{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "ee4c71e9-e325-494b-a787-ac1d5d27ebe3",
   "metadata": {},
   "source": [
    "# When to replace your machine?"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6a7f2ce5-b90a-46e7-8968-e98f96909d90",
   "metadata": {},
   "source": [
    "Your (somewhat strange looking) machine's annual maitenance cost $c(x)$ evolves with its level of wear $x$. At some point, it becomes cheaper to by a new one, for a fixed price $C$. \n",
    "But the evolution of the level of wear is random: from one year to the next, it increases by a random amount modelled by an exponential distribution of parameter $\\lambda$. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "4c9770dd-e5e9-4b89-9b01-5bbca2419949",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt #%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "182fcc4f-c75c-4c02-b68c-587e00f35a4d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gamma = 0.9 # discount factor\n",
    "replacementCost = 150\n",
    "def maintenanceCost(x):\n",
    "    y = 4 * x\n",
    "    if 2 <= x <= 4 or 8 <= x <= 11 :\n",
    "        y += 80\n",
    "    elif 14 <= x <= 18:\n",
    "        y += 50\n",
    "    return y\n",
    "\n",
    "meanWearIncrease = 1\n",
    "randomWearIncrease = lambda: np.random.exponential(1)\n",
    "\n",
    "actions = [\"Keep\", \"Replace\"]\n",
    "\n",
    "xx = np.linspace(0, 20, 100)\n",
    "plt.plot(xx, [maintenanceCost(x) for x in xx]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "3fa848e9-528e-4aae-bef6-3ab9ea848965",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3171.0831454026134"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "randomPolicy = lambda x: np.random.choice([\"Keep\", \"Replace\"])\n",
    "\n",
    "def evaluatePolicy(pi):\n",
    "    x = 0\n",
    "    totalCost = 0\n",
    "    discount = 1\n",
    "    T = int(4/(1-gamma))\n",
    "    for t in range(T):\n",
    "        a = pi(x)\n",
    "        totalCost += maintenanceCost(x) if a == \"Keep\" else replacementCost\n",
    "        x = (x if a == \"Keep\" else 0.) + randomWearIncrease() \n",
    "        discount *= gamma\n",
    "    return(totalCost)\n",
    "\n",
    "evaluatePolicy(randomPolicy)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7254393c-c359-41e2-aec1-e99e943068b4",
   "metadata": {},
   "source": [
    "# Approximate planning using a simulator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "16404c62-e4f8-4da2-9f47-93cd5751f539",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f9745b312b0>]"
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAigAAAGdCAYAAAA44ojeAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAABicElEQVR4nO3deXhTZdo/8G/2tKULbekGbSm70IJQlE2HHUUWR1REGUVlGJ1xHHmBdxTn9SfMy4jjjDoOjuvgAIpTXwcZHXED2VREawtYFqFAgSIthUKbrmmanN8fpydtaLOcNGlyku/nunJRknOSJ4Qmd577fu5HJQiCACIiIqIgog70AIiIiIiuxACFiIiIgg4DFCIiIgo6DFCIiIgo6DBAISIioqDDAIWIiIiCDgMUIiIiCjoMUIiIiCjoaAM9AG/YbDacO3cO0dHRUKlUgR4OEREReUAQBNTU1CAtLQ1qtes5EkUGKOfOnUN6enqgh0FEREReKC0tRa9evVweo8gAJTo6GoD4BGNiYgI8GiIiIvKEyWRCenq6/XPcFUUGKFJaJyYmhgEKERGRwnhSnsEiWSIiIgo6DFCIiIgo6DBAISIioqDDAIWIiIiCDgMUIiIiCjoMUIiIiCjoMEAhIiKioMMAhYiIiIIOAxQiIiIKOgxQiIiIKOjIClBWrFgBlUrlcElJSbHffu+997a7ffTo0Q73YTab8fDDDyMxMRFRUVGYPXs2zp4965tnQ0RERCFB9gzKkCFDUFZWZr8UFRU53H7jjTc63P7RRx853L548WJs3rwZeXl5+PLLL1FbW4uZM2fCarV27pkQERFRyJC9WaBWq3WYNbmSwWBwent1dTXWrl2LN998E1OmTAEAvPXWW0hPT8e2bdtwww03yB0OERGFqQOlVfjgwDnYBEH2uaOy4nFjdqofRkW+IjtAKS4uRlpaGgwGA0aNGoWnnnoKffr0sd++c+dOJCUlIS4uDuPHj8cf/vAHJCUlAQAKCgpgsVgwbdo0+/FpaWnIzs7Gnj17nAYoZrMZZrPZ/neTySR32EREFELqzM34+YbvcKHG7P7gDvzjq1NYMCYT/zNzMHQalmMGI1kByqhRo7BhwwYMGDAA58+fx6pVqzB27FgcOnQICQkJmD59Om6//XZkZmaipKQETzzxBCZNmoSCggIYDAaUl5dDr9eje/fuDvebnJyM8vJyp4+7evVqrFy50rtnSEQeabRYcbyi1u1x8VF6pMVFdMGIiJx7ddcJXKgxo2dcBH46PE3WuZfqmvDPb0ux/uvTOH6hFn+7awTiIvV+Gil5SyUIXsyNtairq0Pfvn3x29/+FkuWLGl3e1lZGTIzM5GXl4c5c+bg7bffxn333ecwGwIAU6dORd++ffHKK690+DgdzaCkp6ejuroaMTEx3g6fiFoIgoBbX96DwjNVHh3/8KR++K8pA6BWq/w7MKIOnKtqwKRnd6LRYsMrPxvhVarms0PlWPzOftQ3WdE7IRJ/XzAS/ZKi/TBaastkMiE2Ntajz2/ZKZ62oqKikJOTg+Li4g5vT01NRWZmpv32lJQUNDU14fLlyw6zKBUVFRg7dqzTxzEYDDAYDJ0ZKhG5sPPoBRSeqYJWrUKPaOe/a4IAlJsasWb7cRw7X4Pn5l6NKEOn3kaIZPvzp0fRaLHh2t7xuGGI85pIV6YNScF7vxqLn6//Dqcq6zH7xa/QOyHK5Tk6rRpLpg7A+AE9vHpMkqdT7yxmsxlHjhzB9ddf3+HtlZWVKC0tRWqqGN3m5uZCp9Nh69atmDt3LgBxluXgwYN45plnOjMUIvKSIAj463bxS8TC67Kw/KarXB6/qeAslr9XhE8PncetL+/B3xeMRK/ukV0xVCJ8f7YK7+37EQDwPzOvgkrl/SzeoJQYvP/QOPxyYyG+LbmEw2Xu6xtXfnAI25aM5+xhF5AVoCxbtgyzZs1CRkYGKioqsGrVKphMJixYsAC1tbVYsWIFbr31VqSmpuLUqVN4/PHHkZiYiFtuuQUAEBsbi4ULF2Lp0qVISEhAfHw8li1bhpycHPuqHiLqWntOVGLfmSoYtGosvD7L7fG35vZC78QoPPBmAX4or8HNL36FX/ykj9tCw/goPQYkR6NvUhQMWk2nxnyx1oydRy9gxw8V+KL4AuqaXLcpUAH42ehMrJg9pFOPS4ElCAJWfXgEADBneE8M7RXX6ftM6GbA2z8fhX2lVWhw8f/IKgj4zT/34eTFOuw4WoHJVyV3+rHJNVkBytmzZ3HnnXfi4sWL6NGjB0aPHo29e/ciMzMTDQ0NKCoqwoYNG1BVVYXU1FRMnDgR77zzDqKjW/N6zz//PLRaLebOnYuGhgZMnjwZ69atg0bTuTcsIvLOmpbZkzuvzUBStNGjc3Izu+ODX4/Dog3f4dA5E1Z//IPHj6dVq5CVGIX+yd1g9CJQKamsw/7SKsitnttSVMYARSZPSxQ7M4shx6eHyvHtqUsw6tRYdsNAn92vVqPGNb3j3R5317UZeHX3Sfz9ixIGKF2gU0WygSKnyIaInMs/dQm3v/I1dBoVdv33RNmrc+qbmvHyzhM4XVnv8jgBwPnqRhwpN6GmsbkTI241JC0GkwclYeKgJPR0Me7Sy/W49eWvEW3UomgFey15as3nxXh+2zHY3HxC6LVqjO6TgClXJWHSoCRZ6T5BEFBRY8aRMhOOV9Si2c2DbfzmNEovNeA3k/phyTTfBSieOlfVgOuf2QGrTcCHD1+H7J6xXT4GpeuyIlkiUrY1248DAG7LTfdq6XCkXoulMj4oBEFAWXUjjpbX4OTFOlhtNtmPGRuhw08G9EBqrGfjlT7zGi3sVi3HlqIyt8EJADQ127D72AXsPnYB/+/9QxiUEo2BKdFwNaciAKgwmfFDuQmX6y2yxtUj2oAHxveVdY6vpMVFYEZOKj44cA5rvyzB83dcHZBxhAsGKERh6kBpFXYfuwCNWoVfdtEbvkqlQlpcBNLiIjCxSx4RiNCJaSSLVYDFamNTLg9dqmsCALz981EYlOr8m+7FWjN2/FCBz49U4LvTl/BDeQ1+KK/x+HHUKqBPj24YmByNCL3rlJ9aBdw+Mj2gK8d+fn0WPjhwDv85cA6P3jgIKbGepUVJPgYoRGHqxR3i7MnNV6chIyF0V+EY9a0BSaPFygDFA4Ig2AOUrB5RiI9y3sRMKn5+YHxfXK5rwu7iCx51d42N0OGq1Bj0S+oGo045NYhDe8Xh2qx4fFtyCev2nMJj0wcFekghiwEKUYj5odyE//fvQ6gxO6/1EAQBP5TXQKUCfjWhXxeOruvpNWqoVWKqp8FiRbRRF+ghBT1TQ7O9HsRVcHKl7lF63Hx1T38NK2gsur4Pvi25hLe/OY2HJ/VjLyA/4b8qUYj5YP85fHvqkkfH3jwsDf2Suvl5RIGlUqkQodOgrsmKxib5NS/hqLJOnAHpZtB2ekl4KJo8KAlZiVEouViHfxWcxYKxvQM9pJDEAIUoxDS0FIPOGpaGuSN7OT1Oq1ZjeEZcF40qsCL0YoDSwEJZj0jpnYRu3J+mI2q1Cvdfl4Un/n0Qf9txHIfOVbs8PiU2Ag9P6sf0okwMUIhCjLlZnCXo16Mbru/PltwA7DUODFA8c7FWDFDkpHfCzW0jeuG5z46iosaM//vurNvjR2TEYcLApC4YWehggEIUYswWMUAx6PhtTSKt5Klv8k0PllBnn0FhgOJUhF6DNxeOwu7iCy6P21RwFicu1KHCg8JhcsQAhSjEmJvFWQKDlgGKRFq+yl4onqmsFT9ME6K4Sasr2T1j3TZrKz5fixMX6lBV39RFowodfAcjCjFSiofFja3sKR4WyXqksmUGJZ41KJ0WFymuGrtUJ68hHTFAIQo5rQEKf70lEaxBkYUpHt+JjxT/DTmDIh/fwYhCjLnlQ5g1KK0YoMgjLTPmKp7Oi2sJ8i4zQJGN72BEIYYpnvbsNShNDFA8UWlfxcMalM7q3pLiucwUj2wMUIhCDFM87XGZsTxM8fiOlOLhDIp8fAcjCjFcxdMeUzyea7sPD1M8nRdnD1A4gyIX38GIQozUB0VJG7D5W0TLhoENTPG45e0+PNSx7lFiiqeqvgmCIAR4NMrCAIUoxNhTPCyStZNmUNgHxb2LLQWy0dyHxye6t8ygNNsElxt4Unt8ByMKMa0pHn64SFiD4rlL7IHiU0adxh4gV7FQVhYGKEQhhkWy7UmreJjicU9awcMCWd+RVvJcYqGsLHwHIwohgiCgiQFKOyyS9ZzUA4VLjH2nO3uheIXvYEQhRJo9AQADi2TtWIPiuUucQfG57uwm6xUGKEQhxCFA4QyKnVHPGRRPVXKJsc9xPx7v8B2MKIRIBbJqFaBVqwI8muBhT/GwBsUt+0aBnEHxGc6geIcBClEIkXqgGLQaqFQMUCStKR7uZuzOJe7D43OsQfEOAxSiEMIeKB2LYIrHY62reFgk6yvcj8c7fBcjCiFsc98xpng8xxSP73Xnfjxe4bsYUQjhTsYda9uoje3GnbPZBFxuCVASu3EGxVdaUzycQZGDAQpRCGmtQeGvdltSigdwXOlEjkyNFvs+PNIeMtR5rSkezqDIwXcxohBiT/GwBsWBsU3AxjSPc1J6h/vw+BZTPN7huxhRCGGKp2NajRp6TcuOxiyUdeoSe6D4hZTiMTfbGCDLICtAWbFiBVQqlcMlJSUFAGCxWPDoo48iJycHUVFRSEtLwz333INz58453MeECRPa3ce8efN894yIwhj34XHOqGOA4k5lrdTmngGKL0XpNdBpxGX/3I/Hc7LfxYYMGYKysjL7paioCABQX1+PwsJCPPHEEygsLMR7772HY8eOYfbs2e3uY9GiRQ738eqrr3b+mRARzBau4nGGGwa617qChwWyvqRSqRAnpXlYh+IxrewTtFr7rElbsbGx2Lp1q8N1a9aswbXXXoszZ84gIyPDfn1kZGSH90FEncMUj3Pcj8c9qQdKIlM8PhcfqceFGjOquJLHY7K/ZhUXFyMtLQ1ZWVmYN28eTp486fTY6upqMXKMi3O4fuPGjUhMTMSQIUOwbNky1NTUuHxMs9kMk8nkcCGi9tiozTkjdzR26xJ7oPiNfT8epng8JmsGZdSoUdiwYQMGDBiA8+fPY9WqVRg7diwOHTqEhIQEh2MbGxvx2GOP4a677kJMTIz9+vnz5yMrKwspKSk4ePAgli9fjgMHDrSbfWlr9erVWLlypcynRhR+2KjNOaZ43GvdKJApHl/jfjzyyQpQpk+fbv85JycHY8aMQd++fbF+/XosWbLEfpvFYsG8efNgs9nw0ksvOdzHokWL7D9nZ2ejf//+GDlyJAoLCzFixIgOH3f58uUO928ymZCeni5n6ERhoe1ePOQogjMobklFsgmcQfE5e7M2trv3mOwalLaioqKQk5OD4uJi+3UWiwVz585FSUkJtm/f7jB70pERI0ZAp9OhuLjYaYBiMBhgMDCiJ3KnkTMoTrEGxT2mePzH3qyNMyge69S7mNlsxpEjR5CamgqgNTgpLi7Gtm3b2qV9OnLo0CFYLBb7fRCR9+wzKKxBacfIFI9bleyD4jds1iafrBmUZcuWYdasWcjIyEBFRQVWrVoFk8mEBQsWoLm5GbfddhsKCwvx4Ycfwmq1ory8HAAQHx8PvV6PEydOYOPGjbjpppuQmJiIw4cPY+nSpRg+fDjGjRvnlydIFE64ise51hQPW913xGYTWhu1cZmxz8XZZ1CY4vGUrADl7NmzuPPOO3Hx4kX06NEDo0ePxt69e5GZmYlTp07hgw8+AABcffXVDuft2LEDEyZMgF6vx+eff44XXngBtbW1SE9Px4wZM/Dkk09Co+EbKlFnsUjWOdaguGZqtMDKfXj8Jj6KfVDkkhWg5OXlOb2td+/ebncJTU9Px65du+Q8JBHJwE6yzkmreFiD0rGLLT1Qoo3ch8cf4pjikY3vYkQhpLUGhR8wV7L3QWENSoda0zusP/EHqUiWjdo8xwCFKIQwxeMcUzyuXaprWWLMHih+IaV4as3NaGpmHZQn+C5GFEJYJOtcBDcLdElK8XCJsX/EGHVQi/sFslmbhxigEIUQ1qA4Z69BYYqnQ0zx+JdarUJsBFfyyMF3MaIQYt/NmH1Q2uFePK5dYg8Uv5O6yV7iSh6P8F2MKIQ0McXjFGtQXLvY0uY+nj1Q/Ib78cjDAIUohDDF4xw3C3SNKR7/685mbbLwXYwohNhX8TDF0w734nGtspYpHn9ju3t5+C5GFEK4m7FzrEFxrZIbBfpdd3aTlYUBClEIYYrHOaZ4nLPZBPu3+kT2QfEb7scjD9/FiEKEzSagycoAxZnWFA+bZF2puqHNPjyRnEHxl3imeGThuxhRiJCCE4Ct7jsiBShNVhuarQxS2pLSO9FGLfQMbv2G+/HII2uzQCIKXuY2MwOcQWlPSvEAQGOzDd004fFvZG624sMDZahqcJ5WOHu5HgDTO/7G/XjkYYBCFCKkFTxqFaCVemqTXdugraHJim6G8Hj7e3//Ofz2X997dGxSNAMUf4pnozZZwuM3lCgMtN2HR6VigHIllUqFCJ0GDRZrWC01PnmhDgDQL6kbhqTFOD1Oo1Jh/uiMrhpWWJJSPKZGseZHwy8SLjFAIQoR7IHiXoReDFDCaalxRU0jAGDOiJ741YR+AR5NeJNW8QiCWJjMJd2u8Z2MKEQ0WriCxx17u/swWmp8oUZsYZ8cbQzwSEinUSO6JbXINI97fCcjChFm7sPjlr0XSjjNoJjEACUphvUlwUBq1sb9eNxjgEIUIuwpHs6gOBWOGwZKKZ4kzqAEBe7H4znWoBD5Q20F0FAl/7zIeCAq0auHtM+gsAbFKXuztjBJ8TQ12+wfhFyhExzsvVD8nOIRBAFfHa/E+q9P4fA5k1f3kZkQibcXjfbxyDzHAIXI10q+ANbPAiDIP1elBn6+DeiZK/tU7sPjnjHMUjwXasX0jl6jthdoUmBJhbH+atZW39SMzft+xLqvTqG4orZT9xXoLzsMUIh8rWQ3AAHQGAB9pOfnNdUDVjNQ7GWA0pLiSVJVAZ/+Dmhy8+ak0QPX/BzoMVD2YylVRMsbbrgEKBUmMb3TI9rApedBwpv9eARBwPo9p/D6FyVul8jXNTXbC+aj9BrcPjIdM4amQu9FY0IGKEShpvK4+Oek/wHG/cbz8/a+AnzyKHCu0KuHlVI8N9dvAr7+l2cn1ZQDd7zp1eMpUbit4qloWcHTg+mdoNFdZoqn0WLF45uL8F7hjx4/RmZCJBaM6Y3bR/ZCtFG5M2cMUIh8TQpQEmT2nOg5Qvzzx0KxUYLMb7xSgDLAfEi8ImcukDig44MvlwD7NwKXSuSNUeGkVTzh0qhNClBYfxI8ustI8ZyrasADbxag6MdqaNQqPHrjQEwYmOTyHI1ahayEKKhDoAkcAxQiXxIEoPKE+LPcACUlB1BrgboKwPQjENtL1ulmixV6WJDR1BIgTXwciM/q+OCKH8QApeqMvDEqnLFlBqU+TGZQLrSkeLjEOHhIq3jOXKrHV8cvOj2usq4JKz84hMq6JnSP1OFvd43A2H7eFdArFQMUIl+qKQcsdWKxa/fe8s7VRQBJVwHlRcC5ffIDlGYbslUl0AoWIDLR9ePHpbecVC2uNoqIkzdWhQq3ZcbnpR4oXGIcNOJbUjw/lNdg/t+/cXv84NQYvHp3LtLjZdSzhQgGKES+JKV34jIBrRdtrNOGiwHKj4XAVbNknWputmG4ulj8S/q1rlNE+iggMgGorxRnUcIsQAmfFI/UA4UzKMEit3d3TBucjDOX6t0eOyorHo9Nv8phJ+5wwgCFyJe8rT+RpI0ACjd4VShrbrZiuLrl8Xtd4/6EuAwxQKkuBVKHyn68oGCuBbYsAWrPuzlQBQydiwi92NMh3IpkmeIJHgatBq/dMzLQw1AEBihEvtTZAEUqlD23T3ahrNliw4i2MyjuxGWIj6PkOpSi/wO+f8ezY8/mI2LCbgDhk+JpLZJliscvPlwC7H/b/XExqcB9HwPRKf4fUwhhgELkS/YC2b7enZ80WOyf0lgNXDop634M9eVIU12CDRqo04a7PyEuQ/xTyQHKiR3inzm3A/2nOT/uo2VAYzVSG44BABpa+kSEMqtNQGUtV/H4TcNloOAfgODB/6VLJ4Hj24DhP/P/uEIIAxQiX+rsDIpGJ67m+fE7cXZDRoCSVvu9OIRu/dFDH+X+hLhM8U+lBig2K1CyS/z52geAdBdprYPvAcc+RmpVAYDhYdHqvrLWDJsAqFUCEmp+AC5Wuz5BpRZroAzdumaASlfyhRicJPQD7t7s/LidfwT2vwVcONp1YwsRsgKUFStWYOXKlQ7XJScno7y8HIDY7W7lypV47bXXcPnyZYwaNQp/+9vfMGTIEPvxZrMZy5Ytwz//+U80NDRg8uTJeOmll9Crl7wVC0RBx9os9hcBvA9QADHNIwUoObd5fFqvOrH/ycW4oejhyQmxLSt5qk7LH2MwOLdPnGkyxIofrK70Hgcc+xg9KsUAJRxSPFJ657bIfdC8Pt+zk7olAzeuBobMkd2HJ+yc3Cn+2XdS62xkR9KuFgOUi8c6/5h1lUBzo+tjNHqgm0fvAEFP9gzKkCFDsG3bNvvfNZrW6uJnnnkGzz33HNatW4cBAwZg1apVmDp1Ko4ePYro6GgAwOLFi/Gf//wHeXl5SEhIwNKlSzFz5kwUFBQ43BeR4lSdBmzNgNYIxPT0/n6kD9sf5RXKZtaLAUpVwtWenaD0FI+U3sm6HtC4eSvLHAsAiLv4HdS4P0wCFPGDbJTuJGCFuPS8m4smX/WVYrHxv+4H9m0EZvwZiO/TNYNVopMt///6THR9nLSVxIUfOvd4hRuADx727NgpK4HrFnfu8YKA7ABFq9UiJaV9oY8gCPjLX/6C3/3ud5gzZw4AYP369UhOTsbbb7+NBx54ANXV1Vi7di3efPNNTJkyBQDw1ltvIT09Hdu2bcMNN9zQyadDFEBS/Ul8X0DdiT0s0loKZcsOiGkMtQeBe7MZmWaxQLYm8WrPHkfqhdJYLV6MsfLHGkjSB0RfNx8QAJAyDNBHQ9tkwiDVGVQ3XeXfsQWBipYeKL1VLSucJjwGXLvI+QnNZuDLvwBfPAuc+Bx4aQxw7S+A6FTXD2SMFWuAvFlWr1SXT4t1JSoN0Ps618f2GNR6jqVB7HfkjR8+Ev9UaZy/Jwg28UvSt68DY3/TufehICA7QCkuLkZaWhoMBgNGjRqFp556Cn369EFJSQnKy8sxbVproZrBYMD48eOxZ88ePPDAAygoKIDFYnE4Ji0tDdnZ2dizZ4/TAMVsNsNsNtv/bjJ5t3U0kV/Z60+8LJCVJPYH9N3Ezf4uHhObt7lTXgQdLKgUotEc66R77JUM0UBEPNBwCagqBVIUFKCYa4DSb8Wf+05yf7xGC2SMAo5vwyj1EXxgcbIFQAiRUjxptjLxiu5u/l9oDcCER8W04pYlYgpjz189e7DmBnHjyXAhBce9rgGMMa6PjeoBGOOAxirxPSIlx7vHLC8S/7z3Q/uMYDuWRuDPAwDTWeDMHvfBU5CTFV6NGjUKGzZswKefforXX38d5eXlGDt2LCorK+11KMnJyQ7ntK1RKS8vh16vR/fu3Z0e05HVq1cjNjbWfklPT5czbKKu0dkCWYlaA6ReLf7saZqn5cO60NYfBp2MVKlS0zynvgJsFrHQ19M0ROY4AMAo9Q9hk+JRwYZES8smc862PbhSQl/g7n8Dt70BDL1DnB1xdpH67Rz7zC/PIWidkDF7p1K1SfN4WShbf0kMOgAgeYjz43RGYHBLg8fv/8+7xwoismZQpk+fbv85JycHY8aMQd++fbF+/XqMHi02QLpyS29BENxu8+3umOXLl2PJkiX2v5tMJgYpvlBTDpzb7/642J7eR/3hxFcBCiAW1p3+UmzYNtyDAsezYoCyz9YPY7VyApR0oGy/8gIUOekdScu3yWvVR9DQ1OzRe5OSVZjMSEIVdDazmBZwVch5JZUKyL5VvLhSXgS8ch1w6kuguSk80jxtV4+5qz+RJA4ASr/xPkA5f1D8My7TfSo253Zg31vA4feBm/4kzowpVKeWGUdFRSEnJwfFxcX46U9/CkCcJUlNbc1ZVlRU2GdVUlJS0NTUhMuXLzvMolRUVGDsWCdTVhBTRQaDcv+Rg5K1Gfj7VKDaww+mB78CUrL9Oyal83aTwI603dnYE6X5AIB9Qn9M1MmYGFXqUmP7N1gP0juS1KshaCMQ31yLfvgR5mabffPAUFRRY0amVH8SlyEuYfe1pCFiCqPughgkKzyl4JGyA2IPFEMM0DPXs3OkOpSLXgYoUnrHky+Kva8HuqUAteVi75VBM7x7zCDQqQoas9mMI0eOIDU1FVlZWUhJScHWrVvttzc1NWHXrl324CM3Nxc6nc7hmLKyMhw8eNBlgEJ+cOoLMTjRRoi/ZM4ukS27Z574PLDjDXZN9a1TsD6ZQWkJUM4fFL+ZumI6B5jOwgo1Dtj6wqCVE6C0fKv2NFANBtU/im/0KjWQ9RPPz9PqIbR02B2lPhLy+/FcqDEjU90SoPhrNY5a3TqLcGK7fx4j2Eizd709WD0msad4vFxqbA9QPNiSQq1pbU+g8DSPrBmUZcuWYdasWcjIyEBFRQVWrVoFk8mEBQsWQKVSYfHixXjqqafQv39/9O/fH0899RQiIyNx1113AQBiY2OxcOFCLF26FAkJCYiPj8eyZcuQk5NjX9VDXeTgJvHPYfOAWX9xftyeNcBn/wOc+QYY1yUjU6ZLJ8U/jXFAZHzn7697byCiu/hNreKQ6z4fZ8XZk+PIQD2MMMhK8SiwBkX6gEgbLv4byaDufR1Qsguj1UfQYLEizvejCwqCIOBCjRm9VS21fZ7Wn3ij7yRxy4ETO4DJ/89/j9OR2gqgZLe4LYRcmWNk7xgOoHX2rs8Ez89JbCnKrjwuzl57GthIyltSPJ6m2nNuB75+ETj2CdBocl/IG6Rk/SudPXsWd955Jy5evIgePXpg9OjR2Lt3LzIzxWni3/72t2hoaMCvfvUre6O2zz77zN4DBQCef/55aLVazJ07196obd26deyB0pWazcCRD8Sf3eWY08XaIpTulb03TFhpW3/ii38jlUr8AD6xXUzzuApQWgpk9wv9AcC7GRQlBSjepHckLYWy16p/QK252YeDCi5V9RY0WW3I1Pl5BgVo/aA+t08s5vRFgO6JZjOwfjZw4Yh35+uigPs/BlKHeX5OU71YSwLIq3+KTQd0kYClXmzmmNjf83Obm1p7qHiaZk8dBiT0ByqLgR8+BK6+y/PHCyKyApS8vDyXt6tUKqxYsQIrVqxweozRaMSaNWuwZs0aOQ/dNSpPAK9N8O5crQGY+Txw1SyfDskvjn8u9r2ITnW+XE2SOkxsPFZfKX4Iy/nFCie+LJCVpI0QA5Sz37l+g2kJUAqsLQGKnBoUqZtsw2VlfNOy2Vo7eHpaoNhWz1yYoUOSqgrVF48DSR7sWaRA0hLjPpoK8Qp/BigxqeIeUhWHxdcme47/HqutL58XgxNjbOuqN09VnwUunQDevgNYtB2ISfPsvNN7AGsTENNL3u+6Wi2+d5YdEAtl5byPXvhBXLFmjG39fXVHJe7ejR1/ENM84RCghDxBAMxe9lgxA/jmVWUEKFJ6Z8gt7puAafViLcrpr4AzXzNAccaXBbISadbkwNvixY1vrWL/FVkpHmNMa4+G6lLA6GIJYzA4fxCovyh++5WWuMqhM+KIegCuth2C/uweYLCMAMVmBfa+BJzZ6/7Y+Cyxm6cnTfb8QOwiKyATXTCDAoizWRWHxfRbVwQoFUeA3X8Wf575vPuZ4Cs1VAFrp4m1TG/PBe77xLM9iOyrxybInylNHCgGKBePApjp+Xlt60/kPGbObWKAUrILqDkPRCe7PyfIMEBpKy4DeFhee3EAQE0ZsG6G+AFurhEbYAWrpjrgaEtHwmwP93nJGN0SoHwDjLjHf2NTMl81aWur93XiN6bqUreH2npdi1PHxQ7PslI8gPj/vrxKbNbmqsdCR6wWcadgKUBzJTIeSM4Rp6lTcsTtAOS+yUuFmFnXe72k9ZAuG1ebDyHi3DcAHvLspNoLwKb7xXoHT2WNB/pP9WqMnXXeZEY8ahCFegCq1tVa/tJnoljzcGKH/1PBNqvY8t1mAQZMF/cNkisiDpj/f8Drk8UAYNNCYN7b7gNKe/2JF7N3PVrqUOQWyp6XWX8iie8jBvFn88UvpWN+Je/8IMAApS2t3rsPmIS+YpfGyyXiDpeDbvL92Hzl2CdiHrR779alrO5IdShnvvbbsBTPHymeiDjgkQNie2w3apr1wP+Kq+O8C1C+964OpXgrULDO8+MPv9/6syFGTB/KIc1wevMB0eKocRhgfgcx57/x7MP0zDfAu/cCNefEmZvrl7guzj38bzGQObkzYAFKRU1ja4FsTE+xgZc/ZY4VN6mrLvV/Kjj/7+KHrj4amPGs98FQ997AnXnA+pni++Iny4EpTzo/vu6iWLAOyCuQlSS2rOSRu9RYmkFJ9qLNQ87t4r/V/reB5MHyz9dFAi0r3wKBAYqv9JsC5L/esu48iAOUopb0Tvatnv9ip18DQCXmbGsvhMxOmT5Tf0lsFw/4fipdrfFo6tlsFjeG06hV0GrkBihSLxQvdjU+ukX8c9BMN9Psgrg8+PxB8Q334jEx2PAmpaoxAAOnuz/OiTMRQ2Cp0sDYUC4ut+/Wfm8xu+PbgK1PiPubJA4A5r4JJA1y/QDGWDFAkZp5BUCFqU0PFH+u4JHoI8WZ1pLd4iyDvwKUqlJg20rx56krxCaSnZF+DXDLq8C7C4BvXxUv7qQMBaIS5T+W1AvlwjGxlsqTfXIEQfzyAHjXLHPIHDHwOl8EbLhZ/vkJ/YGHv5N/no8wQPGVfpNbA5TOTHE2VLV+2LkSmy6/8VJDFXC8pQeNnJxtRHdxP5iKw+JqHiXU2XQlKb0RneZZHtsPzM02AF7MngCtmwbKnUGxWYGjn4g/X7tI3rfKZjNwqUT84JcrOsW7D4gWamMUvhf6IFdVDKz38P/ykFuA2Ws8S99KvVnKi4C6SiAqweuxeutCjRn9/d0D5Up9J7UEKNuBUb+Qd64giKuA3AWsX70AWOqAjDFA7v3ej7WtIT8Fav4IbHsSaG50faxKDeQu8O5x4rMAtVYcv+nH1t87V6pLxQUNal1rgCNHtx7AxMeBg+/JPxeQ133YDxig+Erv68X/RFUtu1x6kyqqOAK8Oh6wmt0fG9MLmP2COHPjqR8+FCvQe1wlv9YgY7QYoJxhgNLOJalA1of1JzKZm8WmY94FKF4uNT77nViwaoi1L9/1mNbgfibCTyJ0GqxrvgFXdbuMSLWbAEkXCYz7DTDqQc+/dHRLal3VUrKr61a1tFFR04gpqgAEKNtWiLNSVovnX6AaTcC/fym+P3lCowdm/dW3O/WOfhAYeb/7gFmt8b51vEYn7nR+8ah48SRAkfqf9Bjk/TYCP1kmXhSIAYqvGLqJH+KnvhCX8XrzYbVnjRicaPTixRlrk9i19K1bgeE/A6b9QaxXcOdgm/SOXOmjge/e8GwFQ7jxR/2JTI0WaQbFi1Uj3gYoUnqn/1T/tFH3E6NOg022sRgy7n48ON5PQWXW+AAHKGb07soUDyAWQEcmikHr2Xz3LQwA8UvZOz8Tf4c0ejGl4IpKLc7OSAWnvqTVA/DzXkI9BojByYVjnn25lNPiPgQxQPGlfpPFAOXE5/KnOGsvAEXvij/fu8V1YVJTHfD5/wLfvCJuCnX8c3FTKFf/iRtNwMmWnLg3b5gZLYWyZfvFZkX6SPn3EaqCIECxp3jk9ECR2HuhXALMtZ6nqX5oWQ0WzDVXHYjQi/9GDU1+bHXfZwLwzcutv3NdSBAEsQZFLXWR7aIZFLVafN4H/yWmedwFKAc3Ae8/LKY8YnoBczcAvTzc20apegwCjvyntfGaO/b6k/DcB40Bii/1myJOcZbsFnPscqYCC9aJMyNpI9z3d9BHAdOfFnOn7z8kfkC+8zPPHidtuHezO3EZYmO3mjJxh91w2BQMEFepSG8SzrRs1BfYAKUTKZ6IOLGws7FazHknXeX+nIvFYpdKtQ7oF5iVKt6KaNkg0K978WSOFXcQvlwizkx1YS6/1twMncWEeGOteEX3LppBAcQ0z8F/AQXrW9MTHbHUtxYRZ40HbnujU3VFimFfyePhUmPOoJDPJGcD3ZKB2vNiKqTPeM/Oa24Sl84BwOhfep7rzhgNPPglsHM1ULjB/aZyWgNw3X95dt9XUqnExzu0WVxuHA4ByokdwEYPe8UA/pl29lBrkayXjcHiMsQ3w6ozngUoUi+drOuDv/vsFaQApcGfAYqxZafbs9+Ksygj7vbfY12hosaMDCm90y25awu3+04Sg9a6CuDYx+6Pv24JMOl/AtbQrsvZe6F4sNS4sbp1ZZ03S4xDAAMUX1KpgL6Txa6fx7d5HqAcfl/cGrtbCjD4p/IeUxcBTP29ePG3dClA+cb/jxVo9ZeAzQ+KP2eMdT/rlDqs66bSO2C2dGIVDwDEtglQPCGldwYqK70DAEZ9S4DizxQPIP7+n/1WnCnoygDF1Lb+pIv/T8akAvd/ItbfuJM0GOg10v9jCiYJ/QGoxHRq3UXXs0bnW3quxKZ33f5GQYYBiq/1awlQTmwH8L+enfPNy+Kf1yz0vlK7K0h1KKXfer6OX4kEQexUWVsuTsn+bFPQ19zYUzze1KAAbQplPeiFUnuhdcO0TvQjCZQumUEBxNTF7j+JMyhduNFmRU1jaw+UrkzvSHqNDL/Aw1P6SPF3req0OIviKkAJ8/QOwADF9/pMBKASG1KZysRvFK6U5gM/FogV7Ln3dckQvZacDei7AeZqcZMuuUuV27p8SsxTXz7l/tj4PuJST2Os948nR8E6ccmjWgfc+vegD06AtjMonUjxAGIjLHeOfQJAEGeNvNmuPsC6pAYFEAvdtRFiuqPiiHedPL1wocbc2kU2gLN65ESPgS0Byg9AbxfL86XatzBN7wAMUHwvKkEsRD1XKM6iDJ/v+nhp9iTn9uDv0KrRit+MTu4U61DkBiiCIE53f/MqcPRjAILn5+5/W2xr7e8VIxeOiZ0XAbHtdepQ/z6ej3SqSBaQt9T4aEttwcAZ3j1WgEXou2gGRWsAMseI7wMlu7osQKmoMWOououXGJPnEgcAxZ+5L5SViow5g0I+1W9KS4DyuesAxXSudW+SUQ90zdg6K2OMGKB8+UJrF1FPVZ12/KXsO0lcAeKqQM7WDOSvFZuh5d0pdvSc/ozYDEsOS4O41NoVwSpuGtbcIM6EjfZwI7kg0KlOsoDnAUpTfeuGfQpM7wBiHxSgC2pQADHNc2K7mOYZ/Uv/Px6AClObFA9nUIKP1BH20GbXM8hSDQoDFPKpfpOB3c+IS1Tfd/EhV3lC/ADOGCtOlytBn4niqqHqM+JFLl0UcPVdwLUymi2NvB/Y+bTYyO7QZnF1TU93/RIEsbV//UWx3bilzvMxRsQDP31ZUTU2nV/F09ILpf6i2GdHH9XxcSd3igFcbIZi3zhba1Bs/n8wqVD+9FeAtVmchfSz6upqJKuqxL9wBiX4SJu01p5vSZe60C3Z/ztRBzEGKP7Qc6T4IddwSWyk5k4XfbPyiYxRwN3/Fmd/5NIZxdklubUkughg6kpx9uSDX4vFYyc+l//4AAA3hYr6KOCWV9zXDgUZs6WTRbLGOHF3YbMJ+PMAsWNnR5pbtmEYOL3Lij59TUrx+L0GBRA3lovoDjRcFveaSXfT48gHtKZTAACLPg46V7suU2AkDwHu+cCzdGrGaEV9UfI1Bij+oNGKKz9O7nR/bExP5e1t09f7re47Je1qYNEOcQl3Y7X74w0xYpV8ZIJ4McYq9kPVnU6neFQqcZO7Hz4EmmrdHKsBht3h3eMEgYiuTPGoNeI+XUc+AL5bK25R4ZQKSB9lD47f+LIEr+0+iSarvJmeaxtLAB1gjesN5WxAEGY8bUER5hig+EvPEa1TeeQ7Gp1iax/8qdMpHgCY+yZQdUosZnYlorui+zIYu2qZsaTPeDFAOfBP8eKKIQaY8xrWnO2HZ7d62G30Cpkasf5E1yNwnY2JfIEBClEI6PQqHkCcSg6DosouW8UjyZkLnPoSqK1wfVzteXHbin/OQ5PlNqjwUzwyZSBm5MhLNybt2gIcBjSJgdtdm8gXGKAQhQB7HxRva1DCiJTiaWq2wWoToFH7Oe1njAFuX+f2MKHZjAN//xWuLv8Xlur+hVtSK9Hn+jcBQ7S8x2to6WUTBsEmhTYGKEQhwCcpnjAhBSiAWCgbZfDv26DNJmDH0QqUXqp3edzR8zX456k5uF0Tj6f1/0CfizuAV38iv9/Qj/vEPwPRRZbIhxigEIUAn6R4wkTbf6OGLghQXvi8GC98Xuzx8QNv/CU0WXeIO5RfOile5NLogcT+8s8jCiIMUIhCgH0GhSket9RqFYw6NRotNr+v5Pmi+AL+ul0MTiYPSrJvVNgRFYAbs1Mwc2gagD7Ag1+JOwJLS7vlSMlRdCEzEcAAhSgkdHovnjATodOg0WLzay+U8upGLM7bD0EA7rw2HavnyNw2ISoBGP4z/wyOSAH4dYsoBDDFI4+/dzS2WG349duFqKxrwuDUGDw5qxMbaxKFKb6bEYUAFsnKI6Va/JXi+dOnR/Hd6cuINmjx8s9G2HuvEJHnmOIhCgGsQZHH2xmUb05WouhH112ML9U14bXdYmHrn24fiswEJ/saEZFLDFCIQoB9Lx6meDwiBSie1qBU1DTi9/85jA+/L/P4MRZel4Ubs5W1pxNRMGGAQhQCmOKRx9NusjabgLz8Ujz98RGYGpuhVgFTrkpGpIvVOADQPzkai65nozSizuhUgLJ69Wo8/vjjeOSRR/CXv/wFAKByshnbM888g//+7/8GAEyYMAG7du1yuP2OO+5AXl5eZ4ZDFLY6vVlgmJFqQt7JL0Xh6Sqnxx08V419Z8Tbs3vGYPUtQ5HTS+Zu3ETkFa8DlPz8fLz22msYOtRx6VxZmeMU6Mcff4yFCxfi1ltvdbh+0aJF+P3vf2//e0REhLdDIQp79lU8rEHxSGI3AwBg78lL2HvykstjI/UaLJk6APeO7Q2thv++RF3FqwCltrYW8+fPx+uvv45Vq1Y53JaSkuLw9/fffx8TJ05Enz6O052RkZHtjiUi+aw2ARaruAMxUzyeWTylP3p1j7DPPDlj1Klx89U90TOOX6CIuppXAcpDDz2EGTNmYMqUKe0ClLbOnz+PLVu2YP369e1u27hxI9566y0kJydj+vTpePLJJxEd3fGmWGazGWZzazdFk8nkzbCJQlJTmw9Zpng8kxxjxEMT+wV6GETkguwAJS8vD4WFhcjPz3d77Pr16xEdHY05c+Y4XD9//nxkZWUhJSUFBw8exPLly3HgwAFs3bq1w/tZvXo1Vq5cKXeoRGFBSu8ADFCIKHTIClBKS0vxyCOP4LPPPoPRaHR7/BtvvIH58+e3O3bRokX2n7Ozs9G/f3+MHDkShYWFGDFiRLv7Wb58OZYsWWL/u8lkQnp6upyhh7Ti8zX46/bjbpdM6rVq/HpiP1yVGtNFI6OuIKUpNGoVaySIKGTIClAKCgpQUVGB3Nxc+3VWqxW7d+/Giy++CLPZDI1GzIF/8cUXOHr0KN555x239ztixAjodDoUFxd3GKAYDAYYDAY5Qw0r6/acwn8OnPPo2B8vN2Dzr8Y6XW1FytO6Dw+DEyIKHbIClMmTJ6OoqMjhuvvuuw+DBg3Co48+ag9OAGDt2rXIzc3FsGHD3N7voUOHYLFYkJrKpkbeqGlsBgDMGJqKcX0TOzzGKgj4w5bD2F9ahV3HLmDCwKSuHCL5EffhIaJQJCtAiY6ORnZ2tsN1UVFRSEhIcLjeZDLh3XffxbPPPtvuPk6cOIGNGzfipptuQmJiIg4fPoylS5di+PDhGDdunJdPI7xJzabG9U3EXaMynB53+mId/v5lCV74vBjjB/TgLEqIYJM2IgpFfvnKlZeXB0EQcOedd7a7Ta/X4/PPP8cNN9yAgQMH4je/+Q2mTZuGbdu2OczAkOek2hOjmx4YvxjfBwatGvvOVGF38cWuGBp1AfZAIaJQ1OlW9zt37mx33S9+8Qv84he/6PD49PT0dl1kqXOkACXCzY6pSdFGzB+ViTe+KsEL247hJ/0TOYsSAliDQkShiO9oIaCx5QPK6GZ/EAB4sGUWpfBMFb48zlmUUMAUDxGFIgYoIUCqQTF68AGVFGO016n8ZVsxBEHw69jI/1gkS0ShiO9oIcCe4vFgBgUAfjm+LwxaNQpOX8ZXxyv9OTTqAvYZFNagEFEI6XQNCgWep0WykqQYI+68NgPr9pzC7z88hMlXJbs8vmdcBOaPymC9SpBqrUFhioeIQgcDlBDQ0ORZkWxbv5zQF29/ewbHztfi2Plat8dnJkTi+v49vB4j+Q9TPEQUihigKJwgCGhsmeI3yghQkmOMeO3uXHzhZrnxNyWVOPijCV8dr2SAEqRai2QZoBBR6GCAonAWqwCrTSx0lROgAMCEgUluO8puKjiLpe8ewN6TrFUJVmYvAlQiomDHr1wK19hmJ1s5KR5PjeoTDwAo+rEadeZmn98/dZ7ZwhQPEYUevqMpXGNL/YlaBeg0vi9i7dU9EunxEbDaBHx3+rLP7586r3UVD2dQiCh0MEBROKlJW4RO47dVNqOzEgCAaZ4gxRoUIgpFfEdTOHuTNj9+ex7VhwFKMOMqHiIKRXxHU7guCVCyxDqU78+yDiUYsQ8KEYUiBigKJ7dJmzfS4yPRqzvrUIIVO8kSUSjiO5rCNchsc++t0UzzBC2meIgoFPEdTeGkJab+WGLcFgOU4MXdjIkoFDFAUbiuqEEBWIcSzFprUPjrTEShg+9oCictM/Z3gMI6lOBlT/GwBoWIQgjf0RRO2iiwK9qcS2meb5jmCSpM8RBRKGKAonD2Itku+PbMOpTgxEZtRBSK+I6mcOYuqkEBWIcSrFr34uEMChGFDgYoCtfQRat4ALEOpWdcBJptAgpYhxI02AeFiEKRNtADoM7pqiJZyeg+CdhUeBb/3vej7HNjInQY1ivWb3sGhSumeIgoFDFAUbiuWmYsGd0nHpsKz+K9fT/iPS+ClOfvGIZbhvfyw8jCV2ujNqZ4iCh0MEBRuMYuLJIFgOk5qfjkYDnKqhtlnVduasSluiaUXKz308jCk9UmwGIVAHAGhYhCCwMUhWvs4hmUbgYt1t57jezzVn90BK/uPomGJhbXyvHh9+dwoqLO6e3NNpv9Z9agEFEoYYCicF21F09nSQFUfUvfFnLveEUNfv32Po+O1WvVTPEQUUhhgKJwjRZlNOmKbAmgGhigeOzY+VoAQHKMAVOuSnZ57HX9EqFRs/iYiEIHAxSFkz7wg30GRQpQOIPiuZKLYmpnbN9E/OGWnACPhoioazFprXCNzV3XB6UzIvRiLFxvYYDiqdOVYoDSOyEqwCMhIup6DFAUrtG+F09wv5TSDEojZ1A8dqpSXPHUOzEywCMhIup6nfpUW716NVQqFRYvXmy/7t5774VKpXK4jB492uE8s9mMhx9+GImJiYiKisLs2bNx9uzZzgwlbDW2NOkK/hmUlhSPhat4PCXNoGRyBoWIwpDXAUp+fj5ee+01DB06tN1tN954I8rKyuyXjz76yOH2xYsXY/PmzcjLy8OXX36J2tpazJw5E1Yrv13L1ZW7GXdGJFfxyFLf1IzzJjMAIIsBChGFIa8ClNraWsyfPx+vv/46unfv3u52g8GAlJQU+yU+Pt5+W3V1NdauXYtnn30WU6ZMwfDhw/HWW2+hqKgI27Zt8/6ZhCFBELq8k6y3IriKR5bTLemduEgdYiN1AR4NEVHX8ypAeeihhzBjxgxMmTKlw9t37tyJpKQkDBgwAIsWLUJFRYX9toKCAlgsFkybNs1+XVpaGrKzs7Fnzx5vhhO2pD1YAOXUoHAGxTNM7xBRuJO9zDgvLw+FhYXIz8/v8Pbp06fj9ttvR2ZmJkpKSvDEE09g0qRJKCgogMFgQHl5OfR6fbuZl+TkZJSXl3d4n2azGWaz2f53k8kkd9ghqbHNipjgn0ER/6txBsUz0pYAvRNYIEtE4UlWgFJaWopHHnkEn332GYxGY4fH3HHHHfafs7OzMXLkSGRmZmLLli2YM2eO0/sWBMHpLrerV6/GypUr5Qw1LEhN2nQaFXSaIJ9BaQmgmqw2NFtt0Ab5eAONS4yJKNzJ+pQoKChARUUFcnNzodVqodVqsWvXLvz1r3+FVqvtsMg1NTUVmZmZKC4uBgCkpKSgqakJly9fdjiuoqICyckdd8tcvnw5qqur7ZfS0lI5ww5Z9vqTIO8iCzg2kmtgLxS3TkkBCpcYE1GYkhWgTJ48GUVFRdi/f7/9MnLkSMyfPx/79++HRtP+g7KyshKlpaVITU0FAOTm5kKn02Hr1q32Y8rKynDw4EGMHTu2w8c1GAyIiYlxuFCbjQKDvIssIO60K3ViZ5rHPalIljUoRBSuZKV4oqOjkZ2d7XBdVFQUEhISkJ2djdraWqxYsQK33norUlNTcerUKTz++ONITEzELbfcAgCIjY3FwoULsXTpUiQkJCA+Ph7Lli1DTk6O06Jb6ljrCp7gT5eoVCpE6rWoNTezUNaNhiYryqobAXCJMRGFL5/uxaPRaFBUVIQNGzagqqoKqampmDhxIt555x1ER0fbj3v++eeh1Woxd+5cNDQ0YPLkyVi3bl2HMzDknNSVNdibtEmMOg0DFA+cuSTOnsQYtYjjEmMiClOdDlB27txp/zkiIgKffvqp23OMRiPWrFmDNWvWdPbhw5q0D0+wr+CR2Hc0ZjdZl1rrT6KcFo4TEYW64M8NkFMNTeIqHqUFKJxBcY09UIiIGKAomlQkq5QUTwQDFI9IPVCy2AOFiMIYAxQFU1KRLNBmR2MuM3aJMyhERAxQFE1xMyg6seSJMyiuSUuM2QOFiMIZAxQFa1TIRoES1qC412ix4lx1AwB2kSWi8MYARcGUspOxRJrpaWjiKh5nSi/VQxCAaIMW8VH6QA+HiChgGKAomLQXj2ICFM6guHVK6iCbGMklxkQU1higKFiDwmpQmOJxj5sEEhGJGKAomL1IVq+Ml9HeqI0BilOnGKAQEQFggKJoSiuSjdCLq3i4m7Fzpy5KmwRyBQ8RhTcGKAqmtBoUpnjca9vmnogonDFAUTApVaKUAMW+iod78XTI3GzFuSouMSYiAhigKJrSimS5ise1s5cbYBOAKL0Gid24xJiIwhsDFAVrVGirexbJduzUxdYW91xiTEThThmfbNQhpbW6Zw2Ka1IPlCzWnxARQRvoAZD3lFYkGy578dhsAr44fhFV9U2yzvui+AIAruAhIgIYoCia0lrdh8tuxp8cKsevNhZ6fT5X8BARMUBRtNZGbcoKUOqbmiEIQsjWWRT9WA0A6BkXITtdk9hNjxuzU/wxLCIiRWGAolA2mwBzc0uKR6uMUiJjS4BiEwBzs00xMz9yScWu91+XhYXXZQV4NEREyqSMTzZqp7G5NU2imBmUNgFJKK/kaS12ZS0JEZG3GKAolFQgCwBGrTICFK1GDb1G/C9XH6J1KIIg2Df8y2SzNSIirzFAUSipQFavVUOtVk4tR4S9F0podpO9UGNGfZMVahWQ3p0zKERE3mKAolBK64EiCfVeKFJ6p2f3COgVUhtERBSM+A6qUK378CjrJYwI8W6y9s3+mN4hIuoUZX26kZ25WeEzKCFag9Larp7pHSKizmCAolANTcrqIiux72gcojMop1tSPJxBISLqHAYoCqW0LrKSCH1ot7tnioeIyDcYoCiU0nYylkTqQncVjyAI9hQP29UTEXWOsj7dyK6Bq3iCzsXaJtQ1WaFSAenxEYEeDhGRojFAUSizwvbhkUSEcIAiNWhLi42AQSHN84iIghUDFIWy16Ao7IMwlHc0LmlJ78jdIJCIiNrrVICyevVqqFQqLF68GABgsVjw6KOPIicnB1FRUUhLS8M999yDc+fOOZw3YcIEqFQqh8u8efM6M5SwI7W6NypuBiV0i2SlFTxcYkxE1HleByj5+fl47bXXMHToUPt19fX1KCwsxBNPPIHCwkK89957OHbsGGbPnt3u/EWLFqGsrMx+efXVV70dSlhS6gyKVDMTigEKV/AQEfmO1puTamtrMX/+fLz++utYtWqV/frY2Fhs3brV4dg1a9bg2muvxZkzZ5CRkWG/PjIyEikpKV4Om6Q+IhF6ZWXppBRPgyX0VvHYAxSmeIiIOs2rT7eHHnoIM2bMwJQpU9weW11dDZVKhbi4OIfrN27ciMTERAwZMgTLli1DTU2N0/swm80wmUwOl3AndZJV3AxKiBbJCoKA0xelJm1M8RARdZbsGZS8vDwUFhYiPz/f7bGNjY147LHHcNdddyEmJsZ+/fz585GVlYWUlBQcPHgQy5cvx4EDB9rNvkhWr16NlStXyh1qSGudQVFWgBKqy4wv1TWhxtzcssSYAQoRUWfJClBKS0vxyCOP4LPPPoPRaHR5rMViwbx582Cz2fDSSy853LZo0SL7z9nZ2ejfvz9GjhyJwsJCjBgxot19LV++HEuWLLH/3WQyIT09Xc7QQ469SFahfVBCbRXPqTZLjJX2mhARBSNZAUpBQQEqKiqQm5trv85qtWL37t148cUXYTabodFoYLFYMHfuXJSUlGD79u0OsycdGTFiBHQ6HYqLizsMUAwGAwwGg5yhhjzFtrrXheYqnlMXuYKHiMiXZAUokydPRlFRkcN19913HwYNGoRHH33UITgpLi7Gjh07kJCQ4PZ+Dx06BIvFgtTUVHmjD2ONCu8kG2qbBZ5mgSwRkU/JClCio6ORnZ3tcF1UVBQSEhKQnZ2N5uZm3HbbbSgsLMSHH34Iq9WK8vJyAEB8fDz0ej1OnDiBjRs34qabbkJiYiIOHz6MpUuXYvjw4Rg3bpzvnlmIU+pePK1FsqG1iqekkgWyRES+5NUyY2fOnj2LDz74AABw9dVXO9y2Y8cOTJgwAXq9Hp9//jleeOEF1NbWIj09HTNmzMCTTz4JjUZZswGBpNS9eEK1D4o0g5LJHihERD7R6QBl586d9p979+4NQRBcHp+eno5du3Z19mHDnlQka1BYgCKleMzNNlhtAjRqVYBH1HmCILDNPRGRjykrP0B2Sp1BidS3xsQNIbKS53K9BTWNYsoqg0uMiYh8ggGKQjUqdDdjo04NVcukSagUykpLjFNjjYpbVUVEFKwYoCiUUotkVSqVfdYnVAKU09yDh4jI55T16UYAgGarDRarWOujtBQP0KabbIjsx1MitbhPZHqHiMhXGKAoUGOzzf6zElMKxhBbycMZFCIi3/PpMmPqGm1TIwat8mJMJTVrK71Ub1+h48zhc+LmlVxiTETkOwxQFKht/YlKpbxluhF6ZbS7P3u5Hjf+ZTfqPBwnUzxERL7DAEWBlNrmXhKpU0Y32d//5zDqmqxI7GZAUrTrvaCGpMVgQFJ0F42MiCj0MUBRIKlJm2IDFAXsaLzjhwp8dvg8NGoV3l40CgOSGXwQEXUl5RUwkGJ3Mpa07scTnAFKo8WKFf85BAC4f1xvBidERAHAAEWBGhUeoEQGeYDy2u6TOF1Zj+QYAx6ZMiDQwyEiCksMUBSoQaFN2iTB3Kit9FI9/rbjOADgdzMGo5uBWVAiokBQ5idcmFNqm3tJMK/iWfmfQzA32zC2bwJmDU0N9HCIiMIWvx4qkD3Fo1VmgGLvg9KFnWS/KL6AdV+dgtXFbttNzTbsOVEJrVqF3988RJFLuImIQgUDFAWSUiNGhc6gdHUNis0m4HebD+LMpXqPjl94fRb6cckwEVFAMUBRIKnVvVKXGUd0cSfZb09dwplL9ehm0OLJWYNdzoxE6jWYNji5S8ZFRETOMUBRIPsMikKLZFtTPF0ToPyr4CwAYEZOKm4fmd4lj0lERJ2jzE+4MNfYrOxOshFduFlgnbkZHxWVAQBuH9nL749HRES+wQBFgRqblN0HpStX8WwpKkN9kxVZiVHIzezu98cjIiLfYICiQErvJNu6m7H/V/FI6Z3bcntxVQ4RkYIwQFEgaS8epQYoXZXiOV1Zh29LLkGtAuaM6OnXxyIiIt9igKJADUrfzbiLVvFIsyfX9e+B1NgIvz4WERH5FgMUBWpUeKv7yJYaFH+u4rHZBGxqk94hIiJlUeYnXJhrVPgMitQHpdkmoKmlp4uv7TlRiXPVjYgxatnXhIhIgRigKJC9BkWhnWTbBlb+SvO8W1AKAJh9dZpia3WIiMIZG7UpUIPC9+LRa9XQqlVotgmotzQjFjqPzzU1WvDnT4+iqt7i8rhPD5UDAG7PZWM2IiIlYoCiQNKsg1J3MwbEsdc0NsteyfPud2ex4evTHh07KCUaQ3vFejM8IiIKMAYoCmRuVnaRLCCu5KlpbJad4ik8cxkAcOOQFFybFe/0OLUKmDQomb1PiIgUigGKAtlnUBRcWyGu5DHLnkHZf6YKAHDP2EyM7Zvo+4EREVFQUO5X8DAlCIJ9N2MlF39KwZWcpcbnTY34saoBahUwtFecn0ZGRETBoFMByurVq6FSqbB48WL7dYIgYMWKFUhLS0NERAQmTJiAQ4cOOZxnNpvx8MMPIzExEVFRUZg9ezbOnj3bmaGEDYtVgNUmAFB2gOJNu/t9LemdAcnR6Gbg5B8RUSjzOkDJz8/Ha6+9hqFDhzpc/8wzz+C5557Diy++iPz8fKSkpGDq1KmoqamxH7N48WJs3rwZeXl5+PLLL1FbW4uZM2fCavX/5nFKJ+1kDCg7xSMV+MpJ8exrSe+M4KZ/REQhz6sApba2FvPnz8frr7+O7t1bPywEQcBf/vIX/O53v8OcOXOQnZ2N9evXo76+Hm+//TYAoLq6GmvXrsWzzz6LKVOmYPjw4XjrrbdQVFSEbdu2+eZZhTBpJ2O1CtBplFsA6s1+PFKAMjw9zg8jIiKiYOLVPPlDDz2EGTNmYMqUKVi1apX9+pKSEpSXl2PatGn26wwGA8aPH489e/bggQceQEFBASwWi8MxaWlpyM7Oxp49e3DDDTe0ezyz2Qyz2Wz/u8lk8mbYQc/UaME9a7/Fj1UNTo+xtaR3InQaRa9Qkbsfj8Vqw/c/VgEAhmdwBoWIKNTJDlDy8vJQWFiI/Pz8dreVl4vNsZKTHVuLJycn4/Tp0/Zj9Hq9w8yLdIx0/pVWr16NlStXyh2q4nx36hL2l1Z5dOxVqTH+HYyfRbTsx+PpDMrR8ho0WmyIMWrRJzHKn0MjIqIgICtAKS0txSOPPILPPvsMRqPR6XFXfrMXBMHtt31XxyxfvhxLliyx/91kMiE9PfQ6hJoaxILR4RlxeOqWHJfH9u3RrSuG5DfSDEq9xbMiWalA9uqM7lCrlTtzREREnpEVoBQUFKCiogK5ubn266xWK3bv3o0XX3wRR48eBSDOkqSmptqPqaiosM+qpKSkoKmpCZcvX3aYRamoqMDYsWM7fFyDwQCDwSBnqIpkahTbt6fEGBU/Q+KOFKA0ejiDwvoTIqLwIqtIdvLkySgqKsL+/fvtl5EjR2L+/PnYv38/+vTpg5SUFGzdutV+TlNTE3bt2mUPPnJzc6HT6RyOKSsrw8GDB50GKOHC1CAGKDFGz/emUSq5q3j2taS+hmfE+WlEREQUTGTNoERHRyM7O9vhuqioKCQkJNivX7x4MZ566in0798f/fv3x1NPPYXIyEjcddddAIDY2FgsXLgQS5cuRUJCAuLj47Fs2TLk5ORgypQpPnpaylTTKKY7YiJCv8eHfRWPB43aLtc1oeRiHQDgas6gEBGFBZ9/Ev72t79FQ0MDfvWrX+Hy5csYNWoUPvvsM0RHR9uPef7556HVajF37lw0NDRg8uTJWLduHTQa5fb18AUpxRMdBjMoclbxSIXDfXpEIS5S789hERFRkOh0gLJz506Hv6tUKqxYsQIrVqxweo7RaMSaNWuwZs2azj58SJGKZGOMYTCDYl/F475IVtogcHg6lxcTEYUL7sUTRKQZlJiIMJhB0Xk+g9LaQTbOjyMiIqJgwgAliJikGpRwSvG4qUGx2gR7ioczKERE4YMBShCpaZBqUMIhxePZKp4TF2pRa25GpF6DAcnK7v1CRESeY4ASRMIqxdNSg+IuxSM1aBvaKxZaDf+7EhGFC77jBxF7kWwYBCiebhZob9DG/XeIiMJK6OcSFKLRYkWT1QYgXFbxtNagXP/MdqfHVZjETSLZQZaIKLyE/iehQkjpHZUKiNKH/ssSF6lDUrQBFTVmlF5yvnszAHQzaHFN7/guGhkREQWD0P8kVAgpvRNt0IbFZng6jRpbl4zHiQu1bo9N7x6J7lFs0EZEFE4YoASJcCqQlcRG6DCCtSVERNQBFskGiZow6oFCRETkDgOUIGHfyTgMNgokIiJyhwFKkAinjQKJiIjcYYASJFo3CmSAQkRExAAlSNQ0MsVDREQkYYASJOyreDiDQkRExAAlWNj7oIRBF1kiIiJ3GKAEiXDsg0JEROQMA5QgwT4oRERErRigBAn2QSEiImrFACVIsEiWiIioFQOUIME+KERERK0YoAQBi9WGBosVAFM8REREAAOUoCAVyAJANwMDFCIiIgYoQUAqkI3Sa6DV8CUhIiLip2EQYA8UIiIiRwxQggALZImIiBwxQAkC3CiQiIjIEQOUICCleKI5g0JERASAAUpQaE3xcAaFiIgIYIASFFgkS0RE5IgBShDgRoFERESOZAUoL7/8MoYOHYqYmBjExMRgzJgx+Pjjj+23q1SqDi9/+tOf7MdMmDCh3e3z5s3z3TNSIKkPSjRTPERERAAAWZ+IvXr1wtNPP41+/foBANavX4+bb74Z+/btw5AhQ1BWVuZw/Mcff4yFCxfi1ltvdbh+0aJF+P3vf2//e0REhLfjDwlM8RARETmSFaDMmjXL4e9/+MMf8PLLL2Pv3r0YMmQIUlJSHG5///33MXHiRPTp08fh+sjIyHbHhjP2QSEiInLkdQ2K1WpFXl4e6urqMGbMmHa3nz9/Hlu2bMHChQvb3bZx40YkJiZiyJAhWLZsGWpqalw+ltlshslkcriEEhP7oBARETmQ/YlYVFSEMWPGoLGxEd26dcPmzZsxePDgdsetX78e0dHRmDNnjsP18+fPR1ZWFlJSUnDw4EEsX74cBw4cwNatW50+5urVq7Fy5Uq5Q1UMqUiWfVCIiIhEKkEQBDknNDU14cyZM6iqqsKmTZvw97//Hbt27WoXpAwaNAhTp07FmjVrXN5fQUEBRo4ciYKCAowYMaLDY8xmM8xms/3vJpMJ6enpqK6uRkxMjJzhB6WcJz9FjbkZ25eOR58e3QI9HCIiIr8wmUyIjY316PNb9gyKXq+3F8mOHDkS+fn5eOGFF/Dqq6/aj/niiy9w9OhRvPPOO27vb8SIEdDpdCguLnYaoBgMBhgMBrlDVQSrTUCNuaUGhUWyREREAHzQB0UQBIfZDQBYu3YtcnNzMWzYMLfnHzp0CBaLBampqZ0diiLVtgQnAJcZExERSWR9Ij7++OOYPn060tPTUVNTg7y8POzcuROffPKJ/RiTyYR3330Xzz77bLvzT5w4gY0bN+Kmm25CYmIiDh8+jKVLl2L48OEYN25c55+NAkk9UAxaNQxaTYBHQ0REFBxkBSjnz5/H3XffjbKyMsTGxmLo0KH45JNPMHXqVPsxeXl5EAQBd955Z7vz9Xo9Pv/8c7zwwguora1Feno6ZsyYgSeffBIaTXh+OLMHChERUXuyi2SDgZwim2D39YlK3Pn6XvTtEYXPl04I9HCIiIj8Rs7nN/fiCbAazqAQERG1wwAlwEzsgUJERNQOA5QAk4pkY7iCh4iIyI4BSoCxSJaIiKg9BigBJrW550aBRERErRigBJiU4mGTNiIiolYMUAKMKR4iIqL2GKAEmKlBSvFwBoWIiEjCACXAasycQSEiIroSA5QAa51BYYBCREQkYYASYPYaFKZ4iIiI7BigBJAgCK3LjJniISIismOAEkD1TVZYbeJejUzxEBERtWKAEkBSekenUcGo40tBREQk4adiAEkFstFGHVQqVYBHQ0REFDwYoAQQC2SJiIg6xgAlgGrYRZaIiKhDDFACiD1QiIiIOsYAJYCkFA83CiQiInLEACWApJ2MOYNCRETkiAFKALU2aeMMChERUVsMUAKodRUPZ1CIiIjaYoASQK19UDiDQkRE1BYDlAAycZkxERFRh/jV3U8e2/Q9thSVuTymzsxlxkRERB1hgOIHZyrrkZdf6tGxBq0aV6XF+HlEREREysIAxQ8+LDoHALg2Kx5/vHWoy2MTuuk5g0JERHQFBih+8OEBMbVzy/CeyEqMCvBoiIiIlIdFsj528kItDpeZoFWrcOOQlEAPh4iISJEYoPjYh9+Lsyfj+iWie5Q+wKMhIiJSJlkByssvv4yhQ4ciJiYGMTExGDNmDD7++GP77ffeey9UKpXDZfTo0Q73YTab8fDDDyMxMRFRUVGYPXs2zp4965tnEwT+c0CsP5k1LC3AIyEiIlIuWQFKr1698PTTT+O7777Dd999h0mTJuHmm2/GoUOH7MfceOONKCsrs18++ugjh/tYvHgxNm/ejLy8PHz55Zeora3FzJkzYbVaffOMAuhoeQ2KK2qh16gxbUhyoIdDRESkWLKKZGfNmuXw9z/84Q94+eWXsXfvXgwZMgQAYDAYkJLSce1FdXU11q5dizfffBNTpkwBALz11ltIT0/Htm3bcMMNN3jzHILGh9+Lsyc/GdCDK3OIiIg6wesaFKvViry8PNTV1WHMmDH263fu3ImkpCQMGDAAixYtQkVFhf22goICWCwWTJs2zX5dWloasrOzsWfPHqePZTabYTKZHC7BRhAEe/3JrGGpAR4NERGRsskOUIqKitCtWzcYDAY8+OCD2Lx5MwYPHgwAmD59OjZu3Ijt27fj2WefRX5+PiZNmgSz2QwAKC8vh16vR/fu3R3uMzk5GeXl5U4fc/Xq1YiNjbVf0tPT5Q7b7w6dM6HkYh0MWjUmX8X0DhERUWfI7oMycOBA7N+/H1VVVdi0aRMWLFiAXbt2YfDgwbjjjjvsx2VnZ2PkyJHIzMzEli1bMGfOHKf3KQgCVCqV09uXL1+OJUuW2P9uMpmCLkiRZk8mDUpCNwPbyxAREXWG7E9SvV6Pfv36AQBGjhyJ/Px8vPDCC3j11VfbHZuamorMzEwUFxcDAFJSUtDU1ITLly87zKJUVFRg7NixTh/TYDDAYDDIHWqXEdM7XL1DRETkK53+qi8Igj2Fc6XKykqUlpYiNVWsycjNzYVOp8PWrVsxd+5cAEBZWRkOHjyIZ555prND6bSLtWb8bcdx2efVmZtx9nIDIvUaTByY5IeRERERhRdZAcrjjz+O6dOnIz09HTU1NcjLy8POnTvxySefoLa2FitWrMCtt96K1NRUnDp1Co8//jgSExNxyy23AABiY2OxcOFCLF26FAkJCYiPj8eyZcuQk5NjX9UTSKYGC/7x1Smvz586OBkReo3vBkRERBSmZAUo58+fx913342ysjLExsZi6NCh+OSTTzB16lQ0NDSgqKgIGzZsQFVVFVJTUzFx4kS88847iI6Ott/H888/D61Wi7lz56KhoQGTJ0/GunXroNEE/oM9LlKPhyb29epcg1aDedcEV10MERGRUqkEQRACPQi5TCYTYmNjUV1djZiYmEAPh4iIiDwg5/Obe/EQERFR0GGAQkREREGHAQoREREFHQYoREREFHQYoBAREVHQYYBCREREQYcBChEREQUdBihEREQUdBigEBERUdBhgEJERERBhwEKERERBR0GKERERBR0GKAQERFR0NEGegDekDZgNplMAR4JEREReUr63JY+x11RZIBSU1MDAEhPTw/wSIiIiEiumpoaxMbGujxGJXgSxgQZm82Gc+fOITo6GiqVyqf3bTKZkJ6ejtLSUsTExPj0voNBqD8/IPSfI5+f8oX6c+TzUz5/PUdBEFBTU4O0tDSo1a6rTBQ5g6JWq9GrVy+/PkZMTEzI/scDQv/5AaH/HPn8lC/UnyOfn/L54zm6mzmRsEiWiIiIgg4DFCIiIgo6DFCuYDAY8OSTT8JgMAR6KH4R6s8PCP3nyOenfKH+HPn8lC8YnqMii2SJiIgotHEGhYiIiIIOAxQiIiIKOgxQiIiIKOgwQCEiIqKgE5YByksvvYSsrCwYjUbk5ubiiy++cHn8rl27kJubC6PRiD59+uCVV17popHKs3r1alxzzTWIjo5GUlISfvrTn+Lo0aMuz9m5cydUKlW7yw8//NBFo5ZnxYoV7caakpLi8hylvH4A0Lt37w5fj4ceeqjD44P99du9ezdmzZqFtLQ0qFQq/Pvf/3a4XRAErFixAmlpaYiIiMCECRNw6NAht/e7adMmDB48GAaDAYMHD8bmzZv99Azcc/UcLRYLHn30UeTk5CAqKgppaWm45557cO7cOZf3uW7dug5f18bGRj8/m/bcvYb33ntvu3GOHj3a7f0q5TUE0OFroVKp8Kc//cnpfQbLa+jJ50Kw/h6GXYDyzjvvYPHixfjd736Hffv24frrr8f06dNx5syZDo8vKSnBTTfdhOuvvx779u3D448/jt/85jfYtGlTF4/cvV27duGhhx7C3r17sXXrVjQ3N2PatGmoq6tze+7Ro0dRVlZmv/Tv378LRuydIUOGOIy1qKjI6bFKev0AID8/3+G5bd26FQBw++23uzwvWF+/uro6DBs2DC+++GKHtz/zzDN47rnn8OKLLyI/Px8pKSmYOnWqfb+tjnz99de44447cPfdd+PAgQO4++67MXfuXHzzzTf+ehouuXqO9fX1KCwsxBNPPIHCwkK89957OHbsGGbPnu32fmNiYhxe07KyMhiNRn88BZfcvYYAcOONNzqM86OPPnJ5n0p6DQG0ex3eeOMNqFQq3HrrrS7vNxheQ08+F4L291AIM9dee63w4IMPOlw3aNAg4bHHHuvw+N/+9rfCoEGDHK574IEHhNGjR/ttjL5SUVEhABB27drl9JgdO3YIAITLly933cA64cknnxSGDRvm8fFKfv0EQRAeeeQRoW/fvoLNZuvwdiW9fgCEzZs32/9us9mElJQU4emnn7Zf19jYKMTGxgqvvPKK0/uZO3eucOONNzpcd8MNNwjz5s3z+ZjluvI5duTbb78VAAinT592esw//vEPITY21reD84GOnt+CBQuEm2++Wdb9KP01vPnmm4VJkya5PCZYX8MrPxeC+fcwrGZQmpqaUFBQgGnTpjlcP23aNOzZs6fDc77++ut2x99www347rvvYLFY/DZWX6iurgYAxMfHuz12+PDhSE1NxeTJk7Fjxw5/D61TiouLkZaWhqysLMybNw8nT550eqySX7+mpia89dZbuP/++91uiqmk109SUlKC8vJyh9fHYDBg/PjxTn8fAeevqatzgkl1dTVUKhXi4uJcHldbW4vMzEz06tULM2fOxL59+7pmgF7YuXMnkpKSMGDAACxatAgVFRUuj1fya3j+/Hls2bIFCxcudHtsML6GV34uBPPvYVgFKBcvXoTVakVycrLD9cnJySgvL+/wnPLy8g6Pb25uxsWLF/021s4SBAFLlizBddddh+zsbKfHpaam4rXXXsOmTZvw3nvvYeDAgZg8eTJ2797dhaP13KhRo7BhwwZ8+umneP3111FeXo6xY8eisrKyw+OV+voBwL///W9UVVXh3nvvdXqM0l6/tqTfOTm/j9J5cs8JFo2NjXjsscdw1113udyAbdCgQVi3bh0++OAD/POf/4TRaMS4ceNQXFzchaP1zPTp07Fx40Zs374dzz77LPLz8zFp0iSYzWan5yj5NVy/fj2io6MxZ84cl8cF42vY0edCMP8eKnI348668tuoIAguv6F2dHxH1weTX//61/j+++/x5Zdfujxu4MCBGDhwoP3vY8aMQWlpKf785z/jJz/5ib+HKdv06dPtP+fk5GDMmDHo27cv1q9fjyVLlnR4jhJfPwBYu3Ytpk+fjrS0NKfHKO3164jc30dvzwk0i8WCefPmwWaz4aWXXnJ57OjRox0KTceNG4cRI0ZgzZo1+Otf/+rvocpyxx132H/Ozs7GyJEjkZmZiS1btrj8EFfiawgAb7zxBubPn++2liQYX0NXnwvB+HsYVjMoiYmJ0Gg07SK8ioqKdpGgJCUlpcPjtVotEhIS/DbWznj44YfxwQcfYMeOHejVq5fs80ePHh2U39Q6EhUVhZycHKfjVeLrBwCnT5/Gtm3b8POf/1z2uUp5/aTVV3J+H6Xz5J4TaBaLBXPnzkVJSQm2bt0qe/t6tVqNa665RhGva2pqKjIzM12OVYmvIQB88cUXOHr0qFe/l4F+DZ19LgTz72FYBSh6vR65ubn2lRGSrVu3YuzYsR2eM2bMmHbHf/bZZxg5ciR0Op3fxuoNQRDw61//Gu+99x62b9+OrKwsr+5n3759SE1N9fHo/MNsNuPIkSNOx6uk16+tf/zjH0hKSsKMGTNkn6uU1y8rKwspKSkOr09TUxN27drl9PcRcP6aujonkKTgpLi4GNu2bfMqMBYEAfv371fE61pZWYnS0lKXY1XaayhZu3YtcnNzMWzYMNnnBuo1dPe5ENS/hz4rt1WIvLw8QafTCWvXrhUOHz4sLF68WIiKihJOnTolCIIgPPbYY8Ldd99tP/7kyZNCZGSk8F//9V/C4cOHhbVr1wo6nU7417/+Fain4NQvf/lLITY2Vti5c6dQVlZmv9TX19uPufL5Pf/888LmzZuFY8eOCQcPHhQee+wxAYCwadOmQDwFt5YuXSrs3LlTOHnypLB3715h5syZQnR0dEi8fhKr1SpkZGQIjz76aLvblPb61dTUCPv27RP27dsnABCee+45Yd++ffYVLE8//bQQGxsrvPfee0JRUZFw5513CqmpqYLJZLLfx9133+2wyu6rr74SNBqN8PTTTwtHjhwRnn76aUGr1Qp79+7t8ucnCK6fo8ViEWbPni306tVL2L9/v8Pvpdlstt/Hlc9xxYoVwieffCKcOHFC2Ldvn3DfffcJWq1W+Oabb4Lq+dXU1AhLly4V9uzZI5SUlAg7duwQxowZI/Ts2TNkXkNJdXW1EBkZKbz88ssd3kewvoaefC4E6+9h2AUogiAIf/vb34TMzExBr9cLI0aMcFiGu2DBAmH8+PEOx+/cuVMYPny4oNfrhd69ezv9DxpoADq8/OMf/7Afc+Xz++Mf/yj07dtXMBqNQvfu3YXrrrtO2LJlS9cP3kN33HGHkJqaKuh0OiEtLU2YM2eOcOjQIfvtSn79JJ9++qkAQDh69Gi725T2+knLoK+8LFiwQBAEcYnjk08+KaSkpAgGg0H4yU9+IhQVFTncx/jx4+3HS959911h4MCBgk6nEwYNGhTQgMzVcywpKXH6e7ljxw77fVz5HBcvXixkZGQIer1e6NGjhzBt2jRhz549Xf/kBNfPr76+Xpg2bZrQo0cPQafTCRkZGcKCBQuEM2fOONyHkl9DyauvvipEREQIVVVVHd5HsL6GnnwuBOvvoarlCRAREREFjbCqQSEiIiJlYIBCREREQYcBChEREQUdBihEREQUdBigEBERUdBhgEJERERBhwEKERERBR0GKERERBR0GKAQERFR0GGAQkREREGHAQoREREFHQYoREREFHT+Px4ZUkNc5iJDAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "d = 31\n",
    "\n",
    "# polynomial basis ??\n",
    "basis = [lambda x, k=k: x**k for k in range(d)] \n",
    "\n",
    "# piecewise constant function\n",
    "bins = np.linspace(0, 20, d+1)\n",
    "bins[-1] = np.inf\n",
    "basis = [lambda x, k=k: float(bins[k] <= x <= bins[k+1]) for k in range(d)] \n",
    "\n",
    "# Fourier basis\n",
    "#basis = [lambda x, k=k: np.cos(k*x/20*2*np.pi) for k in range(d//2)]\n",
    "#basis = basis + [lambda x, k=k: np.sin(k*x/20 * 2 * np.pi) for k in range(1, d-len(basis)+1)]\n",
    "\n",
    "coeffs = np.zeros(2*d)\n",
    "\n",
    "def Qfunction(x, a):\n",
    "    if a == \"Keep\":\n",
    "        return sum([coeffs[k] * basis[k](x) for k in range(d)])\n",
    "    else:\n",
    "        return sum([coeffs[d+k] * basis[k](x) for k in range(d)])\n",
    "\n",
    "for j in range(20):\n",
    "    X = []\n",
    "    Y = []\n",
    "    for k in range(1000):\n",
    "        x = np.random.uniform(0, 20)\n",
    "        a = np.random.choice([\"Keep\", \"Replace\"])\n",
    "        c = maintenanceCost(x) if a == \"Keep\" else replacementCost\n",
    "        newWear = (x if a == \"Keep\" else 0.) + randomWearIncrease() \n",
    "        if a == \"Keep\":\n",
    "            X.append([f(x) for f in basis] + [0 for f in basis])\n",
    "        else:\n",
    "            X.append([0 for f in basis] + [f(x) for f in basis])\n",
    "        z = c + gamma * min(Qfunction(newWear, \"Keep\"), Qfunction(newWear, \"Replace\"))\n",
    "        Y.append(z)\n",
    "    \n",
    "    coeffs = np.linalg.lstsq(X, Y, rcond=None)[0]\n",
    "\n",
    "\n",
    "plt.plot(xx, [Qfunction(x, \"Keep\") for x in xx])\n",
    "plt.plot(xx, [Qfunction(x, \"Replace\") for x in xx])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "c04e4b37-01de-4a31-a933-21ff68c6f31c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2021.7991281396571"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "policy = lambda x: \"Keep\" if Qfunction(x, \"Keep\") <= Qfunction(x, \"Replace\") else \"Replace\"\n",
    "\n",
    "evaluatePolicy(policy)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "effa643b-3d7c-4652-9c85-10a9522621c8",
   "metadata": {},
   "source": [
    "## Evaluation: \n",
    "\n",
    "Upload into folder https://plmbox.math.cnrs.fr/u/d/2666573c3a714a2bbdce/ a file <Yourname.ipynb> adding \n",
    "\n",
    "  - a discretization of the above MDP with a planning algorithm\n",
    "  \n",
    "  - a better basis for the Q-value function, giving a better approximation of the Q-function (compare with the discretized solution)\n",
    "  \n",
    "  - a fitted Q-learning algorithm that converges\n",
    "\n",
    "The computational cost of your algorithms will be taken into account."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "17c392fd-038e-4182-938b-3d93f90df3e3",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f837a9a9-e39a-41ee-8e4b-98f5d20d3578",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python3UMPA",
   "language": "python",
   "name": "python3umpa"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}
