CNN_buy_sell_classifier.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CNN v3.0\n",
    "\n",
    "Using CNN to try to buy and sell movements"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import json\n",
    "import time\n",
    "import pandas as pd\n",
    "import math\n",
    "import matplotlib.pyplot as plt\n",
    "from cnn_prediction.cnn_data_loader_buy_sell import CNNDataLoaderBuySell\n",
    "from cnn_prediction.cnn_model import CNNModel\n",
    "import numpy as np\n",
    "from core.utils import get_accuracy, plot_loss"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load config"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "configs = json.load(open('predict-cnn-buy-sell-EURUSD-config.json', 'r'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "if not os.path.exists(configs['model']['save_dir']): os.makedirs(configs['model']['save_dir'])\n",
    "pre_trained_model_filepath = configs['model']['pre_trained_model']\n",
    "# pre_trained_model_filepath = None\n",
    "generative_training = configs['model']['generative_training']\n",
    "img_size = len(configs['data']['intervals'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "cols = []\n",
    "for indicator in configs['data']['technical_indicators']:\n",
    "    for interval in configs['data']['intervals']:\n",
    "        cols.append(indicator + '_' + str(interval))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataloader = CNNDataLoaderBuySell(\n",
    "    pd.read_csv('data/duka/' + configs['data']['filename'], index_col=0),\n",
    "    indicators=configs['data']['technical_indicators'],\n",
    "    intervals=configs['data']['intervals'],\n",
    "    df_test=pd.read_csv('data/duka/'+ configs['data']['test_file'], index_col=0),\n",
    "    window_size=10\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Open</th>\n",
       "      <th>Close</th>\n",
       "      <th>High</th>\n",
       "      <th>Low</th>\n",
       "      <th>Volume</th>\n",
       "      <th>SMA_2</th>\n",
       "      <th>SMA_3</th>\n",
       "      <th>SMA_4</th>\n",
       "      <th>SMA_5</th>\n",
       "      <th>SMA_7</th>\n",
       "      <th>...</th>\n",
       "      <th>AROONOSC_9</th>\n",
       "      <th>AROONOSC_11</th>\n",
       "      <th>AROONOSC_13</th>\n",
       "      <th>AROONOSC_17</th>\n",
       "      <th>AROONOSC_21</th>\n",
       "      <th>AROONOSC_34</th>\n",
       "      <th>AROONOSC_41</th>\n",
       "      <th>AROONOSC_55</th>\n",
       "      <th>AROONOSC_77</th>\n",
       "      <th>AROONOSC_99</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>time</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>2004-01-09 04:30:00</td>\n",
       "      <td>1.27597</td>\n",
       "      <td>1.27508</td>\n",
       "      <td>1.27617</td>\n",
       "      <td>1.27494</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.000465</td>\n",
       "      <td>-0.000610</td>\n",
       "      <td>-0.000728</td>\n",
       "      <td>-0.000740</td>\n",
       "      <td>-0.000857</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>-0.909091</td>\n",
       "      <td>-0.769231</td>\n",
       "      <td>-0.588235</td>\n",
       "      <td>0.428571</td>\n",
       "      <td>0.705882</td>\n",
       "      <td>0.707317</td>\n",
       "      <td>0.527273</td>\n",
       "      <td>0.376623</td>\n",
       "      <td>0.292929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2004-01-09 05:00:00</td>\n",
       "      <td>1.27510</td>\n",
       "      <td>1.27304</td>\n",
       "      <td>1.27542</td>\n",
       "      <td>1.27240</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.001020</td>\n",
       "      <td>-0.001670</td>\n",
       "      <td>-0.001988</td>\n",
       "      <td>-0.002214</td>\n",
       "      <td>-0.002457</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.888889</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>-0.846154</td>\n",
       "      <td>-0.647059</td>\n",
       "      <td>-0.523810</td>\n",
       "      <td>0.676471</td>\n",
       "      <td>0.707317</td>\n",
       "      <td>0.527273</td>\n",
       "      <td>0.376623</td>\n",
       "      <td>0.292929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2004-01-09 05:30:00</td>\n",
       "      <td>1.27280</td>\n",
       "      <td>1.27310</td>\n",
       "      <td>1.27400</td>\n",
       "      <td>1.27114</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000030</td>\n",
       "      <td>-0.000640</td>\n",
       "      <td>-0.001208</td>\n",
       "      <td>-0.001542</td>\n",
       "      <td>-0.001934</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>-0.923077</td>\n",
       "      <td>-0.705882</td>\n",
       "      <td>-0.571429</td>\n",
       "      <td>0.647059</td>\n",
       "      <td>0.707317</td>\n",
       "      <td>0.527273</td>\n",
       "      <td>0.376623</td>\n",
       "      <td>0.292929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2004-01-09 06:00:00</td>\n",
       "      <td>1.27291</td>\n",
       "      <td>1.27338</td>\n",
       "      <td>1.27432</td>\n",
       "      <td>1.27260</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000140</td>\n",
       "      <td>0.000207</td>\n",
       "      <td>-0.000270</td>\n",
       "      <td>-0.000742</td>\n",
       "      <td>-0.001299</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.888889</td>\n",
       "      <td>-0.818182</td>\n",
       "      <td>-0.923077</td>\n",
       "      <td>-0.705882</td>\n",
       "      <td>-0.571429</td>\n",
       "      <td>0.617647</td>\n",
       "      <td>0.682927</td>\n",
       "      <td>0.527273</td>\n",
       "      <td>0.376623</td>\n",
       "      <td>0.292929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2004-01-09 06:30:00</td>\n",
       "      <td>1.27352</td>\n",
       "      <td>1.27312</td>\n",
       "      <td>1.27371</td>\n",
       "      <td>1.27160</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.000130</td>\n",
       "      <td>-0.000080</td>\n",
       "      <td>-0.000040</td>\n",
       "      <td>-0.000424</td>\n",
       "      <td>-0.001124</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.666667</td>\n",
       "      <td>-0.818182</td>\n",
       "      <td>-0.846154</td>\n",
       "      <td>-0.705882</td>\n",
       "      <td>-0.571429</td>\n",
       "      <td>0.588235</td>\n",
       "      <td>0.609756</td>\n",
       "      <td>0.527273</td>\n",
       "      <td>0.376623</td>\n",
       "      <td>0.292929</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 230 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                        Open    Close     High      Low  Volume     SMA_2  \\\n",
       "time                                                                        \n",
       "2004-01-09 04:30:00  1.27597  1.27508  1.27617  1.27494       0 -0.000465   \n",
       "2004-01-09 05:00:00  1.27510  1.27304  1.27542  1.27240       0 -0.001020   \n",
       "2004-01-09 05:30:00  1.27280  1.27310  1.27400  1.27114       0  0.000030   \n",
       "2004-01-09 06:00:00  1.27291  1.27338  1.27432  1.27260       0  0.000140   \n",
       "2004-01-09 06:30:00  1.27352  1.27312  1.27371  1.27160       0 -0.000130   \n",
       "\n",
       "                        SMA_3     SMA_4     SMA_5     SMA_7  ...  AROONOSC_9  \\\n",
       "time                                                         ...               \n",
       "2004-01-09 04:30:00 -0.000610 -0.000728 -0.000740 -0.000857  ...   -1.000000   \n",
       "2004-01-09 05:00:00 -0.001670 -0.001988 -0.002214 -0.002457  ...   -0.888889   \n",
       "2004-01-09 05:30:00 -0.000640 -0.001208 -0.001542 -0.001934  ...   -1.000000   \n",
       "2004-01-09 06:00:00  0.000207 -0.000270 -0.000742 -0.001299  ...   -0.888889   \n",
       "2004-01-09 06:30:00 -0.000080 -0.000040 -0.000424 -0.001124  ...   -0.666667   \n",
       "\n",
       "                     AROONOSC_11  AROONOSC_13  AROONOSC_17  AROONOSC_21  \\\n",
       "time                                                                      \n",
       "2004-01-09 04:30:00    -0.909091    -0.769231    -0.588235     0.428571   \n",
       "2004-01-09 05:00:00    -1.000000    -0.846154    -0.647059    -0.523810   \n",
       "2004-01-09 05:30:00    -1.000000    -0.923077    -0.705882    -0.571429   \n",
       "2004-01-09 06:00:00    -0.818182    -0.923077    -0.705882    -0.571429   \n",
       "2004-01-09 06:30:00    -0.818182    -0.846154    -0.705882    -0.571429   \n",
       "\n",
       "                     AROONOSC_34  AROONOSC_41  AROONOSC_55  AROONOSC_77  \\\n",
       "time                                                                      \n",
       "2004-01-09 04:30:00     0.705882     0.707317     0.527273     0.376623   \n",
       "2004-01-09 05:00:00     0.676471     0.707317     0.527273     0.376623   \n",
       "2004-01-09 05:30:00     0.647059     0.707317     0.527273     0.376623   \n",
       "2004-01-09 06:00:00     0.617647     0.682927     0.527273     0.376623   \n",
       "2004-01-09 06:30:00     0.588235     0.609756     0.527273     0.376623   \n",
       "\n",
       "                     AROONOSC_99  \n",
       "time                              \n",
       "2004-01-09 04:30:00     0.292929  \n",
       "2004-01-09 05:00:00     0.292929  \n",
       "2004-01-09 05:30:00     0.292929  \n",
       "2004-01-09 06:00:00     0.292929  \n",
       "2004-01-09 06:30:00     0.292929  \n",
       "\n",
       "[5 rows x 230 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataloader.dataframe.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 95%|█████████▍| 140000/147376 [00:00<00:00, 147985.78it/s]\n",
      " 91%|█████████ | 10000/11043 [00:00<00:00, 129126.63it/s]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "((78543, 15, 15, 3), (78543,), (5640, 15, 15, 3), (5640,))"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train, y_train = dataloader.get_train_data(cols, window_size=15)\n",
    "x_test, y_test  = dataloader.get_test_data(cols, window_size=15)\n",
    "x_train.shape, y_train.shape, x_test.shape, y_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA00AAANLCAYAAACdWnYxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzdebhlV1kn/u9LVUgMCQmTUSAQZgQCIcyIbRhkaJsgGgEbFEFpocGg/BRF5qkNQzcO0AyigMogAbSZDAmjEkgMkDmABAizQghDSMxUtX5/7H2pQ3nvSqVWKudW1efzPPepe/fe797rnNzz3v3da5+Taq0FAACA1V1t2QMAAABYz4QmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaNpGVfWxZY9hUVV9uKo+W1WnzF8/vtX6I6qqVdWd55/3qKo3VNXpVfXpqnrawrZPrqozqurMqvqdheXPr6rT5v0fW1XX345xnlNV1x15rJez/6qqP6uqs+exHrqjjgXr0TrsTVevqtdU1b9W1Weq6pfm5U+pqrPm1+kHqurGCzUvmnvQGVX18IXlb5z73BlV9VdVtce8/NZV9fGquriqfm87x/mc7a29Asd44Dz+s6vqD3fksWC9WYe9adXzpqp62cKyf62q7y7U3Gg+//n03L8O2mqff15VP1j4ec19XYFxvr6qjtj+R7pNx3h0VX1u/nr0jjzWrmTjsgews2it3XPZY1jFI1trn9h6YVXtm+TIJCcuLP7lJHu21g6uqr2TnFVVb06yT5LHJblrkkuSHFNV72mtfS7JS1prz5z3eWSSZyV5/A59RFfcg5LcYv66W5JXzv/CbmEd9qanJ/lma+2WVXW1JNeel5+c5M6ttQur6glJXpzk4VX180kOTXJIkj2TfKSq/rG19v0kb0zyqLn+TUl+M9Nr/LxMPe4XrqoHdUVV1YYkr0jyc0m+muSkqnpna+2s5Y4MrhrrsDclq5w3tdZ+d+X7qvrtJHdcWP3XSV7YWjuuqvZJsnlh2zsn2f8K7GtdqKprJ3l2kjsnaUk+Ofem7yx3ZOufmaZttHIloaoOq6qPVNVb56sIR1XVI6vqX+ZZnJvN2z24qk6sqpOr6v1VdcC8/HpVdVxVfaqqXl1VX1qZiamqR837OWVet2E7h/v8TCckFy0sa0muUVUbk/xYpoD0/SQ/leSE1tqFrbXLknwkyUOTZD5pWXGNeR9rPT/7VNXr5ufgtJWry1tt8w9V9cmaZrT+x7xsw3xV5Yy59nfn5UcuXJV+S+exPiTJX7fJCUn2r6qfvLwnCHYV67A3PTbJHydJa21za+3c+fsPtdYunLc5IckN5+9vk+QjrbXLWmsXJDk1yQPnmvfOr+2W5F9Walpr32ytnZTk0m18jn5t7iWnVtXfrLL+cVV10rz+7fOFpVTVL8+96dSq+qd52W0XnovTquoWaxz2rknObq19obV2SZK3ZOpXsFtYh71pW/xKkjfP+75Nko2tteOSpLX2g5UeNh/nJUmeui37WktVPXV+Dk6tqqNWWf+suTedUdMMfs3L/9M5UlX9bG2Z5Tq5pgvoq3lAkuNaa+fNQem4zD2XPqFp+9whyZOTHJzkV5PcsrV21ySvTfLb8zYfTXL31todM/2xXHlhPTvJB1trhyb5+yQ3SpKq+qkkD0/y0621Q5JsSvLIyxnH6+YXxzMXXkh3THJga+3dW237tiQXJPlGki8neWlr7bwkZyT5L1V1nflE4b8mOXClqKpeWFVfmcfyrM5Ynpnke621g1trt0/ywVW2eWxr7U6Zrm4cWVXXyXR1+Qattdu11g5O8rp52z9Mcsd5X73ZrRsk+crCz1+dl8HuaKm9qapWrro+fz7BOXrlxGcrv5HkH+fvT03yoKraez4RuncWetC83z3mx3PMNj8TW2pvm2n26z6ttZXnZ2vvaK3dZV7/6Xl8ydTzHjAvP3xe9vgkfzo/F3fO1HNWozfBFuv2vGlFTbcM3yRbzl9umeS7VfWOOYS8ZCGUPSnJO1tr31jtIKvsa7VtHpRptvxuc4958SqbvXzuTbfLdMH7v83LVztH+r0kT5yfi59J8h9rHFpv2k5uz9s+J628UKrq80mOnZefnukPfjJdEf27mmY9rp7ki/Pye2XLTM4xVbUyHXrfJHfKdAtHMr04vtkZwyNba1+bryS8PcmvVtXfJnlZkl9fZfu7Zmoo109yrST/XFXvb619uqpelOlKww8yncBctlLUWnt6kqfX9B6oJ2VqXqu5X5JHLNStNs17ZFU9dP7+wEy31H02yU2r6s+TvCdbnsvTkryxqv4hyT90nodaZdmaM2Kwi1t2b9o47//41tpTquopSV6a6SQp87gelSls/Ox8rGOr6i5JPpbkW0k+noUeNPu/Sf6ptfbP2/5U/NB9krxtYcbrvFW2uV1VvSDTrTb7JHnfvPz4JK+vqrcmece87OOZeuINM4Wtz61xXL0Jtlh2b0pWOW/KdPvdikdk6hWb5p83Zgofd8x0sfnvkvx6Vf1jprc8HNY51tb7Ws39krxuZfZqjd5076p6apK9M93qfGaSd2X1c6Tjk/yfqnpjpt601gUdvWk7mWnaPhcvfL954efN2RJE/zzTFYKDk/xWkr3m5av9sq4sf0Nr7ZD561atteesNYDW2tfmf8/PdK//XZPsm+R2ST5cVeckuXuSd9Z03+1/T3JMa+3S1to3M7247jzv4y9ba4e21v5LpvcKrHYS8KYk/+mWu63G37t977BMDeIe8xWVk5PsNYerOyT5cJInZrrqlCQ/n+n9AHfKdL/tWgH/q/nRq9I3TPL1zjhhV7bs3vTtJBdmuhqcJEdner/StKOq+2Wa9Tm8tfbDsbbWXjjv++fm431uoebZSa6X5ClrPuq+bm+avT7Jk+bn5LmZn5PW2uOTPCNTjzmlqq7TWntTplmn/0jyvqq6zxr71Jtgi2X3prXOmxY9Ij96O91Xk5w832J7WaZwcmimEHXzJGfP51p7V9XZl7OvtcbfO2/aK9MFoyPm5+QvsuU5+U/nSK21ozK97/PHkpxQVbdeY9d603YSmnac/ZJ8bf5+8ZNJPprkYUlSVffPNOuTJB9IckRt+TSXa9fCp0stqqqNC/fz7pFpuvaM1tr3WmvXba0d1Fo7KNP7Bg5v05sev5zkPjW5RqZA9Zl5HyvHvFGSX8yW+3kX79U/fGX7NRybaSZqZYzX2mr9fkm+06Y3gd96Pn7mx3G11trbM93id2hNbx4/sLX2oUzT8ytXf1fzziS/Nj+uu2e6RXDV6XIgyQ7sTfN7j96VLVdg75vkrLnujklenakn/fBqcE3va7zO/P3tk9w+81XoqvrNTPff/0pr7YdvwL6CPpDkYQvHuPYq2+yb5BtzP/3h7T1VdbPW2omttWclOTfJgVV10yRfaK39Wab+c/s1jntSkltU1U2q6uqZTqLeuZ2PAXYHV/l508L6W837/fhC2UlJrlVV15t/vk+Ss1pr72mt/cTCudaFrbWbX86+VnNsksfWlvdQbt2bVgLSuTV9CMUR83arniPN/er01tqLknwiyVqh6X1J7l9V15rP1e6fLbPrdLg9b8d5TpKjq+prmcLLTeblz03y5po+Vvcjmd5jdH5r7dyqekaSY+cXxKWZZl6+tMq+98x0hXOPJBuSvD/TFYieV2R6v9AZma5uvK61dtq87u3zCcWlme6HXZn6Pmp+8W+ex9F7b9ELkryiqs7IdBvgc7PldpZkei/C46vqtEy35J0wL79BpnuMVwL80+bH9LdVtd881pe11tb62M73Znof1tmZrnA/pv80wG7vOdlxvSlJ/iDJ31TVn2S63W7lNfmSTBc/jp5vpflya+3wJHtkul04mT6c5lHzVd0kedV8nI/P69/RWnteVf1EppOCaybZXNP/KuE27Uc/vCZJ0lo7s6pemOlT+TZlmuX+9a02e2amTxv9UqbbhVbeQP2S+eJRZTpBOzXTewkeVVWXJvm3JM9b7UlorV1WVU/KdDKyIclftdbOXOM5A5Z73vQrSd4yX/hJkrTWNtX0vyX4QE0N6JO5/HOtVfe1mvlWw0OSfKKqLsl0PvNHC+u/W1V/kaknnZMpxCVrnCPV9L+JuXemc7CzsuV9o1sf97yqev7C/p63xq2BbKUu578pV7Kq2jPJpvkP6j2SvHJ+0x7A0uhNwHqkN7FemGm66t0oyVvnqyKXZPp/JAEsm94ErEd6E+uCmaZ1rqpOzDStvOhXW2unL2k8j8l//sje41trT9wVjwusbh32putkuoVua/dtrX17VzsusLp12JsOTrL1/x/u4tba3XbF4+7KhCYAAICO/u157yqJCnZWD25rfUzrLuI++hPstD64y/anE8q5E+ys7t7WPnfykeMAAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQsXHZAwDW4JIGAMC64LQMAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6Ni57AMAaNi97AAAAJGaaAAAAuoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgY+OyBwBrWnak3zxYPzr+0eMD69imZQ8AYJezIzvrsk9LAQAA1jWhCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6Ni47AEAwO5nw7IHMGjTsgcAcJUy0wQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB0bu2svGtz7siPZht38+Mt+/nd2o8/f6H//Zf/+rHvfHqzfZ7C+3z53vGX/gmxa8vF3d8v+77/s469fu/szozMs1+7+/O/I15/TagAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6NjYXbvXVTQKdozNSz7+7h7JNy17ALu6A5Y9gN1c/88H7K60fpZpw7IHsAvb3U9rAQAAuoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgY2N37d0G9/7jo5ls/8H6fQbrrztYv99g/ej499rNj9//9d7x9Xsu+fi7uhsO1i/793tn//3a2etH7ez9adm/v7uuu7912SMYdOlg/QWD9RcO1l8yWD/6+Jddf9Fg/abB+s2D9euYmSYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOio1lpvfXfl5XvDWHmOH6z/7GD9CWPl37lkrP7CsfJcNFg/OPxsHqzfa8n1ew/W7zFYP/r8XbPV4B7Wu8H+9NXBw28crB+1z2D96Atk1GWD9d9dcv3nB+t/MFh/7mD96OO/eLD+ebtufzq6BnvToC8M1n9usP5jg/VfGiu/ZPDcafTUabR+1Oipx7JduuT6G7S1z53MNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQEe11tZc+W9Va6/cBq8eKU7y7sH6uw3W7z9Yf+pg/bcG6zcN1v/bYP15g/UX/NLgDo4YrL/LYP3pg/W3Gqz/qVaDe1jf3jLWn/L+weN/cKz8y18cq98wVp7vDtafNVj/9cH6kwbr/36w/tzB+h972OAORn8BThms/+xg/aZduD+9aaw33fORY4f/07Hy3OVegzs4fLB+9OTtkMH6ax46uIP/PVh/2Fj5ewZfWjcdK88tBus33nOo/Nv1saH667S1e5OZJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6KjW2por31C19sptsGmk+Epw/pKPv2yjz//mwfprD9Y/9laDO1i2by/5+N9qteQR7FjPGOtPO71lN9hlW/bjv3CwftnjX7ZX7rr96Z6D506nDh7/poP19xys32OwftmuMVg/+vj3Haw/fbB+/8H6ZXvrYP0329q9yUwTAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0bOytPHxw59e6/uAODhis33+w/sDB+usN1u+75PrR52+vwfr9BuuvPli/52D96OPfMFi/qxv9/bjOYP3eg/V7DNZfY8nHH/39HD3+ssc/2l9Hxz/a35b9+7sL+9i9B3cw+rt5yWD9RYP1/z5Y/+3B+gvHyi/ZPFZ//lj58NN/3mD96K/PpYP1mwbrHzVY32OmCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOqq1tubK51atvXIb/NZIcZJXD9b/zGD9ffYY3MHNx8qf++nB4w969r8P7uCCwfqLBusvXfLxHzlYf9Bg/XGtBvewvp0w1p9y08Hj/9Fg/bIdPFh/+pUyiu13h8H6Awfr3z1YP/r35aDB+tHn7/8N1r9y1+1PNxs8d/r8XQcHcPfB+hsP1p84WH/OYP3rB+vPGqz/2bHyV19vrP63BqdD2uax+vPHyofrbzlYf0FbuzeZaQIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgI5qrS17DAAAAOuWmSYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJq2UVV9bNljWFFV+1bVKQtf51bVn8zrXraw/F+r6rtb1V6zqr5WVS9fWHb1qnrNvP1nquqX5uWPr6rT5319tKpusx1j/XBV3Xn0MV/OMZ5WVWdX1Wer6gE78liw3qyn3pQkVXVMVZ1aVWdW1auqasPCut+eX6dnVtWL52U/V1WfnHvNJ6vqPqvs851VdcbCz8+vqtPm3nRsVV1/O8Z5TlVdd3sf5zbsv6rqz+bedFpVHbqjjgXr0XrqTVW1d1W9Zz7HObOqjlpYt+q5TlXddeF86tSqeuhCze/O+zmjqt5cVXvNy580v+bb9vYXvWn92rjsAewsWmv3XPYYVrTWzk9yyMrPVfXJJO+Y1/3uwvLfTnLHrcqfn+QjWy17epJvttZuWVVXS3LtefmbWmuvmvd1eJL/k+SBV+JDGTY3t0ckuW2S6yd5f1XdsrW2abkjg6vGeupNs4e11r5fVZXkbUl+OclbqureSR6S5PattYur6sfn7c9N8uDW2ter6nZJ3pfkBis7q6pfTPKDrY7xktbaM+f1RyZ5VpLH79BHdcU9KMkt5q+7JXnl/C/sFtZhb3ppa+1DVXX1JB+oqge11v4xa5/rnJHkzq21y6rqJ5OcWlXvSnJAkiOT3Ka19h9V9dZM5yGvT3J8kncn+fBV/NiuCL1pO5lp2kZV9YP538Oq6iNV9dZ5ZuaoqnpkVf3LfKXiZvN2D66qE6vq5Kp6f1UdMC+/XlUdV1WfqqpXV9WXVq4oVNWj5v2cMq/bsPaIfjiuWyT58ST/vMrqX0ny5oVt75TpxX7sVts9NskfJ0lrbXNr7dz5++8vbHONJK0zjg1V9dL5OThtDmxbb/PKqvrEfHXmuQvLj6qqs+a6l87Lfnm+gnNqVf1T5yl4SJK3tNYubq19McnZSe7a2R52KeutNy30jY1Jrp4tfeMJSY5qrV08b/fN+d+TW2tfn7c5M8leVbXnfNx9kjwlyQvWOEZy+b1pn6p63UJv+qVVtvmHmma5zqyq/zEv21BVr5/70OlV9bvz8iMX+tVb1jpupt70121yQpL95xMv2C2sp97UWruwtfah+ftLknwqyQ3nn1ftJ3PNZfPyvfKjfWZjkh+rqo1J9k7y9bnm5NbaOdv4/OhNO5vWmq9t+Eryg/nfw5J8N8lPJtkzydeSPHde9+QkfzJ/f60kNX//m0n+9/z9y5M8bf7+gZlehNdN8lNJ3pVkj3nd/03ya9swrmdlunqy9fIbJ/lGkg3zz1fLdOXjwCS/nuTl8/L9k3wl05WVTyU5OskBC/t5YpLPz9vcojOOJyR5e5KN88/Xnv/9cKYrNYvLNszLb59pVuuzC8/V/vO/pye5weKyNY778iSPWvj5L5McsezfF1++rqqv9dibMs0WfSfJmxZ60ClJnpvkxEyz3XdZpe6IJO9f+PllSR6a5KAkZ2y17QvnvnRGkut1xvKilce+8vjnf89Jct35+5Xe9GPz/q6T5E5JjluoW+lNX0+y5+KyNY777iT3Wvj5Ayu90Jev3eFrPfamebv9k3whyU0Xlq16rpNpBubMTLPdD11Y/uR52beSvHGVY/ywv3TGoTftZF9mmrbPSa21b7Tpiunns2Xm5vRMf9yT6QrG+6rq9CS/n+n2sSS5V5K3JElr7ZhMJxZJct9ML4STquqU+eebbsNYHpGF2aStlr+tbblN7X8meW9r7StbbbdxHuvxrbVDk3w8yUtXVrbWXtFau1mSP0jyjM447pfkVW2+KtNaO2+VbR5WVZ9KcnKm5+M2Sb6f5KIkr63pNpwL522PT/L6qnpcppC1llpl2ZpXnWEXty56U2vtAdlygrTyHqWNmU6K7j4f961V9cPXb1XdNtNJxG/NPx+S5Oattb9f4xhPb60dmOSNSZ7UGc79krxioe47q2xzZFWdmuSETBeWbpH5pKqq/ryqHpipVyXJaUneWFWPSnLZKvv64UNabdid7WFXti560zwz9OYkf9Za+8LK8rXOdVprJ7bWbpvkLkmeVlV7VdW1Ms3W3CTT2wKuMfeDK0pv2skITdvn4oXvNy/8vDlb3if255lmcw7OdBKw17x8tV/WleVvaK0dMn/dqrX2nN4gquoOmWZ2PrnK6q3D1D2SPKmqzskUin6tpjdCfjtTUFk5MTk6yWpvCnxLkl/oDSf9W2RukuT3kty3tXb7JO9Jstccsu6aaZbqF5IckySttcdnalwHJjmlqq6zxq6/Om+z4oaZp8lhN7QuelOStNYuSvLOTCcXyfRafUeb/Ms8ppVbbG6YqQf9Wmvt8/P290hyp7lnfTTJLavqw6sc6k1J/tNtLVuNv9ebDst08nKP1todMl3U2Ws+gblDplnxJyZ57Vzy85lOdO6U5JPzSdhq9CbYYr30ptck+Vxr7U/WWL/quU5r7dNJLkhyu0z94outtW+11i7N9J7y7Xn/lt60kxGadpz9Mk1BJ8mjF5Z/NMnDkqSq7p/pymsyTY8eUfObo6vq2lV148s5xo+8Z2lFVd1q3u/HV5a11h7ZWrtRa+2gTOHlr1trf9haa5mmtw+bN71vkrPm/dxiYbc/n+RznbEcm+TxKy/Sqrr2VuuvmanhfG++T/lB83b7JNmvtfbeJL+T+QMuqupm8xWeZ2V6o/iBWd07kzyiqvacg9ktkvxLZ5ywu9thvWm+R/8n5+83JvmvST4zr/6HzLNOVXXLTO93Oreq9s90EeVprbXjV/bVWntla+36c8+6V5J/ba0dNtcv9qbDF46xmmOzMBM1XyXe+vn4Tmvtwqq6daaZsMzvmbhaa+3tSZ6Z5NCaPijnwDa9N+KpmW7z2WeN474z08Wpqqq7J/lea+0bnXHC7m6HnjdV1QvmY/zOVstXPdepqpssnNPcOMmtMt069+Ukd6/pE/kq03nTp6/og43etNPx6Xk7znOSHF1VX8s0rXqTeflzk7y5qh6e6b7+byQ5v7V2blU9I8mx8y//pZmuIHypc4yHZTop2dqvZPpwhG2dbv2DJH9T08eWfyvJY+blT6qq+81j+U5+tIlt7bVJbpnktKq6NMlfZLoPOUnSWju1qk7OdG/wFzLdfpck+yb5fzV9XGclWfn0v5fMjawyNcZTVztoa+3Mmj655qxM09FPbD45D3qekx3Xm66R5J01fZDDhiQfTPKqed1fJfmrmj46/JIkj26ttap6UpKbJ3lmVT1z3vb+bf6giDUcNV8c2jyPo/fJeS9I8or5uJvmx/mOhfXHZLrgc1DE2qgAAB8gSURBVFqm91eeMC+/QZLXzY85SZ42P6a/rar9MvWml7XWfuR/67DgvZn689mZZvMfs8Z2wOQ52UG9aZ7NfnqmCyyfmu8Mfnlr7bVZ+1znXkn+cD6n2Zzkf7bpg7LOraq3ZXof+GWZZoBeMx/nyEyh5ScynQ+9t7X2m2s8Xr1pJ1Pbfl7NlWE+mdjUpo+wvEeSV7bWDrm8OoAdSW8C1iO9ifXCTNNV70aZ3gB9tUxXWx+35PEAJHoTsD7pTawLZprWuao6MdMnUC361dba6UsazwMyfcLVoi+21h662vY7+3GB1a3D3vSYTB8DvOj41toTd8XjAqvTm5Z73F2Z0AQAANDRvT3vnCqJCnZSB7W21se07hqO0Z9gp/XAXbg/fUBvgp3WfdfuTT5yHAAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6Ni57AOw4y07Em5d8/E2D9RuulFFsv9HxA6xJg1m/lv3HZ5TfreVa9snXqD2WPYC1Lfu8GgAAYF0TmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgI6Nyx4A69fmZQ8AALhiNg3Wb7hSRsGyLHs6ZBc+eVz2UwsAALCuCU0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAEDHxt7K/QZ3vnmwftSmJR//0sH60fGPPv/Lfv529+NvWPLx173RF9geV8oolmfZv6CMWfYLfNnHZ23L/uO/7JM3xiz7b9su3FvMNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQMfG3srvXVWjYF3asJsfn3Vur2UPYMm8QGB9WvZrc9nHh12UmSYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOjY2Ft50KWDe79wsP6iwfrR8X9/sP57g/XnD9ZfsJMf/7zB+tH//qO/f5sG6+k7YLD+poP1++w9uIPrDtbvP1jfbf9XgZ39+Hsu+fij9fss+fi7sMOOGtzBrQfrl/3fZq/B+g2D9ct+bY3Wjz5/o39blj3+0fodx0wTAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0VGuts3q/3srLd+73h8rz9bHyfGmw/sTB+s8M1v/7YP13l1x/6WD9sm0arL9osH70+buo1eAe1rk/HetPOX7w+KMNZvQFts9g/QGD9RsH6y8brP/eYP3g89/OGqu/YKw85y+5frQ/3XZX7k9nDPamrw0e/yOD9aO9afS1PdrbRo3+8f63wfrR53/0b9MPxsrbN8fqRx/+6H++n1y7N5lpAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACAjmqtrb32XdVZuQ0efJuh8uR/DdY/ZLD+u4P1Zw/WXzZY/73B+nMH6396sP5/jJW/6rix+o+PlWfvsfIvv2qs/kat1dge1rfv1Fh/es3g8V87WH/pYP05g5e8vrl5rP6zY+XD3fWiwfp9l1z/1cH69w3W/91g/YbB+u/vwv3puYO96dnPHxzAMx40uIM/GqwfPPd6wHlD5d8+duzwJ42V56OD9UcO1r94sP5bg/XvHqz/y8H6Xxj94/64tXuTmSYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKBDaAIAAOio1tqaK/+uau2VO4F9B+v3HqzfY7B+w2D9XoP1y3b6YP2JV8oolucTg/UntFZXykDWqb/ZyfvTqNH+NuriJR//gsH6SwfrLxqs39ltGqx/yi7cn64x2Jv+4coayJKcNFj/2cH60dfm5sH6bw/W/6/B+ncP1o8+f18ZrD90sP4nBusf3elNZpoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKCjWmtrr/1idVZug38fqk7OH6y/YLD+K0uu//pg/bcG6780WP/twfJzx+pH//NfOFh/0WD9qENaqyUPYYf6YI31p3sPHr9uNbiDAwbrrzdYf/XB+tFLbht28uNfY7B+j8H6vQbr9x2sHx3/U3bh/nTg4LnTXQaPP/rfZvS1sedg/WhvWvbj33uwfvS1ff0lH3/Zj3/0v98D1+5NZpoAAAA6hCYAAIAOoQkAAKBDaAIAAOgQmgAAADqEJgAAgA6hCQAAoENoAgAA6BCaAAAAOoQmAACADqEJAACgQ2gCAADoEJoAAAA6hCYAAIAOoQkAAKCjWmtrr31WdVZug+fdZqh86T581lj9YTv54//a4OP/1Fj5Sw8fqz9nrDwHDtb/xGD9/xysv6C1GtzF+nbTwf506ODxbzBY/6djw09ePVZ+6OPH6jePlbdTx+rrgLH6c/59rP7fxsoz+tfhmjcfq/+7s8fqvz9Wnsft2v1p8MX9osHDHz9Y//mh6ifV2LnD9Yaqk/cN1p8/WH+fwfo/bU8Yqr+kXjlUf/U9hspz3KVj9T/XRv847zVYf/yavclMEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdFRrbdljAAAAWLfMNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0AQAANAhNAEAAHQITQAAAB1CEwAAQIfQBAAA0CE0AQAAdAhNAAAAHUITAABAh9AEAADQITQBAAB0CE0AAAAdQhMAAECH0LSNqupjyx7Diqrau6reU1Wfqaozq+qoVbY5oqpaVd15/vmRVXXKwtfmqjpkXvfhqvrswrofn5ffuKo+UFWnzdvccDvG+vqqOmL0MV/OMR5dVZ+bvx69I48F68166k1JUlVXr6rXVNW/zj3ql+bla/aTqnpRVZ0xfz18Yfnrq+qLC71ppWftV1XvqqpT5x74mO0Yp94EV6F12KteWFVfqaofbLX8KVV11tyrPlBVN15Yt+preq19XcHxbHftNu5/z6r6u6o6u6pOrKqDduTxdkVC0zZqrd1z2WPYyktba7dOcsckP11VD1pZUVX7JjkyyYkry1prb2ytHdJaOyTJryY5p7V2ysL+HrmyvrX2zZVjJPnr1trtkzwvyR/v4Md0hVXVtZM8O8ndktw1ybOr6lrLHRVcddZhb3p6km+21m6Z5DZJPjIvX7WfVNXPJzk0ySGZXse/X1XXXNjf7y/0ppWe9cQkZ7XW7pDksCT/u6quvoMf1xWiN8GPWoe96l2ZXptbOznJnede9bYkL04u9zW91r7Wk99I8p3W2s2TvCzJi5Y8np2O0LSNVq4AVNVhVfWRqnrrfCX1qHkW51+q6vSqutm83YPnJH9yVb2/qg6Yl1+vqo6rqk9V1aur6ktVdd153aPm/Zwyr9uw2lhaaxe21j40f39Jkk8lWZwFen6mF/lFazycX0ny5m142LdJ8oH5+w8leUhv46p66vwcnLrG7Nezquqk+Wrya6qq5uVHLlzVecu87GcXri6fPAfB1TwgyXGttfNaa99JclySB27DY4NdwnrqTbPHZg5ErbXNrbVz5+Vr9ZPbJPlIa+2y1toFSU7N5b+GW5J95x6yT5LzklzWeY70Jliy9darWmsntNa+scryD7XWLpx/PCFbzq/WfE2vta81nocDqurv5350alXdc6v1+9Q0w/Wp+fl4yLz8GjXdZXRqLczKz8/fSp96aefQD0nyhvn7tyW570qvY9sITdvnDkmenOTgTLM2t2yt3TXJa5P89rzNR5PcvbV2xyRvSfLUefmzk3ywtXZokr9PcqMkqaqfSvLwJD89zwZtSvLIyxtIVe2f5MGZT0aq6o5JDmytvbtT9vD859D0urnJPHPhRXRqkl+av39oppOU66wxjgcl+YUkd5uv/r54lc1e3lq7S2vtdkl+LMl/m5f/YZI7zld1Hj8v+70kT5yfi59J8h9rPJYbJPnKws9fnZfB7mipvWnuR0ny/PkP/tErJzpZu5+cmuRBNd12fN0k905y4MJuXzifDLysqvacl708yU8l+XqS05M8ubW2eY0x6U2w/qyb86jL8RtJ/nH+/sp6Tf9ZpgtFd8g0y37mVusvSvLQ+fHdO9NMemUKaF9vrd1h7lXH1DT79dAkt5371As6x/3h+FtrlyX5XpJVz+lYndC0fU5qrX2jtXZxks8nOXZefnqSg+bvb5jkfVV1epLfT3Lbefm9Mr3401o7Jsl35uX3TXKnJCdV1SnzzzftDaKqNmYKP3/WWvtCVV0t05Tr/9epuVuSC1trZywsfmRr7eBMJwA/k6mBJdPJwc9W1clJfjbJ17L21dz7JXndytWZ1tp5q2xz7/mq0elJ7pMtz8lpSd5YVY9a2P/xSf5PVR2ZZP/5Bb7qQ1plWVtjW9jVLbs3bZz3f/z8B//jmW7LS9boJ621Y5O8N8nHMvWzj2dLH3haklsnuUuSayf5g3n5A5KckuT6mW7re3n96C19i/QmWH+W3asu1/y6v3OSl6wsWmWz7XlN3yfJK5Oktbaptfa9rQ+d5H9V1WlJ3p8p7ByQ6bm5X03vAf2Zue77mULWa6vqF5NcmLXpSYOEpu1z8cL3mxd+3pzppCFJ/jzT1cuDk/xWkr3m5WtNhVaSNyzcu3+r1tpzLmccr0nyudban8w/75vkdkk+XFXnJLl7knfW/GEQs0dkq1mm1trX5n/PT/KmzPfltta+3lr7xfkqz9PnZVu/uBfHv+aLr6r2SvJ/kxwxPyd/kS3Pyc8neUWmZvfJqtrYWjsqyW9muup7QlXdeo1dfzU/elX6hpmuPsPuaNm96duZ/mj//fzz0ZmupHb7SWvthfO+f24+3ufm5d9ok4uTvC5b3jPwmCTvmNedneSLmcLVWuPXm2B9WXav6qqq+2XqU4fP/Se56l7Tj0xyvSR3mmfM/j3JXq21f83Ui05P8sdV9az5os1dk7w904z6MZ39/nD880X3/TLd2sw2Epp2nP0yXUlNksVPTfpokoclSVXdP8nKmwg/kOSI2vLJddeuhU9s2VpVvWA+xu+sLGutfa+1dt3W2kGttYMy3Yt7eGvtE3PN1ZL8cuYrNPOyjQv3Au+R6baUM+afrzvXJNMV37/qPN5jkzy2qvZeGf9W61ea3blVtU+SIxbGdOD8Hq2nJtk/yT5VdbPW2umttRcl+UTWPiF6X5L7V9W1anpD5v3nZcDqdlhvaq21TG+IPmxedN8kZ811q/aTqtqwcttvVd0+ye0zX3Wuqp+c/61MJwQrM+Rfnved+fa/WyX5whqPV2+CndMOPY9ay/w2h1dnOn/65sKqK+s1/YEkT5iPtWGVWfL9Mn2YzqVVde8kN563vX6mO4X+NtMM/qFzz9qvtfbeTOeDh3SO+85seR6PyHSLo5mmK2Dj5W/CdnpOkqOr6muZwstN5uXPTfLm+Q18H0nyjSTnt9bOrapnJDl2/mN9aaZPiPrS1juu6aN6n57kM0k+Nb8F6eWttddezpj+S5KvttYWTy72zDT9vUeSDZmmgv9iXndYpqsZLck/zeNZVWvtmJo+DvgTVXVJpttt/mhh/Xer6i8yXSE5J8lJ86oNSf62qvbLdJXoZfO2z5+bxaZMJ13/mFW01s6rqucv7O95a9x+A0yekx3Um2Z/kORvqupPknwr06xQsnY/2SPJP8997PtJHrVwy9sbq+p6mXrDKdnyvqLnJ3n9fNtOJfmDtuUDJ36E3gQ7redkB/aqqnpxkv+eZO+q+mqS184zUy/J9AEzR8996cuttcN7r+nOvlbz5CSvqarfyNRHnpDptuQVb0zyrqr6RKa+95l5+cFJXlJVm+fH9oRMdxj9v3nGvJL87prPZvKXmXrz2ZlmmB7R2ZZVlJB51arpjcybWmuXVdU9krxynn4FWBq9CdgZ6FUsi5mmq96Nkrx1vgpySZLHLXk8AIneBOwc9CqWwkzTOldVJ2a6hW7Rr7bWTl/SeA5O8jdbLb64tXa3XfG4wOr0puUeF9g2y+pVVfX0TO8jX3R0a+2Fu+JxdwdC0//f3t282HUXYBw/l5nEkFCSRa1gKSjShXXRjSBY3FRciIigYIWupOBCFF0pCNaXrqrUjS/UpYgFBcWgbhSLLxh1USFWY6ARUxNUqE0T04nTxJnjP9B5GPtk/N178/lsD889J5OZX+43tzQAAABB/s/z3rtQVLCqvjev97/0fZfzCVbWhTU+n444m7iFvew/M/4/GP3/9d7e+2wa/WgAAABLTTQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAIJoAAAAC0QQAABCIJgAAgEA0AQAABKIJAAAgEE0AAACBaAIAAAhEEwAAQCCaAAAAgs3RDwAAAEtht9zf6h9HrPGvf41/aQAAAD3RBAAAEIgmAACAQDQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAIJoAAAAC0QQAABCIJgAAgEA0AQAABKIJAAAgEE0AAACBaAIAAAg2Rz8AAAAsBR8ndHbLffv13yj3gW8NAACAQDQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAIJoAAAAC0QQAABCIJgAAgEA0AQAABKIJAAAgEE0AAACBaAIAAAhEEwAAQCCaAAAAgs3RDwAAAKyB0R/H7BzcS4/+pQEAACw10QQAABCIJgAAgEA0AQAABKIJAAAgEE0AAACBaAIAAAhEEwAAQCCaAAAAAtEEAAAQiCYAAIBANAEAAASiCQAAIBBNAAAAgWgCAAAINkc/AAAAwLRb7g/w4yCfNAEAAASiCQAAIBBNAAAAgWgCAAAIRBMAAEAgmgAAAALRBAAAEIgmAACAQDQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAIJoAAAAC0QQAABBsHuirj06y3cH3p7Mz+gFKG6MfAFhbo8+XVT+fWV+j3/uNfu/bGv31e1W5P8DnX/XfWgAAgAMlmgAAAALRBAAAEIgmAACAQDQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAIJoAAAAC0QQAABCIJgAAgEA0AQAABKIJAAAgEE0AAADB5oG++m65l3S3to1yv3NTngJg+bTnW3u+AhyEth0OkCwBAAAIRBMAAEAgmgAAAALRBAAAEIgmAACAQDQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAIJoAAAAC0QQAABCIJgAAgEA0AQAABKIJAAAg2IxXj5evfq3c3yj3rZ3B+1W3seJ7f6Ww3A6V+92b8hSv3K1+PjCW77+D0/7ZM9ro743RZ3Nr9PO3713aP1vXmLeFAAAAgWgCAAAIRBMAAEAgmgAAAALRBAAAEIgmAACAQDQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAIJoAAAAC0QQAABCIJgAAgEA0AQAABJvx6pX/01Msq43Be8baHf0ARDdGPwDAy9gZ/QArzl/nj+X7d0++NQEAAALRBAAAEIgmAACAQDQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAIJoAAAAC0QQAABCIJgAAgEA0AQAABKIJAAAgEE0AAACBaAIAAAgW8zyHy+fSxX34UTef/ljuL5b7P5f7f5b7y938+m657+ZTefs66TfK/ZFyvzg6+AGeX5QvsOR+Vp5PPy7vf7bcP1vuz3Xz//yr22938/p8aM+nVvv8q649n2+f1/h8ulyeTSfL+/+03Lfvvf7QzV8sf7jbs+lGuW/Php1y3773Gf3ea7Q79j6bfNIEAAAQiCYAAIBANAEAAASiCQAAIBBNAAAAgWgCAAAIRBMAAEAgmgAAAALRBAAAEIgmAACAQDQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAsJjnOVz+erq4D5/v5n//W7e/3M1rh8t9m7S75b79+l0q9+fL/Y1yv1PuR/vovBj9CAfqB4vufDpb3v+5cn9l8P23yv2q/3y15+Po598o90fL/aFy/8M1Pp++Vp5N95T3/0e5v1Duz5T78q3fdLXcb5f79mxsf7aPl/tWe7ZeL/ft1/93e59NPmkCAAAIRBMAAEAgmgAAAALRBAAAEIgmAACAQDQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAIJoAAAAC0QQAABCIJgAAgEA0AQAABKIJAAAgWMzzHC5/KF3chxe7+fSucv9guX+43P+m3P+qm79wrdtf6ObTF7r5V77V7c908+lkub+73D9f7p+e50X5EkvuvvJ8ulje/+3l/r5yf6Tc31nunyz33+3mfyl/wl9/uJr/ZHG92t9TrafpzneWL/Dlcv+Go+ULbK3x+XRveTadLu///nLfni3t2di+9ztR7r9a7tvfv3vL/Ze6+cN/rea/fKS7/dtu7/bT4+X+fXu/d/JJEwAAQCCaAAAAAtEEAAAQiCYAAIBANAEAAASiCQAAIBBNAAAAgWgCAAAIRBMAAEAgmgAAAALRBAAAEIgmAACAQDQBAAAEogkAACAQTQAAAMFinuc9L86Lxd4X9/Pix5v1NE3Hyn2ZhPPF8v6lG2NvP10q90+U+5+X+9bVcr99U57ilTs1z4vBj3Cg6vPpaPkAh8r9ijt7Zez9z5f715X7d5f7E+W+/fa7f/D9P7PG59MD5dnUemu5b4/Gu8r9Rrlv/+w+We5b7dlwutxvlfsz5f775b79/rk/nE0+aQIAAAhEEwAAQCCaAAAAAtEEAAAQiCYAAIBANAEAAASiCQAAIBBNAAAAgWgCAAAIRBMAAEAgmgAAAALRBAAAEIgmAACAQDQBAAAEogkAACBYzPMcLn8jXdyHD3fzp651+2e6+XS63F8o98+V+6uD9zvl/nC5P1Tubyv3J8r9kXL/xLwoX2HJfa47n/792e72v+3m09lyP/p8as+H7XL/UrlvXR98/9E2yv3T63w+nSvfOz3UzX//i25/qptPvy7358v9lXK/Ve7bs2233Lfvfdp9ezaM/jjnzN5n0+hHAwAAWGqiCQAAIBBNAAAAgWgCAAAIRBMAAEAgmgAAAALRBAAAEIgmAACAQDQBAAAEogkAACAQTQAAAIFoAgAACEQTAABAIJoAAAAC0QQAABAs5nkOl4+li/vwYDef3tTNj32827+6mz/6bLe/u5tPryn3j5X7Z8r95XJ/qdy33lPu31LuPzbPi/Illl15Po32aLn/VDd/areav+PN3e1PdfNp66HyBQ51828/3u0f+HS3nz5Y7rvf/mm6rdzfscbn058W3dn0xhU/2mqfqNYfWXyx2m9X6/692yfLs+Gbj3T773Tz6clyv9W++Xxtuf/A3meTT5oAAAAC0QQAABCIJgAAgEA0AQAABKIJAAAgEE0AAACBaAIAAAhEEwAAQCCaAAAAAtEEAAAQiCYAAIBANAEAAASiCQAAIBBNAAAAgWgCAAAIFvM8j34GAACApeWTJgAAgEA0AQAABKIJAAAgEE0AAACBaAIAAAhEEwAAQPBfLDXHNROEVYoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 1080x1080 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "fig = plt.figure(figsize=(15, 15))\n",
    "columns = rows = 3\n",
    "for i in range(1, columns*rows +1):\n",
    "    index = np.random.randint(len(x_train))\n",
    "    img = x_train[index, :, :, 0]\n",
    "    fig.add_subplot(rows, columns, i)\n",
    "    plt.axis(\"off\")\n",
    "    plt.title('image_'+str(index)+'_class_'+str(np.argmax(y_train[index])), fontsize=10)\n",
    "    plt.subplots_adjust(wspace=0.2, hspace=0.2)\n",
    "    plt.imshow(img, cmap='hot', interpolation='nearest')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Print labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "close = dataloader.dataframe[\"Close\"].values\n",
    "labels = dataloader.train_labels\n",
    "size = 200\n",
    "closes = close[:size]\n",
    "\n",
    "states_buy = []\n",
    "states_sell = []\n",
    "\n",
    "for i in range(size):\n",
    "    if labels[i] == 1: states_buy.append(i)\n",
    "    if labels[i] == 2: states_sell.append(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA3cAAAEyCAYAAABZOSngAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzdeXxU1fnH8c+ZhCGAiKxmEBcQUbZAMKDBsFh3UbHuCy6DaF3q1l+ttTpu1EqtLdoqUivuFrVqXYq7trgAathxARUCIgHZBYEs5Pz+OJnMTDLZJ7Mk3/frldeduXfmzpkQknnu85znGGstIiIiIiIikto8iR6AiIiIiIiINJ6COxERERERkWZAwZ2IiIiIiEgzoOBORERERESkGVBwJyIiIiIi0gwouBMREREREWkGag3ujDGPGmN+MMYsqeb4+caYReVfs4wxg8KOXWuMWWKM+dwYc13Y/k7GmHeMMV+XbzvG5u2IiIiIiIi0THXJ3D0OHF/D8RXAKGttFjAReBjAGDMAuBQYBgwCTjLGHFT+nN8C71lrDwLeK78vIiIiIiIiDVRrcGet/QDYVMPxWdbazeV35wA9ym/3BeZYa3dYa0uBmcDPy4+NBZ4ov/0EcGoDxi4iIiIiIiLl0mN8vkuAN8pvLwHuMsZ0BnYCJwL55cf2ttYWAlhrC40x3epy8i5dutgDDjggtiMWERERERFJEXPnzt1gre0a7VjMgjtjzJG44C4PwFr7pTHmj8A7wHZgIVDagPNeBlwGsN9++5Gfn1/LM0RERERERJonY8zK6o7FpFumMSYLeAQYa63dGNxvrZ1mrR1irR2JK+38uvzQOmOMr/y5PuCH6s5trX3YWptjrc3p2jVqgCoiIiIiItLiNTq4M8bsB7wEXGCtXVbpWLewx5wGTC8/9CpwUfnti4BXGjsOERERERGRlqzWskxjzHRgNNDFGLMauA1oBWCtnQrcCnQGphhjAEqttTnlT3+xfM5dCXBVWOOVScDzxphLgFXAmTF7RyIiIiIiIi2QsdYmegx1lpOTYzXnTkREREQkdkpKSli9ejW7du1K9FAkTEZGBj169KBVq1YR+40xc8OSaRFi3S1TRERERERSyOrVq2nfvj0HHHAA5ZV4kmDWWjZu3Mjq1avp2bNnnZ8Xk4YqIiKNkZ2djTGm1q/s7OxED1VERKTZ2bVrF507d1Zgl0SMMXTu3Lne2VQFdyKScLm5uXi93hof4/V6GT58eJxGJCIi0rIosEs+Dfk3UXAnIgkXCATweGr+dZSWlkYgEIjTiERERKQmRYVFzB81n6K1RYkeioRRcCciCefz+fD7/dVm77xeL36/n8zMzDiPTERERKIpmFjA1o+2snJitetp1/1cBQUMGDAgBqOCqVOn8uSTT8bkXPE8/+jRo4lF40g1VBGRpBC44QYee/TRqMeUtRMREUkeRYVFrHtsHZTB2sfWsn9gf1pntk70sAC4/PLLU/r8jaXMnYgk3t/+hq9PH/xFRVTO3Xm9Xtq1a4fP51PDFRERkSRQMLEAW+aWU7O7bUyyd6WlpVx00UVkZWVxxhlnsGPHDgAOOOAANmzYAEB+fj6jR4+mrKyMgw46iPXr1wNQVlZG79692bBhA7fffjv33nsv4LJhN954I8OGDaNPnz58+OGHAOzYsYOzzjqLrKwszj77bA477LCoWbPf/va39OvXj6ysLH79618DRJz/s88+Iysri9zcXG644YaK7OPjjz/OaaedxvHHH89BBx3Eb37zm4pzXnHFFeTk5NC/f39uu+22Rn/fKlNwJyKJ98wzUFpKYP/9q8y9S0tLY8yYMWq4IiIikgSCWTtbXB7cFVvWPra20XPvli5dymWXXcaiRYvYc889mTJlSrWP9Xg8jBs3jmeeeQaAd999l0GDBtGlS5cqjy0tLeXTTz/lvvvu44477gBgypQpdOzYkUWLFhEIBJg7d26V523atIl///vffP755yxatIhbbrmlymP8fj9Tp05l9uzZpKWlRRxbsGABzz33HIsXL+a5557ju+++A+Cuu+4iPz+fRYsWMXPmTBYtWlT3b1IdKLgTkcT7+msAfLNn4//FLyoCOS/gz8vj7rvvVsMVERGRJBCetQuKRfZu33335YgjjgBg3LhxfPTRRzU+fvz48RVz3x599FH8fn/Ux5122mkAHHrooRQUFADw0Ucfcc455wAwYMAAsrKyqjxvzz33JCMjgwkTJvDSSy/Rtm3biONbtmxh27ZtFReWzzvvvIjjRx11FB06dCAjI4N+/fqxcqX7/jz//PMMGTKE7OxsPv/8c7744osa32d9KbgTkcTatMl97bEHZGZGdM5MAwIrVuDr1k0NV0RERBKsctYuKBbZu8pt/4P309PTKSsrA4hY823fffdl77335v333+eTTz7hhBNOiHre1q3dXMC0tDRKS0vdeK2N+thw6enpfPrpp5x++um8/PLLHH/88RHHaztH8HXDX3vFihXce++9vPfeeyxatIgxY8bUex272ii4E5HE+uYbt+3dG4yp6Jzp8Xjwt29P5jffwD//WeNyCcraiYiINL1oWbugxmbvVq1axezZswGYPn06eXl5gJtzFyybfPHFFyOeM2HCBMaNG8dZZ51VpSyyJnl5eTz//PMAfPHFFyxevLjKY7Zv387WrVs58cQTue+++1iwYEHE8Y4dO9K+fXvmzJkDwLPPPlvr6/7444+0a9eODh06sG7dOt544406j7muFNyJSGKVl2TSu3fFrkAgQF5eHoGJE92OO+7At9de+Pv2rdJwBWDnzp1RG66owYqIiEhsVJe1C2ps9q5v37488cQTZGVlsWnTJq644goAbrvtNq699lpGjBhRJYA75ZRT2L59e7UlmdW58sorWb9+PVlZWfzxj38kKyuLDh06RDxm27ZtnHTSSWRlZTFq1CgmT55c5TzTpk3jsssuIzc3F2ttlXNUNmjQILKzs+nfvz/jx4+vKEONJVOXtGSyyMnJsbFY/0FEksgdd8Dtt8NNN8Ef/hB5rLQU+veHZctgn30o/P57egF1KWDwer1MmDCBBx98sAkGLSIi0nx8+eWX9O3bt8bHLL1yKWunra02uAMwXoNvgo8+D/aJ9RCjys/P5/rrr6/ogllXu3fvpqSkhIyMDL799luOOuooli1bVmvztsq2b9/OHnvsAcCkSZMoLCzk/vvvr9c5ahPt38YYM9damxPt8VrnTkQSK5i5O+igqsfS0+G22+D88+H77/Httx/+IUN4ZMYMSkpKajytSjVFRERi58fZP9YY2IHL3m2dtTUu45k0aRIPPfRQRcfM+tixYwdHHnkkJSUlWGt56KGH6h3YAcyYMYO7776b0tJS9t9/fx5//PF6nyPWlLkTkcQ67DD49FP44AMYMaLq8d274aqrwBi4+24Kd+6kV69eNU5AVtZORESk7uqSuZPEqG/mTnPuRCSxasrcAaSlwdSp8NBDsNdeFQ1XjDGkqcGKiIiISAUFdyKSOJs2webNbhmEvfeu89MCgQAjRoxg3AUXRD0ercGKmquIiIhIc6c5dyKSOOFZu0rr29TE5/Mxc+ZMCgsLefrpp9m9e3eNj/d6vRWLjIqIiIg0V8rciUjiRFkGoT58Ph/nn39+rY9TmaaIiEjjZGdnV1lyKNqXKmUSS8GdiCRObfPt6mDSpEk1Llzq9Xrx+/1kZmY2+DVERERautzc3Fo7SsarUubiiy/mhRdeAGD06NEEGy6eeOKJbNmypcletynOX1BQwIABA2J2PgV3IpI433zjto0I7mrL3ilrJyIi0niBQABPNY3MghL9N/f1119nr732Stnzx4KCOxFJnEaWZQZNmjQJn89X5YpitKydykpERETqL9iturrsXWMqZX766SfGjBnDoEGDGDBgAM899xwAc+fOZdSoURx66KEcd9xxFBYW1nieAw44gA0bNlBQUEDfvn259NJL6d+/P8ceeyw7d+4E4LPPPiMrK4vc3FxuuOGGqFmzwsJCRo4cyeDBgxkwYEDFIunB8wNMnDiRQw45hGOOOYZzzz2Xe++9F3CZxBtvvJFhw4bRp0+fiucWFBQwYsQIhgwZwpAhQ5g1a1a9v091oeBORBLD2piUZYL7gzN37twqVxTTrK1yBTGZykpERERSSU3Zu8Zk7d588026d+/OwoULWbJkCccffzwlJSVcffXVvPDCC8ydO5fx48dz88031/mcX3/9NVdddRWff/45e+21Fy+++CIAfr+fqVOnMnv27Gqndfzzn//kuOOOY8GCBSxcuJDBgwdHHM/Pz+fFF19k/vz5vPTSS1Reh7u0tJRPP/2U++67jzvuuAOAbt268c477zBv3jyee+45rrnmmvp8i+pMwZ2IJMbGjbBlC7RvD926Nfp0la8oegG/MWRW+sWdCmUlIiIiyai67F1j57cPHDiQd999lxtvvJEPP/yQDh06sHTpUpYsWcIxxxzD4MGD+f3vf8/q1avrfM6ePXtWBGWHHnooBQUFbNmyhW3btlVcwD3vvPOiPnfo0KE89thj3H777SxevJj27dtHHP/oo48YO3Ysbdq0oX379px88skRx0877bSI1wUoKSnh0ksvZeDAgZx55pl88cUXdX4v9aHgTkQSIzjfrnfvei2DUJPwwC3N4yFQXAw33RTxmKYsKxEREWnuol0kbexF0T59+jB37lwGDhzITTfdxJ133om1lv79+7NgwQIWLFjA4sWLefvtt+t8ztatW0eMr7S0FGttnZ47cuRIPvjgA/bZZx8uuOACnnzyyYjjtZ0n+NrB1wWYPHkye++9NwsXLiQ/P5/i4uI6v5f6UHAnIokRo5LMcMHAzePx4D/7bDJbtYJp0+CTTyIe11RlJSIiIs1dlUqZGFwUXbNmDW3btmXcuHH8+te/Zt68eRx88MGsX7+e2bNnAy7z9fnnnzdq7B07dqR9+/bMmTMHgGeffTbq41auXEm3bt249NJLueSSS5g3b17E8by8PF577TV27drF9u3bmTFjRq2vvXXrVnw+Hx6Ph6eeeqrWNXobSsGdiCRGEwR34AK3vLw8An/5C/zqV27nVVdB2C/Rij9MlUo2lbUTERGpXUSlTAwuii5evJhhw4YxePBg7rrrLm655Ra8Xi8vvPACN954I4MGDWLw4MExaUIybdo0LrvsMnJzc7HW0qFDhyqP+d///sfgwYPJzs7mxRdf5Nprr404PnToUE455RQGDRrEaaedRk5OTtTzhLvyyit54oknOPzww1m2bBnt2rVr9HuJxtQ1PZkMcnJybOUJiyKSos47D6ZPh8ceg4svbprX2L4d+vaF1avhlVfglFMqDhUWFtJrn33YFfY7sE2bNixfvlzBnYiItChffvklffv2rddzrrzySv7+979z+eWX8+CDDzbRyGJv+/bt7LHHHoDrtl1YWMj999/f4PPs2LGDkSNH8vDDDzNkyJBYDzfqv40xZq61Nifa45W5E5HEaKLMXYQ99oBf/tLdrlQv7wP81hKceedt1UpZOxERkTqqqJRJsakMM2bMiFji4JZbbmnQeS677DIGDx7MkCFDOP3005sksGsIZe5EJP6shY4dYetWWLcuJt0yq/X997DffpCeDoWF0KmT2//wwxT+4hf0AnYBbbxelq9cqeBORERanIZk7iQ+lLkTkeS3YYML7PbcE7p2bdrX2mcfOPpoKC6G8InTr7zisndpaXgAf16eAjsREWmxUinh01I05N9EwZ2IxF8TLINQo4suctsnnnDbbdvg3XfBGAI//zl5QOCww5p+HCIiIkkoIyODjRs3KsBLItZaNm7cSEZGRr2el95E4xERqV485tuFO/VUt1j6p5/CV1/B4sUuk3fEEfhyc5n5wguu+YqIiEgL1KNHD1avXs369esTPRQJk5GRQY8ePer1HAV3IhJ/X33ltvEK7tq2hTPPhEcfddm7775z+8eOBZ/P3S4sjM9YREREkkyrVq3o2bNnoochMaCyTBGJv//9z23jWQoZLM186ikILjZ66qmh4G7t2viNRURERKQJKHMnIvH144+uPDItDUaOjN/r5uVBz56wYoW736+fyxyWlbn7ytyJiIhIilPmTkTia+ZM2L0bhg1z3TLjxeOBCy8M3T/1VLcNL8vURHIRERFJYQruJKllZ2djjKn1Kzs7O9FDlbp67z23Pfro+L92eHA3dqzbtm/v5uTt2OG6aIqIiIikKAV3ktRyc3Pxer01Psbr9TJ8+PA4jUgaLRjcHXVU/F+7Vy8IBODyyyGnfO1PY9RURURERJoFk0rrWeTk5Nj8/PxED0PiqLCwkF69erFr165qH9OmTRuWL1+uBahTwbp1kJkJbdrA5s3QunWiR+Tk5cHHH8N//wujRyd6NCIiIiLVMsbMtdbmRDtWa+bOGPOoMeYHY8ySao6fb4xZVP41yxgzKOzY9caYz40xS4wx040xGeX7HzfGrDDGLCj/GtzQNyfNm8/nw+/3V5u983q9+P1+BXap4v333XbEiOQJ7ECZOxEREWkW6lKW+ThwfA3HVwCjrLVZwETgYQBjzD7ANUCOtXYAkAacE/a8G6y1g8u/FjRk8NIyBAIBPJ7oP6ppaWkEAoE4j6hxWvQ8wnffddtEzLeriYI7ERERaQZqDe6stR8Am2o4Pstau7n87hwgfBn1dKCNMSYdaAusacRYpYWqyN6lpUXsT9WsXYudR2htYufb1URr3YmIiEgzEOuGKpcAbwBYa78H7gVWAYXAVmvt22GPvau8lHOyMSaJ6rMkGQWuvRbP7t0R+1Ixawc1ZyKDUvW91aRoztfMX3kdRXv1gsFJVomtzJ2IiIg0AzEL7owxR+KCuxvL73cExgI9ge5AO2PMuPKH3wQcAgwFOgWfU815LzPG5Btj8tevXx+r4UqK8T31FH4gmO/yAv6LLkq5rB203HmEBb/9gq1ksbLTtW7NuWSi4E5ERESagZh8wjLGZAGPAGOttRvLdx8NrLDWrrfWlgAvAcMBrLWF1ikCHgOGVXdua+3D1toca21O165dYzFcSTVr18LkyQQAT3lAlAYE+vdP6LAabNYsArNm4SkpiXq4WWbtCotY99EegIe13w2gaG1RoocUScGdiIiINAONDu6MMfvhArcLrLXLwg6tAg43xrQ1xhjgKODL8uf4yrcGOBWI2olTBIDf/x527MA3diz+Sy7BYwx+IPOhh6CsLNGjq5+SEpgwAd/ChfitpXLuLuFZu61b3dy4GCgqLGL+qPkUrS2i4M4CbJk7r8XDyokrY/IaMRP8fiu4ExERkRRWl6UQpgOzgYONMauNMZcYYy43xlxe/pBbgc7AlPJlDfIBrLWfAC8A84DF5a/1cPlznjHGLC7f3wX4fSzflDQjy5fD3//uFpq+6y4CgQB5RxxBwOeDL76A115L9Aijy893a7pV9o9/wJdfwoEHEvjwQzwZGRGH0zyexGXt5syBrl3hlFOguLjRpyuYWMDWj7ay/LfLWffoGiytALAlsPaxtcmVvevSBdLT3dp7NaypKCIiIpLMtIi5JK81a+DKK+GVV+Cii+Dxx0PH7r8frrsOhg1zQclXX8Ebb8CKFXDHHdCpU+Nee/16ePhhOPZYGDq0fs9duRJ69YJu3WD2bDjgALd/yxbo3Rs2boSXXoKf/5wrr7ySadOmUVxcjBeYcOihPJiIn3FrYdQo+PBDd/+cc+Dpp6FSh9K6Kios4pNen1C2qwxMGcburgjuAIzX4Jvgo8+DfWIx+tjo0QO+/979DAX/zURERESSTKMWMReJq2+/hd/+FgYNgn32cYGd1wu33x75uAkToHNn+PRT96G8Xz/4v/+DBx6Ae+5p3BiKi+HUU+GWW1zweMIJMGtW3Z+/eLErF1271j13U/lKInfd5QK7kSPd+YnsnJkGBJYvh507Gzf+hnjrLRfYdewI7dvDs8/C1Vc3uESzYGIBdnd5yaw1EYEdgC22yZe907w7ERERSXEK7iR5bN4Mubnwxz/CokXQti2cdBL85z9VMynt2sH117vba9a4csITT3T3p09v3Fy8X//aBXNdurjXefNNOOIIl8XbuLH2569YEbr91Vcwdix8/rnLNhoDf/mL2xLqnOnxePB37Urm5s3w5JMNH3tDlJXB737nbt98M7z6KrRuDQ89BLfeWu/TFRUWse7RQmxFvxgT9XF2t02uuXda605ERERSnII7SR6BgCuHHDoU3nnHZbxeew2OOSb642+80WWY8vPdB/LXXoN994VVq+qXaQs3fTr87W/QqpULKgsKXMCz555uTHUJdoLB3S9/6bKKH30Ehx3mmqlceCEcemiltx0gLy+PwJ13uh2TJ8e3UcyLL8L8+dC9uyuDHT3afV89HtfMplUrF+R26uTezxFHwPnnu8zmm29WOV3BnQXY4tJaXzbpsnfK3ImIiEiKU3AnyWH+fJcpSkuDRx+Fo4922aOapKfD2We7YMnjcV/nnuuOPfNM5GPLylzmbciQ6sselyxx5Z4A993nArIuXVyAM3u2O//DD7smLzUJBncjRrh5gB06wE8/QZs2rjSzEp/Px8yZM8mcMAH22w+WLnXPq6y42GUOV69uePBnbWSpZWmpC6rBBa5t2rjbp54KTzzhgtrSUtixw2VWv//eBc7//Kd7LyecAFOmAC5jN/fwuaydtgZr0+s2nGTK3im4ExERkRSn4E4Sr6wMrrrKba++GgYMaPi5zjvPbZ9/PrLj4/TpLvM2fz5MnVr1edu2wemnuyBm3Di44orI4/36wQUXuECn8vy/yoLBX8+e7r28/LLLeE2a5OYRVic9Ha65xt3+zW/c6+XluedmZLhgt0sXl530+6Ofo6zMzVssD+DClyOgsNBlRTt1cpm35593gdnSpXDggTB+fOS5xo1zTWCKi933Z8MG997++18XgAfHes018NZbFEwsYNsn27AldQ88bbFl66ytdX58k1JwJyIiIilO3TIl8R5/3AUre+/tAo0OHRp+Lmth4EA3x+2119ycveJiOOSQUEZt771dkNK2beh511/vsnUDB7rum+HHglasgIMPdgHeokXRg1Br3fi3bXMlpl261G/8W7e64G3btqrH0tNds5PNm12wt2GDK5cMd+edcNtt7n0/+ihLb9tE4d8L6X7uHvSZfXb1WcdnngkFxvVxyy1w110U7bE/c4ofx9aygoKnjYfDlh9G68xasrKJ8MorLmN54okwY0aiRyMiIiISlbplSvLassVlqQD+9KfGBXbgGpUEg5R//tNt//EPF5j17Qs5OW79uYceCj1n/nz4619d2eWTT0YP7MBl4i67zAVw1a1Ft2mTC8z22MN186yvDh1ckHHrrTBtGrz/vhv7jh0uSN20yZWL7trlMpHhysrcewX4z38oGjCKddPWQBmsfWYjRcs3u/f/6afue52X575fhx/ulj5oiDvvhDPOoGD7qdji3bU+PKnKMCtT5k5ERERSnDJ3kli/+pVrIJKXBx98UNFFslFWrHDrzLVt6zJVgwa5gO6ll1xp45gxrrvmihUuA5abC5995rJ3f/lLzedeu9ade+dOl+E77LDI4/n5rvRx4ECX3WsKf/iDa/Li97vyyKA5c9x76d4devdm6QfZrOVELF4Mxfi6L6TPV5e77F/Qjz+674HX2+DhFH27hU8O+owy26r2BwPtBrdj6Px6rh0YD9995+Y8ZmYqwBMREZGkVVPmrm5dD0SawubN8Pe/u9vBZQJioWdPGD7cNf4YM8YFdocdVrG2HIcdBp98Ag8+6AKdzz5zc+HuuKP2c2dmwrXXuvlzN98M774beTxY+tmzZ2zeSzRjx7rX/s9/YPfu0ELjL7zgtmedRdH//YF1PWdhS90xi5e1m4ex/09eWofFduy5Z6OHU/DnddhWrSCsJDMpFymvzd57u+0PP0R+X0VERERShMoyJXGmTXPlhkcd5bpYxtL557vt3LluO2mSCx6NCQVx99wDN93kbv/1r5EZrZr85jeufPK99+CLLyKPxSO469fPZQ/Xr3fZOnClosHg7owzKPjDKqwn8tqN3U3MSyKLCotY99i6KnPtkm6Zg7rwet0cybIyF+CJiIiIpBgFd5IYpaXwwAPu9nXXxf78Z54Zyrwcd5xbuy3o2GNd+eLGja6ByYknws9/Xvdzd+wYWnvvs88ij8UjuDMGTjnF3X71VbfNz4eVK6F7d4oOGFIecEWWXDdFwFUwsQBbFr20O6nn11VHC5mLiIhIClNwJ4nxyisuGOnd2wVXsda1q2sS0raty9qFC8/eZWS4ILO+JaHBTOP8+ZH74xHcgSvNBPd9hFDW7vTTKbhrVVwCrlDWrprXSpHsXXZ2NsYY97V4MQYwQ4aE9pV/ZWdnJ3qoIiIiIjVScCdxVbHu2j2PuB3XXuu6VDaFxx+HNWtg8OCqx44+2i3SPWNGwwKxYHA3b17k/ngFd0cc4TKIS5dS9OEXzL+/F0V0pOjI0+MWcNWUtat4vRTI3uXm5uKtpaGM1+tl+PDhcRqRiIiISMMouJO4KphYwNYPt7Ly04NdM4+LLmq6F0tPr35pBWPgwgvhZz9r2LmDWZz5890cLXDbggJ3u6mDu1atKjKeBeP/y9aig1jZ5nIK3sqMW8D14+wfqw0iK14rmRYpr0YgEMBTywWGtLQ0AtUtfyEiIiKSJNQtU+ImWMaHhbUcz/7ntqV1XZuYJJtu3VyHze+/h2++gT59XPv84mJXErrHHk0/hrFjKXrmDdZ90xvwsLb4Z2R8tDVuAVdSLmfQAD6fD7/fz7Rp0ygurroKu9frxe/3k5mZmYDRiYiIiNSdgjuJG1fG57JcFg8rfzqNFGqUX9WQIS64mzfPBXfxKskMOu44CsxFWOvmC1qTxl6j9mLYkmHxef1mJBAI8Nhjj0U9pqydiIiIpAqVZUpcVG6Zb/Gy9sUdSd9so0aVm6rEObgr+qk168wJWNx8MVtqUqKBSTIKZu+8rSIXYlfWTkRERFKJgjuJi2jNN1Kh2UaNKjdViXNwVzCxAGsiF9pO+e9pAgUCATyVFi5X1k5ERERSiYI7aXLVtcxPlVb51Qo2VZk3zy0iHsfgruJ7ujvyv3DKf08TqCJ7Vxlbn/wAACAASURBVH7f26qVsnYiIiKSUhTcSZNrdgtdB/XoAV26wKZNsGpVXIO7Zvs9TbDwzplp5fdFREREUoWCO2lSRYVFrH10bcovdB2VMZGlmcuXu9tNHNw1l8XDk5HP58M/ZAgewN+vn7J2IiIiklIU3EmTKphYgC1J/YWuqxUM7j75BFavdgHffvs16Us2l8XDk1Xg8svJAwIdOyZ6KCIiIiL1oqUQpElt/WArlNX8mFRY6LpawXl3L7/s5t3tuy94vTU/p5Gay+Lhyco3ahQzAZYtS/RQREREROpFwZ00qQ4jO7Dzix+xNh2TVobvFz3o82BKr24XKZi5W7rUbeMw3665LB6etHr1grZtYc0a2LgROndO9IhERERE6kRlmdJkigqLWPdoIda6awh2t6f5zQXr1Qv23JMiOjGfyRRl9kv0iKSxPB4YMMDdXrw4sWMRERERqQcFd9JkCu4swBaXRuxrdnPBPB7IzqaAC9hKFiuX5yV6RBILAwe6rYI7ERERSSEK7qRJFBUWsW7amoqsXVBz7ORY1CeXdZwAeFi70Nes3luLpeBOREREUpCCO4m5osIi5g75DFtSGvV4c8veFXw7HIsBwOJpVu+txcrKclsFdyIiIpJCFNxJzH37228pXluCpVXU480pe1dUWMS6j9tjcR0ybQnN5r21aMHM3ZIlUFZLu1cRERGRJKHgTmKqqLCIH57+AcozWdVpLtm7gokFWBv5XpvLe2vRunSBzEzYvh0KChI9GhEREZE6UXAnMVVwZwHUssA2NI912IoKi1j32Loqa841p8xki6Z5dyIiIpJitM6dxIwLdgoJz9p52ng4bPlhtM5snbiBNZGCiQXYagLZYPauWa3p19IMHAjvvOOCu7FjEz0aERERkVopcycx44KdyH3NtUSxuqxdkLJ3zYCaqoiIiEiKUXAnMVER7JRE7m+uQU5NWbug5hrYNnfZ2dkYYzAXX4wBzPPPu/uVvrKzsxM9VBEREZEICu4kJupSotic/Dj7x2qzdkHNYV5hS5Sbm4vX663xMV6vl+HDh8dpRCIiIiJ1Y6ytvflFssjJybH5+fmJHoZUUlRYxCe9PqFsV/Ut45vz3DtpXgoLC+nVqxe7du2q9jFt2rRh+fLlZGZmxnFkIiIiImCMmWutzYl2TJk7aTSVKEpz4vP58Pv91WbvvF4vfr9fgZ2IiIgkHQV30mgqUZTmJhAI4PFE//WYlpZGIBCo03kq5u814ktz+0RERKSutBSCNNrQ+UNDd267De68E26/3d0WSUHB7N20f/yD4tLSiv31zdrl5ubyxRdfUFxc3KBxaG6fiIiI1IcydxJb27e77R57JHYcIo0UCATwpEde/6pP1q7iHNVkAOuivq8nIiIiLZuCO4ktBXfSTFTMvTMGAG96er3n2tU2f8/j8VQb/Glun4iIiNRXnYI7Y8yjxpgfjDFLqjl+vjFmUfnXLGPMoLBj1xtjPjfGLDHGTDfGZJTv72mM+cQY87Ux5jljTM29xyU1BIO7du0SOw6RGAjPvKWV32/MOSrzer3VBn7K2omIiEh91TVz9zhwfA3HVwCjrLVZwETgYQBjzD7ANUCOtXYA7vPROeXP+SMw2Vp7ELAZuKTeo5fko8ydNCM+nw//scfiAfxdu9aeRVu8GBYurHqOc86hcgjnTU9n/Pjx+E88seoxZe1ERESkAeoU3FlrPwA21XB8lrV2c/ndOUCPsMPpQBtjTDrQFlhjjDHAz4AXyh/zBHBqPccuyeinn9xWwZ00E4F77yUPCGzcCEVF1T+wuBhGjIBDD4Xp00P7rSWwenWVX7ZppaUE5s4l8NJLVY8payciIiIN0BRz7i4B3gCw1n4P3AusAgqBrdbat4HOwBZrbbAN3Wpgn2gnM8ZcZozJN8bkr1+/vgmGKzGlzJ00M75+/Zg5YACZxcXw2WfVP3DxYti6FXbvhvPPh0cecfunTMH37rv4w0owvWlp+D0eMj/5BF96Ov7s7NAxwD9smLJ2IiIiUm8xDe6MMUfigrsby+93BMYCPYHuQDtjzDjARHl61IXSrLUPW2tzrLU5Xbt2jeVwpSkouJPmaORIt/3ww+ofEwz8fD6wFi69FP7v/9wXEHjggdD8Pa+XwHvvwW9/CwsWEJgxI3Ju3+rVUFbWVO9GREREmqmYBXfGmCzgEWCstXZj+e6jgRXW2vXW2hLgJWA4sAHYq7xUE1wZ55pYjUUSSMGdNEcjRrjtBx9U/5j8fLe98Ua4/353+y9/caWcEybgu/RS/H4/Ho/HzacbPRruvhv698fn89GuvAnRTsD37beYtDQtai4iIiL1EpPgzhizHy5wu8Bauyzs0CrgcGNM2/J5dkcBX1prLfBf4Izyx10EvBKLsUiCKbiT5igY3H38sSu7jCYY3OXkwDXXwLRp4PFAv35w332A65yZl5cXdT7dmDFjMCZaUUOIFjUXERGRmhgXZ9XyIGOmA6OBLsA64DagFYC1dqox5hHgdGBl+VNKrbU55c+9AzgbKAXmAxOstUXGmF7As0Cn8v3jrLU1dCuAnJwcmx/8ACXJKSPDZSp27nS3RZqLAw+E5cth7lwYMiTy2I4dsOeerhxz2zZo29bt/+476Nw5dL8GhYWF9OrVi127dlX7mDZt2rB8+XLNxxMREWnBjDFzg7FWZenRdlZmrT23luMTgAnVHLsNFwxW3r8cGFaX15cUUVLiAjuPB1q3TvRoRGJr5EgX3H34YdXgbuFCl9EbODAykNt33zqfPrjg+bRp0yguLq5yXMsjiIiISG2aolumtFThyyDUUl4mknJqmncXXpLZCDUteF5cXMyUKVM0By8FZWdnR50/qX9LERGJNQV3Ejta406as/COmZXL2YPB3dChjXqJYPYuuCxCbTQHLzXk5ubW+m+qf0sREYkFBXcSO2qmIs3ZgQdCZiasXw/LlkUeCy6D0MjMHdScvassWjZP2Z/kU5d/Uy1cLyIisaDgTmJHwZ00Z8bAqFHu9r//Hdq/bRt89RW0agVZWY1+mfpm78Ip+5Ocavs31XxKERGJFQV3EjsK7qS5u+git50yBUpL3e35812ZZlZWzBoJ1Sd7F07Zn+RV079pKv67aR6hiEhyUnAnsaPgTpq7446DPn3cEgcvv+z2xbAkMyiY6fF4PPTv379OWTxlf5JbRfYuLS1if6r+u7XEeYQKaEUkFSi4k9hRcCfNnccDV1/tbt9/v9vGqJlKZcEFz5955pk6ZfFSMfvT0gQCATyVOgmn6r9bS5xH2BIDWhFJPQruJHaCwV27dokdh0hTuugit2D5Rx/BvHkxWwahMp/Px8yZMxk0aFDEfC2PMVV+cadq9qel8fl8+A8+mGB44AX8p52Wkv9uLXEeYUsMaEUk9Si4k9hR5k5agvbtYfx4d/uOO+CbbyAjA/r1a7KXDP9Q6W3dGm+luX36QJk6PvjuO4JL1BcDU555JmVL+5rbPMLatMSAVkRSj4I7iR2tcyctxS9/6bpnvvqqu5+d7bplNpHwOXjjx493HzDLj+kDZWoZ2a5drX94U6W0r7pgpzn/TLa0gFZEUo+CO4kdZe6kpTjwQDjppND9GJdkRhOcgxcIBAjcemvF3K00Y/SBMoUE2rentssAqRQkNKd5hHXREgNaEUktCu4kdhTcSUty7bWh2zFuphJNcA5eZmam+4B5wAF4AP+oUfpAmUJ8W7YwHppNaZ/P58M/fHgok5yWllLjb4iWFtCKSGpRcCexo+BOWpKf/QyGDHHlmCNGxP3lA8ceSx4QiEPWUGKkrAw2biQAzaq0L5CZWfFhIg1Sbvz15fP58B96aGRjnIsvbtYBrYikDgV3EjsK7qQlMQbeeQcWLYIDDoj7y/uyspgJZP7wQ9xfWxpoyxbYvRtfhw7NqrTPN3cuftwHCv/ee6fc+BsisHt3ZEB70UWJHI6ISAUFdxI7Cu6kpenUCQ45JDGv3auX2y5fnpjXl/pbv95tu3SJ2pgjFbN2rFsHy5YRAJdJrsOajClv/Xp8n32Gv3xZEj+Q+e23iR6ViAig4E5iSevcicSPgrvUEwzuunat0pjDCymZtePjjwHwHXEEM40hs7AQSkoSPKgm9p//QFkZgVGjyNt/fwIAn3yS6FGJiAAK7iSWlLkTiZ/993eloatWNf8P081FWHAHkW3104DA1VcnaGCN8OGHbnv00bDPPrB7N3z3XWLH1NRefhkA39lnM/ORR8gEBXcikjQU3EnsKLgTiZ/WraFHD9ekY9WqRI9G6qJScFexfiHlpX1btyZsaA0WDO5GjICePd3tFSsSN56mtmOHm2sLcMoprlOuMTB/PhQVJXZsIiIouJNY0iLmIvGl0szUsmGD25YHd1C+fmHXrq60b+nShAyrwbZtc0FNWhocfnjLCO7efht27oTDDoPu3aFDBzfvtrgYFixI9OhERBTcSQwpcycSXwruUkulzB2Ur194+eWutC/ewd2vfgVnn+1KKWuzfTu8+GLo9zzAnDkuczxkiJtr3RKCu1decduxY0P7DjvMbVWaKSJJQMGdxIa1aqgiEm8K7lJLWLfMCAcf7LbLlsVvLLt2wf33w/PP1x5Ufvcd5OXBGWfAySdDaanbH16SCc0/uCsthddec7dPPTW0//DD3VbBnYgkAQV3Ehu7drkruBkZkJ6e6NGItAwHHui2Cu5SQ5TMHRAK7uKZufvqK/c7G2ouJ8zPh2HDYOFCd/9//4Pf/Mbd/ugjt03V4K6sDC68EI48Etasqf3xH38MGzfCQQdFLoGizJ2IJBEFdxIbKskUiT9l7lJLdcFdnz5u+/XXoYCrqX3xReh2dcHdSy/ByJGwdi2MHg2vvuou3k2eDE884coyAY44wm1TLbh7/HF46ikXsI4aVXOXz5ISuPFGd/v0010TlaABA6BtW/j229C/sYhIgii4k9hQSaZI/Cm4Sy1RGqoAsOeekJnpKiDi1fn0889Dt6MFd4sWuTLMnTth/Hh46y1Xknnffe74+PHu2CGHhN5P9+7QqpVb2HzHjqZ/D42xYQPccIO73b07fPONC/AKCqI//ve/d5m5ffetyFxmZ2djjMG0aoXZsQMDmG7d3L6wr+zs7Li8JRERUHAnsaLMnUj8deni/s9t2QKbNyd6NFITa6vP3EH8591VDu6sjTz++utu37nnwiOPQPli61x5JVx8cSjDGCzJBNc1c//93e3qgqRk8ZvfwKZNbn2+JUvckgYrVrgA79tvIx87a5YL7oyBJ5+Ejh0ByM3NrViEvjper5fhw4c31bsQEalCwZ3EhoI7kfgzRtm7VPHTTy4z16ZN9AqHeM+7Cy/LXL8eCgsjj3/8sduOGRNZgmgMPPQQHHqou3/MMZHPS4XSzA8+gMcec2tFTpnigrV33oHcXJc5HTQIbrnFXTD58UcYN84Fszfc4MpTy4UvQl+dtLQ0AoFAE78hEZEQBXcSG1rjTiQxgsFd5WyDJJfqOmUGBefdxSO427XL/bx4PBDMKoWXZpaVuWwVhObThcvIgPfec50jTz898liyB3fFxXDFFe72TTe55ijg1qt76y3XBfOnn+Cuu9x7OfZY916ys2HixIhTBRehry575/V68fv9ZGZmNuU7EhGJoOBOYkOZO5HEUOaubv79b1dOGM/lBsLVVJIJ8S3LDHbK7N071OkxPLhbutSVLHbvHiqzrKxDBzjpJBcghku24O799yErC/r3d+vxZWe7rOVBB4UapAS1b+9+TmbNgqOOgq1b3Ty7jAx45plQaWqYmrJ3JSUlTJkypcocPM3JE5GmpOBOYkPBnUhiKLir3ebNLrB74gn34X7q1KpzzJpaXYO7eGTugiWZ/fvD4MHudnhwFyzJHD48siSzLpItuHvoIVi82L3n+fPdNlhampER/Tm5ufDuu/Df/7pF3qdPh759oz60uuyd1+ulX79+mpMnInGn4E5iQ8GdSGIouKvdffe5uVMdO7oujldc4eaSVZ5nVp3PP4ezzgqt69YQ1XXKDOrZ0y0zsGpV03eaDDZTCQ/u5s8PHQ8Gd9FKMmuTbMHd3Llu+/LLbs2+WbNcgHfUUbU/d/RoePbZyAXLo4iWvUsDnnnmGc3JE5G4U3AnsaHgTiQxFNzVbMsWuP9+d/uVV+C551yQ98YbbnHu4O+u6uzc6ZYE+Ne/XEAY3mWyPmrL3KWnhxal//rrhr1GXYUHd4cc4soNv/kGtm1z+2MR3C1fHv/saGWbN7sgMyPD/dsdeqjLyoUvQB4DlbN3XsC/114MysqqcU4ewM6dO/H5fCrTFJGYUXAnsaF17kTiqmKNrUMOcetrrVih+TzR3Hefmzv1s5+5tv1nneXK9AYMgNWrXcldTW6+2c1RS0tz2b8TT3SLetdXbcEdxG/eXTC469fPBXb9+7v7ixa5cX79tVuUO5jVq48uXdzfgR9/TPzyHPPmue2gQS54bkLh2bs0IPDDD/Dyy3XqqBlOZZoi0lgK7iQ2lLkTiSutsVUHW7aEFt2+7bbQ/n32CTXTeOih6jNMM2e656eluflXwVb5J50U6hBcV7V1y4T4zLvbudNl1dLSQq8XPu8u2CVz2DC3IHl9GZM8pZnB4C64bEMTCmbvPB4P/hEjyAS4+WZ8Xbrgz8qi5v+pISrTFJHGUnAn8P337srmQw81/BwK7kTiSmts1cH997us3ZFHwsiRkcfOOAM6d3ZzzT77rOpzt21zTVisddm7ESNcWWevXm4e13nnwe7ddR9LfTJ3TRncLV0a6pTZurXbFx7cNaYkMyhZgrvgfLshQ+LycoFAgLy8PAJPP+1+Tr78EgYOJPDpp3X6sKWlE0QkFhTcCfzzn64cZ9Kkhs+R0Dp3InGlNbZqUV3WLigjA/x+dzvaha1f/xoKClx3zZtvdvu6doXXX3dz9l591TXaCM5Tq01dgrvgWndNWZYZPt8uKNbBXXAeaKKDuzhm7sD9n5w5cyaZ++0XWhNv6VJ8nTrhP/HEWjPtLf5ijIjEhII7gXfecdtVqxreLECZO5G4qyl71yI+KO7aVf0FqdtucwHe6NEwalT0x/ziF2777LOR88Oeew4eftjNR3vyycj1zQ4+GP7zH+jUyW1HjIDvvqt9rLV1ywyeG1x2ramakQSXQejXL7Rv0CC3XbzYdZQ0xpWgNlQyZO62bnVzB73eyPcaL+ec476OPx7mziXwyCOhOXlpaXgrzQFs8RdjRCRmFNy1dDt3wocfhu6//nrDzqPgTiTuKrJ3LfGD4sqV0K2ba5RSuXHH1Knw17+6Jhp33139OXr3hmOOcUHiE0+4fR9/DBdd5G7fc49rvFLZ8OEwZ47LtC1c6BYCz8+vebx1ydx17Qp77eUCkx9+qPl8DRUtc9ehgwvIioqguNgd22uvhr9GMgR3wXX7srKiLj7e5Dwe16znjTfggAMi5uSNGzcOT6X/sy3iYoyIxIWCu3qq6FBXy1fKdKj76CP3wSYtzd2fMaNh51FwJ5IQgUCgZX5QfP55VxL5v/+57Nnq1W7/G2/AVVe52488AocfXvN5rrjCbadOdZmesWNdkHPFFXDNNdU/76CDYPZslxksLHTbNWuiP7aoyHWPTEurOWgyJpS9++qrmsfdUNGCO4jsjNmYkkxIjuAuzvPt6iI4J2/SpEnuokz5fm+rVs3/YoyIxI2Cu3pqdh3qgiWZl1zirnJ//HHD2lcruBNJiIrsXfkFGq/HQ7t27aqsnZXSF6Gi+c9/3LZDBxew5Oa6csqzznINQwKBUAauJiefDN27u1LI4cNh40a33MFf/+qCrZp06gRvveWyhz/95F4/mmBJZpcutZ8zGIz873+1j72+wjtlBuf3BTVFcFdQ4P4tEiHO8+3qomJOXmamuyhT/n82DZr/xRgRiRsFd/XU7DrUBYO7M8+EvDzX/e3tt+t/Hq1zJ5Iw4dm7tLIyxhx/fKMuQiV9hcKmTe5CVHq663Y5fLjL3J1zjvtddP75cMcddTtXejpceqm7vWGDC3Kee67u66J5vaG5e//6V/TH1KUkM+jEE902GLzGUrROmUHhwV1jL07usYdbbqKoCL79tnHnaqgkzNyF8/l8+IcOxQP4+/RR1k5EYqbW4M4Y86gx5gdjzJJqjp9vjFlU/jXLGDOofP/BxpgFYV8/GmOuKz92uzHm+7BjJ8b2bTWdZtWh7ocf3LyEjAwX2AU/VDSkNFOZO5GE8fl8+MePdx8UgbtHj27URaikr1B48013IWrkSJclevddOOUUd2zkSJg2rfYMWbhLL3UXpnr0cEFVfX+PjRkDbdq4Ms1ozVXq0kwl6Gc/c7+T8/Mbtlh6TaoryQTIyXHr2h1wQKjbZWMEm7QE577F0/btLpBNT4eBA+P/+nUUuO468oBAmzaJHoqINCN1ydw9Dhxfw/EVwChrbRYwEXgYwFq71Fo72Fo7GDgU2AH8O+x5k4PHrbUN7OKRGM2mQ91777ntiBHuw8SYMe7+G29EltJ8/LGbf1JTy28FdyIJFQgEyOvViwDge/XVRl2ESvoKhWBW66ST3LZNG3jpJfjgA1cmWTkrVZt99nFz3BYvdrfrq1270MWxF1+serw+mbu2bV2AB+53cSwFO2VGC+66d3cLtb/+ev0C4+oEM4ELFzb+XPW1YIHrNjpgQP1/FuLId9RRzAQyv/yyfmsmiojUoNbgzlr7AbCphuOzrLXBSVpzgB5RHnYU8K21dmWDRplkqsvepVTWDkIlmccc47Z9+7qrths2hBb1/fprF/RNnQrPPBP9PLt3u6YsxrgPWSISdz6fj5mzZpGZlgavv07gyisbfBEqqSsUSkpCQc/JJ4f2p6WFLlQ1RI8ejesQeeaZbhutNLM+wR2ELrTFsjTTWtfhE6IHd+Dm2vXtG5vXC2buEhHcJeF8u6i6dHE/dz/9BN98k+jRiEgzEes5d5cA0S41ngNMr7Tvl+WlnI8aYzpWd0JjzGXGmHxjTP764B/IJBDtynZKZe2srRrcGRP6UDFjhsvG/fznri03hAK+ysIXMI/FFV8RaZi993brapWW4nv/ffznnYe30v/JugZmSVuhMGuWW7/u4IPd3LFkMWaMCyxnzapamtnQ4O7tt93SBLHw0EPw/vsuMzhiRGzOWZPGlGUWF7vKkn/8A373Ozj3XPjVr+qc3Sr68AvmM5miA4fW/7XjLThvNRHlqyLSLMUsuDPGHIkL7m6stN8LnAKEX858CDgQGAwUAn+u7rzW2oettTnW2pyudf3DGAcVV7bLP/x409JSK2u3dKlrQNC1q1sHKCh8Mv8ll7g5Gp06uX3VBXcqyRRJHhde6LZTpxL48EM8lRbDrmtg5vP58J9/PpVzdwmvUHjtNbcNz9olgz32qL40MxjcdelSt3Ptv78rKdy+3ZWaNtacOXDdde72I4+4Esym1ru3CyRXr3YdSOvjqqvg6KPhssvcOoXPPguTJ8Orr9bp6QXv78NWslg5NwGLl9dXsHx1/vzEjkNEmo2YBHfGmCzgEWCstbbyb/ETgHnW2nXBHdbaddba3dbaMuAfwLBYjCPeUrqVcXjWLvzq/JFHutLK+fPdOlJ77umuoKaluUAvmKULp+BOJHmccoorL/zyS3xff42/Y0e8rVoB4AX8xxxT58AssNdeVf5IJDJrV1RYxPwpfSiiY2i+XTKprjSzvpk7CL2/yg2utm93XSjrav16N66SErdu37nn1v25jZGWFmpmUp/STGvdvD9wY73jDpgwwd2/777oz/nwQ9fMpqiIom+3sG7TMMDD2tfLKFpbj+9VIihzJyIx1ujgzhizH/AScIG1dlmUh5xLpZJMY4wv7O7PgaidOJOdz+fDP2aM61C3116pk7WDqiWZQW3ahCbzAzz1lLuyOGCAa7IS7Q+QgjuR5JGRAePGudvHHktg9uzIi1DRSqffeCPUOj5owwZ8Dz+MH0KLLaenJzRrV/B/i9i6szcrvRMavxZbUzjppFBpZnBRdahft8ygaPPulixx3UEPO6xu59i92wVIq1e75Q3+9Ke6v34sNGTeXUGBWwy+c2c3z/vWW+HPf4b27V0WMzifLujVV12H1OHDoUMHCg57AIv7Gbe7YeXEJJ/qH565q5RlFxFpiLoshTAdmA0cbIxZbYy5xBhzuTHm8vKH3Ap0BqaUL2uQH/bctsAxuOAv3D3GmMXGmEXAkcD1sXgziRD4059cK+OtW2M3N6KplZSEFsitHNwBXHyx206cGGovnpPjttFKMxXciSSXSZNcxn3GDHwHH4zf78djDH4g8403Isvk/vUvV054+OGRTZMmTYJt2wiMHp0UFQpFhUWs+9c2wMPa3cdStCEJuwvusQeccIK7HV6a2ZDM3eGHQ8eOrtHGsmVu8fFjj3WB4sKFoXPW5G9/cz8H3bq5SoxalreIuYZ0zPzoI7c94ojQHO4993TTBADuvz/02J074dpr3e1996WoqB3rNg7Fll+OsMWWtY+tTe7s3QEHQIcOZP/wA8bjSd61JUUkZdSlW+a51lqftbaVtbaHtXaatXaqtXZq+fEJ1tqOYcsa5IQ9d4e1trO1dmulc15grR1orc2y1p5irS2M/VuLD1/v3szs04fM0tLQGkLJ7vnn3bIGAwdGb/t9xhnu+C23hPYNLZ+YXlNwpwXMRZJDu3YuA1++EHcgECBvxAgCRx7pLkI9/bR73KpVbl4TQGmpy/g98IDL9DzwAAC+P/8Z/3HHuQqFTp0Sl7WbWIDd7ZZosSYteTMywdLMp592XYShYcFdenooUJw2zQV2hWF/KpfUoeAlOD9x8uSGLfHQWA1pqhIM7vLyIvdffbUL9qZPp2jhauaPmk/RLX9xmb6sLFi+nAL/f7HpkQGs3W2T92cF3HsaPJhcXGa8JgldW1JEUkasu2W2TMEraakwIdraUGnONddU/7jKWbhgcJefX/WxytyJJDWfz8fMmTPJvOoqt+Mf/3DB3Pnn6nTECQAAIABJREFUu86TJ58M99zjjl19tWtmUVTkApUhQwjcd5+rUFi/PtQ9N46KCotY99harHUffm2pSd6MzEknuRLC/HxXzv7qq6FMaefO9TtXsDTznnvg229da/+zz3b7Fi+u+bm7d8Onn7rb4aX28TRwoAtevvii7pUt1QV3vXrB2LFQUkLBhJls/XArK+8rX6XpwQcpWr+bddO3YEsjy45TInuXnU2A2j+QpVRHbhFJGAV3sZBKwd2777oSmb33Ds3LqYvgYrDLlrkPg+EU3ImkhpNPdiV6n38OZ53lPkj7fPDoo3DDDa6Tosfjuul6PK40G/AddBAzhw8n01rXvCLOXNYucj5S0mZk2reHN9+Efv1cQDZ2rLuo1rFjRSa1zo4/PtTw6uCD3dzIYOamtszdkiXud3PPnpCo+eDt28OBB7qpAF9+GdpfVubKgX/4IfLxGze6QDAjA4YMqXq+666jiE6sy+8KFtaWHUPRGZdBXp77GSmLPmctaX9WgrKz8QH+/fZLzrUlRSSlKLiLheC8glTodhWetavPYr9eb6jEpnLjhfB17kQkeXm9cNFF7va//+2yKk8+GWrRf8kl7kN3p05w440uoAgKZn/efz+uQ3ZZu3XYksj9SZ2RGT7c/T2YPNnNF4OGBVidOsEvfuEuIL79tivrHDDAHastuJs9221zc+v/urEUranKlCnu4sL48ZGPnTXLbYcNcxcTKxs5koLO1xMM4SweVra7IvQzUlxNcJfMPytQ8RkiUFaWnGtLikhKUXAXC+GtjMvKEjuWmixY4LpktmsHV1xR/+dXV5qpzJ1I6gi2lQeXrTv66Mjjp53m5oj94Q+R+xsa3JWUNKoLYMpmZFq1cmvLLVvmAuU//rFh55kyxXWI3G8/dz+4vMCSJTV/X5MluKvcVKW4OPS9eP11N2cuqLqSzHJFa4tZ9+PwUMMUvKx9/keW37S82p+RoKT+WenbF7xefAUF+E89FW+ljrbK2olIfSi4i4Vu3dyisNu3uzKcZHXvvW576aWuRKi+qmuqouBOJHX06QM33QQXXFBRdllFtOxBbq7LpixcGGrtX5s333Ql4Oed16ChpnxGBtz7nzQpdouud+3q/uZs2+Ya4lQnWYK7yk1VnnoqtEyEta5ZTFAtwV3BxAKsSYvYZ3dbNs7YWO3PSMXjii1bZ8V/vmidtGpVkZENvPQSnkpBe5rHo6ydiNSZgrtYSfZ5d6tWwbPPuoVlr7uuYeeobjkEBXciqeUPf3DlmPVpjZ+REZrvNXNm7Y9/4AHXEGTzZnjhBReM1FNNWbugpM7INJVg9q66pirr18PXX7t1S7Oy4jeuaMLLMnfvdoEuuHJTcMFdSYlb1uCzz1ypcJSAtLpA3xZbyn4qI7cwl9F2dI1fQ+cPbcp32jjlnyF8xcX49923Yu6dF/AfdJCydiJSZwruYiV8IdJkNHmy+8N69tmw//4NO8chh7iSzlWrIifCK7gTaRnqUppZWuo6bl59tStT79DB7QuurVkPP87+MbUzMk2ltnl3c+a47bBhLiuUSPvu6ypFNm6E++5z6/b16uWC/7593fIOM2a4cv+SEvfe9tqrymlStjy3ri680AXiDz5IYM6cirl3aUBg6VKyBwyodQ08rYMnIqDgLnbC590lk23b4PLL3R9VgF//uuHnSktzrbghct6d1rkTaRnCg7uSEpeRO/po1+Lf53MXjvbf331w93pddjBYKfD22/V+uaHzh7qsy18XM5ojGe1/MvUyMk0hGNxVl7lLlpJMcJm4YPYuuHbqjTe6zqHlaywW3f808y/aTBEdo5ZkNovy3NqMHOmym1deia97d/x+Px6PB3+vXmQWF5NrTLWdNMMtWLCg1mAvOztbgaJIM6bgLlaSsSzzvfdc+c7f/+6u3v7lL6FxNlS00kxl7kRahqFD3UWcr75yQdyZZ7rfM5s2wdq1Lqu/Zo3rvvnee25e37HHuuc2ILirEOzQG7y41NKFN1WJJpmCOwhVtuza5RZTD3ZsvfBCaN2agv/ty9YVe7CSC6MGdy2xPDcQCJCXl0dg2jQwhsDSpXgqNVqpTXWLnufm5tYaKGrBdJHUpeAuVnr2dOVH69a5MpPKSkvhttvcH7bf/95d9Y6l5593pSx77ukm3O+zj7uivnKl+0A0bx5cf33jXydaUxUFdyItQ6tWMGqUu11Y6Mrq/vpX1yBjzRpYscIFfqtWhT6kDxvmfi8tWxbZGbE+5s1zWwV3Tr9+bvvVV1X/lpSWhhYvP/zw+I6rOsHMHbjqkeAyB506UTTmQtZxPOBhLcdT1KfqmFtiea7P52PmzJlkjh4NZ56Jr6QE//77461HmW11yycEAoFql1wIKi4uZsqUKcrsiaSgeq6oKtUyxl2dnDnTZe98vtCxggLXLS54NTUQcGtMPfFEqLymsR57DLaW/2ELNi5o1QpuvdWVwMRq3kX4cgjWgjEUbTZ8wWT67WpLlJWJRKQ5+dOfoH9/OPFEF+jVlk1IT4ejjnK/8955x3XrrY+dO93C1h5P4puDJIv27d0FxRUrXOOUYLAHrlRzxw63eHi3bokbY7jg340uXar8+xfsPg9LMQCWNFY+VkqfnEpPb2llt5Xddhv8618Eli3jsTo+pablE3w+H36/n2nTplFcXFzleDDwK6thaSdl9kSSlzJ3sVS5qcq2ba7t86BBLrDbZx/429/cekXz5sGQIa5rXWPXxrM2lElbvNhlD7/7zrUrv+WW2E6o79XLTY5ft85lC4GCVUeylSxWPtnwtaxEJEX06wf33AOjR9ce2AU1pjQz2GWxXz9o27b+z2+uqmmqUvTGp8xnMkWDf5aAQVWjf3/XrfmttyLmZhcVFrHuLU/YunWtUn/uXFPo1w9+/3t8gwfj79atyjp40aRZS6CG9Wxryt55vd5ayza1qLpI8lJwF0vBEoX773eB3J57ujkFP/4IP/+5+5Dyy1+6AOyyy1w5zc03w7hxjSvTXLHCdSLr1s39Ee3WDXr0cK8fa8ZUTILnnHMouvo21m3NBTys/fcu/VEWkaqCwd2777pArT403y66apqqFDy+211sW398AgZVg7PPdhc0w0SbS9fc5s7FzO9+B/PnE1iwAE/rmmtkvIC/pITMY48NrR1Yic/nw3/ssVQO4bxeL+PHj8fv91cb4GlRdZHkpuAuloYPd8HP+vVu/klGhisjevhhePFF11EOXND197/D66+7eWrTp8Opp7pSmoYIZu2GDq37lfTGuPtu15wlLY2CB7YS/NNsy/RHWUSi6NXLlQlu2VJ1nczaKLiLLkpTlaLCItZ90xvwsPaTTkl9sa2mdeuUvatesKQyGHi1atWKtLTIhd3TWrUikJPj5sWOHu2WQgpfGH37drjhBgKvvlrlQ2BwwfSaMnvK2okkNwV3sXTQQe4q2RtvuGzaTz+5bN2ll0YPuk44wbUU79zZBXrHHec+/NRXcPL8sGGNG39dGQPXX0/Rs++wjhNCJTXF6I+yiER33HFuW9/STDVTiS5KWWbB776q+AxvrUnqi23Nft26JhQeeKWnpzNu3LjQoudeL/5LLyVz1izXvGb3bvjVr9wc2TPOgD593AXme+/Fl5aG/9BDIxdM///27j8+qvLK4/jnTCCIWEUQzfgTqIKCCnFBCuIPqq2KrlS7u2KlYrSrVFG3W7v95ViVdkWt3doqtqXquq4/q9KiW6ttVVQUJAqKikjFUMAkKCigwkCSZ/94ZsgkzCQzk8nc+fF9v1553cmdO3OfXMdhzpznOefQQ6mqqvJB5Hnn7TQFVFk7kcKn4C7Xxo2DU06BgQN9AYDOjB7tA8L99/fbCRN8AYFMJGbu8qju6X1xlW2nh+gfZRFJKot1d9H3NrH4tQuIWv+2FRcFhg6FHj2I/u0jFo+vZfNrm2m8d33Cl22FmwEri7513SievQuFQtTU1HD99de3Nj2PZ9V69vTFjx5+2Bfg+dOf/AyiFSt8z9oxY2D+fCKPPda2Yfobb/hG884R2byZkGv730hZO5HCp+CuEBx6KMyf76ctLVkCN9+c/mObmlqnLeUxuGv9x7ntfv2jLCJJTZjgP1QuWNBa2bcTdVe+xkYOZ1Xfy9oU4hB8k/ihQ6ljChtf/IRl/7gQt72pzSGF+mVbOfaty7UdffAikZ2CvTZZta9+1We/Z870Bd5ee83PKlqwAMaMafvYoUOpamqC73wHfvELwvffT01FxY7sXWVFhbJ2IkVAwV2hOPBAmD3b377+et83Kh3Llvm1eoMG+TLTeaIpNSKSkd139021m5vhmWc6PTxaH6Xxse1AiIZNx+gLoySiB4+hkVPBwWerK3Zk7eIK9cu2cuxbl2s7+uDFAq3EYG8nBx/sWyJNmeLrALQrlLLjsQ8+6L9E+f3vd/TFjdx6K6FYxe2KlhYiV13VvX9YEaiuru6w/5/6AErQFNwVkgkT/Ldsn33m34jTke/1dmhKjYhkKT41c9YsiO78/hCtj7L4+MVEG6L+C6Rm/x7jCOkLoyTq6k/C0XERrUL8sm304tGc4E7o9Kfs+9tloH2wl9VjR4zwVTnBF2C56irC06b5zB5Q4xxV6X7xXMLGjh3baasI9QGUICm4KzQ//amvsnnffX6qZmcCWG+nKTUikpUpU6BvX9/MPEmF4LoZdWx8YSMrv7eSxjsbcC2+CqBrDukLo3ai9VEal1QlZOuSB3n6sk0y8q1vwWmnwbRpcO21AER+9CPG77svEYAHHwx0eIWgo0qicVqbKEFScFdoBg70890BLr+88wbnAWTuNKVGRLIyaJCfkrnXXr7Aw8SJsHkz0DojgBZovKcRF227oFdfGLVVN6MOl+Y/4bp2krbeveHxx+H223cUhQuHw8x76CGqAB56qG1bhTLUvh1Fe6ooKkEzV0T/k44aNcrV1tYGPYzu9+mnvsjKmjVw661wySXJWyls2eLXsbS0+EbpKjggIsVg2TI46STfD3T0aLjiCpY/cgAN/9cSK9LkSJaJCvUOMWblGHpVddzEudRF66MsHLyQlq2dfPmXoM/IPprmKFmprq5myZIlnR43cuRIFi9enIcRBa++vp7BgwezdevWne7r3bs3K1euVHAn3crMXnHOjUp2nzJ3hahPH7jxRn97+nSoqoLJk30z9MQqc0uW+GqZw4crsBOR4nHYYfD8836mwqJFRKdcTuOcLQnVd1NMMVQGCuh4arxVGvtesq/Wr0nOaI3ZzlJl7yp79qRPnz6Ew2EVXJHAKLgrVJMnw/e/D+EwrFvn57lffLGvNhcL8KJ/Wcxi/ovo8GMDHqyISIYGD4YXX4Qf/pC6wdfgrKLTh2j9mApaSf5pjVlyya5LRVMTp51yioJhCZSCu0JlBv/5n7B2rZ/CdNttMGSIv/21r0FzM3X3wEaOZFX9SUGPVkQkc+Ew0UsjNL4/Aud6pPWQcs/eqaCV5JvWmCW347rEfq/EVxS9vl8/BcMSKAV3hc7Mr7+75BJ44gno3x/++Eei039E498OBkI0LNhT39KKSFFKJ1hJVO7FmlTQSoLQUfaunAOVyJVX7vggXdGrFxEgfNtt1Jx+uoJhCUx6X5VKYRg8GB5+GL70Jep+9Qnxf96dM1bNWMWQ24YEOjwRkUx0NsUQVESlPa2dkyDEs1R3/OY3bGtu3rG/3AOVsBk1wK+BmgsvpKpnT7jlFiLz53NXU1PSx5RzMCz5ocxdsTnhBKI/uZ1GTt3R30hrLESkGGmKoUjxiEQihCraro0t+0Dl/feJAOM/9zl/HX7yE/j85wnX11PT0kL73F25B8OSHwruilBd3bG4UNukqz4AiUix0RRDkeIRDoepOe201jVmClSgvp4wMO/EE/116NMHnnsO5swh8uijhHq1nXFQ9sGw5IWmZRaZHdOYWtp+exbP3h0UOUjTl0SkKGiKoUhxiVx5JXfNmQMoUAGgvt5vw+HWffvuC1/5CmGg5oILuGP2bLY1NfmCK+efX97BsOSFMndFpqNpTMreiYiISHcJDx9ODf7DY9ln7QDef99v99036d2RSIRQD59HqQAiZ5+dp4FJOVNwV0TU30hEREQCs/vuRHr0YDw+i1f2kmXuEsQL0YSAGqDq+efzNjQpXwruioiKD4iIiEhgzAjvvTfzgKp2xVXKUjy4S5G5A5+9Gz98OBGARx/Ny7CkvCm4KyIqPiAiIiKBGjDAbz/4INhxFIL4tMwUmTt/V5h5tbVU7bYbLF4MdXX5GZuULRVUKSIqPiAiIiKBUnDXqpNpmQDV1dUsWbKkdcegQUmPGzlyJIsXL87l6KRMKXMnIiIiIulRcOdFo7B+PVRUtF6TJMaOHUtlZfuOd21VVlYybty4XI9QypSCOxERERFJj4I7r6HBb6uqIJT643QkEiHUwf2gthKSWwruRERERCQ9Cu68NKZk+rt9xcxU2Ts1g5dcU3AnIiIiIunZay+//fDDYMcRtE563CXqKHunrJ3kmoI7EREREUmPMndempk7f0jy7J2ydtIdOg3uzOxOM1tnZm+kuP9cM3s99vOimY2I7R9qZksSfjaZ2b/F7utnZn82sxWx7Z65/bNEREREJOcU3HkZBHeQPHunrJ10h3Qyd/8NnNLB/e8BxzvnjgRmAL8BcM4td86NdM6NBP4B+AyYE3vM94C/OucOAf4a+11ERERECpmCOy+DaZmQkL0zA6CyRw9l7aRbdBrcOeeeAzZ0cP+LzrmPYr8uAPZPctiJwLvOuVWx3ycBd8du3w18Je0Ri4iIiEgwFNx5GWbuIJa9q6gAoCL2u0iu5XrN3YXAE0n2TwbuT/h9H+dcPUBsu3eqJzSzi8ys1sxqPyj3NxIRERGRIPXr50v/f/wxbN8e9GiCEw/u0szcQSx7N2ECIaBm//2VtZNukbPgzswm4IO777bbXwmcAfwum+d1zv3GOTfKOTdqQAdNIkVERESkm4VC0L+/v13OFTPj0zIzyNyBz9aNByLlHBhLt8pJcGdmRwK/BSY559a3u/tU4FXnXGPCvkYzC8ceGwbW5WIcIiIiItLNyn1q5vbt/m8PhWDvlJPPkgqPG8e8XXelau1aWN/+I7NI13U5uDOzA4FHga87595Jcsg5tJ2SCTAXmBq7PRX4Q1fHISIiIiJ5UO7BXUOD3+6zD8TW0KWtogKOPNLffu213I5LhPRaIdwPvAQMNbM1ZnahmU0zs2mxQ64G+gOzYi0PahMeuyvwJXzwl2gm8CUzWxG7f2YO/hYRERER6W7lHtxlUUyljZEj/XbJktyMRyRBj84OcM6d08n93wC+keK+z/CBX/v96/EVNEVERESkmCi481sFd1KAcl0tU0RERERKWbkHdxn2uNuJgjvpRgruRERERCR95R7cdTVzd8QRvhjLsmWwdWvuxiWCgjsRERERyYSCO7/NNnO3664wZAg0NcFbb+VuXCIouBMRERGRTJR7cJdlj7s2NDVTuomCOxERERFJX9DBnXNw+eVw5ZXQ0pL/83d1WiYouJNuo+BORERERNIXdHBXWwu//CXcfDNcc03+z9+FgirV1dWYGfa972GA/fKX/vd2P9XV1bkds5QNBXciIiIikr7+sS5X69dDc3P+z//AA623Z8yABx/M37mbmmDdOjDzTcwzNHbsWCorKzs8prKyknHjxmU7QilzCu5EREREJH09e0Lfvn565IYN+T13S0trMDdlit+ef77P5uXDunX+7x4wAHp02i56J5FIhFCo44/fFRUVRCKRbEcoZU7BnYiIiIhkJj4188MPUx/zxBPw2GO5Pe/8+bB2LRx0ENx9N1x4oW8nMGlS63TJ7tTFHnfhcJiampqU2bvKykpqamqoqqrKdoRS5hTciYiIiEhmOlt3t307nHWW//n449ydNz4l8+yzfa+4WbPg2GN90HXmmbBlS+7OlUwOiql0lL1T1k66SsGdiIiIiGSms+Du73/3GbWmJnjlldycs6kJfvc7f/ucc/y2shIeecRn8l5+Gf71X/20yVzZuBFOO803Hh8yBC64wO/vQnC3I3vXblqnsnaSCwruRERERCQznQV3K1e23l60KDfnfOYZf76hQ2HEiLZjmTsX+vSBe++FG27IzfnAP+8f/whvvAErVrROQz3++C49bSQSIdQuuFPWTnJBwZ2IiIiIZCaI4C4+JXPyZF+tMtGRR/rAzgx+8AMflOXCyy/77be/DcuXw3vv+QDvvPO69LThcJg+ffq02bdlyxbC4bDaIkiXKLgTERERkczkO7iLRuHRR/3ts89OfsykSfDjH/tpmeeeC2++2fXzxoO7iRP9tMyBA1tbQXTRaaed1ukxaosgmVJwJyIiIiKZySS4W70aGhq6dr4nn/SFWUaMgMMOS33c97/vM3uffAK33NK1c0ajsGSJzwb+wz907bmSmDlzJhUVFR0eo6makikFdyIiIiKSmc6Cu/fe89vdd/fbrmbvEqdkdsTM970DePfdrp3z9ddh2zY49FDYY4+uPVcS4XCYc889F2s/xTRGBVYkGwruRERERCQz6WbuzjzTb7sS3K1fD3Pm+NuppmQmOvBAv/3737M/J8DChX579NFde54OzJw5k169eiW9T1k7yYaCOxERERHJTEfB3Ucf+Z8+fXwbAehacPfb3/q2CqeeCoMGdX58PLhbvRpaWrI/b3y9XTcGd6mamitrJ9lScCciIiIimYkHdx9+uHNfufiUzEGDYPRof3vRouz6zzU1+UblAJddlt5j+vTxRU+iUVi3LvNzxuUhuIPkTc2VtZNsKbgTERERkczssgvsthts3+4bfSeKT8kcPNg3F99rLz+1sq4u8/PMneunVx5yCJx8cvqPO+ggv812aubHH/vWB716+TYL3ah99q4SqDnuOGXtJCsK7kREREQkc6mmZiYGd2atma94JiwTv/yl306fDqEMPrbGp2auWtX5sZGIb53Q1NS6r7bWb6urod2Uye6QmL2rACLtGpyLpEvBnYiIiIhkLlVwF5+WOXiw3yZOzczE0qXw7LM+QxivgJmudDN327bBzJlw333w0EOt+/M0JTMunr0LmVEDVP3lL7BpU17OLaVFwZ2IiIiIZC6dzB1kH9zdeqvfTp3a2lIhXelm7t55pzVjN3Nm67rAPAd34LN34489lsjYsX69YLxCqEgGFNyJiIiISOb2289v33mn7f5Uwd0rr0Bzs7/9ySfw61/D2rXJn3vDBrjnHn97+vTMx5ZuO4Q332y9vXQpPPGED/Dy0AahvYkTJ/Lcc88RfuklDLDzz8fMdvqprq7O25ik+Ci4ExEREZHMjR/vt88+27qvubm1cMrAgX67995+muSnn8Lbb0NDA5xwAkyb5vvgJauieeedsGULfPnLvol4puLTMjvL3L31VusYAW64geirq1jc8F2iuw+Egw/O/NxZGjt27E4tEdqrrKxk3LhxeRqRFCMFdyIiIiKSuQkT/Pb551unNq5Z42+Hw9C7d+ux8ezdPffAF77gs3jgp2o++WTb5928GW680d++/PLsxpZp5u6aa6BvX3juOeoufJaNHMmq3af7gjB5kqwlQntqkSCdUXAnIiIiIpnbf3+f2dq8uTVYaz8lMy4e3N1wg8+mjRkD//Efft+117bN3t10k1/HN3YsTJyY3dj23tu3MdiwwU8BTSUe3H3hC3DppUTpR+NrYSBEQ8NIog3R7M6fhVQNzePU2FzSoeBORERERLITz94984zftq+UGRcP7gAmTYKnn/YtCPr3hwUL4K9/9fe9/z7cfLO//dOfZp85M+s8exeNwooVvsXCoYfC5ZdTV1GDw5/TEWLVjDRaKeRQR9m7Qs3aVVdXJ10bqLWCwVBwJyIiIiLZaR/cpcrcjR8P//zPcNVV8MgjsOuuvsXBt7/t749n7665Bj77DM46C7q6tqyzdgjvvOPXCA4eDL17E23eg0ZOxeEzZ67JaLiroSCyd4WctdNawcKi4E5EREREshMP7l54wfeMSxXc9ezp+8jNmAEVFa37L70U9tzTP37WLLjjDn//9dd3fWydtUOIT8kcPhyAuhl1uFBFm0NcsyuI7F2hZu2gcNcKlmtGUcGdiIiIiGSnqgoOO8xn2xYtag3uBg1K7/G77w7f+pa/PX06tLTAxRfDkCFdH1tn0zITgrtofZTGuxpx29se4ra5wLN3lUDNeecVZNYOCnetYLlmFBXciYiIiEj2EqdmpsrcdeSyy2CPPfzt3XaDq6/Ozbg6a4eQENzVzajDtSRpyUDw2bsKIHLSSXk9f6YKca1goWYUu5uCOxERERHJXjy4e+wxX+WyVy/fCiFdffu2Vs686irYZ5/cjCvNzF1078N81m5biuAuwOxdCKgBqmpr83bubBTiWsHOMooAW7ZsIRwOl9Q0TXPJGkcWqFGjRrnaAn9xi4iIiJSVDz+EAQOI0o+3iDDs4PvptWJhZs/hnG8oPmxY7nrLvfuub9Vw0EGtjdXjtm6FPn0AWH7hUhru/iBlcAdglUb4G2GG3JaD6aJpqq+vZ/LJJ/Pg0qVUjRwJixfn7dzZqK+vZ/DAgWzdtm3Hvt6VlaxctSqwKaX19fUMHjyYrVu3Zv0cI0eOZHGBXXsze8U5NyrZfcrciYiIiEj29toLjjiCOr7um39vm5z5c5j5wia5bBq+//5+G2+snuidd/z6voMPZtOiTzsM7MBn7za+uDF3Y0tDOBxm3ssvU7XLLrBkCTQ25vX8mQqHw9QMGEA8T1YJ1DhH1UcfBTemPfagZvhwOl55l1oxrsnrEfQARERERKS4RY8+hcalJwIhGtYewUENUXpV9Qp2UPHpofX1/ueAA1rvi6+3GzaM0XNGJ398IdhlFzj+eHjySfjzn2HKlKBHlNqKFUTWruWu2K8VFRVEtm+HM8+El1/2xXPy5c03fXXWu+4isnr1jjFlqhjX5ClzJyIiIiJdUrf2xECbf6eUqh1CuzYIBe10b7teAAAQL0lEQVTkk/32qaeCHUdnfv1rwkDN0KGEQiFqLriAqsMPh+XLfd/CZcu69/zbt/sWGsOGweGHw3XXwerVhEeOpOYrX+m0cmZ7hdxbsCMK7kREREQka9H6KI3P7tLa/Ls5lPcCJCmlKqpSTMHdl7/st0895aeSFqItW+Aunx+L/OxnjB8/nsh118GcOb6P4V//6oOuSZNg/nz/d2zaBGvXwooV/vFdPf9ZZ8EPfuCDyP794Rvf8BnP2lois2a1Vh+tqKCyZ89On7IYs3ag4E5EREREusC3EWi7L4j2AUmlaodQTMHdsGGw335+zd3rrwc9muQefhg2bICjjiJ86qnMmzfPZ7wOPhhefRW++U0/xXTuXBg/3jeq32MPvy5yyBA/ZbO6Gi66CO6+G5qb0z/3xo0+u/n449CvH/zhD34a7uzZPjCuqGitPhoKMWXKFEIVFR0+ZbFm7UDBnYiIiIhkaUfz73YFSYJoH5BUksxd9L1NLF5xKdHQXjB0aEADy4BZ2+xdIbr9dr/95jd3LoozcCDMmuUD7Kuu8gEY+Gql4bC/v6XFF42ZPRvOP99n4NLR2AgnnADPP+8D4OefhzPOgCSZuUgkwvjx45k5c2abFgk9e/akol2wV6xZO0gjuDOzO81snZm9keL+c83s9djPi2Y2IuG+vmb2sJm9bWbLzGxsbP81ZrbWzJbEfibm7k8SERERkXwotObfO0mSuav77hts5AhWfe5SX3SlGMTX3T35ZLDjSOa11+Cll3wm7pxzUh+3994wY4bvhdjcDJ98Au+/D++957Nvzz0HP/6xP/bWW/1xqSxd6nsjjhzpg8JDDvHTPYcNS/mQcDi8I6OY2OC8R48eTJkyhcpYgFcZChVt1g7Sy9z9N3BKB/e/BxzvnDsSmAH8JuG+W4A/OecOBUYAiSsp/8s5NzL288fMhi0iIiIiQUqVtYsriOxdu8xdtD5K4++jQIiGzccEn1lM10kn+YzYCy/47FR32rbNZwqnTPH9ADsze7bfnnfejt6BHQqF/E+i3XaDY4+FH/4QTjsNPvsMfvaznR/76KMwYgQceSTcdBM0NMDRR/trEg/k05A4TbOmpobrr79+x1TNCueKNmsHaQR3zrnngA0d3P+icy7ewGIBsD+Ame0OHAfcETtum3Pu4y6PWEREREQC11HWLi7w7N1BBxGlH4uXXUy0Phobs18g6KyAqnp2pn9/+Md/9IHXccfBJZf4giTd4YknfNuFe+/1BVA++yz1sc3Nfr0d+OmUuRAPrG69Fdavb93/wgvwL//i1x326+engM6fDwsWwD77ZHEaP00zEon4YG/KFEJAjRlV8amjRSjXa+4uBJ6I3R4MfADcZWaLzey3ZpYYzk+PTeW808z2TPWEZnaRmdWaWe0HHaVnRURERCRvNr20qSCbf7fRty91PS5kY8twVn77bRrvbMA1+wyNa64IPrOYiQcf9IFPjx5+jduwYd0zTfO++1pvP/UUnH46fPpp8mPnz/fr3gYP9gVRcmHMGD8N9ZNP4Oc/9/vWrYOzz/bB5OWX+4Ips2bBuHFZN75PnKYJEPnxjxnfuzeRlpbCLVyTBnOu4/8pAcxsIPC4c+7wDo6ZAMwCxjvn1pvZKHwm7xjn3EIzuwXY5JyLmNk+wIeAw0/lDDvnLuhsHKNGjXK1tbVp/FkiIiIiUu6i9VEW7jePFlcJ1oK5ZhytxTas0gh/I8yQ24YEOMoMLV3qy/y//DJUVkJtLRxxRG6ee9MmnwXbutUHjlOn+qmPxx4L//d/8LnPtT3+iivgF7+A73wHbrwxN2MAePFFOOYYX0Vz5UqYPBn+8hdfafPpp5MWTMmJqVPhf/4HbrvNZ0cLlJm94pwbley+nGTuzOxI4LfAJOdcPH+6BljjnFsY+/1h4CgA51yjc67ZOdcCzAaOzsU4RERERETi6mbU4eIfd521CeygQNYFZuqII3zwU1Pjp2lOmQLRduOfP9+vRZsxwzf3Ttfvf+8Du+OO8+vu5s1rrUJ5Qbs8TEsLPPKIv/1P/9S1v6m9cePgxBN9sHnMMT6wGzAAHnig+wI78NcMfOBcpLoc3JnZgcCjwNedc+/E9zvnGoDVZhavMXsi8FbsMeGEpzgTSFqJU0REREQkGzsKvrgesT3Jp+8Fvi4wGxUVPmP2+c/7KYRXX91638KFcOqpsGiR3z9mTPrTDO+912+/9jW/HTIEnn3W96h7+GFfmTLxPGvXwgEHwOjROfmz2oj/TcuX+6mX993nA83uFP87Sjm4M7P7gZeAoWa2xswuNLNpZjYtdsjVQH9gVqytQeK8ycuAe83sdWAk8J+x/Tea2dLY/gnAt3L1B4mIiIiIpFPwBYo0ewe+wuQ99/jKkzfd5FsJvPKKX6+2ebOvOjlwICxeDKNGwbXX+kApVYPwxkafIevZs20m7uCDYVrsY/9117Xuj2ftvvrVrNe9dei442DCBH/7mmt8xdDuNmKE//vffrv7CtZ0s7TW3BUKrbkTERERkc5E66MsHLyQlq0taR1flGvv4iIR3x/ugAN84ZMNG3xwdv/9sGWL7wf3q1+1Ht+7Nwwf7tevXXutX9cGPhN4xRW+KufcuW3PUV/vi6Zs3QpLlhAdMJS3Bv83w6I/oNcLj/mpk93hgw98BvKUU3Zun9Bdjj7an/Ppp1uDywLT7WvuREREREQKRbpZu7jAq3p2xdVXw1FHwerVPrA74ww/hbFHD18A5fbbfUbu9NN9ALhliy/C8vOfwxe/2NosPD4l89xzdz5HONwme1d3eS0bo0NYtes0GDu2+/62AQNg4sT8BXZQ9OvulLkTERERkZKRTtYu1DvEmJVj6FXVK48j60bLlvnpmGPGwP/+L/Tq4O/66CN49VW46CJfiXLoUB8AfvGLfqpnYyPsuuvOj4tl76Jbd2Vh6AFaWnoSqmhizJpjS+c6gq+WOXUqnHVW69TTAqPMnYiIiIiUhaJorp5rhx0GdXXwu991HNgB7Lmnr0T5wgu+8uby5a3r2c48M3lgBz57d/HF1PH1HdfXWY/Suo5Q9Jk7BXciIiIiUjKKorl6d8h06mI47FsdjBvn2xpAa5XMFKJT/51GTsVRCYBrojiL0XRkyBC/DnHNGp+tLDI9Oj9ERERERKQ4jF7cDWX5S9Wee8JTT/kpmhs3dlqRsm72FlyoByTMeI1nQYuyGE0yoZCvLvr0076wyhlnBD2ijChzJyIiIiJSrvr08cVUHn/cF2FJYUffwJaKNvuLtpVER4p4aqaCOxERERER6VBHaxlLbg2jgjsRERERESlFO7J2KdYyllz2Lh7cLVrUuh6xSCi4ExERERGRlMquAul++/nm8L/6VdEFdyqoIiIiIiIiKZVlBdLrrgt6BFlRcCciIiIiIimpAmnx0LRMERERERGREqDgTkREREREpAQouBMRERERESkBCu5ERERERERKgII7ERERERGREqDgTkREREREpAQouBMRERERESkBCu5ERERERERKgII7ERERERGREmDOuaDHkDYz+wBYFfQ4ktgL+DDoQZQxXf/g6NoHS9c/OLr2wdL1D5auf3B07YNVKNf/IOfcgGR3FFVwV6jMrNY5NyrocZQrXf/g6NoHS9c/OLr2wdL1D5auf3B07YNVDNdf0zJFRERERERKgII7ERERERGREqDgLjd+E/QAypyuf3B07YOl6x8cXftg6foHS9c/OLr2wSr46681dyIiIiIiIiVAmTsREREREZESoOCuC8zsFDNbbmZ/M7PvBT2eUmdmB5jZM2a2zMzeNLMrYvuvMbO1ZrYk9jMx6LGWKjOrM7OlsetcG9vXz8z+bGYrYts9gx5nqTGzoQmv7yVmtsnM/k2v/e5jZnea2TozeyNhX9LXunm/iP1b8LqZHRXcyEtDiut/k5m9HbvGc8ysb2z/QDPbkvD/wa+CG3nxS3HtU77XmNn3Y6/95WZ2cjCjLh0prv+DCde+zsyWxPbrtZ9DHXzOLKr3fk3LzJKZVQDvAF8C1gCLgHOcc28FOrASZmZhIOyce9XMPge8AnwF+BfgE+fcTwMdYBkwszpglHPuw4R9NwIbnHMzY19y7Omc+25QYyx1sfeetcAYoAa99ruFmR0HfAL8j3Pu8Ni+pK/12Afdy4CJ+P8utzjnxgQ19lKQ4vp/GXjaOddkZjcAxK7/QODx+HHSNSmu/TUkea8xs2HA/cDRwL7AX4AhzrnmvA66hCS7/u3uvxnY6Jy7Tq/93Orgc+b5FNF7vzJ32Tsa+JtzbqVzbhvwADAp4DGVNOdcvXPu1djtzcAyYL9gRyX41/3dsdt3498IpfucCLzrnFsV9EBKmXPuOWBDu92pXuuT8B/EnHNuAdA39iFBspTs+jvnnnLONcV+XQDsn/eBlYEUr/1UJgEPOOeizrn3gL/hPx9Jljq6/mZm+C+078/roMpEB58zi+q9X8Fd9vYDVif8vgYFGnkT+7aqGlgY2zU9lhK/U9MCu5UDnjKzV8zsoti+fZxz9eDfGIG9AxtdeZhM23/Y9drPn1Svdf17kH8XAE8k/D7IzBab2TwzOzaoQZW4ZO81eu3n17FAo3NuRcI+vfa7QbvPmUX13q/gLnuWZJ/muOaBme0GPAL8m3NuE3A78HlgJFAP3Bzg8ErdMc65o4BTgUtj00ckT8ysEjgD+F1sl177hUH/HuSRmf0QaALuje2qBw50zlUD/w7cZ2a7BzW+EpXqvUav/fw6h7Zf7um13w2SfM5MeWiSfYG//hXcZW8NcEDC7/sD7wc0lrJhZj3x/8Pd65x7FMA51+ica3bOtQCz0ZSQbuOcez+2XQfMwV/rxvg0hNh2XXAjLHmnAq865xpBr/0ApHqt69+DPDGzqcDpwLkuVjQgNiVwfez2K8C7wJDgRll6Oniv0Ws/T8ysB3AW8GB8n177uZfscyZF9t6v4C57i4BDzGxQ7Nv0ycDcgMdU0mJzze8AljnnfpawP3F+85nAG+0fK11nZn1iC4wxsz7Al/HXei4wNXbYVOAPwYywLLT51lav/bxL9VqfC5wXq5z2BXyxg/ogBljKzOwU4LvAGc65zxL2D4gVGsLMBgOHACuDGWVp6uC9Zi4w2cx6mdkg/LV/Od/jKxMnAW8759bEd+i1n1upPmdSZO/9PYIeQLGKVeuaDjwJVAB3OufeDHhYpe4Y4OvA0ngZYOAHwDlmNhKfCq8DLg5meCVvH2COf++jB3Cfc+5PZrYIeMjMLgT+DvxzgGMsWWa2K746b+Lr+0a99ruHmd0PnADsZWZrgB8BM0n+Wv8jvlra34DP8FVMpQtSXP/vA72AP8fehxY456YBxwHXmVkT0AxMc86lWxBE2klx7U9I9l7jnHvTzB4C3sJPlb1UlTK7Jtn1d87dwc7rrUGv/VxL9TmzqN771QpBRERERESkBGhapoiIiIiISAlQcCciIiIiIlICFNyJiIiIiIiUAAV3IiIiIiIiJUDBnYiIiIiISAlQcCciIiIiIlICFNyJiIiIiIiUAAV3IiIiIiIiJeD/AV71fdgu0WdLAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize = (15,5))\n",
    "plt.plot(closes, color='r', lw=2.)\n",
    "plt.plot(closes, '^', markersize=10, color='m', label = 'buying signal', markevery = states_buy)\n",
    "plt.plot(closes, 'v', markersize=10, color='k', label = 'selling signal', markevery = states_sell)\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model\n",
    "### Load Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading model =>  cnn-EURUSD_buy_sell_w10-ep500-2020_06_11.model\n",
      "[Model] Loading model from file saved_models/cnn-EURUSD_buy_sell_w10-ep500-2020_06_11.model\n",
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 15, 15, 32)        896       \n",
      "_________________________________________________________________\n",
      "batch_normalization (BatchNo (None, 15, 15, 32)        128       \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 7, 7, 32)          0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 7, 7, 32)          0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 7, 7, 64)          18496     \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 7, 7, 64)          256       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 3, 3, 64)          0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 3, 3, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 576)               0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 128)               73856     \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 128)               512       \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "batch_normalization_3 (Batch (None, 64)                256       \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 3)                 195       \n",
      "=================================================================\n",
      "Total params: 102,851\n",
      "Trainable params: 102,275\n",
      "Non-trainable params: 576\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = CNNModel()\n",
    "\n",
    "if pre_trained_model_filepath:\n",
    "    print(\"Loading model => \", pre_trained_model_filepath)\n",
    "    model.load_model('saved_models/' + pre_trained_model_filepath)\n",
    "    model.model.load_weights('saved_models/' + configs['model']['pre_trained_weights'])\n",
    "    model.model.summary()\n",
    "else:\n",
    "    model.create_model_cnn(\n",
    "        configs['model'], \n",
    "        input_shape=(x_train[0].shape[0], x_train[0].shape[1], x_train[0].shape[2])\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow.keras.backend as K\n",
    "import tensorflow as tf\n",
    "\n",
    "K.set_session(tf.Session(graph=model.model.output.graph))\n",
    "init = tf.global_variables_initializer()\n",
    "K.get_session().run(init)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In memory training\n",
      "[Model] Training Started\n",
      "[Model] 2 epochs, 32 batch size\n",
      "Train on 78543 samples, validate on 5640 samples\n",
      "Epoch 1/2\n",
      "78543/78543 [==============================] - 26s 333us/sample - loss: 1.1050 - acc: 0.3513 - f1_score: 2.7317 - precision: 0.1071 - recall: 0.0037 - val_loss: 1.0983 - val_acc: 0.3619 - val_f1_score: 2.9928 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00\n",
      "Epoch 2/2\n",
      "78543/78543 [==============================] - 26s 335us/sample - loss: 1.0988 - acc: 0.3607 - f1_score: 2.9106 - precision: 0.0065 - recall: 2.1938e-04 - val_loss: 1.0978 - val_acc: 0.3635 - val_f1_score: 2.9892 - val_precision: 0.0000e+00 - val_recall: 0.0000e+00\n",
      "[Model] Training Completed. Model saved as saved_models/cnn-25072020-214942-e2.h5\n",
      "CPU times: user 1min 58s, sys: 13.3 s, total: 2min 11s\n",
      "Wall time: 52.5 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "if not generative_training:\n",
    "    print(\"In memory training\")\n",
    "    hist = model.train(\n",
    "        x_train,\n",
    "        y_train,\n",
    "        epochs=configs['model']['epochs'],\n",
    "        batch_size=configs['model']['batch_size'],\n",
    "        save_dir=configs['model']['save_dir'],\n",
    "        x_val=x_test,\n",
    "        y_val=y_test,\n",
    "        verbose=True\n",
    "    )\n",
    "else:\n",
    "    print(\"Out of memory training...\")\n",
    "    steps_per_epoch = math.ceil(dataloader.len_train / configs['model']['batch_size'])\n",
    "    validation_steps = math.ceil(dataloader.len_test / configs['model']['batch_size'])\n",
    "    hist = model.train_generator(\n",
    "        data_gen=dataloader.generate_train_batch(\n",
    "            cols,\n",
    "            batch_size=configs['model']['batch_size'],\n",
    "            window_size=img_size\n",
    "        ),\n",
    "        epochs=configs['model']['epochs'],\n",
    "        batch_size=configs['model']['batch_size'],\n",
    "        steps_per_epoch=steps_per_epoch,\n",
    "        save_dir=configs['model']['save_dir'],\n",
    "        validation_data=dataloader.generate_test_batch(\n",
    "            cols,\n",
    "            batch_size=configs['model']['batch_size'],\n",
    "            window_size=img_size\n",
    "        ),\n",
    "        validation_steps=validation_steps\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "f1_score\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "precision\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "recall\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Loss\")\n",
    "plot_loss(hist.history['loss'], hist.history['val_loss'])\n",
    "print(\"Accuracy\")\n",
    "plot_loss(hist.history['acc'], hist.history['val_acc'])\n",
    "print(\"f1_score\")\n",
    "plot_loss(hist.history['f1_score'], hist.history['val_f1_score'])\n",
    "print(\"precision\")\n",
    "plot_loss(hist.history['precision'], hist.history['val_precision'])\n",
    "print(\"recall\")\n",
    "plot_loss(hist.history['recall'], hist.history['val_recall'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "78543/78543 [==============================] - 7s 89us/sample\n",
      "5640/5640 [==============================] - 0s 88us/sample\n"
     ]
    }
   ],
   "source": [
    "predictions_train = model.predict(x_train,verbose=True)\n",
    "predictions_test = model.predict(x_test,verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions_train = np.argmax(predictions_train, axis=1)\n",
    "predictions_test = np.argmax(predictions_test, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_right_train = np.array(predictions_train == y_train)\n",
    "pred_right_test = np.array(predictions_test == y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train acc  40.00229173828348\n",
      "Test acc  38.297872340425535\n"
     ]
    }
   ],
   "source": [
    "print(\"Train acc \", np.sum(pred_right_train) / pred_right_train.shape[0] * 100.0 )\n",
    "print(\"Test acc \", np.sum(pred_right_test) / pred_right_test.shape[0] * 100.0 )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}