%Pkua89rJU2 z%|10p-*5dzYHU=>2b+!K@w_6i-bN#L$X&LmbFr-PBq7&Go;EOxrm@Chv#(gGJmNFc9(2_t$;|MsV?SDJNL^u|THH=sb5P}{Fk*Z> z0o%0+h+O3}K9D9hymv)+y2IR@D^*}2t1hGaHlTKd?*wSLOsn~S0NK$Pc9-$M0}}!> zIg|gI6|*T}fgS=hF|(|HQX2s?v;Ue79RW474!A@d0XMT#)YKvYIg|gI6|+6}8Xf{Q zF}GP20dpY%Gq=q)0frj^HMdk!0puG2H@7)%0s07+mIneU4lxQZOl59obZ8(mH8MDp z(GVzq?OR!I8%Gj;=db9;T?FRzeE>lK$=BLhIgaHxP7FWLGOZ9x9ZOOR=il#FRXyZT z6z!2oC4Mj<*3@)YAN5srRrk!uXr-00%2{Os{&0!P6tGEImz8nKW}|FW-lHU|5OFCg z8m(+lF}Q&XDx0EQQpMqtRmtejE2AOc(Yrx^6OWE<#?_%qrxhfPOHKvUhf)*`oJ<)6 z)Td~bM}5{o1L;q-EJ^GKBZ$ST{;F{bZMr9KQi)fco z7*Srd#l&&-K_#P{u})zs&SG-O#Nn4*K8DJmDBUCPQuR|W|hM(CFU6o!77nkyK8 zG?S{hK~ryl7DT+kU^zo&z*)5N0SFg^E0iE8qBeu#l+RR-_60~5v@aN}SU_WqVrWoq zv{Au@NT`BAae*y7!KyusLiW>uB?Uh)I)OpG07f#fa3~nl7*D1HxEM1ULfaTH1y#Tf z3IiGk<2kyJ7(trSz?cl|H1z;5JekpdYM=uA45pcFAp=|Bp~;^Bgtx?al?&OJpel)& zNDSs?;V+ow0B!+YU>bwbp=LynFd*g*ER8dy7p!bV!$^KZv(||?PaNq*5SieqMWh+Q zAB^KsOdc5WLZ=A``Z=Dg_tc7J%dHZr z%+V1E4w@I^`Nh@jWIU%e$0bMO_mk1##Ygpq7O9JNn4zXkb_f*T4$$9hbUs&g)(Dy4@2jFd0kLP>ilXteY*$ITt)4MsDifBi`ntz=wz&%GCP?&5XKo!>v^F7(E{r|2DFzE^C$*}%5!A6u_`W7|{otMF&=$lJbci(DPr za^(5gmLu1v%4475^Dg!Z8}{;dV6V{MUg4{>*ESoi?D(blZ^WW-2eByhk3}JTmH2ls z@7iK#{NslIhu!@@UdKOw_V@qz)%gEfjJgs3FU5Rg&lAHPVBUYv6TkjF&pz=__B#Ga z?;rnUdgI?y^lyy$+ibBqBYdeXHr7JAgSC+QuZ8rtu7#U(Yd;%x_eJ!mK;LwZ>+GXq zR@v-gyS#C<&UXK{%dWTWdWznW?N)ehw`12{T|4%%-O2~P)LvzO!(QnQ?3Mf5EB_Vt zT3cVA-N)`?mp!bhvazPh?rW-~Yc?wV*Hm#|-`V_HOzYUOkL`9|M@H)nBjr0VQt5A` z^0gT0wg;G8E zBMt;*n5ty9n)(FAC?3&Sr9%BjlZN#E+)x7%%3gDKo=Sv&$NnL@OhcteU1B1!4vt(G zF=9ItrkXyB$V1oa!#tIW1{J5E(N!oK;|HVBVdX99;ldJ0F15)(+G$5^LWYEt6kW&; z(y}8j)D?sIcu!h;uIYdIXLjwS`44nrhOC)DmcSqJ@?PKteiIyQ0 z$&^^PKp2>ROty)dwQ{ONL+x22Ls#m{bofkmMg~rv+l6*q?7Y}4brS3pNXjsbqv5L} z%uBf_Kxzd_;wtb9okCLr7c@DYL!YsQf<7#vk55E~6`_0jP=Z@5rRe8i8mGJ^FkDG( zjlkkb3E`Q-NZH1r$JLj_v$bSE-{l;3cHLSF&-7e>N{cer`)o70gtY2Ni;T;xbKzMr z&$SpgK~uoVl<>$c&5p3ypeg9JANHLDNWG+Dc-0W?wxTsvZlIZ6VUrr0c(uwUW6heD zrog`#X?=d0=FDUCpjE#1y&Qu_6q+6BfjubQc-Dz%67%AyOdfi5i z7ix$@6t(ol5KC_HHrk~N&O$|kRV5@_+L@$`mP+QpDIu1V9fH5F zYce5ZQX-**oMAaikwl~P6A|GuB|>90MO8|Fa_FGY(vF1}m*I)}M2#d0drE%M2A%>= zQmSMbikEHVe1;J6xuw>HtUA>OZP9lOkx+Ie%2iQv0!9DY2Jlm%22#WA`D9tM*flqQ zU}8IqA#}W}&(D5lAq6Bd*>A0|3LXDToYYAB+mGKJn|3I^_Sy*pH(8hiwm5E(1jF`_ zTFTGuTxt)6ZSGACtRf+aruHSg56dtWvu4Wv$^A+!+w?TML}bZvm?Fz2TX=7ld3T^` zD^dq%N}EMwK^&`tGi#j0JfXBvqv=zB#;)PM!#iFQ(n`v0pm8jH*UFndxzC<+z2yCl zv`}&|xVTQ7m3hM_WAIKxlo6gO6#UCUU2_)CJp~RTbxC*lx~(H8<5J7>vE09x#V#`} ztClNip^i0kysTX`YzJ=7eB^TBnF0}GYie?q6nDIag%3IP{?*n@i=k~|mxjQ9n;V>c z;9Z#_Ps&^pq6B5z+ZM-wBbkXS`>VheI{<>}k!Ks&W;==hQ~ zZ=iWtBzWFs*EujXWLP?Rc+rG8DV5n*G}zgq_Po)f;aZ-zlYvtUeQwRAWN_3Y}0qW8$rUmr_`U7gT%2NPe}*oY|3fmn>~o>>3T44tumn>>7;c|*Q*uC*FoxXCHdx}G$N_3@vg-W8MIAnS4 z#dZutolsQ~o+&LhXZVD*qA^;IDRTax{Y&Aqvqp3v5G|jwZRALr$|@c*%^U-TN^55V zB_rlDy+U4cDOh&U$R`^La7(}N$*`sTau;jffsn>XzM(0qBsf{ z&H*v5BunD04`T%ye?L0dMAn zuavxkwuPNOQVRTlfT*u$;J}Bm$R|(V;S!&}ZKN#i5(P1#Rl;+B=#W|j%2;`+-Gxd& zLS>XLMQDUA0|y;ny@c$+*2?m%fGkA$J+zX@^5_rND143}O9eh3X0A(`{Y`Vw7vAvR z#Pa6MC*JC^Z^(15Qr5-?x>>L)9W4qc6HYZ9tbE|@IZ`Y<$~xY3uXtO`(i&23Eqh>K z4~k@0DM=x&+IHlBg*qEJ`!n^FENZPhH{ge9!dGfI&8%J}(W{gP?x`(`V&ldXc6i<(s61Eo2!Tf|?9(gpm zKf3(acyjv3Lg9u6p^AG^#xp`?F&RzYPd^bx~KHg?Y zgd0|<6y%~%w(CQ;+x5{mYUBNyHj~-O)!B#X_+vAF{_oXjy7KRiZ#S3ItNFUh?lvkv z|NY_N@d2y6#e^}mz8mQ&zu8&koVKdWdai2|ZnYGjK7Ib=h_%_e7E5aNNof}o-ZOTo zk>0A&kIlcE=gqIp>*hmq+Dw}NH0RA_GjA5n$7Vg~_*(^?CqLhR`Rsvcy2Wtxg-l(1 zF6?4|xvr*VtEMlZ9qD#Wt6K$>qgSsE4jz+MzimxNMcw0UWDM_sR!Skfd|5>CdTT_v z-#lpEjb@x{#=P{Hlck4(Gj!g@R#~Le@q8}a+2s6c-ke=c7n95BIx68-Vd?1ki)V*V z7?pT4D2TF!|F4sdS2^whldwI#P+XdhPcAND|C}w>wYpX0dG*u7M~{DEt+qI@gZ~!# z=V+7Zv2WRjspD3)K4q?31)-lGy!`e4|Q|2vG6g2NfQlNNovmKMuO zv#ndIuKz~YRqDmM9?KB(fAQcpv6u0{0}}!@IhU|)0u;ANcLCH{0yZ+2ux$br5HSib zOl59obZ9alF)}eVIg`-g^3-Wl2Y6mWa4V=U@u|f3gQE5fy{wwAV(l88<35Kg%zF>DDL3stCrYOl<#3^iNz1^zXTR0hwF7 z+5Nj7^jD9+6E%0RxApp;BmcJX&w;LtCznHi>TYCcaSeX8`9xVU5{$m;ZUzC`bgQpK8D-REl zk)4eN$jZUN3FPHq^ZQ?P&D@-wLH4fy2>!R1|M>qb5(wl8GJ{|F?O?_q`Zc{ZywYE? zc%}-9erC#8gZ?)w*wSRrr5>8~pq07N{yF}QI(~r%BGGh^so?h6mh4S`3zNu8WixgeS} zYm9M4XFn9`X2SCPoxL=z0X2G-?gt`hm%bKt#wx7n!c_QaX9YZ>AY>cyk`m9zs;Dx^ zaX&)mmXoHPcWa@8@k)k&A~}o^9g5EjvezC{J|!|8g$$84FEG-~!h!+a9Ua+On*K>7 zxZ`4&_7UJeIaHxM5{Y_ZnOl8?g9%Ust?;a7{kX|V$x^)pKd+F)G9x!op|MY7{~RuI zwA_MVJ2#OVSa@lV8&S0Fcxfg~77A?-Oz(^k>t?CI#GV~1RyTrw-PMq%;iAuM%X<@} z!6WOCf~@Mif2?fZT`A#==eglwdRCVY%Y}=*5rmbV@f%by z(Jtt9;w*+Ppw^CmA54v-d)AnAHMoNUJ1v8laRTogQq?MGekMfYSI@@Oy0 z%r}OSlc5XK;%=!DA%Liio%i?2N*Tv8Y*}S4f>$-?B=HgbkM99^d;>_2mVS@~*@(up zi!#+B-(r^G4(6suMgvL-RGBk7e9Aq~{LYlUWts@PqIcMTmK8?jYq}{^fSF_w3%kFE zQNQA69gav(=e&m%ne$Ikdt`s>W2~bO66`-@i6g8h15GYH3Yho|qit{j>=Q4DRAj}X zWot^;e=8-*BmO|5M5}{Cz%|c zk1u_^KpMR>W}tvKMglkg&7&Gz>hDweM{h&6Q@@=1M1+t^=SyJB z^u6B4^l-M6n(LK(8%svy!G@y3PMzN^;7nG3 zPGH-WHF;Xf!V@dn^7W1PQ+=NRoW^37Q5*N}uB7>0!cS>3ysXOVn*t?yREJ4!E5<#F zb^^C8lxW=6UNx|UTg`JKWNl?=3#}Yf$f^(@GKH+b( z+w`Ho(~_mb${ufk9m-z?&?-ChWD0y2PmT>G(iaHkjYpnLVB#hg;)i>G|4~1HkwIJW zifzzl;q_yBTH^)lhh)YPM(1;@qe^wq6ZEfT>{N6;pUg02&G%H?hzz3_0&nVE@$2=W6||^-d6=CQ&s)R=vu-!< zn?&Vg+w}lU2{2)|vVgZ_QwJOUFwFjTtq|_>N`#*AP?qhO_}MS2-+|oFNFFmBj#4gL zBS;#pPxUx!z0}Y)FM(h0xJI_OA7OmBLt>9%yae@*aT@g5!ru$-z#!y`2Q$4KpF@U8 z*C=d9CSd5%Iz8imcW*w~ppN* z065Wj&Lj(SsC7xU;m^Y0hB@$ktb**U@8khBftf0<_uh5oli?4c#UrL^K2m|txJ8sH zs$}nOYs;UOT9nFM-(R@WJv9bzDWU{TJQIhhsjBzDb zh#5GJUd=zWjin+rEHaC?C(eXg$Y6MwYXj|Y$YKcHnf#dlgHZ)G1tlowC#I5NK%j%9(^(}qSvm^`Q zfdWxf459_=0a1HUzcyRP0zSY-3SdFng@I&II7-9C0R&+F6v6zFxyFnJh zdxX6K9ih1>yVrSYLWwpiydTh&()`k|6W5y_nxuMvLm8!WwPrTv<9j$!IF})Xc5a&f z9u=*ScZLvPTB$FoW}kUP7}3~y<#5U%mrRFG*f;1bM=K#tYoVMJREJc8arbh$hiE0XZ0rUjW2B9M4x3xdS!TH=;vv5#WX?=%p2JR77+>k zpp_~GAQ*UbsF|mKNw&Y3?r4`?FuRMsvR`tCy0PG$$iWh3NE(wg8HR zdD~Do^3O_2V5OiX4q2fBG)0l}8|wf;n@CMeLS5upg{WdAHz_fU>y2JT0lua1=toJE z2o0D0qQR`FjRIGTLg$+l+kFxB17IKLkp3I;`zU5U&T+OQM4j z$>ool)?w4Y8JDPtCqP=LG?{aPdk}%La@e1bDffyN+>~TZ?uW&F^9I=z8cj#D1X8qy zKZld=uC}rvNy+IV#kavy{KwGTx`D&F#rRFutyoLY%Ldidkt!i;eTeUZi|}=Sdz#om zYdlu=cZON_>3?u3JT7?1!jG96zuy*-`9mcEO{GbLK(d=#owTvmXheGizkf)kg2og=!%hK?-@@pfw}aH)6C5&HNRe8mDGP)=47kQXGu z(s)D?Rh5O!Oo>7c>!b7)4GeZA>2;PpQd%J&HjxU@d|=uuk27uQbV8#X{5+ zF__g2+ge}xUeE~IGOIbkB4L5R(-&B*Q!p*+eR)T9=36o8)ze*7^5U)Dd@>arM;$D% zW5Ep@D}t(vDDWb22r{meSLdM%-=QY~V|{9Ian z`qbk}jk}Do^nmh9n{E){!T1UA0HM1!u9cMZFcXcN4-WZNSq3p5WXuRkhb7p!94G-fOP@ zK-PvkfE@;NE~|+g4XR79aK)K4dhaPx&)AOAecyF)<4+GCJzHM-_? zRa(J0kVs#DSLn%vb6vX7Tkn+;StBn=HyTH(!~Xzvj{7=ue)@9%f;ggs$BI8lIYY%b z^Xq8SB#n;mu2+_FTIDkpHeMoP%l{k$L1KavH7f#&d3xlico$S~?Kem3ymr!P zRGID?FDH*gbRJGcxJ0<4%H8RGCdI+K`@_3s9f@Rrb25i(8ci}rN7bI9-Z8(e8X69p zXKu>qVt&`v)UH^7YABNy!-T=!n=Jy@hNrAw@xBiW2_sWUMVTCGA}Mni8S0Rb)Q!+N zldez7E^)F|<|9SG1Epv-X+su9VnBD6_{YHLZP`-bvVTp)LeK$fYxq5Lc?J3#dR;Xr zDS07(4lu~#!7SlSo;m3Ev^XSl-)FRIsk9qdAb)OP<>U^{3qLk~T#WF!N$RLdKtom83yKHm~oS|u6O-*RbR-m0ClCgQqsNP(Ns&dkn)njoh;VF)*uqF8gA)?9k z3zDwq;Ed-~a~EZYF_RQ4vgih-RjC zYz-e)Nmqa`<$)V&=I7|`i}6DZP5}8>TMeCQc1(#)ISo-oYow(e}Ja6l{CJc>Ys_cwHZvMnD$2YTUq?XxXm9+}b-V_oN z4@Z9-Dy}8*l6hf-#g<>mkK8eTo=mR#Ijg!Fqf6V95!pGmb$;JZ%(0B4Rn@*{tYB1g zp>U_W_>N|9PX>y%=evd=TFs6roc!VF7N+lBhG02H#DRYT^ z9Sq5?f{}e4)kreh4N5LuzdFjmqqgRz;;ngj)uEpglcjbDV6f%l;j&tP6K*m4RziCv zzN?<>e2rTrL(1K{VUj4wIy+(F;quvMhPLTQBq4)iGBY7R7q7}C;FogRJBU&Wu6M#b zFsA{vU$)!;Ha5iWv@sJXmMAwG!?qu*g|fRtX3QlGb{@=%%q*_#_9t4WT%ajrpCVW5 z`@td4aVZxi!-UKP?-j0p(`B_+fegFjWc)dKApiKq*q43!XW{RC~ zCw!OWuKb5!SAWy5E94t%kiV6v_zz8#2Q}&fR59HJvY0{!T*fop74eUP$3tTusR!CP z2!xs7vzI|aBUuUs(+Hmfle!ONN3X*fmx%&>J&raGyb*3;T=kiM!!PE3&sBs!?yI zGv9I!Z3A*{g|{+)OZ_oE=_#VE>SUCit;cL{fNbD5;cm1-l@#tsxZXC!(exMN&{%3{ zwG`s|y2ri+0&XL`90Wgdo3LmA%#n@aob;T8e(F5LFRJKE7lY@O5&X8{#%w=f-;y{+ zrp7}~HJ;ka=Df469#>+~jUPN+4ov69+ zaTC2KONbwJpwy2|v)tg$2=(06#l2^@49wTed&}R^fNcz|5w6VSDJDzpHo7PX*f$}d zx@qzl+(EE^O{#?~M56h%oLFFuIT}k*ZA13VXTQltsQulJ4&tOk<#RDkmJ?a)lcnZO z zJZUeFSm%2^YEXA>8p8OMCbntzl;UglO80d8_y>J|W6p%YB8|GwT9TwD)yfVNRWg`h znDA%H`VfBSor9z(a#fm--&IRMkqX<83%@+Ig(MK$ZfN7De5v93j;_>e2E57F$zue0 z=fRte8-&JgF1c+p-R6N$1FHFmDFC_rsM~ z=wgO{eM1R?;SbL7=YE&?zMg1fh9NP9&hfW<8u_rPZlDq+8Nm31d}o7dgwej!nch9Q z2SS{zPc;lXJXsU!8KZbKI;K>mPsD?NiX(EIALkN}C^BKK=-P588kF0&N%d)!Umr+g1o^>@W147#1 z;-tTLcYudPU1qwaUli&cPG28%SteATrmV6&%kJ}7266iqr~{p~6)2ibr-)nMsS4pQ zoP|7Zr@is`uJs==ts}c00-pFZO(ua6SYg99h~B{jyRDaHlThORYW^9C4-1oh&*k-h z5%+3aB*?#d{9~^AXs@rkYgX%93{qbLp0k@+<$w2Lqf|u$Jou~`zYK&pR4 zbDqxV>8fil|4ODf>}Kh~MRG%GJgDbbnhnOIg~(s-3eI|25TDca zXCKX_V5Iw=WwWii8;u1|L@%fLXA1v+S*;EH$*p;vLf{``-^-txW_>b*v_gIZ-dG&* zYA#l3U5Vvz&t`OerWi;jgfYj^*h589<>goOsLn8X7lG*|s6~nT+!qg+jhRFqnm(AY z3#-IDG5e}uwmH(XiG?VrTL;P>f)|Jlm^h*GDac!+)WeB^mmWw6oj>6_E(#BSPJiL& zE|&({^LME-?%fYcZOe68Sf64f?k&kRxQM$mHwb2HtTo5Gl zgYL*oI*K5YrC^W@f9>f{J=|6)M<`3}G1$3u2dTvx6j+U3_uP zXLJP^8s=Zv93Qqr+cQ+fUh6GInqCn@j7EPvDaVr!n*<<GCClA@ zeY5;tS{Ur6ij*k-w{>v*U=4mU6SOr>; zLOzl}3+hyG`QO8K-Ldazeh2Q~D-UUs&Is4-gy6Q*E!M|dzUZHS_3?X);RoG+LX@@- z_7^#h9}eKK|JXB|(fdW#ggFrcwFN*Q<%vHpZmt(z5qi@lP!DG9g|Fd9m zL}u$h+VFjlzDK)IlU>YHSZCI1)tGf(@+oRmT>neq;tXhe{?&&Wf=<@oht|IcMNlMo zwexqk8h1Gg70i!T>~23rfJQsVvQQYMdpleo0$b_q(ZX}tasWu)oPflpo10tY8in9P zQNH?bv0lzl$7X1T^W1b7ch}6Ks!&TUc@PoLCoEVh2u{9~9K8Km%$zm8Pm39&v;6;PiWX4w)(65}c-}if|UmsPZF3 zIRV=aZ6LR|G?`0?e`2bKdeKDHEE-mx;2c47?NMld6;!&ExvY!LLPrDX__#26us7k3I_sewz{a5G(JsqthCPs(2Rk51N&c8#y zON3|t$wx*gb~3p4dTT1GQY7#RPl-Jm*U9@9d(v$=wmmD#1{4wcG96+v??Xta@07dY zuT!pn{lk@HA9LKA_nQBpQi)m!zNp$%Iz)Dhg#DY6);!QTNl_HTjNw{OWbT_HHw@}b zNs_>i9R>jiaK8(Ox)YI>l0@eOtyg0G_3M}%i%R$Ee8IVBQD97$Ag<|iSh(7-&GiY? z2X5*-oD`uro^$%O7nAfY`g-loN{q2V`4dTh3Ji@WYFEb=;{e)(;`h@Wyre(#0Uhw;bY*5luq4YK+xP#D;<`gN}Ck?*r zz(A^6=}4$I`-~YVMr`g~E#i%w2zD{)4?4V&28O+)e|(Lo(>5sAToU?{70s)AhcOzW zMv5z8Hlr4+6JQ&~8k4GJU~%<-0Eog8=^anQmClF6yVTlgsFlbHYT#u?`BQGohP2KB zrL&U8PzxE=l|Ww3TwdgKaeXN0*^fU4DaQC5=J_3~(gL64O89cadA{p}TlAnc?`s3r zlhSHpPxQqMsih2hUR=@MK*y*)13xPST94qA+?vdQ`CfKHr3-0bt`SfgR95~JX}pQFQyd=UuJv0IxW6bhb{#B#!m{NP>{^by<>6?@ z$xYuVlozos7PhJMM?u#9a_oa@M@Yy^xCr(N}DnSx!L?t%xNUmpA zdGH=K9`kdE;yR6VZW+zJGh_KPB(YgriLrd8vX|6zPSwiFS9cmDZNo_~Oh#}Y7&u|Dty~I!cr$4my)bpcNtt!Gx-7$Z4I61n{D>6N zLCb>DE`c#E<-!GrdI-_2P>7EABzXTK3E`FQyPiW9{#=xnOr53roXpn01Fx4X2dhhsEK6U^6 zRXLRQ*!-J+)MCD$2q|eP)s)$;ZaeSGNQ8(XU(To zREvl=KjJ#xrNZ}vIg;R2j))Q8eXs8Q+O~Z|J;-=FKAQKaB`>o@3It^YC@Fd9Kz0R* zcVpO|*C97yhik2klJkLLF@J7DqEGn^1*M-9ueq)w30@7rLpE?x^fmG-q@F^yyr z{ql8x3b4JO;()KdO?95-;Aw80V6V+DHJPPuPzz1B>63-t(K_h95%fpD1P;GLMCt3L zSP*G|u$w~|d;9jn$09lpb;2YB{{QKKMI zgp5_yB9ROQ#gid0IPn*^VjXr;!KKXOT`dx2wD-GLF9H#2ERsjzkeJ~Lt$4(*%YL`f zZqCDhwX{*`Wdi` zXE(vyDJ3!1$3i+h%ltsdD7BQ$@jcEI4T$!&d4v#;ElR0RF29x##9IHl!?%84*h%2= zB4&GIs(=vn5#}AeI5m+QUpX zjI>t6br8+|qf7#;@=TIQQu9$>sW^3lbNSlO(CfZSDn3zoc!N0MfyupbhT7KQRM}1I zsI{1)17%IL@ce2-mNWf3%biV2*dUeyU!vkuN#?Y<3=d6~`q>7Pv2yrld1)1YIMZX| zOyq@I8%yFBa3tz=mQ7jyUjui4Uw4~QPd$Bb203*Jk4ET;Ikf=bL4(Nsy{9(lf;i!@ z%>H3@qTYfiSIqYc1TiYxU;uvNSYl+;$Ov1;g~MFvp!h^$Xa5hAb6=T;=if%k_L?#Z zZwNT`7cc+VmnQ`k)#_j?Gc{;yF~ zOfL%QG##|eZmn-s^&I;VEiU~#f1n?j{Y?;lP>Z-3w~mM6yp<_Wb28W`7=yTvKC_w; zqHvSa-OxD)<{enRWoZaIPb%TOitt#*d@`Of6Y{vMg9c(7juE<=u0n&1e;T$-1amCs z&TbLya;e=~YpdI7_Jla4Wr!WPG$q4_Z0eU6zDwIIXYV5MBygL3?s!7ce-EM2QHq&Q zIf7cM^O_f0R2VodP+KX+Jt%^cps>G!kDp{Jxxz`=vm z+0j|o-1-iT52xV9)kLl@jXumOdwFcWF;@l9$57=h!X>;x-p$*T+=m}Bk6K#gVU?ZC z;Hs~w)h4L0xk)fWlUvN4e=liQGqW-C%#d8FC=@|;n;csi?O${N_B;2@g3ZQu6;~}r z`$t;<-oZ0}Cc7O}f#qQ2*oUf$_)`km?1dp`x^1pqiZBv=6m``T6gqRT<#y5$n8z+> z53l8m7S!{taTqr#GI7&nbv|It@sU3}Z?gM#>L^`2DC2_03v%jne>c1Cqmg_3b%?yA zIq<@V;8t6q;=$}s#B!>H3`5rgGV<(5A0 z`=udKVsH`dRL=FFe`kc!yz<9Qk47F#)3*a`uk0Tl{5=6;nWuXFHL+A8WpUD(dq{ai zjgQ{K)dE)Ak8K=TT&*Q)Jg4m?@KOt82@edPtkKJ??`kVw_F3r_MO2BEr*e;&6SHyy zIl-*xN#xT;IzDS?$f2#5Q#&G~iVei}WZt9a0a@ywqLBNle*=JcTN-&0e^RK4wlhlK z8Ph5Ocd^L9$WIKrrl>o6dbPGk0xvF3E}eu860x-U?kC*M9SOj#5Ewhi{C%~U6FB_7 zj*X%Yl!;YJWITiDS6r@RSHS5~-_)iXrn0~dn@&e``4}$&hSY#3cYbE<_oWQ*!PpcP zOP&jD`!5SJf1q}6=QGGxGceP#E?$a79nTGc7jZL}gkJ3JfMwg%qfq$|AjNy0B|>0Q zKVc)5-H>0vv4P3%{W3qqy2TC(7R`Ei0wnWec=n_r<7&Mr6Tbq`i`*-_4nDP?_f%Ah zN{~wkMubofl+9dG+^qR}QHtERIpMqxqR>uuPC#1&e|~LhNXgrRSHuK5Rg=&vZ$a9` z#4i0)=LnRc!EDy-{n-NXbSJDQY*6&1Ajq~%o;|aDc2lV_R~~VxuBO{$rEB@T=HtWb z{w;57P2<8v|Hm7713TR(5L$9WfMrr%mA%=?Pe(2@$`PSgl8N_ooE$$d&qne7AHIXM zIx5Erf1|G-Kg!%jSZUIUv65q8Z1F|g`leH_XWpkf1E2uwVM5VZIC}({c5}Z7)=U>N zyfUioDSs<%FUD7DU!vd1LgpSWWTU5AiW@m4m2r)DCYp?KIayJcL|A?b)#95iYzU4* zcAdQIefEyh&64e-HKm<>wenl( z;9It`B?TY=#SqA)v?YTN&0 zJt}+a>4M+;YN?AxDHZhDy!XIJqBDX(yAZsZ#dwUUj4P7D!!J_ldDIH+jGB>nr6Qj_ zf3LLipmaX_A?BgP8DBOV>#Q37Vh*@-f)va)B>|8&n5d{VUvx^7S&wP+iDGFO)K@$} z8)ZOx6lfXw>Vof5`_96dJ`fzuJD_}Uab~j@0@X2qD#Vrow5%(|#_qj33;t$tJe3dJ0NLtAJy;2U= z90Srhs3JW&KcOTcjL{ny9LvkcibS_5U|wza#;?>UQymXf8H2lg=o_Lim{gtGe=AEr zw~GchBAEL)b5ALiTgs^v`u9|ypb40xFIE_-ayoGq24I|!e6+ykJ5d*}0jGp7!LiW1 zU2enTY&@%wf^d89Xl?Z`X&WVnYU@q8J$Y^eeK@=$Rwnl#)b&p19i1W8%5LIXjVU{k zl{NX_sn+%Se4NTWpII{abv;|+e`I2D^oN9LFc(iN``m0RKnJ$1FZ-()AKTVpS?gEY zwa=c17T<(IUMqNX?h`W?Q(E?%LorgvPSJ&i5gOwVXns52Ruij_ARz*p@ldzHsUZ69 z5pJoD+(%Kylk}^VzNGfP7PiApne5K`Y4xj@mPh#T8|fMLoI};_LM#69e^rOgvb4$a z%kn2es?fil#HP!?A}CPnUe&cnI;1M=RT~fZa*yXq)h!TyOdqE0UwK(CIpxr6HH;C9 zRESd-yk{?AM9X_e?JJ|rzsMx{`&j-LNcAX(ZnV(X=GT|*I6@xyx*Yj}p9>iqy?hnPSE&V5_B9rS+k+gbb4aI1mWgMg&`_Z?j&W_7@~5!Ci9EUf zT$IlK8uA7jk!}uwz;9mBWCRk*?aF2c&c?LMja??#F%9bQ)xWc#3Wwnu4F43Vn3(}wEzJPtR<>pUQ3XXU>7SASD#@Q}07)}@GiM`P zfTEkRt(6Ht&dS8h-o=a>VD8`yu>H>gVB%nJYV|KoE)4$`VDE0`?D~%|b7u!TfV{Y> zuz!TYPgQ`pD5I(wq{a<8j^uNgXf6@Q4&Yu5? z7~q(h0j5?at^i{*3oCm##(()HZEx-X;P?-0>gM?0)IWk;{?P}Z`o}|RfT@}JKWsN! z+n+{uW&kQt2RlbMS2JgTyo0Hkvpqn?$ba7DA4i<6%>PR@va_=F`u`>WUxu`+(LZJg z+gto&AJcyzD;EhXPcu_RE7yPgHaD{U$ESbe8fO0)j=Y(vm7Cpv>dpQ!<)5*dI@sHK z{ogD9>E+)SGD>R6DhmtK{%;HZoffw@aWJ*Aw*aWP{$r?-v+4hUf2S3Vtp2k?|9=(p z-<1L||G%@mk*l+nCqRdZ;a@w#^snc?k>3AfiHJCO`p~odgVVFJFawy`I5`1a++2SD zAGs!O&dz4`uK#xZpI-iZ{`WSSnR%L-z-<5>On5`B)7ySk`imDYR6)`%%o(cD0+~VA zX8Ue+P_0HRO%0B(aUWE13p@}A=6{2X`S-8(q#s&IjN;7@PN!#Lh$HlXxNoJTRTlnr zJMT(`e;On7Zy?lh>`Z<_+(zEJW^Qf-E7HucW#?Vg$vBnsqe?Qz7*=!*LZa*@tS>*= zOH%7mqGf4+BbXi1)}ky}h85iy3%(s}fQ1)?>?7Qg;}}>LRR%d8hig5uQ-5`Eue+UU zsgc!@!^`$my@hCs2>*nAowd;J{bT(n2o>IBC^+gUy&A647m=fd*?J8l{I6Q?vURtJ z5pS(v9aPGQ+q!)oG=hgu6cSS*J}u*)Pm}pSd+)-1@agTs>b|kCGU`)%bl_-fbuM3c za4f;vxm2>KOXOQyk2WOj5LYk4A_7(z zO^jk(9hpjDX7vG$akFGmS<=zyVpaq^_=2CrBQa(L!BEeer}$BN^RpuqcPExl%&?5b zT1Pl7?vMighp$K(ehEa3LzzBAE)oRG(Y47D{YTmUgjv=DMh)#?jDN;W;F0}Qjn3P$ z|)^{h@>d%u_8U}yI;Te1wIWSuk7NY3t#n^iqCt08ERmILj>Q!&!lQc zr!GWcE*M`10;G#XTz~u}`vcJQfk0h5GXg{z$Q4mjZm*syrLK-40+RLY)zggW*Bqpw z*P`4^c^u|lMU0$!Oyi<@Qx&I+T70BsRAFlt$cZRjJ=sO$mGM8vNp1GX^=wevF|6(a zut|S{?+=CPk{}Y8yVOiV z$B1m@Py}}ZHR$zh`RegAm`}DT%m7;O+<6WhX)og^iQl7fm+Hnim3Dzg@3YlDglhDf zBGxB&7lrKHk3fcddtLpOGkk`6SznuLSU(E5&eKtR8PJ*44XxmHUMD=5bbpfa)Oi>4 zTp)P1o=wf$Uc=moHLNWkdBvUjpmbj+ZTgLRtzM(HOWU|R$rj$2U60g( z?{IOa>MG|CB3JGf*CsK|&cUpY2OaM?G`qBf2B%M0STo4jKJGbQ+&XGe4qauotVM?< zdn>4Xm#z()$gHN##o*(72ukHK7t2{fA=r$xqdf_V7=N9NDoyzI46&T2l3X@(#Is5` zgd3(85|tjDCoNb#zUbP7!?-XOfUiPxsRLK^ad$aph~fo9>Iq+)mKmNx0|9Hax8qE{jlhuE7EuY|F=QSilKrM40_`W4 zc-jZG37&&XVo%ALsmt5vB^eXBs?vL7Tr9t0mw)^(dEwaoI5H_QF;h^79}MaB1RT#&hmY1cFt5R9TitI?-3 z(T&sFJSX6%+X6R`Ng{d<;>J_dCRa>?;8Cv~i95=I$k0(ZIaZ0C{VpU|YyuLn41a#e zwVG2)t_lfcn95#-l}c}6-S;nfM3mxQG;F$(QVtbtW8VZT)x-ms63~Z=Gk=?7L^7n( z%&``+TldgoHRvHaygKZ{#TdON_02Sv9$IoKI_~GE%_ym0`4ro-ZSsfWA+VN4-%!Ssf-7^}4G64ntY@;p^^5+O|rhh%+3iX`X z(Zs0birYDHaUAKIKMMA-0sUDgys4pn`YBe*pzAiXWXpr*rqMgYH^J%b#(^Lje|K(N zfNLvBp=1Nz_RIO#%`yo87saV1;a{U)=|2e&q_Jvycxa(!m6TI&+!o6e0x4DczfYpQ zyY&!qe{rUv-;Jrqzssi$?te>z;CLP4l7ml^Q%$Rg%|r+87I}u-rjux9u8Iq@{>6c8 zZdI5N^*iy)Q0@R_Odpk6S~jUJS^V;C*}ba?CfbPF=Y9osvu@FbwB>91x@W@H z#;(#_dy0dGNR#UJ^42J7W7YxB5wPJq(IX$~(;P4f9>$#v*?+h`G zSnkP3)vRBV=m^~!4dx<;=?x}tl)H9!@QCZwHbAEW>XBsRzItK~TeN$|Tl=$dQvxti z#h`h5Zb|kaXnzfp$+LlVZ>NQW{QPSTmIap8J^_cUF%yaaB4G8x^FwVwtGd^oj+Jb4 zftIN+X)qG%nOR9<)opbmYoD;}p)*5X>#bw8;HlQ!cn`j*gWszFvex2M=r3$XZ1!}@ zj72$J4E)=EQ5<=HkRb;2S0LOPp7sDx^<~lSvxCB`s3sGncu97dgwvgk~$M1!? zjAJxV_(2`Y(A-Y`YVNS7a`bYP0FuoLL(98fRO@*Qg8hOWY1$|WTAIR;^0I8 z&*y;S?SF~*GzXasH0pxqbT?1%MJVj=cfz&V>Wjt-;-ynQL&gDzhb`kUvapwT2V#09 z;h6%V$u}M(DTc9?rGw!tM$|TG!J{}3kA)DtRVX~E0u_!1+l3=glfh%C6803+a%>tj zk~DqB(h8TEHKtZ+^UaWVS z#W<03d$kec09P84VL#ZJrWukflwo$rxw%2ysq4Nw=UrEzsyhy(g{Ju1o9sY6{{Fbq zD3kio&YIaHS)QfOs%MRv6QnxX*qQrSVcDs0S zdrjPJRC9y6&;F7-r^S45m|bKRhRK=o`GvdeaKN!FFLr=vK6OHDBtQBNNX|&f`Z;oO z;piXWX?r(XaNvM+gAX(~DQgoUYN&EE6@MQYCRu%PEH(Uk>Z)Z$9l{}f6GsH$l}6S^ zJ?`c(na*nO$J16y{C?+P@dW|(+{HmRMaeSgty zRr=ARokM)J24VOTdFwieeRahdU_bn4B4?%PZHCJjiVuE`J2*LpT)jT@X?hqLzRY#{ zgTA53yLna4Yu&^F>+@_thF?BMp@n{W#kQNsl>xWFPrUmio0T_6MH)a_tlNIW(@Hh_&Gq-D@Mue5gvqatu}D zN32B9{;z1_z`{X4!R>8rzof&v_Thy*wca#HFMa#z3iuwvLe7RxxC#Ts0mUS2)gM#t ziu?$kb51za=8iW7bW4o5uYZjFr;Cx^2b?MD6_0f9wc|UE;@JlmLD<2kGGLL&*B8kG z7P~BNE*=*zvDY8K8e-Z`fxSQ2f04c0hZxjRZVaSrv@t6X*p1fs8 zJ>3+tWG}RN$`SzNI{@t9#!6dQx?M+Q0;#}I#UqGLTGUsD)xZo z(u9x$W1l|qwdJlerhmg!FX1rEx>Srwz22K3P9?3C9NDJy1}&V?5K%!`6erw2G)`~9 zg1~!Q_wxJ%Wfa0uM^+iY(B@u-DGo9w7-;h4yte~8pj$)WD%2{{n_@pbX_?=Op4~l5 zkoiS5SkPoh>NgF}y%KXn{F>H@(I{P>2H`m${izbvPI$?_uYai`20piMKee4)aAP1^ z*|%HCpo%_G1IXrmFw0?2R04Z*i8tzEP!2;EP;?fw+LH$5Z}<;#_}a!Gd0QN>vTCn+ zRf*-d=a)r08f#CWcIlkM>VP7CoB>iGTnO|^jNo~8`+6N%5Em{nNeLit15-&avb8s% z&Q<=R7diEEL4SyUlap%GvYSc@D^Qx z-v2&ovc?1L`sLK$(`>0waDg9yRN9ncuO0|N5w2o(wmfqr+S!a&a-~G?{D&KTg?Oe% zrTJW9{QB!6uaygC3DI|*C{)om=R<-79d#oWDU=n~-+ydT(56vG{3gF#pPJFMsdINe z;7u@yC;{&m`EQN8=jiP)!5COdUfC}Y8rfk$_PW4hA_3V<1I!fBDtzau^QcdHDu#%> zEhZC^(s89l9N3jJu`QEf)HaW|`U{`h6qEZhgI(+y;}|y1SHU* zGNfk3-pf|EaS*->nwmp7y_?q&Uwc0<+#U*LT`lDNni)hZXMfmx>pLvtd+&Fv_Ir`W zqJI`8404~{AEWRZvug`&FYdQmE>VT|<}5Ln9;a(XPplLoHsd5dh%7gt(wLPfE(@v2 zA#m?(Rl=!a#j2}#U5W4E6EtpZTPSyHJNg>T*tNB0^c6oBKd;#l+*}}}2aDH)6Z|xb0BB8;Zh^97j(4Ws ze8OLRe6;y|IRc$(hHQ31D`GK)C#@Y3#L zFlz<#qFm7rvs8vOP|?BRB$`g`9ecH zn+Wu#Zb$qmxOW2fTLB}k=K)&|yz2EQ@%;}Gfan6krX?K4-qTA$WVOxy=wbD!%JV6< z?V(iO;~!UBsfCdUNYslEwDZVxS-`fp?pDRQ)s=lL?^g^bE27cOE-k6Iq$S?} z%xg#WSNi!4`1Wy8m8;0}x0rgPBWQO>@-zxn&^RtM5dhH zMir(u(7dIqdDNLay71-H#D7v#W5Frf7c7IC@Twa<;;m)**z)01huG@=>Nk?VzT`XF zbIi3ti!GMxY9^zM`EZ%1pMXhB9VbRQoWj9ci7MINEQ0mmiuAkpoiV2=IUZj}e)qnV zN#gMcPjIhPw4KC{XN`!SSLgh~QF&fSAUsmrg;>FmcUXJIUdVjXPk)f@m7VBYRF3-o zF;8*DQ{+&Qvo|%e%Vf&M#tQ@Q8hh2LdX%X>QGu)48Vo5TGxe6_bmjMWT8omuKaEi54lgv} zLk31wPSol9E8xbmtAD4w5YP}g+Q56b3jB5T^fOE9Np=gcjZROTjc)vUGl?QmKRId= znbSmYO0+cjSlF|vQjpumVnQK;pRAbB!_0(*6j~m{BxKi}af$08jd2RLIOCNca(Z@SW3qO;ToX0?ceGRJA&y6RQ%1>c6MuIOC&q)o!n2PxX1t|VHsuiN?CxM0(BD}^H8;mT!U#vf)5DAk8GbP~({!N-m{H-|ykKRhsRqvx_ zl5uquyS2D&xA-PliV&W2qkOGcJ?)&kGo5!2X_%=7PKtunMl6uV-CfMf2rYuj)yE-L zD?Z816r+X`7=Idgnhp(r>z~_>dwDfL{e`$K!rK{(fL{)o+;=aDDGWjbonf4*0cxM_ zJP*QZbihb%RYL#gmm9ECI@s!4VLFUFVw5`6ski%!$NwcOTbC>U5@u11&X_)4{dcTPOSS5PzC72vO%>TlICRC!3=}6{hAf z4lG=24z&1#brCMOW9PKn@lb2t$***PFXnx0B%fxVM5kef&mY%HSX-}27-HB26G7W+ zMidXu`?0q6XZ&33`oF3H=g43+#ns7vF;Fi$gxJi@WBaA*9Re#V3^G3rEG`ai8;uJJ zox8Gn(tmNkI@}HGlN0gXyopn);^|YuZkdmv@;booH9z$|Pt>VN;s%B48$+8*Yf=98 zgq_qTg+-(}R%=hEsyK_wI8MyC-Rwub`LTmUXjaj3oB!G|809j|D4(G*Di+e%8u+!n zOQMqwc2hd_ee!HdUUc9-U2=ku3bZL#F?tu6aFVFI{OusqKj=Bcg=e#*6uq zWePzJy=@6YBEdWE7HZFXvtd~CL`8RPxa_XsD?X|sg??2jbeASBxdCrkj`h3$%t;F8Phd>Xdy3VSE+h6hTtm3}b?Ui?jc+naWw6&RY z(xNt1Rup48$7BcdEGg46H_z?CjNqoof)*Q6x^~SQWn`DhrO~+ADUzqO(dOmgi*Ctu z3t#-`;w0D^ci|2y2MGFWj68jm=FC9lSYtt-xrtl9Ew2pc`LNT zyx4mgq~pWrTayluhP@9V@`~41(}qD@zG8Ci)Nq8v!ZBxf<;gTYrgp4&YATi|M-2@X zEUxUPFnZvKLwd7I#Q369Xb-TfL4W5o3r!C9-FVEm&lSbiz_LmQnq%SS;i5VF~<{W*K+*7YQE?SGSIuXexT zzG$E8i5znHp!|>44XyIh9!CXR|Fr0qC;Pf zP_fp@g$EVI_KbRk4;0z6r!wbPn#+Q)O)U>k7xo^jNA+}@rs9Jq%MKdnXhKq0_nDQQ zlw@vqzpe!IhKLFBE;hJCkAFIjWj4uj$7Lr(cT!B_srytKXfB!*!~&FchWc=0+CAyP z^LUb}svA3_!D?<;Cga7j3kYy%=$|Hj)kitzno^|$bVxQKY^1_y}r27 zEqx<1mKa659usVBOZa6GJd+-&0r6Lwk32pivP8$yv8Gll+20Fk!h}uCsLjU&QPTT$Hk@5KVlHrJdu!gEq+Sqe=_kB=t zew7-K>$v3=J^|B@z|z=vpql&wM;d^=qe!FeIkko_@%j|>5rXDEyi5M&Yuz>3uM|AZ zkV5e4ly}EZjFzp-e-2`l+p;*u=DB(!LRN+`8yM>tuB@C|Sbx>ec{JFzyh+dl!39&( zdk=K!x!xZ;qFTGdeu9Z32X!4{R2XwnA{pfPJ!t}&xSh$Kdsk`%q@1TdD{7uOT*Lpw ztxk?%@q)kj9L6*h5ClK(=&y>(NmE(6QV+M-d1EcuTMZK3_HSdCo4zPDgA{Gx4L9zx zAIH~ZDWlp}Ih^2kCc$Q8l=@YCQuj@dTwJXKec8GiPh^FevTB({+y$(#yM;cW1OyJO0@& z4rB5SX@S($95dn$;(JTt3i^zdpa21nteRhu@l^H)c?fm&tYR$rnD4c=?-)<~cz}~i z{11JY@PF-XrVTuISBCj{uSerJ{G&&j7kd|1IV4V1>$lq62UTO4HxB$SRB@h$$*)6W zT6$GEyAUyT*IS~n)ouT+@m+mO&ZEGFgO?0x_Iu3zRAjhG^Yy|ZL;l2`pk~y0p|-6JjZJ=`H^!Bp1(*V z9F|(NROiT=rw57ddT-&X*DzO>`o=VYHJ-WAuG|9WZgQA{E9bpSAC`@%_)nYpO1w`E zihohpXRcayvV7I|H{c$hXPGpvDH_t+#*NB(q={KY~?HJ7f;{Ym~{AL|2<9 zv@p8&zRP&;80vKnAC4ZxegQI~I2A0$Zhu4$$KH9=1Gi7HDMM}LisDZXml8~l?B;JB zI$S(;y{Y;quf=PW6wqM6LJl$AZM3N;@_x4w{$727nSaG9Na)$9wnv%O;R_xH@1EWV zpGD~YJu^Rx_vP)VgNP@jAln|%RcvhM*leQoyuko|07c>Wi1{9BAGqc;ich7O(|_qU z6GYU6-t}gKNAQ%>po%EU8$8%N^vo^7b>1mottXHmJj(3ULD}X8Runj0GmsfH68gHW z$8z;LqHIHSZH38cD|y5|8iLSr90Z7nwYHEl^J5f5^K?S_*d}p|<+0M3OtZ=|PgQxs zc?BG$6pIrH`|JwMG_NF%VLtG5{eL`UEo=OJ92}>cGX59s&&%=JVvfJPnF@*}IBg$k z$h?L^Q*?|ZaHxM5Of?y|<`FcS+g46CR+Hi3Z!>(kmI%_XaMxz-LUb#cP8XT8?V%^9 z9N`0trs!j^%X~pxVW}+X6#q@E_FYwJ$rW?Sq7q$b#Jt2M!Z%g9mk9J5Vt=qL2Bt(A zoLTd0odp|duZMjY@@xHreSaJMbyQkCdji&sy#tYnKr*Wyrj$KGLt*EvPY}1o`nF!u zu$Vq{;+SEv)?Oq_xv22y<93peH->sUcC?X_PzQ=uXEG9d4Eq$d@nT-+j@QL#&jTMj zif#HJ5BiuQsIFZCzfpMo8%8@GN?%Q_AL)-I$f{mbMnDcJvAX35$ z4-2KSqx*u|?l8i0BlS~y*vh=ImX7fUmH@rQzpK^hHZ!2McB*$};eS8B<=#nsZ4M^+ zdx0C#JTE*eR>=s@d$N$i`n05PaR@P=sVJo$O`8Fny$Z`~lh-6zKsjaPYiPI!Cjc|pRU^G``UYLW3&J7jo zNZ8(A(a0AMzDeA@#Lh-4qyn!N2AO^9g!Q^j+g+b=&lllR z@1gZ05-UoFX1(vZSjg`dIULECSuWop;Hr;L4Xko z>hI00uN8d6dNokX)PKIC?DuHmTVGR--UTpc<0fMdD2LG}YhGEPc&88t7nNkLf@pGi zNwqRMRF+!sRB}Sg3{@YH9jXl{}6Ri zcC0)?4`tbSer|DG$Ep+9Ct<6Yt~=P72)GC5uV__|T(ecCVeY!>Em*Mdg-Z+$TCoH| z-JOl^?VtVd{o5Q=d$kZw5=lf=`6Y&dkFea~s85gU^M4t5+)|Db{B(RaB!^%7e7_k(Y=zgUd;o1NM%bxWFy-!s<;m5~TCQnjHl?pFBUn_#`o!%wgT}nW=-hG@ z%fNuFU&?1bQAYwx%QY$vvug?_bS>rIQ+Om$a(q{k%DK*z3~PHtYr-$_LrceBp@+Ja zhSAuZ3X_63gqCzN#wBh1LIQT4nEjVK@_!w#5$bNl5R@}0yI<9mDyfaV+?36{fmd-Y zJF3m%4DJaZ@K>t5x5FXU=xs^gzHTc}$&M_n4F~BrLA3RQz|Q+(bv84Vz`r*OkVN-Q zh~VRF0<|&#?+rsuma{4B2i;P38{ts+n=J20>@GBlId(>zx&czR@hq`Ir4g38(|;cX zQcTWBSET%dpt;ShJkiI(H`6?)WN(!C@TkBfh8Z%D z#vqT&Wz)aJb{>#$K9Xo8Y;8osL(j=C=5*=0(JGGYRSC{0OKD)A@dlo#jej`>2sv!x z6cp!WWk2=KsjPiz&~HyUQT%is7~3~$$JNGQyRzouY$uxgfApdG1G#mCESu5Yy|7%y zH(Q%HlN6<(UbIlY>|^={#x2FDe}cOUh{x?3Jmi?&gHEvI37(J9?kV4DlUrxmmKUfDvx@O)&jRHZ%dQf4?`+Dv7g^26)``85oOaImgV6X7~jrycvL6#_>!i9dZDq*-PIIp zj6JEYQ20cWU?N*P?bU1#e)SYGe0fQlHG%4&t$DhjtTH$;baauDDBP;=#;t>#I|EVWV59aZ?p!-}5Ognla*5tuFkiIm)8ApMwKl;SeS&R@)}1q=KKJ30A2w#R zZDk(caG0IRucZ$;FPSwRF z+}PfveAG4d#D8x~`GtM+ZYY`CVJ|;P62a%1jik83M^*@%xjW28ew1HkOiL~XGEOZO zlpBIQ@6%z&0)<7hpe@j5U-z^=h8BvijVjH5_y5tOc*p6xh9cjyDQ2OU$8+0v48Es{`Pp>;?d`+7#He` zm==(k9l=6fkL*Sq=MZq;i8A9zl~o)XW7NKfyR2LFz-x4R$n#MMsE~Evp521={6}A^ zqt9`peXEyYi-59;J?UE6Z9VcR>jWLjzQ=a;8-J2H{Y-y(475wSBrPo zZGXg6Wc^4T9A6Kx&-G06;%^~HCf%^CG8&nhceuz8Qx2#r zYtNjUu;=fq9Ecj3!%?tD@;If@R8NC88Gl`|`vn!}!(&0W#eyp+S@_&g_>V$cRMoMq z;MyZZ!2y1kgyd7=}#{3VR0$`2!i2S z6Y&`ieQY64g-RD|d1L*wCcHN&aF;;oa*kNQ_Ky`SoOGOeA-k==!S5pNq-W; zNDQ~dz`MVVjK$Eb*nP8H^7r%M8sC%mO-i0S#F>symII_B)3j>@8`wz(VIJi0_fv4o z-;Mhf6!Wu|T68)v{*NDgqX;oVcZ^1a+YT$;Pz2`DT;Fx@<7Uyak|tfQr^FeBO{YhY zW%xrt6`uh}nV^m3d{Hmm74aL&Sbx)Ws@6=fgjh;G)8?XooFT`=ZJlMiC zx3Y@s&q+&d#+&EU`1yFaOYzn|y0P|`ia$i&%<=CC6?kS%XnswEb3J~qh=1?P(RRjN zsqT*3bYc|ys}evv5b-DbX-$ZEh!sMB>xQj#dkOm-1P41IIF>bDPmy@x_?Ja> z8JY z>v7M6cGkjAj0C3gpAa?k5(z^bN}u%yRV<>x&~`MEs!l#h!$RdYXD(svF*$5va({a|NvZA%EYmt5 zW#Nm_y{KG1!Kr*}c>TUpY!`55@?<+>I9P9aW9A1bZg_){q1xn<-l>=L;HF?MyBAj5 z%&MN% zCSUMC)u{PsI;&=?{eQXcW8sU6{WC{m`i34mHdw$tt=(5(!Q2qP+L9JS`B6cr5N3Lc zp^s(@&)esXuvW!fXu5O*4uWSZ9`824`-S(lWWZXh+B1~Dbs!@9^sPY}Su@gN0@7Pj zQoq~T~d5P_PW zhu0Z};tE#c!wyBOLlC9wUfWQL~JBL(0g1q61 zbbk*KB&473_$Ox9&Vvo|D($_*1bjafKe#iGv+o@+8y0vJJahJk72j5_XL=Hd9xSdBTp6{l-G^gom9NUemnXU%1Q;pZ^d>xaY1%AcYh~ii4#~O3%DhC=Ch5F(4%Q| z_it4#rascKhPpB9TelXlqo#+aM%R!zeQHQ4Un>&nm?!3{7r*2c*{!;Fg3w|i=eG|9 zmsx%e@rs`_AjwnaEmQ!q9z={v8;MfFKJGaoE@jr>eoXq!P?!{7f)PJ=Y?reLm`L28 z!zk^PvVSJjWfRnf@!3M&zAcVqG;loQefvI6v^ru!2~V%7@gjLcww_Hz@)HKqT1%Ls zQr(p2)`Ix{TH=v0>C82oK{`$UMkMUF>561PAk@uaK(FE{q0NcXe=liTGS1i9e-ZfY zzvOFNh)vlKn4mh-YGS0~z$hOqloRp&Ll*prAAbx9=N=|crJLs#i2p()9*ouGZl^#=BJnZzXRnDNjlx1m$;d5+O7s5TcWUqhAlISreeO?7vUQ=2z& z-ukpYp4_@XTMtvh&CaZ>Qz9-{M%hP2qdi#o?n6Gz=;@BC z2+-(8Qq(?Lw88+y2JEpp1Vl7u2bWY3V!@B5J_dvOW%vg=w8Avf8g@@| z&AnTVfI@-0e@POifk${5|5+5#LFl zE(0?^hXYl2wV%JLba-~H%P8{Ld-2GSZ-fU2@Cll0c-rO{Ev@)@|ukyCn z{tUVClWIb2uMws$7c&J;$t9BkJFJ}bbFDTUCl~`yM;QZ706#UF?y4FDJGMa9tT=JB z+qpXgyO`oxT2cBRj|J5aFNOX-N}g511zB*^Ur8io?pN_+(`)|#6PW3_`s>m zcZCOD#ah#MXKg~apAGvCLXI4VAsbH77T+_i`K1K{nopHA@C-tyd^B`73RK*|sDq1* z_e$S)ok7Bhf!HG!Aqu^aOMl_)Hl--$yrjR?L6Qfpw#ycrgetHi7bIa6&sv1gmhE7Rc!W{NKGjx0tQe$)8>`kWLC8aT0@Y?Zg6zThWhgTBbzwIM&5BrAzcY$05hUY1m z!a>vgCqO|aS|rezQWBD7X(+-==OcV2kXLEAJr9i!k#vKZsYpzDxn5jFLH7EQ@DNUg0Mr{8b_^EjIgsw2`gxPt3dY+wWb)Gt zRBVBsevb@*+mD8n`{bOQMl3!Fs;?pAf8dPD`?VMOuy0O+t0qt!p^n*lO_kTfoI@M$3|lAZ*P zsGh4+s&<&wm)-mPQULQ89%N5Sz$eLFdP9(W42iu=cn!0Bo4*bWmj3mPJ5+VU6s?OD z6xK<_#J<-`QVVQISpP~X`$y55VgjL|@50>bsZ#tA%9Fb?5Yl>AL z{jFc%J~(m~|I5neV*W2H8-k0Qo%ufoeh}R3|5>0=1Gdhliujr}91w2J2H=71==8!4 zp$M1Y2+X~P|7`rgAYgGfvOq)vWMLqXtH66r%Tz~qg{oEAVELK<+@UdGqEl6;4 z3bOR#6LaVBXmu9{CH6r1ts(3O-P1o9Gde=O4+q+xBOhI7fGv)Z16>Q|M?3|hF9@t(c`i+KFB|)x4?7v$X@qJcUk$e5GocW6H^bI6)d=>uuV>M>IU#qbYO7- zf$|F7tb%w6Sh;+BJTriX@sk)D?Oi2+a)d15(Gt-m041&g z^4E#J>*g+fDfc2NdR=`ny#P2e%-2M01q8OAE5mol5H1O!o+1_Oxivz*I)N8o&-Ywa z&@ac13&iS*ZlS;{3Gn0Pu0mZt`f=;a>o8mEvn%#3s;Hwb9_kIyxpvc z>HWlpb8UP@er~UT4fzl2I{d}05CrJ>N-6g@tWsq1Lq%TxN4maSe&J$$5n9UNlGpFb z?5GX`&3yOf2hsOMCG08J==l7K(Uo=6(D+SuZom9BaclpNbS*-xk0LuU$U`MtTjPeq zfO-h(B(Y7WxY7-ZM5dmR6|%bqrP1DKK2UEp4cK|3oQ$y_H#w54 zg{lF<{#;WX=f9ZXKCR~ZVz#5M_TsyGP=HXhV z3C9BsVaP8Y)+l0;zw+{`i7{3VW2(x@h;#EoFLrf_3r{=3bE>9WTX3rjVRAL+zGrEe zb!Vgg_`V|nskAtf^3Oz`>$YVAhA*?nwFWh)2Ct%jFPbbMrVKB*e)!O|C?$m#RD=p! zfQV&fVR#?_!|w1mKDo!5G!@m7Kyb#cI;HR<7s5h7e?koPZvZ zF0-W7W#ty~xH7o(Ux&BU`5vv(> z|Jq&7QY_RDKuC$Ho%}cCX^sl@p>5e7(c|C8@3#?E??>V9bH{} zYSzRwWBjy7zH3NDhiJ^#TG^EM9QRjRFs^TMNrPdBy1yhCWwWqniCRkRDTJ%U=)_UG z1R2qe4~P8$%+|9-Fli?Vs(UV%oo4eWdXDCgNHdQc7~0ZPY=65iEgJ=tT7(o+)>3Mx z<>_GwX{H81->tcfw>p(FReeHsgA#}CG8DYw%fsBlE`)7r56_^OwiAJ6gNP73GTVeV zSJHQ&?IoIjRo|AHL>}OW$%km)&!V*&ODbyrq)UycE=h~`SSa_pOmh3i?N};YHk$G3 zP?W|QgPZUE?1{5|?c_${!*Ku=CK$Z~eE<`jI^Y0Ql#+I$-Jc%viZ=Z1gsEC3S-*A_ z15U53tRpxoLO5Zsyc4Z5vhM4iLhV3`GOLN0m$8VDl{!x+1y3R56{b4F8Y1WPXe&j; zm>l_8*QLmeFuX+{P{7wUho@gb5QoGpNS$F&x|xp|^b9h~Tf91%4&tBiR}r1(fLENI zt5LuYJvdz=^5UfY^=}6xypoG_jS~SCZWQ7`;B4ebti&>30_L<5veiK}8taFjNy*jA z?-_AKBeNm#)4cdpt42bYrO6+Yl+9HL?p&$x@htgXU*N-3ft5T%$q_DX0!U@ktn?QS-^bQSmN;BQg|c9f zjDOxEH_?sB=Q$C1-omOl{lmU3G5kr`V-~V0?)K%95x2J3w=EbNLwP-do~8$P{c8%C z^AIL^{EITfUOl5*VwOPibA6S9BBS zNc5{I|5))j=dDZe9Yivb9XP!0;nWFwcJ;c`GMG&D1z%L7`h8?O`l-fj9a|I>)2uhu zieqSkt9uQ(&ACqik3m&=iOL4r?|cX7{8aXwX|fh;wArhuFF8Rr|8C1HVK%=b#qmeU z+9ZZFWkMHe9xizRi=~?>l&GjAI62gBz8jO?#(dfN7?@GV80hZU>46P)9ne+K!T=jd z_0u1DdS{eYR0R+D)?Hs3aOHor!F*x1Wq-c^3^8292oJxQq?+)J?z0WQTYnGeAmlC= z@{`7Ammm8Z<<(m;Os^S+(nImu@PIQb`4*SlpDFU1UqV4PvM$-oO5Z}kS0W7aKL^c;+S~Tr)F&QlBn3v98*UGG zRmFmpPo_}Pd?Tm7+;+-T4*BtiuyutC>;aYt4K}tXbP4z{xn73l{=so>_;*a^m_1o* zfXSrTStc>0D9RG{H^VUvV@@u?@PpT>AGfqnZEeU!Od^NXIu|Na!F304-FNiYZP~s} zh*1|OW=ZS~=zEv$v=!P=Ocb0Vy!Lrc`k^?V$i*H?lR2@^I^!0`DhKskup8z@pLtWM z)zXugkas=8iPnAiDEV&d7qJr$3?X}lbt(VY{cDJ-UGUqJ>XvN>Miz!T6ce-HA?N|m zw2eEQVd~#(k3mkvH%tzoMqAj;>rA@KD|Pe3WIaFst0oViUuUk8@eR4fO~`o32#o}s zg*0m13NLhj4{?19J_rayXDQB~!GBPaNlOo;0neZK@!~19O|Lx_$j&*N%{R{cMJ*E6 zaqi$_CWukhqeIdN0|y?`s)`pG#YK8F+Su)VBzVkI_$n{{LdgJ7W^W)yH2<(C8v{l% z*K_n~;x+1KW>lM6A-BoVXgBQD51UcPZT=KQ9_(;-4JBqTQjEFQ?g>R>XdDOiPmAND zYgrO|xGFf~wcF-;|yuP!&)Fw9|Roco{2kF9e%N;@4Bx4U(ux9-NG5 zjaI8Y38virozDb3^UaC95B11J)>PD;3+~pc6|ktV^ZOiV*(n`w_+?O_x#0D=i$ZU8 z-EuIwKJnWWU*^~zM$jH?2l2Si7@6e@P)eD9qkat!#*g$2BEF@e*@yPi`Frid!NyVk zp77*yW2~NFD*4c_d0}Zx7^fyU-3w z#%*VNOeUtWSF!ufeSiGx)lt8Qg0kea7&J+~$?XEbTw`tL)I>OC!H)NdYE&!N#<`38 zJ66lqdjpFUiFWD1uX72;Ea;s|4wvn*q&W96obU(hX$Q`v8f^o_A=WlY4+Xgpq)S(_ zL)haXX6YeKfz@=|Wa0!y1s~%o62|Cv=q2PB$s1aEN{tFjZE4HO3uGHrr5;gWGW?G# zR;oMTR@814DzvX=2R@eF7HR?tI=bXTAXxTzR5F4^*l=yOI1FpBh?J}Pp-(EueE?s= z+3p_^itX(+i+URV_xj#@k@Is@v1htnuc`caLy|rMTXx-&wiVv;bL^Tc_2m#l`p;%y25PvR41F_>;;PQ zpRc(}i;P7Pa|zwex*?yjw_Fl6jKqbczLZxJaUZD4;6*+fA9kZ{pCgM17_t(fP(8xk zzCKdk$@q6sOS+>o5X5Z^uq^kz>V6FI zU`2P4@eB*p_z?{gFveVye%4Yb1u;_L2<_r*of& zP&GY|9o1>IwY0}qV|*laI^&qv*pr(03{EV!79u^~y?oeFtZF~nSq`I`P@xBc8uHiF z#(iQa$y1DWG(O^WHAHymokZYZK)-9e#N-qtZhSBzd1xsPR*@orPCX5qESbsB9Hqb`Fg7P37T;;hw2I)nyRDA4`OVG0qep{ zQ>(R=1s$0pNBX*)(+oaW&M}T&Hb;BtqyopD+*fBK>awtawWpakjI)XB?raC%I$@uW z$|s$`JL1~KES)rg)KReCeq>H;I*NDSn)$%`taft{65{7vWk>nqWD5bT1c|oiGEGq$ zB5|s+$KWN-M!PKJJ1}f@==$6IXeL9;(B4R~pRknMceyMrSP|pys*sm!Kf^s*bU~V^ zH1>VnB=oj)w~DLeMJrTVc$HTwFGF*1p-`83#fsSHh@;opHII z9Au3Ag=c{6w?3&Ghd-$$K0LBrt(k9cJH_ZY&zoYo>1XEZz%q*EGvszK&|ElIRkK`x zaW5OUXPXHzXq|BXZcX#@NMjfu@;uB?P;kEF?I@?*sHq5zDtT6+oC{l0-q&T=d z>DxMZ%8cTOAyohbUR~DWY`>3U%o_p)2;31<@JG)z}f!~5kJ2cKFN%{st#MKl>4Z_ss^L%&mHh~V2Fd0E&H*Wr}*^#l!f(=3eM%Pqj5tpwdBaolV>TQFsMCQZ?G z@Z-pYM`C~r<$PJ8K?~e0FJ5j2W zx0=5&6deq*;Sneq+)WU@$VfeyC??r*Vym{=u3ay7rlPdWU2W*Jy$B*#B5jum-JSUH z`O1{gFGHY;OU;%gBxHzQZ=afC?XZ}%$s8~!)&FQI_Y#HC`k|+?2rCm3egQWQ@giv7eC#a(Vw${U79!9vhQysLX+vI9POuR*iIY^mEUAD-c~cneSo> zjpvh2Bt$pKA?5c5Yo3`U_WQ>o>XN8QNnn`L6A#q>_WpXL;;r!fy9S4jDKUn?07mM2Qus6mWu`KCVswdtw~*IrcGMxH5HaH zVd9$=#MqE(t{?L^53N9xhjRK zH8Gp`Lw9Ni(tP0l_ST^#JRl6a?5cr<%16Le4Lzu(mhP?NUu10omYY6kh}t{85Jr+p zOka>Pn=3FY8#;PS=qg%Wb8u<;6o)3Laq6)7)$ce2x(!8i@mbb&pv;?K*GyIq%*u^E2d?TFDl~KCiVpA~PEA1_Xh4 z58?hEx5-@Ov4HPRf3!f+BrShrYX+0$Z6weC9V7}=sAUwEor1_Lx2)MHa<rw$BCz5XY;(Nm@xPv0wn~%Fn$&=7=RlX5F->Cs{C8@K31(wwuE#nZFwi&QfC`ww z5wWO7Da--l3L!?;bA#4P42#Dabq-;!^*6Q&VRPzZ^|Tt9ltSCw<`Qb)Q2=k%Q9Wn8 z4Z@A2RVZ_eD6VpPD~XXv?|X-WxY=AYIveC`>pqVm+CT-1q}Mvi^oh4akYu81 zZA6_hh!y9pvEg&`t$6Rwm`Wlx+e#ri`tSAonBLJ9!_2aHQS*?H*Pf1c8y#2c&v6o4 zT%|*Zy0S>!l@Agr<>AKnv-0E_qEqmK?7MgS>Gizb(1nR9q-AE}=FY zvwZc&zQCU##^`pFOfO^=SsWe8HNLIBQN+sAq!Nz6fsZjAtKm6i@T^Qk)WIfgor)Fs zy3U~AVVfOfmjs|k*}5&lztSfKYxTe{AEVBD#2%{)F=VRXe0l6JRSj zq~^VJbCF+7K16c4p8>*_*}KN0vlJE=?F|3+D=x>GePo&rlUV?z+bDXglu~T)rCn^) z*t}WJ1)gP~QJqh+NwD0If(th{@sn%w{^;Jz)+;wc>@3AFIf$?wQ-4Lyy2{8!@a=_j zsT~EvxRYJ62#@Kphv8!9%~&G0*fk6UR35JM?tafG$gx!*pTuttX23L8yrF(#x)^|J zH6-ZIGd+_CJUIi!=|N?|>+^!EEo`^@?SJar14&3$m7!EHWGO3Q^9;G85wdr2tXR|j){7fvs z-`NB(=8cep*rSVdZgVMbi3>~!YGH!ra_dqmnK-h;#R`Ei4c$(R%*W@2foC*7oj%@K zg6oLMj+hj=+kM@XKf(!e9+y{`V70sf(v&e+V6e&ISwyb_7@5l~fntLEXFzGp;nqSW zgr$0kB6em&p@Q0@_RkRa{fpX*6+Io#PU_D62FWK((o`Fg-|5}f`yRPFVY;F%z6V*> z<0Y`XHL-vUps%Mm+5f^ioU@a=(yqru@`mEgV-ImD8-iu5DMxAC{b%}BJ zQJJ8qKqS|$<%|9Un)#$vVQyAc+KpG zz}dXbuSb(pgE#I+S$71ugR%!Qj zkM0$&n{u{Tj~F7()lJUthbFb_uvU6ZXVECzVM1iw0k2Oj(UgqycX@D{vbFv1XhSb7 zve_Aeo23C|;D$Gt0ts4A;Ent8gp5Nk)ZPHJb;>qKI(s7Xe1eb9p5~-uLslT&MK~EQ zu|C6VRW&bEV^CA3qE8xmW~Azw_D0TK28DQ3k^gFzLa5KDt{OWfh zPuW7xE49C$Nitq#ZoOp}?dNd;QnCgh?`EhAVT9;-*2*vHgA@bWj+NuA3*Q;@^ELqg zy~YvqN}crgyvqPcH=sqOSvg8-#MX{yFlS)6>X7=68xB=*gn;SJ*_-?;Q%5xNW<~_$ z9CDH=dke^SmM=f>X6*frQE2e6(xH>MX?!jU2eP}Ah@@ZdOA}qit{Mkf;ZaTd%v0@l z@5myNFWRV5>=E$@=FK;p{R6eUU~zz2&T7!z>RzG@iY0Ea~%sAjFJ~0I9gtQGo(?A>z9q>LEStn&;+@8{J}ciHCDS z*!6=kd8u9VSmdA7DO#nURiddvyrH5Oo!ukV+oh`m9|e>TdO-sz*_iaKo0|Z0Ip6|U z=$1*xZ^rYd9-rx{^3JMb9}RdtUk9Yhcs}#a>D!AU$oCJM9n6>qx>qnMq&>~ zM2YrcX_D8vwAKO<4Q*yxnt;8z^=b!Wdt9Rz_?v7fV-TAJEJFLrZWTGzS*kC5?S{S~ zdrn8FGGaeLPPo{IKF>^!b$HjA-pZnqKfwWv<@sXRbRE>yTtRr;)G6L?`q~<}Q7<`b2QT7_ysS(Xw-)UVgk*M60Q7+xFd$&5o#_vttBKHAd*D zjclq-z?VoMnt`!()jj>)?-2n51yeGUZX-}C4pd{4Fkovr%m_$B(sY<8U-S;96WHsh z8~0bsux0C;s6DVP+oqL{bx;MX>Shewj9S&!wu^4Refl*v zpgJo}vL<*LGL|-h6jr38JnotKeC}$pFLt}`em`-dZi(>Ds9j{yB@v1&dW#iRg?WzS zFAi<#L9L%y?g#iIn_W6ZG#XOgQc2?TLU2Dt33s_#`D-(?Ps_*|7|9nz-K~Vvf4t0@ zvZPqR_L1HOzU?N_q7Xbf^WVmLtHWk$ z(d+K6s0OfGILeyPl?+VuDUs)w5--o#x`(0q)QffIjOokvcz$g7C{ZpA*HprQ<9ZtBTGj~oCdgNmA3Je$O#m^*w ztc6YOz_l;Qe1f+0wk)<^@yb>U9d~DzF?&I?l@;;BY~b%t`h}4=9yApKct%TK%cG@9 zT~BVw1EoHeG-5~Q-aIz;O;(?27n&cV!w%+#hRbzH)jVAxAy>U!_xWeXzPciK1dkCk zTVJ3P4`@RFhpS-y53T|uW%Uybsrh;WY=H&zzxwwW&Hpm9tpBTjkJ;SA0^St-PygPn z89Wx6gO!zqhm|b_sRbPW{{<{9;L)HgDLwykIBalCl9o=toQc^uS>Tv{nA({;TM)Cd zuyXttuu-XH{cDXI>vO7lX4r(+!*ug7P2QZiIUYmQUpHM|h=~gLHaY<@Xan^x?pkiTBYj@g$#sKq37^NCvRB>^7gC3OG@P1)p z8B#XI->gtXs9`00Wxbe2K`H}1)R6FrJt$q{27t^A%qdJyi+CnkQ4e#l3$Oaf4~fKp zTo}hb$UR&~+LLQOi{Kiu#Ipe?mAq(Tk4YH}9XIdnCx=i{IVB!PtRKp8MdsNuS8zD^D zzXP=4nt>Lef!Hu0vJ#O2*9x;zTwC4tJG)nw;n&iFG|_V}7RhoE-b*!Cn+G_8I0#Hr zOx4?n@@xIn(WK87nfz0zZr0kB@Amf(AfnDa5X>elV(EAGh#Q$PbJB*txvmXaKT^$N z6b{QlS+fZgW$;?`$oFj)GWnxboP~QXa{#IAYT*v3E}HH^oDO_Jih2?490Bj}^_EcR zXSnShi0$2WE!Nk#Z~OPTn$hM*UhYBPU%0b-zOu8?y<=1|KXu8C-~EVg2N8t!sNV0r zUK8aVLWZAl6s}Efyi=x*R`(%}xw8x0!;44hvozOT59k%`wB$k8^H)xt-dYTqC*ey4NT}p7|+AO zlB0y(U0Zy{KnJbcJ+u+uCZRA987#NXjr^QmN4uSVW6aymrUA;RT&p-q7BcocK$^=|VIDABrSI`(8*WfDtdn*4mK_?{#t z0}%#d+8L-yR9!vAJ1kGzsYFXUCXO%;)TGk~0+u!b-IeSNbifpgpudfk zm>`4cS>O*b?@pTf>wdA23!r<>itjZVSWw+q3iaK~{zo>xx*96?*K755jJU|aOOKtH zKc09Pe&J3N@*r7TnJ`#hBT=l}8r6F1gABj3P+?OsMXJYkm6LnNH(NBL)SsQu6i{Rg zm4j;Ep%*g_a^OJ#b+mj(0|HLV7E{(#O_XLC>)2yno3sAJ9h%90&aF6$tju8ARI%*( zVnsF|GQFW|iJy%{X@{JZf)}WewmDkBQ|W;5z4q8tb$#^U zvKOnMEdfVJ1M{)J&8A=5n|nzM(X^6eK75%4h@a)GOaSJz?9 zwvicc_tQrn4tJi>){<9$!aD1;@I$!WBfZda)%jb37q?r~tyWv*R|yB+;EhS*wt2$$ z83r$RDgS=zto&+mJIjB(F$YBXy&_Pup`TJ+_#mheK4X91Ax^VmU=R7pThak%n z9p`y$rqLxe$xT{OSAcx+)rb^59H-u3Q2FO1f}@2C7OBycO>9f;rg~YJXO1p^qks}G z)X)S~iWBN@e@+}$squ2VqkwAXp6p%6Cml`7*xZQrzOHm>sbh6kg-NQJm)C5C>8Wi3 z*XZ0k4l6at!ChA@@A%21U}*~rc*J)atuFByy6XD3JECh$k){c*m@G7`O!cyn(GCW%rv373Fmr zrD}Z~0A6M{xLTeDX-vv)jDm72K-s;qR7tr&KI|+i5jfCMgsaAMkrDCy)kx#O?wQfE zcm?7jW9X@o4sz4R6t?}ZT)$ccZZ6{{%162zvhZ?Mt9rAxegE~Bv;)1ek#Vm?`+lBt6d0s%lb>T=)*@Q=%n+2C9D5|`e*96 zI{213aAalbtI}*XUGMcvtELjY6_K#@oBxIfO4zL-Rd?r5+xl7TTM~Thl;4ILl;siO zYj?$++>)_5Nz)wJ-7@lHq!DKmp%`t>WjLxI1La2%v32yU^QQgZ0a)_&==HH*C+Y=m5r0TZ0g^4sCJ`2 zkcgOfjgbEQLN9mH@MM>FG;UPzv>kQkDL~NLtzG{o3O3q)d|>d{NVIQxP5rH$f^Ob) zotA?OKz3G87PNdZIZUuG&uw$G=&lJl;kdYGripj_Tye-cD8%H^2xC-WJ(`#I&E6aL z@T{dCS|!g~A5|FD)bm@63@VzN-w34JALZF7T|S*m z1wA%R*5eb>O0kXfKRpcwC{GGhFE|On!>k9#q+;o5O3cE;49BEFtiwXgM$GarscdiW zOw7W~Ma)dB496sGXJ-E|;{Lx#oLGmSnOlU5Q-Xz!g_TQ!O@vj9nTtbAOhTMXl$DEF zjD?d$kof;!g7|Nv|7(5NdDs$WP-p;xIRTp}L+qiOH&2D`ZL9LK%6VRk#xe$pQgvB1 zA0Y9yNHDy2zycjSaFnv|k(?mXh?pgw9SRjTADhC6C9HDlOWU_~>$q$C*k{V)@^^{m z80oDKGvXUro>S=$OjR+&1<@9s0xx(GG*#NX*%%y*oV*`af$EyD4>MqLig`fww=#@X zB6~$d4pqJ!^kflv|HNe)tFD$+YnybYP$-FyNcgB7``9_F9E!zWxH5cddv$h!M?C#~ zix*l8wT4999jb$i9!3wPS-}&q<8hxZjsh1&WelO#v&P05aJ}%kUs!F10yaVV63g~j zoWewJ9DYKru&2^Q~i)FC8?fgSNh5mLe6a!?lZgW#qN&Fr{Q}SKE>tX)mK?nG@pRruQtX;Z|C$B zNyNb*%y%ZihVnCDph~fbZJ&qqWQszu!ck%dEYs)9%3(nG@9?sOx i|Kf^Wt<> zpX%qS^XZ&n)SxNUgdd>n=>vuk)QO4|fb>rqRXJ6*ywAs1yq$To8W;cdSO`u$VWHkY zkaN=}X>1@M@hksZXt&$@k#Py$ph|wr-^HzM-luSKk{-@ZBTpkX`|-eNLXWHKcru04 zXQ%70i!Ifc>t_};v3?Uf??!8n)u7+aXzw)_p!@ZHGA5x@wVg3SOP5+{G50(Kl;xhZ zntbJdT0Sf-BMC6Ph$>5GGT3Th>!60!Xh7m>`IOqH{tG0IaCs8&td_rP`#v!1dRur- zM#eV{XrNX%$fB)T%bdKoH&HKyH>?Y<6*ObRvB@FE>FsM&Dw9_ckNAme-%xK%9a=z_ zd&(WHuB2gb8QWyvSy7;hdX+~9I6RtKsHd))y7zeGH2qyMNGUeyQ+1xO+DB*|MOW;yv_+6*0~gWPr+IUYQyTwoWk-x#F88)sV2r=b2a5K?p=~ z)Y{rLk@M1ncGOR1IKlj+iUe1~43ouP$3jbyXc=$f6@IsjtPLcoRmgIWlH*#`raDxm zDi;8^;iiu!8s%G?B>Na5pl*VuE3?WAB#G^a7Q*`{puO5n`FCCBh0WuSZ*mM62~Q+Z zP>}vlY%-H2|9G)%Ma7~Zm`2+CIE|y_5&4N^xng6TY1e=IDEU4EZr#FVvUu37)X}`2`_!3ktzPh=BgwY$vmQGXKhmJ!38<) zV9aO5F1QlLrJ%BCv-4n3^-wJg>8lX@iS&LvZussnt<xoz!^nh8HKvDgoB>rZ=MCh68;2(Lvl4b@#uX{Fs!ZS+=QM7bBI{^yG+}n(fYyNEE1b z3L{fclH@P$iwmz|FPDHJimBz;P+!ed6Fg?XaOe4LA+HjqC`fIMMVt;c(UT@eHxK8s&}FYP%XgV zXre&pnN_GK>~mw=`&F{jETUeveK!_ZOl-a06^HQxmSmq$+F8^Qt#oaSr8mt6W`3{; zjZpAzsTPZHWYU^p^qmtx6xU|WQ@i90Vx>f?baY@gPk53n2|zU5ggT5diviqE-7p{y zAM0b-5`4uEjxMD|_fo1cqi6#5HjSUaS@#fJcB9g5pJTb@L*aEJY2v1K z7bRslLSgp-6ZI+AP!vpAJ@i5u=j&-B97hU{C*vrA;QBDm%UnFaNEVqI^yRwt9JBCd zIvglub3m1dZe^jD6hd&s|vzf&nQKno(Ob*_?zp^ANUm&} zUe`YZsIpZY`!nn3+-Nm;2G#w=BQ^;nZ8X=_dF3|wD8rVghn%%W`pn{A>*4S0D#>4N zA~V^i$bK9a{4VvH^ropQ9`+n!86W=S403^kOxhqS-?$l{;g#I?vde>0EWaY*mY*u; zO@mSoNOMI@iOH4r1D>fd9J89ti!P_-aEXV&h>`V_IHRuan5y-!-YZuuuq_W(er^Xy z{<^p&fV_~e57;Kg521S{8OmjyVXzCHS`k^)sEeJ~EGU$kTWPVQRPd)NCe>*=@*{^7 z+D^j0!pd?G1*0IJ*MhmO$6Q5kkCLt^2FBFh(dUgzx0j`c0R3tc9Yea#9jJ!3FAaJw zJ*o`P5Z7&ikwgTKi93*_FqOXe83n){?xjf1l@{OObW~exSUj2cN&8La$GCiDGitUg zw9}+Hm;C-X9Bx3{AvOJ+f}$F-luDx0n~{^5m|iTnzNXk=zH{ zsay-q^NaONz>Q;r*QV?>Gl(%Ys?qs~WaQ@(`qN1^;y+ml-w6VN$=J&f{S0w)gt2%I z?`U4luWkoFEdOG%bywZ z#zuRt$qSm|bxN>8ICkCxEY=Y(gHweGzgBs?N!Vv~N7?@E)9{Tb=AjVVlA=w1?_{*4 z<|;t@qVt7^I5XnkQj_r#mT|1L8zBesy(jj<{FH+euk;tu&DJ5N^Gw+gbr$1x>fI5c z06C&7pu-#b9gUvkNLJ7qQf{ynn>@HQ>;|iPDtSJb(}Ev<#HOYl<_dE`*6Rv^y$62^ zB^LZ6C4#Jk=j}2w(8K?5fc*{AOLy%F8yZ7DxbHkkrJ|A2h2g77=q;(x#pgGZgUC+a zF?P(Uyp{95%*fhMpddVKb?NcXbQ=7wLqebcFb~&DPlWEf%W9{X>kMK-2&yUva8J8Yh2{BHM^ycxqI7nwad8|A%}th?HcZPAsDtdK zHnI*#4BOM5{pHaPrwr170{$(R?DMt$dO;XcrS2iqGX@L7y9Q@tp|VmX{ezJ6YL!+9 zIE*|sk?@@xzbpQ(Wj3@&I9(Z|@-SF?j3JVexI07_9K}|8!ANfR`?@ORjb!9T68O7D zH_QIwe;@H}#B@G_+SaoL&g1gva<*H*`PN_~W(X=*r|L;kkgoA~3|T+Kk(f6-J9D3& zjr|SoJQ^7RXW?k>=I&}?;_$y#PNsOaaI738%p^{xws8FXa7^+Rj#lp0By22ftpB%Q z18Q~kejm1?2JRY}9LY$o>HedX)hR)a<&?>W1S^5I4z`CsTW@Blp~p??wo(EDthmK= z7AFNaN|exOT)7EmkFyJtR!mZB0=-jj?{|i(xpv{ZCJ*^j&o4~+&|Jv&)f&<03f1kG z>{IdZ#;^b<*^1EzO@BDS*ZM95?YS2a6HfIRWaK6_lQbAbpB9TPPA}LtJ3mf>Y2Z3ug3EV zTRJFQidI)3Dd;_oq1ziXV&{%%j%(0L)X*n>pYkT7!f76>>^yvEsvC$oeKwq(U#okc z3#d&Iw@$vOI~%((wh{K+Ogj3V1H5g;lnAIQQ-``<)&^?>`}eBX-U>(D(f9U@i4?y? zprei*+;}-!u!|BZD9Hv>>s3}Ca^MvDZ477UlcnlD)erU~U;&rE@=iC26Cw#_cpA@m ze)j6_567^_H&M50u#sned}9U$)OC-8z2s5zLk5p8arRW4g)p%dyb51bKsxCN5S-GbYM8iow{iJ4 zXYXW5Ry_5WmRB8R>}^uWwQP{JybsjXhM$!hgL<$OYn991(1dJnM#4f1wNhxzwBuMD z-DGuuBa{`7EUP%HeJIa=DTugsyYxJ6Ete!74ZX`}5{NspI1kmzI@ewnUrUY5$|^5% znwow-jIg|S)N4TDs+A=*ouEN#acIDIb zcBFE5`0})s)O8C1&)j%k4bT1+NvG&fWBqH)vl~8!_`0f6`wG53zBuV7efj)C2rj3 zVPX${k&-q7z}c-5@kP*ZD}B{nUp%jX7H2&oAwl5&Og=S|2!%vKVkE-WpbCyCNLrz) zx${TR$xG-n{IjBAs0LMKW@7*yQC6upw+Eh9a6$Pj8b>L@fW$dHM7o|&n4$!IzlLI& zX@9qwaShw%iHq$K@y-~Ir!}I>jv|d)f1F8UvYK{pB149=6GJw;=s5P z52Ae8Z`)}%xZluHY+`_{IjYtj)8ACUqQ~BnSBZHY|A&##q?5uxC97%O z%h5R^L1gB1_X>DySb`IuoL8d`bp=M^4#iAdAPz8hmY=P>__+%ChjB@P+B^ zznXAQEkq&P^>q9Jsv&wTLytaN{--10(4`gw$av_V6;wf^*hrrM;VLM$MY326?Lr(v zy%39saKzk64ddV7lX*Ggz_V$dlP@WZmm3?-f~Sm-=eC!Qk%^Dmt*cxI>Ari7r-#Yo z!sy4G2rr7-6&=f{98*C0n+}rRru%!|RXRo>(ReVZcifO9$p1+t8;{xx${A81qLw-h zNO-JSTAab`W5BJJ1sy)J9MBd+gVtIbdJ9PgX-0Aa!z% zgEE%iqke>AOaDN!k#}a5b-t4cVOu}|QKyBp5z{UZL57z%4pZoPGe4P)xivTdLMo?2 zzhHXWsoWpB%Y}lWL4P%fqwFqo(4akCqoIZVRJvSXPKKE_O`j~oONLH>$BBIwR%hzW zRe*EmgTioNJ?F|loEc;*(Hh;P(Lk1B=h5*_s4_(py{7AJOyGTafvkuk#ae0W-x1@# zm9V~oQ~Z9+{RI7rXIqa7DUrqn-u)Vn^|IAF;#)AMvc(U(BJX~@G`aJ(7~X23px=k* z_*~#04x2wGoyedigz9D0broL>nKH|v5xa4>Jf*_Ysv_?JV592vAXPu75AEWt2fdI+ zx!c!aKmJk@A?1*h;29&|ZEzhfd&2${p#@(SNf*w2&G`=YkVPKm)e4pbe34;m-L_sf zU5_>c$JcI1*_Z`wgm*<_j+$&U^-u`0S}pQO73O!*J=9F2b^@u^F&Q;0!Z*Y6E2D1w zeE;rmnoI5EuZ~6>l$4J__#S{jscN*_ckJ-W@486;>p07zi@4XFw zFI*QDIgVj0^av0B&BI^=K<%B3e{HR?X)9y?BBJZGrNK<3{pa%4R#?Ipe!#A1@?wpm z6_MKa4r5^w9>O68`l-vg+(a9CN>^env@{a`v-8e@bJ1b@_A z;@K(s2YM1v3Bha$KsAISe39p^;P~u7i32cg0i*Sm#?GyBRsQI<0L#lubwKGVe`sA zTXc?49Sp(pyalBLGWg&wwg<4k0R*milG_BvcRolTho(q(xH zgx}RDrumt`4VjvY=_~W-H>)uzWY`50NjvksGs3=|t`@8bq8wYmCCQC{Z^S(9^p^_+ z@v2~GJ*WuGiNT^1lFsmbLqUIvh$fd~J&OUV>Q9tzq4%^@3u0BfZpd=LxI2s8y&oHp z61kC3hR<}L>HUt#>WvQ<$&~Z2u@FQfSAnUyeOv^*_Tklr4Etk8%wR(hvPDyk5l>f; zen@;j{?*&alFl8x8YoXJBpWD&z!ijuPbN4S&A&UA>Uu1&XCZ zX+)Rnl8~c=FYV3UZQ?N|g~=x5jx{dC2!~0PTp#O)HW*b8>JZkxo6Hq=tLO9UK)!gJDaghWF=jq*k@j=QuH;NYi3|KzsU?@AufG zw6A=0;2B&?fnB63iEPWJH8_j6yBy!om&>8*Z_t4t-v0lu1+j%=<4#t_!T{r7`F||O zHQrYIR{LW>ujntltBk2?X@`V#(`1QQ`vp&EZHx)5`Sr%yIQx#IoXm$yKO~unBKj@1 zWSeX-AGDFnj$LWIS~9#(cZttuYc%l^3qbe2I7K z?k_XvX&I|E7Me%4KN+64eDu_}E_WS)?}tylvf`8ySS4qCSygbU`KS%TAF*gAQHjj$ z=bT40zr1c)Tny*#4fKW-Tbr!6xthMoW`FB^BpA$TcW;O^O&fT%JS=UWf*-r*uE0NY zF*TLZ2r84F;ts+TPsiEQ`kjz5O+3oOp?~{2s9)`OFikGD++BBXI{o&+Yx9W%a3p^! z3-(W_MpF%!Op~tEp*oXh@`)az{{0Tjbqct|dvjYe#F+6%eqj$>;oM!**=W(ru_@Pj zloBGhdou{KE|m~2|CenaPiWWYuxiQY7pP(H5^(v(w}-_;!p;|snoNP2r;u%Gw~?EA zY^9w;2_gcHf?e&<>rCiy4808vMCJaj+MLzwN0We0;(Oh*Bq5$ArW1ZZOAQ+fe1c!3 zCauJR!j?Au8Dsf#d>P(i-zs{)6*>$8I!qP!?+4xF z2ccj|upL>Yq5;MN6JhR0YnuA7;B^YsVPKn=(SenYGM^8L)5i@wWJFvaXr>wobDei) zdeLk|BVuC17*bNSY~ov1yJ{s%s`t*_A1_w~FQ9TmS9Z&`-QXo02@ZV<`P-qes&hG7 z70|WSrm%M1wz;xm;Hi}&v9%wfmhPA)WToNX7-NRGx-L$tNh_hAI%G>SNfv#H5MO10 zZ@kf|c_OILV4!1P_<61ioO9DX^U#(b{L^b@Of||BGS4>1+}+DZ+UNP9qZ?!%o**C( zlST#$(}hWo3$seoM2|of5sB~DPM`w1jf!7TPX0v$c9EjIlPT5}3!`mDZx_b)=cUID z2?XZsE4KFC^6wJcK6|(~oT;LQ^V1l7d+0;V0@@TKRv(`?H+SzWpyh!?QNgdSxwyY| z*jkyjDyHkeIg44;tISy@Uj%EGH*1I>D(n3-+cx59byMrT>k0{n89X?3(S{izl`5zg zSA=f@wpxd9{T#=>6z0Bc2u!*b2*%=;JngPcXL0e1kU$9gS%`t`>DP;bynm#QB9Ux- z%6Iv3*SJ}S2i@2L2;V+B3yD@hltNrt*5A`74-G3sST%nPa{3S0V!u@!$G1jt&5{_tE^BM8PG-9g;V8Xi>)g1;JJTz!sT!Z^_ z`K(47-d=`KEKd_!TmF3RKX&ywW5!0i6sg^IG0Q(gvfl?{_r@RjQ<&DT+(#z zBSp=BZz&6B4HnA|(?C|atQg~kex()03VON@E1eoW)30N*F=Io;3F^^e*}er4>gtdK z71TtIw+H17h+kN0CwBU}`9026|M_Dl*-s}KGjOFYem}GFzH8kaB=R5w*4o#m#)Zm^ z`CP1m`|0|qmH!u^@cg+bV!TAiZItI%pIiQi2^Z;V0gGrkH2)a}3y5H59(e>M&Knwe z_qDUSF-`wqg!-8Y3@$VsBK#9F4Wz2l!y)yLpl1|XAg@flZ;hN9Q&wD}1*P&7nmah)W`XW+wa<)m;I50rKwnxZE|F_WS`uClYA*Zwg(M-}P|#zHwgMI>K2 zf~9GpC|eFip~8x~RqWBfZkee1mgPPq)W7+l3Pv=9g>DM&dLe-LX{M`W8=BT)C0?x7 zB!RpZh>MWNm|_3N4CV0?7yMBOy-NR3O4v?-)(sm96=81cC?L)+*9 z+2 z2)21fD>5*UBlz;Z*VCWo>ovx@PUb*9NX%Bjj8c1od|VKm3wrPw+9VU8Lj^DP|02f+4`m1{+PPo>0CZGmbVs z4VB#6g@MQa#?9G~+?=>}zJ8mE$NM%j&m-owc;zM&hW2Y*;f0Mg*h%oZZ{h-;0^rE}Ct3%l!^8ORuV#4ZW(49gbPqTutz|7=d9*2Ve>gIPz z(_ed-aS&=vjy+S2jLvjobEPWIe6u8YDTb)dip_Cz@!6pFFi#v zrx#%kv;5tH?!5uFGS!qNkiGI|pg@o0@%T}5YPWAX zeW>Lekbk|sgBdND{xjtru(C~wbd!jdDVZ9&79i8iy9n<&lR&?zMhvaN_ggb4w!K!v zwBC$`XX|WdemVdXoU%J{>?_Po4U7iz7TLr7D1xMJwtQM9@cSlx;4@M>klgVU{^WHS zU!bB#l(2h5eR_YnKqdn*75}o_(p}_j$c#95QpJSrp-e*s_$E9BA!1;#s={GD@4(?@N=V^ zyFNol2_di%z%u9lwoQM?g)VBU&+U91db`2)iUGvIW!G&Q9Tnc*jjemJ8zQx&U*g}zu z$PQi#;8`pGqSc)v&)=lggcZbk`p6^ZZ((JbMlCChdfYjb)(3szr zTx!p5y6nyA{@eR4Y+}(&#h(_-?~C%x9;=H>HO`fP_^_vR1jRX#-tIg;lK zlrw?v!Bc9$rWf~=RDunc09OpBOoTv{^1L{;Pu|ee?mp#mf|!9YFrU?gz^IL%($>sK z_ZLwj+8rIG8ia6B?EvK0KO*j?fekQu!ZI_G@2{&qr7O;2%|K?mLwdyAYv#G5Cx=St zTaIjQltHPuVEWky=US?%v@F?dXH|>dm0!J$ZwoHK3CivI4?D3xn91gA7^#W#;!xA5 z7^A9mMBMPlAfFXTV%WIG)WS3Ux`!S+$|b`-iEqAt^!X?70+Xl4)<@9?&xzvXVXauv z6mem!+OUbI!47%odmbBEDR7RY_;DLvr3W_}qMlt=+NH~H!-o)lk2Lp*qYivRZ7cDa+=yck;}gj#Wrp*%|5Vt-7Gh+g7|i`i;%~@xAmolI(q(?B3w)nI!gzq`E+SW5+#{kPK| zCyktLXqD=%9e-mzD#hEY>~bv8Rl&V$VY?}*)SZ}o)a8CTf3|VmP+$KOgB2tAYdkkm zvthqWpEWa@Oj+X=owA>JXW5qJaA|oAi+hf0s1hHnJ{T?%#wN5kXX3G1j3KRtY4s5V z*~;#mKFEt9WBZW|L>6`eeK4TCCgB1r(DL4vh|y6LJseS~iHv`sRKX!6S+2u_n@Q0{ zQL%$BUjn=P*rj4vx>^@DlRxR$@W+GnXhVd1hWZ*2@6#KRtAdRS2Uj;dUiCc5pqnmHOb3{8h14n_)bEwVP@#~uOTw{Ih|oU_!A&87;xMEgi}f@I z`?Ni9L1|<%`a5twD^<_58m4WPJ+id?dD*(Z0JrqlfUIKO{xhn5`9eO$qhwdTC0_tVDiOc!81OQ_$4Smv(7%wVV_NaO@S)(X14dJWPZCR~Re2rkza6 z`<{THH(0{)P1}m)b%a)vgLd_V4f!qHN`+fa*ji6C;-grX37(1|PDGhjNajJ8FUMt1 z875MM@GSJ=egsKKs3mv_+fAIt7H6##+GvA>c!D>CMEgFbbqukWOG-Kwzph5xaqAa* zX&DUQ#S5Dq?2};{crLZ;J;j{X3Z<9%n&{vGzqsAMxicvOuDCz+tb!FAO8r+>?xoU> z`gT?s$|M#ag0;N}1$sDH*6SbLzIWEMs4A1|LMyRkf$op`JD$f+)cW%B;2s{#tC83h z^n(;`8bgSyB}` zn~zq+yb_ovy5UE&t-W~PuQsi(CrkKd#*kz(=3OoaJFHLKB48%e82!avyWM|jaSD8=+@<9 zk?tU20StVExeT`Efu_FJenTq8$M^O=P-SmP(x)cEPU;7mHu>7!I?0Gl2yug@X}DJ; zVVC%l?wSv6U%k3^?%cC=m7_eSMkonj?){KU56u6^NIfQEld0~jzDsbSOP!mPd9B$G zYDw{v>WJME&F~D>Fhw{a)y5ZnIKSxO}N2~(bpayXqC9arQe71SR}sGPiPas-dniQrIzW?3pB0?qd2 zApVo+kE6jHM1-gg3{WvaRNsco+Ql(>nsA5~1=)IaT-qQZ^G2Snw!f@E-giDD<-GpK z@!P>^EVVuy8Z!wW91~ROq`l9-Ph@Y_NUK$U(gVf6gvYL;kZ6->q_GoT*x$GKOXsaY z3S1D*$Zu$>k{kN4JY@_~+5SNg9T-cLhA3YgV84HEVQ0P}l8Y=?SMIvvHyfj8c;Yv+ zTLnhx^zLymYbjM6KylIndtON(R@F$R!g!NKp317!Ov~CnA|lKOay~4o&!tG?^}_OZ z=mg4a;`*NpUj61H+#0)}OSM~OS!Qso++U^e{F`~F=TLQxt#usdIcXN!OCSKh$EK5O zg?Om|+diIELZtT~KRd6u`&nr{4}$1}4at$N^A!vfDzxa;22L123P1@`A4ct4g`l*W z!1)J-(6?}7hN*M1@`I5c3Xcc~bFSQHs>2NxD1zC15TJ0ksRtU@ul!t(yR&I2c)iDpjL zfu+bON@V>kVvUT%dML#qc*Nha7~Ot#XuNCOZhagA-P+F~vHW#wbJfi3$QN`!c#@D1 zSt2t;Efzvj3LW!UKAtK>evZU)RyZZSA5R-z>$oP_EP%e&H`tRKo%0es@ON96bwVfO z?kpkvZ;}jf!Rs}#oL>g!@6k{V=Cv=)_awki@RmRJu@;1MCG-1hn*6J}gWnX{FCxw% zs=-}%*L5U3j$_LKv)!O=X6(+ar=rcnV}2Jw%9JEz8+KU`e;UUe+>zB90+FveETey5 zGl|8D$MeAM*G1j#>i$x1GfxQzjw5oZLOc$9TQm%Szh1BvPCg=3p7r|m!?T>%Ex-W` zRLzF4P1%=9@Gz3Ju$nMv#iNM4e(+aznZYRKeaf&K$*Sy@0!f3E^1!b7{#=cFq{KmR z@%n}(dATU`m|bnJL09Viu4tEr^qLT*y2wA0Jgtk#ghbTnvjW6~+ftN--W*)Hwh8F) z`a~t*Y_u^&+LWj8FMNW_EPaU$vJJFg8Vk~VVKe1`VqH@av;4+nXR&Lm^A82j^KF;R zfk+Vj@k6Id7q7-c^M3A^T9AprZt@K2cq(_JL_~5pmhm*C1mz`^FK=@zy zDTz7Q{*NR^`hU3{79N)W&Fx(4>N#z5V*RIyQM!?F!BNR$iWnIY9vWp5i>w=>8}R#G zpqqzr-HP>obUr<-Ju8>6CrO(5G1@e(_x|K2xYO@-TWKJ)v*B5Nri%9o^z*s9Kt!s6 zk?6hDONWOihmO3>HZy_uBZaHCdu119EL^p)G?@9rIBS0Z)%4ExvIp1_&WhJPYbU|b zz}++K6B8aBpCGVhhcLwEl)mXfb$L@(v+<)NTdk!Fb^9JB^|DK%iIRzbfKWs$3kxtz zFrA35cNuq4yZS#Ca5pldb3YLW*jJ~8@kT+pR~+{wwCRfH-Tge3)zML6w*T?KEP;e; zqM=y1YD3rPIOpxG?ZX z%~}mv1<$@94)eepEE~y{xEm=D*TB7O+8kYUKNbvJpM`lb^*y`P3^;{tolr9!ZfG^#Yp2MA{6iQ_ieCM!yz&AjxX7N^6DcgoUj$VLGm& z3Y;Qaj9ltc@p@Pk=HIQNRQu>3MH^CeqxI#g@S36JCBV$!a;7Q__LneW`w%S*20F~X zf~$qyvWW;i~~|>`hUj+sYD(wmm|x>2)SzN{%)>f_I_DFlI)KFjl)e zYI500GFch@|Aq=Kp>rc^E>gM9aZ|bo+*~m~@fr!O&KN%C>`SM6?jAiOBcoU~ z%{BkKSzxXdBl9I-=DxK@ZCUj|A#3gjRAysn;olMpJd__sAxkWef7;FJ+?M5uo ziU7sngEqNeV8*$!hGN2SWE#~s{3~1}@*Qrzk~HNr%W6m*ydA>kqxG+NVd;P{cwWe; z@B8l#RPJRj?uX%)bG;s2=9@yjo>K{dH;r4 z;{*Dz)qMMlEf>B+eF(Xw8#}r?fd2P7L$u&%b+Pd_)QKavT{>^Bdw8yRC(e&Ut6g_0 zRs=!nu6o@%*}7Y6y2rdQaYnzvt9>Ta;NcjluaHME`iFal_35v+iZn887Qad^A3)cw z$JLY3^{{u!aio43c4n9G}|o)$PyO zkw!pXJ((ejn~HFXw>s!y?q8DIkAtm<@P-V7^kLWxSC*cRacq4u@Oo$%}XBlJYIW zfh7H!E_}_dxY4E!JXJHDQvhD^RDU|UWT zF8}^}WzO!`7HoW)v%naSfbCnoLQduFr~UxnwWeBP%{03f;F;Gr_d4zn@ojh*#LwPr zf1BU`Fy-#N|9CvXKn&kbu=_t>ZO;E6*p{0s-NhJ!7P!{hamr>!0p^T?JJZ$e^uyMPx9w@=9dVba=0?*S}s~oa~lR>~((i zZ%}Dw!h6F1#@jVlwhf;J6?{*%D4XM0ZY9aLGQ=a!b^BZ65)Z-b&Tx|FGoFNUkO-#E zcTFh6fG}lJ5>CUx1(*BW1CE%k zkb`^A6vssq5O$8Obm(G1%$2v4`&&G<98yYX+@vN(a@UK}#2jauLizY%4ih)ha0;s^ zs6}mBz@&9l8%c|EOeVCUy%KM90y5McqAk>ENmkY+B{G{Myh zh`r|6e=}{l;LDP)6D`X|h@>#?T81apT?|T0Sw+&i`^COwXfU^gqjg$)`#9Wj6UGRS zw97Q_nQ+^m@5oupZ*CoaSS3fKhRxLOUvc;QxG`J%PaF6o9WsM>Pjjs4aXH7*N~;fg zIqytpN~3$G%^Fl1sc3AM$Sb^)a%0phuye37Z8%&*hzwaJ9p)-UPP09ALn0D)R#)CF zFkO=Rx&sEyzxWS5#W9u<8O4m7&UAg|vtTJ%J9S7CcKhp8{;VSB&Zl!#u?|J^6I@m* zM$$9Iq;?Sc1m`Ei9F8_Q^AmqC{cxh^VCZ6KC2t*jn_sS2XZWBAN-;mxg66I$FmX_Y zit*mgt;`}q;K2%ID#|kXL4oM9O?$h$h%02Dw1+kK;ynK9a&Z6bOV-TKHPqMlUw6qc zCw{gKj$z0?bK#V-e;zE)OyGotH_s*>cDvy>+6vld3{xW&cod<=`xt7kRwS0%@nR`bo1!Q59$Dqz*IFPbNJEsj zlM37|ryl49u0~!}`&_$*z(V{}(qc{)?uMV?4d1__6&*xu@Z0A#iVV8z?wrQpmpJ&p zia2Pe@b5nP%*>d>7;&~1eM>U@GatS{YFE*?|3B54{O1|BX@na^H@yW46bp@ogXjO? z@T?@P>>NBC=>rU)7?7+S-0UpLDsWUlbEk#6zTW14SF{Tf&g;Liwihn8|5q`jOUxM= zZJVK8EQD=AbQt<-d>)P7h zsMMf~!q5fxT}uSU7m&&jomtl2JYYJY?34f~W?|DU*35WEENehh01Q!7vq z9T4^)UJwNj84i!2=B&oD7+!#3$k z__75d{PU{E3X;P9=M#AAhW?V#N?Ar(nb$}(xA<~H1knuQo8K6DQ`-98mwcQ>xaDej zR)@Dr_%uQUVb`pyi}#p*wo&{Qip0%tB2ycH6 zwDV;2{aRXqd-)9pq(KxIJ3zTRgMP^*fqhQdd<#xKmw0tu0#QB4gO5rIT|hXveo+6n zt(nEaomO2)Obl56pW)2`%(MSa|3q|VdvI|A#{4z)^aame{)H_V?!qnHdf`6IZ+*!@k z5Pmh%#L!O2%2e4g({{tp1*@R`hgU;WYALPt6@4$Grz-D(-_|>WYWwhj(BF`yx-(-# zZ*287d5XNS3Bq1J^K^vd^kjTXt+XPMAVcSWx9Do~W$1lm0-YV%Dn3nU5T$=6_l?k#!F*Ru-mVxkJ+Q$y6?6RcECY(^O?{V z$+i9-ECZYYK*>lIuDdZUc>T8e^{K=}!^Ow<1Kp2g20(rQisi3rue-na{sN-S!9d00 z2l2!zqv~p&up6eG7F#u-@Mz?zqm}Hn5X-B*3A(Fw#FTuJA7K8>xc)3Q_BNLZX>5Lc z@(*>6Tjz_<|D2hbUFqoe&#>^CL!VkY1-U9>xp^W zjRL^?YyJ@rzZ;GmIX%pdJUBfVE?o&l-S1dIHTvb>_x<-Nh91Q79uU?9VY?;1%?%sb zO}OsK3V#&ZzNw-A2HmyM`Gj`uMxLIWT7Y|*bG&H*$ijla!4 z5SME2W`=x$?q>3Sf%mGpd_nu)T;3tS^a5~o-$6I*+y8C7`+gi2{Wt#?@cj7pzXCq7 zKb&N$K6qlkd2O}7R?oh>WIn@J401(Ug&pkfpYb1rL@^Zfb87`2b^gr3u6n({&-LvF zeEfz=0f=3`$Fb1CecqT|yQ_aD#zkWHPrlYMYifEr_I*!sysv#8{ARldU;{q!zj7F`B8;^988|7_<#lpuVQz5#(|Vv1PsAXcS~6Wkm{rw>t2rYp7D$+) zn`b!An6Z_!CWo?)CycrGbO(fn>lJznjmna6)tf&*z^+yJ=Qe*H+ehwqgh|fJw6pKP z{(XB2!o+r1r0SICveHM1XQPBZT;GZ-d*6@Y56&^~c;k2Izv&)nvrA%*k{#@W+Ujm= z`}tZEJr2{=X&5ewuj%4SEwF_h z|Ly{pq(ouu`xgyHsm+t5bp0OlzR~zJD0`hKLda_(JrTXS2o+aCzVCJ)JeoZ~@ZrS- zQ+%%#120=+RAIG07W7+2z4x?`(vkmQZA$};^@-EZCE*_r%kMCB-K0)VR?l@$WcWs0 zTC7qILG#d;u`Sk}oMwPL;^(j#DFVhE7hIv@0u|-SN^lPPuuQ%{#c2CzC(o)B|A`I9 zQ!(|g!wsc|&0ZmTPTKHCg1fwbIw&{I%ps=Bt`rd**~6<|RSf&olfumGfm^82p!=uM z=am>vn_B~>KHC32_`fo1^Lptdq&}Fn7X2*|#oIj-p2=%QX5s+w-o932u-v)j1Rf5Z zpNTdF?wDrxyH_lv+FSeM{uos|Ej`jj?00C9zGa^*1xLTreu*Q07-K|+Kmp$%xHj%9 z%ZbQvF}k4f9hh+*5+BdGdYZO+y$OG=IP+kWVYjkKEcQvjA^%yGBPqec_gTwKrESearftZ#}4onDR;3O3d+|co`^<^UHAd}55 zwE8cu`IzqgUjQmV)xUKv;4Rd~$Ol~%-wxaEJ8hGP?9QZG2{UY^41c2tr%0x=S1>A+ zz!=c$$SX|7{S;q>r3vR;_`oczxF&$|gX^SG@;B`$dTx@plCno5JWp)lOWqgQw8nz9 zCtP`Ak~K%Yb1I%%*3XA*wevXjJ}FScQN7PBg*KP<9)woU3?FuG)M_wOb%zt9{xIAzTv{{zR(K4@TL;xJ z$D1QS>O9wXT~M+HDQJUi%f)n^R4gDs$jycK z`(7JFls(wf2Am{aE9=J_Yr3*W3-RY4H-^s1 zF^q$H&nh@i3^HGH!tlwHz6Uq+9*K`G^}mRy*DnY>pmQ?Omw5y4ld}-L@D82*cJgC3 ze4M!S<5uJg7~v4vEUT^R{aM}Y8RBBho=Mpg%a62YMS+SEeejl)D^Fks zw2-T`b|>`C_DSiat<6e8L=L{}&&%p*Rni)W86#UdAGI^%-;LGvI1-FR6B1+jp8xsV z6ViI=ynhrI8l!Y&d@FMinRv2L!=Argf6PfcV@sEZXOgYs<0ezxRsz4fT|ry_qozUI z%H`N?a{UPxROw!gT{K@P#^XDV!f#snnT+xj$Xh!%_w8PpIflJ|2*HUJ){B3oluu zE=Ph%2(wLD>-RzyR+U6NHC)p@3@4Be;e+W&=!9GLuh4ugCT~0Z<#R&9|55kjjgtm4 zoaQ69-=kIF4P``(jUayD&szYLAY71riCx^F)7z8V0N#Jz1+so$-@<*;2 z^MB?p%7=UOB8y3=zDYd~rEQu|6x!C!CoIp8XygcsBm3|igg3tpb*^aQoYNNpiHXdx z5H4yRzNr{fV~%(YymEJ^j_$mMmi<#%PF*mWyWa;e?rmfpd_8dakX8`KrK|+V)4--E zvu~|hZ${$RSyNl&_Y!Z2adm{}%UC_rQ-661J<}@Ob6htX3NqG`8?YYQ9-VQk9G8}- zDJ6;}ZzEAoICwKdG_l58y?CxRFup2U^@@C-BuB$u7C0Luo6zgXa_uaDV%4 zI@|N@nx=}nS$wK$NHs8j8l~ zw}#h&F@oSa@V=`KGzdiyWSO>}yMI+CYC0C5UpTBMEI1RpF|dB05uw{De)9=5WH$Cd zx0>>rLi@>BcG;C2o>}bpi>^3~6myX1IjnaF8)s6K;Tu{fgemJ5UH0^?`44*@gI-)H z_d1`6A_H?I9budmfFli(jy6K7Sx7~m@9!ixHFfJz8_ldU6zB`0@*z6vXn!dE50uWw zaO+qUvtr1a;2vJe4eMQ9xuVb&bIKR^l@uZ4`5fhX=k@Y|xM&UO%LnDEJYktZ@#tcF zjfo%j4+m{lb)NA0yhSHf1imu)m)yFX7;#+m&LSFTVl%Bz6<9fU2t)_cVdE};e<_rf zOt{F=7OGXfQO!VxWUXKCE`KCcK@&D(qeW3;!2X`3hM|O}T8hn^)k$E_4b{=U`F5YH zHI*%N3mp{#Kh!^q{PcqM%cu}lB-6bnkQC};Pj<>kjl9wwXR)Yax)I0=4aHHim4zw* zu*sHyNq2ybx8oDa)=BWAd?JCF{w<_nz)_Ftxji6Oa*TPof`6$zmWKemO+ks( z)rtHP0l^v9;$gYOJ?|b-1d(|4xyo!3n?}g_Si+}zp;M<;{R`m_N8Vi4Cw{>&P{%E~ zel-a)&z2o0-m5wKW$}v)d{0;vqwyo_hmouVu-@YAGld98bsR!nEBc$uB_qbtL)2Z! zd5_zD6`q~h#;67 z4bwRYzeDV8o2s5Yj_g?D#Y^#AMwOqf%XW5v84QcN1YXCXkTZLo0J9L@nf*)|&lBq3 z*ss&Gpwrz{i26>Q8L1wR$03phJ&pL`1hLCdh2~vOdhjBd`F~Dtn5AP>u-%N%W6h0k z5ids#chABiN#hQfEsmi(CR==pR;K5t1)(d4A>ywILgk2=Km8yK=wl15a@;C-8p%j4 z2Bce<+|<^god|hCc;(9%LwZe$0(JAH_;Ig(?U(XP9bmTR(DCd&f43}dL2Cl4=K5o) zY#_EyQ>Spd^M8#RVPdQ7b3j3D!L|0(e!;tjWb14fCn$vv+Vd{{y$2HCV_BY?mY!}n zwmAF1onV&!6n&t%$@nR%`iNljWGbQ4GbWA--}D=g<3eH#4g_97TKimA0Taf-0KOk8 z4`)v@-J(o@K^9=7I4uDtth{Sytjuaot-2RMOFH-$|9^sq&I^{IdyZ0MxA;uF%8Jh0 zqEd6+oX;0-wWbk^p)b6CfjBS`6l&Jfm;Mp;c=kyjOho}L_?Evvb$?v6*-jN9lT-bC z<=Hqv<^F+qA_4TBvK(cMBgrfzHA()d|tol z6pZ8|KYt6Ep`r{a|Ed7kE5xMoWY?LBQ7(!mXu%bIbnHMukncoO@Ym}x*vU3?+s{L( zq}OV1NsyQfLh@W1h3NF%s=3RhE*<}1YgOEO1JyMr+=_Y8 zeqY*J;I_@7-$u$WcQexN0<+f303Ot5`DJAdeg{F)QGM%OAMeZ|Mu;O55=-4wZi za%i<)4TD+2zyj984bL=PE*u*E@}qY$@e{wd>88>7Z=q1p;n>oXRh+6dDtKY)YHBBqD`9uC5H`(AVr!|1mp+1lHhwbn=qZBiTzLdVyxB8vD zk0+hZX+kG+%ab?X%xvJ92GCWNw&BY5^i`igbHWID2;aRfT^1FK!aHAoQ{O^V;9Coyc{Ay{Aw@YYHR+Vk%p8V50BOZ0jF!7f+xlGAN*&D%1*>u2yYo?=RJZX4}oVh9<{eh&Txj z-xVY#zRGXYHKt)&f`LDO?eBw#7Jqbkc1B=+N`>v|kVZd(CnG-Yx{1U*5AFIU16hCF zrdm(UYmQeGkDYS}R%h-&8e6w?DCB-t)|Iaqa3!p5o=9{aWjkjlA5-fbdD%g zu1sggd$d+nwqo?c>HbN|l%Rn`@Du39UYG@lfIsNoqhDk#WY`2uR0Bw_?ox~BG?o8X20>v#w zyG0$=mfuMDA0a*o>pM zRy7q@LW1usN02vuku%(;oEaEZAs5db-)uz2-(~N!blPIU!uW@PGw49!e6jC-N#VjGjfIi zP%D32qle4;`6ObWDA&jCbEH{ahvtBK9WEe#_ibFJ`YV%wV{QV($g_gg^Dk@JFnXM= zHp9UWW;3A!m|5WH-G7n44V2W*d2Ws{_`%IPAZFXrxbo;4S(BA8DMI@%`^+NWPH*92 zzv{{0Xn1qM_15-ndwnvE%w*a}T0KI9>om*tRdCagH=7Mw{JfgFz?XRIAI8}5$aPwn z6?g7?Y)@DxoAit}S0GzWBdi`lCgymLDO`|2lSN7(x{JRMw|_dvI&!UBtX(a7{FTEylPK7{jIKYQ?MHn)t>D z!T9y7n&(t>^&Iwnrc(tUHcWe(mPF84$|j(@+llfCB7YIeqDJ{MVCBMGh#%qbsqhwxY13Y=ydv!l)C#9as3T{gJ;?HosWGS0aghl zwv|^q>3;<5P=bDX@|v*_W6g{nOyYQ9APp=^O+BXj+Fh9UY01y1FtCU{ zha=NzFRXs)aLNM6!c8Tuyb=#-&W8>byJs zL-|^%5d*SmSpGMA08+68;@0~pQRERsf2_RTWprx6-H{_~ za{E;4z;PBEp%%9()yE<_N?b?7D|p{64)YbGP4WwI>+vNz)Q0b{;r%RR6PP-QOc z-kP)RdKW!Je5vY-5DiWlm%~u>lj zI<_PX2u9*DZgQnb#p+ZEo6ntqGJpS;JVDTCUd}hxJgvNf+4=!d{hAP!_4M}l7}sTv zDh5QFZDUyf%lKO1v;y-`{TJ0SSvs2RcrNJfDfr*)Q@{qTvSo;_FXYa#`zw2hBZ98nI!cc28~4EV?j9)uycwjdAb;P`mfX?Q ztS9;!;mQ~R{9_)=`kojo$xas)mGKojY0Z?UsfFSf;}Osx5zE`j?Pb@LM4{y{n>GF-mb(VHTBF&5xANOAL43C zg*cBqIUkAaF001@#@6!RMSHdCIyMnHOg`Va8K3AIvD3&N?}CCLc!VMDM2I+o8fmw8 zgT)+FW&&p6ln=!t>h5gD?uT3b&TV-Xsa5Xc3WLEojeS;ovt2SCpnu_RCcfMy*BZZ7 zUa_q?(hXd})Tqasl_v*XeUcnBRDt0=TK!lhFvJtWr{pBJSxJ>YrkUu8sD zK-#SEAYob+0P5TKXtQsG5CT&4qCulszrX^wpz z0YNFrtzKlWYb>OCRDb_5ix5{qQ}(*SWqMdFZ-_7fp_p(r49VXACB-OA_;Y>_nuYc^ zgVvLP5nGfVdMIYEJ*z;77w~1D3!M_Dq~%$fq^R(?c52|fgB_eO#nkTop<}N{U+SrL z4l_X8NTINGC6}@9>^i$^<>RTdciVMA8)ppZs5>J?pi!~Tt@DWQAk1IK|E0aO(+`$bb-IeF_OptR7YLw_%1?yRY!{-8uX5!26D=F}q#MQlM*c|}Ctdn8r;d(ntBmcJoXytCvGq$?Ynp;}?B!nGH2eEkf^(%&@4| zy=^MyTCPnP;;j5fIHT-+Ls=7ARF<9&m=#sFbE|(pI)BK!v;4~NR#AHS@_g7(lo{q{8g zn`5wJb$=Si7w6w>mP8LwOwQxVZ1Dzc%aAR}5$gLc!OWMN!|UlBVak_@+_|g$eUUFz zQAI$+R56~ieT^m^lS#MM)8V<^$n&_%3L?K$h@kXeKV)K(qU`cX=cX=XlAjvkAqb9m zHSo3|mRU;W5YBoJsba*;9~C%-`Ja5XHwxEB5Pzg!XNtg2zIJ^Q&WpeVEQ!>0uZ6Zd zneVh&+=a?{v>ivBsi-eSF7|>qjq|a+o-+wts*$k<6oTGfhSyuxQ$!5MM;%2iCIn|_ zaC1BWW-7W+d3~>{ft*l!dqxXoDMs)|2CrYUk(-UwcsH;%a)m5b=SbduOPTnrh@?HY zJbyQF4(wxh!xlD(f5-bhZA=;t4u_DG^6-4uvHBUxCL|%e!s)!X`ii3boHqYUBgM&Y zA{xGC5s3#oFhPV)8k%g64-l0;DI8m~Dk0oF>t~zAw#g{K5AZN0$#-K=sJqFPG7ox& z1%m=w?syLbeBBcDXwjQUV~>O(LDogu-hWwSdM&qWjXYrN-8RDenX+-&)$~4_qzx)* zu{7YyLn{QC4Lvz?@HFuf$)XN|7jy08Zew2#h}KFr-AvzTeE4o3X~Qy`kE-Wz?9wmv zA-#wEo54#`)H$SoMYG8bM=9b4v|p_Fk)xAjZ>SX6?`xclBfDWC|)L%+DJ``K-EniB?Vky~=c#pw2!zF6%5zQtYH*Z@*k=o<_iFv4 z+;w{fB)Ht7_#ahffxlF(j^pYu_g~`QdR&eQcgO z)fNar7!$8IbHZ1_yi@M+4<}8pRKrU!S9TDAG<#C_t6i}~@tKV%S9lngG=F-;2RJDq zg(qJ5Ui{cze-W?$M)R~cC3={0Yk7rPf;;e~hNbPkJHD6ubyLTzXQ%Z1U8rDurs-A!i< zoOhx<=GWL;8-duG6r8vpMt`&xZVhbXgK8rYviJ-Z>?~_Cu~Z2?S=3ePinl9HV24LSV|)7BLbx+M%InQ!WS_LujeL8Ci9V-dT;dh z%NyF(sk$bPi$X)!3eYs>PuKRSO}Kl#z3CD6sz zz?nvvjEbvKdp4&jp-lYHl7!o>Xt(^+d>tKQS;HKyD`S=>VJx`umXWG2+-d|!WaB7r zOQzbU&HGSaYCye^%=dj_a{{!Xtl^x#Yq$<`n3E?xPNNKtgbx_2X6Oq!u{aPZOKMLq+M4;0k;k}yS32i~Pn?|FtwhS^HEHn}b3GkVJj^SoJAK<(uO4S)9cqCQHz;<*$yt#8?H z8S>&q3)N*rh64eK4~FdQkM_`NRHc{G;K^!8I!%aXTCH9xEVAi%9*StTpZ||%+rL&b zGe50&;tp}l>ztV=T@lrbhJ;o}l&>k4uzjhI5{ag#Yp?#)k0)@1kqoXjJ-AqBAf5=h-NjsQT`(&ko7`Yw zS#C|Hr(4s2%Py+CJmVSjH#uV)CMUhFjgEJKvMGCt{} zMTEh{xt*lG^h8^#uyWauEtgtPX!Zwj1@N!zW`6h@zA=eg3h430C^3`sLrrw{4N%y? zm%>BXtsXHdC@rxr*0Q=ww%m2V4oMKZ_9YnekGT%&lu_Dk^irL6j0+~+Z2$B}iiR&a zd2JkPOn>GMaituU2~6Z6*lb1?Yda<#wjP2WnR7oGKLnKH{Q#)0+C3FX0va2wvLiIm z{2~1aeYs5#+}4MKDB;2f!}%2Xoj*o<*6dYS$ylsoVi|s%VtmmB)|NXOor}jR3*nl5 z28)!F66}FHmqvN}@uLMcJ@&Nd8s9xx6a0$4A%8?t!*UkaBY^Olt0cQE2;bJBMqnsQ zN`;moj^#MFmD5FEsBv-oT=M49RKX+U8x5)bSv7)R2)E*wIc|AYnJh3?dHs zz{e&pW@HuBbWEO5Qo#i%q?$-60i8>8vM{v_z-l*Emv)_Kd_i*{V*VDG4d7E_mQ zqs3Z{NnKi@odTXWs-Mq1_k~7lOaXN(3xCy%j8g`|z2?-BH{!~73-1{`Ufae6mNj-V z(8JrPm=+SzOD}&i#`O$8BsDz}cV9$#-Nyx5r$=jZD~D2^E!)+&8VHnz*U75(Cy$Eu zBw+82-E#N!GL>(sU$0)1Hu&nLW#+EJy3M*D-v@?)bRv7M*J5JJVsBteRomv{WV(Lq!iU!t#Sun}k*5dCC$U~aMcB;_pY~0(Wdbmo0CG#4HOcI2X5-D$Z8>X#~aIerv_qyct7Xu z)1qg}O`bp;l^6;lu?@}osao;KKA)X;Ig7kina*+xx0R7MTo?43ER`^caqK z(&@+=5%6_=MDiG4D{9LU|9^BnXjiS4!#4&hU-y=zl}{#I10OP`Fs#mh0&4GrdTJfc z&K;9m4X2OIkzc7)Ai5ZG7o?(IlPB&9=>;jPwfE&%6g9p@m`i~jJBHdOLz9@`9fl-o zc*(T8v|xUYVH3Mj&ZYBRNm4VSPofBRK8LFUB}?)3qAW>M~srX zD9spYB=9(&W!7q6jDKcujC;KfE`I)@&OCn1#__FH+9hc(dJvc;;M$A)V_0N9$_7vP zdOPLF+xq9G&7Ie(K7u`C9i|dHcvKGUx1F)V0tD@owL z*>;f&G1z~wD&ZZ(;xxj6}$ph#3t2W-a~o&9Pl1SjpT=pJIv z_5fR>KM@Z3x_@TG_>_O*gzL4RxS~tq+pb=Z!Nh#14uvh28Fm;3g<1ucb#4FZo;m$U zcd${md-fvwmH1H7`uYIm{+kI`H~%!M0q&e@YyN?RwUV|IC>L(fKr1)bE3x z)7h3_0kOfRh7K_h-59qZ(5p~b{8BSz?m$%>X|db}JKgCswlf|Oo18R|VW=_w@Oq?% zORld_{(qVw#XT{TPpgf`S@1_)rH73E*CJYr4CB*nA#|1eH}Mr~)Jf-ScPc5~yn^?p~LV*Tlxq+b-HG1mwAwnU`AytmF&3{Z)()-GIq3mzQP;u1%f z?tf*#95bqyJnXO=-NhT*zNLOzd%kkZg<}mzcO;~~Bfs*rfou!IW>lO@?@{aS`G(p( z>K?#PeCS@M2J5`tAgNP&NE9wm)gk-|w?~pYSDb$~1)^C}( zOuoF?L7dWcxXLq9_V{!hK9-F*=od6hnSoPVctwqG5at)h>3GlKsR z+!-<&hi&(Lvs9;TQa`Da=gSLO`hT~6`$~8~W%06OhAvEj3HZr63pA+Y(Z*+jWm!;8 zsOP5EGW;LkYMWR$Rj!BlXyT~sN6~1ARkRl5V7FUn{dg7@#jdoOy0_DZ4ru!%EpXSw z>l`Izg2|s|J^o1ga{kj})cd<{cLm26G}lcmPX$3(;WzmXdJGOdUI2x}k$>P3eIf6K z-8FWz3$|b?cGnCgvC=N4+29kYSEPs7sd4$Df7TC#&%d!Lu7Ip)@-{tE1z^x<%AMLB zrR67PMhpSyC4p%Y)kfbkYJ3kRm4_X{k6x$O6sHW%T!Lo#xQNGx(bX`Mr5O*cbD35V z`$8F7`5^d&)F?N&{pTWNYkx>QuxboHn7A2zmRXb4_P^Y~-e2yy6`>q*fL3Y$de>|7 zYh<_mmh;!rsED`dSOIhuqIv=4%~7J3D1x)U^o6odt~$!H1oVD*03v0=ernQZ9SN1? z4Y-ME0ApRl2Fr$P};kRIcf@v-Gwturr7O7v7fnXJ` z9$krmLE#VJGco}ptlT+`Llz)S715Cg3}idv9XeFM)%Y3?2D1TErMLS{jX)Y}?xYK9$#kt{StoEK3CO3fI^*wJfO|4S5gm zta!}uOx(>OKX{>zUbwW3Ed9MCZNI;Tu@bWGG9e_zZ%rq!+ve?#{G*I$t$?4FwVmy` z5C7NWa_L>kI4b_vJa;IQ}B6rpgFoq~TayuGp689os>^Gp_J8EUMNTN-`6j^1Egx zcdBlt|Ht#Xlz<_0jTyhr^3{5{`@?%h|Fbq#NYxuKbm4l=FH(&`uy zW^w>WbH_>fKxU)qPvgnwQ35&eHykYQu&m%Bnvl$G*3^Nrg?p5W2n3~1ysoCgM99lz zklGZG8z*~{6I*0sf|i2=A@SyTmO@YTuuVqi-#BkXv!67ZqMVO{5bo1@CgFG4>iUr8 zf{$X4X@l80Xy8cTuc^9>N=JVPv39c24&Eo>2Q@$Rd7g&}K+aBHg6Vg=d1O5Xl3j9v zfhIpZ^M9?`vwuf0gQ(7Hrb4?|KjfXYj`!m5guO0R2>5Mct67KTM!AjUE>kZlF+V*U zovMc>T6?NUOz>n2Dmi;tq?%Pw9#5NONW7LH9ojxle&mA{`-Yn&Y0uH zw1TR;y_jcfK*B0wvL5 zG$6B3DPrF+df0MykZDaDG5f%Uddo!Lo*r+!P2VTM0++MXsRXxJh*wJHXO|asBzVMxmoN!z)&-~)&6ezG$V0yo;2r@L2Rb{Os>#P`U$*Z=}jytBTCQ<`({ zMO?2`Zu~qn@{Qh#`RwUfR#Pi|*0c~}lxUkDE8QI?1-rvhVzm0d*jOi|-Lq?o-N z=()H`ki3q4NmOK;)|H5xZ?f(Hj-SEGUVquH{u->Pk^Ax)%}#vRASv-} zU-8^UD=SBkEwPymIrS|C;+w7^#leXfdFy)p3)#e-W4p)X7mgG*&{t~pBWFCYHoS6i z--I)?i)_Y;Q{uJSrpsLxV*-fb0OOYl1F9mA83a|rtL`L9yK^d+c2ZbDZ&rbyMjdsa z3vH6?&rR^6ri&s%{Gy)zSh70)G=K0y7Udy6EjeC=rsCFtP4nI}pK_N8e^;UvM9-2` zL`0EpwdZIY!uM=F?@yyeqb1f^@8$M~gB?PWp7lYRBv&}>y`NVy^wSO{@sCZf5%-*X z16OCE%-4(b!Tg`4cZajPe^E*3*K5V?=<61SZL0M9x*o~3dEbn4qfyTEn}4-crGM_- zeo;y5zadca(wN#6Vdl@EgJ|0^v-KWv7wfqUL zkt5c`)DS{B|3ny;`)PFjd5&hJ)auez0^+qZHAncD!9y=j>-R)EHJ%L8?$BDBfE>AT zm44Ty2#%*8G~j_IJlz`p*MItgLFhF~NZnH(^py}dkTvsS6kqQNvPeoz(QbY7V-0n4KgM== z!Z;g_sRdY{%`-b*YQtNv*S-*98$LS<<#Sf*a3f*T&gq)*^PLj?Fn?2tL}Si1G!dOt zti9QppwbI!F=ge$!_DVX{?6M+SY~|S3@uke*&O}y2GS@HehhyOGasT|_G*lUYj#(m zt0VnHHpL9$FwaiiO~!Rxb^B7!2{RR9GPOwIht82HcGVr%{3aARRFzl26bhESBzF-R z5k}+xY+SQOMMxiloke3t?i}4&0Fmif5bosrj`%HfJqucZE zk|8$pqq2o)k~rV-MM&^|HHcdMwy3~EJ<_eEi^Pf0Hh*-~j!E_#%n7ZhP-DIi#Pwer zw!eOh+#o`b1BzYk26l5q8Y9)600@kQ`j;m1ijoZEvUBe=tr1$YS zPZArlCQu4Niaf5B?J|p^4W(1_3_+TF~ zO3*9(G{y38q5`jxsBM>I{xeO`A4X~}%Y_xz5PzY$r8nC^TObzn(RI;L^E=~V4X@-4 zpVPU%wSr6Vi+dFf4XV+)uLbeKoU}JKV?SIAF;ByZNm)D>5l)EDf@~1%Cn^Wp``~ke_ z*nfrb2qUS(sNu7P)_vQdr`3X#wWT)_l*w}tUmHqDZF3fhx!8INl)5E_kN{!$Aw!g1 zT$v9fzLd09XF64$7|j3i(Rqlg3EH+(T{g5cE5_9h53FeTyEL>OBFsnGhL@m&^)M6? z|GSay=|aH5J?tDNEOuLa77OU!MM9ksa(^qA&wD$@Y=cCbRD^H>){v!c)g-%_Fl&Eh zkL`+;d%{tPAGuMGy)f3-=F?qsu@pH|A2OBK{$*>P+e*sxBNB7br;WI6D{Q&7rUG#7TC}Xjeq_c zKTOiYUSR8;|Cfq`x>ccphuC67+9c6V+&A+>xoeU$0t4SD!!P!Wu;xYHd7-drru{yp zG{Fvj!&-NHlo?H;OxWo>Bf#(25$NvG*U&^h#QG8Fw3U9T9>4U;jPa37`n|ZHbc87} zjgxyejBm@2imukoEz4rL3RQHesDH^Gp4KuuT(>l+IY3BCj5+(UJloU{zgBtga2q4Q zuYPi<(_MK|F{AbX7aUqc00CH@l@sWUd=p0Dd5IoXpt*6l;&IMu0-A!1eKeE6)f<2Le4qB#BW_uL~jh>kAtePca+POS}47h2Flg^PSe!GF8o?s5l_ z<|_wqui7-1$z&SQnFVUt0c5Rd=VT|#NVDvo4==)KcX+SzO~VCY^?EQ%QqNxT7|9l) zgdRfVmh22e3kqWPLTa#3#kDVic?=MaIXhs(Jtjqp8`#o3>o9XB8_U?rv+Xfhneu~{ zDp5qqMfHtDA(P9`3O_7piGO^yRC>^HcZ_7e*3TVqHOfqQR@i{15Qwb<$V_05d|2;^ zZt`RBTba(y?U|s~N0BW&zFb;Oxn7(c$q0~KjqDyjp) ztDXsAe{@8daWlmfnE!M&@N)bzK5~`lO@1w$xlo2YTwYx1h4)zNxPKW6)71mtjL+i7 zqzVKKk*Yxl4x81qUfQG7XXF&F9Ni;DrYSo4O7gleR%1}Tsz?DIMIzuRVA1xn*SKG5 zomlSA5oPR2%wIBKFncmSx!W6?`RrlwGYYoQmyyA6ugcw&J{aX-aMhv9t z;2kml8_!zo0|l3sD%vH3x%~>J>$DEoH1EYElThi)-uTsEZ(2cb2 zpvt6>@(V`A_Pillr;Dwbz^85=T%1UnF1;aXbcV-0_DS%X5PyqVnM9v~q`qVf>^s|5 z$!w5l!?#m?Gi?&QJ=g&Jcnc!g=?8YqrlU!$*W&{2?XxINThqLmyzLZ;2Il)dc;10nmtO3T@x zXI7PF8V=$U!`k;ikTsBCX8m*V1(qTftJMVA0wk~JVMS3v|Hr?S{)lvnUy?!pZUxc` zvo~#3(q6TTC4*|X)ZCUSiz_btMCWtjYaKi)I{ah8rhoFLatxe#xf9jCu(@yTL!>GK zZmhAosQ<^INy>UQ&!wsFY;X(BsYioBa<`)k$Jb>Sd_GeQdbe8G#guTtgXo-xf~};x zo*2&OZ67`)qgdLMP=dI|2$QJ7%j_WhLsZV4G{AxJmIeXXjBn$0ni5THAsFXKq0L+R znu=$x-hW>?S|>4r=)xvqwkT~jv1zvUXPRb(l*Bjm-Zp5qnM&H?!lOy_Pel}YZs)?5OLvy1hJB7Dft>+onhN7)q?lYcGrk_GcCVZxaZpA3*aAMG79q^yG? z@wj20VB)Lse`6wzl7g?{5LY79v;b48))6g(IiZ(e3p(^cEB#v8pCDp&6&{sxB6fUV zHzL}McYS5;=j@bgnZbTJ@te`3QK9oP^U+1J?>JG|c3E~sNgx$7Qg*O#Kdg+9?yqh! zOMjSt7GN0^$QGP7A0rN5sD0=+l;FHbpZu24wxITcRF3iE3llKDh1)ikUcMXDINV#0 zOrXNs8p{@?p6x>Ox$BO>VpAR8Z9sq{kejMs5PuU``F(bA)P|{`dpOd^2K;2_Equ@u!$~`` zCcQyOazlE>=i~AW)2O#99lfo0bX>37PtbDyn$KK3PV?NOkN&+OAKo}&s^Ig70R6nm zk=CeW415TCohJ|dQ=B@)fLdBYtMLb+F5ktm>_UFO!!32S0fLPv) z*w=Gyn9rGz{E_6S2P3S|{`+jfaa}?fmAhCRc|TI zBQ8?Iyik#ln5mTP%rhYDDFAa^WGeMc2)bi&(%)<+Z)r!WtB2MV(qUC$l7GUn!CFCd zzUNUK8FCIEl5TRbKyv{WGDG5F$+!3J@^QicSx^q%f9znk<5{Cop4HfFp5&m?B_&n| z`VfKzwNjiQ$35FRD|`JlDK|y02gx33<=I7~Ug6(&KySuZqvZgz^8>XnefSe)2~J29 zp(hCA*TT@yN+8zr83!>sOn=zG|1W1FnB2hulc{Zx5o9#uQ%G9lD%AD&xtZPzac}SV zjP|5T&43-|<>V;l!H$_uXCWA^&P)0RFoA0%jWP&Q@$`*d&J8aFi$`Fl!sf2qxaE*~ zXR&AfzA#w1F5vXTh>63blR{o=Y}X_ixP%)72#r=Gf`RVD-wrYpb$>lIl(Vp+R`_f| zf)$D|E`fgOL6`h)BmaYF$XVxiFKc*PER|^GaHwaRj0oGw@@5l^kNH11{VRih=1z8j zJon-;cv`#jTZc?;dR~QwJz#cnd^9dzFbiW?#DL4==mw-0Cob?XDQ4s^JCY!#S>#6- z^bWO@CY_9iK&Mx8V}FT5-j=w;>g**=XqarPo&D>f;)+Q`b{#qX!l){H?5m?V@}%v; z@PZHt^=Rf5&Xg66o(dKv>e~|Kz@lfqPjG(Dle6LIzGadl6?WEX510W9@!nHa^;aPc zp@C;ScQtX5r$p~MLRTFUkI?G>xmD|Ry^M<$wv~ntR|R1YV}F#hZ?%nWDxpn*^pd4% zRZi*`m9S6q=Jr%9PaJNp#Z96!HJs<%5!|@$sU7pv+JZ2he~Tr}`@`%mZ6a@6h`1c9 znqb2H@zZtUjM>F$z4Yy6@;7oV0z+0ohSse$NA2$)l_PAVS zzON+#JfhG4b${|eKwSVcAu8#v8oi-8%E!d;+quH&-UtA6xNkaW7T5DY{V`@`d3}b& z2?AMfM7=5{QChee=eR6O+ML~(A_y|E87CSFlmM&4plci*mXJ)Q8caWP+qtV9q?2b# zNIIDeSaB`WhuPu*LUvfi!5P+ptN!e_E{3k_}a3>Tcgh#;$+Fk&9r@jyi=;E zGrUf{NPmO+T{N3Y=}z4-eXDw!CdTn=#0-M-r|a2Km>W<*+yRk^0t0{#sn64Hw0z{D z>2+ygjJc;QMhHFBCn>{xLECdj(QBMV{Sh?ehmVn5^$5Aa-w3X=vWSdjt|o4!o?E}N z(PD_u(MBEhgV0IVkAFzbZOJWxYL!x1p+TICC4cnzH;bwcb!6S-+3GNsAZj|VNOS|Y zVJ{}gqph77Bsc-1dYOz?8omL%>WzjTZlV-O96-hn&FkzXVQj#5+;VxiKcOYo-9j}$ zyTYEocHR{rOe%Uk1yMXVC0$u|XdNorjjW8+k#`AMAOWal? z^G&)js$VJ!QC7GWQ*?eqZ&Ewwg89&7tC->;E6Wj6MbtCxBlobmr$I}gZljN-IOP+v zX$I-$yds@ZA2Lodm0@sYa@=&y@6+Yi$bbG+$)@Z#A|MVzDNw)lJ#heYhsKuy7&K#F zh6!BstlY%Sw0?BMcbtmksGBQ_+^YOmm4LD+cCu$(KV#%=!2iY z7@1&XnHz>&MDzezp?Ntc^I0*HY0#P9 z!~@KoaoS)VT*G@vP32p{Tf4PC^M4A!Kfb`*0V1L<4e#MT0HQ!$zh@n=nZ&vgC5%v?R|6_-MZE}3X3@#X*e_>HdPO-vV*?@KDZpD! z1m%?!|RDa0X8$xaa&(`a%J|w7J03z-8g>PH(?&6R`t`U;~p?3nEq;)6&u&xeek$y^9yG6ugh3ygRDVV&0->eL;Wcsdy5D&IhgL z=CYGp(R$zJRbBtUDd1{M2}}IkAL1K{S;R8>RD|4Q-gr+tNDd*q_}@+D~vYacT-s-+87&|RrX$O@46(a|XTkYNb>%B-@FF&PN}^8s(bHk;zkeA*PvRb>M6 zQUjrGpOK!m@0L#E{YYD4FR|j0J9ccMx-(?Y7KNT`@$i2|J^3bC58ypOfIHP^`@cw| zlmP@fpyUcUyKbK*F*7xp8P}wP`T{t^6kFH737&#>7d=EcgscVtGMr$0a+@K8x%zzn zZ;6YwZd^dI($YPd`r`x^Fy@;n->5{!vIt#jL|q?;b~UXPBOr?{##37>zGqjtvYKAe z?_xvemQ;WLzHq-7ILcoW6;S9j1iD(w+A{4wBC#Mf%&qTYV%yx8t=gX{GRe*s_dLAF zhk}k&w)5E!zubfb3&TZkrb3@}>s4$Nq)8@BzZPnaP&|Zk`C39 z-6GVE?2D`H-x2&D{?ozH;_vU5kSs0#f$5{zWUe+hR!bfm9Ko_%2=q2RJmze@MNpst zcf?pG`^N5Og@E4lzu(W{D`qJgiG0)%Rsu(~&_R$d=r*cLDmMB?s$9K2*ul+SK3gs^ zm9KwY#!ShGs~PDE{PI=Ze1Iffo{bZq)HAm3^V;59hq?D06{pk(hnxN$xR`nE_fUT@ zWYIgjz|J>K>KN2am2sFLIdmd0q!Fk?@uQv&X##aROe)zh_xB{D1;pEXzMejv z7H{dVWtkYqAu)DCH>%L3-{O2l6Q)hmK2dPWXJ{+K2yC)A+_{|b)I%V*BRMWVp0t0> z#T>%VAGudKB-mj)Yq+RC(tC&$j#KbWOrcX|KL3hi0dIN7(BZ|&39KkJ??0~@wLV1m?z2Gv zQHK#%XkuJAIMN?lm<)jc4vtYS`O!81+DP^^ zZssB3H(S+y?Vux=gtGLH*W&rp!z&uV1VEt-AW?#307HCHb7`I#p_^s{6`9|4CQK<9 zFyP}x2gn*+=F#Is2f`SJZrP&0A=e@feFIzI-@sBb(phitqRffFNdCt-_U3Ht5$^0^4QP1A1wT>}g#NB~Ii+ zgw4NlOy7qdLsG4ZzHDU>ODo0=&6nH^ZL^KZcZ_b$Q9xQy2WOxtoFab<2PF=%<2&Zn zFbb^it1W+?2`b7--5?-AU_lw!Orj~!-aEK?mkxnyf}UgqVr$`|zNu>P71U1npX5+C zrN+{#5qU?5e6c#1VbZ#zaDOAb4`TrEIQ-r!d2W~&VD8sPh>nl?<2Do^ss3{oQ+OxI zHuZFN3m!6o$-LokWEX#sMdBL^_V!OO}^3xP%0tA+5>f6vDRhCh4D; z3=wATnC{IuALA+r$<y}rna}xT=iGNn@3E(QISH}NeiBX>wmdpCv5~`NbYki9Bge8CF0l+HhkOiX~tW)YO zLj3!O`9UgVJ=(7dvR)0(Z`{{QAV@hzJD1_*O=1L`K4gWD-oT>ZjkTgK*buKBza_Y7 z0w%+O*9J#uMjEtt5sGvw5T26If1v86KYs?E*9VHn#Vs?!-sN(xJy8Z-&1dUVIWvOw zr99~PVN~72b7g;s02@GCNo@5oFW-9axsR6?Vdz`ju={2bNwT{nbS|pE@g(^F@EE`G$Tsw`c1qH z^)L}g)Bj+=(g>&{IsyRCe~xU0+qNBUtYzZ`+salj_$YvMPTDhc&gqy9xoDgIlWFN_O2Q3jdgOWYP~><*}&*!I)US18eM6#sk4@qtW1^CxCc82 z0Q1jQu$XAn_W@<7+j~BMGH<)eeUVj?sTx1*R9-6H#zA(2H&)TqUuf`nyB3P^vO??ft1u2JBNVe{u`$;?k0sf_=tU_%j}~| zm=g=RvJgn7=&0qJt*Z_8#?T|w)Ba2{lFIHH*v>%&dMZMM*PxG;?=PDpRrDyrtE`3# zR&;-6Bn~=Rjtw!iyQe_P)N-@UeLxRjyb?8BIC9O2z%+)85~5%|Xvoym*$sEPM|0Jy zoD13=+(A>x@!};b`HEhLogQvgM96Q1XKcIs}{F#_7HPDT_z| z^`pc1{oy-A#^o1qpSiii+_qH=(TXVF8;O7E?A}-V52P5zbpF%FXjGqOZQ#NQae0Q! zibX2qW5M7AG=m^(EQiyegDy(!ww%of5g1>f09cc)-Er_Yjmj8R+Vau2w%}L^1fqX1 zt{&F=#Wv4FAZJ>kzg?NHDI%~mjaAl;C)wos)8bwLA&4l} zsp4_pGay+K7-Y#hpN4q6e|Ns84THo#h%sLUMdKTcGlii*!IKisxIAgtxIWCi?%Cw! zwKodvq>Na4yIzWyQyaeUXPh@MJotaQnI9UQjXPaMnRw};g#~x5zliRkr{+8mj1F(v zOz6YgZ!`rz$J_W5sckcNI;_+H4RI6-#ME*4l;NPn_H=ti6fDn zDO}UqaBQfj=cz*5;BxZ44bOu{-Y5E)(i{(ORykIMJ^(7bP?rY44g_7rQnKah1&dY~ z$|dI|SEQp~g2#fS&7*9{GI@UNS2LqMPd?c5R=sh+~=rdSyLd+)l~=PLNa@Fwic-HM_>++ z=B#o<^M_#rDQ0`yD|x2LdsMi;=qesA8jprUx-CU_5Jt`4TQsVmMo`TA{vDV7s~HiX<+f8`tDKWwq60Uz+nabnnn;1(<%a;xp+TP>~*YHA_8enfx`8wmc=8d&o{Sn1)&7)jzyrT)2k-WIn5pZHnsA+CqD#_8V z(kw2h*W8dC^qjNE-Jow++qs)^B12HgU2w&DYH;3{L@UXLiO>SRwWc{*^b^T|&vy4i zapwfc_XB@JItsaB^&v+fjh!!+tZEvU`D1mX^+y0|e(#nCHrUvagf88DHdvb;xRHE^ zZG3nRP2F?A#u|V1PAc6sEWr`baVQ+qJ+L; znEJPOY#1HeuyL9NR^R_Tt8ogmGWk!yEP1o(>Z(tZ+rmM!vIbfeV6np4y6RljLY2kj z8xy^DAr&Z@^tJsYu;|b}-W^0KVfQiVD-|&ewvh@za>RdhTHv~0D}H@IaE`z}K@>>j zM;Y)?(o%lvP>;y~TvSWQJpO!$HD1S!$?Hn!Jsh^=-&AkBkxFm&)LR%1A;`ienyQ0| z6^eFs;6$O712%;3QC@A3`)#*{8DoB06uw%~S1 zS|3ONr7$|s2Rrv?lBlP>%`|@FFc=NB)N1}QtbK*i);{Ee*m9pQoKio3sxJY3PR4F6vw+&O=u?c zudUJ7Z`4qeOvGr|A>XhSj9|e?n#S-_g4@P|b_)~n2pCC^#ZOan2Z%KKf$CH<@Pt)`5(T-Xn``*q*O8H9D8H;(G4l%su zS|FgdHvjGs+iGeb*fIf3UQ1ML>_ntc_lqh^Ah5Um0XIHvtLv@{S$TRFU+3Z|D(r2M z1chweaY9r!FT~BK6AZmahF&fyaCu(jO>loAaWx=2wrWwV_a&eGA0g{8J}wzC~D$Z+@2x^*>GM+0A}xQe3vx~i9Yr!bPhNNupyqnPGji}7%loCfm-vwLXlsN)yK zH<96&*Jz5w`V0Q2q4nG>H%7hOIq`q41k&Miuz=kK55%vEP#n(P5%3OQ5 z-ZYllukk#W4hXrvAw9S)3DTW?1V_a6$f%iy%63-!wPP9k|2&e`X`5ZT*&g^uFi8qH zq%hSgjG^nq=yN!tphsZhd;Zy>q%qAD#T7tlkg&S7V)TEs*x3V3$4Y>-)&_rU>^w0t zM9P?K%VBJOCK8*AppfU4O+Wvac|k<*G>mnN_zi3b=$;$e?subUPG*yC1M|g5zHW|g z9BXV{3Wyo`T2jt*a3C z>P+AkW8=QSl};0p5^{KvG*EveYR<8rH^US-Q6sx~lkJ5`uu@-tmktAKme+Q@dl2 zQP1cK%1SrCrDIv)hOcaxJxNfmAVXmUfUP8Hfm3%RPk$mi7ZD^-S_^^N(DMJTnGmJA zYfE%&uV9 zR35y)SfPT=v9o+>Zalldp!c>1@ecsll5p8EtrKkZ^ zv5OhRB2TiFK}0UGqPNv=MO!P8_RFBHs!a@MP(_)P>1$JDDac3$#XN z5@>qXyN9kBa;WI9%N;St9CzsG8`KkJrpjDEfbzxOKU^+xzeegQ|9L^49X} zD~>g17~T*vVrFA-qs}a%%pdc9DZ9EXQtkskJA+35bYknj+^3N2qzd!N9009x9{>9_ z({9=!U=+7P4H127;C7iGUiML38VW$i0Zjai)fy>sBD=?s8upmcz3RY3Y11rtepcEH zdK!MH`c8k9kDKO9?5^)TbTZS!5Bb*=rSwhgdSRRT#8dDUAnMVR)%r%$iJM*rPBkM^ zzjdeRODRdZgc-LK%f0{)zkvjR5o|*4X;n%Y4Sg3Bdh>GpMj|Od*-BPv_YyYFL zm3$cmmbNlEJ?lcwI|d@TwJ$Y_y&LK3;E$nq+RuN_Q=#XyF!bUlSjhgk$C)Dn%$u~P zZD(@~9+AR?_}5cJr8GGfjCWxS2jo;)px@qo+M2FB_5Bm}>6tb8&!xhLY|(QC)?bJ$ z$U<5-sbilVwdGZkK3Ji${`W9Td`?P<%XtA&lQ_BCadQ*5slRqVbdtq3Eytue%=AnO9 zuy%G~uGzr-I>E(@6^^ARipdd+8CYI=)qA(b&W-}%kK^3-B;Cz5klJDjheVVLD=}=# zr6z-8=2JdAr4rj~(rcWKoicG%#Thkm{6cK=RfW7RAOA>$zC8c7-&_5zLyW|~XchCGtET?xg zE@22HM^*u69SMG=rLqfT;c>!edFl>aD}TOtc1_VEN7K^}Sn6Uh^M`i>s$GBW?}Dj} z-#WGg`OaFbQ(e!8-!p%c>H( zf+)WPjG}4V_!Mr#aXW9ldE0t!K0$2m6^D1`5V{OZ4a#3S# zM-l(f9dTR4u&I{88GD2zky?L&U{zV4E!ZM`>&wc+k-TJV2^q=%y5N^(Js+#+bZOD^ zQIp}*ny4r(2$$7xmxG+TW$^x6T~u@h8vuxffjIvTx>j|KBr%KNKNQO?dPo1`iFemA z^q-9KG`Kp2abr5joyxRU>F0MYg2Lray0Lx}%)Fez{ZvVyXv+!-{}_Km1e`tL_G|@l zFz?3s8=_|4R);qr1~|MWDx)ArM!dDx@`@x4X7B{9+LHp)5x;UlkMEHI_A}Dk8##0< z1PPABW+Yo>u7OEkS0kLe&`?pNXe1SDh~-RRe#cD)tb%hT(;CVxaK9!eQQ6AW`Q&EJt-G+TSOAABfq31WiK2 zbGZ)vhAU;^WQ_FkiwVl~7wr->58I_cYz6E~`LGk#)%p4`Tpw&u`t3@YA328tWahFfxC0Ya&W9!9MmcEUt$m zA+-rqMq0FDeOAm2;6<%SX&fp?-vw7aus1U>rpMMgXy_0!=e*Md>7wxfBSEa2WtlX$ zuAz(*!JJ|0Sr)}}mtE{zvb@Sn3cVYh!+Rkk5N2cKMcfh8uhssRFqxq8@mwg`hq_%z zw=7EJwu9b#Tjn~y{@z={hD(J;_@2CQ$tj|K;{6MG5NVTWzI%|y%=>PC#)+X;f;nE# z9Mn)K(_q+p;AfjHZK1o_XP0s#0VokUH8Tn?Ol59obZ9alGB`Lkm%%Lo76LRgm%&~H zD1Wt8cUTj7*R^2TC*Bm1|X$GNQfQ&k1(VVuplP$vpc{A!DVTD$=oHbwXLl+5I|BbQ9WFE}$2r}>cj(?F^LENx! zYP0;YDG)MhT$uYgA_&kJOi-YnE9e=6wPYg^Y%KN@?Hw@I$n3E&BnYVv3k!$9OWuNb z1XF86K-WMUz$0J5oDx}76DS}6QO!jO%4fI6faP~5-IxIYWxvBLk01C7C=(geMr z2C;FDEJk1uXkq$=hL})aZGUtS1~dRL5CB6czBqw^rTGz<@d7han22yT#0F?&CKrrg z&_LvZ3g?o8K>+3iff3=~0?Y4Ecs+p1puouJBkO}&if+N8L4d($Gos{g=*v)`HL=K6 zsUWLMg;>l`fC|!3I2#B?Mg;x8*}U&^?GIuyZO8#2`d@4NJ%Jp+V1I`G|CH}sok4+m zv<<`wATz(i7~K7g5RhuifGNJ8Oa0ky0h5vaAh7665NXqr$x*O!Ok}o^^NAri;Q$O? zTlYH*8C!}!3*>SE?awq2Da?1uk!}cd12{JqXIrzM)c$>(mi&n<3PfeF=zykrC1aLkaq$wcG1Zl7_ zgD!Ab!gREjK|T!nS4>muE4bADWkl`e*fOTJ0f1wZk@Mgm>G0nG%P2Y;Uyy%h()<=# zt`He>9Ge4CgD5bS!DTbaq2FM-0FE0J5U}(v`W9W@3|qm0wSTmE8ehBf1jqRqf$PfsY@G+Owr%>f&&cz!4Qyw>g<3hyW;)$m4^!>iFs{BYm^MIBgK+NE|$lH zIr> zgh*&P>x6ZT%_cpIF1oIuGH7`=*R??VVRrSg>aBod%aDmdZQX3D=KV#n_n8VFMb+Io zYk0ps##c#jm_LR*6Wwx6_F3DtmGI)M#Lp?@dP1k_BY*GQ!;&qQYuPvJMBV?Avnt9A z&x)8?#WmBN>=5P7r*)|HUIrpXeZ$6%Z!^8*INR`CYQJ6{=`crzpL#}iwn;0Rl#l|M>Qg&5rNsSZMRE33 z3yYZal7IA@-+S~-g0ADjV0;CTUX$t{IrDt7v`wMX#Yd+ceQM3Fb<+L*qmdGRX7(K~ z=b2Xioker1@0`v1k3USAx?mt7LEvvCG{#U;>*{)x^(#F^7ra_^hHn8c?KYuAn&&Ur zyGHHC;=Hf)KE^+Hm&)XZ4~8~xAFW7kMHgZ^M1RJDBj1vPJ=cfdeXe|l3%{W~uGdTp z+_s~Ne%I4XQPg14=&kUbHg$Q@u_6)IOd$PZ{CbkgfVa~n5w&fNHui^Nog6sJ-6@9d~FB@qZNSvL%_e>>^7y*~o@TS>@7fAf=I{E7FqL zTa<+PO7ugV!~pBZIqo-y3zG|_+b=B=FTTC$)2W4D)R33zPCU6|LrSRh`qS)o@Z+I8 zT~X=9UwY=e*BWhMC|8A`f*u6+KOeoHT&m)wXcKm2{ia94*YD2sTx-<*gt8D(Z*=K^+$hg>d5&nqUWpL%cD zmcqAxYLBs!&67gS?s)Metu`&`fnb3GO=)z3*On6=L+<$5A6#Dv?n-@1$tTwiksg0h}-<;}kCD1ZGjv8wn zL1?(TOPMdXujuT^D3vFlQ+mD2z(=-i&aUNn((SQVrf_tHahb}(gL}t#N9QUh0v=wV z%yTWCa0sW@9?)8GE^wV|WyhL6zifU~Ou@+Z`EB%=$E5mp{b;OTl$pLC@qhL3oqnZr ztIqe+Q8g7opDaz>*Tf!ri4_hra+r==FtRe*U_M2xI#sCV!X^E@wPpepscSfKZphWB z#w@D59J;Q*uk}=Lao(L>=8s4m`7`gb?osNGue6Q}>a~+N;#XmjvOBtRANb_O$)-KD zwNGL;$^VvmBw`Cq*w$&WV}HEGFZs;yZ`bu!q@O3ut@`!x?r+*%O&tvV9t%)oB zMqiycx*=_4&)dP0F+fDJTKrWW^ovBK{=J=2(!8+ON~@_0Zg)KKpiY00bhdc4!9twN zNvinVZI6BaNfCwEGF-q2p6PNDqKK6cIY*kx!p}UY?eDHSw_UdAD1U#vHrkIcJt4|N z*KN(BrY!P}x$P;_A~C>BnZc2t7648Ge%+=|wi&Z1)}c9l^BQOGmYhf8-EkFziq~|? z!xw!{H*Q(vtyr&-Q4BZ4Gzj(@?Wtbs2&s+ zd$>)*mWdl!GZm%TSAWQ_$$sNykp7N%JogdlzZQXj6dI~n7HJ(zn?m7 z`scG4i@+Ni8~L$LqxBEB#1kG>pU%3tG96QC=jW-9Lyhf)`;NOxrp+|IiTFX4aaLBc zSxk|BFl80)@JS`~0SwDaCgH`eH|#W3pE%~YtgpV3Gr=Ai?|;b)PqJvUV{|g@?X|5H zF7YyMKtFgKXlvWe9HJiCli@CV{e54A#Kx(yi0IxvMx|w<4P)a72EBb@TQ9Ez8WX7$ zwQf4Kn|Ay4b*D)o;?H)vFCE{79+&=kt?T2Wdd-wkT6 z7lckxR`nDpOTHS*D}B0o7|O7D$Q82VchjL5rnH(+elfeyIppoYIIjY=t{fMT5#6bo zXE!9gJ9^bGsH8&s!P-{W-;SAO-d<^mEkqqYmR1_m?SG{+4VavZHXi;kp4aoBTj(NT zJ~NAOVIW>#{!9u=h+dFTUZQC#J@|pFvhIVO=+mg;oA*A{+axC*Os;Y27?Anspk2e* zxnO%98o=(B>TU3tTU8mYg|cdofweP2Op2dej~5#%>mVlM!6*=dyyluk^!@fff1Pyr@$T zjNQuhC@j4}*}K-maZEWM_v)s2K&Z>PwYDdAdl@7e>basR%sKIa3wrIeQro?3?>0%B zrqsUz#vo!67SRe$OIKaX2_Gsdh$S52rz%B6G+SM~Uh zTJlEKw|T_2?yjCsh?|qmEZ;$#h-J7Fq7Q6-c_Z$&n~y?K-Tkq;^^GR`O8S0!dOwFB zN|${xv*%QqaoR%KuaT1-vKm(n5PVsCR?O6=7~ zdw+NE--Xq9ifc`{k`8y2y%xFO9DRD|W4q+Kg1954UD<@&)6cnD#0)#Onf^hX_2V&Y z^Q0JocSzRO+*n->drhNM3KQw}wmo}?X#ZnZtEj@}O9x_KI?A~3qnxl^wfk9?;&3oA z@T%mY$aAul1evsvF1h>f51z9IU5em~4}Wf?)Lp6$jFoY4xK;ayuek-U^yt7u&vbg~ zS#aW0dR6}GXz9D%m(6`@6B2k&-bWr}aT>6z^l-LlDV|sKc!R{)!Y9@?m7_kw*6peu z=p)i1y2a8u_sVooX2I-~Ch*?ny@uAC&cmIV?W?U*^@=w6`P6X(Th*|Ud+MRKwSQs_ zuj;oZC^{lRa;MD^bwbY?}9OLlj9ZbEz9@QFxPp0Sfu_OHgAZ?WLJ zy)V<_`kc7zCB_?`a!di=y^@|6H({1vKv_1-94ijxJbnq5)9lk7|euJ=L$8<1eGl zVGq+fEvd{4h<|JvMI>$~RPPiU8GAp;e+v#J6`W|x~R0~WUq5d&Tjml-w#6}N#Q0~-sMjg-H30x~m`u{aaAyZQr~MgcXGu{adBYC!}I z2@f(j3NK7$ZfA68AT&2KG?&3%1So&)TkCJzMiT$7zhXb$;SlZnbtn#qljhOJiDNs7 zoggR_o3Yu*k}Jt=(!c&@W=U~LiL&JQ4*gKTS}m8u-I?D!cPUFIm}Xis$E1?XGwC#w zmMH-PfhiN2VJb@KF$1@hV%EcmW8Q{=@+=6+v|_B79n1;yN$(Uh@O-eac!2Nd889RxJY?bd7%Vfu zj8KtT#gv5gfK(|L2Nq<|61a8n4Gm1Kj1bJCQ1B2;txO2af@l^5a41x?%mKX*o}sKk z3+8}6Dm4IBwelb~;N?I>uRwnk5CXt}@4(=eF%hnTMJwSu;5Uo|Yf?tQ3BFZe0a&tL z4}xYOCR_t64%Putyb^E?OE?QidiZvLFANEoAAk)K_>4#j!ba=@aYVF@7)0ng0679q zLI{H`d{m6s!pBNLmc~6|i24PMV-KE?2K6EF<3KXF1D_c18lF(f#({st=amI~C9o|4 zBy52~3v9tKfIW~(f~839aZOOOGy;$TIY8-%o<^;EkQLN!0)Pj7X@#+aK!$+gz_$ho z0~!QQ1M^u21IVlm$T*s4w16y(IAqoV2Tp+w9n9hku&S|RT^JzSAQM;`Jmo?_dj!)G z1{4iT0O+7Wc-(7LId*@53(z^5A1G7eUm$6q6D@)WmI$Z~e)xg$BgUUjkEe`3W*;>8 z(jntd%Ed?apZ~z^0YCnBT`+zyx+n(x(R8vXCW|?iGjk2-oPqFDAo3$}i(%}p*d*WCOd_^6s*V*;uB3PVlRHyES7!z^{^ znx%GkGfQp1%PckG3E4m?GZdJy!_at4yJiR}>%OdVfAqDyy1DvgTz)CW<>hpG{!1}m zlvkrw+th!4`^f0mghnri5;EB~v=|$eh+EM@b_CjOqGNA}+rr-RaN5`y^aWOx}~n=T?%vh@wBX1~J<+)CZz^xdq$ZJF#dq#EwKVrB9+cy1OEbY2mRi`V4`tO_<%gkep}^8>S`XXD`~$(s3{CyzVOQm=Hh;?L zFpPg|+5myit!bt5$VBxOBovW#3mYB$IM|%{TiOt4jV;=WzM6Zv2CbOdpivGf>B4x0 z09s--4N056HeCsNAOcdhnj+2G*TRMhNh=iE%We3IYn~E#f|aIQ9Z9@U3)T^$0z|Zq z$Orae%&T#NuJ^U13QyXIvd7vm zLnoY5nhHRsUgNF1&m7AMVld>$zb1ViE-pR+B8pU1E6YJtrZeu4D~(GM2Kq&b>TeH zB)x#X0H&y`NMI{U9Tb5Gl{w5{hjai9{pit$5=}KT%~17&#HXS_aA}olU23I>vNhmL zTndp3luRqpw4kDis~DB_HR&;@slb0u<4!k)m=yNRyHL zgXT$N^lUYIN|6ckH2*RMs^#*wjvlIJG7~u$g2bcA@~E8pI=SYOq5?3Ds%C$VlqIS- zv!qTGnsTs?G+$Ckb~g!-<$J6O9Nm{5Rh-;bplKP6;lf0WM#e=L*A`*3-r*c6zYDZ( z6HP^JJ-(s=qoU}AV?b%hgrZGgW6-d8*Wr;Q%B=fRFmmDb|?k=ch1r|vYhzOvXOqpU%?rN!RP%&i27M=l}sEBT&X+}y2hV*O# za%%7`AL&u2NCy?kd3BZD+c|K2sRU6W2 zFUbofJ&o}~kw>d~p7eiQzM`dy1fHvAPjuA6_JOu)*^r~ja~P_AQYO)~K=Sxw04gr2 zhhbSulijW0#W7&ORM%n2RzsQY8+=J2uv*X&OV8#`5+L=d1?6Q`B328{76W|TcKS|I z$h25x1c$*fNQ5YALq%91MNps(X*`*kqwS_*5uOAkLF_>25{ZAH%Gsf{p$>6DWI8x< zBO$4lHsrh%-@T!ok~cKJLvN*UhPUxI94KMlN=dug>dT}_B;g_m2Z;d~3wZo-1^s$~ zeAZXBHki3ERaFDp6*!KO0A>~v1@VG%E(H#oyhdph8*&v)uT!kY>e6|eRJN>hogr`X zkyK$Rli#h?ux5X|4c=4vhyJcXV5|VP|D-2U8zSYqthJP7s}Wj1$Ur85S_@E1W0p=~ z53z^@VZ=O`LL$YOD5%z;sDl+Yd?867C^3c0svmG8vOJyu3W~{TVJXudFL6Pw7Z|u3 znX6t(QwgnVq(1XVhXGYzdD`vJ9?#dyR(Pc9NcieH+OU5-Noi>DzVRrxvD6xL>gHZT z348EC7lbrIsf6mZ)LIfb#Z*K6d_467$wt{d)q@#Db5AdO>CX4G)r{0u+V^!YZA$3c z#42jzIT6v4D`JI)+Cel>g-A^mk3`SGEe=L&^b-0350Mc8Zmhu~T%Zz;0tUGig64xF z#T%l-g{6OVs_LfJ1;6ClHfSV!FtNF3D4gKg9G)Yn6r#a2&I123h^;}?SXF{9vAboS$Vyf&eB_mm!m7Vvwv{<{@MP& zUhE%EuSOFo9_&rW=RhzXUCfzDlza3Zt{uIaampBf_U$Nj4;vj#b{ifJ)B&O3nm8qXi;3f#ota1_(}B;)B-TaU@j0K5NApX*;NSk4*nCpL)Uowq<~Henet7)q z?c0B)^lFpz9;FZX11nF``%L;^TGHd?HqwjOEB!;h#~9SN@Iv*@8*$0@M4GcLz^j0ZjX&#pRD* z7sX^=PSIEni;J7_Xto|PWzbykcW1_8jirAYz>LW85*b>sDd|DWMBwUU`vUekw1P*E4)HyHD3daE&kZ>Zg|%&8_8=%~iTNo1WjCEzZmN^?3A+&u^~M zZ!M~U7T|Ulz)ybu>GbL23h;!+ZD>Pn^gyMPd)gG_zS*%(y~#Qspi5SWi(bTk2D5)W z;P3be|5PkSV38m>FRte$Sf&KCyxhNZ=g<7Qn3dCWvd*s^ z_w~EL^yba+>D$8=Om-tmqQXdY?pyHL=_Cd zk0=N~p&xZ&Abz;KB-_ei8&36mu`zv&YLAc7Q)BT|w?MMW%913Hf^`%FCUi{OflI zdrwcRbawRZ>eF;QzqNXd>;V;z_+J&?XZ8f??QR*(^rM5=d>YNDG%SB8mQpW8PZ+p@ z$?K9PdD*y_&r|QJoIp&zx*0FZ>v0FBoxrTiS=%8Lh;BJ5jigy$_W7 zuJgo0}HsjkCcCE6msomOjTBel^C7aqQM8AK1efs28ts=0~gk+D3 zBB=;p*mYH8+vh;Gbq=(mqI|E}A9bFBf(|O>+*LIzRd-bk++2MED9elObktc{$4S-D~4@tVM)L5S* z&|U^rv8;WQyO7$+En&;UtgA{xDgmplP`xQ)P9|$%&bD@bPigP>6O^M(@x+3vc7boe zD&O%FD7+wUWQNJ2ZrLV_Hgj7ReN}tqRoAbrNc`KXwS9lwRvoM7S?v?7HK$woLD|}a zGaO(ctQ>N64#DvJspmYd=3z@Q6df(W#%X!3#B8WlPG-QIs%tXlOn$-yH31$`u^>=c zh99%wNl#KJ#2fc~yu=+5|APP2%O_ZW^J}bQ5Ow^m9bt0EOfOf5?zCUvWXU@%=% zFy?h6v!Z{lFPZabb;RSkIuu`MV3E2VQ$}obBxXX(W0=+CDUBaQlbvl~nCEm3K;Q*F zPOzb=CeV5y!-%r{_clg@5Zw1h^Md{pqWAqnD`euoa?^ihJSk`M1%@nU?0~-*RTq%! z2K-%lzPOy@e}NNe4GaEf_1cKMiL1W%0v)bQmn(l2T3o3mS5e#pR~rt?+Kl19mUK9i zu7|T=?t!!Zvw{v|QmiP_`}hj?z}FV9#yUa6ZFWsz8-;dV)9c+CyzNy| zhr55U!ky}K1Ks6L?#|sUZ*aO{ZN(YmSoih8%Ibvo)4m(TehaHPbjcg&e58u9fbGK~H)u-ojAFV5le*dm<`)$4jyY{Tz)p~B>W#00Saywq< zaql+FV_iAHI`hPaU-sdWZeMOzUza{C`)2R|twGZ_iZ&?vyz8ilivV62*0}zkxl`=E zurnu~ur5CRfopHgU-7@%=MAPiIGBY-uekYbi2G02(%d$!w zIJo7OSyWu2U}gXx<}x<3n0_FRMYFy(_;l7`1A#riwZBP-OGWE@w+U#RT-!gz&TlFE zWU;o2>VaA)CPD?&Hu!C`}px1*v3~EIJ%Un_xaJ8OrMcKz_w(8B(*AF^ucnW$gSFYw0d&t?tOJ-SC@M$-kWu!Kl$maZ8M(*PiyKpHJSO=uU!&TY*_8qux^{) z9JNfLwB+o&SG8|D67QRGiB?Q)n#8=7>8yX=+}PhO;dTPSQ%+ypwnU|D)tpHd4M8Re z`E}RKyM9!61g{R%lZnmr`qjy!`Zr?Egc;oz-L^ceUwqug+wjfaIpUK!U$q}>QIUBz z>B~WrqMiaa%T?h@>3V5XUq-(6zw%@9-=+%F{fdmPbxz;zhyU^}$h3ZwTzF?*PxO`* zFGS|c z78WLBv*}0US(NI{W4ZK$GpkY+42{jX^uraR4Hb+P41t0{`S~S4K@;GhY7m#cXI@%9 zkY@_zxhO>27&w`^nK~J}7+5-)x;Ysdm|D1+TbR1KI+~d|Tbh{|*eMWJ63c@~9;T*{ zxv<96goM*@e}BE?0VGb)}dX)R)ON_oUJ=bO@+O$Cg6 zcNvV;{O&Yb7jT9h?W(j_^WS^-Yk7G1bwgummb6CAC@*PMiHSE?#Ip$;lG@W0d!wcQ zgIJt1i~It1_XoTs?MMFYTU;O{v?=87dsRi&=neJEw?D}B9W1n9T>e0(=hPlbO@bT_E_WAkxhCoqJIM>#+w~J%_FKH1qCYbJaMps{YT-*B^xP z4zS#5NuAN6I-^N+2dnCgHls5KwC=Q78*q!RI25~s-E_v}v$fy#SDfDW=hH6z-_aju zO`kM>ubjuxZS~ikGnYNSI$>7)-dUFpny+VF2^`-N6W#JFu3zNOjgMKHXWmqA6&3S} z>ldB2ea}3NHGjWVtt!aB`FiG|`uTh2X=kPGjnjB#x?;w`qi1x|(hQ}$BK>TbI9=}^ tESS67@;5)Xg#H?|6agGOFD^+eDyb++P2=J7on?YGUYgC_3vgP8QMdyp=-O{8(D{Ha5!E?1w z##%(TUiVQi%rbq90;m0T`ps5*o#We^Aj6m6{eA(xy2FG?a`M{NG3|_A zo>S4bWNEqTrYu}cw!*Iotep*xmC^ngJn=ss@B|B;@r<&dx&`}NgDN=7g%$CQ5e zyUULxgSgZKIq{gzKcj88A|OPgjTWm?ruXqU#y6^>59f_)tFyRd z0pq3x;bNT3vW<#^9R^bU7A%9kf`y(W-ZawiL+H&S>W@FkANdToa9OT7EvkJLs$qVR z^$Jr2(U9))7}B#JohG@(sk`-}%pizN+)6w#uBJx57VY@=zVfA4w?L~7HCC|jds49!YQfsEYLZsTK7EljrCx*8pe z)#D?G-;cBH_?+>Ij1IC)sMYPL6e9E00vg@#S2zvQK*%iPwLW>of>Q)#V%>#t#9n+} z=_uboQNTKHL~83}BKh|0<8wlUoC{aMT^ycH@oF)acyVMJ4-fKK^m|U25^#>6FEyWt zHhtn~N=i5Bj20MXVk^&??0%G{=r!(OvSkFb6$4XN;c(vj%n?mGC<^-9xi@D09A1;j+q`K`NFrXg43P^s&l2LHJ|wJ1#&{9P%L10 zl<*Yke_XM!xzD=HamfU@3_RTA($+4kG3q3?8x!~;9T3&~(>3sL5}uTtT%{*c)5I)w zZP?2wRn4RKxUz$c8r3oIC}LGt0`Hh1gy%w&Ny7s8F@@}I3$B1jaWf1+5zP&iqjKO4R)#+8gL^#0&nVnd93Di#p} zw`byzqEr87a!kTC@dP~W+v2H7v3FnkZP(H;_gb3t-FgO`2{bk?(8l$830yS>b^GpQ z1SK+`tSO>Bi4pw-=iw^Cp4ow?6mTetIh=Ny@<&EZ%d!1FW10sYA-i&!q~0LLLQYfaCOpE+IqV)Wxobin_c`D&47LJtTo_Rz4_BzdU%Q5I(=M47q&6sks zUTT#4#P&3s)gbbWx^B}6k=3~c&L~UdmsxYE^^>kkL#*=4W=3h?JPj`_^8B63@LWKJ zqDu8Il{yUTYal1MPu?{}&Mn)EW?ximXWw~$6gpo2(>yuKPq3-&qED_PBfeenM42$B zP$(O&S+QMI)QN)F?5}D`t2u*H>zZuam^<1UJp02*Yn#VpQN* z1*O>bQtpI&Mr|UY1m?5ly4F(^1#m+*X$AAb%MBdc@H8uxFd|t)wZMra8T>jL&zlmt z=6=lBs<_z}-Ppdng#-4-pQztnOB_S}?8MBxF>2tjI~-chbJ8{luZz-frmu=Lb+70f zWHiI3vVh9bA!)YBX-Kznv80giFfy45k>tZS_Y(Z zue&QF#w(Z?f4lZrN#^06!U*qGs}QvusF9Vs6x1!GcLz^jSZ)tW2R#Iz>*u-NdG3yZ z;8-Ud!9mb<*rIEzgf18ZnY|b>#uc_X+Ml2Y^sV?ZYQaXmS|Pot(a50Dve~k`gU$K` zJwVEpN|8lU;m%HdIJ8r$*&cY9a5XuSIPJ?~#|bfGR@H)Zgxx3WvV|C?fhFOgRd0#A~LbnMmVkF5+N7b)B1Ic(Xz z->AN!ptQe|!&aqGvKi#sU4*r~|04T#1!!yNK(PPFc`bHaINn0Q?8E;~!=B3#%-bAm zC{=2ktnGT?CyE|<(i6bjXPm=J`mi9+5qzU4w_~TZzYz#H~z55gD^>-=xL(M)5!4kqiLcF*O@@m?R|;Py-W;& z?T+2AWmy@9zuTKvqY;(AMZcgIXMW4$-@$9SB-ew$JF6Z)S7wh>L(4mMqXYg=dlg_=q^g6qV$e3OFtxv(t{OOCG3J#$|pA|xj2)@ z>~a(`0m|%}$~)0ZVxlJI(Q?41LN(&rYwoK2cV;QfX@V@j+HYdwXoWEcR(rs@S#*)} zw7+oD?9Q#q`=8RA%WzF<&IuwzQ&?eX>_?E)n5uer^!Vrb>6?02lXUU-W)%I_O5b}D zdsaR^+EgmuZO6L)!PrFN*jg;N3!@OCpW`9WTYP@i8aWb?xnw+rB`=Xg5U zI_7>>?Ka#wMj+W%t5h9Qu2#(Qda+Mp6a`9ZEp>8v-Wy%*XNFFsVVArXRotRykv~o= zW4GlyJx_Y}KHl5vA5XyCv%jP~4p~C%u>0rCYgbtRadF#e#)xwYmIF=Wr}rRd^x zoNa6yWA_*yndRXrr(H840KCRTBnfu9^-+u9W`Fg@D_y%7Z!fScU&dVM_mIfIOK0aA^2u&kI zdd}btB zE{4}G+#sbQLM1?;FjMa8eNU~Te=<3bETR4^&cK#>l%B0h*`8XoOGvxbLulS$`riSn z9D2w^EkguW0Etw=)JrAgv)xq6i?6Ct+L!Tq*L?%oG*$1yx~MaG>>Hy~m2k18Lo%?z z5d3nrt9-!e74+(gayI;ty|EVlF1ei18sf--Zc=Q%hRK^G zg>0`qulwzS6hRqj{dnrS;!-GvG#2Jk&=4ALbbKx>kK?`$r^?rm5YHxOEWgsflM0HD zjK%WdI2lL`{=gyLCPze*f3`{hRp5m1GO@V{mYtycpG``E)rdrPX7K~qc z-R^mR*v3%5CwZ_)Y^acB2dzVfab-q2U&r#-Sy&--^@VS@qlANp;3;a9YDDpzqF&VN z-tPb4npQDh@MTN`hfFyrl@(Ha4HqhFvS$JwV+}+%V7fOA8(_Y%f&V;(VW!LZh6$BJ zAs|;TZaKZ*yOh3Fe4IW#oYKv21Y71=sZrvJpki`+{oWN!5M`Htx#G0;i@x|U30!Df zb2yU0;#d&TxF&5qiLYK~QNntz0~NnEW??t7eHu!$VSaobQ9>TtZ*o>hae1PrRht{w zWS?gof-Kh@5q7X75z*ZV95N4(0dwqm!u^|Q1#xj8HRXfKJZB{>I^&K=mxYsvg-a$E zuJ}3Mpn^}U_?^oqa@%WAEaOJcU&&A5mOp;k>rG7ACtf}Mu!U}lU@;>KC$f4k7Z2D zQ)OPy+jE(41jSqR2NaHI^uz&3v%z@OKSNC@_=h^FvEq?B=I1WIMf}0780!T@2ig5C z%2PS1|T-^ zGnSJ|cW?}rAjt`PAVU7*MI>+CK+VW03#ff*JW-X6pVZSS2H8lCrl=1;xMPrTj@2Cp zk+R?4U=|TfeTSB`4$JpQk=#~Fco&k`)oJX>1LKRQE4}mGa@qq2pF=dV@8yC?*gUFm(iLs(J7RCwSMmw8W z`4-fh$@>Pf%lWYq4grYHgTXsqqa;rxB($a{Y|6F6BmMZk#1$do#RQ6DrpQ&a*LY8p zHLpwHpZ2rJz`)SZI3M@G;(O zv}3@v=JL9{u$SPy?+!co8JLg#a*c)sF=-2U0{H0_~N*~{5 zDB?PlMZG(0zGtrpoH$W_T%^91go$2jypL2~X$`|&wIqD=m)}JsHsylIx4vn$EEImw zv(Y44bV=!z^(@PbDFnzF3{<|R>%MWtcl3|D=)EbD3wHdwW}K>RwZK=CHsv#E%Q5t! zpiLx4zB^QiBR3#uEs1$NdeY8?5UrG&b-HYu9;v1c`!5$dec}>=Sx6B_09Wjrp`iFwXi;O`Hrw%T}U!|bUvcPBQ;!-Q*%>K+x!uz>E}klvFd zsA@94@Ze!3Zz@SOZ$!)*YOs{SIrL<>!!jkmfm(`ro@C%y_t_eql>pl{=C-G%Nv8k! zk7-v_57|bAQsQ*yWr;a)e?6j1n@UJ@wHo&wfhy6{vn`=aU<>#*d!}m(DzXx^x5fT; zXFYV0ORMe;#g)C7HtJDec|V=A7$}(ip2qR&60}egF!0e3Yz`K3Hb?-8{=CG(=pK?W zeX1FQ0tVbzs+Ftdl*;Bq^MM$92HJXm$Rg7{BgE1M!+Rm-a><(FmbE73Qh;Y`uVj3F z$qFGP9<}~u2)se98FT@^vJ@_A(*#@o$pLZjdHg}|LQgY6Hq2Ucu^z~8-D1-)ZV%t- zri3fhsP$t9pNHNyRI;hm5+SG3r`kJr4(UmYv=oI>c0C5bq7DS@uFl_ktJ)UsH~++0O%Jqbc|6Xy&15`-^$P`s)V>x? z81_hi?izt{_f<>ZXl&8x+(PF7D~Ue0=gH}bkZxSyX!U2i{8uw1hxKUy({5ytV_2ci zNR{Jrf9e9FI0NFC-i;{(LVQ|v4;rOjC?Mc(Y_o#!v+iB#7s-B!5`|X3SCgptEI{}d z0hDgnQMH;9B-pr#kb*t5=0uNkFUa%at-Z0Lx4VkCqYqsFN*)GcNFKmxRRWsosi=Zf+f%%H)O2^TkyAvme*Eo}1Qq4dD<3#*v1R3-nsMYOc5f(s zy&bn_IvToi8b_PS+>UO#^aYbhlw$lr+!c|(!b&tuXm@AHhAT{vZ=%g?&T4Nt^yhLM zx8VXl9jULXI6nEU;^`g{%8rh6>^31!NTucJdqW zIP)M2fs&-k9G&I@p2zTE^Mb@FjH}jNhIkaaoCOCHkAsaw$++pqk=15!=mN0PvrR)t zd)#NxcO%^SjBw*Kkbce-q6%5oGMj_H#3_m>nN;s5g7xCNhAQ$`jr+nkk?tglqc6;w znMgf(qGe}0gY$wbfMrS1ZcOA>EUf_zVEt331j&(B3=%`AQd6iYk%qC1*6 zfo*26N5Q1g=15wV59ktT^N7hFRvFWatETt-SO&Fdw?ctwET(x~W6Y>Eq&ly8w=HdL z)7RsZBt{clUh_$g<#FPhS>bq~VY*c@X(b$^s>gOt#M@}lG_wHaW5-V`TQ}f{J6st8 zaD8J;Fd0iI_5EQw`UFnI6@p)#;-B&6CCj6K{Mgq?h?;ebL%BoeYKY7dA!msmI4D?B zF2g7A`#)nRCvvl6?Ye|dcCEkzIY28_5Rnr80tMDptjQLCi`-Qghxlpd!&(30eI9cN z;kko4^Ofkt`n^KGuI66;c>~y!e|rw6B$H8}LU5Uusf6kPeJbO}4x z<=P@`p$^Dfm5$VsQ9CnzYu?aK7Y!bo$+7Oq1CJhGoRYA^30OH6_)Bc$QCDHZtE(Nt zB#qXPHH8~^XMNM$+f4ipF~Z*S6|n#{&aR%9nG;;WLQeEoUdrakkjpb%g8*`yqevPyFJUb}I2OY#mHSCVB=oD+0+T+@R|xNw0pw3Mg( zX;6M=c4H!fg~odOvj()FyfvSO@|T58Nj4EOaJ!#cy!vSz`q*-FP7=0L7r_UMG~Q<6 zmDSJ0DfekK`9SM=T+w(6%D3uj+w8u6sHKx{Um&n63??q0=+k!B{~T*U$=t)*mJ6Sz znPlB*(hgr$6{{oZ36v1WJWtZaqM?fOhYsL35a3-lW$BR`eg{+$A@f2TrilL)8Bu30 zj7e}`$8lH{Lg)&S=bp_Bk|6=)KsugTsp$HQd4V z%J>YfV80_T*jJW3{?%Cj!z-FUI&(WJiYK+yXxX$J9EY8Aw4)SUGwu-diw{+sCL684 zbP}$}QF2AXd=@ab$GWtOZVsF$x~H$lQNcIefeS1|*x{Tj@-snCL?6d7E*lkf!e-ab zUOFHx$r5=Ib%*`8720E>dZ`g3y7rIs8B8Q&nG#dRzIfvHG-W(-_2{=IL9FhH9Tx5ngzN?MIl+-xu zH#1$Q8A|Lz*Vc1|eCJ-!Ur=GelEIqQFCiHxs0wGUR zFAf>fQtOZdiau;GxSxDa*=a4E^=83IDQ6yED2uI7QNTq?{S*I z@6U@rcU%D`Zh~`d99Dx4J>Iwzt@zy?cFUIT(6IybNGu&j28%u$vPw>A|Khd&!9X4GNu}IaFHzoROjK zM%c*}8DWk|rWlZNA5djNhm85y?||%gj^Yv?zb^t)CI8;@uqkgzhG_AZp(D;{MU%Ev zD47l_;P+|?7#hmKuDWXod2n9WmWtuIyWIM>hU6_1l^Z0IN2n1d|9PN(LC@+V%SrV7 zf_vU1gqo=jXCW&LmX*3~C4<)QwWM3)|KiSvgX+hZ^FEcqQzrgnq+=-c!VriFEA#vq z|62wSmu!Sn3hgbzR;Daw4I$*PGot#SgMjng>$&N#3hIh^>0#$p-#ZWr*eSf*inex@ z;m%q+qvu~!3F|=353s^yK`QStn&V2XFw}G$CW|AFkCQ^A^vl8=Vgm(7UJ8IqUtIL)xpsMVeAYj^1l46J1$_!0_n3o) zzg9S$q_7;#wDiyx@G~3~==Jzjpwc8ie!41IXaESdA|mi*-p{i8~or;Y{k8wd?}(QL6=jAKSAQkusW1}YW^>Q&-1 zj)z_X6h!4vUGF!^QY0+Ja6UJv1#Osq7w|df=rVJw;o8xy&)YNxc_x9@Q~V+Ixrb-i zAs@BzT&J1pvmOw?OIj6liI_RP8v%x$D6iHB6mdocVZ62~lzI{^c}=Mx$%br8x-cEz~%o4G$=N}0Cw z3F>ln*Xe>J#Ri`G>B+7-B9R21-FeKb-;t+ZhnQ|Z>Pl$zQz2TmTJCLh*#ZA#SA1p~ zPtXZ}MghO-;u&23P~u(8)b+HdXPTN=%yL#G1gm*?p7viOFJp{JIAR8#{G<;~oqn2> zFk1meo!_8BWKiMcv}t~davyhhNy^%MDqc?@G1|{+$cRO&juX@>?d%$5<_op}A{g6R zRf(p`_(`#)Zv?x$LB`F3T5AE@RtR%$&)e-F{I?g;*hx`$X=@TN zp((b$J>OIA??cgQuY)1-%-4C}V{>Wkv)OagTEhj-S#tlo|#3Zr3k$Pb{`7xhApl}lJYd~ zk&FEAE0`foMZE>pZveN0ppN3q=4ItoW`RfXWQeUn0ZNN~gPoG_aO`vS`8k%p;WM~W zU$Npyb-~vK5vS$PDnFO<+qIGATmAV%SO}38*{A!T7=gIrth4(3TCoWx0gs{;`Z!>typV$-PBGB_n| zap6X{8=Df5TPtj*!yEPgw(fRYFqSzN)x+?g$|;7?=y<-%S%e3ubuH~NK?`N!T`)*N zd%*ixmT$l1qFJ$3B?wy}EAfo?asS|xb@Fi>{Qi<4&I#FVaWfFf#&2+LWj&^3tq7r3 z{8fYYK4k+rOaTN64P7y_Ch6mvx93|$2ixGnN-FuJ<1T_M^NEbY7mPK?K=4p$>0na% z5)t=!k8iC!|Dce|fQPw7xJ@@rL({f$$u=6$=7rI1Fa1q+|Cvk;ooHz9XxMpt+HcG( zLffBThSAQ162>S#CATJjx(Ap7Uu&)$J`waF(hC053Y7or-GH6re~Sex5>_s*|1usl zz@@JC?{x`u|6Id-NqF3*Hvx2<70@{p)>fE2^a_|W#LV|8L<~XgCMKs6p=rxFWc76^wl|S+53N zeaG`hT2%O@?^({I5y^Or1jZ`C*q=m}*!(i$^jw$Wj%5W>AWxo&{)i(+zE{#Nz~(KH zc?aKOMYS_u&f;h$*m@~5C_mcnmNnrAiIc<%38D#MCbL3Oq+ZK;;zv0Ji^$H#)SA$f zy#>2Wv>2^~P*+YxtXi8ORoz7MMW7kx|r`>qNOs)A^CJWn`r*pv4X=nE3s9}%k;|#ovIL1K6f6oG@zHBPO zO12y-)uBmfjpN!6k5~&nF_m})IvBqXg{4<~wBz2C zBsisAo$g%1FmdsqS|YOoAO^DV@qUu(29j!jTh!FjUw?TK_xJ+1UBi?&_(h!KC7oYw z#Zo0>eaGDRPtV!i!rde?7UP=>1oyFr^NtQqU996F&)q&wrx3h{&61<~aO&)#jR#gB z(4H(N{`Q$RS=F_91HWs_hTRH!G$46}W&TZbU4K>ktB>==N6x9Hsbu7MO={ZC7V?fOds{i) zTDQK#9S+uPn%zjpvqg8ubZP?lSzY9!?}Xo@B_<^Z!dhmRD@wGvzy*SfV1x4FvEX*0 zX;H0#jV4DZuf9TccU^~bPt_Uyj(O=7cl2h1zUnl?&=yBvhV=%*%CcHC!E4lmHZVhzh=N-3Fxw_CrUl$x)=) zyEyenCl2k~NvKi*eIlM?%fwAa5_Hw{UB&TJWktY6&9)#1YDhb41xA6PDO^nPWrFA80#fpY-IurzotB(+jfabaVq6=S+suZ*{D%DWF?mK|c8ih(ZB5EYiSTRPaCy4`6HjsFU z7JHV5%R8&XGuYH@NxL*UyS7z8nRvLgaO}0Zw;*PY%A}%c2c@qtLE3Xe} zAi1W_NNWO1>A@L+EI*Lj48MJinlvYEqBwdbg~4<;RwC2jT}v$18mN z*ax(fe1hg!Tkp%dC@||##S-J|5`h3*q-!=0JT^F{cUUF~G^vpU=T6LV%J}S-q_|Nx znz@b~OI18$jy!Aj<}~t%y>oAW-KKV|yo|p=w{07COZJLIP236Rr%FQ3pcDMeyRT}^ zw9!!LXyw_)+E$EBRK{5(Z8uSm2_#+BL?FrP(z_Ux;l)d>S1{OSRfh$QM1)D45>J2y zTX3Gx`n=2OVX-5a&N;-?G5|F*tbKM%T@mD*5 zwp<|uoYH%>Ze1;J^^A20jVXaTQhX$#rkAE+37jLqTDHMl$>f_oxs);_&Dfqt9pHbi zkMH1~RD^`4(DGfD0PQ{E*JF#ixCxO!yaSA56m&ZKm>Istj*SxH?)te$5+{=+znA$S zRVVzl&?wHlD)DIr1xFC+xov{AM1+2JX{`g$IwEiBU;Av#)BJ{f3jMY=$ zi6j=W-Wc41yx52@LzgC$x_Vg;e?U+iWk&CP|7`C#O5-m9`HqkW0T%SPvB$f;_UltK z~N4UAOWIAi+F z<6nxMoBX(ItV3HCTv~C&@a3xwMdfjCcyQG0#8bROv z3fsrkyu0Ns3{uuh{DB^iA~UfaWq#RCtRZ_CCj(<-I4`LT_Gs2392Sqf6(`t@;D$IU z`O6ve#=FcE=cx4-v^vYZH_gTtws;Id*9Bq~S!f;v)zHdQ;AKDQZAzc@p}ic!Q2(N0 zM4vBSK@{cp(C0DBmwO(FZ)uU(YXG=Cc{TP~VY{d$rJOSs9Z$x>v29+bb@AD*+D>si z7w`O28T5HXt_(suP{?>%MV&%J{5vZ9-D*XfVgM3uDW$_T#gwY+sqE^XDIx2ntgeWT z=_=+eG%jR4&+;es0SKn>MuOfl%z((ePm?d0``I^q75u zrL{Xe*{0nrf8>+^TQe(@DNo*eU$W!a25YI_CQ8;@^qg(a;sWV*GYnd2acq%!m9%Q1 zy?h&v#-}D)7*1~cVPPhRAz%d}_r--MnzkHTbG#D<~FXazbflsm}neV$mWIR<-TD`7U+_ZZ3hujfrfg&AP-pm z=mB*XzWg_g73KnF$y&F7;EdrLDh%)1)i3ZePM?qe>!5S6{tqce|Nn<+SXo#(|6d1P z$KG-OFWP60A!!lulCUOUJmzBpnb7+A)*iC>>=pFoUVVA5RO4jz7A;N0rlFxq5vFts zhU}7u>V{vPRb&S46yJScJiV#C*3cW9tl`&Bzo$)<84lka-Vv5+O=w80hXMZa0O&^k z9T79`0Vr6E{U4kTSc27za?u-u`ynrYvC9R+&w?W>oV$e5e}*Fhx+@9~cokGU`lWC& z3*A~1ed6hcJy6>Q8wOtwn<%g?#j^=U7ody3BIv|_2Q(7iRRwpNh?voC7$gl~A7ajx z=-5B)D3Em6i(*cO{hHssjwAN48IXV}fAAC8URTH5Z z)B)?{%QsLdYVmvd8u~#rz|tqMFsShZdXhez#QM-;(Gp>e$vSf-0M%1*UCD+3`J_a- z^}7<))dNvu=!BVx+yHSoBDgr9DVvJqun_oH;|#pebf?v$^C&KU)K|j8khT5;FXjU5 zq$h=PH(=P4{Z->H=a3?)Tr?=7OxRmfrFxjv!q^9`$XNn}Rw(nBYu}LdEd<`}Ef|5u;JqWo zP^j8#+~13HD?M2M<@>TmUVc7tCHJ>URa7(3Y;`Md`%y_@N7uC(hNfA8ie&rmY9LOmKi#Ie~%mnjqcV;>=y^L@Y3iKR195H{6q zv;g)6+aLq2u=T-CILO{QOS~$uQfd(b2pAz|8C#ZcJnHw+meXDVZi1a4`dncA56eoZ zk+jg9`8$*CQi|4I9{7bS8y;D=IXDDtf-?+jtOIa3UEE&)qU zy(6^z+8r^&^^dF(AnWm{NFi={Bpq293+7!vc`9XsF!Q}JTm_G@Sc>%*2b@J<44FZa z-(JGe$mNynOUNbOt_`?Pha)=(t&RcgQE@Pjtq`Xp-rHyX_|5||Lgvj;CZk5XRnBT6 z5_jYe4#Tx8Cq6&L7Gm({!`1#X@zcph^W;vm%l2q(N&v1oFym@HbI%X%$haMw4D%eJ zw`GGKzOX};A!aE-#oV6VQ=uBI8WysxS7tOK;}w|NFG&=cc<1_Su*$lzt}9b~w8=77 zwXtvEl>|#7!fg>@SW++oS_F}gremgH!(8sHaB2|)?zYGdjt!hy04x666BuYQq-Nn! z(wOc3n$Pea2!?t1lVPUZyD);6ee};gRXfT!azFU@tsNYW7$4fQc~&U>zsOY_bckAR zQY(&69{s3+fglt+{an7v0X?SkqFT4kiB)ebnDygzh1$*@o}62OcMZM1>3>Bmb+!!^ zauIHE29y`hDGWI$Ob{sI5+IG-tlt!7gfiCi{qI)rfKp4XMT{5(6@Es{QWy_u3ggZx zEqCJ=QM2bbz2fx=eRyWSVtrN!A9b=?@~~7Gw~`j8Wor4vK*WDnvrmZ$e?H%;>g(l+ z&6Tfx&UK~%P3qR-uh+kuPl+$i9AaES+1qVh4kSGaK6q-L8Np+X;xhpXz>per92`Ftr; z{=`wHvxG+$|I*tgefUrBOcS}52J&^q1_CBU*+st7&JA82bXS{PBDv9t10nqzJU6Kp zRri;5$MSy00Z)eT>n@ss-5OLu+`SvkaD1QKaXW>ehcsAF)$QV*vig;eAN&$5hI=J`u|EU~`KkaE1Hx^h07TGI;sB<%5kkd7F4ezj?%_NN$s+vXCQ=`)@I{$SOL zl>WrcP)elfR~s$dvr9LTqmKHS-NaLYQ|794O&rYZgQmJ0SF}X>?1EmylVmG7RTfnv z3o5LG>6etq1vGk^S3m!)8F`q(6kXJ}9-3%0Jn`;(=CDBE3hgY3oV~HsS9ckS34w`D zy9y^kTwt8E^%M$_BSX!?^JUdUUG1^_vmrept2N9(#EEa$m9-1&w@4O+OlUhh1LPAw zMKJmK^$k{%X6`RFK3>YZq(_&+idtFb=@}dJ+}__ZsPy~con`62Fr3?;AK59J;m&_- zNudCIdWJ-Ki|tWzos#BnEG9n{=(QJmE_vZ<$aXzHG) zv(~u-z~O$$^Wh-7|M^)_tU_ke3$$I}DA%vuvy>GGv(RX*ADu3B2&3G-AdlGUDXt`8 zp62UikUaE~)o|PH)f?QJ#9wcp(L1*DweD;84|blN&41hr&{uq!x3OOvytuiwt6P66 z>GZeDt6IZNO8kHfO}fBOCsM>4oI!C?)IsgSwWS!Zq?tOpS$p+HA7M4P1w!5V4C-QW z`=w8;P}RqD#5O0gFPFmdX9T_4jm1BIPb;y^kF?-cIM-9zvbZQ<3v@k-d)Ir~NoqRu zh4}nZGGHKZYZ}4SV;nFW1_(5(4xH^qy0-f9_lvGa@D}EB7s%ccEuHB{WB1uGuE1@u z|BWm=#jm?VJe$3cVB$wC(<^eLL(hP%D4ZaKCA;T%9Jrc&z@TXeNB-Yk#LbneoCd;Z zXyZZCf#Bx(pAIHE3p>aEmlS6uVP$9MW@30C)P4 z0m76g*v7a7rALvT;#C#ZWyXLzpD}B-3C@V67ea z3}9*`_yH_48zWe1=Si)I_4Ir=>znyHSZj-$5gv2tj`q4gW1nFc5$- zlK9B>ya9_uR#5MiEDD?@*OzYWFrSq@|An+hw9~-s2?Brwa-lC5o$i)9{?pB?7h>b< z7^ipD`7^^^Fc^}XZmWqHg3qyJRh0yg7QXfhH-vq(o!r)c>qn4H02T_YZfy&xjsb9x z5MtREmtRgJU+d;y#2J7rHhAZDdUbMGYW(;YP@NVZlNj_S=zYK4Vh>U8TFT+?H_4a|$+B}T)~>%{LPTQRpl$=V$Bfyw@Dw+0&mzG{P@K9_qv zV6N6uSL+HP6ZK(u(GK_NxQ?CoJpc}@5d@|$KY{zD^ua*`OiTb1*9znxz~f=|a4Kr@4wqYMrlya*tweSKZjeeEVV)z_;q*F&p(c#3{;jg74J zja_!{e^QBIyhze6eo5Ac1OVKr_ug#wLorKtTX@PHR9xx^T%k?vR z6sWxpiyh2mLlByt82uFNde&yIYXtwn4aUv==>op(b0g@zjsGRd5TLB8qAv?5hdu#A zhczV5_!x2JA(Pl(2)r;oBU@V|NI9o|&{`|2;6xMGg?8|6AGT6p(TJQR!_Sxje-aC^ zdl1b~zAQ<3R=}hC0J$#_0x*gAZZzxwiEBj4FX4-80gSjsAImR$V+(I8jt>|! zXl95n@NMz54E>80@% zaBa5)h#q+se^C#R4+WcPY<9l33V8lXFH&&x)yr zcOSFDB7WqF$(8y8oQYfL%ObE!#Qa!?4(GnnCh0##+y4O)1#gAs3@~HGMzO7E^IPzn z?r@K$yq|MwZ!p6xk`rlCg+#mN8QT3DMiolC2gCvwuNsUqjCT_wjT@?w}~4dy$}`39tA=V@<&!_iJP5k#292{fEVTO{>$nx9Z9RXC9H zhfVnC*BI|~0*CmknMc*#Qdk!BOsSqa4VA|-pKweMP2|6&FR2LV!o!S@0*C$yOivP8 z8NLAs-^e}ODMiU?rR-l%F1TzFosKlL&{0}lEakZhn479zdaG$1-K18}4>5MXGg-X~ zcYl6}$Nkcg6z8xO?5@RSBJ(0^Cwso|-QsMSo*+u-im%|5-0H-fNbNgUa^YDJmJF|7Erx#JV=exd#(kDG<)@phkcJM4!P>nh^87ew+j+U!)P1c zrI(pSe_D)ChEK5m;gZF6VbJujspFJ^9X!>v$QSk>M%QtpcM*zqWUW=%Y5g})<`f41 z6zwPxg!KX2upqyi+vGp*yngI*Y+Ya(#E5VeN{@2sAr10Z0NAjL)}`@8u;kDcjXwZk zCQaBsvXvBcTxboHZD69`CIl3FsHwZB!a7ErHJG{-`2(N$d-a89i-}_Ceq2+~!BDme zcz$rj(v!}%{h4Bduckc;H%+PFxxo8PK9XO1Wm(cuqshVea+G`6X{aJEPD!y?2wZ0V zA;~WE@mJL5#oOC#Xv?fL82z(i$*l|odj-G})z>VSaVUF)e#y)U}GaNQOno_(`u|3eolSVxDh-TU1d)(F(h*Ymq zmPcBo(}#sp{Qm$eK-9lANda{xO%UF^yGM!!Vf<~`h^MO$#1G|Df8$NR{#b9tf%BSs z@P7iOf}CoFASifV!QX2Nja2V!>vG5(g)bZmWJMNC;-~nSa9YPst;2x*D|$lmXvhd~2~2w9PG^ZY^kv zpUgy`7jO0L6Qk(+b)J>5f73}-W(MJrCJgvDa-WEfEcLz!sn*Q%J)*HQ&=$J`9}?3M z+;ETWz4o%Bwmj@OKF2VlGHc)IF%{+Ec0*a-ag767HI_gmX98t?ASVs~8wbPaGYWKARQisnT;FegJn zjy!rx_%8d=1ZXBxYT=CUp5c+yPFwM*neimd=k+_f!YIatKO)yR{Ew67uN(!+ox@Y>YX@V zXRF+n-Y0e4mX)i~yT)AX=LAV8HM&47kv5`Liq+k)iRpx$aJ-3(7Srbw$-d>mFdX5n z9Maz%$%M;-Zl2)B0no0@jw(__jZv>JH98#P<=k#FH$QL&$VO^E2YutSK7XU2Inz5L zalOW8&=g!U`?eekCML)-Y4N!mBCo74M~<`FlEPtNR(!Z*1v7uLZaZ!4aI>| zolY+iTcl4M)259bZLN z2bWQ-VrRs;b?wXnrrS6Vcn3tkdX;aKP_ydHsYZyQa}M^ow_kvnGpSE#KiZg%I;trDt$_ym&D9(FXe8Pa0>wENAz(oNK5hxfVQ zb!WP0&86Fo1L<7tF`ln$il`-sy#lbMLe$cP|7H^KBg=C?!AV8cV&toO`UNu7<)@NC z8p=q>&jjT5r+=^;m}D~|Na`P4+!Sj!Iy$}!L;aXlyu>Rd3mD60E7AE?BkPNUTAQ+b zRHDokl;#(UCc;x6PjGnL|8-UCH*Sx+@PrayC4+b2ox_;{+hz9*f?*mK!^UK;xqTa7 zq%RE?&ho%Zo}_r(Wu~S;mGZ4}DiQ>9%|>S)z7ndS34aSUvI-s6K!OUo0;+Nm7I%94 zM{7>Vw$|;phwmDb83K1uVF7T1y)#I^Ur;v<@=%1*oGW~ZA@#a4l7=f}74F#!gymBV zKyIkWw&Kl96h6qu>4$)AmN*QWBQ)GSk3g38kBcSaag4NY0l9s)Iuy^XKGEW%jLSb5 zN}{Y(@#Tfk; z1`K`sc+C~kK)&yI>JxdEpPwK?%s9iiNavWgqJQ<1ta!s71M{i-YsjYf04s7Py4;&e z%*QM9DI5+bI4p1ZIJkV=BTg7m@v;OHOS|RurhOii6b$lcFsMxj_7ITf49$+4_bXg% zV{3>!X?1tkt1fg9D* za(@Tc;t|8R+hb3{LTPM;6to4ek3KQEwjReYwGY{b_Bs_o<#aBqABonaU&;u$*J zoPK_pV{etP4aW`+W^%}-`@1rzPupjq^tB3jLj_mAas?+*yBL<*KmJPVJIrnePWB~; zj$$d}e9GTG8`9%fJbe@*$8z88n6;_tr+>|fzGdEktEaV1SPZ&hW1;Jt^p468-$ucc z5OFvy`)XCix`u`Y>DWOD4QqD2pU-T&TnahQtOhy8YE&P?Rn`$|K1RV0lklR9a52>! z&Ez$MCd%jUOf6W^=9*4qeM5%U3Gl}RkbgF<3C(at>eF*w#}Lu<(^jxkfiYQl0e_^C zUdh@C#)8ZBsmy)G;l&;GjLO%Wy0i5f^wM7)b0yDY+OjcD0vl(5`Z=BV)jW!i zc$|k>k?8JRrT-F}8EOEQC>glVU0Ktn@{pY2u3IqAm{S)Sdnwe+cA}*MqJf+-Pvy>f zBvNdSzT2THpP`3i%6d(tl3# zPf0NTAZH+=1GVA;rJxa+cRD-5=gRZt&-h^aW_cffRF3@2W@6i+!at3IzHcj<>+lR@ z&LQWhGe!AH@Ib4t?U6ygykE~LA#MdTo7jp8-JZY-xJ`5=!&lG!YH5qcuYZ3~tt+vc zkibuem~NxqZR$qTtoX}ae#;BuXF9B>ie^A1Qiz5xoKl}))>h@dO%;nwt*7$|_Z@+4 zNKU0UbnKk6f%&wz_NYh*!`Gsq(Xhmyg0aR;C>bJ;vZE)SaWOrQY`1U3TUWg(wV3y z>Mae+K54z(+#CCrnwyGoYrN7be(%m#F3vV=qeNUvp*OEbta6&F_k*R?Rkg<@u?0-g-!{(v}f%@>4 zK&zGq8$5eBuzdCCn1$jR^xHi7k&X(-EyGaDPxor0!QN%8(>=>+(0$xO83M5sW73`_@OIukMxeTYN3 zZ;m{=fR(R)uJEr{aDOl0-##}{MPkSkj4&BYFMv~II1E+8%%!L7Z*^57+fBSL_(MkU z1|u^UCG;`tqD5h8rfK%w<08c<-4ecZ$&Dxan=_R>FU(KE**N)6h?R^qkpliwq2MAF zA5`!*XPw8jbt1Qp_{BSY3{#T$K9nCK6x->0uqV1BYaTj1!{_l_S{kEKqb44jDCQv zO3*I138D%|<|WoW3#O>d1(%rV)H4lRtW+qcgt+zWa?pa+Dy8yoge928VhhG-vIn9H z3sK>~#Dva_Pk&Oi94YQ)l!i8+N?veusvnrea<%QyfW4~5A|kR^Qu7{*InG~9+)OSi za%dA@^5^poqkG#bhMZ4nBm!v7j(_}?@BgOIBWn88*GlcGF(M#&4voSP$e*Oc-UtVFJns!h5*^PIS99;JRUhWsqSlcf z4W!g?ud`_qwYMClD4b$nV(2I<8^pbh)en}Rm&-OjCc1%M-kw*|Wj6+T4xiSLFhSx_ zk>ZikR)03#@^WitbT7-yr<)eYm%Pe-*gGkIPR$-b!zksd?m^Oi^paoL{}6onN#{b= zB>jdNu~x>E8&_(Ko5-PhMG`@7TS+cw;S5r_i(o9iDgOh!L9+zw&~YrQc?1@+OV)a3 zBhPsbHrBz$J;jd+j*HvPC_aXAg!%I1ecdut_E(Hdc6aGdN?VMBf$sR)?f} zQbk^Kk=4Cv{YvEU_+-D*4no+Ko*RI*+~cGSE7&V;Js5`D@;yL!!M=x@K@>8<1PlZ< zI)Bnn@7ILfI`RV1wv{<03 zC0l>C?ey#|MR+e4%K@Xl+3A%@oIx%b4S#QPAH!yNHg_j0PxD#840c(JVW2x|At#JU z=JRLm&t6USRUJ(wY`m2xo_KJP-)U4j&B+#*J>tv=(Lt;9niy`{U)|eq>FP?vJHVp4 zi6OrSQ)q;61ZEeC1ZUGFV9TA+q8RyuOV?)Z&+8%CF&MdmCxc>0_`=sm!c^fV|Z?0uBWPs1|E=1wv<~?DA12Vo!ZA8 zr+344FW`nVwb`9LT$e}MdSBP5iho~qEZ>EPi{Oox&O{`T^|bUrn9G6Y>THf!iwZA=w;UHHE%FoF^OMPhoyfXvh>nT?$;q5> z5M;YHWKwFf=vsnWgi`n=O;CXWV@QG5=O?BBwVCBb=#qz(Diam&>K{1kPk$X2kTz9S zLw$bX_uFN5D;24Nv5@*b#iCxfzi2FWICfV=anL!5@WZBUONLDpa;q7IHHIw z2nZ$jUCy$%?9ek_YK!goi4pi?3STdGBQc6J<*Qfyo5&b8yucFez(}zZ5guLPhm8!7 z@@oj)hm0}x`TZ1JaR!<>h<^i^nH~6j-gbV%vLEi-o&yZoY-2R`+h;O42`h03TGN)zDJ`Jll)!8PsCHq4rL>llFsypZV3yFO`0UhjR;eg+7}F;@Lb_?SzNhzu{9MCR3_{&fPeJ<=^xt3zfGc- z@`=i)q^%XV2iDqxib1pE!qz1wAIr`N5=e=S>T8col@Fy9cV>m+zzd-}GMqb-8!`Mr z{}9{chsJToec$*lKzogqBkFlRq%XudiYOVU>KI>_)8csqTAi~kyo%#~Pe8$fHhQI+ z@3stT`XlSryugm7H-FMDLlxG38bQsFu9#xi=Z{a_Sb8<^Ek7Fyh-j&>whp6iY19?K zUl#10`f4-<;L0#6?(=>sEY|BUiK2FpfrS_x3iN58xuSienF_~1a`C%%t;hvWh=aHD z;qoDnqeicNb>p+_L-fBcI$2iV+0CN4o!3nYd=K#)Fwf?0p%-hY~7<1TxS%A5>t{E>}MmD8T~DNFh+d^XbA)Yc(Z~_ zIYWk*KUvA9*N+(upHTd`5f_i^+{P2SxKp%<*KSBB?VK_HVexR|0h?A*ky%A@Sd)1w zCGJvY$*}uI!hhdPBSp?u5arYFch`u^Hq&YI)H~LPBye4`1B-$gC|qT-JbZ7^5??y! zu74THBih`UH+_NeliSZ~9|7vM$GKi`jd%k_{ru-Gd@~M;U#PNB6MO-FyQv-)9$*dK zVi6w2q_5OpZ7&|-J(Mv~=)U~pFRd6e^ zWM2_`EOXGps$Isod6~6oJ=r!C6!j8i2jgT1MdZ43nKhj2t4jXJ9)yN3so80Z^TC{A zAyo}|?GaYBY#)4;0%h6?rCS$iR8G6Kf(Mb;*(R``;Kgw9?uR7en=l)7?I=Sz&shZY zN|{u+RDUI!xgd-n=Ur>eXO%P=uv$C6Ok1~X4avt-8oul)$la3UkUZFG*k&QfW?EA? z0_KpRw`Rd6C7PI8Q^p<#c1cu6Y&BaMA@)nBy6Hgyk;^G-q)8Rs*x034kmb z&u!H2gRO`AjX&sa1anKH_lWtf_SiA>#FSMlXMZ0@HgBC5w3MFpHPjtZgoo!-r4APM z%`%X?1q)InIOLX&iK6%f9;Bh1&#-!Kg-S&J$&Z~2FN z_&-VTU$LcpKHq%(j{lMN!z2)F=8%)kgMTb9@`K#SCbt%@+!emV_NCL8CG%LD7;XJd$lx+w@UHE+E=!J8-HdQ z<5A@m5MHQK1SZf;6KA@1D4w!A&R4L8t{+tpSMGlr_5@rQd&Js*s)PIU{Oyc|jB_~^_8Ob$Rw7tx5GQ+_A9B{*MzAOlBFoPP|>10Jyz z^lUa5&#lgiVRwmnTx8p23&vrkopdJm1k{&`++aBTggCRBpkV9FiSdT#N13J zD*&V#&c51>PUVHnXnz&VaitzAUSF3@)eh@!v@+6SWFyo>YTcFH!8lm&C;ml!Z5WLA ztI%StEF988*E5wN4A6Om1*gKgLQod-;bL|mLjAWG6EXr?k|G<3vXFI}V%+5ATxth( z$Ee%pGMm)6_H2lRjnX1ExK((KxrcoQ0boKgc<>>yao1pRAAj~^P(A^iIk=jrEA7np z-+BSGH_)qPat7-9!#2){{^B|1SxON(SE25Lgm!yr*vPgp8yE$V2I!X_*Vt?%U8PzC zC>IkWIY#{nXMeN#P0#nwP&8xU3-MqS_GB@DfUD0d5!9ZN{L{{5#E)Mu@ur3Z_abjLV0(_g;AnOAvUsN#I!2S63hDi zr=0|lrP6M7<;$m$4htZSJF?Wc@x(-?RXm|Vu{A-xX--XT9hM4BVP~8HbKVTf!q4sS z$?-Us(0}KKOG{&h=hZP@>M2Tn)JOHi>r0GK^buWLfl+@mZDn>!gf*8qo-^3AtAVl9 zrPmA{?VC65E zu&P&h;U0-v7%Gm@wsjLB6V!%@%H$Cn4u;Ss#(%4168gW(HS+w6Z116jp(Qk!3#$yg^v~}rH-}Yd@u@;BP7w%Oop=>V_@ap*bwJ4LG`Qld5ML5A zHRkph&^{dqUD#!QqP)jfV&eVH+M#rG%_}ILf49zkg=&{BzZ$%9OqV?=roiY9_1|XKX%|*I{ zFN~d^%h%**ja$AsU3`X`#%21od>_5?-KofQi$-d&aitMsCVP{IPHE8HDY0!5M43<; zEHJxt6~R?OLR(ZeTkg)4TO?sC!I0-wG_rtgTe#KLu$x(*Jo;2pn6xTod^R%y z$DHWot5`{8@axeNEk4;&GP#rk@M@+7he}(7flfGdiV%3gLx+UZV-X89f3vAGN`L4L zavGr^wSL@hzrvq1j`<$YUfo_a%NDE&DJ>nOQt*v9Dg*J3bs*C;IcAps1i=i^3U|-{ ziG^SdwNf#H$jSYxtux(|pF56$DylbmnM+jiH(GO`a9PSB3J8yqD~mA1dfWBWwI;5$ z4qpUz@zOVqOn`(ozAj#g6JtI_gntIQ9HPkBODNr?zt>c=+rg9D#^^ol*($!?v04cQ z#D!#NWDw!mp_&>1(#))!sX-U5*QLqHNUVf>zuQx$k9vGWnbtqVvkGGpgTOVN7N7N-u?^wd4s#Q^Oy7bdk?cD@F3Z(<$m0MnAu^%R z&5^m+hje#tTk$4?XjRwZ|QY{&!4dPG#0FEhlaSQoA8`b2^A zJ&)M=NJ#--u=h+nX*B2_4zqnbEu9Ev!d&x~DnbIXVd zp8QCA9hIW&*;E0AsQ>|r-;!06wjNw{v7>9+>1|e}c_IfdF($QwS%0vY#7e;gc^ozyAt9zqwj?nna_>nHXV0A@FO4)Fg9|yX?5m^un>o2TzN@?FJ82DAF*L8&snM zy(IDM7K@%}7Kra6K;HeE&C;Y$OosW>8D+a_maExfZ4(Rrk$-t3Iq{-DMv4M5j(3KG z;Srw>TtPhSaHs_V`^$YzUbUyGMltAdkXF0&?eQ|Nu9rHF{^aWN*|59%(Tm+yvs%Du z>I41eOdWRGsE5xm@Pt7eRQ_~&@~Gs4N1d#c34EQy|mnPQqC%b_`E+|v#f|3efvjwSzb3I9D*+ocQrOG z;k6#R>?=mk9qV7d2nmu28QzgNnqB<$|A>#>3$tKqtO7OKCmb1(x;W+ONL0zYOm;{S z1N2hvDEtJWGO{+2?mZsH`oNsxt>-#d{)*xB!MtLIT7TAq0GzT#`eMzi^T-mhz@aG+ zqJE*)_KvqNn(J@LBsYcl)~X`v$XOD+{Mgph;G%m!5w9(c^Ga)^2x044YnPLqsaWjo zQoF@5xkKw-KNnVVt>)*=pvSq( zyNL=j*MF1%V|nMW<&gfzN1s)-9Q3xfIR=`8xF20Y`Maxc8K>2eY;g{s9-V?Esj z#jIf$l?VGS)>~-f_SE)jxaDWnro_)35l-TK&VRX?D7+Hz0#{*c>|JNnN+I?!V4&lD z(xR%UC)mGC$!YUw1`hd!t42(qJ)DXZ=n!A2=tQn(Iy0QP>Eg3rE$77Qd3kO+;w0<@uZ-dJsr$rRSIuD*7UoMCWc!P7IYyXR3 z9Dd?h86Pk=e@_nV69lIYTdDR9N~{=pvK&n!A8x1iqs0&XHk3bUpSuQF zmVReC(P1sF?dn2f&y*TthVmy5u|nr;6U2^|&G?48QhddWIIz>^{9M9Df&+X-9Dm3w z5`V4fP60}OvjdBCb~tC25$R}HmNHUdD)0jiWO5!?^huX;*=IB$ggA0p;gT}YU8+Iq1;Mx$SsxZT z=Av14t)!~=^9+K6v(R7ivfw$HEzYU7zxjIE=RuOC0?Kl{-GH-D=am~k12 zX8FhL)a8pgu~TYQt?r%?jCdP&k#eiDYCcfnl<|$A*uU^E{M?b;cyS1l(N{j;P-zA? ztw(woBiYC9AJNN7riMS6Ws4BCxRe@%9;V92d?FgB>#^&&5dWZX6tc1Y%U-Ufw=3|P zbBZWx!%j+F7?xM7*R{xOMSq*g+{lE9C*q22lznvsVq`ll&BY}YdAwktg`*oCA zoub?kS#>*b;Y`8jb805+`MskJZ@QJlO`5L93U{P7GeDdx@$NDej-Z_8vz68tAOri& zLH9tNESGSmj^pIuq<=S;Bj#Q<&g0Xbpk%L&x>)uY>Min{^(ikFG@?2?M`cc_I*iag z73iKYiLjITp<0O1!MNtwU=wSsS8<*<#%c@Li{Tlj3?6? zvQW=b;e~tTBffR-9!2Hv9sl>4|J1Zl8dZkVq`%2`NBf)N3suRG`PXO zd!NqL>2GyN^?!>FxQ;rz;gTZ2tv8C}J2XhP6~Cd!w!}8dv`p-5WL{OWsjAf68G#;E z7Iut6fH<&jIq=^zE4q(N*~O>bmc8w?hwV_ND#|*)-PluIBPj_G!Y2w`q}Q<7O-~?m z=4+x%USE9HJ*$_dS^w?8aS{|_}5KkiBk~Z6E`PHzvU~iUzY(o5&t0DbtJ{tGDI|4S zyVmF`sn|yEGr0WxDPC!SG=n2s#ok+#@S$eMb~WFOzOaEKD9c5Bai5eJPb1j9CDOl# z$ea%RfbZc0gvue5kiJrA3X#|=@L1*k#b|E5R}bI!6;?d-0T%_TjP@{ic^RBi$%3?e zz<=_AdG)KRj92)w)!(3GW3D1cD(DznQdAQ2gj>JWy@8aVwQKU*G#(J9y5w1Xe z))YjAJ_3$@vDa*XHI~VBCg}jt{6S>rL4V<|Bq<2YFX_B^K-c|_oFeZFky{jr1$X&H zzdZ8;-7&AzoQM+;(TR|PWGs&IFPA(q-UN0FWcqnt!^k2re_m zW8dV|Eye9Dfuk1kTVR`U*te?ccmp{!Y8{7H4(XBr*fwqG?iF@(i-*XgnadV>R+FUg$Z7^KiaM>^eg>fKkphhQ&C$6Y$+cFf z%B63V19&%HG>bA(qCT2^NPl_oJ`RPq2mv#j9Kx9FS}7Te%2A&)hq)#{Q&r~MrzwMU zgWyY!8fBeHdDWm6^eFK*&u^IgWDOTFKWEl(PvpP{IwyJ$2mwEreYJ9~kyP&@gH#UB&q_){Iu4erV}Iw9E|5eEcLd;w z(Ar(T%5IgKnC?6|dLI?X89fT(M7fid(m4S(c3BJ~a|=_mqA3TO_|8@bes7Rgw79SJ z)m`1;`OH;K^_bYHhGqPS9RCs{V=^9snp`UGHXT-PD*x!MS}Kiu*w7Vk6BrlHO&UIb zRV=3*#FSbRa?A@MK!08+9FJ_gZV(<X4#HD-lrP+~r(R>n64l7mT`AXqEB zB^bC@VCzbuDIlUxw3+CdI6^aCF&pk4t}}4f78f??vPAB-pmLDzr{{Lb{Bd0jGttw& ziD!katuC^3&G3z0m{`E_OOdnPnw^Z5MtJ1?p47Lc5=Ou^)_>AQnGr&qxwIRWe2S=- zWrtk!sbj-(tC!`(bWkuRZUl`Y3A<6q(5dAhn!|2Ygv!P6>`Y0bBBQXpspMk{MupNf z8mN!a>!#l4bo(eX&%NUc;8MNjcMkcw)^@(rEvRF7(K3lbhZF~&JhJow7W2h&DAZEfNDdw`GA+o%f#@g+jV_?M>^1} z`?DnpMrx)1=;;2*ArK~i_Id;32gl|Yb}Ivzsg!J#KvrzF(qor3H!!z`#CzFGdgS}u z?^=3#NR)A|d5YEegCc%-CT<96T-&UqIS0&7{pYMX0A{6Nt+Uac_L3Y z-6R+bYUT{9bX!Yj#8GtTF;gB*dZD81oJLy;#eRPmxqGwzu{OV^qBwQCQp%8RIQ4Z# ztyIt}*ng6gj;uMiFiE-(Wr9#{t}sAiz@OH9Z|rMMwtom2T?slyUooX-I--)j9fP6^ zcz^WR!xa+=+WUxYxI5i{4@EWLdjzi;SGtN?;`c>9dGn()z)Sp0ge8-BPTsz)#9SXQ z(-eeI)otZ!P?T;+N&+QDMI|JmFh`W`Bq-NG-C?AaC8bXe?e;uXZb5Wg0d^Le;#2T+ z^W~G&riYMz+^Fl3&etyDLzUj*hZ>Lh^M59)nihx;#Y`tbtkmFoEcD^o{jEY9DSVZp z$|7p|3et0iv`rTC#k%{1Ykr2GK;uAg`!;>fV^V`n9J4P<$Xc$I!>`^b@?RFk4qIcd zCiVpyp>Ca{SQBOJHD7jm!U;cofKV2`)BV|`Rn>I+LCrhcRyu-kDqu#3?4 z_~Lq&>`V+cq};Ydk3C(e@*{Xs(=SiXi1bD=clJVq@(-mG+p(W|!uEU)Xr1gV?ZHGF zH-fxX9h84U)Ut({JpFn!<#+vg!+&q+3XAk`=YihlUAnaq*G-Nny=Ax@Ru?wiqQ_eO z(yCgaX%d86jZ|!}Q}7rvQ7PkX(E%nQiXVczZ2%O^(Pzt4{9!&JM9~}7 zVU?G?77`QZLP#99(0>{k@!*u2<`r}Js*P3I!5CfHT2b zuQgei$R6_pAsDS@v%Oza+EL0NdYvw!RP4h!O%87On*@*gyv%(T@>cHHjR>O0I9UUwQk#+)7dEb?jD+E&^CcY zZc*t><;-kTDS9>o=U1_>&;p!%kagL8l}60qEDBq^otE{}GP3hgw#RDy+LNrm9b94{ zzLtEPjk3U?tsO5yUQ-nTTA{CN+=3!-#LL06aTEKrN@HOP$o8%~nt$SbanK~b!l^_Q zQu+gY21`Qwlx>#kYEfv3!K`}H$1-kNSj1)bC}bgh6&&Ujn5D4gkrFeeEuSHAZal-w z!!bT$#lvE02*fp}SU=`9V+t3gQ|;pLUv#9x8fbX{U#{8HKVj<{9yofkddjt6=?kC7 zlnM^b$p?)JKw*AYl7E~i5)a59KDMYnF9g@D!>CMKDk)E@UDtge#atVZfV&By>OFFk zWWRlI;JtT6mH&V}Ev`uijPrVo6k2Pd1A|jZlFPqfwj4)l^LFdQ0$9g9J|mXEj#$T}vr{#MuCO);qpp$TV5nz0zhU()1(0N&<% z6VwPLow_~Vw{DA7*VoZ4y9Ys?XwplsvV1%4{zN1 z5l%QENzZ3I>XLVjMd!nJBIX1?l;i_@@XLBGyI+apl?3h;hQwl<+X3g-X=i2%cJRYA zM_7h2fO~Lsv`Z?6G~dY`-2D1CaMR1v6x`8t^GG|^$cU9&@;qlmktr-|IV&Tb^5>Np?wqEK>jQGhHL(C8kXrl% zhx5V10)KYIj=s%TOH+glhNlX-;d&IzZFuuR-vT`CY5CL1Fjwa+hUvsZ1CgJT&Y@Mg zh=un_Noow(^mRn6T;0uDSFo2Y&x%Ecn_!|0_n%O0xP9>m&*vzW?ql-&!PpHwR=YDC zQXipUZ%1&_Yev$CJ$*&VPqj0hwEPd0i{O#V!he4I?kxNLqCZ6E^y?*5_tGd(oK-U) z?d2lWT&zNwG$6Cmc}ZTpUpaYzqa_sW36Qfkzs_#FR4F+g*GEPo9=AM7JGOxW{XXiE z9O?Anib9`Q;;Nl-ePYoqN*3#arze_tVrW!YRWuMmIUv)GeVzHZXn&XEq$fAuByz?w z^naN^Mr?YwdeAsaR>JLKGQGc*ykTkxE9BMPL4`u)bBO=jE(P-sK>Fx-pvw0Jb<>M> zGB+$ZC)Ul3hvmGVSEPswvC>EkFvlF~Jjs`)ivsI`Qfa9XH9>>pSeb)^o5#DCt{ zWKur`LHj(aeNELTs+LlcHwnYDQ{OUH3iH5EN`gXj5SdAdSOV3Hz%LK-q#Q2ylFl>wZd#{s*U+{8!wy?l%Tc+W!+7~~+RZRvb$4J6m@=w}IH%Q^<|m^9;w{f%VN115W7qW}pMntI1@ELtJ$u8_3x6R$0bA0# zYnx1c6z5N-%g7eaR9B9cM0(HW*Ys)4Y2d!`c+2zhk8RwVRCOP` z*zqw4!KpDzZdiRdwVB~W%=cdrVk#Fgly&XQR zMeusm+{TTy>>G&Sn~O58w1Ce-*zjq*yqF*|1c0^y-a_O({1wR$rGMWbrL4!Eq|36^ zy&@gJS09aYeVP&$%fp>%wJ^DTfBZZ>@k9tPwaCX!Zbb^chSc3o2r=4DMbh3gs;CNG z(WPCc5$%bQ8GWUJ@PW=1&?~`&g!YCWisNg=ia(|0LJGx^=rFSXu*VGHNnHLT4$F0G zbam(nAliO{H~PzV%YUElFpaTjXjSeQYTq$ z$)tC$pK88;rWp0+pW?^fSkPtJ+CVeXx zB4@xkvBNy5Wq*jMfcs4~hktqx21c*Ty+5)&VG3G`5Y#wryLDbz(G)ZGYRg)u6GH6P?($&Bo@;JluJ@ zU$NK2T3dMp?9uTO9J5l^GR(s~uX6(4&wK2NGJ0N};F1@~i}pj1BrdgdXs_9K(1CoV z_ISN3pwEJy%EAW#zmyYRR{X8ELj$Jy<%)MYtEW3>lOihql>FF6%W9|Um$OPTn4t3R!Gd--LVa|X-76!G9BDs24(id|(<$rhp^n5faO%x)FaEuLILd z=yE@N{Yo*xEtAb{rGNifB@2Fc2LD;u22-P|cKkur0d`lfbVXpym6DW^A1^kD2e?mP zp%BxOuZyBz+DSh)Qe?Ve1iPcu*G`VfXx5e=S96fOGO!QZHktPPq)SJ}TwZ$IYJV|W z*2?%s3a5j-)bV$Dh1IE)oPz0)lkV4!0<{REas}4smeXU zfcg3nod#zCA?kUiu>UB@o`21Y(89={6v_6Hy#PrI@0UXDN5e3f`%rqEodgoaMZ0s?(~v+k?zrsYBVAqVSkFTwc>FxFe1JjP*>mk`Cl|$ht?;5rCXeR zWkg1yzjs|qD(FLB;u5>;Gse(GS=nhV?WIgdU6`Q}A)dFx%zg4_-`%+`U<)8>)O3y@ zo0%Z?_6gnva5{snO7E&Ot$zX0MNr@%W5)@k=uI&R7Ee=eYix6WtT2S$7k{SoS$nt5 z7cPMJ2!Cj+?a?VV)oUIW{_!Hwog4WkC`v)%3q?JnApI{1$CvheF{RC)^re4f*}WQE{ca@$}=p>?Tnwb^ywah<$u<9h1OlmvYBI% z`da#~nnwU_O@Cm^7iHG3Td5E_`>k(Y2Qzb)E{j#FOiUWS=jqw{zp;|9Ae~qlA(836 zo<41^L1v&zfYi*QthnVr*f6^7R0`2jgoX*=)ZZH@V;TxH1Cv*_rdUF|U7~3+Q9etO z>qC5&IjDJ<$k#O*lYcV4-&Ut%WNswyv%$>X2*UvOa56DCeydti@LQheuKuCcVBDl# zo6&WD-<5TcBQ=+5Cy$cGB9QG_qMU(fLidf6p5fXJuUfvcM({K!oN7(C%j6|I-4c9= zm%0`gRyI&A@8EYp7GS2&lEf}S;t#dT5i-9DzWnLZJc29l0e|`O!kQI;;^KhWrfu^DzQd#VVky~}Fh^EI_FzT9pc43=KU)aa6-x{*91;AhsB zsav$LZcBNIOs+YNclIX*UpF*0+gwI~gHel?mrQUQPvITF4weZ#3CXGyr+k10+odF! zABm*H#+v>^cPURAfxDIUbQw;p@qdj(-f|y;;OuTX7s3x)o7(w9!Qk z3lBGL^t;&fIx>nv#@((%R>vBTe?&(3s})tD--Un}ZmLOhOWiC@^RZ9biTwLugOb&K z$+g79VFq|*^hB7T?FN z=NJAZK7aJvZC!+YQ*hYPw{C2wv2CldZ8WxR>zAal&BnIvq_J&VP14x6ckcb4IS*&% zJne7oSx@_EuQeYq*PI5(UsAc%uxMNLWLd+0H#ab@VR?LR48rp2ajTfunVaVS;V$3( zWR7mz;aalT_#`hk_!=j>yStfxV=z+fLco;=OFd`P_D+IfrhiWGJ zR?-h5^boW5EtBqBaF(aNt17;uK+#H5cXS}{A7ej?x)eWW`8NGROxSgdoKo{;bWdE> zL{xUIjOe`6MWsMDQ68t@ZLVri%eQu6t7Bg$QVmC_(;o*|Fau_w58qjO(9gx;i!4)-4V+L46bP3l z>dbp8l|-+qtc9#Jp>IAo2KECwE?ZPKeykVktEm|aSPFe7Ow2uCB1+rNyjyFvtqKNa zj~Ux#PQtp5gn87!yd_rm&kk|S68RVKO|}9p81#$B1X_F4St4=6M`7gPb0`}KwBhMf z7r50u&v}LA^!hfWuRq^z-HA!R6FdH7dGfW?vdw*t0%g%gSKfU80K3>=7^RfQBR#gr zf)&p#kV3CV`JusNAOs_YyO6wGf)D~wQN=S2miMqku$XJ`c?b9T6(6f93D9sD`D%!r zetyqZbty@h_8Mp1x0SYw+R{IXqD2bf+4J#>i=%t@j@?wJ*Ijm)MR$mnX&lmlMYzVO ztT9fA7vB&rdgz`~U6*6D4p)(RYL#|0=a5Uv_~Br?0vfR%1C=@xR~?#dd9wrLB1c+Z z%c@$d`u<1xyU7VJHN{XO1gW8A``Ykc zf7+G^|9t=P93i@93P%b~mx+L#( z+9IXfT`h8O$&yMcT}#fQ&f0T?Sw@1R>g@F|(`vP;_9GbQdVIu@Fh_h~$8v}K(Sm|4 z30C;)Oh1Hmm-SNiHEMuY07L<&y)Zj1=4TPLRH5X)|9IIbI3Xk6p)$+sxqwM_TaBboKgQNm5s7ly;_>srDL4mBYS+zbQ#X~)ofC|()@Xz6K@x5O2dok94 zpFX|yn8%18;|lbw_ocZ0H01l$?qykdNH4Xpd9H}6OW8D;6>4MmELxSPOa(tV44(F| z2g$Wr7?l}^b>0a@D`{KjDc7%lj)!Q*oq0HV9TS-PvkC*#U{9GS!_8Q6I z{YU*9&xf}oP3TR)w4yD8fVc0W#_FNNM_6AaSX-#72z8m%E_1uKLWR*alfW4+qpxwi zt6~>Y%U_^|3!W{>Uqw#JjE}Y?^44yNc8=s)hE6#>$bFJn!pTi1#Zc<24MdyBmn(S7 z*Njl%Fr!XUwh7Qd?}3C7xwMW>bgO$P4ZoH9Wa2-q<+uXA2L6fW_-yXPPT9pDNpkNn zBpC-s$&gLKyLFsVgxNAGcfB2jK2qPacl~J69;#p51VX+*n8t4sU=OEd!UrSTO8w!& z{A51Ara5sExYv;gL&4KV%AedLV-U5O{M~V<(D*PKRJXSY-<4;eyDg6+$Y^-b6e$XX z!FaY|OBn%HX_yC{1fV%;9PENrqTE-nkaSytU6GMnAd5}M^SRe|EWneGKjv5cQ~i;N zOV;`H_>PBbh*C0$Uh*J|bZLBxL-4$C7c^PPbwy-J*_i=^+BSRdBEEi)C*xN7*mJ+@ zn3(YG!P5x`r(eyc|za=*ZFxWsso?}SIfU!O)ew++~K?St8s7g74IAj zvff{36O0wSP`7?vu(#5Z;Q9Q!)BbeEzWsy_o~c>}h79+hgBC5V^^t-lU_W^zqZnyB za?6T-Q9)uFq{89sbSY#5$NY|8I3wU;g8Mog^0kC?wgu9kQC+QG@aij#DaJp$5RRR7 zS`83pVYIxM8@%>A9Wsah+BmjtAjpF?aigYDom)Jp3c||SSEgld7cA?eE&=|4Khtr- z&IET03J}d{byzmvN7cwx^P`MyEVE^*EuZ|FTdWx@cY|7%+B|@EnHM{oiOiPj|J^G{ zJ=&+0XKWfhAv`i!=cA{oO}rQl)gQqHsA-Wf{%AleTp zx#WK9BKVnTC{8q(*@sfFS@fg(Y=V3dzEMcm9>$hMX+dVdS=_cZb8*03Gh$PU) zzprY}OvEXdij7wZb%dlbWv&%qf&G@%Au-R?RhzsOIY4{7k&L?!Q6;Qg$B^ERZYvgN zt-ZjKKrPh0uO=fh{yY8xiasf6m%YF>Ays-jL#-n!u)n+KP~##iSXX)1)&=X#QzNpf!SVHFQEudqeFwT=vyak|H|ho0x0GyRnr~2(j-W?eIo}F%Vmj@ z%C}N#SXBsvpNvzNT+*0yzd?YqmNSxAi5wIo$fgKTx@-fj0+P+*yVHYtJIj1#o+_er zkzu&N?EcK)Gb^E3Uh0x9lN(nVI0Z}2?<GP?ST9Jv;Axe~<|%Bc5#Nk+o(T8?7R!ow9O>g~mc2Q@$-^ zr^vFOmFQu_5yX8m$Zt&jK=3XWyzjMOOffr$0{8J8p3*ZWJlDye^+H$I*Y{v;K90|9 z8K*E@l0)Q4bt5l8WwirX!_YIoo7PKwO|vo)2s`OkzD)Xy#Dw1$oT-t&GDV8DCecwL zU`yGOU6c!%_%%X-9JAMi^C^FSj&YPRaEXNO3|{?B-NZA^$IFG2Ps`=ZJg)AF&NkK| zRWT_Ii-?}oUbY(PpNd$RHg`QId+~d#p}yNb_^)MiO{~oDbC(CuU3jo@)z2P&Fxy~( zFJK_0J8W^H$|#m4lJT`i&q5%Sa2}1Cgi4()TsKQug;M@4+X_RYm>m6M7@#Mdl?r&x zknZH`USpendv{--nR3`5_?Sn!*SW}|b+9`jDM2ajwSWFjzt`@X#`h(((tO<~fnvtK zD}vo#fuyfK2GsN_1 zeUkNZ&^Qn471h;)`lYzrh83pkm%=o_S&FX6y89%^-D~>Pn;o5RHn)tDTn;^I_w;j1 zuf}W_M7~onT*#wiq|{Bc2d@D?Y&@dm*ZQ+hTUvsD9rnoR-3pJt`+l90yYU($WrTrZ~cA_x7_^y`IuvK)3(1qx=n0x?H91@KPE`6v=dGFCqF*Z=ICP!LzVwzz>HIO@~-dqCPUMEf&7TBEg z%m}82@*Ui@R)5JIj((ioUOqe@pe+7)qmo}zaeub?^Rh(AWV%xfcL8x>U9HUW`F*uNpbEx1nNYAy3(pzh2u5*tTI7FcHLMC>>iqopT z5%>GWGBY5ocePxGh`&ZdMm=!mT`jfs?X-C;Wmi2LE5~cGVIMY(19f-v0V^|792zvC*8Ur%Ecr+vk_Vy>oHqQilKr>y6n*^PL)3SJsYVeDEAY9-$pgN%7BX&VV( zcn604E{5RQ5+@Z+G%H3=HT-}Sf_C$_#>SN1vRy#-;%YIlJ%%SSdZiQzjeWVSSARBH zjpP%%Ig4V4%BgG@*$*0JS{hu0i36z^9uQJn!`s=1W9Yxr#W-jcPfP<1t~T2RI=%}+ zO7P)}MH5dfSt)sHb4uh6Y+>Zk3v&bHH3pppBwPAjowMr>?q!~e9Of!B`27`SBjr2l z?}LU!?0jhT*(+QyC8fXo!dV;M>Jch@nd;-uJ-uAcKqVi4h{dA#?gd2!f^E+#3yw%I zCna5L8|y?Ra9NK3`_f2g%6#R2gfvZ($qqp?ZmTx%;R)E`Kc#!c^2VY;EHeT4OHeJ} zW9-3(uOxNwAD;DL1eyM9L9rN5~t{Kt(6RQqgd)Ecz#R` zbk1m@yyTMoI6Jd{d@P-eYoD*=gi*CN$w%vLJ1p|+=B?cg`kryD)91<7PKtS3}_B|1rt@zA{0n zi`m?CxZb`w1HFu858jk;AW5~Tpg{8L_E*|$!`YuoEUF~C_8oH>hqdJ$511{!g!&gx z7^;>|+UeYH6vX!5gQc-P?({~{kCfJH7_|gz!8Z3{{#J)sHt()*7m#xT4#7dnPdCVa z=d2X?57SDh^-FSbL$o&x8olRoH zoRMzADh5Enkmd7hkwRb4t#))+YnZe_UPyHn$(~~_}>E5vI@Hvd2u+MZh z*9jL3Oi|}xMcrl5lkloR^ZT$QJ6)7;S%DUzp-jI`lrhkjW>6>wCT>4(y|-GG%-74g zL(X3p6*$;3XnXu`%OgF69>U*eAkQexI#r<&6FTLQ9Ql2vfgp_3*I z$i(%TWn_q~riU-!Idn#6V`eT0=B7_m;N7p*BVCe_jRfSxnFoX5JEYvS!)SNpNGA0? zp}ES*Y3^cK*crPp^~lT|`1i!QNHrAz8?19fW%pH#FnSNrDfVQ1`Wnw!RnO(U_|{G~PaJwd+8j+9NhV`{<63t|4s?7St?)c%e{ zh0}wbhdzEcuJZIcYqFX1el2E(&FR!h(&V&cnJ7B2Y9cMr)8}!)))+*E-97LGN!*IO z2Hs=RY3;{6dVb7_3>{YAir1neAJ)5vyDYoTY_ z9*r$mAq{UflA??@x}WL0(t?5CAUL?4TtfEg_Z6<8&9Q$*sHI0xiHF0QA^^dN`tKWI zCjkG((v~p7yU!yQ%GL`JO||bjtXVZD-EcOHcafm`?Vyl7C$P&(dwxo>Sp?&QnCZdT zk=PR049*Y2Ib32^;f&7w`WClBVSmOIV6o_u-sZ%q7y5vepOW~FN+_6qn+Gl{UuiNU zQ<<^QVn3`ngS$L_qc1=nn-$FH1X9TYMx8q{M^5cA8JtqSvStK{zQx?N(n7>>D2aM}I#oUJ^W&4Lh%@cu{8=GQ_XdEh4zo+XV6CkFT$UG)?;V-L!!w2!rpTr)bm2O z#LG7&eix5iOcsTYi`K^&iHSRI1@Q;rjT?g2Xnd}5lGowkke7m{H+j~T*;ozbt5d9h zsVYqO26R*PS=NvCdlI;a-* zw}yYsZ{uPn?Cep^geK0w`rdGX)7ZZ59QhGCE8~dsPK$gOLd5Kf@I+b3z0U;Ta=6Q} z2LC3dFO)CZpVNf?VnHz_SeIMqFsz^v$t)C5#L{6NqcV$8QOY&P3U&k^H2S z*`iLGE|41=N$~*=_9!BoStrT|hs3y|1hU>lIGJ0Z?{-Q=u0ro$D9p*S6n~?<kQWgVFB$v-v&{jb#{8oCcG8f6Rr-To_30HwC{7uDkyY ztqy_A%WX>Xj$CL6{EUh3zwV*j(Q1cT@)4YZA*J3FaG7?>6c{^;FHMmFu!$5vZ2yekRacraCGRblsd{H?OTXlzTyawOizY zQCIaORusmES5W$@s1#jn=;y1=_yj}Qihk$FcU!q8WR)H;df+5VAC;~tg~Y z$0naqk!;fmYDEJb&bQuTo`k-cor#lMcOtSN>83GQRO_Vy;@`@DR01==T#P*^tQL9qoJwre>^fPHKR$5w-{sQ!hN4@Ss`Yz%Mj)UyFx zw|t5(vTyAaKaSBQ>Rqb*qA9%;GR1+34>&*WXsy?k+0YPes`NSK95w=!y+6ZULUL%j zpxEi_Nt9U_Dj)p5&$*Frz^RIkIEQqu-D+1&DEkbUE~}K0^cWdxIi_5m8}w!Nbly0jr*LO)(1%%Fd4+YpnLTxV;)l+AiNKv8C?^LBqKf)1fFOJ z(Lmmx2OGc<@Bsog8Y*0ZJPL|}Ws(eM2;Lxs6r>80YY2ioKyJ*8_Z!T*=p}^hcgUOo zIV0FldIMZ|HA!kr6y9{vyN91IEG>C+@Bj48+)#B?31>>gX$s&{^l*l8{U#t()F>?T zAROV4&jJ4A02D?*83^hocpF-kE=)kogo1`e5Dyfr+;06E%84W+CPYDH45nd(3#m3} zUKXa|fei8AuLxrJu^kO~M^e)70MYLR#StA;CovBTx+=~u0ucH-qQje%7ajWwDf7PC zMHJc;aF!D;ki~bybi0CT0&MK?+ao6O#O8pQL7{k>D$p}w|fvq3>l1D%VJ03BmZv)pw(QS$$ir}4bb9Hxz zNGHdIFXxC5m%ukdn>+y9zT-RuqshWTg8+pm3>(}5|3cq?_sGgE&k`NY@?sSklnw;c zqkr!^0iym3=C?LUBt=VY%pN49FiKVk@wxnI25h)C+o2=eYg#&};g*s`Pnwo|!DWL3 zeZn_gT!Mtt^K%f)%*)_)qeIuKbeKb6#;gfd?bJQKsqA7@*!vJRUoLSAT1v!p_g{{A z^P1midojNe>TINl!|w*VrwUqz!#+$L!w&=?0U%HyhA3-_S-E|_UpG^lJvgY8e94IQ zptKSFkb_ogDDC+o?L{BNATFMUo~PG+`xtOGFiMuD*j9d~zDjp>-(qP#Lstrmi$7QK$$?LbNg)ut5o!%5Ps8U4oAd$1K4F>~E(f0+zhMk~Acz53P;7RD%qY01%)~PJr7qSr{Z#?c? z%Xf4nA(ID$&%XE;iUlp=H2717Ss}4>K~~9%{u#X4;VbtoE2GMnnJe~dhapnYZ@3d! zvrAkcQus(T*u_8Z!)tm3P$I>l0>GfA04~(L7$ae35ztk>@A!El_}_YGJ9Jz@?TC<6 zzM+^0sIC4wp72N?xo@Qp^0n;gW!_@w?R)vlM0(CJjLGs)AHbOv{0kb0DzAM5f8mwJ zI@7Y_gf6%lWeyoOW0UIdHG&G@OdBzs1~mO8vFI))1^tdkq)yTn1!Iq1I6#ku09Qw@ zm>}O|e;47yli46X*La*hQuo8ByD^(NDB9lPC$5%u?r+~BmzTh2JBFd8%TjrUbJ=P!Pbz${y9jyB-1n@3VYaRCTWyND+8Z;Ob+Bxt#dHg z43~*R^%3(jM0aw0C)roS?KGO_pN;jxh$39w&SCiLP0{&>$AQc9PBg+D00SRRA=Ern zk>O|^TdqmQ(IQcYqZ2+52DUdou3RQnL$7zqy{3_Ka*$|$w1moR0`O_0fXkKjxO^_M zVCH=APqOqAbaU_s^!XEOrsn(>_D@b?Pyd*3U!%SL(qVK_RknOs@*|>thhVhxgI#oh z>QVY_CVj|^h}|)pfJX@oC{iPzBP-#ds^PMy-#REXsB)LFtPGc9@?$ry?nTUR2156Q z9GKFVpX)YSxSDC4{ubP>H@DqCwE4^(7Ik`bs>R53lbb(roFeZ)zv`*y|GqG1MB5~5 zHEQaf=*X(iCcIE^77Yu!;EYZsXD`zeX^jLniCCo8uPJa}+f+G#EO=z4EF0G(@Z^6s zw0@Unh4e;zMm*GyWy-RsK$Xzu2YLM8qX26C?wO6NinfyZZtIkrdNXmyc}oMr^Xf65 zqYNL)B;3WnH5jj*y?*|D;5nZyLlTA5{)Kr3X{+qh8%=3EE)?`VWD$G?v#To>)^5ED z(=qRl!noR3!~MSiZQen0ri8>!JG;3EF;*U4 zT=^X8w4V_>#Z9<0n1;qU#@EYZ;pSDhUg`($t{~~ORrwq(Plxu?OiZyc_ouHr52MA>vh?N)yj_mf_qCnPJDD$t|3z4- z3Hiy(&}9?Vw+4dxG_Rgd_*Hd3ditJ#$R^V>H}#x(BB6}^C51-@RJIPrdHbFj;4}Ma(M{E?xcmy!ivJu*s!nKOk zs_^Bye%FnqJ-*~z;ZeWNjpz668CN1c0<+f!zK<|2uoc`9@XvVoH0sa2K2v-aIw<_E zmeQZ+_^x)3iju~d=9>gfV1Q?IZ+gd*(HGxJJ3-y(L*!*7{ILy|K|khvu(ryN^KI-5 zI*0dEa)8Om-jYqSE}9<)#mg-Ld&hj zAhWOA!Tsd5uW@T>cpiGYm)!_LMMT$dzB6sLqf}LAMYa*Swf(@&L~byYia@h7$Apx9 z$L*(R&!{9MR6cEa6OaMD4quUs6M;Qiq<7~e@6XfL?ok_7FghDQ@S~#-P9KrFr9)bJ zBqM)|iXEqX!-b}s)W$H^X&kx|XqJirC@Y^Z($Y}z?wdDfxmhJ~`5;my1g=gQFsXf# zT`3yOsEs}*H7{TKxjIsvcOb2M8n&~|a?!VzAg;`$cI<0=Hx1fiy$fCVdszYW{%HzC?Cw|QnJn(qUCZgYcq=g`f|UG z?AK;k#euXvDsXc`1RO17Xnu=M&9b4lA1CoE1|Dal<_H2`Qj9q|JTJ^5ALE;|^I}w=_JD0wpJMwchC>UixbxdmcY8 z9^?2?J4%7$?!SLD+Yi#zJxsj-%GiwU*h;65ie`^a$@Oeh6zn#>_u0N z#|T&lYI$3yx;Eeqqp*~t>Y@E|aK04onBC$LH%jUZmC9<7>k4$R;aYxc1Mj+;$1$^E zAy|SLn*D}O!rEaZ)j3l!5?|s%PnY)WY>p7FxOx&ufBi$LfsNHJNA3drC^~(Hq~TsO zo2h*5PK+ztD}q10I9OB-y;QjiQpYD1bP71FA^N5mv)Q1Is#}xm#CKC{f?w}(nc3BI zIFZG$G$`_@yzRkNI!(?)x0Rf*>QxH5-Z&26fo-UPS}9o<>mVZF`6s}C>pDOxSuB|`^h_P6zadM83Gt-n zFpQlAjnmn{uZBxGeD;8^@uF#z3p+Lf31*7>%2bCLQ$Y5{YsVhY`=+|#v0WLRM{X!9 z<>1TEl;grwJfhp=Zl&mdD$@wDr&VCSF%-PYam&e6^`4*JRn};~*Ngndaaz z{#0U;BY#q2rDp5tR8slUEa@>m6~?4vYc=lKk6sbyuAN1T^k>4;XU&GzUE~>8E~A2R z{Ot0%%vFTV&G-%YJV!pGdsWrjH{D_q;I6~6!#=N_tEO9c9i!`{j|=u`{AF3S7^nT2 z{KEO1ghv9&yU3nHOS!l%EC&xinY{seW+p@Lu(8`A^Y^0V2~1i)Nml*5iTAZ+D9&Qm zyQX2@@0$iWGcM%iD^uy7eoZ~Uv$5u*V`X=7ceU_N*@Y6IJ$JJ?EaK)-rTpY{>{yrK zr-UoHv1Y*SarScAMR5IH^92O&l65GMeW>@}U_vAuKTTQk~tnEyOcC9Oy z24h{TtQ%!O-)JxcrzG77>}D3x?$}2_o_w0V@pJtk->^K8B-KgI9$P;2sB9-`1{c0l zH_0OP9NU9j-Ejq4Z>m-2sTbBNR73n%r^-A?7ByjwX}KW3)|{sd6AxJf2*0KeqJbu@ zrOLDEm<_ZpxA*RA@XpFg;1?*ozwPXQ(nFK_lZacRVL?0alQ>>5N_8kf;nLh;K`~mL zm_aE}(!?qM!73rp>wq*p=6|rRA5@hmEnNuoTPxN(=!sNXR{uYUCJv@(nFjX$ufe?< zOx!%}LktX~brv1m042@18thxEC^5jvpXsPR(?4jM z4i0ISw(lskUrn z&o-~-jl$T3j2R&f2VoeS8HEo<2YYHXRWi}S2>9^NSYr6my4hgcMRjaMTBYLVrvBqN zEp5C)T?dvXnpws(zX?s4+9B9kM3!AF-wsXvv?!Q!U_aG;fP3?Cptw!+qY6_3o~WBy z0)i}LnG}L98XYzpKnP97o)z=$q?SXIk!Gm+-TE*<>r2&cLXRzEbZeJ5XezRX9V`xn zDKotCBEW|xLt9|lde?@Cew7tDwkcq*82Ly5Cn|jU)R?9F zw?E(ITSS&Eh}q@$@tZFD!tUx|BvVn0kn5n1R7gdMbpp)usa6?oBWSuq7}>QNxXdNd zwLrXocPcWI)xng*qRc`d;t}8ejDSF*MY8R~P`hCUh*rt1``NI|D8vU2(9ObmzursB z^rOzl?j30>B=-vgxd_6~Z#{3N-!BZ>x>X7Md$A*|0Z*Rwazcd}tirn?-R}h(|t?EgWIR3e4E++=tRZDoq^TlA6C#?7N?TBs9OAgt$PZeh(LZ32AHf?;t93f zO^h7^szk*`GS`D3j17o1^gbwtb2{-A$xzI>m`5g~3^0`=@x;s6{Xr_K55c~ArO!$$ z(hnWwNoPxLD!=Sb$uflXsPvZHp0lin!{W~P1h6X}Efx+|3PYY@n)$rxD+cFULp1+- z7xy$>(aK$HSjjomvid=GFt!JZpRR>2TXRbRH#_o8>Xo2fV%QGQpyu6EZ9JKQ?B$9> z3`gJ286VV6+ZJ(jm6Krm^^e3+UK>!2LrttucFyx9vUO9iVR-^1dt*c;Y2uG15^m4a$%_e`aW2Lg@5zPi@*x*XFE(D}) zSmgy&Fv!S`7VAYM@W8D@HlZbBv9c|$eWgUbL;Ry7X^W&D>w*0`hz7&W^} zR~6;xJ#P?s5#5C%c%pF13&>Y6xv|BBzNKVJYbexmW(mi$fa0emA zgG^%Ik=D6ayBce``p;G?8ZjCr&T6@Zph)N%6E$V7s*%JJDBf0|hz2N{MCy!^q8WB| zt@#ES-Ar>#>e|IBx~>qJK@xc;(V-S1y=Y~q98yb;STWheNz93ACCJ&RM75LA7EyBP zh-puiN>wpsUgFf@f0V)ggG0_gI&nE^qtRZ3Ks&_w(S1z>VTVxzO8d8IwLofA8Fyb|;fgjd>k^~KapAskD&sc@ zAf6gQfF1{xZhM)fnXYiIWoAyVv!M0R*em1l%Nia?GY-n_4a@<0$c&bzGaeW|rl!pS zE{bq>_|QBo=}{c?4pu4})yC6Ov0z1tkt4zhYhrmJ>W2*Bjj5!<;W;bT&L~k2i3R+7 z1Zzf(w?)Kb`Q#(f^yvJE(zcqhXASV;16z{alig@GK`5XVYvJ7M$qK0x7__N~tNF;T z1?b%2d#GMCbya``UW6*1NVQ~=3rMjw&nlie$)ud3Q(e3vf;%1+BD6=OA?B+nZ)(_T zn~I^HIkakQwOIYb;jB>3tnYY=Q#<~|x~a}EAw2j=gIt_?MbSa7X4A1Y-RpZC*@8FY zU#vJeI?(f2h0i5YRlUEH$v9%#W-$~*qYnu4amil4bfEz6{dro3b@>G}%QJXxr~v3D zNpfZ1nFB)8RBi(@9`tWM`TN`7nAm-q?1#rb#&HNvNbi%G;&RT~eh&w7S&1i|^vhzO zd#h6}Gzks#ah9g&7S`D&GyGY;QNhEmdvMcjXfBw1Vi5E#;R{rF5^}Sl6y$Z|4MHH!F>&b3RbVw6^ zvH3J`&XdSX$w|)GlVe!SrDmrc_|$bt!ljNp9aiH{pK_W$ksE2z4!y0tP8oNm{MhrP=+(1gw&;u|aom^uv?pdZ1K9wa<8+6AU+5|G=$l<&<_K1d zQVMBDjTTs?OdG}3Ng)^&xG{%`#j;2{*YqogIotT6N}Ez!!D_hi&tsZ;F$?!%?V9CM zsnY`2e^{|}p`-(SC|vX=Ce$PGEK5kEl&NH1MyGO$=B1Ohdy*F7Y>39q|gP zER#8F*$A>XcMVVqpws>fu}zIuL+rU$ev z-h(HP(%?fQxYL0#h)qSd%1NTrg?zZz9M6x`9j;f;jy@VLoaY=licmSZ2IaYghdyZz zZZLC-bc>O+QVzDX%)>qD%pbok&(Wv)tYTp7Ou!s@>&d*7dyWeu;}FIik2Ky68S?-W z$OQTbGmaJxUmszmK3|?a9?blBIdoot2M;%Bzo{n=7N_c;zb0{SoZe1EK2rI7UvR_3 z50!+Iac8&C>Ydr=dpG-MmcFmeJw>o*+kPp`vyDTJtOEXy{yaFa?W*1S(bL=WidOCw zg&pY5pwYK(-wRK01b--28dP+jx8VXm3@U4Pn0jKTD|afS-Sp)vqpa1ZDqsE?`D!y% zTtBg9#f@O|g#P88ber$>`FnhN#3Xz6KzRl(t=Q|H4{Z`J@NHglBQb}uH9@~b|BOhj zbE)v08^u%WN&|Z7;=B%5JvVUV)|HjNRv};_?R=t*N3Wl)#vfY90#YpMyx|q_c&L8W z|0so2>423&-tP8#xuFqS$7AgIe79ra$&D*DrG+kBm>6?(=g?nItoP&zb4@ln!_ZYv zwHyPZvz7L)`%9rFk)q_6gBS-8F;EsVMQz<$dA)e~c68+^uj~sLu=PXo_v<9y{cv*O z?(O0czbx@h9( z=|qPmCEcbmEe%<6=H%4D`rG(evT_BUO=S%}IUs8@x5rSA;gXpw5Eud9?3g4DQVsu5 z23`Fj>e&Y7(dcc}dn+hZo8sEg$AbQ(T-K>Sr*mwm?@ozaW(~IX^3h?Y$FORPzRG^% zIOT4gucqG6H`#Z$nG78j7JeidxSUxsq34?UzS$$xH?{YuhweDicrek(lr-0hiyqm{ zr_b?v;QT%Rd$iCb0FVN-icA()W$!wuX{9HHmo(gm-}XwJGNQ(gx?T(aHFjWU`{vNo z$!+T%68ff8b`i%;4|}<#$MrI#jSQ`STy3Qec?=%FcfcFN_&3oUv(B1 zN|gIIDr7qsEsXZ^{PVOSTBgzRj8lL$gwYuf)mxSi_W3t^jsbK2WSblcR)r*c>>r=0 z0J$xo**5D?x3+DKy5cI;i_;$XE9_dZYu#^Y6sI;};YC{^>yK1Kzk}msnA_(Q*!1cl zzuk_BppBQ-WBnb=ZaD#ikJ@x_19X*%dtuqf=k3DB$4{t6*MD)q^@uWo0b1`ylZl(* zYlUVo8xo7LwcnUBhp|4Ho?|JjfS~>qn2=u3D_dm7N&@MXEa+^xx!vT>Pba z3y|~csOarmb!Kb@d4i`-_M#hmQNvVfj&>tL*$}qL+1Dm03t#wFriV{p#w^m3hMtY* z1J5Lk?$)x9k==<}j5_S+BroOeKSSL!IV(H@RRG^LVxYW38Nu)7^X4q<#7??4g1PB0 zm3N`=oDCt#R!GqYYAEKz*r})YtzV1fgjXT9_syQ=HMm{wQ_0D~Nz~Mv1q7&+w8iUv z7Otw)=v$76h?ae16C9TT)Yx6)NUKLkb-bsRUQTYb`Q5>p?cy!qR#^9Z^U^;tXu*N7 zA}+3Z2l#m-Qw-+4w=i37wZ^Z;*nhEExHEmP^ohEfgoY|zrEF;lo*;0q-)7e$EcW1k zyRy6MP*-o2*;=>K&QJM_mM_N(Ic=@o7W7{$gK^JHxKvg)#+XBIuZh%s%NuD%@?KU` z`@{ZLIE_wi*{eQn_oFe@b!DvYXw5>1TVP8;5~y6NBYMCSf7ggZ+Tr*88`r(k=_B)4 zG!lNJE}M@yA5s+Mm;KPfC3(tAWyKJW$NlJ z3~+L>7_ZvrO|^I#5v;MM9qdsZvxC0<`Vr*nn!*=Rb|Jrur+L(Q4ika3FJY8UnvrT} z{AAF^@Hp!B=ZCpoZ!_sF(Z%wk6|dXxDM|G#q`orY-)-)(OWY>~WFaLWj%EWNsabORRL;L3aj7hVWd#%p^2G7#fROvsx{gl3J%ul zC&{K7SWVujP4_z7VD7x?X4C$``p1GI8R|L2%$u}y^hx4-rb%LOM59gVBbABr-t@v5 zr|xoE?Q=T8F$}FJ({;R6(fn)s??F?E{t0vG6#OF9lsLSsGTt}Blo<(cO@v1%<^ZE0 z({my&>aa)H9cu63>>G7H2L4&Lo`3HCttY2}k|WykmD)_$1LXd^!h^kU!!u9mieE-k zf9Q5$H|wMQe#Lco@6?Y(y&6|rr(zA=m(JaaFSV+v^hBFqG&>_1UEtHNijW=AFC60Q zi7eyVHp2sYLd(4me^@cO*wnStI{@%u?O?m?FMmb15re%*$Ez>XrX-8abBnAH$-_N{cgw_{#H}5k z>*P-+6$iobvTxvA2O2NYbwH;LVXJGK&K*qJD`|K;_;C%tz=ii^ro||xJK$fuR$~I; z=JL%B7wcc38<0(O;@~=*@4HunThd%|Mf|{(#9q3lXvcXeY^0B=Xay}%!tK-xWz!IbqtNY z$I~f*QY~@!??AkL@O72L|0H55(ruOBE(DwxtsshHwj^K-Rnz*#%>QC;;YL+v#jmgZ z8AaK0D;>10*a)3o_gJ4@DO13EW+|DliV90iZU11xpy|T=&$i9T?*AeOV{14THkLFn zLx?bF4pvqc?tiS1ctZ&M|0)^{Au6!g;g}?>oLyXr|JwqLm6;{+-}=ifegbZQ8Abf* zB~nXUctP6QE+3|VH}_dn8Y>?WZB|`Q_xVXdG)l$MEh&kYi|r?AD2ft!0GhNa3iT;| zP<)~p&Vi2G3MM0XCA`JC_4iotcvuXA-HDRl^MtZs;RA%S5k`Z~!z;)3x`Hn&nT>0c zt#zyvFG^Cmp`_Ms^VWKsin_}~r>BbxAF#UCvOVV3w3oYQAG3W};V;zE87Dn_p0?6adIWzwXul^e(Jv#3m4=0Gl2hld6@s z88HhR8yu4cu?`C{8!^kjOBDwP*MFDH#LUDha7@zn<_`ZVmj7Rg5$o`YGIMdVNl1L> zVwK<$7h@HZ`1fe+;%pKuBCMj~TwJ08#Q%GX9{hjp9NE~DzBp(Biuq0zRpoDCw-186 zViA@>#s!}8D5%nXHVD>Wuw|nZsARO(XtUF4QU|N6+$IOn#fXS0)gd#)_X6S)cvG#m z{&(5jHrH!ItZZ+18R8|Y&V$s#y5`bf(ZoUo%X#24`G^;@0oZ&Ym2+87zVB6* zepkFIifXf7bBcDpc+D%MdR|q9>|3vzLjM1K3kt=I*P_Cw4_@~a{(SYiuUNI|^#Hcv xHN5H5?Jr#0Pjp$hE^F2einUhyKRQ3`AFEO@9+y{T1Re`EHZ?d3B_%~qMhe^GpwIvS diff --git a/doc/src/LogisticRegression/LogReg.do.txt b/doc/src/LogisticRegression/LogReg.do.txt index 62f125d95..fed5ac392 100644 --- a/doc/src/LogisticRegression/LogReg.do.txt +++ b/doc/src/LogisticRegression/LogReg.do.txt @@ -37,7 +37,7 @@ failure etc. !split ===== Optimization and Deep learning ===== -Logistic regression will also serve as our stepping towards neural +Logistic regression will also serve as our stepping stone towards neural network algorithms and supervised deep learning. For logistic learning, the minimization of the cost function leads to a non-linear equation in the parameters $\hat{\beta}$. The optmization of the problem calls therefore for minimization algorithms. This forms the bottle neck of all machine learning algorithms, namely how to find reliable minima of a multi-variable function. This leads us to the family of gradient descent methods. The latter are the working horses of basically all modern machine learning algorithms. @@ -51,19 +51,19 @@ models, as we will see later. ===== Basics ===== We consider the case where the dependent variables, also called the -responses or the outcomes) $y_i$ are discrete and only take values +responses or the outcomes, $y_i$ are discrete and only take values from $k=0,\dots,K-1$ (i.e. $K$ classes). The goal is to predict the -output classes from the design matrix $X\in\mathbb{R}^{n\times p}$ -made of $n$ samples, each of which carries $p$ features. The +output classes from the design matrix $\hat{X}\in\mathbb{R}^{n\times p}$ +made of $n$ samples, each of which carries $p$ features or predictors. The primary goal is to identify the classes to which new unseen samples belong. -Let us specialize to the case of two classes only, with outputs $y_i=0$ and $y_i=1$. Our outcomes could represent the status of a credit card use who could default or not on her/his debt. That is +Let us specialize to the case of two classes only, with outputs $y_i=0$ and $y_i=1$. Our outcomes could represent the status of a credit card user who could default or not on her/his credit card debt. That is !bt \[ -y_i = \begin{bmatrix} 0 & \mathrm{no default}\\ 1 & \mathrm{default} \end{bmatrix}. +y_i = \begin{bmatrix} 0 & \mathrm{no}\\ 1 & \mathrm{yes} \end{bmatrix}. \] !et @@ -78,11 +78,11 @@ We would then have our weighted linear combination, namely !bt \begin{equation} -\hat{y} = \hat{X}_i^T\hat{\beta} + \hat{\epsilon}, +\hat{y} = \hat{X}^T\hat{\beta} + \hat{\epsilon}, \end{equation} !et where $\hat{y}$ is a vector representing the possible outcomes, $\hat{X}$ is our -$n\times p$ design matrix and $\hat{\beta}$ represents our estimators. +$n\times p$ design matrix and $\hat{\beta}$ represents our estimators/predictors. !split ===== Some selected properties ===== @@ -100,20 +100,15 @@ literature. This model is extremely simple. However, in many cases it is more favorable to use a ``soft" classifier that outputs the probability of a given category. This leads us to the logistic function. +The code for plotting the perceptron can be seen here. This si nothing but the standard "Heaviside step function":"https://en.wikipedia.org/wiki/Heaviside_step_function". +!bc pycod + +!ec + + !split ===== The logistic function ===== -A widely used model is the so-called logistic (or sigmoid) function which ranges from $0$ to $1$, -!bt -\[ -p(t) = \frac{1}{1+\mathrm \exp{-t}}=\frac{\exp{t}}{1+\mathrm \exp{t}}. -\] -!et -Note that $1-p(t)= p(-t)$. - -!split -===== Probabilistic classifiers ===== - The perceptron is an example of a ``hard classification'' model. We will encounter this model when we discuss neural networks as well. Each datapoint is deterministically assigned to a category (i.e @@ -123,7 +118,18 @@ than a single value. For example, given $x_i$, the classifier outputs the probability of being in a category $k$. Logistic regression is the most common example of a so-called soft classifier. In logistic regression, the probability that a data point $x_i$ -belongs to a category $y_i=\{0,1\}$ is given by the so-called logit function (or Sigmoid) +belongs to a category $y_i=\{0,1\}$ is given by the so-called logit function (or Sigmoid) which is meant to represent the likelihood for a given event, +!bt +\[ +p(t) = \frac{1}{1+\mathrm \exp{-t}}=\frac{\exp{t}}{1+\mathrm \exp{t}}. +\] +!et +Note that $1-p(t)= p(-t)$. +The following code plots the logistic function. +!bc pycod + +!ec + !split @@ -141,7 +147,7 @@ where $\hat{\beta}$ are the weights we wish to extract from data, in our case $\ Note that we used !bt \[ -p(y_i=1) = 1-p(y_i=0). +p(y_i=0\vert x_i, \hat{\beta}) = 1-p(y_i=1\vert x_i, \hat{\beta}). \] !et @@ -150,7 +156,7 @@ p(y_i=1) = 1-p(y_i=0). In order to define the total likelihood for all possible outcomes from a dataset $\mathcal{D}=\{(y_i,x_i)\}$, with the binary labels -$y_i\in\{0,1\}$ and where the data points are drawn independently, we use the so-called Maximum Likelihood Estimation (MLE) principle. +$y_i\in\{0,1\}$ and where the data points are drawn independently, we use the so-called "Maximum Likelihood Estimation":"https://en.wikipedia.org/wiki/Maximum_likelihood_estimation" (MLE) principle. We aim thus at maximizing the probability of seeing the observed data. We can then approximate the likelihood in terms of the product of the individual probabilities of a specific outcome $y_i$, that is @@ -162,7 +168,7 @@ P(\mathcal{D}|\hat{\beta})& = \prod_{i=1}^n \left[p(y_i=1|x_i,\hat{\beta})\right from which we obtain the log-likelihood and our _cost/loss_ function !bt \[ -\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n y_i\log{p(y_i=1|x_i,\hat{\beta})} + (1-y_i)\log\left[1-p(y_i=1|x_i,\hat{\beta}))\right]. +\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left( y_i\log{p(y_i=1|x_i,\hat{\beta})} + (1-y_i)\log\left[1-p(y_i=1|x_i,\hat{\beta}))\right]\right). \] !et @@ -172,15 +178,15 @@ from which we obtain the log-likelihood and our _cost/loss_ function Reordering the logarithms, we can rewrite the _cost/loss_ function as !bt \[ -\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{1+\exp{(\beta_0+\beta_1x_i)}}\right). +\mathcal{C}(\hat{\beta}) = \sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{(1+\exp{(\beta_0+\beta_1x_i)})}\right). \] !et -!split + The maximum likelihood estimator is defined as the set of parameters that maximize the log-likelihood where we maximize with respect to $\beta$. Since the cost (error) function is just the negative log-likelihood, for logistic regression we have that !bt \[ -\mathcal{C}(\hat{\beta})=-\sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{1+\exp{(\beta_0+\beta_1x_i)}}\right). +\mathcal{C}(\hat{\beta})=-\sum_{i=1}^n \left(y_i(\beta_0+\beta_1x_i) -\log{(1+\exp{(\beta_0+\beta_1x_i)})}\right). \] !et This equation is known in statistics as the _cross entropy_. Finally, we note that just as in linear regression, @@ -231,6 +237,18 @@ $p(y_i\vert x_i,\hat{\beta})(1-p(y_i\vert x_i,\hat{\beta})$, we can obtain a com \] !et +!split +===== Extending to more predictors ===== + +!split +===== Including more classes ===== + +!split +===== Optimizing the cost function ===== + +Newton's method and gradient descent methods + + !split ===== A _scikit-learn_ example =====