1_density_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "from matplotlib import pyplot as plt\n",
    "from matminer.featurizers.base import MultipleFeaturizer, StackedFeaturizer\n",
    "from matminer.featurizers import composition as cf\n",
    "from sklearn.ensemble import RandomForestClassifier, RandomForestRegressor\n",
    "from sklearn.model_selection import GridSearchCV, ShuffleSplit, LeaveOneGroupOut, cross_val_score, learning_curve, KFold\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.preprocessing import Imputer, LabelEncoder\n",
    "from sklearn.metrics import roc_curve, auc, r2_score, make_scorer, mean_absolute_error\n",
    "from sklearn.feature_selection import RFECV\n",
    "from sklearn import metrics\n",
    "from pymatgen import Composition\n",
    "import pickle as pkl\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import gzip\n",
    "import os\n",
    "import re"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "saveFigures=False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def featCleanImpute(Y):\n",
    "    ''' Convert inf to NaN in feature array, in place\n",
    "    \n",
    "    Args: \n",
    "        Y, feature array, list of ndarray (#samples x #features)\n",
    "    Returns:\n",
    "        None\n",
    "    '''\n",
    "    # Clean inf values\n",
    "    for i in range(len(Y)):\n",
    "        for j in range(len(Y[i])):\n",
    "            if Y[i][j] == np.inf:\n",
    "                Y[i][j] = np.nan\n",
    "            else:\n",
    "                Y[i][j] = Y[i][j]\n",
    "                \n",
    "    # Impute nan values\n",
    "    imp = Imputer(missing_values='NaN', axis=0, strategy='mean', copy=False)\n",
    "    imp2 = Imputer(missing_values='NaN', axis=1, strategy='mean', copy=False)\n",
    "\n",
    "    imp.fit(Y)\n",
    "    imp.transform(Y)\n",
    "    imp2.fit(Y)\n",
    "    imp.transform(Y)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "elemProp = pd.read_excel('datasets/ElementalDensityIonizationTable.xlsx')\n",
    "elemDensity = elemProp[['symbol', 'density [g/cm3]']]\n",
    "\n",
    "def vegardsDensity(comp):\n",
    "    '''\n",
    "    input: Composition object\n",
    "    '''\n",
    "        \n",
    "    density=0\n",
    "    elemSet = set(comp.elements)\n",
    "    subscriptCnt = 0\n",
    "    \n",
    "    for elem in elemSet:\n",
    "        subscriptCnt += comp[elem]\n",
    "        try: \n",
    "            densityPart = comp[elem]*elemDensity[ elemDensity['symbol'] == str(elem)]['density [g/cm3]'].item()\n",
    "        except:\n",
    "            print('element {} not found?'.format(elem))\n",
    "            return 'error'\n",
    "            \n",
    "        density += densityPart\n",
    "        # print('element {} added density of {}: {}'.format(elem, densityPart, density))\n",
    "                \n",
    "    # print(subscriptCnt)\n",
    "    return density / subscriptCnt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build Model: Density"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import feature set, import data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "with gzip.open('./datasets/density_features.pkl.gz', 'rb') as fp:\n",
    "    X_dens = pkl.load(fp)\n",
    "with gzip.open('./datasets/density_data.pkl.gz', 'rb') as fd:\n",
    "    densityData = pkl.load(fd)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "density_model = Pipeline([('impute',Imputer()), \n",
    "                          ('model', RandomForestRegressor(n_estimators=100, n_jobs=1, max_features=12))])\n",
    "dens_featurizer = MultipleFeaturizer([cf.Stoichiometry(), cf.ElementProperty.from_preset(\"magpie\"),\n",
    "                                 cf.ValenceOrbital(props=['avg']), cf.IonProperty(fast=True),\n",
    "                                cf.YangSolidSolution(), cf.AtomicPackingEfficiency()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 390 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Pipeline(memory=None,\n",
       "     steps=[('impute', Imputer(axis=0, copy=True, missing_values='NaN', strategy='mean', verbose=0)), ('model', RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=None,\n",
       "           max_features=12, max_leaf_nodes=None, min_impurity_decrease=0.0,\n",
       "           min_impurity_split=None, min_samples_leaf=1,\n",
       "           min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
       "           n_estimators=100, n_jobs=1, oob_score=False, random_state=None,\n",
       "           verbose=0, warm_start=False))])"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "density_model.fit(X_dens, y=densityData['density'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Export for testing with MAPI data?\n",
    "with gzip.open('./datasets/MG_density_model.pkl.gz', 'wb') as fp:\n",
    "    pkl.dump(density_model, fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(701, 152)\n",
      "152\n",
      "152\n"
     ]
    }
   ],
   "source": [
    "print(np.shape(X_dens)) # should be (701 x 152)\n",
    "print(len(density_model.steps[1][1].feature_importances_)) # should be 152\n",
    "print(len(dens_featurizer.feature_labels())) # should be 152"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Look at feature importances when trained on the whole data set\n",
    "featureImp = pd.DataFrame(density_model.steps[1][1].feature_importances_,\n",
    "                          index=dens_featurizer.feature_labels(),\n",
    "                          columns=['importance']).sort_values('importance',ascending=False)\n",
    "# featureImp.to_csv(os.path.join(os.getcwd(),'results','defaultDensityModel_features.csv'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# make scorer for r2 metric\n",
    "r2_scorer = make_scorer(r2_score)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Generate learning curves, assess initial quality of model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_sizes, train_scores, valid_scores = learning_curve(density_model, X_dens, densityData['density'], cv=ShuffleSplit(),\n",
    "                                                          scoring=r2_scorer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaoAAAEWCAYAAAA3h9P4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4xLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvAOZPmwAAIABJREFUeJzsnXmYXFWZ/z9vbV3Ve/atycZOEkIgkERBowxLXECCoyIiIgODsgmIhgGViSDLiIKIC+MwKjIw7uJvQEA0ChJIAkkgGxACJJ10Z196q/We3x/nVvXt6qpOd7q7qrr7/TzPferWveee+567nO8957znHDHGoCiKoiiliq/YBiiKoihKV6hQKYqiKCWNCpWiKIpS0qhQKYqiKCWNCpWiKIpS0qhQKYqiKCXNkBIqEXlSRC4uth1KaSEik0XEiEggz/6jRWSliDSJyDWFtk/pPiISEZE/ish+EflVse0pRURkrYjML7YdPaEgQiUi74jIPxXiXF1hjFlgjPlZf8QtItUicq+IbBaRZhHZ6P4f2R/nO1Q8mXKzu7wjIouywrwjIm2eMM0iMr4INuYUjiLE9RVgiTGmyhjzvV7as0RE/qU3cQw2ROQTIvKCiLSKyJKsfUeJyB9EZKeI7BGRp0Tk6C6i+zgwBhhhjPnnXtp1q4j8ojdx9BYReVBELu/F8T8Vkdu824wx04wxS3ptXOdzzReR+r6OFwZRiaovMrVenDsEPAtMA84GqoH3ALuBUw4hvkKkpdYYU4l9sb8mImdk7f+oMabSs2zrSeTFvB/9wCRgbbGNgEF3XdPsAe4F7syxrxZ4HDgaK0DLgD90Edck4A1jTLKvjewpfXSvzgae6IN4Sp4ur5cxpt8X4B3gn/Ls+wiwCtgHvAAc79m3CHgLaALWAed59n0O+AfwXeyDfpu77Xng28Be4G1ggeeYJcC/eI7vKuwU4O/uuf8MPAD8Ik8a/gXYDlR2cQ0McITn/0+B29z1+UA98FWgEXgYWA98xBM+AOwCTnT/z3Wv1z5gNTC/m/dismtLwLNtGXBjN+/XOdhMe597PY/NOu6rwKtAzLV5PPAbYKd7ja/xhD8FWAEccK/fd9ztm10bm91lXjfS1e24AL9733cBm4Ars6+JJ96/ACkg6h5/FFDmHr/ZPdePgIgbfhjw/9z07nXX69x9t2fF9f0892MJHZ/TDs+5u/3z7jOyF3gKmORuFzfsDmC/ey+md/PZMMAXgTexz/03gcOBpe51/SUQ6kY6h2Of54+6/yuBjcBnD3L+f8GWXLsKM9y1c0SOff8OxIGEe30v7epaufvuA7a46XsZOM3dfnZWXKtzvRvArbj5gudeXuo+G38/2Lvq3t9N7vV+G7jQs+944FXP/x7dc+By1/64m4Y/ZqfBtf9XwC9cG17DPuM3ufFtAc702HCJa0OTa/e/utsrgDbAof1dG499V+4FtrnLvUBZvnwv733vzgPc2yX75nq2n+hejDnYzONiN2w6If/sJtYHfBJoAcZ5bnASuBqbIUbcbQngMje+L7gXR/JkAF2FXYrNjELAqdgHOZ9QPQb8rBuZQFdClQTucm9sBPg68Ign/IeBDe76BGxp7UPutTnD/T+qG/diMp6MEfsStdLxIyDf/TrKvQdnAEFsldhG2jOvd7AfHYe5afBhX/6vu9dxKvbhPstzjS/yZGZzc9nobpuIfdEn5klXT+K6Atjg2jkc+Gt2mKy4M8+N+/9e7Ff+cKAK+CNwh7tvBHA+UO7u+xXw+y7iymVfJgy5n/OPudf9WHfbLcALbviz3Gtei83AjsV9Z7rxbBg3XdXY2oEYtqZgKlCD/Vi8uJvpPBOb+YwG/hP4dTfO3x2h+hjQ0MX+W/G8p11dK3f/Z9y0BIAbXJvDueLK9W6QW6h+js24I3TxrrphDgBHu8ePA6Z54l7kea4O6Z7jyWdypcG1P+rGEXBtfxu4GfuOXwa8nZUPHe6e5/3YvCP98TwfqM8612LgRfc5GIUV7G/my/fy3tfuPMC9XbJvrmf7D9NGe7a9Drw/TzyrgHM9L/DmrP2fAzZ6/pe7D87YPBlAzrDYTDEJlHv2/4L8QvUMcGc3MoGuhCqO+4K4247AfrWUu/8fAb7urn+VrK8P7BfWxd24F5NdW/Zhv4AMVpAl6341u2H24WZAwNeAX3rC+YCtuF+I7nGf9+yfk+Me3QT8t7v+d+xX8Mg8NuYUjjzp6nZc2FLSFZ7/Z3Z1vqznRrBifbhn/zw8L3PWsScAe3PF1YV92c9p9jV8Ere04LkPrdhqrw8Cb2A/QHw9fE8N8F7P/5eBr3r+3wPc2510utvux36hbyNHCShHHF0KFVDnPm8XdBHmVjoKVd5rlef4vcDMXHF5nvGDCdVUz/687ypWqPZhBb9TJg08R3sJ75DuOd0Tqmc8+z6Kfff97v8qN021ea7X74Fr3fX5dBaqt4APef6fBbzjCd8h38u3FLuNahJwg4jsSy/Yr9zxACLyWRFZ5dk3HfA6J2zJEWdjesUY0+quVuY5f76w44E9nm35zpVmN/ZrqDfsNMZEPfZsxBaxPyoi5dgqt/9xd08C/jnrup3aQxtGYtP6ZewDE8za/zFjTK27fMzdNh5412Ojg70uEzzHea/TJGB8lp3/hm1rAFtFchSwQUSWi8hHemB/Nj2Ja3yWne/mC5iDUdiPmpc9afqTux0RKReRH4vIuyJyACugtSLi70lissh+9iYB93nOvwcroBOMMX/BVik+AGx3G+Ore3Cu7Z71thz/K6Hb6XwQ+87+tzFmdw9s6ISIjAKeBn5gjHm0B4fmvVZuvDeIyHrXS3AftuTYWweo7Hcg57tqjGnB1hRdATSIyP+JyDGuXbXAMdgSSJfp6Id7vssYk/L8h/b7vkBEXnQdW/ZhS4pdXa8OeYa77nXM6pDv5aPYQrUFuN2TIdYaY8qNMY+KyCRslcFV2K+xWmAN9uakMf1kVwMw3BWINId1Ef7PwFkiUtFFmFZsBpdmbNb+XGl5FLgAOBdY54oX2Ov2cNZ1qzDG5GqMzosxJmWMuQdb9P9iNw7Zhn1hABARwV6XrXnSsQVb0vDaWWWM+ZB7/jeNMRdgqwXuAn7tXsMe39cextVAx/s5sQen2oV9ead50lRjrGMK2Oqjo4E5xphq4H3u9vRzm21Pi/vbk2djC7ZtwHtdI8aYFwCMMd8zxpyErb47CrixB+nrLl2m0xWsH2Orkr4gIkcc6olEZBhWpB43xtzew8PzXisROQ1b4vkEMMzNY/aT/16BvV9d3avs47p8V40xTxljzsB+ZG7A5nlgSx7PegTjUO95n+WRIlKGbW/+NjDGvV5P0PX16pBnYN81r2NWt+wrpFAFRSTsWQLYm3KFiMwRS4WIfFhEqrDFYoNtrEVELsF+nfU7xph3sQ3zt4pISETmYYvE+XgY+yD9RkSOERGfiIwQkX8TkQ+5YVYBnxYRv4icja3fPRiPYaulvkB7aQpsNeRHReQsN76w6xpaBxm32iU9SPKdwFdEJHyQcL8EPiwip4tIEJtZxWj/6stmGXBARL4qtn+LX0Smi8jJrp2fEZFRbslsn3tMCnvPHWzbSLfoYVy/BK4RkTo3E1xEN3Hj/0/guyIy2j33BBE5yw1ShRWyfSIyHPhGVhTbvbYYY3Zihf4z7vX5PLYNoCt+BNwkItPc89eIyD+76ye771MQm6lG3euAiHxORN7pbloPwsHS+W/u7+exGdvP85Uq088wto3E5z7PQXdfNbaq7B/GmG7fJw95r5WbhiT2GQmIyNex7XNptgOTRcSbT64CPiUiQRGZjfWa7Yq876qIjBGRc9wPqhi2yi0tTB+mo7ffId1zsp63XhLCtiXtBJIisgCbP6XZDowQkRrPtkeBW0RklNiuOl93r0mPKKRQPYF9sNPLrcaYFdjGuu9j64Y3YuvkMcasw9aJL8VegBlY76dCcSG27WE31qPwf7EPUyeMMTHgn7BfRM9gG0iXYYvEL7nBrsWK3T437t8fzABjTAM2/e9xz5/evgVbyvo37EOzBfsFlb6fh9Gza/V/2Ot/2UHseR3b+Hw/tmTxUaxnVzxP+JQb5gRsA+0u4CfY6hWwnlVrRaQZ6331KWNM1K1yvR34h1vVMVdEJortz5Wv9NPtuLBC8xTWA+sV4LcHu0BZfBX7rL4ottrrz9jSBVhHi4ib1hex1YJe7gM+LiJ7RSTdJ+sy7P3bjf0izif8ABhjfoctNT7mnn8NsMDdXe2mby+2mmU3Viig589FV+RNp4icBFyP9fJLubYa8n8QXITNE34InOaup0sW5wEnA5dIx3593SoFH+RaPYVt+3kDe62idKy2S3cY3i0ir7jrX8N+SOzFtol6PyBznb+rd9WH/djbhq3Kez/wRbem4gw817QX9/y/gOPcZ/+gec5B0tIEXIP90NsLfBrrfJPevwErTJvc843H5p0rsJ6Ir2Hft9voIWkPN+UgiMj/Yr3usr8cSw4RWQWc3tt2AWVwISJPYxu+1xfbFiU/InIK8H1jTI/7YA5WVKjy4FZP7cGWBM7EloDmGWNWFtUwRVEGNa5QjTDGPFlsW0qFglX9ichDIrJDRNbk2S8i8j2xQw+9KiInevZdLCJvusvFBTJ5LNZNuBn4HvAFFSlFUfobY8wyFamOFKxEJSLvw2b6PzfGdHKKcJ0Orsa6O84B7jPGzHEbalcAs7H13C8DJxlj9hbEcEVRFKWoFKxEZYz5O7YqLR/nYkXMGGNexPbJGId103zGGLPHFadnsA3niqIoyhCglAa4nEBHj5t6d1u+7Z0QO8rw5QCRSOSkww7rqutT4XAcB5+v2F3W+h5N18BhMKYJNF39wRtvvLHLGDOqKCfPQykJleTYZrrY3nmjMQ9ie8Mze/Zss2LFir6zrhcsWbKE+fPnF9uMPkfTNXAYjGkCTVd/ICI9GamlIJTSp0g9HUcLqMP2L8i3XVEURRkClJJQPQ581vX+mwvsdzu8PgWcKSLD3FEEznS3KYqiKEOAglX9icij2MFPR4qdBfIbuAOhGmN+hB254kPYHv+t2HlPMMbsEZFvAsvdqBYbY7pyylAURVEGEQUTKnfA0K72G+wEdrn2PQQ81B92KYqiKKVNKVX9KYqiKEonVKgURVGUkkaFSlEURSlpVKiU3qMDGyuK0o+UUodfpVQwBhwHUin7m15SKUgm25d4HDZutPsAfD67iLSv51v8fhsuHTa9frBFUZQhhwrVUMArNl4B8opOeknvS2NMu0B4RSU9vEs43L5uTMclfS7vtrQ9h1oK60oM/f6Ovz0RQBVDRSlZVKgGGmkByLWkxSaR6Cg++cjO7AMBCIW6n1l7BSv9v78z+q7EMC2w2cLYU2Ix2LSpswDmW1QMFaVfUaEqNrmEJ5XqXM2WLu2khcdb0kmTXdIIBqGsbHBljIXI6H0+K9hpsUskOovfoYqhMR2rPrM/FLorhvkEUlEGISpUfU2+0s7+/e0lnfRvdjVbmrQIZWdmoZCtalP6n/4cuTpfqTAW654Q5vpIMaa95Od9ZtLil0zC7t0dS4jdFT8VQKXIqFAdjFxtOwdr38nOSBIJ2LGjYybS02o2ZfDQX5l/LgGMRu16KgX79nWsIs0+Np9N2e2Cfn/u6tBAoHvil96m9BhjDAaT89cxTmbd7/MTDgyej1oVqly0tkJjY+72nfQLnevLNZ/w+HxQUdH/ditDm64E0OeDSOTQ4s0WwHStQG8dZLwC2JX4edsKA4H2moYSJVtEHON0KTApJ0XKpHCMk1kSToIt+7fgGIeUSWXCIbRPfpT1m56t3WAoD5YzsWZi0a5BX6NClYt01VxlZbEtUQYRxhiSJknCSWKMQTyZrWR+7VrHfR23Cfn35dvWKwpVAsxuC8xX/SnS3v6aSkFzsxWwbopYV6KRq3TiFRDv4hUYYwwODo7jdBIRg3uv8wiMIIhIh9/0eUWEoAQ77D8YSff5GkyoUOWjhL/YlNImLUhJJ0kilaTNiRJNxoin4pgOuVQ6fA5Rkfa48j2JBtMuWpLZ2Amf+Iil4mzc/zY+t4+/uG1wQruw+UQy4YFMWCAz22zmeO9xvhzh3ThyimtmGx33+dL7BPHElQ6fERDjYFIxTFMLTiJO07tv4Lj7UiaFE/CTCgVxggG7+H04PsH4fDhiMmJir63kLaWkBcYrIGl7vNsCvkCPhKQ7iAhBf7BP4hoMqFApSi9IOq4gOUmiqRhtyTZiriAZbObvw0fQF6A8EOm7Uk4PMMbgEyHsL+uwDXCttDjGtl2l3DYs7z6TMh2O6yC0XcTZlYCmd+YWXNN5H4BIB/FOkKJBWqxQiBvSSeBrbUMcB3Fs2ICA4EMCQaSszJbGQqH26sSAv38daJReoUKlKN0g5aRIOAmSJkU0GaUtFSWWipMyKfv1LwY/fgI+f9EEKR+SVVKyG4tkTB/jkyYqgz1o/3XcLh6xmK029JJuZw6FrJCl283S7WRK0VCh8vLII3DzzbB5M4wbBzfdBAsXFtsqpYCknBQGQ0uilWgyStSJ0ZaM2tKG2DohPz78EiDiD/dIkCKPP0nVd76Pv2E7qXFjaLr+KtrOWdB/iVE64/PbJZijWi3tzRuL2e4kxpCpC/T7IRSEsrA9NhhsFzK/v9CpGHKoUKV55BG4/HLr8QewbRt85St2XcVq0OEYJ1NlF3fitCVtO1KSJPFUnK0tDfjER0AChP1lHUsjh0Dk8SepueU2fNEoAIFtjdTcchuAilWp4O0on026e0pTU+eSmM8HZSEIudWJwWC7x6KKWJ+gQpXm5pvbRSpNWxtcey3cf7/1AKyqav+tqLC/3m3Z+9L/laJhjMlU2cVTVpDaUjESTtx6VwF+8eEXPyF/kLCU9bw6yZ4IiUaR1jakpdX+trbia7Xr1bf9R0ak0viiUWoW34Vv3z6Mm6mZYAD8Afs/YL/W0+s2TAATsNVRxu+HoPub2RZo35aJzz9gR7gvmVJoVyKW7qfW3Oxx03dLYt4RYsrK2sUrXaWodAu9Umk2b8693XHgyCPtQ9jUBA0N9re52S7dyABOC4WgutoKV7aoHUzk0r9VVbYfTAm1fZQSXk+7eCqRqbKLp+KZMILg9/kJpiASNUhrC760qLS1C8z4dxuoqAh2EB0rOK1Iixu2tdU22Ke3tbYihyAGvgNN1Nz27b68FDkZD1bgPCKYEccs4esQxs1QO4QJBDLHeoWyPb52V/G0UHaOL+CJL5fwBggtXU7Vgz9F4vYe2lLoNwFD2zkf6vdr1m1E2t3js0l3sG5ttfmG9xkRaW8TSy/pkpjSAb0iaSZOhHff7bx9wgR48MHcx6QfwLRoNTW1i5hHzLa+8w4Ty8o6bt+2rf245mZbL34wfL78YncwEcxez/Vl2ENGP/ssXHKJTcv48bBoUd9WkxpjR1ZoabHXuaUF09KC09xEsvlA5jfZvB+nudkVjyi+1lYqozFqWqOumKQFxRWXeLzL0w73mhAMYsojOOURTHm5XSoiOGPHkMzaZsrLcSIRTHkEU1Fuf8vLccrLGXHZNfh37Ox0ruS4Mez8/f8g7jiOkrTjPEoiCakkkkxCMmV/UynPetINk7LHptfT4VPJ9viSSXbu2s/oqkhmP8mkDZNI2uMz52yPI3P+RAJfW5u7L31s+pweG71pSCaRXCNg9BJfNEbtl79G9W3fxlRWUhUMERpeg6mswKmsxFRWYCrKcSorMJWV7q9dnAr3N72tvLz/q+a8nZmzSYtYW5t9xh2HTEksFrMfz+mSWLpNbAB0eO4PVKjS3H57xzYqsCWYRYvyH+MVji7YtHYtE6dN6/r8sZh9WL1i5xXAXL/NzbbRd+vWjtu6QzjctcgdTACfe46j7723XWC3boUbb7TD9LzvfRlhySze/+n17N+s/aalpVMpRQC/u2TTLiZpgYjgVFdhxo7psM14w1VYITEe0dmwu5kjj5mEiURsA3ofcOAr13ZoowJwwmGabrgaM6w2t/d2H7JpUyORqWP7+SxZuO06GZFNi6JHeNOCnEsoR3zuC3mdE6MfOhNpaaVtx25CJolv9x4C725BmluQlhZ8bdE8R2aZWB7xiFhlu5BVlLcLX2ZbRUfhS4evKD+0wZ+7EjFfk90ei9n3wUnRwVUzGOzooZguiQ3S0tjgTNWhcOGF9rdYXn/pL6fhww8etiscx2b2XYlcU1O7KHr31dd3PC6R6PJUnV6vaBS+9rWD2xgIWLErL4eKCrdEUo4zdjSpSBmJcIhkpIxUJIxTXo5TEUbKK5CKCqSiCtLi4pZiTCSCiYT7zIU45mvE1FT3SVxp0u0qJdHeUijcdh3jKb33RJBT48cS2NaYc/v+W+0H5NpNjUzLJcDJpC1BN7cgzc32t6W1fb25JfMrLe3rvpYWArt3e45tseJ6EEzA316Cq2gXNm8JLvc2bymwElPhKeV15VHopOzEpdGo69xhS2KRJ56h6vsP4m/cYWuJbr+9PW8bwKhQebnwQrvs328HkR2I4/P5fO0lot4Si+UXuauuyn/cAw9kRIiKisy6EwmTiIRIBvzEUjHaUq6nnUki4sMYh4D48UuAoNvbfzDRds6CwS1MfUzT9VflLoVe38WzlyYQwNRUk+rtB0faSSYjZB3Fzgpec0dha27G19KCb89efFvqkeZmG64HpbwR4Qj+mqpO1ZVpMWvfVpkpAYZeWU3lDx/Cl67lePddW0sEA16sVKiU/KRLeSNGdN53xx22ui+bCRNwzj0nj+v3PkwMJEbG9bvMHyIig2eUZ6XvKIlSqIhbYo/AqJEcvGzVBelSXotX5LJLfFb49jXsYoQ4mRJfYMvejqXAbpTyAFttePPNKlTKEGXRIlJf/jJ+jxOIEw6z60uXsW//25lt6bHQgr4AYV9ZrpgUJS+DqhTag1Le+nxVmmCdMGKxrCrMZkZ89orcbXr5PJoHECpU+UgkbBVXINDegU9pZ+FCXq+v5+iHf46voZHk2NHsue4K4h/9EJU+7eSoKP2GCITDOOEwjByRKeXla9Nj4sCf7kOFKhdVVTB5crvHjeuaC7T3mUi7iw5Rosko9fNPI3jxRyj3t49tN3SviKJ0k8zUJcb6QHinMslMemnaJ70kK4zdQPaAjU1XXkbNN+/GF/V0dSkvtw4VA5yCCpWInA3ch83PfmKMuTNr/yTgIWAUsAf4jDGm3t2XAl5zg242xpzTb4b6fPYGl5fDsGF2W3qyuHjcCldawNL43fHDBql7aBpjDHtj+9jRtguAikB5kS1SlEOgO2Lh/Q/dEov27eTZR+cZk312ahPEM2Gkf58dJMA7K7J3tmTo+CtC279+HkaOoOqbd+Pf2oCo11/PERE/8ABwBlAPLBeRx40x6zzBvg383BjzMxH5IHAHcJG7r80Yc0Kh7O1Euo9CJAI1NXab41jxSiTaxcs7WsUgGyYlnorT0LqDaCpGVbASkZZim6QUAm/m3Gn0DZNztcvjsuPoMv6s7dlike6O0fkAuhwiPlssRKxgeMUivd0b/iBikfN/9r7u8M6OQ+qq0nbBP9P0yfMwxjCpdlKPjy9VCpmLngJsNMZsAhCRx4BzAa9QHQdc567/Ffh9Ae3rOT5fu2dcutNveqbSRMIW3Vtb7cvU1NT+sA6wdi9jDAfiTTS27SDoC1LV03HwlPZBTdNLTryZa3ZG29W+bsThONDSfAhx0DmTzX5uszPnTDhv/L78ceTK4LPjTh+TLRa7Wu2oKN5j+0oslJKhkEI1Adji+V8PzMkKsxo4H1s9eB5QJSIjjDG7gbCIrACSwJ3GmNIUMe/4XRUV1rX7rbdg6tT2qsPWVitiyWT7MemqwxJr90o6Sba37qQp0UJlsLzXo4gPWtLD4aSFKHv4oPTHSXl55/ucL2POpqtw+fal13dvgEmTDy2OUsbns7UcyqCmkELVVWVumi8D3xeRzwF/B7ZihQlgojFmm4hMBf4iIq8ZY97qcAKRy4HLAcaMGcOSJUv60PxDp7m5mSX/+EfunZnGU8cz8rKHIn4FOsYh4SQAcacp7zg8UzSWZO2mHF5GA5wu02XcNo1cYyyItwop3e6A57d4RFtjrF35enGN6AeiLVHWLl9bbDP6nN6kKz3T8tv+tw8ScuBQSKGqBw7z/K8DtnkDGGO2AQsBRKQSON8Ys9+zD2PMJhFZAswC3so6/kHgQYDZs2eb+fPn90c6esySJUvoti252r3i8XYBS4/pFQz2m4ClnBS7onvYG9tPeWAYAV/uxyTv8DUDFfeDYe07O5g2riZ/qSg9SOgAmndo7fK1TDv5IONNDkA0XZ1JOklto+oFy4EjRWQKtqT0KeDT3gAiMhLYY4xxgJuwHoCIyDCg1RgTc8O8F7i7gLYXjoO1e8ViVsDa2toz0j5s92pLttHQsgMHh+pQ14PtDkhythW5bTLpalvxWW/P7FlcB0iboqIMNgomVMaYpIhcBTyFdU9/yBizVkQWAyuMMY8D84E7RMRgq/6udA8/FvixiDiAD9tGta7TSQYr2e1eabwu82l3+fRAst52r254HjrGYU90L7uie4gEwgN3FIlutRUF2tuKcpWKGvZDbW3hbVeUXvLb9b/lzufvZFvTNibWTOT202/nwhnqnt4jjDFPAE9kbfu6Z/3XwK9zHPcCMKPfDRxo5HKZT6XaS19p8fJO/ZF2mQ8EMlWHsVSMhtYdxFNx1+28xBvSu1Mqyp7HR0tFyiDnt+t/y1ee+QptSdu/893973L5H+2gtANdrAZPJx/Fks6Q0/NNQcd2r2g00/ZlHId98QPsiO8lFAxTWVZeGt5eOUtFHpfp7pSKFGWIcefzd2ZEKk1ropWbn71ZhUoZAORo90ok4zTu20prW5AKM94OuxKNuhO0AUi7IPTH2H1d9SvSUpGiHBTHOGzYtYGX6l9iaf1StjblmM0A2LxfB6VVBiBNsSYamhrw+/xU1Y7uuDPd7pX2OoxGIemZR6e7/b26LBUZCAS1VKQoPSDpJFm3cx1L65fyYv2LLKtfxr7YPgAmVE0gEoh0KlEBTKzRQWmVAUTSSbKzZSf7Y/upCFbgz1VS8rZ7VbvTEaTbvVKp9qrDlpb24Wt8vs5OC7lKRV4h0lKRonRJIpUagGXNAAAgAElEQVRg9fbVvFj/Ii/Vv8Sybctojtv25sm1kzn7iLOZe9hc5k6Yy2E1h3VqowIoD5Zz++k6KK0yQGhNtNLQ1ABAdVkPZz31Tomd9jo0BnavhXHjrMt8uqSlpSJFOSSiySirGlextH4pz657lg1LN2RE58jhR3LeMecxt24ucybMYVzVuE7HLzx2IYB6/SkDD8c47G7dze623ZQHy/N23u0x6REz0qPMK4rSI1oTrazYtiJTYlrZuJJYKoYgTKmYwgXTL7DCVDeHkeUjuxXnwmMXcs7R52iHX2XgEE1G2da0jWQqSVWoqvTdzhVlENMUa2LZ1mW8tNU6P7y6/VWSThK/+JkxegafO+FzzK2byykTTmHra1sH5Ygbh4oK1SDEGMPe6F52tOwgHAhTWTYIR5hQlBJnT9selm9dztL6pby09SXW7FiDYxyCviAzx87kitlXMHfCXE6ecDKVWaPAbCW3B99QRYVqkBFPxWloaiCajGopSlEKyM6Wnby49UVe3PIiL219ifW71gMQ9oeZNW4W1865lrl1czlp3ElEgjrie09QoRokGGM4EDtAY3MjQX+QqrKqYpukKIOabU3beLH+xczy1l47RnZ5sJzZ42fz0aM/yry6ecwcM5OywAAdkqxEUKEaBCSdJNubt9Mcb6YiVKFzRilKH2OMYfP+zbbE5Do/vLv/XQCqQlWcMuEULph+AXPq5jBj9AyC/mCRLR5cqFANcJpjzTQ2NyIiWopSlD7CGMNbe9/KiNLS+qU0NNvuHcPCw5hbN5dLZl3CvLp5HDvy2Nx9EpU+Q4VqgJJyUuxq3cXetr2Uh/rQ7VxRhiCOcXh91+sZj7yX6l9iZ+tOAEaVj2Ju3dzMctSIo7TWosBo7jYAaUu00dDUQMqkqA73sPOuoiiknFSH4Yhe2voS+6J2OKLxVeM5bdJpzJ0wl7mHzWVq7VR1SioyKlQDCMc47Gndw67WXUSCEcL+cLFNUpQBQSKV4NXtr1rHh60vsnzrcpriTQBMrpnMWYefxdy6ucyrm0dddZ0KU4mhQjVAiCVjNDQ1EHfiVJWp27midEV6OKK0R96KbSs6DEd07jHnMq9uXt7hiJTSQoWqxDHGsD+2n+3N2wn5Q506BiqKYocjernh5UwfplcaXskMR3TMyGMyHnlz6+Z2ezgipXRQoSphEqkEjc2NtCZa1e1cGZJ4p1YfXzWeRacuYuGxC2mKNbF823L+39v/j40bN7J6+2qSThKf+JgxegYXn3Ax8+rmcfL4kxkWGVbsZCi9RIWqRGmKNdHY3IhPfOp2rgxJsqet2Nq0lev+dB13/+NutjZtxTEOfvFzwtgTuOKkK5hbN5fZ42fr+zIIUaEqMVJOih0tO7qeM0pRBjEt8RbW7VzHLX+5pdNEgEmTZHvLdq6dcy1z6uZQvq2ck+aeVCRLlUKhQlVCpOeMMsb0fM4oRRmA7G3by5qda1i7Yy2vbX+N13a8xqa9mzCYvMckUgm+/J4vA7B2+9pCmaoUERWqEsA7Z1QkENHhV5RByY6WHRkxWrtjLa/teI0tB7Zk9k+omsD00dM575jzmD5mOov+vIjG5sZO8YyvGl9Is5USQIWqyESTURqaGkikEjrauTIoMMawtWlrRpTSwrS9ZXsmzJTaKcwaN4uLjr+IGWNmMH30dIZHhneIp+m0pk5Tq0cCERaduqhgaVFKAxWqIqFzRimDAcc4bNq7KVNCem3Ha6zZvoZ9MTvKg1/8HDn8SE6bdBozRltBmjZqWrccHrKnVvd6/SlDCxWqIhBPxWlsbqQt0aalKGXAkEgleHPPmx2q7tbuWEtLogWAkD/EMSOP4cNHfZjpo6czY/QMjhl5TK/mXlp47EIVJkWFqpAYY2iKNdHQ3KBzRiklTTQZZcOuDbaUtN0K0vpd64mlYoCdc2naqGl8YtonbElpzHSOGn6Utq8q/YIKVYFIzxnVFG9St3OlpGiON7Nu57pMm9KaHWt4Y/cbpEwKgJqyGqaPns4lJ1xiS0pjZjCldoo+w0rBUKEqAI5xeGfvOyCo27lSVPa07WHNjjWZ5bUdr/H23rcz7uCjykcxY/QMzjj8jEyb0mHVh2n1tFJUCipUInI2cB/gB35ijLkza/8k4CFgFLAH+Iwxpt7ddzFwixv0NmPMzwpm+CGSnjMqnooTCoR0ziiloGxv3p5xcHjh9RfYvHoz9QfqM/vrquuYPmo6C49dyIzRM5gxegZjKscU0WJFyU3Bck4R8QMPAGcA9cByEXncGLPOE+zbwM+NMT8TkQ8CdwAXichw4BvAbMAAL7vH7i2U/T3FO2eU3+dXkVL6DWMMWw5syZSQ1my3v+mJ/wDqInWcNPEkLp55MdNHT8/pDq4MfBKpBG2JNmrCNcU2pU8pZO55CrDRGLMJQEQeA84FvEJ1HHCdu/5X4Pfu+lnAM8aYPe6xzwBnA48WwO4e4Z0zKhwM65xRSp+SclK8ve/tDu1Ja3asYX9sP2DdwY8acRTzJ8/PVN0dN+o4Nr+6mWknTyuy9Up/EU1GiSfjhANh6mrqKA+WF9ukPqWQQjUB2OL5Xw/MyQqzGjgfWz14HlAlIiPyHDsh+wQicjlwOcCYMWNYsmRJX9neLQyGRCqBwXQY6TzaEmXt8sE31Iumq39JOknebX2Xjc0b2di8kTdb3mRT8yaiThSAoASZUjGF9w57L0dUHsGRlUcyuXwyZf4yG0EKaIDNDZtLJk19zVBPl2McMODz+QhIABHhHd7pfwMLTCGFKldrbPaAXl8Gvi8inwP+DmwFkt08FmPMg8CDALNnzzbz58/vhbndJ3vOqLJAWYf9a5evHZRfs5qunpFvygqwVcXrd63v4OSwYdcG4qk40O4O/umpn2b6GNtH6cjhR3bbHVzv1cCiq3SlnBRtCTtaR22kltpwLSF/qJDmFZxCClU9cJjnfx2wzRvAGLMNWAggIpXA+caY/SJSD8zPOnZJfxrbXRKpBNtbttMca6ayrFLnjFJykmvKiuufup6HVz/MgdgB3tzzZsYdvLasluljpvP5Ez6fGV5I3cGVRCpBLBnD7/MzqmIUVWVVQ6btu5CpXA4cKSJTsCWlTwGf9gYQkZHAHmOMA9yE9QAEeAr4loikZ0A7091fVLxzRlWH1e1c6UwilWDDrg187a9f6zRlRcJJsKJhBfMnz+esI87KjOZQV12n7uBKhmgySiKVIOQPMa5q3JCcRLVgQmWMSYrIVVjR8QMPGWPWishiYIUx5nFsqekOETHYqr8r3WP3iMg3sWIHsDjtWFEMMnNGRfdTEdLOu4rFGEP9gXpeaXyFlQ0rWdm4kjXb1xBNRbs85uHzHi6glcpAoS3RRtJJUhGsYFzlOMKB8JD9gCloudEY8wTwRNa2r3vWfw38Os+xD9FewioaHeaM0lLUkGZ/dD+rt6/mlYZXWNm4klWNq9jVuguAsD/M9DHT+czMz3Di2BNZ/LfFNLbolBVK1zjGoS3RhuM4VJVVUVNW06nNeygyNCo4+wCdM2pok0glWL9rfQdR2rhnY2b/4cMOZ/7k+cwaO4sTx53IsSOP7fCMpExKp6xQ8pJ0krQl2vCLn5HlI9ka2MroitHFNqtkUKHqBjpn1NDCGMPm/ZtZ2bgyZxXeiMgIZo2bxXnHnMeJ405k5piZB+1gqVNWKLmIJWPEUjHb/lQ5Th2y8qBC1QXGGPZF97G9ZTtl/jKdM2qQkl2Ft2LLCvY9b+dTSlfhXTTzImaNncWscbMOeew7nbJCAZuvpB0kKkOVjKkcQyQQ0Q/gLlChykM8FWd783ZaEi1UhvQrZ7CQXYW3smElb+19K7P/iOFHMHvYbD4w/QM5q/AU5VDJtD8Zh9qw7f+k7U/dQ4UqB9FklM37NuP3+XW08wFMegy8lQ0reaXxFVY1rspZhbfw2IUdqvDWLl/LtBMGXydSpTh4259GlI+guqx6yPR/6iv0auUglowhIr2amVQpPPuj+1nVuCojSisbVrK7bTeQVYU3bhYnjj1R+ysp/UosGSOeihP0B7X9qZeoUCkDkngqzoZdG7qswvvglA9mROmYkcdoFZ7S76Tbn5JOkkgwwmGVh2n7Ux+gQqWUPNlVeCsbVrJmx5rMtOgjy0cya2znKjxFKRSOcYgmojg4VIeqGRYZpu1PfYgKlVJyeKvwVjbYPkvZVXifnflZrcJTik7SSRJNRhGE4ZHhVJdVa8m9H1ChUopKPBVn/c71tvpOq/CUAUI8FSeWjBH0BRlbMVaHUutnVKiUgpHuSOstLeWqwjv/uPOZNXaWVuEpJUWm/5OToDxYTl21naBQS/P9jwqVcsg8u+NZLvnPS/KOtLAvuo/VjavzVuHNGDODi0+42HakHTtLq/CUksTb/6mmrIbaSC3hgM7cXUhUqJRD4rfrf8u9b95LzLGloa1NW/ny01/m7+/+Hcc4rGxcyaa9mzLhjxx+pFbhKQOKlJOiLdmGIAwLD6MmXKPPbJFQoVIOiTufvzMjUmliqRi/WverTBXex4/7OLPGzuKEsSdox2llwBBPxYkmooT8IcZUjKEyVKntT0VGhUo5JLY1bcu5XRBW/esqrcJTBhzp+Z/CgTCH1Rym7U8lhAqV0mP++vZfERGMMZ32ja8ary+3MmAwxtCaaMUxdv6n4ZHh2v5UgqhQKd2mLdHGt577Fg+teoixFWPZ07qHuIln9uv8SspAIeWkaEu0IWLbn6rD1YT8oWKbpeRBB55SusWaHWtY8MgCHlr1EJfOupTnP/881x11HROqJiAIE6omcPcZd+s0FkpJE0/FaYo1EU/FGVM5hqnDpjKyYqSKVImjJSqlSxzj8OMVP+auf9zFsMgwHln4CPMnzwfg9NGnc82HrymugYrSDdLzP5X5y7T/0wBEhUrJy9amrVz75LUsrV/KgiMWcPcZdzM8MrzYZilKtzDG0Ja0DhLVZdWMqxynMyIMUFSolJz8YcMfWPTsIpJOknvOvIdPTvukfoEqA4J0+xNgx9/T9qcBjwqV0oEDsQPc/OzN/HbDbzlx3Incv+B+JtdOLrZZinJQEqkE0WSUgC/AqIpRVJdVa/+nQYIKlZLhxfoXuebJa2hsbuSGeTdwzZxrdCZSpeSJJqPEk3HCgTATqidQHizXCQoHGZoLKcRTce554R4eWP4Ak2om8btP/o6Txp9UbLMUJS/p9ifHcSjzlzGuchzhQFirpwcpKlRDnI17NnLVE1fx2o7XuGD6Bdw6/1YqQ5XFNktRcuJtf6qN1BIKhJhQPaHIVin9jQrVEMUYw89f/TmL/7aYSCDCTz76ExYcuaDYZilKThKpBLFkDL/Pz6iKUVSVVRHwBVjHumKbphQAFaohyM6WnVz/9PX85e2/8P5J7+c7Z32HsZVji22WonQgM/9TKkFZoIxxVeOoCFVo+9MQpKBCJSJnA/cBfuAnxpg7s/ZPBH4G1LphFhljnhCRycB64HU36IvGmCsKZfdg4um3nubLT3+Z5ngzi+cv5pJZl+iLr5QUSSdJNBHFYKgpq6Gmqkbbn4Y4BxUqETkD+ATwgDFmlYhcbox5sKcnEhE/8ABwBlAPLBeRx40x3rL7LcAvjTE/FJHjgCeAye6+t4wxJ/T0vIqlNdHKv//t3/nFq7/guFHH8asFv+LokUcX2yxFAWzpKZaKkUglCPqCjKkcQ0WoQr1OFaB7JaovApcAt4jIcOBQxeIUYKMxZhOAiDwGnAsdKpkNkJ64qAbIPZeE0iNWN67mqiev4u29b3PFSVfwlfd+hbJAWbHNUhRSTopo0paeKkOV6r2n5ERyTdXQIYDIg8aYy931O4HTjTEn9/hEIh8HzjbG/Iv7/yJgjjHmKk+YccDTwDCgAvgnY8zLbtXfWuAN4ABwizHmuRznuBy4HGDMmDEnPfbYYz01E4CUSZF0kn1WJRZtiRKuKPzUASmT4n+3/C8Pb36YYcFh3Hj0jcyqndVn8RcrXf3NYExXqaXJGIMxBhHB7/Pjl0PrmNvc3Exl5eDzUi1muj7wgQ+8bIyZXZST56E7Jar/AxARH9AM/PwQz5XrEylbJS8AfmqMuUdE5gEPi8h0oAGYaIzZLSInAb8XkWnGmAMdIrNVkg8CzJ4928yfP/+QDN0f3c+Olh1UhCoO6fhs1i5fy7STp/VJXN1l8/7NXPPkNSzftpxzjj6HO06/g9pwbZ+eoxjpKgSDMV2lkKZ06Sk991NtuJZIINKr0tOSJUs41Pe8lBms6TpUDipUxpg/uL+OiHzAGHP6IZ6rHjjM87+OzlV7lwJnu+dbKiJhYKQxZgcQc7e/LCJvAUcBKw7RlkGLMYZfr/81t/zlFgThe2d/j4XHLtSqFKVoxJIx4qk4AV+AkeUjqQxVEvQHi22WMoDoad3WKhH5hlu66inLgSNFZIqIhIBPAY9nhdkMnA4gIscCYWCniIxynTEQkanAkcCmQ7BhULO3bS9f+L8v8KU/fYlpo6bxzEXPcP5x56tIKQXHMQ4t8RYOxA4Q9AWpq65j6rCpDIsMU5FSekxPXWoOA2YAXxCRl4BXgVeNMb862IHGmKSIXAU8hXU9f8gYs1ZEFgMrjDGPAzcA/yki12GrBT9njDEi8j5gsYgkgRRwhTFmTw9tH9Q8t/k5vvSnL7GrdReLTl3EF2d/UQfkVApOPBW3HXPFz/DIcKrKqnTkcqXX9EiojDGfABCRMmAaVrROAQ4qVO7xT2Bdzr3bvu5ZXwe8N8dxvwF+0xNbhwqxZIy7/nEXP375x0wdNpXHP/U4M8fOLLZZyhDCMQ7RZJSUkyISjFBXXUckGNH+eUqfcUidFIwxMeAVd1GKxIZdG7jqiatYv2s9n535Wb72vq9RHiwvtlnKECGeihNPxvGJj9pwrc77pPQb2ptuAOIYh4dWPsS3nvsWlaFKfvqxn3LG1DOKbZYyBEiPWp4uPY2vHq/Taij9jgrVAKOxuZHrnrqOv7/7d06fcjr3nHkPoypGFdssZZCTnpTQJz6GhYdRVValncaVgqFCNYB44s0nuPGZG4kmo9xx+h1cdPxF6tGn9BvpQWGTTpIyfxnjq8broLBKUVChGgA0x5v5xl+/wWNrH+P4Mcdz/4L7OWL4EcU2SxmkJFIJYqkYglATrqG6rJpwoHRGtVCGHipUJc6KbSu45slr2Lx/M1efcjXXz7teG6yVPie79DS2YiwVoQrt4qCUBCpUJUrSSXLfi/dx30v3Ma5qHL/5xG+YUzen2GYpg4z0lBoA1WXV1EZqtfSklBwqVCXI23vf5uonr2Zl40rOP/Z8bvvgbVSXVR/8QEXpBt7SU9Bvp9SoDFVq6UkpWVSoSghjDI+ueZRvLPkGQV+QH3z4B5x79LnFNksZJKScFG2JtvYJCcM6IaEyMFChKhH2tO3hxqdv5E9v/Yn3HvZe7j37XsZXjS+2WcogIJqM4jgOiVRCJyRUBiT6tJYAS95ZwnVPXce+6D6+9r6vcflJl6sLsNIrsickDPlDTBk2RUtPyoBEhaqItCXa+NZz3+KhVQ9x9Iij+cXCXzBt1OCaB0kpLNFkNDOdu3dKjTfkDRUpZcCiQlUk1uxYw9VPXs0bu9/g0lmXctOpNxEJRoptljIAcYxDW6INxzhUBCsYWzm21xMSKkopoUJVYBzj8ODLD3Ln83cyLDKMRxY+wvzJ84ttljIA8U5IOKJ8BFWhKp3rSRmUqFAVkK1NW/nSn77EC1teYMERC7j7jLsZHhlebLOUAUR26Wl0xWjKg+VaelIGNSpUBeIPG/7ATc/eRMJJcM+Z9/DJaZ/UzEXpNjohoTKUUaHqZw7EDnDX63fx7I5nOXHcidy/4H4m104utlnKAEAnJFQUiwpVP/JS/Utc86dr2HZgGzfMu4Fr5lyj/VeUg6ITEipKRzTX7AfiqTj3LL2HB5Y9wKSaSXx35nf5+LyPF9sspYTRCQkVJT8qVH3Mxj0bufrJq3l1+6tcMP0Cbp1/K++ufrfYZiklSvaUGjVlNTohoaJkoULVRxhj+PmrP2fx3xYTCUT4yUd/woIjFxTbLKUE0Sk1FKVnqFD1ATtbdnLD0zfw7NvP8v5J7+c7Z32HsZVji22WUiIYY0g4CRKpBI5x8IlPJyRUlB6gQtVLntn0DDc8dQPN8WYWz1/MJbMu0XYFhUQqQTwVxzEOIkJ5oJxhFcMIB8KE/CHtmqAoPUCF6hBpTbSy+G+LefjVhzlu1HH8asGvOHrk0cU2SykSiVSChJMg5aQQhHAwzMjykUSCEUL+kH68KEovUKE6BFY3ruaqJ6/i7b1vc8VJV/CV935FG8CHGEknmSkxYaAsUMbwyHAigQhlgTIVJkXpQ1SoekDKSfHA8ge4Z+k9jCofxWMff4xTJ55abLOUApByUsRTcVJOCgSCviDDwsOIBCOU+cvUEUJR+hEVqm6yZf8WrvnTNSzbuoxzjj6HO06/g9pwbbHNUvoJxzjEkjGSThJBCPgDVJdVUxGqIOQPacdtRSkgBX3bRORs4D7AD/zEGHNn1v6JwM+AWjfMImPME+6+m4BLgRRwjTHmqULYbIzhN+t/w81/uRlB+N7Z32PhsQu1MXyQ4RiHeCpO0kniOA7xZJyqsioqghWUBcpUmBSliBTs7RMRP/AAcAZQDywXkceNMes8wW4BfmmM+aGIHAc8AUx21z8FTAPGA38WkaOMMam+tPGR1x7h5mdvZvP+zYyrGse1c67l+c3P88c3/sicCXO47+z7OKzmsL48pVIkjDHEU3ESqQQGg9/npzJYSWVZJVsDW5k6fGqxTVQUxaWQn4mnABuNMZsAROQx4FzAK1QGqHbXa4Bt7vq5wGPGmBjwtohsdONb2lfGPfLaI1z+x8tpTbQCsK1pG1/981cRhEWnLuKLs7+o7RADmLQwJZ0kxhhEhMpQJSMiIwgHwwR9wUwpWdDSsqKUEoUUqgnAFs//emBOVphbgadF5GqgAvgnz7EvZh07IfsEInI5cDnAmDFjWLJkSbeNu+HFGzIi5aU2WMsH5YNseHlDt+PKJtoSZe3ytYd8fKlS6ukyxmAwmf8+8WWWrsSoubm5R8/OQGAwpgk0XUOFQgpVrpzBZP2/APipMeYeEZkHPCwi07t5LMaYB4EHAWbPnm3mz5/fbeN2/G1Hzu37EvuYdvK0bseTi7XL1/Y6jlKk1NKV7mRrjAGBSCBCdVk1ZYEyyvxl3W5XXLJkCT15dgYCgzFNoOkaKhRSqOoBbwNPHe1Ve2kuBc4GMMYsFZEwMLKbx/aKiTUTeXd/58Fjx1eN78vTKH1Iui+TdrJVlMFNId/k5cCRIjJFREJY54jHs8JsBk4HEJFjgTCw0w33KREpE5EpwJHAsr407vbTb6c8WN5hWyQQYdGpi/ryNEovSDkp2hJtNMWaaIo14TgOwyPDmVgzkcOHH87EmokMi9hhilSkFGXwULASlTEmKSJXAU9hXc8fMsasFZHFwApjzOPADcB/ish12Kq9zxljDLBWRH6JdbxIAlf2tcffhTMuBOjg9XfTqTex8NiFfXkapQd4O9kaDCF/iJqyGspD5drJVlGGEAXtHOL2iXoia9vXPevrgPfmOfZ24Pb+tO/CGRdy4YwL2R/dz46WHVSEKvrzdEoW6U62KZMCAwGf7WRbHizXvkyKMoTRN18pGt5OtgABCVAZqqQiVEGZv4ygP1hkCxVFKQVUqJSC4e1kC+Dz+agIVlBVVkXIHyLkDxXZQkVRShEVKqXfyDVhYEWoImcnW0VRlHyoUCl9indYIkAnDFQUpdeoUCm9pi3RZqe/ACLBCLXltYSDYe3LpChKn6BCpRw6BvZH91MbrqU2XKsTBiqK0i+oUCmHRFuiDcc41FXXUVVWVWxzFEUZxOjnr9IjjDE0xZqsl14gpCKlKEq/o0KldJt4Kk5zvJlRFaOoq67T6TAURSkIWvWndIvWeCs+8TGxZiKRYKTY5iiKMoRQoVK6JOWkaIm3MCwyjJHlI3V8PUVRCo4KlZKXtNv5hOoJ2halKErRUKFSOmGMoTneTHmwnLE1Y3XMPUVRiooKldKBeCpOLBljVMUohoWH6UgSiqIUHRUqJUNLvAW/+JlUO4lwIFxscxRFUQAVKgV1mFAUpbRRoRripEeYUIcJRVFKFRWqIUoHh4lKdZhQFKV0UaEagsRTcaLJKKMrRqvDhKIoJY8K1RAj7TAxuXayOkwoijIgUKEaIqjDhKIoAxUVqiGAOkwoijKQUaEaxDjGoSXeog4TiqIMaFSoBilph4kxFWOoDdeqw4SiKAMWFapBiDpMKIoymFChGkSow4SiKIMRFapBgjpMKIoyWCnoVPQicraIvC4iG0VkUY793xWRVe7yhojs8+xLefY9Xki7SxnHODTFmgj5Q0yunawipSjKoKNgJSoR8QMPAGcA9cByEXncGLMuHcYYc50n/NXALE8UbcaYEwpl70BAHSYURRkKFLLq7xRgozFmE4CIPAacC6zLE/4C4Bt9aUAikaC+vp5oNNplOMc4pJwUMYn1yXlHDBvBns17+iSuNI5xEBH84qdRGmmksU/j7w41NTWsX7++4Oftbw6WrnA4TF1dHcGguvsrSiEopFBNALZ4/tcDc3IFFJFJwBTgL57NYRFZASSBO40xv++pAfX19VRVVTF58uQuSx9JJ0nSSeKTvqkZjbZECVf0jfedMQbHOAR8AQK+QFFLUU1NTVRVDb6qxq7SZYxh9+7d1NfXM2XKlAJbpihDk0IKVa4c1eQJ+yng18aYlGfbRGPMNhGZCvxFRF4zxrzV4QQilwOXA4wZM4YlS5Z0iLSmpoYRI0bQ3NzcpaHGGAwGyWlyz3Ech2hL16W47mDcyyUISUn2Or7ekkqlaGpqKrYZfc7B0hUKhdi3b1+n56uUaW5uHlD2dhdN19CgkEJVDxzm+V8HbMsT9lPAld4Nxpht7u8mEVmCbb96KyvMg8CDALNnzzbz58/vEMHHd1AAAB1MSURBVOn69euprq4+qKGlWKJyHAef+Aj4A31mV28ZiiWqNOFwmFmzZnUZppRYsmQJ2e/DYEDTNTQoZI63HDhSRKaISAgrRp2890TkaGAYsNSzbZiIlLnrI4H3kr9ta1BhjCHlpAj4AwT9wZIRKUVRlEJRsFzPGJMErgKeAtYDvzTGrBWRxSJyjifoBcBjxhhvteCxwAoRWQ38FdtG1e9C5fufRwlOPZJgMExw6pH4/ufRXsW3e/duTj7pZE4+6WQmTpjIlIlTMv/j8Xin8I5xACgLlGXaoy655BJef/31Ls/zwAMP8Mgjj/TKVkVRlFKhoB1+jTFPAE9kbft61v9bcxz3AjCjX43LwvfoY/ivuBJpbbUbNm/Gf8UXAXA+fcEhxTlixAiWv7wcgG/++zepqKzg+huu7xDGGIPjOCDkdJj47//+74Oe58orrzxomGJgjMEYg8+npUJFUbrP0M0xvvQlmD8/5+L/4OkEL7uiXaRcpLUV/2X/SuCDZ+Rc/NffcEimbNy4kVkzZ3HlF69kzuw5bGvYxjVfvIZ5c+Yxffp0Fi9enAl76qmnsmrVKpLJJLW1tSxatIiZM2cyb948duzYAcAtt9zCvffemwm/aNEiTjnlFI4++mheeOEFAFpaWjj//POZOXMmF1xwAbNnz2bVqlWdbLvxxhs57rjjOP744/nqV78KQGNjI+eeey7z5s1j5syZvPTSSwDcfffdTJ8+nenTp3P//fdn0jZ9+nSuuOIKTjzxRBoaGnjyySeZN28eJ554Ip/85CdpaWk5pOumKMrQYOgK1cGI5elDlW97L1m/bj0Xf+5ilq1YxpSJU7jrrrtYsWIFq1ev5plnnmHdus41nfv37+f9738/q1evZt68eTz00EM54zbGsGzZMv7jP/4jI3r3338/Y8eOZfXq1SxatIiVK1d2Om779u088cQTrF27lldffZWbbroJsCW2M844g6VLl/Lyyy9z7LHHsmzZMh555BGWLVvG0qVL+cEPfsCrr74KwLp167j00ktZuXIlwWCQO++8k2effZZXXnmF448/nvvuu6+vLqOiKIOQoTvWn1viyEXKSeKbMhXZvKXzzokTSf7lmb61xcDUw6cyd85c/OJHRHj00Uf5r//6L5LJJNu2bWPdunUcd9xxHQ6LRCIsWLAAgJNOOonnnnsuZ/QLFy7MhHnnnXcAeP755zMlpJkzZzJt2rROxw0fPhyfz8dll13Ghz/8YT7ykY8A1iPpscceIxqNEggEqK6u5rnnnuP888+nvLwcgI997GM8//zznHnmmRx++OGcfPLJALzwwgusW7eO97znPQDE43FOPfXU3lw9RVEGOUNXqA5C8rbFBL1tVIApLyd12+Iujuo5jmMdJiorKgn47O148803ue+++1i2bBm1tbV85jOfyTmaRigUyqz7/X6Sydx9q8rKyjqF6eirkptgMMiKFSt45plneOyxx/jhD3/I008/DdCpo3FX8VVUVHQId/bZZ/Pwww8f9PyKoiigVX95cS74FKkf/QAzcSJGBDNxIqkf/eCQHSmyMVjHAr/P32nm3QMHDlBVVUV1dTUNDQ089dRTfXJOL6eeeiq//OUvAXjttddyVi02NTVx4MABPvKRj/Dd7343Uz34gQ98gB/96EeA7Rx74MAB3ve+9/G73/2OtrY2mpub+cMf/sBpp53WKc73vOc9/O1vf2PTpk2AbSt78803+zx9iqIMHrRE1QXOpy/oM2HqEK/rdp4WqezSyYknnshxxx3H9On/v71zj6+qOvP+90cSCCgCXoZBwYEo+nIJBkyRakHFvghoVZAWMjigY1DqpVC1gjPWjryWIuVFhFoEEXTUD9iCRGUGqci1lUuIHiAiFBw75TKoQEHubcIzf6yVw+FwEgLmchLW9/PZn7P22mvv9Tz7rL2fvdZe+3nak5GRwXXXXVfhMjz88MMMGjSIDh060KlTJ9q3b0+jRo1OKLNv3z769u3L0aNHOXbsGOPHjwfgV7/6FUOGDGHy5MnUrVuXKVOm0LlzZ3JycqJDfD/84Q/JzMxky5YtJxyzadOmvPzyy/Tv3z86JX/06NG0bt26wnUMBAK1A5VnCKgmkp2dbWvWrDkh79NPP6VNmzan3LcyPVMUHyt2BqrOyQaqKikqKqKoqIj09HQ2b95Mjx492Lx5M6mp5X92OZs9U5S3LSULtdXTQdCr4pFUYGbZ1VJ5KYQeVRVR8g1RWkpadMJEdXLgwAFuuukmioqKMDOmTJlyWkYqEAgEqopwZ6oCSpzJ1k2tmzQukBo3bkxBQUF1ixEIBAKnJBiqSqQkJIcQdVPqVnsvKhAIBGoiyfF4Xws5ZscwLGqggpEKBAKBMyP0qCqBZJkwEQgEArWBYKgqkGSbMBEIBAK1gTD0VwYzC2fSelJr0p9Jp/Wk1swsLD3MR4mHibqpdcsMEb9z504GDBjAZZddRtu2benduzd//OMfK0X+b0rLli3ZtWsXQNTlUTx33303s2fPLvM4r7zyCjt2HI+RmZubm/AD40AgEEhE6FGVwqzCWTz4nw9y6G/OhdKf9/2ZB/7DhfnIaX/8I+CSCROJQnLEY2b06dOHwYMHM2vWLAAikQhffPEFV1xxRbRccXExKSkplaHWGVPidf1MeOWVV2jfvj0XX3wxANOmTasosSqUoqKiMEU/EEhCztqrcvh7w4nsPDmsBbjp5Ku2reJo8Yme0g/97RD3v3s/0z+aHi0Hzu+dEFl/n8WEnqU7u128eDFpaWkMHTo0mpeVlQW4D/yefvppmjVrRiQSYcOGDYwfPz7qET03N5fhw4dz8OBBfvCDH7Bt2zaKi4v56U9/Sv/+/Rk5ciTvvPMOqamp9OjRg3Hjxp1Q9+TJk/n8888ZO3Ys4IxHQUEBkyZN4o477mDr1q0cOXKEYcOGcd99950k+7nnnsuBAwcwMx5++GEWLVpEixYtTjCoo0aN4t133+Xw4cNce+21TJkyhTlz5rBmzRoGDhxI/fr1WbFiBb169WLcuHFkZ2czc+ZMRo8ejZlxyy238Oyzz0brGzZsGPPmzaN+/fq8/fbbNG3a9ASZli5dyrBhw6L/wbJly2jYsCFjx47ltddeo06dOvTq1YsxY8YQiUQYOnQohw4d4rLLLmP69Ok0adKEG264gWuvvZY//OEP3HbbbQwaNIjc3NxoD3DChAmV4hkkEAiUnzD0VwrxRio+3zCEqKM6iPK9iyosLOTqq68udfvq1av5+c9/zoYNGygoKGDGjBmsWrWKlStX8tJLL/Hxxx/z3nvvcfHFF7N27VoKCwvp2bMne/bsYe7cudFwHE8++eRJx+7Xrx9vvfVWdP3NN9+kf//+AEyfPp2CggLWrFnDxIkT2b17d6kyzp07l02bNrF+/XomTZp0Qk/roYceIj8/n8LCQg4fPsy8efPo168f2dnZvPHGG0QiEerXrx8tv2PHDkaMGMGiRYuIRCLk5+eTl5cHOB+AXbp0Ye3atXTr1o2XXnrpJFnGjRvHCy+8QCQSYfny5dSvX5/58+eTl5fHqlWrWLt2LY8//jgAgwYN4tlnn2XdunVkZmby9NNPR4+zd+9eli5dyqOPPsqwYcN48MEHyc/PZ86cOeTm5pZ6LgKBQNVw1vaoyur5FB0rIuP5DLZ+fXKYj0vPu5QFdy0gNSW1widMdO7cmVatWgEuDEefPn2insf79u3L8uXL6dmzJ4899hgjRozg1ltvpWvXrlFXSLm5uSeE44jloosuIiMjg5UrV9K6dWs2bdoU7SlMnDiRuXPnArB161Y2b97MBRdckFDGZcuWkZOTQ0pKCs2aNaN79+7RbYsXL2bs2LEcOnSIPXv20K5dO773ve+Vqm9+fj433HADF110EQADBw5k2bJl3HHHHdStWzeqx9VXX837758cWuW6667jkUceYeDAgfTt25fmzZuzcOFC7rnnnmi4kfPPP599+/axd+9err/+egAGDx7M97///ehxSgw2wMKFCyksLIxGIf76669rrauoQKCmEHpUpTDqxlE0SGtwQl6D1AaMunHUKSdMlEa7du3K9AYRHw4jEVdccQUFBQVkZmbyxBNPMGrUKFJTU1m9ejV33nkneXl59OzZk+LiYrKyssjKyuKpp54C3A35N7/5DXPmzKFPnz5IYsmSJSxcuJAVK1awdu1aOnbsmDCkSCyJ9D5y5AgPPPAAs2fPZv369QwZMuSUxynLz2Ra2vGp/aWFMBk5ciTTpk3j8OHDdOnShY0bN2Jmp/2/xJ73Y8eOsXDhQiKRCJFIhO3btwcjFQhUM8FQlcKA9gP49S2/5tJGlyJEi/NaMPnWyQy6atAZu0Hq3r07R48ePWEYKz8/n6VLl55Utlu3buTl5XHo0CEOHjzI3Llz6dq1Kzt27KBBgwbcddddPPbYY3z00UccOHCAffv20bt3byZMmEAkEiElJSV6sy2J6tu3b1/y8vKYOXNmtBexb98+mjRpQoMGDdi4cSMrV64sU4du3boxa9YsiouL2blzJ4sXLwaIGqULL7yQAwcOnDATsGHDhuzfv/+kY11zzTUsXbqUXbt2UVxczMyZM6O9nvLw2WefkZmZyYgRI8jOzmbjxo306NGD6dOnc8jHEduzZw+NGjWiSZMm0cCSr732Wqn19OjRg6lTp0bXI5HE7zEDgUDVcdYO/ZWHnPY59G/nbuhpddJIqfPNZuJJYu7cuQwfPpwxY8aQnp5Oy5YtmTBhAtu3bz+hbKdOnbj77rvp3Lkz4CZTdOzYkQULFvCTn/yEOnXqkJaWxuTJk9m/fz+33347R44cwcx47rnnEtbfpEkT2rZty4YNG6LH7dmzJy+++CIdOnTgyiuvpEuXLmXq0KdPHxYtWkRmZiYZGRnRG37jxo0ZMmQImZmZtGzZMhruA9wU9qFDh0YnU5TQrFkzfvGLX3DjjTdiZvTu3Zvbb7+93OdzwoQJLF68mJSUFNq2bUuvXr2oV68ekUiE7Oxs6tatS+/evRk9ejSvvvpqdDJFRkYGM2bMSHjMiRMncv/999OhQweKioro1q1bNPZWIBCoHkKYjwQUHSuiqLgIwyrEw0RtfcdxNusVwnwkB0GviieE+ahBGMHDRCAQCCQDwVAloI7qUC+1XtKE5AgEAoGzmbPOUJVnVlgwUIGyqK3D5YFAsnJW3ZHT09PZvXt3uNEEzhgzY/fu3aSnp1e3KIHAWcNZ1aNq3rw527Zt46uvvqrSeo8cOVIrb2xnq17p6ek0b968CiUKBM5uzipDlZaWFvX8UJUsWbKEjh07Vnm9lU3QKxAIVAVVOvQnqaekTZK2SBqZYPtzkiJ++aOkvTHbBkva7JfBVSl3IBAIBKqPKutRSUoBXgD+L7ANyJf0jplFAxOZ2Y9jyj8MdPTp84GfAdmAAQV+379UlfyBQCAQqB6qskfVGdhiZv9lZn8FZgFluSHIAUoiFd4MvG9me7xxeh/oWanSBgKBQCApqMp3VJcAse7ItwHXJCoo6R+AVsCiMva9JMF+9wElwZQOSNr0DWWuKC4EdlW3EJVA0KvmUBt1gqBXZfAP1VRvqVSloUr08VJp88QHALPNrPh09jWzqcDUBGWrFUlrks0lSUUQ9Ko51EadIOh1tlCVQ3/bgBYx682BHaWUHcDxYb/T3TcQCAQCtYiqNFT5QGtJrSTVxRmjd+ILSboSaAKsiMleAPSQ1ERSE6CHzwsEAoFALafKhv7MrEjSQzgDkwJMN7NPJI0C1phZidHKAWZZjPsIM9sj6f/hjB3AKDPbU1WyVwBJNxxZQQS9ag61UScIep0V1NowH4FAIBCoHZxVvv4CgUAgUPMIhioQCAQCSU0wVBWApOmSvpRUGJN3vqT3vcun9/0kEOSY6N1IrZPUqfokLx1JLSQtlvSppE8kDfP5NV2vdEmrJa31ej3t81tJWuX1etNP+EFSPb++xW9vWZ3yl4WkFEkfS5rn12uDTn+StN67VVvj82p0GwSQ1FjSbEkb/TX27dqgV2URDFXF8Aone8oYCXxgZq2BD/w6QC+gtV/uAyZXkYynSxHwqJm1AboAD0pqS83X6yjQ3cyuArKAnpK6AM8Cz3m9/gLc68vfC/zFzC4HnvPlkpVhwKcx67VBJ4AbzSwr5ruimt4GAZ4H3jOz/wNchfvfaoNelYOZhaUCFqAlUBizvglo5tPNgE0+PQXISVQumRfgbZyfxlqjF9AA+AjnIWUXkOrzvw0s8OkFwLd9OtWXU3XLnkCX5ribW3dgHu4j+Rqtk5fvT8CFcXk1ug0C5wGfx5/zmq5XZS6hR1V5NDWz/wHwv3/n88vlDiqZ8ENDHYFV1AK9/BBZBPgS5zfyM2CvmRX5IrGyR/Xy2/cBF1StxOViAvA4cMyvX0DN1wmcB5rfSSrwLtKg5rfBDOArYIYfqp0m6Rxqvl6VRjBUVc/puJKqdiSdC8wBhpvZ12UVTZCXlHqZWbGZZeF6IZ2BNomK+d+k10vSrcCXZlYQm52gaI3RKYbrzKwTbvjrQUndyihbU/RKBToBk82sI3CQ48N8iagpelUawVBVHl9Iagbgf7/0+TXGHZSkNJyResPM3vLZNV6vEsxsL7AE9w6usaSSD+BjZY/q5bc3ApLtY/PrgNsk/QkXlaA7rodVk3UCwMx2+N8vgbm4B4ua3ga3AdvMbJVfn40zXDVdr0ojGKrK4x2gJMDjYNw7npL8QX4mTxdgX0l3P5mQJOBl4FMzGx+zqabrdZGkxj5dH/gu7kX2YqCfLxavV4m+/YBF5l8UJAtm9oSZNTezljjXZIvMbCA1WCcASedIaliSxrlOK6SGt0Ez2wlslXMXB3ATsIEarlelUt0vyWrDgnOg+z/A33BPP/fixvw/ADb73/N9WeECSH4GrAeyq1v+UnT6Dm54YR0Q8UvvWqBXB+Bjr1ch8JTPzwBWA1uA3wL1fH66X9/it2dUtw6n0O8GYF5t0MnLv9YvnwD/6vNrdBv0smYBa3w7zMP5N63xelXWElwoBQKBQCCpCUN/gUAgEEhqgqEKBAKBQFITDFUgEAgEkppgqAKBQCCQ1ARDFQgEAoGkJhiqCkbSBd7Tc0TSTknbY9brlvMYM2K+sSitzIOSBlaM1MmBpN9LyqrE478u6Y4KPF6KpOU+nSFpwDc83in1l/SIpPRvUEcbSUt9e/xU0mSf/11JeaXs01RSkaR74/LPk/SSpM/kPNEvkfSt05DlmbjrI1Ly3VRNQdICSQ295/OhMfln1B4quo3WFoKhqmDMbLc5T89ZwIs479VZfvkrRN32l3ruzeweM9t0inpeMLM3Klb6wOlgzhVTV7+agfvYtrJ5BPcd1JnyK2Csb59tgV+XY5/+wAogJy5/OrATuNzM2gG5wIWnKc8vY66PLDPbH7sxxrNGwvXSONU1VlGY2c1e5vOBoTGbqqo9nBUEQ1VFSLpcUqGkF3Eeu5tJmippjX8afSqm7O8lZUlKlbRX0hi5+EkrJP2dL/OMpOEx5cfIxVnaJOlan3+OpDl+35m+rpOe2CV9yz9lF0ia75+g0/z6d3yZX+p47KanJeWX6CNJMXKMl7Rc0gZJ2ZLmysXX+beY8/CJpNfk4gz9Rs5DRLxMvby+H8nFTjonRo4NcnF5ygxPIamOpF/78u8ScxNNpPMpzmWm1zni684o+X/8IccAN/rtP5L0oaT2MfWtktQuTr4Gkn7rjzeLGAOUqG1I+jHOUelySQtLK3cKmuE+Sscc68uxTw4wHMiQ9Pe+3itxH63+zPzHmGa2xczml+N4ZSIpV9Isubha8+V6ewv9OfrYl3nct79CSQ/7vPhrrEVMOyuU9KMEdbWSi7u2Ti4GVHOf/7qkF/y2zyR1k/SqXPyol2P23ybn6WQMcKX//8dwcntI9dfGal9Xrt+/1DYaiKG6vziuzQvwb8BjPn05zrP1t2K2l3x5ngosB9r69d/jbgKpOO8QvXz+eGCkTz+DcxRbUv5Zn74NF+cGnKPLF3z6KqAYyIqTsR7wIT6UAjAQmOrTHXCuXXoABUBanNzCeeXoFSPHz336UdwNsSnuBrwDaOzPgwFdfLl/j9MjC3czXgo08Pn/CvyLP9YnEP1QvbH/7YP3MBGn2w+A+bgHsubA18Adp9C5tHM5Gegfc87S/f+z1+d9F8iLqfteYJxPtwVWJZDv8Zh6O8b+P2W0jW0lepdVrow2mYvzlv6fOOPTKJH8MeVbAht9eizwI5/uC/z2G14fzwDbOe75ZGGMjP8NNImR7QBwqV/vjPNW0QBoiHOB1YG4awwXvmV+TH2NE8gwHxjo0/cBs336deB1n77Tn7O2vi1FgPax/4evOxJz3Pj28ADHr916OIN7KaW00eq8byXjEnpUVctnZpYfs54j6SPc018b3IUQz2E7/pRagLtxJOKtBGW+g3NSipmVuKGJpw3QDlgoF/piJN4Bppmt8/u/DdxjZn/z+9wkaTXuZnG937+Ed/zvemC9mX1hZkdwcYWa+22fm9lKn37dyxnLtbhz8aGXaaDXaQ/uRvSSpD44r9OY2VwzG5VAt27ATDM7ZmbbcA5oy9TZk+hcfgg8KelxoIXXqSxmAbfLDVX9MzCjFPle9zp8zIn/T3naxumUw9czzZeZjfMxt0JlvzvNAd6M0Sl++O+bEjv0992Y/N+Z2V9i1leY2Z99uiswx8wOmRt2y+N4G4q9xrbgejnPS7oZZ2ziuQZ/jeAemrrGbHvX/64HdpjZBjM7hnt4a3maevYA7vHtbRXOuLWm9DYaiKFc472BCuNgSUJSa1xE1s5mtlfS6yR+9/DXmHQxpf9nRxOUSRQeIB4B6+z4u5Z42uMu8JIhxwa49xydzGy7pGfi5C6R41hMumS9RK54v13x68L1ZP7pJGGlbFwAxwHAD3E3gLJI5CPsVDqfdC7N7DVJK4BbgPclDcYZr8SVmh2UtATXK7sT11Msl3zlbRun0YbiZduOe780XdJGEoc5KSEHuMDrC3CxpFY4o5olqY6/eVc0B8tYL6tdR8uZ2W5JHXAhQn6E+x/uK23HBJSnLZcXAQ+Y2QcnZLoHruDH7hSEHlX1cR6wH/hazqX/zZVQx+9xQwtIyiTx0/YG4BJJnX25uiXvUiT1B87FOTp9QdJ5QH3chbpLbobWnWcgVysdnx2W4+WM5UPgekkZXo5zJLX29Z1nZvOAH+OGy8piGTDAvwe4BNf7K1Pn0pCUYe4dzPPAf+CGmmLZjxuGimUazqh/aGaJnuaX4XqLSLqK4z3TstpGbD2llpM0VtL3EujR0/fykHQxzhlqwpARktoCKWZ2iZm1NOed/ZfAAHOTfdYDT0nRd5RXJqqzElgG9JFUXy5e2u24Yc94+S/CDRP/FvgZLpRGPCvx1whwlz/2mRD//8evLwAeiDn3V8q9my2tjQZiCD2q6uMj3A2zEPgv4A+VUMck4N8lrfP1FRI3/GFmRyX1AyZ6Q5AK/H9JX+HeIdzge05TcDMY75X0qj/Wf+OGMU6XT4Ah/qX0RmBqnExfyE2FfjNmWOpfgMPAW5Lq4R6yHoHoU2lmguG/2cCNXtZN+JtQaTqTeGi0hH+UlIPzkL8DeDJu+8dAiqS1wMtmNtHMVkk6ROJhP3BG7NWY/2eNzy+rbUzFDVluxfUsSyvXAechPZ5ewPOSjuCe5Ieb2Vfe1twsaVtM2bW4GFCxzAFeBX4B3IN7b7pF0mFc1NrHwE3bBv7JXBypsviJpLtj1k9p6MxstaSZQMkQ32QzWy/p8riiLYCXvSE1YESCwz3kyzwBfOF1Om18m10jaT3uQeYpYtoDzvv5pUDEn+svcQY2YRsNnEjwnl6L8U9vqWZ2xA8T/Q5obcfDk1eHTJfjXlhX2vdSyYKkFrhQ922sCi80f2N+z8wqo5ceCFQ5oUdVuzkX+MAbLAH3V6eROpuQdA8wChhWlUYK3LRzKmcoORCoFkKPKhAIBAJJTZhMEQgEAoGkJhiqQCAQCCQ1wVAFAoFAIKkJhioQCAQCSU0wVIFAIBBIav4XR5vG9bD68rMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x13aec4e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.title('Learning Curve, RForest: std features, max 12 features/estimator')\n",
    "plt.xlabel(\"Training examples: density data, SLAC.  Errors omitted\")\n",
    "plt.ylabel(\"$r^2$\")\n",
    "train_scores_mean = np.mean(train_scores, axis=1)\n",
    "train_scores_std = np.std(train_scores, axis=1)\n",
    "test_scores_mean = np.mean(valid_scores, axis=1)\n",
    "test_scores_std = np.std(valid_scores, axis=1)\n",
    "plt.grid()\n",
    "\n",
    "plt.fill_between(train_sizes, train_scores_mean - train_scores_std,\n",
    "                 train_scores_mean + train_scores_std, alpha=0.1,\n",
    "                 color=\"r\")\n",
    "plt.fill_between(train_sizes, test_scores_mean - test_scores_std,\n",
    "                 test_scores_mean + test_scores_std, alpha=0.1, color=\"g\")\n",
    "plt.plot(train_sizes, train_scores_mean, 'o-', color=\"r\",\n",
    "         label=\"Training score\")\n",
    "plt.plot(train_sizes, test_scores_mean, 'o-', color=\"g\",\n",
    "         label=\"Cross-validation score\")\n",
    "\n",
    "plt.legend(loc=\"best\")\n",
    "plt.ylim([0.7, 1])\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(), 'results', 'finalFigs', \n",
    "                             'densityMG-v1.0_LearningCurve.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Split #0\n",
      "Split #1\n",
      "Split #2\n",
      "Split #3\n",
      "Split #4\n"
     ]
    }
   ],
   "source": [
    "#take predicted values from each cross validation set to ensure training not performed on own set.  Take 200 training samples\n",
    "# Probably could have used cross_val_predict\n",
    "kf = KFold(5)\n",
    "rep=0\n",
    "densityData['density_predict'] = np.nan\n",
    "for train_index, test_index in kf.split(densityData['comp']):\n",
    "    print('Split #{}'.format(rep))\n",
    "    density_model.fit(X_dens[train_index,:], densityData['density'][train_index])\n",
    "    \n",
    "    y_densPredict = density_model.predict(X_dens[test_index,:])\n",
    "    densityData['density_predict'][test_index] = y_densPredict\n",
    "    \n",
    "    #print(train_index, test_index)\n",
    "    rep+=1\n",
    "#y_densPredict = "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x15bb5c88>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(densityData['density'], densityData['density_predict'], edgecolors='k')\n",
    "plt.plot([0,30], [0,30], 'r-')\n",
    "plt.xlabel('density [g/cm$^3$]')\n",
    "plt.ylabel('predicted density (RF)')\n",
    "\n",
    "r2Std = r2_score(densityData['density'], densityData['density_predict'])\n",
    "plt.text(7,20, 'R$^2$ score: {:.3f}'.format(r2Std))\n",
    "plt.xlim([0,25])\n",
    "plt.ylim([0,25])\n",
    "plt.title('Gen. 0')\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(), 'results', 'finalFigs', \n",
    "                             'densityMG-v0_Scatter.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compare to Vegard's law, rule of mixtures"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "densityData['density_vegards'] = np.nan\n",
    "densityData['density_vegards'] = [vegardsDensity(c) for c in densityData['comp']]\n",
    "densityData['vegards_diff'] = densityData['density'] - densityData['density_vegards']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Vegards: 0.8614466511737545, Random Forest Regressor: 0.834774621570455\n"
     ]
    }
   ],
   "source": [
    "#densityData['density_vegards'] = pd.Series([vegardsDensity(comp) for comp in densityData['comp']])\n",
    "r2Vegards = r2_score(densityData['density'],densityData['density_vegards'])\n",
    "r2RF = r2_score(densityData['density'], densityData['density_predict'])\n",
    "print('Vegards: {}, Random Forest Regressor: {}'.format(r2Vegards, r2RF))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x15ce0d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(densityData['density'], densityData['density_vegards'], edgecolors='k')\n",
    "plt.plot([0,30], [0,30], 'r-')\n",
    "plt.xlabel('density [g/cm$^3$]')\n",
    "plt.ylabel('predicted density (Vegard\\'s)')\n",
    "plt.title('Gen. 0')\n",
    "\n",
    "r2V = r2_score(densityData['density'], densityData['density_vegards'])\n",
    "plt.text(7,20, 'R$^2$ score: {:.3f}'.format(r2V))\n",
    "\n",
    "plt.xlim([0,25])\n",
    "plt.ylim([0,25])\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(), 'results', 'finalFigs', \n",
    "                             'densityMG-v0_Vegards_Scatter.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Investigating poorly performing predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cu1.5 Si1.5 Ag0.3 B4 P18 Pt74.7\n",
      "Ta21 Co11 Ni18\n",
      "Hf62 Al13 Ni9.75\n",
      "Hf46 Nb2 Al13 Cu29.25 Ni9.75\n",
      "Hf49 Ta2 Al12 Cu27.75 Ni9.25\n",
      "Fe9 P5 Pt6\n",
      "Zr57 Nb5 Al10 Cu15.4 Ni12.6\n",
      "Zr57 Nb5 Al10 Cu15.4 Ni12.6\n",
      "Zr57 Nb5 Al10 Cu15.4 Ni12.6\n",
      "Zr57 Nb5 Al10 Cu15.4 Ni12.6\n"
     ]
    }
   ],
   "source": [
    "largeErrorDensity = densityData[ [ (abs((x-y)) > 2)&(x>10) for x,y in zip(densityData['density'], densityData['density_predict']) ]]\n",
    "for comp in largeErrorDensity['comp']:\n",
    "    print(comp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>comp</th>\n",
       "      <th>density</th>\n",
       "      <th>density_predict</th>\n",
       "      <th>density_vegards</th>\n",
       "      <th>vegards_diff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>155</th>\n",
       "      <td>(Zr, Nb, Al, Cu, Ni)</td>\n",
       "      <td>6.758</td>\n",
       "      <td>12.330885</td>\n",
       "      <td>6.91044</td>\n",
       "      <td>-0.15244</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>458</th>\n",
       "      <td>(Zr, Nb, Al, Cu, Ni)</td>\n",
       "      <td>6.69</td>\n",
       "      <td>12.400876</td>\n",
       "      <td>6.91044</td>\n",
       "      <td>-0.22044</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>606</th>\n",
       "      <td>(Zr, Nb, Al, Cu, Ni)</td>\n",
       "      <td>15.2</td>\n",
       "      <td>6.715432</td>\n",
       "      <td>6.91044</td>\n",
       "      <td>8.28956</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>607</th>\n",
       "      <td>(Zr, Nb, Al, Cu, Ni)</td>\n",
       "      <td>14.3</td>\n",
       "      <td>6.715432</td>\n",
       "      <td>6.91044</td>\n",
       "      <td>7.38956</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>608</th>\n",
       "      <td>(Zr, Nb, Al, Cu, Ni)</td>\n",
       "      <td>15.67</td>\n",
       "      <td>6.715432</td>\n",
       "      <td>6.91044</td>\n",
       "      <td>8.75956</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>609</th>\n",
       "      <td>(Zr, Nb, Al, Cu, Ni)</td>\n",
       "      <td>15.44</td>\n",
       "      <td>6.715432</td>\n",
       "      <td>6.91044</td>\n",
       "      <td>8.52956</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>615</th>\n",
       "      <td>(Zr, Nb, Al, Cu, Ni)</td>\n",
       "      <td>6.68</td>\n",
       "      <td>6.715432</td>\n",
       "      <td>6.91044</td>\n",
       "      <td>-0.23044</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     comp density  density_predict  density_vegards  \\\n",
       "155  (Zr, Nb, Al, Cu, Ni)   6.758        12.330885          6.91044   \n",
       "458  (Zr, Nb, Al, Cu, Ni)    6.69        12.400876          6.91044   \n",
       "606  (Zr, Nb, Al, Cu, Ni)    15.2         6.715432          6.91044   \n",
       "607  (Zr, Nb, Al, Cu, Ni)    14.3         6.715432          6.91044   \n",
       "608  (Zr, Nb, Al, Cu, Ni)   15.67         6.715432          6.91044   \n",
       "609  (Zr, Nb, Al, Cu, Ni)   15.44         6.715432          6.91044   \n",
       "615  (Zr, Nb, Al, Cu, Ni)    6.68         6.715432          6.91044   \n",
       "\n",
       "    vegards_diff  \n",
       "155     -0.15244  \n",
       "458     -0.22044  \n",
       "606      8.28956  \n",
       "607      7.38956  \n",
       "608      8.75956  \n",
       "609      8.52956  \n",
       "615     -0.23044  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "compTrouble = Composition('Zr57 Nb5 Al10 Cu15.4 Ni12.6')\n",
    "densityData[ [compTrouble.almost_equals(c) for c in densityData['comp']] ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "error std: 0.8400961959128355, error mean: 0.16131624945023973\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x15cf42e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(densityData['density'], densityData['vegards_diff'], edgecolors='k')\n",
    "plt.plot([0,25], [0,0], 'r-')\n",
    "plt.xlabel('density [g/cm$^3$]')\n",
    "plt.ylabel('error')\n",
    "plt.title('Vegards Law Error')\n",
    "\n",
    "stddev = np.std(densityData['vegards_diff'])\n",
    "mean = np.mean(densityData['vegards_diff'])\n",
    "plt.plot([0, 25], [stddev, stddev], 'r--')\n",
    "plt.plot([0, 25], [-stddev, -stddev], 'r--')\n",
    "\n",
    "plt.xlim([0,25])\n",
    "plt.ylim([-17,17])\n",
    "\n",
    "print('error std: {}, error mean: {}'.format(stddev, mean))\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(), 'results', 'finalFigs', \n",
    "                             'densityMG-v0_Vegards_Error.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "11.683333333333334"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(densityData)/60"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "error std: 0.8400961959128355, error mean: 0.16131624945023973\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x15fb92b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5,5))\n",
    "n, bins = np.histogram(densityData['vegards_diff'].values, bins='auto')\n",
    "\n",
    "binCenter = [(bins[i]+bins[i+1])/2 for i in range(len(bins)-1)]\n",
    "\n",
    "plt.bar(binCenter, n)\n",
    "\n",
    "plt.xlabel('error')\n",
    "plt.ylabel('frequency')\n",
    "plt.title('Vegards Law Error')\n",
    "\n",
    "plt.xlim([-5,10])\n",
    "plt.ylim([0,1.1*np.max(n)])\n",
    "\n",
    "stddev = np.std(densityData['vegards_diff'])\n",
    "mean = np.mean(densityData['vegards_diff'])\n",
    "plt.plot([mean, mean], [0, 1.2*np.max(n)], 'r')\n",
    "plt.plot([stddev, stddev], [0, 1.2*np.max(n)], 'r--')\n",
    "plt.plot([-stddev, -stddev], [0, 1.2*np.max(n)], 'r--')\n",
    "\n",
    "plt.text(3,60, '$\\mu$={:.3f}, $\\sigma$={:.3f}'.format(mean, stddev))\n",
    "print('error std: {}, error mean: {}'.format(stddev, mean))\n",
    "\n",
    "\n",
    "if saveFigures:\n",
    "    plt.tight_layout()\n",
    "    plt.savefig(os.path.join(os.getcwd(), 'results', 'finalFigs', \n",
    "                             'densityMG-v0_Vegards_Error_hist.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Why would the same composition be predicted differently?  For the same composition one should get the same features, yet the predicted density varies.  This is also true for the previous predictions.  \n",
    "\n",
    "Should probably depend on the specific split used to train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Drop troublesome compositions\n",
    "X_notTrouble = X_dens[ [not compTrouble.almost_equals(c) for c in densityData['comp']] ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Density Prediction, r2 sans outliers: 0.9075911576494425\n"
     ]
    }
   ],
   "source": [
    "#  Examine performance without outliers (trouble compositions)\n",
    "notTroubleDF = densityData[ [not compTrouble.almost_equals(c) for c in densityData['comp']] ]\n",
    "notTroubleDF.reset_index(inplace=True, drop=True)\n",
    "\n",
    "#  Examine performance without outliers in normal model\n",
    "print('Density Prediction, r2 sans outliers: {}'.format(r2_score(notTroubleDF['density_predict'], notTroubleDF['density'])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Split #0\n",
      "Split #1\n",
      "Split #2\n",
      "Split #3\n",
      "Split #4\n"
     ]
    }
   ],
   "source": [
    "#take predicted values from each cross validation set to ensure training not performed on own set.  Take 200 training samples\n",
    "# Probably could have used cross_val_predict\n",
    "impDF = pd.DataFrame()\n",
    "\n",
    "kf = KFold(5)\n",
    "rep=0\n",
    "notTroubleDF['density_predict'] = np.nan\n",
    "for train_index, test_index in kf.split(notTroubleDF['comp']):\n",
    "    print('Split #{}'.format(rep))\n",
    "    density_model.fit(X_notTrouble[train_index,:], notTroubleDF['density'][train_index])\n",
    "    \n",
    "    y_densPredict = density_model.predict(X_notTrouble[test_index,:])\n",
    "    notTroubleDF['density_predict'][test_index] = y_densPredict\n",
    "    \n",
    "    featureImp = pd.DataFrame(density_model.steps[1][1].feature_importances_,\n",
    "                          index=dens_featurizer.feature_labels(),\n",
    "                          columns=['importance']).sort_values('importance',ascending=False)\n",
    "    impDF = impDF.append(featureImp)\n",
    "\n",
    "    \n",
    "    #print(train_index, test_index)\n",
    "    rep+=1\n",
    "#y_densPredict = "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "top features over 5 training splits\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>importance</th>\n",
       "      <th>totalImp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>mode Number</th>\n",
       "      <td>0.052223</td>\n",
       "      <td>0.300349</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean NValence</th>\n",
       "      <td>0.057423</td>\n",
       "      <td>0.288853</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mode AtomicWeight</th>\n",
       "      <td>0.038915</td>\n",
       "      <td>0.255935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mode NValence</th>\n",
       "      <td>0.042134</td>\n",
       "      <td>0.246298</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean Number</th>\n",
       "      <td>0.018850</td>\n",
       "      <td>0.221519</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   importance  totalImp\n",
       "mode Number          0.052223  0.300349\n",
       "mean NValence        0.057423  0.288853\n",
       "mode AtomicWeight    0.038915  0.255935\n",
       "mode NValence        0.042134  0.246298\n",
       "mean Number          0.018850  0.221519"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "impDF['totalImp'] = impDF.groupby(level=0)['importance'].transform('sum')\n",
    "impDFtop = impDF.drop_duplicates(subset=['totalImp']).sort_values('totalImp',ascending=False)\n",
    "print('top features over 5 training splits')\n",
    "impDFtop.to_csv(os.path.join(os.getcwd(),'results','finalFigs',\n",
    "                             'densityMG-v1_featImportance.csv'))\n",
    "impDFtop.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x16061cf8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(notTroubleDF['density'], notTroubleDF['density_vegards'], edgecolors='k')\n",
    "plt.plot([0,30], [0,30], 'r-')\n",
    "plt.xlabel('density [g/cm$^3$]')\n",
    "plt.ylabel('predicted density (Vegard\\'s)')\n",
    "plt.title('Gen. 1')\n",
    "\n",
    "r2V = r2_score(notTroubleDF['density'], notTroubleDF['density_vegards'])\n",
    "plt.text(7,20, 'R$^2$ score: {:.3f}'.format(r2V))\n",
    "\n",
    "plt.xlim([0,25])\n",
    "plt.ylim([0,25])\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(), 'results', 'finalFigs', \n",
    "                             'densityMG-v1_Vegards_Scatter.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1606ba90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(notTroubleDF['density'], notTroubleDF['density_predict'], edgecolors='k')\n",
    "plt.plot([0,30], [0,30], 'r-')\n",
    "plt.xlabel('density [g/cm$^3$]')\n",
    "plt.ylabel('predicted density (RF)')\n",
    "plt.title('Gen. 1')\n",
    "\n",
    "r2V = r2_score(notTroubleDF['density'], notTroubleDF['density_predict'])\n",
    "plt.text(7,20, 'R$^2$ score: {:.3f}'.format(r2V))\n",
    "\n",
    "plt.xlim([0,25])\n",
    "plt.ylim([0,25])\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(), 'results', 'finalFigs', \n",
    "                             'densityMG-v1_Scatter.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Alternate Feature Set, from rfecv and covariance matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['maximum NsValence', 'minimum NpValence', 'minimum NfValence', 'minimum NsUnfilled', 'minimum NpUnfilled', 'minimum NfUnfilled', 'minimum GSbandgap', 'mode GSbandgap', 'compound possible']\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.colorbar.Colorbar at 0x15b305c0>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAlEAAAI3CAYAAAC70Q6ZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4xLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvAOZPmwAAIABJREFUeJzsnXd4XNW19teephl1yZJ7b9jYdAOmG0InAdJISL6QBL5wuWk3hbRbwk37Uki5pEEgECCXhNBxEogDgUAIzRiMK+5FstX7jKbP/v6Y0Xnfsc4BIyRbFuuXhyfLe86cs9s5OrPevdY21lpRFEVRFEVR3hy+g10BRVEURVGUQxF9iVIURVEURRkC+hKlKIqiKIoyBPQlSlEURVEUZQjoS5SiKIqiKMoQ0JcoRVEURVGUIaAvUYqiKIqiHFCMMbcZY1qNMes8PjfGmJ8aY7YaY9YYY46lzz5qjNlS+O+jB67Wg9GXKEVRFEVRDjS3i8j5r/P5BSIyr/Df1SJyo4iIMaZWRK4TkRNF5AQRuc4YUzOiNX0d9CVKURRFUZQDirX2aRHpfJ1DLhGRO22e50Wk2hgzSUTOE5HHrLWd1touEXlMXv9lbEQJHKwLK4qiKIpyYDjvzDLb0Zk9YNdbtSa5XkQSVHSztfbmN3GKKSLSQP9uLJR5lR8U9CVKURRFUcY4HZ1ZeXHF9AN2Pf+kLQlr7ZK3cArjUmZfp/ygoHKeoiiKoiijjUYRmUb/nioie1+n/KCgL1GKoiiKMsaxIpI7gP8bBpaLyBWFKL2lItJjrW0SkRUicq4xpqawoPzcQtlBQeU8RVEURVEOKMaY34vIMhGpM8Y0Sj7iLigiYq29SUQeEZELRWSriPSLyMcLn3UaY74lIisLp/qmtfb1FqiPKPoSpSiKoihjHitZOyweomHBWnv5G3xuReRTHp/dJiK3jUS93iwq5ymKoiiKogwB9UQpiqIoyhgnvybqoAWxjVnUE6UoiqIoijIE1BOlKIqiKG8DhilqTiHUE6UoiqIoijIE1BOlKIqiKGMcK1ayVtdEDTfqiVIURVEURRkC6olSFEVRlLcBGp03/KgnSlEURVEUZQjoS5SiKIqiKMoQUDlPURRFUcY4VkSyKucNO+qJUhRFURRFGQLqiVIURVGUtwG6sHz4UU+UoiiKoijKEFBPlKIoiqKMcayIJtscAdQTpSiKoiiKMgTUE6UoiqIobwN0++HhRz1RiqIoiqIoQ0A9UYqiKIoyxrFiNU/UCKCeKEVRFEVRlCGgnihFURRFGetYkaw6ooYd9UQpiqIoiqIMAfVEKYqiKMoYx4pG540E6olSFEVRFEUZAuqJUhRFUZQxj5GsmINdiTGHeqIURVEURVGGgL5EKYqiKIqiDIFR8RJljDnfGLPJGLPVGPPVg12f4cQYM80Y86QxZqMxZr0x5t8K5bXGmMeMMVsK/19zsOs6HBhj/MaYV4wxfyr8e5Yx5oVCO/9gjAkd7DoOB8aYamPMfcaY1wpje9IYHtPPF+buOmPM740x4bEyrsaY24wxrcaYdVTmOo4mz08Lz6k1xphjD17N3zwebb2+MIfXGGMeNMZU02dfK7R1kzHmvINT66Hh1lb67FpjjDXG1BX+fUiP6/5iRSRnD9x/bxcO+kuUMcYvIr8QkQtE5HARudwYc/jBrdWwkhGRL1prF4rIUhH5VKF9XxWRv1lr54nI3wr/Hgv8m4hspH9/X0R+Umhnl4hcdVBqNfzcICJ/sdYuEJGjJN/mMTemxpgpIvJZEVlirV0sIn4R+aCMnXG9XUTO36fMaxwvEJF5hf+uFpEbD1Adh4vbZXBbHxORxdbaI0Vks4h8TUSk8Iz6oIgsKnznl4Vn9aHC7TK4rWKMmSYi54jIbio+1MdVOYgc9JcoETlBRLZaa7dba1MicreIXHKQ6zRsWGubrLUvF+w+yf+xnSL5Nt5ROOwOEbn04NRw+DDGTBWRi0Tk14V/GxE5S0TuKxwyVtpZKSKni8itIiLW2pS1tlvG4JgWCIhIxBgTEJFSEWmSMTKu1tqnRaRzn2KvcbxERO60eZ4XkWpjzKQDU9O3jltbrbV/tdZmCv98XkSmFuxLRORua23SWrtDRLZK/ll9SOAxriIiPxGRL4sU7X9ySI/rmyFbWFx+IP57uzAaXqKmiEgD/buxUDbmMMbMFJFjROQFEZlgrW0Syb9oicj4g1ezYeN/JP+AGkhHMk5EuukhPVbGdraItInIbwrS5a+NMWUyBsfUWrtHRH4o+V/uTSLSIyKrZGyO6wBe4zjWn1VXisijBXvMtdUYc7GI7LHWvrrPR2OurcqBYzS8RLm9so45RdUYUy4i94vI56y1vQe7PsONMeadItJqrV3FxS6HjoWxDYjIsSJyo7X2GBGJyRiQ7tworAe6RERmichkESmTvPyxL2NhXN+IsTqfxRjzH5JfenDXQJHLYYdsW40xpSLyHyLydbePXcoO2bZ6YUU9USPBaHiJahSRafTvqSKy9yDVZUQwxgQl/wJ1l7X2gUJxy4DLuPD/rQerfsPEKSJysTFmp+Ql2bMk75mqLshAImNnbBtFpNFa+0Lh3/dJ/qVqrI2piMjZIrLDWttmrU2LyAMicrKMzXEdwGscx+SzyhjzURF5p4h82Fo78PIw1to6R/I/BF4tPKOmisjLxpiJMvbaqhxARsNL1EoRmVeI9glJfjHj8oNcp2GjsC7oVhHZaK39MX20XEQ+WrA/KiIPH+i6DSfW2q9Za6daa2dKfgyfsNZ+WESeFJH3FQ475NspImKtbRaRBmPMYYWid4jIBhljY1pgt4gsNcaUFubyQFvH3LgSXuO4XESuKERzLRWRngHZ71DFGHO+iHxFRC621vbTR8tF5IPGmBJjzCzJL7p+8WDUcTiw1q611o631s4sPKMaReTYwr085sbVi5w1B+y/twsHPWO5tTZjjPm0iKyQfOTPbdba9Qe5WsPJKSLyERFZa4xZXSj7dxH5nojcY4y5SvJ/qN5/kOo30nxFRO42xnxbRF6RwmLsMcBnROSuwov/dhH5uOR/lIypMbXWvmCMuU9EXpa83POKiNwsIn+WMTCuxpjfi8gyEakzxjSKyHXifW8+IiIXSn6Rdb/kx/yQwaOtXxOREhF5LP+OLM9ba6+x1q43xtwj+RfmjIh8ylqbPTg1f/O4tdVa6zVHD+lxVQ4uBt5bRVEURVHGIocfGbL/+6eJB+x6x81oWGWtXXLALniQGA1ynqIoiqIoyiHHQZfzFEVRFEUZWawYyarfZNjRHlUURVEURRkC6olSFEVRlLcBb6eouQPFqPJEGWOuPth1OBC8Xdopom0dq7xd2vp2aaeItlVRhsKIvUQZY84v7P691Rizv9mc3y4T++3SThFt61jl7dLWt0s7RbStYxrNWD4yjMhLVGG3719IfnuIw0Xk8sKu4IqiKIqiKGOCkVoTdYKIbLXWbhcRMcbcLfn9tza4HVxX67czpwVl+pSAVFRNdRJXWT/eZoukXI+XXEMpr9KV+EcgkM8Rl0lQcz3OZ9KwbZBO6IdtEnj3tH76bg52IOFeF58/J8H6KonMnWxzSfoyw6+2lN4uWJoeVPV0Fuew1EmGOsPXixNmS1FeXpp07HhHGMfAFF+K7AzsTCUaGwmh0+L9JTjIWPHXVkvJjKmWK+2P5/9hK9C48hDq0huPFJ0DbaK6RKlNZTimrATn6e9BQ2yI+iNJJ6IhXjR1gmOv60LC4qJUamlc11eSr/+i6skiIjJ9+nQpmTbNNfGaj+YV94WlsTbUv7Yc/evz4ZQ2SuPNc4/myaIpaMfatma36ohkzb5VKdosjOcvj7Xfn7eD9ZUSmYS25vjWorosnoy67Ijtcq1KlOaMCVAtkuicxRNfv02+1Bs/K/j+NOWcN9I9V561RgKFezXgw5cPq5yKurS0uH7Xs2O5zjQncpHCQcYrbx/d21QXfzcmgds4iYhkkx6PeZpXR9RNlOnTp8uSJUvs2o7XnzP7csR4um+a8/1hA+7tMBkaJ3780djws5braGg+WJ5v1I/cfbkI3UP9fKOJBCprJDJpms1RHY6gObY1ultEROLJkMfJqS+4Xzye3YE4HZLBeaK9e9qttfWiHLKM1EuU267YJ/IBBU36ahGR6VMC8uKK/NZFy676hHNMpgwzMlPCN5/7AzOQxORsPh931ri6PhER6dxcSxWAyS9L4RbcVfEp+KvmK8f5Sjbhj3yyBjdqiF5Wal5DedN5+G5FDXZWiG2vQnX4/izBP/wxnHPCUfkHFD/QW3oqUJd4EHUJ45rlK8odu+M43NmnHL3ZsdffBUdhzwJ6udmB/gh3oF4d5+Cv7BHTsM3UmjUzHbvoJTSAOlevzj+Ysmd3OWWnTtnh2H9Zv8ixfSHUZeBlWEQk/Aza3bsEdTlhzk7U5dEFjh2fibfB0m14MPrpJfGl6z/v2PPv+6Zjp1O4TUwTXswic3vy37vkG07ZrP/5kbhR2oRxzGGYJEfP6FA37MRJUXw3jEqmnscc5pfzYC8m9EvfQTtm3HS9a30CvflxLfpbT/+o2Qi781z8Bagsh518bhzscXQfdKOtL30TdfnQ87i3mX+umY/v1mAs7bYynOerX3DsGTcPblPpLowR/4HlP9T8gyB0YifKffwXHCRSGKj6iphjP302rj/7J+7jzS/H/BLF93mE5kT08Hzl/CWY45aqxT+QwqVoSPmfcB90vAN9V1WF50zP9hqqF72Q00vGS1d92bFn3vm9Qe3Jn4gmLjXkpU9f69hzv5/f2SpdS78IiFA7BiddhesbegnOVeK7/lKyt+G5m6bnbrgJg+ynF6r+IzFXS1/Fd/mFP1VF7fgy5tjF//i0iIi8uh0vzP4QvZhGcRJ/L+xcKb3I9mF8x63FdSIdGOOnH/2K+y+LEcFI1o6qZdBjgpF6iXrDXbGttTdLfvsIqaiaagdenv5+6y3OMVl6imTotT5LroEc/YRZ+uJVjh14rdKx+zfViYhIkDwspc04R5pe1vqnsjsJtu3EX7vUQjygyleW4vgz8FLQcwTOGdyMumS241d3kF4yIi3osgydMj4ex7Styv9S4j8M5Tthh8I4R6Qd9U2+H/Xy9eJhsuY+vDhNes9ux84tn+7YUy7FBba34Y+mfytezLa/PMexS1AslTtQ91Qlplr3Mfk/AuHVeLj/4xnYpdS+cRvwEO2vxwedZ+EBaWM496uP4cUpORdeqdrn0B99Z+IFhR1ETHAl/jhlJmAeBPvQx6kNhZfgS+iL9UlxY8ax+HW/sQFZg4+bhX5fuQ79WL4KHZmlX7TxGfhHpBn90T+H3hCIUJe7t3PAW8QeHCZxKeZMyQsYm2w/JmcSTmMpa8R8j01x90JsvGOha3npONQhsBHtjs5yf7kxpYP/QC+8cLtjr9o6w7GPn7fTsVdumuXYdiVeRos8yOQxqGjFB31luIflbDp+Gl6umCR5QYu8qQmMx7hz8eMjuid/b+XoOcPejvLd6N90Oc7dcTYqzC/4sd2wDf0oq9qK8yTq3OeGzbjPCVPldbeATFmhz4Luc8B/WJ9jp1vwkhycin4063Hvpcvx4paagusbejbHyXM1ex7us12rJzv2jIvwI23DGjzf+IWf2fjMbBERqWxHX/Qehnk3bhX6ruN4mo80ZtkI6hW6Ah7LCeU9OP5R18srhxAj9RKlu2IriqIoyijBikhudAXkjwlGqkdXisg8Y8yswgatH5T8TtmKoiiKoihjghHxRFlrM8aYT4vIChHxi8ht1tr1nsf7jbP+iSU8v8E7np/e97Li7uZPpXilLZ//DSrM3mvPBY7uX/XKXcbraLwWEnt+matGUs7AIsjiRZvG1bQ+civzgnNqU9HaHI/zZHO0Li2DjqTTF52nuPJcHyrP+AplXBeqo/vwFo2j5YWd1B/ZkLuMYD0W/1iPMeD6Fi3+5rVrLnePeQs/S0ya6uV1fY/FvUULXffnWoXjua+5K3isjce9VLRQ28MuqmLAo44ea6l57heV+zy+8Eaf70cfcR/wWL/hM2RfuA5svkHdiwNePM5B97Dl+cD3tt/9fma82mQ8FoXLG9VdxOlA43efBEVBEmQXlfPpvBaoexxfVBWPe5HvIeMxJwbubV54XjQGnhEZ1O6U+zO4P0OS7QHm7ZR64EAxYhnLrbWPiMgjI3V+RVEURVGUg4lu+6IoiqIoYxxrNTpvJBgVL1HWIIUBR+H5PTQ0v4dmksu6y01u88Yad+mEPejGw/Vc7POn65P0lc2wFElf9bufx1O+kcHlRW54bqdHm7lePmognyeTc79ohk5UJB28Qf/uW+4W8u1VXy8ZpUg+Yxe6z71NtkhGoRBq69FpHnUXl7rnr+UiJHjm+eFjYHJdinLo7EeuoyLepJzn2myWcbM0fz3laK+TexR7pUXzkO282uQm1+W87smisX7jOhZRJN161MWr2/ej3HUfMy9p0+OeYFm7qC4e917RffAmZbs3klGL6ul1jqIHLF2Sji/K4PVW/uZ7yXkeEjbjyIg+9/lTNJf5HmY5j5/BdEwq+2a1YWU0MypeohRFURRFGVlyuiZq2FHfnqIoiqIoyhAYHZ4ogyzknEiTo/C8JDzGekkdbl/lsv1xybNX1ytKo6guJM1wuUf0j9e1XCNC9uPHhPWoL1+0+Bh3KYnd0NZDXvGU8zzkBadNXm0uOgdLnu4ynPHsO48Ipf2IiiyuA/+DTJd278+ZjUf/Fkl1Xorj/shQfLhHhcwbfG73Y26+2d+0nt3ukdF7f5RRN1gy8pTz3vRJ38J3Pc7jKuftxzW9Ik0954zXOb3krqF2vNd1PI/ZH+l76PL4/sw3Lwa+63l/eOm4HnJl0Xff5PNnuLAiklW/ybCjPaooiqIoijIERocnSlEURVGUEUSj80aCUfMSNeDhzHkk0tyvc7Ardz+vN/gDj/L9cI9bL93OiyJpbT/c0G+G/XKnv3F5kevZq3/fbITSUOsy2u7/YRinXFHo1H58YdjkKRdN9U3K2l7VetMMg3pURJG8MzyT5iApMFQBsvdH3t0PGWxE2lS4rGfUIjNcx7xZPNdUuFz3LURfWo9z6+LuscWoeYlSFEVRFGVk0L3zRgbtUUVRFEVRDijGmPONMZuMMVuNMV91+fwnxpjVhf82G2O66bMsfXZQ9+UdFZ4oY0UCybzrc+mLVznlvBceJ9LkKDyWm7affZtjz378SscufSUsIiK1F+5xynavm4TrT0jghHvDjhnaATtThova/hLHjs7K4DyNFajL+29CXR5DXUKbcM5xZzQ5dsP2esf2V6Ucu2RdKeyufB8lq/Humy5D1aPz0qjvbmxol4iivr5W2P4kvtvQWuPY1cvaHXvXtvGO/drFv3Dsw5+42rEn3oH6BL/W4th7H52Oui3Cxb578gMiIvK1v7/PKZv+J4xj55VRx24PVzt2sp4SsbZi/6lgLyUQpL3zbCemd8lFrY7dt7EO5R3uvyMSh8cdu/YpjFn4fc2O3f/whEHfszR/mA1tMxzbhykjr/ag/AsXPOrYv9xwumMnmzAHyqf3OnZFGH3avx1tYjLl7vK4LSv0pYc09L75ax37vt4ljh1sR5+G5qMuucPw5UwH6sv0zXHPqjl9AeZMZ3/EsUty7mPj2xkZVPZq8xzHDvSjUa/2on/LJ2Fe9ccrHdvQePgT+G50NtpUttu1KpLoLXH/gCNZOfEk2c3rcW9JRb5vbAnGyyTQ/t4j8UwItuDevmLJc45956qTHDvQjmP802KO3V1F+7al3WWlqmfd57D1e8hQH4KZi+TrX7Fq8BiJiASjOHdlBOcL9WE8Oo7GPAk347mfrqJkxmWUsJKS3ja+PNmxa7fgkA1C82Ce87dY+jroAUqUdOavlVuCOW72lOOSKVwz0oB7IlVDdazGxOIkx75h168PPYwxfhH5hYicIyKNIrLSGLPcWrth4Bhr7efp+M+IyDF0iri19ugDVd/XY1S8RCmKoiiKMrJkD/riPocTRGSrtXa7iIgx5m4RuURENngcf7mIXHeA6vamUDlPURRFUZThps4Y8xL9dzV9NkVEGujfjYWyQRhjZojILBF5gorDhXM+b4y5dNhr/iYYFZ6odKWV5vPzUlTgNbh1i/ZW8ooKIpslPJb2mpbl3fifb7jYKbvk3Fcd+49NRzp2+8qpjs3yQ/lUuHX7t1ahitbdbT/7r5Alt597q2P3vAMy0WcaznPsD5zxEurTjPpsbZrm2PEB9YiyMoZbeCMt1IWluvAWuNDjk+FiTh+PutgWSDDd1Kn+KM6/+KlPOPYLy37u2L85BvV9oAEe1mPeu86xX6Z2fP3l/DicuHibU7ZzWi3q2AVZNDuDGkISSWQb6tV3GGTMaTMhRTa9OtGxu/pIYqL2pardXeu2G7JH71yUd26B7OqbM/i71uOOYpkm5yGL/PRPFzp2NkIyHB2eehWy696JaHeoy30/rgfe9VPX8kf68mNWG4DEtS6GuX/fi8c79hdOW+HY/Tn0y+OtCxx7XBiSUUs5xq8Ij59sDS1oUy5GMlRF2u1wSY8fXG5oH7jsRJINaRwTm3Hf5mpIwyOCvbh+sAcVznlsd7Zozh7X8oZuyNClJZDioglaCpCFPDRhapeIiKRpz8LOdvd+zFRibtz54smOPXMmJGvB7SZ7OtBujkbzlbrLq30zXIslFyKp0SOIOtCd76i+ue79G2nCDRKfgmPCzSi3FShP0DONpfdckKVAlPP+k71zSdqn+4/voU+8+29Uu684Vs0pedm+h+RlS+3vPAV9F9qFMQ3RnLG9mHvRtXgW9XvtFTnCWDEHOtlmu7V2icdnbg9BL53zgyJyn7WWe266tXavMWa2iDxhjFlrrd3m8f0RRT1RiqIoiqIcSBql6FVfporIXo9jPygiv+cCa+3ewv9vF5G/S/F6qQPKqPBEKYqiKIoysgxX3rRhYKWIzDPGzBKRPZJ/UfrQvgcZYw4TkRoReY7KakSk31qbNMbUicgpIvKDA1JrF0bFS1QgkJVxdX0iItK/CVFGlpUqHnuPeTAQhScCCU9EZFIg7zbf0olzH1e1y7G74/heUe7DwH5EUbCDkVzPFa/Axdt01uC6iIhs6YY0tKgckXrdCbiQWdYxBfe7TaGSmXJyZWdx/SwFDfmhJhQ5UcsikMqycdQrK3BD+6h9JWsgiTWehKnzznJEct3Ugaiy0yfCu5pMQCbxbc+3LzoJlTy2vtGxH90FedBfCekmGyOXP42TIVd9JIDjWWlNdWGMWU0rks0If5wiIKvRCSWtmJTJiYMli/1K4sq5/mjMKnagPDqDxpgiQ4N9OCZbgv4wHhJBzAZdyzsKYZ2c+K8zhfGNNOLcnRlEMFX4Ecna0oc5E0/jOqmsh/bldTt1YB4EUqiPuyBULN25EQigM3LUv8E+Onc1fYHmUi5I56apkSlzv2Y05R6dl0q7P1pTKffyeCrff1mOSOQIvzTJRBSNFm5Av/P9VBnGOKVjHJFHkpjHZM1Ue0ymAHWIxz6aA88LU+4+ekmKXvOV415N1nKGVErqS/e2+PhZJ2TzkgqYfN/y3ON7qCPtHp1XHsw/NJv6aKLQOcqr+x073oz+5br4SHUOd+HLHNn3dsVamzHGfFpEVoiIX0Rus9auN8Z8U0RestYOpC24XETutrYolfVCEfmVMSYn+RH/Hkf1HWhGxUuUoiiKoigjx2jbgNha+4iIPLJP2df3+fd/u3zvWRE5YkQr9yYYFS9RmURAOjfnFxYH3VOU7BecB4oXkQ94oFYdd49T9qVmSKgPHoVF6Mt2ftGxTRi/ZGI7sTizbg1+bbSfA2/OuMdR+dIPQ979t92XOPaeKM7z3FH3O/Y32g537HsW3+7Y5z/7ZccWkx+udAUtoi2hxZP0CzUxjxZk04L3YBBt6t0wzvW7gXosOA+0wNtQ/w707w+bsCj+n69h5fX6c3/p2JdtRbsfOvlGx/4/ZR8XEZENu5Gra3PjTMe+4qKnHfuve7F4+eQFcNU8FDvBsU0Xfo1v7cTiaD/9GK6f0eHYnS8hP0+kySMX0QwslK58Fn0QOwF9U/nS4Fw4pjY5qExEJLgLcyM1Ba7BmhfxK7brKIxN6Xhc32Tg2YlVkzcuQ96qpHs7Pr12kIdcREQumrFeRESak5iPM0vRR42nwTP6t+bDHLsjCm/VmdO3OvaaTuTnmV/T5npNJzfVPgRK0R9pyrsUrkm4HS5lVfFBZbkX4TGIL8Rcrt6Afuk6gvL2xNCnRbtzzEe/p/rIwxBy91jGUiHX8uOmIPAo4yGhrNwz37GTBQ8V5707Yi68s8kMxj2Zhd0yDovP29vITqI/Fi1AXVpjmMtVYff+Ld39xn8WvDyu2YHu6HP3gJY1oi+SpACUoanSvRC2jzyTafIGhrpwnvg0jGvlJspj1gU7UY/vxo5EuyN+9+CFbS/nl+tUH9aJehl4rfjvQTV5kLsWY45b8sb1zyYvGikJxSt9lEORUfESpSiKoijKyGHFjKY8UWOG0ePbUxRFURRFOYQYHZ4oI85i29Jmj7w9xn3xIL9Y81YunAdqYBE5S3jXT3zFsX/WtcixOTdMOgeXdI6kiM7D0W1hkiJ6Z0Pe6ViHvCCXnrcalUSKEvnPVsi63x5Pi7O7IZ+ky8mF3ZNvbCDqnkArSVue+GjbB0MLNbO0cNbSguUA5VphyY8Xzu/ajDYtO5X2VIDiJu/Z/G7HvnHOHxz7+tazHbu7Iy8phMrQd6kJuP7v1iNH0VHT4Od/tnWWY/uTtPiU5gMvFLeUM6Y/CdmF11r3T3WXabKNkK3iE2lONkFuio8fPFdNi/tC4wyNI0/axDjYLKMkEpBmcrS4NtxKx0ymPFEd7ou5L525xrX8zIrB6zD7c6j73U8j/9CS4zDW40uxKjeZQ12OrIV87fNIIhRsdpd4MlWoe9H9V+b+eIo1Ds6fZOopuKAMkmpiHCSjYA/1Ed1CPIqB1ZC7MhP4/nD/Bd8bc19/sCaDZ1HAR3ULYcx8ND+TnYVnB2mL69qRsMlU4l7JJakdtJDZROhmjUBK2tKCAJap47DlCW+xw3gFcNHjsFgCJQYWfFuPA7LUXWnKd5UN07Ze9BzjrW94oTYbJzcYAAAgAElEQVQHvPjoGP57kKD7NhuBXbIdlTj3VDx3GVt4TqYpSKJ0A75XtQzbP3V24rlY0objU7StTrid2uEVMXEA0A2Ihx/tUUVRFEVRlCEwOjxRiqIoiqKMGNaKZEdPnqgxw+h4iTLkPi3zFZUPUDT2HmvjzAREXfBWLgN5oDgKjyW8z9QgZ9SP6+An5jwmvmZIHVUISpK+WXDf1myBe7r9AkgKXJe+JM7zwBG/ceybexCpc001ouB+Qm7rgai8HLm7s1V0AKkoVTN6HLu7G1ElJRG0KbUDsgjnIsolIH0FKfjIVEFSeLoNEXm7WrFly30n/cqxb+o81bF/OAkRd4++ujh/fc7dFMUAH7dgu2PzjufvmoJtZG7ZcQbqRZIG53eyFJVYJC5QPxVJo0S2nLZ1oFw8CZJ4BnZ6Z8xk94gn24K2soxZQtuSxKailrly1iso6nKie4RZqtY98u3hXe6RwKWz8/OzPY05MD8MiaJ6VpdjN1CunJ4YJKBlM3AjvNAC6emsKZtdr5kNe2hAHsWc44kZP6djUFnnakhWZWGK9oN6JYnxGDt/wv2PSeoI5P/JJejxOCnlcrRINut+nkkVkD1TtGdMVQjzY28V6ToDp6E2z5iP8eil7WLqSlHHzVshG0oPSfh0nsq5yFOXoCi/CeUoZzjitwgPCZQZyO3nr3KPesuRHM0Rwfx899E9nA3TcymACgRisHMkY1q/z/UYnnt8D73QP8exl1E9/Yn8dzMUGZuiez/ehui8Ctwq0juXH8Log/5SnMfnMfeUQ5PR8RKlKIqiKMoIYoqS6yrDg74SK4qiKIqiDIFR4YkyaZFwS97dWRQtxRvZe/qPyd4LyaR9JZIuDriKOZEmRwGxhLfj4psde9m6Sx078n/hWj/sn3AHr/8c5JKybyPrWtsqRJK1v4S6MKdvvRb16cUvhO/VUsMpgmXSvHwSwwRtKbFsMiSV1V24zkl1qMusuUh++PPNyxx76oOQBequ3+3Yu25AdOCUz25y7HUrUN7z1BTHLqlD3d/f+DnH9pECcm/1iY498dn88bXPk1xx1ATH3vsMpMK+yWjrljjCAO/7+g2O/R87EBGY+RbOc8WNyx37l996n2Nf+e+POvatv7lQ3AhVQ44tfxZ1CCzpdezwS9WyL8fN2D2oTERkfQQRPIfVtaK8FjJuejzc/35KFGgoyudT70Ldf/7IBY59zdlP0tW+5FjxVePEjZs3nisixRFPFGwnJ56+0bGfexH9HiIJs3UCItn6XsZ1XiyBtMcEPaTTJMlsRVs9pd1/47XsqRlUFqRTtzViXMJQmou2cQp1udelv4ZC0EgSS/vdH5UTHnKPzussx96qvM1HX4Si6Y5Gu+ufz7fV0O4Wrbtwj3E02i7aqaRuF44fvwLLEiSAjszWQ3pKjqt07HQ/nfQpmDXHuSdLjQTdJTpmYKmBjbtHi4aPxDIDXz+k7PRSJFCdeC8auHcZ2nfsMdhGatsfcN8ctnCnY6+rRNLX6r8gwtZ/HO7bq+Y/69i/+BPuoa8g37FMeibfN+YfGN+2IzEfQ1sga/csJDmR5EfeYidYDRk3EHCPXlUOTUbFS5SiKIqiKCOHFV1YPhJojyqKoiiKogyBUeGJskEr8SkFiYxdnRS9YShCyWttXGgHXK99c8jFWnDj8154nEiTo/BYwvv74occe/51/+rYzY0Ix0ieAbduqhGu5FA33k/d6pI/CG1N+2gvL0qsN/0uDFHTlXk39yXzkCDu3mcgk81aiP3O7noZ5Z854QnHfuDoXzv2hWdjX75Tylsc+9U52LwKgoIUSad9M2Gnqzg6hvbgo+SGlvs+ku/73qMhvXXPwbHhDroQveanqjDwv+tc6tgnjYN0+buzpzv25gQil6JTcaKNMZTn3Lc+k0yK5BBSbBIJzJuQS+7IX894ZHChiPymGlLoyaVIXnlZLaTL2udxwhgpwJxw9fE2jA3vZL85hr5kkrXu0oEtJPA0lPCR5ybv95YrxYUylOQ0mkbEGCc5LPG7ZxMMdbvfuIb2gvNTcGM24i4JXb7khUFlD21DBOzSKZBU/xmHRlOzlrVCigyjqC/fJnoukHqVrnR/VLYd6xHdyQkjaQi4j/k+j02OFK6J82VKcWy4k65Daxtik1C+7eqZjp2jpLqRVpIQeYmEh6p02sRt7h8QPo/1FQOSbQntPcmEg5gbFREMdoaSWlZuwjF7z4W09+7xSJD8Qws579Ra1JfrtSM8z7Gvnv9Px15B+5RmqtyjWvcsy8//SCvtT0njkSIlvyhBLP+ZIjk41wtpMZs4eIu7R9MGxGMF7VFFURRFUZQhMCo8UYqiKIqijBxWjOR0A+JhZ3S8RPmtI6nZTtJXeI88LzmPbE4YWT61V/YlthNRKrwXHifS5Cg8lvA2f/RGxz79k1c79vu/+TfHXv6DMx274wjUl+viJ3dzzy7Ux5JkYtpQn/bF6ITI3/NSw/J12Nds1jOIIuvcBA1o9gaUL7//HY79yzPPdexPfwDS050/Q5TKnMuR7HL77XCbx0+EvjFxaqdjd3QjSivdhbpnqqmPezHVgu/LR6d9bCa52DuQ/JRlom2tdY5dWQb3/4arEDG25yz41j/yMUiXT1yLZJ/1X0IC003fWOzYyQ8hKogJb6B9xd7ZjvK/oz7Zd3XKviz9+Rdcz5eqxLjfTLKWXUSSRhl0w0wpRaxR8sPYdyGwZi9E+ZbvUmjRg3Rhj6SHUoiesrQvH0ejbfs15EdzNMl89Thf+12QTlPHY6ybH6TovGUwo7PcpZMS2vcvUc/JTN3lvN+/sHRQmaG94v6xGnPD1EAyS9Sif+PTeSM2mGXbIOdl6FHkd8+hKseetsm1vCNBCW5J3uxNog57NkCCrT0rL8XzH7m9LZjXqan0oEuhwhOmYWnBnYvucOw0ybErori3WMquDiI6V+TzjvXIfSe5tKhY3mSuPwr2QIRn/5bBkasiIlmSJQeSB4uIBPpRHvsvel7uwHj84KYPOHb8NCQKveOW81FO++VVvBNJWX/2MKJwpzxFDcHqjSIuOfNFERHpTGMcn9oMebCmBtfvW4vIVF7GwRp3CEGJ4sejWRkDjI6XKEVRFEVRRhRdEzX8aI8qiqIoiqIMgVHhiTIJn5RsyssnqYX97geRa9S6B4aI7YcM1L+1atDndWtwjs7D0XTeC48TaXIUHkt4T/8SCTnPf9eHHXv5Qz9y7JPvQWJPt7qIiNRTfboW4n22EmqadC+AvFE6J+8TDlBfNFA7ZtUjwm7LEkgFS+ehgXsbkATwnm/BDf6+f4cMdv/Pz3Lsiz+DLHx3/Rn71XXvwfkt7Us14ynqv5NQt/pX0I7eXfnv3r77EqcsNhHtn/RHRFdNXkxS63ZE/Bx7D5JBmi606YFfQlL94s/uceyffvsyx776h9C7fvAAEnUy/TPQjvAf4a7Pnge/fODRQtLHd+F7FWdgDJgFNUiw2RiD1NERQ9ROz3ToRyxv+Ckirv6/MJY7NyGh68QvIUKRsQl3ScyJnOStKimozrwPEqZtQHJLfwdFr70Hx8guHJM9izasI8p2u9clOo/2qKT9ABMz3HUPf9/g81RNwTU700goWV2DOdNXibkU6KE93EgpC7dTYkxSJUt73deSbL/lMNfy0lZ0Zpq/SllBcxdC3uxbnpfZTA7Xn9qEz7MhDBQf0zdtvGN/4kZIcrzPXMU6JM/MjoP0vqeHpOz1MMtOcU+2WVHyxjrUQPQh7+/JZI/B8z2boiS2YcyBsj9jP8fEsXhu1FyAeyv5HCKhyy/AEowQ7Q2YW4779pQrse9m7HTcZ43PYLkC8/hv85Jx+V5cv6qeIlYFsmzuZPSj2cHJV9EH/VNoX0yvvQlHGCsiOc0TNexojyqKoiiKogyBUeGJUhRFURRlJDGS1Q2Ih51R8RJl/SLJmrzbtHwl5A3rEYXnVR6dBRe64YMKXvH2c+CODpciaqdvFuQB3guPE2lyFB5LeH/5412OfeGyKxw791mPaEIq5vqUlsPum41hiayC+7305bwsGK+DA7GuBe7mlnocW5PEhbY/RpFWC1CZiZ+ENPTQjyDhXXbt4479p+tQnrkYfZamCCE/SSPtn4C7vjSA8eg/hqIVCwn3xkUgtextgHv+1Ksgfc0MQ7ZrIZnm2WtPcOzoVJLB3g+J6bZPIvTmkp9Arrz7akQipj7uHnIUbkKbYmf3ObbvFUiz0WVR2ZfYY+5JL1cZlGcp6ouTfQYpUWmKkmRygsbdNyJCyHc+5syeG5C0UxC8Kb64u7M5MyE1qMzQ9dN/rnfswEyUZykCNv53HGMWI3zNPk+RWVBsJTrbPQlngPYJzEZw/tAe90yoNjhYDuEIKamFDMZ7B/qpfT6SSPn+jGEaSpqiS2MB936MvWtwFLCISE01ZF9OXDoxgjnT9ldEEXYvLvQN1bHrBIo2rsb3UiSDTa/DkoOzx7/m2O1pPAvmRSCDnVO62bGfTbjvcchjz/SkPWSoswYXhVvdpdvgVkh1fjzqJdSHZ230Qtxvwc1oR9ejGJzU0Zhv3U9hX0pLf9FydN+uuhfP9dJm9Gvm7MH3gYhI5rT8+H1o4dNO2Y13QLc/77LnHfsv9yNalBPjZmjf00Af7aPXpwLQWGJUvEQpiqIoijJy6JqokUF7VFEURVEUZQiMCk+UyYmEegvvc2d0uR6Ty7m/73GknmmEq1g4OWdB2hv3OCInemfDfVyzBW7Xsm8jyon3wuNEmhyFxxLeI3+/37Fn33uNR11gcn36ZqI+1dtxUNuJlKDwpLx7Op2Gq7wvjmipujq4r9u3QsaYcTj21GvZBVd9462QgI765BrHvv9HZzv2RdchOu9/VyA6j8mR3DTuNiSn61jM0Xlwm7cdk5dpUi34XlktNJWt353t2K9VIYlksBkSyYzfQZbY1oMEmD1/QTLBi0nC+9N3MH4f/tWfHfunD77TtU2JSZCeqp/GvOqnJH+lzxSkBgT+SWopxoCZMQ7zek8PJMGKEuoXinAL9tJeijQFyj6OpKEtWyBjTPqs+35nvO8dMyCVsfzhpz29Mucg2i3TDEnF3097iZ1Abd2LcU8fO1jmFBGJNLo/buKz0QeGovNknvv+a4Ho4HUdwcMhq2V3YrwqTkCkWccm3BO5Ce7SYtV63E/Wj7pE3IMuJfAP98jbWCP6jCPl9qQw5xPnQUoO781fl8e6cieeSzk/pM2qGMqbZ+E6y3dNoeNxnrWvYZwezJ7u2CaD83wEwWuSPdf9GRwMuvcZM7CKIlHnvjFfeiEll02QjEvLGSJPYfx6DkOHhBdhjIPrIRnb41BewnV8rNYxx78bEb8sr4aeL9od1MGszI/r3XchSWealiQ88z/YmzR9Ec33HbgPQp10r5S5y3wHGl0TNfyoJ0pRFEVRFGUIjApPlKIoiqIoI4e1RtdEjQCj4iUqkBCpeS3v4uw5AoOcpiiUbAblNuceqrf9/Tc59uy/XuXYFa/kk+yVfnivU9axDlJI+wVwJbetQgJD3geJ98LjRJochccSnlddIpuQ8C94OZLEJbdBZuuYD3kj/BpkvsBreRd2jlTLUpIH22bDhx/uQN137YHcFWiHXBEfj757YjOSzlW9B3vC3f4s9p9b+6GfOPaRT3zSsWfeSYkhv4bIOvMokmDuxLZX8o2T80kwr3sKiS6nPUoJRL+NdqTXwD2eGgf5c/PTkO2CPZTAkALDbv/rMseuvQL7aP3k0YscO9Tv7t42pZAFfCnUp64Krvv+XLnsS3bn4DIRka170A6ToX3CaA/H6856yLG/t+Y8x06TVNYWhT1xJtq0ai0kUDmF2uGRbDM1IGeR1MzBV++bgajI+1qXODZLfj4fJYKdBsko1kn7DhKJie7S4szpkNw6Kfnom9ksNbEDN0Uwhu+xRFoxG3JwtAGRnoayYfZPov6g6Dx/wv1R2bvIPbqzl7YyLFJQSMkpSiw6ZfB5EhPoD14FPvc3Q9p7z3nPOvY9qzBOfJ+3v4v2A4zT2KTd/6AGVrjve5f2e4wHgl0lV9jzsWKLe38FaU/KTATnC/ahjp1HkYTXgj7q70e9stXoyFQb5ky6B8dXknS6dQ32FS2bhXmQ8pB1B/YAbL8c0cZ8H3LdA2txz6dqUK90Pc49mfYarSqBpLnL9erKoYS+liqKoiiKogyBUeGJUhRFURRlZMmqnDfsjIqXqHSllaZCpEpwM9zs7AZnUcJr56HZj13p2NvPvdWxm87KSzD/Rnu1XXreasf+Y9ORjt3+Ety+fXPgVi6figiQor3w2MNN0ghLeFyXnrOxz9JnGiDZfOjMlahPM+qztRmSWGJAlTO4Dru7hWTOIAU2ma2QEOOTaU+v2aiLbaE93KhJ/hhuumOehlz54pk/c+zfHIv63r/7GBz/XoT8rNqLdnz7lbycduJiRJTtnIZImngXpJnsdNqvi9oX2U4RiodB6pgxC9LQntWQ/HqiJGPQBGL3exFdkEx6SSnr2gzZ1Tdr8He99gzLhXEsy3nMdx5+r2NnaT/CIsltNeSp5okYy1Cnu2z3wMU3uJY/0pcfs7ogZLg1UYzRfS8e79hfOG2FY/dTdtDHW5Essj4MmbOpgu5hxkMN2tWEqDnbj0eSv8JdKktNGlxuAujf7CQal27UN76JIrpq0HcWt4eUdEFWMln3PmUWzWt0LW/oxrVKKQKzLw5JOmYhD02Ymo+IS2Vwza4OkoZp7mcr0L57XsI4zZwJKd1HCVIb2kmeo7nkK3Pv376ZrsWSC9Ic9ngIB7rz9e+b6y6TcRLbBD2LSlpoL8MKlCep3RztlgtROSWv5HurdzbtV0dJXPke+sR7kERZ5MuOVXtqfqlFdz+eGzaEc3SdjL8NoV2YQKEeWnbSi7kXXYvlI7GDF5ynjACj4iVKURRFUZSRw4pITlMcDDtD9u0ZY6YZY540xmw0xqw3xvxbobzWGPOYMWZL4f9r3uhciqIoiqIohxpvxROVEZEvWmtfNsZUiMgqY8xjIvIxEfmbtfZ7xpivishXReQrr3cinz8nFTX5KIjMdrhGWb61tKeUV9BOaBNc5T3vgFQ1KZB3i++JkgxHr3Z9SfLnE7xnmd/Lf23dbY7CYwmvygf3cEMUlTi6ArJAXwrfZXnIRgouZIpUZNnHZMnlT9uO+dizTn0aKYE7P0N7iWWjkDQCdE7/Fkh+nRQBdkYZ9uy6pQfRfPUTIPFwpGVuT74PEpNxnbnV2POupQH94qMouRxHSPEcIDd7iZ9kGjom1YcO4SCjnMs+bCIivhQlnqwk130X5JbUuMHRZvsVUMbbtpFcUbaHosRoD7cMhWAGECwk/j5KyOkhEXgl14tm83MsSJMjRpOmhPY+YwmvxIc500lSB0fSJTIejxUvGaMX84Cj/7I+9/PwHn9u+AMYlyz1b4AiMTNV1C8kcVlS8Ph+yrk/IiSeCbqWZ7K40ZJptCOTcf/dOnBMhpMKcxQyRdJZqm+oleSxqbDLQ5AQM0nqR9rz0muuZsvdB8qSZOr1ZWceht0jMdMVNGfp3k5X8sOe+53DGUm2o2K+h3j8+L5l+B4auA/2pSyY77/muHsy1UgFIuzSJbg/eM7wczfUS39LvPYgHHGMrokaAYbco9baJmvtywW7T0Q2isgUyW85ekfhsDtE5FL3MyiKoiiKohy6DMuaKGPMTBE5RkReEJEJ1tomkfyLljFmvMd3rhaRq0VEgvXub/uKoiiKorx18hsQ65qo4eYtv0QZY8pF5H4R+Zy1tteY/Rska+3NInKziEjJtGk2tj3/IhVkeYUHnOwiZY3+Me4M7BHHkW9buvMRVc8dhb3t/rP1CMd+4IjfOPbpW6/FuUkm6tmFF736NahL+zmIHuO98DiRJteFJbwnFz3s2Nd3znHsPxx+p2Of+09EjAzEKKZpH6ZshGVO2P1zaT8ykj/8Ibi4o5tI06SuDlajTYFGRBDVnY7+/X4T2vQkJepcc9YvHfuK7e9y7PtO+pVjf2T1x0VEZG0DNCtfI/ruigueduzHmw5z7KX1Ox374d4T8F2KqNrag+jKALpAaich5rCnCclHI33uzljfVPj8S1eiD6JHw41f8WpYBlGdGlwmIqFGyAYcXVa9ClJAz+EkG9bj+gGSK6LlpNOy3FTq3o5/3fBh1/KLp60VEZHmJOb19AgSAr52PCK9/tKMzJEdlAzz9CnbHXtdFyIh51UjQpKxZR7ySgS6R4YimiK1cbfDpaxqcHluJSLQ4ggalKqN0He4f31x9/7KLYQEnerFmJmQe91jqZBr+XGTG3BOklB437aVjbhvBuQ8S8+5I+ZC4k+SRJrMwm4eh0jWlhb0QXMS1zliAerSHMXxNWH3/o3s8YpKfONoxdzArRh3P7ZsL8m+tGyhDFWU7kX4ro+WGbCsXULJhPunYf5UbkLfhHqoz8bhu9EFuEdL/e7367ZX8s+RinnYQ7KXJMREA/qxaifVneaYLYfdP4P+fpGkKr91vbxyCPGWXqKMMUHJv0DdZa19oFDcYoyZVPBCTRKRVu8zKIqiKIpyIMhqfu1h561E5xkRuVVENlprf0wfLReRjxbsj4rIw/t+V1EURVEU5VDnrXiiThGRj4jIWmPMQObKfxeR74nIPcaYq0Rkt4i8f39ONqBERVrc5cCioIIilY/2XNuORIgfOOMlx15UnpehvtEGWeLb49c69s09cKsHe3G+tA9uZVsK12zXQlSmtBzSV99MRCvxXnicSJOj8FjC+1ItEk/e2TvTsTn6bqBuAcPSJrnHSdrzdVLUEMl8uRANeSn80/4o2hSOwMWdpYSADbshg51/4gbH7qJ2f2zHOx37O9Px/vyrjtMdOxrNy2CBIPo0RckP79l8rGMvnABZ9IW2mY7to/3OWPXNUptyATqGDrLUBf2T3CORsk00liQFmHbIN8nawVE2psNd3kmX07E0HknkGJVwE+2lloVsZilCk6PmkpMhCwa73H8PXTR1vWv52eX5ch+FOfVTCNqdOxB+edJRWxx7XJiyuBLHjdvtWs4EW90j2dI1qPtAskYRkUyVuyQU21sxqMzUox2RMtyTyVpIrgEP6ZYJrEaCy8wEnDPY5P6o7I667xO4Ng2pOuCnugUxZhzJFe8onIfm8tr26Y7to8SjuSSHEML0l+B+ylEE37Z2JDOdWg1ZuyvhXncvcvvx18IUqmA81t5kSAHPlPFziaIGaUmHIdnVcJQx304ccUhdw/cn338ljZiHZy9zvz8GnhGGExvTvn9lp0KyjrbhWV/SQdG7NL4llCjU557jdMSxYnRN1Agw5Jcoa+0z4pmDWN4x1PMqiqIoiqIcCmjGckVRFEV5G5DTNVHDzuh4ifKJ5ErybtNMqfsh3nIebH8VZCjef6674La+Z/HtTtlN3Yj6uqZ6j2N/r5b2h6qkCLc2SB2VCEqSvtnowurtcP12zHevCyfS5Cg8lvCuqETiye/gUpKuyJ+f88NlK9yjhsooqqSvB27oUIQSbO6GVMdRfv0xijKkGeKn5HhPt8917O0tkPl+v/QWx/5t11LH/vaEZxz74dVHi4hIKoaT+2Jwgx+z2H0/sgsnw/X+661woXOUmi/BSQlxSJL2JDPUZf64uzM1R7Kgr5v29SJZwN8/+LtmQnJQmYiItGLQAjQGoR6U90+mSMswJxkk+a+OintRr3S1uyz5p4bFruVVs/LRf10ZzIH5YUinlZOxp96uPkRx9lCCzfHTEMn2zxZsMHjGpK2u18yWuCcZ5P3OOIli1iMxZd3MzkFl3WvQMWVh3Hsp2ggyQZIfJ1Plh0hyMSLWcgnMmdxE9yiuXNa9jhMq0H+c4LAiiOjOhkrSpwYS+9Jcnjm3xbF7E5gndaWI3Ny8DVGRuThHbsKsqMO+n7E0jhlfhvFjPPJPem+YRwzIaT6PKNVcC+1FR7JdUZJTiizkOeMr+htA8jwl5LSGZDOK7DOlfA+hc56PY0nFyXR6XyHpazyJ/kqRvBvvgKRciaGW3jlUl0rc5/EwRRwm9EVmLDE6XqIURVEURRkxrBXJ6pqoYUdfiRVFURRFUYbA6PBEZUX8sfz7XHw87wWHQ4renz28yiXroAVubZqG00fyLtbzn0XiSo7W+AkrYpVwx06/C93Tvhjvm90LKNpmFaJ52k7EicKvwW29tRl14b3wOJEmR+GxhLfp4zc69uLn84kTqx+qdMrmXbPZsdfdjejDaCV88hHaU7BmMyJTOinKMMsRYLtQ9+gMtLVsNco3dyCppW8cXPcfue1zjh2fifJ72xDtVVmIwOxdQEkn50GiefnxhY6dng15ZeUeZFEMcHQOSUCZiLtEkFkFSSo9kZIuesgOpbtoT7JFqEPpWvRBfPHgZIUTH3SPzuuvoz3cXoWE1nU4JU6dCCmwchXGzJ/CMX0zcM5IKyXhnOku67Y3VruW37TqQhEp3h+S97YLxnDu1mqKoiJ55cnViKJkpeeBckit1x8lrscwvn736CrT4q4r9bxSN6gsS5Fe7TsR8uifSvJrHfqX99djSY4jU6NZil7z+gW/0339wfZdiKzj5xWfpnQeZLb+9vx5DEmIu5oQVWdJ+u4M4f4f9yJJzTzfIyRRbsKmEZQrVbr46Y/gWU/ZlSV/r7HMFhJM+ryUvyOgffG+nNlJGI/3zkTk730vHu/Y0x6hZ9HXsAQj+n08i/aeRvfTdEinVc/gfuqbiY666b4LHfsL/0XtqMhfK7COIkGraKkHNTC2lDbja6c5m3SPLs1VH6TwPGVEGB0vUYqiKIqijCia4mD4UTlPURRFURRlCIwKT1SwNC0TjspHorStmuCU58gbypFWXi/TJV1wscZxGjEDiTIN76tESTUr8L1J85BErelKyC6Rv0MGK52DkJ/Sl2nz5JPIVf0aZJQEqQ82wrILGliU5JPqMyDhiYisW3qXiIj89UjUpSENl7/vcribo2m4lVenISdmjvkpx0UAACAASURBVEJEUJai80KzUHfzT8gFZdMhOZStRHl8PAakrAxu8wxJIIF21JPd/xNeyCdszAVwfZZdqrbB7grBDV9K+27Zd3Q59gdmv+zYd21Z4thz6joce9sKRI9NnoMx3rML/ccUJQJMYZx4bHLpwe765Me6BpWJiHS1QRaIVJCstAHlmSjmJ/eX9ZMUOAfjlOvGd8fPRVsZE3aX+cTk686SOd9Y/Qsxppb27hNKbJggWaJ8PeZbcj6+y+Q8o/OoCiQvcnQVE1jYO6gsvQeyOu87KB0oT5HU4p+IY1iaKSGZL8blYXcJJtzmEd1Jqi63j8tn1GKubN2cv7f4OdBfRjIjJftMTMYJxz+Lce9Ygrmcwq0q9a9iPPonoAKBhHtE5xcvXu5a3pR2l4ZFvuhY/oI0e+Qi9wjbwysRAeonHb4ugEjBW269yLFrlqF9bf+CPtj9CnRt+wGMjY+iHzkpadeJkGnD5bDtRswPpmZ6fmwSe/Dw9lNEZ5rObem+9dFY+/s4wSbPk4PzZzefbFP9JsON9qiiKIqiKMoQGBWeKCMiAV/+VwkvjuRfpVzulSc9WU3vhPQrxxZ+PbMXIRDFSXixdyKNLrlkHraGWb4OWUQC9Is9XodrpskzkeOdKditQLlv0uTt4K1cOE8LLyIf8ECdW4pfXqeuQb0umfqqY/9+OzwyEdqaZumUXY79RBcWoodDOGesjhbj0qJbbqvJ0mJnykMVoD72J6hNtCg1NiXvXfLRmu62Jnj0auCgkmAfjxPK2WMQpJ9/vL0L73zPY9+fgocs0ON+C6RpEWlRPqYaWlzaN/i7503d6Hq+x33ISza7Cr+uX9pMi+V7MX84wIJ/PIZoCxHe1iKUcV/EapPuv5MyA7uMsCeK5mlZJX7Rx/rgDeT+DVK+q3QFJq3xe6wq9rhvrd+62z73L5QEM4PKeIl/hvrCR0MU6kZfJIO0aJwu0x6lSUZ9F/VYJOyVU4m3FqLURZKjnW920HYsfhevm4/mF28Vwgvx+xYgYCI6FedIVaEfe6ejkukK8mq65Dl7PaIZnMfnsbJ8wOOyq6fG9fNEFh0Q8mHyhfwY0/rVeF7tPQnj8cH5qxz7wb8vc+wjl2Jbohd2z3RsS7n96ucj916Ucm557XyTTOfryQFI3GJ+JgR7MR4ZykfF3ifD4zd4+h4wsp6bjChDRT1RiqIoiqIoQ2BUeKIURVEURRk5rGh03kgwKl6i0lm/tPTk9a/ynfyJcTU9z0MyULiFdoMvzzvcivKfkFSQrYJbedlkbFlx7zMnOvasZ+Bibjgc3VbXAnmlL06Lz+lSXBfOB1Wcd4XqQ1u5cB6ogUXkLOE9c+QDjn3Zduz7fNsR2FLmfc9e49hPrlzk2OUNtPAyRNsY0GLZvqlwp5fGSD6ihfn9te7ty1DOLSmnrU4K7u9glXu0AC+kjs9Hv4d2oy61IWiB9+8+Bt9dB/mz5Ey48HnBboSky56AuyzhtRUJS2tubvnfvXr84EIRKdkOSey5KZA6xu3AMV1nQEIroXxUfpI9a4/HAtzWbkig9WUx1+v6o+4ylL9wKVZlWGqK95NO1YcPihZ7V9F2O1THQMBjMfsb7xoiJvfG93xv3+DcTGWNaGeM5lU19W/nKahkaQUtQM7h+FOnY0+npjjmUnVocE4wEZGaze5tTVbxomKUp0hOS5M06yvEC/hpTX4/5WhLUV/nqjDxwm24/pRmlnoxfuFW1L1nPgWT9LkvLP/t7hNdy/37se1LprBdUibrIX9SX+d87oEwu8/DfZ6laX3HM6c5tm8a6vLMSuSV88dpyQHd2z0vIFfWxFORY2pvglbgE6lN+fLSZpJIaYlGvBr9nqEpwEtDkjxNuatV/xlTjIqXKEVRFEVRRhKNzhsJtEcVRVEURVGGwKjwRFlrJFmQwkJhd3f+/ki50XnkN6cvmML2Ahz5kwyxjgFzdRe2EJi1sMmxOzdReT12V2+pR56Rujrk8GmbTe7s3OC65KtIUYERd1c5b+UykAeKo/BYwrtn9t8c+4M7LnDsSCkksapxyHHVGkMyrYmTkLOmOYNtM0JhSAddh/OWJhSFwpKRz/WQIkmo/aj88SxdhPdQTqkMRS0245olXei7WArl9bQbfWstRfyQ3tY/A3YFSQolne6/I2wtST/rIMVFliHHVOyZetmXI2fuGVQmIrK5DMceVY9zbNmBXeRztOVHfBJFAZJsOKcE+Y2aj8F8S+Xc5ROhrWSY1MA2Inxf0X1w2KRWx95dAvkx3oO+OGwi2rGZ7re6cnfpy3pIpzWTe1zLu7vLXMutiwqVrCFpnPI7xWm7HT/Npf6M+3Pm8Vcgd0fq0NfxNtJmlsJMVHvMH68cd1Qe64RkW5kdfGzpbtrSheoYpAjClhNQ7qduZ2m2vBT92E/53Yy7micVIfc5Uxpw3yKJGcgT1dvpPnbRKOZPdRW0ur4Yyie8iIrtOQcddsoxmxz7td9Awjv5X5An7o8vH+3Y5VvQCdkTkVushu6hnRPcQ+XShSUe0SA9Kzqo75Lu5bxEI8d/XY2HfYDJaXTesKOeKEVRFEVRlCEwKjxRiqIoiqKMHNaKZDU6b9gZFS9RxlgJFbZViLTDVc3J9jy3fSE7sxvuWz95pAcS4iXmodBHW5JUzYCccFIdwnnuehlRKrM34LtblkAGq0nS7vFbkTwvTC7eIEWYZEkR658L97ivE/Upm9ft2NFKRK0MRLBwIk2OwmMJ7+5ZTzj2gj0fcew9HZDqyjvRec20/UmkgbaYmIB2VOxyd1zGppALexz6yVKCwtmzIYHGnp886BzxCRSRN54GdQa5/CfhfLPC0AI3NyHypnILjmmci20qWBrprCZpptRdYrJxHJ8ch2OiuyFt+esG6yEbmiYMKhMRMdsgb6xNYKxr2igJ6AyK6Eq6RwfujdI2Q7TDfOcx7rJEYKt7NkFb2KXeR7JWhqJCd7ZjnqRa0V+cQLUoWeQOXKf3MPe6mJT7A7xvI67Fkl/AY9sXN0WipAOF0QmYAxHu3zmQ+/20/ZKlKXDWYiRL7UqhTcfOb6Crfcmxylrdo/P66ynhJ60y4PufE+8OHGNyVN9pOLePos4CUyBHVT2EsSlrwvMkU0oRsyV0HRqa+Dj3+3nLnvGu5Z6cCjNbiM7zhTwiNIloPyS8NEU2dx7OSyFQ4Wc3Q/ouq8Z4/3njYseO0N8Avm/NJtwre0vx7PB5RK9Wbsrf/5QPtCg6zx+naEkKZE2yPMjdS8s4OPL37Ywx5nwRuUHy+5/92lr7vX0+/5iIXC8iA2skfm6t/XXhs4+KyH8Wyr9trb3jgFTahVHxEqUoiqIoysgyWqLzjDF+EfmFiJwjIo0istIYs9xau2GfQ/9grf30Pt+tFZHrRGSJ5Fferip8133j0hFmdPSooiiKoihvF04Qka3W2u3W2pSI3C0il+znd88TkcestZ2FF6fHROT8EarnGzIqPFG+Xp+Ur8hHuSXfj5dJ1m85IZ71yPmWiFKCuS1wFTvJCilqh5P6cRTQrLmIOPrMCZDElt+PKLil85CQc/tj2BNtxuGI5ttFu3+brbTnFHl7i/YYo0g93osusgntWJ2eli+jvfA4kSZH4bGE99qpv3Xs69oQffTS/5vn2P+y4q+O/d3rrnDsj1/+uGPf6Dvbsas2wA1esYv6sgxTKtwMe3tuomNXV+bHcvIfd+McO9Ff6SqKqqHIuGwQ17nj+t859g21SMK38i5InT9Y+HvH/tS9n3Xsb135B8f+4qtXiRv+Cmgwk/9IY3MNpN+qnxUS9X0e3/vusQ+5nu+VBdh1/uyK9Y59deO/OLYvwXsT4ruc4/Cr8/6Cum/+P459w+L76GrfdCxO/sfYQjndVkX3xHeOftixv/TXyx2boxkvOwd7md219QzHPnfWa67XHPeK+2+2bgRaSekeHBM71j3Kr7w8MaisbzG+N3caIgt3tSKq1hfFfJzwV/dH34uTj3LsICWX3VQ137H/8wc4vmu++3kyvDVfUTJG2KX1kKqDsfxc8tPygIqtOHdJJ8pTuxAR3LUQ5eEukhBT0IAzEfRN+V6SCNPuc+P42btcy6eEsczAa++8AakqHHGP5JtWg3O0xfDcjVQhwrb0ZuhmWz6KZ+dnj0f08e0vX+jYXzgOz6j1C6Y49ob/PsKxT/3O8469pAxLNr6w/cOu9SxtHSzVJ2jv0JJu3CuJes7GS7JdCA97jtwO+jEG7j09ZqgzxrxE/77ZWntzwZ4iIqyRN4qIW5bX9xpjTheRzSLyeWttg8d3p7h894AwKl6iFEVRFEUZOayYA73tS7u1donHZ24V2ffN/I8i8ntrbdIYc42I3CEiZ+3ndw8YKucpiqIoinIgaRSRafTvqSKylw+w1nZYawfklVtE5Lj9/e6BZFR4orKlVjqOy7s4fb3wg7PcVeQ+9nAl+1rh+o1PZt0s/3/BINyo2SDOURKBdPPzzcsc+4Gjf+3YvzzzXMfe24DxMwvwUtyyCwkVAxT9V1QXem31UwRLLoShCFF9ajbjPJmj8udZOgVOYN4LjxNpchQeS3jfqIeUNOfTkME2JuAN7TgCbXquczba1IfKR6dTRB7tSWhKqU0lkBcClXDvW3++re3L0I+xybhmAMFHRUkDs1D25NqGix17WgQScOOZuOatbaejvjNw/pv2LMP5OVqKyFJkYfdsVKKvBxJEbu7gLx8eanY938YEIhJ7c2hIphxSQAVFFsamUv+W4pibGiCb8W+vXzejredjyJykgYMoRMHZAP2oC+A697bRD8hKzMc4RRm90DkTVaHTbOyGdMsURV0SvhTtt0gymO12H5x/nf/0oLJbtp3i2EfXNDr2tnL0e9Vr6N9+CkArmmN0SY4Ozgzeri9/nonuz6JchCQeTrAbRHk5LS+ITs0fw9GP8Ql8bpaP6H6jyL+mkyk5JwWdBfs4MoxO6fETuiLgnmwzmsXz1WsfPX8hojLksX9iT5IS1wZRmRTttReiJQEVEyGDtach86Voy7stcUTENvYjIpfv2zCFSPI9ZEPuGUdbTs63r6SDohxJGuf5EIjRspMQjbVFO9p34nnsSxxQb1ARoyjZ5koRmWeMmSX56LsPisiH+ABjzCRr7cAamYtFZCB0doWI/D9jzECo9Lki8rWRr7I7o+IlSlEURVGUtwfW2owx5tOSfyHyi8ht1tr1xphvishL1trlIvJZY8zFIpIRkU4R+Vjhu53GmG9J/kVMROSb1trOA96IAvoSpSiKoihjHCtyoNdEvS7W2kdE5JF9yr5O9tfEw8Nkrb1NRG4b0QruJ6PiJaq8NCmnHL1ZRETW3Ie94tjNXrQXlUeyTU6wmT4ekT1lkfwHvRuQHNCWwY2b2gE38dQHoSVdePaXHfvTH8BY3/MtRFNO/CQi9RpvnevYLF2kZ6MukRK4laObkLhRSLLJ7IZk1LmQ9mUqlD/RhT4qb0DH8F54nEiTo/BYwtv2wZsc+9TPIkpsyecRXbX7J4hKMghcktrjEMUY8KHue7dA0kxNQFtNF6SA3uPz0VXfW3q/U9aYwtg82wU9KpHFJGBJt/fj6LvHz13g2J+8BtFrf/kXSFwT/xt72vX/+ySc//+6SxfhndB15l622bG33I/+WPR/9k1pInL5j651PR/z5w70lz2TJV2SDijxpZSQDP119K95Nw7p+iotEWC1K+wuV5i+wq3PEaJJPA5av4ExsJdSVNIkRJR13zQd9ToN1+m+g+pyJszIae2udUlsxNhnZuJeKV3nnij0By+dN7iQ9mZ8eD02t7PjIBllSV6uPAeyK0dL7W6G7BIuhQSdirtLi5G57vv+Jakv/dTHWUqwmV4D6anu1OZBdWmP4jngn4/+jdADsHcvnl1XnYqBr6KN9J7swJzN0IO03EO2e+6Bo1zLQz0ea3ePgzlQtf61Na6H5tpR9xSaX7TvX8VXIcf27cAz7a8PIatn7kLIfC/8CNJzugznn3sF7tv7bjnLsetfwTPevN892ea7T847OXbEMDdfeW2mYwcr0Xe5nRgnXvIQoIScZY3ou2AcNuIElUOVUfESpSiKoijKyDJakm2OJbRHFUVRFEVRhsCo8ETFO8Ky/q68RDXpPUjAyJEEmaJkm+66bkMrXMi2BeET2Xg+OZ0ltzq7XTMk7dVdj+ufUo793u78Gfale9+/IwnnQz+Cm/ioT65x7Cc2w4XOdcnwfmAsRUZJtouQ+58iQkKz8i7scAgyWVcI7vyJkxClxnvhcSJNjsJjCe+Zn/7KsY/95r869t0/+qFjn/coskp2vQJZyUd7orF6ZCjiJkN71A0kQv3l7y9zymITMRVrXoO7PTqDEo+2od0XPfCkY2+NI9Tqzpshtf7PnZArr/0u2vqftyP56Bf/hISVTKoaDdl9K+TQ2Lmo2/afFmREqEfyyU+6J9tcGtnu2K+lIFH8rglfXtM/y7HLdvKtCXvRDS869s6/He/Y5970D9frSsrjd9JAZBgrNHR/XPY/jzr295/F3Lebkehx7ucQ6dnxBCJA512D/ecY/+9rXcvlWFw3shbjnTiq3+1oWeKSDPK4Kty3nRnIK5t60dcbOiBRmhWIIOTHSU0f6pIuRV2qE9RRFEOUXUWaFBH0SJZqaInCgnO2OPaWP+bnGEtmuXGoWGgPyuN1dL9RlNryH5F2SpS1kKQZxndjfVTJx6he74QMxkyOuEuXjC1EPdcc6S7dVpUgUWqGvCK8Vif9c4ruvBj34RHXrHXsJ1+k+fYZyOoLyvDMfujHeDbf+d8/duzbOiALbnuWtEhixb35+zLSin6vrOBoTQxk/zT0r78fbUpUoO7JwyAN+/30kEQ+4JHHHvA8UW8L1BOlKIqiKIoyBEaFJ0pRFEVRlJHDyqjKEzVmGBUvUdmwSM+CvGs5txwRP+IRhecVnVe9DC7kbnKhZyUfWROoRwgIJ97MJRB5s+sG7IX36hxs6jXncsgx9/8cbuLLrsW+Tff/iPaWew/SVrATPBuFGzhYjQgP3muqP4aEdCW7KPnoP/O++1gduZjb0AHNGcglkQYMLe+Fx4k0OQqPJbyXv36jY5955ecce/6XkBR2RxXkQnmNolOOg6SYSJB2QZx23iYREWlP4XsvvgLJbOHViFkp8WGculLoi+VfwV6G0ckUCXUx5sB/XfsJx/7Vj29w7K9cif0G7Xs9Erem0U9tZ9A+YD2YKx3vHiw3/eKmS13Pd0svyUQkC+ToDrTzSHY5DlFwWUrW+Pi9Jzj2Bz4ICe+hb2DufQVbAxZF9jG+gtTgo2gw3p/yF7fTXqDz0f7MDMgxzz4DSeXksyDtrfrjYnyXpM6+9yCiiqmgxIxTjsPdsvHlGW6Hy8p1cwaVrd+N+5bllYot6GBTQ3sgHo12GEp6GW/CvVd3BKSh1o2Qr5nEXPcIN06ka0jP4/t8/dOI5k0vyp8nRnWxMdqH8mTsOdfXCg1v4Wzck75l+G5HHEsIasp6HfvcOkhfz/YM7kcRkZ13zHMtb2RJk2+b31JxYQrFH6NspkSSI6ixEkFK6IGd+zieIcH1WKLx8ktH4gsnYvxWLcd829yAyOW2d6Cvr/oGliLEJ+B+8h3hvj9j6uj8Xn7fX3KPU/a1G690bHM85mnFM1WOzQmBE3WQ7QI78IEvoy8yY4lR8RKlKIqiKMrIomuihh9dE6UoiqIoijIERoUnypcSKd+Rj+SaculOpzxL8oJXJAdH6u3aBheyP4rIsAFFKNCCyCIhlSNIefSmfHYTbKrj9tsRbXfxZ55y7D9dB2nvoutQfvuziADxxygBG0kzgUbIWVkLO0ijEp0Bl3DZ9Lxb3p/F+fqmovKhMGSMxAQc8/HLITnyXnicSJOj8FjCe/K2Wxx73m8h+XEiuSztnVfyBNzvYVLzDAWkPP/4sfn2tGAQKmZT0tDfIFlj/zT0S2kDJK4zbnvWsbdEMe7r74EE+8Uf3An7U59y7B/d9gvHvuyhz4obOdpbsfY59HHnUkQIjnuwIJm8H9874yMrxY3zqhFZ9HQfkoO2JqFp/H016h54AeW03Zosfi8i3x7+HRKn/se3OcwHCT9Nr7uk6usf/IuUlD059wPPO/YDLyCZYXAvzld3AuSu1fdCUpl3yTbXa5Ytr3Qtp9yx0tAESXrKBU0uR4uUT4gOKrvwRMhU/2zBHJ93DJLC/uMfqGPpBtJdiACpO/FGRPZVJ91l38imEtdyPynARdKXwXyup4Sf3U/mI9KCmOJF+8P5X4B8XkVycMOmmahLC0mBlEOyZyciCO8VLJcIRKmSGG6Z/BH3FJCT9iM6zxTmkD2jy/XzighkuHgac6m8BDpf8lYkw+0+B/fbrFMRlbl6DcZ48tkNjv2RKc859g0/xo353//xG8e+q+Ukx37padyLTPDV/N+KH/wvlkJkjsbn/ucg4cWOwaSxXXhWmBxFHE5H+3xB9wS4I81oy1g+VlBPlKIoiqIoyhDQlyhFURRFUZQhMDrkvIxIuCPvit7eRvtoZeCTtiSD2Zx7qN5rF0OmWfwUIrNK1uRll/p3YP+0XZuR0M1Uwa29bgWifNgNHz8RbuW7/nwG6ngxvvu/K1C+9kM/cexjnkY0mH8LombqTodc0bC7DseUQpYrW42ItLKVef9+vA7vvqUxVLLrcLiSK3bhmBt9iNziJKO8Fx4n0uQoPJbwtnwEUXuH/QNu7im/gaRR8mV8t3EFoqtilGzuwTPz43TZC/+fve8Oj7O4ur+zVVpVq1hyk+QKBmPAGEIJsekQCCSEmgKBJISEEBJIfqR8Sb60D0IKCSn0BFKAGAOhmmYwHVxwxQYX2ZZky5KsLq22z++PXc05a80GYiwjYM7z8HA9+77zTn21e8+cezFHVfMoyOI14DRaFoPTiJ2OaxrmYaz9xO4MjMF4fP9OauNV5PL/C+jKPLt4TaSScmNtxHOrqqGS6h8zVIH0xILZQ8pERBYEUM45HqNVmOszD11q7KersQ572kBDv7oRATknnogcYz+ef56xP8MpO312GioxJrOeSQ0mtK84Z5gQtZmgdHY7OjA3ZceBNlu1ug4XgXGUtsPtgz1uIhSV4emgeFq7C22XS3RdyZCy+VsgAwx0Y41vG4N+BGqwrsJBUr3Su4XvjZajvQVb7TnWojPs6q4UKbBY/Zei3HnRZaAL4/ulaS5FAU+ZGopMx/7xbAMV+YNPIv/kr944wdjhVtCGgXNArbV18ZjaqZ22m+us5a25fnKDNZdkYZqq8j1hD0Ka6KT+UZ67eDcoruZTsCdCGzEG67ZCNajH4328eTkOXlzzAgL45tFft2/edxHaWE0bcAzzrtSPzBCHL8V+j2zFUYWy1+l4yTqspWg5+sH5L4tLsE583lwvneGHo/P2PJwnysHBwcHBwcFhNzAiPFEODg4ODg4OwwctLu3LcGBEfIlKFKek/YS0O9u7Ee5mT46gmgwmK/Z75hJjvzb3j8ZuOiLdzV83n2TK5n4Ueaueb0PQu+7n4BrurUPd1eMRPLNrG7nhc+Qmm/nM14y9+Jg/GLvjKFzzS2rPyaQuen4n2rO+fbyxB0ann6WSlE+vmwfGTt2UrAUV0VdDqrNDQMFwLjwOpMkqPKbw3joaPvybDsCYPdB8sLGPOHOlsZfugOLuwpVfEBGRrx3wvCl7eTwC/zX2ggrwTUegQF6sBdtA+3RC1CaTD0UOtaYnQSe29GJdKfKmx4vtY5bqRf09U+Gi1xtAu8rkoW55VuQwtI/qoFyD7At++KmPoF0lnHwNZtEKUDmbe6FiKuywP/dbcx+3lr/WnaYFS/2gGbb0Qxm3chnm49BDsVf4JbyiAXPqoYCSnlF2iiQXd7qtGTSJELUmOcYyPjo+pEwRbRkpstP9yQZQXAVT7Eqz8BaSxNHcRCvs6+SMfVday9/oxtyUBjHGLWGoLhvomuP2TauCeXxX7RxrbA7Y2U2BPH/y0unGPnAKKGtPZauxG3swvoqGJhgcOo4iIj119nea5mL7cIi/M/2u6ZpuV6AV1eNd1D8OlYR2oLxgFPj5SDl2fUEjU624Pn8HqZ8p/i3vW1YH8x464Gy8dxnVR6SPJfTHQCfqACppPxJzkNeIa3yUA9XfQ0cn1mAOUmArHT4AGBFfohwcHBwcHByGFy7ty56HOxPl4ODg4ODg4LAbGBGeqPxAXA6YkHaf1r8OGiFFcQLZlZzlVqYv1tV34oO/How8S6cVpgMdvvQmaDKhGGtbW0FjBCtIMUKUSjupWnQe3NDebgxhiiiFur+hnr/OQlvmFCBf3bPrEeyysw4Kj/oWUEaeclAjBQVpyrO3G9eGyyioaJ/dVV60lXK1UWBMnwfuaQ9TTJQLjwNpsgqPKbxLS6F6vG7hacY+5KOg1rq7oEqsXJiuZ8GFCH54Wc0zxv7WUtCGhXWgXXra0S7PaKJpxiCAX5IWR5Rypal1oAh1KanNiu0Uk78LY6kmQ9UVXIx1ED1saNDHRCgHPcg58gqoXQNo7+ilmI+OfXBDrIxycIVpXjeijTF7HEtZFx5jL29PKwsDlLeuux/ratRaUupNBb1bHsJY6Dash1ZSkun8HOqjiF3hlr8NfeUgkRzwlKEswQo1qd689PxUO9qV144+Rcfj5eLxor5UAeW8G0Bj4hV26uvl1onW8q4+rPftfvA30Siem8rDc1e3p+cpRorkSAzXRvopqG4+2lK9EGO3Jh9zXVGKtdmxAe83Teq/WMAecDSyn11xmIUcdJ43mh5j/7She0NEpCeENZZfin3bN5rWTxdJQEsxdrE+XOOlYwb8dyJKzLBvLLi9+A7MB++hwX2wK/wZBV17PVGhtDZrpiLQbEMflN7+PrTLS3aold618RyDN9zQTp03HHCeKAcHBwcHBweH3cCI8EQ5ODg4ODg4DB9c2pfhwYj4EjUQDsqqVXUiIhK0x9d7R/B/Dy7W+xuR6Oim9o+JiMgbJ/7ZlJ25jbPARwAAIABJREFU/lPGnn/EzcY+uwmBGNn1He8EdVH7HFzMO78Ml3H5X0A3eb8Hdcx9DVCs3dqNnHqrjkV7vrAZNNjdhyNfHQeGTCTTbm5fEdqVJGqR/YqpcgSU6yrANKsQ6IrtG6DII2ZBfIcgOB/nwuNAmqzCYwqv/tMYy0OWIfDdw3MQCPWSqs+KiMjWdtR91ZILjX3NmXcZ+/ZGRGu8ZNqLxv7DNjzTvxVzs6kbiqYCom/GnYy8W9sfgWpPNdu3gG9ar7G9FPAzPAtUR3BFZrFS7jwOsMco3ITn9I/HYJetpMCQJ4OmGTUaQf48CdzbUQxaQmh9etvtOfIWrJ5hLZ8+MT2XfTGM3aSKdrTlNKzxgTCuad8JddnYfbHGO4i+Cvrt8iN/adRaHqdgt8kYcSYeO+0RKhpaj28RAnB274fnl9P4th9GlBztZ0V/WMZOhmK1rRN9DQT/O0nVMXVQNEaTRPnT+YOXNmNuwtXp+eM/cUdNQA67aIpoviTm+s1ziI6ieWppBn19/JHI29jYj/KyIEnZCAWv51vLVa4YkZ+FGS9Kr23dUGS9tHQj0+2g58q3Y653Ho6xzjouwes9ipGKjMZ+Kn0T5bFl+GMSoKXUMRvrYHqhnXasX5k+rjB+f/xN4eCyTatB4ZWtwzPbZ2OQEjV4TngaSz1pjc+3Pt7hfYQR8SXKwcHBwcHBYXjhPFF7Hu/6TJRSyquUWq6UeiTz74lKqdeUUhuUUv9SSgXerg4HBwcHBwcHh/cb9oQn6goRWScig77OX4rI9Vrre5RSN4nIF0Xkxlw3i4iI0qIzSpzizeR35Zh5rM7LEYRz+4IaYx/86TXG/lj1JhEROWfjGabsxsn/MvZNHaDYPBQn0NcNt2uiFG7aHUdg2EI+uJ7bZxBttgCBCLktlVVwH19Q/wlj/6LmQWP/vRN5wAbq0CDfzrQb3xtBpxPFxMPR0GlyGeftIJd4kBRdVXA3qyTRBRHQBXnEEnEuPA6kySo8pvCWHTLP2F9uRP6+ttfTwUr9+yCQZulM0Cjfe+1MY39zFlR7928HhciLQJOKS5NCMlYCe1s36J4EMQ2RMXaaJrEeFw3U4Zog5dEbGD+U3wjstCvQuC2czy5ahoVdsgq/N8LllEOSFH8FO9HvgWrMvb/f/gvz4ClbreWTC3cOKYuTNK5+IVRn/lmgFgvKQXP2REAfFeRhneb77Uo2vTVkLeeWB2j/xcrtARv7KS/cIHysRq1GG6NlREe32BWPmmjDltUIpKuroR7r7xr6TBGRtg67LPKVmJ1eLQ2BDmb1YW/b0HMMC7uRP5EpzBQFIU2txTpNUh5Gznf49FpIkQ+YBCXtup3oKyNWaqdRk/RzWOVS54XTbYuNta+BflJFxitxjfahchXkQLNYk94oBa+kS1IBGg8f7D7at4EOCua5EW3Y/xDkL2UMKic7+rFmQ69gDcy9cLGxn+g5DM9po78ZYbS3kAKCeuxDM+xwEcuHB+/KE6WUGi8ip4rIbZl/KxE5VsD03ikin3w3z3BwcHBwcHBwGIl4t56o34nI/xORwZ9D5SLSpbUe/AnQJCLjbDcqpS4RkUtERLxl9ozfDg4ODg4ODnsG2nmi9jh2+0uUUuo0EWnVWi9TSs0dLLZcanX8aq1vEZFbRESCdeO1ZHKLxYop8F6uAJs5/Gd9+8Pl/Xoz6LRohp7695FgFX/VCnrp12OQw+3eUuQv05Try9ODdlUuBxUQPlhTObiILeeiXcu2oy3xGOphVeDNGQWhiMjPq6BCu7cNyfYGXejJfBrSQvINU763SZOgKqlPQUniK0YbFSmUcgWJ5JxT/fvgXs6Fx4E0WYXHFN6tE14y9sTy/UREJN5J1Fgz6IxPHwlXeVMMdMwVtU8b+8oVCMjJ+eq8YXabozyPFGMDYGkk2Gqn3+KVRAUwDVRKFFrX0IUYG2fPGxdswFhzMMpgB+ronorrE1U0T0Q3hYMUJJICpA6qonbFivoaa3lgaroNPTEEXKwrRH7IxP6gnZO0Zvt7cf34Kqg4t+/ED6HS6qFUoYhIosQu7/JQwFH9DtJSVE7oHFLW1wilaSqF+vI6MHb9NZRLLYHnsDqvbD+0vbsP63PsFIzNO0FtKdoYIXUeK+K2BrAvjRKRtuHsyaBiO6PYY5OK0MYn2w4wtqKAo0KvhZppeBcwnXNQFag9Bo8NIwdLm4XBd5MnYJ9rXxi0XSKMvRegVIaRMB0/IPVxopACBVMgy0QhrX2i/5jCU5STMTwObWscoOic3M5M4GIPBSTumQz7sfX7G7sAJxGk60BSFhZhwPoKsW85iKvD+x/vxhN1lIicrpT6uIjkSfpM1O9EpFQp5ct4o8aLyPb/UIeDg4ODg4ODw/sSu30mSmv9Pa31eK11nYicJyLPaK0/KyLPishZmcsuFJEHc1Th4ODg4ODgsJeQErXX/vuwYDjiRF0tIvcopX4uIstF5Pa3u8E7oKR0RdrN23Uw0SHsnma2KYcy5Joj7zf2j14/3die+rRb/nMFF5myrnbQRwtWIuhd9ctEkeTDBes/C4EFe7ZC1VJINFHbwXBV/+RIKNN+vvxUY6e2gSL4/Aq0p68PNMmDKxAotLgF7al6LZ23rH8crg30YIx2Hki5815F0MnSYqJLvJjynkPBa+VtQJ1Hn/SWsV99epaxHzgGVN2FK79g7MFceCIIpCkCFZ4IKDwRkc2fSAcTPW/zsaZs4x1QIsmRMOe9CuXLMxOQa7D6FSyCYBfc5ltOw5yNXoq5qT1ph7HrN8KF3z4jh2udAvt5iN7wVSF3XGpgqGJr9DP2iB4RiO2khNRBrXPQ9qIK1F1eANqHFULJdZincSdAFdlz23jrc6XHvsUXb0ir7xTljVuXQO41TzfaqHPksGtpBB1V+xFQQzueANUrx8D0ltipzpLx6Kuf2tOy3X5Wsv/FyiFlcVKUhZ6DErP9CPszs86GEBvU+xrqjpFys+11Us8dB3Mwn+WuWLudKHQ/6KONUQTHLHmLlFzHpRWFhXk4krB0OXJ9hsaBXh0MBCkiMv9TNxj7mqaPG3tdK/beMVXrjb28i44WpOxrv+7BoXSpiIj2kDpO05r4CcxUJqrluAfs+yC0Hf1IFND7tQvlZR+HvfNJ9DXUgme2HY9xL1qGPeGN2PdtAAJT8RGdtriYch8eAdOTmYYpZaBOl7dgDRS8gL0fRWpC8fai7hTniqQgxzqUK2qpw/sRe+RLlNZ6kYgsytj1InLYf7rewcHBwcHBYe9BuwTEw4IREbFcFyUleXz610/eCngJOH5LrthQjO8tOsvYH5mxydh9Y9KekrUN+KUdKMAv1FgnfsmUvQqPRc9B+DX3hTocjL6jAfGmyvPhPYjRL6UfP4e0MtyWyFj8+lrdCG8R/1qN9ZO3aF94KlK+9K8fjmXlL6Hs9fYfxTL2YXgsds7FL9FrD7/P2H++G/GddsbwK6ugBe0657UvG/trB+Aw/oIL4cnjVC4cB4oPkQ96oO6ZiBhQf/gmjZFG/w89AOUFPnS8ZSEOy3afgDg4XzzuWWO/eOuBxk7Qr+7CevzS3XaKPcVFsBG/pKuOxLG+9qcxZxNO2TLkvoKWHKdvFea9uAET1eanuE/r4XnZMoo8HxSUZ9oziIG0YRx+pe+zDmPN8FcNWMtjPek9wQ4FiWEtjXmRYmwdS+k2SjEH5S9g39RXw8MyutnuuSopsqcZ6WhCvwNlGJvgNrs3I14ytP5kCO6kRIgOFPtR7mlDfQd9ZCPKaXyXhXAQPy8Pnqh+L/rKmDN+o7V8Sz9cj4U+eJdaBhDXaVslvIefm/y6iIj4KQDSs0F4XkuDmMcmijV1wevwZv9wxmNoAGWD2RjFe6wyD2u/KmhfMz37lFjLfQN28QLDk4nl1LGv/SWdDMCrGh6N9RbswtycUbXU2HeWYr9Vv4J+R4oxppUIWSexUrw7Cs8BezDwD7z7S9/EGKyfbJ/X405eLiIiAQ/WwMoCzFfPkXSAvIE9Yei3v4eELetJ8JLAesOb2eH9ihHxJcrBwcHBwcFheOFCHOx5vOu0Lw4ODg4ODg4OH0aMCE9UYSAqHx2Xzlj+wougg1JEdWTFicpB7dU8gn9smYDTfrMqm0REZH1TnSmLVREN1ge750C4vrsmgxZ4oh1xQfqrcf12ouQKyvD8CQvsbZlSioOKnia4gWOj4B729OO5pVMRn8Y7Pu0Gbmsmdzv9ssjbBspooArlRVsq0PaxKG+KgXLor8ZSWLwcAYuKJqEtVfNAfb08frKxL6sBLXfVkgvRdkrlwnGgBg+RM4V3+SjExJn2HOrYbyzo1XVtmJuKj4AW2HEE9ZU4zbbD0b91azBm5Yegr8Fiexb3ADEdE4vbqR+gBTjmzyA69rVTUEliDeIh/EOHQd8UNKIfyZ2UHoPOr8dojZWsw9x0zLRTMDpl/+Xp7Unfqz0cS4c+jxINRjSfpng7SdqfBevRbz5Ez+jKkTqlYCvmI9qNa/LsUyPxMkuqHurnwFg6xEsxh3z96EcvxcfyUiygkhLMaQ/FiSqsBG3PWLZzgrW8j1Li+LxoTzSOeY2OBzW6uLMu3fYExQEjBc1aWvssOii5F/Tg7aVIX7V/KdKZPLgKtLaH6E0+QvB/M9H27SfbUyFlxTfKIe4ZXEPJg3qtnzfXYdz9xehHbxx1P7od79pYBdrSOR3z0deL+Q6NpXRJ9F6fRft2aTXe06FWmvse++H6KflpKvDmNzCmfFD8wKmNxl7VXYc+dTFth/oCPXzgPcfgDTtc2pfhgPNEOTg4ODg4ODjsBkaEJ8rBwcHBwcFheOHORO15jIgvUT0D+fL4G2kXLglrslKO5Moazui4GP7/gU64uRdsTfuqLzgVirK73jjU2IfsW2/s7S8iNkteOx7aF4d7fgyp3T76RShANl4zydiNP0dHuC0tjaArLzgF7Zm3HvGYDp7RZOzXn55u7JJN6faMIlZEeyl9BbmJB0ZTvKsSUAQ+YqBe7kR7R72JD6ZfstnYrX8FXRG5BpRGYy8UVd9aihQs15x5l7G/99qZxuZULoNxoFiFxxTe+jl3Gvtr2w439k9rEbf1271fMfb4Z+BQvbkO7veaF6DgO+JrcL9v/G2tsXsnDo05JCLSOxW++Oc3YE14wDTI4oaMkgtNlLx2u4Ip1Ib6EvlobzIPY9A9HddU1iJWjz+J6yNvYv10HQgl4MT77BtEN4bs5WPStKemulnJ1jUJ1IlKojzZR5RxJaXSmA61VPHr2CuMCVX2+ENbo5SqowgUV1Ts9B/H4hlE6Tpa74X2PxQ9+2B8GzpIBUxDV1uONiYpfUzAZ4/tUxWy01aHVtp1V91xjOvyRUjZ4suk4akpwDss34v5PboSKsDNYdDzLZdivMaFEAypoR/9e2Dun439QhiKvwPysCdEfmisunk50r705JD/fg3mYJyo0CtF1kur30SfYiVYm3kdmJvRP0Y/2ltAY+Z1Yh2mKMVVit6BFauwDp+fhX1buwLjNFBOqcVG43rGH1+fKyIip+232pS91IyYUqtWwC6qp+MgtMbotIREKuj4SJTGF69Lh/cpRsSXKAcHBwcHB4fhgxYXJ2o44M5EOTg4ODg4ODjsBkaGJ0ppk/W7fK1dGaJZGKLo2zSZO/NAMSVr4ab1FqddyE9uR1DGAyeAMmMVTO9YGhL6irmpFS70sTNAV9TlrTP2myVIbRJfBSoiWYO2eELo39PNSHUyvQoqNEZ8EgLMdQbSqhI/KVMGpqFu3w5ShtWCekuugRolBS+4RJL4R18taIagB9RFeAL60bK4GM+aDvlaYR1SsN/eeLSxvzkLqr2mGOiFwVQuHEiTVXhM4f153KvG/lsPgt2Fx5KysRATFY2gT70HQBEYTUGtFJ6EtiTr7BRFiBRjBR8FxRNZjnHy7TeUCogX2H/phWmrxYroGqKS/N1Y5J1rsN5YmVpGTJmK4oP+KvvvoUSxnYbyDyqzfKBI/AEKIBi1ByHk/RYnxkaTgi88xk4tbm2osJb720iJSAqwgD1jiyRLh74jwlVY+wOTcWPxSpT7WIlFTeHXSTf1+/CxW4y9vpuiVxJW1NdYy1uqMTjxJJ47vghUVV8txn5lw9C0PVPHUrBI2qs+2p9vLqkz9qY6UIE+oh8vC59v7CsmLTT23e0fMfaxlP2kaxK9JAjeGL0bc6nzMuW9E+3rLhlE3ZFyVFKwHfMUiGKP8fsqkcd7C/cmKF5ulIJtlpdB2RyuBLfGqm+vP0cKlozaszdB787HsWjO/tLLxn60GXmqeA8nijC/gW5S7eVY18MOvUtwXYc9AueJcnBwcHBwcHDYDYwMT5SDg4ODg4PDsCKVK2eaw25jRHyJUgru53Al3KFM4aXYk+yxB+GMVpJrloLvJTO56I7cF6qzl1vhv/7EuDXG3jAAyi9WgjqKKVt7fj2ospY4KC7/DtBasXKiQ6gtqQg6cnjlFmO/1lZn7I+PfcPYS7ahPaHt6XpSxNoFGvCPYCee0zsGg8dBETnoI+cMy2+DaqYzBv94qJFyA55OVBaqkZ52UH6XTHvR2PdvP9jYV9Q+bexnJqQVQpwLjwNpsgqPKbwLihGo9LY45XOLwVbUp9B2UKH9CfBg+ZRJXnbYg1RGqkhlF8EYa6IO8rxDlXjaHrtPfFGmH5iOprZzdXVQS3rpOb6lGGtPHPXk77TT4L4cwQQlw3ynSIHGz0kF7NQJIytXI61xppsZoVH2PH6JHcTZ0K3+vhwv/MTQcg6OGqdApbw2OL+ejtlffSWUu295G9bevmUttsulvMKuzptWikCzccrbGKQIjMGdGPu8ael9lufH57WFHdb7xgVBCT47HuukqgRtidO8XlgLSrwjAYr7stHIM8nI77ArTP39b587b/B97AvbSY4Q5VX0xDCPhdtIAUov9axjHPTe93L9tBwC3RSck/btKNp/KftWycJgUNloEuukaxbeVw+8iQCm+cTqxyvxHmW1a4z6mnNdO7wv4eg8BwcHBwcHB4fdwIjwRDk4ODg4ODgMH7S4YJvDgRHxJcrT55G8F9Nqlo5j4fLPyvvFjEKOheBthfs2fxPRghl/27/7D8O1FPDs1s1zjD3/R7839l0dUImt/SJotVnzoMh7+duos/au9cZe/zxyrOXXcwRRmA/24F6mZm7biACQPuq3Pi6tEguS8qaMJEz9MfR/Yh5oiTt/hYhu32483dg9F0Gldur9cO0/dPVxxp7zF6hQGuZhnAooT5+HAnv+Ydtp1GCUX7kCATmrX0l3qmUhKBLOhceBNFmFxxTe8zfeYuyrWw4y9uuXw/7DvxBk8Evf+Jaxr3vgRpTffLnY4BmN8Rt1NyiQns9BqZd3a2b8TsF9F339MWt9r3WDPj6+bK2xr/3XWcaOk5IusB60HQea/ebP/mns/7n/M8b+/h/uoKf9P2P5wva9Em9Kj7ciaqw/AMrzoi9BWXnHU3ONXbAZr4y5Zy0z9sJHDzH24aevsj4z+HSxtTyO2KdSsAV7pW9/eyBEFRxKK/VMw9jljwVdGx4gupbeJyXP2dWHDdVQ2/mIrnytiOR8ELWJmm9PFLgihOs5V1qSFGbRj+BdV3x3Jr8mrfHFVdVoSz/K46TuLCKaKvQiGqx9IBnuC2I/Jwuwb++PYsyeBgsvU67A+mRUBe3UJUN70+0JTLNfW3MU9nwTBewtDFLb/wdjmrgI3Nv5Vz9l7H9cj0137jefNPYrHQggHLsBdt2Vbxn79Irlxv7+Q1AuZrXzgfT4tYRRR/5ReL96B2D3UGBeSRC5Q3Re8WRQsD7LMQCH9y9GxJcoBwcHBwcHh+GES0A8HHBnohwcHBwcHBwcdgMjwhOVLNDSMzuTy6ufmsRclocVWPZ6/D2kTtuHVBIZ97/qhCubVX0qift+sPlTxj6iHGq+bcfC9aw6kU+ubzzcupu64cL3d9vbIgG4cj3cHu4TtYcVW+dOej1dNyl17muAAq6S8m6tb0ZwwN+XIQDmhHzQUU+fCIpy4wCu76OAoxv6UO4nUVsnUvpJcgxc8f6toIS0n5Vn6FOwKz0e3Sfg+TuOwOecC48DabLSiim8X1atMPaMK5HcbkV0rLF3zkSfOH+YyhFrz0Mu93AlBfOMYs56aoZun5DHTkElaMFFUrwO0SdfP6uSSE1Ea3VRNwae1/BL/ejTyfTc6MQc+c660utWeXiPwVzaCY7NSwqw/kLMb1MYeyJFc90SsedN66u1FmcFJdSKGpGw/8bz58eHlMUoSGeQFG791D/vQI78enaxpHhJTOgJ2V86fePs5awq4/cLpYuU/BDWSn91WvbpG0ADYsR+emlZcXDJoiaM3fZjQYPxvRVrMB7JIAV9TPAZCeAnY+2UdFsKcx8QOyU1eCyhqthO59WGoDgcT7n+RlFSz0UlOAqRNwrrd1sUxw/6JmBMV/bgfVweRD1baN/y/uM9lLLHFRX1jXSg06bFaAvvyYFJOEaheuhoAy1NTerA7i7Y6j1k81ywzT0P54lycHBwcHBwcNgNjAhPlIODg4ODg8Pwwqnz9jxGxJeogmBUDpu8RUREVj4FiicZgO8xK4hhjnWQousn1CEwY74v7WPd2IHgecl8+FS9A+T2/RmCPt51PJQ6n/8C1Er3//kYY/vOxnO6H4frNwWmQ2onIvBe0AvX+sZuag8FAvREqD356NM/N8wWkeyNoNfAb99aBm6qeAMGbMk/Zxu76RiUf+3Sx439t1tAAnlPR5/emAfX9wDlRJt8aAPaTq7yTd2g0DTRsRwcb8tpaff3F4+DIrCIIjfeXPdRY3MuPA6kySo8pvDWHA712iH/+1VccwHUOfN+e6Kx+w/KkeOrAeq4+IkIoupZiYnNP2VoAMZf/+tMa32sLl0TxRqPj8fz/Z0Yo3gZUVxEIyy5EfRt8mBc8+gNHzP2z27Fs4pK7AEue3vT48o0nIrj+Q13Q5WkjgWPWzkeFEz9Q5ONHZuCdb31MUrEhmZJ4X6gchidraD/PBVor39zge1yibflDyljyqxnPWif1CgKVNpDuQYPpnFhEXAHaJdYqT2oLyN4mL1P/aTe8tL8xePYf9Em9LvwY2maPUbcbaQfdYQn0QuQ2hIvxv74zGmLjD0+gHbdtAmTEEugHp/XvvZPW/oVa3m4k8adxmzrRVScUXvWr4eykLElgvdDahRxX0Td5n8JcxPpgoryiX8cYeyCY+gd9U/kLO2rxVjn0b5d+ST2XGEjbcZZdm7tif3vFRGRX1UfYMpuX4z30pQa5DXcsgTvcVZ9MwWbt5OCjFK3t1if7vB+woj4EuXg4ODg4OAwfNDaeaKGA+5MlIODg4ODg4PDbmBEeKLC3XmyakHa3RqdYlc3ZakKcnyb1h3oTvNKuJMHLycmTXSCXfWo/IIbHzL2+gjouWe+DVfuVX+YZ+y/fO2Txj79egpQ+ORcY29bQQoPajqljpOUz04dMI05uaJdRESiCfQzSG7tAKn2mqaAdrpu+t3Gvr0Nrv3HvwL7d3+7ydg//PaXjX3VdX8z9vfvRMDMpichtYqOwvgVtKMfsRJ2YaN89NJ0O1+8Ffmn2g6HsqjmBdBkvQcg0CXnwuNAmqzCYwpv2f8iqOYR377U2E/+8jfG/tjvvy02eGuQM7D8VtBKqSuacc31GTXmSbjv6vPmW+vbGAFNPLsAqs+rHv6csZnCy2vGxLOa59of3GzsSx7GPN35w9/S0640Vs92u1LO0A4s/SO69ILLFxj7hiUI1thZj/k49/MvGHveAuyPE89FrjZG6il7YErZD+tWvYn6Zd8+y8UinrKh7whNdF5lBRLpta5D4Fqm+4tfG0oJiogkKS9lFoWX4+em96FR1vLiLPUvbM6bGDsWdFP+/en96ougjUGiEwN9RAmSUpC2vLzwTQQHTuRh/VRuwDsiMZryX+6k8T0V5gXTFg/pj4jIxGCrtVzkamOlgul2jhrfbb3ywNHbjd2fwGAX+TCnDd+ZauyNF6Pfn72Igm3+/QRjn3XJImP3JED/LfshAsB+9P+wVo8pQjDRr/wbe4gx+3dXiIhIyWZMXlENxnQr5VVMlOKaFB3FiNKRkYF9aI0TvSt/sT5+2ODiRO15OE+Ug4ODg4ODg8NuYER4ohwcHBwcHByGFy5O1J7HiPgSpQNaBurS3FbZK3DxsudRe+zB8RjBU+Fu7uxFLrZYZ9rFW1nbbsrCUXoO1fHnnyGXWd94OOoqv7PN2Df8/Bxjn0EU3iO/gGqv7AI8q7sPLvRYL55bNgYubz7wFyUFTWIZ6IJNT6QVU/EitJgDjIZr4TIONWBqL7v3G+hTLa6v/l/06dvXQJFz82+RP/Cqyy4z9rirGo3d0gvaRa0DdTju5K3G3taNvGV5FACx9qQd6b6l0M91a3DtEV/Dc6Ip0Gf9CQT741x4HEiTVXhM4b3ya9CVc798hbFTYBSzkIihbY3Hw/Z2gh6LnzX0jXT9zWcNKRMRCe2Aa/+RCgQ/LaXcZx2HghYoOBxBUXk9fHXeJcY+/xQkPLvwZ6DwXid1nljyzImIpArT85EVbJPMW+75uLHVNNCoqhbBDO96/khjH/6xN4392MOgla6HmFDGnrnF2pZpflA51Xmg4h5efpDtcpHmoXnvStdjXXfU4PNSLEfp3gcd7JpBPBjBQypSznfWs6nUdrn0HB+2lhdQIE2mUEryoULteBY0f9ux6eu9AayBRBjrurYGlFxrC94JE6qwTgqDmKfuGMagOITAl5+sWGTspzqhamX8844TrOWcv49x3p/pH5nllou6XR0pM3akjAPw0jvte1Az+1aDjp33h+ONHSXF6Px/zEU9Hain83jYC395FOqZCeo5Nc5+fGT0SU0iIvLgvji6cfgfsccmHIp3Z8e/Qe1FKMVilNTJ/na8u/how4cZSqmTReT3IuIVkdu01tfu8vmVIvIlEUmISJuIXKy13pr5LCkiqzOXNmg35+c1AAAgAElEQVStT5f3CCPiS5SDg4ODg4PD8GKkqPOUUl4R+ZOInCAiTSKyRCn1kNaas18vF5HZWuuwUuqrInKdiJyb+WxAa53jV9behTsT5eDg4ODg4LA3cZiIbNRa12utYyJyj4icwRdorZ/VWg+6el8VkfEyAjEiPFGeqJLQpjTN1XsMKUboW3OWqiDHt+nedeRL5RxYmcs7liIPnOacScR4XPx9qJLW9cPd/tZPZhj7kl8/YOx7LjnF2J+9+VFjX7+A5C6WtoiIdDejvZxTi9U88Wr8Y+zktJs7HEPj8wOI3FZESquOUtCZP7v4X8a+adtcY4e/j/79zx1/N/bVF4MG+81f/mTsz//lm9Y26lJ0cPsjUO0lSBg2QCnc6jem6YjCesx1+SEYgI2/RR3hSaAu8rfj+usegPKOc+FxIE1W4TGFt+hW8F0zbvia2MAKmvEL0dneS9GR0lsz83Ax7jv3ooXW+rqToHRrgqB6f/081g/nTAy/hLXB+dx+fBHm8qfzQSs/9eNf0dNAO0jM/jtJdafHOyuPF+2rs896zth/Xw56ThOVdtqxS439xAIEdP3kGS9TpaBdm+fXWdvStT/R0E1YB5OObbJenywdSsV1z0FZXh72REc5lJXePoxFyVo8J+vVQmpYvQXUk13jKFK0KGT/QKOcY1pyatDYXChAyxamr8/KkUdVR56F2riclLy9edjDagv6nfKjr91bsPbuCIKm9XQTFbke5tzzl+zaGxERmZTfZi3PWm+ZpsXm9FivrKkELdkVQbvK8tCW7v9DLrzE2ejT3ONfN/bj87Emjzx7ubFr8xBk9MnvQX38kZ+jTzNDOC7wo0fPtrZz5yPpv9ef+NXXTVnqMHze/Cz+ng8cRHn0IlhAHGw4PgET6/F+aA4mVSilltK/b9Fa35Kxx4lII33WJCIf+Q91fVFEFtC/8zJ1J0TkWq31v/dEg3cHI+JLlIODg4ODg8PwQYva23TeTq317Byf2Rpi/XaplPqciMwWkTlUXKO13q6UmiQizyilVmutN7275u4eHJ3n4ODg4ODgsDfRJCIT6N/jRWT7rhcppY4XkR+IyOlaa+PO01pvz/y/XkQWicjBu967tzAyPFFaxJvxiMa5+B1QeIxgO74TxohiGsyTl9+Mz8PjKadWH+q+/a9wd6co8F70M1C+XHf/p/Cci9DiGx44zdiBMAXVo/xdnKssv5faM4Zz+VEwvVK4irdtTStefN2Ytm5yGQc76DtxCOVXrfyitU+RL8HFfNUjCPqoP417z/k3lH15TDMW45pEMT5QzWhbZAwolmAr3NztM9L2tlPgzg8Wg6rrnQhFTrKOeMAdUPB96ebL8UymSygXHgfSZBUeU3hrvsHSIlBPkz+zwtiXrK839l/mQJE2+WEEzRzEP+YdN6RMRGRgIuYxsAN0bOF+UGhG3kL/xj0HqkcTNXPTW1D/jaI5PuO6/2fslTfguf5Su/ooWZSu0+OltUl77L678aPPX4jn8Pp5+sFD0UbK8Xjv86BafkXjXr6W5pIQ7Eal/n7M37YgHYE4FqY/xG+JNCZVgSbqi0EJparQrj4/+LFuFtvxu4WUU95+omb8dgqmc6Zd/ahiFDyX9ij/1r5wP1BMd3SmFWPBnZRbbwLWTGg9+pSlwt2K/dY5g9pLqkxvDzqbyqMGpCiwKeHR5w+xlmf5DqiaK6dTcSagaVVJr9jQH8dcJ2ncG2hCOolh8+/EXnn0EayrKI3BS/fh7+fC0ej3L6+/y9jffgHU98uPEGv0afv+6KtL1xMrpfZSsNZ4hV3dye85puEjVehrMt+es3BvYAQRiUtEZKpSaqKIbBOR80TkM3yBUupgEblZRE7WWrdS+SgRCWuto0qpChE5StKHzt8TjIwvUQ4ODg4ODg4fCmitE0qpr4vIE5IOcfAXrfUbSqmfishSrfVDIvIrESkUkXuVUiIIZTBdRG5WSqUkzaZdu4uqb6/CfYlycHBwcHD4oGOEJSDWWj8mIo/tUvYjso8fclO6/GUROWB4W/fOofQICGE6e/ZsvXTp0re/0MHBwcHB4QMCpdSy/3D4eo8ib/I4XXPdpW9/4R7ChrN+tNf69l7CeaIcHBwcHBw+DHjvfSYfODh1noODg4ODg4PDbmBEeKLWdDbLtPk/FRER/xKEtdP0FU+/g697kf2goNNdUFV4BzJKpFoonpJNUOokC6GWCJCaifOn5a2FkozVMXk51GgqROqNTrTFQ6odz3gEmEs2o/5UCAoTVt8kCtI/I+Il+FwlSAVUBjWPHsB93iKomZJR6tMWtCtWijo5txOrCaWS8oH1QjXj70KdvmlQ5STWU565ShqPTLC5YCOeH6DYfL1T7eqjSBW1cTSUXqwwSzZQcMUazDfPJQfSZBXeU6l7jZ3aMdXYHz/hXGNf++idxv7GV9MKwecfhTKu7u/XiA3BzQhSGalG/zjoY88+WIdfn/OUscMkiVvfj4CxOyNQV3GewtWn/9TY+3/3emt7BtMWejhoKu2xyExamzTXioN3FqEfmoIMclDL+iuvMvYPV3/S2pbX2uuMPTof6+eFNxBEdevFVxvb1qeC7VgDiSDWb6KAclIidma22FfZ1YeJkH1P1H8LfZp83W+HtCVdD+r08B6l59Y9iv3UcHJafRcvp3dLI8Y9RvkLfVuxlli9qrah3EOiM1bSeiPUFprK+qsQMHPi7xGkNgvsxaB+bL5i6HiMelPsYHEgBTxOBlDhQBXKo/zeCGA+Ai24OTURY5NqxRjscyvyCm4+G7n8Dj1pjbGXPYQgyut+AXXupOvTY5AV2DWOAQu00fskRWusBgOvaIy8LVhYvoH37lzSSDoT9UGB80Q5ODg4ODg4OOwGRoQnysHBwcHBwWF4MQJ0ZB84jIgvUVqLxGPppiSIssmi8Jjay+GRLHsOrtyeKSiPl6Y5i+KXQX8MVGM1BchtXvgyhiSJ6kROQzC/vIfhGu4/HvRD6fOgrzxEH/VMQjUcmDK0BNRTtJzc/11EYe1PrupMnZ4efM65z0Jr0GCub+zDsLsmoa9TzkHCrIbbQV+1zQEtWPYK3NCpjaAce6YSpTgZtJl3cbGxB+rgCg+0oM2D9EbVkQhQO7EY+eSe34DJK/goXPISQVtG3Y25DFdiccRPRPDK8lsxvo3HYz44Fx4H0mQwhffYU8hXd+os5Mj82+JBWgl0XvkiBEVkMEVR8RoFHp0LOiZYj/m759cnGTtJFNPoVzEenTMRoLCimQJQng6zb/LQwJQiIipDTWivPRDkhHl4aOsstDdWhnmvehJrqWsq5iAPU5mFeffNsZYXNeDBLYXY3MVe29X2PoVnYq2lwrQ/gphrTYnrSseBP+Ygo14P+tfVgTVWNAprnJGoilnLOQ9iisc4gXFK/oAGqj1Dx/Zh3AsPwzunfTNySBYfiPviTyDHYt8RoGBDBVhXPT3Yt0Jt4RyDDH+v/QXLRxGUPcaoJDMUaPj0PuvnKaK+EnGaYGVvV3I91nhBIwUZnoFrKp5G/2JFqJ/3LQfYbfvKWGMPXGIPfOnNHLVIttMfARq72FgaO6K4VQ/2BAc/TlHAVbYd3v8YEV+iHBwcHBwcHIYPWtyZqOGAOxPl4ODg4ODg4LAbGBmeqLhHVHPabcquZM49JLlsQt5ZO4zdsQH51wbzGfUfBmpMmkG7sOrLNxtu/kgErtm8RXCbJ08CZeRZDlVU+Gi4sCtKYHeuR1sCnXBh9x1EypqdcONrylUWWg1Xdbwoo84bZac88+e2oe4GuP97L0V7e7tBcW24D+qn/hNBBQjlMus4HG7rquouPHcDxiO4mKi1WRjjINF/rP7zVaWpkfanya3ePMbYnv3RlMhy1KGJlej5HGitaBTz5FlJecKuaDa2txNUa++lGHfOhXf2NtTPKjym8B59/Qljn/yJr4iIyJNIgSa9p9hpjEQjxj1yABQ8BdS/8IEYu8NOQxaDnTHce9g3thi7JQ7qdEUH5Zkj5BNVzVAWFiNrv30VYxetx/r19WL99p6FvRLrx5qJVNmfWXlUs7V88sdBWxX5MDePvjnDdrnkNwytP6+daPggq0tJ4Ub5NAfaSKpHHee8lXoCKMLoBrr+EzBZYcrQzFQl7GrXxIJqY3sOTN+gaJ9En8Ee0/uBNow8R+XHYG/71mA9xPuhPqaUgVl5/HQffUDCSc71yciifnM4NAaV0HmPFls/53HheeL6ErTPk+OwUPuPwL4J1NNFZ4Le7GnC/j91FijxyHfQpxPuWmzshn+ebG2nrM/suXF4/wWaad3RULByM1qB9kZqMWeql48z2B857NDyjnLQOvx3cJ4oBwcHBwcHB4fdgPsS5eDg4ODg4OCwGxgRdJ4nmJT8KWm3dGwt6LEUtY5dyVmqPfJOhh+EBMozGddHM8ENi5fCBTwwGp+ziiJvKdzBAQ4G94kOY/sWgCrrmwv6JvQi0Vop2J6JcCXHyuHuLVpJaroyohTCFHhuBqnzMmqWXK7h/hdBu3gr8MySP8C1npoC+mH/z4Eyqr9hX2O3fwrUXvkDcPn3j0GgR5mMfkQPwxgEV5ACcjyu8XeRUmUg7SqfcMoWU1YWxDMXN9QY27cfXPh5FFQz71bMQU8NxiP/lBZje68H7RE/C+NbeismdvLDm8WGwUCaIqzCA4UnIvL4w//MWP9rykoeRv8ZsRLMad5SUMk75pCCsQHr4fkVhxib90Hns+Nw75GgKKuWEB37AsyBcXbuwDMwqM5DGSuuPDdg7PxH46J4GeqrnIfnRw9C/8reogdRqq5GogUZ3etB5UZIVVraSBd9BmakaigXOTCe1K0DtNZC9sC0FROxn/mwbYpommgv1v6EfaAkZcRqo9ZynSTFlofeXdSGnefj3kFVoIcUqP5JpAhswR4efVKTsdseA42b4n1I6rxID+hgj5feP8W0ZgjBnXZZpJe7muNIRSqY/qDjaLvyj8Fj4SEV5egKOlLxBtZMybN4f7cfCarM8y+opcuJXb1x8X3GPuM6KGifuPgo1H+enbrMOyB9dCHRhz0ZK6c/PIrXG8ZLE13r30FBamnJepLvZbDN9+zRH1i8K0+UUqpUKTVfKfWmUmqdUuoIpVSZUuoppdSGzP9HvX1NDg4ODg4ODg7vLyj9Lr6aKqXuFJEXtNa3KaUCIhISke+LSIfW+lql1HdFZJTW+ur/VM/s2bP10qVLd7sdDg4ODg4O7zcopZZprWfvjWcFJ43T435+2d54lIiIbP7sD/Za395L7LYnSilVLCIfE5HbRUS01jGtdZeInCEig9KmOyVL9+Hg4ODg4ODg8MHAuzkTNUlE2kTkr0qpA0VkmYhcISJVWutmERGtdbNSarTtZqXUJSJyiYhITU2N7RIHBwcHBweHPQLlgm0OA97NlyifiMwSkcu11q8ppX4vIt99pzdrrW8RkVtERIITJuiJv8tkDq+kLNhZB8hBO+ZaBno7DgFqPpSeuUGVUd0tONyrxuIQ5iG1Dca+rfYxYx/+R2Q5L5qDw8v9T+Ewe+xwpIBJbsEBY87onrWGSymOSDtORKoqtLP6AZRHv5COjXTS+HWm7K6Vhxp7Zh0CHa1tRruumfVvY+8XQCyt83/zbWN/7Wu45k83wXk45/MIgvTEAnhmOXN5IoT+JYtwgjJAB1Rj49DX0c+k+1TQgsOnHfuin3ntOOwZL6C4PXTe9aKvY25ClLL+1/8609hXnzff2NfffJaxz71oobH/Me84Y7/5U2Rxr/v7NcbmVC4cB2rwEPmSO7E2UjuQPodxRw9+SxwbQqqZY+ZjDvLa+OAqzCRlkplzynJjL3ryIGN//vRnjf2jGQ8be+INv7G2JxVIj7GizPQ6iHH/8pHPGfvWl5CuJdiKjTXreKzDFQumG3vmKW8ae94RNxv7wMtxQJ8xQD+z/NhC0jcZh9i3fuU7xv7WcqTkGcRj9QguduQEiAUWvYxYU/5e9DXUzEIVDHYMuhLx0bnrGPQu8taPsU6m/uK3Q9oikp0yig8V88HjvInorOel9AO8eBVJ/ziKF7cDbYxU0OFljqtHZ6Q5VVB+qz2+k4fOfi/9K9bwb9edIG8HL50sv2L608ae/Kv0eFQd2DLkHhGRwgDFeqJD7v1xNDj4Vayxdd9FfK6b5yB22/d+9SVjX3nlPGM/3bmfsdfcjLm/7GocMn+lZ7Kxn1x6gLG3Xoo1NtiP0HaKAUWhwgYP0Itkz2+8kANI0RqjvSUJ90Xmg4R3c7C8SUSatNavZf49X9JfqlqUUmNERDL/b313TXRwcHBwcHB419B78b8PCXb7S5TWeoeINCql9skUHScia0XkIRG5MFN2oYg8+K5a6ODg4ODg4OAwAvFu40RdLiL/zCjz6kXkIkl/MZunlPqiiDSIyNnv8hkODg4ODg4O7wbaJSAeDryrL1Fa6xUiYpMwHmcpc3BwcHBwcHD4wGBERCx3cHBwcHBwGGZ8iM4q7S2MiC9RnrhIqDl9PKt21o63uTo31rbVGjtLDZFZOP6tkMwkSEWhW1D+Rj4yq/+1dB9jx4px/b6jcFZ+mYIKrra809gbtyHVQirPnlog0ATZVZaqoxXl4Qq4Xzvb0mk2nvagXcF6tH19AVIkqE14/vJ9MS7rImOtbTk8H4qxW3vQlpNKVxt7QQBOR+1DnzgtSeEm/CNWQql1GtCnyGCWBoW0CKxmCrVBlRWmJeqLor7XuicaO8F5gGgYN0YwN6EdaG93ktL/TIRqkBHcjAYNoBpJNGJcOZXLIFiFx/hCMdbMwgFIvc4/9iVj//ueo40d6CTFYz6es3AD0vMUkXLouTa7KjCVPzRFiogYlZb20Nqk1EqbwrSWYpTChKp7ow17hVOC7OhHihJGcYM9BY32UNoeUmb219iph3U91UPKfn3wvcZ+vhdjxON776MfNXbKT3WTWdhIqivKAePvs7clGbL/VfIOkKqURZdRlI8pQXqTpoLMmqDqvBFKFUTjkgihQt5jnhhL72CyqpXnSec4Efty52RreUWw31rOGKzT77Wvu44BpNIJ+SEP7B7AfiurwR4TSpGyqBcK0PAYlL/Ug7Vfl99u7CVVuIb338+XnYr6ffb5G1Q6BrvweSJESkhqV2Q8+uHrxFpWRJ3FSZXpMtZ+sDAivkQ5ODg4ODg4DDfcmag9Dfed2MHBwcHBwcFhNzAyPFFKJJVhdtY1DnXVD15jzBxfpj3EFqS8Q931HPCRo14GClC+TwXcvkeGNhj7li5c39SPiHwc1G5bN2gazhjPNvcjNoai3VEwUV8+uYdXwrWdX5T2xU8qgcv6lXHI73xgZZuxV0dAlR1f9Iaxe1Jwmz9KFMGbMXBW8SI0kqmRLCqAqANdgLb3jyd6iFzlmmilko3pthU3ILJgPERUaz7RFdSWBNFax5etNXYkhb6uiaK9swsQdPGRClBlNUGMX4AyrTMi1VhMFa+BD4kcgEHIWxqUXcGBNBlM4R1HY3HJg8gon08ef6bwEkR1zpmMNfnSxpnGPrUatGsWvDnoikywTU20hMePudu/cLuxn8kHjcI02NHj0NenimcZ+9CKrdZnxorsv9kSxN7EicLSfjsNnrIojP7nDQSI/WTdKmP/beHHjM3MTf5O1M20Fq83H1FyvCcY/m57n5J5TAtS2+l90REGrRzaoTPPpGtprLWXqSRc4+un9lJwUA42zEE1E2DTsp7FOLx0s7V8rB/HFTzKPjeDr7Gg107dVpUiwGiBF+/dtiCCEyc3UUfouMSpJSuMvXAT9s05ZYuN/XAXAtDmt2AOFg5gD/Meenb9NGs7B8csXmif98hY9M/fjsFm6jSRT2MU4EVgrXLvwJ2J2uNwnigHBwcHBwcHh92A+xLl4ODg4ODg4LAbGBF0nvbAzX3IxAbrNezCz7bxPXBld63YMKikGLUYvvRIOeoIdqH8jTK4d88pm4I27g/qqb0fPnF2zxcF4Z7uL7CrUzjnXOkyyotHeZkC3aCJOveD/9W7Nq3OW7oelFU5ed43bIaqZlQb7ruk6SvGThQSjXEM2nhX8+HGZrVda7QIbawiFz0rjgbwj7KVsKNlsIMdsFvnpH3lbUxXhNGWZB4nPoTJlOe1/0IuPE05quLjUc9VD3/O2KWk7Pv186cYu3C/brGhZC3a0D4Xc1+wHBTMjjlDKQvOhcdglRhTeJvOu8nY0/72VWP7epnKQT0rb0eurwLq921/+7ixr0TaP/F22ulKQ/EwW0Fj/ae2E40d7LVTGi/cdQiuofXw6P1HGPs3YFckXGn/zcbsXKyY6Klur+VqkQ3bhiogHzn6T8Y+9bmvG7uexnfKXZcae8dxdrqpaC29IypQzlQZIzbNzompZsrjScosVu319hNPe2xa+ZaIETXUBzsZhB0vwR5OFhA3xPQnvWciRDcFd6I8nOPkxM3/Psn+QQ4q6Lwf0CUZ+vitt8ZZr+V8mlk/4anu2PcxNyqM6y++6zJjl30Gxy4uux3zqmZjP4fngnp/thc59XgP+SDyzUKsND2WgS7K/1mKRpYtR3n4eOTTjDeAmw524JpEhHJU5qDY9wocnbfH4TxRDg4ODg4ODg67gRHhiXJwcHBwcHAYRmjJdvs67BGMiC9RKiES6ErbS9aAklJxu8ItSxhC7skrT1lg7BseAb1RlKG8Og8E1RNqQNf7xxMdNBpSlrJXQYUkCuB6766By99Prtm2rVDK/fjYfxv7Fw9+2tgF29CP7v3Qnrxmcg+PpUCL1XBJJzLufV8Pru2cA6opFaE6auFk9JAruWgDrkkFYK8Kw6+tp8KdvmgFlFlnHrrU2A8/9RFjj16KCdl2MsavZBXGqZtiQRZVpKmLgfVQORY00rhMJ+ULUTo87/FijJ2vH/3zd8KOl+GGjkOJXuVAeW9BNcfo2QfXc0DT8IGgbwINebIr8trszl0OpMkqPKbw1l9wo7EnPniJsYs2YK1GTkGAxnAPUUZJ+8tR5Yi1OeiD5ncqsaVSM6PZ2FvXg/fx9aB/PQeQ2pX3atw+Bj0z4tbysROgluyLgMr2xe2vp+JX84eUnbUWNGox2BWZ1ozx3ffwLcZ+YxPRTdTeXlJfeklRFdk5VIkpIuLfNLQtIrtQNqQ4ZN9/rAPz5w2nP/ASDZcKoI6sgI5dFNAxDxMcaMB+48CbWfVUUADRHDStJ57jD+07oIKSo9J7N29rwP55kCrJ9fc8aX/XM629YxvetT4KhKzrcfwg2IvB/vc67D85hSj89fbAsIGu9L19tWgA7494Ab1ntpK8lPoUrcNaqqjEvi3JwzvbroN0eD9hRHyJcnBwcHBwcBheaHcmao/DnYlycHBwcHBwcNgNjAhPlC5MSeSItA++cBmCrmWlRGPXbw438J/XIrBekgKd9WWordBo5H6KROD2TRXCJe7tZJoPdSdCqM8XRgNiRBn5iWa7dhUULhx4L0yp6wKVkPxEklD8acq1V7wMLv9Bd/LAaMpJt5rywI3BfR7K0cWUDlOXySJ8ULCFlECHYJx8r2Gcnq5Gzr54Ce7t2Af3jhrdZexwebmxE1WgfsoL0v3eMooC7O0EdVpZi6B+nWtIIlWH8QqshwudqROm8JgiLTgcdYZfQp3jnqN8YP8P5tfnPGXse36NuTzsNAT5fH4F1GkGOdZmrlx4rMJjCm/zGbcYuy8F9//sl79s7E/PfN3YDz4BdWXWc7vtDUpkAqR6mDqhX6mNK7BQC6diThWxYIlGUCGltbgm+grmnfGFw16yli/rqjH2geUI8rlg1Qzr9bYxLtiOxgf6sQZixVgDm54BZa0qaFPwT8ko1mFqFNZsrsCfsQo7X5pF5zGDRXTd5HmgrTedm95DgTLMtW7EO6F4NDjK3jjGfVQ5gld2pFCuKLhkaAzuTVIQ3qTX3icOUsnw2NnYLHg70v0oW2sfF39fjgCqQYxL48eJniMqUnFg0XaaJ3q/Fm4hRTDnvSPmPZygXJB2kaYkMq/VwlpQf707KR9qAHRlgAKuDtQSxR1FefcKvHP6Bt7Dc0nOE7XH4TxRDg4ODg4ODg67gRHhiXJwcHBwcHAYZjh13h7HiPgS5fFoCeWl3aBJ8gIznccu/FzrINpMiaHomkQmIJ1KkDItSG5lClrIisB4IbmVSVXiJapMU0Iups3i20mxQfUnQpQjj1z77Lbm670xakMmfxaPC6WfynJNs6qFaZoU0ZISZJc70XlE8TCL0dMG+i0rByBRaJ4E6uG+KupTx2CwUmpYnIbLn6TAdNxXr10pk2Lal2gUHoMozT3fq/12Z2yYoqhyfsSdMXLpW3ZP0i7iypkLjxVHrMJjCq/QgxtiYTSmg9qSi5bQOXzN5rnaTjsFetDeJM1HwIc1k7VXkvbAmIyexFA1o4hIc69dIaUi9jrjhUPLmGqKejiwIcoDEEhJvITUaznyzCUjvGbsL52cgRO5nBVmb/M3TCl7fXFav5quicZBaykfBdKlQLZeD+0butfvt1NunLMvu3H0LsqRwNSTUVEn7eI88Qbs96V8VJ4jtxzPJedJTZLcldc7t4H3H+8hX45lO/heT6XsZ0oSJMrMpYBllaqvj96p/barHd5rKKW8IlIl9MdQa22P/k0YEV+iHBwcHBwcHIYXOb6jf+ihlLpcRH4sIi2Cr/FaRGbmvCkD9yXKwcHBwcHB4cOMK0RkH611+9teuQtGxJco3eeV2Kvp5HEDtUQX5Ai6lqUwIBdrYQ389bGVCMbmzwhY+kvR3bxWCmBYDd/wZZ9AwM6n2xBosv8ayJIqf7jR2A03IopkwUXbjN3WB6olvoICw1EOrr5CuJWDrfArR0mQ1kvpAH2T0x0JEK1VdiiUN5ODqHx7H4JIfnfq48a+qXGOsZM/qjT2/r9fbOyn7z3M2DM+vc7Yr26EuqloBagZX5ioumJQqgWcpytIqqB16XunPQNlUayMAmC+ifEqI3rMtxRj+s2f/dPYi7oxT0tuPNjY14ZYmNwAACAASURBVP7gZmN/dR6Ubz++6F/Gvukt5OBjrO9HfrbRr0LZd9g3thi789mh+cHmnLLcWt/CDch3OGfyBmNzHi8OpMkqPKYf6k+83dinbzjZ2Hd97vf0tCuNxVQHY5D+Y7UY47LzHzb2b18/3tjRHZiQi05cZOy/vI58gDNP3mSt84FXD7WWT5veZOzNHUgiWTVpp/X68ISh/ImX1Gi9B+BzzuXYOZ1UXFvtPGe0nMcLdfp77deXL81FORKdFmXaGuVbTkOdFYsz5QpcZYIpqA5SQlKsyP4ulI9/Df32JJimRZ3BSrz3iproLMAnYZZ+AgpJRlkeeChPDpfGoBJ558HWj8U7ATR1LIx3Ql4h2jL+brxDmk7Acw6bg3fR8keQC++Uk5cZe8W+kFTnfxvvqA1Xg3+rP+avxt7ndgRjZRRvTM9NrBXvUToskk0h0l9RVkVKEf6uJA5CkF7NqkjetsMNLU6dlxuNImJPpPo2GBFfohwcHBwcHBwc3iPUi8gipdSjImJCzWutf/t2N7ovUQ4ODg4ODh94KKfOy42GzH+BzH/vGCPiS5T2isQz+Y/yd7y9yif7ZphFechVtL0aMptkMN1NTUHWImPxuacfz/zjY6cYm1UXSQoAt+Ut0Fqek/HMlg3IMVZdB2p1B9GF3l7qH9GVUW5PD6Ylv5VURF3pwJfRMrSltQvu5h0Hgx6TNQiSedX6z6GcA/99CvaWhaBazj3vBWM/eBdyTk08kWiX3jHGLtpIfSJV0kA1qfYol9e4E9KChw3jQIeVrEMdXQdiLBQFrOOcXv9z/2fQJVbkHIxnXvIwKLHzT3nR2D+df46xR4Xs/u2dEVAgnTOR46+FAh3uOLJIdsWiJw+y1le0HW1/aSPOKhaQapFz4XEgTVbhMYX3ENG0kx643NhbiKGIl9ilToO5KDUrQSnH22+eRe7J/GoK1liDtXzPRgQb/fzBrxr7ricQ9FYoZdnkfe000ewyCGBOrsIa/vOqObbLpWDL0HdEXgfNI+WH5P4xJTdQbc/hFqJ56jkEe9u3w64s5JyQDKZsmDJN5hPVuIpy9mVeKYNBUEVEijYTHT4JdJdnPd7xlbNajN1YgSCn/E5LUeDh0jGgpjc32fNGti0aay1v5WHnbXPU0PKKFdYqJLoZ+4rpsZQPNPH2oymwMc3ZK6+CElekCH5oDfZT3nrMU/FMXFP0Euo5fTz2UKLQvv97JqfvPfao1abs1fkHGjs8Du+owrcwH6kA5dTzYsCSbXSNo9RGHLTWPxERUUoVpf+p+97mFgMXbNPBwcHBwcHhQwul1Ayl1HIRWSMibyilliml9n8n944IT5SDg4ODg4PDMMN5wXLhFhG5Umv9rIiIUmquiNwqIke+3Y0j4kuUSor4M8H9wpNJMcLKoRwqIka4HrK2QCcFyst40BNEDQXa8XmsDC72S49/1tjr+6uMveEaqEGqv7PZ2Nt+P8XYY74BVdKy1ZPwrA5qC7ErqRDa4+8kN3Ap5f2rQ9tGT0lThAEKvFdZAMVMLIXyjoNBu/x+xnxj37YDVEvndycY+8SbQOH9+ydQY/3g53cb+8fzzzN2YQcphyhWopdyWvkp11W8CH3quS2toNlnHdRoHTNBLUy8Dzu9vwrjkr8Tffr+H+4w9kv904z96A3o350/xJnAC38GxdpTP/6Vsc+4jhLmEbZ1oz0VzXDdr+iA+qdqSVh2xedPf3ZImYjIc23gfU6tBkVw299Am2midzkXHgfSZBUeU3j1n4ISUeTbuDdu3zeD9CqLrHhtXjLnaWP/eclcY/taQEt84kRQePc8gnE/++OcIw/j3vzkBLHhn/tgn/l3YP3MPHqD7XLpmx4bUhapwH3eiaAEk8tAuSYK0cGS9RToloZoAE2R0DpQQylUn4WSjfbyeAHR0NTcZB7t0aOhVKtYmKazvBTsMwzxrJS+jnH3JDFpfU/gCEHNRiwUphMD3SiPFWOzjm6BYkyIAp7zKVDJjNo8HFHwZMmlv2WsweL2j0fEhsnVbcbujYHCG5WHtoT/DzT/ZjDvctYsqPAev/sIY598LBSxoYMw2C9eiT1UdFGrsX9c85Cxz3vlCms7CxvS87f6dahnE3jVZ+VYDVPO0qwgzmR7RoEa9uTIWejwnqJg8AuUiIjWepFSquA/3TCIEfElysHBwcHBwWGY4TxRuVCvlPqhiPw98+/Picjm/3C9gTsT5eDg4ODg4PBhxsUiUiki94vIAxn7ondyo9L6vf9qOnv2bL106dL3uhkODg4ODg57DUqpZVrr2XvjWcHaCXrM1Xb6cjiw9bLv7LW+vZdwdJ6Dg4ODg4PDhw5Kqd9prb+plHpYLGSn1vr0t6vDfYlycHBwcHD4oEOLC7Y5FINnoH69uxWMiC9Rq9t2SO1NacUUq+pyIdc6YPXN/Z+4wdj9Oi2t+fpqBGj8ZN0qYz+4FQqMgWUIWBelgG5SAtmMjqCNngEcK0uFKO8fXXP/6VBUJSmy31fXftbYp45/w9iPNM4w9s4mBHpUeen6NakMvX00XtUUHHAjckWlApS7q4QiiOZR/2J0PC5I/eghWZIP9XxrLgI9rgsj8OaC1Wj7wVO2GntFfQ3qyQQT9VdRPilSX+pGhOFLFKMtvh701RemQKUToQQqKkGdPdspGGbQ3ld/KcZs47n/Y+z9v3u9sfsmY+7zGzEeA+PSqqetX/mOKZt4w2/EhhQFWeSApN5O1MfBXQPdNB4cTDSf55ICOpIKb/M3r8Jzd9ijQc7L5FYc50PwxZfDuPbGZ6HQ/NVJUGhGNNq7qAvBD/O9UEUlNObppkP+bux9fooxZUTGYHx93XglqQlQP/Lc1N4CdaUBrdmiUbgvvKF06LUi4h0q8EtX007BbWnpe0lstvp6qNFq77zWWg/vGx2geUqgfl8l1mqyOb3mPVHOuUe2Xewm0Qpa17T/Fckukx1BsYLUoFsux5qpvf066+WKxpjV0lsu+K6xJ/02vf5TFSQzJOg4FnOojOapE+8roWs4UKmvB+WJkD1I7GDuPhERnTc0x6KISKANa+xn59xl7POm4EjJQY+k11tXJwRanlaMo7cGqmhdj2uy1g/FvWaFpv4v40k7DB+01oOSz4O01lmZDJVSV4jIc29XhztY7uDg4ODg8CGA0nvvv/cZLrSUfeGd3DgiPFEODg4ODg4ODnsTSqnzReQzIjJRKfUQfVQkIu32u7LhvkQ5ODg4ODh8GPD+8xANN14WkWYRqRARPovRKyKrrHfsAvclysHBwcHBweFDB631VhHZKiJHvN21ueDORDk4ODg4ODh8aKGUOlwptUQp1aeUiimlkkqpnre/c6R4opLKKK9YocTKDFbkZYnz6ASbLsDNj/XONHZ7PK2eOLUWCrhjitYaOzQJqpZb1p2I+ljRNUDKsG7YiSrILgLbIMeIVSFfFbelLwmFx+kTkEPt+EK0rWQiVCs3LUNuNVHp5yZIyMKqnVg/plNTrjpN6jxW2Klemn4Pyj10r4dUcAlSUb3WPdHY69pHG3v6xO3Gnly409iBqZibxRvS98Z6SO1Cyjs9Bp3y+2lBkNAq3gQFn3Rh3Ht7KYcaqZtShZgPRQqwZJH9dwSlIRTFaiFqDiszzX2BHHmxeP3SNR4WMVF1iQJa19yWBNuodDAX3q4YVOHtinMKu0VEJJxChTMCa4z95/xjjL1qADnvwkmM9VElyG3XEsdzagOY9yzkohKo37yeCwoHhl4rIuKzjHGU1g+9LHh8U0FSdIVzzLvfbksuZXjMXo+m9cZ7i+34AOWZzKxVRe1lJaamrcprM1lKz4nyB2iwvxyDymOjPPa16t9p/7Og1Tv4c5GpnhXEWXVzzrkkVG15zSiPjWI1I+5llbG/j/JyFuN6ntekFzYrJHkdDu6DXdG1NfOyIVU27/1YP/ZBkPYe79V4Eb1TeX++D09dDweUUieLyO9FxCsit2mtr93l86CI/E1EDpH0+aRztdZbMp99T0S+KCJJEfmG1vqJd9mcP4rIeSJyr4jMFpELRGTKf7wjg5HxJcrBwcHBwcFhWDFSvr8ppbwi8icROUFEmkRkiVLqIa31WrrsiyLSqbWeopQ6T0R+KSLnKqX2k/QXnv1FZKyIPK2Umqa1tse0eIfQWm9USnkz9fxVKfXyO7lvRHyJUoIfevyNXeX4UZ8dJ8ruoirz9Rk7lflgR9T+q3xnHPGE+Jcr/9LP8krlIEH51yL/4qzwI6u834OfVtwezoremcAvNM2eo2T2/0V22RTUfw+1PUXt1T66iOIV8a8zjzfHWqQ+lfrhJQj4cH1fzB6TpieWh2ZmsphzxiHtIU9CkhpMXocUdYTnRlG7Un6qlK+na3hd5cqo7qEh0F77m8cW74W9VlnX0q9+nbSvWV7XHr5Gc9s5fhSV53g5chwoxqAHKuTBL+rWJGLfcGNCHntQpd5kvvWa/pR9DXjsoYOyvCb83EQyR0CdpMUtxJ5UHowcHiSd482nuKs8Tf/t65nWJ79lOR6aorU92G1FyyfrmbxmecnyGiBbU3CxLO8Te+5T9rWa873LHlH7JXiP5Bp3Lqd9xbHQNO3hrH0eV9brs65hhxPvWx53ssMpTHghN3RwPnKspayx5rXH3mFa7/TaF+1xAS9F5DAR2ai1rhcRUUrdIyJniAh/iTpDRP43Y88XkT8qpVSm/B6tdVRENiulNmbqe+VdtCeslAqIyAql1HWSPmxe8Db3iIg7E+Xg4ODg4PDhgFZ77z+RCqXUUvrvEmrJOBFppH83ZcrEdo3WOiEi3SJS/g7v/W/xeUnTil8XkX4RmSAin34nN44IT5SDg4ODg4PDBwo7/0MCYps7bld/eq5r3sm9/xUyKj0RkQER+cl/c++I+BKVK6VPVlkuDyiX0zCu6R9v7I5Y+hByXQixs8JEOUzL22HsFLveiUpjCiHrcC+n8KD0A3G6dVUfDub208HcmvyOt20Pu/Hh82f6SqzXJooodQu5r7MO5Uaps9QPps2ymD1O9dBfZuzuftA6kyowxnHy/9cVoq/rEpk0MTH7gW3lRxv9AQy2l6i3/gBRRuzaZzqNXfG8xWhhpXLkEGK6gO/N8u5baI8s2pdB4+vx2w+5qhx2rtcDr7dcFAyncmEMHiJnCm+0l7zX1N76gQpj98ZBy44qxb2tsWJjH1ZYb29MrtccU1W0txLJHHSTJZ2HjmGtMb08wEuAWcMcPnjanpKkg8xa5XgBpXKU8xwzTUtCCZ3VoMH66Pl0sNxLlfABa0kwbZejXSRE4YWVJdog5EqrlWuNZV+U+X8OClxyrGued8lBnWbRiUSV5ZrLrGf57O2J53jYIEWYRb/yts0x7zx2nPZFvwOK+UOGJkl7ewYxXkS257imSSnlE5ESEel4h/e+IyilVst/+AKmtZ6Z67NBjIgvUQ4ODg4ODg7DCC0jKdjmEhGZqpSaKCLbJH1Q/DO7XPOQpNOxvCIiZ4nIM1prnYksfpdS6reSPlg+VUQW72Y7TtvN+wzclygHBwcHBweHvQatdUIp9XUReULSZ5H+orV+Qyn1UxFZqrV+SERuF5G/Zw6Od0j6i5Zkrpsn6UPoCRG5bHeVeUTjiVKqVkSmaq2fVkrlyzv8fjQivkT5IiKj1qXtyCehJkok4L9NJnO4rQlnTUPcpfmLDzV2flO6m01HN5uye54/0tilE/HMj3xsHZ5PfuJNt+1jbHUW4uDEH63E9Sd0oS21qIfbEmxFn948tNXYf9t8lLGLx5Kar5+UJNPT8V4KihH3ZSAMWmufMahvy07Qbb846EFj39sGirr1J5OMfc7vFhj7T3ecYewTz33V2Jv7y429ctlkY49aizZuOw1++fqFiCWV2B9qSU93moMc8yLRFVH4yrsm5VM56KNUAM+56EvPGHtpZ62xG+5Gny64HH265R7E2zr7LCTmvu/uOcZOb9E0IjMRq2vCPPLLfxVryHNDhua6Ah9/+Uh70u9NYayT/Qvhef5TG+KS1cxA3Y0rxho70IN+X3b+w8b+zbPo0yVznqanXWmsG5893toeEweK+QeiYDaffJuxJz72JWOHNmEsJp2LfTBvIfZT+8dAC55Jz/T3iR1EtSSIwvLkoPOCoaEyvyj1I0lUC1NiajTiwflasMay4v+U0vNJDZZL/ejvsisIk3l2qlr78MotqkOMovj69DrnOFnJEO2PAWoL1ZfMR/moVWiLl1SGKT/6ytSXD0s8a6JKZtpThhXnRazljMFYXHnFUevnugjlQTpaMBDCuqr9B8ao6VjYpx63xNiL7jjM2Kd98UVjP99CoX1uxJ6b+l28j2cdZf5uykGPX27srRfj1sLN6YFKNeH9w2ujaA3aS2Jq6ZtC8eiIdq4dRzHzSPq7UfYyRo4nSrTWj4nIY7uU/YjsiIicnePeX4jIL/ZUW5RSXxaRS0SkTEQmS5oivElEjnu7e506z8HBwcHBweHDjMtE5CgR6RER0VpvEJHR//GODEaEJ8rBwcHBwcFheDFSgm2OQES11jGVEZBkDrK/o9EaEV+iEsUp6TgxHbwx+NooU86CGG8upR7Z83tAVV15NKLAd2T8rQt3gJKbfQhSVjT2Ip/IK4v3NXYqRAq3g0ip04g2+uqIitiBcG3zW+1tCZNk4/Ed+xn7iAPRnq29qL+VXMiDipD+XriYhdKcNARxX6wVaVG+8+T5uL4YVIj+JByRv3z5FFwzDVzA/a+RQpWC4B16KNq7eSpoPqYX/bNAbyZjlJImU8+2YzktAtMfuWRAuP6Op+Ya2zsevIQ6FpzRDUvgiVXTEBz078sPRxsL7fskSePaOgs0SbQeFIH/6KFUzq0vzRlSJiKiqH/P5E83drAXC3jr+mpjF079/+y9eZhdV3Ulvs+b36tXc5VKQ5VUkixZsjxbCBswg2xjcAh20wRCmiFkIEmTCQdCEtK/dNKhm0CATjq/BEhDgIQhNqOJIbbxgB3b2Ei2ZdmSNUslqUo1T2+eTv/xXt21nutcV7lcJZXtvb5Pn7dPnXvvmevWXnftTWNHtNZnHgM9F1+Jvv79z17v2R/FspJPXf8NZ3tmUrlwkExW4TGFd+wGUHupCiid/zkMSuXDN4Bm3D3d63xm6kp3GpfeFVBunm6i9DHt7kCh+f7ZMfA4SO94EUpBHt9KjqiZZrdaiu9Tjs0OdPts+CnWOOUQK+4sqdBSR9HXcE1xx4quEFH5HETWsFKYqMvxKziiI8xAxidoqc+viNFjrc7yEVaesjptJz2rphgtHa0LXUnthZ3h9FKkNO27nh5D59UP7qezaD2u/fqDr6ILYEYux7459ciFnn1fOxSr732FOz5j9sqq8rRUxNiZUUrr1YmxTpzAWcEpwYSo3qEDCGMUWNDXO4olxk+MMX8iInFjzHUi8l9F5AdzXCMiSucpFAqFQvHygD2L/15c+CMRGRaRvSLyG1L9VutP53PhC/JEGWM+JCK/JtUh2ysi7xeRVSLyTal+oPWYiLzHWuvOG6FQKBQKhUJxjlDL4/cVa+27ReQfn+/1C36JMsasEZHfFZELrLXZmuTwF0XkBhH5rLX2m8aYz0k1ieA/PNe9gsGKNNUytpczoKE4uFpdQDUfOo+zjzNt1liTvIymcO8VCSjgOFhkZAwPKpFLvtQJt3JwFO7bcoM7gzgHQuS2RCmh0mga7WmPIXDhZAbt4czzM8Ep67LUUxuzk6Ar+PlR6lOWqIDoKjzTHoT7vbQOlE24H30tUdA+DlLZnsB9RkeQh7ChHWNcR0HOtLEF79acT66cogiiPoHpGo5hrtNJUIid3aDBxolSMOtA+dkBVvy578/0W6ENcxyaxqIstpXk2YgOubcU00GVsLtToSmiNCmJAQePzJ9BX8tr8fzQoLsjORt2lmfKs+tzIE1W4TGFlwygzmPjiHe3diUUXQcm3N9jxuLuv6UKlCMvSArB8VzcVb1OqTYDpokqPvQVU2xFopLq8rDxPNFU+sXa9A30WNc4pymhDAfQnH0/VurVTWOZc8JR/QacLZw/sGg46iPdJ+B2F4TS7k5V8m4asQ61c4HPH0adUjHoVjCWu6jjFEw0OkLnWA/WfmwAE8UULO9bVkUXi9hDmQvc+yZeW6u5AAa+GIJt4mhwJULlHGOV+hSZgh0onUM3zYvPQ7TksNaWjTGdxpjIQhw+L/SbqJBUOcSiiCSkmrRvpyBo1lekmkDwOV+iFAqFQqFQKM4RjovIg7VAnp5XwFr7mbkuXPBLlLX2tDHmr0WkT6r5Zu4Ukd0iMlFLFijyHIkBa8kIPyAiEu5sclVRKBQKhUKxCDBW1XnPgf7av4CINM5Rtw4vhM5rFZEbRWS9iEyIyK0i8mZHVee0WWu/ICJfEBGJr+qx+YerCq98N6lQ6vLG0bN92hTZDJ/pj4egshucrtI6b1iL0GZ58tW/fh3Kh7pAAaXI7TvytbWeHXgbAqdl76NgmztAXwUCaDy3ZYyouteucecYW9ED1dW9T1zu2bmWqrs+HKdgg80YjfNXDnv2sREo5t5x3W7PfmSs17MnPoc+nff7T3v2Q/+xzbM7dgx69pkxvOw+0Qcqxw5jnFZvQcDPqRzKu7ugtBo8WVWhtT8AaqhMFFe2k2kXz6yjN17/dvTpVAbqyqO3IQjoO9/zgGd/nYKrvmXnLs/+8fcRCLUOjaALuu6Eu3767VhjnbfUGgchm1x+LYL6MZ4ehvLuapr3B75+hWdPXQRPcukkxtpQ0Mf3v/E+z/7mYVz7829EUFTGfRNbnOWvbq6qK6fLWI+cC48DabIKjym8f99yu2f//gCUU1/f+s/0pE96VnlPs7hQ3IF+50lVesmlp531K92zgz4GTlNOvw1Q+5XugeIw3Y092XTIfYqUEkThoyn1gSkJkUm/+/A5Rvckuqnci3601PYCsf2Sb3Xk1pP6X4TlGOo0PogGG6L8mD6mFIcSBfNdh+j5k87ypvjcwTZtjY7N98wOiCoiEowRBU00dSEDWq3zLszl8A7M2ZqdJz174E6swwt+/oBnHxgBlZy4Best8w6cP/9lw+Oe3R3BWmGUairxCqk44zTXWc6pR6x+kD5zyHegfxMdlis5n6k4N6h9E5W01n5kIde/EDrvWhE5Zq0drjXkOyLyKhFpMcaEat6oBScGVCgUCoVCsYjwyy79Mkbtm6jL567pxgt5ieoTkSuNMQmp0nnXiMguEblXqskCvynV5IHf972DQqFQKBQKxbnFE7XvoW6V+m+ivjPXhS/km6hHjDHfkmoYg5KIPC5Veu52EfmmMeYva2VfnOtelZBIvr3qtm04RYoNv5iL3A6OaXc+3KSsdssWq3TMk2PIR3ZxGxxkjwwi99r0Y6DB+N6FV5DU4gQFBL2QXNwUBDDRA2qvMwZ6jlVtT42v8uwr2vs8+8FB5H9j133y6So9VmwETUaxEuUg3Tt4DDTN1w4jACT3qXw1Bnj0HlB4r9oJau+JWxGkru0a0IWs/hkahyt+jBSQDTE0rn8ElNu6V1ZpmqMr4XpvOIh7ZLYiv5YllRwH+Lv7dlBZHIiwQLmrbvnRazz7ytc+49l3/AjUk427Xes2BzXPxCZS6qXRzvyls/+qe+JHW2eViYgEKZXYXU34oyfKQiii7VrWgWspknrtS48hx+J7LgOF981/e61nf/pS3DIedItNBotVqoODbQ4R18O58DiQJqvwmML736tAkV6/H0kI7+rGM+027AnG61aBTt/fANqT6XRG5OBs1V6I4nhOP4U9HGPlHY3vNNI61n0fEMMSl2w3B1R0H5XFRvf6qVP5EX3D+zl6AP1I1T5jYBVeHEy6pNdgr8aH0Km2LZiPkU5w3zaD9gYbQa3FE1iIkw7FrIhI4P4WZ/kUKwS5228ku9a0lt1+slfKOUfTGKf5G3oD5TgcQv3T94HCy6/EeDz2CIJnxoYxNhMoltCDOLO/NIE99BtX3O9sZq6rev9NF4NCPHE/fk9YCjxqKrzIYLJaO5zCD86pM0iZRD+0icio1IWOFSsiS/cSJSJirf0zEfmzZxUfFZEdjuoKhUKhUCgUywrW2vcv9FqNWK5QKBQKheJlC2NMtzHmu8aYIWPMoDHm28aY7rmvXCa580xZJDJRfZ9Lr5lbneeH0iiopMEkXNszwfw2tzIdhZvvXHPQsx+NwmUbDcKdf+a7KC/vBNVifwrXd/Fy0HbpMfiqBxpBk+RKGPJNLcQdEJje+E4S6r/85ip1aIgeYIVLRxI+8anz0fY3rgeVtX8CdMnEV+Ae3/SbUJXt/gEovE03HvHsJ/f2enaglQJlUuC5aBjPjYdBI7SshNrrzB3V564YQD9yYGCk6TEKhrcKdcKUB+3Ktz7p2YM5zPWJH4KneeM7QXf98AfIl3fTjQ959q33o5wRTOHvixgYE8l1gdNoOyCzcPGbn5ldKCJn0lgDr+g44dm3f+cqzzZFPHNGrTrr/m/CfHz9DlB4v3DDg1TrZs8qWXfgyXWR6nykKxjrHUmoBkdfC2qac+FxIE1W4TGFd8fWf6Mnfdaz4j9xK4e/NQVqloN87rhxr7O+vXA2LZiaADV10fmgYA7fBWq8EsFaajrq/vuxSGn5OKCrX+68UMpHnddABxabNB3hK6AYi95VpZtCWVROg+2XZB/aWyaWc2oXzocmbLFnSZjRj1IC51JTmqq8m2rv5BsBLfNQ580o1UrXuqV/hvjMphhou+EJqKI77sQkjNBnFKsuxXk5eh8GZ+U1pzybqe/pH6BO81vx+cbGKM7JmX3wbERHq+N98sc49/1+WzKly4FgOfBnOUnBXc+hOk9DHPjin0Tk6yLyC7X/f3et7Lq5LlRPlEKhUCgUipczOq21/2StLdX+fVlEOue6SERfohQKhUKheHlAExD7YcQY825jTLD2791S/dB8Thhrz31vt2/fbnft2jV3RYVCoVAoXiIwxuy21m6fu+YLR6y7x/Z88Oa5MDtGSQAAIABJREFUKy4SDv/JzWetby8Uxpi1IvJ3InKVVF8BHxKR37XW9j3nhbJMvolSKBQKhUKxhNC0L8+F/yEi77PWjouIGGPaROSvReRX5rpQ6TyFQqFQKBQvZ1w88wIlImKtHRORy+Zz4bLwRB1Ln5Bf+umvi4jI/q8gWGElRAHKSNXiF6xseiNJaPj1sPb2bRvw8/AZqKxYRcFB0SITsFPrcW1DHxqT2kBqtFMYztxKagu3lxSHde0ZovZE3cH5KjPlfD9W/oTougIqtT+OwciuQHn8aihTgt9o8+zpt0H91HAbVGXDV1KfYmRTYMpwCwXKPAG1ZKkZ9YPNVWVfcyMSkk1MQJHTQ3n2TvQh91miFaqa6I/RrhQJaJIXIBdW5S4o3Fa/7bhnD3yr17Pb90FxdM/df+zZ/23vTZ59y7cRrLTz1QOeffJo9bvDE7+BlEuX/A7UaIymPqyTQiPmI9MJe+pCqBl/eQfUdlMlKM+++1Pk+tu4BYojziW27399yLPP/wt3e2bWDedq47UUhtBUUldi3GNxqDI5Fx4H0mQV3p7/g7ZUzlD0Q8KnxpDv8JUJqA9/+QH8EXj8PZibCz88u09NJ7C+ApQ3bnId9mQB8RalshUdDATcatd4BIOTyUM1uO+m/+7Z6z7/qVltEZH6/GicZ60Me/W9qDL4tuq+aW2GZG5kGGt8fTeUaX1D2KvlcZLqNVASN0IgQnuVDk9Wyh35xY959taPudeMX/7AvZ/FHG/8ZDXpfccenyC2dI6z4jFL+6CyE/s/RTkkmw7h4slLsQ6T+zE34RSeyzk4VzyOsTn5JpRHh3DPA3+Gfmz5zp+LiEhPK1SGh05BmRoYwrhHpiinXjfWTCCDe8coQGpobpHj0kE9UX4IGGNan+WJmtf70bJ4iVIoFAqFQqE4R/i0iDxUy8JiReQdIvLx+VyoL1EKhUKhULwcoJ4oJ6y1XzXG7JJq2hcjIm+z1u6bz7X6EqVQKBQKheJljdpL07xenBj6EqVQKBQKxcsAqs5bfKg6T6FQKBQKhWIBWBaeqFQmKg8+uVlERBLtJD2jt+YA566ybnvtlkHPPjlIUpzRqpIilICio9TMcj+Y+RWQz5kyhic6Soq8TVBghMZRJ7sB9+9dCzXNiQHKgzYFFV4oDsVIsRXvs6aEMQhkuLzWXJ/cS62rJz17ej8UPBMQPEqgQPnq9lO7Lkd5IymUxi9AlTXroeY7PYDxjZ/GGBSbMQYsIgxk0Y/m7qrMZ+wU8g42nMA9TuRJ7TeM8SqRorJIirxSI+ZsfIjys12A8d0chmpwYhvKo5NQ9jAeGe317MY+jM3GGzAGkwdXybORXTGrSEREbIDyl1F+Nlaaru5BgNzdE2s9e2AaCqXNW5EnbHsb4sB97fwu53Nzq4rOcu/PJ1KLsXJUSOnZuwKKxwLlJivuwFxzvkfOhcdgFR7jI21Q5B0rQjX3nksfcdbn3HEzyLbTPqF+lCAQrVPhFlM07zQEJoX+mbVTnl04gDlghCbduQl57bMKjdVpqdWUZ62mshvOYp3w2j8Zw14JHEX+O7MRkrnKaZQH87h3BbeRcgMGJzLibnvFvSWkOJ/fFrVpCPjkGsw30NlGYkJu4/QINkgwh3nl/Jp8ZmdWoU+hDO7ffJhUmr1o/OatyF15ILLa2c7sSHXhVFpxpgbPYOGxqpX3cJh+H1TCeH6hGXYeR7PiJQD1RCkUCoVCoVAsAMvCE6VQKBQKhWKJod9ELTqWxUuUCVmJtFYjkIX2J+eu77MQxjJwZ1fSRJvVAk8Wp+CODU+6nXDsbg9SULRcJ1zGgTQqleNE/1H5WBo8gs1gmIM5+H5LU/CbhyaIRuCAnMw61igWpvNMxR15lANvJk6jryUMkZR6EUQxvhc/WHMFXNgnB+B7zmxlXsAdCLVcwP9EwPaIJYIjHKx2MNKGAc5PwoUfacSFxSxH54PZcBzl1qB/gQ70yTyDtbQyBmomQUFRw2k377AijuCRg0k8uJEi5eXaZy/E8PSsIhERiY9iUou0BgpNFNA1h/V5STsCaTKOjWE+3tSFh3HwWEZo0r3FvbVt3fRKKY6+nW5CUM0grb38ENb4/oaVnp044uaDOJAmgym89WHM2Q9OXOjZ/+Mi1HdRReG0df680Ez7jWPFNoHe5aCTWUFg084kAl8ebSWamFCJ+hxG/ClCifcKB/al+rVAmbEGrP3CNOauswm03VAz0XllClJJVF0lTOcbBRPlgJzFgvvsCPgxwLSfxX2p129Tdo9LkJ7JdTgIaSiBNlpav0zVFWgcuTw2ArtIv0p4/52exHqODrj3RyBZHYQuOgeOM/NN2y2EZSJ14U7pbA5l/QZM8WKH0nkKhUKhUCgUC8Cy8EQpFAqFQqFYQmgC4iXB8niJygfEHqnSOan1rI6jOpx/ymchRCtwrAUb4ZOecbHGWkHFFBtIRcHUVBH3KMdBGUXHYOfWgQqInCbqYhP8uhWiSbgtZVJpxdtAPbFasFwipdEgKUJqihsbYDkRTM4/FyJ1TvpyPMdOoL2Jp0AL5C4BXbD/MUjf1rwZueKGJsk/TvPBKhSmDgrtLPcCBvurSqMojV2MqI28ELVHFALnNUxtwxwIjVf4GEnftuCmP3j8Us/esBMKt9PRbmcbH3h6s2c3EaN4+zOgmFpOzr4utdGdvyy9ltZYmNREpO4KkfzpR0/iOYZyE3ZtgDrw759ETr+Lrz7kfK7pcSc8a0hW10SJ1HYlooYCZK9rRy6z8RzWzCWXnvbsVBHrdMeNe53P5Fx4DFbhMYX32PZ/pVp/6Vn5ttkHQL7FvScsHRZMcVeOJJ31IzR9x7JQXyaPuZVsoZSbpqnLEcf5LemsGSfVbKi/uhfKhih+Ov+GD1AOyUHMTTpB9DmdUYEi7U+iviypEjk3KCPf5t63ln9b+LGYtTEe3+yj/CNlJauQ6/KF9mGN8Zyl12FyzCT4tEIXyiPbiFu7D4rGwStxn64Yzo7CZVDEMoJ9VVr3wWmcA0EaLqa+83TO1X1+ESEldAOfkc5HKl6kWB4vUQqFQqFQKJYW6oladOg7sUKhUCgUCsUCYKw996+m27dvt7t27TrXzVAoFAqF4qzBGLPbWrv9bDwrtrrH9v76zWfjUSIicuAvbj5rfTuXUE+UQqFQKBQKxQKg30QpFAqFQvEShxFV5y0FlsVL1N7hM7LuC58SERFDgdYMKb3Y9kPgOFQdxRVQxJmawqOhGSq19CkEz1uxEQqNwdPICfeu7VANfeORKz07OE2BHkmZ5qfUKVD+MkNKmbr29KM9Hb3IVTb5OFQ5oa3VgJHRMMZoapqCerKohpqSTEKV+Fub7/fsT+663rO3b0A+qZ89hRxnyS4o3PL7EaSubnxJbZZohPIlPQSlXGcPFF7p/+is3oPySRXb0Cce33ILR4AkhWYUzwzH0ZbiMNZAoI0UfAMIosj3DCdw7eF3/qlnb/ujz3p2aiPqxPugCsp1VeVTx3/7w17Zhx5/p7iwfwrBKFm5eeg0ku01/ZQiodL8cdDATA8kWxxwNLUVMsYTv/xRz57ZV7MQqo0fqcVMDPeO0rjk+ymXGQUNrHRjXUUOkqLqQgQoPPj2/8+zL/wwxpTBufA4UCar8A7/IWiIyplNs+7x7uOv9+x3dD7q2R/6KeYjegBtzHbTuqLfLLyuLAd6pbE58b4/8uzer35iVltE6tekJSVrIIj7X7X2uGfff+i8al1Sz649H7lA+w4hN2LjGgSOLe7CeVUhNRgHg2SVMwe15FyCRz6C8T3/23/h6JE/DvxnzPF5n/yMiIisucIdLHYyi32YiGCMJihQcnEf8hQW1pA8l5TTJuIzTxFOnIi+bt0MJekz+3pwbZTm9Vdo33y+um+ClLs0QIFCeaxDaQoCyopk+pVlSC3JAZcVL34si5cohUKhUCgUSwz1RC069JsohUKhUCgUigVAPVEKhUKhULzUoRHLlwTqiVIoFAqFQqFYAJaFJypQMJI4UW3K1huOOuvwx7h+2HMGH0Sb0OxX7sqjSANgOvEB4NgTnZ4dpsd878jFqB/Hh6jNayY8e3pvO669AB985o7hQ3ETcqdRqPzM3Z6JJ/ExeZnSBRRPV78wxufoIg2n8FFlvhV1o6PoyPSFeFf+xyOvxsXT+Pr0iuY+z36673zPvuGV+zz7W8fxcT2Pr6W0K6H78PF5aA3qpE5ijIst1fJygsaFPr5t2Q8700VpKjC8MrUZH4QW6MNS/nCWU2y0HIQ9+TrM5YYupFFhNPSjbZmLUT82ii2T7Z69xn54dJvzfn992a2e/adP3+TZ/3b1/+/Zb9+HD9Qb+nHvAL6/lSD1NTZGaSU66EtiRrTsLs/X7kOCDVugtUT7jT+oNTRlgdP4SDhEizI1gXJG0wl3W7LtlLYnjfbUpXIh8EfkM/iX3vs8+0+HLvJsHt8bT/6BZ5s8/f1If543PY1xLFNGJ1NxH5V+YfZK/ST44LRI/GF3D32oXNtDAfoY+sRREh0cJhFBAXvMtGJC6tIvEWLDlFqIsqIUy+7xzY2756/uY26ffldqj0oXIs6fT05hXDKUYyefRv0gffAuBfqYnOasuQvihdQgPq7vugT7eWAE4/TBnns8+0++h/RDU1vdHZn5ELzhFJ7JAg/J0MfkW7D4g6cwdpwaphTHc/iD/rMO9UQtOtQTpVAoFAqFQrEALAtPlEKhUCgUiiWGeqIWHcviJcoaZAjffXidV14XG4qTtBt3eYhcrOWVoA5CtXToWXLdxhsQQ6ghhlgkw6dAsV25BhTXA09s8eyxIuKYSBueUz4OCi9MsUPKq+AGD1Jq9ixu6duekeNtnh3pzIiISKkE93y6mdzdNF6pLtQ5r2fIsy9tPeXZ338a9NxYCbGAMj3wQz84uAHPn8Szco0U9ySOPk1egGsbV8LlXqng2sRPqm72EmWgz67GPYpJ3Du7EWNRHAHVEl+N+FV1cbMOwrXf2QH+b2wt3OyxGPixVIGCFBFKUaJaMtgmZSoPZGc7cl/Vc8x5v/unMdk39T7p2T/3k9/27CZ0SSJprJl8AM+ZvogosUMYv+B6jDWjsTXjLLc1ui5AeylCa7NM9Oo4rfdKBs9s3YB4ZtNPgda+6PyTzmcGyj7UCVOEzPj5MPgcB2oGTOH95Yq9nr3hLoxvlGjJYJ4PFJj5ViomOqac8Gn7tPsIrTRwcCYyiV7dlhzw7IcHqzRwjGj41BY0gNnESozisp3EDyyFS2Jqj/vBlL8furrHneUrGrBASxU3iTEzl5d1nnL+PNuGPdwdw2cRT0x0e3bfgV7PznRi/wcbKa5UP9ZksAd02nQO+5n3Le+/6U2YA46D50Kugz6RGMPc5HfgOnOiQVwotvjEjCr5LGzFixJK5ykUCoVCoVAsAMvCE6VQKBQKhWJpoSEOFh/L4yXKwBX9ik3HnVVYnVdvw5m2ZwpUoFD6hEpNhdKyj+iodtA7RXiVJQb2TB7MXoAmtsKt3NIKiUt2N2iMxh3Dnj18gngBagvTJM37SQ3VhvYUJnFpkBVgo1V5SIBmrYXYo2wH7h0fxnUnhuAqP5Jc7dm2HX7+A1OUVuIQHrDpMvTp9Cr0VXgORtG/9j0Y43wbxoCVZKNXVceS08XYjHspNu0h1U4B98hkobxJE41ZIbXS0H4oAluQ1UbG2uF+N13uU6XUwClm4P6vhEFHVBKzVZf3PXSh837v2vmgZ3/17td69tFf/Jxnbx74Lc8uNFFqIaJp2mh8LfW7vBtUsrwdZuYQ6GmGp/gjZiHLIjJSE0Wn3fRD6R6oSGPk0z58FyhguRrm5Dr3HJcg2JJCM6krfU58TuUyA1bhMYV39LovefZ5X/9Nz65LJ0Ro3If1xu1itSvDtBWc5QFKM8TqvOAIxuCLkatwn97qmZJeTWOUgl2ijECciiSzlvoRdqtdw6QojY6gvEhLhjH+WKe73LrL5XUwbU21e+fui5xVIyNYzI8wD8KCvA2UUorOzgCtw5VX4BOFqQdwdpW307W0b8OU+4b30NQUzhG5cXZ7ed6zdFY03UOfP1wLaq/Yh/IoqSJLRAfboL7JvJSwPF6iFAqFQqFQLC30/W3Rod9EKRQKhUKhUCwAy8ITZSoigZpX/GcH1uMH5JKue4P2eZtOroJbNXcQbtpwzQ08fhFc3+FJuFpzK4hWoiCSrXupDtFt001QgATJNTt6gKi9DeDksgdAqbCCcPICUhBOE9VIgTeDHVDtFUaqz41MoO7Yq0m9cgau78xGuLUDRAs0P4M+laOw942CgjGk4HngAdBTkbWgMcvktmZF0egOPDc6iOem186mvgLDaG8oTS728zFPoSlqr09wzmDWUH3chzOtT55PlEoKdVJh4mwIeaJ1bRr9KLSw6mk2xROedv9dcuvtr0Ebaf0yxbTlyuOefeQe7AMOMjpOClN+VinpDujqh0p09ibieLZmBdZdJYe1z6q2dDeeyVnqedwZhVZncV3wwRKJ2qwjYK6ISPRAfFYZB9JkFR6P7+FfAnW64Tu/4dm8ZlIXo9/BIazPbI+b/uOAo3XgsaQxY2rWssKtRv+FaByLTRiMUpIo66h7DYZHQTUTeyVFujaH+J1S8Rlfn7ii80JgTVUNWk65g20W22hgGjCmAVLeGgqwaSMcTBSDN/QMqEU+rwIHodqzNH68/0q0h/y+EQrVzojsShp3Gq/0GlLqDWM9Bmh+c6twFnb1QPEYDKBPbi3vEsGKeqKWAOqJUigUCoVCoVgAloUnSqFQKBQKxdJC1XmLj2XxEmWSZYm8shq4z/6szV1pHpOfyVJQwFa4iks1Ni2QJl8reZWDOTjkIuNMIeKh2bVwzYYmMWwBctVXuvDM1ElyK3NbSH3kCtYoUp+rjINzBldWXeX5MNzHicacZ2fItc8BMLvuRHsz5M5vuu6MZ5s7Vnr29KW4Z2If6IpMFM9tOA90Zb6b8raNg+ostBHdQ22bCfR46SsP45kFPKdvjHgfCMDEFoiW/AmpKyk4Z/oyBN5regTtnbiQch/uw30m3eK1OmqrZQ34tOww1mfH+jF5NhIDPhQJJWWMj2BczlyDdj19ZI1nmw4KPkprJnkCa4aphuaDs6lFEZGgWzzmKbwsnQAkdJXQIMau0Ez5GxvR9qZDlJ+RWPimo+51XdnqDmxYJOon1kQ04pGkq7pku2dTa5xXjSlHVuExhXf0bZ933vv/jEPhuy4CZert45c465c6is5yKdIY+ARatJS78neuv1NERKbLWNeDBZwhAbrJUB7jsv825LnM0mcJTPmFUu75MD48REO/ey0FCnPTYKXBKj1ufCrEB0ix1kBzRpR869U4lwafxoHVfAj3GSPxX+sB2MkBzPfkBxCA1u7HvrUBokzdS0yKtSDKTFNbOlPDA6S4jqBPFaKmI6TEnOqjg1fxksKyeIlSKBQKhUKxxFBP1KJDv4lSKBQKhUKhWACWiSfKSqCmWDDPT2RUB+MW0Hiviuxhns8LuQ2RW5u98+zt9kmDVOcG5vRsgXk8mQNZlh258eiZnJNO3ExkHSrEvIWDcE/XKbN82mjKPm7+INF2dCMOBsnlUqvOedtYscJtN/NJMzWP+WDUzZ/1m0BS5dQNDgXNc1zLVIFfG63fny5MAZHpq5biey5yOi7jk8POt+3zeH7Ab+3PIy9mfX3HfeZxHavw/FAk+VyQDqNSJeiqPj/49aMy+wf8zDJNasBvPfo95wXM03zWkm8TZtaN3z6oq+uuU9c/nup5tcsdlHle50hd42bO2rnH2vfW1Pa6nJDn0Buk30QtPtQTpVAoFAqFYtnAGNNmjLnLGHOo9t9ZAVKMMZcaYx42xjxtjHnSGPNO+tmXjTHHjDFP1P5dulRt1ZcohUKhUCheDrBn8d8Lwx+JyN3W2k0icnft/5+NjIi811q7TUTeJCL/2xjDUqGPWGsvrf174gW3yAfLhM4DQhBXPW+KIpijwHNTlONsJnfVZgSLDD0BWUbhooxnZ1pxXeAA7IYjsGMjFHQNqeik+WnUyayi3GPjKOdge6xW4vbkL8QgxOKQV0VrSr0RUjO9Zu1Rz/7x49s8e+eF+z370dVQFpUpBl7fGShWWqdJiUh5v3g+IpOUs+s4KSEb4KtevRGKpsG9yGnVdsGIZ08/Ug2Utzux1itrbsYcrGtHYLrJPNrS3AbVYN9KXFtH046xagblAco3xnMgRR9Kga5lqpFVRBXHQVHwUfslT6JyoRH34Fxt0xdBmSZ5WockAMu34z4JUlFlMdR18Mv5NkPrGlLv8XhxUFFWizLNV0qgPIZplyLisNYhFCo7y00KE5IVzHfEh5430dmcfxPtvTz9zcrjy4E0WYXHFN7NbdhPt6QQsPdPVv27sy2BSfcRWmnlSaM8iJRHjxfunUNbq1XLuN/7e5BvcW+mx7Pf3PGUZx+b2MSt8SzOtcfBNnn+XAFXRfzXMK8DP9haoFUbdn+XkW/zuUcD2jI2RQuIqueb+XxHeaGJ1uEEKZRp34bHKfAuDVko7aNqna7OWYAD6nbgMKyEsK7CU3QmrEWdYgjfcRQb3XtI4YsbReT1NfsrInKfiHyUK1hrD5Ldb4wZEpFOEZmQswj1RCkUCoVC8VLH2fRCVd9ZO4wxu+jfB55Ha7ustQMiIrX/PmeMCGPMDhGJiMgRKv54jeb7rDEm6nPpC8acnihjzJdE5C0iMmStvbBW1iYi/yoivSJyXETeYa0dN8YYEfkbEblBqq62X7bWPrY0TVcoFAqFQrFMMWKt3e73Q2PMj0VkpeNHH3s+DzHGrBKRfxaR91lrZ9yPfywiZ6T6YvUFqXqx/uL53He+mA+d92UR+TsR+SqVzfCVnzDG/FHt/z8qIm8WkU21f68UkX+o/fc5Ya2RXKHqjm8cYqUXV5q7oakNlFdskuVN1f8UpuGCLXVRYLocDQMpRphGKRHVMQ0mQIotpHALBp3lphwkG/cvTOHluL49qJ8qwy+fnlE3UWDBAQowGu8AJTZewHXhNCnKSDUTS4DLKSZQv+OiQc/OngJPlG8nPzSLArNo7/B4I561EvTbZIqCN66q8jSxGPiaKfp5mRSHV64+7tmPD3d7dgi3liBRjoUWVgeivGkjPLz2OGjMYNqtuipRnr6JMVCtlnKo5adn590LZWYViYiIIe4vRJRgjoKJBjlPWCvmppzjNsKeugL0VGK/O4cbqzGd5cRWlCnnXYBoTs5tx0rBEnWfA2A2HHMfK/GIOzClWQtupjMJyv1YdpWzvs3OnjOmIlmly23kXHgcSJMVcUzhvSOJgLKfm0BQSybQ/HIlFoXoWKLByg1EMVH+zul10VpbMNY/HL3Ys69oPuHZu1O9np3tJJqIGDSmjMJEWeU3YOOYEXd+u/iQ+7AN0p7zVXnVzrf4oHvhMQ1XId8ABy0Obsa6TmdB7fE5lqXPJcwY0Xwp975dQfuP14QfRTnzaUipF52O7cH9ihSAlgM0V4axD3k+ZvK3isjiS2lfpLDWXuv3M2PMoDFmlbV2oPaSNORTr0lEbheRP7XW/pTuPVAz88aYfxKRDy9i0+swJ51nrb1fRJ4dmvlGqfKUUvvvTVT+VVvFT0WkpTYACoVCoVAozhHMWf73AnGbiLyvZr9PRL4/qz/GRETku1J957j1WT9bVfuvker7yVPPvn6xsNBvovz4yjUicpLqnaqVzYIx5gMzXGlpyufPd4VCoVAoFC83fEJErjPGHBKR62r/L8aY7caY/1ur8w4Rea2I/LIjlMHXjDF7RWSvVJOH/eVSNXSx1XmuF1Cn49da+wWpcpXSuHml7WysuvGnGyjnHCvZOPClj2e0oY/q07WlmvLDEF0SYvfqKqK1ghiSYhNsdmUnSI2RDuE9NA4WTII5H9UOubBNhHMxoX5lJUmmqIPRWJUOSZHapyUCLis7DD/15ZvxLnugebNnsyu7kIU7vyVHubn2d6I8j/KGE3huvoPyqVH+sEgUXEp6Aq741efBmTn8WNUtng7C9Z3sBI0TIRXXwUl8T7ilDQP8SCN4sACpxOqCQZI9dQSSo0aqUqeW4kuJzmpsRdvyh0AF9pzfP+u6QvOsIhERCacoXxcpdZj+y41gcbC6iQOVMn0UOoPx86PteN3WoXZLpn04UGFd7Eqqw7EHue2JEyFnfUYm76aPCgew54+2YnaSx3wCXMZmP8AQz1hOkDKW1IlZomI5Fx4H0mQVHlN4v9ly2tkUG3KvnwrRwRJgZSjRSqRIm8pW55IDuL5hpSc+kr3T+Fv0bR34zPTRocs8O09UNn9+UIoT9TWKH9QpCAmZlT4UV95Z/KxK1Wdl17illcWkz5zymU40uaUxyrfQ2seWlDKdqUXKx9fcitx5pTgkm0ytBXwUoDP7KXCacoeuQ+XIiLsfiR48Mz2KfhQN6gf89uTZwIsk2Ka1dlRErnGU7xKRX6vZ/yIi/+Jz/c4lbSBhoZ6oQXKXMV95SkR6qF63iMz+TaNQKBQKhULxIsdCX6L8+MrbROS9poorRWSSPvBSKBQKhUJxjmDs2fv3coGxfknWZioY8w2pBr3qEJFBEfkzEfmeiNwiImtFpE9EfsFaO1b7iOvvpBo9NCMi76+5354T27dvt7t2zVlNoVAoFIqXDIwxu58rDMBiItHVYze96+az8SgREXnyb24+a307l5jzmyhr7bt8fuTiK62IfPCFNkqhUCgUCsUi42XkITpb0IjlCoVCoVAoFAvAssidt3dwUDZ89tMiIhLqSTvrmHkEnshR8MptG6GmSRWq5ekClClTaaguymW8S3Z9D+XDl+Ohl199wLOP/iNUO+mfR/S40AOQZk1tg/Jl26ZTnp0tQUbF7ZmgYJMVao8ch8IjNlxtD6tRWg9CqZQj9UrDEMrHN2OaMyvxp0j8PAQTLO+Gei13HmQ48QN4GOf0u3HLHs9+aGi9uDA81uQsb2ioylNe133YK9s9Aj0228+PAAAgAElEQVRCVwIKlyeOIkdeewfKzbfaPTu1hhSMO6ACDN4GRc7UtZCSNd6HMR2/GCqq47+FeGwbP/kZzy51Uf7Ck5izwrrqOB1/zx97ZZs+jusYrBjjQLCFzRjT8BEKSNrBQVxxbfsuqHw4B1gzhlIe+zxc9uu+8glne6Tg+PupQirACQoQS0IzVj9GJllxSMFESYm4/+MfQls+/ylnU0IUdJLzufF9Dn0Mfer96uw+8VcJZpqUgm2YO1ZalTrcyjTOhcdKSFbhHfpjtKVyhkNvAo/mcf+VJGv7WQ7JNj9yF5z8n7v+SyIikrHYbw9O495p2vT9GZwze0/ifldtOOZsS4AmcO8Q6q9pxv7/4Wv/1rO3fuyzzvv4feey739ijjd8pnqOl5vcEs34SZx/ZZprznua34I9EeinXIoU1DK7CvePD5DyjR6b2opxD47huS37cZ/RV0Bxd+LX/9CzL/jef68+M0TBdYtYG9kBBN4M0TopNWKsA1mOSMz5C89hsE31RC061BOlUCgUCoVCsQAsC0+UQqFQKBSKJcTLTDV3tqCeKIVCoVAoFIoFQD1RCoVCoVC8HKCeqEWHeqIUCoVCoVAoFoDl4YkyUP3kJ6HGkICtqzMnSF10cgJqs0JNVXHFGuSTe7K0yrNXNUL1NZaESqwcwfNHc8gDlxiCYqO1BQqX9CkoNqYuQLO4LSVS3l2xGu3ZW4Rqpovac/QE1GmVmjDM0qzlmzkvF8oznfifEkRfUolDPZLPkxKJVC1ByukXpDR+lRLG9+lJjN9ECmq3N/Qe8uyHC1DErGsZ9+x9/StFROR4mhR2OaiPXtGJJIiDK5FLbXPLsGc/kUDuPO53mvIBNtGfCA0JSvxl0V5TcC+sCs29kJqGn2XLs/8GKcdmFYmISDBrqA4pdQZwAavw2Bayi0luC+aSc4YxzJQ7qZ5N1tYwzalQvr5yDPcL5t1jVEqw4ojKG3z+3A26y+vuTlWsT5q1cHy2sq7UjzmtNKAxARrfugcV3ePF+eSKgrGry4VHYBUeY0cU16YqaM/1iSHP/gjV35Or7vM8JUF8RfKoZ/cXoTTdnDiD647gvBrO4vxhbGwa8Wzeh+1RtxLa+v1pPR8vxkxOxoh7vCqUq5LSHUqQ5qbCc0OqyEoYlYI51OE9F8RxXLdvef/xHhKf3Ie52jnS1gFV7xTl9DMtOBjjpKqd6sRcl6M0BpSLM5A/d74L/SZq8aGeKIVCoVAoFIoFQF+iFAqFQqFQKBaA5UHnWYGrmP2Nz9f1SPRfIlqY9eMS+alDAbhaCxXwBsECURTkjY0GQeGRZ7bunpUQ/YBMbgsHbKtwe4iaKbM/nRnNWhMM/ThAbIL1KWeqpc7FTfQKD7vxmQND49sSRUC8/jAFpCu7l1SOykM17jAZAsUWCrqD8xXLmJsizVOg5A5eV9cnumXFUh0qtz7u/ADRXBW+J5XzeLieyeC5qQteSfSGUMDBurVP9YN5Xp9EEcxe7tXb+NAq3l7h6eLly/2o+NSpuMfId99W3LQgP4vH3Y96sI771I2jcZebvM/48u3ytMaY6g24/97kQJoMpvCSAfBNfaUUmsDnlWMCx0qg55jmGyvh0wJD50Y8hE1foU6FaID5vAr4DLDxWTLPiwrynTzjrkPP5H1Vdy75tavsU4fpcZo+3kPi2MMiOAP53IhEMb65ND4bKDGFX/I5u2nN+vXjrEDpvEWHeqIUCoVCoVAoFoDl4YlSKBQKhUKxpNAPyxcfy+YlamZyTQ7udOunzvNT6lF9VnsVCrO7GY/ANdscyXn2dJzUT0T1TOXJZ0sqkZVxuOdPF9zu6eks5ewidy+71uNhtKcxjPaw93tGnUeefSk0upVjZXibvetERCwrsKgthu4ZixO1YEAdVKj+YAaquXyeVEw0OS0JUH5tUahcDudXVO+RpXsUcY/JItQu3Y0Tnh0NgDYsx3ie0NxikVWJqNMcx5imeTnMg3VgFz2ri2xpDlqJwFQSzwerhqSO8qP+UZViA6n84qQEirmlbMbRxuqzqu1kaixAEk0bwiBZDHvd1ivF3LSLn6rOL2dYneKRKRif+oGggw+huiZK6tIR6gc9x29ceP7KDXiO9VEWci48BqvwmMJbGyIFHTXh6fQaERHJlrEPLm+ESjVDi2ZVBHvCkpJtLAf1GGMohH3WFYPyN19xT5T1+a1gnw8N5bPs+ExlBaylTyEs7zdqC7er7lyMMm3H31q4iRbeQy5KXgRnY1MU58bAvhWendwAGaDJ+UhySZ1nSZFX10bFix7L5iVKoVAoFArFEoG/PVYsGvSbKIVCoVAoFIoFYFl4ogJFkfhA9X2u/Y39zjqskqhYtzv0zNNwt6bKswPP/ez0ZjyT3Pn9zeArzKVwwZom0Fqn93Xh+TeALhi+c4tn564HJRdIw1WetqDE6tpzCu1hquNkE9qT2DTl2evaqoHyjo0gSGWRqI30GEXVJFd2ohNB9ZLkvi4+iSCgW65DkMyn7z/PszuvQ2C//G6MQR8F26zE0IYHj12IcqJGTkRWenbzgerYnO7sxr27MdaP33eRZ6fW4d7REfQp/0pQhXEKpJk/BeqisBMu97F70d7C6zEe77vgZwJ82LN6b8c9yx8b9ezSj9CPkXfNVmbF1k/PKhMRWdWMeRzLYJ6m06ACCmOwN95CHBrh+FswBm1PEpVzdc5VXUKdWWd5MVuljUyIKAfaV429GLvU0WbcL0N0Yi+eGT2APoWvQEBHxup7ncWSWo17hsF8yfhWd/2r1h6fVVbuwT22JQc8+4uRqzzbVohSmSb+mtV+xEuGJ4ka9gkg+pG73uUuJ9vvs4Rjb/2CZ6//3gdERCR5HEfyQ1es9+zYwxTIdxvOmfZHUX/sENZ4iZi96SGUl/GVQ/33Ma+EueJq9xkcI/Wf9TmDZ/b8715xj/PnLTsyzvL+AoKJ/vO3rvHs/Aba21iGdVRvmc7yyAXY271J2KENWOfHHl+D+/goRm2mOq7Hf9yLexB9PX0G54xsxl4NT2DNVML0ucQq7JXYSh8p7dmAeqIWHeqJUigUCoVCoVgAloUnSqFQKBQKxdLBiKrzlgLL4iWqEreSuqDq4kydbp+j9nOgETRbVzcohWwth1ueVHp5pr7IH9f5U/xPejXqtO0ERTB9G+WNuxCu3Fg/KILcGri+uS0cbJPt7Ci1hxQsmRH45Q8fbBKR+lxmAWKPmjhoIQXbDKebPDvVjWs7XgOq7tAPNnl2cRtc6BP3gr4qXgCX9DVbDnj23lGMR2YlxmB6mChVojRK11Qb/e6Nj3llj473enZoEzqypw+UX2wz3PNN32jz7PRKjF3ytRjr+HdAVw7vRJ/a7saYfnn8NZ7952ARpe9NxHuMgkcIXEKBWQOz5UqBB5tnlYmInGpAeeIMnWQ70adgBmvvyDvdW7PjUczf9Hoqv5va+26Y5QG3YitcW0N1rAw1q3gQ1GI4wooq1Gl5AHVS3agTvQvUjLwV5uDb3IEpy+PU9gbsp1B/xFFb5P5D580qY+Xow4PbPNv0Un44yqP3O9ff6bz3nUPgEKfXoV1TWbcC63PXf8lZPpMLT6Q+kOaMCk8EFJ6IyLGbqtRepoK6P8ogP6RcBvPxzDrP/nrhVZ7d0Tvm2bw2mXobPIM90dTmzp03eZtbcThd9PkN/AaYM6rHv73nemfV5HHsH1YQk/BWmt8AZePoU51o72HUGbscZ0T7LtqTxPmd/jnsoeA+nEWtJ9GPyfPdayxU+5SjsA3zwfk3g7uxxjkXZqEV7WI62FB+vaKls17xoseyeIlSKBQKhUKxxFBP1KJjebxEGSvBWmyXyhj9ZeDzV7IfLMXlKJYpy3ftg1Lrl3KAY8xYlAeK7o/ZTYWu9UkzwiiU6ANV/ri1rj10AbeH4maFp2Z/BBmk74nrU4ugXZzmIEipRcKU/yQyiTpp8hqF6Y9VQ33l8eD++UZAoSFLxqp/0YVpwLIleLDWNtDXxYQYpZcR+qs4lKW/BGl8Qznqd4RiB5FDJDrijpVTbKdnpbAmTQvWWCA3+6/YoPv77rr+h+hb7xJ5R4P0kWukDTeqS8Nj8Bc1f+wcJM8jI5B3z4ip1ecUQnXpZagfdXHG/FIO0XfaPB+M1ma352M4izGINeAv/7JxewnsxOxyForERmGnV1O6IdrP02V4loKUh4PTFgVp3P0+pM7YqLOc07QwOA4Uf0Q+44FKBNC3MyV4VbZFT3t2ir4OtwkSoVCsOd6fnA5mJIb6M/vw2Qin3PPHKbH8MDNkkWH3vkoMUpy6iDsVSoTOJUPbkL1VdemX6NpQjn4H0L6N8ZHCx67PvqnU9mI4goeyd6/klxaJ9xM9J0gphIxbM6J4kWJ5vEQpFAqFQqFYUrCTQLE4UHWeQqFQKBQKxQKwbDxRXkoBn5gtdfDhjEyO4uaMUByP2j0vOu+UV/TUCD7OXLcZH1gPncCHn6UEnt8/iA8yuwfgyx3fAV9u03HYuS60ZXyUPrCm/nF79o7gQ9Te8wY9+8QAPrTPNFTvGZgmGoDiKCX6qLwHbWw8TB+wd6FPIylK6dJOaRcoL0oB36SLHYd7/MkRfHyaK4CieHXPMc++e/J8z96+8YRn73q8+mHwvVGK20V8V5y4qU2r8ZHpuiQ+nH20Cx+8cxs5u3q0BX0qZdAnjqGT73HHbImdJNplxwjq34OPfcMbZtNT6TU+VAjRqBVKG2RTaFddGoyT7g/COZVN4zHYmU5X7XoRAqMcrz6LKeiZMhGRMq39UBr3YJov30qUEZaspKEzqMPIcJOzPDyMsS7Q2g750ONrzx+cVXbiKMWI28JULO5XJOXFIC2aMlFf7+950LN/OHqxZ79h5UFnWx6c3uQsf0XyqGePlbD/OZULx4Ga+YicKbwPtpz07NszoB8/2nmfZ995dIdnn6hgbQYo9U0yyZw/+poruo//9Gr3mgn4pRAizKTWqWxzU/JjCZw55Yj7M4PQNNWhdVhoprhLRM0Wmogqo884OlZCZJLqw9jwnuCUMYxKvtqRQsn9+YVtwnWxYYpZtRHnSVFwFpWS1MbC3OO4JNCI5UsC9UQpFAqFQqFQLAD6EqVQKBQKhUKxACwTOs94rtJkH73XsSjJzw1J5VMX+4TTr2U6z5fQXU7pMpWD2iVAFEJsjFzGFF+pHEEbky1wW1eCpBpqJNkHU5SUdZ3bE6D63B6m1mIDVZtVUYVmcjHTYwJZPCc6xoNH1MxmUIGR0+SefhWyxAcfAZ2Y24oxY8UYU2icGT7RCPXPeB70VGJNdcxaopCp7RtGSpmrOxEQhtVMUZLnhNKsPvRMyWzA8yOkMlq3FpRcjmJfJQ661VWFzWjb6DHEhLEX0BobnE1PJc64XfWxUYrbE6S1FMX85roxsU0rsK6KpH4sjOGZmQ1oS8tjbiWbn1rQzjyWFXm8NrJMPxLNyKIzWlbpNbhR3R4mrO8edpafjIEq72xCWpDhAx2u6tJ3qGtWWdNhSrdBp1qJMyElSXlLjQ/QWt6b6fHsK5pBQe+dBs3PSJfd66e/iDXDSr0MSR05lctMHChW4TGF93MJTOT30li/2R6smWCcaExO4UMqvBWNWFesFGY09s2OfyYiEuTjdY4PlEN7ZqfdEhFpOkYKN6KmWUkbuBxrYHQKn2WEp6kOUWJ8Hjacwf+coH274gieO70O/Q74qPNMLW1X1yacG2f6cT9D1DeruMspzDWn8gqm6JnnUJ2nwTYXH+qJUigUCoVCoVgAloknSqFQKBQKxZJCPVGLjmXxEmUCFYklqr7iYhIubEvBHS3FbquLe0d2eBCu1FIT0yfV+3AgvRn1hYhIRwLu4xMQhtT7PgukDKFAlgUKlticxjODZ+C2LzdSWyiQ5XzaMxYBfZNbXfUDByg9SKWZMohTcLnQGtyjcAKu9VwnqeBoILMdFIhwCM9sbiS3+WnMzSQFr4zE4RPPEf1WIRpzQyPc4kf3VKmRUwlQZu0NaO8xSncRIn51TRQ0Y5HaxZQN+1aLCdQZGoQrvj1EqrZ1bt966AT62nTJqGfnfoK2rbj+lDwbuQ73KVVK0PohyrjYjLURmsB6mC5iDiytw5ZJXBs4iPkOlJ/f6TjDunLmGlYHchuZlhB6Dm+P+BAFt3UzXNI31OYsDxzFBA41w04M+tBNa6ZmlaUKULVVYrQPea+QEmsojz3BgSnf3PGUZ+9O9Xr22zqQoojRn3Gn+dmcgOJ3rIRDZVUEa3hqG/bNTCoXDqTJKjym8G6iYLQfO05U4VpSgMVIQZzGmE4Fcf/OpDv4aXq1e9wD7ticdZgZyuwWN49ciWBfcboUnieTAvVfaEadYiMp5egc5fVWaMI52r6e1HmnsW8jk1zf2Uyxtb0wTPS5ydIvIQ7EStsj3oYzLTuNhpUt/apNKwH0UsKyeIlSKBQKhUKxtNBvohYf+kqsUCgUCoVCsQAsC09UcCIoyX+rqjBGr4U71BKNYFnhxv5TerN+7/aHPfurjyK7+UzgxMF2CsBJ9z54GNEBO05QDrlVqNPVA9fwdA8C+63tQPmZ9aAI3nb9Q559y65XeHZkCEN+httDXTp4BO1pfxT1VzxUpZWmt4Caig1TAEHE3ZPm78ElPr6VKBhSo0z14/kxcmtv3dDv2ScP9Hr2x276tmf/+YNv9eyVd6ONz7wDY1PZh/vfOXyRZ3/rP/2tiIi897H3o723ou7gb0IG9MzP8Px7u0H5NVI+wMZTlC+rCfQGq2B6umj+YhjfxAn3Fij3go4o3gEqwL4BXMDwD7urxk5cF552q/OYluDgleUGtJ0pmNb2ac/OF9Gn9AQmqvNyBJ1M3QG6h5HvcCutyi21wSFaRCiwYJmUU4b3Ht2uHEN52xZQnlO73JE/y+Nuns9sxLxysMR0wiev4a7WWWWmlYLOnqSgs2uxCDjf2v7bEAiWcWwCwTOznaj/6NBlnv22z6P+3pMIOsvYcwQqPxMkOp/Uuby3v16onlecC48DabIKjym8p3/77z37pkPXe/ZTp7HG33He45798CgCfCZCbjVzxx53eSUy99/cM56O7lvd+yo8jX1VIVo9mMcYnbqIgriecT+z0ok2FjJYV6MtqB+jfVukIMNM4ZU63fK8+Olq+5tfhXU9OIr7JU5TPlSKi1s4jt8BhuhK04z2ViBGPftQT9SiQz1RCoVCoVAoFAvAsvBEKRQKhUKhWEJY/SZqKbAsXqJKTRUZvabq5jUDUG8IqfOMjyKP8dXdV3l2by9yrqVWVd29I8OgjEyc5EeTcI+vuAMB9o58oBf33vYVz/71f/iQZ1+74hnPvo3y7t2ye7uzLbluDPkg5eMLUq6rShaqK1Yljm6vBr5MUeDPNWdIiQQmVBoG4D6OjeMmA6/C83/1Nfej7Z9+g2cHXk8KvkHYn3r6Os++5Dzk9XoqTsnSyLVe7mIqBU7P/3XqBhER+W8X/tAr+2LLa9CnBBRMR3qhROpqBsWV+A/QAv07ERD0l95yn2c/8PtXenaSAnua43Dhj1/opowMKRFTV4FuCj0FLqCyY3Z+MONmz+qCA4Yy9IMwBTztw7yPVUgVFEKd7kewTk52oN9rD/tE8FvpI6mqqUEN0XkcP7H1SYzL+BW4d7ABY9f4IHiMkU7srSYIMevR4G5j5TTUYxWiNyNj7rlhFaFXxgFB+TIa3/Ao9nl2BU1U3XnCKlyU5lvch85VG445y4ezoHXiIYzZWA5jNnaIPiPoreaFTIQpWCTlwuNAmqzCYwrve5vuQB3K4fbjLM6ZGEWXDBt3csKxLT6BW/Nz/wYO1HLgjW5z/2qJD2FyeEzDaV6HFBC0gYPqEsVMe4JEu3X35H0bfgbj3rkHY3nyRnc/dr51t4iIBGgR/NswlJh2DdpYfhp7lYPURiaxlkLHsMYDPjGhFS9OLIuXKIVCoVAoFEsM9UQtOvSbKIVCoVAoFIoFYFl4ooLBijQ3V12v6T7QKBXKMcave/XBNkn1NEKJvSCOkaZYlfoZyZMsgtzjdcEEQ5SDiyiCoqWgfaQqGSlS0D6iEbgtgV7cJxmBL/dMnu7JKinysnN+qRlVCSu98m2gzyhFl5RI2RQocOBR1Glm/o8wmoXrm+tnhhA0MNAJirKD8gcODtAY0/gJiWD2D9Vyn0HIJ9taBjy7Lw31VSiEwShSri8bgs1qm+7ImGeXYmj8ZIHWVZjuE3XzbxxYMNFASr00xibaMDugYNnNhNSvX951pHxjyo8D+9kw1SlRzq401Qm6+2H8PoKorXlL65rbUp8nDSbnmTMUeNNmqFM+dLsfmKbhueGcZAxe5+6fU4OpT8xeceBNnhsO3Mr1S37zOg9UfAaEVV3BWtRTDvwZIIqfDz1WcbIKjym8RAANPpyHcpPPKz8UZosfq23MzT2xM8sj3+Jed3zW8jnGSs9SkXKc0lxT2kGx9HlAnVibPgFppP2ZC2CweQ+xSpvRE6ueI0+nML68lxrjOCDGqF1M1fH64aN2PrToUsCIfhO1FFBPlEKhUCgUCsUCsCw8UQqFQqFQKJYYVl1Ri41l8RJVzodk8mjVh2zYzc5u2nn4zII9yAV1ehRKimK66m/dtgWKskODCAjYdB4pLTpxXXwIDbgjtc2zG58a9uxNcQQ83PsM1GMjPw8/9MkRUFylPIb8ImrPkREorRo7kBssfwCcV+eeqnt6ai0ovNgQ/MTJBOi2chQDVorD5mCQ945u9uyGQdCbrQ14/uRxtD3yDgSsPDkFn//YIeREu/ZVez37x/u2ePbazRinN3QdFBGRw/kur+z7T17i2d99PQIIfjDzLs9+37qfeva3o9d4dsdTaPvnjrzWszsPQSbWlMDcTB4HZxOccke+KzZhHU5NoX6YKJjcFCdarCI+5D6kmBYlgZTkRil/IqnOEquwJoOc4M4SfRwnJdukW/lWHnMHuAy3V9eSrUv8BbMSJjVRBo0vmgjVIfqvkYMWuo+VQMStBquj54jWrgT9KKHZZbFhtNHQUIRpfItJov5T7gOF64QpKGop7m5LwEeOuZEkiiFq8FAIKsbpIVBFM/PASr5kEnRUYwz0EefC40CarMJjCu/mtqOefUsK59uVsdPOtnfscc9TKOsjPSVUasOdPOmmyZqPYnKKScwBr9/K6yC3GzqCfvDeSq8n5VuW5pXSWU6uwzi19NP8kRlOuqVyX9pXVXpft+EA2t4Mtd9gP8Y6MUk5SFe5x65AQUDnk4NQ8eKB0nkKhUKhUCgUC8Cy8EQpFAqFQqFYWuiH5YuPZfMSZWtu/ObDbueYDZB72EckMtFMMgleLbV8VUNpUCHd7XAZ50oYhnw7BTkk7/X+NFzv5Xbc57rEQc/+bhlUUilL3Agr7wro35kUXPvdLcjJli6iHyQGk0xXtbzYiAGY3AxKKbOCXMZEaST74WJmqqNEHBOrY97Ysc+zb5W1nj08QXmhfBQxJ9Nwc1+0AXQBq44en6hKJztjoKwCFBTxgQxoxt/bcLdnj5Xw/HIDOCCmLgslymm1AnNwU8d9nv3l6A1oV8x9qtQpkah/NkTquKDDde+zNoPkwmdVVnQEF+Q6SK2UQ/9YFRTtxFptWQV6tdBEEsV5YIY+qlPvke1Ln/NSpkfGE+hgKUFrv+5a9+Bw/kAOyGlTbkmccSiqQmDyJd9K48Xju0Lc4KOFpjS/AXSaGXW3Ze+QO3feuhbMTYkGsysGWrlMTOvgmeq+GYnRBqXxWtGIvTIVxIWcC48DabIKjym8dyRxztyZASXfS21nRXAdzPyJCz+Vaq6Ngm02u5+TzaF/nLexRHGYWVVXaIIdG+P9yYE6cWmG9pC17tx55VoOx9c2I5jyvd+5wrOvvWmPZz+yH58ihCnAJn8SwBR+XS5KxYsey+YlSqFQKBQKxRLBigbbXALoN1EKhUKhUCgUC8Dy8EQFrNia0ijXwQEEUaVCNIpv4E0KzhdIUC66WqXmGNzzYxlQDl1JuMqLGaK+iGVoCUOZcXoSiriHcutQv0QXFIlao3xj3N7WGO4znkN7VjSgPeM0Q6Fc9f6hDOVnmsYzub3Zdg5a6B67ZAg+7vQ0+v3Q5EY8M8XqFTw3GkWfChH42duiGKf9I1DfXdoFaq9Yi0raFYUKMBTG8y+KQbX4jdFXevYHV9zr2d/Joz4HzwsRxRYawTjeNQ51ZWCSktdV3MEHmc6KxWj+UuDiGps4CV7t3m52oO5+IQq8l4H4qE45WabgmWEam8ZTmI9jp0DTrBh0B071CyZoaoo/SwFM+Tl1+f34r1eipjln2eQ01kATUWt1z/T5ICMyQuo/CjgamfCh/xwUbNGnn0Uw5nVniB8zxUE4zQgpEVvdE7umedJZ3h7FIHCA0jxF5OXhaGqr1k+SCi9HQSdLNE+dSdw7EcJ68MuFxyo8pvDemHD3yS8YZDA3tzpvplOViHs+QjmirKNcjnuvasK5cDiK9kYnqV00l7FRN20X5X0bwPna1Icxmwi4+xSs7b9HpnEWZrfi98d/9G3wbEM0Y6GT5oCVpqTKDp3D3Hl+uT0VC4d6ohQKhUKhUCgWgDlfoowxXzLGDBljnqKyTxljnjHGPGmM+a4xpoV+9sfGmMPGmAPGmOvdd1UoFAqFQnFWYc/iv5cJjJ0jgqkx5rUikhKRr1prL6yVvVFE7rHWlowxfyUiYq39qDHmAhH5hojsEJHVIvJjEdlsrXX7mWvYvn273bVr1wvujEKhUCgULxYYY3Zba7efjWcl23rsxdf+/tl4lIiIPHzrh89a384l5vREWWvvF5GxZ5Xdaa2d0eL+VES6a/aNIvJNa23eWntMRA5L9YVKoVAoFArFOYSxZ+/fywWL8U3Ur4jIj2r2GhE5ST87VSubBWPMB4wxu4wxu4aHh11VFAqFQqFQKJYtXpA6zxjzMREpicjXZooc1ZzvpNbaL4jIF0REouu7be9XP1EtL+EWpk6RR6oZn9fc5ocgk8GPa9QAACAASURBVJiGaE5KLVU2MdFHQdbo9bFMipzWK/BCd/XKI579w29d5dkNr0ad4u3IwVd+IwLshe5A0MnpXnpWEvKI+GmSHxJYtcJt+4O33jar7j/3Qb3WGIE05dBpRBZ8xYYTqEOKvIe/gyBxW96CoKHHv7LJs1e/55hnD38eHZnqxQDmLoAyrOExqGAKLTRnNK+936+O09T5UJf1vwlBBntvQd2JDQg6GR/D2J33ewgI+uerf+jZb9n1G5793s2PevbXvnydZ7/+XT/z7NvvRwC9Y7//B569/m8+7dmsmiu0og3RmqrswJ99yCv7zH48h/HQOFQ+V7ZgTD//PXw2GCB1aXyQlD2Uo67l5/s9e/g+BHp83X96zLM/d8U/e/a6L37S2Z7wSHUvsFqHlaPNF4969ugxKKRCacx79Hwo0wL3Y72HdiJv3OM/93HP3vqxzzrbUqHAjKxuzLehcUdvxtyc/+2/mHWP3Dj2flc39uH4Y9ifFTrtGvrRWe53gVIpcq62zEpUOvjfMN9+feLzpW6MqQ0rrsZcTt5WnctwivLDrcYzG/sqVI6bd+yB1GtsCwaygCmry4XHgTRZhffQrR/27MoZ7H/GQAlq1zBF212xBv3Y8JnqvtlxFXLOMa5p2+/ZOYu9vSaMOfvb33qnZ5/+dSyI/7z5Cc/+7vdf49mXXYd7rqBgpv9++ys8++KdON+GMpBs9u9GEOXDH73Zs193d3U8Bu+DD6DUgPGKbIGCMD2IgMehafeZ7qcuP6uwogmIlwALnk5jzPtE5C0i8l8sPqw6JSI9VK1bRPqffa1CoVAoFAqFC8aYNmPMXcaYQ7X/tvrUKxtjnqj9u43K1xtjHqld/6/GGJ8Y+i8cC3qJMsa8SUQ+KiJvtdZyRJnbROQXjTFRY8x6EdkkIo+67qFQKBQKheLs4UX0TdQficjd1tpNInJ37f9dyFprL639eyuV/5WIfLZ2/biI/OoLbpEP5hPi4Bsi8rCInG+MOWWM+VUR+TsRaRSRu2pvgJ8TEbHWPi0it4jIPhH5dxH54FzKPIVCoVAoFArCjSLylZr9FRG5ab4XGmOMiOwUkW8t5Prnizm/ibLWvstR/MXnqP9xEfm4388VCoVCoVCcA5zdT6I6jDEcu+gLtW+h54Mua+2AiIi1dsAY45c+PFZ7RklEPmGt/Z6ItIvIBEUQ8BW4LQaWR9oXhUKhUCgULyWMPFecKGPMj0VkpeNHH3sez1hrre03xmwQkXuMMXtFZMpRb8leH5fHS1TZiExWlRqmmeQ5rMgLzD0GNgjFSCVCkpjQ7IRBlTD9D6lz4mF3PilWDTVGoXCbpLx04TAUZkVuSxjPt3VtgZKDlUNMKJfjsAeKVelQqgT5XpDqJnySMq2JIclZiqR/EcpFtToOpdUpym+1isqHOGehDxHM6bvK9CkfD6sNVC8OZSnvXxZjEZ5CjqpgAQMTTqN+VxQqnOEK+pQZhzpwfXTIs0Np9GlDnEJquFN81W25AOVzs0F3nq650EG51FaTEqlua/MzeRnSHLfFcJ8hEgKti42KCybqZtOtqY4r56dkFVkT5ZkcieIHlTzGoimOOlO0n1qonBGanWpQREQoRZwEaAnb53E6GdrvdbknbaerugQK7vOE5zpI3fCba99vP6y7jqUxjoUwydO1cyRI7QqQojVI4xKgtlQi2IistgvmcC3vM04a6JcLj1V4jFUhd55JF+JB9znaGISSN2ZRpzGA8mDR3a4kTQLPTUcU7W2mxJQ8l7z/OA/haZ/9PzhZVfCFaSjKSJspkRDO+kyFznrOz8iqzDDK+Qx5OcNae63fz4wxg8aYVTUv1CoRGXLVs9b21/571Bhzn4hcJiLfFpEWY0yo5o1aUoGb5s5TKBQKheIlDiMvqg/LbxOR99Xs94nI92f1x5hWY0y0ZneIyKtFZF8tWsC9IvL257p+saAvUQqFQqFQKJYTPiEi1xljDonIdbX/F2PMdmPM/63V2Soiu4wxe6T60vQJa+1MAMGPisjNxpjDUv1Gyvc77heK5UHniSx6nHh2pUrN3cpB9epc7D73CCxSm+puY/34o3lcW8PzbRfXD87n2vnc3oeGYvg9yrgCvs2DFvFDRJiumLv+vO7Pa6XiLnddG5xHgwPGTVf4wVJgw4DPwvW9Z+W5B2Rew0X3MH73W6ztO8f4+oGXFNM1vo/hYfTp0lKnrrDP5yw4i0ESOZDmYoP3h5vwe45raY3z0PGeCND9/bbEfM5PL6Cz33KnBli6n6k7OGRu+2zC2hdNsE1r7aiIXOMo3yUiv1azHxKRi3yuPypnKeWceqIUCoVCoVAoFoDl44lSKBQKhUKxZHg5JQY+WzB2Gbj3tm/fbnft2jV3RYVCoVAoXiIwxux+rjAAi4nGlm572et+72w8SkREHrjtD89a384l1BOlUCgUCsXLAefeZ/KSg34TpVAoFAqFQrEALAtP1FNnBuW8v/qMiIiUGjhCGamC5qHeqsRxbWgCUQQDtXiDHPyRg0JywMFKhIJnpjiIGtUPuRvDipFKwt0WVoxwwE+/9pST+EEwU33nDRTxoBI9Z+bnIiJlKjcctC/PqhJ6JgeDo1fruuCZSdwzPI5G8j2LjdwelHPQ0JkxDuTxIH4Oz0GdioqDfVLAOh4P7mslyosDJqtjLD3r+G992LM3fvIzaDuPcRaNmLn/sd/9A1z3KVzHqBtT7hP1o9yKAH7BMSw4DrpYjrnVebyujn4I7dnwmU872+ONgXWUSf3YBXI8YO7Ao/znmEE35OjNNDafdI9N3Z9y3D3aZ9yn8xz3qfgEDeV7BNYg2mdpEJET6/YerQcpU7+pr8f/K9bJnOP7HKjQnptZt/Vrw32dn8qY54nLK36nPFXiefLtkw/42sqZTSIict43ftNZ1/qo1Or2P52LAVpL4jc2vA/KvP+xEPis4bOOn8trbONff2Z2eyngc4DOvBKdeYEcnsNt5zo28vzUuYsJ/SZq8aGeKIVCoVAoFIoFYFl4omzISrGt9trOfyUE6Y2d077wXzBkN+5Gyo/p8/BngEnW7Gn8icOxPYKUasZS+pHoOqQKyBxq8Wz2lMQo90auA+1tPISh5bZIjP70oWcZ+pMn0IIcD9zti7edEhGRE5OtXlmpjHtMjTXgugieE4vjfpEQyjN7cZ/Wi0fQrLuQ69G+DilKQndgDCa2kldqM3Ij2L5Gzy6sxrhye9Z8t+oSHNtCHpZLkcYl8TDuMb0e14XI0xbZjPpdTbCPHkQqptZupKyp3NWOdr0OqZW6mnEto/UZ2Jm3on+x25s8e+zq2ZFuui4ZdN4vHEQ/okGshwMHkBczdgKu0rZ9qM8e1JHLYHc8AXv0BneqlUqHOxqPnfnLnP/SJm9LrAkpNkpHke6D04nke3Dvlt1oZOlapBlidOzxSbVCW8KUUWd8s9sVs+aK2Rkc0gU8/7LOU559526EkSmnUMfUuXDIu0YpmuKDOC+ya9glApSb3Gl1OA1N3Z//NN6/e8U9nv2391wvIiKRYUoFtQ3rLrQHc5DdgrnuvhXnzOg22PkWPDN5kvYZraVKxO0u23HVAWe5XyoXxowH6vC7Puf8+TenceZsiZzx7O9OXe7Zt33+dZ5t3oR0RkE6DKeewH42dP4E2FtEdmcj6pzcs8qzy43u+Su1VfuafAYDltqIumYVzlSbpzmjX6nlsI/bpzQPN+VSwIpIRV1Riw31RCkUCoVCoVAsAMvCE6VQKBQKhWKJoY6oRYd6ohQKhUKhUCgWAPVEKRQKhULxMoCq8xYf6olSKBQKhUKhWACWhSfKlIxERqpNCZ4PtRSrK8w8XqHDqZhnxwdIqdJafVdsOEVxlFBVKkOoG7sYiq5YGIqc8hgpXC5DvJnwYSjJiluhmgnvg1IwRm0pNkLJ0dBPsUa4PYO4Vi7CeFzQVFWz5MpQDZUpY32K+s/oaYVaajKPOpURPL85irbnIcySxjjKS+OYg8ajpApMoL0th9GedDfaGcpA5ZLoryplyhHE6hnoRbtWPgMVUDmKeyQG8Py1r4YKbl1izLOP51Z79iUroOLam2vDtZ1QIqaLJFdicJydCsdGclefQTKSd5aPZdHXrhbMaWQENyxTbKZwimJTkYoq2EPzdAyKrY0rh53PtUX330nh8eqarIuDw3F4GtEPjrvEcXWCMVKsGR/lG9/eZ+zyDdS/Asf5cdefzM5e55NTGN9sG9YMj2+xDfeOD7gbk6c6YYg4pZh014+fDDvLOQZUXWwtilvVsgPnSPJ49f6JQQz2WAJq26ZjKK9E0P/wNNZDnJTCHC+p+SjmKdeGOqGce56uadvvLG8MZj076PNxzUxXWYXH+MVGqH33F7CYbm5H2q87Jq/27LEMFsFmWuPjUajz2howBue1YG8/MQDlK6s3gz0Y98AxjHEdans+hKp1sfFCnfS7oUAxCSlOlCWlaaUR9QNRtyJQ8eLEsniJUigUCoVCscRYBrlyX2pQOk+hUCgUCoViAVgWnigbFCk2V93VxUG4V61PgE2/4HVNcfwPB8cLJKv0UJ5ogGITheHnFCJEOzGVVWyk9BwFDFsQLIKUcigvUVtyqynwZ4Jc6wW4qksNlHaFqIAQueWDNV4lQtEJKwH0o6UZwUFTGfR1OI0xjRNFWUDsTClRXpIiGErJFilAKdEu6TVoY7yFKKZWjF+xE7RcKQOXd6mhes/MCjwz3AS/eaEZg5prp7khqufUNBrfnQBdWWnFM9MltCVHNM1EDvRj2boD33HqiVKRKTeiZhxB8yJBt6s+EUa7GoII1Ff3Z8w8YvAVMkRvUvl0wc19JdoyzvJMubYmOHULmdEQ1lWGU1YE0eAQBW4tEQPdFHNTmsaHxeD0GBxs0/gEJUxEZgd9zERwk+4Y1sMjPL4NqFNq4JRDPmvAh05kMAVbdy2frHRecaDeuvtEZv7LgTFRl88TTv1TCVHw0xbYhWbUKSbR13wz3cenfznrpihjFuPuG3azdnsOpMlgCm8r0fn3UTolHoNkAmspW0K7OEBrmej2gQyC4fK+TTShxcNUR1rcizLSXH1uoRnPZFq7NYl91T+GxV+XvobWhsmhLTY3xzcBSwj9sHzxoZ4ohUKhUCgUigVgWXiiFAqFQqFQLCGsaLDNJcDyeImqiJgaVRPuBiXllwfJD5FpuGljZ0idV6NyGpBSS8oxcq+SP654JRQonJculIHLOBSDazgyTdRQEq7nMOXpiw6SOq8JD2s4iedWIqTqIG9veRX8wx2hqqotQrnXUkX45KfTRFdm8fx4M/JGFahPJLaRCtFa0QnKuxVFn4qT8GcnzuA+qRWgzdr7iZYMoTwC0aOEJ1K15+Dn0+R6j42hfw39qJM8jecnSU3YyhKaEsaxMYS2c5/aYqjfN0mcJoEpBT/62KWy8VP7saJsOApVnd+hViHakCmbWLJA5Zj71hhNJiEzHneWzyhGee0zrZZNoB8BypfHdEWBqO84PX54gvpHyHa6Hd9MnTK15keVTWRm9ymfRluemOjGDzhF3ggeFMz6cKdEqwfyvAbc1f9fe2ceJ1dVp/3fqaquqt6XdGchK9lIQLYYtoCyCLIpiArKOK+OoiCj4zsvoyOMszi+szmO4jALI76oOLghjoDAEJYBWQxbQgJZyL51Olt3p/eltvv+UdX3eYo+Bdh2J53k+X4+fDh96ta959577s2t33Of349rCRb1czfJQAGdy5YUHGxDx54lI143O+miVEMyOkg1LHvplYMkvhvv9Nf9iw3kvP1Tyw54+6sj/jnGDM0ProXHsAuPJbzzyjEWPgY9dE+bUInrNlNOUifd0+bWwJ23q602bPP1x9eQbaN3F4igIBGW4Z8jS2N11tFLczDmfxUjMkgSZSVdOHH/cReHJ+PjIUoIIYQQY4YzMyd33qijd6KEEEIIIUbA+IhERQPL1eRDzm4Nwqv8zPx20pO1nYKlAkpuNpQ4rWMhugJyvkQorDz555Tgbj3W0fsXyLxX+RDG2HMZEieW/xr97SeXGAvJZh0nkKRIYWBHYeAPzVobtr975+VmZta0EjLVjoshY0x6kRL1HU/OlDswrnglTnn1zdA30/86OWznPolw/uCdU8L27ktJZquHRBh0ILTdeiY5EUnuGujDdhsuy3/3ykkI7T/UckLYnvhXcFfFB9lJR5LnnyPZ3lO1GGP5pyE57PjSvLBddgsS9XX+3fSwfeBq89I/Ce0kybfsQpvYSNkYCyRu9F9SDTMwr7KbyYn4ZzQ3SMraeRln+0Rz2k/gaGp5Fzno/g6JBe3dtOESyTZT9fnv8rzji2zm3diPHRfTIpMgozY9Bolk3/mYk42PUgLDa9DMXeCXibpbsXyM3Ktuh1+KTK+tGdYXJcfajvWzwnZqNs6dS+FY1L8L7jGWstu7MJbofOxT0M1eSDC4wC9x5ei4O3oVISC5+T/vfU/Yrj1/n5kVuztj3RhLZBHmjOvBWJpPJLdogGsyk8b5y52L66l/ABLwlJrh89fM7LYbP+Ltj6b9MtRFT6I9JM0+8J1zvctyIk2WzFnCe/Hvbw/b8354Y9iOXLsqbH9+9Zqw/djiiWF73ftODtvJ38f+NdyGY5mqxbFpebd/n4bmRM3lu8O+zq1NYTvTAsnakXTL8jg7XxP7sc1D6pCTkjjqKBIlhBBCCDECxkckSgghhBBjit6JGn3Gx0NUJLBoIYyfrqLkjux6KEpK6HdLJckxNsD1qhL5GCYna3RU4yibRIyz5Tysu+W9FNrfinENLCJHzAaEdTuPQyg+uRdjGaRkcLkEvsvuH8fJ5siVdO+Lp4Xt+vPa8uM6CxJeltwjuy4iW18OssjGTyCEXz0Z8mP3VtKsrqB9WgPXUMdFkEMqNlHyygk0deqwrWgn9TuMJ0dyS+ujeenprjrUuUs1Yh1tezEudm6xazHzSSyfrIfENNABiWnTpyhp6WsUir8a+1TW6k8sOMi1sTbAwZedShLlmsI6Se5adzNq9BXBCR0d9s9RElKWNIISSRmbL6J96sac2XqNb2kzl/O7x9yQG6xEQsvmC0gOokSF7H7cfzoGHNuHudF6ml9879k5XIYzM4tyvTGaP3z9M6mpKU8n1tHXhM8dOUADckXtXQMJqOi9ATocvf2U+LfSP5ZIi79eJTu2+HbFSTgHZ0MubFudn0uO1d0KfLGtC5I8J9Is34P95oS9jqb1vs2Q6nk+bEpQfTtS33Z9pmQqzbckdBle0ub9nGvhcSJNduGxhLfx45D2jm38dNjetgbnOHUrvpuYANmTr9stH+EDS07HXv/8zxZeP2jZhWMXIemb7+N8TIMSTr0sOS2jnukrDl/Gx0OUEEIIIcYO5YkaE/ROlBBCCCHECBgXkSg3GLHo5rwTJzV15KHkdC2eCRNtnEUw305zIj3aDCczXHTq5rB91cRXwvY//gccK/WX7g3bB/4bklTyBLhB+voQSo63Dx+LmVmGwvUcxqdFbPrDCBvvvyF/uj46f3nYd9ezcLucfer6sP2bDXPC9hdOeyJst1JhvEfvOydsn/jZ18L2ipdPCtvHnrM9bK/bDrdb5U4MMtUDySRHjpQouQwzVeiv2JtvT14GZ9OBhXBiJQ9gnzOUNDCIoH3tlx8L27sGIUssvfussP2xT2KZe/7lwrB93oUrwvZDD55pXkj6qdyJy6T3LEgQtU8Od49959y7vKt7qhvW0MtrV4btT/34c2Gb5UpHMgP3n37uurC97PkFYfvDizAnmFhXiQSXBbnQpf3bufw9L4XtXz29OGwnWrG+qRcgW+yup+B4nHIKnJBMzUZ/zbABGC2Lktr2zvQnifQ5DtnRGq3GxR3pJim9DNuv3chfRpNry3HyysE6/3GMHyhVe9HvPAtYzqPkjTWbCuOlXU7R/aysm2rhVfvHEqXXA3KU87V8HyWmJPUx0ekPS3xo/kpvf1UUcz/qSti8CquMlkiOPH8y5gbXwuNEmuzCYwlv62X/L2xfdM0fhO17f/rPYfvG7ZeF7bXP4Jq77KIXwvaOfkjuzy/DNcRUbM4fwKF7lZlZ1xxKBJv2y/A5dlnTtZWZQHUbD5k9LzDTO1GjjiJRQgghhBAjYFxEooQQQggxthzSHFVHKOPiISqImaULyf9cjFwPb6NeHpOtJDcYh9MLZqH4AQTeImQgipF0svln88P2PwVo978LiewGl5Gr7BQ4w8rWkIurDvuRK5FUrkhyLHIIYfnKW3aF7R2vzDQzs18+dR72Yzq++Pr3Eb6urMM6frACIe4UGaRylCj0yReR7NLOwD6tfHU2hjgNMsngBK7BR86TweHH3cws1oP+/Rfm1z9QQ+4cll262THnrxl2962X4rvT6Xidj9pZd//nRRjvBTh/j9wLCW+whGQU34sxdL6DaiVugYTXtmS4zeaWb3x6WJ+ZWd8UjPGJzWeH7Ybf2xe29+yCLBlrw/ZZ4nnlwePDtmvAZHrkJ5Axv3kKlmfJmCkrnA92vTqSuJ/6wen441hsp386BrP7UUh4g5OxTNtTSH5qOAXWeUoJWxINMUVysOv0Oyedp/ZY7STM5Y4WTPLJ78Tx3fc6HJrtJ/qHUkb5J/unkBOy17OwmfVP8TsR2XHIkGnYAnJsti/Kr4fdkiwHsbM4oPtijp2IdO/kpJ69x9JY2CVawv34y/vP8fZHcVso2o9b/pb6C7eFrpWk0RIHEujnezDXwuNEmuzCYwnvsXt+ELYv/uANWP7/YJ0pum55nyI0DaMlXJfJJfn7yCDV5TO6v+fYPVuPFUb3wH3ILutYO7uWvZsUhymS84QQQgghRsC4iEQJIYQQYozRi+Wjzrh4iHJps+TufNi0n8L5JU93qXAofTdOiQhdIYTNUkSEwu25csSVj1u4LWyf0wCn3l3fvSRsV12Kulsdv0YytuCd0AJS+1HfqmgslJitj8bDdb0CSuTW8/Vp6P9IPjx90pmwFj37EiS8JTeQ62zdO8L2Te98PGxvpKJwL3wTrqt5f4QafcsfwHePuRAOrK2voD4bJ/njhJgDE/0yZqYK/dXL8xahJphwrOIYdvjR99gAR+f9I3/8aNhe1QVZac2PIHd9+Pqnwva9d58XtpdcDdflc7841XzkjoVzsPFxGsQHkUQw8rOCNPEJfHzTTfd41/dcF5yN1zS8GLY/d+dnw3ashmQaSk6aJanj0kvgwntgNVyUl1yAfTKDphEd8F8s6ZqCfJ5hmQifv++6Z8P2j59bEraTu3HLOP79cIOueAH7N/k9qMnIVK2Le/v7plDSTnLnpSb5pdagf7jLr2cvpNDodJy7rmcosWk9jmM9hl5EqobktHZyYyV8S5uV7/Y7DrPkgmNZO0cST5aO/YSX8+thuZ/Hwm5iHkuKklcmUCKvSG6L9WObvM5kG91hSYU+9SI4QJnGBCTxSKmXawrdbn6P9+OGSmiCWboXZkg241p4nEiTXXgs4S39rx+G7cvW49WF/Utnhu3TbsS9MU76+AO/ead3nOknG/P7QecuQm5KRzJg2TacgzS5kI1k1+xk0hAl5x1RjIuHKCGEEEKMIUHxQ7oYHfROlBBCCCHECBgXkSgXmEUL4erZ8/a8+cJmlgv88dDmFXDNFdUEKzwq1qynulzskoniWXJ1DdbBIev+ySQVZri+F9aTKEOYON1JydhYMqHof9F4qD9wGE/LuyhBaFk+FP7CjllhX5Rq7v1qBWxZ5Tugsa1ZABmumZKApisxrgWVSCC6YScksf81dVnY/vtnUKAthtx4Rrkure51ctlRElOuoxcdyO9Tqg773zcJ+9H4Klx7g7RMvBPHd1k7XIMTEhhMz0xKjkeZBRPt6J+ZbA/bT0z0/zTL7cN3U9XYj65mHL8JHnXq8QPHD+80s1nlkAF/1YHz5BZ3hu1gCxKhVm0jeZd+6qxcAHk3uQFjrCjhfMsm/bLLkNOx1C/Tp/fOxR+0DK9vfStkl+R+DDKd9UtcZT3+sbCEl2xFO35CCUucx5036WS4MrsHyCG1mBJvboBrr2o35lJAbthkByXb7MF20pX+35sRvznPojitxck2KWFs/HjsX7SQeTM2QO7kLLZZuQf7karB8W2jJKCDdXyfIWcYqatJug4od2YRE5Pd3v7aGGTSSImXLYZenYiUcFbPrcN52t2H8zG3Bv3r3ndy2OZaeJxIk114LOE9fNzDYfvk6j8M25PieNXiyb1wXbPjl+mel58ftWvpBk+L5qjeamYiOSSplmLANfUoQawr4Yo8KOidqFFHkSghhBBCiBEwLiJRQgghhBhjFIgadcbFQ1SuPGd9J+VDxdtXQk5jGYNjZiXUPGugelhdc6nWVF0+5h4/gN0dIHmOk23WPQJX3dYkHEfV74Mck3sACeNyF1Lo+zHUZKpJkaw0m2pX1SD+H+/EeAYbSLajhJXZGXCz5AbzYfxgP+QKDg1XbYSENzgB/Wu/isyCHbOxzNyPbwjb933rgrC9/z0IT//zt64O20maLV3zyFF1DELuqeVVYbtnFiSTeDskiEhB3qy6BokQF9Xg+D69CFLShAZIbwcGoJ+lboOct60JkyP5Xugoy/8CzpsDF+J4PHrLu8P212/9MXbK/iRsHffdA2H74cd+FrYvX3Rx2L79xV8UWjeFfau/A2cj89IknNNyqsfVdx40lQS5OBMd5OzhOmhfhFxQcxLOwbM3UQ1AlAy0IOnXm7IFCZtln6Ib7O1ITBlfRHUSKcFnxT2wK3XgUrHuX/mTbfY3+S/c2k10rVbRB09BOrUrqT8zfD27WzGWXB8mqktg/wO69jqvx3XLrwdEI9i/jnYMprbeL3H1LCyhiXFSSz7GlARzVhXkvF2X58ecpjqUjZMxB7dvhWY+4Vj0J5c2Yixn4TqsJhdc50y4SyM0lkTSX6f0kYdO8/Zzwk+WKP+aEpfmCs7iUnLeyt14tSBD9ed2teH8JX8f0lt2A+YA18Ljg5bgfwAAH7xJREFURJrswmMJb9WX/z1sv+M29E9/GPeU7PV+PTvekD9+nQvJZsnnke672U6cM0dO5QTVTM110Qdy5x1RjIuHKCGEEEKMLU7vRI06eidKCCGEEGIEjItIVKQvYhWr8iHnmZdvHfF61hrCupywckimGGjyJzBkx1GUEmZeP/+5sP0v98MBcvanVoft5T9HLHviVTvC9qZX4aLKlftDxiy5cZI2R/XOap9FOPnAGXmZrWk+nCydL8AhlT0DY3fr4fQ652+fD9tJytp373ch4f3wq98K29f9NawvX/3K98P2H//ik1g/7VJ6DyUWpR86LOE5kjfihaSA/XdD9nl5MmTcmSshJ/Y1QTqtH8TKZ92EbIkZ0n1XPbogbJ/zd8+E7Se+jnp1Z/zNS2H7i+Q4vJoMaVuvxnZZChj4Enb8yn/80/w2b8P3PvflX5iPP6iBdPkEJYt8shtuvvvWvStsk7HQsuU4dhu/DGmm+jnsd/Unsf63QzDkcCMHEUsU825GwsXmFyBRJvZh7H3XQFaKPQe5qfaKFu82J77iT57ZOQu3ofI2HN+9Z/p/NS+cv2tY3+em/0/Yfrobc6CMsiX+/CHUTwvWQXonc56VHcA2J+ZIhi8nC+r70Yy2++v7cT3JoMRP1dhs7Gt0bV46TFKOyp4dkOombqYEvLvQn55ESR9fx3U4EEG7rsXvyAsilET2A2iedAFkfqYxUcItSUQG8zvbVO1PttmbgvRVUYN7UWc/JnzDbZVhe8tHMGcuu+iFsM218DiRJrvw+Lpd/QVIe3NPxH0s6PXfm4Mt+TE0vU5jP4acx5xgmGqKcnJb/jdm6JWSQ44iUaPOW0ainHPfc87tc86t9nz2Redc4JxrLPztnHO3Oec2Oededc4tGotBCyGEEEIcat6OnPcDM7vkjZ3OuemWf210B3VfambzCv9db2a3/+5DFEIIIcTvRGD5nG8H67+jhLeU84IgeNo5N8vz0a1m9qdmdj/1XWlmPwyCIDCz551zdc65KUEQ7H7TjTizXGEka1+dge6s3w1S5CIiZ03VPBSPSq1C+L2sYKzpPQmOlcQWhI8HJiNkfN3834TtpfshtUz9NUK2ve+mkPQeSrpGcfvKY+ESS6/EWDhJZc8CyFaJZnLWNSL02z0L8kmyKr98DyUTnHwOpI16SjrZUoF9XVwJifQ/dp4btptewfLfa0N4vJ+cZD/ae1bYzk6GFlC9Escv1ocT0k7JDSs3YZ/6pmKfYgU5q+51hPwrKLll/wRMy1wZhdBJDbqiEbXinuqEa6dqJ8ZyfjXqAd5zEvbvpArUA/zNg2dgpYjy22kXI/C6/wZIjRf9GHXvln4KEuEQy7rmDOszM/ub5ZeH7XPnwEa66k6yNl2KOdNHLq5UH+bblvMhr14xDb9t/mrGA7Q1SLPx/SUu8aEaZzm/VWjR2dvD9lMTYL1LpzH3PjYb5+B7HTgWcxJIysjsvMS/rfkLsa1dnXBpTUr6nW+vr50+rO/P7vtU2O6eh7nWsArHMbOQ6nJG/GPpJJdhBoqYlXX7l69bV6I2YRU5/kiG5gS3XIuyfmjesqmPXLrdM7EfcUrkmUK+SmtaRbVBM37XZV8T5kPNDn+C1n191d7+TA5jKFU7LyjL9+9cNcX7eXQ67jn7Kdnm0L3NzCxVS3M2i+3s6IcEG6Ghcy08TqTJLjyW8DbRNXTcnTd6x1nZnD/2vcd4P7aApO/EHow3VU//UFWTO7mcJL/oUfSEcRQwoneinHNXmNmuIAhWOVd0E5lqZjvp7+ZC37CHKOfc9ZaPVlmspv6NHwshhBBilHAWyJ03BvzWD1HOuQoz+4qZvdf3safPe9aCILjDzO4wMyufMl1nVgghhBCHFSOJRM0xs2PNbCgKNc3MVjjnTrd85Ilj7dPMzG/VIXJRs1Rt/jkq3kE1hrj2UKkEZRRW7m6Dq+MzVz0RttvS+f7yKEKq7z3ntbD9Qh8kmH978NKwnaklRwW5V5qfRcg4cyHiyvHnEZ5PTUIo9zMfxFh6spBDKqL47oXnrQnbz/djPP9xL1yBwbq8g2eATDUtAwiJb6NtRnogA9605WNYB9Udc1djmc2/QWLKyImQY15+Gk4nm4Lxnng1pLJ1bXAILqyCRHfCOxGA3NmPaOOLNceamdmGOZDwol0YSzARMk60zO9q+bMHrg3b7JSxRdi/G+77DJaZinX+5UNIIGof8ktGyx+AI63/eoxhx48goQ18dHhY/tGXTxzWZ2ZFzrcnN2D+xI6lZai2myPpMkal6Fh+yJCj86PL/nfY3vynWP7/XsPJRME1VXlNqC+Hc5o27Ocpj/xR2P74aaif2Hc8pMVpccglN7zz6bA9Mw73KMPOPmZ9HJpJYjdJI6e2+Ra3IDF8TnSRVFc1CXOwqwvyICtQRUk9CU68y8klI35jobWdVuIDro9GiScdtbm2Wudx+ePqKP9lLkEJeKmfJbxMEz7YyQlJ6VWIMpLKggDLd0T8slLLcr8Ut6vUPZhU7SEzZLbaf91GtuIebexY2wYJseXd5Fqk8/H8MtyLopU4Ng/QvYtr4XEiTXbh8TW0/jp+bRdJczsXFMZWSzJcC+7dmYnoT+yk1zuaMcdTVNfQpfDdXAkpVBye/NYPUUEQvGZm4b+azrltZrY4CIJW59wDZvZ559xPzewMM+t8y/ehhBBCCDH2SM4bdd5OioOfmNkyMzvOOdfsnLvuTRZ/2My2mNkmM/uumf3hmywrhBBCCHHY4oJx8GS6ePHi4OWXXz7UwxBCCCEOGs655UEQLD4Y26qtPCY4c8Fn3nrBUeLRFV87aPt2KFHZFyGEEEKIETAuyr4IIYQQYgwZSrYpRpVx8RC1qWeHXfHM583MbN2zs8P+HI2Ok5sVOfWonWhHYK3+7D1hu6os707ZvALGwaGkcGbF9Y6mPAvHyK7zsL4rz0eSxcf/88ywnXkXMt+5l+AE4rE3nIOxVJbBKbP5FdTXC2j5CNdfqsasr5+Rr1U2mIYdLbUeVp00uQlr1mOFFfuoHtkS7PdVS1BDbunPsU+pU+BuKlsFG1OW6rlNPgumy7IotrtlFRyKuSS2GyO3YKRgiHvPJUjWOLcctd/+dcV52BA7NMlxNOOX5Hz5Ar679ISfh+3F34ZjbeLFzWG79UEc955ZGOPWL/xJ2J596zfDdnQaZUjdAHdR8sR8ctfXrvha2DfnG0h0yXCy2CKnVR0+YGdqhhyYPPdrNmGZrjn4btUO9L/2LdQ+POXBP/eOp2N7XWFgNMYMJa7divX1n4maaeXlmL+ZF+C4HJiEsSTa8N3XqQ7jgv/6a+9Y+luR1TJSRW6oHZhwm26Gc2rmd74xbB0uXco6BmI9GFe6gZxh5JaKdmOe8n2BHaCbv4SxHH/fV73bGuiHY8vR+rOURDXowzUaq8kf1xzN99wg1Z7sJfcqFags34V1XHDF8rA9PQnn5PfWUsLcLLbPSR83Xv0XYfvcJ77o3ae9nXDQ8T6tu+qrYXvOP+Xnf6aBJjlD+xevhTOWnYo5usFn6RhVbMYxTS6BAzT9JGoJds+jBJcNSDg8VAvPDIk0zciFZ2bbPof9zu3JZ139491Qo+5ffmrYPvE4pENc+9KssB3rowTRdD0lUGayOBGqOOwZFw9RQgghhBhblGxz9BkXD1H9g3FbtSUfHahppV8knFYmUiJnFLVzi1HBu7MPP+V3d+d/ddcdh19n6SxWnsnQL75n8Ou3fB9+tbWn8UumqgW/4H5vIfLj/PRHyOnUei2iFx00lj39iFZVU5ka/mXXP4hfXLHV+PU3UKjenqb8QBV7cAB6yqgsQ4mi4Yk27OvW3glhu3wf1vn1xfeE7X+8++Nhu++zGC9XY2/bgojEtBP2hu32XoowUE6auQ35X5FcruE7a1CW5X3HI4dXdwbnYzCL6bq3DxHL5heR1+Ybk5GnqXYrDsL9C7BP7//G58N2qg77wWTrMLZsG4XgpuIXdqYnaW+kosUfEUl0UOkPKgkS78D56JlJkaWZiHDmqNxGah/mzwVn4zi9tsKfn6rjQKW3fyj/Dc+7omhAM/Ytk8YYByIIyeRq8d15J+GX+c7HZ3o3Ob2+w9ufq8e+TirvDtvPdc/3LV6UC2iIymb0DTRSSY427FP/ZMrXxJErqroQoehBZhYiGZFdw8+1mVk85s8T1dCI658jKzUJrHPb47PCduqEfCSqLI71pShqNWkeIi/72xF9rl2CXFoRCneu6cE1cdHs9WH73bWvh+0Xuv0livY+NdXbX9ZDf/A0vwrNoV2tet1/XXHZq1Qt5lIZgp1Wczmy4rTsmhy2K/bi/A3S/dvRva52Le4RnQtxzpqw28WlXGr9EbOhCNS3p8Dw9OjO0zGWExCRrWjBeRpswBhTEyhPVYzm2NuImorDh3HxECWEEEKIMUaRqFFH7jwhhBBCjBuccw3OuceccxsL/x9WYNc5d75zbiX9N+Cc+0Dhsx8457bSZ6eM1VjHRyTKBRYtlCPpOo7C4/zQ/DYeoN0uvATN5U2GvtvhIG1UrEWoN0Uvxe4/iV7urcBGf70B5d1rm7DM7Xe9P2ynTyXJpgXbKhoL0UXdybWQ/Hg8VkslEAplKPhQpKjgOr/Qy/0DjVQKgV5KfeX1WWG7phoh5ltu/1TYztDUG9iOeVxUPoZk1z0kNVQswzHgl6Bf2Zs/T6sq8YJ3bgAreW43aqHkHsFLox2LEEIvPxtyQRDFPt35ImTB6hlY55n/ipeBc4jKFx2PItL0+4LWH99NL/VPGP4bZLBhWJeZmWUq/CH8VB1JTDSU7laS4UgOgkBq9vy9J2P9c/3bjexLePtDCYRVLZp2ReNqw7FOx7D/5Z348vanScIrcVfZ2DzR2x/dgzFuo/FES6geXI4lHBeVcUm0k4Q3CfuRY3NKeQm9uxElj5Jkquib6ZftBtP+ne3qxpmKJyAZ7V6LYxBLYjyVFfmXrKMkewd03ve04Npz/ZjXe9twfTy4n0rc0GSqrYWG9uR/oURK/0JIi7finWnLVPqviWyFt7uYQlmbnjn+41t2gGQ4mm9pDN06tzZhdWQA6ppD531NHZah7xaFBei67T3GP5m4lAsz9BI5S3hrb/z3sL3oaygd030qSYJ0zmJUyoqNIi5xqKJBweEUibrZzJ4IguAfnHM3F/7+Mi8QBMGTZnaKWf6hy/JJvh+lRb4UBMG9Yz1QRaKEEEIIMZ640szuKrTvsqLqtV4+bGb/HQRB31ssN+roIUoIIYQ40gksH4k6WP+ZNTrnXqb/rv8tRjtpqO5u4f/+MDb4qJn95A19f+uce9U5d6tzzh9yHAXGh5yXc5btyQ9lwnLKh0IR2IAko8CxBoFmNIVQZfvZCCdX1eUfTnu3Ie5bex5yN/VTGDy+kWQ1RIytvh7WlJxBCrz4mufD9rPfPiNsZ8oxsANLMJbyaoTQB3ZCc6s8Zz/G00b5WKjqe7qQNybShdPWT5XQ3SDlgOmnHCUd5DiikHxZzSD1Q6Zxp8EtFV2GY9OwAutvo32aMQ+OvObX4KY57xPIrfXwhhPCduUzeamqawkkkpPnwd316krIeVd/+jdh+5evQ76KUh6e/tmQ+ebOQM6o7fshF04/bVfY3v0k+tONfpkmvp+qsR/D4XpayFONPVciVO8ox9XAMdhmwyvYTroSxzcXx/6xFJAtx/r7yCkYPeC/lKMzer39qd78+nl+sTuvejW2n2rCeB3JYP2cwyvx1ln8SkmLnDeL8zFF/KfGcj4JlvLzDJ6Oa7XmfyCL9k7FMmW7yT3GDt8YSZfkPoy3slUY9O+u8va7OszJgV6ss2o2rq3uPbjOo8vzcl2GVLCghiRVOu887yp20f1yKva7uhzX9t4W3Mgu/MCqsP3sDjhcmfiCLn8/ORGDwC+PRQbz/W5Kyvt5jOZSfRU5mHsxyTMtOKY5mldZconyHHAkP3JuOiP5lucV513LTPS784byQLELjyW8FX95e9ie90P0c56odDW2kySXaIlDdyTS+mZlX5xzj5vZZM9HX/ltNuKcm2JmJ5rZUuq+xcz2mFnczO6wvBT4teHf/t0ZHw9RQgghhBhbxlHG8iAILiz1mXNur3NuShAEuwsPSftKLWtm15jZL4MgCJ+Ih6JYZjbonPu+mfkzyI4CkvOEEEIIMZ54wMw+UWh/wszuf5Nlr7U3SHmFBy9zzjnLv0+1egzGaGbjJRKVdRYtSFRtp5Vy5711DLR8J3Ynvh3SQf+efDi9biuWbW9HFLGaUvJ3LkQ8vawTz5jdryExZW4JHDyP/ALlUtKXI5weew0haR5LOoHQfu02bLdnPxwpNcg3aL1nIuQdFCTPsi5yEFL4n915OVJOBpqo/EovJVTcBqmjbzqOe/WzkPB6T8W+5tYh5J7cif3Y0YNj2bAO61/aBWdLJdTK0MHGZT1e7ZyF7W/Bfjy0e0nYJoXCuqi8g+tCrH7bS5DqMiR1tt+H/v5T/FID40iqshQlMaVEeZH+4RKPK2H6GpgG2aCsDfO070LMmfR2nI84zT1eJ5cHqlqPc9A3xf8Tk8tdMIkhp2fELz9m6GsV2ynBZpxkX7pUHSUEZRmDiXf5r2G+tGOkPg5O8O8Tz+EhUgswTx0dRz6+g/sxf7NxconRespojPEDb6OUTLf/d2j5ZpKnKE+nG6A/5lNC1yGnHjt29/sTNPLxYnk+uwbO2HZSKyvIRfnCOkjijvOHfgjN3r3+OdPH8q1HyjYzyxTKVAWDfvkzm0J/Szvp1CSxFb2tQdtkN5/V4xou24abXWYi+rOdOAh8/SX24CJK7PQnBR0q5cKJNNmFxxLexo9D2jvu++hn2XCgqUTZMlGKfzCze5xz15nZDjO72szMObfYzD4bBMGnC3/PMrPpZvbrN3z/R865Jstf3ivN7LNjNdDx8RAlhBBCiDHlcCn7EgRBm5m9x9P/spl9mv7eZmbDUuwHQXDBWI6PkZwnhBBCCDECxkckKmKWqyjEallGiVFCR2oXuaJYdanHM2GxHJJf6MA7oIskyH3VNZcS3JG7I02biXfQ+rYiDs517GwrwuCpeqzHNxYzs47jaTxU044TU1orQtVDbiVOAsouFXZuDU5iWZQkvDglgyMpguuRZSnMHxxAuJvllVgPli/rwTrbFmOf2OHWcTLGE+3O90cHyC1Fx5cTrpZ1knutiZw0GZbY0B0dpH0d4HpqWMZRYs/EPr/sMDgD2iHLhYON2L+gbPivuqL5QMTIPcdO0/QOkk5oLvfP9EuOnGgxF2f51i99ce04ZmgMXCuOZYaeuTgHsU5/gsQof4Fdsv3+bfZP8zuhyujYsCEv8J+aoppk4Tab/bXt+PhGaH25pP885WaQfL0f66yY3u1bPJSv3khXE2mwGf9v1TKqm5iqLyxPi2bnkDTVQ0lOGzDG1Da8NsDHPULTp3+K/xWFVJNfe451+w88Jyt15j/HkcI1lyvxT0tkoIRMTddSQMcgKHLYoc0JWvmacy04Z44ceXyv4XtzRbN/X4dcdlwLryiRJrnwWMJb/0lIezc0nxW29/ZDal27e5J3mweFwyQSdTihSJQQQgghxAgYH5EoIYQQQowdgZnlFIkabcbHQ1TWLFqQlliSMq6RxTEzTsJJ0l62jpLBdUGGGpJ7gip8niK3i9WSzJD2S2/FlgpK1lZDsl07OZSaSoyFVbYqxLNTtK2ghsZDLpehY8ROnUFy5+T4bNLxcnFOkoeFYhT+HyBZYqCRZFSSV7PVJAuQQzBKcl5mBsaeIYkwWo3+oTp57ITi45KCEbJILnFlJJ1QOxjA8Y2Sg2+wHMsMct00liVK3FPYIZSg8zpAMlvZnjIbRgm3m6P5k6FxJdpxfgdn8eBpHnIdv2ocqHSUpIgScl7Obz4KHXRFiS45/2SStBaSnViCYWmzWA72H4NIXwmZiGVRdoCVqmvo6eb5k66j40uS8sAU7Gy8lS4WWl86BpmIpcveNn/huEi/P5if5fPB46X+XBlJ4kOSKc27tOGE8L2ov5vGSMe6lKOTSVMCz1JztRRF7rhSdQ0L5yHrkbrNzIIsyXOUmDZC851r3rHM5+i+l03494MTxvJ1W/TvCl1DqTr/+XMFmZulY66Fxw5Uvp+whPedacvC9r91TA/b3Wmcv83erYvDifHxECWEEEKIMeSwKkB82KB3ooQQQgghRsC4iETF+s0mvJZ/Qo5/HHXYsuwqexsZyiKkzfRQDbfkgXx/32w8MyZb0e6rQJi2rA617XJdCOHHUfLK+qZiO+xwy1ANp2OmtXvHEu+i2mcz/aHn/qRf9hiS8RybnEqF2Cn839gIZ1HrtoawXdlMbr7jIFPF2H1ILrWaWkpouLo+bFfswyD65mM/qvag3VNF0ldFXmtIbiA3JR2XgUa/yzJF8m7NnI6w3dkB2SPZivX0s8uvjdw807FPA5P88yq6F+ssciV145LxSSalasilWd6IY5kMuZUam1CzrHMl7IQxlktPwTnI7ieZuJ6kQCJSIq9opHAsi+Rlun5mTm0N2/vWIw1LnMqqdTRieXZoZqv8xyC5z/+bLUU16ook5kr/r2aWdYbIsFxDTXayTpqOrLpdO1DPNELnMV1NjtFuqoPmSlgFS+nBNEaWxAOWaafgXuMKyTmjNMczVWhHyQ2bJUne1fJ1i+SVUUyTIsmqSL4d9N/+eb4zRW7UErfjUm7FcN0kpbFLNlvJzmmMK0uuuswEcoy20zKTcQwCkr5z5KpNU+LdeDkOgkv56zkmClMliLEMj8+5Fh4n0mQXHkt4n6tDbdD3V60L26gSepBQJGrUUSRKCCGEEGIEjItIlBBCCCHGGEWiRp1x8RAVyQRW3pYPt06qgm7Wl4Fckcoi9BuwzEdx5QjF8fsoRB9N5fsd10AjGYMTwMUoqSeHktn1xa6SMpbzyKlXm0CovpcTFFIGz6LxUJidx5OrY+0uNmzspVyLjpIolkVxMCKc4LKfktdFMUhOwBghF1yM1pPj48dZSclNyPvESSKDgpwXyZA7h9qcMJPlKJaMYlF2EPJYaJvk8imqeUf92fISCQdJVmJ5o2jeZD2aRqaEzsHniY1bNJbaJOZMD22/rJeXJ3cXG5SifhmlVMLKIRkqiPgTZsZJ42K5i89TkYuKjaxR/406NuDtDmspDqNEnJwTJw7BjkCe+zyWaKSE1MQqFU8H2qlIibE73xywYrcZz08+3kmSodJBXisqqkdI0l7RNd/L9wfaJl0r0UF2vtE6SVqMlSohWeK4F53XEtM8GJKqS1wHkQQlqyU5jyVuVkijNMZMiSTLRfe92FuPke91uRJy7NA85/uGS5SY79TmRJrswmMJb0YMCVLF4c+4eIgSQgghxBiiPFFjgt6JEkIIIYQYAS4YBxqpc26/mW03s0Yza32LxY8Ejpb9NNO+HqkcLft6tOynmfb1UDAzCIKmg7Gh2sSkYMkxHzsYmzIzs0e23bo8CILFB22Dh4hxIecNTSLn3MtHw0E/WvbTTPt6pHK07OvRsp9m2lchRoLkPCGEEEKIETAuIlFCCCGEGGPGwes7RxrjLRJ1x6EewEHiaNlPM+3rkcrRsq9Hy36aaV+F+K0ZFy+WCyGEEGLsqI1PCpZMvvagbe+Rnf98VLxYPt4iUUIIIYQQhwV6J0oIIYQ4GpDyNOooEiWEEEIIMQIUiRJCCCGOBhSJGnUUiRJCCCGEGAGKRAkhhBBHPIEiUWOAIlFCCCGEECNAkSghhBDiSCcws1zuUI/iiEORKCGEEEKIEaBIlBBCCHE0oHeiRh1FooQQQgghRoAiUUIIIcTRgCJRo44iUUIIIYQQI0APUUIIIYQQI0BynhBCCHHEE5jlJOeNNopECSGEEEKMAEWihBBCiCOdwCwIlGxztFEkSgghhBBiBCgSJYQQQhwN6J2oUUeRKCGEEEKIEaBIlBBCCHE0oGSbo44iUUIIIYQQI0CRKCGEEOJIJwjMcnLnjTaKRAkhhBBCjABFooQQQoijAb0TNeooEiWEEEIIMQIUiRJCCCGOAgK9EzXqKBIlhBBCCDECFIkSQgghjngCvRM1BigSJYQQQggxAvQQJYQQQggxAiTnCSGEEEc6gakA8RigSJQQQgghxAhQJEoIIYQ4GgiU4mC0USRKCCGEEGIEKBIlhBBCHOEEZhbonahRR5EoIYQQQogRoEiUEEIIcaQTBHonagxQJEoIIYQQYgQoEiWEEEIcBeidqNFHkSghhBBCjBucc1c759Y453LOucVvstwlzrn1zrlNzrmbqf9Y59wLzrmNzrmfOefiYzVWPUQJIYQQRwNB7uD997ux2sw+aGZPl1rAORc1s38zs0vN7Hgzu9Y5d3zh46+b2a1BEMwzswNmdt3vOqBS6CFKCCGEEOOGIAjWBUGw/i0WO93MNgVBsCUIgpSZ/dTMrnTOOTO7wMzuLSx3l5l9YKzGqneihBBCiCOcbjuw9PHg3saDuMmkc+5l+vuOIAjuGMX1TzWznfR3s5mdYWYTzKwjCIIM9U8dxe0WoYcoIYQQ4ggnCIJLDvUYGOfc42Y22fPRV4IguP/trMLTF7xJ/5ighyghhBBCHFSCILjwd1xFs5lNp7+nmVmLmbWaWZ1zLlaIRg31jwl6J0oIIYQQhxsvmdm8ghMvbmYfNbMHgiAIzOxJM/twYblPmNnbiWyNCD1ECSGEEGLc4Jy7yjnXbGZnmdlDzrmlhf5jnHMPm5kVokyfN7OlZrbOzO4JgmBNYRVfNrObnHObLP+O1J1jNtb8Q5sQQgghhPhtUCRKCCGEEGIE6CFKCCGEEGIE6CFKCCGEEGIE6CFKCCGEEGIE6CFKCCGEEGIE6CFKCCGEEGIE6CFKCCGEEGIE/H+neKe7vhXvmAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1607e3c8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Look at feature correlations..\n",
    "X_densDF = pd.DataFrame(X_dens, columns=dens_featurizer.feature_labels())\n",
    "correlations = X_densDF.corr()  # Returns NaN if values do not vary at all...\n",
    "\n",
    "# find columns that do not vary\n",
    "invariants =[ col for col in correlations.columns if correlations[col].isnull().all() ]\n",
    "print(invariants)\n",
    "\n",
    "# plot covariance matrix\n",
    "fig, ax = plt.subplots(figsize=(10,10))\n",
    "cax = ax.matshow(correlations, vmin=-1, vmax=1)\n",
    "fig.colorbar(cax, label='correlation')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------correlated features to drop\n",
      "['3-norm', '5-norm', '7-norm', '10-norm', 'minimum AtomicWeight', 'maximum AtomicWeight', 'range AtomicWeight', 'mean AtomicWeight', 'avg_dev AtomicWeight', 'mode AtomicWeight', 'minimum Row', 'mean Row', 'mode Row', 'maximum NpValence', 'range NpValence', 'range NdValence', 'range NfValence', 'avg_dev NfValence', 'mode NfValence', 'range NValence', 'range NsUnfilled', 'range NpUnfilled', 'mode NpUnfilled', 'range NdUnfilled', 'range NfUnfilled', 'avg_dev NfUnfilled', 'mode NfUnfilled', 'range NUnfilled', 'range GSvolume_pa', 'range GSbandgap', 'avg_dev GSbandgap', 'range GSmagmom', 'mode GSmagmom', 'avg s valence electrons', 'avg p valence electrons', 'avg d valence electrons', 'avg f valence electrons', 'max ionic char', 'avg ionic char']\n",
      "----------------------------invariants:\n",
      "['maximum NsValence', 'minimum NpValence', 'minimum NfValence', 'minimum NsUnfilled', 'minimum NpUnfilled', 'minimum NfUnfilled', 'minimum GSbandgap', 'mode GSbandgap', 'compound possible']\n"
     ]
    }
   ],
   "source": [
    "# Select upper triangle of correlation matrix\n",
    "upper = correlations.where(np.triu(np.ones(correlations.shape),k=1).astype(np.bool))\n",
    "\n",
    "# Find index of feature columns with correlation greater than 0.95\n",
    "to_drop = [col for col in upper.columns if any(upper[col] > 0.95)]\n",
    "print('----------------------------correlated features to drop')\n",
    "print(to_drop)\n",
    "print('----------------------------invariants:')\n",
    "print(invariants)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "39"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(to_drop)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "avg_dev GSbandgap           0.997949\n",
       "maximum GSmagmom            0.510955\n",
       "range GSmagmom              0.510955\n",
       "mean GSmagmom               0.741110\n",
       "avg_dev GSmagmom            0.723227\n",
       "mode GSmagmom               0.796498\n",
       "range SpaceGroupNumber      0.628599\n",
       "avg_dev SpaceGroupNumber    0.552631\n",
       "mode SpaceGroupNumber       0.604314\n",
       "avg p valence electrons     0.531445\n",
       "Name: mean GSbandgap, dtype: float64"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feat = 'mean GSbandgap'\n",
    "upper.loc[feat][upper.loc[feat]>0.5]\n",
    "#returns other features fitting the covariance condition.  (queries rows of cov matrix)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 9min 53s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Try recursive feature elimination\n",
    "RF = RandomForestRegressor(n_estimators=100, n_jobs=1)\n",
    "\n",
    "rfecv = RFECV(estimator=RF, step=2, cv=KFold(5),\n",
    "              scoring=r2_scorer)\n",
    "if True: # Takes forever\n",
    "    rfecv.fit(X_dens, densityData['density']) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimal number of features : 30\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x17df2358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Optimal number of features : %d\" % rfecv.n_features_)\n",
    "\n",
    "# Plot number of features VS. cross-validation scores\n",
    "plt.figure()\n",
    "plt.xlabel(\"Number of features selected\")\n",
    "plt.ylabel(\"Cross validation score (nb of correct classifications)\")\n",
    "plt.plot(range(1, len(rfecv.grid_scores_) + 1), rfecv.grid_scores_)\n",
    "plt.ylim([0.75, 0.85])\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(), 'results', 'finalFigs', \n",
    "                             'densityMG-v1_rfecv.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combine data: original feature order, feature rankings by rfecv\n",
    "feats = pd.DataFrame(dens_featurizer.feature_labels(), columns=['feat_name'])\n",
    "feats['orig_order'] = range(len(feats))\n",
    "feats['rfecv_ranking'] = rfecv.ranking_\n",
    "\n",
    "# Mark features to drop\n",
    "feats['to_remove'] = 0\n",
    "feats['to_remove'] = [1 if feats['feat_name'][x] in to_drop+invariants else 0 for x in range(len(feats))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "# take all features ok'd (not dropped due to covariance or invariance\n",
    "feats_subset = feats[feats['to_remove']==0]\n",
    "\n",
    "feats_subset.sort_values(by='rfecv_ranking',inplace=True)\n",
    "feats_subset.reset_index(drop=True, inplace=True)\n",
    "\n",
    "# Create new feature matrix and list of features (differently ordered from original featurizer)\n",
    "X_subset = X_notTrouble[:,feats_subset['orig_order'][:rfecv.n_features_].tolist()]\n",
    "reduced_feats = feats_subset['feat_name'][:rfecv.n_features_]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### After feature reduction, try re-evaluating model performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "densModel_elimFeats = Pipeline([('impute',Imputer()), \n",
    "                          ('model', RandomForestRegressor(n_estimators=100, n_jobs=1, max_features=12))])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Split #0\n",
      "Split #1\n",
      "Split #2\n",
      "Split #3\n",
      "Split #4\n"
     ]
    }
   ],
   "source": [
    "#take predicted values from each cross validation set to ensure training not performed on own set.  Take 200 training samples\n",
    "# Probably could have used cross_val_predict\n",
    "impDF = pd.DataFrame()\n",
    "\n",
    "kf = KFold(5)\n",
    "rep=0\n",
    "notTroubleDF['density_predict_rfecv'] = np.nan\n",
    "for train_index, test_index in kf.split(notTroubleDF['comp']):\n",
    "    print('Split #{}'.format(rep))\n",
    "    densModel_elimFeats.fit(X_subset[train_index,:], notTroubleDF['density'][train_index])\n",
    "    \n",
    "    y_densPredict = densModel_elimFeats.predict(X_subset[test_index,:])\n",
    "    notTroubleDF['density_predict_rfecv'][test_index] = y_densPredict\n",
    "    \n",
    "    featureImp = pd.DataFrame(densModel_elimFeats.steps[1][1].feature_importances_,\n",
    "                          index=reduced_feats,\n",
    "                          columns=['importance']).sort_values('importance',ascending=False)\n",
    "    impDF = impDF.append(featureImp)\n",
    "    #print(train_index, test_index)\n",
    "    rep+=1\n",
    "#y_densPredict = "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "top features over 5 training splits\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>importance</th>\n",
       "      <th>totalImp</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>feat_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>mode NValence</th>\n",
       "      <td>0.195739</td>\n",
       "      <td>0.940237</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mode Number</th>\n",
       "      <td>0.067177</td>\n",
       "      <td>0.916297</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean NValence</th>\n",
       "      <td>0.171629</td>\n",
       "      <td>0.663548</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean Electronegativity</th>\n",
       "      <td>0.093666</td>\n",
       "      <td>0.361777</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean Number</th>\n",
       "      <td>0.031258</td>\n",
       "      <td>0.275884</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        importance  totalImp\n",
       "feat_name                                   \n",
       "mode NValence             0.195739  0.940237\n",
       "mode Number               0.067177  0.916297\n",
       "mean NValence             0.171629  0.663548\n",
       "mean Electronegativity    0.093666  0.361777\n",
       "mean Number               0.031258  0.275884"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "impDF['totalImp'] = impDF.groupby(level=0)['importance'].transform('sum')\n",
    "impDFtop = impDF.drop_duplicates(subset=['totalImp']).sort_values('totalImp',ascending=False)\n",
    "print('top features over 5 training splits')\n",
    "impDFtop.to_csv(os.path.join(os.getcwd(),'results','finalFigs',\n",
    "                             'densityMG-v2_featImportance.csv'))\n",
    "impDFtop.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x17cc1b00>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot Comparison, report r2\n",
    "r2rfecv = r2_score(notTroubleDF['density'], notTroubleDF['density_predict_rfecv'])\n",
    "\n",
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(notTroubleDF['density'], notTroubleDF['density_predict_rfecv'], edgecolors='k')\n",
    "plt.plot([0,30], [0,30], 'r-')\n",
    "plt.xlabel('density [g/cm$^3$]')\n",
    "plt.ylabel('predicted density (RF)')\n",
    "\n",
    "\n",
    "plt.text(7,20, 'R$^2$ score: {:.3f}'.format(r2rfecv))\n",
    "plt.xlim([0,25])\n",
    "plt.ylim([0,25])\n",
    "plt.title('Gen. 2')\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(),'results','finalFigs',\n",
    "                             'densityMG-v2_Scatter.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x16091978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(notTroubleDF['density'], notTroubleDF['density_vegards'], edgecolors='k')\n",
    "plt.plot([0,30], [0,30], 'r-')\n",
    "plt.xlabel('density [g/cm$^3$]')\n",
    "plt.ylabel('predicted density (Vegard\\'s)')\n",
    "plt.title('Gen. 2')\n",
    "\n",
    "r2V = r2_score(notTroubleDF['density'], notTroubleDF['density_vegards'])\n",
    "plt.text(7,20, 'R$^2$ score: {:.3f}'.format(r2V))\n",
    "\n",
    "plt.xlim([0,25])\n",
    "plt.ylim([0,25])\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(),'results','finalFigs',\n",
    "                             'densityMG-v2_Vegards_Scatter.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train model on difference from Vegards Predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mean difference: 0.161, variance 0.707\n",
      "max difference: 8.760, min difference -2.111\n"
     ]
    }
   ],
   "source": [
    "print('mean difference: {:.3f}, variance {:.3f}'.format(densityData['vegards_diff'].mean(), densityData['vegards_diff'].var()))\n",
    "print('max difference: {:.3f}, min difference {:.3f}'.format(densityData['vegards_diff'].max(), densityData['vegards_diff'].min()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "densDiffModel = Pipeline([('impute',Imputer()), \n",
    "                          ('model', RandomForestRegressor(n_estimators=100, n_jobs=1, max_features=12))])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 374 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Pipeline(memory=None,\n",
       "     steps=[('impute', Imputer(axis=0, copy=True, missing_values='NaN', strategy='mean', verbose=0)), ('model', RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=None,\n",
       "           max_features=12, max_leaf_nodes=None, min_impurity_decrease=0.0,\n",
       "           min_impurity_split=None, min_samples_leaf=1,\n",
       "           min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
       "           n_estimators=100, n_jobs=1, oob_score=False, random_state=None,\n",
       "           verbose=0, warm_start=False))])"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "densDiffModel.fit(X_dens, densityData['vegards_diff'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Make sure is true\n",
    "\n",
    "See if there's any learning at all \n",
    "\n",
    "Everything else must be process dependant, experimental, bad data, etc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train on difference between density and vegards, see what features float up\n",
    "Not including outliers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>comp</th>\n",
       "      <th>density</th>\n",
       "      <th>density_predict</th>\n",
       "      <th>density_vegards</th>\n",
       "      <th>vegards_diff</th>\n",
       "      <th>density_predict_rfecv</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>(Ce, Al)</td>\n",
       "      <td>3.462</td>\n",
       "      <td>4.057876</td>\n",
       "      <td>3.1070</td>\n",
       "      <td>0.355</td>\n",
       "      <td>3.936705</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(Al, Fe, B)</td>\n",
       "      <td>3.539</td>\n",
       "      <td>5.809120</td>\n",
       "      <td>3.6620</td>\n",
       "      <td>-0.123</td>\n",
       "      <td>4.743269</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>(Zr, Al, Ni)</td>\n",
       "      <td>3.384</td>\n",
       "      <td>3.895027</td>\n",
       "      <td>3.5105</td>\n",
       "      <td>-0.1265</td>\n",
       "      <td>4.059783</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>(Zr, Al, Ni)</td>\n",
       "      <td>3.517</td>\n",
       "      <td>3.999587</td>\n",
       "      <td>3.6965</td>\n",
       "      <td>-0.1795</td>\n",
       "      <td>4.178260</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>(Hf, Al, Ni)</td>\n",
       "      <td>2.835</td>\n",
       "      <td>5.478570</td>\n",
       "      <td>3.8505</td>\n",
       "      <td>-1.0155</td>\n",
       "      <td>4.376186</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           comp density  density_predict  density_vegards vegards_diff  \\\n",
       "0      (Ce, Al)   3.462         4.057876           3.1070        0.355   \n",
       "1   (Al, Fe, B)   3.539         5.809120           3.6620       -0.123   \n",
       "2  (Zr, Al, Ni)   3.384         3.895027           3.5105      -0.1265   \n",
       "3  (Zr, Al, Ni)   3.517         3.999587           3.6965      -0.1795   \n",
       "4  (Hf, Al, Ni)   2.835         5.478570           3.8505      -1.0155   \n",
       "\n",
       "   density_predict_rfecv  \n",
       "0               3.936705  \n",
       "1               4.743269  \n",
       "2               4.059783  \n",
       "3               4.178260  \n",
       "4               4.376186  "
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "notTroubleDF.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Need to RFECV with this dataset independently"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 10min 3s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Try recursive feature elimination\n",
    "RF = RandomForestRegressor(n_estimators=100, n_jobs=1)\n",
    "\n",
    "rfecv_diff = RFECV(estimator=RF, step=2, cv=KFold(5),\n",
    "              scoring=r2_scorer)\n",
    "if True: # Takes forever\n",
    "    rfecv_diff.fit(X_dens, densityData['vegards_diff']) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimal number of features : 74\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x193a26d8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Optimal number of features : %d\" % rfecv_diff.n_features_)\n",
    "\n",
    "# Plot number of features VS. cross-validation scores\n",
    "plt.figure()\n",
    "plt.xlabel(\"Number of features selected\")\n",
    "plt.ylabel(\"Cross validation score (nb of correct classifications)\")\n",
    "plt.plot(range(1, len(rfecv_diff.grid_scores_) + 1), rfecv_diff.grid_scores_)\n",
    "#plt.ylim([0.75, 0.85])\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(), 'results', 'finalFigs', \n",
    "                             'densityMG-v3_rfecv.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combine data: original feature order, feature rankings by rfecv\n",
    "feats_diff = pd.DataFrame(dens_featurizer.feature_labels(), columns=['feat_name'])\n",
    "feats_diff['orig_order'] = range(len(feats_diff))\n",
    "feats_diff['rfecv_ranking'] = rfecv_diff.ranking_\n",
    "\n",
    "# Mark features to drop\n",
    "feats_diff['to_remove'] = 0\n",
    "feats_diff['to_remove'] = [1 if feats_diff['feat_name'][x] in to_drop+invariants else 0 for x in range(len(feats_diff))]\n",
    "\n",
    "# take all features ok'd (not dropped due to covariance or invariance\n",
    "feats_diff_subset = feats_diff[feats_diff['to_remove']==0]\n",
    "\n",
    "feats_diff_subset.sort_values(by='rfecv_ranking',inplace=True)\n",
    "feats_diff_subset.reset_index(drop=True, inplace=True)\n",
    "\n",
    "# Create new feature matrix and list of features (differently ordered from original featurizer)\n",
    "X_diff_subset = X_notTrouble[:,feats_diff_subset['orig_order'][:rfecv_diff.n_features_].tolist()]\n",
    "reduced_feats_diff = feats_diff_subset['feat_name'][:rfecv_diff.n_features_]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Split #0, training on 555\n",
      "                        importance\n",
      "feat_name                         \n",
      "minimum CovalentRadius    0.090222\n",
      "mean NpValence            0.052912\n",
      "avg_dev Column            0.043513\n",
      "mean CovalentRadius       0.039461\n",
      "mean GSbandgap            0.034925\n",
      "Split #1, training on 555\n",
      "                        importance\n",
      "feat_name                         \n",
      "minimum CovalentRadius    0.055082\n",
      "avg_dev NpValence         0.043464\n",
      "avg_dev Column            0.043173\n",
      "minimum GSvolume_pa       0.041328\n",
      "mode MeltingT             0.036597\n",
      "Split #2, training on 555\n",
      "                        importance\n",
      "feat_name                         \n",
      "avg_dev Column            0.106061\n",
      "minimum CovalentRadius    0.091468\n",
      "mean GSbandgap            0.056466\n",
      "mean Electronegativity    0.037023\n",
      "mean CovalentRadius       0.034887\n",
      "Split #3, training on 555\n",
      "                           importance\n",
      "feat_name                            \n",
      "minimum CovalentRadius       0.063427\n",
      "mean SpaceGroupNumber        0.047460\n",
      "avg_dev Row                  0.038823\n",
      "avg_dev Column               0.036288\n",
      "avg_dev Electronegativity    0.035690\n",
      "Split #4, training on 556\n",
      "                        importance\n",
      "feat_name                         \n",
      "minimum GSvolume_pa       0.066839\n",
      "mode MeltingT             0.061686\n",
      "minimum CovalentRadius    0.059429\n",
      "avg_dev GSmagmom          0.058920\n",
      "mean GSbandgap            0.037199\n"
     ]
    }
   ],
   "source": [
    "# Train on difference\n",
    "diffFeatTop5 = pd.DataFrame()\n",
    "\n",
    "# take predicted values from each cross validation set to ensure training not performed on own set.  Take 200 training samples\n",
    "# Probably could have used cross_val_predict\n",
    "kf = KFold(5)\n",
    "rep=0\n",
    "notTroubleDF['diff_predict'] = np.nan\n",
    "for train_index, test_index in kf.split(notTroubleDF['comp']):\n",
    "    print('Split #{}, training on {}'.format(rep, len(train_index)))\n",
    "    density_model.fit(X_diff_subset[train_index,:], notTroubleDF['vegards_diff'][train_index])\n",
    "    \n",
    "    y_diffPredict = density_model.predict(X_diff_subset[test_index,:])\n",
    "    notTroubleDF['diff_predict'][test_index] = y_diffPredict\n",
    "    \n",
    "    \n",
    "    featureImp = pd.DataFrame(density_model.steps[1][1].feature_importances_,\n",
    "                          index=reduced_feats_diff,\n",
    "                          columns=['importance']).sort_values('importance',ascending=False)\n",
    "    print(featureImp.head())\n",
    "    \n",
    "    diffFeatTop5 = diffFeatTop5.append(featureImp)\n",
    "\n",
    "    \n",
    "    #print(train_index, test_index)\n",
    "    rep+=1\n",
    "#y_densPredict = "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "top features over 5 training splits\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>importance</th>\n",
       "      <th>totalImp</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>feat_name</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>minimum CovalentRadius</th>\n",
       "      <td>0.090222</td>\n",
       "      <td>0.359628</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>avg_dev Column</th>\n",
       "      <td>0.043513</td>\n",
       "      <td>0.257590</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean GSbandgap</th>\n",
       "      <td>0.034925</td>\n",
       "      <td>0.190607</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>avg_dev GSmagmom</th>\n",
       "      <td>0.033573</td>\n",
       "      <td>0.161952</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>minimum GSvolume_pa</th>\n",
       "      <td>0.014008</td>\n",
       "      <td>0.159870</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        importance  totalImp\n",
       "feat_name                                   \n",
       "minimum CovalentRadius    0.090222  0.359628\n",
       "avg_dev Column            0.043513  0.257590\n",
       "mean GSbandgap            0.034925  0.190607\n",
       "avg_dev GSmagmom          0.033573  0.161952\n",
       "minimum GSvolume_pa       0.014008  0.159870"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "diffFeatTop5['totalImp'] = diffFeatTop5.groupby(level=0)['importance'].transform('sum')\n",
    "diffFeatTop = diffFeatTop5.drop_duplicates(subset=['totalImp']).sort_values('totalImp',ascending=False)\n",
    "print('top features over 5 training splits')\n",
    "diffFeatTop.to_csv(os.path.join(os.getcwd(),'results', 'finalFigs',\n",
    "                                'densityMG-v3_featImportance_190319_4.csv'))\n",
    "diffFeatTop.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# plot Comparison, report r2\n",
    "r2noOutDiff = r2_score(notTroubleDF['vegards_diff'], notTroubleDF['diff_predict'])\n",
    "\n",
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(notTroubleDF['vegards_diff'], notTroubleDF['diff_predict'], edgecolors='k')\n",
    "plt.plot([-5,5], [-5,5], 'r-')\n",
    "plt.xlabel('(density - vegards prediction) [g/cm$^3$]')\n",
    "plt.ylabel('predicted difference')\n",
    "\n",
    "\n",
    "plt.text(-1,4, 'R$^2$ score: {:.3f}'.format(r2noOutDiff))\n",
    "plt.xlim([-5,5])\n",
    "plt.ylim([-5,5])\n",
    "plt.title('Pre-Gen. 3')\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(),'results','finalFigs',\n",
    "                             'densityMG-v3_differenceScatter.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_sizes, train_scores, valid_scores = learning_curve(density_model, X_subset, notTroubleDF['vegards_diff'], \n",
    "                                                                    cv=ShuffleSplit(), scoring=r2_scorer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure()\n",
    "plt.title('Default feats, deviation from Vegards')\n",
    "plt.xlabel(\"Training examples: density data, SLAC\")\n",
    "plt.ylabel(\"$r^2$\")\n",
    "train_scores_mean = np.mean(train_scores, axis=1)\n",
    "train_scores_std = np.std(train_scores, axis=1)\n",
    "test_scores_mean = np.mean(valid_scores, axis=1)\n",
    "test_scores_std = np.std(valid_scores, axis=1)\n",
    "plt.grid()\n",
    "\n",
    "plt.fill_between(train_sizes, train_scores_mean - train_scores_std,\n",
    "                 train_scores_mean + train_scores_std, alpha=0.1,\n",
    "                 color=\"r\")\n",
    "plt.fill_between(train_sizes, test_scores_mean - test_scores_std,\n",
    "                 test_scores_mean + test_scores_std, alpha=0.1, color=\"g\")\n",
    "plt.plot(train_sizes, train_scores_mean, 'o-', color=\"r\",\n",
    "         label=\"Training score\")\n",
    "plt.plot(train_sizes, test_scores_mean, 'o-', color=\"g\",\n",
    "         label=\"Cross-validation score\")\n",
    "\n",
    "plt.legend(loc=4)\n",
    "#plt.ylim([0.4, 1])\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(), 'results', 'finalFigs',\n",
    "                             'densityMG-v2.0_LearningCurve.png'))\n",
    "    print('saved figure')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Look at mean absolute err with this new data set:\n",
    "\n",
    "mae = mean_absolute_error(notTroubleDF['density'], notTroubleDF['density_predict'])\n",
    "mae_vegards = mean_absolute_error(notTroubleDF['density'], notTroubleDF['density_vegards'])\n",
    "print('MAE for random forest: {:.4f} vs vegards law {:.4f}'.format(mae, mae_vegards))\n",
    "mae_diff = mean_absolute_error(notTroubleDF['vegards_diff'], notTroubleDF['diff_predict'])\n",
    "print(mae_diff)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Assess performance of using difference predictions instead of full model "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "notTroubleDF['vPlusDiff'] = notTroubleDF['density_vegards'] + notTroubleDF['diff_predict']\n",
    "r2Comp = r2_score(notTroubleDF['density'], notTroubleDF['vPlusDiff'])\n",
    "print(r2Comp)\n",
    "\n",
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(notTroubleDF['vPlusDiff'], notTroubleDF['density'], edgecolors='k')\n",
    "plt.plot([0,25], [0,25], 'r-')\n",
    "plt.xlabel('density [g/cm$^3$]')\n",
    "plt.ylabel('predicted density (Vegard\\'s + difference)')\n",
    "\n",
    "\n",
    "plt.text(10,20, 'R$^2$ score: {:.3f}'.format(r2Comp))\n",
    "plt.xlim([0,25])\n",
    "plt.ylim([0,25])\n",
    "plt.title('Gen. 3')\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(),'results','finalFigs',\n",
    "                             'densityMG-v3_Scatter.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Try to predict Vegard's law directly"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Train on difference\n",
    "vegardFeats = pd.DataFrame()\n",
    "\n",
    "# take predicted values from each cross validation set to ensure training not performed on own set.  Take 200 training samples\n",
    "# Probably could have used cross_val_predict\n",
    "kf = KFold(5)\n",
    "rep=0\n",
    "notTroubleDF['vegards_predict'] = np.nan\n",
    "for train_index, test_index in kf.split(notTroubleDF['comp']):\n",
    "    print('Split #{}, training on {}'.format(rep, len(train_index)))\n",
    "    density_model.fit(X_subset[train_index,:], notTroubleDF['density_vegards'][train_index])\n",
    "    \n",
    "    y_pred = density_model.predict(X_subset[test_index,:])\n",
    "    notTroubleDF['vegards_predict'][test_index] = y_pred\n",
    "    \n",
    "    \n",
    "    featureImp = pd.DataFrame(density_model.steps[1][1].feature_importances_,\n",
    "                          index=reduced_feats,\n",
    "                          columns=['importance']).sort_values('importance',ascending=False)\n",
    "    #print(featureImp.head())\n",
    "    \n",
    "    vegardFeats = vegardFeats.append(featureImp)\n",
    "\n",
    "    rep+=1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Sort and compile importances\n",
    "vegardFeats['totalImp'] = vegardFeats.groupby(level=0)['importance'].transform('sum')\n",
    "vegardFeats = vegardFeats.drop_duplicates(subset=['totalImp']).sort_values('totalImp',ascending=False)\n",
    "print('top features over 5 training splits')\n",
    "vegardFeats.to_csv(os.path.join(os.getcwd(),'results','densityMG-Vegards_featureImportance.csv'))\n",
    "vegardFeats.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "r2Veg = r2_score(notTroubleDF['density_vegards'], notTroubleDF['vegards_predict'])\n",
    "print(r2Veg)\n",
    "\n",
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(notTroubleDF['vegards_predict'], notTroubleDF['density_predict'], edgecolors='k')\n",
    "plt.plot([0,25], [0,25], 'r-')\n",
    "plt.xlabel('Vegards predicted density [g/cm$^3$]')\n",
    "plt.ylabel('predicted density')\n",
    "\n",
    "\n",
    "plt.text(10,20, 'R$^2$ score: {:.3f}'.format(r2Veg))\n",
    "plt.xlim([0,25])\n",
    "plt.ylim([0,25])\n",
    "#plt.title('')\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(),'results','finalFigs',\n",
    "                             'densityMG-Vegards_predictScatter.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Once satisfied with feature set, overwrite original datasets?\n",
    "\n",
    "with gzip.open('./datasets/density_features_subset.pkl.gz', 'wb') as fp:\n",
    "    pkl.dump(X_subset, fp)\n",
    "with gzip.open('./datasets/density_data_subset.pkl.gz', 'wb') as fd:\n",
    "    pkl.dump(notTroubleDF, fd)\n",
    "with gzip.open('./datasets/reduced_feats.pkl.gz', 'wb') as fr:\n",
    "    pkl.dump(reduced_feats, fr)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Restrict to 12 Features with varying importance\n",
    "Omitting outliers\n",
    "Compare to best model: Vegards+Difference predictions\n",
    "ref: Jae/Nils attempts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "varyFeatData = notTroubleDF[['comp', 'density', 'density_vegards', 'vegards_diff','vPlusDiff']]\n",
    "varyFeatData.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Train density_model on all data several times, get consistently performing features\n",
    "featListSorted = pd.DataFrame()\n",
    "\n",
    "kf = KFold(5)\n",
    "rep=0\n",
    "for train_index, test_index in kf.split(notTroubleDF['comp']):\n",
    "    print('Split #{}, training on {}'.format(rep, len(train_index)))\n",
    "    density_model.fit(X_notTrouble[train_index,:], notTroubleDF['density'][train_index])\n",
    "   \n",
    "    topFeats = pd.DataFrame(density_model.steps[1][1].feature_importances_,\n",
    "                          index=dens_featurizer.feature_labels(),\n",
    "                          columns=['importance']).sort_values('importance', ascending=False)\n",
    "    \n",
    "    featListSorted = featListSorted.append(topFeats)\n",
    "    rep += 1\n",
    "    \n",
    "    \n",
    "featListSorted['totalImp'] = featListSorted.groupby(level=0)['importance'].transform('sum')\n",
    "featListSorted = featListSorted.drop_duplicates(subset=['totalImp']).sort_values('totalImp',ascending=False)\n",
    "print('top features over 5 training splits:')\n",
    "featListSorted.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "featList = pd.DataFrame(density_model.steps[1][1].feature_importances_,\n",
    "                          index=dens_featurizer.feature_labels(),\n",
    "                          columns=['importance'])\n",
    "featList.reset_index(inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#diffFeatTop\n",
    "#densFeatTop\n",
    "# Gather default ordering of features\n",
    "#featList = pd.DataFrame(density_model.steps[1][1].feature_importances_,\n",
    "#                          index=dens_featurizer.feature_labels(),\n",
    "#                          columns=['importance'])\n",
    "#featList.reset_index(inplace=True)\n",
    "\n",
    "scoreDict = {}\n",
    "colList = []\n",
    "\n",
    "for n in range(0, len(featList)-20, 5):\n",
    "    # Grab n to n+12th most important features\n",
    "    featIndex = featList[ [x in diffFeatTop.iloc[n:n+20].index for x in featList['index']] ].index.values\n",
    "\n",
    "    X_temp = X_notTrouble[:, featIndex]\n",
    "\n",
    "    # Begin training with 5-fold CV\n",
    "    kf = KFold(5)\n",
    "    rep=0\n",
    "    \n",
    "    scoreDict[n] = []\n",
    "    print('--------------------------------------------------------')\n",
    "    for train_index, test_index in kf.split(varyFeatData['comp']):\n",
    "        # String for column name\n",
    "        colStr = 'n{}Part{}_predict'.format(n, rep)\n",
    "        colList.append(colStr)\n",
    "        \n",
    "        print('Split #{}, training on {}: {}'.format(rep, len(train_index), colStr))\n",
    "        \n",
    "        varyFeatData[colStr] = np.nan    \n",
    "        \n",
    "        density_model.fit(X_temp[train_index,:], varyFeatData['density'][train_index])\n",
    "\n",
    "        y_pred = density_model.predict(X_temp[test_index,:])\n",
    "        varyFeatData[colStr][test_index] = y_pred\n",
    "\n",
    "        #print(train_index, test_index)\n",
    "        rep+=1\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "varyFeatData.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for name in colList:\n",
    "    num = int(re.search('n(\\d*)Part*', name)[1])\n",
    "    currSplit = varyFeatData[ ~varyFeatData[name].apply(np.isnan) ]\n",
    "    scoreDict[num].append( r2_score(currSplit['density'], currSplit[name]))\n",
    "    \n",
    "#print('avg: {}, std: {}'.format(np.mean(scores),np.std(scores)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ns = []\n",
    "avgs = []\n",
    "meds = []\n",
    "stds = []\n",
    "mins = []\n",
    "for key, item in scoreDict.items():\n",
    "    ns.append(key)\n",
    "    avgs.append(np.mean(item))\n",
    "    meds.append(np.median(item))\n",
    "    stds.append(np.std(item))\n",
    "    mins.append(np.min(item))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ns = np.array(ns)\n",
    "avgs = np.array(avgs)\n",
    "meds = np.array(meds)\n",
    "stds = np.array(stds)\n",
    "mins = np.array(mins)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure()\n",
    "\n",
    "plt.title('Performance with 20 features')\n",
    "plt.xlabel('n where nth to n+20th most important feats used')\n",
    "plt.ylabel('r$^2$ score')\n",
    "\n",
    "plt.fill_between(ns, avgs - stds,\n",
    "                 avgs + stds, alpha=0.1,\n",
    "                 color=\"r\")\n",
    "plt.plot(ns, avgs, 'o-', color=\"r\", label='mean')\n",
    "plt.plot(ns, mins, 's-', color='g', label='min')\n",
    "plt.plot([-5,ns[-1]+5], [0, 0], '--', color='k')\n",
    "plt.legend(loc='lower left')\n",
    "\n",
    "plt.xlim([-5,ns[-1]+5])\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(),'results','featElim_20_noOut.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Sum importances\n",
    "featListSorted['cumSum_totalImp'] = featListSorted['totalImp'].cumsum()/5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure()\n",
    "\n",
    "plt.plot(range(len(featListSorted)), featListSorted['cumSum_totalImp'])\n",
    "plt.ylabel('Cumulative Feature Importance')\n",
    "plt.xlabel('n most important features')\n",
    "\n",
    "if saveFigures:\n",
    "    plt.savefig(os.path.join(os.getcwd(),'results','cumSum_FeatImp.png'))\n",
    "    print('figure saved')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0                       maximum MeltingT\n",
       "1                       mode GSvolume_pa\n",
       "2                 mean Electronegativity\n",
       "3                         mean NUnfilled\n",
       "4                            mean Column\n",
       "5                  mean SpaceGroupNumber\n",
       "6               avg_dev SpaceGroupNumber\n",
       "7                          mode MeltingT\n",
       "8                       avg_dev MeltingT\n",
       "9                          mean MeltingT\n",
       "10                        range MeltingT\n",
       "11                   mode CovalentRadius\n",
       "12                     avg_dev NUnfilled\n",
       "13                        mean NsValence\n",
       "14               avg_dev MendeleevNumber\n",
       "15                         mode NValence\n",
       "16                      avg_dev NValence\n",
       "17                         mean NValence\n",
       "18               minimum MendeleevNumber\n",
       "19                           mode Number\n",
       "20                           mean Number\n",
       "21        mean simul. packing efficiency\n",
       "22                      maximum NValence\n",
       "23                        mean NfValence\n",
       "24                       mean NdUnfilled\n",
       "25                   avg_dev GSvolume_pa\n",
       "26             avg_dev Electronegativity\n",
       "27    mean abs simul. packing efficiency\n",
       "28                        avg_dev Number\n",
       "29                        minimum Column\n",
       "Name: feat_name, dtype: object"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reduced_feats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}