run_analysis_taxiV2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "from agent import Agent\n",
    "from monitor import interact\n",
    "import gym\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimal_sarsa_max = {'algorithm': 'sarsamax','alpha': 0.2512238484351891, 'epsilon_cut': 0, 'epsilon_decay': 0.8888782926665223, 'start_epsilon': 0.9957089031634627, 'gamma': 0.7749915552696941}\n",
    "optimal_exp_sarsa = {'algorithm': 'exp_sarsa', 'alpha': 0.2946281065178629, 'epsilon_cut': 0, 'epsilon_decay': 0.8978159313202051, 'start_epsilon': 0.9803552534195048, 'gamma': 0.6673937505783256}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Episode 20000/20000 || Best average reward 9.493\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.245\n",
      "\n"
     ]
    }
   ],
   "source": [
    "env = gym.make('Taxi-v2')\n",
    "agent = Agent(**optimal_sarsa_max)\n",
    "avg_rewards_sarsamax, best_avg_reward_sarsamax = interact(env, agent)\n",
    "\n",
    "agent = Agent(**optimal_exp_sarsa)\n",
    "avg_rewards_exp_sarsa, best_avg_exp_sarsa = interact(env, agent)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.DataFrame(list(avg_rewards_sarsamax), columns=['reward'])\n",
    "data.loc[:,'episode'] = range(0,len(list(avg_rewards_sarsamax)))\n",
    "data.loc[:,'type'] = 'Sarsa Max'\n",
    "\n",
    "data1 = pd.DataFrame(list(avg_rewards_exp_sarsa), columns=['reward'])\n",
    "data1.loc[:,'episode'] = range(0,len(list(avg_rewards_sarsamax)))\n",
    "data1.loc[:,'type'] = 'Expectation Sarsa'\n",
    "\n",
    "data = data1.append(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "reward          9.49\n",
       "episode        19900\n",
       "type       Sarsa Max\n",
       "dtype: object"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data[data.type=='Sarsa Max'].max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "reward                  9.24\n",
       "episode                19900\n",
       "type       Expectation Sarsa\n",
       "dtype: object"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data[data.type=='Expectation Sarsa'].max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAssAAAHxCAYAAACbAfGNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nOzdd5xU1d3H8c8t07dXFhaWXpYivQiCIthQYlQSYkGjRhODiSaax8QnUWM0lhhN1Bifh9jjY4mxJdgbooiAdJDOUrb33en33vP8MYgSWEXZMuz+3q+XL9k7M+ecmTuzfDnzu+doSimFEEIIIYQQ4iB6Rw9ACCGEEEKIZCVhWQghhBBCiBZIWBZCCCGEEKIFEpaFEEIIIYRogYRlIYQQQgghWiBhWQghhBBCiBZIWBZCCCGEEKIFZkcPoK3V1QVxnPZdSjo7O4WamuZ27VO0Dzm3nZec285Lzm3nJOe182rvc6vrGpmZgRZv7/Rh2XFUu4flz/oVnZOc285Lzm3nJee2c5Lz2nkl07mVMgwhhBBCCCFaIGFZCCGEEEKIFnT6MoxDsW2LuroqLCvWJu1XVuo4jtMmbYsjZ5puMjNzMYwu+fYXQgghxNfQJdNCXV0VXq+fQKAbmqa1evumqWNZEpaTkVKKYLCRuroqcnIKOno4QgghhEhyXbIMw7JiBAJpbRKURXLTNI1AIK3NvlUQQgghROfSJcMyIEG5C5NzL4QQQojD1WXDshBCCCGEEF9FwnIHa25u5pe//HlHD0MIIYQQQhyChOUO1tTUyJYtmzt6GEIIIYQQ4hC65GoYyeSee+6kurqKX/7yGnr37sPll/8YgFtvvYkJEybx0Ucfomka27dvo7m5mYsuuoRTTplFKBTij3+8ne3bt+E4DuedN4+ZM0/p4GcjhBBCCNG5yMxyB7vqqmvJycll/vyrePPN11BKEQ6HWb78Y4477ngAqqoq+etfH+LPf36A++//EzU11Tz66N8YNGgIDz30BPff/z889thD7N27p2OfjBBCCCFEJyMzy0miR49CunUrYNWqT6ioKOfYY6fgdrsBOO20MzBNk7y8fIYPP4Y1a1axfPnHRKMR/v3vlwCIRCLs2LGdHj0KO/JpCCGEEEJ0KhKWk8isWbN5441Xqaio4OKLL9t//Is7zSnlYBgmjmPz61/fzKBBgwGora0hLS293ccshBBCCNGZSRlGBzMMA9u2ATjhhBNZsWIZtbXVDB06bP993n77DZRSlJeXsWHDOo45ZiSjR4/jhRf+AUB1dTUXXvg9KirKO+Q5CCGEEEJ0VjKz3MGysrLJz+/GlVdezr33PsiwYcPp27f/AfeJRiNccskFxOMxrr32etLTM7j44h9w1123c8EF38FxHK644idSgiGEEEII0cokLHcw0zT5618fQilFMNjMpk2buOKKnx5wnxNOmMFpp51xwLFAIIXf/Obm9hyqEEIIIUSXI2E5SWzcuJ6f//wnfP/7PyA7O6ejhyOEEEKINuQoRXM4js9t4jJ14pZNdUOEplAcw9BQDng9BpGoTShqEYrG2V3ZjOMoPC4Dv9dFis/E5zHRNA2/xyQn3UtWmrejn1qnI2E5SRQXD+OVV94+6Pj119/Y/oMRQgghxAFsxyEYtrBsB8t20DQNt6njMg0iMQufx0TXNGxH4SiFbTvYjiIYsQhF4oQiFlv3NhCJ2eytamZnRROxuANAwGsSjFhfOQZD1zAMjXjcQbVwnx5ZHgb2SCUYU+RnpxK3nf39K8cmxx2lKMdNJBSGeITUFB9el4GtFCleF24VJqBFMTSFqWLozVUoK4pmuNB8aWC60XQDdBN0I/EfoMKNgIZTX4qKBiEeAU1HRZtRkWZwbEBDORaaJ4DmTUFz+QCVuM3lTfxsmFgnzwPcrXHaWoWEZSGEEKIFTqgBFaxDBeuwK7eB24/m9qGZ7kSA8KSAHcPsNRIVj+A014CmQTQELg9Gbl8wTLBiif/riVlAAGVbgELFwgf0qbm8ibbdPrBiqEgzelreNxq/chxUtBmnZjd2dQma6ULzZySeW+0eNG8KoKHi4UQAMtzoKdlo6floHj+aLx0FRKI2Po9BOGoTj0aJRKOkpKbidykwXIm+wo0o3Uy8PvtCY1lNiMqaRmzLxtZdOI6iMRjDZerkZ/kpyPKR7XVwgrWJ56sZqGAtSjmgVOI/QPOmovlSUc012NUlqGiQ+lQ/cQLoKVlogazE83IsLNumMaIRilpE4za6phG3bExTJ9XvJsXrQtPA4zLQde3z10opNE0jEo2zfkcNJZVBLDvRv2U5rNhcRV1T9JCvs46DgUMcE1Bk6kG8WhyfFiOuDOqcAGHlRjNMXKZBtyw/xw7tRkF2gHAkTkMwRnqKm5wMH2kehRasRXPi2JaN6Qvg9XrxmQ7pegTTCuLEQsTCYWKRMFYsitFchVGzDS0WxFAW7E2MK1ThJo5BCD+peogAYTSAssN7/9hAk+MmiguXZhPQImhf8ZignkpE9xHX3GjKIaw8xM3uiVfIsYlYkGLHSYlE8VCHrTQcNHxOJRoOtu7FV1sPgW/2nm8LEpaFEEJ0eSrSTHz7MrCioBs49WXYFVtxanbDZ3N4mrY/vH1jmgamJzEbFw1++Zh0ExwbDYWlu7ENL3Hdg214iTk6urJRShHRfOBNwe0LoMeDBJp2Y1hBlKNwE/3KcPNlgo6bGC4q7DTQdLxEKTRq8WsOloJmDSxMdGx0FI4C0NA1RbPjwQT6aTE0FHvsbJocL17lxqfFsLQ49XoUj9HwjcZW+yW31dlpbI13Y7edRcjxYGo2GXqIDD2IW7OJKhPQMEwDnwk99SoynRosDAxl0x+bHsqFhUGdk4KuKca7ddK7WcR8OdguP+5wNZ5IFQodw4mjKwtbd6MpG13Zhx6Y6UbzpaG5vNCkoEnDaaoEOw6annhfOPa+WdiDWfv++4wbcKOB24fZczhaIBPN7U+0Y1uYwTpwLLLCjej+DLRAJo43nYhy4/V50T0+moMR4pYDGoTCMaK6j7qYC1tpROKKesuLQqM5HKe+KUxTU5hoLIbXpeExwGsqXIZGXcyN49jEcOPYCl3T0DQNr9sgEks8H5epk5bhIhi1aArGaAzF8XkMTEMHBQqF2zT4lTcX3zd4T7QVCctCCCE6DbVvNlLTD1wZVcXCxNa8gl2xFRUJork8aP4MVCyEioVRTVX7vkbex+XFyO2De+Qs9Jwi8Kah5/QmFIkRDwfxEmNTWQQPUXJ9DtVb1xOOK5q8+QDYmJhOlFy7Ag3F3ro41fUhPLpNvsfEo9vE01JwHIew5sXtMmkOx4lELYxoAw2WmxRCRJWLqHKRaQTxajF8+2YrXZrC2RdGAlot3vBevPVxmh0PJU4a1SoXv8dFbdTA8aTSYHnZa/bEti1S7EY8Lo1mdx6aHUM5DlHc+FyQ5nZI10NkO9XkeGL4rCawYvS0Kok6Ct3lp9ZfhO3PJRJsZmd1DB8RDJcbw59OpjsGSmEr8NghAn43RkoaBg496kvQ4hF0uxmlu4jqqQSjAT6yhlIa8eNEE/W4DSoFv9+DaZrErUTISjOi+FWIqqib3XYOhj+VWMxGC9eRqQfJ1IPkuGP4/V68bo3e7jomNWxlsrP5gPeBbXixdBeGHSeRzmxQUEEem/QBuAwNze0jNzeTHJ+D7lhkNVWBchJB1/Rg1+6BaB1aIA2jaApogOFCc/tRkWY0w9w3052OZnpQ8Qgq0oiKhlDRYGIGPtKU+HbCcXD1KEYzXSjHTrx3DRM9q2eiP01LPN6KJcog/BmJsO32o7k8YLoP+LbicH0xiHq+1iPbR25uKlVVTR09jP0kLAshhDiqKaWwtn9MfPMHODW7UNEgmiewr+zBRcjtxm6shmgQPbsXttLRo0EINyZKDdx+9PwBuIbNoNbIYcO2SvY2mUTiNvEyh/L1IUqry4AyLNtpYRTd/uNnB3ABiSU98zJ9DB+cTdi2WbStZn+ZgsvQMQyN5lCcrDQPWfsu0Ap4TQpyU/B7Tfp2T8Nt6gQjFim+RMmDrmvoXwhIlu1Q2xQl3dDJdxsHlRi0lRFH+PhcoPe+PyulCEctPG4DQ//qbSCys1PYsqOauqYoBdkBfB7jgNCorBgq1ICKNILbh+7PTJR6fEHiH1cOmfvqboU4FAnLQgghOpSKBrFKViXqdHUTpRxUsI6wchMOBglH4sTTCnHSuxNoKsEf8JHmslCRZqySlTj1ZRANoqVkY6d1p15Lp74xRLAxhB2PYWATVzks5URKdmUTjtp4XAaZqR6aQjHyMn24DJ3QK03sqUoUc3rdBgGvC12HvEw/00dnomsaPq9Jqs9FbVOEIb0yMU2d8poQ3XMD9C1IAzggsNU0RLCVIjfd+7Vn//5TRkrLgc40dPIykumL669P0zT8Xtdh31/XNbLSWl79QTPdaGm5kJb7pX2iSVAWX07CcgcrKyvle987i969+x5w/IwzzuTss7/Tpn2/+OI/8fv9zJx5Sov32bBhHe+++zZXXPETFi9+j08/3cill/7wiPqNxWLce+/drFq1Al3XSUlJZf78qxgyZOgRtSuEODxKKXAsNKPlYKKUA7FwYob2MDif1fI2VhDfsQIcC9VUk6idVA5OYyWaNxUVyKI+7qE2atBdr8FoLker3o5mxw9q0wBSAL+C/5wk/ewyqzI7k53xAkrsHHZpI6nYG8GyHbxugyFFmXTL9tOzWzqRxjC51UG66RoFOQGq6sNU1oXp1z2N+uYotqPISvPQt3sqp0woIj/Td9jhdkBhRou3ZafLMl5CHO0kLCeBnJxcHnnkyXbvd926NYwaNeZL77Nz5w7q6hKXUUyZMo0pU6Ydcb/PPPMkSjk89tjTaJrGmjWruO66n/Hcc//GNOUtKcSRsKt3YpdvAcfC6F6MntENp7EKHBu7YguxtW8kvpaOxzD7jcM94lRUPEJs+3IioRBuXWFYYWI1eyBYyybvSMJGgPyCbjS5MqmsDRG3HLwqTHVDlGpXPjmuGKpuNyP5lN5GxUFjalZeap0U0ow9BFSIVM0mFYgpg712OrutIpZEB1LnBDA1GwMH25/FicNz6JabTl6GF61uNzSUEnGlsXZPhMogmIF0PP40UlPc5NiKYEUTA4syGT84j77d03GZia/yk63+UQhxdOnyyeSDtWUsXnOYa6gcps8umJ4yooDJwwu+cTubNn3KNdf8hMceexrD0Pn+98/jttvuYtGid9m9exd79+6hsbGBb33rLM49dx62bfOXv/yJlStXYNsOp512Ot/97nkopXjggXtZtOhdTNNg9uyz6NOnL4sXL2LFimVkZ+eQm5vL3XffSTgcpq6ulrlzz+eUU2axYMFfCYfDPPro38jNzWPlyhVcf/2NrFu3lj/96Q/EYjEyMjK49tpfUVjYk/nzL6O4eCirV6+ivr6Oq666lkmTJh/wvGpra4jH41iWhcvlYsSIkfzylzdg24kLOe666za2b99GbW0tvXoVceutd1BbW8vPf34l6ekZuN0errzyau644xZs28btdvOrX91Az569eO65p3n11YVEImF0Xeemm35P7959juh8CpEsPrt4za7chgrVgeNglazCaaxAxUKAhmoo/9I29JwizMJhRKNRwts+xtr6EQBxZRBSbprQiWkeGiwPHi2Tgc4nGJqCLYnHD/7PBi0gDBgQwcMb4WEstQZjudNoCMbolZdKTqafgM9FUyhORsBFcaGfTFeM0iaNkOMmM8XNSbZDMGKR5neTkeKmX4/0xBXyn+meDYwEoN+4VngxhRDiMB0VYfnll1/mgQceIB6Pc9FFF3Heeed19JBaVXV1FRdddO4Bx379698yaNBgzjzzbP7ylz9hWRZnnnkOAwYMYtGid9mxYxsPPPAQjuNwySXnM2bMeDZuXAfAQw/9nVgsxs9+Np/Bg4upqalh7drVPPbYU1iWxRVXXMpdd/2ZKVOmMmrUGCZMmMSf/nQXF154CWPHjmfv3j1cdNG5zJkzl0sv/SErV67gwgsvYeHClwGIx+PceOOvuPnm2xgyZChvv/0mN954PQsWPLbvdosHH3yYxYsX8b//+8BBYXnOnO9x7bVXcfrpMxg1agxjxozn1FNPx+PxsGrVJ5imiwcffBjHcfjJT37IkiUfMGjQEHbtKuHZZ++loKA7t956E3Pnns/06TN4663XWb9+LVlZWSxa9B733fcgHo+XBQv+yvPPP8vVV/+iHc6iEK1DKQfVVI1yLJy6MoiFsOv24tTsxmkoRzXXHPgAw8QoGJxYh1c5aEUjcQ+bCbqJtWM5dtUOzB7FVDfGaXJns605hYr6MB+uq8CjejIrdxexWIxN/jGMLi6kpLyJSMyiT0Ea44fk4/WAHYtSUVqOGaykIDslsR4viXVwnabqxNX5poeUnsM4Kao4w6WjaRq1TdEvraOVf8YKIY4GSR+WKyoquPvuu/nnP/+J2+1m7ty5TJgwgf79+7dK+5OHH9ns76GYpo5ltXTF9MG+rAzjwgsv4ZJLLsDj8fDrX/92//EZM07G7/cDMGXKVFasWMaGDWvZsmUzK1YsByAcDrFt21Z27tzO9OkzcbvduN3uQ/Y1f/5VLF26hMcff5itW7cQDodaHO/u3SWkpqburzGePn0Gd9xxC83NzQBMmDAJgL59+9HU1HjQ4wsKuvP440+zceN6li//mFdf/TfPPPMkDz30d0aOHE1aWjrPPfcMu3btZM+e3YTDiQX7MzOzKCjoDsCkSZP54x/vYOnSDzn22OM4/vgTMQyDG2/8HW+++Tq7d+9i6dIPGTBg0Je/+EIkEaUU4df+jL1r1UG3hfwFuH2Z+PqOQ/OkYPY6BpRNSPOzqswmGLZoDMXIcHnoWWfgdmmUa0P5NNyd6EaHpRsqgEqgEo/LYGJxPmdM7k3u4VwU5vXSNy0d+OrPU8oXmjvaLzgTQgg4CsLyhx9+yMSJE8nISFxAcfLJJ/Pqq68yf/78Dh5Z+2hqaiIUChEKhWhsbNz/OhjG51fvOo7CNA1s2+GKK37CtGnTAaivr8fn8/Lgg/cf0GZZWSkZGZkHHPvNb64jNTWNyZOP48QTT+Ktt15vcUyOc6hF+RXOvkXU3e7EFpWapiW+Nv4PDz54P2edNYfi4mEUFw9j3ryL+dGPLmbZsqW43S4WLHiQOXPmctpps6mvr9/fhsfz+WqQJ5wwg2HDRvDBB+/z7LP/x0cffcBFF13KlVdeztlnf4eJE48lKyubLVs2tfg8hGgrSilqG6NkpnrYU9VMSUUTGho7yxuJWw7pKW6UgtrGCGl6mKLYFvKsMnIaNqA5FtV549gYzKI2ohFy57CtwaCqVqEBY/RcMlI9lG6qprIuTCRm0xxOXBx3qD0zTCOxFu/wvtkM7JnOccd0J9XnOuKVGYQQoqtI+rBcWVlJbu7ny77k5eWxZs2aw358dnbKIdrUMc2vXsPxSBxu+8a+mryW7n/PPXcwZ853Ucrh7rtv55ZbbkfXNd5//12+973zCIfDfPjh+/zhD/dgmiYvv/wC06ZNIxaL8+MfX8ovfvErRo8ew9NP/x9z5szBsiyuueYn3Hnn3ZhmYokm09RZtuxjnn76n+Tm5vKvf70EgKYpXC4Tx0lsE6rrib90+/btQ2NjI5s3b6S4eChvvvk63boVkJWViaZpGEbi9W3pudXUVPHYY3/jZz/7BS6Xi4aGBurr6xk4cADPPfcsM2bM5FvfOpOqqirWrFnJhAkTD2rr+uv/i5kzT+acc+bQr19f7rnnLjZv3kjPnj0577wLiMViPPHEw2RkZB7ytdV1ndzc1MM6R//pmz5OJL/WOLert1Tx91c/ZePOWjJSPNQ3f749rt+t0d9bR561GRuNoZ44fdhLihYmrnQ2W3lElYuHPx1EblYKhXkphKMWPbobXD65Dxt21LLwwx1EYjaFeSkMLMpEQ2PqqB4M65dDqt9FWU2Q8poQ0ZhFZpqX3gVpeFxGlw/H8rntnOS8dl7JdG6TPiwfamby6/zSr6lpPmgm1HGcr1Um8XV9nTIM23aorq7i/PPnHnB85MhRDB8+kt27d/Ob3/wOpRSvvDKP1157DcdRuN0eLrvs+wSDQc4//yJ69uxNQUEhu3aVcMEF38O2bU477QyOOWY0AOvXr2fevHNxHMU558yle/eejBkzjgcf/At+f4CLL/4Bl112MampKfTsWURBQXd2797DoEHFLFjwIPfe+yeKinqjlELXTW666VbuvPM2IpEwaWnp3HTT77EsB6UUtp14fe19i/f/52tx1VXXcu+99zBnzpl4vT5cLpMf/nA+hYVFnH76mdx00/W89dYbuFxuiouHsWfPHkaNGntAW+ef/31uv/13PPTQ/2IYBvPnX82gQUN47rlnmTv3bFwuF8XFw9i+fdshz4XjON/o6ni5qr7zOtJzG7cc3ly+m2ff3YbPY3DGsb0pqw2hazB7ch8MJ0Lg3btQ9aXgcaHQ0L0BjJzBuEedzq54JqHqCOkBNzekeSnMDRz0u65vfgqzJvTEdtSBF78B0VCUaCiKC+iZ9Xn5Q1NDmK7+jpXPbeck57Xzau9zq+vaISdXP6OpQ6XRJPL888+zfPlybrnlFgDuv/9+lFKHXYZxqLBcXl5Ct25FrT7Wz3zdmuWv629/exCASy65vM366Oy+6XtAfjkf/eKWQ1MohmHopAfc+4+3dG4t22F3ZTMBr0lGiod1O2rZsLOWxmCMFJ+LzDQvKMXCpbuIxmyG9cniim8VY5Svw6kuAd3APfREIosextqxHPfI03Efc+phr18sjpx8bjsnOa+dV7KF5aSfWT722GO59957qa2txefz8frrr3PzzTd39LCEEEeZpRsqeOmDHZTXhvbX9Q4pyuSc4/vRu1sqocjBm2LsqWrmz/9YQ3VD5IDjHpeB32sSt5z99cKGrnH21N6coK8g/vf7iX9hk43YihdAOXgmfhf3iFPb7kkKIYRodUkflvPz87n66quZN28e8Xicc845hxEjjnQ3+qObzCgL0TKlFKu31vDkm5vRNMhO8+J2GWzYWUvA5+K0iUVkp3spKW/ivVWl3Pzocvwek1DUIifdy1nT+lKUn8q67bUs31RJMBLnB6cXYzkOn5bU0btbGieM7oEea8ap3EZzfT1aYwVGvBlKFmI3VWH0OgZX7zEYvY5BheqJb/kQzeXBNbzl3TKFEEIkp6QvwzhSnbEMQxw5KcPoXN5btZdn39lGYW4Ay1FsL20kxedicK8MGoIxmsNxumcHmHfKIFL9n5deVNaHWbm5ii17Guiel8K6rdXsLG9CQ5H4raExe3Jvzjzu8+3oleNgbfmAyEdPQTT4+SBcXrAt3GPPxDPy9HZ77uKryee2c5Lz2nlJGYYQQrQSx1G8uHgH/15SQo/cAHFbEYlZfGtKH6Ye053MVM+XPj4vw8dJ43py0tge5OVnULqrjNrnbsUXKkN5AqC7MHYbND8WRQtkgmPh1JUCoAWy8E7/IZo/A82fjubygm6i6W270o4QQoj2JWFZCHHUcJRi9dZqErs+K15ZuovtpY1MGtqN82YOxO9t+VfaZ1+ifbbChLItIm89gLVnLVgxQt4ATiSIH3CNOAV71xqcpir0HsXo/kzsqm0o28boNhCz7zhcxSdKMBZCiC5AwrIQIqlFGhtYtyeEx+vmpcU72FbaCCh0FKZpMmtSEWdP64dTX4ZV34TmTUFPL0DTNJxQPdhxtJRsYqsWElv1L/SsQoz8/qjGKqydKzCLRqFn98IVqSYaszH7jMPVZwxq/BxwLDTzy2enhRBCdG4SlpPAO++8yeOPP4Jt2yjlcMopszj33Hnt0vfChS9z6603ccMNv2PmzM8vPnrmmSf585//yLPPvrR/i2kh2ltNTQPGc1djxwq4r3katubiV71XkBvcim5HwZuKEe5H8PlGnKrt+x9n9ChGc/mwdq4AQPOlocKJrdc1TSe+9g1QDu6x38Yz+lvAwTVymm6AbiCEEKJrk7DcwaqqKrnvvnt46KEnSE/PIBQKMX/+ZfTqVcSUKdPaZQx5efm8++5bB4Tl9957h5SU5Nk9R3QuSqnErK3hOuBYJGZTURciPeBhycqdqLUvMdXlUOzeyx+ynkSlFaA1loHbj3v46Tiheuzda9B86XgmzkXPKsSu2kFs1UKIR3CNOAXdn0F06bPg9hE46yb0tDxULIyy4+i+tA58FYQQQhwNJCx3sPr6eizLIhKJkJ4Ofr+f//7vG3G7E1/9vv32mzz11BNEo1Gi0SjXXfffjBw5mvnzLyMtLZ0dO7bxm9/czD/+8RTbt28D4NvfnsPs2d9m+/at3H33nYTDYerqapk793zmzJl70BhGjhzN6tUrCYfD+Hw+ysvL8Pv9pKQkrgy1LIu77rqN7du3UVtbS69eRdx66x0sW7aU++67h0cffYqqqkquvPJyHnzwYfLy8tvvBRRJTyl1wE50TnMN4Tfuw6nZjWf8HNwjTiYSs7jl8RVk12/gGPcusvRmJpk1mC6HeMExpA4ah127B2vbx+g9h+ObOb/F8gizcBju4Scnyi/2bfxh9h0Hjo2elgeA5vah4Tvk44UQQogv6vJhOb75A+KbFrVqm5qmoZTCNWgqroGTv/S+AwYM5LjjpvGd73yLgQMHMWrUWGbOPIXCwp44jsOLLz7HHXfcQ0ZGBv/614s8+eTjjByZ2MK6X7/+3HrrnaxcuYLGxkYefvhJGhrque++e5g9+9u8/PKLXHjhJYwdO569e/dw0UXnHjIsG4bB+PGTWLLkA6ZPn8Hbb7/B9Okz9+8UuG7dGkzTxYMPPozjOPzkJz9kyZIPOP74E3n33bd57LGHWLlyBT/+8U8lKAti694kvnkxZq8RaIEsYp+8BLqB7s9AWTGcmhJweXHSCoh+9H+8uSmM2bCX0yJlFKfuxTG9mFaYaEp3Uo6fR0rBIDRNwwUw8eD376FophvMz5eI01Oy2+bJCiGE6PS6fFhOBtdc80suvPASPv74Iz7+eAmXX/59brjhZqZNm86tt97JBx+8z65dJaxcuQL9C1ffFxcPA6Bv337s2lXCz342n4kTJ/OjH10JwPz5V7F06RIef/xhtm7dQjgcanEM0zx7z44AACAASURBVKfP4OWXX2D69Bm8//673Hnnn/eH5ZEjR5OWls5zzz3Drl072bNnN+FwGICf/vQazj9/DsOHH8OMGSe31UskjgJW+Wbql7+Kp/QTAGLVOxM3BLLRvakoK4ITrANgad45vLXLwzz1T46rewGAaFo27h7j8U6ZhxNuIMWfgeb2d8RTEUIIIfbr8mHZNXDyV87+fl1fZ1OSDz9cTDgc4sQTT2LWrNnMmjWbl156nn/960XGjZvIpZfO4+STT+OYY0bRr19/nnvumf2P9XgSX0Onp2fw+OPPsGzZUpYs+YCLLz6fxx9/httu+y2pqWlMnnwcJ554Em+99XqL4xg9eix33HEL27dvJT09Y38JBsDixe+xYMGDzJkzl9NOm019ff3+Zbjq6mowDINdu3YSi8Vwu90tdSE6GbtyO+G3HgAgEo7gsZrwAO9GhrAwdAxpbkWeqqLGKuLX50zG4zL407OfUFq+h5panTQ/uGdcjr71eVx9x5I6bOb+to195RNCCCFER+vyYbmjeb1e7r77ToqLh1FQ0B2lFDt3bmfAgEHs3r0LXdeZN+9iAG6//Xc4zsEhfPHi93j11YXcfPNtTJgwiRUrPqaysoJlyz7mySf/QU5OLgsXvgyAbdsYxsFX+BuGwbhxE7jjjls566zvHHDb8uUfM336DGbNmk11dRWrV69k3Ljx2LbNLbfcxE9/+nNWrFjGggUPcMUVP22DV0kkA8dRvLliDx6XztBeqRiv3YcVjVASTWeQK7GKxALnbC699GSmKkjxmazYVMVfX1zPj+9ehMvUiVsO580cQ26Gl/490vF7XVD8yw5+ZkIIIUTLJCx3sNGjx3LxxT/gF7+4CsuyAJgwYRLf//4P0HWd/v0Hcu655+D1ehk5cjTl5WUHtTFx4mTeeectLrjgO7jdbqZNm06/fv25+OIf8KMfXUpqago9exZRUNCdsrJSCgt7HnIs06fP5LXXFjJlytQDjp9xxre56abreeedN3G53AwdOozS0lKeeuoJsrKymDZtOmPHjueCC77L1KnTGTZseOu/UKLDbCtt4Ll3t7GttJH4vm9MRrhKuCS1lr81TWez04sJgzLpn6NzwdD+B2wnPX5IPh6XwaLVpazcUs2koflMH93jgAv+hBBCiGSmqc++T++kamqacZwDn2J5eQnduhW1WZ9fpwxDdIxv+h5o7/3qO1pDc5Q/PLWKvdVBxg/JY1CvTKpqQ0wsfYxUux79rNtJT/vqumKlFLsrmynMTUHXkzMod7Vz25XIue2c5Lx2Xu19bnVdIzs7pcXbZWZZCHFIccvm7mdWU14b4oozhzF2cGLZtciS/yMe3o1n6vdxH0ZQhsQKMb3yZd1uIYQQRx8Jy0KIg7y0eAcvLN4BwJVnD2fUgFyUFSW27g3ia1/DNfREXIOmfkUrQgghxNFPwrIQXZCKhYgsfhz3sJk49aXEVr+C5kvDNeQE3q3pxguLd9C/Rzonj+/JyH6ZOOFGwq//GadiK0bPEXgmfFfqjoUQQnQJXTYs/+euYqLr6ORl+ocl+snLWFuXYG1dAoCWlo9qriHy1l+IRQbRp9sMfjYhirPlUZrfW7/vURre43/Q6kstCiGEEMmsS4Zl03QTDDYSCKRJYO5ilFIEg42YZtddD9pqqCC29nXKycWvgvjdOqsK5mKm5xJa/ARTvZ8yNbaJ+LugeVPBE4BoEN8pV2H2Oqajhy+EEEK0qy4ZljMzc6mrq6K5ub5N2td1/ZDrIYvkYJpuMjNzO3oY7SoWt3ljyWYa1rzHSa7lgMaC5hOotz3YjkJV1AF1pHnGM3FgJsbOJRg9huI77eegFKqxEj2joKOfhhBCCNHuumRYNgyTnJy2+4tflrMRyWTt9hpe/OfrXBBYTJY7SEzzUFZ0Mr85biYuU6e0OojPY1JaHaRnXgoZGdOxK2eiZxSgaTpooElQFkII0UV1ybAsRGemlKIpFCfV7+LFxTtYv3Qp81NfJ45J/ZCzKZx0MtlfKEPpU5AGQLesz5eBM/L6tvu4hRBCiGQkYVmIo1RVfZiymhAj+mVT0xDh/TWlFOWnsnhtGSu3VNM9J8DQ5iXMT1sNqXmkz/4VuYH0jh62EEIIcVSRsCzEUWjhRyX8491tAMwYU8iqrdVUN0QA0IAhRZlE9m7i9LRVOPmDSZ1+CboEZSGEEOJrk7AsxFEkFLH4eGPF/qAc8Jq8uWIPAKcfW0R6wMOIftnkZvioeGkRqsZH2qyr0UxPRw5bCCGEOGpJWBbiKLFpVx3/+68N1DZG8XkMfn3hOLpl+amoC1FdH2FIUSa6nlgKMb55Mf7ylbiGnyxBWQghhDgCEpaFOAo4SrHgXxupbYwy98QBTBlegN+b+PjmZ/rJz/z84jyrdCORdxegeVJwj5zVUUMWQgghOgUJy0IkOaUUr328i5rGCJedUczEod1avK9dt5fwK3ej+dIJfPc2NLevHUcqhBBCdD4SloVIci9/uJMX3t/ByP45jBmU1+L9nFA90Y+eBsB/xi8lKAshhBCtQMKyEEnGUYq/v76ZT3fVceEpg1m4pISxg3L50ZnDDtqeXTkOTvVOMN2EXroF4hHcY89Gz2h59lkIIYQQh0/CshBJpDEU4/a/f0JZTQiA2/7+CQDfOq7vQUEZIL7udaIfPQWA5kvHf+ZvZFtqIYQQohVJWBYiCTQ0R3npw518vKGCYMSiMDfA92YM5H9eWk+/Hun0yAkccH/lWNiVO4iueCFxQDPwHDdPgrIQQgjRyiQsC5EE/uflDWwsqQPgeycO4ITRPTANnbuvnHLA/ZRyCL34O5zK7fuPuUd/C/fo2Wi60a5jFkIIIboCCctCdJBFq0vZWFLHscO6sbGkjplje9IrP4Vjh3U7ZMkFgF2+ZX9Qdo89C1e/CWipuWi63p5DF0IIIboMCctCtLPaxgh1zVEeeeVTAJZuqKAg289Z0/ricbU8O2xXbiPy9oOgG3in/xCzzxg0TUKyEEII0ZYkLAvRjjbvrueup1cRtxwAeuWnUF4T4rIzhrYYlJUdx967nsg7C8Dtwz/7eoy8vu05bCGEEKLLkrAsRDspqwly73NryEhx06cgjZ55Kcya1Jto3G4xKNuV24m8/zBOzW5w+/Gf/FOMrMJ2HrkQQgjRdUlYFqKdLPjXBgxd4+dzR5GX8fmGIS3OKFsxwm/cB1YM7/E/SJRduLztNVwhhBBCAFLwKEQbU0rxxOub2FHWxBmT+xwQlL9MdNlzqGAt3pnzcQ2cLEFZCCGE6AAysyxEK6trinLr48uZe+JAonGLF97fQXVDhMLcAJOG5h9WG05DOfF1b+IaPA2z++A2HrEQQgghWiJhWYhvqCEYo6E5SmFeCvq+pd7WbKvhnmdXA3D/82v333f8kDwunz20xSXhvsgJ1hF84WbQwD3qjLYZvBBCCCEOi4RlIb6hR1/5lFVbqzENjRNGFRKNW3y8sRKAvEwfaQE30ZjNd6f3Z3CvzMMKygCx5f+EeATfadeip+a05VMQQgghxFeQsCzElyirCfLkG5sprw3j85hMGdmDVK/B4jVlbCypQ9NgeN9s3li+G4D+hel894T+9OuR/o36iyx+jPim93GNOEXKL4QQQogkIGFZiBaU14b4/ROf0ByOA9CnII2n3th0wH2uPHsEI/pl89rHu3AcxaxJvb9xf7E1rxLf8DZGj6F4Rn/rSIYuhBBCiFYiYVmIFryxbDfRuM01c0eiaRpDijJZuqmK9VurOXViL2oboxT3TpRXnDqh6Ij6smt2E/3oKbTUXHwnXSkrXwghhBBJQsKyEIcQjdl8vLGC0QNzKe6dtf/46VP6MmFQLgAF2YEj7kfFwti1e7A2vQ+6SeDbN0hQFkIIIZKIhGUhvsBxFO+vKeWZd7YRjlqcOLrtdsuL71hBZNFDEA0C4Co+Ec2b0mb9CSGEEOLrk7AsxBe8vmw3z7yzlZx0LyeM6kH/wm92od5XUVaU6OJH0QNZuI+7EBWsxzV4Wpv0JYQQQohvTsKyEPssWV/OM+9spW/3NH51wZj9aye3NruulNDzNya2sZ7xY8yCQW3SjxBCCCGOnGx3LQQQjlr8/fXN5GX6+MEZxW0WlFU8SuSNe8GKYfQ6RoKyEEIIkeRkZlkIYPW2akJRi5+cM4L8TH+b9RP75EWc+nJ8p12D0X1Im/UjhBBCiNYhYVkIYNnGStICbvp/w81EDodybOKbF2P2Ho1ZOKzN+hFCCCFE65EyDNHllVYHWbWlmuNGFKDrbVN+AWDv3YAKN2IOOLbN+hBCCCFE65KwLLq8jzZUgAYzx/Vssz5UpJnY6oXg9mP2GtFm/QghhBCidUkZhujyVm2pZkBhBml+d6u2a5WsREvJRoUbCb9xH8QjeCbORTNcrdqPEEIIIdqOhGXRpUViFnurmpk9pU+rtamUQ+zjfyRmkvfRAln4Zl+Pkd12s9dCCCGEaH0SlkWXtLuymer6MAGfCwUU5ae2WtvxdW/uD8pGj6HYe9fjKj5BgrIQQghxFJKwLLoMpRSaplHfHOW3jyzDdhRpgUTpRa/81tlmWsXCRD95EaNHMb6TfwqGG7t0I0bBwFZpXwghhBDtS8Ky6DJu//snNITiZKV6UApSfC4agzGG9ckiM9VzxO0rxyG27g2IBvGM/w6amWjT7FF8xG0LIYQQomNIWBZdQk1DhM17GvB7TCpqQ5w4upA5J/Rjd2UzRd1S0Y5wxz6nuZbgkz8DwOg1EiO3dyuMWgghhBAdTcKy6BLWbK8B4JcXjCEUidO7Wyou06DfEWxC4jTXoCJNaO4AwX/esP+4d+J3j3i8QgghhEgOEpZFp7ZmWw3/+nAnW/c2kJPupXu2/4hnkQGUbRF6/iZUuHH/Mc+UeZh9xqL70o64fSGEEEIkBwnLotPaUdbIvc+tQanEzwMK01slKANElz69LyhrgELPKMBdPL1V2hZCCCFE8pCwLDqt91aVYpo6d/7oWFZvrWZ43+xWadeu3UN83Zu4iqfjnTIPq3QjekrrtC2EEEKI5CJhWXRK/16yk0WrSxk7OI8Un4vJwwtarW1r60eg6XjGngWA2X1Iq7UthBBCiOSid/QAhGhtbyzbzXPvbSc9xc1pE3u1evt2+Wb03CI0b+uszSyEEEKI5JU0M8svvPACf/jDH8jOTnydffzxx3P11VdTWlrKtddeS01NDX369OEPf/gDgUCgg0crkpXtOLy4eAeFuQGuO280fq+rddsPN2FXbcc1dEartiuEEEKI5JQ0YXnt2rVcd911nH766Qccv+mmmzj33HOZNWsW999/P3/5y1+49tprO2iUItntKG0iFLWYd+ygVg/KSjnUvvV3sG1cA6e0attCCCGESE5JU4axdu1aXnjhBWbPns0111xDQ0MD8XicZcuWcfLJJwNw1lln8eqrr3bwSEUyiMZtlFKEoxaV9WFe/mAHO8oaeey1T/F5TIp7Z7V+n0ufoWn1W7iPORUjq7DV2xdCCCFE8kmameXc3Fwuu+wyRowYwR//+Ed++9vf8l//9V+kpKRgmub++1RUVHytdrOzO6auNDc3tUP67Qos2+GHt71FRW3ogOPPv78DgN9cMoE+vVo/LJds/xj/wPHkz7q41ZagE8lFPredl5zbzknOa+eVTOe23cPyK6+8wu9///sDjvXt25dHHnlk/8+XXnopM2bM4Be/+MVBj/+6IaWmphnHUd9orN9Ubm4qVVVN7dpnV7G3OsivFyzd//PAnhnoGpwyoYh/vLuVcYPz6J0baPXXX8XC2M21eLr3p7q6uVXbFslBPredl5zbzknOa+fV3udW17UvnVxt97B86qmncuqppx5wrKmpiUceeYSLLroIAKUUpmmSlZVFc3Mztm1jGAZVVVXk5eW195BFEnn67S0A5GX6+N2lEzCNzyuJRvRru7WOnYbENxqu7O5YbdaLEEIIIZJNUtQs+/1+FixYwOrVqwF44oknmDlzJi6Xi7Fjx7Jw4UIgsWLG1KlTO3KoogM1h+Ns2FHHqRN7cdvlkw4Iym3NqS8FwJ3Vo936FEIIIUTHS4qaZcMwuOeee7jxxhuJRCL07t2bO+64A4AbbriB6667jgceeICCggL++Mc/dvBoRUfYtKuO259cCcD4wfnt3r+1YwWaNxVXdneoDbd7/0IIIYToGEkRlgHGjh3L888/f9DxHj168Pjjj3fAiESyiMZt7n9+HQATivMp6ta+Rf/xrR9h7VyBa8QpaEbSfGSEEEII0Q7kb36R9LaXNtIcjvOjM4cxemBOu/atlCK24gX07J54xny7XfsWQgghRMdLipplIb7M1r0NABT3zsTQ2/ctG/vkJZyGctzDTkJzedq1byGEEEJ0PJlZFklr/c5aXvmohA076+iRGyDQyjvyfRVr1xpiK57HHDAZc8Dkdu1bCCGEEMlBwrJISpt21XHXU6v2/zy8T9stC3coyrGJLHoIPasQ73EXorXzjLYQQgghkoOEZZFUQhGL/3l5PWu21QAwoDCdxmCMaaO6t+s47IqtqFA9nmPPQzPd7dq3EEIIIZKHhGWRVN5cvnt/UJ538iCOH9Ux6xpbJStBNzELh3VI/0IIIYRIDhKWRdL4YG0ZLyzeQd/uaVx0ymB65AY6ZBxW6afE17yKUTgMze3rkDEIIYQQIjlIWBZJIRKzeOL1zQBMPaY7hXkt79He1sKvJDa+MYtGddgYhBBCCJEcJCyLpPDuylKicZvrLxhDvx7pHTYOJ9wIdgyjx1BcQ47vsHEIIYQQIjnIJf4iKXywtoyBhekdGpQB7PLE7LZnzJloutGhYxFCCCFEx5OwLDpcNGZTWhNkcFFmRw8Fu3wLGCZ6bu+OHooQQgghkoCEZdHhXv5wJ0pBUbfUDh2HE6wjvnkxRl4/NKN9N0ARQgghRHKSsCw61KcldSz8qASAvgVpHTYOJ9xI8B//DdEQroFTOmwcQgghhEgucoGf6FD/XLSdtICbX10whvQUT7v37zSUY+1ei1NfBtEgvtm/wuw2sN3HIYQQQojkJGFZdJjVW6vZureBc2cMIC+jbdYzdhorCb18G55J3wNdR/OmYnYbiFIOAOHX/pQIyoCeWShBWQghhBAHkLAsOkQoEufRVz+lR26AaSNbf5c+pRyculLimxejgrVE3rx//22uQVOJ71yBWTAYp74MPbcPriHHY/Yc0erjEEIIIcTRTcKy6BBvrthDfXOMK88egctsvdJ5u3YPkXcexKnZ3eJ94psWgWZg7VyBq3g6nskXoGlaq41BCCGEEJ2HhGXR7pRSvL+6lKG9M+nTyhf1RZc+sz8ou4bNBDue2GCk7ziUFSO+/k2s3WvxHHseKliL0WOYBGUhhBBCtEjCsmh35bUhahqjnDG5T6u2q5SDU7kdANeQ4/Eee94Bt2umG/cxp+E+5rTEgazCVu1fCCGEEJ2PhGXR7j4tqQNgcK+MVm3XqduLijbjPf4HuAZObtW2hRBCCNE1yTrLol3F4jZvf7KXgmw/ua28AoZdvgUAo9uAVm1XCCGEEF2XhGXRrv7ywjr2Vgc5a2rfVq8Vtsu3oPnS0VJzW7VdIYQQQnRdEpZFu9lV0cSabTWcOaUPYwbltXr7dvlmjG4D5II9IYQQQrQaCcui3WzYmahVnjaq9ddVdoJ1qOYajHwpwRBCCCFE65GwLNrNrsomMlM9pAfcrd6201AOgJ7ds9XbFkIIIUTXJWFZtJuS8iaK8lPbpG3VVA2AnprTJu0LIYQQomuSsCzaXDRuc/OjyyirCdGnoG3CstNUDWhogaw2aV8IIYQQXZOEZdHmtu9tYEdZEwCDizLbpA+nuRotkIlmyNLhQgghhGg9EpZFmyupaN7/59be3vozqrFKSjCEEEII0eokLIs2t2VPPVlpHv7682mYRuu/5ZTjYFeXyMV9QgghhGh18p21aDPrd9bicRms2lLNKRN64XYZbdJPdOnTYEUx8vq1SftCCCGE6LokLIs2UVYT5K6nVgHg85icOrGoTfqJb/2I+NrXADC6DWyTPoQQQgjRdUkZhmgTq7YmlnJzmzrnzRxAis/V6n04DRVE3v4rAP6zbpKaZSGEEEK0OplZFm3i442V9MpL4caLx7dZH059GQCuYSdh5LTNzLUQQgghujaZWRatbkdZIyXlTUwd2b1N+3EaKwDwjJ7dpv0IIYQQouuSmWXRahxHEbcd3l6xB4/bYNLQbm3bX0MluP3gCbRpP0IIIYTouiQsi1YRi9vc+X8r2VbaCMCMsYX4PG379nLq9qKn56NpWpv2I4QQQoiuS8owRKtYs61mf1DulZfC2VPbdhk3FWnGLt+M2X1Im/YjhBBCiK5NZpbFEYtbNi99sBOXqXPj98eRk+7DZbbtv8NCr90DysHsM7ZN+xFCCCFE1yYzy+KIvbh4J3uqmhlYmE5BdqDNg7LTVIVTsRXXsJkYeX3btC8hhBBCdG0SlsURsR2HRatL8XkMLjhlcJv3pxyb+NalALgGT2vz/oQQQgjRtUkZhjgi5TUhmsNxLj19CHkZvjbvL7LoIazNH6Bn90LP7NHm/QkhhBCia5OZZXFEdpY3AVDULa1d+nPqEhuR+GZdK6tgCCGEEKLNSVgWR6SkvAm3S6cgy98u/alIE2b/Seje1HbpTwghhBBdm4RlcUQ+3VVHv+7p6Hr7zPKqSBOar31msYUQQgghJCyLb6yuKcqeqiDD+mS1S3/KikI8guaTWWUhhBBCtA8Jy+IbW/hRCbqmMXpgbrv0p8KJ+mjdKzPLQgghhGgfEpbFN2LZDh+uK2dCcT757VSv7DTXAMjMshBCCCHajSwdJ76WkvImdF0jGI4TjlrtNqsMEHn3fwHQAu1T9iGEEEIIIWFZHDbLdrjpkWUA9MpLwecxGdons136VlYU1VSD2XccenavdulTCCGEEELKMMRhW7K+fP+fd1U2c9bUvnjd7fPvLae+HFCYfcfL+spCCCGEaDcysywOi2U7PPP2Vvr3SGf+2cOJxW1y0tt+x77POPWlAOgZ3dutTyGEEEIICcvisOwsbyIYsThpXE/S/O527Vs5DvGN74LLh56e3659C/H/7d15fJTlvffx7z1LkslGCCREtmAEoaDFtoC1LnHBJQKCWz0exVbPU1DbyrEveRlxO48vKYXaI7yqT126cPTYRz3Vkzx6QKUqVRb3VrEWBYSwJWRfZsss9/38EYgEEh1kZu6Z4fP+K3Nlct+/zMXo9/XLNdcFADi2EZbxlVo7g1q1sU6SNH50UVLvbfra5P/v/y3L366cyn+R4eSfLAAASB6SB77SQ89v0q5Gr2acVq6CJHeVo/u2yPK3yznqm3KdeEZS7w0AAEBYRh+RqKm6hi79fXurDIehb40bqh0NXfrn6eM0fcqopNdjedskSZ5z5vHBPgAAkHSEZfSx+q06/feb23sf+4NhGZKmfsOetcKmv01yuqXsPFvuDwAAjm1sHYc+PtvVLknK97glSS+/s0vlZQUalJfc5RcHWN5WGXnFdJUBAIAtCMvoZVqWPq/v1NmnDNeKW87QgXiazFP6DmX52uTIS87BJwAAAIciLKPXP+raFOiO6sRRRTIMQz+smqATRhTq/KnJX6t8gNnVJCN/iG33BwAAxzbWLKPXa+/vVmFelr4zvlSSdObk4Tpzsn2HgJj+9p6dMIZyvDUAALAHnWVIkrpDUX28vVXTJpTK7UqNfxZm8w5JkmPoGFvrAAAAx67USEWw3Ya/NygcMfUtG9cnHyzavlfBv/xBcrrkHFpudzkAAOAYZVtYXrFihX7961/3Pu7s7NS8efNUVVWla665Rk1NTZKkUCikhQsXqqqqSpdeeqm2bdtmV8kZy7Qs/b/123XiqCJNSPIJfQMJvv5bWYEOuSdNl+HOsbscAABwjEp6WO7q6tKiRYv0+9//vs/48uXLNWXKFK1evVpXXnmlFi9eLEl68skn5fF4tHr1ai1atEjV1dXJLjnjbd/bqQ5vSGefMjwltmiLtu2R2fS53BPOUva0K+wuBwAAHMOSHpZfffVVjRkzRtdff32f8bVr12rWrFmSpJkzZ+qNN95QOBzW2rVrdckll0iSpk6dqra2Nu3duzfZZWe0t/+xT06HoZNPSI1dJ6I7P5QkZX3nUhkOPoMKAADsk/SwPGfOHM2bN09Op7PPeGNjo0pKetbLulwu5efnq7W1tc+4JJWUlKihoSGpNWeyQHdE6z6q19QJpcrLcdtdjqJtexX62yo5Bo9gf2UAAGC7hLXtVq9erSVLlvQZq6io0MqVK2O+hsPRf5YfaLw/Q4bkx/zceCopKbDlvkfqhTc/VzAU1ZXnj7e9Zisa1s7/+4AMQyo595+Vl6Kvod2vExKHuc1czG1mYl4zVyrNbcLCclVVlaqqqmJ+fmlpqZqbm1VWVqZIJCKv16uioiKVlpaqqalJ5eU9OyI0NTWptLQ05uu2tHhlmtYR1380SkoK1NTUldR7fl1v/nW3Rpbka7DHZWvN4c/fUfDP/0eSlDP9x/IPmSR/Cr6G6TS3ODLMbeZibjMT85q5kj23Dofxpc3VlNk6rrKyUjU1NZKkVatWacqUKXK73aqsrFRtba0k6b333lN2draGD7fvoIxMYlmW6hq6NHZEoa11mN4WBV97tPexa8REG6sBAAD4Qsp8emrBggWqrq7WjBkzVFBQoAceeECSNHfuXN1zzz2aMWOGsrKytGzZMpsrzRxb93TI3x3R6DL7/tQRqfurAi+vkAxD2addLSsUkJGdZ1s9AAAAB7MtLP/0pz/t87ioqEiPPPLIYc/Lzs7W0qVLk1XWMeODz5r08POblO12auKYYltqsCyzJyhLUlausk6+0JY6AAAABpIynWUk15/f26X8XLfu+cFUDRlkz6Efwb/8rvfrnDN/aEsNAAAAX4awfAx686O92ryzXXPOON62oGx6WxXZskHu8WcpenitOgAAIABJREFU+6zrU+IwFAAAgEMRlo8hbV3dqnnzc735Ub0mHV+si04dbVstoU0vS5KyvjWToAwAAFIWYfkYsnL1Zm36vEWS9L9mTlSW2/kVP5EYVtCr8CevyzX2e3IUxr4NIAAAQLKlzNZxSKy3/t7QG5QlaVBelm21hP6xVoqGlDU59n24AQAA7EBn+Rjx2gd7JEmTji9W5WT79qkO/WOtQu/+Sc4Rk+QsHmlbHQAAALEgLB8D/u+ft2jrng5dMHWU/um8cbbWEvpwlaSetcoAAACpjrCcwSzL0pKnPtDW3R2SpG+eMMTeesyILG+Lsk6ZIdfwb9haCwAAQCxYs5zBGtsCvUH5+qoJth0+coDZ2SiZUTmKOK4cAACkB8JyBvu8vrP36++ML7Gxkh5mW70kyTGYsAwAANIDyzAy2JbdHcp2O/XwrWfJ4bB/L2Ozfa8kyVF0nM2VAAAAxIbOcoYyLUt/3dKkkyqKUyIoS5LZtldGXrEMtz2nBgIAABwpOssZatueDnV4Q7Yuv7AsU5avTUZOvmRGFa3fLMfgEbbVAwAAcKQIyxnqvc1NcjkNTT5haNLvbVmWovWbFXhxqSTJ8BTKMqNSt0/OSeclvR4AAICvi7CcgSzL0gefNWrSmGJ5spM/xdG6vynwyoov6gn0fNAw57yb5aqYkvR6AAAAvi7Ccgba0dClls5uzT6jwpb7hz9/R5LkHHmSPOfdpNDf/kdyuuU+YZot9QAAAHxdhOUM9N7mRjkdhk4Zl/wlGJIUbfhMroppyjnvJhmGoexTv29LHQAAAEeL3TAyjGVZev/TJk0oH6x8j9uW+1uBThn5xTKM1NiFAwAA4OsiLGcQ07T0h9Wb1dgesG8XjEi3FA3LyCm05/4AAABxRFjOIJ/tate6j3pOyfv2OHvCshXokiQ5PAW23B8AACCeCMsZZPPONknSjbMnqTAvy5YarGBPWDYIywAAIAMQljPIZ7vaVT6sQNO+Mcy2Gg5sE8cyDAAAkAkIyxnCsiztavSqvMzejm5vZzmHzjIAAEh/hOUM0dbVLV8wolGl+bbVYPo7FN72tiRDhofOMgAASH/ss5whtu7pkCRbwrLZ2ajw1rcU3fN3Res/latimgx3dtLrAAAAiDfCcgZobPPrmde2qqw4VxXDk9fRNTsaJGeWfP/9v6VunyTJGFSm7O9dk7QaAAAAEulLw/LcuXO/9GCJJ554Iu4F4ci98u4udXhDuvO678jlTM7KGsuMyvdM9WHjuZcskoMlGAAAIEN8aVi+9tprJUlr1qyR1+vV5ZdfLqfTqdraWhUWEohSxY6GLo0dOUjHH5ecOQlv2aDgG3/ofewYUi6zpU5yuAjKAAAgo3xpWL7wwgslSb/73e/09NNPy+Ho6VqeffbZuuqqqxJfHb5SoDuiHfVdmj5lZNLuGXz9sT6PHcUj5TnvRsmdk7QaAAAAkiGmv9m3tbWpu7u797HP51NHR0fCikLsnnj5U5mWpZMrhiTlflY4KO1fmuO56GdyjjxJ2VPmyFF0nBx5g5NSAwAAQLLE9AG/mTNn6vvf/77OP/98WZall156Sd///vcTXRu+Qjhi6m9bmnXmN4/TpOOLk3LPaMNnkmXJc/Ftco08Sa7R30zKfQEAAOwQU1i++eabNWnSJL311luSpOrqalVWVia0MHy1rXs61B2O6lvjSpJ2z8iujyWHU86ycUm7JwAAgF1iCstXXnmlampqNH369ETXgyPw+d6epTDjRg1K2D2sUEDhT9+Uq/wURXZtUvjjV+Qc/g0ZLvZRBgAAmS+msJyTk6OGhgaVlZUluh4cge31XRo22KO8HHfC7hHeulHdG/+o7o1/lCQ5ho5RzjnzEnY/AACAVBJTWA4EAjrvvPNUVlam3Nzc3vEXXnghYYVhYJZlad1H9fpwa7OmfWNYQu9lNm3v+SIrV+4TT1f2lEtlZOV++Q8BAABkiJjC8p133pnoOnAEXtywQ//95naNHTlIl1dWJPRe0abtco46WZ4Lb5XhSM6BJwAAAKkiprA8bdo0tbe3KxAIyLIsRaNR7dy5M9G1YQDvf9akiuGFuuOab3/pCYtHI9LwmbrfekZm625ln3gmQRkAAByTYgrLK1as0GOP9RxE4XQ6FQ6HNXbsWJZh2CASNbW32afzp4xKWFCWpO4NT8ls2Sn3pPPkPokPdgIAgGNTTO3C2tpavf7667rwwgv1yiuv6Be/+IXGjh2b6NrQj/oWvyJRS6OHFSTsHlakW2bLLmVNnqGc0+fKcDgTdi8AAIBUFlNYLi4uVmlpqSoqKrR582bNnj1bdXV1ia4N/WhuD0iSSgd7EnaPaHOdZJlylp6QsHsAAACkg5jCssvl0s6dO1VRUaH33ntPkUhEnZ2dia4N/fAGwpKkAk/itouL7vlEkiFHGX89AAAAx7aYwvL8+fN199136+yzz9aaNWt09tln67vf/W6ia0M/vMGesJyXoLAc2fmhQu/XyFFyvBw5iVvqAQAAkA5i+oDf6NGj9R//8R+SpJqaGtXV1Wn8+PEJLQz98wbCcjkN5WTFfx2xZZoK7j98JOvkC+J+fQAAgHQTU1i++eab5Xa7dcEFF+iCCy7QhAkTEl0XBuALhJXncSdkJ4xo/WZZHfuUc+6Nco/lLwcAAAAxheWXX35ZW7du1WuvvaZ/+7d/U3t7uy644AL97Gc/S3R9OESXP6z8RC3B2Pa25MqWa8y3EnJ9AACAdBPzSRMjRozQ+PHjdfLJJysQCOjll19OZF0YgC8QVn5O/MOyFQoovO1tuY6fIsOVHffrAwAApKOYOsvXXnutduzYoW9/+9s6/fTTdd1112nUqFGJrg398AYjOq44N+7XjezeJIWDcn+jMu7XBgAASFcxdZYrKiqUn5+vjo4OdXR0qKurK9F1oR+maampPaAhg3Lifu3ovm2S0y1naUXcrw0AAJCuYuos33fffZKkzz//XOvWrdPChQvV2tqqjRs3JrQ49NXYHlA4YmpESV7crx1t2CLH0HIZjpj+SQAAABwTYkpGwWBQb7/9tv7yl7/ojTfe0ODBgzV79uxE14ZD7G70SpJGluTH9brhz9+V2fS5sk/9flyvCwAAkO5iCsunnXaaJk+erOnTp2v+/PkaNmxYoutCP+pb/ZKk4UPi21mObN0oI69Y7pMviut1AQAA0l1MYfn1119XTk6O6urqVFpaqmAwqJyc+K+bxZdr93YrL8el7DgeSNL93vOK7PhA7m+cI8MR8+YoAAAAx4SY0lFdXV1vV3nfvn0666yz9MEHHyS6NhyivatbRQXx29bNikYU+vjPkiT3pPPidl0AAIBMEVNYXrp0qVauXKmioiKVlZXpl7/8pRYvXpzo2nCIdm+3ivLjF5ZDH9RKIb88F/1MzuKRcbsuAABApogpLAeDQY0dO7b3cWVlpaLRaMKKQv/avSEV5WfF5VqWaSr08Rq5jp8i56iT43JNAACATBNTWHa5XOro6JBhGJJ6tpBDcoUjpjq8obh1ls32vVI4KFf5t3rnFQAAAH3F9AG/+fPn69prr1Vzc7N+9rOfaf369b17LyM5nvvLNpmWpfGjiuJyvWj9p5IkZ+kJcbkeAABAJoopLP/617/WQw89pHXr1smyLN188819lmUgsUzT0sa/N2jqhFKdVDHkqK9nWZbCf/+zHMWjZAxiG0AAAICBxBSWc3JylJ2drWuuuSbR9aAfH29vUZc/rKkTSuNyPaurSWZ7vbLPuI4lGAAAAF8iprAcCAR03nnnqaysTLm5ub3jL7zwQsIKQ48tu9v10PMfa0hhtk4ZNzQu14w210mSnCXHx+V6AAAAmSqmsHznnXcmug4MYP2mekWipm6cfZJczvgcGmI275AMhxyDR8TlegAAAJkqprA8bdq0RNeBfliWpU92tOnbJ5bohBGD4nJNs2OfQh+9JOfwCTJc8dmGDgAAIFNxvnEK6/SF1NwR1Ilx2gFDkiK7N0lmVDln/jBu1wQAAMhUhOUUtrvZJ0kaVZIXt2tGG7bKyC2SUVASt2sCAABkqpiWYSTCihUr5HA49NOf/lSS9O677+onP/mJysrKJEkTJ07UkiVL1NnZqdtuu027du1ScXGxli9frpKSYyPo7WnqCcsjSvLjds3ovi1ylo1jFwwAAIAYJL2z3NXVpUWLFun3v/99n/FNmzbphhtuUG1trWpra7VkyRJJ0vLlyzVlyhStXr1aV155pRYvXpzskm2zu9Grgly3CvP6X1ts+jtkhbtjvp7pbZXlbZFz2Lh4lQgAAJDRkh6WX331VY0ZM0bXX399n/FNmzZp/fr1mjNnjm688UbV19dLktauXatZs2ZJkmbOnKk33nhD4XA42WXbYkdDp8rLCgb8vu8/F8j7h/mK7tt62PesoFdW0NtnLLr3E0mSs4ywDAAAEIukh+U5c+Zo3rx5cjqdfcYLCgp03XXXqaamRpWVlbr11lslSY2Njb3LLlwul/Lz89Xa2prsspNu5ep/aHeTT2PKCvv9vmVGer/2194vSer+24vy/eluRZt2yP/CL+R9coGiLTslSaEPVym49rdyFI+UY8joxP8CAAAAGSBha5ZXr17du5TigIqKCq1cubLf59933329X1999dX61a9+pa6urn6f63DEnvGHDInfet8jUVIycEc4Fm982NNZn3bScf1eK9zRqIP7xkOH5Gr7ezWSGZG/5j7JMiVJ7r0fqHjceO38+BU5C4dqxD/fJdeg+O2ucSw62rlF6mJuMxdzm5mY18yVSnObsLBcVVWlqqqqmJ5rmqYeffTRwzrOLpdLpaWlam5uVllZmSKRiLxer4qKYg97LS1emaZ1xPUfjZKSAjU19R/0Y2GalhyGoVMnlmrE4Jx+rxVp2C1Jcg4bp+i+Ldr1xyXSgW6zZcp1wqmyfG3q2FijQLelqK9dORf8VG2hHOkoajvWHe3cInUxt5mLuc1MzGvmSvbcOhzGlzZXU2LrOIfDoTVr1ujll1+WJNXU1Gjy5MnyeDyqrKxUTU2NJGnVqlWaMmWK3G63neUmXLu3W6ZlaezIon53rbAsS93r/1OS5Bw9WZIUqfurnCNP6n2O+8TT5T7pfMnhVOiDWskw5Bo+MTm/AAAAQIawbeu4Qy1dulR33323Hn74YRUXF2vZsmWSpAULFqi6ulozZsxQQUGBHnjgAZsrTbzWrp4dLoYUZvcZjzZ+ru53n1N0z997x5zHndj7tWf6j+X9z3+VIt1yllTIyMmXY/Bw+Z+/V47BI2VkeZLzCwAAAGQI28Lygf2VDxg3bpyefvrpw55XVFSkRx55JFll2S4SNfXzJ9+XJBUX5PT5XvjTN/sEZdeJp8tZWiHXiacra8LZMrI8yp11h6INn8rI6flzgnPwCHku+pkMd99rAQAA4KulTGcZPXbU96zRMQxpWHFun+9F922Rc+RJiu7+WJKUc+YPZThc8pz9o97nOEvGyFkyps/PuUaw/AIAAODrSIk1y/jCJ3WtMiStuOVMuV1fTI/pbZXZukfOYePkGFIuGU4Zzsxeuw0AAGA3Ossp5sOtLRpzXKHyPV8EYcuMKPjmSskw5D7xe8qaXCVZyd3hAwAA4FhEZzmFtHu7tb2+U6eMG9pnPLrnE0V3faSsyRfLUVAiw5Ulw509wFUAAAAQL4TlFPLh1mZJ0rcOCctmW88BJVnfvCjpNQEAABzLCMsp5IPPmjV0UI5GDM3rM26218vIKejd4QIAAADJQVhOEfUtPn28vUWnThzW5yCS8LZ3FN68VsagYTZWBwAAcGwiLKcA07L04LMfyuV0qPKU4b3jViig4OuPSVkeZU2otLFCAACAYxO7YaSAhha/mjuC+ufp4zR00Ben7EWbd0hmRJ5zb5Fr9DftKxAAAOAYRWc5BXy+t1OSNHFMcZ/xaON2SZKj9Pik1wQAAADCckrYvLNNeTkulQ3pe2Kf2fS5jIISOXIKbKoMAADg2EZYtlkkaupvW5p1ytihchz0wT5JijZtl7OErjIAAIBdCMs2q2vokr87osljD9lb2d8uy9siZ2mFTZUBAACAD/jZbEdDlySpYnhh75jZ0aDAKw9JhkPOkSfZVRoAAMAxj86yzeoaulSQ69bggi+Or47s+KvMtt3KPv1aOYtH2lgdAADAsY2wbLMdDV0qLyvocxCJ6WuV3DnKmniujZUBAACAsGyjUDiqvc0+jSnru9uF5W2VI794gJ8CAABAshCWbbSrySvTslQ+rLDPuOlrlZFHWAYAALAbYdlGe5t8kqRRpXl9xi1vC51lAACAFEBYtlFzR1CGIRUX5vSOWaGArECnjIISGysDAACARFi2VXNHUMUF2XI5v5iGaOsuSZJzyCi7ygIAAMB+hGUbNXcENGSQp8+Y2bJTkuQoHm1HSQAAADgIYdlGzR1BDR2U02fMbNkpIztfRt5gm6oCAADAAYRlm/iDEbV1dausOLfPeLRllxxDR/fZdxkAAAD2ICzbZHeTV5I0qjS/d8wyozJbd8kxhCUYAAAAqYCwbJNdjYeHZbNjnxSNyFnMh/sAAABSAWHZJvta/crOcmpwQXbvmOVtkSQZBUPtKgsAAAAHISzbpK2rW8UF2X3WJlu+NkmSgw/3AQAApATCsk3avD1h+WCmvycsG7lFdpQEAACAQxCWbdLW1a2iQ8Ky5Wvv2TbOlWVTVQAAADgYYdkGUdNUu7dbgwsO2WPZ18b+ygAAACmEsGyDDm9IlqU+H+6TJMvXIiO/2KaqAAAAcCjCsg06/SFJ0qC8L5ZbWJYls6NRjsJSu8oCAADAIQjLNuj09YTlwoPDcqBDinTLUTjMrrIAAABwCMKyDTp9YUl9w7LZsU+S5BhEZxkAACBVEJZtcGAZRmGuu3fMbK+XJDkGldlSEwAAAA5HWLZBpy+kLLdDOVmu3jGzeYeU5eH0PgAAgBRCWLZBpz+kwty+eylHm7bLWXK8DIMpAQAASBUkMxt0+kJ9P9xnWTJb98gxZLSNVQEAAOBQhGUbdPrCfTvL3T7JjMjBgSQAAAAphbBsg05/386yGeiUJBmeQrtKAgAAQD8Iy0lmmpa6/CEV5n2xE4YV6JAkGZ5BdpUFAACAfhCWk8wbDMuy1GcZhkVnGQAAICURlpOs/9P7CMsAAACpiLCcZL1h+dDOsuGQkZNvV1kAAADoB2E5ybyBnqOuCw4+va+zUUZ+MXssAwAApBjSWZL5ghFJUm7OQWG5ba8cg0fYVRIAAAAGQFhOMn+wp7Ocl9Nz1LVlRmV21MtRNNzOsgAAANAPwnKS+YIRuZwOZbmdkiSrq0mKRuQcTFgGAABINYTlJPMHw71dZUmKtu2VJJZhAAAApCDCcpL5ghHlefquV5YkR9FxdpUEAACAARCWk8wXCCv3oM6y2bZHRl6xjCyPjVUBAACgP4TlJPMHI8rLPigst++Vg/XKAAAAKYmwnGS+YKR32zjLMmW21bNeGQAAIEURlpPMGwj3HkhidbVI0RCdZQAAgBRFWE6iYCii7nBUhXk9R12b7XskSU72WAYAAEhJhOUk6vSFJEmFufvDcu+2cYRlAACAVERYTqJOX8/pfQc6y9G2PTJyi2Rk59lZFgAAAAZAWE6ijv2d5UF5X3SW6SoDAACkLsJyEnX69y/DOBCWOxrkGMRhJAAAAKmKsJxEB9YsF+S6ZUVCUiggI6/I5qoAAAAwEMJyEnX6Qsr3uOVyOmQFuyRJhqfQ5qoAAAAwEMJyEnX6Qr1LMCx/hyTJQVgGAABIWYTlJOrwh1R44ECSYKckOssAAACpLOlh+f3339fll1+u2bNn6wc/+IH27Ok5mKOzs1Pz5s1TVVWVrrnmGjU1NUmSQqGQFi5cqKqqKl166aXatm1bskuOm76dZcIyAABAqkt6WF64cKEWL16s2tpazZo1S/fff78kafny5ZoyZYpWr16tK6+8UosXL5YkPfnkk/J4PFq9erUWLVqk6urqZJccNweHZZPOMgAAQMpLalgOhUJasGCBJkyYIEkaP3686uvrJUlr167VrFmzJEkzZ87UG2+8oXA4rLVr1+qSSy6RJE2dOlVtbW3au3dvMsuOi1A4qmAo2rvHsuXvkNw5MlzZNlcGAACAgSQ1LGdlZWn27NmSJNM09dBDD2n69OmSpMbGRpWUlEiSXC6X8vPz1dra2mdckkpKStTQ0JDMsuOiy99zel/B/qOuLW+rHPnFdpYEAACAr+BK1IVXr16tJUuW9BmrqKjQypUrFQqFVF1drUgkovnz5w94DYej/yw/0Hh/hgzJj/m58VRSUtDncUd3VJI0sqxQJSUF2tPdIffgksOeh9THnGUu5jZzMbeZiXnNXKk0twkLy1VVVaqqqjps3Ofz6aabblJRUZF+85vfyO3u2R2itLRUzc3NKisrUyQSkdfrVVFRkUpLS9XU1KTy8nJJUlNTk0pLS2Ouo6XFK9O04vNLxaikpEBNTV19xnbtbZckmeGompq6FGpvkqtw+GHPQ2rrb26RGZjbzMXcZibmNXMle24dDuNLm6u2fMCvvLxcK1asUFZWVu94ZWWlampqJEmrVq3SlClT5Ha7VVlZqdraWknSe++9p+zsbA0fPjzZZR817/5lGPm5blnRiKxAp4y8wTZXBQAAgC+TsM5yfz755BO9+uqrGjt2rObMmSOpp6P8+OOPa8GCBaqurtaMGTNUUFCgBx54QJI0d+5c3XPPPZoxY4aysrK0bNmyZJYcN12B/WHZ45blb5NkyWDNMgAAQEpLalieOHGiPv30036/V1RUpEceeeSw8ezsbC1dujTRpSWc1x+WYUi5OS6ZDa2SJEceYRkAACCVcYJfkngDYeV73HIYhixfmyTRWQYAAEhxhOUk6dofliXJ9NJZBgAASAeE5STx+kMq2B+WLV+r5PbIyPLYXBUAAAC+DGE5SbyBsPIPHEjia5Ujn50wAAAAUh1hOUn6LMMIdMrILbK5IgAAAHwVwnISWJYlrz+sgtz9yzACnTJyUudkGgAAAPSPsJwEwVBUUdPq7SxbgS4ZnkKbqwIAAMBXISwnQZ8DSaJhKRygswwAAJAGCMtJcOCo64Jct6ygV5LoLAMAAKQBwnISeAMhSVK+J0tWoFOS6CwDAACkAcJyEnTt7yzn57plBbskSYaHsAwAAJDqCMtJ4N2/ZrnA45bZ1SyJ0/sAAADSAWE5CbyBsJwOQzlZTpkdDZLTJSOfsAwAAJDqCMtJ0OUPKz/XLcMwZHXsk6OwTIbBSw8AAJDqSGxJ4A2EVXDg9L6OBjkGDbO5IgAAAMSCsJwEXn/oi6OufW0swQAAAEgThOUk6AqElZ+bJSsakcJBto0DAABIE4TlJDiwDMPq3n8gSU6+zRUBAAAgFoTlBDMtS95AuOeo6wOn92UTlgEAANIBYTnB/MGILOvAgSR0lgEAANIJYTnBDj6QpPf0PsIyAABAWiAsJ5j34KOuu32SWIYBAACQLgjLCdYVCEmSCjxZdJYBAADSDGE5wXo7yx63zLa9MvIGy3Bl2VwVAAAAYkFYTrA2b7eknmUYZnOdnEPH2FsQAAAAYkZYTrCtuzs0YmieshSR2V4vx5DRdpcEAACAGBGWEyhqmtqyu0PjRxfJ9LZIsuQoOs7usgAAABAjwnICdfrC6g5HNaIkX5avTZJk5BbZXBUAAABiRVhOoD57LPvbJUmOvMF2lgQAAIAjQFhOIK+/Z9u4fI9bpn9/ZzmPzjIAAEC6ICwnkDcYkdQTli1fm5SVK8OVbXNVAAAAiBVhOYEOLMPI87hl+drlYL0yAABAWiEsJ9DByzCsYJcMT6HNFQEAAOBIEJYTyBuIKDvLKbfL0ROWOeYaAAAgrRCWE8gbCKnA45YkWUEvnWUAAIA0Q1hOIG8g0rNe2TRldXvpLAMAAKQZwnICeQPhnvXKIZ9kWTJyCuwuCQAAAEeAsJxAB5ZhWMEuSSIsAwAApBnCcgL1LsMIEJYBAADSEWE5QSJRU4HuSM/pfW17JEmOwhKbqwIAAMCRICwniO+g0/uiDZ/J8AySUUBYBgAASCeE5QQ5cHpfvsetaOPncpaNk2EYNlcFAACAI0FYThDfgbCc45Lla5NRMNTmigAAAHCkCMsJ0uXvCcsF7qgUDcuRO8jmigAAAHCkCMsJ4gv2hOU8IyBJMjyEZQAAgHRDWE6QA2uWc02fJMIyAABAOiIsJ4jXH5bb5ZAztH+PZZZhAAAApB3CcoIcOOpawU5JhGUAAIB0RFhOkANhueeoa0NGdp7dJQEAAOAIEZYT5Iuw7JOyc2UYvNQAAADphgSXIN5AWHket6xun4zsfLvLAQAAwNdAWE4QbyCsAo9bVreXJRgAAABpirCcAKZpyRc8qLOcQ1gGAABIR4TlBPAFw7Is9axZ7vbRWQYAAEhThOUE6PSFJGn/MgzCMgAAQLoiLCdA1/6wnJfjlLr9hGUAAIA0RVhOgE7//s6yKyLJIiwDAACkKcJyAnR6e8JyvuGTJBm5g+0sBwAAAF8TYTkBuvZ3lnOjB8IyR10DAACkI8JyAnT5Q3IYhtzhTkmSI7fI5ooAAADwdRCWE8DrDyvP45Ll75AkGYRlAACAtERYToAuf0h5OW5Z/nbJ7ZHhzra7JAAAAHwNhOUE6PKH9neW2+XIo6sMAACQrgjLCdDlDys/xy3T384SDAAAgDTmSvYN33//ff385z9XJBJRUVGRfv7zn2vEiBF699139ZOf/ERlZWWSpIkTJ2rJkiXq7OzUbbfdpl27dqm4uFjLly9XSUlJsss+Il5/SGWDPbK6OuQYNtbucgAAAPA1Jb2zvHDhQi1evFi1tbWaNWuW7r//fknSpk2bdMMNN6i2tla1tbVasmSJJGn58uWaMmWKVq9erSuvvFKLFy9OdslHrMsfVl62S5a/jW3jAAAA0lhSw3IoFNKCBQs0YcIESdL48eNVX18vqScsr1+/XnPmzNHP/NccAAAOo0lEQVSNN97YO7527VrNmjVLkjRz5ky98cYbCofDySz7iESipgLdEQ3KjkjRiBwcSAIAAJC2khqWs7KyNHv2bEmSaZp66KGHNH36dElSQUGBrrvuOtXU1KiyslK33nqrJKmxsbF32YXL5VJ+fr5aW1uTWfYR8QUjkqQiZ0ASB5IAAACks4StWV69enXvUooDKioqtHLlSoVCIVVXVysSiWj+/PmSpPvuu6/3eVdffbV+9atfqaurq99rOxyxZ/whQ/K/RvVfX9Dsqbksz5QkDR5+nDwlBUmtAYlVwnxmLOY2czG3mYl5zVypNLcJC8tVVVWqqqo6bNzn8+mmm25SUVGRfvOb38jtdss0TT366KOaN2+enE7nF8W5XCotLVVzc7PKysoUiUTk9XpVVBT7DhMtLV6ZphWX3ykWu/a2S5KsoFeS1BEw5G3qP/Qj/ZSUFKiJ+cxIzG3mYm4zE/OauZI9tw6H8aXNVVs+4FdeXq4VK1YoKyurpwiHQ2vWrNHLL78sSaqpqdHkyZPl8XhUWVmpmpoaSdKqVas0ZcoUud3uZJcdM2+gZz21xwhJkowsj53lAAAA4Cgkdeu4Tz75RK+++qrGjh2rOXPmSJJKS0v1+OOPa+nSpbr77rv18MMPq7i4WMuWLZMkLViwQNXV1ZoxY4YKCgr0wAMPJLPkI+YL9KxZzlFPaDaycu0sBwAAAEchqWF54sSJ+vTTT/v93rhx4/T0008fNl5UVKRHHnkk0aXFjS/YE5KzrG5ZMiQ6ywAAAGmLE/zizBcMy+Ew5DKDkjtHhsFLDAAAkK5IcnHmDUSU73HLCvllZLMEAwAAIJ0l/bjrTHfS8cUaXpIvNb3Ph/sAAADSHGE5zr59YolKSgpU9/sAH+4DAABIcyzDSBAr5JcIywAAAGmNsJwgVtArIyfP7jIAAABwFAjLCWBZpix/hxy5sZ80CAAAgNRDWE4A098lWVEZhGUAAIC0RlhOgIi3TZIIywAAAGmOsJwAUcIyAABARiAsJ8CBsOzIHWRzJQAAADgahOUEiPo6JEmGp9DmSgAAAHA0CMsJEA16JYdTcmXbXQoAAACOAmE5AcygX0Z2ngzDsLsUAAAAHAXCcgKYQS+n9wEAAGQAwnICmEGfjGzCMgAAQLojLCeAGfTKyOaoawAAgHRHWE6AaNAnI4uwDAAAkO4IywnQ01lmGQYAAEC6IyzHmWVZvbthAAAAIL0RluMt5Jcsk7AMAACQAQjLcWYFOiVxeh8AAEAmICzHmdkblgfZXAkAAACOFmE5zqxAhyTJyKWzDAAAkO4Iy3Fm+eksAwAAZArCcpxZwU7JcMjIzre7FAAAABwlwnKcWUGvnLkFMhy8tAAAAOnOZXcBmcZ94hnKH3+KAnYXAgAAgKNGWI4zZ2mF8ksKFGjqsrsUAAAAHCXWCgAAAAADICwDAAAAAyAsAwAAAAMgLAMAAAADICwDAAAAAyAsAwAAAAMgLAMAAAADICwDAAAAAyAsAwAAAAMgLAMAAAADICwDAAAAAyAsAwAAAAMgLAMAAAADICwDAAAAAyAsAwAAAAMgLAMAAAADICwDAAAAA3DZXUCiORzGMXVfJB5zm7mY28zF3GYm5jVzJXNuv+pehmVZVpJqAQAAANIKyzAAAACAARCWAQAAgAEQlgEAAIABEJYBAACAARCWAQAAgAEQlgEAAIABEJYBAACAARCWAQAAgAEQlgEAAIABEJbj7IUXXtDFF1+s888/X0899ZTd5eAIPfTQQ5oxY4ZmzJihZcuWSZI2bNigWbNm6YILLtCDDz7Y+9x//OMfuvzyy3XhhRfqzjvvVCQSsatsHIGlS5equrpa0sBzuHfvXl1zzTW66KKLdNNNN8nn89lZMr7Ca6+9pssuu0wXXXSR7r//fkm8bzNFbW1t73+Tly5dKon3bTrzer2aOXOmdu/eLenI36e2zbGFuGloaLDOOeccq62tzfL5fNasWbOsLVu22F0WYrR+/Xrrqquusrq7u61QKGRdd9111gsvvGBVVlZaO3futMLhsHXDDTdYa9eutSzLsmbMmGH99a9/tSzLsu644w7rqaeesrN8xGDDhg3Wqaeeat1+++2WZQ08h/PmzbNefPFFy7Is66GHHrKWLVtmT8H4Sjt37rTOOOMMq76+3gqFQtbVV19trV27lvdtBvD7/dbUqVOtlpYWKxwOW1dccYW1fv163rdp6m9/+5s1c+ZMa9KkSdauXbusQCBwxO9Tu+aYznIcbdiwQd/97ndVVFSk3NxcXXjhhXrppZfsLgsxKikpUXV1tbKysuR2u3XCCSdox44dKi8v16hRo+RyuTRr1iy99NJL2rNnj4LBoE455RRJ0mWXXcZcp7j29nY9+OCDuvHGGyVpwDkMh8N69913deGFF/YZR2pas2aNLr74YpWVlcntduvBBx+Ux+PhfZsBotGoTNNUIBBQJBJRJBKRy+XifZumnn32Wd17770qLS2VJH300UdH9D61c45dSbnLMaKxsVElJSW9j0tLS/XRRx/ZWBGOxLhx43q/3rFjh1atWqW5c+ceNqf79u07bK5LSkq0b9++pNaLI3PPPffo1ltvVX19vaTD368H5rCtrU35+flyuVx9xpGa6urq5Ha79S//8i9qamrSOeeco3HjxvG+zQD5+flasGCBqqqqlJOTo2nTpsntdvO+TVOLFy/u87i/zPRl71M755jOchxZlnXYmGEYNlSCo7FlyxbdcMMNuv322zV69OjDvm8YBnOdZv7rv/5Lxx13nE477bTesYHmkLlNL9FoVBs3btQvf/lLPfvss9q0aVPvesiDMbfpZ/PmzXruuef0+uuva926dXI4HFq/fv1hz2Nu09OR/jfYzjmmsxxHw4YN03vvvdf7uLGxsffPDUgP77//vm655RYtWrRIM2bM0DvvvKPm5ube7x+Y02HDhvUZb2pqYq5T2KpVq9TU1KTZs2ero6NDfr9fhmH0O4fFxcXyer2KRqNyOp3MbYobOnSoTjvtNBUXF0uSzjvvPL300ktyOp29z+F9m57WrVun0047TUOGDJHU82f33/3ud7xvM8Sh78evep/aOcd0luPoe9/7njZu3KjW1lYFAgG98sorOuuss+wuCzGqr6/Xj3/8Yz3wwAOaMWOGJGny5Mnavn276urqFI1G9eKLL+qss87SiBEjlJ2drffff1+SVFNTw1ynsD/84Q968cUXVVtbq1tuuUXnnnuulixZ0u8cut1uTZkyRatWreozjtR0zjnnaN26ders7FQ0GtWbb76piy66iPdtBpgwYYI2bNggv98vy7L02muvadq0abxvM8SR/v/Vzjk2rP762vjaXnjhBT366KMKh8O64oor9KMf/cjukhCj+++/X88991yfpRf/9E//pDFjxmjJkiXq7u5WZWWl7rjjDhmGoc2bN+uuu+6Sz+fTxIkTtWTJEmVlZdn4GyAWzz//vN555x394he/GHAO9+zZo+rqarW0tOi4447Tv//7v2vQoEF2l44B/OlPf9LKlSsVDod1+umn66677tLbb7/N+zYDPPbYY3r++efldrt18skn695779X27dt536axc889V0888YRGjhypjRs3HtH71K45JiwDAAAAA2AZBgAAADAAwjIAAAAwAMIyAAAAMADCMgAAADAAwjIAAAAwAMIyAGSYH/3oR9q6dWtcrjV//nw9//zzcbkWAKQjTvADgAzz+OOP210CAGQMwjIApLjXXntNv/nNbxQOh5WTk6Pbb79d69at05YtW9Tc3KyWlhZNmDBBixcvVn5+vs4991ytWLFCFRUVuuOOO1RXVyeHw6FJkybpvvvuk8Ph0DPPPKMnn3xSDodDQ4cO1d13363jjz9e+/btU3V1tRobGzV8+HC1tLT01rFt2zYtXrxY7e3tikajmjt3rq644gobXxkASDzCMgCksB07dujBBx/UE088ocGDB2vLli26/vrrNWvWLH344Yd67rnnVFxcrIULF+rhhx/W7bff3vuza9askc/nU21traLRqO69917t2rVLe/fu1W9/+1s988wzKi4u1vPPP68f//jH+p//+R/dd999mjx5sv71X/9VdXV1mjNnjiQpEonolltu0bJlyzRp0iR1dXXpqquu0tixY3XKKafY9fIAQMKxZhkAUtj69evV2NioH/7wh5o9e7Zuu+02GYahnTt36qKLLtLQoUPlcDh0xRVXaN26dX1+9jvf+Y62bt2quXPn6rHHHtMPfvADlZeX680339TFF1+s4uJiSdJll12mffv2affu3dqwYYMuu+wySVJ5eblOPfVUST2hfefOnVq0aJFmz56ta6+9VsFgUJ988klyXxAASDI6ywCQwkzT1Gmnnably5f3jtXX1+uZZ55RKBTq8zyHo2//Y9SoUVqzZo3efvttvfXWW7r++ut11113ybKsw+5jWZYikYgMw+jzfZer538T0WhUhYWFqq2t7f1ec3OzCgoK4va7AkAqorMMACnsu9/9rtavX69t27ZJkv7yl7/okksuUXd3t1599VV1dXXJNE09++yzOuecc/r87B//+EfdcccdOuOMM7Rw4UKdccYZ2rJli8444wytWrVKra2tkqTnnntORUVFKi8v15lnnqlnnnlGkrR37169/fbbkqTjjz9e2dnZvWG5vr5eM2fO1Mcff5yslwIAbGFY/bUYAAApY/Xq1XrkkUdkWZZcLpcWLVqkjRs36q233lI0GlVbW5umTp2qu+66Szk5Ob0f8DvhhBO0aNEiffrpp/J4PBo+fLgWL16sQYMG6amnntLTTz8t0zRVXFyse+65R+PGjVNra6vuuOMO7dy5U2VlZYpEIrr00kt12WWXafPmzb0f8ItEIrruuut09dVX2/3yAEBCEZYBIA39+te/Vltbm+655x67SwGAjMYyDAAAAGAAdJYBAACAAdBZBgAAAAZAWAYAAAAGQFgGAAAABkBYBgAAAAZAWAYAAAAGQFgGAAAABvD/AchTN/fZEvLQAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 842.4x595.44 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set(rc={'figure.figsize':(11.7,8.27)})\n",
    "sns_plot = sns.lineplot(x='episode', y='reward', hue='type',data = data[data.episode<1000])\n",
    "sns_plot.figure.savefig(\"tax1v2.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Episode 20000/20000 || Best average reward 9.246\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.322\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.258\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.184\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.232\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.383\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.288\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.322\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.344\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.192\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.288\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.286\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.338\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.384\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.181\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.438\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.239\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.257\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.373\n",
      "\n",
      "Episode 20000/20000 || Best average reward 9.382\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Who wins more often? SarsaMax of Expectation Sarsa?\n",
    "sarsa_max_wins = 0\n",
    "for i in range(10):\n",
    "    env = gym.make('Taxi-v2')\n",
    "    agent = Agent(**optimal_sarsa_max)\n",
    "    avg_rewards_sarsamax, best_avg_reward_sarsamax = interact(env, agent)\n",
    "\n",
    "    agent = Agent(**optimal_exp_sarsa)\n",
    "    avg_rewards_exp_sarsa, best_avg_exp_sarsa = interact(env, agent)\n",
    "    \n",
    "    if best_avg_reward_sarsamax>best_avg_exp_sarsa: sarsa_max_wins += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sarsa max wins in 30.0% of cases\n"
     ]
    }
   ],
   "source": [
    "print(f'Sarsa max wins in {sarsa_max_wins/10*100}% of cases')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Sarsa Max got 9.493"
   ]
  }
 ],
 "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": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}