{ "cells": [ { "cell_type": "markdown", "id": "835e5a45-a9f2-49c1-a942-ab1378c7f055", "metadata": {}, "source": [ "# Exemplar Text Selection\n", "\n", "Toponymy provides several methods for selecting exemplar texts that best represent the content of clusters. These exemplars serve as representative samples that will inform the LLM-prompt for a particular cluster or topic. This tutorial will explore the different exemplar selection methods available in Toponymy, their relative strengths and computational costs, and how to use them effectively.\n", "\n", "Before we start, let's import the necessary libraries and load some data to work with." ] }, { "cell_type": "code", "execution_count": 1, "id": "3f213524-81b5-407f-99f5-c7dcc92b2e26", "metadata": {}, "outputs": [], "source": [ "import numpy as np\n", "import pandas as pd\n", "import matplotlib.pyplot as plt" ] }, { "cell_type": "markdown", "id": "1c9bb8b2-a0a9-4e7a-b19e-f09f7eaa1036", "metadata": {}, "source": [ "For this demonstration we'll use the ArXiv machine learning dataset, which contains titles and abstracts of machine learning papers from ArXiv. We'll just grab a single parquet file, comprising eighth of the full dataset, to get enough data to work with." ] }, { "cell_type": "code", "execution_count": 2, "id": "83d6a3dd-4927-43c7-b62a-410b635f5201", "metadata": {}, "outputs": [], "source": [ "arxiv_ml_df = pd.read_parquet(\"hf://datasets/lmcinnes/arxiv_ml/data/train-00000-of-00008-f3c9b137f969d545.parquet\")" ] }, { "cell_type": "markdown", "id": "876f991f-3f60-43c1-a3fb-03188ffa2205", "metadata": {}, "source": [ "As with other Toponymy workflows, we need both the high-dimensional embedding vectors and the lower-dimensional clusterable representation of our documents." ] }, { "cell_type": "code", "execution_count": 3, "id": "4526cdd8-ee26-4e2f-95e6-b3acc70f04ad", "metadata": {}, "outputs": [], "source": [ "document_vectors = np.stack(arxiv_ml_df[\"embedding\"].values)\n", "clusterable_vectors = np.stack(arxiv_ml_df[\"data_map\"].values)" ] }, { "cell_type": "code", "execution_count": 4, "id": "8dec6d67-8eef-4597-b3de-c5c0812bc50d", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(35227, 768)" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "document_vectors.shape" ] }, { "cell_type": "markdown", "id": "78f654b2-3f48-4de9-aaad-4462ae8dd1c1", "metadata": {}, "source": [ "We'll need some topical clusters, ideally at a few resolutions, to select exemplars for. For that we'll just use the ToponymyClusterer so let's import it." ] }, { "cell_type": "code", "execution_count": 5, "id": "a2f3fd51-adce-4293-92fe-fe9af2455241", "metadata": {}, "outputs": [], "source": [ "from toponymy.toponymy import ToponymyClusterer" ] }, { "cell_type": "markdown", "id": "e5255886-ad55-44de-9a53-cd314eaf6853", "metadata": {}, "source": [ "First, let's create our cluster layers using the ToponymyClusterer. This will give us the cluster structure we need to select exemplars from." ] }, { "cell_type": "code", "execution_count": 6, "id": "84542771-ce40-4f43-b7ab-16ea2a90807a", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Layer 0 found 750 clusters\n", "Layer 1 found 215 clusters\n", "Layer 2 found 59 clusters\n", "Layer 3 found 16 clusters\n" ] } ], "source": [ "layers, tree = ToponymyClusterer(verbose=True).fit_predict(clusterable_vectors=clusterable_vectors, embedding_vectors=document_vectors)" ] }, { "cell_type": "markdown", "id": "09ed3d1c-d597-455a-bb38-039a65feb762", "metadata": {}, "source": [ "Now we can import the different exemplar selection methods available in Toponymy. The library provides several approaches with different trade-offs between quality and computational cost." ] }, { "cell_type": "code", "execution_count": 7, "id": "0701597a-933b-4808-9360-6b18f1bf7547", "metadata": {}, "outputs": [], "source": [ "from toponymy.exemplar_texts import diverse_exemplars, random_exemplars, submodular_selection_exemplars" ] }, { "cell_type": "markdown", "id": "47959ddc-6fef-498d-a649-783470c12273", "metadata": {}, "source": [ "## Submodular Selection Methods\n", "\n", "he most sophisticated approach to exemplar selection uses submodular optimization. Submodular functions have the property of diminishing returns, making them ideal for selecting diverse representative sets. Toponymy provides several submodular functions for exemplar selection, each with different characteristics.\n", "\n", "Let's start with the facility location approach, which aims to select exemplars that minimize the maximum distance from any cluster member to its nearest selected exemplar." ] }, { "cell_type": "code", "execution_count": 8, "id": "6551d309-8932-4a00-9fca-a0e3033cd116", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 18.9 s, sys: 1.05 s, total: 19.9 s\n", "Wall time: 2 s\n" ] } ], "source": [ "%%time\n", "fl_exemplars, fl_indices = submodular_selection_exemplars(\n", " layers[3].cluster_labels,\n", " arxiv_ml_df.title,\n", " document_vectors,\n", " n_exemplars=8,\n", " submodular_function=\"facility_location\",\n", ")" ] }, { "cell_type": "markdown", "id": "b5e8475c-4218-40ba-ba90-d6e778e82449", "metadata": {}, "source": [ "We successfully ran exemplar selection on all 16 clusters in the uppermost layer, and it took around two seconds. That may seem a little slow, but actually exemplar selection via submodular selection scales mostly with respect to the *size* of the clusters rather than the number of clusters, so the uppermost layer, with the largest clusters, is actually the more challenging task. For reference we can run exemplar selection on the 750 clusters in the lowest layer in about half the time:" ] }, { "cell_type": "code", "execution_count": 9, "id": "184e9734-9827-4ed7-b32a-e16319576910", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 983 ms, sys: 240 μs, total: 983 ms\n", "Wall time: 982 ms\n" ] } ], "source": [ "%%time\n", "submodular_selection_exemplars(\n", " layers[0].cluster_labels,\n", " arxiv_ml_df.title,\n", " document_vectors,\n", " n_exemplars=8,\n", " submodular_function=\"facility_location\",\n", ");" ] }, { "cell_type": "markdown", "id": "a8145b8c-20b9-43f4-a6f8-468e4bf17a8b", "metadata": {}, "source": [ "So we are expending computation time -- what are we getting for our trouble? Let's have a look at the exemplars picked for a couple of the clusters -- we can look at the same clusters for other exemplar selection methods for comparison -- and see whether they make sense." ] }, { "cell_type": "code", "execution_count": 10, "id": "bacc5468-c9f9-4df6-b59d-abd8e2241ca3", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['On computing the Hermite form of a matrix of differential polynomials',\n", " 'Gr\\\\\"obner Bases over Algebraic Number Fields',\n", " 'On the asymptotic and practical complexity of solving bivariate systems over the reals',\n", " 'Arrangement Computation for Planar Algebraic Curves',\n", " 'A Symbolic Summation Approach to Feynman Integral Calculus',\n", " 'Polynomial Systems Solving by Fast Linear Algebra',\n", " 'Survey on counting special types of polynomials',\n", " 'Differential elimination by differential specialization of Sylvester style matrices']" ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" } ], "source": [ "fl_exemplars[0]" ] }, { "cell_type": "code", "execution_count": 11, "id": "800e7b70-4ac3-4121-9176-573a0a4bd7b7", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['The challenges of SVM optimization using Adaboost on a phoneme recognition problem',\n", " 'Dependent Hierarchical Normalized Random Measures for Dynamic Topic Modeling',\n", " 'Seeded Graph Matching',\n", " 'Structured Learning via Logistic Regression',\n", " 'Encoding Prior Knowledge with Eigenword Embeddings',\n", " 'Multi Terminal Probabilistic Compressed Sensing',\n", " 'Neural Network Influence in Group Technology: A Chronological Survey and Critical Analysis',\n", " 'Neyman-Pearson Classification under High-Dimensional Settings']" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "fl_exemplars[15]" ] }, { "cell_type": "markdown", "id": "62e77396-6f60-4a1d-87ce-9afeff8a2d98", "metadata": {}, "source": [ "These look potentially reasonable. We have a mathematical topic related to the combination of linear algebra with polynomials in the first cluster. The second cluster seems quite broad, and the question is whether this is a poorly chosen sampling, or a good representation on the breadth of topics covered in the cluster. Hopefully we can get a better idea of that as we look at some of the other methods.\n", "\n", "The saturated coverage approach is similar but uses a different objective function based on ensuring coverage by exemplars that can be quicker to compute for large clusters." ] }, { "cell_type": "code", "execution_count": 12, "id": "08e1296b-fddb-42a5-a2f0-6cf907abab2d", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 17.6 s, sys: 2.24 s, total: 19.9 s\n", "Wall time: 830 ms\n" ] } ], "source": [ "%%time\n", "sc_exemplars, sc_indices = submodular_selection_exemplars(\n", " layers[3].cluster_labels,\n", " arxiv_ml_df.title,\n", " document_vectors,\n", " n_exemplars=8,\n", " submodular_function=\"saturated_coverage\",\n", ")" ] }, { "cell_type": "markdown", "id": "067c9f45-7b10-4811-994d-305a271d8089", "metadata": {}, "source": [ "We get faster results, but how well has it worked?" ] }, { "cell_type": "code", "execution_count": 13, "id": "fc63b738-6bbc-4166-9b95-35b85d468990", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['On computing the Hermite form of a matrix of differential polynomials',\n", " 'On the Computation of Matrices of Traces and Radicals of Ideals',\n", " 'Gr\\\\\"obner Bases over Algebraic Number Fields',\n", " 'Computing the decomposition group of a zero-dimensional ideal by elimination method',\n", " 'Computing solutions of linear Mahler equations',\n", " 'Computing and Using Minimal Polynomials',\n", " 'Note on fast division algorithm for polynomials using Newton iteration',\n", " 'Polynomial Systems Solving by Fast Linear Algebra']" ] }, "execution_count": 13, "metadata": {}, "output_type": "execute_result" } ], "source": [ "sc_exemplars[0]" ] }, { "cell_type": "code", "execution_count": 14, "id": "e91b45aa-c915-4033-aa68-afdb0ac4b3c1", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Minimax sparse principal subspace estimation in high dimensions',\n", " 'Fast global convergence of gradient methods for high-dimensional statistical recovery',\n", " 'Regularizers for Structured Sparsity',\n", " 'Optimal prediction for sparse linear models? Lower bounds for coordinate-separable M-estimators',\n", " 'Statistical consistency and asymptotic normality for high-dimensional robust M-estimators',\n", " 'Statistical and computational trade-offs in estimation of sparse principal components',\n", " 'Sparse PCA via Covariance Thresholding',\n", " 'Convex vs nonconvex approaches for sparse estimation: GLasso, Multiple Kernel Learning and Hyperparameter GLasso']" ] }, "execution_count": 14, "metadata": {}, "output_type": "execute_result" } ], "source": [ "sc_exemplars[15]" ] }, { "cell_type": "markdown", "id": "ab95c86d-4046-4352-991c-2c461a6897c8", "metadata": {}, "source": [ "The first cluster again looks string, with the linear algebra and polynomial flavour coming through with some of the same examples picked. The second cluster starts to show the differences in these approaches. Again the exact nature of the cluster is clearly fuzzy (it is likely a large cluster), but the exemplars chosen by saturated coverage largely focus on statistical methods, giving a somewhat different impression than the fairly diverse examples provided by facility location selection. We'll get more sense of this cluster as we continue." ] }, { "cell_type": "markdown", "id": "d2b895d7-ffe4-4b9f-8cc3-ba560d2b51cd", "metadata": {}, "source": [ "## Central Exemplars\n", "\n", "For a faster alternative that still provides good diversity, Toponymy offers the diverse_exemplars function which takes exemplars close to the high dimensional centroid of the cluster and then seeks a diverse set from within those. The aim here is to get exemplars that capture the core idea or topic of the cluster, without having too much repetition. It is also a very fast to compute approach due to its simplicity." ] }, { "cell_type": "code", "execution_count": 15, "id": "aa624e99-757a-4287-b074-62753a0b7f36", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 14.3 s, sys: 29.3 ms, total: 14.3 s\n", "Wall time: 238 ms\n" ] } ], "source": [ "%%time\n", "d_exemplars, d_indices = diverse_exemplars(\n", " layers[3].cluster_labels,\n", " arxiv_ml_df.title,\n", " document_vectors,\n", " layers[3].centroid_vectors,\n", " n_exemplars=8,\n", ")" ] }, { "cell_type": "markdown", "id": "21908a79-9748-4e1a-b19d-9e31209bee02", "metadata": {}, "source": [ "This is clearly a fast method, and will scale well to very large cluster sizes and large numbers of clusters. How about the selection? We should expect more coherent looking exemplars as we can expect them to be focussed around the central subject of the cluster." ] }, { "cell_type": "code", "execution_count": 16, "id": "ecf5ad96-fb24-45c8-ba96-896df3eb9d44", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Polynomial Systems Solving by Fast Linear Algebra',\n", " 'Algorithms for Computing Triangular Decompositions of Polynomial Systems',\n", " 'Constructive $D$-module Theory with \\\\textsc{Singular}',\n", " 'On the Complexity of the Multivariate Resultant',\n", " 'Real Polynomial Root-finding by Means of Matrix and Polynomial Iterations',\n", " 'Fast Computation of the Roots of Polynomials Over the Ring of Power Series',\n", " 'Critical Points and Gr\\\\\"obner Bases: the Unmixed Case',\n", " 'On the asymptotic and practical complexity of solving bivariate systems over the reals']" ] }, "execution_count": 16, "metadata": {}, "output_type": "execute_result" } ], "source": [ "d_exemplars[0]" ] }, { "cell_type": "code", "execution_count": 17, "id": "42cb651a-276c-45b3-a0a3-e2e62a91bca8", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Fast Algorithms for Learning Latent Variables in Graphical Models',\n", " 'High-Dimensional Covariance Decomposition into Sparse Markov and Independence Models',\n", " 'Statistical Query Lower Bounds for Robust Estimation of High-dimensional Gaussians and Gaussian Mixtures',\n", " 'Beyond L2-Loss Functions for Learning Sparse Models',\n", " 'Group-Sparse Model Selection: Hardness and Relaxations',\n", " 'Computationally Efficient Robust Estimation of Sparse Functionals',\n", " 'Fast global convergence of gradient methods for high-dimensional statistical recovery',\n", " 'Noisy Matrix Completion under Sparse Factor Models']" ] }, "execution_count": 17, "metadata": {}, "output_type": "execute_result" } ], "source": [ "d_exemplars[15]" ] }, { "cell_type": "markdown", "id": "73ba847f-105e-4be6-848a-42fafc0a5c23", "metadata": {}, "source": [ "As you can see we get a similar topic result for the first cluster, with a slightly different selection of specific examples. For the second cluster we get something perhaps closer to the saturated coverage selection with a general trend toward statistical methods, but no clear organizing principle. Perhaps the cluster it just quite varied? We can definitely get a sense of that with the last, and fastest selection method: random sampling." ] }, { "cell_type": "markdown", "id": "edefced1-a4f6-4b26-b727-dd373c39c01e", "metadata": {}, "source": [ "## Random Selection\n", "\n", "The simplest and fastest approach is random selection of exemplars from each cluster. While this doesn't guarantee quality or diversity, it can be useful for quick exploration or if a large number of exemplars are required, or you simply need to scale to truly large data." ] }, { "cell_type": "code", "execution_count": 18, "id": "a46483d3-3d04-4085-9ec6-eb81afa0153c", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 1.11 s, sys: 4 μs, total: 1.11 s\n", "Wall time: 17.7 ms\n" ] } ], "source": [ "%%time\n", "r_exemplars, r_indices = random_exemplars(\n", " layers[3].cluster_labels,\n", " arxiv_ml_df.title,\n", " n_exemplars=8,\n", ")" ] }, { "cell_type": "markdown", "id": "ec65833c-ba96-4b76-88d6-b7f2ed6a12e1", "metadata": {}, "source": [ "As expected, random selection is extremely fast and will also scale well with the number of exemplars requested (where the other methods will tend to be a little slower). This also gives us a chance to see what kinds of things these clusters actually contain." ] }, { "cell_type": "code", "execution_count": 19, "id": "110a0615-cc15-4de9-8c09-76f50974c782", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Parallel computation of the rank of large sparse matrices from algebraic K-theory',\n", " 'Symbolic-numeric methods for improving structural analysis of differential-algebraic equation systems',\n", " \"Hrushovski's Algorithm for Computing the Galois Group of a Linear Differential Equation\",\n", " 'Exact Symbolic-Numeric Computation of Planar Algebraic Curves',\n", " 'Survey on counting special types of polynomials',\n", " 'Obtaining Exact Interpolation Multivariate Polynomial by Approximation',\n", " 'Power Series Composition and Change of Basis',\n", " 'Generating Loop Invariants by Computing Vanishing Ideals of Sample Points']" ] }, "execution_count": 19, "metadata": {}, "output_type": "execute_result" } ], "source": [ "r_exemplars[0]" ] }, { "cell_type": "code", "execution_count": 20, "id": "556dba28-9d22-433f-b3b9-65f025f7fad3", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[\"Total Variation and Euler's Elastica for Supervised Learning\",\n", " 'hi-RF: Incremental Learning Random Forest for large-scale multi-class Data Classification',\n", " 'Linear Spatial Pyramid Matching Using Non-convex and non-negative Sparse Coding for Image Classification',\n", " 'Mind the Nuisance: Gaussian Process Classification using Privileged Noise',\n", " 'A New Measure of Conditional Dependence',\n", " 'Ranking and combining multiple predictors without labeled data',\n", " 'Learning Boolean functions with concentrated spectra',\n", " 'Learning sparse gradients for variable selection and dimension reduction']" ] }, "execution_count": 20, "metadata": {}, "output_type": "execute_result" } ], "source": [ "r_exemplars[15]" ] }, { "cell_type": "markdown", "id": "9a7d3b26-4de8-4984-ba91-94c7e4a7ad36", "metadata": {}, "source": [ "For the first cluster we can see the mathematical flavour favoured by the more careful selection methods certainly holds true, but given a small random selection it is clear that the cluster is reasonably diverse. For the second cluster we see the focus on statistical methods may not be entirely representative: there are a lot of general learning approaches here, from clustering, to boosted trees, alogn with statistcal approaches.\n", "\n", "We can take a larger random sample to see this more clearly:" ] }, { "cell_type": "code", "execution_count": 21, "id": "72bad2f1-dc54-4fb1-be3e-a879856d56c7", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 939 ms, sys: 469 μs, total: 939 ms\n", "Wall time: 15.1 ms\n" ] } ], "source": [ "%%time\n", "r_exemplars, _ = random_exemplars(\n", " layers[3].cluster_labels,\n", " arxiv_ml_df.title,\n", " n_exemplars=32,\n", ")" ] }, { "cell_type": "code", "execution_count": 22, "id": "4cb06ff2-85a0-45e9-99cf-c0cc8759d022", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Matrix Formula of Differential Resultant for First Order Generic Ordinary Differential Polynomials',\n", " 'On the computation of the straight lines contained in a rational surface',\n", " 'Generic design of Chinese remaindering schemes',\n", " 'A Note on the Space Complexity of Fast D-Finite Function Evaluation',\n", " 'The Invar tensor package: Differential invariants of Riemann',\n", " 'Partial Denominator Bounds for Partial Linear Difference Equations',\n", " 'Harmonic Sums and Polylogarithms Generated by Cyclotomic Polynomials',\n", " 'Gr\\\\\"obner Bases for Linearized Polynomials',\n", " 'Divide-And-Conquer Computation of Cylindrical Algebraic Decomposition',\n", " 'Real root finding for determinants of linear matrices',\n", " \"Zeilberger's Holonomic Ansatz for Pfaffians\",\n", " 'Factorization in categories of systems of linear partial differential equations',\n", " 'Computing necessary integrability conditions for planar parametrized homogeneous potentials',\n", " 'GPGCD, an Iterative Method for Calculating Approximate GCD of Univariate Polynomials, with the Complex Coefficients',\n", " 'Special Values of Generalized Log-sine Integrals',\n", " 'A local construction of the Smith normal form of a matrix polynomial',\n", " 'Faster exponentials of power series',\n", " 'About the generalized LM-inverse and the weighted Moore-Penrose inverse',\n", " 'Faster arithmetic for number-theoretic transforms',\n", " 'The Hilbert scheme of points and its link with border basis',\n", " 'Relating $p$-adic eigenvalues and the local Smith normal form',\n", " 'Tropical Effective Primary and Dual Nullstellens\\\\\"atze',\n", " 'On matrices with displacement structure: generalized operators and faster algorithms',\n", " 'Generating and Exploring S-Box Multivariate Quadratic Equation Systems with SageMath',\n", " 'GBLA -- Gr\\\\\"obner Basis Linear Algebra Package',\n", " 'Improving the use of equational constraints in cylindrical algebraic decomposition',\n", " 'Rational Solutions of Underdetermined Polynomial Equations',\n", " 'Implementing Gr\\\\\"obner bases for operads',\n", " 'Comprehensive Involutive Systems',\n", " 'On the robust hardness of Gr\\\\\"obner basis computation',\n", " 'Shortest Two-way Linear Recurrences',\n", " 'Symbolic-Numeric Tools for Analytic Combinatorics in Several Variables']" ] }, "execution_count": 22, "metadata": {}, "output_type": "execute_result" } ], "source": [ "r_exemplars[0]" ] }, { "cell_type": "code", "execution_count": 23, "id": "b982e435-b7de-4c7c-99e4-c58870ffe0ed", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['Onsager-corrected deep learning for sparse linear inverse problems',\n", " 'A characterization of product-form exchangeable feature probability functions',\n", " 'Bayesian Inference Based on Stationary Fokker-Planck Sampling',\n", " 'A concentration inequality for the excess risk in least-squares regression with random design and heteroscedastic noise',\n", " 'A Nonconvex Approach for Structured Sparse Learning',\n", " 'Bayesian SPLDA',\n", " 'The Laplacian K-modes algorithm for clustering',\n", " 'A Slice Sampler for Restricted Hierarchical Beta Process with Applications to Shared Subspace Learning',\n", " 'Robust Probabilistic Modeling with Bayesian Data Reweighting',\n", " 'Quantile universal threshold: model selection at the detection edge for high-dimensional linear regression',\n", " 'Out-of-sample generalizations for supervised manifold learning for classification',\n", " 'Graph-based Generalization Bounds for Learning Binary Relations',\n", " 'General Scaled Support Vector Machines',\n", " 'Jitter-Adaptive Dictionary Learning - Application to Multi-Trial Neuroelectric Signals',\n", " 'Kernel Bandwidth Selection for SVDD: Peak Criterion Approach for Large Data',\n", " 'Seeing the Forest from the Trees in Two Looks: Matrix Sketching by Cascaded Bilateral Sampling',\n", " 'Clustering Small Samples with Quality Guarantees: Adaptivity with One2all pps',\n", " 'Faster K-Means Cluster Estimation',\n", " 'PAC-Bayes Iterated Logarithm Bounds for Martingale Mixtures',\n", " 'Finite Sample Prediction and Recovery Bounds for Ordinal Embedding',\n", " 'KohonAnts: A Self-Organizing Ant Algorithm for Clustering and Pattern Classification',\n", " 'Information Projection and Approximate Inference for Structured Sparse Variables',\n", " 'Estimating Probability Distributions using \"Dirac\" Kernels (via Rademacher-Walsh Polynomial Basis Functions)',\n", " 'Scalable MCMC for Large Data Problems using Data Subsampling and the Difference Estimator',\n", " 'Relative Error Embeddings for the Gaussian Kernel Distance',\n", " 'Thoughts on Massively Scalable Gaussian Processes',\n", " 'Efficiently Sampling Multiplicative Attribute Graphs Using a Ball-Dropping Process',\n", " 'Local Identification of Overcomplete Dictionaries',\n", " 'Randomized Dimensionality Reduction for k-means Clustering',\n", " 'Total Variation Classes Beyond 1d: Minimax Rates, and the Limitations of Linear Smoothers',\n", " 'The Generalized Mean Information Coefficient',\n", " 'Efficient Multiscale Gaussian Process Regression using Hierarchical Clustering']" ] }, "execution_count": 23, "metadata": {}, "output_type": "execute_result" } ], "source": [ "r_exemplars[15]" ] }, { "cell_type": "markdown", "id": "a622fe17-adfd-4c88-a963-2e811cbef1fd", "metadata": {}, "source": [ "With a larger sample we see that while the mathematical cluster is more general than polynomials and linear algebra, that is a reasonable representation of the core of the topic. On the other hand the second cluster remains extremely diverse, and likely about general classical machine learning techniques in general rather than any specific subfield." ] }, { "cell_type": "markdown", "id": "a06884e6-7010-40d3-89a8-377435b07f72", "metadata": {}, "source": [ "## Visualizing the selections\n", "\n", "To see what is going on with the exemplar selections it can be helpful to provide some visualizations. While the actual selections were making use of high dimensional vectors, we also have the low dimensional representations used for clustering. We can look at the clusters in this representation and overlay the different selections to see what kinds of exemplars are being picked out by the different methods. For this we'll need matplotlib." ] }, { "cell_type": "code", "execution_count": 24, "id": "ec0e964c-8f58-448c-9703-b8b221808d37", "metadata": {}, "outputs": [], "source": [ "import matplotlib.pyplot as plt" ] }, { "cell_type": "markdown", "id": "232cbdcd-84f7-4ac5-b215-18efbe5054b6", "metadata": {}, "source": [ "Let's start with the cluster that was providing confusing results -- the second cluster which two of the methods leaned toward statistical methods, while the facility location and random selection gave far more diverse results. " ] }, { "cell_type": "code", "execution_count": 25, "id": "58d4a52d-9da0-42c4-b4cc-f3d6bb18ff9e", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[[], []]" ] }, "execution_count": 25, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "fig, ax = plt.subplots(figsize=(8,8))\n", "ax.scatter(*clusterable_vectors[layers[3].cluster_labels == 15].T, s=1, c=\"gray\", alpha=0.25)\n", "ax.scatter(*clusterable_vectors[fl_indices[15]].T, marker=\"o\", label=\"Facility Location\", alpha=0.5)\n", "ax.scatter(*clusterable_vectors[sc_indices[15]].T, marker=\"X\", label=\"Saturated Coverage\", alpha=0.5)\n", "ax.scatter(*clusterable_vectors[d_indices[15]].T, marker=\"^\", label=\"Diverse Central Selection\", alpha=0.5)\n", "ax.axis('equal')\n", "ax.legend()\n", "ax.set(xticks=[], yticks=[])" ] }, { "cell_type": "markdown", "id": "a77bedcf-6ecd-474f-bf5f-19d848813c08", "metadata": {}, "source": [ "As you can see, the cluster is quite large, with a few different branching subtopics spreading out from the center. The central selection approach, and the saturated coverage, tended to focus on the more central regions of the cluster, while the facility location selection picked out items from some of the branching arms as being representative of the cluster as a whole. This has the benefit of providing a more representative overall sample, but has the downside of providing a less coherent overall topic.\n", "\n", "Now let's examine the linear algebra and polynomials cluster ..." ] }, { "cell_type": "code", "execution_count": 26, "id": "98e5fa4a-7e3d-46ce-9524-733625215b73", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[[], []]" ] }, "execution_count": 26, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "fig, ax = plt.subplots(figsize=(8,8))\n", "ax.scatter(*clusterable_vectors[layers[3].cluster_labels == 0].T, s=1, c=\"gray\", alpha=0.25)\n", "ax.scatter(*clusterable_vectors[fl_indices[0]].T, marker=\"o\", label=\"Facility Location\", alpha=0.5)\n", "ax.scatter(*clusterable_vectors[sc_indices[0]].T, marker=\"X\", label=\"Saturated Coverage\", alpha=0.5)\n", "ax.scatter(*clusterable_vectors[d_indices[0]].T, marker=\"^\", label=\"Diverse Central Selection\", alpha=0.5)\n", "ax.axis('equal')\n", "ax.legend()\n", "ax.set(xticks=[], yticks=[])" ] }, { "cell_type": "markdown", "id": "36f33149-ebb0-4f3a-8814-7a7e2c4338a1", "metadata": {}, "source": [ "We can see why this cluster was more easy to select exemplars from -- while it has some outlying topics, there really is a single core topic, and significantly fewer points overall which makes it much easier to cleanly represent with a small number of exemplars. None the less we have a similar result with the facility location managing to select a few outlying points representing some of those branching satellite subjects, while the other methods remain focused on the core.\n", "\n", "Let's look at a few more cluster to ensure we are getting the right intuitions about how these techniques work." ] }, { "cell_type": "code", "execution_count": 27, "id": "c0ec02c3-5628-4d98-86b6-f2c47d638485", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[[], []]" ] }, "execution_count": 27, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAoAAAAJ8CAYAAABunRBBAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjAsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvlHJYcgAAAAlwSFlzAAAPYQAAD2EBqD+naQAAs35JREFUeJzsvXl8JHWd///qHJV0pzud6Uwm00wmIcwwQwIDAh7ArIICDrqL466rq3KIywKuuyquirrrtcuurq547fpz3UNEBZFVFPSrgAegMIhyM8w9ZBIy05kjPel00p10jvr9EauprlRV1310v56Pxzwy3V3Hpz71qc/7Ve/P5/3+RERRFEEIIYQQQuqGBr8LQAghhBBCvIUCkBBCCCGkzqAAJIQQQgipMygACSGEEELqDApAQgghhJA6gwKQEEIIIaTOoAAkhBBCCKkzKAAJIYQQQuqMJiMbLS4u4tChQ0gkEohEIm6XiRBCCCGEmEQUReTzeZxwwgloaND38RkSgIcOHcLatWsdKRwhhBBCCHGPF154AT09PbrbGBKAiUSifMD29nb7JSOEEEIIIY4yOTmJtWvXlnWbHoYEoDTs297eTgFICCGEEBJgjEzXYxAIIYQQQkidQQFICCGEEFJnUAASQgghhNQZhuYAEkIIIXIWFhYwNzfndzEIqSuam5vR2NjoyLEoAAkhhBhGFEWMjY1hYmLC76IQUpd0dHRg9erVtvMyUwASQggxjCT+Vq1ahVgsxsUBCPEIURRRKBRw5MgRAEA6nbZ1PApAQgghhlhYWCiLv87OTr+LQ0jdEY1GAQBHjhzBqlWrbA0HMwiEEEKIIaQ5f7FYzOeSEFK/SM+f3Tm4FICEEEJMwWFfQvzDqeePApAQQgixyAUXXIDrr7++/PnEE0/El770pfLnSCSCH/3oR56Xyw8eeOABRCIRBgiFBApAQgghNc1VV12FSCSy7N++fftsH/vOO+/EjTfeqPl7JpPB6173OgDAgQMHEIlE8NRTT9k+r1Joeo1S+ALAeeedh0wmg2Qy6U+hiCkYBEIIIaTmueSSS3DzzTdXfNfV1WX7uKlUSvf31atX2z5HWBAEoa6uN+zQA0gIIcRTFhdFvJAtYNfYJF7IFrC4KLp+zpaWFqxevbriX2NjI77whS9g06ZNaGtrw9q1a/Hud78bU1NTFfs+/PDDOP/88xGLxbBixQps2bIFx48fB6DuCZMjHwLu7+8HAJx55pmIRCK44IIL8Otf/xrNzc0YGxur2O8DH/gAXvWqV1m+3q997WtYt24dBEHAxo0b8e1vf7vi94mJCVx77bXo7u5Ga2srTjvtNPzkJz8BAIyPj+Ntb3sbenp6EIvFsGnTJnz3u98t73vVVVfhwQcfxJe//OWyN/XAgQOqQ8A/+MEPcOqpp6KlpQUnnngibrrppopynHjiifj0pz+Nv/zLv0QikUBvby/+67/+y/J1E+NQABJCCPGMfUfy+NoD+/HFn+/BV365F1/8+R587YH92Hck70t5Ghoa8JWvfAXbt2/HLbfcgl/96le44YYbyr8/9dRTuPDCC3HqqafikUcewUMPPYRLL70UCwsLps/1u9/9DgDwi1/8AplMBnfeeSde9apX4aSTTqoQaPPz8/jOd76Dd77znZau6Yc//CHe97734QMf+AC2b9+O6667Du985ztx//33AwAWFxfxute9Dtu2bcN3vvMd7NixA//6r/9aTikyMzODs88+Gz/5yU+wfft2XHvttbjiiivw6KOPAgC+/OUv49xzz8U111yDTCaDTCaDtWvXLivH448/jre85S1461vfimeffRaf+tSn8PGPfxzf/OY3K7a76aab8NKXvhRPPvkk3v3ud+Ov//qvsWvXLkvXTkwgGiCXy4kAxFwuZ2RzQgghNUixWBR37NghFotFS/vvPTwp/v2dz4jv+vZj4j/evV38t3t2if9493bxXd9+TPz7O58R9x6edLjES7zjHe8QGxsbxba2tvK/P//zP1fd9o477hA7OzvLn9/2treJmzdv1jz2+eefL77vfe8rf+7r6xO/+MUvlj8DEH/4wx+KoiiKQ0NDIgDxySefrDjGZz/7WXFgYKD8+Uc/+pEYj8fFqakpzfMqzyPnvPPOE6+55pqK79785jeLr3/960VRFMV7771XbGhoEHfv3q15fCWvf/3rxQ984APlz8rrFkVRvP/++0UA4vHjx0VRFMW3v/3t4sUXX1yxzYc+9CFxcHCw4jouv/zy8ufFxUVx1apV4te+9jXDZas39J5DM3qNHkBCCCGus7go4t7th5GdLuHkVXEkWpvR2BBBorUZJ6+KIztdwn3PHXZtOPjVr341nnrqqfK/r3zlKwCA+++/HxdffDHWrFmDRCKBK6+8EuPj45iengbwogfQTa666irs27cPv/3tbwEA3/jGN/CWt7wFbW1tlo63c+dObN68ueK7zZs3Y+fOnQCWrqmnpwcbNmxQ3X9hYQH/8i//gtNPPx2dnZ2Ix+O47777MDIy4kg59u7dW+FBPf3008v/j0QiWL16dXm1C+IeDAIhhBDiOgcnith/dArpZOuyPGaRSATpZCv2HZnCwYki1qacTzTd1taG9evXV3w3PDyM17/+9XjXu96FG2+8EalUCg899BCuvvrqcpJdaeUFN1m1ahUuvfRS3HzzzTjppJPw05/+FA888ICtYyrrWBTF8nfVrummm27CF7/4RXzpS18qz4+8/vrrUSqVTJVBfk75d0qam5uXlX1xcdHUuYh56AEkhBDiOtOleczMLyAmqPsdokIjZucXMF2a96xMjz32GObn53HTTTfhnHPOwYYNG3Do0KGKbU4//XT88pe/dOR8giAAgOr8wb/6q7/C7bffjq9//etYt27dMs+ZGQYGBvDQQw9VfLdt2zYMDAwAWLqm0dFR7NmzR3X/3/zmN9i6dSsuv/xynHHGGTjppJOwd+/eZddSbR7k4OCgajk2bNhgawkz4gz0ABJCCHGdNqEJrU2NKJTmkWhtXvZ7sbSAlqZGtGkIRDdYt24d5ufn8e///u+49NJL8fDDD+M///M/K7b56Ec/ik2bNuHd73433vWud0EQBNx///1485vfjJUrV5o636pVqxCNRnHPPfegp6cHra2t5Zx5W7ZsQTKZxD//8z/jn/7pnwwd7+DBg8tyCvb29uJDH/oQ3vKWt+Css87ChRdeiB//+Me488478Ytf/AIAcP755+NVr3oV3vSmN+ELX/gC1q9fj127diESieCSSy7B+vXr8YMf/ADbtm3DihUr8IUvfAFjY2NlAQksRe8++uijOHDgAOLxuGo6nA984AN42ctehhtvvBF/8Rd/gUceeQT/8R//gf/v//v/TNUbcQd6AAkhhLjOmo4o1nXFkcnNLBsGFEURmdwM1q+KY02H+0OuEi95yUvwhS98AZ/97Gdx2mmn4dZbb8VnPvOZim02bNiA++67D08//TRe/vKX49xzz8Vdd92FpibzQrWpqQlf+cpX8PWvfx0nnHACtm7dWv6toaEBV111FRYWFnDllVcaOt7nP/95nHnmmRX/7r77brzxjW/El7/8Zfzbv/0bTj31VHz961/HzTffjAsuuKC87w9+8AO87GUvw9ve9jYMDg7ihhtuKHv0Pv7xj+Oss87Cli1bcMEFF2D16tV44xvfWHHuD37wg2hsbMTg4CC6urpU5weeddZZuOOOO3D77bfjtNNOwyc+8Qn80z/9E6666irTdUecJyKqDcgrmJycRDKZRC6XQ3t7uxflIoQQEjBmZmYwNDSE/v5+tLa2mt5/35E8bn74ALLTJaSTrYgKjSiWFpDJzSDVJuCdm0/E+lUJF0oeDq655hocPnwYd999t99FIQFG7zk0o9c4BEwIIcQT1q9K4J2bT8S92w9j/9EpHJ6cQUtTIzatSeK1p3bXrfjL5XL4/e9/j1tvvRV33XWX38UhdQIFICGEEM9YvyqBky6I4+BEEdOlebQJTVjTEUVDQ6T6zjXK1q1b8bvf/Q7XXXcdLr74Yr+LQ+oECkBCCCGe0tAQcSXVS1ixm/KFECswCKSGyefzGB4eRj7vzxJLhBBCCAkm9ADWMNlsFplMBgCQSNTn3BpCCCGELIcCsIaR8jKp5WcihBBCSP1CARgy8vk8stksUqlUVa9eIpGg548QQgghy6AADBkc1iWEEEKIXSgAQwaHdQkhhBBiF0YBh4xEIoG+vj5V7x+jfgkhhKhxwQUX4Prrr/e7GCRAUADWENLwcDab9bsohBASGI4cOYLrrrsOvb29aGlpwerVq7FlyxY88sgjho/xzW9+Ex0dHe4VUgWvRVupVMLnPvc5nHHGGYjFYli5ciU2b96Mm2++GXNzc56Vg3gDh4BrCA4PE0LIct70pjdhbm4Ot9xyC0466SQcPnwYv/zlL317WZ6bm0Nzc7Mv59aiVCphy5YtePrpp3HjjTdi8+bNaG9vx29/+1t8/vOfx5lnnomXvOQlvpQtiPVVE4gGyOVyIgAxl8sZ2ZwQQkgNUiwWxR07dojFYtH6QebnRDE7JIqLi0ufFxeXPs/POVHEZRw/flwEID7wwAO62910003iaaedJsZiMbGnp0f867/+azGfz4uiKIr333+/CKDi3yc/+UlRFEURgPjDH/6w4ljJZFK8+eabRVEUxaGhIRGA+L3vfU88//zzxZaWFvEb3/iGeOzYMfGtb32ruGbNGjEajYqnnXaaeNttt5WP8Y53vGPZOYeGhkRRFMXnnntOfN3rXie2tbWJq1atEi+//HLx6NGj5X2npqbEK664QmxraxNXr14tfv7znxfPP/988X3ve5/m9X/2s58VGxoaxCeeeGLZb6VSSZyamhJFURRnZmbE97znPWJXV5fY0tIibt68Wfzd734niqIoLiwsiGvWrBG/9rWvVez/+OOPiwDE/fv3i6IoihMTE+I111wjdnV1iYlEQnz1q18tPvXUU+XtP/nJT4pnnHGG+L//+79if3+/GIlExMXFRfFnP/uZuHnzZjGZTIqpVEr84z/+Y3Hfvn0V53r44YfFM844Q2xpaRHPPvts8Yc//KEIQHzyySfL21Srv6Cj9xya0WscAiaEEOINC/PAs/8H/OYmYN8vAVEE9v1i6fOz/7f0u8PE43HE43H86Ec/wuzsrOZ2DQ0N+MpXvoLt27fjlltuwa9+9SvccMMNAIDzzjsPX/rSl9De3o5MJoNMJoMPfvCDpsrx4Q9/GO9973uxc+dObNmyBTMzMzj77LPxk5/8BNu3b8e1116LK664Ao8++igA4Mtf/jLOPfdcXHPNNeVzrl27FplMBueffz5e8pKX4LHHHsM999yDw4cP4y1veUv5XB/60Idw//3344c//CHuu+8+PPDAA3j88cd1y3frrbfioosuwplnnrnst+bmZrS1tQEAbrjhBvzgBz/ALbfcgieeeALr16/Hli1bkM1m0dDQgLe+9a249dZbK/a/7bbbcO655+Kkk06CKIr44z/+Y4yNjeGnP/0pHn/8cZx11lm48MILKzyy+/btwx133IEf/OAHeOqppwAA09PT+Lu/+zv8/ve/xy9/+Us0NDTgT//0T7G4uAhgaR78pZdeik2bNuGJJ57AjTfeiA9/+MMVZTFSf3WDEbVJDyAhhBBbHsD5OVF88jZRvP0KUfy/q5f+/e5/RPH//nLp/7dfsfS7C57A73//++KKFSvE1tZW8bzzzhM/+tGPik8//bTuPnfccYfY2dlZ/nzzzTeLyWRy2XYw6AH80pe+VLWcr3/968UPfOAD5c9qXruPf/zj4mtf+9qK71544QURgLh7924xn8+LgiCIt99+e/n38fFxMRqN6noAo9Go+N73vle3fFNTU2Jzc7N46623lr8rlUriCSecIH7uc58TRVEUn3jiCTESiYgHDhwQRfFFr+BXv/pVURRF8Ze//KXY3t4uzszMVBx73bp14te//nVRFJc8gM3NzeKRI0d0y3PkyBERgPjss8+KoiiKX/va18TOzs6K9vnf//3fFR7AavUXBugBJIGB0ceEkKpMjgIj24BGAehcBwgxYM89gNC29LlRAEYeASYPOn7qN73pTTh06BDuvvtubNmyBQ888ADOOussfPOb3yxvc//99+Piiy/GmjVrkEgkcOWVV2J8fBzT09OOlOGlL31pxeeFhQX8y7/8C04//XR0dnYiHo/jvvvuw8jIiO5xHn/8cdx///1lz2Y8Hscpp5wCANi/fz/279+PUqmEc889t7xPKpXCxo0bdY8riiIikYjuNvv378fc3Bw2b95c/q65uRkvf/nLsXPnTgDAmWeeiVNOOQXf/e53AQAPPvggjhw5UvawPf7445iamipfs/RvaGgI+/fvLx+3r68PXV1dy87/9re/HSeddBLa29vR398PAOU62717N04//XS0traW93n5y19uqv7qCQaB1BBmVglxEianJoRUpaMPGHgD8PTtQD4DJNJAvBuINCx9BpZ+7+h15fStra24+OKLcfHFF+MTn/gE/uqv/gqf/OQncdVVV2F4eBivf/3r8a53vQs33ngjUqkUHnroIVx99dVVo18jkQhEUaz4Tm0faQhV4qabbsIXv/hFfOlLX8KmTZvQ1taG66+/HqVSSfd8i4uLuPTSS/HZz3522W/pdBp79+7V3V+LDRs2lEWcFtJ1KoWiUjxedtlluO222/CRj3wEt912G7Zs2YKVK1eWy59Op/HAAw8sO748ylpZXwBw6aWXYu3atfjv//5vnHDCCVhcXMRpp51WrjM1Eau8N9Xqr56gB7CGGB0dxXPPPYfR0VFPz5tKpZBOpxl9TAjRJhIB1l8E9J23JPjExSXxJy4ufe47F1h/4dJ2HjA4OFj27j322GOYn5/HTTfdhHPOOQcbNmzAoUOHKrYXBAELCwvLjtPV1VV+AQaAvXv3olAoVD3/b37zG2zduhWXX345zjjjDJx00knLxJvaOc866yw899xzOPHEE7F+/fqKf21tbVi/fj2am5vx29/+trzP8ePHsWfPHt3yvP3tb8cvfvELPPnkk8t+m5+fx/T0NNavXw9BEPDQQw+Vf5ubm8Njjz2GgYGBimM9++yzePzxx/H9738fl112WUX5x8bG0NTUtKz8kkhUY3x8HDt37sTHPvYxXHjhhRgYGMDx48crtjnllFPwzDPPVMz1fOyxx0zVXz1BAUhso5ecmhBCALwY8DG8bcn7J4m/SMPS5+FHXgwMcZDx8XG85jWvwXe+8x0888wzGBoawv/93//hc5/7HLZu3QoAWLduHebn5/Hv//7veP755/Htb38b//mf/1lxnBNPPBFTU1P45S9/iWPHjpVF3mte8xr8x3/8B5544gk89thjeNe73mUoZcn69evx85//HNu2bcPOnTtx3XXXYWxsbNk5H330URw4cADHjh3D4uIi/uZv/gbZbBZve9vb8Lvf/Q7PP/887rvvPvzlX/4lFhYWEI/HcfXVV+NDH/oQfvnLX2L79u246qqr0NCgb+6vv/56bN68GRdeeCG++tWv4umnn8bzzz+PO+64A694xSuwd+9etLW14a//+q/xoQ99CPfccw927NiBa665BoVCAVdffXX5WP39/TjvvPNw9dVXY35+vlzPAHDRRRfh3HPPxRvf+Ebce++9OHDgALZt24aPfexjy8SanBUrVqCzsxP/9V//hX379uFXv/oV/u7v/q5im7e//e1YXFzEtddei507d+Lee+/F5z//eQAvei2r1V9dYWTCIYNAwsHk5KR44MABcXJy0tRvhBBiBFtBINkhUbzrb5cCPn71aVG86z2i+J0/X/r7q08vfX/Xe0Qxe8DRMs/MzIgf+chHxLPOOktMJpNiLBYTN27cKH7sYx8TC4VCebsvfOELYjqdFqPRqLhlyxbxW9/6lghAPH78eHmbd73rXWJnZ2dFGpiDBw+Kr33ta8W2tjbx5JNPFn/605+qBoHI05CI4lJgxtatW8V4PC6uWrVK/NjHPiZeeeWV4tatW8vb7N69WzznnHPEaDRakQZmz5494p/+6Z+KHR0dYjQaFU855RTx+uuvFxf/kFonn8+Ll19+uRiLxcTu7m7xc5/7XNU0MFJdfeYznxE3bdoktra2iqlUSty8ebP4zW9+U5ybWwrOKRaL4nve8x5x5cqVy9LAyPnqV78qAhCvvPLKZb9NTk6K73nPe8QTTjhBbG5uFteuXStedtll4sjIiCiKL6aBUfLzn/9cHBgYEFtaWsTTTz9dfOCBB5YF4Tz88MPi6aefLgqCIJ599tnibbfdJgIQd+3aVd6mWv0FHaeCQCKiWP11a3JyEslkErlcDu3t7S7KUeIWw8PDyGQySKfT6OvrU93GrzmEhJBwMDMzg6GhIfT391dMtDeElAJm98+WAj6ApWHf4W0AIsBCCdj4OmDTm4FGTk8nznDrrbfine98J3K5HKLRqN/FcQS959CMXuNTViforRIiCb9CoYBcLgeAwRyEEIdpbFoSd8BStO/AG5bm/HX0AjvvBnovpPgjtvnWt76Fk046CWvWrMHTTz+ND3/4w3jLW95SM+LPSfik1QmJREJT1ElRvMlkksEchBD3kERg33lLwk8KDFm5AWhfQ/FHbDM2NoZPfOITGBsbQzqdxpvf/Gb8y7/8i9/FCiR82kiFd5CeP0KIqzQ2AStk01AikcrPhNjghhtuKK/gQvShACS63kHiLJxnSQghJAhQABLiIUyaTQghJAhQABLiIXrBOIQQQohXUAAS4iFmhts5XEwIIcQtuBIIQT6fx/DwMPL5vN9FITKk4eJsNut3UQghhNQY9ADWIGY9R1bnpdFD5S4cLiaEEOIW9ADWIGY9R6lUylL+P3qo3MXPNZbteIXpUSZhJhKJ4Ec/+pHfxagrLrjgAlx//fWunuOBBx5AJBLBxMSEq+fx4lqcggKwBjEr6KwKDavCkQQfO+KeLwYkaFx11VWIRCKIRCJobm5Gd3c3Lr74YnzjG9/A4uJixbaZTAave93rfCqpPSYnJ/EP//APOOWUU9Da2orVq1fjoosuwp133gkDq74a5qqrrsIb3/hGx45XjYWFBXzmM5/BKaecgmg0ilQqhXPOOQc333yzZ2VQoiUo77zzTtx4443+FMokHAKuQbzK68f8gc4RlOF0qRyCIJgS9/Lyc+iaBJFLLrkEN998MxYWFnD48GHcc889eN/73ofvf//7uPvuu9HUtGQOV69e7Wo5FhYWEIlE0NDgrP9lYmICf/RHf4RcLod//ud/xste9jI0NTXhwQcfxA033IDXvOY16OjocPSc1Zibm0Nzc7Pt43zqU5/Cf/3Xf+E//uM/8NKXvhSTk5N47LHHcPz4cQdK6Sxh6vfoASSW4DCfswTFayaVo1QqmfIKy8vv59A1CQ+iKOLw9GFHPVN6tLS0YPXq1VizZg3OOuss/P3f/z3uuusu/OxnP8M3v/nN8nbyIeBzzz0XH/nIRyqOc/ToUTQ3N+P+++8HAJRKJdxwww1Ys2YN2tra8IpXvAIPPPBAeftvfvOb6OjowE9+8hMMDg6ipaUFw8PDeOCBB/Dyl78cbW1t6OjowObNmzE8PFze78c//jHOPvtstLa24qSTTsI//uM/Yn5+XvP6/v7v/x4HDhzAo48+ine84x0YHBzEhg0bcM011+Cpp55CPB43Vd57770XAwMDiMfjuOSSS8rzxD/1qU/hlltuwV133VX2qj7wwAM4cOAAIpEI7rjjDlxwwQVobW3Fd77zHYyPj+Ntb3sbenp6EIvFsGnTJnz3u981de9+/OMf493vfjfe/OY3o7+/H2eccQauvvpq/N3f/V15G1EU8bnPfQ4nnXQSotEozjjjDHz/+9/XPe62bdvwqle9CtFoFGvXrsV73/teTE9Pl3+fnZ3FDTfcgLVr16KlpQUnn3wy/vd//xcHDhzAq1/9agDAihUrEIlEcNVVVwFYPgR8/PhxXHnllVixYgVisRhe97rXYe/evYbr200oAIklgiJYaoWgDKdbLYdyP74gkGrsOb4Htzx3C/Yc3+NbGV7zmtfgjDPOwJ133qn6+2WXXYbvfve7FSL1e9/7Hrq7u3H++ecDAN75znfi4Ycfxu23345nnnkGb37zm3HJJZdUGPlCoYDPfOYz+J//+R8899xzSKVSeOMb34jzzz8fzzzzDB555BFce+21iEQiAIB7770Xl19+Od773vdix44d+PrXv45vfvObmmvaLi4u4vbbb8dll12GE044Ydnv8Xi87OE0Wt7Pf/7z+Pa3v41f//rXGBkZwQc/+EEAwAc/+EG85S1vKYuUTCaD8847r7zvhz/8Ybz3ve/Fzp07sWXLFszMzODss8/GT37yE2zfvh3XXnstrrjiCjz66KOG7hGw5JX91a9+haNHj2pu87GPfQw333wzvva1r+G5557D+9//flx++eV48MEHVbd/9tlnsWXLFvzZn/0ZnnnmGXzve9/DQw89hL/9278tb3PllVfi9ttvx1e+8hXs3LkT//mf/4l4PI61a9fiBz/4AQBg9+7dyGQy+PKXv6x6nquuugqPPfYY7r77bjzyyCMQRRGvf/3rMTc3Z6i+XUU0QC6XEwGIuVzOyOakDpicnBQPHDggTk5O+l0UEkAOHDggPvLII+KBAwf8LgpxkGKxKO7YsUMsFou2jrOwuCB+49lviFffc7X4jWe/IS4sLjhUQnXe8Y53iFu3blX97S/+4i/EgYGB8mcA4g9/+ENRFEXxyJEjYlNTk/jrX/+6/Pu5554rfuhDHxJFURT37dsnRiIR8eDBgxXHvPDCC8WPfvSjoiiK4s033ywCEJ966qny7+Pj4yIA8YEHHlAt0ytf+Urx05/+dMV33/72t8V0Oq26/eHDh0UA4he+8AXV3yXMlHffvn3l37/61a+K3d3d5c9q9Tk0NCQCEL/0pS/plkEURfH1r3+9+IEPfKD8+fzzzxff9773aW7/3HPPiQMDA2JDQ4O4adMm8brrrhN/+tOfln+fmpoSW1tbxW3btlXsd/XVV4tve9vbRFEUxfvvv18EIB4/flwURVG84oorxGuvvbZi+9/85jdiQ0ODWCwWxd27d4sAxJ///OeqZVIeT+1a9uzZIwIQH3744fLvx44dE6PRqHjHHXeIomisvpXoPYdm9BrnABJLcP4f0YPzAIkee4/vxa7sLqxNrMWu7C7sPb4XG1MbfSmLKIplz5uSrq4uXHzxxbj11lvxyle+EkNDQ3jkkUfwta99DQDwxBNPQBRFbNiwoWK/2dlZdHZ2lj8LgoDTTz+9/DmVSuGqq67Cli1bcPHFF+Oiiy7CW97yFqTTaQDA448/jt///vcVHr+FhQXMzMygUCggFostuwYAmtchYbS8sVgM69atK39Op9M4cuSI7rElXvrSl1Z8XlhYwL/+67/ie9/7Hg4ePIjZ2VnMzs6ira3N0PEAYHBwENu3b8fjjz+Ohx56CL/+9a9x6aWX4qqrrsL//M//YMeOHZiZmcHFF19csV+pVMKZZ56peszHH38c+/btw6233lr+ThRFLC4uYmhoCM8++ywaGxvLnl4r7Ny5E01NTXjFK15R/q6zsxMbN27Ezp07y9/ZqW87UAASQhyHLwhEi0VxEdsObcPc4hy6Yl3YP7Ef2w5tw8krTkZDxPtZSTt37kR/f7/m75dddhne97734d///d9x22234dRTT8UZZ5wBYGnotbGxEY8//jgaGxsr9pPm3AFANBpdJs5uvvlmvPe978U999yD733ve/jYxz6Gn//85zjnnHOwuLiIf/zHf8Sf/dmfLStPa2vrsu+6urqwYsWKClGhhtHyKgM3IpGI4bmaSmF300034Ytf/CK+9KUvYdOmTWhra8P111+PUqlk6HgSDQ0NeNnLXoaXvexleP/734/vfOc7uOKKK/AP//AP5Uju//f//h/WrFlTsV9LS4vq8RYXF3Hdddfhve9977Lfent7sW/fPlPlU0OrzpQvHXbq2w4UgDVGUKJJiTfwfpOwIXn/0m1L3q7Vbat98wL+6le/wrPPPov3v//9mtu88Y1vxHXXXYd77rkHt912G6644oryb2eeeSYWFhZw5MgRvPKVrzR9/jPPPBNnnnkmPvrRj+Lcc8/FbbfdhnPOOQdnnXUWdu/ejfXr1xs6TkNDA/7iL/4C3/72t/HJT35y2TzA6elptLS02C6vhCAIWFhYMLTtb37zG2zduhWXX345gCXhtXfvXgwMDFg+P7DkFQSWrk0KrhkZGTHssTvrrLPw3HPPadbxpk2bsLi4iAcffBAXXXTRst8FQQAA3XoYHBzE/Pw8Hn300fI8yfHxcezZs8f29TsBg0BqDAZn1B56ARW83yRMyL1/bc1LnqK25jbML85j26FtWBQXqxzBOrOzsxgbG8PBgwfxxBNP4NOf/jS2bt2KP/mTP8GVV16puV9bWxu2bt2Kj3/849i5cyfe/va3l3/bsGEDLrvsMlx55ZW48847MTQ0hN///vf47Gc/i5/+9KeaxxwaGsJHP/pRPPLIIxgeHsZ9991XIQo+8YlP4Fvf+hY+9alP4bnnnsPOnTvLXkItPv3pT2Pt2rV4xStegW9961vYsWMH9u7di2984xt4yUtegqmpKcvlVXLiiSfimWeewe7du3Hs2LGKgAYl69evx89//nNs27YNO3fuxHXXXYexsTHD5wKAP//zP8cXv/hFPProo+UI6r/5m7/Bhg0bcMoppyCRSOCDH/wg3v/+9+OWW27B/v378eSTT+KrX/0qbrnlFtVjfvjDH8YjjzyCv/mbv8FTTz2FvXv34u6778Z73vOe8jW+4x3vwF/+5V/iRz/6EYaGhvDAAw/gjjvuAAD09fUhEongJz/5CY4ePYqpqall5zj55JOxdetWXHPNNXjooYfw9NNP4/LLL8eaNWuwdetWU3XgBhSANYIkErTytzEqM7zoibygRA8TYgSl909C7gV0i3vuuQfpdBonnngiLrnkEtx///34yle+grvuumvZcKiSyy67DE8//TRe+cpXore3t+K3m2++GVdeeSU+8IEPYOPGjXjDG96ARx99FGvXrtU8XiwWw65du/CmN70JGzZswLXXXou//du/xXXXXQcA2LJlC37yk5/g5z//OV72spfhnHPOwRe+8AX09fVpHnPFihX47W9/i8svvxz//M//jDPPPBOvfOUr8d3vfhf/9m//hmQyabm8Sq655hps3LgRL33pS9HV1YWHH35Yc9uPf/zjOOuss7BlyxZccMEFWL16tekk0lu2bMGPf/xjXHrppdiwYQPe8Y534JRTTsF9991Xjm6+8cYb8YlPfAKf+cxnMDAwUN5Ha3j/9NNPx4MPPoi9e/fila98Jc4880x8/OMfL8/DBICvfe1r+PM//3O8+93vximnnIJrrrmmnCZmzZo1+Md//Ed85CMfQXd3d0X0sJybb74ZZ599Nv7kT/4E5557LkRRxE9/+lNH8iPaJSIaGGienJxEMplELpdDe3u7F+UiJhkeHkYmk0E6nVbtJKr9ToKLk8O8HDImdpiZmcHQ0BD6+/tV56LpsSgu4pbnbsEzR5/Buo51y35/fuJ5bOrahHec+g5f5gISEhb0nkMzeo1zAGuEalGXjMoML04GVEjeROm4VqCIJFbITGdwMH8QkUgEz+eeX75BBDiYP4jMdAZr4muW/04IcRQKwBqhmkhgVKZ7hEkQOfEi4ISIJPVHd6wbbz3lrVgQtSfNN0Ya0R3r9rBUhNQvFICE2CRMgsiJFwF6k4kVmhqasH6FsahWQoj7UAASYhNBEBCJRMppAWodepMJIST8cKYtITYplUoQRdF0YtOw4lREOSPTCSHEP+gBrHGk+WmCIKBUKoVinlrYqLchUaeGvMM0dE4q8WKVAkKIOk49fxSANY5kZOVLy9DYOku9DYk6JXjrTTjXAlLuskKhgGg06nNpCKlPCoUCgOVLyJmFAjCEmIk6lYyr3ANIwkHQoovl5XEil6QknKWh4KBcJ9GmsbERHR0d5YXqY7HYsjVuCSHuIIoiCoUCjhw5go6OjqoJzKtBARhClENnekKh3rxTbuGHGPNyiNTI9blVHg4Fh4vVq1cDQFkEEkK8paOjo/wc2oECMIQoh87MCEKzBM0L5RdGRIrTdeXlEKmR66tWHqvXz6HgcBGJRJBOp7Fq1SrdNWAJIc7T3Nxs2/MnQQEYQpRePTVBODQ0hLGxMQwODlY1xplMBiMjI+jt7a1YB1E6llviMkwYESlOe7K89N4aub5q5bF6/fRSh5PGxkbHDBEhxHsoAENANdGlJgjHxsZQLBaRzWareqyGh4dx6NAhAFgmAKt5G+sFeR1r3Q89ERV04ex0guigXy8hhNQ7FIABRjKihUIBuVwOgDHRlUgkMDg4WDbAWkhirqmpCclkEp2dnarHMipy6gUtEawnoupBOMuvf3h42NCQ+ejoKACgp6enZuuFEEKCCAVggJFEQzKZRDqdNiW6JGOazWYxNTWlmgNQOl6hUEBDQ4OhkHIO11kTwWERzk557owOmQ8NDQFYiiZVOx89iYQQ4g4UgD5SzbjJjagV41ctB6A8DUcsFgu8OAkKVkRwWISzHU+lsj0bSVHU399f/r/T5SGEEKINBaCPVDNudkWDZFTn5uYwPj6uuVZtWMQJcQa9Fw81z51RL5xZsZZIJDAwMKC7TVg8p4QQEjYoAH3EKeOmZqDl32Wz2bpaq5booyfU1F4GjAo7N8QaX04IIcQdKAB9xCnjpmag5d/Ri2KfWpqLZrY9sP0QQkjtQQFYA8iXe5OW1FLOHwy7aPGbsM9FMzs/T47R7f2qo1oS54QQ4hUUgDWAZKDlqTf6+vpoDB0k7F4wL8SZX3UUdnFOCCF+QAHoM056L8IuUoJM2L2obrQNNa9itWTZbsB2Twgh5qEA9BilYXTSexF2kULcw422odd2nfbK6QlKeTojaQoEnwNCCNGHAtBjlIYxaN4LzqdiHRhFr+063a6NCEoOBRNCiHEoAD1GaRjNeGa8ECY0oqwDo+i1Xac9jkYEZdBepgghJMhQALqImmBTM4xuJdo1Uy4JGlF7yZDDTlCv04igrLZNUK+NEEL8gALQRYwKNrXt1IyVU+LMbCLgesNOMuSwY7QtViNoYiufz2PHjh0oFosAavseEkKIESgAXcSoYFPbTs0QOyXO6OUzT73UmdG2qIUk/AqFAnK5nKF93EK5Gk6xWEQ0Gq35e0gIIUagAHQRpWDT8oqoCTsrgqPaknBGhGTQPDdBoV48o3bboiQWk8kk0um0r2JLazWceriPhBBSDQpADzHjSbEiOKotCRfk1RxIcDHTFoMktLgaDiGEaEMB6CFuDyNWWxIuCGUktY2fQksvMTUhhJBKKABrCCeWhKPRJHaxO43AzP7KeX563mv5tgA41YEQUtdQAHqIV8OrTqQx4VxAYhWtdu5GuiOteX7VtgXAqQ6EkLqGAtBFlAbPq+FVJ9KYKI0lxSAxilY7N9oGtfZX8+AJglAONqnmvVY7Lqc6EELqFQpAF1EaPD+HV82KT0EQEIlEIAgCA0OIKbTaudE2qLW/mgcvnU6jr6/PUrnYlgkh9QwFoIsEKaDCrPgslUoQRRGlUilQ10HCg9NBGUY8eJy6QAghxqAAdJEwB1SopdDI5/Pl6OKwXhfxDqc9x0Y8ePRWE0KIMSgAiSr1vBwacQY/PMf0VhNCiDEoAB3AqWGnTCaDkZER9Pb2Ip1OO1hCc2hdD41r/eBEm/bDAx5mrzshhHgJBaADOOUZGxkZwdDQEAD4KgC1rofGtX6w2qY5B48QQsIBBaADOOUZ6+3trfjrF/T0ESNtQE3scZoAIYSEAwpAC7i15FQ6nXbN82fGM0NPHzHSBtTEXpBfHuidJISQF6EAtEAYvRxhLDMJNmpiL8gvD3wGCCHkRSgALeCXl8OOByPInhkSTrwWe3Y9eHwGCCHkRSgALaBm+LwYXrLjwQiyZ4YQI9j14PEZIISQF6EAdAgvhpfowSD1DNs/IYQ4BwWgQ3hhnKx6MDj5ndQC9OARQohzUAA6hN/GSU/kcfI7IYQQQuRQANYIeiJP8koKgoCdO3cCAHp6eigGCSGEkDqFArBG0BuCloTejh07kMlkIAgCYrEYBSAhhBBSp1AA1gjVhqCz2SyKxSJSqRTi8TgKhQLy+TxFICGEEFKHUADWAfl8HoVCAd3d3ejp6SkPF7vhBWTACSGEEBJ8KADrgGw2i1wuh3Q6XSHK3IhYZsBJbUOBTwghtQEFYB2gnB/oZsRytXQ4FBDhhgKfEEJqAwpAUhUzos3IXEQKiPDCZMyEEFIbUADWAdVEVzWB56Roo4AIN37nuySEEOIMFIB1QDXRVU3gOSnawiIgOFRNCCGklqEADCFmxUk10VVN4IVFtDkJh6qDA8U4IYQ4DwVgCHFanNSjwKsGh6rtYUa0eTkFgRBCyBIUgCGE4sR95KKYHijzmBFtym2V9c32TgghzkMBGEKC5LGTjLUgCCiVSjUpkuiBMoZcuJkRbcptlfUdpPZOCCG1AgUgsYVkrCORCERRBFB7IokeKGPIhVtfX5/hdqAUeKxvQghxHwpAYgvJSMs9gHYI4nArPVDGcEq4SXWdzWYrPhNCCHEOCsA6wg1x5bQ44nBreHGyLbAdEEKIu1AA1hFhMKoc/gsuXnpn2Q4IIcRdKADriDAYVQ63BhcvXyDYDgghxF0oAOsIr41qEOfzaRGmsvpFGF4gCCGEGIMCkLhGNpvF0NAQxsbG0NvbG+g0MWEYHvcbJ4IzKLQJISQYUADWIEExsqlUCmNjYygWixgZGQl0mhh6t4xhVyib2V+rHQelfRNCSJihAKxBvPBmGTHCiUQCg4ODyxJFBxHOOTOGXaFsZn+tdkxvLSGE2IcCsAbxwpulZYSVwpDCqrawez/N7K/VjumtJYQQ+1AA1iBeiC4tIyyf9zc4OEjxRyyj1Y75UkEIIfZp8LsAJJwkEgnV5b5SqRSi0SiKxWI5WECPfD6P4eFh5PN5t4pKLMD7QgghtQ09gMRR5PP+JO+g3nzBsMznqrfAg7DcF0IIIdagACSOoxyi0xMTYZnPVW+CKCz3JQjU28sBIaQ2oAAkrqMnJoI6n0tp1OtNEAX1vgSRens5IITUBhSAxHXCKCaURj2M1+Al9ewFq7eXA0JIbUABSIgK1Yx6PQseNerZCya9HEiBM2wThJAwQAFIiAXqWfCo4acXLChinG2CEBImKADJMoJiUP2kmjHnsF8lfg6Reym89J4NtglCSJigACTL8NOTERTxWc2Yc05gcPBSeOk9G2wThJAwQQFIlqFnUPP5PEZHRwEAPT09jhu8oAyj0Zj7h9mXADv3ysy58vk8CoUCkskkvXyEkNBDAUiWoWdQpaXeACAWi5UnvyuNqFVPHofRiJcvAWbOlc1mkcvlkE6n+XJACAk9FIDEFKlUCv39/eX/5/N57NixA8ViEcCLRtSqEQ+L5y0oQ9W1iJcvAWbOpbUt2wIhJIxQABJTJBIJDAwMlD8PDw+jWCwiGo1CEIRyGoxa9+SpCVwKgUqs1oeXLwFmzqW1bVCmLRBCiBkoAIlp5IZdLvTkhrCvr6+mjaGawKUQqKRe6qPWX3YIIbUJBSAxjCT8CoUCcrkcAHWhVw+GUM0bRCFQSZjrQ+m9lD4LgoBSqVTh1XTKY0kPMiHESygAiWEkj04ymUQ6nS7PAZSMFgAUCgUUCgVXIoT9wIxRDsv8Ra+wWx9+CiKl93J0dBRDQ0Nobm7G3NwcVq9ejc7OTs2yWSl7vXhMCSHBgAKQGEbu0ZEM1PDwcNloAVgWIRx2aJT9Q6vuvRCGat7LUqmEQqGApqYmTE1NYXJyEmNjYxgcHFxWDr12o1X+MHtMCSHhgwKQGEYe4St9VhotKUJYHhASZuFEo1wdtwSZVt17IcqV6/t2dHQgnU5jYmICHR0d6O3txcjICCYmJrBjx46yCJTyZBaLRc18gVrlpweZEOIlFIAEgHEjrjReSqMlRQjLPYNhNmo0ypWotRO3BJlW3ZsV5VYEqiTkjhw5AlEU0d/fj87OToyNjWF6ehoTExPo7e0FABSLRYyOjiIWi6FQKJS94KeeeioAYOfOnQBeTJzOlwpCSBCgACQAjBtxNeOltjoIjVxtotZOqq0c47R30KwotyJQpYTnpVKpPN91x44dGB8fx+HDh3HkyBGcccYZZU9gsVhELpdDMplEd3c3pqamIAiCauJ0pXcx7F5yQkg4oQAkAIx7VZTGV0oELRnYycnJ8nAYjVrtodZOqq0c47cn2MrLiDzhufRS09vbi8nJSRw7dgyLi4s4cuQIgCUP4OzsLLq6utDT04NsNou5ublytLA8cbqcINQNIaR+oQAkAKwPdWazWRSLRaxYsQKRSATFYhHZbFb3WHopNYIGU3NUYradKMWXH/UpL7OR80vbKCPZ0+k0SqUSmpubUSwWMTMzg6NHj6KhoQGLi4sAUG7X6XS67AHs6ekp/yaVB+D8UkKIv1AAEtNoJYIGUJESRgvJ8xGJRCCKIoDgekD88NLUkuhUCka/vV7K86u9jOiVUWrbc3NzeO655zA5OYl169YhGo3i6NGjWFxcRH9/P/r6+pZFyCuPSS85IcRPKABJVZSCRG/FDyMGTTKicqMbVPzw0hgRSWEViX57vdReWJQvI3pllOp6x44dmJ2dRalUwvj4OABgYmICq1atKkfAS55A+XGsRseH9X4TQoILBSCpilKQ2DXiVj0ffg8f6qFXNrPlNlK/TnnS7Nap2f399nopz6/2MqJWRrmnUEr/snr1aoiiiOPHj2NqagoAEIlEMDExUQ4IicViFedVegWN1p3fnlNCSO1BAUiqohQkegbSTXEWZCOoVzaz5TYikpzypElRqloJjY3s7/Y9caJtaR1DnspGmrM3NTW1bG6qVE/FYhENDQ1obW1FW1sbIpEImpqaykKvWCzi6NGjiMfjGB0dxfT0NDZu3FhOjyQXnE8++SSy2WzF71r47TklhNQeFICkKkYEiRdCIMhGUK9sbpTbKU9aKpXC2NiYoeAdOXKPmHKY02mcaFtGBLoUxJTL5SCK4jLhNjY2htnZWXR0dKC5uRl79+7FwsICGhsb0dHRga6uLhw+fBi5XA7T09OYmpoqB4comZiYQDabxezsrKHy++05JYTUHhSAxBG8EGdBNoJ6ZQt6uQcHB5cF71TzukmiKZ1Oo6+vz9UyOtG2jAh0aXj36NGjWFhYQKFQKM/lK5VK6O3tRXt7O4ClNa8bGhowOTmJSCSC9vZ2RKNRRKPRspewu7u7nBpGQr6e9saNGwGg4ndCCPEKCkDiCFoih5PXg4/avVPzmOlFf3tdvmoo250RgZ7P59He3l6O6i0Wi3juuecQj8fLnk4A2L17N9ra2tDd3Y3W1lYsLCwgmUyiWCyiu7sbHR0dmJiYAIBlqWTU1tMmhBA/oAAkrhLkeXtBJCiCWU3g6UV/B41q7U5rSbtcLlf2au7cuROlUqmc5DmVSqFQKGB2dhZzc3NIJpNIJpOIx+MYGRnB7OwsTtu0CfMtSWSKk5g6fgyt0eiyOYdBrjdCSP1AAUhcJcjz9oJIUASzmlAJ073UWrJQEn1GlrTr6enB5OQkisVieRm3np4eFItF7N+/H4cOHUJHRwfa2toAAEdnG3HrE0dxrJTF/GIEWJzH48cO4c/EKM7o7/bs2gkhxAgUgMRV6PEwR5BFVhDupVEPabVhbXk9y48pn88oLf82MjICQRDK33V2dmJkZARNTU1IpVKIxWIoNLXj8al5TJWKWNESwZrODszMN2DX4Tz+9zf78d62GNav4nNACAkOFIDEUYIyhBlWgiCygowdD6ly/p20vzw3n/KYpVIJoiiiVCpVHEcewLEoirhz1zRKkSmkY7MQhGYIzU1YXJjBupUxZAoivvfwbvzthRuQ/EMQCSGE+A0FIHGUoAxhktrErIdULRhE65hqq3SonS+RSFTk7XshW8B4qQlrV8QwP7OIaDSGXC6Hubl5rFjRgZVYwK5Dx7Fz+DDO2dTOlyRCSCCgACSOEuQhzFrDDyHht3gx6yE18kKitkqHmfV6jxyfRDaXx+q2CEQRKBYLmJubR3NzE6LRKIQWEUen5tEcW4o03rFjB4rFom6ZCCHEbSgAiaP4MYTptyjxC6PeVifrJ2weXjMvJNUCR7Sud66QR2lmGscXI1iZjKOhoRFAAclkEoLQgvzMHFLJBFataEc2exTFYhHRaJQvSYQQX6EAJKEnbKLEKYyKm9HRUQwNDaG/v7/qkmNOnTMomHkhMZoPUSKfz2N0dBRNpQJOiDfiQG4BqzubEIkseQIXFhYgiiIyuRlsWpPEmo4oppuZB5AQEgwoAEnoCZsocQqr3lY7HsF6C1LRa1vS+sAAcPHAGty1M4d9x4roWRFDrK0NCw3N2HloAtGGBWw+MYGGhkjd1R8hJLhQAJLQo2dU63V4WE5PTw9isVhZxNSrx9QKem0rlUqhv78fwFIdt0YP4e7Hh3FofA7ReDvirRGcmGxEf/M0Zo4cQH5ljPVNCAkMFICkpqHYWS5itHLgVaufoItpJ8snDe8Cy5dzk1BGA5+17gQIM8cxmi2gc3UHTuhKYSF/FOPHCigWi8hms4GsN0JIfUIBSGqaeh0e1sNoDjwlVpZX8xIzYr9aWeXDu9IqINWOkUgkcGJfHxoiI+hNt6NUKiAzOYmurq4KD6ydchFCiFNQAJKaRmsIL+yG1qny242SleO3t9XMtVQrq3x4V+t4asE18sTRysTT+Xx+WZ5Bs+UihBCnoAAkocWOCAq7obVSfrX6shslK8dvb6uZa6lW1kQigY6ODoyMjGBqasrwcQVBQCQSgSAIy8pj5J75XYeEkPqBApCEFqeWBQsjVsovry/ps1UPol0x6TdGyjoyMlIeBk6n08t+7+joQGdnJzo6Osr1USgUli0dJ2HknoWpDgkh4YYCkIQWKyJILlz6+vrcKprrWBEK8vqy6wENuwdVDaWo7e3tBYDyXyWlUgnFYhEjIyNob29HLpdDMplEOp1WbZMUd4SQIEEBSEKLlfQvtShcjKJWX1Y9oGH3oKqhbBvpdFrV8yeRSqUwNjaGYrGI5uZmRCIRdHR06O5DCCFBgQKQ1CRaQq8WhYsV7HqjvPRmeRWwo2wb0nkFQSgHdUjnl37r7e1FqVRCoVBAqVRCqVQKfYARIaQ+oAAkNYmW0OMwnL8R0FbOLaVkGRsbw+DgoGtl1graWFraTSxvI/8tnU6jr68P+XweAFAoFHDw4EE8//zzWLVqFc4999y6b2+EkGBCAUhCT9gDErzGyDC4WyLRyhC8fKjVy2TK0suD3AOo9tvw8DAEQcDk5CSKxSKOHTuGY8eOYX5+nsmfCSGBhQKQhJ56ntdnBSPD4G7VqZUh+EQigcHBwfJwbLVcek6h9xIh/SYl0o5EIigWi4hGozjhhBMwMzODFStWoFAoIJ/PO1ZWvWFpQggxAwUgCT1hn9dnx9tmZV8j3lG36tSqZ1YpuKTv5PgxtC0JsaamJnR3d6OnpwcAsGbNGhQKBeRyOc2VRKygNyxNCCFmoAAkoSfsw712vG1ueeqCWqfKoVe52HOiLsyKyFKphKmpKQBLok8+BSGfzxteAs5o2QqFApLJJDo6OpYNSxNCiBkoAAlxADveJzveNic9dXrX4JR3ze5x9DyBTtSFWRGptmScfJjWSbLZLHK5XNX0NIQQYgQKQFJXBCW4QVkOq2Vx0lOndw1OeRqdOo6a2HOiLlKpFAqFguG5e4lEorwOsIRbw7Rhn+pACAkWFICkrghKcEMQA1e08uClUinHxIdTx3FriDqRSCAWiyGTyVTM3TPz4qAWPVxtfyPHD+qwPCEknFAAkroiKMENWuVw2kNp5nhaefAAoK+vz5HyeCFi7NahmhDesWMHisUiAP3UOaOjowCAnp6eiu30gleAYL4QEEJqGwpAUlcExYuiVQ67Q8nK34wIFy3COuTotJjKZrPlFC96gl1KWA1gWeRvtbq0WtdcdYQQYhUKQEIChJNDyXrCxQh+iWWnPXhmUdap/Hh6cyPn5uYwNzeHrq4u1WFfvWuxWtf0HBJCrEIBSIiCIA3DVkNP7GgJF63yeJ2PUAu7osaucFXWqd7x5NtK3tbGxsaqOQudIqxeWkKI/1AAEqLAjSFEp5ZeU/MqmQ0a0CpPUPIRCoKASCTieBoVo5gRkPJte3t7y3/lOfvcFGdBmdJACAkfFICEKHDaq+Lk0mtOiEmt8gQlH2GpVIIoiiiVSraPZQezXk15fr7h4eFyzj6vvciEEGIECkBCFDjtVTFyPKMCSm07pTioJhK1yuNlPkI9QROUYU07Xk2nr4Fz/QghTkMBSEgAkAsoPXGkJrT0ghaCipagCZKny049Ov0SEYZ7SggJFxSApG4IkrjQwkrqFjNBC0FBa8UNerrUCcM9JYSECwpAUjfoiQs1ceiHYLSSuiUs4kBZn2orbgTJ02VFjIbhJYMQQgAKQFJH6IkLNWPvhzdKL+dc2DEyVB0kMWtFjNKDSQgJCxSApG4wms9N7zu3MSKAwuplCttQtZXyBcmDSQghelAAEgJ1Yx9UgTI6OoqhoSH09/djYGDA9P5yAQnAMzEZ1PokhJB6hAKQkJBjxCOoXLM2k8mgUChgcnLS8lrBtYodDyuHgAkhYYECkJAQkc/nAQD9/f3o6ekBYEx0yLeRPH+FQsHWWsG1SpDy/xFCiFtQABISIuTDv3qRs0ovljK4JJFIIJ/PIxaL+TKXUCqfIAgolUqOlMGpSG4jIk5vGJ2eP0JIGKAAJCTkGEkOHbQ5jlL5IpEIRFGs+F4p4EZHRwEAPT09ppfJs+LNM1Iv8uMC4LAvISR0UAASooMXEbdmztHT01P22unh5VCkHS+b3AOYzWYxNDSEsbExDA4Olpe1GxoaAoCKfIFqZSgUCkgmk45Gcmtdm17UeFijtAkh9QUFICE6eDGpv9o5lILCSDm89O456WUbGxtDsVhENpstD1339/cD0Bdx2WwWuVwO6XS64rh260Hr2pTHlf+fgSCEkDBAAUiIDl540qqdw4yg8MP75FQdJRIJDA4OVsytSyQShlLduHWfrByXgSCEkDBAAUiIDl540qqdw4yg8MP7ZKaO9ASqHfHq1n2yclwGghBCwgAFICEBR0tQqAmmoHuf9ARqkIdOOa+PEFJrUAASIsNNQ1/t2GbPrSaYgu590hOo0ndNzc347bN70RxLYNWKdqzpiKKhIeJpOZWo1TVFISEkzFAAEiLDTS9UtWObPXfQvX1q6AnURCKBw0XgOw/vxq5Dx9HY3IpUMoF1XXFsOa0b61f5J7LU6jrIHktCCKkGBSAhMtwUVdWObfbcQff2mWXfkTxufvgAjuQWsHpFHCsSbZhHI7YfymH46CT+eGMbzlp3gi/XrFbXYRTghBAiQQFIiAw3RVW1Y9eaoDPD4qKIe7cfRna6hIETOhCJvDjkG29pwtMHjuBnT0+gt6Oloo78HIat5/tFCAk/DX4XgBC/yOfzGB4eLq+v69W+TuNUWfy8pj0Hj+HpA2NYGWtAJBJBqTSLXG4CpdIsIpEIelfGkV0QMNvYVrGfNAw7OjoamPtBCCFhgB5AUrfYmcMVpPlfTpXF6Wsy4507dDSL3FQB3fFmAG0oFovI56cAAILQgva2KLLFRaC5peLYgiAgnU6jUCgE5n4QQkgYoAAkdYvROVxq69EGaf6XU2Uxchwzos6MoDyhK4Vk/CjQJAAAotFoxd9iaQEtTY1oE5oqjp1Op9HX14d8Pm9oiTxCCCFLUACSusXoHC619WjdnP9ldl6bkbIYOaaR45gRdWaE6YY1K3HGiTlsP5TDirgIQWiBICx5+0RRRCY3g01rkmhvmsfw8HDZ8ydfMYSeP0IIMQ4FIKk7zAoso+vROoVcZEmf7QY5ODW8a0bUmRFlDQ0R/FF/ArtHj2DnoXms7YwjKjSiWFpAJjeDVJuA157ajYnjlZ6/IMG8gISQMEEBSOoOs2LI6Hq0TiEXWX4INz3c9LR1NMzg/PQidk0JGC/M4fDkDFqaGrFpTRKvPXUpD6AU4xGEoV6l4AvSvFBCCKkGBSCpO4I0f08NvZxzTh7TKZzyfKVSKbx0A3DRihWYnG/CdGkebUJTIFYCUUMp+ILergghRA4FIKk7wjRfLAxldcrzJb/WJCRheRTTzS8KyyB52ZSCLwz3ihBCJCgACSHLMOPVc8vzpSb26GUjhBBnoAAkJCTIRRngTHCIFmY8bW55vtTEXpC8bEHyRhJCiFkoAAkJCcroYDfFhyAIiEQiEATB8WMbJUhiTw16IwkhYYYCkNQlYUzZoSY4pP87fT2lUgmiKKJUKlV8H8Z6cws9gcp6IoQEHQpAUpeEcfhOKTjk/3f6erS8W2GsNz9gPRFCgg4FIAkVTqYckf/1i6Bej5Z3y496y2QyGBkZQW9vL9LptGfntUNQ2hchhGhBAUhChRspR5zAqpAL6vX4fR45IyMj5aX4wiIAgz5/kRBCKABJqPDLs6In8PL5PHbs2IFisQjAnJCrB0+RUXGstl0+n0dzczNOOOEE9Pb2elVkQgipeSgAia+Y9Zz55VnR89Rls1kUi0VEo1HTQs6p6wly0IFRL6fadtlsFqVSCX19faHx/hFCSBigACS+EpbJ8nqeOvlviUTCFzEW5HpUqzu1OtKLcq5lDykhhPgBBSDxlbAYeD1PnfI3P8RYkOtRre7U6siINzQIns4glIEQQuxCAUh8pRYmyysFgR9izM96tCKIjNaRUigGwdOpVQYKQ0JImKAAJKFFK2jAKSNs9FhKQVALotYMRkWZsj7VxJMgCCiVSppiOgieTuZIJITUAhSAJLRoBQ04ZYSNHisIosRPrHrzgBeF3/j4OMbGxhCPx8vLz6mJaa/FtdpLQJByJBJCiFUoAEko8CNowOix6s3jp8To9avV5+joKIaGhtDc3AwAiMfj6OzsDIyIMvNCUe/tgBASLigASSgwGjTgpBG2eyzOCatErz5TqRRWrlwZuLqiV48QUqtQAJJQUM0QB1FscU5YdXp6ehCLxQJ13+TQq0cIqVUoAIlnWBVpRvYLothyw3sURKFrBwosZ6m19kEIcQ8KQOIZVkWakf2COFTnhriR5sz19/djYGDA0WOT8BPEFyFCSDChACSeYVWkGdmPnqTaQ8ubRS+XOvl8HoVCAclkMlAvQoSQYEIBSDzDqkgLqrjzQ4jI58zVOlreLPn30mcz96AWBKTaNWSzWeRyOaTT6dBeFyHEOygAiWuEwdDaKaMfw216YjgM9W0GLc+v/Hsr98CL++bmfFdA/RqCOA2CEBJcKACJa4RhPpKdMgbN4IahvgHjIkdL7Kp9b+YeeHHf3JzvCqhfg1Qv+Xwew8PDNfMiQAhxBwpA4hpBE0hqCIKASCRSXn1CiZ5YCdrQtFTPgiCYEgBeew7NiKNqZbNyD7y4b27OdwX0xV5YXgQIIf5CAUhcI2gCSY1SqQRRFFEqlZb9ls/nsWPHDhSLRQDBN6ZSfQ8PD5sSAF4LBjM5HcMqZqx648w+MxwKJoRYhQKQ2CZMc8+UZdUzltlsFsViEdFoNFTG1KwA8FowVBM5clGjVjY32puRY1o5r1kBq3YOvfOmUikUCgUUCgXk83nV9ZMJIUQNCkBimzB5aaSyFgqFcjRtX1+f6rZy8RH065JjVgAETTAo611ZNrtRwGoYacNW2rn8WqwmNNc7byKRQCwWQyaTQSwWC9R9JIQEGwpAYpugDjnl83mMjo4CWEqfIvf4FQqFqiLCijAKkzc0qFSrd7tRwNWOaWcbJfJrMTI0r3aOaucN6vNHCAk2FIDENkHzIElks1kMDQ0BQNk7Ip+bJXkAnfRghskbGlbsRgEbPaaVbfSwmtC82nmD+vwRQoINBSCpWVKpFPr7+8v/l2NURJj16GkZeXoGK3GqPsyKHz/vA4UaISRIUACSmiWRSBhaL1fPMJv16Gkdyy3PYBiFpVvR1Ubn2A0NDWFsbAyDg4OhqTNCCHEaCkDiCWEUKoBz86vcmqdlVFgGqf7diq42UhepVApjY2MoFovIZrO+14VVgnQ/CSHhhAKQeEJY58ZZHbZTGmi3hv+MCssg1b9b0dVGEmEnEgkMDg6W701YqXY/KRAJIdWgACSeUG+Ril4JLqPCMkj175YYNpoIuxbm4lVbwSZIgp8QEkwoAIkn1ILRNUOQBBfgff376YEKWt27gd4KNkB91AEhxB4UgMRV6nUoqt4ErxKnPVBm2pFe3ddKe6wm8Oq9/RFCqkMBSFylXoaiakVYOIXTHiin5rzVSnukwCOE2IUCkLiK2aWwwoZ0TYVCAblcrvx9rV2nWZwWKNUEpZawM7P2MyGE1BMUgMRVzC6FFTYk4ZFMJpFOpx1fWYQsYWZ5ODnKe6F2nFp8MSGEkGpQABLPCIP3xc7KH04vT0aMoyUQjbQ5CnZCSD1CAUg8I8jzltSGcq2WNcjXaZUwesmMljkMLyaEEOI0FICEQH0o18x+QG17j/y8Tqvi02iZa1GwE0JINSgAiacE1ZNkdXWKevEe+XmdVsVnvdwbQgixAgUg8ZSgesysBgfUi/fIz+u0KuTq5d4QQogVKACJp4TJKxNUsRpG7Hh+nVqPmRBCyItQABLbOLVKQ9AIk1gNOn6IaQp4QgjRhgKQ2KZWDa0TYpVeqCX8ENMU8IQQog0FILENDa02tSqOzeKH5zdM3mZCCPEaCkBiGxpabSiOtZF7RwFvl8+jZ5YQUu9QABLiEGqioh7EsRN5+gA47inVKxc9s4SQeocCkBCHqEVRYUTcOZmnT81T6kYiaHpmCSH1DgUgIQ5Ri6LCiLhzKk+fFwJT69yEEFJvUAAS4hC1Iiokj5sgCBgfH0epVIIgCJrbW7luM149JoImhBDnoQAkpEapJrK0fpc8bpFIBOPj4wCAUqnkaNnMePUkIZfP5zE8PAxBEFAqlTwN4GDQCCGk1qAAJCSkVBMl1USW1u+Sp00QBLS3t1d8Z7dMEoIgIBKJ6HoWtcobiUQgiuKycrtJLc7v9AIKZ0KCCwUgISGlmiipNnSq9bt86DSdTjtaJolSqQRRFE15FuXCVPIAekUtzu90E0n4FQoF5HI5ABTOhAQNCkBCQko1UVJtDpyROXJmPThGhZIVQeVUea3kH+R8QnNILwLJZBLpdJrCmZAAQgFISMiQC5i+vj5Xz2V26FPaJpvN6u7jlqAyUl638w+GGaeGbOUCn/VKSDChACQkZHg5Hy2VSqFQKKBQKCCfzxs6n5/z5Yx4FuXzD+PxeNXt6wmn7h09poQEHwpAQkKGU/PRjHh7EokEYrEYMpkMYrGY6WFgr4MAjHggrcw/rBc417ESp9svg2JIkKAAJCRkKL0rakbF7goe8v3NigJ5+YaHhz33BpoJjmF0byX03FXidPtgeyNBggKQkJCjZlTsruAh37+vr8+ysXLCo+R0IIqayKHHi6jhtEeUHlYSJCgACQk5emvqKg2NUkw5vbybEic8SlYCUYyekx4voofT7YPtjQQJCkBCQo6aUdEyNEbFVLW5dF7OZQqT10SvXjj/i8hheyB+QwFISB1hRkzpiUUv5zJ57TWxY5idqDMKg/qA8wGJ31AAElIHWMkdqCcWw+SVM0s1w6wn0JyoMzeEgRlR6da2pBK3niHeE2IUCkBCQoLUscuXQjPawauJimrH0/O81fJcpmqGWU+gqdWLWfHthjAYHR3F0NAQ+vv7MTAwoLutGQFKL5Y13BRpvCfEKBSAhIQEqWOPRCIQRRFA9Q5eLvKUS3JZOV49UE3cVhNoSuMu1XOhUEAsFgv8snNmBGgqlcL4+DiGh4fLbawePVBmr9lNkVbL3nniLBSAhIQEqUOXe+yqIRmadDq9zPtk5Xh6WFlj16nzeTlUWk2gKY27VB+FQsFVz4xe2Xt6esris9o+Ztdcnpubw8jICKanpxGPx+vSA2X2mq2ssGMUv18gSHigACTEBdwQJ1Y6dj1vgNOGwus1drPZLIaGhjA2NobBwcHAJOpV1rlUz/l8XlWEOYXZoelq+2iRz+exY8cOFItFFAoFNDc3Ix6Po6GhoeIFoJ48UFaSpZtdYYcQp6EAJMQFguIFMbI0mlPo5SN063xjY2MoFovIZrOOXJsRQ15N3GuJLbc9M1aEl5V9stksisUiotEogKWl9TZu3FgxvF1vgsbplzNCvIACkBAX8Ltzl4sUr8So0gh6kR5mcHCwwuvkBUER90qsiBCtffSWFxQEAf39/eU6NzKvkSynHoUyCRYUgIS4gN+du1yk+C1G3cTNYWy3V0kxg9eBFWr1oBVJTBFDSDihACTEB9wOmJCLFDNeHq8JQhnkGBF3foh7t72OyvugVQ+lUglHjx5FT0+PqXIE7T4TQigACfEFpwIm9IRkGHK4BaEMQGU9plIpjI6OAkBZ6PgtYNz2Oirvg1r76enpweTkpKU5l0G5z4SQF6EAJMQHnAqYsCMkgzA0HIQyAMvrcWhoCADKEZp+Cxi7XsdqAtao59PqnMug3GdCyItQABLiA04FTNgRkn7PUwxKGYDl9djf36/6fSqV8t0baIVqAtbofZBvZ6YegnKfg0IY2xCpPSgACfEBpwyAFSFZi8bH7jUp61G5XJr89+Hh4dANZ5rxwBmtS7+9omHGSN05/ZzW4nNP7EEBSIgP+Gk8a9FwO3FNRg2k3eFMPwyxGQ+c0bp0Q1TWC0bqzunntBafe2IPCkBCfMDPOVG1OB/LiWvSSnOiRCmmgrQOrBPIlwgcHh42nfBajaBfs9cYqTunn9NafO6JPSgACfEBP+dE1eJ8LD+vyco6sPK/QUOqSyeHuoN+zV7jxJrTZqnF557YgwKQEINYHcbi8Fewke5PR0cHTj311KqeLyVGxY28HfT19TlSdjdxUrRRfFRCjygJAhSAhBjEaqett5/eklv1lKDZz2uW7k86nUZfX59pz5dRcRM2o692XW4nMK8XqonrIPQBpPahACTEIFY9Inr7qYmCIAgFr1ee8POa1VK9yP+6dZ4w4lQC83qn2jxSN58HiksiQQFIiEGsDmPp7aeXx09PKLjZiefzeRQKBSSTSc9WnlC7Zq8MlVoqHTfO5/YwqF59OVWXTiUwr2fU7oWR56HaMYwShBdMEgwoAAnxETVRYEQouNmJZ7NZ5HI5pNNp1wyE0sCpXbOZa9QyiGH0dlgts159OdVerOSdJJWo3Qsjz4PeMcy0mVrwRBNnoAAkJIS42Yl7YSCcToOhJXD8SLhrBiPeIKPorVZCox8c1O6FWe+w8hhqbUarXTMgh0hQABISQtzsxINiIMyUQ0vg+JFwVw8jc72M5uFTIq8vZSCLm/c0jF5WP3HiXiiPodbO5W2rrS2OgxNFTJfm0SY0YU1HFA0NEVtlIOGHApCQOsau8Q6K8dcyqnY8jW5EaBuZ6+VEHj4nPX7VrplzyvxHrZ1L935isRU/uG8Hdh06DjQ0Ix5twbquOLac1o31q3i/6hkKQEJqEK/Wc3XL+HspLLVEohsR2mbmetkRcU56/KpdM4eX/UVvqPdwEbjj4QMYPTqBGEpY2dGM5mgzth/K4VCuiHduPpEisI6hACQkIDiZY82N9Vzd2F+LIHiVzERoGxWsZoSZX0PxZucPBmXKQL1gNGXM4qKIe7cfRna6hIETkpiZmUE0GoUgNCPe0oS9R6Zw33OHcdLKOIeD6xQKQEICgpM51owKM7vG2y3jHwSvkpkIbSciloOC8loo8NzFbHtQrlmt9awcnChi/9EppJOtaGlpxtzcHI4ePYpkMol4PIF0shX7jkzh4EQRa1MxV66NBBsKQEICgpM51pww2vl8HqOjowCAnp4e08eTDJsgCCiVSqYEj1uiwy3x5UTEslPYvcYgiO96wm570HpWpkvzmJlfQEyIAgByuRyOH58AAMTjCUSFRhyenMF0ad564UmooQAkJCDo5Vjzw2uUzWYxNDQEAIjFYqbPKxm2SCQCURQB+B8kYNTYmq1vuxHLTt5ftwQFcQczgjufzwMA+vv70dPTo7ttm9CE1qZGFErzSLQ2I5lMAkD5b7G0gJamRrQJlAH1Cu88ISHAjzlxqVQK/f395f9b2R9YSmcyMTGBQqGAfD7vWPmtiCa9XHly3KxvNYHl5PnowfMeOwLejOA2k6R9TUcU67ri2H4oh3hLE+LxBOLxpX1EUUQmN4NNa5JY0xE1VV5SO1AAEhIC/DDqiUQCAwMDtvaXjFSpVEImkyl7Ep3weFkRTXq58uR4vTSdk/eXHjzv8eoFzUw7aWiIYMtp3TiUK2LvkaW5gFGhEcXSAjK5GaTaBLz21G4GgNQxFICEhIAwGHW9KGYjKxeYxc0IZre9dMByQen2/Q168EmYcfsFTX7vUqkUstksgOrtcP2qBN65+UTcu/0w9h+dwuHJGbQ0NWLTmiReeyrzANY7FICEEEfQi2I2snKBWbyOYHbayHs9rK+8PxSDzuG2gDeaIUBN5K9flcBJF3AlELIcCkBCiCOYiWIOg0dTidNl9npYX36+IORZJMa9skafLa372tAQYaoXsgwKQEKIYaotj9bX11felsJCHa36chs1ActAEX8xKsTl906KBFbD6jrSpD6hACQkYPi1Pq+R/dxYHs0LgjT/zYv6qna9YfTA1iJWvMB67Ue6r3bWkSb1AwUgIQHDr/V5jexnZnk0L6kmeIIkUr2oryBdL9HGihA30n4EQUAkEoEgCLbKR2obCkBSNwTJC6SHHYGQz+dRKBSQTCZN72/kvGaWR/OSaoInCCLVK8y2gbA8F2QJI89bqVSCKIoolUoelYqEEQpAUjfoiYQgGUE7gspMolgnz2sHo3Wvt101gScfrpZ/9gO3vXNm2wC9hbVHPb3wEOtQAJK6Qa9TrBUjKAgCZmdncezYsUCIWSMYrXsjc5+cOI+cTCaDkZER9Pb2Ip1OG9qnGm4bZ7PHp1hQJ0gvhWZRPg9hvhbiHhSApG7QEwlGlwgLOqVSCdPT05iensbKlStDUX6jAsTodlr3r9r+oijiSOEIVsVWIRJZypE2MjJSXg/ZKQHopqfVStsNwhB+EPH6pdDNfqdWXnCJs1AAEgLjS4T5iREDIa3fWygUHF971y2MChCj22kZu2r77zm+B3ftuwtb12/FxtRGAEBvb2/F36BDQ+8cXntGjd47q2tgh6lPIN5AAUiIgqAOiRkxEInE0vq9koiV1t4NOrZT3+SOY2LkOUy1n4yD03m0RVdgRUMeWJgHGqt3c4viIrYd2oY9x/dg26FtOHnFyWiINCCdThv2/AXBc2ym7eqVNwjX4jdGXzicqiuj985IP6C2LCMA5HK50PQJxH0oAAlR4MeQmNKIqBkVM8Y9qCJWC1ueq4V5jD54Myb3PIz/1zaIR2PTOH8+hfa5vegcvACrN19RVQTuPb4Xu7K7sDaxFruyu7D3+N6yF9DNa3BaaJlpu3rlpSfROFbrSm0taCP7G3m21ZaOSyaTSKfToekTiPtQABISALLZLIaGhjA2NobBwUFVo2LGuIdtXpdlwbowj+Ff/Q+KO3+O0mIjWsTfIIZGoLSIXL4dpd9+H7Ozs+h7zV9pikDJ+1ecLSK5mERxvljhBVTDrkCX8FNo6ZU3bC8QfmK1rqzeeyPPtla+zjD1CcR9KAAJCQCpVApjY2MoFosVQzf1YoCtCtbFiReQ3fkgpucbMJzqxLAwiZNmZ3CguRm7W9uwbnIC2Pkg1p59CRpSJ6oe46nRp/D4C48jHoljanYK7a3tVb2AdgW6hJX77JTXUK+80m/5fJ5LilXBats1kqzZ6r1Wlon3jqih/npLSA0iGTO9tTT9IpFIYHBwEP39/eXOvq+vL3Qdt5t1LIoiDk8fhiiK5e8OYhV+2/JHiLY0Yrghg9lIMxrRiaLYiKGGQxCagUda/ggHxS7VYy6Ki3jgwAOYKkwh2hhFPB5HZ6IT84vz2HZoGxbFRdX9UqmUI8NpVu6zJD6leV1u4uW5ahm158JIsmYj9a/1zAW5vyPBgB5AUjcEfV5T2IZt1XCzjtWidKfnFvBs2ysQbT2AQ6UHkZxPYX5xER0RAUebj2PHyrPwXMsr8Jq5BdVj7j2+F6OlUaxNrkV7oh0tLS0AgNUNq3W9gFbvlRPeOy+9w/XmibaC1TW0jdSt2fl+8vMHvb8j/kMBSGoStaAKq0uk1QJeRXXqGSw7ZdCK0m1rbsQZs7/HodJjmBFa0AUBi1iA0NiMo2jFWOlJnI4NaGse0Dym2CCiu6O7/B0ARJuimFuYqzoXUAuta1VOzndieM8sZu5DLbyUuI3VNbSN1K2RbbSGkineSTUoAOsctXQBtTDfR9kp21kizQ+cEGzyY3jlDdAzWMoymLlGrSjdNTiCtfO/wu8aZtAeSaFlsQhBLKAkxtDe0I79DVm8ZP5XWBO5BEBbxTEz0xkczB9EJBLB87nnl580AhzMH0RmOoM18TWm6kGrvuVG2S8PDT1DzmJ1DW2n0BpKpngn1aAArHPU0gUA4TcMyk45bG/DThhp+TGcuH4zgs1IlKxRb5helC6Sa7A33QWMjaB1ZhKLjQ041tqD+MwYWkszmGxaxN7VK3FJ+wnLytgd68ZbT3krFkT14WEAaIw0ojvWbayCZGjVt5pR9rpNhu1Z8Jtq7d5roaVcnpD3k1iFArDO0UoXEHbUohj7+vr8LlYFeobFSKdezTDJj2HESFU7nhlRqkxro5bnzKg3TPL+JSNJTE1VRunGhTjyK1ejZXYNjuaP4lhkBXKIISl0YKWYRUtiDfIrVyMzc3SZF6+poQnrV6zXvY5qaNWZU8N7TpXH7fPWMkHzmI6MjGDv3r0YGxvDhg0b0NPTE7i+jYQDCsAax+zbaxA6OCcJWuctR69sRoz06OgohoaG0N/fj56enmX32ayhr1ZXZhNRy9PaGPWcKI8tef/mFueQakkhN5NDoiWBIwtHsO3QNlw2cBneOvB2LGz4c4hTh3FEXIHi/CKiTQ1YFTmOSLwbjY2CJS+eEfTqzI/VNILc3sNK0Dxsvb29yGazmJqawtDQkK2VPbjiS31DAVjj1LtBCFrnLcfJsjlxn6uVx2wiaimhtZHr0zq25P1Lt6UxNz0HAFhYWMDqtqUo3aHc0ItRuisHcYqh0jmHXp358ewFub2HFac9pmZFl3L7dDqNeDyO0dFRAPbudb3bh3qHArDGqXeDEOThLrtl6+npQSwWc2z43um6sns8yftXWigh2hRFQ2sDFsVFtLS2oKWpZXmU7sI8MDkKMdmLI8WjWBXtQiQ3ArT3GFoPWI5RI613jX48e0Fu72QJs6JLK+n4wMDyyHYlZqaJkPqDArDGoUGoXWp9+F41SjcCYOYP/+RRutFu4Nn/Q2n/g3hsxQb8SpjGm+O92DjyGNB7HrDpzYZEoGQwC4UCcrkcAPP1Kje6qVSq7Knp6OhAqVTicFudk0qlUCgUUCgUkM/nq7YFOyKtmtikfahvKABJKOHcldrHcJRuSyfw7P8Bu3+GmZlZPHH8LuyNNWMbBJzcnEbD7p8tbWxABEoGM5lMWl7pQxndPDQ0BADo7Owsr2LCNlu/JBIJxGIxZDIZQ/P3lCLNTN9HDx/RgwKQuI4bYq3W5q4ERdDanZ+kyh+GZtHRB0QigCgCE8NVh2YNR+kePwCMbAMaBQx3pvB8NoO+uSJ2NYvY296JjfksMPII0HcesEI/WlIZOW0FpdHt7+8HUOkBdIKgtJl6RK3uvRJmahH2WtDDR/SgAKxDvDYcboi1WnuzDYqgdWJ+UgUL80veuZFtwMBWYP2FwL5fADvvNjU0q0tHHzDwBiw+9V08VhjCohBDh9iKkcgMfp3dgXQpjqYNb0Cso1f3MEafC7OR9UbmalkhKG2mFrGSEkkvr6XyeHaEmZEIe0KMQAFYh3htONwQa7X2ZhsUQWu2HLrbS+Jv98+ARgF4+vYlz9/wNszNz6P05PfRMDOD6MuusCcCIxFg/UXYO/Y4dj1/N1Y2p1Cam8PK5ibsmj2Kx6L9WNd+OvoikYrdJKMsCAJKpZLqvD81IRAU4RWUNhN2rNxjvfypanktpc+FQqEcuGXVw242wp4QLSgA6xBl52XFI8j1RJfQq4d8Pl8OAOjp6QnFUI3ZcuhuPzlaHppF5zognwH23AMk0phuimJ+Yjdanv8NsPE1VYdmdRFFLO69D9te+DXmmlvRLsQwHykh1ixgfDGKPbM78fLcU4D4hyHoPyAZ5UgkAlEUVef9qQmBoAivoLSZsGPlHqvVvV5eS+lvoVAw9PJQLRhJLlTl18HpAMQMFIB1iLyjyufz2LFjB4rFYvk3IwTFC+I3evUgzdUBYCtZqxt4Mg3gD0OzePp2IJ/BbEsKxYUWRFvaEJsaQykaRcNpbwSqDM3qkc/nMTn8DI7v+A52LU4jHe1C03wRTXN5oCGBE6IrMVQ8ioP778bG3jMrhKZklCUPoNEVWfSEV7V65bw9e7hRf2bvsRHUIvSloWBl6iY1jAQjObmMJ9tlfUIBWIOYeZiz2SyKxSKi0agpj0ZQvCB+o1cPqVSqHAAQtHryZAWLPwzNYmIE2HMPigstmJoqAAA6ZsYhDF4CbLq0witn5ToO5hbxRHMLSrOLiM7ksQgAqXVA7gVE50uYExexLR7HyYk0GmT7Kl+EpNUVSqVShSg0s8xWtfWN+eJkDzfqz0tPqtFzGQlG0huGNgvbZX1CAViDmHmYrUY9muk03Xq7DMJbq149yJO1ytckdrusRurFkxUsRBHY9wvM7f81ZiJxNDY1Ix6PIRprAxrTwPAjS17C9RdaFoGpVAqHi4dxZD6NSGQCzxeyQCINtCUBcQbIHwIS3TgYX1leD1hvzpc0HCz9ldeB2XoN8vBxWAlb/ZnxCAOVLwxGptbI95HanPw3o4StXokzUADWIIIgIBKJYG5uzjPRoYdbb5dhemv1sqxGzqVnYKoZg2rzHsu/zY8DO+/GbGkWuUgn2o6/gJbScUQWe4BUHzC+fykaeOXJpucAys9zcvpkXLR4MRL9WxFbnALa0y9uOJkB4qvQ2ChAzIl49LlH0dzcjFKpVFE/yuFguQdQwmq96g0tBuElJkyEbd5jtTZjdxjXqWHgsNUrcQYKwBqkVCpBFEWMj49XTTzrhTDRExR2AlAEQbCcrNdrvHzDtnuuasag2rzH8m89PUDveWjJ342OhXEszC9grP0l6J7aB0GcBxZKQO+FQPsa02UcHR3F7t27kUql0NXVBWFSQKqtC30nvrRyQ2mdYACP7nkUQ0NDOOGEE9DV1YVjx46hUCiUA3Sqzesz296MGNXR0VEMDQ2hv7/ftXQxxD+qPYtWhnGVK82Y3Z8QCQrAGkRtcnu1bd3sNPQMoRUBKu2TTqdNzc/ykyDOM7JKtXmP5b+NTcCmN6MZgLj/1xhb8TLMrj4Hkfm9wIFfLIk/G3kAZ2dnkc1m0dXVpSnM5Mayt3cp2KS3txelUslwgI6f7Y0ewnCgdZ+qPYtqwSLVkPeZfX19uvuz/RA9KABrEDsCwOsOw4oA5XwVf6k277Hitz+IwCNNfTiUjyDd1oZY7xuAtS9Z8vyZEH/yttnT01P+Xp5iR9l+lcYynU6XtzMaoONme+vp6dGNCg3TNIdawUofqHWf3Ipa1lpLWDmn0EqGB1I/UADWOVoJS6XPTqDXCVoRq5yvEhyMTnIXOvuRTvzBGx2JWMr7J6XVOXDgAFatWqWaW1HZfrXEmzxAp9o1ySfbS/sauXYjVGvLfNmpjtOpd6xEyGuJMreilrXWElbOCbSS4YHUDxSAdY7SwLhhcILgxeBQiD5a9WNlSSy1350YPk2llpbAymQymJ6eVh26VbZfKy8L1Zb5ctO4KzFSfuU9qre2bibQQq8+jMz11DqWmijL5/MoFApIJpOm+9Nq91Crn9aaE1gP7YCYhwKwzlGbg+J0ZxEEL0YQRGiQ0aqfakEKynurNFxO3vtEYmkJrPb2dt1zmg0mUhpIKYpeEATN69T6zi5ODD/WS1s3Gpxj9D4ZeVkxPP/1D8fL5XJIp9OOvITI0Wrnav25lymoSLigAAwBYX+jN2qUjVyn1boIggj1E6sehWoo763ScDn9QqE2dKtmLI0MC+7YsQO54+NY1VJC98aXozQ3h9SKFVgcfx7iAsqpYtSuU+s7u1gRb1548YOINCUgGo1icHDQ+LxUDbTqzegLhvI3O/dBvq9erkCtMsqplxcCYh4KwBDg5AMcZDFp5Dqt1kW9zxu06lGoFqSgRMtw2fXKGT2nhCQOxsbGylG/8mNms1nMTE+h+8hv0FXYi6PHn8doy0acFh3D6uO/R0vHqWjpGKwol9F1ne1gRTR44cUPItKUgGKxiGw2a/uaterNjz5Hvu/w8DAymQwKhQImJydRLBZRKBTKz2U1r2+9vBAQ81AA+ozdVRvMEuS3QSPXqTY85ydeiRy7VKtbq2kslMi3HzpwAM/uP4j4ihmcfOJarOmIoqFBf8UPK+1TrYxycTAyMrIsH6bQ1IA1xx9B99xuRGICWo79Bg3NLyApjkJoT2Llsd8DQyvLaWokQQkspY2Ryur0PawX8WYXqb3Kxb2Tx5XfV7P9r+FE6SZfcMbHx5HJZMqf5c+JXr/INkW0oAD0GburNpglyG+DRq5TSnItH57zEzt5DM3sY5dqdWt2GFVvWCqfz+OJ/YewbXgaQ+Mi8MIE4nuLWNcVx5bTurF+lftRr9J8QWmOmCQSJE9efuRZ9Bx9CpGYgPjaMyBkhxE9vhdNK5ZWKZk7vAszO36OyMozEO8ZRCpVua6zMtoyqF71WkBv6T6nczOqPQdqz46VIVdp2oEyLYuZIB5BENDV1VXhmc/n8xgZGUGxWAxMv0jCAQWgz1gxeHY8SGF/GwyagLVSHr1hUr+G6LWGUeWGTF42pQCSb/fE/kP49m9fwFxDC9alU4gJTSiU5rH9UA6HckW85SWr0NEwo3qNTrZPtWMNDw9jaGgIpdlmtPecj66px4B8BmJ8NUoNCSw0NWM2sweLMwUc6noVYvMxxKGdNkZZF054fNRw2qsUJtTq161+wEzAiDTFQJp/KA9KSSaTqilh1NKyKI+l9jmTyUAQBMTj8bKA7OvrK4vKiYkJdHR0BKZfJOGAAtBnrBi8IA/juk3QBKyV8qjN75G+t3tvrQoFrWFU+V952ZS/STnQcpOTeHxsHnMNLRg4IYmWlual47c2I97ShL1HpvCTp17Aq1fPWb5GO8g9eWvWXIjm3TFgzz0oLrRgaqqAhfk5tE4MYbb3fMQ2XYpUZ6fqceT1NTU1pTn8piYUtLA6kb/W+gMjkeR2+wG7Ux7U5h/KvZJqefqU1yFF5iqPJQgCisVieaUb+TM2NjaGqakpdP6hXe7YsQNHjhxBU1MTent7a+L+E++gAPQRRrQSpyM4nRQKSmMolWlubg47duyoMDiSwTs8NYddhybRuzKOlpZWlEqzZa+HILQgnWzFC+M5jCeBfo/ncVYEcqxZg8ThR4HhbUAijWhLGwCgWJzFbPMKJHM7kJzbB8SrDy/qTUswE6hgdSK/0ehVM/jpVfRC0FqJHJcjn2Kg9+yqidZMJoOnnnoKDQ0NOOWUU9DX11dxrGw2i2g0WvYUyveLx+NobGwsv3AVi0U0NTUhGo1y+JeYhgLQRxjRGi7cMIrKe2n33loRCmqoXatUtkcffbQcECEtrSYd87mDE8hNFdAdbwbQhmKxiHx+CgAgCC2ICo2YKS2gOBfRNFhuiQ95IEf7QhaJ5+8GEAESabTkM2gpZBBr7UShZwNiM2PAzruBlSeXVy2RD/HJo4r16lVNKGhR7Thm5wjbEVJ+ehX1PM9uzoW2+4Kkdh+y2crVYwBgZGQE4+Pj6OzsLJ9fywMv/65UKmFxcRHT09OYm5tDd3c3+vv7IQgCJiYmVJeGI0QPCkAfoSfPHdwUEGZWFPDae1LtvGbEpd61dnZ2IpvNorOzE5lMBiMjI+js7ERzczNWdyaRjOeBpiXvXjQarfg7OV1EAxaxKrVCs92bMcRm6lo+/Nt+wmpg/jzMPXc3Zl94Gi1CC5o3XAJheBuE+XEgIgK95y6tV6woVyQSqYgqrlavRuvdyRc7O6tQAP4mudbyPFfzcJr14Bn1sho9rt461PL9ent7y3/NPKeSN3l2dhbRaLQiHVGpVCqfS54ehhA9KAB9hJ48d3DLe2FmgriX3hPJ8BQKBeRyuYpyWDUEetfa3NyMFStWoLm5GSMjI9i7dy/27duHtrY2nLxhA844cTW2H8phRVyEILRAEFoAAKIoYuTYFNItC+jratcslxnxYbauK4zjpjcjN34MC0MPY3rda9H90rcDHb1Lnr/eC8spYJTlmpubw/j4OObm5jRXWPA7MCObtb4KBVDZNzl1LU6PeCiPZ/e50ztPtXmcahG+Wu04nU6XPedmyyf3JquJ5EKhUFPzQYm7UADWIUH1UDmFW55VMxPEled389olw5dMJstLYpkJPlBD71rl1zc3N4cDBw5gdnZ26d/MDNZH5/BcaRo7D81jbWccUaERxdICMrkZdHe04XUb2rBSI7hC7dx6SZhticXGJrS89HLkus9Gsvc0IBIB1l8ErNyw5PlrrOwepXINDw9DFEWMj48vyy+odi7ps9VIb6eSY1tBTdgEZe6y0/Nn9c5TbR6nWoSvnRd8ybPe29tbIRa1jil9n8/nTSVuJ/UNBWAd4tf8Hq/O67dnVWsukFvXLjd88mM7uUqCHPn1ZbNZdHd3IxKJIB6PI5/Po3j8OM5ub8ZYYwzjhTkcnpxBS1MjNq1J4rWnLuUBNLM+qTIJs5E5k0YT+iaSK5BInvfiflNTyE4CwszRZauHSEj7y+cCKs8rX5/WiKfK6aW81ES0FdGmJmyCMne52hw8I8sBGh2SrjaPUyvC1+r1joyMYGhoCLOzs+U2NjU1pSoKlWWl548YhQIwZKgl4FVOSq+GX3MP63nOo9612/UOqnX6RoyWE6RSKRQKhfLnfD6PVCqFjV1daE8m8cL4NJpjCaxa0V6xEogZESGfu2f0Wowm9NXaTznPT46a0JAMvjwViDJBsZ6nymoEsFGsija1l4uwPMd6Xljps56X3OgawIB+aiclWsFE8t+kNC/Nzc3lYx04cAD79+/HzMzMMgHo95QDEk4oAEOGWgJePWOlF82pxO1OxK/zOonVsuoZELdy/3nhDZCOv3v3brS1taGnp6c8TDs8PAxMj2Nlu4C1qVjFUG5HR0fZQ2YEsxPbjQ7Da+WcU3r3jHroVL2MVTxVWvvpbW8Wq0soar1cBP05BV58OSnnqPzD/Fi5kFV6yfWSnZsJNpL+qgWqSEPq0Wi0ot/O5/N48sknkc1msXHjRrziFa+o2P/YsWOIxWKIx+PLzllruSCJN1AAeoRTIkfNUKgNRUmY6RhqfWjYCbxKSQEYT+rsRpnMDJ8BwOzsLObm5tDe3q55XfKh3FNPPdXwEl5Wo4KVx1c7jvI7LXFj1ENn9SXHbVHl5RKKmUwGe/fuRTwex8knn+zLMy3VNwDkcrmK+bESSi+5cr6j/L6aaYPy+XjK+ZPyIXX5OsbStocPH8bi4uKyYwHAhg0bsHLlStV+PixeWRIsKAA9wikjreZN0MNMx1AvQ8N6QQXVcKOsVkRHNa+THbTWLNU6f09PDwBgdHQUmUwG7e3tGBgYWHZdVoZy5dtaCvSochyjx7broRsdHcXQ0BD6+/tVl5RzG6MeUavIjzUyMlKOCu/s7PRFAKoFRlUT3sPDwxXzHbU8tUZfjqREzfL5k2oeZslbXiwW0d3dXV7rV6+sZn4jRAsKQI/wS1yZ6Ric7ETs5uRyE72ggmp4WVa9NqPldXLCqKtN/K92fknUSPWqhnw7M5ipc6PeVD1PoZUyBH0ag1r5lWLZzjXIj9Xb24vZ2VnE43FDgskNpCFvaaqBEdTmO0qYmeMnF5/9/f2q02927tyJoaEhrF69Gp2dnRAEAd3d3QD0X0qD3s5IuKAA9Ih6eUNTy0kXtOu26onyGitv/E54mvUMod75e3p6EIvFIAhC1fx4ZgOXjGK0XszWU7Xh+GoeU6lu3G5vZhNjy/9aaTtqUc+JRKJCdFUTTGYxco1WhryN9tHVXuaNPD+FQgGTk5NoampCqVTSXD9YOdWiWjsjxAwUgDVCUN4M1YZegobSE+VW3Tl9XK3jyb93wtNs5WVFylu2du1aHMofQmm8VD6WHGWUbaFQ0AzwcLL+lPVitp6qDcdX85jK5xyqHcMOVgMX1Ibo5X+1ziHfRy/q2chxq51D7TsjQtXNEZdqz4eR5ycWi6G9vR0rV66sWBIO0I4OB1C1nRFiBgrAGiEogRRG3n6DRrW6s5NDzco9qWZslceTf9/X1+davevVg5S37IWZF7CrcRf+aMUf6Q5dSx5AaeUCNSHopJdOMsrylC2SYDFyf5WCQr6PIAjlSf165ayWdsQI1QSRXuCCnUAUrXuRSr0YaStfh9ZIChW1pdOU9aN2XiPiLugjLnKPsLycanMOpb9TU1OG2hkhRqEArBH8mmOoJOgdrxrV6s6JHGpm0DO2asdz4t6r5Zc0KkCBpXlfi+IiHpp5CDvHd2JldCVeFX+V6vHlniJp5QK1Jayc9NJJ51IbQjNyf5XtWtpHGsorFotVhxtTqeorSqiVudoQoNZ80KmpqYrUL3ZeErXuRSKRUB26tHIv1OpH7bxG+xi/R0WMvJBUQ75dNpv1LJKb1AcUgDVCGIVXUKhWd1YFltV7omdszXpujCI32IVCQTViVa8e0uk0JlsmMfzoMJJiEtuPbMfe43uxMbVx2fHVvESpVGrZHDmz12VEyCuH0PL5PAqFApLJpOZ+equKqEV6apFImE/ObWQIUKuepHlwExMTKJVKFfP0zKLmQdUT6mbuhXzOqLJ+5NdmVtD5PSpSrc07JcIJsQoFoMu4lWqBYs9dzKwCYGR/M/gh5uXGRb6yhxy9ci2Ki9h2aBsWsYiOpg5MRCaw7dA2nLziZDREGlSNl9yLZjbRs9nyyc+tHD7N5XJIp9Omhj/lgshM2Z0QtUbPJRep0rzcWCxm+Nxq6NWFHDP3Qm0Iu9q5pc9mhu29Rq/NA+ZFKV/yidNQALqMk2+hfr/RBgGvRLDdunbqXnl1vXLjYiVide/xvdiV3YXO5k4sLCwg1ZTCruyushdQzXgpBYpUDrfQK4Ne5LKekHDbKCuPb+ZcSpGqV89G2pmWt9RKXk2te2E0j6OVYXuv0WtvTotSOgeIFSgAXcbMA1/tIfb7jTYIeCWC7da13v5mOms/EgirGS69Mkvev7nFOaxOrEaxaWlob7Q4WvYCTk9NL9tfzYumhVsGTirDzp07sXv3bqRSKZx55pnLRFeQjarR4A69ejbyXCm9pfKUT1bzasrLJ0dPbFsdtg8CbrUlMwFGQa0b4j0UgC5j5oGv1gkH3RB5gVci2G5d6+0fRk+uXpkl71+6LY2W5ha0tLQAAFY3rC57AVvzrZr7G6lrq/OpzBi72dlZZLNZwwEaRjC6nJ+d8xltT3r1bOS5Um4jT/lkNq+m0ZQvalgdtq9lUqlUOd+itBKPFvVWN0QbCsAAQQ9fdYIkgq0a8Gr3WX5cuwmE3VyDGnjR+1daKCHaFMWi+OI6ptGmKOYW5rDt0Db8We+fqe5v5/x6hky67mPHjuHw4cNVPajypbec9CrpzXN0yhAbGcauhpHnSrmN2pxKo6hdu5E5fmreP70gnXrpSxOJBLq6ujA1NVV123qrG6INBWCAkA9xWO3IiXdYNeDVjK2Tef2cEhlaZc5MZ3AwfxCRSATP556v+G1+bh6lUglD2SFMnThVnrtlpCxqQThKT5FeVKt03ZOTk8jn8+V0I3rXV22I3Upd6s1zlAJu5Dn0rIhMqW7kK25I5XWzD7HzMlYtclgv56XS+2c232GtovayqFYXQXqJJv5CARhA6KJ3HrNGwUpyYKcwelw/yyjRHevGW095KxbEhWW/ZTIZHDt6DKtWrkJ3rBsHXzhouF3rPQPSb0ZWn2hubsb8/Dyi0SiA5UvRmVmSzkpd6s2/SySW59BzKldf0PsQNRFiZC5gNeEI1G//qVZ/9VoXxBgUgAHErKGp1zdeM5jtCJ2KMrTj0amGkQARt9/2mxqasH7FetXf2mfbEZ2KondFL5oamiradbW5cXrDfEaeD+m6U6kUVq5cWd5WqrN4PA5BEMpL0kn76GGnLrX2VV6LHcFuNJjCKaxE/xpFq77UPMGpVGVycQ5xvgjrguhBARhAtDo/s0uE+UU10eOHYDXbETrVcbp5b6RVKLTy9hnB7r3Q219KRCytXCBv1zt37tQUr9lsFocPHy577aTvrAyLaz1L8XgcnZ2dFR5AL9ullnhxSrB7McwnRZ4C1qJ/raBc93hoaAjDw8Po6uoqi1C1a3dTrAYZDvcSPSgAQ4SWmPDiLc+McawmevwQrGY7QrMiXAs37420oLyd5L5u5Ds0MkdPj1TK2JJgVtBagxVAxRw6L8STk0mwvUR+f81G/9pFmTR6eHgYIyMjyGazy0SomlgEvBOrQYCjQ0QPCsAQoWUEvXrbN2ocqxlrK8bcTDoNNzs9s4LJzXtjN0IYQHkYVFov1ix60bl6c/T0yp5ILF8yzQvPmJcvUpI49ioJtpPI769XuSkllJHH8XgcANDW1rbsvinFotdiNQgEbXSIBAsKwBChNf/FiwfbjHGsZqytGHMjQQHSb252ekGaU+OEKFIO0zpRBjNz9AB3IxXNPCdevkhJ4thIEuyg4eUzoDZ0K79H0rrIasO6SrHotVh1CzP5L+2s/0xqHwpAn7Eq5Lx+s/N7Lome0XFyIn01/K4Hp3GjrszWkZW2bCZpsF5qFC+9x8Dy+g5Le5LXi519rbz47d69GzMzMwCwTMRV8ySHoW7NYuR5MeKFJ4QC0GesCjmvPVF+zyXR68yVv9Vqx6+H2VUxgBfFkJqBcDM4RImVqPcdO3agWCxWnUNXLTWKl95j6ZhhbJtKIW20juT3ysj2SlKpFFasWIHjx4+rHrse57cZeV6CNFJBggsFoM9YfVC9NiR+ziWp147eDGa8AhJ6yXPdCA7RQi7GjGyfzWZRLBbLUcJml0+UP2teeo/DjFq9GKkj+b1Sbm/kuU4kEtiwYQNGRkbQ0dGx7Nj1OL/NSN+vFQnNfpTIoQD0mbB4BOwaRqOdj9p2Rjt6tzq4MHScVr0Ccm+gXBwq8/DpYXYpLjv3WO3YRufQqT1ret7jIN13L9YU1kOtnoyUUTkPT47ePZcfS2ueKsW6OepVMBNtKACJIewKVaOdj9p2Rjt6tzq4IHSc1Yy8Fa+A/P/KoVLlclt6qNWPXnnUtjcTjWxEjDhBEO67kbKYLadbgtFsO9B7rpURvGrbheXlOShQMBMlFIDEE4x2Pmrb6XX0ZleIMILSQAah43TbyFcbKtXDTP2orfIBGI9G9mNuoiAIrq3NbTRBsZkgKPmxvUwcb/Y5MSoOKfScgfVIlFAAkkBhtpOyukKE0WMGpcM0a1ztXIPZe2Bmey3volkvr9UEymbnJiYSiXKCaDeSNhtNUKxXx1q/eZ043kmBUS9iJUjTDEj9QQFILOPk6iBWz+1GniulgdQqu5edt1mDGASvpRp2h/Ok/awmULZSL1bOabRtpFLuJSh2Yug0CAIlCGVwiyC+bJL6gQKQWEbyXoyNjWFwcLCqoZP/deLcbuW5UhpIrbIHZe6VE8f2ysgq69bsutHSP6sJlK14lqyc02jbSCTcS1BsJqDF7nBxkFbfCRPV+kWn6rWWRTSxDgUgsUwqtXzNVi2cHtLx0sOlVXZlGap1sm4aMuXQaKFQQC6XM3wuJ8pmxchUO6/W70bbk5OGz0wbNjsv0u0IXyP1rPYyZ7SNu9m2g+rNdoJqbcqpeq1lEU2sQwFILJNILF+z1ctz+92RKT0s1RLemk2NYgblMGUymTQ1NO5E2awYmWrG3a7x98vwSecyktvQyQhfLYzUs9rLnPI5szuv0Eo7D8Kz7jZa9eKU+K1lEU2sQwFIbGF2SM8OQR7GUCa8Nbu+rZphNXO9U1NTGBsbQ2dnZ1n4ORXIYVSEWDEy1Yy7EeOvV09+Gj4n6s2p8lcTpEZf5uzOK6QnSh27nu5q1IOIJuahACSqBHGN4iAbD2XaCilyFDBWVjXDauZ6R0ZGytGkr3jFK8xfgMmyqeGGkTHSDvXqyU/D50S9OVl+I8FM1ebT2i2Pm17wIGImGEj+lxAvoAAkqlgVW252ZEHuJI0Gjhjd3+wxent7K/4awahx0jL6XhhsI+0wqO0iSF4XrfyLgLcvVnptSW0KRdhFoZlgoERiKdDISs7JsNcT8QcKQKKKVaPqptELkkGthhNlNXOMdDqNdDpd/mzXc2YELyJEjXqMnI4EVzuH28ExbkfSHj58uLx+spwgeOWUUyjk3wfV628Es/2o1ZRTfgVxkXBDAUhUMSM+2HEEDy88Z0b3dysptRcC1O7cTDPldNOIp1Kp8rSE9vb2itQzTtSxXZRTKNS+DyNmXwS1rrfafXCinsIutol5KACJbWqp47AjFoIkhI0YhGrGyYn1h42WxQpuC1CtYVOzxzNaTjeNeCKRQFdXF6ampkwdzysBptWWwuT1r4aR/kHreqvdByemaYRdbBPzUAAS29RSx2FHzDophNU6bjOduROG06nrccuIuy1As1l7y9aZLacT9aRWNqnddHR04NRTTy3/ZkeQyAnSi0+QMZM4X4nVOYJmnuFaEtvEGBSAxDZWOg4vkt9awY6YdVIIq3XcXntawyDsnRIxaseym/LEC7RWSpEjtRvlqjmjo6MYGhpCf3+/rdVI9Nql2We5lsVkKmU8cb4WbnmfSX1CAUhcxcrKAX4OKdsx7k4KA7WO2+vOPEhCRwsn24ryWLVy/W63G73jm70/tTSdREkisTzXolnBa+Ze1rKYJs5AAUhcxcrKAXxrVRdfXgiSMBkNvdQmVghjuzM711N+f3t6eiytp6w8jl671BuSVmtjYbwHZlDWlVnBa6YPqGUxTZyBApC4ipVhtDB4XmqVMBkNrTl6ZjAqZIKK2TLL729fX5/l6zWb387ovmG8B3ZwU/DWupgm9qEAJK5Sbx26Hfz2vjntUXMbpr4wj1OiQOs4RtowhcmLuNk/su8l1aAAJMQgbgs0pRjxWhA64VHzEreiZiX8FuROYzeyXM6yul+Yx+LEC3ju4BxGMkfQm57Fy9ccQ0PHWqCxSX9fQogvUAASYhC3vUVKMcKoX3fxYnWFIFEtslz6bFrwLsxj7OFvY3zHA3gi8nL8DpvwigM/R0L8HToHL8DqzVcsE4FkiVp7ySDhgk8lIQZxUyCpGQKnz+dUYudawczqClLdCYKAUqkUSoOtFVleKBTK/3K5HAATgvcP4u/o4z9CcaEBr2y+Hyd35LB64glMzy1g8fEfAQBFoAZWl34zug0hevCJJMQgagJJrRNeXBRxcKKI6dI82oQmrOmIoqEhontsNUOgJ8isdP615tGyi5nVFaRl1CKRCERRLP8eJrQiy2OxGDKZDJqbmxGJRCAIguFjLk68gPEdD6C40ICFjpMgzh3DuomHMCV0YaFtJYoTz2N8x4NYddqr0ZA6cdn+9S5itNqgkWeVzzOxCwUgITZQdsJPDx3GT556AWPTIhYiDWhtasS6rji2nNaN9au0O2mz3j4rnb+fQ7xBTAhsxuMp1ZncAxhm5PUrXYvkASyVSsu20fJOTWAVHm4+F+c2/wLi3DFMNq7A8cY2NDYKaJ87hkhzIx5uPhftYhfWqpTDi/Wcg4xWGzTyrNbblA3iPBSAhNhA3gnvO5LH//5mPw5PTKF3ZQLdnSkUSvPYfiiHQ7ki3rn5RE0RaHb41Urn7+cQr9kAFze8G3ZERK0Nj6ulg8nn8xV5AbXugfz7Y/Mt2DY/iLXxI9g49Vscb2zDzNw8WgHEF45iX8dmPN14Nl41t6BaDqPtOGjtwW2MtDc32mSQ64Q4DwUgITaQOuHFRRH3/n4/iouNOCWdRCwWQ2NDBInWZsRbmrD3yBTue+4wTloZ1xwOtrPWbxA7bjUvk9EAFze8G14OmQXxfshRq19lmxIEQXVIWL5vZiiDU6cfQffi45hq60Jjo4BWAI2CgKmFLqQnnsQZ8ZVoa1Zfas6oiJF7YM2shatHkIZQg9JeglQnxH0oAAlxgIMTRew/OoW1nXEkWptRKs0il5tANBqFILQgnWzFviNTODhRxNpUTPUYdjpfM/t6ZWz0kg6bmX/nFEZEpZE1qo0EggTdY2VkfmmhUIAoiuUhYbV9B1cMo6XpKczORDDbvBLtc8cQXziKqYUuTDevROP0JDbPPYI1kS0A2myXV5qLKX1nhyANoQZFeAWpToj7UAAS4gDTpXnMzC8gJkQBAMViEfn8FABAEFoQFRpxeHIGR45PYjF/1PFlsMzs65Wx0SuTGYHnlPAxck4ja1QbCQTx24NZUWexKDA5CnT0AZEIIIrAxDDQ3oN8obisbqXzLCwsYHJyEitXrtQ8T/sJpyD9ktcuRQFPPI9IcyP2dWxGeuJJNE5PItq4iPi6c/BCbgGp5rzt/JZO1muQhvWDIryCVCfEfSgACXGANqEJrU2NKJTmkWhtRjS6JASlv8XSAlqaGjFXyCMzPQ7A2WWwrAQ0uG1spDLl83lbw3Z2BasZwaFXN/JhyImJCRQKBeTzec9S6phJWl2us8V5JHK/A0a2AQNbgfUXAvt+Aey8G+g9D9nky5E5fLRcZvnxh4eHkcvlMD4+jvXr12uea/XmKwAA4zsexMPN5+LpxrNxRrwT58w8hKa1Z6FwwmbkDh8FGpoqywbz97NWBUqtXhcJNhSAhFTBSDTkmo441nXFsf1QDvGWJghCCwShBQAgiiIyuRlsWpPEQN8qTBwXTIsvu14wP9e8tSvg7ApWM+fXqxv5b6VSCZlMBrFYzLO61Cub8hpTqRSwOI9Vh38NDN8PNArA07cvef6GtwGIALt/hlV9M0D3q1TnAgqCgJaWFvT29uqeC41NWL35Cqw67dVoF7vwqrmFpTl/hzbhYB5INjQhnU4vu49+e7sIqXcoAAmpgpFoyEQigS2ndeNQroi9R6aQTrYiKjSiWFpAJjeDVJuA157ajWR7Asn2dsfK4MT+bs8JtGvw7QpWNwRH0ESMsjyJRAKJ+XHg2SeWxF/nOiCfwdyO/4eZ5iSEzj605EfRePB3aIxvBFauWnbMdDqNdDqteS5lQEZD6sSKVC954XQsqrQrI/MP7bbFoARVBAHWBdGCApCQKmgZPKXRXb8qgXduPhH3bj+M/UencHhyBi1Njdi0JonXnqqfB9BoGawKDr393Z4TqGbwvTRKbng8gzZkpxoVngO6TrwIsT13A/kMZltSONowh/nZRXSOD6NFaMTx1X+E0XwEC9ms4XtkNCDDSh051RaDElQRBFgXRAsKQEKqoGXw1Azc+lUJnHRB3PRKIEbL4AZ+eLNolNwlm80iMzYGrD4DfX3HgD33oLjQgvn5RTQ1NaB1Lgec/MeIbbwU6ePHLb0YBNmzGjQPbTWMRJjLBTlgfN3msNUF8Q4KQEIMYrQjbWiIaKZ68Qs9Y+5UsIZTwRZeYcULGZbhtFQqBYgiuiafXprzl0gj2rKUhiUaa0PzbDMw/AgSHX1IrL9wKTpYRj6fR6FQQDKZ1E3VAyy1LflnOzj1ohM0D201jESYy59hAI7MayX1DQUgIQYJc0dqRHC5Oc9QiZd1qSXarFxvWDyX5TmAz/wCQARIpNGSz6ClkAEa00AiDYzvX4oGXnkysKKvYv9sNotcLod0Oq17nU7MLQ2LqHYT+TQTraUG1Z5hevWIHSgACakDjAguN+cZOolZwaAlUpQrXRg5rplULL7T3gP0ngfs/tmS2AOADZcseQTH9wMLJaD3QqB9zbJdU6kUxsfHMTw8DEEQVINBpO3kf+UYFcthEdVuYuT5VG5Ta15r4j0UgIQYpNY7Ui0jZPS6vfLqjY6OYmhoCP39/RgYUF9iTI6WSCmVShUrXRgRImZSsVTD9fbU2ARsevPS/0ceAQbesJQHsKP3D3kAL1z6vXG5GUgkEpibm8PIyAimp6cRj8dNB3oYfSEIwnQAv3G6LciPR4FNtKAAJMQg9dqR2rluI4bNbSGkJVK08tJ55QH1pD1JIrDvvCXhF4kg3/0K5GbbkFw7gISK+JPo7e1FNptFQ0NDeZ6fmftk9IUgzFMrnMLJtpDP5/HEE0/g+PHj2LhxI3p6egDUt8Am6lAAEmKQevVUmL1us94Hs8avp6cHsVjM9n1QG1KzY3zN7u9Ze2psqpjjlz1+HJmpBixMTCKRXFH+XinE0+k04vF44DxJteiJt/OMKdP/7NixAwcPHsTMzAwKhYKhIK9arFNSHQpAQgxSr54Ks9ctFwpGDJtZ41cr98HJ63AiAltN4KmV0e8XoKAIUSex84wpA5uKxSJisVj5X7V9qv1GahcKQEIMwrfkJarVQxgEXa3dS8mAFwqFsnfU7DzGavfNjbQvVqhXT7wcrTrQiybWqzfWaX1CAUiIQcwY2VqmWj3IBUa11SL8IgweDytevUKhYPq65Ofp6+vT3TYI9VYrHmArVLtXenVj9TdSu1AAEmIQO0a2ljBTD0H1LAS1XHKs5FXM5/OIxWLlZQv1Vpawch63663WPLNOEwQBTmoHCkBCDCA3TKlUypEghLCiFBt69RBUz0JQyyXHitiSrkvyvOqtLGHlPPJ6c0OsUeDoE4YXFxIeKAAJMYDcMPX19dE4IfgiKuzeJDv1a2RlCbvncUOsUeDoYySilxCjUAASYgAtwySKIo4UjmBVbBUiivVUib8ExZukl7LDLYHqhTh3KnWJHN1yL8wDk6NAR9/S2sWiCEwML614opPPsBZxum2H/WWJWKO+nhpCLKJlmPYc34O79t2Freu3YmNqow8lI1oExZukl7IjCALVKmrPhJ6QsHW9C/PAs/8HjGwDBrYurWiy7xd/WNHkPM0VTWoVedt2QryFvS0Sa9TPE0OIRbQ62EVxEdsObcOe43uw7dA2nLziZDREGnwsKZETlCHqaik7/BaoTqInJCxfryT+dv8MaBSAp29f8vwNbwMQWfoeqBCBtezRUl6bE5H2tdgWSXUoAAnRQOpoC4UCcrkcgMoOdu/xvdiV3YW1ibXYld2Fvcf3htYLWMsGU4ndazW7v5YQtStQvbhnZs+hJyQsX+/k6JLnr1EAOtcB+Qyw5x4gkV76N75/aa3jvvPKK56E1aNlpL6V1+aEeAvKyxLxFgpAQjSQOtpkMol0Ol3RwUrev+JsEcnFJIrzxVB7Ab02mH4KTrvXGhRx4UU55OeQPluew2eVjj5g4A1Lnr98BrMtKRQXWhBtaUNL/g9lG3jD0lrHfyCsHi0j91R5bRRvxCoUgCRUeCkc5B2t8lyS9y8ZSWJqagrtre2h9gJ6bTD9FFF2rzUo4sKLcqRSKRQKhfI/NU+4GSw9v5EIsP4iYGIE2HMPigstmJoqAABaChlgwyVLcwJlQVhhFUVG7mlYr40EDwpAEiq8FA5aHa3k/ZtbnMPqxGoUm4qIRqMYLY6G1gtox6hYMep+iii7BtQNA2ylDr0QAolEArFYTNMTbrbclp5fUVwK+BjeBiTSiLa0AQCisTagMQ0MP7LkJVSIwDBCcUe8hAKQhIogeF8k71+6LY2W5ha0tLQAAFY3rA61F9AqVox6EA1dmIel9bCbqkjPE2623Jae34nhpWhfRIBEGi35zJLnr1E2B3Dn3cDKk8tzAAkh1aEAJKHCb+Egef9KCyVEm6JYFBfLv0WbophbmAutF9AqQRDlThDmYWk97KYq0nvmzJbb0vPb3rOU6mX3z5bEHrA07Du8benzQgnovRBoX2PuuITUORSAhJggM53BwfxBRCIRPJ97fvkGEeBg/iAy0xmsiYfPIEleMCNryMq3l2/nhCfND29cmIeltXA7VZGy3K7ct8ampRQvwFK078AbloZ7O3r/kAfwwprIA1hPkfgkGIT7iSHEAE52rN2xbrz1lLdiQVzQ3KYx0ojuWLet8/iF5AUzsoasfHv5ZyeCBbzwxinbhZ4Ic9M4u3lst1IVaZXZtfsmicC+85aEnxQYsnLDkucv5OIP4OoexHvC/9SQusFqh2a0YzVy/KaGJqxfsd5cwUOEmTVk5dunUindtDl6qNW7F944MwY3m81iaGgIY2NjGBwcdNSgjo6OYmhoCP39/RgYGHDsuE6lKlK7P1p15+p9a2yqnOMXidTUnD+n6y4o6YpIcKEAJKHBaodmtGNlh2l+KFJteycEultDonIxY8bgplIpjI2NoVgsIpvNhqJ9OJWqSO3+aNWd2eXhyIvIxbX8sxKj9Vkrc3OJe1AAktCg16HpdYpGxQQ7THtYFW1u1bvafEa5mOnr6zMVtTw4OFhuY07S09ODWCy27Lh2hJOTqYrU7o+Ze+3Vi1UtCE0jibeNJuf2O2COBB8KQBIa9LwLTsw7Y4fpHk4IdLOozWe0IjblZe/rMzbkaEaMaF2/UjiZOaaTqYrs3h+vXqxqwYOvNqUC0B5mN3vNtSCSiXNQAJJQY3XeWT2SyWQwMjKC3t5epNNp188nNzZ+GGe1+YxWxIxW2fWMqZ3rlXsuk8kkCoVC+Tsjx/QjVZEfAl9J2D34WnVoZJjd6DXXgkgmzkEBSEKNXpJaJfX+9jsyMoKhoSEA8EQAyo2NH8bZqPCo1i60yq5nTO1cr3TcdDpdXoVDPkSsd8x8Po+dB3diKDvkaaqiIAiLIHjwzfYxei9JRq5H+n10dBTA0nQCzgskRqEAJKEmiHORgkpvb2/FXyepFsnrtHF2UsxXaxdaue7m5uYQiUQgCELVfcygZqSN1mE2m8Xc8TlctPIidK/WTkXkdKoiCoslzPYxdl+S8vk8duzYgUwmU26H0suCkbZM6hsKQFI31LuRSqfTrnn+zEby2k0g7aSYN9sulHMLS6WSrfMrUdab1Yh3Lw09hcUSZtuS3ZekbDaLYrGIVCqFrq4uAKjrl1xiDgpAUjc4YaTqfRhZCy3DZyZhsBlRZzUiXA2z7cJsrkQvoRDzFyfSKJlBKSDz+bxqRDkhalAAEmKCeh9G1kIyZPl8HsPDw2WDZCZhsBnviZ7hdPseUWQRPbx8SVTzFrNtEqNQABJiglocRnZzPp2ZhMFOGa+w3aOgeZWDVh45QS6bhJurxri19japTygACTFBLb5huzmfTqu+3DRaYbtHQfMqO1Eet+5v0OpKDTdXjbE7dYIQORSAhNQ5TnrMjIgvKXKxWCyW93GaMHlFvPBYmqkPq+VxM++jPDdiGPJ9tre3o729HalUypG2qHf9YfN4k+BAAUgIwiUYnMapfHlGkSIXo9GoJaNVrRxeCEwnkde/0TrW2s5M0I3W9lY9qG7mfZTnRjS6GotfZLNZ5HI5pNNpJBIJDA8PI5PJoFAo6KZoqXZMresPm8ebBAcKQFLXOLmUXK3jlFfHbqqSauWwKzD9xGgda21nJujGzPmM4GbexzB5uZRllf4WCgXLdR2m6yfhgQKQ1DXKpeQEQaiIYiUv4pQRcir1hda98isXnlH0vHxG61hrOzNBN2bOZwQ3PVFh8nJpRebaSdHitZee1AcUgKSuUYoFabgGoCdQSVCMsFQOtXuVz+fLy2IF1Vui53UzWsda23mdh84P7Iocv0SSF3XNgBBiBgpAUtcoO+V6G2rRM4ZB9yao3SspBQewtCRWWMrtNEG/d3awK3L8FElW7osXATykPqEAJERGGD0idtAzhk7mM3NDkKjdq1Qqhf7+/vL/g4hbbczNKNwgYVfk+CmSrNwXM/vUW/9F7EEBSEgdo2cMncxn5pUgSSQSGBgYcO34QcbNKNwgYVfk+CmSrNyXWr6XxF8oAAmxSC0Ms8mTyco/S/8fHBwsX6MdaMTcx80o3DASxOfTyn3hvSRuQQFIiEWUXi0jBieIRsmJoIRqeG3EgljPbkOhUEktD4MT4gQUgIRYROnVGh0dxdDQEPr7+9HT02MqIa+fyK+jVoRTEOvZTZxcbcLre+/Wed32Ooc5gIoQgAKQ1Bhedrx6HhezCXmdwOq1y6/DSBocs6tQOAWjIbVxQvD6JZrdOq/bHlGp3GorfNTbCwgJJxSAJLSoCQI/O96enp5liV6NJuR1AivXrqxDNeGk3MboKhROC0K3oyHD7LVxQvDaPYbV+gurWNdb4SOs1+QGYX6uah0KQBJa1ASBnx2v2goAXmLl2pV1qCaclNvorUJRKBRQKBTKCZmlIXGjkblOrJJhlTB7bZx4sbB7DKv1F8S5i0ZEi1RutRU+nLqmWhBPYX6uah0KQBJa1ARBEI2JV1i5diOiSrmN1nkSiQRisRgymQxisVjFb0YNmRcBKVrQa2MPs/XntLhRO578OwCGzxeU3Hu1IJ74XAUXCkASSMy8gRPrGKlDM/Us7+xTqVTZMyI3ZIC2IfZzlQy2J3sYqT83k1WrHU/Z7oyez+l2WG/D43L4XAUXCkASSJwyDtW8AnrHroXhF6+pNgxezfDLDffU1BRKpZLj9V8LXpWw4mayarXj6X2nh9Oixa3h8aD1UUErD9GHApAEEkEQEIlEIAiCreOoeZ4KhQJyuRwA/c6YQkEdK528miHTMsRSvUciEYiiWN7fKWrBqxJW3ExWLX95kD77NS/XSHCVEwStjwpaeYg+FIDEFF6lACmVShBFEaVSydZx5EJS6pySySTS6XTVztgroRC2t2a7nXw1wy/VtyAIZQ+gk3BIyhmcehFwkqAIECPBVU4QtJeZoJWH6EMBSExhNAWIXYykIzGCXEgqvQ/VUHbadoaT9QiK0TKK2518kASaU+LcyHFEUcSRwhGsiq1CJBJx/PjSNnJh7UfUr5sERYD4OZfVT4L07JLqUAASQ8iNh5r3zOkOz0g6EiNoDTlZ6TyNTDKvh4ne9dTJOyVyjBxnz/E9uGvfXdi6fis2pjaaaqN6SYmV2zg1tB7EdhuUtulFOYIowN0kiII37FAAEkNInU06nUZfX9+y373o8KwYHK1yGTGYRs4v/86pid7Kjq4eOr6gXqNTIqfacRbFRWw7tA17ju/BtkPbcPKKk021J+m4akmJlds4NbQeFLFVrwRRgBvB6rNeb4LXCygAiSGsdDZOG3UnDY4Rg2nk/GaCG4yQz+exY8cOFIvF8vHroeMze41eCUan2ly14+w9vhe7sruwNrEWu7K7sPf4XpyQOgGAuahVtaTERsvgBUEV+mEkCPdTjtG1ka32Z1o2iG3KOhSAxBBWOhu9B93vh1ZpMAVBwPDwsK3yGB1i1vstm82iWCwiGo0u6/D8etP34l6ZvcZaEcX5fB7Hxo/h/vH7UZwtIrmYRHG+iG2HtuEdp77DdKCV3WkOblMr9y1seNEW9O6tE+l/qo3mqJ2X6EMBSFxD70EPykMrdSrDw8OuJ6U18ptaoIrfb/pO3iunDJHfotgpstksHjvwGLYXt2NF8wpMTU2hvbW97AXcmNpYsa2Z+xCUZ0yOk/ctiALXLexeq1pbcKL+5HPDC4UCksmk6r11I/1PtXnppDoUgMQ19B50rw14tc7O6SFuvePp/ea32FO7JvncMclLCugHvGjNY9TKwWhWrNippyAJh44VHXhh5AU0NDUglUih2LTk/R0tjpbnAjZEGgCYb6NBFMlW75vaPQuiwHULu9eq1hacqD9lYFE6ndYUmE7fo2rz0kl1KACJL3gtdKpF6zo9xK13PL9Fnh5q16TmJQX0l9VS1rc0r7G7u9uTKHI9giQcxubHcFg8jLUda9HS3IKWlhYAwOqG1cu8gGbbTZDbmVnU7lkQBa5b2L1WvbnKdupPL7DI7eesnu6/W1AAkppELxO/lbmJep4xNzogv7xURjyXgiBgYmJCc7hHeRz5vMaenh7V6/FSrATFcEiRv6WFEqJNUSyKi+Xfok1RzC3MLfMC1gNGn7VaErjVcONanTimfC61tAKLhNvPWT3df7egACSBx4mcfXpvwHr76X3vZgfktZdKXsdawylyT2Aul6sY7tHaVk4QhlyB4BiOzHQGB/MHEYlE8Hzu+eUbRICD+YPITGewJr7G8HGDNMRthWw2i6GhIYyNjWFwcFDz+SXBQdlfydsgANsBdmqEvZ0HAQpAEnjsJoBWYmVuotr3bnZAdt6enUpybbdsRkRlPdMd68ZbT3krFsQFzW0aI43ojnWbOm6QhritkEqlMDY2hmKxiGw2G8prqDeUfYJyCogb7THs7TwIUACSwGNFDFn1GGjtp/a9m9GxdjweZsplJZLOaNnYQevT1NCE9SvWO37coAxx26G9vR3t7e2mXjJqrY0ZvbYg1IF8KHh4eLjcn8zNzeHQoUOIx+OOv8yqBafVWhtwGwpAYhu3O6BqgsPMOqhOlNFu+gG3IxrNCAA3I+lqQYiEkbAPl2az2apTDOTbGnlugiCSzGL02rS28+OapeH7hoYGdHV14ejRozh06BD6+/stlcFIoJ3TKbzqCQpAYhu/PT1GOkAny2hXNLkV0agccpXexvUMgJsiLexChPiDmTYp37ZagvVqz//iooiDE0VMl+bRJjRhTUcUDQ0RO5diC6P1oLWdH/2yNHyfyWQwNTWF1atXo7+/H729vZaPJ/9rdRuiDgUgsY3fD6AgCIhEIhAEoeJ7K9nnjbw1271etyIalR2+EQMQdJHmtXeXuIfR+2SmTcq31fMEVXtm9x3J497th7H/6BRm5hfQ2tSIdV1xbDmtG+tX+TusanU7P/rlRCKBwcFBtLe3A4Bm1L+Z41XbP+h9WJChACS2cfsBrGY4SqUSRFFEqVSq+F7eATo5b83u9ZrZ34y4UXb4fgtzJzByP/z2QBNjWJmbakbUWw382nckj5sfPoDsdAnpZCtiQhSF0jy2H8rhUK6Id24+0TcRaBa3ky8bIZFIYGBgwPPzEvNQABJfMdLRV8vbp7UEkZUOMGiiyYzRVF5vLbwZOz0EFBRvYVDK4SVW5qYCxkW9fD6t/LNeXS8uirh3+2Fkp0s4eVUckcjSkG+itRnxlibsPTKF+547jJNWxn0dDjYKX4aIGSgAiSdodcJGOiw9w6GcMG7XsAZNNBkxmsqcW24LC7N1rLUsnJH9nR4CCoqBdNsbFkSkso+OjgLQHx60+iKmVq+jo6MYGhpCf3//Ms/UwYki9h+dQjrZikgkglJptpy0XBBakE62Yt+RKRycKGJtKmaqLH5gpd7kQW3Sah5hbmfEOBSAxBO0DJ6RDkvPwOvln3JjmNVrjIgbL3JuaZ3PiHiRloGTtvdThAXFw+u2NyxIKIOxhoaGAACxWEx3mgWw3JtXDa16LZVKOHr06DLROV2ax8z8AmJCFABQLBaRz08BAAShBVGhEYcnZzBdmjd6ub5iJWOCcj1f6Thm4Xzd8EEBSDxBq2O22tHL9zcqKLU8UYVCAblcztL5g4DaNbslcLSG3PU8vJJHxcrcRKcNRlA8vGbKERTRahVlMFZ/f3/5/8DSPVbzClodBlZu29PTg8nJSdXE0m1CE1qbGlEozSPR2oxodEkISn+LpQW0NDWiTagNU6mXgUBtPV8z6Hla9c5P/KM2WjUJPHoGz0inYHSY08x5pM/JZNJSPr+goDb3zy20crQZ8fBK36vdI61hqHocKlUSFNEqx2pwklqAgJZX0CnhK0WmyvsPiTUdUazrimP7oRziLU0QhBYIQgsAQBRFZHIz2LQmiTUdUVtlCApuZSCwc37iHxSAxHeMdApODHPqRckGzcAGEb2AGz0Pr5noa+UwlJtDpUEVjEEtlxw7wUlK1LyCRvYzg9axGhoi2HJaNw7lith7ZGkuYFRoRLG0gExuBqk2Aa89tTsUASB6eLEsY09PD2KxmOXpPMR7KACJ7xjpFJwY5gxrlGxQBEE2m8Xhw4fLw2Ny7NalchhKvryTUYNl1rsQ1OGooJZLjpOeHDWvoNvIn6n1qxJ45+YTy3kAD0/OoKWpEZvWJPHaU/3LA+gkXrSpsPSn5EUoAEko8HKYM2gERRCkUktZ/tXmUtlFeX+tLO9k1gAFdTjKjXK5NZfSyGozfpVRD+UztX5VAiddEA/USiBmqbZubqFQQKFQQD6fr/jdjSwCQXlpJfpQABJHsJsahGgTFKGiN5fKDmptwYtrDqrHwo1yufUS4eQ6tF6+6Ki1r4aGSChSvWhRbd3cWCyGTCazLPrajSwCQXlpJfpQABJHqPbAKw2CnQ5Cy7jUqqgMklCRl8Wp+lZrC0G6Zr9wsj27Jai1jmvl+fbyRacW21e1+tP63Y0sAkF5aSX6UAASR6j2wCsNgp0OQnmsWknnEjaMGvlqQkZveCoI+PVi4aQXxS3Bo3VcK893LYoyL6lWf1q/OzG9Rm0JOt7L4EMBSByh2pwgpUGw2kGoRaLWSjqXsKFm5PUSzUoof9cbnjKC2wLNrHfbCEb2Cbow1oMCoL7gkG84oQAkjqLVESgNglWjrZaHjulc/EFtOFjphVUKdql9FAqFcsoIpz3CTmPWu20EI/vYFcZ+EuTpGGpl86O8ftdRGKYYEHehACSOYrQjsGq0nUpk6nfnG3a05nQqvbBaiaMLhYJjos2q8dFrA2pDWk6eX75PtehNs8cOAkH2CKmVzYvyOjkP2gnCMMWAuAsFIHEUox2BIAiIRCIQBMGV41fD7843yBgRx3pzOo0M/efz+YqksXbuh9U2oXdOs8O+dsqsl/ImrIY1qMJVK5m53fJqrWQjx8l50E7g9/mJ/1AAEttY8aaVSiWIoohSqeRy6dRh56eNnviRGzoz8y2npqZ0J4m7eT+02qfeOd0Y9tWiFttiUIWr3CMNoGK+spHy6q15rbaSjRyn5kE7hRPn50hKuKEAJMsw+1AHPeWDGn53vkFG795oLfiuFehhxDBK37l1P6zOS3V62FcLt9sijfSLyO+bk3M3pePKPYBKarHP4UhKuKEAJMsw+1C7nfKBBsxb1II7qtW9lmGVhvo7OzvR3NxsSTDZvf9uzEuVzx0zsr2fBMVIB+E5Vut3rM7drHbcWsOvhO3EPSgAyTLMPtRud35BMWD1iLLutRZ81zKs2WwWoiiiubnZ8iL0du+/0fZptt0HuV3KjXVQjLSX9WVEbFp50akHoacFE7bXHhSAZBlBe6iDYsDqESvzlux6WaqVwShmPU5m272dco2OjgIAenp6XHnW5Ma6r68vEM+zl8+xWbHpVFLzWob9cO1BAVgD1HqnFDRBWk/YrXsnhkrdiPJ1AjvlGhoaAgDX8vt5aayD6D0ze/1up6+qBby4f7Vuy4IGBWANUM+dEgk+frXPIHkslEOy/f39AOynHdEylF6KrWorvcjxysCbvX4npglQvNiHtsxbKABrgCAZujDCjttd/GqfekbdjXuud0zlkKw8gtoKQTKUZiJr/S633fuu16a0IuTrDTt1TFvmLRSAIYfixT5+G6VaJ4hD+G7cc71jOm3Y3DaU1fqVailztMqlLLeRBMpOYva+y69T2p99rT5+JHUn1qAADDkUL/bhW2cw8PJlxo17rndMpw2b0eMZqtOFeWByFOjoQ35qCtnxccwd3Y9jc63lcylR63fMRt4CL3rN2tra0NLSUnF8N9qBnUhvALp9rVaEfL3B/jQ8UACGHK8iJGsZvnUGAy9fZty450FsR8o6XfbcL8wDz/4fMLINGNiKbPN6FJ/9Mbqzv0NHz8vR0jGoely1fkc6V6FQKAsho/XR3NxcXhrSiXag1b85Eemt1deaOXYt97+sh/BAARhyghohGXbYMXkPPQfOI1+hYnh4GIVCAblcDgCQiEWXxN/unwGNAvD07ejuPhML2YchtLSg+djvgaGVwKY3A42VpkJv2LdQKBjuWySvmVQu+SoadtqBU/2b8jq9mC5QT7Ae/IUCsE6pNWPrtGBjx+Q9QfSgBREzbV2q0+HhYWQyGSSTyRfXcJ4cXfL8NQpA5zogn0HryAPAijSQSAPj+4GRR4C+84AV1ZN4S+fK5/OGh0LV9nGiHRjt3/x60UulUigUCigUCsjn83Xb7mvNDoUNCsA6pRaMrbzzdjqRKzsmElTsrr1d3keMAwNvAJ6+HchnMNuSQnGhBdGWNrTk/zDvbeANQEevqfJZ6Vu8miOpfP6tzmW0g3R8AMjlcq7lggwDtWCHwgwFIAkt8s7bTiJXtQ6fHRMJKlZeTlTbcyQCrL8ImBgB9tyD4kILpqYKAICWQgbYcAmw/kIgEqmZKRHK519vLqO0jVtlqPDG1hFm2lKttLugQgFINAn6w6fm1ai24oQgCOXJ5hIc7iVhwrGXE1EE9v0CGN4GJNKItrQBAKKxNqAxDQw/gkJLF462n45Csfji3MEQPyNKwac3l9EtYabqja0jzPS37JvdhQKQaCItWTU2NobBwcHAPYDKzttIZ1EqlSCKIkqlUvk7DveSIOHZi9fEMLDzbgARIJFGSz6z5PlrfHEO4ML2HyG7ehGt6Y2h9lbJ67Svr0/zN6lPcbPe62F0Qa8Nm+lv2Te7CwUg0SSVSmFsbAzFYhHZbDbwnZaRzkJtm3rokEl48Mzr0d4D9J63FAU8vn/puw2XLHkEx/cDCyU0nfRqpLo3IbVylemyqIkAo985jV6dWq1vs+X2c0TF63Pr1amZ/pZ9s7tQANYAbj3ciUQCg4ODFZnwg4yRzoIdCgk6nnk9GpuWUrwAS9G+A29YmvPX0bvkGey9ENFNb0Zfo7aZMLr8nXyKhpHvnEavTq1G5Jott5XrdKpv93ooVa++gz61qJ6gAKwBtB5uuw+achmk4eHhUD20+Xweo6OjAJbyjYWl3KR+8D0ASRKBfectCT8pMGTlBqB9zbL8f0rMLn9n9Dun0avTRCKBWCyGTCZjKiLXbLmtXKde326mb/N6KFWvvjmvLzhQANYAaoENgP0HTZ7Zf3JyEsVi0fKx/ECawwgAsVis/F2YRCypbQJhDBubkG9KITsy8uKzYSDvH2B++Tuj33mNEYFkd66glevUKpeybwvTyAfn9QUHCsAaQC2wAbD/oMkz+xeLRUSjUUMu/aC4+FOpFPr7+8v/D4SxJURGUIyh1WcjSMLCDvLr0Oq//AiK06pfZd8WFpSjSjt37gTAERq/oACsAbSMiN3OWdpfmaVfidJ4BEVoJRIJDAwMLPs+TB0mqW3MPqNuvVwFRYgGAa3+K0hBcVp9W9CR1y0AU15M4jwUgDWA32kLlMbDCWNixdBV26dWvBWkfnHr5cqNZ8PIMxzESFq9F2q9oDg3rrfWUNZtGL2YtQQFILGN0ng4YUysGLqgeB4JcQsvPXV2xYqR59GLSNpqmJnbJ49cVpbBjeutNZR1G0YvZi1BAUgCiRVDx2EsUus45akzko/PrlixmpfT7jHNClenRKgb10uIm1AAkkBixdBxiJfUG1a9dEby8dkVK0aeRzciac0KOqdEqBvXa5R6H1om1qAADBFePeTsTAgJB1a9dEby8TkpVrzsU8wKOqdFqB/9p7IdsA8nRqAADBFezR9x+zxqSUzZYRFiHqteOq/z8Xk5983vkQA/5vkp20G1MrC/JQAFYKhwe/6I1CkIguDqwu9qSUzrfXI0IVYIojdKjXqa++bHtSrbQbUyGO1vg9J+iDtQAIYIvc7eiQdV6hTS6TT6+oytBGAFQRDQ1taGeDxuKHUMOyFCrBGUFyu/vXJeEoRrNZu6S4ugtB/iDhSANYLeg2pUQHn15loqldDS0oKVK1eWy6PXYZnthJwSjBSeJOzUk+dNotpzq/e7k+unBzkXoF46Gzn12H7qCQrAGkHvQTUqoLwaTnIy9YNamZwSjHz7JWHHaW9UEMRLNao9t3q/O7V+ut7+QelXjJQjCN5M4h4UgDWC3oOqFFBOpo5wuqxmt1crk1mBaSavVxgMICFuoVzKy+izEKQoYL3f7Xq8nMoF6OeKJ6R+oACsA5QCysnUEX6jViazAtNMXq+gvL0T4hZ64kP+rJh5Fpx8buwu+aj3u12Pl1O5AL3oZ+jdIxSAdYiTqSP8xokymTlGEEUwIU6iJz7UnhUjz4K0jSAIGB4ediRYTa18YaGaiHVq1Map8pDahAKwDgmikAsLrDtS6xh9yTHzLEjbDg8P2xZvtfASVk3EOjVq41R5SG1CAUgIIaSMmy85WvNqlYnh/SqfV7ixBrLd8hQKBRQKBeTz+dDXLzEGBSCpezj8QYg3yMWb9NwVCoVlieFrHbM5Xd0WvYlEArFYDJlMpm7uAaEAJITDH4TYQOsFqtqL1ejoKIaGhrB69Wr09/cDCPewrlO41R+p3Q/5d7UwtE7MQQFI6h52fIRYR0uwGBUy0WgUAwMD5c9uJmoOA271R2r3Q/5dX19fzdYpUYcCkNQ9RodX6sH4EGIWLcFSTcj09PQgFost+91qouZaeT7dGu5Vux98+a1vKAAJMQiHiglZjpZgsZqPz2qiZj6f+qjVdy0E1BDrUAASYhC+LRPiPlYTNfP5dJda8bCSF6EAJIQQEnpq2ZsVBPFFD2vtQQFIag63Okt2gIQQIzjdBwWh76GHtfagACQ1h5HO0koHzQ6QEGIEpwVbEPqeWvaw1isUgKTmMNJZmumg5WKxr6/PuYISUocEYTjTbYwKNqN1QfFF3IACkNQcRjpLM2/UQRh+IaRWqIfnyahgq4e6IMGFApDUJVodtNobeRCGXwipFfg8vQjrgvgJBSAhMtTeyDn8Qohz8Hl6EdYF8RMKQEJk8I2cEEJIPUABSAgY6EFIUPj/27ljHdVhKIqiF/REFyFRIiH+/7PSUFLgjoJMRYsyk8D4zVmrduEm0lZ87YRLItADAQhlGBt64VuEzxCAUI5+oRe+RfgMAQhlGBt68Z0b+sDPCUAAuudoGNYlAAHonqNhWJcABKB7xjRgXdvf3gAAAJ8lAAEAwghAAIAwAhCAbrTWahzHaq399lbgTxOAAHTj+dzL9Xp9uU4owjJuAQPQjbnPvXgXEJYRgAB0Y+5zL94FhGUEIAD/He8CwjJmAAHollk/eA9/AAHollk/eA8BCEC3njN+u92uxnGsw+EgBGEFjoAB6NYwDHU+n+t+v896HgaYxx9AALrn1i+sSwAC0D23fmFdjoABAMIIQACAMAIQACCMAAQACCMAAQDCCEAAgDACEAAgjAAEAAgjAAEAwghAAIAwAhAAIIwABAAIIwABAMIIQACAMAIQACCMAAQACCMAAQDCCEAAgDACEAAgjAAEAAgjAAEAwghAAIAwAhAAIIwABAAI82/Oommaqqrqdru9dTMAAPzMs9Oe3fbKrABsrVVV1el0WrAtAADerbVW+/3+5ZrNNCMTH49HXS6XGoahNpvNahsEAGAd0zRVa62Ox2Ntt6+n/GYFIAAAf4dLIAAAYQQgAEAYAQgAEEYAAgCEEYAAAGEEIABAGAEIABDmC46izdRrc4j2AAAAAElFTkSuQmCC", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "fig, ax = plt.subplots(figsize=(8,8))\n", "ax.scatter(*clusterable_vectors[layers[3].cluster_labels == 6].T, s=1, c=\"gray\", alpha=0.25)\n", "ax.scatter(*clusterable_vectors[fl_indices[6]].T, marker=\"o\", label=\"Facility Location\", alpha=0.5)\n", "ax.scatter(*clusterable_vectors[sc_indices[6]].T, marker=\"X\", label=\"Saturated Coverage\", alpha=0.5)\n", "ax.scatter(*clusterable_vectors[d_indices[6]].T, marker=\"^\", label=\"Diverse Central Selection\", alpha=0.5)\n", "ax.axis('equal')\n", "ax.legend()\n", "ax.set(xticks=[], yticks=[])" ] }, { "cell_type": "code", "execution_count": 28, "id": "105c2ba1-7b34-4eb4-aa4b-1d6249bb7b22", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "[[], []]" ] }, "execution_count": 28, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAoAAAAJ8CAYAAABunRBBAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjAsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvlHJYcgAAAAlwSFlzAAAPYQAAD2EBqD+naQAAlCFJREFUeJzt3X18HHWBP/DP5mGT3WSz6fYh3TZNCIS2CRRsAQUqwglY5A7rnYIoz3KAeqfiIT78zqc7VE5PEPH8cXr+rJ6CqAcocspzQaGI8kyfW0gT0m4a2m02m+ymm2S/vz/CLpvN7O7M7Dx8Z+bzfr14lSS7M9+Z+c58P/Odme/4hBACREREROQZNXYXgIiIiIisxQBIRERE5DEMgEREREQewwBIRERE5DEMgEREREQewwBIRERE5DEMgEREREQewwBIRERE5DF1aj6UzWaxb98+hEIh+Hw+s8tERERERBoJIZBMJrFkyRLU1JTv41MVAPft24dly5YZUjgiIiIiMs9rr72G9vb2sp9RFQBDoVB+gi0tLdWXjIiIiIgMNTo6imXLluVzWzmqAmDusm9LSwsDIBEREZHE1Nyux4dAiIiIiDyGAZCIiIjIYxgAiYiIiDxG1T2AREREhaanpzE5OWl3MYg8pb6+HrW1tYZMiwGQiIhUE0JgaGgIIyMjdheFyJNaW1uxePHiqsdlZgAkIiLVcuFv0aJFCAaDfDkAkUWEEEilUhgeHgYARKPRqqbHAEhERKpMT0/nw9/8+fPtLg6R5wQCAQDA8PAwFi1aVNXlYD4EQkREquTu+QsGgzaXhMi7cvtftffgMgASEZEmvOxLZB+j9j8GQCIiIp3OOOMMXHvttfmfjzjiCNxyyy35n30+H379619bXi47PPbYY/D5fHxAyCEYAImIyNUuv/xy+Hy+Of/t3r276mnffffduOGGG0r+PRaL4d3vfjcAYM+ePfD5fHjhhReqnm9x0LRacfAFgFNPPRWxWAzhcNieQpEmfAiEiIhc75xzzsGGDRtm/W7hwoVVTzcSiZT9++LFi6ueh1P4/X5PLa/TsQeQiIgslc0KvBZPYfvQKF6Lp5DNCtPn2dDQgMWLF8/6r7a2FjfffDNWrVqFpqYmLFu2DB/72McwNjY267tPPvkkTj/9dASDQcybNw/r1q3DoUOHACj3hBUqvATc1dUFAFi9ejV8Ph/OOOMM/OEPf0B9fT2GhoZmfe+6667DO97xDt3Le9ttt+Goo46C3+/HihUr8NOf/nTW30dGRnD11Vejra0NjY2NOPbYY3HfffcBAA4ePIgPfvCDaG9vRzAYxKpVq/Dzn/88/93LL78cjz/+OL7zne/ke1P37NmjeAn4rrvuwjHHHIOGhgYcccQRuOmmm2aV44gjjsDXv/51fPjDH0YoFEJHRwd+8IMf6F5uUo8BkIiILLN7OInbHnsF335oJ259ZBe+/dBO3PbYK9g9nLSlPDU1Nbj11luxefNm/OQnP8Gjjz6Kz3zmM/m/v/DCCzjzzDNxzDHH4KmnnsITTzyB8847D9PT05rn9ec//xkA8PDDDyMWi+Huu+/GO97xDhx55JGzAtrU1BR+9rOf4YorrtC1TPfccw8++clP4rrrrsPmzZtxzTXX4IorrsDGjRsBANlsFu9+97uxadMm/OxnP8PWrVvxb//2b/khRSYmJnDCCSfgvvvuw+bNm3H11VfjkksuwdNPPw0A+M53voNTTjkFV111FWKxGGKxGJYtWzanHM8++ywuuOACXHjhhXj55Zfxla98BV/84hfx4x//eNbnbrrpJpx44ol4/vnn8bGPfQwf/ehHsX37dl3LThoIFRKJhAAgEomEmo8TEZELpdNpsXXrVpFOp3V9f9f+UfF/7n5JfOSnz4h/uXez+Pf7t4t/uXez+MhPnxH/5+6XxK79owaXeMZll10mamtrRVNTU/6/97///Yqf/eUvfynmz5+f//mDH/ygWLt2bclpn3766eKTn/xk/ufOzk7x7W9/O/8zAHHPPfcIIYTo6+sTAMTzzz8/axrf+MY3RE9PT/7nX//616K5uVmMjY2VnG/xfAqdeuqp4qqrrpr1u/PPP1+ce+65QgghHnjgAVFTUyN27NhRcvrFzj33XHHdddflfy5ebiGE2LhxowAgDh06JIQQ4kMf+pA4++yzZ33m+uuvF729vbOW4+KLL87/nM1mxaJFi8Rtt92mumxeU24/1JLX2ANIRESmy2YFHti8H/HxDI5e1IxQYz1qa3wINdbj6EXNiI9n8OCW/aZdDv6rv/orvPDCC/n/br31VgDAxo0bcfbZZ2Pp0qUIhUK49NJLcfDgQYyPjwN4swfQTJdffjl2796NP/3pTwCAH/3oR7jgggvQ1NSka3rbtm3D2rVrZ/1u7dq12LZtG4CZZWpvb8fy5csVvz89PY2vfe1rOO644zB//nw0NzfjwQcfxMDAgCHl2LVr16we1OOOOy7//z6fD4sXL86/7YLMw4dAiIjIdHtH0njl9TFEw41zxjHz+XyIhhuxe3gMe0fSWBYxfqDppqYmdHd3z/pdf38/zj33XHzkIx/BDTfcgEgkgieeeAJXXnllfpDd3JsXzLRo0SKcd9552LBhA4488kj87ne/w2OPPVbVNIvXsRAi/7tKy3TTTTfh29/+Nm655Zb8/ZHXXnstMpmMpjIUzrPwd8Xq6+vnlD2bzWqaF2nHHkAiIjLdeGYKE1PTCPqV+x0C/locnprGeGbKsjI988wzmJqawk033YSTTz4Zy5cvx759+2Z95rjjjsMjjzxiyPz8fj8AKN4/+Pd///e488478f3vfx9HHXXUnJ4zLXp6evDEE0/M+t2mTZvQ09MDYGaZBgcHsXPnTsXv//GPf8T69etx8cUX4/jjj8eRRx6JXbt2zVmWSvdB9vb2KpZj+fLlVb3CjIzBHkAiIjJdk78OjXW1SGWmEGqsn/P3dGYaDXW1aCoREM1w1FFHYWpqCt/97ndx3nnn4cknn8R//ud/zvrM5z//eaxatQof+9jH8JGPfAR+vx8bN27E+eefjwULFmia36JFixAIBHD//fejvb0djY2N+THz1q1bh3A4jK9+9av413/9V1XT27t375wxBTs6OnD99dfjggsuwJo1a3DmmWfit7/9Le6++248/PDDAIDTTz8d73jHO/C+970PN998M7q7u7F9+3b4fD6cc8456O7uxl133YVNmzZh3rx5uPnmmzE0NJQPkMDM07tPP/009uzZg+bmZsXhcK677jqcdNJJuOGGG/CBD3wATz31FP7jP/4D//f//l9N643MwR5AIiIy3dLWAI5a2IxYYmLOZUAhBGKJCXQvasbSVvMvuea85S1vwc0334xvfOMbOPbYY3H77bfjxhtvnPWZ5cuX48EHH8SLL76It771rTjllFPwm9/8BnV12oNqXV0dbr31Vnz/+9/HkiVLsH79+vzfampqcPnll2N6ehqXXnqpqul961vfwurVq2f9d++99+K9730vvvOd7+Df//3fccwxx+D73/8+NmzYgDPOOCP/3bvuugsnnXQSPvjBD6K3txef+cxn8j16X/ziF7FmzRqsW7cOZ5xxBhYvXoz3vve9s+b96U9/GrW1tejt7cXChQsV7w9cs2YNfvnLX+LOO+/Escceiy996Uv413/9V1x++eWa1x0ZzyeULsgXGR0dRTgcRiKRQEtLixXlIiIiyUxMTKCvrw9dXV1obGzU/P3dw0lseHIP4uMZRMONCPhrkc5MI5aYQKTJjyvWHoHuRSETSu4MV111Ffbv3497773X7qKQxMrth1ryGi8BExGRJboXhXDF2iPwwOb9eOX1MewfnUBDXS1WLQ3jXce0eTb8JRIJ/OUvf8Htt9+O3/zmN3YXhzyCAZCIiCzTvSiEI89oxt6RNMYzU2jy12FpawA1Nb7KX3ap9evX489//jOuueYanH322XYXhzyCAZCIiCxVU+MzZagXp6p2yBciPfgQCJkumUyiv78fyaQ9r3oiIiKi2dgDSKaLx+OIxWIAgFDIm/f4EBERyYQBkEyXGx9KaZwoIiIish4DIJkuFAqx54+IiEgivAeQiIiIyGMYAImIiIg8hgGQiIjI5c444wxce+21dheDJMIASERErjY8PIxrrrkGHR0daGhowOLFi7Fu3To89dRTqqfx4x//GK2treYVUoHVoS2TyeCb3/wmjj/+eASDQSxYsABr167Fhg0bMDk5aVk5yBp8CISIiFztfe97HyYnJ/GTn/wERx55JPbv349HHnkE8XjclvJMTk6ivr7elnmXkslksG7dOrz44ou44YYbsHbtWrS0tOBPf/oTvvWtb2H16tV4y1veYkvZZFxfriBUSCQSAoBIJBJqPk5ERC6UTqfF1q1bRTqd1j+RqUkh4n1CZLMzP2ezMz9PTRpRxDkOHTokAIjHHnus7Oduuukmceyxx4pgMCja29vFRz/6UZFMJoUQQmzcuFEAmPXfl7/8ZSGEEADEPffcM2ta4XBYbNiwQQghRF9fnwAgfvGLX4jTTz9dNDQ0iB/96EfiwIED4sILLxRLly4VgUBAHHvsseKOO+7IT+Oyyy6bM8++vj4hhBBbtmwR7373u0VTU5NYtGiRuPjii8Xrr7+e/+7Y2Ji45JJLRFNTk1i8eLH41re+JU4//XTxyU9+suTyf+Mb3xA1NTXiueeem/O3TCYjxsbGhBBCTExMiI9//ONi4cKFoqGhQaxdu1b8+c9/FkIIMT09LZYuXSpuu+22Wd9/9tlnBQDxyiuvCCGEGBkZEVdddZVYuHChCIVC4q/+6q/ECy+8kP/8l7/8ZXH88ceL//f//p/o6uoSPp9PZLNZ8fvf/16sXbtWhMNhEYlExF//9V+L3bt3z5rXk08+KY4//njR0NAgTjjhBHHPPfcIAOL555/Pf6bS+pNduf1QS17jJWAiIrLG9BTw8q+AP94E7H4EEALY/fDMzy//aubvBmtubkZzczN+/etf4/DhwyU/V1NTg1tvvRWbN2/GT37yEzz66KP4zGc+AwA49dRTccstt6ClpQWxWAyxWAyf/vSnNZXjs5/9LD7xiU9g27ZtWLduHSYmJnDCCSfgvvvuw+bNm3H11VfjkksuwdNPPw0A+M53voNTTjkFV111VX6ey5YtQywWw+mnn463vOUteOaZZ3D//fdj//79uOCCC/Lzuv7667Fx40bcc889ePDBB/HYY4/h2WefLVu+22+/HWeddRZWr14952/19fVoamoCAHzmM5/BXXfdhZ/85Cd47rnn0N3djXXr1iEej6OmpgYXXnghbr/99lnfv+OOO3DKKafgyCOPhBACf/3Xf42hoSH87ne/w7PPPos1a9bgzDPPnNUju3v3bvzyl7/EXXfdhRdeeAEAMD4+jn/6p3/CX/7yFzzyyCOoqanB3/7t3yKbzQKYeevUeeedh1WrVuG5557DDTfcgM9+9rOzyqJm/XmGmrTJHkAiIqqqB3BqUojn7xDizkuE+NWVM//9+YdC/OrDM/9/5yUzfzehJ/B//ud/xLx580RjY6M49dRTxec//3nx4osvlv3OL3/5SzF//vz8zxs2bBDhcHjO56CyB/CWW26pWM5zzz1XXHfddfmflXrtvvjFL4p3vetds3732muvCQBix44dIplMCr/fL+6888783w8ePCgCgUDZHsBAICA+8YlPlC3f2NiYqK+vF7fffnv+d5lMRixZskR885vfFEII8dxzzwmfzyf27NkjhHizV/B73/ueEEKIRx55RLS0tIiJiYlZ0z7qqKPE97//fSHETA9gfX29GB4eLlue4eFhAUC8/PLLQgghbrvtNjF//vxZ9fO//uu/ZvUAVlp/TsAeQCIico7RQWBgE1DrB+YfBfiDwM77AX/TzM+1fmDgKWB0r+Gzft/73od9+/bh3nvvxbp16/DYY49hzZo1+PGPf5z/zMaNG3H22Wdj6dKlCIVCuPTSS3Hw4EGMj48bUoYTTzxx1s/T09P42te+huOOOw7z589Hc3MzHnzwQQwMDJSdzrPPPouNGzfmezabm5uxcuVKAMArr7yCV155BZlMBqecckr+O5FIBCtWrCg7XSEEfD5f2c+88sormJycxNq1a/O/q6+vx1vf+lZs27YNALB69WqsXLkSP//5zwEAjz/+OIaHh/M9bM8++yzGxsbyy5z7r6+vD6+88kp+up2dnVi4cOGc+X/oQx/CkUceiZaWFnR1dQFAfp3t2LEDxx13HBobG/Pfeetb36pp/XkJHwIhIiLztXYCPe8BXrwTSMaAUBRobgN8NTM/AzN/b+0wZfaNjY04++yzcfbZZ+NLX/oS/v7v/x5f/vKXcfnll6O/vx/nnnsuPvKRj+CGG25AJBLBE088gSuvvLLi068+nw9CiFm/U/pO7hJqzk033YRvf/vbuOWWW7Bq1So0NTXh2muvRSaTKTu/bDaL8847D9/4xjfm/C0ajWLXrl1lv1/K8uXL8yGulNxyFgfF4vB40UUX4Y477sDnPvc53HHHHVi3bh0WLFiQL380GsVjjz02Z/qFT1kXry8AOO+887Bs2TL813/9F5YsWYJsNotjjz02v86UQmzxtqm0/ryEPYBERGQ+nw/oPgvoPHUm8InsTPgT2ZmfO08Bus+c+ZwFent78717zzzzDKampnDTTTfh5JNPxvLly7Fv375Zn/f7/Zienp4znYULFyIWi+V/3rVrF1KpVMX5//GPf8T69etx8cUX4/jjj8eRRx45J7wpzXPNmjXYsmULjjjiCHR3d8/6r6mpCd3d3aivr8ef/vSn/HcOHTqEnTt3li3Phz70ITz88MN4/vnn5/xtamoK4+Pj6O7uht/vxxNPPJH/2+TkJJ555hn09PTMmtbLL7+MZ599Fv/zP/+Diy66aFb5h4aGUFdXN6f8uZCo5ODBg9i2bRu+8IUv4Mwzz0RPTw8OHTo06zMrV67ESy+9NOtez2eeeUbT+vMSBkAiIjJf7oGP/k0zvX+58Oermfm5/6k3Hwwx0MGDB/HOd74TP/vZz/DSSy+hr68Pv/rVr/DNb34T69evBwAcddRRmJqawne/+128+uqr+OlPf4r//M//nDWdI444AmNjY3jkkUdw4MCBfMh75zvfif/4j//Ac889h2eeeQYf+chHVA1Z0t3djYceegibNm3Ctm3bcM0112BoaGjOPJ9++mns2bMHBw4cQDabxT/8wz8gHo/jgx/8IP785z/j1VdfxYMPPogPf/jDmJ6eRnNzM6688kpcf/31eOSRR7B582ZcfvnlqKkp39xfe+21WLt2Lc4880x873vfw4svvohXX30Vv/zlL/G2t70Nu3btQlNTEz760Y/i+uuvx/3334+tW7fiqquuQiqVwpVXXpmfVldXF0499VRceeWVmJqayq9nADjrrLNwyimn4L3vfS8eeOAB7NmzB5s2bcIXvvCFOWGt0Lx58zB//nz84Ac/wO7du/Hoo4/in/7pn2Z95kMf+hCy2SyuvvpqbNu2DQ888AC+9a1vAXiz17LS+vMUNTcc8iEQKmV0dFTs2bNHjI6O2l0UIjJZVQ+BxPuE+M0/zjzw8ejXhfjNx4X42ftn/n306zO//83HhYjvMbTMExMT4nOf+5xYs2aNCIfDIhgMihUrVogvfOELIpVK5T938803i2g0KgKBgFi3bp347//+bwFAHDp0KP+Zj3zkI2L+/PmzhoHZu3eveNe73iWamprE0UcfLX73u98pPgRSOAyJEDMPZqxfv140NzeLRYsWiS984Qvi0ksvFevXr89/ZseOHeLkk08WgUBg1jAwO3fuFH/7t38rWltbRSAQECtXrhTXXnutyL4xtE4ymRQXX3yxCAaDoq2tTXzzm9+sOAxMbl3deOONYtWqVaKxsVFEIhGxdu1a8eMf/1hMTs48nJNOp8XHP/5xsWDBgjnDwBT63ve+JwCISy+9dM7fRkdHxcc//nGxZMkSUV9fL5YtWyYuuugiMTAwIIR4cxiYYg899JDo6ekRDQ0N4rjjjhOPPfbYnIdwnnzySXHccccJv98vTjjhBHHHHXcIAGL79u35z1Raf7Iz6iEQnxCVT7dGR0cRDoeRSCTQ0tJiYhwlp+nv70csFkM0GkVnZ6fdxSEiE01MTKCvrw9dXV2zbrRXJTcEzI7fzzzwAcxc9u3fBMAHTGeAFe8GVp0P1PL2dDLG7bffjiuuuAKJRAKBQMDu4hii3H6oJa9xL6OqRCKRWf8SESmqrZsJd8DM074975m556+1A9h2L9BxJsMfVe2///u/ceSRR2Lp0qV48cUX8dnPfhYXXHCBa8KfkbinUVVCoRBCoZDdxSAiJ8iFwM5TZ4Jf7sGQBcuBlqUMf1S1oaEhfOlLX8LQ0BCi0SjOP/98fO1rX7O7WFLi3kZERNaprQPmFdwu4vPN/pmoCp/5zGfyb3Ch8vgUMBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBER0Rt8Ph9+/etf210MTznjjDNw7bXXmjqPxx57DD6fDyMjI6bOx4plMQoDIBERudrll18On88Hn8+H+vp6tLW14eyzz8aPfvQjZLPZWZ+NxWJ497vfbVNJqzM6Oop//ud/xsqVK9HY2IjFixfjrLPOwt133w0Vb31V7fLLL8d73/tew6ZXyfT0NG688UasXLkSgUAAkUgEJ598MjZs2GBZGYqVCpR33303brjhBnsKpREHgiYiItc755xzsGHDBkxPT2P//v24//778clPfhL/8z//g3vvvRd1dTPN4eLFi00tx/T0NHw+H2pqjO1/GRkZwdvf/nYkEgl89atfxUknnYS6ujo8/vjj+MxnPoN3vvOdaG1tNXSelUxOTqK+vr7q6XzlK1/BD37wA/zHf/wHTjzxRIyOjuKZZ57BoUOHDCilsRz1WlShQiKREABEIpFQ83EiInKhdDottm7dKtLpdNXTymazYmhsSGSzWQNKVt5ll10m1q9fP+f3jzzyiAAg/uu//iv/OwDinnvuEUIIcfLJJ4vPfvazs74zPDws6urqxKOPPiqEEOLw4cPi+uuvF0uWLBHBYFC89a1vFRs3bsx/fsOGDSIcDovf/va3oqenR9TW1opXX31VbNy4UZx00kkiGAyKcDgsTj31VLFnz5789+69916xZs0a0dDQILq6usRXvvIVMTk5WXIZP/rRj4qmpiaxd+/eOX9LJpP576ot7/333y9WrlwpmpqaxLp168S+ffuEEEJ8+ctfFgBm/bdx40bR19cnAIhf/OIX4vTTTxcNDQ3iRz/6kThw4IC48MILxdKlS0UgEBDHHnusuOOOO2aV7/TTTxef/OQnSy7b8ccfL77yla+U/LsQM/XpG9/4hujq6hKNjY3iuOOOE7/61a/yf9+4caMAIA4dOpT/3ZNPPilOO+000djYKNrb28XHP/5xMTY2lv/7xMSEuP7660V7e7vw+/2iu7tb/PCHP8wva+F/l112meKyxONxcckll4jW1lYRCATEOeecI3bu3Kl6fSsptx9qyWu8BExERJbbeWgnfrLlJ9h5aKdtZXjnO9+J448/Hnfffbfi3y+66CL8/Oc/n3X59Be/+AXa2tpw+umnAwCuuOIKPPnkk7jzzjvx0ksv4fzzz8c555yDXbt25b+TSqVw44034oc//CG2bNmCSCSC9773vTj99NPx0ksv4amnnsLVV18Nn88HAHjggQdw8cUX4xOf+AS2bt2K73//+/jxj39c8p222WwWd955Jy666CIsWbJkzt+bm5vzPZxqy/utb30LP/3pT/GHP/wBAwMD+PSnPw0A+PSnP40LLrgA55xzDmKxGGKxGE499dT8dz/72c/iE5/4BLZt24Z169ZhYmICJ5xwAu677z5s3rwZV199NS655BI8/fTTqrYRMNMr++ijj+L1118v+ZkvfOEL2LBhA2677TZs2bIFn/rUp3DxxRfj8ccfV/z8yy+/jHXr1uHv/u7v8NJLL+EXv/gFnnjiCfzjP/5j/jOXXnop7rzzTtx6663Ytm0b/vM//xPNzc1YtmwZ7rrrLgDAjh07EIvF8J3vfEdxPpdffjmeeeYZ3HvvvXjqqacghMC5556LyclJVevbVBUjosZESURE7mRUD+B0dlr86OUfiSvvv1L86OUfienstEElVFaqB1AIIT7wgQ+Inp6e/M8o6AHM9fb94Q9/yP/9lFNOEddff70QQojdu3cLn883p9ftzDPPFJ///OeFEDM9PADECy+8kP/7wYMHBQDx2GOPKZbptNNOE1//+tdn/e6nP/2piEajip/fv3+/ACBuvvlmxb/naCnv7t2783//3ve+J9ra2vI/K63PXK/YLbfcUrYMQghx7rnniuuuuy7/c6UewC1btoienh5RU1MjVq1aJa655hrxu9/9Lv/3sbEx0djYKDZt2jTre1deeaX44Ac/KISY2wN4ySWXiKuvvnrW5//4xz+KmpoakU6nxY4dOwQA8dBDDymWSalHsXhZdu7cKQCIJ598Mv/3AwcOiEAgIH75y18KIdSt72JG9QDyHkCaI5lMIh6PIxKJIBQK6fqMmmkQkTftOrQL2+PbsSy0DNvj27Hr0C6siKywpSxCiHzPW7GFCxfi7LPPxu23347TTjsNfX19eOqpp3DbbbcBAJ577jkIIbB8+fJZ3zt8+DDmz5+f/9nv9+O4447L/xyJRHD55Zdj3bp1OPvss3HWWWfhggsuQDQaBQA8++yz+Mtf/jKrx296ehoTExNIpVIIBoNzlgFAyeXIUVveYDCIo446Kv9zNBrF8PBw2WnnnHjiibN+np6exr/927/hF7/4Bfbu3YvDhw/j8OHDaGpqUjU9AOjt7cXmzZvx7LPP4oknnsAf/vAHnHfeebj88svxwx/+EFu3bsXExATOPvvsWd/LZDJYvXq14jSfffZZ7N69G7fffnv+d0IIZLNZ9PX14eWXX0ZtbW2+p1ePbdu2oa6uDm9729vyv5s/fz5WrFiBbdu25X9XzfquBgMgzRGPxxGLxQCgZHir9Bk10yAi78mKLDbt24TJ7CQWBhfilZFXsGnfJhw972jU+Ky/K2nbtm3o6uoq+feLLroIn/zkJ/Hd734Xd9xxB4455hgcf/zxAGYuvdbW1uLZZ59FbW3trO81Nzfn/z8QCMwJZxs2bMAnPvEJ3H///fjFL36BL3zhC3jooYdw8sknI5vN4l/+5V/wd3/3d3PK09jYOOd3CxcuxLx582aFCiVqy1v84IbP51P9FHFxsLvpppvw7W9/G7fccgtWrVqFpqYmXHvttchkMqqml1NTU4OTTjoJJ510Ej71qU/hZz/7GS655BL88z//c/5J7v/93//F0qVLZ32voaFBcXrZbBbXXHMNPvGJT8z5W0dHB3bv3q2pfEpKrbPik45q1nc1GABpjtxTTOWeZqr0GTXTICLvyfX+RZtmersWNy22rRfw0Ucfxcsvv4xPfepTJT/z3ve+F9dccw3uv/9+3HHHHbjkkkvyf1u9ejWmp6cxPDyM0047TfP8V69ejdWrV+Pzn/88TjnlFNxxxx04+eSTsWbNGuzYsQPd3d2qplNTU4MPfOAD+OlPf4ovf/nLc+4DHB8fR0NDQ9XlzfH7/Zienlb12T/+8Y9Yv349Lr74YgAzwWvXrl3o6enRPX9gplcQmFm23t5eNDQ0YGBgQHWP3Zo1a7Bly5aS63jVqlXIZrN4/PHHcdZZZ835u9/vB4Cy66G3txdTU1N4+umn8/dJHjx4EDt37qx6+Y3AAEhzhEKhir12lT6jZhpE5C2FvX9N9TM9RU31TRgaHzK9F/Dw4cMYGhqaNQzMjTfeiL/5m7/BpZdeWvJ7TU1NWL9+Pb74xS9i27Zt+NCHPpT/2/Lly3HRRRfh0ksvxU033YTVq1fjwIEDePTRR7Fq1Sqce+65itPs6+vDD37wA7znPe/BkiVLsGPHDuzcuTNfji996Uv4m7/5Gyxbtgznn38+ampq8NJLL+Hll1/GV7/6VcVpfv3rX8djjz2Gt73tbfja176GE088EfX19fjjH/+IG2+8EX/5y190l7fYEUccgQceeAA7duzA/PnzEQ6HS362u7sbd911FzZt2oR58+bh5ptvxtDQkKYA9P73vx9r167FqaeeisWLF6Ovrw+f//znsXz5cqxcuRJ1dXX49Kc/jU996lPIZrN4+9vfjtHRUWzatAnNzc247LLL5kzzs5/9LE4++WT8wz/8A6666io0NTVh27ZteOihh/Dd734XRxxxBC677DJ8+MMfxq233orjjz8e/f39GB4exgUXXIDOzk74fD7cd999OPfccxEIBGb1ogLA0UcfjfXr1+Oqq67C97//fYRCIXzuc5/D0qVLsX79etXLbxY+BUxERJYo7v3LKewFNMv999+PaDSKI444Aueccw42btyIW2+9Fb/5zW/mXA4tdtFFF+HFF1/Eaaedho6Ojll/27BhAy699FJcd911WLFiBd7znvfg6aefxrJly0pOLxgMYvv27Xjf+96H5cuX4+qrr8Y//uM/4pprrgEArFu3Dvfddx8eeughnHTSSTj55JNx8803o7Ozs+Q0582bhz/96U+4+OKL8dWvfhWrV6/Gaaedhp///Of493//93xI01PeYldddRVWrFiBE088EQsXLsSTTz5Z8rNf/OIXsWbNGqxbtw5nnHEGFi9erHkQ6XXr1uG3v/0tzjvvPCxfvhyXXXYZVq5ciQcffDD/dPMNN9yAL33pS7jxxhvR09OT/06py/vHHXccHn/8cezatQunnXYaVq9ejS9+8Yv5+zAB4LbbbsP73/9+fOxjH8PKlStx1VVXYXx8HACwdOlS/Mu//As+97nPoa2tbdbTw4U2bNiAE044AX/zN3+DU045BUII/O53vzNkfMRq+YSKC82jo6MIh8NIJBJoaWmxolxERCSZiYkJ9PX1oaurS/FetHKyIoufbPkJXnr9JRzVetScv7868ipWLVyFy465zJZ7AYmcotx+qCWv8RIwAeBTu0Rkrth4DHuTe+Hz+fBq4tW5H/ABe5N7ERuPYWnz0rl/JyJDMQASAD61S0Tmagu24cKVF2JalL5pvtZXi7Zgm4WlIvIuBkACwKd2ichcdTV16J6n7qlWIjIfAyAB4FO7REREXsI7bcl0yWQS/f39SCaTdheFiIiIwB5AsgDvLyRyFyveUkBEyoza/xgAyXS8v5DIHXJjl6VSKQQCAZtLQ+RNqVQKwNxXyGnFAEimKRxaptwApkTkDLW1tWhtbc2/qD4YDM55xy0RmUMIgVQqheHhYbS2tlYcwLwSBkAyDS/9ErnP4sWLASAfAonIWq2trfn9sBoMgGQaXvolK3Ewc2v4fD5Eo1EsWrQIk5OTdheHyFPq6+ur7vnLYQAkU7AxJquxx9latbW1hjVERGQ9DgNDphgcHMSWLVswODhod1HIIyKRCKLRKHuciYhUYA8g5bHXjpyMg5kTEanHAEh5lS6haQmI7e3tCAaD7I0hIiKSEAMg5VV6aEPLPVbsjSEiIpIXAyDlVQptfKqXiIjIHRgASTX26hEREbkDnwImIiIi8hgGQCIiIgMlk0n09/cjmUzaXRSikngJmIiIyEAclJycgAGQpMQxCakY6wQ5BR+YIydgACQp8QyairFOkFPwgTlyAgZAklIkEkEqlUIqlUIymeTBlNirQkRkIAZAklIoFEIwGEQsFkMwGGQAJPaqEBEZiAGQpMUeHyIiInMwAJK02ONDRERkDo4DSESexfHaiMirGACpomobSTayJKvck8XxeNzuohARWYqXgKmiaoff8NrwHRyvzjl4nykReRUDIFWkt5HMBSG/349oNOqZRtZrgVcWeoI37zMlIq9iAKSK9DaSuSAUjUbR2dlpQsnkxF4lezB4ExGpxwDoQE65xOjVIMReJXt4tb4REenBAOhATunpYBAiK7G+kRWccgJOVAkDoAOxp4OIyB5qT8AZFEl2DIAOxJ4OornY4JIV1J6AO+VKDXkXA6CE2JARaccGl6yg9gRcy5UaHvPJDgyAEmJD5h48sFuHt0aQTLRcqeExn+zAACghNmTuwQO7dXhrBDkVj/lkBwZACbEhcw8e2ImoEh7zyQ4MgOQaMl5u5YHdHjLWBSIimdTYXQDSLplMor+/H8lk0pDPuUXucms8Hre7KGQzNXXBa/sHEVEh9gA6kNr7yvTcf+bknhM7L7c6eb25kZq6wPszicjLGAAdKNeo+f1+9Pf3lwwdegKRUxpFpcBl5+VWr4Vt2ampC7w/k4i8jAHQgXKNW39/f9nQoScQVWoUzQ4taqcvW1B1c9h2K96fSVYw8pjJk0YyEgOgg5nRg1GpUTQ7tKidvmy9N8XrTc2BWrZl0IINEZE6Rh4zedJIRmIAdDA7ejDMDi1qpy97742aA7Xsy1AOGyIidYw8Zjr5pJHkwwBImpgdWuwIRWb0ZjnpQK1n+Y1ePvYoklsZeUxz8kkjyYcBkDzPjN4sJx2o9Sy/0cvHHkUiImsxAJLnOam3zgwyLL8MZSAi8hIGQPI8J/XWmUGG5ZehDEREXsI3gejEtwi4F7etMq4XZVwvzuDW7eTW5SLzsQdQJ96zZDwZHgRIJpPYunUr0uk0AG7bQqzzyjgIuDO4tf66dbnIfAyAOvGeJePJcCCLx+NIp9MIBALctkVkqvMyBSgOAu4Mbn1VpEz7JTkLA6BORt+zlEwmMTg4mA8f7e3tnmsYZDiQFZah3PqXKYBYRab79GQKUGa8cYeM57RXRaol035JzsIAKIl4PI6+vj4kk0mEQiEEg0HP7dQyHMjUlkHtAd2LQVErGcYhtJoMdd0LzNj/vFhfyZ0YACURiUTQ1dXFy48OofaAbnZPlRsCpgzjEJI7mbH/sb6SWzAASiIUCqGnp8fuYpBKag/oZp/5y3QpVC/2jpBZjKhbxSdZrK/kFgyARCYy+8zfDY0Re0fILEbUreKTLNZXcgsGQA9ww2VCUsbGSD/uF6SGG06yiJQwAHqAGy4T2knmoCBL2WQphxbl9gsnLg+ZgydZ5FYMgB5g1xlstY2oLI2wzAFalrLJUg4tyu0XlZZHlrpJRKQXA6AH2HUGW20okCVU2H0JqFzYMLtsaoOO1nLIEKDK7ReVlkeWuinDenQKriui2RgAyTTVhhO7g1eO3ZeAyoUNpbIZ2dCpDTpa15EsAaqUSstjZd0stz1lX48y4boimo0BkExTbXCyO3jJQmvYMLKhKzfvaoKmLOFeq8Jl7uzstGSe5banU9ejHbiuiGZjACTH8dqlHK1B2MiGrty8qwmaZoV7s+uGHb1I5bYnT5LU47oimo0BkByHb9coT01DZ8QyqgmaVq9Ls1/RZ0cvEoMLEZmBAZA0szsg8e0a1TNiGdUEE6vXpdmv6GMYIyK3YAAkzewOSHy7RvWsWkY987HiBMML25jeZPdJK5GMGABJMzMbTxkO1F7o5bFqGUvNx6wnW816atkMpdaBDPuA7LSuI7tPWolkxABImpnZePJA7Q1GPtlaGAac1LNXah1wH6hM6zpyUr0gsgoDIBnGqgcLqmV2Dwt7cCoz8snWwjDQ2dmpe52r2W7FnzFjKByGlcq0riMZenyJZMMASIax6sGCapndw2LE9N0eIo3czkYFJjXbrfgzZgyFw7BSmRHryMp9zO37MzkTAyAZRsZhQZSY3cNixPSdehmw3PZV+psR9cGowKRmuxV/hr11zmXlPubU/ZncjQGQDCPjsCBKzO5hMWL6Tg0W5bav0t9yv0ulUggGg9I//FP8GfbWOZeV+5hT92dyNwZAMkWpnh29vYQy9BwaQe1yaAkWMq2bcttX6W+5/0+lUrafGBhNhu0iQxlkZWV454kCyYgBkExRqidIby+hnp5DGRs/M3pAZehVzSm3fZX+lvtdMpnM9wDaycg6I8N2kaEMRCQnBkAyRTWXPMr1FGmZnoyNnxmXgtxweUmWHhIj64wM20WGMhCRnHxCCFHpQ6OjowiHw0gkEmhpabGiXOQxRva85Kbl9/uRyWSk6gH0Ghl7YctxWnmJiAppyWvsASQpGNnzkptWNBpFZ2en7ulYHQbsDh9mzF/GXthyZOmJJCIyGwMgScHIS1VWjgtnJLvDkhnzN2pbxGIxDAwMoKOjA9Fo1IiiERF5GgMgScHInhcrx4Uzkt33axk9fyN7FAcGBtDX1wcAugOgHT2sWuZpdw8wEXkLAyC5WjWNqtWXA+2+/Gj02xWM7FHs6OiY9a8eZvRwVqpf5eZZ/F27e4CJyFsYAEk3J/RYeLFRrXa7VPP9wvWd60n0+/3o7++vqp5Eo9GqL/2a0cNaqX6Vm2fxd+3uASYib2EAJN2cEK682KhWu10qfb9cQCxc37kexf7+fkPqidpgWupzZvSwVqpf5eZZ/F27e4ALOeHkjoiqwwBIujkhXMnUqFql2u1S6fvlAqLS+rb6oRyzTkyUQpFb6hcvRxN5DwMg6eaUxs9rvRnVbpdK39ca6NSUp3AbAdD9GkGlzxm1/Y0ORTKFLF6OJvIeBkByPZka2hwnh1Izgn/hNgKg+zWCSp8zavsbHYpkClkyX44mInMwAJLrydTQ5oJfKpVCIpEAIE8otVO51/8p0RKgjdr+RocimUKWTGUhImswAJLrydS45XqjwuEwotGoFKFUBsXbqNL2UurVM+rhDyf3zpJ9WG/IaRgAyXPsPFAXPyVrd3msYvQyKvXqVXOpV834hV7YTl5i9PaU8VYTonIYAMlz7DxQK/VGeaHhMHoZjXzaOJlMYuvWrUin02Wn44Xt5CVGb0+ZbjUhUoMBkDxHtgO1bOUxgxXLqPdSfzweRzqdRiAQmDV+YTEvbCcvMXp7ynSrCZEaDIBkKj2XWcy+1CbbgVq28pjB6mXUMmh0KpVCW1sb2tvbXb8d6E1e2O+IymEAJFPpucxi1L1cPLirJ9N6q6YsWp+yjsfjSCQSiEajJT/HJ7eJyI0YAKmsaoOBnsss1Vya4X1a+pi53rTWoWrKovUpa7/fD5/PB7/fX9U0ZQrQRERqMABSWdUGAz2XWaq5NGPmfVpq3lZRzTTtCA65+fv9/qqGpSm3HFrrUDXbUOkp63IymQyEEMhkMlVNkyceROQ0DIBUltU3vhcGkkwmozkYmXlfj5q3VVQzTTuCQ27+0WgUnZ2dVU8HmLscZrw6zqjvqilbuWkaFaCJiKzGAEhlWX2jdC5I+Hw+CCHyZZCB1rdV6J2mlYyaf7npKNUhq3s+jRgkWmkaRgVoIiKrMQCSVHIBorAHUA8zAobWt1XomWYlRi+XUQFf63T09HxWs+xG9LQqTcPuAE9ys/sWD6JyGABJKkYFErsvrZpxv2BuOm6410xPcKpm2Y0IakrT4FAiVI5b9ldyJwZAciW7e2bMuF8QsH+5jKInOFWz7EYENYY90sot+yu5EwMgVU3LoLtmXw4pnIed92SZcb8g4O0QIsuy87IeqSVLnSVSwgBIVcv1dqVSKQSDwZINoxWXQ2S55GLG/YJOYmZIsjuAyVLHiIiqwQBIVcv1bKVSqbINoxWXQ8yeh93hwyrVLqeZIcnuAMbLeu7klX2bKIcB0EPMOsDleruSyWS+B1DtvJ120LU7fKhhxDqtdjnNDEl6pu20ekbW4ysoyWsYAD3E6PBSfNArd79LqcvERpfJ7IDmhN4fI9ZBtcup594ntY2onmkbWS+ccBLgNDIEKL6CkryGAdBDjA4vWg56pS4TG10mswOamvBhd2NmxDqo9uZ1PT2+ZjaiRtYLJ5wEOI0MAUpvnU8mk0ilUgiHw6wT5CgMgB5i9BNpWhrCwsvEwEwQTCaThpdJhqfu7G7MZFgHg4OD6OvrQ1dXF3p6evK/N/KVcVoYuU5kWL9u4+RQHY/HkUgkEI1GWS/IURgASTc9DWEoFEIwGEQsFkMwGHTlAdPJjZnZKr0yDphpUAt/JuPY3TtdipNDNfd3cioGQLKc2w+YDDJAe3u74gNBlRp6u3tPqyFruCrk5PUrKyeHV/I2BkCynN3vv7Vi+l5vaLVs48L17+STAydsc1nXrxPCM5HbMACS9MxuWM2YfjUNrZWNoZ55GV2+wvXf2dnp2ABQapvLFG5k7a1yQnhWS6btTVQOAyBJz+xei+LpZ7MCe0fSGM9Moclfh6WtAdTU+DRNs5qG1srGUM+8jC6frL1SWpXa5m4KN2ZxSx0AuL3JORgASXqFTxD39/cb/s7hwoZ793ASD2zej1deH8PE1DQa62px1MJmrDu2Dd2LrDmYW9kY6pmXlu+o2RZm90rZ3SNT7fa0u/xWsLNn0uj166YwS+7GAEimsPO+Or1n4LuHk9jw5B7ExzOIhhsR9AeQykxh874E9iXSuGLtEZaEQCsbw1AohObmZgynhtEsmuHzze3p1DLgdzEZekMsKcP0FDA6CLR2Aj4fIAQw0g+0tFe9PZXK74VQaBUt9UOGExoiozAAkinsvK9Ozxl4NivwwOb9iI9ncPSiN4NQqLEezQ112DU8hge37MeRC5orXg6OxWIYGBhAR0cHotGo6jJYLdeYHaw5iIdjD2N993osqV8yp4GrZlvKcF+c6T0y01PAy78CBjYBPeuB7jOB3Q8D2+4FOk4FVp0P1Oo/1CqVX4Zg7URK9U5L/eB6JzdhACTDmTUyvtoz61KfKxc69o6k8crrY4iGG+Hz+ZDJHEY6nUYgEIDf34BouBG7h8ewdySNZZFg2fkPDAygr68PAGwLgGoCVjwex959e/GnqT9hZ2onNu3bhDOaz8D+of0AoKuBVMvshlSp19IUufC34/dArR948c6Znr/+TQB8M78HqgqBSuXnZUZ9lOqdlvrB9U5uwgBIhjN6ZHyjeovKhY7xzBQmpqYR9AcAAOl0GsnkGADA729AwF+L/aMTGM9MVSzn/PnzAQAdHR26y1qt4mUt1fPx6tirGBwaxLLQMmyPb8dbWt+CaDQ6q4Ez44EWsxtSIwKmqno3OjjT81frx+FQOzIH+9G49X9RP68dCEWBg68AA08BnacC8zr1Ls4cvMyojx3vuCaSFQMgGU7tQVZtsCvXmGsJh2XLNXkYU4fTSIzXIBJqQiAwEwRz/6Yz02ioq0WTv/QukytnNBrF2972trJlMVvxsiqtw6bmJuwRezCVnUJ9ph7pqTReGHkBlx1zGWp8NaaUI8fshtSIgKkqRLZ2Aj3vAV68E5mD/Th42I+6ujYsbIigITnzXfS8B8naeYireICpGtWeKHnhvkIGOKI3MQDSHNU2BGoPsmp7aco15lp6esqVq2F6HPNqMxg4IDCvOQi/vwF+fwMAQAiBWGICRy9oxNToMJL18xWno1ROuxrV4mVVKtuuQ7uwPb4dYV8YY2NjaGlswfb4duw6tAsrIitMKUcpRq8nIxp6VSHS5wO6zwJGBtC49X9RV9eGqaks0qlxNKRiwPJzgO4zMbh9u+K7kY2kptdXy/eJyN0YAGkOq546VNtLU64xN+pS4oL583Hu8YfxvzvGsWt45l7AgL8W6cw0YokJRJr8OGFxHfYPDaHG51Msj1I5BwcHTW/41SguW1ZksWnfJkxmJ7E4tBjpupn7HQfTg9i0bxOOnne0Yb2AasgYPlSFSCFmHvjo34T6ee1Y2BBBOjWOQLAJqI0C/U/N9BKKJaaXV02vr5bvu50XejyJymEApDkqPXWY+7naA6cRvTTF09B7UA+FQjj9LSuwdMmb4wDuH51AQ10tVi0N413HtKEtAMTjjVI0kNU2Xrnev2hTFA31DWhomOntXFyz2PBewEKlyu3Y8DHSP/O0L3xAKIqGZGym5682+uY9gNvuxbI1H0HwmGNMXT41vb5avu92Mp50EFmJAZDmqPTUocwHzmrL1r0ohCPPaC75JhCt02xvb0cwGDS84a9mOXO9f5npDAJ1AWRFNv+3QF0Ak9OTir2ARvSYlCq3Up0rNT+pem5a2meGetnx+5mwB8xc9u3fNPPzdAboOBMiFAVGRjE2NmbZq/ecHujM3s6OPekgMggDIKlSLhTKxIiDek2Nr+JQL2qZ1QhXs5yx8Rj2JvfC5/Ph1cSrcz/gA/Ym9yI2HsPS5qX5XxsR/AvLXamBL3X5XKoTkNq6mSFegJmnfXveMzMOYGvHG+MAngmsOh/xwb2IxWLw+XwQQgCw79V7TlFquY0Khk4PyETVYgAkzWQ+cMpQNit6qKpZzrZgGy5ceSGmxXTJz9T6atEWbJv1O92hs+AtGaFQCKHmZmCkH/0j04jtfx2AtmAjXc9NLgR2njoT/HIPhixYDrQsBWrr8mX1+/3IZDL5n9XUFemWtwwj636p5fZqICYyGgMgGcqOy3OyXSqUvYGqq6lD97xuzd/TGjqTySTiB4axaP8fEBh+bs5bMhYtWgO0vaNksCl1+VyGkD9Hbd3scf58vlk/lyqz2iGOOjuNG0PQTEbW/VLrzEmBmEhmDIBkKD0NQLVBTbZLhUaPg2glI8sUPzCMzHM/R3b0eSA0b85bMgL9G9HZ2Ah0nK/4fSmDnsEikQhSqRRSqRSSyeSs5ZX9REKJFeHMC/WCyAoMgGQoPQ2AWQ2dXT0FRo+DaCUj77uaXz8BMfEK/IFmYP5RQDIG7Lx/5ulYg96SobZcMoZtYGYdB4NBxGIxBINBxz8ZzXBG5BwMgGQoPQ1AtQ2doy4VFpClgS8MR0bed9W8tBd464dmev6SMRxuiCA93YBAQ9Ost2SgVf8r89SWS8awnWPX21KIyNsYAMl21TZ0Tm0oZSl3YTjq7Ow07r6rgrdkYOf9SE83YGwsBQCz3pIBny//Fa09dWrL5ff74fP54Pf71ZffIrLUAyLyFgZA0k3Wy2pOZdf6VBOidIWUgrdkIBRFoKEJAOa+JaMgBGrtqVNbrkwmAyEEMpmMtmUg1YQQGE4NY1FwEXwFoZ6I5GTdu57IdXKNdTwet7soruC69Vn8lozDcbSm9qDhcHzmHkBg5u8jA/mvRCIRRKNRwy+LmzHdZDKJ/v5+JJNJw6bpZDsP7cRPtvwEOw/ttLsoRKQCewBJN6PvYfN6j6Jd9wSadn+cyrdkoOXNwabNuhxqxnRlvq/Qarm3y+w8tNOWd0kTkXYMgKSb0Y2qmxvU4nCrFHatvhcsVwa/329Kr5vSWzKSbW9FKtOAeUNPwL/iTCS7zkF8cK9lod/IMSNleYhHBrl3Sy8LLTP1XdJEZBwGQJKGmxvU4nBrZdgtFW5yZYhGo+YNNFz0loz4wABiNUdhcsWRaO95a/4VaYA1ob/UetezPfjwxoxc71/6cBrhbBjpqTR7AYkcgAGQbKGmB8yuS8JmzLc43FoZdkuFG8vKUPCWjNy8wpHIrFekWRX6S83PzScfOWbtT7nev7AvjLGxMbQ0trAXkMgBGADJFmp6XMzsJSvXGJox3+Jwa2XvkRXjzKkNF3rWQzXBpfi7pebn5N48tevH6HqdTCZx4OABbDy4EZPZSSwOLUa6Lo1AIIDB9CB7AYkkxwBItlDT42Jmr0y5xlDNfL3+wEoxM8N6NdN2832lOUrLqFQ/jd6f4vE4ntnzDDanN6NzQSca6hvQ0NAAAFhcs5i9gESSYwAkW6jpcTGzV6ZcY6hmvrIECzVB1IqylnunrRHTLvzXqu86hdIyKm1zo/en1nmteG3gNdTU1SBQF0BWZPN/C9QFMDk9aVovIE/AiKrHAEielGsMc2O5aW1IZAkWasKdUlmNbkDLvdPWiGnrnZ6TL+2qpbSMVtTPsZoxpOpSqKuvw6uJV+d+wAfsTe5FbDyGpc1L5/69ClpOahgWiZQxANIcXjpg6u0dkyVY6H2Lhxm9grKEYsBbdViJlvqpd121Bdtw4coLMS2mS36m1leLtmCb6mmqpaWuydJbTyQbBkCaQ+0B0w2NrN7QIsuy6w2iZoQ1WUIxwEZfjVwdTqVSSCQSALStq7qaOnTP6zareGXlypl7a065cst0YkIkEwZAmkPtAdOpjazSk6FaOXXZc9QutxFBt3DA6UwmY0loZqNfWa4Oh8NhxYHAZTnJKUXtPijTiQmRTBgAaQ61B0ynNrJGhDfZl71S423l0CG5afh8PgghKk7LiOBRWIdlDzJ2KazDSutF9pMc2fdBItkxAJJuTj2zNqLhkH3ZKzXe5f5eGJiMWFe57xb2AGope7U9iNUEGSeHx0plr1SHZQ9Ysu+DRLJjACTP8ULDUanxLvf3wsDU2dlZ9boqvF9LTZAqLpvWHsRK09MiN28hBFI1KSwKLoLP59M8HTtU24Nn9n4ihMBwathR65TITRgAyfWc3IujV6XGu9zfzej50RJGisumtQex0vS0yM3rYM1BPLzlYazvXu+YgY1l78HbeWgnfrP7N45ap0RuwgBIrif7vUyyMaPnp5owYmePbSgUQlNzEx7d8ih2HtrpqNebydzTnRVZbNq3yXHrlMhNuMeR60UiEcWnHI2QG0g6mUwaPm03CYVChlxOtkLxNt11aBe2x7djWWhZ/vVmbmRlXfbKOiWSGQMgmc7ukGRm+Mj1LubGIyM5VFPnCrdprqcqfTiN+kw90pk0Nu3bNOu1Z25hVV22Yp0Wbn+7jz9EsmIAJNO5OSQV9i5W09A4sZGSucyDg4PYsmULBgcHNX+3cJvmeqrCvjDGxsbQghZdPVZ61lXxd4QQ2D++P/8gjNHM7CkvZNQ6LafwmOPm4w9RNXgPIJlO9pvRq1F4n1V/f7/uew2Nuk/Rygde3HpvZW6bZkUWm7ZswmR2EotDi5GuSyMQCGAwPaj5vjWt6yqZTGLr1q1Ip9P575j90IQV9wzmev8ms5OINESQmEgg1BDC8PSwofcCKh1z3Hj8IaoGAyCZTuab0UvRE6SqCbpGhWS14/sZsT1kDvbt7e0IBoNVlS3XUxVtiqKhvgENDQ0AgMU1i/M9VmqDWOG6UrMd4vE40umZwBmJRFzz0EThOp0cnwQATE9PY3GT9nVaTvExx2nHHyIrMACS7WQcpkVP71Y1QdeokKx2fD8j5iVzsK+2bLnAlZnOIFAXmHV/WqAugMnpSU1BTKmnOJVK5UNqcVmL39KxI75jzkMTThs6pXid1jTWICuyaGhsQENdg+Z1SkTVYQAk29l9KVEpgMrcu1VONeP7GRXE1UzHrPcDG7UMsfEY9ib3wufz4dXEq3P+PjU1he2x7di9YDeWL16uadqRSASpVArDw8MlB7Yu3I6FD02Es2Gkp9KODEqK69QHYOKN/3zA3uRexMZjWNq81MaSEnkDAyDZzu6wpRRAZe7d0qvSMmkJ4uWClprpVPt2j1KMOploC7bhwpUXYlpMK/49Fovh4IGDqEtrP4SGQiEEg0EIIfKXeJXk1vHBmoOzH5pobHFkL2CpdZpKpZBIJBAOhxFqCqEt2GZTCckuMl4F8gIGQLKd3WHL7gCqxMgDotppFb5xo7+/v+I9aqWClpr1We3bPSpNt9rp1dXUoXted8m/L6lfgnhzXPU9feXKWW4d7923F3+a+pMhD6LYrdQ67U/243DyMCLNEXTO67ShZGQ3u68CeRUDIHmeVQFUS1Aw8oCodlq59aDmaeZyQUvN+tSyzrWsN6u2ZbVPf6spZyQSwatjr2JwaBDRluofRJGVGSdg7FFyFhlPwr2AAZDIIlpCnZEHRK3TUvN5K3ttZe8dMKvxampuwh6xB6JGGPIgiqzMqEuy1xmaze6rQF7FAEhkES1BQc8BsVSvh9Zplfq8Xb0qegOWVeU1q/Gq9CCKlx+aqLRt2aNEVBkDIFnKy5dmzD7LrbbXo9K2satXRe96Ky6vmXXPjHs2W1pbyj6IAgC1vlpPPjRRqS6yR4moMgZAspTZIcUqesphdtmr7fWotG3UTl+WbVRcXjMDrFn3bHZ3ln4QpVqybCc9tDywRETKGABJF72Nh9khxSp6ymF22avt9ai0bdROX5ZtVFxepeUzKgTZec+mXrJsJz1y5S1+XR7Zz8knFl7DAEi66G08qr2/TJZ7e/SUQ/YeNKMum8myjYopLZ9RIcjIS45WXb6UdTupVfy6PJKDk08svIYBkHQxuvHQOlSJ3fSUw8weNCvf4lGJkQ+wmM3pIagaRu5Ldmw/NWMpkvW8vE85DQMg6WJ0EHPjQcPsy+SF0zfqrNuus3cj51vqNXNK26NcPbYj1Djt8lmuvLm3eQDW1RtZTgZpNm4X52AA9BCjGxcjp+fGg4bRl8nLTd+oAG1XEDdyvqVeM6d1ewwODqKvrw9dXV3o6elRNe9q9wmnXT7LlTccDiMajbrqBI7I7RgAPcToxsVpjVUpZvW6mB2mii+BGVF2u4K4kfMt9Zo5rT2ruYcLtKh2n3BaT7gbLsM6rdeVyCgMgB5idOPitMaqFLOCrNlhyo29pkYotV609qyGw2Ecc8wxpjzoU4rTtqnTyqvELSeyRFoxAHqI0QdrNxz8gZnGOpVKIZVKIZlMOmaZeI+aOarp1XLLPiEjp/bUE8mKAdCjvNCQqxUKhRAMBhGLxRAMBqsapkYvPdO342lhL/SWMMTJobiuOrWnnkhWDIAepeVgKmtYNLJclXoBzA4+eqavp+fCa/eokXMV11W76p6sxz+iajEAepSWg6msvT5GlqtSL4CW9VWuwSj1N7/fD5/PB7/fb1iZlTg9wMnSGMtSDlmYsT6K62o1PXXVlE/W4x9RtRgAPUrLwVTW0KC3XHoaAy3rq1yDUepvmUwGQghkMhnDy16o2stddjeG8XgcfX19GBoaQm9vb/53Vgcxu9eDbMxYH0Zdmk0mk7peGVc4niSHuCE3YgCkimR9Y4DecukZ302LcsG01N/Uhlm7g4fdJwORSARDQ0NIp9OIx+MAYMv6sHs9yKbU+pChp3RwcBCxWAyRSETX7RLRaBSdnZ1lPyvDchJpxQBIlqoUYNxwIC0XTKsdosTu4KG2nGZtx1AohN7e3vy0cwr/34o6ZNeDA3bvH6XmX2p92H3CkuP3+7Fw4UJNZXDDbTJE5TAAkqWqfdjCiAawvb0dwWBQdYiyu9Et5JQnFtU2iEZcjld7md0N7F42rfO3+4QF0L6/57jhNhmtZDrWkfkYAMlS1T5sYUQDqDVE2d3oOpHRl7S1NEyV5u3kRs7uoKF1/jKcsFhRBhmW0wh6j3VO3qe8jAGQpGLk07hGsbvRVUO2A7DRl7S1NEyV5m1EoLdrfdsdNOyeP5lLz6D4eh+yIfsxALqEbAFADbOfxjWKExo9WXoptW5TO+59NGJasqxvIiOFQpUHxS+We292IBCQ+iSZ5mIAdAnZGiQ1QcDOMhsRmGUK3bL0Upr1dhIjQ7gR48lxaBDvsGs/VzNfK8Zf1PJ5u4+DpA0DoEvIEgBy1AQBO8tsRPiUKXQbEWqseKOKksHBQezYsQORSASrV6+2fV2Wo2VoEHIHu/ZzNfOVYfxFJ1whIWUMgC5h1k6oNxyoCQJ2HjiMCJ+yhW69rHyjCqBcpw4fPox4PI54PC51Y+KWbU7q5ba13+9Hf3+/6mNhtSdWauoa6yNVgwHQ4yodpPSGAyvDnV33Esp05ltNY2N1I1Jcp9rb2+eUxShGXyKTaZuTNXLbvL+/X9OxsNoTKzV1jfWRqsEA6BGlGsJKByknnGHKdCnWLtWsA6sbkeI6FQqFTHkjC8C6YTaZ7oM1Q+HyVXNvHJGMGAA9olRDqOYVTjLf65RMJpFKpRAOh6U70FrZODqpsbEycDppvZjNjPro9oBduHydnZ3SXgUh0oMB0CNKNYSlDlJOObDH43EkEglEo1HpymnlOmRjo0xpvbi916oUM+qj2wO225ePvI0B0CO0BgSnHPjKldPMhr5w2gAU5+OUdeg1Tjm5MZoZ9dHOEw873/ns1ZMIchcGQFLklB6lcuU0s6EvnDYAxfk4ZR16jV0nDXZzW320M8h78STCzfuGVzEAkmuZ2QOnNG329FWmtRExo9Gx66SBjGVnD7sXe/e5b7gPAyC5lpk9HsXT5gFRHa2NSHFPq9k9EDI17G7tcTFquezs0XRbb6qS4u0k075BxmAApKoVHyjc2nBR9aoZSsOKHgiZGnazl9eu/dQLPUluOAYWbyeZ9g0yBgMgVa34QOGFAzzpY8RrpvT2QDitUTa7x8Wu/VTNcjltWxVz4jGQPX7ewwBIVSs+UPDAUR2nN35mqfaJTKc1ymb3uNi1n6pZLqdtq2JWrlujjhfs8fMeBkCXs2OohNzPyWQy/+5MwPz7t9widyBOpVIIBoNcZxWoDQvFjbIQAsOpYSwKLoLP56s4HxmDeTVlkrmBd/pJpJXrVk9YVqo3Tl/npB0DoMuVOjioGcfOyHkDykOlqCVj42uW3DZJpVKO7gWxitqGq7hR3nloJ36z+zdY370eKyIrKs5Hxl4pGctkBDvCqVOPMXqCm1K9kfmEgMzBAOhypQ4ORoYzLfPWe3bp1oZOSWEPaq4H0MsqNcx6Gq6syGLTvk3YeWgnNu3bhKPnHY0aX03Z7xjVQ2Jk0GCvjXGceozRU/9ZbwhgAHS9UgcHs8exU2rkqjmoevGA5eUz8sL6Y0bDvOvQLmyPb8ey0DJsj2/HrkO7VPUCGsGo5Snuxc/dblFumk7t5bKCkccY2dezl48t9CYGQI9SM45dNQcxoxttHrC8pbD+GB3+c71/6cNphLNhpKfSqnoBjarTRi2Pnl58p/ZyWcGIY0zumJlKpZBIJPLTNYvsQZPkxgDoQVY8NenFHjsvM7ohKqw/Rof/XO9f2BfG2NgYWhpbVPUCGlWnjVoePb343C/NlTtmhsNhRKNR09czAz1VgwHQg/Q+NakFe+yMZ9XZvp756G2ISs3LyPpTOI+m5iZs2rcJk9lJLA4tRroujUAggMH0YMVeQNnqtFJ54vF4/m9kveITl8KREMzYJgz0VA0GQA/S+9Sk18h2ecWqs30989HbEFkx5E3h8uyb3Ift8e2INkXRUN+AhoYGAMDimsWW3wtopGQyia1btyKdTgN4c7sV12Gre4xk24fMVnzMNHt9e/0YTdVhAPQgHjTUke3yilVn+3rmo7dOWTHkTW4erfNacffA3chMZxCoCyArsvnPBOoCmJyeVP1EsGzi8TjS6ZnezMLtVlyHre4xkm0fshp76EhmDIAEwHtn6mrIdvA2OrhbcflVzXzNHvImN4+9Y3uxN7kXPp8PryZenftBH7A3uRex8RiWNi81vBxmKr70WPj7VCqFVCqFZDJp6clfMplEKpVCOByWZh+yGk+2SWYMgAQAGBwcRF9fH7q6utDT02N3caTg9oO3Xb0zdg1C2xZsw4UrL8S0mC75mVpfLdqCbaaWwwyl1l8oFEIwGEQsFkMwGLR8OycSCUSjUVfvR6VYMdg+UTUYAIkgdw+oWWWzq4fTrvnW1dShe163qs+atc6rna6e71da38XTLDUPrfOWrQfdalYMtl+JkfVY5mMk6cMASACA9vZ2T791QuZ7ldS8zk9rme08mDuhZ9Ws+lDtdPV8v9L6Lp5mqXlonbcTtrOZzB5svxy94xGWOy7IfIwkfRgACQAP1jL3Vqh5nZ/egbrNfPrWbtWEXLPqQ7XTNaNcxdMsNQ+Z9xEZqRlsv1ql6rjSeIRq9odyxxS/3w+fzwe/32/4cpA9GACJIHcALlW2ahpkK56+tVs1Adms+lDtdM0ol1JQKXU/oex1xMqebRkuiZaq40oPBW3btq3ifd7ljimZTAZCCGQyGUOXgezDAEjSkeHAaiW9y1tNg2zF07d2Y4+V91h5mVL1vKankB15DXuxCOOT02iqr8VSDKOmdRlQW10TXKqO6z02lPse9yf3YQAkWymFH6/da2Ln8jqhV0cvNy8bKbMypKia1/QUhp78KQ5ufQxP1p+KFxpOwFsOP4O1k09hfu8ZWLz2kqpCYKk6rnRc1Xufd+G0Ojs7dZeV5MMASLZSCj9eO9NUu7xe6xkl0srK0F9xXtNT6H/0h4g//1scRj1OaXgIXXUHsXjsOYxPTiP77K8BoOoQqMTIoZa8dkLuJQyAZCul8OO1nhu1Z/F2HYidcl+VWwKy05fD6eU3SnbkNcS3PY6xSWCqdSmCNWM4auQJjPkXYrppAdIjr+Lg1sex6Ni/Qk3kCEPnbeRJtNdOyL2EAZBs5bWwp4Wa13ipbWyraZSlvK/K4O/KxIjlsDOE6S2/7cFR5b165cpZ+LcRLMKfGt6Ot2UeRI1IYLR2AQ7VNqG21o+WyQPw1dfiyfpT0CIWYpnKIqpdR0YeV3mMdi8GQCJJFQc+pQOx2sa2mkbZytd5GfFks9N7KoxYDqvCsFIg0Vt+WwO8hnv1ypWz8G/phnl4uelt6G5M4KiRJ5DItGBicgqNAJqnX8fu1rV4sfYEvGNy7ptpKg3vUmn4JqMG8yZ3YwAkkpSaM2+1b3nw+/358cDKfU6pwbHqdV7VNk5u6akwYjmsCsOV7jXTsk2NKLOuOvRG+Hv92V8jPV2DU+rL36tXrpyFfxvJ1OAth5/B4rHnMOZfiNpaPxoB1Pr9GJteiOjI8zi+eQGa6ucOyVJpeJdKwzcVv4Ukt07c0ktOxmAAJHIwpbBQ2AjmDvjRaLTsE3xqxhMrnrbRDQgbJ+NYFYYrhTYt29SIMusZ4Dw78hoObn0M6ekaTLceCTF5oOy9emrv2W062Ie1k09hfHIa000L0DJ5AM3Tr2NseiHG6xegdnwUayefwlLfOgBNs6ZVaXiXSsM3FX6/cBtUcxuJDJxUVidgACTNuBPaR+to/mp7VdSOJ2ZmSHN64+RFlUKb2h5qo7avngHO92IRnqw/BafUPwwxeQCjtfNm3asnan14ePotqB0LYmWZ3ah436hpXYb5vWcg++yvMR5/BVM1AjvmnYz2sZdQOz6KQG0W83tPx3htGPH+/lnroNJ61fP33PT13kZSTjKZxODgIICZ4WbM2ld5kmgsBkDSjDuhfdSs++K3AJTrIdQ6RISZlxZLNU59fX0YGhpCb2+vY+obg+uMSvWq1KVKvetMbQ9ZofHJabzQcCK66uLoiv8B+0U9pgQQxMy9epsDa/DU4R6sOnAIKzsWzfpu4Xaes2/U1s1cNgYw9cKDeGh6NbZOnYxTmxfh7ZNPIfLGvYX9g3sNOZ4WlgWYvS6rCelq5PZTAAgGg/ltYPQ+4Jb7fGXBAEiacSe0j5p1r6XR1Xpgtvo+u0gkgqGhIYyMjOC5557DokWLTO1hMApPktQpdamy2nVW6TKt3+9HJpNBJBJBU31t/l69hG8eprJAne/Ne/U601txin8BlixYPWsaxWXu7OycO883QmCw6604fKgep/jqUC8Wo33eaWhZsqLiPYVaFIdpKy+9RyIRdHV15f+/uDx8IllODICkGXdC+1Ra92rOup0U4EOhEHp7e7F161bEYjGMj4/nexhkJsM6dkIvZKlLlWbJhRKfzwchBABgWXM2f69eJtSORen9CItDSE235e/VO6v2BSxvfu+saRSWtWyZa+vQsuwYnLwM6O/vRyw2jEPhKFreeKrYqOOpUlnUrEsj6kkoFJrzfmEZ9gEqjwGQyEXKnXU79ZVOuRDY0tICQHuD4oQgZAZZeyFLbQ+9Tw9rkas7hT2ANcFA/l49JPvhr6/Fq/NOQ3Tk+Vn36tWE22dNQ83l1VLzt+IWCrXlMquesKNAfgyAZCivNrayKNfAyBoI1FDqYVDLqOXWUrermadR+5CsPTBq1o3aExmtveGlQknuXr2DWx/Hk/Wn4MXaE3B88/xZ9+qhQo+dmnLZGYpKla9UPVG6XO604waVxwBIhnJyyHCDcg2M3++Hz+eD3+/XPX0nBnwz7rGqtOzVzNOofcjqsKG2bqhZN0acyGhaj2/cq7fo2L9Ci1iId0xOo6m+B0t958z0/Kl4V6/sx75S5StVT5Qul8u4XKQfAyAZStZeB7fSEsgymQyEEMhkMrqnIXsjp8TMe6zMmKfefcjucK62bqhZN+U+U+3QRiXV1qEmckTRa9maSnzYgPnB2m2mtXxKl8uL2V3nqDoMgGQo3vdhjkqvhgL090pZ1bNlF6MaKavqtt752B3OraobatZPNdtc75h2erabldtMa/nUfN7uOkfVYQAkcgC1b+oop9QB3aqeLcCeHgOvNFJ2h3OZTv6q2eZKY9qZpXibOa1Hze46R9VhACSSWKV3+RrR6FrZcOttmIUQGE4NY1FwEXw+n6Z5eqWRkimAAdWHmWq+X802VxrTzizF26x4/6jmgRc9tE5HtjpH2jAAEklM7bt8nUJvw7zz0E78ZvdvsL57PVZEVmj6Lhspe1Tb82rXgOVqnzi34k0XpjzwUoZXestpBgMgkcTc1nulp2HOiiw27duEnYd2YtO+TTh63tGo8dWYVEIySrV1V/a6rxSWqg2FxfuHaQ+8mDwdcgYGQCKJsfcK2HVoF7bHt2NZaBm2x7dj16FdmnsByXrV1l3Z675SWDK6B03tOjBqXRVOx2n3I5J2DIBEJK1c71/6cBrhbBjpqTR7AUkKSqHLTT1osl4OZjA1DgMg6cKdkKyQ6/0L+8IYGxtDS2MLewFJWrIOvq3ne7KGWVmDqRMxANIsag8o3AndQ802tyPw53r/JrOTWBxajHRdGoFAAIPpQfYCkitUu1/pPQ6r+Z7ZYVbvsssaTJ2IAZBmUXtA4U7oDslkElu3bkU6nQag792sZsn1/kWbomiob0BDQwMAYHHN4qp6Adl7TXZQqnfV7ld6j8MyHL9LLXul/VP2e0OdhAGQZlF7YOBO6AyFB1MAig1QOj3Ts6b33axmyPX+ZaYzCNQFkBXZ/N8CdQFMTk/q7gWUsfeaodT9lOpdtfuV3uOwDMdvI95MRNVhAKRZZDgwkHEKD6YAyjZA5ba71fUiNh7D3uRe+Hw+vJp4de4HfMDe5F7ExmNY2rxU07Rl6P0o5pZGz4ggK0sYNrIcyWQSqVQK4XB4Vr1z+/G23AmoEW8mouowABK5mNLB1MwGyKhGsy3YhgtXXohpMV3yM7W+WrQF2zRPW8ZG1y2NnhFBVpYwbGQ54vE4EokEotGobctk92sYgbknoEqU9k9ZTgrchgGQyEJ6DmTVHPyKD6ZmHzyNajTraurQPa/bqGJJT8ZQqocRQVaWMGxkOWRYJjuCdaUTULVkOSlwGwZAmoNnW+bRcyCz6+Cnpx7I0NCRfYwIsrKEYSPLUWpaVh5r7dg3jToBNarsbNtmYwCkOXi2ZR49BzK7QpWeeiBL403kBFYea+2+3cOoKxnVTIdt22wMgDQHe3HMo+cgbFeoYj0g9phUr9w6NGMfs2qbaQ1TRoWvaqbDY9psDIA0B3txCGA9IG/2mBgdoMqtQzP2sWq3mdrl1xqmjAhfpZ6mVovHtNkYAIlIWuyBspcXe0yMDr1Wr8Ny81OzP6ldfq1hyojwJcPT1G7CAEhE0vJiD5SV+NaFuYwObFavw3LzU7M/yRb6C+uolWXzwsknAyARSUu2xshtGLDncnPoVbM/ybb8hXW0s7PT8LKVCnpe2DcYAInIUlrOrGVrjNyGAdt6dvYsOXF/MruOlgp6Xtg3GACJJOe0SxGVymvGmbXT1pEsnBgInCpXR1OpFBKJBAD39iwZyew6WiroeWHfYAAkspjWsGLVU31GqVReM86s3X65xm0B123Lo0aujobDYUSjUdN6lmKxGAYGBtDR0YFoNGrKPLSSeXt7IeiVwgBIZDGtYaXawFRpfkYfnCuV14wDrtsv1+S2YSqVQjAYlLIh1ULGwG52SCmso2Yu88DAAPr6+gBAmgAo4/YmBkAiy2kNK9UGpkrzUzo4y3zGrsTtZ/G5bXfgwAH09fWhq6sLPT09VU3Tzm1cWCdlqWtmhxSr6mhHR8esf4vZsb7dfoLmVAyARBYoPuhWc+DVegCvND+lg3M1jaGa78rS6DtFbhumUinDpml1r0zxpcncPPv7+6XoHXJLSIlGo2V7/vRs98L9NTcNLftu7nPxeFzTfKvF40x5DIBEFjCysTW64VYKiNU0hmq+q3UZzDqQO62BaG9vz18CrpbVgafUpUlZgpfbe5Fz9Kzvwv0VgK4TPDsuA/PSc3kMgEQWMLKRs6LBrKYxVPPdUstg9ZhcTmsgjAwpVgeeUpcmvRK8itl18qFmfReXrXh/TaVSSKVSSCaTqp/0tyPoy3JyISsGQCILOLnhzjGywSq1DFaPycUGwjqVLk16TTweR19fH4aGhtDb2yvV7RLF+2FhL14kEkEwGEQsFkMwGFT9pL/W45YRy52bZzKZRH9/v2N6+q3CAOhiTru8RXKzorfM6jG5vNr7RNUx4tgaiUQwNDSEdDqNeDxecjp29FJXui+42vcNqyHzbTNuwQDoYqz0ZCTZLz0TWcWIY2soFEJvb++shyuU2NFLXem+4HL7qVHtjtNum3EiBkAXY6UnIzGceRuvKLzJqGOrmn3KzP3OjNcyWrluCpVbFh67lDEAuhgrvXexsSaj8YrCm5SOrU7c58zYpna1O6yf2jEAErmQmQdDJzZ0VD1eUSjPygBi1D7opm2qdll4/HoTAyCRC5l5YOeZtjfJfEVBhkbdyjBl1D4o8zYFzLlEzePXmxgAiVzIzAO7m3oNyB1kaNStDFN690EZgrIWZmxXHr/exABIRJrI3mvgJE5rkGXltUZd7z4oQ1DWwoztyuPXmxgAyRRWN2xsSMmJnNYgy0r2Rl2G41MymUQqlUI4HHZMUK5mu8qwzmXHAEimsLphY0MqDx541fNaz5VX5Y5PqVQq/y5nq/eNeDyORCKBaDTqif2SbUJlDIBkCqsbNjak8rDrwOvE4Cl7zxUZo/AdunaFkuJjpNb9xWn7V7XL6wUMgGQKvQ2b3p2UDak2Zh4M7QrjPOMnWRW+kzbXA2hXGXK07i9O27+qXV4vYAAkqXAntYaZ69muMG5V8FQTntnbQMDcemDFPW2xWAwDAwPo6OhANBot+V2t+4vTr7I4vfxmYAAkqRi5k7IRLk2msGQUq4KnmvDMExlt3Ho50sh6oHZaAwMD6OvrA4B8AFT6rtb9xelXWZxefjMwAJJUjNxJ2QiXJlNYMoKVgUBNeGZvgzZuvRyZ2/5+vx/9/f2q6mepuqy2TnV0dMz6V8t3yVsYACnPKWfVavGgZz+rtoEZgaDU/qAmPLO3QRu3Xo7M1YP+/n7F+qlUx0rVZbV1KhqN5nv+tH6XvIUBkPKcclatFg969rNqG5gRCNy2P8ik2nvjnLZvl6qfSnXMKeGWnI8BkPJ44CGnMiMQyLA/uK1XPsdr4bpU/VSqY3aHW7fWOZqLAZDy7D7wEMnEjv2huPF1a1CSIVzbwcings3i1jpHczEAegjP7MhJvFhfixtftwYlGYNPtdTUVz3hSu9+kEwmMTg4CABob29X/V231jmaiwHQQ3hmR05i5YMdsihufN0YlNxKqb4W1zc94UrNflDqYZLccDDBYFB1PbJ6EH+yDwOgh8h0ZseDBVVSWF+Nqi+ynwQx8DlXJBJBKpVCKpVCMplUvISvtH0r1W01x+1SD5N0dXVV/K5RZN+3aC4GQA+RqXHhwYIqKayvpYbR0Mpt7wd1evndJBQKIRgMIhaL5Xvc9Ia34ulW2ralHibp6enRvBx6ydTBQOowAJIpSjVMud9PTk7C5/PB7/fbWEpyCqMal+LG1OknIk4vv9vouYRvRN2W4eRehjKQNgyAZIpSDVPu9z6fD0IIZDIZu4pIDmJW4yJLr4XenjxZyk8z9NRTBieyCwMgmaJUw1T4aqRMJlP2/i5e3nIPWbelLI2v3p48I8sv6zYiInMwANIcRjQEpRompd+Xur+Ll7fcg9uyPBl68riN7Fd47AXAQE6mYgCkOaxuCCr1FvLylnPlGjS/349oNMptWYIMPZHc3+yT209SqRT279+PoaEhtLS0IJFIAGAgJ3MwANIcVjcEWnoLyVlyJxPRaBSdnZ353/Nyo3y4v9knt5+Ew2EEAgGk02m0tLTwpIlMxQBIc5jdEBQ3/gwD7lXqZELGy42sh6RXtXWneD/R++YPM+sv9w/3YQAkyxU3/jKGATJGqZMJGS83uqkesrG2VrV1p3g/0TMNs+uvm/YPmsEASJYrbvz1hAE2cM5mRC+z0XVAxlCqFxtra8lQd8wugwzLSMZiACTd9DbASme7WhspNnBkdB0w69YHO05W2FhbS4b7J80ugwzLSMZiACTd7AxhbODIKXXAjv2EjTUZhVdb3IsBkHSzswFmA0fFdUBvQ2V2A+eUoEqkhFdb3IsBEO4/w0kmkxgcHAQAtLe3z1lGoy7lEtlJb0NldgPH/YSsZmSbxhMY92IAhPvPcOLxOPr6+gAAwWBwzjKavfxuD9gkB70NlZMbOO5bpKTUMV1PfeEJjHu5OgCqrexObgDUiEQi6Orqyv+/0t9L/c0Ibg/YJAe9DZWTGzjuW6TESeNvkn1cHQDVVnYnNwBqhEIh9PT0lP27mcufOwj5/X709/ezt4LIIG4/eSVjFdcXo3qQK91mRHJydQDkwVEOuYDZ39/Ps08iA7n95FUWTrvUXqrzo7C+JJNJbN26Fel0es7n9Myv3G1GJCdXB0AeHO1VfNBkICcyl9OCilM47dKpmmNtPB5HOp1GIBCo+phc6TYjkpOrA6BTueUgXnzQZCAnO7llvyrHaUHFKZx28qrmWFu4TNXWlUq3GZGcGAAlZNRB3O4Gz2kHTXK+cnVe735l936kBV+raA4nnrxW2q5OXCYyFgOghEodxLUeqO3uDSh3gGGjQ2YoV+f1npAUT1PmusvXKlIOtytVwgAooVIHca07tMw9cDw4kRnK1Xm9PR7F03Rb3ZX5OEH6KW1XmU9eyHoMgA6i9UAtcxc/Gx0ygxl1vniabmtYZT5OkH5K29VtJy9UHQZAB3HTgdpNy0LeUjyURjweRyqVQiKRyP+dSEY88aZCDIAkBSf3oLgB178+uR6VcDiMaDTKhpWkxhNvKsQAaCI2qurx0oS9uP71MXIoDSIiKzEAmkjWRtWKYKp1Hrw0YS+z17/eOif7SRR7VMgqsu8L5DwMgCbI7ah+v1/Ky0JWBFOt82BDai8t619PQ6S3zsl6EkVkNbv2BQZP9/JsADSzUud21Gg0is7OTlPKUk35rehtY4+ee5VriErVS731wayX1xM5jV3HVJ6EuZdnA6CZlVpro6W2LIXT0VL+4vlb0dvGHj33KtcQqXkJvRbF3zNjv/VSqPTSsjpVqW1k1zGVJ/Pu5dkAqLVSazlwam201JalcDpayi/jGZxb7wnzgnINkdlj5FWq98lkEoODgwCA9vZ2R7wxx0pGLyv3R+MVbqPcz3auX57Mu5dnA6DWSl3NgbNSo6W2LMVPHKoth4xncLwnzJ3MHny2VL0vHI+vr68PABAMBvkAUhGjl5X7ozpagnLhNuL6JTN5NgAWUrNzVnPg1Bo2jb4EkPteMplEf3+/FGfrRt0TRtWxogfHim1WOB5fV1cX0uk0UqkUksmk5h57NzN6Wbk/qqMlyCltI7vXL3t63YkBEOp2TisbCbPO+mQ6mzTqnjCqjhV1otI2K2xccmUqbmgqNUDFveP9/f2IxWKqewFJH+6P6ugNyrKs31LHCQZDZ2MAhHxnsWaVR7blJPvJUCeK73kqbmiSySSef/55xONxrFixAj09PSWnNTY2hng8jsnJSfh8Pvj9fsXPmd1wsWH0LqVtL0uQU6t4GUodJ2S7X5G0YQCEfDunWeWRbTnJfjLUCaXGpfD/4/E44vE4Dh8+rPj9ZDKJrVu3Ip1OIxAIQAiBw4cPY3x8HC0tLYhGo3O+Y3bPp0y97WQtN2z74mUodZzg/YrOxgBIurCHg6pVWIcKx8tUuv9pxYoVAGae7C3+bjwez4e/jo4OZDIZHDhwAOPj43M+m5u22T2fMvSskvWSySRSqRTC4bDjtn3hfqK2/sp4vyKpxwBIuvBsj6pR2GsHlK9DTU3NaG7rxHhmCiOTtWjKipJDIhUGvAULFpTsmTC751OGnlWyXjweRyKRQDgcRjweB+Cc42PhftLZ2am53IV1XksHATsT7MMASLqwh4OqEY/HMTIygmw2W/I+PQDYPZzEA5v345XXxzAxNY3GuloctbAZb+8K5V+zqBS22DNBdsjVsVQq5bgT5MJjenEo0xrStHQQsDPBPgyABXgmoh57OLyl1L6hd5+JRCJobW1FOp1GJpNR/Mzu4SQ2PLkH8fEMouFGBP0BpDJT2LwvgX2JNK5Ye4SqeRpdV3mcoFIKh9wKBoOOOuko3E9yT9Hnfq81pGnpIGBngn0YAAvIeCbCxoZkUGrf0LvPhEIh9Pb2zhr+pVA2K/DA5v2Ij2dw9KJm+Hy+me811qO5oQ67hsfw4Jb9OHJBM2pqfCXnY8YbZ2Q8TpBcnH6CXBzKtIa0pqZmjEzWYu/4FJomU1jaGii5nzp9XTkZA2ABGc9E2NiQDErtG2YMkJ5MJrFlzxC27xtBNByEz+dDJnM4/6AHADTXZLBt7yHsHUljWSSoOI3cm0ESiUR+fmqV2+9kPE4QGal43yy3rxafKJW6bWPdsW3oXsQ2TCYMgAVkPBNhY0MyKLVvmHF5devWrdg1PI5DSR9a/VlkaoJIp9NIJsfyn5tMp5Cc9mM8M6U4ncI3g+TuFdSi3H6XW14tN/mzJ5/cqPhESc1tGwyB8mAAlJyMoZSoGpUur6bTabQ2NcJfO4kDI2OorfHle/5y/45nphHK1qHJX6c4bb/fP+shEa0q7Xdae+bZk0+lOPnkoPBEyajbNsg6DIBEZIpSDVtxGFIaf6x13jzsyryGFwfiaGxshN/fAL+/AQAghMBY1o9VS8NY2hqYNc/ctKPR6KyxBbWUTw2tPfPsyadSnHxyUHii9Fo8hVdeH0M03Djntg2/vwHRcCN2D4+VvG2DrMcASGU5+eyUjKG3Dig1bMUD5RaPB1g4/th71nRg4OAYnn91P45YFMaCeS1IZ6YRS0wg0uTHu45pm9OToCVo7dq1C7t27cLRRx+NNWvWqF6u3PJofeiF46GREiedHJSro8OHRhFPJLEgGAZQP+u2Db+/AQF/LfaPTmD40CiyyddZzyXAAEhlOfnslIyhtw4oNWy5gXKj0ShCoRD6+/vzvQS5z+UambZIBKctmsIDr49i30FgbLoWDXW1WLU0jHcdo3xDuVLQKtVojY2NYXx8HGNjY8WTmcOKcMZ9zZuccptPpcHbJ1NJTE9O4FCyDoHGxvztGrW1tUgkRjBdU4+GulpMppKIjR9UnAZZiwFQQjL1BDjp7JSMpXQ/nRblBmNW+jd3ObiwkTm+qw2NkwmEo+0ItMzDZCqJns5FCLdU/0Tv0UcfjYaGBnR0dOiehpHM2tdkOp6QcxW+crH4ZC0SiaCnsw0rlySwJzGNqBD52zYSiRGMjiZxcMqPtx0dRU/nIowc8isOOE3WYgCUkEw9AU45OyXjabmfTg2lg31x/SpuZOLxOOZHIohGggAOIzZ+ECOH/Ai3tJSdbqFSwSoajSIajaoqs9/vRzgcRiqVQjKZNGWfMGtfk+l44hVuDDZKr1wsfn3cB9auwIYn92DX8My9gAF/LaZr6nFwyo+21ia865g2hFtC+f23eMBpshYDoISUGqxqDihuPBiR+YzqkSo1Jp9SvVRqZFKpFFKpFFpbWxV7InONUCqVQjAYhN/vRyaTmfWaOL31vjAEB4NBxGIxBINBR+1H7MW3nhtDt5oe/e5FIVyx9oj8OID7RyfQUFeLtx0dVbxtg3XTXgyAElLa0ao5oLjxYETaaT0RMKpHqtSYfEr1UmmMvcLgpdQTWfz+VZ/PByHErOnppdRAlWusZDzZYi++9bwSbJTqVveiEI48oxl7R9IYz0yhyV9X8k0grJv2YgB0iGoOKHYfjGRsFL3IrhMBpV69wt/7/X709/fn/15czkr1N9eIxGIxjI6OYv78+aivr6/qHqPC7xWGzkrT4MkWAQw24+NjyCbjWBqJIBTikC+yYgA0iNkhp5oDit0HIzaKcrDrRKDSW0SK7wMqLmfh98vtZ5lMBkII1NfX50Ob3nuMtNbZah+YIXITHvOdgQGwDC2hjhW+9PqKRCL5+7jMuoGeKrP7RKCUUoEvmUzO6hkElPezWCyGgYEBzJ8/f0740hJ6lQakVhvkSj0wo/YYwl5yspqZdc7uq06kDgNgGVpCHSt86fVVfB8XGzgqVCqYKtUnpf1sYGAAu3btQjwex9vf/vY5da9UfStuAIufaKx2zMNSy6Ck3NtRuL84h5O2m5mdFrKebNJsDIBlaAl1rPDl1xcDMmml1HOstJ91dHQgFothbGwMg4OD6OnpUTV9rfcallPqMrXaaRZ/jlcUnMlJ243HZGIALIOhThul9VXqZnqiStT2HEejUaxYsQJ9fX2apl/uXsNq6OlJLJ43G2dnctJ2Y/tGDIBwVre90zjpjJjko7ZBbW9vRzAY1NTwmtUAGhEC2Dg7U6XtxraGZMIACIYUMznpjJj0M6thUxuEqg1MRpaf4Y1KYVtDMmEAhJwhxS1nimwM7WVVPTKqYSscTqXwbR5qv6d3OdkwkxVkbGvIuxgAIWdIUWqQ3BIKS3H78tnBqmBjVMOWK6/Wt3kUvw5Oax3SW37WWdJCxraGvIsBUFJKDZLbeyncvnx2sKrHwaiGrfDtILkeQC3fy70OLlemcorDm9byJ5NJbN26Fel0WtX8iIhkwgCokVVn/EoNktsvH8i8fE7t6XFaj4Pe8hYOHl3pYZDctkylUkgkEvnvaxWPx5FOpxEIBHTVWafWKSJyBwZAjezspXJaY66VzMvH3klnUFOHctsyHA5rfm1bqTH+lOZZKeCxThGRnRgANZK5l8qpZO0JqebVYCSvSsGtHC1j/FW6j1eGOqX3oRsicj4GQI1k7qXKkTVQlSJrT0g1rwarxGnbSGZa12U1+7CW0FbpPl6j61Q5pdaR3oduiMj5GABdyIxA5cUXh5tZLllDr13BtJr5qlmXRi1XpbfdFP5Npvt4S60jvQ/dkH48+SNZMAC6kBmNjBdfHG5muWQNvXYF02rmWxhi+vv7FRvWwcFB9PX1oaurS/W7gtXSUna76nqp+laqPAwp5pH15I+8x/MB0G0HusLlAVCyQdRK1sBiNTN7kmRg13auZr65ddnf329Lw+qEfSO3PuLx+KyfS2FIMY8T6gt5g+cDoNsOdIXLA8CwZZM1sFjNbfWlmF3b2Yj5lmtY9bwrWEmphyacUBe01F2GFPM4pb6Q+3k+ALrtQKe0PG5ZNhm4pb64recbKN+waml0y60bJz80oaXuMqQQuZ/nA6DbDnTFy+OmZZOBW+qL23syq1Fu3Sg9NOGUMG1F3XXKuiAiBkDSgAd393BLT2aOkXWz3LpRClG5ew/1vofYTbx0YsHjITkdAyCp5qWDu5vJ1HAZVRYj62ZxyKtURrXvIZZpvZdTqZzl/u62E4tyeDwkp2MA1MEpB3Kjeeng7mbFDZfW+mxk/TeqEbVjzMZYLIaBgQF0dHSgs7Oz4nuInRIYKpWz3N/dcouEGjwektMxAOpgxIHcyhDp9qFLSJtIJIJUKoVUKpWvG1rqs5FBRs0YfmpYMWZjcRkHBgbQ19cHAGhubq64j5kZGHJhdP78+aivr69qX1cqp2yvsJMBj4fkdAyAOhhxALSyN8ApPQ+kjd5gHwqFEAwGEYvFZvVYqa3PRgYAu8fwU1K8XkuVsaOjAwDQ0dGhah/TGxjUbOdcGI3H45g3b17ZclSiVE67XmFHROZhANTBqDHLCnthzLpnKJlMIpVKIRwOe/6M3W2MeHuGnnHszOj5kKlXqfCtIe3t7bN6vgr32ebmZixevBjNzc3579r12sBcGC3sATSSTNtHZl69PYiciQHQJrlemL6+PoyOjqK3t7fsWbfeg0k8HkcikUA0GuUBySR2HfSNeHuGLGQrT05xz1dhzykAS3rF1GznaDSKaDRqyvwBebePbHi1hZyEAVAFsxr4SCSCoaEhpNNpxONxU56o45m7+ew66LNR1qfS/tza2or58+ejtbU137tXvB8VPvlbrhffqHJ2dnYaOm0yB4+35CQMgCqY1cCHQiH09vbOendv8d9LzU9tKGVIUE9v0OdB31kq7c+ZTAZCCGQyGcWB1Qt/LuwRNHo/s+PEws4nwt2Ax1tyEgZAFcxs4EsdMAqHmFC6tMNLDcbTu0550HeWSvuzlv3dzGND8bRLhS07h+Up9XkGQyL5MQCqYEcDXzjEhFIA1Nvw8MBcGnvyvKHS/qxlfzfz2FA87cKwlfs5EomU/H25cpU6Dhj1RDhPUInkxwAoqcIhJpTobXicNoZhtbSUlT15pJcV+0Rh2Crcj0v9vlw5Sn2ueB+otFyl9hmeTBHJjwFQUmY91ee0MQyr5aSyepmTTiqUGFXPyq0HpbBVahifSvu32uOArLdFmFVfjBp6y8l1mbyDAdBByh1YkskkBgcHAQDt7e2GD0ZbyEln92aUlQd44zk9qBtVz9Suh1L7sdr9W83nZB5D1Kz6YtTQW06uy+QdDIAOUu7AEo/H8/cMmvFEYiE7L5VqDV9mlJUHeOM56aRCSWE9q+YEQc96MOvhEJnHEDWrvnDoLfISBkAHKXdgiUQi6OrqQjqdNm1cMhnIEL54gDeem+6/LHyTSE9Pj6bv6lkPpfaJavcVmeu5WfXFiOm6qS6TuzEAOki5A0soFEJPT0/+faVm9wLaxahGqZreEScd4Hm52v1K7RPV7itaHwghImdhAHQZmc/ajaA1fJVqtGToSbSCV5ZTJu3t7QgGg5btg7ntGo/H878z4w0iHPOPyF0YAF3GSb1TVijVaLk9KOd4ZTmtpHdoFDMVjwVoRujnmH/6MSSTjBgAydVKNVpeCcpeWU4ryRh4lOq50aGfY/7pJ2OdIWIAdBieSWrDAERGkzHwFNdzK+s897HKZKwzRAyAklAb7HgmSWSv4nvuzN4PrXgHsNsZva5kGI6KqFoMgJJQG+x4Jklkv9y4m0NDQ+jt7TXt7R+5eakZ5kWWQKi2HFaW1+gTZ56IkxswAEpCbbDjmSSR/SKRCIaGhpBOpxGPx6vaJys9Xev3+xGNRisO8yJLKKn2fcRmMPrEmSfi5AYMgJJgsCNyjlAohN7e3nwPVjUqPV0bjUYVh3MpPmbIEkrUlsPK8hp9fOXxmtyAAZCISAejQoBRT9fmppNMJtHf32/bpWAj30dMROZhACQikpDegGTXpWA99/TJct8ikRcxABIRuYhdl4L1BE9Z7lsk8iIGQCIii1jR42XXpVU9wTMSiSCVSiGVSiGZTDIEElmIAZCIyALJZBJbt25FOp0GYG+PlxlBVE/wDIVCCAaDiMViCAaDhrznm4jUqbG7AEREXhCPx5FOpxEIBGx/UndwcBBbtmzB4OBgyc/kHiZJJpOGzLPU9CKRiOIwN5XkLh/nBuQmIm3YA0hEZIHCS6RO6LGyavBkPT2HyWQSqVQK4XDY9jBN5FQMgEQV8FITGUFr0DGj3uWm2draimOOOQaRSKTkfGQePDkejyORSCAajXKfJNKJAZAcy6pgxicVyQ5m1DulwaX7+/sN65krx8jp6Q2TZh4zeKJITsMASI5lVTCT5Q0L5C3l6p1S2FATQJSmqfQ7q8KM3vlUM0aiUe9wVpo2TxTJSRgAybGsCmZ8YwHZoVy9UwobagKI0jSVfmdmmCkMfXrnozc4GvkOZ6VpF/5LJDsGQHIsM4MZL+eQzNT25Olh9gMWhaFPb5kr9eSV2n+NfIdzMZ4oktMwABIp4OUcMosRJxdqe/L0MPsBi+KnofXMo7gnD8CsdTo4OIi+vj50dXWhp6dn1ncZ1IhmMAASKeDlHDJaLvilUikkEgkA6k8urOyRNrvuGxHAinvyeMJGpB0DIJEC9hKQ0XIhJRwOVxz4uDjwWRlw7Kj7egKuUjlz67S9vR3BYFD1AzREXsQASGQQNixUjtqBoJVeGWfFsCd21l+tAbe4rMVhUOsDNERexABItshmBfaOpDGemUKTvw5LWwOoqfHZXayqsGGhctT2rCm9Mi733dzr1NQO/6KlThpdf9UEytxn/H6/ptfBVVNW3t5BNIMBkCy3eziJBzbvxyuvj2FiahqNdbU4amEz1h3bhu5F9gQnI3o/2LCQEcr1FBYHH6XeQkBfsDK6/qoJaYWXxYPB4Ky/lRvrUGtgLMTbO4hmMACSpXYPJ7HhyT2Ij2cQDTci6A8glZnC5n0J7EukccXaI2wJgbmGKJVK5e8d0tpIsGEhI5Tr7SsMabnwNzIygtbW1llhSOmNH2rnaxQ1gTL3t1QqpWpcQy3LxVsyiMpjACTLZLMCD2zej/h4BkcvaobPN3PJN9RYj+aGOuwaHsODW/bjyAXNll8OLtcQEdlBKQAVhsPC8Fc8Fl6p8GVlKFITKAuXp/ihjWrHOlTTA8mQSF7GAEiW2TuSxiuvjyEaboTP50Mmczh/r5Pf34BouBG7h8ewdySNZZFg5QkaqFxDRGSHcmEnd59gYfhTejBC6XtWnuCoDVhqxzXU0kupJizyvl3yMgZAssx4ZgoTU9MI+gMAgHQ6jWRyDADg9zcg4K/F/tEJjGembCsjL+OSLMrVxeL7BEvdC1jue1awM2Cp2Zd53y55GQMgWabJX4fGulqkMlMINdYjEJgJgrl/05lpNNTVosnPaklUTnG4UXpyWM33csy6FKomYNl5GZYnfORlbGnJMktbAzhqYTM270uguaEOfn8D/P4GAIAQArHEBFYtDWNpa8DmkhLJQW04UjvGYCnlXp1WDTUBy+xeQt7nR6SMAZAsU1Pjw7pj27Avkcau4Zl7AQP+WqQz04glJhBp8uNdx7Q5fjxAIqN44R41sy/DemEdEunBAEiW6l4UwhVrj8iPA7h/dAINdbVYtTSMdx1j3ziARDJSG46qDTnlXp1mNrMvw/I+PyJlDIBkue5FIRx5RrPr3gRCZDSlcKR0SbPakKM2hDnxcirv8yNSxgBItqip8Wka6qVSw+PEholIj3LjA5ZjxD7ihcupPJaQVzAAkiNUani80DARAXPfBqI2rBixj8j+VK8ReCwhr2AAJGmUazgqNTy8z4e8orC3r7+/XzGsGHmZWM0A04WcHqB4LCGvYAAkaZRrOCo1PLzPh7yoVFjRe5lYidZA5/QAxWMJeQUDIEnD6IbD6ZeiiCopfIVhf39/vq5r3Zeq6X0vVSYikhsDIEnD6IbD6ZeiiNQqrutan+pNpVJIJBKzpld4yZf7D5H7MACSazn9UhRROYW9dnrrei44hsNhRKNRRCIRnjgReQQDILkWey7ILZQu0RYGtc7OTk11PTc9v9+fD36lLv16CW8bIS9hACRP4QGenEipV66aHu7c9KLRKDo7O2f9Tc+Jk+z7ldrysfeTvIQBkFyt+MDPAzw5kVLYq6aH2+jbI/TuV0YFx0rTUVs+3jZCXsIASK5WfODnAZ6cqJqwpxSOjL49otp7EHNl0qvSdPx+P3w+H/x+f9mwqPbVe0RuwABIrlbcMPG+QPKacuFITbhR8xm9+5VRg1NXmk4mk4EQAplMZtb6AFBx2XjVgNyKAZBcjYGPvK5cOBocHERfXx+6urrQ09MDwNjbJiqFx2oGp+7r68PQ0BB6e3srTkdpHah94plXDcitGADJs3hph7xAa8gqDldGPGySK4dRIpEIhoaGkE6nEY/HK067eB2EggFgdBCY98ayzZsHHNoDtLQDtXVlv0vkFgyA5Fm8tENe197ejmAwOKdnrDBcaR1iplC58FjNCVgoFEJvb2/++5pMTwEv/woY2IRQz3qEus8Edj8MbLsX6DgVWHX+nBBI5Eas5eRZans22FNIbqXUu1VVuFIx/Ryt9yYW/65w2qr30Vz42/F7oNYPvHgnMNIP9G8C4Jv5PcAQSJ7AGk6uoTWoqb20w55C8hIzTniUplnuBExpn1P6ndKr7MqWeXQQGNg0E/7mHwUkY8DO+4FQdOa/g68AA08BnacC8zpLT4fIBRgAyTXMvN+o8F8iN1OzH2l9elhpmuVOwEo9tJH7tzj4Fb7KrqzWTqDnPTM9f8kYDjdEkJ5uQKChCQ3JN54M7nkP0NrBnn9yPQZAko7eA69ZQY03gZNbae2Zy1ETEgs/ozRNrePxFf6uv79/zjuMVQ1x4/MB3WcBIwPAzvuRnm7A2FgKANCQigHLzwG6zwR8Pvb8k+sxAJJ09B54GdRIdrL1KmntmcspDnSVgqTSNLXs5+XG/dN0j6EQMw989G8CQlEEGpoAAIFgE1AbBfqfmukl7D6TPf/kegyAJB0eeMmtjOhVMjJERiIRpFIppFIpJJPJfBkrTbs40OkJkoX7eTYrsHckjfHMFDB5GA3T41gwfz5CoRCSySS2bt2KkZERtLa2qhr3r3j6eSP9M0/7wofDDRGk9r+CuvQB+BZ0ApHOmXsAt90LLDgaoXn6n34mcgIGQJIOe/LIrYw4uTHy0mQoFEIwGMxPb3R0FOl0uuy0S/X2HThwAHv27IHf70c0GkUymcTg4CCAmeFmCqdXOI39aeCBv7yCV14fw8TUNKYOpzGvNoNzjz+M09+yAvF4HOl0GtlsVvW4f7nyz/lcS/vMUC87fo/J/duRGkngUGglusZi8IspYDoDdJwJtCzVvC6JnIYBkIjIIlpObkoNhZJKpRAOhw3rIc9NJ5VKIZ1OIxAIlJx2rjeuOCSGQiFMTU3htddeQyqVQnNzc35AaQAIBoOKPYZ7Dk3gni0jiB0cRdeiMJYsCCMxXoOBAwL/u2McS5ck0fZGWfx+P0ZGRvK9lbrCb23dzBAvAPyv/AHT7e9C7eKT4ZvaBex5eCb8cQgY8gjWciIiCZUaCiWRSCAajRr2WrZcKE0mk/lBocvdV1cqJHZ0dGBoaAhjY2MYHBxEe3s7urq6AMzt8YxEIsgKgXu2jyN2cBRNU0lMTdSitqYVkVAT5jUHsWt4DA9u2Y+PnH5UvjyZTAaxWGxOoFSznHlvhEB/56noaO2YeTBE9ADL3jLT88fwRx7Bmk5EJKFcaJqcnMTTTz+Njo4Oy17LVipElXv4IhqNYvny5flev1AolH+/cLFQKISRyVoMJnaia1EYUxO1CASCSCRGEAgEAADNNRls23sIe0fSWBYJzpl/qTKrGhOwtm72OH8+H8f9I89hACQiMki1D2gove3i6aefzoeqt73tbfneuv7+ftXzqXTpuPjSLgDEYjGkUqmKvYKFlF4tV8p4ZgoTU9NYsiCM6akAXn/9dUxOTmFychKHDx/GRGYSYwjOPBjyBjVvFlE9JiCRxzEAEhEZpNzbKtQMmqzUe9XR0THrXz3zqXTpuNSl3VQqNWs+lXoRtdzj2OSvQ2NdLVKZKWQPpzE5OYX6+pkmaXJyCllfLUKBRjT51TVTaoaGIaI3MQASERlE6RKllkGTlXqvotEootFoVfOpNG6fUngqvi+weDrV9nYubQ3gqIXNeGHgIBb6J9HQ0ID6+jo0Njairq4O+8aBY5fOw9LWgKrpcfQAIm0YAImIDKIUQkrdt1YYoLT2XmmZj9Lni8NiqfBU/Hult3Hkfq9VTY0P645tw47BYbxyIIX5gRr4J6cwnsni0GGgrbUJ7zqmDTU1PvUTnZ6aed9vayfg8yE5OorRgZfRsqwXofC8/MdkG5CbyA4MgEREJioVrgpDWGdnZ/53ue8YNZ9iRg0lUzyItJoHS4rHBexeFMKVpx2F+154Da8lMhjJTKNmOotowzTevbwJ3Ys0rIfpKeDlXwEDm4Ce9UD3mUi9fC/qtv0Wh/e/HaEzPpp/wpeveSNiACQiskVxj51VoUTpfkA9PWKFg0iXG5alcL5K4wIe39WGVZ2L5rwJpLGhQf2DLrnwt+P3QK0fePFOYKQfkdf+gMN+HxqGNgEvL8iP8ce3DRExAJLL8NIOOUVxj51VoUSp505v+FRzeRuYCX+Tk5NoampCc3PznM/X1PiwLBJ843spRObPx+DgIPr6+tDV1VVyOJm80cGZnr9aPzD/KCAZA3bej/pQFPVt0ZlXvA08BXSeCrzxijceH8jrGADJVXhph5yqMJSYeSKj1HOnJnwqhbrc/YvFl64L90NgZkgZn88HIQSmpqYwNjamuHzF3wOAdDqt2BM4ax21dgI975np+UvGcLghgvR0AwINTWhIvjG9nvcArR0gohkMgOQqvLRDTlQc+Mw+kSneT9T0iCmFuuL/L3w3cOG/wMyr3AYGBpBOp7Fz506Mj4/P6d0r/F4kEkEwGJwzFI1SeUKhENB9FjAyAOy8H+npBoyNpQAADakYsPwcoPvMmQGfiQgAAyC5DC/tkFMUhr7iMGP2iYya3sZsVuTvy2vy16F13rw5ZVL6/1LTi8fj6OjoQCaTwYEDBzA+Pj6rTIUPiRQOhF08FE3x/CKRCCAEsPthoH8TEIoi0NAEAAgEm4DaKND/1MyTwQyBRHkMgERENigMfXp65MwoR26eu4eTeGDzfrzy+hgmpqbRWFeLoxY2Y92xbbPGCcwpdxm3cIDraDSKzs5ORCIRLFiwYM44hkoPiagaoubQHmDbvQB8QCiKhmRspuevNgqE3rgHcNu9wIKj+co3ojcwABIR2aB47D8re65LjUEIzIS/DU/uQXw8g2i4EUF/AKnMFDbvS2BfIo0r1h5RdniWwumVGuC61DiGXV1ds6ahWks70HEqsOP3mNy/HYcPZ1B35Blo3P/cTPibzgAdZwItS7VNl8jFGADJVmpvdufTveQ2dt6uUDwGYa4c2azAA5v3Iz6ewdGLmuF743JpqLEezQ112DU8hge37MeRC5oVB2jO7ad+vx+Dg4NIp9MIh8P5cf/KCYVClZ/2LaW2bmaIFwATWx/CYORUBJf9NToXL5/p+es4Mz8EDBHN4N5AtlJ7szuf7iW3suPkptQ9hjv3HsCLe4bQ1toEn8+HTOYw0uk0amtrMT09jQXBeuweHsPekTSWRYJzppvbTzOZDPbt24fGxkasXr3a8OVSXGdvhEDfguMRnAoiMn8+0NwJLFg+0/PH8Ec0C/cIspXam935dC+5lR0nN6V6H/e9HkdiLIW25noATUin00gmx+DzzTxnEWxqwuGpupkBm1H6ncIHDx5EY2Mj5s2bB7/fP2cYl2p7/kuus9o6NLf3orlwIrznj0gRAyDZSu1lMDMvl/HyMqlhVj2R6eRmycIIws2vA3V+AEAgEACAfA/gdE09GoQPTX7lV6rl1svBgwcRiUSwZMmS/NAvuc8ofa+UOZ97412/81rn4cDhA5jX2jrzAEhLe8kevkrbjfs/eRUDIHkeLy+TGmbVE5mGLlq+dAGOPyKBzfsSaPJPYGJiAoFAAH5/A4QQ2DU8hqMXNGJqdBjJ+vmK4TUej2NgYAATExMAgGw2i0AgoDhkTKWhY2a9tSRxCKG++4GBTYh1nIhHMkNoTm1Gy8AzMw+AlLjHL/d08dDQEHp7e+esazXblSGR3IgBkDxPph4Ykpds9cSMUFJT48O6Y9uwL5HGtn0jCOIwFrQK1GdrEEtMINLkxwmL67B/aAg1Pt+sB0hyIpEI5s2bh0OHDqG5uRnz58+fU8bi0FsqhOXeWrLnld2o33oXGie2oNYfxKatv8DOeoFNkz4cXR9FzY7fz3xBIQRGIhEMDQ0hnU4jHo8rlrfwXyU8SSQ3YgAkz5OpB4bkJUs9yQW/3Nh6gLGhpHtRCFesPQL3PjeA7fsOYX8aaBaTWLU0jHcd04a2ABCPN5YMTKFQCGvWrMk/DZzJZCouTyqVQjgcRiQSUbyv8NCrSQRffwGHg3XY2zIf2w/uwbLUYWyvb8CulvlYkYzPetdvcXl6e3tnvcau+O+V1p9s4Z/ICAyAREQOUmpsPSN1Lwrh2nf1znoTSEvdFEYOxYFABJ2d5R+syIWq/v7+spdfc8uTGyQ6953i+wqPOuGdSPnHUD/wIDYlduBwTT3a/CEc8k1gU2IHjvYtQE2Zd/1WG95lCf9ERmIAJCJykOIBpM1SU+ObNdRLcTArVO4evnKXX3OfUfp31tPDLS0InXoRdmQHsf3Ve7GwphmZ7CQW1Ndh++HXsWtJL1Z0nc7XvBFpwABIJAHeZE5qleuNMrMelboMmkwmsXXr1vyTvgBmlaHc5Vdg7vIU9h7mA2dzM7K7HsSm1x7HZHYK4cxBTPlbUONfiEOTB7Fp9304eroGNWd9Bairz5eL+xRRaQyARBLgTeZkBDPrUangGY/HkU6n80/6Kg0No7YsJV9RN9KPXZt/ju0TryM6NYWabBb+8f1AdgKLpyawvVZg1857sSK0GDj5o0BtHfcpogoYAIkkwJvMyQjV1CO9PWZ+vx+BQAAdHR1zLv9qmWZxT+KsV9RNNWCTvxaZqcMITAtk6+qBhiCQTiBQW4vJ2lpsagri6Ne3o2Z0LzCvk/sUUQUMgERVMuJSE28ydyerL0NWU4/09phlMhkIIfJP+xaWodx9g0rzL+xJLBSbeB17I53wjb2GV5MxwN+IbEMLsjU1qJ1Kw1dbj711AcSOOh1L33gQhPsUUXkMgORYstzjo7bhlKW8ZB0nXYbU22NW+L1Sr4arNM3cUDBtbW1ob2+fs67agm24sOdDmF7+PuD524Ghl5Csa0f6cBpNwSyapg6htvMdaOv9Wz4IQqQSAyA5liyNq9pGTpbyknWcdBlSb49ZuR4/tdMsHgqmUC5Uts1bhNCBp4HEfiB8JA43RJCuHUcg2ISGw3Fg/07g1ceB7jPnhECefBHNxQBIjmVU41pt46C2kVNTXjZU7uK1y5BG9CIWy73K7dCrz6M3fj/88AGhKBqSMTSkYkBtFAhFgYOvANvuBRYcPWcwaJ58Ec3FAEiOZVTjalXjoKa8bKjIyYzoRSyWG0swMT6N0dZjsODAX2bCHgAsPwfo3zTz83QG6DgTaFmqOI38O4WTSe5bRGAAJJLqMp1MZSH5yNRDbFVZCscSbGhdA/QtmHntW897Zi73tnbM9Px1nKn4LuDcNILBIGKxGILBoO3rjkgGDIDkeWZeptPaSHrtkiFpI1MPsZVlmbVfrDp/5p2/rR0z9/p1nwUsWD7T86cQ/nJ4ckU0GwMgkYlkarBJPlpPEGQKMbaVpbZu9j1+Pt+ce/6IqDIGQCITydRgk3y0niDI1ENsRFmsuIycTCYxODiI4eFhCCEA8GSMCGAAJDKVTA02ycfrJwhW9JDnniLOZDKIRqOeXddExRgAiYhs4vUTBCsCcCQSQVdXFwAoDjJN5FUMgEQ2kOlpTiK9rBpDsxqhUAg9PT2mzoPIiRgAiYpYEc74cAi5AesxkXMxABIVsaJR8/q9X+QOrMdEzsUASFTEikbN6/d+kTuwHhM5FwMgURG7GzXeH0hERGZjACSSDO+rIiIiszEAEkmG91UREZHZGACJJGP3JWgiInK/GrsLQERERETWYgAkz0omk+jv70cymTTl80RERLLiJWDyLK0PW/DhDCIicgsGQPIsrQ9b8OEMIiJyCwZA8iytD1vw4QySAceJJCIjMAASETkIb0UgIiMwABIROQhvRSAiIzAAEtmEl/JID96KQERGYAAksgkv5RERkV0YAIlswkt5RERkFwZAIpvwUh4REdmFbwIhIiIi8hgGQCIV+Bo4IiJyEwZAoiJKYS/3wEY8HrexZJUxqBIRkRq8B5CoiNLTuU55YINPFhMRkRoMgERFlMKeUx7YcEpQJSIiezEAEhVxSthT4uSyExGRdXgPIJEKvLeOiIjchD2ARCrw3joiInITBkAiFXhvHRERuQkDIJEKvLeOiIjchPcAEpXA+/6IiMit2ANIVALv+yMiIrdiACQqgff9ERGRWzEAEpXA+/6IiMiteA8gkUq8J5CIiNyCPYBEKvGeQCIicgsGQCKVeE8gERG5BQMgEWYu78bjcUQikZK9e7wnkIiI3IIBkAi8vEtERN7CAEgEXt4lIiJvYQAkgjMu76q5TE1ERKQGh4EhcojcZep4PF7VdDicDRERsQeQyCGKL1Pr7RHk/Y5ERMQASOQQxZepc0EulUohGAyqDoK835GIiBgAiSSjtmcvF+BSqZSmHj0n3O9IRETmYgAkkozaS7S5IJdMJvM9gERERGowABJJRuslWvboERGRVgyARJJhoCMiIrNxGBgiD+EQMEREBLAHkMhTOAQMEREBDIBEnsIhYIiICGAAJPIU3l9IREQA7wEkIiIi8hwGQCKbaX0wgw9yEBFRtXgJmMhmWh/M4IMcRERULQZAIptpfTCDD3IQEVG1GACJbKb1wQw+yEFERNXiPYBEREREHsMASEREROQxDIBEREREHsMASOQRHD6GiIhy+BAIkUdw+BgiIsphACTyCA4fQ0REOQyARB7B4WOIiCiH9wASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHHMAASEREReQwDIBEREZHH1Kn5kBACADA6OmpqYYiIiIhIn1xOy+W2clQFwGQyCQBYtmxZFcUiIiIiIrMlk0mEw+Gyn/EJFTExm81i3759CIVC8Pl8hhWQiIiIiIwhhEAymcSSJUtQU1P+Lj9VAZCIiIiI3IMPgRARERF5DAMgERERkccwABIRERF5DAMgERERkccwABIRERF5DAMgERERkccwABIRERF5zP8HaWJCJ8YPgRMAAAAASUVORK5CYII=", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "fig, ax = plt.subplots(figsize=(8,8))\n", "ax.scatter(*clusterable_vectors[layers[3].cluster_labels == 8].T, s=1, c=\"gray\", alpha=0.25)\n", "ax.scatter(*clusterable_vectors[fl_indices[8]].T, marker=\"o\", label=\"Facility Location\", alpha=0.5)\n", "ax.scatter(*clusterable_vectors[sc_indices[8]].T, marker=\"X\", label=\"Saturated Coverage\", alpha=0.5)\n", "ax.scatter(*clusterable_vectors[d_indices[8]].T, marker=\"^\", label=\"Diverse Central Selection\", alpha=0.5)\n", "ax.axis('equal')\n", "ax.legend()\n", "ax.set(xticks=[], yticks=[])" ] }, { "cell_type": "markdown", "id": "bad276b1-b432-49a8-b62a-2d51599f20c7", "metadata": {}, "source": [ "The general trend seems pretty clear. So, when selection what exemplar method to use you should consider what you most want from your exemplars: a coherent central topic that is easier for the LLM to label, but perhaps not as representative of the full diversity of the cluster? Then central selection or saturated coverage should work well. If you want to ensure the diversity of the cluster makes it to the LLM then facility location is a good choice. Lastly you may need to alter your choice depending on your computation time and scaling needs. Hopefully this walkthrough has provided enough information to help you make informed choices." ] } ], "metadata": { "kernelspec": { "display_name": "toponymy_docs", "language": "python", "name": "toponymy_docs" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.12.8" } }, "nbformat": 4, "nbformat_minor": 5 }