2_linear_model_file_input.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import itertools\n",
    "%matplotlib inline\n",
    "import pylab\n",
    "from mpl_toolkits.mplot3d import Axes3D\n",
    "from IPython import display"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<mpl_toolkits.mplot3d.art3d.Line3D at 0x118012a90>]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeQAAAFKCAYAAADMuCxnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAAPYQAAD2EBqD+naQAAIABJREFUeJzsnXmcU9X5/z8322SZyTAIgsMmuLIJCIUqIi604Ia1oF+x\niNSNr4KIoNalaov2i/6KCwpFxOLSFkFrUSsitZRNRSwIisiiUougMKBMMtlzl98f47mc3LlJbpJ7\nM5nJ8369eNlmkpyT7Xzuc87zPB9BURQQBEEQBNG82Jp7AgRBEARBkCATBEEQRElAgkwQBEEQJQAJ\nMkEQBEGUACTIBEEQBFECkCATBEEQRAlAgkwQBEEQJQAJMkEQBEGUACTIBEEQBFECOHK4L7X0IgiC\nIIj8ELLdgSJkgiAIgigBSJAJgiAIogQgQSYIgiCIEoAEmSAIgiBKABJkgiAIgigBSJAJgiAIogQg\nQSYIgiCIEoAEmSAIgiBKABJkgiAIgigBSJAJgiAIogQgQSYIgiCIEoAEmSAIgiBKABJkgiAIgigB\nSJAJgiAIogQgQSYIgiCIEoAEmSAIgiBKABJkgiAIgigBSJAJgiAIogRwNPcECKLUURQFkiQBAGw2\nGwRBgCAIzTwrgiBaGyTIBJEGJsSiKCIej0OSJNhsNvWf3W6H3W5X/z8JNUEQhUCCTBAamBCHw2EI\nggCn0wlBEGC32wEAsiwjEonAZrPB4XCoQqwn0iTUBEEYhQSZIH5AURSIoghJkiDLMuLxOOx2O5xO\np7plzcSWCbDD4VAfK8syJEmCoigAkFWobTZK4SAI4igCWzwMYPiOBNGS0AoxE9JgMAigMSLmfyeC\nIEBRFAiCAJfLlRIJ88/J/sv/Y49PJ9R2u52iaYJonWT9YZMgE2ULi2i1QqwoCuLxOKLRKADA5XLB\n4XCoUbAsy0gmk02ej4ms9l86oZZlOe1zOByOJufVJNQE0aIhQSYILbIsqxExi3SZEMdiMcTjcfV2\nu92OqqoqJJNJ9TYA6hlyRUWFKtL8P21EbUSotf/4v7O5sGham0xGEETJk/WHSmfIRNmgJ8Q2mw2K\noiAajSIWiwEA3G433G43wuFw1ufkhZKHj6bZP1EUCxJqURSRTCZJqAmilUKCTLR6eCFmMCGORCKI\nx+MAjgpxumQrbdSaieYQavZY7fh6Wd8EQZQeJMhEq4QXKybETIhkWUY0GkU8HocgCFmFWI98Ra3Y\nQs1Kt1wuV5PxSagJorQgQSZaFVohVhRFFVpZltUzYkEQ4PF4UFFRkVaIBUHQTbzixzKLQoRaT6SZ\nwCYSCXUrWy+iZmOTUBNE80OCTLQK+DpgXoiZqEajUSQSCVWI3W53ixAao0LNOorx2Gw29YKC7zJm\nZOubjc2fS5NQE4S1kCATLRomTKIoqtnNTDBEUUQsFlOF2Ov1oqKiolUISTqh1sv4ZjsDjEwRNYMX\napZhzo9NQk0Q5kOCTLRIsglxNBpFMpmEzWbLW4hZKVRLQtsBLBKJQBAE3fIsvYi6EKFm9yWhJoj8\nIEEmWhSsz3QikUgpXdITYp/PB5fLVbAQtPRe1Px7xMOE1UqhZu8dCTVBZIcEmWgR8BFxMplEQ0MD\nqqqq4HA4kEwmEY1GIYqiqUKciZYYPWvhW3jyWCnUeuOnax9KQk2UGyTIREnDWyDyURcANSIWRRF2\nux2VlZWqM1Ox5tYasUKo9ZLSeKFOJpNIJBKw2+2qgxYJNVFukCATJUkmIWZ1xdFo1FIh5qNgEoHC\nhVrP7Yp/rlgspooxL9Ta8fXKs0ioidYACTJRUmQSYrY1zQTZ7XbD4/EUZSFurdGwGeQi1JIk6ZZW\nsc+Qzwvgn4d/PhZN82OTUBOtARJkoiTIJMSJRAKxWAySJMHhcMDn8yEcDqvRVDFggkILvHEyCbU2\nmmYXWSwqTteVLF+h1rpn0edIlCIkyESzwguxVvASiQSi0ShkWYbT6YTX64XT6YSiKAiHw5ZHrayp\nSENDQ0pUx/pgA1ATyWiBN45eDTX7TJ1Op9rQJJf2oSTURGuABJloFvj2lkyI2aKqFeLKyko4HMX9\nqoqiqJpOSJKEiooKdd78+ai24QaV9BSGzWaD0+lMua2QPt+5CjX7DPU+R/osCashQSaKCt/eUivE\n8XgcsVgsqxDz541mw9cys3Gqq6vViwd+rqIowuPxGDojzVQaRBxF732xwpAjnVCzz1BbR01CTRQD\nEmSiKGQT4mg0CkVR4HK54Ha7myUi1jYVkSRJNaJIt+ima7hhhnMTkZnmFGoWZbtcLhJqwjRIkAlL\n4b2ItRm0sVgMsVhMXdg8Hk+TxTUdZjXmyNTdi80tV8x2biqHxb1UnLP4jmL8+68V6kQigWQyCYfD\nkVNErb14IwgeEmTCEnghZrBkKF6IKyoq4Ha7DQuxWVjVZjMTRoUinXNTOQq1meQi1OxsmZFOVLWJ\nafzz8RehbHw9oWYNTwiCBJkwDa0XMYNFs9FoFPF43BQhzjdCzleIiy3U+XbFosU9d9IJtZ5rlvb9\nj0QiuhdKelvf/MUW/xml60pGRxjlBwkyUTBMPOLxOOLxOJxOp7q4MSFm2chutxtut7voW3f5CHGm\n5DGrF8pCumLx56N8i0pa3HMjXX4A+66zJD8z+nyzHu2ZhFrb8IRofZAgE3nDb80xB6ZoNAqXy5US\nEQPWCLGRCNmsrWm2cOY6vtkYEWqWOMfORyVJQjgcbhGJZKU0Fz34919RFLjdbgDWGXJkE2oqs2td\nkCATOaMVYkVRUqKJaDSKRCIBQRAsi4izLTzJZBKxWAzJZBJ2u70oZ8TNCS8UfIZ6OBxWbzOSccxv\nlbbW98oMtBdjhexoAPkLNRub/ZeEumVDgkwYhm+KIctyihCzEiGgUQw9Hg8qKiqKvjXNWzGWgxBn\ngwmF0WYb2o5kxUoka629wptTqLXjp+tMRpQOJMhEVrIJMR8RA1Ddl6xEm9SlFWIzHKCsbEDS3GTK\nOGZb3rmIRLlF1IWeyZsh1NlKqnihTiaTKa1JtbsiFFGXBiTIRFq0Qgwc3d4URRGxWEwVYq/XC4fD\ngWAwWNQ5WiHE5YwgCE2ashQiEvQ55EYuQm20Kxz/XCxCzhRR858jfZ7FhQSZaAJv+KBteKBNkvJ6\nvaioqIAgCCmlTsWYI29qb7UQa5+3nBYmK0SiNbx/xXwNmT4DI81O2FzZb9RIRK1tWMM3TSGhtgYS\nZEIlFyHWO5stxhav1hPZSiE28nrKuZyoUJHQi+BayntZKp97Ls1mAKQ0PDESUbPnIqEuDiTIREYh\n5reEs5UNWSXIbGuN35p2OBxqy81iwBY4WmSyk0tHLK1QR6NR6gttAtrPIJlMIh6Pw+v1Nvkc8unz\nDegLNZ/xTUKdOyTIZUw6IQbQRACb42xWT4jZPMLhsHqubTUsaQ1AE7MJNi8SjexkEmr++KGlJJK1\nxM+bPyPmyWdXw6hQa8fXNjxh82nuz7MUIEEuQzIJMb8lnKsQmxUhZxLiYv1gZVlWu4slEgm43e6U\nFohMMFipF5C+RIXIDL/t7XK5UpKOSrV1aEvLvNf+zrUUsquRr1DrlWfpdSYrJ6EmQS4jmNDxFoh6\nQuxwOEouItZLqrJiUWRCzMQYAKqqqmC325FIJNRFxmazIR6Pw+Px6G4B8vNsrUlNVlJIWRC95/rk\n8/rNEmpeXPnPlN1fK9T8xa/dblfb8SaTSdTV1eGEE05olZ8nCXIZwHfVYkLMfhSJRAKxWEwV4qqq\nKjgcjry/7PkIZSlFxHzPbYfDgVAolPG8XG+hz9TGkn+8XmTXGhcZMzEi1Ebf80KFuiV9VmZfvOYq\n1EYazmiFWpZl9YiIHWls3boVkyZNwu7du019PaUCCXIrRpZlNZmDP68BoPadlmVZFWIzmnnkIsjs\nijifLXKzImQmxMyFim/1qd0eNQovGnxNr5GoohTPSotJIeKY73ueLpLLNJeWtmUNFOcCIpNQ59pw\nhr3H/PM1NDTA7/e32t8DHXC1QmS50c+VuS+Fw2EAjT+WRCKBQCCAcDgMu90Ov98Pv99veWctHkVp\nNHgPBoNqBFpZWQm/31+0NpeK0mh+EQgEEIvFUFFRgTZt2sDr9aoXLUbLnozCFiun04mKigp4PB74\nfD74fD643W71/JR9frFYDJFIBOFwGJFIRO3NzfqHE9nJ9J6z9q5MwFmzG+17nkgkUrrUtUSae96C\n0NhwxuVywe12w+v1wufzwev1pv3us/yMWCyG9957D7Nnz8aWLVtQWVlpSkLn+vXrMXr0aHTq1Ak2\nmw1vvPFGyt/D4TCmTJmCLl26wOv1onfv3liwYEHKfeLxOCZPnox27dqhqqoKY8eORV1dXd5zogi5\nFcGuOnljdHZlmUgkkEgkIMsynE4nKisrm3RkMoNMkas2Ii5kizzfCFlRFHVrWlEUdZFuzuQrtljx\nlNtZabEFo9BIDoCaGNlSdjFKbX7Zjh/YebLdbse2bdvw2GOPIRQKAQD8fj969eqFPn364IILLsDl\nl1+e8/jhcBj9+/fHtddeizFjxjT5+2233YY1a9Zg8eLF6NatG1auXImbb74ZnTp1wsUXXwwAmDZt\nGlasWIFXX30Vfr8fkydPxpgxY7B+/fo83hFAyOGH0DIvDVs5fAs8vlMW+/HxXsTs6tQKIWYEAgE4\nHA74fL6UOWqF2OPxFHRWzV5XTU2NofvnI8SSJCEQCOgmdbFoio+oi4GeSGsjN16cWSMXt9tdcgsy\nDzta8fl8JTdPvYsjveOMUq63jUajAACPx2PpOIoCfPedAEkCkkmgUycF2rdAUYA1a+x4800Hpk5N\n4LPPbDjxRBknnZQqMdrvhCzLeOyxx/Dee+9h1KhR2L59O7Zv346zzz4bs2fPLmjeNpsNr732GkaP\nHq3e1rdvX1x55ZW499571dsGDRqECy+8EDNnzkQwGET79u2xZMkSXHbZZQCAXbt2oWfPnvjggw8w\nePBg7TBZvwwUIbdQtEKsKEqKMPDiA0DdGrIaPnI1MyLONE4m9ITY7XY3iYzSjcGeo1TIpTMTEw3m\nh9xcJUItHW0kx353LpcrxdaylFuHateH/J6jUWz372f/bPjmGwH79jX+d/36pnKyYkUEQ4c2Bgrh\nMPDSS05Mn350HVq4sLGxz403JnDwoICRI0WMHy+qcwZSEyhjsRh69+6N22+/vaDXYoQzzzwTb7zx\nBn75y1+itrYWq1evxueff46RI0cCADZv3gxRFHH++eerjznllFPQtWtXbNiwQU+Qs0KC3MLgM0l5\nIWYCFY/HU8SnoqICwWCwaAsAu5JlSWNmC7FR2HsRjUZzFmIjlJqQ6Qm1oiiIRCLqlnhr3/YuNpnq\nbfOt3W2u9z0QAF5/3YlTTpHwySd23H9/BU4+WcaWLYX9XlascGDdOjsWLXLi4EH9C4ILLhDx/vt2\nfP65DbfccrS1J9/56+g8Azj22GMLmpNRnnrqKdx4443o3LkzHA4H7HY7Fi5ciKFDhwIADhw4AJfL\nBb/fn/K4Dh064MCBA3mNSYLcQsgmxHymMC8+xYzu+DmGQiFTs7czjcn/aLVC7HK54PF4TBPilgQf\n2fEtRnMtESpmg4aWcCGQ7TdVSO2umVn2ogiwKRw6BNTVObB/vwNr1tjxzjsOfP115oi5UDEGgCef\nzN7adsUKB2w2BX/5SwyDB6cma2lfdzAYxMknn1zwvIzw5JNPYuPGjXjzzTfRtWtXrFu3DjfffDNq\na2tx3nnnpX2c3oWEUUiQSxz2I9bzItarnWUlO4xibLtqt6ZZ1rSVQqxX+8t2B2RZLpoQl9J2tlFy\nLREyUkNaqJC2xPcxV/I5bmDw7/eOHU5s2eLA1VeLsNmEJtvIX31lw1NPaYWwyuJXVxhz5sRx0UWp\nr1nvOxEIBFBdXW35fGKxGO699168/vrrGDVqFACgT58+2LJlC2bPno3zzjsPHTt2VKtF+Ci5rq4O\nHTp0yGtcEuQSJZsQRyIRtSxAT4h5rOpqpXdGzBb4YpVRWSnEpXiGbCWFZh6XakKTVRT6+j7/XICi\nCDj5ZFn3uCEQUPDQQxUYPDiJw4eBuXO92Lfv6H2mTi1o+JLhvvviuOaaZJPb9SLNhoYGtGnTxvI5\nsQxv7fis1zoADBw4EA6HA6tWrVKTunbv3o29e/fijDPOyGtcEuQSI5MQS5Kkbk0LgpBViBlmC3K6\nZC1m+pBvQ418CAaDRY2IyxGjZVmlmtBkNul+S//5j4BjjlHAgiVJArZvtyEQEPD66w786lcJPP20\nE199ZcNPfyrihhsaM54nTkzgk0/s+Ogj/e/uM89Yn4zZXEyalMDttyfS/l17HGVmhBwOh/HFF1+o\nn+eePXvw8ccfo23btujSpQuGDx+OO+64A263G926dcOaNWvw4osv4oknngDQWHp13XXXYfr06aip\nqUFVVRWmTp2KoUOH5pXQBZAglwxaIQaOLmhaIWYNDXLJmjRDkFlDD22rTT4atioa184hEokAgNrd\ny8pSLqIpmWpI8zknLeVdiP37BRx3nIJPP7Wha1cZVVXAoUM2bNvmxOjRMmQZmD3bhYceqlAf4/Eo\niEZTLzyeeeboNvIrrxz9zTz/fHEsREuRzp3lJmVRDL0IORgMmibImzZtwrnnnqt+l2fMmAEAuOaa\na7Bo0SIsXboUd999N8aPH4/vv/8e3bp1w6xZs3DjjTeqz/H444/Dbrdj7NixiMfjGDVqFObNm5f3\nnKgOuZlh24F6XsSSJKnWf3xEnGuEoVcbnOsceSF2Op1wu92629KRSASJRML0bSU2B77dpyiKqK6u\ntiwqVhQFR44cgdfrhd1uT2lByjKYSz0qj0Qiah1yc6IXTet1W3I4HEUpy5Jl4PvvBbRrp7+s7d0r\n4IUXnOjdW8bEiam1u+PGJfDSS40iWlOj4MiRlh3xNzd1dQ3Q+3qGw2E4HA5UVDRe6CiKgu7du2Pj\nxo048cQTizxLU6A65FIlkxCzphNMiL1eLyoqKgrq85tvVyteBJ1OJ3w+X8Zo1Irtce0cKisroSgK\nGhoaTBtHD/Z+x+PxlKYr/N+SySRkWW4127FWka4siwk1awlqVlnW8uWN39GzzhKxerUDTz7pQmWl\nggEDJKxb58DmzY3zOHKkQc1EVhTg889t2LnThvHj0zfQYGLc+Hj6vAvhlVciumIM6FdQFCupq7kg\nQS4y2YQ4Go2q3ZUKFWJGriKZTgSLuS2sPafWzoGdVVq11SnLstrdSJIk9ayeLxcSRVH9xyh3c4hc\n4Le9mRjHYl40NDQ2kThwADh4UMHBgwIOHxZw0UVR9O0bVx+rV5b18stO7NjRmGWcTDZ9z9esSf0O\nd+1aiRNOkLF1a+nucrRWli6NYORISfdv2qYgQGPELEkSCTJROEyIk8kkgsEg3G63KrZaIfb5fKab\nLBjtalWoEBcaIesJsV5Ubt1W5tFSMjYGO7NnrTNZpCeKolrvzW/D6mUhazOQyymaZtvDdXWp/w4d\nEnDwoA11dQLef78SkUjm9+Pddz249dYYOnQQ0aGDhLZtRSxd6sQ//lGBWbMC+OorB264wZ/xObQ0\nNAgkxs3Ac89FccEF+mKcDtbGtjXni7TeV1Yi8BEx8yJmZ2e8B7BVQgxkF0kzI2K+VCiX15Epc7sY\npKvpDgQCWR+bbTuWj6j1mm/wYt1SomlJShXZgwcFLF/uwOuvp35elZUKolFAkgp/TR99ZMc11+jn\nQfzzn603E7k1MnhwZjHWi5ADgUCrtl4ESJAtg+8zzYRY22taluWcPIDzJZ0gl+LWtFEhNqtGWNvr\n2mgpWTZyyUJOJFLLPrT9posVTUtSY6/igwdTo9i6OluT6Lauztj7k0gAtbUKZs2K4f77K7BnDzm+\nljtvvXUYbduKiEaz5wZoBbk1b1cDJMimw7e31AoxEx6gcWG2WogZWkG2UoiNRsjNHRHrmU5kcn/K\ntFDkgpFo2sye06IIHDliQyDQKKr/7/+54HYDHToo+PBDO44/XlZF9rvvBMhy6vNVVSlo315Bhw4y\njj1WQY8eMj77zJaDIAv4+mshY5IUUT6MGZPAkCE2yLIjY0kcgyVMCoKgdsSiCJnIip4XsVaImfCw\nloV8f2ErYYJcKhExv1Wfr/FEvhFyrqYT/MWMlTsY6aJpbZesxiQyBYcO2XDokA2HDzvw3Xd2HD5s\nx6FDdhw+3CiWjeezAr7/vhKKkn7e3bo1Cu327XbccUccAwdKaN9ewbHHKmjXTsE553jx0ENxdO+u\n4PnnnZg7t3xrZonCmTo12eSiO1NJXDwex+LFi7Fw4UJ06dIFsVgMb7/9Nvr06YNOnTrl/Ztcv349\nfv/732Pz5s349ttvm1gvAsCOHTtw1113Ye3atRBFEb1798arr76Kzp07q3ObPn06li5ding8jpEj\nR+IPf/hDQeYXJMgFkkmI9ZpoOBwOy8t1tLCFPRAIWC7EmYSSXZiIoliUrXoe7cWIWZ29Ct0yj0YB\ntxu6zRH277fhvPO8eO+9CNq3bxxn3z4BvXpVNrlvdbWM9u1ltG8voV27JE48sVFUjzlGxLHHKujY\n0YkrrvCivj51IFkG1q1r/B78/vcVuO66BP74x1TRvfxyb0GvkSAAYPXqMAYMaFp7rrdrxLyQ3W43\nunTpgj59+uCTTz7Bzp07ccEFFwAAqqurMWXKFDz00EM5zyUcDqN///649tprMWbMmCZ///LLLzFs\n2DDccMMNePDBB1FVVYXt27en1PNPmzYNK1aswKuvvgq/34/JkydjzJgxWL9+fc7zYVBjkDzhhZjB\nhEXbwMLj8aRcFTJBrqqytuG7Nhp0Op3weDyWRsSiKKpbS3yJEi/E7P0oRIhZ0w6fz6c2Dkh3v0KE\nuL6+Xn0MO+tlUXM4HEZFRUXabfbduxu3ic86q2kCSzQKDBniw1df2XDssTJmzozj1FNlnHyyjA0b\n7KiuVjBiRGMC0xNPxPDttwI++cSOcPiogDIuvDCJn/5UxEMPVeDwYRvWrKlHJKLg5Zdd+N3vgmho\nEPCb3/ixZAkJK9E8fPBBGL16NRXjdDBB9vl86jrx6KOPYv/+/bjnnnvw6aef4tNPP8XJJ5+sK6i5\nYLPZmkTI48aNg8vlwgsvvKD7mGAwiPbt22PJkiVqH+tdu3ahZ8+e+OCDD9K1zqTGIGbCIs10Qsyb\nHGSKQvlMa6vmyc/FbrdDkiRUVlYW9fyluSPiTHXMuT6Xlvp64N13K9Cnjw3HH68f4c6f78Qf/+jC\nz36WxMMPx1Fb2/g8//qXHX/6U2NPYwCoq7Phf/83/RnrtGmZM4jfesuJt946elFwzjlHu6Q9/3x+\n3dkIwiy2bAnhhBNyP1oCUo+J6uvr0aZNG3Tv3h3du3fHJZdcYuo8+bGXL1+OO++8E6NGjcKWLVvQ\nvXt33H333bj00ksBAJs3b4Yoijj//PPVx51yyino2rUrNmzYkHcva0p5NADfTSgej0MURfXcTxAE\nxONxBAIBRCIR2O12+P3+jPVyVrovxWIxdS4OhwN+vx8ej0f9u9WwH1A4HEZDQ4OavOb3+y0p6dK+\nJhYRB4NBhEIhCIKQ9fPIRLr5Llrkwi9+0Rb9+rVBdXUV/P4qtGtXialTK7BunR0NDcDbbzeO9/77\ndgwa5MO8eU7s3y/grbccePXVowJ6wgkyrrsugdNPz60ukyBKnc8+y12MgeZ1eqqrq0MoFMIjjzyC\nCy+8EO+88w4uu+wy/PznP1e3ow8cOACXy5ViuwgAHTp0wIEDB/IemyLkDPBdmdgZMZ/Zym8Hu1wu\nuN1uQ4u+Fe0ltRaE/Fy0ZTVWIYqiavpgdRa53nPyEbnVWdvhsIAOHSQcPHh06zuREPD8864mZgEs\nI/nuu924++6mz/XllzZ8+SUlSxGtiwceiKNz5/zXOe1vvFhlT2z38mc/+xmm/uBxedppp+H999/H\n008/jWHDhqV9bK79F7SQIOuQzgKRvdF8uUw+yUFmCbKeEOvNxWpfX22nMaCxu1Wxssi1W+P5ZG0b\npb4eqKpqLCfixZggiKNce20CM2bkHwjorVXBYLAoEXK7du3gcDjQs2fPlNt79uyJ9957DwDQsWNH\ndSeOj5Lr6urQoUOHvMcmQeZIJ8Q2mw2yLKviZ6RcJhNmtJc0IsT8eOxxZqLX8tPpdKK+vt7UcTJh\ndYMVQRAQjSo4dEjAcccBw4ZV4sABAYlE662FJIhC+fBDOz76yIbTT88vV8Zq68VMOJ1O/OhHP8Ku\nXbtSbt+9eze6desGABg4cCAcDgdWrVqlJnXt3r0be/fuxRlnnJH32CTIyC7E0WgU8Xi8YCFmFNJe\nUrtNbiQ6N1uQeTcqbcvPYpxTswsBoPE1WdVylLFggRsPPliBqVMBmw0kxgSRhZNPlnHSSYUlruo5\nPZkVIYfDYXzxxRfqerVnzx58/PHHaNu2Lbp06YI77rgDV155JYYNG4Zzzz0XK1aswJtvvom1a9cC\nAPx+P6677jpMnz4dNTU1qKqqwtSpUzF06NC8E7qAMi970goxcLRJQ7rexoW2VAQaz3RDoRDatGlj\n6PnyFWKGJEkIBAKorKwsaBuZ92e22WzqtrRWCL///nt4vV7TPXi147P3Il+fZyP85S8xXHdd627X\nRxBm8s9/hjF4cGFirPXxVhQFPXv2xIoVK3DaaacVPMe1a9fi3HPPbbJ2XXPNNVi0aBEA4Pnnn8f/\n/d//Yf/+/TjllFMwc+ZMXHzxxep94/E4br/9drz00kuIx+MYNWoU5s2bl6kxSNYr+bIUZN7wIRwO\nQ5ZlVFVVqULMImLAXCFmJJNJNDQ0oLq6OqOoFirEDFmWUV9fn7cga4WQd6rSw2xB5scXBEF1XwoE\nAnC5XPB6ramvVRQFEycqWLqU2j4ShBG2bQuhW7fCpSIcDsPhcKg9BhRFQadOnbB9+3Z07dq14Odv\nJqgOOR3MT5dts7KIOB6PQxAES4RYS7qLIbOEmJHvlrVWCI36M5u1dc1fHOmNb+UWeTKZRCQSxdKl\n7dXbamoUMqQniDSccYaIUEiAGbGb9jgvmUwiFAqhpqam4OcuZcpSkPkaYhYtBwKBlOjLSiFOJ5Ba\nwwOz2juLthnNAAAgAElEQVTyz2+EfIWYUahQaj2JPR4P3G53URqKsNKtxuYvqe87iTFBpGfDBgcO\nHIijd+/Cnkdv7WhoaIDD4bBsN6xUKEtBBqC2PmQ1usVc9LVj6DkPFZo4ph3PiEhKkpSyS5CrEBeK\n9ty+GBdHDEmSEIlE1IzxyspKTJtGW9UEYZRbbglhyJAGRCL6rmS5riN61out2ekJKFNBVhQFwWBQ\njUITiUTRxBg4+kVjW7JWCbFRtFvDhV6c5Boh5+tJbMaWtfZ8nGVs795tw6JFZfnzIIicOe88ETNn\nipDlCtWtibcPBYz7fOu1zSwH60WgTAWZtVMUBAHJZBKJRKLgDiv5EA6HAaAoQqwnXmYLca7k6kls\nJpnOp1evFnDBBdQ5iyCMcMstCfzud3EAjY5NrDtevj7f/N8YgUCgLAS5bHtZs05OVnex4lEUBdFo\nFIFAQJ1DdXU1fD6f5VExL8iyLCMcDqO+vh6JRAIejwdt2rSBx+Mx5QufLXJlQlxfX49oNAqXy4U2\nbdrA5/PlJMb5RMiyLCMSiTR57W63Gzt32jB8uJPEmCAMcv31CTz0UFz3b0xsmfe72+2G1+uFz+dT\nL4BZe19RFNVEVtZjIB6P47PPPsOyZcvw3//+15SmIOvXr8fo0aPRqVMn2Gw2vPHGG2nvO2nSJNhs\nNjz55JMptx85cgS/+MUvUF1djZqaGlx//fVqcFUoZRkh8xRDkPUiwXg8DqfTWbTtaVbSFYlEUpKl\nrDijTSeUuXYYMxMj2+JTpzqwcWPZXqMSRM789a9OzJ4d13U7SwcLhLTrDm/iw6LolStX4oEHHgDQ\nGMAMGjQIffv2RZ8+fTB8+HAMGjQop/lm80FmvPbaa/jwww/RqVOnJn+76qqrcPDgQaxatQqJRAIT\nJ07EpEmT8Oc//zmnuehRtoLMl84A1ghypmStZDJZlKgcgGqQwSwJi1HSxVOoJ3E6jNhYakvI0m2L\n//vfAtavJzEmiFyYOjUBs5YRQRBgt9tVi1uPx4Np06Zh3LhxWLhwIVavXo2+ffti27ZtWLp0Ka6/\n/vqcBXnUqFEYNWoUgPRr/v79+zF16lSsXLkSF154Ycrfdu7ciZUrV2Lz5s0YMGAAAOCpp57CRRdd\nhNmzZ6Njx455vPKjlK0gM6wQZCNno8VoM6nNWrbZbPD7/ZYLMRNKdgEQiUSyekSbTa4XAb/+ddn/\nFAgiJ666KlmQgUQ6+HVREAR07NgRdrsdQ4YMwdy5cwEcTca0YuwJEybgzjvvbGIuAQAbNmxATU2N\nKsYAMGLECAiCgI0bN6p+yflStquQFRGyngFFuiQlKwVZr+0nbx9pNWzrKRgMQpIky4RY7z3Uuwio\nqqrSFeIDB4A5c+zo21fB2rUUHROEUc48U8ScObGctqqNki7Lmm8KwsxkzObhhx+Gy+XClClTdP9+\n4MCBJq0x7XY72rZtW5APMqNsBZlhhiDnIsT8uGYLsnYe/NZ0OBxuUoZgNmyrSRRFKIpiuSex3tiR\nSASSJMHhcGS9CIjHgeees6O+vnVnbhKE2fzlLzH80NXSdPQqXgKBAI4//nhrBvyBzZs348knn8SW\nLVtyfqxZVTplGxbwbx5zdcoVliQVCARyzhY2U5D57G1+Hl6vt2jnxKw/d0NDA4DG11csMebHZuP6\n/f6sEXm3bsAFFxTWBJ8gyo1Nm8I45hhrj9uaw3rx3XffxaFDh9ClSxc4nU44nU7897//xfTp09Gj\nRw8AjT7IdXV1KY+TJAlHjhwpyAeZUfYRMpC7OPJ9r/OtnzVDkHOp47Vqi5z3RGbbSMlkEslk0vKa\nQdb2tKGhIS8/5C++EPDSS8VtwkIQLZnXX4/g5JOtvYjVW6eCwaBp1ovpmDBhAn7yk5+k3PbTn/4U\nEyZMwC9/+UsAwBlnnIH6+nps2bJFPUdetWoVFEXBkCFDCp4DCTKMi5VWiAvJVi5EIPNpqGG2IPNC\nzFpNMjFk59VWwXfXApCTH/L+/cDcuXY8/jh99QkiF554IoZzz5UsHyfdlrUZEXI2H2SteYXT6UTH\njh1x0kknAQBOPfVUjBw5EjfccAPmz5+PRCKBW265BePGjSs4wxooY0HmP/BsYmWFN3I+AqlXwmO0\nw5dZgpyu1WQxOuhou2s5nU4kk0nVos0In35qIzEmiBy58cYErr02WbTx+PVEURQ0NDSY4vS0adMm\n1QdZEATMmDEDQKoPcrp5MBYvXowpU6ZgxIgRsNlsGDt2LObMmVPw3IAyFmSedPWsVggxP6ZRgSxE\niPWeKx/x1ApxJuMJs6PxdO5PsVgsp0S1AweANm1aja03QRSFk06S8Mgj+t24rMDKCHn48OE55Qvt\n2bOnyW1t2rQxpQmIHmUryJkiZCuFWEsmgTTTFznfCDabJ3GmsQrNPGRb86ze0O12p7T3zEX4d+wQ\n8KMfOSGKlFFNELnwj3/UQ5YFAEedm6xEu24wMyCrz5BLgbIVZB62sBdTiDN9qa1oMZmrSDan8US+\n7k/p+PRTAYMGUX9qgsiV3/8+CLc7iXg8tVkHc2sqxF4xE/xzsZ05q7OsS4GyFmSt4UJ9fT0Aa4WY\nHxtIFUirWkxqx8tEuu3hXH5s+UbIRttc5jJOIgFcd11Zf80JIi9+/vMkJk4UUFHha+LaJElSWntF\nXqxzFWnmEMUTDAbh9XrhcrX+i+qyXql4P2KgOELM4MVET4jdbrepna2yCbLZUWkuWHkhMmuWHR9/\nXLbl9gSRN++848B33wmorVXUqJj/TeZrr2hEqLVduqqrq1u99SJQxoLMziVYa0dmulCsRhrsyxWP\nx5FIJIre65lhhRAbjcZZm8toNGpJi81rr3Vg8WKqMyaIXLn00iQeeSSO2tr0v+Fsrk1akebXg3Qi\nrdc2s76+Hn6/3+RXWJqUbejAOjoxL1ygOJ7IbJxksrGEIBaLwW63w+/3o6qqyjIx1opkOk9iM7p7\nGRFk1l0rFAqpn0Wurz/TOCtX2kiMCSIPXnklgj/9KZZRjDPBommn06keO/l8Pvh8Png8HrhcLrU7\nYiKRQCwWQyQSQTgcVncrWcOfcDhsWoScyQtZFEX86le/wmmnnYbKykp06tQJ11xzDb799tuU57DS\nCxkoY0EGGou++e0TqwWZnZEGg0E1c9jj8VgqxAz2GtkZcX19PSKRCJxOJ6qrqw21+zQDURQRDAbR\n0NAARVFUITazxeaRI8Cll1rfspMgWhvbtoUwcqQ1zT+YULMjOa/XC5/PB6/XC7fbnXJGnEwmsWTJ\nEtTW1uL222/HwYMHMXPmTLz22mvYs2dPXq2OmRfyvHnzmoh7JBLB1q1b8cADD2DLli1YtmwZdu3a\n1cS96aqrrsKOHTuwatUqLF++HOvWrcOkSZPye0N0EHIQoVZXwCmKIiRJgiRJCAQClvVe1tuadbvd\naGhogM/ny6mxRb6wpDW2LWTmOa0WvfdTkiREIhG1s5fX682pzaUeLMqurq5WX4eiAN26uVBX1/rP\nmwjCbDZtClveGjMToigiFovB4/Fg7969WLNmDdatW4d3330Xoijiu+++AwDccMMNeOaZZ/Iex2az\n4bXXXsPo0aPT3mfTpk0YMmQI/vvf/6Jz587YsWMHevfuneKFvHLlSlx00UXYt2+fkU5dWRelsj1D\n5rEqQtYKsZ77UTGicpYwBUDtrlUsT+Jid/ZasMBGYkwQefD11w0olcoim82G7t27o3v37jhy5Aiq\nq6vx7LPP4ttvv8W2bdtwzDHHWD4HFsCw+ucPPvjAUi9koMwF2QpPZPY82YSYjWuVIOtF5bIso6Ki\nomjb4/F4XDWZMNJQJN9x2Hv4zjsCpk2jrWqCyIVhw0QsXx5t7mkASG8swc6Qa2trUVtba/k84vE4\n7rrrLlx11VWq77LVXshAmZ8hM8wSZBaNBoPBlGQlv9+vuxVuhSAzIdbOoaqqytILAAY7owYat5Q9\nHo+aOGdl2cLGjcAll7T+OkWCMJv16x047TQfAoHmngl0s6wbGhqK2qVLFEVcfvnlEAQBf/jDH7Le\n3ywvZIAi5JT/XYj7EnM/EkVRjYgdDkfWFpNmCiSLiPk58BcCVkfkfPkUADVZwyrYe5tIJLB4MQBY\nfxZPEK2Rr76y4YdAsFnRE7f6+nr07t27KOMzMf7666/xr3/9S42OAeu9kAGKkFXSGUxkgyUW6WUN\nGyl+N0MgtZnLlZWVaRPUrIjI9cqnrIYvHXvjDQULFvgsH5MgWiMnnSThyy9DsCC/My+062axImQm\nxnv27MGqVauauEvxXsgMM72QgTKPkHlsNltOYsVHo3a7PcUP2CiFCjLvSWxkDmZuGVvZXSsbyWQS\nkUgEkiQhFBIwcWJby8ckiNbK4sUxtG9fGkU0euthMbyQa2trMWbMGGzduhVvvvkmkskkDh48CABo\n27YtnE6n5V7IQJkLcj5b1mYIMT9mPlF5vpnLZkTkLDKNRCJqd7GqqqomQmzF9rj2AsTn82HiRDo3\nJoh86dtXgttdGmIMWOv0lMkL+YEHHsDf//53CIKA/v37p8xl9erVOPvsswFY64UMlLkg82QTR1EU\nEYlETBFifsxcRKvQEiIzzslZZOpwODK2uTRTkLWvm733y5YJ+Mc/6NyYIPLhkkuimD+/Hg6HDbHY\n0RaWdru9WftGN5cXspHgyEovZKDMBdlIhJzrtnCuGBEtSZIQi8Vy8iQuZDwt2l0BqxqoaEnnxawo\nAkaNcmLtWkqBIIh88fttcLtdpppCFIpeUlcwGGxynttaKWtB5tEKMi/EVjW0yPZcZnsS5xq1FnIx\nUmg0zly49F734MFObNtGYkwQ+TJuXBJPPpmA03n0yId3b5IkybAphJktd7WCzHpaF7PsqTkpe0Fm\nwsH+Wwwh1o6txQxP4lzG01Ls7loMrR+ynvOUooDEmCAKYOBACU8/HYP258y7N/FHUVqRtjqa1tYg\nC4KQUn7Umil7QWawq8NgMFg0EWICyS4IeCEGzPdnzibI6baI801YMxohazO2mUOM3uueMIG+sgRR\nCL/6VbyJGGcimxeymdG0ds0IBoPw+/1Fs8Vtbsp+dWOmB4lEAgAsafGYDr5DmNmexOnGK1ZEbkSQ\n9dp76mVsM6JR4JVXSqRYkiBaIMcfL2PUqMLdnKyIpjN5ITdnklkxKXtBDoVCalQWj8dNTdgySiAQ\ngKIoGSNDK2AXAsx4wu12w+PxFOX1a7uK+f3+rD22a2ooo5og8mXChATmzo1bOkYh0TRvESuKImw2\nm9rHulwoj32ADPj9flRXV6sWiFb3emZjMFNuAHA4HEXxJOa3yKPRqNpdy+12o02bNvB6vaaJcaas\ndb6zGesqlkmMd+0SsGBB2X9VCSJvbrzRejFOBx9Js6DD5/PB5/PB4/GgoqIipTFTMpnE3/72N5xw\nwgl48MEHEY/H8eKLL2Lr1q2Ix/N7DevXr8fo0aPRqVMn2Gw2vPHGG03uc//996O2thZerxc/+clP\n8MUXX6T8/ciRI/jFL36B6upq1NTU4Prrr0c4HM5rPuko+1WO1dxZZcHIw5KWAoEAIpGIKkLF6nDF\n0La59Hq9llwI8O9lY1etEILBICRJgs/ng9/vN3ROP2GCA7feSi5OBJEP116bwOzZzSPGmWDRNPOH\nZ0FRRUUFevXqhWuvvRYejwfffPMNJk6ciAEDBqCyshJ33HFHzmOFw2H0798f8+bN011vHnnkEcyd\nOxcLFizAhx9+CJ/Ph5EjR6pHmQBw1VVXYceOHVi1ahWWL1+OdevWYdKkSfm/AToIOQhQ6bRzMRFR\nFNVtlPr6elRWVsLlMrf7kzZpyel0qlvDgUDA8rpeNn4kEoGiKHA6nfB6vZZeBITDYYiiiKqqqial\nW0bP6D/9VMCgQdSJiyDy5aqronj6aTH7HUsAURQRi8VSAoT58+dj69atWLBgAbZt24ZPPvkE3bt3\nx8iRI/Mex2az4bXXXsPo0aPV22pra3HHHXfgtttuA9CYTNahQwe88MILuOKKK7Bjxw707t0bmzdv\nVv2QV65ciYsuugj79u0z2joz66JX9mfIVnkis+fSJi3x3a1YZ5hieSLb7XY1Oi3GOTW7yMk3UWzy\n5LL/ehJEQdx6awiAdY5rVsCvEaxtZlVVFc4880yceeaZpo/3n//8BwcOHMD555+v3ub3+zFkyBBs\n2LABV1xxBT744APU1NSoYgwAI0aMgCAI2LhxIy699FJT5kIr3g+YKchaIdSzQjR7TC16VowA1LNb\nq2Db8uysp5CM8VNPVbBxo9kzJIjyYNSoOI47ruVsbOqtS8VI6jpw4AAEQWhiodihQwccOHBAvc+x\nxx6b8ne73Y62bduq9zGDshdkszyRGdk8ifUwUyC1Pbd5X2ZWfmCFIGu35Vk07vV683q+7dsFvPAC\nlTgRRD48/HAMEycGc6o3bm70yp4CgQBqa2ubbT7ZdvSM3CcXyl6QeQoR5HxcoFgymRkCyeqpWYcx\nvfGLuS3PbsuHeBwYOJDOjgkiH1avDmHgQAWRiAJBaDl5u+n6WFvdNrNjx45QFAUHDx5MiZLr6urU\nLeqOHTuirq4u5XGSJOHIkSNNIutCaDmfVhHIxw5Rr4zHaPYwG7MQgWTZy4FAQD0frq6uzji+WYKc\nTCbR0NCAUCgEQRBQVVWVEpHnS3U11RsTRL5UVYVVe1RW05uPzWspYJbTUya6d++Ojh07YtWqVept\nwWAQGzduVM+szzjjDNTX12PLli3qfVatWgVFUTBkyBDT5lL2ETIvHHwtXDbM6nmdryDn0+bSrK0V\nI6YTfDSey7hXX132X0mCyJvOnWW0b++EzXa0UxZrxdtcDk5GsTJCDofD+OKLL9S1ds+ePfj444/R\ntm1bdOnSBdOmTcNDDz2EE088Eccffzzuu+8+dO7cWU3WOvXUUzFy5EjccMMNmD9/PhKJBG655RaM\nGzfOaIa1IWj14zASIVthvJCLIBfS5rLQLWurTSduvdVBrTEJIk9YvbHD4QTgRDgcht1uh8vlyqnn\ndHP5IWsFWVEUBAIBUwR506ZNOPfcc9VjwhkzZgAArrnmGixatAh33nknIpEIJk2ahPr6egwbNgwr\nVqxIKYFdvHgxpkyZghEjRsBms2Hs2LGYM2dOwXPjKfs6ZABqRjCrndXbItF6EudST5uJYDCobvdm\nwox+14qi4MiRI/B6vXC7jZdC6F0EZHvtiUQCoVAIbdq0MTTHtWsFjBxJ58YEkQ8TJyZw//0JtGt3\ndJkOhUJwuVy6fRX0ek7zwQgfTdvtdjWStlKoo9EoBEFQ1yZFUXDyySdj1apV6N27t2XjFhGqQzaC\n1oKRx0wHpHRjZ0JrSVhIv+tcI2TtRYBZNpBaNm4kMSaIQnj+eRcmTEimCHIm0vWc5sWZRdPJZFK9\nj1akzdzyTrdlXVNTY8rztwRIkDl4QbbKk1hvTL1tcm0ZkcvlMqXFppH5m3ERYFT8P/5YwPDhJMYE\nUSgdOhz9remVEGVDT6QBNBHpdA5OWpHOda3UCjJbg6zOsi4lSJA5mCBHIhHLPInTjclgZUQsSzKb\nJWGh4/GYeRFgRJB37hRw7rnUo5ogCuHttyPo1UuCVbql52OsjaYlSUrp+wwgRaCNRtPaLl1s/S0X\nSJBxNEplWzOxWMxyIebHZg5MrKkH6+7Ft9k0ezyebC0+reDLL4ELLnAiEimNDE+CaIlcdFESZ57Z\n1N84nwg5F7LZLDKRzpZAxou9dl0KBAJl5YUMUB0yAKhWhEyQ/X6/ZQ5IWtjFAKtlBoCqqipD/sD5\njsd/8VkdtV4tcaHjAPoR8t69Ci64wIlvvy2fHxpBmM0vf5nAH/8Ya+5pqPA2i2x3Tc9mUZZlJBIJ\n1YI2HA6rTYTY35LJpGU1yLIs47777kOPHj3g9Xpx4okn4qGHHmpyv2x2jFZAETKgbs86HA7T/S0z\nIYqius3Dmopk6+5lFnxnL6OdxfJBux2/f7+Ik0+uNHUMgig3xo9PYs6c9JaKVkfIucBH06yNMB9N\nS5IESWqM8kVRxOrVq3HdddfhxBNPhCiKmDt3Lvr164d+/fqZItAPP/wwFixYgBdffBG9evXCpk2b\nMHHiRLRp0wZTpkwBcNSO8YUXXkD37t3x61//GiNHjsSOHTtMdwPkobInNHacYl+MYtgh8vW8LGI1\nWh5UKMyPWFEU2Gw2eDweU2uJGVo7S7Yd/+ijbjzySOYSL4IgMnPwYAM8nvR/Z2tMsXb6CkWWZUQi\nEbjdbuzduxfLly/Hhg0b8O677yISiaiBy0033YQ//OEPBY11ySWXoGPHjli4cKF629ixY+H1evHi\niy8CyG7HmCdU9mQEKy0YefRKqARBKEpUzrLGWXak2eVbWtjzSpL0w5Z4Es89V4WqKvrKEUQhdOsm\nZxRjwLo1zCr4iL5Hjx645ZZb1MBo6dKl2LlzJz7++GMcd9xxBY915plnYuHChfj8889x0kkn4eOP\nP8Z7772Hxx9/HIAxO0aroNUR1gtyphIqdm5tpScyX0tst9vVxiJWwkq5otEobDYbRLESM2fSVjVB\n5IPTqSCZbFyf/v1v4xfwpbBlbYR0Tk/V1dVwOp3o27cv+vbta8pYd911F4LBIE499VTY7XbIsozf\n/e53uPLKKwEYs2O0ChJkHcwSRyaGLGHB7XbD4/FY7sDEnk+vlpg5UlmF9jWzjO19+0p/24wgSpUe\nPWTs2mXHK69EYORauqVFyAxt2ZMVSV1Lly7F4sWLsWTJEvTq1Qtbt27FrbfeitraWlx99dVpH2e2\n1aIeJMhIjZDNsEPMpc2l2YKcrZbYLLtHvXH5CwC3241YLAaHw4FvvxVwzjlUb0wQ+TBkSBy1tTKu\nvjqBoUOjiEab9p1uKZFwOvTWpEAgYKpxA+POO+/EPffcg8svvxwA0Lt3b3z11VeYNWsWrr76akN2\njFZBgqyhEMHSi0rdbnfGxhpmCjJrKJKplthsQdY2MuEvAOLxOBQFCIUEhEIte8EgiOaipsaG2bPj\nqKmRIUmOjAYRrBkHOzJqKUKtt2XNtpXNJhKJNHlf+PeMt2M87bTT1Lls3LgRkydPNn0+PCTIGvIR\nrEI6XJkhyCyDWRRF2O32jFniZgoyayYiiqLuBYAgCDj//Eps2UJfM4LIlZkz45g504V//tOBPXvs\nGDJESPl9KYrSxBxCexwVi8V0m3CUIlqRtKoO+ZJLLsHvfvc7dOnSBb1798ZHH32Exx9/HNdff716\nn2x2jFZBKyXQ5Ew3F/MFszpc5SOSWjvEXGqJCzkP0dYw610AvP++gN/8pg2JMUHkyf33V2DChATu\nuCOBbt2arg+CIDRZa1htbyKRSLFb5B+jZ7XY3JG03vrX0NBgSR/ruXPn4r777sPkyZNRV1eH2tpa\n3HTTTbjvvvvU+xixY7QCqkPG0QgXaPwSKIoCv9+f8TF8dOhwOODxePKuXf7+++/h8XjgyVbL8AN8\n+VSutcTxeBzhcBg1NTU5/whzGfevf7XhgQcEfPklCTJB5MvatWEMGJDZo12PeDwOSZLg9XqbNOFg\n0XSmLW8zXZyMEIvFIMsyvF4vgMY1ediwYXjsscdSyo9aOFSHbAT+i9dYopM+C1kURUSjUVM7XBmN\nys1woOK3yI0+TjuukRrmsWNlzJxpjiEGQZQrL7/sxIAB6TtypYNfT1gEzNpa8vfJtOXNOmwVY8tb\nbz2yasu6lCFB/oFMnshA6jZtrtvDRsdORy5Z20bGYs+ZDb3MaSPj7t8PnHBCRc5zIwgilT//2YkH\nH4wj11MwIxfcmba8tQYR/GOs2vIudy9kgAS5CVpx1J7T+nw+01tNpnsuM3yJ042V7QKAT1LLZdz1\n6wX85Cfkb0wQhbJrVwjV1UrOYlwI2Vyc+HNpM7e8tRcQiqIgGAyWlRcyQILcBCbIkiQhFoultLm0\nqtVkNk/kQnyJc0VbOmXUi/nIEeCcc5zYtat0szgJoiXx5Zc2nHVWU2tFI7Be9WZQjC1vrSCHQiHI\nspw1l6e1QYL8A1pRDAQCeZ/TFjK2kVriQscCmkbI2rPxXAw2Nm0ScOGFTgSDLaPmkSBaAu3bl3Ye\nrZlb3uxvjEAggMrKyqIEIaUEhTM/oCgKotEoIpEIAKCiogLV1dVNWl1aAfNEDgaDKZ7IZvgS640F\nHBVkSZIQCoVUF6jKykr4/f6cMsa3bxdIjAnCRD79NIRTTsk9u5pRjDaPerCo2Ol0oqKiAl6vFz6f\nD16vF263W13PRFFEPB5X/ZAVRYEoijhy5Ag2btyIuro6VFdXW/IavvnmG1x99dVo164dvF4v+vXr\nh48++ijlPs3hhQxQhKzS0NCARCIBp9OJZDKZd9JUrrArSFmWTU8W04M9L7M7yyVzOh2TJlFbTIIo\nhGuvTWDgQAk7d9qxYYMdFa0oJ9LIlncikYCiKNiwYQOuvPJK2Gw2VFVVYdy4cejfvz/69++PAQMG\nNDF8yJX6+noMHToU559/PlauXIl27drh888/T0keay4vZIDqkFVYHRxLJrDaE5mv6QUav7Rt2rSx\n/KqW+RQzCtmSHz/egb/+tby2lAjCbJYvj2DYsPzOitMRCoXgcrksFxAz4L2QRVHEZ599hrfeegtL\nly5F165dsXXrVjQ0NODKK6/ESy+9VNBYd911FzZs2IC1a9emvY9FXsiAgTpk2rL+AYfDoSYgANZ6\nIkciEdTX1yORSKiCCFjbd5ZlbAeDQQCNr7dNmzZ5b8lv3SqQGBOECVRWJjF/vh2vvEK/J7fbjdNP\nPx0nnXQSevbsiXXr1qG+vh5ffvklfvvb3xb8/H//+98xaNAgXHHFFejQoQNOP/10PPvss+rfs3kh\nWw1tWf+A1Z7ImWqJ2W1WoG3v6XK51K35fLfkL7/cgb//nRYPgjCD4cMbm19ccUUEF10USZv0ZBQ9\no4ZSJpMXMtBYUtWjRw9TxtqzZw/mz5+PGTNm4N5778XGjRsxdepUuN1ujB8/vlm9kAES5LRYZYeo\nV/FpLAAAACAASURBVNObT/csI/CmEw6HA36/Hw6HA/X19Xm9PkUB/vQnG4kxQZjIxRcnMXVqDAMH\nJiFJtrT9p7U1vi1FcLORzunJii5dsixj8ODBePDBBwEA/fr1w/bt2zF//nyMHz8+4xyL8X6TIP+A\nFZ7IRmuJzRZkvplJuvae+by+r78GbryRErgIwky6dFHw4x8DgFPNW2HNOPg6X9Zvn8ELNN8xq6VF\nyAx+vvX19ZYI8nHHHYeePXum3NazZ0/87W9/A4Bm9UIGSJB1KVSQ+Vpih8ORtZbYrG1yPlFMEIS0\nXcXy/aGOG0diTBBmM3Jk0975fGYyg++Yxdf46nXMAqAmqZa6MKdzeio0o1qPoUOHYteuXSm37dq1\nC926dQPQvF7IAAmySr4WjDz5NtcoVJDZ+XQ0GjXUzCSf17dxo4DNmykHkCDMYNQoEWvW2CEIgNFE\naD2RBpp2zJKkxoztRCKhtvwt9FzaSrKdIZvJbbfdhqFDh2LWrFm44oorsHHjRjz77LNYuHChep/m\n8kIGSJB1yVWwCvElZuMBuQtyvr2uc3194TAwfHjpl08QREthyxYb7rgjgWuvTeCYYwp7Lm3HLLYe\nVfxQzJzJF7lUzqX1jCWsEORBgwZh2bJluOuuu/Dggw+ie/fumDNnDq688kr1Ps3lhQxQHbKK1hMZ\naOyWlQntFrHH48mruYYkSWqrOCMfeqG9ro16PjeOBXg8rahLAUGUAJ9+GkLXrtYsqaIoIhaLwev1\n6m5560XTjHTn0lYSj8chiiJ8Pp9624gRI3Dffffh4osvtnTsIkN+yEbRbllrv6g8rM1mIb7E6cbO\nBithYpnT+fS6NhohRyJA27YkxgRhNo8+6sKcObn7HOeCXu5IpnPpTE5OWpMIM0Vaz+kpEAiUndMT\nQIKcAhMqm82Wsr3DyNcf2Mi47PnTwfsx52r+oDdeNkH+z3+Anj1JjAnCTGbMiGP0aBG9e+ffpzob\nuRxHGXVykiQJyWRS/buZ59J6iWdWbVmXOiTIOujZIebrD2x0PDaOFn5b3Cw/5myCvGsX0K8fiTFB\nmM2jj1Zg0CAZAwZYJ8hmkM3JyexzaT1BpgiZAHBUsLRdrnLxB853TIZ2W9xMP+Z0gswuPFauBAAS\nZIIwm6FDRVx0UdPdNzOxqg6ZOTkxNyc2Fm+3qK2XTme3qN2i5v9/MplEOBxOMXwoF0iQOZhQsS8H\nsyTku1xZPbZV2+LZYEliR47IuPPOjpaORRDlytKl0aKMU6xsaV6kGbmeSzOnO0YwGITT6YTH4ynK\nayglqLBUA8tQZFRWVlriS5xu7EAggEgkAqfTiTZt2jTJlDQDPkKWJAkNDQ1qZvnll7c3dSyCIIBl\nyyKoq2uAgcKGgrGqL75R+DNpdrzn8/ng8/ngdrvhcrnUPB3mdieKIr7++mtcf/31WLhwIbp27aoa\n4VjBrFmzYLPZMH36dPW2eDyOyZMno127dqiqqsLYsWNRV1dn2Rz0IEHmCIfDCAaDkOXG8x0zzmuN\nkEwm1atIu90Ov9+PyspKy6Ji9nrC4TACgQAkSYLP58PMmW2xbRv1qSYIs9m7V4bLJRdNLEup8QeD\nnUu7XC643W5VpIHGcqsjR45g9+7deOyxx/Dll1+ibdu26NGjB8aMGYNFixaZNo9///vfWLhwIfr1\n65dy+7Rp07B8+XK8+uqrWLduHb755huMGTPGtHGNQILM4XQ64fV6UVlZWZTxRFFUo1O29WN1NM7O\nxYHGK0KPx4Pq6mqsXevGvHl0gkEQZtOtmwi/P4FIJIJwOIxIJIJ4PI5kMglJkixxlmtpOBwOnHba\naVi3bh2WLFmC3r17409/+hMuu+wy1NfXY9u2baaMEwqFMH78eDz77LMpSWPBYBCLFi3C448/juHD\nh2PAgAF47rnn8N577+HDDz80ZWwj0ArMwQyyWQ0yi5TNRq+zVzwet/SHpM0UB5ByLj56NHXiIgiz\neeedMAYPlgE4IEmCmviUrge1NkM5H1pC/2ot/HzD4TDat2+P8ePHZ3RgyofJkyfjkksuwXnnnac6\nPgHApk2bIIpiig/yKaecgq5du2LDhg0YPHiwqfNIBwmyDmyr2GyBlGVZ9UTWZk4nk0nd2mcz4M0u\nXC4XnE4nwuGw+iMYN46+BgRhNnPnxjBkCLuoT19GxNf78muANjvZ7IYcpYDeGltfX2+oi2CuLFmy\nBFu3bsWmTZua/O3gwYNwuVxNxi2WDzKDVuIMmOmJzISYZU57PB5TDC0ywZtdOBwOtZkI+9ErioLn\nnrNh2TI6NyYIs7nqqmTGv6fLUM5W66vX3pKHNTdqCRTLC3nfvn2YNm0a3nnnnZwaKhV7t4EEmcMK\nT2SjDUXMFGRtMxGt2QX772uv2XDTTWSpSBBm8+KLUeSTCmKk1pfZLvKP4QWalR21BIrl9LR582Yc\nOnQIAwcOTKkwWbduHebOnYu3334b8XgcwWAwJUquq6uzxAYyHSTIaShEIPNpKGLWBYDRZiIvv+zB\ntGnlV+dHEMVg9Gjzjp8y1fry293JZFJN2GR5KsU2isgXbYTctm1bU59/xIgRTRLDJk6ciJ49e+Ku\nu+5Cp06d4HQ6sWrVKlx22WUAgN27d2Pv3r0444wzTJ1LJkiQOczyRI5EIqr5Qz6eyLn+aHJtJvLd\ndwKmTSu/tnQEUSw++siGQYOsa4+ZySgiEomotxfbKCJX9NbYYDCIHj16mDqOz+dDr169mtx2zDHH\noGfPngCA6667DtOnT0dNTQ2qqqowdepUDB06tGgJXQAJclqa0xPZ6GPytWHs2pUiY4Kwin/+M2yp\nGKeDj4BZvS/Q1CjCygzvXCnWlrUe2tf4+OOPw263Y+zYsYjH4xg1ahTmzZtn+Tx4SJDTYFSQtZ7I\nhfacNnoRoI3EjdowPvwwJXARhNm0by/j0CEbVq8OY+DA0jKOyGQUUQoZ3trnbWhoKIqxxL/+9a+U\n/19RUYGnnnoKTz31lOVjp4MEmUO7ZZ2pDpnPnAZgmidyNkHmI3G73Z5TJL51q4Df/IY+coIwm0OH\nbLj00mSzi7FRYwmrMrzznS9PfX19WVovAiTITeANJtI5IvHntWZZMWb7YmeqYTbC4cPAj39MzT8I\nwgrOPVfErFnx5p5GQZiR4Z2L5SJ7fq3zU7Ei5FKEBDkN2gg53/PaXMZj4/BoLwDyicRFEejcmewU\nCcJsTj1Vws6ddnTooKBz5+YvNTLbejGfDG8AupG03pz0cmYCgQAJMtEIi4xtNpv65WYlTLme1+Y6\nLnD0B6W9ACgkEq+sJDEmCLP5z3/q8fOfV6GmRsGUKYnsD2glZMrw1kbS2TK82fPxzxMMBstWkFtG\nO5dmgH1JmPmDoiiWWjHygpxMJhEMBhEKhVT3J5/Pl7MYyzLgdpMYE4QV/OxnXnz1lYClS7/DSSeF\nkUgkIIqiZT3wjWB2hGwU3nKR7R5mslyMRqMIh8OQJAmSJCGRSODf//43vvvuOySTSdPPkGfNmoXB\ngwfD7/ejQ4cOuOyyy7B79+6U+5SC/aKQQ2lP8+/HFAHWUzoUCkEURfW8thg2jN9//z3sdjskSYLd\nbofX682pzRuPogAeD4kxQZjJ5Zcn8MorjbkYNTUyli1rQO/eSTUqZBR6tpovLOnTCh91s+AzvBOJ\nxp2Fb775BoMGDVIDkIsvvhinn346Tj/9dPTv37/g3tYXXnghxo0bh0GDBkEURdx999349NNPsWPH\nDng8jWWgN910E1asWIEXXngBfr8fkydPht1ux/r16wt+zT+Q9cMnQdYQDAYRDofV/887IlmFXulU\noRcA99xjx2OP0YkEQVjF2rVBnHaarIovv23LzleLLdKiKCIWi5W0IPOEw2E16Pjkk0+wdu1aLFiw\nAF27dsUnn3yCWCyG4cOHY82aNaaOe/jwYRx77LFYt24dzjrrLASDQbRv3x5LlixRO3Xt2rULPXv2\nxAcffGBWc5CsHzSt2BoEQYDb7YbT6URDQ4OlY2lLpwRBgMvlQkVF/pGtKAI33ODASy9RvTFBWMUl\nlyTQp48ISToapzBxFQQBTqczrUiLophXAlQulGqLTC0sIHS5XGr0unTpUnzwwQeQJAk7d+5ENBo1\nfdz6+noIgqC26Ny8eTPZL5YiXq83xRPZiibt6UwnzLgAuPJKB958k8SYIKxk0KCEmvzJ538w8eVh\n92EiDUC3lChdAlQuIt1STCWAzE5PrJlJnz59LBl32rRpOOuss9R2mgcOHCD7xVLGaKOOXOG9ibWm\nE4UaTBw5AhJjgrCQDh1k/PrXEVxxRRTJpJQiKnyky5NJpNO1uDSapdxSIuFM8K/BKi9knptvvhmf\nffYZ3n333az3JfvFEoF9CGZlTKbzJtaOmb/DFHDccZTERRBWsWFDFKedpqCxOMWXUo/L/iWTybxF\nms9Udjgc6pZ3Ln2o2XO2BKFOZyxhZcnTlClT8NZbb2H9+vWora1Vb+/YsSMSiQTZL5YaZnsiZ/Mm\n1o6d7wXAMcdQFy6CsIrFi+M/iPFR+HpcvrNVLiLNrzHscdoxmEgDUEU6Ux9qAIjFYs1iFpEL6bas\nrYqQp0yZgtdffx1r165F165dU/42cOBAOBwOsl8sZQr1RDbqTVzoeL/9rR2RSOn94AiitXDppVL2\nO0FfpAGoAqon0gCanBfnK9LxeLzZzSKMUkynp5tvvhkvvfQS3njjDfh8Phw8eBAAUF1dDbfbDb/f\nT/aLpUihnsi5ehNrx851vGXLbJg1iz5GgrCKGTOS2e+UBSaGmURaL/uaF89sIs2XX3m93pzMIppT\npPkxGxoamkSvZvD0009DEAScc845Kbc/99xzmDBhAgCyXyx5chFI1mErGo1CkqS8e13nIsgffwyM\nG5df4xCCILIzZ04Cv/ylmP2OeWBEpPUi6UwizZoZJRIJVaT5xxgxizDb0Skd6Zye+vbta/pYRo4C\nyX6xxDEqyFpv4nybieRyAbBvn4ghQ3w5j0EQRHZqa2W8/nocvXoVt4xIT6S1Z9La9px2uz0lSYy1\nqeTFl8FH0na7HRUVFYYdnfT6UBeC3pZ1OTs9ASTITcjFE5n3Js6WsJXL2JlS7SVJwr33Ak88QWJM\nEFYxbJiMb78Vii7IerDyKP4in4l0MplU208y4vG4GhVrTSC0As2eX0+ktRne/Dh6Ip1PQxM9p6dy\n9UIGSJB1yeaJXKg3caZx08GytcPhOJ544riCxiEIIjNLlzqwdKkD27ZF0aNH84uyHqIoqsEAa5WZ\nKZLWCihwNCtcW4alFel0Xcf0RNpoa1A9q9lydnoCSJAzohVkvYQtj8djqvcoG4f/32xMWVbwP//T\n3pSxCIJIz1//GkOvXgq6dSs9MWY9DViXPz4YSBdJa8+lGdqOYAxZltOKtFmtQdN5IVOETKTAf1nY\nF9Msb+Js4wJHr1q1Y95wgx8bNtBHRhBWc+yxKDkxZr3vE4kE7HY7KisrsyaNZtrutkqk2X0ztQa1\n2+3q8/DNTIoZIc+bNw+zZ8/GgQMH0K9fPzz11FP40Y9+VJSx00GrewbYlysYDKqtLisrKy1zf2KC\nzDxDWZJYZWUlxo514+23qS0mQVjNK6/EMXBg83ka68FHxcxfuJBcFT2R1svwZuQr0nxrUK1IM4Fu\naGjAgAED0KNHD7Rv3x5vv/02zjrrLJx66qmWrbVLly7FjBkz8Mwzz2Dw4MF4/PHHMXLkSOzevRvt\n2rWzZEwjkP2iDslkEslkEqFQyBRv4lzGZQYTdrsdHo8HTqcTzz1nx803U3kTQVhNKBRBKTW10kbF\n+ZRSFjK2VqT5+ud0Z8X8uTQPfz+bzYZIJAJBECCKIp555hls2bIFq1atUu1vPR4PLr74Yrz88sum\nv7Yf//jHGDJkCObMmaPOuUuXLpg6dSruvPNO08f7AbJfzIdIJKJ+WQDA5/NZ6onMd/UCGuvhvF4v\nBEHA//6vA88/T5ExQVjN22/HSkqMWV8Dlq9SqEd6rvA1zAyjkTQT3XROWOx/s2Dntttuw969e7F+\n/Xrs27cPn3zyCTZv3mzJxUcymcTmzZtxzz33pLzWESNGYMOGDaaPlwskyDqwK1GHw2GpJzKzYYxE\nIuqPLhaLweFwQBAE1NWBxJggisSwYaWxTc2qOJgRjRX5KvliVKT5SFrbEUyWZVXEWWY4AGzbtg1A\nYzvLs88+G2effbYlr+Hw4cOQJKmJaUSHDh2wa9cuS8Y0CgmyDh6Px3JPZN6GkXX1stlsiMViUBQF\nkQjQtSu5NxFEMdiwIdrcUwCQGhWzI6vm7jmdDaMira2X/uijj7B+/XoMGDAAO3fuxKOPPoozzjij\n6JaH/Jyb+70mQc6AFZ7IkiQhEomoV7/arl6NZyoK2rUjMSaIYtGtWz1CIbsqLMXu7VzKUXE+aEWa\nrXuyLKvWkl988QUWLFiA+vp6AEC7du3gdDrx29/+Fv/zP/+DXr16WTK3du3awW63qwYTjGJbLerR\ncj/xImCmIMuyjHA4jEAgAEmSUFlZiaqqqiZn04IgYMYMT8HjEQRhjDffDKsGMCybORQKIRgMIhQK\nqd34+Mxgs2DHVqFQCKIowuPxqE0+WgOsj0IoFALQmI/j8/nU0lEAuPXWW7F48WLceOONkCQJ8+fP\nt3Tr2Ol0YuDAgVi1alXKPFetWoUzzzzTsnGNQBGyDrwnMmCsMXk6WJYkS9jyeDxwu91pr7yfftqL\nF16g6JggrKZjRxkLFyZwzjkCBOHob06vdzS/3cpH0YVE0qz7niiKcDqdhl3hWgqstTA7lmPr3sGD\nBzF16lTs2LEDr7/+OoYNG5by/ullaJvN9OnTcc0112DgwIFq2VMkEsHEiRMtHTcbJMhZYN1ocoVd\n+fIddbJtQ736qg2/+U1VIdMlCMIgr74aR//+TX/bRpppFCLSvDMca71rdUllMWFrXywWg81mU6tU\nFEXBsmXLcNttt2HMmDFYvHgxqqqarnf59MTOlSuuuAKHDx/G/fffj4MHD6J///5YuXIl2rdv3k6I\nVIesA/tCAY2t3BwOB3w+42YOfMKW0+mE1+vNmr7/1ls2/PznredHSRClTiAQQSHVjHodr7QuTFqR\nbkzYPLo2mNl6txSQZTklWZVFxd9//z1mzJiB999/HwsXLsTIkSNb1es2CNUhF0ouloi8+5Pdbv//\n7Z15eFT1ucc/J5M9gbATQMCwCUQhAZJYsFyU+pByL1LcQBGkvVwCMbJIZSsIFQlg0YAIAkFCUQFR\nC7hgqUVka8gikJSyuiJggIBAMhOSWc79I57xzGQmmZDJnJnJ7/M8PD4eI3nPLOf9vdv3pVGjRi6d\nfF98UUd6ungrBAJPkZBgrpMzBtciaUfbmKByRaKiYOUPOIv6ZVnm73//O88++ywPPPAAhYWFNG3a\nVGtzvRbhBWrAFYdsv/0pIiLC5SH+8+cRzlgg8CA9elgYOdJc8w/eBo6ctFr2UnkmVFRUWB21u2rS\nWmFfC1ei/ps3b/KnP/2Jjz/+mDfeeIMRI0b41H1pgfAEDnB1J3Jdtz+VlECXLqKBSyDwFEePltGt\nm2eqb85qqYBDtSv7mnRgYKDNXmNvc2ZKVKw0rKqj4v379zNp0iTi4+MpLCzUfJzIVxAO2QnV7US2\n38QUHBxc61EFiwUGDBA1Y4HAk/TvH0pRUVmd09U14azDWEFZR6guaTkS0lCePeq5Xm9w0uq5aXWH\nuMFgYMGCBWzZsoUVK1bw5JNP+lXneH0jHHIN2Dtkk8mEwWCw2cR0OzrXL7yg48wZ8UEVCDzFmDEm\nJk0y1qszVrJm5eXlVaLimvAVJ63UiqFyjDM4OBhZlsnLy2PChAnceeedHDt2jPbt29erHf6IcMhO\nsI+Qle7BiooKAgICiIyMvG1Zu8zMAJYtEy+9QOApZs82MmWKEQdTNm5DXSsOCQkhJCSkzs7Rm5y0\nsgTHXk2svLycJUuWsG7dOtLT00lJSRFR8W0ivIILyLLM9evXrd2DdfmiHTokMWWKeNkFAk/xwQe3\nSE6uv8UR6hWJymG9PlckauGkHWlsQ+VCiAkTJtCkSRPy8/Pp3Lmze2+2gSGOMU5QUk/qlYhRUVHV\nqmy5giTB2LHesVVGIPB3Ro821aszNplMlJaWUlFRQUhISL07Y2coDjo0NJSIiAgaN25Mo0aNCA8P\nt45XKZvlSkpKKCkpQa/XW+vA1TWulpWVYTAYrKOcwcHBmM1mli1bxpAhQxg7diyff/65cMZuQAiD\nOOHatWuUl5cTGBiIyWQiKirKLV+0r7+G2FjRWS0QeIKrVw2Ehrr/71VHxcq6Vi0ccW1QJCntxUyc\nRdLKPdrvYz5z5gwpKSnIskxWVhaxsbEa35nPIIRBbhelWUGZp6urtqrFAhMnBrJpk3d/aQUCf2HY\nMFO9OGN1+lbtqLwdJU2tTnc7ctLqdDfA4cOHyc3NJT4+npMnT5KRkcFzzz3H7Nmz/Ury0xsQEbIT\nFL1as9nMjRs3XFbdckZREdx5p4iMBQJPUlJiwF39ReqmJp1O51dbmRSUKRJZlq1Nq5mZmSxZsoQb\nN24A0Lp1a371q1/Rt29fHn30Ubp3766x1T5Djac2//o01QPuWMG4cqVOOGOBwMN8/PEttzljo9FI\nSUkJRqORsLAwIiIi/MoZK+lpvV5vbUxTGlijoqIAmDZtGm+//Tbjxo2jtLSUV155hcLCQo0t9y9E\nytoJ9isYb9chnz8Pzz8vXmaBwJPcf38FCQllmEy/dBXfDmoBDPWojz9hNputIkfqca0ff/yRZ599\nlq+++opPPvmE/v37V1mTaDbXjwRpdRw4cIC//OUvfPnll/z444/s2LGDhx56yOZnXnjhBdavX8/1\n69cZMGAAb7zxBl26dPG4rbXFvz5Z9UBdHPJnn0lCGlMg8CBjx5ZTXHyVrVuvYzZXRnw3b96kpKTE\nqiOgbmRyhiJ7WVpaislkIiwszO9S1EpUXFpaCkBkZCShPxfd33vvPZKSkujUqRNHjhxhwIABVQ41\nim63p9Hr9cTFxbFq1SqHB62lS5fy+uuvs3btWnJzc4mIiGDIkCEOl3x4GyJ0c4HabHxSkGWZw4cb\nVNldINCc1avNSFIY4Hg9otFotP6ss6UO6mUJ/hwVK9Ke6qi4uLiY5557jry8PLZs2cJvfvMbr2tY\nS05OJjk5GXAcKK1YsYJ58+YxbNgwADZt2kTr1q3ZsWMHjz/+uEdtrS3+9SlzI64umHBEUZGJsWNh\n0aJ6aPEUCARViIiQuXjRgNp3KBFcSEgI4eHhNGrUiMaNGxMREWHVXlbUtUpLS62RdElJCSaTidDQ\nUL+MisvLyyktLUWWZetrAfDJJ5+QlJREZGQkBQUFPPjgg17njGvi22+/paioiMGDB1uvNW7cmKSk\nJLKzszW0zDVEhOwCymLxmjCbLbz1lok5c8K5dk2MNwkEnqBZM5mjR8v4ufeoWpztMDYajZSXl9sc\nvJWVquqtS1ovdagLzhZe3Lhxg5kzZ/LZZ5+xZs0aHnroIZ+8P4CioiIkSaqyXap169YUFRVpZJXr\nCIfsBPsIuTqHrNRiUlKC2batHsVyBQJBFcrKuG2NavWKREUaNygoqNZSlN4cRTtbAynLMnv37iU1\nNZV7772XwsJCWrZsqbW59YKyl8DbEQ7ZBZylrJUPuiIo37mzaOASCDzJzZsGblcgq7oViXXViw4M\nDPQKB6AsxbG/R71ezwsvvMB7773HypUrGTVqlFfYW1eio6ORZZlLly7ZRMmXL18mPj5eQ8tcQzjk\naqhuJ7LJZEKv12M2mwkKCuL48SiyskKQJBlZ9v0PtkDgC9yOM7aPil1dkWjvpNUqV4qIUHl5uc3P\nO1rq4AmUNHxZWZlN5F/ZbHqYlJQUunXrRkFBAe3atfOITZ4gJiaG6Oho9uzZQ69evQC4efMmOTk5\nPPPMMxpbVzPCIbuA2iErp2pFw7ZRo0b88EMQycnBmEzCEQsEnqK01FDr/6e6qLi2OJOitI+k1Z3d\nnnDS6i7xoKAgwsLCkCSJW7dusWjRIjZs2MDLL7/M//7v/3p1qt0Zer2er776yvpM/uabbygoKKBZ\ns2a0b9+eqVOn8tJLL9GlSxfuvPNO5s2bxx133MHw4cM1trxmhEOuBnWErKR+1Kdq5UtosZhtnHFU\nlMyNGzV/yVq1krl8WThxgeB2+PZbiU6dXBstVLqLy8vLbeqo7kadulb/blectNI8poxf1RYlKlY2\n1Kmj4oKCAiZMmEDz5s05cuQIMTExdb9ZjcjPz+f++++3HoimT58OwNNPP82GDRuYMWMGBoOBlJQU\nrl+/zq9//Ws+/fRT69Yrb0ZoWVeD0WjEbDaj1+utQ+WhoaGEhVXOOZrNZjZu1DFzZgjl5fDEE2am\nTr1Fq1Z6tmwJYf/+MHbtCqZVK5ktWyq46y4L778fyODBZl58MYh33xXnIYGgtnTtaiErq5y4OBlX\n/JYzJSotcTQjre5TcTYj7Qy1opiyhjEgIACj0ciyZct4/fXXWbBgAWlpaV6/lcqPqfFDJxxyNZSV\nlVFSUmKVh2vcuDE6nQ6LxYLFYkGWZebPD8FgkJg8uYJWrX4Rng8NDeXcuSBu3YKePX956UpL4Z57\nwtwSGQcHy1RUiAhb0LD461/LefTRmiUb7aPi8PBwr3ZGt+uklVoxYN0+BXDy5ElSUlIIDAxk48aN\nYgmE9ojlEnVBSf0oETFg8yUJCAhg4UIj6eklNG160yqxp6TDOnWSbZwxQF5eADqdzGOPmXjvvXJi\nYhwLjiQmmlm9upxz5wzo9ZV/rl418OqrFRw7VoZeb+DTT8uJi7Nw8GAZU6caHf49AoG/8bvf1eyM\nTSYTpaWllJeXExISQmRkpFc7Y3BNyERxvoqQyc2bN63R/9WrVwkMDMRsNvPaa68xePBgfve733Hw\n4EHhjH0EESFXg9FoxGQyWb/cSo0nMDDQqvKjLPC+3VSY0QhmM4SGVu5MvnwZWrfGpVScmsuXW3fC\npQAAHl1JREFUYd8+HZmZgbRpI5OYaOHiRYnoaNn6O4xGMBoljEYoLpa4dQsaNZLZsEHsNBX4Bs89\nZ2ThQueHT0UTQGm6DAsL83pHXFssFgsVFRU2Hd2nTp3igQceoEWLFjRq1Ai9Xs+cOXMYOXIk0dHR\nGlorUCFS1nXhxo0b6HQ6a+rLPoUElVFycHAwQUFBPtmx6IzK3a/l/Oc/Zu66y0xYWJjD2crr12Hz\n5kCuXpXQ6SpT6BUV8M47gfTsaaF5cxm9XsJgAL0eDAaJ06crX6egIBmjUaTcBZW0bClz5Yrzz0PT\npjLHj5fRpInj/67IYFosFmvqVutasbtRHzjUOts//fQTr732Gv/617+4cuUKxcXFXL16FYD/+Z//\n4aOPPtLYcgHCIdeNTp06ERgYSL9+/UhISCAmJobNmzcTHh7OkiVLCAgIsNaS4ZcOS7XUnq89ENSd\nmnWJ/N2F2QwVFZV/yssrI/yKCigpgWnTgikrgxdfLPt5BlTm1i0LRqPEkiWN+PFHHQMHGtm927Xu\nyvBwmf/6LwsVFXDxosTJk5UHh6ZNZVq0kDl71n8OXL7Gjz9WCoBERFT9bw0hKobKA4fBYECWZZsD\nx4ULF0hLS+O7775j48aN3HvvvQB8//335Ofno9PpGDFihMbWV0b28+fP55133qGoqIi2bdsybtw4\n5s6dq7VpnkI45Lqg1+s5evQo+/fv5+233+bUqVM0bdqU/v3706lTJ5KSkkhISCA6Oto61qAIBCjU\ntltSS9Qzmr665UYZMVHeB0eNMYD1PQoLC6OsLIjgYAlVqwAAxcWVpYTIyMp/NxorhSiKiiTatpV/\n/n2Vh4affoL09CCeecZESMgvh4jKP5L1QHH0aAC7d+uIjq4cjWvaVGb7dh3Dhpnp08eC0Vj581u3\n6vjuuwAGDTJjNMKhQ/7nYFwlN7eM2FjHjx+lpmrvpPwJ+wOHsvDCYrGwbds2nn/+eZ566inS09OJ\ncHRi8RLS09NZvnw5mzZtomfPnuTn5zNu3DjS09NJS0vT2jxPIBxyXbly5QpxcXFcvXqVadOmMW7c\nOI4fP87hw4fJycnhyJEjNG3alH79+pGYmEhCQgJxcXEEBwc77Jb0Rok99Rc+ICCA0NBQG8lAX0fp\nXjUajVV2ovpSVqOsDHbu1HH9usSgQWbuuqty7MdiUfoD4Nw5uOOOCmS5UqEpNDSUwMAgTKZfDgUV\nFXDrlsR330lcuiRx82blP4uKJDZurBzFCwmRKS//5XUID5cxGJy/Lo89ZmLvXh3Fxe5/7UpKDNif\nC9VjPr56eHQFZ2n4y5cvM3XqVAoKCli/fj0PPPCA135uFYYNG0Z0dDSZmZnWa48++ijh4eFs2rRJ\nQ8s8hnDI7uDll1/mscceqzJMr6R3//3vf3P48GEOHz5MXl4e3377LbGxsSQkJJCQkEBiYiIxMTFV\nxAGU116t3qM0jHniy2Wfnm4IEUZAQID14a3OaKizGu4SavA06gyHehbVncgy7N0bwJIlQQwcaGH6\ndGOVzIIsw/HjEp9+qmPXLh2RkfDxx+V8951EbGwYs2YZmTfPqPp5GaPRwq1bZm7dsnDxoszZsxIH\nD4Zw86bEihUlBAX9cmAym83Wz2xYWBhBQUE+8f7UBvXIljoNL8syH3/8MZMnT2bYsGG88sorRLmy\n5soLWLx4MZmZmezevZuuXbtSUFBAcnIyGRkZjBo1SmvzPIFwyJ5GlmWuXbtGTk6ONYrOy8sjICDA\n6qD79etHv379iIyMrOKkFeyjNnc/WJUO8fp8eGtNbQ4cajUlxUk7mgFV3hNvWsFnP2+rNOD5MjXN\n5AIEBQVZVyl60/tRV5wJmfz000/MmDGDL774grVr1/Lf//3fPnXPsiwzZ84cXn75Zauew6JFi5g5\nc6bWpnkK4ZC9AbPZzOnTp8nJybH+OXHiBJ07d7aJou+66y4AmxqoumHM0V7W2qJO9fnLw9sR7qiH\nq52Co/fDkZP2NOqUptYNePWFelECYP28Ons/fGEloiOcCZnIssyePXtITU1l4MCBrFy5kubNm2tt\nbq3ZunUrM2fOZNmyZfTs2ZNjx44xZcoUMjIyGDNmjNbmeQLhkL0RWZbR6/Xk5+eTnZ1NTk4Oubm5\nlJSU0KdPH6uDTkhIoEWLFlWalBRq0zBmv+EmJCTEb9PT6oeau+vh6oyGfQOfJ0sPlWNpvyjD+Wtn\nsf2iBPtMjn2GqT4OsZ5AfYBUH6xKS0uZO3cu27dvZ9WqVTz22GNeew810aFDB+bMmcPEiROt1xYt\nWsQ777zDiRMnNLTMY9T4xvlfaOQDSJJEZGQkgwYNYtCgQUDlA/bcuXNkZ2dz+PBhli9fztGjR4mO\njrZG0QkJCfTq1cuqxqM4BaVRyVnDmDqKCg4OJiQkxOeiB1dQd9zWV7ToaAWfvZNWLw5w9H7UxaaG\nVPd3tD7QHmcrEdUHWGfbljzZr+EM9UFZvfRClmUOHTpESkoKd999N4WFhbRp00YzO92BwWCo8lor\n3eKCSkSE7KUokd6xY8esae7c3FzOnz9P7969bVLdd9xxRxWnYI/SceuPDTDqKCowMJDQ0FBNo0X7\n+qfJZHLLInv7+/TXzmJls5rS36CsD7xdHB2a6rLIwV04WwVZVlbGwoUL2bRpE6+88gpPP/20X7zP\nv//979mzZw9r1qwhNjaWI0eOkJKSwvjx40lPT9faPE8gUtb+hCzLXL582drRnZubS35+PmFhYTYO\numfPnmzbtg2z2czYsWNt9jlD/TeMeQr7NLwzNTFvwD61qj401bQj19F9+tNYmoIn79OVRQ719R2x\nj4qVz60syxw9epQJEybQtm1b3nzzTTp27Oi236s1er2eefPmsX37di5fvkzbtm158sknmTdvnl/2\nsThAOGR/RnmonDhxwuqk9+zZw4ULF5AkieTkZIYNG0bfvn3p2rVrFdEMX6y1Kdin4euyaF4LHO3I\nddQfIEkSFRUVyLLsk/fpKs6iRU9in+qurmnsdg9+6uhffZ8VFRX85S9/YfXq1SxcuJDU1FSfPSgL\nnCIcckNBr9czcuRIPvnkEwYNGsSYMWO4cOGCNdVtNBrp27evzehV06ZNXWoY86axEnWXuL81M9lH\nberaJ1Dl0OQPD2xvH9mqS2ZDjX1NXB39nzhxggkTJhAWFkZWVhbdunXzyL0JPI5wyA0FWZYZP348\nQ4cO5eGHH7Z5MFgsFr755htrw1heXh6FhYW0b9/eJtUdGxtrFcyoTmFMizEf+3Smv9bDoWpzmiKG\nUZ2gjK9kNtR4Q1RcW2rKbAQEBFRJdSsd8UqnuFITN5lMvP7667z88svMnj2b6dOne9VhROB2hEMW\nVEV5QBw5csQqXpKbm0txcTHx8fH07duXxMREEhMTbXS6nUUI9a1o5QkFKm/AFTlIV1Pd3qyd7u1R\ncW2pST9doaioCJ1OR6dOnfjmm2+YNGkSZWVlZGVl0bt3bw0sF3gY4ZAFriHLMhcvXrTWoh3pdPfr\n14+4uDhCQkJsVK3qqxnGkeSlLz+4nWGfzqxt9F9Tg5I3pbobgpAJ2B4ilfubM2cOWVlZNGnShLKy\nMhITE5k6dSoDBgygdevWGlss8ADCIQtuj9rqdCtOwR0NYw1l1hbqL/qvqUHJ0018DWVFImCjKqZ8\ndgEKCwtZsmQJP/30E2azmTNnznDlyhUAXn/9dZ555hnNbBZ4BOGQBe6jOp1uZWd0QkICffv2pXHj\nxi6lVe0bxhpKetrTaVtHghk11T7d5aSdbSzyN9TqaeqSg8ViYfPmzcyaNYtx48bx0ksvER4ebhUD\nys3NJT4+ni5dumh9CwBcvHiRmTNn8umnn2IwGOjatStZWVn06dNHa9N8HeGQBfWLKzrdCQkJdO/e\nHUmSqm0YU6Jsf561Be9J27qyO7ou5YeGGBXbb6C6dOkSkydP5sSJE2zYsIGBAwd69WHk+vXrxMfH\nM3jwYCZNmkSLFi04e/YsnTt3rrLtTlBrhEMWeBZ7ne7c3FxycnKq1ek+deoUUVFRNmvkfHUFYnX4\ngoOyLz04msV1ZXe02kH5e1SsvKfqqFiWZXbs2MG0adMYMWIEy5Yto1GjRlqbWyOzZs0iOzubffv2\naW2KPyIcskB77HW6c3NzOXr0KC1atKBRo0acPHmS0aNHs3z5coKDg22iNm9tTqoNSqrYVx2U/VpK\nRws11KNw6jnx8PBwn3mfaovJZMJgMFR5T69du8Yf//hHDh48SGZmJsnJyT7zXsfGxpKcnMwPP/zA\nvn37aNeuHampqYwfP15r0/wB4ZAF3kd5eTnLli3jpZdeIiQkhN/+9rfk5uZy4cIFq0630tndvn37\nKrVPX5rD9Uf96Zp2R0Pl4UnZV+wtojLuwj7ToRw6ZFlm9+7dpKWlMXjwYFasWEGzZs20NrdWKDPS\n06dP59FHHyUnJ4epU6eybt06nnrqKa3N83WEQ66OVatWsWzZMoqKiujduzcrV64kISFBa7P8njNn\nzhAXF8fEiRNZsGABjRs3dlmnOz4+nvDw8NtqGPMkDUV/GmwPHcr2JW/cHe0OnDWolZSUMGfOHD7+\n+GNWr15dRZzHVwgJCSExMZEDBw5Yr02ZMoX8/HwOHTqkoWV+gVi/6Ix3332X6dOns27dOhITE8nI\nyGDIkCGcOXOGFi1aaG2eX9OtWze+//57WrZsab0mSRKtW7dm+PDhDB8+3KFO97Zt2zhz5gzdu3e3\naRhT63QrkpP2KyldqXu6C19UoLod1ONpgE0zE1TdHa10lYP3rUGsCXVXvE6nIzIy0tqIuH//fiZN\nmkRcXBwFBQVER0drbe5t06ZNG3r06GFzrUePHvztb3/TyKKGRYONkO+9916SkpJYsWIFUPmFa9++\nPZMnT2bGjBkaWydwhCzL3Lhxg7y8PBuFMbVOtzJ+peh0q9OqCvU14tNQhEzANip2dTytNmsQvSnV\nbTabMRgMVbriDQYDf/7zn9m8eTMZGRk89dRTPl+OGD16NOfPn7dp6po2bRp5eXkcPHhQQ8v8ApGy\ndoTRaCQ8PJwPPviAhx56yHp93Lhx3Lhxg+3bt2tonaA2qHW6FQddUFBAhw4dHOp011fDmL3+tL8q\nUFW3JOF2/7762B3tDuxnxcPDw61RcX5+PikpKXTs2JH169fTvn17j9lVn+Tn5zNgwAAWLFjA448/\nTk5ODikpKWRmZjJq1CitzfN1hEN2xI8//ki7du3Izs4mKSnJen3mzJns37+f7OxsDa0T1AVnOt1X\nrlwhPj7e2iyWmJhImzZtqszhOmoYqy6laq8/HRoa6nWjTO7idqLi2/097tiwVBfUZQf1Aau8vJwl\nS5awbt060tPTSUlJ8fmo2J5du3Yxa9YsvvrqK2JiYpg+fTp/+MMftDbLHxA15Nogy7JfRjUNCUmS\nCA8P57777uO+++4Dqup0r127lpSUFKtOtxJJx8XFERoa6nQFor2DNpvNTuun/oR9g1p4eHi9Nqgp\njWHK73C0UMPR++KOEoT6XgMCAoiIiCAwMBBZljl+/Dj/93//R1RUFHl5eV6jrOVuhg4dytChQ7U2\no0HSIB1yixYt0Ol0XLp0yeb65cuXhci7HyJJEu3ateORRx7hkUceqaLTnZOTw6ZNm2rU6bZvGFP+\n7uDgYL+LkhS8oUFNnbpWcJTqtm/kq20Jwtm9mkwmli9fzquvvsrcuXOZNm2a32ZBBNrSIFPW4Lip\nq0OHDkyePJnnn39eY+sEnsYVne64uDhycnLYsWMHf/vb32xWUyrUpya0J7GPFH2hQc3RWkpXZtar\nu9czZ86QkpKCxWIhKyuLu+++W7P7E/g8oobsjG3btvH000+zdu1a69jT+++/z6lTp2zGcQQNF7VO\n986dO9m9ezcVFRUMGjSIdu3aWaNoRafbUcOY1o1Jt4M3RMXuwJXd0QEBAdZGsqCgIKswhtlsZu3a\ntbz00ktMmzaNOXPm+O0cucBjiBqyMx5//HGKi4t54YUXuHTpEnFxcezevVs4Y4EVnU5Hz549+eCD\nD9i5cydJSUlkZGRQXl5OdnY2n3/+OYsXL7bR6VacdMuWLW2cQEVFRa0bxjyNfVexUj/1VWpKdRuN\nRpta9Pz58zl+/Dj33HMP+/btw2g08s9//pO+fft6xfsj8H8abITsKyxevJjt27dz6tQpwsLC6N+/\nP0uXLqVbt25am9Zg+Oc//8mpU6eYNGlSldqhM53u6Ohoa6o7MTGRXr16ERQUVKPCmHqZhidxNmvr\njzjqFjebzWzatIn333+fwsJCrl+/DkDHjh1JTExkypQpDBgwQGPLBT6OSFn7OkOHDuWJJ56gX79+\nmEwmZs+ezfHjxzl58iRhYWFamyewQ6lHHjt2zGbs6vz58/Tq1csmilZ0uqubwa1vhTFns7b+iDNl\nMagchXz22Wc5e/Ysb775Jh06dCA3N9e6rWzu3LkMGTJES/OdsnjxYv70pz8xdepUXn31Va3NEThH\nOGR/o7i4mFatWrF//37rWI/Au3FFpzshIYE+ffoQHh5eZTZawZ3jPeA9e5k9gXpeXD1DLcsyH3zw\nAc899xwjR45k6dKlREZGam2uy+Tl5TFy5EiioqK4//77hUP2bkQN2d+4fv06kiT53BaZhkxddboB\nl8Z7XG0Ys99WpOgy+yuKshhURsXBwcEAXL16lWnTppGbm8vmzZt58MEHfepAUlpaylNPPcX69etZ\nuHCh1uYI3ICIkH0IWZYZNmwYJSUlYoG4n1GdTre6YSwhIYFmzZq5NN7jqGHM2Q5ff0RRbVNU1JTV\nl7Iss2vXLiZPnsyQIUNYvnw5TZo00drcWvP000/TsmVLli1bxv333098fLyIkL0bESH7E6mpqZw4\ncUKsQfNDJEmiSZMmPPjggzz44INAVZ3upUuX2uh0KzKgsbGx6HQ6m2UajpSslFS4TqcjLCysQUTF\nsizbqKjduHGDmTNn8o9//IM1a9YwfPhwnzyQbN26lWPHjpGfn6+1KQI3Ihyyj5CWlsauXbs4cOAA\nbdq00docgQcICAigS5cudOnShTFjxlTR6c7OzmbFihVVdLoTEhJo27at1UFfuHCBpk2bWju3LRaL\ndY2gp1ZSegp1Ot4+Kt67dy+pqakkJiby73//22dHHM+fP8/UqVP57LPPxGy0nyFS1j5AWloaO3fu\nZN++fXTq1ElrcwRehL1Od05ODkeOHKFJkyb06tWLsrIyvvjiCxYuXMizzz4L4JGGMS1wlo7X6/XM\nnz+fbdu2sWLFCp588kmfuzc1O3fu5OGHH7ZunoLKHgOlp6C8vNyn78+PEV3Wvk5qaipbtmzhww8/\ntJk9joqKIjQ0VEPLBN6IMtqTkZHBokWLqKioYODAgXzxxRfcfffdNqlu5XCndtCOFMaU0Stvfcjb\nN6mFh4dbo2JlfWDXrl3JzMykXbt2WptbZ/R6Pd9//73NtXHjxtGjRw9mzZpFjx49NLJMUAPCIfs6\nziKVrKwsxo4dq4FFriFmI7Vjz549/OY3v2HEiBGsWrWK6OhohzrdkiTZNIv17duXxo0b28hNOlp9\nqBYv0dpJq0e31FHxrVu3SE9P580332Tp0qWMHz/ebxeAAKKpyzcQTV2+jhKx+BJ5eXlkZmbSu3dv\nrU1pkDzwwAN8/vnnDBo0yOowmzdvbrNWT63TnZOTw/z58/nPf/5D586dbcRL1DrdzhrG7JdpeAK1\noIl6dEuWZQoKCpgwYQLNmjXjyy+/bBBlHq0PRgL3ICJkgVspLS2lb9++vPHGGyxcuFCc2n0EWZbR\n6/Xk5+eTnZ1tVahSdLrVDWOtWrWycdDqsStJkmwcdH2kup3JfBqNRpYtW8bKlSuZP38+kydP9utO\ncoHPIVLWAs8iZiP9B1d1uu+55x6Cg4NdUhhTxEtux0lXJ/N56tQpJkyYgE6nIysri549e7rtdRAI\n3IRIWQs8h5iN9C8kSaJjx4507NiRUaNGOdTpXrduncs63Y4UxlxtGFOvhFRHxWazmdWrV7N48WL+\n+Mc/MmvWLJ/eUCVo2IhPrsAtiNlI/0eSJEJCQkhKSiIpKQmoqtP91ltvMWXKFMLCwmyi6D59+hAZ\nGWnTMKZEu+C8YUw5BNy6davKSshvv/2WSZMmcePGDfbu3UtcXJyopQp8GpGyFrgFMRspABzqdOfm\n5lbR6e7Xr591jE9dj1Y3MSrqYrIso9PpiIiIsDaYbdy4kXnz5jFx4kTmz58vRgAFvoCoIQs8g5iN\nFDjDmU53RUUFffv2tXHSzZs3x2g0kp2dTdeuXa2bl9asWcNf//pXevfuzblz57h69SpvvfUWAwcO\nFAc9ga8gHLJAO0RTl8AZ9jrdubm5FBQU0KZNG0JDQzl9+jQzZ85kxowZBAUFcfDgQTZu3EhhYSFn\nz5611pLj4+P5/e9/z4QJE7S+JYGgJkRTl0A7ROQicIa9TrfFYmHt2rU8//zzBAUFMWrUKLZs2cJr\nr73GPffcw7Vr16ioqGDDhg0MGDCAgoICq9CJ0izmTSxevJjt27dz6tQpwsLC6N+/P0uXLrVR2xMI\n7BERsqDBcvHiRWbOnMmnn36KwWCga9euZGVl0adPH61Na3B8/fXX9OjRg9GjR5ORkUGTJk2sOt0H\nDx5kzZo17Nixg6ioKK1NdYmhQ4fyxBNP0K9fP0wmE7Nnz+b48eOcPHmSsLAwrc0TaINIWQsEjrh+\n/Trx8fEMHjyYSZMm0aJFC86ePUvnzp2JiYnR2rwGyddff03nzp21NqNeKC4uplWrVuzfv5/77rtP\na3ME2iBS1gKBI5YsWUKHDh1Yv3699VrHjh01tEjgr84YKg+AkiTRrFkzrU0ReDEiQhY0SGJjY0lO\nTuaHH35g3759tGvXjtTUVMaPH6+1aQI/Q5Zlhg0bRklJCfv27dPaHIF21Bgh++/6E4GgGr755hve\neOMN7rrrLv7xj38wceJEJk+ezNtvv621aQI/IzU1lRMnTrB161atTRF4OSJCFjRIQkJCSExM5MCB\nA9ZrU6ZMIT8/n0OHDmlomcCfSEtL46OPPuLAgQN06NBBa3ME2iIiZIHAEW3atKkiVtKjRw/OnTun\nkUUCfyMtLY2dO3eyd+9e4YwFLiGaugQNkgEDBnD69Gmba6dPnxaNXQK3kJqaypYtW/jwww+JiIjg\n0qVLAERFRQmZT4FTRMpa0CDJz89nwIABLFiwgMcff5ycnBxSUlLIzMxk1KhRWpsn8HGU5Rj2ZGVl\nMXbsWA0sEngBYg5ZIHDGrl27mDVrFl999RUxMTFMnz6dP/zhD1qbJRAI/BPhkAUCf8BisTB//nze\neecdioqKaNu2LePGjWPu3LlamyYQCFxDCIMIBP7AkiVLWLt2LZs2baJnz57k5+czbtw4mjRpQlpa\nmtbmCQQCNyAcskDgA2RnZzN8+HCSk5MB6NChA5s3byY3N1djywQCgbsQY08CgQ/Qv39/9uzZw9mz\nZwEoKCjg0KFDDB06VGPLBAKBuxARskDgA8yaNYubN2/SvXt3dDodFouFRYsWiY5wgcCPEA5ZIPAB\n3n33XTZv3szWrVvp2bMnx44dY8qUKbRt25YxY8ZobZ5AIHADImUtEPgAM2bMYPbs2Tz22GPExsYy\nevRopk2bxuLFi7U2zWdZtWoVMTExhIWFce+995KXl6e1SYIGjnDIAoEPYDAYqghNBAQEYLFYNLLI\nt3n33XeZPn06f/7znzl69Ci9e/dmyJAhFBcXa22aoAEjHLJA4AMMGzaMRYsWsWvXLr7//nu2b99O\nRkYGDz/8sNam+SQZGRmkpKQwduxYunfvzpo1awgPD2fDhg1amyZowNRGGEQgEGiEJEkRwEJgBNAK\nuAhsBhbKsmzS0jZfQ5KkIMAAPCLL8oeq6xuBKFmWR2hlm6BhIyJkgcAHkGVZL8vyc7Isx8iyHCHL\ncldZlud7kzOWJOnXkiR9KEnSBUmSLJIkPeTgZ16UJOmiJEkGSZI+kySpiwamtgB0wCW765eAaM+b\nIxBUIhyyQCBwFxHAMeAZHEjtSpI0E0gDUoBEQA/sliQp2JNGVoOEkAgWaIgYexIIBG5BluW/A38H\nkBytOoIpVKbYP/r5Z8ZSGZX+DtjmKTuBYsAMtLa73oqqUbNA4DFEhCwQCOodSZJiqEwH71GuybJ8\nE8gBfuVJW2RZNgJfAoNV9kk///u/PGmLQKBGRMgCgcATRFOZDvaWuu2rwF8lSfoSyAWmAeHARg1s\nEQgA4ZAFAoG2aFK3lWV5myRJLYAXqUxdHwOGyLJ8xdO2CAQKwiELBAJPUESl822NbZTcCjiqhUGy\nLK8GVmvxuwUCR4gaskAgqHdkWf6WSqesrts2BpIQdVuBABARskAgcBM/i5d0oTISBugkSVJv4Jos\nyz8Ay4G5kiR9BXxHpdDJeWCnBuYKBF7H/wMKt6l+I4D/xQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11072b1d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Let's look at our training data in 3D\n",
    "train_data = np.loadtxt(open(\"training_data.csv\",\"rb\"), delimiter=\",\")\n",
    "x_data = train_data[:,0:2]\n",
    "y_data = train_data[:,2]\n",
    "fig = pylab.figure()\n",
    "ax = fig.add_subplot(111, projection='3d')\n",
    "ax.plot(x_data[:,0], x_data[:,1], y_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "W = tf.get_variable(shape=[2,1], name='weights')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "b = tf.get_variable(shape=[], name='bias')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "x = tf.placeholder(shape=[None, 2], dtype=tf.float32, name='x')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# For multivariate regression we want a matrix multiplication here. \n",
    "y = tf.matmul(x, W) + b"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "sw = tf.train.SummaryWriter('.', graph=tf.get_default_graph())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "init = tf.initialize_all_variables()\n",
    "sess = tf.Session()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# This initialized the variables. You can re-run this to reset the model\n",
    "sess.run(init)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "x_in = [[3, 4], [7, 5]]  # Two input values: 3,4; and 7,5."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-2.79004145],\n",
       "       [-5.3118391 ]], dtype=float32)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sess.run(y, feed_dict={x: x_in})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "y_label = tf.placeholder(shape=[None], dtype=tf.float32, name='y_label')\n",
    "diff = y - y_label\n",
    "L = tf.reduce_mean(diff * diff)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "train = tf.train.GradientDescentOptimizer(learning_rate=0.01).minimize(L)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "eval_data = np.loadtxt(open(\"eval_data.csv\",\"rb\"), delimiter=\",\")\n",
    "def eval(limit=None):\n",
    "    x_eval = eval_data[:limit,0:2] if limit else eval_data[:,0:2]\n",
    "    y_eval = eval_data[:limit,2] if limit else eval_data[:,2]\n",
    "    return sess.run(L, feed_dict={x: x_eval, y_label: y_eval})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8858.1162"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "() () ()\n",
      "(1000, 2) (1000,)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[<Thread(Thread-5, started daemon 123145336127488)>,\n",
       " <Thread(Thread-6, started daemon 123145340334080)>]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "batch_size = 1000\n",
    "num_epochs = 100\n",
    "\n",
    "filename_queue = tf.train.string_input_producer([\"training_data.csv\"], num_epochs=num_epochs)\n",
    "sess.run(tf.initialize_local_variables())\n",
    "reader = tf.TextLineReader()\n",
    "key, line = reader.read(filename_queue)\n",
    "x1, x2, y_in = tf.decode_csv(line, record_defaults=[[0.0], [0.0], [0.0]])\n",
    "x_in = tf.pack([x1,x2])\n",
    "print x1.get_shape(), x2.get_shape(), y_in.get_shape()\n",
    "x_train, y_train = tf.train.batch([x_in, y_in], batch_size=batch_size)\n",
    "print x_train.get_shape(), y_train.get_shape()\n",
    "tf.train.start_queue_runners(sess)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "()\n"
     ]
    }
   ],
   "source": [
    "yq = tf.matmul(x_train, W) + b\n",
    "diff = yq - y_train\n",
    "Lq = tf.reduce_mean(diff * diff)\n",
    "train = tf.train.GradientDescentOptimizer(learning_rate=0.01).minimize(Lq)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "batch 0: training loss 1530.580444\n",
      "batch 10: training loss 1529.425171\n",
      "batch 20: training loss 1528.318848\n",
      "batch 30: training loss 1527.285645\n",
      "batch 40: training loss 1526.303223\n",
      "batch 50: training loss 1525.393311\n",
      "batch 60: training loss 1524.510864\n",
      "batch 70: training loss 1523.694824\n",
      "batch 80: training loss 1522.923462\n",
      "batch 90: training loss 1522.195190\n",
      "batch 100: training loss 1521.508179\n",
      "batch 110: training loss 1520.854980\n",
      "batch 120: training loss 1520.243408\n",
      "batch 130: training loss 1519.664185\n",
      "batch 140: training loss 1519.116455\n",
      "batch 150: training loss 1518.619873\n",
      "batch 160: training loss 1518.126343\n",
      "batch 170: training loss 1517.675293\n",
      "batch 180: training loss 1517.243042\n",
      "batch 190: training loss 1516.840210\n",
      "batch 200: training loss 1516.451050\n",
      "batch 210: training loss 1516.076904\n",
      "batch 220: training loss 1515.748657\n",
      "batch 230: training loss 1515.423828\n",
      "batch 240: training loss 1515.124390\n",
      "batch 250: training loss 1514.843994\n",
      "batch 260: training loss 1514.573608\n",
      "batch 270: training loss 1514.321289\n",
      "batch 280: training loss 1514.082520\n",
      "batch 290: training loss 1513.863892\n",
      "batch 300: training loss 1513.648438\n",
      "batch 310: training loss 1513.461426\n",
      "batch 320: training loss 1513.270264\n",
      "batch 330: training loss 1513.099731\n",
      "done, iteration 333, eval loss 1442.910156\n"
     ]
    }
   ],
   "source": [
    "try:\n",
    "    for i in xrange(100000):\n",
    "        _, lq = sess.run([train, Lq])\n",
    "        if i % 10 == 0:\n",
    "            print \"batch %d: training loss %f\" % (i, lq)\n",
    "except tf.errors.OutOfRangeError:\n",
    "    print \"done, iteration %d, eval loss %f\" % (i, eval())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'my_checkpoint.ckpt'"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "saver = tf.train.Saver()\n",
    "saver.save(sess, 'my_checkpoint.ckpt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7473.311"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sess.run(init)\n",
    "eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1442.9102"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "saver.restore(sess, 'my_checkpoint.ckpt')\n",
    "eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}