{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loaded  44  metafeatures for  166  datasets\n",
      "loaded  166 datasets and  14 classifiers\n",
      "classifiers: ['RandomForestClassifier' 'SVC' 'KNeighborsClassifier'\n",
      " 'GradientBoostingClassifier' 'LogisticRegression' 'DecisionTreeClassifier']\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "# get data metafeatures\n",
    "metafeatures = pd.read_csv('../metafeatures/pmlb_metafeatures.csv',sep=',',index_col=0)\n",
    "print('loaded ', metafeatures.shape[1]-1, ' metafeatures for ', metafeatures.shape[0], ' datasets')\n",
    "# get ML results\n",
    "data = pd.read_csv('sklearn-benchmark5-data.tsv.gz', sep='\\t', names=['dataset',\n",
    "                                                                     'classifier',\n",
    "                                                                     'parameters',\n",
    "                                                                     'accuracy', \n",
    "                                                                     'macrof1',\n",
    "                                                                     'bal_accuracy']).fillna('')\n",
    "\n",
    "data['accuracy'] = data['accuracy'].apply(lambda x: round(x, 3))\n",
    "print('loaded ',data['dataset'].unique().shape[0],'datasets and ', data['classifier'].unique().shape[0],'classifiers')\n",
    "# subset data to classifiers used in PennAI\n",
    "pennai_classifiers = ['LogisticRegression', 'RandomForestClassifier', 'SVC', \n",
    "                      'KNeighborsClassifier', 'DecisionTreeClassifier', 'GradientBoostingClassifier']\n",
    "mask = np.array([c in pennai_classifiers for c in data['classifier'].values])\n",
    "data = data.loc[mask,:]\n",
    "print('classifiers:',data['classifier'].unique())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# can we predict which learner will be best on a dataset from dataset properties?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 166/166 [00:00<00:00, 1147.42it/s]\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/pandas/core/frame.py:2842: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  downcast=downcast, **kwargs)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "methods: ['RandomForestClassifier' 'SVC' 'KNeighborsClassifier'\n",
      " 'GradientBoostingClassifier' 'LogisticRegression' 'DecisionTreeClassifier']\n",
      "metafeatures[dataset].shape: (166,)\n",
      "9  features dropped due to missing values\n",
      "X shape: (166, 35)\n",
      "y shape: (166,)\n",
      "fitting model...\n",
      "confusion matrix:\n",
      "Confusion matrix, without normalization\n",
      "[[ 0  4  0  0  1  0]\n",
      " [ 2 66  0  1  5  8]\n",
      " [ 0  4  0  0  0  2]\n",
      " [ 0  2  0  2  0  4]\n",
      " [ 0 17  0  0 11  1]\n",
      " [ 0 18  1  1  1 15]]\n",
      "Normalized confusion matrix\n",
      "[[ 0.    0.8   0.    0.    0.2   0.  ]\n",
      " [ 0.02  0.8   0.    0.01  0.06  0.1 ]\n",
      " [ 0.    0.67  0.    0.    0.    0.33]\n",
      " [ 0.    0.25  0.    0.25  0.    0.5 ]\n",
      " [ 0.    0.59  0.    0.    0.38  0.03]\n",
      " [ 0.    0.5   0.03  0.03  0.03  0.42]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVkAAAEmCAYAAADIhuPPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzsnXeYFFXWh9/fzICgBF3BRFQQEANJMICCiwnFnGVVMK2u\nOa2uYs45RwwoZkwrGAADIqigBEHM8VOMuKCAIMHz/XFvQ0/T09NVXcM0w32fp57pvlV16lR3z+nb\n5957fjIzAoFAIFA1lFS3A4FAIFCTCUE2EAgEqpAQZAOBQKAKCUE2EAgEqpAQZAOBQKAKCUE2EAgE\nqpAQZGswkupKGibpN0lDC7DTT9LIJH2rLiRtJ+mTYrmepJaSTFLZivJpZUHS15J29I/PlXRvFVzj\nLknnJ2233DXCPNnqR9KhwOlAO2AOMAW43MzGFmj3MOAkYFszW1ywo0WOJAM2NrPPq9uXipD0NXC0\nmb3in7cEvgJqJf0eSRoMfGdmA5O0u6LIfK0SsNff2+uRhL18CT3ZakbS6cBNwBXAukBz4A5grwTM\ntwA+XRUCbD6E3mLVEV7bHJhZ2KppAxoCc4EDchyzGi4If++3m4DV/L5ewHfAGcDPwA/AAL/vYmAh\nsMhf4yjgIuDhNNstAQPK/PP+wJe43vRXQL+09rFp520LvAv85v9um7ZvNHApMM7bGQk0quDeUv7/\nO83/vYHdgE+B/wHnph3fDXgbmO2PvQ2o7feN8fcyz9/vQWn2zwZ+BIak2vw5rfw1OvvnGwC/AL3y\neO8eBM7wj5v4a5+QYbck43pDgL+A+d7Hf6e9B0cA/wfMBM7L8/0v9774NgNaA8f6936hv9awCu7D\ngOOAz/zrejvLfuGWAAOBb/z78xDQMOOzc5T3e0xa2wDgW2CWt90VmOrt35Z27VbAa8Cv/r4fAdZM\n2/81sKN/fBH+s+vf97lp22LgIr/vHOAL3GfvQ2Af374JsABY4s+Z7dsHA5elXfMY4HP//j0PbJDP\na5Xzs1LdgWZV3oBd/QekLMcxlwDvAOsAjYG3gEv9vl7+/EuAWrjg9AewVuYHs4LnqX+KMmAN4Heg\nrd+3PrCpf9wf/88M/M3/8xzmzzvEP1/b7x/tP+RtgLr++VUV3FvK/wu8/8fggtyjQH1gU1xA2tAf\n3wXY2l+3JfARcGrGP0HrLPavxgWruqQFPX/MMf6fcXVgBHBdnu/dkfjABRzq7/mJtH3/TfMh/Xpf\n4wNHxnswyPvXAfgT2CSP93/p+5LtNSAjgFRwHwYMB9bE/Yr6Bdg17T4+BzYC6gHPAEMy/H4I99mp\nm9Z2F1AH2BkX2J7z/jfBBeue3kZrYCf/3jTGBeqbsr1WZHx2047p6H3u5J8fgPuyLMF90c4D1s/x\nei19jYC/44J9Z+/TrcCYfF6rXFtIF1QvawMzLffP+X7AJWb2s5n9guuhHpa2f5Hfv8jMXsR9S7eN\n6c9fwGaS6prZD2Y2PcsxuwOfmdkQM1tsZo8BHwN7pB3zgJl9ambzgSdx/wgVsQiXf14EPA40Am42\nszn++h/iAg9mNtHM3vHX/Rq4G+iZxz1daGZ/en/KYWaDcIFkPO6L5bxK7KV4A+ghqQTYHrgG6O73\n9fT7o3Cxmc03s/eB9/H3TOXvfxJcZWazzez/gNdZ9n71A24wsy/NbC7wH+DgjNTARWY2L+O1vdTM\nFpjZSFyQe8z7PwN4E+gEYGafm9ko/978AtxA5e/nUiQ1xgXwk8xssrc51My+N7O/zOwJXK+zW54m\n+wH3m9kkM/vT3+82Pm+eoqLXqkJCkK1efgUaVZLP2gD3cy3FN75tqY2MIP0HrtcRCTObh/vmPw74\nQdILktrl4U/KpyZpz3+M4M+vZrbEP079o/6Utn9+6nxJbSQNl/SjpN9xeexGOWwD/GJmCyo5ZhCw\nGXCr/+eqFDP7AhdAOgLb4Xo430tqS7wgW9FrVtn7nwRRrl2GGztI8W0We5nvX0Xv57qSHpc0w7+f\nD1P5+4k/txbwFPComT2e1n64pCmSZkuajXtf87JJxv36L5Zfif/ZBkKQrW7exv003DvHMd/jBrBS\nNPdtcZiH+1mcYr30nWY2wsx2wvXoPsYFn8r8Sfk0I6ZPUbgT59fGZtYAOBdQJefknD4jqR4uz3kf\ncJGkv0Xw5w1gf1xeeIZ/fgSwFm6GSGR/spDr/S/3fkoq937GuFY+115M+aBZyDWu8Odv7t/Pf1D5\n+5niVlx6a+nMCUktcJ/ZE3HpqzWBD9JsVuZrufuVtAbu12ZBn+0QZKsRM/sNl4+8XdLeklaXVEtS\nH0nX+MMeAwZKaiypkT/+4ZiXnAJsL6m5pIa4n0PA0l7FXv6D9Scu7fBXFhsvAm0kHSqpTNJBQHtc\nT66qqY/7x5rre9nHZ+z/CZc/jMLNwHtmdjTwAi6fCICkiySNznHuG7h/6DH++Wj/fGxa7zyTqD7m\nev/fBzaV1FFSHVzespBrZbv2aZI29F9GV+DyzknNVqmP+5z9JqkJcFY+J0n6J+7XQj8zS/+MroEL\npL/44wbgerIpfgKaSqpdgenHgAH+9VwNd7/jfWoqNiHIVjNmdj1ujuxA3IfjW9w/6nP+kMuA93Cj\ns9OASb4tzrVGAU94WxMpHxhLvB/f40ZWe7J8EMPMfgX64mY0/IobIe9rZjPj+BSRM3GDTHNwPZYn\nMvZfBDzofyoeWJkxSXvhBh9T93k60FlSP/+8GW6WREW8gQsUqSA7FtezHFPhGXAlLmjOlnRmZT6S\n4/03s09xA2Ov4HKPmfOq7wPa+2s9R3Tux82IGIObbbIAN+86KS7GDTL9hvuCeybP8w7BfXl8L2mu\n3841sw+B63G/EH8CNqf8+/caMB34UdJyn1dz83HPB57GzV5pBRwc58bSCYsRAoEKkDQF6O2/WAKB\nWIQgGwgEAlVISBcEAoFAFRKCbCAQCFQhIcgGAoFAFRKKOgSqnUaNGlmLFi0LtrMkwfGFUuU7XXPF\nkdTdJXlnkyZNnGlmjeOeX9qghdni8gvxbP4vI8xs14KdKxJCkA1UOy1atGTc+PcKtjN3QXLFxurV\nKb5/jSV/JRNmS0uSC7N1aylz9V8kbPECVmtXfpbUgsm35rtCa6UgpAsCRc/IES+zxaZt2bRda669\n5qqCbC1ZsoQdum/JofsXVkkyKZ+SsnP8sUfSsum6dO20eWwbSfuUFwJKSstvNYwQZANFzZIlSzj1\n5BP477CXmDz1Q4Y+/hgfffhhbHv33HELbdpuUhQ+JXlv/Q7rz3PDXop1blX5lB+C0rLyWw0jBNlA\nUfPuhAm0atWaDTfaiNq1a3PAQQczfNh/Y9n6fsZ3jBrxEv844sii8CnJe+ux3fastVaUsgtV71Ne\nSKEnGwhUJ99/P4OmTZstfd6kSVNmzIhXr+O8s8/gwkuvpKSksI99Uj4leW9JUS0+hSBbXEha4kuZ\nTZf0vqQzfE3POLYuSQm1VbD/OEmHx7C7i/dxil9X/Yl//FAcP7PYbyBpkKQvJE2U9Lqkrr5gy+wk\nruGvc0JqHb+k9v71niyplaQ3k7rOimDkSy/QuHFjOnTqUt2uBNKRoLRW+a2GsTImQOabWUcASevg\nqug3AC6MasjMLqhk/1259uc4bwSuyj6+itOZZrbc8LmkspgVje7HqQK0NjOT1AqnRJAoZnZ72tN9\nccWXUyMh2+VrR5JwS7izVfXKyQYbNOG775aVLJ0x4zuaNGmS44zsjH/nLV5+cTivjHyZBQsWMHfO\n7xx/9OHceW/0772kfErKTpKseJ8EpdF7r5LWBO7FVdkynIrDJ7iiQS1xqgoHmtmspDyNy0rXk03H\nzH7GaRmdKEeppGslvStpqi+JBoCksyVN872xq3zbYEn7+8dXSfrQn3edb7soVSnJlz97x+9/VtJa\nvn20pKslTZD0qaScwUfS0ZKek/Q6ywLxOf78qZIuSDv2CN8+RdIdkkrkikJ3xFX7N/86fGFmL2Vc\np4Gk1yRN8nb7+vb6kl7yr8MHafd/bdr9X+3bLpN0qqQ9cZXBTpL0SmaPOZv/klp7e4/gKh+tH+nN\n9WzZtSuff/4ZX3/1FQsXLmToE4+ze989I9s5/+LLmfrJ10ya/jmDBj9Cj+13iBVgk/QpKTtJssJ9\nij+74GbgZTNrh1OR+Ain7/WqmW0MvOqfVzsrY0+2HGb2paRSnIbQXsBvZtbV14McJ2kkTmp7L2Ar\nM/tDGYWZJa0N7AO08z3DNbNc6iGczMUbki7B9ZxP9fvKzKybpN18e4UpCE8noKOZzfLnNAe2wn3k\nXpS0La5u6j54OW9J9+DKri0AJufRK5wP7G1mv/se/zhcacPdgK/NrI+/94aS1vXtm2a7fzN7XlI3\nnFTOTUpTcsjh/8+41/3wbL34fCkrK+PGm29jj913YcmSJRzR/0jab7ppXHOJkJRPSd5b/8MO5c0x\no/l15kzabNSM886/iCMGHFWtPuWHIqcI5Gohb4/T7MLMFgIL5UpX9vKHPYir73t2Qo7GZqUPshns\nDGyR6p3h1GA3xgW9B8zsDwAz+1/Geb/hgtd9koaTUYDav6lrmllKUuRBYGjaIak6mBNxP1UqY2Ta\nz5idgT7AZP+8Hu6n/5o4lc/33K9t6uJqzWbT3cqGgKsk9cAV324mV/R5qm+/CicEOE7SH/6YQZJe\nIFoB7or8/xn4oqIAK+lY3K8QmjVvnvMCu/bZjV377BbBpdx0364n3bfLW0oqK0n5lJSdwUMeLdhG\niqRf75ykZheUp5Gk9M/NPWZ2T9rzDXG1lx+Q1AH3f3cKsK6Z/eCP+ZHyMjnVxkofZCVthJP5/RkX\nWE7yOdH0Y3bJZcP3FLsBvXFyIifilCvzJaULtYT8XtN56e7h1DLvSz9A0mk4UbfzM9rbAh0llVTS\nmz0c9yXT2d/fd0AdM/tI0pa4nutVkl4ysyt82044tc/jccEzHyryv3XGfZbD/9PcA9Cly5ah3uYq\nS9YgO9PMtsxxUhmu2PdJZjZe0s1kpAb8L7Ki+Fyt1DlZObXKu3Ba7obLcR4vJ7KWEt5bAxiFk5VY\n3bdnpgvq4fTkXwROY5lSKLBUJmZWWr71MKIL5VXECOAo7yeSmvoe5yvAgf4xktaW1NzMPsFVyL9A\nvosrJw/SJ8NuQ+BnH2B3wovBycl8zDWzIbgq8p0l1QcamNlwf/+dEvA/EKiceLMLvsPJrI/3z5/C\nBd2fJK3vzGp9XMer2lkZe7J15SrW18KJug3BSQmDG21sCUzyAegXXF7yZUkdcT+9F+J0qs5Ns1kf\n+K+cTpJwMiSZHAHc5QP1l8CAJG7GzF6U06t6x8fMOcChZjZN0sXAK3JT1BbhlGT/z1/7BuBzSfP9\nfWZKmQwBhkmaBkzAyZOA+wK5StJfwEJvsyHwjM9jp2RoCvI/4ssQWJWJODfWzH6U9K2ktr7T0Rsn\nHf8h7v/0Kv+3CldR5E9QRghUO126bGmhQEzlFGmBmImV/LTPSclaLW21XgPLtS147phKbfpO071A\nbZZ1ekqAJ3EDsd/gpnBljr+scIrvkxQIBFYZ3Ayu6PNkzWwKkC0Q9y7Up6QJQTYQCFQfEkqwZ12M\nhCAbCASqlUJrSRQ7IcgGqp2/gIWLI6+4XY5m251a+UF5Muvd2xKxk1QeFZLLOf+xcEkidpJAEiWl\nIcgGAoFAlRF6soFAIFBViBqfk63ZXyGBlZ7vvv2Wvrv0plunzdiq8+bcedstkc5vWK8uj157FFOe\nGcjkpwey1RYbAnD8wT2Z8sxAJj51HpefEl2KphhlY7bcfGN6bdOJ3j22ZOeeW8e2c99dt7BLj87s\nul0XTj72cP5csKBg3ypCiJKSknJbTSP0ZANFTVlZGZdddS0dO3Vmzpw59Ny2Kzv03pF2m7TP6/zr\n/r0/I9/6kEPPuo9aZaWsXqc222+5MX17bU63g65i4aLFNF6rXiSfUhItL7w0iiZNm9Jj66707bsn\nm7TPz6d0+h3Wn38efyLHHHlE5HOz8fTwUay9dvwFdz/+MIMHB93ByLGTqVO3Lice1Y9hzw5l/0MO\nS8S/5RA1Pidbs+8usNKz3vrr07FTZwDq169P23bt+P77/Cr1N6hXhx6dWzH42bcBWLR4Cb/Nnc+x\nB2zHdQ+MYuEiN5D0y6y5kXwqRtmYJFmyeDELFsxn8eLFzJ8/n3XXi1WlMm9qek+25t1RoMbyzTdf\nM3XKFLbsulVex7fcYG1mzprLPRf/g7cfO5s7LjiU1evUpnWLdejeqRVjHjqTkfeeQpf2uauAZVKM\nsjHgfnofvPdu7Lz9Vgx54N5YNtZbvwlH/+tUenRsw9abbUj9Bg3YbofKKnfGR7h5sulbXudJX8vV\nh56Sqtgl6W+SRkn6zP9dq8ocj0CVBllJ60p6VNKXcjIpb0vapwB76UW0c0rHVGKno6+DmnreX9Iv\nWiZr81SqmEwSZLnenpJiFxSWVEuuyPhnckW5304ViPEfvkQKtKT7KamxpPFy8jPbSXoxs+5sVTJ3\n7lwOO+QArrz2Bho0aJDXOWVlpXRs14xBQ99km0Ou5o/5f3LmkTtRVlrC3xquwfaHX8e5Nz7Hw9cU\nJqxYLDw/4nVGvTmBR54exgP33snb46IrBP02exavvDycNyZ+xNvTvmT+H/N4buhjVeCtR1BaWlpu\ni8AOZtYxbQluURbtrrIg6wu0PAeMMbONzKwLruh004zjYuWFzewCM3slpnsdcaX+0nnCv2Gb4gqn\nHBTTdqXXM7Pn02Rc4nApTmlgMzPrDOyNK3KTKBl+9gammVknM3vTzHYzs7z1xOQKq8di0aJFHHbI\n/hx40KHsufe+eZ8346dZzPh5Nu9+8A0Az74yhY7tmjHjp9k89+oUAN6b/g1//WU0ipCXLUbZGID1\nN3A+NG68Dn367sXkie9GtjHujddo2rwlazdqTK1atdhl972Z+O47Sbtajjg92QrYC1frGf9374Kd\nS4Cq7Mn+HViYrpNlZt+Y2a2+5/i8pNeAVyXVk/Sq75VNk6twDoCk8+RkXcYCbdPa06Vjukh6w/eW\nR6SVO1tOGkZSbeAS4CDfcy0XTH3QXwOY5Z+3lJNxmep9bF5J+wFysi7vSxqT7Xr+/m9Lu49bJL3l\ne/ypeyqRk5z52P/0eVHS/r6HfQyuluaf/nX9ycyezHwD5GRuJvre+bG+rdRf8wP/Wp/m20/WMvmZ\nx31bf0m3+WIc1wB7+Xuom95jlvQPLZPJuTsVUOVEJK+X9D6wTdQPkL83TjzuaNq23YQTTzkt0rk/\n/TqH736cxcYt1gGgV7e2fPzljwwbPZWeXZ0kWuvm61C7VhkzI+Rli1E2Zt68ecydM2fp4zdee4V2\n7aMrGmzQtBlTJk5g/h9/YGa8NeZ1Wm/ctvITYyJlnV3QSNJ7aduxWU41YKT/fKf2r3JFuzcFJuXY\n3xnYwsz+5wPbPl4qpRGubN7z/piDcT3BMm9vYroRudqxtwJ7mdkvPmhejhNWgwxpGDPbUU6Haksz\nO9Hb6I8Lgj1wPcRPgWH+/FuBB83sQUlHArfgviErar8A2MXMZkha08wWVnC9dNYHeuDkWp7H1cfc\nF1e2sT1OWucjnIBia+D/zOz3HK9tiiP961sXeFfS095mEzPbzPuS+sl/DrChmf2ZmQYwsylZ7iH1\n+m+C6/V3N7NFku4A+uHketYAxpvZGZmOKV0ZoVnFOdF33hrH448+zKabbU6PrdwA2AUXX8bOu+ZX\nuf/0q4fywBX9qV1WytczZnLshQ8zb/5C7r6oH+8NPZeFi5Zw9AVD8rKVohhlY2b+/BMD/nEAAIsX\nL2bf/Q/m7zvmrFWflY5durHrHvuwR+9tKCsro/3mHTj48Oj+RCHL7ILKinYD9PD/Y+sAoyR9nL6z\nmIp2r7ApXJJuxwWShcDtwKi0MmQCrpC0PW6VZRPct9B2wLMp2RgfeDNpi1OsHOX/8UuBH9L25ysN\n84SZnShn5HbgLFxdym1wAQ9cjdZr/OOK2scBgyU9mXbtynjOqxx8KKe3Be61Gurbf5QTXozKyVqW\nA2+Gk+L5BNhI0q3AC8BIv38q8Iik53BpnnzpDXTBBXFwMjmpYslLgKeznZSujNAphzLCNt178Nv8\n+MtAp346gx79rlmu/ciBhamzF5tsTIsNN+K1cRMrPzAPTjv7fE47+/zKD0wACUpipAjMbIb/+7Ok\nZ4Fu+KLdZvaDiqhod1WmC6bjeqIAmNkJuH/Ixr4pXZqkn2/vYk7u+yegTp7XETDd51M7mtnmZpYu\nnRJJGsYrLAzDCbVFxsyOAwbigtpEOZHGyvgz7XFln7jPgeaSco7+SOqF0zbbxsw64DS46nhtsQ44\nkbnjcDU5AXbHfbl0xgXMfL+AhevRp17/tmZ2kd+3wMyKZ6F8oAgRJSXlt0rPkNaQU/NATpFjZ+AD\n3K/A1ITjoinaXZVB9jWgjqTj09oqGrFPSaUskrQD0MK3jwH29jnA+sAeWc79BGgsaRtYOvJe2W+3\nOeQeKOoBfOEfv4VLWYD7MngzV7ukVmY23swuwCkWNMvjetkYB+znc7Pr4lU4fa/+PuBmn+9Njfwf\nkHF+Q2CWOXXedsDW/thGQImZPY37Mugsp7zQzMxex6l7NsQJIubDq8D+/mdbahpNi0rOCQQA15Mt\nKyspt+XBusBYn+ufALxgZi/jfnnuJOkzXAejkMHlxKiwt1JZT6mynKDPiewN3Cjp37iAMw/3T1w3\n4/BHWCaV8h7wsbcxSdITwPu4rv9yw6U+57k/cIucqmwZcBO5VV1fB86Rk7G50relcrIlOA2h/r79\nJJwq5ln+HgZU0n6tpI1xPbxXve//l+V6lfE0y2Q1vsXlo3/z+wYCl+HSCwtwr+sFGee/DBwn6SPc\nF1FqiLiJ9zv1af4PLsXysH/9BNxiZrNTeddcmNmHkgbiBiFSMjkn4CrTBwKVEjVdYGZfkqHD59t/\npQiLdlcoPyPpW9wIXvorkHpuZhZtBncgMpLqmdlcn3KYgBtc+rG6/UqaTl22tDfGTSjYzrrbnJyA\nN45Q6jA/NmpctyD5mdU3aGOtj76jXNu0S3cqyGaxUWFP1syaVbQvsMIY7kf6awOX1sQAG1jVEaU1\nvHZBXoMbkg4GNjKzKyQ1xc1HS2YoM1AhZtarun0IBKqSuLMLViYq/QqRmzS/A5Aqw/MHcFfFZwQC\ngUD+RJ1dsLKRT092WzPrLGkygJ/cXruK/QqsQpQAtfMbVc7Jt2/eVLgzCZOk/HbD1WslYqdu7dgr\nnBNHgtLSmhdY08knyC7yo8YG4AdhChdkCgQCAWpm7zWdfLoPt+OmEzWWdDEwFri6Sr0KBNJISoUA\nXMHtHbpvyaH7R1dDqAqfkry3JGwVqkQRGdX8dEGlQdbMHsLNy7wO+B9wgJk9XtWOBQKwTIXgv8Ne\nYvLUDxn6+GN89OGHse3dc8cttGm7SVH4lOS9JWUrpUQxYfIHvPLGWwy6+w4+/ij+610ZAkpLS8pt\nNY1876gUN8l8YYRzAoGCSVKF4PsZ3zFqxEv844jC6scm5VOS95aUrUKUKGJRQE9WrqLcZEnD/fMN\n5Woefy7piWIZO8pndsF5wGPABrhasI9K+k9VOxYIQLIqBOedfQYXXnplwRInSfmU5L1VhVpDVCWK\nOChG7YI0TsFVp0txNXCjmbXGlSqt2vJheZLPp+1woKuZDTSz83DVbvpXqVcFIGlu2uPd5OrItpBT\nVfgjtcY+89gc9ipVAJCrW7vcChWl1Y1NGklnytWanSLpXUmH5/Il5jW2lHSLf7yapFe0rCbuvZKi\nKwdWEyNfeoHGjRvToVOX6nZlpSCOEkUcJCgrLSm35XeemuKKGt3rnwtXw/opf0jRFO3OZ3bBDxnH\nlVG+lGBRIqk3rsbrLmb2jV+HPxM4A1c/IS/MrPB6djHwHxr5UoeZ+44DdgK6+Rq8DYDYsj4VYWbv\n4WpJAHTybR398yei2JJUGqciV1IqBOPfeYuXXxzOKyNfZsGCBcyd8zvHH304d94bveRhUj4lqbCQ\npK24ShRxyTLNrZG8bpfnHl8aM52bgH+zrPDS2sBsM0utPf4OV6ej2qnwa0PSjZJuwA12Tfc9l0HA\nNFywKlrk6tIOAvqa2Rdpu+7HFYJZTh5UFVf3T1cAOF/SJ5LGSnpMXm/Mc4DSFBjS2pv53uVnki5M\nu97pcuoEH0g61be19PYfwpVua6YsKgbAucDxqSI9Zva7mT1IBpLulKssP93PDEm1X6VlKgjX+bZy\nig6+rZek4b73/zDQ1b8+rdJ7zJJ2ltMZmyRpqKR6aa/d1ZImAZlVwvIiKRWC8y++nKmffM2k6Z8z\naPAj9Nh+h1gBNkmfklRYSMpWIUoUcRBQIpXb8EW707ZyAVZSX1zVvpVi1WmunuwH/u90XHHnFFUr\n+FM4q+GKTvcys48z9s3FBdpTgPSAl6u6f+qYrsB+uOo/tVhepaGcAgOu1Bq49MpmuJVy70p6ATfn\neACwFe5zNl7SG7g80sbAEWb2jqQuZKgY+F5rfV+JqDLO84tHSnEyP1sAM3C93na+UloqFVJO0SHd\niC+MfDRwppn19b6kXpdGuNknO5rZPElnA6fjJHcAfjWnQ1YOpSsjNK+41lCSKgRJkZRPSd5bUrYK\nVaKIjERZ9MUI3YE9/f9aHaABcDOwpqQy35ttivusVzu5CsTctyIdSZBFuFqvR+GCaSa3AFNSPThP\nrur+KboD/zWzBcACScMy9lekwDDKl2BD0jO4WrWGU3yYl9a+Ha7o8Ddmlvoi+5LlVQzyV/yDA30w\nK8NJ3LTHlU5cANwnNyo73B8bR9EBXJ3a9sA4/9rVBt5O2581rZCujNAlhzICJKdCkKL7dj3pvl3P\ngmwk5VOS95aErUKVKKIioq+KM7P/4Ep0porTn2lm/SQNBfYHHmdlKtrtfxo+7n9afpraVoRzMfkL\nOBDoJunczJ3mFFYfxdU8TZGrun++VKTAkBlAKqt9t1QxIpuKgU8RzJW0US4jkjYEzgR6m9kWuCBd\nx3/Ld8MNEPTF1Z2Nq+gA7rUblfbatTez9FHdeRWdGAgo2cUIZwOnS/ocl6Mtio5iPkN5g4EHcP9M\nfYAniTiU7RvHAAAgAElEQVTosaLx6gG7A/0kZZvGcQPwT5YFw3yq+48D9pBUx+cc++bpzk7eXl3c\naOc4nIrC3pJWl5PP2IdligtLURYVA7/rSuB2nzpATu338IzTG+AC3G9yygp9UscCDc3sReA0fPFj\nZVd0yId3gO6SWns7a0hqk+e5gQBlJSXltiiY2ehUCsvMvjSzbmbW2swOMK/mXN3kM7tgdTMbIek6\nP4g00I/8rRiltZj4XOSuwBhJv2Tsmyknvnaaf15pdX8ze1dOyHEqToNsGsuUCnIxAbcsuSnwsB+x\nR9Jgvw9cD3WypJYZ52ZTMQC4E5c2eFfSIu/v9Rn3+L5cUZ+PccoK4/yu+sB/JdXBfXGe7tuzKTpU\n+pvanEJwf+AxSav55oE4xd9AICeSEi2iU4xUqIyw9ADpLVwe8RncT8sZwHVmVnVi7EWKlikVrI7T\nHzvWzHLJngfyoEuXLW3c+PcqP7ASklIOAKhXZ4UJOa9wFi5Orr5Tw7qlBakYrL3Rprb7ZY+VaxvS\nr8OqoYyQxmnAGsDJwOU4kb3C1iWuvNwjNwG/Di6HGwJsIFAAAspqeE82H4ns8f7hHJYV7l4lMbND\nq9uHQKAmISVbc7cYyaVW+yw5RsLNrOqXggQCgRpPTSxvmE6unmyVrLkPBKqKJNQVkibJ/GdS95ek\nT4WyKgx85VqM8OqKdCQQCKx6uJxs8X05JknNvrtAjSAp9YAkq/4XowpBMSpI5ENZSfmtMvxc9Qm+\nzsbSuhxaWevJBgLVSZLqAUlV/S9GFYJiVJDIh9TAV/qWB38CfzezDkBHYFdJW7MS15MFXD3RqnQk\nEMhGkuoBSVX9L0YVgmJUkMgHAbVKVG6rDHOkakHX8ptRpPVk86ld0E3SNOAz/7yDL1gSCFQ5VVHx\nHwqr+l+MKgTFqCCRD6mBr4yebCO5Ep2p7dgs55VKmoIr5DQK+IKVrZ5sGrfg1un/Cm65JrBDVTq1\nIlAeqgh52NhA0lM59q8p6V/5Hu+PGS1XU/Z9OcWDjrmOX9FIukTSjpUfWbysqKr/K6M/1aEgUVpS\nfqOSerIAZrbEF5Bviit41G6FORyRfIJsiZl9k9G24mqhFTFm9r2Z7Z/jkDWBf0U4PkU/n2+6A7i2\nQDcBkJTIOlEzu8DMXknCVj4kWfEfkqn6X4wqBEkrSHTetDXH9O/H2DGvc/zRmbWHkkOCWqUqt0XB\nV9V7HdgGX0/W7yqaerL5BNlvJXUDzHfRT6WGFv+QUyZ4zZd1fFVSc9/eStI7cuoEl6V6wf74D/zj\nTbVMWWGqL7ZyFdDKt12bcXyppOvk1AimSjopi0tvk/aTRxUrEOwmp/c1UdItWqbeeZGkIZLGAUP8\nNa/1PeSpkv7pj1tf0hjv5weStvPHLqfK4Nv29497y6mFTpN0fypvL6eIcLH3c5qk2L2MJNUDkqr6\nX4wqBMWoIJEPAkqlclul50iN5QvLy1W32wknqPg6rp4srEz1ZIHjcZWamuOqT23t22oit+JqEmwB\nPIJLlYCrun6zmW2Oy/Vk4zh/TEdgS3/cOcAXvs7qWRnHH4sr7t0x7XqZ7IpTechUIOiM09463VfT\nuhvoY2ZdgMYZNtr7cw7Bjbb+ZmZdga7AMXJ1Zw8FRnjfOwBTcKO2TcxsM3/fD6Qb9dcdDBzk95dR\n/nMx0/t5J66ubSzSK/533HwT9jvgwNjqAamq/2PeeJ0eW3Wmx1adGfnyi9XmU1L+JOlTdRBjdsH6\nwOuSpgLv4uoZD6dI68lWWoWrpiJprpnVy2ibCazvJWhqAT+YWSNJvwLrmtliuRqu35tZPbnShMPN\nbDNJhwLn4SRrnjGzz9L3e/vpxz8N3GVmozJ8GI37ENXGlTPs6CVh+uKCWirIpxQIbsUF957+/D1x\n1cH6SroINxibmkf4FLAFTgoHXLGff+KUEu7H6Xg9Z2ZTJK2FC+Qv4lUZzOwvuRKNw3EDobea2fbe\ndm/gBDPbV9LXOCmfGZK2Ai43s3J5XJWXn+ny6ReZGano1PTVVUn5lGS1ssb1axVUMatZu83tjEHP\nl2s7bfuNVq0qXHLiictFYjNbbsRvVcbMHpU0Hlcs/EX/UzwfHa5s9MPJ2FyLC6L7skyB4JD0A/MY\nGEtXJhBwkpmNyDxITnxyd5wEzQ1m9pCkDsAuuF76gUSrvlaRUgQQTX4mUHNx8jPV7UXVks/tvYIr\n4vwqrvDzOiz7B6ppvAUc7B/3Y5lawTs4EUXS9pdDTg7mSzO7BZcL2gJXuax+tuNx007+mUrUK0NB\n19xPjPOBrX1OsyIFgk9wOmAt/akH5bi/EcDxvpeOpDbeTgvgJzMbhNOx76yKVRlSfAK0TPmDq9D2\nRo5rBwLLIRQ5J7uykU+pw3JSM5KGAGOrzKMVx+qS0vOrNwAn4ZQIzsJJsAzw+04FHpZ0Hq5weTZF\nhAOBw+SUCn4ErvDqDOP8YNdLwO1px98LtAGm+nMGkVGUx8zmS7oeOMvMjlIWBQIz+1RumtjLkubh\nclQVcS8uDzxJkvw97g30As7yfswFDqdiVYaUbwskDQCG+i+Kd4G7clw7EFiO1OyCmkzknKykVrj8\nXKuqcan4kFNCmG9mJulg4BAzq/pF3XmiZYoNwgXyz8zsxur2K1+SUkYoxvxnMfpUTDnZlptsYRc8\nNLxc21HdWqxyOdlZLMvJlgD/w42ar0p0AW7zQWw2xacMcYykI3CDYZNxsw0CgaJHokamCNLJGWR9\nUOnAskm9f9kqOB3BzN7Eq7oWI77XutL0XAOBFALKaniQzfn7wwfUF/0StiWrYoANBAJViSgpKb/V\nNPJJ8kyR1KnKPQkEAqsccVZ8rWzk0vgq8xVtOgHvSvoCN+dSuE5u5pSeQKBamfZttkkf8eiy4VqJ\n2ElSEuern+dVflAeNG+0eiJ2kkCKni6Q1Ay36Gdd3HjRPWZ2s58G+QRuBs3XwIFmNitRh2OQKyc7\nATc3Mt5C8UAgEMiDkui918XAGWY2SVJ9YKKkUUB/4FUzu0rSObgB+rMTdTYGub5mBWBmX2TbVpB/\ngUBBsipX/OdE+m7dhsN233Zp2wWnHEn/Pben/57bs/8OHei/5/Yr1Kek7Jx3+vH02KIle/6969K2\nl4c9wx47bMmmTevzwfuTIvtz/LFH0rLpunTttHnkc+MQJ11gZj+Y2ST/eA6uOEwTYC9csW5YSYp2\nN5Z0ekXbCvMwsEpTqKzKbvseyvX3DS3XdsnN9zP4+TEMfn4MPXfeg547912hPiVlZ58D+3HPI8+V\na9u4XXtuGfQoW27dPbI/AP0O689zw16KdW484hXtXnq2W+nYCRiPqy/yg9/1Iy6dUO3kCrKluAIl\n9SvYAoEqp1BZlY5dt6VBw+z5VTPj9ZeeY8e++2XdX1U+JWVny6170HDN8vfWauN2bNi6TWRfUvTY\nbnvWWutvlR+YEKl5shk92UqLdrtzVQ94GjjVzH5P3+dnQhXFbKhcOdkfzOySFeZJIJCFbLIqEyaM\nT8T2+++9zVqN1qFZy2iLF5PyqSrvbWUiRk4WX3/jaeARM3vGN/8kaX0z+0HS+jhpmmqn0pxsdSBp\nSVoB6WGpAr0J2F1aNDsBW4MlfeX9nCLp5CTsVnCtXpK2zWg7XMsKak+WdGaaX/moL+Rz3XJyOZIe\nkyv2fZpqgAzNK8OfZsfd46sRBApHQFmJym2VnuMWSd0HfGRmN6Tteh5XrBuKqGh3rp5s7xXmxfLM\n9wWkkfQgcAJweTX6UxFnmVlOza5sSCo1sygSPr1whVve8uf3wRWt2dnMvvcFYxLXCDGz7/GV5iWt\nB3Q1J7ccmbQpgZFIWn4mxeLFi3lj5HDue/a1yOcm5VNV3dvKRGrgKyLdcVXfpsmJKQKci1MieVLS\nUcA3uKJN1U6FPVkz+9+KdCQHSyVYJNWTk4VJyZrs5dtbSvpI0iBJ0yWNlJOlQFIXOVHC93HBGt9e\nR9IDaT3BHXx7f0nPSRolJ6Nyoh/smywnQZMzYSXpEG/zA0lXp7XPlXS992Mb79cbcpIxI/zPGySd\nLOlD32N83Cf2jwNO8z3m7XAVsc70QRAz+9OXKcz05QI5qZkPJN3jewDLXcO39UzrlU+WVD+j5z8S\naJLyQeVlaCq6l9GSbpL0HnBK/m/5MpKUn0nnvbdG02KjjVlnvehBLSmfqureViokSjK2yjCzsWYm\nM9vCq450NLMXzexXM+ttZhub2Y7FEsOKulyupFJcjzpVOn0BsI9fCLEDcH0qcAAbA7eb2aa4Ii6p\n0YwHcIWqM2sPnIDLj28OHAI8KCepArAZrlB2V1wP+g8z64QL+Ok9xmvTAtPmkjYArsbpv3cEukpK\nTSNZAxjv/RiPK8a9v5eMuZ9lPfVzgE7mJGmOM7OvcSUEb/Qfpje9fxPzeAlvM7OuXpmhLk51eLlr\n+LYzccoGHYHtgPkZtvZkmZROqs5uKjdW0b0A1PaDF9enG5N0rPzo8S8zf6nwBgqVVbnwtKM57qBd\n+L+vPmef7TZl+NAhALz6wrORB7yS8ikpO2f+qz+H7Pl3vv7iM3bo0oanH3uQV156nh26tGHKxAkc\nf/h+HHNotGJx/Q87lL/33JbPPv2ENhs148EHqlbBJVW7IH2raSSiYFoF1PU/A5rg5sClJFoEXCFX\nxf8vvz81TeMrM0v9dJiIKyi9JrCmmY3x7UOAPv5xD1xwwMw+lvQNrr4rwOt+/t0cSb8Bw3z7NFwx\n7hTl0gW+Zz3azH7xzx8BtsfpdC3BJeoB2uIC5Sj/HVEKpKaeTAUekfScP68QdpD0b2B14G/AdH8v\n2a4xDrjB+/yMmX2n/D7wue4F3Aqc5YiijLBrn93Ytc9u+fiyHBffeG/W9vOuvj1re74U4lNSdq67\nY3DW9h37xO8NDx7yaOxz45Ln52ylpVh7sqmcbAtcYE39zO+HEwrs4vf/BKR6n+lqDVklTyKQbuuv\ntOd/FWB3QVoeVsD0tJ86m5vZzn7f7riasJ1xy5mzXW86rvxihfhe+R24HubmuKLgqddquWuY2VXA\n0bge7zjlrzCb616gvPxNILAcJSq/1TSKNcgCYGZ/ACcDZ/hg0xD42ZzQ4Q64IJzr/NnAbEk9fFO/\ntN1vpp7Lybg0x0mqFMIEoKekRj7VcQjZJVk+wS322MZfv5acpHgJ0MzMXsctB2yIm6ucKWNzJS5V\nsZ4/v7akozOukQqoM+XmE6byp1mvIamVmU0zs6txKgf5Btms95LnuYFVHOF6sulbTaNY0wVLMbPJ\nctK/h+Bks4dJmoZTUv04DxMDgPslGW7wJsUdwJ3e1mKgv5n9Wcib7OfnnYPTfxfwgpktN43EzBb6\nQaNbJDXEvQ83AZ/iZG4a+vNvMbPZkoYBT/l0xElm9qKkdYFXfE7acLnQ9GvMlhPB/AC3+iUlS1Na\nwTUu9V9cf+F6yi/hVHMru+eK7mV6/q9cYJVFUFLUXb3CWWUlwQPFQ1LyMxO/Sq7gUlJVuJKkGKtw\n1VutpCCpmPZbdLZHh5f/sdepRYNVS34mEAgEqgqXLqhuL6qWEGQDgUC1UhMLdadTw7MhgUCgqFH0\ngS9J90v6OW2hDJL+JreA6DP/t2jyPaEnG6gxLPorOfntJX8V31hFUrnUD7/7vfKDVhAi1rStwcBt\nOHWEFOdQhAW7IfRkA4FANRNVSNEvLspcMluUBbshBNnASkAh6gFXn3sy+2zbjgF79Fja9vlH0/jX\nQbtw9N69+Od+vfloajQFgSTVA5KyVYidS88+gV26tubgXbdZ2nbPzVey+7ab0K9vD/r17cG410fm\nsFAYUvmNCEW70yjKgt0QgmygyClUPWDXfQ7m6kHlV/befe3FHHHCWdz73GgGnHwOd197USSfklQP\nSMpWIXZ23+9Qbn5g+WJyhwz4F48MH8sjw8fSfYeds5xZOBLZCsTkVbS7IoqpYDeEIBsocgpVD+iQ\nTRlBYt7cOQDMm/M7a6+zXiSfklQPSMpWIXY6d+tOgzWra5woehWuCvgprfpb0RTshhBkA0VONvWA\nGTNmFGTzxHMv5+5rL+LAXltw1zUXcszp5xfqZo1k6JB7OHS3bbn07BP4/bfZVXKN1MBXArULirJg\nN4QgG6gESefJ1eid6ks6XijpyoxjOkr6yD+uJ+luSV/4+rKjJW1VPd5n57+PPcC/zrmMJ0dP5V//\nuYxrB8YqdVuj2a/fUTzz+hQeHj6WtRuvx81XnFdl14oxhesxXNnRtpK+kyvSfRWwk6TPgB3986Ig\nBNlAhfiiL32Bzr727I64ugwHZRx6MPCYf3wvbuR3Y19fdgDQKK4PVaEeMPK5x9neK9T22nUvPo44\n8LUqsHajdSgtLaWkpIS9Dz6c6THkxfNCUFpSfqsMMzvEzNY3s1pm1tTM7ivWgt0QgmwgN+vjBiH+\nBDCzmX76zKyM3umBwGOSWgFbAQPN7C9/zldm9kJcB6pCPWDtddbj/QnjAJj0zps0abFRQfZqIjN/\n/nHp49Ejh9OqzSZVcp1QhSuwqjMSuEDSp8ArwBNm9gau13owMF7S1sD/zOwzSXsCU/LRL/PTco4F\naNa8eYXHpasHLFmyhCP6HxlJPeDS049hyrvj+G3W/zig5+b0P+lszrz0Rm69/FyWLFlC7dVW44xL\nbqjcUBr9DzuUN8eM5teZM2mzUTPOO/8ijhhwVCQbSdsqxM7AU45i4vixzJ71K327t+eYU85h0vix\nfPrhB0iwftPm/OeymyL7lC81sYZsOqEKVyAnvi7udji5n3/iV9bgRB1bADcA35rZ9T7IDjCzfaJc\nI6kqXO988WvBNlJ03TCZ2QPFSJIrvrq1WrOgilkdOnWxkW+8U65tvYa1QxWuwKqD75WOBkb72rtH\nmNlgSV8BPXFaaqlZ7NOBDoquxhtYRUnNk63JhJxsoEIktZW0cVpTR5zUMriUwY3Al2b2HYCZfYEr\npn6xLyaeUhLefQW6HVjJyLLiq0YRgmwgF/VwKr4fenWK9sBFft9QYFOWzSpIcTRuSePnvkrSYIpo\nYniguBCitKT8VtMI6YJAhZjZRGDbCvbNBGplaf8dOKaKXQvUIGpi7zWd0JMNBALVR/baBZWfJu0q\n6RNJn/vShkVLCLKBQKDacMtqowVZP+PldqAPLoV1iKT2VetpfEKQDQQC1UqM2gXdgM/N7EszWwg8\njqsnW5SEIBsIBKqVGCu+mgDfpj3/zrcVJWHgK1DtTJo0cWbdWvqm8iNpBMxM4JLFZidJWyvaTotC\nLjJ50sQRq9dWZm2LOpLSV6fcE7WmbDERgmyg2jGzxvkcJ+m9JFYCFZudYvQpyXvLhZntGuO0GUCz\ntOdNfVtREtIFgUBgZeNdYGNJG0qqjauj8Xw1+1QhoScbCARWKsxssaQTgRFAKXC/mU2vZrcqJATZ\nwMpEUnm5YrOTpK1is1MlmNmLwIvV7Uc+hCpcgUAgUIWEnGwgEAhUISHIBgKBQBUSgmwgEAhUISHI\nBooWOdZPwE6JpKzVxGLYObBQO0kiqVTSaQXa6CqpT5b23SR1KcR2IAx8BYocSR+Y2WYJ2JlsZp0S\nsJPkAoQ2wFm4VVNLZ/qY2d8j2plgZt0K8OM1nGzQNxntLYAHovoTKE+YwhUodqZI6mRmkwu086qk\n/YBnrLCexSuSzgSeAOalGmNKUA8F7gIGAYXI9YyTdFsWn/LV8a6fGWD9+d9Iyy15DUQk9GQDRY2k\n6UBb4AtcABFgZtY5op05wBq4YDY/zU6DiHa+ytJsZhZZV1zSRDMr+Oe4pNcr8CmvHqikz82sddR9\ngfwIQTZQ1Ehqla3d64mt1Ei6CCfN8yzwZ6o9Zq+4ED/uAn4FBqZ6+V6j7WJgPTM7dkX6U9MIQTZQ\n9EjaGmhjZg9JWhtYw8z+L6INAf2ADc3sUknNgPXNbEJEO6sDpwPNzexYLzTZ1syGR7HjbSXSK5a0\nLnAFsIGZ9fEFrLcxs/vyPH8N4D6gKzDFN3fAiWIebWZzo/gTKE8IsoGiRtJAoDvQyszaSGoCPGFm\nPSLauRP4C/i7mW0iaS1gpJl1jWjnCWAicLiZbeaD7ltm1jGKnSSR9BLwAHCemXWQVAZMNrPNI9rZ\nCCeOCTDdzL5M2NVVkjCFK1Ds7A/shh/QMbMZQKQ8qmcrMzsBWODtzAJqx7DTysyuARZ5O3/g8ruR\nkVRL0smSnvLbiZKWE6fMg0Zm9iTuSwQzW0yEgTSvRnwertM1zG8hwCZECLKBYudPnydM5QpXj2ln\nkdeGStlpjA9KEVkoqW6anVak5VMjcifQBbjDb118W1Tm+TRKyqetgd8inH8ITv59pKQJkk6TtEEM\nPwJZCFO4AsXOM5JuBxpKGgAcBdwfw84tuAGmdSRdjushD4xh50LgZaCZpEdwqYz+MewAdDWzDmnP\nX5P0fgw7p+PqqbaSNA5ojLu/vDCz94H3gf/4AH0Q8I6kL4BHzWxQDJ8CnpCTDRQ9fjXSzrif5SPM\n7KWYdtoBvb2dV83so5h21ga29nbeMbNYci+SJgEHpGZK+JzoU1Gnp/lzy3BT3QR8YmaL4viUZq8X\ncCPQ3sxWK8TWqk4IsoEajaQGZva7pL9l25/vdClJ7czsY0lZA2CEif/pNnvjBqy+xAXHFriVV9nm\nvWY7/+9m9pqkfSvw6ZmI/nTFpQ72A77CqcAONbNfo9gJlCekCwJFiaQ3zKynpFn4XGNqF26aU9ag\nmYVHgb64GQHL2QHynS51OnAscH2WfQZEXnpqZq+mpoD5pk/MLEp+d3vgNWCPCnzKK8hKugI4EJiF\nC6zdzey7CH4EchCCbKBYGeD/Frqs8yr/dxMzW1CAnVH+71GFjrzn6IG2lhSlBzrL/73PzMYW4NIC\nXA/6Te/f4X4J8jfARSt6cURNI8wuCBQrQ/3fl8xsSeYWwc7N/u9bBfrzH//3qQLtAPT0f/fIsvWN\nYCf1RXRLgf7sDUwHkLQ97ovpIdwMhaKWoVkZCDnZQFEiaQrup/5JwLWZ+80sr8Ai6R1gKi6QPJ7F\nzsl52hmF+wneFXgzi50987GTJJIeA7YENsDVdli6y7lkW+RpZ0pqMYWfyfGLmV2UuS8Qj5AuCBQr\nhwD74j6jjQuw0xfYEdgFl5eNy+5AZ2AI2fOykZF0Cm7gaw6uEldn4BwzG5nP+WZ2iKT1cKqthQT5\nMkllfhFDb1zueem+AuwGCD3ZQJEjaQ8zG5aAnQ5+Pmihdhqb2S+F2vG23vfLYHcBjsPN2x0SZwpX\ngX6ch1tVNxNoDnQ2M5PUGnjQzLqvSH9qGuFbKlCUSDrEzB4DNpK03E/6COmCf/tlsEdLWq5HESFd\ncJOZnQrcX4GdOD3J1HLc3YCHzGy6L2ST38nSk2Z2oKRpZJ+BkVe6wMwul/QqsD6unkPKVgkuXRMo\ngBBkA8XKWv5vobMLUgsO3ivQzhD/97oC7aQzUdJIYEPcaqv6RFvqe4r/G2WwLCtm9k6Wtk8LtRsI\n6YLAKoikEqCemf1eoJ21gGZmNrUAPzoCX5rZbL9gomlUe75U4Xwz+8tL2rTDzcooaNVXIBnCFK5A\nUSPpSkkNJJVJGiHpJ0mHxrDzqLezBvAB8KGks2LYGe3t/A2YBAySdENUO55tcAsQZkv6By4nG6Ww\nS4oxQB1fBnIkcBgwOKZPgYQJQTZQ7PTxPc6+wA/AJsDZMey093b2Bl7C/UQ/LIadht7Ovrg86la4\n2QtxuBP4Q1IH4AzcNKyHYtiRL7m4L3CHmR3AsrqwgWomBNlAsZMaN9gNeNKvPoqT46rla7XuDTzv\nf0rHsVMmJ1N+IBBZDSGDxX6QaS/gNjO7Hagfw44kbYNTfnjBt5UW6FsgIUKQDRQ7L0n6ANgKGCWn\nnhqnfuvdwNc4McUxcnLXcXKyl+DmpX5uZu/6ylmfxbADMEfSf3A96hd8jjZO0e5TcSvSnvUzFDYC\n8ioyE6h6wsBXoOiRtA7wPzNb7HOqa3qFhELtpibgVwt+IcGhwLtm9qak5kAvM4uTMkjZTGRQL5Ac\noScbKGp8EZX5PsCeg1shFXkFmKRT/ICVJN3na7lGrpwl6Rpvp5akVyX94getImNmPwJPA6l6rTNx\nhcWj+pTIoF6gaghBNlDsXGRmcyRti8vLPgLcFcPOkb53tzNuDu5hLKvQFYWd0wbivgZaA7ECmqRj\ncAVn7vZNTYDnYphKalAvUAWEIBsodlIVt/oCd5vZf1nW84tC+uqqIWY2Pa0tCqmBuN1xBa3jTLlK\ncQJOvuZ3ADP7DFgnhp2kBvUCVUAIsoFi5wdfGeog4EVJtYn3uU2trtoNGBFjdVWK4ZI+xokevion\nyBi3Tu2fZrYw9cRLyMQJjkkN6gWqgDDwFShqJNXDBcapXv5lA6BDVJ2vLKur1gaaxFmt5Rci/GZm\nS+TUcxv4/GpUO9cAs4HDcTUC/gV8aGbnRbWVxXa1DuoFlhGCbGClwAe2OqnnZvZ9DBtrARtn2BkT\nw85mQPsMO5FnBPjAfxRpIpHAvRbjn1LS7rgFCOk+XRLVTiB5QoGYQFHjg8eNQFPgV1yB6s9w6/Oj\n2DkaV1ClKTAFpzb7NhFnGEi6EOiFC7IvAn2AsURcqSWpFLdirB+ulmxsJN0FrA7sANyLkwOfUIjN\nQHKEnGyg2LkcNzj0iZk1A3YlizJBHpyCUzX4xsx2ADrhfqpHZX9cYesfzWwA0AFoGNWIl9Bp4XPM\nhbKtmR0OzDKzi3E1EdokYDeQAKEnGyh2FpvZL5JKJMnMRkmKU25wgZktkISk1Xx+t23lpy1HqtrV\nYkkNgJ+BZjHsgJMCHyfpeWBeqtHMohacme///uFz1r/iasMGioAQZAPFzm9+8Gss8JCkn1kWVKLw\nnaQ1cfNQR8lJjX8Tw8573s4gnJzNXFzaIQ5f+K2EeDULUgz3Pl2LqwxmuLRBoAgIA1+BosZPtZqP\nGxg6HPfTfEghEjCSeno7L6dPoYphpyVuZkGserJVgaTVgDoFzt8NJEgIsoEajZ+VUCG+qlc+dnLq\nbgMvOXQAABdPSURBVJnZpCh+eZvDWH5e7G84FYe7zSzn/Fu/5DiXT89E9SmQPCHIBooS/3M+24cz\npV+VM3im2fnK20lf3ZV6bma2UZ52clW1MjOLUwfhZlwdhsd800G4RQSG6yHnXBor6YFKfDoyqk+B\n5AlBNlCU+ClOFeJH51dqJL1rZl2ztUmabmah8HYNIEzhChQrHYEdzWxJ+oZTIchLhRVA0i6S9s/S\nvp+knSLY+Yek5XqWkg6LI4fjqefLG6ZsNQfq+aeV5oolnS7pqCztR0k6NaZPgYQJPdlAUSInUX20\nmX2V0d4SuM/MeudpZxywd+ZAmS/+PczMtsnTznigt5nNzWhfAxhjZl3ysZNx7m64imJf4NIXG+KW\n1o4GjjGzmyo5fyKwdaZgop97+16+kuCBqiVM4QoUKw0yAyyAmX3ti7Lky2rZZiKY2UwfIPOlVmaA\n9Xbm+QpYkTGzFyVtzLLVa5+kDXblDLCesmyKtGa2UFKcCmOBKiCkCwLFylo59q0ewU4DX92qHD4w\n1o1gp262oOynmMVateWLy5wFnGhm7wPNJPWNYKJE0rpZ7C7XFqg+QpANFCuvSbo4s1HSBbif0/ny\nDE62e2mA9Isb7vL78uU+4ClfRjBlpyXwuN8XhwdwuddUymIGcFmE86/FaYP1lFTfb71wAo9xVsUF\nqoCQkw0UJb6HeD+uNsBk39wRmAYMMLM5edopwwWuo3ErvIRbBnsfcH62n9s5bB2HEyxMDU7NBa4y\nszvztZFh7z0z21LSZDPr5NveN7MOEWz0Ac4BNsNN/ZrufYpUCjJQdYQgGyhqJLXBlfADmG5mn8a0\nUxcnFQNOaTbO0tyUrfoA+Qb6HHbewhWbGWdmnSW1Ah4zs24R7fQws7EZbd3NbFwh/gWSIaQLAkWN\nD6p1cDpWn0pqJinySD6uJOHGfusjaV9JveWUcPNG0im43vBcSfdKmiRp5xj+AFwIvIzLxT4CvAr8\nO4adW7K03RrTp0DChJ5soKiRdBtQC9jezDbxy2RHZE7iz8POC7jcZ2rlVi9cgZcNgUvMbEiedt43\nsw6SdgGOAwbiainkXHabw97auNq2At4xs5kRzt0G2BY4FVdzN0UDYJ8oaYdA1RF6soFiZ1sz+yde\nR8vXGogzml8GbGJm+5nZfrii2wZsBZwdwU66IONDBQgyAmBmv5rZC2Y2HPibpCgFvGvj8sNluCpe\nqe13XN3bQBEQ5skGip1FXqbFYGnPL44AYjMz+ynt+c++7X+S8h78Ypkg44bAfxRDkFHSFrjR/w1w\npRdvB27DBfzr87VjZm8Ab0gabGbfeNslQD0vER4oAkJPNlDs3A48DTT2U7rGAlfHsDNa0nBJR0g6\nAvivb1uDPBUS/AT/C3Cj+V3N7A9cb3JARF8GAY8C+wG/4ORwvgBam9mNuU6sgCslNfD38gHwoaSz\nYtgJVAEhJxsoeiRtiqtZIOAVM/sghg3hglp33zQOeDqqaKGkaWa2edTrZ9iYYmYd055/mW81sFz2\nJPUDOuO+BCaGZbXFQUgXBFYG6uP0qx6StLak5mb2f1EM+GD6lN8KYZKkrmb2bgE26kjqxLJc7p/p\nz2PUpq3lV7DtDdxmZoskhd5TkRB6soGiRtJAXO+zlZm1kdQEeMLMekS0sy8uzbAOLpil6sk2iGjn\nY9x8229wulwpO1EqgyVam1bSybjBu/eB3YHmwMNmtl0UO4GqIQTZQFEjaQpOWXZS2qqoqVF/Ckv6\nHNjDzD4q0J8W2dpTA0/FgqQyM1tc3X4EwsBXoPj50//UT80uiFIcJp2fCg2wsDSYrgns4bc14wZY\nSSfICSCmnq8l6V8x7Kwr6T5JL/nn7YEj4vgUSJ4QZAPFzjOSbgcaShoAjMTVNIjKe5KekHSIX+21\nryrRyMqGX/H1CC7tsA7wsKSTYvgDrmbs0pkNZjYLOCaGncHACNyUMIBPcQsUAkVASBcEih5fBGVn\nXP5zRJziJ8quhxVZB0vSVGAbM5vnn68BvB1nJF/SNGCL1AwHOcmdqVFlZ7RMsia90Ey5GQyB6iPM\nLggULT7ovGxmOwEFVZUys6hzWStCQLq+2BLir/h6GXhC0t3++T99W1Tm+UUaqWC9NU71NlAEhCAb\nKFrMbImkUkkN4q5gkvRvM7tG0q1kUb81s5MjmnwAGC/pWVxw3Yv49WTPxgXW4/3zUcC9MeycDjwP\ntJKT22lMWFZbNIR0QaCo8cGsIy4XOy/Vbman53n+HmY2zK/yWg4zezCGT52BHrigPdbMJldySpXh\nl9FuDUwA2uIC/ydR6uQGqpbQkw0UO8P9FgszG+Yf/mFmQ9P3STqgAL+EC7KRUwWSnjSzA31ONlvv\nOu/8rpn9Jel2n4udHtWXQNUTerKBosQXPemfoL1JmeUIs7XlYecC4ABcPQXhVlkNNbO8ZWMkrW9m\nPyQ151bSdcDbwDNRlwkHqp4QZANFSZwAWIGdPriyhAcCT6TtaoArBB5VheAToENKVdYrLkwxs7Yx\nfLvazM7+//buPNrOqrzj+PdHEsKUBKyAVoiBMChQiYQgFRYyZjEZUhUVSS2DzChTERBaEFQorBYL\n1IHKqgOBJRSCLJB5MBDCEBKGKAEqKFBsSRligBAg+fWPvU/um5Ob5LznnsN5Oef5rHXXue97zrvf\nfbNunrXvfvd+npWda6Cd+cCapIdwC2hyN1toj5guCFW1Rt3+/qWU2N//IjADmEBK0l0zHzihiX69\nSKrUUCvdPZRUALEZe7BsLtu9+jm3QraHNXn/8B6IkWyopDw6e4j+g2wz+/uH1B4GSVqHlEv2sRLX\n11YnjATGkVYCmBQoH7Td8MYGSUcBRwOjgf8qvDWMVO9rUqNtFdqcAOyUD+/OScBDBUSQDZVUXFjf\novbuJo1mB5NGtC8B99luaDS7vNUJNWVWKUgaAawDnEtKS1gzP1d+KEXSeaTAPzmfOgCYYfu0sm2F\n1osgGyqpDUF2lu1PSvoaaRR7ZjOJZlpJqTrtC7YXStoZ+ASppE1DScQL7TwGjLG9OB8PAmZFPtlq\niNwFoapKzUs2YLCkD5MegDX9p7SkHSTdJukpSc9IelbSM002dw2wSNImwKXAhqSKCc1Yu/D9iCbb\nCG0QD75CJdm+FVJQA84CPkr6fa09OS9bSeBsUhKVabYfkrQx8HQTXbuM9MDsYZbeXtuMxbbfzYlq\nLrZ9saRmNjacC8zKeWpFmps9dcWXhPdKTBeESstJspcJarZf7lB/HrD9qVa1BXwfOJ2U6/ZZSbNt\nb9VEWx8mzctCehD3P63oYxi4CLKh0loV1CRtAFxMX42ve4DjbL/Q4PW1NbtfBAYB1wILa+83UTKm\nlvf1SFIWryslbQR80XZDhSIlHWv7kvz9lk7lyUPFRJANlZafnA84qEm6jTTf+Yt8ahJwYM7w1cj1\nLS0Z0wrFDRut2rwRWi+CbKi05QS3ZtbJLpNftVM5V1uVu6AuyLZ0NUZonXjwFSrN9i4tauplSZOA\nK/PxAUDpeV1J/WX/mkcqwf1Ig80cl1/3LXv/OmtL+hvSKqHh9ZUebF87wPZDC8RINlRaXrh/Jn27\nmX4DnG27VFLqnIzlYuCv86lpwDdcsrS4pCuAbYFadq99gceAUaREMeeXaW8gllPtoaZ01YfQHhFk\nQ6VJugaYDdR2VP0tKUFL6fpcLerPVGBv26/n47WAG4E9SaPZLUq0NZ9lpwvmkXItnGS72fW3oUJi\nuiBU3Wjbny8cf1upTHgpA11dULAehQdwwDvA+rYXSFq4nGuW5/vAC6QHcgK+TMpnMJNULHLnRhpR\nqnj7VdJoesn/6SaqPoQ2iCAbqm6BpB1t3wtLNicsaKKd/yAFs1qi7kn5XEOrCwomk8rP/Coffxa4\nIhdU/F3JtibY3rpwfGl+GHeKpG+VaOfXwP3A48Dikn0IbRbTBaHSJI0hTRWMII32XgEOsv1oyXZa\ntrpA0rb0jYin2Z5Rto3cznTgQuA/86kvACfa3r5M32L5VrVFkA3vC5KGAwygoOIdpJFrcXXBwbZ3\na/T+tv8s6QP9vd9k9qyNgX+l72HcdNLutv8GxtZG7w20cwLwOiknQ3Etcek+hdaLIBsqSdIk25cv\nZ8kUtv+lZHvF1QUG7qPE6gJJN9jeV9Kz9NX2WvLaRC6FlpF0DPBd4DX6HqR1tE+hT8zJhqpaM7+2\nJOt/rps1YQDX75tfN2pFf6ClD+NOAjax/X+t6ltonRjJhq4maTXgS8CrpLWtJ5PW3P4eOKdsYJIk\n4EBgI9vnSBoJfMj2g030bUBbfQvt3ApMtP1m2T6E9osgGypN0vnAd0grCm4mJbY+wfblDV5/FWmZ\n1ZqkagSzScF2R1Ki61K7riT9kPQEf1fbH8+lbG61PW4ll/bXVksexkmaAmwJ3MXSc7KxhKsCYrog\nVN1429/M20f/AHwOmAo0FGRJFWm3kjSYVIXgM/n8zZJKrVDIPmV7m1reV9uvSlq1iXagRVt9gevy\nV6igCLKh6mq/o/uQtq3OS3+xN+xtgJwc+8W695pJuv1OLu9iAEnr0vza1ENIc7IX0vcw7qCyjdj+\nWQ70m+VTT9aKRobOiyAbqu6GnLh7AXBUDmpvreSaog0kXURaBVD7nnz8kSb6cxEwBVhP0ndJa1vP\naKKdfh/GSTqetBOsYbk+2M9II30BG0r6O9tTm+lXaK2Ykw2Vl9emzrO9SNIawPBGM/+3sspsoc2P\nAbuRAtodtp8o28YK2n7O9siS1zwMfMX2k/l4M+BK22Nb1a/QvBjJhkqStKvtO4vp++qmCRpK41cL\nopL2t3113T327/+qlbY5B5jTzLUNKDUXkg2pBVgA209JGtLCPoUBiCAbquozwJ2k3AD1TINBtuA0\n4OoGzvWrLmOWCt8PBla13ar/S838aTlD0k/oexh4ICmTV6iAmC4IXU3SXsDepNpcvyy8NZy08mC7\nJttdCzgGOAKYYvukEtf2l+IQUvBevWzAljQ092XHfOoe4Ae2y2YFC20QQTZUmqTvAefbfi0fr0PK\ntdrQwyZJWwNjSCXB/7Hw1nzgLtuvluzP2sDxpNSCVwAXukOVc8P7QwTZUGn91a5qJuuUpCEDWdYk\n6YOk7atfIuV6vbhsdYZWW16NsJpGa4WF9oo52VB1gyQNrf3pK2l1YGgT7Wwn6Szgo6Tf+7KJXf4I\nzCVl8noTOLT4IK5swpoWqe1WOya/FrfnxuipIiLIhqqbDNxRqGd1MH2laMq4jJRG8GGa24RwAX2B\nqz5pTUcCWl5ni6Q96kb7p0iaCZzaiX6FpUWQDZVm+5/y9tfd86lzbN/SRFPzbN80gK5cZvv5/t6Q\nNNCqswMlSTvYnpYPPk2qYBsqIOZkQ+XlXLCb2r49b0YYZHt+yTbOAwaRln4Vk6jMbPD6OcCetv9Q\nd/5g4Azbo8v0p5UkjSXNE9eqR7wKHNLozxbaK4JsqDRJhwGHAx+wPVrSpsCPGq1oUGjnrn5O2/au\nDV6/N2m76z62n87nTgO+AuzVRA7Ylsvl0+n0A7mwtAiyodJyZdrtgAdq846SHrf9Vx3oy27Aj4GJ\nwNdyv/YpuwysDf0aCnyeZavVnt2pPoU+MW8Tqm6h7bdrBzllYemRgaT1JV0m6aZ8vIWkQ8u0YfsO\n0oO3u4GNSTllOxpgs18B+wHvAm8UvkIFxEg2VFpO2v0aafH/14Gjgd/ZPr1kOzeRll+dbnvrHKxn\nNToiLuzSEmkJ2TukVQq1pWDDy/SnlSTNtr1Vp+4fVixGsqHqTiWtT32ctIX11zSXWvCDtq8i5361\n/S4llnLZHmZ7eH5d1faaheOOBdjsPknv+fRJaEws4QqVZnuxpOuA62zPHUBTb0j6C/qSbW8PdMsD\noh2Bg3Il3YX0ja5jx1cFRJANlZQLFp4JHEv+i0vSItJ21mYe6JwIXA+MljQNWJeUcLsb7NXpDoTl\niznZUEmSTiQFj8NtP5vPbQz8ELjZ9oVNtDkY2Jw00uu6Ei2S1gNWqx3bfq6D3QlZBNlQSblQ4R71\nJbtz+Zlb65PGrKCdZZJ/F9kum5e2ciRNAP4Z+EvgJVJ+hidsb9nRjgUgpgtCdQ2pD7AAtueWzPrf\n6uTfVXQOsD1wu+1PStqFlCQmVEAE2VBVbzf53lJsn5lfDx5wj6rrHdsvS1pF0iq275JUqhhjaJ8I\nsqGqtpb0537Oi8K848rkud3l6lCKwlZ7LVdqmApMlvQSsRmhMiLIhkqyPahFTdXSEm4OjCOtMIA0\nffBgi+7RafuRSqafQKrvNYJUCSJUQDz4Cj1B0lRSnoH5+XgYcKPtnTrbs9aTtApwgO3Jne5LiB1f\noXesz9JzuW/nc+9bkoZLOk3SJZLGKzkWeIZUODJUQEwXhF7xc+BBSVPy8USaq7BQJb8g5Y6dTsoK\n9i3SnPVE2490smOhT0wXhJ6Rk1vXymZPtT2rk/0ZqGLKR0mDgD8BI22/1dmehaIYyYaeYfthSc+T\nVydIGvk+3xW1ZMea7UWSXogAWz0xkg09oZ9dUSOBOe/nXVE5l0NtqZaA1UmVdDuefjH0iZFs6BVd\ntyuqhcvcQhvF6oLQK96x/TKwZFcUsG2nOxW6X4xkQ6+IXVGhI2JONvQESWuSdkWtQt+uqMl5dBtC\n20SQDV0vL2+63fYune5L6D0xJxu6nu1FwGJJIzrdl9B7Yk429IrXgccl3UZhLtb2NzrXpdALIsiG\nXnEtfQm6a3Nk6lBfQg+JIBu6mqT9gA1s/1s+fpBURNHAKZ3sW+gNMScbut036cshC7AqMBbYGTiy\nEx0KvSVGsqHbrWr7+cLxvbZfAV7Jy7pCaKsYyYZut07xwPaxhcN13+O+hB4UQTZ0uwckHVZ/UtIR\ndE/5mVBhsRkhdDVJ6wHXAQuBmfn0WGAoKbn1/3aqb6E3RJANPUHSrkAtreFvbd/Zyf6E3hFBNoQQ\n2ijmZEMIoY0iyIYQQhtFkA1dRdIiSY9Imi3paklrDKCtnSXdkL+fIOnUFXx2bUlHN3GPsyT9faPn\n6z7zU0lfKHGvUZJml+1jGJgIsqHbLLA9xvZWwNvU7epSUvr33vb1ts9bwUfWBkoH2dD9IsiGbnYP\nsEkewT0p6efAbGBDSeMlTZc0M4941wKQtKekOZJmAp+rNSTpIEmX5O/XlzRF0qP569PAecDoPIq+\nIH/uZEkPSXpM0rcLbZ0u6SlJ9wKbr+yHkHRYbudRSdfUjc53lzQjt7dv/vwgSRcU7n3EQP8hQ/Mi\nyIauJGkwsBfweD61KfCDXJ32DeAMYHfb2wAzgBMlrQb8O/BZ0lraDy2n+YuA39jeGtgG+C1wKvD7\nPIo+WdL4fM/tgDHAWEk7SRoLfDmf2xsY18CPc63tcfl+TwCHFt4ble+xD/Cj/DMcCsyzPS63f5ik\njRq4T2iDyF0Qus3qkh7J398DXEYqA/5H2/fn89sDWwDTJEFKGjMd+BjwrO2nASRdDhzezz12Bb4K\nSxKCz5O0Tt1nxuevWfl4LVLQHQZMsf1mvsf1rNxWkr5DmpJYC7il8N5VthcDT0t6Jv8M44FPFOZr\nR+R7P9XAvUKLRZAN3WaB7THFEzmQFosmCrjN9gF1n1vqugEScK7tH9fd4/gm2vopaXfao5IOImUQ\nq6lf6O5876/bLgZjJI1q4t5hgGK6IPSi+4EdJG0CqciipM2AOcAoSaPz5w5YzvV3AEflawflsjbz\nSaPUmluAQwpzvR/JW3ynAhMlrS5pGGlqYmWGAX+SNIRUBLJof0mr5D5vDDyZ731U/jySNouMY50T\nI9nQc2zPzSPCKyUNzafPsP2UpMOBGyW9SZpuGNZPE8cBl0o6FFgEHGV7uqRpeYnUTXle9uPA9DyS\nfh2YZHumpF8CjwIvAQ810OV/AB4A5ubXYp+eIyW6GQ4cafstST8hzdXOVLr5XGBiY/86odViW20I\nIbRRTBeEEEIbRZANIYQ2iiAbQghtFEE2hBDaKIJsCCG0UQTZEEJoowiyIYTQRv8PFxAj8QtXb4kA\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f17a28bcb38>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUAAAAEmCAYAAAATPUntAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzsnXd4VFX6xz/vzKSSQIAAktBSgECAQEKxIIoNC3YE7Nh1\nV13buu7Psq5t7WJ3xd6x0kSwooLSixQpCYQSQCkhEEibmff3xzmTTEIqGYSV+32e88y555x7znvv\n3Pve97TvK6qKAwcOHByKcB1oARw4cODgQMFRgA4cODhk4ShABw4cHLJwFKADBw4OWTgK0IEDB4cs\nHAXowIGDQxaOAnTwPwMRuVdE3rHxDiJSKCLuELeRKyInhLLOerR5nYj8Zq+nZSPqKRSR5FDKdqAg\nIktF5Nj93Y6jAB2Uw778v4tIk6C0K0Vk2gEUq1qo6jpVjVFV34GWpTEQkTDgSeAkez3b9rUue/7q\n0EkXeojIGyLyQF3lVDVdVaftb3kcBeigKtzA3xpbiRg4z1fdaANEAksPtCAHA0TE80e25zygDqri\nMeA2EYmrLlNEjhSROSJSYH+PDMqbJiIPisgMYA+QbNMeEJGfbBdtooi0FJF3RWSnraNTUB1Pi8h6\nmzdPRI6uQY5OIqIi4hGRI2zdgVAsIrm2nEtE7hCRHBHZJiIfikiLoHouFpG1Nu/O2m6MiESJyBO2\nfIGITBeRKJt3hu227bDX3C3ovFwRuU1EfrHnjRWRSBHpAqywxXaIyLfB11Xlvl5p46ki8r2tZ6uI\njA0qpyKSauPNROQtEdli5b0r8EESkVFW9sdFJF9E1ojIKbVcd66I/N3Kv1tEXhWRNiLyhYjsEpGv\nRaR5UPmPRGSzlfEHEUm36VcDFwK3B56FoPr/ISK/ALvtf1o+FCEik0XkiaD6PxCR12r7r+oNVXWC\nE1BVgFzgBOBT4AGbdiUwzcZbAPnAxYAHON8et7T504B1QLrND7Np2UAK0AxYBqy07XiAt4DXg2S4\nCGhp824FNgORNu9e4B0b7wQo4KlyDWHA98B/7PHfgJlAOyAC+C/wvs3rDhQCg2zek4AXOKGG+/O8\nvZ5EjKV8pD2vC7AbONG2f7u95vCg+zobSLD38Ffg2uquo7rrsm1eaePvA3dijJdIYGBQOQVSbfwt\nYDwQa+tcCVxh80YBZcBV9jquAzYCUstzMRNjrSYCvwPzgT5Whm+BfwWVv9y2GwGMBhYG5b2Bfbaq\n1L8QaA9EBT+LNn6YbfM4jAJdDcSG5Jk/0C+dEw6eQIUC7AEUAK2orAAvBmZXOednYJSNTwPuq5I/\nDbgz6PgJ4Iug49ODX5BqZMoHMmz8XupWgC8CkwCXPf4VOD4ov619+T3APcAHQXlNgFKqUYBW4RQF\nZKmSdzfwYZWyecCxQff1oqD8R4GXqruO6q6LygrwLeBloF01ciiQilFqpUD3oLxrgv7HUUB2UF60\nPfewWp6LC4OOPwFeDDq+ARhXw7lxtu5m9vgNqleAl1f3LAYdnwusB7YSpPQbG5wusIO9oKpLMErk\njipZCcDaKmlrMVZBAOurqfK3oHhRNccxgQPbVfzVdp92YKzG+PrILSLXAMcCF6iq3yZ3BD6zXdMd\nGIXow1gzCcHyqupuoKZJiHiMtZNTTV6l+2LbXk/l+7I5KL6HoGtuIG4HBJhtu9yX1yBrGJX/q6r/\nU7k8qrrHRmuTqV7/oYi4ReRhO+SwE6PIAjLVhuqem2BMxCj2Fao6vY6y9YajAB3UhH9hukjBL81G\njEIJRgeMtRPAPtML2fG+24HhQHNVjcNYolLPc+8HzlTVnUFZ64FTVDUuKESqah6wCdPtCtQRjel+\nV4etQDGmK18Vle6LiIitN6+asnVht/2NDko7LBBR1c2qepWqJmCsuhcC435VZC2j8n9V9X/aX7gA\nOBPTk2iGsWih4j+s6fmo67l5EPPxaisi5zdSxnI4CtBBtVDVbGAscGNQ8mSgi4hcYAeqR2DG0SaF\nqNlYzBjcFsAjIvcATes6SUTaAx8Cl6jqyirZLwEPikhHW7aViJxp8z4GhorIQBEJB+6jhnfCWnWv\nAU+KSIK1dI4QkQjb9mkicryYZS23AiXATw26etPOFoyiusi2cTlBSldEzhORdvYwH6M4/FXq8FmZ\nHhSRWHvttwDvNFSefUAs5tq3YZT4Q1XyfwMatFZRRAYBlwGXAJcCz4pIYu1n1Q+OAnRQG+7DjIsB\noGaN2lDMC74NY60NVdWtIWpvKjAFM2C/FmNx1dU1Ajge06X9WCpmggPLSp4GJgBfisguzGD+AHs9\nS4G/Au9hrMF8YEMt7dwGLAbmANuBRzBjjSswkzfPYqyv04HTVbW0ntddFVcBf8fc43QqK9J+wCwR\nKbTX9Tetfu3fDRhrcjUw3V5jaGZOa8dbmP8uDzPhNbNK/qtAdzskMa6uykSkqa3zelXNU9UfbR2v\nW0u7URA7wOjAgQMHhxwcC9CBAweHLBwF6MCBg0MWjgJ04MDBIQtHATpw4OCQxR+68diBg+oQHx+v\nHTt2anQ9vhBO6LkbP8EYcoTq6kJ5ZfPnz9uqqq329Xx3046q3qJKaVq0Zaqqntxo4eoBRwE6OODo\n2LETM2bNbXQ9hcXeEEhjEBN58L0aPn9oVKDbFToVGBUmVXcGNQjqLSYibWSltOIFz9Zr508o4HSB\nHRxU+HLqFHqldyU9LZXHHn14r/ySkhIuumAE6WmpHH3kANbm5tZYV5hbiIt20zzaTVTY3i+9S6Bp\npIu4KDdxUW7C3NUrhlDJFMpr+2rqFPr0SKNXt8488djedT07+kmyMtIZkJXBaUNOYN3a6vVUKGXa\nJwjgclcOfyRCtanYCU7Y15CZmaVFZaqFxV5NSk7WZStytGB3ifbs2UvnL1qqRWVaHkY/87xeedU1\nWlSm+uY77+u55w0vz9uyq6xS8Pr8uq3QxMu8ft1eWDm/qNSnu4q8umVXmW4vNOXL82ydjZUplPUU\nlvi1sMSvBXvKNCkpWRf/mq3bdxVrj569dM7CJeX5hSV+/XzqN/p7fqEWlvj1qWee13OGDS/PC6VM\nwNzG/PcS3UYj+99WKTS2zoYExwJ0cNBgzuzZpKSkkpScTHh4OOeNGMmkieMrlZk0cTwXXnwpAOec\nO4xp336D6t5dQ4/LdBkDvcYSr59wT2ULT4HAUJ8IVNfDDJVMoby2uXNmkxxU17DhI/i8Sl3HHDuY\n6Giznbj/gMPZmLf3BpdQyrTPEDmgFqCjAB0cNNi4MY927cq5CUhMbEdeXt7eZdqbMh6Ph6bNmrFt\n294ELi6RSgrNryYtGHtK/UR4XDSPdtM0yk1hyd7s+qGSKZTXZsq1Kz9OTGzHxryaeQ7efP1VThyy\n95xCKGVqFBwFWH+IiE9EFloqoEUicqvsI/W6iNwntTjAEZFrReSSfah3iJVxod2XusLG39oXOaup\nv6mIjLGUQ/NE5DsR6WcJCnaEog3bzl9F5EIb727v9wIRSRGRH0PVzoFChEco8frJ3+NjZ5GP2Mg/\nePzpD8AH773DgvnzuOmWvx9oUaqHCLjDKoc/EAffVFfdKFLV3gAi0hqzybsphr6pQVDVe+rIf2lf\nBFTVqZiN/YhxKHSbqu41zSkiHlXdl6nL1zDUQKmqqiKSgmElDilU9fmgw3MwTMqBkfJqqeqrg920\nLlrB0VctEhIS2bChgvsgL28DiYmJe5dZv5527drh9XrZWVBAy5Z7M1j5VStZfC4xacGI9LjYWWys\nPq/fjMeLQHCxUMkUymsz5Sq6tHl5G0hI3Jsc5btvvubRhx9iytfTiIiIqL6eEMm07xBwN/zDI8Zl\nwysY8l7FsFCvwDAYdcLwEA5X1fza6vmfswCDoaq/A1cD14uBW0QeE+Nn4hdLkAmA9Tmw2FoxD9u0\nN0RkmI0/LCLL7HmP27R7ReQ2G+8tIjNt/mdifSCI8dfwiIjMFpGVUoMPiyA5rhSRcSLyHRVK8g57\n/i+WAipQ9lKbvlBEXhDj36Ir0BtDQa72PuSo6hdV2mkqxsfEfFvvUJseK8aXwyIRWRJ0/Y8FXf8j\nNu0BEblJRM4ArgduEOP/oZKlWZ38YnxXLBORdzEOf9rW9X/27deP7OxV5K5ZQ2lpKR+N/YDThp5R\nqcxpQ8/g3bffBODTTz7mmMHHUR0piNdvlnsEVnxEeFyU+iorQL9q+cxvYAK46vBWqGQK5bVl9e1H\nTlBdH384llOr1LVo4QJu/Ou1fPjJeFq3br1XHaGWaZ+x77PATwNTVDUNyMAYBHcA36hqZ+Ab9ib0\n3RsHegawoQEorCZtB4YO6WrgLpsWAcwFkoBTMJRC0Tavhf19AxiGIcFcQQU7Tpz9vRdjvQH8Ahxj\n4/cBo218GvCEjZ8KfF1FtmlA36DjKzF0Qc2DznkB+yhg6KCOxHzZxlFBlf4yhmzyHOCjGu6NB9hh\n42FAUxtvDayy8RFUpjNvZu/d0mqu/wHgpmriwe3UJH8qhqeub3WyBofALHBRmepnEz7X1M6dNSk5\nWe+97wEtKlP9551360efjteiMtX8XUV69rnDNDklRbP69tNlK3JqnAXescerXp9fvT6/Fhab2d7d\nJT4t2FMx81vq9WuZDTtsevAscGNlCmU9wbO8n4ybpKmpnTUpKVnv+ff9Wlji13/831069uNxWlji\n12MHH6+tWrfWnr0ytGevDD31tNP3mgUOhUw0dhY4NlEjT3y0UqirTvvMrgk8r0HpK4C2Nt4Wwx5d\ne/ta9ZN3kENEClU1pkraDqArxmlNLwzlOJgbdQ0wBFiuqmOqnPcGhsxzHDDPhknAJFUtFZF7MU5z\nxgCLVbWDPS8Fo4QybRf3TlWdISJtgBmqmhrUxjSCusBivHsNUNWr7PFoDINugT0lBqNs4jCccFts\nehSG0HIpcL6qnlfNvfEAW1U1TgzB52hgIEYRpWEcA7XCEJuOBSZaucMwTm5mAZ/b6y8T4791q6qO\nrhIPbqcm+adjfH90riqnlfVqzAeL9h06ZK3MadR6WsBZCF1fhHgh9DxV7buv57uatdeII26ulFY8\n9da1GF7FAF5W1ZcDByLSG2MQLMNYf/Mwzq/y1LCIB4Zd8gPHNeHg+5cbCBFJxvh4+B1jhdygZgwu\nuMyQ2upQVa+I9McQaw7DdPeOa4AYJfbXR/3u6e6guGCcxLwaXEBEbgZeU9W7q6R3BXqLiEtrH1O7\nBPMByLTXtwHjXe1XEemLsdweFpEvVPUhm3YicB7GS9hJ9biO2uRPrXKdlWAf6JcBsrL6/m99hR2E\nEFJdt3drHUrVA2Ri3vVZIvI0Vbq7qqoiUudz9T89BigirTCU58+pMWWnAtdZiwYR6SIiTYCvgMvE\n+HxAgvzC2uMYjNeqycDNmK9KOVS1AMgPGt+7GON6MRSYClxh5URE2olIPPA1MNzGEeNLt4Ma9uHF\nwD32K4eIJMnefl2bAb9b5Xci1reHGCrxQlV9G+OhLVNEYjHd5Un2+vuEQH4HDurGvs0CbwA2qOos\ne/wxRiH+JiJtTbXSFmMU1Yr/RQswSkQWYsa4vMDbGH+uYGaFOgHzrXLYApylqlOs2TxXREoxXcD/\nC6ozFhgvIpEYi+aWatq9FHjJKtHVGB8FjYaqThaRNGCm1We7MF7NFovIv4GvxSzzKQOuxfjdvcxe\nc7aIFNnrvK1K1W8DE0VkMcYn7SqbnoGx/PwY14nXYpTlp2L8W7hquP4Gyd/A2+DgUEYD1/6p6mYR\nWS8iXa1BcDymO7wM854+bH/H11IN4FDiOzgIkJXVVx0yhLrxpxwDbN5JI469q1Ja8bir6qzTGjSv\nAOFUGCQujDOoDpiJxuGqur22eg6+f9mBAweHDMwqmIavA1TVhUB1SvL4htTjKEAHDhwcOIggIbRI\nGwpHATpw4OCAwuU6cHOxjgJ0cMDhB0q9te6SqxfaH31T44WxyJ/zXEjqCdW4HYRujHNP6d6kDwcK\nIoLL7ShABw4cHKJwLEAHDhwcmhAO6BjgAd/b64QDF4CTMfsns4E7qsmPwGyZy8Zsk+tk00/EbD9a\nbH+Pq+bcCWeffXauqq5Q1ew333zzG8x+6oXAl0CCqnYoKSn5cdGiRbpixQq94ca/6c9zF2lBkc+w\nNRcWak7Oar3k0lH6r/se0oIinxYU+XT4yAt06dKl+t130/SiS0bp1p3FWlji0/nL1unC5ev18Wde\n1rCwMMUTpacPv1IX/pqrCxYt1hUrszU/f4e6E49WV1yqtmydoD17ZehXX3+jzzz/krrie6orJlFd\nnnD9y1/+ohs3bdJbbrlFw8LCVETU5XJps7g4HTz4OFVVHTz4OH3o4ce0WbM4BdTj8WhGRobefffd\n2j09XS+5dJRGRUVpRESExsfHa+fOnTU5JUW7dE3TsLAwDQ8P1/j4eJ0zd55u31mkhx9xZGD3giYn\np+jGLTt0+65i7duvv4qIiki5LBOmfqfpPXtpRESEYs9p2ixOF2dv0NEvvKItWrbU6CZNVETU4/Ho\nwGMG6+aCUp27JFujoqIVw6CirdocprOXrdXv5/6qbQ5LKK/ruJNO1dVbivTH+cs1vlXr8vY9YWEa\nFRWt7342Rbul9wrUU4QZyZht//s3MHt1NwPFmGUqN9q8SzA7hPxAjqdlssaP+qBSwCxyDjxHx2G2\naS4B3qRib3wzYCKwCLM99LJ9egcO9EvohAMTADeQAyRj1lItArpXKfMX4CUbHwmMtfE+VoGBIW3I\nq3LeOR6P5/21a9eWqGqyqoaXlZUtVtXuNv9GW+/LTz311JeHtW2ru4p9WlxSooOOPU53Ffu0zOvX\nLl3TdNmqNer1+bRXRm+dNX+xFhT5dOEvS7SkzKelXr+ee94IfeLp59Tr82vG2fdpeK9rtGmzOH30\nyWfV0/FElciWGtZluEb0ulZvfvhDfePT6YorQtum9tOf5q9S8USpu1mSzly0WgedPFLdCUfpHXc/\noCcNOVkvv+IqTUpO1vj4eM3s20979OipHTp00N69e+vnn0/WBb8s05iYGO3du4+mdu6s1994k3bp\n0lXPOecc7dSpkzaJidEff56jG7fs0PDwcJ03b55u31Wsbrdbb7jxRi0s9umJJ52s4yd+rs88+5y2\naNFSn37uRR16+pma2K6d3v7PO/WJ0c/qucOG6/SZc3Vx9gZN6dxF3W63fj9roW4uKNXemX31lbc+\n0CZNYvSx0S/o5oJSHf3CK5qRmaVxzZvr8tzfdHH2Bu3evYduzC/WfgOO0LCwMH37k8k6Y+EqTWzf\nQRes2qgnnDxUmzaL0/fGTdVOyanaMTlF3/5ksp5y+tl6+9336zczF+s7n36h3dJ7amRklC7M3qSr\ntxQpZvtnC2A7FWQjb2D25b8FuGxaa8z+9hXAUOBB4F5PfLK2vuLDSgGzl787Zl3feqCLreM+4Aob\n/z/gERtvZdsPb+h78D+9Fc5Bo9AfyFbV1apaCnyAITUIxpmYry6Y7UbHi4io6gJV3WjTl2J250RA\n+bbCW1577bUJubm5pZivf6nH43kvqP4mWAvE5/O1j4mNRQC3J4x1a3Mp2VNI7tq1tGvfnsR2HVAV\nrr/xJj6fNAEB0rt3p8zOmWT17V/Ohty0SSS+jT/TsnUCXg1DxIW7eWd01zrE5Wb4yVl8+MUsUC9t\nknoTEeYisllb/Ls34Xa7yC+N5rC4MJrGxrBs2VIys/oS3zKePn36sHbNas4bMZKuaWlERkWxYcMG\nuqalUVxcjMfjITW1M70yepObu4Zp06aRv2MHyUnJ9MnM4tdlS4mPj2fFylVs37YNj8fDli1bcLmE\n8y+6mAnjxzFhwgSKi4u4/MqrSe/Zk+3btjHu009Y/usyTj71NHr3yaRVq9aUFBfT5rC2bN5obn9U\ndDQglJWV0qNX7/I/bn1uLqcMPZO45s1p1ao1LeJbMeGzj9m+bSvNW7QktmlT2ia2Y9DgE5n2zVRm\n/fQj3Xv24vCjBtEpKYXSkhKmTBpH9srlnHHuCJJSUjli4DGs+HUprdq0YdvWYK4ChmGIL/YEpZ0E\n3Kd2v7oa6roLMCQikzA7mwrBjAEGB4wyOxPD0lSqqittnV8B59q4ArF2x1eMPafBs0SOAjx0kYj5\nugawwaZVW0YNcWsB5qEMxrnAfFUNEELcDzzRrVu3uLy8vLLg+qdOnXquiKwHLgTuAe49//zzW37/\n3XdEhwsLFi1h/bq1FBcXsWnzZhItXbtfoVOnjmzKyyMyTCj2mplVVT8fvP8OJ5w4hOIy5bNnr+P5\nB64ns3cGjzxtiH8kLAYt2037luF0bN2EKa/9E9wR/JpXyvdzV7Exex4d27VlwpczWTpvGgmtYsnf\nuYfdu3fjV6VJTAzt23egabNmxMfHU1JcwtIlSzn++OOZPGkiERERFOwsoEmTaCI8Qrt27YxSUmXT\npk0MPeVELr3ofLZu3cradevZuDGPyMhIvpg8mQFZGUye8Bl5eRvIy8ujS9c0Jk0Yb5SACHkb1tOz\nVwafT5qI1+tlbe4a8jasp7BwF5l9+5ff2DtuvR6/3897b78esMDZubOAT8a+R7dOhzGofwbz585i\n8aIFuFxudu4sYMTQE+ib1p41q7NZk5NNXFxzcnOy2bBuLRGRkeRv28amvA2kpfdk6iSzo+z9N1/B\n7/fjcrnpmJQcaN4FPA5kichZQf93MjBdRH4TkSki0hlD2tvcMiRdDWQJZh1gcMBs0UzEMMJ4LFEH\nGEUb4PC/FtOdLsUM0fwNiBORr0Rklf1tTh3YrwpQRNqIyHsisloMdfvPInJ2I+oLJiitlc6+jnp6\ni8ipQcejRGSLVFDtfxwgTggFqmnvDBGpm6yx5vrCxBC4rhJDePpzgAxBRHJDSEYQi+kGB4gnEoFv\nReRoEZksIocDj2AoxwLbk1JU9bPqKhsyZMhMVW0PvIth3Dk/KipqdM+evbj4kktp26oFGb0zayTc\n7NmzB6oVzouWLV3CUUcdzZEDjybcIww991Kuv/dVstf9zhMPVL69w08fxLjvlxOWdgH4SmjfOpq0\nlESSBl7B2nXrOaJ7S44+ehA1uQ2PjIwk3ONmyZIl9OjZi3bt2/HQg/fRNiEBMO4jn376GRRwu900\nadIEr7eM3zZvJjY2FhFhdU4OAK1at+GCCy/kh59ms2PnLnKysxGBhx55nDH/fZHXXxmDqhIWFsYl\noy4nMTGRo4/ox/VXj0JEuHjUVcQ2bQrAC2PepHmLlnRN687SxYv46IN3OOmU0zj62ONISGxHZFQk\nv/22CZ/PT3FxMevWruHMYSNZnLuFPv0GsPLXpSxfshiXy8X9jz3DDVddxMwZP+B2u3G7Xfzfv//D\nrJ9/ZOjgwxn77puICPc/Ojp45nYZhg3pDGC0GKq4f2LGBZ/E8EPmY1jMPUAWcBqG2u0EQwjtrhQC\nUKPNRwJPichszD7zwBqeKMx+93AMFd5zmI9qgwhR95sCtKbpOOAHOw6UhbmYdlXK7dNMtKreo6pf\n76N4vTF0UMEYq6q9VTUd81UZsY9119meqk7QCmr5fcH9GMLHHqqaCZyFUVYNQR4VX1Mw/0tVzzrL\nMQPNYCY+ADJU9UfMF/wt4BJVzbF5RwB9RST31ltv/XfLli3j7Ncer9fbPqj+dzGW4xXNmzd/u0On\nTrww5g0SEhJQv48mMbG0Peww8ixdu0sgN3ctvXtn4HFDTITgL9lDWloao0c/ZbrPArOmf4V3y2JW\n/DKTI7K6U7bhB3w71yFhTRg2JIsPp8xFwpqAO5LD05oxe3EuRZ42qHiYnVPE4YcfzsZtRTRvGk2T\nJk1wibC7sJD169fRskULXnzxBWJimnDqqSfjdrmYMX0GnTolEdcsjrwNeeTkZLM2N5cd+flER0cT\nHh7BnAWLefGllwkLC6O4pISEhETy87fTLjGRsPAIMrP6UlRcQmJiIhEREUyYPJVLL78CFJJTUvF4\nPDzy+FP88NNsNm3MIya2KcMvuKj8D9q+bSs+r5emcXEce/xJLJg3lxYtWtK+fUf+dts/mbskB7/f\nT3hEOC1btaJDx064xIXH4+HU08/BExbG7t2F7NxZwDHHD+GzqT/SJ6s/sU2bkZTSmTaHJfDSG2N5\nf8KXrPx1KU1iYhh4bCW7Ixb4zHZTpwF9VHUTpkfxEfC6LdPLpk1V1d2Ycb7VQHUWYHjgWVHVn1X1\naFXtD/wABLrDMcDnapCNmXQ5m4ohmzcx70Wt2J8W4HGY/nu5Xw1VXauqz1qLa4KIfAt8IyIxIvKN\ntWYWi0j5WJSI3CmGan46RtMH0oPp7LNE5HtrZU4NosTZi65eDFHofcAIa/FVUnRWITfBfLUQkU5i\nqOV/sTJ2qCP9PDFU84tE5Ifq2rPX/1zQdTwjIj9ZSzlwTS4xNPjLrTk/WUSGWcv0KgwXWom9r7+p\n6odV/wAx1PvzrFV7tU1ziyGCHQMMFpH7rYzXA1fa6/nAVvEbMMZads/YtAUichjm4X1QDaHqRfYL\nfQ1GYaasXr36iNTUVOnfv/+qyMjIXwoKCq4BJtg6zsQo13WbNm063e/34xLYU7SHrt264w6LoFPH\nDqxft468vPWIKM89M5rwqFgKS5TnXxrD7f/4ByJuir1mMAiB9JNvI6LHpYRFRPP9jLl4EgeixdtI\n6pxO89hoZi5ag3qLAR8zZ/zA0VmpaP6veJq2Y1BmKsvm/8AWb2uKS8ro3r078+fNZcuWLSxYuJC8\njRvJzs7G4/Hw1ptvUVJSAp4ITht6BmHhYRTuLuTiSy8jq28/TjzpJJq3aEHrNq3Zs2cP4WFudu/e\nTf8Bh9OiZUt2FxbSsWNH1O/ng3ffoX//fgw9/XReGfNfAH5dtpTIyAiuvPpa9uzZQ2FhIdddfQVb\ntvxOckoqXdO6A+D1ennv7Tc4a9gI/H4/s3+eQVq3dH7bvInBJwzhh2nfMHXyROLt2OHFo65EXC6+\n/fILCnbk891XU8jftpVBx5/I4UcN4sN33wBgTc4qior2MPyiy9i+bSvFxcVcet5QAEZd9dfy56tg\nRz6YCZD3ba/jKGCZff/GAYMxSmgnRnGNBwbadywM6CAi1Y0BtgYuFJG59p25WswY8z8w9HdgxvvG\n2Of7VoyscGgHAAAgAElEQVRuiLPKF8wMdJuq78Re2I+zjDcCT9WQNwrzNQhQ03uooG+Px/TpBWMu\nLwaiMY6PsqmgqH8DMyYQhqG7b2XTR2CIRKEGunrb/nNV5NmCWaLxG/Aj4LZ5E4FLbfxyYFwd6YuB\nRBuPq6W954Ku4yPMx6g7ZmICe22TbfphGIU8DPMlXVDLfc8F4m08cH+jMMsIWtp7+lXQPcnGzAbv\nxCx7uQ/DOA1G0WbbsBp4y6bfhVnGsBjji6GAilnhFzBjM51OOeUU3bFjx0ZVzXn22WeXAEuefPLJ\n3y6//PLZwO2ZmZlL5syZs2fRokW6bNkyvfuef2nuxq1VlsHk6KWjLtO77r2/fBmM2+3WESPP12nf\nf689e2Xo/919r+4u8enilXm6aMV6fWT0yxoWEa14otRz2AC97+n39LgTTlLCYlUiW6o78Wh1x6Vo\nqzaJ2rNnL50y9St9cvTzGtbRlPGER2mTJk20bdu2eumll6rH41FAXS6XxsTEKKC9emVoenq6Jien\nmKUodlInPDxcf545W9O6ddO2CQkaHh6uUVFReuzg48op7Nu2TdCIiAh1uVzatm2C5u8o0G0Fe7Rb\n9/RK9bRNSNBZ8xZpu3bty9vv3DVN03v20nc+Gq85G/PV4/Go2+1WEdHIqCg9ZvAJesPNf9ek5FR1\nuVzl6Q8++pRuLijVN9//VD1hYQpmWcsRA4/RVZsLddqcZRoT27S8/ajoaO3ctZs+9+q7Gt+qtQIa\nERGpXbv10G7pvXTStzP1+dfeU/scLLLPQmCG9lvM5Ngu+9zOwvQcsM9XGaYrWxTeOlWTbv68UqDy\nMpjH7DO2AuuSwaZnYpZTLcN0t+/HumkIKpNfl57ab3RYInIjkKSqN9vj5zH07KWYKfJjVPUymxcG\nPAUMsje0K8aXx0jMSxxwtPMksFFVH5cKOvvlGAW42jbtBjap6klSA129iIzC+Kq43tZbfmy77s8D\n61T1YRHZivEzUGbl3KSq8bWkvwSkYGh5PlXVbXW09wZGIb1r83apaqwYqvlFqvq6Tf8U4wFvJfCm\nqlZLWioiubburWIo/QNjrp0wrgFWYHylTMbQ33+pqn4RmYKZlRuHUeaFVeSseg25GDaOkZglCQHy\nySiM97h7RcQLRKhqrXuv+mT11e9nzK6tSL3Q5ogbG11HAM5WuPohuVVUo+iwIg/rrO0ueqZSWs4T\npzaoTqlwXXEVcKyqbrJW6DRV7VrbufuzC7wUo6UBUNW/YqhqWtmkYLr0C216lhqXl78BkfVsR4Cl\nasbveqtqT1UNpnNvEF29mi/CRIwybjBU9VqMhdQemCci9fEhWBIUr2tZfDam69C0tkIicixwAnCE\nqmYACzCU+PkYUtRpmJm0V+wpp2EUfyYwpwFjs4JRyIH731VV77V5xXUpPweHOgSXq3Ko8wyRJmJY\nzBHDRH4SpoczAUOECvUkRN2fCvBbIFJErgtKq2lmNUDfXiYig4GONv0H4CwRibIXfHo1564AWonI\nEVA+Q5peh2y7qH3SYCCmWwjGuhxp4xdiusc1potIiqrOslbrFowirKu96jADONeOBbYBjgVQs9bq\nVeBpO3aHiLQSkapOkpphugB7xDA2H27LxmMWp36CUdSZYhin26vqd5hxlmaYQeb64BtgmBgfzYhI\nCxHpWMc5DhwAhhHf43FVCvVAG8wSm0UYtvPPVXUKhgn6RBFZhfn41znRWONXvi4LQ1V31pGvYtYF\nPSUit2OUwW7MCxZVpfi7VNC3z8V0a1HV+SIyFjPG8Dswp5p2Su3EwTMi0sxe02iMBVoTvgPuEEOt\n/x+bNkJEBmI+Chsw43QANwCvi0jAQ9tldaQ/JmbNk2CUwyIMjX3V9urCJ1RQfa/HbAcKeF67C+N5\nbZmIFGPua1Un71OAa0UkMH4y06YnWrkDT9o/McMG79j7J8AzqrpD6uH/VVWXichdwJdSQd3/Vwwj\nrwMHdaI+Vl8wVHU1Vfz22PRtNJAQtcYxQDELVpXKXbLAsap1Eelg/0FEYuxYXEvMl+4oVd18oOUK\nNZwxwPrhzzgGGJ3QRVOvfKFS2uL7T2xUnQ1BjRagmgWrDg4sJolIHGZd1P1/RuXn4FCH4D7Y+QBF\nZCSQrMZ/bDugjarO27+iOVDVYw+0DA4c7E+INLwLHErUqXrtgt3BGF+4YFZwv1TzGQ4cOHBQfzR0\nFjiUqI8FeKSqZorIAgBV3R6YfXTgIBRwAeH1m/2rFet/HN14YUKMULqgbBZdL6fhdSIqvOFe2PYX\nRMDtPridIpXZ2T2z48gMyDfegYMDBw4c8MdbfcGoz2f3ecySjFYi8m9gOoYBxIGD6lCJZbqGMsMx\ny3uWYna3lGNn/lby1uawPjebl//7Ynm6xwXhbghzGdqn9LRUjj5yAL//tplwt8lrFlVh2QgQG+ki\nLtpNXLSbYAMzMkxMepSb6PCKDLfL1BEX5SbYSPpy6hQy0rvSIy2Vpx57mOD31SWg3hIuuXAEPaxM\na3NzAQizMoe7jTxfTp1CL1vPdVddVl7nY48+jLtKPd06J9ErPY0BWb057piB/LpsGQBzZs9mQFZv\nundJoVmTCDq1O4zHHn2Y4uJiBh7Rn/6ZGWRmpPOvu+/kogtGkJ6WSvu2rejdsxv9+vTi/BHDKCws\nBODVMS9xRN8MBg7IZMhxg1j+q2lj3pzZDByQycABmRzVvw8Txxtyn+LiYgYPPJyj+vdhQGZPHrr/\n3or7LfKqmP3vv4hhU4qx6deK2d+/UESmi0h3m95fRBYi1XaBu9m6llqdU2sbjUJde+XsMpl0DN/W\n3zAMJAeExdgJB31wq2qOWhZoVV2klgU6KHRWQ6ja3B63zszM0qIyVa/PrxdffLEuW5GjBbtLtF+/\n/jp/0VIt9ap6fapFZaqjn3leb77lNi0qU31v7Me6bv16LSozeVsLy3TLLhOKSn26s8hbfrzV/u7Y\n49WSMl9FetA5ZV6/bt9tz7d1FhZ7NTk5WVesMjL17NlL5y9aqkVlqsVlqj6/kenKq67RojLVN995\nX889b7j6/KolXq1UT1Jysi5fmaO7CvdoeHi4Tvh8SqU6A/WUeFXHjHlVzz1vuBaVqX706Xg98aQh\nWlSmuq1gt+4oLNak5GT9/sefNT6+lfbo0VPnLVyiW/J3aVGZ6s49pdqhY0c944yztKhM9aWXK+q6\n4W83l7sXWP9bfvn+6vc/+kyPP/EkLSjy6aZtu3TbrhItKPLpitUbNL5VK922q0R37PFq3pYCLSjy\n6dadxZrVt78Cv6oq2L38Nv4k1sVClfQzgCk2Hg14Ytp11UFPTq8UMPySYPb5zwIOr62NxoT6Dry4\nMQtcS3FIVB3UjP5UECfUxDJ9FaZXkW+PfwdjIRXu2sWmzb+RlJxMeHg4p595FpMmjsftgoDXzEkT\nx3PGWecAcPbZZ/Hh2LGBFwL7gwBhbqHEW7EGLxCL9AhFZUHpNhrmFrx+xVdlcGfO7NmkpqbSoZOR\n6bwRI5k00eywcrvA5zcyXXix2YF1zrnDmPbtN6gqwUsA58yeTUpKKp1Tk5k3fyHtO3Rk4YL5leoM\n1OMWuPDiS8rr2b17dzlHYnR0NAvmzyclJZXWbdrgcgnnnjeczydNICbGGERlZWXkb9/OkFNPAyrq\n8vv9FBcVldfVtGnFXoc9VdrweMzoWHFJcXm6iFRqo8xbwXerdmOE3UsfFbjlWnnDRJOg9D2q6q3B\nAgz8C2E2aA1tICILRGSSjSeJyCwRyRaRsfWZq6jPLPCdwPtAAoYz7j0R+Wdd5zk4JFEflukuNszA\n7E45GcxgeOHu3Tz8n/8Q7jZd3nbt2pOXl4dguprhbrjn7rtISkoCwON207p1a1zqJdwNER7zorpc\nhjQ1JsJFXJSbmIjgbq4Q5haaRblpFlXRNQ4sRWsaac4JjMtv3JhHu3bty7vgHdq1I89S8IuVe9PG\nPJI7tccl4PF4aNqsGVu3bivvAntcFfUIsGlTHokJCWzeaK4tMdHUuXFjHu3am3rEbRRQWudk7vzn\n7TzxVAVhwA/fT2PunNn07dOTZ55/iQ4dOpKXl4fP52NAVm86JLQmPDyck4acbO6Tx0NJSQkdEw9j\nxYrlXPOX68vrGvPSC2R078w9d97Bo088XZ4+d/YsBmT25Mi+GTz1zAvlCtHn8zFwQCapHQ5j8HEn\nQNCefhF5HUNDlQY8G5T+VxHJAR7FsEQF0gdIDXuB7a6p3zFEIbNqaEMxTDEBPIJhoErFfGCvoA7U\nx5q7BOinqnep6p2Yr/yoepx3QCAihUHxU8XwAHYUwya9J7BntWrZWuqbbBcj11ZmmlTQdgenl/P+\nhRoicpsYrsCFIjJHRC6pTZZ9bKOviDxj4xEi8rVUcBq+EhjPaSA8QGfM3ubzgTEBq6JVfDxjP/yQ\nUp9RLqmpKZVOLPXBx598Quv4CqbzHj16sDW/gFIfRIe7jPLAKJ3iMj87inyoUmmsT4CCIh+7S3zE\nRrrL08Lcwq5ic47bRflYn4hRqKU+Y5oET1oK5i0s85kxvwBcYqzWwLVUHedXNXWFVZmQdUkF43XT\nZs2YMXMODzz0CA8/9EB5mdTOnTnr7HOZ/vMcHnvkP5SVlQKGWXnWvIVk526gqKiIlSuWl5/TqnVr\n5i5cTFpaNz79eGx5+lXX/oVFy1bx7wf+w2MPP1ie3rf/AGbNX8x302fx5GOPUFxcXN7G9FnzWZa9\njvlz50AQaYkadqcEjFIaEZT+vKqmYLbB3hWUPksEPG5XpWDzemMMrv4i0qOaNtZhCA9eMf+RCIaD\n9GNbNGSEqJuoPFvssWkHNUTkeAyJ5ymqGtiXuhW4tSH1qOqpqroj1PLVBTGo9v8RkWsxDM397YNy\nPHWzyDQYqjpXVQNf7D42rbeqjlXVK1V1WZVTamSZFpHAq74Bw9pRhmHxXdmhQwdUYU9REQsWLADA\npxAZEU5iYiJKEA3+sl8J85iqfD4fX3/9NS1aGMKdMp/icQk+S5sf6DaXeP3llp5flVKfqSyQL7a9\nMp+Wd5V9fpOekJDI+vXry9vfsGEDiYnGqA3IFVzG5/Wys6CAFi1bVtSlkJiYyIYN61GgbdtENm7a\nSEJCIgLk5Zk6ExIS2bhhPT41ZKc7Cwpo2bIlw0eMZOKEceU3NSHB1JXWrRsxMTEsWDC/XCaAuLg4\nDjusLRPHm656oK7WrVtz3oiRTBj36V7/9bnDR/L5xL3JU7qmdaNJTAzLli6plB4XF8fRxxwLhjij\nHGrYfz6gwnlRMD6gGqXkdkmlAMSLyFzga0xv4V/VtNER2EZFd7klhg8wsF+wut7HXqhRAYrIU2L4\n97YDS+0XfwyG+HBrTecdDBCRQRjG46FaQdcOxi/BCBFpUc05F4lhjl4oIv8NvLAS5GNDRO4WkRV2\nNut9sf5JLM6TIObpoPT21ipbJSL/CmrvFjHM0UtE5Cab1snW/xaG3qe9GMboJXYm7WZ7+v8B1wXG\nRFR1p6q+SRWIyItiWHWrzqY9LCLL7Gza4zatEpO1TTtWRCZZq/kdoJ+9PynBlqaInCQiP4eFhT2/\nefPmo3744Yd0IHzZsmX/Pumkk1JEZD4QYKsZh2W2wZDfdsnLyzMuvmJj2b59G7lr1oD6+HLqVE4b\negZ+rbCgrrnmWjZs2ADAlClTGXLyyeVjVEb5qbGutMJSC/e4ypVdqVcJsxmBOhUo8xrlGYBLTHrf\nfv1YtWoVa3PXUFpaykcffsDQ088AKJfrtKFn8O7bb+IS+OSTjzlm8HGV/Jq4BPr27Ud29ipyVq+h\nd+8M1q1dS2afTEpKS/lo7AecNvQMTht6Bm+/9SZ+hReef7a8ni8mf05qamcActesoXefPmRnr2LG\n9OmsWL6cH6ZN48gjB7Jjh/lOFxUVocDqnGxUlReff45jBh8HwKSJE+jcJQ2AnOxV5TJO/eJzkgNt\n5K7B6zV6ZN3ataxasZyOHTuxdcuWSm18983XAMX2Y51qnwXBTHYst8edgx7J04BVNj1JRDzGOpZK\nAcNi1Bc4GjOe/GaVNoZiFO8sGona1gEGVP5SDHFmADOrKXswIQL7kqnq8ip5hRgl+DeCvioi0g1j\nsh+lhpLrBQzF1VtBZfphvmoZmIHZ+Rin4AF4VLW/GOdH/8LQ8YAZMuiB2UEzR0Q+x7xblwEDMIbG\nLBH5HjNu0RnDND1TRLIw7NI9rAxxYlh6YtUwYtSFO9UsXHdjXA/0wlhkZwNpqqpB3ft7gCGqmle1\ny6+qv4vIlRg27qFWlsB9icd0a04oKyvb/Y9//OO122677Xug4NNPP9391VdfrVTjQjPgMnEqcNK2\nbds2btu2Le7xxx/flJ2TDYBPhW+//ZbNmzYxYfw4/Ajd09N56KGHuOrKK2jTpg1nnXUmt99+O+PH\nj6N58xZ8PvmL8iUre0r95ZMYhSU+YiLd1rpTCotNRrFXiYkQ4uySmcISk65AUZm/PD1g3Xk8Hp58\n+jlOP3UIPp+PUZddTpdu6dx37z1kZvXlrDPP4OqrruCSSy4mtXMqzZu34O13P8Drp1wuVTOm99TT\nz3HqKUPw+3yMHDmCm2+6gS1btnLyqafSPT2dse+/y+LFi0lPS6Vw1y5imzZjQFZv4po3Z8xr5vv2\n04zpPP6YYXk67eQTaBYXx4jzL6Bps2b07pFGRGQkMTExjDz/ApYv/5X0tFR+/+13EhIT6NunJz17\nZvD4088D8PKLzzPtu28ICwsjLq45L415HYCZP03nqccfJSwsDHG5eOLp52gZH8+Sxb9w7VWX4ff5\n8Pv9nH3uefz80/QC+wy/aZ9NwTAgBWjwrhfjvKwM83wH+PoGAncggmfvhdBdReQXjIH2oapOsr2h\nQBuHYSZUWgAXYdjin8Z4hfNYK7A6Hzd7Yb8xQh8oiMgeDBdhjqr+LSj9XowCfAVDfd8Tw+IcIyLX\nUzOrcS6G+fgioLmq/svWF8xOPY2amaePU9XA+Nx9GItagZZawXR9P4ZSawLwnaom2fTmVGFvxvD0\nrVXVal3+WVluU9W5tqt8NeZD1xZD4fUxRnHPwzBqT1JDKVYdk/Wxtq6hwfHgdjAP4xuYLgcY4oaf\nVfUKe++OCRqCqBZZWX11xqy5tRWpF0LFlgIQE7lPvrr+J1DqDd0+hmZR7kYxtzTv1F2Pu+ftSmmf\nXtG33nVWeUY/Aj5R1Q/s8/yLqr5Q2/n1mQVOEZEPbHdpZSDUR7gDBD9moW1/Efm/qpl2PO89DGdd\nALWxGtcXNTFPV/3C1PXFKZ9V02rYm223t1BEkqs/3UBEkjAK6nhV7YVRoJH269gfowiHYngD0X1j\nsgZz774KunfdVTV49m13TSc6cBAgQwjRXuB/ALeISDZmTPDVuk6ozyTIGxjXdgKcgrEQxtZ2woGG\nGtbk0zCepaqbCn8S48EsoKjqw2o8AzhdRCLFrEAfWk9xTrT1RWEGgGdg2KPPEpFoMZTeZ1PBNF0O\nqYa92Wb9B3jedgcQ41XvkiqnN8UonwJrlQb8BscAzVR1MnAzllhSqmeyrg9mAkcFjc80EZEu9TzX\ngQM8Llel0BCo6rRAr0RVV6tqf1VNVdXz1HpNrLXterQRrapTReRxO6Fwl5gZmrsbJOkfDDv2dTLw\ng4hsqZK3VUQ+wygAtB6sxqo6R0QmAL9gfJYspoKhuTbMxmwlbAe8o6pzAcQ4QwqwgL6iqgtEpFOV\nc6tjbwZ4EdMVniMiZVbeJ6pc4yIxBBbLMWvzZtisWGC8iERiPmq32PTqmKyPqeviVHWL7eq/L8Z1\nIRhlfTD3EhwcJBCRkBJGNLj9usYAReQnzIDlp5juUh7wuNbhbenPCKlgaI7G+Cu5WlXnH2i5/tfh\njAH+sTiYxgBbJqfraQ+8Xynt7QszDjwjdBBuxsy43Ag8iJl+vnx/CnUQ42Uxi38jMWOGjvJz4KAR\nMIvWD2I6rKBtKLuoIEU9JKGqFxxoGRw4+DNBJLSciQ1FbV7hPqOWGUtVPWe/SOTAgYNDCgeSD7A2\nC3C/7GF14GB/IRSs0qFGKMfbQnV9oZSpsTjQkyC1eYX75o8UxIEDB4cezBjggftwHXyfTAeHNAKs\nyemWKbkqSkpKypmOg9mXA/j6yylk9epG7/QuPPnY3sTlzz39FP379ODIfr05/ZQTWbe2YpNK8yZh\n5UzIw84+IyQyhUqekcMq0yo29j4FEOY27NjNo91Ehe1tiUV4hBZNDEt2XJS7nHIslPC4Koe6YNfi\nzpYqrNGyD3yAB5pB2AlOIMAIHWBNDjBCB7MvB0J17MuBvO2FpdopKVkXLlulWwqKtEfPXjpr/uJy\n1uOCIp9OnPK1btq2SwuKfPrE08/p2eeeV57XpEmT8nhVJud9kamw2BsyeUIlU1GZljNgb9lVpl6f\nX7cVBjFiBzFkb9lVpjuLvLqnxFcpLTgAcxvz37dJTdfbJi6vFOqqE2M4xth4OWs0ZpPGSJv+EoYw\npNb2620BBi1ydeBgvyDAmhxghA5mXw6gJvZlML4sklNSSEoy559z3gg+nzSh0vmDjhlMdHQ0AP36\nH87GvNr3yzdGpjmzQy9PKO5TAB4X+PwVzNUlXj/h+8HCqw0ChLmkUqgLahDg8gxmjQ49H6AY5yWL\nqaCxyRCRZ+s4zYGDBiPAmhxAgCl5rzLtTZkA+/K2bdvK8xIrnZ/IploUyttvvMaJljUZjNOfY47q\nz/GDjmTC+HGNlimU8kwK4gNs7H0KwCVSibbf0HtV3w2Oi3ITG+nai9i1sQhMglTHBxgUrq7mPLcE\nsUYDOewDH2B9FkI/g9n3Og7Kt1gNrtfVHcQQkUJVbZRXKRFJAJ5R1WE15McBF6hlpKirvC0zDcPc\nUozxq3GVqi5sjJyhhGW0+UFVvz7QsjQGY99/hwXz5zH5q+/K05asWENCYiJr1qzmjFNOoEePngeP\nPCefQO+MDJJTUmqpJfQo9SolXh9g/KnERLjYWRzaWWT33mbYVq1jJ4gaUtTe9h37DEOR32DUpwvs\n0r3pjHz70tifDaq6sTZlBsQBf2lA+QAuVNUM4AXgsUaKCYCIhGRvl6res7+UX4DpOIAAU/JeZdab\nMsGsyYG8vErn59E2cW8j4Ltvv+bxR/7DBx+PIyKiYmQnwZZNSkpm0KBjWbhwQaNkCqU8Awcdw8KF\nC0JynwLwq+7l4tNfpZscfFTs1eq4+xoFETMRExwaAjXsTt8BR2D5AG1WvfgA66MA14tIf0Ct2XkT\nf9KN7mIYmb+11F/fiEgHm54iIjPFsDI/INaXiC2/xMbTpYJR+hdLLPAwkGLTHqtS3i0ij4thYf5F\nRG6oRqSfCTLjxTIvi8h8EflIKnyvnirGP8g8EXlGKrxk3Ssib4vIDOBt2+ZjYnyI/CIi19hybUXk\nByvnEhE52pbdi43apg2z8ePFeOVaLCKvBcaJxbBo/9vKuVhE6vV17tvPsCbnrrHsy5YpORgB9mWA\nT6uwL2f27UdOdja5lr3504/Gcuppp1c6f9HCBdx0/XV88PE4WrUudw9Dfn4+JSWGPGTb1q38/PMM\nunXr3iiZ+vYLnTwzf/6Jbt26h+Q+BeD1m10YASUY4XGVuwsIIPiUcLfs5TWvsRDALVIp1HmOSCtr\n+SGGZelEjB+S74CAgXEpsDfHf1XUNUsCtMZw+W+14QMg/kDPHDY2AIXVpE3EsDGD2e88zsYnAefb\n+LWBc4FOwBIbfxZjuYEhBY0Kzq+m/HWYAVuPPW5hf6cBfW38JuAhG4/HEDA0scf/wLA4R2LYXpJs\n+vsYklOAezHEp1H2+GrgLhuPwJCtJmH8pNxp090YxpgsDM9fQPY4+/uGfcgC7Xax6W8BN9l4LnCD\njf8Fw3ZT5yxwUZnqZxM+19TOnTUpOVnvve8BLSpT/eedd+tHn47XojLV/F1Feva5wzQ5JUWz+vbT\nZStyys8tKPLpR59N1JTUztopKVnvuvd+LSjy6e3/vEvf/+gzLSjy6TGDj9dWrVtrz14Z2rNXhp5y\n2lAtKPLpl9/+qN3Te2iPnr20e3oPffG/r4REplDJ8+yLL1ea5W2MTMGzuDv2eNXr86vX59fCYuNH\neXeJTwv2VMTLvH4t8/q1pMy31ywxjZwFTuzSQx/9LqdSqKtOoBewAMPMtAS4x6YnYxiWsoGPgIi6\n2v/TMULXF9WNAYrIVqCtGlr8MAxjdLyIbAPaqKpXDAffRjVM0p0wyqaHiFwA3IlRBJ+q6qrgfFt/\ncPlPgJdU9asqMkzDjAGGYyivequhqR9KNczLGMX7tKoeY88/A8NSM1QMC7aqamCd1MeYhydAT98M\nw4tYjHEV8A5G6S+UatioVdUvhsZrEmZS7FlVHWTrPh74q6qeI4YJ+igr9wDgQVUNuAgIXOfVGIVM\n+w4dslbm1EoaXS/82XddhEqmULLmtIoNaxRzS/u0nnrrmMoz4zcPSj542GDEOELaS0uq6l4zM4cy\nVPU9EZmFIWKdbLuX9fHbUR0uxFhuj2EU3DlUMC+fH1xQRHrXUVcwI7NgLLOpVQuJcSR1GvCGiDyp\nqm+JSAYwBGP1DqdhLEA1MWQDoKovAy+DocNqQL0O/kQQqp0E+cNQn6a/xhBkfoMh1WxNxcP9Z8NP\nwEgbv5AKluaZVLj5G1n1JAAxFPWrVfUZzNhDLwyDTmwNbX0FXBMYtJUqnurUmOZ3A4fbMbSamJdX\nAMlSQaY6gpoxFbjOWreISBdbT0fgN1Udg/GZkik1s1EHsALoFJAHwxT0fS1tO3CwFwRp8BhgKFEf\nOqxK9Pci8jYwfb9J9MchWkQ2BB0/iXEa9LqI/B1DC3+ZzbsJeEdE7sSQwlbHBD0cuFgMQ/NmzNjd\ndhGZYSc+vgCeDyr/Csbn6S/2nDFUIaBQ1SIReQL4uxonQ6OowrysqitF5C/AFBHZDcyp5ZpfwYxD\nzhczIr4Fs1j0WODvVo5C4BJqZqMOyFYsIpcBH1klPgez+t6Bg3ojMAt8wNpv6BigiKRgxoP+2AVJ\nBzyrnysAACAASURBVBBiGKCLVFVFZCRmQuTMus77oyAVTNWCUbKrVPWpAy1XfREqRuiDcbztYJTp\nYBoD7NStl97z1qRKaVf073hQjQHmUzEG6MK4dbxjfwp1ECILeM4qmB0cfIzYV4nIpZiJkQXAfw+w\nPA4c1Asi/OHd3mDUqgDtC59BxYJCvx6C08aq+iPWe9rBCGvt/c9YfA4cBCCA5wAqwFptaqvsJquq\nz4ZDTvk5cOBgf0JC6Re4wajPoMJCEemz3yVx4MDBIYd92QkSStTmE8SjhlmhD8b/bA5mTZlgjMOq\nyyIcODigWLy+Pm6a64espOYhqSeUNP1rft9dd6F6oEN8dEjqCQVEGt4FFpH2mA0HbTDzEy+r6tN2\nKdlYzEqHXGC4qubXVldtY4CzMWu/zqiljAMHDhw0CtVRcNUBL3Crqs4XkVhgnoh8BYwCvlHVh0Xk\nDsxk7T9qbbuWPAFQ1ZzqQkMldnDI4GTMIulsql8tMAqz/nChDVcGMjwuKC3aRd7aHMa++2YlqneP\nC8LdEOby8+zTT5CelsqLz41G/KU23WzmD6BVbDiZHZuS2bEprWIrmNEFSGkdTWbHpvTp2JSWMWEA\nhHuEHokxZLSPpXeH2EosKd98OYWM9K4kd0igfdtWlWjoPS5QbwmXXDiC9m1b0aJpNH16deeEYweS\nvWIZ4W4Y+/67dO+STNPocKLChKgw4e8330iYCz76cCz9+vSiT89uZKR3Jb55LHExkaR0akev9DR6\ndutCy2YxZPRI4+wTDmf8R+8yatgpZHVuwwN33oLP5+Ock47kukuGsfSXBZx5fH+GHNWLB+++jQfv\nuo2szm145tH7OOuEAQw9NovDu7ejX5+eDMjKYOoXkwF46YXniIlwkZbakYGH92Xad98ChlMwLiaC\njgmtOKJfHzoktOKs00+lf2YvuiR3oHlMBEC6JdGYYgk5fhGRlSLyqw3PiciblhTjVzGEIStt2XNr\n6AKX/2EicqmIrLLhUptcCLwmhg/wR6Aj8ABwJoYIFepJiFrbhuMNwC01hcZsgHbCnza41Xwgk1U1\nXFUXqWr3KmVGqepzwWmZmVlaXKbq9fk1JTVVl63I0TKvTy+77HKdv2iplvlUy3wVVO833XyLFpWp\nTvnqGz3/wou0qEy11Ku6ZWeJTl+5XWdm52tRqVdnZufrzzb+c3a+Tl+5Xddu3aPrtu3R6Su3l5ed\nvnK7btpRrNm/Fer0ldt1Xu4O9fst8UCJV5OTk3XxspXaKSlJu3bt+v/tnXd4lFX2xz8nhYQWem/S\npHew/+wVEQtFXV17211XV3ctu7rq6rq2dXXtvWNFUMGCBSsWRDpKL1KUIi2BkEK+vz/unTAJAWYm\nWTOQ+3me95l578x73jvJ5OTee+75Hn0zaYp69OipaTNc3yLy8yt/2aBnX3hJQ4cN16jRb+qoo4/Z\nJmHftp1atmqlN954U+nVqqljx701/tMJatmqlX5csUr/vf9BNWrUWMcNPF4jRryk/v0HaMiw4fpl\nwya9/NooHXX0Mfp08nzVq99Az4x8Vzfcdq9+c85FuvqG23T8ScN0yBHHqkfvfnrprfGatSxbvfrt\no30PPETVa9TUxNkr9P3yHA074xwdM+hknXfBRfp26ky1btNGGzYXqG+//jr4kMP0/Euv6pvJ09Ws\neXPl5BXpd3/4o4aderouuuT3yskrUq/efVSnTh0tXr5Kjzz+lHr16i1cXjzA0bgZ5QHed9yJE9aY\nB4z377kVt41sL9zgq2GHrr309syVJQ7cUtsk3JauPFwiQj1camm96O+Ot5WP+8e7Pqrdos93dOxs\nBJiKS8avvYMjECjNPriR30Lcl/Jl3H/lmNi8aROdOnWmbbt2pKSkMGCffRk75k1SzUk3gZN6P/Fk\np3h04EEH8/577yI5WffIelvdGmms31xIYZHYWiTWby6kXg232tMkK4Nla7cU37MwShI5Up4xLcWK\nN75+++1E2rXvwJo1a+jQoSO/Petsxr33rpOhf+tNCou2yc9nZWVxypChfDz+I3I25RTLT307cSIN\nGjakc+cufP311+y//wG02WsvXnlpBB06dKRRo0aMeetNatWqRXq1apw8ZCiLFi3k448+pHr16mzZ\nsgUzIy9vCykpKfQdsD8ZGZls3pTDpx+9x5DTzyY/L4+c7I306rcPRUVF5GzcSKMmTQGoVTvLfxoj\nJycbM2Pjhg00a9acRx68nzN+ezatWrcGoGvXbmzJzeWbr79i1apVHHHkUQDMmzuX1atXk5qWxuZN\nm5BEfkE+/veMpPflYgbCOa3WOMWhVNyOujS2jf43SiqStCayD7DUCFByG6HvBJ6RdK/cWt4HOEfn\nPo2TgxsLbMSleRYj5wV3uWtlZ2uAP0m6eVcGAoEoWuAksiIsA/Yt431DgINxupJXgPumLlu+gtde\nfYWMVNgqyMisztq1awE31Uwx+Mufr6RNmzauLUrqvWnjhqzbVAA4R5hXsC0Dw9W6SCl2cK0bVKdO\n9TS2FBSxcPVmCraKH3/JpVuL2jSrk0lqChR4yd/ly5fTulWrYhn6Vi1b8tU33zBgwL589+03pKXA\nTyuW07aNk59/4rFH2bBhA9defRXjP/6E9BQ3laxVqxYtW7bi1Vdf4bzzL+Dj8ePJz89j7tw5LFm8\nmOXLl7Nq1UqyN2aTUc19rvXr1tG5YztyN28iK6sOJx6xL3fc9zhpae7P9ruJX/GfR55jU042eXlb\naNLMSUe++PQj7HvQISz/cZvCzr2338QnH7zLhvXrWDB7Fq+/9gpPPTuCu/99B+++P57fXXQ+AG+M\nfp1evftw49//xpNPP8/H45327cjXXmbosOEM2Hc/9u3XEzMjNzcXoLFXGbpC0lJJX3nHfwLwE07U\nty0uPbQ+TlPgQx9UvbRjt15lrQFGolllfZ9aAPh89tf96+9JkpmtNLNmkn4ys2Y4ufydsss1wMrA\nzLbaNnHOMRHxwwqwWyxIWgG2njGzRb6fU83ssoqwu4N7HWpmB5RqO8u2iZVOMbO/RPUrFtXpWO7b\n3H+5I+cv+TWeK8zsZjM7cmfX74AxuGlLT9x/9GfBfdmysmpz5Z+vIm+rc3ZNGjcmJSUFM1evIn8r\nTJs2jQb165QwWCMzgxSD5eu3sDMMyEhPIXtLIdOWZpO9pZC9GlYH3Jrhqo15TFq8ge9X5JCe6q4p\nknPOEQesqDFFpF/y70tPgUt+/wdatW7NtdfdwC23/JMiQaTO0OrVq6hevQbNmrdEguqZmdz/wMOc\n+ZtTWbxoIenp6aSlpRTX6aidlcWEr7/lzrvvZb/9D+DVdz7l8QfuJm/LFmbPmk5mZnW69Sy5Q23V\nzz8xbuwbHDWw5MD7T9fexLmXXM6+Bx7COedfwOtvvs3ZZ57GP275Fym+Lu+ypUu54W/Xst8BB3LM\nMcfRomXL4utHvvoKJw8ZxhOPPsKEbyYz/Yf5nHXOeQCRkZn7Pbrc9Zq4CG0LnCOsD3T3v4KuOD3J\nr4B/u7rAVuLAOcsd/x6dh30SJ4LaBqeBCfAWTggVYhRE3dkI8IhdXfw/JFdSbwAzexb4A279INm4\nStLIXb+tJGaWKlfTIFYOxS38fumvPw63LnK0pBVeHOGsePuxKyStwCvsmllTYICkDju5ZDnQKuq8\nWJY8altVdGWeJ3DTHFIM8vMLWLBgPgBbiyAjI526desiUewUpk6bXrxXrLCwkN69elG3Ti3yt25z\nTvmFRdSpkV58k4y0FDZsLiieEv+S40aKa3LyaZLlVnOaZGUwa0U2ANlbtv1qmjdvwdKlS2nctAU/\nLl3K8uXLnNS9fyySe8+SH5fSpnXLYvn5886/gOv/ejVbBS1btiAnJ4cFC+ZzwYUXs3z5MlJSU2je\nogWDTjiBgYNO4ISBxzBv7lwaNmzElvxCNmzYABINGjRg+Kmncfmlv+O6Ox6iRo2azJvzPUsWLWDZ\n0iUcuW9X8vK2kJO9kbS0dH6YOY0lixdw5cW/JS8vjy25mznmwJ6MmzCd119+llvufohbrvkj19/w\nD3Jzczn7zNNITU1l9apVvPTiC9xw4818//0svpzwOY8/9jA5OTlsyc2lWkZGsaOM1CUZdurpPPn4\nozUjv0cv1jEUty0lopq+CZc+uxKnQzkO6I8TLD0/EgQp/dWL+j4dWur79AlwIE59aB5uqv2kmf0N\np8D+qpmdDyzBCZTslB2OACWt3dXFvxLFsvBmVsucVH1Eav1E376XjzA9bq5Q8vvmpLIxs37mCihP\nwzlSfHummT0dNYI6zLefY2ZvmNkH5qTdLzWzK/17vrZSslWlMbPTvc2ZZnZHVHuOmd3t+7G/79en\n5mTsx/khO2Z2mZl970daL5uTuboEuMKPNP8Pp8zyF++gkJQnJ2VVui83mJO/n2lmj/n/nNvdw7cd\nEjWanWJmtUuNmN8HWkT6YCWl8fuZ2afp6ekPrFix4v8ee+yxfYBq8+bNu+7II4/sYmaTgMu9nWZR\nXRyM+y+OgFatWrJo0UIWL1qEUcSbb7zB8YMG+2pl7oJzzjmXH2b/AMAn4z/koYceomBryT+g9ZsL\nqVsjrbjKWGRNEGDtpgLqVHf/9+tWT2dzvnN2eYVF1K3unGb19JTi6U9Efr5hw4bMnzeX559/jqOP\nOY7XXnmZQScMJsWc/PyLLzzL3LnziuXn33v3HTp06EiKORurV69i+bJl9O+/D6++/BJLFi9m8ODB\nrFzpZmlHHHEkq1evorCgkFGvj6RFi5YcevgRLFm8mLFj3qJDh44sX/YjCxfMpUWr1hwz6GROHn4G\nH37zPXc/9Az7HXgo7TrsTd169flsygK69erL3Q8/S2b1Gjz6/Cj3g2/RihefeoS9O3Vm9g8/0LBR\nI2bPX8KXE6eQkZnJpZf9ib9c81eeevYFZs9fwvdzF/Gv2++i496duOT3l9K8eQtmz/6e1atX8/NP\nPzH+ow/AiekOBlYAVwO3AAeYWZqfpjbAKcgLJ657ODAbN8D6HjNSSh1RjAOONrN65gR6jwbGSfpC\nkgGjgDsk9Zb0jqRfJB0hqaOkI2PyYUkQOdzuYJvkfCruP8Wx/jwNyPLPG+IW3A03pSrEqSeDK5B8\npn8+HTjYP7+LbZL0fwae8s87Az/iZN7P8XZrA41w6xGX+PfdwzbZ92eARWzbztEDaO7tNPJ9HQ+c\n5N8v3MZMcHVMvwQa+fNTo/qyAi/lzTYZ+ptwDi/y81kL1NnBz+4ZYKh/Xj+q/XnghJ3cYwxOxRlc\n8CuNkhL+xc+j71P6s9x55523rVixYoOkBf/+978XAg9JulnSYH/tbZJmrVq1aunXX3+9sXfv3jNb\ntW6t3AIX6d24MVvz5s3TE088USz1fsdd/9aq1Wu0tUjKLyjUhRdfonbt2+vrb75RQUGBthZJW4uk\nX7LziqO7c3/O0ea8Qm3OK9Tcn3OK279duF7rN+crZ0uB1m3K18SF64ojvxs2FyhnS4GytxQor9BF\ngbcUSGPGvq2OHTuqadOmqt+ggdq2a6eDDzlUr416U9df/3e9+uprGjJkqLLq1FFmZqY6d+6iQw89\nVNNnzFThVmfjlltvU3p6utLT01WvXj3dcss/9bfr/q6WrVqpRcuW6tSpk/oPGKCaNWsqIyNDDRo0\nUPsOHdWqVSvVqFFDe+/dSV2699J9T76k5i1bK6tuPVWvUVNNmjbXrfc8okOOOFavvvOZOnTqolZt\n2uo351ykWcuyVb1GTR01cLA6dOqiNu3aK6tOXXXq3EU9evbSm2PfU05ekf5+081KTU1T6zZtiqX5\nFy79WTl5RXrk8adUu3aWvpv2vXLyinTv/Q9p706d1bBhI9WqVUtALq4WxxLcetxUXOmMdcD3OEHf\n14BZuBHbQv83+RHQunP33vpq3roSByWjuef5v8f5wLmlvusLgc7l8TVJKYlvZluBGbiR3w/AYZK2\n+v8o9+AW0IuATrgF1kycWnJHf/01uD/MB4DpkiLFjXoCL8pJ0o/GSbqP9699jhsh9sU5ggt9+4/A\n/nLy7ucBPSX9ybw0fPQU2I9Ih0g6y5+fD3STdKWZFeKczlYz645zGhHF6FRc0OloM3sPN919AydP\nn2NO2j5H0r+93bW4GiDbpT5E98vMhuD+K9fArcPcL7dJtKx7XAucDIzASfovs5IS/sXPo++D+2++\no8/yCXCjpJ0KpVaUHNZ3i3a66T8uKioTpCJJxkyQWhkp5ZKu6tKjj55545MSbft1qPuryWFVohj1\nTomsAbbBjfAiU9czcKOrfv71lTjnByVVqsuUYY+DaFtFUedF5bC7RdvW/QyYJTd07y2ph6Sj/WvH\n4zT9+uJSEMu63yycRNcOMbNMXARuqKQeOMHVyM9qu3tIuh23Kbk6MMFirOS2i88CJSX5A4HtSLGS\nx69671/3dvEhaTNwGfBn7wjqAKvkihYdhnOQO7t+PbDezA7yTWdEvfx55NyctHxrXAZDeZgIHGJm\nDc0sFTidsmXi5wCNzGx/f/90c2U1U4BWkj7GpfDUwU1HS0vr3wbc5QMTmFk1M7uAkkSc3Rpz+6Ui\n63Vl3sPM2kuaIekOnLpzrA6wzM8S47WBKo4BZlbi+DWpkGLZ/0skTTGz6ThnMgIYY2YzcDvFZ8dg\n4lxc2oxwC/kRHgIe9rYKgXMk5ZXnFyC3/+ha3JqIAW9L2i4ULynfBxDuM7M6uN/Dvbh9cS/4NgPu\nk7TezMYAI/0U+4+S3jGzJrj9VIZbX3yq1D3WmytoNRO3rSAilZ+6g3vc4v+pFOFGmO9SMmCxo8+8\no88yK/afXKDKYpBSicOwpFwDDFQtwhpgbOyJa4Bde/bVi2NLTpL6tMlKHkn8QCAQ+F/hpsCVd//g\nAAOBQKVSmTVBkjoIEggE9nAs/iCImT1lZquiNuljZvXNJS/M848xrWGEEWBgj6GgqOJKUG4tSr61\n8Ypau/t+2cYKsVMRGAltfXkGt8f3uai2a4lTDBXCCDAQCFQy8RZFkvQZLhsqmvjFUL2xcISjUo++\nffspt0B6c+y7at6ihdLT01W/fn3dfOttTpTUH+tztujkU4aqZs1aysjIUI+evTR73iKNffd99enT\nV207dlHHrj114ZXXq9Ve7dWiVRs1adpUPXr2UpduPZRVt76GnHWxxk1frpZ7tVN6ejVVr15Drdvs\npTp16ujJl99UyzbtVC0jQ5nVq6ttu3ZKT09Xk6bNVK1aNaWnpyszM1OpqamqUaOGBp1woi75/R+U\nmZkpMys+UlJS9Le/36iMjIziNkCZmZm6+ZZ/6cQTT9Ts2bM1b948XXPNNUpNTVWPnr10wIEHKT09\nvfiaU045RZL00KOPKTMzUw0aNND48eOVnZ2thx56SL+/9DI1aNBQqampqlatmh599FHNmTNHc+bM\n0b///Z/ivg4fPlyzZs3SkiVLtHL1GlXLyFS1jEy179BR48aN08KFi7Q5r1D7H3hw8Wv/+te/NGfO\nHG3OK9S74yeoXv2GqlYtQ0cceZSmTJmidRtztC57i3ACB1MbNGgwY/To0XkrVqxYK+mH4cOHv4xP\nE7366qtXzZkzZ6ukWZLuxAlgzOrfv//Cbj37aM7Pm0ocwJK2bduu7tGjx5YePXrkDhky5GH/XblH\n0lRJU+fOnbuwS5cuWyPfIRIQQ5UUHGA4Kv/o27dfCdXkqTN+UPfuPdSx496aPG1WsQO8974HdcCB\nB+mCCy/Wsy+8pH323U9Dhg3XVxMna8GS5fp49ho9MfoTpaSkasQHk7Q+e7M6dO6mZ8ZOUHZugTp1\n76V7n39Ll99wp0449Wz9sCJbDz7+rDp06aFjT/mNHhs1Xq99OkMfz16jbyZPV+MmTZSSkqKJU2Zo\nbfYWde/RU99OnaneffrqwIMO1mNPPqP//PcBnXfBRcrJK9IZvz1bjRo31sGHHqYJ33yneYuWKSev\nSB98/LnMTI0aN9b3cxdpa1GRNuUVKSevSDNmzNDQYcN13Q03Kce35eQV6Xe/v1RffvmlJk+erM35\n29pXrFqrSy65RC+88ILe+/CT4va8wiI98MCD6t69p84651zl5BXpw0++0NBhw/XzypVq2rSp9tl3\nP3348eeauGC9Ji5Yrw2b8nXMMcfo1vue1qSF6zVpoWtfuHKTxr7zrjp366WJC9brhTc/Vs+++2ri\n/LVat3GT2rXvoIdHjNHkRRsETPK/x2evuOKKxbi8+2qS6vr2w+bPnz+7Zs2az0hi4MCBxwMT7rvv\nvmpFRUXTu/fqo3krN5c4WrVqtULSP/31KZIalv7OjB8//saWLVuui5yXdnjAuli+e2EKHEgKolWT\nO3XuzPDTTqfNXnsxdsy2feRjx7xJfn4+Z/z2bE4ZMpT58+fx8Ucf0qt3b5o3bw5Abu5mLMVo17oV\nWy2VQ487mS8+epdvps5kw9o19Oy/PxM+epdjTjoNgONPPIVFc3/g8IEn07FrTxo2cXu/u3btxob1\n62nYqDFdu3ajWrVqDB1+Ks8+/SQrV65k5oxpDBp8Em+PeYszfusk6ObNncO6tWsZOuxUevXuQzPf\np+9nzSQ1NZW2bdux115tijUEp0yZzIsvvkif3r0YNvz0Ej+Pnj268/LLr5BVp6T24T9vvglLSaGg\noIADD/q/4vZUxDXXXA2IocPcZzMzjj9+IPfddz/duvekoKCAuvWcmFFmegrr1q5m/PiPOXLgSRRF\nSY41yqrGX668kmFnXQTAlvwC8vO2UK9GCp9+9jkbNmZTv2HjaDXtOvn5+Yffe++9qbgMq3yc9D3A\n76666qqtmzZtGgHwzjvvrAUyTzvttIEFBQWzYHs1mPr16zfEZTuB25i/pvT3pVmzZoM3bNiwPqpp\npW1TVIpJDBXCGmAgSYhWTQZo0aIlRVuLWL58eYn3bNiwnpatWpGWlkadOnWoVbs2v/yyTWLws/fH\nUrd+Q2rXrE5eQRGNmjZnzcqfeOP11zjx5KGYGWtW/URjr56ct3E1IAYeuk+JokpvjH6d1NQ0unfv\nUdzWokVLvvryC7p378Ghhx9JVlZWsVL0j0uWsHjxIoqKijj4kMNKfLZHHnqA2rVrM/y03/iasq79\nrttuZdOmTbRv34EOHTsWv//HJYtp1Kgho0ePplGjxiVsvT7yVRo1akS79h1KREyzszdy++23M+LF\nERx7jNOp3Xe//enTuw91smpzy803MXr0KLp17QI4B7hw/nzGvf8B3VrWpmX9zGJbqSpk6NAhXH3Z\nRXRsWoMB++xHv/3+j3GjR7B8yQI++fgTjj+kX3FBKaDtzz//bB9//PEGSZNx+oA1AfLy8rrts88+\nrYuKiv4JfCqnCfnxHXfc8dLIkSOHRIIgZeQC3wJMxinJNCn1dWmTnp7eKjs7O3pneNxiqBAcYGAX\nmNl15jQWp3stwBvN7LZS7+ltZj/457XM7FEzW2BO6/ATMytLFr/CWTRvNh+OeY3ufbe/3Ttvvs6J\nQ0rqY67Jyefuh58hs3oNsrcU0blprWI7f7vmKjBKqCIDLJg3n40bNzDs1NNKtI987WV69OxFRmYm\ndett24Hx2aefMGf2D2wtKuLkIcNKXPPRh+9Tp149OnToWKK9qCCX22+/g9pZWdSsWbO4/ccfl7Bq\n5UpyN+fSoePeJa6pW7cui5cs4aGHH6VIRkYqLJg/n7Vrf6FLly6069Sdyy//E2zNJzUFirYW0rdP\nL378JZfvl+eQkZ5Cw9rOoUlF1K3fiB9WbGLNxnyaZxmLF8zl4COOY/8DDuC8Cy9i5Dsf07xeBj16\n9MgA0po3b95ixowZd+HqiG/CVwRcv359g27dui00s/2Aq/Lz80eZWZcbbrjhltNOO20FbL8NJi0t\nLSKx1hevHB39Wfv16/dm3759awKdzGyZV126HTjKzOYBR/rzXRIcYGCHeIGDQUBfST1xX6yPcfqF\n0ZzGNlnyJ3ARuo6S+uFysRvu6l7NmzvV5GXLXAmIiGpyixYtSrynTp26LFu6lMJCp5qck51NgwYN\nWLZsGTdcehbn/PEaNmVvJK+wiIz0FFb/vAJLSaFo61Y6de8FQMPGzVj103IKi8RHb48CjNzUWtTO\nTGX1zyu44dKzGHTCYA497HBWRI1Av5v0LWZuqnvscccX92nZsqWMfPUVNm/eTGpKCg0aNHCfYdky\nzvrNqfTu04/+/fehSZMmCJf58Nmnn5Cbm0u1tDRat21b4mfRtEkT3nzzDSZ88TkpBhlpbmR0+623\n0LBhI8ygUaNGxe9f+fPPbNq0ibFj32bYqaeztci9f8ybo1m8eDFLflxKrVq16N6zFz/9/BOZ6am8\nN/YNpk+fTtc++wGwflMBNaqlUlhYyI8//kjtxq5I0rrNhdTMTKN77/7MnDGNFavW0Wffg5j09Zdk\n526lf//+Nc4888ysn3/+ufDSSy+NRGFH4pwXc+bMqfHTTz89jpv1T8zJyam+9957T8/KylqYkpLy\nKQapKSUPSUU4sVNwI8C+0T+f7777Ths2bDhKUrqklpKeVCJiqAQHGNg5zYA1kvIAJK2R24KwrtSo\nbjjwkpm1xxVBut5/iZG0SNLbu7pR/wEDWLN6NT/88D1z58wpVk0+ftDg4vccP2gw6enpjHj+WUa9\nPpL27Ttw6OFHsGHDBk4ZfDwX/vkGjh96JsuXLGTu/AWkaiufvDuavNzNnHbaaazxUvgHHH4s4954\nmZ8Wz2PNyp/of9ChNKqdwYpVv3Dtxadz4Z9vYNK3E7nk939kwfx5LF60iPz8fEa+9gpdu3bn2IGD\nyMx0U8aBg07gofv/y9q1vzBtymQOP+IozIz169cz5KRB1Klbl5q1ahaPGCPq1iOee5p27Ttw2mmn\nUbP2tnW+7yZ9S6PGjenTty/Lfv6FIkFeobtu7Ftv0KbNXvTfp+QI987bb2XcuPfp07sX++63vxvh\nAXXr1+eVV17h1FNPpaCggBlTp9KkcRO2FBRx392306x5i0gNDmr7IlFjR45gzJgxHHmUUzSrnEGc\nMwAAIABJREFUnZnKug3ZfPfNF9x/738Y0L8f0yZ9TcdOXaiZmcqMGTNyR4wYcXReXt4KnD4neLVn\nM+s8atSoggsvvDAyj987IyODuXPnDpg+ffqHRUVFPcpSg9m4ceMGtknhO+XobXTGlcj8alffqZio\n7AhgOJL3wElxTcWp1DwEHOLb/wLc45/vh48E4qTRR8do+yKcos+kiCL06LfeVrPmzZXmVZNvuvmf\n+ut1f9fQYcP12qg3tS47VyeedHKxanL3Hj30/ZwFuvEft6hGjRpq37m72nfurqYt26hZqzZq2Xov\n3fiPm7XXXm11xjkX6p8PPa9Fqzfp23krdcgxg1Wnbj01adpU07+fo7Wb8nXxlX9TZvUaatN+b6Wl\npalHz1568tkX1KFDR7Vt205169ZT3/4DNOzU0/XKyDeUk1ekNRs2q3OXLqpeo4bq1auvGT/ML1ZZ\nrl69utLS0pSamqqu3boXqyzn5hdp7ty5Wrp0mSZ8+XVxFDc3v0jHnzBYrVq11oB99lVOXpEKtxZp\nc36Rvp06U2amH3/8UQUFhSoqKiqOJjdu0kQDB52g+QsWaOtWd82mvCL96cqrlJGRoaeeflrz5s3X\n8uXLlVtQpE9nLJelpGjcF1O0aUuhNuUVavXGPH27YL06dumuI447Qes25WtTXqGycws0bfF6HXTY\nMW57zO13aOnyFdqUV6glqzfL/w4XPvLIIydJmiRpuqQ3JNUDbqpZs+adkl6QNFPS5M2bNx8JPAr8\ncNFFFy3v2buvVqzPK3FkZGRMl/SZt/WRpNZR352bJN1eUd/xoAYT2Cle1/D/gMOAi/E77nFrNG2A\n/wBLJd1tZoNxsuUnx3OPilKD+XrBL7t+U4wMaLvT0i+7NRWZCbJP+/KpN/fq00/vf/p1ibamdaoF\nNZhAciCnYv0J8InXTjxb0jNmtgg4BFfjd3//9llAL4u/6l2gimJGWXWBfzXCGmBgh5hZJzOLDlH2\nxhW/ARf0uAdYKGkZgKQFuCnRP7xQa6Ri3/G/YrcDuxlmJY9fk+AAAzujFvCs+RKauKLWN/nXXgO6\nsS36G+EC3L6t+V6t4xli3JQaqHoYVly+NHL8moQpcGCHSPoOOGAHr63BVd4r3b4RuPB/3LXAHkRl\nCqKGEWAgEKg8/BrgDgqj7/gys2PNbI6ZzffyVwkRHGAgEKg0XCpcfA7Q70x4EDgOtyxzupl1TeT+\nwQEGAoFKJYG6wPsA8yUtlJQPvIzTA4z/3olcFAgEAhVFAnWBWwBLo86X+ba4CUGQQKUzefJ3a6qn\n25Jdv5OGlCGNlADJZqcibf3adtqU5yZTJn83rkY1K50rnmlm0TvjH5P0WHnusyOCAwxUOpIa7fpd\nYGaTKiJDINnsJGOfKvKz7QxJxyZw2XKgVdR5S98WN2EKHAgEdje+BTqaWVszq4ZTI3orEUNhBBgI\nBHYrJBWa2aXAOCAVeErSrERsBQcY2J2oqHWgZLNTkbaSzc7/BEnvAO+U105QgwkEAlWWsAYYCASq\nLMEBBgKBKktwgIFAoMoSHGAgaTFHswqwk2JmZaraJGBn+K7f+ethZqlmdkU5bQwws+PKaB9oZv3K\nYzvZCUGQQFJjZjMlda8AO1Mk9akAOxW5+Xlv4CpcNkXxjgxJh8dpZ6KkfcrRj/G4UgZLSrW3AZ6O\ntz+7E2EbTCDZmWpmfSRNKaedj8xsCDBK5fuv/6GZ/QV4BVf/FgDFWIaxFK8BjwCPA+UpITDBzB4o\no0+TY7y+dmnn569fYrZdmtoeRRgBBpIaM5uFK7e4APfHbYAk9d3phdvbyQZq4hxNbpSdrDjtLCqj\nWZLaxWPH2/pOrnZyuTCzj3fQp5hGbmY2X1KHeF/bEwgOMJDU+FrD2+Hrj+zWmNlNuHIBo4G8SHuC\no8ny9OMR4BdcPWf5NgP+ATSVdNGv2Z9fk+AAA0mPme0H7C3pOTNrANSU9GOcNgw4A2gr6RYzawU0\nkzQxTjs1gCuB1pIu8kWjOkkaG48db6tCRpNm1gT4F9Bc0nFeHHR/SU/GeH1N4ElgAK4ONEAvXIGr\nCyTlxNOf3YngAANJjZldDxwItJe0t5m1AF6RdFCcdh4GioDDJXUxs3rA+5IGxGnnFeA74CxJ3b1D\n/FJS73jsVCRm9i7wNHCdpF5mlgZMkdQjTjvtcIWuAGZJWljBXU06wjaYQLIzFBiIX9yXtByIa93O\ns6+kPwBbvJ11QLUE7LSXdCdQ4O1sxq0nxo2ZpZvZZWY20h+Xmtl2haZioKGkV3EOHkmFxBFU8VX/\nrsMNiMb4Y493fhAcYCD5yfPrUpG1qRoJ2inwtSQidhrhHUac5JtZ9Sg77Ylav4uTh4F+wEP+6Ofb\n4mWTXxqI9Gk/YEMc15+OK4H6vplNNLMrzKx5Av3Y7QjbYALJzigzexCoY2bnAucDTyVg5z5csKGx\nmd2KG1len4CdG4H3gFZmNgI3PT8nATsAAyT1ijofb2bTErBzJU4Pr72ZTQAa4T5fTEiaBkwD/uqd\n56nA12a2AHhR0uMJ9Gm3IKwBBpIen6VwNG6qOU7Suwna6Qwc4e18JOmHBO00APbzdr72NZITsTMZ\nGBaJaPs1uJHxbvHx16bhtgsZMEdSQSJ9irJ3KHAP0FVSRnlsJTPBAQb2aMwsS9JGM6tf1uuxbjkx\ns86SZptZmc4pjk3H0TaPwAUvFuIcVxtcRkZZ+/rKuv5wSePN7JQd9GlUnP0ZgJsODwEW4aqtvSbp\nl3js7E6EKXAgKTGzTyUdYmbr8GtbkZdwW0XKdGhl8CIwCBe53c4OEOuWkyuBi4C7y3hNQNzpYpI+\nimyj8U1zJMWznngwMB44YQd9iskBmtm/gOHAOpzTO1DSsjj6sdsSHGAgWTnXP5Y3Fet2/9hF0pZy\n2PnAP55f3gjpTkZuHcwsnpHbOv/4pKQvytGlLbiR5+e+f2f5tMElwE2/9sbsX5MQBQ4kK6/5x3cl\nbS19xGHnv/7xy3L256/+cWQ57QAc4h9PKOMYFIedyD+J+8rZn5OAWQBmdjDun8ZzuEhyUkvjl5ew\nBhhISsxsKm76+kfgrtKvS4rpj97Mvgam4/7IXy7DzmUx2vkAN60cAHxehp3BsdipSMzsJaA/0ByX\nK138kuuSesZoZ2pkI7ePuK+WdFPp1/ZEwhQ4kKycDpyC+47GVDd4BwwCjgSOwa0DJsrxQF/gecpe\nB4wbM7scFwTJxinC9AWulfR+LNdLOt3MmuKqo5XHAaeZWZrfQH0Ebq2z+LVy2E16wggwkNSY2QmS\nxlSAnV5+v1t57TSStLq8drytaT517RjgEty+xOcT2QZTzn5ch8u2WQO0BvpKkpl1AJ6VdOCv2Z9f\nkz3auwd2X8zsdEkvAe3MbLtpahxT4Kt96toFZrbdf/s4psD3SvoT8NQO7CQyAouk0A0EnpM0y4s2\nxHax2auShpvZDMqOlMc0BZZ0q5l9BDTD5UdHbKXgliD2WIIDDCQr9fxjeaPAkc3Ok8pp53n/+O9y\n2onmOzN7H2iLy8KoTXzpeZf7x3gCJ2Ui6esy2uaW126yE6bAgSqHmaUAtSRtLKedekArSdPL0Y/e\nwEJJ6/1m7Zbx2vNyVrmSirzMfmdc9Lxc2SBVgbANJpDUmNltZpZlZmlmNs7MVprZbxKw86K3UxOY\nCXxvZlclYOcTb6c+MBl43Mz+E68dz/64zc/rzexM3BpgPCIGET4DMr1U2PvAb4FnEuxTlSI4wECy\nc5wfqQ0CfgK6ANckYKert3MS8C5u2vnbBOzU8XZOwa3b7YuLMifCw8BmM+sF/Bm3leW5BOyYl+U6\nBXhI0jC26foFdkJwgIFkJ7JOPRB41WclJLJuk+619k4C3vLTw0TspJkr1TkciFsFuhSFPuBwIvCA\npAeB2gnYMTPbH6d4/bZvSy1n36oEwQEGkp13zWwmsC/wgbkqZYno7z0KLMYVRvrMXMnHRNYAb8bt\nu5sv6Vuv4DIvATsA2Wb2V9xI9G2/JpiIIOqfcJkqo30kuR0Qk6BCVScEQQJJj5k1BtZKKvRreHW9\nMnR57UY2/1YKfhPzb4BvJX1uZq2BQyUlMg2O2KyQAE9VIYwAA0mNFwzI9c7vWlzmRNyZIWZ2uQ9e\nmJk96bX44lZwMbM7vZ10M/vIzFb7AEbcSPoZeB2I6O2twYm2xtunCgnwVEWCAwwkOzdJyjazA3Dr\ngCNwxcTj5Tw/Kjoat8fwt2xTiomHo6OCMouBDkBCzsbMLsSJKzzqm1oAbyRgqqICPFWO4AADyU5E\n+WUQ8KikN9k2YoqH6KyL5yXNimqLh0hQ5nicWGgi21Yi/AEnqb8RQNI8oHECdioqwFPlCA4wkOz8\n5BVKTgXeMbNqJPa9jWRdDATGJZB1EWGsmc3GFTD6yFxxpUR1BvMk5UdOvKx9Io6rogI8VY4QBAkk\nNWZWC+e0pntJ+uZAr3jrgpSRddEAaJFIFoffBL1B0lZzVeqy/HpevHbuBNYDZ+Fybn8PfC/punht\nlWG7UgM8uwvBAQZ2C7zTyYycS1qRgI16QMdSdj5LwE53oGspO3FHbr1TPp+ogk/AE0rgj9LMjsdt\nfo7u083x2qlqBDGEQFLj/7DvAVoCv+DEP+fh8l3jsXMBTjygJTAVV9XtK+KMBJvZjcChOAf4DnAc\n8AVxZnCYq1H8nKQzcFqACWNmjwA1gMOAJ3AlMSeWx2ZVIawBBpKdW3GBgjmSWgHHUoYicwxcjlNz\nXiLpMKAPbvoZL0NxoqE/SzoX6AXUideIl/Vv49c0y8sBks4C1kn6By7HeO8KsLvHE0aAgWSnUNJq\nM0sxM5P0gZklIkm1RdIWM8PMMvx6YqddX7YdEdWVQjPLAlYBrRKwA64c5gQzewvYFGmUFK+4Qq5/\n3OzXSH/BafsFdkFwgIFkZ4MPhHwBPGdmq9j2Bx8Py8ysLm6f3Qfmym0uScDOJG/ncZzEfg5uKp0I\nC/yRQmI5wBHG+j7dhVOoEW4qHNgFIQgSSGr8dpVcXJDgLNx08/nyyNKb2SHeznvR21ASsLMXLgKc\nkB7g/wIzywAyy7k/scoQHGBgj8ZHj3dIrDVvzWyndTokTY6nX97mGLbf97cBp1796K7qGNv2dYVL\n9ynW+sJVluAAA0mJn6KW9eWM1LvYqWOLsrPI24nO+oicS1K7GO3sTF1FkhLJK/4vLq/5Jd90Km4D\ns3Ajy52ms5nZ07vo03nx9qmqERxgICnx20R2SJzF0ZMSM/tW0oCy2sxslqQgavo/JmyDCSQrvYEj\nJW2NPnDqyzFVOwMws2PMbGgZ7UPM7Kg47JxpZtuNyMzst4lI9HtqeQmsiK3WQC1/usu1STO70szO\nL6P9fDP7U4J9qlKEEWAgKTFXpvECSYtKte8FPCnpiBjtTABOKh008cKqYyTtH6Odb4AjJOWUaq8J\nfCapXyx2Sl07EKdsswA3JW+LS4f7BLhQ0r27uP47YL/SxY/83sJJsZbFrMqEbTCBZCWrtPMDkLTY\nCxDESkZZEWNJa7zzipX00s7P29nklVjiRtI7ZtaRbVktc6ICHzt1fp60siq/Sco3i72+cFUmTIED\nyUq9nbxWIw47WV5lpQTeaVWPw071shym36aTUDaHF1K4CrhU0jSglZnFU+M3xcyalGF3u7ZA2QQH\nGEhWxpvZP0o3mtkNuClirIzCla4sdl5+Y/Uj/rVYeRIY6aWmInb2Al72ryXC07i1vsg0fDnwzziu\nvwtXS+QQM6vtj0NxxZoqsoD7HktYAwwkJX5k9RQu13aKb+4NzADOlZQdo500nFO5AJf5YbjUtSeB\nv8dTPNzMLsEVH4oEKnKA2yU9HKuNUvYmSepvZlMk9fFt0yT1isPGccC1QHfc9plZvk9xyYVVVYID\nDCQ1ZrY322rczpI0N0E71XHy9eAquiWSThexVRsgVie8Eztf4oQVJkjqa2btgZck7ROnnYMkfVGq\n7UBJE8rTv6pAmAIHkhrv8DJxdS/mmlkrM4s74oqTreroj+PM7BQzO8JcxbmYMbPLcaPIHDN7wswm\nm9nRCfQH4EbgPdza3wjgI+DqBOzcV0bb/Qn2qUoRRoCBpMbMHsDVyj1YUhef2jau9AbiGOy8jVtr\ni2R0HIoTM2gL3Czp+RjtTJPUy8yOAS4BrsflJu80VW4n9hrgtAkN+FrSmjiu3R84AFcX+J6ol7KA\nk+OZSldVwggwkOwcIOlifN0Nn7ubSNQ1DegiaYikIThBU+EKrl8Th53o4krPlaO4EgCSfpH0tqSx\nQH0zi0cctRpuPTINpyYTOTbidAsDuyDsAwwkOwVeOl5QPGJKpJhRK0kro85X+ba1ZhZzIIRtxZXa\nAn+1BIormVlPXJS2OU6e60HgAZwzvjtWO5I+BT41s2ckLfG2Q2H0OAgjwECy8yCueHgjvy3mC+CO\nBOx8YmZjzexsMzsbeNO31SRGZWi/ufgGXNR1gKTNuFHYuXH25XHgRWAIsBon0b8A6CDpnp1duANu\ns1AYPSHCGmAg6TGzbrgcYAM+lDQzARuGczgH+qYJwOvxFiAysxmSesR7/1I2pkrqHXW+MFZVmp3Z\nM7MzgL44B/1dSIXbNWEKHNgdqI2rd/GcmTUws9aSfozHgHd0I/1RHiab2QBJ35bDRqaZ9WHb2mFe\n9HkC2oLRhdEfkFRgZmFkEwNhBBhIaszsetyorb2kvc2sBfCKpIPitHMKburcGOdoInqAWXHamY3b\nT7gEV8cjYicehZoK1RY0s8twgZxpwPFAa+AFSf8Xj52qSHCAgaTGzKbiKrhNjsqWmB7v9M7M5gMn\nSPqhnP1pU1Z7JAiRLFgojB4TIQgSSHby/PQ1EgWORwghmpXldX5Q7OjqAif4o26izs/M/mCumFHk\nvJ6Z/T4BO03M7Ekze9efdwXOTqRPVY3gAAPJzigzexCoY2bnAu/jcoTjZZKZvWJmp/sskFNsFzU1\nysJngozATaUbAy+Y2R8T6A84zb/iCLSkdcCFCdh5BhiH21YDMBe3OTqwC8IUOJD0+IT/o3HrbeMS\nSfS3sutnxF03w8ymA/tL2uTPawJfJRJxNbMZQM9IJNpcGYDp8Urh2zYZ/WhRhRKR5kDZhChwIGnx\nDuE9SUcB5VI3kRTvXr0dYUB0PZKtJJ4J8h7wipk96s8v9m3xsslvEI840v1w1eUCuyA4wEDSImmr\nmaWaWVaimQ1mdrWkO83sfsqoMifpsjhNPg18Y2ajcY7vRBLXA7wG5/R+588/ILGC5lcCbwHtzZUA\naERIhYuJMAUOJDXe0fTGrf1tirRLujLG60+QNMZnf2yHpGcT6FNf4CCcQ/1C0pRdXPI/w6e+7QdM\nBDrhnPKceHQOqzJhBBhIdsb6IyEkjfFPN0t6Lfo1MxtWjn4Z29cbju1Cs1clDfdrgGWNSmNeT5RU\nZGYP+rW/WfH2paoTRoCBpMQn+J9TgfYml5asKqstBjs3AMNw+cmGy754TVLMUvZm1kzSTxW1p9DM\n/g18BYyKN7WvqhMcYCApScQ57cDOcTjpquHAK1EvZeFEVuNVX54D9IpUb/NK01MldUqgb3dIumZX\nbTHYyQZq4gIyuSSY5VIVCVPgQLJSo1S+bAniyJddAUwCBuMEUCNkA1ck0K8VOIXqSPnKDFwxo0Q4\niu21CI8ro22nSKqd4P2rPGEEGEhK/KjmW8p2gInky6ZHAgNmVg+nBTg9jusjUeTWwABcxFY4JzZR\nUsybqs3sd7gC6O2B+VEv1cbVBzkzVltRNgcDB/vTT7zAamAXBAcYSEqiN/VWkL1PcKPANNxIcBXw\npaSYRoE7iiJHiCeabGZ1cHWPb8NJV0XI9orXcWFmt+Oc8gjfdDowSdJf47VV1QgOMJCU/A8c4BRJ\nfczsAtzo78ZERBUqEnNV4JZJyjNXz7cnTmY/JoHWKDvTgd6Sivx5KjAl6AHumpALHEhW4loHi4E0\nM2uGC4YkPD00swPN7AMzm2tmC81skZktTNDc68BWM+sAPIarV/xigrbqRj2vk6CNKkcIggSSEknv\ng3M4wE1AG9z3NRLhjFdB+WacYMAESd+aWTtgXgJdexIXPPmOkilxiVAkqdCLMtwv6X4zS2RT9W3A\nFK8zaLi1wGt3fkkAwhQ4kOR4AdLtHI6kXyqpP99I2reibAH3AtfhtAoXmdlMSd0TsNUMtw4ILijz\nc0X0cU8nOMBAUlNRDsfMWuKKhUdqgnwOXC5pWYzXR/YkDgdSgVFAXuT1BGTsI7p9l+DUZF4ys7bA\ncEkxFX0ys0slPeCfd5Mr0RmIg+AAA0mNj3CW2+GY2Qe49bVIAfQzgTO80kws11eojH1FEL1ZvKI2\njlc1ggMMJDU7cDyJ7APcTh+vsjTzKioXuJQDrNCoeVUhBEECSY2kwyrI1C9mdibwkj8/HYh7HdHM\nylKh2YArQzk1RjOX+8dB8d6/FHXN7GTcbo6s0grXkkaV0/4eTxgBBpIav2n4RrZlOXwK3CwpLsFP\nLzxwP7C/b5oAXKY4y2ua2YtAfyCiMjMImA7shRNFuDMee+VhByrXEeJWu66KBAcYSGrM7HVgJhDJ\ntPgtTowg7noeFdSfz4CBknL8eS3gbeBY3Ciwaxy2stl+CrwBl7v8Z0mJ7i8MxEiYAgeSnfaShkSd\n/8Ncqcy4KG8UOIrGRAVjgAKgiaRcM8vbwTU74l5gGS44Y8BpuPzgybjCT4fGYsRcZbmzcKPQ4r/p\nBNSuqxzBAQaSnVwzO0jSF1C8MTo3ATtP4xxNRAT1TN8WUxQ4ihE4Sfw3/fkJwIu+ONL3cdoaLKlX\n1PljPjBzjZn9LQ477wBfAzOAojj7UKUJU+BAUmNmvXHT3zq4UdJa4BxJ0+K0U2FRYDPrz7aR5ARJ\nk+K14e18BdwDjPRNQ4ErJe0XT9/CFpjECQ4wsFtgZlkA5SiO9BFuxBcdBT5X0hGx3l/SRjOrX9br\nCaq4tAP+y7bAzFe4rJflQL/IqDcGO1cAObgc5+i9knH3qaoRHGAgKTGzMyW9sINtJ0j6T5z2oqPA\nAr4kjiiwmY2VNMjMFrGtFkjxYwK5yRWGmf0BuBVYz7agSqX2aXchrAEGkpWa/rFC1I59nY3B5bh+\nkH9sWxH9gQoNzPwZ6CBpTUX1raoQRoCBPRozywROBdbh9u5dhdtTuAC4JV6nYWYGnAG0lXSLmbUG\nmkqamEDfypWeF2XnfeAkSZvj7UNVJzjAQFJjZncC/8RFft/DiYZeIemFGK9/FbdVpSZOhXkmzhEe\nhBMRjSsbw8wexkVaD5fUxcvrvy9pwC4uLctWhQRmzNVO7gZ8TMk1wLANZheEKXAg2Tla0tU+5Wsx\ncArwGRCTA8RVfutuZmk49eVDfPt7ZhZXJNmzr6S+Ed0+SevMrFoCdqCC0vOAN/wRiJPgAAPJTuQ7\nejwu1WyDm4XGTD6AFx5dUeq1RARNC7zkvADMrBGJ7707D7cGeA/bAjPnxGtE0rPeCe/tm+ZECkAF\ndk5wgIFkZ6wXRc0FfucdzpZdXBNNSzO7DxetjTzHn7dIoD/3AaOBxmZ2K27v3vUJ2CkzMGNmf8Jl\niMSMryfyLG6EbEArMztb0meJ9KsqEdYAA0mP33u3QdJWM6sBZMWqeFyR1dyibHYGjsA5m48k/RCv\njZ3Y/lFS6ziv+Q74jaQ5/nxv4CVJ/SqqX3sqYQQYSErM7HBJ46MlnkpNfWOSeoo4ODMbJum1UvcY\nVvZVu7Q5G5idyLUxENf83pMecX4AkuaaWXoF9mmPJTjAQLJyCDAel2tbGhGjA4zir8BrMbSVSSnl\nFot6ngZUk1RRf0uJTMkmmdkTbAsMnYFTlAnsgjAFDuzRmNlxwEBcLY9Xol7KwkWI90nQbi3gD8DF\nwGhJf47j2rJksMA51urxOlMzy/B9Ocg3fQ48JCledZoqR3CAgaTGzP4F3ClfLNzvu/uzpJgCD2bW\nC+iNK4t5Q9RL2cDHktbF2Z+6wJ9w8lMvAveokirUBcpPcICBpKasWheJqJ+YWXp5toaYWUNcytmp\nOK2+++NVpa5odlRTJEKstUWqMmENMJDspJpZRmQ6Z2bVgYwE7OxjZjeReIH1JcBqnKLMZuD86KBM\nvOIMFUQki+UP/jE6pS6MbGIgOMBAsjMC+Ciq/sW5bJPHj4cnKaPAehzcxTanUlqgoVKcjd9HiJkd\nVWqUfI2ZTQaurYx+7U4EBxhIaiTd4VPWjvRNt0gal4CpDZLeLUdXnpS0tKwXzKy81d3Ki5nZgZIm\n+JMDcJXiArsgrAEGkh6v5ddR0od+I3SqpOw4bZSrwLrPRjlW0uJS7ecC10tqH09/KhIz64dbl4yo\nZq8Dzov1s1VlggMMJDVmdiFwEVBfUnsz6wg8EquSc5SdchVYN7OBuBS14yXN821/BX4DHJeAhl+F\n40uIUtnBmd2J4AADSY2vALcP8E1kncvMZkjqUQl9OQJ4FDgJuMD36/h4t9L8D/qVAQxh+6pwN1dW\nn3YXwjpBINnJk5QfOfGyVnH/1zazJmb2pJm968+7mtn58diQ9BEuCPMJ0A6nCVipzs/zJnAiUAhs\nijoCuyCMAANJjRdEXY/bePxH4PfA95Kui9POu7gtLNdJ6uUd6ZRYR5JR2RuG24ZTgIsmR7bTZMXT\nn4rEzGZK6l5Z99+dCSPAQLJzLW7/3Qxc2tk7JCY/1VDSq3jtPkmFxLEdRlJtSVn+sZqkmlHnleb8\nPF+a2a++JLAnELbBBJIaSUVm9gbwhqTV5TC1ycwasE3IdD9gTwkWHASc4yvW5bFtVBoyQXZBcICB\npMQXH7oRuBQ/UzGzrbgUtEQW968E3gLam9kEoBFOzHRP4LjK7sDuSlgDDCQlvh7wccBFkhb5tnbA\nw8B7ku5JwGYa0Ak3QtrjZOPNrDGQGTlXjDWPqzLBAQaSEl906KjSZSu9JP77pQUSdmJnO2HVaCTF\nqyuYdJjZYOBuoDmwCpfv/IOkbpXasd2AMAUOJCvpZdXslbQ6TrXjihZWTUZuAfYDPpTWxu2JAAAF\nr0lEQVTUx8wOwwkiBHZBcICBZCU/wddKIOlG/3huuXuUvBRI+sXMUswsRdLHZhZXYaWqSnCAgWSl\nl5ltLKPdiFrn2hV+LXGHVJKMVUWz3itUfwaMMLNVhI3QMREcYCApkZRaQaYi0lWdgAG4SDC4KfHE\nCrpHZXMirmzoFbh6IHVwCtiBXRCCIIEqgZl9hsvbzfbntYG3JR1cuT2reMwsBThd0ojK7kuyEzJB\nAlWFJpRcO8z3bbstZpZlZn81swfM7GhzXAosxBWBCuyCMAUOVBWeAyaa2Wh/fhKJKUsnE8/jtP++\nwqnT/A23RnqSpKmV2bHdhTAFDlQZvHBopHTkZ5KmVGZ/yku0LJiZpQI/Aa0lbancnu0+hBFgoMog\n6TszW4qPIptZ6908W6I4k0XSVjNbFpxffIQRYKBKUEa2RGtg9u6cLeFzoyPbXQyojqtYV+kSXbsL\nYQQYqCrscdkSFbhVqMoSosCBqkKBpF+A4mwJoH9ldypQuYQRYKCqELIlAtsR1gADVQIzq4nLlkhh\nW7bECD8qDFRRggMM7PH4LSIfSjqssvsSSC7CGmBgj0fSVqAoUjc3EIgQ1gADVYUcYIaZfUDU2p+k\nyyqvS4HKJjjAQFVhFNvETyPrPlZJfQkkCcEBBvZozOxEoKWkB/35RFxBJAHXVGbfApVPWAMM7Olc\nzTYNQIBqQD/gUOCSyuhQIHkII8DAnk41SUujzr+QtBZY67fGBKowYQQY2NOpF30i6dKo00a/cl8C\nSUZwgIE9nW/M7MLSjWZ2MXuOJH4gQcJG6MAejS8W/gaQB0z2zf2ADJxw6MrK6lug8gkOMFAlMLPD\ngYj01SxJ4yuzP4HkIDjAQCBQZQlrgIFAoMoSHGAgEKiyBAcY2KMws61mNtXMZprZa2ZWoxy2DjWz\nsf75YDO7difvrWtmv0/gHjeZ2V9ibS/1nmfMbGgc99rLzGbG28c9meAAA3sauZJ6S+qOq/1bItvD\n186N+3sv6S1Jt+/kLXWBuB1goHIJDjCwJ/M50MGPfOaY2XPATKCVLyT+lZlN9iPFWgBmdqyZzTaz\nycApEUNmdo6ZPeCfNzGz0WY2zR8HALcD7f3o8y7/vqvM7Fszm25m/4iydZ2ZzTWzL4BOu/oQZnah\ntzPNzF4vNao90swmeXuD/PtTzeyuqHtfXN4f5J5KcICBPRIzSwOOA2b4po7AQ74K3CbgeuBISX2B\nScCVZpYJPA6cgNsr2HQH5u8DPpXUC+gLzAKuBRb40edVZna0v+c+QG+gn5kd7GsTn+bbBgIDYvg4\noyQN8Pf7ATg/6rW9/D2OBx7xn+F8YIOkAd7+hWbWNob7VDlCLnBgT6O6mU31zz8HnsSVwlwi6Wvf\nvh/QFZhgZuAEEr4COgOLJM0DMLMXgIvKuMfhwFlQLLa6wczqlXrP0f6IFF+vhXOItYHRkjb7e7zF\nruluZv/ETbNrAeOiXntVUhEwz8wW+s9wNNAzan2wjr/33BjuVaUIDjCwp5ErqXd0g3dy0QWQDPhA\n0uml3lfiunJiwG2SHi11jz8lYOsZXNbKNDM7B6dkE6H0Rl75e/9RUrSjxMz2SuDeezRhChyoinwN\nHGhmHcAVTDKzvYHZwF5m1t6/7/QdXP8R8Dt/baqX2s/Gje4ijAPOi1pbbOHT8j4DTjKz6mZWGzfd\n3hW1gZ/MLB1X0CmaYWaW4vvcDpjj7/07/37MbO+gfFM2YQQYqHJIWu1HUi+ZWYZvvl7SXDO7CHjb\nzDbjptC1yzBxOfCYmZ0PbAV+J+krM5vgt5m869cBuwBf+RFoDnCmpMlm9gowDVgFfBtDl/8OfAOs\n9o/RffoRJ+qQBVwiaYuZPYFbG5xs7uargZNi++lULUIqXCAQqLKEKXAgEKiyBAcYCASqLMEBBgKB\nKktwgIFAoMoSHGAgEKiyBAcYCASqLMEBBgKBKsv/A4a3aSV3ZupSAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f17aa6fb8d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# get best classifier for each dataset\n",
    "from tqdm import tqdm \n",
    "best_method = dict()\n",
    "for i,(dataset, group_data) in enumerate(tqdm(data.groupby('dataset'))):\n",
    "    best_method[dataset] = group_data['classifier'][np.argmax(group_data['accuracy'])]\n",
    "\n",
    "# print(best_method)\n",
    "\n",
    "# make new dataset combining metafeatures and best methods\n",
    "y = np.empty(metafeatures.shape[0])\n",
    "methods = data['classifier'].unique()\n",
    "print('methods:',methods)\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "le = LabelEncoder()\n",
    "# le.fit(methods)\n",
    "\n",
    "print('metafeatures[''dataset''].shape:',metafeatures['dataset'].shape)\n",
    "y_str = [best_method[ds] for ds in metafeatures['dataset'].values]\n",
    "\n",
    "y = le.fit_transform(y_str)\n",
    "\n",
    "metaf = metafeatures.dropna(axis=1,how='all')\n",
    "metaf.fillna(value=0,axis=1,inplace=True)\n",
    "print(metafeatures.shape[1]-metaf.shape[1],' features dropped due to missing values')\n",
    "# print(metaf[:10])\n",
    "from sklearn.preprocessing import StandardScaler, Normalizer\n",
    "\n",
    "X = Normalizer().fit_transform(metaf.drop('dataset',axis=1).values)\n",
    "print('X shape:',X.shape)\n",
    "print('y shape:',y.shape)\n",
    "\n",
    "# set up ML \n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.model_selection import cross_val_score, StratifiedShuffleSplit, LeaveOneOut, cross_val_predict\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "# dtc = DecisionTreeClassifier()\n",
    "dtc = RandomForestClassifier(n_estimators=1000)\n",
    "# dtc = KNeighborsClassifier(n_neighbors=1)\n",
    "# dtc.fit(X_t,y_t)\n",
    "cv = StratifiedShuffleSplit(n_splits=30,test_size=0.1)\n",
    "print('fitting model...')\n",
    "# print('mean CV score:',np.mean(cross_val_score(dtc,X,y,cv=LeaveOneOut())))\n",
    "# print('mean CV score:',np.mean(cross_val_score(dtc,X,y,cv=cv)))\n",
    "\n",
    "print('confusion matrix:')\n",
    "import matplotlib.pyplot as plt\n",
    "import itertools\n",
    "%matplotlib inline\n",
    "\n",
    "def plot_confusion_matrix(cm, classes,\n",
    "                          normalize=False,\n",
    "                          title='Confusion matrix',\n",
    "                          cmap=plt.cm.Blues):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    \"\"\"\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=90)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "        print(\"Normalized confusion matrix\")\n",
    "    else:\n",
    "        print('Confusion matrix, without normalization')\n",
    "\n",
    "    print(cm)\n",
    "\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, cm[i, j],\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "\n",
    "# Compute confusion matrix\n",
    "cnf_matrix = confusion_matrix(y,cross_val_predict(dtc,X,y,cv=LeaveOneOut()))\n",
    "np.set_printoptions(precision=2)\n",
    "\n",
    "# Plot non-normalized confusion matrix\n",
    "plt.figure()\n",
    "plot_confusion_matrix(cnf_matrix, classes=le.classes_,\n",
    "                      title='Confusion matrix, without normalization')\n",
    "\n",
    "# Plot normalized confusion matrix\n",
    "plt.figure()\n",
    "plot_confusion_matrix(cnf_matrix, classes=le.classes_, normalize=True,\n",
    "                      title='Normalized confusion matrix')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# can we predict the best score achievable on a dataset from dataset properties?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 166/166 [00:00<00:00, 366.18it/s]\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/pandas/core/frame.py:2842: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  downcast=downcast, **kwargs)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 38 iterations, alpha=3.126e-04, previous alpha=4.224e-05, with an active set of 23 regressors.\n",
      "  ConvergenceWarning)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 40 iterations, alpha=8.702e-05, previous alpha=4.364e-05, with an active set of 21 regressors.\n",
      "  ConvergenceWarning)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 45 iterations, alpha=8.415e-04, previous alpha=2.058e-05, with an active set of 28 regressors.\n",
      "  ConvergenceWarning)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 37 iterations, alpha=1.792e-04, previous alpha=5.883e-05, with an active set of 24 regressors.\n",
      "  ConvergenceWarning)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 44 iterations, alpha=8.357e-04, previous alpha=2.484e-05, with an active set of 29 regressors.\n",
      "  ConvergenceWarning)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 42 iterations, alpha=3.816e-05, previous alpha=3.540e-05, with an active set of 27 regressors.\n",
      "  ConvergenceWarning)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 42 iterations, alpha=1.294e-04, previous alpha=2.623e-05, with an active set of 21 regressors.\n",
      "  ConvergenceWarning)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 41 iterations, alpha=1.121e-04, previous alpha=6.574e-06, with an active set of 26 regressors.\n",
      "  ConvergenceWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "metafeatures[dataset].shape: (166,)\n",
      "9  features dropped due to missing values\n",
      "X shape: (166, 35)\n",
      "y shape: (166,)\n",
      "fitting model...\n",
      "mean CV score: -0.194007105864\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 46 iterations, alpha=6.347e-04, previous alpha=2.807e-05, with an active set of 31 regressors.\n",
      "  ConvergenceWarning)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 42 iterations, alpha=2.737e-04, previous alpha=4.824e-06, with an active set of 25 regressors.\n",
      "  ConvergenceWarning)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 48 iterations, alpha=6.191e-05, previous alpha=1.402e-05, with an active set of 23 regressors.\n",
      "  ConvergenceWarning)\n",
      "/home/bill/anaconda3/lib/python3.5/site-packages/sklearn/linear_model/least_angle.py:334: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 46 iterations, alpha=6.347e-04, previous alpha=2.807e-05, with an active set of 31 regressors.\n",
      "  ConvergenceWarning)\n"
     ]
    }
   ],
   "source": [
    "# get best classifier for each dataset\n",
    "from tqdm import tqdm \n",
    "best_score = dict()\n",
    "\n",
    "\n",
    "# print(best_method)\n",
    "\n",
    "# make new dataset combining metafeatures and best methods\n",
    "y = np.empty(metafeatures.shape[0])\n",
    "for i,(dataset, group_data) in enumerate(tqdm(data.groupby('dataset'))):\n",
    "   y[i] = group_data['bal_accuracy'].max()\n",
    "\n",
    "print('metafeatures[''dataset''].shape:',metafeatures['dataset'].shape)\n",
    "\n",
    "\n",
    "metaf = metafeatures.dropna(axis=1,how='all')\n",
    "metaf.fillna(value=0,axis=1,inplace=True)\n",
    "\n",
    "print(metafeatures.shape[1]-metaf.shape[1],' features dropped due to missing values')\n",
    "# print(metaf[:10])\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "X = StandardScaler().fit_transform(metaf.drop('dataset',axis=1).values)\n",
    "print('X shape:',X.shape)\n",
    "print('y shape:',y.shape)\n",
    "\n",
    "from sklearn.tree import DecisionTreeRegressor\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.linear_model import LassoLarsCV\n",
    "from sklearn.model_selection import cross_val_score\n",
    "\n",
    "# X_t,X_v,y_t,y_v = train_test_split(X,y)\n",
    "\n",
    "# est = DecisionTreeClassifier()\n",
    "est = RandomForestRegressor(n_estimators=100)\n",
    "# est = LassoLarsCV()\n",
    "# dtc.fit(X_t,y_t)\n",
    "print('fitting model...')\n",
    "print('mean CV score:',np.mean(cross_val_score(est,X,y,cv=5)))"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [Root]",
   "language": "python",
   "name": "Python [Root]"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}