MDN_Ages.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Import packages."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "# import packages\n",
    "import numpy as np\n",
    "import time\n",
    "import pandas as pd\n",
    "\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.colors as colors\n",
    "import matplotlib.cm as cm\n",
    "from matplotlib import gridspec\n",
    "\n",
    "import torch\n",
    "from torch import nn\n",
    "from torch import distributions\n",
    "from torch.nn.parameter import Parameter\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define plot properties\n",
    "from cycler import cycler\n",
    "import matplotlib.cm as cm\n",
    "\n",
    "from matplotlib import rcParams\n",
    "from matplotlib import rc\n",
    "from mpl_toolkits.axes_grid1 import make_axes_locatable\n",
    "\n",
    "def rgb(r,g,b):\n",
    "    return (float(r)/256.,float(g)/256.,float(b)/256.)\n",
    "\n",
    "cb2 = [rgb(31,120,180), rgb(255,127,0), rgb(51,160,44), rgb(227,26,28), \\\n",
    "       rgb(166,206,227), rgb(253,191,111), rgb(178,223,138), rgb(251,154,153)]\n",
    "\n",
    "rcParams['figure.figsize'] = (9,7.5)\n",
    "#rcParams['figure.dpi'] = 300\n",
    "\n",
    "rcParams['lines.linewidth'] = 1\n",
    "\n",
    "rcParams['axes.prop_cycle'] = cycler('color', cb2)\n",
    "rcParams['axes.facecolor'] = 'white'\n",
    "rcParams['axes.grid'] = False\n",
    "\n",
    "rcParams['patch.facecolor'] = cb2[0]\n",
    "rcParams['patch.edgecolor'] = 'white'\n",
    "\n",
    "#rcParams['font.family'] = 'Bitstream Vera Sans' \n",
    "rcParams['font.size'] = 23\n",
    "rcParams['font.weight'] = 300\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Download DR16 spectra.\n",
    "\n",
    "> Download in parallel."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "# open batch script\n",
    "f = open(\"ys_run.sh\", 'w')\n",
    "\n",
    "# initiate text\n",
    "file_text = \"\"\n",
    "\n",
    "# loop over different batches\n",
    "for c1 in range(26):\n",
    "    file_text += \"sbatch run_script_\" + str(c1) + \".slurm\\n\"\n",
    "    os.system(\"sed 's/XX/\" + str(c1) + \"/g' run_script_helios.slurm > run_script_\" + str(c1) + \".slurm\")\n",
    "f.write(file_text)\n",
    "f.close()\n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Check which spectra were not downloaded."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import os\n",
    "import astropy.io.fits as pyfits\n",
    "\n",
    "#--------------------------------------------------------------------------------------\n",
    "# list all the files\n",
    "list_files = os.listdir(\"../apogee_dr16\")\n",
    "print(len(list_files))\n",
    "\n",
    "# load catalog\n",
    "catalog = pyfits.getdata(\"../allStar-r12-l33.fits\")\n",
    "catalog_id = catalog['APOGEE_ID'].astype(\"str\")\n",
    "telescope_array = catalog['TELESCOPE'].astype(\"str\")\n",
    "print(catalog_id.shape)\n",
    "print(np.unique(catalog_id).shape)\n",
    "\n",
    "#--------------------------------------------------------------------------------------\n",
    "# not downloaded\n",
    "ind_missing = []\n",
    "\n",
    "# loop over all spectra\n",
    "for i in range(catalog_id.size):\n",
    "    \n",
    "    # monitor progress\n",
    "    if i % 1000 == 0:\n",
    "        print(i)\n",
    "        \n",
    "    # check \n",
    "    apogee_id = catalog_id[i]\n",
    "    telescope = telescope_array[i]\n",
    "    if (telescope == 'lco25m'):\n",
    "        filename = 'asStar-r12-%s.fits' % apogee_id.strip()\n",
    "    else:\n",
    "        filename = 'apStar-r12-%s.fits' % apogee_id.strip()\n",
    "        \n",
    "    if not(filename in list_files):\n",
    "        ind_missing.append(i)\n",
    "        \n",
    "# convert into numpy\n",
    "ind_missing = np.array(ind_missing)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(ind_missing.shape)\n",
    "np.savez(\"ind_missing.npz\", ind_missing=ind_missing)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Inspect network architecture."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define container\n",
    "class model(torch.nn.Module):\n",
    "    def __init__(self, num_neurons, num_features, mask_size):\n",
    "        super(model, self).__init__()\n",
    "        self.deconv1 = torch.nn.Conv1d(1, 8, mask_size, stride=3)\n",
    "        self.deconv2 = torch.nn.Conv1d(8, 8, mask_size, stride=3)\n",
    "        self.deconv3 = torch.nn.Conv1d(8, 8, mask_size, stride=3)\n",
    "        self.deconv4 = torch.nn.Conv1d(8, 1, mask_size, stride=3)\n",
    "        \n",
    "        self.batch_norm1 = torch.nn.Sequential(\n",
    "                            torch.nn.BatchNorm1d(8),\n",
    "                            torch.nn.LeakyReLU()\n",
    "        )\n",
    "        self.batch_norm2 = torch.nn.Sequential(\n",
    "                            torch.nn.BatchNorm1d(8),\n",
    "                            torch.nn.LeakyReLU()\n",
    "        )\n",
    "        self.batch_norm3 = torch.nn.Sequential(\n",
    "                            torch.nn.BatchNorm1d(8),\n",
    "                            torch.nn.LeakyReLU()\n",
    "        )\n",
    "        self.batch_norm4 = torch.nn.Sequential(\n",
    "                            torch.nn.BatchNorm1d(1),\n",
    "                            torch.nn.LeakyReLU()\n",
    "        )\n",
    "        \n",
    "        self.mlp = torch.nn.Sequential(\n",
    "            torch.nn.Linear(85, num_neurons),\n",
    "            torch.nn.LeakyReLU(),\n",
    "            torch.nn.Linear(num_neurons, num_neurons),\n",
    "            torch.nn.LeakyReLU(),\n",
    "            torch.nn.Linear(num_neurons, num_features),\n",
    "        )\n",
    "    \n",
    "    def forward(self, x):\n",
    "        x = self.deconv1(x)\n",
    "        x = self.batch_norm1(x)\n",
    "        x = self.deconv2(x)\n",
    "        x = self.batch_norm2(x)\n",
    "        x = self.deconv3(x)\n",
    "        x = self.batch_norm3(x)\n",
    "        x = self.deconv4(x)\n",
    "        x = self.batch_norm4(x)[:,0,:]\n",
    "        x = self.mlp(x)\n",
    "        return x\n",
    "    \n",
    "#-----------------------------------------------------------------------------------\n",
    "# initiate model\n",
    "model_conv = model(num_neurons=30, num_features=1, mask_size=11)\n",
    "˜\n",
    "# forward model\n",
    "spectra = np.zeros((100,7214))\n",
    "spectra = torch.Tensor(spectra)\n",
    "spectra = spectra[:,None,:]\n",
    "y = model_conv(spectra)\n",
    "print(y.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Infer stellar ages with MDN.\n",
    "\n",
    "> Plot inferences."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7214, 2556)\n",
      "(556, 1)\n",
      "0.14189097\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAnAAAAHiCAYAAABoVfF2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzs3Xt4lNW1+PHvzo1wiQSSwEAgkBguCUQJeIlYilWjoFbPqS2lVKttbY+t1va0tj3tr7XWXm1ta6lW6umxVSyHUrVHVFSCFkUwIIEokXAJCQQIkwswECAhmWT//njnnbwzmXsml0nW53nyhMy8l50JMCtr772W0lojhBBCCCFiR1x/D0AIIYQQQoRHAjghhBBCiBgjAZwQQgghRIyRAE4IIYQQIsZIACeEEEIIEWMkgBNCCCGEiDESwAkhhBBCxBgJ4IQQQgghYowEcEIIIYQQMSahvwcgeiY9PV1PnTq1v4chhBBCiCgoKytr0lpnBDtOArgYN3XqVLZv397fwxBCCCFEFCilDoVynEyhCiGEEELEGAnghBBCCCFijARwQgghhBAxRgI4IYQQQogYIwGcEEIIIUSMkQBOCCGEECLGSAAnhBBCCBFjBn0Ap5SyKaVuU0r9Xim1WSl1TimllVLlEV5PKaXmK6V+qZR6Ryl1XCnVrpRqVEqtV0p9VimlApyvg3zYI/9uhRBCCDEUDIVCvkuB30XxelcDGyxfVwM1QDZQ7Pr4jFLqVq31+QDX2Q74ev54tAYqhBBCiMFpKARwpzECru2uj+nAz3twPYURsD0KrNZaN7ifUOp24L+BG4GHgO8GuM6ntNYHezAOIYQQQgxRgz6A01o/BTxlfq2UurOHl9wGzNBat/u410ql1GTgZ8BdSqnvaa07e3g/IYQQQggPg34NXLRprU/7Ct4sXnV9HgsEbUYrhBBCCBGuQZ+B6wfDLX9uCXDcD5VSEzF+BkeBN4G/B1k3J4QQQgghAVwv+Izr8/ta69MBjvuC19d3AD92bX7Y0TtDE0IIIcRgIFOoUaSUmgfc7fryl34Oex1YBkzDyNZlYAR91cBUYL1rHZ0QQgghhE8SwEWJUmo88AJGVvOfWuvVvo7TWi/SWv+v1rpKa92qtW5yHXs5UAukAT8Kcq8vK6W2K6W2NzY2Rvk7EUIIIcRAJwFcFCilRmNsXsgCyoA7w72G1roJ+IXry38PVAxYa/2k1voSrfUlGRmyT0IIIYQYaiSA6yGl1CjgNaAQ+BC4Psjat0C2uD6PdX0IIYQYQMrKyvp7CEIAEsD1iFJqBPAKUATsA67VWvekk0Kb5c+ywUQIIYQQPkkAFyGlVDLwIvBR4CBwjda6p31MZ7s+tyIttYQQQgjhhwRwEVBKJQLPA9cCR4CrtdZHenjNBOBbri/f1Fo7ezZKIYQQQgxWEsD5oZR6Ryl1UCn1Da/H44G/ATcAdozgrSbEa/5SKXWHUirF6/HJwHMYU7FOjD6qQgghhBA+Dfp1Vq7gaKfloWGuz7OVUk2Wx3+ltf6V5etJwBQg1euSS4BPuf7cCvwlwIbRr2mtrfeeidHg/n+UUtXACWA0MANQruvdpbXeGsr3JoQQQoihadAHcEA8Rm21YI+PCPF6wyx/nur68Ge019dPYGTtLgUmus49j7F7dQPwmNb6QIjjEEIIIcQQNegDOK31QYzsVrjnTfXz+F+Bv0Y4ltcxOjEIIYQQQkRM1sAJIYQQQsQYCeCEEEIIIWKMBHBCCCGEEDFGAjghhBBCiBgjAZwQQgghRIyRAE4IIYQQIsZIACeEEEIIEWMkgBNCCCGEiDESwAkhhBBCxBgJ4IQQQgghYowEcEIIIYQQMUYCOCGEEEKIGCMBnBBCiEGlrKysv4cgRK+TAE4IIYQQIsZIACeEEEIIEWMkgBNCCCGEiDESwAkhhBBCxBgJ4IQQQgghYowEcEIIIYQXXztZS3bbWbGxipLd9n65vxBWEsAJIYQQQZTstnPf6p28XHGM+1bv7JMgTohAJIATQgghgti0v5GW9g4AWto72LS/sZ9HJIY6CeCEEEKIIBZMy2B4YjwAwxPjWTAto59HJIY6CeCEEEKIIIrzbSxfWshNsyewfGkhxfm2/h5SxGR93eCQ0N8DEEIIIWJBcb6NsS25zIvh4E0MHpKBE0IIIYSIMRLACSGEECGqrKzs7yEIAUgAJ4QQQgw4EiiKYCSAE0IIIQaAaG8ukM0Kg5sEcEIIIYQQMUYCOCGEEEKIGCMBnBBCCCFEjJEATgghhBAixkgAJ4QQQggRYySAE0IIIYSIMRLACSGEEGJQGEqlUwZ9AKeUsimlblNK/V4ptVkpdU4ppZVS5VG49gKl1ItKqQalVKtSqkop9Rul1Ngg5yUopb6hlNqhlDqjlHK4xnZHT8ckhBAi+oZSYCBiw1BoZr8U+F20L6qU+grwGEYQXAd8COQB3wSWKKU+orU+5OO8YcBrwFVAp+u8JGA+MF8pdQ1wh9ZaR3vMQgghhBgcBn0GDjgNbAB+CXwS+H5PL6iUKgT+gPH6fQ2YpLWeB0wC3nB9/ruf03+BEbzVAhdrrS/SWs8EFgKngNuBu3s6RiGEEEIMXoM+gNNaP6W1LtZaf09r/TxwLAqX/SEQD6zSWj9mZsu01icwMn7NwOVKqRutJymlMoB7XF/epbWusIzzbeA7ri8fUErFR2GcQggheplMr4r+MOgDuGhTSo0CFru+fML7ea11E/Cc68tPez19C8Z06QGtdYmPyz8DnANsGBk5IYQQok9IIBpbJIALXyGQDLQBW/0c85br8xVej5tfv+3rJK11K7DNz7lCCCGEEIAEcJGY7vp8SGvd7ueYA67POUqpRB/nVgW4vnnujAjHJ4QQohdIhkoMJBLAhc8sEXIiwDHmc3HABRGeOyb8oQkhhBC9qLoabr0Vzp3r75EMeRLAhS/Z9bktwDGtlj8Pj/Dc4f4OUEp9WSm1XSm1vbGxMcClhBBiaCnZbWfFxipKdtv7eyg9UlNT4/e5fssEvvkmzJ8PH/sYDPf7FiX6iARw4TMDrKQAxyRb/twS4bkt/g7QWj+ptb5Ea31JRkZGgEsJIcTQUbLbzn2rd/JyxTHuW70zYBAn06Fh0Br+8AdYtgxWrYJ77wWl+ntUQ54EcOE76fqcFuAYc6q0E6MOXSTnngxwjBBCCC+b9jfS0t4BQEt7B5v29+0Mhb+gMNRgsbKyEjAC0Tcr64NmEUO9bo+C1fPn4UtfgiefhC1b4OqrI7+WiCoJ4MK31/U5y2uDgtWFrs/VXhsdzHNzA1zfPHdvgGOEEEJ4WTAtg+GJRgnN4YnxLJgW2QxFf07DmlnEnYcdQbOIwZgBIUQYxNntRsB24gS8+y7k5EQ8FhF9EsCFbydwHmMatMjPMWYNt3e9Hje/XuDrJKVUMnCZn3OFEEL4YAYnxfk2li8t5KbZE1i+tJDifFvY1wpnGrY39HcW0W37drjsMiguhueeg1Gj+mccERoKU+QSwIVJa30Go5cp+Gh5pZRKx2jZBbDG6+m1GBsYLlRKFfu4/OeAEUA9fmrFCSGE8K8438bdV+VGFLxB/wdQwbKIfRKYrFoFixfD734HDz4IcRIqDETyU/FDKfWOUuqgUuobPp5+CGN92zKl1L1KGas5lVJjgdVACvAe8Ir1JK11A/BH15d/VkrNttzvo8CvXF/+RGvtjOo3JIQQIqhgAVReXl6v3Ldkt53nyw4DsHxpIbkj2yLOIkasowO++134wQ/gjTeMciED1FDIsAUz6AM4pdRkpVST+YHRhB5gtvVxpdR3vE6dBEwBUr2vqbXeAXwd0K7rHVFKlQFHgGuAOuDTZo9UL9/DyK5lAe8rpT5QSu3B6N4wGlhFV5AnhBCiD0U6DduTgMKctt1U1cR9q3cCcGl2Wq8Eb37X9zkc8J//Cdu2GR8XXRT1e4voGvQBHEbT+TTLxyg/j48I56Ja68eAq4CXgGHAbOAY8ChwkdbaZxEfV7usa4BvAR9gbFqYiLHm7Qta68/6CfyEEGJI6utsS0+nYcPlPW27atsh3qs5zuP/eD2q9/G7vm/vXigqgkmTYP16SE+P6n1F70jo7wH0Nq31QSDsgjVa66khHPM2EaxVc02P/tb1IYQQIgaUVjdRXutgyfDMqAZ349vqGJ4YzxkgKT6OzVXHOVbfzCPr9zJ91sVRu5ev9X3FB3fAnXfCz34Gc+dCor/iCmKgGQoZOCGEEKJHSqubeGT93rB3p4ZSG64oJ53lSwspSDnPlRem0dbRCUCrszOqmyg81vclxHHb22vgrrvgn/80ar2JmOI3A6eUeirK99Ja6y9G+ZpCCCFEryuvddDqNAIrd/YqwsyYr6CuON9Gad54ii6fQmmN0RI7OSHOvYmirKyMefPmRTj6rnssX1rI82sb+dE7f2bCsUOwdStMntyj64r+EWgK9U6MRfrR6pehAQnghBBCxJw5Wals2FNPq7Mz5CLBkazdM4Os721L4fPXzYj6OrziC5yMffoXTCgshDWbYERYy7/FABJsDdxh4C9RuM8XMHZ1CjFkReM3aCFE7wj277MoJ537rzMycUsW9255j+J8Gy9kp1GUE/lmArMsyfi5duorK43yJ1u2UHbLLbB0KSxfLv1MY1ywAK5Wa/3jnt5EKXUtEsAJIYSIUZWVlRTl5VGUk868MIK3SjN4CnJMNH+5M3ebNlY1sWf1Tm4e6SBv7//Bn/5k1Hj7yEeCBm/yC+fA11ebGCTMF0IIMaANhOKw0RiDdbdpe+t5hq9eBc88A2+/bQRv/Tg2ET2BMnCFwJko3ecOuuqvCSGEEKKXmGVJOttaWPmPBylPaoOnn4aZM0GCsEHDbwCntX4/WjfRWh+I1rWEEEII4V9RTjpzE0/z/u9XM+HWz3Bg4WWQktLfwxJRFtYUqlLqAaXUD5VSw3prQEIIIcRAUrLbzqPPrqW0uonKysr+Hk5w//oXV3xpCaOWfZqpTz0O8fH9PSLRC8JdA/cAsFRrfb43BiOEEEL0B3+BmbkhYK+rM8Kuo44+HlkYOjvhoYfg17+Gdevgyiv7e0SiF4UbwDUAZ3tjIEIIIcRAY90Q0OrsZMuB4x7Pm83hS6ub+mN4Xc6cgSVL4NVXjQ0Ll17av+PpB0Ntk0W4AdxbQL5SamRvDEYIIYSIpp6+qS+YlkFnQ9cy7v31ze42Wtbm8I+s39t/QVxNDcyfDxdcABs3Dtlm9KXVTazYWBVym7NYF24A9xPXOY8qJRUAhRBCDG7F+TYunpRKx7lTADg7tbs/qXd2rrw28unViAPNf/0LrrgCvvhFyr76VRg2NJeol+y287NVG9y9avs9I9oHwg3g0oGfYrTZKlNKfVMptVgp9VF/H1EfsRBCCNFD4WxGWDTbRnyckbNIsvQntTaHT06IY05WavQH6o/W8Pe/G10Vnn0Wvv71HnVWMKeCYzV7tWl/I22WXrU9CaZjRbBODN420tUf9WLXRyA6gnsIIYQIgVTL7xtFOenMmZTKzNkTGOtMcLfRMvuWrnn1NHOyUkNqfRWVdVrt7fDlL8Obb8KWLXDhhX4PNVtqXTssA39/VcypYMfhY7xzaifLlxYytuej7FMLpmXw54Q4NDA8Mb5vg+l+Em5wVYsRlAkhhBii+itwtN63L8Zgvcf40cO5+6pcKivbPY4pzrcxtiW3x/cKOSN4/Dh85zswdSr85S8Bg7ddRx38bJ/RUqv8zF6mz7rYZ2BmnQpuae9g0/5Gbomx5pfF+TZuL5rCiYQMliwuZGzL0f4eUq8LawpVaz1Va50dzkdvDVwIIYQYKNZseJefPrs+4inIXUcdvLh5V+C1W2Vl8LnPGTtMX3gBRgbeT7jP3uyxRs9cu+fNOhU8PDHePUUcawoyU7n7qlx3hnSw66teqEIIIcSgLPVQWt3EytJDbKpq4r7VO8MO4kq2fsDK0kPsrW/mZ6s2+D7/tddg0SL4z/+Eu++GuOBv39NtKR5r9PwFZuZU8E2zJ7B8aWFUAqDB+HMeaMLtxDCztwYihBBCxKLyWofHAnp/mS5/ao+fc5/f5uxk1bZDvFdz3MjGdXTA974Hjz8OGzZQmjMn5M0GBZmpLF9ayILcdO6/bkbAwKw43xYz2atgGy5ifUNGqMLNwH2olNqklLpTKTWiV0YkhBBCuPRGJsdc2B+NN/iysjLmZKWSlGC8nUYyBZmVNsJ9fkKcYnPVcfbWN/P4S+U0XrMI3n0XnnmGX9uH8Yt1lWGVyijOt3HrvMkhbbCIBdbae76ynaXVTQGfH0zCDeDOA1cC/wMcU0r9SSl1efSHJYQQQgQXSrbFGgSaAcC/tn8YtTf4opx0bi+awoLc9IimIHPHpXB70RRmjE9h2rgU2jo6SWk9w8/X/pbqEWlQUkLpyQ5WvH2ADtc5wUplRDNIHUh8bbiwKq91BHx+MAk3gLMBXwV2ACnAl4AtSqkPlFJfV0qlRXuAQgghBqeeZteCZWN8CRYARKK0uol99mam21IinoIsyEzl0uw05uemcc3BHVy/fysvz7qK0488ComJlNc66OjsKgIRB9hPt/r8ns0slLkmL1j/1lharxZsw8WcrNRBsSEjFOHuQj2ttV6htb4UuAhYDpwAZgO/BY4opVYrpa6L/lCFEEIMFpEEDd6lNryDsTWvvhX0GqHuuAylrEdZWRml1U08sn4vm6qMjQw9ynhpTfpbb/CLlx7lrexC1s+c737KGpjEAXFxiu21J30Grt5ZqH325sjHNMAE23BRlJMe9Q0ZA1XEu1C11hVa628AE4FPAyVAIrAEeFUpdVAp9YBSKis6QxVCCCG6eAdjoRRvNQOAwsmpft/gwwkuN2zZQatlA4J3Rm/z5s2hXej8eXjjDZK3b+XfPvcIjaPG0m5p22UNTOZmjcHpysb5yiJ6Z6Gm21JC/n5iQbANF7G0IaMnelxGRGvdrrX+h9Z6ETAV+BnQAUwGfgRUK6XWKaUW9/ReQgghhMk7GxPqQv3ifBtX5413v8H3ZApxui2FZNcGBHWmgQXTMtzXK61u6tpNGsjp0/DTnzIxOZm6r32Lk+kTiB+V1q30hxmYLJptC5hFNIM9c00eMCR2ZQ41UWtzpZS6CPgisMxy3ePAKGARcL1SaiPwKa31iWjdVwghxNBldkKYl2+jrCyy6vul1U1sCNJuyp+CzFSycjLYsOUsFzjjKXaN49ev7+H36yo5Vd/MI+uNLgjF+TbjXlt2cO38uca93n0XVqyAxYth9mxmZY/n6rmF/LzqFT7pp/RHUU4602dlsubV0yxZ7DuLWJxvo37eZABWlh5Cpzp559RO7i2IoygnXdqwDQI9CuCUUqOBzwJfAAoxeqRqYAPw38D/YQRwnwO+DVwF/Ab4fE/uK4QQQpgCrVerrKwkLy8PgGeffZbbbrvN4/mS3XYeWb+XZrtnuynznFAU5aQz+vxkamqcgBEQrii3u3eMWrsgmPd679QeckrLWLzyUbjlFqNIb00NYARfpXnjA2YUrYFrIGaT90TMnaunB01JkaEuogBOKXUNRtD278AwjMDNDvwF+LPWusZy+Eng90qp1cA+4IYejVgIIYQIIJzM0qb9je41bGagNb25MqwAzpuxY9SzzVVKcqL7XnG6ky+UvsBMexVbnn4OyrtvvsjO7upEWbLbzpqNVSwZnhl2k/mh2OR9qAi3E8MDSqkaYD3wGSAJeA34BDBZa/3/vII3N611PbALkNBfCCGEh74sZWG914JpGe41bN5rzrwze6E2nLduIjA9tbmGlOREUs+f5T9Kn2fyqXpuueM3vNY5JuC1vEulhFK816o430bxJBX2OkEx8IW7ieFBYApwFHgIyNZa36i1/j+tdUfAMw3bgbfDvKcQQog+9uyzz/bZvfqzDllxvo37r5tB3oQLmD1xtN/jTp48GfI1zU0E40YluR9rae8geV8lfyx5jGOjx/Hja79Ee8rooHXKvEulBCre60/uuBSPXZmy9m1wCDeAewn4ODBVa/2g1vpwOCdrrb+htf5YmPcUIuYNld58Qgw0NTU+J4W6OdB4xl1XLVjh21CdPNfu/vPiqq38x4/u4oJ77ybty3ey+KJJIdUpi6RUSjhCDZ7l/7CBJ6w1cFrrW3prIEIMVuYUiOPwMd45tXPQF5cUItaYzejNhf7ehW9DnTq12rS/kXZXrbal5a/zw0Nl7Prdb2H2bAoqK8nLMzYglOy282ZlPbbmYdiP1HNZkueUqlkqxdxxOrbF2Gnbl1lL+T9sYIpaGREhhG++WvfIf35CREc0Apk5Wam8YlnoPz01xSNo23XUQXXDGXYddbDk2itCuuaCaRms0U4Ka3ZyWVIyFS+sZ2RK10qjyspKTgzP5J5VOzhWeYBhzcNwOhzsbTnEtbvt7Drq4K16Y+NCpKVSvLOPkW6GCPf/sFhqzRXLggZwSqlkYATg1Fqf9nrugQCnvqC1rujh+ISIeQumZfCPsiOcZ/D35hMiVl2YMYrUCWO479OF1O847n68ZLedp17bSvPpFp7ecpDdLSnc5wqqAike0cqT//ojy0ck4/j1H/jMVXMoKyvzCAxXbT1EW0enx3ltzk5+/8Z+dlhqty1fWhj27lPr+N+srKdpzEHWNzS4s2j3FsSFXPNO/g8bmELJwG0ArgC+BTzq9dyDGHXffCkGFkQ8MiEGCe8pEMm+RY8UIx1aQl3PFo6uOnCnSdGnuj2/aX8j7U7jbc7Zqdl+yFgnFzCoKiuDH/2IkcuWwcGDXDZzou/jVNcfnY569593HzvlMaW7aX8jt0wK/3uramhm1eqdHDvsYP+OIySMvxDoqgcXqp78Hyb/RntPwABOKVUEzAfKtdbewZvpOOBdxGYOMF8pdZnWelvPhylEbAu16KYQwlNpdRPltQ6WDM/slev7rANneX7BtAyeSFC0Wh4LGFQ98QSV3/0u855/HsaOhV/8wu+9l102hc1VRrbPrIIP0Km7dhi6M14t4XeZqD1+jpZRxtRnh9YMU13XDHczhPwfNvAE24V6K8bfqUcCHLNHa/0p6wfwHYy/j0uiNM4eU0otUEq9qJRqUEq1KqWqlFK/UUqFlZlWSk1VSukQP/7i4/yDIZyXHL3vXAghYpOZHTNroFU1NAc/KQSl1U28WVlPyW57wDpwYAQucyePISkhnnhlREA+pxGdTli7Fv7wB/jRj6C4OOAYampqKM638fiyuUxKdnJZ9lgSExRORz3DE+O5Jm+8u3ZbpFn7rLQR7h2syQlx3L0wd9DWgyvZbef5ssNh18mLZcGmUK/AaEz/cpjXXQuccZ3f75RSXwEewwhY64APgTzgm8ASpdRHtNaHQrxcK7A5wPPJgJkv3hLguAqge77e0OnncSGEiHmhLnK3Zsda2juoPX7O77HWllkmc/3X2JlN7vVeZlB48rDDPRV6/3Uz2LDlLNfON3qP/uT5GncnhJLddrYfOkmbs4PR6ZOZMzmV+z5tBFXPPrvBuOiJE+5MW9mTT8Jh3xW2zJ6r020pjHI9VpxvY1bmaC6elkHieQdn40fz/aXGOjxzp2oofL2mxZdfxG2Wvqr3XD+TsvSzPeobOxCV7LZzz6odnKhqouz0Hnc7tMEuWAA3HTjovXkhGK11h1KqGsiNeGRRopQqBP6AEbx9DXhca61dmbc1wDXA34GiUK6ntbYDHwlwvzuAvwItruv68zWt9cZQ7imEiH2xthaoN9abhWvBtAxWJsTR6uxkeGI8WWkjQj7XLH1x7LCDg65m8tA9KNy0v5FbXL1M83xkpVZtO0SHqxyIU2tQeGbEDh6Eb38bLr8cCgth1Khu1zDHY/ZB3XrwBMWTlPvxD4+eYuSEZiaPHcnEieON4HBHyN8q4DnVbA1eQumrGutWbevaDNLeqVm17RD3Fgz+IhvBplBHY6xx8+cr+J9ePec6v7/9EIgHVmmtH9NaawCt9QlgKdAMXK6UujFK97vT9fmFcANfIYRByhAI6OqSYE775Y5LCem8srIyj9IX1mby1ilT61RoTU2Nz3pvTc3nAYhLNgIzx9m2ridLS+Hhh+Eb34BPfhLi/L+lWgPHNmcntcfPuYPMQyfOsra8jvLa0Ls9WHlPNQ+laUSg+1ZKf1srB5lgAVwL4PvXCUBr/Set9Vo/T18AHus++5xSahSw2PXlE97Pa62bgOdcX346CvebCix0ffnXnl5PCCGiYaAExGY1/3ACjKKcdI82UNbrBOoKsGBaBqqpGvBc22YGhYWTU0NaX5aeMszj69SRSdDZCd//PqxeDf/1X5z4+B08X3Y44Bo9M3B0OupJSogjK22ER5CpgYq60xGt8/POKnq32zKng/tKaXVTn3ZtWHb5FJLijXAmMU6x7PIpfXLf/hYsgKsHspVSieFcVCmVBOQADZEOLEoKMdaktQFb/Rxj7qCNxnq9OzA2b9QCbwY59m6l1MtKqTeUUn9TSt2tlArt10shhIgx1qbsj6zf6zOICyXQtF7ny79b4zcYtAZq9183wyNQK8pJ5+q88QGDt11HHTxfdpj8CaOJjzOmOxPjFLNVI9xyC2zeDA89xK6E0dy3eiebqppYW17HM+8edF/DmtGzjqd4kiJ3XAoLpmUQbykl0nHOEXCdnz/eWcVot9sKR39kA83NIAty0/nuoplDplRTsABuKzAcuD7M6y7GCJxKIxlUFJm7wQ9prdv9HHPA9Tkn3EDVSimlgM+5vnxGax1sI8KngRuBq4FlGBnCaqVU4K1LQgxRAyWLNFj09evpPaUZSVP2cK9jBmqhrP+qq6tz/7mqoZmVpYfYVNXEf2+qJn3kMDJShvGji1OY9affwqRJUFJCzfHj7LM3e2TRnt9xxG8vVXM85lRwcb6NuxfmYhaEi49TYa3zM3lPNffnerdg2cDeUpxv49Z5kwf1Wj9vwQK45zD+Zv1MKTU8lAsqpUYAP8X1d7lnw+sxcy3niQDHmM/FYUz7RmohRtYRAk+fbgG+jLELdiQwBvg4sBNIB9YqpeYGupFS6stKqe1Kqe2NjY09GLIQQoTHe/oy1EDQ2pRdOY6GnSUyd5SmJCd6lMbojWxT7fFztJnr1To6qW9uRdU3kP/At+D66+GJJyApCYDpthTPLJrW3XqI1l+GAAAgAElEQVSpBvLt62cyLyuVwsmpzJmUGvI6P2/mVDPQr03nB1I2cLALuE1Da71WKVUOXAy8qpS6TWt9xN/xSqlJwN+AfOADrfWLUR1t+Mx6am0BjrGu0wspSPXjTtfnTVrrA/4O0lov83roHPCyUuoN4B1gLvAr4NoA13gSeBLgkksuGSLLNcVQJtm3wPpqh6uvpub+yjV4j8lazX+sMyGsTElpdROP7Wrg2GEH9s01fOHKbPZVnGZOVipFOelUVob3i6xZB65ot93dHN76y3BW2gj2tmgjiNOaeUcqGV63j3V3f58ZV0xzH1fV0MzphASuv2QCz1UYUVxyQhzTbeEFYeNHD2dk2gg27jvHCzuOUBRh8GW+TubP5+aR4WW/ovF3yMwGltc6WLK40P36iugLZZ/tEmAbRlusKqXUS8DbQA1wFiOLNBUjA/VxIAlwMDCK+JrBWVKAY6xFc1siuYlSaiRG0WOIcPOC1rpFKfX/gFeBjymlxmitI9uSJIQY9HozqPV3bV9NzcNp8WRW86+s9LeixbfyWgct7SPd921ubXdnm0w1NTXd6sCB59QodK8Dd29BXLdgMndcCrZJY9i21859637Pmw3VPDsxl8VXzYfzRqD369f3sLa8jvjUDjJSGrhm5jha29O4dv4MRp8PL6Dcaz/Nkfo6zjef51TTWe5ZtYMrTn3g8/sJxPt1CicTaGVteh/JerKinHSKctJ7td6cv2Azlkr19FSwKVS01lUY7bT2YARCn8DoifoiRp/UF4Hfux5Pch13pdZ6fy+NORxmAJQW4BjzF8hOINKyH5/C2K17DvhHhNeArsK/cXRNxwohxIBgnQbtaVPzcHYqzslKDXhfXzXrzCD08ImzHvcJdY3WqKZDfPGt/+WKiSPYfeMSpk5Mcwd6JbvtrHj7gLtaRUt7B63tHd3WYO066uDDo6cCLuQv2W3nQOMZNBA/wqi81dbRGdFmBu/XKdxMoDkec5PIfat39ttUrAguaAAHoLXeA1yEscvydYwuC8ryccb1+B1Agda6ezGd/rHX9TkrwAaFC12fqwNsdAjmTtfn57TWPen1Yp3qHfxVCIUQPdaX08vmNGhPWzy98kEdv1hXGfJOxaKcdJYvLQxY+sM70wZGkLj5QJN7x+quo47Q1mgdOQIrVsDcudhefZGrCiYxdlRXOZFN+xvdxX0B4hXdgqWS3Xae3nKQQyfO8vBre/wGQmZ9Oquk+LiINjOYr5P58ynIDH/9ma8saywaCpm4kAI4MLoraK1Xaq1v0FqPxijSOwkYrbUe7Xp8pda6o9dGG76dwHmMzKC/Tgtm3bZ3I7mBUiob+Kjry79Gcg2L2ZY/+11rOBjImiYhYlNxvq1bXTZffBXFBSOoemNPA+YbRag7FYvzbQFLf3hn2sCYUnR2GIFWq7OTffbmbnXguq3Fe/ppWLkSPv5xo1yIUnizZiLjlHLtJMWjF+eqbYdwuoI8szuALwumZbjLlACMT0nm8WVzu21mMF/PYIFJqD8fk3cmNJpZVtG7Qg7gvGmtm7XWdT3MOPUqrfUZ4DXXl3d7P6+USgc+6fpyTYS3MWu/HQQ2RngN03dcn3drrWXlpxCi1/gLsHpbea2DTu2ZvfKVBQvnl7yqhmZ3ps2a0ZuTlUpCfPfNBdY6cGYAU+84B88+Cz/5CbX/tpQ3Geu3HIiZiSycnMrnr5zKnMmprKnqZFNVE4+s32ucF2J3gOJ8G3MmGbtQF07P4LYrpvRZHTPvmm0lu+1Ry7IOZgMlARFxABdDHsJY37ZMKXWvq14brl6oq4EU4D3gFetJSql3lFIHlVLf8Hdhr9pvT5ttugIcf79S6mtKqTSvx9OUUn/CWEsH8EDo354QQsSOOVmpJLmmMONd2atQdqSaZUR8TUXWHj/nzrR5Z/TGpSRzyZQx3HxxJvvszR7nP/6P13lk/V42bq8ifssWmg8c5A+/fJbnDrex87CDlaWHut3PXG9XnG8jK20E++zNrNp2yL2uzsz0Lbt8CgmWAsCBugOMHz2cq/PGM3nsSPdjN9xwg9/jozU96KsvLISfxRP9w28Ap5T6nFIq3AK+/q51vVLqc8GPjD6t9Q7g6xi///wBOKKUKsOYorwGqAM+7SP4mgRMAQItIlgIZLuu/XQIw5kELAcalVLVSqmtSqkPADtGbbhO4Lta6/6unycYOL9lCTGYFOWkc3vRFG6aPYHvLZ7Jt6+fGfSc0uom7lu9k52unaMlu+0eGcSstBHuTJu5rq202siG1TlaeP+wg3/uPMKmqiZ3hi47O5vyWgcj95fx25cfpXnYCFYv/BSP7jyOBjrOHKfN0kPVW8luO+sqjrGpqok3NpWSGOeZ6SvOt3HH/KlMGTvSoztAVUMzb1bWR9QyK9r89YUVsSHQQvm/YtQlez0K9/kBxk7WZ6JwrbBprR9zBUr3u8YxGyOAWwv8VGt9PMJL3+n6/LbWuvs2qO5WY0y3XgZkYdTX6wCqMVp6/VFrXR7hWIQQMaCvarZFW7TGXVlZSUFmKnl5ucEPdvEuj7Fpf6O7zQ5A8eUXcbzBzoTZE9y1x1ZsrHJnl9otGw6MDN1pFo6Hj9XvYd67a/hj/kL2tTYzRyV4bE4ASElO9NnUe9P+RtqdmgTX9S/JGkPCBec8yogsufYKDry30Z1hLK1uYl3FMdqdmopjp9xTlr0lWBkS75ptknGLLUNmp6PW+m2M+nWhHj81hGPupCuIC+WapfR/ezEhhIgpc7JSeedUJw66MkX1O4znSqubWP3WARoazrA4K5ViV+2xOVmpbNhTzxkgQSmclkmWPfZTzFz/Elft3MrLd91DclIaC+sqSRkznMNe935qcw1ZBxupbjhDaXWTe7frgmkZPJGg0BhZt0WzbYw+H09egMLC5bUO2p2ujQ1Ozab9jf1e6NZas03ElmABXIFSKlhT9lAUROEaQgjRp2I1W+aP9fvxVTttoCrKSWf6rEx+XvUK33ctrH92hxG8PfzaHs41nOa8o4WHX9vD9FkXM9Z1zv3XwV8Ov8OErFS2HzLKgg5va+XWF/7IWUc9O375KBeO7ORCYHNrLfb2Doj3vHdLewfbDp6k5XQLj6zfy6zTZwEje3XD7AmcThjDtfNn+O0IkZHRNS05JyuV/0tQ6FHjUGcajClL6VQgIhQsgBsNXBWle0nLJyGEiAHWVlO9Oa1WWVnJvHnzKK1uorzWwZysVPzFy8X5Nkq9yoiU1zo8pkfbO7VHd4iinHQqstOYPctGxdFTJNRX89y2/+OVUWN57Iol3HQmnoUjO93nT7elsOdsPA6g49wpwNglay6RbnV2sr/hjPv43HEpZGdPJs/HJgxfa2iLctLdQd8Fznh3tlCISAQK4D7fZ6MQQgjhk78soDXoCaevaDBmaYmmygPct3pnr5eSMO/X6uzklXe2M33WxSHfb05WKq/vttPiqAeM3Z6+slpFOen8eNIu9qx8nbULPsvyzHzUqTqjfIml7VVBZirXzi3k51Wv0KRHkZubzoRZE/jfTV3Xsp9q7dHaNTPoq6lxRnS+1WDKDovw+Q3gtNah7KoUQggRBeFM11qDng176rn/OvxmrsLlq7REbwRw5i5S6/3MXZ+h3q8oJ53PTI9n/YmRnGg5yndcuz27ZbWee45ZK1ZwaMknueS/fsJNr77FWGeiz2lPM9P3fpudC20prN3b4DF91Kl1r70mfc3seRoo8ykGrqFQB04IMQiV7LaH3EvTn1gtFWMNelqdnSF1MghVpKUlIn0trfdLSojzuJ/5Mw7UaqsgM5VPzJ3EpVPTumci29vhrbdg9Wp46imYNs1d48y7zVRjY2O3XZv77M3utlKmhHjV7TWxljQprW7y6MgwceJEn+POzs72+z15XxOC7ygNdK63kt12HlxbwT2rdvByxTEeWb9Xep7GoCGzC1UIMXiYDbcdh4/xzqnen+YbaBZMy2BlQhytzk6SE+J89/OMUHG+jZsvzuSFDxP4wpXZvf66WktZjHUmuO9XWt3EY7sacBw+xoY99UyfdXF4Fz55Er7zHTh7Fp55BkaNgoqKsC4x3ZbCh81GcKnbWrhkyhj0uXSK823uwsKXJY1xB4O7jjpYU3WCZnsT5Wf2cv91vq/7ygd1VNQ5OJaxhxGnm/jw6ClGTmgm/O6noTOzu12v60mG2Yya8q1hZj4Dkaxe35EATggRc3w13B5KAZw16JmTlequOxYNJbvtrH3/KI4WJ09trmHO5NSovrY1NTUe2aeS3XbL99Huftxa+80MMBK9rlVa3cSGssNc4PQqirt3L3zrW7BoESQlGcFbGKoamvnw6CkSc84ZnbRdFs2yUVE/0v0LxLHDDva2HOL2IsjLMzJ2rc7h7jG/VmHnzP56xs5s4sRwI7B5//06qs8bodrjG6voqD/A8RNnsb9bwZUXpuMnYefBDJKWDM/0+7OxHmNlfV1NyV6ZT3+CTfFbf7HasKeecxdMYV9F4HGKyEkAJ4SIOQumZfCPsiOcJ7oV5GOpbIhZvwvwW3csEtFaAxdKmZJdRx38bF/XG/6S3K5VPWbtt/N0BRilH3ieu7L0fdqcnagzDcxMaDGe+Mc/4J57jOzbddfBn/8c0jgOnomjtNoItNZVHOPsibMcr6wnPt2I4DRG8JOA5y8Qba7WWQDJiZ41SMoPO2g+7ODAa3tIfr+D5qPHaKlvJiG1K9dm1qdzdmjsp3yVDPYUSvbZ+rq+dXIH4+sPMH6usfHC+romxcdRMGUMi2bZohJgWV+XVmcnK94+wLm6oZkl7wsSwAnRR2IlMIgFZsPtNa+elgryUTa+rY7khDjO0PvtlfbZm2lJ7nrD32c/637OrP225tXTzHEV6C193vPcNrPTglPT1Hoe/vhH2LABHnsMZgZv0QVQsvUDSo5omuubeWT9Xq46m+outtsJJBnts4lTRlBZUdX1C4QDY93edFsKAEdOtmBN2ZnBWXunJq6jE18SXNdPiFfYRicHHW8o2Wfr69rW0Uml/bR7R7H1dV2yeG5UCwlbf7GKB3dXi6GYJe8LsolBCBGTYqnhdixtlijKSefmizNJHe57DVw0v5fpthSGu7JWyZZAyGT+jIty0rs1s59uSyHJ3GzR2c7EynJOb9nKxmdfYYU9IeDGB6va4+fcgWCrs5Om5vMkJhhBVVJCHLfOnUTh5FSmjB1JUU46EydOdP8CUTg5lduLplCQafRe3W/pb5qglLs/amKccje2B+ONd8yIRO65KpdlM+LJz57ElRemezSz92fBtAz3a+YvwLa+rqa+aFZvvi43zZ7ArfMmBx2n6BnJwAkhBjyz5lksrqWJpeANjNe6N9fAWZl118wsm6+1fJWVlZwalsEj6zdz0tXMfvnSQgoyU7m9CD7csZfcN15nS9Iwvlj0eRJer6X56DFe323n4kmp6BNnPa5n/jzMXZ1ZaSOoPGtMKQJU2psZO3wYI1KGsbRoCkuumMq6k+PZWJ/mcR2z3Ei2awNDea0DZ6d2r9Obk5XKolk2nmvZzycXzWT72VReq9sPGJm92RNH8+3rZ/Ld8g1hvWa+ss/eGwfM13X5mgb2Or12FPdy54fifBtjW4w+tzcNz5QseS+SAE4IMaBZa57JWpreV17r8LsGrqfBaFlZGVUNzew4cZjpthQKMlPdb/hm+Q1zrVawMU0HCk4cZtpLf2HlxHwq4hMYTTzDOroa2G+vPUnrgSZKq5v8FjvOHZdC9flOymuNr52dmqbmVjrOtoX8fdXU1DBn5mW8khCHxsjc5aSPorzWQVbaCGPacngmGzdvpT11vHvatWS3nZWlhzh3NokDTU1cCSFtYjBfs3mu4M26ccDcrVucb2Psx2dzYngmfzz7PtcWxPV55wfrOEX0yRSqEGJA87WoXvSeOVmpEdWBC8QM/Eqrm1hXcYxNVU2sLD3ErqMO9+OPrN/Lpqom7lu9s1tNsm5jyk2HV1+l5vHHqbv9i+yePAMwpiqT4j3f1pwdOmCdvLq6OuZfmEZCvPJ4vKNTuzcnAIwZM8bdYuzxf7ze7RpFOencXjSFm2ZPYOG0DNa+f5SXK46xruIYpdVN7vIsE0YPY+G0DAoyU9m0v9E9fevs0Lx38CQv7DgSVk02740D3v8+ivNt3DpvsjuAHcxrcW+77bb+HkKf8hvAKaU+Go2PvvxmhBD+xdpUninSwrKDTV/9/Iwm8DOYMT4l6tnO8lqHe4OAdfemrwybqaamxj2mwsmpPPbveRT/+nvw9tvw5S+Ts+BSLh+nmDJ2JN9dNJPHl83lkilj3OvPQDFiWODJpoLMVK68MJ1Lssa4A8D4OOWxJq/+lNHMfudhh8/Ct5WVlRRkpnL3Vbm0tne4v592pxFAmuVZjp06z1v7G9l11MGCaRnudXwd505xrs3JYeco7lm1wx3cBmNdExdqORAxOAT6W72Rnjeg10HuIYQQAVlrng2mtTQDoWSJvzGYTeCj/VrPyUrllfQJtDk7PXZvzslKZcOeetoB1VTNgmmX+OxnemLSCK75ylKYPBkefBDq6gCYPHYkyR2jKcpJZ55r6vCZdw/yXNlhQPPPnUeYPj5wLbjJY0dy183GlOPPq9fhTE51F+itamhmX/1p4iZ1bXZYte0QB12FfL1Nt6VQfqaTVmcniQmKJYsXdmsZts/ezMP5Nm4vmsLzG5uoa4D4UcY6u7aOTo/sXyCea+IWUpxv49kdnseE08UhWvr77/ZQECi4qsV/AJdpOdcJHAfSLI+1A3XRGKAQQpg1zwbLWpporCULlbWg61iv5yorK/v0jbYoJ52F0zKoqHMwe2JXgGRk2GDDlrNMt/lZq1VRAU88YdR3+8EPyNuxg5o6/28z1Y1n6MQIito7Na9V2LnMzzverqMO3qs5zuzqJu751Dx3L1TomvY9e76D0a7jE+MUm6uO0+Aq5Ds36Sx5eXnuFlYFmalk5WRQXuugbdQEdyC8MiGOdjxLjxRkpmLPzeD5mn3u8STFd9+RG4isNRuaAjWzn+rrcaXU74B7gRXAcmCf1rpTKRUHTAfuA+4C/qm1/mbURyyEECIk3kVf7y2I69f2RqXVTbzlWvd1/GwjSS1NmMuWinLSGX1+su8TX3kFfvc7+PjH4Yc/DHiPsrIySqubqPXafYprRrWystIjI3X4xFl22A+568B5t+wyp33jRxjhW/qoRPImpbKrrSubZm/pXoDX/KVj3TojMDMzyUaQmuLRi9U2KYu0Ue+RfMEwRiSc4VfL5lK/40TIwXWsLo8QPRPWJgal1F24AjSt9Ve11nu01p0AWutO19dfBb4EfF0p9YXoD1kIIQaXaL4BW6/lXfQ1mk3vI1Fe63Av2m9zdlJ7/FzgE5xOeO01ePJJWLHC6Ffl4q9hu7khouFM1y7SxDjFolm+s1P2U60edeC8NwHMyUp114VLTojjytwMFs22udedJSXE+SzA62t8RTnpTLelsM/e7F7jZnSUOMSJjmQamtsYlzIMgOfLDkuDeRFQuLtQ7wbqtNZPBzrI9Xwd8JVIByaEEGBkkVZsrAq5MOtQUlrdxIMvVfCntw74fLP3LvpqNr03i+KGulDeH/Nn431va9Fda0A5JyvVvWg/KSGOrLQA7dtPn4Ybb4Rjx+DppyE31+PpXUcd7u/h8ImzfHj0lLteoLnWDIyM2XcXzfRbRiTRa9dqSnJXx9W8vDyKctK5YfYEZoxP4f7rZpA7LoWinHSPQr5mAd5ga81Kq43dt+Yu3JLddo+OElprttWc4J5VO/zuyBXCFG4AN4PQ17bVAaH1MhFCRM1gmk4xpwBfrjjmc+ffQBLodS+tbvIZ6PREaXUTD7+2h+2HTlJpP809q3Z0u761Mr7ZRsl8TXcedriDiEhYfzb3rd7pDrCt17c+DrhLbRSknOf2oinkjvOzzquyEu64w8i43XEHpKZ6PG3WT9t52MFfNx/krX2NHDpxlkfW72XEsAT3rmUzYwawYmMVVZZOCeb0ZHtHJ05Hvfvx5tZ2dzN7c+y541K45coCjyCwON/GF28oYsm1V4T8mnlnIDftb2S6LYU41VXCxHnmOG0dUjZHBBduANcOzFBKBWzY5np+hut4IQaswRTsDEbBalzFAnNKzzvQ6anyWgftnV37zNo6fL8+xfk2Fo5vdy+k927Evml/Y0T/Dta8+pbP6dlg07YFmalcnTfeYw2Yh5074aMfhc9/Hh59FOK6v01Z66d16K7XoNXZybnzTu6/bgY3zZ7A/dcZ9eHM139dxbFuWUfb6GSPKdKU5ETWVRxzB4TRzPx6ZyAXuOrBXTNznPuYONVVy24ol80RwYUbwG0GUoDHXZsWulFKKeAx4ALX8UIIERF/Na5iKfD2rnEWrXVoc7JSLbXOjJ2LobzZL5iW4Q4QEuJUt3PMbOFh700APu7va3rW37RtUFrDiy/CU0/B2rVw882BvwdXIGTNnsUrxZysVIpy0t09VGuPn/OoyWaW5ygrK6OyspLJY0d6TJE2t7a7a9W1Ojt7/PPKzs52/9nMQC7INT6bQfWNF01k4fQMCiensmBaOo8vm8uC3HTpOiICCjeA+zFG2ZA7gSql1LeVUjcqpS51ff42UAV8HiP79uOojlYMCv7WzQjhzToFeP91M8J6MxsoQZ53F4GQA5oginLS+cz0eC6ZMoY82wU8vmxu2G/27ZbgB7ralr1ccYzNB5o8/o16L8o314FZp2eh62dWODnV4/GAWlvh+9+H7duN+m5X+J6WNIOhYlf9tOz0kcRbgthb507qdr+stBHu1z8xwSjOW1lZyZoN7/Kntw/w5p56jjfYuTQ7jaKcdI81cEDQIsBWvnaNeq+LK8hM5dZ5k7tlICePHcnVeeO5fPY0d/cECd5EIGEV2dVab1dKfRJ4GpgK/NLHYQo4Bdyptd7e4xGKQcW7rIH8himCsTbHjkVmjbPyWgfTZ2dTXvG++/GeKshMZUleHpWV8d3+Hfkr0rtpf6N7jVVHp2bT/kZumdT1XKultZO1D6ov1vpjZWVH3fc0G70H672ZnZ0NtbXwxS9CTg784AcwbFjI3/uouZOoamimtOIsrfGtPov15o5L4bZbl7Dm1bdoGzWBUW0n2XUUnt5yEGen5ryjhTpHC7MnGrtWm1s9V/6cO+90/9lspTV2ZlOvlGPJzs6mpqYm+hcWg1LYvVC11i9hbE74MbANOA10uj5vcz0+U2v9YhTHKQYJ7/UxsbimSYhwFeWkMycrlac217g3ZPTXrlrrFGdigmJ8W53Hc2a2KiHemF7trUxmXl4e7N0Ll19OZWEhPPSQ3+DNOxAt2W3n+bLD7k0JDc2tNDaf9/u6FufbuPuqXPemiX32ZpydGqejno5zpwCoqDM2LSyYluGxJm5OVirZ2dlUNTQHbKUlRF+LqJm91rpBa/1jrfUVWusxWutE1+crXI/XB7+KGIq818fIAl3hbbBOsZfXOjw2ZPR0bZW/OmjBWKc4b5g9odvOSnMDwJUXpnfLvoV6T2sZEb9eeMHYpPCXv8CNN1K5Z4/fQ61BpJnF31RldEj44MgpnB3+16xZ16CZf55uS8Haut5sYVVe66A438YNsycwZexI7r9uhvv1sa6ls26oKa1u4r2a40ED8p4EwoH+TQyUpQKi70UUwAkRKe+yBjJ9Kqy8S1MMpiDOuujfzOz0l+J8G1fnjfdZxsPcAGDWNgtXoDIiAHR0wEsvwd/+Bg88AIsWAUZdt+fLDgetTWfN4pubDRLijXAsdfIM9+saKNgsyEzl2rzxHkFcnGsDBBjTrrMyR3sEt+amCVNKciK7jhrZuL2uLg7m31fzewknyzpx4kSfjw/mfxOiZyIO4JRSE5VSS5VS9yulHojmoMTgZk5nSPAmvHmXpujPKfZoZwKti/6tmZ1wxhBJtqWsrIyampqQz400s2cKVEZke3kVDf/vQQ4cPsr2H/+W5+uclOy2u7sRWAvc+uM9BXzRpNFceWE6U8aODH3TBMbOz5vnTCQjZRjZ6SNZMC3d49yMjAyPqdsRaRM8zm9ubWefvRmdnmN8bT/Epv2NHsV6ozFV3hfLTqTxfGwKO4BTSo1WSj0DHAL+BjwM/MjrmL8rpZxKqYt9XUMIIXzxLk3Rm1PsgYKj3sp6mL+8hBq8+SqUa+rJYndzitNa2NbKnBaM5Pv2V0bk3RfeYPI3vsoHyRn801bAQxtr3d0Gthw43q3ArS95eXnuLP6CXKNDAhjtsNJHJYX9S2HuuBQunZrGJ+ZO8sg43nDDDd2OnW5L8SgQvGBaBtNtKd1KsliL9UZjqlyWnQh/wu2FOhx4E/gs0AJsBHz9evE/rmv/Ww/HJ4QYQrxLU/RWljZYgNbXm218dWoIlMkyd0OG0wrLvMevX9/jnuJcV3GsW2BYsvUD97RgJIWHfZYReeMNLvr8p3hm7g28nLeAThRms6uW9g7QdCtwG+wet84zGt+vqzjG3vpmyo84AgacpdVN7o0PwVpeedt11ME+ezM3X5xJ4eTUgCVtrMV6/U2Vh3N/WXYysAykNbrhZuC+DhQCW4AZWutrgH0+jvsX0AZc07PhCSGGGu8p9t5YpB0sQOtJ1iPc8Vprr1mDyUB9TM3dkCtLD4UUYJk7KF+uOMaKtw94rCHzzhBZF+tbA8dQMn7m9Ku5xq545jijCf1vf8uuP/8vW6df2u2ceAXzc9N8FrgNpvb4Ofc6OLMkireamhoe/8frxnSmY7jPoDUQ65To2vePkpU2wp1B3Wdvdpdkcbruby3WG+pUeTCy7GRgGGjrEcMN4D6NUcj3s1rrY/4O0lq3AweArB6MTQghekWwAM2a9bi3IK5X3zittdeswaSvPqbex7c5O3mtwiipYb6Z+FrDZg3KOjo1rjX/JCaobhkis/Btx7lT7sDRnHL1l/EzsxIez58/D7feCu+9B888Q9Gnruf+62aQZ7vA3fszTinuXphLQfthtzgAACAASURBVGaq3wK3gWSljXCX/Ii3dJXwfg2s3TB8Ba1VDc1+d5KW1zo412S83bU6O6k9fs793HRbivvvkTrT4L6/+b0EC97McijhZFJF/xloZbDCDeCmAdVa69oQjj0FjA9/SEII0btCmZYKZ71aT1hrr3kHk74yLx612uIU7x9xuNeS+csIWLsRDE+M5+6FuT7LiICxLuz+62aQnz2J5UsLAdxTrr4yfqXVTe6shHsDwoED8OSTkJ4OTzwBaUaZjqKcdP5j4YUsmJbOTbMn8Pkrp/Lt62cC4U0rgjGt+cGRU4wdnkTu1CwWXnRhwGlNazcGa9Baf6rFPQ3rq77bnKxUd5CYOnkGWWkj3M8VZBrTxDPGp3DD7AmMbfFftNjX+M1yKCtLD/ldjxiJcF9LEZqBth4xrE4MgAY6Qjx2DHAu6FFCCBGhnuyes3YRCMZfV4NoMGuvldc6WLI4+Bon8/jnzu7DOSqFqnZjCtHMCEz3cU7uuBQum+l5j9Zt46mrq/NxtBFoVWSnUZxvc+0MNhb4tzk7ebb0oPsYMOvbdT1/9LmXOL/8++ydmkfJNx5ibGvXPUqrm9hQdhiAu6/KpbLSs+uBqWS3nTUbq1gyPNPv809vOUirq19r/Nk2Mm3K57HmWM1uGG2jjKC1stLInjSdaaO11QhKzfpuZmcK89wvLLqcffZmvrq0kNLnuxoM5eXlMS/fxgvZaUz0UZIlkBMJGe5sTpuzk9rj58M6X/Q98xe/Na+eDunfam8LNwNXA+Qopbr3K7FQSk0EpgP+KzMKIcQAMBAKoZq110J9QyjKSefqvPHMvzDNb/aup/cwWXcGAxw80eJRHsP9vNZ87OAOPvXo9/nyDd/i5eFZ3Pf3cvdx5tq9TVVNbD7Q5HcdmpmZMtcZeWemysrK2LS/EWendj/W0amxn2r1eT0zSDW/f+/ad+mjkkiIV8SPSnPvLvVmTolG8w3b+romJcR5ZPbEwDWQ1iOGG8C9DAzDq2yIDw+7Pg+YdlpKqQVKqReVUg1KqValVJVS6jdKqbERXOtBpZQO8nF3gPMTlFLfUErtUEqdUUo5lFKblVJ39Oy7FGJgGghBUjC+doL2lUhrrxVkGjsiF+SmR2WHornWzQyuysrK3DuD00d1NXm3lscoyknnD5/I55Hda7muaS9/+uUzvJU5G/DcBOHdZ9VfeY199maPdUbWNWemBdMySLA0sY+PU9hGJ4f9/VY1NNN0po2Z4y/wubvU59q+KCnKSecLV2YzYfQwFk7L8FlUuS9IDbjYFe4U6m+Au4BvKqVswJNAIoBSKgO4CPgmsBg4DDwRvaFGTin1FeAxjIC1DvgQyMMY6xKl1Ee01ociuHQDsN/Pcz43eSilhgGvAVdh9JD9EEgC5gPzlVLXAHdorbWv80XfsE7hDITftETvMrNDrc5O3jm1M6bKNRTlpDP6fM+zQ+YOu2OHHRxcv5dZp8+6nyvOt3FlbgYlRzQar/IYTU1c++C9lI0aTuVPHmLq/EKGH9mJA8/dswumZbAyIY52jM4Jc7JS4XzXInAziJ1uS2HP2XjOu873lZkqzrdxx/yprC81snPjUoZhb7BTsttOqL+R7zpqlFE5e+Is9vZEbs7Jck8Ll1Y38eaeet5YtYPmo8dcWc7Qg7i8vLygQfmJ4Zk8tXknjafOc/xsI0mTFJauX0IEFVYGTmt9HLgJaMSoBbcRMPeF24H1GMGbHbhZax29VZkRUkoVAn/A+F6/BkzSWs8DJgFvuD7/PcLLv6q1/oifD3/Zx19gBG+1wMVa64u01jOBhRgbP24H/GbvRO8baFvFY0ksZNp88bcTNBrMzF5/NK8P5+exaX8jjsN7ASPD5j0lmTsuhduLpnh2knjvPfjc52DxYnj4YUhO7lYHDmDFxioA7r9uBgUp57nywnS/m0PMjQHmBpPccSk+g6GCzFQ+MXcSF00azY7DJ0OqW2fWzyutbmKfvdldgsTZod2ZPjOYr3O0uEuEtDo72Wf3/XZWstvOezXH/W5C8Jfhsu5o1KmZPjON/aknmTnJ6vWNsDsxaK23AQXArzHWxCnLx1Hgd8AcrfX7URxnT/wQiAdWaa0fMzNbWusTwFKgGbhcKXVjbw/ElaW8x/XlXVrrCvM5rfXbwHdcXz6glIr3Pl/0jYG2VVwYevKmECyQCbQTNJLrmaw13nztcAyVOZUX7SCwsbHr77b1NUhOiMM2Orlb4FSQmcrC8e1G8LVuHdxwA3z720ZP07iutxOzDhzg/v7vWbWD1z60k5U2ImifVXOdUSi7Oq214MwpW1+7MEurm9z18x5Zv5fkxHj37tKEeEVW2gjmzZvnEcybkhPimG7rPsVp7sDdW98cdn05647G5CiugfNVaLY/lwdIMNd7IuqFqrVu1Fp/V2udC4zCyGKlaq2ztNbf0lo3RHWUEXJttljs+rLbdK7Wugl4zvXlp/tgSLdgTJce0FqX+Hj+GYyduzaMjJzoB321VTxWs1WDkbmzM9rV7q3BQGuAFlGBWMt09CQIDMZ8Dcy1YH6DrI4OePRRo0Dvm2/Cxz7m95oeNes6Otl+6CTrKo5x+MRZv+f4suuow28Am5QQh1LGJgRzytYMPM3gobKy0qMWXKuzk9b2Dm6YPYEpY0dy5YXp7jVoZiAbPyqNpPg4LskaY2QOfdSnM3bgehZFtgYs3sGL9WtrS7D7r5sR9ho4X4GRr/Zr/gpFi9gXbiutB5RSP3St4wJAa31Oa12ntT4d/eH1WCGQjNEVYqufY95yfb4igutfrJRapZR607VB4idKqVkBjjfv8bavJ7XWrcC2HoxHRIG0rhmaIt2lGYh3VivUXwaswb01SAglCAzUMSHYuixzd2tRTjoTJ070eK6qoZkXN++l+ecPw5498MwzUFAQ8HrW79/U7vS/Y9RX5qyqoZmVpYfcAWxpdZP79alqaGZX8zC01ijgC1dm+52atdaCMzNqueNSmJU52iNYtQayjy+by4M3zw54TbMXanxc96LIwZgtwaJVa9B79qA+aWKvLg8Q/SvcDNwDwFKtdawUrDFLIh1ydYfw5YDrc45SKtHPMf7MAT4DfAy4GfgBsEsp9Ts/U6DmeKoCXNMcz4wwxyKiaCBtFY+GgdS/r7cNpMymNbMXqH9mINZyE95BYE8a2geTbVlRX7Lbzpateyha+Rjvtyez9ZsPQmrwYMX8/i/JGuMOdBIT/O8YNfuVmn9PS6ub2FzV6LM5/K6jDo/nNP+fvTePj6q+9/+fn2yEJTJAAoFAgBCWIJElinFBrQJW2tp6UfRaqVb9tS79tra1tre9bdXW2t7r7WJde611K1qU1noVlaAVWQxIIAgmEEICCZEJDDAQIJDt8/vjzDk5czLLObMkk+TzfDzymGRyzpnPOTPJvOe9vF7QdDrYv3ldC64zw1iY46L6YBOfNhzrkhG8+7orNSuwAM9X9cGmhPxbKioqClg9CNYeoEqbvR+nU6gHAWe5755FH0g6EmIb/XdJwFnAYRvHdQP/BfwdLeBqQgvO7kIbQLgHaKWzpy2S9QwLtoEQ4hvANwByc5VbmSI0elnFW3+g101Y9gSxFu0tzgvcsF9ZWRn0ccy/K87LZMrZOSx/+zizcl0smJ5NWZl9xf9Y4H7571y7dRXvT7+EjbmFtH92gvMDKQZbKKlwU17n5fMzsvn22TNZ/nYTLUNGM6jNv+G/oKCA5as/Ynn1EZrcHna+spWFI0/xj637OX6ilRQXtHrqycgez6xcl+FPesr3O9ACw3mTs8DSN2fOJBbnZXLEl2Fcvnq3aQrVw4iRof8mCgoKKKlwa5O4J7S/pUz3JlpGaM9Te0dwaZTuIpjQrBOhaEXvwWkGbg0wXQgRugM1cdA/5rWE2Macyx9o56BSyqd8PYAbpZQeKeUZKeV2KeWdwA99m31XCDEhivUEXYuU8k9SynOllOdmZfWslYci8ekLQxmR6qR11+M4yXDqDeVOtMXMtl76Y8XSeikoUsJzz7Hk6Qd5c8ZlbMwtJC0lieFt4V9D1Qeb/Hr3gIBCujpV7ia/Ut8HuxppNYn1DhuUyp++u4TivEze2eE2Mm8AmUNSWTRjNAumZ4d8Du1MoYZi7e7OjF9zaztIjIyX2aIr1AeAYAMWZq9Z3SM1kixfoOpBuPaARMpaK+zjNID7hW+f3wshgvuWJA56MJQWYhtzLr85Bo/5P2hacyloZdVI1xOLtSgUCeff1x105xuSnuH8xwebushYWKdHzQ3lhm+oA8wDDSt3HIhrEFe7axc88QSsXs2Aso+Zc/E5zMvPZGnxeFuG83WHTznq3ZuSneFX6jsr3f/fZPbQgSyYnk1pjYdt+zuD35QkwUX5mhBuKHP4cFOoc2dOC3tO8yZnkWZa44X5I/y8UO30slmDu+0NXh7b3mF4za7bfcjwSFVDB4pQOC2hZgK/BB4AioQQLwGVhCir+uQxeoqjvtsRIbbRy5odQNSDGFLKdiHERuAaYHIU6zkaYhtFNxFPD8zuItH8+3qSeDyf1gxneZ32b8Rcun5r3WZ+cuN8GtPSOqcyTUHNirJ65g/IItzSzL6jrW3+WaOYnpvbDX/+M0yYAM88A+PGkT8yg4kTxwFa0LGmURO6DiacmztiEO5UTZDX6N3zlTetAxKgSZTk5mWxesNJ7rphNlWfJvGbd3bSjBakXThphHENzJm5yaMyyB+ZRPXBJpa9spVD1R427j3C0mL/4webQt102s2k3ExbQemC6dksLR7PkZQsllw1m+HNIyP2QtWpcjfRnN6pB7fHvY22HP+MeX/+m1UEx2kG7gO0LFwyMBNNC+5N4F9Bvt6P1UIjZJfvNjfEgMIk321NiEEHp+glUmuArK8nP8S++np2hdhGoXBEXxvKgMQp+1gznHoZzRzYtfia780N5WkpSWSkpxrZFjsSIeaBhtQU0UU7TLfCKqlwG6VEc9nQFvv2UXnTTTBjBtx5J6T7Dxxsb/AaU6GhhHPzR2YY09yRDHAU52Xyw89PY/Y4FzdfOMEIsMzTpKkpnYGdOePX0tbBkRT/THOwKdTMIWm4j522XdIuzHF1+VuqP3LS2TU2MSU7o3PAI0lwXmFBzDPmvf1DqCIwTjNwdWjDPr2FrcAZNP/WYmBtgG10vbWPYvi4M3y3+y33fwTcCswLtJMQIh2YG4f1KBSKOKFnOJ9oKueuG2Yb4rOjWj5joC8DleazntKnMt/51I035RgVnx3rUmYMFeiYBxrq2odRd/gUpTUehqIFao9tP8iBes0M/urBXo4N0ALDoz57rCX5nZ/Z9QnM0hoPxXmZWkD8+uuwbBncdRcMHQq+Thk9MJybNoxN23bS0jaAVDozjoGC6YkTJ7JgejbDm/0/r1YfbOJ4yjCG+9ats73B6zfE8K3CJM0mbJF/Kk2bJoXVG05yVlsyhTkuamuPGhm/E6brHWi/105Wcc7MHKr2VHHq8CG21nsRAzvYv2Ev8yvcjgPNkgo36/d4kAM72LtqF1POnhnVB6XcEYO49Uudr6dwxyqt8WgDCsr2r9/hKICTUk6I0zrigpTyhBDiHTQB3TuwBHBCiEzgWt+Py2PxmD5HB10LbpXl128AjwOThBALAoj5fg0YBDQSRCtOoVAkBnrQUlRUxILp2TQWjfObEDUHW8PbUvz6o3Y0HKPJfZzGPYeN7ItdnbgF07Op+tTF628epbVNGoHZkZRUo7za3NpOlbuJIyla2TDFNYom9z6qhmhrKKlwGxOYj6zaxQ8ub+P8fzwPGzfC7bfDrFngkyjResfWc7Tey67mfRRmJJGWkoTEP+Noh9IaDyt3HEAOaaf8hH9AqQ0xDDTWX153PGhPme7/WlvbZtyXPzKDmxZrgc+U7IyA+xbnZbJpxCDe2NZAk9tDu/cIHVKSDLR1SJZt3Oc4CFq7+xBt7dox7AThVqrcTbSkd9B+4jCtHZIqdxO/Mb2eQtGbPXwV0RORE0Mv40G0/rYbhRDf0ocvhBDDgVeADOBj4C3zTkKIdUKIvUKIeyz3ny2EeFoIMdNyf5IQ4t+BZb673pRSfmzexudQ8YTvx2eEEDNM+1+CJk0C8AspZRsKhaIL8bKVigd66drcX2XuxWpp7+Ci/BGGGr/dN9/yOq8xQSkz86hyN/mVVwemJjMlO8OvbJhmsoNau/uQsX/aiePk/Ow+aGiA558Hy2S733rbOmhp6zA8UR+9YbafAXy4yUnzuq3+otYhBj0w3N7gDTqYYEUXxg3Vz1Z3+JRxPl3KSRGM5s2bnEVKsrajHoQ//uq7tqdIp2RndJbFkwSeky2UVLgDTqtaUSK9/RunTgy1QohPzE4MiY6UcgvwHbS/1T8C+4UQZWjlzSvQJkav1z1STYwFxgPW/wSpaBps5UKIw0KILUKITYAHLXg7Cy3TtzTIkv4DLbuWC2zzXc+daBItQ33HeCLIvooY0p/EbfsKZqugeNpKxRNzUDUwNZkb5453rMY/K9dFaoogxTXKCNaK8zL9TOQLc1yGeK11enTe5CySkwQZZ07x2zd/h5h+Nvzud5DRtRHfGgTmjhjUpQ9Mn/AMNzmprxu6+osW5riMteqBoa73trbaE9HUrhm9Dyx3xCDjfMzxWkqS4Ma5442fzULGoVgwPZuLJmUaAsGArWuhU5ijPV9jXFr2sfLA8ZC9hWacevgq+hZOM3AjgdO9yIkBACnlY8BlwP+h9cPNAA4AvwfOkVI6kTPfi+a48BbgRRtImIU2uPA2WuD2OSllwI+LPrusK4DvA5+gDS2MwdcfJ6X8aoBgUhFjrJ6B3REIJErTfW/GPBhgR5qiO66508cwB1WRlryK8zJZNGO0cQw9MNNN5K0aYIGyUvmH9jG3fgcvFy2i/qbbIDmQeYy/g8HS4vEBddz0LF2btzFkJsi8bt0NoaTCzf3/t4On12gmNItNpcPyOq+hu9YS4vmeOHFiyIDL7FiRPzKDc8cPR+Cfgbt+wQURlx/HDR9sWJDZyYpZhwoWTM8mY0CKMV2rlZDtZRzj4eGr6B04DeD2opUcex1Syg+llFdLKTOllAOklJOklN+VUgZ0XpBSTpBSCinl/Zb7vVLKh6SUX/Qd4ywpZZqUMltKuUhK+ZKUsj3MWtqklL+VUs6WUg72HeNCKeVfYnjKihD0BXHbQPTVrKJeTjJPfDrxFk009KDK6RuuWaQ2f2SGcQxruU3XQ7NmcWpra6Gjg5QHH+CSPZvZPHY6700sChss6B6puv2U+dhFRUVdsop2n5ftDV7uXraFzfuOUuk+zm/e2elXKp2V6zJ019Ji8HyX1nj4+5b9bNjjQQIprlEAJA8Zwakz/l0rdkqYgYg0K5Y9NB3XuKnGfnZ7C+Ph4dsTqA+4znEawL0KTAlj2K5QJDx9Udy2J7KKscRO8KlPfEbjLRoNsXiTiaekQ1lZmeGAoEuTvPDRXq0fa+Mn0NIC117L7F2bWX3tjzmWPoT0ABObwdje4GXljgNdZE/MWbpQmSB9iEHff8Oew7S0dzoq6E38OsV5WulXLwFH83zrDf+1npN0mO5PHjQ0IiP6YOhZMacZ1nHDB/PoDbP9SsgKRSicyoj8GrgKeF0IsURKuTUOa1Io4k5fFLcNlFXsLedl9Wv9VmFSUFHbQNIUiUosswqlNR5Wl9Uzao47qHgu+PTQhnSWmZeX1dPq8ZDh3sPo6rXwb1/B9fLLXPK/b/LZ9ne5emaOrWCh+mATW44209omSaHrxKXuMxrqNacPMej7kwJpyUnoPTmpScKvLw60HjE7IrvhMJc2dQQwfHAal51jz0XBLvqUrNO/v0DTzApFMJwGcE+gmbfPATYLIT4ltBODlFLeFsX6FIq4oQcCRb0kyAnHvMlZvFq2nzP0vqxiMDeD7qKkws3yD6oZ3uYlwspZTAgW8JVUuHlo2Wpa2joMo/dPfLpsVsx6aDqTPXVcWv4OG8cXUvK9X8Keo5qUxpl2VpTVA/C1CyYEXdcLH+3ljW0HSB460rgvkhL2rFwXr6cIpG//C/NHMP9Lc3h0+SG8Kcf42Xevp3HLatvHszNocMKnXbdobirJFgfIaaPPYsbYMY5cFAKVVuuPnGTnsUaGT/MYHzwiLcEqFHZxGsDdgtb3qf8VzKBTtDYQElABnKILShk89vTmrKI1+IxVOcsOugCut/4AwttAbt7ksJZWoaisrIzo9V1ZWRn0Td9qor5iy35O+HTZ5ld0ZuRKazzUHT7FrV+cyAs1A/BIyRd3rmP21rd5ctrF1LuyWVut9a7p2ah24OVVH5F+8gC5eZN5v7KRa6fNBWqpra3l2IAsVpTVY56tGj10AHdE6K6waMZojqcMI3eSJqY7H7j/SzOorExmeHMDjTaPVVBQwPLVH1HlbmJKdoaRpdve4KXK3cT8AVm879nJX9bvpeWoF/f6Wi7IG8Hq3dr+aSlJXDhpBMf3RtdqEEjI15oh1T8gKLFdRSxxGsA9EJdVKBSKmNBbs4rW4FN3MwiFHihF4wFaVFTEUx/8zRDA1S2vwj1mNDgtqxYVFXFkoJtnUpJoaesgWcAZXzClT2Z+eWynlMfRei1YmT1yCDPfWk7+wb38+OIbaWw5TWqKMLJmz5oeo0NKNlQfZnl1h+HacEl6E/kjMyiv82KeyhJCsKhwTMQlx/yRGZxIy2D5tgZOtw00XBeGWrYLFdAChsRIS1sHG/ce4Sc3Tqakwm3cV3Z8JykjW+mQnZOdrqGpXD1rDGkTRnOisYUqdxOtR04SwJo1KPprTmft7kMkjy6gzdtolJW/PLZze2t7QLC+uED+sApFKJw6MagATqFQxAVz8Blt/49uLzQr1xU2mzYr18W6Yx1+llc61jfrWGGeJtWpra0NGrDoJupV7iYuvDSfPy7fY5i8j2r5DMj0E9wdctTD+SueZvrYHP5+88PM8R7Evb+O3BGDjOBhcdE4/neDVkxJS0kCAadbO83e6w6fIn+kJga8emcjzWhTb0UThkXdk1blbkJmFoK72iiZXzrK2TGsEiPldV4a0zozla0d0m9KTx9UOHIqg+G5Lh5at5mWtg7aD3gYMTL8B55ggfu8yVk8k5JEG6aysukDSG/uTVUkNk4zcAqFQhEXYhUo6Zmo020drN7ZGLCkZSaU5ZWd9dnNyoXbJlw/l97MP2pc4OBJD7TOOnGYf77wfV4+71yGP/wAtyclUVlZSW2Gv3zn1y6YwJZ/upg2YzTD27S3gv3VHTSjDRO0Dc5i3e5D5A7ycvXMHD456WLuzIkMaTka9lzDcVbbUcMn1iiZn3Em5TMr18VbvqykHnhPmZzFn5IEbR3+UpoCuOOSSRRnnmTlTv/gr61dUnf4VMTnsmB6NpdOzuLj0x6+MDOnywBCPHtT9dKsnQ8qir5HVFZaQiNTCJEbqwUpFApFNJgzUXbEfiGw5VWisnb3IUPwta1DGiXf4rxMfj3wM5ZUfMChB37FqFtuhCTtX7xZyNbMqKEDmZXrMqQ77l04lYmZWjm50n2cjbVHeHPHAd7Y1mA4MMSC/JEZhhyME8kMs42aWWLkJzfOpzgvkwXTs5k8yn8gYfTQAVw9aww/uHKacZ9ZXy7dlUnuiEERn0tJhZs1uw/hOdHKG9sausjgmKVvwsmK6Pp9gaR0rB8A+oIriSI6IgrghBBX+EziT6AZr9dYfv9DIcSzPr9RhUKh6DbMorJOJiXNZU0925aIvqtmoVij5NveDo8+Sv6KF0m7/VZm3nO7rWM1Hms2bJ9eLN1HVeMJmk63GAGijl5SDUVpjYf3KxvDBhJ6plEPmu2WE7c3eP0CltIaD4U5ri42ZBdOGmEEZwNTk1lUOKaLe4Q5+Fs0Y3RAdwm7mAdMgn1gsHOuZv0+O1ZaTl1JFH0PxwGcEOJnwCpgITAQLTtttQA+BtwMXB3tAhUKRf8lkgBKF5WNVuzXnOF4+O2dvPDR3i7bBOpli+Rx7Jq1Q6dQrG5tVZw1gKL774dPP4UXXoBs7Xy3N3hDXrsjA3OoajzuZ1S/oqwez4nWLtsmQ9AsVVlZmW+St4Ot9d64iUhXuZv8ApbyOm/AnsHCHO266BmvYFlDPfgDeL+yMeIgfd7krM5sXhRuEXWHT1mkdEK/HvqKK4kicpya2V8J3I+WefsOmtn7hgCb/h0tqPtylOtTKBT9lGhKRLq9UDTirGt3H8JbvwuAdikdBVlgL7jTz1HPgNkNIgxrK3EKbrkFJk2Cxx8Hl8s47oul+4xrV32wyW9//XGbznTOlyYJ4T9t6rtNFoLFReOMLFWgXr3yOq8ta7rSGg8ryuq7rEe/P9T5T8nO8AtYQknNFOa4uOMyTezZ/HjWMqTuLLG13stj2zsiCjz1ARPdzD7SDwy5Iwb5ucOEk9LpaVcSRc/jdIjh22jabrdIKf8BIIToYrwupTwohKgHpkS/RIVC0R8JVCKK9ZuUWZ/Lypmda0gWgnafDEU7+Nk8BTveCp9bgh3M56hPUtoOOquq4PXX4TvfgV/+EizSFi1tHaTiP1GqB1/mxwWtT2zGGBelR5NoRguQigtG4d5/hmuvmkZxXiYrV1YEXYo+yeuls1F/eHNXf1ZtuGQgbXsrKPHp1+n3N7k9lJ/YxZL8wHmFwhwX8+doUjOzcl1Br5OuA5frHciqgwc5VO1BnDhISYW7y+unyt1Ea1unzEikr7HCHBdDWkZF9YEhf2QGNy2ezRNN5dx1gz0pnd7kSqKIPU4DuLnAIT14C4MbmOp8SQpF30cJe4bHPL0XjxKRVZ/r6sFev3JbYY6LxYMHsqKsnnbfGqZkZ/hN/pm1y7Y3eHmoaiuHqj3sfEU7Xjgxfv0cT9BVwiQoUmql0tdfh3vuga98JeBxn0lJMhwPcocM0qywjmjBpf64+H6/aJqmQXZiUAb7PYO5feFUhp45RG3GGVtBiT7J+6vqt/ix0ajf+bqurKykKiPNKNe2tklDL81scXW6rYMqjo51HwAAIABJREFUt7+xjx6QTcnO4KabQgcsZh249GODSBk1ye/xrH9rU7IzWJciaG2TPeZeYi4DKysthROc9sBlAPttbpsMpDo8vkLR5+ntpvOREEmvWKASkR2pEbuPZdXnCpRd+9oFE/iPRQUUu5qNzNA3frfcKE2aS6rmHq1gxwt2jrpZe9hg6cwZWLoUVq2Cb34Tpgb+jKyX9fRrBxgm8t9+RbOwfvSG2YwfPtj4/Yul+9i87yj7vZFJaiyYns3lIbxQzcMXZkFh8/16kKyji/XqJeZAfyvmAMg8UNAuJcm+OrD58cwU5rhYNGM0s8e5HBnPOyEeOoIKBTgP4A4BE8JtJIRIQcu+qY8QCoWFQMKefQF94MD6JhtKGiEc+vReNKWpYOsyN4EPTE1mSnZGwKb44rxMFheNozBHk9vwzxZ1BmnmHi39eHaCyQXTs43jQ2c/WJfrdfiwViptb4dnnoGhVu8Cf/Q+sOK8TOoOnwpYKjw7ZyjFeZlUuZuMwEcOGUl5nZftDd4uzf3bG7yOewHN56kPlyyaMdoIlvT75+VrwydmS6yXSvf6ifWG+1uxDhTccWm+MWkaLDhbcP45IQNPRd+gLwbSTgO49cAwIUS44YSbgcHAvyJalULRh7EGDn1heixYVtHcpN8T2cbSGo/fuszBiFWfqzDHRWmNJ+Tk5pTsjKDZosIcl5FNCzX9GG69uqyH3/XasIGCX/4S5s6FZcsgPd3RcXNHDCI1RUtHBXrNTcnOMAKftJQkBg1I4cXSfWyt9xqSHdUHm/yyYU6mNvXAWB8uCSTrYZYD0Uuhe480G9uk2Sij65lHPRj8wZXT/AYwIsGOSHNBQQHVB5sCflCIxfEVikA47YF7FFgCPC2EaJJSvm/dQAixBPg9Ws/vH6NfokLRt+jNpvPBCGYXFAsbId0Wa8nAnJCOCoHQpiMHG49fXnccgJdeeombbrrJz77rh//nZXn1EcPBYUl+UpcetsIcF7l5WYZN11CLe4C5h+mlLQ4Xi78IsXG9SlfCj34E//mfMGIECKtqUye6FluxJYjIH5nBohlwPGUYd/lKhSUVbj5tOGboqS0thiMpWQxvS+HImTY/bbPyOi/HU4bR0qYFbeE8Y6PFPIQB2pDFosIxQXvDzJZnumNFQZCsbTjHi0gorfGwcscBUk5o/ZTfKkyKKmusSGwSJeh2lIGTUm4AfgGMBEqEEHuA6QBCiJVCiDrgZbTs24+llDtivF6Fok/gVMQ00QmWVYw026j/g3z81Xd5ZNWugBk0O8zKdRmPn5achPv46aAZklDlUTN2JUrM7gehTNmt601N0gI04a7i9tf+AL/+NaxZAxdfHHS/kgo39//fDn7zzk621nu5e9kWnv5wj9/1yh+ZwWJfcKlnRvcdOWn08ukl18Icl79TgW5TZcnS2Rq4iBBzKTQtJYlFhWN6zCUjWAneTHmd169EHc/gNlISJehQxA7HQr5Syp8DXwc+AyYCw9Akgz4PjEVzZrhFSvnfMVynQtHt6OW0/jBkEC3B7IIWTM/mW4VJRlnRacBqzUg5fWMsztMe99zxwwDYXHeUb7+yNWAPV6jyaDic9vnpvWTBth/U0szP3n2KQbV7YONGAo2z6pkkvUy8ed9Rw0Ghpb2DygPHjfKnFatEizVY1Z0KdG2z4rxMQyC3YPRZTMocYus8wT9wsBtEmIcwlhaPjyp403v5Ivk7tjtwNCvX5Veijmdwq1DoRGRmL6V8XgjxV6AYOAdwoYn77gDWSSlbYrdEhaL76dSs6mDdsa1xm1DrS5jLkWaK8zIZemZcRNdPN2g/3dYR8RvjgunZLM9IZ7unMxAMlF0LVx61UlrjYXVZPempyZRXbcXrHWjIh4TbT5e62PmK9trSKa/zkuNp4Pq1y6iaNIc/fu+3PDBsWMjjmcvEVvTyp9Ws2iojEihYDaZttufQCVraOnhk1S6mnD3TcVk7HHq2sjDHRUFBPpWVXZ0h7KL30p3yaA4R5mttB7stAMV52qBE2oTRLLnKnoabQhEtQTNwPi/TH1nuyxVCjASQUrZJKddJKZ+QUv5KSvmolPJ9Fbwp+gJmbaq+NCkaL2I54WU+ltkWK5Dpud3HNZdS9QnRQNgtj5qHDd7bedCRfEh5ndfoL7O+tj7n/pSHVj3B21Mv4JXir3DxtPBBr/ncUpMEozIGkJbsX/60omdMdRkRuxku87SqLq7cXZnqcGXooqKiLrZkZlmRSP6OnbQA5I/M6FNtEYrEJ1QJ9RZgkeW+vcCr8VqMQpEomLWp+sqkaG/tgdGDqkjeGHUZD72UGsofM5x3qJl3PnUbAX6HSW8sUHBoDXDM/WXGa6ujA/7xD6a+8DT1P3uYQZddYtseyXxuP/z8NG66YAKP3zjHmMQszsuk+mAT71c2+pWOzTIidjH3wbnGTSUjPdWvR9FqkRUIp5qABQUFxnNTUuEOOikcyJbM3EsXyd9xsNYAK731b0vRuwlVQm0DBgS4P/gIlELRR9C1qcrrvH1mUjRSysrK+sQblLnEa50Q1Utt0tXGW+s285Mb5xNMZa20xsO2+s5ASABXnj2a1hEnWXLVbBq3HDZ+t72hc7JVL8Xr/WVV7iZtIjR3CC/d/Ec4ehReeIEZmZkkV1YGnaI0H7vK3cT8AVncfV2R4VCwcmfnNGxBXqYxIdnaJtnVvI/5ASylrMdd01jN8DYv1k63JfMvADDWbs1U1x2OTAQ4FHrJWbraWHN0C6cP7Ka1Q7Lu2FlMOTvHKFcGsiW7+zqtl27TtmbDIaLMYvFVUFDgN3BiJVBrQF/4e1D0fkJl4A4A04UQY7prMQpFIhFN5keRmATL/phLbeEkMsrrvMawAGjm0P/adZBZua6AXpvmAGf522sALRBaXDSOBQOb4aKLYOBA+MlPINNeNuzIwBxDk+2RVbtsT0iGE8PVg6U3dxzgxdJ9ATNqhTkuY5rVmqnOHTHI1vrtoJdMzSXnlvYO49pbS6Lmcqd5SrYwxxVWqDeQtIgK0hSJTqgA7i00OZBdQoiPhBC65luhEOJ9m1/vxf8UFApFX8dJ2S1UKVTPtFgzLlbZilDDErNyXUbQohNsQtY82dplCKOiAi64AG69lYLf/hZS7TsPmgPO02GCslm5LgZljjbOLVQZ0S9Yauvwy6gVFBR0eR7M7gqP3jA7KsHcUOs3npvkJENmxVoS1cud11w2154tWYIRS306FXz2D0KVUH8MzAAuBs433T8UuMzm8WX4TRQKhSI2mEuhuhivHXTZCl3Itjgvk8rKwEGRNlgB7+xw8/GR/bR1yKATsubJVmM6UUp49VV48kn4+99h/nzwDWPoosXD27yBlEMMDLN6Vw7pelBmmXysra2loKDAKNlu2tbM3JnjA2aiamtrmThxIrNyXbyVkoQEkoQwAqdQFOdlUpyXSdH0bEpXhN3cMcV5mfzkxvm+aziHqk+T/VobzMK+ernTzuRqoIBUoehNBA3gpJRe4BIhxAygABgE/AWoAh7unuUpFP2XvtJ71l2UlZWxdn+aoeCv6ZudtL2/E9kKPWhZPqjJ6AcLJh1hDnDKSvfCb34Dn3wCP/85ZcOGoT/D5n454W0gN28y5qd/+eqPqHI3MWqO2y/g1Eu3gRwKzOc2pGUUE8NMmxbnZXLp5CzWHoJWKdlSf5QXPtrLqTNtzMp1Uefruws2xTtx4kRbr1sn2SbtWGXGNRze3GB835OUVLhZ/kE1Swbm+AXF+v2zcl30hj/fYOehSHzC6sD53BR2AAgh/gIclFI+H++FKRQ9jdmeR9E7MDJTBNc3i+Vzqts2WQOokgo371c2MjdtWGcmrbER7rgDXC74y1+grs7vWFq/3ECgax+enlk0a8fpAaeVaEtxp1vbafd939omWbFlP+1S8m6FmxZPPW0dko17j3QZhtAlPOYPyEqowGV7g5dNPnuxWAUo+sSrt/6AMZwCWgb1se0H8dYfYPXOxrjo5MWSQOehgrjeg1Mnhq8Dv4rHQhQKRWBUEGkfs4L/1TNzqHI3dbuTxvYGTTR2a72Xv6zfywsf7YWdOzUj+vPOg//+bxjcVXjX3C9n7cOLVs/MCeZ1CKBdap0wrR2Sto7AwxC6G4SdoYpg2LUbc4Ie+G6t9wZ1UigqKnL82IEEfkEXVe50uEh0/chg59Hd2LErU3TFqRfq81LKd+K1GEX/QgUm/ZNYPO96ic6qsVZS4abK3cSgASm8sa2BtdWeiDxUo6HK3WS8KXZICas30nrnXfDII3DHHZTuPcJTH1T7abKV1njYsOcwY4cN5NzcYV2a8KPVM9PtpAJZiFkpzNHss+blZzJ34nAjmEtNEqT4BgiswxCJGrjEK/ANJvBrFlVODzMwkghE6lUcS+zalSm6EpGVlo4QYhgwhBDacFLKumC/U/QtVM+WojuxaqwtHHmKVQcPcqjaQ/qxQUbmSJsQPR6Tx9N114JNOJZUuPGcbCFpIAg6+GLlWn5waB+P3f5jiq+7jtJX3zUs2vQ+tyMD3fzmnZ2cOnic1NahpCadYvqUZL/j6plFQzsugJadeQDCer9uJ/Xpib1kvruTH1w5DYCsLO0Nu/pgE1uOaOXPoXTan9XWtnH1tKmGvVhdTbLRA2cutc3KdbHuWAcnsBe4WDNe5naFWP4P0UvqbZgClBjYXOkTr8vfPm4MUzRumUhBXiZTzs5h+dvHbfUmxopIr1mg8+hu7NqVKbriOIATQuQBD6C5NITzX5GRPIZC0V30hYCzvwbO5p6x5tZ2NtV6aE7LAbSyXzLQSmzMxUs2fkLJfqn1pp3Yxb0LtSBHD+pGzXFT5et/OnTgOGMzTnHppx8wXiSz5PY/8NVizc+0vM5r6MLpfW6NaYf8dOVaO2RQr1a9367L+kzevcLbwHxTFsMsDdIhJU+tqWbWuM7jvPDRXt4o/4xkVztlx3fy71OS/SZg9SEMgCM7N7G4qGuPXbEvcHmiqZz5F9pzkOgO9MDXT8g3RgFVMO9f/f7eQrDz6C50X94z9B3Xm+7CUQlVCDEd2AzcCAxDy7yF+nLaY6dQKPoxTgJRq8ba3ImZfuWrxUXjmJefGdBD1Sl1h0/56a6V13mNzJZepn3nUzfNre1knTjCa89/H4a62PbVb/LAbZcZj2/WkMvIHs+sXBfzJmcZ2maglSqtwxdWj08rZkcEa3/arFwX6Vm5pLhGAdAuMX5/KiWD18rqDb2n1g7JhurDBCLcc7NgejaLi8ZFfK3N5XC9JyoWpe/CHBe3LSoOGFTGo+9Opz9+qIoEu3Zliq44zY49hJZ12wH8J7ARbSpV6b0pFIpuJZDG2hcHdmaBivMyqXQ1h826VFZWhm1kP9nS7vfzoAEpXUzpSYMF+7YyZ8PfeGHB7XRMTGfx3Al+j69ryOklSV0O45rZY1m59gDZWYO54bxchp7pDMD0HqFD1R7KTwwOaIU1b3IWL6Ykcbqtg4zs8cybnEVjk5Ypm3L2TCZuqmfnIa0kYs5y1B0+RYfVMNEXS+rTm8OneeIuimvOIK45qtWGmxoin+QM9Vzq5dpEnzLvTwFgT2cBeytOM2SXAGeAK6WUb0gpG1XwplD0fRL1jc5qdxZpFihUxqe0xsPuRv+Spq6LZgwWpCTx9Zp1PFbyGGVX38S8//pxUFcCfc36Gksq3LyxrYGjp1rZf7S5y/bmHiHzgIA5SDE7Ity7sLOEqU+HVntOkJwkKMg+yy/LkTtikF/2L0UILpw0wsgubq338siqXV2uS6wzV34ZxPYOWtrtuUzYIZ5ZNoWiJ3GagUsHdkopD8RjMfFECDEPuBe4ADgL2A/8E3hISnnEwXGE7xhXo7lUFPiO5wW2As8Dy4IFtkKIcAFvo5RSfQxRKLqJ7Q1eHqraamh3LclPoqCg09Tdffy0n6VMshBG9mxp8Xiq9x/mj9ue5UDVDgaUfcxlH34IYOjAmdEHDfT9wT940cuzl47q3EfvEQo3IGDuVdPRpkM1yZK2DknmkDS/7F3+yAzmTpvGM0cqGJJ1FjcsmMbQM4dYY8ounm4L7Q3rFP26LhmYY2TWzBnEtGQtKD5jPt8YDB7Em1gFiirgVNjFaQC3C23qtFchhLgTeAwt4/gZ8Cla4PU9YIkQ4mIp5T6bh7scWG36uQaoBSYCC3xf/y6EWCylPBPiOJvR/kdZCdyAolAo4kKVu4nm9M4MV5X7pJGBkq42UpMEyb4sVRKweM5YivMyKa3xUFe9nyvfeZkxF13AgT//GXJzjYDwQL2XXc37mF/hZjj+ZcLVOxuNQQg9eGlFC1hm5brAVELVe4QiGRDQp0PPoEl/BBI2Ls7L5MicsUycqE1RQibHBnh4yze9mW7SpItWJNhsdbbu2Fa+VZhEUVFnBlG3ywJY/nZTt05yKhS9DacB3J+AJ4QQ50spN8ZjQbFGCDEb+CPa/97/BzwupZRCiOHAcuAK4G9Asd1DogVsvwdekVIeND3WUuB/gS8ADwI/DHGc66SUe52djaK3ouxqumKenjVfHzv9TmbV/1DlUqtpfSCmZGew82SykfFJT03mpdK9hiVXa4ckd/ggJkzIZEp2BksumEBpjYd/vvguV6/6Ex/kzWbiT3/H8NOfAf4BoT5Q8OWxgTNtxXmZRvCyesPJzt49iw/rgunZNBaN8wVYnegZvWDXrdgkazG8LYXCMFZa5v306c1rF06NWQ+crsuWSld5F7PlGBDXSc5E6y/TX8+j5rgT2rlBkVg4FfJ9CngR+LsQ4t/is6SY81MgGa2s+Zhe2vSVTW8AmoDzhRBfsHm8TcBUKeWj5uDNd8wX0QI3gNuFEGoKV9HtQpWJ9uYUDuv1CTd5qG+vq/6bt9ffCJ1MLxbmuIwpuKtn5rBm9yH2HunsRUtPSeKcsUNZXDTOCIBOrXidH5b8L68WLmDVpLmsre58vCnZGcY0rFnwdt7kLGMCNT0lieFtnUFacV6m49690hrt/ENdt6KiIhZMz+aOy/KDBm/BXi/69GYsBxisgsSRyLv0ttd3OMyv5+4WnVb0bhxl4IQQ7/u+zQReFUIcBfYAwRyjpZTyiijWFxVCiCHAVb4fn7T+XkrpEUK8hmYRdj3wVrhjSinDKYK+jTatOxzIAhqdrFnR91BClaGxXp9worvm7Zvc+yiv0xrGdB9KbVpT02ozE2oQQ5+Ce+qDaiNDpHP1zBwyjx7Tfmhvh0ce4cI1H/KjL/w/Ks+c6QzSfH1ahTku5s+Zza+q32LuzPGGyb25TDgr1+U3aRoJZk05/brFMtgqKCigsrIyZseDTl22IylZxuRwXyHSqVb99ZyaOS5motOK/oHTDNFlvq9UtFLicOA80/2BvnqS2WiDFy1okieBWOO7vSBGjznQ9H3XkbJOfiqEeFsIUSKEeE4I8TUhhHWgX9EHSAS7mkTGen3CZWXM25s9Q612TpE03s/KdXWxlak5dILqg028sX4XTb98GPbuJX3ZS9xw/aXMy9dKjdaAfMH0bC4vGNUl62WdQNXRs0pOAiazplwsxIq7i8Icl9/kcH8hWObQ6etfodBx2gP39bisIn5M8d3uk1K2Btlmj+82TwiRGmI7u/y773ZbmGzdrZafbwYe8A0/bAm0g6J3kgh2NYmM9foEy8oYfXJX5RhN/VOyM/xEcs12TrNyXRzZ6WwtxXmZjB02EHORe3fVTtp2V3B3xXq2jJnEO/NvZ6GnRWv+39nGRJt9ZbHGrCnXXdms2traqAcZgk1ZxlqXzU6pNd7lWDu9r3Zf/wqFFUcBnJTy+XgtJE7o/aChZEL03yWhyYFEPAUqhCgC7vD9+Osgm72LJjXyMZqUyRBgPlrZNQ9YJYSYLaWsj3QdisQjFkKVfdkyy3x9Ak0cmqU+1h3byqM3zGZx0Ti/bax2TsV5max0GMABLDw7m5er2mlF00UbXV1B8bYSVs24nI/HTid1fxOfuLuWaK2Y/UXjJYRrbvy3M6nZFyUqAv1NJMLfid7bZn7NhgriQr3+FdHRV4fI+nqTfbrvtiXENqdN3w8MulUYhBCjgL+jBcX/kFK+Emg7KeXnpZQvSymrpZSnpZQe37bnA3XACODnYR7rG0KIzUKIzYcORddHo1BESrzFfc3iulXupi59hDrmsmM4Id/PPvss7OMW5rj44eenMS9/BN+uXcPiHe+zqngxH4+dbmwjM/NClmhLKtys3HEg4KBFrEiEIKU30FOZuEC9r4rup7uHyLqTvh7A6cFZWoht0k3fh+pZC4oQYija8EIuUAbc4vQYUkoP8LDvx2t8gsHBtv2TlPJcKeW5WVmqn6q30l/egEtrPKwoq3f0j9P8T/eRVbtIT0326xPKSE8N6Q0aLcWjB7N4/euct7+SP19zH/tcoxFoIr76GkL1Ki3buI/WNk3+N5J+vO0NXsMXNJ5EWw5VBKe7el/7y/+RSOnLgXTQEqoQ4mu+b49JKf9puc82UsoXIlxbLDjqux0RYhu9zNoBOB7/8U26voM2MPEpms1YpGNEG0xrGo4S9VU4JNHKrLp4bZPbw85XtDKSHZ0r8z9dmZnH6dPbefQWrU9oyoyJPLu+lkPVHjbuPcJSuwqOdjl0CB54ALKzyXj+99ze0MTqDVsMEVzzBGWgEu32Bi/rmzr/dFOThKPG9NIaD8urOzjdFr70Fmv0106k2dVAJdq+Wr4Kh+p9TQx0J5Mz9L0hslA9cM+heR/vQrOcMt/nhJ4M4Hb5bnNDDChM8t3WOB1gEEIMQpMeKQaqgPlSymiCLnOp1+mAiUKRcJjFa/VPv18eG34/6z/dKa4Mo0/on/tbjeCupa2DKndT6INZeOGjvby1oYb0d3fygyun+QcdZWUUPPQQ3HwzzJoFAwZQnDeAoWc6++0KCkL3KlW5m2hJ7zB+njnO5agHzioPEk/ZGTs9caGGC/TgbFaui0CfGwL1gUUiVJvoxvPBUCbtPU9fDqRDBQkfogVrdQHu6y1sRbOrGoAWZK0NsM2lvtuPnBxYCJGOFtheAuwFrpBSRlvvmOG7PY3Kvin6AGabKOPTr40pu66TeSP9j/nmvwBNRuSstqMBj1FUVMTKlSuN78vKyqg5mconZfU0n2jl8Q+qAbhcj61eew2efhp+8QsoLgYbkh6BSpC6s4PXt77Pn+3sDWNWrovVOxs53dYRVcYgFgMLgY6h32cOzlbvbGTK2TO7BGeByld2AnidRMomK3ovfTWQDhrASSkvs3NfIiOlPCGEeAf4Mtp0qF8AJ4TIBK71/bjc7nGFEKnACrTp0f3A5VLK/dGsVQiRAnzf9+P7Usq2aI7XG0m08h+oN5BYMCNnKF7O4me+UmBZWYOtbEqwyTyz9dSU7AyGtAQO4KyU1njYtv8oYnznfasr3VxenA533AHvvgvPPgvjxgU/iA2sQr5OJ1Ct8iB2MwbdnaUyB2enTZZhZgKWr0wBvPr7Uigip68PMYBmbdUB3CiE+JY+HODzQn0FyECT9PBzYRBCrBNC7BVC3GO5Pxn4K7AIcKMFb+ENF7V9fy2EuFkIkWG5fxzwGlqWsI1OOy6FokeIRSCgZ2g27zvKHs+JGKyqE916yq63p2Ek3+pfQLh6dCrceSd89hk891zUwZtOMCFfu+iCv5GUe7orKDI36aebLMPM6JnUL84Y3a29fApFf6DPB3A+UdzvoJV+/wjsF0KUoWXOrgA+A67XPVJNjAXGA9b/wEuA63zfnwb+4gv2An3Ntuw7Da2P8KgQokoIUSqEqAT2oWUJTwO3SCmDuUYoeoDe2HsTS5y4A5iDB3OGRjd1D4cuHfL4q+9GvRYz5l48gPRUwQNjW7j7h1+FOXPg9ddhyJCIjq2jn3uw0qVZFqW3o3us6sHZvQunhtQ464/OCwpFvOkXjfJSyseEEJ8A9wIXovWa7QfeAH7pcPDAbHc1wfcVjKGWn59Ey9qdB4zx7XsGbXp1NfCYlHIPCkUfQC+fnfDUk5E9Pmwvl+5lGqqnKpq1POPVSnfpKUksSvNy8/3fhMcfh0mTICnyz7J2es2s53bvwq7/HMz0lqZ9vcxtl3BDDwqFwj59PgOnI6X8UEp5tZQyU0o5QEo5SUr53WDBm5RygpRSSCnvt9z/nO9+O18fWPZ9V0p5h5SySEo5WkqZJqXMkFIW+taigjdFzOjpAEDP0MzLzwyZodGxepmGy9htb/Dy55WlVB8MPIVaUuHm/cpGSircLJiezaWTs8gcnMrP6taQXboWSkpgyZKIs3pOSpWBfFqtgV9tra1ODFvo2T6zjlygQFM/h2j14OxcC6u2Xzw17nr6tR8Nqi9QYZd+E8ApFIruwfzmGc4ZwcysXFfAnqpAwUhpjYcXS/extd7Lyh0HugQDpTUerf+ucg/ffmUr//3uTkor9nPRphKSyst59bxFlKR1X0nPem6hdOGifQOPl/K8HhBHWgIONPSgUCgiRwVwCoXCj1hkLyI5RnFeJt8qTPLrqdIDMT0Y0Z0Xyuu8tPh62lrbZJdgwJzxam5tp+LDzdy95q+cHDCQ/7zyTupOJ3H3si3dZqtTnJfp1y8WL29UiI/yvP48bK33Rpw9szP0oFAo7NMveuAUCkXvQDdn19ECscGAFozoor2zcl28lZJEG5CaIrrIU8zKdbHumBbgXVG3jUdX/o4bJ8xmjSubgcnav72Wdi0LNKUbzkvPqpn7xUprPJpUSIwdCsJJd0SC+XnQs2dO12zW9puV64rbUEOg/sFYaOIpFImGysApFIqExVx6HJiabNhZFedlsrR4PLPHuVg0Y3SXYKA4L5NHr5/FvFOf8cSqRxn8+goGXbUQs8FwWnLXLFBpjadbJkVLazSTe7tlzpIKNyvK6m2tKx7SHcHK207RJ1LjmYFUKPoLKgOnUCgSluK8TKacnWM4MjRu6Zw5KsxxMaRlVOAdz5xhwaM/4u+e/QzYsgXGj+cL9fWkNXs4MmIY+72D+a8b57BgejYvbdF20YNIx4ULAAAgAElEQVSq020drN7ZyJL8JOKVuAlkl5UaZFu9p+1QtYfyE7uYcvbMsEFZtMrz1iyW/jz8qvotrrUxkNJfUQMIiu4k4gycEKJYCPEjIcRjQog/W343WgiRG/3yFApFT2In8xNoyCCW2NER81vD0aPwjW9AczNccw2M77ReyB+Zwf1fmsG/zRnb5XjldV5kZh6glQmdeqw6YVaui/QU7d9vOLusRGn+18WJY5k9c9IrqYIjhcIfxwGcECJXCPEhsB54CLgLuMWy2YNArRDigqhXqFD0Q+IdFNldw7df2craak/QxvVYTjxGKudRfbDJWMMz//MyLf/5M7jkEvjb3yA1WF6rK9YyoV6ujQd3X3cl9y6caqvMaW7+d42bmtDN/3oJuidftwpFf8FRACeEGAZ8AFwMNKC5CgTyAH0eEMBXolueQtH/sAZFPaXcb838LH97TchtYjXx6JTjKcNobm3niupNPPHy/Xx86RfhtttAiPA7m7BOikZqg2UmVPO81S4rmBabWU/PGuwlUlZKtyuLtXxJILr7vBPpOisUOk4zcPehuQf8E5gmpbwNzQbKygagGbgkqtUpFP0Qa1BUXuftkXVYZR8CaZeZtwlXCowXU7MGcf+//sz15SXc8rXfcNZF59veVy8R68FGJE321jf3eLzZ63p6PdV7Zuec1u4+ZJSgeyqYD4YK+BIfdc2c4zSA+wrQAtwupTwVbCMpZQdQjeYnqlAoHGANikKJvsYTq5NCoKBG36bY1dwzZuXNzcx4+Rm+lHSEd+/7Nd/+9lcozHE5cgZYW+2xnemMtxxFPI8f7prYfexgx0mEYF6h6E84DeDGA7tteoeeBNRfsELhEKsMRE9KLiyYns2U7AzK67xBAxwn2aGYGro3NMBTT0FODiM+fI+vL5rVZQ1jxowJurue6WzzNobMdJbWeFhRVm+ICPcXnGZE4iFfolAoguNURqQFfzP3UIwE4jfGpVD0YcwyEGVl0YmwRkNJhZsXS/chXW2GwXykb8x6xsts6G4nRgg4qfjhh/DLX8L8+bB4MaQ4V0TSBW+9BM906tIiTW4P5ScGk5vn6ZMaZrHK/EUrX6JQKOzjNANXBYwXQoTMrAkh8oE84JNIF6ZQ9GZD6r7C2t2HDMuqJve+qPqarEMREfX2SQm/+pX29f3vw5w5Ea9HzxjNHufi0Rtmc/d1V3bZpjFtjKHX5nTNixYtcrSeWARRwbJmvbG/qDeuWaHoTpwGcP8AUoH/CbaBECIFeAyQwKuRL02hUNglXsHuvMlZpPn0ytKiUOCvrKy0NRQRkpMn4bHH4PXX4YUXID8/5Oa6+XqoaUhd2yxYVnHe5CxDry2iNScYKihSBEK9LnonTgO4R4Ea4KtCiA+EEEuBDAAhxHlCiNuAj4GFwKfAs7FcrKJvoDJrvYcF07NZWjyeL84YzdLi8VH1NZl7pK6emUN5nde+1MSBA3DxxZqu24cfwsiRITfXy7Vb671RSVosmJ7NvQunhhzkCEUsddH6g5+nCiQUCvs4ahyRUp4UQnweeANNImSe6delvlsBVAJfklK2xGSVioSnpMLN8g+qbRtzO91e0XMU5rgoKMinsrI16mMtmJ5N1acuHlq2GunKYd2xreEb3rdsgf/4D+1r5EhITw/5GAUFBfzTIsWybOM+vJ/Uc1ab87bc4rxMhp4ZR0GA4C1UUGW25tLPc3iIx4mXuX00qIBKoUhcHDsxSCmrgdnA3cC/gMNAO3AMzZ3hO0CRlDKQPpyiD+JUjb+0xhMz9X5F76O8zmv01YXVC3vtNbjvPrj/fq3nzaY4r7lcm5acxPo9h1lb7WHljgO2J2ADBWdOAppAfqc61mnc7hTBtWJ3MthOSdoufSUw7CvnoeidROSFKqU8I6V8Uko5X0o5UkqZJqUcLqW8REr5Rynl6VgvVJG4OFXjL6/z9rh6v6LnmJXrMvrqrHphhp1WayvceSe88go8+yxcENiVr/pgk58Qr455QOGi/BG0tGuBVGubtDWIoL8x68FNJBIiwfxOzR9gdIuytbsPBQ324on5w1cwuzR9u7uXbWFrvZe7l21RH7oUigQgYjN7hULHqYCn2XNSCX52pa/3CBbnZRp9dQHLp0ePwl13wf798NxzkJsb8DjbG7ys3HEgqBCvPqBw49zxxustNUXYHkQwB1ovlu5zrF1XnJcZ0O+0MW1MF3N687BEd/5NWCeDgwWOyzbuM4LglvYOlm1UBRaFoqdRAZwiapwKeJo9J5XgZ+9Ez0xFmokpzHFxx2X5DG+2aNzt2wdLl8LMmdq06ZAhQY+xofowrW0SCG05ZnaUWDRjtO1BBHOmuCWAhIida2D1O4WuFmXzJmcZwxLd/TcRaC0BsVaundnMKhSKOOBoiEEI4WSqtB04DtQCa6WU25w8lqJ34VTAUwl+BqY3ZN/Mgrx6c34sKGhooPLXv4Yf/xgWLoTkZIqKirpck5IKN4++sYOqxs6BhLTk0BIfC6Zn01g0jtraNtvrWXLVpax7ZStn0CRUllx1KaCVVwNdA7tBlx5QLn/7OLNyXcZ+xXmZFOdlduvfhHktS666NOg53Dh3POurNQOetOQkbpw7vtvWqFAoAuNUvvwW36003Wf+LBbofgkghFgP3OobglAoFL2UQD2PU6I5YEcH/Pzn8Mwz8MMfasFbEI4NyOKhV7birT9Ku+z8d3NR/giK87R/Z7FqLDcHN8PbUvyCm0DXwEnWTP8AEwuiPV87H6YWTM/m8Rvn8Kualfz4xjkqa+4ANeigiBdOA7gHgGHAnUAysA4oR7PMygBmAhejZd+eBFqBAmCB7/73hBBzbHqpKhQKB4STZikrK4vqzaS2thbotKA6Q2e/VuOWCA964gRccw0cOQLPPw8HD4bcXCtrDva7L0kIpo8eima/HJztDV42VTYyfJr9XrbOQMs/2Ap0Dfo6C6ZnUxpC9FihSGT6YiAdiZDvF4CdwNlSysuklPdIKX/qu/0ccDaa5dYXgYellF8CJqPZao0Fvhu75SsUsScRy5jh+q26U5olZqbl+/fD178O2dnw3nswYkTYXcwDMMlCIIAOKXl2fW3IIQPd03VrvZdHVu2i+mB0Ns12rkFRUVGffNNQKBSJgdMA7n5gAvAVKeWuQBtIKauAL/u2e9B3Xz1wE1pZ9UuRLVWhiD2JGKxZsROcdbc0y4Lp2X7N+Xp2zjY7dmjB25Il8NRTkJZm/CqUe4F5AGbKqAyjZyPUEAP4e7qebuug7vApZ+sNgPUaxBoV/CkUilA4DeC+DOyQUob8b+37/afA1ab7dqDZcE10ukiFoj9jJziLtzSLoc8WLVLCsmXw5JPw8MNw3XV+4rzbG7xhBW31wOnC/BGkpmj7DkxNDjnEYPZ0TU9JInfEoNicj0KhUPQQTnvgRqE5LthBAFbDwsNAjsPHVCj6NbNyXaw71hGy36o4L5MpZ+f4pgm1qdCnYmBVVlLhZkVZPVOyMwgu6GGTM2fgjjtg3TrNWeHcc7tsUuVu4nTbQKAzWP3y2M616D1+uiXVWNcg4AT/dcPsrpIkFiZlDYHMydywcCpHdtr9N6ZQKBSJidMArhE4WwgxWUq5O9hGQogpaL1w+y2/GocWxCkUfihv1MCUVLgpr/Ny60UzqdqhBWcLpmcHHEjQG+6PQMQSF9bH/vYrWzlU7WHj3iMsGBte/Et/HmfluvBbntcLl10GOTl8/NNHeHdrBfNrPF002aZkZ1B+ooPTbR2dwWpzA6U1Hh7bftA4p4UjT/Fq6T5OeU4aWbiw53HguJGFmzhRFQIUCkXvxmkJ9R++fd4QQswMtIEQ4hzgdbQM3ArT/dlANhA08FP0T5x6qfZWnPY0mXvfnl1f66cZFgqn1maBHvepD6pZtnGfn5BtuL6xoLZMe/ZQ8OCDsGgRJfc/ym/W7mNttWb0bh08KMxxBRS0tZaRN9V6jJ621jYZ8hzN1yOQIG+siKaf0q4fqUKhUOhEMsSwG5gKbBFClAkhnhVC/MF3uxnYCkxDm0R90LTv/+e7fTfKNSv6GNEGHH0Va9BiN/Bwam1mxmyqvn7PYdKStX8RaZa+sUDBSkBbphdfpOD3v9f03X76U9buOWx4fp4OEkwFci+w9vjNnZhpZNNSU0TAc9TN1zPSUzuN7VM0wd9ARvXdQaAg3q4fqUKhUJhxVEKVUh4TQlyMpvF2DTDb9yXxF+5dAdwlpTQ3mvwBeALNnUGhMOiPmlp2sPa+2fXw9FfXd1Y+NZuqt7R38LmpWXhPZmo9cC1HQ+5rfh4HJcGtrz8BH67Spkzz841tXkxJohVtmMDuOVl7/IY3N5B+8gCbtjWTO2KQr6zc2QOnB0UH6r2419dy60UT2dCknYddK63uIlDg29faCNREbXjUNVI4xWkPHFLKQ8C1Qog84EpgCjAEOIGWdXtXSlkTYD8VuCkCEk3A0ZcJFLTYJVKrMj3A0nvQbpw7nsaUcQDU1oYO4PTn8fW/f8aM5x9hwvgc2LQJ9u712+behVNZveEk8y+c6hdMldZ4WF1Wz11z5gc8V/M5lZU1UJjjYkjLqIBrsWZ1m063srhonKNr0V2YA9+QfqQO6a8BgeqnVfQXHAdwOr4g7ckYrkXRj1HeqIGxBi1Wgg4NRPF49y6cSnmd1wimX3LgsrBAHGX4nx6gsqAA3nqLkipPl/UV52Uy9Mw4CkzBm166bXJ72PnKVr5VmBTV+ehBkRd/t4iCgoKA/qo9STBvVIVzovGoVSh6G0574GwhhBgghPh3IYTqd1Mo4kS8eqcC9aDZEup980249FJNoHfpUkqqPLbXZy7d2u33C9XHpgdFs8e5Ar6JJ1p2Ste2S7TyrpVYTO/G89qrflpFfyKmAZwQolgI8TTgBl4C5sfy+ApFotITGZ2AQwM9QNnmzfDhh/DNb8I//0npjAtZUVbPsk37bK9v3uQs0n1DCYH6/SJ5018wPZvLlXdn3EmkYDiaAR6ForcRdQAnhBgjhPiREKISWA/cDgwFTgEvRHv8WCGEmCeE+KcQ4qAQ4rQQoloI8T9CiOHh947tMYUQKUKIe4QQW4QQJ4QQXiHEeiHEzZGuRdH/ML9ZxbJ3yhGnTsFPfgKffgobN1IydCKPrNrF2moP66s7p1jtrG9GzlAKss/i0Rtm285E9XY9t0QKfvoCMfPpVSh6AREFcEKINCHE9UKIt4F9wENo0iIAHwJfB7KllF+PzTKjQwhxJ/ABmrVXK5rN1xjge8A2IcT47jqmEGIAUAL8DpiJZi/mBi4EnhNCvCCECK+Y2odIpH4kK+FM5HsS85vVvQunxuTNKpgeWfXBJt6vbPS/Dm43ZXPmQFIS3HYbjB3bZYr1okkjwq5PLwVv3neUPZ4TUZ+DmXABni41kojPryIy4u1Rq1AkCo4COCHEXCHEk2gBxzK0KdRkoNq3SaOU8nNSyuellCdju9TIEELMBv6Idq7/DxgrpSwCxgLv+W7/1o3HfBi4DKgDZkopz5FSTgMuRbMpWwrc4WQ9ivgQK4HheAaoseydCtZTt73By8odB9ha7+28DuXlcMst8PnPwy9+YZjRW0uhN54/Puz6rEK7dkvB0Wav9PP1Oy+FQqHoJYQN4IQQo4UQ9wkhPgU+Ar4BuAAv8DRwkZRyaqhj9DA/RQsyl0kpH5NSSgAp5RHgBqAJOF8I8YV4H1MIkQXc7fvxdinlDv13UsoPgft8P/5MCJHs8DwVMSZUQ3QiZw0jJVhPXZW7idY2CWjX4dTjT8K998LPfgZLl/qZ0etTrPPyM22XsMyl4LRuLAWrhneFQtGbCRnACSHeQssUPQwUAG3Am8B1wGgp5Z1Syo/ivsoIEUIMAa7y/dhF8kRK6QFe8/14fTcc88tAGrBHSlkS4PAvoPUOZqNl5BRxJFwQ1t8aooP11E3JziA1RZAkO3ho9dPMf/uv8MwzcOGFAY9TnJfJ4qJxtktY5lLw0uLx3Vb66m3Pb1/rl0vk9gSFojcQLgN3FVqmyQvcA4yRUn5ZSrlCStkS99VFz2wgHWgBNgbZZo3v9oJuOKb+84eBdpJSngY2OVyPIk70t4boYD11hTkuvjhpKDfvXsPCQc0M3loGEybE/LHvuCyfwhx7zgyxesxQUiOK+NFf/I8VinhiV8jXBXwfGCmEeFFKuSuOa4olU3y3+6SUrUG22eO7zRNCpIbYLhbH1PetDrCPed/L6BwKUfQg/U1gWD9fP/btI2/FX2HGDLLefxeS+051f8H0bBoXFavgrZsJVL5Wz4FC4YxwGbhLgL+g2WSNA/4DqBBCbBJCfEsIkdiqk6DLeRwJsY3+uyTgrDgf08m+w2yspc/RF3vLejUrVsCvfw3z58PChQGDt8rKyh5YWFdiVWLsa6XKRKS3la8VikQkZAAnpVwnpbwNrSfrZjTZDIBz0czpG3w6aNfGdZWRk+67DVXuPW36fmCcj+lk36BrEUJ8QwixWQix+dAh1XitiAMdHdqQwne/C/fdB+ecE3TT7Q3ekFIcpTUe1evUj7ATAPe39gSFIh7YkhGRUjZLKV+UUl4BTATuR9MvSwW+RKdkxiAhxMXxWGiE6MFQWoht0k3fN8f5mE72DboWKeWfpJTnSinPzcrqm59cVYOz/UxQzAOkkye1oO399+Hjj2HixKB6aqU1Hl4s3RdUikP3OE3UXqeSCjcryuoTbl39AaXXplBEh2MhXyllnZTyQSnlZLRJyecBXfPtLGCNEGKPEOJnQoielkk/6rsdEWIbvazZARyP8zGd7Hs0xDZ9GtXgbB9rgGQV4HXMnj2al6nLBe+9B6NGAcGzbOV1XlpMHqZWKQ6rx+na3YcSpuSqv87WVnvU60yhUPQ6orLSklKu9bktZKO5L3zg+9VE4OfA7qhWFz36sEWuECI1yDaTfLc1NgYYoj2mvq+lSzzgvr1lUMQRdjJrSp/LPpGYwAflvfc0aZBrr9XssQYMALTgLViWbVauizSTcK+1l8kq7JtIvU7qdaZQKHozMTGzl1Ke8rkvXA7kAQ+iWWz1tCXUVuAMWsmyOMg2ut6aXT27aI6p/zwv0E5CiHRgrsP19BqsmbVg2SLV4GyfcCbwtpASHn0UvvpVeOUVWLLET5z3SEpW0CxbcV4mS4vHkz+4JWAvky7sa6fXqbTGE/NyZkFBQdDfOXmdqcEGhUKRaMQkgDMjpdwnpbxfSpkHXB7r4ztcywngHd+PXeypfFO0+gDG8m445htoAwyThBALAhz+a8AgoJEgWnG9GWvGI1i2SDU428caIDm21DpzBm6/XRPm/egj+NznumwSLstWmOPivIkjgj5PxXmZfr1O2xu8rCir533PYCMbq5eC9XKmHtzHsxdSf505cY2IBCfBnwoUFQqFXWIewJmRUq4Jv1XceRCtF+1Gn/SJABBCDAdeATKAj4G3zDsJIdYJIfYKIe6J1TGllAeBJ3w/PiOEmGF6vEuA//L9+AspZVsU55yQWDMeobJFPdHg3FslTKwBkm3cbi1gO3oUNmyAIIMKepYtlODtmDFjbD1kSYWb5dUdrK328PgH1UY2dtmmfV1KwXZ7IcMZ1usEysYtmJ7tyDVCoVAoEoW4BnCJgJRyC/AdQKIZ0O8XQpQB+4ErgM+A63U/UxNjgfFoIsaxOiZoWnofArnANiHEJ0KInWjuDUOBZXQGeX0Ka2YtFgbsigjZvBnOOw+uvBJeew2GDAm5eWGOi8sLRkUd6Jh79nSaW9tB0qUUrHrUFAqFIjh9PoADkFI+huZu8H/AAGAGcAD4PXCOlLK2u47ps8u6As3Z4hO0oYUxaD1vt0opvxok8OsTKOmA0HSLZtpf/wpXXQV/+AP8/OeQ1H3/Bsw9ezoDU5O58fzx3LtwqlHOLM7LTJheSFXW7CRUT6FCoehe7Fpp9XqklB/ioK9MSjkh1sc07dcG/Nb3pVAAnZIgp9s6WHdsa+z7strbNX23117TNN4KC6M6XFFRkeOys96zt3rDSS68LJ+qHSdZctVsn4VXA0PPaOXMsrIGI2O7/O3jxjZlZQ3GYysUCkV/pt8EcApFohNIMy1mAZzXC/fcA+npsGkTZPZc+bo4L5OhZ8Zx05XTKMs8GdJnti970aogVKFQREO/KKEqFL2BjPRUkn3yHTEtGe7cCeefD7m58O67PRq8xYPuLuv11mEXhULRt1ABnEKRAJRUuHl2fS3tUpIM3HrRxNhk31auhEsu0UqnP/gBpAbTno6ccF6ofRmVRVMoFD2FCuAUChM95cFqnrhsB5pO2zEFCYGU8Nxzmsbb66/DbbdFdbhgQVo4L9RIUYGRQqFQhEYFcIp+i3Xis7TG02MerOaJy/SUpIjLp0VFRXDqFGVXXQWrV8PGjZo9VhSECtLCeaEqFAqFIj44CuCEEO0OvlqFEIeFEJuFEL8VQkyL10koEgOnWZOezLJYTeBLKtyU13l7THfMrJF378KpkZdP6+th3jwQgqItW2DcuKjXFipIC+fSEG9uuummbn28YPRU5lahUPRfnGbghIOvZGAYMAdN9HabEOLO2CxboYiOQBOfs3JdPao7pmvkRSxwvH69Nqxw/fXwi1/AwIHGr6IJMAYNSCHJNFyRkZ5qHCuQS0M8A/Nojh2vddl1jOgLqNK2QpE4OA3gJgLfA1qBDcCtaAHaZN/t14F1aH6f30MTt70OeBdIBf4ohLggJitXKKLAagI/b3IWxXmZvdeD9fXX4ZprNE/T++7zM6OPJsAoqXCz6uAgOqREwP/f3p3H2zXd/x9/fZKIBDGEoAlJDDGLIOYp5lZbbVFj1FCKX7W0tNWa8vWllJhq+IqiQRExtKrmKYgISYgaSgWRCCGJhEQmST6/P9ba7s7JOefec+6Z7/v5eOzHvmcPa6+z9z7nfO7aa2DPjdfklhc++Cat0e9PL9koDZVQjgBEI0aISDUUGsCtQxiv8zp339Xdh7r7eHd/L85vdffdgevidl3d/T53PwC4Oh7vlyV9ByJFyBwEPgk+6mGkiKVK077+Gi69FG67DZ57Dg44YJntWxNgpPd14L1ps5dKa/ykWa1/Q3WuVCNGqHRLRApRaAB3DjAb+G0z250Vtzs7texcYAGwS4HHFCmLogeBr5Jtt912qdK08255ls933TPUe7v1VtgkezXT1gQYu/XpRsf24WuifTtjn03XXiqtfj2XGSq4zckc47de7icRqW+FBnDbAxPiUFA5ufvXwIS4fbJsDvA2sGahmRSRICkR6/35x9x902m81b0PXHUVdOmSc59SBhj9Yl23JK1s9fWeeGsq942bXBN1wSpVqlUPJbci0lgKDeCWJwy83hI94vZpBswv8JjSRugRUvN269ON77/3Ehc+dj3X7PkT5l10MbRv3+x+xQYYz787jYWLQ2OPxUv8m+G9cqWVlBA+P2F63vp2pRo9Qa0/S0OfPZH6U2gA9xbQ3cyOz7eRmR1HCPTezFi1PjC9wGOKCMCSJex7zxAue/4WnjzlD+x30a/LVuKT/KCnH78u18GaffxayQr9ban1p4hIpkIDuGsIpWhDzOxKM9vSLDR3s2ALM7sCGEKo83xNsqOZ7QisBLxcmqyLtCFz5sAhh8Bjj9HplbEcMnD/ijyuSx6/7rbhGhywxbeaPWY64LPp7xdcob+Qkjm1/hSRtqygAM7dbye0Jm1PaE06HlhgZl8QGii8RujzrQNwtbv/LbX7tsBDwG0lyLdI2/HRR7DTTrDaavDMM/Ctb1X08PtutjYHb7suG665dD275PHl61NmLbVtEvC1qkPiFihV609pPHokLG1Bh0J3cPdfmdlThJaoO8c0km/2JcALwKXu/lDGftcRuhcRkZRx48bl/sF5+mk4/ngYNAhOPXWp/t2qafT707n29c+YNfkTFk+fzNarNY3duu9ma/PptuuyabEdErdQEiwOf+RLDv2OWn+KSNtS1Fio7v6v2N/bKsDWwG5xvoq775EZvIk0kopUnHeHa6+FI46ACy+EX/yiZoI3YKlhx5YAL0+c2erzUUypiVp/ikhbVXAJXJq7f0V4bCrSJiQV52dN/oSRX7xann6/Fi6EE0+E0aPhxRdh5szSpl8C/XquytMzmkrd3JtaqIqISPkVVQKXiA0X1jGzTeO8dooIRMogW8X5fCVyBZfWzZgBJ58c5i++COuvX8rst8p66633zd87rr8GJ++xIe3jRz6zhWqpugkREZHsigrgzGxrMxsOzAI+BN6I81lmNszM+pUwjyIllwRWo98vrFebzIrzXTotl7Mri4K7uRg3Dn7yE9h+e7jvvryd89aC3+y/Cb//zibNtlBVhXIRkdIrOICLfcCNBg4mNF6w1NQFOBR4KfYFJ1Jz0oHV4MffKajuVuaoBrPnf52zK4uCurm46y749rfh178OJXDtWlU4XjE7rr9G1haqIiJSXgX9SpjZ1oQ+3pYDngMOBNYDOsf594ERcf2QuL1Ii1SqV/10YDV/0ZKC+w9LV5zP15VFi7q5WLwYrrkG/vAHeOop2HvvIt9VdYx+fzr3jZvMhM9mVzsrIiJtSqH/5p9J6APuT+6+Z2yN+qG7L4jzh9x9L+ASQgOJM0qdYWlMlexVPx1YderQrlX9h+UbZ7TZMUi/+AIOPBDeeAPGjIG+fYvORzU88dZUBj/+Ds9PmM7Db3yikRBERCqo0FaouwMzgXOb2e484CRgQBF5kjbkibemMnzEBBatvmiZx40/WKc8x0z3H9av56qtbjm572Zr03XehmybJZ2c6/773xC87bMPnHcerFHePtMK0dI6a8+/O435i8I4qV8vUitUEZFKKjSAWxN41d0X5dvI3ReZ2QRgq6JzJg0v3SVHlx5d6Ni+HQtIPW6cN6Vsx04Cq6p45BE45hi46KLQXci4cdXJRyvt1qcbt3dox9fECrCdlsu5rRoyiIiUVqGPUL8AerRw2x7AlwWmL21Iui7awsVL2GWD1apazTUAACAASURBVHM/bmwE7nDZZWFkhfvvD8FbHdt3s7U5cKsetDPDgVte+KDZx6jFtv4VEZGlFVoCNw7Yz8yOdPc7c21kZkcB3YFHW5M5aWy79enGPeM++qbU7cgdetF1XoesjyLLLXmUe2jnHnQtxwHmzQsB23/+Ay+9BD17luMoFTd3wSKWuANNj75zBd/pEtcn3/6UjTbfqjznukDqs656VDIrUrxCS+D+j/C05BYzu9jMlvqmNrO1zewS4GbA4/YiWTVbyb9CMhtQZJYOtfpH5tNPYffdQ4vT558vS/BWrR/Cfj1XpWOH8DXS3IDyrW39KyIiTQoK4Nz9n8D1QEfCYPZTzGy6mb1pZtOBKcBv4vrr3f3BUmdYGku+sSwrFZRk9tc2ftKs0iU+alSo73bwwXDnnbDCCqVLuwbsuP4aHL1jL7Zed9Vmg/BsrX9VAiMiUpyCewt191OBE4CJhNK4rsCmcW7AB8Dx7v6L0mVTGkmt/Whn9tfWr+eqpUn4llvghz+Ec86Bs86qqcHoS2nLHqvy0wN2bLYENV3ieuZ+G1c1aBcRqXdFDWbv7rcQHqNuDGxMGIFhNvCOu79TwvyJlF26W5FDv7M1XVvb+nXRIjjjjNDa9NlnYe7c0mS0ykpRV6yqrX9zUNAoIvWoqAAuEYM1BWxS99L9tY0b14oAbsYMOPRQWG650FhhtdXqtpsQERGpXfUx4KJIPXjjjTAQ/TbbwEMPheCtBFRCJCIimXKWwJnZ7qU4gLs/V4p0RGraiBFwySVwxRVw9NHVzk3NU1AqItI6+R6hjiB0BdIa3swxROrbkiVw4YVw3XWh1G377audIxERaQPyBVeTaH0AJ9K45syBY4+FKVPgttsUvImISMXkDODcvXcF81FWZtYLOBv4NrAWMAN4BrjY3d8oIr0+wEHAAKAv0A2YR2jQcT9wrbvPybHvCGCPZg6xqbu/XWi+pII+/jgMibXNNvDMM/Dmm9XOkYiItCEN/3jTzLYDngRWJozl+jrQCzgSONjMDnL3hwtIrz3w39SiqcBrwNrAdnE6wcz2dvcP8yT1LvBZjnWN0e9EoxoxIpS8nXsunHZayft3U/2wpel8iIgsq6FboZpZZ0KJ2MrA34Du7t6fME7rpcDywDAzW6uQZIEv4/6buvu33H07d18X2InQkfEGwN3NpPNHd981xzSpoDcqleEe6roddhj87//C6afXROe8tRDgtJXxRJ94ayo3jJjAE29NrXZWRKSNa+gADjgRWIcQVJ3g7nMB3P1r4CxgFKET4jMLSHMxsJ67/y7zMae7jwYGxpc7mFm/VuZfasXChXDSSXD99fDCC7DDDtXOUU4KMsojc8xcnV8RqaZGD+AOjfNb3H1BeoW7OzAkY7tmefB5nvWjCI9qIYxSIfXu009hr73gs89g9GjYsLZGEkirVJBRC6V+lZY5Zu7z705ban1bPCciUj0NG8DFumr948tcfdE9G+c9zaxHiY7bAVguvsxXl+0QM/uHmT1tZsPN7Ndm1q0UeZCWa/ZH95VXQuvSvfaC+++HLl2qm59mNBdkSPEyx8zdrY8+riJSPQ0bwAG9CXXcACbk2GYysDD+XarSsh8AKwCLgBfzbPfduO2ewI+By4EPzGxgnn2kAr4JooYNg/33h8GD4YILoF3tf1zKGWS09RKmZMzc723xLf58+Nbsu9na1c6SiLRhjdwKtWvq76yPPN19iZl9QegGpNXjHpnZKsDg+PJmd5+eZbPXgQeAJ4APCX3tbQecA+wF3GZmM939oTzH+RnwM4CePXu2NtuSafFiOOccuOsueOIJ6Fc/VRmTIGP4I19y6HeqF2SMfn864yfN4tDOPRoq0EmPmSsiUk2NHMB1Sv29MOdWMD/OO7fmYPGR7Z2Ekr+JwO+ybefuv8iy+Bkze5YQ2H0PuMrMHo719LKlcSNwI0D//v3V2XIpffEFHHVU6KR3zBjoVn+PycoRZBTSyvSJt6Yy+PF3mL9oCSO/eFWlVSIiZVCTz4TM7EIz8yKmialk5qf+7pjncEmgN68V+TVCQHUAMBP4vrt/kX+vpbn7EuA38eWGwJbF5keKNGkS7Lgj9OwZSt7qMHgrl0Ienz7/7jTmL1oCqB6eiEi51GoJ3FeE0RIKlX5UOjP19+rAlMyNzawdsEqW7Qv1Z+B4YDawfzGjOwC4+9tmNpPwOLcP8O9W5EkK8dhjYWSFSy4J3YVI0Xbr043bO7Rj/qIlquwvIlImNRnAufvFwMWtTOYDwqPTjoQSrWUCOGBdmkrn3inmIGY2GDiV0OL0AHcfU0w6Kcnj3pq8Ng3HHa64IjRUuOwyOOGEaueoRWq5QcG+m63NmfttHOrAVbEenohII2vYIMHdF5vZGGAXYHeaugxJS8Yknezu2QK8vMzsQuAMwuPaA919ZLH5jemtCawZX37UmrSkBRYsgGOOgddfD/27Tc/W5qRxVDLo+/mP92fcuHGq7C8iUiY1WQeuhO6N8+PMbPn0ilhvLXlWNrzQhM3sD8DZhBKzg939qdZkNPoNYaiumUBrS/IknylT4MQTQxA3ciT06lWVbGjUBBERKUajB3A3Ah8D6wE3mdkKAGa2HHAJsDMwh6auP75hZoPNbKKZDcuy7pfARYS+3g5394dbkhkz+4mZnZ3ZabCZrWhmFxBK8wAucvd8LWelNUaPDp3zDhgQ+npbccWqZKMRhmaq5Ue5IiKNrGEfoQK4+1wzO4jQ59pA4PtmNgHoBaxBKD070t2z/XKuEbebmF5oZt2Bq+LL2cAZZnYG2d3i7rekXncFLgQuNLOPgE8IozZsQlNr2Gvc/fKC3qi03NCh8Nvfws03Q/fuVR2MPtuoCaovJiIiLdHQARyAu79kZn0JHeXuT+ie43NgGPBHd3+9wCQ7Eh5zQmgtukuebZ/MeP04cCmwI6FUcIuY1lTgBeBGd8817Je0xqJFcOaZ8NBDMGIEbLYZjBtX1Szt1qcb94z7iAVoaCYRESlMwwdwAO4+ESioeaG7HwscmyOtoopt3P0tcnTwK2X0+edw2GFhKKyXX4bVWj3oRkkeHdbKqAn1To9xRaQtavQ6cNLWvflmqO/Wt28ofStB8FZK+262NicP2LCug7dCRmkQEZHSUAAnjeuBB0JDhfPOg8svhw5tosBZRETaAP2iSeNxh4sughtugH/9C3bYodo5EhERKSkFcNJYvvoKjjsOPvww1Hfr3j3nppWqO6U6WiIiUmp6hCqN48MPYZddoHNnePbZvMGblI4CVBGRylMAJzWl6GDguedgxx3D0FhDh0KnTs3uIqWjIE5EpLL0CFXq3w03wPnnw+23w377VSULCmBERKSSFMBJ/Vq4EH75y1D6NnIk9OlT7RyJiIhUhAI4qU+ffQaHHAKrrhrGNl155VYnqVI0ERGpF6oDJ/Xn1VdD57y77Qb/+EdJgjcREZF6ohI4qS933w2nngrXXhuGx5K6o5JOEZHWUwAn9WHJEjj3XLjjDnjiCejXr9o5EhERqRoFcFL7vvwSBg6EWbNC57xrrlntHImIiFSV6sBJbXv33dC/W/fu8OSTCt5ERERQACc1IGedqMcfh113DV2F3HADdOxY2YyJiIjUKD1CldrjDlddBZdeCsOHwx57VDtHIiIiNUUBnNSW+fPh5JNh/Hh48UXo3bvaORIREak5eoQqtePjj2HAAJg7F154QcGbiIhIDgrgpDa89FLonPd73wt9va24YrVzJCIiUrP0CFWq77bb4Iwz4Kab4Ac/qHZuREREap4COKmeRYvgd7+DBx6AESNg882rnSMREZG6oABOqmPmTDj88DDCwssvQ9eu1c6RiIhI3VAdOKm8t94K9d022wweeUTBm4iISIEUwEllPfhg6Nft7LPhyiuhgwqBRURECqVfT6kMd7j4YrjuuhDE7bhjtXMkZZZzhA0REWk1BXBSfl99BccfDx98EOq79ehR7Rw1PAVPIiKNTY9QpbwmTQrjmS6/PDz3nII3qXsKjkWkFiiAk/J5/nnYYQcYOBBuvRU6dap2jkRERBqCHqFKeQwZAueeC7ffDvvvX+3ciIiINBQFcFJaX38Np50GzzwDI0fCRhtVO0ciIiINRwGclM60aXDIIdClC4weDausUu0ciYiINCTVgZPSeO210DnvLruEobEUvImIiJSNSuCk9e69F045Ba65JgyPJSIiImWlAE6Kt2QJnH8+3HYbPPYYbLNNtXNUl9QthYiIFEoBnBTnyy/h6KNhxozQOe9aa1U7RyIiIm1Gm6gDZ2a9zOxGM5tkZgvM7GMzu8PMtigyvQFm5s1Mw5pJ40Aze8LMZpjZXDN708zON7POxb3LCnrvPdhppxC0Pf20gjcREZEKa/gSODPbDngSWBn4Angd6AUcCRxsZge5+8NFJr8AGJtj3dt58nQxcFZ8ORGYBGwODIp52t3dZxWZp/J68kk46qjw6PSUU8Cs2jkSERFpcxo6gIulWfcTgre/ASe5+1wzWw64EPgtMMzM+rj7p0UcYqq771pgng4kBG8LgSPd/b64fF3gX0BfYAhwWBH5KR93uPpquOQSuPtuGDCg2jkSERFpsxr9EeqJwDrAB8AJ7j4XwN2/JgRRo4AuwJkVzNOgOB+cBG8xT5MJQdsS4NBiH++WxYIFYTD6v/419O+m4E1ERKSqGj2AOzTOb3H3BekV7u6Ekq70dmVlZhsCW8eXN2Sud/e3gWcrmadmffJJCNhmz4YXXoDevaudIxERkTavYQM4M2sP9I8vn8uxWRIs9TSzHkUcZmUzG2JmT5rZw2Z2jZntm2f7neL8g1jili9PO+VYXzljxoTOeQ84AIYPh5VWqnaOREREhMauA9cbWD7+PSHHNpMJddE6AhsDUwo8xmrAzzKWnWpmTwGHu/v0jHXJwKC58gPwXpxvXGBeSutvf4Nf/Qr+8hf44Q+rmhURERFZWsOWwAFdU39/nm0Dd19CaJkKIRhrqXnAbcD+wLqEQLE3oS7dV8DewD9jKWC2PGXNT8a6nPkxs5+Z2VgzGztt2rQCst1C8+bB7beHAekVvImIiNScRi6B65T6e2Ge7ebHeYv7X3P3l4CXMhZ/CFxuZqOA5wmPQI8gtH7NzFOr8uPuNwI3AvTv399bmu8W69w5jKwgIiIiNakmS+DM7MIWdJSbbZqYSmZ+6u+OeQ6XBFXzSpF3d38RuDe+PChjdZKniuVHREREGk+tlsB9BcwoYr/0o8mZqb9XJ0v9NjNrB6ySZfvWGkXoEqRPxvLkGKvn2Td5zFrK/IiIiEgDqckAzt0vBi5uZTIf0NRAYUOyN1BYl6bSsHdaeby05BFp5vlNjrFhnn03KEN+REREpIHU5CPUUnD3xcCY+HL3HJvtEeeT3b3QFqj5JJ3wfpSx/MU47x1HXsiXpxdzrBcREZE2rmEDuCipi3acmS2fXmFmBpwUXw4v1QHNrDswML58PL3O3ScAr8WXJ2fZdxOaArh7SpUnERERaSyNHsDdCHwMrAfcZGYrAMSxUC8BdgbmAIMzdzSzwWY20cyGZVk33Mz2M7MOGcu3B54k1Kv7mKaRHtLOj/Mzzezg1L7rAncTrsm97v56oW9WRERE2oaarANXKnHg+oOAJwilYt83swlAL2ANmgaUn5pl9zXidhOzrNsP+DEwz8zeIzS66E6oUwfh0en33P3LLHl6wMwuA34D3GtmHxD6otscWA54k6aSQREREZFlNHoJXNJnW1/gZmA2sCWwCBgG9Hf3B4tI9izgLkJw1x3YFuhCaH16FrClu7+Wa2d3/y3wI+BpQoe9mxBGYLgA2N7d83X0KyIiIm2chTHdpV7179/fx44dW+1siIiISAmY2Th379/cdg1fAiciIiLSaBTAiYiIiNQZBXAiIiIidUYBnIiIiEidUQAnIiIiUmcUwImIiIjUGQVwIiIiInVGAZyIiIhInVFHvnXOzKYBH5Yp+TWA6WVKW+qH7gNJ6F4Q0H1Qbr3cvVtzGymAk5zMbGxLeoOWxqb7QBK6FwR0H9QKPUIVERERqTMK4ERERETqjAI4yefGamdAaoLuA0noXhDQfVATVAdOREREpM6oBE5ERESkziiAExEREakzCuAEM+tlZjea2SQzW2BmH5vZHWa2RZHpDTAzb2YaVur3IfmZ2W5m9oCZfWZm881sgpldbmZdaylNKb9SXjczG9SCz/vJ5XgfUjwzW9vMBprZ1Wb2gpnNjddqfAnS1vdCBagOXBtnZtsBTwIrA18AE4BehI4aFwAHufvDBaY5AHgm7j82x2ZPuvug4nIthTKzU4BrCf+0fQxMBTYFOgMfAbu6e0EdQpcjTSm/Ul83MxsEnA98BrybY7PL3P2BVmRbSszMTgeuzLLqNXfv14p09b1QKe6uqY1OhA/UZMCB24EV4vLlgD/F5V8CaxWY7oC478Rqv0dNDrA1sChek1Np+setKyF4d2B0tdPUVLf3wqC439Bqvz9NBV2344EngIuBg4Hfx+s4vhVp6nuhgpMeobZtJwLrAB8AJ7j7XAB3/xo4CxgFdAHOrFoOpRTOBdoDd7r7tR6/Ud39c+BwYDawg5l9t8ppSvnpugkA7n6Lu+/r7r939/uAT0qQrO6vClIA17YdGue3uPuC9Ir4wRuSsZ3UGTNbCfhOfPl/mevdfTpwb3x5WLXSlPLTdZNy0v1VeQrg2igzaw8kY9k9l2OzZ+O8p5n1KOIwK5vZEDN70sweNrNrzGzfItKR4m0NdAIWAi/l2Ca5zjtVMU0pv3Jft63M7E4zezpWYP9fM9u8mIxKXdL3QoV1qHYGpGp6A8vHvyfk2GYy4cPYEdgYmFLgMVYDfpax7FQzewo4PP5HJuW1UZx/GB+NZ/NenK9vZsvl2a6caUr5lfu69YtT4kDgbDO7GjjT3RcXll2pM/peqDCVwLVd6ebcn2fbwN2XEFqmQgjGWmoecBuwP7AuIVDsTahL9xWwN/DPWAoo5ZVc56zXOGNdO0Jr5GqkKeVXrus2FbgU2BHoRiiF6QvcABhwOqGivDQ2fS9UmErg2q5Oqb8X5tlufpx3bmnC7v4SyxahfwhcbmajgOcJRehHAH9rabpSlOQ6t+QaQ8uucznSlPIry3Vz9xuyLH4dOMXMPiC0aP+VmV3v7hNbkqbUJX0vVJhK4OqQmV3Ygo4zs00TU8mkP0gd8xwu+VDOK0Xe3f1FmiqyHlSKNCWv5Dq35BpDy65zOdKU8qvGdbuc0BdYB8IjVWlc+l6oMJXA1aevgBlF7Jcu2p6Z+nt1stRvM7N2wCpZtm+tUYRWSH1KmKZkl1y31fNskzz6WELo968aaUr5Vfy6uftiM3sJ+BH6vDc6fS9UmAK4OuTuF9P6OiUf0NRAYUOyN1BYl6b/pt5p5fHSkiJ23X/ll1y3nnkqDW8Q5++3sFJxOdKU8qvWddPnvW3Q90KF6RFqGxVbhI2JL3fPsdkecT7Z3QttgZpPMsbqRyVMU7J7lTCkWUdCJfNskuv8YhXTlPKr1nXT571t0PdChSmAa9uSumjHmdny6RVmZsBJ8eXwUh3QzLoDA+PLx0uVrmTn7nOAR+PLZQYUN7M1gEPiyxZd53KkKeVXjesWe9xP+oLT572B6Xuh8hTAtW03EioYrwfcZGYrAJjZcsAlwM7AHGBw5o5mNtjMJprZsCzrhpvZfmbWIWP59oTx8FaJxx2Sua+UxQWEOidHmtmpMTjHzLoCwwjDpY0BHkrvZGYj4zU+vVRpStWV9F4ws81jZ91bZSxvZ2ZHAHfGRf9y9zFI3dP3Qg2p9mCsmqo7ATsQKpM6MAsYC0yLrxcA38+x39C4zYgs62bFdXMJ3QmMBibFZU7oIHirar/3tjQRBpZeEs//FGBcvD7J6/Wy7DMxrh9UqjQ1VX8q5b1A6Lg3+VzPAF4BXiY0mEqWPwesWu33rWmZa7ouMD01zY7Xa1HG8t+25F5ozf2lqbhJJXBtnIc+2/oCNxM+wFsSPsDDgP7u/mARyZ4F3EX4oHcHtiX85zUqrtvS3V9rdealxdz9WmAA8CChY+UtCINXXwX0dfcPaiFNKb8SX7eJwDmEEpVZhAZR/QgNFx4Bjgb2dPdZJcq+lE57QovRZFopx/IVCklU3wuVYzFiFhEREZE6oRI4ERERkTqjAE5ERESkziiAExEREakzCuBERERE6owCOBEREZE6owBOREREpM4ogBMRERGpMwrgREREROqMAjiRAsVxAN3MBlQ7LyLVUOufATPrHfNX9p7qk+OkphHlPmY5mNn4zPdS7TxJfgrgRNoIM9st9eX8SLXzU6vM7FgzG2Rm/aqdF6krY4EXCOM/52RmfczsEjN7ycw+M7OFZvaFmf3HzO4wsyPNbMXKZHkprxLyP7YKx5YidKh2BkSkYo5N/b2fmfVw9ynVykwNOxbYgzDO5/iq5qR2vQfMJwxSXou+Bt6p8DF/7O4Tc600s+WAwcDPCeONQrjHJhLGG10X2AQ4EphmZke5+xNlzO9S3P24mM/egMYrrQMqgRNpA8xsBeDH8eUswmf/6OrlSOqZu+/t7pu4+8vVzks27j4l5m+TaucFwMzaAX8HfgksBM4F1nb39dx9e3ffAugK7ALcAawGbFWt/Ep9UAAn0jYcDHQBPgL+Jy47pnrZEWlTzgK+Syi13MvdL3T3T9MbuPtidx/l7gOBnYB3q5BPqSMK4ERKzMwONLNHzGxarN/ykZndaWbbNLPf0Wb2spl9ZWafm9mjZrZriSpkHxvnf4vT18AmZrZjM3la1cyuNLMPzWx+nP/ZzLrGemJuZkNz7LuSmf3ezMbEOj7zzOztWP+na6FvwMzWMbPTzOwxM3s/pvdFrEv0azNbPsd+nc3sd2Y21sxmx2vySXw92MzWi9sNiOd4j7jrXzMqdS/zPs3su2b2gJlNjelONbN7zWyHHHkZGtMaZGarmdkVZvaemS3IrPxuZt3M7LJYN2qumX0Zz+UZZtYpR/o94vV5J56feWY2ycyeMbM/mFnnLPt0NLP/Z2bPx/tuQTy/15nZOjmOk7URg5m1N7OTzWykmc0ys69jPa9/x/T65jkfq5rZVTHt+Wb2rpmdFUuvMLPlzeyceD7mmdmUuP0y9cVa8pkxs2+b2f0xnQVm9qmZvRjv2TVy7VcoM+sC/Ca+vMDdRze3j7uPdfcHUmkcFN/P+2ZmeY51Z9zuT6ll33xO42dhkJm9Fe+pia14a1Jt7q5Jk6YCJkKdFQcGZFl3bVznwBTgZWBmfL0I+GmONK/M2G9M3G8R8KtkXZH57QUsiWlsFpf9M74ekme/NQn1iDzu/zrwGrAYmABcHdcNzbLvhnEbJwSLE4A3CY+PnFDHpneB72Nw3HduTO/lmE7y3kYCy2fs0x54PnVu30/ttyAuGxi33Tqm8UVc/t/4Opn+kEq3HXBLKt3PgFcyrvXxWd7D0Lj+2piXxcAbhArkj6W22xL4JG67MK5/K3W8scDqGWn3BD5N7fOfuN3H8TgOrJPlGo+J6xYT7u3XgHlx2XRg25Z+BoC7UnmcHM/1u6n0zslxPq4C3o75fgX4MJXOdUCneA2S8/Xf1HV/NEv+eif7Z1nXHvhrKv3P4zl4j3CvOvDDAu7LJJ2s9zNwGE2fg65FfoY7pO6HvXNss2rqPG+cWj4oLhsWr4cTPtdjgTcLOXeaamuqegY0aaq3Kc+P1/GpH8+jAYvLOwJ/Sn2Jb5ux3/dSP6AnZux3FU1BjxeZ3/Pi/mNTyw6Jy2YBnXLsdy9NgcwmqeUbEIK5JF9DM/brRFOwcQewVmpdN+CBuG5kge9jb2B3oH3G8o0Irecc+H3Guh/RFExskSWfPwa2y1g+Iu5zbJ68nB+3mQDskVpuwMmEAG4BMWBOrR9KU4D3MrBeal3nOF8+nnMHngG+ldpmq9T9d39G2klA/QSwRsa6bsCpwGoZy5+O+zwGrJ9aviJwI01Bb8fmPgOEANgJAfDuGdt3AA4A9slxPhYCz2a816Np+lz8nfAPQJ/U+n1S9+B+Gen2JncAd1FcNyceo11q3QqEz/EyQWuee6G5AO6auP7VYj6/qXQujunckWP9z+P65zOWD0rdc+8BfTPvuZaeO021NVU9A5o01duU48fL4pejA5fm2G9EXH9vxvLn4vI/59jvhdZ8odJUEnZaatnyhJIHB47Iss8GNJVwbJNl/SY0leoMzVh3SvJDkv5xTK1fkRBQObBzia7JhjG9/2QsPysuv7KAtJLrdGyO9asDXxECtC1ybHNFTOPGjOVD4/L5ZJSGpbb5SdzmK6BblvUDUkHDlqnlj8ZlP2jh+/wOTQH6ilnWt6OpdO7IFnwGDo/L/l7AuU7Ox9xs5wMYTVMJcP8s6++O66/KWN4722cGWCueewcOLdG911wAdz9ZAu4ijrN+PA/zgFWzrB8Xj3NMxvJBqTwucw6zpJP13GmqvUl14ERKY2PCFyyEUrNsLo/z/c2sPYR6YoSWZwA359jvpmIzZWa7E4KxRYTHWwC4+wJgeHx5bJZd9ycEpePd/ZXMle7+NuGRVjaHxPlN7r4ky75fEUqJIAQjLRbrSZ1kZrea2eOx3tZIQiAAsHFGPa9Jcb5PCes1HUAoqRnl7m/k2OYfcT4gx/on3f2jPOlDKGmZlrnS3UcQHjNCCMISyXs9yEKXFc1JrtMd8ZpkHmcJ8GB8OaAF6SXH38FCVxSFeDTH+Uje53h3z9Y/WbJ+gxYe57uEf17ec/fhzW1cIl3ifJlzDGBmnWzZzoA9s36au79PKDHtROhqJJ1GP2Ab4Evgnhz5eDPHOZQ6pX7gREpj4zif6e4f59gm6eBzJaAH4QdvQ0JJR1K3J5vW9EV2XJw/5u6fZay7DTiJENysk/EDulGcv5Yn7fGER5qZkorqvzazE3Ps2yvOs1aSzyZWmL8HyBeIGaE7hqR/u38QSiC3ACab2VOEksGRwEvuvqilx09Jy8aJMwAACUtJREFU3t8mMXjMJmlkkOv9/SdP+sm9lOt+gHAvbUMoCU38mVB69xPgADN7jFB6+6y7v5UljeR9HGFm++Y4zlpx3pLr9CLhvO4KvGtmzxEei44kBLvz8+z7Xo7l05pZn9zTK7UgfwCbx3mzDQlKaHac5+qcdwnhOiVWAzbLse2NhKoExwPXp5b/NM6HuXuuvvny3XNShxTAiZRG8l/2p3m2mZpl++SHZ567L86x3+wcy/OKrfOSUpbbMte7+ygzexfoQ/jR/2NqdZKvfMfOtW7VOO+bY33aCi3YBjNbmabg7TFCncLXgVnuvii2VEzO3zelT+4+18x2I9RZO4xQAvPduHq6mV0J/CnPuc8meX9rxymfZVp9RllLY6JC7qVkW9z9DTPbmfBe9weOihNm9jZwtrvfn0ojeR8b0xQ05tLsdXJ3N7MDgLMJ99NecQKYbWY3AuflCDBynQ9v4fqcLTMzrBznM1u4fSkk/0z0zrbS3RcSgl4AzOx7NJV8ZvoHIajd1sz6uvu/LbS+Tkrk8pXW57vnpA7pEapIaSTBzFp5tkn/2Cfbz4nzzslj1Sy65FjenENoCsTuzvaYhhC8wbJ9wiX5ynfsXOuSfXdyd2tmOraF7+UAQvA2idBC8Bl3n54qQVs9147uPtXdT4n79wN+ATwS97mIUEeoEMn7G9KC99fSwCKtkHtpqSDa3V9x9x8QgrMBhA5jxxNK6u41s32yvI8jWvA+BrQk4+4+293PcvfuwKaERjn3EkokzwCGtCSdMvoyzlfNu1VpJaW0W1oR3eekxWAv+Wfs+Dj/EaHU+XV3H9Oa9KW+KIATKY1k2J7VzKx7jm22iPM5NP1XPoHwCKU9TY93MhXbI/uxcT6bUJqTa3Jgo1h6k/hvnOcrRcuVr+TR35aFZzmn9eJ8bI5HcXn7s4NQp8vdX3P3a939AEKrTAiNLpbatJmkyvH+0pJ7aYs82yTr3s620t3nu/uzHjqM3ZpQ3zFpIZso6/tw97fd/SZ3/zHww7j4qNgvWrUk73mnCh7zEULL3A6EKgut9Zc4H2hmHWl6fJqrDq00KAVwIqXxDqG7BYDTc2zz6zh/NHlk5+5zaKr/cnzWvXIvzylWIt8jvvy+u6+dayK0goWm+nIQHlM6sLVlGdTdzDYi9dgnQ1KJ+tT4A1MKyWO3XI8sf1VEmqPifHVbuhPgeXGe6/HnvwgtGXc2s3IEAg/H+VFm1i1zZWyYsm3Gts15Mc7T/1wk1+k4M1ut4FwWJjnXBnyrzMfK5yHCtdvAzA5pbuNScPcvCX0YApxnzXSe3YL03iF8ZlcnDM21F6FF9O2tSVfqjwI4kRJwdwcujC9PN7OByTozW87M/gjsSWgNenHG7pfG+almdnzS07qFHvIvpwWlS1kcS/ixnEhTgJbL0Dg/NGnB6e7vEbo/ABhmZt/UkbIwcsG9hJLDbP5CqDDdF3jQzPqkV1rorX9XM7vJzHq08P0k72FnM/smoDWzFc1sCDlKVCyM0HCmZYwoYGarELoYgfDoaUFqdVJhfkC2Xu89DIF0SXz5QOwlf6nvUjPrFY/708z9W+BuQue3KwDDzeyboNXMtqTpev093QrWzIaY2TIlXGa2PuFRJoTOWxMPEvuZA560jJFCLNjGwmgR2zWXaTMbaGb/k+V6dyb0RQgwg3BPVkVsyJO0Bv+rmR2ZvnZmtoKZHWdm22ZPoWgXE7p56QQ8bWbnpq9r6vibk9HCNIcbU+m2A/7h7p+XKrNSJ8rVP4kmTY06kX8khuto6nMpGYkh6W9tEXBCjjSvyrHfIkLJnQOLW5g/I5QGOvA/Ldh+JcJjXQeOSi1Pj8SwGPg3oVVq0iFo0nHszVnS7E14XJW8p/cIpUD/JpSm5e07K0c+70jtN5nQR9kcQiB5bLY0M85rMjLA6zT1WD8b2CXjODvT1AfeJEKr1RHAWRnn+JpU2jNjfsbS1GO+A4My0h6abXmW97oloaGC0zQ6wZupfI1j2ZEYxqeu1X8JLS3fSe3zDhn9yhFKcZJ+CJNzNDqm9UVq+YCM/SZmLieUPCfbfxrPx3hCvTMndGJ9SCHng6Y+zIbmWJ9c9xFZ7r+sfZkRqivcmsrrjHhfTKCpY+CSjcSQ2q4j4fthcWqfD+KxxxDGKU6WTwdOzpNWJ5q+V5yMDpILOYc5PrvqB64OJpXAiZSQu/+cUN/nMUJ/U/0Ij2yGATu4e9ZWYu5+OqEhwVhCNwJ94t97x7SgqQJ2c/agqc7YMq1Psxx7DqFEDVJ9wnkordiREARNIbRU7Er47397wo9d1ny5+0SgP6F3+BGESuPbEgKG1wmlILsShkxqqWMILRzfJVTwX48QXO3j7kNz7PN/hIr8zxACmS0J53YS4ce0r7unu3DA3UcRRmgYCaxC6KdvD1JddnjwC0I3KnfGc7AloT+yGYQ6Z0cSOvQtmLu/TijBHEwIxjcldL3yCmFczV3cfUbGbqfH440jNDDZllC69gpwDqET16X6lYtp7EkYkeBRwj27DeFR9QTC+duP3H3+pd0HnEmo8/UVoSuMTQnByK2E0Q3uzb17ZXgYNP4Y4EBCKeQiwud0RUIgdRbhvir1cRfG74fNgMsIn++VCCNYbEQ4Z3cBA4Ge7n5DnrTmE8Y0hhBMP1Xq/ErtS4bsEZEaFevq3EMYimeb5ravFDP7F6FLjtPc/c/Vzo9INcTW3BCGRZtYweMOJ/yjcb67X1DCdHsTSgbx4lpRS4WoHziR2pfU+Sp5qUCxzGxdIOn8tWbyJVJF95jZAsI/Wr8o54HMbE3gB4THsX8tUZp/JZROL9/ctlIbFMCJ1AAzOxUY5+4vppZ1AS4gDJe0iAr3oRUr+p8G/MXdP0kt34Lw+KYjoYf9VyuZL5Ea1T/Oixndo8Viw5oLCZ+/+9x9comS3priuyySKtAjVJEakHocOYemBgibEP4bduCX7n5thfO0Bk1DGU0lVHBfnaYxX6cAe7n7f7PsLiIlZGbfJtTP60mo/7kA2NrdNURWG6USOJHacD0heNue8OXciTDO4yjg6syK9hXyJaHRwP6ERytbEVoSvkHoC+2KzErxIlI2axMa08wntFw9S8Fb26YSOBEREZE6o25EREREROqMAjgRERGROqMATkRERKTOKIATERERqTMK4ERERETqjAI4ERERkTrz/wGofUGx/jVFzQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 648x540 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# load models\n",
    "model = torch.load(\"models.pt\",  map_location=lambda storage, loc: storage) # load in cpu\n",
    "model.eval()\n",
    "\n",
    "\n",
    "#==============================================================================\n",
    "# restore validation set\n",
    "temp = np.load(\"../source_apogee_rc_spectra.npz\")\n",
    "apogee_spectra = temp[\"apogee_spectra\"].T\n",
    "apogee_spectra_err = temp[\"apogee_spectra_err\"].T\n",
    "apogee_spectra_all = temp[\"apogee_spectra_all\"].T\n",
    "apogee_log_age = temp[\"log_age\"]\n",
    "print(apogee_spectra.shape)\n",
    "\n",
    "\n",
    "#==============================================================================\n",
    "# extract input and output arrays\n",
    "input_array = np.copy(apogee_spectra).T\n",
    "output_array = np.vstack([apogee_log_age]).T\n",
    "all_array = np.copy(apogee_spectra_all).T\n",
    "\n",
    "x = input_array[:2000,:]\n",
    "y = output_array[:2000,:]\n",
    "y_train = np.copy(y)\n",
    "\n",
    "x_valid = input_array[2000:,:]\n",
    "y_valid = output_array[2000:,:]\n",
    "x_valid_count = x_valid.shape[0]\n",
    "\n",
    "x_all = np.copy(all_array)\n",
    "\n",
    "\n",
    "#==============================================================================\n",
    "# scale the labels\n",
    "x = x-0.\n",
    "x_valid = x_valid-0. # convert to the write float format\n",
    "\n",
    "y = y-0.\n",
    "y_valid = y_valid-0.\n",
    "\n",
    "#-----------------------------------------------------------------------------\n",
    "# make pytorch variables\n",
    "x = torch.Tensor(x)\n",
    "y = torch.Tensor(y)\n",
    "\n",
    "x_valid = torch.Tensor(x_valid)\n",
    "y_valid = torch.Tensor(y_valid)\n",
    "\n",
    "\n",
    "#==============================================================================\n",
    "# plot inference\n",
    "y_truth = y_valid.numpy() -3.\n",
    "y_category, y_normal = model(x_valid)\n",
    "\n",
    "# extract parameters\n",
    "y_category = np.exp(y_category._param.detach().numpy())\n",
    "print(y_category.shape)\n",
    "y_normal_loc = y_normal.mean.detach().numpy()[:,:,0] -3.\n",
    "y_normal_std = y_normal.stddev.detach().numpy()[:,:,0]\n",
    "\n",
    "#-----------------------------------------------------------------------------\n",
    "# the dominant mode\n",
    "y_category_mode = np.argmax(y_category, axis=1)\n",
    "y_pred = np.array([y_normal_loc[i,y_category_mode[i]] \\\n",
    "                   for i in range(y_normal_loc.shape[0])])\n",
    "y_error = np.array([y_normal_std[i,y_category_mode[i]] \\\n",
    "                   for i in range(y_normal_std.shape[0])])\n",
    "\n",
    "#-----------------------------------------------------------------------------\n",
    "# plot results\n",
    "y_truth = y_truth[:,0]\n",
    "plt.scatter(y_truth, y_pred, s=10)\n",
    "plt.plot([-0.5,1.0], [-0.5,1.0], color=\"red\")\n",
    "\n",
    "# plot error bar\n",
    "for i in range(y_error.size):\n",
    "    plt.plot([y_truth[i],y_truth[i]],[y_pred[i]-y_error[i],y_pred[i]+y_error[i]],\\\n",
    "             color='k', alpha=0.2)\n",
    "plt.xlabel('log Age asteroseismic [Gyr]')\n",
    "plt.ylabel('log Age inferred [Gyr]')\n",
    "\n",
    "# print 1 sigma spread\n",
    "print(np.median(y_error))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.2107642\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x540 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(y_truth-y_pred, bins=20);\n",
    "print(np.std(y_truth-y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Plot PDF."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_normal_mix(pis, mus, sigmas, ax, label='', comp=True):\n",
    "  \"\"\"Plots the mixture of Normal models to axis=ax comp=True plots all\n",
    "  components of mixture model\n",
    "  \"\"\"\n",
    "  # x = np.linspace(-10.5, 10.5, 250)\n",
    "  x = np.linspace(-0.1, 1.1, 250)\n",
    "  final = np.zeros_like(x)\n",
    "  for i, (weight_mix, mu_mix, sigma_mix) in enumerate(zip(pis, mus, sigmas)):\n",
    "    temp = stats.norm.pdf(x, mu_mix, sigma_mix) * weight_mix\n",
    "    final = final + temp\n",
    "    if comp:\n",
    "      ax.plot(x, temp, label='Normal ' + str(i))\n",
    "  ax.plot(x, final, label='Mixture of Normals ' + label)\n",
    "  ax.legend(fontsize=13)\n",
    "  return final"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}