sec3_1_preprocessing.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load libraries\n",
    "import utils as u\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.decomposition import PCA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.rcParams.update({\n",
    "    'axes.titlesize': 14,     # Title font size\n",
    "    'axes.labelsize': 14,     # X and Y axis labels font size\n",
    "    'xtick.labelsize': 12,    # X tick labels font size\n",
    "    'ytick.labelsize': 12,    # Y tick labels font size\n",
    "    'legend.fontsize': 14,    # Legend font size\n",
    "    'figure.titlesize': 14    # Figure title font size\n",
    "})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle as pkl\n",
    "\n",
    "# load the stored data\n",
    "file_name = 'sec2_data_output.pkl'\n",
    "\n",
    "with open(file_name, 'rb') as f:\n",
    "    data = pkl.load(f)\n",
    "\n",
    "wl = data['wavelength']\n",
    "df = data['df']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['manga_id', 'teff', 'teff_err', 'logg', 'logg_err', 'feh', 'feh_err',\n",
       "       'flux', 's2n', 'teff_class', 'teff_label', 'logg_label', 'feh_class',\n",
       "       'feh_label'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = np.stack(df['flux'].values, axis=0)\n",
    "\n",
    "y_t = np.array(df['teff_label']).astype('int')\n",
    "y_g = np.array(df['logg_label']).astype('int')\n",
    "y_m = np.array(df['feh_label']).astype('int')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature selection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wavelength range: 3621.60  10353.81\n",
      "Number of values per spectrum: 4563\n"
     ]
    }
   ],
   "source": [
    "print('Wavelength range: {:.2f}  {:.2f}'.format(wl[0], wl[-1]))\n",
    "\n",
    "spec_size = X[0].shape[0]\n",
    "\n",
    "print('Number of values per spectrum:', spec_size)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "   **Lines to be selected**\n",
    "   1. Ca II K (3934 A) & H (3968 A) lines: Key indicators in A-type stars, showing strength variances linked to temperature and luminosity effects (Gray & Corbally 2009)\n",
    "   2. Fe I (4046 A): Often used with hydrogen line Hδ for temperature classification, especially in F-type and later-type stars (Gray & Corbally 2009)\n",
    "   3. 2 spectral lines can help us classify stars from B to M (Brice & Andonie 2019):\n",
    "      - H_delta (4102 A): present in A, B, F, G\n",
    "      - Ca I (4227 A): present in F, G, K, M\n",
    "   4. G-band CH (4300 A): Prominent in late-G to K-type stars and is sensitive to surface gravity (Gray & Corbally 2009)\n",
    "   5. He II (4686 A): dominates O-type stars (Gray & Corbally 2009)\n",
    "   6. TiO band (4955 A): at least one TiO band is needed for M-type (Gray & Corbally 2014)\n",
    "   7. Fe I (5269 A) & Fe II (5018 A): for metallicity classification (Santos et al. 2004) [DOI: 10.1051/0004-6361:20034469]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wavelength bounds (3934): 3926.45 - 3940.94\n",
      "Wavelength bounds (3968): 3960.96 - 3975.58\n",
      "Wavelength bounds (4046): 4038.31 - 4053.22\n",
      "Wavelength bounds (4102): 4094.49 - 4109.60\n",
      "Wavelength bounds (4227): 4218.91 - 4234.48\n",
      "Wavelength bounds (4300): 4292.40 - 4308.24\n",
      "Wavelength bounds (4686): 4677.35 - 4694.62\n",
      "Wavelength bounds (4955): 4945.38 - 4963.64\n",
      "Wavelength bounds (5018): 5008.41 - 5026.90\n",
      "Wavelength bounds (5269): 5258.96 - 5278.37\n",
      "\n",
      "Final shape: (59085, 170)\n"
     ]
    }
   ],
   "source": [
    "spec_lines = {'line_wl':\n",
    "    [3934, 3968, 4046, 4102, 4227, 4300, 4686, 4955, 5018, 5269],\n",
    "              'bounds':\n",
    "    [8]*10}\n",
    "\n",
    "new_X = u.select_features(X, wl, spec_lines['line_wl'][0])\n",
    "for i in range(1, len(spec_lines['line_wl'])):\n",
    "    # print(f\"Line {i+1} of {len(spec_lines['line_wl'])}, Current shape: {new_X.shape}\", end=\"\\r\")\n",
    "    new_X = np.concatenate((new_X, u.select_features(X, wl, spec_lines['line_wl'][i], spec_lines['bounds'][i])), axis=1)\n",
    "print(f\"\\nFinal shape: {new_X.shape}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# PCA plot"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is only done to plot the PCA plot and choose the number of features. Neither the scaler nor the PCA object fitted using the entire dataset are used later. New scalers are fitted using only training sets to transfrom the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = MinMaxScaler()\n",
    "# fit the scaler and transform the data\n",
    "X_scaled = scaler.fit_transform(new_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "pca = PCA(n_components=X_scaled.shape[1])\n",
    "# fit the PCA object and transform the data\n",
    "X_pca = pca.fit_transform(X_scaled)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "More than 99.95% of the variance in the data is explained by 9 PCA components.\n"
     ]
    }
   ],
   "source": [
    "n_comp = np.where(np.cumsum(pca.explained_variance_ratio_) >= 0.9995)[0][0] + 1\n",
    "print(f\"More than 99.95% of the variance in the data is explained by {n_comp} PCA components.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAnYAAAHWCAYAAAD6oMSKAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAAB6fElEQVR4nO3dd1yT1/4H8E8S9t44QcU9UKlI1YqjrlpHvSpqW3eX1lF7bdW21lG1Vq1Wb2+1w6rV+rOOWkevraMORLTu1roHQwQBgSAIgSTn9wclNSQgPARCwuf9euWFOc95Tr4JKF/PlAkhBIiIiIjI4snNHQARERERmQYTOyIiIiIrwcSOiIiIyEowsSMiIiKyEkzsiIiIiKwEEzsiIiIiK8HEjoiIiMhKMLEjIiIishJM7IiIiIisBBM7IiIiIitR5RK7rKwszJkzB3369IGXlxdkMhnWr19f6vszMjLw2muvwdfXF87OzujWrRvOnTtntO7u3bsREhICBwcHBAQEYM6cOVCr1eVqk4iIiMhcqlxil5qaivnz5+PKlSto3bp1me7VarV4/vnnsXnzZkyaNAlLlixBcnIyunbtihs3bujV3bdvH1544QV4eHjgP//5D1544QUsWLAAkydPltwmERERkVmJKiY3N1ckJiYKIYQ4ffq0ACDWrVtXqnt/+OEHAUBs27ZNV5acnCw8PDzEiBEj9Oo2b95ctG7dWuTn5+vK3n//fSGTycSVK1cktUlERERkTlWux87e3h41atSQdO/27dvh7++Pf/3rX7oyX19fREREYNeuXVCpVACAy5cv4/Lly3jttddgY2Ojqztx4kQIIbB9+/Yyt0lERERkblUusSuP8+fPIyQkBHK5/ttq3749Hj16hOvXr+vqAUC7du306tWqVQt16tTRXS9Lm0RERETmZvPkKpYjMTER4eHhBuU1a9YEANy7dw+tWrVCYmKiXnnRuvfu3Stzm8aoVCq9Hj2tVou0tDR4e3tDJpOV4Z0RERFRdSWEwMOHD1GrVi2DjqairCqxy8nJgb29vUG5g4OD7vrjX4urm5mZWeY2jfn4448xb968MrwDIiIiIuPi4+NRp06dEutYVWLn6OhodM5bbm6u7vrjX4urW3i9LG0aM2vWLLz99tu650qlEgEBAYiPj4ebm1tp3hIRERFVc5mZmahbty5cXV2fWNeqEruaNWvqhlkfV1hWq1YtXb3C8rp16xrUbd++fZnbNMbe3t5ob5+bmxsTOyIiIiqT0kzjsqrFE23atMG5c+eg1Wr1yk+dOgUnJyc0btxYVw8Azpw5o1fv3r17uHv3ru56WdokIiIiMjeL7bFLTEyEUqlEUFAQbG1tAQBDhgzB9u3b8eOPP2LIkCEACjY83rZtG/r376/rPWvRogWaNm2Kr776Cq+//joUCgUAYPXq1ZDJZLp7y9ImERERWQchBLQCUGu10GgF1FoB7d9fC59rNAIaIaDRaqHWCqg1AlrxWB1NwVddHc0/92qF0Huueex1jL1eVtbDUsdeJRO7zz//HBkZGbrVqXv27MHdu3cBAJMnT4a7uztmzZqFDRs24M6dO6hXrx6AgiTs6aefxtixY3H58mX4+Pjgiy++gEajMVjEsHTpUgwYMAC9evXC8OHDcenSJXz++ed45ZVX0KxZM129srRJRERETyaEQL5GIE+jRZ5ai/y/v6oe+3Ne0a9luFbQjkCeWvP3NVHQfpHX0/1Zo59YabTC3B+RHq3qUanryoQQVSt6APXq1UNsbKzRa4WJ3JgxYwwSOwBIT0/HO++8g59++gk5OTkIDQ3FsmXLDPasA4CffvoJ8+bNw5UrV+Dr64sxY8bgww8/1PUASmmzJJmZmXB3d4dSqeQcOyIiqjRCCKiKJEKFiVRBMqT558+P1Stapv9cA1V+kTYfq1PwVYM8jRb5amFQj0pPq3qE+M8iSpU/VMnEzloxsSMioscJUZDw5ORpkJ2nwSOVuuBrnhqPVBpk56nxKE9T8Hj82t9fs1Ua5ORrDBIy/eRMg3wNf9VbsrIkdlVyKJaIiKiqUWu0+omVLvEq+vyfpOtRXkEylpOnQbaq4Fp2nlrvubqKDfuRNLYKGRRyGRSygq82CnnBV7kMcpkMNn9ft5HLoJDLC8p1z//5qtArK2hDnZuN1aWMg4kdERFZPY1WICtXjczcfChz8pGZm4/MHPXfX/ORmav++6t++cNctS45y1Nz+LAyyGSAnUIOOxs57G3ksP37z4VltkWvGS2X/X2P4u9rMtjb6NezU8hhayOH/d9f9RMrecnJluKxBO7vBK0iZWZmYvW40tVlYkdERFWeVivwUGU8+TKWlClz/knMMnPy8VClNvdbqLIKE6PHk6fCBOnxssef29so/rmnSB17vXJFkToyvWRL91oKBWxtZLBTyGGjsKqd2CodEzsiIqoUQghk52nwIEuF1Kw8pGXn6RKwf5Ix4wlblkqN6jIj3MFWDic7GzjZKeBsZwMnewWc7BRwsrOBs50Cjn9/dbIv+OpgqyiSTD2enCkMEjS9xE0h59nlVoaJHRERSaZSa5CWnYcHWXlIzVLhQVYeHmSr8ODvsgdZ//w5NUsFlRUNZ9op5HCyL0i+HO0UBcmWnQ2c7R9LvnTP/07S7BRwtrcxeF6QuBXUV1TwsB5ZNyZ2RESko9EKZDzKw4PsxxK1wuQs++8/Z/1z/WGu5Q1xKuQyuDnYwM3RFm4OtnBztIGbgy3cHW3/LvvnmquDzWOJV0GSVtibZsshQ6qCmNiZU1oaoNFIu9fBASjuMOCMDCA/X1q7dnaAu7vxa0olkJcnrV1bW8DDw/i1hw+B3Fxp7SoUgJeX8WvZ2cCj0m/qqEcmA3x8jF/LyQGysqS1CwC+vsbLVSogM1N6u97egNzIL5q8vILvnVSenoCNkX8q1GogPV16u+7uBT9vRWm1wIMH0tt1cwOKOxEmJUV6uy4ugKOj8WupqZA8TujkBDg7G79mgn8jhBDIUql1PWmpWXklJGsqpGXnoaov0pTLYJCU/ZOY/f286J8fq+tkp+DwI1ktJnbm1LkzcPmytHsnTgT++1/j1154ATh6VFq7Q4YA27YZv/bKK8D27dLa7dIFOHLE+LWZM4EvvpDWbvPmwF9/Gb+2dCkg9XQQH5/ik4BvvwUmTZLWLlB8ArB7NxARIb3d5GTjSeOJE0C3btLbvXQJaNHCsPzaNaBlS+ntHj4MdO1qWP7gAeDnJ73drVuBoUONXytPu59/Drz5pvFrzZoVJHdSzJkDzJ1r/Fo5/o3YFz4IH/V5E6nZeVVyNae9jRw+LvbwcPon6XLXJWC2RXrU9JM0ZyZmRMViYkdEZIVSHubhnlJiT7gECrkMXs528Ha2g4+LPbxd7ODtXPjVDt5/l/n8XcZeM6KKwcSOiKgKyFapEfvgEeLSshH74BFi0x5hYnoO6pgxJndHW71krDBZ83H5O1Fz/qfM3dG2wvfyIqInY2JHRGRGP5yOw9IFB5CaZTh/dUyexPl1xXCyU/zTk1aYlP2doD3ey+bjYgdPZzsuDiCyQDwrthIZnBXLxRMFuHiiABdPFLDgxRNqjRZJylzczXiE+LQcJGTkID7tEe6m5yAh/REe5RvOdcuxsUeOnYPRZt1zHkIhSjc/zlYhR20PBzS1zcOi94fpXUu4FgPPerXhZMf/yxNZorKcNc/ErhKV5RtDRFWTEALxaTm4mpT595Bp9t9DqI+QkJ5Toed+ujnYINDbGQHeTgj0ckKgtxMCvJwR6O2EGm4OBUOhKSmGi0SKW1hDRBahLPkD//tGRFQMrVbgzoNsXEpQ4q97mfjzrhJ/3VMiswL3bqvp7oCAv5O2QG/nf/7s5Qx3J9sKe10isg5M7IiIAKg1WtxOLUji/kxQ4q+ETPx1T4lsE89zs1PIUcfL8e8et8cSN28n1PF0goOtwqSvR0TVCxM7Iqp28jVa3LifhUsJSly6p8SlBCUuJ2Yi18gcOClcHWx0vWyFw6YBf/fA1XBz4JFRRFRhmNgRkVVTqTW4npSFPx9L4q4mPSz3pr1eznZo6OtikLgFejnBw8mWe7QRkVkwsSMiq5Gbr8HlxEz8laDEpYRM/JmgxPX7D8u9oMHP1R6tarujRW13tKzlhlZ13FHDzYHJGxFVOUzsiMgiZavUuJyYqTcn7mZKFjTlTOJqeziiRS03tKztXpDM1XKDn5vx7UiIiKoaJnZEVOVl5ubjr4RMvTlxt1Oziz12t7QCvJzQsrYbWtT6J4nzdilmDzwiIgvAxI6IqpTcfA3Oxqbjj7v/JHGxDyRuNP2YBj7OaFHbHa1qu6FlLXe0qOXO7UOIyOowsSMisxJC4EZyFo5dT8HR6yn4/U4aVOVY2CCXAUG+Lnpz4prXcoOrA5M4IrJ+TOyIqNJlPMrD8ZupOHY9BceupyIpU9qRcjZyGRr5u6Ll33PiWtZ2R7Oarjw6i4iqLf7rR0QVTq3R4uLdDBy9XpDM/XE3A2Vd42CnkKNJDVe0rP13ElfLHU1quHJDXyKixzCxI6IKkZCR83ePXAqibqaW6RguuQxoVccDwbXddYsbGvu7ws5GXoERExFZPiZ2RGQSOXkanLzzQJfM3UrJLtP9tT0cEd7YB+GNfNGxoQ/cHTknjoiorJjYEZEkQghcu/9QN0/u95i0Mp3m4GArx9MNvBHeyBfhjX0R5OvMDX9NwcUF+PxzwzIiqhaY2BFRqaVn5yHy70UPkTdScD9TVab7m9ZwRXhjX4Q38kW7ep6cH1cRHB2BN980dxREZCZM7IioWGqNFufjM3TDq38kKMu0KbCnky2eaeSL8EY+CG/sC3+e4EBEVKGY2BGRnvi0Rzh2oyCRO3HzAR6qSr/oQSGXoW1dj4Jeuca+aFXbHQo5h1eJiCoLEzuiau5Rnhonbz/Asb+3IrmdKmXRgy+6NPZBx4Y+cONGwEREZsPEjqiaEULgatLfix5upOD0nXTkaUq/6MHRVoGnG3jpeuUa+HDRAxFRVcHEjqiayHiUh//7PR6bTsYiISOnTPc2reGKLn8ncu3qecLehoseiIiqIiZ2RFbuxv2HWHciBj+eu4vc/NL1zHk52+GZhj5/r2D1gR8XPViO1FSgWTP9sitXAB8f88RDRJWKiR2RFdJqBY5eT8G3UXcQeSP1ifUVchmeCvAs2CC4sS9a1nKHnIseLJMQBcld0TIiqhaY2BFZkWyVGtvP3sWGEzFPXARR18tRtzlwxyBvuHLRAxGRxWNiR2QF4tMeYcOJGPxwJh4PSziT1UYuQ99WNTG2Uz20qevBRQ9ERFaGiR2RhRJC4NSdNKyLuoMDl+9DW8Jom5ezHV5sH4CXnw5EDXfOlyMislblTuwuX76Mq1evIjs7GyNHjjRFTERUgtx8DfZcvIdvo2JwJTGzxLpNa7hibKd6GNimNo/vIiKqBiQndqdPn8arr76KP//8U1dWmNgdO3YMffr0wZYtWzBgwIDyR0lESM7MxaaTsfj+VBweZOcVW08mA55t6o9xz9RDhwbeHG4lIqpGJCV2f/31F7p37w65XI5p06bh6tWr2Ldvn+56586d4ePjg23btjGxIyqnP+5mYF1UDPb+cQ/5muLHW13sbRDRri5GdwxEoLdzJUZIRERVhaTEbs6cOQCAs2fPomHDhpg3b55eYieTydChQwecPn3aNFESVTNqjRa//nUf66Lu4Exseol1A72dMKZjPQx5qg5XthIRVXOSErujR49i8ODBaNiwYbF1AgIC8Msvv0gOjKg6KjwdYmN0DO4pc0us26mhN8Z1qo9uTfy45xwREQGQmNg9fPgQfn5+JdbJycmBRqORFBRRdVPa0yHsbeT4V0htjOlYH01quFZihEREZAkkJXZ169bVWzRhzLlz5xAUFCQpKKLqoCynQ9Rwc8DIDoEY0T4AXs52lRQhERFZGkmJXb9+/bBq1SocPHgQPXr0MLi+detWnDx5ErNnzy53gETWJlulxo5zd7E+6smnQ7QN8MDYTvXxXMsasFXIKylCIiKyVJISu/feew/bt29H3759MXr0aCQlJQEAvvjiC0RHR+P//u//UK9ePbz99tsmDZbIkpXldIjng2tibKf6aFPXo/ICJCIiiycTQtrp0Ldv38bIkSMRHR1tcC0sLEyX3NE/MjMz4e7uDqVSCTc3N3OHQ5WgLKdDeDrZ4qWwQJ4OQeWTkgIUnQOdnAz4+ponHiIqt7LkD5I3KG7QoAGioqJw4cIFnDx5EmlpaXBzc0NYWBhCQ0OlNktkFXg6BJmNkxPw95ZUemVEVC1I7rGjsmOPnfV7kKXChuhYfH8ylqdDEBGRSVR4j51SqURsbCwaNmwIJyP/E8zOzsatW7dQr149JjBUbfx+Jw2vbTyDjEf5xdbh6RBERFSRJC2zmz9/Pjp16lTsPnUajQadOnXCwoULyxUckaXY92ciXl57qtikLtDbCXP6N0f0rO74sH9zJnVERFQhJPXY/fLLL+jZsydcXY1vkOrm5obevXvjf//7Hz755JNyBUhU1a2PuoN5ey/D2KSGTg29MbZjfXRr6gcFT4cgIqIKJimxi4uLQ79+/UqsExQUhAMHDkgKisgSaLUCn/x6FV8evW1wrWsTX8x8rima1uBUBCIiqjySEjuZTAaVSlViHZVKxSPFyGrlqbV4d/tF/HThnsG1l8ICMG9AC9hwQ2EiIqpkkhK7pk2b4pdffoEQwuhqPq1Wi3379qFJkyblDpCoqnmYm483Np1F1M0HBtem92qMN7s15CpXMp+0NKBzZ/2yyEjAy8s88RBRpZLUpTBixAhcv34d48aNg1Kp1LumVCoxbtw43Lx5Ey+//LJJgiSqKu5n5iLiy5MGSZ1CLsPSIcGY1L0RkzoyL40GuHxZ/8HRE6JqQ9I+dvn5+ejWrRtOnDgBDw8PhIaGonbt2khISMDp06eRkZGB8PBwHDhwALa2thURt0XiPnaW7WbyQ4z+9jQSMnL0yp3sFPjipRB0beJXzJ1ElYgnTxBZnQrfx87W1hYHDx7EBx98gK+//lpvkYSbmxveeecdzJ8/n0kdWY0zMWkYv+EMlDn625n4uNjh2zGhCK7jYZ7AiIiIHlPukyc0Gg2uXr0KpVIJDw8PNGnSBAoFj0Uyhj12lumXS0mYuuU8VGqtXnl9H2dsGNseAd48romqEPbYEVmdSjkrtpBCoUCLFi3K2wxRlfRddAzm7P7LYI+61nU98O3odvB2sTdPYEREREaUO7EjskZCCCz99Rq+OHLL4NqzTf3wnxfbwsmOf32IiKhqkbzR1sGDB9G3b1/4+vrC1tYWCoXC4GFjw198ZHny1Fr8e9tFo0ndiPZ18eXIp5jUERFRlSTpt9OOHTswbNgwaLVaBAYGomnTpkziyCpkqdSYsOksIm+kGlyb1qMxpjzLPeqIiKjqktRjN3/+fDg6OuLgwYO4c+cOIiMjcfjwYaMPKVQqFWbMmIFatWrB0dERYWFhpT6ebMuWLQgJCYGDgwN8fX0xfvx4pKYa/pK+f/8+xo4dCz8/Pzg6OiIkJATbtm0z2ubBgwfRrVs3+Pj4wMPDA+3bt8fGjRslvTequpIf5mLYl9EGSZ1CLsMng1thag/uUUdERFWbpMTu2rVrGD58OLp3727qeAAAY8aMwfLly/HSSy9h5cqVUCgU6Nu3L44fP17ifatXr8aIESPg5eWF5cuX49VXX8WWLVvw7LPPIjc3V1cvMzMTzzzzDHbs2IHXX38dy5Ytg6urKyIiIrB582a9Nnfv3o1evXohLy8Pc+fOxcKFC+Ho6IhRo0ZhxYoVFfL+qfLdSsnCv744gb/uZeqVO9oq8M2odhgWGmCmyIiIiEpP0nYntWvXxpAhQ7By5UqTB/T7778jLCwMS5cuxfTp0wEAubm5aNmyJfz8/HDixAmj9+Xl5cHf3x/BwcE4cuSIrmdl79696N+/P1atWoXJkycDAJYuXYp3330Xhw4d0iWnWq0WTz/9NOLj4xEbGws7OzsAQK9evfDXX3/h9u3bsLcvWAGpVqvRtGlTODs74+LFi6V+b9zupGo6G5uO8RtOI+OR/h513s4Fe9S1ruthnsCIpOB2J0RWpyz5g6QeuyFDhuDgwYNQq9WSAizJ9u3boVAo8Nprr+nKHBwcMH78eERHRyM+Pt7ofZcuXUJGRgaGDRumN1zWr18/uLi4YMuWLbqyyMhI+Pr66vU4yuVyREREICkpCUePHtWVZ2ZmwtPTU5fUAYCNjQ18fHzg6OhokvdM5rP/ryS8+PVJg6Qu0NsJOyZ0ZFJHREQWRVJit2jRInh4eGDYsGGIi4szaUDnz59H48aNDTLS9u3bAwAuXLhg9D6VSgUARpMtR0dHnD9/HlqtVlfXWD0np4KNZs+ePasr69q1K/766y/Mnj0bN2/exK1bt/DRRx/hzJkzePfdd8v+BqnK2HQyFm9sOmuw8XBwHXfsmNAR9XyczRQZERGRNJKWsrZq1Qr5+fk4efIkfvrpJ3h4eMDd3d2gnkwmw61bhltGlCQxMRE1a9Y0KC8su3fvntH7GjUqmNgeFRWFsWPH6sqvXbuGlJQUAEB6ejq8vb3RpEkTHDx4ELGxsQgMDNTVjYyMBAAkJCToymbPno07d+5g4cKFWLBgAYCCBHDHjh0YOHBgie9FpVLpEk6goPePzE8IgU/3X8fnh28aXOvWxBefvxgCZ3uu8iYiIssj6beXVquFjY0NAgL+mVBubKqelNPKcnJy9IY9Czk4OOiuG+Pj44OIiAhs2LABzZo1w6BBg5CQkIDJkyfD1tYW+fn5untfeeUVrFmzBhEREVixYgX8/f2xdetW7Ny50+A17O3t0bhxYwwZMgT/+te/oNFo8NVXX+Hll1/GgQMH8PTTTxf7Xj7++GPMmzevzJ8BVZx8jRazfvwT28/eNbgW0a4OFg1qBRuF5O0diczPwQGYONGwjIiqhXKfFWtqLVu2hL+/Pw4dOqRXfvnyZbRo0QJr1qzB66+/bvRepVKJUaNGYffu3bqyl19+GY8ePcKPP/6I9PR0eHh4ACiYy/fGG2/gwYMHAIAaNWpgzpw5mDBhAqZOnYrPPvsMAPDGG2/g5MmTOHfuHOTygl/4+fn5aNGiBTw9PXHq1Kli34uxHru6dety8YSZZKvUmPD9ORy7nmJwbcqzjTCN25kQEVEVVOGLJypSzZo1kZiYaFBeWFarVq1i73V3d8euXbsQGxuLo0ePIiYmBhs3bkRiYiJ8fX11SR1QsADk3r17+P333xEdHY3Y2Fg0aNAAANC4cWMABStt165di+eff16X1AGAra0tnnvuOZw5cwZ5eXnFxmNvbw83Nze9B5lHykMVhn910iCpk8uAj//VCm/3bMykjoiILF6Vm0jUpk0bHD58GJmZmXqJUGHPWJs2bZ7YRkBAgG6YOCMjA2fPnsXgwYMN6tnZ2SE0NFT3/ODBgwCAHj16AAAePHgAtVoNjUZjcG9+fj60Wq3Ra1S13E7Jwuh1vyM+TX8Y38FWjs9HhKBHc38zRUZERGRa5RqKjY6OxsGDB3Hv3j29IUdd4zIZ1q5dW6Y2T506haefflpvHzuVSoWWLVvC29sbJ0+eBADExcXh0aNHaNq0aYntTZgwAV999RVOnjypl8QVdePGDTz11FPo0qUL9uzZAwDQaDTw8fGBn58f/vzzT93edllZWWjWrBlcXFxw5cqVUr837mNX+c7HpWP8hjNIy9bvWfVytsPa0e3QNsDTTJERERGVTlnyB0k9dmq1GiNGjMCPP/4IIQRkMpneQonC51ISu7CwMAwdOhSzZs1CcnIyGjZsiA0bNiAmJkavrVGjRuHo0aN6r7t48WJcunQJYWFhsLGxwU8//YT9+/djwYIFBkld8+bNMXToUAQEBODOnTtYvXo1vLy8sGbNGl0dhUKB6dOn44MPPsDTTz+NUaNGQaPRYO3atbh79y42bdpU1o+OKtGhK/fx5uZzyM3X386krpcjNoxtjwa+LmaKjIiIqGJISuw+/fRT7NixA+PGjcPEiRPRrl07vPXWWxg2bBiOHTuGxYsXo0ePHvjkk08kBfXdd99h9uzZ2LhxI9LT0xEcHIy9e/ciPDy8xPtatWqFnTt3Yvfu3dBoNAgODsbWrVsxdOhQg7qtW7fGunXrcP/+fd2K2nnz5sGvyI7t77//PurXr4+VK1di3rx5UKlUCA4Oxvbt240O71LVsPlUHD746U9oi/RHt6rtjm/HhMLX1XDlNRERkaWTNBQbHBwMAPjjjz8AFJzaMHfuXHz44YcAgL/++gvt27fH8uXLi13BWh1xKLbiCSGw4uANrDp0w+BaeGNfrH6Je9SRlcvIAF54Qb/sp5+AxxaPEZFlqfCh2Js3b+KVV17RPZfJZMjP/+dIphYtWqB///5YvXo1EzuqNPkaLd7f+Se2njHco25wSB0sHtwKttyjjqxdfj7w2LGIujIiqhYk/Zazs7PTHb8FAC4uLkhOTtarExgYiBs3DHtNiCpCtkqNV787YzSpm9y9IZYNDWZSR0REVk9Sj13dunURHx+ve960aVMcO3ZMt2ACAE6ePAkvLy/TRElUgtQsFcatP40/7ir1yuUyYP7Alnj56cBi7iQiIrIukrowunTpokvkAGDYsGG4du0a+vXrh//+978YMWIEjh8/jj59+pg0WKKiYlKzMXj1CYOkzt5GjjUvP8WkjoiIqhVJPXbjxo2DRqNBQkIC6tSpg8mTJ+PIkSPYu3cv9u3bBwBo3749Fi9ebNJgiR53IT4D49efxoMie9R5ONli7ehQPBXIPeqIiKh6MelZsWfOnMGtW7cQGBiI9u3b6x3DRVwVa0q/Xb2PN78/j5x8/ZM/6ng6YsO49gjiHnVUXaWkAEW2bUJyMuDra554iKjcKnxVbHHatWuHdu3ambJJIgM/nI7DezsvQVNkk7oWtdywbmwo/FwdzBQZERGReXFDL7IYQgisOnQTKw5eN7jWuZEPVr/8FFy4Rx0REVVjpfotOG7cOMhkMixatAj+/v4YN25cqRqXcqQYUXFWHLiOVb/dNCj/V9vaWDw4GHY2HPonIqLqrVRz7ORyOWQyGa5cuYLGjRuXeu6cTCaDRqN5csVqgnPspEtS5uKZT36Dusjw68SuQXindxPdNjtE1R7n2BFZHZPPsbtz5w4AoHbt2nrPiSrL8ZupekmdTAbMG9ACozrUM19QREREVUypErvAwMASnxNVtBO3UvWe925eg0kdERFREZImJTVo0ABvvvmmqWMhMkoIgRM3H+iVdWrobaZoiIiIqi5JiV1qairniFGluZ2ajaTMXL2yDkE+ZoqGiIio6pK0N0RwcDCuXzfccoKoIpy4pd9b5+9mjyBfZzNFQ1TF2dkBQ4YYlhFRtSApsZsxYwYGDx6Mw4cPo1u3bqaOiUjPiZv68+s6BflwFSxRcdzdgW3bzB0FEZmJpMQuPT0dvXr1Qq9evfDCCy8gNDQU/v7+Rn/Zjho1qtxBUvWl1QpE39bvsesQxPl1RERExkg6K7ZwX7uitz6e2AkhuI9dEdzHruwuJSjR7z/H9cqiZnZHbQ9HM0VERERUuSr8rNh169ZJCoyorKKLzK+r5+3EpI6IiKgYkhK70aNHmzoOIqOiiuxf17EhV8MSEREVh4drUpWVr9Hi9ztpemUdOb+OiIioWJJ67B6n0WiQmpoKlUpl9HpAQEB5X4KqqYvxGXiUpz9Hs0MDJnZEJVIqgVde0S/75puC1bJEZPUkJ3Znz57Fe++9h2PHjiEvL89oHZlMBrVaLTk4qt6iipw20bSGK7xd7M0UDZGFyMsDtm/XL/viC/PEQkSVTlJid+HCBXTu3Bk2Njbo1asX9uzZg9atW6NGjRo4d+4cUlJS0LVrV54pS+VS9HzYTpxfR0REVCJJc+w++ugjAMCpU6ewa9cuAMCgQYOwb98+xMTE4I033sClS5cwZ84c00VK1UpOngbn4zL0yng+LBERUckkJXbHjx/HgAED0KxZM11Z4Z52jo6O+Pzzz1GrVi289957pomSqp0zsWnI02h1zxVyGULreZkxIiIioqpPUmKnVCrRoEED3XNbW1tkZWX906hcjq5du+LQoUPlj5CqpaLz61rXcYerg62ZoiEiIrIMkhI7Pz8/pKen657XqFEDN27c0KuTm5uLR48elS86qraii+5fF8T5dURERE8iKbFr3rw5rl27pnveqVMn7N+/H9HR0QCAK1euYOvWrWjatKlpoqRqRZmTjz8TlHplHTm/joiI6IkkJXbPP/88jh07hsTERADAjBkzIITAM888A19fX7Rq1QoZGRmcY0eSnLr9ANrHjiG2t5EjJMDTfAERERFZCEmJ3RtvvIGEhAR4exf0orRu3RqHDh1Cnz594OPjgx49emDPnj0YNGiQSYOl6uFEkfNh29XzhIOtwkzREBERWQ5J+9jZ2trC399fr6xjx474+eefTRIUVW9RNzm/joiISApJPXZc7UoVJflhLm4kZ+mV8XxYIiKi0pGU2PXs2RMBAQGYOXMm/vzzT1PHRNVYdJFhWFd7G7SqzTMuiYiISkNSYjdp0iSoVCosWbIEbdq0Qdu2bbF8+XLdYgoiqU4U2b8urIEXbBSSfkyJiIiqHUm/MVetWoV79+5h9+7dGDJkCK5fv47p06cjICAAvXv3xqZNm7iHHUkSxf3riIiIJJPcFaJQKNCvXz/88MMPSEpKwjfffINnnnkGBw8exOjRo+Hv74+RI0eaMlaycvFpj3A3PUevjPvXEZWRrS3QpYv+w5anthBVFzJReMiridy9exdffvkllixZArVaDY1GY8rmLVpmZibc3d2hVCrh5uZm7nCqnC2/x2Hmj//M2fR2tsOZD3pAJpOZMSoiIiLzKkv+IGm7E2OEEDh48CA2bdqEnTt3Ij8/HwoF9x6j0iu6f12HIG8mdURERGVQ7sTuwoUL2LhxI7Zs2YKkpCQIIdC8eXOMHDkSL730kilipGpACGGQ2HVqyPl1REREZSEpsYuPj8f333+P77//HpcvX4YQAv7+/pg6dSpGjhyJtm3bmjpOsnLX72chNUulV8b964iIiMpGUmJXr149AICDgwOGDx+OkSNHolevXpDLuS0FSXOiyGrY2h6OCPByMlM0RERElklSYte1a1eMGjUKgwcPhouLi6ljomooqsj+dR05v46IiKjMJCV2PFKMTEmt0eLUbc6vIzKJhw+BmTP1yxYvBlxdzRMPEVUqk62KJZLq0r1MPFSp9co4v45Iotxc4Isv9MvmzmViR1RNcFIcmV3R+XUN/Vzg5+ZgpmiIiIgsFxM7Mrui58N2Ym8dERGRJEzsyKxUag1Ox6TplXXg+bBERESSMLEjszoXmwGVWqt7LpMBHRqwx46IiEgKJnZkVkXn17Ws5Q53Jx5YTkREJAUTOzKroseIdWzI3joiIiKpSrXdyfz58yU1LpPJMHv2bEn3kvXLUqlxMT5Dr6wj59cRERFJVqrEbu7cuQZlj58KIIQwKBdCMLGjEp2+kwa19p+fHVuFDKH1PM0YERERkWUrVWJ3+PBhg7JPP/0U+/fvx8iRI9G5c2f4+/vj/v37OHbsGDZt2oTevXvj7bffNnnAZD2iburPr2tb1xNOdtwzm4iISKpS/Rbt0qWL3vNvvvkGR44cwdmzZ9GiRQu9a6NGjcLUqVPRsWNHDBw40OBeokKcX0dERGRakhZPrFy5EsOHDzdI6gq1atUKw4cPx4oVK8oVHFmvtOw8XE7M1Cvj+bBERETlIymxu3nzJry9S+5d8fb2xq1btyQFRdbv5G393jpHWwVa1/EwTzBERERWQlJi5+vri3379uktmnicVqvFvn374OPDHhgyruj8uvb1vWBnw913iIiIykPSb9IXX3wRf/zxB/r374+LFy/qXbtw4QL69++PS5cu4aWXXjJJkGR9DObX8XxYItNQKIDmzfUfCoW5oyKiSiITxXW7lSA3Nxf9+/fHoUOHIJPJ4OzsDF9fX6SkpCA7OxtCCPTo0QO7d++Gg4NDRcRtkTIzM+Hu7g6lUgk3Nzdzh2M29zJy0HHxb3pleyc/g5a13c0UERERUdVVlvxBUo+dg4MD9u/fj2+//RZdunSBnZ0d4uLiYGdnh65du+Lbb7/Fr7/+yqSOjCraW+fuaItmNatvoktERGQqkjcNk8lkGDNmDMaMGWPCcKg6KHo+bIcG3lDIZcXUJiIiotLibHWqVEIInLjJ/euIiIgqQrkSu507dyIiIgLBwcFo2LChrvzq1atYsmQJEhISyh0gWZc7qdlIyszVK+P5sERERKYhKbHTarUYNmwYhgwZgh07duD27du4c+eO7rqnpyfef/99fPfdd5KCUqlUmDFjBmrVqgVHR0eEhYXhwIEDpbp3y5YtCAkJgYODA3x9fTF+/HikpqYa1Lt//z7Gjh0LPz8/ODo6IiQkBNu2bSu23R9++AEdOnSAs7MzPDw80LFjR/z222/F1ifjoorMr/N3s0eQr7OZoiEiIrIukhK7FStWYNu2bXj99deRnp6O6dOn61339/dH586d8fPPP0sKasyYMVi+fDleeuklrFy5EgqFAn379sXx48dLvG/16tUYMWIEvLy8sHz5crz66qvYsmULnn32WeTm/tNLlJmZiWeeeQY7duzA66+/jmXLlsHV1RURERHYvHmzQbtz587FiBEjULduXSxfvhwLFixAcHAweyQliC4yv65jkA9kMs6vIzKZ7Gxg7lz9R3a2eWMiosojJGjZsqVo37697vncuXOFXC7Xq/Pqq6+KWrVqlbntU6dOCQBi6dKlurKcnBwRFBQkOnToUOx9KpVKeHh4iPDwcKHVanXle/bsEQDEqlWrdGVLliwRAMShQ4d0ZRqNRoSGhooaNWoIlUqlK4+OjhYymUwsX768zO+lKKVSKQAIpVJZ7rYskUajFW3m/SoCZ+zVPbaejjN3WETWJTlZCED/kZxs7qiIqBzKkj9IPlKsc+fOJdbx9vbGgwcPSqxjzPbt26FQKPDaa6/pyhwcHDB+/HhER0cjPj7e6H2XLl1CRkYGhg0bptcD1K9fP7i4uGDLli26ssjISPj6+qJ79+66MrlcjoiICCQlJeHo0aO68s8++ww1atTA1KlTIYRAVlZWmd8TFbicmIn0R/l6ZR15PiwREZHJSErsHB0doVQqS6wTGxsLDw+PMrd9/vx5NG7c2GADvvbt2wMoONnCGJVKpYvNWLznz5+HVqvV1TVWz8nJCQBw9uxZXdmhQ4cQGhqKVatWwdfXF66urqhZsyY+//zzMr+36i66yPy6et5OqO1h+H0gIiIiaSQldm3btsWvv/6qN2/tcWlpafjll1/w9NNPl7ntxMRE1KxZ06C8sOzevXtG72vUqBFkMhmioqL0yq9du4aUlBTk5OQgPT0dANCkSRPcvXsXsbGxenUjIyMBQDd3Lj09HampqYiKisLs2bMxc+ZM/PDDD2jTpg0mT56ML7/8ssT3olKpkJmZqfeozqKK7l/H1bBEREQmJSmxmzJlCu7evYvBgwfj7t27etdu3bqFQYMGQalUYsqUKWVuOycnB/b29gblhadY5OTkGL3Px8cHERER2LBhAz799FPcvn0bkZGRGDZsGGxtbfXufeWVV6BQKBAREYETJ07g1q1b+Pjjj7Fz5069eoXDrg8ePMA333yD6dOnIyIiAj///DOaN2+OBQsWlPhePv74Y7i7u+sedevWLfPnYS3yNVr8fidNr6wT968jIiIyKUmJ3cCBAzFjxgzs27cPgYGB+PTTTwEAfn5+aNy4MSIjI/HBBx/ozWErLUdHR92w6uMKeweNDaEW+vLLL9G3b19Mnz4dQUFBCA8PR6tWrdC/f38AgIuLCwAgODgYmzdvxq1bt9CpUyc0bNgQq1atwmeffaZXr/C1bG1tMWTIEN3ryOVyDBs2DHfv3kVcXFyx8cyaNQtKpVL3KG5+YHVwMT4Dj/I0emUdGjCxIyIiMiXJGxR//PHH+PXXX9GvXz84OTlBoVBAq9WiT58+2LdvH+bNmyep3Zo1ayIxMdGgvLCsVq1axd7r7u6OXbt2ITY2FkePHkVMTAw2btyIxMRE+Pr66s35GzJkCO7du4fff/8d0dHRiI2NRYMGDQAAjRs3BgB4eXnBwcEB3t7eUCgUeq/l5+cHALrhXWPs7e3h5uam96iuip4P27SGK7xdDHtmiYiISDrJZ8UCQM+ePdGzZ09TxQIAaNOmDQ4fPozMzEy9ROjUqVO6608SEBCAgIAAAEBGRgbOnj2LwYMHG9Szs7NDaGio7vnBgwcBAD169ABQ0DPXpk0bnD59Gnl5ebCzs9PVLZzr5+vrW8Z3WD1F3dSfX9eJq2GJiIhMrsqdFTtkyBBoNBp89dVXujKVSoV169YhLCxMN08tLi4OV69efWJ7s2bNglqtxrRp00qsd+PGDaxZswb9+vXT9dgBwLBhw6DRaLBhwwZdWW5uLr7//ns0b968xB5EKpCTp8H5uAy9so5BHIYlIiIytXL12KnValy7dg0ZGRnQaDRG64SHh5epzbCwMAwdOhSzZs1CcnIyGjZsiA0bNiAmJgZr167V1Rs1ahSOHj0KIYSubPHixbh06RLCwsJgY2ODn376Cfv378eCBQv0euYAoHnz5hg6dCgCAgJw584drF69Gl5eXlizZo1evddffx3ffPMN3nzzTVy/fh0BAQHYuHEjYmNjsWfPnjK9t+rqTGwa8jRa3XOFXIb29b3MGBEREZF1kpTYCSHw4Ycf4j//+Q8ePnxYYt3iEr6SfPfdd5g9ezY2btyI9PR0BAcHY+/evU9MElu1aoWdO3di9+7d0Gg0CA4OxtatWzF06FCDuq1bt8a6detw//593YraefPm6ebOFXJ0dMRvv/2Gd999F99++y2ys7PRpk0b/Pzzz+jdu3eZ31t1FHVTf35dcB13uDrYmikaIiIi6yUTj3d5ldL8+fMxd+5ceHh4YODAgahTpw5sbIzniHPmzCl3kNYiMzMT7u7uUCqV1WohxcDPj+Pi3X82tJ7UrSGm925ixoiIrFhKClDkP6hITgY4H5jIYpUlf5DUY/ftt98iMDAQZ86cgbc350pR8ZQ5+fgzQf+UEs6vIyIiqhiSFk8kJSXhhRdeYFJHT3Tq9gNoH+sTtrORIyTQ03wBERERWTFJPXb169ev9sdjUekU3b+uXaAnHGwVxdQmonKTyQAfH8MyIqoWJPXYTZgwAXv37kVycrKp4yErc+IW968jqlQ+PgXz7B5/FE30iMhqSeqxGzhwICIjI9GxY0d8+OGHCAkJKXYyX+FGwVT9JD/MxfX7WXplHTi/joiIqMJIHoqVyWQQQmDs2LHF1pPJZFCr1ZKDI8sWXWQY1tXeBsG13c0UDRERkfWTlNiNGjUKMs7ZoCc4UWT/urAGXrBRVLnDToiIiKyGpMRu/fr1Jg6DrFFUkfl1HYI4z4eIiKgisfuEKkR82iPcTc/RK+vUkPPriIiIKlK5zoolKk7UTf3eOm9nOzT2czVTNETVSE4O8O23+mXjxgGOjuaJh4gqVakSu+7du0Mmk2HDhg2oU6cOunfvXqrGZTIZDh06VK4AyTIV3b+uQ5A35HLOyySqcFlZwKRJ+mUREUzsiKqJUiV2R44cgUwmw6NHj3TPS4MLLKonIYRBYteR8+uIiIgqXKkSO61WW+JzosfdSM5CapZKr4zz64iIiCoeF0+QyRWdX1fbwxEBXk5mioaIiKj6YGJHJmc4DOvNYXkiIqJKUK5Vsbm5uTh9+jTu3bsHlUpltM6oUaPK8xJkYdQaLU7e1k/seD4sERFR5ZCc2P33v//F7NmzoVQqjV4XQkAmkzGxq2Yu3cvEw1z9Y+R4PiwREVHlkDQU++OPP2Ly5MmoW7culi1bBiEEBg4ciEWLFqFPnz4QQmDw4MH4tuheSmT1ThQ5baKhnwv83RzMFA0REVH1Iimx++yzz+Dn54fo6GhMmzYNANCmTRvMmDEDP//8MzZt2oSffvoJgYGBJg2Wqr6i58N2ZG8dERFRpZGU2P3xxx8YMGAAnJz+Wemo0Wh0f37xxRfRvXt3zJ8/v/wRksVQqTU4HZOmV8b964iIiCqPpMQuPz8fvr6+uueOjo7IyMjQq9O6dWucO3euXMGRZTkXmwGV+p89DmUy4OkGXmaMiIiIqHqRlNjVqlULiYmJuueBgYE4f/68Xp3Y2FjY2PAo2uokusj8upa13OHhZGemaIiIiKofSYldaGioXm9cnz59EBUVhY8//hh//fUXvvzyS/z4448IDQ01WaBU9UUZ2b+OiIiIKo+kxG7o0KFQqVSIiYkBAMyaNQt16tTBBx98gODgYEyYMAEuLi5YsmSJKWOlKixLpcbF+Ay9so7cv46IiKhSSRorHTRoEAYNGqR77uvriwsXLuCbb77B7du3ERgYiJEjR6J27domC5SqttN30qDWCt1zW4UMofU8zRgRERFR9WOySXCenp545513TNUcWZii58O2resJJzvOsSSqdL6+gBBPrkdEVolnxZJJGJwP25Dz64iIiCpbqbpUvvvuO8kvwCPFrF9adh4uJ2bqlXH/OiIiospXqsRuzJgxkMlkZWqYZ8VWHydv6/fWOdoq0Kauh3mCISIiqsZKlditW7euouMgC1Z0fl1ofS/Y2XCUn4iIqLKVKrEbPXp0RcdBFiy6yPy6Tty/joiIyCy4bJHKJVGZg9up2XplnF9HZEYqFbB7t37ZgAGAvb154iGiSlWuxC4xMRFbtmzB+fPnoVQq4e7ujrZt22L48OGoWbOmqWKkKuzETf3eOndHWzSv5WamaIgImZlARIR+WXJywTYoRGT1JCd2//3vf/HOO+9ApVJBPLZn0qZNm/D+++9j2bJlmDhxokmCpKorqsj5sE838IJCXraFNkRERGQakhK7LVu2YPLkyfDx8cH777+Pzp07w9/fH/fv38exY8ewcuVK3fWIov9zJKshhDDosevEY8SIiIjMRiZE2bcoDwkJwd27d3HhwgXUqlXL4Prdu3fRtm1bBAQE4OzZsyYJ1BpkZmbC3d0dSqUSbm6WP1x5OyUL3T89qld28O0uaOjnYqaIiAgpKYCfn34Zh2KJLFpZ8gdJe1JcuXIFERERRpM6AKhTpw6GDh2KK1euSGmeLERUkdWwfq72CPJ1NlM0REREJCmx8/DwgLNzyb/AXVxc4OHhIaV5shDRRebXdWroU+aNrImIiMh0JCV2AwYMwJ49e6BWq41ez8/Px549ezBw4MByBUdVl1YrDPav68D964iIiMxKUmK3ZMkSODs7o1evXjh58qTetejoaPTq1Quurq5YvHixSYKkqudKUibSH+XrlXVkYkdERGRWklbFtm3bFnl5eTh37hw6deoEGxsb+Pj4IDU1VdeLV7NmTbRt21bvPplMhlu3bpU/ajK7oqthA72dUMfTyUzREBERESAxsdNqtbC1tUVAQIBeedHFFEUX3EpYgEtV1Iki8+t42gQREZH5SUrsYmJiTBwGWZJ8jRa/30nTK+MwLBERkflJmmNH1dvF+Axk52n0yrhwgoiIyPwkJXbR0dGlqrd69WopzVMVd6LIatimNVzh48IDxomIiMxNUmLXpUsXLFiwoNg5c2lpaRg4cCAmTZpUruCoaoq6yfl1REREVZGkxK558+aYM2cOunXrhoSEBL1rhw8fRuvWrbFnzx68+OKLJgmSqo6cPA3Ox2XolXVqyGFYIiKiqkBSYvf7779j0qRJOHbsGIKDg7Fjxw5oNBrMmDEDPXv2xMOHD7Fx40Zs3LjR1PGSmZ2JTUOeRqt7rpDL0L6+lxkjIiI93t4FZ8M+/vDmf76IqgtJq2Lt7OywcuVK9O7dG2PHjkVERATq1KmDu3fvIiwsDN9//z3q169v6lipCig6vy64jjtcHWzNFA0RGZDLAV9fc0dBRGZSrlWxffv2xdtvvw0hBOLj4+Ht7Y1t27YxqbNiJwzm17EngIiIqKqQnNg9fPgQL774It577z3Url0bI0aMQGpqKtq1a4dffvnFlDFSFaHMycefCUq9sk5cOEFERFRlSN7upHXr1tiyZQteeOEF/PHHH/j++++xbds2qFQqPP/883jrrbeQl5dn6njJjE7dfgDtYwuh7WzkCAn0NF9AREREpEdSYhceHo779+9jzZo12LFjBzw9C365Dx48GH/++Sc6d+6MVatWoX379iYNlsyr6Py6doGecLBVmCkaIiIiKkrS4omWLVtiy5YtaNKkicG12rVr4/Dhw1i4cCHmz59f7gCp6jA8H5bz64iqnLw84MQJ/bKOHQE7O/PEQ0SVSlJid+rUKdiV8I+ETCbDBx98gJ49e0oOjKqW5Ie5uH4/S6+sY0POryOqcpRKoFs3/bLkZK6UJaomJG93UigrKwvXr19HdnY2OnfurFcvLCysfNFRlRFdZBjWxd4GwbXdzRQNERERGSN5VWxMTAwGDhwIT09PhIaGottj/0OMiopC8+bNceTIEVPESFXAiZv6iV1YfS/YKMq1Ww4RERGZmKTfzHFxcXj66afxv//9DwMHDkSHDh30zo0NCwtDamoq/u///s9kgZJ5nbhdZH4dh2GJiIiqHEmJ3Zw5c5Ceno6jR49i+/btBnPpbGxs0LlzZ0RFRZkkSDKv+LRHiE/L0SvjwgkiIqKqR1Ji9+uvv2LQoEHo2LFjsXUCAwORkJAgOTCqOoquhvV2tkMTf1czRUNERETFkZTYpaWloV69eiXWEUJApVJJaZ6qmKgi8+ueDvKGXC4zUzRERERUHEmJnb+/P27cuFFinT///BMBAQGSgqKqQwhhsDExjxEjIiKqmiQldj179sTevXvxxx9/GL0eGRmJ3377DX379i1XcGR+N5KzkJql3/PK+XVERERVk6TE7oMPPoCjoyPCw8OxcOFC3Lx5EwCwb98+zJ49G3369IGPjw/eeecdkwZLlS/qpv78utoejgj0djJTNERERFQSSYldvXr18Ouvv8LT0xOzZ8/G5s2bIYRAv379sHDhQvj6+uJ///sfatasKSkolUqFGTNmoFatWnB0dERYWBgOHDhQqnu3bNmCkJAQODg4wNfXF+PHj0dqaqpBvfv372Ps2LHw8/ODo6MjQkJCsG3btie237NnT8hkMkyaNKnM78sSFR2G7RDkDZmM8+uIiIiqIkknTwAFe9XduHEDe/bswalTp5CWlgY3NzeEhYVh4MCBJR459iRjxozB9u3b8dZbb6FRo0ZYv349+vbti8OHD+OZZ54p9r7Vq1dj4sSJePbZZ7F8+XLcvXsXK1euxJkzZ3Dq1Ck4ODgAADIzM/HMM8/g/v37mDp1KmrUqIGtW7ciIiIC33//PV588UWj7f/444+Ijo6W/L4sjVqjxcnbRebXNeQwLBERUZUlqphTp04JAGLp0qW6spycHBEUFCQ6dOhQ7H0qlUp4eHiI8PBwodVqdeV79uwRAMSqVat0ZUuWLBEAxKFDh3RlGo1GhIaGiho1agiVSmXQfk5OjqhXr56YP3++ACDefPPNMr83pVIpAAilUlnme83hQly6CJyxV++RpMwxd1hEVJLkZCEA/UdysrmjIqJyKEv+UOXOhNq+fTsUCgVee+01XZmDgwPGjx+P6OhoxMfHG73v0qVLyMjIwLBhw/SGCvv16wcXFxds2bJFVxYZGQlfX190795dVyaXyxEREYGkpCQcPXrUoP0lS5ZAq9Vi+vTppnibFiGqyP51Qb7O8HdzMFM0RERE9CRVLrE7f/48GjduDDc3N73y9u3bAwAuXLhg9L7CPfMcHR0Nrjk6OuL8+fPQarW6usbqOTkVLAo4e/asXnlcXBwWL16MTz75xOh91iq66DYnPEaMiIioSqtyiV1iYqLRRReFZffu3TN6X6NGjSCTyQyOMbt27RpSUlKQk5OD9PR0AECTJk1w9+5dxMbG6tWNjIwEAIMTM/7973+jbdu2GD58eJnei0qlQmZmpt7DUqjUGpyOSdMr4zYnRBbA0xO4dEn/4elp7qiIqJJIXjxRUXJycmBvb29QXrjwIScnx+AaAPj4+CAiIgIbNmxAs2bNMGjQICQkJGDy5MmwtbVFfn6+7t5XXnkFa9asQUREBFasWAF/f39s3boVO3fuNHiNw4cPY8eOHTh16lSZ38vHH3+MefPmlfm+quBcbAZy87W65zIZ8HQDJnZEVZ6NDdCihbmjICIzqXI9do6OjkaPIsvNzdVdL86XX36Jvn37Yvr06QgKCkJ4eDhatWqF/v37AwBcXFwAAMHBwdi8eTNu3bqFTp06oWHDhli1ahU+++wzvXpqtRpTpkzByJEjERoaWub3MmvWLCiVSt2juPmBVVF0kfl1LWq5wcNJ+kpnIiIiqnhVrseuZs2aBkOhQMEQLQDUqlWr2Hvd3d2xa9cuxMXFISYmBoGBgQgMDETHjh3h6+sLDw8PXd0hQ4ZgwIABuHjxIjQaDUJCQnDkyBEAQOPGjQEA3333Ha5du4Yvv/wSMTExeq/18OFDxMTEwM/PTzc3ryh7e3ujvY+WIIrHiBEREVmcKpfYtWnTBocPH0ZmZqbeAorCodA2bdo8sY2AgADdObUZGRk4e/YsBg8ebFDPzs5Oryfu4MGDAIAePXoAKFg0kZ+fj06dOhnc+9133+G7777Dzp078cILL5T6/VmCLJUaF+Mz9Mo6cH4dERFRlSd5KFatVmPFihVo37493NzcYGPzT4544cIFTJw4EdevXy9zu0OGDIFGo8FXX32lK1OpVFi3bh3CwsJQt25dAAVJ19WrV5/Y3qxZs6BWqzFt2rQS6924cQNr1qxBv379dD12w4cPx86dOw0eANC3b1/s3LkTYWFhZX6PVd3pO2lQa4Xuua1Chvb1vcwYEREREZWGpB67nJwc9OrVCydOnICPjw/c3NyQnZ2tu16/fn2sW7cOXl5eWLBgQZnaDgsLw9ChQzFr1iwkJyejYcOG2LBhA2JiYrB27VpdvVGjRuHo0aMQ4p8EZPHixbh06RLCwsJgY2ODn376Cfv378eCBQsM5sg1b94cQ4cORUBAAO7cuYPVq1fDy8sLa9as0dVp2rQpmjZtajTO+vXrW11PXaETRebXta3rCSe7Kte5S0TGqNXAtWv6ZU2aFCyqICKrJ6nHbtGiRYiKisLHH3+MpKQkvPLKK3rX3d3d0aVLF/z666+Sgvruu+/w1ltvYePGjZgyZQry8/Oxd+9ehIeHl3hfq1atcOPGDbz//vuYPn06MjMzsXXrVrz//vsGdVu3bo1169Zh4sSJWLduHSIiIvD777+jdu3akmK2JlE3Dc+HJSILkZ4OtGyp//h7qycisn6S/gv3ww8/oFu3bnj33XcBwOih8A0aNMD58+clBeXg4IClS5di6dKlxdYpXOjwuOeffx7PP/98qV7j//7v/yTFBkCvl9DapGfn4XKi/n573JiYiIjIMkjqsYuLi0O7du1KrOPq6gqlUikpKDKf6Nv6vXWOtgq0qethnmCIiIioTCQldq6urkhOTi6xzq1bt+Dr6yspKDKfqJv68+tC63vBzqbKbXdIRERERkj6jf30009jz549yMjIMHo9Pj4e//vf/544J46qnqLnw/IYMSIiIsshKbF75513kJ6ejmeffRZRUVFQq9UAgEePHuHQoUPo3bs31Go13n77bZMGSxUrUZmD26nZemXcmJiIiMhySFo8ER4ejs8//xxTp07V65VzdXUFACgUCnzxxRd46qmnTBMlVYoTRVbDujnYoHktt2JqExERUVUjeWOjCRMmoGvXrlizZg1OnTqFtLQ0uLm5ISwsDBMnTkQLHkJtcaKK7F/XIcgbCrnhimciIiKqmsq1Y2WzZs2wcuVKU8VCZiSEMDK/jsOwRERElkTSHLtDhw6ZOg4yszup2UhU5uqVdWrIhRNERESWRFJi17NnTwQEBGDmzJn4888/TR0TmcGJIr11fq72CPJ1MVM0REREJIWkxG7SpElQqVRYsmQJ2rRpg7Zt22L58uVITEw0dXxUSYxtc2LsRBEiIiKquiQldqtWrcK9e/ewe/duDBkyBNevX8f06dMREBCA3r17Y9OmTXj06JGpY6UKdDM5S+95WAMOwxIREVkayUcKKBQK9OvXDz/88AOSkpLwzTff4JlnnsHBgwcxevRo+Pv7Y+TIkaaMlSpQ8kP9+XV1PZ3MFAkRERFJZZKzolxdXTFu3DgcPnwYsbGxeO+995CXl4fNmzebonmqYHlqLdIf5euV+bnZmykaIiIikqpc2508TgiBgwcPYtOmTdi5cyfy8/OhUChM1TxVoJQslUGZnysTOyKL5O4OHD5sWEZE1UK5E7sLFy5g48aN2LJlC5KSkiCEQPPmzTFy5Ei89NJLpoiRKlhypv4wrJ2NHO6OtmaKhojKxc4O6NrV3FEQkZlISuzi4+Px/fff4/vvv8fly5chhIC/vz+mTp2KkSNHom3btqaOkypQ8kP9HjtfF3uuiCUiIrJAkhK7evXqAQAcHBwwfPhwjBw5Er169YJcbpIpe1TJiiZ2nF9HRERkmSQldl27dsWoUaMwePBguLhwE1tLl1JkKJbz64iIiCyTpMSOR4pZF4MeO1cHM0VCRERE5WGyVbFkuQzm2LHHjshyabXAA/2TZODtDXCqDFG1UKrEbty4cZDJZFi0aBH8/f0xbty4UjUuk8mwdu3acgVIFS/FoMeOiR2RxXrwAPDz0y9LTgZ8fc0TDxFVqlIlduvXr4dMJsOMGTPg7++P9evXl6pxJnaWoeipE1w8QUREZJlKldjduXMHAFC7dm2952T5NFqB1Kw8vTLOsSMiIrJMpUrsAgMDS3xOlistOw8ardAr41AsERGRZZI0m3b+/Pk4duxYiXUiIyMxf/58SUFR5Sk6DCuXAd4uTOyIiIgskaTEbu7cuThy5EiJdY4dO4Z58+ZJaZ4qUdEVsd4u9lDIeeoEERGRJaqw9e95eXlQKBQV1TyZSEomV8QSERFZC8mJXUlniebl5SEyMhJ+RZfcU5VjsCKWiR0REZHFKvUGxQ0aNNB7vmLFCqxbt86gnkajQWpqKnJzc/Hqq6+WP0KqUDx1goiIyHqUOrHTarW6XjqZTAYhBIQQBvVsbW3RokULdO/eHbNnzzZdpFQhkosOxXIPOyIiIotV6sQuJiZG92e5XI5p06bhww8/rIiYqBJxKJaIiMh6SDor9s6dO/Dw8DBxKGQOPCeWiIjIekhK7LhBsXUQQhhJ7DjHjoiIyFJJSuwKRUdH4+DBg7h37x5UKpXBdZ4VW7Vl5qqRp9bqlXEoloiIyHJJSuzUajVGjBiBH3/8EUII3WKKQoXPmdhVbSlF5tcBHIolIiKyZJL2sfv000+xY8cOjB07FmfOnIEQAm+99Raio6PxySefwMPDA0OHDsWtW7dMHS+ZUNEVse6OtnCw5abSRBbNzQ3YulX/4eZm7qiIqJJI6rH7/vvv0bJlS3zzzTe6Mg8PD4SFhSEsLAx9+/ZF+/bt0b17d7z++usmC5ZMy3APO/bWEVk8e3tg6FBzR0FEZiKpx+7mzZvo2rWr7rlMJkN+fr7ueYsWLdC/f3+sXr263AFSxTHY6oR72BEREVk0SYmdnZ0dnJycdM9dXFyQnJysVycwMBA3btwoX3RUoQw2J+aKWCIiIosmKbGrW7cu4uPjdc+bNm2KY8eO6S2gOHnyJLy8vMofIVUYDsUSERFZF0mJXZcuXfQSuWHDhuHatWvo168f/vvf/2LEiBE4fvw4+vTpY9JgybSKDsVyRSwREZFlk7R4Yty4cdBoNEhISECdOnUwefJkHDlyBHv37sW+ffsAAO3bt8fixYtNGiyZlkGPnRuHYomIiCyZpMQuJCREb2GEra0tdu/ejTNnzuDWrVsIDAxE+/btIZdL6hCkSpJiMMeOPXZEFi8lBfDz0y9LTgZ8fc0TDxFVqnKdPFFUu3bt0K5dO1M2SRUkJ0+Dhyq1XhmHYomIiCwbu9SqqaLz6wD22BEREVm6UvXYjRs3TlLjPFKs6io6v87RVgEXe5N24BIREVElK9Vv8vXr10tqnIld1ZVisHDCHjKZzEzREBERkSmUKrG7c+dORcdBlSw5s8ipExyGJSIisnilSuwCAwMrOg6qZIabE3OrEyIiIkvHxRPVVNHEjitiiYiILJ+k2fLHjh0rdd3w8HApL0EVzHBzYiZ2RERElk5SYte1a9dST7TXaDRSXoIqmOEcOw7FEhERWTpJid2HH35oNLFTKpU4d+4cjh07hueff56bFVdhBqtiORRLRERk8SQldnPnzi3x+vbt2zFmzBjMmzdPSvNUwfI1WjzIztMr41AsERGR5auQxRNDhgxBt27dMGvWrIponsopNUtlUMahWCIiIstXYatimzVrhujo6IpqnsohOVM/sbORy+DhaGumaIiIiMhUKiyxO3/+PORy7qZSFRnb6kQu56kTRERElk7SHLu4uDij5Wq1GgkJCVi/fj1+++03vPDCC+WJjSpI8kOeOkFktVxcgM8/NywjompBUmJXr169Erc7EUIgKCgIK1askBwYVZyiQ7G+nF9HZD0cHYE33zR3FERkJpISu1GjRhlN7ORyOTw9PREaGoqBAwfCwYEJQ1WUksXNiYmIiKyRpMRu/fr1Jg6DKlPRHjsOxRIREVkHrm6ohlIM5tixZ5WIiMgaSOqxe5xWq8X9+/eRn59v9HpAQEB5X4JMzOCcWPbYERERWQXJid2mTZuwbNkyXL58udjzYGUyGdRqteTgyPS0WmF4nBjn2BEREVkFSYndsmXLMGPGDNja2iI8PBw1a9aEjU25O/+oEqQ/yoNaK/TKOBRLZEVSU4FmzfTLrlwBfHzMEw8RVSpJ2dh//vMf1K5dGydOnECdOnVMHRNUKhU+/PBDbNy4Eenp6QgODsaCBQvQs2fPJ967ZcsWLFmyBJcvX4arqysGDBiATz75BD5F/lG7f/8+Zs6ciZ9//hkPHz5Es2bNMGvWLAwdOlSv3o8//ogffvgBp0+fRlJSEurWrYt+/fph9uzZ8PDwMOXbrhRFh2FlMsDbxc5M0RCRyQlRkNwVLSOiakHS4omUlBQMHjy4QpI6ABgzZgyWL1+Ol156CStXroRCoUDfvn1x/PjxEu9bvXo1RowYAS8vLyxfvhyvvvoqtmzZgmeffRa5uf8sGMjMzMQzzzyDHTt24PXXX8eyZcvg6uqKiIgIbN68Wa/N1157DVeuXMHLL7+MVatWoU+fPvj888/RoUMH5OTkVMj7r0hFEzsvJzvYKriGhoiIyCoICVq3bi1Gjx4t5dYnOnXqlAAgli5dqivLyckRQUFBokOHDsXep1KphIeHhwgPDxdarVZXvmfPHgFArFq1Sle2ZMkSAUAcOnRIV6bRaERoaKioUaOGUKlUuvLDhw8bvNaGDRsEAPH111+X6b0plUoBQCiVyjLdZ0pbT8eJwBl7dY/eK46aLRYiqgDJyUIU9NH980hONndURFQOZckfJHXVTJs2Dbt27UJsbKzpMsy/bd++HQqFAq+99pquzMHBAePHj0d0dDTi4+ON3nfp0iVkZGRg2LBhepsn9+vXDy4uLtiyZYuuLDIyEr6+vujevbuuTC6XIyIiAklJSTh69KiuvGvXrgavNWjQIADAlStXJL9PczFYEevG+XVERETWQtIcu9GjRyM5ORkdO3bExIkT0bp1a7i5uRmtGx4eXqa2z58/j8aNGxu01759ewDAhQsXULduXYP7VKqChMXR0dHgmqOjI86fPw+tVgu5XA6VSmW0npOTEwDg7NmzJc7nS0pKAgCDeXuWwGBFLLc6ISIishqSl7JmZmZCqVTiww8/LLFecVuhFCcxMRE1a9Y0KC8su3fvntH7GjVqBJlMhqioKIwdO1ZXfu3aNaSkpAAA0tPT4e3tjSZNmuDgwYOIjY1FYGCgrm5kZCQAICEhocQYP/nkEygUCgwZMqTEeiqVSpdwAgWfmbklG2xOzMSOiIjIWkhK7D788EMsWrQIvr6+GD58uEm3O8nJyYG9vWGyUXjubHELFnx8fBAREYENGzagWbNmGDRoEBISEjB58mTY2toiPz9fd+8rr7yCNWvWICIiAitWrIC/vz+2bt2KnTt3lvgaALB582asXbsW7777Lho1alTie/n4448xb968Ur3vysLjxIiIiKyXpGzs22+/RePGjXH69Gm4uLiYNCBHR0e9Xq5ChatajQ2hFvryyy+Rk5OD6dOnY/r06QCAl19+GUFBQfjxxx91sQYHB2Pz5s1444030KlTJwBAjRo18Nlnn2HChAnFvqfIyEiMHz8evXv3xsKFC5/4XmbNmoW3335b9zwzM9PoMHJlSsniHDsiIiJrJSmxS09Px/Dhw02e1AEFQ67GhkITExMBALVq1Sr2Xnd3d+zatQtxcXGIiYlBYGAgAgMD0bFjR/j6+urtOzdkyBAMGDAAFy9ehEajQUhICI4cOQIAaNy4sUHbFy9exIABA9CyZUts3769VD2U9vb2RnsfzUUIwR47IiIiKyYpsWvVqpUu0TK1Nm3a4PDhw8jMzNRbQHHq1Cnd9ScJCAjQnVGbkZGBs2fPYvDgwQb17OzsEBoaqnt+8OBBAECPHj306t26dQt9+vSBn58f/ve//1VIQlsZslRq5OTrz3nkqRNERETWQ9J2J++//z5++uknnDt3ztTxYMiQIdBoNPjqq690ZSqVCuvWrUNYWJhuKDMuLg5Xr159YnuzZs2CWq3GtGnTSqx348YNrFmzBv369dPrsUtKSkKvXr0gl8vx66+/wtfXV+I7M7+iW50APCeWiIjImkgeiu3Zsyc6duyIkSNHlrjdyahRo8rUdlhYGIYOHYpZs2YhOTkZDRs2xIYNGxATE4O1a9fqtXv06FGIx47KWbx4MS5duoSwsDDY2Njgp59+wv79+7FgwQK9njkAaN68OYYOHYqAgADcuXMHq1evhpeXF9asWaNXr0+fPrh9+zbeffddHD9+XO/0C39//1Idc1ZVFB2GdXWwgYOtwkzREBERkalJSuzGjBkDmUwGIYQu2Xp8U2CgYD6XTCYrc2IHAN999x1mz56td1bs3r17n7gnXqtWrbBz507s3r0bGo0GwcHB2Lp1q8H5rwDQunVrrFu3Dvfv39etqJ03bx78/Pz06l28eBEAsGTJEoM2unTpYlmJHbc6ISIismoyIcp+OvSGDRtKXXf06NFlbd5qZWZmwt3dHUqlstgezor0TeRtLPj5n9Mynm7ghS2vdaj0OIioAqWkAEX+g4rkZMCCp5EQVXdlyR8knzxBlsfgODEunCCyPk5OwJw5hmVEVC2YZldhsgjJmRyKJbJ6zs7A3LnmjoKIzERSYhcXF1fquoXbjpD5GfTYcUUsERGRVZGU2NWrV89gsYQxMpkMarVayktQBeBQLBERkXWTlNiNGjXKaGKnVCpx8eJF3LlzB126dEG9evXKGx+ZEIdiiYiIrJukxG79+vXFXhNC4NNPP8WSJUv09p0j88rN1yAzV7/3lEOxRERE1kXSyRMlkclkmD59Olq0aIF33nnH1M2TRClGTp3w5VAsERGRVTF5YleoXbt2+O233yqqeSqjovPr7G3kcHPgomgiq5OWBrRoof9ISzN3VERUSSrsN/utW7e4cKIKSSl66oSbfakWwBCRhdFogMuXDcuIqFowaWKn1WqRkJCA9evXY9euXXj22WdN2TyVA1fEEhERWT9JiZ1cLi+xt0cIAU9PT3z66aeSAyPTSs4smthx4QQREZG1kZTYhYeHG03s5HI5PD09ERoairFjx8Kv6HmFZDbJRYZifZnYERERWR1Jid2RI0dMHAZVNMOhWCZ2RERE1qbCVsVS1WI4FMs5dkRERNamTIndwoUL8d577yE/P7/YOnl5eXjvvfewePHicgdHplO0x86XmxMTERFZnVIndgcPHsSHH34Ib29v2NraFlvPzs4OPj4+eP/993H48GGTBEnlo9Zo8SCbQ7FERETWrtSJ3XfffQdPT09MmjTpiXXffPNNeHl5Yd26deUKjkzjQXYehNAv41AsERGR9Sl1YnfixAn06NED9vZP7umxt7dHjx49EBUVVa7gyDSKzq9TyGXwdrYzUzRERERUUUqd2N27dw8NGjQodcP169dHYmKipKDItFKy9Lc68XGxg1zOUyeIiIisTakTO7lcXuKiiaLy8/Mhl3PRbVXAFbFERETVQ6kzr1q1auHSpUulbvjSpUuoXbu2pKDItLiHHRERUfVQ6sSuc+fO+O233xATE/PEujExMfjtt98QHh5entjIRIqeOuHHrU6IiIisUqlPnnjzzTexbt06DBkyBL/88gt8fHyM1nvw4AGGDh0KtVqNCRMmmCxQkq7oUKwvh2KJrJeDAzBxomEZEVULpU7sQkJC8NZbb+Gzzz5D8+bN8cYbb6Bbt26oU6cOACAhIQGHDh3CV199hZSUFLz99tsICQmpsMCp9Aw2J+ZQLJH1cnUF/vtfc0dBRGZSprNiP/30Uzg4OGDp0qVYuHAhFi5cqHddCAGFQoFZs2ZhwYIFJg2UpEvhHDsiIqJqoUyJnUwmw6JFizB+/HisW7cOJ06cQFJSEgCgRo0a6NSpE8aMGYOgoKAKCZbKTgjBxI6IiKiaKFNiVygoKIg9chYi41E+8jRavTI/N863ISIiskbcaM7KFZ1fBwC+LuyxIyIiskZM7Kxc0a1OPJ1sYWfDbzsREZE1kjQUS5aDp04QVTMZGcALL+iX/fQT4OFR+bEQUaVjYmflDE6d4ObERNYtPx84etSwjIiqBY7JWbmiK2K5hx0REZH1YmJn5QyOE+NQLBERkdViYmflDIZi2WNHRERktZjYWTmDzYk5x46IiMhqMbGzcsmZ+kOx3MOOiIjIejGxs2LZKjWy8zR6ZTx1goiIyHoxsbNixk6d4Bw7IiIi68XEzooVHYZ1tlPA2Z5bFxIREVkrJnZWzHBzYg7DEhERWTMmdlasaGLHzYmJiIisGxM7K2a4OTETOyIiImvGxM6KpWQW3ZyYQ7FERETWjImdFTOcY8ceOyIiImvGJZJWzODUCQ7FElk/OztgyBDDMiKqFpjYWTHDOXYciiWyeu7uwLZt5o6CiMyEQ7FWKk+tRfqjfL0yDsUSERFZNyZ2Violy/DUCZ4TS0REZN2Y2FmpoqdO2Cnk8HCyNVM0REREVBmY2FkpY5sTy2QyM0VDRERElYGJnZXiqRNERETVD1fFWqmUTJ46QVQtKZXAK6/ol33zTcFqWSKyekzsrBQ3JyaqpvLygO3b9cu++MI8sRBRpeNQrJUySOy4hx0REZHVY2JnpQw3J2aPHRERkbVjYmelkjM5FEtERFTdMLGzQhqtQGoWh2KJiIiqGyZ2VigtOw9aoV/GoVgiIiLrx8TOChWdXyeXAd48ToyIiMjqMbGzQkVXxHo520Mh56kTRERE1o6JnRVKKbpwgsOwRERE1QITOytksNUJV8QSERFVC0zsrJDh5sRM7IiIiKoDJnZWyGAPO251QkREVC1UycROpVJhxowZqFWrFhwdHREWFoYDBw6U6t4tW7YgJCQEDg4O8PX1xfjx45GammpQ7/79+xg7diz8/Pzg6OiIkJAQbNu2zWibCQkJiIiIgIeHB9zc3DBw4EDcvn27XO+xInEoloiIqHqqkondmDFjsHz5crz00ktYuXIlFAoF+vbti+PHj5d43+rVqzFixAh4eXlh+fLlePXVV7FlyxY8++yzyM39J9nJzMzEM888gx07duD111/HsmXL4OrqioiICGzevFmvzaysLHTr1g1Hjx7Fe++9h3nz5uH8+fPo0qULHjx4UCHvv7w4FEtERFRNiSrm1KlTAoBYunSpriwnJ0cEBQWJDh06FHufSqUSHh4eIjw8XGi1Wl35nj17BACxatUqXdmSJUsEAHHo0CFdmUajEaGhoaJGjRpCpVLpyj/55BMBQPz++++6sitXrgiFQiFmzZpVpvemVCoFAKFUKst0X1lotVrR6P3/icAZe3WPMzFpFfZ6RFTFpKcL0aWL/iM93ZwREVE5lSV/qHI9dtu3b4dCocBrr72mK3NwcMD48eMRHR2N+Ph4o/ddunQJGRkZGDZsGGSyf/Zs69evH1xcXLBlyxZdWWRkJHx9fdG9e3ddmVwuR0REBJKSknD06FG9eEJDQxEaGqora9q0KZ599lls3brVJO/ZlDJz1MhTa/XK2GNHVI14eABHjug/PDzMGRERVaIql9idP38ejRs3hpubm155+/btAQAXLlwwep9KVTD86OjoaHDN0dER58+fh1ar1dU1Vs/JyQkAcPbsWQCAVqvFH3/8gXbt2hnUbd++PW7duoWHDx+W8p1VjqLz6wDAl4kdERFRtWBj7gCKSkxMRM2aNQ3KC8vu3btn9L5GjRpBJpMhKioKY8eO1ZVfu3YNKSkpAID09HR4e3ujSZMmOHjwIGJjYxEYGKirGxkZCaBgsQQApKWlQaVSPTGeJk2aGI1JpVLpEk4AUCqVAArm+FWUmHsPoFU90j13c7BBXk428nIq7CWJiIioAhXmDUKIJ9SsgoldTk4O7O0Ne5gcHBx0143x8fFBREQENmzYgGbNmmHQoEFISEjA5MmTYWtri/z8fN29r7zyCtasWYOIiAisWLEC/v7+2Lp1K3bu3Kn3GoVfpcQDAB9//DHmzZtnUF63bt1i76kI7osr9eWIiIioAjx8+BDu7u4l1qlyiZ2jo6NeL1ehwlWtxoZQC3355ZfIycnB9OnTMX36dADAyy+/jKCgIPz4449wcXEBAAQHB2Pz5s1444030KlTJwBAjRo18Nlnn2HChAm6eoWvJTWeWbNm4e2339Y9z8jIQGBgIOLi4p74jSHjMjMzUbduXcTHxxsM11Pp8DM0DX6O5cfPsPz4GZafJXyGQgg8fPgQtWrVemLdKpfY1axZUzcU+rjExEQAKPFNubu7Y9euXYiLi0NMTAwCAwMRGBiIjh07wtfXFx6PTSAeMmQIBgwYgIsXL0Kj0SAkJARHjhwBADRu3BgA4OXlBXt7e91rlzUee3t7o7197u7uVfaHx1K4ubnxMywnfoamwc+x/PgZlh8/w/Kr6p9haTuEqlxi16ZNGxw+fBiZmZl6H/CpU6d0158kICAAAQEBAAp6yc6ePYvBgwcb1LOzs9Nb7Xrw4EEAQI8ePQAUrJRt1aoVzpw5Y3DvqVOn0KBBA7i6upb+zRERERFVoCq3KnbIkCHQaDT46quvdGUqlQrr1q1DWFiYbn5aXFwcrl69+sT2Zs2aBbVajWnTppVY78aNG1izZg369eun67ErjOf06dN6yd21a9fw22+/YejQoWV9e0REREQVpsr12IWFhWHo0KGYNWsWkpOT0bBhQ2zYsAExMTFYu3atrt6oUaNw9OhRvRUiixcvxqVLlxAWFgYbGxv89NNP2L9/PxYsWKDXMwcAzZs3x9ChQxEQEIA7d+5g9erV8PLywpo1a/TqTZw4EV9//TWef/55TJ8+Hba2tli+fDn8/f3x73//u0zvzd7eHnPmzDE6PEulw8+w/PgZmgY/x/LjZ1h+/AzLz9o+Q5kozdrZSpabm4vZs2dj06ZNSE9PR3BwMD766CP07t1bV6dr164Gid3PP/+M+fPn48qVK9BoNAgODsbbb79ttGdtxIgRiIqKwv379+Hj44MBAwZg3rx58PPzM6h79+5dTJs2Dfv374dWq0XXrl2xYsUKNGzYsGI+ACIiIiIJqmRiR0RERERlV+Xm2BERERGRNEzsiIiIiKwEE7sKdvr0aUyaNAktWrSAs7MzAgICEBERgevXr5s7NIu1cOFCyGQytGzZ0tyhWJxz585hwIAB8PLygpOTE1q2bIlVq1aZOyyLcePGDQwfPhx16tSBk5MTmjZtivnz5+PRo0dPvrkaysrKwpw5c9CnTx94eXlBJpNh/fr1RuteuXIFffr0gYuLC7y8vDBy5EjdcZDVWWk+Q61Wi/Xr12PAgAGoW7cunJ2d0bJlSyxYsEC3mX51V5afxUL5+flo3rw5ZDIZli1bVjmBmkCVWxVrbT755BNERUVh6NChCA4ORlJSEj7//HOEhITg5MmTTE7K6O7du1i0aBGcnZ3NHYrF2b9/P/r374+2bdti9uzZcHFxwa1bt3D37l1zh2YR4uPj0b59e7i7u2PSpEnw8vJCdHQ05syZg7Nnz2LXrl3mDrHKSU1Nxfz58xEQEIDWrVvrNoEv6u7duwgPD4e7uzsWLVqErKwsLFu2DH/++Sd+//132NnZVW7gVUhpPsNHjx5h7NixePrpp/HGG2/Az89P97N56NAh/Pbbb5DJZJUffBVS2p/Fx/3nP/9BXFxcxQdnaoIqVFRUlFCpVHpl169fF/b29uKll14yU1SWa9iwYaJ79+6iS5cuokWLFuYOx2IolUrh7+8vBg0aJDQajbnDsUgLFy4UAMSlS5f0ykeNGiUAiLS0NDNFVnXl5uaKxMREIYQQp0+fFgDEunXrDOpNmDBBODo6itjYWF3ZgQMHBADx5ZdfVla4VVJpPkOVSiWioqIM7p03b54AIA4cOFAZoVZppf1ZLHT//n3h7u4u5s+fLwCIpUuXVlKk5ceh2ArWsWNHg/9tNmrUCC1atMCVK1fMFJVlOnbsGLZv347PPvvM3KFYnM2bN+P+/ftYuHAh5HI5srOzodVqzR2WRcnMzAQA+Pv765XXrFkTcrm8WvcqFcfe3h41atR4Yr0dO3agX79+uhODgIITgBo3boytW7dWZIhVXmk+Qzs7O3Ts2NGgfNCgQQDA3zUo/c9ioZkzZ6JJkyZ4+eWXKzCqisHEzgyEELr986h0NBoNJk+ejFdeeQWtWrUydzgW5+DBg3Bzc0NCQgKaNGkCFxcXuLm5YcKECZyDU0pdu3YFAIwfPx4XLlxAfHw8fvjhB6xevRpTpkzh9ACJEhISkJycjHbt2hlca9++Pc6fP2+GqKxDUlISAPB3TRn9/vvv2LBhAz777DOLHMJmYmcG33//PRISEjBs2DBzh2Ix1qxZg9jYWHz00UfmDsUi3bhxA2q1GgMHDkTv3r2xY8cOjBs3DmvWrMHYsWPNHZ5F6NOnDz766CMcOHAAbdu2RUBAAIYPH47JkydjxYoV5g7PYiUmJgIo6PksqmbNmkhLS4NKparssKzCkiVL4Obmhueee87coVgMIQQmT56MYcOGoUOHDuYORxIunqhkV69exZtvvokOHTpg9OjR5g7HIjx48AAffvghZs+eDV9fX3OHY5GysrLw6NEjvPHGG7pVsP/617+Ql5eHL7/8EvPnz0ejRo3MHGXVV69ePYSHh2Pw4MHw9vbGzz//jEWLFqFGjRqYNGmSucOzSDk5OQBg9DgnBwcHXR1rOe6psixatAgHDx7EF198AQ8PD3OHYzHWr1+PP//8E9u3bzd3KJIxsatESUlJeP755+Hu7o7t27dDoVCYOySL8MEHH8DLywuTJ082dygWy9HREUDBUXqPe/HFF/Hll18iOjqaid0TbNmyBa+99hquX7+OOnXqAChIjrVaLWbMmIERI0bA29vbzFFansKfTWO9coXTBArrUOn88MMP+OCDDzB+/HhMmDDB3OFYjMzMTMyaNQvvvPMO6tata+5wJONQbCVRKpV47rnnkJGRgV9++QW1atUyd0gW4caNG/jqq68wZcoU3Lt3DzExMYiJiUFubi7y8/MRExODtLQ0c4dZ5RX+vBWd+F94NnJ6enqlx2RpvvjiC7Rt21aX1BUaMGAAHj16xLlgEhUOwRYOyT4uMTERXl5e7K0rgwMHDmDUqFF4/vnnsWbNGnOHY1GWLVuGvLw8DBs2TPe7pnA7qPT0dMTExCAvL8/MUT4ZE7tKkJubi/79++P69evYu3cvmjdvbu6QLEZCQgK0Wi2mTJmC+vXr6x6nTp3C9evXUb9+fcyfP9/cYVZ5Tz31FICCz/Nx9+7dAwAOcZfC/fv3odFoDMrz8/MBAGq1urJDsgq1a9eGr68vzpw5Y3Dt999/R5s2bSo/KAt16tQpDBo0CO3atcPWrVthY8NBubKIi4tDeno6WrRooftd07lzZwAFQ9v169fH5cuXzRzlk/G7XsE0Gg2GDRuG6Oho7Nq1y2InY5pLy5YtsXPnToPyDz74AA8fPsTKlSsRFBRkhsgsS0REBBYvXoy1a9eie/fuuvJvvvkGNjY2uhWfVLzGjRtj//79uH79Oho3bqwr/7//+z/I5XIEBwebMTrLNnjwYGzYsAHx8fG6IbBDhw7h+vXrmDZtmpmjswxXrlzB888/j3r16mHv3r0cvpZgypQpeOGFF/TKkpOT8frrr2PMmDEYOHAg6tevb57gyoCJXQX797//jd27d6N///5IS0vDpk2b9K5b4h45lcnHx8fgLxoA3V52xq6RobZt22LcuHH49ttvoVar0aVLFxw5cgTbtm3DrFmzODWgFN555x3s27cPnTt3xqRJk+Dt7Y29e/di3759eOWVV/gZFuPzzz9HRkaGrnd4z549uuGtyZMnw93dHe+99x62bduGbt26YerUqcjKysLSpUvRqlUrrtrGkz9DuVyO3r17Iz09He+88w5+/vlnvfuDgoLYqYAnf44hISEICQnRuycmJgYA0KJFC8v5fWPuHZKtXZcuXQSAYh8kDU+eKLu8vDwxd+5cERgYKGxtbUXDhg3FihUrzB2WRTl16pR47rnnRI0aNYStra1o3LixWLhwocjPzzd3aFVWYGBgsf/+3blzR1fv0qVLolevXsLJyUl4eHiIl156SSQlJZkv8CrkSZ/hnTt3Svw9M3r0aHO/hSqhtD+Ljyv8bC3p5AmZEEJUePZIRERERBWOiyeIiIiIrAQTOyIiIiIrwcSOiIiIyEowsSMiIiKyEkzsiIiIiKwEEzsiIiIiK8HEjoiIiMhKMLEjIiIishJM7IiIiIisBBM7IjKbMWPGQCaT6c5jtHT79+9Hp06d4OnpCZlMZjlnSxKR1WBiR2QFYmJiIJPJIJPJ0Lt3b6N1Tp48CZlMhjFjxlRucNVETEwMBg4ciNu3b2Ps2LGYM2cOhg8fbu6wSILCv0/8u0KWyMbcARCRae3fvx+//fYbunfvbu5QqpWDBw8iNzcXn376KV588UVzh0NE1RR77IisSL169SCXyzFjxgwIIcwdTrVy7949AECtWrXMHAkRVWdM7IisSJMmTTBy5EicOXMGW7duLdU99erVQ7169Yxe69q1K2QymV7Z3LlzIZPJcOTIEaxbtw6tWrWCo6Mj6tevj1WrVgEAhBD49NNP0aRJEzg4OKBRo0b47rvvio1Bq9ViyZIlaNSoERwcHFC/fn3Mnz8f+fn5RusfO3YM/fv3h4+PD+zt7dGoUSN88MEHePTokV69I0eOQCaTYe7cuThx4gR69eoFDw8Pg/dUnEuXLiEiIgJ+fn6wt7dH/fr18dZbb+HBgwe6OoXDdnPmzAEAdOvWTTcsfuTIkSe+Rl5eHlasWIHQ0FC4urrCxcUFzZs3x9tvv4309PQyx1Oo8PuqVCoxYcIE1KxZE87OzggPD8e5c+cAFCSjL7/8Mvz8/ODo6IhevXrhxo0bBm3JZDJ07doVd+/exYgRI+Dj4wMnJyd06tQJBw8eNPq+UlNT8dZbb6F+/fqwt7eHn58fIiIicOnSJYO6hXMt79y5g1WrVqFp06awt7dHYGAg5s2bB61Wa/Q1du3ahWeffRaenp5wcHBAy5YtsWzZMmg0Gr1669evh0wmw/r167F//3507NgRTk5O8Pb2xujRo/U+v/Xr16N+/foAgA0bNui+l49/Pwt7Zlu3bg13d3c4OzujXr16iIiIwMWLF43GSlRpBBFZvDt37ggAonfv3iI2NlbY29uLhg0biry8PF2d6OhoAUCMHj1a797AwEARGBhotN0uXbqIov9MzJkzRwAQAwcOFO7u7mLUqFFiypQponbt2gKA+Prrr8XEiROFv7+/GD9+vJgwYYLw9PQUAMTRo0f12ho9erQAIPr37y+8vLzEG2+8IaZPny6aNGkiAIjBgwcbxPTFF18ImUwmPD09xahRo8T06dNF165dBQDRsWNHoVKpdHUPHz4sAIiePXsKW1tb0atXL/HOO++IYcOGPfEzjYyMFE5OTsLGxkYMHz5czJw5U/d5BAUFiZSUFCGEEOnp6WLOnDm6a6NHjxZz5swRc+bMEXfu3CnxNR49eiQ6deokAIhGjRqJyZMni+nTp4uBAwcKJycncf78+TLHUygwMFDUrFlThIaGilatWompU6eK4cOHC7lcLjw9PcWVK1dEQECACAsLE9OmTRP9+/fXxaFWq/XaAiCCg4NFQECAeOqpp8SMGTPEuHHjhLOzs1AoFGLnzp169ZOTk0VQUJAAILp27Spmzpwphg0bJhQKhXBychKRkZF69Qt/DgYPHix8fHzEmDFjxJQpU0RAQIAAIN577z2Dz27mzJkCgKhdu7YYN26cmDZtmmjXrp0AIIYMGaJXd926dQKAGDRokLCzsxODBw8W//73v0VoaKgAIDp16qSre/78eTF16lQBQLRu3Vr3vXz8+xkREaH7TKZOnSreffddMWLECFGjRg3x9ddfl/g9J6poTOyIrMDjiZ0QQkyfPl0AEP/5z390dUyd2Hl5eYlbt27pyuPi4oSdnZ1wd3cXjRs3FsnJybprJ0+e1CVwjyv8he7r6yvi4+N15SqVSoSHhwsAYvv27bryv/76S9jY2IjWrVuL1NRUvbY+/vhjAUAsW7ZMV1aY2AEQ3377rdH3aIxGo9ElJr/88ovetXfeeUcAEOPGjTP6uRw+fLjUr/Pvf/9bABAjR440SKYyMjLEw4cPJccTGBgoAIihQ4eK/Px8Xfknn3wiAAgPDw8xbdo0odVqddcmTJggAIgdO3botVX4Gb744ot69S9evCjs7OyEr6+vePToka587NixAoCYNWuWXjs///yzACAaNmwoNBqNrrzw56B+/fri3r17uvKUlBTh4eEhXF1d9RL2/fv3637es7KydOVarVa88cYbBj83hYmdjY2NOH78uK5crVbr/lMQHR2tKy/8+1T074oQBd8XmUwmnnrqKYPvmVqtFunp6Qb3EFUmJnZEVqBoYpeWliY8PDyEn5+fLjkwdWI3b948g/rdu3cXAMSGDRsMrjVo0EAEBATolRX+Ql+wYIFB/cjISAFA9OvXT1c2ZcoUAUAcO3bMoL5GoxG+vr7iqaee0pUVJnYhISFG319xjh07JgCI5557zuDaw4cPhZeXl3BwcNBLNsqa2OXn5wtXV1fh7u4u0tLSTB5PYWIXGxurVz8uLk4AEC4uLiI7O9vo63z44Yd65QCEQqEQMTExBq8/fvx4vURKpVIJBwcH4e3tbdC+EEL07NnT4HtY+HNgLPkuvPbHH3/oygYMGGD0vQnxT+L1eG9vYWI3atQog/qF11atWqUrKymxUyqVul6+x5NcoqqCq2KJrJCnpydmzpyJmTNnYtmyZZg7d67JX6NNmzYGZTVr1izx2qlTp4y21blzZ4OyDh06wMbGBufPn9eVnTx5EgDw66+/4tChQwb32Nra4urVqwbloaGhRl+3OIWv2bVrV4NrLi4uaNeuHfbv349r166hVatWZWq70NWrV/Hw4UP06NEDnp6eFRKPp6cnAgIC9OoXfo8aNWoEJycno9cKF4I8LiAgAIGBgQblnTt3xtq1a3H+/HkMHjwYV69eRW5uLrp162bQPlAwB/HAgQO4cOGCwff9qaeeMqhfp04dAEBGRoau7OTJk3B2dsa3335rUB8AHB0djf4clLb9kri5uaFv37743//+h5CQEAwdOhRdu3ZFaGgobG1tS9UGUUViYkdkpaZMmYLPP/8cn376KSZOnGjy9t3c3AzKbGxsSrymVquNtuXv729QplAo4O3tDaVSqStLS0sDACxcuLBMsRprvySZmZkl3leYABXWk6LwfdWuXbvC4pHyPQJgdNFKca9dWF74fsrz2ZUU0+MLItLS0qBWqzFv3jyjrwEA2dnZktt/km3btmHRokXYvHkz3n//fV3bY8eOxaJFi4wmtESVhatiiayUo6Mj5s2bh6ysrBJ/Acrl8mITrseTqop0//59gzKNRoMHDx7A3d1dV1b4izkzMxOiYCqJ0UdRpV0FW/R1jMUFAElJSXr1pPDw8AAAJCQkVIl4nqS41y4sL/w+VUasbm5u8Pb2LvFn4M6dO5LbfxInJycsWLAAt2/fxu3bt7F27Vo0adIEK1euxLRp0yrsdYlKg4kdkRUbPXo0WrRoga+//ho3b940WsfT0xPJyckGyV12drbRrS8qQmRkpEFZdHQ01Go12rZtqysLCwsD8M+QbEUpfE1j25VkZ2fjzJkzcHR0RJMmTSS/RpMmTeDm5obTp08bbGtijnieJC4uDrGxsQblhd+7whibNm0KBwcHnD592mD7GeCf92BsuL60wsLC8ODBgwr7+VQoFABK14tXv359jBs3DkePHoWLiwt2795dITERlRYTOyIrplAosGjRIuTn5xc7zy40NBT5+fn4/vvvdWVCCMyaNcvocFZFWLlyJe7evat7npeXpxvievxYp4kTJ8LGxgaTJ09GXFycQTsZGRl6c/Kk6tSpE4KCgrBv3z6DfdoWLFiABw8eYMSIEbCzs5P8GjY2Nnj99dehVCoxdepUgyRCqVQiKyur0uJ5Eo1Gg/fee0+vR/SPP/7Axo0b4evri759+wIA7OzsMGLECKSmpuLjjz/Wa+OXX37Br7/+ioYNG6JTp06SY5kyZQoAYNy4cUb38EtKSsKVK1ckt1941m98fLzBtZSUFKN78aWnp0OlUsHBwUHy6xKZAufYEVm5AQMG4JlnnsHx48eNXp80aRLWrVuHV155BQcOHICvry8iIyORkZGB1q1bV8qGq08//TRat26NYcOGwdnZGXv27MG1a9fwr3/9C4MHD9bVa9myJb744gtMmDABTZo0Qd++fREUFISHDx/i9u3bOHr0KMaMGYM1a9aUKx65XI7169ejd+/e6Nu3L4YOHYrAwEBER0fjyJEjCAoKwuLFi8v7tjF//nycPHkSGzduxMmTJ/Hcc8/B3t4et2/fxi+//ILjx4+jTZs2lRZPSYKDg3H8+HGEhoaiR48eSElJwQ8//AC1Wo2vvvoKjo6OurqffPIJjh49igULFuDEiRMICwtDTEwMtm3bBicnJ6xbtw5yufR+hT59+mD27Nn46KOP0LBhQ/Tp0weBgYF48OABbt68icjISCxYsADNmjWT1L6LiwtCQ0Nx7NgxjBw5Eo0aNYJcLsfIkSORnp6Otm3bonXr1ggODkbt2rXx4MED7Nq1C/n5+Zg+fbrk90VkEpW/EJeITK3odidFRUVF6fYiM7aFw2+//SbCwsKEvb298Pb2FiNHjhT3798vcbsTY9t6FG5NYWxjXmNtFda/deuWWLx4sWjYsKGws7MTgYGBYu7cuXrbdzzu999/F8OHDxe1atUStra2wsfHR4SEhIiZM2eKK1eu6OoVbncyZ84co+08yR9//CGGDBkifHx8hK2trQgMDBRTp0412AxYCGn72AkhRG5urli2bJlo06aNcHR0FC4uLqJ58+bi3//+t8GeaGWJp6RtbACILl26GJQXt81HYf34+HgxbNgw3fYqHTp0EPv37zf6GikpKWLKlCkiMDBQ9z0aMmSI+PPPPw3qlvRzU9LneuDAAdG/f3/h6+srbG1tRY0aNUSHDh3ERx99JOLi4nT1Crc0WbdunUEbxf2MXLt2TfTt21d4eHgImUymiyE9PV3MnTtXhIeHi5o1awo7OztRq1Yt0adPH7Fv3z6jnwVRZZIJwQMliYioeDKZDF26dCnVEWlEZF6cY0dERERkJZjYEREREVkJJnZEREREVoKrYomIqEScik1kOdhjR0RERGQlmNgRERERWQkmdkRERERWgokdERERkZVgYkdERERkJZjYEREREVkJJnZEREREVoKJHREREZGVYGJHREREZCX+H/5zkzVdz4R9AAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pca = np.cumsum(pca.explained_variance_ratio_)[8]\n",
    "\n",
    "plt.plot(range(1,21), np.cumsum(pca.explained_variance_ratio_)[:20], linewidth=3)\n",
    "plt.plot([9,9], [0.88, y_pca], color='red', linestyle='--', linewidth=3)\n",
    "plt.plot([1,9], [y_pca, y_pca], color='red', linestyle='--', linewidth=3)\n",
    "plt.xlabel('Number of components')\n",
    "plt.ylabel('Cumulative explained variance')\n",
    "plt.ylim(0.99,1)\n",
    "plt.xlim(2,15)\n",
    "#plt.grid(True)\n",
    "\n",
    "plt.tight_layout()\n",
    "\n",
    "plt.savefig('PCA.pdf');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Preprocess and save data files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data saved in file sec3_1_data_output_teff.pkl\n"
     ]
    }
   ],
   "source": [
    "# t_eff\n",
    "u.preprocess_set(new_X, y_t, 'teff')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data saved in file sec3_1_data_output_logg.pkl\n"
     ]
    }
   ],
   "source": [
    "# log g\n",
    "u.preprocess_set(new_X, y_g, 'logg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data saved in file sec3_1_data_output_feh.pkl\n"
     ]
    }
   ],
   "source": [
    "# [Fe/H]\n",
    "u.preprocess_set(new_X, y_m, 'feh')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}