solutions_02_channel_estimation_digital_mimo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Inteligência Artificial e Aprendizado de Máquina Aplicados a Redes 5G e 6G.**\n",
    "*Aldebaro Klautau* (UFPA). Minicurso 5 do SBrT - 25 de setembro de 2022."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a href=\"https://colab.research.google.com/github/aldebaro/ai6g/blob/main/solutions_02_channel_estimation_digital_mimo.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Clone the repository if running in Colab and install all the dependencies\n",
    "if 'google.colab' in str(get_ipython()):\n",
    "    from google.colab import drive\n",
    "    drive.mount('/content/drive/')\n",
    "    import sys\n",
    "    import os\n",
    "    try:\n",
    "      !git clone https://github.com/aldebaro/ai6g.git\n",
    "    except:\n",
    "      print(\"ai6g is already in the contents\")\n",
    "    %cd ai6g\n",
    "    !ln -s /content/drive/MyDrive/ai6g_files/files_02_channel/* ./files_02_channel\n",
    "    !pip install -r requirements.txt "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'EM-BG-GAMP algorithm')"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAj4AAAHFCAYAAADyj/PrAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/NK7nSAAAACXBIWXMAAA9hAAAPYQGoP6dpAABkWElEQVR4nO3deVxU5f4H8M8ZGIZ9HVYBAREVcUVLXEEDNUvL3LJFc0vTvF2zn6nllpYtZmWpdXNrcanUXCtwyRUVFfddQVAWBRSQZRiY8/uDmBwBGZThwMzn/XrNvc6ZM898+Tbqx3Oecx5BFEURRERERCZAJnUBRERERLWFwYeIiIhMBoMPERERmQwGHyIiIjIZDD5ERERkMhh8iIiIyGQw+BAREZHJYPAhIiIik8HgQ0RERCaDwYeoFq1cuRKCIFT6+Pvvv7X7+vn5QRAEhIeHVzjWDz/8UOH7KvPgZ9nY2KBZs2aYPXs28vLyyu2vUqnwzTffoFu3bnBxcYFcLoeLiwvCw8Px7bffIjc3t1o/e0JCAiZOnIhmzZrBxsYGlpaW8PPzw8svv4zdu3ejspvIf/XVVxAEASEhIVX+bMOHD6/w9Tlz5mj3SUxM1G4fPny4Tk8UCgWaNGmCmTNnorCwsFo/38MIgoBZs2bV2HjVFR4ervM9ys/Px6xZsyr83syaNQuCICAjI6P2CiSqReZSF0BkilasWIGmTZuW2x4cHKzz3M7ODnv37sXVq1fRqFEjndeWL18Oe3t75OTk6P25AwYMwNtvvw0AuHfvHvbs2YM5c+bg1KlTWL9+vXa/27dvo1evXjhz5gyGDRuGiRMnws3NDZmZmdi1axf+7//+D/v378ePP/6o1+du3rwZQ4cOhVKpxNixY9G2bVsoFApcuXIFv/32G7p3744dO3agR48e5d67fPlyAMDZs2dx+PBhPPnkkxV+hp2dHX799VcsWrQIdnZ22u2iKGLlypWV9srKygq7du0CANy5cwdr1qzBnDlzcOHCBaxbt06vn6+uW7x4sc7z/Px8zJ49GwAqDdZERkskolqzYsUKEYAYFxdX5b4NGzYUe/fuLXp7e4vTpk3Tee3KlSuiIAji6NGjRQDi7t27qxwPgDh+/Phy21955RVRJpOJBQUF2m1RUVGiXC4X9+zZU+FYGRkZ4o8//ljlZ5bVam1tLbZv317Mzs6ucJ/du3eLJ06cKLc9Li5OBCD26dNHBCCOHj26wvcDEF9++WXRyspK/O6773Re27Fjh/a9AMSEhATta8OGDRNtbGzKjdelSxcRgHjjxg29fsaqABBnzpxZI2NVR15eXoXbb9++XWlNM2fOFAGIt2/fNnB1RNLgqS6iOkwmk+HVV1/FqlWroNFotNuXL18OHx8fPPXUU4/9GQ4ODhAEAWZmZgCAuLg4REdHY8yYMejatWuF73FxccHLL7+s1/iff/458vPzsXjxYtjb21e4T3h4OFq1alVu+7JlywAA8+fPR8eOHbF27Vrk5+dX+nM8//zz2iNEZZYvX45OnTohKChIr3oBoEOHDgCA69evV7rP7du38cYbbyA4OBi2trZwc3ND9+7dsW/fPr0+Y//+/QgLC4OlpSUaNGiA999/H99//32503EajQaffPIJmjZtCoVCATc3N7z66qu4ceOGznjh4eEICQnB3r170bFjR1hbW2PEiBHa18qO7CQmJsLV1RUAMHv27EpPE6anp+PFF1+Eg4MD3N3dMWLECGRnZ+vsIwgCJkyYgBUrVqBJkyawsrJCu3btcOjQIYiiiE8//RT+/v6wtbVF9+7dceXKFb16Q2RIDD5EEigpKUFxcbHOo6SkpMJ9R4wYgZSUFPz111/a965atQrDhw+HTFa938KiKGo/7+7du9i0aRNWrVqFIUOGQC6XAwBiYmIAAH379n2Mn/BfMTEx8PT0RLt27ar1voKCAqxZswbt27dHSEgIRowYgdzcXPz666+VvmfkyJE4dOgQzp8/DwC4e/cuNmzYgJEjR1brs8v+gi4LCBXJysoCAMycORPbtm3DihUrEBAQgPDw8CrnXJ06dQqRkZHIz8/HqlWrsHTpUhw/fhzz5s0rt++4ceMwZcoUREZGYvPmzfjggw/w559/omPHjuXm4aSmpuLll1/G0KFDsX37drzxxhvlxvP09MSff/4JoLRfsbGxiI2Nxfvvv6+z3wsvvICgoCCsX78e7777LlavXo3//ve/5cbbunUrvv/+e8yfPx9r1qxBbm4u+vTpg7fffhsHDhzA119/je+++w7nzp3DCy+8UOlcLqJaI/ERJyKTUnaqq6KHmZmZzr4NGzYU+/TpI4qiKHbr1k0cMGCAKIqiuG3bNlEQBDEhIUH89ddfq3Wqq6JH7969xXv37mn3Gzt2rAhAvHDhgs77NRqNqFartY/i4mK9fmZLS0uxQ4cO5baXlJTojFdSUqLz+g8//CACEJcuXSqKoijm5uaKtra2YpcuXSr82caPHy9qNBrR399fnDx5siiKovjNN9+Itra2Ym5urvjpp59WeqqrrIbbt2+LX375pSgIgti+fXu9fr4yxcXFolqtFnv06CE+//zz5eq7/7TSwIEDRRsbG53TSSUlJWJwcLBOjefPnxcBiG+88YbOeIcPHxYB6JwC7datmwhA3LlzZ7naunXrJnbr1k37XJ9TXZ988onO9jfeeEO0tLQUNRqNzs/l4eGh8/35/fffRQBi69atdfb94osvRADiqVOnyn0mUW3iER8iCfzwww+Ii4vTeRw+fLjS/UeMGIHNmzcjMzMTy5YtQ0REBPz8/MrtJ953RKfscb9BgwZpP2/v3r346quvcPToUfTq1QsqleqhNW/atAlyuVz7cHBw0L724BGs+0/LVaZ///46402cOFHn9WXLlsHKygpDhgwBANja2mLgwIHYt28fLl++XOGYZadsfvzxRxQXF2PZsmUYNGgQbG1tK60jLy9PW4Orqyveeust9O7dGxs3bqzyZ1i6dCnatm0LS0tLmJubQy6XY+fOndojTpXZs2cPunfvDqVSqd0mk8kwaNAgnf12794NAOVOQz3xxBNo1qwZdu7cqbPdyckJ3bt3r7JufTx4xK9ly5YoLCzErVu3dLZHRETAxsZG+7xZs2YAgN69e0MQhHLbH3b6kKg2MPgQSaBZs2Zo166dziM0NLTS/QcMGABLS0ssXLgQW7ZsqfTUzZ49e3TChFwu15kv4urqqv28Ll264M0338RXX32F/fv3Y+XKlQAAX19fAOX/ggoPD9eGpmeeeUbntR49euh8ZtnckrLxKvrLbsGCBdrxHnTlyhXs3bsXffr0gSiKuHv3Lu7evYsBAwYAQLl5PPd77bXXcPv2bXz44Yc4fvx4lae5rKystHWcOnUKd+/exbZt29CgQYOHvu/zzz/HuHHj8OSTT2L9+vU4dOgQ4uLi0KtXLxQUFDz0vZmZmXB3dy+3/cFtmZmZAEpPTz3Iy8tL+3qZivZ7VC4uLjrPFQoFAJT72ZydnXWeW1hYPHR7Td4mgOhR8HJ2onrA2toaQ4YMwUcffQR7e3v079+/wv1CQ0PLBQkvL6+Hjt2yZUsAwMmTJwEAkZGRmDZtGjZv3oyoqCjtfo6Ojtp5Og/+pfjgfX3uP5IRGRmJb775BkePHtWZ5/Pg5fn3W758OURRxG+//Ybffvut3OurVq3C3LlztROy71c26Xv27Nlo0qQJOnbs+NCfXyaTVXv+EQD89NNPCA8Px5IlS3S263N/IxcXF6Snp5fbnpaWVm4/oHTujre3t85rKSkpOn0GoHOEhYgqxuBDVE+MGzcO6enp6NatGywtLSvcx87Ortp/iZ84cQIA4ObmBgBo164doqKi8L///Q+DBw9Gly5dqhyjSZMmlb723//+FytWrMD48eOxY8cOnXvsVKRs8najRo3w/fffl3t969atWLBgAf74449yR57KvP3227CyssLAgQOrrP1Rld3w8H6nTp1CbGwsfHx8Hvrebt26Yfv27cjIyNCGF41GU27idtlpq59++gnt27fXbo+Li8P58+cxffr0R6q9sqM3RKaAwYdIAmfOnCk3/wYoPQpS2ZVErVu3xu+///5Yn5ueno5Dhw4BKD3lcOLECcydOxeOjo547bXXtPv99NNP6NmzJ5566ikMHz4cPXv2hJubG3JycnDq1Cns2LGj0kvTK/qZ1qxZgxdffBEtWrTAuHHjtDcwvHXrFqKjowFAO94ff/yBlJQUfPzxxxXeXC8kJARff/01li1bVmnwiYqK0jlaZQjPPPMMPvjgA8ycORPdunXDxYsXMWfOHPj7+1f43/Z+06dPx5YtW9CjRw9Mnz4dVlZWWLp0qfYO2mVX6zVp0gRjxozBokWLIJPJ0Lt3byQmJuL999+Hj49PhVdZ6cPOzg4NGzbEpk2b0KNHDzg7O0OpVFY4b4zI2DD4EEng/pBxv//9738YNWqUwT73/lNHcrkcPj4+6Nu3L6ZPn46GDRtq93N1dUVsbCz+97//Yd26dfjll19w79492Nvbo3nz5pg4cWK16uzbty9Onz6NL774AitWrMDs2bOh0Wjg4eGBJ554Ahs3bkS/fv0AlE5qtrCwqLRHSqUSzz//PH777Tekp6dXOFemNkyfPh35+flYtmwZPvnkEwQHB2Pp0qXYuHFjlZezt2rVCjExMZg8eTJeffVVODk54ZVXXkG3bt0wZcoUnYnjS5YsQaNGjbBs2TJ88803cHBwQK9evfDRRx+VO+VYHcuWLcM777yDvn37QqVSYdiwYdp5XkTGTBBF3lSBiKguiIqKQmJiIi5duiR1KURGi0d8iIgkMGnSJLRp0wY+Pj7IysrCzz//jJiYGO3dqonIMBh8iIgkUFJSghkzZiAtLQ2CICA4OBg//vij3kuBENGj4akuIiIiMhm8gSERERGZDAYfIiIiMhkMPkRERGQyOLn5ARqNBikpKbCzs+Pt34mIiOoJURSRm5sLLy8v7U1AK8Lg84CUlJQqbzdPREREdVNycnK5te3ux+DzgLJ1hJKTk/W+Jb8+1Go1oqOjERUVBblcXmPjGiP2qnrYL/2xV/pjr/THXunPkL3KycmBj49PlesBMvg8oOz0lr29fY0HH2tra9jb2/M3RhXYq+phv/THXumPvdIfe6W/2uhVVdNUOLmZiIiITAaDDxEREZkMBh8iIiIyGQw+REREZDIYfIiIiMhkMPgQERGRyWDwISIiIpPB4ENEREQmg8GHiIiITAaDDxEREZkMBh8iIiIyGQw+tSQ1uxCXswWkZhdKXQoREZHJYvCpBevikhC+YC++PmeG8AV7sS4uSeqSiIiITBKDj4GlZhdg6obT0IilzzUiMG3DGaRmF0hbGBERkQli8DGwhIw8begpUyKKSMzIl6YgIiIiE8bgY2D+ShvIBN1tAgA/pbUk9RAREZkyBh8D83Swwkf9W+iEHzMzAUXFGumKIiIiMlEMPrVgcHtf/P12V4xvVoJQX0cUl4iYuuE0RFGs+s1ERERUYxh8aomngyWCHEV8/EIILOUyHLyaiV+OJktdFhERkUlh8KllDZ2tMSkyCAAwd9t53MrhfX2IiIhqC4OPBEZ08keLBg7ILSzGjE1npS6HiIjIZBhV8Jk1axYEQdB5eHh4SF1WOeZmMnz8QkuYywT8eTYNf55JlbokIiIik2BUwQcAmjdvjtTUVO3j9OnTUpdUoWAve7zeLQAA8P6ms8guUEtcERERkfEzuuBjbm4ODw8P7cPV1VXqkir1ZvfGCHC1we1cFT7afl7qcoiIiIyeudQF1LTLly/Dy8sLCoUCTz75JD788EMEBARUur9KpYJKpdI+z8nJAQCo1Wqo1TV3FKZsrPvHNAMwt28whi6Lw9q4ZPQJcUeHAOca+8z6qqJeUeXYL/2xV/pjr/THXunPkL3Sd0xBNKKbyfzxxx/Iz89HUFAQ0tPTMXfuXFy4cAFnz56Fi4tLhe+ZNWsWZs+eXW776tWrYW1dO3dX/uWaDAfSZVAqRExpVQILs1r5WCIiIqORn5+PoUOHIjs7G/b29pXuZ1TB50F5eXlo1KgR/u///g+TJk2qcJ+Kjvj4+PggIyPjoY2rLrVajZiYGERGRkIul+u8lltYjKcXHUBajgqjOvthSs+gGvvc+uhhvaLy2C/9sVf6Y6/0x17pz5C9ysnJgVKprDL4GN2prvvZ2NigRYsWuHz5cqX7KBQKKBSKctvlcrlBvsAVjessl2Pe8y0wctVRLD+QiH6tvdHC26HGP7u+MdR/A2PFfumPvdIfe6U/9kp/huiVvuMZ3eTm+6lUKpw/fx6enp5Sl1KlHs3c8WwrL2hE4P/Wn4K6hGt5ERER1TSjCj6TJ0/Gnj17kJCQgMOHD2PAgAHIycnBsGHDpC5NLzOfDYajtRznU3Pw3d5rUpdDRERkdIwq+Ny4cQMvvvgimjRpgv79+8PCwgKHDh1Cw4YNpS5NL0pbBd7vEwwA+HLnZVy9fU/iioiIiIyLUc3xWbt2rdQlPLb+bRvg9xM3se9yBqZuOI21oztAJhOkLouIiMgoGNURH2MgCAI+fL4FrC3McCQhC2vikqQuiYiIyGgw+NRBPs7WmBzVBAAwf/sFpGVzBXciIqKawOBTRw3r6IfWPo7IVRXjvd/PwIhvt0RERFRrGHzqKDOZgE8GtITcTMCO8+nYdporuBMRET0uBp86LMjdDm+EBwIAZm0+izt5RRJXREREVL8x+NRxb0Q0QmM3W2TcK8LcbVzBnYiI6HEw+NRxCnMzzH+hJQQBWH/8BvZeui11SURERPUWg089ENrQCcPC/AAA0zaeRp6qWNqCiIiI6ikGn3rinZ5N0MDRCjfuFODzmEtSl0NERFQvMfjUEzYKc8x7PgQAsOJAAk4k35W2ICIionqIwaceCW/ihufbNIBGBKb8dgpFxVzBnYiIqDoYfOqZ958JhrONBS6m52LpnqtSl0NERFSvMPjUM842Fpj5bOkK7l/vuoIrt3IlroiIiKj+YPCph/q28kL3pm4oKtFgyvrT0Gi4nAUREZE+GHzqIUEQMPe5ENgqzHHs+h18vfsKDl7NQGp2gdSlERER1WkMPvWUl6MVpvQqXcH985hLGPq/w+g0fxfWxSVJXBkREVHdxeBTj3Vv6qbzXCMC0zac4ZEfIiKiSjD41GPXs/LLbSsRRSRmlN9OREREDD71mr/SBjJBd5tMAPyU1tIUREREVMcx+NRjng5W+Kh/C53w06OpOzwdrKQrioiIqA5j8KnnBrf3xYF3u2Ni90AAwN7Lt5FcwSkwIiIiYvAxCp4OVvhvZBDCAlygKtbgg63npC6JiIioTmLwMRKCIGB2v+YwlwmIPpeOvy/ekrokIiKiOofBx4gEudtheEc/AMDsLeegKi6RtiAiIqI6hsHHyPznqcZwtVMgISMP3+9LkLocIiKiOoXBx8jYWcox7emmAEoXMU25y5sZEhERlWHwMULPtW6A9n5OKFCXYN6281KXQ0REVGcw+BghQRAwu28IZAKw7XQqDlzJkLokIiKiOoHBx0gFe9nj1TA/AMDMzWdRVKyRtiAiIqI6gMHHiP03MgguNha4cuseVh1MlLocIiIiyTH4GDEHKzmm9C6d6PzFjktIzymUuCIiIiJpMfgYuQFtvdHG1xF5RSX4cDsnOhMRkWlj8DFyMpmAOX1DIAjAphMpOHwtU+qSiIiIJMPgYwJaeDtg6BO+AEonOheXcKIzERGZJgYfEzE5qgkcreW4kJaLHw9dl7ocIiIiSRhl8Fm8eDH8/f1haWmJ0NBQ7Nu3T+qSJOdkY4H/61k60fnz6Eu4nauSuCIiIqLaZ3TBZ926dXjrrbcwffp0xMfHo0uXLujduzeSkpKkLk1yg9v7oKW3A3JVxZj/xwWpyyEiIqp1Rhd8Pv/8c4wcORKjRo1Cs2bN8MUXX8DHxwdLliyRujTJmckEzO7bHACw/vgNHLueJXFFREREtctc6gJqUlFREY4dO4Z3331XZ3tUVBQOHjxY4XtUKhVUqn9P++Tk5AAA1Go11Gp1jdVWNlZNjvkoQjxtMTC0AX49dhPv/34GG8Z2gJlMkLSmB9WVXtUX7Jf+2Cv9sVf6Y6/0Z8he6TumIIqiWOOfLpGUlBQ0aNAABw4cQMeOHbXbP/zwQ6xatQoXL14s955Zs2Zh9uzZ5bavXr0a1tbWBq1XKvfUwNx4MxSUCBjoX4LOHkbzFSAiIhOVn5+PoUOHIjs7G/b29pXuZ1RHfMoIgu4RDFEUy20rM3XqVEyaNEn7PCcnBz4+PoiKinpo46pLrVYjJiYGkZGRkMvlNTbuoyrySMLsrRcQnabA5MGd4WxjIXVJWnWtV3Ud+6U/9kp/7JX+2Cv9GbJXZWdsqmJUwUepVMLMzAxpaWk622/dugV3d/cK36NQKKBQKMptl8vlBvkCG2rc6nq1YwB+PZaCc6k5WLjzKua/0FLqksqpK72qL9gv/bFX+mOv9Mde6c8QvdJ3PKOa3GxhYYHQ0FDExMTobI+JidE59UWlE53n9Cud6LzuaDJOJN+VtiAiIqJaYFTBBwAmTZqE77//HsuXL8f58+fx3//+F0lJSRg7dqzUpdU57fyc0b9tA4giMHPTGWg0nOtDRETGzahOdQHA4MGDkZmZiTlz5iA1NRUhISHYvn07GjZsKHVpddK7vZsi5mw6Tt7Ixi9HkzHkn6UtiIiIjJHRHfEBgDfeeAOJiYlQqVQ4duwYunbtKnVJdZabnSXeigwCAHz85wXczS+SuCIiIiLDMcrgQ9XzalhDBLnb4k6+GguiL0ldDhERkcEw+BDkZjLM6RcCAPj58HWcuZktcUVERESGweBDAIAOAS7o28oLGhGYwYnORERkpBh8SGva081gY2GG40l3sSH+ptTlEBER1TgGH9LycLDExB6NAQDz/ziP7AKuO0NERMaFwYd0vNbJH41cbZBxrwhf7OBEZyIiMi4MPqTDwlyG2X1LJzqvOpiIdXFJSM0ukLgqIiKimsHgQ+V0bqxEiwb20IjAlPWn0Wn+LqyLS5K6LCIiosfG4EPlpGYX4GzKv6vcakRg2oYzPPJDRET1HoMPlZOQkYcHr2YvEUUkZuRLUxAREVENYfChcvyVNpAJutsEAfBTWktTEBERUQ1h8KFyPB2s8FH/FjAT/k0/jtZyKG0VElZFRET0+Bh8qEKD2/ti/7sRWPlaezhZW+BOnhq/Hr0hdVlERESPhcGHKuXpYIXwJm6Y2CMQALBo12UUqkskroqIiOjRMfhQlYY+6QsvB0ukZhdi9WFe1k5ERPUXgw9VSWFuhjf/Wcpi8d9XkF9ULHFFREREj4bBh/QyINQbDV2skXGvCCsPJkpdDhER0SNh8CG9yM1keOup0qM+3+65hpxCLmBKRET1D4MP6a1vqwYIdLNFdoEay/YlSF0OERFRtTH4kN7MZAImRQYBAJbtT8CdvCKJKyIiIqoeBh+qll7NPdDcyx73VMVYuveq1OUQERFVC4MPVYtMJuDtqNKjPqsOJuJWTqHEFREREemPwYeqLaKJG9r4OqJQrcHiv3nUh4iI6g8GH6o2QRDwTlQTAMDqw0m4ebdA4oqIiIj0w+BDj6RjoBJhAS4oKtFg0c7LUpdDRESkFwYfemSTe5bO9fn12A0kZuRJXA0REVHVGHzokYU2dEZEE1eUaER8yaM+RERUDzD40GOZFFk61+f3EzdxKT1X4mqIiIgejsGHHksLbwf0au4BUQQWxlySuhwiIqKHYvChxzYpKgiCAPxxJg1nbmZLXQ4REVGlGHzosQW526FfKy8AwILoixJXQ0REVDkGH6oR/3kqCGYyAbsv3sax61lSl0NERFQhBh+qEf5KGwxo6w0A+OwvzvUhIqK6icGHaszEpxrDwkyG2GuZOHglQ+pyiIiIymHwoRrTwNEKLz7hAwD4LPoiRFGUuCIiIiJdRhV8/Pz8IAiCzuPdd9+VuiyTMj4iEJZyGY4n3cXui7ekLoeIiEiHUQUfAJgzZw5SU1O1j/fee0/qkkyKm70lhoX5AQAWRF+CRsOjPkREVHcYXfCxs7ODh4eH9mFrayt1SSbn9W6NYGNhhrMpOfjrbJrU5RAREWmZS11ATfv444/xwQcfwMfHBwMHDsQ777wDCwuLSvdXqVRQqVTa5zk5OQAAtVoNtVpdY3WVjVWTY9ZVdhYChndsiG/+vobPoi8iIsgFZjJB7/ebUq9qAvulP/ZKf+yV/tgr/RmyV/qOKYhGNAN14cKFaNu2LZycnHDkyBFMnToV/fr1w/fff1/pe2bNmoXZs2eX27569WpYW1sbslyjll8MfHDcDPklAl4OLEF7V6P5mhERUR2Un5+PoUOHIjs7G/b29pXuV+eDT2XB5H5xcXFo165due3r16/HgAEDkJGRARcXlwrfW9ERHx8fH2RkZDy0cdWlVqsRExODyMhIyOXyGhu3Llu65xoW7LgCHycr/PWfTpCb6Xdm1RR79TjYL/2xV/pjr/THXunPkL3KycmBUqmsMvjU+VNdEyZMwJAhQx66j5+fX4XbO3ToAAC4cuVKpcFHoVBAoVCU2y6Xyw3yBTbUuHXRiC6NsOpQEpLvFGDTqXS8+IRvtd5vSr2qCeyX/tgr/bFX+mOv9GeIXuk7Xp0PPkqlEkql8pHeGx8fDwDw9PSsyZJITzYKc4wLD8QHW8/hq52X8XybBrCUm0ldFhERmTCjuaorNjYWCxcuxIkTJ5CQkIBffvkFr7/+Ovr27Qtf3+odaaCa89KTvvCwt0RqdiHWHEmSuhwiIjJxRhN8FAoF1q1bh/DwcAQHB2PGjBkYPXo01qxZI3VpJs1SboYJ3QMBAN/svor8omKJKyIiIlNW50916att27Y4dOiQ1GVQBQa188G3e68iOasAP8Rex9hujaQuiYiITJTRHPGhusvCXIb/9AgCACzdcxU5hbzXBRERSYPBh2rFc629EOBqg7v5aizfnyB1OUREZKIYfKhWmJvJMCmy9KjPsn0JuJNXJHFFRERkihh8qNY8HeKJph52yFUVY0HMJRy8moHU7AKpyyIiIhPC4EO1RiYT8HZUEwDAT4euY+j/DqPT/F1YF8fL3ImIqHYw+FCtau5lp/NcIwLTNpzhkR8iIqoVDD5UqxIz88ttKxFFJGaU305ERFTTGHyoVvkrbSATdLeZCQL8lNbSFERERCaFwYdqlaeDFT7q3wL3Z58P+4fA08FKspqIiMh0MPhQrRvc3hdbJ3bWHvlp6e0oaT1ERGQ6GHxIEs29HNCzuQcAYF1cssTVEBGRqWDwIckMbu8DANgYfxOF6hKJqyEiIlPA4EOS6dLYFV4OlsguUOOvs2lSl0NERCaAwYckYyYTMLBd6VEfnu4iIqLawOBDkhrYzhuCABy8momkCu7xQ0REVJMYfEhS3k7W6ByoBAD8cpRHfYiIyLAYfEhyQ9r7AgB+PZaM4hKNxNUQEZExY/AhyT0V7AZnGwuk56iw59JtqcshIiIjxuBDklOYm6F/mwYAgLWc5ExERAbE4EN1Qtk9fXZduIVbuSqJqyEiImPF4EN1QmN3O4Q2dEKJRsTG+BSpyyEiIiPF4EN1RtlRn1+P3YQoSlwMEREZJQYfqjP6tPCErcIc17PycTVH6mqIiMgYmT/Km1QqFY4cOYLExETk5+fD1dUVbdq0gb+/f03XRybERmGOZ1t5Ys2RZMTekmGi1AUREZHRqVbwOXjwIBYtWoTff/8dRUVFcHR0hJWVFbKysqBSqRAQEIAxY8Zg7NixsLOzM1TNZMQGt/fFmiPJOJkpILtADaVcLnVJRERkRPQ+1dWvXz8MGDAADRo0wF9//YXc3FxkZmbixo0byM/Px+XLl/Hee+9h586dCAoKQkxMjCHrJiPVytsBTdxtoRYFbDmVKnU5RERkZPQ+4hMVFYVff/0VFhYWFb4eEBCAgIAADBs2DGfPnkVKCq/MoeoTBAEDQxtg7vaLWHf0JoZ3CoAgCFKXRURERkLvIz7jx4+vNPQ8qHnz5oiMjHzkosi09WvlBXNBxIW0XJy5yVnORERUc2rkqq5r167h7Nmz0Gi4zhI9PkdrOVo6l17PvjYuSeJqiIjImFQr+KjVasycORPPPvss5s2bh5KSErz44oto3LgxWrZsiZCQECQmJhqoVDIlYe6lwWfziRQUFJVIXA0RERmLagWfd999F0uWLIG7uzuWL1+O/v37Iz4+HqtXr8batWthbm6O6dOnG6pWMiGB9iJ8nKyQqyrG9tOc5ExERDWjWpez//bbb1i5ciWefvppXLp0CU2bNsW2bdvQu3dvAICbmxteeuklgxRKpkUmAAPaNsDCnVewLi4ZL4R6S10SEREZgWod8UlJSUGrVq0AAEFBQVAoFAgMDNS+HhQUhLS0tJqtkExW/7ZekAnAkcQsXL19T+pyiIjICFQr+JSUlEB+3w3lzM3NYWZm9u9gMhlELrJENcTD3hIRTdwAAL/EJUtcDRERGYNqL1nx119/wcHBAQCg0Wiwc+dOnDlzBgBw9+7dGi2OaHB7H+y8cAvrj9/A21FNYGHO5eWIiOjRVTv4DBs2TOf566+/rvPcUDebmzdvHrZt24YTJ07AwsKiwpCVlJSE8ePHY9euXbCyssLQoUPx2Wef6X3/Iap7Ipq6wdVOgdu5Kuy6kI5eIZ5Sl0RERPVYtf75rNFoqnyUlBjm0uOioiIMHDgQ48aNq/D1kpIS9OnTB3l5edi/fz/Wrl2L9evX4+233zZIPVQ75GYyDPhnYvNanu4iIqLH9Eirs0th9uzZAICVK1dW+Hp0dDTOnTuH5ORkeHl5AQAWLFiA4cOHY968ebC3t6+tUqmGDWrngyV/X8XeS7eRcrcAXo5WUpdERET1lN7BZ/PmzXoP2rdv30cq5nHExsYiJCREG3oAoGfPnlCpVDh27BgiIiIqfJ9KpYJKpdI+z8kpXSJBrVZDrVbXWH1lY9XkmMbqwV55O1jgCT8nHEm8g3VHrmNCRCMpy6tz+N3SH3ulP/ZKf+yV/gzZK33H1Dv4PPfcczrPBUHQuYLr/rk9hjrd9TBpaWlwd3fX2ebk5AQLC4uHXmL/0UcfaY8m3S86OhrW1tY1XidXrdff/b0KMhdwBGb48cAV+OVfhIzrlpbD75b+2Cv9sVf6Y6/0Z4he5efn67Wf3sHn/nW4duzYgSlTpuDDDz9EWFgYBEHAwYMH8d577+HDDz/Uu8hZs2ZVGDruFxcXh3bt2uk1XkUTq0VRfOiE66lTp2LSpEna5zk5OfDx8UFUVFSNnh5Tq9WIiYlBZGSkzi0BqLyKetVdXYJNn+xBVmExHJs8ic6BLhJXWXfwu6U/9kp/7JX+2Cv9GbJXZWdsqvJIc3zeeustLF26FJ07d9Zu69mzJ6ytrTFmzBicP39er3EmTJiAIUOGPHQfPz8/vcby8PDA4cOHdbbduXMHarW63JGg+ykUCigUinLb5XK5Qb7AhhrXGN3fK7lcjufbNMAPsdfxW3wKIpp5SFxd3cPvlv7YK/2xV/pjr/RniF7pO94jBZ+rV69q7+VzPwcHh2otUqpUKqFUKh+lhHLCwsIwb948pKamwtOz9JLn6OhoKBQKhIaG1shnkLQGt/fBD7HXEX02DVl5RXC24W0KiIioeh7pbnDt27fHW2+9hdTUfxePTEtLw9tvv40nnniixoq7X1JSEk6cOIGkpCSUlJTgxIkTOHHiBO7dK13KICoqCsHBwXjllVcQHx+PnTt3YvLkyRg9ejSv6DISzb0c0KKBA9QlIjYcvyF1OUREVA89UvBZvnw5bt26hYYNGyIwMBCBgYHw9fVFamoqli1bVtM1AgBmzJiBNm3aYObMmbh37x7atGmDNm3a4OjRowAAMzMzbNu2DZaWlujUqRMGDRqE5557Dp999plB6iFpDG7vAwD45Wgyl0chIqJqe6RTXYGBgTh16hRiYmJw4cIFiKKI4OBgPPXUUwa7c/PKlSsrvYdPGV9fX2zdutUgn091Q9/WXpi77Rwupd9DfPJdtPV1krokIiKqRx75BoaCICAqKgpRUVE1WQ/RQ9lbyvF0C09sOH4T644kM/gQEVG16H2qa+3atXoPmpycjAMHDjxSQURVGdLeFwCw5VQK7qmKJa6GiIjqE72Dz5IlS9C0aVN8/PHHFV6unp2dje3bt2Po0KEIDQ1FVlZWjRZKVKa9nxMClDbILyrB1pMpUpdDRET1iN7BZ8+ePfjss8+wa9cuhISEwN7eHo0bN0aLFi3g7e0NFxcXjBw5En5+fjhz5gyeffZZQ9ZNJkwQBO0kZy5cSkRE1VGtOT7PPPMMnnnmGWRmZmL//v1ITExEQUEBlEql9iormeyRLhQjqpb+bb3x6V8XcSL5Li6k5aCpB29ZQEREVXukyc0uLi7o169fTddCpDdXOwWeauaOP8+mYV1cMmY+21zqkoiIqB7g4RmqtwY/UXq6a2P8TaiKa39hXCIiqn8YfKje6trYFZ4Olribr0b02XSpyyEionqAwYfqLTOZgIGh3gCAdZzkTEREemDwoXptYDsfCAKw/0oGkrPypS6HiIjqOAYfqtd8nK3ROVAJoHT9LiIiooepVvAJDg7WuTHhmDFjcPv2be3zW7duwdrauuaqI9JD2T19fj16A8UlGomrISKiuqxawefChQsoLv53iYC1a9ciNzdX+1wURRQWFtZcdUR6iAx2h5O1HGk5hdh7+XbVbyAiIpP1WKe6RFEst81Qq7MTVUZhbob+bUsnOa89wtNdRERUOc7xIaNQdrpr14VbuJXLo45ERFSxagUfQRDKHdHhER6qC4Lc7dDW1xHFGhELYy4hNbtA6pKIiKgOqtaSFaIookePHjA3L31bQUEBnn32WVhYWACAzvwfotrmr7TB8aS7WHMkGevikvFR/xYY3N5X6rKIiKgOqVbwmTlzps7zitbreuGFFx6vIqJHkJpdgI3xN7XPNSIwbcMZdA1yhaeDlYSVERFRXfJYwYeorkjIyIPmgbn2JaKIxIx8Bh8iItJ6pNXZH7Rnzx7k5eUhLCwMTk5ONTEkUbX4K20gE6ATfmQC4KfkfaWIiOhf1Zrc/Omnn+oc9RFFEb169UJERASeeeYZNGvWDGfPnq3xIomq4ulghY/6t4DZfZPt2/k58WgPERHpqFbwWbNmDYKDg7XPf/vtN+zduxf79u1DRkYG2rVrh9mzZ9d4kUT6GNzeF/vfjcDMZ0u/oyeSs3E7VyVxVUREVJdUK/gkJCSgZcuW2ufbt2/HCy+8gE6dOsHZ2RnvvfceYmNja7xIIn15OlhheEc/tPZxRFGxBqsOJkpdEhER1SHVCj5qtRoKhUL7PDY2Fh07dtQ+9/LyQkZGRs1VR/QIBEHA2G4BAIAfD11Hnoq3WSAiolLVCj6BgYHYu3cvACApKQmXLl1Ct27dtK/fuHEDLi4uNVsh0SOIDPaAn4s1sgvUXLWdiIi0qhV8xo0bhwkTJmDkyJHo3bs3wsLCdOb87Nq1C23atKnxIomqy0wmYFSX0qM+3+9L4KrtREQEoJrB5/XXX8eXX36JrKwsdO3aFevXr9d5PSUlBSNGjKjRAoke1YBQb7jYWODm3QJsO50qdTlERFQHVPs+PiNHjsTIkSMrfG3x4sWPXRBRTbGUm+HVMD8s3HEJ3+29hr6tvLi2HBGRiePq7GTUXglrCEu5DGdTcnDwaqbU5RARkcSqFXzMzMz0ehDVFc42FhjczgcA8O3eaxJXQ0REUqv26uwNGzbEsGHDOImZ6o1RXQLw46Hr2HvpNs6n5qCZp73UJRERkUSqFXwOHz6M5cuX48svv4S/vz9GjBiBl156ietzUZ3m42yN3i08se1UKr7bew0LB7eWuiQiIpJItU51tW/fHkuWLEFqaiomTZqEjRs3wtvbG0OGDEFMTIyhaiR6bK93Lb20fcvJFKTcLZC4GiIiksojTW62tLTEyy+/jJ07d+LMmTO4desWevXqhaysrJquj6hGtPR2RIcAZxRrRCzfnyB1OUREJJFHvqrrxo0bmDt3LiIjI3Hx4kW88847sLc33NyJefPmoWPHjrC2toajo2OF+wiCUO6xdOlSg9VE9cvrXRsBANYcSUJ2gVriaoiISArVCj5FRUVYt24doqKi0LhxYxw/fhxffPEFkpOTMX/+fJibV/u2QNX67IEDB2LcuHEP3W/FihVITU3VPoYNG2awmqh+CW/iiibudsgrKsHqw0lSl0NERBKoVlLx9PSEnZ0dhg0bhsWLF8PNzQ0AcO/ePZ39DHHkZ/bs2QCAlStXPnQ/R0dHeHh41PjnU/0nCAJGdw3A5F9PYvmBBIzo7AeFOW+/QERkSqp1xOfOnTtISkrCBx98gCZNmsDJyUnn4ejoKPkVXhMmTIBSqUT79u2xdOlSaDRco4n+1beVFzzsLXE7V4VN8SlSl0NERLWsWkd8du/ebag6asQHH3yAHj16wMrKCjt37sTbb7+NjIwMvPfee5W+R6VSQaVSaZ/n5OQAANRqNdTqmpsHUjZWTY5prAzZKwHAq2E++OSvy/h271X0a+kOmax+L2PB75b+2Cv9sVf6Y6/0Z8he6TumIIqiWOOfrqdZs2ZpT2FVJi4uDu3atdM+X7lyJd566y3cvXu3yvEXLFiAOXPmIDs7u9o1rF69GtbW1lV+BtU/BcXAzONmUJUIGN20BCFOkv0WICKiGpKfn4+hQ4ciOzv7oVNuJA0+GRkZyMjIeOg+fn5+sLS01D6vTvA5cOAAOnfujLS0NLi7u1e4T0VHfHx8fJCRkVGjc5XUajViYmIQGRkJuVxeY+Mao9ro1fw/L2LZgeto7+eE1SPbG+Qzagu/W/pjr/THXumPvdKfIXuVk5MDpVJZZfCp1qkumUxW5erWgiCguLhYr/GUSiWUSmV1SqiW+Ph4WFpaVnr5OwAoFAooFIpy2+VyuUG+wIYa1xgZslejujbCD4eSEJd4B2dS76GNb/2/+zi/W/pjr/THXumPvdKfIXql73jVCj4bN26s9LWDBw9i0aJFMNQBpKSkJGRlZSEpKQklJSU4ceIEACAwMBC2trbYsmUL0tLSEBYWBisrK+zevRvTp0/HmDFjKgw2ZNo8HazQt1UDrD9+A9/tvYYlL4dKXRIREdWCagWffv36ldt24cIFTJ06FVu2bMFLL72EDz74oMaKu9+MGTOwatUq7fOyRVJ3796N8PBwyOVyLF68GJMmTYJGo0FAQADmzJmD8ePHG6Qeqv/GdA3A+uM38OfZNCRm5MFPaSN1SUREZGCPfOfmlJQUjB49Gi1btkRxcTFOnDiBVatWwdfXtybr01q5ciVEUSz3CA8PBwD06tUL8fHxyM3NRV5eHk6fPo3//Oc/Br2pItVvTTzsEN7EFaIIfL//mtTlEBFRLah28MnOzsaUKVMQGBiIs2fPYufOndiyZQtCQkIMUR+RQY35Z/HSX4/eQOY9VRV7ExFRfVet4PPJJ58gICAAW7duxZo1a3Dw4EF06dLFULURGVxYgAtaNHCAqliDH2KvS10OEREZWLXOA7377ruwsrJCYGAgVq1apTPn5n4bNmyokeKIDE0QBLzeLQATVsfjh9hEjO3WCFYWXMaCiMhYVSv4vPrqq1Vezk5U3/Rq7gEfZyskZxXg12PJeDXMT+qSiIjIQKoVfKpaIJSoPjI3k2FU5wDM3HwW3+9LwEtPNoRZPV/GgoiIKvbIV3URGZOB7bzhaC1HUlY+/jyTJnU5RERkIAw+RACsLczxaoeGAIDv9l412I04iYhIWgw+RP94taMfFOYynLyRjcMJWVKXQ0REBsDgQ/QPpa0CA0K9AQDf7rkqcTVERGQIDD5E9xnVJQCCAOy+eBuX0nOlLoeIiGoYgw/RffyVNugZ7AEA+G4vl7EgIjI2DD5EDxjTrXQZi00nbiItu1DiaoiIqCYx+BA9oK2vE9r7OUFdImLFwQSpyyEiohrE4ENUgTFdGwEAVh9KQm6hWuJqiIiopjD4EFWgR1M3NHK1Qa6qGGuOJEldDhER1RAGH6IKyGQCxnQtneuzfH8iioo1EldEREQ1gcGHqBLPtWkAVzsF0nIKseVkitTlEBFRDWDwIaqEwtwMwzv6AQD+t+8al7EgIjICDD5ED/Hykw1hbWGGC2m52HPpttTlEBHRY2LwIXoIB2s5hrT3BQB8tfMyDl7NQGp2gcRVERHRo2LwIarCiM5+EATgeNJdDP3fYXSavwvr4nilFxFRfcTgQ1QFM5mA+6f3aERg2oYzPPJDRFQPMfgQVSEhI6/cthJRRGJGvgTVEBHR42DwIaqCv9IGMkF3m5kA+CmtpSmIiIgeGYMPURU8HazwUf8WOuGnf6g3PB2spCuKiIgeCYMPkR4Gt/fFgXe748UnfAAAuy/cQg7X8CIiqncYfIj05Olghdl9QxCgtEHGvSJ8ueOy1CUREVE1MfgQVYOFuQwz+zYHAKw8mIhL6bkSV0RERNXB4ENUTd2CXBEZ7I4SjYhZm89yKQsionqEwYfoEbzfJxgW5jIcvJqJP86kSV0OERHpicGH6BH4ulhjbLdGAIC5W8+hoKhE4oqIiEgfDD5Ej2hct0Zo4GiFlOxCLP77itTlEBGRHhh8iB6RlYUZ3uvTDADw7d5rSMrknZyJiOo6Bh+ix9ArxAOdA5UoKtZgztZzUpdDRERVYPAhegyCIGBW32CYywTsOJ+O3RdvSV0SERE9RL0IPomJiRg5ciT8/f1hZWWFRo0aYebMmSgqKtLZLykpCc8++yxsbGygVCoxceLEcvsQ1bRANzsM7+gHAJiz5RxUxZzoTERUV5lLXYA+Lly4AI1Gg2+//RaBgYE4c+YMRo8ejby8PHz22WcAgJKSEvTp0weurq7Yv38/MjMzMWzYMIiiiEWLFkn8E5Cx+89TjfH7iRQkZORh+f5EjAtvJHVJRERUgXoRfHr16oVevXppnwcEBODixYtYsmSJNvhER0fj3LlzSE5OhpeXFwBgwYIFGD58OObNmwd7e3tJaifTYGcpx9TeTfH2ryexaNdlPN+mATwcLKUui4iIHlAvgk9FsrOz4ezsrH0eGxuLkJAQbegBgJ49e0KlUuHYsWOIiIiocByVSgWVSqV9npOTAwBQq9VQq2tuEcqysWpyTGNVX3v1TIgbfjrkgPjkbMzdehYLB7Wslc+tr/2SAnulP/ZKf+yV/gzZK33HrJfB5+rVq1i0aBEWLFig3ZaWlgZ3d3ed/ZycnGBhYYG0tMrvrPvRRx9h9uzZ5bZHR0fD2tq65or+R0xMTI2PaazqY696OAInks2w9XQa/DU3EFiLBxrrY7+kwl7pj73SH3ulP0P0Kj9fv1uKSBp8Zs2aVWHouF9cXBzatWunfZ6SkoJevXph4MCBGDVqlM6+giCUe78oihVuLzN16lRMmjRJ+zwnJwc+Pj6Iioqq0dNjarUaMTExiIyMhFwur7FxjVF979UNy3NYG3cDMRmOeGNgB5ibGfYagvrer9rEXumPvdIfe6U/Q/aq7IxNVSQNPhMmTMCQIUMeuo+fn5/21ykpKYiIiEBYWBi+++47nf08PDxw+PBhnW137tyBWq0udyTofgqFAgqFotx2uVxukC+wocY1RvW1V1N6NcMfZ9JxIf0efjmeimH/XPFlaPW1X1Jgr/THXumPvdKfIXql73iSBh+lUgmlUqnXvjdv3kRERARCQ0OxYsUKyGS6/4oOCwvDvHnzkJqaCk9PTwClp6sUCgVCQ0NrvHaiyjjZWGByzyZ4//czWBB9Ec+09ISLbflwTUREta9e3McnJSUF4eHh8PHxwWeffYbbt28jLS1NZ+5OVFQUgoOD8corryA+Ph47d+7E5MmTMXr0aF7RRbVu6BO+CPa0R05hMT6Lvih1OURE9I96EXyio6Nx5coV7Nq1C97e3vD09NQ+ypiZmWHbtm2wtLREp06dMGjQIDz33HPay92JapOZTMDsfs0BAGvjknHqxl1pCyIiIgD1JPgMHz4coihW+Lifr68vtm7divz8fGRmZmLRokUVzt8hqg3t/ZzxfJsGEEVgxqaz0GjEqt9EREQGVS+CD1F9NbV3U9hYmOFE8l2sP35D6nKIiEwegw+RAbnZW2Jij8YAgI//vICcQt7gjIhISgw+RAb2Wid/BLjaIONeEb6IuSx1OUREJo3Bh8jALMxlmPVs6UTnVbGJuJSeK3FFRESmi8GHqBZ0DXJFVLA7SjQiZm46W25iPhER1Q4GH6Ja8v4zwVCYyxB7LRPbT1e+fhwRERkOgw9RLfFxtsbYbo0AAPO2nUN+UbHEFRERmR4GH6JaNC68ERo4WiEluxCLd1+VuhwiIpPD4ENUiyzlZnj/mWAAwHd7r+F6Zp7EFRERmRYGH6Ja1rO5O7o0VqKoRIMPtp6TuhwiIpPC4ENUywRBwMxnm8NcJmDH+VvYfeGW1CUREZkMBh8iCQS62WJEZ38AwOwtZ6EqLpG4IiIi08DgQySRN7sHwtVOgcTMfMzYdBap2QVSl0REZPQYfIgkYmcpR0QTVwDAurhkdJq/C+vikiSuiojIuDH4EEkkNbsAvx37d8V2jQhM3XCaR36IiAyIwYdIIgkZedA8sHKFRgQSM/KlKYiIyAQw+BBJxF9pA5lQfnsKj/gQERkMgw+RRDwdrPBR/xYwE0rTT1kG+mDrOdy4w6M+RESGYC51AUSmbHB7X3QNckViRj68HC0xcU08Tt7Ixvifj+OXsWFQmJtJXSIRkVHhER8iiXk6WCGskQsautjgm5fawsFKjpM3sjFv23mpSyMiMjoMPkR1iLeTNb4Y3BoA8EPsdWw6cVPagoiIjAyDD1EdE9HUDW92DwRQenn75fRciSsiIqoZqdmFuJwtIDW7ULIaGHyI6qC3ngpCx0YuyC8qwbifjyNPVSx1SUREj2VdXBLCF+zF1+fMEL5gr2Q3bGXwIaqDzGQCvnqxDdztFbhy6x6mbjgNURSrfiMRUR0jiiKOX8/Cu+tPa+9dphGBaRvOSHLDVl7VRVRHKW0V+GZoWwz+7hA2n0xBez8nvBLmJ3VZREQPlZ5TiFM3snH6xl2cvJGN0zezkZVXVG6/ElFEYkY+PB2sarU+Bh+iOqydnzOm9m6KudvOY87Wc2jh7YjWPo5Sl0VEBADIvKfCqZvZOH0juzTs3LyL9BxVuf3MBKBEfHCbAD+ldS1V+i8GH6I6bmRnfxxNvIM/z6Zh/M/HsfXNznCysZC6LCIyYqnZBUjIyIO/0kZ7RCa7QI0zN0sDzqkbd3HqRjZu3i1/qkomAI3d7NDC2wGtvB3QwtsRTT3ssOnETUzdUHq6SyYAH/YPqfWjPQCDD1GdJwgCPhnYEhfScpCYmY///nICy4e1h6yi9S6IiB7TurgkbUARALT0dkB2gRqJmRXfUT7A1QYtG5QGnFbeDgj2soe1Rfl4Mbi9L8L8nfDL9t0Y9HQEfJV2Bv5JKsbgQ1QP2FvKsfilUDy/+AD+vngbi/++ggndG0tdFhEZidxCNeISs7Dj/C2sPvzv1VYigJM3srXPfZ2t0cLb4Z+g44CQBg6wt5Tr/TmeDpZo7CDC08GyJsuvFgYfonoi2Msec58LwTu/ncLnMZfQxtcJnQKVUpdFRPXQPVUx4hKzcOhqJg5dy8Tpm9naK64qMrV3Uwxq52MUp9kZfIjqkYHtfHA08Q7WHU3GxDXx2DaxC1ysuZ4XET1cXlnQuZaF2GuZOHMzGyUPJB0/F2u09HbAlpOpuP8VM0FA39ZeRhF6AAYfonpndr/mOH0zG+dSczBh9XH88Fqo1CURUR2TpyrGset3EHut9IjOqRvlg46vszXCAlzQoZEznvR3gZdj6UTjToFJmLbhDEpEEWaCINkkZENh8CGqZyzlZljycls8s2g/jl6/gwUxl9FS6qKIqNbdf+WVg5Ucx67fwaFrmYi9Whp0ih8IOj7OVujg74KwRi54MsAFDRwrDjOD2/uia5ArEjPy4ae0NqrQAzD4ENVLDV1s8NnAVnj9x2NYduA6RgQJeFrqooio1vx06Dre33QGZTd0r+g+Od5OVugQ4PLPwxneTvrfM8fTwcroAk8ZBh+ieqpncw+M6RqA7/Zew+qrMryYmYfGHo5Sl0VEBiCKIq7ezsO+y7cRfTYNsdeydF4vEQEPe0t0ClSiQ4AzOgS4wMe59m8OWB/Ui+CTmJiIDz74ALt27UJaWhq8vLzw8ssvY/r06bCw+HeylSCUv6/JkiVLMHbs2Nosl6jWvNOzCY5fz8LR63fx5pqT+H1CZ1jKOdmZyBhk56tx4GoG9l2+jb2XMiq8WeD9Fg5uhbBGvNKzKvUi+Fy4cAEajQbffvstAgMDcebMGYwePRp5eXn47LPPdPZdsWIFevXqpX3u4OBQ2+US1Rq5mQxfDGqJXl/swYX0e3j/9zP4dGArqcsiokdQXKLByRvZ2HvpNvZevo2TyXd1LjG3MJfhCT9ntPZxwDd/X8X96xaXLv9gU/tF10P1Ivj06tVLJ8wEBATg4sWLWLJkSbng4+joCA8Pj9oukUgy7vaWGN5Yg8XnzfDrsRto7+eMQe19pC6LiPRw404+9l3OwN5Lt3HgSgZyCot1Xm/sZosujV3RNUiJJ/1dYGVRekTXx9naqK+8MqR6EXwqkp2dDWdn53LbJ0yYgFGjRsHf3x8jR47EmDFjIJPJKh1HpVJBpfp3QbWcnBwAgFqthlqtrrF6y8aqyTGNFXtVPWq1Go0dREyMCMAXu67h/U1n0MTdGsGe9lKXVufwu6U/9kp/D+tVanYhrmfmo6GLNTwdLJFfVIzDCXew70omDlzJwLUM3WUgHKzM0amRCzoHuqBzoPKBOxxroFZrAAD9W3sizN8JSVn58HUuHbs+/Lcy5PdK3zEFURQfcq/Guunq1ato27YtFixYgFGjRmm3z507Fz169ICVlRV27tyJGTNmYOrUqXjvvfcqHWvWrFmYPXt2ue2rV6+GtTUnhlH9oRGB7y/KcPaODEqFiLdblsC63v7Thqj+i00XsO6aDCIEACLcLEVkqgSUiP/OR5VBREM7oJmjBk0cRPjali7gSdWXn5+PoUOHIjs7G/b2lf/DT9LgU1nouF9cXBzatWunfZ6SkoJu3bqhW7du+P777x/63gULFmDOnDnIzs6udJ+Kjvj4+PggIyPjoY2rLrVajZiYGERGRkIu139dE1PEXlXP/f3KUwPPL4nFjbuFiGzmhm9ebFXhpH9Txe+W/tgr/VXUq8TMPER9cQAV/QXr7WiJzo2V6BLogg7+zrC3Mp3+GvJ7lZOTA6VSWWXwkfTfgxMmTMCQIUMeuo+fn5/21ykpKYiIiEBYWBi+++67Ksfv0KEDcnJykJ6eDnd39wr3USgUUCgU5bbL5XKD/GY31LjGiL2qHrlcDldrORa/HIoBS2IRc/4WFu1OQFigC/yVNjz/fx9+t/THXulPLpcjM78EPx5KxKqDiRWGnoWDWuG5Ng1M/h8khvhe6TuepMFHqVRCqdTv0rubN28iIiICoaGhWLFixUPn7ZSJj4+HpaUlHB0dH7NSovqjpbcjZjwbjPd+P4NFu69g0e4rkAnAR/1bYHB7X6nLIzI6oigiIRd4a90p/HUuvdwdk8uYCQI6NHIx+dAjtXoxAyAlJQXh4eHw9fXFZ599htu3b2tfK7uCa8uWLUhLS0NYWBisrKywe/duTJ8+HWPGjKnwiA6RMeve1FXnuUYEpm04g65BrjzyQ1RDioo12HY6BSv2J+DUTXMAaQCAJ/2d8VonP2TlFeH938/yyqs6pl4En+joaFy5cgVXrlyBt7e3zmtlU5TkcjkWL16MSZMmQaPRICAgAHPmzMH48eOlKJlIUomZ+eW2lYgiEjPy+Qcv0WO6navCz4ev4+fDSbidWzpH1FwQ0a9NA4zoHIDmXv/ePy6iqZvRrnlVX9WL4DN8+HAMHz78ofs8eK8fIlPmr7SBTAAePOJ+KT0HYY1cpCmKqJ47fSMbKw4mYOvJVBSVlF5W7m6vwND2PlBmX8CgfiHl5pkY85pX9VW9CD5EVD2eDlb4qH8L7Q3OSi+mBWZuPodCtQZjugZwngGRHopLNPjrbDpWHEjA0et3tNvb+DritU7+6B3iAWhKsH37BQmrpOpg8CEyUoPb+6JrkCsSM/Lh62yFFQcS8f3+BHz0xwWkZhfi/WeCYcYbhhBV6E5eEdbEJeHH2OtIzS4EAMjNBPRp4YnhnfzR2sdRu69aUyJRlfQoGHyIjNj9h9nfeyYYHg6WmLvtPFYeTMSt3EJ8Pqg1FzUlApCaXYCEjDxoNMC20ynYGH8Thf/cJdnFxgIvPemLlzs0hJu9ZRUjUV3H4ENkQkZ1CYCbvSUm/3IS20+nIePeEfzvlXZwsOZ9Wsh0rT2ShKkbT+PB2/k297LHa5388UxLT/4DwYgw+BCZmL6tvKC0tcDrPxzDkYQsDFh6EKtGPAEvR07AJNPz15lUvLvhdLntS19ui57NPTgXzghVfRdAIjI6HRsp8eu4MHjYW+LyrXvov/ggLqTlSF0WUa25nJ6LUauO4vWfjlf4uoOVBUOPkWLwITJRTT3sseGNjghyt0VaTiEGLonFwasZUpdFZFCp2QWY8tsp9PxiL3acT69wQVAzQYCfkotUGysGHyIT5uVohV9f74gn/J2RqyrG8OVx2HIyReqyiGpcdr4a8/+4gPBP/8a6o8nQiEDP5u6I/m83fPxCC5j9c3SHd1g2fpzjQ2TiHKzl+GHEE5j0ywlsP52GN9fEIz2nEKO6BEhdGtFjK1SX4IfYRHyz+yqyC9QAgCf8nDGld1OENnQCAAS62Wpv/cA7LBs/Bh8igqXcDF+/2BZz7M5h5cFEzN12HqnZhZj+dDPIeK8fqodKNCI2xt/E59EXkfLPfXiC3G0xpVdTdG/qVm7+Du+wbDoYfIgIACCTCZj5bDC8HC3x4fYLWLY/Aek5hVgwqBUU5ryUl+oHURSx++ItfPzHRVxMzwUAeDpYYlJkEPq39eZNO4nBh4j+JQgCxnRtBHd7S0z+9SS2nkpFxj0Vvn2lHRyseK8fqtuOJ93B/D8u4EhCFgDA3tIc4yMCMayjH+/DQ1oMPkRUTr/WDaC0VeD1H4/h0LUsDFoai5Uj2vNUANVJV2/fw2d/XcQfZ9IAABbmMrzWyQ9vdAvkzTmpHAYfIqpQp0Alfnk9DMNXHMHF9Fz0X1x6o8MgdzupSyMCANzKKcQXOy9jXVwySjQiZAIwINQbbz0VxBtyUqUYfIioUsFepff6Gbb8CK7ezsOAJQfxv1fb4ckAF6lLIxNUtp6W0laBzSdSsGx/AgrUpQuEPtXMDe/0bIomHgzm9HAMPkT0UN5O1lg/riNGrTqKo9fv4JVlR7BwcGu0beiIhIw8+CtteAqMDG5dXBKmbjgNzQPrabX1dcS7vZvhCX9naQqjeofBh4iq5GhtgZ9GPYn/rI3HX2fTMX71cQgARAAyAfiofwsMbu8rdZlkpFKzC/DuhvKLiM7v3wKD2/twaQmqFt65mYj0Yik3w+KXQvFC2wYASkMPAGhEYNqGM0jNLpCuODJaGfdU+L/fTpULPQDQ0MWGoYeqjcGHiPRmJhPwQlvvcttLRBFnbmZLUBEZK41GxJojSeixYA/2XS6/hhzX06JHxeBDRNXi72pT4cKOk345ie/3XYOquKT2iyKjciEtBwO/jcXUDaeRXaBGsKc9xoc34npaVCM4x4eIqsXTwQof9W+BaRvOoEQsvYTY1U6B9BwV5m47j1WxiZjSqyn6tPDkaQiqlvyiYny58zKW7UtAsUaEtYUZJkUGYXhHP5ibyfByWEOup0WPjcGHiKptcHtfnUUd3ews8duxZCyIvoTkrAJMWB2P730SML1PM7T349U2VLUd59Ixc/NZ3LxbOlesV3MPzHg2WOd+PFxPi2oCgw8RPZIH/xIa3N4Xz7bywv/2JuDbvVdxIvkuBi6NRc/m7pjSqykCXG0lrJbqqpS7BZi95Sz+OpsOAGjgaIU5/ZqjRzN3iSsjY8XgQ0Q1xtrCHP95qjFefNIHC2MuY11cEv46m46d52/hpSd9MbFHY7jYKqQuk+qA4hINVh5MxOcxl5BfVAJzmYBRXQIwsUcgrC34VxMZDr9dRFTj3Ows8VH/Fnitkx/m/3EBuy7cwqrY69hw/CbGRTTCiE7+XDTShMUn3cG0jWdwPjUHANCuoRPmPh+Cph72EldGpoDBh4gMJsjdDsuHt8fBKxmYt/08zqbk4JM/L+Kn2OuY3LMJnmvdALKKLhEjo5RdoManf13Az4eTIIqAo7UcU3s3xcBQH34PqNYw+BCRwXUMVGLLhM7YdPImPv3zIlKyCzHpl5NYtj8B059uho6BSqlLJAMSRRGbT6bgg63nkXFPBQB4oa03pj3dlKc+qdYx+BBRrZDJBDzfxhu9Qzyx4kAiFu++grMpORj6/WF0b+qGqb2borG7nXYhSq4BZhwSMvLw/u9nsP9K6U0IG7naYO5zLRDWiAvdkjQYfIioVlnKzTAuvBEGtfPGol1X8NOh69h14Rb+vngL7fyccDTxDjQi1wCrz1KzC3ApPRd7L93Gj4eSUFSsgcJchje7B2J01wAozDm/i6TD4ENEknCxVWBW3+YY1tEPH/9xAX+eTcORhDva10vXADuNrkGuPPJTj6yLSyq3oGiXxkrMfS4EDV1spCuM6B9csoKIJOWvtMHSV0Ix69ngcq+ViMCszWex/3IG1CUaCaojfZVoRKw5fB1T1uuGHpkAfPxCC4YeqjN4xIeI6oSeIR6Ys/UcNA+swv3X2XT8dTYddpbmCG/ihshgd3QLcoWDlVyaQklHoboEvx27gWX7E5CQkVfudY0IXM8sgJcjFxSluoHBh4jqhIrWAHulQ0OoijXYcf4WMu6psOVkCracTIG5TECHABc81cwNTwW7w9uJf6nWtqy8IvwYex0/xCYiM68IAGBnaY57hcW4P7tyFXWqaxh8iKjOeHANsLK5PRqNiPjku9hxPh0x59Jx5dY97L+Sgf1XMjBryzk087RHZLA7Ipu5I6SBPRdHNaCkzHx8v/8afjmajEJ16enHBo5WGNXFH4Pa+WDrqRRteOUq6lQXMfgQUZ1S0UKUMpmA0IZOCG3ohCm9miIhIw87zqUj5nw6jiZm4XxqDs6n5uCrnZfhYW+Jp4LdEBnsgQ4BzjpXEKVmF+JytoDU7EL4KnmqrDpOJN/Fd3uv4s8zadrTkSEN7DGmayM8HeIBc7PSKaOVhVeiuqLeBJ++ffvixIkTuHXrFpycnPDUU0/h448/hpeXl3afpKQkjB8/Hrt27YKVlRWGDh2Kzz77DBYWFhJWTkQ1zV9pg9FdAzC6awCy8oqw+8ItxJxLx97Lt5GWU4ifDiXhp0NJsFWYo1uQKyKD3XEnvwgfbD0HjWiGxef38lJ5PWhEYNfF21h24DqOJGRpt3cLcsXrXQMQ1silwqNrXEWd6rJ6E3wiIiIwbdo0eHp64ubNm5g8eTIGDBiAgwcPAgBKSkrQp08fuLq6Yv/+/cjMzMSwYcMgiiIWLVokcfVEZCjONhZ4IdQbL4R6o1BdgtirmYg+l46d59NxK1eFbadTse10qs57NCIwdcNpPOnvAj8lrzZ6kKq4BOuP3cSXJ82QfigeAGAuE9C3tRfGdA3gmlpUr9Wb4PPf//5X++uGDRvi3XffxXPPPQe1Wg25XI7o6GicO3cOycnJ2qNACxYswPDhwzFv3jzY2/M3KpGxs5SbIaKpGyKaukGjCcGpm9nYcS4dm07cRPKdAp19NSIQ8dnf8FfaoLG7LYLc7dDY3Q5N3O3gr7SBhbnp3e0ju0CNnw9fx8oDibiVqwIgwEZhhpeebIjXOvnxKA4ZhXoTfO6XlZWFn3/+GR07doRcXnqePjY2FiEhITqnvnr27AmVSoVjx44hIiKiwrFUKhVUKpX2eU5O6WrBarUaarW6xmouG6smxzRW7FX1sF+Va+5hg+YeARgU6oXwBXvLXSovAriWkYdrGXn462y6dru5TEBDF2s0drNFYzcbNHazRaCbLfxcrCE3Kx+IUrMLcT0zHw1drOHpYGngn6pm3F+zKIpYGZuEX47eQF5RCQDA3U6BDs75mDa4C5ztSgMPv2MV4+9B/RmyV/qOKYiiKFa9W90wZcoUfP3118jPz0eHDh2wdetWuLiUrvcyZswYJCYmIjo6Wuc9CoUCK1euxIsvvljhmLNmzcLs2bPLbV+9ejWsrXkJJpGxiE0XsO6aDCIECBAxOECDYCcRqfkC0gqAtHxB++vCkoqvCjMTRLhZAh7WIjytRXhYAWkFwB/JuuOGudftP1bv7wV0/hfwtBLRvYEGbV1EmOBBL6rH8vPzMXToUGRnZz/0LI+kwaey0HG/uLg4tGvXDgCQkZGBrKwsXL9+HbNnz4aDgwO2bt0KQRAwZswYXL9+HX/99ZfO+y0sLPDDDz9gyJAhFY5f0REfHx8fZGRk1OjpMbVajZiYGERGRmqPUlHF2KvqYb/0l5yZiw3R+9E/qjN8XOwq3EcURaTlqHD51j2dx9VbedqjIVV5tqUHGjhawclaDidrCzhay7W/drKWw87SXO9L7qtzNEldosGdfDWy8oqQlVeEzH/+PytPjaz80l+nZRfi1M2ccu9t6+uI8eEB6BJYOmGZ3yv9sVf6M2SvcnJyoFQqqww+kp7qmjBhQqWBpIyfn5/210qlEkqlEkFBQWjWrBl8fHxw6NAhhIWFwcPDA4cPH9Z57507d6BWq+Hu7l7p+AqFAgqFotx2uVxukC+wocY1RuxV9bBfVfNxsUNjBxE+LnYP7ZWv0gK+Sjv0uG8VDY1GRMo/i29eSr+HS+m5iE+6W+HdirecSntoHeYy4Z8wZAEnm9Iw5GxjAUdrCzjft+3Y9TtYuucqNCIgCMDgdj5o7G6HrDxVabC5V6QNORn3VMgpLH7k3rzTs2mFK6bze6U/9kp/huiVvuNJGnzKgsyjKDtQVXa0JiwsDPPmzUNqaio8PT0BANHR0VAoFAgNDa2ZgonIZMlkArydrOHtZI3uTUv/MZWaXYBO83fpzB0SBGBkJ38Ua0Tc+ecoy91/jsLcyS9CflEJijUiMu4VIeNekd6fL4rA2rjkqusUACdrCzjbWMDF1gIuNgo42/z7XBCAGZvO6qynxbsrkympF5Objxw5giNHjqBz585wcnLCtWvXMGPGDDRq1AhhYWEAgKioKAQHB+OVV17Bp59+iqysLEyePBmjR4/mFV1EZBAPLrNRdqfih90fqFBdgrv5atzJL8KdvCJk5RfhTr4ad/4JRqXb1LiRlY9rFRxN6hDgjCbudnC2UcDZ1gIuNv88bC3gbKOAg5UcZrKHn0azMJPx7spksupF8LGyssKGDRswc+ZM5OXlwdPTE7169cLatWu1p6nMzMywbds2vPHGG+jUqZPODQyJiAyluncqtpSbwcPBDB5VzNep6GiSmSBg4eDWjx1SeHdlMmX1Ivi0aNECu3btqnI/X19fbN26tRYqIiL6lyHuVFzZ0aSa+hzeXZlMVb0IPkREpohHZohqHoMPEVEdxiMzRDWLt6ciIiIik8HgQ0RERCaDwYeIiIhMBoMPERERmQwGHyIiIjIZDD5ERERkMhh8iIiIyGQw+BAREZHJYPAhIiIik8HgQ0RERCaDwYeIiIhMBtfqeoAoigCAnJycGh1XrVYjPz8fOTk5kMvlNTq2sWGvqof90h97pT/2Sn/slf4M2auyv7fL/h6vDIPPA3JzcwEAPj4+EldCRERE1ZWbmwsHB4dKXxfEqqKRidFoNEhJSYGdnR0EQaixcXNycuDj44Pk5GTY29vX2LjGiL2qHvZLf+yV/tgr/bFX+jNkr0RRRG5uLry8vCCTVT6Th0d8HiCTyeDt7W2w8e3t7fkbQ0/sVfWwX/pjr/THXumPvdKfoXr1sCM9ZTi5mYiIiEwGgw8RERGZDAafWqJQKDBz5kwoFAqpS6nz2KvqYb/0x17pj73SH3ulv7rQK05uJiIiIpPBIz5ERERkMhh8iIiIyGQw+BAREZHJYPAhIiIik8HgUwv69u0LX19fWFpawtPTE6+88gpSUlJ09klKSsKzzz4LGxsbKJVKTJw4EUVFRRJVLI3ExESMHDkS/v7+sLKyQqNGjTBz5sxyfWCvSs2bNw8dO3aEtbU1HB0dK9yHvfrX4sWL4e/vD0tLS4SGhmLfvn1SlyS5vXv34tlnn4WXlxcEQcDvv/+u87ooipg1axa8vLxgZWWF8PBwnD17VppiJfbRRx+hffv2sLOzg5ubG5577jlcvHhRZx/2q9SSJUvQsmVL7U0Kw8LC8Mcff2hfl7pPDD61ICIiAr/88gsuXryI9evX4+rVqxgwYID29ZKSEvTp0wd5eXnYv38/1q5di/Xr1+Ptt9+WsOrad+HCBWg0Gnz77bc4e/YsFi5ciKVLl2LatGnafdirfxUVFWHgwIEYN25cha+zV/9at24d3nrrLUyfPh3x8fHo0qULevfujaSkJKlLk1ReXh5atWqFr7/+usLXP/nkE3z++ef4+uuvERcXBw8PD0RGRmrXNDQle/bswfjx43Ho0CHExMSguLgYUVFRyMvL0+7DfpXy9vbG/PnzcfToURw9ehTdu3dHv379tOFG8j6JVOs2bdokCoIgFhUViaIoitu3bxdlMpl48+ZN7T5r1qwRFQqFmJ2dLVWZdcInn3wi+vv7a5+zV+WtWLFCdHBwKLedvfrXE088IY4dO1ZnW9OmTcV3331XoorqHgDixo0btc81Go3o4eEhzp8/X7utsLBQdHBwEJcuXSpBhXXLrVu3RADinj17RFFkv6ri5OQkfv/993WiTzziU8uysrLw888/o2PHjpDL5QCA2NhYhISEwMvLS7tfz549oVKpcOzYMalKrROys7Ph7Oysfc5e6Y+9KlVUVIRjx44hKipKZ3tUVBQOHjwoUVV1X0JCAtLS0nT6plAo0K1bN/YNpX82AdD++cR+VaykpARr165FXl4ewsLC6kSfGHxqyZQpU2BjYwMXFxckJSVh06ZN2tfS0tLg7u6us7+TkxMsLCyQlpZW26XWGVevXsWiRYswduxY7Tb2Sn/sVamMjAyUlJSU64W7u7tJ9aG6ynrDvpUniiImTZqEzp07IyQkBAD79aDTp0/D1tYWCoUCY8eOxcaNGxEcHFwn+sTg84hmzZoFQRAe+jh69Kh2/3feeQfx8fGIjo6GmZkZXn31VYj33TRbEIRynyGKYoXb65vq9goAUlJS0KtXLwwcOBCjRo3SeY29Olr1QP8w5l5V14M/s6n2obrYt/ImTJiAU6dOYc2aNeVeY79KNWnSBCdOnMChQ4cwbtw4DBs2DOfOndO+LmWfzGvlU4zQhAkTMGTIkIfu4+fnp/21UqmEUqlEUFAQmjVrBh8fHxw6dAhhYWHw8PDA4cOHdd57584dqNXqcqm4Pqpur1JSUhAREYGwsDB89913OvuxV7q9ehhj75W+lEolzMzMyv1r8tatWybVh+ry8PAAUHokw9PTU7vd1Pv25ptvYvPmzdi7dy+8vb2129kvXRYWFggMDAQAtGvXDnFxcfjyyy8xZcoUANL2icHnEZUFmUdRdqRHpVIBAMLCwjBv3jykpqZqvwjR0dFQKBQIDQ2tmYIlVJ1e3bx5ExEREQgNDcWKFSsgk+kelGSv9GfsvdKXhYUFQkNDERMTg+eff167PSYmBv369ZOwsrrN398fHh4eiImJQZs2bQCUzpfas2cPPv74Y4mrq32iKOLNN9/Exo0b8ffff8Pf31/ndfbr4URRhEqlqht9qpUp1Cbs8OHD4qJFi8T4+HgxMTFR3LVrl9i5c2exUaNGYmFhoSiKolhcXCyGhISIPXr0EI8fPy7u2LFD9Pb2FidMmCBx9bXr5s2bYmBgoNi9e3fxxo0bYmpqqvZRhr361/Xr18X4+Hhx9uzZoq2trRgfHy/Gx8eLubm5oiiyV/dbu3atKJfLxWXLlonnzp0T33rrLdHGxkZMTEyUujRJ5ebmar83AMTPP/9cjI+PF69fvy6KoijOnz9fdHBwEDds2CCePn1afPHFF0VPT08xJydH4spr37hx40QHBwfx77//1vmzKT8/X7sP+1Vq6tSp4t69e8WEhATx1KlT4rRp00SZTCZGR0eLoih9nxh8DOzUqVNiRESE6OzsLCoUCtHPz08cO3aseOPGDZ39rl+/Lvbp00e0srISnZ2dxQkTJmiDkalYsWKFCKDCx/3Yq1LDhg2rsFe7d+/W7sNe/eubb74RGzZsKFpYWIht27bVXoZsynbv3l3hd2jYsGGiKJZeoj1z5kzRw8NDVCgUYteuXcXTp09LW7REKvuzacWKFdp92K9SI0aM0P5ec3V1FXv06KENPaIofZ8EUbxvhi0RERGREeNVXURERGQyGHyIiIjIZDD4EBERkclg8CEiIiKTweBDREREJoPBh4iIiEwGgw8RERGZDAYfIiIiMhkMPkRk0rp27YrVq1c/dB9BEPD7779Xa9yvv/4affv2fYzKiMgQGHyIqNbdunULr7/+Onx9faFQKODh4YGePXsiNjZWu4+fnx8EQcChQ4d03vvWW28hPDxc+3zWrFkQBAGCIEAmk8HLywsvvfQSkpOTq6xj69atSEtLw5AhQ6pVf9nnCYIAc3Nz+Pr6YtKkSdqFhwFg9OjRiIuLw/79+6s1NhEZFoMPEdW6F154ASdPnsSqVatw6dIlbN68GeHh4cjKytLZz9LSElOmTKlyvObNmyM1NRU3btzAunXrcPr0aQwaNKjK93311Vd47bXXIJNV/4/CFStWIDU1FQkJCVi8eDF+/PFHzJ07V/u6QqHA0KFDsWjRomqPTUSGYy51AURkWu7evYv9+/fj77//Rrdu3QAADRs2xBNPPFFu39dffx1LlizB9u3b8fTTT1c6prm5OTw8PAAAXl5eGD16NCZOnIicnBzY29tX+J6MjAzs2LEDCxcu1Nl++fJljBw5EkeOHEFAQAC+/PLLCt/v6Oio/UwfHx/07dsXx48f19mnb9++iIqKQkFBAaysrCqtn4hqD4/4EFGtsrW1ha2tLX7//XedU0MV8fPzw9ixYzF16lRoNBq9xk9LS8OGDRtgZmYGMzOzSvfbv38/rK2t0axZM+02jUaD/v37w8zMDIcOHcLSpUv1OuJ06dIl7N69G08++aTO9nbt2kGtVuPIkSN61U5EhsfgQ0S1ytzcHCtXrsSqVavg6OiITp06Ydq0aTh16lSF+7/33ntISEjAzz//XOmYp0+fhq2tLaytreHp6Ym///4b48ePh42NTaXvSUxMhLu7u85prh07duD8+fP48ccf0bp1a3Tt2hUffvhhhe9/8cUXYWtrC0tLSzRp0gTNmzfH1KlTdfaxsbGBo6MjEhMTH9IRIqpNDD5EVOteeOEFpKSkYPPmzejZsyf+/vtvtG3bFitXriy3r6urKyZPnowZM2agqKiowvGaNGmCEydOIC4uDvPmzUPr1q0xb968h9ZQUFAAS0tLnW3nz5+Hr68vvL29tdvCwsIqfP/ChQtx4sQJnDx5Elu3bsWlS5fwyiuvlNvPysoK+fn5D62FiGoPgw8RScLS0hKRkZGYMWMGDh48iOHDh2PmzJkV7jtp0iQUFBRg8eLFFb5uYWGBwMBANG/eHNOmTUPr1q0xbty4h36+UqnEnTt3dLaJolhuP0EQKny/h4cHAgMD0aRJE/Tp0wezZ8/GunXrcOXKFZ39srKy4Orq+tBaiKj2MPgQUZ0QHByMvLy8Cl+ztbXF+++/j3nz5iEnJ6fKsd5//32sWbOm3GTj+7Vp0wZpaWk64Sc4OBhJSUlISUnRbrv/EvuHKZtPVFBQoN129epVFBYWok2bNnqNQUSGx+BDRLUqMzMT3bt3x08//YRTp04hISEBv/76Kz755BP069ev0veNGTMGDg4OWLNmTZWfERAQgH79+mHGjBmV7tOmTRu4urriwIED2m1PPfUUmjRpgldffRUnT57Evn37MH369Arff/fuXaSlpSElJQV79uzBnDlzEBQUpDNZet++fQgICECjRo2qrJmIageDDxHVKltbWzz55JNYuHAhunbtipCQELz//vsYPXo0vv7660rfJ5fL8cEHH6CwsFCvz3n77bexbds2HD58uMLXzczMMGLECJ1J0zKZDBs3boRKpcITTzyBUaNGVTpX6LXXXoOnpye8vb3x4osvonnz5vjjjz9gbv7vXULWrFmD0aNH61UvEdUOQazopDYRkQlIT09H8+bNcezYMTRs2LBGxz5z5gx69OiBS5cuwcHBoUbHJqJHxyM+RGSy3N3dsWzZMiQlJdX42CkpKfjhhx8YeojqGB7xISIiIpPBIz5ERERkMhh8iIiIyGQw+BAREZHJYPAhIiIik8HgQ0RERCaDwYeIiIhMBoMPERERmQwGHyIiIjIZDD5ERERkMv4fDUZhnuUPkn0AAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "\n",
    "from files_02_channel import gamp_curve\n",
    "\n",
    "snrdb, nmse = gamp_curve.nmse_data()\n",
    "\n",
    "plt.plot(snrdb, nmse, \".-\")\n",
    "plt.grid()\n",
    "plt.xlabel(\"SNR (dB)\")\n",
    "plt.ylabel(\"NMSE (dB)\")\n",
    "plt.title(\"EM-BG-GAMP algorithm\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sum = 0j\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from files_02_channel import dataset\n",
    "\n",
    "data = dataset.data()\n",
    "\n",
    "antenna_channel, angle_channel = (data.get(key) for key in [\"Harray\", \"Hvirtual\"])\n",
    "fig1, axs1 = plt.subplots(3, 3)\n",
    "fig2, axs2 = plt.subplots(3, 3)\n",
    "for i, ax in enumerate(axs1.flat):\n",
    "    ax.matshow(np.abs(antenna_channel[i]))\n",
    "for i, ax in enumerate(axs2.flat):\n",
    "    ax.matshow(np.abs(angle_channel[i]))\n",
    "\n",
    "angle_ch = angle_channel[0]\n",
    "\n",
    "antenna_ch = np.fft.fft(\n",
    "    np.fft.ifft(angle_ch, axis=-2, norm=\"ortho\"), axis=-1, norm=\"ortho\"\n",
    ")\n",
    "\n",
    "print(f\"sum = {np.sum(antenna_ch - antenna_channel[0])}\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import math\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def mse_db(y_true, y_pred):\n",
    "    return (\n",
    "        10\n",
    "        * tf.math.log(tf.math.reduce_mean(tf.math.square(y_true - y_pred)))\n",
    "        / math.log(10.0)\n",
    "    )\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from files_02_channel.low_res_mimo_ch_est.models import dense_model, residual_model\n",
    "from tensorflow.keras.models import save_model, load_model\n",
    "\n",
    "\n",
    "def get_model(Nr, Nt, T, model_type, should_include_x=False):\n",
    "    if model_type == \"dense\":\n",
    "        num_neurons = [256 for _ in range(10)]\n",
    "        return dense_model(\n",
    "            Nr,\n",
    "            Nt,\n",
    "            T,\n",
    "            num_neurons,\n",
    "            should_include_x=should_include_x,\n",
    "            activation=\"tanh\",\n",
    "        )\n",
    "    elif model_type == \"residual\":\n",
    "        return residual_model(\n",
    "            Nr, Nt, T, [400 for _ in range(3)], should_include_x=should_include_x\n",
    "        )\n",
    "    else:\n",
    "        raise Exception(\"Invalid model name\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.callbacks import (\n",
    "    EarlyStopping,\n",
    "    ModelCheckpoint,\n",
    "    ReduceLROnPlateau,\n",
    "    TensorBoard,\n",
    ")\n",
    "\n",
    "\n",
    "def train(filename, training_generator=None, validation_generator=None, epochs=100):\n",
    "    model = load_model(filename, custom_objects={\"mse_db\": mse_db})\n",
    "    model.compile(loss=mse_db, optimizer=\"adam\")\n",
    "    history = model.fit(\n",
    "        x=training_generator,\n",
    "        validation_data=validation_generator,\n",
    "        epochs=epochs,\n",
    "        verbose=1,\n",
    "        workers=1,\n",
    "        use_multiprocessing=False,\n",
    "        max_queue_size=16,\n",
    "        callbacks=[\n",
    "            EarlyStopping(\n",
    "                monitor=\"val_loss\",\n",
    "                min_delta=5e-3,\n",
    "                patience=9,\n",
    "                restore_best_weights=True,\n",
    "            ),\n",
    "            ModelCheckpoint(\n",
    "                filepath=filename,\n",
    "                monitor=\"val_loss\",\n",
    "                verbose=1,\n",
    "                save_best_only=True,\n",
    "            ),\n",
    "            ReduceLROnPlateau(\n",
    "                factor=0.5,\n",
    "                min_delta=5e-2,\n",
    "                patience=3,\n",
    "                cooldown=5,\n",
    "                verbose=1,\n",
    "                min_lr=1e-6,\n",
    "            ),\n",
    "            TensorBoard(histogram_freq=1),\n",
    "        ],\n",
    "    )\n",
    "\n",
    "    return history\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import files_02_channel.low_res_mimo_ch_est.snr_generator as snr_gen\n",
    "\n",
    "\n",
    "def test_by_snr(model, snr, test_generator=None):\n",
    "    test_generator.set_snr_generator(snr_gen.constant_snr(snr))\n",
    "    return model.evaluate(test_generator)\n",
    "\n",
    "\n",
    "def test(filename, test_generator, snr_db=[]):\n",
    "    model = load_model(filename, custom_objects={\"mse_db\": mse_db})\n",
    "\n",
    "    return [test_by_snr(model, snr, test_generator) for snr in snr_db]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                   Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      " input_1 (InputLayer)           [(None, 256, 16)]    0           []                               \n",
      "                                                                                                  \n",
      " flatten (Flatten)              (None, 4096)         0           ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " dense (Dense)                  (None, 400)          1638800     ['flatten[0][0]']                \n",
      "                                                                                                  \n",
      " leaky_re_lu (LeakyReLU)        (None, 400)          0           ['dense[0][0]']                  \n",
      "                                                                                                  \n",
      " concatenate (Concatenate)      (None, 4496)         0           ['flatten[0][0]',                \n",
      "                                                                  'leaky_re_lu[0][0]']            \n",
      "                                                                                                  \n",
      " dense_1 (Dense)                (None, 400)          1798800     ['concatenate[0][0]']            \n",
      "                                                                                                  \n",
      " leaky_re_lu_1 (LeakyReLU)      (None, 400)          0           ['dense_1[0][0]']                \n",
      "                                                                                                  \n",
      " concatenate_1 (Concatenate)    (None, 800)          0           ['leaky_re_lu[0][0]',            \n",
      "                                                                  'leaky_re_lu_1[0][0]']          \n",
      "                                                                                                  \n",
      " dense_2 (Dense)                (None, 400)          320400      ['concatenate_1[0][0]']          \n",
      "                                                                                                  \n",
      " leaky_re_lu_2 (LeakyReLU)      (None, 400)          0           ['dense_2[0][0]']                \n",
      "                                                                                                  \n",
      " concatenate_2 (Concatenate)    (None, 800)          0           ['leaky_re_lu_1[0][0]',          \n",
      "                                                                  'leaky_re_lu_2[0][0]']          \n",
      "                                                                                                  \n",
      " dense_3 (Dense)                (None, 128)          102528      ['concatenate_2[0][0]']          \n",
      "                                                                                                  \n",
      " reshape (Reshape)              (None, 16, 8)        0           ['dense_3[0][0]']                \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 3,860,528\n",
      "Trainable params: 3,860,528\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "from files_02_channel.low_res_mimo_ch_est.pilotgen.qpsk_pilot_gen import QpskPilotGenerator\n",
    "from files_02_channel.low_res_mimo_ch_est.channelgen.random_sparse_gen import (\n",
    "    RandomSparseChannelGenerator,\n",
    ")\n",
    "\n",
    "from files_02_channel.low_res_mimo_ch_est.XYGenerator import XYGenerator\n",
    "\n",
    "rand_gen = np.random.Generator(np.random.SFC64())\n",
    "\n",
    "Nr = 8\n",
    "Nt = 8\n",
    "T = 256\n",
    "\n",
    "should_include_x = False\n",
    "\n",
    "model = get_model(Nr, Nt, T, \"residual\", should_include_x=should_include_x)\n",
    "\n",
    "model.compile(loss=mse_db, optimizer=\"adam\")\n",
    "model.summary()\n",
    "\n",
    "filename = \"model.h5\"\n",
    "\n",
    "save_model(model, filename)\n",
    "\n",
    "x_generator = QpskPilotGenerator(Nt, T, rand_gen, is_constant=True)\n",
    "\n",
    "snr_range = (0, 21)\n",
    "\n",
    "num_clusters = 1\n",
    "channel_generators = {\n",
    "    name: RandomSparseChannelGenerator(Nr, Nt, num_clusters, rand_gen)\n",
    "    for name in [\"train\", \"val\", \"test\"]\n",
    "}\n",
    "\n",
    "trainConfig = {\n",
    "    \"num_batches\": 1000,\n",
    "    \"batch_size\": 32,\n",
    "    \"x_generator\": x_generator,\n",
    "    \"channel_generator\": channel_generators[\"train\"],\n",
    "    \"snr_generator\": snr_gen.uniform_snr(*snr_range, rand_gen),\n",
    "    \"quantizer\": np.sign,\n",
    "    \"constant_batches\": False,\n",
    "    \"should_include_x\": should_include_x,\n",
    "    \"generator\": rand_gen,\n",
    "}\n",
    "\n",
    "valConfig = {\n",
    "    \"num_batches\": 100,\n",
    "    \"batch_size\": 32,\n",
    "    \"x_generator\": x_generator,\n",
    "    \"channel_generator\": channel_generators[\"val\"],\n",
    "    \"snr_generator\": snr_gen.constant_snr(3),\n",
    "    \"quantizer\": np.sign,\n",
    "    \"constant_batches\": False,\n",
    "    \"should_include_x\": should_include_x,\n",
    "    \"generator\": rand_gen,\n",
    "}\n",
    "\n",
    "testConfig = {\n",
    "    \"num_batches\": 1000,\n",
    "    \"batch_size\": 32,\n",
    "    \"x_generator\": x_generator,\n",
    "    \"channel_generator\": channel_generators[\"test\"],\n",
    "    \"snr_generator\": snr_gen.uniform_snr(*snr_range, rand_gen),\n",
    "    \"quantizer\": np.sign,\n",
    "    \"constant_batches\": False,\n",
    "    \"should_include_x\": should_include_x,\n",
    "    \"generator\": rand_gen,\n",
    "}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-09-06 14:29:11.318875: W tensorflow/core/platform/profile_utils/cpu_utils.cc:128] Failed to get CPU frequency: 0 Hz\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 998/1000 [============================>.] - ETA: 0s - loss: -10.8007\n",
      "Epoch 1: val_loss improved from inf to -12.99279, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -10.8045 - val_loss: -12.9928 - lr: 0.0010\n",
      "Epoch 2/100\n",
      "1000/1000 [==============================] - ETA: 0s - loss: -13.8526\n",
      "Epoch 2: val_loss improved from -12.99279 to -14.42486, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -13.8526 - val_loss: -14.4249 - lr: 0.0010\n",
      "Epoch 3/100\n",
      "1000/1000 [==============================] - ETA: 0s - loss: -14.7661\n",
      "Epoch 3: val_loss improved from -14.42486 to -15.46465, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -14.7661 - val_loss: -15.4647 - lr: 0.0010\n",
      "Epoch 4/100\n",
      " 998/1000 [============================>.] - ETA: 0s - loss: -15.5965\n",
      "Epoch 4: val_loss improved from -15.46465 to -16.07193, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -15.5974 - val_loss: -16.0719 - lr: 0.0010\n",
      "Epoch 5/100\n",
      " 995/1000 [============================>.] - ETA: 0s - loss: -16.1003\n",
      "Epoch 5: val_loss improved from -16.07193 to -16.69714, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -16.1025 - val_loss: -16.6971 - lr: 0.0010\n",
      "Epoch 6/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -17.0234\n",
      "Epoch 6: val_loss improved from -16.69714 to -17.15092, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -17.0209 - val_loss: -17.1509 - lr: 0.0010\n",
      "Epoch 7/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -17.8069\n",
      "Epoch 7: val_loss improved from -17.15092 to -18.10004, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -17.8088 - val_loss: -18.1000 - lr: 0.0010\n",
      "Epoch 8/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -17.6494\n",
      "Epoch 8: val_loss did not improve from -18.10004\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -17.6495 - val_loss: -17.4995 - lr: 0.0010\n",
      "Epoch 9/100\n",
      " 998/1000 [============================>.] - ETA: 0s - loss: -18.4352\n",
      "Epoch 9: val_loss improved from -18.10004 to -18.30621, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -18.4365 - val_loss: -18.3062 - lr: 0.0010\n",
      "Epoch 10/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -18.8631\n",
      "Epoch 10: val_loss improved from -18.30621 to -19.37655, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -18.8627 - val_loss: -19.3766 - lr: 0.0010\n",
      "Epoch 11/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -18.7101\n",
      "Epoch 11: val_loss did not improve from -19.37655\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -18.7128 - val_loss: -19.2588 - lr: 0.0010\n",
      "Epoch 12/100\n",
      "1000/1000 [==============================] - ETA: 0s - loss: -19.1092\n",
      "Epoch 12: val_loss improved from -19.37655 to -19.85994, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -19.1092 - val_loss: -19.8599 - lr: 0.0010\n",
      "Epoch 13/100\n",
      " 998/1000 [============================>.] - ETA: 0s - loss: -18.9874\n",
      "Epoch 13: val_loss improved from -19.85994 to -19.86510, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -18.9918 - val_loss: -19.8651 - lr: 0.0010\n",
      "Epoch 14/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -19.4228\n",
      "Epoch 14: val_loss improved from -19.86510 to -19.97620, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -19.4229 - val_loss: -19.9762 - lr: 0.0010\n",
      "Epoch 15/100\n",
      "1000/1000 [==============================] - ETA: 0s - loss: -19.4680\n",
      "Epoch 15: val_loss did not improve from -19.97620\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -19.4680 - val_loss: -19.6173 - lr: 0.0010\n",
      "Epoch 16/100\n",
      " 995/1000 [============================>.] - ETA: 0s - loss: -19.5076\n",
      "Epoch 16: val_loss did not improve from -19.97620\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -19.5073 - val_loss: -19.2154 - lr: 0.0010\n",
      "Epoch 17/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -19.5721\n",
      "Epoch 17: val_loss did not improve from -19.97620\n",
      "\n",
      "Epoch 17: ReduceLROnPlateau reducing learning rate to 0.0005000000237487257.\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -19.5716 - val_loss: -19.8705 - lr: 0.0010\n",
      "Epoch 18/100\n",
      " 999/1000 [============================>.] - ETA: 0s - loss: -22.6294\n",
      "Epoch 18: val_loss improved from -19.97620 to -22.73547, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -22.6317 - val_loss: -22.7355 - lr: 5.0000e-04\n",
      "Epoch 19/100\n",
      "1000/1000 [==============================] - ETA: 0s - loss: -23.1404\n",
      "Epoch 19: val_loss improved from -22.73547 to -23.10129, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -23.1404 - val_loss: -23.1013 - lr: 5.0000e-04\n",
      "Epoch 20/100\n",
      " 998/1000 [============================>.] - ETA: 0s - loss: -23.2616\n",
      "Epoch 20: val_loss improved from -23.10129 to -24.12231, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -23.2584 - val_loss: -24.1223 - lr: 5.0000e-04\n",
      "Epoch 21/100\n",
      "1000/1000 [==============================] - ETA: 0s - loss: -23.2940\n",
      "Epoch 21: val_loss did not improve from -24.12231\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -23.2940 - val_loss: -23.7611 - lr: 5.0000e-04\n",
      "Epoch 22/100\n",
      " 995/1000 [============================>.] - ETA: 0s - loss: -23.4945\n",
      "Epoch 22: val_loss did not improve from -24.12231\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -23.4830 - val_loss: -24.0928 - lr: 5.0000e-04\n",
      "Epoch 23/100\n",
      " 999/1000 [============================>.] - ETA: 0s - loss: -23.6666\n",
      "Epoch 23: val_loss did not improve from -24.12231\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -23.6606 - val_loss: -22.8914 - lr: 5.0000e-04\n",
      "Epoch 24/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -23.7714\n",
      "Epoch 24: val_loss did not improve from -24.12231\n",
      "\n",
      "Epoch 24: ReduceLROnPlateau reducing learning rate to 0.0002500000118743628.\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -23.7796 - val_loss: -23.9452 - lr: 5.0000e-04\n",
      "Epoch 25/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -25.6070\n",
      "Epoch 25: val_loss improved from -24.12231 to -26.70644, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -25.6074 - val_loss: -26.7064 - lr: 2.5000e-04\n",
      "Epoch 26/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -25.8227\n",
      "Epoch 26: val_loss did not improve from -26.70644\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -25.8342 - val_loss: -26.1932 - lr: 2.5000e-04\n",
      "Epoch 27/100\n",
      " 995/1000 [============================>.] - ETA: 0s - loss: -26.1561\n",
      "Epoch 27: val_loss did not improve from -26.70644\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -26.1360 - val_loss: -26.3635 - lr: 2.5000e-04\n",
      "Epoch 28/100\n",
      " 998/1000 [============================>.] - ETA: 0s - loss: -25.9264\n",
      "Epoch 28: val_loss did not improve from -26.70644\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -25.9236 - val_loss: -26.6076 - lr: 2.5000e-04\n",
      "Epoch 29/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -26.0305\n",
      "Epoch 29: val_loss did not improve from -26.70644\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -26.0386 - val_loss: -26.5801 - lr: 2.5000e-04\n",
      "Epoch 30/100\n",
      "1000/1000 [==============================] - ETA: 0s - loss: -26.1747\n",
      "Epoch 30: val_loss did not improve from -26.70644\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -26.1747 - val_loss: -26.4033 - lr: 2.5000e-04\n",
      "Epoch 31/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -26.3066\n",
      "Epoch 31: val_loss did not improve from -26.70644\n",
      "\n",
      "Epoch 31: ReduceLROnPlateau reducing learning rate to 0.0001250000059371814.\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -26.2967 - val_loss: -26.6353 - lr: 2.5000e-04\n",
      "Epoch 32/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -27.5998\n",
      "Epoch 32: val_loss improved from -26.70644 to -27.92367, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -27.5987 - val_loss: -27.9237 - lr: 1.2500e-04\n",
      "Epoch 33/100\n",
      " 999/1000 [============================>.] - ETA: 0s - loss: -27.9343\n",
      "Epoch 33: val_loss did not improve from -27.92367\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -27.9319 - val_loss: -27.8400 - lr: 1.2500e-04\n",
      "Epoch 34/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -28.0471\n",
      "Epoch 34: val_loss improved from -27.92367 to -28.74924, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.0300 - val_loss: -28.7492 - lr: 1.2500e-04\n",
      "Epoch 35/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -27.8221\n",
      "Epoch 35: val_loss improved from -28.74924 to -29.00038, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -27.8169 - val_loss: -29.0004 - lr: 1.2500e-04\n",
      "Epoch 36/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -27.5780\n",
      "Epoch 36: val_loss did not improve from -29.00038\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -27.5875 - val_loss: -28.6485 - lr: 1.2500e-04\n",
      "Epoch 37/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -27.9172\n",
      "Epoch 37: val_loss did not improve from -29.00038\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -27.9038 - val_loss: -28.3758 - lr: 1.2500e-04\n",
      "Epoch 38/100\n",
      "1000/1000 [==============================] - ETA: 0s - loss: -28.0477\n",
      "Epoch 38: val_loss did not improve from -29.00038\n",
      "\n",
      "Epoch 38: ReduceLROnPlateau reducing learning rate to 6.25000029685907e-05.\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.0477 - val_loss: -27.8130 - lr: 1.2500e-04\n",
      "Epoch 39/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -28.7018\n",
      "Epoch 39: val_loss improved from -29.00038 to -29.96140, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.7106 - val_loss: -29.9614 - lr: 6.2500e-05\n",
      "Epoch 40/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -28.6606\n",
      "Epoch 40: val_loss did not improve from -29.96140\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.6675 - val_loss: -29.5919 - lr: 6.2500e-05\n",
      "Epoch 41/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -28.3571\n",
      "Epoch 41: val_loss did not improve from -29.96140\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.3500 - val_loss: -29.8720 - lr: 6.2500e-05\n",
      "Epoch 42/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -28.3365\n",
      "Epoch 42: val_loss did not improve from -29.96140\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.3215 - val_loss: -28.9528 - lr: 6.2500e-05\n",
      "Epoch 43/100\n",
      " 998/1000 [============================>.] - ETA: 0s - loss: -28.7655\n",
      "Epoch 43: val_loss improved from -29.96140 to -30.02369, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.7672 - val_loss: -30.0237 - lr: 6.2500e-05\n",
      "Epoch 44/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -28.5261\n",
      "Epoch 44: val_loss did not improve from -30.02369\n",
      "1000/1000 [==============================] - 9s 9ms/step - loss: -28.5347 - val_loss: -28.6686 - lr: 6.2500e-05\n",
      "Epoch 45/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -28.5822\n",
      "Epoch 45: val_loss did not improve from -30.02369\n",
      "1000/1000 [==============================] - 10s 9ms/step - loss: -28.5710 - val_loss: -29.9647 - lr: 6.2500e-05\n",
      "Epoch 46/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -28.5527\n",
      "Epoch 46: val_loss improved from -30.02369 to -30.32782, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.5376 - val_loss: -30.3278 - lr: 6.2500e-05\n",
      "Epoch 47/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -29.0086\n",
      "Epoch 47: val_loss did not improve from -30.32782\n",
      "1000/1000 [==============================] - 9s 9ms/step - loss: -28.9947 - val_loss: -29.7548 - lr: 6.2500e-05\n",
      "Epoch 48/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -28.6663\n",
      "Epoch 48: val_loss improved from -30.32782 to -30.70066, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.6544 - val_loss: -30.7007 - lr: 6.2500e-05\n",
      "Epoch 49/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -28.7637\n",
      "Epoch 49: val_loss did not improve from -30.70066\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.7831 - val_loss: -30.1652 - lr: 6.2500e-05\n",
      "Epoch 50/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -28.9214\n",
      "Epoch 50: val_loss did not improve from -30.70066\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.9234 - val_loss: -29.7383 - lr: 6.2500e-05\n",
      "Epoch 51/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -28.6985\n",
      "Epoch 51: val_loss did not improve from -30.70066\n",
      "\n",
      "Epoch 51: ReduceLROnPlateau reducing learning rate to 3.125000148429535e-05.\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.6873 - val_loss: -28.1111 - lr: 6.2500e-05\n",
      "Epoch 52/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -28.5820\n",
      "Epoch 52: val_loss did not improve from -30.70066\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.5916 - val_loss: -30.5460 - lr: 3.1250e-05\n",
      "Epoch 53/100\n",
      " 995/1000 [============================>.] - ETA: 0s - loss: -29.3460\n",
      "Epoch 53: val_loss did not improve from -30.70066\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.3498 - val_loss: -29.9554 - lr: 3.1250e-05\n",
      "Epoch 54/100\n",
      " 999/1000 [============================>.] - ETA: 0s - loss: -29.0101\n",
      "Epoch 54: val_loss did not improve from -30.70066\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.0013 - val_loss: -29.7857 - lr: 3.1250e-05\n",
      "Epoch 55/100\n",
      " 995/1000 [============================>.] - ETA: 0s - loss: -29.0987\n",
      "Epoch 55: val_loss did not improve from -30.70066\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.1101 - val_loss: -30.4935 - lr: 3.1250e-05\n",
      "Epoch 56/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -29.1784\n",
      "Epoch 56: val_loss improved from -30.70066 to -30.86630, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.1739 - val_loss: -30.8663 - lr: 3.1250e-05\n",
      "Epoch 57/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -29.4200\n",
      "Epoch 57: val_loss did not improve from -30.86630\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.4139 - val_loss: -30.8035 - lr: 3.1250e-05\n",
      "Epoch 58/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -29.1928\n",
      "Epoch 58: val_loss did not improve from -30.86630\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.1964 - val_loss: -30.4888 - lr: 3.1250e-05\n",
      "Epoch 59/100\n",
      " 999/1000 [============================>.] - ETA: 0s - loss: -28.9457\n",
      "Epoch 59: val_loss improved from -30.86630 to -31.22572, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -28.9506 - val_loss: -31.2257 - lr: 3.1250e-05\n",
      "Epoch 60/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -29.1060\n",
      "Epoch 60: val_loss did not improve from -31.22572\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.0994 - val_loss: -31.0377 - lr: 3.1250e-05\n",
      "Epoch 61/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -29.2835\n",
      "Epoch 61: val_loss did not improve from -31.22572\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.2976 - val_loss: -29.3217 - lr: 3.1250e-05\n",
      "Epoch 62/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -29.1491\n",
      "Epoch 62: val_loss did not improve from -31.22572\n",
      "\n",
      "Epoch 62: ReduceLROnPlateau reducing learning rate to 1.5625000742147677e-05.\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.1531 - val_loss: -30.7328 - lr: 3.1250e-05\n",
      "Epoch 63/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -29.0381\n",
      "Epoch 63: val_loss did not improve from -31.22572\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.0221 - val_loss: -30.7547 - lr: 1.5625e-05\n",
      "Epoch 64/100\n",
      " 995/1000 [============================>.] - ETA: 0s - loss: -29.5384\n",
      "Epoch 64: val_loss did not improve from -31.22572\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.5375 - val_loss: -31.0671 - lr: 1.5625e-05\n",
      "Epoch 65/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -29.4035\n",
      "Epoch 65: val_loss did not improve from -31.22572\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.4153 - val_loss: -29.8849 - lr: 1.5625e-05\n",
      "Epoch 66/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -29.5141\n",
      "Epoch 66: val_loss did not improve from -31.22572\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.5316 - val_loss: -30.5427 - lr: 1.5625e-05\n",
      "Epoch 67/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -29.7718\n",
      "Epoch 67: val_loss improved from -31.22572 to -32.03428, saving model to model.h5\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.7772 - val_loss: -32.0343 - lr: 1.5625e-05\n",
      "Epoch 68/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -29.2467\n",
      "Epoch 68: val_loss did not improve from -32.03428\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.2541 - val_loss: -30.6492 - lr: 1.5625e-05\n",
      "Epoch 69/100\n",
      " 999/1000 [============================>.] - ETA: 0s - loss: -29.0753\n",
      "Epoch 69: val_loss did not improve from -32.03428\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.0804 - val_loss: -30.8861 - lr: 1.5625e-05\n",
      "Epoch 70/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -29.8901\n",
      "Epoch 70: val_loss did not improve from -32.03428\n",
      "\n",
      "Epoch 70: ReduceLROnPlateau reducing learning rate to 7.812500371073838e-06.\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.9012 - val_loss: -30.3751 - lr: 1.5625e-05\n",
      "Epoch 71/100\n",
      " 995/1000 [============================>.] - ETA: 0s - loss: -29.4821\n",
      "Epoch 71: val_loss did not improve from -32.03428\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.5024 - val_loss: -29.7720 - lr: 7.8125e-06\n",
      "Epoch 72/100\n",
      " 997/1000 [============================>.] - ETA: 0s - loss: -29.4637\n",
      "Epoch 72: val_loss did not improve from -32.03428\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.4530 - val_loss: -30.0083 - lr: 7.8125e-06\n",
      "Epoch 73/100\n",
      " 996/1000 [============================>.] - ETA: 0s - loss: -29.7933\n",
      "Epoch 73: val_loss did not improve from -32.03428\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.7981 - val_loss: -31.0834 - lr: 7.8125e-06\n",
      "Epoch 74/100\n",
      " 999/1000 [============================>.] - ETA: 0s - loss: -29.5148\n",
      "Epoch 74: val_loss did not improve from -32.03428\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.5155 - val_loss: -31.1303 - lr: 7.8125e-06\n",
      "Epoch 75/100\n",
      "1000/1000 [==============================] - ETA: 0s - loss: -29.6440\n",
      "Epoch 75: val_loss did not improve from -32.03428\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.6440 - val_loss: -30.4584 - lr: 7.8125e-06\n",
      "Epoch 76/100\n",
      " 995/1000 [============================>.] - ETA: 0s - loss: -29.6625\n",
      "Epoch 76: val_loss did not improve from -32.03428\n",
      "1000/1000 [==============================] - 10s 10ms/step - loss: -29.6493 - val_loss: -31.3711 - lr: 7.8125e-06\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x15e715f90>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train(\n",
    "    filename,\n",
    "    training_generator=XYGenerator(**trainConfig),\n",
    "    validation_generator=XYGenerator(**valConfig),\n",
    "    epochs=100,\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000/1000 [==============================] - 5s 5ms/step - loss: -4.6134\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -5.8960\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -7.7463\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -10.2767\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -13.4843\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -17.2965\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -21.3356\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -24.7559\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -26.8552\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -28.3163\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -29.5174\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -30.8894\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -30.6560\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -30.4079\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -29.4765\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -28.8351\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -28.2879\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -27.8156\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -27.7849\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -27.4840\n",
      "1000/1000 [==============================] - 5s 5ms/step - loss: -27.3880\n"
     ]
    }
   ],
   "source": [
    "nmse_db = test(\n",
    "    filename,\n",
    "    test_generator=XYGenerator(**testConfig),\n",
    "    snr_db=np.arange(-30, 31, 3),\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1628a4790>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig_res, ax_res = plt.subplots()\n",
    "ax_res.plot(snrdb, nmse,'.-')\n",
    "ax_res.plot(np.arange(-30, 31, 3), nmse_db, \".-\")\n",
    "ax_res.legend([\"EM-BG-GAMP\", \"DL approach\"])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "52e85753",
   "metadata": {},
   "source": [
    "## Authors\n",
    "\n",
    "- [Marcos Yuichi Takeda - LASSE/UFPA](https://github.com/yuichitakeda)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.8.10 64-bit ('3.8.10')",
   "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": "3.8.10"
  },
  "vscode": {
   "interpreter": {
    "hash": "5aaf80bf5bb851baa857fc3e2a05055a4723060c18483edce1107b583200cb68"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}