Estimate_CO2_profile.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "37932aac",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from matplotlib.colors import LogNorm\n",
    "import glob\n",
    "import os\n",
    "import random\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "import plotly.express as px\n",
    "import tensorflow as tf\n",
    "import keras\n",
    "from keras.layers import Input, Dense, Flatten, Reshape, Convolution1D, Conv1D, \\\n",
    "    MaxPooling1D, UpSampling1D,Cropping1D, LSTM, RepeatVector,Activation, TimeDistributed\n",
    "from keras.models import Model, Sequential\n",
    "from keras.callbacks import EarlyStopping, ModelCheckpoint"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0d794d47",
   "metadata": {},
   "source": [
    "# Process description\n",
    "Below is the scheme of the pilot plant (adapted from Process summary, the ChemEng Discovery space, Imperial College London). There are six sampling points between stages, and all the sampling points are fed to the same gas analyzer. However, the gas analyzer can only process the sampling stream one at a time, i.e., at any given time, there is only one position that the gaseous $CO_2$ concentration is known.\n",
    "<img src=\"images/pilot_plant.jpg\" width=\"500\">"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "e5d536eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Records in the spreadsheet are are already labelled\n",
    "data_path_list = glob.glob('./data/withLabel/1*.xlsx')\n",
    "list_of_origin_df=[]\n",
    "for i in range(len(data_path_list)):\n",
    "    xls = pd.ExcelFile(data_path_list[i])\n",
    "    data_df = pd.read_excel(xls, sheet_name=0, index_col=0, header = [0,1])\n",
    "    data_df.columns = data_df.columns.map(''.join)\n",
    "    data_df=data_df.rename_axis('time').reset_index()\n",
    "    tmp_name=list(data_df.columns)\n",
    "    tmp_name[-1]='label'\n",
    "    data_df.columns=tmp_name\n",
    "    list_of_origin_df.append(data_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6f6e6f03",
   "metadata": {},
   "source": [
    "Labelled process record:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "eb0da692",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>time</th>\n",
       "      <th>AT100(pH)</th>\n",
       "      <th>AT300(pH)</th>\n",
       "      <th>AT400(CO2 %)</th>\n",
       "      <th>FT100(kg/hr)</th>\n",
       "      <th>FT101(kg/hr)</th>\n",
       "      <th>FT102(kg/hr)</th>\n",
       "      <th>FT103(kg/hr)</th>\n",
       "      <th>FT104(kg/hr)</th>\n",
       "      <th>FT105(L/min)</th>\n",
       "      <th>...</th>\n",
       "      <th>TT400(0C)</th>\n",
       "      <th>TT401(0C)</th>\n",
       "      <th>TT402(0C)</th>\n",
       "      <th>TT403(0C)</th>\n",
       "      <th>TT404(0C)</th>\n",
       "      <th>TT410(0C)</th>\n",
       "      <th>TT411(0C)</th>\n",
       "      <th>TT412(0C)</th>\n",
       "      <th>TT413(0C)</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2014-01-20 17:23:02</td>\n",
       "      <td>10.468116</td>\n",
       "      <td>10.856668</td>\n",
       "      <td>0.109407</td>\n",
       "      <td>49.641636</td>\n",
       "      <td>0</td>\n",
       "      <td>114.306496</td>\n",
       "      <td>253.188339</td>\n",
       "      <td>244.473846</td>\n",
       "      <td>0.076785</td>\n",
       "      <td>...</td>\n",
       "      <td>117.106621</td>\n",
       "      <td>116.955292</td>\n",
       "      <td>116.753510</td>\n",
       "      <td>22.537117</td>\n",
       "      <td>104.836342</td>\n",
       "      <td>22.658978</td>\n",
       "      <td>34.482868</td>\n",
       "      <td>36.524200</td>\n",
       "      <td>33.270100</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2014-01-20 17:23:46</td>\n",
       "      <td>10.429050</td>\n",
       "      <td>10.856927</td>\n",
       "      <td>0.098878</td>\n",
       "      <td>49.669559</td>\n",
       "      <td>0</td>\n",
       "      <td>73.167267</td>\n",
       "      <td>253.357178</td>\n",
       "      <td>244.543427</td>\n",
       "      <td>0.076773</td>\n",
       "      <td>...</td>\n",
       "      <td>116.413254</td>\n",
       "      <td>116.233284</td>\n",
       "      <td>116.142960</td>\n",
       "      <td>22.537117</td>\n",
       "      <td>105.931946</td>\n",
       "      <td>22.678768</td>\n",
       "      <td>33.612118</td>\n",
       "      <td>35.468010</td>\n",
       "      <td>32.640736</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2014-01-20 17:24:29</td>\n",
       "      <td>10.380840</td>\n",
       "      <td>10.868021</td>\n",
       "      <td>0.001587</td>\n",
       "      <td>49.750290</td>\n",
       "      <td>0</td>\n",
       "      <td>30.941877</td>\n",
       "      <td>254.384949</td>\n",
       "      <td>244.684982</td>\n",
       "      <td>0.076896</td>\n",
       "      <td>...</td>\n",
       "      <td>115.733566</td>\n",
       "      <td>115.700668</td>\n",
       "      <td>115.546623</td>\n",
       "      <td>22.537117</td>\n",
       "      <td>106.712189</td>\n",
       "      <td>22.689386</td>\n",
       "      <td>32.948505</td>\n",
       "      <td>34.523560</td>\n",
       "      <td>32.053650</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2014-01-20 17:25:12</td>\n",
       "      <td>10.319781</td>\n",
       "      <td>10.888022</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>49.734818</td>\n",
       "      <td>0</td>\n",
       "      <td>65.672806</td>\n",
       "      <td>254.687668</td>\n",
       "      <td>244.545029</td>\n",
       "      <td>0.076057</td>\n",
       "      <td>...</td>\n",
       "      <td>115.841347</td>\n",
       "      <td>115.708611</td>\n",
       "      <td>115.522621</td>\n",
       "      <td>22.537117</td>\n",
       "      <td>107.563896</td>\n",
       "      <td>22.699842</td>\n",
       "      <td>32.655727</td>\n",
       "      <td>33.851078</td>\n",
       "      <td>31.548174</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2014-01-20 17:25:55</td>\n",
       "      <td>10.248368</td>\n",
       "      <td>10.903030</td>\n",
       "      <td>0.000364</td>\n",
       "      <td>49.566025</td>\n",
       "      <td>0</td>\n",
       "      <td>15.200113</td>\n",
       "      <td>254.401810</td>\n",
       "      <td>244.331406</td>\n",
       "      <td>0.076959</td>\n",
       "      <td>...</td>\n",
       "      <td>114.879135</td>\n",
       "      <td>114.479225</td>\n",
       "      <td>114.496765</td>\n",
       "      <td>22.537117</td>\n",
       "      <td>107.848602</td>\n",
       "      <td>22.714924</td>\n",
       "      <td>32.635029</td>\n",
       "      <td>33.495365</td>\n",
       "      <td>31.267054</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 92 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                 time  AT100(pH)  AT300(pH)  AT400(CO2 %)  FT100(kg/hr)  \\\n",
       "0 2014-01-20 17:23:02  10.468116  10.856668      0.109407     49.641636   \n",
       "1 2014-01-20 17:23:46  10.429050  10.856927      0.098878     49.669559   \n",
       "2 2014-01-20 17:24:29  10.380840  10.868021      0.001587     49.750290   \n",
       "3 2014-01-20 17:25:12  10.319781  10.888022      0.000000     49.734818   \n",
       "4 2014-01-20 17:25:55  10.248368  10.903030      0.000364     49.566025   \n",
       "\n",
       "   FT101(kg/hr)  FT102(kg/hr)  FT103(kg/hr)  FT104(kg/hr)  FT105(L/min)  ...  \\\n",
       "0             0    114.306496    253.188339    244.473846      0.076785  ...   \n",
       "1             0     73.167267    253.357178    244.543427      0.076773  ...   \n",
       "2             0     30.941877    254.384949    244.684982      0.076896  ...   \n",
       "3             0     65.672806    254.687668    244.545029      0.076057  ...   \n",
       "4             0     15.200113    254.401810    244.331406      0.076959  ...   \n",
       "\n",
       "    TT400(0C)   TT401(0C)   TT402(0C)  TT403(0C)   TT404(0C)  TT410(0C)  \\\n",
       "0  117.106621  116.955292  116.753510  22.537117  104.836342  22.658978   \n",
       "1  116.413254  116.233284  116.142960  22.537117  105.931946  22.678768   \n",
       "2  115.733566  115.700668  115.546623  22.537117  106.712189  22.689386   \n",
       "3  115.841347  115.708611  115.522621  22.537117  107.563896  22.699842   \n",
       "4  114.879135  114.479225  114.496765  22.537117  107.848602  22.714924   \n",
       "\n",
       "   TT411(0C)  TT412(0C)  TT413(0C)  label  \n",
       "0  34.482868  36.524200  33.270100      1  \n",
       "1  33.612118  35.468010  32.640736      1  \n",
       "2  32.948505  34.523560  32.053650      1  \n",
       "3  32.655727  33.851078  31.548174      2  \n",
       "4  32.635029  33.495365  31.267054      2  \n",
       "\n",
       "[5 rows x 92 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list_of_origin_df[0].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "84bd61c6",
   "metadata": {},
   "source": [
    "The measurements consist of 53 temperature transmitters, 11 pressure transmitters, 19 flow transmitters, 4 level transmitters, 2 pH analyzers and the previously discussed gas concentration analyzer. The measurements cover the pilot plant process including the elements shown in previous figure as well as the various gas handling and storage systems and the auxiliary processes. Key measurements of the absorber column include (with descriptions on the P&ID):\n",
    "\n",
    "* FT103: Flow transmitter, $CO_2$ lean amine solution, valid range 120-12000 kg/hr.\n",
    "* FT104: Flow transmitter, $CO_2$ lean amine solution, valid range 100-1600 kg/hr, connects in serial with FT103.\n",
    "* TT210: Temperature transmitter, temperature of $CO_2$ lean amine solution, K.\n",
    "* TT211: Temperature transmitter, temperature of $CO_2$ lean amine solution, K, connects in serial with TT210.\n",
    "* FT303: Flow transmitter, $N_2$ absorber inlet (volumetric) flowrate, valid range 38.7-575 kg/hr.\n",
    "* FT304: Flow transmitter, $N_2$ absorber inlet flowrate, valid range 4.4-438 kg/hr, connects in serial with FT303.\n",
    "* PT403: Pressure transmitter, $N_2$ absorber inlet pressure, valid range -1-4 barG.\n",
    "* TT304: Temperature transmitter, $N_2$ absorber inlet temperature, K.\n",
    "* FT301: Flow transmitter, $CO_2$ absorber inlet (volumetric) flowrate, valid range 7.14-71.4 kg/hr.\n",
    "* FT302: Flow transmitter, $CO_2$ absorber inlet flowrate, valid range 1.56-156 kg/hr, connects in serial with FT301.\n",
    "* PT402: Pressure transmitter, $CO_2$ absorber inlet pressure, valid range -1-4 barG.\n",
    "* TT104: Temperature transmitter, temperature of $N_2, CO_2$ mixture at absorber inlet, K.\n",
    "* PT111: Pressure transmitter, bottom of the absorber, valid range -1-4 barG.\n",
    "* PT403: Pressure transmitter, top of the absorber, valid range -1-4 barG.\n",
    "\n",
    "Note the pressure transmitter of $N_2$ and of the top of the absorber is the same. The lean gas ($N_2$) is recycled from top of the absorber and mix with the $CO_2$, and forms the $CO_2$-rich feed gas."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5d4f59cd",
   "metadata": {},
   "source": [
    "# Kinetic (dynamic) modeling\n",
    "Instead of implementing complicated kinetic correlations, we fit a pseudo reaction to approximate the absorption process\n",
    "$CO_{2,(g)} + MEA_{(l)} \\underset{reaction_2}{\\stackrel{reaction_1}{\\rightleftharpoons}} \\left[ CO_2-MEA\\right]_{binded_{(l)}}$\n",
    "\n",
    "$R_1 = k_1 \\cdot \\exp^{\\frac{-E_{A1}}{8.314\\left(\\alpha_1\\cdot T +\\beta_1\\right)}}$,\n",
    "$R_2 = k_2 \\cdot \\exp^{\\frac{-E_{A2}}{8.314\\left(\\alpha_2\\cdot T+\\beta_2\\right)}}$,\n",
    "\n",
    "The vector of parameters are combined as x0 = [$k_1, \\alpha_1, \\beta_1, k_2, \\alpha_2, \\beta_2$]\n",
    "\n",
    "kinetic_model contains the Simulink model of the absorber. Please note that the parameters in the Simulink model are outdated and need to be run by Matlab script to update these parameters.\n",
    "The result_generation ran the kinetic_model with the fitted parameter: x0=[500.6,1,1005.5,4380,1,1263]. The results are tagged as \"time stamp.csv\". A schematic of the Simulink model is shown below:\n",
    "\n",
    "<img src=\"images/kinetic_scheme.png\" width=\"1000\">"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "09a304d7",
   "metadata": {},
   "source": [
    "# data-driven modeling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "c0880690",
   "metadata": {},
   "outputs": [],
   "source": [
    "def reset_random_seeds():\n",
    "    os.environ['PYTHONHASHSEED']=str(0)\n",
    "    tf.random.set_seed(0)\n",
    "    np.random.seed(0)\n",
    "    random.seed(0)\n",
    "    \n",
    "def rmse(pred, label):\n",
    "    return np.sqrt(((pred - label) ** 2).mean())\n",
    "\n",
    "def avgOutPoint1(df):\n",
    "    #Change point 1 readings to Avg. of Point 2.\n",
    "    label_list=list(df['label'])\n",
    "    conc_list = list(df['AT400(CO2 %)'])\n",
    "    p=0\n",
    "    num=0\n",
    "    i=0\n",
    "    sampling2_avgcon = list()\n",
    "    while i<len(label_list):\n",
    "        if label_list[i]==2:\n",
    "            p=p+conc_list[i]\n",
    "            num=num+1\n",
    "            i=i+1\n",
    "        else:\n",
    "            if p ==0 and num==0:\n",
    "                i=i+1\n",
    "                pass\n",
    "            else:\n",
    "                sampling2_avgcon.append(p/num)\n",
    "                i =i+1\n",
    "                num=0\n",
    "                p=0\n",
    "\n",
    "    i=0\n",
    "    k=-1\n",
    "    while i<(len(label_list)-1):\n",
    "        if label_list[i]==1:\n",
    "            conc_list[i]=sampling2_avgcon[k]\n",
    "            i=i+1\n",
    "        else:\n",
    "            if label_list[i+1]==1 :\n",
    "                k=k+1\n",
    "                i=i+1\n",
    "            else:\n",
    "                i=i+1\n",
    "    #set new AT400\n",
    "    df['AT400(CO2 %)']=pd.Series(data=conc_list,index=df.index)\n",
    "    return df\n",
    "\n",
    "def columnSeparator(df):\n",
    "    # linear interpolation to fill missing values\n",
    "    for i in range(1,7): # 6 sampling points\n",
    "        new_con =list()\n",
    "        j = 0\n",
    "        while j<df.shape[0]:\n",
    "            if df.iloc[j]['label']==i:\n",
    "                new_con.append( df.iloc[j]['AT400(CO2 %)'])\n",
    "                j = j+1\n",
    "            else:\n",
    "                new_con.append(np.nan)\n",
    "                j = j+1\n",
    "        df[str(i)+\"_sampling\"]=pd.Series(data=new_con,index=df.index)\n",
    "        df[str(i)+\"_sampling\"]=df[str(i)+\"_sampling\"].interpolate(method=\"linear\")\n",
    "    # Notice the df will contain NaN for the first few rows for each sampling point column\n",
    "    return df\n",
    "\n",
    "def getU(ens,reduced_dimension):\n",
    "    # assuming ens has shape of [sampe_size, sample_ravel()]\n",
    "    u, s, vh = np.linalg.svd(ens.T, full_matrices=False)\n",
    "    truncation_parameter = reduced_dimension\n",
    "    print('Truncation parameter: ',truncation_parameter)\n",
    "    u1=u[:,:truncation_parameter]\n",
    "    s1=s[:truncation_parameter]\n",
    "    vh1=vh[:truncation_parameter,:]\n",
    "    return u1\n",
    "\n",
    "def dfPOD(train_df_list, test_df_list, _train_feature_list, _label_list, reduced_dimension):\n",
    "    print('Train df(1st one) shape before POD: ', train_df_list[0].shape)\n",
    "    print('Test df(1st one) shape before POD: ', test_df_list[0].shape)\n",
    "    \n",
    "    label_list=_label_list.copy()\n",
    "    train_feature_list=_train_feature_list.copy()\n",
    "    #train_feature_list.remove('AT400(CO2 %)')\n",
    "    #label_list.append('AT400(CO2 %)')\n",
    "    \n",
    "    for i in range(len(train_df_list)):\n",
    "        if (i==0):\n",
    "            train_full_values=train_df_list[i][train_feature_list].values\n",
    "            train_label_values=train_df_list[i][label_list].values\n",
    "        else:\n",
    "            train_full_values=np.concatenate((train_full_values, train_df_list[i][train_feature_list].values), axis=0)\n",
    "            train_label_values=np.concatenate((train_label_values, train_df_list[i][label_list].values), axis=0)\n",
    "    for i in range(len(test_df_list)):\n",
    "        if (i==0):\n",
    "            test_full_values=test_df_list[i][train_feature_list].values\n",
    "            test_label_values=test_df_list[i][label_list].values\n",
    "        else:\n",
    "            test_full_values=np.concatenate((test_full_values, test_df_list[i][train_feature_list].values), axis=0)\n",
    "            test_label_values=np.concatenate((test_label_values, test_df_list[i][label_list].values), axis=0)\n",
    "    \n",
    "    n_set = train_full_values.shape[0]\n",
    "    train_index = list(set(range(0,n_set,1))- set(range(0,n_set,5)))\n",
    "    test_index = list(set(range(0,n_set,5)))\n",
    "    \n",
    "    x_train = train_full_values[train_index]; x_val = train_full_values[test_index]\n",
    "    x_test = test_full_values; \n",
    "\n",
    "    u1 = getU(x_train, reduced_dimension)\n",
    "\n",
    "    x_train_encoded = u1.T @ x_train.T # shape [truncation_parameter, sample size]\n",
    "    x_val_encoded = u1.T @ x_val.T\n",
    "    x_test_encoded = u1.T @ x_test.T\n",
    "\n",
    "    x_train_decoded = (u1 @ x_train_encoded).T # shape [sampe_size, sample_ravel()]\n",
    "    x_val_decoded = (u1 @ x_val_encoded).T\n",
    "    x_test_decoded = (u1 @ x_test_encoded).T\n",
    "    \n",
    "    print(\"Train: \", mean_squared_error(x_train,x_train_decoded));\n",
    "    print(\"Val: \", mean_squared_error(x_val,x_val_decoded));\n",
    "    print(\"Test: \", mean_squared_error(x_test,x_test_decoded));\n",
    "    \n",
    "    encoded_train= (u1.T @ train_full_values.T).T#.reshape(train_full_values.shape[0],train_full_values.shape[2])\n",
    "    encoded_test= (u1.T @ test_full_values.T).T#.reshape(test_full_values.shape[0],test_full_values.shape[2])\n",
    "    encoded_name_list=['encoded_{}'.format(i) for i in range(1,encoded_train.shape[1]+1)]\n",
    "    upper_index=0\n",
    "    bottom_index=0\n",
    "    for i in range(len(train_df_list)):\n",
    "        upper_index=bottom_index\n",
    "        bottom_index=upper_index+train_df_list[i].shape[0]\n",
    "        train_df_list[i]=pd.DataFrame(np.concatenate((encoded_train[upper_index:bottom_index,:],train_df_list[i][label_list].values),axis=1),\n",
    "                                     columns=encoded_name_list+label_list)\n",
    "    upper_index=0\n",
    "    bottom_index=0\n",
    "    for i in range(len(test_df_list)):\n",
    "        upper_index=bottom_index\n",
    "        bottom_index=upper_index+test_df_list[i].shape[0]\n",
    "        test_df_list[i]=pd.DataFrame(np.concatenate((encoded_test[upper_index:bottom_index,:],test_df_list[i][label_list].values),axis=1),\n",
    "                                     columns=encoded_name_list+label_list)\n",
    "    print('Train df(1st one) shape after AE: ', train_df_list[0].shape)\n",
    "    print('Test df(1st one) shape after AE: ', test_df_list[0].shape)\n",
    "    return train_df_list,test_df_list\n",
    "\n",
    "def dfAE(train_df_list, test_df_list, _train_feature_list, _label_list, reduced_dimension):\n",
    "    print('Train df(1st one) shape before AE: ', train_df_list[0].shape)\n",
    "    print('Test df(1st one) shape before AE: ', test_df_list[0].shape)\n",
    "    \n",
    "    label_list=_label_list.copy()\n",
    "    train_feature_list=_train_feature_list.copy()\n",
    "    #train_feature_list.remove('AT400(CO2 %)')\n",
    "    #label_list.append('AT400(CO2 %)')\n",
    "    \n",
    "    for i in range(len(train_df_list)):\n",
    "        if (i==0):\n",
    "            train_full_values=train_df_list[i][train_feature_list].values\n",
    "            train_label_values=train_df_list[i][label_list].values\n",
    "        else:\n",
    "            train_full_values=np.concatenate((train_full_values, train_df_list[i][train_feature_list].values), axis=0)\n",
    "            train_label_values=np.concatenate((train_label_values, train_df_list[i][label_list].values), axis=0)\n",
    "    for i in range(len(test_df_list)):\n",
    "        if (i==0):\n",
    "            test_full_values=test_df_list[i][train_feature_list].values\n",
    "            test_label_values=test_df_list[i][label_list].values\n",
    "        else:\n",
    "            test_full_values=np.concatenate((test_full_values, test_df_list[i][train_feature_list].values), axis=0)\n",
    "            test_label_values=np.concatenate((test_label_values, test_df_list[i][label_list].values), axis=0)\n",
    "    \n",
    "    n_set = train_full_values.shape[0]\n",
    "    train_index = list(set(range(0,n_set,1))- set(range(0,n_set,5)))\n",
    "    test_index = list(set(range(0,n_set,5)))\n",
    "    \n",
    "    x_train = train_full_values[train_index]; x_val = train_full_values[test_index]\n",
    "    x_test = test_full_values; \n",
    "    print('LSTM x_train shape: ', x_train.shape[0])\n",
    "\n",
    "    noise_factor = 0.1\n",
    "    scale_arr = np.var(x_train,axis=0)\n",
    "    x_train = x_train + noise_factor * np.random.normal(loc=0.0, scale=scale_arr, size=x_train.shape) \n",
    "    x_test = x_test + noise_factor * np.random.normal(loc=0.0, scale=scale_arr, size=x_test.shape) \n",
    "    x_train = np.clip(x_train, 0., 1.)\n",
    "    x_test = np.clip(x_test, 0., 1.)\n",
    "    # AE shape default design\n",
    "    input_img = Input(shape=(x_train.shape[1]))\n",
    "    x = Dense(64,activation='relu',kernel_initializer=ReLu)(input_img)\n",
    "    encoded = Dense(reduced_dimension,activation='relu',kernel_initializer=ReLu)(x)\n",
    "\n",
    "\n",
    "    encoder = Model(input_img, encoded)\n",
    "    encoder.summary()\n",
    "\n",
    "    decoder_input = Input(shape=(reduced_dimension))\n",
    "\n",
    "    x = Dense(64)(decoder_input)\n",
    "    decoded = Dense(x_train.shape[1], activation='sigmoid')(x)\n",
    "\n",
    "    decoder = Model(decoder_input, decoded)\n",
    "    decoder.summary()\n",
    "\n",
    "    encoder=Model(inputs=input_img, outputs=encoded, name = 'encoder')\n",
    "    decoder=Model(inputs=decoder_input, outputs=decoded, name = 'decoder')\n",
    "    autoencoder_outputs = decoder(encoder(input_img))\n",
    "    autoencoder= Model(input_img, autoencoder_outputs, name='autoencoder')\n",
    "    autoencoder.summary()\n",
    "    #Optimizers \n",
    "    opt = tf.keras.optimizers.Adam(learning_rate=0.001)\n",
    "    autoencoder.compile(optimizer=opt, loss=\"mse\")\n",
    "\n",
    "    # Set callback functions to early stop training and save the best model so far\n",
    "    early_stopping_monitor = [EarlyStopping(monitor='val_loss', patience=30, verbose=0,restore_best_weights=True)]\n",
    "\n",
    "    history = autoencoder.fit(x_train, x_train, epochs=500, batch_size=40,shuffle=True, \n",
    "                              validation_data=(x_val, x_val), callbacks=early_stopping_monitor)\n",
    "    autoencoder.save(os.path.join('DAE'))\n",
    "    autoencoder=keras.models.load_model(os.path.join('DAE'))\n",
    "\n",
    "    plt.plot(history.history['loss'])\n",
    "    plt.plot(history.history['val_loss'])\n",
    "    plt.title('model loss')\n",
    "    plt.ylabel('loss')\n",
    "    plt.xlabel('epoch')\n",
    "    plt.legend(['train', 'val'], loc='upper left')\n",
    "    plt.show()\n",
    "    print('Train RMSE: ',rmse(x_train,autoencoder.predict(x_train)))\n",
    "    print('Val RMSE: ',rmse(x_val,autoencoder.predict(x_val)))\n",
    "    print('Test RMSE: ',rmse(x_test,autoencoder.predict(x_test)))\n",
    "    \n",
    "    encoder = autoencoder.get_layer(\"encoder\")\n",
    "    encoded_train=encoder.predict(train_full_values)#.reshape(train_full_values.shape[0],train_full_values.shape[2])\n",
    "    encoded_test=encoder.predict(test_full_values)#.reshape(test_full_values.shape[0],test_full_values.shape[2])\n",
    "    encoded_name_list=['encoded_{}'.format(i) for i in range(1,encoded_train.shape[1]+1)]\n",
    "    upper_index=0\n",
    "    bottom_index=0\n",
    "    for i in range(len(train_df_list)):\n",
    "        upper_index=bottom_index\n",
    "        bottom_index=upper_index+train_df_list[i].shape[0]\n",
    "        train_df_list[i]=pd.DataFrame(np.concatenate((encoded_train[upper_index:bottom_index,:],train_df_list[i][label_list].values),axis=1),\n",
    "                                     columns=encoded_name_list+label_list)\n",
    "    upper_index=0\n",
    "    bottom_index=0\n",
    "    for i in range(len(test_df_list)):\n",
    "        upper_index=bottom_index\n",
    "        bottom_index=upper_index+test_df_list[i].shape[0]\n",
    "        test_df_list[i]=pd.DataFrame(np.concatenate((encoded_test[upper_index:bottom_index,:],test_df_list[i][label_list].values),axis=1),\n",
    "                                     columns=encoded_name_list+label_list)\n",
    "    print('Train df(1st one) shape after AE: ', train_df_list[0].shape)\n",
    "    print('Test df(1st one) shape after AE: ', test_df_list[0].shape)\n",
    "    return train_df_list,test_df_list\n",
    "\n",
    "def getSampleSet(df_list, callback, train_feature_list, label_list):\n",
    "    conc_label_list=[]\n",
    "    for i in range(len(df_list)):\n",
    "        # The 1st element in label list is label\n",
    "        nset = df_list[i].shape[0]-callback\n",
    "        print(\"Number of set: \", nset)\n",
    "        conc_input=np.zeros((df_list[i].shape[0],6)) # Only the true reading is non-zero\n",
    "        conc_set = df_list[i][label_list[1:]].values # Excl. 1st column, label column\n",
    "        conc_set[np.isnan(conc_set)]=0\n",
    "        parameter_set = df_list[i][train_feature_list].values\n",
    "        for k in range(conc_input.shape[0]): # Only the true reading is non-zero, default for onehot\n",
    "            conc_input[k][int(df_list[i]['label'][k])-1]=1        \n",
    "        conc_input[np.isnan(conc_input)]=0\n",
    "        parameter_set=np.hstack((parameter_set,conc_input))\n",
    "        sample_set=np.zeros((nset,callback+1,parameter_set.shape[1]))\n",
    "        label_set=np.zeros((nset,1,conc_set.shape[1]))\n",
    "        for j in range(nset):\n",
    "            sample_set[j]=parameter_set[0+j:callback+j+1]\n",
    "            label_set[j]=conc_set[callback+j]\n",
    "        \n",
    "        if (i==0):\n",
    "            total_sample_set=np.zeros((0,callback+1,parameter_set.shape[1]))\n",
    "            total_label_set=np.zeros((0,1,conc_set.shape[1]))\n",
    "        conc_label_list.append(label_set)\n",
    "        total_sample_set = np.vstack([total_sample_set,sample_set])\n",
    "        total_label_set = np.vstack([total_label_set,label_set])\n",
    "        \n",
    "    return total_sample_set, total_label_set, conc_label_list\n",
    "        \n",
    "\n",
    "def getDataSet(path_list,df_list, callback, test_df_index_list, mode, reduced_dimension):\n",
    "    # Pre-processing: avg. out Conc @ sampling point 1\n",
    "    print('Timestamp in selected Test df list: ')\n",
    "    for i in range(len(test_df_index_list)):\n",
    "        print(os.path.basename(path_list[test_df_index_list[i]]))\n",
    "    for i in range(len(df_list)):\n",
    "        df_list[i]=df_list[i].set_index('time')\n",
    "        #df_list[i]=df_list[i].drop(columns=[\"time\"])\n",
    "        df_list[i] = avgOutPoint1(df_list[i])\n",
    "        if (i==0):\n",
    "            origin_feature_num = df_list[i].shape[1]\n",
    "            print(\"Origin input DF feature num(excl time, incl. label): \", origin_feature_num)\n",
    "\n",
    "    # Separate df_list into Train & Test df lists\n",
    "    test_df_list = list(pd.Series(df_list)[test_df_index_list]) # The 140207_1 is selected as the testset\n",
    "    train_df_list = list(pd.Series(df_list)[list(set(range(0,len(df_list),1))-set(test_df_index_list))])\n",
    "\n",
    "    #Min-max scaler, calculated from train_df_list\n",
    "    for i in range(len(train_df_list)):\n",
    "        if (i==0):\n",
    "            tmp_full_values=train_df_list[i].values\n",
    "            tmp_conc_values=train_df_list[i]['AT400(CO2 %)'].values\n",
    "        else:\n",
    "            tmp_full_values=np.concatenate((tmp_full_values, train_df_list[i].values), axis=0)\n",
    "            tmp_conc_values=np.concatenate((tmp_conc_values, train_df_list[i]['AT400(CO2 %)'].values), axis=0)\n",
    "    tmp_full_values = tmp_full_values[:,:-1] # excl. label column as well\n",
    "    general_scaler = MinMaxScaler()\n",
    "    conc_scaler = MinMaxScaler()\n",
    "    general_scaler.fit(tmp_full_values)\n",
    "    conc_scaler.fit(tmp_conc_values.reshape(-1,1))\n",
    "\n",
    "    print('Conc scaler max: ',conc_scaler.data_max_)\n",
    "\n",
    "    # Process into + ['label', '1_sampling', '2_sampling', '3_sampling', '4_sampling', '5_sampling', '6_sampling']\n",
    "    for i in range(len(df_list)):\n",
    "        df_list[i].iloc[:,:-1] = general_scaler.transform(df_list[i].iloc[:,:-1].values)\n",
    "        df_list[i] = columnSeparator(df_list[i])\n",
    "        df_list[i]=df_list[i].fillna(0)\n",
    "\n",
    "    nan_num = 0\n",
    "    for df in df_list:\n",
    "        avail_index = (df['6_sampling']>=0).argmax()\n",
    "        if (nan_num < avail_index):\n",
    "            nan_num = avail_index\n",
    "    print(\"Minimum call back is: \", nan_num+1)\n",
    "    if (nan_num+1<callback):\n",
    "        print('The smallest feasible callback is: ', nan_num+1)\n",
    "\n",
    "    # Separate df_list into Train & Test df lists\n",
    "    test_df_list = list(pd.Series(df_list)[test_df_index_list]) # The 140207_1 is selected as the testset\n",
    "    train_df_list = list(pd.Series(df_list)[list(set(range(0,len(df_list),1))-set(test_df_index_list))])\n",
    "\n",
    "    print(\"train list: \", len(train_df_list))\n",
    "    label_list=['label', '1_sampling', '2_sampling', '3_sampling', '4_sampling', '5_sampling', '6_sampling']\n",
    "    train_feature_list = list(np.sort(list(set(df_list[0].columns)-set(label_list))))\n",
    "    if (mode == 'PCA'):\n",
    "        if (reduced_dimension == 16):\n",
    "            train_feature_list = ['TT113(0C)','TT300(0C)','TT302(0C)','TT202(0C)','TT304(0C)',\"TT400(0C)\",\n",
    "                                  \"TT410(0C)\",'PT103(barg)','PT402(barg)','PT403(barg)','PT102(barg)' ,'AT300(pH)',\n",
    "                                  'FT302(kg/hr)', 'FT303m3/hr','FT304(kg/hr)','FT103(kg/hr)']\n",
    "        else:\n",
    "            train_feature_list = ['TT302(0C)','TT300(0C)','TT401(0C)','TT303(0C)','TT400(0C)','TT113(0C)',\n",
    "                                  'TT301(0C)','TT202(0C)','TT412(0C)','TT309(0C)','TT110a(0C)','TT410(0C)',\n",
    "                                  'TT112(0C)','PT103(barg)','PT402(barg)','PT401(barg)','TT404(0C)', 'TT214(0C)',\n",
    "                                  'FT304(kg/hr)','AT100(pH)','FT303m3/hr','AT300(pH)','FT105(L/min)','FT103(kg/hr)',\n",
    "                                  'PT403(barg)','TT304(0C)','FT301m3/hr','TT107(0C)', 'PT110(barg)','PT111(barg)',\n",
    "                                  'TT210(0C)','TT211(0C)']\n",
    "    if (mode == 'DAE'):\n",
    "        train_df_list, test_df_list=dfAE(train_df_list, test_df_list, train_feature_list, label_list, reduced_dimension)\n",
    "        train_feature_list = np.sort(list(set(train_df_list[0].columns)-set(label_list)))\n",
    "    if (mode == 'POD'):\n",
    "        train_df_list, test_df_list=dfPOD(train_df_list, test_df_list, train_feature_list, label_list, reduced_dimension)\n",
    "        train_feature_list = np.sort(list(set(train_df_list[0].columns)-set(label_list)))\n",
    "    train_sample_set, train_label_set,_ = getSampleSet(train_df_list, callback, train_feature_list, label_list)\n",
    "    test_sample_set, test_label_set,test_conc_list = getSampleSet(test_df_list, callback, train_feature_list, label_list)\n",
    "\n",
    "    \n",
    "    return train_sample_set, train_label_set, test_sample_set, test_label_set, conc_scaler, train_df_list, test_df_list, test_conc_list\n",
    "\n",
    "def GCfunc(i, j, corr_length):\n",
    "    r = abs(i-j)/corr_length\n",
    "    if (r>=0 and r<1):\n",
    "        return (1-pow(r,2)*5/3+pow(r,3)*5/8+pow(r,4)*0.5-pow(r,5)*0.25)\n",
    "    if (r>=1 and r<2):\n",
    "        return (4-5*r+pow(r,2)*5/3+pow(r,3)*5/8-pow(r,4)*0.5+pow(r,5)/12-2/3/r)\n",
    "    else:\n",
    "        return 0\n",
    "\n",
    "def covLoc(B, corr_length):\n",
    "    ls_size=B.shape[0]\n",
    "    for i in range(ls_size):\n",
    "        for j in range(ls_size):\n",
    "            B[i][j] = B[i][j]*GCfunc(i,j,corr_length)\n",
    "    return B\n",
    "\n",
    "def trainSequenceLSTM(state, train_sample_set, train_label_set, test_sample_set, test_label_set):\n",
    "    n_sequence = train_sample_set.shape[1]\n",
    "    n_feature = train_sample_set.shape[2]\n",
    "    n_set = train_sample_set.shape[0]\n",
    "    print('train_sample_set shape: ', train_sample_set.shape)\n",
    "    \n",
    "    train_index = list(set(range(0,n_set,1))- set(range(0,n_set,5)))\n",
    "    test_index = list(set(range(0,n_set,5)))\n",
    "    x_train = train_sample_set[train_index]; x_val = train_sample_set[test_index]\n",
    "    y_train = train_label_set[train_index]; y_val = train_label_set[test_index]\n",
    "    x_test = test_sample_set; y_test = test_label_set\n",
    "    print('Train len: %d, Val len: %d' %(x_train.shape[0], x_val.shape[0]))\n",
    "    print('x_train shape: '+str(x_train.shape)+ ', y_train shape:' + str(y_train.shape))\n",
    "\n",
    "    \n",
    "    if (state == 'train'):\n",
    "        model = Sequential()\n",
    "        model.add(LSTM(100,input_shape=(n_sequence,n_feature), return_sequences=True))\n",
    "        model.add(keras.layers.Dropout(0.1))\n",
    "        model.add(Dense(100))\n",
    "        model.add(keras.layers.Dropout(0.1))\n",
    "        model.add(Dense(100))\n",
    "        model.add(keras.layers.Dropout(0.1))\n",
    "        model.add(Dense(6, activation='sigmoid'))\n",
    "        \n",
    "        #Optimizers \n",
    "        #opt = tf.keras.optimizers.Adam(learning_rate=0.0003)\n",
    "        opt = tf.keras.optimizers.Adam(learning_rate=0.00001)\n",
    "\n",
    "        #model.compile(optimizer=opt, loss=my_loss_fn)\n",
    "        model.compile(optimizer=opt, loss='mse')\n",
    "        model.summary()\n",
    "        # Set callback functions to early stop training and save the best model so far\n",
    "        early_stopping_monitor = [EarlyStopping(monitor='val_loss', patience=70, verbose=0,restore_best_weights=True)]\n",
    "        \n",
    "        # fit model\n",
    "        train_log = model.fit(x_train, y_train, batch_size=40, epochs=500, shuffle=True, \n",
    "                               validation_data=(x_val, y_val), callbacks=early_stopping_monitor)\n",
    "        LSTM_model = model\n",
    "        #Note the label\n",
    "        model.save('LSTM_MSE.h5')\n",
    "\n",
    "        plt.plot(train_log.history['loss'])\n",
    "        plt.plot(train_log.history['val_loss'])\n",
    "        plt.title('model loss')\n",
    "        plt.ylabel('loss')\n",
    "        plt.xlabel('epoch')\n",
    "        plt.legend(['train', 'val'], loc='upper left')\n",
    "        plt.show()\n",
    "    else:\n",
    "        model = keras.models.load_model('LSTM_MSE.h5')\n",
    "    \n",
    "    # Evaluate the model on the test data using `evaluate`\n",
    "    print(\"Evaluate on test data\")\n",
    "    #results = model.evaluate(x_test, y_test, batch_size=36)\n",
    "    predict_set=model.predict(x_test)\n",
    "    results = mean_squared_error(conc_scaler.inverse_transform(np.mean(predict_set,axis=1)),conc_scaler.inverse_transform(y_test.reshape(y_test.shape[0],6)))\n",
    "    print(\"MSE:\", results)\n",
    "    print('RMSE: ', rmse(conc_scaler.inverse_transform(np.mean(predict_set,axis=1)),conc_scaler.inverse_transform(y_test.reshape(y_test.shape[0],6))))\n",
    "    \n",
    "    val_output_set = model.predict(x_val)\n",
    "    \n",
    "    print(\"Evaluate on validation\")\n",
    "    results = mean_squared_error(conc_scaler.inverse_transform(np.mean(val_output_set,axis=1)),conc_scaler.inverse_transform(y_val.reshape(y_val.shape[0],6)))\n",
    "    print(\"MSE:\", results)\n",
    "    print('RMSE: ', rmse(conc_scaler.inverse_transform(np.mean(val_output_set,axis=1)),conc_scaler.inverse_transform(y_val.reshape(y_val.shape[0],6))))\n",
    "    \n",
    "    R=np.cov((conc_scaler.inverse_transform(np.mean(val_output_set,axis=1))- \\\n",
    "              conc_scaler.inverse_transform(y_val.reshape(y_val.shape[0],6))).T)\n",
    "    R=covLoc(R,2)\n",
    "    np.save('R.npy',R)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "f41c502e",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "data/WithLabel\\140120_1.xlsx\n",
      "data/WithLabel\\140206_1.xlsx\n",
      "data/WithLabel\\140207_1.xlsx\n",
      "data/WithLabel\\140207_2.xlsx\n",
      "data/WithLabel\\140214_1.xlsx\n",
      "data/WithLabel\\140214_2.xlsx\n",
      "data/WithLabel\\140227_1.xlsx\n",
      "data/WithLabel\\140313_1.xlsx\n",
      "Timestamp in selected Test df list: \n",
      "140207_1.xlsx\n",
      "Origin input DF feature num(excl time, incl. label):  91\n",
      "Conc scaler max:  [12.03671837]\n",
      "Minimum call back is:  1\n",
      "The smallest feasible callback is:  1\n",
      "train list:  7\n",
      "Train df(1st one) shape before POD:  (117, 97)\n",
      "Test df(1st one) shape before POD:  (118, 97)\n",
      "Truncation parameter:  32\n",
      "Train:  0.00023401074560962606\n",
      "Val:  0.00022780908376132073\n",
      "Test:  0.005939000544029034\n",
      "Train df(1st one) shape after AE:  (117, 39)\n",
      "Test df(1st one) shape after AE:  (118, 39)\n",
      "Number of set:  100\n",
      "Number of set:  33\n",
      "Number of set:  33\n",
      "Number of set:  50\n",
      "Number of set:  33\n",
      "Number of set:  201\n",
      "Number of set:  211\n",
      "Number of set:  101\n"
     ]
    }
   ],
   "source": [
    "reset_random_seeds()\n",
    "ReLu = tf.keras.initializers.HeUniform()\n",
    "data_path_list = glob.glob('data/WithLabel/1*.xlsx')\n",
    "\n",
    "list_of_origin_df=[]\n",
    "for i in range(len(data_path_list)):\n",
    "    xls = pd.ExcelFile(data_path_list[i])\n",
    "    data_df = pd.read_excel(xls, sheet_name=0, index_col=0, header = [0,1])\n",
    "    data_df.columns = data_df.columns.map(''.join)\n",
    "    data_df=data_df.rename_axis('time').reset_index()\n",
    "    tmp_name=list(data_df.columns)\n",
    "    tmp_name[-1]='label'\n",
    "    data_df.columns=tmp_name\n",
    "    list_of_origin_df.append(data_df)\n",
    "\n",
    "for string in data_path_list:\n",
    "    print(string)\n",
    "list_of_processed_df=list_of_origin_df.copy()\n",
    "test_df_index_list=[2] # Select the index of test dataset\n",
    "callback=17 # length of past records excl. current time step\n",
    "# getDataSet(path_list,df_list, callback, test_df_index_list, mode, reduced_dimension)\n",
    "# mode ={'PCA', 'DAE', 'POD'}, reduced_dimension = {'32', '16'}\n",
    "train_sample_set, train_label_set, test_sample_set, test_label_set, conc_scaler,train_df_list,test_df_list,test_conc_list \\\n",
    "=getDataSet(data_path_list,list_of_processed_df, callback,test_df_index_list, 'POD', 32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "d3f2e34e",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train_sample_set shape:  (661, 18, 38)\n",
      "Train len: 528, Val len: 133\n",
      "x_train shape: (528, 18, 38), y_train shape:(528, 1, 6)\n",
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "lstm (LSTM)                  (None, 18, 100)           55600     \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 18, 100)           0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 18, 100)           10100     \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 18, 100)           0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 18, 100)           10100     \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 18, 100)           0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 18, 6)             606       \n",
      "=================================================================\n",
      "Total params: 76,406\n",
      "Trainable params: 76,406\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/500\n",
      "14/14 [==============================] - 5s 49ms/step - loss: 0.1698 - val_loss: 0.1625\n",
      "Epoch 2/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.1621 - val_loss: 0.1549\n",
      "Epoch 3/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.1545 - val_loss: 0.1476\n",
      "Epoch 4/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.1474 - val_loss: 0.1405\n",
      "Epoch 5/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.1403 - val_loss: 0.1337\n",
      "Epoch 6/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.1340 - val_loss: 0.1272\n",
      "Epoch 7/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.1274 - val_loss: 0.1209\n",
      "Epoch 8/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.1215 - val_loss: 0.1148\n",
      "Epoch 9/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.1157 - val_loss: 0.1090\n",
      "Epoch 10/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.1097 - val_loss: 0.1034\n",
      "Epoch 11/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.1042 - val_loss: 0.0980\n",
      "Epoch 12/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0992 - val_loss: 0.0929\n",
      "Epoch 13/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0941 - val_loss: 0.0880\n",
      "Epoch 14/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0894 - val_loss: 0.0833\n",
      "Epoch 15/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0849 - val_loss: 0.0789\n",
      "Epoch 16/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0805 - val_loss: 0.0747\n",
      "Epoch 17/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0762 - val_loss: 0.0707\n",
      "Epoch 18/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0724 - val_loss: 0.0669\n",
      "Epoch 19/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0687 - val_loss: 0.0634\n",
      "Epoch 20/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0651 - val_loss: 0.0601\n",
      "Epoch 21/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0621 - val_loss: 0.0570\n",
      "Epoch 22/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0591 - val_loss: 0.0541\n",
      "Epoch 23/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0564 - val_loss: 0.0514\n",
      "Epoch 24/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0537 - val_loss: 0.0489\n",
      "Epoch 25/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0510 - val_loss: 0.0466\n",
      "Epoch 26/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0489 - val_loss: 0.0444\n",
      "Epoch 27/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0467 - val_loss: 0.0424\n",
      "Epoch 28/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0447 - val_loss: 0.0406\n",
      "Epoch 29/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0429 - val_loss: 0.0389\n",
      "Epoch 30/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0413 - val_loss: 0.0373\n",
      "Epoch 31/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0396 - val_loss: 0.0358\n",
      "Epoch 32/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0380 - val_loss: 0.0344\n",
      "Epoch 33/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0370 - val_loss: 0.0332\n",
      "Epoch 34/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0356 - val_loss: 0.0320\n",
      "Epoch 35/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0342 - val_loss: 0.0309\n",
      "Epoch 36/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0333 - val_loss: 0.0299\n",
      "Epoch 37/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0321 - val_loss: 0.0290\n",
      "Epoch 38/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0314 - val_loss: 0.0281\n",
      "Epoch 39/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0306 - val_loss: 0.0273\n",
      "Epoch 40/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0295 - val_loss: 0.0266\n",
      "Epoch 41/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0289 - val_loss: 0.0259\n",
      "Epoch 42/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0281 - val_loss: 0.0252\n",
      "Epoch 43/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0274 - val_loss: 0.0246\n",
      "Epoch 44/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0269 - val_loss: 0.0240\n",
      "Epoch 45/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0263 - val_loss: 0.0235\n",
      "Epoch 46/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0258 - val_loss: 0.0230\n",
      "Epoch 47/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0251 - val_loss: 0.0225\n",
      "Epoch 48/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0247 - val_loss: 0.0221\n",
      "Epoch 49/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0243 - val_loss: 0.0217\n",
      "Epoch 50/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0238 - val_loss: 0.0212\n",
      "Epoch 51/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0234 - val_loss: 0.0209\n",
      "Epoch 52/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0230 - val_loss: 0.0205\n",
      "Epoch 53/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0226 - val_loss: 0.0201\n",
      "Epoch 54/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0223 - val_loss: 0.0198\n",
      "Epoch 55/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0219 - val_loss: 0.0195\n",
      "Epoch 56/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0216 - val_loss: 0.0192\n",
      "Epoch 57/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0211 - val_loss: 0.0189\n",
      "Epoch 58/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0210 - val_loss: 0.0186\n",
      "Epoch 59/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0207 - val_loss: 0.0184\n",
      "Epoch 60/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0204 - val_loss: 0.0181\n",
      "Epoch 61/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0201 - val_loss: 0.0179\n",
      "Epoch 62/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0200 - val_loss: 0.0176\n",
      "Epoch 63/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0196 - val_loss: 0.0174\n",
      "Epoch 64/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0194 - val_loss: 0.0172\n",
      "Epoch 65/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0193 - val_loss: 0.0169\n",
      "Epoch 66/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0189 - val_loss: 0.0167\n",
      "Epoch 67/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0187 - val_loss: 0.0165\n",
      "Epoch 68/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0186 - val_loss: 0.0163\n",
      "Epoch 69/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0183 - val_loss: 0.0161\n",
      "Epoch 70/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0181 - val_loss: 0.0159\n",
      "Epoch 71/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0179 - val_loss: 0.0157\n",
      "Epoch 72/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0177 - val_loss: 0.0156\n",
      "Epoch 73/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0175 - val_loss: 0.0154\n",
      "Epoch 74/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0175 - val_loss: 0.0152\n",
      "Epoch 75/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0172 - val_loss: 0.0151\n",
      "Epoch 76/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0170 - val_loss: 0.0149\n",
      "Epoch 77/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0168 - val_loss: 0.0147\n",
      "Epoch 78/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0167 - val_loss: 0.0146\n",
      "Epoch 79/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0166 - val_loss: 0.0144\n",
      "Epoch 80/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0163 - val_loss: 0.0143\n",
      "Epoch 81/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0162 - val_loss: 0.0141\n",
      "Epoch 82/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0161 - val_loss: 0.0140\n",
      "Epoch 83/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0160 - val_loss: 0.0138\n",
      "Epoch 84/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0157 - val_loss: 0.0137\n",
      "Epoch 85/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0157 - val_loss: 0.0135\n",
      "Epoch 86/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0156 - val_loss: 0.0134\n",
      "Epoch 87/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0153 - val_loss: 0.0133\n",
      "Epoch 88/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0152 - val_loss: 0.0131\n",
      "Epoch 89/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0151 - val_loss: 0.0130\n",
      "Epoch 90/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0149 - val_loss: 0.0129\n",
      "Epoch 91/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0147 - val_loss: 0.0127\n",
      "Epoch 92/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0146 - val_loss: 0.0126\n",
      "Epoch 93/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0145 - val_loss: 0.0125\n",
      "Epoch 94/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0145 - val_loss: 0.0123\n",
      "Epoch 95/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0143 - val_loss: 0.0122\n",
      "Epoch 96/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0141 - val_loss: 0.0121\n",
      "Epoch 97/500\n",
      "14/14 [==============================] - 0s 21ms/step - loss: 0.0139 - val_loss: 0.0120\n",
      "Epoch 98/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0140 - val_loss: 0.0119\n",
      "Epoch 99/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0137 - val_loss: 0.0117\n",
      "Epoch 100/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0137 - val_loss: 0.0116\n",
      "Epoch 101/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0136 - val_loss: 0.0115\n",
      "Epoch 102/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0135 - val_loss: 0.0114\n",
      "Epoch 103/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0132 - val_loss: 0.0113\n",
      "Epoch 104/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0132 - val_loss: 0.0111\n",
      "Epoch 105/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0130 - val_loss: 0.0110\n",
      "Epoch 106/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0129 - val_loss: 0.0109\n",
      "Epoch 107/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0128 - val_loss: 0.0108\n",
      "Epoch 108/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0128 - val_loss: 0.0107\n",
      "Epoch 109/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0126 - val_loss: 0.0106\n",
      "Epoch 110/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0126 - val_loss: 0.0105\n",
      "Epoch 111/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0124 - val_loss: 0.0103\n",
      "Epoch 112/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0122 - val_loss: 0.0102\n",
      "Epoch 113/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0121 - val_loss: 0.0101\n",
      "Epoch 114/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0120 - val_loss: 0.0100\n",
      "Epoch 115/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0120 - val_loss: 0.0099\n",
      "Epoch 116/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0119 - val_loss: 0.0098\n",
      "Epoch 117/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0118 - val_loss: 0.0097\n",
      "Epoch 118/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0117 - val_loss: 0.0096\n",
      "Epoch 119/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0116 - val_loss: 0.0095\n",
      "Epoch 120/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0115 - val_loss: 0.0094\n",
      "Epoch 121/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0114 - val_loss: 0.0093\n",
      "Epoch 122/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0112 - val_loss: 0.0092\n",
      "Epoch 123/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0111 - val_loss: 0.0091\n",
      "Epoch 124/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0110 - val_loss: 0.0090\n",
      "Epoch 125/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0110 - val_loss: 0.0089\n",
      "Epoch 126/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0108 - val_loss: 0.0088\n",
      "Epoch 127/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0107 - val_loss: 0.0087\n",
      "Epoch 128/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0106 - val_loss: 0.0086\n",
      "Epoch 129/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0105 - val_loss: 0.0085\n",
      "Epoch 130/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0104 - val_loss: 0.0084\n",
      "Epoch 131/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0102 - val_loss: 0.0083\n",
      "Epoch 132/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0102 - val_loss: 0.0082\n",
      "Epoch 133/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0102 - val_loss: 0.0081\n",
      "Epoch 134/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0100 - val_loss: 0.0080\n",
      "Epoch 135/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0099 - val_loss: 0.0079\n",
      "Epoch 136/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0098 - val_loss: 0.0078\n",
      "Epoch 137/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0097 - val_loss: 0.0077\n",
      "Epoch 138/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0096 - val_loss: 0.0076\n",
      "Epoch 139/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0096 - val_loss: 0.0075\n",
      "Epoch 140/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0094 - val_loss: 0.0074\n",
      "Epoch 141/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0094 - val_loss: 0.0074\n",
      "Epoch 142/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0093 - val_loss: 0.0073\n",
      "Epoch 143/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0092 - val_loss: 0.0072\n",
      "Epoch 144/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0092 - val_loss: 0.0071\n",
      "Epoch 145/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0089 - val_loss: 0.0070\n",
      "Epoch 146/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0088 - val_loss: 0.0069\n",
      "Epoch 147/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0088 - val_loss: 0.0069\n",
      "Epoch 148/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0088 - val_loss: 0.0068\n",
      "Epoch 149/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0087 - val_loss: 0.0067\n",
      "Epoch 150/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0086 - val_loss: 0.0066\n",
      "Epoch 151/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0085 - val_loss: 0.0066\n",
      "Epoch 152/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0084 - val_loss: 0.0065\n",
      "Epoch 153/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0083 - val_loss: 0.0064\n",
      "Epoch 154/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0083 - val_loss: 0.0063\n",
      "Epoch 155/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0081 - val_loss: 0.0063\n",
      "Epoch 156/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0081 - val_loss: 0.0062\n",
      "Epoch 157/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0081 - val_loss: 0.0061\n",
      "Epoch 158/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0080 - val_loss: 0.0061\n",
      "Epoch 159/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0079 - val_loss: 0.0060\n",
      "Epoch 160/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0078 - val_loss: 0.0059\n",
      "Epoch 161/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0078 - val_loss: 0.0059\n",
      "Epoch 162/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0077 - val_loss: 0.0058\n",
      "Epoch 163/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0076 - val_loss: 0.0058\n",
      "Epoch 164/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0076 - val_loss: 0.0057\n",
      "Epoch 165/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0075 - val_loss: 0.0056\n",
      "Epoch 166/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0074 - val_loss: 0.0056\n",
      "Epoch 167/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0074 - val_loss: 0.0055\n",
      "Epoch 168/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0073 - val_loss: 0.0055\n",
      "Epoch 169/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0073 - val_loss: 0.0054\n",
      "Epoch 170/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0072 - val_loss: 0.0054\n",
      "Epoch 171/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0072 - val_loss: 0.0053\n",
      "Epoch 172/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0071 - val_loss: 0.0053\n",
      "Epoch 173/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0070 - val_loss: 0.0052\n",
      "Epoch 174/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0070 - val_loss: 0.0052\n",
      "Epoch 175/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0069 - val_loss: 0.0051\n",
      "Epoch 176/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0069 - val_loss: 0.0051\n",
      "Epoch 177/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0069 - val_loss: 0.0050\n",
      "Epoch 178/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0068 - val_loss: 0.0050\n",
      "Epoch 179/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0067 - val_loss: 0.0050\n",
      "Epoch 180/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0067 - val_loss: 0.0049\n",
      "Epoch 181/500\n",
      "14/14 [==============================] - 0s 11ms/step - loss: 0.0066 - val_loss: 0.0049\n",
      "Epoch 182/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0067 - val_loss: 0.0048\n",
      "Epoch 183/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0066 - val_loss: 0.0048\n",
      "Epoch 184/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0066 - val_loss: 0.0047\n",
      "Epoch 185/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0065 - val_loss: 0.0047\n",
      "Epoch 186/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0065 - val_loss: 0.0047\n",
      "Epoch 187/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0064 - val_loss: 0.0046\n",
      "Epoch 188/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0063 - val_loss: 0.0046\n",
      "Epoch 189/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0064 - val_loss: 0.0046\n",
      "Epoch 190/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0062 - val_loss: 0.0045\n",
      "Epoch 191/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0063 - val_loss: 0.0045\n",
      "Epoch 192/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0063 - val_loss: 0.0045\n",
      "Epoch 193/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0062 - val_loss: 0.0044\n",
      "Epoch 194/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0062 - val_loss: 0.0044\n",
      "Epoch 195/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0060 - val_loss: 0.0044\n",
      "Epoch 196/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0061 - val_loss: 0.0043\n",
      "Epoch 197/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0060 - val_loss: 0.0043\n",
      "Epoch 198/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0060 - val_loss: 0.0042\n",
      "Epoch 199/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0059 - val_loss: 0.0042\n",
      "Epoch 200/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0059 - val_loss: 0.0042\n",
      "Epoch 201/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0059 - val_loss: 0.0042\n",
      "Epoch 202/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0058 - val_loss: 0.0041\n",
      "Epoch 203/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0058 - val_loss: 0.0041\n",
      "Epoch 204/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0058 - val_loss: 0.0041\n",
      "Epoch 205/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0058 - val_loss: 0.0040\n",
      "Epoch 206/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0057 - val_loss: 0.0040\n",
      "Epoch 207/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0056 - val_loss: 0.0040\n",
      "Epoch 208/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0056 - val_loss: 0.0040\n",
      "Epoch 209/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0056 - val_loss: 0.0039\n",
      "Epoch 210/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0056 - val_loss: 0.0039\n",
      "Epoch 211/500\n",
      "14/14 [==============================] - 0s 25ms/step - loss: 0.0056 - val_loss: 0.0039\n",
      "Epoch 212/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0054 - val_loss: 0.0038\n",
      "Epoch 213/500\n",
      "14/14 [==============================] - 0s 11ms/step - loss: 0.0055 - val_loss: 0.0038\n",
      "Epoch 214/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0055 - val_loss: 0.0038\n",
      "Epoch 215/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0055 - val_loss: 0.0038\n",
      "Epoch 216/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0054 - val_loss: 0.0037\n",
      "Epoch 217/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0053 - val_loss: 0.0037\n",
      "Epoch 218/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0053 - val_loss: 0.0037\n",
      "Epoch 219/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0054 - val_loss: 0.0037\n",
      "Epoch 220/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0052 - val_loss: 0.0036\n",
      "Epoch 221/500\n",
      "14/14 [==============================] - 0s 21ms/step - loss: 0.0053 - val_loss: 0.0036\n",
      "Epoch 222/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0053 - val_loss: 0.0036\n",
      "Epoch 223/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0052 - val_loss: 0.0036\n",
      "Epoch 224/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0052 - val_loss: 0.0035\n",
      "Epoch 225/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0052 - val_loss: 0.0035\n",
      "Epoch 226/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0051 - val_loss: 0.0035\n",
      "Epoch 227/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0051 - val_loss: 0.0035\n",
      "Epoch 228/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0051 - val_loss: 0.0035\n",
      "Epoch 229/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0051 - val_loss: 0.0034\n",
      "Epoch 230/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0050 - val_loss: 0.0034\n",
      "Epoch 231/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0050 - val_loss: 0.0034\n",
      "Epoch 232/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0050 - val_loss: 0.0034\n",
      "Epoch 233/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0050 - val_loss: 0.0033\n",
      "Epoch 234/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0049 - val_loss: 0.0033\n",
      "Epoch 235/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0049 - val_loss: 0.0033\n",
      "Epoch 236/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0049 - val_loss: 0.0033\n",
      "Epoch 237/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0048 - val_loss: 0.0033\n",
      "Epoch 238/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0048 - val_loss: 0.0032\n",
      "Epoch 239/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0049 - val_loss: 0.0032\n",
      "Epoch 240/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0048 - val_loss: 0.0032\n",
      "Epoch 241/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0048 - val_loss: 0.0032\n",
      "Epoch 242/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0047 - val_loss: 0.0032\n",
      "Epoch 243/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0047 - val_loss: 0.0031\n",
      "Epoch 244/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0047 - val_loss: 0.0031\n",
      "Epoch 245/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0047 - val_loss: 0.0031\n",
      "Epoch 246/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0047 - val_loss: 0.0031\n",
      "Epoch 247/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0046 - val_loss: 0.0031\n",
      "Epoch 248/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0046 - val_loss: 0.0031\n",
      "Epoch 249/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0046 - val_loss: 0.0030\n",
      "Epoch 250/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0046 - val_loss: 0.0030\n",
      "Epoch 251/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0045 - val_loss: 0.0030\n",
      "Epoch 252/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0045 - val_loss: 0.0030\n",
      "Epoch 253/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0045 - val_loss: 0.0030\n",
      "Epoch 254/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0044 - val_loss: 0.0029\n",
      "Epoch 255/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0045 - val_loss: 0.0029\n",
      "Epoch 256/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0045 - val_loss: 0.0029\n",
      "Epoch 257/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0044 - val_loss: 0.0029\n",
      "Epoch 258/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0044 - val_loss: 0.0029\n",
      "Epoch 259/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0044 - val_loss: 0.0029\n",
      "Epoch 260/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0044 - val_loss: 0.0028\n",
      "Epoch 261/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0043 - val_loss: 0.0028\n",
      "Epoch 262/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0043 - val_loss: 0.0028\n",
      "Epoch 263/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0043 - val_loss: 0.0028\n",
      "Epoch 264/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0043 - val_loss: 0.0028\n",
      "Epoch 265/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0042 - val_loss: 0.0028\n",
      "Epoch 266/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0043 - val_loss: 0.0027\n",
      "Epoch 267/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0042 - val_loss: 0.0027\n",
      "Epoch 268/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0042 - val_loss: 0.0027\n",
      "Epoch 269/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0042 - val_loss: 0.0027\n",
      "Epoch 270/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0042 - val_loss: 0.0027\n",
      "Epoch 271/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0042 - val_loss: 0.0027\n",
      "Epoch 272/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0041 - val_loss: 0.0027\n",
      "Epoch 273/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0041 - val_loss: 0.0026\n",
      "Epoch 274/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0041 - val_loss: 0.0026\n",
      "Epoch 275/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0041 - val_loss: 0.0026\n",
      "Epoch 276/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0041 - val_loss: 0.0026\n",
      "Epoch 277/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0040 - val_loss: 0.0026\n",
      "Epoch 278/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0040 - val_loss: 0.0026\n",
      "Epoch 279/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0040 - val_loss: 0.0025\n",
      "Epoch 280/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0040 - val_loss: 0.0025\n",
      "Epoch 281/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0039 - val_loss: 0.0025\n",
      "Epoch 282/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0039 - val_loss: 0.0025\n",
      "Epoch 283/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0039 - val_loss: 0.0025\n",
      "Epoch 284/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0039 - val_loss: 0.0025\n",
      "Epoch 285/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0039 - val_loss: 0.0025\n",
      "Epoch 286/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0039 - val_loss: 0.0024\n",
      "Epoch 287/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0038 - val_loss: 0.0024\n",
      "Epoch 288/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0039 - val_loss: 0.0024\n",
      "Epoch 289/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0039 - val_loss: 0.0024\n",
      "Epoch 290/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0038 - val_loss: 0.0024\n",
      "Epoch 291/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0038 - val_loss: 0.0024\n",
      "Epoch 292/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0038 - val_loss: 0.0024\n",
      "Epoch 293/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0037 - val_loss: 0.0024\n",
      "Epoch 294/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0038 - val_loss: 0.0023\n",
      "Epoch 295/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0037 - val_loss: 0.0023\n",
      "Epoch 296/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0037 - val_loss: 0.0023\n",
      "Epoch 297/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0037 - val_loss: 0.0023\n",
      "Epoch 298/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0037 - val_loss: 0.0023\n",
      "Epoch 299/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0036 - val_loss: 0.0023\n",
      "Epoch 300/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0036 - val_loss: 0.0023\n",
      "Epoch 301/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0037 - val_loss: 0.0023\n",
      "Epoch 302/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0036 - val_loss: 0.0023\n",
      "Epoch 303/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0036 - val_loss: 0.0022\n",
      "Epoch 304/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0036 - val_loss: 0.0022\n",
      "Epoch 305/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0036 - val_loss: 0.0022\n",
      "Epoch 306/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0036 - val_loss: 0.0022\n",
      "Epoch 307/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0035 - val_loss: 0.0022\n",
      "Epoch 308/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0035 - val_loss: 0.0022\n",
      "Epoch 309/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0035 - val_loss: 0.0022\n",
      "Epoch 310/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0035 - val_loss: 0.0022\n",
      "Epoch 311/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0035 - val_loss: 0.0021\n",
      "Epoch 312/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0034 - val_loss: 0.0021\n",
      "Epoch 313/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0035 - val_loss: 0.0021\n",
      "Epoch 314/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0034 - val_loss: 0.0021\n",
      "Epoch 315/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0034 - val_loss: 0.0021\n",
      "Epoch 316/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0034 - val_loss: 0.0021\n",
      "Epoch 317/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0034 - val_loss: 0.0021\n",
      "Epoch 318/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0034 - val_loss: 0.0021\n",
      "Epoch 319/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0034 - val_loss: 0.0021\n",
      "Epoch 320/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0033 - val_loss: 0.0020\n",
      "Epoch 321/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0033 - val_loss: 0.0020\n",
      "Epoch 322/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0033 - val_loss: 0.0020\n",
      "Epoch 323/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0033 - val_loss: 0.0020\n",
      "Epoch 324/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0033 - val_loss: 0.0020\n",
      "Epoch 325/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0033 - val_loss: 0.0020\n",
      "Epoch 326/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0033 - val_loss: 0.0020\n",
      "Epoch 327/500\n",
      "14/14 [==============================] - 0s 21ms/step - loss: 0.0033 - val_loss: 0.0020\n",
      "Epoch 328/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0033 - val_loss: 0.0020\n",
      "Epoch 329/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0032 - val_loss: 0.0020\n",
      "Epoch 330/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0032 - val_loss: 0.0019\n",
      "Epoch 331/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0032 - val_loss: 0.0019\n",
      "Epoch 332/500\n",
      "14/14 [==============================] - 0s 22ms/step - loss: 0.0031 - val_loss: 0.0019\n",
      "Epoch 333/500\n",
      "14/14 [==============================] - 0s 22ms/step - loss: 0.0032 - val_loss: 0.0019\n",
      "Epoch 334/500\n",
      "14/14 [==============================] - 0s 21ms/step - loss: 0.0031 - val_loss: 0.0019\n",
      "Epoch 335/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0031 - val_loss: 0.0019\n",
      "Epoch 336/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0031 - val_loss: 0.0019\n",
      "Epoch 337/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0032 - val_loss: 0.0019\n",
      "Epoch 338/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0031 - val_loss: 0.0019\n",
      "Epoch 339/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0031 - val_loss: 0.0019\n",
      "Epoch 340/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0031 - val_loss: 0.0019\n",
      "Epoch 341/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0031 - val_loss: 0.0018\n",
      "Epoch 342/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0031 - val_loss: 0.0018\n",
      "Epoch 343/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0031 - val_loss: 0.0018\n",
      "Epoch 344/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0030 - val_loss: 0.0018\n",
      "Epoch 345/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0031 - val_loss: 0.0018\n",
      "Epoch 346/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0030 - val_loss: 0.0018\n",
      "Epoch 347/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0030 - val_loss: 0.0018\n",
      "Epoch 348/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0030 - val_loss: 0.0018\n",
      "Epoch 349/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0030 - val_loss: 0.0018\n",
      "Epoch 350/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0030 - val_loss: 0.0018\n",
      "Epoch 351/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0030 - val_loss: 0.0018\n",
      "Epoch 352/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0030 - val_loss: 0.0018\n",
      "Epoch 353/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0030 - val_loss: 0.0017\n",
      "Epoch 354/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0029 - val_loss: 0.0017\n",
      "Epoch 355/500\n",
      "14/14 [==============================] - 0s 21ms/step - loss: 0.0030 - val_loss: 0.0017\n",
      "Epoch 356/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0029 - val_loss: 0.0017\n",
      "Epoch 357/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0029 - val_loss: 0.0017\n",
      "Epoch 358/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0028 - val_loss: 0.0017\n",
      "Epoch 359/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0029 - val_loss: 0.0017\n",
      "Epoch 360/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0028 - val_loss: 0.0017\n",
      "Epoch 361/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0028 - val_loss: 0.0017\n",
      "Epoch 362/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0029 - val_loss: 0.0017\n",
      "Epoch 363/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0028 - val_loss: 0.0017\n",
      "Epoch 364/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0028 - val_loss: 0.0017\n",
      "Epoch 365/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0028 - val_loss: 0.0016\n",
      "Epoch 366/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0028 - val_loss: 0.0016\n",
      "Epoch 367/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0028 - val_loss: 0.0016\n",
      "Epoch 368/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0028 - val_loss: 0.0016\n",
      "Epoch 369/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0028 - val_loss: 0.0016\n",
      "Epoch 370/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0027 - val_loss: 0.0016\n",
      "Epoch 371/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0028 - val_loss: 0.0016\n",
      "Epoch 372/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0027 - val_loss: 0.0016\n",
      "Epoch 373/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0027 - val_loss: 0.0016\n",
      "Epoch 374/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0027 - val_loss: 0.0016\n",
      "Epoch 375/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0027 - val_loss: 0.0016\n",
      "Epoch 376/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0027 - val_loss: 0.0016\n",
      "Epoch 377/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0027 - val_loss: 0.0016\n",
      "Epoch 378/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0026 - val_loss: 0.0016\n",
      "Epoch 379/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 380/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0027 - val_loss: 0.0015\n",
      "Epoch 381/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 382/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 383/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 384/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 385/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 386/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 387/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 388/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 389/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 390/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0026 - val_loss: 0.0015\n",
      "Epoch 391/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0025 - val_loss: 0.0015\n",
      "Epoch 392/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0025 - val_loss: 0.0015\n",
      "Epoch 393/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0025 - val_loss: 0.0015\n",
      "Epoch 394/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0026 - val_loss: 0.0014\n",
      "Epoch 395/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 396/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 397/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 398/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 399/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 400/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 401/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 402/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 403/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 404/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0025 - val_loss: 0.0014\n",
      "Epoch 405/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 406/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 407/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 408/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 409/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 410/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 411/500\n",
      "14/14 [==============================] - 0s 12ms/step - loss: 0.0024 - val_loss: 0.0014\n",
      "Epoch 412/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0024 - val_loss: 0.0013\n",
      "Epoch 413/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 414/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0024 - val_loss: 0.0013\n",
      "Epoch 415/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 416/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 417/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 418/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 419/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 420/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 421/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 422/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 423/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 424/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 425/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 426/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 427/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0022 - val_loss: 0.0013\n",
      "Epoch 428/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0022 - val_loss: 0.0013\n",
      "Epoch 429/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0023 - val_loss: 0.0013\n",
      "Epoch 430/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0023 - val_loss: 0.0012\n",
      "Epoch 431/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 432/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 433/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 434/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 435/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 436/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 437/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 438/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 439/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 440/500\n",
      "14/14 [==============================] - 0s 21ms/step - loss: 0.0021 - val_loss: 0.0012\n",
      "Epoch 441/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0021 - val_loss: 0.0012\n",
      "Epoch 442/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 443/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 444/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0021 - val_loss: 0.0012\n",
      "Epoch 445/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0022 - val_loss: 0.0012\n",
      "Epoch 446/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0021 - val_loss: 0.0012\n",
      "Epoch 447/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0021 - val_loss: 0.0012\n",
      "Epoch 448/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0021 - val_loss: 0.0012\n",
      "Epoch 449/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0021 - val_loss: 0.0012\n",
      "Epoch 450/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0021 - val_loss: 0.0012\n",
      "Epoch 451/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0021 - val_loss: 0.0011\n",
      "Epoch 452/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0021 - val_loss: 0.0011\n",
      "Epoch 453/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0021 - val_loss: 0.0011\n",
      "Epoch 454/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0021 - val_loss: 0.0011\n",
      "Epoch 455/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 456/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0021 - val_loss: 0.0011\n",
      "Epoch 457/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0021 - val_loss: 0.0011\n",
      "Epoch 458/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 459/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 460/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 461/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 462/500\n",
      "14/14 [==============================] - 0s 22ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 463/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 464/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 465/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 466/500\n",
      "14/14 [==============================] - 0s 19ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 467/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 468/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 469/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 470/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 471/500\n",
      "14/14 [==============================] - 0s 13ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 472/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 473/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0019 - val_loss: 0.0011\n",
      "Epoch 474/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0019 - val_loss: 0.0011\n",
      "Epoch 475/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0020 - val_loss: 0.0011\n",
      "Epoch 476/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 477/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 478/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 479/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 480/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 481/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 482/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 483/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 484/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 485/500\n",
      "14/14 [==============================] - 0s 14ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 486/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 487/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 488/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 489/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0019 - val_loss: 0.0010\n",
      "Epoch 490/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0019 - val_loss: 9.9585e-04\n",
      "Epoch 491/500\n",
      "14/14 [==============================] - 0s 15ms/step - loss: 0.0018 - val_loss: 9.9673e-04\n",
      "Epoch 492/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0018 - val_loss: 9.9210e-04\n",
      "Epoch 493/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0019 - val_loss: 9.8452e-04\n",
      "Epoch 494/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0018 - val_loss: 9.8594e-04\n",
      "Epoch 495/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0018 - val_loss: 9.8445e-04\n",
      "Epoch 496/500\n",
      "14/14 [==============================] - 0s 16ms/step - loss: 0.0018 - val_loss: 9.7309e-04\n",
      "Epoch 497/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0018 - val_loss: 9.7018e-04\n",
      "Epoch 498/500\n",
      "14/14 [==============================] - 0s 18ms/step - loss: 0.0018 - val_loss: 9.7262e-04\n",
      "Epoch 499/500\n",
      "14/14 [==============================] - 0s 20ms/step - loss: 0.0018 - val_loss: 9.7227e-04\n",
      "Epoch 500/500\n",
      "14/14 [==============================] - 0s 17ms/step - loss: 0.0018 - val_loss: 9.6791e-04\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Evaluate on test data\n",
      "MSE: 0.04703987324763109\n",
      "RMSE:  0.21688677517919597\n",
      "Evaluate on validation\n",
      "MSE: 0.06301061585842911\n",
      "RMSE:  0.2510191543656163\n"
     ]
    }
   ],
   "source": [
    "#LSTM training\n",
    "trainSequenceLSTM('train', train_sample_set, train_label_set, test_sample_set, test_label_set)\n",
    "model = keras.models.load_model('LSTM_MSE.h5')\n",
    "#np.save('results/set2_DAE_16.npy',conc_scaler.inverse_transform(np.mean(predict_set,axis=1)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b073f5ab",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'CO$_2$ composition(%)')"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "predict_set=np.load('results/set1_DAE_16.npy')\n",
    "plt.plot(conc_scaler.inverse_transform(test_label_set.reshape(test_label_set.shape[0],6))[:,5], color='tab:blue')\n",
    "plt.plot(predict_set[:,5],linestyle='dashed',color='tab:blue')\n",
    "plt.plot(conc_scaler.inverse_transform(test_label_set.reshape(test_label_set.shape[0],6))[:,4], color='tab:red')\n",
    "plt.plot(predict_set[:,4],linestyle='dashed',color='tab:red')\n",
    "plt.title('Data-driven model')\n",
    "plt.xlabel('time')\n",
    "plt.ylabel('CO$_2$ composition(%)')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "76aed33f",
   "metadata": {},
   "source": [
    "Comparison between different dimensionality reduction methods"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "f7ee082e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 864x288 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "DAE_result = np.load('results/set1_DAE_16.npy')\n",
    "PCA_result = np.load('results/set1_PCA_16.npy')\n",
    "POD_result = np.load('results/set1_POD_16.npy')\n",
    "discrete_record = list_of_origin_df[2]['AT400(CO2 %)'][17:]\n",
    "sample45 = np.where(discrete_record >0.5,discrete_record,None)\n",
    "fig, axs = plt.subplots(1,3, sharey=True, figsize=(12,4))\n",
    "plt.subplots_adjust(wspace = 0.05, bottom=0.15)\n",
    "fig.set_tight_layout(True)\n",
    "axs[0].scatter(range(PCA_result.shape[0]), sample45, marker='x', color='black', label='Measurements')\n",
    "axs[0].plot(PCA_result[:,5],linestyle='dashed',color='tab:blue',label='Estimation @ point 6')\n",
    "axs[0].plot(PCA_result[:,4],linestyle='dashed',color='tab:red',label='Estimation @ point 5')\n",
    "axs[0].set_title('PCA')\n",
    "axs[0].set_ylabel('CO$_2$ composition(%)')\n",
    "axs[0].set_xlabel('Time')\n",
    "axs[0].legend()\n",
    "axs[1].scatter(range(DAE_result.shape[0]), sample45, marker='x', color='black')\n",
    "axs[1].plot(DAE_result[:,5],linestyle='dashed',color='tab:blue')\n",
    "axs[1].plot(DAE_result[:,4],linestyle='dashed',color='tab:red')\n",
    "axs[1].set_title('DAE')\n",
    "axs[1].set_xlabel('Time')\n",
    "axs[2].scatter(range(POD_result.shape[0]), sample45, marker='x', color='black')\n",
    "axs[2].plot(POD_result[:,5],linestyle='dashed',color='tab:blue',label='Fused @ 6')\n",
    "axs[2].plot(POD_result[:,4],linestyle='dashed',color='tab:red',label='Fused @ 5')\n",
    "axs[2].set_title('POD')\n",
    "axs[2].set_xlabel('Time')\n",
    "#plt.tight_layout()\n",
    "plt.savefig('reduction_compare.png',dpi=1000)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dbb10e5f",
   "metadata": {},
   "source": [
    "# Model fusion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "ebb92183",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "kinetic_model\\140207_1.csv  in test\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fused RMSE:  0.10850019921956107\n"
     ]
    }
   ],
   "source": [
    "dynamic_path_list = glob.glob('kinetic_model/1*.csv')\n",
    "def VAR_3D(xb,Y,H,B,R): #booleen=0 garde la trace\n",
    "    # xb: priori, Y: observation, H: obs. matrix, B: priori estimate uncertainty\n",
    "    # R: Measurement uncertainty\n",
    "    dim_x = xb.size\n",
    "    #dim_y = Y.size\n",
    "    Y.shape = (Y.size,1)\n",
    "    xb1=np.copy(xb)\n",
    "    xb1.shape=(xb1.size,1)\n",
    "    K=np.dot(B,np.dot(np.transpose(H),np.linalg.pinv(np.dot(H,np.dot(B,np.transpose(H)))+R))) #matrice de gain\n",
    "    \n",
    "    A=np.dot(np.dot((np.eye(dim_x)-np.dot(K,H)),B),np.transpose((np.eye(dim_x)-np.dot(K,H))))+np.dot(np.dot(K,R),np.transpose(K))\n",
    "    vect=np.dot(H,xb1)\n",
    "    xa=np.copy(xb1+np.dot(K,(Y-vect)))\n",
    "    return xa.ravel(),A\n",
    "\n",
    "test_name_list=[]\n",
    "for i in range(len(test_df_index_list)):\n",
    "    test_name_list.append(os.path.splitext(os.path.basename(data_path_list[test_df_index_list[i]]))[0])\n",
    "    \n",
    "dynamic_df_list=[]\n",
    "dynamic_train_list=[]\n",
    "dynamic_test_list=[]\n",
    "for path in dynamic_path_list:\n",
    "    dynamic_df_list.append(pd.read_csv(path, header=None))\n",
    "    if os.path.splitext(os.path.basename(path))[0] not in test_name_list:\n",
    "        dynamic_train_list.append(pd.read_csv(path, header=None))\n",
    "    else:\n",
    "        print(path, ' in test')\n",
    "        dynamic_test_list.append(pd.read_csv(path, header=None))\n",
    "\n",
    "for i in range(len(dynamic_train_list)):\n",
    "    if (i==0):\n",
    "        train_full_values=dynamic_train_list[i].values[callback:,:]\n",
    "    else:\n",
    "        train_full_values=np.concatenate((train_full_values, dynamic_train_list[i].values[callback:,:]), axis=0)\n",
    "B=np.cov((train_full_values*100-conc_scaler.inverse_transform(train_label_set.reshape(train_label_set.shape[0],6))).T)\n",
    "B=covLoc(B,2)\n",
    "np.save('B.npy',B)\n",
    "\n",
    "for i in range(len(dynamic_test_list)):\n",
    "    if (i==0):\n",
    "        test_full_values=dynamic_test_list[i].values[callback:,:]\n",
    "    else:\n",
    "        test_full_values=np.concatenate((test_full_values, dynamic_test_list[i].values[callback:,:]), axis=0)\n",
    "\n",
    "R=np.load('R.npy')\n",
    "B=np.load('B.npy')\n",
    "R+=1E-10\n",
    "B+=1E-10\n",
    "max_val = np.max([np.max(B),np.max(R)])\n",
    "fig,axs=plt.subplots(nrows=1, ncols=2,sharey=True)\n",
    "#im1=axs[0].imshow(B,norm=LogNorm())\n",
    "#im2=axs[1].imshow(R,norm=LogNorm())\n",
    "color='Greys'\n",
    "im1=axs[0].imshow(B,extent=[0,6,0,6], cmap=color,norm=LogNorm(vmin=1E-10, vmax=max_val))\n",
    "axs[0].set_title('B')\n",
    "#axs[0].text(5,0.25,str(B[-1,-1])[:4],color='white',size='small')\n",
    "#axs[0].text(4,1.25,str(B[-2,-2])[:4],color='white',size='small')\n",
    "for i in range(6):\n",
    "    if (B[i,i]<0.01):\n",
    "        axs[0].text(i,5.25-i,'{:.1E}'.format(B[i,i]),color='white',size='small')       \n",
    "    else:\n",
    "        axs[0].text(i,5.25-i,str(B[i,i])[:4],color='white',size='small')\n",
    "im2=axs[1].imshow(R,extent=[0,6,0,6], cmap = color,norm=LogNorm(vmin=1E-10, vmax=max_val))\n",
    "axs[1].set_title('R')\n",
    "for i in range(6):\n",
    "    if i == 2:\n",
    "        axs[1].text(i-0.6,5.25-i,'{:.1E}'.format(R[i,i]),color='white',size='small')       \n",
    "        continue\n",
    "    if (R[i,i]<0.01):\n",
    "        axs[1].text(i,5.25-i,'{:.1E}'.format(R[i,i]),color='white',size='small')       \n",
    "    else:\n",
    "        axs[1].text(i,5.25-i,str(R[i,i])[:4],color='white',size='small')\n",
    "axs[0].set_xticks(np.arange(0.5, 6.5, step=1))\n",
    "axs[1].set_xticks(np.arange(0.5, 6.5, step=1))\n",
    "axs[0].set_xticklabels(['1','2','3','4','5','6'])\n",
    "axs[1].set_xticklabels(['1','2','3','4','5','6'])\n",
    "axs[0].set_yticks(np.arange(0.5, 6.5, step=1))\n",
    "axs[0].set_yticklabels(['6','5','4','3','2','1'])\n",
    "cb=fig.colorbar(im1,ax=axs.ravel().tolist())\n",
    "#plt.savefig('cov.eps', format=\"eps\", dpi=1000)\n",
    "plt.show()\n",
    "\n",
    "H=np.eye(6)\n",
    "conc_set=test_full_values.copy()\n",
    "for i in range(conc_set.shape[0]):\n",
    "    conc_set[i],_ = VAR_3D(test_full_values[i]*100,predict_set[i],H,B,R)\n",
    "print('Fused RMSE: ', rmse(conc_set,conc_scaler.inverse_transform(test_label_set.reshape(test_label_set.shape[0],test_label_set.shape[2]))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "6b369c9f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1c2464d1100>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 864x288 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "discrete_record = list_of_origin_df[2]['AT400(CO2 %)'][callback:]\n",
    "sample45 = np.where(discrete_record >0.5,discrete_record,None)\n",
    "ssae_set_6 = np.load('results/set1_6.npy')\n",
    "ssae_set_5 = np.load('results/set1_5.npy')\n",
    "fig, axs = plt.subplots(1,3, sharey=True, figsize=(12,4))\n",
    "#plt.ylim(-0, 4.5)\n",
    "#axs[0].plot(conc_scaler.inverse_transform(test_label_set.reshape(test_label_set.shape[0],6)))\n",
    "plt.subplots_adjust(wspace = 0.05, bottom=0.15)\n",
    "fig.set_tight_layout(True)\n",
    "axs[0].scatter(range(predict_set.shape[0]), sample45, marker='x', color='black', label='Measurements')\n",
    "axs[0].plot(test_full_values[:,5]*100,linestyle='dashed',color='tab:blue',label='Estimation @ point 6')\n",
    "axs[0].plot(test_full_values[:,4]*100,linestyle='dashed',color='tab:red',label='Estimation @ point 5')\n",
    "axs[0].set_title('Kinetic model')\n",
    "axs[0].set_ylabel('CO$_2$ composition(%)')\n",
    "axs[0].set_xlabel('Time')\n",
    "axs[0].legend()\n",
    "axs[1].scatter(range(predict_set.shape[0]), sample45, marker='x', color='black', label='Measurements')\n",
    "axs[1].plot(predict_set[:,5],linestyle='dashed',color='tab:blue')\n",
    "axs[1].plot(predict_set[:,4],linestyle='dashed',color='tab:red')\n",
    "axs[1].set_title('Data-driven (DAE) model')\n",
    "axs[1].set_xlabel('Time')\n",
    "axs[2].scatter(range(predict_set.shape[0]), sample45, marker='x', color='black')\n",
    "axs[2].plot(conc_set[:,5],linestyle='dashed',color='tab:blue',label='Fused @ 6')\n",
    "axs[2].plot(conc_set[:,4],linestyle='dashed',color='tab:red',label='Fused @ 5')\n",
    "axs[2].plot(ssae_set_6,color='tab:blue',label='SSAE @ 6')\n",
    "axs[2].plot(ssae_set_5,color='tab:red', label='SSAE @ 5')\n",
    "axs[2].set_title('Fused & SSAE')\n",
    "axs[2].set_xlabel('Time')\n",
    "axs[2].legend()\n",
    "#plt.tight_layout()\n",
    "#plt.savefig('compare_set2.png',dpi=1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "de92e058",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-4-66e0fdabe42e>:46: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  df['AT400(CO2 %)']=pd.Series(data=conc_list,index=df.index)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA1gAAAEYCAYAAABBWFftAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAADS8klEQVR4nOzdd1zV1f/A8de5g8teMmSoKCpuERE1Z44cDUfD1d5D21nWr7T6qmnDtpUtK0vLkWZa5s5y4F7gXsiWve84vz8u3BBBQYELep6PBw/uZ78/l/G57885n/cRUkoURVEURVEURVGUK6exdwCKoiiKoiiKoihXC5VgKYqiKIqiKIqiVBOVYCmKoiiKoiiKolQTlWApiqIoiqIoiqJUE5VgKYqiKIqiKIqiVBOVYCmKoiiKoiiKolQTlWApiqIoiqIoiqJUE5VgKUo9IIToKYT4VwiRKYRIE0L8I4ToUontpBCi+SXWGS+E2C6EKBRCfFttQSuKoij1Vk1dd4QQBiHEV0KIU0KIbCHELiHEkOqNXlHsS2fvABRFuTghhDuwHHgM+BlwAHoBhdV0iHjgf8AgwKma9qkoiqLUUzV83dEBZ4A+wGlgKPCzEKK9lPJkNexfUexOtWApSt3XEkBK+ZOU0iylzJdSrpJS7gUQQtwvhIgRQqQLIf4UQjQpnr+xePs9QogcIcSo8nYupVwspfwVOFcL56IoiqLUfTV23ZFS5kopp0gpT0opLVLK5cAJoHMtnZui1DiVYClK3XcYMAsh5gohhgghvEoWCCGGAy8DIwFf4G/gJwApZe/i1TpKKV2llAtqN2xFURSlnqq1644Qwh9rQnegek9BUexHJViKUsdJKbOAnoAE5gApQohlxRelR4DpUsoYKaUJmAaEl9xNVBRFUZSqqq3rjhBCD8wD5kopY6vvDBTFvlSCpSj1QPGF7F4pZTDQDggE3geaAB8IITKEEBlAGiCAoPL2I4RYWdxtI0cIMa52olcURVHqm5q+7gghNMD3QBEwvkZPRlFqmSpyoSj1jJQytrja3yNYHxSeKqWcV8ltVaUmRVEUpUqq+7ojhBDAV4A/MFRKaazGcBXF7lQLlqLUcUKIVkKI54QQwcXTjYAxwBbgM2CSEKJt8TIPIcTtpTZPAppdYv86IYQjoAW0QghHIYS6+aIoinKNqunrDjAbaA3cLKXMr/YTUBQ7UwmWotR92UBXYKsQIhfrBW4/8JyUcgkwA5gvhMgqnl/6buEUYG5xV447Ktj//wH5wEvAncWv/68mTkRRFEWpF2rsulP8rNYjQDiQqLqtK1cjIaW0dwyKoiiKoiiKoihXBdWCpSiKoiiKoiiKUk1UgqUoiqIoiqIoilJNVIKlKIqiKIqiKIpSTVSCpSiKoiiKoiiKUk2uyVLMPj4+MiQkxN5hKIqiKNVkx44dqVJKX3vHURF13VEURbn6VHTtuSYTrJCQELZv327vMBRFUZRqIoQ4Ze8YLkZddxRFUa4+FV17VBdBRVEURVEURVGUaqISLEVRFEVRFEVRlGqiEixFURRFURRFUZRqohIsRVEURVEURVGUaqISLEVRFEVRFEVRlGqiEixFURRFURRFUZRqUqsJlhBisBDikBDiqBDipXKWtxJCbBZCFAohni81P0wIsbvUV5YQ4uniZVOEEGdLLRtai6ekKIqiKIqiKIpiU2vjYAkhtMAnwEAgDogWQiyTUh4stVoa8CQwvPS2UspDQHip/ZwFlpRaZZaU8p0aC74cUkqEELV5SEVRFEVRLkKazeT++y8aJycc27VD4+ho75AURbkG1eZAw1HAUSnlcQAhxHxgGGBLsKSUyUCyEOLGi+ynP3BMSmm3QSVzNm4k5ZNPaPLNN2icne0VhqIoiqIogCU3F42LCwCJk6dgjI8HvR7HNq1x7hSBa7/rcYmKsnOUiqJcK2qzi2AQcKbUdFzxvKoaDfxUZt54IcReIcTXQgiv8jYSQjwshNguhNiekpJyGYf9j8bVlYI9e0n5+JMr2o+iKIqiKJdHmkxkr17N6Yce5ujgwciiIoRWS6M5XxD86Sc0uPcehE5P+o8/krNhAwCWwkLiX3yJ9PnzKTh0GGmx2PksFEW5GtVmC1Z5/elklXYghANwCzCp1OzZwJvF+3oTeBe4/4IDSfkF8AVAZGRklY5blnNEBJ63307a3Ll43HIzjq1aXcnuFEVRFEWpJGNSMhkL5pOxcBGm5GR0fn543X4HliIjWgcHDKGhGEJDcevXDwBZVISloMC67dmz5PzzD5lLlwKgbdCA4I8+xDkiwm7noyjK1ac2W7DigEalpoOB+CruYwiwU0qZVDJDSpkkpTRLKS3AHKxdEWuc33PPovXwIGHyZHUHTFEURVFqkDSZMOfkAFB06iSpsz/D0CqM4E8+pvnaNfg+OQGtq0u52woHB7Tu7gAYmjWjxd8bCV31JwFvTUfr6sqZhx7GmJxca+eiKMrVrzZbsKKBFkKIpliLVIwGxlZxH2Mo0z1QCBEgpUwonhwB7L/SQCtD6+mJ/0svEj/xRXLWr7fdKVOU+kSaTKR+Opvcf/4hYNpUDKGh9g5JURQFsLY85W7eTNaqVeSsWYv7TTfR8P9ewblLF5qvWY0+MPCy9iuEwKFxYxwaN8al+3Xk/r0RvZ9fNUevKMq1rNYSLCmlSQgxHvgT0AJfSykPCCEeLV7+mRCiIbAdcAcsxaXY20gps4QQzlgrED5SZtczhRDhWLsInixneY1xv/lmdH5+OHftWluHVJRqY0xKJv6558jbvh3h5MTJ0WMIen8Wrj162Ds0RVGucYlTp5H5669YsrPRuLjgev31uPbpDVgTpMtNrsrS+/vhedttAOTv3g1C4NSxY7XsW1GUa1dttmAhpVwBrCgz77NSrxOxdh0sb9s8oEE58++q5jArTQiBS7duABQcPIguIACdV7k1NhSl7pEWjPHxBM6cgXPnzpx57HEKjxzBtUcPiuLiKDxyBOPp0zg0a4Zrr172jlZRlKuUOSeXnA3ryd+xA/9XX0UIgdDpcBs4ELcbBuJy3XVoHBxqNAYpJYlvvEnR6dM0/vornDp0qNHjKYpydRNSXlG9h3opMjJSbt++vdr2Z8nN5Wj/ATg0a0bjb75GYzBU274VpTpJs5nM337D45ZbEBqNtepW8QcXS0EBwmBACMGZJ8aTs2aNbbuA6dPxHDHcTlEryqUJIXZIKSPtHUdFquO6Y0pPR+vqitDrqykq+7Hk5pK9dh1ZK1eSu2kTsqgIbYMGNF28GL2/fbrrGRMSOHX3PZgzMqxJVvv2dolDUZT6o6JrT20WubhqaVxcaDhlMvk7d5IwaZIqeqHUSab0dM48/AgJL00iZ/16AFtyBaBxdLQNnu3z2GM0+elHmm9Yj3P3biS88gq5W7bYI2xFqTFCiMFCiENCiKNCiJfKWd5KCLFZCFEohHi+1PxGQoh1QogYIcQBIcRTNR2rtFiImzCBU3fdjfHs2Zo+XI2wFBZizskFIOfvTcS/8AIFBw7gOXoUTX74nhYbN9gtuQLQBwTQZO63aD08OH3/A+Tvq5VHupVrTLd53Wg/t/0FX93mdbN3aEo1UglWNXEfPBi/F54na8VKzjzyKAWxsfYOSanjpMmELCqqlWMZz57lxK23khcdTcM337hkURandm1x7tQJvb8/jT7+mAYPPYRTp061Equi1AYhhBb4BGt12jbAGCFEmzKrpQFPAu+UmW8CnpNStga6AU+Us231xqvR4D1uHIVHj3J8xEiyV6+uycNVG2k0krNxI/EvvsSR63qQ/sMPALj27UOTeT/QfN1aGr78Ms6RkQit1s7Rgj4w0JpkubuT/lPZITcV5crlmnKrNF+pn2r1Gayrnff994NWS+qns8leu1aNj6Wcx1JURMGBA+Tv3EXutq3kb98BGg2Nv5xT4w9VJ06fjjkjkybz5uHUvl2VttW4uOD3zNOAtRuhOStLVdxSrgZRwFEp5XEAIcR8YBhwsGQFKWUykCyEuLH0hsWVaxOKX2cLIWKAoNLb1gT3IUNwbNeOs888S9z4CXiNG4ffxBfqZLd0KSVJ/5tK1u+/Y87IQOPmhtvgQTh3tY6konF0xLlzZztHWT59UBBNfvwRnbf1mWoppa11X1EUpTJUglWNhBA0uPdePEeMsPWRz9u+HWmx4BJVK8NzKTXEkpeHxtkZgLgJEyg6exaHRo3xnTAeQ/PmFW6Xt3Mn5sxM3K6/HnNKCqfGWEcmcGjWDPebb8J87hz6oKAajd2YmEje5i34PPJIlZOr0qTFwsnb70AfGEijzz+79AaKUrcFAWdKTccBVS4JK4QIAToBW8tZ9jDwMEDjxo0vK8iyHBo1IuTHeSS/+x7Z69fh+/RTUAsJVv6ePeRs2Ig0mfB79hkAUr+YQ+6mTVjy8rDk5lr/T7q5Erp8OUIITGnncOnZE/ehQ3Hp2aPGC1VUp5KuiqbUVOKefpqGL7+MY5sabaRUFOUqohKsGqD18ACK7+DNfJvCw4dpNPtTXLp3t3NkSlVJi4W0b77h3Dff0vSXn9EHBKAPboQ0msjdvJnsNWvwHjcOn/FPoHVzs21nSk8n+d13yVy4CM9Ro3C7/np0gYEEf/oJju3aXdACJI1GjAkJOFTTh7DS9A0b0mzlCtvv5eUSGg0ew24h+Z13yd26DZeu9eemgbRYSHrrLXL/3oRj+3Y4d+qEU0RnDC1bqDvT167yfvBVqvokhHAFFgFPSymzLtiZlF8AX4C1yMXlBFnucR0c8J/0Er5PTkDj4oKlsJDcTZtw69+/ug4BWK9huZv+4dycOeRt2waAxsPDlmDJwkKkxYzW2wt9cDAaF2d03v8V+w167716//dlKSjEGB/P6fvup/G33+DYurW9Q1IUpR5QCVYNEkLQaPannL7vfs48+hjBH3+kyl3XE1JK8nfuJOWDD8nbtg23G25AODoC4P/iRABMaWmkvP8Bad99h0PzULxuv53cLVvIXLqMnHXrMGdn4/3A/fg+/jhg/X2o6NmnxKlTyV71Fz6PPIxTeDiG1q2r5W5v0cmT6Js0qbYufV533knaD/NIfucdQn5eUG8+PJkSEsj6bTkOzZpZBy5d9hsAwbM/xe366+0cnWIncUCjUtPBQHxlNxZC6LEmV/OklIurObZK0bi4AJD+008kvzUDj1tH0vCVV2yt7Vcq6/cVxD//PDo/P/xefBHP229H6+piW+47YTy+E8ZXuH19+f9wMQ7BQTSZO5dTd9/DqbvuJui9d3Ht3dveYSmKUsepMu21wJSezun7H6Do6FEaff4ZLtddV2vHVqpOmkycGDGSwiNH0Li44P/yJDxGjqzww0LBoUMYmjdHaLVkLPmVlPffx6FpU/wnTcIxrGWljll44gRxT4yn6PhxAIRej/tNNxE4fZr1GAcPYmjdukofWExpaRwbNBjPO27H/4UXKr3dpWQs+ZWESZMImvUe7kOGVNt+a5opJQWtjw9gLfqRu3kznsOHI/R60n/+mcIjRzG0bIGheXMcW7a0fXhV6oeqlmkXQuiAw0B/4CwQDYyVUh4oZ90pQI6U8p3iaQHMBdKklE9X5ng1ed2RJhMpn3zCuc8+xyG0GUHvvYdjy8r97ynNUlhI5q9L0bq54j50KJbCQrJ+X4H7TTfWq+59NcEYH8+ZJ8ZTeOgQQe++U6/+9yl1S7d53cotaOGic2HLOFWtt76p6NqjEqxaYs7M5NRdd+PYrh2B06bW6rGVC2UuW0bmr0txiuyMY5s2FOzbbx10tzihSfnoY/RBQbgPuqFWP2gbk5LI37OH/N170Pn40OD++5BScrhzJB7Dh+P/f68gNJUr/pnw6mtkLFlCs2VLMTRrVm0xSrOZEyNGYmgVRtDMmdW235qQ8sknSKMR36eeumhyGj/pZbL++AOZnw+A1suLpr/+ateS0UrVXM44WEKIocD7gBb4Wko5VQjxKICU8jMhRENgO+AOWIAcrBUHOwB/A/uK5wO8LKVcUdGxauO6k/vvv5yd+CKW7Gwaf/sNzp06YUpLo+jkSXR+fuh8fcstiGHOySFj/nzOzZ2LOSUVt0GDCP7g/RqNtT6y5OWR/M67+DzxOLoGDS69gaIoVz2VYJVijwQLigeJ9PCo9AdkpfpZCgvRGAxk/Porad98S+HhwyAlCIFT5wiafP31eWND1QXSbCZpxgzSv/se91tuJnDq1EsONFpw8CAnbr0N77vvxn/SBcP7XDFTWhpaLy8wmUiYPMU6hpbBgNBqQKPFpXs3XLp3x5ydTeonn2IpLMCSk4slJwdpMdPgnntqvCX33NffkDxzJh7DhxMwbeol/+6kxYIxPp6Cffs4+/wL+D3/PA3uu7dGY1Sqz7Uw0HBlmFJTSfi/V2nw0IM4d+5M1ooVnH32OdtyrYcHOj8/At95B8ewluT++y9xTz2NJTsbl+u60+Chh3Du1u2q6N5Xk6TRSPKs92nw4APovL3tHY6iKHZS0bVHPYNVi3Re1pKvxvh4kt+bRcMpU87rz16fmbOzwWxG4+FRZy/M2WvXkvS/qTT++is8hw/Hc/hwzJmZFMQewtCieZ29SAqtFv9Jk9B5NyDl/fex5OQS9N67aIqfCStLSkni1GlovbzweeLxGomp5L0yFxaR+++/yPx8LIWFYLEgLRaEgwMu3bsjCwrI+PlnhIMDGjc3NK6uWDIzsRS3FMmiItDrq/13Jv2nn0ieORP3oUMImPq/St3UEBoNDsHBOAQHY2jeHIeLVIdUlLpK5+NDo89m26adu3al0Zw5mJKTMaUkY0pOxpicjNbdWpSn8OgxXK67jgYPPnhFVUavNQWxsaTPm0f2H38QPHt2pbuDK4pybVAJlh0UHj9B1sqVIARBb9ftLlZlSSkpOnGCnI0byd+xA///exW9vx8ZCxeRPGMGGmdn9E2a4D7oBjxuvrnGS5CXp/D4CfRBgWgMBrL++JO87dtxbNeWxCmvY2jeHF2pgg9aD496UQ1PCIHPo4+gcXcj6Y03yfhlId533WlNZsokD6akJIwJ8fg+/RRad/cajUvr6kKL9esqXK7z9SVs547z5kmzGYpjTv3sMwpirc80aJycqiWmjCW/kvj6G7j260fgjBmXNXipoUULwDp22bX+7IlSv+kaNMC1V88Kl3vffRfed99VixFdHZzat6fJD98T9/gTnBozhsB33r7kAO51TdafqyiIOYihaVMMLVrg0KxZhTfuFEWpGtVXzQ5ce/bA+957yPr9d4rizto7nEopOnWKxGnTODZoMMeH3kjyWzMoiD2EKTEBAJduXfF78UU8brsVjYszKe9/wLEhQzHn1N7I5JbCQlI+/Ijjw4Zx7quvrHGfOE7Gzz+T8NIkdA0a0Oiz2dVWYcsevMeOpdFXX+I+ZDBgfZbs+MiRZK9fT0l3X33DhoSuWIHnrbfaM9QKCa3W1mKl9fEhZ906Tj/4EOasC6pcX97+dTpc+vQmaNZ7l+xKeTGZvy3n6PX9MKWnX7DMlJJC1qpVpH33nbX19iKklJwcM5aE11/Hklt7fw+KotQsp/btCVn4Cw7NmhH3xHjS58+3d0gXJaWkICbGNp3911+c++xz4l98iRMjb+VQRGdO3XefbXnutm0UHDqEpajIHuEqSr2mnsGyE2NiIkcHDMT7zjvxf+lFu8ZSGYUnTnBqzFicOnbE9fq+uPbujT4wsML1i+LOUrB/H+6DrYlAyocf4X7jUAyhoUiLhcKjRzE0a4bQ6Sg6dYqCmFiQFttglZaCQrxG3YHWw4OC2Fhyt2zBnJaOufjDrmPbNnjccgsaZ2dyt2wl6/flZP+1GnNGBu433YT/pJdsDyFbcnPJ3bwZxzZtLhpzfZS9bh1Jb72F8dRpnCMjce3XD69xY8t9kL2uylq5krMTX8TQrBmN5nxRLSXlpZRX3O2w8MgRjt98Cw0efQS/p5+2zU+cNo307763TTtHRdHoyzkXtHTl79mDY9u2CJ2O1C/mkPLee+gbNybwrbdwjuh0RbEpF1LPYCn2YikoIPGNN/G+6846OU6WMSnJWthpya8UHT9O02VLcWzZEnNWFsLREePp0xQePUrhkaMIgwGfhx8C4Gj/ARjPngWtFocmTXBq3w63gQNxGzDAzmekKHWHKnJRSl250J197nlyNmyg+fr1dfJZLFN6OplLl+J9zz0IIZBG42W1CBgTEjg+bDiWvDycO3emIDYWS2YmIb/8jFP79qT9MI+k//3vgu2ar12DPjCQ1NmzSfngQ9Dp0Hp5QpERc04OYduj0Tg5EffkU+Ru2oTr9dfjefttuHTrVg1nX39Io5GMhQtJ+eRTzKmpeI0dQ8PXXrN3WFWS888/xE14EkPz5jT9eQFg/f0reW6xMpJmzMShaQhed9xRbXHFPfU0uf/8g8ctN9PgoYfQBwSQ8/cmCo8cwTmiE4XHT5Dw8sv4v/yyrZuVOSeX5BlvkfHLQgKm/s/Wkpi7bRsJL03CmJhIg4cewnfCeIRO9dKuLirBUuqKtO++w33IEHS+vnaNo+jMGRLfeJPcf/4BiwWnyM54jhiB++DBlaqOW3D4MIVHjhQnX0fI37Ub90GDaPjaq0iTifiXX8apY0ecI7tgaNFcFfBSrkkqwSqlrlzoCg4eJOeff/AeO7bOjbljSk/n5G23Y0xOptmSxRiu8IF/U1oaKR9+SP6u3Th1aI9T58649umDzssLc0YGxqRkhEagcXFB4+yMcHREODggNBprFywp0bi5WRM9KTElJ6P39wfAmJSM1tOjXrXa1ARLbi5ZK1fi2r9/lRKTuiJ/337yd++2PlsmJUd69qLBfffS4MEHL7lt7tZtnL7nHrwfuL9ax/wqiI3lxPARoNcT+NZ0PG688YJ1cv75B5du3RBaLXk7dxH/4osY4+Jo8MD9+Dz++HldUs05OSRNm072mjU0+3UJ+oCAaov1WqcSLKUuKIo7y/Gbb0br4UHwJx/j1LZtrR1bWizk79iBNFtw6dYVc04OJ+8YhdugG/AcPhyHJk2ubP9SIgsK0Dg5URR3llN33okpMRGwPs/sFBlJgwfuxzkiojpOR1HqBZVglaIudBcnLRbiHnuc3H//pfHcuao7k1LrpNFI3PgJ5Pz9N02+m4tzZMWfmy0FBRwfNgwkNFv6a7UVyyiRt2MH+sDASyZD+Xv2cHLUaPRBQQTOnIFz584VrmtMSkbv74eUksKYGBzbtKnWmK9FKsFS6oqCmBjOPP4E5vR0At96C/fBg2rsWNJiIX/3HrL+WEn2H39iSk7GqVMnQn760bq8GrpLV3hsKTGePUte9HbytkeTt307DV99DdeePcjbvp3U2Z/hHNkZp86dcerQQRXQUK5Kqkx7HSVNJrJWrEDn549Lt672DgeAtK+/JmfDBvxf/T+VXCl2IfR6At99l5O33srZ556n6a9LKmyVS/34Y4ynTtP422+rPbkCLpoolZa9eg2eo0fh9/zzaF1dL7puyQDGGb/8QuJrk9EHBWEIC8PQsgXuQ4aqks+KUo85tm5N058XEDd+AmeffpqiZ5+1PddUHUonTfEvTCTr99+tQ2P07oX7kCG49e1rW7cmh00RQtiGtvAcMdwWG1hb600pKaR8+JF1rEm9Hqd27Qh6/33bzaW6OqSLolQH1YIFGI1G4uLiKCgoqP1gpMSYnIzQatH5+NT+8cuGY7FgSkpCGAxXNC6Uo6MjwcHB6K+gipuiFMTEcHLUaFy6dyd49qcX9PEvijvLsUGD8Bw5goA337RTlJfPkptL+vz5FBw4SMHhQxSdOInG0ZHQP1ba/fmN+ka1YCl1jaWwkMTJU3C9/nrcB91wRfsqae3OWrmSrFWrCJk3D52PDzmb/sF8LhXX/v0veWPHHsyZmeTt3En+jh3k79tP46++ROh0JE2fTu7mLbgNHEiDRx5Ww2Eo9VadaMESQgwGPgC0wJdSyrfKLG8FfANEAK9IKd8ptewkkA2YAVPJyQghvIEFQAhwErhDSnlhXeWLiIuLw83NjZCQELvcUTH5+WFMTETr4oo+MMDuD75bmjWzPv90GeMHgfVCcO7cOeLi4mjatGk1R6dcSxxbt8bvxYnkrFmLJS//gmIwDsFBNPric5w6dLBThFdG4+JCgwcesE0bk5LJ/fdflVwpylVAYzAQ+NZ023TWqlU4dexoe364Msw5OWQuXkL6ggUUHTsGWi0u3btjzsxE5+ODa88eNRF6tdF6eOB2/fW4XX/9efMNLVtScOgwqZ98QvaaNQS98/YVP+utKHVJrbVgCSG0wGFgIBAHRANjpJQHS63jBzQBhgPp5SRYkVLK1DL7nQmkSSnfEkK8BHhJKS9a97zsncSYmBhatWplt+ZqKSWm1FRMyckIjRZ9cBBaN7daj8GSl4/WpXrGiJJSEhsbS+s6WLJWqV+klCClteBJTi4aF2eEEJhSUq7qRCRvxw5kYSEu111n71DqBdWCpdRl5pwcjvUfgHBwsBa/uMRNIUteHhpnZ2sr/cCBOLZvj+ett+J2w8B6WcSoItlr15HwyitY8vIIeu9d3Pr3t3dIilIlFV17arOmZhRwVEp5XEpZBMwHhpVeQUqZLKWMBoxV2O8wYG7x67lYk7Mqs2dfYCEEel9fDKGh1jLodojFnJZG0YnjmHNyqmV/qm+1Ul2EEAiNBktBAafvuYfEyVPI27mTo/0HkL12rb3DqxFSSpLffY8zjz1O4fHj9g5HUZQrpHV1pfH33yEcHDh1191kLv/9gnUs+flkLFrEidvvIO6ZZwBrK33on3/Q9OcFeI2646pKrgDc+l1Ps2VLcevfTxX7Ua4qtZlgBQFnSk3HFc+rLAmsEkLsEEI8XGq+v5QyAaD4+5WPUmoHWq2WiG7d6HrbrXTu2ZOTJ09iTErCmJSENJur7ThTpkzhnXfeOW+epaAAY2IiGlfXOlcuXlFKCIMBlx49yPj5Z07dcy9anwY4R0XZO6waIYQg+IP3EY6OJLzyf0iLxd4hKYpyhRxbtrSN/xj//POkfPghAIUnTpA4bRpH+vQl4ZX/w5Kfh2ufPrbtHBo3tlfItULn60vQe++hDwhAWizEv/giORs22DssRbkitfmwT3lNGlXpn9hDShlf3I3wLyFErJRyY6UPbk3KHgZoXAf/WTk5ObF7927btLX8aTzmjHTMmZkYmjWrkWezpMWCMS4OodHiEBSkWp6UOksIgd+zz6APaEjy+x8QOHVqnXyou7rofH3xn/QSCS9NIv3Hn/C+c5y9Q1IU5QrpvL1p/PVXJL75JiUfi7L/XEX6T/Nxv+EGvEaPwiky8pq9FpszMiiIiSVz6TK8xo7Fb+ILqry7Ui/VZgtWHNCo1HQwEF/ZjaWU8cXfk4ElWLscAiQJIQIAir8nV7D9F1LKSCllpG89eG5DCEHLnj3IcnVFGo1sXrmSvsWlVzds2EB4eDjh4eF06tSJ7OxsAN5++226dOlChw4dmDx5sm1fU6dOJSwsjAEDBnDo0KHzjmNKTsZSUIA+KNDaPVFR6jivMWNouWUzLt272zuUGucxbBguPXuS/N57GBMS7B2OoijVQDg40PCNN/CZMB4Ar7FjaLF+HUHvvoNzly7XbHIF1gQ05Jef8b7nbtJ//JETt91GQWysvcNSlCqrzRasaKCFEKIpcBYYDYytzIZCCBdAI6XMLn59A/BG8eJlwD3AW8Xfl15poH1LjSFR4o477uDxxx8nLy+PoUOHXrD83nvv5d577yU1NZXbbrvtvGXr16+/5DHz8/MJDw8HoGnTpixZsgSw9tvWu7lh2b8fabQ+mvbOO+/wySef0KNHD3JycnB0dGTVqlUcOXKEbdu2IaXklltuYePGjbi4uDB//nx27dqFyWQiIiKCzqXG9REGAzofX7Tu7pV8dxTF/q6VDyBCCAJen0LWH39e1QU9FOVaU/p/mLr+nk9jMOA/aRIuvXoTP+kl4p58itAVv9u9wrKiVEWt/bZKKU1CiPHAn1jLtH8tpTwghHi0ePlnQoiGwHbAHbAIIZ4G2gA+wJLif0g64Ecp5R/Fu34L+FkI8QBwGri9ts6pOpXtIlia1tsbnbs7FI8B1KNHD5599lnGjRvHyJEjCQ4OZtWqVaxatYpOnawDA+fk5HDkyBGys7MZMWIEzs7W6oC33HLLefu+2h6YVZSrjT4oiAYP3G/vMBRFUWqVa88eNFu2DFNCAkKnQ5rNWPILLhiuQ1Hqolq9HSClXAGsKDPvs1KvE7F2HSwrC+hYwT7PAdVa1/NiLU7Ozs4XXe7j41OpFqvK0Ol0WCwWa0lqNzeEVovFaOTZceMY2KkTf6xdS7cuXVjx00+Yi4qYNGkSDz/0EFgstjs977//frl3+825uciCArReXhcM3qooSt2TsXgJeTu2Ezh1qr1DURRFqRU6Ly/bjeCUWbPIXreeRp98jENIiH0DU5RLUJ+s67CQkBB27NgBwKJFi6wzpeTooUO0Cw3l+UceIaJdO2IPHOCGvn35+uuvyUpJoSA2lmMbNnB2/3569ezJkiVLyM/PJzs7m99++8067lZCAqbU1IscXVGUusSUkkLmosUUnTlz6ZUVRVGuMi49e2E+d44Td4wi5+9N9g5HUS5KJVh12OTJk3nqqafo1asXWq0WAI2DA7OXLqXz8OFEjRiBi48Pwx58kCEjRzJ27Fh69u9PlzvuYNwzz5Bx5gxt3d25fcQIwsPDufXWW+nVqxcyPx9LQQE6f3/VeqUo9YTHzTeBEGQuW2bvUBRFUWqdS7euhCz8BX1AAGceeYRzX39jHYheUeogcS3+ckZGRsrt27fbpmNiYmjdurUdI6oZ5pwcjPHxICWGFi0QGg1FZ89iyc5G6PU4NGtWo8UCrtb3VVHs5dS992GMjyf0zz+umUIflSWE2CGljLR3HBUpe91RFOXyWHJziX/5FXLWraPZb8twaNLE3iEp17CKrj2q+eIqpnV1xdC8OQ4hIbaWKpmfj9Dq0AcEqA9oilLPeAwbhvH0afJ37bZ3KIqiKHahcXEh6P1ZhPzysy25Mufk2DkqRTmfSrCuckKjQWMw2KYNzZtjaNEcTXFVQUVR6g+3gQNxvf56hFb9664OQojBQohDQoijQoiXylneSgixWQhRKIR4virbKopSc4QQOIaFAZC5bBnHBg8hb+dOO0el1HX5Bw5w5vEnsOTn1/ix1FVaURSlntC6utBo9qc4dSy3qKpSBUIILfAJMATrcCBjhBBtyqyWBjwJvHMZ2yqKUgsc27RB4+LMqXvuJf2XX+wdjlJHZS7/nVNjx1EQE4MpKanGj6cSLEVRlHrGmJRE4fHj9g6jvosCjkopj0spi4D5wLDSK0gpk6WU0YCxqtsqilI7DM2b0/Tnn3GJiiLx1ddIfONNpLHsn6xyrZJmM8nvvkv888/j2L4dTRf+Uitl/lWCpSiKUo9IKTk5ZgyJb76JNJvtHU59FgSUrnkfVzyv2rYVQjwshNguhNiekpJy2YEqinJxWg8PGn3+Gd733Uf6jz+Su3mzvUNS6oikGTM4N+dLPEeNosnXX6Nr0KBWjlurAw0riqIoV0YIgfe4O0l++21OP/ggQW+/jc7Hx95h1UflVfmpbFndSm0rpfwC+AKsVQQrH5qiKFUldDr8X5yI+0034tS2LWCtOKhxcanR41qKitA4OCBNJpJmzsSSmYUhLAzHNm1wbNMarbt7jR5fuTjvO+/E0Lw5XnfcUavHVS1YdcTUqVNp27YtHTp0IDw8nK1bt7J8+XI6depEx44dadOmDZ9//vl52wwbNozu3bufN2/KlCkEBQURHh5u+8rIyKjFM1EUpaY1eOB+AqZOJX/nLk6MGEmeKv99OeKARqWmg4H4WthWUZQaVJJc5e/dy9H+A8j6669q23fu5s2c++ZbEl59lZNjx3G4W3fiX3wRsCZ4OWvXkfvvvyTPnMnpe+/lcFRXkme9D1i7quVs+gdTenq1xVPdLLm5FMTGkrVqFee++gpjcjIAxuRk8vfts3N0lZezcSMJr76GlBKHxo1rPbkC1YJVJ2zevJnly5ezc+dODAYDqamp5ObmMmLECLZt20ZwcDCFhYWcPHnStk1GRgY7d+7E1dWVEydO0LRpU9uyZ555hueff76cIymKcrXwvHUkju3acvbJp8ha+QfOkZGYs7PJ3bIF586d0Xl72zvEui4aaCGEaAqcBUYDY2thW0VRaoHOvyH6xo05O+FJCsePx+fxx2xD1lSWOSODvJ27cOt3PQBJb79N4cEYtF5eOIQ2w+2GG3Du0sW2fuhfqxBCYEpLo+BgDAUHD+LUvh0ARSdOcObBB62xBQbg2KYNzpGRuA8Zit7fr5rO+tKkyUTRyZMUHDqEU4cOODRqRO62bZx99jnMqannrWtoGYbez4+kqdPIWb+ewBkzcB88qNZirSopJWlffUXyu+9hCAvDkpWF1sPDLrGoBKsOSEhIwMfHB0NxOXUfHx80Gg0mk4kGxX1FDQYDYcUlSQEWLVrEzTffjL+/P/Pnz2fSpEl2iV1RFPtxDAsjZNFChIMDAHnR2zk74UkAHEJDcY6MxLVXT1z791fj3pUhpTQJIcYDfwJa4Gsp5QEhxKPFyz8TQjQEtgPugEUI8TTQRkqZVd62djkRRVHKpff3o8n335H42mRSP/6YwkOHCHxreqW6DJrS0kj75lvS581Dmky02LgBracnQe+8i9bLE52XV7nblfyf1Xl749qzB649e/wXT1AQjb/9hoIDByk4eJCC/fvJWb0GhyZN0Pv7YUxMxJyRgSEsrNr+X0uLBaHRYEpJIfmDDyiMiaXw6FFkYSEADadMwWH0KPQNG+Latw8OjRrj0KQx+kaNcGjcGK2bm3W9114lLimJs08/TdFzz9LgwQfr3DWl4NAhUj74kJy1a3EbMpjAqVPtOiSRkPLa6xYeGRkpt5fqUhMTE0Pr1q0BePrpp9m9e3e1Hi88PJz333+/wuU5OTn07NmTvLw8BgwYwKhRo+jTpw8PPvggy5Yto3///tx0002MGTMGTfHdlwEDBjB58mT8/f257bbb2Lt3L2DtIjhnzhx8fX0B8PLyYt26ddV6PpVV+n1VFKXmWYqKKNi/n7ztO8jbHk3+jp1YcnMJmf8TTuHh9g6vRgkhdkgpI+0dR0XKXncURakdUkrS5s4leebb+L3wAg3uu7fCdU1paZz7Yg7pCxYgCwpwHzKYBo88imNYyxqJrej0aXQNG6JxcCDlw49I/fRT9MHBuPXvj9vAATh16oTQai+5HykleduiMSUmUHTqFAWxhyiMjcVj+DB8n3wSc04Ox24YhCGsJY5hrTC0CsOxVSscmjY9b6zUi7EUFpIw6WWyVqzA47ZbCZg8GaHXX+lbUC2k2cyxGwZhzsjA5/HH8L7//lpLACu69qgWrDrA1dWVHTt28Pfff7Nu3TpGjRrFW2+9xZdffsm+fftYvXo177zzDn/99RfffvstSUlJHD16lJ49eyKEQKfTsX//ftq1szZDqy6CinJt0jg44BwRgXNEBDz8ENJkIm/Hzqs+uVIURamIEIIG996Lc0QEjiXFLwoK0Dg62taRUiKEwJKTQ/qPPxYnVo9gaNasRmNzaNzY9tpr7Bh0Df3JXrOG9B9/JG3uXPSBgdZuh1otWX/8SdHp05gSEzAmJGJMSsQ5PJyGr72GEIK4J57AkpMDGg0OTZviFB6OIawVAFpXV1r++88VxaoxGAh8520cQpqQtfIPLHl5dut+B5C3axfpP/1EwJtvojEYCHp/lrXVzY4xlaYSrDIu1tJUk7RaLX379qVv3760b9+euXPncu+999K+fXvat2/PXXfdRdOmTfn2229ZsGAB6enptueusrKymD9/Pv/73//sEruiKHWT0Olw6RoFQO62baR9Oxe/Z59BHxho164TiqIotc2pQwcAjEnJnBwzmgYPPohr796c+2IO5vR0gj/6EIfGjWm+YX2FXQBrks7HB6877sDrjjsw5+SQ+/ffGBMSbS1YKR9/RNHRY2g8PND7+6MLaIg++L9aO43mfIHW0xN9QMB5yWN1EhoNvk8+SYMHH0Tj7IylqAhzSgr6oMqOcHHlcrdtI3X2bPI2b0Hr6Unh0aM4tW2LU/v2tRZDZagEqw44dOgQGo2GFi1aALB79278/f1Zv349ffv2tc1r0qQJAD/99BN//PGHrYLgiRMnGDhwoEqwFEWpkDE+npz168lZuxawjhvj2rcvAdOnVfnBb0VRlPpK4+KMY8swkt54kySNBqHV4nn7bUizGaHV2iW5Kkvr6or7kCHnzWv85Zdo3dwqfIbMuVOn2ggNwHaDLmn6dLL/XEXwxx/jHFGzxzdnZnLmiSfI374DrY8PfhMn4jV6VJ29WVjlBEsI4QIUSCnVCJfVJCcnhwkTJpCRkYFOp6N58+Z88MEHPPLIIzzyyCM4OTnh4uLCt99+y8mTJzl9+jTdunWzbd+0aVPc3d3ZunUrALNmzeKHH36wLf/1118JqYVRqxVFqbs8hw/HqX17CmJiMcbHk79zJxoPd6TRiKhkH3xFUZT6TuvqSvAnH3NuzhzMmVl433sPen9/e4d1SfqGDe0dwgW8776b3H//5fS99xIwfRoeN95Y7ccwpaai8/FB4+6O1tMT/1dewfP222qsla66XLLIhRBCg7UE7TigC1AIGIAUYAXwhZTySA3HWa0uVuRCqT7qfVUUpbaoIheKoii1z5SeTtyECeRv34HvU0/S4NFHq6XAhCklhaS33yZnzVqarVhRq6Xsq6Kia09l+oWsA0KBSUBDKWUjKaUf0AvYArwlhLizWqNVFEVRakXu1m2kzZ1r7zAURVGUekjn5UXjr7/G/ZabOff1N5iSkq5of9JkIu277zk2ZCjZK//A6+670Lq7VVO0tacyXQQHSCmNZWdKKdOARcAiIUTdqNOoKIqiVEnW77+TsWgRLtddh6H4OVBFURRFqSyNgwOBM2ZgPHMGfcOGSLOZs88+h9uA/rgPGmQbq/FSLPn5nBw7jsKYGFx69MD//17BUFzQrb65ZAtWecnV5awDIIQYLIQ4JIQ4KoR4qZzlrYQQm4UQhUKI50vNbySEWCeEiBFCHBBCPFVq2RQhxFkhxO7ir6GViUVRFEUB32eeRuPqSuL/plJb4yKac3KRFkutHEtRFEWpeUIIW9l5U1IShYcOEf/CRI7070/Kx59gSkmpcFtLQQEAGicnXHv2IOiDD2j05Zx6m1xB5boInkcI0U0IsVYI8Y8QYngVttMCnwBDgDbAGCFEmzKrpQFPAu+UmW8CnpNStga6AU+U2XaWlDK8+GtFFU9JURTlmqXz8sL3qSfJ27qV5LfeQppMNXYsY1Iy8ZNe5nC3biTPmFljx1EURVHsRx8YSLMVv9Nozhwc27Qh9eOPOdKvP/n79p+3njSbSZs3j6N9r6cgJgYAv+eew33QDbU2UHBNuWQXQSFEQyllYqlZzwK3AAL4F/i1kseKAo5KKY8X73c+MAw4WLKClDIZSBZCnFeGREqZACQUv84WQsQAQaW3VRRFUS6P16hRFB07Ttrc73Bs3wGPmy6/ElTye7Mwxp3BkpsHGg0aFxcc27Wlwb33Ihz05Kxfj6FFC9J++AGPW0fi2LJlNZ6JoiiKUhcIjQbXXj1x7dWTopMnyVi6FMfW1oGPMxYtxlKQT8aiRRQejMG5Wzc0Tk52jrh6VeYZrM+EEDuAt6WUBUAGMBawAFlVOFYQcKbUdBzQtQrbAyCECAE6AVtLzR4vhLgb2I61pSu9nO0eBh4GaFxq5GxFUWqPMTmZU+PuJOTHeeh8fe0djlJMaLU0fPX/cLvhBpyjugDWbnxa1/LHWykrf/8BnNq1BaBg/36M8fEIZyeQYMnNtQ2UqfPyovn6dVjy8jg2eAhJ06fT+Ouv6/2dSkVRrm3d5nUj15R7wXwXnQtbxm2xQ0R1i0NICH5P2Z7uIWPJYvK370Dn50fQrPdwGzz4qrsOXDLBklIOF0LcDCwXQswFnsaaYDkDw6twrPLeuSp1+BdCuGItrPG0lLIkuZsNvFm8rzeBd4H7LziQlF8AX4C1XG5VjqsoSvVI/XQ2xrg4Uj6dTcDk1+wdjlKGS9coAIrOnOHkHaNo8PDDeN97T4UXPlNqKkkzZpL12280/vYbXLp1o/HXX130GBqDAY3BgO+ECeT+8w8yPx9RRweKVBRFqYzykquLzb/WNfnuOwpiYnBoElLpG3n1TaWewZJS/gYMAjyBxcAhKeWHUsqKn1i7UBzQqNR0MBBf2Y2LKxUuAuZJKReXii1JSmmWUlqAOVi7ItY7iYmJjB49mtDQUNq0acPQoUM5fPhwrR0/KyuLV199lU6dOtGpUydGjx7NgQMHKlz/448/pnnz5gghSE1NrbU4lfrLmJxM5pIlICWZixdf9IFXxb607u44RUSQPGMGcU+Mx5yRYVsmi4rIWLSI0/c/wJHefchauRKf8eNx7ty5SsfwGjeWRrM/RaOSK0VRlGuK0Ghwatv2qk2uoBIJlhDiFiHEJmAtsB/roMMjhBA/CSFCq3CsaKCFEKKpEMKheD/LKrOhsN4+/QqIkVK+V2ZZQKnJEcUx1qhfd52lx1trafrS7/R4ay2/7jp7RfuTUjJixAj69u3LsWPHOHjwINOmTSPpCscSqKy0tDQGDBhAUFAQ//77L7t27eKFF17gwQcfZMuW8pu2e/TowerVq2nSpEmtxHi1MCYnc3TgDddkcpH66Wxb5ThpsZDy6Ww7R6RUROvhQfDHH+H/8iRy/v6bo/36E/f0M8ULtSS//z5FZ87Q4KGHaLZsGb7jn0DoqzZaR0mrWNHp02StWlXdp6AoiqIodlOZFqz/YW29uhWYIaXMkFI+C7wGTK3sgaSUJmA88CcQA/wspTwghHhUCPEoWAtqCCHisBbS+D8hRJwQwh3oAdwF9CunHPtMIcQ+IcRe4HrgmcrGdDl+3XWWSYv3cTYjHwmczchn0uJ9V5RkrVu3Dr1ez6OPPmqbFx4eTq9evZBS8sILL9CuXTvat2/PggULAFi/fj19+/bltttuo1WrVowbN85WYjk6OprrrruOjh07EhUVRXZ29kWP/9xzz/H666/z6KOP4lT8kGHnzp1ZtmwZEydOLHebTp06ERISctnnfK0q3UXuWmJrvTIWj+hgNKpWrDpOCIH33XcT8uM83G+6CedO4db5Wi1NFy4kdNWf+D3zNIZmV1ZGN+Pnn9E4qVYsRVEU5epRmSIXmVhbm5yA5JKZUsojxfMrrbiE+ooy8z4r9ToRa9fBsjZR/jNcSCnvqkoMV+rtPw+RbzSfNy/faObtPw8xvFPQZe1z//79dK6ge83ixYvZvXs3e/bsITU1lS5dutC7d28Adu3axYEDBwgMDKRHjx78888/REVFMWrUKBYsWECXLl3IysqyJU3lycnJ4cSJEwwZMoStW7cyfvx4fHx8CAgI4PXXXyciIoKdO3cSERFxWeem/KdsFznfxx+7Zgo9lG69KlHSiqWexarbnNq3x6l9+/Pm6f39q23/fs8/f+mVFEVRFKUeqUwL1gisBS1MWItbXNPiM/KrNP9Kbdq0iTFjxqDVavH396dPnz5ER0cDEBUVRXBwMBqNhvDwcE6ePMmhQ4cICAigSxdrJTB3d3d0uorz6JiYGFtyN3HiRBYtWsS8efNYu3YtZrOZsLAwjh07ViPndq25lrvI5axb91/rVQmjkZy1a+0TkKIoiqJUExdd+c8SVTRfufpVpgXrnJTyo4utIIQQsqR/2lUu0NOJs+UkU4Gel1+/v23btixcuLDcZRd7Ww0Gg+21VqvFZDIhpaxSqUspJdriEsoajcZWwr5rV2sF/eTkZNq0KTsetFJVFXWRu1ZasVpsWG/vEBRFURSlRqhS7EpZlWnBWieEmCCEOG/wKCGEgxCiX3Hp9ntqJry654VBYTjptefNc9JreWFQ2GXvs1+/fhQWFjJnzhzbvOjoaDZs2EDv3r1ZsGABZrOZlJQUNm7cSFRUxYUSW7VqRXx8vK2VKzs7G5PJdNH1S9Y1m83ExcWRkZHB1q1biYuLY926dXTv3v2yz02xulgXOUVRFEVRFOXqUZkEazBgBn4SQsQLIQ4KIU4AR4AxwCwp5bc1GGOdMrxTENNHtifI0wkBBHk6MX1k+8t+/gqsD5MvWbKEv/76i9DQUNq2bcuUKVMIDAxkxIgRdOjQgY4dO9KvXz9mzpxJw4YNK9yXg4MDCxYsYMKECXTs2JGBAwdSUFBAfHw8Q4cOvWB9d3d3AgICWLp0KTNmzGDEiBGMGTOGIUOGMGvWLL788kscHBwu2O7DDz8kODiYuLg4OnTowIMPPnjZ538tUF3kFEVRFEVRrg2iKj37isei8gHypZQZNRVUTYuMjJTbt2+3TcfExNC6dWs7RmRfSUlJ3HjjjUycOJGRI0ei0+mIjY1l9+7djB5dpTom57nW31dFUWqPEGKHlDLS3nFUpOx1R1EURan/Krr2VOYZrJIdGLCWag8BdCXP+Ugp36imGBU78ff3Z9WqVUyfPp0ZM2ag0Who3bo1kydPtndoiqIoiqIoilKvVDrBApZiLdm+AyismXAUe/H29ubtt9+2dxiKoiiKoiiKUq9VJcEKllIOrrFIFEVRFKUWCSEGAx8AWuBLKeVbZZaL4uVDgTzgXinlzuJlzwAPAhLYB9wnpSyoxfAVRVGUOqoyRS5K/CuEaH/p1RRFURSlbhNCaIFPgCFAG2CMEKLsmBRDgBbFXw8Ds4u3DQKeBCKllO2wJmiX/8CqoiiKclWpSgtWT+De4gqChYAApJSyQ41EpiiKoig1Jwo4KqU8DiCEmA8MAw6WWmcY8F3xOI9bhBCeQoiA4mU6wEkIYQScgfjaC11RFEWpy6qSYA2psSgURVEUpXYFAWdKTccBXSuxTpCUcrsQ4h3gNJAPrJJSrip7ACHEw1hbvmyDuCuKoihXv0p3EZRSngI8gZuLvzyL5117ypa2r0Kp+4okJiYyevRoQkNDadOmDUOHDuXw4cNXvN/KysrK4tVXX6VTp0506tSJ0aNHc+DAgQrXHzduHGFhYbRr1477778fY9kxnhRFUeo2Uc68sv/My11HCOGFtXWrKRAIuAgh7rxgRSm/kFJGSikjfX19rzhgRVEUpWJpBWnkm/LtHQZQhQRLCPEUMA/wK/76QQgxoaYCq7PWTYc/Jv2XVElpnV43/bJ3KaVkxIgR9O3bl2PHjnHw4EGmTZtGUlJSNQV9cWlpaQwYMICgoCD+/fdfdu3axQsvvMCDDz7Ili1byt1m3LhxxMbGsm/fPvLz8/nyyy9rJVZFUZRqEgc0KjUdzIXd/CpaZwBwQkqZIqU0AouB62owVkVRFKWMPGMeG+M2MjN6JiOXjaTPgj4MXjSYlSdWUpVxfmtCVYpcPAB0lVK+JqV8DegGPFQzYdVRUkJBJmyd/V+S9cck63RB5mW3ZK1btw69Xs+jjz5qmxceHk6vXr2QUvLCCy/Qrl072rdvz4IFCwBYv349ffv25bbbbqNVq1aMGzfO9ssUHR3NddddR8eOHYmKiiI7O/uix3/uued4/fXXefTRR3FycgKgc+fOLFu2jIkTJ5a7zdChQxFCIIQgKiqKuLi4yzp3RVEUO4kGWgghmgohHLAWqVhWZp1lwN3CqhuQKaVMwNo1sJsQwrm40mB/IKY2g1cURbnWmCwm9qTs4fM9n3PvH/fSY34PnljzBAtiF+Dt6M2EThMIcAlg4saJPL7mcc7mnLVbrFV5BksA5lLTZsrvPnH1EgIGF7dUbZ1t/QLo+ph1vri8t2P//v107ty53GWLFy9m9+7d7Nmzh9TUVLp06ULv3r0B2LVrFwcOHCAwMJAePXrwzz//EBUVxahRo1iwYAFdunQhKyvLljSVJycnhxMnTjBkyBC2bt3K+PHj8fHxISAggNdff52IiAh27txJREREudsbjUa+//57Pvjgg8s6d0VRFHuQUpqEEOOBP7FWAfxaSnlACPFo8fLPgBVYS7QfxVqm/b7iZVuFEAuBnYAJ2AV8UftnoSiKcnXLKsri9+O/syV+C9GJ0WQbsxEIWnm34q42d9E9oDud/DrhqHME4IF2D/BT7E98uOtDRiwdwRPhTzCu9Th0mqqkPFeuKkf7BtgqhFhSPD0c+KraI6rrSpKskuQKrii5upRNmzYxZswYtFot/v7+9OnTh+joaNzd3YmKiiI4OBiwtnidPHkSDw8PAgIC6NKlCwDu7u4X3X9MTIwtuZs4cSKLFi3C1dWViIgIXnvtNcLCwjh27FiFCdbjjz9O79696dWrVzWetaIoSs2TUq7AmkSVnvdZqdcSeKKCbScDk2s0QEVRlGuURVpYenQp7+98n7SCNIJcg7gh5Aa6BXaja8OueDl6lbudVqPlzjZ30r9xf/639X+8s/0dfj/+O5Ovm0zbBm1rLf5KJ1hSyveEEBuAHlhbru6TUu6qscjqqpJugaX9MemKkqy2bduycOHCCg5XcbdDg8Fge63VajGZTEgpEVWIQ0qJVqsFQKPR2Cpdde1qLaaVnJxMmzZlh4axev3110lJSeHzzz+v9PEURVEURVEUpSIHzh1g2tZp7E3ZS7hvOJ/2/5Q2DdpU6fNtgGsAH/f7mFWnVjF963TG/j6WO1vfyRPhT+Csd67B6K2q8gwWUsodUsoPpZQfXNPJ1dbZ1m6BkzOs30s/k3UZ+vXrR2FhIXPmzLHNi46OZsOGDfTu3ZsFCxZgNptJSUlh48aNREVFVbivVq1aER8fT3R0NADZ2dmYTKaLrl+yrtlsJi4ujoyMDLZu3UpcXBzr1q2je/fuF2z35Zdf8ueff/LTTz+h0VTp10hRFEVRFEVRzpNRkMEbm99gzPIxnM0+y9SeU/luyHe09WlbpeSqhBCCQSGDWDp8KSNbjOS7g98xYukI/o77uwaiP98lPxkLITYVf88WQmSV+soWQmTVeIR1iRDg6HH+M1eDp1unHT0uuwVLCMGSJUv466+/CA0NpW3btkyZMoXAwEBGjBhBhw4d6NixI/369WPmzJk0bNiwwn05ODiwYMECJkyYQMeOHRk4cCAFBQXEx8czdOjQC9Z3d3cnICCApUuXMmPGDEaMGMGYMWMYMmQIs2bN4ssvv8TBweGC7R599FGSkpLo3r074eHhvPHGG5d17oqiKIqiKMq1y2wx8/Ohn7np15tYfGQx41qP47cRv3FL6C2XlViV5WHwYHL3yXw7+FsMOgOPr3mciRsmkpqfWg3Rl0/UZhlDIcRg4AOsDxR/KaV8q8zyVlif9YoAXpFSvnOpbYUQ3sACIAQ4CdwhpUy/WByRkZFy+/bttumYmBhat25d+ROR8vxkqux0PZOUlMSNN97IxIkTGTlyJDqdjtjYWHbv3s3o0aMve79Vfl8VRVEukxBih5Qy0t5xVKTsdUdRFEWBPSl7mLZ1GgfPHSTSP5JJXSfR0qtljR2vyFzEV/u+Ys6+OUztOZUhTYdc0f4quvZUZRysGZWZd5HttcAnwBCgDTBGCFH24Z404EngnSps+xKwRkrZAlhTPF2zyiZT9Ti5AvD392fVqlVER0fTtWtXunTpwrRp02yFMhRFURRFURSlupzLP8er/7zKnSvuJDUvlZm9Z/L1oK9rNLkCcNA68Fj4Y/w24jcGhwyuseNUpYrgQODFMvOGlDOvIlHAUSnlcQAhxHxgGHCwZAUpZTKQLIS4sQrbDgP6Fq83F1hfhZiUYt7e3rz99tv2DuOaYkxO5tS4Own5cR46X197h6MoiqIoilKjDqUd4udDP7P8+HKKzEXc1/Y+Hun4CC56l1qNI8g1qEb3f8kESwjxGPA40EwIsbfUIjfgnyocKwg4U2o6DuhaDdv6Fw/8iJQyQQjhV94OhBAPAw8Dtkp5imJPqZ/OxhgXR8qnswmY/Jq9w1EURVEURal2BaYC/jz5Jz8f/pm9KXtx0DgwKGQQD7Z/kGaezewdXo2oTAvWj8BKYDrnd7/LllKmVeFY5fWjq+wDYFeyrXVlKb+geCDIyMjI2nvwTFHKYUxOJnPJEpCSzMWL8X38MdWKpSiKoijKVeN4xnF+OfwLS48tJbsomxD3ECZ2mcgtobfgYfCwd3g16pIJlpQyE8gExlzhseKARqWmg4H4atg2SQgRUNx6FQAkX2GcilLjUj+djbRYAJAWi2rFUhRFURSl3isyF7H61Gp+PvwzO5J2oNPoGNh4ILeH3U6kf2S1VAWsDyrTRXCTlLKnECKb/1qNSt4dKaV0r+SxooEWQoimwFlgNDC2GrZdBtwDvFX8fWkl96kodmFrvTIai2cYVSuWoiiKoiiXrdBcyNrTazl47iBaoUWn0f33Jf57rdVobdMmi4kCcwEFJutXvjmfQlMhBeYC8k35tvmF5kKEEAgEQgg0aECAQKARGoR1AoHgQOoB0gvTCXYN5pnOzzAsdBgNnBrY++2pdZVpwepZ/N3tSg4kpTQJIcYDf2Ittf61lPKAEOLR4uWfCSEaAtsBd8AihHgaaCOlzCpv2+JdvwX8LIR4ADgN3H4lcdpLYmIiTz/9NNHR0RgMBkJCQnj//fdp2bJmq6mUyMrK4u2332b58uUAhIWF8eqrr9K2bdty13/ggQfYvn07UkpatmzJt99+i6ura63EWt+Vbr0qoVqxFEVRFEWpqkNph1h8ZDHLjy8nqygLB40DFiyYLKYq78tB44CjzhFHraP1e/FrB611PFSJxGKxYMKElBKJvOB7ZMNIbmtxG90Cu6ERlS5WftWpdBVBIcTtwB9SymwhxP9hHavqTSnlrsruQ0q5AlhRZt5npV4nYu3+V6lti+efA/pXNoYr1W1eN3JNuRfMd9G5sGXclsvap5SSESNGcM899zB//nwAdu/eTVJSUq0kWGlpaQwePJj777+ff//9FycnJ3bs2MGDDz7IrFmz6Nat2wXbzJo1C3d3a+Pls88+y8cff8xLL9V8hfyrQc66df+1XpUwGslZuxZUgqUoiqIoykVkF2Wz8sRKFh1ZxMFzB3HQONC/SX9GthhJVMMoW2JjtpgxSRMmS5mv4nl6jd6WRBm0BrQarZ3P7OpRlTLtr0opfxFC9AQGYR2r6jMqXwnwqlBecnWx+ZWxbt069Ho9jz76qG1eeHg4YE2+Jk6cyMqVKxFC8H//93+MGjWK9evXM2XKFHx8fNi/fz+dO3fmhx9+QAhBdHQ0Tz31FLm5uRgMBtasWYObW8UNkM899xyvv/46Q4b8N9ha586dWbZsGbfeeisbN268YJuS5EpKSX5+/jXTp7Y6tNiw3t4hKIqiKIpiJ1JKvj/4PSeyTtDQuSEBrgEEuATQ0KUh/s7+thajsttsT9rOkiNL+OvUXxSYCwjzCuOlqJe4qdlN5RaN0Gq0aNFi0Bpq47SUUqqSYJmLv98IzJZSLhVCTKn+kK49JQlSeRYvXszu3bvZs2cPqampdOnShd69ewOwa9cuDhw4QGBgID169OCff/4hKiqKUaNGsWDBArp06UJWVhZOTk4VHjsnJ4cTJ04wZMgQtm7dyvjx4/Hx8SEgIIDXX3+diIgIdu7cSURExAXb3nfffaxYsYI2bdrw7rvvVs+boSiKotQJUkqwWKD4uwTrtMWCtEiQJa9LrVPqNVKWsx6lpiXnFwQW538rfeNOCOuXRoPQakGjRWg1oNUWT19kfh27ASilBLMZaTaDyYQs+9pkBnPJa5N1XZMZLGak2fLfd2mxblfyMylZZrGAEAiNBkTxszIaDRRPC03xeyk0oCler+R9K17PNk8jQKu17q/k/dWVfr+t34VWa33PNdb3vi6+73WFlJK3tr3Fj7E/4ubgRnZR9gXr+Dj52BKuhi4NcdQ6surUKk5lncJV78otobcwsuVI2ni3Ue9zHVWVBOusEOJzYAAwQwhhAK7dzpW1ZNOmTYwZMwatVou/vz99+vQhOjoad3d3oqKiCA629qgMDw/n5MmTeHh4EBAQQJcuXYD/WpoqEhMTY0vuJk6cyKJFi3B1dSUiIoLXXnuNsLAwjh07Vm6C9c0332A2m5kwYQILFizgvvvuq+azVxRFUQAyFi8h4ZVX/ks0hPVhc8r5EmB9La3Jiy2FKZ5Gygte25KnkuToalL8oV+UJAWlEwFt8QP68N97WPL6vO9VOJ6k4gTKYgGz+ZK7uGqUSozP+50tmS5vndLzzvsq3iXiwp9VmeWy5LdeUvHvPdZnimzrlF2v9N9P6Xml/2bKW1YSU7nnDPnmQvqZCxmid8TZQQPCEzMSi5CYsWCSFsxkYyQDk9yPUZqwIAnXO+Hu6IOrgxsa7S6E2M3Jkv1qBEJo/nsvyjln27lW8F5UuN5552gpvvFS6n2QxTdOSv5/SHnezRkpLdb1bdPl3LgpPa/kvS09r+zv1MVeF3+v6PdIlFo/YPo03G+4gZpQlQTrDmAw8I6UMqO4JPoLNRLVNaZt27YsXLiw3GWy7C9WKQbDf02+Wq0Wk8n60GFV7mZIKdFqrX1uNRqNbRDmrl2tPT+Tk5Np06ZNhdtrtVpGjRrF22+/rRIsRVGUGmIIa0mDRx8p/gCC7cOO7QNL6Q+JsqRl6OIJgzjvA2rxhzNN8QdC/vtAKIo/KIKwJiQlLSMaUf7rkv1csEz814JSqnUFuPADXemWrZIPfraWGjOYLUiL+b9ExjZd5rvJfOF8c/F2xdMlx5NlPmiW+yGzktdXoStO5LQ663um1Z03z/pah9BqEDqd9fUFy0u9Lmk5EprilqYyLU7a0suK732Xbmm0tTCWalEs+QBb0vpV0iImZXHLmPyvRawy77u5pJXN/N8HbuR/LZflTRf/rtp+hy+YV+p32/azKPOzKi+JuiABw/b6v89I5SRoJR/OS37fy0nizltW5hgl+y57PrL4Z7A9IZojaYdp7dWKjj4dbLGfn4SUSmKKW37NFhNaNNZlJcmNxWLbr23dkmSn9N93eedRcvq2ZKSc/xNlExZN6f8LJedb8veuOf+90WpL/e+w/g+wvveaS29f5v9P6Z/ZeZ+Jz3tdZl7J/8Dy/j+WWu4QXG7Zh2pR6QRLSpknhDgGDBJCDAL+llKuqrHIriH9+vXj5ZdfZs6cOTz00EMAREdHk5eXR+/evfn888+55557SEtLY+PGjbz99tvExsaWu69WrVoRHx9PdHQ0Xbp0ITs7GycnJ3S68n/UrVq1Ijo6GgCz2UxcXByurq5s3bqVuLg41q1bx6RJk87bRkrJsWPHaN68OVJKfvvtN1q1alWN74iiKIpSmlPbtjhVUNVVUZS6TUrJ9G3T+Sn2GPe0uZ9Bkc+prn1XuUp38RNCPAXMA/yKv34QQkyoqcDqKhedS5XmV4YQgiVLlvDXX38RGhpK27ZtmTJlCoGBgYwYMYIOHTrQsWNH+vXrx8yZM2nYsGGF+3JwcGDBggVMmDCBjh07MnDgQAoKCoiPj2fo0KEXrO/u7k5AQABLly5lxowZjBgxgjFjxjBkyBBmzZrFl19+iYPD+Q9bSim55557aN++Pe3btychIYHXXlPV75TyGZOTOTrwBkwpKfYORVEURVFqlZSSaVun8VPsT9zb9l6eU8nVNUFcrAvaeSsKsRfoLqXMLZ52ATZLKTvUYHw1IjIyUm7fvt02HRMTQ+vWre0YkX0lJSVx4403MnHiREaOHIlOpyM2Npbdu3czevToy97vtf6+KlYJU14nY8ECPEePVuN8KTVGCLFDShlp7zgqUva6oyjK1a8kuZp/aD73tr2XZzs/q5Krq0xF156qFKkQ/FdJkOLX6rfkKuDv78+qVauIjo6ma9eudOnShWnTptkKZSjK5TImJ5O5ZAlISebixaoVS1EURbkmSCmZunWqSq6uUVUpcvENsFUIsQRrYjUM+KpGolJqnbe3N2+//ba9w1CuMqmfzrY+0Iz1AfWUT2erVixFURTlqlaSXC04tID72t7HM52fUcnVNaYqRS7eE0KsB3oWz7pPSrmrRqJSFKXes7VeGY3FM4xkLl6M7+OPofP1tW9wVdRtXrdyBxN30bmwZdwWO0SkKIqi1EUquVKgakUuHIG+wPVAH6Bv8TxFUZQLlG69KlHSilXflJdcXWy+Uj8IIQYLIQ4JIY4KIV4qZ7kQQnxYvHyvECKi1DJPIcRCIUSsECJGCNG9dqNXFKWuOS+5aqeSq2tZVZ7B+g5oC3wIfAy0Br6viaAURan/ctat+6/1qoTRSM7atfYJSFFKEUJogU+AIUAbYIwQouygf0OAFsVfDwOl7w58APwhpWwFdARiajxoRVHqtLkH5v6XXEWo5OpaVpVnsMKklB1LTa8TQuyp7oAURbk6tNiw3t4hKMrFRAFHpZTHAYQQ87E+W3yw1DrDgO+ktdzuluJWqwAgF+gN3AsgpSwCimoxdkVR6pisoiy+2PcFvYJ6qeRKqVIL1i4hRLeSCSFEV+Cf6g/p2pSYmMjo0aMJDQ2lTZs2DB06lMOHD9fa8bOysnj11Vfp1KkTnTp1YvTo0Rw4cOCS202YMAFXV9daiFBRFKVaBQFnSk3HFc+rzDrNgBTgGyHELiHEl8VDl5xHCPGwEGK7EGJ7iqqgqShXtbkH5pJdlM2TEU+q5EqpUoLVFfhXCHFSCHES2Az0EULsKx4j65pSnYOnSikZMWIEffv25dixYxw8eJBp06aRlJRUDZFeWlpaGgMGDCAoKIh///2XXbt28cILL/Dggw+yZUvFD/Bv376djIyMWolRURSlmpX3CajswJAVraMDIoDZUspOWFu0LniGS0r5hZQyUkoZ6VvPCrsoilJ5aQVp/HDwB25ocgOtvFvZOxylDqhKgjUYaIq1wEWf4tdDgZuAm6s/tLot9dPZGOPiquWB/XXr1qHX63n00Udt88LDw+nVqxdSSl544QXatWtH+/btWbBgAQDr16+nb9++3HbbbbRq1Ypx48ZRMmh0dHQ01113HR07diQqKors7OyLHv+5557j9ddf59FHH8XJyQmAzp07s2zZMiZOnFjuNmazmRdeeIGZM2de8fkrSl3norugceKi85V6IQ5oVGo6GIiv5DpxQJyUcmvx/IVYEy5FUa5BX+37igJzAU90esLeoSh1RFXKtJ+qyUDqk7KDp15p2en9+/fTuXPncpctXryY3bt3s2fPHlJTU+nSpQu9e/cGYNeuXRw4cIDAwEB69OjBP//8Q1RUFKNGjWLBggV06dKFrKwsW9JUnpycHE6cOMGQIUPYunUr48ePx8fHh4CAAF5//XUiIiLYuXMnERHnf3b4+OOPueWWWwgICLjs81aU+kKVYr8qRQMthBBNgbPAaGBsmXWWAeOLn8/qCmRKKRMAhBBnhBBhUspDQH/Of3ZLUZRrRFJuEvNj53NTs5to5tHM3uEodUSlEywhRCTwCtCkeDsBSCllhxqKrc6qzcFTN23axJgxY9Bqtfj7+9OnTx+io6Nxd3cnKiqK4OBgwNridfLkSTw8PAgICKBLly4AuLu7X3T/MTExtuRu4sSJLFq0CFdXVyIiInjttdcICwvj2LFj5yVY8fHx/PLLL6xfv77aztOYnMypcXcS8uO8ejdGkqLUJxWN6VXW1T7Gl5TSJIQYD/wJaIGvpZQHhBCPFi//DFiBtafGUSAPuK/ULiYA84QQDsDxMssURblGfLH3CyxYeKzjY/YORalDqlJFcB7wArAPsFxi3atWTQye2rZtWxYuXFjuspJuf+UxGAy211qtFpPJhJSySg9XSinRarUAaDQaGjduDEDXrl0BSE5Opk2b8ysX79q1i6NHj9K8eXMA8vLyaN68OUePHq30ccsq3eWyppJVRVEqP3bXtTDGl5RyBdYkqvS8z0q9lkC5fX6klLuByJqMT1GUui0uO47FRxZza8tbCXYLtnc4Sh1SlWewUqSUy6SUJ6SUp0q+aiyyOqomBk/t168fhYWFzJkzxzYvOjqaDRs20Lt3bxYsWIDZbCYlJYWNGzcSFRVV4b5atWpFfHw80dHRAGRnZ2MymS66fsm6ZrOZuLg4MjIy2Lp1K3Fxcaxbt47u3c8fP/PGG28kMTGRkydPcvLkSZydna8ouSrb5bI6CocoiqIoyrUmISeB7w58x4rjKziWcQyTpeLrv3LlZu+ZjVaj5eEOD9s7FKWOqUoL1mQhxJfAGqCwZKaUcnFldyCEGIx1cEYt8KWU8q0yy0Xx8qFYu2PcK6XcKYQIAxaUWrUZ8JqU8n0hxBTgIawlcwFeLr4rWSMuOnjqZba8CCFYsmQJTz/9NG+99RaOjo6EhITw/vvv07t3bzZv3kzHjh0RQjBz5kwaNmxIbGxsuftycHBgwYIFTJgwgfz8fJycnFi9ejVZWVk8+OCDrFhx/lvj7u5OQEAAS5cuZcaMGYwYMQIfHx+GDBnCrFmz+PLLL3FwcLis86qs2uxyqSiKoihXEykle1L28EPMD6w+tRqzNNuWGbQGQj1DCfMKI8w7jDCvMFp6t8Td4eKPDyiXdjzjOMuPL+eu1nfh5+xn73CUOkZcrAvaeSsK8QPQCjjAf10EpZTy/kpurwUOAwOxVmCKBsZIKQ+WWmco1n7tQ7E+UPyBlLJrOfs5C3SVUp4qTrBypJTvVOpEgMjISLl9+3bbdExMDK1bt67s5ledpKQkbrzxRiZOnMjIkSPR6XTExsaye/duRo8efdn7rcz7akxO5tjAG5CFtpwdYTDQfPVf6lksRakB7ee2r/S6++7ZV4ORVC8hxA4pZZ3tslf2uqMoV8poNrLq1Cp+OPgD+8/tx03vxq0tb+X2lreTb8rnUPohDqUd4lD6IQ6nHSa9MN22baBLIC29WxLVMIrRrUaj1+jteCb103Prn2PT2U38cesfeDl62TscxU4quvZUpQWro5Sy8lfmC0UBR6WUx4sDmg8M4/zKS8OA74r7vW8RQngKIQJKqjYV6w8cuxa7J9YUf39/Vq1axfTp05kxYwYajYbWrVszefLkGj/2xbpcqlYsRVEURTlfekE6Cw8vZH7sfJLzkwlxD+GVrq9wS+gtOOudbeuFeYdBqPW1lJKU/JTzEq7Y9FjWn1nP78d/561ebxHiEWKX86mPYs7FsOrUKh7p8IhKrpRyVSXB2iKEaFO6xamKgoAzpabjsLZSXWqdIKB0gjUa+KnMduOFEHcD24HnpJTpZZYjhHgYeBiwFXJQ/uPt7c3bb79d68etiS6XiqJUzEXnUukqgoqi1B1H0o8wL2Yey48vp9BcyHWB1zH5usn0DOqJRlz8kXohBH7Ofvg5+9EruJdt/upTq5myeQp3LL+D5yOf5/aWt1epUNa16uPdH+Pu4M7dbe+2dyhKHVWVBKsncI8Q4gTWZ7CqWqa9vL/Ysv0TL7pOcTncW4BJpZbPBt4sXu9N4F3ggm6LUsovgC/A2lWjnOXqn0o1qmzX0xYb1tdsIIqinOdqLr2uKFebAlMBa06vYfGRxWxL3Iaj1pGbQ29mXKtxNPdqfsX7H9BkAB18O/DqP6/y5pY32RC3gdevex0fJ59qiP7qtDt5NxvjNvJUxFPqWTalQlVJsAZf4bHigEalpoOB+CquMwTYKaVMKplR+rUQYg6wvKqBOTo6cu7cORo0aKCSrGogpeTcuXM4OjraOxRFURRFqXcOnjvI4iOLWXFiBdlF2QS5BvFUxFPc1uI2PB09q/VYfs5+zB4wm59if2LWjlncuuxWpnSfwvWNr6/W41wtPtr1Ed6O3oxtVXZcckX5T6UTrOKCEh2Bkrblv6WUe6pwrGighRCiKdYiFaOBsr+dy7B295uPtftgZpnnr8ZQpntgmWe0RgD7qxATAMHBwcTFxZGiyoNXG0dHR9sgyIpSH1U0IO/VPgCvoij2kVmYye/Hf2fJ0SXEpsVi0BoY0GQAI5qPoEvDLpfsBnglNELDuNbj6BbQjZf+fokn1z3JbS1v44XIF857rutatyVhC9sSt/FilxfV+6JcVKUTLCHEU1jLoZeUZf9BCPGFlPKjymwvpTQJIcYDf2It0/61lPKAEOLR4uWfYR3wcShwFGuZ9vtKHd8ZawXCR8rseqYQIhxrF8GT5Sy/JL1eT9OmTau6maIoV7GKnlO6FgbgVRSldlikha0JW1lyZAlrTq+hyFJEa+/WvNL1FYY0HYKHwaNW4wn1DOXHoT/y8e6P+Wb/N0QnRjO953Ta+15JjbOrg5SSj3Z+hL+zP7eH3W7vcJQ6ripdBB/AWho9F0AIMQPYDFQqwQIoHp9qRZl5n5V6LYEnKtg2D2hQzvy7Knv8+kTdPVfqM2NyMqfG3UnIj/NUuX1FUZRSTBYTe1L2sOHMBladWsXZnLO4O7hza8tbGdliJK28W9k1Pr1WzzOdn6FnUE9e2fQKd628i0c6PMJDHR5Cp6nKx8ary8a4jexN3cvk7pMxaA32Dkep46rylyIAc6lpM+UXpVBKMZqNaIQGrUZbpe3U3XOlPkv9dDbGuLh6VW6/opsaiqJcnfJN+cRlx9HEvQkOWocaPVZOUQ7/xP/DhjMb2Hh2I5mFmeg0Oro27MqTnZ6kf5P+de5De5eGXVh0yyKmbZ3Gp3s+ZeHhhQxqOoihTYfStkHba+qZdYu08NGuj2jk1ohhzYfZOxylHqhKgvUNsFUIsaR4ejjwVbVHdJU4nnGc/239H9GJ0QA0dmvM7yN/B2DJkSUIIbi+0fW13vyvXD3OZJ1hydElZBZm8lLUS+i1VR8o0mQxsT91P+F+4bbpiu5Qmi1mpm+bzh1hd9DSq2WF+zQmJ5O5ZAlISebixfg+/li9aMVSyZWiXN0yCzPZnbybHUk72JG8g4OpBzFJEw4aB9r5tCPCP4JOfp3o6NuxWq7NZ3POsv7Mejac2UB0UjQmiwkPgwe9g3rTt1Ffrgu8DlcH1ys/sRrk5uDG9F7TGRwymEVHFjE/dj7fH/yeRm6NGBwymKFNh1ZLNcPqYrKYyCrKwtvRu1r3u+rUKg6lH2J6r+lqUGalUqpS5OI9IcR6rOXaBXCflHJXTQVWl0kpicuJw1nnjLPeGUet43l3cn479huv/fsazjpnHmr/EHqtHje9m235vJh5HEo/RLhvON8P/d4ep6DUI+kF6exK3kVKXgrnCs7RO7g37XzacSzzGF/t/wqLtOCgdeDFqBcrvc98Uz5Ljy5l7oG5xOfG8/uI3wl2C+blv18mJi0Gd4M7ZosZszTTyK0R7/V9D61Gy46kHSw6sognwp/g3rb3lpuMlR48uqJBo7OLstmeuJ2+jfoihOB45nHyjHm4O7izLXEbm+M3czzzON8O/rbGbkJkF2WzL2Ufe1L2sCelKvV6FEWpD5LzktmZtNOWUB1NP4pEotPoaO/Tnnva3kMzz2bEpsWyO3k33+7/li/llwgEoZ6hRPhF0Mm/ExF+EQS4BJx3nTdbzGQVZZFRmEFmYSbpBem218n5yWxJ2MKR9CMAhLiHcFfru+jTqA8dfTvWy252fRr1oU+jPmQWZrL29FpWnFjBV/u/Ys6+ObTwasGQkCEMbjqYRm6NLr2zapSan8relL3sSdnD3pS9HDh3gHxTPh19OzK8+XAGhQzCzcHt0juqgNliZmfyTj7a+RHNPZszJGRINUavXM1EZccruppERkbK7du3X/b2BaYCuszrct48nUbHwx0e5rGOj3Em+wxz9s7hqYinaOB0wWNjWKSFbw98y6wds/hx6I/lPjzafm7FD5Tuu2ffZceu1B8peSk8ve5p9qXuQ5YaMu7/uv4fo1qNwmg2km3M5vM9n/Nj7I+8f/379G/c/6L7PJd/jrkH5rLk6BIyCjPo4NOB+9vdz/WNr0cjNPx86Gf+Pvs3ReYitEKLVmgJdgu2JW9pBWlM3TKVVadW0a5BO56LfI4I/whbdStjcjLHBt6ALCy0HVMYDDRf/RfS25M/Tv7BkqNL2JW0C5M0seSWJTT3as7/tvyPBYcW2Lbxc/bjofYPMbrVaGvZ/4Jz1TIuS+nx7m7/7XZi02IRCJp7Nbd9GLoU9Rxk3SSE2CGljLR3HBW50uuOUjlmi5kPdn3A6lOrOZN9BgAnnRPhvuFE+EfQ2b8z7X3a46i7cBiRPGMe+1P3szN5J7uSd7EnZQ+5RmvLtr+zP/4u/mQWZpJRmEFWYdZ5/5dL02v0dPTtSN9GfenbqC9N3JvU3AnbUWp+KqtOrmLliZXsTtkNQAefDtwQcgPtfNrR3LN5td4gM5qNxKbFsjf1v4TqbM5ZwPoZrLV3azr4dsDL4MXKEys5lnkMR60j/Zv0Z3jz4UQ1jKpUJUajxci2hG2sPr2atafXklaQhqPWkQ/6fcB1gddV2/koV4eKrj2VTrCEEHOBp6SUGcXTXsC7UsoLBvWt6670Qmc0G1lxYgV5pjzyTfnkm/IxWUx0adil0n98ucZc+v/Sn+sbXc/0XtMvWK4SrPrBaDGyOX4zRrORqICoKt8pO555nHxTPjqhwyRNbDyzEWe9M/e0vQezxczjax4n3Dec7oHdCXQNxMvR64LuCUXmIu5eeTdxOXH8MfKPC7qcFJoLOZd/jkDXQFLyUhi6eCi9gnsxttVYOvt3vqx+9H+c/IOpW6ZitBjZOGojDloHTmaexPT+lxgXLgOj8b+V9XpSx/bntdDdpOanEuIewoAmA+gR2IOOfh3Ra/Qk5iayP3U/aQVpRPpH0tSjqS2uXw7/wqwds3j9utcZ2GRglWMFa7eRBYcWsPDwQuYNnYez3pmdSTspshTR3qc9LnqXWvmbM1lMxKbFsidlD4fTDxObFsu7fd4l2E0NaXClVIKlSCl5ffPrLDqyiF5Bvega0JXO/p0J8w67rG5dZouZIxlH2JlkTbgyCjPwNHjiYfDAy9ELT4Pn+V+O1u/OOudr6vkkgPiceP44+QcrT6wkNi3WNr+BYwOaezYn1DOUUM9Q2+vyEi+TxURqfiqJuYkk5iWSlJtEYm4iSXlJJOQkcDj9MEWWIsCa8Hbw7UBH34509O1I6watz3uGTUrJ/tT9LD221DaWWIBLALeE3sKw0GE0cj+/pa3QXMi/Z/9l9enVrDuzjuyibJx1zvQO7s2AJgPoFdRLlWVXylUdCdYuKWWnS82rD+rKhW761umsPbOW30f8fsEDtqqKYN10POM4C48sJMAlgLva3EWhuZDIH6x/Vx4GD1urS8k/+kJzIScyT5Ccl2z7yjfl81zkcwDc/+f9tuf0AASCwU0HM7P3zCrFFZcdx5nsM3QP7M6htEPE58STY8xhZ/JO/jzxJ20atOHLQV8C1q5xV9JlokSuMZcj6Udsz2+NWDqC42lHCUmStIiXnHODsDjJsK0Sc6Af37wWycgWI+ke2L1K47mcyTrD8xuf5+C5gwxsMpBJUZPwda78M137U/fzxuY3iEmLoUvDLrxx3RvlJjQ1nWBtS9jGU+ueIseYA4CXwYsw7zAmdplIC68WfLbnM/JMedzX9j68HL2u+HjXGpVgKR/u/JA5++bwUPuHeDLiSXuHc81KzE3kSPoRjmUc42jGUY5nHudYxjHyTHm2dXycfAj1DMXdwd2aSOUlkpqfikVaztuXk84Jf2d/Gro0JMwrjA6+Hejg24GGLg0rHU+huZB1p9fx69Ff+Tf+XySSzv6dGRY6DGe9M6tPrWZj3EbyTHm4ObhxfaPrGdB4ANcFXVfnCo8odU91JFh7gL5SyvTiaW9gg5Sy3g2OUFcudFlFWThqHWu8epFyZaSUbDq7ia/3f832pO3oNDpGh43mxagXbXfJCswFfLXvK/6J/4cg1yDS89PJM+eVu79mHs1YMmwJGqFhT8oe0vLTMEkTFmkh3Dccfxf/K4r32fXP8tepvwDrxal/4/7cEnoL3QO7X9F+L0ZKyZaELexM3snOpJ3sS92Hv7M/w5sP54H2D1zx/o0WI3MPzGX27tkYtAZe6/4ag5sOvmRMU7dO5edDP+Pj5MNLUS8xsMnACu8sV3RTQyBYfftq/Jz9qhRzQk4CP8T8QNsGbRnabCjJecnM3jObqIZRRPhF4Ofsd14sr29+nUWHF6HX6BkYMpDbW95OhF/ENXcn/HKpBOvaNi9mHm9te4tbW9zK5O6T1d9NHWORFhJzE60JV8ZxjmYc5VjGMXKMOfi7+NPQuaH1u0tDW0Ll7+yPu4N7tf4sE3MTWX58Ob8e/ZVTWacA8Hb0pl/jfgxsPJAuDbtcVsEo5dpVHQnW3cAkYCHWQX3vAKZKKetdlYa6dqEzWUwIRJVLuSvlyyzMZHvSdhJzExkWOuyKqzRN3TKV+Yfm4+/sz9jWYxkWOqzcZ+sAtiZs5aNdH120aEJNd/Es6XboqnfF18n3qurWcCrrFG9sfoP72t1Hz6CepOSlkFmYSSP3RiTnJpOUl0RaQRo3hNwAwOOrHyfYLZgJnSZcVqtdbFos434fx9LhSwl2C2ZezDx+Pfornf07E+kfSSe/Tng5etla5JLzkll7ei1rT69lW+I2AO5vd3+l76YfyzjG/Nj5LD++nBxjDs90fob7291PrjGXtafXcl3gdRX+7l3rVIJ17fr9+O+89PdL9GvUj3f7vlsvi0gotUtKyd7UvZgsJsJ9w9XnL+WyXXGCVbyTNkA/rFUE10gpD1ZfiLWnLl3o4rLjuP/P+3k28lkGh1z8jnxdlZibSIGpgBCPEFLzU5n09yTC/cLpEdiDdj7tauViZ7aY+WLvF6w+vZoj6UdsDx+PChvF/3X7v0rvR0rJnpQ9LD++nLGtx9LMoxl7U/ZyPPM4Nza9sdJ3ttQzdDWndKGKj3Z9xBd7vzhvuUZoWHfHOrwdvbFIS5W6I5YnMTcRHycfdBodq0+tZsGhBexJ2UO+KR+wthJuG2dNpu774z62J223PWd2R8s7CHANqPIx84x5/HnyTxq5NSKyYSR7UvZw54o7cdA4cFPoTdzZ+k5aeLWocHuj2UhWUZYtGUsrSMNN73bFd2bNFjO/Hv2V5PxkW+t7M49mdAvodtG7zFlFWUQnRBOXE0dcdhx3t7272quNqQTr2rTp7CYmrJlAuF84nw38THXpUhSlVlV07anSJ9/ihKpeJlV1VYBLAFqh5aeYnxjUZFC96tZgkRZ+iv2JD3Z+wIjmI5jUdRIeDh6kF6Tz+Z7P+WzPZ7g5uHF3m7t5uMPDV/xBtzyF5kIMWoO1hHjyDjwNnjwR/gSRDSOxSAstPK0fQpPzkjmZedJafSh5D+cKztG/cX8eaP8AecY8VpxYwbbEbWxP3E5KfgoGrYHO/p1p5tHM1udbqRtK/42Maz0Ob0dvco25+Dv74+fsR2P3xrYxUKrjd650X/8BTQYwoMkAjGYjB84dYF/qPgrN/1VMfLbzszjrnWnm0eyK/pad9c6MaDHCNt22QVt+ufkXfjn0C8uOLWPxkcW092nPZwM/w93BnSPpR1h3Zh0Hzx3k4LmDJOQm0MS9CctHLAfghQ0vsD1pOwEuAeg1eowWIy29WvJhvw8rHVNGQQYv/v0i/8b/e8Gy6xtdzytdX7mge2tcdhwf7vyQNafX2B5Od9O7MbDJQBq5NWJn0k58nX1rvbSzcnXYm7KXZ9c/S6hnKB/2+1AlV4qi1BmqHd3OtBotY1uPZWb0TEYtH8V97e5jYJOBl93qU/ruflZRFm56t2pJ2orMRZilGSedE2aLmdPZp5ny7xR2Ju+kZ1BPRrcaDYBeq2fhLQvJLMxkc8JmVh5fySe7P7E+d+IfcUUxpOan8nfc3+xP3W+7E56Sn8Ky4cto6NKQ2f1nl3uH3mQx8cy6Z9ibuhewjkni6+xLVlEWYP0QPm3rNDwMHnTx70KPoB70b9y/zg8AqVj7zo9rPa7Wj6vX6gn3C7cV+ChR3pAL1UGn0dHKuxWvdn+VCZ0msPDIQrYlbMNVb/0d/e7gd/x69FeauDch3C+cEe4jCHILsm0/ptUYOvl1Ii4nDrPFjF6jJ8QjBLDepPhq31fc1eauCrtRHjh3gGfXPUtKfgqvdX+NEc1HUGgupMBUwLJjy/h498cMXTwUX2dfzNLMyOYjeSz8MWvrXuI2RrYYyZCmQ86rHmaRFl7951USchMY2nQo/+v5vxp575Sr0/GM4zyx5gkaODbgs4GfVUvhHkVRlOqixsGqAyzSwuIji5l7YC4ns07axsbaHL+ZYxnH8HP2o4l7Exq7N8ZJ52Tbbt3+3/j362nkDL2OeGMqcTlxBLsGM3fIXMA6zk9yXjLtfdrTI6gHt7e8vdKJm9Fs5Ndjv7IjaQeH0g5xMvMk71//Pn0a9WHd6XU8ue5J3PRuTIyayLDQYRUmcVJK9qXus7UAxZyLIceYQ6G5kEJTIYXmQtr5tKOxe2NS8lL4/fjvGLMzSf15Pg4jbyJfLxnRYgRtG7Rl+fHlTPp7Eu4O7jR2a0ywWzCN3Bpxe8vbL9oNS0ppLaVuMVrHyCinQltCTgINXRpWWwvi1dxFsKJiEKWpape163TWabwcvS7rQ+bfcX8zfu14vB29GdB4AJ39O9PZv7OtUuOGMxt4dv2zeDt5M6vvLNr5tMNoMaLX6JFS8uLGF1l1chVmzLZ9uju40963PUEuQTjrnXFzcMOgNZBnyiPPmEdWURZ7U/ZyNOMoAF6OXqy5fc1lldIuoboIXjsScxO5a+VdGM1Gvh/y/QUltxVFUWpLtTyDdbWoqxc6i7SwI2kHXRpaBzF+Y/Mb/HL4l/PWaeXdil9uts576NMb2OKSgI/JiSZBbQh2DSbMO4y72twFwMLDC9mVvIu9KXs5mXWSdg3a8b+e/yPUM/SicSTmJnL/n/dzJvsMfs5+tPJuRZhXGHeE3UFDl4acyDzB+jPrGdp0aJUq3u1O3s19f9yHSZrOm/9a99e4veXt7E/dz5jfx9jm66UGF0d3pnSfQv8m/ckuyiYpN4lQz1CEEORs3EjavHk4deyI15gx6LzqTmnrulZm31JURP7u3Ti2bo3W7cru9F4seSytvieS9pCQk8Ce1D3Wrn6pB4nPjaeBYwP8Xfzxd/Yn2C2Yxm6NaezemECXwGp7MPtA6gE+3v0xO5J22J4t2zxmMyezTnLPyntwN7jT2b8z6QXpnM05i06js3U/nLFtBnqNnmGhwzhXcI6D5w5yOP0wRzKOkJyXTE5Rjq17IICj1hFnvTNhXmF0DehKt8ButPJqdcXnohKsa0NGQQb3/HEPyXnJfD3oa1o3aG3vkBRFuYZVW4IlhBiItYLgJ1LK3UKIh6WUX1xqu7qkvlzoLNJCVmEW8bnxnM46zamsUyTkJvBqt1expJ5jx7CBOGUXYdAaaL76L3S+FY8N9OfJP5m6ZSpTrptCv8b9OJ55nLT8NDRCQ44xh6S8JIxmI2Nbj8UiLbyy6RVuatCbCHMwppQUTMkpADh16oShRXOEpurPtpTctdZr9Bi0Bhx1jhi0BnydfHF1cMVsMZOdeIZTNw9Dk1+EXl/+eRUePUrSjJnk/v032gYNMJ87hzAY8Bg+HO87x2FoUfHD/wCyqIiCQ4fJ37uHgr17KTh4EHNuLhhNSLMZodej9fJC6+WJ3s8P1759ce3TB41z/avGJy0WslasJGXWLIxnz4Jej0vXrrgNGIDbDQPReXtXeZ8qwapYrjGXA6kH2Ju6l8TcRNo2aEu4Xzgh7iEXtI4azUZOZZ3iUPohohOj2ZqwlbicOMDaJbClV0sauzUmrSCNpLwkknKTKDAX2LbXaXQEuwbTxL0JIe4hdAvsRlTDqCsa9qFkIOS9KXvpHdybcSvGUWAqIN+Uj6+zL4EugQS4BhDqEcrDHR6udItvkbmIQnMhTjqnGit6czkJlhBiMPABoAW+lFK+VWa5KF4+FMgD7pVS7iy1XAtsB85KKW+62LHqy3WnLssz5vHQXw8Rey6WzwZ+ZrsZqSiKYi/VmWAtAe4D/g9YAdwmpXy8WqKsJVd6octevx4sEn1gAPqAADTu1TtOQ2UkTHmdjEWLwGgEvR7P224jYPJrF90mpyjH9lzRpL8nsfz4ctsyl3xJWJYr7zd+hvzdu8nbtRPjqdPl7kfr4YFTZCQuXaNw7trtshOuqp6XNBpJ+ehjzn31FRpnZ3wefxzvcWMpOn2atLlzyVy6DFlUhGO7dngMH47bgP5YsrMxxsdTFBdH4aHDFBw4QMHhw9b9A1pfH5zatUfr4YHQ60CnQxYWYU5Px5yeTtHp05jT0hBOTrhd3xe3QYNx7dWzVpItaTKRu3UrWStXUrBnD85donAfOgSniIiLvt/SbKboxAny9+0n/YcfKDhwAEOrVjS4/z4KDh0ie/VqjKdOI/R63G64Ac9Rd+DcpUulf4ftkWBZpIXkvGRS81NJzU8luyibMO8wmns2r5HiKVVxPOM4a06vYc3pNcSkxdgGynTSOdlagzwNnjRwbIBeq0ev0ZNjzOF01mnM0tqtzk3vRmTDSLoGdCXcL5yWni0veJ5QSklKfgqnsk5Zb7hkn+JM1hlOZVunC82FuOhd6BnUkzGtxtDZv/Nln1NmYSZ3rbyLc/nn+H7o94S4h9j9fb6UqiZYxcnRYWAgEAdEA2NKV8cVQgwFJmBNsLoCH0gpu5Za/iwQCbirBKtmGS1GJqydwOb4zbzb510GNBlg75AURVGqNcH6Qkr5cPHrt4D+Usp6dRvpSi90x0eMpDAmxjatcXFB37gxDsVfGnc3hF6P0OnR+fni2qMHGheX6ggdAGNyMscG3oAs/K96mTCU39pTdOYM6T/8QN7u3QiNFqHTgU5Lfn42xoJ8KCpCm5aFyMiybaP19sYpohPO4eE4NGuGztcXnZ8fsqiIvB07yIuOJi96O8bTp23ru3Tvbm3p6dUTradntZ+XJT+fs889T8G+fXiMHInf889d0PpiOneOzN9+I/PXpRTGxl6wf427O45t2+DUti2O7drh1KEDuoCAiyYW0mwmb/sOslauIPvPVZjT0xEGAy49euDatw9OHcMxNA9FaKtvDI2ikydJ+/FHspb/jjktDY2zM47t25O/ezeysBCdvz/OnSPQBwaiCwxE4+yMMe4sxjOnKTp5ioLDh5H51g/2usAA/J56Cvebb7YlZVJKCg8fIWPRQjJ/XYolKwuHZs3wGjMGj+HDLtmFsDYTrLSCNBYfWcwvh34hPjf+guXuDu5E+EfQya8T7X3a06ZBG1z01fe3VhGzxczy48v5ev/XHM88DkAHnw50D+xOR9+OtPdpj7vBnZOZJ63ddFP3kl2UjdFsxCiNOGodaebRjFDPUJp7Nqe5Z/Mr6iJXaC5ka8JW1p5ey7oz68gszGRS1CRGtRpV5X3FZcfx/IbnOZx+mC8GfkFkwzrb6+48l5FgdQemSCkHFU9PApBSTi+1zufAeinlT8XTh4C+UsoEIUQwMBeYCjyrEqyak2fM48W/X2T9mfVM7j6Z21reZu+QFEVRgOpNsIZJKZeWmp4gpfyoGmKsNVd6oTOlpmJMSMAYn4AxIR7j2XiMp09TdOoURWfP2lpHSgiDAdfevXAbOBBDWBj6oCC0rtaWJGkyYUpLw5Kbi9bT09qSconWoPNaeUqUau2RUpK3dRtp339Pztq1oNXiHBEBWs153eCEgwPCwQGttxeGZqE4NGuKITQUfXBwpVozjPHx5G7dRt6WzeRs+gfzuXOg0eDUoQMOzUNxaNwEh8aNQKPBkp2DJScbKSX6hgG21j+tj4/tWOWel06HU6dOFB44ADodAW+8gfvgQZeMrSA2lrytW9H6+KAPDEQfGITOz/eKWhqlyUTejp1kr15N9urVmBISABDOzji2aY1jyzAMLVtgaN4cQ1hYlZ51kkYjOZs2kf7jT+T+/Tfo9bj174/70CG49u6NxtERc04uOevWkfXnHxQeOYIpPgFZ6r3SBQTg0KgRhrAwayLZrh0OTZteNPmz5OeTtfIP0ufPp2DvXoSzMx633Iz7kKE4R3RC6C8sOlDZBGvbuG3sTdnLjqQdnMs/h4fBAw+DB54GT0I8Qmju2fyCZEhKyamsU+xI2sGWhC2sOb0Go8VIVMMoBoUMwt/ZHx8nHxx1juxP3c+OpB3sSNrB6Wxrsi8QhHqG0tG3o7XKn284jd0bV1vri5SSNafX8NGujzieeZzW3q0Z0WIE/Rr1q9LziDUp15jLxI0T2Ri3kbGtxvJClxcq3S1vxfEVvLnlTQSCab2m0bdR35oNthpdRoJ1GzBYSvlg8fRdQFcp5fhS6ywH3pJSbiqeXgO8KKXcLoRYCEwH3IDny0uwhBAPAw8DNG7cuPOpU6cu/wSvUUm5SUxYO4FD6Yd4scuLjG09tsJ169qzr4qiXP2uOMESQjgCzQEJHJNSFlxikzqrJu8kSosFWVSENBqRJhOFR46QveovsletwpScbFtPU5xImTMyoPTPQKNB6+WFY6tWOHfriku37ji2aX3eh+QjffpiSkq64NhaX18aPHA/GQt+pujECbReXniOugOvMWPR+/vVyPmWkBYLBfv3k7N+Pblbt1F06hTm1NRLbqcLDMAlqivOXbuSPHMm5vT0ctdzjowk8O2Z6AOqPmhrTZBSUnTiJAX795G/bz8F+/ZReOQIltzii7sQGFq0wKlTJ5w6dkQfGIDOxwdtgwYIvR5ZWIgsLKTo5EmyVq4ke9VfmDMz0fn64jl6FF533HHRZ+rA+p6bUlOx5OSiDwrErNOQXpiOXqPHVe9a5UFl8/ftJ/3HH8n6/XdkUREaV1dcunfHqXME+qAga2Lc0J8Bv91EBrmYNdbzLI8GDRqhwSRNCAQeBg+yi7JtXeIANBZJp3w/QtL0NEgpwDulAG1uAYd9jRwJFJxr6k3PNkMYHTaaZp7NrOdcVIQxOQVTSjKyoABZVISloJCs00dJObCToiNHIS2DRDczZz3MJHkJ0ho4oG/SBO/mrWnbxJqoOeur3sXzUNohpm6dyq7kXTT1aMqEThMY0HhAnRy7zmwx896O9/ju4Hd0D+jOa91fI9gtuML10wvSeXf7uyw9tpSOvh2Z0XsGQa5BFa5fF11GgnU7MKhMghUlpZxQap3fgellEqyJQAAwVEr5uBCiLxUkWKWpFqyqO3juIBPWTCDHmMPbfd6md3Dvi65/NVdvVRSlbrrsBEsIoQOmAfcDpwANEAx8A7wipTReZPM6yR4XOmmxUBATg/HMGYxnz1IUFwcWia5BA7Q+DdC6umLOyMSUnoYpJYWCPXsoPGItYaxxc8M5MhLnqCicIyPRBwVaW7q0WkznzpGzfj3Za9aSu2kTsqgIp/BwPEeNwn3IYDSOjrV6nqWZc3Ixxp2xnoOrG1o3a6udMTERY3w8xjNnyNu+g7xt26yJZhlaDw/cBg/G46Ybcercudqe86opUkpM/9/eecfHUZyN/zu719Ulq9pyr+BuwHRMi00vgQQCCaQRSCD1R94QUkiFhIQUaoDwUgKhmgB5CR1Tgg0uGGyDe5Usq1hdd7q28/tj9k53OkmWbHXN9/O5z93tzO7OMzu7O888zzxTUUFwyxYC69cTWPMhgY8+wmpu7no/r4fgMbNpPWk+3uOPoyh7DPne/BSXMUta8RDX2xu2xxeV3dGwg5pATXxdrxhu080o7yjGZ41nQuYEJmRNoDitmKK0IorSikh3pneoHESbW5RV8u13aH7nnbilLgUhcBQV4ZkxA8/06ZTnGzxT/RprgltIzyti0fhTmZ07k8NyZ+ANWIT27KZl5zaat2/B/8kGzO1lmKG2iJLNOR4iXhdZ+5oQlv1ccjoxvF4MrxcZDhOtre20Hs28PNxTpuDIzye8t5zArp1Qk5y/wQdVuSbO8eOYMPNYSg47EveECTjHjcNwdRwcoiXcwl1r7+LRTx8lw5XBd+d/l/Mmn9ejYA3RxkZCO3YQ2lNGuGwPVjCoLLzjxuIaNw4zN7dHipqUUlkvLUt9pASnE+FwJN0nz2x+hls+uIWojHLZjMv42qyvxdehAuV69fAnD/PghgcJRAJ8deZX+ebcb/ZZIIq+pD9dBIFvA18EIoAHyASWSikv7+x8WsHqGa/vfp0b3rmBbHc2t59yO9Nypx1wH61gaTSa/uZQFKw/oVwgvielbLK3ZQJ/AAJSyu/0oBAHHbFJCLETaAKiQCQmjBAiF3gCGA/sBD4npezYDGIzVF50kepqWla8j/+DD/B/oKxCcYTAzMoi2tAAUuIoKSbjlFPJvuizeKZPH7hCHwTSspTLW4JVTng8+ObORXTS6R0qyGiU0O7dRKqq8VeW8+mW5exu2Mmu1goqI7U0+GDdeEHYmdy5NoWZtNCxZVm0RFriARRijMscx5TsKYzyjiLPm0euJ5ewFaY51ByPDrmzYSc7G3fGAy7EcBgOMpwZpLvSKUor4pyJ56RYd6SUROvriVRUKKW4shLZGkSGQ1jBIK27dlK3bg2OsiqMbnobG5mZeKZNw3PYDNwzZuCZOhXXuHHxeYpWSwuB9RtoXb+eaH09ViCAFfAjTAeOokKchYU48vMxfL64m6ujoKDDiIiW309oTxmh3bsI7drFvo1rqNm8Dtfe/eQ0J1uOHXl5mLm5KoJkRgZBIuxpKWNX8x5qHUFGjz2c4w4/g8zisTiKi3GWlGBmZ6coRlYwSKS6htDOnfhXLKflveW0fvppiqUaq+1aCp8P1+gSnKPHYGZlgWkiTAMQyGgUGQkrBXN/LeHKfUQqlQWvQxwODK8XMyMDIyuLaIaXTa5aVhm7aRrlI33sRPx5aYRz0vlw/0fUttZy6thTuW7edQdcwqEjog0NtuJYRri8jHBlJZHKKiKVlQiXC9ekibgnTcY9cYKaN1hcjOF29/g8B+IgFCwHKsjFqUA5KsjFF6SUGxLynAVcS1uQi79KKY9qd5xFDFULlpTJVuj2/wcAKSUPbniQP61Wa6799ZS/Mso7qlv7agVLo9H0N4eiYG0Bpsp2Ge0ITBullF3HxE7Of9ARm2wF6wgpZU274/4eqJVS3iKE+BGQI6X8n67KMihfdN0gXFnJ3h/fiP+//8V9+OF4Z83EMSqfjFNPwT19+qB0VToY/GE/Oxp3sD+wX31a9+M0nOR4cshxq8VUHYYDQxiYwiRshWmNtBKMBmkMNVLlr6LaX01Naw21gVpqW2upa60jIiMYwsDAwGW64vOBstxZZLgyyHBlkO5Mj//OdGWS48lhas7Ugx7Rl1KyYf8Glm5Zyos7XqQl3EKuJzc+R2hqzlQ8pgeX6cIQBrWttexr2ce+ln00hZri19QQRlLZSjNKmZ47vdsLy8ai8MWOXdFSQX2wnpZwC02hJjbWbmR7w3bSnGmcOeFMloxfwrzCeR0u/BpbPPqlnS/xwrYXqA/WM9lTymUZp3JK9pE4GgO24m+BMECAmZ6Os7QUV2mpco8d4LZa31rPC+uf5J3lT2Ds2cfkBi+TIzn4Wix8LWFEc4BAqAUhJT7DQ1bQRDSlzu0QLhfC61VzGp1OrJYWrMYES6LTiW/uXHzHHI1n+gxcpWPUHEeHQ1myd+0itGs34fIyQmXlhPfswWpuRtqWKSkthMOJMFWAGjM3F2dRIY6CQhVMxjTagpdEoso1ORTC8vuxmhqJNjQSqaslXL43xWU3YoI/0016wWgy8ktsxTIdIz0DIyMdw+NFOB32PDyhFN2WFqzmZsIVFarsu3djNTQkHddIS8NRWIijsADpDxDcvh2rqSkpjzlqlFKQDQMcJmZmFuMfe/SQrulBhmk/E/gzatDvASnlb4QQVwNIKe+xB/3uAJagBv2+LKVc1e4YixiKCtabN0NrAyy5WSlVUsJLN4AnC06+YUCKFI6G+dWKX/Hs1mdZPH4xvz7u1yx6YlG351VpBUuj0fQ3nb17utNrlO2VK3tjVAjRkwgZRwFbpZTb7QI9DpwHfJKQ5zzgYft8K4QQ2UKIYillJz5K8X0W2b8fApYBXSpYh8r3l32fsqYycj255HpyyXJnxcMvxz4u04XLcJHjyWF2/myK0ooO+bx14UZWV6yk+nBBXc5moosOo9kso3XfvQTKAoStcHx9qdh6MwKBEIKoFaUl3EJLuIXmcDMt4Rb8YT8tkRaiVjSurDgMBz6nD5/Dh9fhVeGlbctItjsbr8OL1+HF5/RRkl7C2IyxeBwH54YYtaLsadrDtoZtbKvfxpa6LWys3ciuxl1IDm0B7Jh7XJ43j+K0Yg7LOwyn4cTCwpIWwWiQ+mA9Da0N7GrcRVO4ieZQc9IcoRgZrgyOKT6G40cfz8LihRSndR15sCnUxAcVH/BO+Tu8W/4ulf5KPKaHz4z/DBdOuZD5BfP7XcEwhBF3DewIKSVrq9fy9OaneX7b8zy1+SnSnekcW3IsM/JmYElVb/XBet7Y/QYVLRU4DAeLxizic9M+x8LihYM+jHci2Z5svnjEVVy24Gv8t/y/PLX5Kd5u3Ik/7CcQieI00jhn0iVcPPVixmaOBdT8r0hdPZGqKsL7KpRVr7ISGQzFFRvD58ORPwqHHVzFO3dupyH9XePH4xo/vt9ktgIBZYXcawfmKS8nUlVJpL6eaF09oV27sBobiTY3QzT1PohjmjiLinCNG0vmmWfgGlOKc2wprjFKeWwf3EVKSaSqmtCOHSo4UMVeIhUVWIFWsKLIqDVgrsxSyhdRy40kbrsn4bcEvnWAYyxDvXf6lFhEyjxvHnke9UzO8+Z1e92zUDTEvpZ9lDeXs7e5nPKaD9hb9RGV/1xOTskCSqu3M3bXB5ROOZPSpr0Uphf16z3dEGzg+8u+zwf7PuCq2VfxrbnfwhBGh8oV0Ol2jUajGQx0R8H6RAjxJSnlw4kb7QnBqbGwO2c0sCfhfxnKSnWgPKOBClRwjVdspe5vCYsbF8YUMDt0bofRHNpFc+pBsVMZnzmeYDRIbaCWHQ07aAg1qPDLVrhTxaDAV8Cc/DlMy5nGlJwpTMmZQnFacZJlxJIWda111ARqKG8up6ypjLLmMnY07GBr/VZqAjXweVCDrRLfjhfIySyMKz2x9XViC4NGrSgSiSWtuMtZujMdn9PHKO8o0pxp+Bw+HIaDqIxiSYuIFcEf8RMIB2iJtLC7aTdrq9dS11rXoWwCQXFaMROzJzI1ZypTc6YyOXsyRWlFZLra1geLvdy31W9jbfVaPqr+iA01G5IWTh2dPprpudM5c+KZTMmeQoGvIMntrb61ntrWWprDzfGyWtJSyqzpwmN6yHBlkO/LJ8OZ0WMlRkpJIBKgOdxMU6iJplATFS0VLN+7nHfL3+WVXa/Er+X8gvlMy52Gx/TErTwb6zaytmot2+q3IZGkOdM4pvgYvjnmm5w+7vRuW5sGAiEE8wrmMa9gHjcuvJHlFct5p+wd3i57Oy43KLfCY0uO5dp517KodBGZrswBLPWhYwiDE8acwAljTjhgXuFy4SwswFlYgHfWzH4oXe9ieL24J03CPalrN0ApJTIQwAoG1TyvSARpWRg+H0ZamrLa9eDeEkLE601z8KyqXMUvlv8iZXuGM4Ncby55nrz48zLPk0fYCrO3ZS97m/dS3lxOtb866RluCpOizFEUNO1ly5ZK3nQ6iOTnQf37sHQxLsPF6IzRlGaUMjZjLGMyxjAmfQz5vnxGeUeR68nttbl6uxp3ce3r11LeXM5vj/8t50w656COk+ZI69TapdFoNP1Jd1wERwNLgQCwGqXoHAl4gQuklOXdOtEhRGySUq4WQpRIKffaCtSrwHVSyreFEPVSyuyEY9RJKXO6KktfumpErShhK0wwGiRshdnXso+Pqj/io+qPWFe9jrLmsqT8btNNmjMNU5hxN7ZE0p3plGaUMtlbSs5jrzKmIkJBgyS3CbxGx2tf9QURK0JTqIlgNEggEqAl3EJZUxk7Gnews2En2+q3sa1hGxGrrfxeh5d8bz7+iF8phzYOw8FhuYcxO38203KnMSlrEhOzJ/bL+kUHi5SSzXWbWV25mrVVa1lTtYZKf3IkxwxnBrMLZjM3fy4LChcwt2Buhy52QwkpJcFoEFOYyr1SGAPu3jeYCFdVseuyyxn/2KMqxLz9uz/uyb6gO2GuE2UeTHIejItgf3Ko751gNEi1v5r9rfupDdSq79bauBt17Hdtay31wXqlQKUVUZJeQklaCaPTR6vf6ep3ga8AhzDhF9mAmty87wcb2N20hz1NeyhrKmN302722P/bz+E0hEGOOyeucOV71fco7yjyffkqkqntzZHo2eE0nEkeH+tr1vP9t76PgcGfT/4z8wvnJ51Hu/1pNJrBzKG4CHqBHwAu4HBAAP8BwqjoSd2lDChN+D8GaL9yaKd5pJSx7yohxLMol8O3gcqYG6EQohioYgAxDRPTMONuc6O8o5g5aiaXzbgMUBHJttZvZUvdFqoD1QTCAfwRP2ErTJ4nL/5yKkkrYUzGmLgVqOKmX1C/Egi3KcTSaVF9190U//xnfS6Xw3CQ40nWW2eOSh7FD1thpWw1bKOypZIqfxVV/ip8Tp960aeVMDZzLDNyZxy0W+FAIYRgWu40puVOi6/D4g+r6xa2wkStKPm+/CHlJtcdhBBD7lr1JzV33U24rIzqu+4GKeO/++Oe7Au6446VKPNQlXMo4jbdyorURbj9GGErjEB0bWGKzbmyMYHR797B6CU3c0zJMe2ySva37qe8uZwafw01gRqqA9XUBNp+b67dzP7W/R26WR+ICVkTuPPUOynNKD1wZo1GoxkCdEfB+jPwYynlx8AbsY1CiCPstO7a8lcCU4QQE1ARmy4B2q8Y+DxwrT0/ayHQYCtOaYAhpWyyf38G+GXCPlcAt9jfzzGISXOmMSd/DnPy5/Rov+Y330xZwJhwWC0kPEg6OU7DGXd/HAkczFpKmuFDuKqKhmefBSmpf+YZBICUNCxdSv43rxlU1p3eIlHm4SznUOeAlvOYcvX+3bDwGhXoIvYf2gJf2Agh4taprohaUeqD9dQEamgJtxC2woSiofhAVCgaImJFkrabwuT8KecPeXdjjUajSaQ7CtZ4W7lKwl7Jfnx3TySljAghrgVepi1i04bEiE2oycZnAluxIzbZuxcCz9quSQ7gMSnlS3baLcCTQoivAruBi7tbpqHElLeWDXQRNBpNAjV33a2i/QGEw0i7Qyqt/rMs9zeJMg9nOYc9QqhogTHlSgj1DWr7QboBm4apgnB483qtqHpelUajGYp0Zw7WVinl5J6mDWYGXbhcjUYzpAhXVbHt9M8gg8EO04W7/+ZH9iZdzXdZc8brKTIPJjmH+xysPmEQroPVV0StKJX+SnY17mJP0x7OnXQuHoeHt8ve5r/l/+WIoiNYULiAXE/qenoajUbTGYcyB2ulEOLrUsr72h3wq6igFxqNRjOiSLJedcBwsu5kN0t++UiUqk9vS5F5OMk5ImmvTA0D5UpKyUfVHzEhawJZ7iyW7VnGbatvo6ypjLDV5mYfW4dwW/02lm5ZymMbHwNgcvZkFhQu4EdH/ajXoiRqNJqRR3eeHt9FueddRptCdQQq6MUFfVQujUajGbR0OCcykUE2P7K7dOSO9dl3LQoaoOnlVwb9PFDNyKUl3MIL217giU1PsLV+K3eeeicnjjmRTFcmE7Mmsqh0EWMzxqpP5lgKfGrZgC/P/DKXz7icDfs3sKpyFav2reLj6o/jytXvV/6eYCTIkUVHckTREQech6bRaDTQDRfBeEYhTgZiYeM2SCnf6Cr/YGZQumpoNBrNICPRFXIwuQN2hHYRHJkEIgFuW3UbL2x/gZZwCzNyZ3Dp9Es5fdzppLvSD+qYUsr4chQ3vHMDb+55k5awGniYkjOFCydfyOWHXd5rMmg0mqHLobgIAiClfBN4s1dLpdFoNJpBiw5qoRmMRKwI2xu2MzVnKh7Tw9rqtZxSegqXTL+EWaNmHfJafYn733zCzUSsCBtrN/J+xfssr1hOXbAOgHA0zHVvXseCggUcU3IMM3JnYBrmIZ1bo9EMD7ptwRpO6JFEjUaj6ZqOAnkMZiuWtmANf2oCNTy9+Wme2vwUgXCA1y5+DZ/TR9SKDohiU95cznfe+A6b6jYBkOnKZGHxQr4y8ysp60RqNJrhySFbsDQajUYzcugokIe2Yg0sf179Zzbs38DC4oUcU3wM03OnD1uLSTiq5vs5TScr963kzrV38lH1R0SsCMeWHMsl0y7BbboBBqwORqeP5ulzn6YmUKOsW3uXs7xiOcGoGpSoba3FY3r0mokazQhEK1gajUajSWEoLG4+0sjx5FATqOEva/7CX/gLma5MFo9fzM+O6fvrcceHd7By30qm5Ezh8LzDOSzvMCZmTzzwosbdoDXSyrqadWys3cjG2o1sqt3EtoZt/OGkP3Dq2FNxGk4iVoTLZ1zOZ6d8lvFZ4w9doF5klHcUZ008i7MmnkWiV9CtK29l+d7lfH3217l46sW4TNcAllKj0fQnWsHSaHqJox89utMFMVdctmIASqTRHDx6cfPBxxWHX8EVh18Rt5i8X/E+Hocnnn7lS1cyJn0MR5cczaxRsyhOK+5xpz4cDbO2ei3v7X2PNZVruH/x/TgNJw7DgSUt/r393zyx6QlAhTR/9rxnAXhv73vkefI6VLqiVpSNtRvZ599HZUtl/Pvk0pNZMmEJlf5KvvLyVwDI9+YzLXcaJ4w5gbEZYwEVUv0fZ/7joOutP0mcv/X5aZ+nyl/FLR/cwkMbHuLqOVdz7qRzB0349/rWerY1bCNqRbGwmJs/F4/Dw+7G3Wxv2I4lLaSURKVKP3H0ificPppDzQCkOdMOeb6bRjNcGRx3uUYzDOhIuepqu0aj0RwMiRaTGIFIgFxPLsvKlvHctucAEAi+OfebXD3navxhP/etu4+S9BJGp42mJL2E4vTiuJvd2qq1/H3d3/lg3wf4I34cwsHs/NnUBmopTCvk6jlXc/Wcq7Gkxe7G3Xyy/xOiMho//0/e/QnVgWrcpptpOdNwGA6OLTmWb8z5BlEZ5ZL/uySe12k4KfQVMjt/NgClGaXcc9o9TMudNqzCoM8tmMvfF/+dFRUruH3N7fz8vZ9T3lzOdfOuG9Byrdy3kjs+vIO11WuxZJsb8IsXvEhpZimv7HqFv6z5S8p+b3/+bXxOHw+sf4D71t2Hw3CQ7c4m251NjieHu0+7G7fp5u2yt/mg4gOC0SAhK0QwGsQUJr85/jcA/HXNX3m3/F1C0RAAPqePorQiblt0GwBPbX6KPY178Dq9pDnS8Dl95HhyOHXsqYAaBHCah2451Wj6Eq1gaTQajUYzxPE6vNy26DZlLarbyNa6rext3su8gnkAVPoreXD9g0RkJGm/m465ic9O/SxhK8y2hm2cM+kcji05lqOKjuowzLkhDMZnjU9y05NS8vfFf+eT/Z/wyf5P+LT2UyxpkeZMA8Blurjz1DsZ5R1Foa+QXE9ukuXDEAbHjT6uD2plcHB08dEsPHMhy/Ys4/BRhwOwoWYD1YFqThpzUp9agcLRMKsqV/F22dssHr+YuQVzMYWJP+Ln67O+zryCebhMF4YwyPep4DXnTTqPY0qOwcDAEAZCCAwMstxZAJw45kSy3FnUtdZRH6ynPlhPY6gRl6GspW/sfoMXd7yI23TjMl24TTfpzra2lOZMo8BXELeu+sP+pDIv37ucd8reoTXaGt82OXtyXMH62itfY1PdJgp8BRT4Cij0FXJY3mFcNuMyAPY07mF0xmgMYfRRrWo0B0ZHEdRoeolZD83qNG3dFev6sSQazchDRxE8MFErSnWgmvLmcvY276W8uZxFpYuYnjs9ae0nTd/z43d+zAvbX2Bi1kQKfAV4HB7GZ47nB0f8AIAnNz1JfbAej+nB6/TiMT0U+go5qvgoANbXrCdiRXCZLpyGE5fpIt2ZTp43j3A0zH92/odle5bx3t73aAm34Dbd/PDIH/K5aZ8bQKl7RsSKEIgEaAm3IKWkOL0YgKc3P83W+q1U+auo9FdS5a9ies50bj/1dgAWP72YqIxy2rjTOG3sacwrmDdsg8FoBh4dRVCj0Wg0mhGMaZgUpRVRlFbEgsIFSWlauepffnHcL5hfOJ9Xd71KS7iFutY6TNGmBDy75VnW71+ftM8RhUfEFawfvfMjdjXuSko/YfQJ3HXaXQgh+P3K3+MyXCwZv4RFpYtYWLwQr8Pb94L1Ig7DQYYrgwxXRtL2i6ZelJI30Vjw7fnf5uWdL/PUpqd49NNHyfPk8b0F3+O8yed1+9xSSva37mdz7WYaw40sGb8EgA37N+Bz+CjNKB00c+k0gxPdOjQajUaj0Wj6Eafh5KKpF3WoLAD88+x/ErbCBCIBAuEAgUggqUP/6+N+TXO4mVA0RNgKE4qG4vPXHIaDJ89+kqK0ohHjJpc4QBCbn+gP+3m7/G1e3fkqOZ4cALbXb+ehTx7itLGncXTx0ThNJ8FoMD4X8clNT/LKzlfYXLc5vqB0jjuHxeMWI4TgNyt+w7qadbgMF5OyJzE5ezJHFh3JBVMu6H+hNYMarWBpNL1EmiOt0yiCGo1Go9H0BKfhxOlykunKTEmbWzC3y31L0kv6qFRDB5/Tx5LxS+LWJ4BtDdt4eefLLN2ylAxnBjmeHPa27GX5pcvxODxU+avwR/ycPPZkpuZMZUr2FKbkTIkrcD8/5udsrN3I1vqtbKnfwvv73scf8ccVrAueuwCvw0uhr5ACXwH5vnzm5M/hyKIjATXfTK+LNjLQc7A0Go1GM+TRc7A0Gk13CEVDLN+7nNd2v0ZLuIUpOVO4fMblKa6I3SVshXEaTqJWlJs/uJmdjTup9ldT5a+iOdzMF6Z/gRsW3kAoGmLBPxaQ5kwj35tPoa+QTHcmZ0w4g9PHnY4/7OehTx7C5/DhdXjjn2m50yjNKCUUDbGrcVc8eIjLdOEyXHgcnm65K8ZC7kdllKilvsNWmHRnOi7TRUOwgT1NewhGgwSjQcLRMMFokKOKjiLbk004GsY0zAGxiq7ct5KXd74cj1iZ7c4mx53DvMJ5eB1eLGkNmLVWz8HSaDQajUaj0YxoXKaLk0pP4qTSk3rleLF130zD5CdH/yQpzR/2xyN3RmWU7y34HtX+6nhwjur6amoCNQDUB+u5a+1dKcf/4ZE/5IuHfZE9TXu48PkLU9JjkUA37N/A11/+OqZhxpUoS1rccsItnDruVJbvXc43XvtGyv5/O+1vHDv6WD7Y9wHfX/b9lPRHzniEuZ65/Hv7v/n9yt8za9QsZufPZk7+HGbnz45Hl+wtdjbs5LXdr/Hqrle5ceGNzM6fTVlTGa/uepX6YH3S0gKvfPYVvOleHlj/APd+fG982YBMVyaZ7kx+ddyvSHOmsaZyDdsbtpPpyiTLnUWmK5MCXwF53rxeLXsiWsHSaDQazYhECLEE+AtgAvdLKW9ply7s9DMBP3CllHKNEKIUeBgoAizgXill6sJBGs0I4l8flnPry5vYWx+gJNvL9Yuncf680f2ad7CR6A7odXj5ysyvdJq3JL2ED7/4Ia2RVgKRAP6In0AkEJ9bV+Ar4A8n/YFQNEQoqtYXC1thZuWrCMZZrizOnXwuESuCw3BgCANTmIzJGANAaWYp1869FtMwMYWyRLlMV3zJhbn5c7nz1DvjofVdpgu34WZ0hqrrCVkTOHPCmXxc8zH3rbsvrui8/fm3yfHksKl2EwCTsif1OABIc6iZRz55hFd3v8qWui0AzB41m2A0CMAFUy7ggikXELWiNIWaqAuqJQJG+VTdzBw1k89N/Rx1wToagg00BBvYWr81vnTAf3b8h8c3PZ50zoumXsTPj/l5j8rZE7SLoEaj0WiGPD11ERRCmMBm4HSgDFgJXCql/CQhz5nAdSgFayHwFynlQiFEMVBsK1sZwGrg/MR926PfO5rhzL8+LOeGpesIhNsWn/Y6TW6+cFaKMtRXeTX9hz/sZ8P+DWyu2xxff+y7b36X13e/DkCmK5NsdzYTsybGw+c/u+VZaltr4y5+PqcPgWBh8UKC0SAnP3EyU3KmcPq40zlt3GkUpRX1WnlbI63UB+tpCDbQGGqkMdhIYVohM0fNPORjaxdBzaDh6EeP7jQYxIrLVgxAiTQazQjkKGCrlHI7gBDiceA8IFFJOg94WKqRyBVCiGwhRLGUsgKoAJBSNgkhPgVGt9tXoxkx3PrypiQlCCAQjnLry5tSFKG+yqvpP3xOH0cWHRkP3gFw/ZHXc9q409jduFstQN1an2TBe27bc6yuXJ10nKk5U3nm3Gdwm25evfjV+OLkvY3H4aHIUdSrStuB6FcFqy/cMYQQNwFfB6rtw/xYSvliP4ijOUg6Uq662q4ZWLRCrBmmjAb2JPwvQ1mpDpRnNLZyBSCEGA/MA95vfwIhxFXAVQBjx47tjTJrNIOSvfWBbm/vq7yagWV0+mhGp3eu9P7v4v8lEAlQH6ynLlhHY7CRabnT4ul9pVwNFP2mYNnuGHeS4I4hhHi+nUvFGcAU+7MQuNv+jgA/SHTHEEK8mrDvn6SUf+gvWTSakYRWiDXDlI5W1m3vM99lHiFEOvAM8F0pZWNKRinvBe4F5SJ48EXVaAY3JdleyjtQekqyUxc37qu8msGNEAKf04fP6RsRywj0Z0zDuDuGlDIExNwxEom7Y0gpVwBxdwwp5RpQ7hhAzB1Do9FoNJqDoQwoTfg/Btjb3TxCCCdKuXpUSrm0D8up0Qx6rl88Da/TTNrmdZpcv3hav+XVaAYT/eki2JfuGNcKIb4ErEJZuuran3wouGpoVyyNRqPpN1YCU4QQE4By4BLgC+3yPI96vzyOel81SCkrbHf2vwOfSilv689CazSDkdh8qO5E++urvBrNYKI/Fay+cse4G/iVne9XwB+BlDiYQ8FVQ7tiaTQaTf8gpYwIIa4FXkbNC35ASrlBCHG1nX4P8CJqTvBW1LzgL9u7Hwd8EVgnhFhrb9PzfzUjmvPnje624tNXeTWawUJ/Klh94o4hpayM/RZC3Af8u3eLrelt0hxpnVrqNBqNpr+wFaIX2227J+G3BL7VwX7v0vGAoEaj0Wg0/apg9Yk7RkLIXIALgPV9KYTm0NHujkMLrRBrNBqNRqPRdJ9+U7D60B3j90KIuSgXwZ3AN/pFII1mhKAVYo1Go9FoNJru06/rYPWFO4aU8ou9XEyNRqPRaDQxpAQhOv9/sHk1mt5Et1PNIKI/w7RrDkBnLlfaFUuj0Wg0A8KbN8NLN6gOKKjvl25Q2w8lr0bTm+h2qhlk9KsFS9M12hVLo9FoNIMGKaG1Ad6/W/1fYndM378bFl6TPOrfk7waTW+i26lmEKIVLI1Go9FoNKkIoTqgoDqgsU7pwmvU9sSOaE/yajS9iW6nmkGIdhHUaDQajUbTMYkd0hiddUR7klej6U10O9UMMrSCpdFoNBqNpmNi81MSSZy/crB5NZreRLdTzSBDK1gHg2XB1tdh57tQtgr2rYO6nRBsHuiSaTQajUbTO8Q6orH5KT+vV9/v353aIe1JXo2mN9HtVDMI0XOwDoZIAP5xYer2E/4fnPpTaK6GOxaAMw3cGeDNAV8uzP8STDsDQi1KOSs8HLLG9H/5NRqNRqM5EEKAJyt5fkrMtcqTlTq3pbt5NZreRLdTzSBEyBGorR9xxBFy1apVB3+AaATKV0EkCNEQhAMQaoaCGVAyDwJ1KtxnuAWCTeCvVduOvQ7mXAIVH8PfTlDHKpwF05bApFOheDa4dEh2jUaj6SlCiNVSyiMGuhydcSjvnX99WM6tL29ib32Akmwv1y+exvnzRh9y3m6j1xfSDAV0O9UMAJ29e7QF62AwHTD26M7TvTlw5u87T8+bDF99FXYvh00vwTt/hLdvha+8AmMXwifPwRu/AV+esnx5s8GTDcd9B9ILoOpTqNwATq9SyNyZ6pMzXpXNstSDor8eFpalFE2nR/1v2qceVqYLTCe40sEYZN6olqUskeEAhP2q/rzZEKiH7cuUlTHsV4qzvxYOvwBGz4eKj+CF76p0KwwOr5L79F/C+OOhdjtseBYcHuUyGrI/x14HuRNhy2vwzh/UeSNBkFFw+uDC+yB/qnI9XfsYONzqGK40VX9HfV21heZqpbinF7XVt0ajGbb868Nybli6jkA4CkB5fYAblq4DSFGcepK3R7R/l3T1bulJXo2mN9HtVDOI0ArWQODyQelR6nPcd1QHfs8HUDBdpbsz1G9/LdTugEAttDaqTjbApv/A679IPe7/26IUsGW/VUqbJwt8oyAtHzKK4NzbwZ0OzVVKkTBMMBxgRcCKqg4+KKVv74fQUAat9UoZSC+AC+5R6U9/BXa/rxSQsB8irTB6AXz9DZX+j4ugcl1buUwXTF0Mn/+H+v/Cd6ClJrnssboAeOQCZfEDwFYUp54BJ12vNv19MSBBmCAMpejMOBeOvRZCfrh9viqTZSmlyZcHC66EBVdA/W645wQlVyJLfgdHXw1NFfDUFclpDo9y5xw9X/325UF2KRhOdZ5wQMkISvF9/ZcJ+9pK8OxLlIJlOlRed6ZSooSh9ne4VX7/fti7BiKh5Pqde6lSsFY/CG/+2m4nWeq6ZI2Gi/5XpQeb1PHiCyhaqq4yS9T/2h3qHJFWCLcqZc1wwPSzVPqu5aruRYJC7MmEcceq3w1lkFYADhcajabvufXlTXGFKUYgHOXWlzelKE09yavRaDSavkMrWIMBX65yE4wx6RT16YwjvgIzzlFWlFALBBuVAubNUenjj7cX06tXiox/P1RvVJYSUMrZh/9IPqY3F/5nh/q99lH49AWllHlz1H6+3La8BTOUouH0tVnRskrb0k++QVmxrIiybLVUQ0ZxW3rdLmiuBBJGjLLHtf12+pTyBIBUsji9benuDHVcaamP09tmzTFdMPk0VT5hKGXBv19Z0kApR7MuVt9Ory2DB8YcpdJzJ8I1y5US7EyzLUi+tnPnT4PLn+782sw4B35cocrnSlcKVSITF6lPZ8z+nPokYkXbFJ4ZZ6vr0lypFOWWKmiqVBZOgFd+opSwRLw58D871e9Xf6qubSJZY9sUrGU3w463ktMLDodvvqd+P3kFVKyFvClQeJiaQ1hwOMz5vErf+CJEg+r6Gab6ziiCkrkqfdNLqm5Mp1JQTac6Rt4klR5qUUrpYLN4ajQDxN76QLe39ySvRqPRaPoOPQdrJLJ7BTSWq457NKw6wq501XkHpZQ407SVYiiy+WVo2GP/sa1/zrQ2Bah8tbKMmi6lPDo8yqqZM16l1+5Qinnic8HhVhY8UArSnveh6hP1adqnBgO+8IRK/+N0ZQVM5PAL4OIH1e+bS9WAQCLzLofz7lTn/GWe7TZpK7cZhTDvS7DwKtVWP3pcKcfeHCWDYSrrXHqBmhvpr1HbrYit4IdtF9ssZbXd9Z6y3kValYsmEiacpBS85mqlXDrcKn/hzOSBBc2gZrjOwTruljcotxUkkyhRDEAwOtvLf390Sqd5E+kor0aj0WgOHT0HS9NGV/PHoM0Sphl6TF3cdfroBV2n507oOn3akmRrKyjFJsaV/9c2t8yKKgujJ6st/cv/AWSb8hMNKRdWUPlP+7ly84zNXWusaFP0/bXw/LWpZTrtJjj+e0qx/Ovc1PSzboMjvwr1u+DxS1PTL7xPKVjVG+GZryanZY1VrrHjj4Oy1fDRP+2yRpRrqmXBoh9BzjgoXwMb/09tj8Y+QTj1JkjLU9a99U8ra6TptudPpsNx31VumPu3Kcuk06fcNg0TEG2uw9WboLVBWStjSqPp0nMHhjnXL54Wn1d1s+N+LjDfpRkfLiML7smBjBK47EkAbp++gZVrVlEf9dCMhwBuWowsFi++Uh2sZqtqn06f+rh82mKs0Wg0fYBWsDQazaGR6AYZc/XrjKKZXR8nNg+vI9JGwXfXKZfPQJ1S7KyIctsENTBw1m1KsTEdtpLihDFH2mWbAlctUx1Kh9ue9ybaFMDRC+BbK5V1y78f9n2sgppkFKn0+l2w/hlAquPGlKDId1V62Sp49zbb9dGlyuDwwAk/UApWc6U6nhVVimWwGUJNcIytNK55GP7751S5f1qjXCk/uBdW3p+c5kqHH5er3y/fCJtfUuWKzU9My4MvPafSl/1OnT/m1utKU9a/Y69T6cFmlWaYaAYPsblTt768idcb5xN053FcqZuJmZZyDY+5PwPzw6uZa/4bw2gb9GhOG0f6vBvVnxe+A7veTT5B8Vz4hu0W/LcTVRtJZPwJcOW/1e/bj1Buye5M5artzlDpp/5Upb/1ezWo4s6w55iaKqjTBDtq7vql6jvmPmyYynpeMEPdFxv/3TYwY0XVQE0sOm+4Vd0jVrhtgMaKwIQT1RzRlv3w1i329jBIlBf6rIuVW3bjXlh2S8KAhG3hn3OpmgPcUK7c42MBhmLf445Tc279tWqQw53eFnzIlWa7tOtBDo1Gk4xWsDQazdDAMCF7rPp0hDdbWao6w+VTHbWu0mOBXgAmnZycPvNC9emMhVepT2cc8WX1SSTRFfPIr8LEk1SQEiuqOo/xXiJwzLdg6hLl6thar6xZVkJAg5zxUDTb7lzacxc9mW3prfVKSYxHyPQry1tMwXr0IhXZ1JWh6tKVrpTO8+9U6f/+vlISTadS3oQBRbPalOKnrrTnXtodY2nBxJOVVRLgictV2WPum94cNfdvwZWd15kGUEqWUrQO4OZ38YMYUtpLh7RAuIX0xDZy6k+VohGLnhpqSbYwz7/Cnh+bQOL9dsSXVaCgYJNy9Q022YF0bFb9LzTtTd5/zqVtCtbSq1T7TOTIr8NZf1Dt5skvpcp03HdtBcsP/7k+Nd0wlYIVCcDHT7bN7xRC3QOltsdGsEkNQIB939n3yLjjlIJVvxve/E3q8T/3iFKwytfAo59NTb98KUw+FTa/Ai/9T5vyFZvje9ovYNTkNgu3y9dmQXT6YOpn1DWo362i0MYHbxxqkCZ/hrLi+2tVnQsjeY6rL09ZICNBdV8lBn+KBNvcnJv22QGO7OVloiEl/8STVPreD5XHQCIOt5INYM9KaN6XXHfudDXnGVT03ZYae260ncebq+QDZcEP1NrPJjv4UkZxm9fF+mfUNQLiz7ysMW3nX/e0PXgWG8ByqQBPRbNUesXHqk5Mt90G7KkP3myV3trQVq+Gw65H+zyWpQbtwv62eyMcUMfPHqsGMjY8q/aJH8OE4jlqUDHYDHtWJB/fcKhnctoodaz6PW2BxWLfnmw1BzwaVh4bMaU/9u3w2pGh7ekcsXqP1Z/TZw/yBdW9DG31Ly117U2nqtdAXdu89Vh6zniV3lTZNm0k9nyWUXVvmE4VuXr/1rZ50w63qv/RC9T5W2pU+WPXJVa3MW+okD/hvrfrXBiq/YAqeySYUD5L5csobGu7oZa2cktLlSM2oFuzRZ0/JpsVVfdZrG3sfFdd/9igjBWF9PyuYx30AnoO1kGwtaqJQMjqME3SeX0eqKp7Ogh2KMcTCQEmDmbwrbvNpqv6ONjjH+iIieIkrS9Ix4K2L2PSou9J27u4tl2UJ3m37p3rQMTHYEXbltjvzi5nsixtW6VUafHAg92Us+trFDtYz/aLkSJL/L+SUwCik4abWP6YXBIlpAQsmShzN1poXAYZL79sJ1ficRLrNvl/Z3IKDKHkMURshQURl7G9lCnntOWzLCWbJVWaJWP/uyiHFUXa7mFy53Lw1yBDAWTYD5EgMr0Aedj56hwr70M212BZ0fi5Ze4k5OzPqzpZ/RAy3ArCQAo1T0jkTcCYdDKGAGPFXZhEMITANARmqAmjZA7GzPPxOExOO6ywi4twYIbrHKwhRzSiLLORkOqkOTxtnfyaLW0dnJgbsS9PKfpSqiisidYtYagOsjfH7gTXJigftiLVm+6N0Yhy640E2+ZqpuUri5y/VgX4CdmKaahJfR9+oVLA9nygrMyhlrZormG/ivBaMB1WP6QsiO2fONeuVgrYf/+qghC15/sbIbNYra351i2p6TeUqfK9fCMsvyM1/ef16qHy3LXw4SPJaa4M+HGZ+v3Ul2HD0uT0jGL4wUb1+9GLYcsryel5U+A6u00/cAbsfi85vWSe8hoAuOd42LcuOT3ROvrXeUrBTGTqGfCFx9XvW6co62kisz4Hn71P/f51kVKyE1nwZTjnz6rt/LKDqQ/Hfhs+8yt1bX/fgXv8yT9R0YsbyuBPh6emx6IPV34Cdx+Tmn7uHTD/i0o5/ftpqekXP6jmKG99Hf7RweDdZc/AlNPU0j0dDT585WU15ePDR+G5b6amX/1f5TXy/r0dD0585yOlZL39B3jjV6np129XXhCv/UJ5Z7TnJ1VK2Xrxh/DB35LTTBf8tFr9fvYa+Oix5PTE4FtPXJ4afCt7rPJYAXj4fNj+ZnJ6wWHwzeXq9/2nQdnK5PQxR8HXXlW/7zwaqj9NTp90Cnzx2VSZDoLO3j1awToIzrvjXT4qa+jFEmk0GoXExMJAYtDxIIbKJYhixCf890U5DCROIphYOIhiYOHAQiSUq83ZSNofMLAQQu2vZLGSjhX7NoWFSRQjYV/RXgFHEMJBUDoJ4iSEkxAOwtJBGAcRDCI4CGPaRwALI16e2Llj5TCxcBPGK4J4COEliEeE8RLETRg3YVxCfQukLXXsKG0yx44VO66DqJJNRHARsY8VwkkUh4hi2ukxmSTCrkmDqDQIO9K47KZ/HtIVG+4KVnl9gPXl3X/vdP/VnjAgcYABkcRBhtS9Ox+c6egYXZWzs7InDi50dP72gyEdn//g6bCsPTmiJGGOprIiSW+uUhhbG5At+0mavyotGDUNTAeyoVwFELIsoM0KIUuPVsro/q3KShJLEyaYDsS44xGGUGmBOoSpLCjCdCAMF+SMU4M6LTUQDsQHsAAwTUSmclEV/mpEJGgnqJEgYTjiVgbh329b3VW6ECgFOKZcB+rsiLi2hQaUkuzOUL9bGxKsX7Hzu+JWDhmog2gYaUWR9jdOHzItX12BinVgRez0CFJakF6gBoEsC3a8hZSW2m7FLDjjkPnTkNEo7F6OtNfulIb97RuF9OaogahWOzhTopXFla4sJZGQbd1T22U0qlpGegHSnQUhP3L/NpUWyyMtZO5EZeUL1COrNiYNCEqJKpsnG1r2I2s2203IAAFSCmTBdHClq3ZTv6vtLWIP3JE3BVxepUA27bPrPXaBBTJngrKOBurblmeJW9AM5R5vmBBqQYb8Sj7LUm1URpFZtnW7udK2kEXBsmVDIG0LkqzfA/665EFKw6GUJLCt4o3xdoUQYLjaLFSN5WqwAqEGHYVAmC7ILlXZmyoQ0SCGEPYgrACnB5FRFG/bAgvDMBDCVG3X4UK40zlzVjFTCzO6eQN3jFawEjjUF13Vo9/AWbdVjdTGTbrER227YxIS8Xpv951wPdp3tojnTLA+tUtp+9m96yoTj2K7VEj1ZLS/E01AsZtXpJ476XwJMsRcBexuVSxfZ7KlHk+25U3aFvufWn+x/B2fISaDkSBLO7N8knTJdSranU/YD0thv9SUjIm/ZdIxUuWWyW2h/TmkjJ+jfR12Rtv1Uw9LKUz7oSzsNhvbnnodk65XwlpaQkYQ0kJYkbi8woomlNlKLVfifZBUH1abXLHjY7X97yFSGEpG23VG2p+4JSXhWgv7fEo+y5bZ/m9FEVYYYUUQMtL5CTVdojopbqTtRiLtOWHSXmpAxkyA0op3Ji13Ju5vf3BI5x3uCtZTq/Zw/dMf92KJNBrN0CHx/SgBgcMwE7w9kt0+OuqFdtQ17cyzpyM63r+zvCKmx8XzKeUo+X9PiHtOyLbfMc+U2P9ED5WYd0eix4rVrpty+6XzOGdOSY/K0R4dRbAXyc3w8unOIEjio+wCSX66i/x0J5GoZEtVzJdY3QgABZke8tJchKIWu2r9NIWbQESQRBEiCmYQRARDOBifcZjdOEVcofc4TJyGIColwYiV0rBdDgPTMIhaklDUSnIzMgSYhujgZkpVXtTv2Aha++0J+ZP970i61RIUlXX7NxCV0UQ1S6UYJvML5nVR0wmKXwfH7VQpaq8YJsoa9x9P8BOXHciWVIwkFYRoQpbV1WuJSEvdxEJie0ZjGg5OGHMSUqjRepFwmGjslAnHNAyBaSiFPWLFMtujkMLAMAw7XeUXnbnFJMlkJY+EymhCx9ZK3a99fSaOZCX6nceUlsTfor0lqQMlMJ4v9ts+fuJx4v9Fcv4U+dp8rZVClDDqG3dBisncrh0nniupHPach9gciJgff5JfvZlapsR2174+EmUzY3MrEo9rtpOz3QCNtOw5E8G2uRORUNscivhk/3By3aSUI+FcDk/yGnAOb9t3LPiH6bYHXKy2dtT+WhoO5Z4lzDbZhBlvm31hVxzJnH5YIf/37eN7tE93O08dPWI7cyMXHWyjg7ztz5zYmRLAWc+elZCa/Lx48cIXOy17h528pPJ3MADYPo+dWnnbbTQ+/zyZ551H4fe+1+H5unX+bu3Z1UFTy9bR+brqzHZEe9flRNftjrbLpO0y2fLX7pGeaLlLTesbEttezF2cxG0J1z6Wpy1/20ESt7flT243SfdBB12R9mVK2d6uLPFjtTtOR+Vs67vB7Idnd3IGWHfFuk7TNF0Ta999GZ9GK1gHQeSMP/LH/auTtkkJ584p4bMLxtASCPPHJ9fGFZqolEhLctERY1gys5iaOj8/fOxDPqrcjBXJBEstkusuegZXzkqigTH4d6aGo/7rpfM4d04JK7bWcNn976ek/++VR3Ly9AJe27CPbzyyOiX98auO5uiJeby0voJf/9+nOAx7LoQhMITgzsvmMyk/naVryvjTa5sJhq34aADAc9cex5gcHw+/t4M73txG1JKEoxbhqCQqJSt+dAq5aS5ue3Uz97y1HcMAUwj80QYwwqRN/CPCiBDcfyKRhjkAzGiaAYDDEDx3reo8/OnVzby+MXmidbrbweNXKR/nm57fwDtblG9vbJSiINMdT//+k2t5f3stAE5T4HIYTMpP5+7LVYjy6/75IZ/sbQDaHmTTizO5/VKl7F167wq2VDVhSYjY8h07KZe/X6kWIz7xltcpr29NKp8jYz3eMWrx5ubNP0ZaPoxaR3y05KIFY/jDxUrmaT9+kUi7YZQrjx3PTeceTjAcZfpPX0q5dt9cNIkfLplOXUuI+b96NenF4DANrv/MNL5+4kQqGgJ87aFVKa4+31w0iXPmlLC1qplr/rE6PlcnakmiluTGs2Zw5qxiPt5dx7efWBsfaYo97G8653COn5zHql11/ORf6/E4TTxOA6dpYAjB9YunMXN0Fqt21nLfO9sxhFD6j32eG8+awYRRabyxsZK/vbU9IU1iCMGfPj+X0lwfL62v4B8rdsfldpoCj9Pk1+fPJC/dzbJNVby1uRqnaSCEal+GEFx7ymQ8TpP3ttawtqw+/sJSc5Mk1548GYTgpfX7+HB3HZGopRRdS+IwBD89+zAQgidX7eGjPfWYhsDncpDuNslLd3PpUcoVYtmmKioaWuNzpgSQ4XGyZKaKNPjW5mqqGlvVSJoliViSbJ+Ts2epEbJnVpdR1RSMd04EguIsTzxS3HNry2kOqnlKhv3iLcrycOJUFcp+6ZoyWsOWmtNkCIQJY/N9LJyYB8C/1pQRirUtuxFMyE/nyPG5YFk8+5EKQGBEBUTADAkm5aczoziTqCVZvasO0wDTMDCFQAiTgkw3BZkewlGLHTUt8UPHRgYLMtzkedy0hqN8UtGQMgetNNdHQYaH5mCEdWUNWFK1OYeh7s2J+enkpuk19w5EdtUHZK96oJPUTnoJHfYeDqCh9Abd8KD4Xf2mDksjgHFvfjt2oC6O24O0DvJYwSC+ve/AfAtR/lfSnvsQw915O7SCQfwrV+I78kgM1wC01x55G/WVitOejgY/u8reqTrSzbwDOWzTwWB0PKmztthZnk6OlTKgDffsU/POREeX9OHzUjZZoRCt69fjmTkTw+mk6/uELtIOoQ31WP5uHfTgy9MB8WH4k29MDWjVS/SrgiWEWAL8BTCB+6WUt7RLF3b6mYAfuFJKuaarfYUQucATwHhgJ/A5KWVdX8nwrw/LufXlTeytD5DldSIE1PvDlGR7+XBPHbe9upmLmx/hM+4QrrN+x/nzx8Cbv4VNL0LVmfCmoKS1nmcnwKz0FwGQlhMZSUeYfgAMVxXPXHMMUautAxyxLA4rVhHBphVlcOcX5mNJmfQMOqxEpc8ek8VfLpkbV4BCEYvWsMWEUWkA5KW7OWpCrjpuVB2/vD7AF+5dQVVTkNw0F5leB/uDEfyhKC67M3v8797kp2nP4c0t4ZTpF2LuW4vTasUpLAx/Ne5lb4NhsaBlDM7cCkLRCFFMHJaJlC4QahRcGK0Ip1JwirOUcmkabYJkep0UZnji/yVKwYoxJsfL9OK26GgCyEvooE0rzIh3sGPyF2S449dv2cYqmoIRvE6T/AwXVU1BtlQ1887maoSAqYGPmGQ6cY47msmFGTj3fsCk6hfgjddACK7J9xLINRBOL7fWrbGvWXX8/M6cFSBdXDX7a3EFdkZCea9frMKKGwnm8ti1dRiCG8+c0SabnT63NBsAj9PgO6dOSTKVhy2Lw0dnxuuxOMtDRUMr26qaaY1YOAzBDUs/5tv//JCCTDduh0F1U4jSyA5yjCiO0vkUZnrgzd/iW7eC2c7PKUtjNKT8vA0naa7Z8NINuINFlOYeT2s4SjBs0RyJYNltFKAlFGXXfj9RS8blM4SgNdxmAZF2OZ32AIREYhix6yUJ2HmllISjktZwNP5o3biviadXlxGOWkkK4h1vbiXb6yQQjhKMpLoajnr3Jjxn/44P99TzxPs7MGUYw3RgIPGZUX5qPgzebD6uP5uX12wlgoHfUtbm/Iw2BeuR5bt4fWPyROsJo9LiCtbdy7aywlbuYxxeksnZs5WC9eB7O1nXbh7Nwgm5cQXrz69tiSsxMU6ZXkBtS4hbX97U4SKy4/J8RKKSi5sf4a7IeYRwJqVfOmY/Rx5ehtz4It/bdUPK/l87fgI/OfswAuEon/vb8pT075w6he+dPpXalhCf+dPbKek3njmDr584kfL6ABfe9V5K+m8vmMUXFo5le3Uzl963IiX9L5fM5by5o1O29wd98U7qM/z7qd26Mn5/gHr2GQKKMtXzsi4QJminx56opiEYla6ej/X+EOFo8v3hMAQ5PqedHiZsWW39H8BpGvH0/S2h+L0eO4fTYZDtVem1LaH4oFIsl9thkulRz++allB8/tSssP3MNsJgqLk9MqLmQuzb2NYOfW4nmR4HUsI+e/AiNrAmEWR4HGR5nfZ7rLW90zg5PhdZXifhqEV5fSDmYI6MRKEwi6xgM2mRVlq3raHS9Cm5Et6ro9LdpLsdBCqrqcl0IrZuQNiKmADy0l34XE5aw1FqmoP2/m3Wi9x0Fx6HSWs4Sn0gnGIhyfE5cZoqvSkYSVEhsn1OHIZBIBylJdTOkmzLZwqBPxwhEEp8zqoj5aa5MAS0BKP4Y23HvgYSyM9wI4Cm1khS24qVP99+dza2RpKe4zELy6g0NyBpCIQJxZ+96vimIcj1qbpqCKS2PdMQ5NhtpyEQJmKlts2seHpy2wPVNjPsvkFDaxjLUiYJYXsOOU1Bmsthlz85eqUQanDS61DLUjQFY1Fb2/r9LoeBJ1SHtCyaHDkQs2a1qmiOrvRsPA4TS0qag1H7mraZu9wOA5fDxJLQEozEL7qIp5s4TYOolATCMdkT9nc6SLcspDRAOlNUjEhrCw7TIGLJ+LUJV1YqL6c9e8gYU4LDEIQtSWvSe1GdwyMsIps2YU6dRpA2L4rYedLcTkxDEIwkv1dFPN2BIQShSJRQNPm5gBB4XXZ61CJsyfiebekmApUeibaTToDPqa5NMGoRjaaam3wuld4atoi2U9ZEQnogHE1pO4YQKt1Mfl/2Jv02B0sIYQKbgdOBMmAlcKmU8pOEPGcC16FeZguBv0gpF3a1rxDi90CtlPIWIcSPgBwp5f90VZaD9YX/14fl8QUfO0fyM8cjfMXxEg9bZ5B53q2cv/JSFT2ncCaMPx7evweAWRM6CTdN/5p+uycXJMq2bcLlTCpIj8tC4SyotMt81NXMqn6xW+cefHJ2//odSM6BMt/3hZwI1P+F18CSm/vWrt5NuivnjeY/+JrjPzwiey5nyFb4Yi/5en+IQDgat8BJqToJJdleACobWwlFrLi7itNQL9Esu4Pa2klZPfaLpLYlRMRWHmMWvmUbq/nNi59263qebqxiqVzEkeOyGbvnXwgkafMuIXvf28iKdewadSLy848h37kN66PHic6+lKzTrqco20skarFiey1RKYlaFpYFUSmZlJ/G5IIMWsNRXv+0Cmk7vcYsnNOKMpiYn44/FOH9HbUY7VzBJhekU5LtpTkY4eOyekyhBh0i9gDQtMIMCjI9nUrWXXo6B6uv3kmdna83ogje/8523t1aQ9SS8Tboczm4/wol9i3/2cjybTVEpYoqKYHCTDcPfllZ3//n6Y9ZvbtORZa03cHG5fni6dc+toaN+5qUhdS+joeVZHLb5+YC8K3H1rBrf4t9bjUIMm9sDjdfqCazX3Lvciobg0SttsG/k6bm88vz1Pp3593xLk2tERCws2EHoKz/7gIVna5l+3WAwbTcafGImufMLuEbJ00iFLH4/L3L41Zrw1D33rlzSvj8kWNpag3zPdv6btrpAOfOGc2SmUXUNAe56fkNAMhgkOY330REo5y2exULqjdTmV3IP7/+WyynC8tS915UwlUnTGRhlsVrF3yRO2echTRMXIfPRDocWFLyozOmc+ykUazcWcuPl65L8Q748yVzOXJ8Lq9s2Mf/e+oj5RlhWfH0Z645lnljc3hy5R5++EzqHLtXvnciUwszeODdHfzy36nN678/OoXR2V7ueGMLf3hlc0r62p+dTrbPxc3/+ZS/vbU9JX3Lb87AaRr89F/reWTFrqQ0l8Ng86/PAJRnyNI15UnpuWku1vz0dAC+8cgqXt6Q7HlSmuvlnR+qUNiX3/8+726tSUqfXpTBS989EYDz7/wva/fUJ6UvGJfDM9ccC8Dpt73FlqrmpPQTp+bz8FdU2z325tfZ25DsWXLmrCLuukx5rsy+6WUaW5Pn1l68YAy32p4lk378Ykon/Mpjx3OT8xFaV9zP9OBDtOdbiyZx/ZLp1DQHOeLXr6Wk/3DJNL65aDK79rdw0q3LUtJ/ed7hfOmY8Xyyt5Ez//pOSvofL57DTevPIOIfR2DXNSnp91y+gCUzi1i2qYor/3dlSvqDn53GoiMn838fV/Ctx9akpN/n+pQxTz3AOxd9i9+Gx6ekd7ft3f76Fv74amrb++hnnyHL5+y07W39zRk4utP2nljL0g87b3tXPbyKVz7pvO1ddv8K/rt1f1J6Yts7VAY8yIUQ4hjgJinlYvv/DQBSypsT8vwNWCal/Kf9fxOwCGWd6nDfWB4pZYUQotjef1pXZTnYF91xt7zR4QhyKm2d1zhFs1JClA4WBav7ckGHsnVAV7IlMjjl7N71g8FzDRPpCzmBQaVcgZYzlQPcm+1lHmRyHioHoWD1yTups/ONmDDt3WTWQ7M6TevrZ2fFTb+g/plnIJxg1XA6yb7oIop//rOu83eR71CIKXUxzwRQbrhOw8AwBOGEUf7E+U8eh4lhWxnCsXQp4y7MPpeJEGr/RM+CmPt3zN3Zaj/73ybmXXCg9M76kjFr3oH2jyk37Y/jMJWmHIpYSUtOSJQsscGpptawsp7GBr8AhynI9KjBrb31gaSBCUtK0t2O+ODOVlt5S6ybTI+DvDQX8j83sGfFM/F55Nacy5AnfJ9sn4u8dDeRqMWWquakgTeJpCjTQ0Gmh9ZwlA17GyEp4IIa3CjM9NASjPBxWUPSHDlQg1OnPnskMuoh2jomKQ3gnSseZ1S6m9qWEOvLG6h95B80v/uuWm7AdLDgpPnMuOlGqppa+bSiiUSi9fVkXvNF0loa2J+ZT/jOBzCzspLmi80bm02a28He+gA7alra5LOv09ET8/A4TbZXN7OtuiXJswYkp0wvxOUw2LivkW1Vtns5bdNOzppVjGEIPi6rZ+d+f3K7EMQ9P9bsrqOsLpBk3XU5DBYfrjxHVu2sZV9jm3ItUNapk6cXAPD+9v3UtoTi1xfb8n3c5FH0BoMhyMVoYE/C/zLUiOCB8ow+wL6FUsoKAFvJKujo5EKIq4CrAMaO7V7nvz17u62ECH4Z+WJyR+eqtzteh2EQ0H25oEPZhgjD9fq1p8/kHGSdcS1new5wb7aXeZDJOQD01TspTm+8d4YraY40WiItHW7va5rffDNZuQIIh2l+4w1opziFq6poePbZtvzhMA1Ll5L/zWtw5Of3WpkMQ2B0McfIaRrYukSHuB0m7i56dAfaP6boHGz6gSLCHWj/tmkCHedzObpe7yzD07WrV8zLoDMmF6R3mibOuJmxH9zdtuHCnyU9Ox2mkTQNoD0ep8mCcZ33H9LcDo6ZlNdxmiONFlpwpG1N2T4qXblv5qa5OCZbsu35h5DBYDyPeHYbkW9dRUF+PgUZyV4CFTfdRX1IKTV5gXqyX3ys00GDkmxvl/U3MT+difmd19/0okymF3VeP7PHZDN7THan6fPH5jB/bOf1d8T43E7TgPgc5f6mPxWsju6a9kManeXpzr5dIqW8F7gX1EhiT/aNUZLt7dFIchL3ppoi0yyLlg6iwfXHCyaR7ssFHcp2kAxeObt3/WDwXMNE+kJOAF66YVB1yrWc7TnAvdle5kEm5wDQ5++k3njvDFdWXJY6H6+/mPLWsm7nrbnrbrWOUgLSsqi+6+5et2JpBiFSqmdlIv347OzufdKTdtpfgwYjnV5cAv2AlAGlCf/HAHu7maerfStt10Ds73ZLffce1y+ehrerISCg/dyWf52zvs01p3AmLLw6nnNF4dmsKziTdTt2s67gLNZ96WPWXbGu31883ZML2s/BSpSFwgR3j6OuTt01gXVXrBvEcnb/+nHU1awoOjvl+g2EbIn0hZwsvBrev1u9WPrJrfhAaDkTSZRzCdsmXt6WdOQ32mQumgU/q1PugYNMzgGgr95JmmFEl9YuzfAmply9f7d6Zv68ftA+O3vSTrtSxjS9R39asFYCU4QQE4By4BLgC+3yPA9cK4R4HOVu0WC7/VV3se/zwBXALfb3c30lQCzSV2dRBE+ens+bG6tpbPbxpHk2mefaUQQbz1QHmHqmGvFYeLUa6/Rmw6IfAQI8WQM2ktxerkRZ2ssp3VlsG/dFJn3xdlh2S5ssleuS5Epz+GiJ+FPONZCWne7K2RgaWtevPcO1nbZHy9m5nJMaHoH8BLmmJchsGGr0FQaVnANAX72TNMOInli7NMMMYb8LEuerDtJnZ0/aaU9cZDUHT78FuYB4RKY/o8LaPiCl/I0Q4moAKeU9dkjcO4AlqJC4X5ZSrupsX3t7HvAkMBbYDVwspUyOk9yOfpls3H4FM8siHtYoVuex9L5e7ay3SSxvoixDXa5EhvP1S0TLOXzlbC9Xoszt8w4Dehrkwt6n199JnaGDXGg0Q5T2z8ph9uzUHBoDHkVwMKFfdBqNRjO8OBgFqz/R7x2NRqMZfnT27unPOVgajUaj0Wg0Go1GM6zRCpZGo9FoNBqNRqPR9BJawdJoNBqNRqPRaDSaXkIrWBqNRqPRaDQajUbTS4zIIBd2iN1dh3iYUUBNLxRnuKDrIxldH8no+khG10cyvVEf46SUg3aVTP3e6ZLhKJeWaWigZRoaDGaZOnz3jEgFqzcQQqwazBGr+htdH8no+khG10cyuj6S0fXRPYZrPQ1HubRMQwMt09BgKMqkXQQ1Go1Go9FoNBqNppfQCpZGo9FoNBqNRqPR9BJawTp47h3oAgwydH0ko+sjGV0fyej6SEbXR/cYrvU0HOXSMg0NtExDgyEnk56DpdFoNBqNRqPRaDS9hLZgaTQajUaj0Wg0Gk0voRUsjUaj0Wg0Go1Go+kltILVQ4QQS4QQm4QQW4UQPxro8vQ3QohSIcSbQohPhRAbhBDfsbfnCiFeFUJssb9zBrqs/YkQwhRCfCiE+Lf9f8TWhxAiWwjxtBBio91Ojhnh9fE9+15ZL4T4pxDCM9LqQwjxgBCiSgixPmFbp3UghLjBfsZuEkIsHphSDy6G47tHCLFTCLFOCLFWCLFqoMtzMPS0bQ8VOpHrJiFEuX291gohzhzIMvaE4dp36UKuoXytPEKID4QQH9ky/cLePqSulVaweoAQwgTuBM4ADgMuFUIcNrCl6nciwA+klDOAo4Fv2XXwI+B1KeUU4HX7/0jiO8CnCf9Hcn38BXhJSjkdmIOqlxFZH0KI0cC3gSOklDMBE7iEkVcfDwJL2m3rsA7s58klwOH2PnfZz94RyzB/95wspZw71Na4SeBButm2hxgPkioXwJ/s6zVXSvliP5fpUBiufZfO5IKhe62CwClSyjnAXGCJEOJohti10gpWzzgK2Cql3C6lDAGPA+cNcJn6FSllhZRyjf27CdV5Ho2qh4fsbA8B5w9IAQcAIcQY4Czg/oTNI7I+hBCZwInA3wGklCEpZT0jtD5sHIBXCOEAfMBeRlh9SCnfBmrbbe6sDs4DHpdSBqWUO4CtqGfvSGbEv3sGKz1s20OGTuQasgzXvksXcg1ZpKLZ/uu0P5Ihdq20gtUzRgN7Ev6XMcQb8qEghBgPzAPeBwqllBWgbnigYACL1t/8GfghYCVsG6n1MRGoBv7Xdpm8XwiRxgitDyllOfAHYDdQATRIKV9hhNZHOzqrA/2cTWW41okEXhFCrBZCXDXQhelFhvP9fa0Q4mPbhXBQu2h1xnDtu7STC4bwtbKnXawFqoBXpZRD7lppBatniA62jcg490KIdOAZ4LtSysaBLs9AIYQ4G6iSUq4e6LIMEhzAfOBuKeU8oIVBbsbvS+yX2nnABKAESBNCXD6wpRr06OdsKsO1To6TUs5HuT5+Swhx4kAXSNMldwOTUG5bFcAfB7Q0B8Fw7bt0INeQvlZSyqiUci4wBjhKCDFzgIvUY7SC1TPKgNKE/2NQ7j4jCiGEE3UjPyqlXGpvrhRCFNvpxahRh5HAccC5QoidKLedU4QQ/2Dk1kcZUGaPNgE8jVK4Rmp9nAbskFJWSynDwFLgWEZufSTSWR3o52wqw7JOpJR77e8q4FmGjyvosLy/pZSVdsfXAu5jiF2v4dp36UiuoX6tYthTDJah5gMOqWulFayesRKYIoSYIIRwoSZiPz/AZepXhBACNb/mUynlbQlJzwNX2L+vAJ7r77INBFLKG6SUY6SU41Ht4Q0p5eWM3PrYB+wRQkyzN50KfMIIrQ+Ua+DRQgiffe+civKRH6n1kUhndfA8cIkQwi2EmABMAT4YgPINJobdu0cIkSaEyIj9Bj4DrO96ryHDsLy/Y51bmwsYQtdruPZdOpNriF+rfCFEtv3bixqo3MgQu1ZCyuHgZdB/2KEu/4yKBvaAlPI3A1ui/kUIcTzwDrCOtjlHP0b5/D4JjEV1Ki+WUg6bCbLdQQixCPh/UsqzhRB5jND6EELMRQX8cAHbgS+jBnNGan38Avg8KtrTh8DXgHRGUH0IIf4JLAJGAZXAz4F/0UkdCCFuBL6CqrPvSin/0/+lHlwMt3ePEGIiymoFyrX4saEoU0/b9lChE7kWoVzOJLAT+EZsTsxgZ7j2XbqQ61KG7rWajQpiYWL3HaSUvxxq/SqtYGk0Go1Go9FoNBpNL6FdBDUajUaj0Wg0Go2ml9AKlkaj0Wg0Go1Go9H0ElrB0mg0Go1Go9FoNJpeQitYGo1Go9FoNBqNRtNLaAVLo9FoNBqNRqPRaHoJrWBpNEMIIUSeEGKt/dknhCi3fzcLIe4a6PJpNBqNZnih3zsaTc/RYdo1miGKEOImoFlK+YeBLotGo9Fohj/6vaPRdA9twdJohgFCiEVCiH/bv28SQjwkhHhFCLFTCHGhEOL3Qoh1QoiXhBBOO98CIcRbQojVQoiX2638rtFoNBpNp+j3jkbTOVrB0miGJ5OAs4DzgH8Ab0opZwEB4Cz7ZXc7cJGUcgHwAPCbgSqsRqPRaIY8+r2j0dg4BroAGo2mT/iPlDIshFgHmMBL9vZ1wHhgGjATeFUIgZ2nYgDKqdFoNJrhgX7vaDQ2WsHSaIYnQQAppSWECMu2yZYW6r4XwAYp5TEDVUCNRqPRDCv0e0ejsdEughrNyGQTkC+EOAZACOEUQhw+wGXSaDQazfBFv3c0IwatYGk0IxApZQi4CPidEOIjYC1w7IAWSqPRaDTDFv3e0YwkdJh2jUaj0Wg0Go1Go+kltAVLo9FoNBqNRqPRaHoJrWBpNBqNRqPRaDQaTS+hFSyNRqPRaDQajUaj6SW0gqXRaDQajUaj0Wg0vYRWsDQajUaj0Wg0Go2ml9AKlkaj0Wg0Go1Go9H0ElrB0mg0Go1Go9FoNJpe4v8DFWrOUn5RA60AAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 864x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def columnSeparator2(df):\n",
    "    df = avgOutPoint1(df)\n",
    "    arr = np.ones((df.shape[0],4))\n",
    "    for i in range(df.shape[0]):\n",
    "        label = df['label'].iloc[i]\n",
    "        if label < 4.5:\n",
    "            arr[i,label-1] = df['AT400(CO2 %)'].iloc[i]\n",
    "        else:\n",
    "            arr[i] = None\n",
    "    arr = np.where(arr!=1, arr, None)\n",
    "    return arr\n",
    "df_set1=list_of_origin_df[2][callback:]\n",
    "df_set2=list_of_origin_df[1][callback:]\n",
    "set1_arr = columnSeparator2(df_set1)\n",
    "set2_arr = columnSeparator2(df_set2)\n",
    "\n",
    "discrete_record_set1, discrete_record_set2  = list_of_origin_df[2]['AT400(CO2 %)'][callback:], list_of_origin_df[1]['AT400(CO2 %)'][callback:]\n",
    "ssae_set1_1,ssae_set1_2,ssae_set1_3,ssae_set1_4 = np.load('results/set1_1.npy'),np.load('results/set1_2.npy'),np.load('results/set1_3.npy'),np.load('results/set1_4.npy')\n",
    "ssae_set2_1,ssae_set2_2,ssae_set2_3,ssae_set2_4 = np.load('results/set2_1.npy'),np.load('results/set2_2.npy'),np.load('results/set2_3.npy'),np.load('results/set2_4.npy')\n",
    "\n",
    "set1_kinetic, set2_kinetic = np.load('results/set1_kinetic.npy'),np.load('results/set2_kinetic.npy')\n",
    "set1_fused, set2_fused = np.load('results/set1_fused.npy'),np.load('results/set2_fused.npy')\n",
    "set1_dae, set2_dae = np.load('results/set1_DAE_16.npy'),np.load('results/set2_DAE_16.npy')\n",
    "fig, axs = plt.subplots(1,2, sharey=False, figsize=(12,4))\n",
    "plt.subplots_adjust(wspace = 0.05, bottom=0.15)\n",
    "fig.set_tight_layout(True)\n",
    "axs[0].scatter(range(set1_arr.shape[0]),set1_arr[:,0],marker='o',label='Conc. @ 1')\n",
    "axs[0].scatter(range(set1_arr.shape[0]),set1_arr[:,1],marker='x',label='Conc. @ 2')\n",
    "axs[0].scatter(range(set1_arr.shape[0]),set1_arr[:,2],marker='s',label='Conc. @ 3')\n",
    "axs[0].scatter(range(set1_arr.shape[0]),set1_arr[:,3],marker='^',label='Conc. @ 4')\n",
    "axs[0].plot(0,linestyle='dashed',color='black',label='Fused')\n",
    "axs[0].plot(set1_fused[:,0],linestyle='dashed',color='tab:blue')\n",
    "axs[0].plot(set1_fused[:,1],linestyle='dashed',color='tab:orange')\n",
    "axs[0].plot(set1_fused[:,2],linestyle='dashed',color='tab:green')\n",
    "axs[0].plot(set1_fused[:,3],linestyle='dashed',color='tab:red')\n",
    "axs[0].plot(0,linestyle='-',color='black',label='SSAE')\n",
    "axs[0].plot(ssae_set1_1,linestyle='-',color='tab:blue')\n",
    "axs[0].plot(ssae_set1_2,linestyle='-',color='tab:orange')\n",
    "axs[0].plot(ssae_set1_3,linestyle='-',color='tab:green')\n",
    "axs[0].plot(ssae_set1_4,linestyle='-',color='tab:red')\n",
    "#axs[0].plot(test_full_values[:,4]*100,linestyle='dashed',color='tab:red',label='Estimation @ point 5')\n",
    "axs[0].set_title('Set-1')\n",
    "axs[0].set_ylabel('CO$_2$ composition(%)')\n",
    "axs[0].set_xlabel('Time')\n",
    "axs[0].legend(loc=(0.01,0.42))\n",
    "axs[1].scatter(range(set2_arr.shape[0]),set2_arr[:,0],marker='o')\n",
    "axs[1].scatter(range(set2_arr.shape[0]),set2_arr[:,1],marker='x')\n",
    "axs[1].scatter(range(set2_arr.shape[0]),set2_arr[:,2],marker='s')\n",
    "axs[1].scatter(range(set2_arr.shape[0]),set2_arr[:,3],marker='^')\n",
    "axs[1].plot(set2_fused[:,0],linestyle='dashed',color='tab:blue')\n",
    "axs[1].plot(set2_fused[:,1],linestyle='dashed',color='tab:orange')\n",
    "axs[1].plot(set2_fused[:,2],linestyle='dashed',color='tab:green')\n",
    "axs[1].plot(set2_fused[:,3],linestyle='dashed',color='tab:red')\n",
    "axs[1].plot(ssae_set2_1,linestyle='-',color='tab:blue')\n",
    "axs[1].plot(ssae_set2_2,linestyle='-',color='tab:orange')\n",
    "axs[1].plot(ssae_set2_3,linestyle='-',color='tab:green')\n",
    "axs[1].plot(ssae_set2_4,linestyle='-',color='tab:red')\n",
    "axs[1].set_title('Set-2')\n",
    "axs[1].set_xlabel('Time')\n",
    "plt.savefig('compare_upper.png',dpi=1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "1c610e9d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Timestamp in selected Test df list: \n",
      "140206_1.xlsx\n",
      "140207_1.xlsx\n",
      "Origin input DF feature num(excl time, incl. label):  91\n",
      "Conc scaler max:  [12.03671837]\n",
      "Minimum call back is:  1\n",
      "The smallest feasible callback is:  1\n",
      "train list:  6\n",
      "Train df(1st one) shape before POD:  (117, 97)\n",
      "Test df(1st one) shape before POD:  (50, 97)\n",
      "Truncation parameter:  32\n",
      "Train:  0.00020657505898268987\n",
      "Val:  0.00020325999653631073\n",
      "Test:  0.005828281701551702\n",
      "Train df(1st one) shape after AE:  (117, 39)\n",
      "Test df(1st one) shape after AE:  (50, 39)\n",
      "Number of set:  100\n",
      "Number of set:  33\n",
      "Number of set:  50\n",
      "Number of set:  33\n",
      "Number of set:  201\n",
      "Number of set:  211\n",
      "Number of set:  33\n",
      "Number of set:  101\n",
      "RMSE of SSAE on test:  0.4119852671537521\n",
      "RMSE of SSAE on val:  2.381124839767557\n"
     ]
    }
   ],
   "source": [
    "# Note: Switch to set3 scenario i.e., test: [1,2]\n",
    "list_of_processed_df=list_of_origin_df.copy()\n",
    "test_df_index_list=[1,2] # Select the index of test dataset\n",
    "callback=17\n",
    "train_sample_set, train_label_set, test_sample_set, test_label_set, conc_scaler,train_df_list,test_df_list,test_conc_list \\\n",
    "=getDataSet(data_path_list,list_of_processed_df, callback,test_df_index_list, 'POD', 32)\n",
    "n_set = train_sample_set.shape[0]\n",
    "train_index = list(set(range(0,n_set,1))- set(range(0,n_set,5)))\n",
    "test_index = list(set(range(0,n_set,5)))\n",
    "x_train = train_sample_set[train_index]; x_val = train_sample_set[test_index]\n",
    "y_train = train_label_set[train_index]; y_val = train_label_set[test_index]\n",
    "y_val\n",
    "\n",
    "set3_ssae = np.zeros((conc_scaler.inverse_transform(test_label_set.reshape(test_label_set.shape[0],6)).shape))\n",
    "for i in range(6):\n",
    "    set3_ssae[:,i] = np.load('results/set3_'+str(i+1)+'.npy').ravel()\n",
    "print('RMSE of SSAE on test: ',rmse(set3_ssae, conc_scaler.inverse_transform(test_label_set.reshape(test_label_set.shape[0],6))))\n",
    "set3_ssae = np.zeros((y_val.shape[0],6))\n",
    "for i in range(6):\n",
    "    set3_ssae[:,i] = np.load('results/set3_val_'+str(i+1)+'.npy').ravel()\n",
    "print('RMSE of SSAE on val: ',rmse(set3_ssae, y_val))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a5949bd2",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.8.8 ('base')",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  },
  "vscode": {
   "interpreter": {
    "hash": "ad2bdc8ecc057115af97d19610ffacc2b4e99fae6737bb82f5d7fb13d2f2c186"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}