helmholtz.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-01-09 01:00:04.507956: I tensorflow/core/util/port.cc:110] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.\n",
      "2024-01-09 01:00:04.542708: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
      "To enable the following instructions: AVX2 AVX_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2024-01-09 01:00:05.196810: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n"
     ]
    }
   ],
   "source": [
    "from modules.models import Helmholtz2DPinn, create_dense_model\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from modules.utils import PrintLossCallback"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def rhs(xy):\n",
    "    return (1 - np.pi**2 - (6*np.pi)**2) * tf.sin(np.pi * xy[:, 0:1]) * tf.sin(6 * np.pi * xy[:, 1:2])\n",
    "\n",
    "def u(xy):\n",
    "    return tf.sin(np.pi * xy[:, 0:1]) * tf.sin(6 * np.pi * xy[:, 1:2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-01-09 01:00:06.334157: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n",
      "2024-01-09 01:00:06.354784: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n",
      "2024-01-09 01:00:06.354935: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n",
      "2024-01-09 01:00:06.355960: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n",
      "2024-01-09 01:00:06.356075: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n",
      "2024-01-09 01:00:06.356137: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n",
      "2024-01-09 01:00:06.397755: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n",
      "2024-01-09 01:00:06.397876: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n",
      "2024-01-09 01:00:06.397942: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n",
      "2024-01-09 01:00:06.398001: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1635] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 4274 MB memory:  -> device: 0, name: NVIDIA GeForce RTX 3060 Laptop GPU, pci bus id: 0000:01:00.0, compute capability: 8.6\n"
     ]
    }
   ],
   "source": [
    "n_samples = 512\n",
    "\n",
    "tx_samples = np.random.uniform(-1, 1, size=(n_samples, 2))\n",
    "\n",
    "tx_bnd_bottom = np.concatenate([np.linspace(-1, 1, n_samples//4)[:, None], -np.ones(n_samples//4)[:, None]], axis=1)\n",
    "tx_bnd_top = np.concatenate([np.linspace(-1, 1, n_samples//4)[:, None], np.ones(n_samples//4)[:, None]], axis=1)\n",
    "tx_bnd_left = np.concatenate([-np.ones(n_samples//4)[:, None], np.linspace(-1, 1, n_samples//4)[:, None]], axis=1)\n",
    "tx_bnd_right = np.concatenate([np.ones(n_samples//4)[:, None], np.linspace(-1, 1, n_samples//4)[:, None]], axis=1)\n",
    "tx_bnd = np.concatenate([tx_bnd_bottom, tx_bnd_top, tx_bnd_left, tx_bnd_right], axis=0)\n",
    "\n",
    "tx_samples = tf.convert_to_tensor(tx_samples, dtype=tf.float32)\n",
    "tx_bnd = tf.convert_to_tensor(tx_bnd, dtype=tf.float32)\n",
    "u_samples = u(tx_samples)\n",
    "u_bnd = u(tx_bnd)\n",
    "inputs = [tx_samples, tx_bnd]\n",
    "outputs = [u_samples, rhs(tx_samples), u_bnd]\n",
    "\n",
    "mean = tf.reduce_mean(tx_samples, axis=0)\n",
    "std = tf.math.reduce_std(tx_samples, axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.QuadMesh at 0x7fe5d53b5d00>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkQAAAGgCAYAAABL3XhTAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAACHo0lEQVR4nO29eZhdZZXvv/beZ6pKUgm5GYpoZL4MMgThJoa2hdvJQyLcbrhybaK0QBrDBYk2BhliMwgoAeFyabi0sWkmn4bG4QJOGMFo9KoxaAQHDDxCMyoVhJhUUlVn2vv9/eGPIue8nyRv6lRVAuf7eZ56oFbW++69zx7OW3t911qRc86ZEEIIIUQbE+/qHRBCCCGE2NVoQSSEEEKItkcLIiGEEEK0PVoQCSGEEKLt0YJICCGEEG2PFkRCCCGEaHu0IBJCCCFE26MFkRBCCCHaHi2IhBBCCNH2aEEkhBBCiLZnRBdEP/zhD+2v//qvbdq0aRZFkT344IM7HLNq1Sp717veZcVi0fbff3+76667PJ9bb73V9t57byuVSjZr1ix79NFHh3/nhRBCCNE25EZy8r6+PjviiCPs7//+7+3973//Dv2fffZZO/HEE+2cc86xe+65x1auXGkf+chHbM8997R58+aZmdmXvvQlW7JkiS1fvtxmzZplN910k82bN8+eeuopmzJlStB+ZVlmf/jDH2zcuHEWRVFLxyiEEEKI0cE5Z5s3b7Zp06ZZHA/zOx03SpiZe+CBB7brc9FFF7l3vvOdDbZTTz3VzZs3b/D3mTNnuvPOO2/w9zRN3bRp09yyZcuC9+XFF190ZqYf/ehHP/rRj37ehD8vvvhi8Hd+KCP6hmhnWb16tc2dO7fBNm/ePDv//PPNzKxardratWtt6dKlg/8ex7HNnTvXVq9evc15K5WKVSqVwd+dc2ZmNv3KSy0ulYbxCIQQuy1Ob4OF2K2J3A5dsnLZXrziMzZu3Lhh3/xutSDq6emxqVOnNtimTp1qvb29NjAwYH/6058sTVP0efLJJ7c577Jly+zKK6/07HGppAWREO2CFkRC7N4ELIgGXUdA7rJbLYhGiqVLl9qSJUsGf+/t7bXp06d7fg7CkS6BEwQ2GhsFjo3iLGhsFPu2JIGx4BfDhRbTfE37EuJjZhYb+cEx0H6AjfwIGjvcZPBF6gK/XNPMvzB4PhgLF1Wa+WMz2EaoX/O+OBqXwrgUjgHGGtgcjI3AZoG2yL8ccb4obcUPbLjdFvxCx5INrh/ys2zHPqHzwy3P2yRoLN0Ew4wL/RIFN7zlyY9kLYF+wTbal4CxLgmcn/xaGRv43eoSeF7kRv66eJ3dakHU3d1t69evb7CtX7/eurq6rKOjw5IksSRJ0Ke7u3ub8xaLRSsWiyOyz0IIIYR487Nb1SGaPXu2rVy5ssH2yCOP2OzZs83MrFAo2FFHHdXgk2WZrVy5ctBHCCGEEGJnGdEF0ZYtW+zxxx+3xx9/3Mz+nFb/+OOP2wsvvGBmfw5lnX766YP+55xzjv3Hf/yHXXTRRfbkk0/aP//zP9uXv/xl+8QnPjHos2TJErvtttvs7rvvtnXr1tm5555rfX19tnDhwpE8FCGEEEK8hRnRkNnPf/5z+6//9b8O/v66jueMM86wu+66y15++eXBxZGZ2T777GPf+ta37BOf+IT90z/9k7397W+3f/3Xfx2sQWRmduqpp9of//hHu/zyy62np8dmzJhhK1as8ITWQ8GB5sXlQd+S9wPlSd4XACTgl8v5fsVC3bMVEt+vlK95to6cbysl/nyduaq/jdjfRkdSbfrdn78Y+/PnQQBRiv2x5Ee2GMQI5EckzUKJnSCFvxFI80N+NQieoy3zb7sK2Vzet6W+30AW5tefFjxbud7oV4Zx5bo/f/M4M7NqDWx1//jrVd8vrfmfpwNbBDarkkDDN0V13w8u0WBb4t9SBrfGTmwD9HU0Xx30elV4TqUwX9M20AdsEWwzysK2aeRHeiGyDf1WDv5z31EtG7qkSN8Sgy2Ba5nGgi3Lgy0XNhZub28sPCq2sU3fL8Wxvg0eW7hvDo4rJbFa8/N3BJMjRnRBdNxxxw2muBNUhfq4446zxx57bLvzLl682BYvXtzq7gkhhBBCmNlupiESQgghhNgVaEEkhBBCiLZnt0q73+UE1hKKQS+UK4KGpuALBTrBNqZQ8Wxj875AoStf9mzjwDY28ecbnxsAP5gvbrSNif25xiX+XKUIjivyjwG1RqgrAp0ECENoRQ/h9WBQAgF+NYhjV2FvylAMvgwB+n4H+h6w9aZ+IdH+zC8psTnz/TbVOz3blrRx7OY6jKt2+OPq/r5trvr70V/1/Qbg3FYj0BWRnADqGlGBNpIigGQONTqoDfJvA/SDW89yFbiW0QbaRPCLq+TnH1xUBVstbfLxP4CoDh9UDT6oOthSGIs2/xicgzsN9EfBgL4niuCJAX5GuqIcfF2CXsjyIKLJ+Vo6V/Dnc3nfLwO/rABaxyKNbfRLC/6x1kukRwK9XRH8oJoNnFqEyoxFVLeMBIEjhN4QCSGEEKLt0YJICCGEEG2PQmZbg+XQqYWG70btLHLQVqOQo5R4P9w0Ng+hKgiPdeV828Rcn2cbn/R7tglgaw6HjYv8+cfEfqxgTATHBZ9dCT68ovmvevNRmC0e5jV9BgGyFF7l180PA5Sdb6s5/7Pqh/hLn/M/537Iae2M/LEUHitlQy950EwK4cEMbpY6tAapp5B2n4NWI9QeBMIWDmJhmIXbQqsJ+kiCw22UOg82SpPHMFoZSlJAeCyu+Ocbw2GVJlsV4n4UHqv68zsY6yg8BqE18nMUHqMwWivA8ydK/GsUbXn4uoQwWlSAkBmE0SKyFcFWg9AnlMKgMgjmhezpnvKHGYShqdUGlQTARwqdxsB7dDTRGyIhhBBCtD1aEAkhhBCi7dGCSAghhBBtjzREWxMY0+TsUEhJhBTCFHQW1cyPV5ehTnohsGXG5sjXlFArjBBS0HHUQPNTgxT7qkFbEWrJAX4F0CTlIa4do/DLJwG/FE5uBrYUqq3XwA+kIlZ2Yan4fZmfnk6p+H2YYu+nxW+G9PwtaGucbwCuuyqUCaiCNqgGttB7IKN0WxI3gI1S7PGyCJvO4JQF2+BWtphaNIAeI4IWChF8VtTiIoJrKgQqWWAxHQRobyCVnFL2SS8UkdYIjstl8Nyi7gd0HEBE6fSgF6J0etIVYSo+aY0oFR/8MO0edEWuAH40trl1B2h+sHUHHCq2KaFXKqG2wHt0NNEbIiGEEEK0PVoQCSGEEKLt0YJICCGEEG2PNERbQ3ohqC+e1aHeShVaNMAmMtIQ1f2A7QDUsthS8zUlG3N+O4bOnK/n6UigtQb0GWj2K8XQfgTqEBXBj9p5cC0cXy9UAD/SQSVY4GLopPA3QgaBcvKrOdDVgK0M9YXKzrdVyA9sA6Q/Ai1QXx1aazS14OiH+iZkK4NtoAb6oxp8JjVo0wF+VNt/iFI4M9uGDgiegCRbQX0h6Y9IL0QdH6D1AtxCllTJz/+sqP5RVPMPpLkmUgR9FuI6zAV+UZ3ElOCHPVjCbBH5tQJpjajmDtXAgnOLflBny4F2h/wytME1Ba01QvxS0guBvIlad8BjBq/tNNAP9UcoCBw99IZICCGEEG2PFkRCCCGEaHu0IBJCCCFE2yMN0VZEoFmwGsSDqeYQ9GJKIf5fhfoj/Tk/6JpAH7QYeoORXy7x9TfUVy0P/deae7JRj7YEhBw56uUGthjEGOgXGEsO9WsFqjHViq0OwXPqA0b1emgs+tH1SNtt8sNxtG+go8tgLGnwHPnVqScS3I/gh9o/qhGUCxMC4VjQWWTQ8ov6OGFvNDpe7KsW1isqtCdbsx/1v8K5Qm1Yy420QeQHtuEmsM4N16ci/RH5tWILux5Bmsi1g5psoXPh/KQDChzbynyjid4QCSGEEKLt0YJICCGEEG2PQmZbgSm9NXhlDTZ+Feu//6NXsSm8P4a33a2VOg/2C3hv3cJr52FnNNI0h/lAWtrl0LGh+9w8X/D8Q98mHX8rn0lo6w5KsU8pjBa6fy18VsHHi7GbsKFB22jh2g4/Z0O8FkeCUXgmBX+kw/0cDfGDk9bS/K0cQ7DNbf/3YURviIQQQgjR9mhBJIQQQoi2RwsiIYQQQrQ90hBtDaS4QkZ4eHosaZICU1WDxwbaUMcQMrYVPQWNBbeWNBaB8yEt6Alaibvj7gXG0zFtlsYO0a+VlOGWUpApVRfKTOBx4XxhF1/oNgzmo7ERjaVtwAUe4XyB2widr8kPfWh6mh9tMBhopVwGbZeg8iihULkM7jZCfmCDbeD+0TYymA9KYeD+NY2lufDBAN8NOBZs+P1DJTQCx45mNw+9IRJCCCFE26MFkRBCCCHaHi2IhBBCCNH2tLeGyEUN8VMssQ81h2Iq2Q+2uBZoo7GBNtxu3Q+68nw79otS8AEbxn5hfmoVwBoiGAvbRYZZQ+RIGBFadp9C9gn4hdqwJQW0DIE7O8SPfKh+D88fZnPUBoNstL/g5/KBmhKoOeTy/oUbwXxxzn84JDCWWukU8v7Nl4f5CtByp5j4Y4s531YCv0Lsz1dKGh9AOfApwsMiDzc47VsM4pM8PFhJQ0QtgQhq/0NkgTd4DXpepHCDk66oBuK3Clz0ddgG+VXBrwp+5TrYUt9WabJVYVw19bdZq4fZ0hrZ4CEFrX6oDY+jun/UrmeE0BsiIYQQQrQ9WhAJIYQQou1p75BZExGkAWKoqRpmS8hWIT94fUxjy+BXg9f7FXiVX4OO8lWyNb7ejmr+626yWR1CD3XwS0Nt8Po8o1zQYc7JpPAY2RL4WyKG3HHyy0FLlxyk0eb829PlYWzBt2V5eOVfgK71xUZbWoAwFdjqRd9GY9OiZ8Iu8ZTmS1DIMKKwHL1lh9R5Co8lBf96zBcgdFXw49+dZMv7N3NXwX8QdOZ8v3G5sj8WbGPhwdIJtnFx49jO2PcZA7YSxPpLkW/LQ9OhAoTMKIyWQCislfR8gsJeVWixVIMLrQax4zLaCr4NYr19mX9z9Gf+2M1ZybNtqYMNbrbeJr/+OvjUfFtfzd+Pgap/DP1V36+aQMiwCs8t+NwxPX8U0RsiIYQQQrQ9WhAJIYQQou3RgkgIIYQQbc+oaIhuvfVWu/76662np8eOOOIIu+WWW2zmzJnoe9xxx9kPfvADz37CCSfYt771LTMzO/PMM+3uu+9u+Pd58+bZihUrWtvR0PYTlJ4fmHaPeiHQFeVIL0TaoDLYKpDmWoEU2bJvi6pNtiocBNlqvs2BzWr+Nl0dPjzQFTlI2XekP2qBCFLnqVdClPjx7ygPtxP5FfxYfJQnG8xX8GP2ruj7xQXflpFfU9psVAI9Uko1BnwTEliyIIPTSPdZaFsaLHcAbSpi6M0TQ+o8pcl3QDo96YUmFEEHlB/wbaAN2iPf79nGJ/7Y8YnvNw78uuJGW7OmyMxsDGiDOiE9fwx8xnk430XQiuQj3xZDmnyCfUqGTs35x5GBgK3m/M+g7PzPqgwaxn7QH/VBrQnSC/WCXmhz1uHZNiadnm1T6vt1NJVZ2AhfVDHcVKGlDVI41hSeFxloKVNsaRO02RFjxBdEX/rSl2zJkiW2fPlymzVrlt100002b948e+qpp2zKlCme//3332/V6hsPlddee82OOOII+8AHPtDgN3/+fLvzzjsHfy8WQb0phBBCCBHAiC+IbrzxRlu0aJEtXLjQzMyWL19u3/rWt+yOO+6wSy65xPOfOHFiw+/33XefdXZ2eguiYrFo3d3dQftQqVSsUnnjNUxvb+/OHoYQQggh3sKMqIaoWq3a2rVrbe7cuW9sMI5t7ty5tnr16qA5br/9dluwYIGNGTOmwb5q1SqbMmWKHXjggXbuuefaa6+9ts05li1bZuPHjx/8mT59+tAOSAghhBBvSUb0DdGrr75qaZra1KlTG+xTp061J598cofjH330UfvNb35jt99+e4N9/vz59v73v9/22Wcfe+aZZ+xTn/qUve9977PVq1dbApqNpUuX2pIlSwZ/7+3t5UURxS9DbQTpGLDODWhjArdLdVlYPwF1bqA1hDXZIhiHtXWg5HwEdYNIB0SHShHsCD4n0vzQNkJBDRFcU1ybCD4DqCVk9Jni5wzbRY0Ttfiggj0B7UECr7Fhv1eA0HuAbFS+xsGEZAuFatqE2qiVQ0a1b6itBNy31EKC5mtuSUG1daqgKck731YOLB5FGpUMtDd50AuRXyikDUphvgyeNmXQGtXArwYXH9cw8m1VOGd0HkNtdL7rTX7kgzY4rtBrm8DTGGobRXbrwoy33367HXbYYZ4Ae8GCBYP/f9hhh9nhhx9u++23n61atcrmzJnjzVMsFqUxEkIIIcQ2GdGQ2aRJkyxJElu/fn2Dff369TvU//T19dl9991nZ5111g63s++++9qkSZPs6aefbml/hRBCCNGejOgbokKhYEcddZStXLnSTj75ZDMzy7LMVq5caYsXL97u2K985StWqVTs7/7u73a4nZdeeslee+0123PPPVvbYQxdhXUSj6ALd+jrP+qSDm/KsR1DAu0YsqJviyHdP4bOx83tPCJo+RFRmw5q8UGtNmgs+EXUuoPeu5KthZBZaEiKQ1yhYcnA1h3UpgO2QS05glt35Bv3L8v7+5timw7PtI22H+CHHevBBvcAhu8CwZR9aNdD3borcdijki5HSk0uQ0mFMnwwvTk/DftPdT/lmlt3+CUAxiaNqePUfqMT+hBx6w7fj1pyFCI/1Tu4dUdoTxcgM//zTCEUROenhu084JxBSw4KhfVDm45yYCo+tekgvz5oy7G5ydZfh/mhTUc/te6o+ftbqUKbDuhsn4HNwb03zJ1adpoRL8y4ZMkSu+222+zuu++2devW2bnnnmt9fX2DWWenn366LV261Bt3++2328knn2z/6T/9pwb7li1b7MILL7Sf/vSn9txzz9nKlSvtpJNOsv3339/mzZs30ocjhBBCiLcgI64hOvXUU+2Pf/yjXX755dbT02MzZsywFStWDAqtX3jhBYub/tp+6qmn7Ec/+pE9/PDD3nxJktivfvUru/vuu23jxo02bdo0O/744+3qq6+WTkgIIYQQQ2JURNWLFy/eZohs1apVnu3AAw80t43sgo6ODvvOd74znLsnhBBCiDZnt84yG21InwAhYiZwbAZdKiDEbnEK8dW6b6OWIRGUTo9TSFmnFiRpsw+Mg7A+7gdpeahaO6Xit5CS2UocOrjcAfpR+nvYWNKRgWQhWG9G+psM9EzN28C5YD9Q40b7S2Nb8UsCS1TQNQD3lFUpDRnc4J6q5/wPoZzzb/rNga1AyJaD1iKFHGhyoLVGIQFbkx/OD3PlwEbtHcgvgRuSxmILiRZu5tCU8NC0cyqBgCUVKMUeLuZ6Rn7+Nqop6JnArwZ+9abrFueq0zgo9xCqDYJ7xcF3VwRaPdL0jSZq7iqEEEKItkcLIiGEEEK0PVoQCSGEEKLtaW8NUeQaBCdYMwZi2CnpHahcBsSXsRYK6WoodA5jg0PstC8hOp2hjtuWH9FK7YnRqFvRQqsWIrhbRLB2afjGtjR/oF7K4kAdUPB8vgmvPdAxkK7I1eg54JtAAmIZbBhkSjZAOxhYAos+P772duxH7XBCr2Mci46h84X5tUILnUDC68qhSHDofnyz0YbJ1jgWjz+0HlDgWNaJtrCNUURviIQQQgjR9mhBJIQQQoi2RwsiIYQQQrQ97a0hagJrnFAtlFbi+oF+odvAuHuoLiBEd4C6hlD9w/BqDHDo7qQ7aEVjEDjfsOsOsgCNQYA2wWxb+jgS34Rtg3QHwVo12j+ou8VavUC9Q6BfqG4w/NiG7ufd3qNR72s0tIShtPK8aEUjuAv0gERLWr1A3SD2BEW/wPnoO3OE0BsiIYQQQrQ9WhAJIYQQou1RyGwr8HVdDl7/5f134BH4RVCyP875tgRsOSi7T6X9qTx/jloF0HwB5f65jD+V+/fbCeQCS/HTfoSW9ieGu9x/aGn/lMr4U7l/GEt+VMZ/uMv9V+uNjwCan0r7ky2Fkv2h5f45/R38KHWe2gJQWxpsmxPYDofa67TgB5f8Nlr4hPoNsTUPtc0JbMMTHh4cemueYW/DQwSGpKhtDod94BmCbXjABqVfWmrN0+QX3DYntIUPzhd2DNgWi75vm3URwSd259EbIiGEEEK0PVoQCSGEEKLt0YJICCGEEG2PNERbQy0FSC8EtlzRD7zn834Qv1jw/TryNc/Wma96trFg68z5tnG5imcbA7bOGLaRK+/QpxTB/sb+/AUQMZRif2weRAt588cmIFBISGuEed1hZPA3QgpCgRoE8asQZCe/siv4Ngio92fg53y/LWnJH5v6Y7ekRc/WV2+09ddhHNj6ar5toOrvW7nm2yo1/3OqV0BrRVIBEl6A/gj1LeAHl6Ml/iVvcHlbAmPhdrGk6u8MbSOpwLVcg7Fgi6ug16uBrckvSuH5VvPvvagO91QKfqBlMtgG9jqimg+gXQoGNT+BtsS/zlwM114uzM/l/evb0dg8aO4KYIOxWQF0jU1j4bFgKY4DHRCNBR0Q+Zn/6EGoHc5odvPQGyIhhBBCtD1aEAkhhBCi7dGCSAghhBBtjzREW0PhZdAVUS2hHNhILzS26IsRxhV8W1fet43PD4Bf2bPtkevzt5v4fhOSfn9f4oGm3/1xY0BQMQZ1RaAhojpE8MGXIj/mHoNfAvH/uIV1fmb+cdScf25TiGxXwK8Muogy6GD6QBtEGqI+0B9tTjs828a00/fLfK3Rn2pjGn7vrcNcNd9WSvxre2Pkzx/aWsWBYCiDukYkIQouNwOyFWznQboispGuCHRAuTLogMAvKYNGDrRByQBod6pQy6vi73RUbTpvNSg6VIWDrYGt7m/T1f35HI0lvRBoklwLGqIINURwAYFflIOvxsR/JkV53y/Kg7AG/Az8XNH3cwXflhX9saQ1ikuN+5wUfZ862Kj+VR1FfQDVZgqsa8Q1htS6QwghhBBi1NCCSAghhBBtj0JmWxPY4Tc0DEAtJNAW2KaCbDl4509+efALsVHqfB7nBxscV2h4LA+vtikUloexrUDtN2LYl5rzP5cssM8AhdsKMF8NyhEk8J459HxTiYLmtil4jVFrldBru4XeCxF1zUYbDB6F1hCh28DtYosL8KM0dmqFMdQ0dghTYUo82Cg8hvNB2r0jPwg5txIyg1vKIgp7wQVEW+UqEBBuo/R8skGLJS5bEHa+I4gne9dZ6D2AfnQthsbEW7CNInpDJIQQQoi2RwsiIYQQQrQ9WhAJIYQQou2RhmhrKH6J6cBQ6hxShKlFAbWfIDKIWGcQI65k/jYGoJ76lsRPid6U89Opm9PsqXUHpd1TS45SRG0/oE0H2vzPqWC+X0w50i2QQUw8vHWHnxJfM2rdAS0uoHVHX+bXu+8HG6XT90ObDmrd0VtvHLu5DnNB647eKuwbtfOAe6BWhzYdlGIPdfwjuB9RA0Gpv9QWgNKBwZaRVI1aFIRqKsAtpnYROSg1Ae0d4gKUqYC2Kc3p+VHNP48RaIOodUcEafekF4qyMD/UPJGtFUgAivoessFFADaXAz9s0wFjwZbBuaUU+5TafjSl1Kd5eL6VqHWHZ+J2HtS6A1YVdE+FanZHE70hEkIIIUTbowWREEIIIdoeLYiEEEII0fZIQ7Q1JO+pU40KPyBaC2w9QPqJ/ipoNHK+luO1xI/FF3J+vL8IbRWo1UIJxhbietPv/jaLsT+OatWQH9XHoVo1CZyM5po524LqOhGk0yJSEJ+Q1oh0RRn8zVFJQVcTqA+rg5ilDJqxKvr581XqjTbat2oK1ztcx1XQC5G2Lq3B5wnzuRrohUC/R6ebpDyoDQp9AsKfjjSWNBUpPEPqYAMpncV1qO0EnxX77Xg+bmcCdW9CbVhzieohgV+oXihUVhRaIieBZzxpjVCXBmNB90X6tYy2Cza8zmAbIX7ULoPHBdpIQwS20O06aJU1mugNkRBCCCHaHi2IhBBCCNH2KGS2FfQ63tG7d3g9bfCqs17xX23XYwhJ0GtCWqrC++gIxqINGz9De5Dmsdi6BOb33dgPbf7Y4W75QOC5DYRCZryNsO2ijeaD0CyNzagzNW0ja/6dfMLmohIV0I0BU+cN7j0M57TScTvnf6IQ0bMMUpMjSEPm/YN9CWzdEXy8GG4K9dvxNoPbj4BbaDhr2Fum4EZa2ERgSjjuSujYUD8qF4HfDzv2w3GB8+P+ho5NoOVO8HE1X7QjF1YblTdEt956q+29995WKpVs1qxZ9uijj27T96677rIoihp+SqVGPY1zzi6//HLbc889raOjw+bOnWu/+93vRvowhBBCCPEWZcQXRF/60pdsyZIldsUVV9gvfvELO+KII2zevHn2yiuvbHNMV1eXvfzyy4M/zz//fMO/f+5zn7Obb77Zli9fbmvWrLExY8bYvHnzrFwub2NGIYQQQohtM+ILohtvvNEWLVpkCxcutEMOOcSWL19unZ2ddscdd2xzTBRF1t3dPfgzderUwX9zztlNN91kl156qZ100kl2+OGH2xe/+EX7wx/+YA8++OBIH44QQggh3oKMqIaoWq3a2rVrbenSpYO2OI5t7ty5tnr16m2O27Jli+21116WZZm9613vsmuuucbe+c53mpnZs88+az09PTZ37txB//Hjx9usWbNs9erVtmDBAm++SqVilcob7SZ6e3txu9gWgDK9IZgaHNYMjM8TuI1QHUxoRmuAX2h27AiGet9gNLYxzOXkW5AuBe8L6bKC/EL1V62U3W9BTxE6n4GODttvIGGCmeDziLnooWMDbaHbHepcQEvXMTEaD4xh32mflr4L0G8Yn/EtfP8M+/ceMCrfGdthRN8Qvfrqq5amacMbHjOzqVOnWk9PD4458MAD7Y477rCvfe1r9m//9m+WZZkdc8wx9tJLL5mZDY7bmTmXLVtm48ePH/yZPn16q4cmhBBCiLcQu13a/ezZs+3000+3GTNm2LHHHmv333+/TZ482b7whS8Mec6lS5fapk2bBn9efPHFYdxjIYQQQrzZGdEF0aRJkyxJElu/fn2Dff369dbd3R00Rz6ftyOPPNKefvppM7PBcTszZ7FYtK6uroYfIYQQQojXGVENUaFQsKOOOspWrlxpJ598spmZZVlmK1eutMWLFwfNkaap/frXv7YTTjjBzMz22Wcf6+7utpUrV9qMGTPM7M+aoDVr1ti55567czvoooa4KOmFIiixTyXxcWygraX5sPQ+bGOI81HZ/fAaKr6Na7KEtQBAdpGGiDUv1AIAxobW7kBbWFsAalMR4odtDALbVuA2A9tlhI+FeibYAoDmg4uF5gO/CPyixL/AIxgbg1+SAxvUBcvBWLIlcLOhX9M2cjAO94NsMJbqh7ViG25C64eRXz2DljPwcCA/slFLoJT8wFajbUBRrXpT2x1qpUM1y9Cv7tsctdKBseRH363UKgu/R0aIES/MuGTJEjvjjDPs6KOPtpkzZ9pNN91kfX19tnDhQjMzO/300+1tb3ubLVu2zMzMrrrqKnv3u99t+++/v23cuNGuv/56e/755+0jH/mImf05A+3888+3z3zmM3bAAQfYPvvsY5dddplNmzZtcNElhBBCCLEzjPiC6NRTT7U//vGPdvnll1tPT4/NmDHDVqxYMSiKfuGFFyyO31hR/ulPf7JFixZZT0+P7bHHHnbUUUfZT37yEzvkkEMGfS666CLr6+uzs88+2zZu3Gjvec97bMWKFV4BRyGEEEKIECLnQpOo3zr09vba+PHjba9rP2vxVououOr7xlUImdXIz7clNB91oMax8EoZtpvUwI/GUjfsAL+4DmEBtEFIgfwyeP9JXbPJj/tg+LZWwB4nYTYKN1kMr49zYEvgFTV1zSY/aDXh8vDqHfyax6YF8vFMPBe0t+Cx4EdjA+fLCpBiDzZXgGsKbFEewkMFP76cz/s3cyHv+3Xk/Ru3lPNtncE2/4ExJlfx/RJ/bEfTw6YED5Ui2EoR2MAvD3F4siXQVySm8B31H2mBFGSzGYSuqhCvrUFstgZ+ZedfpBW4cPvhAi+D3wDcMH1pMcivv15o+h3mJ1vVt5Vrvq1ah/ZUVf9zSqvwzKv5tgj84lrjsyYrl+35iy+1TZs2DbseeLfLMhNCCCGEGG20IBJCCCFE26MFkRBCCCHanhEXVb+poNLkEMImHRDqhcCWg/6zScXXO5CGKFcGzU8F4u5o8+P4UdW3xZVGXUBUg9z8GnwANRA4kV/dt7kUtpGS1sj3c6Q1IkhrBDqgCDQ/qCFK/Nh5DDbLkQ1uuzyIY/K+nyv487kC+UEcv+iPzZpsaRG0R6ArSkuQ4gunOy1S3QHfhKUN6FQEt98InQ/uPdABFQr+wXUWQMsDtnEFX9/TVfAfBOPg4dAFtvG5Ad+W9PvzJbCNuHHsmBi0RxEcFzzMSqANKsEDE+RmlocTnsB9FpMf2NLA+hsZ+NXg2VAFWw22W4aLqh90QH3O1wv1Zb4OqB9sG9NOz7Y58xOINtV9v956abu/m5n11jo825acv79bqv6+bamADsqzmDmotUHp/sPdJmln0RsiIYQQQrQ9WhAJIYQQou3RgkgIIYQQbY80RFtDYWiykWwluJ0F6ICo1QbVF6JaQmQD3U+IXujPfk1aCfBhvZBvc1XwIw0RaI1IV4RaIzfMdd0jqI0B7TIi0As50AaRX1SgAjtwHBmcR9AihBIH1FPC1iBQXymD2lExlOenwwpu/RJ6P1J5KmytAvWuqIVEDLoiaF2RgF8+Af0R2UCI2AF1g8YmvsZnLGmDAvRCf7Y1+nXF/rhOqC80BvRCnXA95UHkVQJbznxbAvdePMx/s2fwoK45/9jIVjHfVga/AmwjD355mI9asISSQY2y5jpJVDepCvqeagrnJ/AeoBYscGpZLyQNkRBCCCHErkULIiGEEEK0PVoQCSGEEKLtkYZoa0JjmrCMhNAs2qg/VZoPE0GgpoJ2JoK6NKQNgZ5acb5xpyOoZxNBnxurg96FtEZQXyiiAjaoqYH+VKQhAr9gSC9EAXCI11NtIrRhHSI4Z1DDCGsO5cnPt2EdokLc5APXJ/Y3o+sY/OBQs8B7heoGhd6jdK84uKcc1EJJU9BiQM+mauJfe9QXKgfaixiETwnf4P6+YJ8t6KkVQz2cuLGWTH/i15YpZX7NIapNVAIdVCmC/m5GPc/8+eh0U3vAUEDmhvLPFC6gGlwrZfzcob4Q2Ki/2ebUr/9DtYmo5tCWlOoQQT2hpp5nm2swVw3qIcF1XK37N3Md7pU0g75lqLsNrFE2iugNkRBCCCHaHi2IhBBCCNH2KGS2FQ5SCCnlOA3MfKZX/hQuwPBD3bfVOyjF3p8vrvsbTqoQ+qLU6aY0fkr/p3Fsg/AYhLNoLL06jULbdLSSiY9hGgi1UAp7YMo6hy8h3ERjKQUerh++bimkFTX97rng/NCdgG0wH90/oWMx3EZhFbp+oCyAg3sF3KwKG8nARqG1gZp/cL05P7S0oeKHPEoJhKVgbCf0DirC2I6mBwb5FCP/oVKE8FiebJCeT7YEblJK1ya/UNLAv/cp3JjBwzs4VAkXM/lV4AIfSP2xA3DDVOFG6K9DqC5t3AaFdMtwfVLafQW+Q+pwvac1/7PLahDWhxuthaoDw4LeEAkhhBCi7dGCSAghhBBtjxZEQgghhGh72ltDFLnG/FzQdmSQEk+tHBx8khA6xxgpZKxvYyzEXAP9sLUI6nS2//u2bKz5CfQLbdEARG7k8zRRLxSYDhyaOk46GBqL8wX6hWwDx7WQJt+SH9roAvJNCOjyItoZ0DbUQReRVkBnATrEiNobwHHEkMaPY7HdSGBbhSY/8qH2ESyZC2txQmDLlMCyA61Aui+CSjTQWPJL4fmbwUVK6ekZlYEI9MP9a/KjkhJUeoL0PQ72g+6VUG0Qafrwu2sU0RsiIYQQQrQ9WhAJIYQQou3RgkgIIYQQbU97a4iaCNUx7PL64mKU0fl+s4H6ONDbYdGhYFHS0KErinZPiF3Brrkrdj16QySEEEKItkcLIiGEEEK0PQqZbQWl9GKaL/lR0/nAsRGkqtJYSsFNWkrV3XHabIiP2Ta6d1MKcmC6LfkRuypVl9JtCUyZxflgLFxUmNIL2wj1a94XSsHNIFU3w7RcKvcQmJZLoatAW3BKL5a3CPUDG263Bb/QsUMsoWFmXvmNVuYPLrVBYPmNXVRCgwgsjYF+oaU2QstlBJbQCCm/MexlMFoqtUHft+CXGz3Jgt4QCSGEEKLt0YJICCGEEG2PFkRCCCGEaHukIdoKB5oXR6078lAmP+8LABLwy+V8v2Kh7tkKie9Xytc8W0fOt5USf77OXNXfRuxvoyOpNv3uz1+M/fnzIIAoxf5Y8iNbDGIE8iMS6lMSSAp/I5Dmh/xqEABHW+bfdhWyubxvS32/gSzMrz8teLZyvdGvDOPKdX/+5nFmZtUa2Or+8dervl8KrTEc2CKwWZUEGr4pgtYdcIkG2xL/ljK4NXZiG6Cvo/nqoNerwnMqhfmatoE+YItgm1EWtk0jP9ILkW3ot3Lwn/suJgEo+CWgh6M2TglcyzQWbFkebLmwsXB7e2PhUbGNbfp+KY71bfDYwn1zcFwpidWan7+B+s2hoDdEQgghhGh7tCASQgghRNujBZEQQggh2h5piLYmsJZQDHqhXBE0NAVfKNAJtjGFimcbm/cFCl35smcbB7axiT/f+NwA+MF8caNtTOzPNS7x5ypFcFyRfwyoNUJdEegkQBhCK3oIrweDEgjwq0Ecuwp7U3agyYEAfb8DfQ/YetOSPzYrerbNme+3qd7p2bakjWM312FctcMfV/f3bXPV34/+qu83AOe2GoGuiOQEUNcogtoyJEUAyRxqdFAb5N8G6Ae3nuUqcC2jDbSJ4BdXyc8/uKgKtlra5ON/AFEdPqgafFB1sKUwFm3+MTgHdxroj4IBfU8UwRMD/Ix0RTn4ugS9kOVBRJPztXSu4M/n8r5fBn5ZAbSORRrb6JcW/GOtl0iPBHq7Ivj5tzydWoTKjEVUt2wUWyfpDZEQQggh2p5RWRDdeuuttvfee1upVLJZs2bZo48+uk3f2267zf7yL//S9thjD9tjjz1s7ty5nv+ZZ55pURQ1/MyfP3+kD0MIIYQQb1FGPGT2pS99yZYsWWLLly+3WbNm2U033WTz5s2zp556yqZMmeL5r1q1yj74wQ/aMcccY6VSya677jo7/vjj7YknnrC3ve1tg37z58+3O++8c/D3YhHe3e0sWA6dWmj4btTOIgdtNQo5Son3w01j8xCqgvBYV863Tcz1ebbxSb9nmwC25nDYuMiff0zsxwrGRHBc8NmV4MMrmv+qNx+F2eJhXtNnECBL4VV+HXqTl51vqzn/s+qH+Euf8z/nfshp7Yz8sRQeK2VDL3nQTArhwQxuljq0BqmnkHafg1Yj1B4EwhYOYmGYhdtCqwn6SILDbZQ6DzZKk8cwWhlKUkB4LK745xvDYZUmWxXifhQeq/rzOxjrKDwGoTXycxQeozBaK8DzJ0r8axRtefi6hDBaVICQGYTRIrIVwVaD0CeUwqAyCOaF7Ome8ocZhKGp1QaVBMBHCp3GwHt0NBnxN0Q33nijLVq0yBYuXGiHHHKILV++3Do7O+2OO+5A/3vuucc++tGP2owZM+yggw6yf/3Xf7Usy2zlypUNfsVi0bq7uwd/9thjj5E+FCGEEEK8RRnRBVG1WrW1a9fa3Llz39hgHNvcuXNt9erVQXP09/dbrVaziRMnNthXrVplU6ZMsQMPPNDOPfdce+2117Y5R6VSsd7e3oYfIYQQQojXGdEF0auvvmppmtrUqVMb7FOnTrWenp6gOS6++GKbNm1aw6Jq/vz59sUvftFWrlxp1113nf3gBz+w973vfZbS61ozW7ZsmY0fP37wZ/r06UM/KCGEEEK85dit0+6vvfZau++++2zVqlVWKr2hk1iwYMHg/x922GF2+OGH23777WerVq2yOXPmePMsXbrUlixZMvh7b28vL4oCY5qcHQopiZBCmILOopr58eoy1EkvBLbM2Bz5mhJqhRFCCjqOGmh+apBiXzVoK0ItOcCvAJqkPMS1YxR++STgl8LJzcCWQkuBGviBVMTKLiwVvy/z09MpFb8PU+z9tPjNkJ6/BW2N8w3AdVeFMgFV0AbVwBZ6D2SUbkviBrBRij1eFmHTGZyyYBvcyhZTiwbQY0TQQiGCz4paXERwTYVAJQsspoMA7Q2kklPKPumFIvrjFY7LZfDcohYfdBxAROn0oBeidHrSFWEqPmmNKBUf/DDtHnRFrgB+NLa5dQdofrB1BxwqtimhVyqhtsB7dDQZ0TdEkyZNsiRJbP369Q329evXW3d393bH3nDDDXbttdfaww8/bIcffvh2fffdd1+bNGmSPf300/jvxWLRurq6Gn6EEEIIIV5nRBdEhULBjjrqqAZB9OsC6dmzZ29z3Oc+9zm7+uqrbcWKFXb00UfvcDsvvfSSvfbaa7bnnnsOy34LIYQQor0Y8SyzJUuW2G233WZ33323rVu3zs4991zr6+uzhQsXmpnZ6aefbkuXLh30v+666+yyyy6zO+64w/bee2/r6emxnp4e27Jli5mZbdmyxS688EL76U9/as8995ytXLnSTjrpJNt///1t3rx5I304QgghhHgLMuIaolNPPdX++Mc/2uWXX249PT02Y8YMW7FixaDQ+oUXXrB4q7ju5z//eatWq/Y//sf/aJjniiuusE9/+tOWJIn96le/srvvvts2btxo06ZNs+OPP96uvvrq1msRkV4I6otndai3UoUWDbCJjDREdT9gOwC1LLbUfE3JxpzfjqEz5+t5OhJorQF9Bpr9SjG0H4E6REXwo3YeXAvH1wsVwI90UAkWuBg6KfyNkEGgnPxqDnQ1YCtDfaGy820V8gPbAOmPQAvUV4fWGk0tOPqhvgnZymAbqIH+qAafSQ3adIAf1fYfohTOzLahA4InIMlWUF9I+iPSC1HHB2i9ALeQJVXy8z8rqn8U1fwDaa6JFEGfhbgOc4FfVCcxJfhhD5YwW0R+rUBaI6q5QzWw4NyiH9TZcqDdIb8MbXBNQWuNEL+U9EIgb6LWHfCYwWs7DfRD/REKAkePURFVL1682BYvXoz/tmrVqobfn3vuue3O1dHRYd/5zneGac+EEEIIIdTLTAghhBBCCyIhhBBCiN26DtFoE4FmwWoQD6aaQ9CLKYX4fxXqj/Tn/KBrAn3QYugNRn65xNffUF+1PPRfa+7JRj3aEhBy5KiXG9hiEGOgX2AsOdSvFajGVCu2OgTPqQ8Y1euhsehH1yNtt8kPx9G+gY4ug7GkwXPkV6eeSHA/gh9q/6hGUC5MCIRjQWeRQcsv6uOEvdHoeLGvWlivqNCebM1+1P8K5wq1YS030gaRH9iGm8A6N1yfivRH5NeKLex6BGki1w5qsoXOhfOTDihwbCvzjSZ6QySEEEKItkcLIiGEEEK0PQqZbQWm9NbglTXY+FWs//6PXsWm8P4Yu7K1Uuo82C/gvXULr52HndFI0xzmA2lpl0PHhu5z83zB8w99m3T8rXwmoa07KMU+pTBa6P618FkFHy/GbsKGBm2jhWs7/JwN8VocCUbhmRT8kQ73czTED05aS/O3cgzBNrf934cRvSESQgghRNujBZEQQggh2h4tiIQQQgjR9khDtDWQ4goZ4eHpsaRJCkxVDR4baEMdQ8jYVvQUNBbcWtJYBM6HtKAnaCXujrsXGE/HtFkaO0S/VlKGW0pBplRdKDOBx4XzhV18odswmI/GRjSWtgEXeITzBW4jdL4mP/Sh6Wl+tMFgoJVyGbRdgsqjhELlMrjbCPmBDbaB+0fbyGA+KIWB+9c0lubCBwN8N+BYsOH3D5XQCBw7mt089IZICCGEEG2PFkRCCCGEaHu0IBJCCCFE29PeGiIXNcRPscQ+1ByKqWQ/2OJaoI3GBtpwu3U/6Mrz7dgvSsEHbBj7hfmpVQBriGAsbBcZZg2RI2FEaNl9Ctkn4Bdqw5YU0DIE7uwQP/Kh+j08f5jNURsMstH+gp/LB2pKoOaQy/sXbgTzxTn/4ZDAWGqlU8j7N18e5itAy51i4o8t5nxbCfwKsT9fKWl8AOXApwgPizzc4LRvMYhP8vBgJQ0RtQQiqP0PkQXe4DXoeZHCDU66ohqI3ypw0ddhG+RXBb8q+JXrYEt9W6XJVoVx1dTfZq0eZktrZIOHFLT6oTY8jur+UbueEUJviIQQQgjR9mhBJIQQQoi2p71DZk1EkAaIoaZqmC0hW4X84PUxjS2DXw1e71fgVX4NOspXydb4ejuq+a+7yWZ1CD3UwS8NtcHr84xyQYc5J5PCY2RL4G+JGHLHyS8HLV1ykEab829Pl4exBd+W5eGVfwG61hcbbWkBwlRgqxd9G41Ni54Ju8RTmi9BIcOIwnL0lh1S5yk8lhT86zFfgNBVwY9/d5It79/MXQX/QdCZ8/3G5cr+WLCNhQdLJ9jGxY1jO2PfZwzYShDrL0W+LQ9NhwoQMqMwWgKhsFbS8wkKe1WhxVINLrQaxI7LaCv4Noj19mX+zdGf+WM3ZyXPtqUONrjZepv8+uvgU/NtfTV/Pwaq/jH0V32/agIhwyo8t+Bzx/T8UURviIQQQgjR9mhBJIQQQoi2RwsiIYQQQrQ90hBtTWj7CUrPD0y7R70Q6IpypBcibVAZbBVIc61AimzZt0XVJlsVDoJsNd/mwGY1f5uuDh8e6IocpOw70h+1QASp89QrIUr8+HeUh9uJ/Ap+LD7Kkw3mK/gxe1f0/eKCb8vIryltNiqBHimlGgO+CQksWZDBaaT7LLQtDZY7gDYVMfTmiSF1ntLkOyCdnvRCE4qgA8oP+DbQBu2R7/ds4xN/7PjE9xsHfl1xo61ZU2RmNga0QZ2Qnj8GPuM8nO8iaEXykW+LIU0+wT4lQ6fm/OPIQMBWc/5nUHb+Z1UGDWM/6I/6oNYE6YV6QS+0OevwbBuTTs+2KfX9OprKLGyEL6oYbqrQ0gYpHGsKz4sMtJQptrQJ2uyIoTdEQgghhGh7tCASQgghRNujBZEQQggh2h5piLaG4pehtsD5sA0E1NrAOiqoi6BtkB/UuYHWENZki2Ac1taBkvMR1A0iHRAdKkWwI/icSPND2wgFNUSgA+LaRPAZQC0ho88UP2fYLmqcqMUHFewJaA8SeI0N+70ChN4DZKPyNQ4mJFsoVNMm1EatHDKqfUNtJeC+pRYSNF9zSwqqrVMFTUne+bZyYPEo0qhkoL3Jg16I/EIhbVAK82XwtCmD1qgGfjW4+LiGkW+rwjmj8xhqo/Ndb/IjH7TBcYVe2wSexlDbKKI3REIIIYRoe7QgEkIIIUTbo5DZ1mDoKqyTeARduENf/1GXdHhTju0YEmjHkBV9Wwzp/jF0Pm5u5xFBy4+I2nRQiw9qtUFjwS+i1h303pVsLYTMQkNSHOIKDUsGtu6gNh2wDWrJEdy6I9+4f1ne398U23R4pm20/QA/7FgPNrgHMHwXCKbsQ7se6tZdicMelXQ5UmpyGUoqlOGD6c35adh/qvsp19y6wy8BMDZpTB2n9hud0IeIW3f4ftSSoxD5qd7BrTtCe7oAmfmfZwqhIDo/NWznAecMWnJQKKwf2nSUA1PxqU0H+fVBW47NTbb+OswPbTr6qXVHzd/fShXadEBn+wxsDu69Ye7UstPoDZEQQggh2h4tiIQQQgjR9mhBJIQQQoi2RxqirSB9AoSImcCxGXSpgBC7xSnEV+u+jVqGRFA6PU4hZZ1akKTNPjAOwvq4H6TloWrtlIrfQkpmK3Ho4HIH6Efp72FjSUcGkoVgvRnpbzLQMzVvA+eC/UCNG+0vjW3FLwksUUHXANxTVqU0ZHCDe6qe8z+Ecs6/6TcHtgIhWw5aixRyoMmB1hqFBGxNfjg/zJUDG7V3IL8Ebkgaiy0kWriZQ1PCQ9POqQQCllSgFHu4mOsZ+fnbqKagZwK/GvjVm65bnKtO46DcQ6g2CO4VB99dEWj1SNM3mugNkRBCCCHaHi2IhBBCCNH2aEEkhBBCiLZnVDREt956q11//fXW09NjRxxxhN1yyy02c+bMbfp/5Stfscsuu8yee+45O+CAA+y6666zE044YfDfnXN2xRVX2G233WYbN260v/iLv7DPf/7zdsABB+zcjkWuQXCCNWMghp2S3oHKZUB8GWuhkK6GQucwNjjETvsSotMZ6rht+RGt1J4YjboVLbRqIYK7RQRrl4ZvbEvzB+qlLA7UAQXP55vw2gMdA+mKXI2eA74JJCCWwYZBpmQDtIOBJbDo8+Nrb8d+1A4n9DrGsegYOl+YXyu00AkkvK4cigSH7sc3G22YbI1j8fhD6wEFjmWdaAvbGEVG/A3Rl770JVuyZIldccUV9otf/MKOOOIImzdvnr3yyivo/5Of/MQ++MEP2llnnWWPPfaYnXzyyXbyySfbb37zm0Gfz33uc3bzzTfb8uXLbc2aNTZmzBibN2+elctlnFMIIYQQYntEzrW0Zt4hs2bNsv/yX/6L/Z//83/MzCzLMps+fbp97GMfs0suucTzP/XUU62vr8+++c1vDtre/e5324wZM2z58uXmnLNp06bZBRdcYJ/85CfNzGzTpk02depUu+uuu2zBggXenJVKxSqVNyq59vb22vTp022v6z5jcWmrKqC0ig19o7PbvyEKHOu9IRriuG35EXpDFDyf3hCF+bXWMDnwrUnodgP99IYozK8V9IYIfEbhDVEr22j2y8ple/7iS23Tpk3W1dUFkwydEQ2ZVatVW7t2rS1dunTQFsexzZ0711avXo1jVq9ebUuWLGmwzZs3zx588EEzM3v22Wetp6fH5s6dO/jv48ePt1mzZtnq1atxQbRs2TK78sord7i/mNJLqb+tPKQC/UK3gQ+R0IdcyEMUH9KhD/PhfWDi0N3pIdrKAzNwvmF/iGYBD8yAB61Z+B8F2I0h8DV7+AId9g/KTPAfHoEP70C/0D+Chv2PjwDbqJS3GI0/jEJp5XnRyh88u+CPG6KlPzwC/wgK/6MgcD76zhwhRjRk9uqrr1qapjZ16tQG+9SpU62npwfH9PT0bNf/9f/uzJxLly61TZs2Df68+OKLQzoeIYQQQrw1aYvCjMVi0YpFv/GdEEIIIYTZCL8hmjRpkiVJYuvXr2+wr1+/3rq7u3FMd3f3dv1f/+/OzCmEEEIIsT1G9A1RoVCwo446ylauXGknn3yymf1ZVL1y5UpbvHgxjpk9e7atXLnSzj///EHbI488YrNnzzYzs3322ce6u7tt5cqVNmPGDDP7s0h6zZo1du6557a0vxi/zEE8NO+LAiLwi6Bkf5zzbQnYclB2n0r7U3n+HLUKoPkCyv1zGX8q9++3E8gFluKn/Qgt7U8Md7n/0NL+KZXxp3L/MJb8qIz/cJf7r9YbHwE0P5X2J1sKJftDy/1z+jv4Ueo8tQWgtjTYNscfy36h84X5wSW/jRY+oX5DbM1DbXMC2/CE66WG3ppn2NvwEIEaHWqbwzoYeIZgGx6wQemXllrzNPkFt80JbeGD84UdA7bFou/bZqFo8IndeUY8ZLZkyRI744wz7Oijj7aZM2faTTfdZH19fbZw4UIzMzv99NPtbW97my1btszMzP7hH/7Bjj32WPtf/+t/2Yknnmj33Xef/fznP7d/+Zd/MTOzKIrs/PPPt8985jN2wAEH2D777GOXXXaZTZs2bXDRJYQQQgixM4z4gujUU0+1P/7xj3b55ZdbT0+PzZgxw1asWDEoin7hhRcsjt9YKh9zzDF277332qWXXmqf+tSn7IADDrAHH3zQDj300EGfiy66yPr6+uzss8+2jRs32nve8x5bsWKFlbZOoRdCCCGECGTE6xDtjvT29tr48eO9OkT0Wk8hM4XMFDLbNSEzCwyZUdgrPBSmkFnI/AqZKWQWHjILm49CZo6+b5vmy8ple/6Sf3zz1SF600EF42jxA7Zc0X+K5PP+E6lY8P068jXP1pmveraxYOvM+bZxuYpnGwO2zhi2kSvv0KcUwf7G/vwFeCKXYn9sHp7AefPHJvC0TWjhhIVuwsggzyCFp14NnkhVeBKQX9kVfBs8Hfoz8HO+35bUfzPan/pjt6R+pmVfvdHWX4dxYOur+baBqr9v5Zpvq9T8z6legYUjfaHRtwgspvDLGvzgcrTEv+QNLm9LYCzcLpZU/Z2hbSQVuJZrMBZscRX++KiBrckvSuH5VvPvvagO91QKfrAwM9gG9jqiv81hIRYMLmACbYl/nbkYrr1cmJ/L+9e3o7F5+AOiADYYmxXgj7SmsfBYsBTHwcKMxsKihvwsMMmb2uGM5hsbNXcVQgghRNujBZEQQggh2h4tiIQQQgjR9khDtDUUXgZdEQmjc2AjvdDYoi9GGFfwbV153zY+PwB+Zc+2R67P327i+01I+v19iQeafvfHjQFBxRjUFYGGiETV8MGXIj/mHoNfAvH/uIV1fmb+cdScf25TiGxXwK8Muogy6GD6QBtEGqI+0B9tTjs828a00/fLfK3Rn2pjGn7vrcNcNd9WSvxre2Pkzx/aa86BYCgDkTZJiIK1s9QcmUTLpCsiG+mKQAeUK4MOCPySMmjkQBuUDIB2pwqJCRV/p6Nq03mrgYK6CgdbA1vd36ar+/M5Gkt6IdAkuRY0RBFqiOACAr8oB1+Nif9MivK+X5QHYQ34Gfi5ou/nCr4tK/pjSWsUlxr3OSn6PnWwkZi/jqI+gITmgSJtVsK/RXqZCSGEEEK8GdCCSAghhBBtj0JmW0Nv6wKzNAmqh4O2wJo7ZMvBO3/yy4NfiI1S5/M4P9jguELDY3l4tU2hsDyMbQWqJRTDvtSc/7lkgUVTKNxWgPlqUI4ggffMoeebShQ014DCa4zqRIVe2y0UkolgrEMbDB6FOjeh28DtYr0e8KM0dqrrM9Q0dghTYUo82Cg8hvNB2r0jPwg5txIyg1vKIgp7wQVEW+UqEBBuo/R8skG9OC5bEHa+I4gne9dZ6D2AfnQthsbEW7CNInpDJIQQQoi2RwsiIYQQQrQ9WhAJIYQQou2RhmhrKH6J6cBQ6hxShKlFAbWfIKjfFfXUqkAzmQGop74l8VOiN+X8dOrmNHtq3UFp99SSoxRR2w9o04E26Jdmvl9MOdItQL3Mwlt3+CnxNaPWHdDiAlp39GV+vft+sFE6fT+06aDWHb31xrGb6zAXtO7orcK+UTsPuAeoDxqm2EMd/wjuR9RAYI8psAX3iYJtUIuCUE0FuMXULiIHpSagvUNcgDIV0DalOT0/qvnnMQJtELXuiCDtnvRCURbmh5qn4W61SQJQ1PeQDS4CsLkc+GGbDhgLtgzOLaXYp9T2oymlPs3D861ErTs8E7fzoNYd1N+M7p9Aze5oojdEQgghhGh7tCASQgghRNujBZEQQggh2h5piLaG5D11qlHhB0Rrga0HSD/RXwWNRs7XcryW+LH4Qs6P9xehrQK1WijB2EJcb/rd32Yx9sdRrRryo/o4VKsmgZPRXDNnW1BdJ4J0WkQK4hPSGpGuKIO/OSop6GoC9WF1ELOUQTNWRT9/vkq90Ub7Vk3heofruAp6IdLWpTX4PGE+VwO9EOj36HSTlAe1QaFPQPjTkcaSpiKFZ0gdbCCls7gOtZ3gs2K/Hc/H7Uyg7k2oDWsuUT0k8AvVC4XKikJL5CTwjCetEerSYCzovki/ltF2wYbXGWwjxI/aZfC4QBtpiMAWul0HrbJGE70hEkIIIUTbowWREEIIIdoehcy2gl7HO3r3Dq+nDV511iv+q+16DCEJek1IS1V4Hx3BWLRh42doD9I8FluXwPy+G/uhzR873C0fCDy3gVDIjLcRtl200XwQmqWxGXWmpm1kzb+TT9hcVKICujFg6rzBvYfhnFY6buf8TxQiepZBanIEaci8f7Avga07go8Xw02hfjveZnD7EXALDWcNe8sU3EgLmwhMCcddCR0b6kflIvD7Ycd+OC5wftzf0LEJtNwJPq7mi3bkwmp6QySEEEKItkcLIiGEEEK0PVoQCSGEEKLtkYZoK7AtAGV6QzA1OKwZGJ8ncBuhOpjQjNYAv9Ds2BEM9b7BaGxjmMvJtyBdCt4X0mUF+YXqr1opu9+CniJ0PgMdHbbfQMIEM8HnEXPRQ8cG2kK3O9S5gJauY2I0HhjDvtM+LX0XoN8wPuNb+P4Z9u89YFS+M7aD3hAJIYQQou3RgkgIIYQQbY8WREIIIYRoe9pbQ+Sihrgo6YUiKLFPJfFxbKCtpfmw9D5sY4jzUdn98Boqvo1rsoS1AEB2kYaINS/UAgDGhtbuQFtYWwBqUxHih20MAttW4DYD22WEj4V6JtgCgOaDi4XmA78I/KLEv8AjGBuDX5IDG9QFy8FYsiVws6Ff0zZyMA73g2wwluqHtWIbbkLrh5FfPYOWM/BwID+yUUuglPzAVqNtQFGtelPbHWqlQzXL0K/u2xy10oGx5EffrdQqC79HRgi9IRJCCCFE26MFkRBCCCHanvYOmTWBoaZaoK3q2xKwYQdqHAuvlGG7SQ38aCx1ww7wi+sQFkAbhBTIL4P3n9Q1m/y4D4ZvawXscRJmo3CTxfD6OAe2BF5RU9ds8oNWEy4Pr97Br3lsWiAfz8RzQXsLHgt+NBZsFL/MIMWewmgUHrMChH3yEB4q+A+HfN6/mQt5368j79+4pZxv6wy2+Q+MMbmK75f4YzuaHjYleKgUwVaKwAZ+eXiIki2BviIxhe+o/0gLpPAOIIPQVRXitTW4qGrgV3b+BV6BG6EfLvAy+A3ADdOXFoP8+uuFpt9hfrJVfVu55tuqdWhPVfU/p7QKz7woTCcQ3K5nGNAbIiGEEEK0PVoQCSGEEKLt0YJICCGEEG2PNERbQ6XJIYRNOiDUC4EtV4axFV/bQBqiXBk0PxWIu6PNj+NHVd8WVxp1AVENhFU1+ABqIHAiv7pvcylsIyWtke/nSGtEkNYIdEARaH5QQ5T4sfMYbJYjG9x2eRDW5H0/V/DncwXygzh+0R+bNdnSImiPQFeUliDFF053WqS6A74JSxvQqQhuvxE6H9x7oAMqFPyD6yyAlgds4wq+vqer4D8IxsHDoQts43MDvi3p9+dLYBtx49gxMWiPIjgueJiVQBtUggcmyM0sDyc8gfssJj+wpYH1NzLwq8GzoQq2Gmy3DBdVP+iA+pyvF+rLfB1QP9g2pp2ebXNW8myb6r5fb7203d/NzHprHZ5tS87f3y1Vf9+2VEAH5VnMHNTaoHT/4W6TtLPoDZEQQggh2p4RXRBt2LDBTjvtNOvq6rIJEybYWWedZVu2bNmu/8c+9jE78MADraOjw97xjnfYxz/+cdu0aVODXxRF3s999903kocihBBCiLcwIxoyO+200+zll1+2Rx55xGq1mi1cuNDOPvtsu/fee9H/D3/4g/3hD3+wG264wQ455BB7/vnn7ZxzzrE//OEP9tWvfrXB984777T58+cP/j5hwoSRPBQhhBBCvIUZsQXRunXrbMWKFfazn/3Mjj76aDMzu+WWW+yEE06wG264waZNm+aNOfTQQ+3//t//O/j7fvvtZ5/97Gft7/7u76xer1tuK+3FhAkTrLu7O2hfKpWKVSpvxMp7e3vZkcLQZCPZSnA7C9ABYf2jwFpCZAPdT4he6M9+TVoJ8GG9kG9zVfAjDRFojUhXhFojN8x13aE2RgTtMiLQCznQBpFfVKCiO3AcGZxH0CKEEgfUU8LWIFBfKYPaUTGU56fDCm79Eno/UnkqbK0C9a6ohQTUNaJ2Fgn45RPQH5ENhIgdUDdobOJrfMaSNihAL/RnW6NfV+yP64T6QmNAL9QJ11MeRF4lsOXMtyVw78XDHMTI4EFdc/6xka1ivq0MfgXYRh788jAftWAJJYMaZc11kqhuUhX0PdUUzk/gPUAtWKjkEOqF3qoaotWrV9uECRMGF0NmZnPnzrU4jm3NmjXB82zatMm6uroaFkNmZuedd55NmjTJZs6caXfccYe57RToW7ZsmY0fP37wZ/r06Tt/QEIIIYR4yzJiC6Kenh6bMmVKgy2Xy9nEiROtp6cnaI5XX33Vrr76ajv77LMb7FdddZV9+ctftkceecROOeUU++hHP2q33HLLNudZunSpbdq0afDnxRdf3PkDEkIIIcRblp0OmV1yySV23XXXbddn3bp1Q96h1+nt7bUTTzzRDjnkEPv0pz/d8G+XXXbZ4P8feeSR1tfXZ9dff719/OMfx7mKxaIVi37KoBBCCCGE2RAWRBdccIGdeeaZ2/XZd999rbu721555ZUGe71etw0bNuxQ+7N582abP3++jRs3zh544AHLU42WrZg1a5ZdffXVVqlUWlv4hMY0qQUL1EchG/WnSvNhIgjUVNDORFCXhrQh0FMrzjfudAT1bCLoc2N10LuQ1gjqC0VUwAY1NdCzijRE4BcM6YUoAA7xeqpNhDasQwTnDGoYYc2hPPn5NqxDVIibfOD6xP5mdB2DHxxqFnivUN2g0HuU7hUH95SDWihpCloM6NlUTfxrj/pC5UB7EYPwKeEb3N8X7LMFPbViqIcTNz4f+xP/eVnK/JpDVJuoBDqoUgT93Yx6nvnz0emm9oChgMwN5Z8pXEA1uFbK+LlDfSGwUX+zzalf/4dqE1HNoS0p1SGCekJNPc8212CuGtRDguu4Wvdv5jrcK2kGfctQdxtYo2wU2ekF0eTJk23y5Mk79Js9e7Zt3LjR1q5da0cddZSZmX3ve9+zLMts1qxZ2xzX29tr8+bNs2KxaF//+tetVPJPYDOPP/647bHHHnoLJIQQQoghMWJZZgcffLDNnz/fFi1aZMuXL7darWaLFy+2BQsWDGaY/f73v7c5c+bYF7/4RZs5c6b19vba8ccfb/39/fZv//Zv1tvbO5gRNnnyZEuSxL7xjW/Y+vXr7d3vfreVSiV75JFH7JprrrFPfvKTI3UoQgghhHiLM6J1iO655x5bvHixzZkzx+I4tlNOOcVuvvnmwX+v1Wr21FNPWX//n8vO/+IXvxjMQNt///0b5nr22Wdt7733tnw+b7feeqt94hOfMOec7b///nbjjTfaokWLWt5fBymElHKcBmY+0yt/Chdg+KHu2+odlGLvzxfX/Q0nVQh9Uep0Uxo/pf/TOLZBeAzCWTSWXp1GoW06WsnExzANhFoohT0wZZ3DlxBuorGUAg/XD1+3FNKKmn73XHB+6E7ANpiP7p/QsRhuo7AKXT9QFsDBvQJuVoWNZGCj0NpAzT+43pwfWtpQ8UMepQTCUjC2E3oHFWFsR9MDg3yKkf9QKUJ4LE82SM8nWwI3KaVrk18oaWDOEIUbM3h4B4cq4WImvwpc4AOpP3YAbpgq3Aj9dQjVpY3boJBuGa5PSruvwHdIHa73tOZ/dlkNwvpwo7VQdWBYGNEF0cSJE7dZhNHMbO+9925Ilz/uuOO2mz5vZjZ//vyGgoxCCCGEEK2iXmZCCCGEaHu0IBJCCCFE2zOiIbPdnsg15ueCtiODlHhq5eDgk4TQOcZIIWN9G2Mh5hroh61FUKez/d+3ZWPNT6BfaIsGINpBiHU4QL1QYDpwaOo46WBoLM4X6BeyDRzXQpp8S35oowvINyGgy4toZ0DbUAddRFoBnQXoECNqbwDHEUMaP47FdiOBbRWa/MiH2kewZC6sxQmBLVMCyw60Aum+CCrRQGPJL4XnbwYXKaWnZ1QGItAP96/Jj0pKUOkJ0vc42A+6V0K1QaTpw++uUURviIQQQgjR9mhBJIQQQoi2RwsiIYQQQrQ97a0haiJUx7DL64uLUUbn+80G6uNAb4dFh4JFSUOHrijaPSF2Bbvmrtj16A2REEIIIdoeLYiEEEII0fYoZLYVlNKLab7kR03nA8dGkKpKYykFN2kpVXfHabMhPmbb6N5NKciB6bbkR+yqVF1KtyUwZRbng7FwUWFKL2wj1K95XygFN4NU3QzTcqncQ2BaLoWuAm3BKb1Y3iLUD2y43Rb8QscOsYSGmXnlN1qZP7jUBoHlN3ZRCQ0isDQG+oWW2ggtlxFYQiOk/Mawl8FoqdQGfd+CX270JAt6QySEEEKItkcLIiGEEEK0PVoQCSGEEKLtkYZoKxxoXhy17shDmfy8LwBIwC+X8/2KhbpnKyS+Xylf82wdOd9WSvz5OnNVfxuxv42OpNr0uz9/Mfbnz4MAohT7Y8mPbDGIEciPSKhPSSAp/I1Amh/yq0EAHG2Zf9tVyObyvi31/QayML/+tODZyvVGvzKMK9f9+ZvHmZlVa2Cr+8dfr/p+KbTGcGCLwGZVEmj4pghad8AlGmxL/FvK4NbYiW2Avo7mq4NerwrPqRTma9oG+oAtgm1GWdg2jfxIL0S2od/KwX/uu5gEoOCXgB6O2jglcC3TWLBlebDlwsbC7e2NhUfFNrbp+6U41rfBYwv3zcFxpSRWa37+Buo3h4LeEAkhhBCi7dGCSAghhBBtjxZEQgghhGh7pCHamsBaQjHohXJF0NAUfKFAJ9jGFCqebWzeFyh05cuebRzYxib+fONzA+AH88WNtjGxP9e4xJ+rFMFxRf4xoNYIdUWgkwBhCK3oIbweDEogwK8Gcewq7E3ZgSYHAvT9DvQ9YOtNS/7YrOjZNme+36Z6p2fbkjaO3VyHcdUOf1zd37fNVX8/+qu+3wCc22oEuiKSE0Bdowhqy5AUASRzqNFBbZB/G6Af3HqWq8C1jDbQJoJfXCU//+CiKthqaZOP/wFEdfigavBB1cGWwli0+cfgHNxpoD8KBvQ9UQRPDPAz0hXl4OsS9EKWBxFNztfSuYI/n8v7fhn4ZQXQOhZpbKNfWvCPtV4iPRLo7Yrg59/ydGoRKjMWUd2yUWydpDdEQgghhGh7tCASQgghRNujkNnWYDl0aqHhu1E7ixy01SjkKCXeDzeNzUOoCsJjXTnfNjHX59nGJ/2ebQLYmsNh4yJ//jGxHysYE8FxwWdXgg+vaP6r3nwUZouHeU2fQYAshVf5dehNXna+reb8z6of4i99zv+c+yGntTPyx1J4rJQNveRBMymEBzO4WerQGqSeQtp9DlqNUHsQCFs4iIVhFm4LrSboIwkOt1HqPNgoTR7DaGUoSQHhsbjin28Mh1WabFWI+1F4rOrP72Cso/AYhNbIz1F4jMJorQDPnyjxr1G05eHrEsJoUQFCZhBGi8hWBFsNQp9QCoPKIJgXsqd7yh9mEIamVhtUEgAfKXQaA+/R0URviIQQQgjR9mhBJIQQQoi2RwsiIYQQQrQ90hBtTWBMk7NDISURUghT0FlUMz9eXYY66YXAlhmbI19TQq0wQkhBx1EDzU8NUuyrBm1FqCUH+BVAk5SHuHaMwi+fBPxSOLkZ2FJoKVADP5CKWNmFpeL3ZX56OqXi92GKvZ8WvxnS87egrXG+AbjuqlAmoAraoBrYQu+BjNJtSdwANkqxx8sibDqDUxZsg1vZYmrRAHqMCFooRPBZUYuLCK6pEKhkgcV0EKC9gVRyStknvVBEWiM4LpfBc4tafNBxABGl04NeiNLpSVeEqfikNaJUfPDDtHvQFbkC+NHY5tYdoPnB1h1wqNimhF6phNoC79HRRG+IhBBCCNH2aEEkhBBCiLZHCyIhhBBCtD3SEG0N6YWgvnhWh3orVWjRAJvISENU9wO2A1DLYkvN15RszPntGDpzvp6nI4HWGtBnoNmvFEP7EahDVAQ/aufBtXB8vVAB/EgHlWCBi6GTwt8IGQTKya/mQFcDtjLUFyo731YhP7ANkP4ItEB9dWit0dSCox/qm5CtDLaBGuiPavCZ1KBNB/hRbf8hSuHMbBs6IHgCkmwF9YWkPyK9EHV8gNYLcAtZUiU//7Oi+kdRzT+Q5ppIEfRZiOswF/hFdRJTgh/2YAmzReTXCqQ1opo7VAMLzi36QZ0tB9od8svQBtcUtNYI8UtJLwTyJmrdAY8ZvLbTQD/UH6EgcPTQGyIhhBBCtD1aEAkhhBCi7VHIbCsiar9bg9eflGIPrQdSeN1dhXTb/pz/jjGBth8xtMIgv1zih5uojUge2o00tyChliQJxC1y1LoEbDHEHtAv8NVpqF8rUEmFVmx1eFdMbS8oPZ3Goh9dj7TdJj8cR/sGYeMMxlLI2ZFfnVoAwP0IfhjqppT4XFjcC8dCWCGDDhfUtgBbgdDxYhuRsNYIoS1Imv2o3QPOFWrD0iUUCiM/sA03gWndXI6Bwm3k14ot7HqESDynyjfZQufC+SnsFTi2lflGE70hEkIIIUTbowWREEIIIdqeEV0QbdiwwU477TTr6uqyCRMm2FlnnWVbtmzZ7pjjjjvOoihq+DnnnHMafF544QU78cQTrbOz06ZMmWIXXnih1aGjshBCCCFECCOqITrttNPs5ZdftkceecRqtZotXLjQzj77bLv33nu3O27RokV21VVXDf7e2flGanmapnbiiSdad3e3/eQnP7GXX37ZTj/9dMvn83bNNde0tL+Y0luDGD7YODbtB0QpNp1CQB3C/62VOg/2CwjktxCHH3ZGI01zmA+kpV0OHRu6z83zBc8/9G3S8bfymYS27qAU+5R0RaH718JnFXy8KGYJGxq0jRau7fBzNsRrcSQYhWdS8Ec63M/RED84aS3N38oxBNvc9n8fRkZsQbRu3TpbsWKF/exnP7Ojjz7azMxuueUWO+GEE+yGG26wadOmbXNsZ2endXd34789/PDD9tvf/ta++93v2tSpU23GjBl29dVX28UXX2yf/vSnrVCAIghCCCGEENthxEJmq1evtgkTJgwuhszM5s6da3Ec25o1a7Y79p577rFJkybZoYceakuXLrX+/v6GeQ877DCbOnXqoG3evHnW29trTzzxBM5XqVSst7e34UcIIYQQ4nVG7A1RT0+PTZkypXFjuZxNnDjRenp6tjnuQx/6kO211142bdo0+9WvfmUXX3yxPfXUU3b//fcPzrv1YsjMBn/f1rzLli2zK6+8spXDEUIIIcRbmJ1eEF1yySV23XXXbddn3bp1Q96hs88+e/D/DzvsMNtzzz1tzpw59swzz9h+++03pDmXLl1qS5YsGfy9t7fXpk+f7jtCzQ8okRNeL4Q0SYG1O4LHBtpQxxAythU9BY0Ft5Y0FoHzIS3oCVqJu+PuBcbTsY4IjR2iXys1VFqqyUK1S6DuFh4Xzhd28YVuw2A+GhvRWNoGXOARzhe4jdD5mvzQh6an+dEGg4FW6ofRdgmqFxcK1Q/jbiPkBzbYBu4fbSOD+aA2GO5f01iaCx8M8N2AY8GG3z9UUyxw7Gh289jpBdEFF1xgZ5555nZ99t13X+vu7rZXXnmlwV6v123Dhg3b1AcRs2bNMjOzp59+2vbbbz/r7u62Rx99tMFn/fr1ZmbbnLdYLFqx6PdxEkIIIYQwG8KCaPLkyTZ58uQd+s2ePds2btxoa9eutaOOOsrMzL73ve9ZlmWDi5wQHn/8cTMz23PPPQfn/exnP2uvvPLKYEjukUcesa6uLjvkkEN28miEEEIIIUZQVH3wwQfb/PnzbdGiRfboo4/aj3/8Y1u8eLEtWLBgMMPs97//vR100EGDb3yeeeYZu/rqq23t2rX23HPP2de//nU7/fTT7b3vfa8dfvjhZmZ2/PHH2yGHHGIf/vCH7Ze//KV95zvfsUsvvdTOO+88vQUSQgghxJAY0TpE99xzjy1evNjmzJljcRzbKaecYjfffPPgv9dqNXvqqacGs8gKhYJ997vftZtuusn6+vps+vTpdsopp9ill146OCZJEvvmN79p5557rs2ePdvGjBljZ5xxRkPdomBc1BA/xZ5DUHMoph5GYItrgTYaG2jD7db9oCvPt2O/KAUfsGHsF+an3kmsIYKxsF1kmDVEjoQRoX2IKGSfgF+oDXt0QQ81uLND/MiH6vfw/GE2R33ByEb7C34uH6gpgZpDLu9fuBHMF+f8h0MCY6m3YCHv33x5mK8APQiLiT+2mPNtJfArxP58paTxAZQDnyI8LPJwg9O+xSA+ycODlTRE1CORoH6IRBZ4g9egCVgKNzjpimogfqvARV+HbZBfFfyq4Feugy31bZUmWxXGVVN/m7V6mC2tkQ0eUtD7kPoSOqr7R/0LR4gRXRBNnDhxu0UY9957b3NbffFNnz7dfvCDH+xw3r322sseeuihYdlHIYQQQgj1MhNCCCFE2zOib4jebESQBoihpmqYLSFbhfzg9TGNLYNfDV7vV+BVfs23xVWyNb7ejmr+626yWR1CD3XwS0Nt8Po8o1zQYc7JpPAY2RL4WyKG3HHyy0FLlxyk0eb829PlYWzBt2V5eOVf8G1psdGWFiBMBbZ60bfR2BRkfRm1HQyLlmDIMKKwHL1lh9R5Co8lBf96zBcgdFXw49+dZMv7N3NXwX8QdOZ8v3G5sj8WbGPhwdIJtnFx49jO2PcZA7YSxPpLkW/LQ9OhAoTMKIyWQCislfR8gsJeVWixVIMLrQax4zLa/G4JZYj19mX+zdGf+WM3ZyXPtqUONrjZepv8+uvgU/NtfTV/Pwaq/jH0V32/agIhwyo8t+Bzx/T8UURviIQQQgjR9mhBJIQQQoi2RwsiIYQQQrQ90hBtTWj7CUrPD0y7R70Q6IpypBcibVAZbBVIc61AimzZt0XVJlsVDoJsNd/mwGY1f5uuDh8e6IocpOw70h+1QASp89QrIUr8+HeUh9uJ/Ap+LD7Kkw3mK/gxe1f0/eKCb8vIryltNiqBHimlGgO+CQksWZDBaaT7LLQtDZY7gDYVMfTmiSF1ntLkOyCdnvRCE4qgA8oP+DbQBu2R7/ds4xN/7PjE9xsHfl1xo61ZU2RmNga0QZ2Qnj8GPuM8nO8iaEXykW+LIU0+wT4lQ6fm/OPIQMBWc/5nUHb+Z1UGDWM/6I/6oNYE6YV6QS+0OevwbBuTTs+2KfX9OprKLGyEL6oYbqrQ0gYpHGsKz4sMtJQptrQJ2uyIoTdEQgghhGh7tCASQgghRNujBZEQQggh2h5piLaG4pehtsD5sA0E1NrAOiqoi6BtkB/UuYHWENZki2Ac1taBkvMR1A0iHRAdKkWwI/icSPND2wgFNUSgA+LaRPAZQC0ho88UP2fYLmqcqMUHFewJaA8SeI0N+70ChN4DZKPyNQ4mJFsoVNMm1EatHDKqfUNtJeC+pRYSNF9zSwqqrVMFTUne+bZyYPEo0qhkoL3Jg16I/EIhbVAK82XwtCmD1qgGfjW4+LiGkW+rwjmj8xhqo/Ndb/IjH7TBcYVe2wSexlDbKKI3REIIIYRoe7QgEkIIIUTbo5DZ1mDoKqyTeARduENf/1GXdHhTju0YEmjHkBV9Wwzp/jF0Pm5u5xFBy4+I2nRQiw9qtUFjwS+i1h303pVsLYTMQkNSHOIKDUsGtu6gNh2wDWrJEdy6I9+4f1ne398U23R4pm20/QA/7FgPNrgHMHwXCKbsQ7se6tZdicMelXQ5UmpyGUoqlOGD6c35adh/qvsp19y6wy8BMDZpTB2n9hud0IeIW3f4ftSSoxD5qd7BrTtCe7oAmfmfZwqhIDo/NWznAecMWnJQKKwf2nSUA1PxqU0H+fVBW47NTbb+OswPbTr6qXVHzd/fShXadEBn+wxsDu69Ye7UstPoDZEQQggh2h4tiIQQQgjR9mhBJIQQQoi2RxqirSB9AoSImcCxGXSpgBC7xSnEV+u+jVqGRFA6PU4hZZ1akKTNPjAOwvq4H6TloWrtlIrfQkpmK3Ho4HIH6Efp72FjSUcGkoVgvRnpbzLQMzVvA+eC/UCNG+0vjW3FLwksUUHXANxTVqU0ZHCDe6qe8z+Ecs6/6TcHtgIhWw5aixRyoMmB1hqFBGxNfjg/zJUDG7V3IL8Ebkgaiy0kWriZQ1PCQ9POqQQCllSgFHu4mOsZ+fnbqKagZwK/GvjVm65bnKtO46DcQ6g2CO4VB99dEWj1SNM3mugNkRBCCCHaHi2IhBBCCNH2aEEkhBBCiLanvTVEkWsQnGDNGIhhp6R3oHIZEF/GWiikq6HQOYwNDrHTvoTodIY6blt+RCu1J0ajbkULrVqI4G4Rwdql4Rvb0vyBeimLA3VAwfP5Jrz2QMdAuiJXo+eAbwIJiGWwYZAp2QDtYGAJLPr8+NrbsR+1wwm9jnEsOobOF+bXCi10AgmvK4ciwaH78c1GGyZb41g8/tB6QIFjWSfawjZGEb0hEkIIIUTbowWREEIIIdqe9g6ZNYEpvZT628pr7EC/0G3ga+bQ1+Ahr9nxNX7o6/7hfaWOQ3en1+ytvFIPnG/YX7NnAa/UA17Fm20rHEyxprBt0Gv24NAs7R+UmeDQdODr/UC/0DB5+LEN3c+7vUejvMVohM5DaeV50UpIfBeEv4mWQtOBYXJsgYV+gfPRd+YIoTdEQgghhGh7tCASQgghRNujBZEQQggh2h5piLYC45c5iIfmfVFABH4RlOyPc74tAVsOyu5TaX8qz5+jVgE0X0C5fy7jT+X+/XYCucBS/LQfoaX9ieEu9x9a2j+lMv5U7h/Gkh+V8R/ucv/VeuMjgOan0v5kS6Fkf2i5f05/Bz9Knae2ANSWBtvmBLbDofY6LfjBJb+NFj6hfkNszUNtcwLb8ITrpYbemmfY2/AQgRodapvDOhh4hmAbHrBB6ZeWWvM0+QW3zQlt4YPzhR0DtsWi79tmoWjwid159IZICCGEEG2PFkRCCCGEaHu0IBJCCCFE2yMN0dZQSwHSC4EtV/QD7/m8H8QvFny/jnzNs3Xmq55tLNg6c75tXK7i2caArTOGbeTKO/QpRbC/sT9/AUQMpdgfmwfRQt78sQkIFBLSGmGhmzAy+BshBaFADYL4VQiyk1/ZFXwbBNT7M/Bzvt+WtOSPTf2xW9KiZ+urN9r66zAObH013zZQ9fetXPNtlZr/OdUroLUiqQAJL0B/hPoW8IPL0RL/kje4vC2BsXC7WFL1d4a2kVTgWq7BWLDFVdDr1cDW5Bel8Hyr+fdeVId7KgU/0DIZbAN7HVERLNAuBYOan0Bb4l9nLoZrLxfm5/L+9e1obB40dwWwwdisALrGprHwWLAUx4EOiMaCDoj8zH/0INQOZzS7eegNkRBCCCHaHi2IhBBCCNH2aEEkhBBCiLZnRDVEGzZssI997GP2jW98w+I4tlNOOcX+6Z/+ycaOHYv+zz33nO2zzz74b1/+8pftAx/4gJmZRRDz/fd//3dbsGBBaztM4WXQFVEtoRzYSC80tuiLEcYVfFtX3reNzw+AX9mz7ZHr87eb+H4Tkn5/X+KBpt/9cWNAUDEGdUWgIaI6RPDBlyI/5h6DXwLXQtzCOj8z/zhqzj+3KUS2K+BXBl1EGXQwfaANIg1RH+iPNqcdnm1j2un7Zb7W6E+1MQ2/99ZhrppvKyX+tb0x8ucP7TXnQDCUQV0jkhAFl5sB2Qr2NyNdEdlIVwQ6oFwZdEDgl5RBIwfaoGQAtDtVqOVV8Xc6qjadtxoUHarCwdbAVve36er+fI7Gkl4INEmuBQ1RhBoiuIDAL8rBV2PiP5OivO8X5UFYA34Gfq7o+7mCb8uK/ljSGsWlxn1Oir5PHWxU/6qOoj6AajMF1jXiGkOjpyIa0QXRaaedZi+//LI98sgjVqvVbOHChXb22Wfbvffei/7Tp0+3l19+ucH2L//yL3b99dfb+973vgb7nXfeafPnzx/8fcKECcO+/0IIIYRoD0ZsQbRu3TpbsWKF/exnP7Ojjz7azMxuueUWO+GEE+yGG26wadOmeWOSJLHu7u4G2wMPPGB/+7d/671VmjBhgucrhBBCCDEURmxBtHr1apswYcLgYsjMbO7cuRbHsa1Zs8b++3//7zucY+3atfb444/brbfe6v3beeedZx/5yEds3333tXPOOccWLlyIoTQzs0qlYpXKG++3e3t7eYM0PDBLk6AWEmgLbFNBthy88ye/PPiF2Ch1Po/zgw2OKzQ8lodX2xQKy8PYVqD2GzHsS835n0sW2GeAwm0FmK8G5QgSeM8cer6pREFz2xS8xqi1Sui13ULvhQjGOrTB4FFoDRG6DdwutrgAP0pjp1YYQ01jhzAVpsSDjcJjOB+k3Tvyg5BzKyEzuKUsorAXXEC0Va4CAeE2Ss8nG7RY4rIFYec7gniyd52F3gPoR9diaEy8BdsoMmILop6eHpsyZUrjxnI5mzhxovX09ATNcfvtt9vBBx9sxxxzTIP9qquusr/6q7+yzs5Oe/jhh+2jH/2obdmyxT7+8Y/jPMuWLbMrr7xyaAcihBBCiLc8O60+veSSSyyKou3+PPnkky3v2MDAgN1777121llnef922WWX2V/8xV/YkUceaRdffLFddNFFdv31129zrqVLl9qmTZsGf1588cWW908IIYQQbx12+g3RBRdcYGeeeeZ2ffbdd1/r7u62V155pcFer9dtw4YNQdqfr371q9bf32+nn376Dn1nzZplV199tVUqFSsW/ZKYxWIR7UIIIYQQZkNYEE2ePNkmT568Q7/Zs2fbxo0bbe3atXbUUUeZmdn3vvc9y7LMZs2atcPxt99+u/3N3/xN0LYef/xx22OPPVpf9FD8EtOBodQ5pAhTiwJqP0FkELHOIEZcyfxtDEA99S2JnxK9KeenUzen2VPrDkq7p5YcpYjafkCbDrT5n1PBfL+YcqRbIIOYeHjrDj8lvmbUugNaXEDrjr7Mv577wUbp9P3QpoNad/TWG8dursNc0Lqjtwr7Ru084B6o1aFNB6XYQx3/CO5H1EBQ6i+1BaB0YLBlJFWjFgWhmgpwi6ldRA5KTUB7h7gAZSqgbUpzen5U889jBNogat0RQdo96YWiLMwPNU9kawUSgKK+h2xwEYDN5cAP23TAWLBlcG4pxT6lth9NKfVpHp5vJWrd4Zm4nQe17oBVBd1ToZrd0WTENEQHH3ywzZ8/3xYtWmTLly+3Wq1mixcvtgULFgxmmP3+97+3OXPm2Be/+EWbOXPm4Ninn37afvjDH9pDDz3kzfuNb3zD1q9fb+9+97utVCrZI488Ytdcc4198pOfHKlDEUIIIcRbnBGtQ3TPPffY4sWLbc6cOYOFGW+++ebBf6/VavbUU09Zf39jgcA77rjD3v72t9vxxx/vzZnP5+3WW2+1T3ziE+acs/33399uvPFGW7Ro0UgeihBCCCHewozogmjixInbLMJoZrb33nubg1ei11xzjV1zzTU4Zv78+Q0FGYUQQgghWmVEF0RvOkjeU6caFX5AtBbYeoD0E/1V0GjkfC3Ha4kfiy/k/Hh/EdoqUKuFEowtxPWm3/1tFmN/HNWqIT+qj0O1ahI4Gc01c7YF1XUiSKdFpCA+Ia0R6YoySOSspKCrCdSH1UHMUgbNWBX9/Pkq9UYb7Vs1hesdruMq6IVIW5fW4POE+VwN9EKg36PTTVIe1AaFPgFBUkJjSVORwjOkDjaQ0llch9pO8Fmx347n43YmUPcm1IY1l6geEviF6oVCZUWhJXISeMaT1gh1aTAWdF+kX8tou2DD6wy2EeJH7TJ4XKCNNERgC92ug1ZZo4mauwohhBCi7dGCSAghhBBtj0JmW0Gv4x29e4fX0wavOusV/9V2PYaQBL0mpKUqvI+OYCzasPEztAdpHoutS2B+34390OaPHe6WDwSe20AoZMbbCNsu2mg+CM3S2Iw6U9M2subfySdsLipRAd0YMHXe4N7DcE4rHbdz/icKET3LIDU5gjRk3j/Yl8DWHcHHi+GmUL8dbzO4/Qi4hYazhr1lCm6khU0EpoTjroSODfWjchH4/bBjPxwXOD/ub+jYBFruBB9X80U7cmE1vSESQgghRNujBZEQQggh2h4tiIQQQgjR9khDtBXYFoAyvSGYGhzWDIzPE7iNUB1MaEZrgF9oduwIhnrfYDS2Mczl5FuQLgXvC+mygvxC9VetlN1vQU8ROp+Bjg7bbyBhgpng84i56KFjA22h2x3qXEBL1zExGg+MYd9pn5a+C9BvGJ/xLXz/DPv3HjAq3xnbQW+IhBBCCNH2aEEkhBBCiLZHCyIhhBBCtD3trSFyUUNclPRCEZTYp5L4ODbQ1tJ8WHoftjHE+ajsfngNFd/GNVnCWgAgu0hDxJoXagEAY0Nrd6AtrC0AtakI8cM2BoFtK3Cbge0ywsdCPRNsAUDzwcVC84FfBH5R4l/gEYyNwS/JgQ3qguVgLNkSuNnQr2kbORiH+0E2GEv1w1qxDTeh9cPIr55Byxl4OJAf2aglUEp+YKvRNqCoVr2p7Q610qGaZehX922OWunAWPKj71ZqlYXfIyOE3hAJIYQQou3RgkgIIYQQbU97h8yawFBTLdBW9W0J2LADNY6FV8qw3aQGfjSWumEH+MV1CAugDUIK5JfB+0/qmk1+3AfDt7UC9jgJs1G4yWJ4fZwDWwKvqKlrNvlBqwmXh1fv4Nc8Ni2Qj2fiuaC9BY8FPxoLNopfZpBiT2E0Co9ZAcI+eQgPFfyHQz7v38yFvO/Xkfdv3FLOt3UG2/wHxphcxfdL/LEdTQ+bEjxUimArRWADvzw8RMmWQF+RmMJ31H+kBVJ4B5BB6KoK8doaXFQ18Cs7/wKvwI3QDxd4GfwG4IbpS4tBfv31QtPvMD/Zqr6tXPNt1Tq0p6r6n1NahWdeFKYTCG7XMwzoDZEQQggh2h4tiIQQQgjR9mhBJIQQQoi2RxqiraHS5BDCJh0Q6oXAlivD2IqvbSANUa4Mmp8KxN3R5sfxo6pviyuNuoCoBsKqGnwANRA4kV/dt7kUtpGS1sj3c6Q1IkhrBDqgCDQ/qCFK/Nh5DDbLkQ1uuzwIa/K+nyv487kC+UEcv+iPzZpsaRG0R6ArSkuQ4gunOy1S3QHfhKUN6FQEt98InQ/uPdABFQr+wXUWQMsDtnEFX9/TVfAfBOPg4dAFtvG5Ad+W9PvzJbCNuHHsmBi0RxEcFzzMSqANKsEDE+RmlocTnsB9FpMf2NLA+hsZ+NXg2VAFWw22W4aLqh90QH3O1wv1Zb4OqB9sG9NOz7Y5K3m2TXXfr7de2u7vZma9tQ7PtiXn7++Wqr9vWyqgg/IsZg5qbVC6/3C3SdpZ9IZICCGEEG2PFkRCCCGEaHu0IBJCCCFE2yMN0dZQGJpsJFsJbmcBOiCsfxRYS4hsoPsJ0Qv92a9JKwE+rBfyba4KfqQhAq0R6YpQa+SGua471MaIoF1GBHohB9og8osKVHQHjiOD8whahFDigHpK2BoE6itlUDsqhvL8dFjBrV9C70cqT4WtVaDeFbWQgLpG1M4iAb98AvojsoEQsQPqBo1NfI3PWNIGBeiF/mxr9OuK/XGdUF9oDOiFOuF6yoPIqwS2nPm2BO69eJj/Zs/gQV1z/rGRrWK+rQx+BdhGHvzyMB+1YAklgxplzXWSqG5SFfQ91RTOT+A9QC1YqOQQ6oWkIRJCCCGE2LVoQSSEEEKItkchs60JfYUX2l0cu3VDymg+7J0/N4OmXGJIw6ZQCLSQiPONOx1B+nYEZd2tDuEdCq1BOn1E+doYQoIWDRQyA79gKDxG73vh9TSl4qMN0+7hnEHKPqbY58nPt2HafSFu8oHrE9t50HUMfnCoWeC9Qmnyofco3SsO7ikX2Om7Bi0KqtBNntogUKf4GOJ8SWC3dwp7YAuJGNK/48bU6f7ET6UuZX6KPaXilyDsV4qgnYlRiw9/Pjrd1A0nFIjqotohhQuoBtdKGT93SKcHG7Xz2Jz66e6Uik8p9ltSSruH9PmmFh+bazBXDdL/4Tqu1v2buQ73SppBmw6UmQSW5BhF9IZICCGEEG2PFkRCCCGEaHu0IBJCCCFE2yMN0VY4SCGklOM0MPOZNBCkn0A9Rt231Tsoxd6fL677G06qoAWi1OmmNH5K/6dxbAO9EOh7aCzFkqPQNh2tZOKjbgW0J5TCHpiyznou0N/QWEqBh+uHr1vS+ERNv3suOD90J2AbzEf3T+hY1B+RzoSuHygL4OBeATerwkYysJHWaKDmH1xvztfabKj4GpBSAjodGNsJvYOKMLaj6YFBPsXIf6gUQS+UJxuk55MtgZuU0rXJL5Q08O990l9l8PAO1m7BxUx+FbjAB1J/7ADcMFW4EfrroF1KG7dBGrcyXJ+Udl+B75A6XO9pzf/sshroHOFGa6HqwLCgN0RCCCGEaHu0IBJCCCFE26MFkRBCCCHanvbWEEWusWAJaDsyqBFErRwcfJIQOscYKZTw2cZYiLkG+mFrEdTpbP/3bdlY8xPoF9qiAYjcyBeuQL1QYH2U0Fo6pIOhsThfoF/INnBcC3WDWvJDG11AvgkBXV5EOwPahjroItIK6CxAhxhRewM4jhjqGuFYbDcS2FahyY98qH0ES+bCWpwQ2DIlsA5TK5Dui6CaVTSW/FJ4/mZwkVK9nozqYgX64f41+VGNLarFRfoeB/tB90qoNog0ffjdNYroDZEQQggh2p4RWxB99rOftWOOOcY6OzttwoQJQWOcc3b55ZfbnnvuaR0dHTZ37lz73e9+1+CzYcMGO+2006yrq8smTJhgZ511lm3ZsmUEjkAIIYQQ7cKILYiq1ap94AMfsHPPPTd4zOc+9zm7+eabbfny5bZmzRobM2aMzZs3z8rlNzoyn3baafbEE0/YI488Yt/85jfthz/8oZ199tkjcQhCCCGEaBMi50ZWhHHXXXfZ+eefbxs3btyun3POpk2bZhdccIF98pOfNDOzTZs22dSpU+2uu+6yBQsW2Lp16+yQQw6xn/3sZ3b00UebmdmKFSvshBNOsJdeesmmTZuGc1cqFatUKoO/9/b22vTp022v6z5jccnv7dK4Y7s2pimE2AGjoD0RQuweZOWyPX/xpbZp0ybr6uoa1rl3Gw3Rs88+az09PTZ37txB2/jx423WrFm2evVqMzNbvXq1TZgwYXAxZGY2d+5ci+PY1qxZs825ly1bZuPHjx/8mT59+sgdiBBCCCHedOw2WWY9PT1mZjZ16tQG+9SpUwf/raenx6ZMmdLw77lcziZOnDjoQyxdutSWLFky+PumTZvsHe94h2VbheK2id4QCbF7ozdEQrQNr39vj0Rwa6cWRJdccoldd9112/VZt26dHXTQQS3t1HBTLBatWCwO/t7b22tmZi9e8ZldtUtCCCGEGCKbN2+28ePHD+ucO7UguuCCC+zMM8/crs++++47pB3p7u42M7P169fbnnvuOWhfv369zZgxY9DnlVdeaRhXr9dtw4YNg+NDmDZtmr344os2btw427x5s02fPt1efPHFYY9H7q68rqHSMb/1acfj1jG3xzGbtedxt+Mxm71x3C+88IJFUbRNzXAr7NSCaPLkyTZ58uRh3wkzs3322ce6u7tt5cqVgwug3t5eW7NmzWCm2uzZs23jxo22du1aO+qoo8zM7Hvf+55lWWazZs0K3lYcx/b2t7/dzMyi/7/iWFdXV1tdXGY65naiHY9bx9w+tONxt+Mxm/1ZWzxSxz1iouoXXnjBHn/8cXvhhRcsTVN7/PHH7fHHH2+oGXTQQQfZAw88YGZ/Xpicf/759pnPfMa+/vWv269//Ws7/fTTbdq0aXbyySebmdnBBx9s8+fPt0WLFtmjjz5qP/7xj23x4sW2YMGCEVktCiGEEKI9GDFR9eWXX25333334O9HHnmkmZl9//vft+OOO87MzJ566inbtGnToM9FF11kfX19dvbZZ9vGjRvtPe95j61YscJKW6XG33PPPbZ48WKbM2eOxXFsp5xyit18880jdRhCCCGEaANGbEF011132V133bVdn2aVeBRFdtVVV9lVV121zTETJ060e++9dzh20cz+LLi+4oorGkTXb3V0zO1DOx63jrl9aMfjbsdjNhud4x7xwoxCCCGEELs7u01hRiGEEEKIXYUWREIIIYRoe7QgEkIIIUTbowWREEIIIdoeLYiEEEII0fa85RdEn/3sZ+2YY46xzs5OmzBhQtAY55xdfvnltueee1pHR4fNnTvXfve73zX4bNiwwU477TTr6uqyCRMm2FlnndVQdHJXsrP79txzz1kURfjzla98ZdCP/v2+++4bjUMKYijn5LjjjvOO6ZxzzmnweeGFF+zEE0+0zs5OmzJlil144YVWr9dH8lCC2dlj3rBhg33sYx+zAw880Do6Ouwd73iHffzjH2+oB2a2+53rW2+91fbee28rlUo2a9Yse/TRR7fr/5WvfMUOOuggK5VKdthhh9lDDz3U8O8h9/iuZmeO+bbbbrO//Mu/tD322MP22GMPmzt3rud/5plneud0/vz5I30YO8XOHPNdd93lHc/WNevM3hzn2WznjpueWVEU2Yknnjjos7uf6x/+8If213/91zZt2jSLosgefPDBHY5ZtWqVvetd77JisWj7778/lvXZ2eeEh3uLc/nll7sbb7zRLVmyxI0fPz5ozLXXXuvGjx/vHnzwQffLX/7S/c3f/I3bZ5993MDAwKDP/Pnz3RFHHOF++tOfuv/3//6f23///d0HP/jBETqKnWNn961er7uXX3654efKK690Y8eOdZs3bx70MzN35513Nvht/ZnsaoZyTo499li3aNGihmPatGnT4L/X63V36KGHurlz57rHHnvMPfTQQ27SpElu6dKlI304QezsMf/6179273//+93Xv/519/TTT7uVK1e6Aw44wJ1yyikNfrvTub7vvvtcoVBwd9xxh3viiSfcokWL3IQJE9z69evR/8c//rFLksR97nOfc7/97W/dpZde6vL5vPv1r3896BNyj+9KdvaYP/ShD7lbb73VPfbYY27dunXuzDPPdOPHj3cvvfTSoM8ZZ5zh5s+f33BON2zYMFqHtEN29pjvvPNO19XV1XA8PT09DT67+3l2bueP+7XXXms45t/85jcuSRJ35513Dvrs7uf6oYcecv/4j//o7r//fmdm7oEHHtiu/3/8x3+4zs5Ot2TJEvfb3/7W3XLLLS5JErdixYpBn539HIm3/ILode68886gBVGWZa67u9tdf/31g7aNGze6YrHo/v3f/90559xvf/tbZ2buZz/72aDPt7/9bRdFkfv9738/7Pu+MwzXvs2YMcP9/d//fYMt5MLdVQz1uI899lj3D//wD9v894ceesjFcdzwoP385z/vurq6XKVSGZZ9HyrDda6//OUvu0Kh4Gq12qBtdzrXM2fOdOedd97g72maumnTprlly5ah/9/+7d+6E088scE2a9Ys9z//5/90zoXd47uanT3mZur1uhs3bpy7++67B21nnHGGO+mkk4Z7V4eNnT3mHT3T3wzn2bnWz/X//t//240bN85t2bJl0La7n+utCXnWXHTRRe6d73xng+3UU0918+bNG/y91c/ROefe8iGzneXZZ5+1np4emzt37qBt/PjxNmvWLFu9erWZma1evdomTJhgRx999KDP3LlzLY5jW7Nmzajv89YMx76tXbvWHn/8cTvrrLO8fzvvvPNs0qRJNnPmTLvjjju8auO7ilaO+5577rFJkybZoYceakuXLrX+/v6GeQ877DCbOnXqoG3evHnW29trTzzxxPAfyE4wXNfhpk2brKury3K5xsL1u8O5rlartnbt2ob7MY5jmzt37uD92Mzq1asb/M3+fM5e9w+5x3clQznmZvr7+61Wq9nEiRMb7KtWrbIpU6bYgQceaOeee6699tprw7rvQ2Wox7xlyxbba6+9bPr06XbSSSc13JO7+3k2G55zffvtt9uCBQtszJgxDfbd9VwPhR3d08PxOZqNYOuONys9PT1mZg1fgK///vq/9fT02JQpUxr+PZfL2cSJEwd9dhXDsW+33367HXzwwXbMMcc02K+66ir7q7/6K+vs7LSHH37YPvrRj9qWLVvs4x//+LDt/1AZ6nF/6EMfsr322sumTZtmv/rVr+ziiy+2p556yu6///7BeelaeP3fdiXDca5fffVVu/rqq+3ss89usO8u5/rVV1+1NE3xHDz55JM4ZlvnbOv793Xbtnx2JUM55mYuvvhimzZtWsMXxPz58+3973+/7bPPPvbMM8/Ypz71KXvf+95nq1evtiRJhvUYdpahHPOBBx5od9xxhx1++OG2adMmu+GGG+yYY46xJ554wt7+9rfv9ufZrPVz/eijj9pvfvMbu/322xvsu/O5Hgrbuqd7e3ttYGDA/vSnP7V8z5i9SRdEl1xyiV133XXb9Vm3bp0ddNBBo7RHI0/oMbfKwMCA3XvvvXbZZZd5/7a17cgjj7S+vj67/vrrR/RLcqSPe+uFwGGHHWZ77rmnzZkzx5555hnbb7/9hjxvK4zWue7t7bUTTzzRDjnkEPv0pz/d8G+74lyL4eHaa6+1++67z1atWtUgMl6wYMHg/x922GF2+OGH23777WerVq2yOXPm7IpdbYnZs2fb7NmzB38/5phj7OCDD7YvfOELdvXVV+/CPRs9br/9djvssMNs5syZDfa32rkeLd6UC6ILLrjAzjzzzO367LvvvkOau7u728zM1q9fb3vuueegff369TZjxoxBn1deeaVhXL1etw0bNgyOH25Cj7nVffvqV79q/f39dvrpp+/Qd9asWXb11VdbpVIZsYZ7o3XcrzNr1iwzM3v66adtv/32s+7ubi9TYf369WZmb+pzvXnzZps/f76NGzfOHnjgAcvn89v1H41zTUyaNMmSJBn8zF9n/fr12zzG7u7u7fqH3OO7kqEc8+vccMMNdu2119p3v/tdO/zww7fru++++9qkSZPs6aef3uVfkq0c8+vk83k78sgj7emnnzaz3f88m7V23H19fXbfffdttxn66+xO53oobOue7urqso6ODkuSpOXrx8ze+llmr7Ozouobbrhh0LZp0yYUVf/85z8f9PnOd76zW4mqh7pvxx57rJdxtC0+85nPuD322GPI+zqcDNc5+dGPfuTMzP3yl790zr0hqt46U+ELX/iC6+rqcuVyefgOYAgM9Zg3bdrk3v3ud7tjjz3W9fX1BW1rV57rmTNnusWLFw/+nqape9vb3rZdUfV/+2//rcE2e/ZsT1S9vXt8V7Ozx+ycc9ddd53r6upyq1evDtrGiy++6KIocl/72tda3t/hYCjHvDX1et0deOCB7hOf+IRz7s1xnp0b+nHfeeedrlgsuldffXWH29jdzvXWWKCo+tBDD22wffCDH/RE1a1cP861QZbZ888/7x577LHBNPLHHnvMPfbYYw3p5AceeKC7//77B3+/9tpr3YQJE9zXvvY196tf/cqddNJJmHZ/5JFHujVr1rgf/ehH7oADDtit0u63t28vvfSSO/DAA92aNWsaxv3ud79zURS5b3/7296cX//6191tt93mfv3rX7vf/e537p//+Z9dZ2enu/zyy0f8eELZ2eN++umn3VVXXeV+/vOfu2effdZ97Wtfc/vuu69773vfOzjm9bT7448/3j3++ONuxYoVbvLkybtV2v3OHPOmTZvcrFmz3GGHHeaefvrphrTcer3unNv9zvV9993nisWiu+uuu9xvf/tbd/bZZ7sJEyYMZv59+MMfdpdccsmg/49//GOXy+XcDTfc4NatW+euuOIKTLvf0T2+K9nZY7722mtdoVBwX/3qVxvO6evPuc2bN7tPfvKTbvXq1e7ZZ5913/3ud9273vUud8ABB+zyhf3r7OwxX3nlle473/mOe+aZZ9zatWvdggULXKlUck888cSgz+5+np3b+eN+nfe85z3u1FNP9exvhnO9efPmwe9iM3M33nije+yxx9zzzz/vnHPukksucR/+8IcH/V9Pu7/wwgvdunXr3K233opp99v7HEN4yy+IzjjjDGdm3s/3v//9QR/7/2uuvE6WZe6yyy5zU6dOdcVi0c2ZM8c99dRTDfO+9tpr7oMf/KAbO3as6+rqcgsXLmxYZO1KdrRvzz77rPcZOOfc0qVL3fTp012apt6c3/72t92MGTPc2LFj3ZgxY9wRRxzhli9fjr67ip097hdeeMG9973vdRMnTnTFYtHtv//+7sILL2yoQ+Scc88995x73/ve5zo6OtykSZPcBRdc0JCivivZ2WP+/ve/j/eDmblnn33WObd7nutbbrnFveMd73CFQsHNnDnT/fSnPx38t2OPPdadccYZDf5f/vKX3X/+z//ZFQoF9853vtN961vfavj3kHt8V7Mzx7zXXnvhOb3iiiucc8719/e7448/3k2ePNnl83m31157uUWLFu3Ul8VosDPHfP755w/6Tp061Z1wwgnuF7/4RcN8b4bz7NzOX99PPvmkMzP38MMPe3O9Gc71tp5Drx/nGWec4Y499lhvzIwZM1yhUHD77rtvw3f262zvcwwhcm43yZsWQgghhNhFqA6REEIIIdoeLYiEEEII0fZoQSSEEEKItkcLIiGEEEK0PVoQCSGEEKLt0YJICCGEEG2PFkRCCCGEaHu0IBJCCCFE26MFkRBCCCHaHi2IhBBCCNH2aEEkhBBCiLbn/wOnhcvDPB6BHgAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X, Y = np.meshgrid(np.linspace(-1, 1, 100), np.linspace(-1, 1, 100))\n",
    "XY = np.concatenate([X.flatten()[:, None], Y.flatten()[:, None]], axis=1)\n",
    "plt.pcolormesh(X, Y, u(XY).numpy().reshape(X.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "class UniformSine(tf.keras.initializers.Initializer):\n",
    "    def __init__(self, c=6.0, seed=42):\n",
    "        self.c = c\n",
    "        self.gen = tf.random.Generator.from_seed(seed)\n",
    "\n",
    "    def __call__(self, shape, dtype=None):\n",
    "        n_units = shape[0]\n",
    "        return self.gen.uniform(shape, minval=-tf.math.sqrt(self.c / n_units), maxval=tf.math.sqrt(self.c / n_units),  dtype=dtype)\n",
    "\n",
    "    def get_config(self):\n",
    "        return {'c': self.c}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "backbone = create_dense_model(layers= [tf.keras.layers.Normalization(mean=mean, variance=std**2)] + [512]*3, activation=tf.sin, \\\n",
    "    initializer=UniformSine(c=6.0, seed=10), n_inputs=2, n_outputs=1)\n",
    "    # initializer=tf.keras.initializers.HeNormal(42), n_inputs=2, n_outputs=1)\n",
    "backbone.layers[2].weights[0].assign(backbone.layers[2].weights[0] * 30)\n",
    "pinn = Helmholtz2DPinn(backbone)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "scheduler = tf.keras.optimizers.schedules.ExponentialDecay(1e-3, 1000, 0.95)\n",
    "optimizer = tf.keras.optimizers.Adam(scheduler)\n",
    "pinn.compile(optimizer=optimizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-01-09 01:12:12.375856: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'data' with dtype float and shape [?,2]\n",
      "\t [[{{node data}}]]\n",
      "2024-01-09 01:12:12.764194: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'data' with dtype float and shape [?,2]\n",
      "\t [[{{node data}}]]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 000000, Lr: 191928736.00000, Lb: 0.71259, MAE: 0.80878, loss_total: 191928736.00000\n",
      "Epoch: 001000, Lr: 17647.08594, Lb: 0.12915, MAE: 0.48212, loss_total: 17647.21484\n",
      "Epoch: 002000, Lr: 25639.36719, Lb: 0.08984, MAE: 0.45893, loss_total: 25639.45703\n",
      "Epoch: 003000, Lr: 13745.17578, Lb: 0.06141, MAE: 0.44142, loss_total: 13745.23730\n",
      "Epoch: 004000, Lr: 12630.12891, Lb: 0.04325, MAE: 0.42980, loss_total: 12630.17188\n",
      "Epoch: 005000, Lr: 8884.65723, Lb: 0.03119, MAE: 0.42281, loss_total: 8884.68848\n",
      "Epoch: 006000, Lr: 4560.90674, Lb: 0.02280, MAE: 0.41752, loss_total: 4560.92969\n",
      "Epoch: 007000, Lr: 2487.73120, Lb: 0.01696, MAE: 0.41391, loss_total: 2487.74805\n",
      "Epoch: 008000, Lr: 2145.19214, Lb: 0.01275, MAE: 0.41139, loss_total: 2145.20483\n",
      "Epoch: 009000, Lr: 1604.16479, Lb: 0.00970, MAE: 0.40932, loss_total: 1604.17444\n",
      "Epoch: 010000, Lr: 1301.85620, Lb: 0.00735, MAE: 0.40778, loss_total: 1301.86353\n"
     ]
    }
   ],
   "source": [
    "callbacks = [PrintLossCallback(1000)]\n",
    "history = pinn.fit(inputs, outputs, epochs=50000, batch_size=1000, verbose=0, callbacks=callbacks)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10.6 ('tf2-pinn')",
   "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": "3.9.16"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "e186c38e7d8b9112f5aec0d01a09336163341258afaa29e2c2eecdf076e8d0c2"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}