15_guest/15_02_lsa.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SVD and Latent Semantic Analysis (LSA)\n",
    "\n",
    "Singular Value Decomposition (SVD) applied to text documents is often referred to as _Latent Semantic Analysis_ (LSA).\n",
    "\n",
    "Let's $A$ be our $N \\times n$ feature matrix with $N$ samples and $n$ features, and let's write the SVD as follows, trunacted to some dimension $k < n$:\n",
    "\n",
    "\\begin{align}\n",
    "A = U \\Sigma V^{T}\n",
    "\\end{align}\n",
    "\n",
    "Each of these matrices represents some grouping the data:\n",
    "- $U$ ($N \\times k$) represents how each datapoint relates to some latent concept,\n",
    "- $V$ ($n \\times k$) represents how each concept relates to each original feature ($V^T$ would be $k \\times n$) \n",
    "- $\\Sigma$ ($k \\times k$) represents the strength of each of those concepts in the dataset.\n",
    "\n",
    "This reduces the number of dimensions from $n$ to $k$, since instead of modelling our data for each feature, we now model our data for each concept. Our intuition is that datapoints with similar relevant features will get similar concepts.\n",
    "\n",
    "A very concrete example of this is text classification, in which case this method is often called _Latent Semantic Analysis_ (LSA).  As always with Natural Language Processing (NLP), we transform our $N$ documents into a $N \\times n$ feature matrix, where each column represents a certain word or n-gram (e.g., using `CountVectorizer`).  \n",
    "\n",
    "This notebook demonstrates SVD and LSA with a simple example.\n",
    "\n",
    "\n",
    "This example is over text, but really, you could even use this for **recommendation systems!**. If you had a matrix of users x items, where each entry is the # that they purchase, you could apply the same technique."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Import code and data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.datasets import fetch_20newsgroups\n",
    "from sklearn.decomposition import TruncatedSVD\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer, CountVectorizer\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import Normalizer\n",
    "import seaborn\n",
    "from matplotlib import pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 4 true labels\n"
     ]
    }
   ],
   "source": [
    "categories = ['alt.atheism', 'talk.religion.misc', 'comp.graphics', 'sci.space',]\n",
    "dataset = fetch_20newsgroups(subset='all', categories=categories, shuffle=True)\n",
    "\n",
    "documents = dataset.data\n",
    "y = dataset.target\n",
    "\n",
    "true_k = np.unique(y).shape[0]\n",
    "\n",
    "print \"There are %d true labels\" % (true_k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[u'From: healta@saturn.wwc.edu (Tammy R Healy)\\nSubject: Re: who are we to judge, Bobby?\\nLines: 38\\nOrganization: Walla Walla College\\nLines: 38\\n\\nIn article <1993Apr14.213356.22176@ultb.isc.rit.edu> snm6394@ultb.isc.rit.edu (S.N. Mozumder ) writes:\\n>From: snm6394@ultb.isc.rit.edu (S.N. Mozumder )\\n>Subject: Re: who are we to judge, Bobby?\\n>Date: Wed, 14 Apr 1993 21:33:56 GMT\\n>In article <healta.56.734556346@saturn.wwc.edu> healta@saturn.wwc.edu (TAMMY R HEALY) writes:\\n>>Bobby,\\n>>\\n>>I would like to take the liberty to quote from a Christian writer named \\n>>Ellen G. White.  I hope that what she said will help you to edit your \\n>>remarks in this group in the future.\\n>>\\n>>\"Do not set yourself as a standard.  Do not make your opinions, your views \\n>>of duty, your interpretations of scripture, a criterion for others and in \\n>>your heart condemn them if they do not come up to your ideal.\"\\n>>                         Thoughts Fromthe Mount of Blessing p. 124\\n>>\\n>>I hope quoting this doesn\\'t make the atheists gag, but I think Ellen White \\n>>put it better than I could.\\n>> \\n>>Tammy\\n>\\n>Point?\\n>\\n>Peace,\\n>\\n>Bobby Mozumder\\n>\\nMy point is that you set up your views as the only way to believe.  Saying \\nthat all eveil in this world is caused by atheism is ridiculous and \\ncounterproductive to dialogue in this newsgroups.  I see in your posts a \\nspirit of condemnation of the atheists in this newsgroup bacause they don\\'\\nt believe exactly as you do.  If you\\'re here to try to convert the atheists \\nhere, you\\'re failing miserably.  Who wants to be in position of constantly \\ndefending themselves agaist insulting attacks, like you seem to like to do?!\\nI\\'m sorry you\\'re so blind that you didn\\'t get the messgae in the quote, \\neveryone else has seemed to.\\n\\nTammy\\n',\n",
       " u\"From: jk87377@lehtori.cc.tut.fi (Kouhia Juhana)\\nSubject: Re: More gray levels out of the screen\\nOrganization: Tampere University of Technology\\nLines: 21\\nDistribution: inet\\nNNTP-Posting-Host: cc.tut.fi\\n\\nIn article <1993Apr6.011605.909@cis.uab.edu> sloan@cis.uab.edu\\n(Kenneth Sloan) writes:\\n>\\n>Why didn't you create 8 grey-level images, and display them for\\n>1,2,4,8,16,32,64,128... time slices?\\n\\nBy '8 grey level images' you mean 8 items of 1bit images?\\nIt does work(!), but it doesn't work if you have more than 1bit\\nin your screen and if the screen intensity is non-linear.\\n\\nWith 2 bit per pixel; there could be 1*c_1 + 4*c_2 timing,\\nthis gives 16 levels, but they are linear if screen intensity is\\nlinear.\\nWith 1*c_1 + 2*c_2 it works, but we have to find the best\\ncompinations -- there's 10 levels, but 16 choises; best 10 must be\\nchosen. Different compinations for the same level, varies a bit, but\\nthe levels keeps their order.\\n\\nReaders should verify what I wrote... :-)\\n\\nJuhana Kouhia\\n\"]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "documents[0:2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3387, 2189)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Feature extraction\n",
    "Extract features from the training dataset using a sparse vectorizer."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dimensions feature matrix: (3387, 2189)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[u'00',\n",
       " u'000',\n",
       " u'01',\n",
       " u'02',\n",
       " u'04',\n",
       " u'08',\n",
       " u'10',\n",
       " u'100',\n",
       " u'1000',\n",
       " u'11',\n",
       " u'12',\n",
       " u'128',\n",
       " u'129',\n",
       " u'13',\n",
       " u'14',\n",
       " u'15',\n",
       " u'150',\n",
       " u'16',\n",
       " u'17',\n",
       " u'18',\n",
       " u'19',\n",
       " u'1988',\n",
       " u'1989',\n",
       " u'1990',\n",
       " u'1991',\n",
       " u'1992',\n",
       " u'1993',\n",
       " u'1993apr15',\n",
       " u'1993apr19',\n",
       " u'1993apr20',\n",
       " u'1993apr21',\n",
       " u'1993apr22',\n",
       " u'1993apr5',\n",
       " u'20',\n",
       " u'200',\n",
       " u'2000',\n",
       " u'21',\n",
       " u'22',\n",
       " u'23',\n",
       " u'24',\n",
       " u'25',\n",
       " u'256',\n",
       " u'26',\n",
       " u'27',\n",
       " u'28',\n",
       " u'29',\n",
       " u'30',\n",
       " u'300',\n",
       " u'31',\n",
       " u'32',\n",
       " u'33',\n",
       " u'34',\n",
       " u'35',\n",
       " u'3539',\n",
       " u'36',\n",
       " u'37',\n",
       " u'38',\n",
       " u'386',\n",
       " u'39',\n",
       " u'3d',\n",
       " u'3rd',\n",
       " u'40',\n",
       " u'408',\n",
       " u'41',\n",
       " u'42',\n",
       " u'43',\n",
       " u'44',\n",
       " u'45',\n",
       " u'46',\n",
       " u'47',\n",
       " u'48',\n",
       " u'49',\n",
       " u'50',\n",
       " u'500',\n",
       " u'51',\n",
       " u'52',\n",
       " u'53',\n",
       " u'54',\n",
       " u'55',\n",
       " u'56',\n",
       " u'575',\n",
       " u'59',\n",
       " u'60',\n",
       " u'600',\n",
       " u'602',\n",
       " u'61',\n",
       " u'621',\n",
       " u'64',\n",
       " u'65',\n",
       " u'70',\n",
       " u'75',\n",
       " u'80',\n",
       " u'800',\n",
       " u'88',\n",
       " u'90',\n",
       " u'92',\n",
       " u'93',\n",
       " u'95',\n",
       " u'9615',\n",
       " u'__',\n",
       " u'___',\n",
       " u'____',\n",
       " u'_____',\n",
       " u'_the',\n",
       " u'ability',\n",
       " u'able',\n",
       " u'absolute',\n",
       " u'absolutely',\n",
       " u'absurd',\n",
       " u'ac',\n",
       " u'acad3',\n",
       " u'accept',\n",
       " u'acceptable',\n",
       " u'accepted',\n",
       " u'access',\n",
       " u'according',\n",
       " u'account',\n",
       " u'accurate',\n",
       " u'act',\n",
       " u'action',\n",
       " u'actions',\n",
       " u'active',\n",
       " u'activities',\n",
       " u'activity',\n",
       " u'acts',\n",
       " u'actual',\n",
       " u'actually',\n",
       " u'ad',\n",
       " u'adam',\n",
       " u'adams',\n",
       " u'add',\n",
       " u'added',\n",
       " u'addition',\n",
       " u'additional',\n",
       " u'address',\n",
       " u'addresses',\n",
       " u'admit',\n",
       " u'adobe',\n",
       " u'advance',\n",
       " u'advanced',\n",
       " u'advantage',\n",
       " u'advice',\n",
       " u'aerospace',\n",
       " u'age',\n",
       " u'agency',\n",
       " u'ago',\n",
       " u'agree',\n",
       " u'agreed',\n",
       " u'ah',\n",
       " u'ahead',\n",
       " u'air',\n",
       " u'aircraft',\n",
       " u'al',\n",
       " u'alaska',\n",
       " u'alexia',\n",
       " u'algorithm',\n",
       " u'algorithms',\n",
       " u'alink',\n",
       " u'alive',\n",
       " u'allan',\n",
       " u'allen',\n",
       " u'allow',\n",
       " u'allowed',\n",
       " u'allowing',\n",
       " u'allows',\n",
       " u'alpha',\n",
       " u'alt',\n",
       " u'alternative',\n",
       " u'altitude',\n",
       " u'amateur',\n",
       " u'america',\n",
       " u'american',\n",
       " u'americans',\n",
       " u'ames',\n",
       " u'amiga',\n",
       " u'amusing',\n",
       " u'analogy',\n",
       " u'analysis',\n",
       " u'ancient',\n",
       " u'andrew',\n",
       " u'angeles',\n",
       " u'animal',\n",
       " u'animals',\n",
       " u'animation',\n",
       " u'ann',\n",
       " u'anonymous',\n",
       " u'answer',\n",
       " u'answered',\n",
       " u'answers',\n",
       " u'anti',\n",
       " u'anybody',\n",
       " u'ap',\n",
       " u'apart',\n",
       " u'apollo',\n",
       " u'apparently',\n",
       " u'appear',\n",
       " u'appeared',\n",
       " u'appears',\n",
       " u'apple',\n",
       " u'application',\n",
       " u'applications',\n",
       " u'applied',\n",
       " u'apply',\n",
       " u'appreciate',\n",
       " u'appreciated',\n",
       " u'approach',\n",
       " u'appropriate',\n",
       " u'approximately',\n",
       " u'apr',\n",
       " u'april',\n",
       " u'arbitrary',\n",
       " u'arc',\n",
       " u'archive',\n",
       " u'area',\n",
       " u'areas',\n",
       " u'aren',\n",
       " u'argue',\n",
       " u'arguing',\n",
       " u'argument',\n",
       " u'arguments',\n",
       " u'arizona',\n",
       " u'array',\n",
       " u'arromdee',\n",
       " u'art',\n",
       " u'article',\n",
       " u'articles',\n",
       " u'asd',\n",
       " u'aside',\n",
       " u'ask',\n",
       " u'asked',\n",
       " u'asking',\n",
       " u'aspects',\n",
       " u'ass',\n",
       " u'assembly',\n",
       " u'assertion',\n",
       " u'associated',\n",
       " u'association',\n",
       " u'assume',\n",
       " u'assumed',\n",
       " u'assuming',\n",
       " u'assumption',\n",
       " u'astro',\n",
       " u'astronomy',\n",
       " u'atheism',\n",
       " u'atheist',\n",
       " u'atheists',\n",
       " u'atlanta',\n",
       " u'atmosphere',\n",
       " u'atoms',\n",
       " u'att',\n",
       " u'attack',\n",
       " u'attempt',\n",
       " u'attempts',\n",
       " u'attention',\n",
       " u'attitude',\n",
       " u'attributed',\n",
       " u'au',\n",
       " u'aurora',\n",
       " u'austin',\n",
       " u'australia',\n",
       " u'author',\n",
       " u'authority',\n",
       " u'authors',\n",
       " u'available',\n",
       " u'average',\n",
       " u'avoid',\n",
       " u'aware',\n",
       " u'away',\n",
       " u'az',\n",
       " u'b64635',\n",
       " u'baalke',\n",
       " u'background',\n",
       " u'bad',\n",
       " u'bank',\n",
       " u'base',\n",
       " u'based',\n",
       " u'basic',\n",
       " u'basically',\n",
       " u'basis',\n",
       " u'batman',\n",
       " u'bay',\n",
       " u'bbs',\n",
       " u'beam',\n",
       " u'bear',\n",
       " u'bears',\n",
       " u'beauchaine',\n",
       " u'beaverton',\n",
       " u'began',\n",
       " u'begin',\n",
       " u'beginning',\n",
       " u'behavior',\n",
       " u'beings',\n",
       " u'belief',\n",
       " u'beliefs',\n",
       " u'believe',\n",
       " u'believed',\n",
       " u'believers',\n",
       " u'believes',\n",
       " u'believing',\n",
       " u'bell',\n",
       " u'benedikt',\n",
       " u'benefit',\n",
       " u'berkeley',\n",
       " u'best',\n",
       " u'bet',\n",
       " u'better',\n",
       " u'bible',\n",
       " u'biblical',\n",
       " u'big',\n",
       " u'bil',\n",
       " u'billboard',\n",
       " u'billboards',\n",
       " u'billion',\n",
       " u'bit',\n",
       " u'bitnet',\n",
       " u'bits',\n",
       " u'black',\n",
       " u'blew',\n",
       " u'block',\n",
       " u'blood',\n",
       " u'blue',\n",
       " u'bmd',\n",
       " u'bnr',\n",
       " u'board',\n",
       " u'bob',\n",
       " u'bobbe',\n",
       " u'bobby',\n",
       " u'bodies',\n",
       " u'body',\n",
       " u'book',\n",
       " u'books',\n",
       " u'born',\n",
       " u'bother',\n",
       " u'bound',\n",
       " u'box',\n",
       " u'boy',\n",
       " u'brain',\n",
       " u'branch',\n",
       " u'braunschweig',\n",
       " u'break',\n",
       " u'brian',\n",
       " u'bright',\n",
       " u'bring',\n",
       " u'british',\n",
       " u'brothers',\n",
       " u'brought',\n",
       " u'bruce',\n",
       " u'bs',\n",
       " u'btw',\n",
       " u'bu',\n",
       " u'budget',\n",
       " u'buffalo',\n",
       " u'bugs',\n",
       " u'build',\n",
       " u'building',\n",
       " u'built',\n",
       " u'bunch',\n",
       " u'buphy',\n",
       " u'burden',\n",
       " u'bureau',\n",
       " u'burn',\n",
       " u'bursters',\n",
       " u'bus',\n",
       " u'business',\n",
       " u'buy',\n",
       " u'ca',\n",
       " u'cacs',\n",
       " u'california',\n",
       " u'called',\n",
       " u'calls',\n",
       " u'calpoly',\n",
       " u'caltech',\n",
       " u'cambridge',\n",
       " u'came',\n",
       " u'canada',\n",
       " u'capabilities',\n",
       " u'capability',\n",
       " u'capable',\n",
       " u'capital',\n",
       " u'card',\n",
       " u'cards',\n",
       " u'care',\n",
       " u'careful',\n",
       " u'carefully',\n",
       " u'carnegie',\n",
       " u'carried',\n",
       " u'carry',\n",
       " u'case',\n",
       " u'cases',\n",
       " u'catholic',\n",
       " u'cause',\n",
       " u'caused',\n",
       " u'causes',\n",
       " u'cb',\n",
       " u'cc',\n",
       " u'cco',\n",
       " u'cd',\n",
       " u'ceccarelli',\n",
       " u'center',\n",
       " u'central',\n",
       " u'centre',\n",
       " u'century',\n",
       " u'certain',\n",
       " u'certainly',\n",
       " u'ch',\n",
       " u'chance',\n",
       " u'change',\n",
       " u'changed',\n",
       " u'changes',\n",
       " u'changing',\n",
       " u'chapter',\n",
       " u'charge',\n",
       " u'charles',\n",
       " u'charley',\n",
       " u'cheap',\n",
       " u'cheaper',\n",
       " u'check',\n",
       " u'checked',\n",
       " u'cheers',\n",
       " u'chemical',\n",
       " u'chemistry',\n",
       " u'child',\n",
       " u'children',\n",
       " u'choice',\n",
       " u'choose',\n",
       " u'chris',\n",
       " u'christ',\n",
       " u'christian',\n",
       " u'christianity',\n",
       " u'christians',\n",
       " u'church',\n",
       " u'circle',\n",
       " u'cis',\n",
       " u'city',\n",
       " u'civil',\n",
       " u'claim',\n",
       " u'claimed',\n",
       " u'claiming',\n",
       " u'claims',\n",
       " u'class',\n",
       " u'clear',\n",
       " u'clearly',\n",
       " u'cleveland',\n",
       " u'clinton',\n",
       " u'close',\n",
       " u'closed',\n",
       " u'closer',\n",
       " u'cmu',\n",
       " u'cobb',\n",
       " u'code',\n",
       " u'cold',\n",
       " u'college',\n",
       " u'collision',\n",
       " u'color',\n",
       " u'colorado',\n",
       " u'colors',\n",
       " u'com',\n",
       " u'come',\n",
       " u'comes',\n",
       " u'comet',\n",
       " u'coming',\n",
       " u'command',\n",
       " u'comment',\n",
       " u'comments',\n",
       " u'commercial',\n",
       " u'committed',\n",
       " u'common',\n",
       " u'communication',\n",
       " u'communications',\n",
       " u'community',\n",
       " u'comp',\n",
       " u'companies',\n",
       " u'company',\n",
       " u'compared',\n",
       " u'complete',\n",
       " u'completely',\n",
       " u'complex',\n",
       " u'complicated',\n",
       " u'compound',\n",
       " u'compression',\n",
       " u'computer',\n",
       " u'computers',\n",
       " u'computing',\n",
       " u'concept',\n",
       " u'concepts',\n",
       " u'concern',\n",
       " u'concerned',\n",
       " u'concerning',\n",
       " u'conclude',\n",
       " u'conclusion',\n",
       " u'conditions',\n",
       " u'conference',\n",
       " u'confused',\n",
       " u'confusing',\n",
       " u'congress',\n",
       " u'connected',\n",
       " u'connection',\n",
       " u'conner',\n",
       " u'consequences',\n",
       " u'consider',\n",
       " u'consideration',\n",
       " u'considered',\n",
       " u'considering',\n",
       " u'consistent',\n",
       " u'constant',\n",
       " u'consultants',\n",
       " u'contact',\n",
       " u'contain',\n",
       " u'contained',\n",
       " u'containing',\n",
       " u'contains',\n",
       " u'content',\n",
       " u'context',\n",
       " u'continue',\n",
       " u'contradiction',\n",
       " u'contrary',\n",
       " u'control',\n",
       " u'conversion',\n",
       " u'convert',\n",
       " u'convince',\n",
       " u'convinced',\n",
       " u'cookamunga',\n",
       " u'copies',\n",
       " u'copy',\n",
       " u'core',\n",
       " u'corp',\n",
       " u'corporation',\n",
       " u'correct',\n",
       " u'correctly',\n",
       " u'cost',\n",
       " u'costs',\n",
       " u'couldn',\n",
       " u'count',\n",
       " u'countries',\n",
       " u'country',\n",
       " u'couple',\n",
       " u'course',\n",
       " u'court',\n",
       " u'cover',\n",
       " u'covered',\n",
       " u'crazy',\n",
       " u'create',\n",
       " u'created',\n",
       " u'creating',\n",
       " u'creation',\n",
       " u'crime',\n",
       " u'critical',\n",
       " u'cross',\n",
       " u'cs',\n",
       " u'cso',\n",
       " u'cult',\n",
       " u'curious',\n",
       " u'current',\n",
       " u'currently',\n",
       " u'cut',\n",
       " u'cuts',\n",
       " u'cwru',\n",
       " u'd012s658',\n",
       " u'dan',\n",
       " u'dangerous',\n",
       " u'daniel',\n",
       " u'darice',\n",
       " u'dark',\n",
       " u'data',\n",
       " u'date',\n",
       " u'dave',\n",
       " u'david',\n",
       " u'davidians',\n",
       " u'day',\n",
       " u'days',\n",
       " u'dbstu1',\n",
       " u'dc',\n",
       " u'dead',\n",
       " u'deal',\n",
       " u'dealing',\n",
       " u'death',\n",
       " u'debate',\n",
       " u'dec',\n",
       " u'decide',\n",
       " u'decided',\n",
       " u'decision',\n",
       " u'deep',\n",
       " u'defense',\n",
       " u'define',\n",
       " u'defined',\n",
       " u'definitely',\n",
       " u'definition',\n",
       " u'degree',\n",
       " u'degrees',\n",
       " u'deleted',\n",
       " u'delta',\n",
       " u'denver',\n",
       " u'department',\n",
       " u'depends',\n",
       " u'dept',\n",
       " u'depth',\n",
       " u'derived',\n",
       " u'described',\n",
       " u'describes',\n",
       " u'description',\n",
       " u'design',\n",
       " u'designed',\n",
       " u'desire',\n",
       " u'despite',\n",
       " u'detailed',\n",
       " u'details',\n",
       " u'determine',\n",
       " u'determined',\n",
       " u'develop',\n",
       " u'developed',\n",
       " u'development',\n",
       " u'device',\n",
       " u'devices',\n",
       " u'dick',\n",
       " u'dictionary',\n",
       " u'did',\n",
       " u'didn',\n",
       " u'die',\n",
       " u'died',\n",
       " u'diego',\n",
       " u'difference',\n",
       " u'differences',\n",
       " u'different',\n",
       " u'difficult',\n",
       " u'digest',\n",
       " u'digex',\n",
       " u'digital',\n",
       " u'direct',\n",
       " u'direction',\n",
       " u'directly',\n",
       " u'directory',\n",
       " u'disagree',\n",
       " u'disclaimer',\n",
       " u'discovered',\n",
       " u'discuss',\n",
       " u'discussed',\n",
       " u'discussing',\n",
       " u'discussion',\n",
       " u'disk',\n",
       " u'display',\n",
       " u'distance',\n",
       " u'distributed',\n",
       " u'distribution',\n",
       " u'divine',\n",
       " u'division',\n",
       " u'doctrine',\n",
       " u'document',\n",
       " u'documentation',\n",
       " u'does',\n",
       " u'doesn',\n",
       " u'doing',\n",
       " u'dollars',\n",
       " u'domain',\n",
       " u'don',\n",
       " u'dos',\n",
       " u'double',\n",
       " u'doubt',\n",
       " u'doug',\n",
       " u'douglas',\n",
       " u'dr',\n",
       " u'draw',\n",
       " u'drawing',\n",
       " u'drawn',\n",
       " u'drink',\n",
       " u'drive',\n",
       " u'driver',\n",
       " u'dseg',\n",
       " u'du',\n",
       " u'dunn',\n",
       " u'dwyer',\n",
       " u'earlier',\n",
       " u'early',\n",
       " u'earth',\n",
       " u'easier',\n",
       " u'easily',\n",
       " u'east',\n",
       " u'easy',\n",
       " u'eat',\n",
       " u'ed',\n",
       " u'edge',\n",
       " u'editor',\n",
       " u'education',\n",
       " u'effect',\n",
       " u'effective',\n",
       " u'effects',\n",
       " u'effort',\n",
       " u'electrical',\n",
       " u'electronic',\n",
       " u'element',\n",
       " u'elements',\n",
       " u'email',\n",
       " u'end',\n",
       " u'ends',\n",
       " u'energy',\n",
       " u'eng',\n",
       " u'engineer',\n",
       " u'engineering',\n",
       " u'engineers',\n",
       " u'engines',\n",
       " u'england',\n",
       " u'english',\n",
       " u'enjoy',\n",
       " u'enter',\n",
       " u'entire',\n",
       " u'entirely',\n",
       " u'entry',\n",
       " u'environment',\n",
       " u'equal',\n",
       " u'equipment',\n",
       " u'equivalent',\n",
       " u'eric',\n",
       " u'error',\n",
       " u'errors',\n",
       " u'especially',\n",
       " u'essentially',\n",
       " u'established',\n",
       " u'et',\n",
       " u'eternal',\n",
       " u'europe',\n",
       " u'european',\n",
       " u'event',\n",
       " u'events',\n",
       " u'eventually',\n",
       " u'everybody',\n",
       " u'evidence',\n",
       " u'evil',\n",
       " u'evolution',\n",
       " u'exact',\n",
       " u'exactly',\n",
       " u'example',\n",
       " u'examples',\n",
       " u'excellent',\n",
       " u'excuse',\n",
       " u'exist',\n",
       " u'existed',\n",
       " u'existence',\n",
       " u'existing',\n",
       " u'exists',\n",
       " u'expect',\n",
       " u'expected',\n",
       " u'expensive',\n",
       " u'experience',\n",
       " u'experiment',\n",
       " u'experiments',\n",
       " u'expires',\n",
       " u'explain',\n",
       " u'explained',\n",
       " u'explanation',\n",
       " u'explicitly',\n",
       " u'exploration',\n",
       " u'express',\n",
       " u'expressed',\n",
       " u'extended',\n",
       " u'extra',\n",
       " u'extremely',\n",
       " u'eye',\n",
       " u'eyes',\n",
       " u'face',\n",
       " u'facilities',\n",
       " u'facility',\n",
       " u'fact',\n",
       " u'factor',\n",
       " u'factors',\n",
       " u'facts',\n",
       " u'fail',\n",
       " u'failed',\n",
       " u'failure',\n",
       " u'fair',\n",
       " u'fairbanks',\n",
       " u'fairly',\n",
       " u'faith',\n",
       " u'fall',\n",
       " u'fallacy',\n",
       " u'false',\n",
       " u'family',\n",
       " u'faq',\n",
       " u'far',\n",
       " u'fast',\n",
       " u'faster',\n",
       " u'father',\n",
       " u'fault',\n",
       " u'favor',\n",
       " u'favorite',\n",
       " u'fax',\n",
       " u'fbi',\n",
       " u'fear',\n",
       " u'features',\n",
       " u'federal',\n",
       " u'fee',\n",
       " u'feel',\n",
       " u'feeling',\n",
       " u'feet',\n",
       " u'fi',\n",
       " u'fiction',\n",
       " u'fido',\n",
       " u'fidonet',\n",
       " u'field',\n",
       " u'fields',\n",
       " u'fight',\n",
       " u'figure',\n",
       " u'file',\n",
       " u'files',\n",
       " u'film',\n",
       " u'final',\n",
       " u'finally',\n",
       " u'finding',\n",
       " u'fine',\n",
       " u'fit',\n",
       " u'flat',\n",
       " u'flight',\n",
       " u'florida',\n",
       " u'fly',\n",
       " u'flying',\n",
       " u'folks',\n",
       " u'follow',\n",
       " u'followed',\n",
       " u'followers',\n",
       " u'following',\n",
       " u'follows',\n",
       " u'food',\n",
       " u'fool',\n",
       " u'force',\n",
       " u'forces',\n",
       " u'forever',\n",
       " u'forget',\n",
       " u'form',\n",
       " u'format',\n",
       " u'formats',\n",
       " u'formed',\n",
       " u'forms',\n",
       " u'forum',\n",
       " u'forward',\n",
       " u'forwarded',\n",
       " u'foundation',\n",
       " u'founded',\n",
       " u'fraering',\n",
       " u'frame',\n",
       " u'frank',\n",
       " u'fred',\n",
       " u'free',\n",
       " u'freedom',\n",
       " u'freenet',\n",
       " u'frequently',\n",
       " u'friend',\n",
       " u'friends',\n",
       " u'ftp',\n",
       " u'fuel',\n",
       " u'fully',\n",
       " u'fun',\n",
       " u'function',\n",
       " u'fundamental',\n",
       " u'funding',\n",
       " u'funds',\n",
       " u'future',\n",
       " u'gain',\n",
       " u'galaxy',\n",
       " u'galileo',\n",
       " u'game',\n",
       " u'gamma',\n",
       " u'gap',\n",
       " u'gas',\n",
       " u'gave',\n",
       " u'general',\n",
       " u'generally',\n",
       " u'generate',\n",
       " u'generated',\n",
       " u'generation',\n",
       " u'genocide',\n",
       " u'george',\n",
       " u'germany',\n",
       " u'gets',\n",
       " u'getting',\n",
       " u'gif',\n",
       " u'given',\n",
       " u'gives',\n",
       " u'giving',\n",
       " u'glad',\n",
       " u'global',\n",
       " u'gmt',\n",
       " u'goal',\n",
       " u'god',\n",
       " u'gods',\n",
       " u'goes',\n",
       " u'going',\n",
       " u'gone',\n",
       " u'good',\n",
       " u'gospel',\n",
       " u'got',\n",
       " u'gotten',\n",
       " u'gov',\n",
       " u'government',\n",
       " u'graphic',\n",
       " u'graphics',\n",
       " u'gravity',\n",
       " u'great',\n",
       " u'greater',\n",
       " u'greatly',\n",
       " u'greek',\n",
       " u'green',\n",
       " u'gregg',\n",
       " u'ground',\n",
       " u'group',\n",
       " u'groups',\n",
       " u'gsfc',\n",
       " u'guess',\n",
       " u'guest',\n",
       " u'guilty',\n",
       " u'guy',\n",
       " u'guys',\n",
       " u'half',\n",
       " u'hall',\n",
       " u'hand',\n",
       " u'handle',\n",
       " u'hands',\n",
       " u'happen',\n",
       " u'happened',\n",
       " u'happens',\n",
       " u'happy',\n",
       " u'hard',\n",
       " u'hardly',\n",
       " u'hardware',\n",
       " u'harm',\n",
       " u'harvey',\n",
       " u'hasn',\n",
       " u'hate',\n",
       " u'haven',\n",
       " u'having',\n",
       " u'head',\n",
       " u'hear',\n",
       " u'heard',\n",
       " u'hearing',\n",
       " u'heart',\n",
       " u'heaven',\n",
       " u'heavy',\n",
       " u'held',\n",
       " u'hell',\n",
       " u'hello',\n",
       " u'help',\n",
       " u'helpful',\n",
       " u'helps',\n",
       " u'henry',\n",
       " u'hey',\n",
       " u'hi',\n",
       " u'high',\n",
       " u'higher',\n",
       " u'highly',\n",
       " u'hill',\n",
       " u'historical',\n",
       " u'history',\n",
       " u'hit',\n",
       " u'hitler',\n",
       " u'hmm',\n",
       " u'hold',\n",
       " u'holds',\n",
       " u'hole',\n",
       " u'holes',\n",
       " u'holy',\n",
       " u'home',\n",
       " u'hope',\n",
       " u'hopkins',\n",
       " u'horus',\n",
       " u'host',\n",
       " u'hot',\n",
       " u'hour',\n",
       " u'hours',\n",
       " u'house',\n",
       " u'hp',\n",
       " u'hst',\n",
       " u'huge',\n",
       " u'human',\n",
       " u'humanity',\n",
       " u'humans',\n",
       " u'hundreds',\n",
       " u'i3150101',\n",
       " u'ibm',\n",
       " u'ico',\n",
       " u'id',\n",
       " u'idea',\n",
       " u'ideas',\n",
       " u'ignorance',\n",
       " u'ignore',\n",
       " u'ii',\n",
       " u'iii',\n",
       " u'il',\n",
       " u'illinois',\n",
       " u'image',\n",
       " u'images',\n",
       " u'imagine',\n",
       " u'imaging',\n",
       " u'imho',\n",
       " u'immediately',\n",
       " u'immoral',\n",
       " u'impact',\n",
       " u'implementation',\n",
       " u'imply',\n",
       " u'important',\n",
       " u'impossible',\n",
       " u'include',\n",
       " u'included',\n",
       " u'includes',\n",
       " ...]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Remove english stop words. Remove words that \n",
    "vectorizer = CountVectorizer(stop_words='english', max_features=10000, min_df=.01, max_df=0.5)\n",
    "X = vectorizer.fit_transform(documents)\n",
    "n_samples, n_features = X.shape\n",
    "print \"Dimensions feature matrix:\", X.shape\n",
    "vectorizer.get_feature_names()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Dimension reduction\n",
    "We could reduce the number of dimensions by performing a SVD. We hope that we could describe most of the data in a much lower-dimensional space, in which the dimensions represent topics (or concepts, or _latent semantics_) rather than just words."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "n_components = 150 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Explained variance of the SVD step: 83%\n",
      "Dimensions feature matrix (old): (3387, 2189)\n",
      "Dimensions feature matrix (new): (3387, 150)\n"
     ]
    }
   ],
   "source": [
    "svd = TruncatedSVD(n_components)\n",
    "lsa = make_pipeline(svd, Normalizer(copy=False))\n",
    "\n",
    "X_red = lsa.fit_transform(X)\n",
    "\n",
    "explained_variance = svd.explained_variance_ratio_.sum()\n",
    "\n",
    "print \"Explained variance of the SVD step: %d%%\" % (explained_variance * 100)\n",
    "print \"Dimensions feature matrix (old):\", X.shape\n",
    "print \"Dimensions feature matrix (new):\", X_red.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAEKCAYAAAAYd05sAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10b287e90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "f = plt.plot(svd.explained_variance_ratio_.cumsum())  # these are the singular values \n",
    "f = plt.title(\"Explained variance by number of components, total: %d\" % svd.explained_variance_ratio_.shape[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Great, we can describe almost 90% of our variance by just 150 dimensions, and even better, 50% of our variance by only a handful of components.\n",
    "\n",
    "Each component is made out of a certain word combination with different weights. Let's have a look at themost important components."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(150, 2189)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svd.components_.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABZkAAAHvCAYAAAA2B9GQAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzs3XmYXGWZ9/FvCAQIiYFgkBADQkhuVFBRkE1AGIbFhW0E\nARcWHRYXlFdRcUAQRVBRBx1ZRUDWkXEXEAVkF9yQnTsojCA6EOiABAQC9PvHcxoqbSfp7nQtp+v7\nua6+0lV16jz3qXT/6vRdzzlnTG9vL5IkSZIkSZIkDcdS7S5AkiRJkiRJklRfNpklSZIkSZIkScNm\nk1mSJEmSJEmSNGw2mSVJkiRJkiRJw2aTWZIkSZIkSZI0bDaZJUmSJEmSJEnDtnS7C5AGEhFjgY8A\ne1J+TscBPwE+k5nPtLO2kRQRbwXemJlHDvDYTODbwGRgHvDezMwWlyhJCzCfF1hmP2DnzNyxdZVJ\n0oLMZYiIDYH/BMYDY4EvZua5LS5Rkl5gNkNEbAV8CVgG+AdwcGb+psUlqoWcyaxOdRKwEbB1Zq4P\nbAgE8K22VjXyNqQ0kQdyLvDNzHw1cCTwvZZVJUkL1/X5HBGTI+Jk4OutLUmSBtTVuRwRY4D/AY6o\ntn8H4KsRsXaL65OkRt2ezeOAC4D3Z+brgM8DZ7e4NrXYmN7e3nbXIC0gItYEbgVWzcx5Dfe/DNgk\nM38YEZOAbwKvBXqBS4BPZ+ZzEfEU8FXgbcBLgEOB3YD1gL8Cb8/MJyPiWeBrwFbACtXzf1CNdQSw\nB/AsMBv4UGY+GBFXAtcDmwGrA9cAe2dmb0RsChxXret54KjMvCgi9gF2AZ4DZgLPAO8FJgA/pMy2\nOCUzj2jY1mnAHZk5qeG+e4FdM/OmJX2NJWk4zOcXtvcAYBVgDvDWzHz7Er+4kjQM5jJExLLAXpl5\nRsN9twMfyMyrlvQ1lqShMptf2N6x1faMAT4M7J6Zb1ryV1idypnM6kSvB25vDGOAzHwwM39Y3fw6\nMCcz1wM2oATzx6vHxgF/zczXACdSPin8CPAqYBLQd1jzUsAjmbkBsDvw7Yh4aUTsC2wPbJCZrwVu\nA85sKGWtzNySEvBbA1tExErAGcC7M/MNwE7ASRExvXrOFpRQXw+4Djg0M28ETgYu6B/GwHTKm0ej\nvwDTFvPaSVIzmc9le0/JzM8BTw3lxZOkJuj6XM7Mp/s1mPenNEhuGPSrKEkjq+uzudre56rG+l+A\nLwJfHvxLqDqyyaxO9ByL/9ncHvgvgOp8RidTDo3r03dqiXuAWzPzb5nZC9zLgody9K3jVsonjVtU\n6/52Zv6jWubrwL9ExDKUTxh/Uj1nHvBHYGVgE2BV4EcRcRNwEeWTv/Wq5/wuM/uaxr9vqGFM9dXf\nwrb/uYXcL0mtYD5LUmcxlxtExKeAoyiz/J5ezOsiSc1iNleqxvo0YFPgjOraUxqlvPCfOtFvgFdG\nxIR+h5ZMA04B3kEJ7MYgG8uCP8+NO5XzFzFWY9N2KV58MxjT7/6lG+77R8NjvdX9SwF3ZubGDfVO\npRxK/e5+z4HFNy7uowR8o2mUTwAlqV3MZ0nqLOYyL5wy40xgHWDjzLxvcc+RpCbq+myOiJcA/9J3\n+o7MvCkibgbWBe5e1HNVX85kVsfJzAcoF737dkRMhBcC6kTg4cx8CrgU+GD12LLA/sAvhjHce6t1\nvJ6yU3plte59I2J8tczBwFX54hVg+4dpL+VwvJkRsXm1vvUpwTl1MePPp1xpdQGZ+RfgTxHxzmp9\n2wHPVZ9OSlJbmM+S1FnM5RdcCEwENrPBLKndzGagzII+vTrPMxHx6qq+Gwe9Zaodm8zqVB8A7gCu\nrw7VuIFyHqH3V48fDKwSEbcCtwB3AsdUjzVezbK33+3+No2I3wGnU05C/1j1/WXAryPiDuB1wLv6\nrXMBmfkw8G/AlyPiD8BZlHMZ3T9ADY23Lwd2jIgTBqhtD+DAahs/RznRvyS1m/k8+G2QpFbo6lyO\niM0oF8eaAVwXETdVX/+6iG2RpGbr6myuZnDvDPxntf2nA3s2nHJDo9CY3l7/NlJ3iojngZdmZk+7\na5Ekvch8lqTOYi5LUucxm9VpnMmsbuYnLJLUmcxnSeos5rIkdR6zWR3FmcySJEmSJEmSpGFzJrMk\nSZIkSZIkadiWbncB6k4R8b/Arpn5+0Us81ngj5l59giN+VPgwsw8ayTW12/dNwFbZubfR3rdgxx/\nQ+A/gfHAWOCLmXluRHwKeGfDoqsAEzJz0gDr2A/4OCUXLgMOzsxnm168pI5iPjdPREwGTgJeCzwB\nnJGZ/zXAcpMoF0cJyoSAszLzS62sVVJnMZubKyI+ALwPWB74HfC+zHym3zJjga8C21L2l4/PzFNa\nXaukzmE2t0ZEfB94IDM/PBLLqXlsMqtdFnuelsw8sgljNuX8MJm5fjPWOxgRMQb4H2DfzLwiIqYB\nv4+IGzPzOOC4arlJwK+B/QZYx7rAUcD6QA9wLnAI8OWWbISkTmI+N8/XgL8Dr6Tsg/0wIu7NzIv6\nLXc0cF9mviMixgO3R8RVmXlji+uV1DnM5iaJiF2BDwGbAo8BF1L2g7/Yb9EDgBnAq4GXAL+KiN9n\n5m9aWK6kzmI2N1lEfAJ4E3DBSCyn5rLJrEGLiLcD/wGMA54EPp6ZN0TEGcD4zHxnRLwauAJ4M2UG\n7auBl1VffwDen5mPN6xzDGUG7kbARGBMtcz1EXEmcGtmfiUingKOBf4VWA04ITNPqNbxPuAgymyv\nR4APZWZGxGrAWcBU4M+UWbz9t2kScB8wKzMfrO67ATgSuBf4JrBCNeYfgHdm5tMR8TTwQ8pstHcB\nvwFeCjxNmaU2E5gMPA7slZmzI+JK4HpgM2B14Bpg78zsjYi3AZ+rtuEJ4MDMvCUiNqU0iVcAngeO\nGqAZMa66/wqAzHwgIh4GpgF/bFjuK8DFmXlp/9cB2An4UWY+Ur0GpwBfxyazVAvmc8fmc3+vBz6Y\nmb3A/Ii4GHgHsMDzMvMj1Yw5KFm+LKXxIalGzObaZPN7KbOSH62258Dq/6y/nYFTMvN54NGIuAB4\nd7UtkmrCbK5NNhMRWwHbAScDKy3pcmo+z8msQYmImcAxwA6Z+XrKJ/nfj4jlgQ8Cr42IvSmfGn0k\nM++snroR8G/AOsCzwGf6rXojYNXM3DgzXw18B/hU9VjjJ3TjgDmZ+SbKH+THRcS4iNiSsmO4eVXX\nl4HvV8/5JnB9Zq4LHFzVsIDMfAz4AWUHkYh4JbAq8HPg/ZRDmTcF1gbWBN5SPXUZ4MeZuU5m/q5h\nldsDPZm5SWYGJaQ/1LA9a2XmlsB6wNbAFhHxMuBsSjC/ttqGYyNiReAM4N2Z+QZKI/ikiJjebxue\nzswz+m5HxP6U8L6h4b5XV8/v//r3eTnwl4bbD1T3Sepw5nPn5vMAbgTeGxFLR8QEyuu/6kALZuZz\nEXEOcAvwS2D2YtYtqYOYzbXK5pnAyyLikoi4mXJ036MDLDcduL/htvvLUs2YzfXJ5qq5/p/AXsBz\nS7qcWsMmswbrXymfnF0R5Tw951B+gdfOzCeBPYDTgBsys+/whF7gu5k5p5q1dTrl06UXZOYNwBER\ncVBEfJkS3CsspIYfVf/eRJnVNQF4KyUor6/q+iKwUkSsBPwLcGY1zp8on0QO5DRg7+r7fYFvV/V+\nEngkIg6lfCK2WjVmn2v6rygzvwd8JyI+HBEnUD75bNyen1TLzaPMMl6Z8gngbZl5S/XYDzLzrZRD\n9lYFflRt20WUT/3WW8h2EOUczEcBb8/Mpxse+gjwjcZPW/sZKAsMaKkezOca5HPlY5TX/ibKHw4/\nB+YvbOHMfDdlNsnKLPxDQkmdyWyuTzaPA7YBdgM2oMzaO2aA5dxflurPbK5BNkfEMrzY6H+QMjN8\n2MupdTxdhgZrKeDyzNyj746IeDnw1+rmOsDDwOsjYpnM7PujuXHHayzlU78XRMRbKZ86HU85TOMu\nqk/fBvAPgCyHYUAJkKWAszPzU9X6xlBC81HKm0HjzuCAF7HLzOuqWWVvBPYENqkeuqCq+b8pQTid\nBUNrXv91RcRBwL8D36Cc1/gR4BX9t6HSW63vGfqdUykiXlPVfmdmbtxw/1RgzgDjLkt541kH2Dgz\n72t4bCywK+Uw7YW5j/Jm22caC85sltS5zOcOzud+XgJ8IjPnVs/5JHD3ALVuB9ySmX/LzCeiHJK9\n62LWLamzmM31yeYHgB9UzRIi4lzgiAGWu4/yWvWZxoIzmyV1PrO5Htm8QTXe16rXaFVgbEQsm5n7\nD2M5tYgzmTVYVwDbRvWbW4XoLcCyEfEKSqBuQwnTvotkjAF2ioiXRMRSlJD6ScM6x1TP+UmWKzP/\nDtiFEoB9jy9KL2UW2J4R0Xe48UGUN41e4GfA/lW9qwNbLWJd36IE6M2Z2ddc3RY4OjMvrG5v1FDb\nQMZUzzkzy+krZgM79ntO/23qpRw+/cqIeFVV686Uw0xuAGZGxObV/etTmhFT+WcXUs79tFljg7my\nHjB3gPsb/RjYMSKmVG9o+1MOt5HU+cznzs7nRgdSLupHdUjh+4HzBlhuN8o59Po+RNwduHwx65bU\nWczm+mTz/wC7R8Ry1X7wzpSLZff3I2C/iBgb5fDvd1KaSZLqw2yuQTZn5q8yc/XMXD/LxQhPBi7o\n3zge7HJqHWcya1Ay844o5/q9oNr5mg+8nfJp1XnAl6plPgjcGhGXUYLmQeBiyuG+VwFfaFhtLyUE\nzoty/rPngKuBXasxevstS//bmfnziPgi8IuIeJ5yYaRdqmU+CJwREXdQZuXetIhNPKuqbY+G+z4N\n/CAieigXBLiScgjLwurppXxyeWpE7Fttz++AdRexHWTmQxHxLuCsiFi62oZ3ZubDEfFvwJcjYjnK\nh0LvzswFZkxExGbA24AErqveL6HMlvtFVfO9/ceNiB2BAzLzrZl5a0QcTXnTXYbyRtD/itqSOpD5\n3Ln5PIBjgbMj4lbKzvmRWZ3/LiI+W415JOW0GidXy/VSZtidsJh1S+ogZnOtsvlEyikyfkdpovwO\nOAT+KZtPAmYAN1NOsXFyZv7TYeaSOpfZXKtsXqh+2awOMqa3959+NqQRERFHAStn5ofbXYsk6UXm\nsyR1HrNZkjqP2SwNnqfLUDM1XkVVktQ5zGdJ6jxmsyR1HrNZGiRnMkuSJEmSJEmShs2ZzJIkSZIk\nSZKkYevaC/89++xzvXPnPtnuMoZspZXGU8e6ob6117VuqG/tda0bYMqUiYu7erAWoq65PFh1/rke\nDLevvkbztoG5vKTamc3t/Nl0bMd27OYym5eM2ezYju3YzbCk2dy1M5mXXnpsu0sYlrrWDfWtva51\nQ31rr2vdWjKj/f/d7au30bx9o3nbtOTa+fPh2I7t2KN3bC2Zbv25cWzHduzO1rVNZkmSJEmSJEnS\nkrPJLEmSJEmSJEkatq49J/Ps2bPp6ZnX7jKGbO7cCbWsG+pbe13rhvrWPpy6p09fg3HjxjWpIrVC\nXXN5sOr6+zhYbl991XnbzP7ma2c2t/Nn07Ed27FHhjndHGazYzu2Y4+kkcrqrm0yv+ew8xg/aZV2\nlyFpBDz52EOccOiOzJgxs92laAmYy5KGwuxvDbNZ0nCZ081jNksaKSOZ1R3ZZI6I7YDVM/O0Zo0x\nftIqTFhpWrNWL0mjTrOz2VyWpNaKiBuA3TPzvoUtYzZLUucxmyV1oo5sMmfmpe2uQZK0ILNZkkad\n3nYXIEmSpNGhI5vMEbEPsD0wA/gr8HLgksw8PCKmA6cAywP/APbPzL9ExBHAzsAcYDxwRGZe1Y76\nJWk0MpslqTNFxPLAd4CpwP3AFsBbgf8CngWeAv49M++PiGOA7arlXtqeiiWpviJiFnAGMB9YCjgV\n2Bt4DlgVODUzT4yILYHPVMtMAPbKzLsj4nBgJ0o/5qTMPDUiPgzsSfnw74LM/Eart0uSltRS7S5g\nIXqrr9UpYb0hsHVErA8cD3w9M7cCvgIcFxGvoTQ+NqA0M6bizAxJGmlmsyR1pv2BP2Xmm4CjgJcB\npwEfyMw3AycCX42INwCbZ+YGwHuBie0pV5JqbRvghurfI4FJwGrA24GNgUMiYgrwKuDd1f7x94Hd\nqv3m7YE3Vl8zI+JVwO7AZpQPCXeuGtmSVCsdOZO5wS2Z+ShARNwIBLAu8OmI+CQwBngGeCXw68zs\nBZ6KiN9Wj0nqEpMnT2DKFP9WbhGzWVJHMPtfsA7wM4DMzIh4GJiambdUj18DHAfMAn5bLfd4RNza\njmIldY9RmtOnA5+k5O5jwM+B6zNzPjA/Im7jxSP/vh4R84BpwHWUHO7bP54PHBoRuwNrAFdU618R\nWBuY3bpNktTNRiqrO73J/Mrq8L9ngI0oh6TcBRyfmb+KiAC2BG4HPhwRY4BxwPo4W07qKj0985gz\n5/F2lzEad6IHYjZL6giDyf4uyeXbgE2AH0XEDMppMH4fEetl5q2UTE7gDuCDVS6Pp8yyk6SmWVhO\n1zybdwKuycyjI2JP4BjgkYhYClgOeDVwN/AjYK3MfCIizqRMtrgLOKjK4aWBi4CPA7dn5g4AEfFR\n4BYkqUX6snpJs7nTm8zPABdSDvm7MDNviYiPAydFxHKUc38enJm3RcTFlENWHqZ8Iji/XUVL0ihn\nNktSZzkdODMirgL+TDk3/r8D/1U1MuYD78vM/42IS4DfUGbYPdiugiWpxn4LnBURzwBjgW9QTiV3\nCbAy8LnMfCQizgGujYgnKHk7NTNvjoifUWY1LwWcWO1LXx4R1wLLAjdSMlqSaqVTm8zjKE2MBzPz\nbY0PZOa9lHMYvaA639HczNwoIpalzOa4v1XFSlKXMJslqTOtD5yemb+IiJnAJpl5M2UG8wIy8xjK\nrDtJ0jBk5j3A5n23I+LNwBszc89+y31sIc8/jnIKo8b7jqdc40SSaqvjmswR8RbgYOBA4MuDfNrD\nwIYRsS/lUOzTMvMvi3rCk489tER1Suoc/j43Xyuy2f9HSUNhZizgHuD8iDgSWAb44Eit2NdZ0nB1\nUX70XRy7ZbrotZXUZCOZJ2N6e7vz9JizZ8/u7emZ1+4yhmzy5AnUsW6ob+11rRvqW/tw6p4+fQ3G\njRvXpIoGb8qUiV7YbpjqmsuDVdffx8Fy++qrzts2mOw3l5dMO7O5nT+bju3Yjj0yFpbTZvOSMZsd\n27EdeyT1ZfWSZnPXNpmB3k64SNhQTZkysSMubjYcda29rnVDfWuva93gDvMSqmUuD1adf64Hw+2r\nr9G8bWAuj4C2ZXM7fzYd27Edu+ljm81Lxmx2bMd27GaMvUTZvNRIFSJJkiRJkiRJ6j42mSVJkiRJ\nkiRJw2aTWZIkSZIkSZI0bDaZJUmSJEmSJEnDZpNZkiRJkiRJkjRsNpklSZIkSZIkScNmk1mSJEmS\nJEmSNGxLt7uAdpk9ezY9PfPaXcaQzZ07oZZ1Q31rb2bd06evwbhx45qybqlu6prLg9XJGWgWSVqY\ndmZzO3PTsTtvbN+rpBeZzd039gorrGwGquO1vMkcEdsBq2fmaa0eu9F7DjuP8ZNWaWcJ6nJPPvYQ\nJxy6IzNmzGx3KVJTRMTXgK8ATwCXA3Myc7uFLW8ut4dZJI0OEbEPEJl52BCfdwOwe2beN9DjZrM6\nge9VGo0iYgtgbmbeGhH/l5mrDva5ZnN3MQNVFy1vMmfmpa0ecyDjJ63ChJWmtbsMSRq1MvMQeGEH\n+p7MfMeiljeXJWmJ9DbjeWazJDXNfsD5wK0MMcPNZkmdqB0zmfcBtgfWBO4DXgFcAKwLrA9clJn/\nERFbAp+hnDd6ArBXZt4dEUcAOwNzgPHAEcAfgNOBydUwB2fmba3aJknqdhGxPPAdYCpwP7AlkMDB\nwNeBqRFxZGZ+tn1VStLoFxEfA94JPAtcnZmfioijgE2BFYD3Ae8BtqPk9UvbVKokjQoRsQxwMrA2\npX9xODAF+ACwDKWBvAuwHvBF4GngMkoOvy4i7gCWjYhzgdWBR4B3ZOazLd4USVoi7bjwX2/19QrK\nJ3dvAz4HHAJsRNnxBXgV8O7M3Ar4PrBbRLyW0qDegNJonlot+2ngsszcGjgAOKklWyJJ6rM/8KfM\nfBNwFLAKJeufAT4CXGGDWZKabhawG7BJZm4KzIyIt1Ly+PbM3IwyeWPzzNwAeC8wsW3VStLo8H7K\naeG2pPQpTqQ0nN+amZsDd1Aayr3Aspm5RWYeDfwM+ERm3k/J5sOq5SdRJuBJUq2088J/92Tm4xEx\nH3gwMx8FiIi+w0T+Cnw9IuYB04DrgHWAX2dmL/BURPy2WnZdYKuIeGd1e6WWbYW0BCZPnsCUKc39\n267Z62+Wutbdxdah7CiTmRkRDzc8NqY9JWmwRiqLRvvv7WjevtG8bV1kDPBa4CeZ+Vx13zXAq6vv\nZ1f/zgJ+C1Dti9/a0iqlYWr2fnM7c7Bbxx5F1gPeFBEbVbfHAvOBs6p+xjrAr6rHciHr6Gk4N/7/\nUY7all7Qit7BwnRrRnXr2EuinU3mxTkVWCszn4iIMyk7zrcDH46IMcA4Xvx07y7gnMw8PyJW4cXZ\n0FJH6+mZx5w5jzdt/VOmTGzq+pulrnVDfd8MRsBtwCbAjyJiBrBym+vREIxEFtX593YwRvP2jeZt\ng67K5V7KKeQ2joixwPPAFpRTGb22ug1lf/oD1f70eMrRg1LHa+Z+cztzsJvHHkXuBO7PzGMjYjng\nC8DHKJPllgJ+zouTLp5veN7zlIY0DP+8+uoSze4dLEy7c8KxWz/2kmjH6TL69C7m+3OAayPiWsph\nfFOr8yxfDNxAOYXGfMqh2McAu0fEL4FLKCfOlyS1zunAKyLiKuBI4KmGx/pOkyRJaq67ge9SjgC8\nEbg3M39YPdYLkJk3U/aXf0O54NSDbahTkkaTU4B1IuJK4HrK6TGuofQtrgae5MVTfTbuE98IHBsR\n6/DP+8ruO0uqnZbPZM7Ms4CzGm4/BazVcHu16t+P9X9uREwB5mbmRhGxLGXm3P2Z2UM5kb4kqT3W\nB07PzF9ExEzK+UC3rh5L4Kr2lSZJo1+1j93na/0e+2y/28dQJmlIkpZQZj4D7N3v7m8tZPEX9okz\n81TKEdwAqzXcv+eIFihJLdLJp8sYyMPAhhGxL+WTvdMy8y/DWdGTjz00ooVJQ+XPoEaZe4DzI+JI\nylW0PzjUFfg70R6+7pIWxYxQJ/DnUFqQvxPdxf9v1UWtmszVBf/2G4l1nX3sXvT0zBuJVbXU5MkT\nalk31Lf2ZtY9ffoaTVmv1GqZ+SCw9WIXXIS65vJgdXIGmkWSFqad2dzO3HTszhvb9yrpRWZz9429\nwgpe8kadr1ZN5pE0a9asWl7kps4X56lr7XWtW6qbuubyYJklkuqondnc7gvfOHb3jC3Vjdns2FIn\naueF/yRJkiRJkiRJNWeTWZIkSZIkSZI0bDaZJUmSJEmSJEnDZpNZkiRJkiRJkjRsNpklSZIkSZIk\nScNmk1mSJEmSJEmSNGw2mSVJkiRJkiRJw7Z0uwtol9mzZ9PTM6/dZQzZ3LkTalk3DK326dPXYNy4\ncU2uSFInqWsuD8QMkzRatDOb27nf69idMbbvp9LAzObRP7b5pzrq2ibzew47j/GTVml3GRrAk489\nxAmH7siMGTPbXYqkFhotuWyGSRpNRks2q358P5UWzmwe3cw/1VVtmswRMRa4DJgFfAq4HzggM/cc\nzvrGT1qFCStNG8EKJan7RMRRwN+AG4AdM/NzEbELcENm/m0o6zKXJWnhImI7YPXMPK2V45rNkrRo\nEfG/wKzMfGYE1rU58Ghm3rqo5cxmSZ2oNk1mYBowMTOnAUTElm2uR5IEvQCZeTNwc3XfwcAdlOaz\nJGkEZOal7a5BkjSg3hFc1/uA84FFNpklqRPVqcl8MjAzIk4GbgLu6nsgInYDDgGeA67NzMPaU6Ik\n1UtETADOA1YEbgc2Ax4BDszMjIgDgZdl5mcj4ljgDcDKwM2ZuV/DerYEDgTOBl4HfCcivgXMzMxP\nVEej3ARsMBKzPCSp20TEPkAAfwX2pDQ1LsjMb0TErsAngPnV43sARwJ/y8xTImId4KTM3CoijgHe\nTPk74HuZ+aWWb4wk1VSVxTsDE4CXAp9reGxd4CvA2OqxgzLzVxFxN3AtJcMfBP6tWuZkYG1gKeBw\n4HFgO+B1EXFHZt7fos2SpBGxVLsLGIKDGGBmXESsBBwFbJ2ZmwPTImKb1pcnSbX0AUrDeAvgO8BE\nSuOib0ZGL0BETAR6MnNbYENg44hYrf/KMvNi4A/AeyizMHaOiKWA7YErbDBL0rD1AmsBu1M+ENyC\nkrGzKE3lL1X7wj8FXsLCZ9btRWlSbw482uyiJWkUWj4ztwG2Bb5KaRiPAV4FfKx67IvAvtXyawKH\nZ+amwBTKvvT7gTmZuSWlaf3NzPw98DPgEzaYJdVRnWYyj1nI/WtTgvqSiIDSIFmrVUWpOSZPnsCU\nKRPbXcYLOqmWoapr7XWtu4ZeQdmZpZpp8TQL5m3fh5H/AF4WEecB8yizN5ZZ1Iozc15EXEWZkbEP\n8NkRrbyDLSzDRvvPtdtXX6N520aRMcAGlOy9orpvRcq+8P8DDouIg4E7gR8O8Nw+76I0P1YFLmlm\nwdKSauXfBO3MwW4du6Z6gasBMvOhiJhLmaHcSzmS5IiI+AelL/FY9ZyHM/OB6vv7geWAdYHNI2Kj\n6v6xEbFyi7ZBNdA//7o1Jxy7XurUZF6YeylBvU1mPhcRe1MOyVaN9fTMY86cx9tdBlB+uTullqGq\na+11rRtq+WZwC/Am4McRsR6wLKWhvBowG3g98BdgB+DlmblHREwBdmHhH/49T5nRAXAa5WKtkzPz\ntqZtRYcZKMPq/HM9GG5ffY3mbYNa5vLC9FLOfb9cZr4FICI+Sjlv5/7AUZk5pzq13C7AU8DU6rmv\nr5YfB+yWmXtGxBjg9og43xlz6lSt+pugnTnYzWPX1BjK6eOIiJdRmskPVfefALwrM++KiM8Ca1TP\nGejIkruAv2TmsRGxPPBpoIcF96PVxRrzr92/q47dXWMviTqdLqO3378AvZn5MOUQlasj4gbKIdl3\nt7o4SarV67MTAAAgAElEQVSpb1FmKF8FHFrd9w3gxIj4GeV9ohf4NbBWtdyFwJ8ojWj453y+nnJO\n5hUz89fADODcpm+JJI1+CVwREddGxG8os5gfoGT0TyPiMuBlwE+A/wbeEhG/BNan7Dc/A/RU+8xX\nAJfaYJakIekFVq3y9ieUU889Vz12DnBhRFwNzOTFD/oGWscpwDoRcSVwHfC/mdkL3AgcF9Vh2pJU\nJ7WZyZyZfwY26Xf3VdVj52IDQ5KGLDOfBfYGiIhlgbsy8xIGPoT6jQPcd33D932ZfARwRLXOpYAn\nKOdnliQN3zjgmcw8Hji+32M/rb4azWWA3M7Mz9FwoSpJ0pBdlZmHNdzuO13n16qvBWTm1Ibv92x4\naO8Blj0VOHWE6pSklqpNk3mkPfnYQ+0uQQvh/43UNmNY+IWihiwi1gS+D3w7M+ctbvnR8rs/WrZD\nUueIiLcABwMHtHpsM03t4s+eOtiI7S8Pl78fo5v/v6qrrm0yn33sXvT0LLbn0XEmT55Qy7phaLVP\nn77G4heSNKIy8ylG8MKpmXkv5RDtQalrLg/EDJM0kjLzYuDidozdzmxu536vY3fG2L6fqtNk5lnt\nrgHM5m4Y2/xTHXVtk3nWrFm1vMhNnS/OU+faJTVfXXNZkkazdmZzuy9849jdM7ZUN2Zzd40t1UWd\nLvwnSZIkSZIkSeowNpklSZIkSZIkScNmk1mSJEmSJEmSNGw2mSVJkiRJkiRJw2aTWZIkSZIkSZI0\nbDaZJUmSJEmSJEnDtnS7C2iX2bNn09Mzr91lDNncuRNqWTcsWPv06Wswbty4NlckqZPUNZf7mGuS\nRqN2ZnM793vrNLbvP1L3MZvrM7YZrW7StU3m9xx2HuMnrdLuMrrSk489xAmH7siMGTPbXYqkDlLn\nXDbXJI1Wdc7mbuD7j9SdzOZ6MKPVbUZFkzkivgZ8BXgCuByYk5nbLeo54yetwoSVprWiPEka1SLi\nSuCAzMwlWY+5LEnNERFvpuT0nkN9rtksSRAR+wCRmYc1Yd1vZogZbTZL6kSj4pzMmXlIZv4FWA+4\nZ3ENZknSiOqtviRJncmMlqQl08wcNaMljQq1m8kcEcsD3wGmAvcDWwIJHAx8HZgaEUdm5mfbV6Uk\n1V+Vt2cAqwPjgI8CBwJrAmOBr2bmdxuWXxE4B5hIeX85PDN/GRG3UXL6meHMopOkblLNltsZmAC8\nFDgaeAT4PPAc8CfgAGAMJaMXyOTq6JI7gXWqZd5Z/du3/t2AQ6p1XduMWXmSNFpFxBeADYCVgZsz\nc7+IOAr4W2aeEhHrACdl5lYRcQtwJfAaSiN5J+Bx4BvAhpT96yOBx4CZEXExsArwE/sZkuqojjOZ\n9wf+lJlvAo6ihHAv8AzwEeAKA1mSRsSBlKNDNgX2oHyo92BmbgZsA3w+Ilaulh0DHA5cmplbArsB\np1ePrQAcbYNZkgZt+czcBtgW+BrwbWCXzHwz8ACwD6XRPFAm9wLXZeZWwH8Dn67uIyJWouw/b52Z\nmwPTImKbFm6XJNXZOGBuZm5LaRJvHBGrsfCZyBOB8xqyewfKh4grZ+ZGwFaUhjXAcpQm9ObAh5q2\nBZLURLWbyUyZlfEzgMzMiHi44bExAz9FnWby5AlMmTKx3WUMWp1q7a+utde17lFmFnAJQGb+sdqJ\n/kV1e15E3AHMaFh+HeDs6vG/RsTfI6LviiRLdL7mOhhMro32n2u3r75G87bVUC9wNUBmPhQRTwBr\nAxdGBMDylCxeCbisWq5/Jl9R/fsrStOiz9rAFOCSal0TgbWauTFqvpHer25nHjh2d41dQ73AKhFx\nHjCPcsTJMv2W6d+TuKn6935KI/kVlGwmMx8FPlOdk/m2zJwPzI+IZ5tSvdpipDK6W3PCseuljk3m\n24BNgB9FxAzKYSqqmZ6eecyZ83i7yxiUKVMm1qbW/upae13rhvq+GSzEnZRZGj+OiLUos5mfAn4Y\nERMp58G/t9/yWwA3R8Q0YEXKId4Az7es6jZZXK7V+ed6MNy++hrN2wa1zOUxwBsAIuJlwLLA3cBO\nmfn3iHg7pbnxKsqMt4EyeQPgx8BmlH3nPvdSGh3bZOZzEbE3LzZAVFMjuV/dzjxw7O4bu2bGUGYe\n352Ze0TEFGCX6v6nKKfzBHh9v+f1n+V8J+WIPyJiEuWIk2MHWE6jxEhkdLt/Vx27u8ZeEnU8Xcbp\nwCsi4irK+YueanjMi09J0sg5BVirOr/nmcB2wMoRcQ3wS+CozJxTLdsLfAHYusrnHwD7Z+ZzmMuS\nNBS9wKoRcRnwU+AgyjnxL4qI66rbtwKnsvBM3qfK7h2AY/rWm5kPA18Fro6IG4DtKQ1sSdKi9QK/\npuwbXwVcSDlH/lRKo/gtEfFLYH0Wvu/bm5k/BuZW2f0zyimR6Pcc950l1VIdZzKvD5yemb+IiJnA\nJpm5dfVYAle1rzRJGj0y82ngXf3u3meA5bZquLnLAI97KLYkDc1VA1yQ7xcDLLfPQp7/qcyc3bi+\n6ovMPBc4d4krlKQukplnAWctYpE3DvCcNRu+P6zh+4MHeP5VDY+vNswyJamt6thkvgc4PyKOpJz/\n6IPDWcmTjz00okVp8HztJQ2kztlQ59oldaSOmcVmvnU2/3+k7uTvfj34/6RuU7smc2Y+CGy92AUX\n4+xj96KnZ94IVNRakydPqGXdsGDt06ev0eZqJHWauuZyH3NN0kioZsstyfO3WvxSg9fObG7nfm+d\nxvb9R+o+ZnN9xjaj1U1q12QeKbNmzarlRW7qfHGeOtcuqfnqmsuSNJq1M5vbfeGbbhxbUj2Yzd01\ntlQXdbzwnyRJkiRJkiSpQ9hkliRJkiRJkiQNm01mSZIkSZIkSdKw2WSWJEmSJEmSJA2bTWZJkiRJ\nkiRJ0rDZZJYkSZIkSZIkDZtNZkmSJEmSJEnSsC3d7gLaZfbs2fT0zGt3GUM2d+6Ejqh7+vQ1GDdu\nXLvLkDSK1DWXzUNJo1k7s7md+73tGnv69DVaPqak+jGb6zG2fyeo23Rtk/k9h53H+EmrtLuMWnry\nsYc44dAdmTFjZrtLkTSK1DGXzUNJo10ds7mu+t5Tpk1bud2lSOpwZnPn8+8EdaOmNpkjYjtg9cw8\nrZnjDMf4SaswYaVp7S5DkmotIsYClwGzgE8B9wMHZOaeQ12XuSxJgxMR+wCRmYctYpm9gZ7M/MlC\nHt8f+HZmPruoscxmSWquiFgLuBj4FfAVYKXMvGZRzzGbJXWipjaZM/PSZq5fktR204CJmTkNICK2\nbHM9ktQNehe3QGaetZhFDgPOAhbZZJYkNd2bgJ9m5scj4ijgb8Aim8yS1ImaPZN5H2B7YE3gPuAV\nwAXAusD6wEWZ+R9VU+IzlAsRTgD2ysy7I+IIYGdgDjAeOAL4A3A6MLka5uDMvC0izgBmAMsDJ2Tm\nOc3cNkkSACcDMyPiZOAm4K6+ByJiN+AQ4Dng2kXNuJMkLVxELA+cAawOjAP+B9g4Ii4FpgAnZeZp\nEXEbkMAzlDz+P+B7wHeBMcBywIHABsCqwPnArq3dGklqrYiYRcnQ+ZSew6nA3pR91FWBUzPzxEX0\nJQ4HdqL0T07KzFMj4sPAnpQP/S7IzG9ExEzgW8AywJPAHsDxlN7FZGBH4EvAy4GpwI+rWj4NLB8R\nPVVdz0TE7zLzt01+aSRpRC3V5PX3Vl+vAPYD3gZ8jtJ02Ah4X7Xcq4B3Z+ZWwPeB3SLitZQG9QaU\nRvPUatlPA5dl5tbAAcBJETEB2BzYpXrOc03eLklScRBwB2XGxQsiYiXgKGDrzNwcmBYR27S+PEka\nFQ4E7snMTSlNi38A8zNzO8r+70er5VYAju53yqI3Ag8DOwAfBFbIzNMpDeg9WlS/JLXTNsAN1b9H\nApOA1YC3AxsDh0TEFAbuS6xP6TG8sfqaGRGvAnYHNgO2AHauGtnHA8dUWX0CZWJdL3B5Zr4JmAj8\nKjO3p/RDDszM+4FjgXMz8wvAmcBXbDBLqqNWXfjvnsx8PCLmAw9m5qMAEdF3qN9fga9HxDzKodfX\nAesAv87MXuCpiOgL2XWBrSLindXtlTJzXkR8FDgNeAngLOYmmzx5AlOmTBzy84bznE5Q17qhvrXX\nte4uNGYh969NmV13SURA2aleq1VFtdJQ8nC0/1y7ffU1mrdtlJgFXAKQmX+MiMeA31ePPUg54q9P\n9nvuJcBM4EeUWXyfb26pWhKTJ08A2vs76diOPQqdDnwS+BnwGPBz4PrMnA/Mr44CmcHAfYlZvNiX\nmA8cGhG7A2sAV1TrX5GSs7Mo51Wm73z4EbEXL+byXGDDiNgK+DuwbHX/GBbcp17Y/rVqZrh9k4F0\na044dr20qsm8OKcCa2XmExFxJiVUbwc+HBFjKIcFrl8texdwTmaeHxGrAO+LiFWBN2TmrhGxHHBf\nRHwnM59v/aZ0h56eecyZ8/iQnjNlysQhP6cT1LVuqG/tda0b6vtm0AT3Ui4CuE1mPlddgOqmNtfU\nFIPNwzr/XA+G21dfo3nbYNTk8p3AhsCPqwtEHUM5n/JA+u//vhn4W2ZuFxGbAF8Atq6WG9uccjVc\nPT3zANr2O9nOPHDs7hu7hXYCrsnMoyNiT0qGPhIRS1FOI/Rq4G7Kh3H9+xJ3AQdVfYmlgYuAjwO3\nZ+YOANWEt1soWf1G4PKIeBewUjV+3+S6fYBHM/PAiFgb2H+AWp+n+Uecq0WG0zcZSLt/Vx27u8Ze\nEq0Kr97FfH8OcG1EXEuZ7TY1M2+jXGH1BsqhKvMp55c7Btg9In5JmZlxa2b+H7BqRFxH+VTyyzaY\nJaklevv9C9CbmQ8DXwWujogbKIcZ3t3q4iRplDgFWCsirqQ0l7/S7/GFXQiwF7gZeH+17/wlSpMZ\nykWlLhr5UiWp4/wWODoiLqeccvMblPMmXwJcDXwuMx9h4L7EzZQZ0NdRcvOczLyF0ki+NiJ+QzmC\n7wHgUOCwKm/3BM6txu/L6MuA7SPiKuBEYHZErMaLpxkF+B3wIS+mLamOmjqTubqq9VkNt5+i4XDp\nzFyt+vdj/Z9bnRNpbmZuFBHLArcB92dmD+Xcc/3HOmgotT352ENDWVwNfO0k9cnMPwOb9Lv7quqx\nc3lx53qx6pgtdaxZUv1k5tPAuxby2Av715m5ZsP9n21YbNsBnrfPYMY251rH11pqjsy8h3INJwAi\n4s3AG/udv37AvkR1/3HAcf3uO55yDuZGf6Kc97nRvg3PuQN43QBDNPZMLqZMtlsk86Lz+X+kbtQp\np8sYyMOU8xXtS/lU77TM/MtIrfzsY/d64XC0Opk8eUJH1D19+hrtLkHSKFPXXDYPJY1m7czmdu73\ntmts31OklmicOVxLZnM9xjbT1W06tslcnVh/v2atf9asWbU8/+BoP2+ipO5V11yWpNGsndnc7nMS\n+p4kjU6ZeRXVkXd1ZTZ319hSXXhCeUmSJEmSJEnSsNlkliRJkiRJkiQNm01mSZIkSZIkSdKw2WSW\nJEmSJEmSJA2bTWZJkiRJkiRJ0rDZZJYkSZIkSZIkDZtNZkmSJEmSJEnSsC3d7gLaZfbs2fT0zGt3\nGUM2d+6EltQ9ffoajBs3runjSFKfVuSy2SZJQ9POfeZW7fe2e2zfmyQNldnc+WOb7epGXdtkfs9h\n5zF+0irtLqMjPfnYQ5xw6I7MmDGz3aVI6iLNzmWzTZKGzn3m5vK9SdJwmM2dzWxXt6pVkzkirgQO\nyMxc0nWNn7QKE1aatuRFSdIoEhHbAatn5mmtHttclqTmiYhPAldk5m8a7lsOuDMz11zY88xmSRoZ\nEXEmcH5mXrqk6zKbJXWiWjWZgd7qS5LUBCOx0ytJ6jyZ+cV21yBJXc5+hqRRrSOazBGxPHAGsDow\nDvgocCCwJjAW+Gpmfrdh+RWBc4CJlG04PDN/GRG3AQk8k5l7tnYrJKl+IuJ7wAmZeXVEbABcDpyY\nmYdFxIeBPSk7wxcA5wGXZeb6EbExcHFmTo6IlwPfAt4GnAKsTbmw7OGZeZXZLEnNExGzKPvR8ynZ\n+27gM8DLganAjzPziL4ZdMB1wLnAisAf21GzJI0mA+TwXsAhwGbVIudl5tcblp8InA5MAlYDvpmZ\nJ1dHbj8ITAa2y8znW7YRkjQClmp3AZUDgXsyc1NgD2BL4MHM3AzYBvh8RKxcLTsGOBy4NDO3BHaj\nBDTACsDRNjEkadBOA/auvt8X+DRARLwK2J2yc7wFsDOwMvBI1VTeAfhzRGwI7Ah8H/h3YE6VzTsD\n36zWazZLUvNsA9xQ/XskZRLGrzJze2Ajyn42vDh77kDgliqrT2lxrZI0GvXP4Z2AV2TmxsCbgL0i\nYt1q2TGUCRnnZ+Z2wHbA/6se66U0pP/VBrOkOuqImczALOASgMz8Y0SsBvyiuj0vIu4AZjQsvw5w\ndvX4XyPi7xHRd9b7JT5fs2Dy5AlMmTJxxNfbjHW2Ql3rhvrWXte6a+jnwJcjYiXKTvDvqvtfDawB\nXFHdXhGYCfwAeCuwCXAcsG31/X7AUcCbImKj6jljGz4g7Ihsbla2DdZo/7l2++prNG9bFzgd+CTw\nM+Ax4LPAhhGxFfB3ylGCjWYCFwFk5q8jYn4La9UABnpv6tb3KsfurrFHkf45fBNwDUBmPhsRNwCv\nalj+IeCjEbErJacb+zIdsc+sJTfSf3d0a044dr10SpP5TmBD4McRsRZlNvNTwA+rQ0nWA+7tt/wW\nwM0RMY3S/HikesxP/EZAT8885sx5fETXOWXKxBFfZyvUtW6ob+11rRvq92aQmc9HxIXAyZQG8nN9\nDwG3Z+YOABHxUeBm4A+U02Y8BFwKfByYm5kPRcSdwP2ZeWx1GqRPAz3V+joim5uRbYNV55/rwXD7\n6ms0bxvUL5eHYSfgmsw8OiL2pOT0lzLzwIhYG9i/3/J3UD4c/HFErA8s09py1V//96Z2/k46tmO3\ncuxRpH8OHwP8HvjPiFgG2BQ4i3IkIJSZy7+qTpGxFWUCR5+O2GfWkhvJvzva/bvq2N019pLolNNl\nnAKsVZ2D6EzKISMrR8Q1wC+BozJzTrVsL/AFYOuIuIrSFNk/M5/Dk+hL0nCcQTm9xber272ZeQtw\neURcGxG/oRzW90BmPgAsC1yemY9Szj13UfW8U4B1qiy/DvjfzPQCJ5LUXL8Fjo6IyykN5c2A7av9\n5BOB2dVRglDy+GTKfvc1wAcoEzskScPXP4d3Be6NiOuBXwEXZuZN1bK9wE+AD1b7zB8B5kdE/6NO\nJKl2OmImc2Y+Dbyr3937DLDcVg03dxng8bVGtjJJGv0y835K4xjKLIu++48Hjh9g+Y0bvt+04ftn\nePH8zo3Lm82S1CSZeQ+web+7XzfAovs2fP/O5lUkSd1lITn8hwGWa8zh9QZY1VYD3CdJtdERTeZ2\nePKxh9pdQsfytZHUDs3OHrNNkobO7GwuX19Jw2F2dDb/f9SturbJfPaxe9HTM6/dZQzZ5MkTWlL3\n9OlrNH0MSWrUilw22yRpaNq5z9yq/d52j+17k6ShMps7f2yzXd2oa5vMs2bNquVFbkb7xXkkda+6\n5rIkjWbtzOZ2X/jG9yRJncps7q6xpbrolAv/SZIkSZIkSZJqyCazJEmSJEmSJGnYbDJLkiRJkiRJ\nkobNJrMkSZIkSZIkadhsMkuSJEmSJEmShs0msyRJkiRJkiRp2GwyS5IkSZIkSZKGbel2F9Aus2fP\npqdnXrvLGLK5cye0pO7p09dg3LhxTR9Hkvq0IpfNNkkamnbuM7dqv7fZY/veI2mkmc2dO7aZr27W\nsiZzRIwFLgOWAd6amY81YYyVgO0z8/zFLfuew85j/KRVRrqEUeHJxx7ihEN3ZMaMme0uRVIHi4iv\nAV/NzPtHYn3NzmWzTVLdLG7/OSK2AOZm5q0R8X+ZuepI1+A+85LxvUdSM5jNncnMV7dr5UzmacDE\nzNygiWO8FtgRWGyTefykVZiw0rQmliJJo1tmHjKS6zOXJemfLG7/eT/Kfu+tQG8zCjCbJanzmM2S\nOlErm8wnAzMj4mTKDvNLqvEPz8xfRsRtQALPAHcBM4GVq69vAv8GzAL2zswbI+JY4A3V4zdn5n7A\nfwCviYj3Z+a3WrhtkjQiImJ54AxgdWAc8FHgQGBNYCxl5vB3I+JK4A/AusA84BpgO2BFYFtg5+pr\nAvBS4OjM/H5D1j5drffbwORq+IMz87aIOAOYASwPnJCZ50TEMcCbKbn9vcz8UlXD/sBDwDnARBbM\n9VuAK4HXUJofO2Xm30f6NZOkUaxv//lbwCrAcsBU4HDgfkruvy4i7gCWjYhzKe8fjwDvAFYATuef\nc/7PwJ3AHZn5/1q5QZJUdxHxPco+8tURsQFwFPB/lB7GUpR94asG2hcGXg8ckJl7Vuv6W2ZOjYhd\ngU8A84G/AntkZlM+PJSkZmnlhf8OAu4AHgd+nplbArtRdnyh7AQf3Re2wBOZuQPwPeAtmbkjcByw\nR0RMBHoyc1tgQ2DjiFgN+DxwhQ1mSTV2IHBPZm4K7AFsCTyYmZsB2wCfj4iVKTuqN2bmNsCylMzc\nlpKzW1aPj68e3xb4anXYdV/W7kX5YO6yzNwaOAA4KSImAJsDuwDbA89Vde0F7Fk99mh1Xy8whtLs\nuHSAXJ8InJeZbwYeAHYY0VdKkka/vv3n8/4/e/cdJ1dZv///tQEChIRIMFQjSkjegCCCIL0EUUSk\nSguigAUE7IoaRCl+NHa/2OhN6kd/KooFFKQ3xULnilg+ICgENgRCRAKZ3x/3vWTYzO7Ozs7MmTN7\nPR+PPHZ25sy53/fu7HXu3KcBX885fwRwjKQ/AlcAn8yXLRoPzJK0PTAR2BQ4jn45n9f7CmCmJ5jN\nzBpyJnBofnw4KYsfz2PhvUkHyUHtsfBAE8cHAV/JGf5z0kF5Zmal0s4jmXvy1w2AiwAkPRIRT0VE\n38WEVLX8H/PXJ4F7qh6vAPwHWD0iLiYdwTee1JcerCkmTRrP5MkTmr7eVqyzHcpaN5S39rLW3QTT\ngV8BSHog70D7Tf5+QT5abWpetjon782P55FyEuC6/L7HImIeMDk/35e1GwMzIuLA/P0quY2PkAbP\nK5OOUAZ4B/BlYI2++qqsD1yQ2+qf63/KXx+qqqswrcq2enX759r9K69u7lvJ9Y1t/w0cHxHvIU1Q\n1BrD90p6sGr5caSzXV6S8/nr45Lmtahm66eRbc9o3Va57dHVdon9GvhqvifUdqSs3i4itsyvL5MP\nCoGhx8J9Of8xYFZEfIh0psllLancWq4b51Lc9uhqeyTaOcnc517SkXB/joi1Sad2P5FfW1xj+R6W\nnjzeDXiFpIMiYjLpiLse0hF37Tw6u2v19i5g7tynm7rOyZMnNH2d7VDWuqG8tZe1bmjKxuA+0hka\nP4uIdUlHNTwLXJbP4tgY+HtedqhT6F4PEBGrk46keCw/35e19wG3S7okTwq/JyLWAF4vad+IWAF4\nMO/Q21/SzIjoAe6JiEv71bwDcEeNXO+o0/xakW31KvPnuh7uX3l1c9+gvIP0Kj3A54EzJV0REYez\n5Ai6xaRLKUHtvL0fuLAq599d9T5rk+Fue4r8m3TbbrudbZeVpMUR8UPSJY1+Qhr3PiRpdr703XFA\nb168fzY/S7rsERGxDksuZ3QEcKKkufkSo/sA329tT6wVum0uxW2PvrZHot0TshVgNrBzRFxHCuQj\nJL3A0uFbqfra//HvgHXzOn4I/BVYK3/dOO/9MzMro9NJ+XYtcB7pepurRsQNwDXkwWed61ojIq4i\nnXJ3tKTFvDRrvwAcEBHXkI5OvkvSv/P7biIfpSHpOaA3Im4Ffku6NEbf0XIV4IsML9fNzKx+FeAH\nwNdyzu5CuicJwG3A7IhYn9qZ2z/n7656zczMGncu6dIY55DG7+vn8ftNwD8GuJ5yBbgdeDKPq08E\n/pZf+x3w8zx2Xx24vKXVm5m1QE+lMjrHmDPe/b2K78Za24J5DzP7iK2YOnVaU9db1iOlylo3lLf2\nstYNMHnyhI64bE9EHAqsL2lW0bXUq9W53Kpsq1eZP9f1cP/Kq5v7Bp2Ty2XlMfPINLLtKfoIJrft\nttvUtrN5BJzNnakb51Lc9qhre0TZXMTlMjrCwvmPDb3QKOWfjVnXKNVexFZnj7PNzGz4nJ0j45+f\nmbWCs6Uz+fdio92onWS+YPbB9PYuKLqMYZs0aXxb6p4yZZ2Wt2FmrSPp/KJrGK525LKzzcxseIoc\nM7dr3Nvqtr3tMbNmczZ3btvOfBvNRu0k8/Tp00t5ami3n9JqZqNXWXPZzKybFZnNo/VUVTOzoTib\nR1fbZmXR7hv/mZmZmZmZmZmZmVkX8SSzmZmZmZmZmZmZmTXMk8xmZmZmZmZmZmZm1jBPMpuZmZmZ\nmZmZmZlZwzzJbGZmZmZmZmZmZmYN8ySzmZmZmZmZmZmZmTXMk8xmZmZmZmZmZmZm1rBliy6gKHPm\nzKG3d0HRZQzbvHnjm173lCnrMHbs2Kau08xsuFqZy845M7PGFDlmbsW4t9lte/tiZkVwNrePc96s\nfqN2kvmdsy5m3MTVii6jcAvnP8Ypx+7J1KnTii7FzEa5VuWyc87MrHEeMw/M2xczK4qzuT2c82bD\n05GTzBGxE3CkpJn9nv8m8A1JD420jXETV2P8KmuPdDVmZtYkzmUzs87jbDYz6zzOZjPrRB05yQxU\naj0p6aPtLsTMzMzMzMzMzMzMBta0SeaIWBH4PrAm8BCwIyDgMWAVYD/gLGAisBbwXUmnRcS1wH3A\n+kAPcGD+Oi0ifgmsBlwu6aS87JFAL3B+XlcP8C5gdeDrwHPAQmA/SeW76LKZWYMi4jBgb2A88HLg\n88BJpCz+L3AUcBEwgZT/x0u6JiLelpebD8wD7gSuBb6S33cG8CxwNLAcaUfgPsDGwKz82hTgNGBn\nYBPglJzxXwB2yu39SNJXWvgjMDMbtSJiOnAusIh0c+9DgM8BryCNz3+Wv58DbCHpyYg4Chgv6avF\nVCTed+8AACAASURBVG1m1h3yOPytwIrAVODLwD9IuTuGND4/mDRX8gNgZWAc8BlJv4mID5DG1ysB\njwP7SFrU3l6YmY3MmCau6wjgr5K2A04kTQ5XgIslvZkUtJdI2hXYFfhYfl8FuEnSDOB/gePycysA\newHbAx+oWhbgeOAySdsCHwfekJe9lDS5fSppYtvMbLRZUdIuwJuBb5B2xp0s6WDgs8CVknYE9gfO\njogxwCnAWyTtDPyHJVm7vKQdJF0ITAN2l7Q9cC8pxyvA2sC+pAns40mTGruRdghCGkzPJGX5ky3t\nuZnZ6LYLcGv+egJph+Itkt4CbAm8X1KFtLOx75J07wDOa3+pZmZdaWVJewB7Ap8GNgAOyXMdPyaN\nv9cFVgX2IGXxshHRA0wCdpG0FengjC0KqN/MbESaebmM9YErACQpIh7Pzyt/fQz4SETsCzzVr+3f\n5q+3kCaLAe7Oe+4WRcTz/dqaTjoqGkm3ALfko54/A1wNPAzc1qyOdbtJk8YzefKEtrTVrnaarax1\nQ3lrL2vdBasA1wNIeiwi5gHBkhxeH7ggv/5IRDxFOsLtKUlz8zI3AGvkx33vA5gLnB8RC/J6bsnP\n3y3phYiYT9rR+HxEPEnaUQhpAuPLeZ2/ampvh6GdOTeYTqihldy/8urmvo0iZwOfIo3H55POUNki\nImaQxt7L5+XOAS6NiOuBR6vy3xrQqu1LkX+TbtttW0MqwJ/z43+SxsKPAN/K4+e1gRsl3RsRpwOX\nkM4Q/JakSkQsAi7Jy76Czr206ahTnfOj9W/VbY+utkeimcF1N7A18NOImEraO9fDkiPiPkY6muK0\nPNjdveq9m5NO4ds2rwcGuC5zdh/p6OW7ImIH0mkp/wTOk3RsRMwiHVl9clN61uV6excwd+7TLW9n\n8uQJbWmn2cpaN5S39rLWDYVvDHqA1wNExOqko9geAxbn1+8DdgDuiIi1gZcB/wImRMTLJT1OyvG/\n5+UX53VNJJ2hMoV0Bsyvc1swSFZHxFhgf0kz8xEa90TEJc24eetwtSvnBlPmz3U93L/y6ua+QeG5\n3E57ATdIOjkiZpImO74i6f0RsR5pbIykB/POwM+QD9qwxrVi+1Lk36TbdtvtbLsLVY+Le4AzgamS\nFkTEecCYiNgImCDpbRGxJnBTRDwI7CVpq4gYB9xOc886txHoy/mi/17ctttuV9sj0cxJ5rOB8yLi\nOuD/SNforLAkaC8Hvh0RB5FOmV6UJyAADouIjwELgHcCr+WlAd3/8ReBcyLiENIkyHtIl+c4KyKe\nAV4gD6TNzEaRCrBGRFxFus7b0aTLB/Xpy879SNeLO0LSonwNuF/mo5HHkK7X2bc+JM2PiJtIp2E/\nT7ou/pqkyegBs1rScxHRGxG3ki7DcWURE8xmZqPE7aQzTp4jZfm2wKkRsTXp+vpzImJNSf8iTXx8\ni3S2iZmZNUf/cfEFwA15juJR0vj5L8AJEXEAKas/CzwAPBMRN+b3PpKXNTMrlWZOMm8KnJ0vWj8N\n2Dpf3xMASdeSbhL1EhEB8GlJc6qevi7/63vvWvnrjKpl9uy3qn+QjsAzMxvNrpM0q+r7dfseSJpH\nuqFIf68DtsuTwhcAD0rqn8MHDtRefv1+0k3/kPQksGF+/HnSDQjNzKyFJP2NdP37aq8bYPFlSeP2\nwc4cNDOzOkk6v+rxs8CrB1l8/xrPvbHpRZmZtVkzJ5n/RrqG0Amkawsd08R1N93C+Y8VXUJH8M/B\nrOs0MmHwNHBrRCwkHZ38v80tqT6tyiPnnJnZEhHxRdKNst9Wz/LO0IH5Z2NmRXH+tId/zmbD07RJ\nZkmPko9iG+b7Zgy9VPNdMPtgensXFNH0iEyaNL7pdU+Zsk5T12dmxag+gmKY7/su8N0mlzNsrcxl\n55yZWSLpuOEsX+SYuRXj3ma37e2LmRXB2dw+znmz+o3aO5ZOnz69lDe56fab85jZ6FXWXDYz62ZF\nZnPRN77xNsnMOpWz2cw6ke9YamZmZmZmZmZmZmYN8ySzmZmZmZmZmZmZmTXMk8xmZmZmZmZmZmZm\n1jBPMpuZmZmZmZmZmZlZwzzJbGZmZmZmZmZmZmYN8ySzmZmZmZmZmZmZmTVs2aILKMqcOXPo7V1Q\ndBnDNm/e+IbrnjJlHcaOHdvkiszMmqNVuezsMzNrXJFj5pGMe9vRtrcvZlYUZ3NzOc/NmmPUTjK/\nc9bFjJu4WtFltM3C+Y9xyrF7MnXqtKJLMTOrqRW57OwzMxuZ0TZmrpe3L2ZWJGdz8zjPzZqn0Enm\niFgd+JykYyLiH8B0Sc8NsOy/Ja3R77kTgX9JOn24bY+buBrjV1l7+EWbmdlSIuIwICTNqnruEuBd\nwBrAJpJ+Ptg6nMtmZq0RETsA8yTdVWtMPRhns5lZ/fL4d13gnZLmtKodZ7OZdaJCJ5klPQock7+t\nDLF4rdeHeo+ZmbXHUnksaSZARLwRCGDQSWYzM2uZdwOXAHfh8bOZWSu9UZIPMTazUWnEk8wRsTJw\nFjARWAv4ATBT0ob59e8AVwHzgM+RbjY4HjgYWARcImnrqvVtBHwdWAZ4OXCUpFuA5fNewSnAnZKO\n7lfHbGC7/L5vSPr/Rto3MzMblq0j4ipgZeAk4DvAa4BPAytGxE1DHc1sZmYDi4jlgNOA9Uhj6uOB\nycDRwHKkCeR9gI2BLwP/JY3DdwVeFxH3ksbUFwGvBJ4A9pP0fJu7YmbW0SJiOnAuac5iDHAGsHvV\nQRT/krRmRJwHTAJWBe4AJkbET0hn853NknmS70o6LSK2BL6Z1/kw8A5gGnAK0EPK5XdLeqpdfTUz\na5YxTVjHVNJE8a7Am0lhekdEbB8RywM7AT8DNgQOkTQD+DGwP0sfSdGTl/u4pF1Ig+PD82srAp+U\ntB2wakTs0femiNgNeJWk7YGdgc9ExMQm9M3MzOrTAyzI2b07aYK5B3gBmA1c5AlmM7MRey8wV9KO\nwN7A90gTzrvncfC9pAnlCrC8pB0knQxcQRpHP0Q62GNWXn4isGkB/TAz63S7ALfmryeQ8rKWCnC1\npG3zgXC9kvYhZXPfPMmuwMfy8qcDh0vainSW3wakCeyj81zJr4BPtqhPZmYt1YzLZTwGfCQi9gWe\nyus8EziUdB3On0paHBGPAN+KiAXA2sCNNdZVAR4BPhsR/wEmAPPzaw/mgTHAzaRTr/tsBLw+Iq6p\n6tc6wJ1N6F/XmDRpPJMnTyi0hqLbb1RZ64by1l7WukexCnATgKS5ETGftBMS0mRzT1GFdUL29emU\nOlrF/Suvbu5bl9kY2C4fCQfpDL5FwPl5jL0+cEt+TQOso1fSg/nxv4FxrSq227Vy+1Lk36TbdtsG\npKOQP0XaSTcf+HW/16vHtrXyttY8CcDqkgQg6VyAiNgAODUiIJ2V0rJrOVtt9eb5aP1bddujq+2R\naMYk88eAW/KpHzNIR1L8NiK+RppM7rusxRnAupKeyaeU1DqKuod0msg7JN0fESeRJosBXhERa0j6\nN+myGGcDfQPs+4FrJB0ZEX2nDv6tCX3rKr29C5g79+nC2p88eUKh7TeqrHVDeWsva91Q3o1BE/QA\nWwBExBrASsDj+bUXaM6ZMw0pOvv6lPlzXQ/3r7y6uW/Qdbl8H/CQpNkRsQLwReDjpDH3GNIkSN/E\nx+Kq9y0mTUiDr8ncNK3avhT5N+m23XY72+5wewE3SDo5ImYC7yOPZyNiHdIlMvrUytWP02+eJD//\nSESsJ+mBiPgUaYL6ftKNAv8ZEdsAa7aoTzaAevK86L8Xt+2229X2SDTjP/2XA8dExLXAh4FFETEW\n+CGwnKS/5+UuBG6MiBtJRyj3BWel39cLgR9GxPWkaxP1LfcE6Ujom4F/SLqi732SLgcW5PfcDiyW\ntKAJfTMzs/pUSNddvhq4DDgyP1ch3Whqr4g4oMD6zMy6wenA+nncfTPp8hg3kE7pvh5YyNJjbIDb\ngNkRsT5LT4Z40tnMbGm3Ayfnse0RpEtYPBkRtwIn8tKD2io1Hv+MpedJliONkc/Jz28C/IJ0YN4F\nEXED8CXStZ3NzEpnxEcyS7qWdOpef7Pzv77lPj7AKrbJr6+bv/9m/te/nVfWeO6kOtZvZmYtJul8\n4Px+T/fl+p9Jp3CbmdkISHqOdEm6amcNsPh1Ve87g3RWIaQbUPU9P7OpBZqZdQlJfwO27/f03jWW\nO7zf92vlr9dSe57kdmCHfs/9EZjRaK1mZp2iGZfLKKWF8x8ruoS2Gm39NbPyaUVOOfvMzEbGOVqb\nfy5mViRnUPP4Z2nWPKN2kvmC2QfT21u+K2pMmjS+4bqnTFln6IXMzArSqlx29pmZNa7IMfNIxr3t\naNvbFzMrirO5uZznZs0xaieZp0+fXsqb3HT7zXnMbPQqay6bmXWzIrO56BvfeJtkZp3K2WxmnagZ\nN/4zMzMzMzMzMzMzs1HKk8xmZmZmZmZmZmZm1jBPMpuZmZmZmZmZmZlZwzzJbGZmZmZmZmZmZmYN\n8ySzmZmZmZmZmZmZmTXMk8xmZmZmZmZmZmZm1jBPMpuZmZmZmZmZmZlZw5YtuoCizJkzh97eBUWX\nMWzz5o2vq+4pU9Zh7NixbajIzKw5WpHLzkIzs5Epcsxc77i3HW17e2JmncTZ3DjnuVnrjNpJ5nfO\nuphxE1cruoyWWDj/MU45dk+mTp1WdClmZnVrdi47C83MRq6bx8z18vbEzDqNs7kxznOz1irFJHNE\n7Aq8UtKZzVrnuImrMX6VtZu1OjOzUSEirgWOlKRmr9u5bGY2PBGxAnCfpFe3qg1ns5lZ53E2m1kn\nKsUks6Qri67BzMwAqOR/ZmZmZmZmZmZAiyeZI+JHwCmSro+IzYETgX8D00g3HTxe0nURcTcg4Dng\nO8DX8+OFwH75X0iaFREfBw4Engeul/TpiDgReBWwGrAO8FFJv25l38zMuklErAicC7wSGAt8BHg/\n8GpgGeAbkn5QtfzLgAuBCaRtyfGSrhkqzyWV72L4ZmYdJCLGAxcBLwMeyM+9jpS5zwPPAu+T9FBE\nfBCYSdo5eKmkb0fEvsAngUXAI8BBkrzz0MxsmCLiMGAPYAVgTeAUYC9gI+ATpHHyh4H/An8BjgB6\nSGPul4yx89mCf8rvXRnYX9KDbeyOmdmIjWnx+s8EDs2PDweuAB6XtCOwN/Dd/NpKwMmSZpJC+VJg\nR+BUYBXyUXMRsTGwP7C1pG2AaRGxe379WUlvJYX4R1vcLzOzbvN+4G85Ww8iZfCjkrYFdgH+JyJW\nzcv2AMcDV+Y83x84O782VJ6bmdnIvB+4M+fv6fm5M4GjJe0EfA/4RkRsABwAbAvsAOwdEdNJGf8V\nSdsDPydNZpiZWWNWkrQ78GXgKEn7kiaT30c6yG5GztsngSPzv1pj7Apwm6Q3Ab8h7SA0MyuVVl8u\n49fAVyNiFWA70sTEdhGxZX59mapJi77re34R+AxwNfAwcFvV+gK4VdIL+fsbgNfkx3/OX/9J2pM4\nqk2aNJ7JkycUXcZSOrGmepS1bihv7WWtu8SmA78CkPRARKxFGuAiaUFE3AtMrVp+feCC/PojEfFU\nRPTdfaSePG+LTsvCTqqlFdy/8urmvnWhacAvACT9LiKeB9aUdGd+/QbgS6Sj4dYBfpuffxmwHvAx\nYFZEfAi4D7isjbWXVru3J0X+Tbptt211q7BkHmI+KVMhTSivCNwt6Zn83PXAm4HFwFVQc4z9p/z1\nIWCN1pY+eo0kz0fr36rbHl1tj0RLJ5klLY6IHwKnAT8BngAekjQ7n5p9HNCbF1+cvx4CnCfp2Ij4\nNGkv4P/l1+4HPh4Ry+TldwC+D2yCrxH6Er29C5g79+miy3iJyZMndFxN9Shr3VDe2staN5R3Y0Aa\nFG8B/Cwi1iUd6fYscFlETAA2Bv7eb/kdgDsiYm3S5MUT+bVaeT6LlOcnt7wnVTopC8v8ua6H+1de\n3dw3KHUuD+ReYGtSXm8KLAc8EhEbS7qLdPaI8r97JO0GEBEfAe4iZfGJkuZGxGnAPqTxtA2induT\nIv8m3bbbbmfbXWKweYgNI2KcpIXATqRcfh7YntpjbM9ptEGjeV7034vbdtvtanskWn25DEjXG9ob\nOId0St/6+XpDNwH/yNeAqw7T3wFnRcRVwAzg/Px8RdLdwA/ye28D/i6p7+iL6nU4nM3Mhud0YN2c\nz+cBuwKrRsQNwDXkCYm8bIV0lPLOEXEdaSfiEfksk4HyfCeW5LmZmTXuNFJe3wAcDfyHdFr2dyLi\neuCDpPuT3AlcHRE3RsTvSUcxP0zK5p/nbF4duLyITpiZdYlK1dfqx4uAE4BrIuIWYBLp8nFnMPAY\nu9Z6zcxKo9WXy0DSQ8DyVU8dWmOZdase/450dEa186te/ybwzX7vP6nq8f3AzkPVtXD+Y0MtUlrd\n3Dczaw1J/wXe0e/pw2osN6Pq231qvD5Ung+o2dnlLDSzbpTz+sAaL+1YY9mvAV/r9/TP87+6OEv9\nMzCz2iRVz1NcCVyZH98B7JZfurTGWw+rsa4ZVY9P7/96f86lxvjnZtZaLZ9k7lQXzD6Y3t4FRZcx\nbJMmja+r7ilT1mlDNWZmzdOKXHYWmpmNTJFj5nrHve1o29sTM+skzubGOc/NWmfUTjJPnz69lNcf\n7PbrJprZ6FXWXDYz62ZFZnPR1yT0NsnMOpWz2cw6UTuuyWxmZmZmZmZmZmZmXcqTzGZmZmZmZmZm\nZmbWME8ym5mZmZmZmZmZmVnDeiqVStE1mJmZmZmZmZmZmVlJ+UhmMzMzMzMzMzMzM2uYJ5nNzMzM\nzMzMzMzMrGGeZDYzMzMzMzMzMzOzhnmS2czMzMzMzMzMzMwa5klmMzMzMzMzMzMzM2uYJ5nNzMzM\nzMzMzMzMrGGeZDYzMzMzMzMzMzOzhi1bdAGtFBFjgO8BrwX+C7xX0l+rXt8D+CzwPHCOpLMKKbSG\noWrPy4wDfgO8W5LaX+XS6viZzwQ+TPqZ3wUcLalSRK391VH724FPARXgIknfKqTQfur5rOTlzgCe\nkDSrzSXWVMfP+6PAe4C5+akjJc1pe6E11FH7FsDXgR7g38Ahkv5bRK2dqMzZXI+y5nc9ypzx9Sjr\ndqBeZd1e1KvM25WiNJLH9X6OWtF2fv6PwPy82N8kvafZbedlXpLT7ep3rbbzcy3vd60MJ41l2vH7\nrrn9aFO/l8r2Nn7Oa25X2vU5z8u9mPnt/Jz3bzt/P+J+d4Oistm57Fx2LjuXR5rL3X4k897AWEnb\nAJ8mTfoAEBHLAd8A3gTsCBwREasVUmVtA9YOEBGbA9cDryZ9+DvFYD/zFYHPAztJ2g6YCLytkCpr\nG6z2ZYDZwBuBrYGjI2JSIVUubdDPCkBEHAlsREk+K9lmwDslzcj/OmkiYLDPSg9wBnCYpO2BK4B1\nCqmyc5U5m+tR1vyuR5kzvh5l3Q7Uq6zbi3qVebtSlEbyeG9g+cE+Ry1qe3JErABQ9TtsdAKqkZwe\n8u+nVW23o9+DZHg7ft81225Tv2tl+6q0p981tyvt+pznGvpnfls+57XabmK/u0FR2excdi4P2rZz\n2bk8VCPdPsm8LWmCB0m3AZtXvbYB8ICk+ZIWATcCO7S/xAENVjvAWNKHpNOOgBus7meBrSU9m79f\nFvhPe8sb1IC1S3oBWF/S08BkYBnguSKKrGHQz0pEbAO8ATidtNezUwz1GX89cFxE3BARn253cUMY\nrPbpwBPAxyLiWmCSJzKWUuZsrkdZ87seZc74epR1O1Cvsm4v6lXm7UpRGsnjbYFfDfCeVra9I7AJ\nMC4iroyIqyNiyxa0DbVzeqj3tLLtdvS7VoY/S3t+3wNtP1re70GyveX9HqTttnzOB8j8tnzOB2i7\nWf3uBkVls3PZuTxY285l5/KQ/e72SeaVgaeqvn8hHyLe99r8qteeJu2d6RSD1Y6kmyX9s/1lDWnA\nuiVVJM0FiIgPAitJuqqAGgcy1M98cUTsC/wJuAZY2Ob6BjJg3RGxJvA54AN03oTBoD9v4BLgSGBn\nYLuI2L2dxQ1hsNpfDmwDfBvYBXhjRMxoc32drszZXI+y5nc9ypzx9SjrdqBeZd1e1KvM25WiNJLH\nQ/2cW9n2M8BXJe0KvB+4qAVtD5TT7ej3QG23vN8DZPhvhqq3hW1fRft+3/2z/Zmh3tPCthfShn4P\nkvkt7/cgbTer392gqGx2LtfftnPZuexcrqHbQ/spYELV92MkLc6P5/d7bQIwr12F1WGw2jvZoHVH\nxJiI+Brp9IO3t7u4IQz5M5f0Y2BtYHngXW2sbTCD1b0fadLzl6TrCh0cEWWoG+AUSb15T/kvgE3b\nWt3gBqv9CdJefkl6nrS3sNG9jd2qzNlcj7Lmdz3KnPH1KOt2oF5l3V7Uq8zblaIMN4+fHOI9rWx7\nHjAHuAhA0l9I29w1m9x2M9/TrPW0pd8DZHhb+j1A2237fdfI9rb9vmu03Y5+18r8Q+upt0VtN7Pf\n3aCobHYu178e57Jz2blcQ7dPMt8EvBUgIrYC7qx67X5gWkSsEhFjSaeY3NL+Egc0WO2dbKi6Tyf9\noe5TdepFpxiw9ohYOSKujYixSjexegZ4oZgylzJg3ZK+LWlzSTOALwEXS/p+MWUuZbCf90Tg7ohY\nKdI1jncGbi+kytoG+5z/DRgfEVPz99sDd7e3vI5X5myuR1nzux5lzvh6lHU7UK+ybi/qVebtSlGG\nm8c3D/GeVrZ9C/Bu8jUEI2It0lE5/2py2818T7PW065+18rwdvW7Vtst7/cg2d7yfg/Sdsv7XSPz\nL5J0/mDvaXHb3yfdmLUZ/e4GRWWzc9m5PFTbzmXn8qD97qlUynhfl/rk/0T03T0R4HDS9fjGSzoz\nIt5GOiR8DHC2pFOLqXRpQ9Vetdw1dNDd0Qerm/SfudtJF83vc4qky9pa5ADq+Ly8j/RHtgi4A/hg\nDp5CDeOzcigQko5rf5VLq+PnfQjwIdKdT6+SdFIxlS6tjtr7wrkHuEnSR4uptDOVOZvrUdb8rkeZ\nM74eZd0O1Kus24t6lXm7UpRG8rjWexrJsQbbXg44D3gl6aY0n5R0a7PbrlruxZxuV78HaLvl/aZ2\nhv8/4Gf939Psfg/S9i9pw++7Vrbn5drxOa/V9rK093P+YuYX8Dmvbrspn/NuUFQ2O5edy3W07Vx2\nLg/a766eZDYzMzMzMzMzMzOz1ur2y2WYmZmZmZmZmZmZWQt5ktnMzMzMzMzMzMzMGuZJZjMzMzMz\nMzMzMzNrmCeZzczMzMzMzMzMzKxhnmQ2MzMzMzMzMzMzs4Z5ktnMzMzMzMzMzMzMGuZJZjMzMzMz\nMzMzMzNrmCeZzczMzMzMzMzMzKxhnmQ2MzMzMzMzMzMzs4Z5ktnMzMzMzMzMzMzMGuZJZjMzMzMz\nMzMzMzNrmCeZzczMzMzMzMzMzKxhnmQ2MzMzMzMzMzMzs4YtW3QBZrVExDLAh4GZpM/pWOBy4HOS\nniuytmaKiN2BN0g6YZBlXg38AdhF0h/bVpyZWT/OZoiIPYDzgAfzUxVgB0kL2lehmdkSzmaIiEnA\nt4ENgBWBL0i6sM0lmpm9aLRnc0RsCFxU9dQywEbAvpIua2OJ1kY+ktk61anAlsDOkjYFtgACOKvQ\nqppvC2DSQC9GxArAhXiHkJl1BmczbAN8VdKm+d9mnmA2s4I5m/POP0mbAbsA34qItdpVmJlZDaM6\nmyXdWzVe3hT4DXCxJ5i7W0+lUim6BrOXyEfu3gWsUf0f94hYHdha0mURMRH4LrAJ6SiyXwHHSXoh\nIp4FvgG8DVgZOBbYH9gYeATYQ9LCiHge+CYwA1gpv/8nua3PAgcBzwNzgA9IejQirgVuBrYFXgnc\nABwqqRIR2wBfyutaDJwo6RcRcRiwD/ACMA14DngXMB64jLRH73RJn63xszgbuAX4DPB2H8lsZkVx\nNr/Y3+vysqsAC4HPSLphpD9fM7NGOJtfPIr5n8AkSc/m59YDHpb0nyb8mM3MhsXZvNTPY3vgAmAj\nH5zR3Xwks3WizYB7+oePpEer9np9C5graWNgc1IwfyK/NhZ4RNJrge+R9hR+GNgQmAjsmZcbAzwh\naXPgAOCciHh5RBwOvAXYXNImwN2koyP6rCtpR1LA7wzsEBGrAOcCh0h6PbAXcGpETMnv2YEU6hsD\nNwHHSroNOA24dIAJ5vcCy0jqtj2dZlZOzubkceA7ub5ZwE8iYu06f4ZmZs3mbIb1gH8BH4+IGyPi\n98DrPcFsZgVyNr/U10gT4J5g7nKeZLZO9AJDfzbfAnwHIF/P6DRgt6rXf5S//g24S9K/JFWAv/PS\nUzn61nEXaU/jDnnd51QNTL8FvDEiliPtYbw8v2cB8ACwKrA1sAbw04j4E/AL0p6/jfN7/iDpkby+\nP1bV0JP/vUREbAYcCRxV9fRSy5mZtdGoz+a8/rdL+ml+fBPpSJA3DfFzMTNrFWczLAe8GpgvaTvS\nkXvfzONpM7MiOJuzfHT0qpIuHuLnYV3A13m1TvR7YIOIGN/v1JK1gdOB/UiBXR1ky/DSz/N/qx4v\nGqStF6oej2HJxqCn3/PLVj1XfVREJT8/BrhP0lZV9a4JzAUO6fceGHrC+J2k02JujgiAtYCLIuIT\nkn4+xHvNzFph1GdzPq3xGElf7FdH19y8xcxKZ9RnM+nUcchH6Un6a0TcCLyBNBFiZtZuzuYlDgTO\nr3NZKzkfyWwdR9LDpLuQnhMREwAiYmXSaSKP52utXQkck19bHjiCdCH54XpXXsdmwPrAtXndh0fE\nuLzMh4DrtOQOsP3DtALcCkzL1xoiIjYF/gKsOUT7i0hHX7yEpI9KCi25SP4jwMGeYDazojibAVgA\nHB0R+1atbwvgirp7ZmbWRM5mkPR30mTyYXl9q5Nu0vr7+rtmZtY8zuaX2BG4uq6eWOl5ktk61dHA\nvaQjef9ECry7gffm1z8ErBYRdwF3AvcBX8ivVd/NstLv+/62iYg/AGcDB0ianx9fBfwuIu4FvDV9\nQgAAIABJREFUXge8o986X0LS48Dbga9GxJ9Je+oOkfRQjRqqv78a2DMiThmkRjOzTjGqs1nSC6Tr\n030i9/GcXF/vIH0xM2u1UZ3N2T7AmyPibuAa4CRJfxikL2ZmreZsTtYD/jFI/dZFeiqVwT6rZt0r\nIhYDL/fkgJlZ53A2m5l1HmezmVnncTZbp/GRzDaaeQ+LmVnncTabmXUeZ7OZWedxNltH8ZHMZmZm\nZmZmZmZmZtawZYdexKx1IuIfwL6SBrzzc0ScBDwg6YImtflz4IeSmn6H03ytpR0lPdXsdTdQy0TS\ntZiCdNbC+ZK+UmO5FYHvApvn5W4Djsk3IzCzUca53Dr15m29+W1mo4ezuXWczWbWKGdz60XEFNL1\nrDeudVmQiPgh6brPfdYFrpG0d5tKtCq+XIYVbchD6SWd0KxArmqzJYfwS9q0gwL5ZOBBSRsDWwBH\nRcSWNZb7DDBG0muB1wIrArPaV6aZdRjncuvUm7f15reZjR7O5tZxNptZo5zNLRQR7wJuANYYaBlJ\n++e6NwWOAHqBY9pUovXjI5lt2CJiD9JgbCywEPiEpFsj4lxgnKQDI+I1wG+BnYADgdcAq+d/fwbe\nK+npqnX2AP8P2BKYAPTkZW6OiPOAuyR9PSKeBWYDbwLWAk6RdEpex3uAo0g7T54APiBJEbEW6c6o\nawL/B6xWo08TgQeB6ZIezc/dCpwA/J10dMNKuc0/AwdK+m9E/Be4DNiEdLfW3wMvB/4LnApMAyYB\nTwMHS5oTEdcCNwPbAq8kheahkioR8Tbg87kPzwDvl3RnRGwDfCnXsBg4UdIvBvs9SfpwRCyTv10b\nWB6YX2PR63IfkbQ430l2g8HWbWadxblcjlymzrwdRn6bWQdzNjubzazzOJvLkc2533sBuwH3DLZs\nXn5s/jl9RNLDQy1vreEjmW1YImIa8AVgN0mbAUcCP86nmR0DbBIRhwKXAh+WdF9+65bA24H1geeB\nz/Vb9ZbAGpK2kvQa4PvAp/Nr1XvqxgJzJW0H7Ad8KSLGRsSOwLuA7XNdXwV+nN/zXeBmSRsBH8o1\nvISk+cBPgENyPzcg7S37NfBe4FxJ25BOw3g18Nb81uWAn0laX9Ifqlb5FqBX0taSghTWH6jqz7qS\ndgQ2BnYGdoiI1YELSAG9Se7D7Ih4GXAucIik15OC9tR82sigJL0QERcCdwLXAHNqLPMbSQ/kfq8D\nfBj44VDrNrPO4FwuTy4PJ2/ryW8z61zOZmezmXUeZ3OpsvkRSW+v+h0M5T3APyX9tM7lrQU8yWzD\n9SbSHrTf5uv1XAi8AKwnaSFwEHAmcKukS/N7KsAPJM2VVCFdy2zX6pVKuhX4bEQcFRFfJQX4SgPU\n0BcafyIdQTAe2J0UmDfnur4MrBIRqwBvBM7L7fyVtEeyljOBQ/Pjw4Fzcr2fAp6IiGOB00h7/8ZX\nve+G/iuS9CPg+xHxwYg4hbQHtLo/l+flFgAPAKuS9gTeLenO/NpPJO0ObEPaQPw09+0XpL1/Gw/Q\nj/61HELaG7kqS28MXxQRrweuB74t6Zf1rNvMOoJzuWS5XG/e1pvfZtaRnM3OZjPrPM7mkmXzMHwE\n+J8mr9OGyZPMNlxjgKuVr3mjdN2brVly+sL6wOPAZhGxXNX7Xqh6vAxp79+LImJ3loTNZaTwG+jz\n+R+AHJiQTkUZA1xQVdNmpGulPUnaKFSv6yVt95F0E7BsRLwBmAmck1+6FHgf8A/gG8Afc5t9FvRf\nV0QcBZyVX7sIuKTfe/5T9biSX1tEv2srRcRrc+331fiZ/7pWP6reu2tErJn79kzux2YDLHtQXt+n\nJH1psPWaWcdxLpckl/P7h8zb4eS3mXUsZ7Oz2cw6j7O5RNlcr4jYFFhW0vXNWqc1xpPMNly/Bd4c\nEQEvhumdwPIR8SrSdYh2Ae4n7X2DFDh7RcTKETGGFHCXV62zJ7/nckmnA38A9iGFd9/rg6mQAmpm\nRPRdEP4o0sajAlxBugA8EfFKYMYg6zoL+DZwh6R/5ufeDJwsqe+0uS2raqulJ7/nPEnnkk6j27Pf\ne/r3qUK6k/UGEbFhrnVv0ukmtwLTImL7/PymwF9Ie2AHsz/pGkxExPLAAcDV/ReKiP2AU4A3Ve2t\nNbPycC6XJJeHkbd15beZdTRns7PZzDqPs7kk2TxMOzLwEd7WRr7xnw2LpHsj4gjg0kgXt18E7AE8\nB1wMfCUvcwxwV0RcRQqcR4Ffkk4tuw74YtVqK6Q9fRdHxB2kvYTXA/vmNir9lqX/95J+HRFfBn4T\nEYtJN+HYJy9zDHBuRNwL/JN0WspAzs+1HVT13HHATyKil3RjgGtJp7IMVE8F+BpwRkQcnvvzB2Cj\nQfqBpMci4h3A+RGxbO7DgZIej4i3A1+NiBVIO4cOkfTQIP0A+DhwWkTcldv7iZbcVOAkoCLpRJb8\nLs7O21qAGyV9cIj1m1kHcC6XKpcHzNucy0g6gUHy28zKwdnsbDazzuNsLlU296/rRf2ymdyfvw9j\nfdYiPZXKUp8Ns6aKiBOBVT1paWbWGZzLZmadx9lsZtZ5nM1m9fPlMqwdqu+mamZmxXMum5l1Hmez\nmVnncTab1clHMpuZmZmZmZmZmZlZw3wks5mZmZmZmZmZmZk1zJPMZmZmZmZmZmZmZtawZYsuoCjP\nP/9CZd68hUWXMWyrrDKOMtYN5a29rHVDeWsva90AkydP6Cm6hrIqMpeL/My5bbfttlvLuTwyZR0z\n16vMY456dHP/urlv0P39czaPTFmzucyf67LWXta6oby1l7VuGHk2j9ojmZdddpmiS2hIWeuG8tZe\n1rqhvLWXtW4bmSJ/727bbbvt7m3bRqbbf3fuX3l1c9+g+/tnI1PWz0dZ64by1l7WuqG8tZe17mYY\ntZPMZmZmZmZmZmZmZjZynmQ2MzMzMzMzMzMzs4aN2msyz5kzh97eBUWXMWzz5o0vZd1Q3trLWjeU\nt/bh1D1lyjqMHTu2xRVZOxSZy0X+rbhttz0a23Z2l0dZx8z1KutYqV7d3L9u7ht0Rv+c1Z2rrNnc\nCZ/rRpW19rLWDeWtvci6i87tUTvJ/M5ZFzNu4mpFl2FmI7Bw/mOccuyeTJ06rehSrAmcy2ajg7O7\nXJzNZqOTs7qzOZvNrL9OyO22TzJHxK7AKyWd2e62q42buBrjV1m7yBLMzEopIg4DQtKsYb7vVuAA\nSQ/Wet25bGbWeZzNZmYDK2p+w9lsZp2o7ZPMkq5sd5tmZtZUlTa/z8zMzMys43h+w8xsiSKOZD4M\neAvwauBB4FXApcBGwKbALyR9JiJ2BD5HujnheOBgSX+JiM8CewNzgXHAZ4E/A2cDk3IzH5J0d7v6\nZGY2GkXEx4EDgeeB6yV9OiJOBLYBVgLeA7wT2BV4CHh5QaWamZVSREwHzgUWkcbEZwCHAi8AawBn\nSPreIOPm44G9SGP+UyWdEREfBGaSdvxdKunb7e6XmVm36DvDD3iEftkaEfsCnyRl+CPAQcAJwL8k\nnR4R65OyeUZEfAHYiZTXP5L0lbZ3xsxshMYU0GYl/3sV8G7gbcDngY8CW5ImJQA2BA6RNAP4MbB/\nRGxCmqDenDTRvGZe9jjgKkk7A0cCp7alJ2Zmo9d0YH9ga0nbANMiYndSvt8jaVvSRMf2kjYH3gVM\nKKxaM7Ny2gW4NX89AZgIrAXsAWwFfDQiJlN73Lwpadz8hvxvWkRsCBwAbAvsAOydJ7LNzKwxFWBd\namfrQcBXJG0P/BxYmYHP7DuYNEm9PfBkq4s2M2uFIm/89zdJT0fEIuBRSU8CRERf6D4CfCsiFgBr\nAzcB6wO/k1QBno2I2/OyGwEzIuLA/P0qbeuFmRVq0qTxTJ7sucs26wE2AS6X9EJ+7gbgNfnxnPx1\nOnA7QM77u9papZl1LGd33c4GPgVcAcwHfg3cLGkRsCgi7gamUnvcPJ0l4+ZFwLERcQCwDvDbvP6X\nAeuxJLfNzF7krK5LD+kguOVYOls/BsyKiA8B9wGX1Xhvn3cAXyadpfKrVhZsZt2r6NwucpJ5KGcA\n60p6JiLOIwXwPcAHI6IHGEu6vAbA/cCFki6JiNVYcjS0mXW53t4FzJ37dNFlvGiUDMQrpMsUbRUR\nywCLSUdtfJ80+bw4L3cPcHTO7HGkI+3MzNqa3SXP5b2AGySdHBEzgS8AT0TEGGAF0s69vwA/Zelx\n8/3AUTmDlwV+AXyCdLbJbgAR8RHgzjb3ycxKopVZXfJsrlYB7gBWkPRWeDFb7wKOAE6UNDciTgP2\nAZ5lyRnZm+XlxwL7S5qZM/ueiLhE0kNt7ouZldxIc3uk2VzkJHNliMcXAjdGxDPAo8Caku6OiF+S\nTht8nHRUxnOkAffZEXEE6RSUE1pdvJnZKPcX0pFyN5EuvXSDpMvyZY0qAJLuiIhfAb8nHWX3aFHF\nmpmV1O3A+RHxHLAM8G3SNZl/BawKfF7SExFRa9x8R0RcwZKc/p6kOyPi6oi4EVgeuI2Uz2Zm1jiR\ndgBWZ+vDwO+An0fE08DTwOWkyx79IF9L/w9ARdJzEdEbEbcC/wGu9ASzmZVRT6Uy0CWBOk++5tx+\nkk6NiOWBu4EZkv453HVt+fYTK+Mmrtb0Gs2sfRbOf4xTjt2TqVOnFV3KiyZPntAz9FJWi3PZbHRo\nd3Z3Uy5HxE7AkZJmtqtNZ7PZ6NTqrO6WbI6I9wGvkNTWA92czWbWXzNye6TZ3MmXy6jlcWCLiDic\ndKTcmY1MMANcMPtgensXNLW4dpg0aXwp64by1l7WuqG8tQ+n7ilT1mlxNdYuReZykX8rbtttj8a2\nnd0N67uBdtuUdcxcr7KOlerVzf3r5r5BZ/TPWT24iHgr8CHgyHa3XdZs7oTPdaPKWntZ64by1l5k\n3UXndqmOZG6ySiddx7VekydP6Kjrzw5HWWsva91Q3trLWjd0z1EZBSksl4v8zLltt+22W962c3lk\nSjlmrleZxxz16Ob+dXPfYFT0z9k8MqXM5jJ/rstae1nrhvLWXta6YeTZPKZZhZiZmZmZmZmZmZnZ\n6ONJZjMzMzMzMzMzMzNrmCeZzczMzMzMzMzMzKxhnmQ2MzMzMzMzMzMzs4Z5ktnMzMzMzMzMzMzM\nGuZJZjMzMzMzMzMzMzNrmCeZzczMzMzMzMzMzKxhyxZdQFHmzJlDb++CossYtnnzxpeybmhv7VOm\nrMPYsWPb0paZNUeRuVxkto72tp3XZp2trGPmepV5bF2Pbu7fcPvm7Y11k7Jmc5kzqay1D1W3s9Ga\nadROMr9z1sWMm7ha0WVYCyyc/xinHLsnU6dOK7oUMxsG5/Lo47w263zOZusG3t5Yt3E2WzM4G63Z\nOnqSOSJWAO6T9OoBXj8COEfS88Nd97iJqzF+lbVHWqKZWdeIiGWAq4DlgN0lze/3+g7APEl3RcS/\nJa3RzPady2ZmSUQcBoSkWS1Y907AkZJm1rO8s9nMrLaIWB44RNLZEbEK8BZJl0TEecAlkq5sVdvO\nZjPrRGW/JvMsYJmiizAz6xJrAxMkbdd/gjl7N7BWflxpX1lmZqNOKzPW+W1m1hxrAu/NjzcB9syP\nnbNmNip13JHMETEeuAh4GfBAfm4H4ATSpPh44GBgB2AN4JKI2A84A3gFKeh/Jumz7a/ezKzUTgOm\nRcRZwGrACqRMPR54CNgVeF1E3AssHxEXAa8EngD2A1YCzgYm5fV9SNLdEfF/wH3AvZI+1s4OmZmV\nWUR8EdgcWBW4Q9K7I+JE4F+STo+I9YFTJc2IiDuBa4HXkiY49gKeBr4NbAGMJY2n55Oy/pekrL9c\n0knt7ZmZWWfJZ5DswZLx7ymkHN0I+ARpzLsPabz7eH78GWDDiPgssD2wSUS8L6/yyIj4JDAROErS\n7yPig8BMUkZfClwMXCVp04jYCvilpEkR8QrgLElvaUPXzcyaphOPZH4/cKekHYHT83Mbkk5DmQH8\nGNhf0tnAv4GDgCnALTmEt8zrMDOz4TkKuJc04P26pDcDRwDHSPojcAXwSUkPkXb4zZK0PWnwvClw\nHGmgvDNwJHBqXu8rgJmeYDYzG5axpEsUvZk0SbxVRKzFwEfITQAulrQT8DCwG7A3sKqkLYEZpAlr\nSJMoe5EmRT7Qsh6YmZXLSpJ2B75MmhjelzQWfg+wCrCLpK1IB+ttAfwP6SCKzwNfAK6WdGZe1+2S\n3kja0XdYRGwAHABsSzpgbm/SDsQn8qTybsD/RcQWpCOif9yWHpuZNVHHHckMTAN+ASDpdxGxCHgE\n+FZELCCdzn1jv/fMA7aIiBnAU8DybazXOtCkSeOZPHlC09bXzHW1W1lrL2vdJdeTv/4bOD4i3kOa\nzKi1reiV9GDV8uNIR3rMiIgD8/Or5K+PS5rXopqt5Jqd1/UqMmPcttWpAqwWERcDC0g795brt0xP\nv+//lL8+RJpIfhVwC4CkJ4HP5Wsy3y1pEbAoIoZ9bxOzMipqezMSZau35CrAn/Pj+aSz8ACeJO30\nW0Q6i3oB6QCKZXlpBvf0+/4P+Wv1OHkd4Lf5+ZeR5j5+AuwObA18CXhzfnx4k/plNqhOz8ZOrm0w\nZa17pDpxkvleUqj+LCI2JQX6mcBUSQvyRfT7jsBeTLom82HAk5LeHxHrkfY22ijW27uAuXOfbsq6\nJk+e0LR1tVtZay9r3dAVG5Me4PPAmZKuiIjDgUPza32ZC7WPpLsfuDDf8GQ10jWc+95nVlMz87pe\nRWaM2y6m7RLqIR15/BdJB0XEZNKp2T3As6RTuQE26/e+/tl8H7A/QERMBP4XmF1jObOuV8T2ZiTK\nPB6uR4dm80DZuDywt6StImIccDspjxfz0rmJWmeK9008C7hH0m4AEfER4A7SxPbFwGPAlaRLc8yT\nNHfk3TEbWidnY1lzsKx1w8izuRMvl3EasG5E3AAcDfwHuAC4ISJuJJ0K2DewvoF01PNVwFsi4jrg\ne8CciFhzqTWbmdlQKsAPgK/lTN2FdCofwG3A7HwN0P6D8ArpNMEDIuIa4FfA3VWvmZlZ/SrA70hj\n4uuAHwJ/JY2B/xd4a87aTRk4YyuSfgbMy+PqK4BvVq2fGo/NzEazStXX6seLgAV5PuLXpDOt1wIe\nBcZGxGzS/aQ2jogP11qXpDuBqyPixoj4PbAe8LCkh0mT2FfnM04Wkc/sNjMrm55KZXSOK2e8+3uV\n8ausXXQZ1gIL5j3M7CO2YurUaU1ZX9n3QpWx9rLWDTB58oT+py5bnZzLo0+z87peRR9V67bb3rZz\neQSczdYNitrejESZx8P1cDaPjLPZmqHTs7GsOVjWumHk2dyJl8toi4XzHyu6BGsR/27Nysl/u6OP\nf+dmnc9/p9YN/Dm2buPPtDWDP0fWbKN2kvmC2QfT27ug6DKGbdKk8aWsG9pb+5Qp67SlHTNrniJz\nuchsHe1t///s3Xm4ZFV18P9vMzR0c5uWxkag00Fou5cSNTGOoIwSwRgGjWhACTiESRM1iQMqIjgg\nicMLSVQkKIhAfvqGvNGoIA5AgzjFOcBqB4wgQjfetqVF5vv7Y58L1Zc736o6daq+n+fpp+rWGfba\nl7rrbFbts8t8LfW2po6Zp6vJY+vp6Of+zbRvXm/UT5qam5uck5oa+1RxmxvVTgNbZF61alUjp683\nfNp9Y2OX1Hl15uUBXkZgINuWNH1NHTNPV7/non7uXz/3TZpKU3Nzk/9umxp7U+NWM/XiF/9JkiRJ\nkiRJkhrCIrMkSZIkSZIkadYsMkuSJEmSJEmSZs0isyRJkiRJkiRp1iwyS5IkSZIkSZJmzSKzJEmS\nJEmSJGnWLDJLkiRJkiRJkmZti7oDqMuaNWsYHt5Ydxgztn79UE/EvXz5LsyfP7/uMCT1kTrzcp25\ndZDb3mab7b2WSD2uqWPm6aojDzqOljRXTcrN5jxpcDSiyBwRjwLelpmvatc5jzrpIhYu3qFdpxso\nd25Yy5mvP4QVK1bWHYqkLoiIY4DIzJMm2edoYDgzPzPB9mOBj2bmfROdw7w8WLyWSDMTEcuBP8zM\n/5pg+77AcZl5RET8MjN3iogPAO8HNgIHZebFM23X3Nxe5j5JoyLijcCXM/ObLa9tDVyfmbtOdmxT\ncrM5TxosjSgyZ+ZtQNsKzAALF+/A0HbL2nlKSepXI1PtkJnnT7HLScD5wIRFZvOyJE3q2UAA4xaZ\nx5OZr4MHC9CHADMuMpubJakzMvOM2R5rbpbUiyYtMlez1w4DhoBHAu8ATgUSuBs4AbgQWFSd662Z\n+ZWI+DPgbcA84NvA8cDewDuB+4GfAMcBuwEfA+6lrA99ZHXe/686duvq2A3AxZm5R0R8H7gCeCKl\n8HEocAfwL8CTgVuBXYGDM/N/5/C7kaSBFBELKLn594H5wP8FnhERlwFLgQ9l5jkR8UPK9eAe4AZK\n/v134JNsmsOfAuxIKW68oLu9kaTeFhGr2HQ8/FLKOPr3gJ2ATwOnAG8CFkTENcD/AmdScu2vgJdP\ncO4rKGPutwBPjIhXApcBZwMLgN8Bx2bmzR3qniQNnOnk9cw8OSLOo4yPr6HUVR4B/LiOmCWpHabz\nxX8LMvMA4DmU2+0WA6dl5pHAycBlmbkPcDhwbkRsAfwT8KeZ+VTgR8By4CPA8zNzX+AXwDHAAcDX\nqsdTqnM/FbgdeC5l9vI2bDqLbhFwUct5nkuZmbEkM58OvKJqb8qZd5KkcR0P/DQz9wT+glKEuDcz\nDwSeD7y22m8byvXgiJZjn8aYHJ6Z51IK0H/RpfglqUnGjocXAddm5kHA04HjM/MB4HTgwmq5jHOA\nEzNzP+BzwBsYf+w7Uv17J+WW7H8F3gucVR37PuA9neycJA2gKfN6td9o3j4e+H5VVzm7y7FKUttM\ntVzGCHAVQGaujYj1lNv0str+WOCCavstEfEbYGdgfWbeXr3+3ojYgfKJ3aciAsrMiS8A7wLeCFxK\nma38ZuDzwErgPymf/L2TMkuj1Xeqx5soM+UeDVxbtXd7RNwwk1+CZm7JkiGWLl004+Nmc0wvaGrc\n0NzYmxp3n1hFycVk5o8jYgPlrhSA24CFLfvmmGPHy+HSuGZ7LWmHOnOMbWuMc9l0PHwq8NSI2A/4\nDbBVtd88HhoTPw74UDWu3hJYM8n5W48DeALw5mot0HmUu1HURd3Off38t9fPfYP+718fmyqvj/0W\nvJXAZwEy8xsRcW8XY+24sTmvye/rpsbe1LihubE3Ne65mqrIPI+yBMXol+8tAtYCD1Tbr6csg/G9\niFhGub3jl8AjImK7zFwfEWcCnwBuBg7JzDsi4mDKF5AcCqzOzNMi4ghKIr4A+GVmHhgRewDvBl42\nJq6xMzV+CBwFnBkR21EKJOqg4eGNrFt3x4yOWbp00YyP6QVNjRuaG3tT44a+uZhcT7mr5NMRsRvl\nA8GJ1lx+YMzP+/LwHL5/td/mnQlXTTWba0k71JljbLuetnvc2PHwd4F/yMzjI+IxwLHVfvfz0F2I\nNwBHZebNEbEnZTLHZFqPvR54b2ZeG6VKvU87O6OpdTP3NXlMNZV+7hsMRv/62HTz+qjrgD0oY+8n\nUT487ButOa/J7+umxt7UuKG5sTc1bph7bp7OTOYdI+KLwLbAicCHWra/G/hoRLyQMjv52My8NyJO\nBD4bEfcD387Mb0bEa4DPRcRmlE/z/rI65/kRcQ9l4Ps64OfAv0XECVV8p7bEMm6MmfnZiHhutUbd\nrcCdlBl0kqSZO5uS26+gFIbfR1mXf9SE+Rj4HuPn8NWUGRr7dyJgSWqwb7HpePiZlFnKe1C+q2RN\nROwE/AB4S0T8N+V7US6olqkboazJvIyH8nNrnh6hfB/KEyLib4C/r86/NWX8/jed7qAkDZjp5PWd\nq31HgA8DH4+I1ZQPEe+qIWZJmrOpiswAV2bmSS0/7zb6JDPXU9bn3ERmXkq5NaT1tcuBy8fsejuw\n1zhtPmec1/aszrNryzlPAqhmYazOzFdHxPaUmc23T9InSdIEMvNu4CUTbLuL6jowJh+f2rLbw3J4\nZh7T3iglqT9k5k95+Hj4j8bZ9ZeUpepG7Tdm+4+BK6tz7lw9tu6ze8vzg2YVrCRpSjPI6613bL+4\ncxFJUndMp8jchC/Quwk4IyJeS5l194bMnHQm850b1nYlsH7k705SJ5hbBov/vaVm8G+1vfx9SmqH\npuSSpsQpqT0mLTJn5kRrcPaUzLwTOGwmx1xw+pEMD2/sUESds2TJUE/EvXz5LnWHIKnP1JmX68yt\ng9z2NttsX0vbkqavqWPm6aojDzqOljRXTcrN5jxpcExnJnNfWrVqVSMX4m7yAuKSNJk683LdX0pm\n25J6VVPHzNNlLpLURP2emyU102ZT7yJJkiRJkiRJ0vgsMkuSJEmSJEmSZs0isyRJkiRJkiRp1iwy\nS5IkSZIkSZJmzSKzJEmSJEmSJGnWLDJLkiRJkiRJkmbNIrMkSZIkSZIkada2qDuAuqxZs4bh4Y11\nhzFj69cPtT3u5ct3Yf78+W09pyTNVJ15uRO5tR/b9nohDZ6mjpmnq1052PwoqZuakpvNjdJgGdgi\n81EnXcTCxTvUHUbt7tywljNffwgrVqysOxRJA8683Nu8XkiDydw8NfOjpG5rQm42N0qDp6eKzBHx\nduCXmXn2NPd/FPC2zHzVTNtauHgHhrZbNtPDJElTiIi9gfWZ+YOIuDUzd5zOceZlSWqvqcbKEbEv\ncFxmHjHROczNkjQ9EbE1cH1m7jrN/cfN0RHxnuo85090rLlZUi/qqSIzMDKTnTPzNmDGBWZJUke9\nHLgY+AEzzOuSpPaZxljZHC1JNZkkR5ubJTVSW4vMEbEtcA7wCGBn4IPAi4HvAI8HtgUOz8yfR8Tp\nwJOB7YHvZebLW87zLuAXmfnBiNgOuBw4CPgkMA/YGjge2ABcnJl7VMfsW/Xp3zPzH9rZN0kaNBGx\nJfBh4DGUL4p9K7AUOBHYkjIAfj7wBOAM4G7gi8CBwB9FxHXAVhFxIfD7wK+AF2bmfV3zbB6yAAAg\nAElEQVTuiiR1REQsAD5GyXHzgddSxqi7ApsD78/MT0bEFcB3KePhjcBqSq58BPAc4DDgYMoYdyfg\nTODQav+/z8xPR8ThwOuA+4GrM/Ok6i7APYFtgFcAL6yO2wL4EPAFHhorv5CH5+95nfrdSNIgiIgh\n4EJKPv9x9drewCmU8fMQcCSlnrFdZp4WEVtRrgmHAB+vcvSfA28B1lGuJzd0uy+SNFebtfl8K4B/\ny8wDKQPmv6UMYr+emX9CKRYfERGLgOHMfA7wVOAZEbFzy3n+FfjL6vmRwCeApwG3A8+lfNq3DZt+\nwnckcASwF/DrNvdLkgbRK4F1mbkPpQDyQUrB+XmZuRdwHaVIMgJslZl7Z+ZpwKXAGzLzJsrA+qRq\n/8XAk2rohyR1yvHATzNzT+AvgH2A2zLzmcABwDsjYnseGg8fAGwF/LYaB19XHTMCDGXm8ygf2p2Q\nmS8AjgVeVk26eDuwf5VPl0XEAdVx/1O1t4BSxHha9W/sIpgrGT9/S5Jm73jg+9V4eXTZz92Bl2bm\nfsAlwOHABcCLqu2HAJ8B7gGIiC2A9wHPrmopd2J+ltRA7V4uYy3w2oh4AfCblvN/p3q8CdgR+B3w\nqIi4iDKbY4gyqwJgJDNvjIg7IuJxlOLxwcB6yuD4P4F7gXey6eyLl1AG5TsCn29zv/rakiVDLF26\nqCttdauddmtq3NDc2Jsad595AvCsiHh69fPmlPx7fkRsBB4LXFttywnOMZyZP6+e3wos7FSw6rx2\nXy/q/Du37cFqu4NWUY07M/PH1aSJy6ufN1Z3dKyo9v129fhrSpEXyvh26+r56Hh5A3B9y75bUz7g\nWwp8PiIAFrWcd031GMA3MnOEkqtfHxGPbol1HePnb7VBN8fTM9WrcbVDP/cN+r9/fWIl8FmAzPxG\nRNwL3AKcVeXbZZS7T34dEd+JiGcBR1Mm5I3WM5ZSxszrq5+/Sp/caTJebmzy+7qpsTc1bmhu7E2N\ne67aXWT+W+DazPxwROwH/NkE+z0X+L3M/IuIWMqmt+uNPp4DvA24KTOHq/P9MjMPjIg9gHcDLwOI\niPmUZTiOiIh5wP9ExMXVLDpNYXh4I+vW3dHxdpYuXdSVdtqtqXFDc2NvatzQdxeT6yk5+PTqi0ze\nDfwdZbC8GeU27NGc/UDLcQ9QCtLgLIy+0s7rRZ1/57Y9eG130PWUu/I+HRG7UWYz3wX8v+rOvScA\nN1b7TpUPJ9t+I2WyxgGZeX9EHE251fowHsq/1wPHV2PhLShFj1fDg0vavR1YzsPzt9qgW+PpmWry\nmGoq/dw3GIz+9YnrgD0o14EnUZa6OAdYUX3YeB4P3UF+DmXZo60zc03LB4FrgUdExCMz83bK3Sg3\nd7EPHTM2Nzb5fd3U2JsaNzQ39qbGDXPPze1eLuMzwKuqdedeQ5lFMZ9NB80jwDeA3SLiSuBTwE8o\nazjTsu9/AM8Gzq1+/h7wyoj4CvAPlGIHlJnP9wDDEfE14MvAZRaYJWnOzgYeW+X0r1IG0auBrwFX\nUW7l26natzXPfx04PSIey8OLJhadJfWTsylj2iuA8yhLUGwfEauBrwBvz8x10zzXSMvjJmPnqujw\nfuCqarx7EA/NYB4ByMzvUZYruoaSqz9BWSt/JDN/U70+Vf6WJM3MhynXgdWUde9/R1kaY3VEXE25\n82QngMy8CvgDyvVi1Ehm3k/5UPCyiLic8kGhuVlS48wbGenN3BURC4ErMvNpnTj/fi//4MjQdss6\ncepG2bj+F5x+7DNYsWLssn3t19RPc5oaNzQ39qbGDbB06SJnhs2Sebm3tft6UffMVtseqLbNy3Ng\nbp5aN8fTM9XkMdVU+rlvMBD9MzfPQRNy83i5scnv66bG3tS4obmxNzVumHtubvdyGW0REXtSPhF8\ne6fauHPD2k6dulH8PUjqFeaj3uZ/H2kw+bc/NX9HkrqtCXmnCTFKaq+eLDJn5leBJ3ayjQtOP5Lh\n4Y2dbKIjliwZanvcy5fv0tbzSdJs1JmXO5Fb+7FtrxfS4GnqmHm62pWDzY+SuqkpudncKA2Wniwy\nd8OqVasaOX29ydPuJWkydeblAV5GwGuKpEk1dcw8XeZBSU3U77lZUjO1+4v/JEmSJEmSJEkDxCKz\nJEmSJEmSJGnWLDJLkiRJkiRJkmbNIrMkSZIkSZIkadYsMkuSJEmSJEmSZs0isyRJkiRJkiRp1rao\nO4C6rFmzhuHhjXWHMWPr1w89GPfy5bswf/78miOSpPaoMy+35tZ+a9trhaS5aOqYebqmysHmUEm9\nqJdzs3lTGlwDW2Q+6qSLWLh4h7rDmLU7N6zlzNcfwooVK+sORZLaoul5uRd5rZA0V4Ocm82hknpV\nr+Zm86Y02LpaZI6I3YDPAdcC7wO2y8zV3Yxh1MLFOzC03bI6mpYkjcO8LEmzFxHHAJGZJ83wuK8B\nL8rMn4+33dwsSTM3NidHxIHA72fmOe04v7lZUi/q9prMzwL+KzNfBrwQ2L3L7UuSJEn9aKTLx0mS\nJrZJbs3My9pVYJakXtWWmcwRsQr4GHAvpXB9JPA64JnVLhcB/wG8GVgQEcPA0cA9EfEoYGlm/nVE\nvAnYIzMPjYiXAL8PfBp4P7A58EjghMy8NiL+F7geuA74AHA2sAD4HXBsZt7cjr5JkqanmrFxGDBE\nydenAU8G9qVcb/49M/+hrvgkaRBExN8BLwbuA67KzDdFxNuBPYFtgFcARwEHAjdR8rUkqQMiYiml\nFvJRYCXwYeDfgJ8DK4BvZOaJEfFISt1kPpDA/pnpmhOSGqVdM5kPAL5WPZ4CHAo8OjOfQZm9fCSw\nGDgduDAz3w2cR1ky4x+Avavz7A3sHBGbAwcD/w78AfB3mXkAcAbwsmrf3wOOyMy/Bd4LnJWZ+1Xn\nfE+b+iVJmpkFVb5+DuUDwKOAI4C9gF/XGZgkDYBVwOGUSRt7Aisj4nmUGXX/k5nPpHwQuFdmPgX4\nS2BRbdFKUn/bEfhPygS8+1teXwm8HHga8KfVxLu3AJdk5r7Apxjg78+S1FztSlznAm8ELgU2AN8B\nVgNk5n3VWm+jS2PMaz0wM++KiDUR8RTgHkqxeh/KekVrImIH4OSI+B1lELyhOvT2zFxfPX888OaI\neGN1/nva1K+etmTJEEuXNuv/C5oW76imxg3Njb2pcQ+4EeAqgMxcW9218irKB4Q7Ap+vMbaBNdW1\nos6/Ndu2bbXVPOAPgc9k5mgxYzVlwgbAmupxFfAtgMy8IyJ+0NUoG6aJ4+2xmh7/ZPq5b9D//etz\n8yh3jNxCuSu71Y8z87cAEfFLYGvgsZS7wwGu7laQndDLY8+5amrsTY0bmht7U+Oeq3YVmQ8FVmfm\naRFxBPAu4NvA/4mILSm3551HGfiOeoCHku1/UGYjXwLcCLwbuKzadibwksy8ISJOBXZpOX7UDcB7\nq2U0glKk7nvDwxtZt+6OusOYtqVLFzUq3lFNjRuaG3tT44bBvZhU5lGWx6CakbEY+IvMPCIi5gH/\nExEXZ+ZNdQY5aCa7VtT5t2bbtt3NtgfECPBd4BnVXYEPUO4S/DhlDD46dv4f4MQqLy/E70iZVNPG\n22M1eUw1lX7uGwxG//rcCHA+cAHwSeCDY7aN9UNK3eT7wDM6Hl0H9erYc66aGntT44bmxt7UuGHu\nubldy2V8CzgtIr4EHAu8ALgxIr4KXAt8KjO/W+07mlD/G3h1ROwD/BclkX4BuAJ4EqXgDPAJ4FMR\ncRXltpKdxpwH4O+BUyLiCspA+ntt6pckafpGgB0j4ouUvH48sK66m+XLwGUWmCWpo35EKWZcA3wd\nuDEz/1+1bQQgM79HubPkm8DFwG01xClJg2AkM6+j1DQ+wEM1jLFF5hHKkp+HRMSXgVdSvu9Kkhql\nLTOZM/OnlPU2W313nP3Ob3n+OeBzLZu3bnm+Vct+H6Ak5LHn2rnl+Y3AQTMOXJLUbldm5kktP38B\neEddwUjSoGgdZzNm7JyZp475+V2UOw8lSR0wpvbxHjb93qg9W7btARARzwXelpnfiogDKEvNSVKj\nDOxi8nduWFt3CHPS9Pgl9a3xbv+bFvNa+/k7lTRXg5xHBrnvkrruRuCjEXEfZVnRv55s517NT70a\nl6TuGNgi8wWnH8nw8Ma6w5ixJUuGHox7+fJdpthbkrpnzCy6GaszL7fm1n5r22uFpLlo6ph5uqbK\nweZQSd2QmTfQMsN5Kr2cm82b0uAa2CLzqlWrGrkQd5MXEJekydSZl+v+UjLzuqRe1dQx83SZgyU1\nUb/nZknN1K4v/pMkSZIkSZIkDSCLzJIkSZIkSZKkWbPILEmSJEmSJEmaNYvMkiRJkiRJkqRZs8gs\nSZIkSZIkSZo1i8ySJEmSJEmSpFmzyCxJkiRJkiRJmrUt6g6gLmvWrGF4eGPdYUzb8uW7MH/+/LrD\nkKSOqTMvr18/1Jdte+2QNFdNGzPP1OLFj687BEmasV7MzY47JQ1skfmoky5i4eId6g5jWu7csJYz\nX38IK1asrDsUSeqYJuXlJvDaIakd+jk337lhLRecPsR22+1UdyiSNCO9lpsdd0qCHioyR8RWwEsz\n89wZHLMvcFxmHjHT9hYu3oGh7ZbN9DBJUoeYlyVp7iLi1szccRr7bQcclJkXT7afuVmS5iYijgEi\nM0+axr7TqouYmyX1ol5ak3kn4JUzPGakE4FIkiRJDTXd8fEfAod0MhBJEjCzusVs6iKS1BN6ZiYz\n8BZg94g4GXg6sIgS31sz8ysR8SfAO4C7gF8BLwfmjR4cER8DVgALgDMz8xNdjl+SBkJErAI+BtxL\n+bDySOB1wDOrXS7KzLMi4jxKzn40ZcB8TGZ+p+sBS1JDVLPdDga2puTNM4FDgccDfw/8PvB8YBvg\n9ur5SyjjYoC3t5zr3cC2mfnqiDickqfvB66uZtO9BXhiRLwyM/+1452TpAap8vFhwBDwSEot4lQg\ngbuBE4ALeXjd4jrgKuAPgGHgCErd4hkRcRmwFPhQZp4TEfsA76Tk5p8Ax/FQXeStmfnOLnVXktqi\nl2YyvxO4DtgWuCwz9wEOB0ZvEzkbeH5m7gtcCbyV6hPBiBgC9qIMtA+iJGlJUmccAHytejyFUgB5\ndGY+A3gWcGREPJ6So3+WmQcB/wQcW1O8ktQk22Tm84AzgBMy8wWU/PkKYDvggCrfbgE8lZJrhzNz\n78z8MkBE/COweVVgXkIpPu+fmXsByyLiAMrY+8sWmCVpQgsy8wDgOcD7gcXAaZl5JHAy49ctFgCf\nqPLtDZTC8Qhwb2YeSKlZvLba9xweqnH8AjiGqi5igVlSE/XSTObRWcmPo3wiSGbeEhG/iYgdgN9k\n5i+rfVYD7wL+q9pvY0S8lpKktwX6bhbzkiVDLF26CODBxyZqauxNjRuaG3tT4x4Q5wJvBC4FNgDf\noeRlMvO+iPgasHu17+jM5Zt5aKazuqT12jGROv/WbNu29TAjwHer5xuA66vnvwbmU+4guTgiNgK/\nB2xZbc+WczwKeAJlVhzAYygz5z4fEVBm3e025piB1u/vzX7uXz/3Dfq/fz1uhDIjmcxcGxHrgeCh\n3PlY4IJqe2vd4t7MvLra56vAc4FrgW9Xr90GLIyIpZQ7Vj5V5eYFwBc63qsOms64E5r9vm5q7E2N\nG5obe1PjnqteKjLfT5lZfT1lVvJ3I2IZ8AjKbSbbRsSOmXkrsA8tA+OI2BF4cma+ICK2Bn4eER/P\nzAe63osOGR7eyLp1d7B06SLWrbuj7nBmpamxNzVuaG7sTY0bBuZiciiwOjNPi4gjKB/6fRv4PxGx\nJbAncD5lUK0ajV47JlLn35pt23Y3226Yidbu3Ao4LDOfERELgW/x0CSN1jHvbZl5UERcEREHAv8N\n3ESZAX1/RBxNKWRvS2/d1Vibpo45pqPJY6qp9HPfYDD61+PmAU8GiIhHUT6gW8tD+fZ6YG/gey11\ni18BW0bEEzPz+5Q7/H5Y7T82t99Oyc2HZOYdEXEwcEd1/kbm5qnGndDs93VTY29q3NDc2JsaN8w9\nN/dS8lpLmaGxLbB/RFwJ/AdwbGbeB/wVcElEXA3sT1kTCWCkKjzvGBHXUD79+8d+KjBLUo/5FnBa\nRHyJcgv3C4AbI+KrlJkan2pZe3mk5dEva5WkqY2XN0cos5g3VmPhLwC3ADuPOab1+SuAf6YULN4P\nXFXdaXIQZbLGT4AnRMTfdKgfktRkI5QawxeBzwAnsumynO/m4XWL0e1vjIjVlJnKH2k534PnzswR\n4DXA56o6xgmUgvQ6YH5EnN6hfklSx/TMTObMvBt40iTbvwR8aczLV1b/yMwTZtLenRvWzjTE2jQp\nVkn9LzN/SrnjpNV3x9nvZS3PLwMum+y85rr28vcpNU9mnt/y/MG8mZnfAw6c5jl2rh5/AqysXr6w\n+tfqFh5a2mhC/ZxL+rlvktriyuqLUkftNvokM9dT1lcez8sy856Wn1tz+12j58nMy4HLxzl+wrrI\nqF7LX70Wj6R69EyRudsuOP1Ihoc31h3GtC1fvkvdIUhSR9WZl5csGerLtr12SJqrpo2ZZ+rRj340\nGzbcXXcYknrTbO7C68qde72Ymx13ShrYIvOqVasau0aKJPWjOvNy3evFej2S1Kv6fcw8f/58wCKz\npE213lkyw+N2m3qvuev33CypmXppTWZJkiRJkiRJUsNYZJYkSZIkSZIkzZpFZkmSJEmSJEnSrFlk\nliRJkiRJkiTNmkVmSZIkSZIkSdKsWWSWJEmSJEmSJM2aRWZJkiRJkiRJ0qxtUXcAdVmzZg3Dwxvr\nDmNKy5fvwvz58+sOQ5I6rs68vH79UE+1be6X1CuaMmaejeXLd6k7BEmalV7IzY5XJY01sEXmo066\niIWLd6g7jEnduWEtZ77+EFasWFl3KJLUcU3Iy91g7pfUS/o1N4/m2mXLtq87FEmasbpzs+NVSePp\n6SJzRGwFvDQzz53BMfsCx2XmEZPtt3DxDgxtt2yOEUpS/4qInwGrMvOeNpxrL+DXmfmDifYxL0tS\n7zE3S9LcRcTWwPWZuWs7zmdultSLen1N5p2AV87wmJFOBCJJA6id+fQVwM5tPJ8kSZIkSeoRPT2T\nGXgLsHtEnAw8HVhEifmtmfmViPgT4B3AXcCvgJcD8+oKVpKaKiKOAQ4DhoBHUnLr6LbHA+8DNq+2\nnZCZ10bEj4CrgQBuA/682ufDwGMoH2S+FbgDOBD4o4i4LjNv6lK3JKnvVPn6T4EFwArgDOBnwNso\neXcIOBK4CfgksC2wEHhLZl4eEa8Gng9sA9wOPD8z7+1uLySp/0XEEHAh8Ajgx9VrewOnsGm+3hdY\nmZlviIjNge8AT2nH3YSS1E29PpP5ncB1lMHxZZm5D3A4MLp8xtmUgfG+wJWUYoYzmSVpdhZk5gHA\nc4D3UwrG84Ddgb+rtp0BvKzaf1fKh357AkuBp1LuPllX5evDgH/JzG8DlwJvsMAsSW2xbWYeDBwC\nvAl4HGWJuf2ASyjj5d2A7YGDgSOALSJiHrAEOCAzn0GZvPHUGuKXpEFwPPD9alx8dvXa7jw8X18M\nHBYRmwEHAV+2wCypiXp9JvPorOTHUT4BJDNviYjfRMQOwG8y85fVPquBdwH/1f0wO2fJkiGWLl20\nyWtjf26Spsbe1LihubE3Ne4GGwGuAsjMtRGxnjJDeQS4BTg5In5HuaNkQ3XM7Zn5i+r5TcDWwOOB\nvSLi6dXrm0eE36o0Q+Pl/k6p82/Ntm1bszICfLd6fjMl994CnBURG4FlwNWZeV1EnE0pXmwJnJWZ\nIxFxL3Bxte/v0fv/P9AxS5YMAf3/3uzn/vVz36D/+zcAVgKfBcjMb1T5d7x8vTEirqTc+XcMcGpN\n8c7IbMerTX5fNzX2psYNzY29qXHPVa8PKu+nzLa+HtgL+G5ELKPcbjIMbBsRO2bmrcA+QNYWaYcM\nD29k3bo7Hvx56dJFm/zcJE2NvalxQ3Njb2rc0OiLyTzgyQAR8ShKMXlt9fqZwEsy84aIOBXYpTpm\nvDtHbgBuzszTI2IB8GZKvn6AMjNa0zA293dKnX9rtm3b3Wy7D7Xm33nAOcCKqlBxHrBZtdTRosz8\ns4jYCbgmIn4OHJqZz4iIhcC36P07GztmeHgjQGPHHNPR5DHVVPq5bzAY/RsA1wF7AJ+OiCcB8xkn\nX1f7nkO5M2VJZv6wjmBnajbj1Sa/r5sae1PjhubG3tS4Ye65udcHlWspiXhbYP/q073/AI7NzPuA\nvwIuiYirgf15aA1Rl8yQpJkZAXaMiC8CnwFOpHzQB/AJ4FMRcRVlRsZOk5zjbOCxEXEFcA3ws8wc\nAb4OvCcionNdkKSBMTLm+QXA6mpMvIiSp38E7FuNnz8JnExZE/S31X5foMyomyinS5Lm5sPAbhGx\nmjK2/h3j52sy8xuUdfYvrClWSZqznp7JnJl3A0+aZPuXgC+NefnK6p8kaWauzMyTWn7erXr8QPVv\nE5m5U8vzI1o2HT3Ovh8BPtKmOCVpYGXm+S3P76Ksjz+Rw8d57dltD0qS9DBVPePF09m3Wo/5t5Ql\njiSpkXq6yNxJd25YW3cIU2pCjJL6Sq13gZjzCn8PknpJv+akfu2XpOaJiF0pXwL40czcOJ1j6s5h\ndbcvqTcNbJH5gtOPfHAdtl62fPkuU+8kSXPUOjOuLnXm5SVLhnqqbXO/pF7RlDHzbJhrJfWCzLyR\nSe7gHk8v5GZzqKSxBrbIvGrVqsYuxC1J/ajOvFz3l5J5PZLUqxwzS1LvMTdL6kW9/sV/kiRJkiRJ\nkqQeZpFZkiRJkiRJkjRrFpklSZIkSZIkSbNmkVmSJEmSJEmSNGsWmSVJkiRJkiRJs2aRWZIkSZIk\nSZI0a1vUHUBd1qxZw/DwxrrDmNLy5bswf/78usOQpI6rMy+vXz/UM22b9yX1kqaMmVuZRyX1u7pz\ns3lW0ngGtsh81EkXsXDxDnWHMak7N6zlzNcfwooVK+sORZI6rgl5udPM+5J6TdNys3lU0iCoMzeb\nZyVNpKtF5ojYGrgBuAT4QGbeNINjfwasysx72hHLwsU7MLTdsnacSpIGSkTsCxyXmUdMY99HAxdn\n5h5T5XHzsiT1HnOzJHXGTMbUY5mbJfWiOmYyj2Tm387muLZHIknqJvO4JEmSVDg2ltRXOl5kjogh\n4ELgEcCPq9e+AhwPPBJ4H3APcCfwwurfYcBQtf20zLyk5XyPr47ZvNp+QrXvX2Xmi6p9rgH+PDNv\n7XT/JKnfRcQq4GPAvZQvjP0IsDIiPgfsAHwmM0+NiH2At1X7DAFHVsdIktosIi4ELszMz0XE44B/\nBG4FVlLy8Fsz88qIeCFwIrAlpaDxfOAJwBnA3cBHMvMTdfRBkpoiIo5hTJ0C+BXwTuB+4CfAccA8\nyrh5V0rN4v2Z+cmIuAK4Hnhstc+Lq8fR8x8OvK4619WZeVI3+iVJ7bRZF9o4Hvh+Zu4DnF29NvqJ\n3aHAvwH7AB8Ctqu2LcjMA4DnAO+PiM2r/ecBuwN/V20/A3hZZl4OPCEiHhERfwCss8AsSW1zAPC1\n6vEUYDGwNSWH7wW8utpvd+ClmbkfZVmkw3GGhiR1yjnA0dXzlwNfBW6vxtyHAf9SbVsJPC8z9wKu\nAw6k5OatMnNvC8ySNG2tdYoPAB8Fnp+Z+wK/AI6hFJpvy8xnUsbO74yI7Sl595pqnPz/AW+uXiMi\ntgPeDuxf5eplEXFAF/slSW3RjeUyVgKfBcjMb0TEfS3b3g28BfgSJSl/vXr9qmr/tRGxHlhavT4C\n3AKcHBG/AxYBG6ptnwCOAHYD/rVjvemyJUuGWLp00Savjf25SZoae1PjhubG3tS4+9S5wBuBSyk5\n9wvADzPzXuDelrx+C3BWRGwElgFX1xFs042X9zupzr8127ZtzcmVwD9FxCOBP6EUmZ8VEU+vtm9e\nFTbWAedXufmxwLXV9ux2wN0y0zza7+/Nfu5fP/cN+r9/DTPCpnWK3wKPAT4VEQALgMspE+e+WO23\nMSKuA1ZU5/hy9XgtZbLGqMdQah6fr861iFLX6FlzGa82+X3d1NibGjc0N/amxj1X3SgyXwfsAXw6\nIp5EuVVv9LaQlwLnZebrI+Ik4Fjgf4EnA0TEoygJdm21/zzgTOAlmXlDRJwK7FJt+xhlWY4FlGJI\nXxge3si6dXc8+PPSpYs2+blJmhp7U+OG5sbe1Lihby8mhwKrM/O0iDgCeBcPfSjY6iPAbpn524g4\nj+7cLdN3xub9Tqrzb822bbubbfejzByJiAuAfwIuA24GbsrM0yNiAWWW3L2U2XHLKTn5Czw0Dn+g\n60F3yUzyaJPHHNPRz/3r577BYPSvYeaxaZ1iK+BHwKGZ+ZuIOBjYSLmzby/g/0XEIsryRDdW53gK\n8GngmcAPW859I3ATcEBm3h8RRwPf6XyXZm+249Umv6+bGntT44bmxt7UuGHuubkbReYPAx+PiNXA\nDcBdlE8BR4BvAP9afQp4P6XIvC+wY0R8kXJL9omZ+UBEjN5y/QnKp4XrKYPp7QEy85aI+A3w1czs\n20GzJNXgW5RZcPdQ1pY7C3hay/bW/Hx1ldNvA3Yas12S1F7nAe+gFDF+BpxTrfu5LfAvVeHjGsqS\nR/cBw5TcfCPmZkmaiRE2rVOcUL322YjYjHK3319S7uQ7p6p/LADenpnrqhnKx0TE31KK0UcBTwRG\nMvP2iHg/cFW1VOiNwCe72z1JmruOF5kz827KovYT2aP1hyr5Xjl2ofvMHL1d5APVv/FsRrmtW5LU\nJpn5U8qMjIm271w9/t0Eu+xZbd+1/dFJ0kDbErgqM9dUPx89dofMnGgcfmXHopKk/vSwOgVliYyx\njpng+De15GsoefhKgMy8kHJntiQ1VjdmMs/GjGZWVLcErga+VBVDpnTnhrVT71SzJsQoSe1izvN3\nIGn6IuIFlKUwjutkO03LS02LV1Kj9MwdIHXmOvOspIn0XJE5M8+fxTG/o6xvNJJXDXMAABUySURB\nVG0XnH4kw8MbZ9pU1y1fvsvUO0lSH6gzLy9ZMtQzbZv3JU1HZl4CXNLpdpoyZm5lHpXUbrOpU4w5\nfr92xQL152bzrKTx9FyRuVtWrVrV2IW4Jakf1ZmX6/5SMq9HknqVY2ZJ6j3mZkm9aLO6A5AkSZIk\nSZIkNZdFZkmSJEmSJEnSrFlkliRJkiRJkiTNmkVmSZIkSZIkSdKsWWSWJEmSJEmSJM2aRWZJkiRJ\nkiRJ0qxZZJYkSZIkSZIkzdoWdQdQlzVr1jA8vLHuMCa0fPkuzJ8/v+4wJKlr6szL69cP1dr2Ntts\nb86X1JN6fcw8lmNoSYOgztxsnpU0kYEtMh910kUsXLxD3WGM684Naznz9YewYsXKukORpK7p5bzc\nSeZ8Sb2sSbnZfCppUNSVm82zkiYzsEXmhYt3YGi7ZXWHIUmNEBFbAUcBy4BbM/PsdrdhXpakh4uI\nfYHjMvOIae5/LPDRzLwvIl6Vmf8SEccAkZknzbR9c7MkzUxEbA1cn5m7dqoNc7OkXuSazJKk6dgJ\neGXdQUiSpnQSsHn1/K3V40hNsUiSJGlA9NxM5taZFtUngDcAZwBHAw8A38zM10TEcuBsYAHwO+BY\nSn8+A9wOfC4z/7GGLkhSP3oL8DjgqcBlEXE4sD1wcmb+V0S8BHgNcDfwI0pOfimb5vOOzuiQpH4Q\nEauAjwH3UiaEfARYGRGfA3YAPpOZp0bEFcCxmbkmIo4HHgXcBOwI/FtEfBNYEhH/Anyj5fx/DRxB\nKTz/W2b+U/d6J0n9KSKGgAuBRwA/rl7bGziFksuHgCOBfYGVmfmGiNgc+A5lfP0pYFtgIfCWzLy8\n232QpLnqxZnMY2dajADHAK/KzD2B66tk/F7grMzcD3gf8J5q30cBf2KBWZLa6p3AdcBpwM2ZeQDw\nWuCEiFgCvB3YLzP3An4NHIcz5yRpNg4AvlY9ngIsBrYGDgX2Al5d7deaY0eAkcz8KHAr8OLMfDcw\nnJmvAuYBRMTuwIuAZwJ7A4dVRW1J0twcD3w/M/ehTIYD2B14aVWzuAQ4HLiYkns3Aw4CvgysoEze\nOJjyIWDPTQaUpOno9eQ1r3p8OfD3EbErcG31+hOAN0fEG6uf76n2vTEz7+t6pG22ZMkQS5cuGnfb\nRK83QVNjb2rc0NzYmxp3H5vX8vzb1eNtlNkWuwH/k5m/rV6/CngO8PUJjtcYk+X8Tqvzb822bVvj\nOhd4I3ApsAH4AvDDzLwXuDcixhvnTjVxZLQg/QfALpSiBpQZd48B1sw16F4xm3za7+/Nfu5fP/cN\n+r9/fWYl8FmAzPxGRNwL3AKcFREbKd9rcnVmboyIK4EDKZPpTs3M6yLibEoBekvgrDo6MF1zHbc2\n+X3d1NibGjc0N/amxj1XvVhkvouy9ifAH1ePfwUcn5l3R8SlwJ7A9cB7M/PaiAhgn2rfB7oabYcM\nD29k3bo7Hvb60qWLxn29CZoae1PjhubG3tS4oa8vJvfzUBFj7AzlG4HdI2JhZt5JuQ0wGT+faxwT\n5fxOq/NvzbZtu5ttN8yhwOrMPC0ijgDexaYf2o26C9iZUiD+Y8pSGVDGwqNrMs8b85iUDwWfCxAR\nrwW+3/Ye1Gim+bTJY47p6Of+9XPfYDD612euA/YAPh0RTwLmA+cAK6rC8nk8NJY+B3gTsCQzfxgR\njwcWZeafRcROwDVUBeteNJdxa5Pf102NvalxQ3Njb2rcMPfc3ItF5kspt1+vBv6bMoPjB8DqiLgD\nuJlyC+FNwIeqdT4XAH9THe/t2ZLUfmspg+WtGXOLdmb+KiJOAb4SEQ9Q1mR+AyU3j83nkqTJfQs4\nPyLuoRSLzwKe1rJ9NAefBXwwIn4O/KJl+2pKcWJ/4LqIuAD4IiVffz8ivhQRVwNbUYrXt3S0N5I0\nGD4MfLwa995A+d6o/0upY/yWcgfgTvDgTOcVwD9Xx/4IOCUiXkQpRJ/c7eAlqR16rsicmRsos+DG\nOnfMzzdS1jAaa892xyRJgy4z7waeNOa1GyhFDDLzYsotfq3uYfx8LkmaQGb+lLL28kTbd64ePw98\nfpztx7Q833+c7e+lfLeJJKlNqrHyi6ezb7Ue82+pxs7VsYd3LjpJ6o6eKzJ3y50b1tYdwoR6OTZJ\n6pRBzX2D2m9JzdCkHNWkWCUNpup7pi4BPpqZG2d7nrrynXlW0mQGtsh8welHMjw865zeccuX71J3\nCJLUVXXm5SVLhmpte5tttq+lbUmaSq+PmcdyDC2pl2XmjYy5O3A26szN5llJExnYIvOqVasauxC3\nJPWjOvNy3V9K5vVIUq9yzCxJvcfcLKkXbTb1LpIkSZIkSZIkjc8isyRJkiRJkiRp1iwyS5IkSZIk\nSZJmzSKzJEmSJEmSJGnWLDJLkiRJkiRJkmbNIrMkSZIkSZIkada2qDuAuqxZs4bh4Y11h8Hy5bsw\nf/78usOQpNrVmZfXrx+qpe3ly3fpepuSNBO9MmYe5dhZkurNzeZhSRMZ2CLzUSddxMLFO9Qaw50b\n1nLm6w9hxYqVtcYhSb2gF/JyN41eA5Yt277uUCRpQr2Umx07S1JRV242D0uaTGOKzBHxRuDLmfnN\ndpxv4eIdGNpuWTtOJUmahoj4GvCizPz5eNvNy5I0OxGxNXB9Zu46wfZjgY9m5n0zPbe5WZK6KyJu\nzcwdJ9vH3CypFzWmyJyZZ9QdgyRpTkbqDkCSBtRJwPnAjIvMkqSuc8wsqZFqLzJHxDFAZOZJ1SyM\nG4AzgKOBB4BvZuZrIuI84GJgJ+BPgQXACuCMzDw/Ip4G/DNwB7AWuCszX9bt/khSv4uIBcDHKfn4\nJmBv4HmUHHwfcBfwV5l5U0S8Cziw2u+R9UQsSf0nIoaAC4FHAD+uXtsbOIXy5d5DwJGUHL0jcHFE\nvBD4CPB7lBz+6cw8ufvRS1KzTHf8C6wDPglsCywE3pKZl0fET4GvUWoYPwReWe1zLrCkauZvMvOH\n3eqTJLXbZnUHwMM/pRsBjgFelZl7AtdHxOYt+40A22bmwcAhwJuq1z8MHJ2ZzwZ+0vGoJWlwHQv8\nJDOfBbwdeBRwDnBiZu4LfBB4f0Q8GdgrM58C/CWwqJ5wJakvHQ98PzP3Ac6uXtsdeGlm7gdcAhye\nmecCtwJ/ASwHrs3Mg4CnV+eQJE1tWuNfYDdge+Bg4Agemti3DHhrZj6d8iHgYcCbgS9m5v7AccCH\nutUZSeqE2mcyjzGvenw58PcRsStwbcvro75bPd4MbF093ykzr6+er6YMpHvekiVDLF06s7rLTPfv\nJU2NvalxQ3Njb2rcA+KxwKUAmZkRcTslB3+/2r4aeA+wCvhWtd8dEfGDOoLtZUuWDAH1vt9t27YH\noe0+tRL4LEBmfiMi7gVuAc6KiI2UgsbVY45ZDzw1IvYDfgNs1cV422I2Y+ep9Pt7s5/71899g/7v\nX8NMa/ybmddFxNmUu7C3BM6qtv88M39aPf8qEMDjgf0i4sXV69t1oR9zNtc83OT3dVNjb2rc0NzY\nmxr3XPVCkfkuyi0nAH9cPf4VcHxm3h0RlwJ7jjlmvDWKboqIx1WF5j06E2r7DQ9vZN26O6a9/9Kl\ni2a0fy9pauxNjRuaG3tT44aBuZj8kJJn/zMiVlCWwfh2RDwhM38A7AMkcB3wqoiYR7ldcPe6Au5V\nw8MbAWp7v9f5t2bbtt3NtvvUdZRc/OmIeBIwnzKrbkVmbqyWmhu9a/EBYHPK3YK/zszjI+IxlJl5\njTLTsfNUmjzmmI5+7l8/9w0Go38NM63xb0Q8HliUmX8WETsB11A+EFwWEY/KzNuAZwIXAEuBT2Tm\nxRGxA2WyXc+bSx5u8vu6qbE3NW5obuxNjRvmnpt7och8KXBCRKwG/hvYAPwAWB0Rd1BmK38daF1f\neWSc5ycCH61mbtwD/KLTgUvSgDoXOC8irgT+F/gd5cPBf64KyvcCr8jMn0XE54FvUmbX3VZXwJLU\nhz4MfLwaQ99AycX/lzKG/i0l545O5FhNKXK8GrgoIvYA7gbWRMROmfnLrkcvSc0yrfEv8EvglIh4\nEeWDvtF17++u9h1dtugzEXENcG5EHEtZn/mUal+/+E9SI9VeZM7MDcC+42w6d8zPD/sSv8y8i7Lm\nEcDTgIMz8/aIeAcliUuS2u9JwLnVl5isBPbIzO9RZnBsIjPfBbyr2wFKUr/LzLuBF0+5Y9n3mJYf\n/6gjAUlSf5v2+Bc4fJzX7srMTV7PzGHg+WN3zMyd2xGwJHVb7UXmNroN+EI1k/nXwNGT7XznhrVd\nCarXY5CkWfgpcHFEnEJZa+5V7TjpoOXEQeuvpGbqpVzVS7FIGjhzHf+2dXZyXfnQPCxpMn1TZM7M\nfwf+fbr7X3D6kQ+uhVmn5ct3qTsESZqRai25/dt93jrz8pIlQ7W07TVAUq/rlTHzKPOmpDrMdfzb\n7tnJdeZm87CkifRNkXmmVq1a1diFuCWpH9WZl5v85QyS1EmOmSWp95ibJfWizabeRZIkSZIkSZKk\n8VlkliRJkiRJkiTNmkVmSZIkSZIkSdKszRsZaeuXnEqSJEmSJEmSBogzmSVJkiRJkiRJs2aRWZIk\nSZIkSZI0axaZJUmSJEmSJEmzZpFZkiRJkiRJkjRrFpklSZIkSZIkSbNmkVmSJEmSJEmSNGtb1B1A\nJ0XEZsAHgScCdwOvzMyftGw/GDgZuA/4aGb+ay2BjmOq2Kt9FgKXAy/PzOx+lA83jd/5EcBrKL/z\nHwAnZuZIHbGONY3Y/xx4IzACXJiZZ9US6BjTea9U+30E+FVmntTlEMc1jd/364BXAOuql47LzDVd\nD3Qc04j9qcD7gHnArcBLM/PuOmLtFbPJx9N9b3ei7er1bwMbqt1+mpmvaHfb1T6b5PJu9Xu8tqvX\nOt7v8a4FlL+Xbvz3Hvc61KV+P+w60sX3+bjXsG69z6v9HrwOdfN9Prbt6uc597vfNHncPB1NHVtP\nR5PH39PR1DH6dDV1LD8dTR7v94Im5+Wm5twm59Om5som58Cm5rhO1jP6fSbzYcD8zNwTeBPllwRA\nRGwJvB/4E2Af4NiI2KGWKMc3YewAEfEU4CpgV0qS6BWT/c4XAO8A9s3MZwGLgT+rJcrxTRb75sDp\nwLOBPYATI2JJLVE+3KTvFYCIOA54PA15r1T+GDgqM/er/tWejFtM9l6ZB3wEOCYz9wIuBXapJcre\nMpt8fBiw1WTv7Q61vTQitgZoef/NtgA1m1w+5d90p9ruRr8nuRZ047/3uG13qd/jXUe2pzv9Hvca\n1q33eRXD2OtQV97n47Xdxn73myaPm6ejqWPr6Wjy+Hs6mjpGn66mjuWno8nj/V7Q5Lzc1Jzb5Hza\n1FzZ5BzY1BzXsXpGvxeZn0n5hZCZXwee0rLtccCPM3NDZt4LXA3s3f0QJzRZ7ADzKW+MnvjEr8Vk\ncd8F7JGZd1U/bwH8rrvhTWrC2DPzfuCxmXkHsBTYHLinjiDHMel7JSL2BJ4GnE35JKpXTPUefzLw\n5ohYHRFv6nZwU5gs9lXAr4C/jYgrgCU9dDGp02zy8TOBz09wTCfb3gf4Q2BhRFwWEV+KiKd3oG0Y\nP5dPdUwn2+5Gv8e7FtxFd/57T3Qd6ni/J7mOdLzfk7Tdlff5BNehrrzPJ2i7Xf3uN00eN09HU8fW\n09Hk8fd0NHWMPl1NHctPR5PH+72gyXm5qTm3yfm0qbmyyTmwqTmuY/WMfi8ybwv8puXn+6tp4aPb\nNrRsu4PySVSvmCx2MvOrmXlz98Oa0oRxZ+ZIZq4DiIi/BrbJzC/WEONEpvqdPxARLwC+A3wFuLPL\n8U1kwrgjYifgbcCr6b2EPOnvG7gYOA7YH3hWRDyvm8FNYbLYHwnsCfwTcADw7IjYr8vx9aLZ5OOp\n3iOdbPu3wD9m5oHA8cD/3979g1hxRXEc/7kkWkQTLGMgjcipBYs0FkkpNqKVhIhCSgM2FnZ2KSyU\nFBJCIBHWUoKgjULSGFPYRCxyEkibyr+gjfFPcd/KY9/MvLvD3PvmDN9Ptcu+u+fcN7Pnnr3vvZn1\nArHbanmNebfFLj7vlrXg1rJ8C8a+rXrHe/M68nzZmIKxX6jCvDvWoeLz7og91LynJnLfnCNqb50j\ncv+dI2qPnitqL58jcr8/BpHrctSaG7meRq2VkWtg1BpXbD9j6g31M0m75r5fc/fXs6+fbvrZLkmP\nayWWoSv3MevM28zWzOyC0sc0jtZObomlz7m7X5P0iaQdkr6qmFuXrryPKRWJm0rXXzpuZhHylqRL\n7v5o9sr8DUn7q2bXrSv3h0rvKnB3/1/pFcK+79Cbkq3W4ydLxpSM/VjS35LWJcnd/1E6rh8PHHvI\nMUP9nirzblkLqsy7JXa1492wjlQ73g2xa8y7aR06kZNvodhDzntqIvfNOaL21jki9985ovbouaL2\n8jki9/tjELkuR625ketp1FoZuQZGrXHF9jOmvsl8R9IhSTKzzyTdn/vZX5L2mdluM9uu9NGSu/VT\nbNWV+5gty/t7pYJ2ZO5jJmPRmruZfWhmv5nZdk8X9n8u6dVq0lzQmre7f+fuB9z9c0nfSrrq7ldW\nk+aCruf7I0kPzOyD2TWBvpB0byVZNus6z/+VtNPM9s6+PyjpQd30Rmmr9fj3JWNKxr4r6ZRm16Yy\nsz1Kr/b+N3DsIccM9XtqzbtpLag176bYxefdsY4Un3dH7OLzbliH1t39564xhWNfUboByxDznprI\nfXOOqL11jsj9d46oPXquqL18jsj9/hhErstRa27kehq1VkaugVFrXLH9jG1v3oztutnDmR3IjTsm\nStJJpWui7HT3H8zssNJb79ck/ejul1eT6aJluc897leN5A6VUnfeSn9Q95Qu8L/hkrv/UjXJFhnn\ny9dK/5i+lPSnpNM+gjvJbuFcOSHJ3P1c/SwXZTzfX0r6Rulup7fd/fxqMl2UkfvGIrhN0h13P7Oa\nTMejTz1uGtOn1vWM/b6knyR9qnSDibPu/sfQsece966W15p3S+zi81bzWnBR0vXNY4aed0fsm6pw\nvJvWkdnjapznTbHfU93z/N06tILzfD72IOf51ETum3NE7a1zRO6/c0Tt0XNF7eVzRO73xyByXY5a\ncyPX06i1MnINjFrjSu5nTHqTGQAAAAAAAABQ1tQvlwEAAAAAAAAAKIhNZgAAAAAAAABAb2wyAwAA\nAAAAAAB6Y5MZAAAAAAAAANAbm8wAAAAAAAAAgN7YZAYAAAAAAAAA9MYmMwAAAAAAAACgNzaZAQAA\nAAAAAAC9vQWaUjgWZ+800AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1058dc290>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "words = np.array(vectorizer.get_feature_names())\n",
    "top_words = 10\n",
    "n_rows, n_cols = 2, 4\n",
    "top_components = n_rows * n_cols\n",
    "plt.figure(figsize=(20,7))\n",
    "for no, component in enumerate(svd.components_[:top_components]):\n",
    "    plt.subplot(n_rows,n_cols,no+1)\n",
    "    s = pd.Series(component, index=words).sort(inplace=False)[-top_words:]\n",
    "    s.plot(kind='barh', title=\"Component %d\\nexplained variance %4.1f\" % \n",
    "           (no, 100 * svd.explained_variance_ratio_[no]))\n",
    "f = plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Very clear results! \n",
    "\n",
    "Let's look at our documents and see how an article is represented in these dimensions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABH4AAAEKCAYAAABkPpJpAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10cea8f90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "f, axes = plt.subplots(1, true_k, figsize=(20, 4))\n",
    "for label in xrange(true_k):\n",
    "    X_red_class = X_red[y == label]\n",
    "    no = np.abs(X_red_class[:, :top_components]).max(axis=1).argmax()  # pick most outspoken sample\n",
    "    f = pd.Series(X_red_class[no,:top_components]).plot(\n",
    "        title=\"Most outspoken article in %s\" % dataset.target_names[label],\n",
    "        kind='bar', rot=0, ylim=(0, 1), ax=axes[label])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Indeed, the article has stronger representation in the component that resembles its category.\n",
    "\n",
    "This is just data magic!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}