From c2f533862f02c01467a4f8178f923b1a4d3e7192 Mon Sep 17 00:00:00 2001 From: mhjensen Date: Sun, 29 Dec 2019 21:56:55 +0100 Subject: [PATCH] updating --- doc/pub/DimRed/html/._DimRed-bs000.html | 46 +++++---- doc/pub/DimRed/html/._DimRed-bs001.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs002.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs003.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs004.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs005.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs006.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs007.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs008.html | 53 ++++++---- doc/pub/DimRed/html/._DimRed-bs009.html | 46 +++++---- doc/pub/DimRed/html/._DimRed-bs010.html | 54 ++++++---- doc/pub/DimRed/html/._DimRed-bs011.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs012.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs013.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs014.html | 48 +++++---- doc/pub/DimRed/html/._DimRed-bs015.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs016.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs017.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs018.html | 46 +++++---- doc/pub/DimRed/html/._DimRed-bs019.html | 59 +++++++---- doc/pub/DimRed/html/._DimRed-bs020.html | 44 ++++---- doc/pub/DimRed/html/._DimRed-bs021.html | 49 +++++---- doc/pub/DimRed/html/._DimRed-bs022.html | 47 +++++---- doc/pub/DimRed/html/._DimRed-bs023.html | 95 +++++------------ doc/pub/DimRed/html/._DimRed-bs024.html | 103 ++++++++++++------- doc/pub/DimRed/html/._DimRed-bs025.html | 96 +++++++++-------- doc/pub/DimRed/html/._DimRed-bs026.html | 98 +++++++++++------- doc/pub/DimRed/html/._DimRed-bs027.html | 73 ++++++++----- doc/pub/DimRed/html/._DimRed-bs028.html | 59 ++++++----- doc/pub/DimRed/html/._DimRed-bs029.html | 67 ++++++------ doc/pub/DimRed/html/._DimRed-bs030.html | 73 ++++++++----- doc/pub/DimRed/html/DimRed-bs.html | 46 +++++---- doc/pub/DimRed/html/DimRed-reveal.html | 78 ++++++++++---- doc/pub/DimRed/html/DimRed-solarized.html | 101 ++++++++++++------ doc/pub/DimRed/html/DimRed.html | 101 ++++++++++++------ doc/pub/DimRed/ipynb/DimRed.ipynb | 50 ++++++--- doc/pub/DimRed/ipynb/ipynb-DimRed-src.tar.gz | Bin 191 -> 191 bytes doc/pub/DimRed/pdf/DimRed-minted.pdf | Bin 260550 -> 262362 bytes doc/src/DimRed/DimRed.do.txt | 51 +++++++-- 39 files changed, 1281 insertions(+), 874 deletions(-) diff --git a/doc/pub/DimRed/html/._DimRed-bs000.html b/doc/pub/DimRed/html/._DimRed-bs000.html index c0e2e1fdb..fcef4f476 100644 --- a/doc/pub/DimRed/html/._DimRed-bs000.html +++ b/doc/pub/DimRed/html/._DimRed-bs000.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -218,7 +224,7 @@ MathJax.Hub.Config({
    [2] Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University

    -

    Dec 28, 2019

    +

    Dec 29, 2019


    @@ -242,7 +248,7 @@ MathJax.Hub.Config({

  • 9
  • 10
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs001.html b/doc/pub/DimRed/html/._DimRed-bs001.html index 604451fe3..95607135f 100644 --- a/doc/pub/DimRed/html/._DimRed-bs001.html +++ b/doc/pub/DimRed/html/._DimRed-bs001.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -250,7 +256,7 @@ visualization.
  • 10
  • 11
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs002.html b/doc/pub/DimRed/html/._DimRed-bs002.html index f281284d9..020984f1f 100644 --- a/doc/pub/DimRed/html/._DimRed-bs002.html +++ b/doc/pub/DimRed/html/._DimRed-bs002.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -241,7 +247,7 @@ ensures that all features are exactly between \( 0 \) and \( 1 \). The
  • 11
  • 12
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs003.html b/doc/pub/DimRed/html/._DimRed-bs003.html index d2bfcf82f..974ed102f 100644 --- a/doc/pub/DimRed/html/._DimRed-bs003.html +++ b/doc/pub/DimRed/html/._DimRed-bs003.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -244,7 +250,7 @@ techniques.
  • 12
  • 13
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs004.html b/doc/pub/DimRed/html/._DimRed-bs004.html index b9946be9a..3c420051d 100644 --- a/doc/pub/DimRed/html/._DimRed-bs004.html +++ b/doc/pub/DimRed/html/._DimRed-bs004.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -319,7 +325,7 @@ svm.fit(X_train_scaled, y_train)
  • 13
  • 14
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs005.html b/doc/pub/DimRed/html/._DimRed-bs005.html index c1f31f395..c2ac3ee56 100644 --- a/doc/pub/DimRed/html/._DimRed-bs005.html +++ b/doc/pub/DimRed/html/._DimRed-bs005.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -269,7 +275,7 @@ svm.fit(X_train_scaled, y_train)
  • 14
  • 15
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs006.html b/doc/pub/DimRed/html/._DimRed-bs006.html index 2f361bebf..05eccf8b5 100644 --- a/doc/pub/DimRed/html/._DimRed-bs006.html +++ b/doc/pub/DimRed/html/._DimRed-bs006.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -248,7 +254,7 @@ logreg.fit(X_train_scaled, y_train)
  • 15
  • 16
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs007.html b/doc/pub/DimRed/html/._DimRed-bs007.html index 1c3b2e6e7..7efbdf565 100644 --- a/doc/pub/DimRed/html/._DimRed-bs007.html +++ b/doc/pub/DimRed/html/._DimRed-bs007.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -303,7 +309,7 @@ applications.
  • 16
  • 17
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs008.html b/doc/pub/DimRed/html/._DimRed-bs008.html index ec4603204..3949fe0c4 100644 --- a/doc/pub/DimRed/html/._DimRed-bs008.html +++ b/doc/pub/DimRed/html/._DimRed-bs008.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -201,6 +207,15 @@ MathJax.Hub.Config({

    Basic ideas of the Principal Component Analysis (PCA)

    +

    +The principal component analysis deals with the problem of fitting a +low-dimensional affine subspace \( S \) of dimension \( d \) much smaller than +the totaldimension \( D \) of the problem at hand (our data +set). Mathematically it can be formulated as a statistical problem or +a geometric problem. In our discussion of the theorem for the +classical PCA, we will stay with a statistical approach. This is also +what set the scene historically which for the PCA. +

    We have a data set defined by a design/feature matrix \( \boldsymbol{X} \) (see below for its definition) @@ -233,7 +248,7 @@ We have a data set defined by a design/feature matrix \( \boldsymbol{X} \) (see

  • 17
  • 18
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs009.html b/doc/pub/DimRed/html/._DimRed-bs009.html index 05ce56b2b..5494b0ae6 100644 --- a/doc/pub/DimRed/html/._DimRed-bs009.html +++ b/doc/pub/DimRed/html/._DimRed-bs009.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -203,7 +209,7 @@ MathJax.Hub.Config({

    Before we discuss the PCA theorem, we need to remind ourselves about -the definition of the covariance and the correlation function. +the definition of the covariance and the correlation function. These are quantities

    Suppose we have defined two vectors @@ -282,7 +288,7 @@ In the above example this is the function we constructed using pandas.

  • 18
  • 19
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs010.html b/doc/pub/DimRed/html/._DimRed-bs010.html index 9fcb9607f..0b33892e5 100644 --- a/doc/pub/DimRed/html/._DimRed-bs010.html +++ b/doc/pub/DimRed/html/._DimRed-bs010.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -202,7 +208,9 @@ MathJax.Hub.Config({

    Correlation Function and Design/Feature Matrix

    -In our derivation of the various regression algorithms like Ordinary Least Squares or Ridge regression we defined the design/feature matrix \( \boldsymbol{X} \) as +In our derivation of the various regression algorithms like Ordinary Least Squares or Ridge regression +we defined the design/feature matrix \( \boldsymbol{X} \) as + $$ \boldsymbol{X}=\begin{bmatrix} x_{0,0} & x_{0,1} & x_{0,2}& \dots & \dots x_{0,p-1}\\ @@ -227,7 +235,11 @@ $$ $$

    -With these definitions, we can now rewrite our \( 2\times 2 \) correaltion/covariance matrix in terms of a moe general design/feature matrix \( \boldsymbol{X}\in {\mathbb{R}}^{n\times p} \). This leads to a \( p\times p \) covariance matrix for the vectors \( \boldsymbol{x}_i \) with \( i =0,1,\dots,p-1 \) +With these definitions, we can now rewrite our \( 2\times 2 \) +correaltion/covariance matrix in terms of a moe general design/feature +matrix \( \boldsymbol{X}\in {\mathbb{R}}^{n\times p} \). This leads to a \( p\times p \) +covariance matrix for the vectors \( \boldsymbol{x}_i \) with \( i=0,1,\dots,p-1 \) + $$ \boldsymbol{C}[\boldsymbol{x}] = \begin{bmatrix} \mathrm{var}[\boldsymbol{x}_0] & \mathrm{cov}[\boldsymbol{x}_0,\boldsymbol{x}_1] & \mathrm{cov}[\boldsymbol{x}_0,\boldsymbol{x}_2] & \dots & \dots & \mathrm{cov}[\boldsymbol{x}_0,\boldsymbol{x}_{p-1}]\\ @@ -277,7 +289,7 @@ $$

  • 19
  • 20
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs011.html b/doc/pub/DimRed/html/._DimRed-bs011.html index 77d5296ca..7eefc658d 100644 --- a/doc/pub/DimRed/html/._DimRed-bs011.html +++ b/doc/pub/DimRed/html/._DimRed-bs011.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -265,7 +271,7 @@ C = np.c
  • 20
  • 21
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs012.html b/doc/pub/DimRed/html/._DimRed-bs012.html index 7a8763b59..d2d517c11 100644 --- a/doc/pub/DimRed/html/._DimRed-bs012.html +++ b/doc/pub/DimRed/html/._DimRed-bs012.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -267,7 +273,7 @@ The above procedure with numpy can be made more compact if we use pand
  • 21
  • 22
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs013.html b/doc/pub/DimRed/html/._DimRed-bs013.html index 2ab2ddff5..9a02ba499 100644 --- a/doc/pub/DimRed/html/._DimRed-bs013.html +++ b/doc/pub/DimRed/html/._DimRed-bs013.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -249,7 +255,7 @@ We expand this model to the Franke function discussed above.
  • 22
  • 23
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs014.html b/doc/pub/DimRed/html/._DimRed-bs014.html index d7c6c6753..7fc40d5f4 100644 --- a/doc/pub/DimRed/html/._DimRed-bs014.html +++ b/doc/pub/DimRed/html/._DimRed-bs014.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -256,8 +262,8 @@ columns since all matrix elements in the design matrix were set to one

    This means that the variance for these elements will be zero and will cause problems when we set up the correlation matrix. We can simply -drop these elements as follows and then construct the correlation -matrix. +drop these elements and construct a correlation +matrix without these elements.

    @@ -285,7 +291,7 @@ matrix.

  • 23
  • 24
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs015.html b/doc/pub/DimRed/html/._DimRed-bs015.html index 86ccf8390..c98730ba2 100644 --- a/doc/pub/DimRed/html/._DimRed-bs015.html +++ b/doc/pub/DimRed/html/._DimRed-bs015.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -265,7 +271,7 @@ It is easy to generalize this to a matrix \( \boldsymbol{X}\in {\mathbb{R}}^{n\t
  • 24
  • 25
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs016.html b/doc/pub/DimRed/html/._DimRed-bs016.html index 6f742428c..983a50bab 100644 --- a/doc/pub/DimRed/html/._DimRed-bs016.html +++ b/doc/pub/DimRed/html/._DimRed-bs016.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -273,7 +279,7 @@ features/predictors.
  • 25
  • 26
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs017.html b/doc/pub/DimRed/html/._DimRed-bs017.html index 3d8d8fdb9..6392f4885 100644 --- a/doc/pub/DimRed/html/._DimRed-bs017.html +++ b/doc/pub/DimRed/html/._DimRed-bs017.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -253,7 +259,7 @@ $$
  • 26
  • 27
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs018.html b/doc/pub/DimRed/html/._DimRed-bs018.html index 10609edc8..9796a01be 100644 --- a/doc/pub/DimRed/html/._DimRed-bs018.html +++ b/doc/pub/DimRed/html/._DimRed-bs018.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -229,7 +235,7 @@ X = np.r Make thereafter a small Python code which plots the data. Note that the function multivariate returns also the covariance discussed above and that it is defined by dividing by \( n-1 \) instead of \( n \).

    -Now we are going to implement the PCA algorithm. We will break it down into sub-steps and across multiple cells. +Now we are going to implement the PCA algorithm. We will break it down into various substeps.

    Compute the sample mean and center the data

    @@ -319,7 +325,7 @@ Finally, try out your own PCA function with other data sets.
  • 27
  • 28
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs019.html b/doc/pub/DimRed/html/._DimRed-bs019.html index cce5ff3ab..0281b99d7 100644 --- a/doc/pub/DimRed/html/._DimRed-bs019.html +++ b/doc/pub/DimRed/html/._DimRed-bs019.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -202,8 +208,11 @@ MathJax.Hub.Config({

    Classical PCA Theorem

    -We assume now that we have a design matrix \( \boldsymbol{X} \) which has been centered as discussed above. For the sake of simplicity we skip the overline symbol. The matrix is defined in terms of the various column vectors \( [\boldsymbol{x}_0,\boldsymbol{x}_1,\dots, \boldsymbol{x}_{p-1}] \) -each with dimension \( \boldsymbol{x}\in {\mathbb{R}}^{n} \). +We assume now that we have a design matrix \( \boldsymbol{X} \) which has been +centered as discussed above. For the sake of simplicity we skip the +overline symbol. The matrix is defined in terms of the various column +vectors \( [\boldsymbol{x}_0,\boldsymbol{x}_1,\dots, \boldsymbol{x}_{p-1}] \) each with dimension +\( \boldsymbol{x}\in {\mathbb{R}}^{n} \).

    We assume also that we have an orthogonal transformation \( \boldsymbol{W}\in {\mathbb{R}}^{p\times p} \). We define the reconstruction error (which is similar to the mean squared error we have seen before) as @@ -215,7 +224,13 @@ with \( \overline{\boldsymbol{x}}_i = \boldsymbol{W}\boldsymbol{z}_i \), where \ \( \boldsymbol{Z}\in{\mathbb{R}}^{p\times n} \). When doing PCA we want to reduce this dimensionality.

    -The PCA theorem states that minimizing the above reconstruction error corresponds to setting \( \boldsymbol{W}=\boldsymbol{S} \), the orthogonal matrix which diagonalizes the empirical covariance(correlation) matrix. The optimal low-dimensional encoding of the data is then given by a set of vectors \( \boldsymbol{z}_i \) with at most \( l \) vectors, with \( l < < p \), defined by the orthogonal projection of the data onto the columns spanned by the eigenvectors of the covariance(correlations matrix). +The PCA theorem states that minimizing the above reconstruction error +corresponds to setting \( \boldsymbol{W}=\boldsymbol{S} \), the orthogonal matrix which +diagonalizes the empirical covariance(correlation) matrix. The optimal +low-dimensional encoding of the data is then given by a set of vectors +\( \boldsymbol{z}_i \) with at most \( l \) vectors, with \( l < < p \), defined by the +orthogonal projection of the data onto the columns spanned by the +eigenvectors of the covariance(correlations matrix).

    @@ -243,7 +258,7 @@ The PCA theorem states that minimizing the above reconstruction error correspond

  • 28
  • 29
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs020.html b/doc/pub/DimRed/html/._DimRed-bs020.html index 536339c85..08aa77aa6 100644 --- a/doc/pub/DimRed/html/._DimRed-bs020.html +++ b/doc/pub/DimRed/html/._DimRed-bs020.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -245,7 +251,7 @@ where the vectors on the rhs are known.
  • 29
  • 30
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs021.html b/doc/pub/DimRed/html/._DimRed-bs021.html index 2ecf2204b..3a436e092 100644 --- a/doc/pub/DimRed/html/._DimRed-bs021.html +++ b/doc/pub/DimRed/html/._DimRed-bs021.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -232,7 +238,10 @@ $$ $$

    -We are almost there, we have obtained a relation between minimizing the reconstruction error and the variance and the covariance matrix. Minimizing the error is equivalent to maximizing the variance of the projected data. +We are almost there, we have obtained a relation between minimizing +the reconstruction error and the variance and the covariance +matrix. Minimizing the error is equivalent to maximizing the variance +of the projected data.

    @@ -260,7 +269,7 @@ We are almost there, we have obtained a relation between minimizing the reconstr

  • 30
  • 31
  • ...
    • -
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs022.html b/doc/pub/DimRed/html/._DimRed-bs022.html index 285b05f8e..29dc48b88 100644 --- a/doc/pub/DimRed/html/._DimRed-bs022.html +++ b/doc/pub/DimRed/html/._DimRed-bs022.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -247,6 +253,9 @@ discussion in chapter 12.2 of Murphy's text has also a nice link with the Singular Value Decomposition theorem. For categorical data, see chapter 12.4 and discussion therein. +

    +Additional part of the proof for the other eigenvectors will be added by mid January 2020. +

    @@ -272,6 +281,8 @@ chapter 12.4 and discussion therein.

  • 30
  • 31
  • 32
    • +
  • ...
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs023.html b/doc/pub/DimRed/html/._DimRed-bs023.html index 67b68fa28..9f8585645 100644 --- a/doc/pub/DimRed/html/._DimRed-bs023.html +++ b/doc/pub/DimRed/html/._DimRed-bs023.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -199,59 +205,11 @@ MathJax.Hub.Config({ -

    Principal Component Analysis

    -
    -
    -

    -Principal Component Analysis (PCA) is by far the most popular dimensionality reduction algorithm. -First it identifies the hyperplane that lies closest to the data, and then it projects the data onto it. +

    Geometric Interpretation and link with Singular Value Decomposition

    -The following Python code uses NumPy’s svd() function to obtain all the principal components of the -training set, then extracts the first two principal components. First we center the data using either pandas or our own code -

    +This material will be added by mid January 2020. - -

    import numpy as np
-import pandas as pd
-from IPython.display import display
-np.random.seed(100)
-# setting up a 10 x 5 vanilla matrix 
-rows = 10
-cols = 5
-X = np.random.randn(rows,cols)
-df = pd.DataFrame(X)
-# Pandas does the centering for us
-df = df -df.mean()
-display(df)
-
-# we center it ourselves
-X_centered = X - X.mean(axis=0)
-# Then check the difference between pandas and our own set up
-print(X_centered-df)
-#Now we do an SVD
-U, s, V = np.linalg.svd(X_centered)
-c1 = V.T[:, 0]
-c2 = V.T[:, 1]
-W2 = V.T[:, :2]
-X2D = X_centered.dot(W2)
-print(X2D)
-
    -

    -PCA assumes that the dataset is centered around the origin. Scikit-Learn’s PCA classes take care of centering -the data for you. However, if you implement PCA yourself (as in the preceding example), or if you use other libraries, don’t -forget to center the data first. - -

    -Once you have identified all the principal components, you can reduce the dimensionality of the dataset -down to \( d \) dimensions by projecting it onto the hyperplane defined by the first \( d \) principal components. -Selecting this hyperplane ensures that the projection will preserve as much variance as possible. -

    - - -

    W2 = V.T[:, :2]
-X2D = X_centered.dot(W2)
-

    @@ -276,6 +234,7 @@ X2D = X_centered30

  • 31
  • 32
    • +
  • 33
  • »
    • diff --git a/doc/pub/DimRed/html/._DimRed-bs024.html b/doc/pub/DimRed/html/._DimRed-bs024.html index 2e151571d..2be659898 100644 --- a/doc/pub/DimRed/html/._DimRed-bs024.html +++ b/doc/pub/DimRed/html/._DimRed-bs024.html @@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source ('Proof of the PCA Theorem', 2, None, '___sec22'), ('PCA Proof continued', 2, None, '___sec23'), ('The final step', 2, None, '___sec24'), - ('Principal Component Analysis', 2, None, '___sec25'), - ('PCA and scikit-learn', 2, None, '___sec26'), - ('Back to the Cancer Data', 2, None, '___sec27'), - ('More on the PCA', 2, None, '___sec28'), - ('Incremental PCA', 2, None, '___sec29'), - ('Randomized PCA', 2, None, '___sec30'), - ('Kernel PCA', 2, None, '___sec31'), - ('LLE', 2, None, '___sec32'), - ('Other techniques', 2, None, '___sec33')]} + ('Geometric Interpretation and link with Singular Value ' + 'Decomposition', + 2, + None, + '___sec25'), + ('Principal Component Analysis', 2, None, '___sec26'), + ('PCA and scikit-learn', 2, None, '___sec27'), + ('Back to the Cancer Data', 2, None, '___sec28'), + ('More on the PCA', 2, None, '___sec29'), + ('Incremental PCA', 2, None, '___sec30'), + ('Randomized PCA', 2, None, '___sec31'), + ('Kernel PCA', 2, None, '___sec32'), + ('LLE', 2, None, '___sec33'), + ('Other techniques', 2, None, '___sec34')]} end of tocinfo --> @@ -175,15 +180,16 @@ MathJax.Hub.Config({
  • Proof of the PCA Theorem
  • PCA Proof continued
  • The final step
    • -
  • Principal Component Analysis
    • -
  • PCA and scikit-learn
    • -
  • Back to the Cancer Data
    • -
  • More on the PCA
    • -
  • Incremental PCA
    • -
  • Randomized PCA
    • -
  • Kernel PCA
    • -
  • LLE
    • -
  • Other techniques
    • +
  • Geometric Interpretation and link with Singular Value Decomposition
    • +
  • Principal Component Analysis
    • +
  • PCA and scikit-learn
    • +
  • Back to the Cancer Data
    • +
  • More on the PCA
    • +
  • Incremental PCA
    • +
  • Randomized PCA
    • +
  • Kernel PCA
    • +
  • LLE
    • +
  • Other techniques
    • @@ -197,37 +203,61 @@ MathJax.Hub.Config({

     

     

     

    - + -

    PCA and scikit-learn

    +

    Principal Component Analysis

    +
    +
    +

    +Principal Component Analysis (PCA) is by far the most popular dimensionality reduction algorithm. +First it identifies the hyperplane that lies closest to the data, and then it projects the data onto it.

    -Scikit-Learn’s PCA class implements PCA using SVD decomposition just like we did before. The -following code applies PCA to reduce the dimensionality of the dataset down to two dimensions (note -that it automatically takes care of centering the data): +The following Python code uses NumPy’s svd() function to obtain all the principal components of the +training set, then extracts the first two principal components. First we center the data using either pandas or our own code

    -

    #thereafter we do a PCA with Scikit-learn
-from sklearn.decomposition import PCA
-pca = PCA(n_components = 2)
-X2D = pca.fit_transform(X)
+
    import numpy as np
+import pandas as pd
+from IPython.display import display
+np.random.seed(100)
+# setting up a 10 x 5 vanilla matrix 
+rows = 10
+cols = 5
+X = np.random.randn(rows,cols)
+df = pd.DataFrame(X)
+# Pandas does the centering for us
+df = df -df.mean()
+display(df)
+
+# we center it ourselves
+X_centered = X - X.mean(axis=0)
+# Then check the difference between pandas and our own set up
+print(X_centered-df)
+#Now we do an SVD
+U, s, V = np.linalg.svd(X_centered)
+c1 = V.T[:, 0]
+c2 = V.T[:, 1]
+W2 = V.T[:, :2]
+X2D = X_centered.dot(W2)
 print(X2D)
 
    
 
    
-After fitting the PCA transformer to the dataset, you can access the principal components using the
-components variable (note that it contains the PCs as horizontal vectors, so, for example, the first
-principal component is equal to 
+PCA assumes that the dataset is centered around the origin. Scikit-Learn’s PCA classes take care of centering
+the data for you. However, if you implement PCA yourself (as in the preceding example), or if you use other libraries, don’t
+forget to center the data first.
+
+
    
+Once you have identified all the principal components, you can reduce the dimensionality of the dataset
+down to \( d \) dimensions by projecting it onto the hyperplane defined by the first \( d \) principal components.
+Selecting this hyperplane ensures that the projection will preserve as much variance as possible. 
 
    
 
 
-
    pca.components_.T[:, 0].
+
    W2 = V.T[:, :2]
+X2D = X_centered.dot(W2)
 
    
-
    
-Another very useful piece of information is the explained variance ratio of each principal component,
-available via the \( explained\_variance\_ratio \) variable. It indicates the proportion of the dataset’s
-variance that lies along the axis of each principal component.
-
 
    
 
    
 
@@ -251,6 +281,7 @@ variance that lies along the axis of each principal component.
   
  • 30
    • 
   
  • 31
    • 
   
  • 32
    • 
+  
  • 33
    • 
   
  • »
    • 
 
 
diff --git a/doc/pub/DimRed/html/._DimRed-bs025.html b/doc/pub/DimRed/html/._DimRed-bs025.html
index a1f2ee2be..f9daa99eb 100644
--- a/doc/pub/DimRed/html/._DimRed-bs025.html
+++ b/doc/pub/DimRed/html/._DimRed-bs025.html
@@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source
               ('Proof of the PCA Theorem', 2, None, '___sec22'),
               ('PCA Proof continued', 2, None, '___sec23'),
               ('The final step', 2, None, '___sec24'),
-              ('Principal Component Analysis', 2, None, '___sec25'),
-              ('PCA and scikit-learn', 2, None, '___sec26'),
-              ('Back to the Cancer Data', 2, None, '___sec27'),
-              ('More on the PCA', 2, None, '___sec28'),
-              ('Incremental PCA', 2, None, '___sec29'),
-              ('Randomized PCA', 2, None, '___sec30'),
-              ('Kernel PCA', 2, None, '___sec31'),
-              ('LLE', 2, None, '___sec32'),
-              ('Other techniques', 2, None, '___sec33')]}
+              ('Geometric Interpretation and link with Singular Value '
+               'Decomposition',
+               2,
+               None,
+               '___sec25'),
+              ('Principal Component Analysis', 2, None, '___sec26'),
+              ('PCA and scikit-learn', 2, None, '___sec27'),
+              ('Back to the Cancer Data', 2, None, '___sec28'),
+              ('More on the PCA', 2, None, '___sec29'),
+              ('Incremental PCA', 2, None, '___sec30'),
+              ('Randomized PCA', 2, None, '___sec31'),
+              ('Kernel PCA', 2, None, '___sec32'),
+              ('LLE', 2, None, '___sec33'),
+              ('Other techniques', 2, None, '___sec34')]}
 end of tocinfo -->
 
 
@@ -175,15 +180,16 @@ MathJax.Hub.Config({
       
  • Proof of the PCA Theorem
    • 
       
  • PCA Proof continued
    • 
       
  • The final step
    • 
-      
  • Principal Component Analysis
    • 
-      
  • PCA and scikit-learn
    • 
-      
  • Back to the Cancer Data
    • 
-      
  • More on the PCA
    • 
-      
  • Incremental PCA
    • 
-      
  • Randomized PCA
    • 
-      
  • Kernel PCA
    • 
-      
  • LLE
    • 
-      
  • Other techniques
    • 
+      
  • Geometric Interpretation and link with Singular Value Decomposition
    • 
+      
  • Principal Component Analysis
    • 
+      
  • PCA and scikit-learn
    • 
+      
  • Back to the Cancer Data
    • 
+      
  • More on the PCA
    • 
+      
  • Incremental PCA
    • 
+      
  • Randomized PCA
    • 
+      
  • Kernel PCA
    • 
+      
  • LLE
    • 
+      
  • Other techniques
    • 
 
         
       
@@ -197,45 +203,36 @@ MathJax.Hub.Config({
 
     
     
     
     
 
 
-
+
 
-
    Back to the Cancer Data 
    
-We can now repeat the above but applied to real data, in this case our breast cancer data.
-Here we compute performance scores on the training data using logistic regression.
+
    PCA and scikit-learn 
    
+
+
    
+Scikit-Learn’s PCA class implements PCA using SVD decomposition just like we did before. The
+following code applies PCA to reduce the dimensionality of the dataset down to two dimensions (note
+that it automatically takes care of centering the data):
 
    
 
 
-
    import matplotlib.pyplot as plt
-import numpy as np
-from sklearn.model_selection import  train_test_split 
-from sklearn.datasets import load_breast_cancer
-from sklearn.linear_model import LogisticRegression
-cancer = load_breast_cancer()
-
-X_train, X_test, y_train, y_test = train_test_split(cancer.data,cancer.target,random_state=0)
-
-logreg = LogisticRegression()
-logreg.fit(X_train, y_train)
-print("Train set accuracy from Logistic Regression: {:.2f}".format(logreg.score(X_train,y_train)))
-# We scale the data
-from sklearn.preprocessing import StandardScaler
-scaler = StandardScaler()
-scaler.fit(X_train)
-X_train_scaled = scaler.transform(X_train)
-X_test_scaled = scaler.transform(X_test)
-# Then perform again a log reg fit
-logreg.fit(X_train_scaled, y_train)
-print("Train set accuracy scaled data: {:.2f}".format(logreg.score(X_train_scaled,y_train)))
-#thereafter we do a PCA with Scikit-learn
+
    #thereafter we do a PCA with Scikit-learn
 from sklearn.decomposition import PCA
 pca = PCA(n_components = 2)
-X2D_train = pca.fit_transform(X_train_scaled)
-# and finally compute the log reg fit and the score on the training data	
-logreg.fit(X2D_train,y_train)
-print("Train set accuracy scaled and PCA data: {:.2f}".format(logreg.score(X2D_train,y_train)))
+X2D = pca.fit_transform(X)
+print(X2D)
 
    
 
    
-We see that our training data after the PCA decomposition has a performance similar to the non-scaled data.
+After fitting the PCA transformer to the dataset, you can access the principal components using the
+components variable (note that it contains the PCs as horizontal vectors, so, for example, the first
+principal component is equal to 
+
    
+
+
+
    pca.components_.T[:, 0].
+
    
+
    
+Another very useful piece of information is the explained variance ratio of each principal component,
+available via the \( explained\_variance\_ratio \) variable. It indicates the proportion of the dataset’s
+variance that lies along the axis of each principal component.
 
 
    
 
    
@@ -259,6 +256,7 @@ We see that our training data after the PCA decomposition has a performance simi
   
  • 30
    • 
   
  • 31
    • 
   
  • 32
    • 
+  
  • 33
    • 
   
  • »
    • 
 
 
diff --git a/doc/pub/DimRed/html/._DimRed-bs026.html b/doc/pub/DimRed/html/._DimRed-bs026.html
index 99b82253e..e59ea0bfd 100644
--- a/doc/pub/DimRed/html/._DimRed-bs026.html
+++ b/doc/pub/DimRed/html/._DimRed-bs026.html
@@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source
               ('Proof of the PCA Theorem', 2, None, '___sec22'),
               ('PCA Proof continued', 2, None, '___sec23'),
               ('The final step', 2, None, '___sec24'),
-              ('Principal Component Analysis', 2, None, '___sec25'),
-              ('PCA and scikit-learn', 2, None, '___sec26'),
-              ('Back to the Cancer Data', 2, None, '___sec27'),
-              ('More on the PCA', 2, None, '___sec28'),
-              ('Incremental PCA', 2, None, '___sec29'),
-              ('Randomized PCA', 2, None, '___sec30'),
-              ('Kernel PCA', 2, None, '___sec31'),
-              ('LLE', 2, None, '___sec32'),
-              ('Other techniques', 2, None, '___sec33')]}
+              ('Geometric Interpretation and link with Singular Value '
+               'Decomposition',
+               2,
+               None,
+               '___sec25'),
+              ('Principal Component Analysis', 2, None, '___sec26'),
+              ('PCA and scikit-learn', 2, None, '___sec27'),
+              ('Back to the Cancer Data', 2, None, '___sec28'),
+              ('More on the PCA', 2, None, '___sec29'),
+              ('Incremental PCA', 2, None, '___sec30'),
+              ('Randomized PCA', 2, None, '___sec31'),
+              ('Kernel PCA', 2, None, '___sec32'),
+              ('LLE', 2, None, '___sec33'),
+              ('Other techniques', 2, None, '___sec34')]}
 end of tocinfo -->
 
 
@@ -175,15 +180,16 @@ MathJax.Hub.Config({
       
  • Proof of the PCA Theorem
    • 
       
  • PCA Proof continued
    • 
       
  • The final step
    • 
-      
  • Principal Component Analysis
    • 
-      
  • PCA and scikit-learn
    • 
-      
  • Back to the Cancer Data
    • 
-      
  • More on the PCA
    • 
-      
  • Incremental PCA
    • 
-      
  • Randomized PCA
    • 
-      
  • Kernel PCA
    • 
-      
  • LLE
    • 
-      
  • Other techniques
    • 
+      
  • Geometric Interpretation and link with Singular Value Decomposition
    • 
+      
  • Principal Component Analysis
    • 
+      
  • PCA and scikit-learn
    • 
+      
  • Back to the Cancer Data
    • 
+      
  • More on the PCA
    • 
+      
  • Incremental PCA
    • 
+      
  • Randomized PCA
    • 
+      
  • Kernel PCA
    • 
+      
  • LLE
    • 
+      
  • Other techniques
    • 
 
         
       
@@ -199,33 +205,44 @@ MathJax.Hub.Config({
 
 
 
-
    More on the PCA 
    
-
-
    
-Instead of arbitrarily choosing the number of dimensions to reduce down to, it is generally preferable to
-choose the number of dimensions that add up to a sufficiently large portion of the variance (e.g., 95%).
-Unless, of course, you are reducing dimensionality for data visualization — in that case you will
-generally want to reduce the dimensionality down to 2 or 3.
-The following code computes PCA without reducing dimensionality, then computes the minimum number
-of dimensions required to preserve 95% of the training set’s variance:
+
    Back to the Cancer Data 
    
+We can now repeat the above but applied to real data, in this case our breast cancer data.
+Here we compute performance scores on the training data using logistic regression.
 
    
 
 
-
    pca = PCA()
-pca.fit(X)
-cumsum = np.cumsum(pca.explained_variance_ratio_)
-d = np.argmax(cumsum >= 0.95) + 1
-
    
-
    
-You could then set \( n\_components=d \) and run PCA again. However, there is a much better option: instead
-of specifying the number of principal components you want to preserve, you can set \( n\_components \) to be
-a float between 0.0 and 1.0, indicating the ratio of variance you wish to preserve:
-
    
+
    import matplotlib.pyplot as plt
+import numpy as np
+from sklearn.model_selection import  train_test_split 
+from sklearn.datasets import load_breast_cancer
+from sklearn.linear_model import LogisticRegression
+cancer = load_breast_cancer()
 
-
-
    pca = PCA(n_components=0.95)
-X_reduced = pca.fit_transform(X)
+X_train, X_test, y_train, y_test = train_test_split(cancer.data,cancer.target,random_state=0)
+
+logreg = LogisticRegression()
+logreg.fit(X_train, y_train)
+print("Train set accuracy from Logistic Regression: {:.2f}".format(logreg.score(X_train,y_train)))
+# We scale the data
+from sklearn.preprocessing import StandardScaler
+scaler = StandardScaler()
+scaler.fit(X_train)
+X_train_scaled = scaler.transform(X_train)
+X_test_scaled = scaler.transform(X_test)
+# Then perform again a log reg fit
+logreg.fit(X_train_scaled, y_train)
+print("Train set accuracy scaled data: {:.2f}".format(logreg.score(X_train_scaled,y_train)))
+#thereafter we do a PCA with Scikit-learn
+from sklearn.decomposition import PCA
+pca = PCA(n_components = 2)
+X2D_train = pca.fit_transform(X_train_scaled)
+# and finally compute the log reg fit and the score on the training data	
+logreg.fit(X2D_train,y_train)
+print("Train set accuracy scaled and PCA data: {:.2f}".format(logreg.score(X2D_train,y_train)))
 
    
+
    
+We see that our training data after the PCA decomposition has a performance similar to the non-scaled data.
+
 
    
 
    
 
@@ -247,6 +264,7 @@ X_reduced = pca
   
  • 30
    • 
   
  • 31
    • 
   
  • 32
    • 
+  
  • 33
    • 
   
  • »
    • 
 
 
diff --git a/doc/pub/DimRed/html/._DimRed-bs027.html b/doc/pub/DimRed/html/._DimRed-bs027.html
index 95a52b810..56acdc987 100644
--- a/doc/pub/DimRed/html/._DimRed-bs027.html
+++ b/doc/pub/DimRed/html/._DimRed-bs027.html
@@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source
               ('Proof of the PCA Theorem', 2, None, '___sec22'),
               ('PCA Proof continued', 2, None, '___sec23'),
               ('The final step', 2, None, '___sec24'),
-              ('Principal Component Analysis', 2, None, '___sec25'),
-              ('PCA and scikit-learn', 2, None, '___sec26'),
-              ('Back to the Cancer Data', 2, None, '___sec27'),
-              ('More on the PCA', 2, None, '___sec28'),
-              ('Incremental PCA', 2, None, '___sec29'),
-              ('Randomized PCA', 2, None, '___sec30'),
-              ('Kernel PCA', 2, None, '___sec31'),
-              ('LLE', 2, None, '___sec32'),
-              ('Other techniques', 2, None, '___sec33')]}
+              ('Geometric Interpretation and link with Singular Value '
+               'Decomposition',
+               2,
+               None,
+               '___sec25'),
+              ('Principal Component Analysis', 2, None, '___sec26'),
+              ('PCA and scikit-learn', 2, None, '___sec27'),
+              ('Back to the Cancer Data', 2, None, '___sec28'),
+              ('More on the PCA', 2, None, '___sec29'),
+              ('Incremental PCA', 2, None, '___sec30'),
+              ('Randomized PCA', 2, None, '___sec31'),
+              ('Kernel PCA', 2, None, '___sec32'),
+              ('LLE', 2, None, '___sec33'),
+              ('Other techniques', 2, None, '___sec34')]}
 end of tocinfo -->
 
 
@@ -175,15 +180,16 @@ MathJax.Hub.Config({
       
  • Proof of the PCA Theorem
    • 
       
  • PCA Proof continued
    • 
       
  • The final step
    • 
-      
  • Principal Component Analysis
    • 
-      
  • PCA and scikit-learn
    • 
-      
  • Back to the Cancer Data
    • 
-      
  • More on the PCA
    • 
-      
  • Incremental PCA
    • 
-      
  • Randomized PCA
    • 
-      
  • Kernel PCA
    • 
-      
  • LLE
    • 
-      
  • Other techniques
    • 
+      
  • Geometric Interpretation and link with Singular Value Decomposition
    • 
+      
  • Principal Component Analysis
    • 
+      
  • PCA and scikit-learn
    • 
+      
  • Back to the Cancer Data
    • 
+      
  • More on the PCA
    • 
+      
  • Incremental PCA
    • 
+      
  • Randomized PCA
    • 
+      
  • Kernel PCA
    • 
+      
  • LLE
    • 
+      
  • Other techniques
    • 
 
         
       
@@ -199,15 +205,33 @@ MathJax.Hub.Config({
 
 
 
-
    Incremental PCA 
    
+
    More on the PCA 
    
 
 
    
-One problem with the preceding implementation of PCA is that it requires the whole training set to fit in
-memory in order for the SVD algorithm to run. Fortunately, Incremental PCA (IPCA) algorithms have
-been developed: you can split the training set into mini-batches and feed an IPCA algorithm one minibatch
-at a time. This is useful for large training sets, and also to apply PCA online (i.e., on the fly, as new
-instances arrive).
+Instead of arbitrarily choosing the number of dimensions to reduce down to, it is generally preferable to
+choose the number of dimensions that add up to a sufficiently large portion of the variance (e.g., 95%).
+Unless, of course, you are reducing dimensionality for data visualization — in that case you will
+generally want to reduce the dimensionality down to 2 or 3.
+The following code computes PCA without reducing dimensionality, then computes the minimum number
+of dimensions required to preserve 95% of the training set’s variance:
+
    
 
+
+
    pca = PCA()
+pca.fit(X)
+cumsum = np.cumsum(pca.explained_variance_ratio_)
+d = np.argmax(cumsum >= 0.95) + 1
+
    
+
    
+You could then set \( n\_components=d \) and run PCA again. However, there is a much better option: instead
+of specifying the number of principal components you want to preserve, you can set \( n\_components \) to be
+a float between 0.0 and 1.0, indicating the ratio of variance you wish to preserve:
+
    
+
+
+
    pca = PCA(n_components=0.95)
+X_reduced = pca.fit_transform(X)
+
    
 
    
 
    
 
@@ -228,6 +252,7 @@ instances arrive).
   
  • 30
    • 
   
  • 31
    • 
   
  • 32
    • 
+  
  • 33
    • 
   
  • »
    • 
 
 
diff --git a/doc/pub/DimRed/html/._DimRed-bs028.html b/doc/pub/DimRed/html/._DimRed-bs028.html
index 0c9c7f308..0e6a50276 100644
--- a/doc/pub/DimRed/html/._DimRed-bs028.html
+++ b/doc/pub/DimRed/html/._DimRed-bs028.html
@@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source
               ('Proof of the PCA Theorem', 2, None, '___sec22'),
               ('PCA Proof continued', 2, None, '___sec23'),
               ('The final step', 2, None, '___sec24'),
-              ('Principal Component Analysis', 2, None, '___sec25'),
-              ('PCA and scikit-learn', 2, None, '___sec26'),
-              ('Back to the Cancer Data', 2, None, '___sec27'),
-              ('More on the PCA', 2, None, '___sec28'),
-              ('Incremental PCA', 2, None, '___sec29'),
-              ('Randomized PCA', 2, None, '___sec30'),
-              ('Kernel PCA', 2, None, '___sec31'),
-              ('LLE', 2, None, '___sec32'),
-              ('Other techniques', 2, None, '___sec33')]}
+              ('Geometric Interpretation and link with Singular Value '
+               'Decomposition',
+               2,
+               None,
+               '___sec25'),
+              ('Principal Component Analysis', 2, None, '___sec26'),
+              ('PCA and scikit-learn', 2, None, '___sec27'),
+              ('Back to the Cancer Data', 2, None, '___sec28'),
+              ('More on the PCA', 2, None, '___sec29'),
+              ('Incremental PCA', 2, None, '___sec30'),
+              ('Randomized PCA', 2, None, '___sec31'),
+              ('Kernel PCA', 2, None, '___sec32'),
+              ('LLE', 2, None, '___sec33'),
+              ('Other techniques', 2, None, '___sec34')]}
 end of tocinfo -->
 
 
@@ -175,15 +180,16 @@ MathJax.Hub.Config({
       
  • Proof of the PCA Theorem
    • 
       
  • PCA Proof continued
    • 
       
  • The final step
    • 
-      
  • Principal Component Analysis
    • 
-      
  • PCA and scikit-learn
    • 
-      
  • Back to the Cancer Data
    • 
-      
  • More on the PCA
    • 
-      
  • Incremental PCA
    • 
-      
  • Randomized PCA
    • 
-      
  • Kernel PCA
    • 
-      
  • LLE
    • 
-      
  • Other techniques
    • 
+      
  • Geometric Interpretation and link with Singular Value Decomposition
    • 
+      
  • Principal Component Analysis
    • 
+      
  • PCA and scikit-learn
    • 
+      
  • Back to the Cancer Data
    • 
+      
  • More on the PCA
    • 
+      
  • Incremental PCA
    • 
+      
  • Randomized PCA
    • 
+      
  • Kernel PCA
    • 
+      
  • LLE
    • 
+      
  • Other techniques
    • 
 
         
       
@@ -199,18 +205,14 @@ MathJax.Hub.Config({
 
 
 
-
    Randomized PCA 
    
+
    Incremental PCA 
    
 
 
    
-Scikit-Learn offers yet another option to perform PCA, called Randomized PCA. This is a stochastic
-algorithm that quickly finds an approximation of the first d principal components. Its computational
-complexity is \( O(m \times d^2)+O(d^3) \), instead of \( O(m \times n^2) + O(n^3) \), so it is dramatically faster than the
-previous algorithms when \( d \) is much smaller than \( n \).
-
-
    
-
    
-
    
-
+One problem with the preceding implementation of PCA is that it requires the whole training set to fit in
+memory in order for the SVD algorithm to run. Fortunately, Incremental PCA (IPCA) algorithms have
+been developed: you can split the training set into mini-batches and feed an IPCA algorithm one minibatch
+at a time. This is useful for large training sets, and also to apply PCA online (i.e., on the fly, as new
+instances arrive).
 
 
    
 
    
@@ -231,6 +233,7 @@ previous algorithms when \( d \) is much smaller than \( n \).
   
  • 30
    • 
   
  • 31
    • 
   
  • 32
    • 
+  
  • 33
    • 
   
  • »
    • 
 
 
diff --git a/doc/pub/DimRed/html/._DimRed-bs029.html b/doc/pub/DimRed/html/._DimRed-bs029.html
index e605c052c..1c29effce 100644
--- a/doc/pub/DimRed/html/._DimRed-bs029.html
+++ b/doc/pub/DimRed/html/._DimRed-bs029.html
@@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source
               ('Proof of the PCA Theorem', 2, None, '___sec22'),
               ('PCA Proof continued', 2, None, '___sec23'),
               ('The final step', 2, None, '___sec24'),
-              ('Principal Component Analysis', 2, None, '___sec25'),
-              ('PCA and scikit-learn', 2, None, '___sec26'),
-              ('Back to the Cancer Data', 2, None, '___sec27'),
-              ('More on the PCA', 2, None, '___sec28'),
-              ('Incremental PCA', 2, None, '___sec29'),
-              ('Randomized PCA', 2, None, '___sec30'),
-              ('Kernel PCA', 2, None, '___sec31'),
-              ('LLE', 2, None, '___sec32'),
-              ('Other techniques', 2, None, '___sec33')]}
+              ('Geometric Interpretation and link with Singular Value '
+               'Decomposition',
+               2,
+               None,
+               '___sec25'),
+              ('Principal Component Analysis', 2, None, '___sec26'),
+              ('PCA and scikit-learn', 2, None, '___sec27'),
+              ('Back to the Cancer Data', 2, None, '___sec28'),
+              ('More on the PCA', 2, None, '___sec29'),
+              ('Incremental PCA', 2, None, '___sec30'),
+              ('Randomized PCA', 2, None, '___sec31'),
+              ('Kernel PCA', 2, None, '___sec32'),
+              ('LLE', 2, None, '___sec33'),
+              ('Other techniques', 2, None, '___sec34')]}
 end of tocinfo -->
 
 
@@ -175,15 +180,16 @@ MathJax.Hub.Config({
       
  • Proof of the PCA Theorem
    • 
       
  • PCA Proof continued
    • 
       
  • The final step
    • 
-      
  • Principal Component Analysis
    • 
-      
  • PCA and scikit-learn
    • 
-      
  • Back to the Cancer Data
    • 
-      
  • More on the PCA
    • 
-      
  • Incremental PCA
    • 
-      
  • Randomized PCA
    • 
-      
  • Kernel PCA
    • 
-      
  • LLE
    • 
-      
  • Other techniques
    • 
+      
  • Geometric Interpretation and link with Singular Value Decomposition
    • 
+      
  • Principal Component Analysis
    • 
+      
  • PCA and scikit-learn
    • 
+      
  • Back to the Cancer Data
    • 
+      
  • More on the PCA
    • 
+      
  • Incremental PCA
    • 
+      
  • Randomized PCA
    • 
+      
  • Kernel PCA
    • 
+      
  • LLE
    • 
+      
  • Other techniques
    • 
 
         
       
@@ -199,28 +205,14 @@ MathJax.Hub.Config({
 
 
 
-
    Kernel PCA 
    
-
    
-
    
-
     
+
    Randomized PCA 
    
 
 
    
-The kernel trick is a mathematical technique that implicitly maps instances into a
-very high-dimensional space (called the feature space), enabling nonlinear classification and regression
-with Support Vector Machines. Recall that a linear decision boundary in the high-dimensional feature
-space corresponds to a complex nonlinear decision boundary in the original space.
-It turns out that the same trick can be applied to PCA, making it possible to perform complex nonlinear
-projections for dimensionality reduction. This is called Kernel PCA (kPCA). It is often good at
-preserving clusters of instances after projection, or sometimes even unrolling datasets that lie close to a
-twisted manifold.
-For example, the following code uses Scikit-Learn’s KernelPCA class to perform kPCA with an
-
    
+Scikit-Learn offers yet another option to perform PCA, called Randomized PCA. This is a stochastic
+algorithm that quickly finds an approximation of the first d principal components. Its computational
+complexity is \( O(m \times d^2)+O(d^3) \), instead of \( O(m \times n^2) + O(n^3) \), so it is dramatically faster than the
+previous algorithms when \( d \) is much smaller than \( n \).
 
-
-
    from sklearn.decomposition import KernelPCA
-rbf_pca = KernelPCA(n_components = 2, kernel="rbf", gamma=0.04)
-X_reduced = rbf_pca.fit_transform(X)
-
    
 
    
 
    
 
    
@@ -244,6 +236,7 @@ X_reduced = rbf_pca30
   
  • 31
    • 
   
  • 32
    • 
+  
  • 33
    • 
   
  • »
    • 
 
 
diff --git a/doc/pub/DimRed/html/._DimRed-bs030.html b/doc/pub/DimRed/html/._DimRed-bs030.html
index 283c55794..f9bca9bdc 100644
--- a/doc/pub/DimRed/html/._DimRed-bs030.html
+++ b/doc/pub/DimRed/html/._DimRed-bs030.html
@@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source
               ('Proof of the PCA Theorem', 2, None, '___sec22'),
               ('PCA Proof continued', 2, None, '___sec23'),
               ('The final step', 2, None, '___sec24'),
-              ('Principal Component Analysis', 2, None, '___sec25'),
-              ('PCA and scikit-learn', 2, None, '___sec26'),
-              ('Back to the Cancer Data', 2, None, '___sec27'),
-              ('More on the PCA', 2, None, '___sec28'),
-              ('Incremental PCA', 2, None, '___sec29'),
-              ('Randomized PCA', 2, None, '___sec30'),
-              ('Kernel PCA', 2, None, '___sec31'),
-              ('LLE', 2, None, '___sec32'),
-              ('Other techniques', 2, None, '___sec33')]}
+              ('Geometric Interpretation and link with Singular Value '
+               'Decomposition',
+               2,
+               None,
+               '___sec25'),
+              ('Principal Component Analysis', 2, None, '___sec26'),
+              ('PCA and scikit-learn', 2, None, '___sec27'),
+              ('Back to the Cancer Data', 2, None, '___sec28'),
+              ('More on the PCA', 2, None, '___sec29'),
+              ('Incremental PCA', 2, None, '___sec30'),
+              ('Randomized PCA', 2, None, '___sec31'),
+              ('Kernel PCA', 2, None, '___sec32'),
+              ('LLE', 2, None, '___sec33'),
+              ('Other techniques', 2, None, '___sec34')]}
 end of tocinfo -->
 
 
@@ -175,15 +180,16 @@ MathJax.Hub.Config({
       
  • Proof of the PCA Theorem
    • 
       
  • PCA Proof continued
    • 
       
  • The final step
    • 
-      
  • Principal Component Analysis
    • 
-      
  • PCA and scikit-learn
    • 
-      
  • Back to the Cancer Data
    • 
-      
  • More on the PCA
    • 
-      
  • Incremental PCA
    • 
-      
  • Randomized PCA
    • 
-      
  • Kernel PCA
    • 
-      
  • LLE
    • 
-      
  • Other techniques
    • 
+      
  • Geometric Interpretation and link with Singular Value Decomposition
    • 
+      
  • Principal Component Analysis
    • 
+      
  • PCA and scikit-learn
    • 
+      
  • Back to the Cancer Data
    • 
+      
  • More on the PCA
    • 
+      
  • Incremental PCA
    • 
+      
  • Randomized PCA
    • 
+      
  • Kernel PCA
    • 
+      
  • LLE
    • 
+      
  • Other techniques
    • 
 
         
       
@@ -199,14 +205,32 @@ MathJax.Hub.Config({
 
 
 
-
    LLE 
    
+
    Kernel PCA 
    
+
    
+
    
+
     
 
 
    
-Locally Linear Embedding (LLE) is another very powerful nonlinear dimensionality reduction
-(NLDR) technique. It is a Manifold Learning technique that does not rely on projections like the previous
-algorithms. In a nutshell, LLE works by first measuring how each training instance linearly relates to its
-closest neighbors (c.n.), and then looking for a low-dimensional representation of the training set where
-these local relationships are best preserved (more details shortly).
+The kernel trick is a mathematical technique that implicitly maps instances into a
+very high-dimensional space (called the feature space), enabling nonlinear classification and regression
+with Support Vector Machines. Recall that a linear decision boundary in the high-dimensional feature
+space corresponds to a complex nonlinear decision boundary in the original space.
+It turns out that the same trick can be applied to PCA, making it possible to perform complex nonlinear
+projections for dimensionality reduction. This is called Kernel PCA (kPCA). It is often good at
+preserving clusters of instances after projection, or sometimes even unrolling datasets that lie close to a
+twisted manifold.
+For example, the following code uses Scikit-Learn’s KernelPCA class to perform kPCA with an
+
    
+
+
+
    from sklearn.decomposition import KernelPCA
+rbf_pca = KernelPCA(n_components = 2, kernel="rbf", gamma=0.04)
+X_reduced = rbf_pca.fit_transform(X)
+
    
+
    
+
    
+
    
+
 
 
    
 
    
@@ -225,6 +249,7 @@ these local relationships are best preserved (more details shortly).
   
  • 30
    • 
   
  • 31
    • 
   
  • 32
    • 
+  
  • 33
    • 
   
  • »
    • 
 
 
diff --git a/doc/pub/DimRed/html/DimRed-bs.html b/doc/pub/DimRed/html/DimRed-bs.html
index c0e2e1fdb..fcef4f476 100644
--- a/doc/pub/DimRed/html/DimRed-bs.html
+++ b/doc/pub/DimRed/html/DimRed-bs.html
@@ -104,15 +104,20 @@ Automatically generated HTML file from DocOnce source
               ('Proof of the PCA Theorem', 2, None, '___sec22'),
               ('PCA Proof continued', 2, None, '___sec23'),
               ('The final step', 2, None, '___sec24'),
-              ('Principal Component Analysis', 2, None, '___sec25'),
-              ('PCA and scikit-learn', 2, None, '___sec26'),
-              ('Back to the Cancer Data', 2, None, '___sec27'),
-              ('More on the PCA', 2, None, '___sec28'),
-              ('Incremental PCA', 2, None, '___sec29'),
-              ('Randomized PCA', 2, None, '___sec30'),
-              ('Kernel PCA', 2, None, '___sec31'),
-              ('LLE', 2, None, '___sec32'),
-              ('Other techniques', 2, None, '___sec33')]}
+              ('Geometric Interpretation and link with Singular Value '
+               'Decomposition',
+               2,
+               None,
+               '___sec25'),
+              ('Principal Component Analysis', 2, None, '___sec26'),
+              ('PCA and scikit-learn', 2, None, '___sec27'),
+              ('Back to the Cancer Data', 2, None, '___sec28'),
+              ('More on the PCA', 2, None, '___sec29'),
+              ('Incremental PCA', 2, None, '___sec30'),
+              ('Randomized PCA', 2, None, '___sec31'),
+              ('Kernel PCA', 2, None, '___sec32'),
+              ('LLE', 2, None, '___sec33'),
+              ('Other techniques', 2, None, '___sec34')]}
 end of tocinfo -->
 
 
@@ -175,15 +180,16 @@ MathJax.Hub.Config({
       
  • Proof of the PCA Theorem
    • 
       
  • PCA Proof continued
    • 
       
  • The final step
    • 
-      
  • Principal Component Analysis
    • 
-      
  • PCA and scikit-learn
    • 
-      
  • Back to the Cancer Data
    • 
-      
  • More on the PCA
    • 
-      
  • Incremental PCA
    • 
-      
  • Randomized PCA
    • 
-      
  • Kernel PCA
    • 
-      
  • LLE
    • 
-      
  • Other techniques
    • 
+      
  • Geometric Interpretation and link with Singular Value Decomposition
    • 
+      
  • Principal Component Analysis
    • 
+      
  • PCA and scikit-learn
    • 
+      
  • Back to the Cancer Data
    • 
+      
  • More on the PCA
    • 
+      
  • Incremental PCA
    • 
+      
  • Randomized PCA
    • 
+      
  • Kernel PCA
    • 
+      
  • LLE
    • 
+      
  • Other techniques
    • 
 
         
       
@@ -218,7 +224,7 @@ MathJax.Hub.Config({
 
    [2] Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University
    
 
    
 
    
-
    Dec 28, 2019
     
+
    Dec 29, 2019
     
 
    
 
    
 
@@ -242,7 +248,7 @@ MathJax.Hub.Config({
   
  • 9
    • 
   
  • 10
    • 
   
  • ...
    • 
-  
  • 32
    • 
+  
  • 33
    • 
   
  • »
    • 
 
 
diff --git a/doc/pub/DimRed/html/DimRed-reveal.html b/doc/pub/DimRed/html/DimRed-reveal.html
index 589de6f51..9c2731ba5 100644
--- a/doc/pub/DimRed/html/DimRed-reveal.html
+++ b/doc/pub/DimRed/html/DimRed-reveal.html
@@ -148,7 +148,7 @@ MathJax.Hub.Config({
 
    [2] Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University
    
 
    
 
     
    
-
    Dec 28, 2019
     
+
    Dec 29, 2019
     
 
    
 
    
 
@@ -518,6 +518,15 @@ applications.
 
    
 
    Basic ideas of the Principal Component Analysis (PCA) 
    
 
+
    
+The principal component analysis deals with the problem of fitting a
+low-dimensional affine subspace \( S \) of dimension \( d \) much smaller than
+the totaldimension \( D \) of the problem at hand (our data
+set). Mathematically it can be formulated as a statistical problem or
+a geometric problem.  In our discussion of the theorem for the
+classical PCA, we will stay with a statistical approach. This is also
+what set the scene historically which for the PCA.
+
 
    
 We have a data set defined by a design/feature matrix \( \boldsymbol{X} \) (see below for its definition) 
 
@@ -534,7 +543,7 @@ We have a data set defined by a design/feature matrix \( \boldsymbol{X} \) (see
 
 
    
 Before we discuss the PCA theorem, we need to remind ourselves about
-the definition of the covariance and the correlation function.
+the definition of the covariance and the correlation function. These are quantities
 
 
    
 Suppose we have defined two vectors
@@ -606,7 +615,9 @@ In the above example this is the function we constructed using pandas.
 
    Correlation Function and Design/Feature Matrix 
    
 
 
    
-In our derivation of the various regression algorithms like Ordinary Least Squares or Ridge regression we defined the design/feature matrix \( \boldsymbol{X} \) as
+In our derivation of the various regression algorithms like Ordinary Least Squares or Ridge regression
+we defined the design/feature matrix \( \boldsymbol{X} \) as
+
 
     
    
 $$
 \boldsymbol{X}=\begin{bmatrix}
@@ -637,7 +648,11 @@ $$
 
     
    
 
 
    
-With these definitions, we can now rewrite our \( 2\times 2 \) correaltion/covariance matrix in terms of a moe general design/feature matrix \( \boldsymbol{X}\in {\mathbb{R}}^{n\times p} \). This leads to a \( p\times p \) covariance matrix for the vectors \( \boldsymbol{x}_i \) with \( i =0,1,\dots,p-1 \)
+With these definitions, we can now rewrite our \( 2\times 2 \)
+correaltion/covariance matrix in terms of a moe general design/feature
+matrix \( \boldsymbol{X}\in {\mathbb{R}}^{n\times p} \). This leads to a \( p\times p \)
+covariance matrix for the vectors \( \boldsymbol{x}_i \) with \( i=0,1,\dots,p-1 \)
+
 
     
    
 $$
 \boldsymbol{C}[\boldsymbol{x}] = \begin{bmatrix}
@@ -843,8 +858,8 @@ columns since all matrix elements in the design matrix were set to one
 
    
 This means that the variance for these elements will be zero and will
 cause problems when we set up the correlation matrix.  We can simply
-drop these elements as follows and then construct the correlation
-matrix.
+drop these elements and construct a correlation
+matrix without these elements.
 
    
 
 
@@ -1026,7 +1041,7 @@ X = np.random.multivariate_normal(mean, cov, n)
 Make thereafter a small Python code which plots the data. Note that the function multivariate returns also the covariance discussed above and that it is defined by dividing by \( n-1 \) instead of \( n \).
 
 
    
-Now we are going to implement the PCA algorithm. We will break it down into sub-steps and across multiple cells.
+Now we are going to implement the PCA algorithm. We will break it down into various substeps.
 
 
    Compute the sample mean and center the data 
    
 
@@ -1103,8 +1118,11 @@ Finally, try out your own PCA function with other data sets.
 
    Classical PCA Theorem   
    
 
 
    
-We assume now that we have a design matrix \( \boldsymbol{X} \) which has been centered as discussed above. For the sake of simplicity we skip the overline symbol. The matrix is defined in terms of the various column vectors \( [\boldsymbol{x}_0,\boldsymbol{x}_1,\dots, \boldsymbol{x}_{p-1}] \)
-each with dimension \( \boldsymbol{x}\in {\mathbb{R}}^{n} \).
+We assume now that we have a design matrix \( \boldsymbol{X} \) which has been
+centered as discussed above. For the sake of simplicity we skip the
+overline symbol. The matrix is defined in terms of the various column
+vectors \( [\boldsymbol{x}_0,\boldsymbol{x}_1,\dots, \boldsymbol{x}_{p-1}] \) each with dimension
+\( \boldsymbol{x}\in {\mathbb{R}}^{n} \).
 
 
    
 We assume also that we have an orthogonal transformation \( \boldsymbol{W}\in {\mathbb{R}}^{p\times p} \).  We define the reconstruction error (which is similar to the mean squared error we have seen before) as
@@ -1118,7 +1136,13 @@ with \( \overline{\boldsymbol{x}}_i = \boldsymbol{W}\boldsymbol{z}_i \), where \
 \( \boldsymbol{Z}\in{\mathbb{R}}^{p\times n} \).  When doing PCA we want to reduce this dimensionality.
 
 
    
-The PCA theorem states that minimizing the above reconstruction error corresponds to setting \( \boldsymbol{W}=\boldsymbol{S} \), the orthogonal matrix which diagonalizes the empirical covariance(correlation) matrix. The optimal low-dimensional encoding of the data is then given by a set of vectors \( \boldsymbol{z}_i \) with at most \( l \) vectors, with \( l < < p \), defined by the orthogonal projection of the data onto the columns spanned by the eigenvectors of the covariance(correlations matrix).
+The PCA theorem states that minimizing the above reconstruction error
+corresponds to setting \( \boldsymbol{W}=\boldsymbol{S} \), the orthogonal matrix which
+diagonalizes the empirical covariance(correlation) matrix. The optimal
+low-dimensional encoding of the data is then given by a set of vectors
+\( \boldsymbol{z}_i \) with at most \( l \) vectors, with \( l < < p \), defined by the
+orthogonal projection of the data onto the columns spanned by the
+eigenvectors of the covariance(correlations matrix).
 
 
 
@@ -1195,7 +1219,10 @@ $$
 
     
    
 
 
    
-We are almost there, we have obtained a relation between minimizing the reconstruction error and the variance and the covariance matrix. Minimizing the error is equivalent to maximizing the variance of the projected data.
+We are almost there, we have obtained a relation between minimizing
+the reconstruction error and the variance and the covariance
+matrix. Minimizing the error is equivalent to maximizing the variance
+of the projected data.
 
 
 
@@ -1255,11 +1282,22 @@ our basis of eigenvectors is orthogonal, see Principal Component Analysis 
+
    Geometric Interpretation and link with Singular Value Decomposition 
    
+
+
    
+This material will be added by mid January 2020.
+
+
+
+
    
+
    Principal Component Analysis 
    
 
    
 
 
    
@@ -1316,7 +1354,7 @@ X2D = X_centered.dot(W2)
 
 
 
    
-
    PCA and scikit-learn 
    
+
    PCA and scikit-learn 
    
 
 
    
 Scikit-Learn’s PCA class implements PCA using SVD decomposition just like we did before. The
@@ -1348,7 +1386,7 @@ variance that lies along the axis of each principal component.
 
 
 
    
-
    Back to the Cancer Data 
    
+
    Back to the Cancer Data 
    
 We can now repeat the above but applied to real data, in this case our breast cancer data.
 Here we compute performance scores on the training data using logistic regression.
 
    
@@ -1389,7 +1427,7 @@ We see that our training data after the PCA decomposition has a performance simi
 
 
 
    
-
    More on the PCA 
    
+
    More on the PCA 
    
 
 
    
 Instead of arbitrarily choosing the number of dimensions to reduce down to, it is generally preferable to
@@ -1420,7 +1458,7 @@ X_reduced = pca.fit_transform(X)
 
 
 
    
-
    Incremental PCA 
    
+
    Incremental PCA 
    
 
 
    
 One problem with the preceding implementation of PCA is that it requires the whole training set to fit in
@@ -1432,7 +1470,7 @@ instances arrive).
 
 
 
    
-
    Randomized PCA 
    
+
    Randomized PCA 
    
 
 
    
 Scikit-Learn offers yet another option to perform PCA, called Randomized PCA. This is a stochastic
@@ -1446,7 +1484,7 @@ previous algorithms when \( d \) is much smaller than \( n \).
 
 
 
    
-
    Kernel PCA 
    
+
    Kernel PCA 
    
 
    
 
 
    
@@ -1472,7 +1510,7 @@ X_reduced = rbf_pca.fit_transform(X)
 
 
 
    
-
    LLE 
    
+
    LLE 
    
 
 
    
 Locally Linear Embedding (LLE) is another very powerful nonlinear dimensionality reduction
@@ -1484,7 +1522,7 @@ these local relationships are best preserved (more details shortly).
 
 
 
    
-
    Other techniques 
    
+
    Other techniques 
    
 
 
    
 There are many other dimensionality reduction techniques, several of which are available in Scikit-Learn.
diff --git a/doc/pub/DimRed/html/DimRed-solarized.html b/doc/pub/DimRed/html/DimRed-solarized.html
index a70ae5c1a..af3fb6759 100644
--- a/doc/pub/DimRed/html/DimRed-solarized.html
+++ b/doc/pub/DimRed/html/DimRed-solarized.html
@@ -124,15 +124,20 @@ div { text-align: justify; text-justify: inter-word; }
               ('Proof of the PCA Theorem', 2, None, '___sec22'),
               ('PCA Proof continued', 2, None, '___sec23'),
               ('The final step', 2, None, '___sec24'),
-              ('Principal Component Analysis', 2, None, '___sec25'),
-              ('PCA and scikit-learn', 2, None, '___sec26'),
-              ('Back to the Cancer Data', 2, None, '___sec27'),
-              ('More on the PCA', 2, None, '___sec28'),
-              ('Incremental PCA', 2, None, '___sec29'),
-              ('Randomized PCA', 2, None, '___sec30'),
-              ('Kernel PCA', 2, None, '___sec31'),
-              ('LLE', 2, None, '___sec32'),
-              ('Other techniques', 2, None, '___sec33')]}
+              ('Geometric Interpretation and link with Singular Value '
+               'Decomposition',
+               2,
+               None,
+               '___sec25'),
+              ('Principal Component Analysis', 2, None, '___sec26'),
+              ('PCA and scikit-learn', 2, None, '___sec27'),
+              ('Back to the Cancer Data', 2, None, '___sec28'),
+              ('More on the PCA', 2, None, '___sec29'),
+              ('Incremental PCA', 2, None, '___sec30'),
+              ('Randomized PCA', 2, None, '___sec31'),
+              ('Kernel PCA', 2, None, '___sec32'),
+              ('LLE', 2, None, '___sec33'),
+              ('Other techniques', 2, None, '___sec34')]}
 end of tocinfo -->
 
 
@@ -174,7 +179,7 @@ MathJax.Hub.Config({
 
    [2] Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University
    
 
    
 
    
-
    Dec 28, 2019
     
+
    Dec 29, 2019
     
 
    
 
    
 









    
@@ -541,6 +546,15 @@ applications.
 
 
    Basic ideas of the Principal Component Analysis (PCA) 
    
 
+
    
+The principal component analysis deals with the problem of fitting a
+low-dimensional affine subspace \( S \) of dimension \( d \) much smaller than
+the totaldimension \( D \) of the problem at hand (our data
+set). Mathematically it can be formulated as a statistical problem or
+a geometric problem.  In our discussion of the theorem for the
+classical PCA, we will stay with a statistical approach. This is also
+what set the scene historically which for the PCA.
+
 
    
 We have a data set defined by a design/feature matrix \( \boldsymbol{X} \) (see below for its definition) 
 
@@ -556,7 +570,7 @@ We have a data set defined by a design/feature matrix \( \boldsymbol{X} \) (see
 
 
    
 Before we discuss the PCA theorem, we need to remind ourselves about
-the definition of the covariance and the correlation function.
+the definition of the covariance and the correlation function. These are quantities
 
 
    
 Suppose we have defined two vectors
@@ -616,7 +630,9 @@ In the above example this is the function we constructed using pandas.
 
    Correlation Function and Design/Feature Matrix 
    
 
 
    
-In our derivation of the various regression algorithms like Ordinary Least Squares or Ridge regression we defined the design/feature matrix \( \boldsymbol{X} \) as
+In our derivation of the various regression algorithms like Ordinary Least Squares or Ridge regression
+we defined the design/feature matrix \( \boldsymbol{X} \) as
+
 $$
 \boldsymbol{X}=\begin{bmatrix}
 x_{0,0} & x_{0,1} & x_{0,2}& \dots & \dots x_{0,p-1}\\
@@ -641,7 +657,11 @@ $$
 $$
 
 
    
-With these definitions, we can now rewrite our \( 2\times 2 \) correaltion/covariance matrix in terms of a moe general design/feature matrix \( \boldsymbol{X}\in {\mathbb{R}}^{n\times p} \). This leads to a \( p\times p \) covariance matrix for the vectors \( \boldsymbol{x}_i \) with \( i =0,1,\dots,p-1 \)
+With these definitions, we can now rewrite our \( 2\times 2 \)
+correaltion/covariance matrix in terms of a moe general design/feature
+matrix \( \boldsymbol{X}\in {\mathbb{R}}^{n\times p} \). This leads to a \( p\times p \)
+covariance matrix for the vectors \( \boldsymbol{x}_i \) with \( i=0,1,\dots,p-1 \)
+
 $$
 \boldsymbol{C}[\boldsymbol{x}] = \begin{bmatrix}
 \mathrm{var}[\boldsymbol{x}_0] & \mathrm{cov}[\boldsymbol{x}_0,\boldsymbol{x}_1]  & \mathrm{cov}[\boldsymbol{x}_0,\boldsymbol{x}_2] & \dots & \dots & \mathrm{cov}[\boldsymbol{x}_0,\boldsymbol{x}_{p-1}]\\
@@ -840,8 +860,8 @@ columns since all matrix elements in the design matrix were set to one
 
    
 This means that the variance for these elements will be zero and will
 cause problems when we set up the correlation matrix.  We can simply
-drop these elements as follows and then construct the correlation
-matrix.
+drop these elements and construct a correlation
+matrix without these elements.
 
 
    
 









    
@@ -1001,7 +1021,7 @@ X = np.random.multivariate_normal(mean, cov, n)
 Make thereafter a small Python code which plots the data. Note that the function multivariate returns also the covariance discussed above and that it is defined by dividing by \( n-1 \) instead of \( n \).
 
 
    
-Now we are going to implement the PCA algorithm. We will break it down into sub-steps and across multiple cells.
+Now we are going to implement the PCA algorithm. We will break it down into various substeps.
 
 
    Compute the sample mean and center the data 
    
 
@@ -1071,8 +1091,11 @@ Finally, try out your own PCA function with other data sets.
 
    Classical PCA Theorem   
    
 
 
    
-We assume now that we have a design matrix \( \boldsymbol{X} \) which has been centered as discussed above. For the sake of simplicity we skip the overline symbol. The matrix is defined in terms of the various column vectors \( [\boldsymbol{x}_0,\boldsymbol{x}_1,\dots, \boldsymbol{x}_{p-1}] \)
-each with dimension \( \boldsymbol{x}\in {\mathbb{R}}^{n} \).
+We assume now that we have a design matrix \( \boldsymbol{X} \) which has been
+centered as discussed above. For the sake of simplicity we skip the
+overline symbol. The matrix is defined in terms of the various column
+vectors \( [\boldsymbol{x}_0,\boldsymbol{x}_1,\dots, \boldsymbol{x}_{p-1}] \) each with dimension
+\( \boldsymbol{x}\in {\mathbb{R}}^{n} \).
 
 
    
 We assume also that we have an orthogonal transformation \( \boldsymbol{W}\in {\mathbb{R}}^{p\times p} \).  We define the reconstruction error (which is similar to the mean squared error we have seen before) as
@@ -1084,7 +1107,13 @@ with \( \overline{\boldsymbol{x}}_i = \boldsymbol{W}\boldsymbol{z}_i \), where \
 \( \boldsymbol{Z}\in{\mathbb{R}}^{p\times n} \).  When doing PCA we want to reduce this dimensionality.
 
 
    
-The PCA theorem states that minimizing the above reconstruction error corresponds to setting \( \boldsymbol{W}=\boldsymbol{S} \), the orthogonal matrix which diagonalizes the empirical covariance(correlation) matrix. The optimal low-dimensional encoding of the data is then given by a set of vectors \( \boldsymbol{z}_i \) with at most \( l \) vectors, with \( l < < p \), defined by the orthogonal projection of the data onto the columns spanned by the eigenvectors of the covariance(correlations matrix).
+The PCA theorem states that minimizing the above reconstruction error
+corresponds to setting \( \boldsymbol{W}=\boldsymbol{S} \), the orthogonal matrix which
+diagonalizes the empirical covariance(correlation) matrix. The optimal
+low-dimensional encoding of the data is then given by a set of vectors
+\( \boldsymbol{z}_i \) with at most \( l \) vectors, with \( l < < p \), defined by the
+orthogonal projection of the data onto the columns spanned by the
+eigenvectors of the covariance(correlations matrix).
 
 
    
 









    
@@ -1145,7 +1174,10 @@ $$
 $$
 
 
    
-We are almost there, we have obtained a relation between minimizing the reconstruction error and the variance and the covariance matrix. Minimizing the error is equivalent to maximizing the variance of the projected data.
+We are almost there, we have obtained a relation between minimizing
+the reconstruction error and the variance and the covariance
+matrix. Minimizing the error is equivalent to maximizing the variance
+of the projected data.
 
 
    
 









    
@@ -1198,10 +1230,21 @@ discussion in chapter 12.2 of Murphy's text has also a nice link with
 the Singular Value Decomposition theorem. For categorical data, see
 chapter 12.4 and discussion therein.
 
+
    
+Additional part of the proof for the other eigenvectors will be added by mid January 2020.
+
 
    
 









    
 
-
    Principal Component Analysis 
    
+
    Geometric Interpretation and link with Singular Value Decomposition 
    
+
+
    
+This material will be added by mid January 2020.
+
+
    
+









    
+
+
    Principal Component Analysis 
    
 
    
 
 
    
@@ -1257,7 +1300,7 @@ X2D = X_centered.dot(W2)
 
    
 
 
-
    PCA and scikit-learn 
    
+
    PCA and scikit-learn 
    
 
 
    
 Scikit-Learn’s PCA class implements PCA using SVD decomposition just like we did before. The
@@ -1289,7 +1332,7 @@ variance that lies along the axis of each principal component.
 
    
 









    
 
-
    Back to the Cancer Data 
    
+
    Back to the Cancer Data 
    
 We can now repeat the above but applied to real data, in this case our breast cancer data.
 Here we compute performance scores on the training data using logistic regression.
 
    
@@ -1330,7 +1373,7 @@ We see that our training data after the PCA decomposition has a performance simi
 
    
 









    
 
-
    More on the PCA 
    
+
    More on the PCA 
    
 
 
    
 Instead of arbitrarily choosing the number of dimensions to reduce down to, it is generally preferable to
@@ -1360,7 +1403,7 @@ X_reduced = pca.fit_transform(X)
 
    
 









    
 
-
    Incremental PCA 
    
+
    Incremental PCA 
    
 
 
    
 One problem with the preceding implementation of PCA is that it requires the whole training set to fit in
@@ -1372,7 +1415,7 @@ instances arrive).
 
    
 









    
 
-
    Randomized PCA 
    
+
    Randomized PCA 
    
 
 
    
 Scikit-Learn offers yet another option to perform PCA, called Randomized PCA. This is a stochastic
@@ -1387,7 +1430,7 @@ previous algorithms when \( d \) is much smaller than \( n \).
 
    
 









    
 
-
    Kernel PCA 
    
+
    Kernel PCA 
    
 
    
 
 
    
@@ -1416,7 +1459,7 @@ X_reduced = rbf_pca.fit_transform(X)
 
    
 









    
 
-
    LLE 
    
+
    LLE 
    
 
 
    
 Locally Linear Embedding (LLE) is another very powerful nonlinear dimensionality reduction
@@ -1428,7 +1471,7 @@ these local relationships are best preserved (more details shortly).
 
    
 









    
 
-
    Other techniques 
    
+
    Other techniques 
    
 
 
    
 There are many other dimensionality reduction techniques, several of which are available in Scikit-Learn.
diff --git a/doc/pub/DimRed/html/DimRed.html b/doc/pub/DimRed/html/DimRed.html
index d300d5422..e9a79a081 100644
--- a/doc/pub/DimRed/html/DimRed.html
+++ b/doc/pub/DimRed/html/DimRed.html
@@ -129,15 +129,20 @@ div { text-align: justify; text-justify: inter-word; }
               ('Proof of the PCA Theorem', 2, None, '___sec22'),
               ('PCA Proof continued', 2, None, '___sec23'),
               ('The final step', 2, None, '___sec24'),
-              ('Principal Component Analysis', 2, None, '___sec25'),
-              ('PCA and scikit-learn', 2, None, '___sec26'),
-              ('Back to the Cancer Data', 2, None, '___sec27'),
-              ('More on the PCA', 2, None, '___sec28'),
-              ('Incremental PCA', 2, None, '___sec29'),
-              ('Randomized PCA', 2, None, '___sec30'),
-              ('Kernel PCA', 2, None, '___sec31'),
-              ('LLE', 2, None, '___sec32'),
-              ('Other techniques', 2, None, '___sec33')]}
+              ('Geometric Interpretation and link with Singular Value '
+               'Decomposition',
+               2,
+               None,
+               '___sec25'),
+              ('Principal Component Analysis', 2, None, '___sec26'),
+              ('PCA and scikit-learn', 2, None, '___sec27'),
+              ('Back to the Cancer Data', 2, None, '___sec28'),
+              ('More on the PCA', 2, None, '___sec29'),
+              ('Incremental PCA', 2, None, '___sec30'),
+              ('Randomized PCA', 2, None, '___sec31'),
+              ('Kernel PCA', 2, None, '___sec32'),
+              ('LLE', 2, None, '___sec33'),
+              ('Other techniques', 2, None, '___sec34')]}
 end of tocinfo -->
 
 
@@ -179,7 +184,7 @@ MathJax.Hub.Config({
 
    [2] Department of Physics and Astronomy and National Superconducting Cyclotron Laboratory, Michigan State University
    
 
    
 
    
-
    Dec 28, 2019
     
+
    Dec 29, 2019
     
 
    
 
    
 









    
@@ -546,6 +551,15 @@ applications.
 
 
    Basic ideas of the Principal Component Analysis (PCA) 
    
 
+
    
+The principal component analysis deals with the problem of fitting a
+low-dimensional affine subspace \( S \) of dimension \( d \) much smaller than
+the totaldimension \( D \) of the problem at hand (our data
+set). Mathematically it can be formulated as a statistical problem or
+a geometric problem.  In our discussion of the theorem for the
+classical PCA, we will stay with a statistical approach. This is also
+what set the scene historically which for the PCA.
+
 
    
 We have a data set defined by a design/feature matrix \( \boldsymbol{X} \) (see below for its definition) 
 
@@ -561,7 +575,7 @@ We have a data set defined by a design/feature matrix \( \boldsymbol{X} \) (see
 
 
    
 Before we discuss the PCA theorem, we need to remind ourselves about
-the definition of the covariance and the correlation function.
+the definition of the covariance and the correlation function. These are quantities
 
 
    
 Suppose we have defined two vectors
@@ -621,7 +635,9 @@ In the above example this is the function we constructed using pandas.
 
    Correlation Function and Design/Feature Matrix 
    
 
 
    
-In our derivation of the various regression algorithms like Ordinary Least Squares or Ridge regression we defined the design/feature matrix \( \boldsymbol{X} \) as
+In our derivation of the various regression algorithms like Ordinary Least Squares or Ridge regression
+we defined the design/feature matrix \( \boldsymbol{X} \) as
+
 $$
 \boldsymbol{X}=\begin{bmatrix}
 x_{0,0} & x_{0,1} & x_{0,2}& \dots & \dots x_{0,p-1}\\
@@ -646,7 +662,11 @@ $$
 $$
 
 
    
-With these definitions, we can now rewrite our \( 2\times 2 \) correaltion/covariance matrix in terms of a moe general design/feature matrix \( \boldsymbol{X}\in {\mathbb{R}}^{n\times p} \). This leads to a \( p\times p \) covariance matrix for the vectors \( \boldsymbol{x}_i \) with \( i =0,1,\dots,p-1 \)
+With these definitions, we can now rewrite our \( 2\times 2 \)
+correaltion/covariance matrix in terms of a moe general design/feature
+matrix \( \boldsymbol{X}\in {\mathbb{R}}^{n\times p} \). This leads to a \( p\times p \)
+covariance matrix for the vectors \( \boldsymbol{x}_i \) with \( i=0,1,\dots,p-1 \)
+
 $$
 \boldsymbol{C}[\boldsymbol{x}] = \begin{bmatrix}
 \mathrm{var}[\boldsymbol{x}_0] & \mathrm{cov}[\boldsymbol{x}_0,\boldsymbol{x}_1]  & \mathrm{cov}[\boldsymbol{x}_0,\boldsymbol{x}_2] & \dots & \dots & \mathrm{cov}[\boldsymbol{x}_0,\boldsymbol{x}_{p-1}]\\
@@ -845,8 +865,8 @@ columns since all matrix elements in the design matrix were set to one
 
    
 This means that the variance for these elements will be zero and will
 cause problems when we set up the correlation matrix.  We can simply
-drop these elements as follows and then construct the correlation
-matrix.
+drop these elements and construct a correlation
+matrix without these elements.
 
 
    
 









    
@@ -1006,7 +1026,7 @@ X = np.r
 Make thereafter a small Python code which plots the data. Note that the function multivariate returns also the covariance discussed above and that it is defined by dividing by \( n-1 \) instead of \( n \).
 
 
    
-Now we are going to implement the PCA algorithm. We will break it down into sub-steps and across multiple cells.
+Now we are going to implement the PCA algorithm. We will break it down into various substeps.
 
 
    Compute the sample mean and center the data 
    
 
@@ -1076,8 +1096,11 @@ Finally, try out your own PCA function with other data sets.
 
    Classical PCA Theorem   
    
 
 
    
-We assume now that we have a design matrix \( \boldsymbol{X} \) which has been centered as discussed above. For the sake of simplicity we skip the overline symbol. The matrix is defined in terms of the various column vectors \( [\boldsymbol{x}_0,\boldsymbol{x}_1,\dots, \boldsymbol{x}_{p-1}] \)
-each with dimension \( \boldsymbol{x}\in {\mathbb{R}}^{n} \).
+We assume now that we have a design matrix \( \boldsymbol{X} \) which has been
+centered as discussed above. For the sake of simplicity we skip the
+overline symbol. The matrix is defined in terms of the various column
+vectors \( [\boldsymbol{x}_0,\boldsymbol{x}_1,\dots, \boldsymbol{x}_{p-1}] \) each with dimension
+\( \boldsymbol{x}\in {\mathbb{R}}^{n} \).
 
 
    
 We assume also that we have an orthogonal transformation \( \boldsymbol{W}\in {\mathbb{R}}^{p\times p} \).  We define the reconstruction error (which is similar to the mean squared error we have seen before) as
@@ -1089,7 +1112,13 @@ with \( \overline{\boldsymbol{x}}_i = \boldsymbol{W}\boldsymbol{z}_i \), where \
 \( \boldsymbol{Z}\in{\mathbb{R}}^{p\times n} \).  When doing PCA we want to reduce this dimensionality.
 
 
    
-The PCA theorem states that minimizing the above reconstruction error corresponds to setting \( \boldsymbol{W}=\boldsymbol{S} \), the orthogonal matrix which diagonalizes the empirical covariance(correlation) matrix. The optimal low-dimensional encoding of the data is then given by a set of vectors \( \boldsymbol{z}_i \) with at most \( l \) vectors, with \( l < < p \), defined by the orthogonal projection of the data onto the columns spanned by the eigenvectors of the covariance(correlations matrix).
+The PCA theorem states that minimizing the above reconstruction error
+corresponds to setting \( \boldsymbol{W}=\boldsymbol{S} \), the orthogonal matrix which
+diagonalizes the empirical covariance(correlation) matrix. The optimal
+low-dimensional encoding of the data is then given by a set of vectors
+\( \boldsymbol{z}_i \) with at most \( l \) vectors, with \( l < < p \), defined by the
+orthogonal projection of the data onto the columns spanned by the
+eigenvectors of the covariance(correlations matrix).
 
 
    
 









    
@@ -1150,7 +1179,10 @@ $$
 $$
 
 
    
-We are almost there, we have obtained a relation between minimizing the reconstruction error and the variance and the covariance matrix. Minimizing the error is equivalent to maximizing the variance of the projected data.
+We are almost there, we have obtained a relation between minimizing
+the reconstruction error and the variance and the covariance
+matrix. Minimizing the error is equivalent to maximizing the variance
+of the projected data.
 
 
    
 









    
@@ -1203,10 +1235,21 @@ discussion in chapter 12.2 of Murphy's text has also a nice link with
 the Singular Value Decomposition theorem. For categorical data, see
 chapter 12.4 and discussion therein.
 
+
    
+Additional part of the proof for the other eigenvectors will be added by mid January 2020.
+
 
    
 









    
 
-
    Principal Component Analysis 
    
+
    Geometric Interpretation and link with Singular Value Decomposition 
    
+
+
    
+This material will be added by mid January 2020.
+
+
    
+









    
+
+
    Principal Component Analysis 
    
 
    
 
 
    
@@ -1262,7 +1305,7 @@ X2D = X_centered
 
 
-
    PCA and scikit-learn 
    
+
    PCA and scikit-learn 
    
 
 
    
 Scikit-Learn’s PCA class implements PCA using SVD decomposition just like we did before. The
@@ -1294,7 +1337,7 @@ variance that lies along the axis of each principal component.
 
    
 









    
 
-
    Back to the Cancer Data 
    
+
    Back to the Cancer Data 
    
 We can now repeat the above but applied to real data, in this case our breast cancer data.
 Here we compute performance scores on the training data using logistic regression.
 
    
@@ -1335,7 +1378,7 @@ We see that our training data after the PCA decomposition has a performance simi
 
    
 









    
 
-
    More on the PCA 
    
+
    More on the PCA 
    
 
 
    
 Instead of arbitrarily choosing the number of dimensions to reduce down to, it is generally preferable to
@@ -1365,7 +1408,7 @@ X_reduced = pca
 
    
 









    
 
-
    Incremental PCA 
    
+
    Incremental PCA 
    
 
 
    
 One problem with the preceding implementation of PCA is that it requires the whole training set to fit in
@@ -1377,7 +1420,7 @@ instances arrive).
 
    
 









    
 
-
    Randomized PCA 
    
+
    Randomized PCA 
    
 
 
    
 Scikit-Learn offers yet another option to perform PCA, called Randomized PCA. This is a stochastic
@@ -1392,7 +1435,7 @@ previous algorithms when \( d \) is much smaller than \( n \).
 
    
 









    
 
-
    Kernel PCA 
    
+
    Kernel PCA 
    
 
    
 
 
    
@@ -1421,7 +1464,7 @@ X_reduced = rbf_pcaLLE 
+
    LLE 
    
 
 
    
 Locally Linear Embedding (LLE) is another very powerful nonlinear dimensionality reduction
@@ -1433,7 +1476,7 @@ these local relationships are best preserved (more details shortly).
 
    
 









    
 
-
    Other techniques 
    
+
    Other techniques 
    
 
 
    
 There are many other dimensionality reduction techniques, several of which are available in Scikit-Learn.
diff --git a/doc/pub/DimRed/ipynb/DimRed.ipynb b/doc/pub/DimRed/ipynb/DimRed.ipynb
index f688683cf..ec03b06a5 100644
--- a/doc/pub/DimRed/ipynb/DimRed.ipynb
+++ b/doc/pub/DimRed/ipynb/DimRed.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: **Dec 28, 2019**\n",
+    "Date: **Dec 29, 2019**\n",
     "\n",
     "Copyright 1999-2019, Morten Hjorth-Jensen. Released under CC Attribution-NonCommercial 4.0 license\n",
     "\n",
@@ -409,6 +409,14 @@
     "\n",
     "## Basic ideas of the Principal Component Analysis (PCA)\n",
     "\n",
+    "The principal component analysis deals with the problem of fitting a\n",
+    "low-dimensional affine subspace $S$ of dimension $d$ much smaller than\n",
+    "the totaldimension $D$ of the problem at hand (our data\n",
+    "set). Mathematically it can be formulated as a statistical problem or\n",
+    "a geometric problem.  In our discussion of the theorem for the\n",
+    "classical PCA, we will stay with a statistical approach. This is also\n",
+    "what set the scene historically which for the PCA.\n",
+    "\n",
     "We have a data set defined by a design/feature matrix $\\boldsymbol{X}$ (see below for its definition) \n",
     "* Each data point is determined by $p$ extrinsic (measurement) variables\n",
     "\n",
@@ -419,7 +427,7 @@
     "## Introducing the Covariance and Correlation functions\n",
     "\n",
     "Before we discuss the PCA theorem, we need to remind ourselves about\n",
-    "the definition of the covariance and the correlation function.\n",
+    "the definition of the covariance and the correlation function. These are quantities \n",
     "\n",
     "Suppose we have defined two vectors\n",
     "$\\hat{x}$ and $\\hat{y}$ with $n$ elements each. The covariance matrix $\\boldsymbol{C}$ is defined as"
@@ -535,7 +543,8 @@
     "\n",
     "## Correlation Function and Design/Feature Matrix\n",
     "\n",
-    "In our derivation of the various regression algorithms like Ordinary Least Squares or Ridge regression we defined the design/feature matrix $\\boldsymbol{X}$ as"
+    "In our derivation of the various regression algorithms like **Ordinary Least Squares** or **Ridge regression**\n",
+    "we defined the design/feature matrix $\\boldsymbol{X}$ as"
    ]
   },
   {
@@ -592,7 +601,10 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "With these definitions, we can now rewrite our $2\\times 2$ correaltion/covariance matrix in terms of a moe general design/feature matrix $\\boldsymbol{X}\\in {\\mathbb{R}}^{n\\times p}$. This leads to a $p\\times p$ covariance matrix for the vectors $\\boldsymbol{x}_i$ with $i =0,1,\\dots,p-1$"
+    "With these definitions, we can now rewrite our $2\\times 2$\n",
+    "correaltion/covariance matrix in terms of a moe general design/feature\n",
+    "matrix $\\boldsymbol{X}\\in {\\mathbb{R}}^{n\\times p}$. This leads to a $p\\times p$\n",
+    "covariance matrix for the vectors $\\boldsymbol{x}_i$ with $i=0,1,\\dots,p-1$"
    ]
   },
   {
@@ -848,8 +860,8 @@
     "\n",
     "This means that the variance for these elements will be zero and will\n",
     "cause problems when we set up the correlation matrix.  We can simply\n",
-    "drop these elements as follows and then construct the correlation\n",
-    "matrix. \n",
+    "drop these elements and construct a correlation\n",
+    "matrix without these elements. \n",
     "\n",
     "\n",
     "## Rewriting the Covariance and/or Correlation Matrix\n",
@@ -1104,7 +1116,7 @@
    "source": [
     "Make thereafter a small Python code which plots the data. Note that the function **multivariate** returns also the covariance discussed above and that it is defined by dividing by $n-1$ instead of $n$.\n",
     "\n",
-    "Now we are going to implement the PCA algorithm. We will break it down into sub-steps and across multiple cells.\n",
+    "Now we are going to implement the PCA algorithm. We will break it down into various substeps.\n",
     "\n",
     "### Compute the sample mean and center the data\n",
     "\n",
@@ -1208,8 +1220,11 @@
     "\n",
     "## Classical PCA Theorem\n",
     "\n",
-    "We assume now that we have a design matrix $\\boldsymbol{X}$ which has been centered as discussed above. For the sake of simplicity we skip the overline symbol. The matrix is defined in terms of the various column vectors $[\\boldsymbol{x}_0,\\boldsymbol{x}_1,\\dots, \\boldsymbol{x}_{p-1}]$\n",
-    "each with dimension $\\boldsymbol{x}\\in {\\mathbb{R}}^{n}$.\n",
+    "We assume now that we have a design matrix $\\boldsymbol{X}$ which has been\n",
+    "centered as discussed above. For the sake of simplicity we skip the\n",
+    "overline symbol. The matrix is defined in terms of the various column\n",
+    "vectors $[\\boldsymbol{x}_0,\\boldsymbol{x}_1,\\dots, \\boldsymbol{x}_{p-1}]$ each with dimension\n",
+    "$\\boldsymbol{x}\\in {\\mathbb{R}}^{n}$.\n",
     "\n",
     "We assume also that we have an orthogonal transformation $\\boldsymbol{W}\\in {\\mathbb{R}}^{p\\times p}$.  We define the reconstruction error (which is similar to the mean squared error we have seen before) as"
    ]
@@ -1230,7 +1245,13 @@
     "with $\\overline{\\boldsymbol{x}}_i = \\boldsymbol{W}\\boldsymbol{z}_i$, where $\\boldsymbol{z}_i$ is a row vector with dimension ${\\mathbb{R}}^{n}$ of the matrix\n",
     "$\\boldsymbol{Z}\\in{\\mathbb{R}}^{p\\times n}$.  When doing PCA we want to reduce this dimensionality. \n",
     "\n",
-    "The PCA theorem states that minimizing the above reconstruction error corresponds to setting $\\boldsymbol{W}=\\boldsymbol{S}$, the orthogonal matrix which diagonalizes the empirical covariance(correlation) matrix. The optimal low-dimensional encoding of the data is then given by a set of vectors $\\boldsymbol{z}_i$ with at most $l$ vectors, with $l << p$, defined by the orthogonal projection of the data onto the columns spanned by the eigenvectors of the covariance(correlations matrix).\n",
+    "The PCA theorem states that minimizing the above reconstruction error\n",
+    "corresponds to setting $\\boldsymbol{W}=\\boldsymbol{S}$, the orthogonal matrix which\n",
+    "diagonalizes the empirical covariance(correlation) matrix. The optimal\n",
+    "low-dimensional encoding of the data is then given by a set of vectors\n",
+    "$\\boldsymbol{z}_i$ with at most $l$ vectors, with $l << p$, defined by the\n",
+    "orthogonal projection of the data onto the columns spanned by the\n",
+    "eigenvectors of the covariance(correlations matrix).\n",
     "\n",
     "\n",
     "\n",
@@ -1371,7 +1392,10 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "We are almost there, we have obtained a relation between minimizing the reconstruction error and the variance and the covariance matrix. Minimizing the error is equivalent to maximizing the variance of the projected data. \n",
+    "We are almost there, we have obtained a relation between minimizing\n",
+    "the reconstruction error and the variance and the covariance\n",
+    "matrix. Minimizing the error is equivalent to maximizing the variance\n",
+    "of the projected data.\n",
     "\n",
     "## The final step\n",
     "\n",
@@ -1460,9 +1484,11 @@
     "the Singular Value Decomposition theorem. For categorical data, see\n",
     "chapter 12.4 and discussion therein.\n",
     "\n",
+    "Additional part of the proof for the other eigenvectors will be added by mid January 2020.\n",
     "\n",
+    "## Geometric Interpretation and link with Singular Value Decomposition\n",
     "\n",
-    "\n",
+    "This material will be added by mid January 2020.\n",
     "\n",
     "\n",
     "## Principal Component Analysis\n",
diff --git a/doc/pub/DimRed/ipynb/ipynb-DimRed-src.tar.gz b/doc/pub/DimRed/ipynb/ipynb-DimRed-src.tar.gz
index 2d4521bc7d318258c541cecd9a138e864c4e8a1e..3639478816bf1f89af9b8877b8273f05403c73f3 100644
GIT binary patch
delta 172
zcmV;d08{_J0lxtrABzY82NMZi00ZsM%?iRW3z(w{TgS|JN^=&<3iWnw9INXO
zdzM$=nSbKgNDI5e;2Nz!X(uYJFHp#~6g-mMpeo>6*r^3Z+x`0Hmp&-1*mJpgjir3Ie=2mk;0gbaX0vHDb|C_87vrq}e3IaAaldx?mf8AMGkK4!*exF~#
zM+%7AWFKU64+LxPI`9Gk9P9ua!0Q0C#IcA-;#j0Sp8WcJ)z!^I!;!}DAPMr+2fM4Q
z>-egwnM5C>B>Kx$;@g+kS1(?tc~poZlUa0q6Q$aUJe5&qav_z8uGi7K`1gfXakW?~
zrQ%hyZ@a^B)m3Yrf9N)K%e`IM9qKRKKeqM$gva5F0B8~S8na7=H0p$9Hdmk=yAy_Y3u&Dg&|j5%?$o~Re~t&oj2pBiExq@HW>c>iNlhkXW~O_7zF>!0A#zzv_c+V2nYSBs2D`za?6}=HprjsIeFHdmgLT=J+-s|j
z903qhq$>Vv337lP!j0`cuc$pK7aXYLL9{nIeJTz(a+Y|R18Q0905%S;Hg@i;in!#4
zf1+O9c8pc@y}t>4Z4RTwHvU*q;5j&Oz^q>i_B9asjn|*MCw*Sm?dsS%mGFm~M_nZ~
zA|ogwrWs$zgB_slcyQxbakJh$C|C?IOGDtwpSU-~RrG6Y*!m^7ZXZYU0FmgG{pXQ1
zXr9~>_bgYmZ9`Yp1qXvU95;y)S>eM@e?V&BP2(YXQ+h0i<|pX~W>(eScR{^t4#5m)
zW*lfSWMPbVK*^>Quxh-zhFWy?Fg4OY@^WCNn
z@rk|$BUx8nb=cMWkOO;3&tRD@gmP}r>F$gO!MEy}slGrLdk*M6ZP+2ELsvVWU4#@u
z4Ji`bx;-bNWtstIEZRb(`iU|ce}k&ABFjyrwG=Y<{sJ?VWRzNFuppArJr7)-SoTpJ
zGcMsmTeko76C1zrrrG+y%<&pipN>`Q^13ws4#}KCN^+?BZJ~QpF=!lk=#}I}H}Q4w
zuG1{ih;mAL(1PT$L2^#p&UDMi)R;-36BCrQCd;t#4?l1_r&I;cQ6slVe**%PU8rAb
zhjN_o9h{SVso{!MKoPj=>QH+xIsp)}(m$V9Ij{||{&hMZgMH9B{4xzXe?zGV*l8O=
zZb}+ARa@~~&E|O^{PqTnUoqBOmz^{1QMFXWM9!s)6y7>AtHL=W
z8ZKTI)tm%%dls7aJQ*PO+5at*Bs$&pNhE|m&dPB68AH7
z1+I|R-oDn&Y%af)fUhKvu!gk2YI?&7In`Lo5oT$Ov-wvIE&062)KDbRJj=u_7*$9#
zadX@|2HVStJjF((Ghw~jUxq5srb88ozE2r=z^tKz;6@!z(@|dlf4az83NzX3YedWC
z7%j)Tdhh3um9*8C<6g_3OU|XZZ1(5P1$78p6rng92NKpJXVasHgjK^K)*}K-a(GG5
zgVRllP8ZF*NL`+xson{;#dSG0e>yJG0EXaVrmXjG62K;?!&eE%TOWKJy?i8f^fx8f
z$HUBe6!vqfaIdc0aDrloB
zIXR>Um#lDA@~j;LC6YuBZzA^OP2_Cq2IdXi*GuQs9N$u;f2I#N|gFWPY#$U)C#M5{k)q?FNIC>6FZtlU6gQHSn&M}1ysw@o|wRd2NaOHe`|ct
z;rd5jr^bZROdv*6E}7^e1WC>kLRGO1h?eqB(_(#~Eo%~4$2Z@%RNIjP4e;Ff7Z
zk6P#XLOPVBe;4^O-zt4Gu;tQyOQCO??0=O2XOpwhG0!AFC7&QarSt{12HSfc`JGrT
zPzDQ-;87Ed*37a}q?8Jd%=`NpNLW(AXC+ojgWF}i@{Y6aNhMo-DR;NqYKG^+y4ng!
zs{nXDpNBlBB@Qq^hv&$S0cjq|xYqpT`s&kF|B`1lf1omw$^fuTwAx+0d!IyWe52
    iX`e!Kb`|67^dBM_1qVq(hYg$J;Ll$0wQkTWB_N}Eqs0d|K~+Nr%S
zF_Oen%0_E&@4@ub854r7i`-_=7g#SQb^e4UtkQ(}C~H!`te!cKAicNwACtK-k8%RB
z5lwE)fADHBqjiP+Q=gi}m?tB$f(m~K(@a>JR6<4=j7N)tsHRGWRin;-rfXglbTmjKhU792HPSm#`T%eR84CKXm
zh%^6t5T~9E;>=Sa&Rl{x^=uGF(OqMqrKR~dVO$`UQ%qD}0dg}EW#2|<420y2sQ72v
ze{!^HK7&E&rRk~Dp@4r+>HjZK5nuL?Hg|`*JAGKYP-s0~6cy+ez9=&Pe?^f!TNF{L
zs(e+HP9W`h;4C%vazSH6UgZCcf+jEIH&itLk%TujMRral5MGk&uP~a2)?!7T_+lh`
zaxr2@0sYaRHM!6!y^w05f_16M2B6%(1p&Oj-&&LU9VY`hH#L*7Z7F{oS&!Ss`F(!{
zA1NRj&wO*zMqir7MbZXElLZpM_Jb^O*F<$}D9LO0*Y}%))Q}n-D@$uR=*yCv^Sh6k
zPfN4=jAnnoq2c-0+nX1!na&DQC{Wq$UB(Cn?4lXVm@q^$$duFsxXf;6*_-_KLpd07
zE%WhW06PCTkRl(|)o6b{4(+Sae{4s&^q+pfblrTv9}I2p%8q$-1IxMZ`EOY-?3>C*
z^NDY&=5Hw%^U96>?At=8%4eei7~iV$)dxFd9X6Skqsidi?N1gP#v61vCt8_I2}!_s
zY*uzhDDsDT@qSv)mj3Ip!sXoCbg>L=i#z|EyM1QmfaP^H9!-Cn`Q^A;R*Sj+yj#?v
zeN?Zi@p?L{r=P=$qx!z|uSgo{Nkaxh!HfLImG7vQE%2%jJU%%DUW^yBS+iw>$r51-
z>#sh9mzGsBVimT}MytB|~`X4R@ZO*U?HNxKndkm^4+LUVL20xP(&)(40x4TLJ)xy!hp{SfE0mt
zfp=C9O#T-~EhrE`?AE+!ptKzLI2f8d1uR2_H0C*h#r}VrA8v2{xM>*6DJ5KFvQR`7
zIvdY!-n^sP1p9uKP6eA3*98a@fZ~pB~4v#@;ml8x$<#&8Y?8tB!>$^!Wh5z(l5S5_5VSq0xhzd|MFdrSqVpcn+X3$mEr&eU;s6Ik(P2T4
z$sFH!W(a?sx)$qb-}SvG}f^kw+kLnq273FvIfsuLWACKl?5vNMFsd0n;%Q-(g
z;E;b;l^Du!R&2XfAArPubY-!7w0{=9$D=!5E$U@TlbF>Y-LrNbrWC}O*4o*qENSDo
zhMxNL7Ll5@i=-eKDoU8q&Wm&OG;ZWSQi6pheX~~5#}mW!A+M<};^KQQ4}A&RGbRWX
z0i(ARL}EPvqD_$vHEB?1c2z}p1~L+y7yW-yEcKIOgKTK5dLf`_(h4b$h_zUM1G8g;
zgM8TKio1+JlCcCQkX#x?6eYC)*|I_=VJetKFk;%PghUU~DxqfBuoH;VVdRe?Tn+k&
z&e11AAu_kB=m(UaMze=$qqHBK*TpygjPyBj&iM5htuS6-gkG=5t0rb%FKzJd5rTi1
z4I1O(jY2%24mTRu!HA<JyI{^izWv>T?$Tx6r
zk~x|Gtz1_3^A|V*NibTi>ySwLtWn2<#HKj*qDFjeo>N3R}0q6}|WnR=NIDi0cb>Ww4;_^?!W8AhE
zHmHK8)zOW7iF|@?Eza5^p`3Uw-dPUX1Ubx@MNI64)B=OoR*`>C@tG63$;*Gz9>UTL
z1K}Zt{hXT4T3rtTXhqwII%$`YRhWr!G`BJ=1U{lbhG>K&Fhn(tm_9Tr-f1&OwjDE9
zpKaWN-P6}Z=45B
zfCG|9!?S3gUC@Z5pkYTt!*+i`BfbVSkb(x~6jQorpO=^(D(xT2J^Y&&%ta2p=5
zCqn)Z1RtUinA>XJjEK?$)foe@m)JvQIFbOLlnBM1B8%**WRYE!EV6G_7TGr`i*Qx4
z2v;SG@Xg91d{tSbm|j3v7`IVWv^b*GKMWPyV|!!50!2Z$T7Aw&C=-9(Nb^&IN|6+V
zQS$_)giD~FK&dYmrIeasdL|IA2-wS`q|b*EyAqVn$fcyhWBR5=j8a0DQ^Fi8d<>G%
zF3I@{=zodT(EF{1y%1rLc45U{GrbFUaSBiuQ&4dZl;rcG#I6J-bv~31D5yBQf<7<$
z=TuPMJA_oYZ9zZ;<>Y^G!V$70e}hfFGW(-IwxE%#ieagKiqipfG}{X;-;L5aqA
zt~04F@54)gJc>Z1uNk?hsct
zhuip4h60jmFP1HjGr|wZ%V{|)UHQ(+-pWZP
z?5!N+VXV8AbCZ9{$GTdTy`?7ZHm^P1tjpwu9Jj7OBiCGf{@0q<1fOczn~V=K{_!0195`w>bV-Hg|SqD5R|x^W}HGExyz9&JP2E
z`NG$UxYms)#1DIsx~)<%I9%GevI`CDDNW*vqhnXy36Zkx+()B3E0X){Bo#M|fRWuv
zuDoV6wM2Q*R9w26JzpamVjtSC!39EQ`LlysZU(Uq50F+BWJyw9V9U(YCyBv>nr3fOI={Ftl}5g8Py$M<`qK
zy-WsWK(%)$ZN<0qPl=?LVp>@A!0OOY$8hiP3+`pSq9?8z(|WrcD=a5akwT-ktU{SmF@}oUw5lc{&&)kI@DgPeO_8#E-~FYuI({I
zZ}G$0Pg_6PuP8k`3E4A>J9D{qwTC~g+V?9TK+AAj99e-azDOy=c5;WUe0wYh
z4aXiU;#ijAw@Z_*37{Tn-49MU1R&KHL`?gS3eI^$aa9J`QU3tJI41Mk+4%Ea={!sP
zbe!V;pI%>{W&+vNp`7Y?>_wbtkNY!DIvu$#XVAZoD>%ZZiEe&&_pO%MwEuRcf0t^1
z2X2t&(x5el=rF9E@x$%S{{TWH?P{|XDf9{gG%}Nclq!=oG!K7>%ZJ3}O*xfRIku{j
zs`Vad2|L4BA}1u3+4Zka<3$k^M$+uV@rfYVY&81ubra31xG$>WFLzbA{&{ov>NS>y
zMp_V2Y~B@^DI_r|1eb_VUTk*7oARG4MN6|&k9E8A^mTvx%L>TC+`D7D9hs~^ux0}DddQEVF
zG18n9EAJXJq!}yLL?VHim-m~l@6Emryd^LHUXOkA-jAP}@xhP2Ur}7bX6nJ;`+dQ{
zOAUk9E6u#rKpEyHshu>5@+zd21jSYb$Ve%ISpl*Vii>~E!RwLv&8xEQ#xQy?eHe`o
z^=L;r>P?l~pa3q*x^L=sYdnEhyefY+edkB-x;}CVG(T08^}`C6fobn}J%u-e73~uH
zsxj{frKopS%7?G8;HoITR)%zbZ6@KQAoDo&7+c4_CHIoIJ{$?iGq5F;-w
zswy5O4w1@V@;r>^#Qbx);Pz=rCasRJG3=XevCv;<8Hh8iP
zk515_0){B3QMeqsetQ4#SUY_a2Y`o^KE{7BP`Vf;9vpLKW*VGQ2Qzz!diBoKN+r0M>_;LjImo0aiwa(Y#6itO!y_%J
zax5#v2rHPA%+YQ#``4&qv{=umzV=i!LF@`~ETWpe&FWDwa>MC!Mh
zJVU^hrzQWG@x(R30@2Dl8ja=B#@2skopT!^xWWkI0kQ=nuGnOY~?O01f0EBYZj_e0Q7#bgwVu#D|FA{>^Kz%hRf{<-WN
z3C%3nFdb}PRk?aafi`I|BfF6q&M>oDkQxIac)Gwh@e3&pUlESKh#Ai47kDvnKnTG}
zxjYNI6%5vMGhDEiO$bXDN<0L@m}76N3;M7~EmHkvc!5j|V&#gh&Uxju0j$7XA(RB4
z66nJpWAU(MnTm8
zMa=$aGH3uRGC|$SlC$3k#C_^vmqA>35NFTjUw$S3Ld6GR4yb}SSn?C{Z)8+K38ydT
zU-5$cORp(mmVeomNRl4+)J5C_C!?C&27L5XEffD~TdW1DV*N2;Ax(cJ5#T}uF&Q%N
zA2L$d5nYyDFBc$(pOirLDCYv)hId(HcX;=5=)9nq8QCW&CJ`KeKEbmI2~(&&nSN15
zViz*fBpdsv%=?T|+2>IzyP%1W6#Rmh-o$KKge{@~UkO9qV1w`_0vD>)
zS+SrJnAcc^Vj)m|mCJvRw5pH-PfhaU(?heh-^33-$u;2^>fc9yj&}~w0(@Drkm~5=
zFGu#`FUOnek!1J2hL{n3l`Vi6Q7>ByXZCcq);yms>m)ERpKANhr=jJywa9&2FSDDM
z<(aCUrbh=hTb`0zILKSAaaKg&%qsD5R=lQJtVlJT*sN7{eL#O!tgkK4Z*d}dix0^!
z;zJe@AU8lg1s8YT=AT^IJJfGPsh4dY|ArCtB0Kx)y3g8JB3h|0rTOzh5FSI3rwP9q
zb>#)rC6XYa#1eH)Jm;HQ3xK*$y=~HDEKL~_grwf_#X*$}V3K9i?wW1wPpM_>p|}%cmxSG1c$jUM~M2lBgDsyY5`s$fXB$YTP7jQ2b2q%e7M-(%
zl9vsv+D?C~c6-10;0bGgaR=4;eh-!!;jo$B3|nx&_TDyL=gtbQ*?7L;%qiQu!IgHI
z#5ZsF@4k71IZy#+esc2$18y;fKX>!yqYjV$@68)_Q}c#>Li5J{3_zD|Ofe!*j5$C4
zn+Z@G8#S%};UoI)a6CAKCDR`Q^KyrVe(m&^gXw=)I5t&Je?SbTZ_F_N@qv_GKR$H#
zT^oOh82h>%?3*rjhumKz%mSH4#D20KP8AqYd=kly4Fs^;#0P&rbcgU{z9tEse~`D?
znv1E{t_dNOto`GaTuUHcY{y~#4qB}l<>kizJYfZL;AB^K;?!|BJ7fP#Y3*-K_O`wx
z9TO!L5|n18OOqL--ijy5Z2G6`%xi4@NHkLMcta%QfDr!!^)U33li@QHvt>P34FWYX
zmx1a7DzoTHtp|TZ$nkQ%J$g^*kJqs2T~88$0-VLsj)dBBy8h3@r9TFJ=m9YHFbFz0
z|HRY=7BFsV6UTI&i}hr+%Jq1m53oq(LQq8*^{t9el)3FVOxs|a4m=w@TurjD3Nyye
zXBr2N?uSV|2@8{5jFqGb&Y@@|g-D{r2+$(=)1WdU;3Ilt5c;hyr);i{N
zLRx*23DJE1quQ7xn>uKY?uQlr3MgcP$p;xDap>3Z7&;hX6i$S+bmhq0r-*#Km>PSM
zBDyuBG=?XKWaq*nF!bJO!EH$d@23S~q=Jg&@DpT6{6#Ygnyt4gHS$jKx^
zEn;b-uPT3nWH2hU=jPIOn%1T@e=Q+6M`&Q%;`}!n>Mr>>gR-8iPQXfTD{koSQjikn
zlGq4XK?khxkh~F{4r}Y5zCxxxnFQsTcuV
zkJp*JJ8#8&WoH^`y6~v|u4(1Un`GS8wXWA6Dy3DvJn>f336_h_UR~p_&ik7>bS!o=S>zvP#^hDoRNKW&60$A_Hkws>~{zEjAZ#-hc9J3H`4!&Qw`scV23f83qqLTW3FB
z{3jS&7UMP2Aks4H$kyA0E{+L!Z9!#P(CgV8R{NT?D&bY>RN;r@OiP_?6{g$LAMrFtYWh{z}eAx;fcxMwNYRjZPoNhjOPnSVlJ3235`ucO$A!I
zY$Wvgho{htmpU^gSW3Cya)RPW1*jIyM1My}=@Y&HlS>$n(0FP90~=@k21-#*EpT)&
zBp6D->J8c8*TmKs#+@ARpG-^%`=?bo*}o<+79ONBVudGm1?k$2H~3i!a+4GNdqeI%
z9@oE_$L)c~QzMKB;Vz+1!2m`R*z?4uYfz*d!~UxTw+O~1fjoLrV>v5~%Pu$Y&3{j`
zLV8Zo0$3q7)|*||W1FMnZ0jf3HU~-qnIlSqp50YnHQRbgV}zd;-Rrs{hYTH8csBF8
z$FJ}3PhD@rOq*ugY?@z#R|iRS+5P~JR8ZiBZIKyJrG#FuLR&9(+pcf-i#UtT>$css
zp;>L0p%Fp=?Z(c(p2)Z;V||#Al7HM(ecOCwvpHBj|K6_{H1v!1Ydv15f?d^Q>mde-
ztAl1k3|{}TZvr*1!5(1Lo|t_Xx3iMenxk~otbZ8dOQgI>JUe|r^Hy1^R
znYO!jGnb(g)J^awtycri3V$mn>%EUzRA3f8VV3(yLmLIqn2fvHk;1!ZHE|4kZ6QKb
zwC&B7g7Tm)@C_axRp#IAhy;K_K6olY9LXXbwCS+JKAU$oP-x
zRXD2;SWP(y?jew!Ecd6hK?WUF2|mkBzwxIOX1wzFqv@+{_}#Vr&3|qMFhF9x7ih*X
zQH0ULy3rKM5kZ(d04fqJbX=k0pD3~r3GF11q{5YshJN=hVGhI0!R39HKbp*Y;*a8F
zM=^*2t0`uM8O0z-JC^aHn8qb44N3`UG#D0)G<$}ow;{N@AI`EgOzWw+5sg;lFE_W|
zHSIbgSrcIxtiME}kAI1!$$_%4VZEb+=YLnLw%V@hq#Mv0*buUyu)RKVU-!-Jx~a(x
z1W!mAYPB5*oLbG6yulu+haLu^JPi#Bh?-nTT1yd3`YIWzXkY`{cqGaMI20?~x!}$aLZHt}2tH@uKg(qv8^N8x`_$@_)AUuc7GMAc
z@XZ`jLo0`*r%-wI{xteq)-8pNee+?&A$JXgBfizv-EHtni=IB-C3AmB4t4QG;F~A_
z+4ckshcbqS41WPB6nt31GV=mrQ1A?MD!urRznRTJw-j{=$&@r_aF_YAL`yy7*y83M
z6%mwVrgRWh6c8?y4;2wWvXJhfGJ+~0S5uj?1+&qV-z0r^q19t5J`P_2ZW(qT0tQ*c
zY~7QyOko9lXgNUT566`gPCDVr11%jDC`Ys;li7G#h<`8T(}`v$eEQ{{0BbOIJQVF}
zs+=(0)pVUwW2sH}Gg`z#)26PrBVvk$$SGnJhFw_6^NHpkwfd4at(la!~91$PI-|o9D}>J>=Lt>^cCe9AeTgzMv02%&Z<~
z@7z;1rhm}>tJz%S7+!CaAIDsJ9x?@{{kjlIRP3XuT#8Ux385uEhAdK0Dkk$9Cs{#M
ztpqGdNpy<75(6^6Y$8QW9|e|xKw`Rb>>sC<_mw{s7xo!h=9jM)dVW_6qQI;L#c2ojQ7XW@^dWk5)B``?<8qN?{s6f|j
z@PE|9XGMP7EO629@hO(Ipu28Xp+pQvsvH;$B*NJ#qGaid*k!!&8`LQxgL#>lDU3c_
zSM3T&SOjyG;|Z|_yWZDq&}GU@QUSg>G1H~5p2!kb-1XJfy6J8_wU?v|zF9Zh_o0hQ
zXTC^y=M}@^$3&QRfG!iT$CrS|BFBYSe1Cbp$I@HqNFexlGo*!(ni%`lJ+;7L2O%Ir
zPMCbuP~EThb>a;bGJ|_5PO5^wq%KE$;wy*C&^an)ejKTeufllcPfr^j?kBgL&!
z$C`p~DoQ;Rw-1%^1Kcz}{KJGB--K{NTS@aW*2(v?IN)EGTaPtDt~5;
zW?0r;v)VqruB(3E5(Xcw-40rSGc2QreVy7bU~e&aLUvMsB^nVW+Gv0-F2lqt1s9Zs
z=WBn~Oxzi0?9g(dN#`W`Iy{)&L)B*ysu+`*&*N)x4u5`~jv{68dA>oOHx~qo4k#Ob
zZ1{1H5xBC(C&d%}L-7-gN%w?g*?)mRmhr7Ge*1QHJ5Lr}1>RXmLI#j3B`Gozoy@C1
zIt1`Td;Kv)%_+}EpJ+blfAQ_T7#G;yB(;JmR!5Qh{R~H;t&&(BU$Cz))LL{wQOH@CY@qLe7|a+7C;kSg2UHk*G%KN)v0
zV{hy2waU`ybGKD88cl!f&8SCz>pG(AuEBwdTH8I)=*fWFH(1XG*8K1(DK?d;a+#yN*RfDp3EN7cD%y3s
zw*%}_DmrKo((0CGPqy`=G`ibf%OWx~(M@9-T5-3PW%SH@ZA0IG(H)Y1PrBzXGM2F*
zQI@)6QDh;MOi7k6z3pmy^db~l^kKSCXF!QbG!p%$)9v#BV!ym#EUFUp5^sQ23bs+n
zCMznDl)&MzDmMUm3jgXsn^l5XvPNT;Ah6A{ZGKD$dEa2@)WICGIk-i>#=hgLxsYmvIOp+A_cnFO^1t71N
z0A_X+L4V*5!TzNv61mC14^+Ymy!9u+={w-;+ScJIPY`TG{qj8Sy>fwCJFk+4H=G8q
z1EBoy2
zwoKDLVA0UQ9xL8|q!OWWHpy`oZmQz^0S8UNFwh6LqhB@y40>QnN`u*8(AUtmsl+uM
zk(`%rR4U>WP&@S)Mu;
zjU2xr>p?1;Ic*Ah=Rec2hMF>NO%mRPz%&3Qp0ykAzbv~*cwa^
zSL_??BX+l75qy+Sa^6Axu(fx+?jcw60gdb4kElguqoDU&a6Jar*z1X64kv^(9j|emQpz>-h;wkT
zORnpGm=PpZ2|#V7UXgSlu6G{kX-p_vc$6g{1fpeOXW+^g3UPj^7?p_v@}g(A_sSss
zPlbdmB_uSoV~2*6rLP++Up1BNR5)N)dP6M|P%%%dCC}#V8c-Gxt5vq#GxHggj(w(L
zpQ#u@APByiHfV=r^>rVN_WlOTo34hL#NUE{*n%ley_qT+Jubp3U0TF-msR)?_k!Q|
zk25gSp>+2(ca2~6M0wn1Iiz!|Dd?uPX~Y;m4EC-y>yNXe*1OhFB
zz;}vp)`Thm2Lo_jV92$KB{PYY#|XIRKr@Ru@eu4nG+rtvoCexg*z(&)8aL*g2DlS{
z^2`4?`FbG6j_UyrG&Fmnad7d=e6R9c#MKm14Z+1786R?|VaT}+;IyNlDrU&;
zr}}v0k{TuxI#TYPB_*Uyh_{w?fF98DjUa`VQ6vGdrljP%kg5FejfJxuYR`S^z*4`3avHcvo*=f10U
z#1KmR;mqhYpGrz-?TSzY`OYYz6a&9|C>j43j%$fV0AG);CrE}Y;&|YDG1R=2p$y_L
zXQ-DGLip1bxSy=3jp&vIGXFh=X5uBfBi^b24g`8Cy6Hk;m8B7u3hYzhVwq
zrO?&|pwTs|0D(ZNrKZwLO;yymVR?mnKmr)qmc0hnKUz)aurs_?vp_qPX8DC6-;Ri}
zaYT1~2^7%Z#z`KUqyMa&5A9AN!-7YU1VPQqPg+`SxXMiZ=`-mtuj4m=fh@&dTwYv|
z)I~vdE`sw*FX8d%u$9c@?Ll_q_YUab!r!
zIAWmj1+=4gju|`s%?=ZfwGu`aF&OrxXDRjl)b(eIiyz@gl|&CzSor8GLhp%BfOw^z
z#vifoSbsU0g9I8$rlJUci1*Je6nT8zsqpWp9H+{^H(bh*&hnXh=huVdi-qtRi6uGDZstopE={0}WV^28Vx$b&^N+~?7H{Och
zkDXs#&tc;2G>;p9ywhr^_$RA{oGKAA=6dRLH;hJmF(j9k5)#Xu9r0nIeI^1W0T&cS
z-XOz$WJLb~IAFZ+{|@@)iHpV*cRKn#SH5yqo6?G_@Y4HQTqH>)WN0ybbd^bTZ#vUc
z4T@7}qINJRbSwD<8-A&Ca@-k(kN!U8YxgO+Ay>p1EqN(GB8_N0nlW0okT9y{`L0V{{h9F+CmCtZe(+Ga%Ev{3T19&Z(?c+F)=V8
zFd%PYY6?6&3NK7$ZfA68ATl{LF_W=vD1WV5S&!Su5q|ft&`Yo!WQNP`=D9!~lB}IL
z>tuuVWPu|g7)1&8qK(=pW5O)WH))cE+9aFX{pMf6FujXRTGh6_(s?j8pMS1y
zjL|_;_79e)hpQXxbp2R&Ekn9{9><*=d4H88LD|=3TM6V{lteas2(NirB)brr=vESJ
z>%DCU;9NHKNVI)o_V0+$xb~%j0NS`7d6%d8UVCE=&NhiSfm#GHQv`U$i7!OV$gYDnbhb#l3+B|WJ
ztP{bKuJtAGbVLI@*zuczH-DDq@*SiBmHFpYu2_?B6sFK@FX5dOSs15`wlFF$=ykgA
z?JF3=!utK)%!Ea>zT5KomA6Qss<{Pqm8%TJ_nn40w`%67}F=_kjdTvQW~@&ST@8Goc&2Y>EJmLHrz
z-;);E9$FYjPL93hoAn?Ud2o6*RdjlHQA~LfCQ;0|h)#~9Q_B=S<3O$kvQJb74>OFm
zBy7`+wFoDM1K+9Y`%kPx0J3V~RO>J})%V5qLtnR5eJC3SldRZ{w`BG1`Z(}W?`yq_
zP1n9Ef$k)6apJ3T=YOk=*aj3Bw!wx04ql(}j#b$*^l44isR?s1bM@8#yQ;2UTG9-c
z5vr_Mf8LXL=Qt>?zw~j!BwZ=IkOR&oJ5E_E!6OPIx#qMI5U9-cS!bH0DohGdoVwlC
z6~dDDoU;e)DOS-mmH=wVtE9^y94`ex{&?8QBXaT3lP(MgwGl|hJc`jsWlw4SuknL3q(Mv003*aFd5N5G`JYwVsE6gYV1sO5cI
zxmf5JL6CZXXdD)#$P%{+PRyXu*}6M8__s((4kDxY_`zFnRGi`(AxfuecD2McpttK9
zH&sgqj}206H-BOND$$
z2Ao@X$pyuQ3gbw4{pu>$l*CPKIa*{USdp3L$x)z=l}?~Wd?hB7$i2pj(l>_8m<+t$
zm2{IWpCMugPWiFiieqg$@o8Ifd`Neep649?86tEd{C{Wvf!BruWek|c!X>~Kf}j&l
z5SY|VQ^%(%X@CyhT-_*O2#$^wZSSnm!KR6W_tph|2jcOyup?fH0UK~CbB=e4eF_&;
z2>0pRckY8lKvszG;+zJJJdDUdTzagKnYnlapdY8f-cKZ%C>9+7(5L8h`A9!bfPR-;
zD=jq9SAUwJ<6XxNf*cMm*^g%H{md1`h{Q=O0_fVNZpHhMQq-XhC12yDIX-$P@`_L~
z;}g1461diW;0c91@>2pY`@Wu?5ElwhIHH0b(N7?*+tyy2Fo7
z7ce2G)o%T>t&9hFXgc<>-uog+OS0jfv&YN*FkUifc$?59#;0^QCW%*x3YLu)0H*@Wgjw
zgpepw>pOeCVAS#g@QE{@ue^&QeeTr=hYEj4a$;Bkz`jd%F
zoTbnLy}ovRQR9(Enj-hfoX_(Sb-6ITRIlZT+@xyD>(B63<%R1htnNI+TUD4)XW9DB
z1@$yZ`n-BB4z*`RgpYi=c501E(Bz+yMB15TPim)*bx3;s6%5o+Xp^U3#Xz0Jp?^8W
z@JlFvMmMRT&GPIF<!$r8<
zb-h|YbHy-uZ2P|Mrt_^J68S
zH1UyL^8UvwF6vdlPg2U?OQ@FeVc@ak0W#cK5u}{A%owWp>t818D+{hxl`ElWtE#2z
zOOG4|*`!-ON4t>W8U#13yl-0z$;LMlm84AS*u55inkf2SlZQIkfZQHhO+qS;8ZQGh@PTRKaX-!+x#``xf?&97}
zMpZ;c)T%18GS4}g7U`2{r=HW3kLVVfMh@9^~
zI2HE5jgV%;lq57l;E~J-l<_pSYRq{5ZFf%B3Qnon$q2ira!u2>Sv+f(8M9QGb|pTW
zHshreCGlSQN1Yn*9px(atUcC6!9)Ssg$~0}LxLu!6-ptF?n)L&d9ZFlf?6f{$CSsw
z{h<6CZ0ugS84mRF7sW{hTy%+hMa6EbTO&ijnft<>vT6$0+Wfax%L7v2W
z5twEuI$pHq)>tG63~dponGgCgx)*Iem)g-8zxB5!d?*euGupxJ#iyO{skdiSSIi{d
zw!6Cee;1Gy`k9W<(pit_nickHkSIg$l+U&I(wlI+ql2BQh
zrTbY1`4MEHyy6}zpDCD;(GXw|+Q|U^WTCg`HIp%%Cb9esB<}_OvJ%QD<6RruOPw8Y
zCa^F`k$rN0TS|IYIeA$^&RFWV@W=0__~}Vs*@tqAB$yCQI(+GzIV0_+%Labm{zenc
zG<)|^jd+~P#l*x8rpV_Hw*f~As`eH!aUYGG_Aql<*0f*#-XkL)&{eL8`atGp8}B3=
zY0nvDzd*+fRI`tF3!yprUZOv5?+>g_@-fQapQ!Q+D@2hme_u8-Z%Z>FtTDi@^!2=c
zQ4Mgg=1~xnwxp3tgHQS>U%I@@xW!HqCpB8K%tvecD&`bU9OHmC{zt~i;PK;0Rfbv^
zu$m*-5gpjmqk`6O24s#;^RS8(KF|CTC%?r@;1tz>zf_|(#{DdK9_EA34bor278q#ZT#TM^A*hv0$oMFJ>s>MN}n1PbrIQU#g5(A&A&P
zJVy{JN^y_!Ii+-+`~aW#L*QbPwfO(Y;8~edO%l++I5;zd{=@tLT5@h%oXEYWTDLGW
z+Au^QL_Oty$`q&+k)9qGA3;=FqH;)eP~$E>-mXnE)5N#(ZTOaA!?-;Ac5a4P5fDYN
zKW-_HzHe@JyFZaFL`Y)U6+^q;iPI#K2lTPUMbd^@f20zxWXGWqY!ieUjbtiEmfKT&
zNNMFdq>vA`_}R7town60?zs+mvDX7_(6}8pP=-BqQEYeZJ6mM-F~inftL*Gt>YvK-
z%m`B4e~cqWyWWop?Wqr-)?uL~2SrPdr_!*CwK`^;yo7@Bn%Sb&Rg!2K{^aTV)YTU2
zD&p6|FHNPg#B-U{IS4hugYWf)mSbBQxlP6c{y^pEKw|wi6Yt3Gi
zB#}TY`vF{jg)?8~ht~WjO6#F#9k?#|Rj1(L{cplLUTw}GPGl_UZe^}c2Kv*h1`GSz
zDC~mPKlWSzB`%@Ww&964^95dCog8T1IrA#k`~LlipLq3M3M=}U93+ht>JkE?mcJ%j5F?;eLc*L@XtK~_O
zTUJ?RaYUoG)itw{7Vr-=yi37MLsgz{MdyAfdA~}yyE87yL+etoCYbVWdd$fT)a5U1
zZ!KhJTt{BdqW!$r5_(^Wqa2kYUM6BtC3Bq-b!HEn)09wR(w
zkdTPD<7G~|gB3nfAYV9#QkwEtxuO@Gf;!_r^*dpRq#>@IgR*wzRPHMMCWM_8nH%o@
zbuwCf-VOS03o+UEAKKZR7Sy<>^_yJ^eGuvy_$IPBY+EzlG4@J*@GYcsxe$qhjB=y5
zW#heFQ1WlMjpkjZ-sCe+xkVzK?iM=S%-_?zdX{g{I22JH3ZIZ54W0>}gsXOU_s!gS
zdV28t2>~MRj;VNdc~!enyeMB2a@U&#lF-g8pT!p}Kx2mr&vnf!ns06@Dl^uv8~J&Xs@8LbrsmeGhQEyG{o;=-+3{X1
zH^p6;6!2)LTsW*Hpt^m@1KKF`BbMLmlZ(Cki}
zQdU}LQ#nG*R(Rbsr!H>_ad>J_E5@vxo_JBm9J}+!L&&Hwud505m
z@Y$U=x+)wj@I(wS)9tM;&GH0V;^4%nDX=uWBLXTc8fIe?(9L0(OGUq)Yuwq#nIM8}
z^X_z$%`o`;rgxK8b0_gm>|I)Qh!zU-dyghc9dLD@g;mbxQ0-J|?pC78VAI|%Cj^dt
zWhX~9?zsI`B%~I(uCt55uMOqv=4oLxLsAt8!~`k%q|kH-bw@{
zKZ1NXvtAeIx$^Kx^?Ltwl(PM&jV6A>b);sf6=jB6(mAKT4hPDXhq;
zaGDlRz5)tyR8iMb@Ff)&!R;AG2oQPL=DMA|@)v2TS!rp-ba3K(RbI%HCe0;ysB6
    E>lQVdGsU@^LB4e>U+jEaatDze8C
zqPb$PRQml@S49NnTS4DpUkDsjK_AT6X}v8{@8HTfO;r0en<+7fqZ@rNi*exzg3{W9#-|BDdmgb}7&gcJ;sM7A4MzV}
z1BZ!dK=6Ql__3#ti1Gb?xj%x=1Wlf-ndJsA{f
z3dYgofNA0u+{uE(y;8@cySHv`e^UW+54}qyzlhDwU()rQMlY;i4}Hj+J?m@Z)8#sA}{*n
z-KZTh#q*K6)|I8On$p#LH{trvvKLkI&GcjH<>3WD3lu(wH+-?JpRIoe2eAvlXZ42=
zCPdv{j$rSH7WRC)`-e+E)@R7Ho0v!>kODb6rgd>~Y|4-K{dI5k^*zLN@o(m;aMYz2
z`^)e1>n$Q2JEPW%>>HP;FTk%S>-+ubQ6hKD?ase>XAWz@2*!u8X+8R|Ez
zJ*u{=NR_42x7T{%$J-2Maye~W`5v`Pd~Na
z&$86tZ#BCPLb=W0iR``n8+8ryXM(XW@BRX89KAV!?w-UBLD-f%)*+`q?5G!9Ya(;#=XUgy+vk%TbX>?H>g@YfUa^XO?Z+y0{UDVjkL5fWN;?Pg?dg>m!qKMz
zxDW4)@U91*wX7}yf)2L|5d)9FQBs$+g7C8EmmW@_>80833T_kVu*y%qloTR-h0q?m
z%OWKp?P!IkE?XwT{RnBc%O1x^p#N#I|?GuM8W_Y%%ic~1U7eL{ir4cM(s90TqZXf>3qj?|tr-TR-SU
zbVYXm?_hQU-lR8vi2N?=iJV}fn1DpqDh^{kRDy{hG$#jvH))2Ced@gD66dcH>TW;V
zjH^(>TpYof2eQ%q{dBk;k~`pGo3x5+6(cAYN4s_mD4WIJop5f7g8bZIQPDXEt;J$2|sOn)D8+
z2DoHGHJcKR9^)^r9XFX7u#oWK5(mC@9E#Glm=#kyZWGX!HXuz+TSVE$)g!`7ZNA4H
z2wFj`rN0zR@n>pkXfejArcRh>&G)o${(*sLjR@0E8^+ej6RI5E;av}7_bcYX-3E;_
z?LcD4VQ7v^OS@yX8**81Z1(EnW&CyrQ)D6EC-;`7%AB>Lz`
z*{1HsANMgBDtxQ~^HFBff->|n-9y&cdkJ3ApTUjIhZ4~`G2^q|yIzM#m=Gq|kPZGP
zN+>l>CE7wi4ND{tz6e7|^=NJF(Yjy$<;M*ODxTbRkz&*Vr>z2eKcYx}k!zCB6y?ck
z4zql4L|WM1Xj7Oz+o&a}-i7QL-S5rD>mha(P_4>GV@8`5nFLT6!<7alq*Jv<3Ddw}
zirYm6F-Kb2Q)Wtp<_J)+j-w)g3QACpMaSMqFu60=VB`vuO*sD1&Nv&!J(v^CD!7Tj
zt>5yjOhkGrR9Wann7fi%W#5Jz4bTuz-#=Z#5^nFwX@??F=W
z>y{CoJWL5EwIkiHdS3h>{ksjUdF%Ld)Wj2RE4mx&BwR-iJuT2o04INa`P#dzt$SM)
z3oZpUsp*yu%H~>=oa1-|5zWCF1Up``==n?=@cscTPsr3g%Ebgo>%F9LZV=O$uKCWG
zbU&IsrHpb9NG(G(WR%jwPSat)UJPl;raT_zpQ1x0O7b2?!bzLx2czdq>ToXy9;!Lh
z{FtQmFtB+I?VS)#+FqwEeS((_vn;7e5t7Ycdz1;0l~YSM7!8%wOHb2I)9qq^3?C%{
zev=(#(K9$`qR>sM)rQ2t7F@?u@-3sD|9%8KzCID2`MkdJK|D&<`m9c^u#_$tkJd?K
z`>6ktI2@>LW;M4jH+V!}Jn*8n=82WDdiZ-s!&d8DL-K%;8`OYQlxoukkTrM2w5nQd
z{wQSqX|N!tJkUMaGcOZcT8uxGTTDEyp~tZt$Wk#cp9V47fKIpgX9x^uZM**bm}P7%
z^PbP)cWWvAbAH=>dhwtCV(9ud4Om)l3$vk`hyL^zG~B0b;r~&bVduz1o&-m4jw-65
zq~!WvtyUIdR#r}y|DTDo692d1&Rl2*$ADzz;$UV=ZS
zBhr*bvNh+2XJM!nps*b%!@FZnmIw47g(5o^{1r3@qzuwXSlSN
zK*$f)2l2FW3+irskSZbziVhA&%RnbaY=PtrM92Y&;|FOHwi3fY@*^e-8oXsv{^jOc
z!NKbOh7Gluqq7bM&0n;d%q3i*i=fI$lh4G0Z(3~`*l
z-*uq%8BZv-=hjvaT3{gI9nH-GSOPt{xp?{6v8inN`K37d
zJ^zf})TbHpKpZ_2G=mboq>4Che-5^<=lG{BZQ#HBg#EII6&@YLw!VP)KqI$&{wb21
zs2T_nu%7=7kX@D{&jKecUzvRUurMD#*-s)AfssjS>TwWz1P1&3;I@diQKE#|h*?HO
zkkCGRBBmnP*FFd^&_Tw81i*FtYcJ^!u=r8vPSTZ_{#V=Y6}$8ECIM%&liTO3l=HjC
z-i%L=3ZBZU5}1L3$kNpY|E}_SYqr8{12)VdkI8`%@pKJq3qjnu49(T+Qr_w_3HYES&Ew
zAz5JmkA;&J5;I4OC)k^I7MRekaZetCK`@Aw6jo8wpARdjxRD|A1EHJlui)#=9q0{ZuNUgP%!#fhl{
zgOxRS*~?AH*U!x2`0&;-;Qcc(miIPMJqGxe`W2p<70x#=J}|hk{$Zr?nQ5@Naa;Nq
z(qX19n27H_Q|8|sPt%w6dZOIe;aomsk%hU*dtqauA$waBBrgx>`WFV&N7cah*(db}
zQ7eeA4U3Emjl%i2O=e%7xI_GvJiYx4hf}JJA5D9nKZvZ
z2QClX6TOfuofEwjFTMS@T0U_y|3H3A%*GK4+x!Io&wMi+53TLq{%L4|%HjQXu=x1>
zZ7=mbgSNmvTL3rGvTGb={??oSNo{Q?)#Jbpy_$m)M3dTFEuGJ*ZE8vSp{|I{U~
zzx$CC2>zB6!xM1E=I`sj5*L}@KeX!Z`U+%$
zoiHM)r|*|JGEe{nq7yvcB=h7dt|OE_6<0JSK*Ehmj#!%Kui_;03wxRAK&C3ymUC~E
zxdRI5>%&Lx$Kv??r8#-~f2z?R4)E;Kq4V{7tGR!ss2t_dA=`cgSOmBrOF&Aot(>WY
zFPdY0%JN20BbBqdvBT?BEa{r+-GbJXN)6^Y
zX;i8B6|Uv+{Igq3Ne_Z9bJ*(jV4W&VSn6n-NRWnk7F|D$@#$GIItU30LlHPZ6Jg76ZN
zM@ojfFc2J!vwA&MA*NUT;{uY(S8a8ny4K
zOINWWMXqtgWrDoUcd6f1gj)D(Lp8?jxK7QX6!-ue3rRvLoI$ms1c#cY;m;kO`$
z$ej9{T3Cf?sWPYd#{~7Gz3ko2#k)`!JSba-h~SJD3du`b8o6JS!A4jj!6|xZvy}Or
z*r0%msm#?K@j7iz${$_!(1+QuF3(8OfQ6+Vczn3IDQy?8$%)Az@Lzh-4L8xplG9WCOMzqWYOnX~L}u$+|5iFl6)G=L@N{!f6x
zm7s+v1Z_#Rw=?HKD&RAJGY>i)lnAY^R60dQjC|fXq;w4AJDm*1)k>IM(DKx=ti0eZ
zOa*pM`<;z<>^Kp}KV|6eFT92_XY2!aCHJXfJ11t+*3~re=Q4vfojv`fQ_dXiC`l7g
zP-?EyK~4+}c^J5m?Q}9-tDm`OA$!22vE+Ic`kpmH5{HGTcDpZG$#v!so2HrmvLp(Q
zMszU>Lv;>WZfMcU&`An=+RLsfIq}5V02_UngHFJ7Ee^!nCn5&MyeF+H4OLuM2V@`$
z*c{^_LC%w_*O1s-d5@2483A^T#q6Ib&V4|_3GR&6g(+mYp>)t|vF@5wlM`5@@P<_R
zy(L9=Y)K%AbeIwO8~;@yvbs7&GgjJ2L*E4c=OcIh{>OFiXFiiX^(GGbg)PqmNR2Pd
z`f_vLQAnbF5k56q=6S~lSbV?=?BHAy7=HAVZ7lS1NR23nMVt^odB!i9K_b)$jE=mzNf
zlw|S!xWa@{)78ovR))87YWMYDXQoF+nS31}C*X9p-Fv~9Qkpwfz9Q0`>}2c7r!s<8KGm=*gM6d4VPpW8Tq{uL(
z!1F4m{pC&(JVeWwoKUm?*gMM7v@J~L&4y1Ff!Qka8ZG6D2vDHlsc(
zZb5f#+lD($?5rG=euuiZ<)4+BA#=VY;sYssE+mx{lu43hLF%k&9tS#8LmVspYNR;q
zZ?cl|9UWCZ$&Aqp{co1X4Eh1LzON_8&TBodT3&kw(U&_
zv?3Kzj_44Fyu8yGOiwm5#J{Z9O3(mk86j9-guG!<`@ZE@JQm-S
zx%D|#dTlwfRN4+Luo>n#6mc~-y*A7vaK$ErR=wkP`Ud6&mjYYgW4#2F>y^ss+tyi(
zv(3#GH*&gNn%>mgyNtTr6{tM_9rLz)XJ1-kRiD{9nn$BRQP5+=aK-T%Pi
zh#Rlt*AFN(*Bfa+0pc2x;lki7^z;3>UUSe))~Ye%WeN9xP#!CnR3jqswkmbnuWt_D
zKQ~f@=NB5I9FaxEsG6x(!%6*4Q5wZ$%?RB2i)w6Y#z;FzqLGPRd}$6@U=2FEhET8N
z#CYxH3afEX0)(@xONtLgdRto)A|)sfdy&Z=dSnbT*8gstPoj1q2oQRt&97jgP2mV>
zYORzg_lWDfq@y++s`IB9?TzY`wpS1M^apl+$4JpMNs~t)|NeN|vg+~qw~3nX-8sQH
z#A3uOQ7~z>sFunjSFTA0A?wa)Uyecyu0_BCF}oE+gVaGON7aloo2O{pqRv;Vui*NJ
zYnAYi@2c*QeSf?B-%rJ`OAp)?CF!B%kVej33H5aWIk_q3Qk5eV*3tP{pK=XO4u0U?
z)bgtBQg|X}>gY*pF{-#-vf^&$$*|b;{u$a+m&bo6lP$sI9oJTG#LSbUB6v<@e@Uoc
zKoE&i{m03mg}+YAh!T&D)@|2_MDlV3icW);=Oe^8NORwS;6i`PnOj;)YJi<$oO>>{
z3Ndj5KE<}{P&95@2tJB$RrwJ6jugNEYJq+5M=g>V5DZr@qNA1d{zE=nd3p*lI5v)F*PB0T%f!!r)E(ZL9gLN1y&aaF=2I7vWv9_4au
z)REz*j!GD8JPhM#<3KV8i#XEjj#!1_uL>R_+v>0Y9WTgZ=x`h;^oaRt@
zEeJImKUFvUJ*2hYV#S1|Vjko;e_b+5$m!zv5-6*)aqwH>qwDMCgTyhsTv`otX4CP)
zl4#~$^GSW%#ZwE^MF(`lHR0#y@?&d}NFmyCT=hW#(?zb3P(w!o#Vx1$%{q)N^c;p%YTI^+7j@<7skW0MsI6CTM&WzHbDyFP8^W+eUp{B(RIgb$)7%%cI^BN
z5Hw%o%cUl4S<==Nr@C&{+Zj5S;cB2x_Oz!k$~uvO1-lD@rfIl`l00sjs!v@Vk3cq`
zHQ>DWvTwqh88rSP)ch0veF`Vux{Y)68s@}9zjudy#8h08f&dukuiuwIPkCvih=9AN
zdiUy(Gz5X!H^+=U2m1vcE@sQ9xG-8J-|aP58l&`y8IDFb3%FOzR*;_l;5X-UMnh<|
z?bB8>eQYMV*{HP1fNh&f1`EHv!kZ%W-YQT@HCf(lqrLGEEhA;8r^P?Juy6`?$2&r~
za*3jtWWD7y{Jbe#Rob`
zk-WIz5sY=q5G5VMn&N_rCyDKWK4>DjX>nGSgtw{aGO9Jp5aQ3GFASN#bkGjoda-x%
zHuP$|(H
z#n-Vf2mqh^TCF(tMSM{d8?(E?jjOnILB5s$+3uL9KoNdXmHYYX1WDI$h&N4>vvo+H
z;qFm5H?q#}dPd?)V;`U0=3$C{aK=bX=wn&$Wxi92&>Y!6R6kG^f_Zt(q0ZJqFQtZ`
zw-s=o1pM}M`HymgJN0B0*wqlL$g~@65<9QF3z*)($JF>-wzefm{`(?OkT_h3fV?;V
zJJwAbGAYg0)^3xQb~A|s%9lcsx;QZdB1EwfM}}g2o{~OpFjb1D&M+=g=3U2=;KL(8
z2VGPDEtvvAmpl&{#_dN~dH%{5b!!yz*ArD%PLPpSOaqyz{
z04v|adLQiKzgfVU=9x_H($nMarFOLF?gd?k50s-f~cC}&*d2X3yKspV-UwZ}43wfw@R3G1Rl
z!=&RMMkzM~>=kSj0w2FLH$L7XSdSy0JmpBfzgFw?jdAgmlyiqO<#it8iY0IhfCvW#
z!=nJ}_6|+mAQGJrrhJ9VcgG99GFwL8WC$7B4V=75*QT$l6OZ4b;St^|V|4$*f7ZJW
zj%jO3{@Z%4e5@4x5Va!6M&ZdM2t8e8V(b#iggB>PDa$63&lWbfBb9r1H_!k)Z-UHQ`VC>{8KfnM_#=k*O=0T2d8OCPO*
z;UTRHTb2ZHto@d1q7=r>Z7*)dqI<hkzT7bXdOsO<*pMO!1+Yz@5|AE`_DxKh2
zXyXjDYO1VXqJ$cN0Mbuv7#$9GRE0XNjMooJu-Pb<>QD4}GVrj+)o}Ni)>DO3dT|j*
zW=5&BJH>5KP@u6UVZWFed)+BKp{@rr+WO>3kQU#@i#RPS#=y3LhpeO7Zn?}Tf1GB+
zC9D&aX2i>N;?AzQ&+|f7-bPb`G&#_f6-v(J83n|x$9hU5fT_D{VN*G@0Mdw{*h9}3
z-T|mDq_+C4u(DtCl1wk8tp)Lz-EY19W_5QEk~FJe%#R^OMtk_2;62?d>(9gGH2Q;h
zkDmQ+QObbv6c>ZFi&_X9dF}ks8hzIUTS-ZkI+#B>pm8pPr14mpeG32k$ZwWM;L@^b
zN6r)`fno|5fbFoS|Ei=c++oeqin7lAq8GlxfZX=WXIT$0m-Z}5e7Vn$Z
zq+dM@*nhAWm{~sJbR2Nq_~gwTBW3kEXWLi>DkOeJqvcX5|ue$j#kw!_6MAwA6$FbxZJVhkZU9Yv|
zx6{fz@?-zDWvQAKiEXu$Y~g{re1ouPWV>M11E$BhRBKfdqu;xBpp4Gt$5f+~cJGZS
z^z?R$>PK8X>?QaQoZjBi@J*ba)1v-mrfB+5GWy}&pBNKl0ezy;tDp)pG=~^JD9P#6
z1F9aBOqM~APK~QHibHIq3DCY^5qoyKE&>_>&6O)qRB_QJnT|y1Mok3EMJh=m4WR#l~2{PEefzaSn_oiA}O-kuWKi<~u%ahts4KChr
zd$pCUjHUl#P#c6Uy|*FC=4a1%u|LcjP$BZ$RTq{c`ceSdsc}Ai7KylAS
z&Sf)g&Sgv`JoSHKkZ698J(aADGdDA`d%7NkJIUah%O5D+aP^%1Xf;G(3xoj;>lD7Qv1cm2#Y$3+O}_Wyw{W1Bvs@1I!05H*meE*
zY4iQ(bOoj?Hdjv{{hEbuhBq)7^{5PZT@>YjYzFR%kWBcdw1r=kHWl8MZEPn
zOL?}U#uYdnfJd3HeU@{XPY!c4%kxa@q$YP<2{peTf4wSx|2yWnaiu9Z6A$)0`syr}
z{hiNvK$~PbDq;_Wb^JhUSAo
zC&_{56nM}PgrN3I6`8(HHxjeGdl%~p=7`mz^r45}70h-Dy`DC94j>7>^evA?D}x|hnl+s%(bCEzA{RTV1c
z(T!MTeIpSXPOJdG6j~Dr#z3E--Y)a6ta*ZxzKexKteq?Tl78nUp@m}6O)0&C>PLUFjd+2*u
zTl+znW%_8{gWaZ>5)iEBZiUnwc<##Qm01DUi9z{Sc9Npn}63NrESYoiNTp`SK-<<^KT(NU#AoU
zZUm0{=5RNo!2H8jTvDefZJm7F+TPR8|VCuKwq+bUptUu!46OGQ*k)
z6Po35)pks=30=M@QT_Y$fN9rTeb*?*JBr?5XnI+v={XwT6j0c7^UJ!zL|KHG3T=a)7!_QqgZ)(ni
zUP4J%U~rPjdIu~K-1F?+!9{{j1kvRBLJ5Vou*;kT0j5&veSyV;EH2^L7U49clLyK(
z;QMogq1&V5OCDwxYQ~X+@EOF;xbqVY+mDxieO|WXI@OSyRKj)TzcyK6aIh3Pcv$YiHDw9j
zVibnEP;;I^7E$8G_WS4t`MtzPU^MOIz?@`#4kcB#5JI4Mf*7b1WQP3TOZ^fq7P2*b
zt88*MaUnsQ)jY}eU8b{g`y>{U$!>=)Q^qokONOL+cm&D>250eNb_L{i`sJ9KvMfQu
zdxKa6XW5k2gHQ3`W~{&F<$5pe!I_4AJsv{b$^v~DIfYi*zzAoJ
z^J=bk;fJ1i0uuw(z2Tqy{Y;tk-hKkJl%`;lcsIRwiRF)N`B&KZ!AJicRN?-1(FWGs
zvK>XRHBaieYdBF14jnwXw#aOsr9Mh!NyfST^&;-1CUy9634E)&kzx9U^y)*(izy7K
zzx)mnlYwMlmVJ5}1~iAdQNNF8fa^z9qb-e-vA3D{;bJ&(M`aUjb7%r7)<0Mk=2AT@eEQMURsr6RQvRk
zRMMR#-OitX99~wuZTQC-2R)YqeD=Y`0*;xLS{t>n9M$Bggc4f#D>t#%KmvaQOW0G#
zB%B5!2=FwVa3Ppbzh&NY-#)V;>^dV`-)UmIN+I=FqhEiKei7=F=X
zM00_HZ%LOM<DGAph#s$Z9M;xxMb^5864P
z&MRKW&*c6qSu#hvRW@zcT$|LFz+D+aX2kU#6JYV}oEk(>zh1)e$q<%fOIzzjyWTw|
z>B;QSydlBMJDt`0Te+7R-U?&kD7%@+<68ttkB|j!LyIx6Q2+7{;M*QFF`d$G)gAXn
zGN0PmDPRWk5%C%@4<8%^nLoBajGql*Pj)0b=(4j18C<+hfn$`JioN$C)R6aV3j0Py
z%;x5Hk1?u&L4x6q=NWi1yU^97?}&+RYUxn8et>tzOw0MMnq2mt_z&{!i?4Vb#pp>x
zNW!YujVDj$XlMJMK%e-aJHBE|^2%WKX~$#hy6n7hcKDkcvq>lj-4=G&xk%!hp3hI0y|Flrk@MGJ#COS4RxK8ezcOV^Va1OS{wM=~R_I7`mDw5Hb
zh;s8TU=2K+?S(Ou2PfWSHqcu@bnbPKV6@5(_H<unS~fv?}F(dqQ{or
z?on_0D_cw3!bo^$YTSAivJH^oYxrpEMmB99
zyvcK4#=qP1(sp`fJNiJ7yTB>1;N&UJc1tYe3Xkj`4EJ`8iORR*zt?WhoF!96Pu`|4Iew(KK
zTr6Q}BU7=8umD05vdUkw%+3yDBH6Ytvr>uYh9pIEAkDK8ZD9d|35cW{g`H8jojlJT
zJ*v3H3#NDU;U^kp=ut8AAv&Jns9;#I)-5W{Z0XcSRRUb0{WA_2~D_
z|H|decQ##Zp7F!qMPc=A$u#+x7^)$F?Xf29)4y(}Q2$9b!lW1HVa{AMzLc{H(X6LA
zN9bfujLzIZ-kSIf(Fc&!hMFS$w8`ROObpg%@Ce3)QVXZgtU$l50wjx=-!=|(nDx}2
z>}LFZCnTf-Vn<-KyVZZM!sA-Oic8i!lS-p8bnuSE`UI5W_NqF+9{-uNELTtg&}|rj(8ii{Qimefi|0Os`e7nez%$%1!ZgJ6$3b&f%9s7c*7I)%dw`^UtU=1O_i6I`%`wZ}SD7oC;14@s
zJ|GyA)z!bV;5m}D8<;~xjuj?e)`2PneyPgQdE|+aEtUIe7s12amT*fXwtXmLis~
za$Q~NXMSwK$>PLNbeT2bAzEvbkXQb}vEl>Mw$#hJ6@$_s91T9>`JPsADhg`zYKq$w
zZ`Qfqw6A_7yI+eQa~k
zGZHwMU?h4b$u^L!uBL+J0Gtw1i9;&oklDE;m}(d}qPEC3?<2a`)z-qJw`CX3?rFjU
zv=AS-zg)RHR^$Zg2id*01Db+~Gu>o~vw_YfSR;dUxq6*_%&a82&5A9dQ)~8%%qp#+
zz+fPL*qH}y$1H_(7I&_Nt`eMjhId^nI@*xi(sXSa#&M|u-HNmea81P9$53_kh3M4~
zS1s-xm#|T
zyEO-ch=w(0kCH}FUH|XxaU=iV562QCrO^1>WqNvnNzDG)uU2RK#wdPzr!vv4NIoM`
z%OYrZvd2SIQ)V7xfDl}OMwsu%Yy{mA4t}W9DEX7@YVMkLPK%ZQUC(Fm(!4|RmaCk;
z9y^LoyTBs&+rIu=zbm2b+rl`Df_fHP!LW8OqclnT{{uomy}z~I)Un`gT$!pxw&K2I
zKe0*`X`yYrS`6~O2}d$R8j}A<6rZe@I2&XUc?T^&&w;np`bARMQ0UjkDXcD|lk`Tsl{zr>bvOwkk+`d?
z?5ZjLwfjBPbnyysiW(U5xv#}Fchq{?LDdG1rD3`-e2{Ip}SNn!u;K(wybd7X8hQU_kw1LhAD(p=kSaV)~44qIAckOZ9TSe^oE_uPl};f!{Yre5JLCEgzAk{XmkMbsnr9iV*Tc)V$)A7d3wm
zSx~J=p6w=*2T{^+ho8@xRWt%=SLl0>a2Hu)MBwA9IR3oEw}RkFA>mFj5XEK9g|+uD
z94jnYcMr}7t>Q_%QGi7aR97flV&6iy_jytnf4IcD??yc3YdE&xj#NuP-M
zR(t329*qXkuZe%!-KeD4`Jirx`?J4mmyqk&=CWl9xv1qjOpWqkt&Nu(rr+m?whf6h
z2u{V)#=n+H^?v58*?-sY_2By@mNM;bCqf~(Uz8WLdsf=zY9V2;wAm7@onwD8xuRez
zTOx-4iT%OFOs5pd-t;DKTCpg)bwDR9SB+SUqrAmi^^u%P2B}z_;&(C1+;Vk%C`!ok
z!^mvj3K(&?mx*(DBbv90uTuL0OAA|q!LJ?^Dw|AxJ7C0m=B%eg_?7vYqjY&|V
zYuEulF9+!)+R^)^ZXAcv4NiZ4TgxPnE`L=%LSe(2df#QN*0GD^{o1b}R90nX%c4(R
z)yx^jIUCR_rOS(;Ast8H=V-FzFg0?>cSVdP7xJ4wpp(Qbp!u`UiUfC0Wm^ZC7?&ZM
zYO)rJwYkf?8lKoFs*f{D6q^jsh9xv)!W6sfN?h9`moJ)()8?iJRjGe&1+~92ujGPy
zHf|Sx{<6Lkm{Qha=}5vP6ZW#8<0!U~4IWtaGZz_}XP`2@J1p3OnBf7l&mj^*{51`0
zUl#GpUCeh;3P_Epb*XnPg|Ty3vWizofS`myAgxJ5lRD`hv!d^ATsHXi=2t&*ubqRk
z4NOG;xKs_ybo(s(PPKnSeIrB~SjjHSmHRz{uK~;^2GQ%i&K&vLlecjjBqLGREL;JX
zS0wu+ICw?}qgYThg=P>3nBdG#CzhdthQj7F@`sY(ySFb6lCW#~7QrZ8BOo}3vsSf`
z%EF+|(@NV<07vRcuu|H#N%RvZzmg+F7J8N6Jp2+NxEcv~B;J2obb#={4eB|vpx>xU
zZX^WD3+HE8euA6|XTt2|W7`H+*v%0dEm%~UT&b%sS1_>@Q@v_BcKSVw$s1a2`Q*cB
zgKfz=>qwYAU*bKJC|^>R#(|K5SE>@fwAvW$`(4chJ(V-YjAGt<*qJAgB1%wg6V)39ovj5ebn&*LU6bjz2&vGQpf~3!Ljo
zb)YokOTs2w3veC1(*g$0$y3FI!$TLy*Chb@W@5_@79)S)Ce}QM+4kvHge<+Hjw5_<
z8qW9l_%6xkiCKBf6SL#}&?;7CA}zzd1Nc0lX!tqIpiSM01LchXWS$6{AJT0fT?;H<
zPmjbR@5#I`YfyqY^hbi;C!v!hUFzenhJp%n^2iF(!ClNsq6>?*L)J%l?dBBeG+!zg
z5G0WRc?^Fl9yvelye&jANuJ37GzT^-ESeFK6>|;C{sluAgW}3A_l;jzraNt)ddK<@
z_^xZVppiZYiHt=lOb{f?CD9p0ehHc#82#cg5g3B!)f(V6ox7&j4#G2zJ#9nW@zMyB
z#3yfcNVZR|*x1Ms(%r==kFAEE9o&yiV^8l$*nTv=X|(Jb`+NJhx?Z>^x|Tc!9(lYZjQK2NMEFa5+4ctwSfLp2BgH&wZqM4#Sd+7I>&-`q<-m!R`5`4+Q
zOxS-3#wD#_(<{sqRqo277xy)^VAb=fgN_Ny!89)R@)Nife5bwGFUL<|P^dRh2|#iv
z(aD}UeYE#PqL=;jHnJNSU&`#}D)x};{CSFkmf%n8cQtA#N-bV;o^yYj
z&kdQwgheM)(YoRby@^qSN+Y?cz`+_DPl$)kv_=)`80q}$yuBVmYvzid6ooj}{5PHS
zZOCLDnTbDE2Nw`vS_1W-VMgt|KMB%zxvUBii_e3@OA}uiH^(}P;Y*odBs4-Rb7c0z
z7dY4EQ|k$ZH*kGVM+=ZYV;02^Tm*k1GE{9#$7Z`_!)bn3(hTDMR<=2yQxF
zWpsy5UEGf$&cMZEdEFCcT9HXzm=y^11_zQ!$wq{Lcg9d*p;B*1Qfac)?QEh1Sg1M_
zl#0I|d}7^aIcj-a8~iQYL_c_`!32HNL2ZDfP&Gj7il{#VpQQwoKN&42ZoGfzOWv`(
zKmcpk;@Ihi3)^UVx|J0}9LC?xzRzRex4vFwx-0z66cdA@hcPo9T{xw1tiQX=TtvLL
zODYaE1?l(%31ZMv1i2>>wWxoqHDEcaKQ9vCS3j
z`}dohTGzoyvs&DoRWh1lewOBOfE^S=i6E>gIV58;EpVa#;
zP?!od(C#Kl-*s?(Z2~oX0n;4C(cgT0p(!mg8!NYJ^&Rm!nE;;pJx}BtR)awuEk<{G
zMNO@Gp^PY@FJL-HNOOPv>XCOK8bi18R}er-cv#7nvAHg}ND%UJ-8@&!95>?%qF_zc
z(B${PBAgre;W%|lUe2O@Jr3OxP
zBT?HP@GBijU@je2AMpRhOy$kJ7TY1
zZFcq*tH#XBgjmnkSQwv*=@w^5%vE%4Iwi``$F_2#nL}CZRKjs+K1m|LLpF_H`$Kf=
zZoNX-y4g79x+}iN#B1+d|+0(3?e4N_ZlB;HGF(6(qD-EB6caQn2~Q&`F$!HYU^u#>UF%c0ehi0
z#jo3nqi1xq;d)lfpNo-X7{ajNN?1IsY|uHxp%`zG9m&(Nl=QW&Yv~y)q1nqsgG@G5
z`n9Os`c*T9LQphAY|aEKTbxviL@zvRvXyBXV|RZC0;w`QSzY%2LBAYiEjubjW0rW6
z&AEh7;6Jm(Scv*%uwaWPKGS?^hZTS!J@6`amDzz?-UBXX=*OWVaC)_SL$V%v8psOp
z<^^TM0P9X(E9HeWB|SjD1P*FA5)xmKx_+W5topTot;vPzK`ACuqc0Bo=i?Z5Ea7HU
z(K3Ha(0Gg`YFQ!CnJIOtUBwoNZw>!=x=~ca&fx
z{h18@RwPkO&`Our^P38Cl_xQi6*x6;#=MM_h@d?qYecD+roCM2EIO9WScJ`dL>7S*
zrHNb?qVVdt7=c3yUDDn6J57I#h~7S%Hf6Tk2fjw5MrSPN5UYWP@;1x_
zg~Jmk4Uo<_swokuzi(IRe5Vj*?vYI^nu){DS@)KF%EyE}*NVe@gH}(+jA69d;3Uuz
zr~+?&?i|7Ow
zjTQtc9dczkKh91o)gd{)SCF>hx$U8%=3S9QHL&XDG;TRER#H*JhPYIb8(}4
zAFC6&4uhN~8+TOT@}AMS;|yLC$0UE+?0vsHcJb5UMqbXOa5kEP=pvdl+&nNcAHD0J
zM8F~Gu0cDf&}lK)iu5cLvbk0PGiT39+A;N1vhz7mEgh+QFIQ++N2KaYu`#|rvwf|-&cdpsN6?Q-0+P1;asw1(v-&F-y3{>Fdt{H%3Z
z!3UMn%^CsfZ40Pu1tmypxvp7jcd8NJLvW~EPK+KKRVw-`z}Z7P-x2bKz4kslbPvvdMC=U
zqxsO+n6wCXIZ2m8-pm==C)j@jv6QL6XNm4r)9;}*MvqhTlFEPAi(sxebI_Cb=Y~gP<<)U@0!`a=(5W3f+aR}-Gto82yTjB`n>af
z(f~o&B~`8SYGxmEBNy9qb1QmhQY|sHsY}H4u7-(2a<|B_Uq;GTGC&tg@uyzBI{L%%
z)(8owglQ)qhYnk3{ywi)EgAXQX)87o5udxz&^+o;N%
z;DZOGtJQ*vW4g>XI{`(w}7Z
zfbiXQ&emTIXpTGt8WUy)ES*&j+<$uj<1!BxN|6d#Duw4nDAi&r*
zIyO4C-LY-k#*J;;wr$(CZCf*MF=saKD{50!La%>(naeb%Mb+q?95w9$}TRTIM@mWX=iHgfnpo@6)hCXGfVliIbahHi6rL+
zrau(52s${~N*s5}dIOO_OhG!UzcOZ%4Wys%TG40r%!41)97caR+ZbCm1B8iT%?`x=
zxCVbIO#oyb*PjHfsGI0STeTOqK_^>+$&o?3X646In;Zrw(>G9di;6xHkY%bVuxtd4
zlp1Gh)?Gwh4X@}jW34C5BF%_I-3?fC?HBdpCo#qw
z|C2LF^8$`@k^C8nWzn?({c`^L
z6Id62feFMrWBN{3ueM}=kJO(na>d~HF{3p@nx!Z5l`D{G#qMJtoD6cdzB6Im`D{Te
zr5>FmR35+q^U1M_3_t#Q5DlI9(WZaOHQx^;yksh*RF|y)_4Z6F3N*c|RW%uB#^nS~
za9O!?N$TvII?ZfPPxgGW+D$Pf1d!XCHty>G1L&n`;9unUyzB
z_B}@7scqdCB7O!(twf?!W4K7+T-4SBGV6-9ME$Nypn|mlyQce8$@4tjs=IN|81
zi!(C~$(-G#xE((4EKFw>=bZ2xEJ0K1Gf$$9ac=Of3TUA|L(mzsOh4%^_Ktu4!H}kX
zxUe)lqo&+7aDw^zW&N&4R6~;0?UZhkfr6+do9W!_?lDnvR4lw0u7`Y4q8Rwf6*oX?
zvR?BZ2xUXCr8hCxc{|hSZ{UB{G|38lsju3LW$O;D74mfXvp$8=G2mwzaWB{BQ-&^-
zKQuDB^t0}JqJ)xGb3zUr^ueXu
z;TFM~78(44D|2Q2O}IxxY{;D0`p%zaDnIwz@(^?2DroWCfj$PKa?XF3b9Z5OxX)&F
z7L0(dwUe8IA`M)a+c|_BBu^J*V0WTduI=l1fe26Pnhuv4*FlcuG2a(PUS|!*I48_)
zm$}q0>#6-*u-DVNZtPP06l*d$iG@_T=UAh2K;;{25@|i%5(jo*sOeYfWIBJWf-gqKA+M^MDC!Pu{)97RozHP#!M|nbAb}<#)CS&N(72R8
z+s9nEHRp{x8p=|l$!z#v!TNv5%5w#Kt-K-*si{B1kufm#HjxV9uYmb*K(PO9xk~ma
zCh69HjDdQHz1a{8dUH2EGtX+--n+gISmI%BChUk|8drUS^f#BU?y(C
z=%hW3eZv{rIJUL51oPSV**8S04l9XYL*w%8$cZ@vUn!ELLCa|ze0Q77?d3+eYf1xHyN0
zaK)yTs*DkDdcSoyx<0p5+}O1GI31
z2(EO)|5bkkGRXi7jIArd4r(PgX4o+=M{7EuX$L9wvuMX_=>>uQFd;d0we3$%24AZ#
zqcDLg;kxZ*lnOT2S}Q#gw9mFii~e9+I)S?ygw)`Ccz#=)06#b)G~{6mZuWxWS0s9~
zjpvQtls6beNz&^i7~)A%rTDw1Zox2*sw>tHBA9>wizVFq=brE_fDUsoUi+U#wqxtO
z@bD<=pp7)b+F8XW!$V}Sk{@K15Ezpj@rO{o$3R5e%Rg#Af~vtg^5s4mAO(sVvk;Vj
z)@}j4$YWZEuUV|ZhZ@9zBW|z)q1QRBmc7jeD(W^Hx~_xE0>J76j^PJz>|Ef0WfTBl
z%SC?{_r^NrNO*<%q+q-#=BpKFxnv&w5zlzg)_1&Dyt+>6#x&H^17+U^rT?^>>U#hI
z8CsT7(JgnlZ+SDDdzUgVzQYqJT~3+%yUS&(xEWGqRpdEn?WzRYB{q4z_r$>HL@tEU(AK8iLlqe*%fY2`30TidW8t*w7Y
zyy7(nKzuMSwL=fq>GdqP4rPLGhXa)*)}Cakwv;4=w~2ofQ}QqbPM3!7ED8`k1F9w@whOPW85BIF@RB5J
z@SeMY*2O>TFbX$_5r*u=Bf$NK$ZbcdY8_o*h5w{}*4%%@3?)9UNym~%|sf{mi5tI6i_>>Otz$0W#J?0y(zsY3P?pjMei
zUM`Lu_o+9v#r}BVj&tdmZK5|Xu*2GKcTlC!uw)x;BTnv{(laVu>>K0c3^mZx8+Y2gz{j;T1z~rb3uLV#B4pgi$=m%JbPP~&;I#Z5
zOBvtLc#eX6va4m#$L5q@kyn2*pga?t&7*wPhTj^=wdjTk993|5V%~q5F6i;2PpEi`
zUpI2xYqU-i`Ijp#KR$+SCXWih
zG?>HAjxmhA%63$}F=fZ0MFpg_1;FUz-{~U(UTsnGBdnfSmA`+SaA-{M!*mY|w(7~;
zpcKrV(ITm~NjOM`>~||}MwL$(QKC9%9nsI?8i-dnF&W@_Win=BgSUca2wCmTepbtz-naa4X
zIew!vq3^@1v2lOkLD{nSr!~1=*Yj*b6j*S9{TO3P-d;tN#_09PUYiE0WAHu12vM{$
z0o=u!UdEr
z0FCZf2lQ$F#c=Ey+EX~}o)`r0{b#83F#`HqEzGja0R>dX{vA}Q?
z3Dnv=6GrqD+PF}!pFk%=ognEvSkJ`KSuENM14)AlcKApAJQaf~=gH-U>j{oDZE+Tp
zC@ppp9{GQxd|5|l4cFLw+J65}U4tMAzt;r}EDV72_;sK#5yshid9D>j{m5)Cq)eNhf0CMg3FOc9W$Mn
zPbqs#&$)$$qVKd;uu|wlcUC+Q#%*l=CJ?H`Z=-(`;{pvS3gy+LM|$lDF%UPOxKi1S
zO^!G*(GRLJi~G_dPJReD1Y)U9*r23L;Dmpe?HSgd$Qs*m&av&LJCMdu5FAD~`k7B#
z^ro1JkEZ37#>A5C6|l5)Vm5X(bU=s%h?zAzF7ZoxzdZukwdRDm46~k`xP0vD$i?}a
z)CYfTc$U5Z`JYP$`mM#K81I?Lx_0mC!>aHENV3Dc6_?dI&x1~81pAVS%mRm9A{f+z
z3>>sGS2%=kf+A4-7^r2@YS6YFm-#ISB_3YgaZ7&g2
z-B}Cjhx01(wFV_Chxf5znimSwl$bv=psRmO`lcSORXO++3{R3s$cfe$x(_f5yMD45
zz~e}jk$0+TWELQ}y^?0R@Qc~Q{?i3o!W%&B$n^yPh5N_w2Yd*f|1TKmbKowopLZk0
z+#tVo)Eqr{_bYrcpr71wZ2HM*s%8UyoX%{n4jG{W{s!({zyLj-F2Yvu0;@vX#J-%E$Aey4hR>+cO-)NTCtW33(cuKj@M;v+z(hV-q4
zw}C4f4Cjqd7{R+cuEx04#uhnICjjN&tMQsp
z37ugW=e71Q-(jpjTswA_G}yXY`-JZGqNEnqzAnfW|D}(Pj+oZi#m&lWC`3QL9UHg_+DP0~IT=-SLOD2+i7n
z5N_k%5y{Acriw8KH~Flbyh3(fZZvOifaKc86*;eR$*(~pIO6HIcadTohnR%>7t&x_
zq_|1tdN(aWd9IWsSmmRH=;mM?*zED&4W3QAvOyVhvrEyF5QBf>$#Dp7j0WM8OFVYD
z&F?NGmaAYG%sDBbh^&Jdxpq1D)0Oo3DC%Xjkd3VZ0e@M`z6;1>D&2Afxtt{sfr5C_
zp!{R)nMIWBA^_$|`21JXv}O*bk7gUdrk;u477qxu0k|W`Y&-Rh;8ubBmmZ-Trkj;}
z*GU||Ql7QcH`9N5Ok<`jYDOca8(igdN0HJmq{ceRid#prX|~E3IMox?+~CR+I2-g*
zy-{WWF3N&Q`|0DLg$-3FbHN#Rxd^ciw!9B*gGl7_u(+$guLnv#c&(>ngx%JvQwB4v
zcYjMq#2`mQ%xEv~WAF5;>MGX%LkeZPsI+R88(bK*A(nrxsO$+xYH-X$YWrZgfI!nZ
zPO~SOfs)T^l=>3$CUEu-c0sM-{`5&(fZu{TLbt){WD2cpiI=`NtBIA4Ujg)*2q`t;
zo(wt{KDr3o7Ck+{-H(8Yoof7-3PVct_yk)SOZ4(g#fQ%qXbYscJg{=K1xy|UxELbA
zZXA40W0QZms4Tz7Oz6^~3woAdD#s_V^!yx_eh~3~t0wc!Zehy}OxVwrV;*U2UMW2n
z8Tj_vP2(^t5U!jSL6JNXMn(!tplEjSO|`pydI?^d5g>S_Ioavj634GZ3xMm;^zu3&pyo
z*FvI1Kp%!|7&mxcN!!_7Ia6(SUmbDXjx2h8GO_k@>lq06x!87#jR8oFfjDo@0R
z2PesC`-QBFr39#%Ero+5el;s~y71WHiIJru?^@^oep)@i%0UrJ`SZ4$ZIfbV_2Xk6
zx*>m=nJ!iK$E^C^jnqJx9;v(+^|N<+dp%3;l^X)ISZ;$$q$(O5T>0R2eAnG{=3m40
z$Lh`cV2o756pl2q%phczY5vGTzjXYaxG2)mepM7n*9}U&)jHRA#FS~?_FY!FlVRC(
zaP>F=I@gw8VW+*$6ls>)pyX6zYNS)RPKSS$r1-MQV{2@*B|UhXj;%?7JLb0Lg**pv
zsTI5|P1Eln*y2^WZGteo2XrvNhi1d1_S6yhjC+v?3S(?0Mzg!~40v@b7;I
zM@+_rN`oU6C%H~a(edkKpnFKb&z~vEEX}yX<3EHvN}tW}KX~FwPy{K-
z5$=@k=CZyY0NXM-kv+#fS98vYm|K5p3h!LR04n`_!%ngy@?mGaGHec6hJ6Px>iJa`
zU$F=fsVib)ZjR|*@&v)Y4YT8(K7n%eda1_M0CbGQ4Qw1p!ZW2`2+&YB;?O@k`J0r4
zje;>mPnwX_HrnqG@yh6uzT1LTH7G*qhwm5tT)R4o5>>DR^=fYxaqGh99oT;!&?@TR
zAnC*yGUG6;#?uSn2;!T+gd=4dVUao`*Ll&Tm;raKz$p-cRY;d6g3OI?wSJI|51tz0
zf&jI8F3vivm;qZN-@oFIo@TL;g#4yKXARi(cLUN%XA*~oF%eE`?PFFg($vv7O4Rt>
z(t$T0vj2^l3)?x+ttvtnFB5;XHT#&1Jf^pQRHt!({l`c@10C|D$Vn5ldrsZG%_lnd
zkX=ZOhfT2m_|UyeU1eqrpoL&3;O!$MOHhNjJO#fMc)12!U<;<8rZQP9W;xTtn~+0A
znd!&9^I8HD!xn9ssGbrPrcGl#ks5PPh4iqV-yVu8a-Nu+tVxZfmQ#NR9JZ>sYb9Pi
zd~TJ=h3aAaf{7SNI6lrThF#u7-fYK+x!1yHho-K3i=Yq
zX{sc8JAm>LSr80(nCg@hoVMJ;PZRD?rjW#~or~JWK%oDO?q7d8LE^tDWee|_Yoj>>
zQ>&H36f+Xk$$NxGEX7mgDQ~-oN3P8E^EMM%0k2~$;cYT0@D5dd-5&8;ku#7}i-7_D
zaotvCQ^Uxd!K@kUvu1!O3`}K_+YhOT%;J!grEtYH4sqPp*N-+02X`muzRgWqY>DYH
zp?c#=#>O?B)%SlK!RI4Y7ykNspgV#cdP;Xr;t$Qgth%az9c@_mP41%q2vi6$sJP;o
zRiAYBxAUuy9qAbV!gR3&qde*I9P{D*f)9|+KQa)((HWXz(F2on=Ef41dns|d*ANq_
z+yJx)%ZmoQC_hva!b2}%JD#>2H>_lPBeOyNUUz8M_R)WL9&uc
zzev@Dnz*5ypz{B27s&rHRT0FIdDu0H6#
z85@7AsSdl5qe^UrBZ%f@u`x~~*9ea6r-E6d9E#d_;T$F!5
zXiE1g9Y=dBip^$Ow!y3I_>y?S7#ytAl=@nVQqLI3-rB90p<(C#hr92;0z$R+y-NCtXR<;*}lI1(;myFv%
zS-Q{+#p@-tz&468a55m>KaL$BO&EXmV0`W&ukl)*PGEbMTE=CHe~?Ag-?^a32S`{@;!HIe#43G9cQt&|Gz9C
zyE|Pv5|kyu)Yoy
z?w!wL)S-PW6dD`cyKZY?crOe-Mz^m4fdYPYJ8rSw1(sOE`sAt%2p3
z(mG}04^A_FI9h?0b2_tQ1azJ`rc|M
zNlt>AhpSh8yH>d)S+B<@YakZfOroIMuO22#UUFmop?g90iw35(=1Fco5MkHpf1auDV1IN3_LdiETR`acy>&M~;^vjAIYo&m
zE|qz)3~Qj39DEx@>#)fGH*&3(Pq)3xYd==EwMbyZX@%_1=Uz^yrxml5Rb%gA_xf@1
zR?FYpJ~m=>qT>aejop7Y5uH%-bRel#5Rt?g>)`!-u#T%Eu7)_0pZr{w&Z3fx~P)rNdBGy0@mdaM;in6gCMS*D%-$p^>u;B
z3BJZDwn34ESvnT5s1(uD=CN^!JC&o2+wKtPsd|1O(3L&)ZL@!_o|IMT&37t{>e8*#
zXqqj^nKg)e43;8uNXliJ@_thLm8QkgnzB-ckb(wch&#b7l|w6`MlDuLQ(S|pOgN_r`UNa;u-!A^UGTryXlwy;^PF+NpdMXcgaz%6WbC78*g_jIf%
zA^#3Mr!Ik#rGy%~`(yXKL|XGWaJHGq@_o+~iG5OH$Bus=ijnnG>+}AO15!I6nI>BT
z0n~3kxg;6r7+0%T)<3wywVv_Y{DUEZJ$spR;cl{lpa|J+n7`xYj65GssP-#heIb}b
zGsTyq)=3z`^f#_6ZS}G?-W1ODw-QYETg@N7b-m_FjKjm}nR|gr!dL{?dZ1@{)%`N)
zYo@sbW@LZT33Bb8X1`Lyfoy>N8j>+th#D&5LhICdM>7Gvng-rrSAZ5g1M#GnJ~()L
z-T-YVYffQDcB?0XK&Ff?=DKGmc0#3=wv$f{-|gYhgY@p@MuX+&H9obUgAIA;(8bTG
ztQy*Pa*6}#*;C=#?1ygyNOgckgQ?2E-uQ))bq{}A=V8Iap&My1;lP|l6tkCf<$;>M
zgCxxM5klF<8;Nf3gW>PJLtEdHFv}{8l%s}j9lVdYWLSZ?POa_s$lPNG_C7%$43rPc
zQnPs&n>*}OMHi&=TM3^cA^uU}#TW}azB-wYm86(=pi$pWnYQ---C?vZC#g4&x_|#f
z@6CVIsTDKM6?MH2X{C8`A^K{W?I!=`uMaQR-^+w3?+@t#kKHJEYAo>@+Wy3UkDx=w
z>eA6Fnn(@MU-4G!)=lEg)*C1Lg0j>EAMvyXZ((+DmQY}nfqu@rNY&AkRwS&(u4kzQ
zHMIJS=^w-&ATRHFf674SRWBluB=<^S&1HWuIk4DiE;_Q`y1J|pdfi$}BpWae&zjZV
zufbgl+$Zj_+ev_aO<#xdq-3rYMtKX9RSoQ*(s?mv#@&9;Dyo08b?i%*D87e7M2kum
zDAeA3tX)xXKLw!Du&pa!J5Gh(e>r#|BBZ99sKM}vl-Zf=Pq(jq{)y-CM3U%%2Q7b%
z!MMh+lh6B6KIDy7qjzfzA-*l>PLR~w{OU05d`vY9xRLq+6~B$X(8mO2pN39=3?j2;
z5q{AiIlZtVs`Sq*+yml%5uhV!E7xD!O0RO^IMI+gof$7_G4>7x&~@LFelFszS==dc
z;2{ZUtGwJ|s@aA?rhkqd6IR6%|8#%U^xM6gpijCrhZC4FIQz3^FW9aqB6O5X@Ckq>$End&S+z~y5u*!W{AIn0TNY=VE~&$7(q
z!aPHk-}V|S*=EHu4|)Z^}B=JJ}Zf%lwhOv@G1Z_XJMecX-lTR`x4epAht3D01-AEYCZ)r>N|C
z-lr2sd||%a~6N2re$F<+kxk?vm~g&B0&V^LS)
zg9wGsRDYAmIm-j+h&$LY}5)~~yPF24}4
z`n$dIoLSnC*xyW(g-Kmh6IqydR6K|sEG~AHg$(oW>Yy6AIzip6
z`>ubJ-=sfzm<(8CA-*q%@L+j$QrOhn15XNbUn}Qpgj#rHb9rtFtZp5T=RY4Umqg$V
zjM0fB69O^L===;GYtWXgqok&t%8UfhYmJtRChi!snG+btcYTh@QRk~Nn=gdNP^WleEZ
zC9Z6JNDO#9Qo~SzK+UIyatyq4SIO9{fn|Td_52~_5YnHaM5{@;73nC8VLqnVw(yg7
z0ej^#P(_3w3Zd?1IkX4IYk+LVC0f*}mu35Rm2RuCr-U)}WfWG(oh%2hXS1`cUHKZVtb69m~=K63xr#NH-rZl@DtyDAl
zbld2n8Ao0Fhy83G#L&&H)2(8buc-MF_CjKVj^Q{Ng~l=|&#zgylZM%)C&z$D)@x!b
z$7gi`r>lPy%9(JLagSR~;a46$PVKg3g$-<7^U7NYlX36;o|j>Xp?f3BuPM8{{mVts
z(8IDC?IAVu^fNv&yV>>3(cLdowcXz3Fk^tqnn2s~(SSb0HgjZa{5p9-8!Qr()!Qad
zELuCQ>3p30BD4KO6J;hLcMh_Z-k)m{YY(=}F;#!;K%iJM3eNFg3O%W;6yG0XOn0xD
z4F7N^xU0srB3a>3O+C%^d(NCyU2-g
zA2bhv9T3tNEVUJ@C8xl5%K;7d1PB0jR3tgOe7kJ4@szjMxZ(UzJF9&bj&B}$(2
z?ep?lP#}$JsQD=&gn`8nG@UBa!j-73+3$u~I&V}9DzkG}jb@i4VfrSKE2)EHe-+Oi
zca4%BVC=vg+LIT7`xE`EyhXQ9AtU9FItr$bH$tI7!r3L
zn`$}HU-%NOLnvnVD>g@7@1*{o0>j)FUP@oj&P
zZ*cPsMmH9|&bPW6&_y#%)oh;fT9!;o#^CoeHaXaKF%Q93jE8?k_Ym?_nNCmH`ED+|
z1(^3mqGhnG*7Z!)0GcW@M@_Q#;{S^7%^=V;1(EZIn~q(5e5T&UaJ;;XSo2C2NGo2X
z<4uNQGLH&f1ceqDR!`qx+C0hy-G+Y=EeC~Ug&rjX$yE>)kA@`xY_$kxFzO!AXI-j48NrfK|yNgQ>)V%>n^a6_n-=A?W>!6jU{hD8j{W2T)VDd
z10ON+3bOOIu9te3Yc`njAZu(}sbvjd0=ljfQpM6db&F>c7hR!LvBQx`iP;gg99&|8?ZXAB2iz
zIwn5?heY9wY3RHLp1qC;x;Vw8zR&>^Uz?oAes7I@le}k#EuH)iXTa~1Q9TnWF*G?K
zFd%PYY6?6&3NK7$ZfA68F(5KCHaH3|Ol59obZ9alGB!0cFqcux0Tcx?H#j*old*Uw
zf3$gJRGdw=H35RV2Jhf*!Ciy9yEYC@aDsavxVuYm3+^t#-QC??KJw0;o6M~H_ggGz
z_OZR|)Tw%)o9v^!5}lwi$PoA$WaB``M9;_tkd{#dSsU0eG1Ccwtc(H7^o)!waAah{
zc0dCMbC8XQfdh~W!0uoMkTY_4OSA(pe={<2!jS>QfHpw8H>WYc&=nv9bTCkMwFNQ(
zC=LDr@*sN$Izt2dH#gA6)Z7M0^_C(GvURmHH#KwkUE>oS-S0}jvxVpZk_JYWAZL3^
zbAW-3F+h@Dh8`dba(*+K11Lc@07Iaeft3jWWCBnIssmJ%L=}|)Vv2Gq@=8?ne{Y4A
z9BpkucK@SASV>t$j20jwD61?A0IJdg#8i}&fB#el+PvvEr3J_;zxjXjyaoPFml0JK
zRMwCeWn%cf1^^Sl320|;{+sr{w2{9l1N@QpmTF=Lvi^qvKxyXSV9Uk8;Oy*7Z|Z38
zKo7DrrMI>EN1w8pxjn!cWM>I@e|xn9S^@uQjH8Y5TRR=hfd4A+drJV)=0-ppd*JV+
z&!B%ft>0Stmh@(K_+Mdfjd1u~)9PR00DBeIl
z9Sj^C?E(7#*xtT@#uWb&2m}Z_+S&c)komXE?tf|iEnNuowlZxiHxC2nfB)Vw0~<&C
zuYdLCKbLI;vavU}cd-9gMIgY$+zR+xz5VY!Gq?H2CL<^-@mW+!nNIp`@NDR0KyTf#
zp?7d`_$T^zJwXv^E&vB38-R(E1;F?=q@p&)!XRtwH`VrVzuPBb{?;c4kew^Te@(Te
z4anKX?Z2%i<~GJAzq@Yie`w2~Vq^Zr5hx+@Ke2BnxIZ>ipaXyr0Q>>~x)_-;{HFS6
zQhuA6ew*J~;NfNqvIUqJSlI(T%uRrAFK}-522MbLgPkML!|iX!eYyfwj4n>wm`mSBM($w^YjiBiY>kv$+ecH>S5y
zHU_%d>={0qKGHu(dY-|Cx~|K)yLgXv%Jt-RSEvbUz0x!RfmZT^aQ
zv;Cd_wie63f8d*7tH0nIoAn>a^rrN$QcQ13L4UzFrM7>#-(qbI-Y#M*pozmDE7Sj2
z|8=zgWwp
ze?0*Ixpd!dhkxVWPagmf=mIo?o1Xz0ar;{&)CLxMisp3GKZ*yO8|`D0qWtTe=yj$Ron2y7a7%6Q@-i3Tq
zREJ?_h2}Dcg4=>OH`emb;q@LPBGjN$aGrWne?L1A>z5>yZATl52Eips6baFM=~+Uz
z$6395>j<6?xrk>Org2Bd4~ElPd_Bz4DtS^hXql0SvHfaeSs8GiqEg-qCOeA!bb`^?e|tYuBOB0uzL6z=-7L@Jw!SabL(iC9hvc}0ak~g?4?f^AGbz22!f)_DI^r7TbP6b`5b&!r)8D~qQ^_Zh=XVA
zenf%YleJ@bHRv<%fuE$R57AP8q~)~B=FCw;%LJY0-Sk`Tec2@O+Zm5Hqp{`
z6HWY7L?@jOrA?zQB|Q?cBs>F68OSH7h5#S~VRTO4G6aK{s4j>SclGoG22jOTNbvCycP
zO8D8f$pYdDntOWU(tbbpmetKv-LwUL4oj1UDue{+99+1P8&
z5e#P~Wc4Ix8qY_&$|S;6!3pOo=-8ABn~`0OeQr6lT+9A*v#X^yGPRo)1UBi3KRf44y~h?ztk7ZJSg
z(I7nTs8dvEyHf+Ej>m0(1KO{ZcxdlxtQZY5^M`A5>K_j#
z!cVi;x^}T)bUm6+fAUz5wBqGy0XQTvy}osvhr)w%UC#naRnt88sI2rfKfVI*VpHH<
zKOES3Y-NV6xqV^NieN%v(zwxKEXeuL@!s-=V;In+ItL;?<}2w2IU)<_mB*8Zd*{Lq
zhpb&%niYlWO|0#GS)4EoIfAh$A2^d9U=djVQBl&dR9ffae`l}%DVeN2oEu@+j1(Cu
zbpJIbT?*F-Xd+o`?u7R>)h(`tsv;!@9?|!rD=*Ie*`SpX_fEKm^C9my
z6h0xw-pubm9}!jxrarTw(1{gCHZbH83&-+R?|NEwg&jB1)wepk#8}!stkYL+L~%P=
zWjA&`sAxAXe_Z^!smsQCiV=ISOzV#+)J(KUzPK4YG9I%Lf-{oZX#8|2);-@Bj4im9
zMe^91Kse9u>JEP32j#-#peRmM7xwa8rO75*%IP|BRfOG3I#{h05YJLee+NumL_N^7Yk{&nkEHQn)$N#w
zTSr6Z`in9mjLHMM$Ag*oHCa%Y1uu?w-}rkK$srx+k^@3S3KKc&N1NB6FgP=PQQ%mc
zup{2OUekl
z#2G+ze>4L$xV>Vtdpz3cFWV2*oRYUEwoaV7rd^{cUs|D(cY3?g?kps1+?}zx5EkHv
zKAZ87B!G>QrQVn|Tn|L9FvZtP?M7b{Vyp8{{a|uQjOWO4Nhoqkb6Bg+PFjww#<*{M
zu){8~otqyc6Uh_3`G$1N!kO%+f-%(K%5k}de^xGI)*;Y4LV|)h&vV*CJgRz}e~ID7
zVGt?T_Wi`VdSVyDb(jmh6{1I_!aGc$L22ShHAvqn3+u$w$Irxx#OqrR&fVf=dY9f7
zfy*csaPxI6)#deyvV@}fj42F8B1WCqJe}zpRo6v1o8faDuEK>V`7I9ohiddC^&}o-P!RrpS?QxHD(I1;>)XA7aAVMW_Mz{5dUS8&7K1coB&O9As+TPKF>Fm+
z7DR?@qb#G3dZT;g9me}k9oBu(h7v9poH*IiWv9eIQQhSGGmH7K32Ss48pif1KB|?>
zPd9H}27PB#vr*SklqI?!7p<{@$$7S`f7E#aAJKd8p8`6zFtJ7iX@8=$fEzMxP^XUF
z7|&XBXm?;kI#s$2=V}`xsPkhj0Bk7{)YahPjr@u-+_z&K6_v~fEmTrYk=~!<7xqz8
zghFZ&kl7us`9)Xe!^$`(QuVzOD(xf5&H(RlOW!fn=vL?~jl=?Wu7CWm2-Yta*YmiFzPc
z6eL^G21ar(q=S?_z&djjI`uy4hb=dM<`&%9!r>?enpeNL#96u2C**-4+Lto7BsI|67`27DI=g5bA$ygH
zU1tsk`9NJO8%~0I9AIu*f0VTEvZRLV&s+$n)5$M2%v)Ij#?ee}qd0B*{7$Xay^Q=g
z$M=%OU5+WXF<-_eKqou#;8iW^M^n6%-`@}C0Gh~BaZ|lEhu)!r`W_Wgz?Z0r~C;Sq{e^&^fyyO9{O%tAD
zmgZQ*QA7OEq+6_>-cfPo{TDnAKTfH_8NUvLTyKrc&)nFTt9u39pZC!l(x^FhpL$L6
z>QQQe3hAC0@@w!7V-#`hPF#a}=$P_*ERc{JFb(Y$Ss$(-m|L2JaSLF5cE9Fr?}CJP
z8RsX*#Kx+ROil^df1`{NA4B(5*6BimOAm3^k4K|gT*4yAaSVTQ*iJ`>VS#_hPH38J
z&8A1&@5b>!=3s4)rJfP@(oO*k;V$q)lhW
zVl!TVSW@BprAy5)xf225a1^MQtPp9hU=7Th5gG3N1W-B2ciIHQM;7IgY(WVcfu>0W
zbzVP0?yPG4i!Xw`)C5G5yc~qoi!@+22c6vIllo|wTy7|CJ+|P3Z8H*_R11o3$nTd7A$+b%VIrB^Ot<
zqX%th?YC*91GAMgl9+D=L=1l&r1Thz_{8sJW?q<^Mj>!{_MXrO~TDFYPrbvifQZw5~4#TnUj
z#&yQOe^zZ}=p47$CGZ_j&W~LGG_ruD??zRS-Jp5&>-%DWq1V3kgsJxF95SGhecUL;
zz{00$-A|%N&s8eUkoG>{e8g5{UoMbqO}|6Qzlb}5F!sbs-I2!IPeTwXK>G2_N9e?B
zZRH8%0F}EHpRgOvAY(ly;`m`4Px@3JsRmt`f79^VDH`=XNPWKMK
zs=l#b1}#Gs8C8yt(jKFT-m3}T(4dohAY>}y@_c|M_r1P$g8cN}*a?HJJ=nP;LU!-^
z5eV#;m0_dexL2RQ_l@CXqW_XPxcuHv6#WrhK&yxvQzdwU@KX$_980jQ$kg7SyH6Va6-Rbcl>UcqACxc${|XE@uMW!xqurBsuE9bZ8Oef&<+E^cqt44
zomV_Nz0w}HD0`ED4SSdo*t|=KC=_#QQwBNQoG~<)1v*Kqfj#--;7%r=N^EzLf7t|k
z!6@?5$jw&*&eQqSldH`_rk1i(Ug)QECWGTv;!C5own$6%_j6;c_)Nvps67Z3j8g)=
zo2h2#S-jK!@o9SNRZ@IHDW8dPf(j7ol68JkZL%1^!I(S0r@_u-HZ-;4@FAb3;6t4(
z+Hj1b6pDVtyPz)sN>d~NQ_!`tf38OIFZevb6y=mz{@g!|v7-er8xmlV5l?-L8sDFaf5N3Z6I#)C
zwlx}6Mvq7Se5>F-Iv3|v&geHD4~MpY#A7r)`b^UoMR6FNFZ!dNhUK!LCH?ZD39<$T
z{}URt;3)&ihUX$DqKU`)JlF-5s*49q5+hsR1DPfh*%#Xdt*WAm+)v%`jmx7Ht(1cO
zjnWD~z#FR1FW@MFzR_T*e;s5c0=P)XR>rN`VND>&)K7$24tLmGO}S>EaFKyBGiAz>
z)-U~DS~90nZm7wR@>JXyb!>9@3U)hVvvOCm#qR<>b7Eyo>rH5$TKGqw0d^dqN0|xV
zokUQg7Pgn4c09fNChk0zgR1`ZCcz<1K(o{$LhmKx28Z~1`>8C`e>2;g0t#$RQG5lT
zq`o@5R>$nnVcI?Ky&(fPn1G!nVf#ymEmgQKBOIq>I>CfmW_+PZ4@JMqTAml*`y*BH
zJVvITCZAHZA3&?}_3tOV0xEv3VTSUv2#5}2n0w6c;L
z3AE0j?FEr0i7;-~fBCk-{B&HWUnl~w4iY)p@_(7tieJ8tCRgJOTsCJ8tbf}+MC(Yp
z4L_pjf3O~eZ$!R(U42E9C!25aI4lL3Jt6eH)1_w!z`a)MA6N%Pk$9C(qVz-zpeo+!
zYqsGTc}AZOq#^~dBkkxA-L>7_x0+_)V17BMlJ^PO?%Kr$f1_O=gG0`f^wLfFTASAA
zwV&-*{i8Q?eu<9fOU;gGKYjF>+M!-R$_dFrp|+}bh65^PY{`?ihOHz)c6aV)$x2(EFUEmA@ZLEd;XGvBf`_bdgn?
zi2)lQzqo}Ge@6^97Vpl&0$_HfR%+bE>R}|>3*3&$LIY8Vj4y9$msfL49-K_#kO5vu
zix)pDgEt;}o7*R{6}-DVoury9LQaPFf1PUIBrnHyepaGpi(~4(3{l<0
zVZX)aAe14$kV8D15`x@Nl)}bq9a{gTlQlOwty@oz5Re&6p1y4M>1ykHSzwSD=P1z$
zvaB_6?tbd|Bpb+Y(dpfMPUfUjfF*#JNH~C^w*3CL7fPnRUr<(@N7(ykD^^ISsRp7`cxb30)mS8pi^K
zb{Z&G+p-H`^xyhhqr}>k(xp
z?(bPVInUVFs_)bcGrwU2Yo<gul!|g)H=vt`l^8s>3^T2=5VsJ|7#V#?
ze;?${-rkj)etL$pFao-oZV`y|=N^3gi0>bW(NpRFlRgha@mgTeas4~XhuQ->>f@g)+2kW4LHy^=klmAJ%Nq@he^aBa4bKRt7Du7Q_#&^-K^Ip$B&r%JC>9J4
zFYUwM6PFNQfR8F(tdT-2s4p?~(=V~axWj=q<65H(jpE}<+ct%v-%|arcwX0Or<+gC
zNHlRro9
z@vC@tfQ9FUol5GeAmW?W{;&ccf3?d+{j3+(J$x>Wn+6}V|D#_i>biFU?*=^kQR5Bl@X*@F)&thc;dZhlXn@XQ`sFbmsaKKn18
z>z~6O-+Ofz)v_7Q!>b>>FG6y-kzkI+6*MU-O*F1AUJ4mVv_7Tk@!w_}B
zP3YqLiLE)q;ujL2t+btKEVKRH=Dk$G^T!|F57_vx6am4$i5)1MxLHR$~79f{vHsdT0LU$
znk%Cyda3-ee|ANPi|~Sm4$iOJ*XS9#`LwY*FKJmUm55|V>4ZPj@Btk^K7RFf01M27Vz+VpMw%k4ypVtt?Ok@
z$NDZ_)bNKHrCMM1;2Sb0*i{>F^E6lQE>VQ80YOhmSRAkS`eT#6EJgT9VC~WEjpjC7
z1mIS^5)$#Wx6yPqaEnB)flvu5kolaX4|M6KRg4tt#vW3AhM4c%rKC>(
z$kOrg>iN4XBK?j@p#fpy6Tzs#di|r(DySv~3t7dy0^y*6%Xh~pM(?M&S6C$g4TyzW>PZ7H0%;NiKZ!R+xy8|zyM7WVS|=5$
zf2sK6{h0Ws-!=4Q14sV^X4+!Pfhg%%FDvn5yZ|}Qls7t5O^dfhY^%*9r9IDP>Sg3y
z@rq=n$}d8kRZqH?#R!yEsF!ID3Z98bhdCP0I83YANFIElbFJ|2
zT#74;USyJ)ptwCb&|?qnn%F|EY$M?*$@9#(rmhr5pA3gW5_kysk*d!a*biw=ASb8M
zW9yv7kJR;#UL_W7Y;4B>6x5rbA`7A{(K$M_@1!MRx{=8rr-@l+Qn0gK)9>BFfA>K+
ziJ2iQRW(HO?y+7x<&xNg$luo)g9vcmO$J<;tr=s$aLHg-e3FRkP)+%YA(L7GlNvRX
z#Ue-$8V|pK0<+rwmA_0$iRw0A$I-Ni(hp53+Rs?#L}X8i
zhIR_=aPs=@PGs2Vv-8nC6bWq;f1(#(-}ju-9+i62NXTj(N68QQm^l+%!!E`;;kF8D
zxnGrjGEg~VB?})lgc$oQ@^3iz^p#F*JTerAKK#e0*=KAWOGzhsqrMf3Ol((0JSOaJ+SsL
zCw1}^1=nQzpsCX*@Xbqus}<@=C}Ry2MZ9}=gc-t>Cfa$%EtJ~SyKjUmm3Pxj;!iE&
zM@|xtB&;1-e-2&xj>e}8-@4=KJY;;3QFzj~JNtv|@?ejKOW@F=a0VDTq}+A1dw
zyOxmK2Infg|8p6&2RA5MlY+)0mXYJ=)Lr=i`sTSZHGVOZdb45`0c90~r<~1?rb;=8
zLwn8U2u^9o!bJh6)Igf14obFYUk>>Q7f-4vs+`1xXLXYl+eQm)f6Q49evu!Ecvk7T
zm(>7w#6G0nl^#_Ro?El$g`8T#I3?q*zdQcQwRXNkfUi(GC4NZ3thh8{rLd6ZGc1F<
z`FnheSiy1UsV?O%r*;Yqbc{-SM+$_bz`4m!s59-e$#9va+Aw=Hj^4)ai*9R}jh%we
zFVIcY+Q~{uO{CL4e<#MuaV?HOUrQ_X3{dD>8QYiwW*^wo*6dMTm?JX}U^v8KM}Z84
z`t)9;g}uu;DE{&-x3Wt1Pg7_fM6yNHHHRo|-n0Q9i!CGiHCxS6yy5+TXBi<25W&t@
z<6shrV(Tch?T!y44g@ZnU730b8=XW}q_(l`;bh{JVbO0_M+J;ha`b?y~y|cx)
zt1MxukVH^sx;)Hxxcr$cvlAOChv%0AVHBfWc92=bomXS7c#ZthXB{puotC}hTWqpi8*YL|M
zeWjoU_3gBsZqXTjoZ9(7@-nVG)0+K~1tF2fGqi2aCL!1oi=;Sx%_%&j*QSWPtHiI1
z4J7|`2bUES0ilG#ThB8ZC48Bct!|*uSx&Znm?JVN7rNTaeM#T+uEo6fJD#z&_6tgK
zUyh1-e+*8aG=D%!nnbh^!__T8PZ3)rx(wN?BcrD;p``uNX6oAHeah44+p+=`98ylaK^xpFV*&b!<^HVEItPuQ<7>J
z{vYox(p#bk?8orpeop3uNDwBDnPiC{c}&IVp;F
zf8H52X)<$!cN-;rLml}@nt-dA_v#!Ia41utX?x?uGKux!I9eZb^cycTKT}%VY|5d%
zqK)>u7U`aZ>bTViESSCid5GJC-!&Av**$8I*dD67S$Z
z4p7Sto89$rr1bq(`IkU2nQCPWD;vCRVaEUPRFoK*1V+NUt9Rwj|nGm<81
zfA!l?4G@n?(wvdn%A%&8Pcxt#7A;rT-_i)Yayp!@xY|*0^&4a=5gsz-GM-t3}6&;V3oCu+;dC4E@wbw;vF@%V$eTY@&!)U38
zo_{&!u`2=gPk-irwav_ocZNXGf394lk-EcIk|63xr}l7w)izW&~B
zBIOpR*@|gUKr)qqGw5|DYT2IKHk@FodTB0UVa6kDjs(^w^qUZj@(N}-72%9yK1+uJ
z4lh-a4V0WqY2GA}JXUMJC-KLY15E|-JTy+(UfHiCNzS!<6x4nlSrDacf4I-$ud)}4J}*P#
zB05@XdmLcSxQnYUy0DE;`(o@0!v1taj*z)X6RQZWALxTxkw9m`WW>?j_{lVA8_*i=
z@*zk;fI+1z93j(2u<}5LfD{&lAdY529@o+hCdv2Z+_EuUAa-ehBCz*0R!V3lxB^f~
zz)|{Z33nFjR+HAae?rl^Xyn6frSId|Sb++<-R(E*X$CZ?%+yl3V1hH2qU;O$6U}ne
za1x8#k9Uh->h7h;`DnlLrh$uRI_(a~7;Wu}4^~5N7u0u(y^G-5QJ#+>Y^C=!Qy8WTaRp_c&O%cPb)p5)lIrDQf_p2Su}!jg63(^CN&as*en5V{V;U5c~NgW
z+j35I%Y_||f5Owehd@akf?i%zZ)oa|9Nz?b7?7{mJ-g7MEtoLCK>=0@$$M(&ef8i)<++?Ep$aP+pNjU26H#^t+C{
zA$Wr`FPvWM$Rd}o$v`eS$6gjO3}W)1rQjIfg8BO$f3^l#rl_NW?d^q2)|jUqD?>EW
z;yAX0?Nm==wso0sj&dJ%861`G4Vw{~>6mClxJ5j!P{|%Fhn%S`^xs7Xk_+}9hb&Ql
zG1-eTh>*o!AGsV0kZ8+Ix8J7N&leA=aA@?zf%(CNLQFIHe|#=E&Cu)Ue*ZAzr0n!5
zT{b9@f01N58<$GxS=Yg8iLg6;22}*?%1(ClOmD4q%Kf-`JdaF#BTeQ!qhSet)VLOY
z$j&R_X+#oB9W7{3aR5Kgi@jvpH9&ftU~`6H?UH`aU)f%A+fTV11vZpZ#nu$@A(~l?
zr3pRDdJJ2WQ1f$lvI*C>h!ErxNN%@|dEec^e}Wjw709#O9A4>Z_|bkM>0W-C^g{Zw
zjk2VKFLk*0)yR~UR0stNKDQ#!u?8O?*_xw4Ff$q9UGbVlS}#BEURtk`OAOmfD;>MH
z3k@c$$dRW`gF}7sNy5)$Zd>ey>+C+8*Gc4itrw<2+$psA;FkG+kMMGyhSu3qXR}{?kSJ8$M4QLuVYZ#fVHM+`3U$>-LhHua*%#@%)FB&mR*j>oJNf;UXR
z{coM^tn)ha1=n=lCb+y&jslxmC=JPjfh>188wbu^{6|5!c$kLGM~1chuZx|Tf5S!x
za#7Hr)vUo;KBg9Cs-C+T9G|-TPM5O)9*7x;V&MYEm2Xdpoqnl1HjnksMTOM-3F8ad
zD^_3$M|eGohxbvaTH)-D`K7HXAYTMcvb!dste|{KwREjyGWjGJ69szZTAyq+-eLD%?>kkqDF(d6s$mi!t
z?6j_nP2EIgm6Jyl2zR0}d?+45BpnKc*@d-`>w&%}mN(9;z;_-OUT8vu?Jb>Cwoa41DK-b?
z!93Y#jvM0dpV8rOEU&_Y$$C}#s&IW89z|lDf5}?idPcP#^XTkt8xoz2
zdqMLhNo1NpL4R(~r`R;fekMiI+a9nQ;UoMALk3-^l1@y48{Pm>Cyp>7*>S2i#UhU}
zsaukAw6#Ms+@{|P>Fl>I*{z$YvA~04>G9&{!FZO1d{z`_O=WYJ=5T4SdYT)+Va14T
ze!sS!|B7?-f4yQ!o-f#E(G=nO*H}8X4Y~HYcs5vM;5#5m@}dqOv(y}**;pc*au+G}
zy9p#HTH>Ws(qKFHb|^oL;4CM-r8w^fsS=Sux3ZO7h&bUxSe>e+P;SL9<64z>HZdIQ
zDk<9!uN-)D)K*_QO$B0&LE&3!PpY~icW2%a8fI
z(Q;sJR__8~Wz#vFpdvabJE2<eHW2)#^zTGR7xkV@;)1kNBH)ydbnzcOKW2Uf+HU`M!H^biA&4H)t;g<$*k`v
zR>C{lG0{31T~Pe?egtgC1D(1O0*s3+ANya{c4}ytdbo11$L|dDxi7J3h
z`9oD3sppJu-8HbG8q-nw%!Ysi*8j0qqPXhfQUwea9*i6Gt+dGI!GVa*)1`DDf7;&*
zw7}ZSos@mvmmEy&5+DV3A++{n>lbSi1V-^fAIO@QSXnuks%=?dvmvl`+W-|3&Gq}@s8qsV%VLD
zE7gX&98jL*`AmGoB9<;Yf(SbX3dGg^AEeKXRV@XRg|&!i!ucyrl-O0Z`d7oA4YDZn
zj`4hZ+5s`Y=PLt3cZbdJTtkg2Ie~&moo#w7<
zwL=(Ird7C+9y5E=_z;LbHfmg=HRL(;DG#M0i&5$aGDYCBsFd#4PFoY3n9}tm^
z#S;k!m#XQtl^@{ibxZ|dRaX5>02y~yd;lGu;d9!((*?|(TZ#9%yCxtF6LO%s>wo)
z4M_nOjd3g4@)vzXU9)+&*Ml!G
z0YOxmyNEN4B-P)Xe+-=W>5;{IF?TmU5Mlb^A4O)+b6IPhWs!3`VnuLJ=nj1Cr1SD^
zmtH&E3e*}fs;
z^icVedY5g!TjsLfm`pM0!k#ZEz_*=&%u#>BRyg>Is;$4He|YM^|FN%Jt~RL;ymiZC
z108HFkb1Cyw*QqZ95b@;&iC7XH*L`V5gnDZdc5AsG-rlX(8)2o#ruO*m_wu6?1J05
zarK?!@h`;d_VJG>7}6&t1);ch;nU$(T1b~V{!>3)*Ss3&nVD5zL>I|Ch+Z4o=d}ol
zZ`FboYe@fiOcd;|AlPoWx=PWuE%rIJ0g)P*F1me4hu&#w~xSguoBQ(X2`ROqc
zA0!w8^@i2AoXlwWsA{>~AGV8_N2nQ3q!wN@l{VsT=tn+nJo*X~D|rV@%G)D_Qdisa!3u<^l_(+Z1SFJvF@xi#9+z&S%jz$Q%uoxHD#Zc
z7{`=%e?UPZt)rTFU23v*DLKQ!KexE9V6Uy#(##M)5}Q2J1tgW0ui~du1#go=0c&4#
zPk*`+T~@0VYuYOIMyAA`;oV1`#ZR2eNPg&6I>Q7Uj5Ft_vm*JE5?mp$OG8xDkb0uL
z=+6=qA<4U5Z&sMb{0Nrcd@8V1Vn7lECp@7ve`ky@wGvl(#O#eC8rHIbr;*b;dk}o*
zu{f52LIAC@;`y_8Ar9Jw{x;E~+;}OX?(?j=p(h+Y`7JdVgDv;oWKlFGDCaKHXdiiP
zv>7k`E9VGr*pG`(`B<#oKjj`nVs1_PWeqv!sShlB&6st@m?ryNR*M&_j;ST$*8B{M
ze}DP^*ks0QApnSNrlEHb>rSK;UFRDJ(5ndz2b#Lc(sIBl`~K|uI2<$5`aC<(ZEslH
zuq8xccTDlvHg&A%_xk=>`P#)Kk{p@#T3@wg*LPCX1I#5B46N@mtmFcY*D3-JQAYy3
z;tLXQ6vzzRRX<2L+@^h#gtZcYB_2sJe-Guw04qTt6fSa>%uTwrHG=H!(0PNBMfkZu
zdt1LOw;lWN-WvN;HDp_zZ#HchY6Q2CA!(iuWoSWD5VeSW
ziKIWr{q$!WZXG~%B-7UjJXjz5NZEI7-73n^3O(29jUjPLG+X?;M(~>wRbP^cKGX|yIb&~UQy7%7`;Duh;p!o)
zaxVaOpTrK&SB!6`aTH!y`mWc;u|6Wt*G*HDsz-$G;v=TAXF(nilOk<|hFl8C2|G3ycWROd{vv-YBOOirhWw#7%Y^kOJuz-EqO^2n9|9Z{}Q8$FK(sGS;VBYb5
zvkMCzSrJn9qcY(u5SB|DI1Wx7wI6Jw>6GZXq6Xg4^T%D6Wf3v4gQ_j4l
zk!3kiX}C8Kfnp&B9T0?MT-C&Edem7f^~dc^ex#{a81Y}hk+i-eODF0j?bQRp6)c(x
za5>mV!j4j}g=lTb#sV~p=r7tfx4q<-*HBvezhCdHyc-~Y1SnMBhB+=SZ5qI0*nTSk
z$xbxy=c15r)tm(d)bSlpe@hHo@mxm$lFoLW3qF*Ew0tRPB&o9D#hxNK%+Q>9&4RT_
zKLaVBGGLrVei0IV#UBz
z9T|&;&zwH8SHPt|fAsr?lJzYM($xzhyi=I>$cS_PKR!DW{D9vl)Y$^vZ8T3
zO^n?iQtgYl(QbyX%`Nq=Uza5m#Wc4yCi4RNoeXVYgnZqq%-MaNk
zzbAUi2$Q36fB*dYbbn^$nzb$Ns0TLX_Vvbbkm?eO0e8{7PslkxAcHcZWCFg-YT}vH
zlq5*r#Ztjel8kvUNvr*QqqxPI!nRs!4*!B%Bkr`*L!C=<_i@GG9$aI?8&M)_FA^*Y
zdX*_#M%=i@SWDKGUg5Ulx~7x}tPhoV7g^zT1gNmpLB
zM;d+%&4$ALVBBqSaB`9wZ|wT8@x{|^K_KC&{;8I=`}8W)uKLmmSuaOx*vaceHWB9V
zOqrvpZAf1G=6`sU6s@9=gU&65oms8i&*KKGb$IQz@5Yvw@VZ3KvB5&)CZJ!}(FY|e
zrCGWEN-z~xt}G4k`L+_$KRx(}T#j{hP!8vYMZIRUj-*zLJir=^sCk8#YCI3U8P43$
zq`bE{EkLx+H1U{9Cfi@?%B?dU%dtqX;pPh$tc#U+$$u;MzEkxIcZXIU_|18h`nt)Q@nx944-{Nb&~
zXWNyvO(_18(sTNu+r*2u&#CIG<@+9pBV5%e%hr+RSX<^eLbJ5
zF=H=IT7R9DHhpizj>W#+yE5bXf`nOByLD2~npge4vN`)={rfe|v^1SrrZO~VU?=ei
zmtbeG>)Dh0ZK9fynjxLgrel7P0VFFOCBnCRtWIzFDQgP2U7DsUAul@d4jiShT#`Be
ziO>O5{M6mRz&p~G5b|iW7nl*%M%bI21{JtyNq=8tRkgN_We2T$aJMw+N*UB};$E16
zrhgj>La3Y0k|QHk;Vf51tCe^^H?oafw^_lbeUeD${r>__E3nk&?V1aDh99r~zqWOt
zbgsC7IPn%L(bgxbC3Mqn@COaU0`HSu<;l^zXPmmeN#dYcE2Qn+6tQIe@wQ58BLGXn
zioP^e?@bRVJ?_eg#mL()q5O&(I$p{rXI
zd6`9=Y#a1FcIW{x{1&qP*D3~=p<-Bk?AfP~y}d}WdtgDlishJUInK8bKbI@gu}626
ztqutTa)fjQKg8k&I5U+(n5p4crj}CL~BactS1aDiv;I;}!C<#1neLX@=r1
zwPmF>Da6T94OaG^7T}uMsVx_^{J`AwWDx6Gx@p|b3s=toFO!R9X1#%Ime&NM4^)({
zO=*{`zlM&{Q~YV0y~Og;C%36Ug@49Z(Z=MU`Wm5?+3_v|MmaFhT0L>$>4OTvqaE81
zv_29G?KInF=++SVrW~7(R_YW`aG9~&?uA!}4Th3MXdq)`vI{dJ!v+s=c^`$^Pc8hw
zC#;U`d{&!KUsuJMefFmUtM
zb0~VrnSYk(%0%0!OqB$iYJWnR-qN@gNKw}xyVx2>K(?rXSuz~0j5p%m5Ds!rnuQ3C
zLXOEKNx0OpFyqY5rQXYLRR4>_OqSWoL|mtEb^5=`FISfBom8r8A%&UkhV|_UXCcy-
zn3S6p{E39?7q}zpE*3g-yhAG8fhlpHpAm;9Um#3_?-5a(2J1u<0)J_NlmbIa&Vn5#
zHU^_~wn_owT%3Dx(+p9|)7pkC{lNFJBM2I)x9zK4(%B8F1F&0LhGMTN7E%iVr@~Xk
z(FOP?#lcXV^q#)*-4pcHqw2gKyw6yW)GeB)za3s~BKhOl^A1oDH=CC08v|H-A_f1+sfi1r#Td%ZcccFztGm%Y|@`-XcSHnKtnTLRR$;rJ9?!s
zt;Gwi!I{L(qr$8@M%vf92Auc~Dq?TC
zFSCQqQMY2=%6|Zt!yDjTepoLzcj4~-!f7xV`CCraNhyLAl!VIV3QdckcCp}7^0#I$LIf#%l!{WB6oF6GOu|mWE^Ywfm>pVy&`_@vpIjyZs
z5hHnR_O4-cR8qUBc~S_p&oL%TKM>3^#{p#_T`dk^(Wv4ab?E40YDOQ*F30(NCzZ?^
zCV`D8_!OZ&aDIr(#j-`R-7QQE4r>`&zWu&#BVqLssL8zCWq-VMQeXmDxv{n>(w)eg4<>7NWU
zbANwi=1{Wx&6^r}i9>96y^M`BEep)i1Z~uO=1^T53a#dXm+UXZ25?nI4zOd5}e4(!3b2MKpPnvA)
zgG0a3pfO!6^P@t@!{F?AK#0k=+c=*)@PEc+kP>|=2~UhQ+gWmqRwqD3OmtM$e>QNT
z6eB8v*d+N9d>GXlyrDh47!tCWp7D_&@p$9vDKe^G9s%gjbvV)fE4VhnZ;HG4Jt-g3
zJTA5R1``D`4VAzu{OOl1A!;CgDEZ<=L2FhErGRsArHJG{r-6A{wO7rTu|ZUC^?xd^
zXFNKOq+#?Xl`DtM;-gZVb5Jqu3qSQe=cxKc?AV;fX$V*9;8cposNJ~%oVjekbW0{D~}r$HJvi2Nu%E0
zUk1dRmM9bYe(Kx$Jk#!1oC7&Q~!?=O(K2euh?wCP+~9xN-SPdncZS_iauGi@)2cmP+^=5dS^_iaH!mMzxhR#Tz|^2*u=y~
zL8Ud>W#o?!O3@StZh8Hm@?DbSrhKv;Sdhjn+6Rf`2vRd(?SiTgE)B?Q3L|Ko9wrt)
zXG)_&;07b^ak{rft}E-LbE`!l823FmXuBn=D=fCDjl`{0uJX}?#*SiNyc>KS;wtDA
z=ycKhkLy&gHUmNA=sW3bQh!ADm3@*g924EEvWYei)NvLRvXXi5`;9@D2!qXoZ5C*_
zTh>DN^$Ot^x9FXh{xI_a)4uAq`Eg=(>xQ6Sk9STVc&-VQs2t}%-_6QQBpbpm-F{GBY83aJ+4Hx8_Jx+L
zoe3wenU41+M{qqm(toj8FMuI_{LNY7^gkLRT2_yWc4uAPBy!`x`0dK0S3amjYv%a8
zMy;E~XA9OT&ftCq--BUJE)mP6U}6{_eCZV0U@HP#*BO7Z?cyQ=xS@71)Ej@YAJVu_
zF6xAs1MFN*HpCv`#4B6kldapqQa>O${ilK`|9{e$*kSY>Gk;t1C_m1{Vl*g)p6B|&
zzwY&gT$=x?Iphvza^lBY)1gDw^E}xw+UvG5iJsBfXL?VKg(P>0m97*uaurIow>*q<
z1w#iONIbc*3VkuvCO}3|7PqSxN%LH1Q~OE+p%ZxZNP}f
zh=}gHbjd2gq<{1CEC9dgUc&H{s|M7L2`EJMa60H!xFd93@AK_SRG+@bj9J{L28mJH
znh;n7fPbx~S1ulWlyK~eGMZwUIb$UPzMoqt$z^BL!XH$D@w
zw|RxWp<$!?tX!TW_BZgh{YNB+w-2oU*`1R4@xWHBSP2!B43=N
zU+W-Zv;WA-d)6&ugA^4wdjd;vcIM4xPQ3jAYJWdNl;PpC=p>-kg(4?;&*|6W!|-yA
zVU(%P*TUqKgT>n6*%Wk0j4>_gBBW5?#iBj~>*}UJ&m-8K94`<0vpmRY_BqvwVJwrg
zoS^I`bIvo|3ga8XFT}7k+CE|*z&Cf+*@U{
zuYY2P$Tk@PJKN6sAebSH2OwUU8c=Hk$`}=lp?aDXNAcK!6cVEA1diwu53>W_tXW0>
zEgJYlaVqfl4O$_2A)T?aLT|m9rhU6IrE*AbSnkineE^%cw_7%~mVmKZYS?2sMTzOo
zrUPL4=`7%)pfJ`qToh@(8*gM^*;+P31%Ha#X~pndyRU5E_q*S!o!rjz9SXB~+D3w#
zC^FO|@q_}FOB>xQpiZ&RAxAArHHD7==Av^T&rw@@iunx&g6N+Pk`d}f2aK97)1CRz
zi$2*~w?lk{s@Yi@B9iUP(4G&eQkODlXOQ$WPD*6OzD}Ucm
zo9p^Xp!BSCMuJ~Z3obtr6{8$Ea28UUiKSH1kjvZ*kG@oJa{V~wy!>RmJyGeq3KwB7
z-pJ1N9f#Lsd5wz=^Q5yItVT~mJco2*yM$tgR^8W3@(T2k&Je}35kcy8wqX(!FMHar
zL@2d(Ks{~f9e$rLNJj!V{&x3&bALb)_ILn)^TBZXPJ9jSSNZpuiv3z-){r-QW3I5EPR(wn*`F5ljQYqG%YD>4H(iP=M*P_V
zJT>fiycdtbPPF(Qh)By_Q`c&)b%Lro!6fF-
z${z2@oZLbyXXBF^7qX+J;SA^BU}x*s{y6{Y{T2L}f=N`yvx0vgnDSs0q)k}V`Muh5
z7{VVB4$X$S23pY+5_+$j_3$vadePQjnRp@1rFR6FW|UZ4G%C0kDG
zbrFq5kfMqi5Ct}Y*Sg2gy@L@3!oii&E)l=_F38AJlgARD!x4sJ11$y2c8#!JdvMKX
z<8i88F}Td;m$z+~Hhwn3F0^#Wj@4`X4^HFQ?Qc-7V6Q#^w&f&uO@Bj~6Fzdax7O}#
zla^nR;;$Qe4nh$`EWxLsZW{8KY0Rx;g;b&_gKQFzu1>$aR5|Re0=5!ACDP@WjRooL
z@OzGLVW<_GPnSI|i+xrDcEITApDHfl*6}I=nHk=obLCPIbVAqy-%`%zEi*iZfaF_x
zBHn?`YY+ApY!l#9OMf?9-aMw=aY833Bmpf=ZwP>b)|KRe`*^r(sw1aG#=u2k}
z4~fCBiewi+5Mn+)d;gp->qTX}PxGf5sPf$T$M(iRl0N=e2wRum0Pesd8X7wAc-WpH
zXX$tAeBiGEh>$tB0{Pj#KAT0;m9AX4vn}2~up=j<|Bx%t-ijL7*UjUunAK8nZy|j
zLYFUJegC!jQd{Cm=Ng_K;lXQ7xO__9t$bRM=~lC|YU(uORbLLC)chtwTN^D5#RuhW
z@`pHWv40+YC<%x+Kuo(L{(obvFvHAo(YC0&{gynEH6KlMp54>`9((b!;@=a*woIG-
zQ33$M;6I>IU8crEjSsQLhhR+!x6t%BkW0Bk%j#wI;f_Q5a5{eTkDTl*u)q34Y`CkR
zfOPS;5SZ#sz1nSj5HkszwD=tC*v;}G*KM5@Mt?|)eg&(lMT}hMm!l_WUF;XP$gGIJ
z0V`AZ(TRal8TCWEW=KB~6~6Pvnm?NXszSYuczyE|WZz$(!dNZ5L%$dYYll2~}P;^(GZqvfG@*3L7ZTF9Al$7efA2
zgETj-zGkI6=SaloX*AY)qi2<0*3hN@R0@Z3lk@*>+2B?|6(sSjEIzpD>O@ZQrhjru
z@&YgCtX~kW#x&bnPQ;X9sK4DapD6L`Qxpfuy9g%oo*JN@4%zj
zmybz%0$sG}Bx5NWA%{MK^7gRgUQ6TReC*>Sk$hU2^rIu=PFrokvPBI-y*ZrLh*KBH
z47dQIm#dydwnNWwPZzzx`i*OiPo$*ct5hoDm2v&e!10`o0)@-Y2T?pCC4W1YQ<}8@
zHm01qff;LVt~|$^Mo=j%k1*_;BqJKg;uO*l#`Ee}vp>>h)Goc$GWU6@2xyzm^l9KQ
z^hb7wWwsf=|1eM%2kA_Em|LK6p#lg_IN{$L^vJ>^uACF!=r6KRX$z8i1GA;TGE6CE
zN1^Tz&RSn=dvZ%j63$
z!w>P?uABI8(%g3hLx6TT(GmVNSlmHnO-Xfe{ph6P_+6o*6ZlG>R_E1FTs2~4e97#x
zA}|COC?gpoE2)G8-7?W6l-1I?!R>*s)p}D{{+c=jmOJc^W>{eYY=8gmY(pSOEi4R%
z!vdp$aUts*SYLwa)U2CSBrzvd#J59({RvCEA|=XKLDG1&qKb!*1`-4TUrvaVh?)MCo)?}
znDJKiO;Oiu^(Cm8hJTb==~%MB1-v|U^q=qJCyxmI(mV0g$V4BI<$fyM7+i!-pMo1u
zenOJJuSzDiZ8s?{4%ar$O3M<^+#DH@HMt9771F=|RXS6j1;5i0TbfdnO2lp^j@O`Z
zi3>KpF%23RObAI6m5m}-UWNQ9P8*_10~2xz&N7FEiY0PWr+)^LB6bK5Nk)^lTrdSk
zlejMQ8PXtN3}-&K;-7RPA~RH-JcLIB-VU7n2pR7_jtvhM`_kU3jU=l2ImQm*Q=qp-
zVEs?vt@AHi{K2?vIL}{>6;$N55Q`hnwa4k_4~_ZcuP3XKf`KnlZLf>@@b}q4S7AS=
z{LID_ZhJ^Nr_B?o9#pD{_-w$@GvA`7ij6NoR-%G{ICnGcO$yEwg+>C;rRw6-5iLXi
zBE)huDZJm;Wt%6=-E`{R*lhXVSDQU@G@N8NAh?psxPRgPyf60b%Di?z_sNxWiwG++
z@^dx8M(A>km+>LqA|-u34DO`%A%zV094T({($V0M1uKu9DYX3Ni#DCwP_JM5)ZPUK
z+*4EY(Z)Jm($4@)HF_0j(HT$GLn%m%K12XAaE1qMp<45Xo&rV^F=8oRTpNvBAC3yb
zA1h{YQlLz@3n1VG~qkv-%=+m({+ZkgA$AN7skl9FeH76G!z_N
zflJQ%AAwk-UP
z3M-VGK>;cVt3aV=hV~>{kbZM;>hvZGX^Qtb$GnF|R)f{PJKhbGr?VVjR3tzly{8{mqb
zy=PCSP^3^`GyGJN5+lgZDgAwBpFf?=Y1(1LBIYsi3coobos!H%jFx4F73g)q?zG`U
z49+yufsh8#|1=WS1e7v_!DWkY^CZCkIDgXXDyACKx7(i@?O)TLVvq`*>cfAe`*Y^F*hU%uy(VcsIGhs3Dbd^iq2I~ji@~|8vPS*XF-46F*1mE0SO)BJ!(fS
z%`k+pCghv2yoII(w_I!Uv~Sj@Ju$BmDt$u`0SuA?F2QEV>+B_AY`}Kha(TEvp?@XT
z-9j}$xoj(|SOp36xf^UQ9X)`VOF>65@Tx==ipV-jFJFswg^JU)?hLLOKCtXcH(e2h
zD|P%b8KZZ8eLw=>
z+J45UwJ-dyi3h|zHr(b0*w~-0LSQp+jPR`~`FxPhYq$r7MF#kl7sG?wOPwpXylM0D
zPOv+)4d?j#`|Q82Nb8a*u%t?N=tWVv@3V(|Aem0WF#td@VXTVl7_geh7=JY5dz0Rg
z_~yO%XH8P^RF`??^tI-671KS+YH-?@M%A3kK1q&da(Ip6_8#{DZHew~k0{mXc4*aUvbvwt25Q+Q)f=w`%qWZs
z>i|@x3TCJ9<_6L2b$97V`_FI8a{d)6>6-;%rFyc_Xl~w`G2NDIS156fQUC
ztfp$Rc{h%Ckt4)ob#&UHX%@K5ZNns$)g*m~dojgW>tHBv$ji47l_KSTR7eRe?8}?5
zM1|00QaE5q)tcieN?ym*4^yl%x3hm9M{74M!01cLh9}hDD}My*KWwz{sHY0JkQp=X
zp>a9q_~7jjF@rNWE~G}~x*wuyC7CEztG#iI)!EGJy?yhWNiNa=%8|c%6QiZRH2=o>
zMLGZWj^uc=zl#6aThh*!63j3#;FXcsf+$D9BRxA3*q@ALSYj<_fKb9+Lm2sD049_B
zXZrdckKpcVj(>sXqr3URQcM<4FH>o4auk>v26($6IJf@=RczgVuYcGBIQ^b!tYFYc
zhS3H$Lf0ef90s{a>vG!OG1>z8=m~HF4~$Hml4IQHa8HHx4I^k;eV@_CJZ^J&yQ&VMTszh)hKkh
z!xKzd`smw^`fz(g)9a_%7(85PdIW22qRJ%TAfWx+z)6jCSxV!quyhs=TJh}YdP)J+Y%1;$C{PFuL%vlJBi5S)@Vr`1v+G4O7j
z%pj}Y5r6xf6^oLgiZj;Yog>@a#0hzy8ChoPTa-Uy7Z;~-r=}X4)|O;MCZjWHR&@bDq<&9J`pAu)l(%JV2c3Z-2}v>rq@W2N_uVJ=V^mTROKXc~QQz
z3%)>dGZQ9|Rx@-)x`;JZW(ok6_a_w+fKzFll_%)&x)K}xEoQ(=#ROVYd^U9i_Xf)l
zJ0~J>!?R7fEj6TRsvVW#zuuz)IVyw<10h}()4;4ByE254Gt1@~4Gf$@PC_JdT%Bgb
zw0{D=$`|$x01l!VnFUbFl%K1x20d&T+@&dKq+>ExncNVUCf%=_Qrka1({%$hHHHN0
zA@?T@!&)`~D5#c-WzFd3WbYybidy~gIkZpYD8q|Z`Y$pt3gSgbk6L}RAaCtEDXo&d
zDF{N6Z^c#kxgf#pIwI2p>G~S9bYV@JLVpy^4!Ghpif}6!@+7R(S0-=8eZu_8zlI9Q
zKpq`h-j@u7MNYW4bzJr$Q)G-Gw?sr3O0V;X9!z&Q8ssgcjaG)+%2WQS8p{%{d(T9`
z_Iz5UmO>wtaIn@{zdPcCINFWyr<&EqEC<2DL#
zbyZiOD{5Pd`o^&;@Fp(xSnpB-jON6atTG!@$WaVJ!qhRJP^NCZRiaMOr(TcI{GzOO
zB5Pgu5G2N3eE^RBmP3^?`~m!Yh}!`C3J#)H
z`%IB-LvQ&?^HN0$iFksS9UeiIzDGnwvy3`fCpc9#bbh0ubGU`?mVBdUJ=8R?9*F_S
zoEXgj>%N7N!9S@APQ9`{&wnKJmnFvq|A+Vod=SKsoJfE86*#|5CZPw6th
zj}J?bw2f;Kv;EGOEhaICKNaNalC$0~t_`OFA*@4}M289yPIxPNoPOt$~73TN(_AY2>_6nJ-}XZw8)5r}R_q
zSmnG;-VSxN(&%Se7|=k9F;J!Nxzr)g)kj)&TIFIScPz^z>)gBHC1cK|N
zf?1%P_G7NM?(!7JxJVRkANi@h9!o9#gIb(M&1<*5T(G^x-^i)s2g?}G5p5N&T@PMi
z5(!*va>yw;8Db{PP%z0Cv%U1DS#p!YHk$kV%
zWPk~Vht<&CYR?ZJe*-5@%evrgxyHWqA~FP0fN#M@gO9P?22f&`!_GAA0}tZR
zV)$h$pz7WKQe;*RYuBfAWLeiVVqrrQC!Q5WY;wuf(0>I~Sn+y(Gy3;EoN&+wb?q9?
zx71!CdCeVg=LqCOyfPEg<+x^1(A+@Ev|IqF2p@^WhQ#;UIH*+)90NhI<>>S-28Wn!
z)Au5^?_1n;)|uTqHzh{8%0bQ=GjI2&*_V)Y|t8$y|pikqtg3hw@{C
zMc3|rO@Fy9@+j@B2!Y)#BY{OItFXJZruKomd_t=h^B~_ROR)jVju>phl-Y#Z?VSKV
zW$_`Y=Lp*MkibpCaaup8r=6fa1PG>}hLvW{m{V$epziZ!RZ1pr7@bsoF^(vXqKby7ivE=MCEC61mY?TK{`k&m3h|UUHUWdGC>Z;EOHoYBHt?
z6MyJU7R$Lws|`du8{wS-O7i9ul>P%dde>i&5Yt&pQX@`cfWP*BBMoJ++od8MhAgcI
zo*?Bo?(p9j7sciGD)3W|L;Xz4xaF4G?!-j7A>aFBz~N7CoT-oMKI$zH|6z|N;Rfwm
z36X3#)hzL-NTMz6#P|7<_+Lj5pT%5{H-9(IIpLr7Ma#{_ZsYLmo6|Ht3ubFdD2tZ8
z?SB_!BWCWnkw8;vFkNRx6&)GM1MA}P?&Too{EvBll?%&{y(PN7?!o8|5Okpt^#^&i
z(!{`d6~XT0{Tw?y#=>_KPAwl6+qT{wT$byx4(T>?-JnHd6sG_%wR4mOZO5mc2YBBN_g-}vSJop!zq`uKH%gKOmX#JBntR}m#_87vZmH?u-+@b^
zP`dbVx%&>6xTw9Im{=J*!Neh@shr?K!4@Fb-@`7arQyzifnvA|hs(nXlIVdGH!V;)
zq$PL$pHzNBAVE%#+H&Sn1e?~a;(xnrO_~)J#GwoxZzpdJPwHZ57wXuaF@>2jW(_5F
zXI`U=Z;9n0Ox0!B?lSkwmM6xAhC!YBBl&{{5sN3akwz{OFTMvjG-aO_M
z!c)0p$)EAM0X=XEpPg`goir9Ez05SLYFvTz6-kRA*C}_A2tqf;I`y@R9)Ak-3y;V1
z)89FJC}V)Yu9n=;xR#8h-*S7Gx~^i`_;$N3bp@UrvP*>Wy>D5$$9z<;FxPV`$CtaF
z^-ayGS7g67JE$X#VS_ioN$q
z;b|f4J3m@DM5|{ptfM=ph<_mvy;P>);;|)y83HJLZkYV~l;AnAw(!Xo2~f9o5kz`F#_7qlfDB}WY?u7R2M+#bklQ(I-$Ogda>KG
z)nd~usks3R(o1}7Z>m1{#D*z}jkd`0vAix51zfJyiH^QJ-)M}zQ2zdWHTQc7sA&C`
zN=nlw4#OUizy&`U(tqO1T<9tF3E8)mjyMh|wTcgEvtGV4eDJrRjwE+ot3u|*CJP&F
zO^Gw>H@;$!$GkH+_iw{O7#EyS#}YE``St*kIV-9L&U$p|JlC*_8Ay9oC`Jba6>Oq3
z5N-Q*akes#V|3qm?w`S4uCn>HG{KQzZ28R@h
z{vhPv7%Jbjrbx!geT{U22eF{IdIL$js#OJ~6_Pgq#}rH4fC^=9WOHh?wvG-2}3FYd8hC4W7E@KD_2wXys$zzfE$Gb
z*g_p42p9kb1AiDn?*WVrRSXOPY6g17cMW+2v4IUeTwPJ<|8P+2PiA(8mR!F
zCj0<3V?(3MUn3|I%in3zqoN?
znE`%N!@9!ID3@Oe0NjokjH?t7=;h@l=-}aw5k#RK1b7`-R$b{Y4N8Q1U>dFLUVpSB3sN
z&3~yYqJOZJSt9%bAYT7IF$mJb-S>~){IhI36w)2;j&c7hA`}3FBcPY;-7n7!j{GIl
zQP9;;RWURY(8kUlQa}fV?G9297
zwoe(3?Gpxt_5uEn0e42CypVqXOACV|?O~UvV1Mu73N%K--8`Tg%K!1fib#LU9H1Bg
z2mo~hK)vl8ftRU&P0pnld@07ZB*4!VVZJ$LR_HS|94~mZ4GgOBYgf3`+wP)LN9r9>!Q#u5X8T9aCcR>
zH`M+v9AoGBSCjve{}oCB>41O&1i&JKAYn0~zcj{|w+(?EYwT)*UoN*xz1Y8W*on1s
zMnc`)0V1GZDk!!y|6-3lj!W(Upt_#2l7F@t-~Skf=>QNC6$3!fXowFf
z2s<>6hr?jMYrT|1(Qwos)q%0~|4|7TTgh)$BG|^cBOvaM
ze;8n6VE%yE+B|3PwB6}!Kj=u&kDVh^RVVYDrXK#TV3jYVB$k$-aU>`I>huQNDI0n+9FwQ$@ll6$cTl0$RhbG!Z+?d=u
zr_tPtsh}0S7)Q&;7cDXrd~e(YQnXgckP=-7IU3UfJsL8kv^coi%x?wf8}G*KefUoN
z0(Sj@X|&Sw6R^)7t|v7P8#5>6C%a|csoqZF#Rk$UBmKlZV8qSjbAJi18R;%n(~0>8
zfT!>3M8G}X84i@D!N;JNtkH5eKTbV8`x@uNT=wmlakjQxQNq!Om#>S=d;F(8A$!NY
z{__(zsg$cQQS#B4$6YFUvN1%R7x!UvqvLkfCP7@3HJ)ECZh7vUGL7M>F47K)=QG07
zmp6zB)K7b~%*RxO!hgq{JGef`=aqjrWS5k5+U!gkR&76Gr)CXii8xQq6yYP#T+NnQ
zd|os>`osBV{UA6a#$`x6KaNgCYN4zMX5ZKFr1pg%g3`Q%)0;rie8WJBQ9W5o0MG23
zUV@IO+yM8QDP|27{`G92KHFSDE_c*QaiJ1sn2NdfhGbWUxqtL5L71zm`@5gD?Y+L*
zXznOp!1Y1pRblJgfrubdIm8gC5W0|?Dv83k3=9poHn$R2_c*ua(=hQx#)L8HkJafp
z?M2jG9L3_5!S;g4=Z1y=O!G(r*U^(({VvYU&@4#M;E06|1xs-uQg=
zDAq+A_crfRxPLsu`I6c>UJu-vjg7;j6mD1}&W^H*@A)F}2mx!_x*-kOHPP-FBkNbc
z2gB4VCQPCo`^+4X<7QkZM?FI9rhJZpxIQfM!%S>s=`>qoPC@Y*Nin;CW3c8saMC%_
zUyrq|Nn7VfamUtjdzrxtt=WrdhHFK&y&+m=HD*Uoz<+qRMCKb0J7P_e<}E8n1$eC=
zJf(|Z+^}JjU3|tGHz3cugD+1UT;+)JopK)=@5UA2}ofHa0R1K(k)9o`x
zKG@hNA1t}WGn1_bFY0>wHR??i1($kYZt$>mKyB3QVNW2&@ZwpijZO_~2&xcwnEZWe
zQfulg=zps2U?O`+_}7Z#lIJp!6abaf1RI*(PfZ#mf+J7MZ&c{PY?Vujf!B%N7QCj;
z`!e!~O-;Qdp3d$bdn1AEhy;q&poDG
z0)Iwz6aDJ0PeOaXM-%uIxQo)|*d;CMs1*U{h3MvSa0#e;lW}HeSe+x!Nz$@~gbW_#
zR^rPlQ+p4*kmgZzujAiT7h5-l;1kh4tx!dIz7~spoCpqme=Sc@^h5o$oG5YwCCQw|)mrA-uaa?$x_>en9Fg*J`H#7M`_cBWTRq{yFn({kJ(??O
ztFaIoToFDY?nUeUW4VsRCGKl|^<(2jEEIX~a(3g_EueXcsY{t|nlx+n)Fsn~!uWL~&3638=Hd%Hu?j0L#er1T3&nz36|A`}J#ZMt0
z&~>-@7I>Vl-4=yZq*I#mGJ&=c3dlas?iJgv##whDSd1Q=&6w8|(odRg0e@a)q;KcY
z>VVutu3m3dxIwMF1meG^?Fq*z{+
zNaXzE8NDtQ-o;f(gZ6O`G=K2Lky038d+nY3Rt5FXeh!{R1VZ$E@Z*@4!O!?OhL2+c
zuN4fLMhNm$Sy^mCD(UxMUk?s#0R5?<6O>M!C{f
z^6oc6X&lFT`>k08rhlRoo0hYkL64OBiI?v#-u1bf9AeJw6y2TNKRh3UoNfqekU#hu
z)_lyFwA}mDL{og;xOcmrrnl>2TSZpkx(dNoY3^;sP(v|&#>YIt`TTrt4nuqVlaEGV
zJOzmgdh)Ajl-vr@{5;e_Uv_Te`%%+Nm+Yy_v*?!Q`b8So1%G@Dwq4={eNNasq}bko
z9h0Gh+Dl{SKgX`hz*I#G7f8O~kgI7uZIs44!;PpNoW7HC>?qVG3?!T}`z*t**QB>!
zU-pjFrT{A5>1i=8|FWJ}+F}bg_=dZ=;&t%3ou?Nb>d~%^57(k4{`QaX*d)z(5(V
ziVd5E>J)6>f|4j#P8CPq8oa|31@C7VU4aktR8ZA-+<(K}v1fOE4s2oB%}t8bxbtZO
zZ?$*hDKR4&Qu4Oy=KmdEzbj37%2PdVq;MzzA}-4KjM
zHBnrSlETg~4U?jMl^^BW*TStenxDn@L_hmyznpmTEh{R(pvs)U(D*9S*9l-BvnMvp
z=W)8wIoyAfdJf>tQOk3kkiJlbGtrXl!Yz2C0Dq{L)3?0VnYBIDQfhDJtqiso9Gtk)
z5Qh#Jpdp%FHj2li$vq9yDKES=|DCmagRM_`r+J}cJeG5UsWWBFsryyGqe(xJWOlw~
zcYK>59mR^3<4}(Sex;s*MVGyD5Zz7Untc!}zc{Bjg-ftb-MqzdG5QhV$B_Z<0;Q@7
zX@8K0-=nq}P&Y+Ll~jk=s4!t;fe-aED?DYUEbq2HM
z$zU?V*~72a3=c9AZe|_Y4stU#I10FDG
zt%TCcdkQfrnyL5{JK1W^$&QUR@hl$$U0KIs
zzulWO|D_W1rDIzjm@6pY!9q(Jd$=&;n1OF@CbDVy%gYgy@y;E@lHhX4`@zYKx$t98
zZbWI88p*t=k;6r)m#~|@Ltm3Os(PQ-(LwX^gTmx*@@eVib0cwEw3vHy_J(h+fPYLT
zAsGf321L$u--`Aykf5!shmnf_U$$;zn4#n-5h6=|@&b}3yd+X3ADMcWb*mtYS`>^)
zzmxmHg^6t9RgUQzkifGuVk-LJCY5KyBwEW1ZJmpFZ+6G3K9UJC@+#X;;<~%<$PZ)K
zW*F-#59IU#leJk+9Cu7Ath#n0#DAC@9uyKD|Gm^{*lJz-FwUzLewxRkXr_)U5Nxjg
zB1+-?r
zVeAlKEnZq|H$Z4xPV6y_TeULq*~9lHTS%W*8W#*_`LF*rD6uUk{8UvFul2bLVO`>v
z2>mVmdJdfISYD#?nm3D+uCN8yKU+k;#Cj3nZlE+
zyb3a~pXU-sWD-SLf%^#PvBugnj-pt_U{_dZ)zkTt`T5J_MD?#f`U}ouBn5H~3Umsl
z%STtO)>SstZwVh@L}Oc-gW
zX$Y?**=E2z0*#I#c|k79J6eQsXXG3?f&gBMF;BhKDs@Wv*@ym75!}ACQmkEIU_SkH
zq@;x;VjPDAyN{NV1;>-vR_b`BJtZE?w{w+`I^*jwho=xoG`k%H{5T0*?
zry%vP&%@Pdii|rh8y{2gc^wV4nZ)QXM#fA(kSO4QQG7GmTC|7E@{dyqDcucr4UZgl
zO?OTyZF$zWu?KI^n&&pE*?MbJoWa)}t?~ZALdQ6w{I+P~0cgTpR@cL4C}i47gzg|d
z`|tk{Xu|jNNJosfe+Q3-OiR_M++}Nv(~`pi@2nLglboRoOa)F3TFCxz1zO6Kozowg
z)JK59GXHpMBx7H*AKU)stg+DL>70P#=4riRNPE&T>q&M$W<^HJ{s#$Dl
zDN@dxt#}BZSnIh=&*#7Q`RxF7M%KspJ(HF6$2Ql;D5!@Ju(`esF
zvw}i=X_CR$!mo0;ddK%;Oy`eN!ZcO#Ft*~a`9*Wq3g&13tD;QTE?`x=ppdUfxLH|W
zV1n6}@tb(t{yFtuYq1B~Yl;oIvMF%hzV0!wO(^qB;A2`aY#fCCi(;Tp4+*TPFH7EY
ze*85#2zrB#U%hp$S(o%0sHq*Q+NQ+^BQCK*+YgjFEuvc+)>Fs(jZr@?+en?jBez}p
z%d5%fs~BO#ZqzOtJ@;3~{_swZhfqTStx)0U{-73G37HVk_
zejZry7-c0bc4SXR_N!Cxuc>B>jfXyp<=F+-;}WaVE%4=3{y3-|8P45U=FDysHQT1H
zMU6EUt`tp1kGzF)9qn%9w2Hrtxqh((KdIx_l&2Jeqcw4*d{a@1rqd9fd(*l5NGezcqOYp6~
zCi+{XaO+QW9|#oWgYk)Ht$O-M>R
zA%le&Ic;ocNzFl)<{ZGDqp%U@O@^Yed3OCj)yTrYU_skTzKPy|B3HHcS~zfmTW!@G
z3053a#XqtZz1H9(bYn4Q%eVa9L6R8Can{}Yy)lH|;v{?=*E^IP0{=38av3hB?pARK
z`9+Y$X45W?D=wq7w!02{JkgqUI6Iui=Ftp&gb!|9`6Of^_<#ei!(&uht*j-dFUMOC&H*&MQM$)hjj>~eqixf`g0lh$3;mvZy=vb}HuZg=-1ndgz<9Xad!i({L=
z1#vK`(z}87t;kmxbln*qG;4cRVR}Z^ej|s4jw^Slw?p;Xf`Wg28EVE66mYM1IQPuq
zHQsNDRsQ^vvWWfscHn{?l}-N8<%0+QTGoD&q3kl}CVf00%qgmLK+?^gwssACLrheT
zdGbX*0}U$SLeX}4%NT*O3+pomC7wxmAJzySQ4!#TQL$-1#!w=m`>L@h+tkBEVg0_W
zk@iX>&A)vXeFnq9?-N=ZHe1n7^KH*C%|h>i`&S<3hglWp+4vbmFd>w*1pDmGT6=VD
zXGpdBdByl-?T{+?XeH@=^xZ>dQIbAQ
zNQHpRZKcPiP}&HGjOy$$JyW_@q<^dR%-wKpoZ&vk+pr{t9dz
zrt*;zUEy^kUaJ&8@I@Mv)|RM^)E
zpL#-CvMB_ZtLK~s_UC0sQ0l^!f>*V7tf$95Jy`GpM8=-Xa
z{(-@Adw#{pWFOm2xowwaPxUjEKviT>#ayB2Fk{@Z@ct8PL)_0h-w1xKb7e0kOYKTa
zesc|atLC77gP{jWvQx;E3l|{Ys!qjmFT5Rn+=BU8Ye42oYI{+#=IBBfjbp3i5uM{^
zjP+Llwc0lFprM(aY0WXv;PbRFL^`+lkfFr3B^025YqNM$P^p{2
zLwh$A$ym@Me5WhhKMyu-g=N3N39YQ0L8;qy++^KK)nP&|c9VfLuMz89Umjk>+6jyn
z!X6K6P+3Z$`K0hz8>9T;o%`~1()PI6DQWcR8`$mD+8H-obwgcpm*_<*IOb}5VKXH)I&=PhPx_LIuxM5;UqUKpFnRY1zM);tQwaz~6xV{?
z0n>rmX3iR0^cw4U0~5gCDi$PzLkm!=iub#C3VVIf6Anc6KnAka159q6bND)4JMM+g
zqJuxL>O=(Aq|8E>B{J^!Qs_1{FNER^B0I0P7+7AWJTKuQ`V^Kd{QsF&2UmlA{I{mu
zn;NCrw0qOO7Qzwu+aQ?JRdvDC7q2+m;qyFja|9NSE)pa>@3fBAigkM2!_X}_5%{h2{
zHX^N0(y_n)Xs%emfzKtqi~v3~Z8H9RoJvB`zYGUYoW|V9s@bc!C{6vFNGRdY`2P=;
z;P}6YL<;dZIAn|PBq%ltCD;EAEm(+Ixw%;XA3wpy#>1R4*9wja$;QFK%AHW>P6af!
znyTn*G;@S^c+K`xw03lK@M35wv_k*g?F;A#27yJBg1NdPgt6<}x>SAYem1#w?&+zS
z%RL$J%rTB`JF*g&DXShLHAHNN7!Sze*WlJzg9II?O*f!|H%Iz|C7WACHQJBZJ2pN&
z9VH?|cL>#ZQwz;2qwNj?3F=Mw2L#}tLzZGB56IF53C1FZc7bXa&RhnO(f5O!9T3@A
zfgA-jsC|Ji4cy|_pl8@5vCD0!>SBvUjVFctbL-mB(sci#8yFFM=vK6|7GDV>4o@dp
z+tN|M-`l^2n>$+G52pa>B^vnz0#(b_i)LXCS`X@H3$iZc;}@MAS&StlrwqtS>PEz2
zDrrnC3_)Hz32GHKHo6|9g3hCrud{;WGO`KDNy@xG>)YnCiS;nB38~`ke@}tKeq>dX
z7BJ-5<}wcSJ$e1dhx=jAPHla0T>_Q`fz;q1aZtaS9Km@%#X$^e?{6GXO^olKp3Isa
zU7igfJJ#yk{l4Z)Y|XEcZUCqb9x&n@J0X;tnql{!+M?^HhR`1!z1O;+)EnBMSzG=+
z2_}62Zx(S?5V`OL;WsLjWzfiVdT!4E=#|iJkHrQ3=J&X&nz}j^x8^FZ7f2REW)-nN
zAr@iw707*$O{{({$L9{1KM0LHSIZ7Q?f1FUJMz@GbP9y1ni;|!02uFm?(mxJU7TM1
zFFWD2re8@RG#m4y5C?{&T1Aj|-F+
z@B6nPKIhZ)GG~t$V1IeqU;NMJ`V3Yj_AAFB6!tJ?3hM;a02)LWREUU
z8?k3AM796W?PUbPgOBCCiRwZ`x%1z@w%db*aBuYh+j;q3xU~=J?}M&wZn^V|{CUut
z`=P4aw+4lP^0I-Tc-K??jRO|a6HQBYkf{}zCEN{Sor?$uWYDgHv!Wq)UHsy<-FLwGPe=H{ynWC&8$Yk$ZE%Syz{Z
za(Y4_dxy^up8N_yboltUL2P>MsC&2)Q=oe(Z#N#e_91J3eF=MmaFz*%67^DkLc@Zr
zntTPmtb*Fq11MB}vR}yiV6`+K!kZwRLxM2<)eiu)@M_XGF-*re5rm=oXCU!~)wl2m
z%Ad#oO<(^Do4)=FpZ*J1UjGZ*0hA!!Yvh0%oDXPNA#W@IMJ(-*Xb@D`E>;uD)6b8S
z5V+9!uKV_({rcyZ*!+gP?~l;{7D%jo--Bg$wSz@fA{c~!5&Nk14K70M56}t$$)F$D
z;LtPIAXL`yZXJRf&l?A4mY{Dj!SJLgtW?DCC4fpu|`yzviNBBy?tN)QAXeY&j4L@r7SR;FM^g?vr
z(~=6`e+cICe*X3$PkpV=>2;G6br8Zn0twNE`|Th?UkyB6aNR+)e?Ka`%1>U4RA(C?
z@AmNUa1(cZN*?nNM|~)#9dm*GXD40Yi*<1G$`LdoPTvy##>YWIf>{8F0(L|X4UeHd
zMh=8M_xBLLB*6+pj&p?nI($8L@7Y2Ac{evE={S1)@BE$5f@Vlci^oJL{b%pUFl!h+
zg6SklAAWWK-OKZW%I7oQkiI&_d(O!Dcc;~7a1-BVZ?q7!O9?aDn}tM5Uf((_FqZXkI&jO`fl_G
zgfIFZ?@8bw5N{w=3ci#6RYCm(^F0#22#GY=rG6!t2K<$zC`;Q%9T^u&5J#Z-YrN$@*?ldmNV+XiO93>g01`5m%$kXE^z{F{
z-#T*=L#-vL`X)F2UbG!FWnuWaY@JLLv-7^#GdxcLKG_ncI&%*VDBB$_A^-5Q@nA;G
zK_gm|Obqie*Z&XKL)LP*7yjDvc5l#xxo;hb?NrPc&-a2bP)e%<{
z;Ge;CutA~r*JbifS^vuD_RJGlLa=2We6Tl=*=7uNzlP>|!gra+gOZzJ6H+g4Roujw
zvFUqkSMx7@vbwPTknq?RYPWqcjK)ZizRL8OorYu6!3SPlCW}K!$x=fI`Cu4c#clt)
zTT@sQ7+BiDp_9VIsLQd>m1km4>h3(Kyh5y6ztBkBILE2GB67iWjRC2XVlqV24r;OYNq>X
zK}r_vbA0J(@?N$E?+%8G3Pc4$Dej?sb>o8>j*p_3FQW$DWX9%O0u04gfp=E6l~*yT
zBc}%w;=#usnIkHjAldIQZexQ%j+mwDLJ-y*KYKHUga`^0-
zoFAk^hhu6k6FNlg9HBlf@`E?xr;}KNhlw`0Ykv{qOe5hdHmof5TK#fFl2VoxVXV<0
z4u7P&-6+J_njwk&jWSSu^G)|#&-}#(t6lLmNfCYWO}7wpL?L*+L>l6Co^2D=JoofbN|{ebxz(tC{)OpnzQI#5!s^p+TZgM
zjz7+9XZRc72@3tbD71|A45b$|ctM%x4fAF+nZ_q9Klz}pPE8fo&lRULmzixxodABg
zpFgXSygEl4JTdcJJ$qB^Hy7#Y7YPUF><7eysLM!U9?{UG4JRJz6zqMtVaqp$wK=;+
z7D=z$t?@!ThltQqC9}>c4OWlMsO@LM)##Qgt5RNP&vM!^+RBU`%P>EaWtb~gMU8`t
z*RzGk8Mbcf^I^giRTepRL_5VD$XvB)JZKu}Gg4BDuQbOSAaNa8z)8p0=k6jV$^tB?
zs@1;@U1?k2eomX$xC@l5)=B(`qYr#?q2KD)u8(J($H1FO;yEBWHr37){{SByyzFMR}~+
z+-Kh=IrbZrB0GlQD~*cKrgEz7;dZ2$MT>7a#>SFl{yNeMe%wWZCS1xhry1=Q3;_=r
zYD@ATq)E$M(?c^jmYsY4O0GO7ZWABa_%Qk$kL8a*`q^-%Gx~eZO|_i{>&(i77
z`?OUTc7X{t4Wa92>DAq>=Gnq*Vjyb(T|Kh2d*YjDGsR
zx#>fusA^}8KJyZ^&Hk2tm`rB6loRVb#QDdi>W6nZiR)^(O%7JDCX45PIArCY_qR`J
z;Ow!sBHeAQR13;Mg97}+o20;TBYjMFWk&Te*Spcsk5$QYE4SR#x|D4Zrm;fjThv=v
zxGtH9A@pmR@-DYWEj*cGNnpYRVv`y(#WPC)*90)uK60CD9L`i?O}YLn*4usq!m2<4
zYS@(Zy@-T3x{oM&BYb#+dAb}!GRO?Em*;?AkSPJwHv*=#NaAhv&N;!Jk8nC{V&b+~Mwd%zq2pWtnLcOJ0|-3LCm`DBpee@JnSe`JDs*
z*kJu%bO#F2STGM)%p5R$b>+-S@qF}4yTsTM@eZ#=%--$tm!rVuwoYGj)T-BA6H3c^
zHNSOxO!u}E8{S3-pErvJa{tWJFsSDCZPP?e(-dP*$za1(pUDj9X-I^EVcF?#>;E7@
zfp++oqaFs&laji^lS$Nu#8?DNOB6;&n7Y)%%XB|VGd-2`94cU$SVNKn_u1rVN)!EU
z-*Ar$65JFMXt0d6*BsbIdBT4^!d@X2-7S=q}-LDf=z0goV{&Q^(Kj@LtM9d$IOg_hSiU9~wOed+1@>NAWea>P1z9+O
zw;iUsr-UIFzbc)OM5*;C_Z`7vzP2+jr_wKP;JWkAZUWk#>PyK(>m2XUSu5;tiGmUT
z&C({_DPCPzLJ%y1;eZ2Y#rM_2iX~5*Cr4}bmA#}AT1C_8jMK`oa_MD83~v6mNX8<+
ziJmN+wY518{E8oG#xs193FJd=4)7vmeG|~irvketd*a6Jc$lLS*4Ap*(-DPuZ*Z2<
z?jz)5$N@alG4}=w#-BSTioD#9FhmG~56WxziNaZZ-}q7Hgh3hNQawwQB>A_6H^
z+t7N50b9Yl@?E
z^eyy|0^xg-+sN|bh#Zv5Rl4Q}>+{V*C}247AM8|xFxuJqph`k{-oZbyi1(|zM+y7b
zPiIQvn+|ZgP})yoB{yCsmpsQk$>*Btu^R&7leMt0)oQ`%nHHGhH58;B2HWhVPhA>0Iu{hxUpRp7{S
zb=jF%se;smBmIK)_PwMgXGvExcCgLA{x*0yioW0IQm*jJa_5Bme%BqF#d6bVX3`tV
zR~3n1F}3S3PUT$kqb;y{PO1Ml(;#gVNMMvTCdY82*najZ%u)xba%J50xHjS0coI)<
zUyYhnO0YE)Mj^cU^zMV%uprE*DKPOFFPd@yaC!h)-duWU>e9j*z$pB`UrOEhUBbyX
zViv0{sUUQJ>cU^yq^6c1U<$7An7`gGccEv9-uvyE!`o{hvR5gq2zMMv1M$IB<3{O|
zKL>qp!b~A)7zCzqo0pJW0Mx3-qPBkj-*ct(x-S#132+i=Z
z$vuy$wJ8z5O_RAE1hrGj;n}<~XN(icew*6f5qUJ^5E}lm3^L7{jXxB-ilW%U42@hq0ZcT%%*DLp5YI(i}^y_fl0R|0tu46
z@NRLqwkLu{XQ=BOAsVa_gYduZD5x!_&b#)26D;nH4-U-boV2SwI!
z%@P{YWNNJmOn9~qx5h7r_=-^GIt^>B!`q4P4Yf?qw*|voby~)$D<+m)lkeYzV3fb4
zpADYnP08FjS;o}g=vm9w90R&Ti7a}&;pyBNGm_0Va45B`df%>P9|Si9B_R^2FSiO)typg^=I
zhpUfg6?6O4SzLbibF;}=ElP0IlNr*DKyCGUC5ziVhli7>hs3a@;8=K?M1>9oQ(a;K
zLeI@*=Wm&!GhkLkaih&pbvuK$8PoG8I0jGl8TW-^q=#XD^^S!xOR@sr%K!(KHUm`O
zP)-Ap$0PGC-2uuczo=zC2B+(lvlc+Yo~Q`>!^9|{7{~>wX?xZs1+p;GA2=<}kbu;{
zwxA1VTjo?F=UCj2nKtrp^N;c+8bl)?3Hq&459?SM2~dv}R)Rh8uD};)(zMhPEk$nt
zxmXCYV2p^2>Ob`;ktgLFqVJD;@)V0PoTi>Np6jc4D4jg$
z8lwhn190UQcWm1C=G_qG>u*F;L*TkhyRI=j8M5BQK6;YP_Ld3j<%I9~-?Oq^$FbYf
z2Nq0htK=R*+KrmR^c`Rj&C&EtOs1=cG*{RkIhUB*uvdEzWZn_yUQEAJY#W!O{=ucw
z!1k^$*LY5KPvUv8HAf(L95TQ8gmvhMp2E=P1FgaZ0e&+
zTltA7u^Go!9Sk3+bMvd=UOfXet{&YmTZy4d`BgFJnZ4aaqm*`j+u@J?OA%5+gVZ+p
z5>OtLRn@)H;HY95C=q#&S(43CmLlJ{invVmVDG*NW8g*FLZ!Rkg{1bVxe4&~fcnT9
z%j-9z647=zbj3ewu(z2I3y4r
z>Qt^M!zfcX6*J9Da7|OO*ilDnkn1>%9w11YT|0g>`8(2J%JX)iqG@q5j
za=l5l!_vK)f-HZAtIz(k_tblYgzUEJHZ>-BEN1YxeZ7sif4qAjfD+_GVwe
zJcoUGzNjTQztgMe+%mE3h*5w|eNC-)`r%x)>+hpZ)bs{Mheu4Rl|cFx=Rjxm(}KA1
z+gWr@P{OR9oW?AeeY4qUX_5u6-vHrdOq-5+SCTs?6bIAt6eSJMMWEitf8Wsm=}mvv
zr)j@$l)sJnSTqr(jrnpop7Sr{n!IeCrr|Wnc}V=YXz;OBe-dBn+w0$*aT(%YMBiuY
zW*H;Gb{S2e-Swkrg&q1*5&pRCa>^7lL=ApVCRfz#7As40uvjNQFM<4Tl@RE!^raQd
zEZ2N3PqlTVTTQcd7Wwh_k|9M8J2kdD@o7?=)xMD#4Rz-2$L)*H@r)y;u#KTw&&$F%
zKANQliQD#wl*^yLoN)O*))r7UNsXK#j~y)T3k~O*z7@H~2DSS!6S-?)AGT3!W6u(G
z!!?sFM724^Fsd~1m+Hp-@NPg)?oVy9{Atgi5%EUvj#PDR>vDRDz10jf&&HA*{9JNu
zlfUZ}0t8Ym^;83}%_lCe^|fM=*O@ZLu)VKRFIYv}JG~37^7TKhJl4A*OKX{hGn(UR
zBX>*My7OX%YV3B8hMt9qd#s{p)Dd`@sN~OIAj;BN)7>Q6T@TA9c$@U3ZTK~6
z8|!&F-psKBwF=#PJ9N;T9mCQAAczPo=$qYt`d}^ai{{YnMprquf>~{B%vTTy{el*j
zJC<-c7cNitBaMWMj@E&24oW1EE6mRWjXLVi1Du}>vAc+Cq=O_;~3&-8#D4?*^NIl1u}p5b8%
z`D`QJuES_=XL1D+r#ya|KUAbneNeY{dT2aZP6FZ^_5=?Jl@&uX`p%>gtq^Q(QqKF<>QMd5Hc&NYdW8ON-m*|
zBnbQw+xUTvoxMtm1azL9S^qzO<5+7bUprc}&5H#|Z|)RdG}zGK!#u(>O{BDHv7MSN
zLr$H>z>fFzY_?fLJG&j=DPeNXj^aQVzGHVdwY?DHRmDtEyI!H(jAiFdb!FZsANE@b
zLdALxS@lZ^xY1PR);#aaHiOOI&tFH+{JD7-`Dg$YBzX`5OI6oOpS#|tiKJ^L@ZBVL
z7#03Lx8CEX{uL>G^P|EI!*{bxmqs}4upA4%R?TQ@I$gDUL4N&ouz6S-?36=I|4y2K
zg8~Agb?Khom1Ydxd+LPs{Z}6I_0M*=3jDv_k3ALyBCG0nI{f+{M#9Gna?0d^rPC{?AAA*=CiTM_(yx-kG^{HDi#Wac1q
zr>nVtA}RhSN?+AV363H>M{Ny!Kpv;(K%Vq#F@?fD2U%~MBT~$p4=oPM(4JC#t=Vr~
zY@@e-z4Ava^61a{)tX7(A5*A8Yde4j?9Z$CKR@n?O@)8VjLu%O{AP{sSc%DN+bOC4
zO8ocCP@mQXJ)S`HfF#NKVB@Y=;YN^AgS?6U^M-SRNGI{N*}QT)h2MIni-)cGRa0}@
zkfa8UAj&EKcaHwFTUcB?Jua=-uFi7tM8*Eqk0U=7B+cYnoPUg~#*+cLXM=En8NBIL
zSBI_APSEqZ)y?ry$IzI>F#7=#Y?0R5un)oca$yq}9JtCnU(0p;x})wUh6VD`VLi0B
zLb1_;uMs>a%>?rWgRkXd`-GrLlfWZHWXf*!R(gD(WPB#o&d-M-=W{K^NKUQ}hskVb
zNVW1d1+*yF^MQ?_JBqp*A^#v8Z$`>@MZ=Me>U71$_>zOQEQ-QGCim
z_qVr7nd+90WJJmTXR&Nv8rrT&QZleUFWR7j<0*@2q1}rvfY&47ree%6Yv)3QJIb&b
z;kLBs64zDjp(L7iDBM9vhmZGzt;8byO9n@6OTgV1_bJVwV6em+xMD?+6Aju2b8tw`QyA6I&O9`+I2=)UiCb4|
z5AJ1wT`ye#=NS|)x%ER2#!TP(t^t?Y~0-iPi^&7wROlxGDw%^SS
z!tDhNcaanX9d-JuTFSJ-msdF8bpfetPHUOWCw@Do)|ZwWtBNk6u15vgbmeBF8_C;M
z9aZv1*Z$4DaufwC8=Lvyu0TQFn4PFjm6m={;9_}K&oU#e82$h$5%IF%$MikxYAKB`
zN5^@|N@7B=&-5eUzmN0cEOoU9Z+F*_u34VJ8ukHF7q5@j5(c)GRmG7uzND;9?2`PI
zqJT+OA-t#sR&1Rrj)l%4%n0SbV%ATz*31cdYHL|1v1r18DVL7MjFoUoAq*Y|d+m|7
z-T$Y2!W5=6g(Efn;7#g#Zq){nz-wJkUD$i(kb1Eu>a7o0Rr8h0zVkB$&TN#8h`Rl+
z62e5x>=EF2N^V9mljm_tg2rbwPp0QlRU+6x9A7q`br~Y{3u1BxYblypCJX`{C$S8jR(QmFG5567pN_?76s0<75jno3<54fVqkiB-zHRa83O%YKiOe#;
z%ev+l&GHwiPk)#nkcwIF79?HgP7i;zxK?-TE2_uH@ZJBgbLX19N-!&29}%)raUQj0qlC18y43X;
z!lOOU^B<8GnJXy{DP>rMCX)X*NW7kZVCI*S5=>_m^fu=oYe8wMK?n@D6sXqIi|c#Z
z@*e@bi40~P76h!*N26$Iwi+v2EI=Aq_D2RSYPl1M%(4Cct(sjc98J$Bl6`SpU)Uk3
zH3U^oU5Ub^iZY$CKZ@r`KD-Z5#K)RLFIv*P;8dTc&ilQrlI_G8_=)zN*gOW1V%rT?
zh=Q=iSjkhVjieUyCTK&fkS^prTd~H!hY75_(2+wXEWq*DslDcXKR4rN~L`XnQQ?0@)lLgs1USxXyYF+mrwMrg>
z_FHTk>3-iWTl9aI^M`^=tGNLdxY-OJmVSRw#go6A%JUbk&?%o?*_H9YMg)0ZpVs29
zMzNm_!0NS8`EV+rZ6esCk6G3KNvgC)VEH#F%4c-Q1$%5_v!U~OHb2Yu1&X)>(}Zi4
zk2O?=Y>9_FEW8{|k^0i0{&Z#flv!+iHRObmLphfku7?#N7UZ2M2jK&VvEIdXhiLXA
zL&`eYtGawhZW~*qG=xIpXW3rhbC!8WD)!iY{PT(WO20E@f2o#gt%bCXo|c>cbCtPZ
zbap|Szfi4M#|szY*h(@MdE{A8$N@!~w&XVV0J$(*nIR?5RnrP*dFQi?o?OQ2_HivI
zL=X^&bVE>Yf3xZ}^dldLna3I6!4rTv&1=N#(J^|V0Pk5gc*-a(!DINf0*RgqqmAls
zieUNglGggCOLK0nU}e45qk7y}Ies{2LWKZWMKJWVs|DzAg073bhw%D6siDy4^FpFq
ziM0e7`svLWOj3A@fS^=Cgn^`hdT0yKA!vug2=>&oS6j_V->e60FHNwOdYh~EaIUez
zV-I@M%AUL&h;P7%ZVlZ=Vk4eS8owIZg)#OE7F;h%K1rp4WdAcrrU%PjSu@ELsn
zpw(+%WcS#imqAGKS}-+6ChCEa#9_mj1gRglb=e3Xn#0D$C_j{b;R)!U<)_u&x1B;K
zYDD|YYh`R~1?L8i)=M;8Gj1_nWs9Xkr@Cx(F+B>qPei^jVlI|YF8YSoZaJSiY?_uq
zw@x>c#v?#OIOAnKw;bo0(1Q0lL;Xi&I;p7?3V)=4o!U^CE_F*{+r~Vv(c!))(S2ye
z3`q~Gfu8b}&gHNCp0PUp+C2T2yr|DvB|Yg&vG_;IZjc+`-=7lvx@%D-RXN3@%4tp6
zS2hhL!l2Q0E}=e&?i4Yfz7|~hFiH5NI8Op2qbSodxLIiVNY8uLYTBLGXSyQ;D-mN9
z*)RYWYsNUyepmz#f)RS(dth<%fW`ajv-=;n2@Vg{XkiQRfveYpQ_QcVnyu+orCt2fXDU^}|CmkCnjO1Dnmx3EqP=`M0dF
zmDC*%s!-l#i}niJS-LlJ7wzueVEjQc@IQoh5n&%9^MCbjir$5}<1&l9y}+xnx(!$N
zbbbT*Odd2^;nNDqTqe-lKTr27!5Aw3o!8j@2rB?Lt93OzIHMv*TZ2Db_@u#RCMbu7SiRF{I(Mrt#uK3e+X9h7>SO5
z^rskb&afm@p7>jHg6An|^q?JkXUx^jx?>A$eKh?A0ogvK8e0+~Yqhe>u1#NX<-ul=
zAe-yrI-zyRJ*(DjS*6-5T%M!;%v|Df^spI$v&+YBZY1a
zZo@2>QPf9>Q<1D$Q!Y;gC8^4RWYnt1gzj(=+5cQSSsDft`pA#~rM5MA-K0Nti46&i
z;oG|z@N&Kql6c)D>P~F6A`&0dbLmV<_}ag`7_+a~&#KOEj+x>Zdu@yxB@!mfjddqA
zyP%c#x|-S5bPNTW97b|yhgIT&rW+(Mos%t$rODa4U|yiqD-^=JVqG80xKDNGc)Ey-
zDdI^GzsjrBP?iz7fN8Nu=bAZ)6Z!)_0Q)`kb2RR<9D4`(P49c9B}Y?dv21~foGeF+P|jQ0!eIc<2|;$M
z&-|*7TGQN^zJ{!ql5cbLYM}L3&Y5-w#th!GyxV`8@<+R}m8V1#b}Vfv8xm{yNv@ru
z>5!x}hEbwiZ)VgHxxp@sv7eJa*%viLPZ6a?!BDW8eD>~f=gPc=l*PMhgtK#DiDTgY
z?V5Fvp$fY_xw19>`r1>$T=oRwq&w2wC!r+?C{(IfDvp@e!Gxk41~sct9HFwjRZBUf
z*^_pCL;HS&_lo4xsBnt7E_hnstr34Yw3kf0Vg}-SEs$JO(ypSUr($dK_8VP9ksSr2
zKRoznA8Fjh;Qzgb7&^a*7Zo_3K&Ow(d~{(Z`>Q2t-4H+60iU0Pzge;lE@Ehqq&>S^-tglCxRCk?tC598aZ+q1)ZO9@GNP+w|d%|XCK)u5-!`O3R}n_k~_%s
z)U>yg>r`touV-?k3vDV|N*nQ6pFv#Z(lIU)mih<#awj|!3=M8Ce@JV(d>F@+a~*8~
zgBZkt;t*GICC#T1GH0jjyo`=p#;297=`48qx}nf>Bt^KxY;OmYwj=3o%an%)&&kCY
z1ltOeR~V?2|NP@Qag2T7RMbFwPe;!xvwrMBd>s0krC7BSfR-K<9uyVSei}A`4f(qV
z>$juW(vz4fN;7)p?WHohboJGfBCl*KpGZTl8|6%GWo2sQ1k4b1QQy7KtMl0LLCCIh
zvoqvjN@P|6{Yefmf4MrU(~bI(3Hkh(IAKvXb7)cW2tB4mN_oNO${$3Z16?7CN70Iz
zhC=;&@LodXxyg@b(%#2SFJJ4O8+pV=6CsPYuGGy%ORXYHY84vuZnPe|M6(Vnu$TBu
z=?qK^jFa*A`HwMfc@~gfdUr^P=mp_;+s+7z`s~YSBYG4-c)XMmGlXqWqU+bmcsvrz
zYphgFk!Y7F
zn;kfZqyB#wT#x$$O|Z)#VYwV&+UUha?8#s0@OHrq?G{SbmITi6$$@wl;ZHMlJO5)$
z^~h&rJy}VtEeLUvI4(^(w
z0AIG}Ck*-)PT4dqbI5*)M2Bfq$!b}MPiLoiW3gHX0wkz?`!J?WVJ-SDocUxXeE0U6jylmZwbuwA&x
zT4f7x39ag&Sc!iYgiDSeF-YtMaiHfuyQlD;B@V?;*(l9+CV#~!nskkfxS
zdtmcZNoij~9v%(3NB-+5ZT)b6KPjP;oaw92dAEmrKWA7btUTf!p~us2t$vjwOMmn^rw9yrA|`eSDM0@Uki7tr*ODfj$4)aUEaK%asXw#82lkBec+y-XO^h
zw1MCQgwXHJ5=u)hEDFIVJtUF5Y0oEan;JN~O@AwD2xHYg{FCu&9f?yI=&`Q1`S_z?U
zL&qJLgWHT=i^{qZdOUaDcwvKvWIG39&pzs4<`#m|>^Mx5Ejy(uCS9myMJ591UI^_=
z27ih);+_vRic@6i6zC$-xHpBWJbx+{3_3gA6t0%@tQm$&SD#9Un-SD7YryLoH%qO=
z|EMcEU-1b%{|bjd3Qg2jOgL?c`D3}=R}ibwhy?uJ8J=vW>0yocz?;~SKBYgL$qv9J@qj{gJq8fH`Y2j8PzFE
z;%_%FA(v`op}ZM&rr+Yk9NxsIkhJrnHM9
zRQ5x;d@iI|5{;!f)qF`^ukP4i@_!{iMAz-=2?iIj@J_r-{0QuK@j{xQuYV8`+`*LM
zu1_fS$zU-%&=U9WN=lCO!w=p#fh($?MpBJziiEz7qY~4f$>O*&3ehihlv&TS4hO
z#TC{_O~Z-q!B3rV?SYQ1X-PZsT(w}MxvdYLxP&}+Vo2}T5NVT2*#$%~?w+x^4;P+?`dxeX8gQh%%&1t$zT92-G9-gD
z5+|WP-zPB2QLgp7Wia8~XMZzMML!9V3wM@Gw6dM{90TkH>{xKbFq;cKf;At_;dvnMT*P$4Qe^z<+faYYfL(Roka9
zn|1bKQm&>dRNYvS!^M19yHEV*a=?~5|95JwIxyOzN`3qCj{rX0sVPtLmL*4(6!mb=
zycKmOfj4d9N=42*%
z-{mW#2Kuh7(+i}n2GiuI|IdMi_96{G$)DUCHUQ*O)4R-
z_cH`&$)3(`E?{*5Eu*HVM!W(E)f_Zf=A7~v(ofpeULvI|QjtZeyk%EljksPiEH{ye
z?@j}5kwOF}ca33wz5_a8wax{LLj^nPCVy!0QOy9>VEn4jPJc>>GFLdeOv!Kno`^Z)
zyI~cVI4H7XtHHi5U|awMnR%!NbU{a(^>;~V3+zMVoCPbNlo{GhAeU|{shWjPhR7LU
zz$>|IUZ6E;td(EiR)C9QUYmUf3QcD)1WILLg!W7K#qz037?)#2Bwn%Xxh{UyT`LhB
z#LAxDjd>HQ_J2A@vb%!vu#>Q{2Chb=GrNAnNg=;Y+c;Q4dzo7VbEG_;F}}DtrwHZu!%V>%}}BBM1SjKZKw>6^C6;>_7%pBSTQ`E
z!r;w~A9jSJ7q@n+nSGF*2#>iy_#8T(+0BVO@ZuOsy9Ooi!WqGhIz95uRajklMeNdj
zXloV$zxdkHZZFQ?9Ov1H9M6u4rn)Q94XF!K
z>Sp=N#$lI(h&6&Dr}|!@JZE#XL^CJ6FXeVaR3&|XS@O-|+(!qUyHspG%Z4Yve!DU&42Nu@ILw$@*_ul64XvgXNv(`_dEv<
z?VZ{;YICFwRQ%QwX1~yN|7eL3tQ9gv1mPX{L7I+tYWtj4-R&`p7UkljK3z7u>#vIX
zb%#!teQ?4Za|=3xQr~#eOuy+96h|coAos`5eE7N`mbpJKm4*|Gh6jw>+Pyle;3jdy
zG=JQk64zgq7q$+vmjY}BHO8Clgpp~~yF}Uvg+(Q8e#Pj0g&+DI+yZK*M=}%5uQ4LUE=V55+-&lh+9u)1muV4ddq3Qo{aDZtD8@g
zWLG<8&K|YVWVS;ce(mkJ7zWdDJ&uJkB7fs4vlpN^L!(Lp4(@GEeAn6wc>0nm0&1M_
zPyOa|K^T)Cs0tAh_t_&G;x(Jhh9!lAmSV)BU^}>9?-kV$>5*&0Tc#pWsUX52D&(GD
zioVg@k~}s^yME$&1*#(vQBLOmj*?p7nWZkKFD_<4B4<+lIJGk-e*
z9(xU{`N#pHkzaVa-DY(H(ILkP&O6QPQqHr~UK*KT>FEKqdURlWQg%>vO=Lk`@=sir
zo~qjA9(Y8`dYHe{&S_^i$r-lY3SPQ|3L%dmGQN*2h
z7o~CPmDLrRhh_{8I{^*Ih=gF4^K@uI=^PtP8;4IVY-Xiua<_HxazzCJ^d1wYw6HvD
z#>?+F+a4u0!{k2CbcLkfM$Y9Y7^kcE4DJGD>_Jt*lDJk+@ADqZ!peoSrhoY^-Ruz#
zMD*XKOc4Zro|g7X`1aF5mN20Y)@vTei+hx3aEwv`rGM|~J}@$h!&c9gnn
z**A1<^u7FWc`JcS8=;ih^Gmmp&|5uInpK67qKB$s%DxXpL-1i`wSV)oXEMVpB#uS)
zR$`P6H3kiJ<*1NyQnH($MHvV2K_c1*g&3&EJ96nNM|whbF?B-qZ=3F6;GTM
zu2#_O#MiN?n1FjtJMj}O+f=Z~bpTeqyZ6Vu)=xuCjd4Vv@ds!blx^}gQ*J@IO|VWb1$}Z%OZ#-qwruj@W;i5XQv}V
z*5;wUP$FbD>EQmc%7KlowrS_O-q(c*R~#$I4X<5N<|8fz1Z#1)DPDh<2H@NT=Q`9f
z-of9Tn_#Yv0qCycy>S@3ic6AYwCU6_Bv4!e$d{5_L2`#|A^HWmv9X#KCL}5K4sb<}gXpbX*Oyzoc-R-E-45RUtT~5sF5Lze5E*arK
z3)rc<<(ET_nFgawe|SkFjrL^ZH&T!1S;eK#UZWt3f&QAWO!r&VHp6M-lYX5Tw`h-b9T<>u?CB|I#8hI)=b)i5`u
z-IM9<_bT8T1;?{ZK|+Le_3eSfUjEv)m26s`2D~O}c(3U5hTzrynbnHR=79hl8+UAn3$#IIIhzt
zk6gXSUBb+Mp|45Mty6u6aoD^#_PVf1)*!U=b?w3ajI3R)`C>N@aU2mer-?r+?}O3C
zSqP`v>@i4(5z#aq=`0o(DRl^YoFqbuk5>OsCE?*)B;qc{F2p7@I@SyM*tgnX_4Qb`3_En0iZ%~s{Gx%naJJpq0NR~?q#|j
z8UIy#C{4BITgshzBUjM#)1Ek8@06O&xSh@>;Wu2wJROGynmr9LRBnW9u-&IffRvHH
z%YQsCHPHv%!YuRai(+##MWBNHR&`}`6Mw!#o42G}zZd#0?~^Z}yb^twzpt#Tb3~N!
zqje(YvpPsG;+4ks)_k`73mq(5Vv*w#_fu0M6c4=j%45U2VJNh7Nvh!%Yj4hbt5^dH
zx+jOYEE@hBY@oI*s5l2)(j11`M;e37O@D?Eiqt?Ga+h`a106Q1adipsE(>wKo$B@#
z4ZD?*$50fm*47x7(5Heuk6*V6BX^x4+Ava^P6h~BPIgX58pWM(aJO)VT8(}fTz)1W
zGdd7~)w^Fl!xwady}hb;7IWcJIBd({&3CsOJ!ANKaOU}jZmRc%vKN92SE1N)kbjfq
z)6?}azE&qm$hr3r%Xo25*;~yM=)O%Xk6K)rbVFGq4zkwFTie}bV&@0o!nX{c){KLe
z6%D?HnY6A;aMvD?DK-J9MOhDHk$VV9EBkF`()4(64`7!QgUc4(
z6j@ZYxbvMoLa)T?qw3HBzeLjLb3J+pAtrWdKWY4tJZYS(YFHlh+0QzUa>TKZBRc(c
zS#G3GKm=(9d|D_4`M7bZIzg%#M|qZMvG)foW0tMJoEb|8ADAXX-C(YJw*y2EW4$H39ewqtTGYb{Vde0Srq
z|9E-fXYf6c_N@R5)yuY_0)H!$PRT1$JH)^(M9q3^bU5`0>P!hmL>5miYy>mHU@6zT
z&n;N-DA5fJAc+q+Q(sra)l_bR`A%AovK;99_O^lz%
z>s$6Rw_|co8kMFm27do$p?`_9Th9yrF05n0yH7B0=L$`@-6nBbE%PfmEk4$-MkZ6^
zUz(GZNSFy?>ES|z%$ve56Ae3^dtw+-KY`n2LPkusZdB5pOA*o>{~z1tuqBED7{Re^
z+qP}nw(YrN+qP}nwtsEgcHSb{;>c+~wCS`s-G)!8gGHws>B-`mpC;^?%}Q_kfR1tt=tNgs@)mQ6*2&
zZ&adyF##?{3nmb-_AaVp?Ss^AeiwGv3lJbD?
zqs3iQwFCwWsn-ojDk;U+rCueqAky+{OcRlk_o08_baC5{|sbtngD|-1XPfH9+4hpC()oOpHoItOKUQK$MXn-kNAE+)`kun`nw%oR*lbR_<
z4F|%-ckVwK$Ml!*+Uh10e!2>^wqCkY{C6k6(A5SBL%h)Rh_*8cqlE)UThdD8IfCM%
zUVpB23l3|Q3SOJBX}VMf|1!SXm^L|sY8|%T_xwmyPu1TSS0kslxyfG1|HxC28hX*`
zdyOmPe~EA^ukResfK6HBJnOTCjR^{`i8%4v31kO1-&FN(xEu53q-Q@Nv?70MqMsFX
zM2lj=tL%=i;f5`W8_rIY&JC;OCsZ5pH$3vft=8VYof&hzKC
z>a8_QP56bkq0y14ndORlCIkViAoLTk(N{ZO+_!c|)={{~Ru@Tt%L}*!XsPQ=0m2=0
z$=Mj-og`$<1?6ODTk!@~D>SwMG6X6(5oxfhrdQEbUZmiCazPzEL#HD}sg8{Y_J2WG
zf(Wc>!j=lnBLol1PxJe5+toWn>~J&NcG#6kHD+|Mh<7>(CG|^2l1if>QUS`J72|ZgRyu4~}R2
zEE*&b6Nu`E0K2^j7ktY*<*t)LN`J1lpW3ukqH2phN=;`2|FE2I>0-t8X=gY7NUZs5
zHT}jUx{M{^$c+g8>Y)5k6RT_RxksTLVn=wBbNi|``MKILWi*fS-y#d?Mj|{uMr>Vj
z=hP5h3^U%YZWEa=TU`fFxi*g6Addx#>_7U?tZ)^%thkP50VE?ME}ShnN`E?)t!WRc
z%PkeoH?VdCVEjy7HRtd+-XLD@G3rOCQT4#O1AJ+m;*b*d0h1`ry7yR7oPN@6j<10I
ztustSMYKyUR-m-Fx_pH?G{s#Fw3RNJf0=eDPJ8V(p2KRike!NGd9saW_#*R8WB^j9
zvva|RmeFf?RS=3WzrGajYJb1?SRa`5f-+6g27>n3Ro-)@;ai|AeSETtS-N6`WUNKZ
zVy8H!7?3yuyvAL-wC+q_Vs#+=sVj0I9CNlf=thJs(X>g(nTNXgQ-L?XaOc$+Vy3y&
zY$Cn~6R?CM^`E3zdu2KcZq6LJiauCXtomi>F0!4@#A1-|$o(G5{D0APQEvm;rb8i|
zyI6i`Cn;Pl*A{|IY%Fsc(J9|xSjyg5m|qV2POV1L#lQgQlO+VSzgb3t#;AaP06O*H
zrcmH{LAvwZ(*Q3>(l8{%YoauI$)7A#(ay3bGaqq?K3%%~+SNA|&P6Za59SR9!WM6M
zGkBOm&kESmfs(69{D0-HBU+JcpzMxRJKr<-XmI~rZ{#O7}+NAE-vwsb$i>#Nzkiu2f#grr%Y
zK+`aEuw%1XwSxk+@~2(xAZV8Zfia$zuIf&L7pG4;6hF+~2!9~}?4d(D!8)VH|9_r%
zMWHi7`Qa`=FX33Dp5%q(1Zfcl#|!gV#g(ln?vLqZFz$%pA+VJ-nzqC5btu{<-rszt
z$<2pdp7yql?F_kPt$3xjOUyxoov$ze@{69*O}L2tX#sFJIPWjLQfSd78x*}wr_wVf
zX$6@NfV_dnvwxV3(RTTQkGP1$>XI<@EJevOC-Sb)o@NQtxhF=;;VN*`UwCq&uWAps
z$TK8OfY(0{%Y34TF7@S4x5UkwGA%8E<67{|xut~f9IU^cCy(DZi
zi(YF0UP*o;)bgV`9l@Z3_GmffFVE*9*}Cwve3#gzZX7p^X94P#L$6?6ZxY=yDmUEn
zxeHq)vrt&o{hVQ4<6iE(!s3*AQRwA{Gv66)%R-w>zYqnSF!sE=En$?Zib0b3
z=m!{aJN2ostyBFVL0UM+L-@55H2+HXMZ#pgmw(^(tM)P_ro3RKa`5vv7Tf>TA(vz&
z|3&x1=-IM{5xy(s=ph7Dc~K(8C(EG`Vj)cvs0Qv$v>cD`T?3|EUEQS-=VP~Q$LDv&66y(wp?|Ab
zCWHQg9+`tH)bViGC*TLSd#%VznV86g9F#6ZJE=XFI7Or(%sC*fCjlJGyMKFuaX=Azn$q(Yp~EjrdWyp?juEk8#@U#K_eOLznY&
zm`g@PwsrnN?YE9Vo@Tz=^PU{Qx_`;Th%TW&)!w{eZ^-cHgR6$tjw02t739$MIY{;M
zS=B4d$puHInr%0Od#(bs5zHdHgUOq#ksf1Z;CpB`bw%rOl1XUFDX_aRc6g9@48D1Qgj9GAOl
zMi~=y&9fMsprlsF8+ywGB#x+mLYZ>$djdp-8iD2UFTLBTd-?Fwu9Cl@xE3C{cL(9n
zd%rK`QkS&DuSl!cO64X%Ka1&}FTtnemMH|aE+dy;X@wvEMFf}9-oP^~`isjTS{VIl
za1^Qomch!4f|v3C$z25JC4c<;Un5^>tPH6x7kGL9QCHC)_T1S$rZ+l+R~qAT3!tkf
zB_ccK3jUy{YP^oi_sHpw7AOhpE2HnIw2$Wusz4-*WejMucdmSm$IP4O)_3C`6ukna
zL&4qPC7rz^Hv(moT`Fdkyos|V`9-@Y4FGRk*n*2pD&!D6uQ4#h)_<&zV?#|AS@MQ*
zpG*0tIwvUvmXLunq{;M_4D~OC+N`Ca-fyJeG4=kQEHVi~bXk{;dHmmMt!jdHaC~gv
zRw8};N9689s6>O?2Vwsn9(CDb{}o|4w$?QMK$VQm27Io_-Btx;s5i1u7Xw#kx_k(I
zl@u-9Auv_OCyUZI`hNoyW1jE~bhYzdca#qg)rTbJGt)rHK$e+7T8N0Sc90xCmp-%w
z()f}U@`?dj{_#2Fre^L{=Ayoo7Kq(3K1$H{lQ>6k?&gBKEB(r$=-e0UrZG3Oodm3>
z%o=F8qr7*H+@*!_8U}4Ct$2Ss(l?;yKKM(K71QC-W4{3LoqxbABTtx0yY6l*qXL6s
z1nLcvTgvKp^lcUSvM96FKlJBO&}M)ChW4sT;S|Opd>TV%zoZa200dQs+QP`bKX`|b
z>K5{sqFMrWD2>?&`4=6izxokY9rY?h^UnR^d%@Q7c`IdUpIQ5XXC)77x3C+b~
zCyCBVDqCR&Na_A3iTrUr1LHD@d(UE-=FZgS01QFPm0z68H^}_66V2Dc%cZ)hqJuRl
z3rG2Tava1I8B&^f9P}ni%C}U_Crr1ct3}|624Jn@WPgsaKHiqS14}`D*8wVk!(<6Y
z#&0=&=9YzR>#=mjI&PHKhJIgZq_TfH6xNG{t<*m}2yp*QT{BAbd_5KQAHcLj`!uiyjP5AbU<+AN@_*T}eTKvsFp?E?s62k=^;dHmnh7M;
zv_=p2iER{np+eO=JuvaeRb|8OjEPxv*2=!a$ghDoI0DT2DKwM@0X|0*mrb|q5GA>F
zU0fZevSR8_sxv&y1*G6avz
zZnD|ie=!u>iVVdQ+b5aO(2J5u=H9n6ak2s$nZOsE!FFbVqQpKFD!hGJ!QQ7Dfl%#0
z4Hx+EtB@H?$5wt+XH#9S1RVgFW#H=`<9}{>mj6HyM1HP}(Z9YNzYQM`f>yHIAd{6>
zv^h}P$c`^=A$XvW!}d-^3syy!z17zvfvwBgz6YRDTAvYx7DKE)LN<@P6g>xCt5$M*
zyQ&Rg49RRoWqY`XGqU5qa-vzU|ExWkU{f&7wZ2>k#sV1qmv$gHm?`r5oc~+NCx5%%
z1MwFD1V#ULPKezO0~F{tV8KgimDYAp9tsC~!s={SMyeYy92Cbhwf$@Q$mmrRSOIho
zQKg+&ZT)~CO*5<}j4N2lb!nT`52v!(Ln*7glzyf0dNtIX2y+X#%lR9$HIQ#DxviI(
z_oFPcxb$2g34;h1!1$or0Jfeecz^BZidz5B85TsBkcaH^P$&zd%IpNywZ=5>BTj#<
z5W~Y=u-c4XfvVb*Mw9T>#L-s0WggFF?+m}jj+0jM5s-+wg)B!dp~l=$zG*mEL-p#L
za82DM2knlGnrn5s&0yt$pC6r&7yc7Ue64uwoe2$t8n}b3t4y!guXeF|et+Y8dbhre
z;cfIRQ2Tvc26
zlBaA(#`*AWAWlCF0iM{>wfHN=Lnz`MX5oZ(ClmWSvzxNmLLVt*7*+izCr0h0{^o$|
z{K&pBP!Vqk-K_`n!U{xBjDP+W!`G6e_CzPY7l(A#v8h^>BWx0fejs@Du
z!4X=2FlXkQ(5R&GQzQ(NObPZ6fY%;ZtaNAPXvfTxL@v|E*J+{4m3r@pby+f-@ZuJa
z_ACjIa$Uu=0o~kVbg%kosF?z8uGP!{ZZ-WOOLqII1qkl0BX4<1UT|#Xh(<}jl9U!J
zFTU>{zhxZP2~7Sj6Mr#QViRB9g_PUKv6C8df(v>Qc=-jFWVVg4zBxJwa)>JEZSm&l
zaE3U*Th4bSmiv0GH#KK{@>QI_8Obp5SG{CqMGJbo`t&S~W+w@wP&^*SGPY@XNTSK|
zkA&Z*S09E!rmne
zd~=Fmm2-XxjkL?%+8%x}aQ{
z`)t{1`9y_CajX|pC}vhg&#e|>JmgVBl;a`p^8RVEI1Kk2w}3ewyk0w(<_Z1yS&Uk!
zX`GY>qgX##nt!uGy_l*7QA2CgOZb%St^z2C
zcK!B-7@B;4&*XW{O8>m@TdTHxxCcN~d?d}2_XZ7O{RH{w*wOp}IIj6Xzj7En{5j=~c(wCNc`)_*`bm_B=$tUSet&4d`^R;S=w
zD@>JpmzU7AC7dc_qyRYfig>JU;v#NWq7=2RmY)X)+B|Hy{MJZD1yH!Ol8=(rhlk4q
zxTDNmcQkk1uoA?h6w-k~XDF+iyR;lKUF{Qu4~4&=q~#vcqI^^0ECSBpJ
zc0uL@g!jSDveGEr0C{XVjDA(8b#A&~B&>lyUVo)k1$YeFIDc^W-e4ingSMU~-0_3}
zdv^_CVcKZk**D~_oV&$p6!|6r5M5YKiBk3Gw{Em~9pmm{^L<$Sg24={J)2G?fA`m(
zp(X~lglC>zYttM}oc+@qe~KvR)R=VwIf=IC7ZuLyu!FqUk{yHND`N@q%pEi3CN2s2
zvwzV+^R9Cgadqq)BO)O|h=!qM09mEulOkhovZ{2Xpj%`)DyI!2@Z)i*f+6=;g)5rW
zYee3scAqA{Y`qTX3wC&eSmNc@%upZI?OJHCd$sjeoscPAFzU5dKE4g31-h`zrpyKT
zI9CdoOsCVZWD+12Dt5OgJM}qXo2rb9{eL?#@ZdBy*;tgzrj=6hF00io_4qTI@=
z^Pv@K0Iu|8jwRoF)macB_}lMkk@K^;)PgtX!6PvI$n#T2_E1w}Wu&i6A_)}wFd?{J
zz)2~7{>0N7YiD#unCdJwVi7EjC2CGDJQqzj?Uaai#w#*v#^c@$-dXOoGLP{&Xn$fw
z881C6N;gvb2e#qXdV#Gfp;j9UfSSI>UUpL3;
z@u=sk%AZkq1{9S?-9+PZjtQW0ZZ?mq+M*tw`ZQu_0=yC}hq;~Zk*-__O
zB?6Otzy0P1l7kI{_d7AY@3>H(y!#RdgB?qa}ZU1m!n_&0H6
zqGIJNMcgWBSXCW>cTL;m4)4=
zFxRwItTiVi{yQYmXXXu6L{tE}#tvjRz)L@Xp@Io*pUWTWdmt~dE`KI?igcV7!)vvq
z2jOPS8apiV5U>2R&@fPz6D1g9&5m)!_(QN+bXc
znR*=_STw5Q5dUwZ-Y&)4~F;p~g)PH0g(z_y}Y6xADY3eO718gwc
zE>0BgQ&$D0$TT7x*UAFTwp-1nKw%#DYPsfZ?sK@lPJ*CWTl`;C&}7PFM4K$yUQg|1
zd}lE)Q*4t(uv<{FN`Ggh(80K`@t`OMN*?bE`R5S&>DGI&?lXta^LtDOlACYJULEE@
z9hjDYmw(M>EWEcP`rxdGVGMbj-;J+m;KI+#Rpo$9F@hC-0{MxrK#Dn+*;q+s{@BLJ
zQT4}OMje=EBY5N|y)1M(y52Ds1jbfJ1deGZWU_Cp`TL}BW9+Tev!
zZ6Z%WTTtp;L1j~_-9vdpjpYjScMC0VWQ+1YZlRyp^5ctOo8JOYh?6^((UrUG2)6~>cb)T;vNUA_9jld
zURn3nCk6fqP8EVyLn}_ABdxFreL14FZDv~|C_kFC-~a-2vtU5>XOAQHQv+
zeShE+?rhzu$BH~F+PC&uYYOb4M#~cIz+Z3^EwCkw99@fmeVPtxF;XbG!&dol9uT@=
zZtiF)rSDtmIna#(w#GCAN=*E6Zm9kWmM4kZFQqjqIZ-M+5LzGaOs^o5FLchiV4wtq5S
z$PmOiF0Z1fV78_er@vh1*F+Jrp;G_mBCCr7VA-(?U=X)s3#tZknA6iI@c;=hv^{3R
z=EfRvH$cdRYwdjjB`|-l!uGQkhV(0kDZ<_sp;e9F`m%7{zI}I{YZ3!>IV{*v5GQ6_
zcrt!JQBs;=ICIE@uUm=}h&ff>Gk-BZ|9hbn5%@Zo1%wOxb_hR5`T?~?D$S$*mc@%VmLS)QZ1y0Fk%2R-gE*@bgRqT
zWkTBrwho!RtWrG%Fi+rAN5{R9ftPBcN7Ih!(Vc2@kPpg>luwjb;R8Do`&ng50~Lvi
zGYNFthkFB&22>hib$_eKAb%1`qTMd^*;%10rpE~vi4
zuqViFro6Qh4Bqk;j``Vv&84u#3w~@hR}$ukAw4gfE{+@cKV>7kp?@ca@}JE{*${|6
zMlfphJA&iHek$ac=cVB|TM+??$g_zSuaI5CS&tveRQP+}({GfgyB`W-#Pf*hPgl&X
zhCI;}_{L2E+j5=!wn00KkK@pU8{#cku0Fd?weH>eCrYZw1&B(Mzblww1e^5W1Vw&>
z+Q+?dfK1{cAXq-94N8h2fHq)q>nWHvmd{<
zx4Z*(Kb0`Hwo3~pYEg6%7@FUQJk^<&i3`G5t|NN%vq_1}b$<{t{oD^_*2P8p>?g{`
zXI5xbW)71EzXG-9McMHXB*6Kr&A(FzdCpQG<`I{%1J!Wn{lRzP
zA|9*%ltL+qugJnXR)Sc=##GFk^FSD&FbmkQLgN-)iuE1ut{
zH68KdVL(qo`I7ZQr@H)KqeD4+8wvhB7YuCZXaP4tFMr7Ox4FwVi;3Y$-6`1K`v`u1EHe
zCN{;G&VL>rZ7s&&x2I&;EL%xR?}FkX#%*lyPfWf=J-P93hwGw8g!R
zOI7|cjC?fBC(3GXph>UvPuV@UW6=WuCkbJ~vG-0CQ0^jZcjw+orgNx2_=$a4HAl*t
z9GR@}tO@yw<1k}M;@;Tq!+{>yw!Nj3G}x%|^M4Y9tVs~Mh?&KIVw@T2_Gz=Nq=;&d
zVgNRKr|y{K9wl`x<{P1&;g>UE&0QN}^-c{ZbpR(`hU;{(QF
zZ70WM6oI5mRpstRH{t!Fukz|AtD{r)4;qu|_PU>DBZfe*$
zGZjHBmqHE!ao(@88l%9DFAUu7_lB3sMD}Eo?uYfCh#YEbiQpx=0c03C;j-`kzbPVi
zzztCPf1#nVK58Jl>x8v3Nm)wf{?Z_jK^BAF)04W4bg|F^9@YLy0RG;;(r=TE3nMCM
zAY>FbGofMDnM>5~jzVun2S8Ew(tnvOuk(`R;D1JhKpDkCJY3Vr*WnUb_c0eO5ed1c
z6iyXyA8}yt<)Ay76-@UVx{^q0_%00kkc^MtQMvqG8w>%h1`#d6KKaqgQzm8k0ko1(
zIIJic>?e(kRsxZJJrD{Z>%E<$>`Mc)b*ar7csL0t-l7B<&ig3TFrmigKYxkG;0$O?
z&j;mqs$I3szKF@lJ1J^@4XfswD}W4?-=6Z;ubz<%$oCW(ranC711q?2Sr{P5y;Q{)
z!KKKbbDW{C3Jg|Ab1c518|v8=$`SNlS2RxIHR^?P6!fQi1tIXRLM4!g?pt-P%cFwY
zSsJZJSGwac)<5d#@VI5jIe($4EjG(%Fmw;A(f~d_SK%r@-*u4c+Ao;jNv_Q72?0NV
z_Pc={pWPJ|p)J_s!fesjLS;=}3jFes=0riBeg36Yh`t&nzk_Rq&Gjz*qlWYO7H7;y
zhs~78+mEYE*?evoqfE49!oM8mdlcSL!~$!K$C4^yEtT8U4>9}`bbkat%B8BRY7m*5
z&6^+XWZlXeUZQ0`VZec0sK$}U4&P?7N|79kyX_Cy48opleh0MFg6$vVz8)_%2I0nu
zjZ-S^2RQv_wXR=NW@e~?P161s(s{0*J`b{q1pLK8XsqndFJv6vh9P-vjQ)%985t-H
zO|zq_j6ynk`0PSf>wn9W<8oorXz+sM3u*c^bAC`x4p=6YUYK
z(-;P5R@JXAnHXv)(rBD{0K`_FP9N}a#g5wfBVjcDm{~Y=0Drx;5H(>wVtX_p{3ul-
zt9YAR7f{^2qE%~=7w+)Kpjf5g1|~SUKlbTwC@EuoawKJQ33yTC>hYZmkgflU>QlTi
zL0KYa3&pXTCp94pjf)XX1SZel9qF*IACV}VM~QmZCv7bnHX+6uOiw)@k%hbmGUrY=
zyGIYqlDd$zi+|m}83kx`UsXh|Z#5q}qIC$GuN#E-h%eZ*mDb!%0KtO^?FM8)dC<
zeSZz+AndLeGuRYK<;ws{RDhM;wmedsqqPop%~&$;9B
z(2|-$2@wODE4e&eLZlV`BN}#Mq3oXj*;I(_AutMx5Ve(?wRJ#Ks~L_xN@Y@9*~$JR
zgzvmC$A93+ss{PEE0eS}cE{NY5l-(+uYaqZP_zV4EHtC-#YD(;2q1=qk2JSVj0_Kk
zz)bzEsJ#;c^y7o<&l)n4=!{&WHh1(j<$|=+>dg?WH=Y3wqRYwh&Ls-{{*J>4D2hbI
z2WDyFfRpMtJyor{qrg@W+AC@hjje#SXl&rpmGW#SpI~-!UCI~{tD!}wVGPYW0Dm#e
zck#*$6_KC_2qO6GW}>3vH|jH3eH6~6RoHHo(;3OrNMSNu4ROn14kvD>EQ6r7c%#;g4zg-QzTA8FTQJ_1B1n0xLDYj(17Nn`=|BVp?t`gW&lG{
zBVFz&uL(ppx|^CNItDGy!&I{Q-hcmL++C*8m{0QtC04#eejmtb_(>pwxRsQ<*3ryS
zw&jMG-gBMuwNXYBw74t8r7jAeQ>G1D%8vd=TAzdA42iM=%ZU{U&L3K-AmOG
z%{mss6()B61=T2Ax7Ta*afY&cn2w*pPl)+0%*pI%1vY^#x4^QaK&9m|P=7PeqKZDG
zU9r|`(p%R4T&-_G`iT*Pa&2qoaz)yh_WIZkJRz-xtN?rF!L0Qx9uH6>QAYR4YG}`t
zt5~z)rn(5`j~P{F);~@fjQ%ddTsn9>XG<&3(#opbRL!-6yrDw6n^PFkWq_>X0MR$e
z6
z8%~xUV|lSj*hDJ8*s<(lwa^0ezi#2L^MjhZO#2Z9HA>dIil@TDRL77~n^{2QDI@`N
z_HP04LY!DI1dw7#m@gvlL1~|uSrcHXIzAjUO$`eUx9hqwp&j-_b$=oD0Omv}s%W9p
z+#dO&qisxYGDLh*>$WP{xjTtM#K&&@dTQR%olr_%q@3L)C<7>~eEn>@*^Vxn31_$z
zHlGt&Wf5EXc*97piw7_`!h@9d&!k5;_R^Vig1I0PWEe?gk{h{E`x!9eKYQ@ZrP8s1
z;yS|CbqswMRNHk&k$=_aB!G&LsA2AJj%d_Tg>@FC4(usjv{@f)qMCaDbcm%a>xKl@
zvS|IJg^LV>luYPQLm%o%mZKVG%3GeIP>Xm1{4{JrvE;0xoQy}KO^^UiU8*yUW>tmV
z65+fv&9J(AFV*`c+*GS(62VhK$gbP0+;P3P;x$IrFC#%q(to_d2db3V;f#q9c1#|Q
zf#qj=al2|xFWPV+c|jWZ>28<3aj7HyrcPXy$xI0#N6ybofCCMg$2`juoMEWw(A)d^
z7Ax6z%imQ_f?#JWf$438Mr{z*(qW$DtP;4678a~i_@->H^g$H-O(M4UNOCC88?1x?
zAqI7RtmgweuzzMtvCu1VvF43I63)xkLC+)4FcjC6uD&Hv;G#QB!F7}
z+OA}%BB>E;P6y&e!@iJbE>C7uhFj9v(MExIMFzW81bmwr$($pkusI$H^NT
z9ox2TTOHfB)k#jiI$xce^H=?MYwWeF_P*G)tHxaOnFBFJN!Abu%WChwcJ>NNfX~hV
zGE2{63q=5WE}Vd0ROZ<o!7ty
zQmvLfEN+b@;_pt4^G-Ip8b{t~KVXBX>%oKPIN2kJX@cSc8!n0;17;w5X_{%MyWt`k
zao^eG#2YMJx%~(f3G`5eL7Icwmo2-Rj<=Rx!_BY!92-y&n0l2w(U&wzJj(X`^^nvI
z{D}AlY_2F*P$dGqxVp5me0u)vx0*jn;(b^qLBWxGQV`?XehKsq_i%btGTyN$O!iIM
zQi(L*pVD1rP98CfaDfFLLFIu(fl=%nIU2tXs%(38H-Z9Rs^&{Y)X<~4cJKdqrxTVc
zbj*HfS8*~FMzuQWXQ+QJL^s&=`cK69XVJ5-LK~h!G`(;O-}P#kkJ1Az@z1H%nwZ_S
zro^`qQKBPn)?_0nx>uLu7GcW2s1Xb4KjhwVU
zeXY>whWj>vo)A`2_4kbD`_pyQ?(XkgA*y^>5*LV@$EUuc5VrJ3kAR*mj1
zffhioZ#G0b@WNss%)5pEr`*e77C!IjGbXcPng-I`5LN`Z99K!k^-je!-a@9IRtWPh
z#@0K#`Q%3PuZ7FQDf{ak8;_VJ{N0qc??305(}ZgY04Wqy=lxhAwv?BPbe2eK4$x5k
zfU;2B(fa@^$98iRdx4+!8v}|*pa;wh00%Vwj?Cnp!7m~`Sh>ozA_YQ0;`?CcrvUR^
z(9jt5hZ(QZ7lA^5^#c|*o2-*P3k>?QgG+iv2w?R`qZ#IVol%GrITUTT|6+6e7KfRj
zZ4v4VLv?6gvV*RGvRlIG{vsi{M>MUsMbxSNlpN2cG(OziS7>w#v)8$dU4x
zQyFrxbRAMSZNw;<6ANRIGJD|nZ@_YzUfWGt
z9FDO6}H0D$xC1Qz$nosewnjc-1qoETw>DW*s_Ga$PPF|hpt#ia2nCS9Ss)v
z=JdP}=nD`!BGkU5SSVJnILsM>9Gn_(A!xepH$SMAJ;!6H>#_U1Gy{jK_^BvPD6}vD
zUa$gVC#>#gO_j#`V+_ZyF~Azs&<{$Y2W;8<)Ivs7iNc%ddi(~8tCI3sEyFv=lRsr(
zTT8zi9L1K6SUKjB5C8z=Lyz}J;iZms5+dKfh3z0n9z5%+(U$C4L)%-&TlRcsV_7yi
zkBCo6?M_{#yI1dWcL=SJx}+8X!<3ZDXPjO;46dac;5~G8hvzX_V8BH;zFHnzQF1!U
zE<%3OKmZ&q<=z_$+oY$#IVMHo@aSVC$Ky^DukNZ^$i(zM}a;A^x@G}QYzJh_Gcl2%4H382PTMqTL)%-<%FTreiyPfvG
zx{oQRm|(EJng(zAK49veJZs+Tn5cx$aRliLm#K+cTcFs91G$)T81M%4`&fIO`%Z`LuhX7B6n)uhC{zp|t(2P7j)vLrBK1n76>KD)xc7
zQ}BJGQh%w5YOp;XRA^I7FZ@#h)d`0Y91haWBe
z@jfN|cheKmpcx$cd!6b#=kaNvV(}zb=yMoWfIcB6Z}hp4XWQlK`2xtN@MiPs@6(!h
z-M+61<`>w2^$}o14{6Fq8ef{yvQszf@6e+o)CW;)+krN=$>h~1sw698ErggC=#SLk
zuceyxhM?;=39mSTZcumiD{#W~x8t1D?8)Tl1}d=7SMUBqQVS$3s0)sD&j7Src#;@2{^9wMLkMY@|$9Evr8f
zg2fhbt8|pP>u7~4GN;7>`QuXY0{m{+S2I{b*)|@W#>mJ$A$;K&$S(xU}!>x_(@H{Xf!JTpjN$k?E!GfGeq>#;
z29@(Yb3TOE-Y&*)w)LqY)@1>xuD#QK6T$nARu;Bq>N~@5@|qRoaL10^xK`c4=HxiN
z=!(qP&@^4bPjnz{r=@t_Bv}Efk`_+W4~1hn
z0{L!9d{&pU?e+v}uil#}_fFyv4X`lQ#OMZ*etBHZU~Gy$
zj-d}@-zCrB8k3FLL0p!ZHvZv%jGhuL@eqZBU7IRGy~MWGXdxC$G>=si|k7+PT{K@A+Skw&&ZXmlqYEnh?}lsD{WPHh788Lm_k1
zbA>{DvAaoRQfW?t*kjOW<4``b{aO(>A>uh=dW-VL81au`FKy-HEbiJ58cond&cKV=
z0qf0jk{EGfhZ#Y?vA`}mb7n63oavb}ugS`PwUb`C5y48H2nmvvGX(y<>`S+G(2{nrSadJQ>jbFTg=?f6e+Ym4Uim5hX
z-oxw8I=Ne%h|cmuFi6W3uPOUaTX-P3mQ;uw5)0{Q(}*=THOZ+F^N6%Lz$=V@tuKCOvGTUhr4_zynWJt_CnWF@KB>A0%Ue-N=48Ht`CQWrqwUGz2`b4fu
z*l(r%=f;iMIUGreXJ7Qu5Mk~Z1@Y>O!!w5Z)QEYvCOW*5%8o=
z`y@2oQB>um0{;(f&5ZxDa8VqzWru@yOovEt4uR`77B3pyY3thor-cZ{s#nOZrh!KN
zUn0>uoCnaiR~Moy>{!~XXjUNM{6nJozeX`XA-t~_pm(_P90jgy^FERYzA5gTa4!d(
zppQ_D@9mP{=85F#a-R|a3(6Jf+~4xG7t$ZuX+M13rrzS(9hK)W#sP~9<_M*|jeUQ0
z47Duk%OpWrZ5yn!%Nqw@{l}itKqNnxCxgWCCUn|R1`IxF*RSh-;Q;7;50lXX?OM+_
z6OIOf>&T+)bNRl#^2REsK9>k*A?l`EJP!-%S7>PUv&LatnQkw$7x~k%KNM2u0anr~
zM`atuBwqq}7Uuz43PT^q7c)&PooDpJ121TBSwfCGXOk6&#ia?=cx|qY>49h6_(6r(VJXG5=9Y;xmGxX4+DFd+g1kx$lq^!hr^Osi+OUK)
    gP-c3+6&oP`(U)Jd3GQs;!43@SLI+-5!$#hN1?
zRK|*L(;y%=*a28(pKa&AL|nD1bs0HCMx_Vt0_I3*K7ouqcv9B1noe+^OJV`PV7w3h
z4eQlEdk#Trn@A$(#@tXc@55%*uYy8y|6%oSTj;Erja0=)>n*%EQo$Z{b7fxLJPhAG
z2vVEBvW%kCdqaQ9`+)BIY8}p*hT~@U#+Q^8cDC&NW6&(Udl_EUTB85!O85HIyQXVO
zJiakZWfADVvR!@8ruemU^i{lJn3QI=h9ptj999HBT;
zYfH#}-=`)l35QOk+FLFSWZ=Roh8)bu(L+Y@+5mnQ{;NF1aUt;*D=l%H3cLohc`K_D=C%DyiQwodc%?Yp{ZmntY|d?hXd)a3@G5Gl)qdo?nXZ#
zD;cQspadk>#JOMtRg`-DY;59ac4ceqK8BJ>~0Eum|k4X7P!{dLJZ^#A{>Y#bS_|B>-GSopLAq$cY{U8I-O6`zJ)bNs=;}-H5s^
zAi%14`*zfP=5K5mBDJQKE{s`8VOWifNL#8qYy9!Wh&vdA$9ol8!^$v5f2s=}*ukv>
zPxj#%7=x{n-@rp-6r$uyN-isCb2&<+q0(8!1K|QzrvxT`j67`$)fh
z;r_F@!3DaxNY#3=wI(|5G(ub@d8TzIiV?3gWzrDc-Q(=A3Kge>AN~D{w7n4YY8W7NRW%VdtpojUf?YjB=U%Mp;D0o?Lio@XJB9{uo#lc#?Vt
zXQ&y#as2MWnPf^nxmRq2+|-+pWW=tU7rjW30M5z=^+lZtWBNScUO;wqK4Q{)sJim3
zynujSnj;X?aHQr@3W6+6PvDUcL~Ty*q-Sw@N~YqT=dG<4gqoy{IL7#+<@CjCfmp2e9q)3Xjk%Gt*!rsA#B%i`laS!
z;`aX6;wbFIgxcxFzq%?^W5{v?-J^Lqg@%DsocCF;R-X-#FQ8~J4C54^kBprFs=1pf
z5GH!+pOU6Zkodzph&giRBn>
z8FHt}!cDpfdSxc7ny`yYD(b51G<{>=h8i<=TuE+{DVM8V}wz^SkLXZTo_rgnS6kSS(SQREz|C^K-D!
zzeW?u>egJ!kRMcTC`P)HOAwk_$hEm~5>%Kp$uqp$LA2X-nx+KMA5PhUGReuW^8xBIoXJhU%v-;S{
zgNyGGa}Kj8wgIf8S>^J$X&Z8Mze_-JrBfI0EZq#1FfPo>{jip%WQo*4t5AZyYY(-|
zvbJArgU5t>FcnCnKP-_`Uf6mR!euj({luG2j>o^S-TKvDjn|-1&Vk+-54zVSG)~a#
z|Coub;hYJqZB~kilg6S|C}Gl0Cb_kmGx>Tc
z>B5nG#b=CKUtnm~#;-~v1=G;kr%|}=wK99S3S6dLM*wy>OnYWr3#Jtc^#+uc*)K2S
zNcb|ja^jr(>ouIaqV!OWh`K=vKe#I+r+=mV;EbUWn@HV-IBCs4Xch1rCt<4!@OLoOG1<12t$Zkh3mR?L#
zqofS?=K93PVi_E_5toMQNm@7v9A5!NLqrHa>#rn~+-cKB!==Y^``6&m?H^H*$)-r*
z2+K!DVJI8Wm6mRGbHPKT_GL9+?H9ZDlVn5Cg(-J0w#FtJ
z8cwC?148yK^*^jKb5z`
z0l*n-cBj_jTbvN=e;fo)KjQr^gJYySAT#-uxS*32zt8`HEgdQ%36VnSd!#5F3lpzL
zWpw)#z~s}B%e|%KBgD@`O*GzbdK1hg?Ft};zJ5o?+*mregqJ)B1U_|NJy%EqPms
zewMx8{|tipD%7i9(+(WE@Wzh!B1smgkBoyl2Fb?mc*NF=*-8RMRIXol7M3WcEdEUUr072
zUN$>TrCKpkfv130bePW2de_UD47@=jc-1O}*vnaa6A62i$Ms)8*XrS$;tY~*Hg({6
zKv`EP3xLw-_xk5Jfj|_LO5dL5$xV7ll?quQ@WX!2xURT~V1rykMD4UukO&1vjqoOS
zB4N1smhHIfXNV^bt^ldV-$Ezp_HtUgMa(uNLby>>ba@;dw*uV6$?{1%kq+h@PzCqBdlvX-%7d1jSi
zj)wR~T9oES=*!Uz`(oI60~FyqCjN65ce-_##wR_FN79Bk)+)ZM5$Nt0VtA`oSiy<2
zk7rcxGDt0xoz?knqTw5$+|;3aO{_8OdW8sXlo-=P*Eq<&lqhHDKWAAXG&-r!V3
zD9%sN4#~B;+f`Kpf+%6|@9ft4MlT08mBZs6)9I$A7U!l1&Jo7}8kix|F?%o5?TxYl
zizp*xi>sW!&6R|^MYwY6ZU-$ok|4Sw7+h#g-NFj9Q&v$SKp^H;C1v@3zOBlQHG|-^
zZ!ytItfCynmW)dr2B4$sny%%rLMC>#O4$IcQM;$mr(uet!$zu3|0YY2;=gXF
zQg)_w?tz*SX(L-T0M@LN4;yW!jjG_Ei%jelKfi-6Q*
zrf$=SW!Bvma0*cUlw%2c*WS%UO?QE>zCf${v0*lyrC{rIVy^al;nC2kC0WuOCRpWv
zX~hZ^W244wCFjd=C$C-nBt;
zw^X!6PWy|2q?|@k+PSvJ$#kLd?98A^3{zwu-UMlcsVp?$B)yQXnIDblXX)0&*g=!3Q
z^TRR8o24i$g2S|In1iYEg0iHbjDwT4sN{il%7C)9Af1DWTYz%3sHuYMn1gctXA`5P
zuN7Po6_lstTQI?cva|>aJk`_B(0pBT}U`snE&$<_7*OdBy6l)+$sKN;Iu#;=lM}A-{+bWDW{V7t(Nk#
zGBQ|P7>HDgAdsOz6T0v#_W^jZqvh|qxmgFLjw?c$9*WMU^>5%YpH0gYCV7+tWf%sQ
zAf-y?KndBj90GtEH7h(Ctk69)$2n(w*z;ejgydsPfBw9|PQH{BQdC;K*20I&P
z2sBT55C=+tIk*l4Q@l26VDDxOGloRdUHB?Yqqg49cVMTDatUwFpL9tKe93V#kADz$
zD4cx|)Eo?h8}Y@M)B@5d$PsB+Mv~tYWEo`!(FrKodC*SIemm~2o#^fzJFu~3nFyb)
zoeCBFu;hP%MdX1Jqt`oAHt;Rwh8>|~N52VyO6yJU>?fLSTf+L)
zM`r}3Z4?cHI=Kjk0~!MM@z)^}7Xb$n>K3@ekqd4WAkppGZ5deqv7rH{)4^+FtXM?f
z52J-xyw9&nB7&hnp-emB<^8nMb9d+Y9LY1Hj$VXtC
z_|YBXPB=N1KzF9PGg4&IGuYxcN;S}83K)|*8A?T2H|#L#-R(Nm_O7NMuX1UHnySo}
zNWMY(>R-#10N(}q>THCLtUwBoqjX4i$axy^!loE(hOX`)m`iT
zd)6op4PzIKAd(4b(8|r(nw7+O0t4g`k=V10y>om(Q*JZ838};LNl0U+NARnSaRtpT
znec*Uk%9PIi7xU%8#NHjV;=90Uo*T)Y29T)t{Z$;VL=;&l&T(B7OX_%EuE(%1y$Sf
zi}cA8m)^c8>1NBx*1MX$I=>&^nuC@<5@G25ZhU2V`nt#M6{;CjX
zMlwOsvS$kF<=}4Ubn$pQW_sF?&6Yu;UUm_HB`NciJ25$T2ymj^5L~H
z1KDE7m-&tczh^SC!^@I)KCcx?aQ*n@W|96Jd-Hgp!xu?l){$ibS__AZ@1fmW;Ya47
zV21ty+g%jVoYn(JS1$1>#|KcpQ!0NmcyNYVD}`7wg(5|?ikUrwDF)MLNSGp|#eV{TT)
zqAMKd{>A;XL|44;`=qP_rM}NcvwzsJU%-_kYbIXYYz1aLi$em$z4*xqE&VzQVx?Wd
z8AGbs2|mybRd7W$?9DaSpd$!Mel_pQ*?MZ$n4?r;G#&Ulo^K(sbdA~45{VM0d*4n1
z>Q$08%Gbujy3p#K+{l+zb(j^%bCq989CGvUhH
zVI3N~Qfd~&?VE7^pL4VwJr(Dh^d_+V
z4YB;R5pqC};J%TgL`nuf*}vIJKi?Vv$_SES{*jL%Vf{%%w6|A=AaJV
zyw;{D6>rW_{0w$}0|?@3ApRa`BH!gg<6m;GJvR&p2ga~x_G_}
zBxEf7tyc(KEKbyXrB*HCz}NqP(f`NJ{rp-#s*}uhVvlJNIizrAT0zfD*OfQ6r1)cJ
zer)3TdIsC}mBF?w*v<$M+_vqf`9wpsEg^N@4H%eFU;|1{L?MEH((Gb=|3xD6mQQOd
zuZMy8aAauTRH@LW?yg5YF&uvkK2wG^+OP2(xtMZbR+g=M%8{Q28`Qile|D6w`13eb
z!kkduJ5@%IOB@rr6s|Ym=))JOEH4L{Sr<8ibnvvVQ;2T(BlPbT3P5noOT*ADqM=66
z+t=4!{qVQm7--+p{%-9o7}uAQh>e$UIhmar3PoKPgGeekO#TrHPIV2pFT3IHV=bV6w*Zx<;MhG
zxO7LUdDw&1VSV+Dza0R#WZchi_)2@(Xu%Np>xuO-_=Z@vM)@;HlxE3>6f{UR2B(ca
zD{QW*%ca1RLNPJpOqmz$_>_9@x2gy#0G}HBkv#H2JDsMYYk12k_K9-wNd52(KB~SS>wD^|TKp7}f?wJWgW&a1J)nOw=ILy*f
z2KR|ryp0oD{39uQ*wj&$Uui1LU1uE#U~to;GxMj#DYuArlHVBjwhPe;1KelmSej4|
zo(SqP=vwi3UmFw^aJa|y8q%qP_nTN=&OO;US4gsOd;
zQfwWYI{qb)6Cl*iF!%^>cSuYVfLvJsw!3P*yB(9$y9Mn!RR-n&(a+k=Z~aM$!#mlo
zKV>b)7tUq~;iS22Xz?ZiT5x|RAr
z@fACP&q{pnfn0T2RC~3Q(SK?rHbNbhsEZzyJmeTZr(g-ya*@nI!Wn=r|bBy7px8&7_rX|4Vb#rXH24}3T
zVD9$cNc_rSOhdvV`R5@ybwxFMaUckuuKZ_LKPQb68lc7cM=eZKul>X_itL)9GyXP$
zOS&hx!gv?4h7QWesh(z-C{J3;F^+IJ3_)8Xp9lVu)fOwHF@j9$55jg?TbflANf$s<
zGEftqXu>jGDyZ`sEOrL+S(Z3^8PUXl*rvNV*1Y8i7t}n1CPggs8UL!Zz#lc*9B+kc
zUlhxg;xdPd+0&x8qDDJB%No@LXHYFZ%kuj{G+%5<$4wi^B@BfXU8aPrzSd4q>Zy4R
zPHSht3Hq&U!3^E%3G3D?knE-C{gkw*#X)ium-+maw*3&i-+&>uy(NIr-Pt^GI&%nB
z+%C|Dk+NTQpzc2x;-o{!&-cbqkk0PLc5Z(k$~<_Qe}|KIot?O@pnc>l`^Z{f*uj|A
zwt#M!0KmCfC-8gYDCVZQ^2?&^p^?ZX2sv|;e>a0RRK|i2)-w|9B>kQPsj_wJgIj{I
zPE#K59g;iVEduuJ%%luhr1jmDOyK7~>+aKO(TI`8Dq{N{{o{_b3}7IH2V#EUFpohf
zyY_NlEqUKNCiWGZyDW#FIwv_6Fy1aN<%KKFNOooVrXF1%HTbd&>c87(BIkr2j@Gw+
z5F<*f`TY;ZjECjF#w_LR2^=4R)c}r3&C1h^goTGS1@##m!{$H6O%;ww#@-z8eUbUU
zN}NQGUrdrmgolfpQ<9m5T|`8bN8;N{F>{KEGBdMqO0uws3zGcrw|qbDzebFmhb`sL
zGdLZvO({tPV|e%8cK{^i(h#MSc6=|yP+z%~hBgu|AmSY?fNtSvcqJ}XdqvTN{vUWP
ziKexwnTd%-Nmh7La4?Hxbd!p2iO1cvoB2`37%p1_FiocuOv@Nb{>f!XIsn0mb>h7R
zM5Ce*v6jd)7ruFb%F@Cx$hy>D7Y^ST{$dBH$F?x$1mEo=5NuZnWf?>@D&}v^WyR1q
zsymfZ<*!+uRU2cBUk1*}cF5Op&(JL3hr#=m~<|^=G78%?D2FeT=
znd3bU$_*pt03o?sTdDf+V}Kb69TfP@IHR#?!^{I4^PO;A_HH6Ip
zYuZty=hdv0#Ebuej^Xfnmm@JkC(?S)I6v}vk$)VH9rxzvO1+k4
z>6X9wEuj1rsd=+PFnl#N0iVyIAJEtfKW+a157PNNw+S2$fs=)kg$<69QbI`*?*9OP
C;oYwQ

delta 124306
zcmV(?K-a(8fe^;?53n-=0fnBxj7M9smq>Ia|1y>m}Tr5&;n2>sLxmmnh{<6yCvRtjD
zl*^*tw_S56x^m+ux_|4c^`l+hHPxLTAKGgFY4zdqrx&k7GA{dE<#98|hFZ2=-j%y@#h3e)Fv~8m0CN4pJ&|1`%w`~yEckFXrJ2YUmpf4Y
zSCxON_xZN^llWark;DUGLSU5U(z0plXy`#}|+-G8BR%(#~}r=|DdrQVgG
z8Oi4#S8M!OAG$#Gx++$}F0cLTRhffS>149tED@1;DVa&LMT(~s+H;>b#EzEB`ih=n
z*_7Mz^GYiSqA2~87YWw*jf~p}uQ~A>KHt|JTdg%$%a@KVOc9r6SzgsmpmR&#hV?g1
z?S3xGG7^unA%AN9Y*+hlH|6_8s(tzBp#*2r_%jtXZG=N;uvqzXsD$Gj;CJK4Mcv?}
zc~{kYn(&&cin7HifZXM!Qz8AZE%Qwn9$Nn~F-7kVOEf69dD~V+zV*MqbvoY{)eUAG
z!^hkvmMI=o@T%V3c!lnPI6ONKz#jmr`!<@!`$Q*iUw{382lx2J<|J9Z-1+z0s=E#l
zzPZ_g89Ez|<1Gk#bC``5WE#l=qG!fRZyPUj@QzQ7URIkjZ~f2#0#9!+7`$!bk=(0i
zTXN4fo}Hf`v5mJe+YfE@NRgP-DIQXI$-rv}2MG(zMO%H^zX+z{=YsQbTq-r@GO#E!%}i#dJpOw6SU?9$uy*nd38gK|
z$elBjtpn|?pM%YKcB`(P1(4R92@~)hj$4t!qo8{S{o=Lc6Yt3^4!7n2U>5ZMKH=u*
zo5^u#)H~82^MWuMpcaG=_}#c;)RlLp{?~}zO@G;y&92%tD$!9#8v;hQ
z{O&vdoYScUn0uWL8tbm}F2?F=yX~DDLI-xft6;F@_C7+aEt<*)#&QzrJxKp~TH^^q
zqO@P8@^M}eRIajwDUfz@dqWNs?6mbih<^`G47H`7t2{cJiXy(pF9J9uZN`$ulHHI8
zV14PNfYG6iv`3kNT@pDLOzg65I)asTwupxh;`qe*icQF^CYpCANPz&8`V&Nk$&@3*
zgSAZ!YGUxWz2-I)31N(_3$!%jV%7%fUq3@f@Y%p7;A>OQ#;vu0_e#tH837|-N`G2<
z<6{q{6VlQ3!WajxS2Zp9yh`Pe574|9iCZvA2592yuy_2m=VKOmV+F5j(jc^#G2_Yj
zknv!=soLVuw*JNQh5=$D4TPIPEQRMHv{*g)HsnK@Pd#Wk)>Zp3S09?T+I%p$ePt)bdv;&CZbH)gA^_#)zT&`
z(?N&Y3}9v)@(tx8E|}8ye!{OUy1EHEg~OSN@Elfv+G4`WZ!<1#
zov?HWLwBcPB9Uw?lJb%hJcq11v-%G_5IR5+v-Dfc`^r9r`$@FBO)j&l1^<@hjp1N#~|TM*Ez
zImQL!x(`=F&N2Xs0N&2fPMLulBCnKbG1*|E)Qn}aSJAr}I%I?92u#z6$~4?dSn%x(
z1;v!ej!-b+A!S1KHS-4_ihsZN>*OTYO&Q`ZrHP5kVjL7|h9W9pjU>r1Yq(z+b+gW9
zPst5x=IpDF`9M(eF$y>)PpCRO&6h-m&d>6tzgGNuV2e<`&M2-S<8a~aF9>ip^BQgQ
zO!AZB3G$OtondPL-}A`t2)2S*EI{^V={=xU&$41U<#{7~e>>w7V1Lt`WdbV&8F?s_
zZhcDaqM57EFevZx-OaX~;n`-cZdyY=0v_o3OyoJCI5+~@+$c>>EhNCcko@uT;*X2y
zlSQWBEEmFB29Hdzd_gw&=GS2fP5u?BK?ZMb?_r=*!
z&aZ|BcvBYncI(TjQ8hj)tRJRfXR@aw{97i;l*SG+lop6Q^A0j)DoMX2f*tv{^o)PA
z5FLt6A>nek`hOY^7byt~P}W-UcX9}&yiz1{3)`p^p|Nj+5=44R>q<<$eR7Deq_B_sv>+F0vqnWYh
z?)E8jpAy_hIl2D51Czm4T@o+xVDdSqh_{074Wo&cy
z*Pd$H=Irv{UGpp4$oU1^LwpbSPeaG7EJGLc&(d{OO*(W+t9?ip;+5LaMomY(YD3o#
zwB!&rp?^!%*!Sb|sKLV((nM5BzUN%~e)UaLf83V4*1v_zO^mj68|GYd5>MAtI=Gws
z#^1sE@5`bJU1z-ddX47urmKp>HgDY30a47GPi5bL*IqVkwN}zD--PaY)!(M`^q70P
zy-2dyp4#o=mhmJlcnb8rcg0}tikjwo?}~AMZB0r0hxU8zE;u#^oV6Ib{`PNX-_~*o^wcFNf$_(RzM?1Zf%M%m>054?~ln;gaN=}v_ME@7C
zVnS(?upTD^H#Req@M9@|9NCiF#_@f>0*_#oIhy;hDlc*DRGhd{vA4>ulJr0=u`5(~
ztU*#M{e5~4fCF#>cateqI{%BEmQjO_lZI8w6qkuBt!y%<9pYk%g)iIDlMnoLQW=n`=(yF_Ol
zqBj%X9or6A4(uN005>fbbePQGiDep5ka0KZTR(}Y_#6=ZKtaMaQiK9WBv2YeZ3$~`
zKNOQGliDdpv6yYk^~w+2?^bvFw{qiuw;zfy@=$)9P?gtzukWqMe)i9RJI`l}U9m~v
z)i6{l;2LRzW1(;=5y8zgM-q$n#3W6d1qyMc+Y2tECJ~q%bEIeopTIXp5@2!yt4U`8
zJ#q-O>Xal!`V0zEVmb@nVvLU*!w49o4rnAn!R8z$k*1wuGHk~at?A$ImwHUb`XRaMz_1!
z!jC>y>-URd8Rk5d+Yj~$=i4>>!g4>p05vE1L*d`N;)C$R*akoYmgT=rw8+EvyV)vi
z;#A2F&5O5~@l}|x{ZRM`u}V6f=pZEwMVv5?)N;0et;$cJe~jh#>&iZGzm*-?TI`SQd{NwGmT1&d1+0176}2r(kmBHaUY&jCmI(C3QVi~y361Sf!83Pcnn
z)lku}LL^}_#3C3m?R7$;r<7XFAn?E-U|@dcKUP5vy*`3-d~bmQWrTmcD*6e;8IcIo#igvhLuQuCicegEn=6+zeDOV2xAsVwmDMa=P@d!FxYhVi_
z25wC>T8zR@V1*p3sY4n9H&8C-v{dU5P<_{mD3r*5m>8b#0f!N5H{8SDicR^jdId9p
z2D9z1D*X6Q-l`o!m9X*L2*jB5=Fn1{YMifkm4BEQ6$C6oi>pi>1D5N5
z-Np}8#Y0tWHqDCJ;$dBZ8*nZG-WTP6t;Z<*+rO)Mxtdj<{oB8a*=Fk>fBAhkt9&$c
zV8yv+h-nYCx7#fLxtu=)NvW(K*#lvRVs_WAW8M-lJMIX2-w+eVmJu=A8YwW+hSfa&
zE#-q568rm6+Ei0Qk%W+R3IVvrvb9-%+g4MsXH9IIw2&C?6*6OB_(TvV>U}$?paM|B
zO-a^+0(R7iF*S1Dn6(gcsDKqu#X=N9ao{K1I;n2kCVROv%5oV@4w69XDM8!>Bj$zcwFL8gT2V$%&9Z2EnJ%47g>gMiHYxUAE>>I!Np(u{K4*FGgIFQL-r7
z+g(f+#VGC{f#z(o_`=W}kwtNKSsV@d3(4XsqDZew6zNroBK^aPBK@kO2w@`x!7aOZ
zh@!-1#Hjz#m>%LlQrz8SPN3A6i&9F>;GV<*
zu7tn897^hZD3L2c>C3sK1{Izltk_^21V1V*;VJC7LKKUyf&Q0R4K-{vj4g7|bTsum
z*7N58^(RtL^5;{~3k8Q(LHQp*LCK3MDC=YcB^+xI4G6JRra&cb`zmpN+dn#Q%-SeZ
zkjf=$#4yOIK$`bw5+PJEBFYtUk}+jqWylq|YD9k#j0lM^oYGqJTdGvlO2z~sWV)55
z=eU{hbAr(d5df7TNyK=ON}2Q)uOE*Pg<^sj?_l_iO?8P-s)UyqT;D4w?$`vtGO$x#fi;SMwJQS
zYen6<1D}wY_%ck3|UXO4@lG1lTHDI
z2ojftJ|zvOE%YQ(Qf+(D>okY=B7J+2)AypIocW=T1BY>N-AWLPfq;{4xyTGc^q4eY
z1grBnE3l}NHyg2k1p2mhiMXp_Eeh#slK>kCX4MN{a80%G19|mw>{>OnZ`FX&y;!HD
z8PK%pns~#0n!#ieYFxK_=O=3CS0>cJAj}y=K3uXfnxGCGhCWph4$~*cbF*biBw-^i
zFOfGq=V|o$r-{MTPg337$7G-h9F*;Z-V0qF%2b^fNzI&p+KG{Br}}%UtiRWtrYZ`q
zP|;9){56g~HldIgDs`nTKW@J1R`PgoeGm?;*1lOIl>cf!2rssnd8kk^IM($+P}zcp
z4)g~x$-u44_JT;)de2ed%uMo}UQ#hl5Y?i+$o06)7M3Wl>b6H;gPpDJiaPRf9SyOI
zr+}vdo_Ab-JPj_M>|Hz!E}nX0Nb}{|2P*#Up;#4_ZTGP+hv9`^(N{O!H=JQaC22#|
z7i(*IB-SSKI%4gZ?wksDVhcmt+{BshyK;nD6tj7{$W6fsYZ7?w17Pc&5*J0q
zX@Bi&(~W{7XizTGT|4{O)3qkyv>)I`cXJl3uRVw)eX`L}jix_0{@M?wm`vgJBvXEY
zb-2trkaRi{jB&A_?MR>^#2eoww$ey5adPw6aLds^?-#kk&IZ-)Y+&t1KBI~6Y%pKY
zE!}f}3+M!8>>yl@CRxgAp`6aD{uX)g%7U`zov(+^AKi{Nd=?ptD`x80p94Uo#RR!t
zTERJX75
z>?9;FD2^wYnGh5t)lR{UiC#$UFJ$5_OJ^~EM(x$LYQ6fJq`YFSm**$pK-zClLVg~t
z=VP>t=nD}{jENQ1|LJv7ygA~UNzuLu<++Z>*5P=Q?|W8upjpCW`?muPDm=+^{-FkMUfOnkDXb^CqZDd(dft5
z4K$1FAuF=K+!f*X&#SvP@371?(t?O=b)R9Tki;YtTp~hww%TOx^M5WB&COy-Dg1nV
zD(j6O_m5TQN4sXRBr4ySZFthyUj`K6wRe_9TYuKSE`Th|-S6vlUp2M=x2d}IzUxfj
zUH;-YHkSUs?j^MNu=?wpcLZk`Bh5Ln@-8t$nz3w2Bodf;dEYi|Yqn+JEqVU;vTv)8
zetf9Ok2n65Z(+{v`cLFyE@EvRagQDwp+VLx2w^nq)-{x8WAb{u69wmMC4{`
zcBZz=^qrrnYX9q{g2Bdg)kByZzj|;+Zhr$+XZrA7WxeoVAcF$pl7CxJMPGo){ga>Y+k%PpB+ylKD707#ENr-uJW`f{qN{dK+fROS(>6zr
zok=?g*n;t(?zw`%FM^-kH{0zfynkpu0!b6xYTeXb-|p9a#KwgO6&O!9VgWU5BmsFv
z0J{LrKbk{Z1-Bn#*V|^{3@_-&@E+yfINLj#G*F)UC)g45DYWr|gJ#G8ZC=-)01-@qlW~%LY%@;n5KiRKO19f)9RnO*_25dn%nCi37gfls?8UP=C7EBpwt~
z=4Be3Q3vySXnJ#R%6{LPu_3%hAkQ2%q7^Enj4}JKl+GyPltv2hlv5^4Fe}9Lk}v>>
zn7xA0>4;7tMS0}(KEVh81`on~OmGm0i-E%j2xh7BZ@fw*k_<8xiUMFeefQn!?mu@?
zT|kXA&6vapW16jZcke$G*?$I}+5A8OyAF{dD&aUtdh1FftMtNu?%uIN6Ux
zSaOg_Q5F@v1c`&1iH1j7Qsr0{h!K`CCz+#NCHAjT!DzM|QGMg7=me%Yr5ho97C}%ejm$wnUpi8dguygo#X#_5(2W@B
z85~LTi~?SYDHk##qA^Kz<+TX{A1ZU5}H}CVLI5pC{lHb0&UW4M0O=JoM2|PATxjYHG1q_yBGn}!OO$c)rns^9=F~{CkXY^r_T1@qu
z-~}=m#m`IleFzfkc(
zm;z;5R)PE{wX1a9npE&b-Ms5{G!VYI=o9FyTiNNq4SJlMr1Eg
z%tUbfdV(ht5~fglGJR75;QY@$m%xkPdu@ES~PUS~h6n{v8r%v+Y;jvoVGx4KOa#gri
z`1ju5-<<-q0AH3Yq&j;3Uh!N4($pVNGb=z7vv8R)@#`$z!CxL+o72@-(cuBKZkz%;9S*qmv
zfUH9FA+H7*X)E^6T)6KZqWiJu-8e99zUVq;Y~)WGK`Yy=c1@py
zYizfhZq^d|;qAA>Ow*cOy3a$VZ62&K!XVYO&BJ(Z`#I&pd%M5WVJyZEi>-g4n>Tyb
zfpqBfz}Pl*!wd6f*o2bj6|7nht2SG|cz@>!OaJ38s?+@*EH%PmGp*^?;C}UkZMcq|
z6(%HuopCM
z>_-4}?#2`&0>zl~ju*~^WIN0w1&Xuh%f+oxfZ4o`g)Y>Du1=$B3jIyfP==6;s4_^i_=tHA`vByez*g7}N4<_tfcA2ho2_ae1ay{Pq15r}B5L6LH
zeLv%~X>QvR(_7fi1kXluSGX+p!i+hLGu;J8i^Qa!#EwZp#+1@TaL|8rn!-j+G4`5*
z*6uaWbncZi0;XKEg@HX|ym1>*YaMfhA@@E>m*`OcQEf~nPI&`Ii^K|lPZZL{Jg&@rpuY6+#fN|WTUu#gR-8iPGC=N%W`M|Q@9i6lGq42
zBH0R1a8Q9_m+Utc%T9$VEe<70i2r+okoaH#K&$z`CGBNTNR{vmoxXUkzc3|w6lgH8
z5e2fJIm@o?_6cp)3L&7F4T0i0k1$Hf7b+#=Xo|m!QY5Yk(fjnx4N1a_jsqMjz=y-R=
z%u)CLwS^b2-T1DnPZ$W5@HkxjB#J7W@mQGeHOE;^4HPzK7IAmTi1{yhUA=o8;k5hp
z7=F|EZval&_R>vE#tCKOoEB&0lZp{w`gonmyYqg{S9X7)eyC
z=NOr;ETn3er^CH9M#34w-guif!y5wgwP$M$p&VF?d00rHLa<{3nD(YV{9(1uq_T+6
zIL;iKWN1o|^6a;JtHDHWA)oY+V)A)@>}ErkS2p$FIu$o
zGc=R%V<~@HTaVnf6@K4ep*~fEdT8-3rZ5n+wNo3-Lz6{g!1052M>{KfMzfJLUOT_O
z=a7dQX@+~rrU;s%FFO=ohh3)?u=!Bfmm8^u*6@lPgVj;mw
zV2lVpCsr^_n{kEri>0t-{sQSTZkbgE2&@WkIZmIwJU{#CjKdd2#xtO1PG&|**vDj-
z+p~Xn?~7~&@7`nuldjBeJ>@n7MoJiLvhUCS83ZoM={ad3p_p{XDTHf
z>g{3)Cw@zuIe6u)Ed7!NQ$l5maV`Vxd4hk(3MI4Ugz810lGEeyf@$E7#Pl*LN2A1}
zHSxfyxd+fFt7B4u+km;SA7?k+G_zZ%TV0nAe?Oc3EV!B5G+
zjAnLUB$N&yZQRuEze<1$XN4dmkKun*Yr(jHZ?^Ew4~tTGP*EIUAr3a%T|eNs!06f4
z&v2{{m;^LOm;^Jsi=k@TdPQr5pAKW}rlP0}6Bp>5c|G9QxA>>7+pyENX`5~Ha|rD~
ziE-O|@Pq{jan_WXRt1Ehz(H4E?%ICn_Lp%Jlh<9h>%y>VS78v+0A;5MKs|q>v6063
zR3SyVt%k1o%oa-!dj73HFfi!P?2meUP64XqDjamoL*nUR*bsu(Kkb_U%^MI0$aElN
z-vzpD_4yzhkN>pR(GkbAXF_t*!6LBahx&5x+QDq88Y)IM#Su3gSOwZk0bpZUSrR6l
z`$eqIeR5T8!b3mQHwjxS0yKX{0l=gxVP$DYQ`2ZGqAWqa`^^f6AoH&I)Idu2CA=X{
zYkGN29{`{Bz$eZipEHsvpWYI1^2b6KUcE%x)fXtd&@|7-O4bptmQf?j?2E;)Rk}m)Ij$i;l
z!Xi;hBf!!og7%(h^AMJlIskSeUp#
z&c9G{!4vEWfP9HZov!`iLqZ^$=|k9iASS7q;mkHZ#3Uk$n;qpL8gd`W3^$WJ1hEH>
z8KFlcLhD3q!a4}&X3p9Rl)ecW-rXiGoMuWLNgOTG@ci}m#s`0>T~Ao6qbUXjtPt;G
zjj3}$E*uyi?%??!)w-+Nb)Ad`N@5;kNW(sX-we&orl~0!gm_5BYSoT7PNSx!ps+^(
zqK6SLPZNX^awn@o{?CI@-z4)D1&~0hN4!j!2UM2!Y34E8>cl((7>?Qcv6n)jsmq75
zL;k4r(Ed-RhxdQT8fsxs>xAe}2@}E5ME61`y|u(D@(HRfcnX`q6OB9US)_C^JJ)Yr
z7CaC)?-znSG4E5U&%EiYWJ>^YbPL}s!3CTgc`RjKeLPJ9S9M3(KD1?8pgk+XJVT78WIxzm@uNO<;
zEd^a%Fe&sSuuDI!+>-Z|TT!`3tpq8Ub~cm7GE?yTS_wc|3VUBOK{}||WF{fg5YdFc
zPDbxi$%j~c*nBwzX3#$545ExRy9Z~HvK0uTZZwtO9ahdT=>#heuyoX?oWPPu*5t0_
zPsP(2XD5Gn`t(SEoftD+inN+)D$KQ-ZdWSoxN#T=27nPLw{_J{fGPGP=N2O|@4|+j
zuQz|MJ(#p;^{mMyPWsmN5mv!d`yE@wtkwC8O1$Ilkxev@gqW$iz^e~_6ox{0F}-e+AIFM%8EOTl{JAEUL|M7u
zp}`VTOMDEKB&Tvr;wVD{Ny-4EiLcrf=DPd
z1UQ{6IO7Es&ws$uSLd&N`{K}3!|?_?3s*yUm=p+KPRa>|J!{9fqVm&6`h=AWW}J99
zfyRGQOg|=p9;>rd(04wqfkev6$K7E(wuhzi8?1QJZLC1hBrou=@Zr9F?s)nCD&oHr
z;Qk{`ZK1ReO5Ph@U!B&|M_rW}yods}}nT{r8{Dh8i&BZL7(aGo+!I^WE8@#b&j
zr;HHhX=14``D{~lYd~Qhh
z;msWl!BK^f5g{ndK5D4>H~4n(mb#fCzLY1eLEl;z6P@_>;VMi{`k9|6tmCV&o*TQ*
zGH7~=r?gU3l|15=<6=f;VJttTep`1
zy;tFWsGw%nWQSGVH|zHKRb37Hj!b{>!T#;YL>YRDo{P6Bbpd+wF%q(qCM>}Sx1h-~
z8ZMqf$BP9Uc!lRH?`b;f1qM45To{s<#88KB+27ZI<`{O=Y~#me`utw>RI>#BIENI!
zu*4US+iS^<1;B&js$wNTB-gV8TM|a9s
zpM@bR%V(}Lnc*9dnIa<$7N_#n0;`x&tBcl2qnU2RWQdvhmq9evjyPtZ!EzBAnoe|$$8|Ik(K
zi?i
z;CnwOI{Eb72l%8okVoN%2VC+(?R_;+V>}p|b?hB)$7h_THcV>l(R9xCaE={m7r4wW
zjS~^X4+;`lT$Cw^KbU5!e@9pbiqL6PwU7N^$HOtcxzkAzeY67}DYb~~4XF#84d2cF
zoJeIHt@UDJ<8UML=mka0BQu=Zsxx6mHBzg!VO8Ds&Nv<(`}0;_`b*JM6_=7V-&A9D
z3-hIns;8N8z`t>yQUR}SKdUP~IvO)tlp%$nBy%%uzwNf
zsn}(S0Hi|anHXT2`yl3wd
zV&Q#30aAj^(qetIduq^;ycfN8+@<6)({Y)X>*f2Y3ziQTN1u8aE6Zb5+tT1{kJsL$
zOB~d+6my>IU6=)2(z~#z{_Bs9y!JgIG{X&9MO9;df^bs(G3juDF&0OY_xa
zWFWX;=4(p}k;XdFAeT>ECvfR%D<47bNpcyc^?YbQ4RlnGf3iGxEE@TJL&k$lwn$~<
z$~w#AG-c~SM#6PEs+^mS-Z`PeLF-%oW{_)8m1E`1NPWvw&#EvD?=@~*nn>$7%~oh1|HP}V$esx4}QQoQd
z{^^^oeHf|%GBt0|^xgY$JAh;wha3PJd^8*9!egF-D?s}Kt(sMC%Na7Lz06-C+sXyE^N~3p3DYXhmmW+F4Zf8!
zlRSpJ#3=R}w><=Uw*}8*R872|DB`d~Fw^mzo)b#31szEiYjt_*HeduvWvW5Z@`|L6
zPJQq&e@|_K*xW-bxgZQ$7IuNGe4!BK7m87(j%C62J$Pl1{$~~;3IXX1?bx9qMIlj)
z(o?2_oeCT5N^ht|03>FrT%XyzO#nqMV_oX?nuX7xbnG)F`%KCB0YUKDv_U&0t0#TX
zI`|u`Z@Lv_s*te~Yz;8QG?9Fe*?m^^v_ZoW1hfbNzrnzR
zCM3X?W*&2aAa^nmd`KcaM8G{A^(>~uW3UU2@l-LRG>m+;AFI*UkhNcUe_lmOchn`k$7Og0{I2o-syPxsu7b_M>M8uS
z5IO%yw)n7T*eZNt8OPZY^9ojbflRh@V8^3p2t3yFcgZ(EZMQVM0OEqd1g%=v#<}xC
z29rcx+&~s3Xm`p_!CT%yzzxMGDK_0+5SZ$er&P7+)rXFD?TN2z?poecfKr2DaF9|0EF?^uw6@I1n~9P4+P1OMS5=dLJT!8L|{Su%Ngp$
zj1Yde74D}SYNK(>0-66!p^|uw?pSLn!4BmyPeSz{Xq8qGXI$`(7a9QWwVzs=d9ua_
zD>R7#AlXBA0`JFrbv)*14zmNKe|w&8g0?-u@yD>qC}qRQ;Ay+I=USAthsrVaC(v
zo}M@s$5mggo6p!HFjAB;6mqW|OUmC)T)&4}d<$F3H2Olc!Uta->P~$9!z&Hc{vNB2
z?MIU(M8GI8v{>Z)I`FR>f1r4L-I?}xM8=8t4~B^x=`35Q=Yed&^+_J~d6-zc6^1ZE
zd9!?yK?JYbv#D-_%@^fM(0g4u-jp5$aCcbk#%sw+Hh8ojpeP}iF!&f!|
zJ|Z%G_>W0}FA$-|VkdxTA&(
zh>TjcyL|TKf24YT6Qto#OY4PSgsii4CxzJuvaBvhY@b0ZVoUhK%{GMZ;9u~_X;p9;
zm26*4B}=cKcS#~aU4L!pN?%XrP){=Vh2JhqN|?4RDODO$yW7l0DZsc&GDE}Tga45@dX9da3WBxzkPNK}J
z+})`pP9*b$RIDcb>HgjS0R{h37L)PF69F)j0TdJiI5RMl@M9={ty$e~3VWhDXt<*{x6J}|?N|P+qCRyF?R{spD
zdS~0A?%H8}ZFCZU^!C?7-P?+0X)ty??t8md!C?E3Yn=s_fkpBg{^;&`@3@tfu`lbk
zZXXyt*zwze7e20&Bq;m3Y&Z7%^>u8L;2-N;1>J#BH{GGB*8jTw%SpQ9B-A;RjjLL0
z$qPE2HhlLMOEb3p-pqtWwA|Y={z6PiSmfAnTVeg*C(g@%!Z_0&VbksQU2EHM5KMDg
z;5r83S-iOXcNd7&NfO3Iw!p7W5IesjjlEJmrqWiP-`C8{t7CIoF042JL^F<@~|W9!-lZ8XVruex@21J*4K3YINBa3qBr_Pl3P-#@br0m!BWtkzNB=#aDf<@SAFx0`xjHVh_R
z0S_0quwd)9j{~DV(*&kRj6`8kq%%%pwMw-LO~!mY5zA#uzausqaShlnT!Z}r0=(j-
zXPP*Z#`>mg8G6}Jb!x&KvYcP~e>c>{Npp@3I96GIvAEqcc<01tnTe8NA0m@am!zhTVA&tqOWp-AV|I2H;xLjVv;Ljp0If#rB;0N!%QSp-2M1?BXQ=BEKfxK|OJ(!W^^nYNc?Tj}crN}VrUIqrw#jtG-$F|qy*SZ5rp9sfsdMb>KJmC~b{dgGc{ed(S#bP4>Hf8@{W%H4LV1WODbgQ)RM7K0U$E!{p
z1ljMKXZ{rL{VWtE5gxEqF$-PW)U8B+A2NzMwBZ#`yOsmW6rCt4!o*A<2<5AF
z6b#fG+q*f=85Pk67?BR2=yA-
z%5N{RmPBD1WnU%2ILX5}H(w{hREH>!UK-CA@NQlqCevY5oM8Q8A`@qSDXc&*w_TmJ
zc%+e9UQn@t0et)+hr_{uxVTok@1lI(4i=vg@x9poT%4JpC#G>Ld=$OA5b)^H*$>s@W{h
zPH_H85uAsad0{v7@DxRV;NE{2o=cu`$t^HeeMvE$-lb=T+~eVJFI-L)uzLe?-_yt`
zV%v{(cNk8l3^A$~RB%#%&of0TL8w=8qEbudc5|xE&%-2|5NbJ%OK?2E{Ms>a*wOpi
z`)a8?bzc#Oko(e1&z6=aGm&eR$66Vfs+};w<1sd%Px8U
zcJZmlR%@GT35%KiG+gY}gKSF6gEP_(qE$L84|vL9rT
zcKICbOonTar$xrcHrEkwRS7pq;)rU30!%#fY+FBUuc^yMOb!0^#=@{KxneDxyGKO{f2dp^g)6)YR8(zM62;Gd6eqdRq|phJ&?ZexY9Qz&
zV!{=j5PZ}1sFHo}h*Cp$SjVd+PbC98_YN8qLn;+d?DZE|$B4hl!x4gRr&S*{K8YnWT}vX{mqM{c`Vq
z(#?wIU7^T-6;(AGudmS~pZq)jdYsw?wX_B#-Gy=8HFmjC>IewJmlEbEX*W5F(zotj
z24sy)S0^HQ%HrwB|G+$@(+#Oe*C7pNCQ-X4mPo(1)xm85C9Xy-d
z&4FID=@1?lA&&Ro^SkGcSV#((SH4@r2t&8CRI?A~OkSs1Xkw`{EmybBWb)w9_FW?Z
z3)89`ORCv<=54e<6tE21G~FPd^9V~**SW62UB!T1+1B@6Q<0joh`K(CNyeJ}RFdfs
zBe8OS9&qXfr{25CMesPP>0CQ=YC	5FL;db7N
zG%qV=Kpq9%OtCf&rq}#ZTj1vA&6zg9%l@85uA-E<_YJe<(;yh<%@m`u;LVPNAg*3j
z)JI}@WX`N{r}W1rhsb7~0+UL6%!e}xrpG>ianJ1?EG>fL%q6eSy#}hlYmlTq<2pKS
zq-T!nrbO!BZhrV#XpIc#2zU@$N)V*|W
zSwN=08DnF`l0BbCpll0ur2OMR+feWK)+s_2++I@ospti0Q$PCFqw;@`sDiW9HT|JB
zFyNSse>y%9eB{`Y*IN?AVH~!$XPzoPI;H_nDXBeEq^tvqV92u7xKCo
z4p|O^VixBg6oGvvNdIbB7rHr@oKi6}{ly`#b`1GYc0iq6=ZbP!91N7Fji1G}n9^;O
z$40dBB>(I2k%L{(zBy9UI2wj+y`LJDLV*0`HjEO#X8(Ovj6w3+b^e;lj%D4*DZ>`Y
zB4YDAH9Dt%t&x_-p^CEAb%M~9DUN8$ERf#Ze)B&Ipvftd@yHVaF_Qrl69h3dGc}X&
zV<>;ES4)%IHVnS!SLi6utQk=xC9$`3cDGF@n;z`RrRjl|xfUB)QuRo>`Sk?|$`56?
zGs*N~kpu|>AMk;25*?x>`sGy;zCUkYy}nO*#5vPSrP1~=;zF=IE22#0EEOu+?xR2A
z-^a@KtF_2gJlb+4cw9B^@5v6UKexZVzUO~=q?yjr3>73yYYy6zhD`;J+
zXX|cgaj>895Gdrasly$Qbn}okws
zZmLcH6KX#N^Yy*pmvMmmV1^UC?_AlTsz*n_d(`1-rNJ@4zd6w4
zW1^CW+YNm9^GQ$y2n9fw6-A`9Vo85M5CS)qkUeX`=KwFF%Qb`sgYMCV+1Nkf-B-a;
z1aYfZ*ONe#79c)j5C*)3G?%gn&hVF;y4iJS+{^~D&zik%qyC$4q|AqLP
zFeKR355WEwGQe~%a4Mg@)&)!QEK-?ZT#(acZ_P;lKvL4@(3`HJ;qJs46_>ct50zi_x
zBktY#V|Kp})LNt+#56h4SPKSJjuaT4YDqX@5g6B-_jkSw01*5lnlRbBmy;vOnGl-v
zTor&2aCr{6x(K)o0ax)SA7{+vjur-YgnAs_DRjI!M-f)ax*aU3FwK9S)*#UHDYO%L
zwy;w|*#f}GPIsXZnbw(pWUIB5I({1{;ere+cI`Qr#nXUR?rHMqLjXHIyI}&4zZ!hL
zZb(RPh>B6MD<|^iM8gcv;EQMK7jczP3}qT{bzjk8DZ%&2=cnxlTJSH7lvf_f@P@&1
zg}lCg06;o)rVI)(@|J%c7}_lhvJaNYC#3IuGcx2ky2+d~nFm}S-oM>!L#aKQ1K}$v
zhBQoXF~$B5W_?pw<V8GK5XOh-o!<8QSxrgk$9G=uRaSMCKnG&QVFXp0XEZF=z~YDwZ~<
zIfH&Myj$oy^=WAqSeg2&>Y19R^KGzLn&u=8qfnP=0wkM{w{&)I5!EKh4Sg#A7It*9
z?`qS8n_3ELOWS{_FN99=R2RMhpDXBV#AVl!K_vpvTCe;sTaL6C9tl9zjK-IX0=wY(
zZT6Nl-qvtn&X=(=OBLOP_EV;Oadl1mK2aFDi_|(5tKr3vB+%rhAtaBK?HLuf&=n7}
zcX)LR=?YEn4atU!esh9=B4q36U(via~ZR}o1TBA3xfGF#3>od(K&
z--#@mS5PEEgBnbk`elC-i3I&R;Pf+Yi~D_oFZZI7yubXaU1YfwqEBC<@~U
zaY#X;Tj$^RoI68uRwSJ@Wv2n8;c)iOotbm)YnSVlR#7>lk}|F+qtQ1Ce=ay>igG0?
ztCe?IS)+V5%32kyS6PLa7B;CAi?T)KXsD)WubfrIXXTtK#VY52q19oHfX}RzjZhB-
z&}eMR%4@8g^ul9VZ=CXw_aP~V9RQE0o!vC^PRFD2g5#Xgz|{*2^)`?pORHED4R`G5qfe2D%dJd6LAvb^dbQo
zi^C_J(?S?3ozq!=B~Pq8a8NGda0%lPnBZtG8OS?)VskjUD+Rz|>pI6p#!(G{^Z0~N
z4$4~tJPSUt$(#U1!F7BOg~%!&u@1UELV1pWz!GZZ0rvt>2aUsmuwX%D!I3Q%D}VqS
z1{W%XU_pWbR-SBNM$DP?05!);fHC19Mh6wL)eA#_GaxE|;JawbW*|c@U@Fm2jbMsh
zf>jkQ=?04t6rcn6oaG4ca*_50;Jxt+W1uGh69emtF+mUPD<;EPkV#k*-$^jE78)c!
zurmP@r(8S_se@xq_4)mjNgoZU*Zc+d&PxsHtN!bFu-DMNJGKN
zVaY=VCN}0F*$lCZx8H4`a{TM<>ZkGfvuCov`21#byIzeK!~Z}0xY^u2{AnDQ
ztKq-5Ki%H(5X1WZ^~i7I^B;b=T`m9paCd#P{3ABn_PR&wAOHLpxfl|>3Y-m~clGwo
zo2yfQl1?OHAA^Af&JabQ8saRaUi-lzV$BdmTp665C#o#~E)!7&V|cT^zaT0wW0~{v
zdb3&HuVCEr_0R7%UteslH%pMlfyMX&_M>QXFKD$*>;;V9xNRSH>$iPaX=r`K8nZP{
zI%;ZtbWCo21RcB8gFs`e9&C4fdB487Si)j|u*a`nsPWt7yUk;(dt~7@KQ?h2Kg7oO
zfZ8WS>V@s!t{$jMI*
zr_nwpJL>buj%I$zWscNb9;C!POE$QFG^AW+$+DL6;Nw$;-kz9_$WulzA$aGhkG#J#
zj)`aQ4Cq|44?=I87kZO@O!oZH8;^uu+@lTtqZX|_mZzL(j+35p%n3hzJ@Zsf(m}LM
z?mSM?{AinwMOz!;fU`uM$pdbHKPrD6zLTuj^Li?~c6?8Q=?OnN&R-u`&*U
zxx^pbk>+?N@$>IU^RYY9P==_>ELmsGz{hL9MH$o+*q1~nGT_8pdyhU*3qLsumxCx=
zJ}3(BADb2Xq~~UaxAT0G=;M%o93KsrA!9El!1HYR@CRo<-)`>imk$pEI>Yk&x8rx~
z^-mAu&(}9U-maFPy;)x0uWnbbKQq(kXJvjeDaVqD@qD{KlcjVY1diIxsizbl
zAuAZI4~#3H7adBO&q_Ix77{!@yLqxf4_TX8vL5Ni=R8Z+nJ`4-XUSTBM5timr+T<~
z@)63;iq2;IpwQ+I5pyUv^-*dE3`6u>(jUq0A5{6|=Lvhn8-L`{^HE_3&Cc4H!pRTC
z&(v(a!Cl)Y%T1>qfSrs1PkrcR%%Me?vn|4WfJK-*FAjVnzdA3H?i;1XyVpP5-M{`$
zUGDZXezpA1?alJl*I&qgPhpoaB3S}CoJysXhTK}0^H5Yk{H@=sT=9Mgh^=<8+wp!T
z$FCNbxme6~t_)5|A7kEderprwAGZ8Wz_qp?cTUFnvlxH##7eC#iL<0d!skAu<MMqQ)+pM{_!`4{YX#2>5Rcmp7^UD8*|9jgkT+C^k
zeAB&KLsqP97A%dCFAa79whft3Ki>CE%Ye^X$kWH5EzBwT6uH`
zx5-h+GITt+E!->+(Cq2LF>BJDG|fjl>Zxb)47a?Ng2)*sPD}@i+3uXLh)WR>zNtWY
zUX3lp^z9y%8YlwhfD>Cts}qe@3(};X`IawMkfX>jQH=n9wo~PB2T(FsPJ9nSQ%C0c
zgXWn%?h;tS8ji=zkQZrnAqkfvnNV;Qu>gz|a?h0wLmc*3aHEsOp@~dvZYW_A=Yu6w
zw-Z!L;yS6BK7cDh3*;0}mK(~Ehs{_X@F8mQwU;Q7gp#oa5?!2l2CXroD&RXTVz)BV
zx5mhuA834ka>U#*C=(anVmL0M3(*L?8lu=M5G?W&ew5$bk?IrWBM^fn`z+q?xWiGv>NC1z;j|nmZIzIt)_mN5W#vh3E?IDoLwit(s|-;!9wsA<
z6%wCoT$Q%k8yp=B(oBpMnzQ&{DonGnNTL%@&3wAWw~e?xPrGD%H#_Ex3t~?+NZ2)j;@~PNaSTqZE0vvz|Lfgwn9D*
zTvTs=-?c@CYRgc#_Ij!QFrhOLxz=s(+*Hx~TAwo;JWLn5k|d*%r>IFF1|BTc;0U=E
zGP`qvc2&eAL{A-)xlMbm&;;BXXmGjXS$2?
zbcouoAU&n6dTN_W5trJxKC{+pBS}k5DMh|#A(}E$vs{Z(&ShPz0Lr;kt2QIEc|%~O
zWVc&AbzZ+>kf4HjO?#D-7Pp3S8{Q3nMN!_v(E623k(q!5#l99R(qk>_jPI^Ccsb@-
ziq`LWx2?Toz3#P@R|q{{>K;;er5>R6r3HCi)q!u_Lk#$Mc+`(l`R7oid}pA@h7c9!
zC5~nxLhht;9j|5<K>|mh*IX_B2M@)8B9rmR6ah4|N20zaCpI83Aa7!73OqatFHB`_XLM*W
zATlyDG72wDWo~D5Xfhx&H8?mom(j2R6a_LjGBP)l@na{GY{@Zy2ZmR5RoA<#tNU~2
z%#ac)DANj>*ck!E>};KB8R;0f05Y;lb~c8#j106wcGe~UCOQTN7FbeJVMm~$vxS|l
zh@mr(3&8Gd4v;r?{s?pgFflN2!jb~Sfwn-$k3kcFkq1B)=xnIsVGm>kP#FFN3U*G;
zv_^(bAHzUfGYeaPAmvAhu${e!qlKBd^Pf4GX=(q=^e0+~4j^S{Y-Q)>WMu&`v^4=p
z(aF*QR(F*YK>Ik$3{#6(kTa%A+I-3LkHNl@P0mxVw18tpve}cs9{xxXx
zQOb{?54-b!e}sJ$!uijf*8g$`I01qGmB!rA>91HB1qB&^jiH6DGtk!1*7(EG+0fa=
z31IM-?c*0{LjEs-K!C7|qvM|(vi~>a_@6ZYCtb+yV`VzlUOt9y|GQ&`wk}Sd|ESG>
zFWcD8*2%)j+38<10s*EL*1$jNo&M~Zh3#K9SwT5}Nik7n6-E1@Qwv)Y(?3-=aj~aYwY6|?0ZNMe+xNo+`?t*u=nP;003858cVlzNFaesv(#zR7e>4Nd|37E!e{zYr
zSX;{(+5jp3yP*H;WoTnz?eX7!|Kp+#{3Dfr;{OqB;Us3^4m44)a5gsoS5^OImvlD#
zXnH|gGi%_-j{Id({d112KN|Mq(y{n+X8~v#S=jz(?4uWrt!#l#P5`!lg#kZm@IQJ#
z3jaqffL=jGUP@Ms`hVKxFDFr3V>=TITQdL?D;vPj(b3QYmf@oZm{?f>UW^~DYyxzD
z|Eoy=^mMj%&L1HFdlzRPfT^7$?4ONfV+YU+|1tfGH~{n_e-kHwUi3f2&IF*B_?wsk
z^pbxQ3xHngZ(;+`%lu6r8D#&aj|_7EAr6L*wf#*W8I=B}j||Fx(?X{jQ|G^}
zjQ_U&>%9HP{2|I7=xAa0k1{ZRtjzHr_^}M9f8a-i^FQ!o87}|8k7c<2jZ7bZTHODE
zA0j;dfgid(|HA+7ZG}IV`d@bz!{5#R@2A~gtnBP)X9ZNZF!^}W|HDPr(Am+#U7O+K
z!eRWdfBgL4f4cvbfb{RX@*mMcLU!(6v@9HK09qzaRsbX0#||+vGBf-9SFG{B9z1_t
zb|0tm|M<_X4*&w)fyS^)^LEC6+(DK}4I#z8qPdeL5Tu-RCuJCX8d9;~OR1A>IA27v
zwupfIdHzj)X=HhJG7?<6z6t)e1sbG5=+?rctJx5LZuzr}jh`K@
zQQa4jvDX=Gj
zqr$kBj&Nj|J9|`z$;{Ie#)L1iz5gWmyAaULM3oOT92t&0gfb2H>@xgpf)SA~+;1u*
z)R~9F>l(&OfPCm(s(91EGCnMWSjKL-<>v^`yzFSmA4wa)RiO@laVxRH`!MI1SfD)v
zsz4Te%eBGI3d_Vt#`DNQNwpp#Cj^7lQ^`<7Zq>FkQEM*qD%?+*(+gn37{glZ7P(6Z
z8!qjui%-hQq;rn)j#gt|8eX_iZ-e!mh;4C)0hjfr$m#8>O@HOUuZy5S$*l=zlTKZX
zfH<+F_IMC%6-~*1aV?s`d1u01yF;PA=;_`C4u$7sthhVnWM#ksSPyo$4@n-alIr3D
z&IzffwGeR^E}81)Q#~ewXKU;3OZ`(8RQh|Ivaz0DiyzUpP}U}XB_NdJP6a?Nr|
zU9`9M4p`8n?PtjrgtjGO*u6<0t#y|_Y%oQR{;ozQ-_%rpKN9S_=(>{>+{=O#=60X*
zQMLyM*BK{zY_{a&V$N;j|tXMB*~;95@DU^4Odb)jBvUqy5-QdHyDvxYf7Sf(k|PqZ?5Tc8%-=kqp*Mg
zl$%jun~EwX1O-9v$`P%z*UbsRZ)GqeY*dPGGzQ3jnld3QT)h!d_siiKt1&KWqlk>E
z#L$SARLzmiLD1@9b4eOXd~Y!lkYx1W85Y8}tUl(|321>
zXBY7fO?$b!;UWFf-aI+d4h+RI=W#?E@V}xr1$33T7eQ**QQ)Kp2UMWnAPuCRqydYo2pes|yg&KEWm?F@=_wVZhXRXqf3
z>RC>5!ACYw4wQ*Y^;~{|#=hzV;b%yJ61@qRaPY`E#FvXDaAvazJlHaUhmujkz4lPD
z%X)Sh!;6WB8fHW0Y?w}3Vi~-^l?t>H+~K!>rblVZ2}Hil9*%g&bs|l)*sdVs?124c
zqfmL|rsOZ#_Uh{kOFXDaUiWmk7kZbZ&a6MpzH?_>yl5~JB%DPi(?hZm^KJXEb*Bt=
zNv~y>kK(k(ERv6<4z0;H#OD)KyTqHH8AhZC>Wklai_t@MQg=#e4tD8CA8I0e)QI_i
z*BT~Vn(H5Mwiouvvxi@giv|?@5sBwnaeYnK&4FIqF0Jm}g37wp?(&3WU7$UzLSp&Z
zZPI7GtoeiAV}28MyhZY1Qr;ONAz{ZL0lx!tqDO#`boMv)+)tPGYVfeNX|KfJpfy|6
z!Ba*`1T$6bDszSNE`EbB7EPt>N~UgqLD2;{oJ%|aU!D(b`<%QV(q{dEJGui7^`!cQ
zP(6th!pO!v)hsThdWgQUoGv|lSHV8L&7IN
zb<`JdYTcj>2Qx*@!HK?vc#~1eZ3`G}*Z!CuXH+uPD=1UL08jk=sqZQDEWqP`=N;m;
zXsv7RG#tty^wI#9ZWLeI`m-r7m`)>yd69+9f|DbIFi11cNoO(68j2TJMj00CL!yM+
z5z*4_EU6tMOXGE}{}_5pmf*Qxe|tAw_pQ5C9kv!B{O1)NoWZ!S!5m-_k#Zj9lQTNO
z#+x@o6r5h>kI!EA{iF?@-~FF|HS0AcI&EY9i1627$MW5OgrU!1RXVUH3iQfif5r(^{`eJfhxr%XKEc{
z-`6k}7CqD@88gG~98$0~t@L}O^pR+0<=K`-eC*YtlnG103x5wCLPT(Xz0&ravVW7!
zR?V|5TTP`CtTx&NOES$8M%8!UG?-$A&@kaMU*t8eE;3K{p6&l*-wqS(Q>aRAG)x!g
z+1ax9C9Pd6JQQwk7cCr9i5#T#5q;`-wo%cu*2`LB&XlGS&)%<=_L_c0K3-FpmX8EK
zjmE4!;HKHuFD&~%qAH_*LZE>d7HHtjyXS)!G$}=7Qb+mm2>R1$DZ|5!?b9q=_jmFA
z{%lDg=;&BPHwdAjli3sNLXo}mh}a>eu?{*OP}MZmW0BmDLLy%h17gf89PVF~x8
zAsx1*&r}WbKA_B0v0DD>dFpUiKTcjv_3%kflyFdWMp_amQ&7l(*z;6}w*V48Zf
z1wS`f{IwxoBo;?n)|v((JTx?q2S+I4yAdVvI!j@f*4tQUNR+saz8`v(PSY3_$r?e~
z+S8)aI!0ag6%?^yR=yHrUqoFbHmbgUzV2JaqZC;r5<`f8TYmfPv);UVZHZ1i-DH>H
zm@k+jPvtY|smYwBQmYXHB{~b>=_<0YYuc;~`g4h!CA||axUClp;Gp3doVjg5
zNP1+Eq}**W(4H_{8#;~hSVabI0xqSBdqVk0$N>!Lk+4AUV
zq08@C^H$q+ds>6_+yhs4kN$pY`dUzU28lW_czAH32)hTmU8D%~S&hnY!=`7;6q`Dx7mI%4HJyXr;%-Bza>|i4P`OiT7+yK`oevlf@#X7iwkAcX)G(E6ZhJ!?qT($8&{M_JGPQ9O!K4!fq21#6jylzqjK5h)=6t*LkKeG
zE*tqfHrb74VK9FWtMS7Ba|NPW?hNvL!nYij?|2OImrzkjhAu8_dC`OAyDG_lpaHr`
zJm0JkS3m)_S;gb~z&$^4>CB|4IG3}?`0v?Mu!!FwFXP179y(KJQ<3uq9xXFjw!gwE7GXGYF%2RJS8FfMZ$$bhr}-75
zAWp`>;tFgt!KW-Q)lg4U`y%e(--C}OW+Is=NB!kQ(vxd5L6Ck>|B9sam4i0Ct;Kkz
zG?ppnxM&`#zA5CY(LCE+mZTzp{PMLmmp-hmUaF|NL9K_5PTLSPM#U42qXT<%%ra(ZBNsG7lNpR}j
zw;Mi{qsrlkyg#M;kPLZWh(VGTRJ4la^38D1Kr!D`HSon$_+{Z%&oTai_iI~E9%M{e
zT0OU4z=3CfYtwQ)2Yjc0q2sl84MtP=I0$&AS6nyFrjo*S)Xo0{6t>1p+d46daC8?Ck36&1JF@OkhS@{l;nrYOK?`>G8mbvumZK>S
z&h0*V$iLgYIL*<2jb-~Yzg^cwrx)+UutjJnZ?X3UMb$j^T3bhdqk8|2rXulZ!vcA-
z^kZHXL1m;FTii5C%6H|AfE~$_N$KQffS;0V#*qjxb;(NPuI14UsutuRQasii$Gn`L
zl>^n*f1%23Wo>74q!yyl>;UgIjXm`gS1K}M&VS~=&e-L(#swD@NC;sw;{wB#s~R1N
z-s*qiEq)rJ+n9lW7igGK7s`WvazmGE^s!Ttz=+RX@8l)@ENWxhD^p+%-kFsV3VARr
z7Wklqm>JIqX{5zBFHPznl?ZPq#h|~DSYxQoS3kdfA$ohYMe$3K7iY*mxmOHnRbe80kI5)9!Vmn8Z)cXkq%*v==Wm6PF8P6Id%sCk?l)n#htrzv_;v8m3_>0~s}
zPcO*whA+wYU>ehzk{*6ot89LKLz(hJd%LBOe}8LME-;OaC#X&0OQtM(#?T@SPs6q*
zFY-P#+*~VvEx?d0!I#b^_If@FA!q-+g
z;bSZ-qx$XpQ~FaOexb9D%!r)Jj=^*57nj|iO9FM4YSPRPrc59*GY{XQ0cIAH?HtDR
zS`Y#fNfynY-AK}*?@wYNc5f#6#+JkVZ187)0v)WSPEPtmaTojyUscXkVXPE}
z$Nj-ZqPyiQNst=%;#c3$Xr%g8d4fI#onEsjB5N_xH77pYK!*||2Za{Az!zy3t^MgoA55A|1WLYKCCrQTnsM
zg#xsHPul68479yzRm3J_GPU*h<3!(~}~1;~_mCZOvXSp&g+gr`59fP43F`}JW>Blh>FZDqiX
z$dF{7gxR6tq#mhmrOd)SPwmwRrH?TL8O3#fG8i0=>`FzIk_-|0{ty7kdLZx-a|H44
zO9u6quB~H}$>m;10u`!2ZXALK3$;R;wbataSO`Ugt?|3ZtN2-vfy{0Hz^QZ#8+Ub%
zCYVi@4DTM}Ar*XXy-iExnZRCZh8O9YhxRs3Q$~4DI8-`+)XUV|u1j)r&2_y=`&4;<
zY127I5LelewRffT{w8t8Emnh$u=mx_u7&7{FliTh5qyi2wVs5&-#HiohS2+
zGOUc?TB|5*_K3kQ03`dFirlvNlh3SwS{L{Er-aN1Gb(!)&ZLEe@>B(R3$*3WS`04$mS-8ZTeP|uNU6kXlXVI
zkZ%^6bWq7)u;rT;w74n38Io0K@@tmeR4M^JlD{hTq=zc{)6DaQTk<{UoE~;F
z+3qDV1jmwPjf7k_7XMR|&u$ri%qv8;D>d8Gr&${G8&wRSGwE9X_gbw5^j3uNR$A72
zOoq|O;n|G|?6eJaZ~3AP2hwAF3|y+0q~MjG(_>*Vz~CDLE?{cU!`%F>Dr)8@+bY@SDm}WkiI^M4^`DcW$HQ#5vQ(CXqXs&#;i7txgDl4w?m~ZT@|_Pl3Q84q+ANl=4XgRZoNw->?kG;BZzbF
zW|z~)maRGF*5>%sKAm8HHy|ZRBx%FyWdw!#rQQbi@V3&#OYZ`Xb!wmIE_=hr758h}
zPF$5FdLE2hwHQ?ApvBrXBibATc71}f*qL>iEoS5!snw{*#QoiWZi6p%rei6@CvW+$
zlC7<7(AHwPI<_L7@HhSa368M0drHLmOG&Fd6PLXWd2_H1%)eiuMC*X2q!jUr1{0DM
zsy@{Dr!Z=OhjTF0s!+i+q>c&y1czSl-SxiMKdb8Y00|-?*`(Y@q*yBkGwBnfkGii9
zB(%5<`OpYhh##3!eY08NF$XzywM(
ztJJiwbJJBqi$DS-pMh%N|M^R%OnTLWvHdl|Xey#UZq^YAPd#*;cIe$A8w<(22
zFK+AF5z-jobzv-kl}cY%#I)C5Wm
zZ~Hvg@Ns0vgeJ;lR`BZeb*c_iT2z7H8QgtJh5?&@Zri&*UxK^Ed3ZXRlkPGA;GC{8
z3s8cP6gXW(dBo(l<2XwUU~t3g!D#&UR%5Fj~aB`>lejbur&sd^W{f3l})x^-yI3OW6AvN6W-2K*&G|1EpJ-0-YkhR^0
z*R=of^J)ar#my})3ip_uW*xss33Ac}lg2JHW48)^72C~KuU0Hc9Cq+v?UW>+EGQ75
z6@)u~+`7g{#mp_nhYVv@+xC@rx3RoBYf)TXjoI{bW(Slgm6rSu_;~a?o(LH16OiqG
zXA|`5sT)7$z73O3EObrwP7;K7=bU?$x2dJ_Qv+!5Mt5r>`CG;N*lO9p%Z&u}#CG&W
zmVQjV9(){oI
z^>u>>rWie1FNUp?WMAME-PyyepSs-6-KFW@nx+#Inagf>NxN7+HC0BfDGalgWW>wG
zuBcdraHt1=K`f?^&8i!^FrfB(MGPs!oUEsE0Y~1We#F>`R;Sc8A5POZjQOisbXJob~mSnI-#!>QzIylUFgo@R!rT6
zT?m{mb>k9CPTu~lcri^R)!#(3T56;CNkSJOaaz55$fzy#@?+rTb$cO}Dzv^}um5>T
z@LY!*7n#DlG4%X4Bc!g73M&(T7WD7!GT_{
zgrZR7^XR8l2UcPWP8(R7np%Q(-+fFa7}b0hus{xiNTUw{40!$UG-D#aR?DmR5*~_8
zxUYwioUrl>TDQRCfF1gOep_78N{~aC`}$La%I7Xw2m`|D(|&YJ-3X&&xf;W{pl*(V
zUEhCyaSLxSuGQ#!$~$S{xHrziaI$KV)yxT6}BUD
z*o2In&d$VUOi#3Bi?do~kOuCQFCs#xpFu^rB4mK9bJS00?XrY_O_sa^BPPGLM~_HS
zWff0gTqsX_C%wBAO(T)uLo)wB*KAsWF`{$m<|ECdNZ35*EO9)V2^Lj8U(m9a><+xO
z5#BAZ>wACQI!}44@|a}^FEcunbDCA-YbeEpcy{3YK-A1UJ?TLZWgkbQ(W(=dE_YXl
z>Td7xN(sh|t+j)HQ*Lhes1ra^j;hh^HW`oV?p39!+c=E=DWV7(8uYQv6=n5tzCPD&C>uCMSEnxsxj+go$+J4}%|nk>gOnTY)80Iy;x>P0YwO
z$5|%H4AGBInl+~puJ;RiDH~)=h%#@RznBYY1$zr*>jHy`V+$NnC`}UtE(^7yDk-Lq
zhkVx}1Duj%ZFoOX;Os>@S?Lx&ZlDDn6WNJxw;_aJNQd2ihh4ph%?h34lORqcA;Rw*(L{u8^zD
zCTX{mz_lF;qq~AoMNQVmlI>w;-t?~tl#$CbI5sE_q%73}auLT_d@(UgbY-fz$@+4I
zK{`*Hp|anIfZ6X>Rac6*7UFGrvpc=i*LIcpvHrGyF?k0mn~mj6TU5eyTc{777XlCl
zvQ~zN7&V?_UO=N~wa7PKyZ6k_8V{!OzL55x-3&|8?GVDD^v78;7)bMj?FJ>P6Cy6N
z{Sjom*-H0#ea+pOjx9&IWq2=p@D5vGXph_F=~eb>LQN`(gh$a8NA>%nyCaWl4559l
zvO<-A1Hhj>vE#arW9v3quRLyZHjrGiQmby5nsvz$y(-r!g-1x`g((_ez_}g1ym;Yy
z?^nut1c%wmCL*6sYz%rwb;cPOJKnbS4&LD-H|`A2dW@fxAPH?RY%G#{p%{3$j8Q%}
zEfA&WB$6~D1~}g#Q3jeTq;m-$nZ(n^eWkB|)PWotME0r>OB0E|Zxv>2s5!hlFb`mN
zI0|PdHpeASx{>I!$~GzU+UzRiuOrT>;KM=Q2yO@9larkcv?oeDPZ
zvXLE3+iZ4`jt(=X?=6%s%)(hkGN119fuQbWVLXA>Ul1GEw&OgSZ1OA+(F#}ziV0mp|moe`)RNJ
zJqMiQ;)|GgI}(WjVstPX)Br%rjRBMnN@PcpBf2K}p-KwjD`wzPsUIV|)9;F-U?-FF
z4J?dVGMF5XgY8+1=pTAuz5&0fKjFZn&X_esOfph#?A1A%w@yVf!
z+J}jDeBbpz;S9dZWllz8n6n>VkI-|zaECGKTEclXF&aU5$%G4!sxtw9GSx0Z?gtA7
z5_d0^Y`W5_3ayJryck$*yCC=9NCvlVG3vIG++a@Ic^fMftyTAhZj-Bw%Iwxq)Zw8l
zcep+i4twK1n~sX^U7-vX%xUJDSY9=;#?)Mvaq;73M2V@;Y0>%2z+GFHR6`PhJefi0TRi3qBSA~XLwHWGSgUA{_Cm+P2^48BCHz@NqFV=>bJ
zGp$oiR28B2JeN?aO3QxcEkI*qyPpa|xjZF7RWYcEGnfj`_Pe
znnEm~>c=?5rFsavbsB9d0+`6&u91%+ca91|?uE$@68nXt-*%>dGzTG(xc4#oMj=m{
zEcoUH5Ym^}kSSrGN65%2tP%`d+i_5)8ou+iLY#J|9fW~dp18E|kTEamvTE_m(TgW#RiK}{>Y^ahTe6!%
z;j2jiVQsG`mKUhmiiD+`qo!xgG|Ob4tG4Zg;A9RRI5mh0RqL4aw_`l$`qOhn*h4FY
z{A_*c5?QylT-rTH;l5A;m)eNc&Pz)w0r#;i)_MAzn1Kj?>^(wdD5pj9dMtja%ufn7
zC7fZ_WwLSedjWh{-Y`pF$m*aN!)qF5j>}^pS4&->g~3!I%cUs=JnEkjD2#q~coxjo
zl>J&WpI&m&QQu-Q0v0dTt@VP&Spt!f>b(fkL`$wt@nT*dsfMohuvRP?U6onc+
zeWrQX$-GT}2W2|W+7%~JluBHEu{4Sqo8892YJq(>IUGuB>6APx%1aFm3JH?~1kSSw
zBzhzZ1&j&OBB9|9$1S^k9u&(JsyK|CI59n67H`R&)zf4k~}JzQy2$v
zo5ef>*a&Oue$0m32yPb{!Qt`eBRgwAlb^rT(n=lY(8zS@9v0{V!n@`oapuev4oaBH
zB3v(5^^3*Rhj~FCv7hwiP+!Kg(S*>%$e4=lDD<{hj3jZcGp+gq_83G#Xu+i&*bxX5
zs+@IyH82c4Z3d=Ig;Sh!SkA|_Z4nbxLrrxag6*`g&bDHND^TJFz7$VK}yb>0aF=`k$WOB;kIrMmu$R729%QuRCZ9h+&K+V9Ns
zL32@tg+RzwSY%p3t(=%oslwBPXxpU8@R+)Pk$@E8fCE+#w+=nUUHf)4J)YTahWEMS
z{K3!80N+w6-Nj+D1pkOUR@+2V7~G1oC|p+-yFhFUQYjTHo1(K3Nm{j<$6IYCev-9{
zSC6nC2O)Z$O0IJW5+=@Onbwm6l_ef)Eycv}+%)qlWkvaD*Ge@~oco8)ZwqR>zr&q>
zD`P5=-VnIq94BM9_q>ak1gl~d__wMM^<6wFZc=&WiGvZQKP8&Qmyc4Fzl?Z9)2Vl<
z+2-~$g2TkcFNos4oK7Dp)eS@3m*VAAlJT(hTkFo`@vwtldTPz|E%NwcC_TkQ#
zkQwL!-3Yz7gcwW?n8!d*3eZl2v*8MVc?QSen=y&KlJ({_mL}~PUwm(W;52kwCTk<8G>WDg<^poqICXpV~3ifz4L*p2#Wc;EG~C6F&LUv#JBGvC&Km(<&O
zLarvoC+j8723bVjLCeo`;4QU&krXx*`t@-NtIOyly%BGv4orO=PJ&1z?&>PLYKnjD
zeh)QWyaJq}28Mj@YjMpTwVrlRwSnVl<+4Dd)1Z-W;Wk4l;i0Q=|2tFS0GB$oOX(?^
zfl;FsxwCcp5P9`xxXWX5T^}ERdXb#Z!TRdE>^zbbYLdC*ZJKdso+$n&gSRW>Pe#IW
z8pwTdXD46N4kN#-O@#EvHfWt))kR*tpA5uL+o(*_bP4?qYU`0^7I+o5DRmTANAbUwQaHhB?STNkp^uCM4tm8^=tJoZW-zY%972KU>
z!ppD>2?*L?1}%g#Fk?}t_p_Hg4$_`G>u3-~XKLQplS~y!$*1yFN!?i^$gaQRi!Y{}
zc~R}AO9}SD{f)-_p3%L$!;n1%CnOcVMFMmU;3%+MAb7qV4y5uI;tU&Q__px=&yPen
zb?P*wz=w+AfZq%ogPSUU-qKq@bscoV9VvSb92aJIBC8@vFXf`
z>{_DR>3cYQ6yIp)%QhC^@qsj@A;D{5}#ok`VlyaC?M=59P?uZI;q
z;0ds`7pzv|fr&q|-~4-5=ngc_7!bL0|B%K#2fgi3Qy38*hKFyLy5?ggb}FU}^E|Gr
zFr}fBjO)yhU|q*nbdOQfe-cZLuBp<+o0J!yZ$998a+wZ+^$VlTvqQn9SA^bdwqMZ#
z0pZ7b%ov4Hn5_?HfIpAw=*r1nGAo*otm{)rZW+VMu@h?$fbY38+3Qq`xI*KU^ouc2DxaYmO?q#)rl3b3Gi$Dfz@RuDWXB-|+mqPVQNu=f6i
zV}&K_?!oz>RXm9|3b3ew>I!8`>|5yeK2Hkc4;R|6B-Q{HJ2%nNfVtZ0HUG|%atYn?
zCH_h-!TjcOZilA=b?6K9+D=M+i`QnZZ|!`hwD~Vje>H?x9r(|VcLLJb1z>3-=@T*E
zYVTa$qtPJxHStfo869Ybo8IJ2D;7n!4(NpCsu62(l(%@RK9W<(AQg*K{4PeBTds}|MG0Ac
z7@5sm0V59gGI0)XMDtehRcc>gX<KvBZ0jHs<1$23
zP1ZuOHg|bf!xI}t^>IdtVw2(7u!M$8m|}NbiEDf0@MWU)Fa5Q_5N_9Z8sE!d@119K|-W!2_#)<|0G$3{<9fhXq>@Gdy7SIYdH;zoudB
z%OZZci}@}}0jUwSF7>XZFm?`0R`Ci65R@5_V1BA{eD>1O(@B)~Xg#
zSs2uLT50#JyE#Im1&b<^D|Pkd3MQ6fs#i_NPQPa{c|)r$pL`f?
zuq|0<9SO7NOT1?iJ4*Q(TjIIuFav#fmc{#K!##8kFvOnuLZ1OWMol
z-zn%f*0DWe6>6u)n5*9EJ(>4q4N5SF{z%aKBy^IbOMU#+P*7n`9$7&;xQkgybYanU$odGc-JBww=1b)Q
zf+P|kk3q#F=ck>wg(xP;e=`|?=D=o!MKdC@Vye;S_8bMbJz6RL3qZor)`KkUK(MN
z_~fk)$@a+=8yh)7y1O{#vDNUigZr^*426;U=xl^J&NLF>zn@JWe;6oxZR!uXSe*Bh
z-KLio+mFUKjh0-j!!D36aP7
zf=)r*TmL|jMzOp*DwHIZqf-gCk
z2|K~Kq!ny>g?XaNe_dJh;=YC!ta?6m&@o{-n8w9kege0G@3a^D<@hNK3iSpm0Z0xd
zI@vR)kM^EO^s=AcMs}mfcd(jh;ez{&(#OZ#Me+I+gv>~1^f<1K&(`UjXrHV$>bfTv
z8G}`53kZWDnV&f~JGVD1YTMtIrE)G6z*1}yEw>;(IVy+he^zKloy_$0*kVzLGqjwr
z?H;K_*X(>c_MPUF#iy%aXs%`1mY`1JUf6b6f=&U3f6LbDPW$3;f!A-}j
zjPB5>i~BLe8Mt^XuY1BwD>A7IvjU;s;6O4d*@zJE&KN2zRO$^$DowV!olTSg3sr}L
zQt{V=PptbaM=g(QgTIBF=m#$~n4oVus11-5ss?CX5%ovlvy@=+C!^)WjrV-XJC+v+
zVC`BQe>>f9VH-_Px3Xf0!}z<|_jwHb*4L{{cZI*1Vq#GAFlMHs3#Sy0^>>$Qv{|7ZNpcbsg_7q_
zNKp-kT01q@Q|~%i=P5QYs#U{uTf?oTR2;m|fB1A2{p5V5BDVwaHosb6u|#;qyS0OL
z+^V*iCaOM&MKeyAPfEtIBlEjltBzi8ba;k-HegUnIn*cvEL=xB-V{$o-PO)1wz;Bx
z|9*2*>pHl~=30kJkcq+nmJ4Z_cQEA@N@6}LkwsuUUNcw46|7>;%YpJIyR%bYKmy8I
zfB25`_MXWdWR>`gX)X4nK@5BnfuZqI_03+PY-D!57K_H*rH#vb!sc@tY_#0@lX|}e
z3R8gw+TA4SyAH0eO`wJ^V49;i`kRk0G^IsmW93$@z9T*-6Tnly=ZSp7YA~py#pq72
zsHs&ilo2KL1x)7%X|7*A@(x5}=vMv;e*#De4=dR+HrFK=2|`}3o9Bv|<7Qkz6s)Nl
zn*1JEgmVL*JZdV!OXu)OlMWd-H$oA0`)B`_u+#T%ds)emYK54fOkWv^PZ7tj)WC^u
zBx>6Oex)M`%%!7KSKpesl=i(Mw8Xn3WF4SSFkd-4{J_lUrM*RXTV3z#qdb0~|QN;nS9CrJc&$fogYe~50~
ztyc(JHyg*CoI64=k_~=@jew*ze})CiiG0>FoFX6mI_O@3NpVQL)^(iQa$T^fYGit+
z*T=%-+qD4IWKV+f@d-ZxR&`+4h8aFJbfHq+{Ck
zD|G>qL;Hk^9nYmyR`Mg2ntmfvq^rZ{yMBAcmdU!4g#Pe=r`QH2hFG
zIe5ay)3^cFUWh=jFymAii86D
zeQ|O*=^Kd;%u1I*#63GxBXJzfVO&ZGEjzy^eP_U@!Eh
z_;p)x^o))+T+eFxb1{+(Ll_oZ35$o74LXN76yq(jBY8TOlD@WeEj?o;G<%t7kjaKh
zzZR8SziOsX2#RKi&6z-Di<3%`=!IuZwlYm)?CwAyRfZ?4%icfef0u)-Wk;oG%o1<1
zIhPO${AZRJ3sJue7HsjvXPQs#umUio2VTXlGCOd~d%(pE{Ww$vPOo-vNY+D716cvy
zyr7I2VBN`UrM!@)qzCAiz(Ea1LgEWj*H1KsRloMHHMvkdD8)o-^u=NSd>zA%CEScE
zT4o6vkFi88t0a;Te@eMtfpd&_r`qQzsLGhdAOI&U*lU1@X%2K+h^0J%y#?0f7fW#=#1qYVl~iE-iDc=
zaCqXR0n!;qH6;S|_w6d3?-at!J+f&)w)2`IwOBT5*_f(CX=!F^o1FoCI0|
zRp719okQ4U4Sgu(A{=>|c^0~_E{rSzH8p_&4+IbMu?!>1In);&=SB17#f~-$T9~02
z$IjO}H&A$de{+48)J%RHX_^7@L634@1$X@d=CI8(_93z_ZC4#AHx?+hY;znfmH5`T
z(SjhQL#{06$JuG6IwZ&U3eq+_w>?zUyepEZ23Fmi#w|z23JXxse)|)JXZG@YE^c)1
zV|60eVUW{g`>fY
zPK_AM`jaPPU4D}M=P^R0_C%O9F*~CG|mSR3WpJHyM-z7hBLx42$E5mz=LCH
zo%j+HFnhJvmB|nSv^zDiaV7??gFv
zG#~mJlNP})C+Tv?n>j=K1bZNsG8OnN(Y=cNe@hm}lVr}P#b9*lJ-jKEMf3CvElnI`lGUp!g|sa^
zf6C4cQ%mmCZ(_}Zogp_DD+Fg#P8$tlvra)v%!!gVkPl9(976@8jd(x5nua*Q)ul&I5qa;
z59-2E$540j?av_1Oo$2CCP_@l#vsTge}H$ojaE5pH3&1MnvqNDvFYROacjq^&S{`S
zkcDJIIbv6Kuy*y{bm?}!<<^b=ml(;NN&z!3)yVc^%vC6g8Ij4yT$6jpstTD&o4R(`
zHHXrhAg#w(r4r-#9-WDC_JD$vCQp_gK#<&V`aQJ9=y7UZQu%rj%oS%2nsUQef5pTt
z8G`DO=9ifVG9x#MR>)2bS8G1A4mu*9aWn~)nB!A#_N_k%6e^;b?3QccalZ1T8HiKLlq>4%}V`a!-@9=zV8&#Qe
zyp^Yv@pJH~`v*9Yd5a&M?JJ4gj0xI6C$A1&8l6)(@Kj&IivTi&UDx|&K;FHq#9c>yhluaT
zngW;h(z+pVO;j5lyd^JqE&NE=a%N$@_^SxKz6gqv8jno*|7Y$QG$msric=aab;e6x
zbONiNOz%|NiS_gae~L}9KT&k-Q!t|MrLuR9hmbOj)G3rsNJwj>g-TJ$qY88J56bj=
z2#ur^1{JJ!vrLS0uKoW(fr>eOWBHFPbw?zVGT7ld3s|S
zil;uiOsU{=poBp?sqsx~wM?~G&Dn4+BCGvBEkz0W0(}mj>vD?oE5vielrOOau@Ks@
z3&Os9gfXEdHne1qc98x?wlkHMo}2jX6kyyO+yjKqmy
z=x*z;E3i9~rEL!_#8YF}HQ+i^jR9`M*4F9SvPQ!mwC8TYm{DkGu2x&hC}Qsz*JDvH
z*yT?39OI44YouQ2)nCoDDIlAX=(XN_nH=Gz`6>SITZ+_J7AhDe_wKVn08xh#RK{XiS|hUaDn_+<_iyI{dp!F?&+A^d%4KYF
zN;T59V7k0Ic$Bcb2S3qjj@{FfFP1NM)yZDeOOu-?VQK|B89EwA2kC{x`g_G!Axrru
zuOTy_3N|EG_+Kw1uyTIcq$x?j+bADbf9pS`f7o2H$o03?0XtxytdGH(ShhhKmQqt|
z!4QIj2gBC^VeoQ5Rp*GA-SfP|O%xey!qPL%wtzDzqt!0$zOjwi1V$Jgt&VLe0Uh(<
z<0rV$jcu?gBf%O&rRn3l@EnZnsvU+fN9{*WN7PFM1-R@$U}bNo
zfAOZGKniELXM2j8LP(yo4#oh-0jJ5{3kM*=Pb#KV*`#w=P)DFLvnS0U2N+7SLiHWR
z_wyOsf|UfL>7I}Rm1A5lp>;o=(`w;VPKe8QeLd{}xo}D@&}Ou0{}KgJ&xVG&RYHi@
z6tj-zpd+^|s-!#CzuVRbvYBt}Sa;2`e?D+8SD@#zkOIS`&
zE-*=rTTvfVN;~b=pJ37A+H%+HDnrP#ewzbOJn<*cSb4zQck%{TgXmKG+GRXjD2(p>
z-n>-|ycRSm1Is-joBlC&ESPEGe=nDPCxk?$7JK{4|99Fij9hw4X_w<@ySC7w`fAJ^
z^@tC~VV32OG;VO*65fz=4Xxs#_@?`>;60&`7TFZjkj#*X!f4VR9Q-kc+;CtcQANs;
zebA0pxrYWTwQ@JRvju84>3%tyeh8_Mhcjr`h5NE~jnS~b?efFc-1Z#5e=n+MYWFn=
z>2OVM0pt9XI-@NvT#WkTd?ef3KC!>$15b!e4h}jfR+r=8NN1QysM%#K5?P`0CB|eN
z4v}4p#6#LBv+%(d@pOgzf0p;#*i+nfe6yTu!4l3RrfC{g8+CL_YDu{HX{Ns$W`gc=
zZN6{D5x~=<96w1m@UGzG2?#L18ihF?pEji@pu{
z$Pzh){MP0MvB1bcf8O4+sQ9B_<%=z{B^(3wdBUc)4f@oO<6Vshz@gFT3X!prD;3&f
zFeX>?xZ&9&9EaPKmBC|Gp88V>|(0E}~Y}cDe_gua4
zIUFKqSbHj>^+h7F;_cRN-bT+;3kflR`{Ukh=6f5HIv1DymvwRM}H-^}j{
zl8ia-7>pi!8Q4E_k4-RK=wEaemz>gMOw!aDOBffW)VST#F8f8bXWj@JdY-S90dMnU
zf9(qYA2a(ChBHCYC(fuE2q9qT$Yong^*>{zjDJr2uk#;nVW%-Ooo1PfwCO76EM|zM
zlq!M_$Wu6Ye}^)J0X&Mm%3}I%FM!@PqBD%x)x@Gb1;ZyqkWvrp;C8tBtn~!i5HPi=
zb2||sm@)oD!810y!?0GDgfa*e#QK@7!dKzIKJLidFAv&nQI}F`*K^EI3ot;=IjXn|~25^@Eee*{@|cfa@X4I~d!Acx!U5wZQT
zC>@t-At0v2ZPb8V1W#BHw{reBQa~sB7zg6(xm>=T3;JBZR3VyRYTNwepHD2f@#00j
zHaUM(Y1%o%%4%%O{TE7CER(y9u2BfNodc>ENJ_E4J@4vOc03_G5vC~V{1Aj!u?rT|
zqqJmRf6QM?G(dhmP^;Ta=Dtf`mm=AGd9e}O_`8pqHHKDCGEUBQ;;al1HZx1m3dU%J
z0<02=%M9;AcDw4>;cveR4DZ{L-tUxTw|?|e;aJyrb?;VLz>l+jyYNqZPV)~-;|V$Q
zx{9T%#X@&7IOB$Pug&SLR)ZPbZ-wTsXzgj2f99Ds>AKns--53DyPJQA_4g9=Qnf5}
zL!*xh7-uu6QaO^#zis{Pi{4Ic+NFp8{Lh*Isf_wPW1Hu5-g8B2me+8%TZkA>URJqk
zXH>z8*w$++->oxBy4d{hwqNYMq`Qgj`r_85EHS_vI@d3?gI8>jrK5aW)qa}mDu);L
ze^U@W%Kq6nk0vcUT|1Ml+_Pr0@NRy8ZJPo9m?;tjgBSE-k9Ba<+v8h5U^yKV!fn)f
z16$zVz)~{OS#R*7%!$BA{>M1>;JVK~yV#)tCzs2pmKJbZ9T4my9U-Ve`H(i8@D9WCn+h+3S3>?8&X;@50xCV0#Bah%lX@-oApnf7~ReeAX)@50kJii3e+OOez_Ok{cDyds(d5FsiK1H{H
zGBZ%{I=y2vIAaYX;C&re-gvzsRum<+bG38)c?y`jk9W~w_9B4?$t$`tX+RtFc^Qx?
zv&AZf8%dsfAeo_)pu=O
zvA0b>$bH|WQ7z7Phd&qpm|5c+%Vl&S;7kb=#7yO7f```nrcOgx^87WeUbwmS3nav9
z`zw4s6w-hb6kly_Kk)iE+ES*yz^80KcYoh1+Lg7vG5tx@F}mgiPPXu8oYwl0{@{iUGe@izPSi<`Ca(&4B
z8v(ArHltKYgnFQv&YHhqgq)Q<8PV7WG@O>C{+UKOLh1z*1`a{L0A?kh
z)zC|^KJ3}Th;%?utFkcHW|PR}fGsb*G_AZ1NX~sU+upDOv|E!kG{CgQ9o$g9H@j{^
zxc(B&;5wNgn|zFpe~1BUp`#pvSb7e4iS}pE`jswpv4Vg)CAO?l_-{Li|lPEQs>idgp8X)?~xNFz>iYvCkZJL&5@a;v+wN
zFMplgc^c@DjCF<#jxa>k+0MhVV#b98KOby;x-Gj^{WNYne>rtC@o&Zt$N(W(HkB#H
z99o_4%&f(V97Kky0=So@?MDf=+~>PNs2U;oq3iQ0Jbf-vWgy2PQU!p>W3IXKZ4R=T
zo>p593Av9wf56m|Y8S*RVQ#b=F7)ck+V2lUzDN%xd{G=egx?)^6@r68dibY^-P38j
zWl7l3C6?Zle>?_yQsLrG_RsRO*$Wgz=2S@-n5Ozxl&!jFLWRt{n=6bg@sX8z!0NuU
z282H&4P7A!Wv33R7mX3Ve>8C#AVf0SA$4}yhZ>0OWn>5ty?jP*Uqe2{Rp>lMYiW;J
zb~UDT)JrW47bi=qszpI)@mx2z@y#(62^ce{&Y)XyXOy!D9&zHNZzLxCu#Y
z_?o@;LfPSwM3*NxUB3}G$(jek#cH6T6E1lp+M|@`??l79cizVLnzaJi80!!)3s8C)
zazqu7>m~5_wba&WSgJCe-fXLov;H{bpLOgAJ3CY1&F{B5o$0@MEEBZ6%@Z2O?~zKD
ztBOTYf2ncY^81P!=|1*T4%A*(WJ8`E
ze+?3dqk!T%uVVA9=tjV&1xs~nTpl_Gh7BUK46P5{neo};c9+*^ip2U0{-*!V-~Wrx
z-yJb8!2AJX^nbM2*#k|+mQ%&gGdc*YderG5eqMdO^swh|fTv?
zN>H`IKf^zIw7C&w-}@6MgDCNa>{BkUe^baK5FZ-$Q(O*b)l%Jmd35&`(GDgL`+d;t{{|ez>~Y=%P4vS&y+>;scwTUlFMc=6rQhy
zJ!fhXT&_atEq-!Tel|gqeKmD!X4cHxbxCRe7r%Cp0QS#YTf(33nlO!P{;c
zT+|Vz^~(prY6=g1i4(NxRQoAge_qu|j7n&?GeD0&5OtFT1i+ZkC1#uUfP---lZaK`IO~2SJQD@hr$HY%~
zMQnuH^md>MmN4NEp;n{p?pYwaum6lE9`^
    5_cI@`^Ff6#JXzMc3UZW{sz&
z3`B1h+FnPrlvZBIB~&xQ@yS&v>rOAg5iSAdI+f%GOw`<21Ns~Ff6O$PIiu~aMF_8*mBBM01PuV*7FBXK}+QJ$vC??Rt0$zZNakKHh9X1O(9
zS)tjFA8=D?eL=fK;N9GuABv_~#^7xoXzBN>wQ_sVA@rN*(jh`I*c6-bfmE_3ja43X
z$)yb$zG<uxmCm5u9BS7LB;=plGN>>jhH`jsMUpTIO`U#bT#ETAc0j?p#>9g@6AgEhP^}tZ4vNXU3gOU_pxsNhjkDl
z&ut~kJOI~XJ095gRB`i{Nmr8&^2Bgb{N`X+7-M(iYTaz>@;OOwBRuGK4SLq
z(|gNtG>Y^cnPg*}t7?;IAKQbAF8hY3>ph$rkC!G!2loEir9g`tV3;Gi@+6AYL?q!0
z@Oy}3EQ6fke=-%Hqqh|yJZUu8)kSnVX+8*rK0@9%$Vk)Z;PM%QAA>l?vcHJ8k~0=V#n{CZed
zZzIha&iFpoZmX!z6HdtIDf3!w$I*tKP6U(>MbVcDe~r2ceA=}>eBw|SN@g5)OI2)t
zg+23asorb*U6mya5bkkiy^TJw`6@v|g$Yzw`T38cBR2y9s@9QCA2!@=3ee~-Uj@$L&VZ@PNHvKUOSquazL&qkjV`M0i=7s6
zT(+6Se{BXx=l8W%?N5$VI1!UGI-;n(!ouJIxl{NsCv>fl6v+TO4S
zWTTC%LXvm4eN>pY;@+wjS$+a!r}=o;FlD{j34ban(b5ne&Vs8?j(va&?Un%49~=S8
zz$b3iG(DJCRp(AM(mB8aW7XF*NF6r3-}+~me;JaZp`v)Q={WDHQ;+z5zx&$&6|fCR
zRWWQsN$sad5YQf1BSf!mi%a{JI_-I<^i^Sw9g1sp7^!G>0h-FiD0ENBsG0_k4q!TV
zA~F`r!~1s>s)mx99Z%l=r0QV}iXCK6za@T*l3)1pMDts5w|apUuE860h!{xhuxeg&&bK2OR%*_P4u7
z_g`fUoBOSVvHtb9AJ@3&sz#@j
z0OEN-dPN%@hHs&lpjGW?`j2gc`>M}Ae|Jbj+-YVp0eN{~FW5%uhAQ(^{A}w5ZJ|O}
zN#tevt@=y#P*o@2yB|FZ>2ZN7dV@s`z-L@V!hah*CWB`*VZe|{il+v!e{RO;al{KZ
zMM`oeZp{z*f!V}Qi=*kC9bqJ+wBG>>whPXBNp(X#8qe8f^Td53vbf}na{79Tf3DBb
z_bdMrs{r;7nSjeX!+`y(c>`Lk0<;B?qNeU8Lb!$9)nz9qr1HcbgmM5pOVu=Y{1{aI
zq(E$b(a0>i82)LQnkS97%IfU8pW&3Alo2fa$&A6*xax(chjFF8K{Z10H|XI(Fn)Hy
zKGSwpfo@%=ulN*HM553FE36~Je}-vAZhByXk%hmiNOhWf#Pc**bO7m{zwvrZD}N3|0_
z{5^Do@=Cl7m9fUD+n$dqNOzPdNIqJX9PxaH2+%dTA;!v3#O+Yejdjnde`+#r==l~x
zVTV|)vhyjw%8^Wp`~*(wbVFKi&kTM{1RJxf%xKQ8=nEUrY9IrJdCvW}kXVN$q5wIj
zR%SKwGKkx%V328qTrc#-4%gY89U@-3agYPpWmr)QXWRJO~N2KOi
zp*e$={4Q%~Z@Thke|gzqp+lvV@-i?(#$M;>OXg4?N2X>8Qh_Uk2JjO-s0`*>b;>3q
zZTaQB_Y#QXa0ye@kftiwuMd}e+5Bo-k?MebdS{bL|Jvo!AYPt*f31o<2zW8{XCi6g
zWx-dMXnb}up^48#pd}!WyKob6DfNT%Da13mUB;3Jg`*}}EeB(L$q3tDD!2;(h&Szz
zZ975CfIMqn?7nT7#D7q?dtpid`e!J&cHt48791RlKc9N2TZE
zqvs81aZmk^JC?kBe`*n7csPene}w)}oa@H!K={A9tCi~v7PJ3-FWa*@Z3q3lzRi=bQkyAKr)q)Nud-O|A*K<7_oMjKxguW~%PMffdJ`5|JB=jn@3
z-)|++yH7n^Pe{(7JF8+?;h&*fEAjR=Zn_aEaHQlhS!2t?kZEz_vyE?J)&yr&5u)f8MQ%P(vn}^KYoyKB
zz)?q*cUlRlS>2AZ+Gd#w5GP#$F|V^c5R-B;1Ex)^)%^!c9Qx-eDpLl_Gw1qg{BXdw
ztRGz$o;BS4x!07uyA8NorX3QChi5D$+IIx|1
z<;?I3e>NjWr&#p}ljD&&inQzm1(Rst9|t+B=kHXfgg+KhdsMMCkh#x^xgI?IEcym;s^9`ad3F242
zl1zUKE}pzMgnB4YvJQJ8b^+L48Bn*yGOZlYf4k1tPPNUeQE3|laD5gT5GJXw+ca5=jIPzT+JUCX7RN_l%WN7Vu0fRfAed}y^Qn^udJ`ktugKAfemT$LV!ihV4fq+H+d6cl@Gg_
z{@+>{T(1@gQjT;$Nc_A+AtV;IVLS+|1Bhat8mj{^3(XZ2I51`GM-;#QUEWFwIMjJ{
zLc9>?`Ohy43t2yV2E3ncm&3RsHq^uN>HooVQFe)C?t~3Wl!!A8o3_XLlL&1de_ux?
z*~mI9#v+>~WY=JAo{`19V{c@&oqDNU9n8S>C{+|Y-0$C*5=(`M&#tzi+DbBH0y@}ZC(iJftO(>&(6l}964zH|r$DV76aSXYV=I}({vJCeG`zKiOcXR@yT
zbXp^q+fC#X#m|q6!F7xFeMa4ZEnL%rmU(b1Jb|~@Y%$PAvc>jk5-F$!1@nGF%;-BGWD1!e-cmeGHeH$
zrrei$bqg*4eX=a06n}z|eo9n9S63I~07#SeZi!Q{$gjjZVT%u3KmeGy-Erl)NZYCu5*{mE
z{C6@~%$58EeV$$r1dSoxf4J3WtlhypkE{FFznj=wNTeOM#UjQ5GEs*t3!;FK@$OGw
z-kbgPIG-5K#2=6SWxkUu+_&u)gHq7C24_k`*~huQ=UJ<($zuYrL1X+wxfC_Ps76`#
zFg+hk7_UB8+$-7ZT7J21o$o%b#?=Sd(S#j1XCPo-rZkAuq(}xne}u2h3se1||qVAwbu7FZ%SN&+7Uiu53B|SIm^Iu>QAkVq`$BkXhVlOdXWKnPuSW8*^f5Rqu7=ga@qCI%t(DX|-
z$km;xna~iAo>5HUXRv}U;jioHlACd)Q)RVw?YDS}pSbX3{qU62xAs&W^?jF2Nyvj&
zsegD!oy4lW7!9$p8KS^K`i?CJCxLq}xo)4{j3|Upv4b63l#)Mmo<7z7{v_$zKzJyI
z3G;;a7NqXUe-;BN{Qmd^9D`TC#nGF^F5LZe!?sw3q}>WH9fr{Yt!^Kxr2!wH0_MRz
zC(|qau~+_6jm#ma6ru4tzEDktP#2foadQ_9O+@AQb>a#ESKo7Zdp{iCe^Oge73_lV19KlVlw;{iT6CNg
zD)xZ7Mcc)e5S_Us!~@N$nhz2=SYU2E
z)CkY+U&&D%R0IV6645
z2H1MP!yT{%aMiWJ@oWLNlhpRQd$!58U@rOee`88^EVQ7XL_&Q)IFP{n{SryuWB*fX
z{*wXSbetviMqU>}o^w`|YR%K_2KTN*-s#N-WsOXG&S}8^6ZC9YQph;GJKIgwieigo
z_#!IA%I#GEP2u3?q7R?26H@dc>}w(GpssE|Cs*uRYoyCxUYsGUS`V8SCqInNh(9k8
ze@QRm)M-3qHid;IfEhd~`So?L2Nx5;u&ik7{+6
z#^6MaJ!;D1>)p4jE1{yQ7pKQK{&ye0G$nvoxTWzg%Qaxj?$7lXX12lqNBYdHf7&PJ
zO*N}24`t~8*1Tf(Cs<@CR+Dp7Ee2XSkH}r;5~pe8%hhLyX@-Ik?{AFCnHwp^Fk4`D
zd~EcmH#keUu`t6OCOV%k@D}$(;*)?LvrHtqKVkd<8q9j3)zR
z8};U}QYTxI*>W(XbML|ar#Fc!f4?4xQCvuYsA=-!d+(}zkv(&I_2d&&|F!JdQ3FS;
z&{?BoYVx=>kY)1^oEHsxRGb9g0l-;4DcIqAQwOHoGs6-_=}>^^?luY@|F-%zD6#9F
z6Key7CIko*Mcq#+BFJFt4uo$0a#CL;GmjZt$ydC`j2UXzQi~WcUSbsAe`ijoOdmt)
zZ^d+XMI6rOF54k_S85_P32O?!$`Mo)>xO43=T~>3P^rA*6*2d(>Q!oSLYA{}C);%w
z0{@PEWwc6Ndm^?ih;01ovXvU+LWSSuN9<*2m({*HbUym#bo8T{Vqc@f6N+S)#gCEfd&LSi84x6v7C)>!soAq`gW#%!Y9RozB+Jn2~eC652I>@p3Tk+6RgjAIi9Bo!ZoaJ*Z#qr
zw3Bq)*>
zn@#GxcR~xMw0GNof8W+xyvVkVPRlBH&l>__;3c;fo-IoYeryY}NX`(6=rp8nQfORN
zGynXXWuyy-GYMKpp!oKS$+Xuuq9`2?!
zWt4C0xs)2+NzW!l?ps&%TFZSlbz#r#*&TU@H&d(Trx#fZZt4yi8O=*j!ezkv*vp#f
zJ?ELo0z0e8f7tuRHmDjk-aW)r-m+O{$ixL}w-f7ii=a4YxH;W|nxGrgP7|Qphw%hd
z-IjYN1ebU4>f7puFG(njNDSkI2z%j!$lpOiOK=9lkiJiv@*K{=;{464V_y`r)g~-5
zA*Pw}4Q5a6{~p)fHpf-*O3gI(U#p9}Y(w4!M%fBwlk>}hey7?eUt8S|GvS1!z`Uut5$8shLliViAKQ4
z#sDa4W9>-8KugaFkd{%jvC_9@V5AYWu`~iO($dp2!;z8-*#q?*&1|fN^&Nqn05(Tc
zfSjS@yQ4jTk)ED^1CA6R2DAp+zblOZ25tZuprgLBn=Oz5K%xH+khgJgq%qKUcvl0h
zP0Xx;l=wPj*hmRbaXB*F0>|24vw@o_9nEpmjCcmHZ^kq
zxY*cR0Ny`;?SYoSe;VUtZS>wwM^oUx0{lJ_fV7z*(Aoj`+ey^sUrMX@R=zvE%N_rh
z*n1-!f5)`^mpQ-z2>g#Uruq*5_)5#mO9QO*&8!`P*80|l?}m>0j!q5$y?EHda>etR3Kfw@=vYy-$ud_HK0lHPsf@HZIm4|1C8(
zvoV&#jsSW9&<+4}H8iFBo$8-S`7LJnEq-r-
zmxry5Ex=gc(gEmYW(<7)fb($BcLo9+?VW&L9)BzT6TvaC0gTKH9p8ui{i4AA6J5gE
z*apDyFY!B*f2;oI_$mImq?GSB)X2u#(hXn)G=`&-f3fEa)~-wTFUBM
z0V)2oq5o~AZ)Ilb_MdkDWugZBO_k#Ra5i%gHFE_T$(uPEn*OV+|B_2M>c0=YfVGJw
z@cl&o5vlyXN0#pc`@VF{e(x**4Fem?f9c+5(a^#g=->cgW&KA5eDA@3@qKUnZ?*tB
zWl;eEe_17}{~DKnj6|#rZH&yUO#qB6tN?v`dwn-J`u7=NWMKh#Fuael5zzIYApy|Q
zTH83jy8vvR9K8U>HuiA850aGyKqv59^e@B;pcDLq*Z_1we-Jx>PWTVv0MLp2FJhwy
z(24#*OaMBGKj=M!)F1SoLHZAR&mi*$y~mXMe_zD@9#j4gVg%4B{z2~Q**_YVC*
z?-^A8p!W=F|BE=@k^Ud_UXj5c^q$@De-X>O&d|p4{p|iNVP^gY$f|ACC}@r}*?Qoq+=@)vv$Z~7JMaiXZlwthIghmf5CUAwtsTJ+uG{CuU<=_vEv^p!~aPCb(#P8zL#P57kn?n
z{xA5>+2K!&cghZy`VOXl8N8p?-_hPJoc@CEMLGWk--~kjtE2BkUH^jb4Bh^M??t))
z1OM|@$^U+w{If0T|6H^GdCvZWl^pGDe=LA%W=8LSWBg?zqwi>M=Bh>izUmm>rT_946tr>mpkaO=e;P)14gdqo`{=*-j?L>od=3Bg$opr_zVCy7EK1mw%}e{1qh
zCCjss7U%rpo#1C(picS~)ADt-Ilg@2+DL(aOW#|@8y`tTU?fwOR>eP6=9h0t2NC6g
zWM+;=A@g)3j^%n|WJN~c9
zc2LK+2aKq2{T_jZ&og>CL0BV_e^9pFohTX~uZUxaKP{A>C-r%qH+go9VpD#_jFf`O>@
z2Bgy11|i}Om|YJef%2CN;Z*#w=zO-IHn&)Hq+N}QA5d_j`fHs+#1{X#fAK*1O?w9R
zhZ~sg`76l?5Thg@bU
zmG?tQkE5ARFu5Oo`z>pN#Kv8&Y1;_dX
zoyR68lDw&H`3q)ZM02rwr!>2IF6JF^L0yS)8rFWmwgkT)V#sf-@AF*Sgz~jNQ$%D@
zq1QvCfY6YM%zCpz2V-c3A@YzecUE)z6snvVJJ2=hjRqlgwQwYGf8YfrGvL?1^V3FG
zxuc?*xSvPBm>(2e-n%)k=!`y
z820E$!DOi1&yIaYSc#+;5bSp_9+yE~p@-b2#^u*i7#_|umx+-{5QCMO{=~}|!JN}1
zFcsTrk#=!udr;$S&Cgeb84vi5Xu`U7VD
z@H3Q6VOr`>)ExFX9Jy*}SvD8>w}Te@c3Z>&yHkm#eAJr>gGhV{BFRi8G|~l7+EnUN
z(&JIfLi5lRK`e70>G)(;xRHLcp45x%P!6MJ#JS7Lxzxk5e@EuNWM)Gr)#t1{VM`Hw
zTCvqUCu6T-_;$!xHH%f_kpM9m+WkzQYkpbdf^VuNtKBEA4e$J~uZ$8*`>Yt4{ddP`R9*POire~e>JN{Z>ho~Er8J1^I8^USAnPf<@alV*C-^T%Ntz-9Q2q{e@MF`xe)eDNsy7k58mQ3r0@)Z#*(FG
z&iL+W9trJ~RjG0Ci2j$onWhb;}cb|W=hj(9>)cm?TyWPPi9LRc-F
z6=g-C5i5;urpqG|is!A}^S11bIBBJ6>Tq$5v#@>Kpsn7F;c~XjY3Y4b(QaA19Jy`G
zf5Cc=6N6Ny{)#Es_GyW1X)APmDsD3jcRa1d==n&jZ($%5TVOq#_^Bg_V1dug6YMYm
z%9X)UQT$V5#Oq6qCaXv}huidZ3HA??;aaW0L~g4Svd<@94v1W@@M+cfei
zhkMjK<8g?daZMN5Es8J_HM7PO`sYU!Vz~LCeOPAvn;jkXD~c%lB#;Txv|H2W
zo1y4chQubRz1SN9Y<0fbA_mvwMD|?Qq!QnDrHQ&fBAom6Qf`*aG&)PO{ks}Tw!>y>qpCVemk|Qo!-N6
z8{-7)0Pk0+@{QncR+>Im4c2qc#ya)(3ov#j_6h07yXNZ82h*Kvr%V4-c)p1qMp%YEOQ@9i%yTgUYsYPE_FQKh8#1k*we~LE_xA|~T
zQU0WgYV*Y28}gtp)DjHIHu{ZMm*#!k;*drS#k5+(byGwLBi6NLZAg)=m1Xo$Z*{JH
zBX~g7VLg|uDd6(Jh*B(Ecgyq@)s4TsFq@4Tv&43yVeFiJK(&;q^YGQ7({n*Jop75#
zS*9tv?1&Fa$+unmoF7Ptf8PJ$Ik0;h6Kh<6x(=ls%z$C@bK2ys(N8OO?QU!c=W36!
zJZ&Qcbv~>`fGq`rx*A-fVL(Zy=T4lHqLSILxk~C866D2);(^cP;SgFMN$rneRxwDX
z1rb$1UEJhqS35c~1RzUhtjFe
zANHFssXyWLxC@NSb63)O7u-6W=rUh)Ps8h_VA8Hm8Iu@scYh5_6HR
z##^OuqwpOOoS|m5GZ$YTh0llic
zWM}~FK<_l-(+kR)e{L=^e~NR3FCoO2uC#=q3Mtt;mV5!3WL+CK6eL@bW_mIoq{GyG
zzy@Os`sV{woNbS^+O*vgYn6XO
zWZ$%8!Fe@8z07|RhwTb1qtSb0_&Agr1Ja$Jb|xKctAvHGVM=v#`Ae6k@DO;<#FKCoThU0V2t49SLw!G7Anv@@!58Fs9(p(AiLkc9;>X^1-~5C
z-aYg2$BkN}H`#vO*5BlmF*P_p&H1``7$Ec-!&?lWf3oZarcD)@Yntv5`0JRi`lpXaNQCCY_;wYFc#C3=9~oc@`8@A-#GeiKRqP$9z`Lw+5;
zd6GPV&6#sp7adc6pBVyT6Q;SVDjVk-oUy%C2(J*qrXyK6)g8LmMcyX{N6#$s%KkQO8D7p{Z&(j&t`P6L&{`_CC;_{pL?kuEAhBT%&pqb}%0$(`41*!d$k
zNKJ!(mzM*VdX)z3<)V|hGO15Q$mNCOHDL=p+O{FVNwuR$duw)SZ>JeL?PnsDQmM8!
zf5ixo`ysgg8U}0k+^o3EATJ#KXknV)d;`%jBkqMkGVI?ps+Wan0+sCPh1iUK(RyFl
zoa3~`tkpuu#dGT^fO?=RrO`;z4vevQJv-Ckkw
z916XFnjWN!4VHGYnD1-Q%g^rd*f1`4e}fI(^Bvx=0)vr-CremmD`c>I7fy-Ap>U1f}9_lZ!y&NRnV<~<-&Z}+c|wlofnuogaS5-h;X->
zw@>0dnOPXWsWUW(rR_siklmzu3K+1{$I$KCe#TT|x_}65VVg2c)i?L6*$9y6*L9Oh
zFra=2yco9?K9CFIT-WPX`dY%3L=b;!sqRGO8=xV86e#_4?k9Nav%dOl;|P_votU&2
zODAJBDeQzZg)e=khg6R)#9?sbe;kVniPD7sQ6sFpKu|dWYqxKgPgTz-Ad{M|hLkeb
zPidcCSoh5se{|T{GYBFLab+P;lj}iGJ4t?SfAW;h)&buf8;d7*9}3}Gf>_8@ym~ep7u~-$PafX*9wh@DbEBvWmiwl)$H?9m=JZK
z5nhF`CjthLI5fH^M~-HTSiStu{G=w?6jeAXD_Yde{^n(D=;%X;Ql_r%A3){s9rHn7)s=N&2lm|XzTxeuEvdaBG3Ie4kx#`B
zZ;|Gtwgn%9u?+gA9MarUQY(3vdR#{t(<$N2M@`>vJ9Y@GGsR*tM`lyTa$cRG
zX$JxK3s@MRI(VkYe=l!WudZVjgZ-ih-B>!#jq=H5D)xxQTbtRFJ(Q;P4=6AF3S|xX
zrQo333&F(o*^)Y2q*Hyv^29zxVua}CU%?j*ep4ruJ!0zQ4}yTTX4XDV$k>|
zoI$UTiYpebLv>~)MmL9J#Vjn+K7!>qZoMYZPO^bYWCnQpe_S~mlReZW9z$nv0L9|R
z6)Kp_mY8SySh>F@=986+l&~np=W8+=QT5lKq{=w7M=c&
zhErer&{IKCJ41G~WfjlaZ;sHoK#khKrdyI&Ge6fe#LJ?SeC@sLF${S-y@iRa)D*{3
zc4vj@uIkxxe+x4RNuk|GSUQ4-X*bO9ly*{A2TYH5aH=vpIRtz6)Ix$sQPrqwZ}
zm^A4y9dA}1(kTc_nnm4rg$ynLYd>df=0?oJ69vJ4f1j?RtM$
zvP4!jf17QT9JEaW(n>`b>D$Xqg?Z2YT%ZGL5$ti*q#(_rZv?BRqbY}0bL<-p7)@oR
zqU%I>`3&ZLid_jOCRA&Zs<7Wk!UmwT(~k5BEEdAJLhfcmz5ixy*pG-*aNO^5OtH>A
zeAPy5w^%|8ni6%?Pa1cQD`6GHZYz&trbqADf6&eSd!{htdYBm785jG5<3dQ@qM!G5
zC;W((?I~{XyhA3TApyMGD!W!_)bNH!crsN$LiPQLkk-1l_OnFm_Lm-JAJdpm7Nu9)
z$#_)Ja%#6A;JDp@QE7)qH!#t-P>xHNQt}A|g}X=+^Tz-TRJFWXGaQD#2Y2S)4cHBq&ct{}#?Dk;ol)j|^l0ol}I?s$5$
z$&OA7kh`SuOU=V5Kf`qVwSGhaOof6PF_*>YVwi(gGcZCy!7CRbuWXoJqFRf6MyG
zJ*MtXJbBZd2i4@bZZCo*27J9YP0|X}gy_8UhLNv6MQaqjfX5rCSLfL((7FsyC&>yN&E(2_@zVZq80l=dJ=5W
zjLKd+&lmP9$TOOSIm1#ac3=YOe9xzp^cXHDZ})jP(};1i?WCZ&aQb=0P_N`M6I#z-
z(=HX6jqeja*3%+>4gZp8dOMWt;6t)8wT3khlwx73$L#?|{7930VBjW8f0^68HDcU#
z&wB{YF6#*v@J|?(M3Z#^kIq<^L{DynO%vL1xtXzc-)hkPg+K>tF2sTi!dyLZY=ZCO
zkHD_i;7E~((Jmi2n=uAqR2M&nS8r6zInp>Ow}{^S5nkyUADQ9wc4vh3GF<@;EXkri
zu=izjC3jMqZlESxVMO9He??j}D`e*c><-feuvWEb0jwiFPon=Q>T|3sYBVq*I(ki
z2p;TXCjIK3Aj7Ba4l**mD<;B<30DJSnQXlwoBj{{twK4^^{w?{({!>
zQ$Eu$a)&~})pzD2_k+z|=N9bqx&9z5dLB#NX%62XAYpFCvhU)n^xjIZ7*}jR
z_gz6%C`B6<$NOD{iS+BpLvb7}eJSG}U=QGwwUgW^CCeI7e~ER4S4DI;y4C>KChXJ@
zxi7AO8#4%Hg5iUEg3YJU$hnm--?zLzj(aVD^U+LpC~+enEK-NA(RjYwFv7eJ1|t<=
zQ_8c_(BoA&s`-+Nk1eetab4>$IVhMpfFA>wkG~X%VCDUipc}~dEvpN~SglUG`NU_)
z0;!7%g3fKvf7BQ31$4pVLcPE)Zef}tE;J~rl^i&0Z3)9iGO>GqXxr`5lX$9@P6tra
zmCkKm%%JHxyH4v^{BmmV-mqBUdl2
zggp&z;J9qCojDP!trOo;fh$zo#R?P$=#a@Lmn%ubEB$`8jQ({5jm78lh0gt|4;DNf
zKr!A8>u&IjX#-oWe&L!Aw9AoniATmVm^_41GVmPt2nt&F0&OdD#q;_ecAW>iJ_9Hi
z3i;6te`GJ$ix^h92R#@nfZl_mw8kla1~==#g4-o@+iWdR@pVOXReblhl)%vDy)81j
zD%K4sg4C=>7~Qme9SMC(Wcsc)s6_?FbJKsnI>@{;sib}@DZG4nA~$7jm`++?9ytFn
z(H;E;qw}Q>9&jD|WL%{T7hKHNq5st4D2zwPe{6n2LL?fY($;s%0&<-4fE{f@;A|;K
zrEFK&P8j6AQ_080-c}kQoq-5d@PrZ7(%9VYYaZWW{Y2rwy_I$q{i}3UvRY;A<{0Ey
ziNf~9euvTI!vh4Z{iqycv^L`ccvF0^(!PT~-Q~vMYBF=6++{3V#!_!j$O~C$9uPiJ
ze~`UsU%5eDf8440ba19Sg0
z1^PlsxU=pe)uVMLZJlBP>)w{M`ry@@{;o_=BcSPCvJ*
zGtL7ZXQyxtO85rC$RmD$pIM%odn%D;)9p$1*qJ>
zB3?>uSNaSrMYM3dfHlv>Oe?X|e@OPhhn0d2CxbT%56;_1%D|Myqw*On%j63!aQ6Z4
zM$m!;C1$a;a7M|&P#J8W){8|iV$@G*>+*S4*ozpI`YKfx-%YiOR)jRBW2p`X&@(e$aRZvT>Ftn^QjDYa(ba^Gyz_TIxQLc5Av1&z)
z0F#jy{j;XB#`zER<;9`Bmatt?7JQ3M>*$78lcfNHZe{f|%qmUm6TeB${LS{C+!yM8S0>Pb5UN|hGUm7P))wl!T
zMuch%?QrGM@8r6?g9wvL6>(yW}suoR73tDgX&t#Z6J4HCffAIOgp3_NIYvCvU{jCd`(kw^l%=p9NYiV|lCdD?9ChB?&(s
z8C$vPY(_;`j^Lpw4!cvva$(tM6%}nk*$AcM`!q$&h=2Vpe+^krkm(TcgSoAg1(8CF
z8plITp*HzkJm-(GjWL^bIc59gU7f9;Amfpa0*1Ih0b2tZOY1PMAc5=y-`O|A@lBix
zrGC*Fm7f-0@s&*(R<{OX!R$VcIUk%k{W2HkZ`I(J#;(RrYpyqk>4U`MF1!*jGKN6k
zwf{9MyHuE!4%7!!Uf%>zwYeI?
zbIjhQS=q-_uro_tNfd*XuKkVdIP+nYO|X>f4xL9j*AS6#kc?C-S*m2N=n#j
zJN5wctooV0>_xu(O>iu1R{9|AYr#}XYYm#iY#$uJm4AK(Ie^>Lh+=S!YIa>gtbFx5
z*iLTO&h`0%jWq2Zwp7-p8x4{7iht<}dKd*-na*cFVdB#z=%3*Sxc#|_cgbf&E1VS(
zpD5J@e>fW=@mqiOkRniYb?nugda?Ts(GbB@CHv>A_eFm|Y;`cwT;)miNh8*tTaa84
zZ>uz6($;UhtZ7glKvuHdD1ig}!Hh{qu;**9#ODt0_6)|`jhW<~s)LkS#3^<(57$3G
zvmMuy9BV-qSS%5+=}_eg$@Cy34O+=$=!L87e;Ey}F
zz1by52B9k0!BA9v$(XSU*#6|lq;?5H_j9I=JYkq}3$;+a743pnBh5Sck)$>|J+p;*
zh1SSAs*t%fhL$)0lTp`rWm~_^!iC2rWoDV?qfar;Pk_Rb_u5q2?RY0$(Y6C=+O{TG3eDJSX3({Kra4oDWxCc^2*u%5MQ$z)q+J5)XidhtfAVt7MUVT*LkD!-K#sPHeDOP(r{J-O`@S@|
z(ILt!B%`x`CFGfd(CfPC$5eAbG9~_QKYGPz6tYq~S83rNxLs$u0K($QAZ3cm;2Ul_
zF)|X16eUZ++NDlUVq>w^3L}9OYB~nyOubIsNZ!hTB71@o5MI6q#;?$NS_{wt==NU9w&D;T+A#&W;#a%a_
z5**%5GiyA;@eDU0F*yW#pI>g(f5QP%y-RgtGo4h5G0jo090(MHRZlJVCJqgE#bfG@
z+;sMIGmvVkSuI}?)2?(=HFTC|*kag^?g2qVIFUG4=`DF=W1veNm6LgKNNTII$dvu*
z`n)DhSfAm1ohwk*KymfstvOy+_a$R<_9-nz%c-1Oz-6+%$};6gvZ0Eue?}yZ^d`$J
z>x>se+exxHIrinaT~jS;QJ51+JQu1QfHK=20LwIV@$E(~sE=QRxasojY>-LKjt5QzZsuvpYjk3sM!?Y0cT
zK9#FOBFDI@p_?YSr>lOuidD>>5fE7#x1X9h#evxnGWilH(c&I^|6-+vd
z@nogzG%(5fy-+DxDs^vMW>e51ktoLO!WHM8;%!%b*-it0vMrN-2tiroKH}d_?g5>x
zR+wYFOdT60JBUACf?xsIL$6FeZ%)q5d5SA%qe_EZ;V`)+f?sc2v~`gyHVPyrndEjoPMq*KuVk!qY2UF
zew{&bgP(5vy+*z@LTsuG>$=HIKkm_`(?a%V
zi-SJ|s6jYKd|#S!V0!(t6<6xG3A)&B0e&WzD9@?zND%_DjFmT{x$R;c=nms9VlEof
zPX<|>gGlU#P?E_rM~$J@1RVB7X7cdq(}TUr1dSp@f3X4Q5k*-k`JXaIUd&|Uvs7lU
zpY$Ec973_4*lV!A3BISHPq>^o93K0
zk9s9Ee-O92`UFRXtHN1hdS>&wHkL+x$cbb1*o__v%7G43nbFT1%7q*#wbmzbA|iw9
z{f!+^sTSJZ{)nH)ClXs69zplU=Lz@qDFRv2-F9xv;0!`u&2l?@-8LN;W6oSbux3yF
zNbMVoENh1Q9CLexT9Ng1e5o~u2P<^bC2Uyie;{BXr;)r~56txTmKbBO
zbw)*0u)gs6s5$O#>D>I6iU}iYviuO^g%;?^c8`(mQJ={8xW4Z?%A*gjM~c|QTGGX@
z-$mKRredlti4(FY?{mN}`>PG9he>&q;c-X}Ur
zD!JZ+m5=~^)R2An-us$IY5ZO9bc+{Ne?IF?XweLL-2NI?EF@!IJH`5@V3d7W-N^7B
z$H$+^NH1!&*O5J6Yw{*;jbu3D`XI~V1L2zc=9#=U@`u!w3rEkK2iv5m*x_II;vlx)
z%YV68V%9qg)wk3ohOInbIc2~w1fkmEliv|vxtfEw1Y*+2%p`Xyc6QYvcMd!Ie{c~T
zI#(+~+i!KbY96F~hqw2%v;|xUI1%Z*=Ej0msfvL}Mxie;@x};HjT{CB7w|QDrjE2w
zv~;ETwO9E+Dlq4oR3~?SR1aC~ezA}6F1U~C%UZVjEQaU$5ZT^S9E7^dewEQyu@pux
zwRpFpU!CSb;ay=cTD+Rv2Wuuse;T;fzF75X#UoyF3DQ4Hen*J~0f>M@;b|=tTXvfA
zA*%5uF5&nf_{jc>e4Q6A9{>0Uc!dXh)WN+TaNE?@s-=C_<#imDhBtUfRT4FlQRycvYJYx2ie}YZx$&u#A
zB36wpHzjI?`R1mtBV4e|UD5FlVDUZeX}d^FZ^xNwra-F>7)FNQ94gj;L}uE-bGx8yzrt
z5-~fbNC{+j(GB~b2;D+l1W)yf=e<*Z!hPGpB)I}Iph%f_iROSnf1=2>tF;vtpBU;=
z`=A#1CJC48)_t$=JQR@|v;{qSom`Qb&^~kUo91LC7#&1aM2{DvIW)_lU4rlVO5Mx}
z#>uNly3Xc36D(2j$pxcA=cB%p?lt?zqWs#hDe5{{xjXZ`-Nq7oSBLjh^4f-B`OM_Y1a!WDup82Vi(oDi!RwHdl+3
z>WpN{)1`Eg>?z6#gGXawp7xyBI*wm^`56r4;6@08FghGDO_JNT{5+=|e_<==70i@n#mCyEb`ACK6r2qH
zkN0&6J_D%AQ(pOPWxb2un%PjFJ;i2-_NXon+%LFVMcSP6rb(G$07P;3JcB!TU*%Vo-KA#+rUKfCR5uFCBlziv0nCkU
zC&hg!fAt2-dgIT-c#@!RSQzesss4Es0V!J&`C{7I>B0kAjUP7gc+)=O)xgjH7|x~W
zRJ(pUVitgl0NXrrnBALm$TAP{wGq{?syuS1vSZ@wlFh^85j%0anh4}4E)qv*kw&Z^
z?~adcl8EqI@b?ABnPLC+^+Xd>e|Ywd5g>$gX##kwr&apGw;bB59drNXLjas`wSLT&{mpQ=>I^H`#PTsWnV*j`LZD>#dc
zY9G*Biv1Ya5RQquHON23?Pi779P#5je+@vrHS}u8!364DhfMPS;2hD-krfz30n+gn
z!uJfK1}>!e8C0|oF!fz1HI4P`X>vrH_k{7}IGmkM#5hHQy^_*BJG>|%XbE|f?`*Tc
zutGg^37wN5K~I2a?wI0zSBY1UFbkU=p=lXI^#;Ty>(=OCbIfE@KP!SLlk)4@^r
z)9XlftPJC%hH%tugIODJ^bA_!%3^7)zobGk`x}a-(5`k|?3c`5P_{=ci4qyO0g}iM
z^(bl*TL_pN6NxJVps3OeLm)Daf3?RR$mf~~P>jeu8zxbQ$JX|F@I<8!Jm+@v%~S>#
zKwK&p&k&~t(4n)VR}fcN-aIeFe4WWopGd==VT78gwQL!15Xw2Y584yX6`PnfW%%x=)2=Ea2~yqU3~6e<}$MCERRH
zS3ZR8+I-%hL>jBq~FR|Wx-&f*Plhy^%j}5QE_bhvTS7%@8
zuIH)zxxR_*4yJXkkqK+ns77sTY2*fO$oG0W9!a*i-;A=MkWTd)wv;7*>)v{iPwKtF
zlXX*`SYL#)>pXYV>&=q!BV~6o&oUu;?F{@pY*_R%NZfclGGDk)f1y=nvJ9%6F4#a@
z8$deygddm@HoW>gLorlndg&+(_S&A9&bOxh(2dpnBi2%pyWeC8{uum%UF5l;AU9FX
z^X-ijgKcKU1=wy8GHIR$`&AXmC3OU39nXyYkBTskZ`Fb9Lp=H<(tz!5niS
z4acng1J#pv5qLWlf8@LX80Q$huCU#uYt?84;jsJi@`XVts;$n-9r^}F!kv{6B?ba@
zb&ps2w*l|C7cY$EXA6`CvR_=T<-21F?0t835EZaBOIak)dB91_Xe*)qF9FBA>OWyr
za;t{~yRfnoi!sN|aK<~a>%RYD@if)IaRQK`_h#%~rc66yRtWEo#9Itma4I(|o
zEFu9n=GDoG=)sn{tS{iqWYzLgZ#5T{xV<$0}-99qBtf1#62Et_A2KqX3rVdFn;Fx!!D
z9fZzu-;_dW(AojoRvc!ThiCaF&2z~>~r1@JGI4Iotu?S_ujHn8T+nSNbt>jld&&i{D8Ot
z0tZj7G|Pft?y-1z;L4nNPrWMlLsIhs%6u2EQx
z*UQWMqV$e{p`!v;Y4PfX75PBrQAg?PA|Z`ef4b@C+YWj5#v66J7`smFqfr){8jqf>
z&VmM;`Z^2E!#KVcYWQ#0*2sEMfHNGMW|a}2r-9O5J#>x`$E`%`DmU&C^-90&__GR!
zhx-rPv0Ky{ZxH`7=x`2U2I-rIU^Lr&Cdg8L5*Wa>?}Q48ECX>oNvFA`YUZ8FeSG`Q
zf5E##mf!XQ{K?ZG!M7C&>9RI*6Ob_8fF&SwNyAY#M|UJ$GLZQw)&#ZiBjMy&*li`q
zgaV{=;KV-6X>?#jhcDG#9Z^q%a{eblQRN_u?H!>6EWR|@g^a3rM`%Ke@bxNV2l0`&
z$$pAhzFbv(qICMf^g_Kq?)ocp|F)8`f49yZ<3p;0jK8T43h9S0bQ3k70wDawC9kM$L*TDl~rojTwU(d;NMHjLR82iry%wO#}VDe$g0;VUh;+`L_;
zCWX2m$j-pY;4&U|Wz=vbMIt-lf1#De?feD?0gSP6fJtZ3rVB~`Jclk5hwAKO%FQw^`24cCv2%R*
z$b=00SgHa(GeEbeC3M5mBb7px0KmEPe!UaSry5L^ff
zjlHpw6g=1D6X!Q|0g*Z|e%s+mJEUH2d$*@K)(-
zKblz%j&sT8{xTxz(-jpWNQ!}8ZSEBMgjD1h3t!ok914w132o&hPg(}4GV#F!LdZtm
znIlRVGo{dre?0!Y>vo&Lf8|>WBhl=2Ynov2F`)<%Wsqm|jLsz(ZT1q{Dt5<|j=pc*
ziv++;dd>S$#9-NU9B-EyR&=`KgX`TEA|oy&D#PdugSd6=HxksyPtH!(oO~|lATFFG
zyw9q)ziyqc8v0oHtnjjOv4vmY4$yT}a;)k%5?k#Qf6MD1$&(@tFn5HZIIZjZqgv)6
z@=N#7lvab<6Cpp)n10>GJ-=$ta!sJrRBE=v8T`4Y>bm2#akqE%$u>pG4%2+Ed>LEH
z9I0^OP($4>Q+u83%tE8V%$|||2r9ed1VzB
zkZJ`ve@`^5M0Tx~2k>R2s=maZIEkbj$R6S96d&!R6SM_GYaB)`$4rDF8iS(*bX87k
zScjdYfS-4MDUg{>6m^C(by=^$pVvMs=kf>L`p-C=MQKNX0lV&4nJfZ7XAPlPO)cKZ
z7)eJ}yRmVBT2`=-#YV~!QQzD*D@!Ox5#Hcke|=9!o?B-Xq9z*+vCcODz#q?ka>x&l
zMMQp^e=@n5;4#`>3Ato0y?sfM_xR<<1z7){liRSuZxQX50A=RP05$Zh+6t~L1oN>5
zFaa?vnA}>Y6|+u(BeE71N}e?Xh33(;aT8@~#glylA1SRK=rFROHI~2TX1u2t(z*dY
zfBHzrN~nD!2+vVQ1=ip1I6|N<6Y58+d|VvfAMG{TvK6KO)2%pnSjY<#g-0?{R+=a%
zY97<+?K(az`=l`1?1NT$-VR<3_(=vAbfR`?v{_L5Nz+BBRSm@Q4F(vBi(VN+(Fx~$
zsK=Cb3lY45WvSZRG(7ZIbHjo=&cHbhe}!+rVJSu*N^1iR#5;WEQIlCGQ&k{-qp=H5
zT(U5KGA|LiFp?mP){yonp;4e-V`jP!J(avO0Rf{z1ZZg2!ejN+}DXP$g22Cr_G=^j1oGgstk}Nr6j#ae}|7a
z6HKkQnNt=#%~2*@JU(TMFSw}=f^X@kC!H!Z2}nD!0@*9ap2)mEG1{#;>ULjv$u1T;
z+fyH8E3$E+=0hVvnM)&wWR=?L81Yh|k!V+UC-+$7ne;y%&Y|YEIbua8bu;Gn4?}zT
z3U;e8Qpgo`8P^M0DYw@Q1cn81f2P!Z{SjqLc#yp@nk2tetBowd2HrkV@A}dwT`2C&
z!y%TI=M15Vh+!u96$f-qHF<0P*jOnTl4!r)UvS_m^&8kU(}x(WHTvcZae?Ha&-&_B
z`UT&iodA>#nf2{SU?9jDtHIQ)1M;ZKt`IlP<%b@8&5Ph1vhD04h5~Q3<@=7SuR9dfR6p^9y=~7~>3(iqX;mM=kYcSd&
z$+T`O(|At|SW4+tS)(GWf1vHk2US!q`;n}pq7hjn6og7Ww7bz#U!9JbKn<~VOKKB+
zZ>68?rQ}T+F4#Wq>dks#T~z>+&n9HGME#Dzkp>}*jrxPp#{f02^$%OfOSf=U6V
zDKpSgpMjU;>;;|4)0h9(wmG#`0w_>4+jdQMO}1^@uF1Aeb+T>S#x&WsZQHu{<-Xm2
zvDdfP$}+{I^%}uXPMi#FZubykhiFb3OwaxTHN#jD!COg-wLhR~G=$L;EFGj_qWKgI
zOR_(~gf51$zgwdre?GI|Q6wQ?b{CkFX*;{?d6~dlP??Xx
zV>ei-)K!NcLYlXxE+eVdG!cA~xdWgCDF?p`yJA5mz~+D}wN2@wJ*JU+4TRs+%d5r+
zd84FINnh`u{Kb<7U27k-_rSMAftG;`kfeZS&V%^HuCQ1ak`SOlLw^@j56G|-L>w{1
z-#CRj!lw~m#=sY!FRlCVH}*|5Z?>S(dqHTV{;@qvcP3-BIT6@kIoGeH6$mv96TLvOkVUE5_9?E=-&BI2%~kr
z;6qSUQwX0afFE8LRDZ)pGoV8kmiW=wNA-4!&k}--my+Crm}RFWPR<}QpPvO#cw5Nj
zS^c~7L=Ipg=anEc84u3E
zV?&oilH!5S8X+5~c0JuK7svIwK~q~QKPs*W3iK;H4ufa&P26z=UA&!gq*1L+hGl|t
z6Q408g(IfkTjfYVjio)pfQdafW)JQU*>&lnFBoy@c7GkB9grIIu_xTtSJ%$I>6|8j
z!N#9?Wl^0>iy|%<-*nz-MNLcT01A-@cJTXnj0uf-Ed54jl=z)M_$UVcDbA<|N6AO#KU%U-YcxhpK)zc3_bRjj>s*
zl7Bo;bk_vKfElkW*QdYnx0G{o)~ejnrRc+T7S9iADUG_9koaQS;hD808YE)sFY^BL
zxT0V)onKi;26>>CReQ=SBGgyOPha*f-QqIm@BV<)933Pz``+~_I>^aC7iuzZ*Ff=X
zfl4<4Yf5oCbM#KhSg;iqW=}cSp3beJkbk|3V^8(C5aFGfYOiRlIoe4O&IVCh@0|Ti
zHFh(qb35I*r+gRn%U1YAei<<_G;i!eS%s=`H;dLbgJ;q8(i1~dmQy;zh|2#Z=&;V>fc+OcJ?Ap-5K7Sge
zWVokBlhoRI1YCTmjH}dV-7)OSX}J-T-zniLiHK$UML&*g*>U|LxWvXi1WZ$O@WCjL
z#S;D^HqGDje<2DD*=j-p%B3qN70xha+*$EfYJF?)F+y<5SGD^N3D>FcP{bHZQ*27H
za;jam5WAi7nQU;w93_}C+)58H3V-6jkDum1?MT=rq;p5j8d=F{#w_r1WB&bsjWw^*k0JWfKS6=`hH&(oBWMM;fyyZ
z#21A;I~J@L$PizZHcb?=w12|s46Lxa1`S)4)*2Xa@NquqB*n8SnQ5XsXA?|+yW~l>
z;?sjM@04dYHlD0gEvNCa4}X+6;1hE-o(r|z
zUz0YbepbT}a^GnNjgiw)8lGD#PrsSwfXx^V#@Yjt4m+R4#L}57h&0)iL9vcdUszkn
z=ANZcSJjLOWf_J=i`cS~7@~NH^*Sn^;0x{%{JN7~C9*C~24Ez106P?U_we)3qPZcAjGEq$Uq)r4VBpFCtFbPv=z1S$5vB7gmhM7|F_h^adNp2n$I
zVn<*0{^9?%|Cs+*)NiQPIk%8fM_P}1o}r4jNLXiGKBWiMR^)p2ALa`6$|`$};QGID
zP>T11h=TgYew#^3cYDk%tHO(=hd`7zB?uCMY%0aR+ov4wuw~qD@RJIGfE_9!>ss$-
zl;f__JNoccR)6R8HP*(=%4Ac@)Q~vkt%Foh=9>x3rRtCNFe|&9DytNs+CZ+?tJv$d
zLJ?dfq1@IYptN)V>Ii^1gssX4U?T~#8R$8kxjiA@w4i1SZGVs+UX;dtJd?;tE|=hqE}!q+
zK1QEg21hi=QClXY*2&Wj%-pN1a20x@67-1`s-
zAx;Pld^i`K52+D+N6j)7^;-7&iPiYz>^6fh;?IB5k}zN+EySjG5Mlosk8YmeO4Hw13db6`0oWXwFvf;v+HOu!hoWUmBuyx3kc8_2nWP
z@+I-q`hsMmsfEnVy>EWCvdDA#W3%^kbH|z03}Y-?D{q~7z1xi-e|u*$MhfFSJKBWG
z!hfr74cd!7{kC*r1_pflhQN8lbJ;~kEoUaBM4?rRQ#gNRbaZBZXj4nMHv~03CMC-7
z4krCVkR<;12T@IF*beY-OP(CJJ{!SR&Sf1lh8I0y-GbpdO7EULnw|^XX6GdRU-?jn
zZ43xUxAJ2mOIowk^|UohX-nUa_KIMQo`2yYsOqh-+II&<+Rr21dw=!{>=mrW;4r8m
z)i-+czah;G0OQApF9acIg`Ri?TU0x6z1Oc!xEzh6t4-vpA1vt8%qUNQ9|&LwSh7pNnoL#hR?i;eVK#7pbr&nX)ZuY}?oeQMEelOkWl*OA%l^8{(T_J+oe4}Xi|BOIKAazKMCRL-@hG-&USL5k;4rZooT@#60$
z(L-fMwO-U)$toj$1@3^rYJ|7iy?+GDq3d`>C2$odB`m+qzkI#^QueRrUk2pRwl!c7
z(R73UG0%;<3h#QQHjtBS)}he%`rPs87OhN`I8^swz$5HXmBR{8!DEFoO`9f_~VGnh-M=3^#5}sP&^qQL$GSw>VCjR
zs2f}(0y1seHVYOt!HZT}`Lmjtrr*xsQsN|NRSaWMJV5k16Twt46Mr+Szj_g-rkLM4
zJ-V(1cs^DRKB~U?y26AZD(aK>isN|fLxFxj_mvYXdO{h
zt;H*LbO$zGw$86|PPTNEe0+KMYO4n0O>^txqn02Q9Exk%61;a5jUT`pFe}%VK{&|+
z=BX|g^+62>>5K0tv41e>?-{v@Cojb+%)D3i>tmvj@Hn!8Kt5A_f)j}TZHWqeD-^Qv
z6~Y>>-W#zP7LK}QuwQg}**)A?X0IY3I3wsK7a5dZGmgExd?^tcOC0U>${B2pN&*A&
zActFd08TiD{6cfY@O0AxB-XJPjwxh-F(5$kZG7u0y(U+o2Y<tAL>h-I#(?f|Um
z!CR|@_-nBw<~aflZYbMlBx}mR5!!aEa-H4RWZU#pINvvXMjNv7^1in77wQQQ63myqAKr9>w>rN
zL{d_%gF?y2+s0*yX$nY|n!oDt0=_xml|lelm<-aQzbpn>)~xO|LVXQm4*yq)L_wr^Mp&sYe5g&xBW
zou^d;{%6Q9%?K?}!!GHn$-`0#H~P5rB~~55c18~7f^P8}EBXD|u+D?`3S&giJsPEa
z_P#m&-^PS-q6Jv0ig0C4xV8~7ul}D@BVVDKD1TW#zA1U2d5@}MA+T6H$27kr*>Q{QJY}T|g4{$3>ux7D=Q{t^*_lc}i6C8T
zMSn#@*Iq;VeGK3|@2>Iy+rP-zWQ?oC#PS)4d=vF+)M~wojTvN2bI;MNqMyTsIbXri
za1%1Hm+HbMT@_B=l4&Z}tz9`EwnKyCMx77YtR=yD1ylygO>}S3Y8*}v$ZP$!9!!Yy
zLa`JM*!s=-_pNr4=xdUi(282OfOe-@$rfX`$=A5xZK)eMZ70C=qi5U4cqU
zgJZ2j{?o9h#Cn6_I@=*j`#Uxy`}K7fD@u*G-LbU3&{uFbg+7F1_2DD$>QTq)U59gR
z`VHdh(qGi6{+KWrb@3vrmNI(6&54@P)~15NA7cI^^%p;_!kDRVc6+OwC_mV;5W~(*+Rtk)
z`}|-w(O2RFYE|1POoKC4Z#eh#Y?L@tk692}p7Xi2#6^`4x{#n(!U!j?=!xEhY@EY&n`F{uL>YAX5v=JCj7Zk@|D4rZ)Tjf%CUwx%gmjh*)
z{I3qF?Y=(_*c_yGK389RzA&)Cv1ff+Ck|HB5JImSeuIkA)cKM_M1MZgEvSn3
z^iK=%{UD?AO{%BDJc(T(E1^6PIXz9Yj^zv
z9F?`CG&dIx8z2T-BVSBArA?jrWqQD%xTe%-f?zKkOuUn1_T24zp;j_Us4LMa%X`Di
z6<|*7bg(NgpwX^KOoVHab>EAis&8|)XIQE%;UFfP0B_4x7p!dW=mYA;cc%d=am6dw
zULMV69r<+Y)0OF?0Dt|h^cUfvc`ncBxc;-aTJ5Jv&}hJV}x|4+ZQyG(_~)xGmj%HbIe#DSy9ZJPt-nz>>vSNBzL+J$5QZ
z5eiOT5^)EiyDTX$a4IgworH@8v2s?$Tz{$AY{hicJ05^V)+w!%HY$Z>>&@<$SkxfI
zmuhX6sy@>t(BCo+1on2x!u(cu6Ci=*cPdm<8Se(G5vF`0PqXnH^XagvIUhc~CVG9+
zbHF5{PJctjd2$PEW&ESt&Tvz~+^=g3=XXv?VJ&O|)`zT0_6soM7g;Ui>ZJGS-r&B6
zB6mbBfjG<1(RdWsQrn{2n}w$88fDzPiy#(9$?k5SaU;soZsZW+Qu&X>l2z&{PwIqt
zTxBR*1-QnD7u`~j0&a82@KM*W7yUGehI^1S@PEK6r8cx8t>L-tCUaZ6#D9tf^=hW?
z84JCYzcyJS9|CMxSfhxW``dEg2vBt%DPnO|M`pvK<+xnWPn0-g*`Dr&c2D6tv!Y54OBiMr%E|aaIo}R0~6BQA(?xoq$EhQOcgR`8L
z6yh~ajhaVR#(5`c^hs9zklqQwZ(3NJEVsQ8T`$Goxy!*hDaX7+xvnT+-Sip1~VMGihB>Kd{YA{N;
zs#eb=MW%s~U$tGIr6Nr!7<#uo5hhnw2T!nQhrxO78x%5kzoMfgaf5b*7^z9a@qLu;
z#ZIO&ReD6~O*F#%1!GR%uzy1HU|1MRU|l*}I#|K8o$<2;a_{>MIg%c{jD?vNmISxv
z4#hn4gSAnMLbwE1#HL=VI4x9ZcKqUVvYG7s&UOGOE3n$I)x-pY&&b~gp{TPi~tB|L%J3l}$
zuC6IvfK?_CIPKGh&5Btp4Bq`D%z=ZvhT{=k#r+yl8lwHlwDKEY`1bhb`3?iT$y)_+Z+|@cdM~2KcJ@f)8P_}p
zp$ovv)mlhu(TAO$e)}rm&*)lN$Nx1tpIY;9B&ky;a0&mk<}1k}_mBr>Z(S-7P3@hF
zUT>tRqdzik&2ZPM(4w)|ysX)ybDuhvP#U|~lYL${TKG)}b^4EG5{OC-pyWU~rC>s4
z@PFP^(yR(Q6o25=nZb_zoo=FNNssDO3PO0_aPKzStD1Ndx$Wvv5D^~4*+MhDS81?*
zO<7HSJwE+zKov+u=k|c(ASs>4B@^k-L3_;6faiM8H!{YhEFjas_WM=8h8kf19sW~E
zvb$Nyk)&7k2cEUZQa*i;oW?}3tX4>6gSLHmm*qM6M1Lb1zngWq_kQV#y*9~;F;o|S
z%l1sLt31pD;kMBVS$wh*iA!ZA4yux;~+eY+DgYsn1LuHq<9
zXs?MpX=^cz6_k`SP0i&o?%MP7dbm*)0lWavbN(jx!p**X)36V&`X+x4Tr=EWCOw`b9@e6!7oxu-%i%WYy5S4*;
zE$_U8qu(VMbRsq>c|!d8sW)xGB^^7!E@xztdiCS`Dh%CNJG%PdyC)Ju6mC)T;3
zAAhX7%~U4w0wGIDR32f&YpLP@lp`a*lJ-*Np!gMMcu9#ofc-=o+d5-I;%is08Vyh7
z@MZ#r^W+zne+qe?MV3sDZJ#1;Os97^RPCu6BEsX&`1Fyo089)^-^%Yk6_~S~wBG7u
z(>a1B^T*#3ZjixWu)2lgc?y~9I&MsWpnqUx(KyJabK;)40vAfVf8w%q+Zj6MX0=cA<~57>8~tgYU3H1Ch8U8)B@H45npcRXtrX_}a2qh)
zyxecIKdxLZRuK1^4Tfe~&Sbek;e)>%2eCSvT^-)`hprB
z@wCqA?m4dtv9<#URtM-PmsM)CIe*ozMPP)V)jZvA@C{0dhI){VqCs9YC?72*8P`!2
zse3O0|NB>cW49uEFgqp~zLP=mIjd*TMAU^6Ool(rJ;LE{dY-r_H-CI)48Ux;(UGcE
zdyXIOOmjH`d+Rc4NX5)E#dGTp-$h=OF7vz5_uO&wogt!OBZVgQYF1PP&VPkDKmzdh
z^-{%WxMiz(-#foO1Yva@v58>Rgg|QERt8JnZ*T?%8_LOp3h6?%Ljwg+6+->Km4i;W
z=#ev?Y`IN~)6o5(uUmcrl?v%(r}_mq?)lB}YQ=xg4`s{9$Hc1w{#pLZv6KmxX!E`N
z+{}kn;dw{Yt
z>&Fn>z2p^5xx5`YS7|jBzfrTTRST!UThdFBfVKVWCy;hBEdvdW-<9yyceo#k5^)g-
zEF)l&2+6H}etzo(p9>*Lm3O*Ebdd3MK{3LCqBCk)LjS_wTaeSZxMYQ@0xXKyL$
zxRjk3Zb4gCWbB1?$-6Y#XB#_EOd*!jI>h
zUgJXWp#SD4aSl!5OMg{nS(kc2w;NnhST!#F@VA6mXSz6{)$VvA^;ZjQh}L!09>k8V
zLdC#Y&rjL)dGSVa8yQ_Yu$n%Va{)$<+I^i)3NoS@kp|oBt+duSRC^}aW95+R*4AVn
zeub3XE5>}{%W6KAzDe?bd`c1Rhq(uj|Q{V#SGUw=lkRLi7-2ax4r1CwaT
zGu7A$g$r;>4d6g+7rLrm&O@>Q@=Qv?c+14}P&j=~P&JwPdmMUgBaH*KTGMYn>{LhVYb#%8ThvG*=?gmr6Vjy-uwUJTU%M{Dr=
zqeg=C;s`TO{?!Q4P=JZsH_o7(9!j`FenfdI?|-LPMC>a0%<`JZ77!%juDzX)c?*9h
zmgzW;))R>qFGlKAV<>V6_j`E`-ePYi_}9I)KXx~hzHgK()Xj(uIeBi3#AW$Zo}K(e
zb3y3O|2A&%hRen)iH94LIM{?nb*=_fO~-KMp-Zmu+Sn`TXRGA=!V!kOpq>)`(W
z<8ceQnzklJv6%bDY0~~41J7H~SR=xp^0e4&uyaNFE_P*f7PYB=u(6FXL!H0WbP0nNw
zq+cO&-e4H1ZLgv(_}o=s5kz)>vwtAN_S5RS@AlB-q!Yxfy=e9-8_KJmZI`K=sEmS$
zUh@dput_JAV|5D~euaD-?nYm!(>H0!7)Kl~CP$jleTgPdXLNfpH$8s6h%siWtSjdQ7Mz%4u6y}&rLts7Tpork
zO3m{@cRdH@0o)fT7|$_!x!;nr1=xqU8>4_f-L2~dm3YllSV&tRyN@bHb~F|z>T$7m
z`(lqHhrEQ53^;qja*zA#f`4ROLSX5>`@RM&wmgjOob{G+ptj52>AzdhV#6v-So7IX
z_eEe&0B1^*RItSyAZ7pQe`G1S!+LjDu=nnU?eg1P=0;Oeihhv}BO^$6jc|%pOuzIy
z{3P@FHP^IYV@KzFYfrOZ(YmbipDkR3k7sUA>Nos;yiq(eokBd^I)9?_Xj3*;To8mC
zm77WUbkHK2$J-@6wcA0gz6d)lEH$~r>QhfslSExoa|cI9m{V>K&Jqes;d`bj*k<3P
zgW@6$U%zD@Q#P9E5rvfDdC{bjMR^wt^&UCGAnUwKd33fI^*c4Wh2=T#bFsl#3{?4j
zpoGpc6Po-_j1Ei+`+xlbKDC&il){j^*mvAjI0K11ksHD5b8&KhUi$K}5pW#Oz77
    *YL2fl6#gO31nr|VLcUiY3^SFQ2Otn=xnLMUrj>T%%Y$lvT
z=_((d?;bDa!3M^k!SI=NkIG)yFi7==xBS(XSqjlUQX5u_Du3L(+$3@5E+mhw4DxsK
zgp#v;`BwpTP3CcsJ{0!~Ef*VBG7MUHgCs?PmNfq&7ZUoEZ$P_rIWhO13&!&jMY6q(
z#fiu^Q`kjwu@{vjeYCT3v<%;2up62afuk)J7KG5{wq2^rygM>g3}oD1bu7
ztxq*MOH0wlKYxAAR)A%oR3MxL{V6qOuam%ybL&izx)8zczg+!yA=O}^<`gbl30TR}
z*jBAS;@Px2&|zM=Muj14fE+^-h6LKqLzD^}JE9ZfsP&WX$F3Nw>8j;LO=;~TQ=?c$o7;D6m6pHdK*%!3`ppO1J^`AG~d
zwLqIO{+L8|TnD_!H!KwtS8%zaeYJN@$17wyebmnzoqqVqO#Jei>bVxSUfD9IbdG{h
zo=zj0e(`V#ZgXBeHQkheCI(9d&LAq@p@G1O=|}5z-cog+Kp6oROlip!j1%ZeaKt=I
zyZImXxKQgAv*H_OtR4qNS)q?Qbm&rg?^A`dQ}fHhR3h>0K*p<<{V$eOhV1EL&0GJx9
z85sjKj0{X~8}kX{0vmg{xuMYi;d0B^#8g8NpsK8Iq6UDP2?8`sjZH2-O`u2|e@8)p
zz6nl$kq2jZ;jX75ctX(__(j@(apT4@1N^3jbA_Q%uD=ohc%86V
zHyI$%%gamH(F21OMxh;r-G30j_?bAtF#s2M3>>WX8HA-!myDje4*EDG%d{2v4Ef<$>C
z{r;C021h!;E>6M0!+#BEiiEp+K($o=cSN>JET<+dBa-QvaHq3o-aY
zjBAO%pBu^z0D~YfP=7cKiu)qhJfb;-82VED3Oc+hcJPI-HNh{I+l5}@Upm~x+Pfg37z{uR^h*WBb>?5}amR7N
z9RSocR5#Nw6@U02!}QBo4QY>ZfFm6NBH|JN2pSFXAqC+kL_}O1;0MMHz5~?z*9ZZC
z!blVr=K^r^!1@DVC^YHCF-wX8fZ7+LzmPNlXz(|Zk^%rN{zlT003hTy!dcq?jl^(K
z>~YuRw;l`v0e}vFKrjFZ{Tt#s26Vi50^l~lZxe9<5P$v$#AQeP4Mo5g5aMd*@H>YD
zE(PMElfQj%tX%)_xqu$O)nJ?h@(+mPg!%*GNV)xX#+@9{4T8qPAqWRJ4EDR$3po@G
zNBvP97+3!vm4I=T{AMMFYa9jv!8rY4fQy0s1LA7){0&8LG`;_TIGR3xKpaiqU+|w(
zbNgbc{(ssLpxY*tT%hJ~2i&vv4-?#Iq2bw{7TQUK^VwrFm}O&ld`ayh^6PO>}4kjrSZCb34o2SUW!U{Ddwc7Sf|U
z&woF8uP~fpQ2Vl5cPH&Ser$>2&lukI9MC2~tdn)Zd5;{0K&`t_w$3IQQflC9vDQ1G
zVJ$gQT0H!{8&8B+J8s7wc7M8(31bQ}kGYlp1MG8%?@5ivaf6%kwfzVEg^~U%>+PhC
zCWdK;z{u-aXHs5EvIFYoZ&up@p1xae{C_R@m$*>cMy~=Nvd1V~f3=Wux)|?s{DaSv~iEGS3mHuhULSmNCNfKI~i})ciTDb7xLXBx25`kEdIy^jY^ePHAc9-GBa!
zY4zS?PHOfCY>{U<1!4k(+FM0(>*Q{;
z;_!r2I0DxBnMD?IV5U1fM>eKu0mHVaB)vsDjaoV(=Ph}D91n|dnhQ7u;QO#CO*3TQU
zms1{YIYM<>o*xxLk&WXT-t=D>qk?w6T%D|D2w@
z{x%$dH9k+Rxu@669)haCpT63flkp;F8APr>nZ_9su~_%LI$bV`0)L>ElXQ<}>y(utXH4h_xfq(Q@Thm~<%8Q#0LgRO+K{x8#PLKzaW
zd$)zT`d@jii$uWjh@a9I5NYIPkG*x9;r946PSOK%){gtBVT=F1h^9R^Q-3w(%PrEw
z;Rzat{_<7^Bklpr|m>h1y*!x7&p~zuk5WLgm%ofxYO$x
zoR*CX88g3BrO6@(KDa)sVc`mrCDe=)9a*3EWrA{5?`B)+@
zAq^bbN>{3)_KJU&1lu6y^mw;G{eD}AY_MT&ar9NAThhfHTP-(Q7Nz4E_2smGPTX7H?d@%tL-gVnA-xd1O-t*Fpw$05~vQd=Q7at^k
zw1SqVU0t&|qYy4#Jwtp_ifMvx$R{v6oc_b_pN8BL{*sBF7Q!++kNCsoxs0<&*hV4AX=rG+$g
z$};u|lK?j0&{08uBTVtoIP^8~^`txRG#)*6dZzZQ=vx>0+V&UU*O()ASXN?o*S&?8
z66RS~Ury+{+de5FFzJ4$XFMctt3RswNPfZic(26}Y);hQhAEEx40|sQh8lat5cyKr
zeifjKEq}XTw!^afP*v}l2QBi)_iO>nfCG*zO+_50)Y+6T%ZFh$uhCv~(VIz5-LE>W
z`xy_vo9r%Li%}|gIX-M{Y5_P=98jR(%f=w9xg0<5hhs0Li+bLvDv{
z8cQaPQq+hH>%Fa6{c_oOMH&vQS+422|7Eif&*lCh<|A!mOi~MxuV19N0Df4a>bs*^
zJ6DMu7WhWXTZJV2YnN@#C@oC`v6<1<-{%`zPQ0c<{;0D9sHW{7PCGX0P-QjbGZ){>KJcEV
z{276EE=AXPrZ{2x)>vaVRD9n>6BLn^>=Y8k
z;aAUoWouKJ?oOxdy9fyp5jJ*}^Y20*-sG`AaxtwdvH%Eg-&1Aks?>Qsp`
zlt3O?CkVdxv5#$1j1_T4mKBEmUEcW&O6DV$^~gTCnppaudmV~ejH3~RQr=IO*M5i~#R
z%z{Inz@|&BYBrCwv_(sn(0{5rcLYCa@|80W4Cwh~?@NV^9p<3tuBoGQE1?W9K
zwV{xsr$>JE5lzUvZOaYdgIH=g@<-I4hKOwdq|V!bVU@@+IyNPS8j%xwk^?quwxJ#w
zx2I_M{6X@Gjx4oNdwQYUIG1ynsN=o2#QjDgIX`;b8q8w%sy}aRgnzF+6FH>**1qia
z?IT63Nuc288N?n3fw;oK`%4RbV;m&AL}5Ya&T-Ny4+}l$x@Vi$@xC?Ph`8=}w45Ex
ze)M{HnMsw2n*Yt(R~RNhJFn$~&`CH?u?T0*;fO+F*OA;hMLk(j(z{2~^e*kTn)k%#
zpWK>fe-}ZyL%-+AHGi>79Qa6p!b@LgB;c`5V||)Tv~NQWYzC}IZ<2U3iJ!L%_|B?9
zW)I5scC(^csTwvZ#AQLcnuR8JnB+{w)-ua_$4ebrb$uDI5crBA>p7&n+fIEgpOH4^
zo0`x431Z{@^n&xN{>ere&HntN@Jtf!qINQN&rpoD^hobb*?%dCS1&OAi_ls@`5G<3
zz&#R`@D%QxhxJzJGx{@26qyBdyz`HcT2BI??Y3oQ@fu$q+IYD
z+}%*=(9KRFbAP$N<)-Lo+DHnlmT
zcQ)~=C9g8$MQY5IMi7?|eXp*+N41jHnr0d*?l>k{ZGTBwE}Jlv)ZD&MmV=Dw_`uRu
z3!yhEuvWI=oDUE6>GSorsgFL*=ssaRp{>{%19-qy@u5FdZBeJpZ}NG*?s!Jz->Pv}
z;mCxssvx*&_U9z*o_6&l-uUdZ^US61<@Mz_h2@);dThPbT%a58vO>2G>^b|Ig3dGV
zDN>nfcYpR-oxcujvuC~>!M@m(dnk|C8Zr{mZ=Py!Q!wLyC<6(Yx;w<&U0L$Uq($DO
zY9OwM{W5{N>Gc58MMorXVz1AY!DY7zEdLro_T0J
z{kWA5iks1*1q7soNGz7$OaO7JvN6mhy9S#p|Y={tu&pG4%BtVJ>U2xsDEZ)~S82H=c1n7pdr{Bkrv*DUR+S>>ToCE@{lj|sen*DtZz#c>
z?pgM$-+Zkf9f%;bPU<+s==z*9JbQGmP*#(>K9SKq6w7)PZ0TM0lP=2JQ#bfWcu4JQ
zuYZPH55XK|p!yqnNrwUTr1@FruKE!s%y)hX&e
zt1xkG7=Cr;8msxJ0Kuv>)Q}X6KYnv$9=iE+MU=fRo-OeIIQ_FtmVR&zO^0}?0X_;!eBy*I
zx|J)Emn;u9OBZZf
z8X9eWUHO*SUdYvG}jW{q}
z{YJHTs~vk`W3)IO9hmnqd#cOvF%`!$2ooB*(8au;>nY5)D<;0~*m|n+Hh({n-Se0_
zfG{h6Z)8+W_B^rM^rlAY^@RSHgeSQ|6|cY88@5K6I2afW$mTFDL6)+Vlt>1w2vUDc
zk+`+Eb$?54rs`?KgOr`xX1eFgl58m2j`paGWoax7bV+Aah-}jxo1NL^zdcRxKS2n2(jlY6^b2L0{u-23yfMo
z(^~hMI5ihJITPCb;N8oW=Sw^r3^^ubiiYck(Bvl}Ja-P%1Tr$MOemFjPw3CCS7(lG
zqldMkO0(_P^FrE5oa(Hr6et@ni?3`wR(_Vj%27Y6@HxP!GuG#n`hOMOeez;6H+SN+
zO0zc-uXb;@r61H(3<6kt9ZyuFKGQIVCVQ22d!Lpo>Pzx?GJE#!&_1{NS>feTU|Vna
zf$t~~5(K==;c<1OfT+y#3qjN|H%PfjGi;b;(npN|KtR90@zLCQxpXUE3ind^F3ny{
z5?$U{&77d9VKuKZm)$#$N?3mrs4P!dIeNcWAiRf@Nux5Ujr+$ALI{iLybWM4H{pD4
zLMjp7Iw?-gvPBuQ7Q|Sy##jR^L2CYByYBTB-w!H!vS4ah;Lp&Ke4cCU%1)GOLa;Tn
z_?T4`U(3oKSj$zbbS*Wst|(^sqdo>7Au96Xvdms**T+V(ouh}7qHll8$pV&dFC8{Z
z+>oMo3hTUv-@HTOrhdmErlNf3`S;m;0L^=APU8l*Q}#z8^sjULyg!KK2=~!`rilbe
zNI$S+!d#j3>`+E+lZ)qel;whSmG0&E#p{n0B}EQSu??1ZQ*i~;1#GJ{&F$UGk(*Ks
zvNNXENsMbhy?Olnvn_u)IOxG5r)qZs+w7E#Em5zftl9&eyzwm0V-2Xv4MPT?@3izr
z6u!?;yVWAXCf7(i#;Ip0b%E9A?j^{UV6QBnWFd@ay@*Gf6VG2uaQ<$l|NDv7PHzRO
z=1Zn-c4{ePB6}DERW62o|57r-FmE}EQu$B)^ezD11fNbsaBY9$0d>-8v+n&u@1PEn
z?VkhVq59FH(sE_NrzghK^PletkPEUJc<~ocwY@1ko
zdv`+Hv~D}N*UW!_yn&3nP+_p@=i3dPLZE`%l;Wv{{D&jahs900o02@t-$pV|bsjP-
z?B&w*#FQ)tT49riBuDr@ql13q%%H4(n_yzocU2ewy7Snoc4X^hL7}Y$or=EBJZ1+gy*|v>(AA-Wb}x-
zGk6xyh@~g@>0Odwi>?B$S}A$vHhoo2^VOO!B%&gJcTKU%MRxSH3Lwk=+?BXpqwwehu}pW$%-98I4QVp<(YOrUy;mRvY(1wi|rE
zO-Ok;DGn+gepK{9t*bj@#u#NBm~=czgsj(|Dz74LKT&>EK&-)(J+aUnNv@flJV2^c7WPe4h6rs^DQ6^UO=LeC{osE*{<3z+XV0KEo??Hw
z{Ri>L%`b#uPO9l);9aBroOd=Vv}c)W+_}MtVH1xA#bX|83Zm{9Nj8VYF)TkMh#2U0$|I
zM(cr;c9THz@HwXWkcVThb6;1km=3iLDGDE-xB`$dXfa3SO%=_rfHob*h7^BDvvRcP
z+~D06tW6Nv>%0ImErSXb8pipFa3kL=;}QP1daA=zbeGj1+MEPT2OLD}7?*yOY<~!~
zW;1$xlXN&?WsHh744OIlnCZ~&4tAQqSZ=VTejcG-${3kc;A1#YK6013Rwh05`fOM(
zA^XcE`l!8E%=u{Eft7WKWQTv|O0U=}_=DOqqMd?$lEvr=L(9$3y@)nGy|cG*Hj}G_
zDK!nSxVnluskeqB~nDrRBV(K|i7J^HxN^U8beG
zs+m^hJ~bA!nULVShw*=lJAxpiesWPM`>vb+^v$KF_KQFYy(@!$l_3(hx-=yY!kW9Y
z_*P#+rQY*{k8b!c$KI}bt3b@*8#r~RM6EUMb5Yk=o8D7C8@IcPcH?DaWAr86?5?jsvle!W)5@ku3COk+IVw)y{joS8hT+SeIn1lVJJ*W#4;lwx@!74gW~C7$kBVE#m>r1$#`0=uS=
z&|Q?PoOv*27j)%Jfi=Jd_Otp*Rn@s0F*lR(V<&&Kxph<&YS%tYH%Nnk
zFd*IC-5t`MLk=@^N|$tZOGqjp-6GvchcuGX((#Swyzk>V&-?rLTPzr^UDv+%-gm4u
zG?eO^ED}}_3!p3n?8d^u$}R{{R(S()GzW9Av%CS?xP#1H0GzDs?A&NHG*T`=b2mE(
zSlZkTC|RR5a?YWUk-MMg_POJ7}vgYEYk
z02}}hpo^>BZ`%LLjTV|2;7@8$S8ErD<6i^-dRsR)CqXtgPft%)8+TVXR)~uYs}tz2
z{IqQCTmgTc5Elmk^wR|h0{&GPcd!+-oNl(jzXSa43P9P;5(st${&tdu{H=6^mJ;d&
zmAm~XF=!#&e#Zp;Z4Ph+0{=y0Ywr4&ud=$jGQiQ?4(tX5n}aQ(hHmC=?ydl{zhux~
zpcUQU2?7C9?k+CBIaK~vQ8FYDj&B&3xE
z0sQQ|01g3e06TO@Wx!Ta5JyL7)~;y3%O`CIt&dr;tBTse<-Z&z*g43
zYj1z$?!=}IwsUp|DoFp2IaGx9M`it&H$j7r7hcU!oTL_x0vI%7+MHFUnhtY
zz}g(-3iPwH20}m3d|k~wfB-iacc7o|KNbH*XdHY1D?3X!=%_=t3))}N6~NXIfWY5k
zXeR%w`kx`7|7)u1z_Re=KU`nbSy0$z(7}50QcW+KxiTUD`RN&f9DHelhc3J
zRZ^5<`oClImz@mQ5@KZswgGVR@B++TT+F@E*rBt)$-@Ki<$w;h70~OiVF9qQf+22D
z7l4zyn;*a$;)3?Op}af*Hi_S&zY#BhP4W-o1F%W`LHqzV=|4ySz$Wt_#K#U`ll_A@
z0c`Sr5Ep<=;SYlHDg8lEKIK0M%BO$w2SNE%{~#!z+J6u~lu!K+g7UrjgP?qxe-Jl-
zP3sSWvTOfAPR-f8<>Mk^j99
z{-c0~w)_v`fd;dLfS_~ozY=cl-x5d1KZG3Y(DhpV141c*e;IHwXPk!UatbIvF5y$A3hImeTqUo@fHL(RbjP0adFNS@#B
z5ONe^@|ecK?9T#o7M@LL1@m4+t$D=pPW8?mrg8
z0Zr~tWFDw17`pfVp@)_P@+W^8s59hWYUpB|{-~kBoXnxu)4!U`&GA3df47(m%H;%f
z`Rl^`k16k8GCRnhMsh=o;{<|UhJQGr66ZfX`;FXzu74dP|FQWU$k`p@2K+ZC2M4sa
z|EN3%w2*%y^ZZ6YkAGB}2WswW=k+HUXwI&ne~t_X2bAd#!SA7Rvvq#~{$ui>HF5KV
z{KEkn!~M^oLplF?VYpgCT>e>DXxBXc0ilKW{Kw2egL?e~LRr240ijLt`I9p=k`K`3
z?+pHZeyaa|L;Q6)vi~_y|L1Y_7uIxhfj9tl?W~{=%YT@tm_xU#mmxd!X5)a$p+Eoo
zpYgv4X#QNv|L~TSgm{1XvT#GU1q-JD^c?1fRs=doe18A(wfy^?^Vh8kJ^B9^|9)fu
zfIu&xCEDT~#8N2KKD{x#++U_>vI3Dtfc2<~P*hJb5n(ZFvYnJzx?qzEC{`TQ9GFd8
z3{jRBH17Hnd%Yo~kAdxc3Uy1bBsmkl^m3wHus7jN~&
zwU&d^Wsa76_R<}w1v4GxeBNY-+q8)_d5NHV3Z*j201~nb4;Aa~RQSiTnSZ(8tUuQ6
z893^XpgnD(+%znjp+u#SgR>92Sm8#B!XJ;>6wzb^(=dMnhhQZ?tF4n>v)N#qz~iV8
zzp@udfH&Yht2&Y03oRpN2S`RLNh0OWzB5d(k(*ZQa*8c8MW#2KQp4y;5gI59S#jf+l0_GMCbr@Avr;G8{4nJ3qOF`bH`BNhk+S-f
z`mUYJmT7-E9Ff4pm>6$UNHHT};mtDR0E@fS&sgrk&8&CsaiP+qHz)>({tm`=J}Qh3VG
zc}T%~%X1gzDwz;Bp#t*~h-9Tki}S2%Qt?7ceoRW1j`>+A3j8WmI&RXhcE?{-A^6~~6QB9nn96j?7d~{Un
znZw`_LwJ?`6=P5*m@JbJz4o%qU1oc3`pBQX>&?w|R)ibjRd$<6F|sD6`oU9LmworS
zXIX!dt_LyicIG){>G%CAH2CVtlSGa>uoNdJh1zV6MgxH+`T@yjH}4D5JaEd692)us
z0{mmctIildL=jTc>$tLE*o$#6d7Q48$zD$|C#gb{NUms3?^hV(L*=6P!O<+yV^s>5
zAMc!D5lsMN8T9}e6*T6zN&;N%K~*nK$zFeHe6yHqHkfa5a>tmQJ6FeAuB7=Yduwk-
zUx@9d=@r#4-wpny5!!m!7?2IB*I4;*`
zcLgFTD`~u*pSdXV#dlmRdMwe;7n4(>>)VNNN1Joh1trJVFo^ofCNGdxewn~)S8#vz
zcc0Jx;`x|t%Nk}#`W>{dZQ&Byo|Tw+_U@6np)*6^yMW`4#8eQh@_9^k~cMVY;?JjW8HON*i_(yq*-zZ_&*Y^fAEb9!$S6HRzaQ
z{S&rXo11Za`9&Zz4ud^<$5(FNBK?2V1`NrA=>&!#<>>L3tFOSH*5y1(H~B60DVZr?
z#Pe}NKk~#vjI=nH%NMtdFk-P6v!#W|S{T$8lxYo<6;X7nw-fN+8Z;jcxsJsO$nIcj
z>h8QNa+RQ2#Tdu1eNR4=s9ck1`qLU3UNcgyre&SXe*Lf-e&QqeiYKQ8w)dd}em
zpB#(v3L}J(IN??=G1r3|r0YL9(;YY(#{VjC)yVMIq&bLX2>axlkAJ}YRs+>KsL7c*
ziH;&D`tc#^1~sSs``Gh_$wq&En>TipzFnk_RyUd{d&v##vP?P`A2FYxP<&c6zIcCP
z?VYp7`gjr@W(1=laI-{QiYa|8!0y_n)aRmgojJL9F5d@*kagi6q_0W}!tOj@(F%T>
z#T$a;h0ftE@>MU22+IIahyK>z*KOA4H796clZ(pZf+}RKiFbJfDv
zOGlpeqt7SF0?%oV@h=|ZJtwO%TPI!(ktTd0aRlP__>!+aR*Q<-a5Eur4y8T(a24Vs
zy)fhEG2GnUeafcf-a3Ej{NhkjgX-s!;OVM0CC-P}mn+!08GT2-9^_Knvu{OP?(Hcz
zv$?a&!xx2pFZ^U@^ft>gw)Q)dmg!|naxCjUT=pE0cH})e{A*g1TCu^ja#!Jw{D}AW
z1Jo82Uvk7OBqW2{8-$DPgUn}ndnRE?n3t61=SegOH_#0y+?juQbJm4EyP5lZgM+b*
zz24H#x2U1F4|xjr+v8f|ha^|CQIi^#H5i`9b`ZLt*Y+}JrFZISG0U9l%GVYaavk|S
zkJ6YR9-9Av;Yk63+2O1pW3aAKRV~$T+9&76b9aX(KlHPJ$e+bOtk7+DQ;CZbSLYNk
z>3KHr$xU<6&TN0vJ^Hjh890+)GxpM%em~$*=#V`0#YWHM
z5`mG>j*L>B={4Ip4-sGEt8ia7KU83-w1{B&^nTtugV5#IJYDF^NztSRa*J{C9%HW;
zX3N1x)S3g11zRUnb@|s>3f!QKHJ6H`QL3Rh6t=C~$mq8e2|=lr^0h+S<}K
z9IF_b;mkZmk|xG>a|^D=S$;zz!4H7t;g)Oz!2cU~mjz+32zglANu{II?J
zK>CGp(#U_z9z2I%58L$60tLn6=yxJCkR6(PA@DvIBWcybZMTJ+R8XtTT${j=fTBWf
z&R~`Y&SMP0Ym~_6Nq5SSjM_h-Uuf((Lebq>ZJPRV!CFFf_VhdRv#eu9Vn{B0-26zV
zU2>b7_6VP{Ec?9#->j(z%Db0=jNM18vI$}BUrT>eRI+|JeDCKvLF)pNP8*=QIt&jN
z(5;^`^v(0Jo-5qJ*?bYl%i}B>WFrYC)$7!>33H?2(XQyvQ8QhloZ4&^6ZTVZ5&o>6
z&ybJ>lWk2irkR&AWr)}AHr5G#uuiHy!e!5eR_L^@N
zw+Vl~%~GQ7R>-e4#{(#k$gh5bcfNCgD}hOT`lP8`Lcg;R_ZVg!DW+3LW{PxywRsv=
zdU_{-bdH1XTPQmpW6{XgSyUhsa?vXjvosM<*#8`L#fj+-$Ie|TT$=#`flJyv`|T5(P%`0zgs1W4
z`MUv94d1a~`mVfBx+oO7%eV8YZ`p~gh%M8Eiz@4enJAAJ&*g
z@JOl+Ag}zu)MzherW0b7T7bjBvwSzsxJ>Gr&xq@`yRRZD(;cUpnoY2
zUhHD+#Y%vhjmb}}x+Y-kRm9K-_r(KY7aK*LmjaJ$QW(_+!k6vpdnSOfO14Vyeo!K9^mLX96@)?~Y(#8|=tpqh>WZfQ(ayfsKr6Y(0
zRmQg)^DV9IhBX&KJ^R!kCX2vLT*QpFc(?~LdIc*
z^H6gHXU*D%J}|0(&gXTKYK<{W3xzEj3rcsS6tC*P6Eic}8dStCZ_vH}HW>c~K%?JFR~r3ljC%DZouV*MDV7{_OZw?YS@NyW>4K;)NAzy{B;I0rh(l
ziTCTdVj-q?4ptvdauty#mugL_-1dB%wqw-OACLS!Qt+sE_r!pb12cI@FG!QW4t-rw
zHbi6-lYFViB9oFGBj}C%f#!2lkfrgQ`Xxdu{@4BK*<_ClU7>%wex+`V?q4c5aLV*o
zB#bA$k6&FC`L{Fdh`zel-jv>dS{Eryz)g!qp-veYwBE)Bu6Ds_95X8&zW#|fWprhr
zHO|}1L{;T+v}+4ijXlh|H-YpVRoETWo!EPxIkX-7k+R#;@u21qsTGKdD`UI?UUh%H
z3sHN`nUxmDZ@GWG&Ob|dH^JnL7&DgcF}Nqkza!$%ULKRqqgjSC_#Fi%o1Q>nPFM{!
zY2x)2(G2EfdtapwGiKY3|G*p`wnr
zq{=+Lh%*+h$~>fp9=9TPzlA4Uy{F6SV*w9wID_;Pi$Q2RGv7zntivxtQh5r!6MTQal$y#sgI&yU9HH_pNq5@GA!`jc
zOx=r!8xMV#({frL`fa5)x(H|Htued)$yJob;#WUDNuC3m1E>-OrR0j=yw-my!=&9{
z$Q?f8T(~gAg_t-y|7+`Qxa?$zyHtDMxBPb>_ceNCSg0z}{5Zc}_@s=Gvq)Yi)~bY@
zt)+kJT05w8!Xn3bi@%;Ru9r_R-wo6w@@;s~c+2)kZb!DvdljhawRG@sehq3&V^9BX
zEk0a&;rvV9ZOQnOp5zRpMjO9bD_173^|(jp4AZ*ZYgrFB@tAi&JR|h<&D65wcaYAv
zbQ7$z3)A-|*^joClxZ9n60vQ0BFjIGY3+aVdb2}g*|HqGepH#i(feUJ^iBKzsKEEf
zUD7SXvt8N9l#bCwvps@#3+^jY?;5xHNSXZb7wN4Qt?hfD3tXpnPl797_9zK{)VI=1
zi5b5%ymw32pvxd^5BT;osjU=fyIbEsy`JnjKjvaA
z2Ro{IN&OhL<;<`4TDf(fsL?WfZgou!JHEMx=^wB)b}4(KYvXcZ@n<^Y9q9`38C@Ia
zPYBxQNfmIw6J7`7R9BFcrz*!t6uzf$-IgRS&6g-K6%J0DVy#{PxKJp}>ZO11BdD5|
zt%=zbbwI#uTa&`LY!O_#JrE+g5}Gq7$-J?`aAbU|W8~x1FDnx&f;hsgKC!2jIi%9?
zLaw|I;h609ofO0eRt%YfO*{2h%m$yCE=qnBno
z_c)K=h*T+(?AD?}3wJ-GgROs%(P!O4W5U{-F6PzNTPusuIQG8@J=oDr4N{T|vcv3>
zoDQa2+P%UJH2>oCd10A^0Gvy00l+OMia@LP)ps}Qm;ChZ`?Os6Xow~k@2j2&Q{Hk@vTrr%yoIK6r6O>Psk-Rg~9iyLF^qF7|2&sRWnftp&G>a8E
z_G~i1R{_s(&-%oaiYf2>zTCoNrNPb>&R0MVGu)g;jvWX`uJ>x&8eGi7xg372b#}o%
zdSRi4V52n=+$sCFIY&!^01fm}lic36fpHidC6SXZs5mtPhxC_DFNF(5ka;rv6gLS(
zJzgdn{UE^S3lr#9G%J5+`4qdP=8cNo3TB#pp;Yzsc1i8YYUNs=g0QQPykI#yz;ubR
z`S`r7>(vym!~{SPGjTqG**lgQ3qOf(=me`J1lLIKOa2j2@Hj7uTa6T7LkS69wEUJJGMS%8Zg}rjbr}pfJD{CW%pF(3*9Kr*6re
zJ!|AAVrj?o+o*rketKV_Ru?nt_>mY`529S(`8L9*H5^1nYts_Ja8egoX8aR{4b*th
z!YM>K8yIV8zv7i4RZc{~HRaX=t~5LB
z#{S4cb4kEJdvy`Vcld#C=p(w5YeLURcaL++XM{5yMA3f_#4a=(bNyB~qaHBA0$9q{
zjzKziMZegWakMT>q`0<*3fqQpYi~ik6ybt(b)x|e4=a*e!+t`%r%DQE3r=x{6rP?l
z*-G--L2s#&kX5u^2dgDTAw>62&qbtAPpdKSrOXlMl_HT?qV!|ly~W+VwhiC|nE@RBrw8)xoSlhum_3pC}N_a%GZ2$Pp1uOtiI$(sc*9B22zJgRkf|
zQra@H_{bWB{p5h|Ug(z6+rf5$PDB^0scG9aj3~9X_8pGf-$C%BRk^~>;M-w=j}%1h
zMRk9Tly7p{KPg)v21ZfUp(10YV4h~ZafH
zvK$0q=lNh{$lu5}M?%d__t092$|A!>q0fKpLwv-vH>QE?B;#n;0jxZ|Xf>+^#hiM<
zD=B`&#@dD6UO9;po<-nx{0qe?d}OfqV+O&9IzZY%IPEzn6M@pIpR48Bvb|A?wwBvn
zH|{47?4(`HhPtN1dHunvUn=;eMhS+1sT1#L7Q5PHpZ3B#cXq>y*(~L6c}-@6r#OEq
zFK1d|%{6_LShXP^TYZ12EusKlX#Jcx4WL+&6YPw~*TlB(;H^t!XUVx%GKb~orf~>a
zT{kXqDuKX+VB5$PlGxYq%w_iu#A=*=*hpNep32Qc6$+A8iq4(=z%~b4-$P0H$a$JA
zADW(Pju*V>uz{=UKd%|iv;P(m{a}AqJk4KVVoDYQ4((Zwy7A+I37m)?V!@>Su(||U
z&-eS9pU?8c-kMnJ0$`ONJVS_vOb3IFPYCz5RF_VC88dK`0`E?W)ECF!%jYp0k{nAE
zf}Is4l3}6kR{C3I*B6X@8L_$W!|aAXO~ikS!}n00
z*=7h=;%?i(RdiK85~P){%i0N!aUA!`kji6mUV9I&z8*06#oktRMm9cfVC+Zfc5vY;
z?Uf}>>IHgoE9O6PKETCBTvl31kBZ)20P8mpC7-N+I*2DTsL$%|fw|`|-Le3flbEUb
zXNu>~y{H|8qg!m`dNk6srzL;Q=`S&JeYQvAf1h?~g?sAtEe6>Bs?0g6N!8=oO2I7<
z`(1RU+$Hi}2qg(oU%i4Eth^Jp4+7oWXs1^}U8ow5{XP6ZA3sw1$}+nibXUeonn+DZ
zQN3tVDEhYyCpz;%+fn$ty@EkHofc`p!5YYLq1C*F1vN+f`I9+w^ZS3{adhiI#(vjcxE+BF9{~($K*Ybr-E6jRn#}(nX68m
zM3aP&*JtofNz}HCQ8)n^r*c&n^5PV#A4_~AwOjrAf-QC!*+&xur%$^Lcoi|b60WwG
zzrBmN-sD>sPX!l~Y+Qe>>H_mRsH!!kwd5MG5-R~YUJ&bG^kkd{U>QYva*(98O#sh6NQip1Eervc1fxyZadRiITv!C}_5zRmJtibq?+v+)T4K
z(^?X%>>L4HpW%O0uQ9VOgNHlvqEHN`J{UV=O@gJ_nMNl*tX0?u|I}IBeEPL0m`)mR$40w9?;SeSy&=K=b`6yhWrn?ZJ|hvD8Z9%Pk{I
z#(}ufkm>T=uG=Pb74)_Ko*LCGr*|nUDwl}ukbz}gz_EW9kwqk}Xq@-73FgoSVLN(y
zyi~c(S>V5A~MS4wkJr#k6xpKiZ
z{Mx2QjkJGspv(I-@HnCR4ojVuW(dINaS$ps&1<=hnsnc*38JSJDI|y1*-p-0-^&i(hKlY0uE+K+IbDWMR|3($0%L
zO)RZLdKA{0QT>__imibGH^Sm~j3Le5HY5=)uaSShNAIZtFQuAd2Hq?rz-FyTx~Ox0
z3xy^14FDn*lFKSwGFul=nk}VI0}>J!l1Y2WAZbbOYKP(lgYsEMWVmpVh5ZYue#Lzd
zXTADR)5D3c6>R4Y3t(sDqk}JWU+O*(uKJj<kpqL5exznDr%36@x!$p<1gzjq(r8mXf$(Ni$|^lFeRUf_ujffad5g-1hA
zCD)42ieae0k9&15j;3-1^aopns0&Uw+!?FU8
zHJGz8L;NHFzpcdytJ-4LeloLq&tI$bj{Px08RUj=VFFtKpEq4mh1e=wI_CL2hPt#_t~MutWpgKmS>;8
zjxcW|Vd62zpcgTx;%M}ag68|s!FQ7ta;(N`$-b)qjM)qPK<(OL9df3kcT3!}y{qUrd$xBQTkMcn+`}s@(JzJv
zU(7R(I|B2~q7RTp-_vJ(CBgD??NWbzLa!z}3-@y3^8JySPRNE7&;e$QGPd1^(pffl
z{)%HfkPwFfap#HfNu5z*huN4d^6iOuN3*NVx3=pfb@j6sq;msQr(B{LIMz46mg0P+
zmT@*dYCVwLETZ!u&Q62QS%0~<(9)WWmBDX6IJAGyrEmp~
zb2NU=ZSgm`o);ZSTMc%iFZLBUayHU1Pe!MXv(^2YPIBjUc#Ux{Kx1;@ci&_NyTvhz
zemD|iL0&DuFwCXT$x2hQ|NMr`<6GZy;`BQWIaya)gG)_jE;9onNGb+gX!LT0*7FOR
zLXm2z*KGdV3*z;>ly2IbwvK=2cLnC!oCiBO!No1c3Z};!JgH#~{>BV)b<=qAXyQf|
z`=y+ix81Ev;bfP?goi1Drv{F`@-auADdc-)eRemg0aO@kXRQX$I*AwxExJUQPt;?B
zPd~jJdneb~NKPuEe3{bH$$3^u_EtjEGEqZfW?$Cc@{hgKs$l?hzPo=>-1gr7fSMsq
z&*1XC5|;!4Q{?TCm(x?7D@AeJ;z80D^q9L<5lI;uBo5Y+NNp#}O8uhky5czV3DzBn
zH<>Z$mR_tEf7t*zCX01C_jW2;bkVR$>TPJn%8#7p-VK)gy!4&r3VLU2lWb;;cbuq0
zVCFYWH#_$-99&p%LuG%5V<5qK5(>`-&d9u}MQD>^!>BzYQ$OPKxDcpy428fbq(;U)
z4~0!4f0VYNVYk4dJuc3Uo}6BgA`CsKZ9~k|aqI_Q-EZDi_yY~5C6gj@@8L~uTHVb~
zswMOEU=tHIG&O^=q+s6Pg?&k%lorbAO!S3aP)JVAy!KAmNY;Oo)WM&@B_0EWF_G(}
zgbU%F<|Fbmz;o`~Duc5{XE!}Ox#$pXXuXzK8q`g~XmMRr+QuWBFa8>!%yR31#PPZ3Z(W1buUMc-8!%wGpIlk0vrm%Fr
zqkQ(5lQLRkfnc6dbB}z6=x*I2eOEiy~QR=
z$GxArU*N7KFz&%wlO<}=D@rsiGBp$5cgo=^-fnit%&Eq?1Qs=(-ls>sz`+E6pR=L9oE1NuKeSAJ^dl#
z<2Q%nH35IADGH(CC*IPW)}WUT;E%!s*|O>k%<=3;d9AzJn~kc|gbN#=T=~0~wcU)x
zWhm;WB=k77M+_I(;W|J&Ji)FBgY<
zHvCqeHgF2ZRew&4ecvbRXS!GhNHeO+g{VI-2UQR|HDPbp4)$4Kzwo5MbY|^hFCcm~
zpJCPzyVqk=QLQd6wi3)1%J(^RQv{Zrkz5lJQQbwdSb;FV3P_U>Z2mlSfS2w?s8kea
z+=zeG(nm69@wqS4n{m#;*o4(Py8qodg&O{0r+lw$@MOGhuacEi0@kb&WqHHqWIdq}
zKl@PRnmAf}M!2|b`)seQyK_J~rC_sWGl>W5v@_ir75uGmu8`mQ9?66X$2Xo>HnUxc
zxqd!;MzS9pSj}JVnAXZ8Q`~?nMsv@fYaD+meGYCZV)m^=yPBdWc{V?85T6#DRkC!L
zAC1#P{0PHGRS`JxNvO(Lj5%^1HFdw25c&&ayfn^17_`vp?r^ujz6jWKvTLD=3okok
zPoMjMQu{_~fKGlz)_GeWiKv2Jv=VMQXW!
zu^DiU)YT*juN6k;HIV-V5C4f2#k_ytIB(DV>J>}^=hL>n+g?#sg%%IE?$(i=^(Il+
zp%Hca_8Rz@jztmN1I*&Tx#HANFt&Ruin&M{E-m9|`I;Vd7?CZsoD?K6y(5|A=Rv{0
zIF@G_-)3d1qI$*J1zMZ(&{ivlXmjFr+u3Nv!Bqa@p5ez*YKOP56mZPXL`i?@ncp`g
z$g-_Xds`mM7^9Ov)-ux=s|48YaoLX8pPVOULYh~|PNczuYhS{7kWM`CX4Y0-_N+NG
zNqF}irFB|EJlwQ|VG#&>XnSm8UkV=@1}IRzda!xAFVIt@NTu^F-%t@T{(LVZ=wA
z%nnyDR=QZjN6$~hzxIpE_ET(P3n8*vdsL(xYR@j2#CBt7NENH?gr$G)?~p-$sasRn
zQ(bheZ61>hNZ{?B5hE#m^Y{od4@4iq`h3M@^@BKT0QkU+>47q%%$n9C>sr3imv89Q
z!JO@9ad&7=%oETQe6Do;Q%i9}Ue-Bqg(~?jh$VtB;e??8@%1bChF3x+i!qLH+~gyo
z!KijG$5)?D-)*7Iu_Aw~8-6HPljg#d8cTA@C>NcK9upd_=81EiL#YEfIk?dSM@Pe9A1A}Uy#C=#7S;0yrpbC{pq!k&vhn^AoAjvx9tK2nU)%KqWp_mAE##p
zuD}2{+^{1HH*R59FlCT2?F~8%rwI9p%Aon}-sE~(d#v$~kXL^i^-kfoLLu)Gw;Hkf
zUgA+LdxN4~Y)^KM9Wc7RNPgm}cm8}IQAN?3nt)kfsOG`{{Se&B7EEs@ocW5O;Sd~S
z8GHef7A%qX!8taCpIO@0t4Rs?DQy@Aj>_PV-_e&C
zG=F;gyW3q8SfGEn9lEotWf=vBSEW9xn?8>xlbYceTq42}4*(iDvn>S_9|^eq20r1*
z{VKrw72OARv|PzNcZts&l9wTx@6#LnX?gPP%H6h=lh*pGYcUXkxiqBqb%CzA>Lh)t
zI90}q$^lWH{<{~sE)p)~@O|A|5z60H7b#_h(Qx_mhTeZ`YoC}vB1_AX+^18#==;lU
z4CC7jCYlTa3qg2Mb2lF=f3U=nzf*BbCh8mBb`2sKNlGF?ELp1Wbsg0K77=uj5(;O3
zb}9Fa>F-RwD_%@!Bx_B~EV~QLxmCQQzw0K%a$0e*e`wd)O%p<8(AYhXlW1pfeGDd9
zN+tucBXxh-tWPG!XL3^_)bCJ`aHLUUVc=QTfH9Qov-s2)F$jhC!);6U!$0@Je*8+Z
z9|P(#ZoWe4mM0OMUSgDci0IpzARd>b`^nH!Ho>qgF3G7o&F$(m-e*NxU#)nVj`!m8
z6-*hf4`P}7_o|}j>ZZ;JX-G9te)9axsaqSR07ZWaF;l{;8gwy^7$subbH=kA(wnrG
zKovqr$O!l(5>uFdh+UU(!NHyA>o{=rag&8OX@W(QPmct#&Q~o}wL+>4=Pi2f+*)u$
zZ$I`swh^&LKkBNVi$PcM1XUY7P81;Ai$`rKb~2ZaUc<(04`jLwi|v<0ODH#fJhX3$
zx@mvy*AZ@EtyBb~kQfUEvubLLifU#dKih?kIK~NS7cNv|LT_LvhNvnmAwq(Di?zsY
zm%^*Rz7>-itxlECx;aPZEhA`3W4{Jp6pq6`s}cs!r-F6iO#Fn_caT3SVWscs<}cN4
z@$!VhS#vULbt|=({p|WYnql{=^LZrPBNl%HsN&qyy6q7rv=P2q!;m$vrMpdD+0A1<
zmcjI}@7dTH-P|OAT7g5<6s3*yl8FFqB8YZXNQ6J1iUhHAu2Y(TKOWU(O024uk|&pxq
zrx@$fG8R8~n`XKhk$s9>RH~6};RY{OZnJCzBKsj!2gb&UXLhQo=OGFRvjcr#%whK2
zPXxm+Xcx!q{shoh`7TIHv{i9J^q`iGx~A%uYI^w|%q%C_sE-y))+;wgbA6t5%?HP^
zc!E71{dJ_*F2d3iYkATL51v2tnRtJQ_&OC5Sey+3WQyLxLR3
z&gH>I%W^(2u}oir1zxIqIO}poy7W5u%hUq5#z)cQSi`!nq(v@K?l>LiF>n%_J81^RNHqhl|I%#(h?nT;{jLw^**ARWnXq4Y8AyK=-=bzw
z6Qag4RfcP?D&qQBSIuiAL=oGGX45KVjoY0`*uKD8!+dgUT{vB5IuFmvN*9R30Q@>7
z8qRHM{oGn#wV2Oi)u~J%IE1fpM%6JpKs^%pniJ%wP9av$xe#8+dc{Tg=5z~_f3xON
zLnyXnDEn7oy+!9O_5=rwmSleyLbWj@X=ZgJc)0TZXs88cG)2`-LpdLe
zm!XP9l{h;msqZ#gG%W;ho4GG5I++J2-k29f
znB}I}?im!J9Qq!Snne1oNKg@oz^S|+i9HS7^6E(Dgn+U9;pWUXb0
zNv~9En-~Xg1no*b1&uelJ(mslJd!Y22-nivJqsz)OuZ)`YK`td5jP!oT!tO^QnY68`VOT|+AA&V~CAj?uG`twDYH^5T*qr8D9@eAPX
zG_A@rDsVlmgaLydr?T
z?tP{WKkK3+#)4$SWRTk~j(tQGax2z4Be|`l)sm^Li%k;A%NwzjBSKKjQ(bFBA!SR4
zPh}1k#>&Rj^WXW1E|vEX@;0-c>ft1X^O|Pj!FRx4;JkO
zafkqydafOWR~O87>oce7)-NsTuoF6*DcdzV^Pf_JqqUY!RLID
zt^VFIREsp|F2@mJtFOe-jjv7*`I3UTOxVjpVQRg5>RR>Vs!DwpWepsANCskB
z9r|&wu(^LphOHaw;9(BrY|2SWe=TjSv17`d*6Z%i(HK2k6Lh*ZpS6apg)V{?2?Th=
z`j_6s_ZF-;GRl#7-NtqFl!p^e+JubpVQivQ^cY}qUSL(9BceNh$nakaO7ubHVHf=p
z%umzB%4S5{UUz~Gsxb^~AU~9_lUUQ_;rlG?Efs%OMNQQNOPHG-rONd)X75OFF=CwH
zP1I`PK-5;V8R3mwsL8X+;V2379r|eM)>E|{Z_`KSaJtiOUEll1l`KzbnUNeGc_edr
z`%-m}!|OJrn5>Cy#w3Fo`jA`#R$N$^lod*{x{;Lk?$pIS8umdJV#b^xt$C3UpkC@U
zxjco}
z7Zr=g&$Vg7URe4;e5K1PB@eU+ADxzV5HubVE(1eN4mNq_7}78yI+0pAIKbtpL@-;EPwOGaAY-8Gv5gG&37>-L7O^&YM|Qko0soq}
zUZCZFoe{2A%)L2oAAGr?psOgazxrJ{qkh`W7z3PAV4M3|BRd<+UZ0Pne9#6-%h{;;
zY?8O2Bq?y6{4P(U8kg88_zt~;ZZw1HNN|*0Gs)qS2b1?(9BXvK1pcXIZgiOjf7Xgb
zJn{odUYwdR+F4{vlUFC|;aFehUKjH&%l^}UnwGC^F~wKeLRCc75!cV8T4$D_v*LP*
zJe0R^$zc>X8Q-FlD@#r2)#ce}YnL6JYD<_B?*-%UiV!>XC?}eia*whdkFmbWB+=)E
zp&0dFqjP%Ke~>R#><%Q0VtU}qk}pN1Ss_t+b^UQ;a5fh*fT2Twp#3>MZ!u=>d6kTR
z-%~b^oPI6kXWw*%5`Ub!MxlT>jqEZY?m>8S3E$8`eCgG%eX=7vyx^qsiw+=`0_(8N
ztiUhxI2v6TJ{q}WVT<@IE;Wm_jPWnziFAVa7VDhnSlU{+!3W3Dsg|~g60F|aS;3%~
z;2$xZZn!HE3jpk8gxXVjp={IhemvPNQcr-qEQR6#Y81ecRw#(E36u_Vg_l!M9
zES_3;S0*5RbgW%=iNcXRx)p>S`RU>Z$~WqYnpeYHsOz7;E)XZ12sE?fXcrjzFw`zi
z=9*RB&~lwyH~g~eRHI1}JXtn>F;j20KyzxtqZ|>!AKzVCL&_$^$|ZVg3H)__w~EBl
zHA9KU@7#n}{6(|qb>u0x@lJ|(XK_H?piAj~9{>y5d{|@jS}_jtxy%nFaI@LGas%-zm?M2-@Qv
z69}AYgbxGetLx`!o$<7I%(7l=>Q8{uedD_;cDk;%R
z{2kZw{7tQc-ri%8e$)~Q*%&@d`_j2&Tb;9n<|&3L_SIIApK0TF7&L2x<70`;E900G
z?y2{z2uv_qjBTSzQHQTk8h)N&p`E;XSykuvAt}f>gZh#@sEz%9z~)zZe@RNI&QyxM
zS=@V|2`gjlXr~sB81Ll|8B@PS_?I){OUOIx0rAtO1m#>wzF<(-c!h@JgRP
z+l;%Ae843OoYGoPH9jL3#Gw2}7W`0Y5)q33GZTQBVmLYteg%{Ji*9C*t-1=t&p@r0
zM&?OqL;gPfaLCR$Y*{I;$)e-^=n|WW!IFiOqvEx_-Y}(q7p0eXLhh3=Ec1C_X6QR|
zUGwZ*{xjN5oG7{xl6T(zxf?~Q;Hq|!oH-`pp`&UWVBn<~=Cbk`sTotgLVKR5S$@81
z%KaSpExAr@ltvH3G%8L-mmC+kx42W|kwv!RR>9SzQJAQol1_`{EOi2YH)X*cZj7l^
zw8D@gDVSD&N6u_mx>OG1S%RD4^N5LJQ$A6{k@F4RN4+Z~_?KtL!u(2}!R-h64QUnl
zSPjv~VLOd!uN`^n3D>;t8fx~lSphXD!XuyepEf>HrF(u2^0{fAp+*m|&DtCJ8j(gI
zhGTM_9ZYCF_jwC|-q<8o8e6|m(Q7v2sv=mo8`%hd@ciO1a5C#@_gT)!UzUx%-iTOr
zTNc$03pUOg&UPw@?|U5SDtAv8!{a-C%F0zy$o_&~CJknndXi}>Z>B;vL?E`QNp4wT
zEc7k
zGL`#(w)OlicfhL{=>b1)T4&^lm5M2HsacV$*)+Ca-lf_##X=lRWa7$*9^}n3Kk@bDuZ}`UgT4Sl&~<`O0`q;Tc=VM3;Q8f2E41|
zx5L8Wxq4v~AY-t_h4HCX%;s0;9KPhi_8~Ju#;(gleQ{&-3|O24D}_K7lN~`dk=BV^
zSu(4h;bqy((GoaaHsvF?!dGz;%Js3Yw!y?`2_F<3Uw{@i12LC+qa~Oif*mb?
z&*xfo2^3ibD%7)pbr4(1^!hO#|YSrzc*R3y0HkID{6uXf=$GDn;lM-HB
zV;`_vkUXV!hIq0lLPk5v_OSnLZ5P;Ea~u^Z-2Q=v3orm;0*-!4DQrRK#6|16L7`jj
zd=r0{yY9rF=8`P>p(mOnn6Z4tt@uKJ$!0u{p@q0SVCXr>&>z8?nUbE%pT>8&CwLB@
zE%s7iC>MJT5%Bd_JK=bG@sHOZvss8*fx9nyW@8sTSn*y)+tEcjpdNz1|9TZ&#hOdl
zJYl&g^UI}ncPsBQ7~`n59O)pB%aza|=e0VD+eSZo;rvg`NUyu(Z5D?BLT_?^kI`M#
z(HK0=b9!nV`lZe&{GYhDBExW5{p}%$PPE?)b97)C2fTKq9F7&HjFGG83m#@Ecp3}=|m
z!FStE(!Q$a?@I@#0LdPNd4a-S$yJUa;T_tCmgCzpKtlV#N~POdiE9B9{namet6cP
zBd}C(_jL{G{A>;TF5^H5XNBO6)k%Bkz^7PF+_|KoI$2rcvxTaeVyZxW*nTb0G6)T&QwKS)zbkL}MR|((D&?VgjX#2jeClFK-(M5jGaE
z#(K+{>N1jI3BI%1EIUit>3n-8j#Amv!woM?OQs1r@#FsJK`&ukHVChT@&B=L4%?w9
z3KooQ+qP}nwr$&gwr$%yv2EM7o%EaBS${^=-m4garA4ErR`W;1zzr}O-~1Cp4mW_-
zJQt1~Re_ert-uODsnY*dq4^#VuStC5_*p7W|MopFD}7XMpQ0h_nSkyHN>w21#NqZA
z5D3%mdFWXYbRJJG<0SV9#S6V+2pJmdT55o=QU&&3w88R!Ui0M?KJbd<8wD$!Yc6o)
z4jU|iP7rp6zsCN|4+JPo|JK@O&}?L0m5HVRlcOL|CuV+b;CVh*McuBJ%p2g-B5mjJBp=u{PIp5n6lq@H(PaH
zu2uwNy4Mtc6_ol4g(z#=;5@unF|%_#Vf3=&d0`p`kTa+zc|SSybN8)V17}W5Jftzm
zU37wrZZ4#TiW+C{iIc>QcZAWGKXZc%A}RmzDJKcTmQlA<8uNl20SL;UScofr
zgjvPGzX3>+zEZ4>N!Pt4k%Hub+{q!7yWCzbECnQZ>V<7!PkZ~jG`0fx?$(i`9!H-}
zun4UNO^Jtr=LR!l`EC;M+S*$3Pvd6R%e3Nux2Jg|^}O)NzK2jJ)&yKi;1CjCJAQhZ
zK*-Pg1xAtS`Z0v>FZ~NzGBS`sa(i5fyQ;6q<3e(V0(C8UeJ8d!1wFeR$_OKOAOCNWC
zI`m%bq69v^k1`@JI5W?6lhehiQwhFzULqdC#Xfk%$Wqy5J#db@QaYG!Muj_*pT;UD
zThUBHf4@_C_IGF#h>R=_vF&tW^kP8b$;+0l`FZ)!c?v#yPRl;pOt&;nt%G53e$_%y
zyFuw6i4+*n3SH`w%o2-gECKYv`6qRMvH^8ft#sYQBZzl?mo%J;Siw~U**MYzuKA)A
zBnZD=;Prxr
zcRo0yW)|;iJJLCa6W*0?VRSw-<6yo2wDu|-Uv84}i-g9FxDzw_gP8KbrB-EsLeLnD
zV)*Vd1gv467OSA$zX9r+NS{+F$-3L6|6(lW-**TI*nxN-bVXjtd3)I9Mzxv0=`=m-
znXM$~rjN`?_e=Kf$F&&YiY@pg!R}>1MXPfwC6E`45sPOfiJLw&
znv}6c^Ia5rRB=8Pmw{R_;3PYvATDRo+|Ya+S$`cPjYG`3&md|Xs7)vS=FIj9d;E6g
zcr5_;8Bn(i;FI2)lXSK1FY`1<@I*uz7v6pwpBu?HdVUoIBnIk!pkaf5S-U>u7mt+I
zuaP&NgrFanAw(d;9imGZ4agW**RvFsw~qYVZGZT_2)Da#c(ht)s;Yc*F}>(}?t8&$
zv@{u4H$}nj(HBGh
z*dsU95(g`8z#_mdSIl?u~g{O=i63!>LhZkOaWiO)rvG|(3Eb*j>qDYVFc1|$vkYm@T|rY#h~
z%+i8c_po;ao=EE-Ob$U(k|Wc9>m!TD%N)@Sr2(SaD&S=^wFU<;Mj?TGu_qh_fXh4n^O;s&jTRC7
zS54cC0QrChrLltr4XrNPF)ei`M+nqH*491V-4P#KyvemJM~ROj@2O=
z;fJv?i;^!hNAr0W#z^tji)#<?z2)
ztj@`rUVNYE?^?ZZ0d0oO{tE)eba~jNSn(>Z_-3}q5OPz1-itX0d@GFE$>A#!BSm5v{u4RB=ESBd|6U|_%}b;qC-tk>pwFayd+FKc6ITi
z5}YBC-&zt$1e!I~BkD%F;cp+SstOy4P5RH{&Tvn?XPox2-{wpz;SBqqVKUyESUC5X
zA66Dr;50C{UyC&|b7^j1?0pDAsM-e*g?nQLyH2g}d{H1}Yy9hvf;r_rY~Cv9SL@r<
zXh`yZwp;@tJXZ)Xku?zLUfD#&^pKEKw?zC@#wtqe=oOyQ(
zo6Lxy!>T^C)9=xEofCK{s|&ZrYu>SY@Wh*xJTJdVMpTO?(eJ@5OJ^!GYr0!^22Owj
zLHU9OXkJBHP|HkXa%XVb*gzyVVnaHZ3HKiLmjlGMJx&!#88|&Ukq%#5h<+Y)k*76(
zHNl;jj=FU`_S!#(w81-p?9GTYEaGqe9V%ftIx`egq5bU3t6TPvJtoY9s|Q-vn+$<_
z{AU08wg^c2(VC%aYTo64^tFCLx|CVI!)#txV?4owq?S)kvh^&FNiVZ*_TJcG8=mAx
z`qwq3Z9J%9e}?yn|JgCP35!fWbRK_yr$WWu3mV0e*0t6BTE(0M#V|v8G++R|gBwR;R?HOn)!%2FP
zv9LqxWJO9@S=^S_#?tyN_wMnj$*?kF9yel(>XvN9Tdvt~cZn?{jxUR6P@SWHSmN5&rcYf)G1Y3~A=j!e=^1PA7nk
zSs^U7zblv2DboNmzu<04Z8CL@BztN4QhUH4hHC%C=$f1koLw<2ltQx4a@*bV#HrVh
zOE?|6TjDq{`ddtCJ`~cE20QtGJTt9jzxyusi07X)UlHEo*MHWH$h7ydeS4wB>SiFw
zSLp$4F5|u7kah51cB;cmUUU8zAz-Xx{D1<*VzgeIJmwyk)bJ)cYR#ZN^$J}iSw8=<
zk6t2G1YRfZuy&+bi5~$D?&R<9JN5|+hImTIHo>ad*-ur?g6n!OEzz7RJd$oK8zX_I>Xj%Ix
zyk7gp4r=AOyVcUcfbgf`rs%ELXuc^m!i~t~V2u73WD5~^`jv`HmzYv*(<50sP0I8+
zCc)fJ^mn3zg)+}WE^zaIAafcH4;JSx_`p4p52uq({_!O=p*@p?8ZO6$wPACU#URd>
zr_MctsmYU|ugBCrC)P%+E
zT&KN;2K14DPk8|tV#Lcqp*vV==fNY$lPmI1z&4*U-D@r@c{Uh2Sy9{WR8)gd)Clx_)hsVBB405-h$7O#+md52z3TV{Y)5o$
zDXl9x@5HwS1m?kifpsPK1efZQ4E4RXcUJ8*e))~!0;khMQ`e+(ClUIQ0QrjO9yb?Q
zj*NkE0h`#B*c{Db@9INx`pC2c@}-QtaMd(nVwrBp2|da3iwwdHp)`7QOZ|A98A#c4
z6y68g6*Eh|!uD+@ietaLBii7pm#i8K=r&@m@9!|4?vz}vxIc6z0)
zq>>E?-vsu5EmVum|7#oO(C?{wbHX4Jh-Y!2B_OP+`CbKmiz7`-EACLAuI84S
z^P!D8$jzORxQ>bmYC{HbJfyGndHK72JLN*Vmdyokem57%xo?kre=J#nADrmhr~5Tn
zkp%=b@S$9^`@xSM>qSyc;~rXy=q}!uGv)E|Ui`8wqS+Ui>7VCMvYfJ3ga%nd^_JD`
zUKVtJV9MFe7gZ7M`Uf=hS50*u5;91UaS+U-Hv|~%M0Ho$3yfa`H2Y`f?>1cdIb2*W
z#)A=_J|I@b1K76$;|FCff*HzalZ{*|CGoZf62Y`^(W9}0qmK0vkqqge`5S9pw_Cx&^BsN^BsB4#hh?rF1z6npBcY8kU!`aIX@!9lKC(GAH
zIOaeLWH;d56TVrrRJ8FdDkCyh7EUvO(xjYpH3O7BT(dsR|Ks{tD~4xeQn(L9(>b^P%f^yZxpp99-Y;Tr|%8$ih97{wLjgV^U4IoJK;)%Yf*C(le
zwGsG^Gck;Jg)Ig&7nPJMWBzn2i*5%c9WwFkrx%W6Zd0=r1!LwjD~{d{4xJ1E0&uA7
zP-r@&tb|&Lu$DUx>Z<1~xx}Qi{yTh^R7kE7jTgD~MePHAvo~$3J)%S5poJ(7gor|v
z)ow?i>R^duU>sb(5fC&GFB;EC0zlY*oFQ!_C;1otarj2csfn&SYq>7S;fq^vR|H35
z_AUm^j%UXzyf!ewx(2Ccx%*`E)~o2~DRZaMnZ=HX_AN0X=L5>iP7_*&I$b-5{^
z*y11ygHDi69#Ll$g0TZ+bs2>*Az!$(0Y^9&vm+zr+~clGuXmt~fG*2zY<=cfBVBiR
z|MBI=2e+fxHO@*l`A03tvRGP0a2n<-Qonu@9x;bjP)-7=Nv^{gW@NK}@G34bvoaX{
zA^K#UM-xN&8)V19(YNs6{bNl-`xJg7ZQj=
zChID>7dTpjl9_@f6OoPW#OgUaepCM?)!L1XhmxBFnZP;
zn=HK;Cd?;NE(5gZO=5U?8#j_zNxX2+a8XAVHc{`8~oEjI8CetXmNBn;ltPz8;{
zPQ7q|_=e(TeLxx}QbsY^8D+FbQ3@AB>e=Kg-4b+sa%=wEYE!{QU8~2?)~UfB#N74I
zrdSI_#^0C4xD2R&hXd~&>~8;!hR%_N4ITz9qIPp_KMKWLJ13>ja9c6xJ}b~dSM2dZ
z)2SeNr5z=2tsQ23AIF4^oVKoDP}eXimp^iz`MQZVbS|tx>faPna+P;}fg}~L%bv$K
zr)qc6=yJkYWgb(d^SOu1mBQc)_K@(d$qg
z5OI!E`CCsgzyyLF20@J#bwvEXB)NFf-5xo}!!XL1BA~u0c}S_xR{E2<#WVM1f~rDr
zp>hn2;c?u}!iKDBr)!Lzj|sVYPQNdg>)JiZw#B2%;IPUmb;!x!C)0Y(aQL*mwvOeUVMu(~<&hu(H&h3nmWHSc(sgdk8iAML@X=j{+1
zniBV)_FdmL|JOt3?E2d#6Q_q=`#9v3Q)cMuoHsm=*`jv*!*FG5Y}8v3hp-$=)5M1g
zDAXBUT=Pu&!wV~Gxjp?>f$vkzLLmO^Se`89mN~b79Kc1|hypn435W4wP`z^(P;ylA
zvaj+~KrIXOQb#d{`ct<AnUVp*a<8YCw?xAt(2@^9&YLF1WUCThpm)w4G(-zE
zJ@A`6ozv~a<@fdgDA3(02AH*qp;&im!?6~hvMcQZVVuOMaFi3Psc=XBjHg0rdj(%(
z?8a$-0xHK6%U(M4h?gkF9uf@BP-)#Sxt9}P^Bt?mn_%XgtQp9b0V?*#4{!*pBJLUc
zx`<~}e{_k*m|i8@1kP9v0StQ9>lQqH(Q+KcfySkmcq3~VdElV(;ukDaBm|vWjn`XN
zJ0dl7Zv@!ibfQn`+f*g5LbN%?PRGfYtyjJ2dHj3X
zcIYkH9X^$^Z-CbhUgph82zIYh|FTq`l(iL3dyUa+NTSECInJ3eIsk!{Oy+nLWe$;l
zowy?!JQY9&S@3L^u*>H{W~Xs!sVQYojfaB3j^fzFapaVkIf7ji-pW|Ni{o;qvg}i_
z7m@YKse3H91wo0uprKlECxVO@`R$a}aEN+5y?Jce#~4}`^&b$|7$fHfF7vYy_e8vr
zbEs;x6_mQMnS{Hx2^?;pfah_T2u
zScZHh80QwfjYAOaGxSt=n#72IDp%_83-^--u1o&5;qMmXN=sDWMr1PY75#j9ulTW=
z(HrydE?>EJc|>wE%$S*2Rp`$2w-24>I=Bw&4Mzm-i9PiF
zAh_3wfY6g4lZ>tTW4x(gS^3$aS;H6j7h;CI_g?$IHCe(*mJc%Ct|<{YD&uZ&V)l%_
zSBPrlsP4f?ba=;_fQGq$y*)EgHPM$&e#U^i9PNwlw-9~&cDv^dGrdlLy;v(6B1QNp_a&RvW|W=H
z?yK3{4+`aHwOZqU$a2eap-l*dzi5nG_da%_SDc?l@PHHxalG(!hYk7N=I;S0X93NB
z@da*jl29}}G`=(!MKI{fR|Fuj=7|dI*?9lb_uuJ4KeeVz|H2u*5CeLB!yrEg?IK}m
zz>X&gu^~Ier#Xh5R74@YnHX7LeX!AyE>h)K&0-8?#6&26;g*wz`NoJZCMGA4{)i)npk&od^f)xzIcW0?
zaNHz*gmA&e7;6BEW}z5Xg|(4{njHX6sdyL>as=gdGH4xvclBEC
z_mgUqW`lZPLSO9hdcyQ%QRiD;cH;qxv>Gc;Q=uV$h8bS^H$Yen>skAbwShPfAVaueR_a+QrT#&O2oWXabH#4jvh=cQe%*Nu0+un@ZEOZ4JS3Yzf
z_)?@%HjHhgu;bn$A(AHDfPOcD>KrNm;no*-(B(~pI)WYGMsGWPzZiNg{BvUQo&B=L
zqWvO&FC}p5BODNvjn@^dEtqvo*5|oFE3tJr(a!t>5Q
z6TJO&r|~O|OlfEWJqEwMZ;Cw1b*Z{6)w&xRfLr$AJzOje0D400ElCo3JA>~6ixx1o&
zsGvjJ)`S55+zy??wn|ZvHLiyS74r*|1S&}tRCNm;!zTnk4P0IANHC!~T=j!aymX>f
zG&}-MI(RP`(=uM!)6L4xuYwigs}nb@!-(<_X_q#hS3ijdON*iyOrU;W<5gpbdu74|
zPuFH8r}BMibxD8Tzjm4V7OV7;h{2414;ts<$*@&;d{A)204y{^htC=&auwAcZ&OH0
zPlS(6CD-nzZ3+(oV%z5$13^rne3eNhy+Xejn%xmo-W;-7B}-7p*d6fERSUm+}
zDOuI0f(2#_U6GF!+^MW$oqu$HMD97ro7Y=~R=3$Lh&Gdrn!6PXsD^m7e{UQmO-wdX
zvqf7Umx@Tz>Fy{W>!Df>*c0Zk9~Y*ebBwF1U;6S39pd*Fh~*kV*`0xeSp4uu{7)YV{t<8wTD8ksOe)Pi+UG}f+GB()5%s7)K;d0S*!#LqP?u+uEB|#
zQh$!h<*myiG3L&2*I@mB4BNs@B>K4|OIVEecHO2Q(?EjLo@1UX%b`rReH{4+)A=QGbAP`
zR11?&=SvlUm7}W37yhg6ynZXo$7#l;lg^VYe&pSfV_S={TRg>o5hB8`asuo%m@QTa
z3Zi(k4a(bbjoIzu_H>V#7#;`s9Di&J537DqsPc3RGxjDE^ww`BO4x*|4PyjdvJGrQ
zSI>~3){z6v&Pp5(kG>X$qiak5+E{@hP7{>|b;yUt99xw!NHC%1F0a*!wZp?PTW)pi
z#qN8>c}Cmn=VdB?JcFaMDZ5HD;<$D)jB`IQb((T;N#yQKRygM6xss2cqk_^a%|MaQoskzgLWIfC&X?W%rT7dxF>x0PLo8zE
zKc_u_ilVlEJi+(X7NM>w=_(93<*FN|IeOdR07NKV4lC82AAZ4kCT|xA+myX++--lB
zR~Jpq-TQ;K-X1flsD-Drj`0Nomhe8V(8z%nUIX;nMUE3{DH^A~1r_x#
z@KBgbt)y`ry*TD_BBmVAT0i^Us6-W9%wvCg*8*aY-_~WGV)mr}^4rO(T!6bm3RO1W
zq-$`${*vZ2(&}t{L^$|?Alt}jCs1Q%%;#R&SjhT$lnee)j8cV7`m+TbKSvlBrbeX~
zV=okceW0wQjv*T|7OFjHqKa`l54P}x6&WaP?#9i})}-W|Cz|O>l25_I>mlK$T(^&c
zt&0`ZXLn=62`XSae*;o8x$@J#QCzZG?7Q>M?)#sZCsw-%%!Ro4OhSdMbBepSx3bIW
z$R@(?v!41@7EjI{ms#8b?xJ>pwuF?F`jsqyqkQ8^Inj|~4+(V17jK*#A8d|;%rFe~
zRJ2rvIhuH{VFQ%RaNfh2kuyvwDwX)PC{_3jPc*#&^%u${)vW4y>sItlUEA*HlSJMf
zkyGiKk|?zxtG+52p2rk~z0wR~P=BrfjP27^UOl7BJBaEKo4>}n3O64{u+%kD^zXcX
zSwB_ZL-J;O4c
zjSwgJBqqgwh9A2XT5S~{uB2v(SL4bplg>Y}KC`|P^*zjA8
z*<)jL+TWdWg{@6HpgDFX1e11MV!f!(9z%s^*v#3XEMsqfOlMdD
zD1=GMO0hcKRFQkuo&ewV;ug76B%LPShPs{E{T#p?(K7~q1aO+D2_NW^Ci-*i{Y0-8
zMD3ixxAi#o4bu#GhmB-Xf0IF-Tz`u}G@gsYq)hNi@t6wNuTz^wPF!?{H42!(-HSOS
zG9j1Ya-UjIMR`S(5pWZQ{Z4L
z?v;H6UEnHHtS*8+WEnJtz{0tnCSC_>5gp%KjZivG)}Wl*`)7FZ103QwZQH0m**5E-
zu9O56PP9~!5dz%OAgg&RAI=CKr15Hsin}W2%4Xl!BulJ?goJr*2F1MVZIyI0<|3=<
zMQsGyu?B9BSUVW!DDgpmY2E)$=PW3y`ph4qnee>&O?)4D45t7CcP*>^FwRqSwHg{%
za4?Zjv9hAJHjBH~U*Lg2Sp_F3L?1C=_1MveK4*y6;T_ueVc4nkRv>>h-R>UVm7UIP1p3bWV1gC@l?3Wb9I;?RXI3+1bN9pPNk!{%qKlRy;fSyk$j
zj%t_*R}opUMG(^faSgJWc=hFZiqHGUg~$xxlt}Pp4m
z=UY`xsg@691wqX6k*P&P&UgqPXDGAKVTv4T{RJKmQbJ~b(Q%}N=&V|wP_K-r6LQph
zj|R!(QHlY}Nr3h`n|0#U7YnDA$h!+QQb81C4~Nntw+9fpa@=g
z&niAeuh(;iKL_Q&ul|{4Zdv8oqWbx_p?;cjJ$e(v4?2&+EbtQ0mZ5`oNS;lx;!WL`
z!haYO5MQ}}>z_3un!UBRH3m1WO#My;?eFHCz3qrQ4Gzm-{HFp}1wCKZZxkTE0y0Ct
zHi|EDaScg|TlV0>rFhrA1+1n&Ts=DjfSU|@B3uY>&^;=!=KdvqY8>aQv)kYr0$V*6
z5Pr9^J+8JOmctNb5p?!}AwFYaW1sP2^b~UEk*F7clUv3p8CZlf>$7$
zg?of4Pv+|_eO!8l&Olqc1FL2Iz*gxOr;tIhqzw=0>DS$y7Vi;q%Uf<1!e)(i+5~zn
zDYXuBdN|$SL090|0z~`CB#}U&4-=B#AIjEK19&y4m}z-J`zg!^zYZ7C3X
z_?qGxaYCdk=e$D{t!3OoI=u#_Ee0CRak)_mFoL7I=U!@%3moSIUDf|OAn1N-Flc{a
z2hV-Gk>-(xLGi^Oni0J==N>MW49CI3)c!hu(4JkR!jcr^5k&;@zoYV{e!*W^XPN$8
z?z820al#UDcYkkER&)R|VOq|-C~06It*xOgar+b8{=5B18cC=nTKnNhkf#Is!K9|c
zH9v*T^!oPV+1D6N^pzztyyGT+ewQLQU=>br8-38nIZ8~kPzQjz#!=p%kQ!EBB)T4d
zt!8?3k<0Fc9WAh{ycUBZV|-nz73OdvxVzr7#-Z>p2@RZ+Zznz^j%ro3wey?Yd`wlB
z>sk!d0*@*Mcw!DgtQxBRTqQ)fBKkzO%5Qpuy8&O3%gM-@{uPr{tw-Tgh#Tme_ug3
zH+L;5SW_ku17#4Oghe;~8}~iKj)ET}+c#-mVCK6nrtP(;P>l5T;nT48cJE*nL#3TPHLu2kl-3b7YXtic*k&69W^>
zdbt32`l8lG)6i0hH2Qy&XH>oIyrz74JPrsYMb1xAmn3i8vk|^DVl$$b&UXt|o|l(}
zKn|6p(Xrc2J`@N%K{gQ)sF7#WZgH^oVv~4#r
z73M4j3CgP=y*Zq*EsCkS_4s>#Mi+rss=B}0u=|s^txegu)Xz=6OWCJ!*P98M^q7k7
z&0}M}Gts4?OKkQ)FGjb45#KBmVwwgbIC+3->#V-kLP$NPeRW;*VNYp*!&OYB%lMM#
zlR#wi?N)>RJ|DHBdtd<_4&UphYr+pOI@K#rb6?KMWlPf)3MF1v(H5P5#AlV*u8`Xw
zFxbNv7&r|=I*lRd@dPkJG!F=}4j8YySJn6$YoQ=8aA36Nb;#LHq;c5OsI9(-8i*=i*c99Th
zQK_I$+}4e`7v5I$_=VF0SY9Z!uxnT8YnxL+@L5Y)pwAqRg~@95GA45A
zbikt(Gb^gJMm->LWTJ_B^HYo2!opIL|6sO|eOfm)0!Ac3i@dM3rCa6m8xs{`T>Gn|
z@u$heg-Jnjz9OlA2!PX;J`&4C?bWP;HJv^-^RqsOmG4mrE*F@}@>tJ1
zB%X0X`4EUbu1^Y&1e8!;OExv467(z$SGmD=rl@gX4wSBuzl>Tz6=Nh`_{s
zW+@S&Y6fgp!{cChsq74&u}v3IL5#4l)juC*qQ|qdc22r~`g$u?ZaE<)PsT>4r~Vsb
z`4h<3%paAIS<1zzz;}*J&wGYg^Eh$lI2g&YhwL`mZ-7eves?s+%;Cd4N1+}}@D3jT
zTR5iu+5R+)?sm{ilhCteZPj0k>NO!EUZ*m35=Br~?+t0q9OWmq*ioA`pp<5KP)l93
zJuABvMYz9z`aq(y`9-YoSP_-er3P(JIcCpR7d!;56iryoM$o!PCSNbAB>%Pp_Z`;(
zEWMMf{2`cyFZ%u`F{;t$aly5g!Ak%~g0>)0V4gWYG@S0BDSDpGcVKxU4YV%_q*vl+
zE>X19Ys>yDMf{*MqNu}cForC(GB4sFTA$1sHK)6O?-M>YAyRB9$jD8_f7}YsO>0OD
zOV&>AHgQfOIh%GLwI+_wWpxtpHmow^t}C1y1()6Bq7oVbXLVlkJetyFU6rbv_t$km
z7-+irz)LnLJ?Pf0?^J^I2H&4;L5*M<^iLI
zmmeqElw+_xDpQGAO3cG*a2??A^|3a_)BEOsSehd2XNO}AkNU+3wB^MJ#hj7Bi?m{v
zTM7b_CWiD-uW$A%n{{26-U_MV+_kqY*V_b0ssngA*X`dfbwp
z&a-UVUvY{S;|t2zQcXJ;8$+H3@{r)>cd!Ml0e>@e=#Oj9O`!D_vI)sPCJ^jmQ(2*Z
z>iezEMX3g9l7#X3WPR@-x=Ot{R2{G-A@VpgRI{O=)nL^DacL$~`H;x3CRxw&XyIEL
zCj&&5=$X3CrH^5F$sUCDa4Y)Zc3GPdi)5~d>zWRxAXa;(m}>RPk-lQVEcmtzh*d{R
z-JZTRiisZ(IEih8^xC_40~q#&9l&RQCEBK8cSMstl7a_+=u@+TX_WGd&R6RBwOruO
z*I11)YsSW*pEJ;#u?El?tPP9T()P`WjKw4Rt{BoDM}`4iywxCPqu^}$QEfOH-|6Go
zkTv{h>#P$Ww9S;kBK&Ho-h66z&B&iLiFOmaDuTcMC_Gj!i3Wfj4Bsz|+vk3NV2ceL
z93m#l
z&~V*PRbfuvwzB=U2a_IJ53;>46or26N~%~7+*#{j+ge}51F-6r@F|2d?pWIipKDkp
zxryjGj?&KzIsBD*U!hrr9>I=($6__eSmv3T08^~Jg9ZQ^)ciotlBs4mw^fN>e^w;^
zRZ+$|?@mSo#N2SGl+_-_Kq00YUr*)?v27396KOz?)!SJ7dEy`c2!YCx5DWT(5`tQ|
z=6XA-h+Z!hhr>JWs-8eP5^
z#cvm;PfUH^4GPQQu8e?%j_oyfp-2|DUpVZ7+y{Bl{ZIv#6ENDiEIj}N$QK=y$<0aR
zX4KkvWgvqtp70cz<}Yg8YdLwbS)8wBulb2rya7xF!Ctg9@SrB-^GEQ(OM{Rm?K&gx
zWGZa(5Y5DD-q4YY^k~L^+m$l%Nik)JYK#g1Qr|3Sn^lGF
zf`c4$gId%MLmz~H08u^c!GulyoD|s*l?t3XQH@@ToUCL*#}(k$k?6MIOU8_K%cfFQOM29
zy11PArnA7+J2~Stx8(VXzIhsHgIa#e@^1yjA!C_GKLNWmxMWvm9%Np8`o@v2f+xkp
zf$J8l)f8x2Z_2VMcDP`hv1v8(BEw~O%4EEuaiwO$y_3v3PS@33ck{nd`U%_<6z8JT
zE3-K>3w=^%BxfM1TtiV!p@lR(M_n&c+r!uLJE+vI@-#CDU*r06-hsvCk+8aKo?Xf6
zC@3l!AP@@9yYeGUNHbLPj*NtxCuY*21EP75D|VbPRcn44j>T
ziZQMvLtLELS&~~?BnmPRMdv5vxLxpE9s_+uOe9$0&i%~s1+(tG$!5fkMp^J$+vc9K
z6_}tsNsN!U_&)P+JZ1?KF~Du6db)r?!Bn~aK6K|T2w*tGtIJ@GgAI*XqabOwhSypM
zt!bnLgRluA_^ni03#pRU-c%&}-XXgZpCTaLjas+aYi53&nw8MWR|h62eH@5g#p;}}
zMenA#2+@1$NBQ_F(T+>cmM>u5@x9wc0mb4&sF|R6<5xUY4hz?)J8pMX34Tf*RQ=M&
zU0Worh1t7`4ruRfebDMY%bS77rR1=AN{CIYu6A}y_OjfjqMdUDJ
zJsCiR-d-e`ARW6z!Tdi`V{$!$Yd=xz$23F%;sZcITKs2c~~t
zeAO&1k;!~Mb|l(se1NUPd#2Kyi3d27B`cP0UAkEDX+=s@F|Jt~kBZnU0d_Qs2O
zL#q3A4c0-nh*fIu*2M1^if#K`AhHtK%NkY&Yl3Ju!-k~>ID}z>I{F4h{?kogc0_to%shC*a!F
zV^f>94$Cfnc{O&{`C{)6ta5Tyty5Db<*4cM&NOCe3_gfN{&7+GokLHBGAsXDj6ue|jz~eC<|#fLAR)l!N-Vf-X}z5?v@5p~Nrmz)?IjpPcY-I+(HU+fn1XZ5
zTQoFY;Jn~5{%=8L>mj^)mHl|yE{0}Bj0lS;9aWnrV)UgBY|>x`kV0|R>jB;)1EZWv
zs(;;5D$qi6qu(&ZL0V|RxTzz(%}v{2*Ps6p
zl3?IWa<6ck(FEU2&Gix?DE@?-e$6ki)<_HMG(7SA8{FOH&3eu^WYgu+2}I7;derV^
zK|zDl<+$pGk&ienAjh3Zxt?zMWjfw2ZhI;Haj9^vxpq*q#^n
z-=YV{d2BTOb&?v+Ds;z0YRcS^Yt5EIFA9UTmmnlH>p3}a4Vaj%)jI!+eD%lXw4Zf=
z#S9*6JTSw4S1n2HpGYohm5_JSY|CW|dfMtTyU;gJf>87YVDEhpouwj|GgH9^`30m(
zG4SpECwl(fq^F~B7tH*0)E2^e+liRiRI^Jj#+(|5FA8rl5egd0Zwi<$vfu)LpQlJ1
zy3JGS14!dVs-7LW0Mhz+E7nf>vPG-{pA-?cHRuf<4UE8jiYz$Jgd$EpF-eJ(+9@z8F@G
zSj6WN90kd-6mt^Bwwr&~p}$L!qRk)=>FIH9M{uis(r(FSt_dHRW&Q=g`$v7i6vxVa
z<$}A;N$jN8!}2UHDkO|Iir9b5{rs&4j(@&vHRG
zTllGD3%n~+lUr3fy-e*woFDAE_bR>-H2SEZSOq#GOZlDJ({cxwu45MtL(F*4ZX6F2
z*>+9{h6@aFFQ(l?YuS~N7#P(rksmT
z?3V5`;nywepKa@sK$^y^YBCZv0OPFHyl9na_QTYRH~eW*uC&0BvV;j$0~c~!!2R4Z
zf@d
z1ZWPZz+Ji5X-metY&M&PwS}{+PaMe{TAymh8)4K^Cv-Wj3Ld)pQ2?`81;v@7L@{n%
z!&LG7Ez4D0&lD|MCv7MUW0&aij3O@D69bacxN!#o+T)D;bOt`zxu(%{RFBvTNi1!d
zXokG+RF^k3yS+Z}n@8Jyyz#j`3A-Ps2~hS+z~2Xi0PvRI_wow7`s?QP9~0THY0_+d
z2$v5R|6-~01iD@NnQD(+AfKmvB;8Y-*GSTrr`3iN^xBCI`E(RaIDgHIpvLF;?edH5
zvA=_{F`^GiI^s2)xMqk12UX!Rzj(1ic*ZgQg7nY+8o^O1q2=$2?FyY8YWO`lc?Y;R
zzKz^QBLLwgi;>a#brw-`k!ahdPZ>(Vm4>#f5?Ac{P4v#iqLumLwW4K7qyQKzq=H2-ZM*wb^Na
zuNx-c=WpUK%`HgI-w7h6W{>MNz>>iFUk@|P?vmz2G6MwL<(VzygFhdR&z&vYZ7CK6
zm&8?j#z_N}0sZqVow8Vf**<%nt#cf>cZTbV+aYPDH6{}Y)ktQseg3BrL!RkOFQo7w
zQU+=L>{*wTtEq=3*S~aZG9IXkX@by*TJh+ulQ`R|uEuxEX#=UbwKl)37!x)rj!itG
z{t_(k_@izF^gzY4RgwvPPMc!GPmKeZjHGjv>bMd7vCLr6G5jWL#cCDb_{^mF3OAN9
zQXK6RbhoO&*tX;My&?pwM=21PJsAi~jz3Y(KnM?49cg{aL{Q~11@kERnhf?xy70qh#NIE~6=k84#jU11NcLtiK=tJWu)z=Qa*W;>)U^Povv@K}qQiv;
zRE*jAU$PE=_oN0RaQK&R4hkZ^BpMw#Uj$CXQWLX%;Eloq9`k4uL(9DPBzj})h@VGo
zY>b!5GHw4>a36@FEZHBq64?y}lPcUocPbw&xgI>1fK`u*8V!%5@;jBq)IqiafKvT`MgVDzPSG=GpFr_sZs2_JLGY7+OD1O$#p%UVhMjvokl(l3{!krvzqH|L
zDZklq*Bw;jMb7&;w^ZWXab4-4a2u`?Su+)MZD9_vB;a6;4h;;2@4sHS-*CJIM*GGP
z1P4o_9+cCpAf52?n{47YZZF@ia_~%>WBW4s1
z@#VU=cRO275Ix&?a3b;3Z}
zmb=_p&^d9zx3UF+xlkQ)%b1~QRX;joj(7RhvlQ={ecr%3@Te6`Y?lkyw5;|;(nH*(
zHGM@XrEn7%dV+_t6lsc5GI*@s`cX3>i8^2CRiiMw8xDBkJs-<{a$z2SWhshACOp|1
zvZhSUk0&t2EW_wjP6tO{-l+v4o~b_)dQ|ZJZ_yAmxI}2%zU9S)AKm!rt4ws0;Ne;{
zMvjT#aHU>u@-HKZux*hT&WH8=_FP0
zzo4moFfmZ(5f4sx@xIphb7SPk`x3Fb9quGHgwx`Z=*-T;dR+7sP#MuGo
z^nuH2c5?&9A3EyW?$$lU*3hb`$!d11^1}B-n{_~+ClDx-kz$;no%aoogmG1u8|cD7HD75_vYQCyp~6v9fTa5n}kooy5Us1JrZpK-GvsxjoyO%Ag<*c=}HgN>gkcx
zia%~-cV5`orGRNMEy*16oy-p1g&<4(s%{Ogl$1I2#;TiH^MCtKT*xEidL6(WasGoJ
zt*!06*dV#4o9!jHkn#ng-T`*(jk8EuR!sx*nmIeYb}dy9=sDO9f@wm{3MQxgpb4l
zI(j|kPw7N*`Vm%r#(ns5;d8AQ{~yc14p<|`#T{ELX5U7KD>pL_kKR(XZFrdh!<->M
z%RqWBCf61}J3+}!^r1+74}5(|lq6Wqpa-3wBpf>9Xc6wzBd7kl{gA+&KLZbLMGEDc
z3BEk$ov|Rxz&t9F-+co-O>l+>4eGF<_Yk8vl^^=hum?IQKbK3plkR}_#aoGfLUG=x
zGv#C$w(4%%5=*qWwNoW-Gti0FmGD%^D!HqLIomYLoU`eR@e5ETj?qYl*JDf@%r$fA
z;{JNQ4ahsnV21#gr!nBDgyp{qKQ$UfMh7V_oQkd8mQx55M#a%9hdr*n==q5fj
zv*3o-9Cbv;b-o_mtk?ICOK%_
zbXiO4F)ev4-LME1%x4Jx?VrA=JRTGs-SrxmLg+7ts4@ZXcIpOb`0kZi4`YtViOpTksp&`O9<0KRmW
z3+Qlo00##rCpUN6B{cE>*6CE2(3J$dh^*4qF0O9mTzq_ptcpJzEZwZgIoa9SlP}Ha
zf!%swm$f!b|7Gnh_8xVf2$J%wtgma4qt2#uyUfbpyz^;R_epFb(_(~-fSr)?P;zPw
zn9J1w?0R&7zmFI%4J}p@IxSy4rZ9}!z(5%#6Fad3vI(V$)9#mfVjpa-oG=HPFL;Ec
z-=duC>qgPfb>>VaFeprNtffPE*kq6>fyBY1&6J-(2JDjZ;jYAqT7A1=7^!`s1d@q;
z2?>%FFm+B+DXATUZ4~_d-EHJqgTjhzg(XRKUuOv^1_PBusjUQtj3Z4V;i`t5P
zpe9g7ogRsU+3E13sJ2HSG<1gLikJHmTmmtuou$d?6>{!`*jVKPu`CKwoiqYNr7*{p
z!gVvUtl+4$^23Om)Ep72kUYs@6+^(dPVAC>Oq5rQIx87hA)Y;7&+Y@kc^!x>HrlpR
zt<%~Je@N-Je~YS~P){2>pwTy)0##BT0Qc)3d-LyvOVPFRQ9?WyS1U}7D^y8Bwb?@O
zgj!CH5Oi$0vVHq~T*V+ZmHy=XgpmH^)q0^|t{r&gdcCb(T)HsdNeyAx74W_7yWGbrmwp(~e1ApPrY-+rVCV2e8L3Ui+R
z)UdO)s)Lhcm(L%jQ78qk@Q=66d?_aeH^!#r?k7#}{?LDIQN8gmgWY$SruxqeH4J`W(2{4
z*i{M7g@CL);}XeE5N`T4@Iebeoo4p9W%xEXGZ@`;O8v2N;spN{C6_>Rz?vo{JQvE9=}JKj2d(d&7*
zH&0J>YX!*YnWo)v7S2K)CHX6?@B(mo+W+@F=@?#C=7_chlP@C$^bw)pP~-#P3n$ff
zqMLZO4~!uPjnnI7o~v|#jSfF)~-bjv}3Fy(Q_+&pSkU6Q{c+dzQ^R>=l5VS3_lFXV|_J
zxOlx9Z47z!dfNU6ygh7vC_5Mo;5QyW@2@+oPSX^J+1x`_gxEkbj?&T7PQ=YVo1<0G
z-6PaBn&-X6rxq@_IARbgB0J(63y@RyQRr-v;q;NYID+=~A`~)*&YJ(2>$MhpaT5)9
zlax6E5MhHixYlExvG^xOp6X;qMK@EX592jqj_tp639kMHigfCrk1by+uPp$d=y>0-1;FA3!aD8$&W}@7>vdm=d`=q%`feE4ngX!3l
z)E0oq$3;2d+(2#I5e`%7gGY|X$8@Reuzi4$t=8InTdL4V^^w))wW2iRpS9*V>ymbZ
zDPEIi=hewh>mVd0i~Ko;gr!NkB1-??^b#~{^t+F>MJFP)@;_iod90zn^`R4DfnJ_L
z^tl9t?sLCmvDnUt*bW`c#~yZ>ep#5YZT-Zx{%A=D&QWG7_NBZumT;@wG#&1FdOBNW
z=wJW7gpj-88S)2o2=>;XTQ1#<$gkFUcoD4TQ(6_BAybNxGr>zsYl^k<%f5F#-=6MBxN<-22&Cm{(I=_qv^g@yK0P&PK?(w
zKxSqFl6hLbbwcITk_x=tbDcbZ0vawFoASYCcrj>2U*Go@KHdV^^g*55c3safCuJM#
zOvoSCVsL5vA+|4(VGC1#JT<0h(c^n(8W1{!*OB|-Ulkp}V6+vhBoP)nzh|~-4*oyU
zqy`@Z;~EBJ=7K5M?aOB9v~yLXj|QjGjoV-pAB^B!u|X8g
z-GJjkAOyCbrn0YrY
zLNu~q-;&*{%RG>O_9vA)kr_3P<8c!>9Yq;jR5138U;Xj~vEv5RQoK>S(pV+g`7@$h
zJ8`1re1Eo>{uVs=wKf|Q^-Bw?QILzQef|~@L2nKEo95;3Vtv?+x0&=&qfKp^4fGhc
zBPYbaCOWWi{|v?F?u|pN#S`7bi^{H=D9gZK*EbD_b8@<$&Te+AL=m77mcTb9e5~=_*myD2Vzx1
zUGg9|_9P=XWGDxw8VM*3a@wop{qonDcQpt=lPe_GgfXi($=!m_evc0o=_)`!O6+TH
zAjG3aoK?N-uLv#Y{V&#TPdt=Bd9uj8A3|rYGNvSNiaAl==yw!%L#5u7h9Bt5)1H=E
z4-!|ug-+Yb3?#PKtOACkD{wiz;5=)~){#_|3dD)q=82%@$5ec>8ATnRV
zTC&euLJF$CV3;7ZRxqN;z9u}1(LW}%DFuMLhUQ$TWzBqCRnl;|My@e~xOb>JtAagG_w5kqHW~#Vg(wMyrFGyH~29UEPh>TcEI5A?^NThY|8DlB3L4
zjnLjI7~9sy+JKs@4%J6M8Ik!F>48gz5eo$scOXD?er9^$eT+7++x<^Nm<8kMurgSc
zNqbFV{T8YzuC~C?yqd;l1y1w^3u3v~s%qEvP^X;-D%2Ojj|i!!1Bb`274rg<1X7U@
zV)!0{S5w*D&w->zr~8@)WdGa4=1_4^aTF7QyZNJ9^Oe2_IZzH*Ylzm_iKkxf;9(q<_hbWe
za>i~&=PUpYvf(-pBv2kDAttM+qHvm}K2`X8@aCtt=E7bZ#E3jbQk7IlV}eFn6i8w&
zh>BL;R-I20ydXebEYF@lYH>H{60E=cF$I-iyyn9B^8q++s=%u7;C?d|9K1%KQ_6G11_yKyRjbKP{#VoY{%gI
zI>Lx?NupX!rO8vlwrm0h@(VnH2i_!+vQ8G?1%caxV&8?gg?efPx4Uzj2R#H9!Dn@_
z&mxsCDXJ@L=+b76&~5#ja8Bke6#RB0YhY=nt96eB2$h9pIHy7ucu_2O28fLENeI74
z4y>XMDr&Tc;RkyfW{9oQPmVm3xCO##psRBA$HL9dswp~@fpJ`X-sxvfa
zpoFlPA)a8T2FdPt8iEJ0f2VKz)8!;j7I7cOf;oZ*XXi~=i?owZ2krF!K?`b##5DBx
z{EcHmoc(~5HT}~2A3>AubODW${_+Nm2W0<_$f|Db^Mjn7j~$U!n_QoroRj=NkKY^}
z-N@Mi0CG0+Z-}gN4i=7|9pC?SlH~e=JREGie7xdf(p>CZ?CervY?7Q3d~6Z`Np3L!
zCm#o|F!}#K@*U;>)D0Jahy1_Q4awXe#~3y2(W@KZ^Yxbb%p;A|!W18(qG78F9|Fyb
zU$~lj*u&@oNJRB5y;_*{pJ}r@Xy~?cl6Gg3cJ(Q4`fs^^=z3preDep=7%vlMgG0hXwb-{VCW7P>ugZwK9*(UIIAtQ7X93A#n*;${%d36Dq%VMoQiDA^=;u@Y!z4p=U;qUE3TlK5KE-C#3Z
z*@VtV^rm@>d&F@+8TRpns}~Wl1WcdabtoS@M)R9wDZfThc`+7U@;X=)TY>i7CSNdb
zj5gdrlVC%J-}(|WmUYtu+(VBUFaf}sIceM6MM*3!PfU`$?kQ;_S6PvhFZm>(1RRo(YWyyNHY*KqPg_`9=
z_67G+g&i0T5|u
Jq*SC4{|g(XdvX8(

diff --git a/doc/src/DimRed/DimRed.do.txt b/doc/src/DimRed/DimRed.do.txt
index 077e1be6c..621e66422 100644
--- a/doc/src/DimRed/DimRed.do.txt
+++ b/doc/src/DimRed/DimRed.do.txt
@@ -337,6 +337,14 @@ applications.
 !split
 ===== Basic ideas of the Principal Component Analysis (PCA) =====
 
+The principal component analysis deals with the problem of fitting a
+low-dimensional affine subspace $S$ of dimension $d$ much smaller than
+the totaldimension $D$ of the problem at hand (our data
+set). Mathematically it can be formulated as a statistical problem or
+a geometric problem.  In our discussion of the theorem for the
+classical PCA, we will stay with a statistical approach. This is also
+what set the scene historically which for the PCA.
+
 We have a data set defined by a design/feature matrix $\bm{X}$ (see below for its definition) 
 * Each data point is determined by $p$ extrinsic (measurement) variables
 * We may want to ask the following question: Are there fewer intrinsic variables (say $d << p$) that still approximately describe the data?
@@ -347,7 +355,7 @@ We have a data set defined by a design/feature matrix $\bm{X}$ (see below for it
 ===== Introducing the Covariance and Correlation functions  =====
 
 Before we discuss the PCA theorem, we need to remind ourselves about
-the definition of the covariance and the correlation function.
+the definition of the covariance and the correlation function. These are quantities 
 
 Suppose we have defined two vectors
 $\hat{x}$ and $\hat{y}$ with $n$ elements each. The covariance matrix $\bm{C}$ is defined as 
@@ -409,7 +417,9 @@ In the above example this is the function we constructed using _pandas_.
 !split
 ===== Correlation Function and Design/Feature Matrix =====
 
-In our derivation of the various regression algorithms like Ordinary Least Squares or Ridge regression we defined the design/feature matrix $\bm{X}$ as
+In our derivation of the various regression algorithms like _Ordinary Least Squares_ or _Ridge regression_
+we defined the design/feature matrix $\bm{X}$ as
+
 !bt
 \[
 \bm{X}=\begin{bmatrix}
@@ -437,7 +447,11 @@ with a given vector
 \]
 !et
 
-With these definitions, we can now rewrite our $2\times 2$ correaltion/covariance matrix in terms of a moe general design/feature matrix $\bm{X}\in {\mathbb{R}}^{n\times p}$. This leads to a $p\times p$ covariance matrix for the vectors $\bm{x}_i$ with $i =0,1,\dots,p-1$
+With these definitions, we can now rewrite our $2\times 2$
+correaltion/covariance matrix in terms of a moe general design/feature
+matrix $\bm{X}\in {\mathbb{R}}^{n\times p}$. This leads to a $p\times p$
+covariance matrix for the vectors $\bm{x}_i$ with $i=0,1,\dots,p-1$
+
 !bt
 \[
 \bm{C}[\bm{x}] = \begin{bmatrix}
@@ -624,8 +638,8 @@ columns since all matrix elements in the design matrix were set to one
 
 This means that the variance for these elements will be zero and will
 cause problems when we set up the correlation matrix.  We can simply
-drop these elements as follows and then construct the correlation
-matrix. 
+drop these elements and construct a correlation
+matrix without these elements. 
 
 
 !split
@@ -773,7 +787,7 @@ X = np.random.multivariate_normal(mean, cov, n)
 
 Make thereafter a small Python code which plots the data. Note that the function _multivariate_ returns also the covariance discussed above and that it is defined by dividing by $n-1$ instead of $n$.
 
-Now we are going to implement the PCA algorithm. We will break it down into sub-steps and across multiple cells.
+Now we are going to implement the PCA algorithm. We will break it down into various substeps.
 
 === Compute the sample mean and center the data ===
 
@@ -835,8 +849,11 @@ Finally, try out your own PCA function with other data sets.
 !split
 ===== Classical PCA Theorem   =====
 
-We assume now that we have a design matrix $\bm{X}$ which has been centered as discussed above. For the sake of simplicity we skip the overline symbol. The matrix is defined in terms of the various column vectors $[\bm{x}_0,\bm{x}_1,\dots, \bm{x}_{p-1}]$
-each with dimension $\bm{x}\in {\mathbb{R}}^{n}$.
+We assume now that we have a design matrix $\bm{X}$ which has been
+centered as discussed above. For the sake of simplicity we skip the
+overline symbol. The matrix is defined in terms of the various column
+vectors $[\bm{x}_0,\bm{x}_1,\dots, \bm{x}_{p-1}]$ each with dimension
+$\bm{x}\in {\mathbb{R}}^{n}$.
 
 We assume also that we have an orthogonal transformation $\bm{W}\in {\mathbb{R}}^{p\times p}$.  We define the reconstruction error (which is similar to the mean squared error we have seen before) as
 !bt
@@ -847,7 +864,13 @@ J(\bm{W},\bm{Z}) = \frac{1}{n}\sum_i (\bm{x}_i - \overline{\bm{x}}_i)^2,
 with $\overline{\bm{x}}_i = \bm{W}\bm{z}_i$, where $\bm{z}_i$ is a row vector with dimension ${\mathbb{R}}^{n}$ of the matrix
 $\bm{Z}\in{\mathbb{R}}^{p\times n}$.  When doing PCA we want to reduce this dimensionality. 
 
-The PCA theorem states that minimizing the above reconstruction error corresponds to setting $\bm{W}=\bm{S}$, the orthogonal matrix which diagonalizes the empirical covariance(correlation) matrix. The optimal low-dimensional encoding of the data is then given by a set of vectors $\bm{z}_i$ with at most $l$ vectors, with $l << p$, defined by the orthogonal projection of the data onto the columns spanned by the eigenvectors of the covariance(correlations matrix).
+The PCA theorem states that minimizing the above reconstruction error
+corresponds to setting $\bm{W}=\bm{S}$, the orthogonal matrix which
+diagonalizes the empirical covariance(correlation) matrix. The optimal
+low-dimensional encoding of the data is then given by a set of vectors
+$\bm{z}_i$ with at most $l$ vectors, with $l << p$, defined by the
+orthogonal projection of the data onto the columns spanned by the
+eigenvectors of the covariance(correlations matrix).
 
 
 
@@ -912,7 +935,10 @@ we have thus that
 \]
 !et
 
-We are almost there, we have obtained a relation between minimizing the reconstruction error and the variance and the covariance matrix. Minimizing the error is equivalent to maximizing the variance of the projected data. 
+We are almost there, we have obtained a relation between minimizing
+the reconstruction error and the variance and the covariance
+matrix. Minimizing the error is equivalent to maximizing the variance
+of the projected data.
 
 !split
 =====  The final step =====
@@ -965,9 +991,12 @@ discussion in chapter 12.2 of Murphy's text has also a nice link with
 the Singular Value Decomposition theorem. For categorical data, see
 chapter 12.4 and discussion therein.
 
+Additional part of the proof for the other eigenvectors will be added by mid January 2020.
 
+!split
+===== Geometric Interpretation and link with Singular Value Decomposition =====
 
-
+This material will be added by mid January 2020.
 
 
 !split