3_absolute_model_fit.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os\n",
    "from bambi import Model\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.patches as mpatches\n",
    "%matplotlib inline\n",
    "\n",
    "from plotting_functions import plot_correlation, cm2inch\n",
    "from analysis_functions import compute_gaze_influence_score, compute_mean_rt, compute_p_choose_best\n",
    "from analysis_functions import make_sure_path_exists\n",
    "from analysis_functions import q1, q3, iqr, std, se, add_best_chosen"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.random.seed(52) # from random.org"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data I/O"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# load all data and divide in even / odd\n",
    "all_data = pd.read_csv('data/data_aggregate.csv')\n",
    "even = all_data[(all_data['trial'] % 2) == 0].reset_index(drop=True)\n",
    "odd = all_data[(all_data['trial'] % 2) == 1].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/armin/anaconda3/lib/python3.5/site-packages/ipykernel/__main__.py:15: FutureWarning: Sorting because non-concatenation axis is not aligned. A future version\n",
      "of pandas will change to not sort by default.\n",
      "\n",
      "To accept the future behavior, pass 'sort=False'.\n",
      "\n",
      "To retain the current behavior and silence the warning, pass 'sort=True'.\n",
      "\n",
      "/Users/armin/anaconda3/lib/python3.5/site-packages/ipykernel/__main__.py:16: FutureWarning: Sorting because non-concatenation axis is not aligned. A future version\n",
      "of pandas will change to not sort by default.\n",
      "\n",
      "To accept the future behavior, pass 'sort=False'.\n",
      "\n",
      "To retain the current behavior and silence the warning, pass 'sort=True'.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# load prediction daat\n",
    "prediction_multiplicative_list = []\n",
    "prediction_nobias_list = []\n",
    "\n",
    "# join prediction dataframes\n",
    "for subject in odd['subject'].unique():\n",
    "    subject_pred_multiplicative = pd.read_csv(os.path.join('results', 'predictions', 'out_of_sample', 'multiplicative', 'prediction_{}_multiplicative_oos.csv'.format(subject)), index_col=0)\n",
    "    subject_pred_nobias = pd.read_csv(os.path.join('results', 'predictions', 'out_of_sample', 'nobias', 'prediction_{}_nobias_oos.csv'.format(subject)), index_col=0)\n",
    "    subject_pred_multiplicative['subject'] = subject\n",
    "    subject_pred_nobias['subject'] = subject\n",
    "\n",
    "    prediction_multiplicative_list.append(subject_pred_multiplicative)\n",
    "    prediction_nobias_list.append(subject_pred_nobias)\n",
    "\n",
    "prediction_multiplicative = pd.concat(prediction_multiplicative_list)\n",
    "prediction_nobias = pd.concat(prediction_nobias_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# discard trials with negative predicted RT\n",
    "prediction_multiplicative = prediction_multiplicative[prediction_multiplicative['rt'] > 0].copy()\n",
    "prediction_nobias = prediction_nobias[prediction_nobias['rt'] > 0].copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Compute behavioral indices for observed data and predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def aggregate_subject_level(data, n_items):\n",
    "    \"\"\"\n",
    "    Aggregates a single dataset to subject level\n",
    "    \"\"\"\n",
    "    data = data.copy()\n",
    "    \n",
    "    # add best chosen variable\n",
    "    data = add_best_chosen(data)\n",
    "    \n",
    "    # Summarize variables\n",
    "    subject_summary = data.groupby('subject').agg({'rt': ['mean', std, 'min', 'max', se, q1, q3, iqr],\n",
    "                                                   'best_chosen': 'mean'})\n",
    "    # Influence of gaze on P(choice | value)\n",
    "    subject_summary['gaze_influence'] = compute_gaze_influence_score(data, n_items=n_items)\n",
    "    \n",
    "    subject_summary['dataset'] = data.groupby('subject')['dataset'].head(1).values\n",
    "    \n",
    "    return subject_summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# compute behavioral indices\n",
    "odd_summary_list = []\n",
    "pred_multiplicative_summary_list = []\n",
    "pred_nobias_summary_list = []\n",
    "\n",
    "for di, dataset in enumerate(odd['dataset'].unique()):\n",
    "    odd_di = odd[odd['dataset'] == dataset].copy()\n",
    "    n_items = odd_di['n_items'].unique()[0]\n",
    "\n",
    "    odd_summary_di = aggregate_subject_level(odd_di, n_items=n_items)\n",
    "\n",
    "    prediction_multiplicative_di = prediction_multiplicative[prediction_multiplicative['dataset'] == dataset].copy()\n",
    "    prediction_nobias_di = prediction_nobias[prediction_nobias['dataset'] == dataset].copy()\n",
    "\n",
    "    pred_multiplicative_summary_di = aggregate_subject_level(prediction_multiplicative_di, n_items=n_items)\n",
    "    pred_nobias_summary_di = aggregate_subject_level(prediction_nobias_di, n_items=n_items)\n",
    "\n",
    "\n",
    "    odd_summary_list.append(odd_summary_di)\n",
    "    pred_multiplicative_summary_list.append(pred_multiplicative_summary_di)\n",
    "    pred_nobias_summary_list.append(pred_nobias_summary_di)\n",
    "\n",
    "odd_summary = pd.concat(odd_summary_list)\n",
    "pred_multiplicative_summary = pd.concat(pred_multiplicative_summary_list)\n",
    "pred_nobias_summary = pd.concat(pred_nobias_summary_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# save files, if not existing\n",
    "odd_oos_summary_path = os.path.join('results', 'predictions', 'out_of_sample', 'descriptives',\n",
    "                                                'subject_summary_odd_oos.csv')\n",
    "multiplicative_oos_summary_path = os.path.join('results', 'predictions', 'out_of_sample', 'descriptives',\n",
    "                                                'subject_summary_multiplicative_oos.csv')\n",
    "nobias_oos_summary_path = os.path.join('results', 'predictions', 'out_of_sample', 'descriptives',\n",
    "                                        'subject_summary_nobias_oos.csv')\n",
    "\n",
    "if not os.path.isfile(odd_oos_summary_path):\n",
    "    odd_summary.to_csv(odd_oos_summary_path)\n",
    "if not os.path.isfile(multiplicative_oos_summary_path):\n",
    "    pred_multiplicative_summary.to_csv(multiplicative_oos_summary_path)\n",
    "if not os.path.isfile(nobias_oos_summary_path):\n",
    "    pred_nobias_summary.to_csv(nobias_oos_summary_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "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 tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"8\" halign=\"left\">rt</th>\n",
       "      <th>best_chosen</th>\n",
       "      <th>gaze_influence</th>\n",
       "      <th>dataset</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>max</th>\n",
       "      <th>se</th>\n",
       "      <th>q1</th>\n",
       "      <th>q3</th>\n",
       "      <th>iqr</th>\n",
       "      <th>mean</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>subject</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",
       "      <th>0</th>\n",
       "      <td>1483.665778</td>\n",
       "      <td>535.686925</td>\n",
       "      <td>471.0</td>\n",
       "      <td>4115.0</td>\n",
       "      <td>11.295782</td>\n",
       "      <td>1117.25</td>\n",
       "      <td>1733.00</td>\n",
       "      <td>615.75</td>\n",
       "      <td>0.890222</td>\n",
       "      <td>-0.002591</td>\n",
       "      <td>krajbich2010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1502.737959</td>\n",
       "      <td>642.897107</td>\n",
       "      <td>423.0</td>\n",
       "      <td>4639.0</td>\n",
       "      <td>12.991134</td>\n",
       "      <td>1062.25</td>\n",
       "      <td>1790.75</td>\n",
       "      <td>728.50</td>\n",
       "      <td>0.780000</td>\n",
       "      <td>0.047497</td>\n",
       "      <td>krajbich2010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4427.150800</td>\n",
       "      <td>2895.781847</td>\n",
       "      <td>664.0</td>\n",
       "      <td>28717.0</td>\n",
       "      <td>57.927224</td>\n",
       "      <td>2411.75</td>\n",
       "      <td>5542.50</td>\n",
       "      <td>3130.75</td>\n",
       "      <td>0.846400</td>\n",
       "      <td>0.015188</td>\n",
       "      <td>krajbich2010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1921.084545</td>\n",
       "      <td>504.680828</td>\n",
       "      <td>838.0</td>\n",
       "      <td>4738.0</td>\n",
       "      <td>10.762278</td>\n",
       "      <td>1617.75</td>\n",
       "      <td>2145.00</td>\n",
       "      <td>527.25</td>\n",
       "      <td>0.688636</td>\n",
       "      <td>-0.009199</td>\n",
       "      <td>krajbich2010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1740.098400</td>\n",
       "      <td>794.662275</td>\n",
       "      <td>500.0</td>\n",
       "      <td>5851.0</td>\n",
       "      <td>15.896425</td>\n",
       "      <td>1198.00</td>\n",
       "      <td>2084.00</td>\n",
       "      <td>886.00</td>\n",
       "      <td>0.840800</td>\n",
       "      <td>-0.009003</td>\n",
       "      <td>krajbich2010</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  rt                                                   \\\n",
       "                mean          std    min      max         se       q1   \n",
       "subject                                                                 \n",
       "0        1483.665778   535.686925  471.0   4115.0  11.295782  1117.25   \n",
       "1        1502.737959   642.897107  423.0   4639.0  12.991134  1062.25   \n",
       "2        4427.150800  2895.781847  664.0  28717.0  57.927224  2411.75   \n",
       "3        1921.084545   504.680828  838.0   4738.0  10.762278  1617.75   \n",
       "4        1740.098400   794.662275  500.0   5851.0  15.896425  1198.00   \n",
       "\n",
       "                          best_chosen gaze_influence       dataset  \n",
       "              q3      iqr        mean                               \n",
       "subject                                                             \n",
       "0        1733.00   615.75    0.890222      -0.002591  krajbich2010  \n",
       "1        1790.75   728.50    0.780000       0.047497  krajbich2010  \n",
       "2        5542.50  3130.75    0.846400       0.015188  krajbich2010  \n",
       "3        2145.00   527.25    0.688636      -0.009199  krajbich2010  \n",
       "4        2084.00   886.00    0.840800      -0.009003  krajbich2010  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_nobias_summary.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "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 tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"8\" halign=\"left\">rt</th>\n",
       "      <th>best_chosen</th>\n",
       "      <th>gaze_influence</th>\n",
       "      <th>dataset</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>max</th>\n",
       "      <th>se</th>\n",
       "      <th>q1</th>\n",
       "      <th>q3</th>\n",
       "      <th>iqr</th>\n",
       "      <th>mean</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>subject</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",
       "      <th>113</th>\n",
       "      <td>2189.337197</td>\n",
       "      <td>2153.038110</td>\n",
       "      <td>286.0</td>\n",
       "      <td>17284.0</td>\n",
       "      <td>12.665149</td>\n",
       "      <td>1144.0</td>\n",
       "      <td>2389.0</td>\n",
       "      <td>1245.0</td>\n",
       "      <td>0.855052</td>\n",
       "      <td>0.213650</td>\n",
       "      <td>tavares2017</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>114</th>\n",
       "      <td>3341.584506</td>\n",
       "      <td>4984.396269</td>\n",
       "      <td>289.0</td>\n",
       "      <td>40259.0</td>\n",
       "      <td>27.691518</td>\n",
       "      <td>1380.0</td>\n",
       "      <td>3250.0</td>\n",
       "      <td>1870.0</td>\n",
       "      <td>0.826173</td>\n",
       "      <td>0.149680</td>\n",
       "      <td>tavares2017</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115</th>\n",
       "      <td>1271.193815</td>\n",
       "      <td>1127.552996</td>\n",
       "      <td>243.0</td>\n",
       "      <td>9399.0</td>\n",
       "      <td>7.145720</td>\n",
       "      <td>771.0</td>\n",
       "      <td>1353.0</td>\n",
       "      <td>582.0</td>\n",
       "      <td>0.865663</td>\n",
       "      <td>0.287849</td>\n",
       "      <td>tavares2017</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>116</th>\n",
       "      <td>2420.559574</td>\n",
       "      <td>2655.438637</td>\n",
       "      <td>317.0</td>\n",
       "      <td>21727.0</td>\n",
       "      <td>14.640130</td>\n",
       "      <td>1248.0</td>\n",
       "      <td>2556.0</td>\n",
       "      <td>1308.0</td>\n",
       "      <td>0.833769</td>\n",
       "      <td>0.326199</td>\n",
       "      <td>tavares2017</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117</th>\n",
       "      <td>1458.849035</td>\n",
       "      <td>823.192480</td>\n",
       "      <td>267.0</td>\n",
       "      <td>6268.0</td>\n",
       "      <td>4.555819</td>\n",
       "      <td>944.0</td>\n",
       "      <td>1716.0</td>\n",
       "      <td>772.0</td>\n",
       "      <td>0.770567</td>\n",
       "      <td>0.160005</td>\n",
       "      <td>tavares2017</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  rt                                                          \\\n",
       "                mean          std    min      max         se      q1      q3   \n",
       "subject                                                                        \n",
       "113      2189.337197  2153.038110  286.0  17284.0  12.665149  1144.0  2389.0   \n",
       "114      3341.584506  4984.396269  289.0  40259.0  27.691518  1380.0  3250.0   \n",
       "115      1271.193815  1127.552996  243.0   9399.0   7.145720   771.0  1353.0   \n",
       "116      2420.559574  2655.438637  317.0  21727.0  14.640130  1248.0  2556.0   \n",
       "117      1458.849035   823.192480  267.0   6268.0   4.555819   944.0  1716.0   \n",
       "\n",
       "                best_chosen gaze_influence      dataset  \n",
       "            iqr        mean                              \n",
       "subject                                                  \n",
       "113      1245.0    0.855052       0.213650  tavares2017  \n",
       "114      1870.0    0.826173       0.149680  tavares2017  \n",
       "115       582.0    0.865663       0.287849  tavares2017  \n",
       "116      1308.0    0.833769       0.326199  tavares2017  \n",
       "117       772.0    0.770567       0.160005  tavares2017  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_multiplicative_summary.tail()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Mixed effects analysis of prediction biases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "n_samples = 20000"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## GLAM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "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>best_chosen</th>\n",
       "      <th>dataset</th>\n",
       "      <th>gaze_influence</th>\n",
       "      <th>is_pred</th>\n",
       "      <th>mean_rt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.822222</td>\n",
       "      <td>krajbich2010</td>\n",
       "      <td>-0.020529</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1704.177778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.673469</td>\n",
       "      <td>krajbich2010</td>\n",
       "      <td>0.247233</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1728.489796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.840000</td>\n",
       "      <td>krajbich2010</td>\n",
       "      <td>0.028877</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4078.660000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.704545</td>\n",
       "      <td>krajbich2010</td>\n",
       "      <td>-0.022031</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1893.545455</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.840000</td>\n",
       "      <td>krajbich2010</td>\n",
       "      <td>0.103622</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1846.640000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   best_chosen       dataset  gaze_influence  is_pred      mean_rt\n",
       "0     0.822222  krajbich2010       -0.020529      0.0  1704.177778\n",
       "1     0.673469  krajbich2010        0.247233      0.0  1728.489796\n",
       "2     0.840000  krajbich2010        0.028877      0.0  4078.660000\n",
       "3     0.704545  krajbich2010       -0.022031      0.0  1893.545455\n",
       "4     0.840000  krajbich2010        0.103622      0.0  1846.640000"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Reformat data for mixed model\n",
    "mean_rt = np.concatenate([odd_summary['rt']['mean'].values,\n",
    "                          pred_multiplicative_summary['rt']['mean'].values],\n",
    "                         axis=0)\n",
    "best_chosen = np.concatenate([odd_summary['best_chosen']['mean'].values,\n",
    "                              pred_multiplicative_summary['best_chosen']['mean'].values],\n",
    "                             axis=0)\n",
    "gaze_influence = np.concatenate([odd_summary['gaze_influence'].values,\n",
    "                                 pred_multiplicative_summary['gaze_influence'].values],\n",
    "                                axis=0)\n",
    "\n",
    "is_pred = np.concatenate([np.zeros(len(odd_summary)),\n",
    "                          np.ones(len(pred_multiplicative_summary))],\n",
    "                         axis=0)\n",
    "dataset = np.concatenate([odd_summary['dataset'].values,\n",
    "                          pred_multiplicative_summary['dataset'].values],\n",
    "                         axis=0)\n",
    "\n",
    "df = pd.DataFrame(dict(mean_rt=mean_rt,\n",
    "                       best_chosen=best_chosen,\n",
    "                       gaze_influence=gaze_influence,\n",
    "                       is_pred=is_pred,\n",
    "                       dataset=dataset))\n",
    "\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "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>Unnamed: 0</th>\n",
       "      <th>mean</th>\n",
       "      <th>sd</th>\n",
       "      <th>hpd0.95_lower</th>\n",
       "      <th>hpd0.95_upper</th>\n",
       "      <th>effective_n</th>\n",
       "      <th>gelman_rubin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>is_pred|dataset[folke2016]</td>\n",
       "      <td>6.183238</td>\n",
       "      <td>193.424643</td>\n",
       "      <td>-418.517774</td>\n",
       "      <td>439.787199</td>\n",
       "      <td>2793</td>\n",
       "      <td>1.000765</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>is_pred|dataset[krajbich2010]</td>\n",
       "      <td>-13.927032</td>\n",
       "      <td>181.691662</td>\n",
       "      <td>-425.657753</td>\n",
       "      <td>376.940723</td>\n",
       "      <td>2220</td>\n",
       "      <td>1.001247</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>is_pred|dataset[krajbich2011]</td>\n",
       "      <td>-15.680129</td>\n",
       "      <td>189.891561</td>\n",
       "      <td>-454.574960</td>\n",
       "      <td>390.925857</td>\n",
       "      <td>2408</td>\n",
       "      <td>1.000426</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>is_pred|dataset[tavares2017]</td>\n",
       "      <td>9.238693</td>\n",
       "      <td>194.331711</td>\n",
       "      <td>-414.882331</td>\n",
       "      <td>443.869557</td>\n",
       "      <td>2276</td>\n",
       "      <td>1.000827</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>is_pred|dataset_sd</td>\n",
       "      <td>226.660761</td>\n",
       "      <td>251.766992</td>\n",
       "      <td>0.016741</td>\n",
       "      <td>686.596580</td>\n",
       "      <td>1047</td>\n",
       "      <td>1.002964</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>is_pred</td>\n",
       "      <td>-9.132202</td>\n",
       "      <td>183.321313</td>\n",
       "      <td>-409.726087</td>\n",
       "      <td>344.452495</td>\n",
       "      <td>1592</td>\n",
       "      <td>1.002040</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Intercept</td>\n",
       "      <td>2944.900317</td>\n",
       "      <td>743.890997</td>\n",
       "      <td>1488.746634</td>\n",
       "      <td>4416.812658</td>\n",
       "      <td>1052</td>\n",
       "      <td>1.002927</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1|dataset[folke2016]</td>\n",
       "      <td>2370.965658</td>\n",
       "      <td>751.982456</td>\n",
       "      <td>948.822934</td>\n",
       "      <td>3931.385639</td>\n",
       "      <td>1118</td>\n",
       "      <td>1.002462</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1|dataset[krajbich2010]</td>\n",
       "      <td>-781.187041</td>\n",
       "      <td>748.869686</td>\n",
       "      <td>-2221.514423</td>\n",
       "      <td>752.620702</td>\n",
       "      <td>1086</td>\n",
       "      <td>1.002634</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1|dataset[krajbich2011]</td>\n",
       "      <td>-521.260399</td>\n",
       "      <td>750.008881</td>\n",
       "      <td>-2026.287790</td>\n",
       "      <td>944.050806</td>\n",
       "      <td>1107</td>\n",
       "      <td>1.002577</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1|dataset[tavares2017]</td>\n",
       "      <td>-1044.707300</td>\n",
       "      <td>753.842420</td>\n",
       "      <td>-2516.386525</td>\n",
       "      <td>457.643558</td>\n",
       "      <td>1111</td>\n",
       "      <td>1.002368</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>mean_rt_sd</td>\n",
       "      <td>1033.254267</td>\n",
       "      <td>48.512640</td>\n",
       "      <td>940.716954</td>\n",
       "      <td>1130.169352</td>\n",
       "      <td>40000</td>\n",
       "      <td>1.000045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1|dataset_sd</td>\n",
       "      <td>1605.862863</td>\n",
       "      <td>505.958319</td>\n",
       "      <td>775.372014</td>\n",
       "      <td>2607.922055</td>\n",
       "      <td>5296</td>\n",
       "      <td>1.000281</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       Unnamed: 0         mean          sd  hpd0.95_lower  \\\n",
       "0      is_pred|dataset[folke2016]     6.183238  193.424643    -418.517774   \n",
       "1   is_pred|dataset[krajbich2010]   -13.927032  181.691662    -425.657753   \n",
       "2   is_pred|dataset[krajbich2011]   -15.680129  189.891561    -454.574960   \n",
       "3    is_pred|dataset[tavares2017]     9.238693  194.331711    -414.882331   \n",
       "4              is_pred|dataset_sd   226.660761  251.766992       0.016741   \n",
       "5                         is_pred    -9.132202  183.321313    -409.726087   \n",
       "6                       Intercept  2944.900317  743.890997    1488.746634   \n",
       "7            1|dataset[folke2016]  2370.965658  751.982456     948.822934   \n",
       "8         1|dataset[krajbich2010]  -781.187041  748.869686   -2221.514423   \n",
       "9         1|dataset[krajbich2011]  -521.260399  750.008881   -2026.287790   \n",
       "10         1|dataset[tavares2017] -1044.707300  753.842420   -2516.386525   \n",
       "11                     mean_rt_sd  1033.254267   48.512640     940.716954   \n",
       "12                   1|dataset_sd  1605.862863  505.958319     775.372014   \n",
       "\n",
       "    hpd0.95_upper  effective_n  gelman_rubin  \n",
       "0      439.787199         2793      1.000765  \n",
       "1      376.940723         2220      1.001247  \n",
       "2      390.925857         2408      1.000426  \n",
       "3      443.869557         2276      1.000827  \n",
       "4      686.596580         1047      1.002964  \n",
       "5      344.452495         1592      1.002040  \n",
       "6     4416.812658         1052      1.002927  \n",
       "7     3931.385639         1118      1.002462  \n",
       "8      752.620702         1086      1.002634  \n",
       "9      944.050806         1107      1.002577  \n",
       "10     457.643558         1111      1.002368  \n",
       "11    1130.169352        40000      1.000045  \n",
       "12    2607.922055         5296      1.000281  "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result_file = os.path.join('results', 'mixed_effects_models', '3_abs_fit_bias_glam_rt.csv')\n",
    "if not os.path.isfile(result_file):\n",
    "    results_mean_rt = model.fit('mean_rt ~ is_pred',\n",
    "                                random=['is_pred|dataset'],\n",
    "                                categorical=['dataset'],\n",
    "                                samples=n_samples)\n",
    "\n",
    "    bias_analysis_glam_mean_rt = results_mean_rt.summary(ranefs=True)\n",
    "    bias_analysis_glam_mean_rt.to_csv(result_file)\n",
    "else:\n",
    "    bias_analysis_glam_mean_rt = pd.read_csv(result_file)\n",
    "bias_analysis_glam_mean_rt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "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>Unnamed: 0</th>\n",
       "      <th>mean</th>\n",
       "      <th>sd</th>\n",
       "      <th>hpd0.95_lower</th>\n",
       "      <th>hpd0.95_upper</th>\n",
       "      <th>effective_n</th>\n",
       "      <th>gelman_rubin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>is_pred|dataset[folke2016]</td>\n",
       "      <td>-0.006850</td>\n",
       "      <td>0.025211</td>\n",
       "      <td>-6.294348e-02</td>\n",
       "      <td>0.042713</td>\n",
       "      <td>14227</td>\n",
       "      <td>1.000141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>is_pred|dataset[krajbich2010]</td>\n",
       "      <td>0.010806</td>\n",
       "      <td>0.025016</td>\n",
       "      <td>-3.388074e-02</td>\n",
       "      <td>0.068218</td>\n",
       "      <td>12820</td>\n",
       "      <td>0.999994</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>is_pred|dataset[krajbich2011]</td>\n",
       "      <td>-0.008073</td>\n",
       "      <td>0.025001</td>\n",
       "      <td>-6.704170e-02</td>\n",
       "      <td>0.036807</td>\n",
       "      <td>13663</td>\n",
       "      <td>1.000039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>is_pred|dataset[tavares2017]</td>\n",
       "      <td>0.003912</td>\n",
       "      <td>0.024995</td>\n",
       "      <td>-4.385866e-02</td>\n",
       "      <td>0.061252</td>\n",
       "      <td>13825</td>\n",
       "      <td>1.000016</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>is_pred|dataset_sd</td>\n",
       "      <td>0.029641</td>\n",
       "      <td>0.031191</td>\n",
       "      <td>6.969388e-07</td>\n",
       "      <td>0.088930</td>\n",
       "      <td>8430</td>\n",
       "      <td>0.999975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>is_pred</td>\n",
       "      <td>-0.022139</td>\n",
       "      <td>0.023374</td>\n",
       "      <td>-7.026238e-02</td>\n",
       "      <td>0.023049</td>\n",
       "      <td>12781</td>\n",
       "      <td>1.000022</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Intercept</td>\n",
       "      <td>0.767203</td>\n",
       "      <td>0.051055</td>\n",
       "      <td>6.602676e-01</td>\n",
       "      <td>0.868185</td>\n",
       "      <td>7143</td>\n",
       "      <td>1.000083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1|dataset[folke2016]</td>\n",
       "      <td>-0.110091</td>\n",
       "      <td>0.052479</td>\n",
       "      <td>-2.205632e-01</td>\n",
       "      <td>-0.007200</td>\n",
       "      <td>7728</td>\n",
       "      <td>1.000155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1|dataset[krajbich2010]</td>\n",
       "      <td>0.050345</td>\n",
       "      <td>0.051821</td>\n",
       "      <td>-5.546481e-02</td>\n",
       "      <td>0.154463</td>\n",
       "      <td>6947</td>\n",
       "      <td>1.000062</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1|dataset[krajbich2011]</td>\n",
       "      <td>-0.031545</td>\n",
       "      <td>0.052142</td>\n",
       "      <td>-1.387941e-01</td>\n",
       "      <td>0.072376</td>\n",
       "      <td>7432</td>\n",
       "      <td>1.000085</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1|dataset[tavares2017]</td>\n",
       "      <td>0.094510</td>\n",
       "      <td>0.052351</td>\n",
       "      <td>-1.174856e-02</td>\n",
       "      <td>0.200963</td>\n",
       "      <td>7580</td>\n",
       "      <td>1.000098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>best_chosen_sd</td>\n",
       "      <td>0.096000</td>\n",
       "      <td>0.004509</td>\n",
       "      <td>8.746727e-02</td>\n",
       "      <td>0.104966</td>\n",
       "      <td>33190</td>\n",
       "      <td>0.999979</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1|dataset_sd</td>\n",
       "      <td>0.101423</td>\n",
       "      <td>0.036252</td>\n",
       "      <td>4.363020e-02</td>\n",
       "      <td>0.173076</td>\n",
       "      <td>10656</td>\n",
       "      <td>0.999986</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       Unnamed: 0      mean        sd  hpd0.95_lower  \\\n",
       "0      is_pred|dataset[folke2016] -0.006850  0.025211  -6.294348e-02   \n",
       "1   is_pred|dataset[krajbich2010]  0.010806  0.025016  -3.388074e-02   \n",
       "2   is_pred|dataset[krajbich2011] -0.008073  0.025001  -6.704170e-02   \n",
       "3    is_pred|dataset[tavares2017]  0.003912  0.024995  -4.385866e-02   \n",
       "4              is_pred|dataset_sd  0.029641  0.031191   6.969388e-07   \n",
       "5                         is_pred -0.022139  0.023374  -7.026238e-02   \n",
       "6                       Intercept  0.767203  0.051055   6.602676e-01   \n",
       "7            1|dataset[folke2016] -0.110091  0.052479  -2.205632e-01   \n",
       "8         1|dataset[krajbich2010]  0.050345  0.051821  -5.546481e-02   \n",
       "9         1|dataset[krajbich2011] -0.031545  0.052142  -1.387941e-01   \n",
       "10         1|dataset[tavares2017]  0.094510  0.052351  -1.174856e-02   \n",
       "11                 best_chosen_sd  0.096000  0.004509   8.746727e-02   \n",
       "12                   1|dataset_sd  0.101423  0.036252   4.363020e-02   \n",
       "\n",
       "    hpd0.95_upper  effective_n  gelman_rubin  \n",
       "0        0.042713        14227      1.000141  \n",
       "1        0.068218        12820      0.999994  \n",
       "2        0.036807        13663      1.000039  \n",
       "3        0.061252        13825      1.000016  \n",
       "4        0.088930         8430      0.999975  \n",
       "5        0.023049        12781      1.000022  \n",
       "6        0.868185         7143      1.000083  \n",
       "7       -0.007200         7728      1.000155  \n",
       "8        0.154463         6947      1.000062  \n",
       "9        0.072376         7432      1.000085  \n",
       "10       0.200963         7580      1.000098  \n",
       "11       0.104966        33190      0.999979  \n",
       "12       0.173076        10656      0.999986  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result_file = os.path.join('results', 'mixed_effects_models', '3_abs_fit_bias_glam_best_chosen.csv')\n",
    "if not os.path.isfile(result_file):\n",
    "    results_best_chosen = model.fit('best_chosen ~ is_pred',\n",
    "                                    random=['is_pred|dataset'],\n",
    "                                    categorical=['dataset'],\n",
    "                                    samples=n_samples)\n",
    "\n",
    "    bias_analysis_glam_best_chosen = results_best_chosen.summary(ranefs=True)\n",
    "    bias_analysis_glam_best_chosen.to_csv(result_file)\n",
    "else:\n",
    "    bias_analysis_glam_best_chosen = pd.read_csv(result_file)\n",
    "bias_analysis_glam_best_chosen"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "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>Unnamed: 0</th>\n",
       "      <th>mean</th>\n",
       "      <th>sd</th>\n",
       "      <th>hpd0.95_lower</th>\n",
       "      <th>hpd0.95_upper</th>\n",
       "      <th>effective_n</th>\n",
       "      <th>gelman_rubin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>is_pred|dataset[folke2016]</td>\n",
       "      <td>0.005786</td>\n",
       "      <td>0.035830</td>\n",
       "      <td>-5.896349e-02</td>\n",
       "      <td>0.082816</td>\n",
       "      <td>548</td>\n",
       "      <td>1.002607</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>is_pred|dataset[krajbich2010]</td>\n",
       "      <td>-0.009886</td>\n",
       "      <td>0.036081</td>\n",
       "      <td>-8.436952e-02</td>\n",
       "      <td>0.052604</td>\n",
       "      <td>429</td>\n",
       "      <td>1.004118</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>is_pred|dataset[krajbich2011]</td>\n",
       "      <td>-0.000542</td>\n",
       "      <td>0.035602</td>\n",
       "      <td>-6.562848e-02</td>\n",
       "      <td>0.070636</td>\n",
       "      <td>396</td>\n",
       "      <td>1.003242</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>is_pred|dataset[tavares2017]</td>\n",
       "      <td>-0.000806</td>\n",
       "      <td>0.035546</td>\n",
       "      <td>-6.716548e-02</td>\n",
       "      <td>0.074427</td>\n",
       "      <td>454</td>\n",
       "      <td>1.002978</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>is_pred|dataset_sd</td>\n",
       "      <td>0.035896</td>\n",
       "      <td>0.043408</td>\n",
       "      <td>5.817650e-07</td>\n",
       "      <td>0.112897</td>\n",
       "      <td>390</td>\n",
       "      <td>1.001424</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>gaze_influence_sd</td>\n",
       "      <td>0.143388</td>\n",
       "      <td>0.005333</td>\n",
       "      <td>1.340886e-01</td>\n",
       "      <td>0.152977</td>\n",
       "      <td>17016</td>\n",
       "      <td>1.000082</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>is_pred</td>\n",
       "      <td>-0.022051</td>\n",
       "      <td>0.035455</td>\n",
       "      <td>-8.253339e-02</td>\n",
       "      <td>0.042121</td>\n",
       "      <td>431</td>\n",
       "      <td>1.003064</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Intercept</td>\n",
       "      <td>0.240041</td>\n",
       "      <td>0.047946</td>\n",
       "      <td>1.449192e-01</td>\n",
       "      <td>0.340855</td>\n",
       "      <td>765</td>\n",
       "      <td>1.004340</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1|dataset[folke2016]</td>\n",
       "      <td>0.081390</td>\n",
       "      <td>0.051597</td>\n",
       "      <td>-1.977554e-02</td>\n",
       "      <td>0.186141</td>\n",
       "      <td>813</td>\n",
       "      <td>1.004182</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1|dataset[krajbich2010]</td>\n",
       "      <td>-0.039964</td>\n",
       "      <td>0.049367</td>\n",
       "      <td>-1.455025e-01</td>\n",
       "      <td>0.056396</td>\n",
       "      <td>855</td>\n",
       "      <td>1.003416</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1|dataset[krajbich2011]</td>\n",
       "      <td>0.003646</td>\n",
       "      <td>0.049217</td>\n",
       "      <td>-9.820974e-02</td>\n",
       "      <td>0.103785</td>\n",
       "      <td>825</td>\n",
       "      <td>1.004046</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1|dataset[tavares2017]</td>\n",
       "      <td>-0.034492</td>\n",
       "      <td>0.050139</td>\n",
       "      <td>-1.386139e-01</td>\n",
       "      <td>0.065620</td>\n",
       "      <td>850</td>\n",
       "      <td>1.003859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1|dataset_sd</td>\n",
       "      <td>0.080697</td>\n",
       "      <td>0.040012</td>\n",
       "      <td>2.181458e-02</td>\n",
       "      <td>0.165967</td>\n",
       "      <td>1623</td>\n",
       "      <td>1.000586</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       Unnamed: 0      mean        sd  hpd0.95_lower  \\\n",
       "0      is_pred|dataset[folke2016]  0.005786  0.035830  -5.896349e-02   \n",
       "1   is_pred|dataset[krajbich2010] -0.009886  0.036081  -8.436952e-02   \n",
       "2   is_pred|dataset[krajbich2011] -0.000542  0.035602  -6.562848e-02   \n",
       "3    is_pred|dataset[tavares2017] -0.000806  0.035546  -6.716548e-02   \n",
       "4              is_pred|dataset_sd  0.035896  0.043408   5.817650e-07   \n",
       "5               gaze_influence_sd  0.143388  0.005333   1.340886e-01   \n",
       "6                         is_pred -0.022051  0.035455  -8.253339e-02   \n",
       "7                       Intercept  0.240041  0.047946   1.449192e-01   \n",
       "8            1|dataset[folke2016]  0.081390  0.051597  -1.977554e-02   \n",
       "9         1|dataset[krajbich2010] -0.039964  0.049367  -1.455025e-01   \n",
       "10        1|dataset[krajbich2011]  0.003646  0.049217  -9.820974e-02   \n",
       "11         1|dataset[tavares2017] -0.034492  0.050139  -1.386139e-01   \n",
       "12                   1|dataset_sd  0.080697  0.040012   2.181458e-02   \n",
       "\n",
       "    hpd0.95_upper  effective_n  gelman_rubin  \n",
       "0        0.082816          548      1.002607  \n",
       "1        0.052604          429      1.004118  \n",
       "2        0.070636          396      1.003242  \n",
       "3        0.074427          454      1.002978  \n",
       "4        0.112897          390      1.001424  \n",
       "5        0.152977        17016      1.000082  \n",
       "6        0.042121          431      1.003064  \n",
       "7        0.340855          765      1.004340  \n",
       "8        0.186141          813      1.004182  \n",
       "9        0.056396          855      1.003416  \n",
       "10       0.103785          825      1.004046  \n",
       "11       0.065620          850      1.003859  \n",
       "12       0.165967         1623      1.000586  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result_file = os.path.join('results', 'mixed_effects_models', '3_abs_fit_bias_glam_gaze.csv')\n",
    "if not os.path.isfile(result_file):\n",
    "    model = Model(df)\n",
    "    results_gaze_influence = model.fit('gaze_influence ~ is_pred',\n",
    "                                       random=['is_pred|dataset'],\n",
    "                                       categorical=['dataset'],\n",
    "                                       samples=n_samples)\n",
    "\n",
    "    bias_analysis_glam_gaze_influence = results_gaze_influence.summary(ranefs=True)\n",
    "    bias_analysis_glam_gaze_influence.to_csv(result_file)\n",
    "else:\n",
    "    bias_analysis_glam_gaze_influence = pd.read_csv(result_file)\n",
    "bias_analysis_glam_gaze_influence"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## No Bias"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "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>best_chosen</th>\n",
       "      <th>dataset</th>\n",
       "      <th>gaze_influence</th>\n",
       "      <th>is_pred</th>\n",
       "      <th>mean_rt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.822222</td>\n",
       "      <td>krajbich2010</td>\n",
       "      <td>-0.020529</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1704.177778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.673469</td>\n",
       "      <td>krajbich2010</td>\n",
       "      <td>0.247233</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1728.489796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.840000</td>\n",
       "      <td>krajbich2010</td>\n",
       "      <td>0.028877</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4078.660000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.704545</td>\n",
       "      <td>krajbich2010</td>\n",
       "      <td>-0.022031</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1893.545455</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.840000</td>\n",
       "      <td>krajbich2010</td>\n",
       "      <td>0.103622</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1846.640000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   best_chosen       dataset  gaze_influence  is_pred      mean_rt\n",
       "0     0.822222  krajbich2010       -0.020529      0.0  1704.177778\n",
       "1     0.673469  krajbich2010        0.247233      0.0  1728.489796\n",
       "2     0.840000  krajbich2010        0.028877      0.0  4078.660000\n",
       "3     0.704545  krajbich2010       -0.022031      0.0  1893.545455\n",
       "4     0.840000  krajbich2010        0.103622      0.0  1846.640000"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Prepare data for mixed model\n",
    "mean_rt = np.concatenate([odd_summary['rt']['mean'].values,\n",
    "                          pred_nobias_summary['rt']['mean'].values],\n",
    "                         axis=0)\n",
    "best_chosen = np.concatenate([odd_summary['best_chosen']['mean'].values,\n",
    "                              pred_nobias_summary['best_chosen']['mean'].values],\n",
    "                             axis=0)\n",
    "gaze_influence = np.concatenate([odd_summary['gaze_influence'].values,\n",
    "                                 pred_nobias_summary['gaze_influence'].values],\n",
    "                                axis=0)\n",
    "\n",
    "is_pred = np.concatenate([np.zeros(len(odd_summary)),\n",
    "                          np.ones(len(pred_nobias_summary))],\n",
    "                         axis=0)\n",
    "dataset = np.concatenate([odd_summary['dataset'].values,\n",
    "                          pred_nobias_summary['dataset'].values],\n",
    "                         axis=0)\n",
    "\n",
    "df = pd.DataFrame(dict(mean_rt=mean_rt,\n",
    "                       best_chosen=best_chosen,\n",
    "                       gaze_influence=gaze_influence,\n",
    "                       is_pred=is_pred,\n",
    "                       dataset=dataset))\n",
    "\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "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>Unnamed: 0</th>\n",
       "      <th>mean</th>\n",
       "      <th>sd</th>\n",
       "      <th>hpd0.95_lower</th>\n",
       "      <th>hpd0.95_upper</th>\n",
       "      <th>effective_n</th>\n",
       "      <th>gelman_rubin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>is_pred|dataset[folke2016]</td>\n",
       "      <td>-2.526226</td>\n",
       "      <td>195.730639</td>\n",
       "      <td>-442.559707</td>\n",
       "      <td>419.515893</td>\n",
       "      <td>5191</td>\n",
       "      <td>1.000126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>is_pred|dataset[krajbich2010]</td>\n",
       "      <td>-17.880493</td>\n",
       "      <td>184.558756</td>\n",
       "      <td>-431.111345</td>\n",
       "      <td>369.979637</td>\n",
       "      <td>3924</td>\n",
       "      <td>1.000357</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>is_pred|dataset[krajbich2011]</td>\n",
       "      <td>5.378380</td>\n",
       "      <td>189.572513</td>\n",
       "      <td>-407.977578</td>\n",
       "      <td>422.470261</td>\n",
       "      <td>4321</td>\n",
       "      <td>1.000615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>is_pred|dataset[tavares2017]</td>\n",
       "      <td>5.896717</td>\n",
       "      <td>193.936155</td>\n",
       "      <td>-413.332875</td>\n",
       "      <td>443.886545</td>\n",
       "      <td>4985</td>\n",
       "      <td>1.000215</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>is_pred|dataset_sd</td>\n",
       "      <td>227.115535</td>\n",
       "      <td>246.036493</td>\n",
       "      <td>0.000107</td>\n",
       "      <td>691.711755</td>\n",
       "      <td>2288</td>\n",
       "      <td>0.999977</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>is_pred</td>\n",
       "      <td>14.652198</td>\n",
       "      <td>185.818392</td>\n",
       "      <td>-354.925326</td>\n",
       "      <td>412.874884</td>\n",
       "      <td>2788</td>\n",
       "      <td>1.001059</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Intercept</td>\n",
       "      <td>2930.310779</td>\n",
       "      <td>795.489915</td>\n",
       "      <td>1289.818888</td>\n",
       "      <td>4439.705363</td>\n",
       "      <td>923</td>\n",
       "      <td>1.000255</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1|dataset[folke2016]</td>\n",
       "      <td>2374.522983</td>\n",
       "      <td>804.441179</td>\n",
       "      <td>828.028235</td>\n",
       "      <td>4005.160960</td>\n",
       "      <td>966</td>\n",
       "      <td>1.000341</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1|dataset[krajbich2010]</td>\n",
       "      <td>-771.767522</td>\n",
       "      <td>800.579567</td>\n",
       "      <td>-2365.051578</td>\n",
       "      <td>796.446894</td>\n",
       "      <td>948</td>\n",
       "      <td>1.000307</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1|dataset[krajbich2011]</td>\n",
       "      <td>-487.608329</td>\n",
       "      <td>802.376406</td>\n",
       "      <td>-2062.665692</td>\n",
       "      <td>1115.052138</td>\n",
       "      <td>948</td>\n",
       "      <td>1.000252</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1|dataset[tavares2017]</td>\n",
       "      <td>-1032.601397</td>\n",
       "      <td>803.738353</td>\n",
       "      <td>-2579.287276</td>\n",
       "      <td>612.248273</td>\n",
       "      <td>962</td>\n",
       "      <td>1.000242</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>mean_rt_sd</td>\n",
       "      <td>1057.685253</td>\n",
       "      <td>50.216135</td>\n",
       "      <td>959.967093</td>\n",
       "      <td>1155.933592</td>\n",
       "      <td>40000</td>\n",
       "      <td>0.999999</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1|dataset_sd</td>\n",
       "      <td>1623.546398</td>\n",
       "      <td>517.883354</td>\n",
       "      <td>782.440179</td>\n",
       "      <td>2679.704160</td>\n",
       "      <td>3999</td>\n",
       "      <td>1.000089</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       Unnamed: 0         mean          sd  hpd0.95_lower  \\\n",
       "0      is_pred|dataset[folke2016]    -2.526226  195.730639    -442.559707   \n",
       "1   is_pred|dataset[krajbich2010]   -17.880493  184.558756    -431.111345   \n",
       "2   is_pred|dataset[krajbich2011]     5.378380  189.572513    -407.977578   \n",
       "3    is_pred|dataset[tavares2017]     5.896717  193.936155    -413.332875   \n",
       "4              is_pred|dataset_sd   227.115535  246.036493       0.000107   \n",
       "5                         is_pred    14.652198  185.818392    -354.925326   \n",
       "6                       Intercept  2930.310779  795.489915    1289.818888   \n",
       "7            1|dataset[folke2016]  2374.522983  804.441179     828.028235   \n",
       "8         1|dataset[krajbich2010]  -771.767522  800.579567   -2365.051578   \n",
       "9         1|dataset[krajbich2011]  -487.608329  802.376406   -2062.665692   \n",
       "10         1|dataset[tavares2017] -1032.601397  803.738353   -2579.287276   \n",
       "11                     mean_rt_sd  1057.685253   50.216135     959.967093   \n",
       "12                   1|dataset_sd  1623.546398  517.883354     782.440179   \n",
       "\n",
       "    hpd0.95_upper  effective_n  gelman_rubin  \n",
       "0      419.515893         5191      1.000126  \n",
       "1      369.979637         3924      1.000357  \n",
       "2      422.470261         4321      1.000615  \n",
       "3      443.886545         4985      1.000215  \n",
       "4      691.711755         2288      0.999977  \n",
       "5      412.874884         2788      1.001059  \n",
       "6     4439.705363          923      1.000255  \n",
       "7     4005.160960          966      1.000341  \n",
       "8      796.446894          948      1.000307  \n",
       "9     1115.052138          948      1.000252  \n",
       "10     612.248273          962      1.000242  \n",
       "11    1155.933592        40000      0.999999  \n",
       "12    2679.704160         3999      1.000089  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result_file = os.path.join('results', 'mixed_effects_models', '3_abs_fit_bias_nobias_rt.csv')\n",
    "if not os.path.isfile(result_file):\n",
    "    results_mean_rt = model.fit('mean_rt ~ is_pred',\n",
    "                                random=['is_pred|dataset'],\n",
    "                                categorical=['dataset'],\n",
    "                                samples=n_samples)\n",
    "    bias_analysis_nobias_mean_rt = results_mean_rt.summary(ranefs=True)\n",
    "    bias_analysis_nobias_mean_rt.to_csv(result_file)\n",
    "else:\n",
    "    bias_analysis_nobias_mean_rt = pd.read_csv(result_file)\n",
    "bias_analysis_nobias_mean_rt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "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>Unnamed: 0</th>\n",
       "      <th>mean</th>\n",
       "      <th>sd</th>\n",
       "      <th>hpd0.95_lower</th>\n",
       "      <th>hpd0.95_upper</th>\n",
       "      <th>effective_n</th>\n",
       "      <th>gelman_rubin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>is_pred|dataset[folke2016]</td>\n",
       "      <td>-0.009518</td>\n",
       "      <td>0.024767</td>\n",
       "      <td>-0.067449</td>\n",
       "      <td>0.034684</td>\n",
       "      <td>10982</td>\n",
       "      <td>0.999993</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>is_pred|dataset[krajbich2010]</td>\n",
       "      <td>0.014496</td>\n",
       "      <td>0.025427</td>\n",
       "      <td>-0.030271</td>\n",
       "      <td>0.072145</td>\n",
       "      <td>7709</td>\n",
       "      <td>0.999977</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>is_pred|dataset[krajbich2011]</td>\n",
       "      <td>-0.003072</td>\n",
       "      <td>0.023852</td>\n",
       "      <td>-0.054723</td>\n",
       "      <td>0.046599</td>\n",
       "      <td>8274</td>\n",
       "      <td>1.000024</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>is_pred|dataset[tavares2017]</td>\n",
       "      <td>-0.001081</td>\n",
       "      <td>0.024223</td>\n",
       "      <td>-0.052884</td>\n",
       "      <td>0.051231</td>\n",
       "      <td>8060</td>\n",
       "      <td>0.999978</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>is_pred|dataset_sd</td>\n",
       "      <td>0.029804</td>\n",
       "      <td>0.029438</td>\n",
       "      <td>0.000001</td>\n",
       "      <td>0.088121</td>\n",
       "      <td>7866</td>\n",
       "      <td>0.999981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>is_pred</td>\n",
       "      <td>0.001294</td>\n",
       "      <td>0.022764</td>\n",
       "      <td>-0.044332</td>\n",
       "      <td>0.048286</td>\n",
       "      <td>7507</td>\n",
       "      <td>0.999988</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Intercept</td>\n",
       "      <td>0.769141</td>\n",
       "      <td>0.049428</td>\n",
       "      <td>0.669974</td>\n",
       "      <td>0.869689</td>\n",
       "      <td>7060</td>\n",
       "      <td>0.999981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1|dataset[folke2016]</td>\n",
       "      <td>-0.113719</td>\n",
       "      <td>0.050585</td>\n",
       "      <td>-0.214910</td>\n",
       "      <td>-0.011406</td>\n",
       "      <td>7618</td>\n",
       "      <td>0.999978</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1|dataset[krajbich2010]</td>\n",
       "      <td>0.049531</td>\n",
       "      <td>0.050381</td>\n",
       "      <td>-0.050653</td>\n",
       "      <td>0.152601</td>\n",
       "      <td>7309</td>\n",
       "      <td>0.999980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1|dataset[krajbich2011]</td>\n",
       "      <td>-0.030279</td>\n",
       "      <td>0.050391</td>\n",
       "      <td>-0.132406</td>\n",
       "      <td>0.071070</td>\n",
       "      <td>7459</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1|dataset[tavares2017]</td>\n",
       "      <td>0.089115</td>\n",
       "      <td>0.050368</td>\n",
       "      <td>-0.009148</td>\n",
       "      <td>0.192995</td>\n",
       "      <td>7568</td>\n",
       "      <td>0.999986</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>best_chosen_sd</td>\n",
       "      <td>0.089742</td>\n",
       "      <td>0.004200</td>\n",
       "      <td>0.081760</td>\n",
       "      <td>0.098106</td>\n",
       "      <td>34822</td>\n",
       "      <td>0.999975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1|dataset_sd</td>\n",
       "      <td>0.099323</td>\n",
       "      <td>0.034384</td>\n",
       "      <td>0.043781</td>\n",
       "      <td>0.168421</td>\n",
       "      <td>13472</td>\n",
       "      <td>1.000056</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       Unnamed: 0      mean        sd  hpd0.95_lower  \\\n",
       "0      is_pred|dataset[folke2016] -0.009518  0.024767      -0.067449   \n",
       "1   is_pred|dataset[krajbich2010]  0.014496  0.025427      -0.030271   \n",
       "2   is_pred|dataset[krajbich2011] -0.003072  0.023852      -0.054723   \n",
       "3    is_pred|dataset[tavares2017] -0.001081  0.024223      -0.052884   \n",
       "4              is_pred|dataset_sd  0.029804  0.029438       0.000001   \n",
       "5                         is_pred  0.001294  0.022764      -0.044332   \n",
       "6                       Intercept  0.769141  0.049428       0.669974   \n",
       "7            1|dataset[folke2016] -0.113719  0.050585      -0.214910   \n",
       "8         1|dataset[krajbich2010]  0.049531  0.050381      -0.050653   \n",
       "9         1|dataset[krajbich2011] -0.030279  0.050391      -0.132406   \n",
       "10         1|dataset[tavares2017]  0.089115  0.050368      -0.009148   \n",
       "11                 best_chosen_sd  0.089742  0.004200       0.081760   \n",
       "12                   1|dataset_sd  0.099323  0.034384       0.043781   \n",
       "\n",
       "    hpd0.95_upper  effective_n  gelman_rubin  \n",
       "0        0.034684        10982      0.999993  \n",
       "1        0.072145         7709      0.999977  \n",
       "2        0.046599         8274      1.000024  \n",
       "3        0.051231         8060      0.999978  \n",
       "4        0.088121         7866      0.999981  \n",
       "5        0.048286         7507      0.999988  \n",
       "6        0.869689         7060      0.999981  \n",
       "7       -0.011406         7618      0.999978  \n",
       "8        0.152601         7309      0.999980  \n",
       "9        0.071070         7459      1.000000  \n",
       "10       0.192995         7568      0.999986  \n",
       "11       0.098106        34822      0.999975  \n",
       "12       0.168421        13472      1.000056  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result_file = os.path.join('results', 'mixed_effects_models', '3_abs_fit_bias_nobias_best_chosen.csv')\n",
    "if not os.path.isfile(result_file):\n",
    "    results_best_chosen = model.fit('best_chosen ~ is_pred',\n",
    "                                    random=['is_pred|dataset'],\n",
    "                                    categorical=['dataset'],\n",
    "                                    samples=n_samples)\n",
    "    bias_analysis_nobias_best_chosen = results_best_chosen.summary(ranefs=True)\n",
    "    bias_analysis_nobias_best_chosen.to_csv(result_file)\n",
    "else:\n",
    "    bias_analysis_nobias_best_chosen = pd.read_csv(result_file)\n",
    "bias_analysis_nobias_best_chosen"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "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>Unnamed: 0</th>\n",
       "      <th>mean</th>\n",
       "      <th>sd</th>\n",
       "      <th>hpd0.95_lower</th>\n",
       "      <th>hpd0.95_upper</th>\n",
       "      <th>effective_n</th>\n",
       "      <th>gelman_rubin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>is_pred|dataset[folke2016]</td>\n",
       "      <td>-0.064004</td>\n",
       "      <td>0.052485</td>\n",
       "      <td>-0.175414</td>\n",
       "      <td>0.022771</td>\n",
       "      <td>1838</td>\n",
       "      <td>1.000588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>is_pred|dataset[krajbich2010]</td>\n",
       "      <td>0.015505</td>\n",
       "      <td>0.042666</td>\n",
       "      <td>-0.078539</td>\n",
       "      <td>0.105360</td>\n",
       "      <td>2085</td>\n",
       "      <td>1.000359</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>is_pred|dataset[krajbich2011]</td>\n",
       "      <td>-0.010303</td>\n",
       "      <td>0.043774</td>\n",
       "      <td>-0.109497</td>\n",
       "      <td>0.075273</td>\n",
       "      <td>1854</td>\n",
       "      <td>1.000669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>is_pred|dataset[tavares2017]</td>\n",
       "      <td>0.017728</td>\n",
       "      <td>0.043824</td>\n",
       "      <td>-0.075261</td>\n",
       "      <td>0.111455</td>\n",
       "      <td>2190</td>\n",
       "      <td>1.000292</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>is_pred|dataset_sd</td>\n",
       "      <td>0.068428</td>\n",
       "      <td>0.048503</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>0.164987</td>\n",
       "      <td>2123</td>\n",
       "      <td>1.000267</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>gaze_influence_sd</td>\n",
       "      <td>0.108833</td>\n",
       "      <td>0.005229</td>\n",
       "      <td>0.098903</td>\n",
       "      <td>0.119399</td>\n",
       "      <td>40000</td>\n",
       "      <td>1.000112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>is_pred</td>\n",
       "      <td>-0.227179</td>\n",
       "      <td>0.041045</td>\n",
       "      <td>-0.308115</td>\n",
       "      <td>-0.131912</td>\n",
       "      <td>1640</td>\n",
       "      <td>1.000545</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Intercept</td>\n",
       "      <td>0.236765</td>\n",
       "      <td>0.036684</td>\n",
       "      <td>0.156990</td>\n",
       "      <td>0.311178</td>\n",
       "      <td>1972</td>\n",
       "      <td>1.001390</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1|dataset[folke2016]</td>\n",
       "      <td>0.069280</td>\n",
       "      <td>0.041886</td>\n",
       "      <td>-0.005757</td>\n",
       "      <td>0.161977</td>\n",
       "      <td>3436</td>\n",
       "      <td>1.000709</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1|dataset[krajbich2010]</td>\n",
       "      <td>-0.029083</td>\n",
       "      <td>0.038281</td>\n",
       "      <td>-0.109351</td>\n",
       "      <td>0.049900</td>\n",
       "      <td>2128</td>\n",
       "      <td>1.001224</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1|dataset[krajbich2011]</td>\n",
       "      <td>0.004569</td>\n",
       "      <td>0.038552</td>\n",
       "      <td>-0.073522</td>\n",
       "      <td>0.087757</td>\n",
       "      <td>2283</td>\n",
       "      <td>1.001138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1|dataset[tavares2017]</td>\n",
       "      <td>-0.029795</td>\n",
       "      <td>0.039038</td>\n",
       "      <td>-0.109882</td>\n",
       "      <td>0.050387</td>\n",
       "      <td>2330</td>\n",
       "      <td>1.001483</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1|dataset_sd</td>\n",
       "      <td>0.066907</td>\n",
       "      <td>0.034044</td>\n",
       "      <td>0.013615</td>\n",
       "      <td>0.137498</td>\n",
       "      <td>6478</td>\n",
       "      <td>1.000045</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       Unnamed: 0      mean        sd  hpd0.95_lower  \\\n",
       "0      is_pred|dataset[folke2016] -0.064004  0.052485      -0.175414   \n",
       "1   is_pred|dataset[krajbich2010]  0.015505  0.042666      -0.078539   \n",
       "2   is_pred|dataset[krajbich2011] -0.010303  0.043774      -0.109497   \n",
       "3    is_pred|dataset[tavares2017]  0.017728  0.043824      -0.075261   \n",
       "4              is_pred|dataset_sd  0.068428  0.048503       0.000006   \n",
       "5               gaze_influence_sd  0.108833  0.005229       0.098903   \n",
       "6                         is_pred -0.227179  0.041045      -0.308115   \n",
       "7                       Intercept  0.236765  0.036684       0.156990   \n",
       "8            1|dataset[folke2016]  0.069280  0.041886      -0.005757   \n",
       "9         1|dataset[krajbich2010] -0.029083  0.038281      -0.109351   \n",
       "10        1|dataset[krajbich2011]  0.004569  0.038552      -0.073522   \n",
       "11         1|dataset[tavares2017] -0.029795  0.039038      -0.109882   \n",
       "12                   1|dataset_sd  0.066907  0.034044       0.013615   \n",
       "\n",
       "    hpd0.95_upper  effective_n  gelman_rubin  \n",
       "0        0.022771         1838      1.000588  \n",
       "1        0.105360         2085      1.000359  \n",
       "2        0.075273         1854      1.000669  \n",
       "3        0.111455         2190      1.000292  \n",
       "4        0.164987         2123      1.000267  \n",
       "5        0.119399        40000      1.000112  \n",
       "6       -0.131912         1640      1.000545  \n",
       "7        0.311178         1972      1.001390  \n",
       "8        0.161977         3436      1.000709  \n",
       "9        0.049900         2128      1.001224  \n",
       "10       0.087757         2283      1.001138  \n",
       "11       0.050387         2330      1.001483  \n",
       "12       0.137498         6478      1.000045  "
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result_file = os.path.join('results', 'mixed_effects_models', '3_abs_fit_bias_nobias_gaze.csv')\n",
    "if not os.path.isfile(result_file):\n",
    "    model = Model(df)\n",
    "    results_gaze_influence = model.fit('gaze_influence ~ is_pred',\n",
    "                                       random=['is_pred|dataset'],\n",
    "                                       categorical=['dataset'],\n",
    "                                       samples=n_samples)\n",
    "    bias_analysis_nobias_gaze_influence = results_gaze_influence.summary(ranefs=True)\n",
    "    bias_analysis_nobias_gaze_influence.to_csv(result_file)\n",
    "else:\n",
    "    bias_analysis_nobias_gaze_influence = pd.read_csv(result_file)\n",
    "bias_analysis_nobias_gaze_influence"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Figure 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def figure_5(observed, multiplicative, nobias,\n",
    "             figsize=cm2inch(18,11), colors=None, fontsize=7, alpha=1.0):\n",
    "\n",
    "    datasets = ['krajbich2010', 'krajbich2011', 'folke2016', 'tavares2017']\n",
    "    \n",
    "    if colors is None:\n",
    "        colors = np.array(['C{}'.format(i)\n",
    "                           for i in range(observed['dataset'].unique().size)])\n",
    "    color_idx = pd.Categorical(observed['dataset'], categories=datasets, ordered=True).codes.astype(int)\n",
    "    \n",
    "    fig, axs = plt.subplots(2, 3, figsize=figsize)\n",
    "    \n",
    "    for m, model in enumerate([multiplicative, nobias]):\n",
    "\n",
    "        # a) Mean RT\n",
    "        axs[m, 0].scatter(observed['rt']['mean'],\n",
    "                          model['rt']['mean'],\n",
    "                          marker='o',\n",
    "                          color='none',\n",
    "                          edgecolor=colors[color_idx],\n",
    "                          linewidth=0.5,\n",
    "                          s=30)\n",
    "        axs[m, 0].scatter(observed['rt']['mean'],\n",
    "                          model['rt']['mean'],\n",
    "                          marker='o',\n",
    "                          color=colors[color_idx],\n",
    "                          alpha=0.5,\n",
    "                          linewidth=0,\n",
    "                          s=30)\n",
    "        axs[m, 0].set_xticks(np.arange(0, 10000, 2000))\n",
    "        axs[m, 0].set_yticks(np.arange(0, 10000, 2000))\n",
    "        axs[m, 0].set_xlim(0, 9000)\n",
    "        axs[m, 0].set_ylim(0, 9000)\n",
    "        \n",
    "        # b) P(choose best)\n",
    "        axs[m, 1].scatter(observed['best_chosen']['mean'],\n",
    "                          model['best_chosen']['mean'],\n",
    "                          marker='o',\n",
    "                          color='none',\n",
    "                          edgecolor=colors[color_idx],\n",
    "                          linewidth=0.5,\n",
    "                          s=30)\n",
    "        axs[m, 1].scatter(observed['best_chosen']['mean'],\n",
    "                          model['best_chosen']['mean'],\n",
    "                          marker='o',\n",
    "                          color=colors[color_idx],\n",
    "                          alpha=0.5,\n",
    "                          linewidth=0,\n",
    "                          s=30)\n",
    "        axs[m, 1].set_xticks(np.arange(0.4, 1.1, 0.2))\n",
    "        axs[m, 1].set_yticks(np.arange(0.4, 1.1, 0.2))\n",
    "        axs[m, 1].set_xlim(0.3, 1.0)\n",
    "        axs[m, 1].set_ylim(0.3, 1.0)\n",
    "\n",
    "        # c) Gaze Influence\n",
    "        axs[m, 2].scatter(observed['gaze_influence'],\n",
    "                          model['gaze_influence'],\n",
    "                          marker='o',\n",
    "                          color='none',\n",
    "                          edgecolor=colors[color_idx],\n",
    "                          linewidth=0.5,\n",
    "                          s=30)\n",
    "        axs[m, 2].scatter(observed['gaze_influence'],\n",
    "                          model['gaze_influence'],\n",
    "                          marker='o',\n",
    "                          color=colors[color_idx],\n",
    "                          alpha=0.5,\n",
    "                          linewidth=0,\n",
    "                          s=30)\n",
    "        axs[m, 2].set_xticks(np.arange(-0.2, 1.1, 0.2))\n",
    "        axs[m, 2].set_yticks(np.arange(-0.2, 1.1, 0.2))\n",
    "        axs[m, 2].set_xlim(-0.25, 0.85)\n",
    "        axs[m, 2].set_ylim(-0.25, 0.85)\n",
    "\n",
    "    axs[0, 0].set_ylabel('GLAM\\n\\nPredicted Mean RT (ms)', fontsize=fontsize)\n",
    "    axs[1, 0].set_ylabel('No-Bias\\n\\nPredicted Mean RT (ms)', fontsize=fontsize)\n",
    "    axs[0, 0].set_xlabel('Observed Mean RT (ms)', fontsize=fontsize)\n",
    "    axs[1, 0].set_xlabel('Observed Mean RT (ms)', fontsize=fontsize)\n",
    "    \n",
    "    axs[0, 1].set_ylabel('Predicted P(choose best)', fontsize=fontsize)\n",
    "    axs[1, 1].set_ylabel('Predicted P(choose best)', fontsize=fontsize)\n",
    "    axs[0, 1].set_xlabel('Observed P(choose best)', fontsize=fontsize)\n",
    "    axs[1, 1].set_xlabel('Observed P(choose best)', fontsize=fontsize)\n",
    "    \n",
    "    axs[0, 2].set_ylabel('Predicted Gaze Influence\\non P(choice | value)', fontsize=fontsize)\n",
    "    axs[1, 2].set_ylabel('Predicted Gaze Influence\\non P(choice | value)', fontsize=fontsize)\n",
    "    axs[0, 2].set_xlabel('Observed Gaze Influence\\non P(choice | value)', fontsize=fontsize)\n",
    "    axs[1, 2].set_xlabel('Observed Gaze Influence\\non P(choice | value)', fontsize=fontsize)\n",
    "    \n",
    "    patches = [mpatches.Rectangle((0, 0), 0.5, 1, fc=\"C{}\".format(i))\n",
    "               for i in range(len(datasets))]\n",
    "\n",
    "    axs[0, 0].legend(patches,\n",
    "                     ['Krajbich 2010', 'Krajbich 2011', 'Folke 2016', 'Tavares 2017'],\n",
    "                     loc='upper left',\n",
    "                     handlelength=1, handleheight=1,\n",
    "                     fontsize=5,\n",
    "                     frameon=False)\n",
    "\n",
    "    # Labels\n",
    "    for label, ax in zip(list('abcdef'), axs.ravel()):\n",
    "        ax.text(-0.4, 1.1, label, transform=ax.transAxes,\n",
    "                fontsize=fontsize, fontweight='bold', va='top')\n",
    "        ax.spines['top'].set_visible(False)\n",
    "        ax.spines['right'].set_visible(False)\n",
    "        \n",
    "        # Font sizes\n",
    "        ax.tick_params(axis='both', labelsize=fontsize)\n",
    "\n",
    "        # Plot diagonal\n",
    "        ax.plot(ax.get_xlim(), ax.get_xlim(), linewidth=1, color='black', alpha=1.0, zorder=-1)\n",
    "        \n",
    "    fig.tight_layout(pad=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeEAAAEgCAYAAACD/ssfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4xLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvAOZPmwAAIABJREFUeJzsnXd4VFXawH/n3qlJJr0QQkLoVZpY\nQFBAsWLDrruua1lX99vVdS24FnRFxd4XWV0F1HVX2VXsSFeaSpEWSiCQhFSSSZlMn3vP98cdQo+J\nJpkA83ueeZi598y57xlO7nvPeZuQUhIlSpQoUaJEaX+USAsQJUqUKFGiHK9ElXCUKFGiRIkSIaJK\nOEqUKFGiRIkQUSUcJUqUKFGiRIioEo4SJUqUKFEiRFQJR4kSJUqUKBEiqoSjHHMIIW4QQpRHWo4o\nUaJE+SlaVQkLIS4RQuQJIXYLIaa3Zt9RorQQuxBirhCiTAhxSqSFiXJ8IYToL4RYLYQoF0I8FWl5\nonRcWnslLAEVEMDvhBCprdx/lCjNxQJcDiwA7omwLFGOPx4GiqSUnYApkRYmSseltZXw74AlwK/C\nn+Nauf8oUVqCDL+iRIkE+kH/RolyCK2thP8LXAn8qZX7jRKlpQSAj4DxwHMRliXK8ccUoHvYN2Fy\npIWJ0nER0dzRUaJEiRIlSmSIekdHiRIlSpQoESKqhKNEaQIhRHchxD+FELMPOj5QCPFe+DUwUvJF\niRLl6CaqhKNEaQIpZYGU8qbDnLoD+ANwO/DH9pUqSpQoxwrHvBI+99xz93rIRl9t+Nq6davMzs6W\nr7/++vHiZJAgpayVUtYBjsM1EEL8TgixSgixasCAARH/PzrWX4FAQF522WXyvPPOO17mYIuI3gvb\n5/XOO+/IzMzMZs/BY14JV1VVRVqEY56NGzcyduxYHnvsMX7/+99HWpz2ok4IkSCEiAdch2sgpfyH\nlHK4lHK43W5vZ/E6Lt6ARo070Kp9+nw+Jk6ciKZpfPzxx63a97FC9F7Y9rz55pvcf//9LFy4sNnf\nMbWhPFGOA9auXcv555/P888/zzXXXBNpcZqNECINiMdIqBBsol0K8DgwVAhxP9BfSvlr4CXglXCz\np9ta3mMBTZc88XkeJbU+hACB4L7z+tA1JfYX9evxeLjkkktISkri3XffxWw2t5LEzUMIcQcwUEp5\nixDiISnlY63cfyzwd4ywu8VSyvfCx88Hbgg3e1NK+XVrXjdKy3jttdd4+umnWbRoEb169Wr296JK\nOMrP5vvvv+fCCy9k2rRpTJw4MdLiNAshxK3AKMAL1AJZQog64Bkp5c6D20spq4FDlvdSyo3A9W0s\n7jHF60t2ENQlfToZu/e6lDzxxWZe/9WJCCF+Vp8NDQ1MmDCBnJwc3nrrLUymiNzSegDF4feHNU38\nQiYCs6WUnwoh/gO8Fz4+FrgL0ID7gKgSjhDPPfccr732GkuWLCE3N7dF3z0ulXDupM8PObZr6gUR\nkOToZenSpUycOJG33nqLCRMmRFqclvCDlPKAvObhlUbnCMlz3JBf4Tpg1asIgVlV2F3jJTs5psX9\n1dXVcf7559O/f39ef/11VFVtTXFbgsTIVT6QtplHXYAN4ffafsf/DXxIOE3w4b4ohPjd3nM5OTlt\nIFqUxx9/nJkzZ/LNN9/QpUuXFn+/1W3CQoh+QoiXhBBzhBAfCSE+FELcLIQ4qhT+1KlTqamp4b77\n7sPn8wGwf2KT6upqfvjhB6ZOndp4bPny5dTX1zd+3rVrF//+978P6HfNmjU8+eSTPPLII4RCISZN\nmsSkSZPQNI0XX3yxsb9//vOfPPHEEyxbtqwth/mzWLhwIZdeeinvvvvu0aaAkVKuEULcvvezEOJO\nKaVbSpkfSbmOdfIrXawrrmVeXjlripzo4b8lXUps5pYrT6fTyVlnncXQoUOZPn06asUGyPsUPM7W\nFr05PIehCH8N3N8G/e/GUMRw4D37fuAM4HRg0uG+uL9fQlpaWhuIdvwipeShhx7iX//6F0uWLPlZ\nChjaZiV8NvCIlLJm7wEhxChgNLCoDa7XJui6zmOPPcbjjz/OypUrmTdvHpdccgkrVqxgz5493HTT\nTezYsYPKykqmTZtGKBTC4XDQrVs3nnrqKbp06cJZZ53F119/zZIlS5gyZQopKSkMGzaMYcOG8cAD\nD7Bu3TrGjx+PlJJ169Zx5513Nirhqqoq/vrXv/LUU09x2mmntfr4FhQuYH7RfKSUdE/szo0Db8Sk\n/PR0mDt3Lr/61a/48MMPGTNmTKvL1dYIIZ4BRgohcjFunD2BFyMp07HOtgoXT325hV4ZcRQ5vdR5\nQyzcUsmI7inEWFTSHNbGtt7Nm3G+9RYyEASzmeRf/wphNlM3Zw6KzUbSVVdRazYzfvx4zh43jvtP\nOw3nn87C0T8JS1ZnxLr3od8FMOS69hziICnlJAAhxHns25puLf4HvCqEuAD4VAjxTtgv4X/Am+E2\nX7XyNaM0gZSSe++9l3nz5rF48WJ+yQNOqythKeVLQogMACHEGcBuKeXS1r5OW6MoCmazmT179gBw\nwQUXkJmZiRCCgoKCxnYWi4XbbruNRx55BIfDQSAQICsri9tuu41du3Yxbtw4OnfuzPbt20lJSQFg\n+vTpXHnllYRCoSNe/+fayJrD3J1zWVi8kK7xXQGo8lbx7A/PMumUwz5MN/LJJ59w8803M2fOHEaO\nHNlm8rUxrwI/AksxtvYqIivOsc+s5bvol+lAVYxF3K5qDy5fEJcvyGMXn9DYTqurY88LL2IbOBAh\nBFJKiv/6INYuWcT064cMBdn1wAPcu3Ill1xyCTe5GvB9/S6qOZbatW7s5buIHzsSNn/W3kr4DOCz\n8PvRwJet2bmU0g38dr9D74WP/wv4V2teK8pPo+s6d9xxBytXrmThwoUkJyf/ov7aKkTpLiFET+AK\n4JGmGgohcoQQHwsh3hJCTBJCXCuEeEMIMUsIERt+zQwfuy78nQPatNEYmDJlCi+88ALV1dUoisLO\nnTsRQuD3+/fKTiAQYNq0aY3/ERaLhdLSUqZNm0YwGMRkMqEoSuNW9ldffcW8efNYtWoVgwcPZv78\n+SxYsIDBgwfz/vvvs2zZMgoLC0lNTeXJJ59k1KhRrT6u/RUwQKI1kQpPBQ2BhiN+Z/bs2dxyyy18\n8cUXR7MCRkpZiPHwWQ48AzwQWYmOffwhrVEB90p30CnehtWsclK3FOyWfVvRtXPmYMnJQQhBZb2P\nz9eXUuDWWVXSwPKdNbj8ft5Zvpxbu+byx549MXfOxGTyI6wxKPEJeMt8BGucYE9q7yGmCSHOFEKM\nAzLb++JR2g9d17n11ltZvXo18+fP/8UKGNqogIMQYiawC3gHuFVKecR6ruEtliQp5bthzz9FSnmF\nEGICsPevqXavZ6CU8iohxIf7t5FSvnOk/ocPHy5XrVrVamPby4wZM+jfvz8nn3xyq/fd1tyz5B66\nJXQ74NiO2h3cd/J9pMekH9L+vffe4+677+bLL79kyJAhzblE2y3jWwEhxNPsM42Ma2p+thZtNQ+P\nBl5fvJ2yOh+KgI/WlqBLCOmSWIuJEd2TefnaYQBUTf8HoZoaREwMn60rJcmq4Ni8DndiKnXJGWxb\n/z0n98zkxIwM1Ph4TGlpkD8PVDMoKrrXS/zAJOypAq6c0W5zMBwrvnfp/Z6Usr6p9pHieJ6DrUEo\nFOLGG2+kqKiITz/9FIfjJx3hmzUH22ol/BawTUq5HVjxE21XAjcJIRZi2DX2PhUUYjgjdGGfjWWv\nZ+DBbQ5g/0xFe7eTW5sbbrjhqFTAAJ3jOuMOuhs/61JHk9phFfBbb73Fvffey/z585urgI8GEoBb\nMRRxy91yozRSXuejst532HOltV5eW5iPWVWo8wb5YkMZIV2iCEGi3YwmdeZuKuemGd9TVO0m4ZKL\nCRYVUlnvw+UPsbPGR5ESQ2FQpXhHAelde9Oj/xCW1sBnvnjKSyoh+yRwV4KuQdCLyaGCLaGdfwVy\nMOZUOnBne188StsTDAa57rrrKC8v54svvmiOAm42beWxLICTwg5ZEsOB4Ej8FpgspfwmnCR/bwHs\nHAyvQDAU7Y8c+tCwf5tGpJT/AP4BxtPfzx3Escptg2/jie+eoMxdhipUAlqAO0889N4xbdo0nnzy\nSRYtWkTv3r0jIGmb8Regn5TSJ4R4MtLCHI1UN/h5eM5G6r0hQJIUY+GxS04gIcZIlPHFhlI+XltC\nTkosQZefOm8QIQRZiXbMqmB3jY9Yq4qUsMflZ/Inm5j+6+EkTLyM0ln/wd1gJqWhhnivi5yaEoIm\nM2pJDd+qVlyXXoPicLDgP+9wkstLr+5nEdq0FHOCxNxvNAxr9/Dtu4DngSMmfYly9OL3+7nqqqsI\nhUJ88skn2Gy2Vu2/rZTwrcCfad6k/Ap4RAhxLcYW9hohxDTAjpEgH/bzDAx//vgwbZrPI4d5Un6k\n7oCPU6dO5dZbb2Xq1Kk8+uij2Gw2pJSNDlPV1dUUFBSwYMECJk0yHJqWL1/OwIEDiY+PB4wQpZUr\nV3L11Vc39rtmzRrmzp2L3+/nwQcf5MEHHwSMWLNXXnkFn8/HpEmTmDFjBlu2bDkgBKq1sJls/O20\nv1HlrcIT9JDtyD7EEeyFF17g5ZdfZvHixXTv3r3VZYgwL2BkH/oBI8yj5XPoOOeBjzawpdyF1WQ8\nFxfXeHn4kw28dPUwdF0y58dS+mWG/84skGA3s7aohvR4GxX1PmLCtmBN10l1WEl32Fi4pYJzzzqT\nZVvddP7PDHIaKlC1EAGTBV2ouMyx+Lr1xhx2cJTX3sDc5cvpaaol4U+PEzN8OLShQ2MTbAwnbzmA\ncLa1G4GRGIuRGuB9KeX8dpYvys/E6/Vy2WWXYbfb+eCDD7BYLK1+jbZSwt8BqcBe20j1kRqGJ+/l\nBx0+2ONvf8/AdvEKjHSI0g033NAmCnh/Uu2pxmPMQTz55JO89dZbLFmy5FgN8HdhZMsCOPxeapQj\nIqXkx921ZCfFoISVnq5Lvt9pRCXWeoOoyoHKUAjBgKwENpfWIwDFrOD2aYZzY0hnxY5q8ov2UPrp\nV5zyvzeRwRCWoB8hwCxAEeBTIdblpNGQoqqU9RhE+vXD2zSaoBmMFUKMAfyAlFJeGT5+F0ZCjeek\nlLoQwgFcLIQ4R0o5N0KyRmkmbrebiy++mPT0dGbNmtVm2djayiY8GOMJ8DYOk/LvaKClIUrV1cZz\nxv4hSmazmXHjxnHFFVewffv2xu/sDVGKNJWeSrbXbEeXhgVASsnkyZOZNWvWsayAAaowYoWfZZ/5\nI0oL0DTZqIABFEWg6YblJ8FuRjvoV5VSku6w8t4tp+CwqVTU+QnpGoqADbvr6FSynYs2zCV54WcI\nvx80Q0GbFIFVC2LVQyiajt9kPaBPq0mNtAIGuAZ4TEp5BYbiBUBK+QBQAAwLH4qXUr4bVcAdH5fL\nxXnnnUd2djbvvPNOm6ZDbaueS8MT8KhmypQp3HvvvYwaNYqsrKwWhyidddZZRwxRMplM/OY3v+Gh\nhx5qvNb+IUobN25k2bJlbNiwgRNOOOHwAv5MAlqAx1c+TkOwAZNiwhPycNug25j5zEy+/PJLlixZ\nQnr6oU5axwpSyieEEB8bb+XmSMtztCGEIMFuxu0PEWs1biEN/hBJYXuwqgjOGZDB15vK6ZYWRzCk\nU7U7n3u6FpAVDNI/y04odhUNISc+n0rfApXL8n4kxqxgq3di0kKYpYaGAF3HhEAVCmmxJhZ16UtM\nvReBoKLexz3n9I3kT7GXpswbL4bPrQL+StT00eGpra3lvPPOY/Dgwfz9739HUdq22GBbhSitxUiC\nUA+w3/ZMuxMNUTqU19a+RkOwAYfF8PDbWbuTRdsW4d/h5+6r7ub6odf/0tVFxJcmTSGEuAc4GWMV\nvEpK+UxbX/NYCw9589sCPl1Xgi9oLHntZpXLh3fhV6fmNrbZUeli9upiziiexpB4F7aEToTcVfy+\nuAFPl96oigm5cjPDt1fSvSxEvN9PXMCLAExSRyIBQVA14bLHk9s3l85vvMHi/Gp0CeP6pjeV8rI9\nQ5Sexwij/JsQ4jkp5V+ac669OdbmYFtQXV3N2WefzejRo3nhhRfa5T7YJithKeXQtui3I3HDDTdE\nWoSfTbGrGE/IQ5m7DJffRXVVNQFPgIkXTGSnbyfT1k3j9iG3/3RHRy/p4a1DwlvSUVrIzaO7k+6w\n8k2+Ya45s1865w08sHZBj3QH9w1wQX0t1JdCTT5CqKQqkjJvGtaCGk7ZWEyCJ0AnnxeLDkJIdF1g\n0jV0RaArIJNj0IefzqY9tZTP/opzrr2wI2xB708VcPoRzBtNnYsSKdxVMP9R8NWCUKH3eCoyz2L8\n+PGcf/75PPnkk+02x1pdCQshnsKo7LFOShkUQqQCFwHFUsp5rX29KC0nrzoPq2ol1hxLfnk+mtDI\nzM4kPiYeu8nOVudWQnqoWbmkjzbCXvaZQohfhw91iqQ8RzMXDcnioiFZTTfaPh+9dB1FwXhcAYgx\na2RZq6jZup1+WwN0qvWS5vFiC0qENJxUJAJdSKxxElN8LN5UG+bSVTTE9uOzIg9ffbKJv108sF3G\n2BzC5o3/Yewsbm7uuebSRD3hTIztbwH8W0rZ8aq9dESkhM/+DCk9IN54cPRs/ILJdz3AxIk3MHny\n5HZ9yGuLu+yTGNVE7g1XTioBZkgpV7fBtX4WJ8w81Ma64TcbDtPy2MPpc+KwOCisL8TlcqGrOglx\nCXhDXuwmw1VaCEFACxyTShjDa38e+5wSow+GbYjmrSe/TuBTdEyKwOnVyA3aqCn2kVPpJqPBiz0o\nEezLwCOF8TngAc3kQTWlEW9WqXSXkzpoADvrAmwpr6dvp/gIjmwfQoj3McSPCee8vqQ551rAkeoJ\n343h6R/HYfIlRDkCe7aAagHVcPKrq6tj5v8WctuYHgx++JF2F6ctCjjUAq+EX0ctjz/+OLGxsQwZ\nMuSAakFTp07l6quvPiT+d38WLVrE8uXLcTgcXHfddTz77LPY7XYefvhh7rzzTk499VSuvvpqZs2a\nRVVVFZdeeindunU7bF+tTaW7kkpPJe4qN9IssdgteENeOsUYC0Jd6qhCJcZ8bCaSklLOjLQMxzrV\nDX6q3QF6pMXxbbmKn05kKS4Cmm6sdP2d6VRlIqWhFmtYAQONilgAUhXoqkJQKOCswZ+VSLBvGsJi\nJTFGsL64tsMoYSnlNXvfCyHuau65FnCkesIDgDuASow86Dcf/MVoPeHDoAVAGM/gNTU1zJw1k1NP\nOZXBfSNTUvyYXOq0JklJSbz88ssEg0H69evXeLyuro4pU6Zw00038dJLL6GqKg8//DBWq5WxY8dy\n+umn87e//Y3Fixdz8803s2jRIpxOJ3feeScrV64E4PPPP+eUU07BarUe6fKtToO3gT179mAOmBnU\nYxA6OgV1BdT4ayisL8QT8vCXEyPmOxLlaEXX0VzlTFlYQblbx2JScPtC+GuGMtpUg0fUo2su/MJK\nNQ6ciplhvkIEIuyAtc+LJaiaCFoFCElVuo3E2FjEiT3xiFwAajwBhnVt9yINR0QI0T/81gT0b+65\nFrC3nvDBWQN3YyQAaQAOm8Ypmj3wMGScAH4X1VV7mPXuu4waNYqTemdCbusXy2kObWETHi6lPOpd\n8FRV5c47jVSOH3/8MZMnT2by5MnY7caW7bRp0/jyyy/59ttviY+PNyq/VFaSnZ0NwNNPP80tt9zC\nd999d8RrJCYmcssttzB9+nTuvvvuNh+Tx+Phzw/9GXGSQGQI8mvzsapWUu2pdHV05caBN9I7qXdH\nc3ppdYQQnYFhUsrPhBBDpJQ/Rlqmo5rt82HVW+xwagz2JpGRfBLejNGA5AtnGvP04YxhDX1kGch6\nqPaT7fegC4X9tYIEdKEQNEt0oVOToFARK9GUalwuqOo7hrJqN5kJNnqmt17u3lbgivC/Pg7dAWzq\nXHM5Uj3h54GnMX66aT+z7+MLXz2s+xfVITtr5r3NpSP7k5ttg+QeMOjwO5tNUVrr4aX5+TT4Q5gU\nhatPzmZEj9QW9dEWK+EbMWLijhmSkpJ47rnnOOWUU1i/fj0Ad911F9OmTeOmm25i9erVdO3atbGw\n8/Tp0ykoKGDFihWMGzeO5557DrvdTnJyMtOnT2fLli2cc8459OnThxdffJHx48e3+RhcLhcTJkxA\nOVchIz2D9Jh0XAEXIRki3hLPmOwx9Enu0+ZydBCmYKwgPsOYr386UsMmnGLOB24IN3tTSvl1Wwrc\nYfHVwfdvQOYQdlWXk+GAThUfkefcyiJ9CK5gOuXeNPo2mEgQCXgUC5mlVcTF6fhirJjcOlY9iCIE\nfqCqsw2pBXDbVaqTLHilhUWdMwkmnclJ1hiuH5TOiR1vFTyb/XbSDzr3IfsW+T9rJdpEPeE89s3B\nKE2wYc8G5m+dzYn539BLyWH2/z7j/NMGsyd3AqVDr2dEz7QWLz6Cms6DH2+kX2Y8nRLsgOStZTtJ\nd1jp0YKHxFaPEw7HCB/g7CKlvLdVL9ICorFxxtb5eeedR59BfUi4MgGLaiGvOg+LakGTGrrUmXPJ\nHBTRakHpHXopfVDs5stSyqaU8K85qJRm+PgzwEsYNrr7pJRNVs85Zufhun9D0UqIy2DJhgIy69cT\nL+upCNnZLrqxNDiAgQUVdA/m49YEifX16HYbwZgYzNUNuAMKyb5q9ASdjFQfy4cmsoFY7BWAUChN\ntrMxZSAB5+ncOGIgt4/t2VzJ2mUOCiEms08Bg5EA5m/7nYP9lO/ec5HgmJ2Dh2FreT2zVhQSCOlU\nm74iOWkPOaUbqfOHqN+1lTPjRhEwp6JrIZZm3Yw7BM9cMZg4a/PXpfPzyvk6r4LMhH25f0Oajjeo\n8fCFAyCCccKbgNfaoN8oTeAJepi2bhqVnkoUoXBR94sYkTUCp9PJ2WefzYgRI5j81GSe+P4JchNy\nyY7PpsZXQ6w5lgp3RWsq4KOBbcD1QoiBwPc/0fZITjH/Zt8q53eH++Jx4RRjTQA9BMBJ2lq8WjUm\nk5kY4eMU1nNCfR4Nahd0fy0+PQa7RcXsdqHEBtiQZsEVEoQkFJ+aihbjZFRQo0ipJT87C4tJJYiJ\noD+B3KQ0Zq/a3RIl3F7YgV1SytcPPiGlfDQcITIUo2Rm1CbbDmwpq+eZuVvplxmPX/eSV7SDjPJ8\n0MrA5SEhM5MCTwNDHVkEPH6y7EHcajzTFm9vUQY2t1/DdFA2LVUxcqG3hLZQwhVSysI26DdKE9yz\n5B6KXEUACAQFtQVUOit58OUHyb06lztuvIPUmFQUoRDUg5gVM6n2VOr8dXRLbB/P7I6ClPJ1IcQ7\ngIrh1NIUR3KKuR84I/x+BnDtYa5z7DvF9BoPq9+G2HRi/FWodjs+rwe7DGFRVdSglwq3C3ODRld/\nNYrAeJTRNKQtSJJJo6xPIvTMRqkT1LmqOD0AThnEryqU+LNJs5+Mqgj0jqnDOgEXCiHGEd6OPihD\n4GtAZ4yUlicB37S/iEc3pbVeZq/ejdWscMWJ2STHNl3JaNaKXfTLjEdVBNXOQrrpO6nXLJjrfSQk\npWCSftzh5+kG3Qy2eGIVEyU1nhbJdWb/DD5ZV0qaw8LeRe+uajfXnNyyB+62UMIPCSFOkFJuEEL0\nAR6QUh6xwKcQQgEeA+IxbMlBYCxgxSgAAQfZ5MJlDxvbhG0mzWZz336HHOu35cA4+mAwyGuvvcaC\nBQs488wzueOOO5plM9i/3OHBFBQU8MEHH1BeXs6LL77Io48+itfr5b777mPOnDmNpQs//vhj8vLy\n8Hq9PPbYYz95ze2129lcvZmc+BzCsYjsrtvNA988QK+BvRgzagyvrXuNM7LPYNLJk3j2h2fxaT6k\nlHRxdOGWE275yWscSwghHgVexlCe1TRtVzuSU8z/gDfDbb5qO2k7Jq6Ai5KGErrGd8U+4UVY8hRI\nDav0YU3NhOpypGrCY5UoLj9WzY6ZACbVREAPEdKhvrNESwhBr3QCIR2POZ0Sk4urQzo7a5JZah1F\nTPJwhGKist7HSd2SIz3sQ5BS3iCEOK2JRBkuYEPY9HFPe8p2LDAvr5wPfthNbmoMtR7J3R/8yJ/O\n7MWQnCP7BQQ1HVWB9VWrcTkXY5cuNK9Oakp37FQRVMxkBVz43PUUJo4GxURI17FZWqYO46wmbhrd\njZnLd6GGC5ic2j2lQzhmzQRWCyEeB9ZjFFBviosxVhrVGKuO30sprxBCTMAIUodDA9UvPajNO609\nCLPZzJ133onP52Po0KE89dRT9O/fn7y8PCZNmsQTTzzBaaedxooVK+jfvz/r168nKyuL1NRU8vLy\nMJvNdO3alT179jBy5EgGDRpE9+7dmTRpEg888ADV1dV06dKFM844g0WLFh1QuvCSSy7B7/czYMCA\npoV0V8Om/7GqdhNWk/F0WFxfjCfood5XjxKjMOqEUZhVM72SerG4eDEXdr+QqadPxRvyYhImzKq5\ntX+6o4FE4DyMh78rmmrYhFNMm5fT7Ki8sf4NNlVtwm624wl6OL3L6Vx20UuQkA1rZ4Gi4jDr1Hgb\ncMfGYTeBWdMxCUC1EZcZh8kOcgCgKFRVVVKjJiPQyfB0Z5tJI/HCqWyYXwS1AXQZoGuKnScubd1C\nJq1ILyHE7YSXQ1LK/XdF1gIIIeYApRGQ7ahFSsl/15TQv/O+ePATuiTw9rJdvJSTZGS+yvsUChaA\nJZbV3Ucy37mJGrOZ7SWxKPZCVFyEfEEa4hU8MRpea1cSnYV0VRP4LPEq0jO74AtqbKtwMfnCn7jf\nHoaRPVIZ0T2FGk8Qh82EWW25Wa8tlHCllHKqEGKGlPLBZrTvAyyXUk4XQsxmX37VQmDvX93BNjl5\nmDaNtLYtzuVykZ6ezurVqxkwYABz5sxh0KBBBxy3Wq389re/ZdKkSfTt25eioiIGDBjAZ599hs+3\nr2TtRx99xNixY3/ymhs2bOCqq646coNtc2HVW5CYwyn15cz2lFIa8hPUdOqq6jDFm7BZbCwrWcZZ\nXc8CQBUq9YF6kmxJjdmxjlMKgZOklO8KIX76P+M4ZJtzG//N/y8BPcC4nHGMzhoNwOry1eTX5tM7\nubfRUNf5ZvcSRnYeSeaoO6EM1Z0xAAAgAElEQVS+BL1iI1WuACYZIHfwOdRWFxBwetC8cVhzMlDs\nseh1lfT2d+VT706Wi0H01PbQTbNitvbmjdhR3JaTwaJ7urOptJ50h5UuSR06eUwfKeV1RzhXIqVc\nzL4sV1GaiT+kN1af24cgsLdO5sLHoL4cEruwpPx7vildiMwZQdeMVFY7v8Be1RtnqY+zMxz0sNkR\n/grOiulK/zgLpl9Pp25HkKXb95BgtzD1skFkxB821PonEUL85BZ5U7SFN87ZQogPMOq1fhB+3xR7\nA87hQMeXnPC5vTY5OFTevW0OQEr5DynlcCnl8L1hQ7+E/Px8VFUlFAoxYcIEpkyZwjnnnHPA8b3l\nrs4880zKy8vp1asX1dXVxMfHs2PHDsBQrNOnT2fjxo0kJydTUlLCP//5T8aOHcvnn3/eWLqwpqam\nsTTiYZESVs+AkB8KV9K1fg8jdCs+dxVVNVXY4+0oqkJWXBYhPYRfM0ovalIjwZrwi3+PY4AZwDYh\nxB+ANyIsS4djRekKXvvxNeLMcaTaUvl8x+fM3GQkGptXOI8cR46RdWjrV7DpI7ruXsfSz/8AepDg\nuU/z+4UxTN7cE/uNH2NOSEdJSkXpNQb7mVeiaG70+ioqLRpul4W6wK1kKFdQ6v89nsx7cPa5ltT0\nzszdVIHVpDIsJ6mjK2AAsxDiJCFE//2Sc+ylrxDiX0KIJ4QQLV9qHcdYTcphTHvSWG16nFCVD4nZ\n+EJ+5vpK2a7ChpLlfF++ErNaj7PuW8YM7UvPIafT2QI2k43Bwob5zIcRsamcPyiTJyYO4r7z+v5s\nBdwatEkpwxYJIEQMRhC7B9iCoZBHY3gd7q29+SpGsPvS/WzCjW2asgkfk275Hie8Phri0sFsB13D\n6Szk/2wahY50khKS6GTvRFnDbnQBo7JGU+Gp4MIeFzK+a9vHJNPxQ5Rmss+ee7OU8jdtfc2jaR7e\n98195DhyDrgBbnVu5dkznuXtTW9T568jdudSUExI1YIv5OF8GUuvmCwufzMfTdP48MMPsdmMG5vU\ndaqm/wP/ZsPvQto13sg+kaSew1lb4qHOG8RmVqjzhjhnQCcqXT5O65HKhMG/KI1ge5YynLzfx8YQ\npf3OC4x49KlSyl++KviZHE1zcC9zN5Yxe81uuqfGEdIlO6vc/HFsT4baSmHpi5DcnR/KVzHVt4M4\nWxJKyE+ZFk+xt5jOiZ25tN+lAPg1P3X+Oh4Z+Uh7ih+5UoYtQUrpAW466PDBtrb9bXLHtT0OgNpi\nkLqhgIEGj5f87eVc3SuV1cPOIlXTUItW4pEhAr5auqpbuGX8s6THZkRY8A6DS0r5LYAQ4vJIC9PR\n0HTtkBWISTHREGxgYs/LuW3uPQzNKybG6UeTAmEPUTegPyuXzkZVh/HBBx9gsey3PScEheddyff9\nnAzukkCneBu1C7aTpFoZkGXmyw1lqIoFKSUN/hBVDX7OHXh0FLfaLyEHHJSsI3z+GSALWAj0al/p\njn7OGZjJgKwEZq/ajcWkcNuYHqTGWSEUS8jv4qOSxSwNVlOrQH2gFjUoKdxdQ1qvNKqCVczZPofM\n2Ew6xXbi/lPuj/RwDkvElXCUn0FtESR2BXcVLs1CXt5G+uaksSYljR8r1yKqd5CqS4arDq7PvZAY\ntxN2LIFBV/5038cH64UQ/8W4Yc6PtDAdjThLXGMYGxgOMprUsAoHd/13HT22DSetYCMhs8QiJGmB\nRH74fgNpKSb+8/x/CElBdYO/0U728JyNuHwh0hxW/ru6BAC3PwRILKrC2QM6sbKgGqFAnFXl+SuH\nYPoZDi4R4goOStYB7L8SfkVKWdTuUh1DdEmK4Q/julHprSTBHp4XJgsfqH7Ka4tIQCPJrFKtCKo1\nQVKvJPz4OTnzZHom9mRX/S4u7nFxhzXFtYkSDidBmABYILJZYo5Jck+Dde9Trieye/WX9OjZm6Wp\nMTSk5KD6K4mxxBEw2dkQ9PFu2VJu6DwGS0FUCe9FSvkPIcQyDHPMxkjL09H4w5A/8OiKR0m0JqII\nhWpfNTcOvJFZK3aRmWAjqaqEkNlOkqIS1CR7qp0k2kzsOjGHlxbsYFulC1URpDRs51rzItLqs0nu\ndiZYYolJNlFW52VY10TWFNbisJkIapJe6XE8fukJ2MxqpIffIqSUj/7E+V+sgI+UOjV87gRgAdBd\nSvlTMe/tRnWDn1cXbqfWG8SiCm4a3Z3eGT8v3/dH2z9iSfESzIqZgBZgdNZoLvd40b3V2ByZpLnK\nSQ6F0HWItduoD3lJjMukf0p/zIqZ/sn9+bTgU0ZkjWjlUbYObbUSvgsjuXiwjfo/volJZgvd2fHt\nqww6eQypGcms8BVAUjZUVmMWKggFl5C4Y5JY48zjVFt0K3ovQogXMULiEEL8rqm0lccjaTFpvDT2\nJb4v+x5PyMPIrJHYTXY+XLaeDIcNKQVBSwINQRe6vw6HRcFki6VAZpBWV0vfTknYSlbSz/8t31cJ\nhqZU4tv5FnlZl6Mn5pKZYKPSFeAf1w9nc1k9iTHmo8H5KpIctp6wEMKMUb7wy0gKdzAhTee+/66n\nZ3ocOfYYdCmZ+sVmnmyBB3J+hYv5myuIsdexxr34gLz2y0qXcVJ5CRZhJuitxh0QpHoDJCTZcEqF\n5JDGwC6nN+7kCCHQpHakS0WctlLCG6MrjF9OlbeKj/M/JiRDnJhxIvML5+MNeRF7BH+/Yzqz35lF\ndq4Vkrvh3jgNn68GmzUR3E6QRp1Wi2qlqL6YU09pMrXx8YYmpXwMQAjxdKSF6YioinrIyiHDYcXt\nD2HrORBl3UZ2N/gx21JIT5BU2svpUb+NSs8jLN+TTm+fm16xqcRaFZxejfR4BxnliyhL/C17XH7q\nrV8z6dsZCCEYkjaEaxKvOWqrd4VXqv2klKuEEFlSypLw8d9whFSVUspZLbjEkVKn3o2RdOahJmRr\n99SpC7dUkhxrwWoydjUUIeiV4eCdFbu4uxlpIact3sGGkjqyEm18VbAKr2olN15r7C+gBXjcs40c\nWypFbjdxUiM7NRVFD3CqOY3iUB1mdV95WHfQTVZcVpuMtTU4ohIOV4k5ACnlF83sd6wQYgzg59A0\nblGawY+VP/LG+jfITcil3l/P+1veZ3jGcLRKjU8XfMoVr1/BGedc1Ni+R2IPil3FlLpLsab0JFCz\nk3ipU+cuJ6fXOdC1Y27FRAjTfon3j+uA6Wah67DuX9zu+o5ZOx3syDibbQ2d6GNXibfswWutQu8T\nS7lIRhcaUm7lR7POHq2CXqqDnnTF7dew6z6qGvysqv2I0f1MpNiNdKnbarbxTt47XD/giIn1Ojov\nYmwVrwL+yr6ojl2t1P+RUqcOATKAk4FbgecO/mIkUqdWNfgPMStYTQo1nkCzvvtjcS19Oxlb16kx\niTiliTWFNYzokYon6KGgroBu9mTqypxIi6Q01oIactE1pDGmyyBqa3fxQt0O4ixxhLQQVpOVh0c8\n3CZjbQ2aWgn/HVgOfI1xs2r2f6CU8sJfKNdxi5SSmXkzeSfvHewmO+XuciSS3Phc1hStofDrQq6+\n+mr8cX7KGsrIjMsE4LbBtzF93XS2VG+hyF1GakIX4hNycQsT553WMb0CI4WU8o69MZtSyk2Rliei\nbPwINn8Cuga2eBj3EDgOMl18/heQOtakzlxTvYKavJlsP6EHvXJzyep5CXpdIa/UbSBNsbPGVYJu\nMWMOeKlXNXbbJScmlpOl9aKsqpoBWXHUx1eTYt+3IkqPSWfdnnXtPPBWxQXUht83ZuaRUi5ppf4P\nmzp1v4peM4DprXStX8z4/p1YtGUPSTH7POSLnB4uO7ELVd4qCmoL6JXUiyTboaknN5bUEW/bp5Z6\nJw5gQfmPSBkkVFmJd+m39NHqcAUslGZ5iHMkYAv60EQIErvyRvk39Bj8a8bEdaJ7Yg86xWXSOe4X\nhbq1OU0p4Z4YzlVnA+Utca4SQkwCLgXcQKqUctAvkvI4Ys72ORTUFhBriiXFnoImNfKd+fg1P+UV\n5Vxz7TVkdc5it2s35Z7yRiWsKiq3D72d24fezs7anfxQ8QN9kvowKG3QUbvN11aEtwnPA6QQQh6U\nZvD4Yee3sHkOpPbBFXCxpGod22dfin7C5dx0wk2GN6mzALw1kNaHhtKtlG76FmtSBjK1jqWWSix5\n73BeyhAAzOYg2ARaSCekWDFJLz0SU9jtczLOvIs+1z9DID6TlYdRTbpsWeWZDkYVcLoQ4ln2Zfxr\nNY6UOnW/8ze09jV/CWkOKxMGZ/L5+jLsFhV/UGNA53jWNfyHj74rwK7aqXRXkmhL5C8n/YVOsfvC\n0XpnOPggUNz42aKYGVHcg84b36eLcw8lqQKsGpn1XgoTk/D26kFIC1ItQ8RaU1las4UxWz4iRtfY\nEvJxVeYoOPc56MD3wCMqYSllSAhRgZFEwyHCd6tm9tsJ+EpKOVkI8VO5o6Psxw/lP5DlyEIVKlJK\nVKGih3R2lO6ga7euZHU2bBuugIt+yYcWogDoltjtuKuM1EJOl1JeHWkhIs76DyClNz7Nx4zK5RCb\ngUVoeF3lPLjsQZ4f8zzmqu1gicFZ46R4xRySMrJYkaSzA4WGoI3k2BReqFlDP1s6OxDYhcRhCuJL\nzMZqsqMoseyKS4Zz3wCTBQsQZ44jqAUb85a7g24yju4Y9ucxVqsKsH3/E0eyFx/rXDwki3MHdqKw\n2kNmgo38ug3848cCKqvjKAtsxmzyEGPbzUPLHmJ4xnBuHXwrAJ0T7WQnx1Dk9JCdZCewcD7dirbR\no86LzxJDamU9tZ0kWoqFlN0u8vv7cUkPY7LHsLRkKamBBnxaJnUhASSwoPh7eq2cBiNuj+wP0gRH\nDMYTQqwF/gSsABZjrByaSxWgCSEeBE79JQIe61S6fLy8IJ9n5m6hsMpI/OUL+eiR2IM93j0UlxXj\n2uMiNSeVzMRMShtK2Vy9mYt6XoTNFLlUa0cr4W3oBiHEiCOkGTx+kDoIWO3cjBaTgioUEAK7UEmx\npbCgcAFkn4SrajczZswgs1MnauJMbJVWGkQCmi4pdUFl0gASHJkke10E9QBltjgsipmhnYaz1WLh\nihGTwLRva/Ivw/9CqbuU/Jp88mvycQVc/GnoUe2gnoehaPOAg/Plv8i+Vexf21WqCGM1qfTOcOCw\nmfnPpi/YWWbDL+swmzRUYmnw2AnpIbbVbKOgtgAAX/52rl/wT2766BHGT7+Ky9a9ytj4PBqUAPXe\netAg22PDolqwBQVl1bsYmTmSGFMM/oCbOK+CpitYzSpWs0KRP4aSzSsj/Es0TVPb0S+G/40DYlvS\nqZRySth9/gTgpZ8p2zHLrrpdvL3pbcrrGthanMCItHNJsDl47IuN7LGX4BOrsJvs1NbW4ixxktMr\nh3tG3EOyNRlPyMOwjGHEmKMhHT+TywEnhpkFDk2ucPzQazxsm0ttyIPZHAdICPkgIYuYoIdydzmb\nCkqZ8cFqrj19AKQlUe7cjOZIRNoSUICQYmG7tzOFQ/tz48AbuGHeZFaWLWe7u5hg1W7uGHkfPVMO\n3LFJsafw1OlPUeurRVEU4i3xhxXvKOI9oGt40XHwvudh7cXHG4VlicTHeKn01mIxGzsg3pCOiVgy\nYjJYVrKM3JguVD77LLZ4Nxn6j0jVhC+go+wpwK+qCGkiNiYWTWrEWxPwWAV9bNlkfrQCgiE6ZTTg\nTTBRodWQKGOIMdlJFBY21JjouL7RTW9Hz4TG7ZRrgQuAZrnVCyEeA7pJKX8lhHgB+HMryHpMUOOr\n4ekfnqZXYh++3F1IUJYyt2wn52bdhFfdhteTRrdMKxuLNlJbW4ujm4MeKT1YWLQQT8jDwJSBjOoy\nKtLDOGqRUj4qhBgppVwOIIQ4ft3GB1wCtYX03FHMuvpiEk2x0G0MCIUydxmD9cGM/+NdDLrkVf7b\nMxNLTT6b/DuIj/8BRyCIT1hpSBiI2+tmbM5YWPU2ir+ekTlnMhIACavehZ7ngXrorSbRltjOA24z\nXFLKF4UQZ2N4Iu8fMtSm9uKjhU6mk8hzf4qqWPBrdZgUE5qmkmzOZn1ZKbvrtpI+ezp9PSCqi8Fs\nQygqitmP7glhTYQ9ihmzL0CDVaXU7KYyJ4mz51Vz+pBL2FSzmeCOBhZ2C+HKlNTrLrq6/ZyqZ5Mf\nM5hzI/0DNEFTIUqnAr/CWAmnAxcdqe1hcADbwu+jCTv248NtH5IVm8O/1y/DL31YzSYCiocvdr9D\nqj0Fs2ph5/ZCautr6dajGw1aA10TuuKwGC77W2q2sLl6M/1SDm8PjtIsLsXw/Ae4EMPkcnxy2h30\nGfF/xK95mXxXERbNx9Zt69Bqu/LbGdMZMvEPjB8e9mROPplKVxcKg8mkxJdgNkkCQUGi2pPchFzY\n/QMk7h+LKsCWAGXroMuJkRhduyClfFoIkYpRgOaMg849IYT4H6CEt6uPS/pnplK8M5fiQAk+GlAV\nFavMZEVhEV5vJSNFLfPiTLipZZiQxGoBpLARtATxxev4FBMl6YJ5WTolXazEJiZz5nceuvc+GWEy\nsb1uB36LpHuxzpIUQbpuwq8nsNN2BjJ3dKSH3yRNbUe/C0zB2Gp5SUoZakG/EsgUQkzAcNKKEqbW\nX8v6knoUNYiqG7F0ZlUhiKQ2sIdAqBLds5vB/QZT4a/AE/JQ569rVMLZcdl8vevrqBL+ZWQIIXpg\nzNOOHb/QHigq/zf8z5TU7+Gu2UtJ9tuY+/H/OHXib6kIxbIvNbLghC6JfFfQn0G5p+L01uAO2bjv\n7CGN/SDlgZ6oMgTWQ9MVNvhDzNtUgcUkOKt/RmMihqORcJGGU4F8jCINo/c796iUcnL4/QNSyscj\nI2VkObWvhw8LV5NmzUKITKoCRfipQAn050RTiBhLDFglZY4GCjUz/QJBGjxeVBssOtGGjo51wARU\nrRzhKqZ7ck/OCfhJjc8hoAWo9deQaE0iIwiKKZkyVVKuZpKQcjJ/G9ex62Y0pYT7ARdj1F4dKIQ4\nRUr5XTP7nYyxhZ3NvsD1KMAF3S7g87xnMSlgsZqo9wWRUsEiHLg8VSjWIob2G4LZbMYUNGE1WdlW\ns40uDqOkckOwgd5JvSM8iqOeB9k3Lx+JoBwdirnrG4jzxfDlx7OZOHEi3eODlG9bTNz6IP6MoQQz\nBpObEkNI08lJTKBXahqXDetCQoxh42PwNfD9PyCtDyAg6DH0d9qB8/X7ndW8vngHmYl2NF0ye/Vu\nHprQn+5pce0+5lZCxcjp/NBhokH2D4ZNaUeZOhRfFX7JRQP64w7ouP0h0hyd+XJjMfXSjdXfgD1Y\ni4JOZU/YXQA990gssRKXKkndDA1mM9K6ipNPvZih6UMRQpDUyY3u81LkLcGsWABJnUPSMzMRTzCE\nK1TM1Ms6fnRsUzbhIDAbmC2E6ILh4feTSlgIsdcXfO/j8K8xEn809Z1YYAnGDbEP0A0wA78HMoFn\nMNK1vS2lXBSe6I1tWhA6FXEGpA6ge+xg1jmXEmMLYTMrBP0xuMnDZCrHkWin1FtKgpZAekw6Tp+T\noG7s6PtCPkrdpdw1/K4Ij+LoRQgxQkq5Arhvv2MOIEtKuSVykrU9Uko2VW3CHXIzLH1YY4jQXpas\nzWPepx9zxRVX0NtWw8CKL9CVStSGEELZzQavk9XmE7njrF4M6nIYe27uaRBogA2zQWoQ1wkuOtAv\nU0rJ28t2cUKXBPbeItIcVl5dtJ3nrxzSVkNva3YAqhDinxyaga0oXL9aBza3u2QdhGRrMk6fk3hb\nHPG28LwTQWwBiNHcpFZ4cdSC6tdIrRfs9GYSV+mDuBB6ohWz307axhqqUr6hJnsAIzuPJO0Pp1L2\nwINYZD3pMo5doRry+iYTDLrwa36GZwyL7KCbSbNyR0spdwshMpvZ53VAIfA+UEfzChvfB3yAETI1\nTEp5nRDi/4BRwBhgKrAJeDdc/ebgNt82U7aIs6V6C46kItRaCw1eP7HmOHxaKULzcu2JV7K8cjkC\nwYkZJ5JsT8bpdbKhagOl7lISLYk8dtpj2E3RTIu/gN7hB0UJ1AOpwE7g1YhK1cbU+Gp4ZPkj2M12\nVKEya9Msbh10K4PSjZXCp59+ytfvvcG51/+R3NyudP7xQ7K1bWgWEw6bCc27iU7+XUy4+TpyUh0Q\nCsCix41kHgCpvWHM/dD7HON1BPwhHV3uX/nPyC3sDXTcBPs/hZTyNQAhRDL7PKH3nns2vMjYm3Tj\nuOTy3pfz16V/pV9yP1RFJegsoL/2Ayc5Y6mrCJFSLrAGQiTWSuKkmV3mdBzsQW+QxMQK6qw61oAV\n/84iXBk5jMsZh6qYyHr5JRyrf2DOD8+QMuxc+obqEQicfie/GfibSA+7WbR6AQcp5WlCiCEY29Ex\nwB+bai+EGI8RZ2cDEoA94VOFGPlSuwDFUko9nPkp5TBtDu6z3ZOWN4eAFuCVta+QZOpGr4QEGvx+\ndrq+RdQHuHHs1VitVvok9WGzczNrK9fSJ7kP3pCXN895s9EmHOWXEfb6nymEiMNwIKyQ8sjpmo5U\nRi78UHo/hjb5t5RyWZsL3wJ21OxgeelyeiX14pTMU3h17at0S+iGQLC8dDmekIc7Ft3BBd0vIKcw\nhz/+4Y/M+eRTPiy0UFzj4STfLjSzGatZoVOCHRLsUL0d4sO3jAWPGKku08Nh1t4aWPwEnNl0jl6L\nqnC453KT2nEzGjVFeFfuReBfGAPbSXiXJVwcJHdfU3Hc5tFPtCVy38n38fbGtzHlrWL02l1cSQpW\n33ZqCsDpt6H6FMwhSchiwicEXilRhMDhtaNoOg6bh4w6C8NHP4lJMeahEIKE4Sfz+z4v8fd1f8cb\n8mJSTFzT5xp6JvaM8KibR1Pe0R9yYL7ooc3pUAhhAfpixBbvasZW8Zhw2/4YW8414eM5wHrCycuF\nEPXh49UYq5f92xxAJJKWN4fvSr+jpNpMUaCOOKtKUWERuj2Zbr0lVqtR9aN7YncSrYm4gi4u7315\no/0jSusQDhXJBb6QUr7VjK8ctowcRgUbF0b0wO62kPXn8sqaVyh0FZJmT2OLcwsfbf+IoB4kxZ7C\n4uLFWBQLqfZUFKHw4+YfmbZgGnPnzmXIkCGccrLkh11OuheUExvjwB6sgWodYtNAqIaakRKcu6DT\nCcYFdQ2qd0DVNkDAqbdD7OHNn4oiGNc3jW+2VdE9LRZdl+RXNnDdKV3b6+dpbXKllJoQIi8c/vbC\n3hNSynuh0dyBlNIVKSE7AtlxOVy7tjOmf+9AWKwE7R6Ceix6vQe7rqFLFUUKTH4dYQ7hFLGkU49U\nBLFqDHH+EEMm3oGqWg7pu7OjM1NGTYnAqH45Ta2Ev8CIC34VQ+m90cw+CzEybC0Fw0YspTyiTVhK\n+UC43Q0YMXW9hRAvAVaMFcgOjO3oEPCmlDIghFhzUJsOg5QS97LluL/7DvugE3CMG4dQDc/Pak8I\nl0eQ6TCxKS8PgaR3p4GUun5A0zVURUWTGqXuUp4c/aSRuzdKa5MgpbxcCPF3oDlK+Ehl5AYAdwCV\nGD4LNx/8xUjsyOys3UlhfSHdE7sD4LA4qPJWsbl6M90SulHjdeP16kh8uP0eGtbuZuzEsQwZMmSv\nzJzcLQXSuxihRSab4e3sLICELlCyBlbPNP6tL+X/2TvP8DiqqwG/Z6tWXbJsyb13G0w1JBBwoRkD\nJhhCiYmBEEiAkNACAQIJfCSUEDAEMBBieg0QeseG0KuNbXAvclfvW+d8P2ZkS7JkS/auVpLv+zz7\naDV7594zs3fn3HIKAw6B5e+C12+XjYXhxV/DifdBam6zMv7sgH4My8/ghW/W43O7uOLo4Qzp0WlX\negRs/3Pn/0Yawgng0d95X9iWGPwN6mhpNeZKbNuYPOBiVe1Qg8GGzC8s45nZL7L3Z28xvNaDpyaG\nt9JCNEQ0LOAWcHuIxmKgQo+6cmoC6ZRqGr6yCMU9c/jfuGMZPPpEsrGzLf3rw1WU1YYZ2D2NXxw8\nYLvMTZ2FHSnhfVX13yISVNWTG47wdsIuxeRV1TktfLQBaJTjTFVv35U2Eo2qsulPf8KKRPDkdafy\ntdepePU1+tx2K5ZLWF1UR00Ivlu7ELfLzaiRoygL1jDMPYPi4HKC0SA+l4/LD7jcKODEMdTZEx5R\nb0S4o0EiLaeRW4e9alONvZWyHclYkfly85fkBhorv7xAHtn+bH4oXkZplZATcFNeXUFdsZu++0wk\nJa0ZV35fup1NSZxHREoUvAH4+C7ouZc96xWBRc/bvsDisbMweVMhbyh8/gAc/oft63XYp18O+/Tb\nPotOJ6RERM4DvsBOKVjS5PNcVT0Xtroy7QrNrsao6t+cek8EJgCPNj0x2VtzGotR891C/vX+OkaU\nrKJ3bSFejSECVkyJhRW3SwiLj4grDXWHUSuCumNUZOeQOWwEIZebyuow63qPJ8XrpqwmzOXPzmdY\nfgb5mSlsLA9y2bPzueu0zrlquCMlXP/QqDfFba0RV7zSd3U6ar/4AquyBKsuSPX8BeB2Y9XWsuSP\nl3H3UVE04qeycj1kCd2z+lJWIxR4f8Te+aP57Y9mJFv8PYXrnL+tTWHYbBo57KD9t2D/Tu6Nv5i7\nxojcEXy15Suy/dusl9dWrqUyXEl5ZYBYMI/SUDk1RS72Grw3VqQC//oQ1JVDwD7HUosXo0VszMnF\nClfT1xKm9D8K3/cvw8DDAIHBE2DFexCugWgEslNg8ES7QV86VG1KwtW3P06SminAJGCRqt7XpEi2\nk7ULdt1FqaXVGBzbhlNwFG0z8rX7QHBLZZAnPluLFK7iyJdnQzTGAf4CsmrXkOaJElMX7lgMRPF4\nBdTCk5lNsKqGqDcCauHyhAhECimtzKJbdh9coTqGZrhJ8bqZPW8lg7unb535ZgW8VNZF+HRlKQcP\n7nxeYDtUwiJyFPCFiBzIHhxyrVWUFxJ+6grcFRbhNRW4/X5iGYNxZWWx+tsP6T/hGJ5+/Q1Se46k\nILcfgaoaTgxVkB57kZKuaL4AACAASURBVCN7rQDr9+BqMZ+GIU60dZDYUho5J/rRzPhJFh/Gdh+L\nb4mPzbWb6RHoQWFVIZ9u/JTjBh/H/FAtpZFVlNSl0b/XYLqV/MABlrJvSg68cB7sdxYMP5oHv3uQ\nMq+HjNQ83PRgrRXlxS1fcIpa4HUm/W4fDDsaipdAeSEMmbxNiKpN0O/A5NyAJKCqr2Fv3zXHeWyL\nU37eLjbR7GqMiGQC/wSu6Aj7zUs2VfHUM4/iLl5KN6lgxPwl1JUECbiVPuVVeL21UB2yQ5RHwS2C\ny2PhGboPXq+HilXfEc1MxQoGCWRkkRKppaL4a5a7ShkVSWPGCbZZUlFVkG7pjfeF01M8rCqu7pRK\neEdP/cuAwcAN2PsOl7WmQhGZ0vQVBzk7PMG3r+Wr/sriqmI2uZXNoTAb1q9laWEpqzJT+eHZl8jP\nz2fmUeP5kX8lo+Rz9sr3MmXfoaRUr7ctSw2GOHDdj65jfMF4ykPlrK9ez/GDjyfFnULl2sWUlgcZ\n1DOfE0Lp/CJ1AD2kD4MHDISe4+DrR8CKsaxsGRkDD7OVabgaTyzEsnAZ1XufDuVrt7ZjqcWmYCml\nQybAlkVQvcm2oMaCvUymSIfDsEOjTgdm72IdzwMnici9OKsxzvE5QDZwtYhM3F1BW8NnK4u56vkF\nXPmfBbz/w5atx8NRizlP/5cB5V8wPL2W/SM/0KOigqgrxCaN4olugaowEY+HiEtQvwdPqht3wEvu\nz04hb8YM1o/pQd24wdTsM4wN0XIisSiWJ0qKu5L3J4YoCtrtHTgwl00VoUZybakMctjwHu1xC+LO\nzoJ17IrRU3fsSFufAiHsEG4tjRK7BtEQz1b8QGF2Jmn9fIycbwGKR2NsCmSxATfVedlMmXI0IkJ2\npAhPZipjejn7vmndYfN324f8M8Sd5gaFzkymy+B2uZkyaApTBk3h2v9di9/t591332XFsmVMPvZk\nFpcU4a1dTVE0jWEFaduCJ3j8ULWRylAlqxF6jpyKv2wNWFGKuvVDDvktfHwvbPyGNdE6Hguvoyp/\nNGEppVuv/lyZPQ5391FQMDq5N6AdaUV/+iX2UnGoabnWsoPVmJ/uap27wkvfrufNxZvpl5PK6pIa\n7vtgOauKazg9L8TCm+9kfNVGAoEawvkpBGIhat1KMKakVykeS7GsIJZPqMnx0506Ai5Fsnvjzskh\nbfx4vE/dTHVUWVNZS6ggHVR5f2ge3oJReP0uHl/8OJcecClT9+rF56tLWVFUTZrfQ3lthP0H5NI7\nu3PGT0iEn/DDIjJSVW8BEJHOabLWBmq+/y9bakoJaZDa/i6soJe8IiWkaayp9FDl6c/y3jF+gm28\nVR6rY2pGEx82xSjh9mGPGiQOyhrE028+zcaVG/nFL35BamoqVmAjVa5V1Hj8LI/Bt+uFaQUH44kG\nueabWRQHi9lYs5HFpYsZkj2EvhkDyAiVk+ZNg8MuIxqq4tb3f8/IHhPJcbZQyoJlPCo1zNyDFLBD\ni/1JRFKxE9mkYXt3dGreXLSJgNfN6ws3kub3oKrc+/4yDl3+KN1kNR6pw1MXIbjaC/1iVPqV3C3g\nidpb0RG3UutRotSxNF8ZEMgkP1RFqmcl7vRJjJ3+K55/9iHCLqEkw83iXpmEcgZSG1HyA15WlBQD\ntpvbjdPGsmJLFcu3VLNv/1y6Z/iTeWt2i7grYQcVkYewTeo3JqiNjkGwkvBXT1IU7UaAOjyxGEVD\nLTbkZCHFubiy8tg4fAJaW8WWWjux1Alpg9g7vYGlYjRkG8WYPeGEsycNEi3L4pN7PmFz5mYOP/Fw\nSmOlrC5ZjaUW0m88Keu+hLTubFHlxbXvEMobQnYgl4lZ/fly85eU1JUwf8t8Mn2ZXHnglVvrXVK1\nloy07kiD/pqTksPSsqXNidGl2Ul/+qfzt36vSYGz21O+eBKzlG8Ky8nPdBReuAailZRVrqH/gAKK\nglsIaAVSF8JaoqSXg9sxI4u6IOSBiAciKDGNsTZWR//RQ/FsnAfWOQw77nRWlg6iatU7+DI+Y5g/\nTCSynI1WNjHtQ5rVOGnN4B4ZDO68rm1bSZQSvhrbkKCcLpzIOrRmDRv/djXFFRs5SCN4qsOk1ypB\nv/DNsCDSUxiYVseA0Bt08+/Dn3/suBIGK+H1K+xODPYy4NE3J+9C9jy6/CAxFovxy1/+khUrVvDB\nqx9QruVsrt3M8JzhnPrqqWxyeSCQQkHdFvbyZPJJXj8ycvvSy58JwAEFBwCwvmo9xw0+zp4FO2T5\ns4jEGrs1WWrhdTWORb0H0Wx/UtWzHAX9PYCIDE+WgPGgvC7M5oogxWUVTOYzxqYU4dYgqZTj0u70\nzE6ltqgYV4WFesAXVsIe8EYFUdvLTYLKl8MtikbGyEqJcuDg/lBVBNE68KWRJUGOCqyiKJTNClcp\nVd4Y/a0iNpT35JeT23X1vd1IlBK+A0hT1XNEZDa7bhXYYYlVVvL5FdfxVSyNfaphWFktEYXaVAjU\nWExaVMOKwUGCfS38YvGrlHf5w3Pj8XpcnH5gP0adeB/UloJakJa38wYN8aTLDRIrw5XEYjFeWfkK\nP5T+wIfzPoQqeP3110lLSyODDPpm9uVvn/8NRclNyYXqIqpidfwQrSNANinukUSt6NaQgAChWIie\naY3DxvfJ6IPf46c6XE26Lx1VZXn5cmaOntnOV91h2FF/OodtRq2/AP7YjnLFjZe+XY9aUBuO8WPX\nEga4N1FdF6NvLEwgM4RuXoTH5UNxI6IE01IoqcmkR2QzLlWwPZGwXIo3BlEPpLkh05sJFNn+5cA5\nBctZtcUiNZbK2LoAoeowaX4Po6wVDP/kCipqw9y/rh/Pxw5BEUYUZHDPz/cj1ZcoVZZ4ErX+GcOO\nnAV2Eocux8pHnuLTaDoRTzruaAxXzMIXVXIjMXJjMTJCMUaXrsfl7PG+WZrP/HVlpHpd/OOdpXy+\nqsSOKGQUcDK4A7hOVSuBu5ItzO5QHa7mmv9dw3UfXcdpr53GUz88xYdvfki0MsrY88ayqnbV1rJR\nK8qG6g30TO1JbeU6qCvF702jCosjIm5mblrPkrIlW2e5xXXFZPozt6bRbMifDv4TqZ5UCqsK2VCz\ngZ+P/DnjenTaLEi7y476U3cR8YtICp04t/pbizdTFYrSJ9vHCHchmdW1DFi2kcCyEjx1ASx3NbXe\nIHSLoAWK3xMhxV+DJT47MIdHCaYoVelCelAYWB5lXNgipeh7ovuetdUWZsSIsRzWRxjSI50emSmM\n6dmN47pXMs6/gVhmP14u9FMQ28AB8gP9u6WxpqSWCx77Osl3Z/dI1PAhBIx0shx1ibA4AGtLanjw\nw1XURmK4PltPStSFKxYi5vEQExeuGFhBF7gsUvxhwv4c+nVLZU1xDWW+XqT5PHyyooSJI3vw5OeF\ndnhAQzLoMoPEf3z1D3JScsjyZ7GibAXrV61H85SZh8zE5Xbx7NJnGdN9DACK4qopZt/KItZXFVLu\nchHxBuhtCZP7HYJsWcyVe/2RJ1a9TEmwhNHdRjN92PRm2/W7/Vy474XteakdmR31pzuABxu875TE\nLCVmKT2z0/BVCz3Wl0JMCfi9BPETqwsQG+4m4IngyrEoW+MhMzuC1oI3YkHAwuUTchV61kToWVlD\n2agz+V3wSGo+TsXz2Vecsn9fJgw/kFTP/fyolwsCPSEahPkLYOx01pTUEolapKd4GRlazmZrHN0z\n/CwvqqaoKtRpjbMSpYRvwLYSdNG6+Lwdni2VQf7030XUhqMUFlVSXjCOo2s/JlfrqE4JkOmtxheM\ngiW4cGFFLKpTfWiomnXRLFZ1OwCP20V5XQQQLMvEPkkiXWaQWFxXzPDAcDZWbmTt2rX4xEePfj1w\nu92ICGErvLWst6aEfTavpCKnL31qS+jjDRANVjE8vTeCgHjom5LLHw5sOdykoVma7U8i0kdVv8HO\nqV5/TIB8Ve1UIcXS/R7cLsEC1tb2Yj9+wHK7cIngClVTSwqhWg+BzDD4wTc0yuKaPvTwltKtvBJ/\n2EuauMCKkt0tSFpeOlcFpzG6V+bWNh7/bA0jembQc9o98NlsOzRqSjb02Rd8aahWEYzGCFVb1FkW\nq4qq6N0tE5cIwUjnTYWZKCV8aX2gchG5iU66D9KQOR+vpqq0lCHfz2Vq1RJcHqU210NKSQSNCiUZ\n2aSkh0iLBAm7XPh79GNVeS4l3abyTiiD7k6kIQGCkRg5adtnAjG0G11mkCgihMNh3njhDdxD3Ywa\nOIqSYAkiQlW4ir4ZfbcV/uZxTu11KC+XLWSjFSOrtoKRMWHfiAs0Zr8yOu2KaTJpqT/NFJEh2BGv\nytmW1vBeoFMp4cuPGs6lz3zL0s3VVKcN44fIDwz2FZHrCxOMplEU8eP3h6mPdlwrAdan5eMbmkLW\nqloyqutQlyIFsKlXT97yTWVQXmqjNgbmpfHsl+v47aSh8OPfbvvg7esgVEVlMIoqeNywWnvj9vpY\nurmaMb0y6ZvbuK7ORNyVsJMCcZSIjMH+Rrbs5JSOjyqV65cx8YenOahsESlWCHd1DNaBZMIWbzZW\nrVCbnk4wO5toaj5lgf7kVJfyQ9ogxvWP8fWaMkQg4HWzprSGm0/aO9lXtSfTZQaJA1MH8vBzD9Mj\ntwfjRo/j66KvSfems6piFT5x8ztXD5h7M4w+EQC/28f0Xj8hFgsjy9/FFa2G2hIo+gEm/3knrRla\noNn+pKo3iogLGAlkAg+ralHL1XRcemSm8Mg54/liVSnvLt7I4DWpDO6zH2W+NGoXvMnmnHQ2e7KJ\nqJc0CbLaKkDFRXrP7nyc2pNesWLSUjyURZRl2pvXYocytryOAXlpLNtcRThq0TsnldzmJieHXwWv\nXkJlUSmDs9L5pCyTDxlLuC6Kxy1cdlSnNjpPSLCOk0VktKougs7rhxmKxnjrmxX0+eZ2hslmpn2z\nmvTKOtLqgkSrXcTCghVx4XXF6NutmHJfKqvKc5Fx++J2e3CXVXDgj/emMCeVxRsqGN0ri8yAh5P3\n68M+/XI6ZbaPrkBXGiSWlpby6IWP0vf4vgz7yTAsLKYPnc5+BfuRZwn93/srZIXBlwbv3Wjn/y1Z\nDj33tnOyDj8a1n0JE6+F3q1KF25owo76k4hcBozHVr6vJEnEuCEiHDioGwcO6oYeeQffPfQk73y7\njvU5BxHoEaQ3ZSyyBuAlRoVkELE8LEsZyQuxMQy01kPNFmpSexNK74OvJsLcJVvIW59CZsCDiPDd\n+k1cO3XU9g17U2DaPbzx2Pv0yEghKzWX02NKTJXC0jp653TeWTAkbjn6HLZlX/o/4MqWCorINOBY\n7JHiv4Cx2LGqvcD5QE/sfK0x4N+q+r6IXNqwjKrGNTtIUVWIP/xnPuPLXycjFOSzr0voHa7FXx4m\nVO3B5bfAEgSw6lz4rQg56XWsjAmV1WHqIkHGDsgn71fncr4JwNGh6CqDxKKiIo444ggmT57Mrdfc\nut2gLvLaFWxKz8Pv8ZDjTYUeI2Hjt3aShgVPQSwCbo+97GcU8C6zk/402vl8NrDLSngH+YTHAFc5\nxf6qqgt3tY02y+Tz8XT+vnydWc3KVWUE+kwmjTpqwhY1pBLwKBmpHibl9yFlXTmLqnvRM3/w1vPT\n/B4U8LmFunAMl0uYOKIHry/cyPF798Ll2n6SMm74YL5aU0ZPBI9biEVjuETo1UnDVdaTKCXccGiS\nvqOCqvoi8KKI5GCnh/Op6hmOkcMhwOHA37BTzz0mIh9h5zpuWObDeAp/13vLGJ6fTr+SDXi/W0Vq\nXQVpniD4IBLzoBHB5VZiMUE8ijumqMsikhfg48NPYfygXH58yKB4imSIL60eJHZENm7cyOTJk/np\nT3/KX/7yl+0U8Debv+G/Rf8jkJKDqzZGr2iMU3oditeXZu/5njwnOYJ3XVrqT3lObOn8+hjTuxin\nvNl8wsDFwAXYM/BbiHM8hmWbK3l1wSZ6Zadw/LjeW1MHgh2N7cP/fURR4Wp+MWMGH62qoCacgtZE\nyPUKFXVRKkLKB0uL2LtPNuvL6gjHLHxuF9GYRUVdhHS/hyNHF+Bxb5uo1GyOUlEXadZm5pT9+1Id\njPLlmjIsVbIDXm6Y1vnDpCZKCX8kIs9jd46XW3nONdim/Cc7/6/BdoDvAxSqquU8bLoBRU3KNGJ3\nElnHKivJmvcW7roavJuWkrGpGG84Rq3LRyA7hDvFTkjtSYvhiUEs6kHdQsAfY+ovf8opE/YzS80d\nn1YPEjsa69atY+LEiZx55plcc801WGoRtaJbo1VZavHvRf9meKA74vGDuCmyIswt+pojJB1yByb5\nCrokLfWn57BjS7/g/N3VFbuW8glnqWo5gIg0G79xV5+FD364kq/XlNE3N5UF6yp4Y9Embp2+N93S\n/ViWxXnnnUfVpiiTf/YbAoEAhw3388HSIqpDMVziYkTPVA4d2h2XCDXhKBl+Dy6BqmCUgM/FEaPy\neWXBRjzuxs/KmCoZKc2rJRHhnEMHcc6hrb6MTkFClLCqPgo8utOCbDXZ/xvwOvAF8Gvno37AApxc\nmiJS6RwvAfKalGna/i4lso6Vl7PhD1cyuELxF64mo7YCr8Ygqli4qIuk4EmL4s2I4Uq1CLt8+Api\n1KWn0nP8JGTCaSYBQ+dgVwaJSWf16tVMnDiR3/zmN1xy6SXM+noWK8pXAJDqTeXS/S+lJlJDijsF\n6XcwLH0DAjn43D5W1W6AYYdDSlZyL6Jr0mx/UtWH41R/s/mEgQoRyXLabTaf8K48CyvqInyxupQR\nBbb7UIrXTXqKh3vnruDKo4dx9tlns3btWj54+WUe+XITCworcLuF3tkB/jR1JA/8bzWjem5zPUrz\neejXLRWP20X/XC8iwoqiGi6ZPIw3Fm1iaH46HpewsriGw4d3bzQz3hNIhHX0LWwzxRdAVfWUHZxy\nETAZyAKGAF+LyJ2AH3sfZAW2ko4CD6pqWESalokLJY8+RrSqkuELvyESBTegIojb7ruWJVguF1WD\n/HzrHcwy31ACA4cT6DWKC6f9JF5iGBJMWwaJHYVly5YxefJkLr/8ci688EIe+u4hKsOVDM+1LUMj\nsQh/++xvXHvwtbZvcEomjDoeNs5HQ9Ws6T0WDvldkq+ia9IO/el54G4RORYnn7CqzgDuZFuErlvi\n1diKourtwkD6PW62VFRzxhlnUFZWxmuvvUZqaioXTMggFI1RHYySm+bDUnjwf6u2qzPV7+Hu0/bl\nq7WlhCIW4wd1w+t28aOheTz+6RpqQjF+ffhgxvbOjtdldBoSMRNu6ueww9GXqs4CZu2gyAbgzCbn\n3L5rou2YmlWrCc7/Go0JHkBQW/qYC/wWLq8FmfCNDuH93IkEhh2OZSm3TNkrEeIYEsAuDBKTzvff\nf88RRxzBddddx7nnngvAkrIl9M/sv7WM1+1FUcKxML3Te1NcV0xeIA+rzwEsL1/O6SNOT5b4XZrW\n9CcRyceeMKCqa9vaxg7yCS+kybMxHgzKS6Mu3Dj4RV0ozAfvvkVBTQ0vvfQSKSkpWz/ze9z40+39\nYrfAwO7plNWGyUm193W3VAUZ1zcbl0s4YEDjKIG9swNccfSIeF9CpyIRSvhuYAy2qX4Y6A/sm4B2\n4sqs17+jZFkVJ1uCC0cB16NK1OWCgJtSfwafuUdz8kmn0jsnldG9zfJeJ6NNg8Rks2DBAo4++mhu\nvvlmZsyYscOyqorb5ebS/S/lmSXPsLB4IR6Xh7NGn8XePYxfeoLYYX9ysiuVYK/kKZ3AJz071cfY\nPpl8v7GKAd1SKa8J8cLbHzKoYiH/ef55fL4dBxq64qgR3DN3OUs32Svk+/bPZuaPjC1CSyTCT/gs\nEblRVa8BEJGL491GvKksr6L0X/dzXPmnuJvzdhKFHGFT91ze7HEAnvEncOSYntuXM3QG2jRIbMk9\nxPlsLPAuMEhVq+Mt6FdffcWxxx7LrFmzOOWUxpP1/fL3Y1HJIvJT8wGoi9YR8ATs7EjAqSNOjbc4\nhubZWX9amKiVu0RywYShfLu2nBe+WsPrzz3KGNnC40/MwePZucpwu4SLJg5tBym7Bomyjh4oIr/A\n7pQdeq1BVVk87adMrS0irTbUbJlwtpu3x4/nLT2ATZrHt6eNb2cpDfFiFwaJzbqHiIgX+CW2QWHc\n+fTTTznhhBOYPXs206ZN2+7zk4edTPj7MN9s+QZLLfJS8/jjQR1+ktXlaEV/OkFEBgM1Tvkr2lvG\nXWVwjpv3/3ERew0ezIMP/hu3u1O61Hd4EqWEzwaOcN536Jlw2U2/IlC9EV8sau/oNCHmFRaPHshc\n3ZcN3oGcf/gQXCYAR2enLYPEltxDLsO2Zbi2pRN31T3kgw8+YPr06cyZM4cpU6a0VDc/H/Vzfj7q\n562u15AwdtSfzmTbEnWncZ0oLy/nmGOOYe+99+aee+4xz7wEkqg7ez5wghOqrcOmZCn86ElWLF5A\n2OcFUcSltkk0gFshTQn281Fc0IdQ9lB+N3kYF5pllq7A2dj7dFXsfJBY7x4CjX8v47At+w+khSAJ\nqnq/qu6vqvt37969VYK98847nHTSSTz55JMtKmBDh2NH/Wkv4DZs6+VOkXC5pKSESZMmMX78eO69\n916jgBNMombCg4FC532zTuTJJhKz+N9HH9IrO5cUbwmRmAc/djJz9YC3b5Tlqf04ctY/2b9gFBcl\nWV5DXDkfGKOq54rItdhZcFqiWfcQVf0ZgIjMAWbHQ6jXXnuNmTNn8vzzz3PooV0sIkHXZkf96bh6\na2kRuRf4bzIEbC2bN2/miCOOYMqUKfz1r381gYfagUQpYQUCTmzTXglqY7f4cnUp1Z5sPEP6Ubey\nDCuaQtTjxq8RNufk8Pm48Yw79QpcBcYAqwvS6kFiS+4hDT6fGQ+BXnzxRc477zxeeuklDjrooHhU\naWg/dtSfAiJSvxeR1n4itZ0NGzYwadIkfvazn3HdddcZBdxOJEoJ/x34DXYy6w5pLeLzuFiTPo7A\n8n9jDUwnv1pxV0fY5Mrh3X1n8rsZE9i7757nOL6H0KEGiU8//TQXX3wxr732Gvvtt1+yxTG0nR31\np+th60LajlZcksratWuZNGkSZ599NlddddXOTzDEjUQp4dNVtUMHxR+dn8ZZC4pYpSM4cZgQlTCF\nkVT2OeZsfj66Qxt0G3afDjNIfOSRR7jyyit566232GsvE/Slk9Jif1LVFcDlyRCqtaxcuZJJkyZx\n8cUX87vfmahq7U2ilPAhInIVUAGgqnELLRkPgsEg06dPp19KJgUnX857EQuXSzhmTAH7jU76xMiQ\neDrEIPGBBx7gz3/+M++++y4jR45MtjiGXadD9KddYenSpUyePJmrrrqKX//61zs/wRB3EqWEb01Q\nvbtNbW0tJ5xwAt26dePRRx/G6/UmWyRD+5P0QeLdd9/Nrbfeyty5cxkyZEh7N2+ILwntTyLSiyY5\n1Rt8dit2LvZs4Jeq2mwih+ZYtGgRRx55JDfccANnn312PEU2tIFEJHC4HNtt42HHRanDUFVVxdSp\nUxkwYAAPPfSQcT7fc0nqIPG2227j3nvvZd68eQwYMCCZohjiQ6L70zk0yKkObFXCqno5gIj8HtsF\nqlW51b/99luOOeYY/v73v3P66SaueDJJxEx4lKqeLCKzgQ6jhOudz8eMGcPs2bON79seSrIHiTfe\neCOPPPII8+bNo0+f7VJhGzoZ8e5PTijUvzY5XETjnOpNzykA9qeZRDjNBYz54osvmDp1Kv/85z+Z\nPn367ops2E0SoYnyRGQKkC8iU5z3SaW0tJTJkyez//77GwVsGKWqJwPHtXfD11xzDU8++aRRwF2L\nuPYnVf1OVac2fAErsXOqb/fgEpHe2DPxC1Q11kx9jQLGfPTRRxx77LE8+OCDRgF3EBKhjZ4DugMv\nOH/z4t2AiKSJyMMi8oCInLGjstFolAkTJjBhwgRmzZplFLAhKYPEdevW8corrzB37lx69jS+512I\n9uhP/wKuAB50XohIff7il7BXNP8mIjtMlVVVVcW0adN49NFHOe64dh+DGlpAtLmsQR0cEZkBlNcH\n1a+PXtQcgUBAL7vsMv7yl78Y5/P2o8PeaCfGb0NUVR9JdLtpaWlaWFhIbm5uopsy2LRLH0xWf9oV\nvF6vvvXWW0yYMCHZouwptKoPdlYlfBXwuqp+KyJPqOrpTT7fug+CnWZsYZxFyAOKO3B9iaizLfUV\nq+rRcWy70yMiRcCaNp6WiH6RzHbas60UVR3TDu10Gprpg/H4LjpCHR1BhubqaNVzMFEuSommPqj+\ntzSzpK6q9wP3A4jIl6q6fzwbj3ede6qMexKq2roMDg1or3vent9te15TotvobDTtg/H4LjpCHR1B\nht2po7Mq4UZB9ZMtjMFgMBgMu0KnVMLNBNU3GAwGg6HTsSeYCt/fCercU2U07Jj2uuft+d12xWvq\nrMTjHnWEOjqCDLtcR6c0zDIYDAaDoSuwJ8yEDQaDwWDokBglbDAYDAZDkuiUhlmtQUTSgHuAMDBX\nVR9v5XnTgGOxM5P8CxgLDAS8wPlAT5pkNBGRSxuW0WbW+B155mEn+R6+O3U64etucGT8EogAEwA/\nUJ+PrNG1i8jpDcs4xm0N5euHHXu2FFgKrN3dOg2tZ0f91Ykn/C4wSFWrE9GOiPQErsIOMPCUqn6U\noHamADOdYg+q6lu7045T5yDgaiBLVac3OD4G+5oA/qqq8Y4X0OnYnYxMO/hOW32fd1DHldjPuzzg\nYlVd15bznc9a9TuJx28grv1bVbvkCzvB9nHO+6d34fwc4N/A487/FwKHAtdiK2YX8ATga1qmhfr+\ngh167vjdrRM4EXgYuB2YBDzrHJ/qXPd21960TDPyHQv8vP6ceNRpXrvfX7EHYXc633d6Atv5O/B/\nTlv9E9jOrdg+/j2BO+J8D59r8v8D2AolC5id7O+4I7yaPmtaKPP75p5jO/hOW32fW6qjwecn7uhZ\nEo/fSTx+A/HsALCv3gAAIABJREFU3115OboPUOi83y6weSu4BjtOa5Hz/xqnzj44GU2c492aKdMI\nETkCWAxswe6ou1vncOBjVb0Ee5aqTco2d+1NyzTlU+AcEXkPeCNOdRpaT0v99TLsFYp4WVC21M5o\n4BHslZprE9jOU8Cz2LHlH4pDOzsiS1XLVbUCyEhwWx0OERkrIq80fAGDaPysaXpOfUamj5v5uKXv\ntC33ucXnsoikA6cAL+7C+W35ncTjNxC3/t2VlXB9VC1ow3WKzc3A68AXbEtA0c+pcx2NM5qUNFOm\nKYcDBwGnO68eu1nnOqDMed+wAzSqzznW9NpbkvEs4DpVnYg9K45HnYbW09L9HQdchJ0u77wEtlPf\np6qBlAS2cxVwGPAT4Mo4tLMjKkQkS0QygVYnu+8qaJwzMtHyd9qW+9xsHc659wJXaJNl8FbK0Jbf\nSTx+A3Hr313WRclZs78bCAL/09bvCf8W+AW2Av4WSAX6s21vtCd2gu0o8JiqvicilzQsoy3cVBGZ\niR1bdNju1CkiqcBdQC3wA3bHORQIABc4xRpdu7N/u7WMbr8nPAZ7BFiM3Qm/3t06Da2naX8FjlbV\nGQ0+nwNcqPHZE96uHREZhb1doth7WfHYE26undOB+ni672gckh2ISDfsZcQjsFevRjltjcG+JoBb\n1OwJ1+8JN33WPOrcr6+w7UEqgHtVdX6Tc1v6Tlt9n3dQx/PYS8rrgWdU9b22nN/g8zns5HcSj99A\nPPt3l1XCBoPBYDB0dLrycrTBYDAYDB0ao4QNBoPBYEgSRgkbDAaDwZAkjBI2GAyGPQgR8YnIXSJy\nh4jMEZEDnePP7WJ9bdIjTdsRkQmOHHeLyH0i0ubc2w3qGiAit7Xw2aki8m8R2a+lMsmgy0bMagkR\n8WE7ZcewHczvUdXPReQ5bRBtpw31uVryuWuhfKN2ROQH4HpVfUpELgJ+pqqHtFWOJm3MAULYbk6z\nsX2TJ2Fbj74NfK+qdzYofwfwR1WtbWX9BdjW0PHwJ92j6ID973NsT4C+2NHb6oCrVPWKJucNwLY6\nvaytMu6KXG0pLyIPqOq5iZCri/JL4DVVfV1EPMB/gBOA4SJyIzDYKXMNtqtOoareLiJXYz9TMrAD\nejyJ7U+8XkTGquplIvJrYAnQHdstMxPbkyMT+BWwrKEgIiLA74Bp9R4gIuIWkR8DxwEFwI3Y/s0/\nAQ7GDsixEduVMgD8R5tEpnL66yPAS9j+v7/F7t+LsT1OGvVpEfkbcB8wtGG9QC9sl6OVgKWq/yci\nl2O7ZZar6rVN78tOXKy2Y49TwnSgDuiwCLtzPQWMATYBOHUNw35QXwvsjd0ZegCXYEfeatQ5mtR7\nqSPrFar6e+BZ58F1fsNCjnsHqlrrKO+1wEjgA+yO5gH+gN3x1wEfqeqLItJbRHyqGm7FPTdso6P1\nv7WqeoHjWnEw0Bt4DraGMfRg94kXgANE5BZshX06dh88zpHzeqe+67HdNl7Gdne5zjn/EWwXlJlO\nnZ+o6mMN5Gh6/T+i8cMwD9vfvgo7KMMoEbke2yXpOxE5UFU/3+ndN4D9nHkGQFWjIlIrIm6gWFWv\nEZGTsb/bPthK7E0RGYn9nPoE+/seib2SeotTx1POjPgn2AP/edghJIPYvrsTsGMRZAH/bCBLd+w+\nrmKHfDyDbeEgfdhumD9V1VtEZCV2/3zUkesroNypv7nwkItU9TYR+TuQDszF7ts7cvP7bZN61wFv\nqOrTIvKk2AFFxqrqmQAt3Jc29cM9UQl3pA4Itk/aWhH5KfaXf6TzRZ8JvOmU2Re7U7qwH2STneNb\nO0cz13krtg/vzmYXB2OPDuu5D/uH8QtVvcRZOvIDadgBTD5wyq0E9gE+20n9hsZ0tP7XV0TuxH4W\nXAHMAWY7PpNhVb0cts4alqrqFc7KSQFwpqqeJCL9sQMlAFyrqqtE5FlsP8oy7PCIC0XkUWCFU24f\noKESbnr9p9P4YSjAAuC/qlooIotV9XpHtq+dazRKuHUsBvbD7lseIFVVY/akFNgWdeps4MfYE4TL\nsZXa9fWFRKRWVaPOv+855deoqiUidU3KHobtmxxqIksRdh8UVX1NRHpghwyeCZyGPRibICIZwE3A\nOY7CdgE3Nmi/OerjFkRwZr9NCLFNB6Y5fxvVK3Zsh/p6xHk19Ot1Nb0vbWVPVMIdqQPW8xj2A3Y0\ncCT2F72+SR3/VdUTROQX2AFEoHHnaMrl2LOKnwF/bvFu2LOk8gb/VzrHKusPqGqViMxwZLsbe1mn\nzClnaBsdrf8VqurFDcp6VTXkLBM2XeaucP4GafxQU7b1wXr5VVXnObOXc0Rkb+yZzZ2qWkbL1J+/\n3UPWqeNWEbmWxg9C0xfbxgPA352ZZxbwV+d4noj8H3YihXOxA6C4gJWqukhELBG5HXt14qYmdT4L\nLMceDAE8JiL3Y29vvIo9OLwFZ6WvHkeh3gHMEZFy7Bnr37AHkn/GVo5lzvso8AcReRV7NeRBESkF\nvlTVJ3bhPmwCssQOjLSXc6xRvU1PcJ6Fi0XkH0Cpqt7Q9L6oamHT83bEnqiEO0wHrEdV14nIIFWN\niEj9F/25iNyF/XB7CFjsLEmOBN5pzYWq6nwRuVRE8lV1cwvFljWQu1nEzi7yR+x9zEXO4cHAK62R\nw9CIDtf/mrBeRHo4bQacJek1NP9dPyYis7EHhTc4x24QkVrgSRGZgL1cnYtti3AzcJeIbAZWq+pd\nDepqev2lNH4YpmPv11nYYV0XiW1ccyd2X1zaimszAKoawk4M0/T42CaHrmjyeVN7gOkNPivDjnlf\n//8j2FsQDfmAZlA7k9P7TQ43t3XSlDebHlDV1dhxpKn/q6r14SOvb1C0vsxZzt/bnb+rm6vXqedU\n5+/NTY7vlp2EiZi1h+PMeO5tulfcinMert8XMXQdROQAYG9VfTDZsrQWZ7D6p53MsA2GDolxUdrD\ncSwSHxc7HnVr6Y49AzF0MVT1C7Zlh+ksvGYUcPIQkZliZ2l6WESucI5dK3aO8qZl5zg2Lzuq79Y2\ntn+4s3fblnOuFzvmdWvK3iS74Ta1M/bE5WhDE1T1wzaW34KdltHQBVHVZpfjOiqq+nqyZehqiMgQ\nGlu6z6exy8/Z2ngZ9T5VfUVE/uMYGg5R1bUisi+2C1IxtiU72Pu6w53/52Jvl1Rg77H+BXtLAtlF\nV6B6Q8EmHgO9sI0BM9iWkAaxLeyfcwwHn1LVU5u2g22XcRb2lk7cMTNhg8FgMDTlfGxL918CP3eO\nLVLV27D36wualD9XRB7AdmUciZ0NCWx3yvNV9RJVrffCuA/bZe4kbGPPD1T1d0B/sVMa1ucWHquq\nFzkKuN5DoALb2nlkS4Kr6hq2pWv8CbaixzmvN7YybpYW2lmEbaWdEMxM2GAwGAxNaeiKU/93Ry4/\nD6jqKwAi8iO2eVw0dekBW8FFnTqa+7y589rqCtTUY+AUVT1eRK5jm3cJOG5Kjp1LoKV2GngvxB2j\nhA0Gg8HQlNk0sHRv47nLsYNuAPwDuMexiJ/TTNm3gPtEZCy2u1xlAw+R3XEFauoxsNHZrz4Q2/2p\nntexA3QswjZUbs4ToQLbTSohGOtog8FgMMQVEbkHe1821g5tHQ4MUNU5Cap/BrBZm4TGjBdmT9hg\nMBgM8eZeID/ZQsSJ4kQpYDAzYYPBYDAYkoaZCRsMBoPBkCSMEjYYDAaDIUkYJWwwGAwGQ5IwSthg\nMBgMhiRhlLDBYDAYDEnCKGGDwWAwGJKEUcIGg8FgMCQJo4QNBoPBYEgSRgkbDAaDwZAkjBI2GAwG\ngyFJGCVsMBgMBkOSMErYYDAYDIYkYZSwwWAwGAxJwihhg8FgMBiShFHCBoPBYDAkCaOEDQaDwWBI\nEglVwiLyqYhcn8g2DAaDoaMhIgNFZIPzGp5seQwdl7grYREZLCLLROQzoCDe9RsMO0NERovIFyKy\nRkReEZHUZMtk2ON4GugJWMCWJMti6MAkYib8a2Az8OME1G0wtIZrgaGAYvfDI5IrjmEP5Arn7xBV\nLUuqJIYOTaKWo9V5GQzJQIDXVXUA8BNgblKlMRgMhhZIhBK+D3sZ5osE1W8w7Iz/A0aJyGbsZUFf\nkuUxGAyGZhFVM2E1GAwGgyEZmJmqwbADRGSQiPxLRJ5rcnyMiDzuvMYkSz6DwdC5MUrYYNgBqrpS\nVc9p5qOLgQuA3wAXta9UBoOhq9DllfDRRx9dbyRmXgl+vfDCC5qfn7+n7G9kqWq5qlYAGc0VEJFf\niciXIvLl6NGjk/79dKTXl6tL9dr/LtQ/v7RIl2yqjFu969ev1xEjRuwpfbBNmGdh+7wsy9Lzzz+/\n1X2wyyvh4uLiZIuwR/DUU0/xm9/8hjfeeCPZorQXFSKSJSKZQFVzBVT1flXdX1X3DwQC7Sxex+XJ\nz9cw5+PVZAe8pPrd3P72UuYu2X1X2jVr1nDYYYdx1llnxUHKrod5FiaeWCzGOeecw6JFi1p9TpdX\nwobE8/DDD3PJJZfw9ttvs88++yRbnLgiIt1E5D5gHxG5SkQedT66E7gLuNv5a2gFlqXMXVLM4O7p\nuETwuFyMKMjkP1+v2616V6xYwWGHHcZFF13EH/7whzhJ2zEQkTQReVhEHhCRMxocnyIizzivI5Mp\nowEikQgzZsygsLCwTZMRTwJlMuwB3H///dxwww289957jBgxItnitBoR6Q5kAmtVNdJSOVUtAc5v\n5vhC4MzESdg1Cccs7FW7xkRju76CvGTJEiZPnszVV1/N+edv91UlHBG5GBijqueKyLWqekOcm/gp\n8JyqviwiTwOPO8cnAJcAMeAPwFtxbtfQSsLhMKeddhp1dXW8/PLLtGXlyyhhwy4za9Ysbr/9dubO\nncvgwYOTLU6rEJHzgEOAOqAc6C0iFcCtqroqqcLtAfg9Llwi2IpYALBUCfjcu1TfwoULOfLII7np\nppuYOXNm3ORsI4OBQud9s/YBu0kf4DvnfazB8aeAZ7Fv5K+aO1FEflX/Wb9+/RIgmiEYDDJ9+nQ8\nHg8vvPACfr+/TefHXQmLyEjsmcMA7LipUeBNYI6qRuPdniE53HLLLdx///3MmzeP/v37J1uctvCF\nqs5ueEBE0oBeSZJnj0JEOPuQgdz7/nIKsgLEVCmqCnLt1FGNyqkqIrLDuubfcw8f3fZ3/nPggfRb\nuIjg8hWkDEnKYFCBgOOqloh+tA5bEX9L4y3Eq4DDnPdzgNO3E0z1fuB+gP33398YrMWZ2tpapk2b\nRk5ODo899hher7fNdSRiJnwkcH3DeKkicghwKPB+AtoztCOqyg033MATTzzBvHnz6N27d7JFahOq\n+rWI/EZV7wEQkd+p6h3AsiSL1mUpqQ5x79wVlNVGSPO7+fVhg7nr9H15Z/Fm/F4XE0fk4/PYuqXm\n448pe/oZNBJBvF6ypp1AxoQJ29X55VNP8flttzHm6GMYMnIkallsufVW+t41C/G1e4C0v2O7qs3A\nVozx5nngbhE5FnhZRB5V1RnO8QedMnuMRWRHobq6mqlTp9KvXz8eeughPJ5dU6dxV8KqeqeI5AOI\nyGHAOlX9X7zbMewewWiQt9a8RWWokqMGHEX31O47PUdVufrqq3n55ZeZN28e+fn57SBpfBGRW4Ef\nicgA7GW8IcAdyZSpKxOKxrjiufkM7ZFBn5wAkZjFH1/4jjtP24cT9mk8gIsWF1PyyKOkjB6FIERi\nFqvmPE7KF9+QVrIZEDImT2ZBIIW5V17JhKlTGTbctkMQAU9uNtUffkjGpEntfZl7qeqVthxyDNuW\npuOCqtYADU2+H3eOPwE8Ec+2DK2joqKCY445htGjRzN79mxcrl23cU7UnvAlIvIAcDKQhT1CbBYR\n6QfMAkqBpcBabIMDP3ZGJoB7gDAwV1UfF5HTG5ZxOqmhlWyo2sCNn91IQVoBPrePGz69gWMHHstR\nA49q8RxV5dJLL2Xu3Lm8//775OXltaPEceVu7GW9/2Hvr21Orjhdk0jM4u73ljFvaTFlNWEAxvTO\nRoBw1OK3T3zNRZOGEozEGNM7m6yAl/IXX8TXvz+CsKKomkUbKskqDcL7XxAeMJgJI3qw7JFH+Od7\n7/L7Y45hwNChdmOFn0LFely1YVyffQejC6BgdHte7mHAK877Q4HX27NxQ/tSWlrKUUcdxUEHHcSd\nd965WwoYEqeEC7AV7x3AeTspOxbb8u8xx/JvP1U9WUSmYlsFwvaWgSc2KfNo81UbmmP2d7MZmTsS\nt8s2hsnMzeS1Va8xqf8kPK7tu4RlWVx44YV89dVXvPvuu+Tk5LS3yHFDVdeIyOHAJux9tCXA9UkU\nqUty06vf2xulXheudB8bKoIUlm6gqDpMXrqPkuowv3n8a3rnBKioiZDiczF+XZCxZcXsNzLAog0V\ndE/zEaitpC4nD3W7ePebFSyYN4/LJ0xkxIUXUjrnYfyZEagpRtN64LI2knrAeHj3z3DGM+15ud1F\nZBL23nDP9mzY0L4UFRVxxBFHMHnyZG699dad2i20hkT5CT8ELFXV5cAnOyn7KXCOiLyHva9Rbzyw\nBtsYoQ/blnfqLQOblmlEw0hFRUVFu34VXZS6aN1WBVyPz+2juG57Z/5YLMa5557LggULePvttzu1\nAm7AaGAi8AiQlmRZuhzBSIz15XVkBbwM6ZFOWW2YjeV1rCyuxed2sba0DkuVAd3SWLWlhoDPTV3E\nQg48mLIQvLNoE14sKn9YQqSykqrqINWlJSxavoZTTz2VHnl5pI4bR/qECURXfIsVCuGOFJNzxAGI\nxwuBdu+jvwWGAcOxw5kauiAbN27k8MMP57jjjoubAobEzYQFOMAxyFJsA4KWOAu4TlU/cILkW87x\nfthWgdC8ZWDTMlsxFoE7xuvyYqmFS7bdznAsTG5KbqNy0WiUmTNnsmHDBt544w3S09PbW9REkYW9\nQnMqMDXJsnQ5wjHLcUOC3DQ/0ahFfbK2UDSGZVnUhpX1ZbVELAuXS0j3eVhdbRGYPBXvK8/Ta90y\nsiO1uKMRBq1fQnFRBr5BQ1lf7WJeLJulj31F/25DufyUfXBlFCAN+jJqNSNVQumH3af8wO+Av7S3\nAIbEsm7dOiZOnMiZZ57JNddcE9e6E6WEzwN+D7QYBKEBbwDXO/u8q4GvReReIIAdIB8aWAY6/7/Y\nTBlDK5kxcgZ3fXsXg7IG4XP5WFO5hoN6HoTPvc2qNBwOc8YZZ1BdXc2rr77aJufzTsClwEhVDYrI\nX5MtTFcjM8WL1y1ELQuPy0V6wEtOmp/VJTVYCl6PG0uVmnCMunAMn1uoCcXICnhx5/ai0uVnSCyE\nJxIiPRrCY0XpW1dCZHEZHx50HPz0JIb5fFTWRZhTeyjnhF+DvOF247EwhNvdROQS4HaaPO9EpBtw\nNvAj7MlIGfCkqr7T3gIadp3Vq1czceJELrjgAi699NK4158oJfwZkAdUOv+XtFTQiTw0vcnhphZ/\njYLBGqvA3WNEtxFcf/D1PLPkGaoj1cwcM5Mxeduy8YVCIU4++WREhBdffLHNzuedgH9gG/p9ge1S\nYgZycebqY0dy4yvfE1OlNhwl1edmREE6C9dXYgGhiIULSPG62FQRpK+1gZFlJZQU92C/lV+SHaoi\nNRwEIOZyoyLUeQO4fX4sxwUpM+Dls/JBnDN2Gnz/EmgM/Jkw9R/tfbkLnedYUy7BDqjxd1W1RCQD\nOEFEjlLVN9tXRMOusGzZMiZPnswVV1zBBRck5jGRKCW8N/YSTQh7BPjHBLVj2EW6p3bngn2271R1\ndXWceOKJZGZm8vjjj++S83knoAo7WhZAMJmCdEaqghGKq8P0zErhy1WliAjjB+XicW9bEi7ICnD3\nGftSUh2irDbM/736PauLq/G4XUSiMfweUFcNfsKcXfUaA2KVeFfUkr6xiJqYB2IgqohLELUIu7xE\n3F4ClaU0nOcqio4+ARkzrf1vxDYmOMZ+IUBV9RTsN1c7gWD2Bb4EMlX1seSJaWgL33//PUcccQTX\nXXcd5557bsLaSZQS3qCqVyeobkOCqK6u5vjjj6dXr17MmTNnl53POwHFwE9E5Da22SAYdoKq8o+3\nl7J0cxWhqMV36yoY3jODnpkB/vXRKv40dRQD8hrbuXVL99Mt3c8dp+7D8Xd9iEsgPwdqXF8wqvB7\n9lu1iSG1m8iIpFCRnoUWKZmRIJWSYhu+qKIi1Hr84PNS1q2A+k2TmlCUgqxA3AxkdoPTsLc3vhCR\npoaid2CvunyJPRkxqy6dgAULFnD00Udz8803M2NGix62cSFRT9kpIrIfznJ0/cjQ0DFQVW75/Bbm\nrZuHx+Vh3/x9+fWIX3PS8ScxYsQIZs+ejdu9a7F8OwOqepOIvGi/1e+TLU9n4fXvNrGhoo7hBZm8\n+t0G3C74ek0ZBVl1ZKV4+cfbS7nztOazaGUFvIzrl8OqomrWhD9kr8LvOW5+IamxICmVLjzBMLmu\nYtwoopDlClKT4kMiUOcLUBbIYqA3QvDgw1i+oRIEemT4ufrYke18F5plR9sbZtWlk/HVV19x7LHH\nMmvWLE45JfGqKyFKWFW7Vj67Lsbv5/6eJaVLyE/NJ6YxPl73Mc+/8TyH7n0od9111247n3d0RORy\n4EDAEpEvVfXWZMvUGfhweRG9swOAsrmiDrzlSOoWytWFLzaU+evCOzw/J+BjucZQdznj1hRBDDIr\nIqTW2e4UYjkm1AKiFgGJUpSRwSv9D8c9fARfjN2Lm6aPI+aUc7uSPgOuZ0eK1qy6dCI++eQTTjjh\nBO6//36mTWufLY5EJHC4GTuzx3xVjYhIHnA8UKiqb8e7PUPbqAhVsLB4IQFPgDWVa3DhYsvGLbgG\nuPAd7uORxY9w5ugzG7kvdUF6qOrJAM7D0dAK/B43UUvxuISQdylpWWtxKQR8bsqiG4lFWhh7h2vh\niwc5Zv13FJaPYWNmAAsX/nCYzKCd06VenQrY7kwC6oI0b5iBBZlYhxzE+vJaoEMp33paVLTOqsvz\ngJhVl47NBx98wPTp05kzZw5Tpkxpt3YT8aT9K3Aw8LjT+a7DVshGAXcANtZspCpURVW4ChcuNpdt\nRrKEgD9Aj7QeFFYXMuubWckWM2E4rm49RWSGiMzAju5maAVnHNSPJZuqqAxV4U9bS8yytzb8Hjcu\ndxR35vztT4pF4YVfQcU6ymIBfpm/jONCa5jfszuZoSgu3aaAYVsUHlHsveCcLPzFRfz3m3V8t76S\nS5+Zz9qSjhWlVlVvwvYP/peqXt7wMxF5EvgT8FdnC6TNiEiaiDwsIg+IyBkNjvcUkVkicpeI/Hh3\nrmFP55133uGkk07iySefbFcFDIlJ4FAO3OW8DB0Mj8uDiKCWsnnLZtyZblJ8KYStMJn/z96Zh0dV\nnX/8c+6dfclkX0kIEPZVFkXcZVFxA5dal7pvpf6qtlq1lqp1rVpRW6tWa11bFRfUCkVBFktFZAfZ\nIYSEhOzbJLPee35/3IEkmISAhAlhPs/jk5m555773vEw7z3vec/3tcQhhGBz1WZCegiz0i0zo5OB\nL2l6AI09HHaQvqlubjsjj78s+S8Wqw9FOjGrCiFNJ9FpQTU1tmhfWF7L5k+eItfnJduuEx8oQprT\nOdlm4SvnidSYdxOvVGDSZbPqwoACUhEErCa2iEyKAwK3VeX0gWlICQ9+uoGXrx6FWe0a0ZqIo5WA\nQwiBlHJfHFNKeXmzdr86xEtcxA+lewHuwgiFu2hFtChGx5g9ezbXXnstH330EaeccsoRv363TX+N\n0Tq763eT68hl3Z51WOOs6KpOUAuS5cral2UqhCCkdU8nLKV8I9o2HM2MyEngxayzuX3BXCoqndT7\nw4AkQZQyyeqF7z+BAeey4OuFVC19hyzKsAZ3UlO5hp62RLyVu7GYUxjszSA/IZ44rRZ3MIQlJFF1\nCJkVhCrwmWwEHckEdTuOrHTOHBSRZBaQ7LaycFMZEwd3jSBGe45WCLG3ULIJaFk0ueP0ANZFXmvN\nPh+MIZNZBjwF3Lj/iUKIm4GbAXJycg7x8t2XWbNmccstt/Dpp58yduzYqNjQGWvCo6WUyw93vzEO\nD26vm9XLVtO7T2/iUuKoDdTiMruIt8YDoOkaZsWMw+yIsqUxuipm1cxVA6/gn5v+SR+LG4pWMNzX\nwKXuM2HbPPR1H1BXoJKZnIRsCGP1b0BHobGxEb85gfhdWxiyw05ioJIit4NE3YdF0zCHdbzxLtCd\nKCEXPqFS77YgRwyCZnNlVQFfqOvkOB3A0V4a+evn0KODRbQu3VuEocLlBWytnRiT8G2b9957j9tv\nv53Zs2czatSoqNnRGTPh6zH2xMXoYmzatImLp11M9tXZaG6NhlADAS1ASAuR5c4ivzafsAzzm9G/\nibapnYoQIhMYKaX8txBihJRydbRtOtoYlzWOMelj2LBtDj2L8onvdYZxwOwgGNaJD+eDSEYEvFRV\nO7CVByGsYbVU4mswMcRWSKUeQgmY2GN34UuAkFVD05LxNfbEqmok5NYxwFNPbcFMSgtTsY7+GYrJ\nTGmtnwmDUqP7BbSkVUcbcc4zaYq0H6oT/Ihm0r1CiLeklD/DkMp8MtLvi4fY9zHJm2++yb333ssX\nX3zBsGHDompLZzjhE4UQTzb/QErZvX/VjwLWr1/PpLMmcfxjxzNp+CS8QS+76nfhMruwqTamHTeN\noBakh/sHRam6I49gzCD+jfHQ+Mu2GkYUj1rUs458Phm4NtLsVSnlF51pcFfErJoZ7q2BuJZjxuJK\nwKOtxIektiyIq9hMI2ZCmoZaq6FrCrrHCvEq9f4QYd3E/KwsdqUnYQqOJQ0PlwTeoXdWOqqiYK7z\nY/KWs2nFJ9Tkns3Pz8jDYekaK2kRR/sBTVP15o52r3Nu/tlBF3eI1EtvLt37TuTzDTSNwRgd5JVX\nXuGhhx5i/vz5DBwY/X3mnTGSvwde6IR+Y+yPlFC5DUw2tLhM/r3j36wpX0OSLYkrBl5Bgs0o6bZy\n5UomT56T2jbrAAAgAElEQVTMH2b8gS2ZWwBwWVwMSjIiZ/m1+aQ6utTMorOpAWo72LatpJgzMLSB\nNeAeoNs64fW7a5m1ajfxDjNXnNCTRGdToQ+yRkL+IrAnQrAeX/5SaqorsesCWfg1oRo7jboZG0HC\nUkUROuGAioaZJMWOZm1ggzmNrYkDSTX1J9mZRUnJDhpddazz+8hSE0iNSyDVbSW5roDJV47sCgpZ\nzbkUWuSVSSKOVkr5kBDCBBwHODj0mXCMw8Rf/vIXnnrqKRYuXEheXl60zQE6xwmXSikLOqHfGM2p\n3A5f/h5MNtDDvOArYGliGn49RG2glk+3f8rzZz5PzdYaLrjgAl566SUmXzCZexbd06IbKSUmpWvM\nKo4gW4CrhRBDgGUHaNtWUsy7NIUab27txO6QFPPG/3ayLL+KXskOahpD/Pr91fx28kD6prmNBulD\nIFAPq98mFArS6A9Ro6RSmzyU+MrVpIgCqvBQEbYRklYcNg3VH0IoCgKBNeSk3HkccYxBhFXKtTr0\ntBWsDvtIcKnkh/fQv7GOgbZsFNXa1RwwGJXcdkopX2rj+AtAJoaa1hhg8ZEyLEZLnn76aV588UUW\nLVpEbm5utM3ZR2fk+E8XQgwFEEL0F0K82QnXiDH/YUgdCEl5VLpS2KzVEawpoLShFF3q1AXquOmz\na7nugYt57bXXmDp1KlbVyrCUYeyq34WUkpAWYlP1Ji7rf1m07+aIEvnBnIiRTfrMAZrvTYqBlv9e\n7gNOA04F7m3jOn+TUo6WUo5OSUn5cUZHAX9I43/bK8hLdaEqCk6riSFZHl75ekezRpGAQs44qgNQ\noyQR1BX0ii2EFDu+9Hikt4HE6gaSKhuwlPtRPAJzvMBvCVPuiCd+4nguHZ3NecMySUrZRlqCC02x\nEw4FUU0Wtpu87K4sJW/SD5J/uwLpwP8JId4XQswUQry/3/F6YJ2U8g/EHHDUeOSRR/jb3/7W5Rww\ndM5M+A1ghRDiUWAtRu3WNhFGNe6HgTiMhK4QRqjPCvw80qzFmlyk9vC+NpE1k2MHfy3IMAhD37mo\noYRK1USd5sNhS0WEAyjBBpw+P5MvTOZssRD0c0BRuH7o9XxT/A3zCuZhU23cNeouenp6Rvd+jjBC\niIeA54HXMcpsXttO87aSYj4CXo20+U/nWRs9Suv8WE0tNcQVIfCHm2Umb54DcZngSqNU2YASDuCh\njkQ9RAA7NdKEM07FaXag+YP43TnUOgKIOg33ZVO4YPwN/O+jdWRJiRCCoO7FH9A5+fhJ7Nq8kkBD\nOXWqYMyES+k7cPgR/gYOjJTyWiHESVLKJW00WQUghPgEKD5ylsUAI9I3ffp0Pv74YxYtWkRGRka0\nTfoBneGEy6SUTwghXpdS/q4D7S/EmGlUYsw6bpVSXiqEOA9jPQ5+uCY3db82b3XCfXRdTDbw18OG\nT0APk64HUFyCsElgAjRfDQ2BEKlmK0GLiw01W+mz7GUcY41nmhMzT+TEzBOjew/RJR44B+Ph79L2\nGraTFNPta1pneOz4Q1qLz8K6jsPSzDFbPaAb0pNVmps+lGIniBQSi+anXqTg8qRjy0kHXxWugZNI\nBvybN5F17k0oVit3nTWAV/+7g5CmIwSMyXVgs5joN/R4ADZXbWbCsAlH6rYPhb5CiGlE1oWllFc0\nO7ZbSrmQplyCGEcIKSV333038+bNY+HChXTVaFRnOOFJkZDMiL2hmQNUUeoP/E9K+bIQ4gOatFcL\ngKGR1/uvyclW2uyjO6zFtYtQoXon2D1gSSRF6pzQsJX349wEGmsgFCLOakJHsE36mGPSqN3yDlNz\nxzA6fXS0re8KFABjpJRvCyHOiLYxXYGVpSt5b/N7hPUwFtXC9UOvp298X84aks7c9XvISXTgD2ns\nrvHx8JQhTSf2nQgrXwdnKj2UCsyaji6hUbfgECGSlAYUbOAth8wRTedJEJFa1YMy43jmJ8axxtAA\n7v/v/ez2NuIwOSjzlXFy1sm4Le4j+G0cNP2llFe2cWxA5PdoJ/COlPL7I2fWsYuu6/zyl79k2bJl\nfPXVVyQmJkbbpDbpDNnKvgd5ShFGqBkMJ7s38yKHJim21jaq79+muQ3de4N6/mLoOQ68pVC/BwU4\nN64/qwPFLNcb6Gs2Y5YS1ewg1+zGZXLitAre3vg2o9JGdcXkliPN68DlQohfAK9E2ZaoU+It4bX1\nrzEwcSBCCHSpM2P5DJ4+7WkuHtmDsb2T+HT1bhKdFu4/dxD25jNh1QTnzoBFfyRDllIjnOz2mVGs\nbhSHis1fji3NQV1VTywJvQBJuLgY+3HHIVqp1uUwO/jT6X/i25JvKfYWc3OPm0lxdM0ZTDPMQogx\nQAPs2zpE5PVLQoiXMbbCLQS6/M0c7Wiaxq233sqGDRv48ssv8Xg80TapXbpCWuxHwJ+FEKdgJC5U\nCyFexMg63FuXc9+aXOT9rFbadGsWFi5kTv4cdKmTHGjkDj2EtceYfce9a1Ziq6mm96AR2BoKiAsF\n6OVOwW12grcMkXsKqualJlCzb+vSMcwMmtZzZwDXRNGWqDNzy0z6xPfZ93CmCIVkezIP/u9B+sb3\n5fw+5/Hz05tt59jzvZGdnzsOnMkQlwH9z8G84d+Ea+tRbRnY7Q58QJrDiuWyRwnkBymd9W9cFoWE\n00/F045IviKUo225pB7Ye0P7tigBCCGeArKAr4CDnaDEOEjC4TDXX389hYWFzJ07F5fLFW2TDkjU\nnbCUshG4Yb+P919ra74md0ysxzVn+Z7lfLb9M/rE9wHAH2rkxcL3uSMhF4TKylUrWVD1P5JGTmJo\n7zGghZi/8V0CvlLG6CnGrNmdRriqBpel6w/KI0C9lPJrACHEJdE2JtoEtABW1brvfZW/im+Lv6Fn\nwI+3YBmvrnyV05y5nHL+yzD3txAOgMWJvuwfaJoFs6ihToljzfYqRvaw08NUjM+WgtlkxeLM482C\nRJYU1pEw/Ex6VX3N0KIvGFfWB1L7R/GuDw/NVLHgh2IdAH+WUu46slYdm4RCIa688kpqa2v5/PPP\ncTiODunddp2wEMIcqQncB6iMVEg6IJH9l+cBFoBIen6MQ+Sz7Z/R29N733ub2UF+9kh2l2/Avz2f\n4s1b2Hn2WAb2jsyMVTP9ck5naclSjut7FqpQKazfxci0kd2yKMMhsFYI8SHGD+a8aBsTbS7scyEv\nrX2JvPg8CiobWLhrBbZwALdmR0uIJ9Gisr6+hHFvX4qaMQTpzqR29lxCe0pB01CdGrXOGhLyjmO9\no4INegMVNghYVEJJmewpqGWYyGdQ/TxsHhfVDUHe/fg6lmfkosel0z+hPzcNu+lorWHdplgHQMwB\nHxkCgQCXXXYZmqbxySefYLO1KqXdJTnQTPgxIcSzGDJ/GnBVB/v9Fcb+y9CPsC1GBE1qP1jHNdkS\n+EtpAu+9upT585czf2dTDWBN13BanPSM64k35CWkhzi/9/mclBUrOQpGzoAQYglGofX10bYn2vRL\n7McZ2Wfw6da5bCpS0aSNPM2BxWJlV3UjfVNdhExWApWFOByn4V26kmBJDcKeRNhbRfWeALb0eP7r\n2sM3TieaKY4GzceovpPZVOxFZRNZFSuwueMAKBR72KrE07u+AnJOptxXzitrX+GW4bdE+Zs4eKSU\nD3X2NdqSTo0cGwrMB3pLKb2dbUtXxOfzcdFFF+FwOHj//fexWCwHPqkLcSAnHIexhehxOu6AAdbH\nftwOH8NThrOhcsO+BBUpJSs3rCT/hXwWLVpEdnY2fWv6UuWvosJXwfaa7UgkUkrG2cZx8/BWBZ2O\nWSIPlpWR1zdLKdvUjj5WOL/P+azb1JueGTrfVnyO0/8/kDpOvR5feQUumxWbYoGQn2BJHUKRhBpr\n0QMNxNlVtoVM+IRK2GTBIlQ89mRWla8hwzmarbUF2PEBbqTUKVQbsZltII18zERbIluqt0T3C+gE\nhBDX0IZUpZTyYESMWpVOFUKYMQRn5vxYW49WGhoauOCCC0hLS+PNN9/EZIr6CutBcyCLF2I8Ya0V\nQmw9iH7PEEKcDgQAeYAtSjFaIayHeWH1C+TX5oOEPY17qPRXYlbMrPt+HUUzi1i0sGnz+a0jbmXG\nihmsLV+L2+Im3hrPmPQxbK/dztLipYzNjE6tzC6KJqV8GGD/YiPHApXeAFvLvOSlukh2Na0Fa1Jg\nM5sZmnAyW3atZWhwIyZFwRYKcnZjDUpCHnLrXAKNNYR9YVSpgWpBWCw0mIMIPQRaCISOSOhFOFhD\n31Q364oKqcFNHBJdSoJ6kB5uDwT8+64tj2JZ5chMdaCUcrkQIktKuTtyaOdhukRb0ql3YYjOTG/H\ntm67XbOuro5zzz2XvLw8Xn31VVRVPfBJXZADOeE5wEQhxNUYDrVDSCnP/1FWHeOE9TD3LL6HKl8V\ng5IH4TQ7yXRl4tf8VHxWwZ7P97Dwy4WkpjYVXVCEQrY7mwk5E4i3xe/7PNudzfzC+TEn3BKTEOIB\njFmKPdrGHEn+8tVW1u+uw2FVaQxqDM5w83/j+wFwyagePDd/K/3TUumvZeEMlOHQyjnJLvCoTijb\nRIGexJbUNFRfOULq2HVBim4mnODHopoJ2DzgSsUsBALBrvp87h8/he3flpFe9B4Wm5MRDo0ybwn0\nGQ9AXbDuaK/e9SxGqHg58FsiOzaklIsOU/9t1RMeAaQBxwO3AH/a/8Tuul2zpqaGs88+m+OOO44X\nXngBpZXtbkcLB3LCMyJtvgQmAB0KoQgh7gWmYuybS5ZSRrdg41FEjb+G6f+bzpaqLSTaEllctJhe\nnl70T+jPnP/Noe6LOhYuWNjq5vMkexLfV7bUAghqQdzmLi10cMSRUt4uhBgceX3MiCes213DppJ6\n+qc3jYfNpV7W7ipnWIqZgRnxnD04nf+s38NZgUrM1oEMyz4RT9VSCDYgwwEaQn6+TWxEibeSXKlj\nCmmEbJLjh4xkZaiEcX0ms6J0BYX1hfT29GZK3hRO6XEK9AYCk2DjZwyVYZ7ybqY6WAlVlWS5svi/\nEf8XvS/mx1OPUZkLjJrCh5tWpVOllJcBCCFeB17uhOt2SSorK5k4cSKnnnoqM2bMOOp1Dw7khMsw\nklfeFEIczKNqOvAfKeUDQoh2taNjtOSVta/QK64XBbUF2Ew2bCYb+TX5bPlmC9Whar6a9xWJ8a2r\nv5zW4zRmbZ1FgjUBi2pBkxrba7bzwLgHjvBddG0ia3XnAFIIIfeTGey2zFm3h5yk5ts2JIMavkV+\n/BikGwGBKafexYUjRqJ/CGpSBvjrDK1yZzKhinwKpI+Q1cZuFXZ4FGy6QJEexlpsXGnuw/t6kFFp\no7ik7yXkePYLf1rdMOIKXMBDGBEfgUBVjs4wYjMqgFOFEE/TpPgHtBuq7jBtSac2O37twZt8dFJa\nWsqECRM499xzefzxx496BwwHdsKLAE0IMQujTnBHqQAUIcTvgK6nut6FqQ5W08PVA4/VQ0ALYFEs\n7Cneg+bTOH/i+W06YACTYuKhkx7i5TUvU9pYilkxc8foO461WsEd4VQp5U+jbcSRJjvRzobiOhKd\nxjqwqWwd/RtXoiT1gdR4oz71V48ifvpP1LHTYNEfQbGAUKgv2c53+Y2Y+rnYaQWfUBFI6oREV7wU\nNe7hvIv+yb3u9A7b041KaD6DMVtVgG37HWs1VB3j4Nm9ezfjx4/n8ssv5/e//323cMDQjhMWQlgx\nErMA/stBFKSWUj4SydwbCjz3Yww8FiiqbiSkSXKTHFhVK5qucXz68XxT/A1r89cSsoXIGpGFalYp\n8ZaQ4Wq7EkiiLZH7TrjvCFp/dBEJQ3uFECcCtdBSZrA7M/W4Hny5YQUOiwmbWcVTtRavbuXUdGPr\nEEKAPREKlxlqWFNehKUvUZO/nPeWl3PBJdewzLuMkPQSkhb8ioOwxYmw+akccg0chAPuZmwAfi2l\n/FgI8Qfg982OdXao+phg165dnHnmmdx4443ce2+rlUOPWtp7FP0thjYzNDng6zvSqRDiYaCXlPIq\nIcQM4M5DN7F7smBTGe99V8jaohpURTAgw40iBFPGTuDP6x7FoTooWleElqgxPH0443qMI6SHeHjp\nwzx92tM4zEeHGkwX5BKgCpgUed9CXKE7YzOrPH3pCF5etJ3iGh9jxB5GDeqBWW2W1CI1sETGljOZ\nNwp7ULS4gl9cfArxSW5sISc5CL53JaEATlUlN64fNtPRI47QCbwD9IxE/vafnrUZqo7RMbZv386E\nCRO4/fbbueOOO6JtzmGnTScspdy3kCiEGMYPwyzt4Qb2bvyLCXbsx5rCGt77rpDimkYyPDb8WgPf\nVXxBUnwtv128jlN69mXJqiWEkkPkpuYyrsc4AMyKmR7uHszaNosrBh4Ty5iHHSnlQ0KIcVLK/wFE\nZsTHDIlOC/dNHghA2ZYphFa8gT09Ut1IC0DYBxnDQQux5pmLSPt+CRdddBLuuHiwxtF/8BXMrlvD\n5MQBSGk8m2+o3MA5vc+J1i11BeqllM8KISZhZCLv2zIkpXxMCPERoBwrEZfDyebNm5kwYQL3338/\nt956a7TN6RTaC0fPAJKAaqAEGELHBTskkBGp93vMxqja4v3lhfRNc7GtvB6PCUpZgsPho4EibKqd\nBZsXYjNbyUjKICzDhPTQPrlJh8lBha8iyndw1DMV+F/k9fnAN1G05YhTG6jlsW8fI6yHUZwqKbs+\n41ZrLnHOVDjvWRCC7566iHUrvmXKT2/AnZgIoUaoLSL39Hu4YNd8Zu+YTUgPYVEs3Dj0RuIscdG+\nraghpXxSCJEMbAJOa35MCPHQ3gmNEOJ+KeWj0bDxaGT9+vWcddZZPPLII1x33XUHPuEopb1wtFlK\nebUQ4m9SyieEEM+303Z/HgCuALKJJSL8AE2XSKlTp2+l3FeEcOxAyESQQRobyzFbFBJSE3CYHeyq\n30VZQxlZ7iwACusL+fnwn0f5Do560iJ66BLIjLYxR5pnVzxLujMdqzBDWCdkdvNHexyPnv4UAE89\n9RSDNy3loiuvJ94T2XNudhiVk8JBxueMZ3zOeKSU3SY55scQqZQ0FtiKUSnplGaHm5csSzqSdh3N\nrFq1inPOOYdnnnmGK67o3lG/9pxwTyHEZCA18je3Ix0KIabtfRn5+zMM3dMYEc7on8rT37yOO7GS\nkL8Ok8VLIFRHuD6I6vKR7ElGIom3xVPpr2R56XI0NPxhP6PTRpOXkHfgi8Roj9/R9HD4YBTtOGJI\nKXlrw1usrVjLqtJVZNiSOKGmDNXixCxUnCUr0TPm8Oj7y3nnnXdYed/pODzxP+yomdONOeB9qBia\nztNb2ZK5SwjxBsZ68MYjb9rRx3fffcd5553HX//6Vy6++OJom9PptOeEP8AoQP1x5O+HHezzSqAA\n+BdG9ukB/6VG9tItwvhB7A/0AszArUAG8BSGXNs/pJQLIgN9Xxu5d3HqKCCshymWc/BaF0AoBZua\nhDcUQPMHsdh8WO0qDaEGMpwZlDeWMyZtDKqict3g68h2Z8cSsn4kQogTpZTfAPc0+8wNZEkpN0XP\nss7lnxv/yY7aHfT29GZr1Vao281WPcSABh9InXihsPHte5n5cZiFCxfi2PQ6NJSDI9noIFAPcZmg\nNlXhWlW6ii8KviDdmc5FfS/CbTlmRWG2A6oQ4u/sp8AmpXw68vu2d79vjHZYsmQJU6dO5e9//zvn\nn39sCC+2l5j1xt7XQojrpJT/6EiHUsqThBAjMMLRDqAjUjj3AO9j7LMbKaW8UghxG3AycDrwBMY+\n5bcj1W/2b/N1R2zrCtz+5XTW7a5GE2ZQa1BMpbBtD3E9nLgSXGhoZDoyibfF08vTC7Nipj5YT//E\no7/2ahehXyRaI4E6IBnIB/7SWuO2KtgIITKA+zAeMt+VUi45ArYfMqvLV9PL0wuANGca9TWFNIb9\n6GocSBi1pxhTvY+Fn31FYno6pN4Ni56Esg2ABE8OTHpkX38vr3mZ7bXbyXZnU+It4Z7F9/DAiQ+Q\n5kyL0h1GDynlCwBCiESatiPt1SXPbXorYjr67bBgwQJ+8pOf8M477zBp0qQDn9BN6Ohu+Q7LTgoh\nLMAAwAnsPNAsVQgxEWOfnQ3wAOWRQwUYeqk9gEIppR4JfyW10mb/PrukaPm6sk2sK6rHYtapDddj\nUnQa/EFs2W4uHH4+brObJHsSy/YsI9udjS/sY0ftDn4/9vcH7jxGh4g8XL4hhHBhZPGXSinb2zrS\nagUbDPH8esCFoe3bpWn+z3B4ynAK85cgdI3cYAPJFXWYK4LkZSegpmUDUB2s46PkVBo8Ti7ocwG5\nntx959f4a9hYtZF+CYbmdJw1jv6m/vxj/T+494TutYfzQESics8C/8R4IMsnEmWRUv4m0sYdeV8f\nJTO7PHPnzuWqq65i5syZnH766dE254hyQCcc+bH6gTB4OxRgiHz8N3L+NClle2vCp2M47EEYIefq\nyOc5wFoi4uVCiLrI55UYs5fmbVoQLdHykKazaHMZgeISRjfsJn7YYKy9eu07/p8t61BViV+U41I9\nVPtKMZsUwqpOpa+Su8bchVkxMzZjLHN2ziHbnc2do+6MhaAPI5H9mrnAbCnlax04pa0KNoOB2zGk\nXZ/CKCm3/7W6zMNgz7ie1AXriLPEIYQg05ZA35pS+uxpoKExyLB+uUYVmlADmypLeX7V8+TG5WJW\nzTy78lkm5EzgvD7nAVBQX4Dd1LLuhVk1U+4rb+3S3Z1cKaUmhNgQ2f42o/nByN7hnpHXhVLKg96T\n3k405l6MZblk4HYpZZd/GGyNzz77jBtuuIFZs2Zx0knHXs3zdp2wEGI6TQNoVwcH0EHJAUop74/0\nfy3GxvZ+QojnACvGwNuOEY4OA69KKYNCiJX7tYk6u6sbmf7J9+RuXkHP3Vv4TzjM8M//Q3rPTNJ+\nayhYJdk81ASq8NgFpeWVuN2JWCwWhLQxIHHAvm1IveJ7MW3EtPYuF+PQ8UgpLxFC/BXoiBNuq4JN\nEcYDoxcjivMDovUw6Av7eGvDWxR5i3Cb3Vwz+Bqm9p3KcyueY3f9bjSpoRMksVqQGfAxbEAeaupA\nCHnBnckbS+5ncNLgfYlX/RL68eWuLzmn1zmoikpvT28aw40trukP+/FYPUfqFrsSAoz955H3+1eU\nT5RS3gT7sqgPhVajMVLKJyL9TgXOAN46xP6jxocffsi0adP4/PPPGTNmTLTNiQoHmgknSClvhI4P\noEMt3yWlfL2NQ8XA1fu1feZQrtGZPDd/K/3MAXqXbsOW4EaWl1GUX0pcdRkF/57JM87/YlNtaKKa\nMl89jjgnVpuVQNDE2J59SHXG9J2PEH0ja8ID9mbyHyBS02oFGwy94Ccx1pZf7GyjO4KUkkWFi/jj\nd3/EY/UwKm0UilC48vMrGZA4AJfFRYW3Al3Xqdxpxu6Aspwc0t29SQvUwvE3g6IQ0kM/yHw2CRO1\nwVoSbYm4LW7GpI1hRekKslxZNIYbKWss4+GTHo7SnUeVSiHELcB3GCUFK/c7Hh8pGAKHvkWprWjM\n3kjlT4hEXPanK0Vj9uef//wnv/71r5k7dy4jRoyItjlR40BOeO8AkkDblQNi0BAIk5q/A5vdirpl\nA3F6CB8QDIXZ8dc/kvfwz6iurqVxeSHuMSnE614Ge32k2xT8xcWcPfaX0b6FY4W9SnAdKkjSVgWb\niPrRtYfVsh/JC6tfYE3ZGhwmB2bFzMLChSTbkvFYPXiDXgYkDmB9+XqKiotw+pz0mjwNtb6I570l\nPHrBq0aVI8CiWtCljiKaJv6a1PBYmma6Vw26ipOzTuaLgi/on9ifs3LPwqLuPwns/kQqxU0GxgPf\nSylf2q/JLTRJpN5yiJdpNRojhIgDXgB+09Z6c1etJ/zaa68xffp05s2bx+DBg6NtTlRpTzHrAQyl\nrNsiH33RkQ4jA7IFUsrZh2TdUYRJVZBJSYTWVeDRgtSEJCFNUqRplMbZqFi2ine++S8TJ06kt7mA\nMm8Nw13pJClmzogfhGX+H2BKl4isd2sOY6H1LkWNv4btNduxqBbMqhmTYiLZnsz2mu3kxefRGG4k\nEApQWFSIisqpk0/FZLZAYm98ir7PAQNcN+Q6ZiyfQbY7G7NiprC+kHNyz/lBycFcTy43D2t1AnZM\nEfl9a+s37jQMVTYrcDEd1N/fj7aiMa9jbNO8XwjxvpTyq0Po+4jz4osv8thjj7FgwQL69esXbXOi\nTnsz4Xcjf98HfomR8dwRUoALgaVAAEM9pts74ctG92De598iwyWIxkp04UJgolJa2ONMZNvanZxz\nzjkMGjQI1m/DY0/mZ5nNFO5qdoGug6K0fZEYMdqguKEYm8lGhjODwvpC7CY7ilCQSOpD9WQ5snjv\n3fcw9TaRlZWFy+ICjBC2WZhb9NU3vi9PnPoEn277lIZQA/ccfw+ZrmNOWOxwcSNGODhwqB20E425\n6MeZduSZMWMGzz//PIsWLaJ3797RNqdL0N4+4c1CiJ8B5wJ/7ug+SCnlG0KIgVLKJwGEEEd9xe6O\nMG7rM4zS57JnoE5xwExcuJGdphxqnUk01lVRMjiekwYOREpJjR7gLGfPlh0I0UKNKEbn0F0jNb09\nvWkINZDtzibLlUWRtwiTMOE0Oan317Nk/hKSk5Lp1a8XG6s2srZ8LQhDi/wXxzUpy26t3sq/Nv2L\noBakt6c31w+5HrNqbufKxzbtjSchhAOjkI0TI7H0mOaJJ57g1VdfZdGiRV1ufTqatDntEkIspynj\nztPaYGsHKYR4TQjxEsZezO7N+o/Qv30VUVZKqr+MxF5e4uIa6GfZTg//BrTMNOqVUylpKKGkoYST\nUkYw2tIsR6OxAlIGxpzwkSEFY3YyBEPn91DCg10Om8nGhJwJbKraRA93DwYlDcJhdvDI6Eeoer2K\nhKwERp0xCqvJSv/E/qiKilkxY1bMBMLGJG1r9VaeW/kcCdYEMl2Z7Gncwx+WHhNVHn8M7Y2nFzD0\n8zQss8AAACAASURBVB+LvG5VEKa7I6XkwQcf5I033mDx4sUxB7wf7YWj/xz5m9xOm7a4HyORoIZu\nXsha99YSfPc+fEVWfJUmGgIC1eQnpa+PfDUNnyuDIY564sNb+c3JTxonSQnfvgRF3xnv04bASd2v\nTmZXpDtHai7Iu4BxWeNYuGshqY5UBpDOv351Fn8ans6plz0COSdw58I7GZHaMhN11vZZjMsax7ub\n3qV/Qv99a7/x1niq/FXsqttFTlzsh7M12htPUsrrIsc2Ro4dc7J3Ukruu+8+Zs+ezcKFC0lLO/YU\n1Q5Eh2QrD4FnAaeU8gYhxMscelZgl6b63Xfxzv0IsTtIwG+mSpgIKQqEJOpmneAAC9LmwKSqjLcV\nNJ0oBIz9ORCrhhQlpBDiNQzxg5JoG3M4SbYnc0n/SyjfupKNT5/CqAH9Oen0SYjvXiGwc3EL5ay9\nhHUjUhrQAj9IvrIoFsoay2JOuH3aG083YKirAVwD/PZIGhZNpJTceeedfP311yxYsICkpFgRqdbo\nqGzlwaJhKGeBUcSh2+Hfto2SL+dREgqTo5lpDAgUCdIhEQICIZW6ugDhZEm9P0S+5uLlN7/jllP7\nMLJnQqwCTXTp1pGa3bt3s/CBc8npN4yTTjsDgYCkPCwla1GS41uUINSljlW1ApAXn0dxQzHx1qbq\nSXWhOoYkD4nKfRxFtDeeUoQQVgxRj2Omtrqu60ybNo3Vq1czf/584uNbqcgVA2hnTfhHEgAGRgos\nJByo8dHCsh2VTH5uMeMen89Lj/yR0pptuILl6FZBSBcoAYG9XmDxCmRYoBOgZ5KDLJfgk9Dx7Kho\n5DcfrGXaOyup9YWifTvHMs8CD0gp62hadukWFBQUcNpppzFsQG9OOe1MwwFHEBYHV6afzIaqDVT6\nKqnwVbChcgM3Db0JgCsHXUkgHGBHzQ6K6ovYWLWRi/MuxmZqVRAsRhPtjadngVeBVzDWhbs9mqZx\nww03sGHDBr788ssOO+DCqkbeWVrAtzsqW43YdFc6ayb8MMbWJIWOSQN2efLLvdz9wVri7CayXRJP\nYi3W2gYUqRG2KmgWMPkhLAWaUAjpKt6AEyVk5tOG0QRThpAkBBXeADmJDp79cgsPXHBsb1KPIt0y\nUrN9+3bGjx/Pr371K4b2LYawH5o70GADo/ueR/9+5zKvYB4W1cL4nPH7nKxZMfPQSQ9RVF9EWWMZ\ng5MH75slx2iXVseTEKKHlHIVRk31vZ8JIE1KuefImtjJaGHkN39hzdpVfF0QYI+jH2N+cRtaB8fP\nS4u2s7qwhjS3lRW7qnn72wKevnQ4VlO3Sdlok85ywr/eqzMthHiMbrAO8tbSXSgCnPU1ZO78hoG2\nnZhdIdRyDYmC06RTkmQmpAgCqkq1zUN5sB8FuddSsr2ChGbhZ5tZZU+tL4p3c8zT7SI1mzdvZsKE\nCfzud7/jlltuIVBXzPaPryfZkUaKOwtRswv6nAlWF25gat+pbfbVw92DHu4fFCeL0TZtjadrhRB5\nGIpXNTSVNXwR6F5OePbdbChtYNmOaqxhP7/OK2YVFfz24+84bVQRQgjO6XVOqzWnS+v8rN5VTf/0\nOADiHRa2l9Xz05e/oXeKi+wEBz/vW4N17dughyH7RBh5VbfZTXLYnbAQYiYwSAgxBEPusuxwX+OI\nEQ7A/IegZhcTikI0FORwfPH3ZMs9JOr1hOpMhKwqmq4gEBQ48yi32lGsdoR0kSNCfK/raLoRWvGH\nNBIcZkCixEQ5okm3itSsX7+es846i0cffZRrr72Wb0u+5a0NbxHf6zi06nxE3Vp+e/oTuNJja7ud\nRKvjSUr5iBBCAQYCccAbUspuVWpqU0kd7y9azeQd3xNsrKaX1Yrfk4FVb8Rc+DlrPQpxe9xsrd3A\nc8tnYFZUesTlclzacfx8+M9xmB18t3g1Pb/7LzaC1FtdBFQzO0UC5iQ3PWxWKrZ9w+vf/BVnigVn\nnIvxOwPEVW+HCQ8SCPn5+t8vEywrYsTES1lFOQoKZ+ScQXWDpLw+QF6qg0+2f8imqk0klG3hKl+Y\nZMUKziSY8BDYo7tefdidsJTyUiHEYCnl93CUbwH5YjooKqQOxlOwgFu9yzFJDRmAcL2CKQwmS5hG\nl516ix3dZ8JvScYjzNgbazn7tsvID5mQSIpqGsmMszGmVxIbSuq58ZReB75+jM6i20RqVq1axeTJ\nk3nmmWe4/PLL0XSNtze+zcDEgUbyVZxRl/rFXXO4O+aEO4tWx5MQ4i7gBAzn++8o2ndISGk4MZfN\nhMNiQtMlS7dXsLGkjjHeBWQVLWbZdisJ8RmYGvYQVmykKA14QmuorbZj0t0ML7Yw6OMSTq0HrwM2\n9IJ6SxXJ+StYFH6JcJyN6sY+aGo8npoSBnrr0XU4y+zD5fajL4VdDRY0IQm7G2iwVPIv927syi7m\nrv47o9f/GRMhFCTL1v2bNXkKpSlu7l84gx7alaTbc9nm/Q8nONeQHapCCwd42+bm2vgRxBcug7+O\ng+wx0Gc8jLr6wF9KJ9BZ4egbgF9FXj8KHH2VvnUNaosgfSjenbtw7NiGvS5IoNaEHlLQ/QpSA8Ur\nMTvDFJuS2RGXQa07CdWu0HPSaSRMnMBvgfvOGcDHq3azdEcldb4Qt4/vy5CsY7LsW9TpTpGaZcuW\ncf4F5/OTGT9hecZyli9cToItAatqbZF9bzfZ2VW3K4qWdl8OMJ4GRyYlLwOH7ITbqSc8BLgv0uxx\nKeX6Q73G/mworuO5+VswKQJNh0SniaJqP+t215IZzEf1zqOsqJw81Ywt/C3ES1JENXZzENWpoe/W\n6e0NcfJOiSmSg6oL6LMbhJCEVdAVE1p1GOJ2kOWTJNaFMQeNKLMS1NHrBVqpJN4RwGsXOGoAB9SE\nNALmalJ2zkSVGhZ06lRQfYLkPZKdHi/gYw9/oagxAbeoY7dXkGkKoJptCF8Nq6rncoI1G2/YT5Uv\njrz8xZgcSTDw3IP6nkpqfDw3fwvegIZJFVx+fA4n9Dq4rVid5YSbV6F3tddQCDEFQxozDvg7MBSj\nULUZuBXIwCiargH/kFIuEEL8unkb2YmpdL5131O56FusMmyEn2vNqDYNxSTRNQEaqGGNONWHdLnR\nJ0xGj7cx9bympCshBBeN7MFFI2PrbNGmu0RqlixZwtSpU7n8z5eT2jOVBJuxFFnuK2dj5UZ6e5p0\neaWUP9j/G+PwcIDxlBxRGkzbqzh4iBKprdYTBm4HfoHh/J/kMOkxaLrkT19sZnBm3L6HuU/X7KbS\nG8SkwDB9G0kldfg1E/6wRl5jCMolwSQVuyIJeM2YHRpZRRqmkBLJz5cIKVDDhrFIkApYdEgNhlAB\nRQMhASkJ6QKhgKoJlJBR/ULRjf/sfkm5LUReRSFBu45ZQlAIJJBSC5oAhI4i/MSFy9BVwR5pYmc4\nQB+zA0Ux4Q96KfIFMGGjpF5jc5WVs+Sn2A7CCYc1nftnrWNgRhxpcQogeWXxDlJcVnqntOv2WtBZ\nC5NLhBAfCSE+BJa311BKOStS9PpW4HJgpJTyNoz6mSdjzKqfwCgbd5MQwtJKm8NGfoWXu2au4bZ3\n1/DZJisVXywl7ANdClSbhlAleligmCUWl4awgEuGOS7By/U3nsojU4fy4AVDUJTukTTQTbmh2etH\no2bFIbJgwQKmTp3K22+/jUgV+xwwQIo9BYDSxlIANF1jS/UWLu57cVRsPUZoazx9gCFr+XHk76Go\nD4KxB7kw8rp5PWGPlLJGSllLG/LAQoibhRDLhRDLy8s7thy9saQOt83UIpqiCkFjMExI03H76zGF\ndEAiEEgfoEM4pKJJlZBPJdwAlpDAcI3tzJF0MEdUtaVoarn30sZ7gUmLHBcQMEFYSDRVEhSCRgF6\n5Lw6B/uuKKXRj1lqhIVOpckEWpCQLkkLKNjNCg3YsdkdJLmsLC8Jduj72cuiLeUkOa2Y9uX3CPql\nuXl7aUG75+1Pp8yEpZRvYWhOHwy/w9hPd2nkfQHG4OsBFEop9cigSALK92vTgkMtZN0YDPPsK/+h\nX/EWrFVl+IurKa+3Yw8GCQkblrgQikVHtepYPGH8mNFtFrQ8C5kD80g9+UxQOyu4EOMw0uFITVdj\n7ty5/OxnP2PmzJmcfMrJvDXnLbaylZ5xPffV8+0T34fjUo5jRdkKLIqFW0fcysDEgUYHgXr49m9G\n1S5PNoy9pUUZwxiHRKvj6UeqDjan1XrCQK0QwoPhcw5bPWGn1URYb9lUCOMqIQ22mXowRG5FQSCE\nRJMKmlBQVZ0QKtKEMc1VJSIsmhwrEh2BrhiHBaCrUOcClx8CAoQOqmaErnUVdFWiqxA0CUIKNFoU\nqpwCiSA/Gxw+QUK9RABBE6zvKdjr+hUpkLqKS+oElDCaxYwvpNOvoYZQOJPqkJmGxEGGccFGVtuO\nP6gZndcfxqy2nMeaFEEwrLdxRut0Rnb0kzSl4gtASil/0k57gTHTnQN8R5OWYw6wlsgAFELURT6v\npOmJcm+bFhxqIes5n3/LiA1LSJABXNs3EOevxaQZj2ASCNaZES4dh0ejxuJEs5oxZVrYbc8kZ+TV\nMQd89LBECPERxo/XZ9E2pqN8+umn3HjjjcyaNYsBIwdwx8I7qA3V0lDXwNbqrQxOGkyKI4UUewpT\n+k5hSt8pLTsIB+CjmyG+J7jTjcIhH90Ml74Opth+4B9BZ4+ntuoJP0eTOMiTh+tivZKdKAh8wTB2\ni/GbFmcz47CoNAQ1Fun96Jm0m9GVWzEJCAozQZOZetVBCjWIOImmgLUhBBJECBQkmiKp9OgETIK4\nBgUVnYoMC2v6uhjxfT1qWKPWDU6fwGcCvx3CDklQEWzMNFFrVVB8AhNQmGglkG4mNxxgtx/8usaO\nJKizmjHrxv8Iq65gw0SjUBhtUhgRgtNsmcT1uozf7hzBCQ0LSGysJohkl2sklaknHdT3dObAVD5Z\nU0yKu+nfTkFVIz8ZfXDLjp3hNR7a7/2BnOD/ARMAD5AHrBRCPIexDPBXYDuGkw4Dr0opg0KI/dsc\nFhzzZ2Mr30ZG+U6sjWFjfQLALIzYhgRMgg/TTqHQk03fdDs+Tx/q3Hmc2P+Ew2VGjE7mECM1UeWD\nDz7gtttuY/bs2YwePZpHv32UPp4+5MXnsWT3EkJ6iGWlyzgl6xTuOf6e1jtZ/4HhfG2RpECbx3i/\nbiYcd9WRu5luxoHGkxAiDeO3CinlQWfItVNPeD3QKSm9j108lGe+2MyOci+KIrj19D7UVlVx18eb\nEGY776edx2bLVkbrNSTHNTCsdjUl0kS11U041Yxq18hy7qGxwoa1MUyirKZhiBOvxU5FqRl7VTUJ\nLhvLz7qNssp8iphL/10l7HQpFAzKY7EzDVfpHnIDYfQ4E9XmeHY6/NRJ46dYNztJCplIF/nE231M\nMgX5hwWqtEzig40M8AukamKLyUGjLZXMwf24OO0UTCn9wZFI3wVbWVKSTG6infqARmF1I0+c0eeg\nviO3zcx143J5e2kBJlUQ1iRjeiVyUl7KQfXTGU74Lxhlvcowsvl6AiPbaiylfB54vp3+itlvoEkp\nn/nxZrbEH9IIlBWSU7oTmy+MaP7soIFuFWAR1Lpc/FcfjqPvqcSluNhZ2cDD42NbP44WDjZS0xV4\n5513uOuuu5g7dy7Dhw8HoDZQS4LVWAs+Pft0NF1jZ/1Obh52M06zs/WOyreCbT9tEnsClG/pTPO7\nNQcaT5HCDpUYkwjJUbIdzmU18fvzm5JLd+7cyZnnncdPbrqTwqTROK1mnNZMNoc0rpk6lESnheSC\nEt5dX0GcXsmV/QVqj+PIr9ZIl37iM9MQ6GhSQW2mgtVUVmo6AHneEmZumck5UqO35U6+2SLQdMkF\nvZO4aGQWv//0e3xaBRvrv0YXXnboOaQ3WJl0/lmcnDKYm/71JQN7ZuCRAlQLuaqVmsYQ/3f80Bb3\n94sz+rJ6Vw2fryumV7KTu88egMt68O7wlH4pnNw3mQpvEI/djMV08GlWnbFP+DohxCNSyt8BCCFu\nP9zX6AymPTWTMcJHf9/efyvN0AEFwg4T5XEeEuP9pCc5yUywceekfsTZYkXPjyIONlITVV577TWm\nT5/O/PnzGTRo0L7PzYq5RSEGVVHRdI1EW2LbnfU7G757FZKaPfHX7ILRN7R9TowDcaDxtL4zJg1H\nkq1btzJhwgTuvvtubrvtNkKazsqCauxmlaE9PPvGYI/cTO7KzWxx7sAWz4MqB8rRz3Bl8MuRv9z3\n/sKWvpPbx/flng99DHZegMWksKfWz5STszDlZBIHTBkygkWby7EmOfEFNQqr63hkSuuTpBE58YzI\n+fFCHUKIFiHpg6WzFjF7CSGuwZgJD+ikaxw21q7bwSnfvcfYkk20+pus6vh729mU3JMd7kwunDqF\nc8fGZr9HKQcVqWlrj2bk2FBgPtBbSuk93Ia++OKLPP744yxYsIB+/fq1OHZZ/8t4Ze0r5MXnYVJM\nFHmLGJQ4CIfZ0UZvQM4JsGUOlG8yVIJ8NRCfAzljD7fpxxIHGk8XCiH6AA0AUsrfHHELfwQbN25k\n4sSJPPDAA9x0k1How6wqnNA7OmUJk11WXr5qFF9tKqOyIcCdE/qR5GpygJcf35OxvZL4ZE0xOYkO\n7p08AIela+fqdJZ11wMTI6+79ExY9/nQr76Y0SIAgdYmRYJggpnF/Y5js8zBmZDBuWOHttIuxtHA\nIURqWt2jKYQwAzdiJBQedmbMmMHzzz/PokWL6NXrh+pqw1KGcdfou3hv83vUBGs4u9fZnNbjtPY7\nFQImPgRV+VC8CjJGQFLv9s+J0S4dGE9X02znzRE17keydu1azj77bJ544gmuvjo6alKtYVIVJg1u\nuypkrxQXd0zo1+bxrkZnOeFbgSFSypuEENMxtFW7HPX1dWy4fSoWi4YpqIMiIzu9mwhZBf/pN4Z/\nyfPolZnGR784rNuSY0SHg4nU9MDYjw4t92jehZHLML2tEw91q9zjjz/Oa6+9xuLFi8nOzm6zXU9P\nT35z/CFMrBJ7Gf/FOFy0N56GYVRR0oF/0VRtqUuzYsUKzj33XJ577jkuu+yyaJvTreksJ9yHps3l\nXXITopSSD/7xFL2FQJjMWEMhFFUiJcZPrVnic1jwuVO4+Y9P8Zvk5Bab12Mc1RxMpKatPZojgDTg\neAyloj/tf+LBbpWTUvLggw8yc+ZMFi9eTEZGxoFOidE1aG88nb83UUsI8SLwyZE07FD45ptvuPDC\nC/nb3/7GlClTDnxCjB9FZylmScAe0TbNPFDjaLB+dy2EAuBxoJlNBCxmpElBmCXYoW6Qi+Wnns1J\nX35FQkpKzAF3L24FLoyI6rexn2cfHwEXR35APxNCvAUgpbxMSnkHsAx4+ccaJKXk3nvv5eOPP2bh\nwoUxB3x00d54sgshcoQQOUAbaetdh8WLF3PhhRfy+uuvxxzwEaKzZsJ/AqZhhGG6ZEp+jS9Eg+qG\n7DS8+ZXICithk4oQkmVpg1g5fDw3XhYbhN2UDkdq2tqj2ez4tT/WGCkld9xxB0uWLGHBggUkJUUn\n6SXGIdPeeHoQQwsBuuiy3F7mzZvH5Zdfzrvvvsv48eOjbc4xQ2c54SuklF26ctKY3ETeUoehbFtC\n0qBUGus0RJ2fApHOshNv4cmfjiTDY4+2mTE6hy4TqdF1nWnTprFmzRrmzZtHfHx0a5vGOCTaHE9S\nyu3A3VGx6iCYPXs21157LR9++CGnnnpqtM05pugsJ3yyEOI+oBZASnnYVK0OF6XFRXz59ovsmXQx\nExPLcCVU40rJ5raf3YaIVZzp7nSJSI2madxwww3k5+fzxRdf4HZ3yfSJGAemS4ynQ2XWrFncfPPN\nfPrpp4wdG9uudqTpLCf8VCf1e1jYtm0bEyZM4Ne//jXTfnEju6oaSXRa8NhjohvHCFGP1IRCIa6+\n+moqKiqYM2cODkc7+3tjdHWiPp4Olffee4/bb7+dOXPmMGrUqGibc0zSGQUc7sbIGH0jkqjQpdi0\naRMTJ05k+vTp3HzzzYAhWB7jmCKqkZpgMMhPf/pTAoEAn332GTab7UhePsbhp1PHkxAik/1qqjc7\n9hRGLfZ44EYpZavVlFrjzTff5N577+WLL75g2LBhh9PkGAdBZ8yEB0UKXb8MdCknvG7dOs466ywe\nf/xxrrnmmmibEyN6RC1S4/f7ueSSSzD/f3vnH21lVebxzxdFQ1FWBCKi6RiTYjBOZayxVqbN4ApI\no9JimPBH2YwtNROVcpJR81cCCS5IQChAQCmQSg20CQasbDR15RSsxopR0AmT4Ye1yJ8888ezD/ec\n955zOefe99xz7r3PZ62z3nvfd+9n7/2++32f/et5du/erFq1ioMPjt2LugH1rk+FPdU3AkuBfUrY\nzK4GkHQFbjb3k2oEzp8/nxtuuIG1a9cybNiw/HMcVE09lPAASWOAQemIma2uQzo18dRTTzFmzBhm\nzpzJ+PHjG52doEE0cqRmz549jBs3jv79+7NkyRJ6947pj65O3vUpuUK9NXP6JUr3VM/GORI4hTIb\n4ZRzGDN79mymTZvG+vXrGTp0aEezHHSQetgJrwQGAt9LxwFtB68/jz32GKNHj2bOnDmhgIOTzOxc\n4KzOTHTv3r2MHj2awYMHs2zZslDA3Ydc65OZ/crMPlr8Azbje6q3+l5LGoL3xC8xszfLyLvLzE4x\ns1MGDhzI9OnTmTFjBhs2bAgF3CTIrKk3kSlLW071s5x44om2Y8cOFi5cyNixYzstjz2cpvVsIukB\nYA7eO7gLOmekpm/fvjZhwgTmzp1Lr1718pETFNEpdbAz6lOaEy7sqb7UzNZJWmJmEyU9CTyDz0fP\nMbOnK8kZMmSIHXrooaxbt46jj65t4/mgXVRVB7uqEp4I7Co41Tezis5Ne/fubatXr2bUqFGVggT5\n08xKOLsYwMzs7nqnO2jQINu2bVt4Xus8OksJN6Q+tYc+ffrY5s2bwxtb59GtlfA1wBoz+6Wke8xs\nQub6vnkQfJuxX+echQHA9iaWVw+ZtcjbbmYfyTHtLo+kl6jdeX896kUj0+nMtN5iZrHfaBFl6mAe\nz6IZZDRDHsrJqOo72NwbLVamklN9oNRxvqQnzOyUPBPPW2ZPzWNPwswG1hqns+55Zz7bzixTvdPo\namTrYB7PohlkNEMeOiKjqyrhVcBsSWOBBxqdmSAIgiBoD11SCZdxqh8EQRAEXY6esEzzri4gs6fm\nMWibzrrnnflsu2OZuip53KNmkNEMeWi3jC65MCsIgiAIugM9oSccBEEQBE1Jl5wTroZaHHpk4o0D\nxuJO0b8FjAD+CugNXAwMJuNMXdKVxWGszPBCys8GfJPvEzoiM3nOuTHl8QngdeAM4GDgCylYSdkl\nTSgOk+bVi/P3dtzt3Q7c+H9LR2UG1dNWfU2uDNcCx5vZn+uRjqTBwDW4beNyM/tZndIZA1yQgi0w\nsx91JJ0k83jgq0A/Mzun6PxwvEwAt5pZ3qaKXY6ObAbRxjOt+j63IeMr+PduAHC5mT1fS/x0rar3\nJI93INf6bWbd8ofv7XlW+vs77Yj/VmAhsCz9fynwQWAKrph7AfcAB2XDVJD3NWAycHZHZQIfBxYD\ntwN/D6xI5z+ayt2q7NkwZfI3FvhMIU4eMuPX8fqKN8LuSM+7bx3T+QZwc0rr2DqmMw03LxwMzMz5\nHq7M/D8fVyj9gHmNfsbN8Mt+ayqEuaLcd6yNZ1r1fa4ko+j6x9v6luTxnuTxDuRZv7vzcPTRwNb0\ndyufqlVwLbAAd54ObuR+dEGume1N599WJkwJkkYBm4A/4hW1ozJPAB41s0l4L9UyYcuVPRsmy38C\nn5O0DngoJ5lB9VSqr1fhIxR5Ld6olM67gLvxkZopdUxnObAC9y3/7RzSaYt+ZrbLzHYDh9U5raZD\n0ghJDxb/gOMp/dZk4xQ2g3i0zOVKz7SW+1zxuyypL/Ap4PvtiF/Le5LHO5Bb/e7OSrjg0ANqKKec\n24A1wC9o2YDi7Unm85Q6U/+/MmGynA78HTAh/Y7ooMzngZ3p7+IKUCIvncuWvVIeLwSuM7MP473i\nPGQG1VPp/v4tcBm+U8+/1DGdQp36M5DHBseV0rkG+BBwGvCVHNJpi92S+kk6HKh6n93uguW8GQSV\nn2kt97msjBR3DjDZ2t4TOY/3JI93ILf63W1XR6cx+9nAK8BPrfo54S8C5+MK+JfAIcCxtMyNDqa1\nM/VJxWGswk2VdAHu1uydHZEp6RBgFrAH+A1ecT4I9AEuScFKyp7mb/eFsdZzwsPxFuB2vBI+1VGZ\nQfVk6yvwETObWHR9EXCp5TMn3CodSSfh0yWGz2XlMSdcLp0JQMGV348tBz/Lkt6GDyOOwkevTkpp\nDcfLBDDVYk64Q5tBtPFMq77PbchYhQ8pvwB818zW1RK/6Poi9vOe5PEO5Fm/u60SDoIgCIJmpzsP\nRwdBEARBUxNKOAiCIAgaRCjhIAiCIGgQoYSDIAh6EJIOkjRL0kxJiySNTOdXtlNeTXokm46kM1I+\nZkuaK6nmbT+LZB0naXqFa+MlLZT03kphGkG39ZgVBEEQlOUiYLWZrZF0IHAf8DHgBEk3Ae9IYa7F\nTXW2mtntkr6Km04ehjv0uBe3J35B0ggzu0rSF4D/BgbiZpmH45YchwP/DPy2OCOSBHwJGFewAJF0\ngKQPAGcBRwI34fbNpwGn4g45/oCbUvYB7rOMZypJx+E2v/fj9r9fxL0TbsItTgphLk35/jowF/jr\nYrnAUbjJ0WZgr5ndLOlq3Cxzl5lNyd6X/ZhYtaLH9YSbsBX4G0nj09+XSfppe/KRkblI0jxJ90k6\nU9K5qYX5+3S8PBN+ZjJ7qlb+kZJu7Gg+eyJNWP8el/RNSfdLOkrSWyVNLROvYg8jD2otf3F4SfPz\nz1G3ZjhugomZvQHskXQAsN3MrsX3az8bt4N9FFggaRiuBHfjbnKH4fpjqpktpsX2+DRgPe7pcAa9\nzAAABtlJREFUbzfuoGgkbop5Ea7oihmIK3mTNEbSMlx5v4Z7DtwDfCIp2UW4O90luFLdhSvjkRXK\nudHMpuOuePumfN0J/G8b96ac3IfM7EZgeHIoMsLMLksKuNx9qYme2BNumlZgYiP+EJfjL8c2gCTr\nnbg7uCnAyXiL7AhgEv6SlLTQMnKvTHmdbGZXACskrTSzi4sDJRtLzGxPsrHbglekR/DW3oHAl/HW\n5/PAz8zs+5KGSDrIzF6r4p4HLTRb/dtiZpck+8ZTgSHAStjnS/hAvE58D3hfUtDH4E5nzsZ7K2/B\nbcxJx1eAB3Cb0+tS/LtxO9ALksyfm9nSonxky/9+SnskA3CnN3/CPSOdJOl63C74V5JGmtnj+737\nAXhv8L3Aw6kOHmJmb3qnFGjxOvVZ4AP4t+lqXKldXwgkaU9S4gDrUvjnzGyvpL9kwn4It01+NZOX\nl4BjJMnMVks6AncZfAHwj3g9OEPSYcAtwOeSwu4F3FSUfjkKfgteJ/V+M7xKiw48NB1L5Mp9OxTk\nKP2K7Xp7Ze9LrfREJTwc+C54K1BSSStQ0rm0tALvxytqobXzc/yDU9wKfEPS8qJW4Dx8o4a1+Mdo\nJL7JwYW4y8pvZvJjwBZJnwCeBM5Mra3zgIdTmPfgLcNe+IfsH9L5h8zsO5LuLVPOabgjjXPKXCvm\nVPylLDAX/4ifb2aTUo/jYLySrsGVM7jyfzfw2H7kB6U0W/07RtId+LdgMt7bmCd3XPCamV0N+4bu\nnjGzyZJm4sOE55nZJyUdi3srAphiZv8jaQXuzGAn7qP415KWAL9P4d4NFCvhbPkn4O/DrlQGAf8F\n/MDMtkraVPjwSXoqlTGUcHXMB74h32ygH3BrOj9A0s34Rgqfxx2g9AI2m9lGSXsl3Y43jG7JyFwB\n/A5/DgBLJd0F/AX4IV4vp5I6GQWSQp0JLJK0C++xfh2vwzfg352d6e83gC9L+iHeEFsgaQfwhJnd\n0477sA3oJ3eM9DfpXIncbAQz+5OkTZJmADvM7MbsfTGzrdl4bdETlXAztQILLMU/sO8CzsQ/OC9k\nZPzAzD4m6XzcixeUttCyXI33Kj6NV+BKHI5/6Aq8nM69XDiRKt7ElLfZ+NzKzhQuqI1mq39bzezy\norC9zexVeYay/oV3p+MrlPYsjJY6WMi/mdkGSZtxn+Qn48OLd5jZTipTiN+qp5NkTJM0hdLeSNTF\nGjCzV/Hh4uz5EZlTkzPXr8pcP6fo2k7c533h/7vx0Y9iHqEM5js5/UfmdLlRmywPZ0+Y2bO4H2kK\nRzMruI+8vihoIcyF6Xh7Oj5bTm6SMz4db8ucz96XmuiJSrhpWoEFzOx5Sceb2euSCkrvcUmz8I/b\nt4FNaUhyGPDjagpqZk9LulLSIDN7sUKw3xbluyzyLb7+FfdTvTGdfgfwYDX5CEpouvqX4QVJR6Q0\n+6Qh6eco/6yXSpqHNwoLawRulLQHuFfSGfhwdX/g34HbgFmSXgSeNbNZRbKy5d9BaY+kL75oZi/u\nW32jfI76DrwuPlNF2YKg6Qi3lT2c1OOZk50rriLOYjM7r345CxqBpPcBJ5vZgkbnpVpSY/Xf9tPD\nDupEmjc9h9Q4MrOpabRisZltyYRdxP59O08rTINUmf7pwHFmtqiGONfjW1/u15+4pFuAGWb20v7C\ntoee2BMOikhzMsskHWJme6qMNhDvgQTdDDP7haT+jc5HjawOBZwvkoZSusjuaUpNfj5rpT24uWb2\noNwi4wBgqJltkfQe3ARpO76IDnxe94T0/3p8pGY3Psf6NXw0BLXTFKiwRiGzWPEofB3CYbRsSFOi\njCUtN7Px2XTwKaEL8dGk3OlxJkpBa8zsJzUoYMzsj2b2ZD3zFDQOMys7J9asmNmaRuehG3Ixvsju\nIuAz6Vyxyc+RmfCfl5uKLcanzF5I5ycBF5vZJDMrLACdi6/W/yS+zuQRM/sScKx8S8PC3sLtMgUy\ns+dobTJFijcEV8ZlqZDORnyVdl2InnAQBEGQpdgUp3Bsy+Rnvpk9CCDp/bQs9sya9IAruDeSjHLX\ny8Wr1RQou1jxU2Z2tqTraFnYCslMKU2x9amUTtHCydwJJRwEQRBkmUfRIrsa4/4O+Kf09wzgzrQY\nb1GZsD8C5koaga/Uf7locWpHTIGyixX/IGkybu62oSjcGtxBx0Z8jVS5RZC78RX4dSEWZgVBEAS5\nIulOfF72zU5I63RqXJhVo/yJwIuWcY2ZFzEnHARBEOTNHGBQozORE9vrpYAhesJBEARB0DCiJxwE\nQRAEDSKUcBAEQRA0iFDCQRAEQdAgQgkHQRAEQYMIJRwEQRAEDSKUcBAEQRA0iP8H9QmPTi6/NDkA\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1402f2390>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "figure_5(odd_summary, pred_multiplicative_summary, pred_nobias_summary, alpha=1);\n",
    "plt.savefig('results/figures/figure_5_absolute_model_fit.png', dpi=330, bbox_inches='tight')\n",
    "plt.savefig('results/figures/figure_5_absolute_model_fit.pdf', dpi=330, bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [Root]",
   "language": "python",
   "name": "Python [Root]"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}