LADMM_TestCSI.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "CMzd-RxcPG7z"
   },
   "source": [
    "# Import libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "executionInfo": {
     "elapsed": 26659,
     "status": "ok",
     "timestamp": 1614104821257,
     "user": {
      "displayName": "Juan Ramírez",
      "photoUrl": "",
      "userId": "10349054824212831472"
     },
     "user_tz": -60
    },
    "id": "vHzrw1PePx5V"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import numpy.matlib\n",
    "import matplotlib.pyplot as plt\n",
    "import scipy.io as sio\n",
    "\n",
    "from models.LadmmNet import LADMMcsimageNet\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "\n",
    "gpu_list = '0'\n",
    "os.environ[\"CUDA_DEVICE_ORDER\"] = \"PCI_BUS_ID\"\n",
    "os.environ[\"CUDA_VISIBLE_DEVICES\"] = gpu_list\n",
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ipwDM8CZuH9i"
   },
   "source": [
    "# Utility functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "executionInfo": {
     "elapsed": 26663,
     "status": "ok",
     "timestamp": 1614104821263,
     "user": {
      "displayName": "Juan Ramírez",
      "photoUrl": "",
      "userId": "10349054824212831472"
     },
     "user_tz": -60
    },
    "id": "S-gREclZuW8v"
   },
   "outputs": [],
   "source": [
    "def imread_CS_py(Iorg):\n",
    "    block_size = 33\n",
    "    [row, col] = Iorg.shape\n",
    "    row_pad = block_size-np.mod(row,block_size)\n",
    "    col_pad = block_size-np.mod(col,block_size)\n",
    "    Ipad = np.concatenate((Iorg, np.zeros([row, col_pad])), axis=1)\n",
    "    Ipad = np.concatenate((Ipad, np.zeros([row_pad, col+col_pad])), axis=0)\n",
    "    [row_new, col_new] = Ipad.shape\n",
    "\n",
    "    return [Iorg, row, col, Ipad, row_new, col_new]\n",
    "\n",
    "def img2col_py(Ipad, block_size):\n",
    "    [row, col] = Ipad.shape\n",
    "    row_block = row/block_size\n",
    "    col_block = col/block_size\n",
    "    block_num = int(row_block*col_block)\n",
    "    img_col = np.zeros([block_size**2, block_num])\n",
    "    count = 0\n",
    "    for x in range(0, row-block_size+1, block_size):\n",
    "        for y in range(0, col-block_size+1, block_size):\n",
    "            img_col[:, count] = Ipad[x:x+block_size, y:y+block_size].reshape([-1])\n",
    "            count = count + 1\n",
    "    return img_col\n",
    "\n",
    "def col2im_CS_py(X_col, row, col, row_new, col_new):\n",
    "    block_size = 33\n",
    "    X0_rec = np.zeros([row_new, col_new])\n",
    "    count = 0\n",
    "    for x in range(0, row_new-block_size+1, block_size):\n",
    "        for y in range(0, col_new-block_size+1, block_size):\n",
    "            X0_rec[x:x+block_size, y:y+block_size] = X_col[:, count].reshape([block_size, block_size])\n",
    "            count = count + 1\n",
    "    X_rec = X0_rec[:row, :col]\n",
    "    return X_rec"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "rRQCOuAhSFkJ"
   },
   "source": [
    "# Measurement matrices"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "executionInfo": {
     "elapsed": 38687,
     "status": "ok",
     "timestamp": 1614104833288,
     "user": {
      "displayName": "Juan Ramírez",
      "photoUrl": "",
      "userId": "10349054824212831472"
     },
     "user_tz": -60
    },
    "id": "FJSePPPcSSjK"
   },
   "outputs": [],
   "source": [
    "fname1       = 'H.mat'\n",
    "data_path    = os.path.join(os.getcwd(),'data/GrayscaleImages/Clean_measurements_25')\n",
    "H            = sio.loadmat(os.path.join(data_path, fname1))['H']\n",
    "H            = torch.from_numpy(H).type(torch.FloatTensor)\n",
    "H            = H.to(device)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "QX09iovOblld"
   },
   "source": [
    "# LADMM parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 39696,
     "status": "ok",
     "timestamp": 1614104834303,
     "user": {
      "displayName": "Juan Ramírez",
      "photoUrl": "",
      "userId": "10349054824212831472"
     },
     "user_tz": -60
    },
    "id": "ivEsq0T4b0_l",
    "outputId": "bf83b91b-e7cd-444e-c2bc-29e5782263a6"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "M = 256\n",
    "N = 256\n",
    "L = 1\n",
    "\n",
    "layer_num               = 16 \n",
    "learning_rate           = 0.001\n",
    "epochs                  = 512\n",
    "num_training_samples    = 80\n",
    "num_samples             = 80\n",
    "batch_size              = 1\n",
    "num_processed_images    = 1\n",
    "compression_rate        = 25\n",
    "\n",
    "model     = LADMMcsimageNet(layer_num)\n",
    "model     = nn.DataParallel(model)\n",
    "model     = model.to(device)\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)\n",
    "\n",
    "sample_path     = os.path.join(os.getcwd(),'data/GrayscaleImages/testimages/Clean_measurements_25')\n",
    "data_path       = os.path.join(os.getcwd(),'data/GrayscaleImages/testimages')\n",
    "prmt_path       = os.path.join(os.getcwd(),'train_parameters/GrayscaleImages/LADMM_Net_layer_%d_ratio_%d'%(layer_num, compression_rate))\n",
    "model_dir = \"./train_parameters/GrayscaleImages/LADMM_Net_layer_%d_ratio_%d\" % (layer_num, compression_rate)\n",
    "model.load_state_dict(torch.load('./%s/net_params_%d.pkl' % (model_dir, epochs)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "bnFw4wIvf7bD"
   },
   "source": [
    "# Testing stage"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 144
    },
    "executionInfo": {
     "elapsed": 41216,
     "status": "ok",
     "timestamp": 1614104835827,
     "user": {
      "displayName": "Juan Ramírez",
      "photoUrl": "",
      "userId": "10349054824212831472"
     },
     "user_tz": -60
    },
    "id": "ZnekgvXkgAso",
    "outputId": "88316f2a-57ea-475c-edbc-4af018bbc6a3"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAggAAAB/CAYAAACKRDozAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nOy9SY8laXKdfe48zzFHzpk19cRBTZALQQC51W+QdtJP0UZ7/QbtBQFaiBBYgKgWJDW7ya7qGroys3KM8cad5+FbxPdYmL+REZVZnVFsEm5AoW5G3OvX3cNfe82OHTuWWK/Xii222GKLLbbYYvOW/Mc+gdhiiy222GKL7Q/P4gAhtthiiy222GK7ZHGAEFtsscUWW2yxXbI4QIgttthiiy222C5ZHCDEFltsscUWW2yXLA4QYosttthiiy22S5Z+lzcnEol1IpHgtVarlRKJhNbrtfh5MpnUer3Wer1WOp0WbZTr9Vqr1co+y89Cy+VyWiwWSiaT9p5UKqXlcqn1eq1kMqnlcilJSqVSl47Nv/33ZrNZrddrLRYLpdPpyPE4zmq1ivx8uVxGzp/v/D7GPXqflslklMvltFwutVwulc1mlUgk7Jz5OwwGA02n08R7/XJJjUZjvbe3J0mRey/J/nZS9L6F98B/zn9mPp9H3sfzEh7juuN5e9u/3VWfDy38Xn9+b/td4TVedfzrzumq3131rE2nU83n8/f+LCQSibWkyJr1a9z7DP4L1+hV5x/6lUQiYeteUsSn8Dn/2n8P7+f36XTa3uPPnffznbxeLBZvfK//rnCtv+nfofl79F2/u+690rlfyOfzWiwWWq1WKhQK5vf8OdyUX9jY2FjfuXNHkrRYLCK/u8oXhO972/V0lZ8J70+4N7zp5+Hv/DH891z3vtlsFnlfeF1Xnd/vu96vu46rrt3bZDK50i+8a4CgdDqtfD6vyWSier0uSRqPx0omk5pMJqpWq0omk7YQM5mMut2uUqmUZrOZ8vm8EomEFovFJSe5sbGhxWKhVCqlcrmscrmsUqmkfD6vbrer1WqlwWCgfr8v6fyhyOVyGo1GkRuQTqd1dnZmP/vZz36mFy9eaDQaqVqt2mcWi4Wq1aoKhYIODg5ULBY1n8+VSCTU7XaVTqeVSCSUTCY1HA5VKBQ0HA7f5ZaZpVKpS4HM72Pr9VoPHz7UxsaGfve73+mjjz7ScrmMOIharab/+l//63v5vtD29vb0n//zf5Z0/vf3VigU7HW327XX4WLn7xh+5vXr11cez39XeC8nk8kbv8ufgxRd8P4Y4XVc9bcKF753vv75CD/v/31wcPDG8wmPz7MtXV7gg8HgjecXvo978Q//8A9vfP/va8lkUtls1gJVgvFCoaDJZKJ0Oq1isaj1em1B7Wq1sk2MNY9fSCQSms1mtkGXSiVLSDY2Nuw7i8Wi0um0ZrOZBoNBZCMcDAbmc9jEk8mkut2u3Z+f/vSnevXqlWazmZrNphaLhcbjsdbrtfL5vMrlsjqdjpbLpV3LcrnUZDJRIpFQoVBQv9/XarXSarWKBA+r1UqZTEaLxcKSDp9w8HciAeJzJEcEFhyXJMYHQT5o8s/Qz372M0nSt99+q5///OcaDofK5/P2/nQ6rb/+67++kWfhzp07+vTTTyVJJycnkd/559X7/vB9mUzGXvd6PXsdrqdyuWyv/Rr3AWT4vX5tTafTyPv8v7PZrL0uFouR93nf4r/r2bNnkfe12+03fm8ul4u8z//On2u4jsPzxcI9yd9b7xPD/Zbj//KXv3zjcaV3DBAymYyq1apl191uV5lMxhZrpVLRbDazqLXdbqtUKtlGPpvNlMvlbFGt12uVy2UNh0MlEgmNRiPN53NVq1VNp1MNh0N7qDOZjEqlklarlVqtlorFomazmdLptAaDgV6/fq1sNqtSqaRXr14pnU7bjf/kk0/0/PlzTSYT/exnP9Pjx4+1Xq81mUzUarV0dHSk+Xyu0WikUqmkly9fqtlsWpBRLpeVyWQizvpdbblcWoB1VWT5rvbs2TNtbm4qn89rOBxqc3NTk8nEHHS32/29kI/rLJPJCATh6Ogo8jv/wF61OCWpUqnYax8shO/zjsE/8KH58/CfCe+Bj/S9I6hWq5H3+evwiysMJLwD8d8bLlwf6PjvDR2ff878uTcajcj7/Hn4DSLMZG7aksmkyuWybYTT6dQ2tGw2a9l3oVDQeDzWZDJRoVC4lIEvFgv7L5/P2898sNDpdOyz4/FY9Xpd6XRatVpNmUxGq9VKy+VS5XLZkoDFYmG/A4FYr9fa39/X0dGROp2OPvzwQz158kTSueOs1+s6ODiwgGU6nardbiuXyymdTms8HkeQTgIYvmO5XFrQk0wmI5u5R/h8YOEDAY+y+M/wfr6PYMIjnq9fv9b+/r4lQjxrICAePb2pZ0GKbu5SdL0fHx/b63BDvwohCd/nfUatVrPXoY/wG6v/TKlUunTumL8/od/n2ZSiazD0M+H5YmFg78/dW/i9V+0b29vbkX+/fPnyjZ8Jv5cE/zpE6p0ChPl8rkwmo/V6rWq1qkQioeFwqOl0ag9Do9HQYDBQJpNROp22aHs4HGpra0vD4TCyCHCiLMJ6va7pdKqTkxOVSiVNJhNls1llMhkdHR2Zk9nb29N6vdbu7q7y+bwt6NFopFarpdPTUzvvFy9eqFwuazKZGNpB1lOv19Vut5XNZjWbzWwj6HQ6kevmPK6Dhr/LQrjz97H1eq0/+qM/siAmlUopn89rPp9rPp+bU44tth/C2CyTyaQ9h2xGqVTKAvlcLmdrXTp3qmS3BLeJRMKQPHwFmx+JA2jZbDZTr9ezzLhUKmlra8uSGel8w2DT8GVEkItsNmvIZjKZtGCCc/DIJAGIPzc2b4IJfwzpcgDIMbmG9Xpta3a5XFqwwObPd/ryoSQLGLg/8/lcq9VK29vbhtbyO4KIsLwTW2zX2TsFCLlczhCB4XBoDyQZVC6XU6fTsUVDKUKSWq2WUqmUGo1GJMLBVquVSqWSbewsjnQ6rWw2q8FgoNVqpX6/b9/T7Xb14sULbW1taWdnR+VyWdlsVt1uV9Vq1UobZ2dnqlQqOj09tfPOZDKqVCp68eKFZTXSRZ10vV5bMDKdTi1Y4Xe5XO5KyOc6e58Z/dHRkTKZjFKplHK5nAaDgcG0IDaxI4jtpi2ZTKpSqWixWGg6nRqCweaUyWSMKwMqyDoiu2UD8/Aym18ymdR0OjUUcTAYKJlMKpfLaTweK5/P21rsdrvqdDrK5XKqVCra399Xv9+3bD2TydgmPplMbP2Q4IAu/P81ejuv/5+/EeFAsKnzXy6XiwQelFI4Zrg5cx/8+z0iwD3IZrNWjpFk35HNZi+VHlj3oKv8TeAoXZctxhZbaO8UIMxmM52cnGhnZ0fVatUevtVqZZk5ZYPxeKxCoaBisahut2vwHHDIm8g2o9FI9XrdAgPpHM55/vy5ksmkWq2W8vm8ZQ2NRsMyk1//+tfa3t5WPp9Xo9GwICGTyej4+FiFQkH1ej2SxUwmE81mMwsOstlsBBYGPspkMspms4Z6jMdjlUoly5p+H1Th+1qz2VSlUtHx8bFu3bqlTCZj9V4cWCqVuhLmii2292VkuLVazdbE2dlZJHsdjUZKJpOWsedyObXbbc3nc4PieXZns1mkrg4HAP4TvAWCDgJkfl6r1TSfz9Xr9dTv95VKpSKBNJv87373OxUKBVWrVfV6PeNR4AMIAFj30gWhmWvzQRBBjIf/fZnA8wq4L1eREEETPBncozSeYOl9U7PZVDab1cnJifb29uy6+RtRxogDhdjext6ZgzAYDHR2dqa7d+/qm2++UaFQUKFQsIWxtbUlSdrc3NTx8bFGo5FFw7VaTcPhUNVq1bgC/kEtl8tKJBIWiY9GIw2HQ21vb9vDjaOBwLhcLjWfz1Uul3V4eKhms6lSqaRSqWQZAA6nXq+r0+kon8+r1+tZVoBdV7vtdrva3NzUfD43LkSj0dB6vVav1zNIlcV407a5ualsNmvEKUkGQa5WK6vb3tS5LBYLIxc9f/488jtfO7uKSCdFoVdPVApJhf7v4sl9vqYpRWuKHt0JOQOeVOj//teRfTzy47kEN2E+qLuObHkVYTO8L+H1v29jo+v3+yoUCsZNIrOdTCa2We7s7Ojly5eRer2v0efzecuGPdxOJj6fzzWdTi3T9uhELpcz3wBpESifTTSXyymVSllGTlmDEiJBRljnZ7P2KMBsNrPyhEcn+D8lVF8m8J1X/NyXXH1QIV10UqTTaTsPyg0cx9+vVqulUqmk4XCodDp9iRfC+d2UzedzIxmHNXTPDfDncN3z6p//ELH1x/NrP1yfzWbTXvu1FR7P+wW/nq4jPV7nF/wx/Hd5DkN4HX4/DN/nz8P7xPA+e/Lmde/jeO+NgyBJ9+7d0/Hxsb799lt98MEH+s1vfqPVamVEC+p9yWTy0oY7HA5VLBZVLBb17Nkz5fN5IzlKirCC4TAUi0WNRiODLn1LEezfyWSiSqWiXC5n3QelUkmNRsOIh5PJxG7QeDxWsViM8BSuMzZ/6fyB4DxSqZRKpZIxtIvF4iXC3k1Ys9nU7u6uvvrqK1UqFSOIlUoljUYjI4oCicYW200aWSxcHRIBAiw2wOl0qhcvXtgmt1gsLMgn+22321qtVpH6eaVSUb/ftwCuUqkYaRieDcdng5/P55GNdDqdar1em5/gvHxGz3kQIEgyhJCkBeie4JsNOJFIWClAOnfurMnVaqXRaBTJ5Pk+zp3vxrzzZoPHJ3qiIvdfOifZVioVff3116rVasYF8ffBf29ssX2XvVOAsFwu1e/3tbu7q9lsphcvXujjjz/Wy5cvdXh4GIlwwtatWq2mUqlkm/OPf/xjIzuy+Gq1mi0E2Kaj0UgHBwcWbddqNZ2enmpjY0P9fl+j0UiNRsOygUKhEKl14rh8awlZ/7uY5yXAYoZsVSgUjDxF29VN2Hq91q1bt1StVvXpp5+qUqnoo48+Ujqd1nA4tDZQMohSqRRzEGL7QWw2m6lYLGo6nWq5XGp7e1tHR0fq9/uG5LF585o2QDodKpWKtre3VSwWDUan5Rl/MplM9Pr1a+MGsWnDCapWqzo+Pjbyo0cB/OZIcgG6OJ/PjejndRLm83mk/dkjGtIFd4KuKjgVq9XKOikmk4mhFh4N4Ji+s0KKogJ8D+fL50FXCFA+/PBDJRIJffbZZ0qn0/oX/+JfSLrolvHlxzhpiO1t7Z0ChGw2q1qtpk6no42NDaXTaY1GI21ubqpUKunJkyfGYg5hk36/r263a9Dg8fGxwfW9Xs8QCAhHkIx83ZDjbmxsGNTYaDQ0m81skULgoQxCSQFYiKj6+8BscA/m87mdC7A+TgOIFGLVuxpBVqVSiXRSSBdB1xdffKFarabbt29HvgeY1bOib8rm87levHgh6TJL20OGPlsJ77m/P/51eN5XwXMh98O3G3qYMoQIfQB3XbnAB5UefrxOL8Ef7zp9A38O4fv8Ma6CPaUodOrv0Q9R4vLGGoU5v1gs1Ov1zEe8ePHCNkLf+gisDurINRBEhHA9GynaKNVqVaPRKMIFANkjUKd8wf1m7XtY35MEQQt926F/zjhv396IP+D60um0IZ5cL+WGTCYTOZ4P4PEj+Eg2c45JYEBwRYABenJwcKB8Pq8PP/zw0jGl8+diOByaj7gJI1GTLrcb+uea7ivpMpTu15f/THjOfg35tRV+r29f9mVE3/IYvs8f269BKbq+fGnj8PAw8j7/d/aw/3VlV/+9od/yvvSq1mopej/9Z8LvJUm+DlF6pwCB6LVararT6ahUKqnf76tSqSifz1v5AcKc/5x/SNvtthKJhIrFohKJhLa3t9Xv91UsFjWZTCwTpt6cSqVULBYNtqN+CWERCM6ziKvVqs7Ozmwh80emC+L7kPd8y5Mk+z6+m+vD6Xyf47O5JhIJ7e7uajKZqNPp2D2kA2R7e1ubm5tWS00mk9ra2rLsqVarXVoAscV2U8bahUgHX6hcLmt7e9v4E9Pp1FA3iIaSjMgIEbFQKNg6JwsGhn/16pX9rlKpRPQAfKsgjjOTyZhuASUN3zVAYILIGF0PBAe8F2JhSBzEL/gWSi+yhi/wiq1v4l/47g4vHgXJkiABsSn8QiaT0dOnT60kS7LF9yJexbXEqGJsb2vvFCCkUik1m81IO91isVCn01Emk7HA4LtIUcgEDwYDpdNp9Xo9U0gkE9/a2lK9Xlez2bSgot1ua7FYaDAYqFAomDIirZbz+dyU0BBbYlEQTHBub5vds+glGdyYy+U0mUwMjSAYYWGj/PiuUB6lD6JmNCaki+AER3J8fKx6vR4hg71+/doEaTyLOrbYbtIoD6B7IsmChfF4rEwmY8RBkAbKDqwRjxb49sFOp2O6ApT3IEKv12tTOvSEP3wIm+t4PDbeA6RFSg58L4ELrY+0cHtYn4BjPp8bKsCa55gQHcn2+U6f7bOOPa/JmydDIhTluyUGg4FxIXyARdmCtm7ukSeK3yRBMbZ/fvZOAQLZAZAgC7PRaFgNfHNz8zuJerlczhwGmYV3FLQSFotFffnll8pms8YWpscXIuRgMIgIonD809NTlUol2+CLxaItEo79plbLN50r4kOTyUTFYtHU22BTI7TkSw/fV1QJGBDz92V3d1eVSkVfffWV2u22Pv/8c/35n/+5/S18XZfa7k1ZMpk0eC38e3v0xJcbQolR37ngEZ0Q3fF8Ea8mGLL6/ftCWNAbcr1SFM4MERd/Hq1W642fkaLXeJWaoxQtJfjrCCFR/z7/NwyfJ9/R4aHOsDR100bnEc8gaxPzZD5P/vPdBV5LYDQa2d8Sf0CwgcHKp55PUODLBf7ekY37EoMnIMJn8KiA72gAvfRdACAMlCUQW/PnyPr1ZVePOHj0w5dFCHa4jwQZs9nMroV7Wa/Xtb+/r6+++krD4VAvX77Uo0ePzBfw3FD+uckgwbeoh8//VahqmKxdJUscrmn/b7+Gwvf5de2vPVQg9N/rfUnoR69SXAzLlSGKjvlygxSF/v31hl0H/tx9eTE07xf9OYWlHO7LdUnkOxWpJ5OJ2u22NjY27Itpb1oul3ZBXGQulzMxD2/j8VgnJydvdOKpVEr37t3T1taWVquVZcOTycQi59BQUuPzs9nMMnnqj2QB/hzeZgOlvMFDQX+1JFvQ2WzW4D8WNY7huo0Kq1ar35npP3jwwOqcH330kVqtlnq9np4/f2695ZwvTsf3b8cW203ZeDzWixcvjIhHNu4z+0KhYF1L6CB4IR84CKCEviQJjN5qtdRoNGx9gQiSNJDFk/2Px2NDEOEx+ZIkGgl8h+9+wF/4GRGsbdBSkhZ4VIPBwMiKBOjwJXh/sVi0z+fzefOVJCDSBeHToyCS7LsJpiRpf3/fuBgffPCByuWyer2ezs7ONBqNrDQBKhITFWN7F3vnYU1ffvmlEomEPv74YyMFjcdjY82nUinV63ULGoDveCDJHDAyUaKZVCqldrt9aUDGm5QLPeGn3+9bT3QikTCxFJwFA0swWp5QfrzK4EZQxlgsFioWi0qlUoYo4ISq1aq1d+EMQBWuI6b0ej0lk0nV63UrkeRyOdXrdVNvpIxzdnamWq2mu3fvKpvN6osvvlCz2Yy0fEqKICaxxXbT9vTpU52enmp3d9eyyNFoZJkyGySwOxLD0oWEMdktSAM/p1MAzQIyfS9F7I8DokCWTWsynQW+a4FWxNPTU8u4KR0i7OQ5BxjkSt9uDQ+DpIjkgZZMv7mDeHCelEY8EuHRQ77Db/bL5dKQUtZ+Pp9XNpvV48eP9ZOf/CQSGHCP49JjbG9r76yDIMnIcr7mWKlUIhBapVIxiVLgfAiI3ugRJgCYTqdqNpuWURQKBYPV0ETAQsiVf9P6xDAXBjGRyQBVttvtCDHoKuOcR6ORHj58qHK5rKOjI5XLZdXrdQsKyFDIGmAMSxfjYq8a+EQ9dW9vz9jUvV5P3W7XhtN4+On09FT1el0vXrzQ06dP9cEHH1hrF87D6zfEFttNGbwYCMCUFHwrIJshXT4EEaxVNk4+40sRvvcflMJn1rw/1CXwhELposxAaRA0Ei4Cx2Ete/Ezzo3v4XwJwlGAlc5bukkgSFggRrL5c+7D4dBKiqAOHN8TFUEZUXulHZzkBx4IicrJyYmOjo4iPoP7cJPDmmL752XfK0CQzrNeNsLRaGRcABwDUB71f4gzb2LWs7Bh3y6XS4PNcrmcbt26pc8+++ydYDE0ESAppVIpVSqVS3XrXC6nTCZjA2BCXgKQpV+cxWJRm5ubGo/HliXhUBgd64mSELXIjq4ica7Xa7169SoyPIaMh5kSyEUTLG1uburs7MzGUcO2Bj25KaO1Sro8Fc1/r3dQ4XVfpRcR1uh8/d/zFsLP+6lo/rtC9MbXEP25h+Wgq9pEwyDXt2z5OmFYc/XH8+8Lr8PXDX0t9Lp2TY+uhc6f678pRIlsXjrnP4QzAvx68qU+1iY8H59N++4cgmMklxkUV61W9fjxYzuWr+vje/z9oLTApksHAwHAeDyOiLT5IMOrN+Kv4CBAFoZbRXLDPYE/sbe3Z8gDmz/Eao+qeA4CSARzZXiuPZERhJaOjsFgoHw+b2gjCpPeL9xUiYGAT7pcu/fP6FUqpVK09u6R5HAd+2vw3xVyLK6ajrq/vx95n29T9D4o9Av+e69bd/5z/ntDLoA/9+uQ5qvG3od7qv8uf66hUiznd92z8L0DBL6QCYgw8KfTqWq1mkF3bJDpdPrKtjvqfyzEW7duqdFo6PXr1zo5ObE+6u8yn4GsVitVq1WbLLlYLHR0dHQJhWDxNhoNW4yhQhv24MEDQ03u3r2r4+NjnZycRJjJrVbLUAogRUZjs7HDRMYBAsGORiMrY3S7Xcs2aFOSzjcjaqrIyWYyGWN7k8Hl8/m4pSm2H8x8l41H8lhHiIgRkJNdewEhPuON9TGbzfThhx+qXC7r2bNn6na7ev36ta0dL4XMuYA8sP7hMvB7NntIlX5aIsdjrXmn65GERCKhVqtlsueZTEbdbteE1CgvkkzhxyhPkAxQlvRlAE88JvDx3RDcP4KoXq9nQSSlnLOzM92+fdvubXi/Y4vtOvu9AgTpIhIEXk8kEur1etYOWSgUdHJycq1oD/V6+p9PTk701Vdf2RRHHAstP2+yRCJhixQ7PT1VtVq1xVguly9lf5w/iw1CkXTOBoVDUCqVdHh4qGw2q3K5rK+//tqg/Gq1qkKhoM8//9yyXU/G6vf7Bq2enp5aJwRRINkCCAuZw3q9vhTQpFIp7e/vG5lrPp+rUqnYPApqpzio2GL7IYyM3EP9/Jy1FbYJUvbzYmNeJIiAv1armV94+vSprTt0SDxSwcZHoOCPTaAB7wm+Dxu97yDw/AOfnXsyI1wnNmLf+u35Ucvl0sbTJxIJW/vMaQC58OOiuRa6oUA6fOnF+zK6OSBI+qmVJBigi29DnI4tNuk9BAhX2enpqfUKvyk4YGwrC5PZ7qlUypAGFiEEpevIhLdv39aTJ08iMOx8PtdisTC2tHQ1rE2JAJheku7evWsZynq9toCBgKNQKNj5v3r1Ss1mM5IxsPFzTMowdIMAlRIoTCaTCBkTchXOglYphqEAS9LWRgaC9OxNlhhAZCRZXzrmoSz/Nwvvvc/KfHtgvV6PvM+XKcLBLt78d/nXu7u7kff5MpMvS4TP6VViWuE5+LZCH6CG7wshPiwk33qkzbcyhn9PHzj6jDAMDL+PaNe7GsJFfD+kOYJtauqMZ+Zvz8+9zDGlPvwCZTdIeJ5jwAbPPfCqiqvVShsbGzo8PDQ0j2DBD1FKpVIm0kQboz8/UFE/Z4GAw4+a5jxAHCaTic2GgcAoyTQMWM+0H/rpkcDQtGwSYHmyJEEUgQWEbAIViKJeFIpA5KbMD3EL25qvUj4MkWVfIvAltuvahq9qeQyP51+HRHi/bvzxQrTFlwT8uYclVF+m8KVR73PCc/L3JSw3eH/pjxeu93B4Hha2YV7Fh/N2Y1q8fuMP+00l2QL1/x6NRleWIa4KDpLJpAqFgo6OjpTJZLSxsRGpKyPvLF3Abm+ybrcb0U1ngbJJT6dT04SHCNRqtUzgqVwuq9FoKJPJqF6vq1AoRNjU8/lc29vbJtXsoVNqtn6krb+PqCTeuXPHrg80JJU616oHPYDz4cfoxhbbTZrfpKWLTR84m7o365DyQqVSiWTLBPLhZi7Jkge6EQiWPHkRWB59BOmciAwCSdAMv4G1sVgsrDTHf6AZg8HAggSCEu9PuDaCddY+ViqVtLu7q2TyXIIehBKUkGFUlARZ09zHROJCY4LvhheVTCa1vb2tH//4x6pWqybi5tEULxzF8cONO7bYrrKbE+vXhXzqcrm8Vtjh97GNjQ1zJGyMXlNbikZK29vbl0giGG1Yp6enarfb6vf7mkwmNlcep0QE+eTJE0nS/fv3de/ePRPgyWazEQIf7GUWOOeHOIukCCErjIBpsdza2tKjR4/08OFDVatVNZvNiCAMgYgfAhNbbDdtvvYPAoB5xj/GMw8fh4BCusiYPdOfzc6T98j+QQv4PJsnr/13szGzFiES+v/4ft8KiQicL5sQgHiu0nK51Hg8VqfTsbbqe/fumcCbX+OgFXRnEbyAAnIOHNtzEAiiQCLS6bQePXqke/fumd6D/7sQMHglydhiexu78WIU3QQ3AWuxmbKhQkx8EwEHDsPR0dGVBB1IhLCSPZsabgVZxXA4NJ15BFGQgYbwhJ4DoijUQ1+9eqVGo2HzLCAoFgoFK0egf4CdnJyoXC6bvgEQLA4CciPwKOd9U5bNZnX37l1JURhcugy1YddB5Ht7e/Y6hNw9hO/RoTAA8iUBD++HNVf/O9+BEHa4eCjRlz1ClrSHTj0CFpYU/PE9dHp8fBx533dJlb/p+P4aw+slIL2OB/Q+zLcU+to+//atgjyjbMpszF5N0JMcaemjJs+Gh5YBQYPvXODzCAwRTAC3c69YK7yHIMOXR0D6vMQyJEI/r2Eymahardp/iKj58oInHkoXQRTXAH+A33F/aCX1JMNnz57Z8CvI4gQP9Xrd5t1wj0K0531bKpUyCKYxJTkAACAASURBVP06Rr5PgsLz8evLr9XweP59VzH8w/f5dRxC7H5N+uOFfqbdbr/xM16hNfyd91vhOryqdT/cM71v8UEeQ/O+6zrCEu/b+IMbDxDW63MtcN+697YO8LvMt/3w706nY3A/zh/9A0k2M95vwBAKWZh0GIRzECAe1et13blzx9qhsMVioc3NTeMYwF2gDPD06VOdnZ1pe3tbs9lMu7u7Jnxy584dy2IODg4iMtTS+WbAeO3j42PrHikUCtra2lKpVLKsh2uK+51j+6HMO1E2Ija96XQa2YjJkiHSEohLF0x7NkbIeQTDbPi+NIGjI1Om1OA5CaPRKKKsyqZKYE27Mgbpj/JfMpnUaDTSdDqN6BV4AaJcLmcTVj2qQTeDJJu4yHVVKhV98cUX6vV6ajQahlrQVYXfYkKmH53N9TKX5ejoyDgH6XRa+/v7Vp7wwQ6JSmyxfZf9IHRWP+XNCxW9D/MZRjhgBSPzXq1WarVaWiwWOj09NdIUXAkWMq1aCBQBZU6nU33yySdKpVI6Ojqy2ma9Xo+IKQ0GAxteAznx66+/Vjab1cOHD60zI5/P2whpSaYo2Wq1lEwmIwECTuF3v/udpPNstFwuq1qtmhiUF5yJOQix/ZDGOiEw5dn3g5eAzXkvXQw+ww2zJhADnmUPkftj+vKGz8z9XIXpdGqJAJsmhMlGo6HFYqF2ux0hAxJkULfHVxAcgGYwJ6XT6dj3FYtFCxT4DySR8/vbv/1bpVIp3blzx2SnGWjF5zOZjPr9vrLZrKrVqknLc63j8ViPHz+OaC/UarWI7LPvIuGexBbbd9kP1u/Cwn+f9S+cA9E5sCCiR1ir1bJshXoidcHpdGoSybRTbWxsmHNJpVLGSWCx12o1Ey7pdDo6PDw0EpJ0DuughT6ZTDSbzXTv3j3dvn3bEAy0IcrlskajkQUEtDpmMhk9fPhQr1690mQyUavVss9ls1lVKhUlEgkTStra2jIHCyTpOQ6xxXZT5tsTPYuekoIn/iEOxGYOpO7b9sLWSP9z1iDfFZIZwzHJ/J5ggmQlkUiYTosvJTUaDXW7XU0mk4h0M0EAEvKgGr1ez3hAyMpL5zD3YDDQ4eGhdUKlUilTXaVl+9GjR6rX60omzydXMhmSsgT+JpvNWmcCg+68oJPnLCQSCT169MjQCHQg0um0+Z3YYnsb+8GflPcZIADVASuyYBjcgp2eniqfz6vRaFgb4Xg81tnZmc2HJ2tvt9vqdDpqNBqq1+vqdDp2zIcPH6pUKmk4HJoKV6vVUrVaVTKZ1IsXL4wHsL+/r83NTUnntTMvajIajYz53Gw2DbbEEbbbbVNcrFQqKpfLVoevVCpar9c2CjuZTOr27ds2ZttnJzfJVkbB7U121c+vm3Z2XT0s/FtiIRfAk0/9tETPMwj/7dGs8Nn0Nb9nz57Z61AhztcA/TmE3AJv/rtCJUqeG0nW0ipdntLoORf+ddi+xb29qayR9edVCr32AJu/5wv4/3stFZ+5w1PwQkjSBVoYzl7gZ74N0Nfe0QfxBMPVaqXt7W1rXabDolQqRTQDhsOh8QsajYYJoZ2enqrZbBpBem9vzxIQNEzgLknSzs6Obt26peVyaeJmBE/ZbFa9Xk/5fD4SWB0fHyuROFdgBE0BxcA4xu7urv390W7wHQ2e+3BTzwLH9614UrRd0D//YefaVfyEEF3y68YfI7y+kDSLhRyhq9oXw2m1njPgeRHeN4Xn6689vA7vx66bSunbtZ8+fWqvPedAiq5z71dDcj6fu46T8s4BAovupgxBI8RDrrJMJmOLfTabqVwuG4vYG21CiDH52REnJyc2Z55shVLEYDDQYDAwkZFWq2WKiNVq1TKV7e1tdbtdq2s2Gg01m02tVisjMkJg9K1aOITVaqVms2lQ4PPnz41sNBwOjXSZy+VMsInsQZLu3LljuuzAmAQfscX2QxgogCcmJhIJe755DwgfyFbIH5Ci8q/Uy+EyoBtAxu6DajbAXC5n2T/vQymRLJoaPdMUIQdD/oO/QEcAn6Es4jVZWq2W8vm82u22Wq2WptOpCb4R4PBvAnbQzFqtFtF8WSwWKhQKms/najQa1kGBL2JjCtskuU+5XE6VSkWZTMYEnLxAEuTMeEZLbG9r7xwg7OzsvLX08bsa7T5AcdcZi6jb7RqjP9wUK5WK6Q6QyfAdz549s9+zsQJfQlzimL4XGbIQkrHdbtcCARiy8BqYsugZ3GQXOLdyuax2u62joyOrX/r7ATkSOdb1eq2trS3V63UrO0DCRMMBRxwHCbH9UMaGDXyPDornCUgXJQApWprwMLnvfJAu5JZ5r3ShxkhCwPri2ScAAVWg5OHRRq93sFqtVCqV9PTpUwtuCA7YYLlGAg6uh8Di4OBArVZL8/nchrnBYfD3AWTCkxzhb+D3yuWyXr58aaVKAicQEZAaX0KpVqvK5/PGf+K6Q+6Hv4+xxXadvVOAQJ3sph6u0WhkLNy3sU6nY208ZOBYpVKxoUlov/Pzo6MjEy6iLgliATOYefXUBXO5nJUqiOzL5bIKhYLu3LkTgTrJdjiGdI68DAYDQzNYvAcHB3ry5Elk0aKaCGHSL36mxk2nU41GI6tfEgxAaOL4N5UpjEYj/fKXv5QUHaYkRdtpPMQVQv0e0ruuDdbrWvjAMQyAfIuQLwmEbZj+nHw5JIQIr2olCtuZfDnDHzt8nz+ef8bD6/CQoy8TQVDFPLzp28HCEsgPYW+CUz0nho3b9+N7/oD/z18/Gxvtgdwr5pQAm7LZerIk50XwL53/fWjDKxQKxglKp9M6OTnRYrFQqVSywWygfqAG5XLZyM1cA6OcaS/k/ID6ITP67gvffjmfz1UsFk08br1e65tvvtHp6andS4Za+W6w9Xptui5+UJ4XaANl8MjITfqFxWJxaZ1jfu36NR1C3/7v79fJdcPevGphWOL0a+jVq1f22rcrStGSiPcFYWkvFAXDQqg/LB1iYWnUlxJ8GSFsS7xKtTHcj/35+XJIqOzqJw1fZe8UILDJ3XT0+baZ75taJtfrtWkLAL0zVRHCYS6X06NHj0wKFofKGFXKKMw7uH//vjY3N1Uulw1CJOpnARJMSLLZCMvl0lq8vDpjoVBQr9fT8fGxERT9+ReLRaVSKT148MCcFTX/UqlkDml3d9ccgnSxMTQajUi2EFtsN2kErGGm6omCbFKUI7zSotdBYPOUFBmyxAbH88yEVrJ7vwF6jQIQCT6HPLrXFkC/JJPJ6O7du4buQVTm2obDofr9vpUIm82mtTvj9AkElsul6vV6pFPDK0QSLFBO9e8/OTm5tAEQWKRSKRNK49ooY3DNlElBQShrer2IGEGI7W3snQIEL3zyh2j0FfuFSG1POofs+/2+kQql8+iVBUnN0RPDPvzwQ21sbNhnS6VSpK5HSYRoERW4QqGgTqejWq1msxFwcP67fYSXz+d19+5dQ0VAGrxz9IIx3W7XRswOh0NVKhUrf7Tb7SsVI2OL7X0bmbZvOfQlAgJlj9LxH8hAGMzyXq8EyGeow9PazBqhO8nzG2DyQ2AkKAE9RO0UxE6SZfjMQpFkqEEmk1Gr1dLu7m5EC2W5XJoPwS/QpinJvpt1DdJBZ4TvyIJHQcnzgw8+iEg1g1RKsp+RMJFIMKcFxIEy5U0KJcX2z8veKUBgU/zHMroWKpWKtQ96gyOQTqdtkRSLRWtJPDo6srkQCKdQI9zd3TWxlJ2dHUnnGQxjWpfLpck643ioXZbLZR0fH5tTmM/ntnFLF1ARMCD9ypVKRbVaTZ1Ox6STeS+MY/quU6mUcS52dnZMLKlUKtlAGNoh6X64Cup7X8Z9uG4euYcIQwlpz7b3MGAI6YUKh1fZVV0HIeTo4XA/Az6ECD/44AN7/ZOf/MRe37p1K/I+353h4f3w/ntI038mvD4Pl/rf+Z9L0fvpM86r7vNNwcqsIV6zoUsXf3+vHeDfB7/At0WGn8fvwMGB1+O7JqRot4ZXZJQuhI/4Pkkm4OY3dlow+Ttubm4av4BZKblcThsbG6pUKlYSXa/XhhyS7YMOcn5esAiUAi7Fer229mZmuuRyuQgvy3cg+PsMwdITQymLUMJg1sRNayDQCSJdhsj99/pnNITcvfmgMQwgfUnRr+MwAPLf5QcZ+XKDFF1D3jeF5+fVGL0vCLuHvPnOpLCc6s/P37PrlF2/+OKLK7/LJ51+MFToB/1Y9qvsn0RDLI6DG4lQiI/OK5VKJHDAsVBzTKVSun//vqTzG8/PyMZHo5EqlYqJixDlU9/zWYd07rhOTk7UaDTU7/dtOBNkJ19r5TxhKNNxcXR0pI2NjUjdyZc8cHR8xndQ7OzsRBwkNd5Op2OZy3ULL7bY3pfhYDwywAYWchP4HYQ/H9SzsXlBH08G5OcePpcUKTNwLh6tazabVq7wXQ+MmG80GtZGSMlwNpvZlFLPl2CDbjabxgfwRsDhhYzY4LkfHknxMs3IqE+nU21ubkZaQz1JUrrYtOBA+C4G6WJYnieBesLmHzISHNsfjv2TCBDeBImxyLDpdKpisWg1emB/yEy1Wk2j0UiJRMIgfKDPYrFo/AJaJT1vgAwJp0KUt7u7a9G5hzXpIEBFkZIG7GdfnwXihB1Ny+VgMFCtVjOdB+qMx8fH2t3dtSDGB0M4L38NscV208ZGF2Z4tCX6wJpAm+zWQ/qepEtw7LMbnm82a37vxZc8AkGrItonkI3xDX4jRuWVEgClCo5NeyCb9pvGMcNp8G2SXLcnKoZlE9AzAhj+z7r3xEeuH8OPERD5wIzOJk9evGrcfWyxvcl+sB2Ejfm72he/y4ACJ5OJcrmcCZqwQTNfncVEuYFFyDlQoyPTpiOCCN1H7D4A4P/A/zgOBF3IgnAu1WpV4/HYtNUhOhGkYL58s16vVa1WLQgg+xmNRtra2rJaoleag5SJWhpQapwpxPZDGRC67zwgA2YjI0Dm2eU59Zk9n/Gf8xumb3X0gTtyypAipQuZd9qSfa2f9bVYLAyZY14DcP10OlWhUDBZdN8hQWkAoxzIzwgA+JknXXrSoXSBksIj4n74oIPgBB8FWiHJ/AkEUSyfz1vpjMCF+xxbbN9lP1iAsF6vLThYr9fa29tTu922uoh/8N9kyI3m83n9m3/zb/SLX/xCn332makijsdj9ft9gwU9cSiRSKjf76tWq2mxWKjRaJgWAYSnzc1Ng+WoJXoNA5yWFyqi3ndycmKbNllDoVCw0gWiKZxXpVLRYDCw4IbFClpB94UkG9+8vb1tDnU4HFoWwSTIZDJpio1hhnUTlk6nL7XxYZ5D4OtrYTuk/51/HU5Zu2qKWfg+387o64lhrdHXA+/du2evHz58GHmf/50v14TtS759yJ9DWEP0z7c/Xlgb/Pbbb+2151L4+qkUvcarJttJ0uPHjyVdnqb5vgwI22f0BAi+tc+X/XyPvnSBEjL2WLoQP+L3HINj+wC/UCjo3//7f69f/vKX+vTTTyPr5PXr12q1WkqlUgbjSzJV1FqtpmTyXCodJECK+qTlcqlyuRzZuNlwCVi8gBGcAl8S8HoOobG2R6ORqtVq5Hu8OqQk67zw7dh8L9/nX/v2UM/LuAlbrVaXEA/Moxe+NBMmjVdNNAzf53/nfUTIJfJr0nMVQt/o1zvlaEn66U9/Gnmfr+v7VuPwvvp16M895Fj5313nF3wbpvcR13Ea/OuQw8S5X7fvfq8AAWj7uwxoL3x/sVjUycnJpQcBSDI0WLj/+l//a/35n/+5/tt/+296/fq19S7jdEql0iX50XK5rEQiYX23flFJssybTgOQBpzPYDCwIAElRtoUa7WaaRZwvaEjLJfL6vV6ET10Mpf1eh0JFLxRZyXIgOeAKBJBBMxqSi6VSsVmUVz3h48ttvdpZK8ECb59UZKhWWTTlM7Y/Pgsa8T36xNQ8D7KCRzvr/7qr/SXf/mX+u///b/rV7/6laSLQBLyoA9UaEsmKACZKBaLkfd4cSPgech+JBCsu/l8bu3PXqOBUgqoCaUHr6hKmyg+ZjAYmL8EWfAETOnCB3vEgmMRQITBi9c/iJHF2N7GvleA8F3BAU7h3r176na7l4RqIMyERt3cZ4nNZlP9fl9/+Zd/qVQqpb/5m7/Rp59+ahv15uamDg8PI61F0nn0PxqNNB6PVavVrMQBVIfC4mQyUbFYNHTh5ORE9Xrd2L+tVisi3sT3IJhEVlAsFm2QUqinTcZJYNFsNtVut60FcjweW5BBaQLJZBayn4JJZoV+PMdIJs8HPtF65R1EbLHdpJHF0mlA4O15MTyLu7u7Oj091Wg0ivyc15QXPOnR8xgI5NPptP7iL/5C9Xpd/+W//Bf97//9v82veDZ/uVyOBOB0Q5XLZesO8IGDh/6Z6OonQUqXxXx8kMM5U+7k/fCQvEaDJOMg+bIlm7znGmGglNwrOEtk7JRG/N8D7ZRYRTG2d7H3VmJYrVaqVqt68OCBut2uFouFXr58ad0CIaz3JvPM/1u3bqnX62k8Hmt3d1e/+MUvdHp6GhFVobOBxUpbEHX5crlsmyUlhVarFWk1omUFoaNaraZ6vW4ZeyKRUL1eN94CNb5qtaqzszMVi0XLEOA+kFUsFgtrk2LuQ6vVMr6D14j3wksECMysR6uBoAZiE+pyOEuyDBjUN6mYBmIjXYbCPATv2xxD84GUh+qua8/0g1PCwNND7h7u++ijjyLv+9GPfvTG34Xn6tuWms2mvQ7RHg9vXgedeujTKzOGpZLPPvvMXntYMQzM/TV6tOh9jlN/G2Mj99wCymzT6VQbGxu6deuWlQNp5/MdCv555llmI/Nro1gsGuGwWCzq888/19/8zd/Y7zkmgb8kG8hGBk/w4NcNPoHNnnWHQitBCbwKkAJPPGZtsvl7pJJNP5fLmV9gY6d0ORwObd37eSr+78414nO4F57n0Ov1LEDw3Q4QOEFvbsLgXUmX25/9d14Hzfv3Xde+64M0fzy/tqRoqc9/5tGjR5H3/at/9a/s9YMHD+x1uN69f7rOv/m2Zv86HP7kS6D+2sNhV3/7t39rr32ZIvTxVylMhv4D/3SdLsZ7CxDu3r2rvb09PX/+3KDvXC6n58+fRx6A60iKbIzFYtHmPVCm4KFn06QeCPxGYMDgJg9Z+p+TEXgy42AwULvdVrlctiicboVyuWx6A6lUSq1WS71ezwYpSRetRiAUtVrNxr1Calqv1zYBDlIkvc5AjZJskA31K/gOIBqwlAeDgUGayMDy8HOsWEUxth/CPDmO1yBee3t72t/fN+Jdp9NRpVLR4eGhfYYgWZJt4p5sBxqWTCbV7/ct2x8MBhZ0sakToPgOAsi7y+XFkCdfcigUCpGAhdd+qJQXG+KzBBcEB5QkyPy9VLokCxAgNsNZoMQBd4jAg+smcPJIC+RHroeSDAgkXRN+9gLHCDvAYovtKnsvO8hHH32kcrmsJ0+e6NGjRwalnZ6eXtqkrou4JBkL3y8GIH4QAhYwkbRXHKNGyOLZ3t42sSNaCf08hE6nY4iAdEHoIeNALAmk4fj4WJVKRRsbG5Y9LJdLI6qxWAuFgmUujInFeaZSKdXrdftekAHveMbjsWUGLHY6HdLptBGZcIyQFeE+8PPYYrtpY3PypcHFYqHbt29bQF0oFAzmpiTo2xolRTZtSRGtAxA5T4Ikm/baB9LF2Ogwe2bDLBQKqtfrVjJgU/ffBarDuiN7ZfPFN3g00pPySERYg74VkjVMJk8QwtpFEVI6R5c8UdL/PyQjg6Iwf4J7yd8GVCQODmJ7W3vnWQy1Wi0C2VCvY2DRp59+eu0xvF77m34nKUJWzGazVlJAYIjhOLPZzDJ336/c7Xa1u7urbDardrutnZ0d+75SqaTZbGaz1fP5vE1d9MzkTqejnZ0d5XK5SMa+tbWl5XKps7MzQxe8hCmLnw07nU4bzMPkSaJ4HB3Kh2RHOCuyDC+mMh6PLTujzklLlnQxrMk7iJswesmlyxCV/7dHjEKI0MNkHoLzzP3Q/PFC6O9P//RP7bXvLPBdC9dZOMzEm4cFw+6Eq9jFfma7FC0l/Pa3v73yeP7f/hr/UAO+dDptGgJeDrlYLOru3bv65ptv9P/+3/+LQPm+nMaz7vkGPiv3sL8vJfhOIukC9WOz9KUPOqCQJQZNYBPPZrO2qcMDAFFkhDTnSOnCIxZszHQsIYzm2zNJYHq9ngVQ/X5fBwcHtsZBRpFfbjQams1mhoL4gINyB/eTsoPXYvDiUiRIN9niCEoiXS6d+QDSw90hmdonkb5MEZYyfVnBDzILfZ5XRP3444/t9Ycffhh5n1dz9X7mbcne/nyk6MAnBttJl7uRjo+P7fV1XVDeF/h7GZrn9/l7Fv7dud7r9oh3nsUQnjSs4l/96ldv/EOHsru+F9cbGT4bop2g63Mul8sG25NpQ4ba2toyQg8T19iI/RyD4XCora2tiIYCECcP5nA41N7eni1MuAu9Xk/lclmj0cgIhplMJtK7jJZCOp02R+LFVabTqTqdjsrlsjm8XC5nDyRsZqSieY8nc1EXhVyFQ0AYhZ9dt+HFFtv7MjYfn9nO53PV63X9+te/jpTWWBuSIuJCZNGeoEdpjsCBDVm6UDckSCVQL5VK6vV6lvlDUCQIoJVRkpGKmaZIQI2vIZFhDVLyRH8A/+IJz16fhFIrPCZKopQpQBoJ7tFZmEwmFpigscJGPxqNTNOBwAo0w3dccF8pS8JXIDGJLba3sXcKEMKNHUTh9PQ00pqDeWJitVrVcDi0uQE+mGDx+IEukkwKFY7BfD5XrVZTJpMxSWVf1+O9o9HIskaya4Y4+YmKvgbJlMbBYKDd3V2bazCfzzUajTSZTFSpVNTtdi1DpzXTIwcoM7JoUUqsVquq1+t69uyZBQDMeFivzydQ9vt9i+oQeIGDAKTqiVLesRFE4MggVsYW200bGxDrgAD+t7/9bWRYGs866BYbI2vBb2b8x3PP76ULjoAvp9ES7OWZ+RwoIBstpECCdF9CZC2yyYNAgDzQygyPifewMfuyCTwnfsY6lS7QDsqmvV7P1rS/RjZ5T2YGkYA8SUkklUoZKuIlnwkI8H1xF0Nsb2u/F0lxuVxemqntjYecKJ0+fZwApBqChmTyXDf99PTUnAZZBxkILT9kLOPx2AIM6aKFifIBQUMmk7HgYjAYaHNz05TYiLZTqfOBSMw+wCqViiEHzGhH8GixWFgGkcvl1Gg0dHh4qHw+r0qlomq1ql6vZ3A6GQwwJCqPXk2O8wXVIPCC6cwx5vO5sXzJNrg/XtY1tthu2jy0zaZ6dnYWIej5OjyveWalC+0RSnpshJKMnAxR2ZcyKHFks1l1Oh0jFYIYsGH6oUWUBzc3N7W5uRk5T88F8vMU/LCk+XyuZrNp5Ui6EeBN4FcGg4GtQ64d6XXQSd/d4HkQlGN86zdlRYIez5XyQQTIi7/HscxybO9qN6qkCNlvtVppMBhEomM/AlWSoQTAfrTpEExQ0yOzHg6HmkwmunPnjjmkdDqtw8NDW8wQDKvVakSBDWeB8+A8aR9E44DgA54AaozJZFLtdlulUkkbGxs6OTkx5TQ4AjiT4XCo8Xis4XBoyMlwODQnSRkCNAKYkfvC/9F18JkUNU7unyQTUMIZ3FSAsFgsLOgJ24p83cujGKH2xcuXL+21r8OF9TVfKvnkk0/sdajk6Nsr/ZS1EFL1fALf9hSen0e5EOCRLtdW/+Ef/sFee15F2KbkP+dbpcK2xH+KMzQoEWCIGHEPacHznUfA755859sGvbiQZ+CH5Qe4A6PRSO12W8vlUru7uxHCHxs0ZUDp/PlhY/ati9JF+cMjc5Is8PZ6CePx2J4xX/4cDofmXzgmAYok8wvL5fmApuFwGOlE4tq9jDr305M8QS8JFvi9t5tWUMSWy6XxBkLOgCdx+nUSPu+eg+TXYLg+PT/B8wkgnGPeZ3hfEioL+vP1KojeN4X29ddf22vPP5KiExe/+eYbex0Gat4X+L9R6Bf8/fMt2NdNyfXPgfd10vU8BuydPJGH0TwB5irj9yFv4U0iSdT7y+Wywe1nZ2dqtVpKJBJqtVq2KZ2enmo+nxvkX6/X7aYUCgWL0EejkTY2NjSbzayPnbKBdEGuYoGenp4qkUhoOBxqZ2fH5iEAfQJjTqdTbW9vK5fL6eDgwLJ8BiyBLJAdUBMlAHr9+rUqlYr1WafT6cjwGpAVP8bVE6NwtNJFLdbrJuAUvdBSbLHdpIGqhXVvsn0v++2fSyB4NlZ+76F0Wp0934bXIA7j8VjffvutlRr9sCXf1sjMAq9+ygZLQMJYZ7J41qhH+zy5kgFrkiyZYROAUwRnAD4GQmckQpPJxCTTQTO5Z7yfAIBN0xMWuW6umQDDcz68744ttrexd+YgfJ8WmfV6rXq9bgxedAlCw7mwOY7HYx0eHlprFFBds9nUn/3Zn6ndbltWxsL74IMPrATRaDRULpetNRJnNJlMTF1xPB5ruVzq1q1bFrFlMhmTPKWueXx8rFQqpePjY+XzeY1GIz1//txe0xmxXl/Mjm82mxEn8PXXX0daLqULjfnVaqVms6mXL1+a8xsOhxYYgKzgiKmTcoxQ5382m8WtjrH9IObb+diUeAZ9yYH/S7Isezqd6uHDh5Z9+c0cVJBNjY2SoD2RSOjJkydG6qvX6/r444+1Wq2spi+dByHMYphOp5ZQ0OboEUYCcnQG4DewWdNVtV6v1e12jZzIfRgMBsYd8msc8TgQDxAU6UL7JNSCoDzg2fWSIgGV1zgggCDAIDDzZHHKlrFfiO1t7EaxzGQyqVqtZjwFggLft+vhN5wH7T25XM5qe8ghZzIZ7e3taTgcajAYqFwu6/DwUJubm8pmszo8PFS5XNa9e/c0HA6N5QsUCNOZiYqdTkcbGxvqdDoRCVPgq36/bw4wlUqpUqno6dOnKhQKFvAMBgNNpIskwgAAIABJREFUJhMrI2QyGXW7XbXbbYN1gLM8tIamAjyJfr+vRqOhbrdraAVQIpkOm74nQHnolfojqMRN9TzPZjODAkMeioeufOtPODTJQ+6+DBC2H3nIzA98CofB+BKDhxJDCM7/7smTJ/Y6LAn4liNfDvEDX6Ro25K/39e1R3k485+LsyY48BuSFz6SLiB4nuNcLqeTk5NImTAssQHlU7bwkD+8oXK5rAcPHiiVSun09NTWA+sGhG17e9uCDcjLBAOQBumI8EgphD+QTpIcEIL5fK6TkxPb/OfzuYbDoQUrfJ5rBgVEUdUHCKCP3C9J1plFd0ZYSsCn4kfpgpjNZpGZD/z+pvwCLeDS9SUBj2z6NcgxMH+eXqVUiq5jX2705UUpWnLwviQsFfpAzJch/9f/+l+R9/393/+9vfb+PPRvPgn23xWWDnw5w/8u9G/+3/59IUp8lV8NS63vvc3xXY3Mtlwu6+zsTPl83hQPpeiFQKKhTrdcLrW/v6+joyM1Gg0jBUoXDh80ABlX2MDIKS8WC+3v75tDICun5zmfz+vu3bsGC0rnD9NgMFAqldLLly+VyWR0enoaQRXy+bz6/b56vZ4tPGB+eo0pbZydnVkZw7dNeiIUhEvKF0CkdIZQemDQU7FYjCx4SQYl4rA8qSm22G7S2MR99u2fO9+S63U9JNnmFW54PM9saB4y39/f12QyUbPZtCSBZIJygXQeMMLT4Xh0HdHF4NsBJdlxCUQQUeK7IQaH3AjWMPNf+v2+BegEg2zc+ADpAj3wfsHzNuBkEHR55UQ/0p5z8RwG+BLoOPjNNvYLsb2N3TgbCt31P/uzP1On09G3334bIRqxMFOplPEMRqORBQ/ValXT6VSffPKJ1eX4Xb1et3q71yJfr9f69ttvtbOzY1KsOAvfez0ajSywSKfThgZI58ELxCcc4BdffGE6DNQ9l8ulDg4OrOZPPZBFnE6n1W63bXOHMIXDom2KMgVR8nQ6VaPRsEmQZBxAnNJFdI3z9DoJXNc/l+w0tj9sCztw2BQ9isJzD6KwXq/1V3/1V3r8+LHxBzA2eTZBnm0CbYL0P/3TPzVEgU0SgTIfZHgtAEjElCl8GyAoAahBuVxWu922AIHftdtt8wsnJyeGEhCgn56e6uTk5BKaks/njVRJkM95kmj4Dgp0Dzypm6AHlVXfWsl9C0mhvrvpnyIJNrZ/HPtBnpRkMqlnz56pUCjo9u3bevXqVYTQhCgR5QEifOqFCAkBFaHDzqL3g48ajYZGo5FarZZBj5CPqMkxJyKdTiubzdpGDyTU7XZNo+D4+NhmQfDds9nMmK3T6VSTyUTj8dhmNkjn5YNaraZ+v28LvlwuW2sjNdBcLmfnQ90ynU6rXq9rMBjY+3AknsgFGlMul40ISgDBMKm3Yap+H1utVlfqLLx+/dpee/g9ZNE+fPjQXm9tbdlrXyqQohC8hxn39/cj7/PsZ/86PE8PBX711Vf2Ohz+5N/n4dFwCM1V9o8VnIXwMffzpp4FygdscH5UsSfR+k2QbPizzz7T5uamCZH5+QMQ7+j6AaGYTCbWduwJgSQHbLLoo0iKBNYosYJ20H4NWuBbjA8PD23KLM/Rer02uWjKi6vVyvwI7dcEJF46mYFMlAm4J/gmSabPAIeDIAF/4bN/Ag2uz7drSjLeBRoJKE/eFIIwnU6tZBA+/35YmX8Ww3PxkHutVrPXvjwQmi83hH7Glxg9zB5C/V7d1JcV/uf//J+R911VKvHXF5q/F6F/u6rcE3JPfMnGX2NYyvTdGf4+h997VceLt3cOEEJ1xLcxWniIuBEe4TivXr1Ss9nU3t6eLThJVkfne5mbUCqVDPIHbUC5kM18c3Mz0m5IYFAqlUykxCsSttttbW1t6e///u+NNT2dTvX69WsrQZA10IdM54J00aYJqQi9BToLQD78Js9QFiBGronfeadG8MEfk3swGo20Wq1M16HT6ZiWw8bGRpwtxHbjRhZOS58vceH8whZH3tvpdOyznnAnyUpmtCV7/QOCCDY+WgrRB6BEICkyf6BarZpGgkclKE147gLdR+l0Wt1u184V7RZEoBBSky7qzl7cKLx2BjH5MqDfuAh0KD/4VnDPxyDxkS6QFnwNqCJBFpshwUiMLMb2NvbOu8e7BgdAeZD4IBt6yA/1Mi+JPJvNVKvVbLN+8eKF1uvzMdCMTl6v1yalmsvlLFv/8Y9/rO3tbVuMMIYh+iCJClR3cHCg//Af/sO73opLdvfuXSUSCYMhw5ofWQlkx1qtpsViYWOtpfOFXqvVTIhlOp1qNpupUqmYs0ImejQaWUuldDFqFeEUHFpssd2kkcVCBJQuhp6x4fp2O5/lw/AnOIbMB3yOCBnZuW/tS6VS+uyzzwxxI2jnPaBv8HgajYb29vZsowXN8DwBOgdQh/2P//E//t73Z29vz76Lzd13HEgXWVy5XI4IoeFvi8WitWiSYbPuUXj0JUpP9lwuz4fJcQ7j8TgOEGJ7K7vx9NJDXdVq1SJ/2oS2traszu9FTSACIXEsnS+Ely9fanNz0z4H+ZCFhyoadX0/8ASobjabqV6v6/PPP48sqt/XgPJ8lE4JxNf+6LP2pCfp3HG2Wi1DPlCFrNfrms/narVaEXiQYMq3NnnpVe53bLHdtFH/DqWOyXLJeH33AXA8QbOkCPrgGfroD/gOApCA1Wql4+Nj7e/v2yyUSqVicDyIBUEM67JUKhncTxAxn8/16tUrdbvdSxDv9zWvMsk98SRIrhX0EE0E3+kCh8ojCr6Nk+N5ToJHHPADlHZii+1t7MYDBB7y+XxuGX61WrVJhWdnZ0qlUnr69Klu375t3ASChJ2dHRuWxMyEJ0+e6NWrV/rZz35maEClUtHu7q6k88ia7oFSqRQZjgRi8fr1a2sf9EpW/+7f/TtrXVksFmq32xEyIMRFn9Usl0s9f/7coEQyeOpElA+A+wqFQmQ4DAGTpAiz2js6Sh5Ao6lUyjgI9GljlHB6vd6N1hofP34s6fIUM5+d+PYjzzOQou1Ivr4YKqF5DgGCV9Ll6YtXTXcLOQi+TcmrQPqWJemy4hn2Q2Zf17Wj+fPw7wsVJnd2diRJX3755Xs+u3MD8qc7AWEhP7nRlwY8n4ZAniCBoJkNtd1uG9pGPb/X62ljY0OJRMLqrYvFQgcHB5rNZrp//76JGOVyOdXrdUsCfDeDF3daLBYm0S5djGvG/u2//bfqdDomAoUfIRGgHEECBGry/PnzCNGSTg6un/JnNps1lIC5NScnJ5YQoNsiKcJDoHwCT6Jer9u5SRfS0RglzZt6hmezmfGOQrQ5nL+D+TUtRZ9fn7yFNXQvwOd5C2FZ1XOivCpiyCX6v//3/9rrX//61/Y69AvX1f+9+QDTr8/Qr/j3ecQ3vH/+GP51yM24qs0x3Avu3bsn6TL3ytt7CxCoj9MGGPaeSzJWcDKZVKfTUSKR0Pb2tskhP3782HqZDw4O9OrVKxWLRTWbTRM1SSQS+qM/+iMdHh7ag761taVMJmOERCSTJVl3AWUJmMi+lOEj6rOzM2WzWZ2dnZkUrBcukS6mKdIS6eutyDQT5HheAQ7NLw7EVVarlX76059atgSaAlRYr9etZVS66C/HaZKdrNdru//XkXpii+19GYEzSAGbFpsQ2a3ftAgCfItwOp1Wv983hdNisahGo6F2u618Pq9araajoyN1u12dnJzo1q1bFjDn83ltb2/bZkArM0E57YEY57NYLDQej+13nhvg348YEwE9SB1wPZv8aDRSvV430jVGgsB9YW1TTqCcSql1MpmYb/n444/N9+GDKFPCy2DGDfwD7rVHD+Bo3BRZNbZ/fva9AwRPVkQdUDrP1piZ4C2ZTKrVamk8Huv4+FjJZNLGGRcKBZ2dnalcLqvT6ajZbFpEOJ1OTUyIRd/pdHRwcKCPPvrINndmJiA3il7A4eGhoRWUHHzfNkxhDOIgC53yBxv+YDCwzoPT01PjG/BZUAqGQxG1VSoVC6JAGGazmYbDoUajke7du2dtm8CAHC+ZvJhXP5/PrYMBRAZo1Gs9EAzF/c6x/RAWdigQJLO+KaOx0fKsShekPDZb2gGRTKeTgc2UDZDPrddrVSoVU1b9kz/5E0kX5D9KEj5zJxDBB/BzOhm8NookQw59kEM5AwIla20wGETQLAIDgnjarf39QfuEoH88HmsymajRaJguA8EFn8efsP59W6ckQzS8gBXvjy22t7HvHSCwqRYKBdtEIRWFQzokWQRPHRIyDuSZdDqtRqMRURFjsYBM7OzsaD6f69tvv9Vf/MVfWG2/0WgokUjYLAc/DIkFxTCkg4MD64UmiwnhHxwO8xGk8xaWbrdr7YxkQWzWGI4LGJDWJOkCmmJBDwYD42FAnkLhLZlMRohFZDuFQkGJxPm8CJwU34ljQMkRSPcmDKhXugzp7e3t2WsPH4YQoYcPfbYWlhg8JOdbD//P//k/kff5oUn+fSFE6L/LqyeGsOs/FpHL308Pt4blBt+m5e+th1slRdbTTZoXRGIj8lkzgQOweajfwUCiXC5n5GPp3Hd4MjLZOsdBd2E4HOpP/uRP7N9kzGgkgDRQAsEXgSLgLzwyh43H40u6CZPJJNKJ4ZEHnzgQqDNdEv8gna8Bro1jMHMmk8noxz/+sXVqgCRC6vaoCGVG1ChJKggSfPJwk4kDPkm6DJH7f/u1HyKdIcKK+VKwFFUW9GvjN7/5TeR9vk3RK6KGfsGXLDxMH66bq1olw/f5JNmv6dBfvmk2Ufg9UvT+XRfk+c/58kX4vfiM6/zc9w4Q2OAXi4UJDS2XS3W73Tc6okwmY/Wvn/zkJzZpsdPpRFQSYfHXajX1ej178JfLpT7//HN98MEH+pf/8l9GMmo+Wy6XdXJyosViYagGAUa9XtdwONTW1pa+/vpr0w4gaMEYFINU6tnZmTqdjnZ2dvTRRx+pVCrp2bNnhmIMh0OrZb18+VL7+/umilapVKx1CsdGEDCZTGxOwx//8R9H1BEJEuAxzOdz7ezsWOAFpCnJPgNBUjpHcSqVirWBxRbbTZrnFkjnaBnZMUEAa4K1jLwxXQbValUPHz40+ePj42Pb8Ngc4SE0m03lcjmDy2ezmW7fvq179+5ZSRFNFE8SxvAbXj+A5CGVStmo6dChsoEPBgOdnJxoPB5rf39f9+/fN52Ebrerly9f2jlI55sNaqokKt4ncJ2gK6APP//5zy1Z8Pc1n89bO2epVDL0JpRnhiMhyTq6uBcxshjb29j3DhDQK9je3rYMz2sc+M1MOo+mUqmUfvSjH2l3d1fr9Vq/+93vlEgk1Gw29fz5c9Xrdd2+fVvSOaGKBfPq1Ss9ePBAH3/8cURXnZafw8NDVatVmwRJMODZvkSfDFLifNLpdMR5QLLqdDrq9XqqVqu6deuWlUNev36tg4MDPXv27FLkB+Eqk8mo2WyqVqvZoCg/xtZDpNQdWdQItoA88L04AYiV/X7fHCROBKU0UBTeH1tsN23wgajL37p1y0YZs7F6JUUCf1CH+/fvGxfp+fPnFmAQPHhEAc7C2dmZ7t+/r3q9buuFbgfeA0LB7wlU6DQiQSAT9JunTxxSqVRkKFs+n9fHH39sx0El9uDgwK6T7M1zlMLAhXtHwEB5kECA4AVf4f/PZ5LJpHVnkbXC76BLg4Aottjexb5XgEAdTZKpAvb7fS2XS+3s7ETIObT05PN5bW5uajgc6vj42MSGptOpnj9/rtu3bxscyNAjMvCPPvpItVrNYP1cLqdGo2ELldpdv983kg8ROZwA4MTNzU11Oh3L8MkWsKOjI52enmqxWGhra8s2+W63q6dPn+rg4MAcS2gMk4I8hAOAhCTJAhKIW7du3TIHWq1W7f8EEVwbgRTOqFarRXq+uceTycQgz5sMDrLZrJUSvNqZFC0rXAeFecjMZ2vhcBRfOvAsZD9jXYoOQLqK8StFIbU/hH7w8Px8ieA6hTgfoHoo8apruslr5Vlj46L8BLcAnpEPFgh2CeAnk4lOTk5s3Q6HQz148MBKlyB+EPFarZatOTZEr6FAJwOBSyaTMYVD32ZJEFMqlWxWSygmhAAbBGgSmfl8rl/+8pc2e8F/xncdZLNZbWxsqF6vG/zvhZm8QNN8Ptf9+/etNEoAQKDg+Q8EEVwvfg7eAegL18h53ZRvyOVyevDggd0zb75U6OH3MHDx7/MlAd+ZJEWHxPn3+S4lKfrc+9JBeA9ClMlfkzd/XWFy6c37N/+7sKvKlwevW8e+LOPVEsPr8PfPf4Yuv3ex7xUgAOttb29rPB5HOAih/rgkI/L0+30bj0wtb2try3p8gdc3Nzd1cnKibrerW7duWSbNolgsFjo9PVWr1bLNer0+HwMN6Y+s2kfOZBTU7jqdToRdLZ3Xufb39y1r//LLL3V6ehpBJDBEXPhjbW9vR9jJGKUCZJlZvEyWRMURrQOmO7ZaLYMnvWxssVhUvV7Xq1evlMlkVCgUTNmtUqmoWq1G2jNji+2mjQ0/mUzq7t27evXqVWRIEOqBBLVktpKMU4PzpmSJ2BmiQZ6hn81mTS1VuqhHU7pAEIjjU6ojIAiHF/ks23MiMCavIub0m9/8RmdnZ0Z8li6CXDZgJo9ub28bkshmTlsnmz+b+WQysY1+OBzaWHdfnkQXhQTEd2qAJEgX0zWZXIsy7U1KLcf2z8u+d4mBKYeffPKJvvzyS6v153K5S2M0IQuxYfvIerVaqVAo2KhjyEDtdtsQAjJur8+O+hm67CAFfEY6JxZSb6SViUClUqkolUpZ5I/dvXtX/X5fz5490+HhoTmRcEF59jQZMwuZ/wiUqGuSAUiyAIFrgruxt7cXISN6JjefpU10Y2PDnB2lCTKk0WhkolGxxXbTRrfAZDLRl19+qZ///Of6u7/7OyMnEhR48R7KC55Pg8JqsVi0rMtvemzw5XLZjuM3WNYp30Xpg8yf44NesiHTBUCyQBcBtrGxoV6vp2+//VYvX760YMWXMeBbeLEiSba54xe8/oEfMgf/iWORSDErgs/Dv6IVGkK051Vwz2mtDtGE2C/E9jb2vQMEYJLPP//c9Aeo0b3JCAzY9GAqJxLn09Du3r2rYrGoo6MjI+lsbGyoUqnYOFdaIKvVqnq9nsGG2WzW6vitVsvahHA8yCrzGaBEnMDp6amdJ5PlrusVpm1rPp9HdN7J5iFwSjKBJ2qj8CqAFA8PD/Xhhx9qOp1aDdKLKaGI6KczejIix8JBM5K62WxGZprHFttNmldMlKS/+7u/syFqdAVgrC0IiugeSDIkko1TutBRQLKdjicCAtZhuVy2zgJKBtIFqXc6napUKkXEgij5gW5wrqH4V6/X02effWYbuA8IKDdSLiEI8UqHuVwussn7jd2XFzg3EgpQBoiMIAEER2FC4ssqnmhJchZyrmKL7Tp7pwABhAB4MJ0+H4D0x3/8x/rFL35x7WeB4/3mCDIAh4H5A7PZzLJ86tsnJyfa2tqy8oZ0XmuBmLharXR2dmY1WyJpIDwQDj9lrdfrWZ0f++abb1QoFEwE6U2Lie9HPY4SA+Qj5tJ7PgQMbAKKVqtleu+PHz9WpVLRgwcPtLW1ZfwM7sd8Pje2Mtfl643SOQ+AsdI4rFBR731aLpfTBx98IOly+6LnE/gJZ54/ICkSmHnlMq9uKCkS6HhYOOSBvG1W9I9VevHn57kFYfuWvy7fehVyPTwnwV8Tf5fQQs7G+zKCAtYDrY7379/Xr371Kys9kNGHo4shI1JKlGSlQ9CDUqmkfr+v0WikW7duGWmRjiW6jkAB0BYAWfRCYgTTnAcbMkE/KIVf+7/97W/t7+f5C5QvfAcSqCbrgHkpnAeoIIERXUhowzD8iUx/b2/PZqwQ2IAG4BNo4eTZ4f+FQsGCMM6dmQ43Yel02hDVkIPgW4q9zw0VDT3XwKt//upXv4q8z3MQfJk49AO+ru9bKkP/4dsmPQoeJr1+3fljh1yKqyStwzZk/7mreBDhufu/X3i9vk3Ucx/CkQIE6teRV98pQFiv15Hom9IBY5qvGoPL9EQch3RBZgI+Qw0MghOMZhYhmzDOBllSn8G3Wi0LLHq9ns2MJ+sGaut2uxaEhDfozp07SibPpVPr9fqlTY3sBmeCaJIk646glkitEJETL+wCZ+BHP/qRMpmMdnZ2DDEBKaD/G0VH/8AtFgvVajX7HcROkAgmXMa1xth+CAPGD8WJ2MThEfD8+yxckimhSrLNT1KExT8YDCKEXyB2Pku5YTweW3JAaYHPontAdk1WzeaOX1gul5dmIaTTaYP9ydIJKPBDlAx9IER5lQ4vMn26DDx34PT0VKlUSo1GQ/v7+yYY5ZFCSRHyMsJr3CtQEK//wGsCtdhiext7ZwRBulAMZCMGFr8qQPAbMFEv0QwKhxsbG8rn80a8W61W2t3dVSp1PuQJciKMYC9xihMim5BkkTlCTjgBWorK5bKazaa63W4kg/voo4+0XC5N173b7Rr6wLXQLYCTYdFCKJLOA4l+vx/ZxHF8oCVwB7a2tsw58jMgUhQhiVT9yFsPKeKMqtWqXr9+bY4yDhBi+yEM8i9thNJ5K3SxWNTZ2Vlk4ih1cG9k9GTbIQEY1JIgIJvNRtomvQ6IV0slSCiXy5rP55dKHRCWSViKxaIF+j6b+8lPfmIKq+PxWM+fPzdhJzZcH5BTQpXOEyTaF6Xz7JS5JFwHGjL4KBAIAgR8Aefo9REoVcBd4l4jFuWllyk9xBbb29g7BQgEBfl8XicnJ2q324YqXFWzLxaLJm7CMBdIQkCARM8MbTk5ObFsmfYmVAWz2az6/b5Ne1wul0bYgYQEU7dYLBofgIi62WxaKaDRaFjGjd27d89KACcnJ2o2m2q326pWq3r27Jltynfu3DEm8vb2tqSL7g4/rZHJjAcHB3rw4IENekqn09rb21Ov19Nf//Vf6/79+3r48KEePXpk10T9cTweGywJl4NgA5lY0BnPJkdY5aYtbNvxJQHfcuQhQSk6JMQrH143AMXbH2Im5Ms6HlKVoufrob8QSvQlGh90hxChhzqva2Hi+Dd1v9iM/cazXC7tb0pGDYue2rnvGOA14j/JZPISPE6gT3nG64oArwMNEzD4yYmeJ4EAE699SeD27duXhjXduXPH9FFAQwkWaG9EdEm64B1wbQQC3AdIm91u176LkgJIxjfffGOicR7x8MOwKNFQOgEtxT96fhSlVZKLmzL+pmF5NlQuxP7H//gfkX97//HkyZNLx8WuKrGF7cAeefUlu/B4/p5c1wrt9zr/jFzXbujPIRTg8ubXePg38td43TG8uqP3C6H/IJm+7ll4pwAB2G29XttwkaOjI4uk+TIgNyJuWoO4sNBRtlotIwkdHR1psVjowYMHRmhkISLFzPH5L51O29S35XJpQQkZtNcd8N/farUi7VCS9J/+0396l1sSMVqPyBxAN7zT4tpevHhhQdFPf/pTPXz4UBsbG+p2uwYpwmze39+3Tgx0Fvr9vr0mg6KdaWNjQ/1+35xSbLHdpIUlR9aZJNuwqNl7ETVKA5D6yuVyJLi+deuWrc+nT5+q1+tpf3//EgIpXQSVHAcUzrf8cU7wG/gsyATBPXLGvovh/2vvy2MkO6/qTy1dXfu+T3dPt2dJe5aOlwTjIGJkQEmsRAoIKcFOMAEjgggoxgiE5AgsWYos4yDE4kQWUVji4H+CLYjsxIaM5RDbGBKPPPa4ZzwzvS9VXfva1bX8/mid2/e9nu7pdqajwO9dqaXqqldvq/d9373nnnvuE0888a7vD1u2kwtFmXJdts3cPZ17CqFFIhHE43GRkef9YuUDnR6Wj3POCIfDqFarhp4XrJyymjVZtlfbdxUD1QJJBCKpkCgAJwrK/LIumU4FBzAf3FQqhY2NDayurqLT6WBxcdGQV+NgphQxEQV6ZFyEo9Eoms0m3G438vm8ODD0yLWyGlvFMhJvtVp4/PHHpfcCPftms4nZ2VkhQLIkslQqodPpIBqNwul0Ip1OCyObx6VVKhVEo1FJxQQCAczPz2N8fBzFYhG1Wg2pVEoUKROJBABINEBVSEKrLAklQ5yICslPa2trUi7FUi/LLDtI43jn2GdkSwddw9wcG5rwrNn2wOZClkqlJOJl90a32y0EOAYg6+vrCIVComfCsc45SSs80nGh40wUA9h0EhKJhIw7QvFf/vKXhQxMh56iaSQ8VyoVUZaldoHT6cR73/teEVfiPGCz2aQSgUiB5kJEo1F4vV4Ui0VMTU3hxIkTIpFMMjb7MDDy06kVTXrU3Kdms2lIyVpm2V5s3xwElgIBm+kDDp5isYhIJIJyuQyfzydVCoFAQCaFer0ujXj4PbZXTqVS+OEPf2iAznw+nwx61vQzJ0+okIsgS5263a60RuYERAZ0PB436JSTWJRMJlGpVIT8SK0EYDOn3263Rf+9Xq8jk8lgfX1dcquE2IlWnD9/HrfccgtarZYQFROJhBAvh4aGcOnSJTidThFISqVS4nixjwLRAZ1z7HQ6cLvdgsqQNKUjqkAgIHDeQTH2u92uQOhm5bKLFy/Ka51+0FULwHaW8/8F00xjcxXJThwd830wk+No5g6pGjLUPBpzVcShQ4cAbG+ec72M0Cqdbh21ksjLBZkLNcl0RMWIGlCOmBwfm80mnReZf3c4HGg0GhgeHkYwGDToKrDDIZ0WcoRYHqmDF1Y1sXMqv8comykR3Sm13++jXC6L8BHl2DnHkVtUr9dRLpelK6zTudmrgSqpdKYymYygLM1mE/V6XRAHXjcJipwD4vG4nBPPd2NjQwIfp9Mp/AqSKMlTYKfbgzKmSoDt1UgzMzPy+vLly/L67bffNmynK590+s0M4etSVHN6S5tOWerqIfO40/vYjaexk2KrOVWokWkdNJr3vVMQZ3bk9LH0b2hubqfPT1+7ef4ggnZdqxhjps7cAAAgAElEQVSArUmQOcF2uy2kIi7aNptN+qKzppd1whoO42RRLBZRLBbR6232P/f7/ULuYbtXsvt1jTFrh7k4RyIR1Ot1gSaZhuDN8vl8UlpJ54KLLgBxWMhCpiYBFzqWIZEVzR+ApVh+vx/BYFCch2azidHRUSm3Iqx6+PBhNBoNrKysSFvqkZERA0Pa4/FIWsLtdqPVasHv9wt5koQkNsMpFAoIh8MolUoGYRrLLDtIIzrAHDo7GTIq5qKvK5K4WDHlRyVAcgeICPR6PZFhPnbsmBxDawPwPQYImsBns9lksQS2NBXIfSKkT50SrTXCcmObzSaLUbVahd1ul4WecxnRBe6bmgx0Pohm6r41Wi7X4/EI76jb7cr3S6US0um08J2IDhA1oGNDp4aODe8L29lrx8Eyy/Zq72r1IGueD5wWD4lEIkIo7HQ68mB6PB6BxThIiDAMDQ1heXlZSh4jkQgASOkeGbxECphu4PFJaorFYsjlclJPzdpjpkVCoZAs+Ex9cADRG2d/hsFggGQyiWAwKN46v8fIx+12IxwOw+/3I5PJIBaLSTMVCh0RKaFeA6MKLvqHDh2SDpPUPtCELjoddCwomMS22iRHUSuBUUooFLL4B5b92IwLLbCVcgAgqS5WEmh0gZEteTs62mFFwNraGmq1mizeHPM6dcHjENqnsfSPhF7yerjYEn5nlZDWB+AxGPQwKOIcRlIw98lrIt+IqQpyBjgvsMyRLaHpgJTLZRF3Y/WV1+vF5cuXhT/BCJKcBaKtPDavjXMUHR2+z/MyE0ots2wne1cOAgk+unSn0+mgWq0KYkCOAgCB7Zmf1+ppzWYT58+fF1i03W7j7NmzMvBZBcDFltEDJxc++FzsQ6GQDH72RmDPh3g8LiQeDnw6EgCQzWalxCgWi8k2gUAA7XZb4DlWY3CC0h3ieHyiHx6PBysrK3C73QiFQkilUohEIojFYqKXQB7Et7/9bfh8PtGD4H0l8SsQCAhSQ012j8cjk2MwGBTxKs3TsMyygzZGr3QO9MKtq27IpeH4Z94c2GpJvL6+jnw+jytXrmBubk62nZ6elhJHCiFxDiCLn44+UwuaHMn3ufgyjcnonWlRLugsh2b5MscZ0Q69f6YHGdHrSJ96Lbp6gOkJh2Oz8V0kEpG5lM5IMBhEp9PB888/Lw4Jr0tXQ9Bp0nwO/rFsVMs4a5VKyyzbzd41W4X5FZaNsHaYuX560uQB6PQDGyVRMYxlkqdOnRI9hf/8z//EJz/5SYn+CaNzICeTSdTrdVlkSVoiA1kfj+xkpimAze5flG9mrn9mZkZko9k4aXZ2VoRLSFSkdDOhx0wmIxMXkRWHwyFdzTweDxYWFqQKIxwOY3FxEbFYDBMTE6hUKkin0yiVSqhUKuJkRCIRVKtV4TKwnIpIArAZabEkk45Ps9mUvO9BRQq1Wg0vvvgiAOC1114zfKZzXVr9z6wEqMsAtfqXuWzyfxNXYW5uTl6b0zsjIyPyWpdbmXkimnfAhj/A9pykdgD1Pthp0LzdQaabuG868YxgGfnyOdSLF9FHooNMCw4GA5RKJUm13XjjjXA6Nzs+nj17Frfddpssngw2tECS7u+gnQc69jrC5ljlOfI+0lHQaqW8j7FYDKurq7LQcttms2kopaToGR0gBivsysq5juRDjcKwJ0uv10OxWEShUMDIyIgEXjxnkilZKsn9sMSQaANTINR9OKh5IZ/P4ytf+QoA41jgPaVppUJzOaTO5evn2uzU7MTPMe9Pf6bTOuZyas1p0Pwe873SaRo9Bs3H1Z/p75jHu+Yn6DFuTgdp3oEey2ZOieZZaH6Cfl8fd7dn4V3PGLp20ul0ol6vS9dCl8uFcDhsUA7rdrvI5/Po9/uoVqsYHR1Ft9tFvV5Hr9eTUkeS8k6ePIl/+7d/E1IivXiSHqnoyAWX0QD5AeFwGOvr6wgEAtIfodVqoVwuo1qtSkVFJBKRScPn80k5pd1uF4VI/mj0vgkf2mw2IQbpHCGwWfObTCbh8XiQTqcxNjYmD10kEsHIyAhWV1dlUgM2B8Z//dd/CSeC186JiGgKuQV0mur1umgeMOoBNhdaK1Kw7KBNE9O0eh+jWgAG2JxBAscNc/jVatVApCN5lxyBVCqFdDqN//mf/5GxpoXJNKeATgHRBqb9COHrJmwMJMxOgn7NYxAx1OqI/X4f4XBYHBO73Y5oNAqXyyUly5p/wN4xLpdLurdS6TUQCBi0G1h2+eqrr6JerxtEkTjXkK/EtAh/B54L0RYGR3SOLLPsWvaunxLt3TC/5vV6pYeBri0m8bDdbmNhYQEbGxuYm5uD2+3GiRMn8J73vAeZTAbz8/NoNpuIxWIYGxvDqVOn8PTTTwOAeOIaCmROj5EzF3pG4CRD0YPWED2ZzEyJsCTI5XJJRQPTAaFQSLx87TD4fD7Je+qSIqYaFhcXpeeDz+dDKBTCyMiIOCMnT56E3+/HyMgIYrGYTHZPPfUUlpaWZCJheoUTLqMhOkXkK2i1NDpq1kRg2UHb1YRxNCmOOXRqeXARp1M9GAxQrVbFmQiHwzh+/Dh++qd/Wqp7Wq0WvF4vMpkMJiYm8K//+q/CNSKMz/mAcxFlmOmYM73AMcrjkXvEMaadHfZkINGRYyuVSsnYBCBB0fDwMKLRqDSPYpt4HlOrKbLnSyAQQCKRgNvtRiAQwPj4OIaHhzEyMmLo5/Lss89iYWFBggA2gmPalE4ZnQJ9n4mUEKmxAgfL9mLXpSCW7Zbz+bzoAVAnoNvtSnQ7OjqKlZUVTE1NiffPzowURhkaGsL8/DxGR0elF0Eul5PBl8/nkc1m0Wg0JGLWJYtslkIvmcSmUqlk0C9nnpSeeLFYFGguGAyKlHOlUhFkQrenZg7S4XBgdHTUUHnQarWQy+Vw4sQJ+P1+lMtl2ZaRw2AwkJrpaDSKcrmMRCIh5Ynf//73cfLkSdnGzNRmiRY5B4RYmQopl8uisnYQVq/XJcVg5jpomExD6WaIXEOJ+jOSVGlacfF/k5mhxL2qRerfTG9n/o5ZGY1m/j208uBBmWbIc+wxH0+yMMcHdQsItXMBnpiYkLFHlJBjuN/vY21tDbFYTIjBTA/qNAXLmxlAcDGkE0MHhfeO6YZarSZoABdV/T/Jlrwm8osYuLARG9/nMcgn4vGJXlJqmfOYRiCpF8Hqh3g8jrW1NbRaLbzzzjt4z3veI0JO5CJohUSSuYmAcO7QnImDehZarZaUPWs4HzBC4TrANDd70ykBjVSbFRJ1GaHeh37fvD+t5qihfcBYoqwdKPNxNSRv7o+jzfw92m73fqfvAEYHXJ+7uXxa3ws9R5idwr04iT/yU8KqAC6QfAiY1wc2L5pw/9GjRzE9PS0/Wq1W2wZR1ut1vPHGG8jn8/jgBz+Ic+fOieZ6IBDA5cuXBYKkR08EgbA/YXhd6hcMBgX6I1+ALONQKIR2uy0IBb103mASA+v1OkKhEAKBgECIxWJRov21tTV4vV5Eo1EsLi6i3W7L/vjgMv9GYlO73ZZ9UC+h3W5jenpanCBOdBRr0o4GxaNIEg2Hw0in0yLkYpllB206lcCxzOiW75MXQ6ifYzcajSIQCGB5eRlLS0sGoiPnEGqZzM/PI5fL4bbbbsMPf/hDkW3v9Xool8vCH9KCbTq/zzw9F0tWWmnxImoy0LEgB0pXAOgqAl0dkUqlJCXKeYbROyuhGBDxPJmuJNrBc2WaIBAIIBqNIhQKoVKpiMPM+0yEgWkSpnm08i2vS98Hyyy7lv3ICAKjcVY2kEPAz+hJ+/1+hEIhdLtdnDp1CpVKBYuLi0ilUgKB1Wo1ZDIZpNNpEdlYWlrC+vq6NClipEHp0UKhALfbLV5TJBKRpif0+DnwqLPAhX8wGEhr5mKxKC2TNRM5EAgImkDBIkYrXPQjkQjy+bxMBBRJicfjmJubw+joqLCVmRtttVpIJpPi9Y+OjiIUCqFQKMDv92N4eBhvv/02+v0+JicnJY1DLXueE3OiLGuk06GlZS2z7CBN57QZXdFRYBTLCNZmswkyAGyJSbFUud1uo9VqiV4J04dUUVxaWkImkxG+ArAluqY7OuoyRkbklDnmOKUDw/f1eNJOfb/fR6vVQiAQEC0T7g+AyL1TpEl3iGw0GjJnAJBAqVKpiCQ85dgZLNjtmwqqnGM0udrhcOCtt95CpVLBzTffDGCLG0GHiGgNgxz2j2AKVXOlLLNsN7suKwjhbkIgsVjMQKQjQalYLCKRSEiUT+gwEAgIzLi6uopisYjJyUl0Oh2MjY2J3Co7stHr52BnlOJwOKSDooazyuWypDqIWJAo5Xa7USqVZJJptVpCmgQ2o3wO8nK5LDnObrcrnRepQcABTr4CoU4AglIQ+aCWApvAxONx/OAHP8DIyAhuuOEGzM7OIhAI4Pz58xgbG0M4HJZUC/fJvC7JleQjFAoFcYoOCkFgugPYzqLVjolmzpqbgmiGvv7MPHlpCG0nlbCfFDPDlto0HKmvyQw56s/072dmIevtdIWIuSkOF+GDgpXpBGh4XEPdXKx4DqxGIrk2FAphY2NDIm5dncAFt1AoYDAYSAtkACImpEmR5mY3dE64EDO65hzC6Nos5kT0kcfJ5XKCJJi5Fkyd2O2bbeK13gvl23n9nHMAiKNCXRemAxjAsKxyZWVF3mNKYmZmBkePHjWMcf6+REB4rkQYNFJ7UAgCq6oAY1WA2fQ4MT+XO1XnmOcZPf7NnLidttOsfvM42anhk3k+0ueu76N5ftup2sEcuOn96X2YU4j6OvS9MM8LuiJsN+P+d5sX3vWMwZ1yIBApiEajInC0urqK5eVlXLhwARcuXDCIKhFKGxoawsLCAmZmZkQQ5dixYwLj1et1+Hw+LC8vo1KpoN/vS77Q6XQiFAoJEYhEolarJQOp1WqJaiNloCnXSk0Gch2o1RCLxVAsFlGpVFAulwUGzGazcLlcopgWCoWEhFitVqXEMJ/Pw+12o1AoIJPJyARaq9VQLBaxvr4uC93S0pIISx05ckQ06Lvdrpz7mTNnZBIi9MlJj2gCjc4Bz9tCESw7aGP0qsXLgK1yVQ1p87ll6d709DRqtZqh2onPPrk0jNCpcsgFOhAIyLzByLnZbAqsrvVKeGxz6aUuf9QKrVwsmN/nd+i8NBoNGXdMmWhBNO6Dc976+rogrSRx8xzYWK7RaAgngwgGpZpZWcXPbDYbzpw5Y7g2/hbAlmIkK8l0aacuO7XMst3sXa8elC2mB8vFq1wuIxQKYWVlRXLhtVoN1WpVPOFbb70Va2trCAQCeOWVVyTqvfPOO6Uckc1TSL5JJBKSe7vzzjsxPT2NwWCAcDiMbDYrnSCpIqiRCxKZ2A++1WpJFEPkg5UOLpdLdN7ppJB/QM4CJzzKHVN7nVE7uQperxdLS0siYESyVq/XQzKZlGNxomFlBbDVaZK8gunpaZw8eRIAhNfgdrtRr9cFwWHJJCFZrb5mmWUHZXT0GX1zzGlnlk51v983NG2rVCp48803RUjM7/fD5XLhrbfeEkTgAx/4gKTWyE0oFAqw2Ww4dOgQ+v0+5ubmMDU1Jd1jGTzQMSdPSjdqYmdJiqw5HA5Uq1VJBXIxZbDBfQCQQEQTAAnpF4tFcUyokkjHhMhjt9sVtBPY4nLxGCQZskwSgFSJNZtNrK6uotvt4p133sGJEyfQ6/VQq9UQi8Wk2okoAr+rG+hZZtle7F0jCCyxI3mHC1IkEjGU/BQKBbRaLaRSKSSTSbjdbnznO9/B4uIier0eWq0WotEo7rjjDomcA4GADISlpSUsLCygVCphYmICsVgML730ksBzjUZDcm6JRALBYFAWxQsXLshAZ+MTThKsG2YjFWo4cJAzz0jyH1Mm9Xod2WwWsVhMtBt8Ph9OnToljVzYOW14eBhDQ0PSZbFYLKJUKgmpqtlsIhgMotFooFAoIBKJCPkplUpJtUSn08F3v/tdER5hTTcAiVi0GAvVFYvFolXmaNmBG/k6mu9D+XIdqRLO53sU7qE4WS6Xw/z8vEDh2WwW2WxWeEZ02umEVKtV1Go1QSDeeecdAFtyyOQs0WHRDj/TC0w3akSUcwTfd7lcwlmipgLLt51Op3CWKGBEpILzC6uQqDyrCYtMmdJpoLNBR6JYLAqSGAqFhOxJifozZ85ISTSrBrRipK5Y4H6ZbrDMsmvZvhEErUTG5kEU72EUUalU4HQ6MT4+Lg2dXC4X4vG4POyXL18W4g1z9RQ7ASDcBAA4d+4cisUiFhcXJccVCoUk705SHqMHRgynT5+WbnHhcBgul0sibtYh9/t9xGIxafgEbKksckJgjwOHw4Hjx48LjE8IkTLRFDYiVMlIvlAoCI+B94/llA6HA5VKRRAPKquxxWs6nUaj0UCj0cC3vvUt3HvvvVLNQPY10QNGWPw9dNnO9bahoSFR7NPle4CxzEbnw8zpjp1yiObuZBMTE/K6WCzKa/NxfxLMrBapLZfL7Wkf2qnTCNBuZY76O+YIkbngg4wcNWTPyJsLvdYJ6Xa7BgSB3UgdDgdWVlbEUbbb7ZLCY7UDF1t2Zrx06ZJcl+YNMVKvVCro9XpIp9Oi1spSY6IZ5NJw/PD7HMuaG1Gr1WSe05oGlGfvdrsiSlStVg2EYqYweD+4eJOUSESDHA4ABr0XiqK53W4hODqdTpTLZTz99NO4//77DcJUVJBlAEdOBRGQg0oxkEsBbFcW1M+fnpvMz7V+lvUzbt5Ol1Hupkyq8/pawdEcPOn5Sd8fcxdEvZ1+beYgaN6Bvg7zPKidNZ0uNt8/vd1u56evS5eMmzkcTHPvhjLv20FgCkATb7QaGQec3W6XHgScPBg5sy/6+fPnEY/HEYlE5CJnZmbEwwe2CBjJZBLr6+vIZrPyQzAi4GRElrFm9bJ0Edhc1KgdEIlEsL6+Ljm6fD4vJL9IJCJCRCwjZHkjYcHl5WUZ2OFwWJo8MSXSaDQk58iJcHh4GKOjo+IoMacYjUaFZUwH5sqVK5iYmBDUwuVyYXl5WVINHBwcNIR0OWlwkvxJJPJZ9n/PuBABmxMR5cCZ8uJirDujulwuUVbkOHS5XJidnZUFlETFubk55HI50RkgQfPGG2/E7OysjH8uyJRijsViwknQCATLLOnEs+cJKxdY/UC5Yz2pM+UIbCkWskKA2/MYgUBAOFCc04gO8LXX68X4+LgsKFRSZeoWgCAS9XrdILXs9XoxMzODd955R7rB8n7qeZDH0iWclll2LXtXHIRwOCy5cwr30ClgXwSHw4FwOCwPJfNx1BAIBoNYXFxEIBDAysqK7GNiYgLDw8NYW1sTOI4yzYcPH5b8P6EzHS0QzdBlVayooDNAJKNarQoUyjIiohCDwQDBYFAIVaVSCZlMRsoJFxcXEYlEJNr3er0ol8si2ESvn42gCIGePn0aJ0+eFGU4DnCiF7lcDolEAvV6HUeOHJFcI6s4NjY28Morr2BqasrABidJlNepSxzNXq1llh2Esasgy57ZdZV5cMLq2rmnsQKASCPJxHSa2+02gsEggsEger2eIEz5fB6vvfYajh07Js6G1gbgfER+FNMU5COR1MwonvOB1lDgXMayynA4LHwKjnc6ECyPHBoaks6urETg/rRyJJUfb7rpJkE7iTY4nU5UKhWDw8GyzXq9juXlZTSbTVQqFbjdbjzzzDP47d/+bUm9MmhiioRdYIniWCRFy/Zi+3IQOPiBrXpeQkqMlgl9ARDmMWWOqfbHyYCtodPpNAqFAvL5PEqlEsbHxxEOh9Hr9XDhwgUcO3YMi4uLOHbsmHjVZi11RhpcmMliJhFQi7MwgqGmeqvVErIhABFNIjQajUZlUWdZIdEFKrPxujihtNttad7idDpx66234md+5mcwNDSEK1euCHxJ0RaPx4NsNit66ysrK0J+1GmL+fl5mTg52WkoV090VJI7COPvBwCHDx82fLZTyZEZWtMKiTupogHGcj69D63EaN7HQZoZkttJuUzDmebv7fV30ddkvn87qbiZy574XB9U1EjnVD93Wq2UYzMYDMq44jkSVSR3ptlsIplMotfriYPBqiW32y0dWqk7sri4iMnJSUndsSwS2F5mxsWR84JOMxBtBCBwvFY21AJKLEtkqpKpC3KxmLIAYBBlYpqAKZVWq4Ubb7wRN910E8LhMEqlkhwPgEGqGdhELqixQi0Wkp4BGNI4/E04X/PZYSB3UPPC0NAQMpkMAGMZs/n30POCuTRYP8v5fF5em9N3q6ur8lqPDfP41M+9PpY55aZLqDUcb0556vJNvZ35/PT41K/N5Z/aWdP7MM8fO5VUmufLncohzcqWe5kX9vWUUATI5/NJPT4nBLZk1ogCIwKKnnByYF6aP8jy8rKQBLnPbreLZrOJbDaLWq2GdDqN1dVVKWeip14qlWRRJdGpWCwKDFitVmVg0/vWuTlC+nRwWHGg+xywwxwjIAorMTe4srIi9dlUdKMxEpicnJSoIxqNGhwWohuEKunAsJPj4uIiOp0O1tbWBC1g+2l9XQ6HQ/Ti6ThYUKJlPw5jnp7EO6J1jGgZYbPKh2OLCyYndZ3DJmEXgCAKvV4Ps7OzGB4eht/vRzwex+rqqowd3ZOAzjobmrETK2XOSQ6kg00Hg4u0zqXzvCnGpJVfuT3f17wGkhDpKOmSz16vh5MnTwoxko4DJ3/eG4olEbUtl8tYX1/H2tqadNBlmoRzFf94Dfo8rcomy/Zq+0IQCNkvLCxgfHxcSDh8OAnljYyMSGQ+NzeHfr+PbDaLUCgkkJ7T6UQikcDy8jImJibEaVheXsZgMJCui6xRHh0dxeuvv46f+7mfE/UzwnTUQmCesd/vC/GPxECqJgKbERZrjlnqCMCgjkingnXKnOhYVcD+DEQOVldXJadJZ4gpi2g0Kgs2UzBETAAII5u6DaVSCaVSCYuLi1LyxAktlUoJdEhCIlMijCo8Hg/y+fyuut6WWXa9zOHY6qnC3gQkLFNZ1e/3I5VKYX19Hfl8XjRNNMmZuXouYKxoouw454h6vS5RcCaTwaVLlzA6OiqpivX1deltsrGxgYceeshAbr1edi3ne6fPiT4AwL/8y79sg/uv9j393tVe2+12/M3f/M2ezjuZTFqpR8v2ZPvmIHCxa7fbSKVSqNfrWFtbk+jZZrPh4sWLAv/GYjH4fD4pMwIgegnxeFwWWpvNhvn5eXS7XWGnc7GuVqs4e/YsUqkUotGokJCIXDQaDQSDQUMVRTgclsggl8shm80CgBCPCPlQS50kRCq/MZIn7ElHgAgGyVaESFOplJRjEgmhjsENN9yAQqEgEVa328Xa2ppEPmxQUyqVpCkL+Qe8Z4x0PvjBD0oukiWUtVoNpVIJXq8X9XodsVhMfqeDghLdbjeOHz8OAEin04bPNFSn0wDmiVBDY6wEAbbD7/ozfSzNXAa2hHmuto/raVQmpJmbI9HMKRDNKDaf+15st8hPX6+GXgHgxIkT27a5nsZxRo0BIlzaMefzraWXqQXC8QZAUmp0yDudjvRYIDLHIKPb7eKNN95AMpkUcTZ2ciTCRjn2UqlkWJi1XW2B5nvm1ztd/9W24eudFnetNKtlo83nqcsSr4YKDgYDQyC0mw0GA0QiEZkPr7d5vV7cdNNNcixtOmAxjw1t2pkrFApXfR8wKiHu1jRJz0c6ZWdOxen5Y6cKIcA4DnVKwOx07ZT22K2JlU4rmFMgev+7VU/olIg+rnme4vy7m5jevhwEQv+RSETajlJ3gLm2aDQqN5SkIXY3Y6lgs9lEt9vFwsICotGoVBQQliOUNzExgVdffRUTExNYWlrC2NgYZmdnpUskSUIkBgKQ7meMpuv1OhKJBIaHh9FoNOB2u9Fut7G+vi6lhoTpKaCkmzVxm8XFRSFitdttVKtVgxIaOQfcf6/XQ7VaFfhwdXVVBvDy8rIopwWDQczMzIjT1O120Wg0hFzFc+10OshkMrj99ttRqVSEsJXP56VlLJ0dyrvq/KVllh2U6Zw5ybdMNfKZZ1qB/AQuaBy7RARIdCQPgJMXnY+xsTEAwPz8vJCNO50OCoWCyJ53u10JEBh8AFcv7dvL+Nht0b2aE7Af4xyi04J8f6/715yC3a5nJwfJMst2sn05CJQJDQaDyGQyyOfzsqCzTpkCQH6/X/LsjBw4cIvFIrxeL6rVqtQCsxFRJpPBDTfcgI2NDbz11ltIp9O44YYbkM/nsba2htOnT4uIEc3r9aLZbMp+mP9kfTLlV6mIRuSDcs/hcBj5fF6iG5/PJ7LJDocDCwsLBn1xkiHpLPj9fsmLMvonsY4cBWBTX4EVINRsn5ubw8WLF6VDZTAYxOHDh6W0EoCUQv7e7/0eVlZWpLENGd3MeTL9QSEXttm2zLKDNC7E4XAYQ0NDKBQKQqTl53TQe70exsbGJDfP1BoALCwsSETGVBlz95lMBseOHUOtVsPFixdx+PBh6U2ghY60ABLRQr24vvjii7jvvvswPj6OYDCIr3/965iensb9998vY+q5557DU089hYcffhhvv/027HY7fvEXfxHPPvssvvjFL+KZZ55BOBzGkSNH8JWvfMVwH7Tdc889yOVy6PV6ePzxx3Hs2DH8yq/8CsrlMlwuF5544gmMjo4KL6Db7eIzn/kMcrkcbrnlFjz88MPo9/sYGRnB1NQUAODJJ59ELBaTY9HZAoDPfOYz+NM//VOcOXMGjzzyiMi8P/nkk0ilUgYCpJVisGwvtq/VYzAYIJ/PY3V1FcPDwyJAUq/XMRgMcOXKFemzQH30SqUiBKZGo4Fz586h0Wggl8tJdFCpVJDL5URzwGazYWVlBTabDaOjo0Ig0nrsumNjs9mUKCKfzwshqVarodVqSYkglRzJXSA0Oj8/L2xg1jJz0JZKJcTjcUkzkM1MfXXyDrSUMyFUOkutVgvFYhG1Wg0rKysolUooFAq4eGeuGdwAABS6SURBVPEiXnrpJbz88svI5/Mol8tYWlrC/Py86LWzpvkTn/iECER5vV6USiWBUW02mwjIkCRpdXO07MdlHAtLS0vodrvioFLgiKk/ljuzYRpRt7W1NSl15oLPlBq7mw4NDUlvFyKXwGbQohdZ/gFbnCKz3X333Xj++edx++2345vf/Ca++MUv4oknnsALL7yAZ555RhZcj8eDp59+etv3H3nkEbzwwguw2Wx48803d7wvX/va1/Cd73wHf/Znf4a//du/BQA8+uijeP755/HAAw/gr/7qrwxOxTPPPIPTp0/jueeeQ7PZxOuvvw5gM0X0rW99C88++yyi0ajhOzuN8QceeAD/8R//gd/8zd/EN77xjW2fW+Rly/Zi+1pBCAuy9CgcDiMajUq3NbvdLhoBXLCBzRxLPp8XwtLQ0BCSyaRULPh8PmSzWbzxxhvw+/3SZ+Ho0aMAgCtXriAcDuO2224Tsl40GsXy8jISiYQgBiMjIyJn6vF4kMvl4HK5pAU01bsYcVBApVwu49ChQ6LLzj4SXHx1J8lutysVDPw+F+iFhQXRdydPodvtShTBfKrb7cby8jLefPNNUXS02+0IBoPw+/0iU00HYXx8HJOTk6LExhQDiZRaHpYkL+ayDgpBCAaD+PCHPwxge3mhLl/UPANz1KJzkpSRBoy5QPM+9GdaYREw5tt24zT8qPfErHCmz0nnNTXnADDmK3XeUedIAeN90ccy50x1Hldfu7mbG491UIsC5wSigWNjY/K8M72gI3q73S6iR1QndLlcIiVMzYBut4tUKiWy7EzZMRU5MzODsbExZLNZ2O12lMtlRCIRSRcSReS1m+H1qakpnD17Fl6vF//+7/+ObDZryN/ee++9+OpXv4pf/uVfvup1ax7J5z//efzFX/yF4ThcvOv1Ok6dOgUAGB8fl3tGvQJuf/nyZZw8eRLdbhdTU1N47bXXMDU1henpaXz4wx/G7bffjoceeki2t9vtmJubwz333IN0On3V7onVanXbc8PjH4R5vV5BO8w8g7W1NXmt8+vmc9HjmmqZwPbnmkqugJGfYOYqUIIbMM5V1WrVsJ25eyLNzP3R93OnEkrAeI27lThrXpU+J7PyoZ4/9LNnvi/aadR8B/N2WipgJ9uXg+B0OnH69GnMz8+jWq3KSY6OjiKRSMjieeXKFQCbkDotGAwim83i6NGj+N73vodSqQSfz4dkMonFxUVxJi5duiR9GM6dO4dyuYybb74Z0WhUyoRI7MlkMiLRzPIiv98Pm82GpaUlIe9Q3pULvpaLLhQKIuhUq9XE0aG0Kp0NOh61Wg3Dw8OSflhZWZHoxePxoFAoGNjbJBDS+WCaptPpYHR0VBQWfT4fQqGQAdno9Xq49957MTU1JQqOFIRhKoECNOwxT+SAaRcrUrDsoM3lcuH48eOYnZ2F3W6XhSCTyWB4eFiIzLlcDhsbG3jzzTdht9sRDofh9XoRjUYBQBowuVwuJBIJFAoFGVdE6YDNVESv18PU1BRGR0dFdZBONZsv6Yn4apPg9773Pdx666349Kc/jYceeghf+tKXcOedd+Kv//qvAWxOqEePHsV///d/G8bRH//xH6NSqeADH/gATpw4gcFgIM4BsEUk7HQ6+MhHPoLl5WV84xvfMJQ3Pvroo/jLv/xLA+nw6NGjeOmll/ChD30IL730EiYnJ2Gz2XD27FlEIhH8/u//Pp599lncddddUnL92GOP4dFHH8Vtt92GW265Rfb12GOP4R/+4R8wPz+Pl19+edt9sOYFy/Zi+3IQ6B0nEglpMDQYDMQRyGazmJiYQDabFTi+2+0ik8kIDPnKK6/A4XDA7/cjFArB4/FIL4RMJiODmov65OSk6JvzHJjrJDmHaYZQKIRer4dAICAiKwCEAEmSJD0s1i6HQiHMz8/DbrdLx7S1tTWBS2u1mqQTdFlRsViE2+2WaJ1d21hrzTIsAIIikFTJcyehstPpyHFsNhv8fj8++9nPIpVKoVKpSIUFsNVeliqKRBYoBc3eEUR8LLPsoM3pdOLw4cNotVoiYsSqJPJqNIuajcrIr3n99dexsbGBSCSCeDwu6QGWSLNqodPpYHJyEqFQSIh9JDMy/cf0no7OaYPBAF//+tfx8ssvY3JyEh/96EfhcDjw+OOPYzAY4HOf+xyef/55Gee/+7u/i4cfftiwj0ceeQTvf//78clPflJ4T+bj0NF54YUX8IMf/AAPP/wwnnrqKQDAn/zJn+Duu+/GkSNHDJHlXXfdhRdffBEf/ehHMTY2hmQyCQDiQH3sYx/D2bNn8Uu/9EvyncuXL+Pmm2+W4I3n8cADD+C3fuu38MYbb+DBBx/E3/3d322roLDMsmvZvpPUbIdKVbN2uy2RwYULF+ByuTA8PIzh4WFZoJeXlwFAdAWi0SiOHDkiC67X6zXk0UlMisfj0pSJokfhcFicgmKxKLALxZiazSbq9TpyuRySySRKpZLUU+vUAlEAn8+Hy5cvo9/vi+PBCMjhcKBQKIhyo+7SSGdgaGgIlUoF5XIZHo8HpVIJnU5HonibzYbx8XEcOnQIhUIB9Xpd1NjC4bBoLhAJYTvs973vfYhGo+JUUI0O2IS0/X4/arUaotGopDvYdY4lZtqhud5GARcAMpHRdEpAK6GZz0XDuXSkrradttnZWXl94403Gj7Tym3nz5+X12aoTpdH6UYne+VsmEsUNeSom9DsptSmSyXNKRouCMDuEKZeXPRvYIaU6VgedNTI55cQf61WEzEykv0Y7QMQpx3YHGvpdBqpVErGFbk81EJhOSCrojiedQMlBiKE73VDJY6fe+65Bw899JCc98WLF3Hs2DEJfrSo0NGjR6WrrEYIAoEAPv7xj+Mf//Efcd999227t9pxCQaDEuD8/d//PWw2Gz71qU8B2CJVApuL9mOPPQYA+NznPodf+IVfkMooh8OBV199VeB7nsfExATOnj2Ln/qpn8K5c+dkn7xWKjTqYx1kJYPL5RJlVXOKTY93Xc5nTtntpKpqVirU16FTEebUhh7/umGaucmR3p9O85lTnnqe0NuZ96dTFvqczAqT+l7o7czzqnaw9X0x32ft/OkUg9kp3Eufnn0LJfn9fpkICB+OjY3JzSGxj5MDb6DH40E6nUY6nTa0N6XMMYlJjAp8Pp/oE1BRbTAYSEmhzWYzlCYyYmdZUzgcFkeA5Y4svaIymcvlEgeAmgqVSkXyprzmxcVFxGIxUWAsl8uCjuTzeUN0A0DQi263i3g8LnC/VpQkQYtwKK/pj/7oj4SYyPInVm2QA0H+RiwWM+Q7uS/d4tUyyw7aHA6HwPuUUe52u0in00L47fV6QhxmdO12uxGNRhGNRhGLxWRx39jYMEy2lHRn7wI6DQDEKWAKkGlFADIuaVfTI7DZbPjnf/5nfPvb34bb7cbY2Bi+8IUv4J/+6Z9kof3sZz+Lu+66a9s+PvGJT+BjH/sY7rvvPtx///2GNEO73cbHP/5xOcaXvvQlAMD999+P973vffjQhz6En/3Zn8WDDz6Ixx57DL/6q78KAPiN3/gN2Gw23H333RgZGcHZs2fxO7/zO1L2rR0bm82GP/iDP8CnPvUppFIppFIp+ezP//zP8eSTT6LT6YjTcbX7YJllu9m7EkqikhkJMLVaDYFAQMSEjh8/LlAf8+QkFjEdwAWQiym1Dyg5SngxHA5LxMHIodlsIhAIiDBKv98XWeJarYZgMGjI97E3AjUNSGAipNlqtaQFM2WXqXvOVAi5ApyYKGsKGD1gQv66FJETDcsgqQ9Bx4HR0eHDh0U5kcdjKoIIA9MooVBI8peEVjmJ6tbVlln24zBWK+jIWZfcut1uxGIxWeCJcOnnm8REVhPxmec8onss0Kknx4gaIOQYEZXTDoLNZsMdd9yBO+64w1AO+eCDD+ILX/iC4Xp+7dd+TcbQz//8z4vDorcLBAI4c+YMABicA2AzenzuuecAGBdklnzr8fmHf/iH8prfob33ve/F97//fQlY9PcGgwGOHDmCV155xfDZxMQEfv3Xf/2av5llll3L9t2siVLBFPIh+54NhwgHcnLgBEBhICIPunMZc3iEBtk9LZPJyIQxGAyEqES5YZYVMq9vs9lw/Phx0TQgqkAZ4qGhIeRyOcnRk+jj9/vR6XREJ4EiQ0wBVCoVaedMIZhWq2XQKdCkRjo4RBl4bMKhrLqgoBEhq0AgAJ/Ph3K5jGQyiVarhUAggF6vh1arhVQqJS1pWf0BbJGiGE2xlHM3tbIf1dxuNyYnJwFsZ+Fr+EvD8fPz84bttOJfIpGQ1+b96evQ0LwZmtRQoobpzaxmXeGgv0MFT5pORejtzJCe/l/Dj+YmL/qcNNRpVqLU/+/WhEajRDr9YGZjHzp0CADw3e9+FwdhfN6ZGtQOsYb5OQb4rPI55pyhuxZSeI2LHtEJtmEHtlRR6XDooIDE4auhBnytIXf9mXkB30vEvdNxrrU9j6X/N3/fvI1+/2qv9XtXOxfz9V1Pc7vdkvozV1XolKCG7c0Qvp4z9LywGySux4YeW4AxDaDh/SNHjhi222le0M3iAGPVha46MDdN0vC+HsdmRUN9Tjq9Yv6NdlJINKPFevzrygXz/piK3I2nti8HgeIlzLET0gMgXdIoR0yPnxE5ux4Wi0VhGpOTQO5ANBoVGJ+Eu3a7jWg0CrfbjXq9Lq1LuSgy5cGbfunSJTku97e0tIR0Oo1KpSKqjYQ4O50OSqUSAoGARDr80Wu1msgtszEK+0DwutkngdKwhPdZhUCIlRO4rqDgPSUUeuTIEeFgMPVC5Un2eiehMRgMisOlf3zCtxRQspAEyw7a6FTrVsnmbqvA1rPPRV07A1QMpWYKuTe6NTH5B8BWAyXunwEG5x/KNV9tUdnNOTBvt9P3dtpuP47Cbgu8PtbVuET6npjfN5d2mp2fq33PMsuuZvvu5sgSQL/fL4OcaQRCfizVczgcqNVq0kudRqTB5XJJV0j2WWfenTBhIpGQtqckORWLRSEwcdKhF0qNBZ2O8Hg8Iti0urqKXC4nDke324Xf7xcNeEL8bCjldDpRrVZlkJH0yCiXzZlYgcBOlpygKJPM++JwOETngG2pmbJh50ctMNNutxEKhWSyczqdSKVS8Hg8hv4WZtNd9Cyz7KCNSp4cJ3TaOUcwimO00+v1ZEHncxoIBETLg2RGli3TkeCY9Pl8hnbM2hlnBQPFy3az3aLyvWy/m0Ow333p75lRhb1G/WYS4k7HsKqbLNuL7ZvJFg6H5eFifpBePgc/5YhZcsj8IHOKHOiEFympynQEF39OFkQu/H6/oQMcpYQpFkQxpI2NDQQCASwvLyMej0sKo9VqifNSq9WEOMX8Jwl+Xq9Xygrr9brA9lRF7Pf7QiwkOYtiMSRJ8RrD4bDcs0AgYIh6eA1ECuLxuDCeNbOVlQ5sqc1SyN3M6/XKuVpm2UEaK3aIDJAzwOecDgDLcTmuNfpos9kM/CMAksLkmOKcwKCE6TmmF4g+MHjR6Y29GpHJ3RwBvd1On5vfu1r1gF74d3MA9hPt74aIWGbZfm3fHARC8yT6AZCSRjoB4XBY+gJQewDYYtq32234fD5DGRgheEYejJiZryeRiVEzIxM6HW63W0r/6LgcOnRIShMpMFQsFiUXzGNQVIWfVSoVyXkCW+WOPA+SLQn/kzTFvgy9Xg/RaFQWdOaB6DBxUmNZpIZRKcxEhIFwKfO7nU4H4XD4mpMGnaqDqmbweDw4ffo0AKMKImDMsenXOp8IGHODFNcCjEqMgDGnqPOVZiU0naPT25m7p+lSRDb/AYyd4wDg7bffltc6J2kuZzJzDWjme69zmfoczApnWmBM8xHM2+l8pc47knNAI0fiIKNGjkGOG45PrUvAxUvPB3Rg6WSzEsfv94vDoTkMdAJ0IyiWAfIa6XzTSYhEIgb+x37tWqkI87a72bV4A3o/3JfmYuy2v70a05UHYU6nU543/YwDxnPVY9I8nvT3WCIPGJUYAWM5rx6f5mvT40bzG8ycAW0aeTJzevS4262LpB5veq4zn5+eF/SaaC7r1M+WPj/zdjvN+eby5+uupEiojwPb7XZLvpE/KksKNzY2JOJnnwMNk1MhrdVqCSxIyWC2gPb7/YbInG1cgS0InYuny+WSigBWSng8Hvh8PlSrVeEOUHEwGAxK0xSmCZgS4Ws6N4yCeB2UiNZkOsoxBwIBuQ4u0PV6HclkUkhvTA/U63U4HA7hZzD1QqhVpyFISmSvimsZ769V7mjZj8P0GNRR/dVajnM8sRRYcxi0ciIAQ0dHTZLmpE1Sc6fTkbQijcf9/Oc/L2OWPCN+xn1qFILzAasg+EdeERFE/T7nuXq9LulJpheZPmFaJJFIyNzEFKsOADY2NpDP5zE9PQ2n04lPf/rTGB0dlcCBOiput1s0Z/jHKhIuTrx/mvtkt9vx/ve//2AfCMv+T9i+Vw8++ITvPB6PLNL0tIPBoJCGOEBITCSEzoEYDAZF6Ie5evYr4DaE5wHI/geDzQ5yjUZDtBBcLhcWFhYMwkpMd1Cbgd0W+T6dFaYcqtUqNjY2hHfAdASlkgkFlstlSQMwiuFiTpIV+Qh0GNxuN0ZGRuB2u0V+mZOL3W7HpUuXEIlEZED7fD70+32EQiF0u11JxezHLLjRsoM2jgkumlx8yath2oFpR6J2fO6ZKqBjzuCDyJ9OkxFp1CW+bB+v0wOs/OFizvHPUsx2u22oRCJfgY3d1tfX0Ww25a9er2N9fV3mAu082Gw2ec1jsvQSgJRjE9VgyoX3ieqtDodD5o5cLidpmpmZGbkvrIgaDAaiqMr9UlqdKCfRV+0c0KGxzLK9mG0/C4jNZssDmL3mhpb9JNnhwWCQuPZm+zPrWfhfadazYBnNehYso+34LOzLQbDMMssss8wyy/7/MCtBbZlllllmmWWWbTPLQbDMMssss8wyy7aZ5SBYZplllllmmWXbzHIQLLPMMssss8yybWY5CJZZZplllllm2TazHATLLLPMMssss2ybWQ6CZZZZZpllllm2zSwHwTLLLLPMMsss22aWg2CZZZZZZplllm2z/wdZjqNNWqnG9AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 648x288 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fpointer        = 3\n",
    "filename        = 'samples_%03d.mat'%(fpointer)\n",
    "y               = sio.loadmat(os.path.join(sample_path, filename))['y']\n",
    "y               = torch.from_numpy(y).type(torch.FloatTensor).to(device)\n",
    "HTy             = torch.t(torch.mm(torch.t(H), y)).to(device)\n",
    "\n",
    "[x_test, loss_layers_sym] = model(H, HTy, 33, 33)\n",
    "\n",
    "fimage          = 'gray_%03d.mat'%(fpointer)\n",
    "I               = sio.loadmat(os.path.join(data_path, fimage))['I']\n",
    "[Iorg, row, col, Ipad, row_new, col_new] = imread_CS_py(I)\n",
    "x   = img2col_py(Ipad, 33) / 255.0       \n",
    "x   = torch.from_numpy(x).type(torch.FloatTensor)\n",
    "x   = torch.t(x).to(device)\n",
    "\n",
    "\n",
    "#x1 = 110\n",
    "#x2 = x1+33\n",
    "#y1 = 99\n",
    "#y2 = y1 + 33\n",
    "#wd = 3\n",
    "\n",
    "#x3 = 120\n",
    "#x4 = 150\n",
    "#y3 = 180\n",
    "#y4 = 240\n",
    "\n",
    "\n",
    "x1 = 132\n",
    "x2 = x1+33\n",
    "y1 = 132\n",
    "y2 = y1 + 33\n",
    "wd = 3\n",
    "\n",
    "x3 = 120\n",
    "x4 = 150\n",
    "y3 = 180\n",
    "y4 = 240\n",
    "\n",
    "I[x1:x2,y1-wd:y1]=255\n",
    "I[x1:x2,y2:y2+wd]=255\n",
    "I[x1-wd:x1,y1-wd:y2+wd]=255\n",
    "I[x2:x2+wd,y1-wd:y2+wd]=255\n",
    "\n",
    "plt.figure(figsize=(9,4))\n",
    "\n",
    "plt.subplot(1,4,1)\n",
    "plt.xticks([])\n",
    "plt.yticks([])\n",
    "plt.imshow(I, cmap='gray')\n",
    "\n",
    "plt.subplot(1,4,2)\n",
    "plt.xticks([])\n",
    "plt.yticks([])\n",
    "plt.imshow(I[x1:x2,y1:y2], cmap='gray')\n",
    "\n",
    "\n",
    "Prediction_value = x.cpu().data.numpy()\n",
    "I = 255 * np.clip(col2im_CS_py(Prediction_value.transpose(), row, col, row_new, col_new), 0, 1)\n",
    "\n",
    "Prediction_value = x_test.cpu().data.numpy()\n",
    "Io= 255 * np.clip(col2im_CS_py(Prediction_value.transpose(), row, col, row_new, col_new), 0, 1)\n",
    "\n",
    "loss_discrepancy    = np.mean(np.power(Io - I, 2))\n",
    "psnr                = torch.mul(10, torch.log10(torch.div(255*255, loss_discrepancy))) \n",
    "\n",
    "Io[x1:x2,y1-wd:y1]=255\n",
    "Io[x1:x2,y2:y2+wd]=255\n",
    "Io[x1-wd:x1,y1-wd:y2+wd]=255\n",
    "Io[x2:x2+wd,y1-wd:y2+wd]=255\n",
    "\n",
    "plt.subplot(1,4,3)\n",
    "plt.xticks([])\n",
    "plt.yticks([])\n",
    "plt.imshow(Io, cmap='gray')\n",
    "plt.text(100, 235, 'PSNR: %.2f dB'%(psnr), fontsize=8,\n",
    "        bbox={'facecolor': 'white', 'alpha': 0.95, 'pad': 3})\n",
    "\n",
    "plt.subplot(1,4,4)\n",
    "plt.xticks([])\n",
    "plt.yticks([])\n",
    "plt.imshow(Io[x1:x2,y1:y2], cmap='gray')\n",
    "plt.savefig('test.png', dpi=300)"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "authorship_tag": "ABX9TyPyLR2hGbCHPKnrkw3hANer",
   "collapsed_sections": [],
   "name": "LADMM_TestCSI.ipynb",
   "provenance": [
    {
     "file_id": "1C86j5nzwPWq3IsqnijPhfiet_YCIAepF",
     "timestamp": 1614076785431
    },
    {
     "file_id": "1s39Kdq7BAIrAzSXuh8y0scz7g-oKr4HH",
     "timestamp": 1614030180900
    }
   ],
   "toc_visible": true
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}