FewshotClassification.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/IPython/core/magics/pylab.py:160: UserWarning: pylab import has clobbered these variables: ['log1p', 'isfinite', 'isclose', 'trunc', 'copysign', 'hypot', 'nan', 'sinh', 'isnan', 'cosh', 'sqrt', 'pi', 'ceil', 'cos', 'tanh', 'fmod', 'log', 'e', 'inf', 'fabs', 'ldexp', 'log2', 'sin', 'isinf', 'radians', 'degrees', 'gamma', 'frexp', 'expm1', 'log10', 'tan', 'exp', 'modf', 'floor']\n",
      "`%matplotlib` prevents importing * from pylab and numpy\n",
      "  \"\\n`%matplotlib` prevents importing * from pylab and numpy\"\n"
     ]
    }
   ],
   "source": [
    "%pylab inline\n",
    "import numpy as np\n",
    "\n",
    "import lasagne\n",
    "from lasagne.init import Orthogonal\n",
    "import theano\n",
    "import theano.tensor as T\n",
    "\n",
    "from math import *\n",
    "import pickle\n",
    "\n",
    "import time\n",
    "\n",
    "import os\n",
    "import sys"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "DIM = 2\n",
    "SPARSE = 2\n",
    "MINSEP = 0\n",
    "MAXSEP = 2\n",
    "\n",
    "class RepeatLayer(lasagne.layers.Layer):\n",
    "    def __init__(self, incoming, n, **kwargs):\n",
    "        super(RepeatLayer, self).__init__(incoming, **kwargs)\n",
    "        self.n = n\n",
    "\n",
    "    def get_output_shape_for(self, input_shape):\n",
    "        return (input_shape[0], input_shape[1], None)\n",
    "\n",
    "    def get_output_for(self, input, **kwargs):\n",
    "        return  T.tile(input.reshape((input.shape[0], input.shape[1], 1)), (1,1,self.n))\n",
    "\n",
    "def getRandomParams():\n",
    "    vec1 = np.zeros(DIM)\n",
    "    idx = np.random.permutation(DIM)\n",
    "    for i in range(SPARSE):\n",
    "        vec1[idx[i]] = np.random.randn()\n",
    "\n",
    "    vec1 = vec1/np.sqrt(np.sum(vec1**2)+1e-32)\n",
    "\n",
    "    vec2 = np.random.randn()\n",
    "\n",
    "    vec2 = vec2/np.sqrt(np.sum(vec2**2)+1e-32)\n",
    "\n",
    "    sig1 = np.random.randn(DIM,DIM)/sqrt(DIM)\n",
    "    sig2 = np.random.randn(DIM,DIM)/sqrt(DIM)\n",
    "    mean = MINSEP+(MAXSEP-MINSEP)*np.random.rand()\n",
    "    return vec1,vec2,sig1,sig2,mean\n",
    "\n",
    "def mkData(N, Nt):\n",
    "    vec1,vec2,sig1,sig2,mean = getRandomParams()\n",
    "    \n",
    "    data = []\n",
    "    tdata = []\n",
    "    tlabels = []\n",
    "    \n",
    "    for i in range(N):\n",
    "        s = np.random.randn(DIM)\n",
    "        lbl = np.random.randint(2)\n",
    "        example = np.matmul(s,(lbl==0)*sig1+(lbl==1)*sig2) + mean*(2*lbl-1)*vec1 + vec2\n",
    "        example = np.hstack([example,np.array([2*lbl-1])])\n",
    "        data.append(example)\n",
    "\n",
    "    for i in range(Nt):\n",
    "        s = np.random.randn(DIM)\n",
    "        lbl = np.random.randint(2)\n",
    "        example = np.matmul(s,(lbl==0)*sig1+(lbl==1)*sig2) + mean*(2*lbl-1)*vec1 + vec2\n",
    "        example = np.hstack([example,np.array([0])])\n",
    "        tdata.append(example)\n",
    "        tlabels.append(lbl)\n",
    "    \n",
    "    data = np.array(data)\n",
    "    tdata = np.array(tdata)\n",
    "    tlabels = np.array(tlabels)\n",
    "    \n",
    "    return data,tdata,tlabels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdMAAAHICAYAAAAV7wD1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzsvX+UHOV55/utqp4etDfHl6QhVsgGdIiJf+ydE4SJQgXb\nNBHGKwJBe5Uf96xvWlfCDEJIztiOtVFynZ1c7mqyMjETEJiRQDqaE3LP3XvmWCAsjMxYhWWmiSIk\nOXIMiU2OUWxHsTNnWe86qLun671/PPNOvVVd1V3dVd1dP57POX16pru6urrqqfd5n5+vJoQAwzAM\nwzC9ow/7ABiGYRgm7bAyZRiGYZiIsDJlGIZhmIiwMmUYhmGYiLAyZRiGYZiIsDJlGIZhmIiwMmUY\nhmGYiLAyZRiGYZiIsDJlGIZhmIgUutn4iiuuEGvWrOnToTBp4tVXX/1nIcSV/f4eljlGheWOGTRh\nZa4rZbpmzRqcPn2696NiMoOmaW8O4ntY5hgVljtm0ISVOXbzMgzDMExEWJkyDMMwTERYmaadahWY\nmqJnZvjw9WCYFvJwW3QVM2USRrUKrF8P1OtAsQjMzwOmOeyjyi98PRimhbzcFmyZphnLIgltNunZ\nsoZ9RPmGrwfDtJCX24KVaZopl2mqZxj0XC4P+4jyDV8PhmkhL7cFu3nTjGmSz8SySEKz6DtJE3w9\nGKaFvNwWrEzTjmlmVzrTCF8PhmkhD7cFu3kZhslHuiXD9BG2TBkm7+Ql3ZJh+ghbpgwTF2m17vKS\nbskwfYQt0yxSrWY/2p800mzdyXRLeexZTbdkmD7CyjRrpHlQTzN+1l1aznte0i2ZRKHO+YH0ix8r\n06yR5kE9zaTdustDuiWTGNQ5f6EACAEsLQG6Djz2GDA+Puwj7B5Wplkj7YN6WmHrjmFW6BRpUuf8\ntk2vCUF/79gBjI2l7xZiZZo1eFAfHnm37jhWzyBcpEnO+Ws1QNNIkQpB7zWb6XSosTLNInEN6jw4\nMmHhWD2zTJhIk2kC09NkhTab1GpQKtXR0XQ61FiZpp1+KTweHJlu4Fg9s4xfpMlvmFpcJLeubZMi\nvfde4Oqr0zt3Z2WaZvqp8HhwZLqBY/XMMmqkqVQCZmeBgwdpKFGHKa/IVCrpHmJYmaaZfio8HhyZ\nbuBYPaMgL//69cClS048VB2msiYyrEzTTD8VXtYkPSn04pZPS+w67wlYjAs515eKVNNah6ksiQwr\n0zTTb4WXJUlPAr245Tl2zaQUda5fKABbtqTfldsOVqZphxVeeujFLT+s2HVarGEmseTNucXKlGEG\nRS9uefUzhgFcuECKrp8jE1vDTEzkaa7Pq8YwzKCQU/UHHwyvoORn7r2Xgk4HDpCi6+fKNLyKDMN0\nDVumSYXdbMOh3+e9l6m6adIxLS31391brZL1axj0P2dyM30gi8MbK9MkEpebLYsS20+6Oe+DPrfe\nbI5e3L3eY/b7X+0+fu+92c4YYQZGtUr1pgCwdi0wMZG9KAIr0yQSR9IJx726J+x573Ru+6FopbtX\nVsAfOAAcPhz+unqPeXq6dUSzLGqWattUz3D11SwzTGSqVboV6nX63zAcEavVstMPhmOmSURaIYbR\nu5uN4150F09NhY8vhj3v7c6tVFqf+Uzvsc2g4zZNUnDNZvfX1XvMc3Otv6FUcpbwsG36n2EiYllA\no+H832w6tae2DZw61d8UgEHBlmkS6SWnXPWjVCrcwagXyzzseS+VaOFFIWjfpRIpv3I5mldBXkO/\n3msS73VVv7vd93g/t2kTcPKkWz4si36XbdPz4mK44/b7HVkML2T1d8WM9zSVy8DIiL9lCgDPPAO8\n8EL6nWesTJNKN4kq1Spw663kMwFoMLYsxy2YR3pVap3Oe7VK7tFmkxTOzp1ud+n0dG+TGKn8g3qv\nqcenNj4NCj55RzS/icLYWKtyGB2NNgHLanghq78rZuRpqtVIae7bRwt9W1ZrzFSKuhDZaP/NyjQN\nhF1pV9JoONsfPkzvdRNfywL9sszluZZLXZw751bai4u9VaqH6b0mkfucnHRinOpoFDTweycKfv9H\nrbLP6gIJWf1dMaOG3W0buP9+en1szL0izNhYqxMm7c4zVqZJp9uVdgHyqUR1OaadfrVfaeculVm2\nALB7d+d9qZOkbnqvqdN/6ZJVR6NurrufBRvlXGU1vNDud7H7d4Vy2XHjAvS8fTu95o1cmCaJeGZO\nnRAi9OP973+/YAbMnj1CGAZ5QwyD/vdjYUGIbdvosbDgvLZqFX1u1Srn9RgAcFp0ITu9PhIpcwsL\ndB3U87xtmxDFYvhz7XdtvPsNQpUJXRfi9tvdnwl73fslH2F/Rw8MVe78fle7c9jH85BkZmZILKUT\nV9Po0WkISyphZY4t06TTzUzfu7Ju3ppjxkEYK8PPPWpZrVm27c63n/W4e3e4a+SVicnJ3ty1/fJc\nZLWHnN/vCjqHOY6xjo/T8/btZJkWChS1yIo7NwhWpkknzMDY7sbN6sDWD6IMgN4M304jRi/u0GVF\nf75UxtnN83jPRQuvry7jOpgw4Z0HhLjuWXXJDpKgc5jnEAsoJirdvZoGPPoopRNkeU7PyjQNqAOj\nn+WU8xs3Nno9j94M3+np+Mpw1O9Yvx6iVsfP20Vs0+axIHZD14HRw/49GORPCtx90DFwDDA8fueQ\nWzKuOGqEoOfFxXBpBGmGlWkakINbUClE0OyYB8Xu8DuPYc6hN8M3bH1mF16DN2ct/NylOnTRxAjq\n+JCwsAATtk15SHNzTj5SrUaZkjKRu62RrU7IJDl1T/aMd7IbpSVjiu/ZahXYuxf4/vfz6fRgZZp0\n1JtT05ycc9VyCpod86DYHd7zCIQ7hzGNHOqc6exZeq1SoefdB8s4JooYQR0NFPESnO+wbeAHP3Bn\nUF68GNLI9srJ5s3Rmk6kVBHEhurdALpryRjmnk3oOa5WgVtucTodnToF7NoFXH45ybOcqyXokGOH\nlWnSUW9OXSfXkaxB9Ha/USWVXb/+dBqM1PM4NRXuHMaQ6OXXswEAnnqKitxPL5lYj3n8qmbhf767\njOb3TeAUbXMTqqico4r4WVRwSjexenVI/e6VEyBa04m8T96iTKw63bMJPseW5W4ZCABHj5KC/eM/\npgWPEnbIscPKNMn4xV6mp8mNKF2+tRop17vuoqmglNReXZZZptvBqJuBsYdEL/VyyGJ3VZECNECd\nWlaar8DEK8LER/8nx3K9CVWcQBmjIEW4BYfwEeMEKhUzXA2fWhhoGGQK91L8x5M3ws+7EabdI9BZ\n3hJ6juUwVSiQ0pS89ho9JAk65L7AyjSJyB6thw6RdKqxF4Ak8uxZJ0gGAEeOAM8/D5w44e/6BRI7\nqx0Y3Q5GUQbGDvgt4iLb4nbi6aedv8uwMIIGtOX/i6jjD5Ym8RPnJzE2boY7TE1zP7ebGARNyMJO\nPPIwoZPnr9vJWycPR8ICkd5W0oYBrFkDfOc7rdu2a+iVGcIUo4p2hcxMvMgicFnlrFY6qwXiskGA\n096SPhNUER22+UNIkMamDVGaFMTc4EC9HJpGPR9mZoQYGaH/DUOI6693X16/x01YEG+jKGxg5dGA\nLuojq9xNJYKaB3QjF53OQacmBTGcw1TJXcz3nBAiMY0ggoapjRtbZdQw3L1k0kZYmWPLNGlI60n6\n+9QpnTe54a67KDAh/2839UvYrHYohK3Z9Xu/FxebZ1/e7oGFglM+cPAgee2vu47+n5ggb/65c627\nvQlVlGHBQhmvwMStsFDBLNbiDG7EaRRgw24qy7O1s466kYtO56CTqzuhbsq+EebcBslb0OsJqRsP\nGqZWr6a/5eu6Djz+uNPIIcuwMk0a6g1oGMDWre7UevW91atJUtXUTzVFX12mYXHRibdm2cXWiU4u\nzCDF0+1kxLOv89Pz2L0TuLlhYfdIGVOWiS1bgJkZGngaDSorkOzYAfzWb7XulmKkt65k9d6KExRL\nhYmbUMU81gOowbZ1VP+mhFtgdVaAYZOnok7I8jah63Ru/eQNcId4EhqSCWolDVBJVq1GivSxx/Kh\nSAHE5+ZNiPch+YQ5Ue22kX1gR0fbu9tGR92+Fl2Ptf8qkuxu61UYO7nlutmvZ1+vv3ejqGFENKCL\nH2OVOLxtwfcyqY93vav1tcexTSwvBSlsQDyOba73P4YZUUNh5Xu+vWsm3v676jno5TxHHCgSLXfd\n4pW3bdvah3gSNMD6tQJX30vQoUYmrMzFYpkmOGM7WYQ9Ue06Hpkm/b+0FGxtSB+Mirc2NatEEcZO\nllM3LjZ1X5qG615/BhrEcqJQDbfAwo/OA59qWPjKsrtWRdeBa68Fvv3tcF8nuQKL0CFQgA2BOv7h\n3CJ+Psg66iUZqNfkGu/nmVZ5A9y+U4DMvlLJWa9Y14Ff/3V35r7KABK8qlXafaNBC1StXes4wVTn\nWB5qS1ViUaZ5C4X0TLcnKmjA8vpYLlygbeV3lEruJdmA1mW6skoUYYyhXrRlX7OzwIED0JcHSAFA\n0w38zcUSbtu+Hn9s1/GHKGI95lcUqq4Dv/d7wOc+1xofnUUFW3AQI2iggRHMouL++SijjiLEshu4\ntKnsf3xRZ8B800fHL1tc+khlWvfSEqVvy3vZtilz/9ixcHWoQGszl4jyPTvrzNXrdWDbNkf/HzxI\nfXiD1qzPNGHMV9HB9dHHlb6yRbcnqp3bUV32S9fp/UKBXESjo5QaunGjEOvWCbFrV+x+FyRtKSz1\nvSQJ4549rvWompou7i/MiD/Q9ogG6NrWYYjfxx4B0CVbWBDiixtnRBXrxCWMiAYM8WOsEjdhYSWD\n9w+0PeJ/u2bB5eL96EdJHOT7X9i1EHw+omaaDuk8D1XuBsHCAi2pp7p61b/Vx7p17vMe5DaW12gm\nHpf/tm3+hyMfP/VT6V5yzUtYmYtNwLLmJ+8b3ZyoTgOWZ6B2PTZu7OtgN7RBLcwgniRhlIHR5XqX\nL26cEYZBCu/HWOVSlLq+PPDMzLhKXbwK1+8hP+v66XJgljKijmxxKEO/dV391vuM8VpkXpkKQedq\nZMStTIMU6siIE7/ets2ZUBeL9L+qXNeti0XLzcy0V6bqYSdhPhuVgStTpk90ssIKBX9JXrcu/ho3\nhaENav2o3etEFIWgDowjI+KvZxZWdNgHCwviD/U9KxZnsbj8FbffvnIdpUJVLdN2Y6rre1etchSp\nptFGMzPx/C6/3+m32HnME7pcKFMh6DoVCk7ioF8Bp3firE6spXdKnv+Rkdb3e7we7ebw3gS6tCtS\nIcLLHJfGJJ12CRumSbnnO3ZQ7EquWjIyAtxzD3D+fPbKEAZdXhE1tjg76zQtbTQwdnYW8/MmLAu4\ncMHEnxww0QRdtq1bl3e9aRNw/DgAUMLSxo14Y8MuXH/WxOqLFC5bWnIut2EA+/Z5DkvGNG2bgrAA\nycjEBC02KeUqrmCWXwwV4Lhqr4yP03VS46kvvNDavBkA/u7vnGstWVqiMrj5eaq5evZZ9/tbtvSc\njFYuA6OjztobagtBlU9/Ol+Xm5Vp2pE33ewsLRWyerWTUidvxiwt2xBnklAY+pBooybEqsukyTq9\nlcK8uTlSrOPjGAPw+eW3q1XgW7NVvP+bs6hdAkbuqWBs3Fx5z7KAO0tljMlJh647ky31N8SZ+Rk0\nyclTXWnceCc78/PA5OTKRGuF11+nGZUQzuRJPd/PPedWpIWCI2ze1qUhJozqLXjhglMrrWIYNPzI\nr8h6B0kA0d28SQpP5ZZ27rQ+JYogL+62qOdvYYH8tzKO5fl8T/ePt0B1eb/eQ/3rmeWd+yWe9EMu\nOGbaf/z6+KkJRzMz7vPt9cnquuPqD+oJGBA68buUQfFTvw6oaY2fhpW5SJZpkAcsNzORpDA7C7z9\nNv196RLNXicn6eTPzjquIXa1dU9US1jWBQd8vidPq7eOuNEALAsWTJcR/dyiibHdyztXXYam6V5e\nrlZzy0yv+P0YriuNF0/J1UorUWl5etsNSZ+sX0uioJ6APh4EdayXjdnWriXniWwfqGnOWgl+HVAz\nP/yE0bjy4Z2t+eWCZGEmMjTCzuK9XWi8SUgyjW5mxm3BFApsIQyauCwz7zUPYZm2/UpvglLUDllx\nJzMF7IvlTmFmxr3YRVBS0cwMJbWpyWdCuAdrmf0bcP3UsV6t1vF7LhRajV9db81/SwthZS6SgPnd\nvMNItswEYUdC73bbtvmnzeu6uyxCvsbKdHDENbMMypT16efWUhrTaRWXoNIZ73bt9uPNPI0iYx3O\nGcudB/X+D2qB6Xc+5TX1uoUDCPIuy2HlXe8KFqM4xWMYhJW5SG7eIA8Y5xz0QFh/iHc7gLJ3/doH\nXnml43eRr83OZtjPkjDi8nH5uep37/bd14pXNUwWsmmSa/fkyeAbttN+qlXKJpcpnbVaNF9ervyC\nMVCpuLPYwiwoDrQupjs7Sw+5KIYnJKF6lw8dosiCzHUaHaXM3YkJuvyaRjmPkrNnndWRsnxJI2fz\nekMig062zAxhSz6821Uq9JidBc6cAU6dcrb9L/8FuPlm4KtfHcAPYFrodE3DJBdUqzR6ieW4lmGE\nm6GGVUqdbthO+7EsJ24H0OgaZQadt5VlotLp+vmdT/Wa1mrA/fe7s32lhvRMnORYX6k4RQJS754/\nD7zznbQwuG0DH/+4k8178KAjvoVCdi9pX0pjusk54GSlZcLOQtQpIkBSvLjoKNUPfcixEppN4H3v\nA155xelK7U2JB9zdqZn4aHdNw9avWpZzPV3FqB0I6t/caWGFdvvxU25qgotvwWuX8Gy8ezrVordz\nH9p2a12LLKGanfW9Dt6v278fuO8+9y5qNSpvXbfOmWtpWmt5a6YI4wsWfYojcLJSj3gTD3Sdkg+2\nbaM+vLLbiV98TfbsVWOpaslGyEQScOwqGmGTC7q9SbyJSrJ/c5xtA7t9P0ZY7mJCykZQEHR0NFhu\nPNd73brWXUixllVZuu5OTEoTYWVu4E0bVEs0l+GRqKZ4tQps3+52rdk2TQVnZoDLLiPrQC4YLvGu\nRqGyXFoBgNfSGxRh3ZndWGp+1u7VV5OsRLnJOrmauPwlXcgxKAhdBzZsAI4ebZUbKWPLpTZvfPIx\nnDkz7rubZpNKZ3buBB5+uLUBV9YYqDL13uvT0zkLj8Sx8Ktl+StEwInwnz3rKM8nnwTuvLO13ZjK\nyEiOZzdDIkhJ+k22wiorv+vHMUhGxVswWizSZFoalAA9r17tLOOo6xQgrVYpYW05GU7YNtbsvR//\nB4AnMY6bUMW/HbXwpZqzPu+XvwycOOHfgCtrDFSZWpZjHNVqTuvI3IRH4lBW5bI7e7dQIGUpG7Zq\nGrUVlCfatqkvZ6HgtBtT0XVagJBTsXvDq/y68Tx4laR31eVu5cNPcXIMklFRxyAAuPde8l6USu5F\nSCsVyux94AEaV7Zvhw0dsJvQlpWuBkCHjSdwP67FG/hdPIpirY5PK+vz/rKo4tYlCyf1MqqGmelh\nZaDKtFRyxnLbBt56K2f3eBxWguyo400e2r/faXj//PPuz9g2cNNN5AK+8krgpZeA736X3tM0mtXI\nfedl4I0j883P1RJlVWTvqsvdljEFXT92wzISv2oAVTZkP2jZPW05+U00m9DQhAagCR2AgA6xolB3\n4bPLfwsI1HGrZuF/EefxGHbAEEuAKOCFX9+Hn9w1nllRHKgyXVx0Wk8BwJ/+KY3zxSK5ArJ6kleI\nS1n5DY6Li44lurQEvPvdwGuvOe+//LJjmcraU29D7KB9Z4043O1Aq6dhbm74bvI8XD+md9qFF+RE\n8ORJeu3gwZWPaQAEgCY01DCKP8NO7MJD0GGvKFEsb6PpOja/+xTe9fpR6IIUMOwG7njuAWDXGIBs\nyufALVM1C1t6GtQ0arV2KZNjQr8GO++Mc2KCIv+NBilN1cUrBL12443ADTfEfyxJJ67YsPecb9rU\nvgFCJyoVqiltWUaGYWLEbwzymxgqSY5y2H4Gd+Oz2AUAuAvP4d/gmyvbkNWqQYeNd7/+TEvJjb20\nhH/aO4uf+UIWB/YhWKa67p8Hc/Sos+ReQM0w0w6/Gae6BJtsTyJPcKEAfP3rwKuvUrJSnk52XEk5\n7c55r03xT5wIrkvNg/udGQ5BE8PlZCMNwBJ0vKqtgyaAF7Eeo6gBAJoA9OX3dU1At5uuXUuVqgH4\nySOHcH6/s2RglhioMpX13X6LynorPbKc9dU3/NpRyf9Vxbq4SEX8ctWJvJ3sOGPDfucc6H39WD+r\nIQ63NCtjph1BE8PZWdhPHYLdaMCGgR+IEm6BhSLqMGBjCTpO40Z8H1fhgz/7HVz5vXMtuxaaDiFs\nGAAMLGFxzgJYmUZDvV4XLlDOjJ+V6hfKYyLilzkqy2e8zTTzQL/c7b0ovk6KLqpb2lMb6FqGq5vj\nYLJNwGT8SxfX4rYjD0BHE3+GCfwuplFHEYZeB/QC1jbPYR1ehf69VrewhmXDSRRgQ6CBIkqbyoP8\nVQNj4E0b5PWSY7ns3y3RdeDXf53KnJgA5KAXJcBsmpR9KjOAs1xNPUi8im/vXuBf/oXcZkEKrJPy\nDeuWDlKGak2abdM1917ruJKymMxxBRahQ4DUYR1XYBEbRuYxe4+Fa3ABBWVdVQFSoBIBQDMM/OgD\nv4YLl1Zj5B6Pi1eVWSDdk7kwbZLkox/tBLdta+2MJ5frSWv7qb4S5zqUEdbLA7d180dt/+ddZ9ZP\nmLtpK9ippV9Q20Hvmre63vo9KVk7keVusCwsCHFLcUH8GKtEHYb4MVaJm7DgiIhnTdQlzRA2IJY7\n/tKzrgfLpfxsu/aFQyaszOnDUuJyQlKpAI8/TjXquk5NOZaWnAqPHTto22oVmJqi51wjLR+1YFdd\nWqkbSiU66UF+dT7p/rQ7LzKW8eCDrZnSc3Ot20urU3ajaddWMGDZNQDBS23Jzz72mHOTjY76N6wP\ncxxMJgkSacsCvtY0sR7z+I/ag/iIMY9TurnSFEnK+5v3PojZrRaOfvok/kpb596JbfvLpVdm1ZVs\nehnPhk0YjStinq0FrXW8Zw/1aff2XJa9ujXN3ZM9lyws0ElQl7zvZQl71cL1cwHwIs3+dNN4fmam\ns2Uq9xm1UXyY40pQw/peya3c9ZFOTg31vV27Whf69m7zhV0LomEUhS3HpnYN8+UHR0Y63ytDks+w\nMjfwmCngP4nevZvem5x0b2sY1B0vSmOYzCGE+++lpe5jnqqFq3ZB8r6fx2zfdnRzXmSMVHaV8YuZ\nAvEkQ4XJUOaG9YwP7UTaK1azs60LfQPuz792uYmNJ63OsdCgjFRdbx2P1A5vCa2bHIoy9cunkOeq\n0XBv+4lPAD/60TCOMqFYlrumCAhewr5ddmaYdSq5T28r3Z6X8fFgJQrEm0HLypDpgU4irSaNBq1T\n3/L5oJIxL96MVL+DqFadHsEAZa1OTtIjQfI+FGXqne0A7nMl0TTg8suBjRtbG8O0G4MSn+Ef5QBl\nnFMtzPWLeXbKzuxkyeSpT283tDsv3V5XzqBlEkCnW12K9YUL/uvUV6vA5s3kQey5CqPdQViWe7wT\ngpajOXkyWfdMGF+w6HMcYc8e91rVfq5z75rHahLYtm3uta0TveB4lANU45yGQYt8z8z4xxH6nJ0J\njl256eW6piSDNkmw3A0WT7KuGB1tzXVRiwv6ktfizUZXVx8fwD0TVuaGYpl6kZ2RatSdauVsqa5z\n1Wtw//1OfWqzSWtiy454qv//0iVncZXEGFhRYpFqnNMwqJlxkAuxXKaWgbZNz+ym7S+9XFd2pTMJ\nRxVrALjrLqds2jQpA1gtLpARp1jzWmQ2uoyZyrhqwu6ZRChT1cL3LqvnPVf791MXPG8OjsymljpE\nBsmffJJcxEtLCfGkRRlAu/2sPEnqyWL6Qy/XlV3pTMIplcilK8sWn3+extKTJynfUYq9t/lO7IyP\nt7ZETdg9kwhlCvi3kfULSW3f7nafS9T1UTdsAJ55xrFcvdlnQz3/UQbQbj4r4wzyJAz9h2ecXq8r\nJw0xCUNtsDYxQcOH7Dh68aJ7LN29m8R+dhZ46inHaIl9waPEJ8IkSJmqqOOLeg5lWrYfmgY8/LDj\nAS0W6cIahtNUPzFegbADqJ8Ahf2snDLKfqx56707DFgxMilHzYnTNGc+LgTwj//obCcjR/v3O5Vf\nL73UJ30XJlEvAW0JE6lMJd5z+JGPBG8rE1yl7/7ee4Grr05xy8eomZ6mSeuZPvRQb3WoTDZJwQyf\nGR5qjFTXSaH6uW+3bAHOnwfuu4/+P36ccldkv4C+HVRQCaAcK73W0wDjekNrJxgG7zlcvZoSlfy4\n6y56T3bHW7vW6b7WqRNbImnXHi4M+/eTIrVtd1BZhdsF5gs56HzmM/TM153xUC47+qhQAH7v92g8\nVSkUaHz97Gfdr/t1y4ztoNq1ulTHykYj2rgZgUQrU+85rFRo7eQ9e4CPftRRnKOjpGh37nQmJdu3\nA7/8y6RTUkmUXqnVKmW+qevb6bp/HSoPrPkh6gSNyQVq/cnGja0FA3feSY6uN95wv75pU58OSO13\n3W5VJcOg/tND6jGdaDevN8tXeqekK+FDH6Kg95kzlOELODHVZhM4dYoeTz8N/MmfpMwyjZKo5C1y\n1nXgk590L1jN7QLzRy8Zx+wWzhWzs04XukaD/l+7lnSU7OQHONm7ug5cey3w6U+3b/QVGW8+gncZ\nyulpJ8MX4JipH/JceMOHAM2O1JRsTfPfx1e/Sue1k75I3LjRa0KLWrhrGNST8dFH3SeQaxzzR7cT\nNO7QlHsuXnQyenWdvH/T086YOzIyhF7p6mL3subU2693CHKaeGUKBHun6nV3cFz6+r39fQF6TVWm\nXsWZqXHDO2j6nUCZ0+4dWBM3o2BixW+C1m5RcfZe5IpKxd26dfVqpymDEG6nl6ZRItLARUJtXgO4\nl6EMM0Hs0/iWCmUaZETJyg+58IlhAI88QoHw48fd+zAM6i0pQ4NexZm5ccM7aPqdQD/XSWZmFEwo\n2l1z9l7kDtOkvBS1yuTgQadE5uxZGkuBPtWThkEt++umG1Kfx7dUKNMg79T8PC0c8OKLzmLii4v0\n2smTTnvnNlFAAAAgAElEQVTCn/s54Hvfo9TtQ4doNuVVnEMfN/ppEYZ172VuRsG0pVqlm0UOSp3W\n32JZyB2mSQ3tZ2ac1R7vvps6mQ5NJLzJNGG7IfV7fAvTwFc+ktj8Oai/+MKCe1FxNUdt9WraXl3g\nVv2M2jh/qD9iUN/tt4JAh+MANxxPN94O5d6bIaGw3PUXvyFgYYEa3MvxM/ZG9oOix3E2rMylwjJt\nR9DkWSasSveEysWL9GwYFExXJydyST3ZOH8gM69hWYR+bg+2RPKBGnfSdeC22xK3PiQzeNSh6NIl\nYO9eskK9LVpT6bSKc/lEH1KvTAEn9Cd7EEgX7YULwRm+AI0j6oLuXkEaWJbasHzMQYlJqbtLBkwW\nkrS8MseKlEHrQiFHjpASlY3uhUh5+DwoAS+GWGomlCng31Gq0XD3LfAihaJaJcV58aLTllCuOFOp\nDGCM6Wdsqt3AP/RAcQrJSpIWx0MZH0yTrNAjR5zXpDVqGNSmdSBj4iCJyTOYGWWqng91bT0VTaPH\n1VfT433vo/6SH/+4k6ykWrJLSwO0TvvRJL3TwB80oGbB8uoXWUrSiiJzLCOZZfVq/9eFoHEzc5c7\nJqMiM8pUPR9ey1S2HfyFXwC+9S1y/37nO8DXvkbbqnWpXgV85gyNG6kUoDADP5fHdAdb8ywjGadS\noXKYRsPpyysE9UXIpLjH5KXJjDL1ng/AyZw+e5ZKYl57za0sZSFy0MoIAHD6NI0bqRwvehn4s2R5\n9QN2j7KMZByZvOkdSzMt7jF4BjOjTIHW8yH/npoil63adlD+3Wl1+G6aaySOXgZ+trw60w+XfJpg\nGck8QWMpE0ymlGkQ6r1fKFDThosXnVTvdmhayscLeRfIHoyd7gq2vJhOsIzkFr82rCwGRC6Uqd+9\nX60Cx46Rgm3H3XcDu3alWFB6iW/l3fJiOsMyklmCFKR3KJmepib4HDonEr2eaZyYpruEUrbJ8tah\nyuQlgJ6/8Q3K+A3LwNfb7vSFQasEMJ3pdG55cXUmY7Rb5tg7lMzN8dCikgvLVOKdcVUqTscj6f5d\nu9a9tNu3vw3cdx99vtN6fQNPcuz0hdUqpS7LamtNo4wspjNhzi1ntDIZo13jGm+ofNMm6oHOoXMi\nN8pUHfuk4qxU/EM/Y2P03re/7Xz+qac691MeeJJjuy9U1/yT2DbNFMbGeODvRKeLyRmtTAbxdkA6\neNBp0uAXLhsb45ipJDduXnXsq9VoFYT16+k9bwc906SV41XOnPF3fajImZthDGimVio5RbTeL7Qs\nZzUQ9cH+mHB0upgDv9gM039MkwwNGeqSfXjV973hMu5ASuTGMpVjn3TfCtHeoJAu3bk54F/9K+Do\n0c5GyECTHKtVsjKbTVKm3o795TIN9Go/xdSnJg+QTheTM1qZDKGGwNTwFw8X4cmNMpVj3+yss9ht\nJ0EZH6dHtQq88IIjXKWS01C/U0OhvqGu+qFp7o798kD27QMeeIC2KRQo4ypzjTX7SKeLyRmtTAbg\nxaPiITfKFHDGvkqlu1V4VCPkrbeAHTtIGY+ODjHvJEzh/Pg4BTVmZ+n/doqUC8gYJpfw4lHxkCtl\nKvEzKML0hAeAD36QhA6gkOTEBHDDDUMw+LpxM3ZapNX743fuBD73ObJohzpjYBim33BDq3jITQJS\nJ8KUY87OOooUIF1z6hTwxBPArbcOodwwTPQ/zA/zZmc99BD1X7Rt+p8Tlhgms8h5+YMP8rw5Crm0\nTP2IOjtLbHVEmB+mbqNp7hmDrvNUlWEyDof/o8PKdJkwXtNKhVafUUs3JYl1j4T5Yeo2pRL5rms1\nygbet4/vMobJIZw20R2sTBXCzM5kk/xjx8gTqmnAzTfTQuOJJcwPU7fhSmyGyTXc4Kt7WJmGpFql\nuKgUrkcfpWoUaci9/HJwfk/qYJ8Pw+QabvDVPZyAFIC3h/nsLHk+haDn55+n3J/FRW72zDBMtuAG\nX93DlqkPfi4OL0eP0na9Ji6Fjkdw4IJhmAFjmtRUbW6OGtrz0NMZVqY++Lk4KhXgwAEn0dW2yVr9\n/Oe77xYSOh7BgQuGYYaA7FZar9PKMLw2RmfYzeuD18VRKpGy/NSn6DWA3L1PPUVC122z59BLjPJa\npAzDDAEeerqHLVMf/CpFpHF4883AV79K2zUa7vX+whLaNcytSRiGGQI89HQPK9MAZELr1JR7hnbp\nUjz7DuUa5pVJGIYZAjz0dA8r0w54Z2j33AOcO0dW6cgIxVJ7IXT1ST/LVDi5KXnwNWEGTLsFPlgE\nw8PKtAOZXV2ek5uSB18TZsCwyMUHK9MQeGdomZixcVV28uBrwgwYFrn44GzevMJV2cmDrwkzYFjk\n4oMt07zCGQbJg68JM2BY5OKDlWmeyYS/OmPwNWEGDItcPLCbl2EYhmEiogkhwm+saT8E8Gb/DodJ\nEdcIIa7s95ewzDEeWO6YQRNK5rpSpgzDMAzDtMJuXoZhGIaJCCtThmEYhokIK1OGYRiGiQgrU4Zh\nGIaJCCtThmEYhokIK1OGYRiGiQgrU4ZhGIaJCCtThmEYhokIK1OGYRiGiQgrU4ZhGIaJCCtThmEY\nhokIK1OGYRiGiQgrU4ZhGIaJCCtThmEYhokIK1OGYRiGiQgrU4ZhGIaJCCtThmEYhokIK1OGYRiG\niQgrU4ZhGIaJCCtThmEYhokIK1OGYRiGiQgrU4ZhGIaJCCtThmEYhokIK1OGYRiGiUihm42vuOIK\nsWbNmj4dCpMmXn311X8WQlzZ7+9hmWNUWO6YQRNW5rpSpmvWrMHp06d7PyomM2ia9uYgvodljlFh\nuWMGTViZYzcvwzAMw0SElSnDMAzDRISVaRaoVoGpKXpmmITCYhoAn5hM0FXMlEkg1Sqwfj1QrwPF\nIjA/D5jmsI+KYVywmAbAJyYzsGWadiyLbsRmE6jVgMlJnuEyiUMV03qd/mfAJyZDsDJNO+UyzWh1\nHbBt4MUXaabLCpVJEFJMDYOey+VhH1FC4BOTGViZph3TJNfQbbcBmkYKtVbjGS6TKKSYPvggezJd\n8InJDBwzzQKmCWzaBBw/Tv/bNlAqDfeYGMaDaaZcV1SrNEktl+P9Iak/MQzAyjQ7LC46rl5dp/8Z\nhomHqIlCvSjifilvpi+wMs0K5TIwOurc7Bx7YZj48EsU6kYpdlLEXsXJWb6pg5VpVpCxF57JMkz8\nyEShXiarXkU8O9tZcUZR3sxQYGWadqpVujkBoFIBdu8e7vEwqYW9im2IMllVFXGhABw8SEqyneKM\noryZocDKNI3IUa9UAnbupBsOAA4dAk6c4JGQ6Rr2KoagU6KQOrFdu5byFqTinZ+n986cAU6fptyG\ndoqTPU2pg5Vp2lBHPV0Hlpac99gdxPQIexUjUq0Ct95KZWkSTQMuu4yUIgA89RTQaNDfut5ZcXKW\nb6pgZZo21FFPCLopm016j91BTBeobl32KkZE3pcqQjg136dOOYoUAN7zHuDJJ1lxZghWpmnDO+pN\nTwNnz9J7lQrfkEwo/Ny67FWMQKlEytOLYdAJPXLE/fpP/ASf5IzByjRtcCyFiQE/t+7u3SxOPbO4\nSG5dVaFqGvCJT9BJvecesk4l99wz+GNk+gor0zSSd5cQp51GJolu3VRdVu/BlkqtylQI4M/+DNi4\nERgbo+fvf58U6dgYLbuWih/LhIGVKZMuOO00FqI4OKIovaDPpuqyVqv0AxoNUqAf+AC9Ztut29Zq\nwN69wAsvOD8OSNGPZcLCypRJF5x2Ghu9ODiiKL12n03VZZ2ddZKNhAC++tX225896/5xc3Mp+rFM\nWHjVmAFTrZJ3p+sV0nr+YMbgJauGSpTlN9t9NtOX9R/+gX6Y/HHXX09Z+PLBi1JkArZM+4SfO6vn\nWX2qfGB9hhOwhkpQrDWM67ddnDZVl7VSobIWtca7HbYN3HEHsG4dKc6JCfqsEPSYmKDt1CYPTOpg\nZdoHgnRfz64s5YOiVsdLkxZKm4CxRSufN1/eE7CGiJ/SCzvX66QwU3NZTRP42MeAJ54I/5nnnwd2\n7SIX8aVLTqKSrEXdsYOUbt4nyymGlWkfCFKaPWdQLn9Q1Op42y7i//lyCQ8fXw+h16GNxnPzpSqT\nkhkqXqXXzSQxNQqzE5WKu6NRJxoNskDPnGktn5GNV2ybFO3sbEZOUr7gmGkfCIr/yJn5gw92qf+W\nP/jSbQ/idn0ePyUWUUQdmt1D4MoHaVl85jP0LMOyHKZlwpDpeGcQpgns20c/2g9Nc/9v21Rn6nUN\nFwrAJz9JzwAp2kOH+KZLIWyZ9oF27qyeZ+amidFJE2dOAuISUBdF6HodegyjV1BiCIdpmTDEHe9M\njZdkfJzct97uRoB/NyQ/bBu4/HJg61ZgZoY+t7TEGb4phJVpTHgHgH64s0yTugc+8ICJDzfnsV63\n8JvTZYxF/CI/93OqShVyRidlE7cyCrO/uOQ9Vbl21Spw7Fjvn1eb3QPA4cPJ6qLBdAUr0w6EGUj6\nNgD4fPniIk1eF4SJvxQmVi0CY10er5cgyyJpHXL6jvfkJdBE6iRrccvioJVbqiZxlhU+o9eLpgG3\n3QZMTjo/MDXpzIwfrEzb0G4gUcfZvgwA6pcXCsCWLUClgnLZDFRyUQY+r2WRqlKFOPCevOlpShip\n1ykutnVrIhYS6CRrccvioJVbEtscBlIuk3Xp1/lIsno18IMftG5TKLgVKZCh7Kx8wglIbQiKJXoT\ndkol/wSMXhJ45GfenHW+XNRqEE88AaxfDxPVwCSmKAX1fphmjpqfe0+et0vNzIw7O2tIdEr26TUZ\nKEhWB51c1HOS3qCRs+nf/u3221286K9sZQP8oH1z5l/qyL1l2s6TFzRL9o67i4u9196pxyDruet1\n4AWjjBdhwACtVaoBaL5dw+GKhaVPm9i9u3U/qZrVJw3vydu0CTh50qkJFCIRfscwtZrdehSCZFXK\n5fT0YPsJJN5AU0+Yrrc2uA/D5ZfH3NmFGTa5Vqad5FYdmEolKv+anQXWrm1VWr3W3qnHoGk0ibVt\n4Gsw8aWf2Yo7vjsDHQICQBM6Dny7jFfuo8+Oj7v3lTvXbJz4nbyxMbrghw5RbCwhM5ROyqbT+/v3\nk+G9aRPJEGdzd4l6woRw6kTDYhg0oMTa2YUZNrlWpmHkVv5/663UqAQg2X/0UWdNbj9UQ0fXKXu+\nVGpVgOox6DrwK1oVH8UsdAFcuGIt6t8toog6mtDxAB7DK6ADmpujfQ0iizg3+AWOTZNipSmboQR5\nXPbvB+5bnowdP07PnM3dJd4T9ku/1LnZvUTTgMcfJ1M/1s4uzLDJtTItlUiBCeGfzONNMJI0GqRI\nZSb74cPBVu3evaRIT51y1gZWFap67/yKVsWLzTJGUAdsoHlOh43l4m/DwDeaTt7u9dezR2hgpGyG\n0k4u5ubc28pJ2fw8GeEXLwZ7X5hlvC6rnTvDfU7TgLvvJo8H4H+C2b2UWjKvTNutnzgx4ViEO3e6\nE3a8iZ3FomOZjozQoCPDaUEdwEwT+Jd/cb8mBy91GzmQXXHAgiEaUn3CgA0DFC/V0cQffcjCnV8z\nIQRZxj/6EVsPQyGBJTMq7azKTZsci1T+Lzl0qNX7ImOlAK9l7cvZs+FdvELQzPrYMboosXd2YYZJ\nppVpmPUTZaLdww87faY/8hFHUcoEoxMnSOEBNGvfvt3dq3r/fnrd68b1Dl7XX986KMlQyRdFGX+I\nEeggM1gDIAAITYNWLOKH7ytDe5mOU1rKYa2HhI//6SEF7oB2nkIpn2rMFPD3viwuUjZ3Cn7yYPGW\nrXlbB3aiXqfB5POfz/mJzBaZVqbtZujqgKNpTp/pWg149llHUcpcActyygzvv791MmrbwAMPkAdH\nvT/Uwev662m27zcolUrAKcPEx+1HMYFpvEf7WwgAKIxAv4dqTK+DiaLSJKVSCRfO48EwRlIQTOzk\nKRwfb530yftB9b4EZa8n8CcPFvWEtKsxZXJFppVp2PUT1ZIUwK0ob7rJeU8qoiBsmyac3kFMDl5T\nU/6DknQ5/1+N/4BP4yHoELALI/j62q0YuaeCsXHakQn/QbLTwMaDYYykJEGkW0+haTrel4sXqdeA\nRP3JhQJw4QLJbG5lqFymWbbM5g2DWj4zMkKzYCZbCCFCP97//veLJLGwIMSePfQcZRt1u40bZVEh\nPdatE0LX6W9dd/Y1OiqEptFD1+kxOipEsSiEYQixalXrdy4s0Ovq+wsLQmx974KYw0ZhA8Je/uIl\naGK3tsd3P72cJ+/3RgXAadGF7PT6GLjMxSlUQyLK4UlZ0XUhCgUhZmac17dtay/fgyARcicHAHWg\nUB+aJsTIiDM47NpFJ3LdOhpgEio3jD9hZS7Rlmm7OF83CxKHmUHL7WTv6kbDcXXJLFzbBt56i47p\nkUfcyRmWRTP2AweCLUCv+w0Afv+WKp5vrMdleBuAEyfVIfBDUUKtFt2S5ARBhX4JVUKC0lFd+pZF\nrl5Z77xjhxO6sCyS7dx7OCzL3ZNX09xx09FRd6cLwLko58/TIuFM5kisMu00KPTLdSkHDbUsRrbf\n1DR3otL0tLOdTNToZuEHywJublgoog4dy8lGcBTqDTiLw0Y8nkQ18Ur9P1f0S6gSFJSOel942802\nm1wC2YJ6ItS+zYD/hCoovsNkisQq006DQj9vbK/hMTrqJGYsLZEvp1ajhCMh6H664w6KM/m1XvNr\nFyiV8ZdGyqg3itDwNnQAal6grtH6w3Hcd92O9wkxtOKlX0LVz6B0lxei3U8Iu5TaY4+RRdpskuxz\nCaSHdifC76TwLCQfhPEFizBxhJgJE+frV+hK3e/CAoU5DINCIDJ2ahj+4ZJi0X086u8oFJz4q2E4\n3/GfNy6IN3563Uq81AZEU9PFX8/E98P27HGOWX53u9/f6dwjCbGrbumXUPUjKB1hv34/odtdJTUs\nnEq5EyK5J5TpSFiZS6xl2sssuJtJfLtmDmoJmRAUPxXLiXi6Dtx4I3DmjP9+63XqerRuHVmic3NO\nDErXyYoFyGVcKgEmqrhqtYXXrirj6h+8Ch1NNGHgwqcfX8nijYNuJseZzf4NI1RBQfZq1Sk09i7F\n1i9h7fFC+P2EbnfFfQNihk9o9gmjcUW/ZmsR8M60Z2bCz7zbzdJVC05m66pJeoUCWarq6+0eqjW7\nahUl9n3AWBC7tT3i/sKMaOgjorlsjTYAUUNBjGszbS3HKOcsbGZzJi3TXllYIJeDvKgjI5Ta2qtl\nG9ZM7NKcDPrqfmXizswIcfvtTsbvIMiV3DGJIKzMJdYy7YR3pu1dfrLdzNtvli5fl2uTqpapbDko\nMxyPHaP3Go3OxymWF5W47TZaC/hbs1X8x+Z6FFGDtmSvxEkFKIC9BBvvNBb7ElbpJrOZY2MKluW+\n2I0GrW/qbcocNjAd1kzs4kK0W0ZN9bTce2/3a5z7GdF+DfO9jSAYhUwmITAqQ1WmUeQraPnJMG5M\n72ffegv40IdIUfpltatlL7ZNY+C999J7Fy86+/3iF1sVrKbRPicn6Tcaey0UUUMBNtRyb6lQoen4\nzcfKGPMkLw36HmSvlEK5THVSar894bO+aVgl2Y3PPeSFCJogTk46YQYAuPpqR8mGkasgJR3UMJ/x\nIUHZ3kz/GJoyjSpffpP2sbFwA4S3+9GOHU7Z2NtvU+/qz3/eOU6A+u4aBg1KhuFkwqu/wTTdKzHp\nOg0wctupKcBAGWuhrZTASEixaig88dhKrJTvwYQg66Vke6Bjx0hreRVhWCXZB9Pf+9VyuUw1Xi8P\nqRu5CpoftGuYz3jIbBICozI0ZRqHfHkn7d1YU3LbqanWPruHDrUqS+nyBZz6bPU31GrAyy87+zAM\nWrZQrjnq7MfEL2h34W5xZGVbWV+qzzzhmt7zPZggVOEKMuu6UZIxm/7er5ayIxWpDDNImVflSrbA\nLJVay7qC5gdBDfMZH7g0JhcMTZnGLV+dGtvI9wD3duUyuWHlKjEAWanSTebtZy2E8776G3TdUcqa\nRm5g74ocspXnN6/egLvfPLJinWqaBu2JJ1pGJL4HB0xY32c7RThE/7j3qwsFktuREVKkACnSt94i\nGdV12ubgQZJpqXhHRx1rtd38wK9hPuMDJyHkgoEoU78xqlv56rULnF+pi/TQye3keqIHD7Z67/yS\nkeT7Xnex2pBB7WMtleLaS1V8yLbw3jdPwQZgADSq3Xef76jE9+AAqVaBW291LuCJE0NrvBAXcnIo\nBHWxm5hw3L6aRjK9YQNw9KgzWZTL+6leEI6fxwCfxMzTd2XaTtGFla8oXeDU91TrUt1OPvyWM/P2\n0pWKU1qu6m8IitmaJvCX01Vcd/96FMQlGMsRUgFA82peD3wPDojZWafNVa3mFAtHVYB9DnwH6WnL\ncjwhS0vAZz/rTkSSrwPO0mve2CrDMOHpuzKNI+7XaR/t3KHyvVrN6UcthP+A4ae4VKVcLtOjl8nB\n2KIFYV+CBuHqv4sNG1hbJpGjR+nRjQL002xBabbKdmEMV79t2ulpVe5tG3jjjdbVwoQAnnsOuPNO\naoX5jncA585RDJRFkmG6JEwxqnz0UsgcR6e1qF3gZmaozl4ukxZUb++3H+93b9sWvi1fy0EoS6yt\nPDZuDLmDZIGsFc/LxgyaRhfW2/cxzOf9hFR9vVik6y27JxSL4h83bhO3FBdaZVsRxKBd79kjxM36\ngvh97BE36wsth7mwQE0V5E9p11hEHpLsSZHUrneZkzsm8YSVub5bpnHERsPso51VuLjoNFxYWnJq\n7fy+2zvT9xoWQI9JQc8/D8BdDgPAvQozMzxk+YtfADxMt3jLcsxAdd08NSh/6BDwzDOOidhs4p1H\nZnAMh7Ee8/irukkfg1sQv7V5HvW62eKZubNUxe/a61FEHXW7iDdK8wCcYzJNSjx66SXHgw24V4UB\nnLCHpNEgL/cXvhDHiWWYfBBZmYZdiSJqbDRK7DBsVqyfR8772UrFP7ba8YcdPdr6uqa1jZcyA6ZT\nALydgJZK7iyeUsm9X8tylhxS0CAwgjo2YxYf1izcWSq3COItsFAsmi3yO7ZoQeh1aHYThl7H2KIF\nVZnKr96wATjiVGLhxhuBq65yv+bl+98Pe9IYhgEiKtO4cyv6VVcZ1jr2U7pBn+3quGQ2iJdPf5qD\nU0nFb/bWTkAXFx2TT9fpf5VSiV4XgmpVNmwgb8XSEnTdwMfsQzDsJdgfL+LYhml8pFCEARLEaypl\nzPtN4MplaKMksFqbWaKf82PDBuCFF8hi1XXgt38bePpp5/2gCSd3xWMYfyIp07iUn7reZ7/qKoMs\nW+/gEKQ4Iw0cqpUi2bgR1Y3/GdYUD0ypoVOm2+hosGt4YsJp8vzII043j9lZGGfOAKdPA7YNu1bH\ny88sYu/IPA7fa+FHa8t4zjJRLtMC9C48AluF6StPlQp5mOt10uWnT1OpjHft3Z/9WeChh2ibRx8F\nNm5074c7cjFMMJGUaRxNBbw3qN/i2v0iaHCI9XurVaqUV9F1nN+wq29ljUyfaOfiME0SXtkSyK92\nSxZ4qlbr4cMrsVZb09EUBfxrcQFLS8CfYDcOK6Hbv5yukivXZ6bXKURy4gTFT1980aklXVx0K+jL\nL6fD86s1VX8Gd+RimFYiKdM4mgp4b1DvDd5P+j44VKt0YtTsDgB497vx+FnTVdY4O8sDUypo5+KQ\nSUsnT1LM1Vun4p11qkpW1/E/3nMjit88h3txAJvtw9h30Uk8uqFWxXt2rAdsf7MwjCxfey15mJeW\n/Ce/nSbH3JGLYYKJnIAU1ZIb5g3a9++2LP912iYmgLMxfxfTHXEH/7zaTDa8bRc/8AjgOz50A8Tr\nr64kFP32aguTy4lHv6pZKDTrgO1fr1oum4GyrFqthgHcdZd/HLXT5Jg7cjFMG8LUz4g+116FXbQ6\ndd/tXVRa02h1cOEuaywWk1vXFwTSXO8XR/Fzu32OjoZfiVsVQJ/jkm//9YznvZmZwG29X6cueK/r\nVEca9qcP8970I9Vyx6SSsDKXiMXBe7Vu4zAu+tquzzQpk2P7dqfj+MaNK2/Jskae5Q+YOP37qhBK\ns00ufhtm/14B9Jh+ZrWKqy5YeOlCGZiexzvOWngJZdxy1sI1nt9glkHbzpYBmL4GsKY5rTUDGjK5\nfhonHDFMOBKhTHsh8Tf6cqYmzpxxms00m66BlfvuDom4/Pt+Qrh7N71++HB3TR8kqlBUq2jeuh4/\nW6vjN1DEhpF5vKLtRrMJfMAA5pXyGZRKrm3vODiPKctRqJs30/PatRRl+J1L+/EbYg7Fv9mEW/94\n3DcRjhOOGCY8iVWmncacxN7o1Sq1j3n2WXebGYACVpy1MXziCv4FCaF3/wBldHs7K3mXN1quD3vz\n7CJZn7Dwc/U6DDQB1PAHjUn8MSZRhYklAXz9xs244QasdBHRlrcVqOPmhgXLon2r+r5SAc7v3I9r\n995HzaGfPo7fAfAkxlsS4TjhiGHCk0hlGsbqTOSNXq0Ct9zin3QEOKuKM8MnDrdAOyGU+1eFWS56\n6609kdvUahC2jZ+Fjt/AKH6vMI1powh9qQYDNtbjRXwQJ/EJbRoP2xNYdboOnF/WkOUyRLGIRq2O\nBop4eaSMqbK/vt99bs612MImzOFJjPueIk44YphwJFKZhrE6E3mj790brEgBZ1XxRBwsExm1724Q\nqjAL4SxdpHop1BIZAAXYEKjj8uYi1uvz+D8xifV4EQXYMPQ6/sO1c1j1Rg2a2ge4XIaxZTN+eBE4\nvrqCqYrj4vXq+zfe2oRrjx+HbGx4RN8Ebbkxk7e7JYciGCYciVSmYa3ORN3o1Sq5doPghSKzi4yP\nHj7c6kZRhblQcFZcUL0UnvXSlqCjgSJeQhlV28QkJvFBnISm1WGMFnHt/3o9sPc4fda2gbfeWrF+\nVxeLWDtdwXMWdTlaXGxthDJljeM7GvDvxBy+oG2CPj6O/3R1gialDJNCEqlME2l1dsKyWmOkAA2a\nd687ObEAACAASURBVN8dz0LTTPJQLM/mpTp+uHcWq9dZzrVWhVlm+Qrh9lIo2/z9WyUsPLeImdfL\nqAoTQgCndBN3FOZxeKuFayrlVktYOQZRq+P/e8DCf7LNlTbBo6O0e8AJ3T542Tieqo9TGKXCYskw\nUUmkMgUSZnWGwa//LgBcdhmwa1fKfgwTmnIZzUIRdrOOpijg8iMHIZ5tUgN6tT+ljI0qWb7nS2U8\nJ3vpAjh1CvjU0TG8bJsri8voOnDbbcDkpIlrpAx5lelVV5EZWq9jSS/iK82yawEb2UNCTTAeZNtO\nhskDiVWmqcO7SghAI+H0NI9WWcY08fSWefztjIV/LS7gXhyAZpOV+t1ZC9+HqXhYHAv0fKmMX56g\n7kYfMKp4UazHDY06XkAR6zGPV2BC08iqnJz0iFClAhw8SPH5kRGarG3YAMzN4cL1m3DmURN6zVnA\nplikjw2rbSfD5AFWpnHhZ5kK4a9kmUxxXcXEtsMmbqhVsdk+DIE6GqKIylNlvHKQFJiTlU5W6nNT\njnK72bagCaespQwLf2WYuPde0psuRSpLaB59FDi73JPy/PmVkpufP3kSf7FzDFOWicsuA973Piep\nKKj0lWGY6LAyjYvFRYqPqos/Fwo8auUAx+A0se/UPP7bMxa+Isr4yyXSgkK4s9LP76/ivUcs/IpW\nxoJh4mWjDCGKaDSorOVrRhmPP06rtKmc30/N7gvNOrSRgtMIRLY1EgLiUg2v/qmFU0367ldecRRy\n6vIQGCZFsDKNCzUjE6DSh337eNTKCU5Y1MT6F8h9W1T0XbEI3Fmq4uK/m8V1Rw7ivWjidhSxb+M8\nPrjLRAHzeHPWwt9cLOHPV1u4ZgygSCpRrQLPP2Dhj5bq0NCEqNkQAHQICGjQZKGLsPFPTcdL0mi4\n85xYHBmmP7AyjYraqumRR4CnnqKEEE46yiV+zY8sixTp2MR6iLcvARDLDRPq+IXvL3cqKpswK8A1\nssHDYXeWkGWZ+Ipdxh+gAA02mijAhrbsGtYwojWhCQHoOt6pLQJN+m5dD86NYxgmPliZRkHtblNQ\nzJDz50mZMrnEawGaJoApC6jXoUFAAGhCg4CGF14t4cCrZLm+tllpXl+rAQ88AAiBZqEIY8M8DAOA\nTRaoDQ078SiuwCIWUcJjhQmM2HVoxSJ+a7qMbzwPHD1KIjkx4V5elWGY+NGHfQCpxrJo0JMpko2G\nu20Tw0hkGMAwIAojgKbDgI0/bU7gl5pV1OvAS3C2gWFQOm6zCbtWx397xsIHbQsjWhMGBAw0cQUW\n8SfYjcPFcVz4xDT+6/vX49hHpvE/xkysW0df610dJjTVKhWlVqvxngeGyShsmUahVHIaNQjhWKec\nLsl4Ufy/+oULEPsPQBM2Rpazd88WTFxXMYHK/ErDe0xMoHmJMoO/IsowBGAXijDsOvRCET+5oYxt\nq4Hta6tY8/EJiFodZZzEHcfG8O8fDV4svCOJX5KJYZIHK9NeqVaBuTkng1fXgY99DLia+7Ixbpyw\nuglzNzVvWHrqMGDXV9oGbtkiRUbxEY+N4buzFjYfLOOvmiY+YFSxeMdmrF4NGJUKdsntpizYnhVj\nFhfN3rN3E7skE8MkF1amvVCt0gjVaDiKdHTUpyiQyTv+Rp6J1/fN4//dbuErzTJeLZr40+Va0PP7\nq1ics1DaVMbYOHU9mqoA35qt4qOH1sM4WoetF3DuDDByDzA2bgauGNNz9m4il2RimGTDyrQXZmdp\noJH8zM8Af/RHrEiZFoKMvP8xZuKhgom6DRQEidQ/Hani9r3r8V7UUT9exHnMY2ycVn8xLQtYoh1p\nzSZ+8dQMaqcOr2xjnJjHd2ctvIQyrRiDKiU99eIl4aJUhukaVqZx8L3vAR//OKdMMi0EGXmWRb3u\nhSAHx8wM8FOahTtRR2HZXbs4ZwHj7pW67bcvQYOAAYERdRuTrNjrqmTFrju0HsZShJgnF6UyTFdw\nNm8v/Pf/3vpardZ+XUsmN6iJsNLIe/BBt06TSlauxCYEYIky6iiiAQMNFFHaVHZ2apo4Pz2Pk++7\nDzXPNvL79u8nl/LfzlgQtXr4zHLO3GWYyLBl2i379wNPPz3so2ASSlAibNDi9rOz1LO+2QTOFk0c\n3zmPy885MVPXfidM1Gom1qGCW2Hha4Uy/neYmFj+Pl2n/XxFlPGHKNL6p51inpy5yzCxwMq0W+bm\n/F83DKejOJNbukmElUq2UlHDkybUNoLe/do28ApMvAIThgBWzTnfJ3Ph/kozcYehrH/aTjly5i7D\nxAIr07BUq2RG/OAHre+NjHAfXgZA/xJh1dbP6tJqmzYBJ09iZSm3vXdY+GeU8G9WL3ZWpP08YIbJ\nGaxMwyBLYdQMXoBGtPFxLolhVug2ETasl1Xdb6nkXth7bAz4r3v3498+ux36keWmvJoGPGkAjz1G\nG3ibBcsPtztgte80yzfDtIWVaRgsi1IuvQhBTRp4oGEUukmElR0pbZuew7iFW15HFTi6HbCbzotC\nULrw9u0Ugmg26VnT6HVvQNe7Y46lMkxXcDZvGMplcuV60XV2izGRUDtS2naIFV78Mm9nZ0lZ+mHb\nTs/oRqM1Phr0HZOT7r7T3GuaYdrClmkYTJMGk717gWefdTI9Hn+cZ+tMJBYXSZRkHHRxsc3GftYi\nQOnAXnTdkVPZ8tJrmfpNBOV3eIOzPGlkmLawMg2LaQJf+ALHkZieCBKbcpk6UYbK//HLvAVardJi\nkdZde/hheq9QALZscbLN28mvmjas68Btt5GVyrLOMG1hZdot3BmG6ZJ24ceuEpaCMm/la4YBbN3q\n1NrYNj2Wltyx/XZf4v0OVqQMEwpWpmGQZTEAZ+4yXdOplDNoftZizQZp3iBtXCiQMi0UwrtpuS8v\nw/QEK9N2SCX65JM0uweAQ4eAEyd4kGHaoirCXko5A63ZoHZK3pIWNSlJiO4Onr0vDNM1rEyDkKPZ\npUvuwYi7xDAd8FOE3Rp7QdZsx5C9n9w2myyzDNNnWJkGYVmtihTgzEamI36KcPfu7nSZnzV7fn8V\nzz9g4St2GQ+Omv6ln/LLpdxqGssswwwAVqZ+VKvAl77kVqTXXw/cdBPHTHNOmGTuODr0tYQuUUVj\nx3r80VIdv48ibq/Nw7LM1mNQv1zN4u2XzHJ2O8MAYGXainSTvf22+/Wf/mng858fzjExiaCX1n9R\ndIwrdDllodCsQ1te6/RXdQvlch+/PAzcJYlhVmBl6kW6ybxs2jTwQ2GSRS8rwqhEMuLKZWijRVqn\n1CjiN/eVMRa0j7gTiIIOnFecYZgVWJl6Ud1kmgbccANwzz3U0J7JNVHct5GNuGWLU7MsjJTLGItL\naXXS8O0OnFec6Rn2jmcPVqZeuM6OCSCKaMRixPXD4uyk4dsdON8rPRHUFZJPY7phZeoH19kxAfQq\nGok04lRFWatRtyNvx6NOB873Ske8Vqh3fjI7Cxw+zKHntMPKlGEGQCKNOO+K4y++SCuN99zvkPHi\nZ4Wqp13TgIsXOfScBXgJNoYZEKbZfb1pX5GK8rbbnKVr/JZbS9yBp4cgL/n0NLVStm3g2DFncR/D\ncIx/v9X2whLls0xvsGXKMHnGNMm1e/JkwnzQ2SDIS7646F6HQNPodfmsWrTdlgt3m+zGyVDxwMpU\nlSSApYrJBt2MkOzK7RveUwuQxVgquYsGbJt6xCwtOY4B6X1vNoGZGWoR/thjnQsLLMv5bK3m7zaW\n4lEq0Wp9HK+NTr6VqTqF8y6czFLFpJVe6nA4kahvyFPrvSzT02ShehVaqQQ8/zwpQ4lUtDt2AGNj\n7S9VqeR81raBt95yv79/P+2n2aQhT66HEKR4mXDkW5mqAQ11gWXOAmDSDDdTSCTey7K4SKFogBSk\nail6G7BJ5JoFcn9+DrXFRVKSshvqQw8BP//zZNHu3w/cf79bUUtUxbt/PzA3R71q5LG1c1oErVKZ\nJxdyvpWpN5tRomYBMEzaSGQdDtPuskjrdWrKvwGbptGjUABOnaIwd7Pp71Arl2m7RoM+a9vA9u30\n944d/opU8vDD9Lx3Lz0fP07JUQAwOuo4OapV2ubv/g644gpgYaF1lcojR0iRyyV1w7io00y+lakM\naExOAl/+sjOV27o1+9MoJrtwDDSRhLks3vm9XPTnkUeAs2dJUT3zjDNUScUohHuFon37gG3b3Kvw\nPfWU2wHnR7NJSk9Ffod0AwPABz8YvK96nRTtkSPOa2Fd1Gm2ZPOtTAG6Yps20RRMsnbt8I6HYeKA\nY6CJpNNlURVuqUQuW6lYpqZIKamr642MuC1Tae2Oj1PcVVVoV11F1qWsbwXcClEmQv34x/7HZtvA\nn/85LajVTikXi8Df/m3r680muYKDlGXa101gZQqQxMo6O12n/xkmL6TZHMggQQo3aHU9wP/y7dpF\nClUqp1276CG3PX+e3L/SDbt2LbmQ2/HNb3Y+/t/9XXL/vvaa+3XbJuvYtv3bKIbJQgaSK66sTAG6\nKqOjHGNi8kfazYEc0c5NHLQU4IkTrdvLZ+myFYIU2A03AF//OimyKDz3HJXxPPtsa3xWxnHVNoq1\nGsV+f+u33FnIpVLrvrsR10ErXVamAMeYmPzCmb+polvvfbvtvQlRlQo9ZFbuO95BMdof/rC7Y3z9\ndXr+/OeDM4cLy5pHWqK2DfzFXzhZyEEOwrDiOow5IitTCceYmDzCmb+5JciGUIfBjRuBW291Moxl\nvFZFLcORyESos2eBJ55o/czSEilr2VJR3beuk6PQTxTLZecz7YouhjFHZGXKMHmGvTK5JkxClHQV\nyxrYWo0U3ic/CVx+uRN/feABUnKqIly71klHUWk2qQznE59wymcAUsy33da6eJGKt/WiH945YqlE\nCVz9FHFWpgyTVcIGjdgrw7RBFY+gBg6m2fpetUrKFyCF+s530go5arnO5ZeTO1h2ZBodba9ILYvi\nrkLQs9+qgfJ45ufJZX3xIvDxj/e/uR0rU4bJIpxYxPSBdvMu73vS1SqLJP75n533VFeunyJWUeeE\n3laJX/5y66qBKocPA5cuOQq8ny0T86dMk5pXzaSC1IgPJxYxQ0Z1tcoewDK5yOvKDVLS3jnh5s1u\nt7HarML7eXkLqPHcdlnCUe/rfClTnq0zEUiV+HBiETNkvA0o1Gb+7Vy5Kt45IeA0npAWb5B4y1tA\ntUz9soTjuq/zpUx5ts5EIFXiw4lFTAIIE29tR1D5jl+HKL/vlnHTgwfpvvVTvHHd1/lSpjxbZyKQ\nOvHhxCImQfQijmHKd8J8p1TAfoo3rvs6X8qUZ+tMBFh8GGbwxDEn7JQ4Fcd9nS9lCvBsnYkEiw/D\nZI847ms9nkNhGIZhmPzCypRhGIZhIqIJv2aLQRtr2g8BvNm/w2FSxDVCiCv7/SUsc4wHljtm0ISS\nua6UKcMwDMMwrbCbl2EYhmEiwsqUYRiGYSLCypRhGIZhIsLKlGEYhmEiwsqUYRiGYSLCypRhGIZh\nIsLKlGEYhmEiwsqUYRiGYSLCypRhGIZhIsLKlGEYhmEiwsqUYRiGYSLCypRhGIZhIsLKlGEYhmEi\nwsqUYRiGYSLCypRhGIZhIsLKlGEYhmEiwsqUYRiGYSLCypRhGIZhIsLKlGEYhmEiwsqUYRiGYSLC\nypRhGIZhIsLKlGEYhmEiwsqUYRiGYSJS6GbjK664QqxZs6ZPh8KkiVdfffWfhRBX9vt7WOYYFZY7\nZtCElbmulOmaNWtw+vTp3o+KyQyapr05iO9hmWNUWO6YQRNW5tjNyzAMwzARYWXKMAzDMBFhZcoE\nU60CU1P0zKQevpxMXLAstdJVzJTJINUqYFlAuQyYpvv19euBeh0oFoH5eff7TKrgy8nEBcuSP2yZ\n5hl5V3zmM/SsTjMti+6WZpOeLWtYR8nEAF9OJi5YlvxhZZpnvHfF7KzjuymXadppGPRcLg/5YJko\n9Otysrsvf6iyZBjAhQudr3+QnGRJftjNm2fkXVGv011x6BCwtOT4bubn/V3ATOowzfgvJ7v78omU\npdlZGjIOHAAOHw6+/kFykjX5Ycs0z8i74sEHga1bSZGqvhvTBHbvTreEMyvEfTnZ3ZdfTBO4+urW\nIcOPIDnJmvywMs07coStVNity3QFRwLyTdjrH7Rd1uSH3bwM0Q8/IJNpWGTyhzf5P8z1D9oua/LD\nyjQPBJW/+L2XdolmBgqLTH4IinEGXf8wQ0uW5IeVaZapVilL4OBBCkx4o/xZywBg+k67eRmTbjpd\n23axTy5TZ2WaXaQ0X7oECEGvqYlFgP/dkXWJZ3omjwNkWog6yQlzbdXk/2IRKJWCP2NZQK0G2DY9\n52FoYWWaVaSilIpU01qj/N67I+0ZAExf4blXMoljkhPG6vTGONvJQ6lEihSg51Ip+u9MOqxMs4qq\nKAsFYMsWYO1a5y7xuzt4ZGTawHOvZBLHJCes1emNcXo/MzVF+1pcBHSdFKmu0/9Zh5VpVvEqSiDc\n3cEwAfDcK5nEMcnpxupUXcryM6USMDHhHMP0NDA6mq+JFyvTLKMqyqkp9tExkeG5V/JQFWGp5HY+\ndbufIKtTKkM/l/Lu3e7h5dIl4OzZ1olX1pPXWJlmFa/kelsHyoaaWZRqhskZ8jaOM0Fs82Z6rlTo\neWqKhg2pNGs1YHKSHuUyDSvNJqVpHDpEn9u9mz6bh+Q1VqZZQVWegL/kdtNQk2GYVBEldrp/PzA3\nB1x/PfCjH9EQ0WiQgnzHO4BHHyXlqWn0kPHQF18ETp6kYWTrVmBmhpTp0lL+CgdYmWYB77Rv8+bW\n1WCkovVrqJk1qWb6StbddWml19jp/v3AfffR38ePu9+zbeChh0hBysIAXQduvBE4fZrel8NIpULz\nc7/vz0PyGivTLOCd9gHBq8FMT2dfqqPC2iKQPLjr0kqYBDE/0Z6ba79f2yZrVP3/qqtaE4zafX8e\nktdYmWYB77SvUqGHZVGQ48ABR9EuLmZfqqPA2qIteXDXpRFVSapxSm8CkJ9ob9rUapF6+cVfBM6d\nc78W1G/XL+s3D4UDrEyzQLtO0tVqq+8l61IdBdYWbcmDuy5t+ClJgF6r1cg5tW8fzaP9RHt8nLaf\nngZef91x50o0DbjpJuAb3yAHFwAcOwbs2uUo7jDHlPXbiJVpVmjXSZot0fCwtmgLi1PyUOd/ly5R\nisTVVzvt/GwbeOAB4LHHgkV7bAz49rdbFSlA265d635PJhjJ7/fKQh7npKxM8wBbouFhbdGRfosT\nh6w7o56jcpmanMmylIMHKfvWMNwt/RYXyfp86ilStPfcA7z73WRhWhZ93ouuA488Qp9VlalhtO/N\nm8c5KSvTJMCjR7IY4OSDL72bPLoHu8XvHG3Z4pSlNJuk/PbtI4vUtilZqFQCdu50chQB4LXXgC9+\nEfjEJ9zKVyIE7atcpn3UaqRg27mNAcBEFa9ttvASyriuYubiGrIyHTY8euQWvvSt5NE92C1+58iv\nLMU0yX1rWcBbbwGf/axbkUoaDeBznwu2TEslch1/5CPA6tX0XTIdw9f6XBbsa+p1VIpFoDIPIPsX\nkZVpRKRlUSo5M7iubv5uRo/IX8YkiTQrjn5Z1Hl0D3aL3zlql4N45Aiwd2/w/kxUUV6ycAJlvKIo\nPcMAPvUptzU7Oup0RDJNchvPzVFG8IocpFmwI8DKNAL795MbRcYqdJ2ErSsLI+zoIc0YmVXQ05cx\nSSKtiqOfFjWHrDvjd46CJjf795NF6uVd7wLeeAP4ZVHFi1iPIuqoo4j1mMdfaiY+/GFqE2hZZLlK\nVN1YrTrN7U+eJCvYNJFewY4IK9MeqVaBHTucVHHA3Q0k9CAQdvSQsz01oyBHs74s4Fd3163iSEKM\ntd+GB+fLdcZbz+ktgxkbI9fsgQP+pS7XXgv8/d8DZWGhiDoKaEKgjls1C1+/zMTkpLP/kRF3Lxip\nGwPlIKczIlamPeKX/abrPU7E1AaW6v8qcranWqY5mvWlnSBrrhvFEVRPGPeYJfu0btrk1CCq5NTw\nGBrVKilGwB2vlNfdstxlMNu3k1JtNPxLXS67jHrwHj8OWCijjiJ0vQ59pIj3bCljvuJ2E1tW6/cD\nHeQghzMiVqY9oma3GQZlw11+eftBLdCqCOM3U2d7b71F7UhcgQomycRhzXn3MTvrTjrxik0vVqxf\nn1avQs2p4TEUqlU6x9IyPHSISlV27iRlOTJCrlaVZpOUqqpICwXgk590xiipHF+BifWYx+9cZeGD\nnynjujGzZU4fewl7Etwr/UAIEfrx/ve/XzAOCwtC7NlDz2G2XbVKCMOgZ9dn9uyhNwB63rOnxx0N\nDgCnRRey0+sjKzIXx2Xz7mPbtmCx6fX7br9dtjSnx+23d3+c/SRvcrdnjxCa5lwPTRNi3Tr3NTIM\nZxtNo/8LBXoeHSU5Ua//woIQxaJ7H4AQIyP0eluZ6WbQ8yMh41c3hJU5tkwj0I0nw9cygZKdG9Zv\nltNMubQThzXn3QcQvEpHr2Li7dO6aVP3x8nER7ncGrO86ir3NtIKlcujCUGW6JYt1LlocRE4f97t\nFvYrg2k0nM/7ykw7D1pYazPD4xcr04iElSFvfOHOkkcwp6fDlbuUSs6CghywShW9hJE6NQsPUtC9\nxjXHxmggXlqi57Gx7o6XiZfz5ym+edllwPveRzHL8+ep3EVSKDgru8i4qUyMnJhoLQCQC0fJ1yUj\nI7SPZtNfZt6ctfBzl+rQhUcRdpPereZ+aBqNZxmBlSl6d+F3I0Neq2LMO0NbXAzuGq1+4cQE3QGG\nQXeF9wuzGo9QycNvRPhQepzxLMtyYm1CZMpwSBydxFiNXwPUbN6ygFOn3Nv92q8B69aRXpKlKsUi\nvedXAKAuHFUqAWfP0vuyftTvmKpVYPfBMo6JIkZQh14owmiX1hu0I1mcumMHbT8xodTUpJvMK9NO\nAttzzVy1itqkhRtqZbxsm6E8Fu6Br9zZdPAevJq2B9BdEcuPSRF5+I3LRPWI9WIJc6buYAgjxt51\nRh96yL2uqGT1amceLjseqWEAvwIAtYBAzdAFfFy7loVvXSjja01KVvpVzcK7t5RRCUrrbde0F6Bx\nS5rQGXL1ZlqZhhHYngas5R3fUqvjuF3E7fo8zhRNdzutTiZBJ9PB7+BLJfc00+siyXA8YoU8/MZl\nhqHYOFN3MPhlZnvPuSxfkQjR2jsXAN7xDufvoDCAaoECwWOja+iCs9FHC0UcNObxNZj4etHEfAXu\nL1WFptM9mtEZW6aVaZhxt6frurxjzW5ilV7H/32bhdHJ5WbO3fp+OzVp8LpOdN2ZZnot04wKqYs8\n/MZl1DGqVGpfhhyn5zuHJYIDRxVjw6CSl6Uld/3wo4/Ss6YBH/4wdRmq11uV6rlzwddfXku1W9Hh\nw8Dmzf7Dixy6DAN48l0W/v2lOgzRhIE6Dt9r4S+uNv1lzCs07e7RjM7YMqdMvUsTdRp3213XwAFK\n2bFWLKI8WcZKS8u4LKeggx8dzZ2Qusjwb/STN/ncbn42CM93TsLUA0MV4wsXqFORV7HVavQsBHDi\nBNWynzsHXHkl8PTTzr6uv747D9ylS8DFi63Di7pNswk8/s0yNsGJkV5TKWN3u2uvCkmne7TdjC2l\nwpYpZSoLnGUxs2WFG3f9rqvfvla2MU2cn57H4pyFt64v4zXLRHl5P14leL5Uxtn7q7gFFq5Z20WD\n+iClEUVIs0IGf2M7hRg0P5NjzqlTNEAGljT08diY3lGtRr8SJ+mEAshqffhhkgHDAD76UeCHP6TS\npXZLoUnKZfeap8eOkeXrHY6KRUeWZEOHW2HhPWqMFGhVeH5C0imh0o80C1uYYlT5SEohcxAbN7qL\nkLdt631f27YF70vWHes6vafrSv3xwgJtvG2b+OuZBXFLcUH8GKtEA7qwWzZOL8hZ8Xy/ade3w6/O\n3SuD8lEsRqun96vH76anSL/JqtzNzFCDjJkZOv8bNwqxZg01YZCNGNTmDYWCc53C9kHYts3ZR9B1\nlMPXyIjzXaOjnn36fWFcQpIkYVsmrMxlxjKtVoGjRwfzXdJS8Kacf2u2CvOwM6s6iwpubshG0jaE\nunGGE2eY7mkXkvBzUkxNuWUQoNja1q3099RU+wx2r3NDroAkF5JWDYIchakHgp9RJ+OZL71EVqhf\n32/Aed3+/9s7/xg5zvO+f2dmd1YqCpfN2g2VwiJQgY1kZ+uQcQhNGJNjnyrXjIxcQhiBLWAPJEPq\nLMn2sXUY0akKAmp5BuM4F0VKfZZ1Fx5iJ0HCRA4dK1K04lBn3uqHLSq62pbMqqGY2mZRXMG0NsSd\n3Zm3f7z77rwzOzM7szv7854PcCD3x8zuzjzv+7zv89P1ppCkno+wnqdBDIMHHj0AC3961cTfbzXa\non1DTSVZCckYC9vEKFPL8tei1DQvb6obymUeFCDuqXyuu4pVvKVYqCgm1pjRCjnfC8snZHthYSlv\nwrZ1ADVooAL1E0+X/p5OE2LQsi3nvssJ+Tt2pPevAv4OSLWaf603wW7qgRN2/WXdFKypK3Ac4NZb\ngddf568XCu0Lrm48R23iKjX2PhZlZg1TeFkJyRgL28QoU9P0Cs+rKvDoo73dBwNVvH7AwnmY2F42\nvHNVqyjNTeFnXBsP5nT89dEKvruFR7htgwmc9oRsW9nEfNnAn69U0vtMifGjx3JraWIygpG+Qqw6\nxb9FBYnLOyFVDY9tI5HtnbDrXyzya86Y1+0liOsCr73GX9+xAzh0qLv7Id/HUHG1LG+Fdv06b2oq\n92MTJ4nqRJ6FkIypsE2MMu0mKjfs+WqVm2vvXp7CtoaNsq4D5Qpa4bpSWkxesTG9xcJ0K8Qt8CUA\nGNY8jLIJGF0444nxIi5SKEFQRZycpmnfFmcliwsSl/thTlig5cgQVt9gbo6LjKrya/rcc9779+zh\npQSfecarc/DSS7ykoCj12O39CBVXOZedMeBv/5bn5ARldtAKbwwEb2KUKRAdlRuVnBxm7pqa+bPy\nbwAAIABJREFUAo5et8CYDSBkeR9n05dvuDiZSNo6eLC91AgxWUTJRqdwXNNEFUbqaN4wkpiLuwkS\nB8Y70HJUCF7/YPzFhQvee3Ud+Oxn+f9XV70oWxG13akFXydCxdXa8IcRZxUi3osyHBPBGwll2s9F\nR9REFGXusm3gWWbit6BDUWxoQsqCOVSiIaD8I0QujaYBO3d65hLHARYXueQnLWhPjB9Rmips1gpM\nEJdmKrBtI1Rhpo3JkFMuwgKRol7vJI5plDrRjjyFyFkj4t6qqmduVxSePwpwl9N3Z3hA0INfN1qF\n6AF/7ujKSrr7ES6upmemCNYg7OWH99LVflwEL0nIr/jrR7h4v9vbRZ0/Lt1A0xjbq6+xy7Mneaz6\n7CyPDxdvXlxsPziYSxP2p6peo8ExT4/BhKYo9I1g3kkgBeDy7MnYcZC2jWSncdXNuBuFVpTDlLte\nWnnGXTuRFnP33TwlRU6526uvsUbBO/DVxbXWdwj2JW1LYekW8UMXF3vrXSoIprvMznYWJPliD1nw\nksrc0Hem/V50xPnKw55fWOAFpvfvN7CtBL6iEvYVgH/JM2fal4RhKAqwbRtw+TJ/LJwe4jyjusIi\nsicqHFcKVquUk0fzdqKTZfnKlfTjLutAyzFwg7Xo1dIYdz/kNmmiJ6lop7a7bkGBDTTbnpU2LJSO\nex988CA3ejHGo7EzmVKCUUrCbCd+SNobFjStAPHCF3axxyDCd+jKdBBpRVETUfB5Od9rdRX44IyF\nbbWaP1Y9l+NlR559lj9mjOfQPPwwL5Ukh+IxBvzDP3gdd2U0jdJjNjMhmslAdvNEJ8tyLsdFUKRi\nXLvWfo5gCECYibITaYOqRpVeF/2d3Omyi1JRvB6lFzQTTNGBRvgEmSR3tGvkm6Rp/IvJBYSTXAAh\nALJ7C4j/0mEX+/jx0RYQjIAyHXRaUdxqOHgPv321iG1SVjwD4NRdvPkGcIusHOt13pJBVf3t7sWf\n6LYsH7Nv38gLB9Fn+hgRGTauRKEH4ZfbvZtHjjIGnDoF3HILcOQIfy2reTSLoKpRoNdFfyd3elgD\nb657DGiIniD7On/KN6mbgKQ4AYj70mNauGHoyhTob5R1VIBt2KQQvIfv3upFtjEACgDNrUM/9RAY\nXLRaC4ryJEJhKgqfgRjjJ1pYAB5/3N/Vd+vW/vxgYrTJyLZZrXrehagg8TDLstiNahr3UMicOeMp\n017n0eA5ajV/yuK4zZdZKK2weU4+bzEyDT1+guzb/CnfpOCKqtMNq1b5zRarhKDgRKVeiAsxM8Of\nG6MMiJFQpv0iWCItrO1QMMpRHjC8CEMBeOst33n/Jf4n/4+wxzzyCE/6kk0XwajdUsmL9s3neyvP\nRIwnPdo25bnmk5/0uoosLSVXcKK5tKJwcZTXdyJ6FOhtHg2eQ8ynzzzjpSwC4zdf9ktpDSRls5tF\nXHBCBJKdQ8h5mojgsGMKhfh5ctSc7kmilMTfICMr4yLnOkXVieBaEUAmIuOigshiz9c8maPlmQOw\npouJOYrKQ/Dkg5J8sSyi40YAUDRveqKKeCeQCzmgMZfzFz1XlGT1wMM+/tgxPjYUhUeGzs6Gj4tu\nRXdtjQ8TEaGaNJgzikmVu75ODYOOhpUFTQ2ZJzsdI/7iCt0P8DcllbmRFLC465Q05F+ebOQuC8EJ\nYnaWTyId78naGvv2nln2FgrMhsZ+DB6mvlmZ1Emtr3TKxwoIoCyrwflJnneSdorp1OxDKOZCwa9U\ns/7Z8kJXKNekimQS5a7vemHQnVg6TeBhNzu2FVcIA/xNSWVuJM28ccEJnQIXxOusGesjW2Jlc72o\nsWDb/LnbUcUHrlu4tGLCCDEZVGHgxA0GfqSUsYdZWFVN/NKGgdKomRqIwZH23neKCpIEOhgAtG8f\nl2PA8yJcvMgfx5lJgzEDYaZVXfdX16nVvBojWUTZBn/2+jofl6rKf9vycvrgpkmi78FYg3JQRzUH\nB7ici9qJYW6OZM7jwf+mNCTRuGzAq7UkO1NV5QnOi4vRxwbNVjJyjYXbIXqOajxBOnBA1KLp1cVm\n1rSwkWWV5DwGYAJ3CKnIajsRcZ6wHWOYPHdyh8hjQa47EjyHqEsiW3T6seAPjt/pae93KkrnHsST\nKHcDsVj228WUpDqOXJFCCFevlTAGMN8mlbmRFrCo67S4yM22YZYAMTF0MlPNznIl+gBOsjOYZg4U\nT1MGbPyRLoBg1SNNm4jqRkmYxEktFUnMTEkHe8j7wtwVSZqGR33FYCGuuMbQidweXRJWDCdNFZ9J\nlbuxD6eIGg/ByVOuABdWSW4ESSpzI2nmBeIj3DY2vGJCcpcgwB8sGRcIdu+OKm7BFHTRZxQ8j1QJ\nhh0aRptF4cQJXi8TL7/cSplpHcsyKgxNjDZyqKqqctOUTJrI3RBhF1avlRW/GVS2ZnUyD5omNw0H\nG027bvvXlb9GOaYSUxxJrN7BsSTGaOZVfMaMgUT09pOo8SA/r2nA0aPAli3J+gWOG0k0LhvSai2K\nxUX/SltRwgMbgqtv3+rv5EnmqvzNrQjd5h8D+GvSCXzHNrcErqIyF2AOFFZDnjXyfVzSjxiY0B1C\nKuJMJBkGSHSK2dC06KCh2dn2YLyonWkvpDFVBn9PmmNJ7kaYqPEQ9fygo4y7JKnMqUPW5QC87hXV\narL3b2x4dRIAbzMI8EWQpoV3R5uaAh58kP+7XjShFHS4Cj9RAypsFGBDRx0a3nJ1nHrR5N+pWoVh\nzeO4WeULp+aKSmEuHKh4Gv8WH1DP48uHLOChhzZnFMVmZGODC5+clC4QK/IwYUyJYYRXUxO718OH\n+dd47DEu2/I4Kpd5P0wxXkT6XtbxGmGbjKS/R/wOGjpjTtR4iHp+wm780M283eSxm2Z4l6By2TNR\nFYv+exYc7F/bMFCqVPDNUxbOPXEN78Er+EtlP/4bStjDLFgwoTwBbDn7cexSl6C5jvcFmxMlq9mw\nXR3/WT2BlwsGtpcx9gJBpCAuojCqZE6GFZDEaW6+mct1WCsu+Wtcuwa88govLZ21mGZRbo+GzpgT\nJQTFIp+kGQsfJxNy44euTLsxmyeJog4q6LuKVRSxAhfAn2hlmKaBKgw88NfAk5iCDhsmO4/TygH8\nIbgj5xlMoeBch+pIHWMsiy+rKxUoloU3ijxF5rfNiZEJIimdaswFJ4q0K8cIxRs8zcIC3wA7jtd3\nQU59Ef+KY1ZXeUGuLOW1rzViNwETkWEXJgSie4jjcIW6sJDtDxyhCzd0ZdrNirbT9Qsq6EsrVZSX\n34+fcXj9tcNsCRoszFsGdtct6LCRgwMNDg6zL2AGy1jGAeiwoYGBAWCKAkX+gs2JsgSglMF1IMaU\nNCvrNCtHORFa04A/+INW4dyVFS8v1Lb5YrJTK65BxHpM0CZjoIxbB51YgkIgBE/0l7t4sb1jfbcK\nccQu3NCVadoVbTCZ/eDB9qT1oILeC4v7OJuva406YFkoFg38DxThQoELHpWrAiighp/EVdjQwWDD\nQQ7f+fkD2Hloh992TBBpSLNyXFnxAgEcB7j3XqBUQhUGlpe9oiRyJ7+4rlZpPnqEFvubgokKag0K\njxzNCwBf+pJn7hVFmrtViCN24YauTIHuF/eO46/UIl4XBTgurVSxFxa27Sj6b2g+D5gmblip4vcw\nBxUuGNSmSuX8L2zFFCowwf2nv2kCO+dGZxVEjClx1d2D5YpkXBewLFgw0GjwpxSFLyYBLuvWBy28\nttXE9rIRGazUSUkGm0N0K+akkJMztGI+Wd+kqJ3iwgJw//3+NpTCuX/zzd0rxFGrgpQk5Ff8DTtc\nfHGRsV27wgt9T0/7o6xfXWymr6gas/M3MuvuRfatXbPsB9Ne/sDl2ZOsDp6+UIfKHEVlDsCuI8du\nx5ovleDcnQOubznigFIUognLZUlaVFpOaM/nPSFvVjMIe9tePb6CV5qvLX9k0gL6wZ/dz1z8SZW7\ngRdt6EdaSpLCDfJfoZBMWDqV+QobaxlezKQyNxI70yR88YvAPfd4j/fs4Quhep3fmbNn/dHXG2cs\nsJoNxXUA18bffHkDp9T/ikIBeOFDVZSseTTeVvSZcnXVheIyMKZBbVZiEKkExf0msDpCqyBiNIla\nncsmlWDIbVjvx40N4Pz5tqalBvw7TMuCz+/vRKzuk2xCLAutXS/Ax1NYcYdOP1tVvT6oI2B9GwsG\n7m/O2kRarQJXrvgLSIs5Ut5BAl4VkUaDy3mcyaSTX7TXIL8MGRtleuaM//ENNwCHDnlBF8K/LdJk\n3vmzRTjPqAAY6tBhwYTrAu95q4rtH58Cg42bNR33Kwv4CbaBXXgRv+x8FQoYCmoD/+UOC9/bb7Qi\nhUuGAZQoXJHoQNQk1SxHxJoht+7jS9BE+Z+o3o/yRCGSsU0ThuE34x7Pm7BtvihUcjpWrpjYXm0e\nWq3izRULx5dMfMMxYucXUTGpXveeO3MmWeSvHBTFmotQRaF158iSpYk0GMhy+HB7OLmcfiEXuw/K\neZC0Sn+YftQk21fxNwyTm9ixHzvmtxAsLrYXoFcU/v9H7l5jjcKNrA6V1ZBjv47F1nEPwDPtOqrG\nHsydZLvVNVZDvlUJKXFPq00MJtTc1jMx5rMfTM+yRrMGtA2NXZ49GVn4OaziVpQpbG2NsdOza+yF\n6ZNsr77W5upwFN4y8HasdfRQiGI1ilSqOklFo0LBG5v97PlAcpchWZhD19b8DWuTdCtI87lpzdF9\nMF8nlbmRFrDgdTl2jN83uVOMuJeyD/Uz6knWULTWpHUcJ9ltt7FWh5i3oLMGFNZQNPbfj3Ffqitr\n6ne9i5RpB2hSiyFisjg9y32boh/u6dm10MEffOrybDJ/fdBlJfv5bWjsM8rJxPNRsKF3nAKWPzfJ\nXNoLJHcjRHA3k7RbQTefI5zxSZTw4mK7ouiBpDI3EuUEowju2LdsAZ56qpVu1yrz9/GfrYIx77hz\nrglH1dGACgYV13JFzM1xqwKA5rMMKnNwy8InsPOnrrbSZgAA3/lOe102gkhKRP2/7WUD+/QKTigP\nYZ9ewfZy8/WZGW4aa9pfg3J/HmarNKHTNOOGiWawgmFxv/eEWtDx0/eYPhNvVBlPw+Au3EIhWTVE\n+XNvuCG+wQQxQQRzSAUi0TkrhJtkbs6rBxs1N4siEZUK/3eAc/hI+0wjzfrVKnfSLC0BjoNfUnUY\nqKAKPktoGvD/dn8Q//wbZ6G6Dn5f+QS0ixdh/H4ZbzxuIfdiw1Oe9Tp+iK14h6ZDc7xcVIqcILLG\nMIB5y4BlGZg3wTsPhbQ5Csr99rIBlCt4c8XCzJKJbzxmQA9p3B1Mf5H9/JppohxTRanTueKGAVU/\n2qTIgqppXKGGtTcC2tO+0gpLUl/oEH2mo6NMpYu9vs6jcYv7TSwsGDhzRqonKmYBEe0AIMds3JGz\n8IJr4BeUKs6pU9BW+esKANRtYHERpRtOo7SwAPydDtbMOXXUPD769TIarIyytoJDch1eipwgMsYX\nazFvtQY+q9m4OLeC4k4LRtlEpWIE5hsDX7EMNBpV/IY7j9WaCcsKzyf1Pdd8oloFrHlPpIPBw2Fz\nTpoIU6p+tAkJrqKA8FrUch/BXI7P246TLtq2WPRHmEbNzUPMPc1UmXadAywtk101h+11htvgwH5a\nx/35Ci64hldPVKw8hF1XUaAUdHxkwcSNG8DUixaUr9pQGAMDWpWNFObVXlt/+Byq967AdYAvo4y1\nhgHXBZ7XDNxwqIzyzd38CIJIidQwoebm8K4Xl5B70YGzrMM4V4Fx3C9/dxWr+JTL60jbro6nr1Uw\nP290FFV5FyrmskajPXiYIFITXEWFpalIGx+4zcI4Yj5OsnMUplvX5TvguPq+QzSTZKZMe6rh7Str\n5CIPNGvi2thdt/AcDC81r2x6K49cDjhwACiXuUnri1U8f/YK/g3LwQXgIIcn8SF8WHsSOXjmh69Z\nBh6EAQeA0pxQhG9oe5mW2MSAaA788ycsvPb0Ffw6HovNFX3bRS+flMHGN3/HwmcRn+4C+C1f8lym\nqsAddzSb3ZPIE1kTsvGBrvt3pklWcUHf7MZG/PuHZCbJTJmmMVUHFe8LCyZKQkGqOdTrDC4c1KFj\nVTOBZjeMpSWgXDZgRLS2uvX+Kdzm2GhAw5dwGCso42LBwEsPV1Ha8N5vwt9lQ1V5zmpYhTeC6AvS\narJw4jj+xKqibJ8Ggw1VTDLCRAYA5TK+fbWIm6Ci0cydrjgmHHBzbdx4ky1k+bx/LiNFSvSNoE9V\nFFIH0u0co0y3WZRDzLCkYmbKNI2pOqh4RW9RWBZU08SldeCNxy1876dMvB0GlK96E8CllSqMMDOs\nZSHn2FDggAG4gpvxPAzoDPhRyb9SMQxg3z7giSf4Y9flJSJpUiEGQmA1+U8XKrjtoIFHrlbwa1st\nbCub/H2iawwAPP44/h1UuHDgQsW/VxfwvMsF1nVDKhU1J4n1oolPftJodcB6+GEe+CRiEkok9ES/\niDO5ppG7sPNkUeko42pJmSnTpKZqUXVK0/jjluKVtuY/AvCxdQP2t/gEIPrK/qJWxd3LU0Aj5Meb\nJpQC9z850PEcM4GmAg6u2qtV4Otfly5CjnxGxACRVpOsZuPP7rPwGDOg6wbeVzGwzQDPWZFLEdXr\nUBUFKhgcRcFHb72I93zn4wCAP1LK2NgIL6n209Cxw6ngeXCFeu3JKkpPicamOo/2JYVK9IteTK7B\nXaNQovPzXIn0GrWbceRvpgFIna5bMBAiWHVKvKl2wsLOmokLLp8AFIW//9Q+C9rZiB/f1OaKZeFS\n0cTFOQNaxC7ZsrzykIrC3a40nxA9k9RkJJlxGqqOZx0TTiCqdr1o4jY176Vr5fN8VdloQMvl8Iuv\nfwl7wAvpHmDLuFQ8B8Boq/OrwoYJC88308Zu+p41Um2rCCKUsF0j4FcgbTuylGQc+TvQ1Bh5IQCE\nmFabF3BvzcbTro47mrmjwsT7lR+YeK+ag+q64dvJpjYvAaiUoue14DXsNcmc2k0RiU1GQlgWFoCN\nDbxWNPFyYOFXrQJTcwZ2Mgsz2gp+dfdVFN+1FdixgwdfXLkC9QtfaOVEFxSbxwRU0Vbn11V0WI7Z\n+vgf/msT+PvhpA4QRGLCdo2A9xxjwHvfC+zc2X2wS8aRvwNVpp0WAm+uWHjndRsqc3CjauPYey38\n2itGK4z/pZcAmzEUlGaqSwxxu+Qsr+GINXsnhkWcyUgo0GCR70oFJcPAAuDLpZ6f52+54Bp4t7KO\ng6v3As81I4YsC288sY5/BUCMAKbmoJimNH7cVqju6/tP4OVPGFDqfHP7vmMGcIwqLBBDJMnuI0pZ\n6Lq3WPzmN4H19d52Q1lG/iapOSj+sqhXGVXjeG3N68toS30Z5TqhcpF6V9XYuTtPdlUCMst2d1Et\n/CYdUI1UP1EFtuXn8/m2grdhh4nndqtrrA7NXzd6dpZ9a9csc5oF8x0o7Fu7ZiPHD2O84P25O0/y\nwvcdfsJAe2p2AcndmJOmEH1Ur9I0haMzIKnMjYyACaV0O9bYZ5STvAh4E3ly+TFuZI7KC4XvVtdS\nNwbIuqlAP3rsjgM0qYUQGPxra7zYvKtKXWHEX1NYohZjrWMhdXAAGJueZo28zlyAuQB7CwX26uJa\n9PiJE1Dp+46LHJPcjTlZ7D4GLKxJZW5kygmKXf1LtoG/0w1UpJ27gSq+O2PhPEy8saOCjTMW/uMz\nPEBJkyxqSRsgZxl/QXVJNxexMiaZjIT5f2fNxNOujgJqUF0XioimW1gAAHzsyjye0kx8o1l8wTTR\nbOBgAfuLwPmmWQvgptsLF6DVebqMCwU/3vMh7i8tAg/pRvv4iRL4gH/i0kwFtm1QXBLRXzrljBaL\naDWRHsEqR7Ek0bhsQKu1sF39q4trzM7fyFf3ES2qZNOYqnJrmui+EzznuKzARx1swh1CGtmRF+C3\nY439De5kdUimqdnZ1skaBd6Ozde7VAjysWOMTU8ztm2bf4cK8Nfzed73rODtUH3fK+pLB3YIl2dP\njsW42IxyN3FETcqij6aipBPCPvsnksrcyOxMgebCHs0VCkxUYeDJ+yz8p0azGEPNhmJZMI4bbQuT\n+XnPL+26wH338XMG4j1GdlFDjD5prBpiAX79OvA8M3ACJ/A+rEJRbGi6zt/UPJkGm9eDNgxe/F4W\n5M99ju9kRfEGme3bebtAAKjVULq4glIZrfHT2imHCXxgh7CtbKJS9jYHIngyyfigaHYiFcGgH8vy\n1+9lrHNZL8EIRYAOTZkGB2C1yqsb3b08Ba3hmZ6edU08AB0MNqDpyDfNAsH7YZo87UjUHnVdHiEZ\nNvlRhwuiG9KkpQkdJjoFvuQY2KdVcPqgVOHo9On2k5mmvzekaOsSxtvf7n989Wp4bl5UBZqAkhWv\npq2xPSJzGTGqhK225OeKRU+RClQ1Xd3eXv0TGawIh6JMgwNwYYHvII9et8CYDYBfmL2wMFs4jjtr\nFXxAtfCRR6LLnxkG8MgjfEfquryx8f79wOoqpdQR2ZDWqiEWbeWyOMbANvmgsBJplgXs3g0895z3\nPk3jk41YKQr+8R8BeCky//fy/8E/Eyt82+aaXFbYFZ637X2k38drWekLy2Qdg0BMGJ2KL+g6MDPD\nF5BCoSoKcNddyc6fReGFjFaEQ1GmwQEodpDPMhO/Bb1lCvNMTwZM00CpQ5DRkSO8TZv8evAxQfRC\nN1aNyGPkF4LlwfJ5r9Hypz4F/O7vtivTV18FA28x6EDBP3nlAhiaPXxzzaEtlOv163jxlIU9XzNa\ni00xZwQ/Ok1hmSG2jyTGgU7FF4TVReSPAlxe/+qvgKee6qzYsvDbZbQiHIoyDQ5AsYN8yQ6Ywpqm\np6j5ptl9rc16FfeY/DvESBIsD3b4MC8RZpr++pcCVQVcFwrEzlSBIhSpqJH5tre1VvuMMbz+xLfx\naczDgokXa0Zrzoj76CS7b4pBICKJK74QLEG3uOjve5pUsfXqt8toRTgUZRo2AL0dpN8UJnehAqS2\npw6/9qdPJ9+Vk3+HGFnCalzKwin7UQHgppuA73+/ZeL9Cj6K/fgLaKoNJd8c1mIX0ORj+GMwKLCh\n44NKBaZpJProKOSF6fHj6X8ysQmIWm2FPXf6tK8U5sBMHRmtCIcWgNRpBwnwwSp3ocrn+Y7UlUrC\npNmVk3+HGFniBrRhAB/+sNczEAC+/30AaO1M33nnu/HG/vtQurgC5/ElKIuPAQqgSh+hgEGDCwYb\nv/NhC7uan9HNXEILUyIxYZN7mAIQQpgk13QQ3zElI5Ua46PZPebnbJ4iA3AX0j338JeXltI1awfI\nv0OMOHED+tgx3jdQtGWToh8VVYV5wgQMAz/8lRX8i3qdt2pjSsunqgBgWg6O60LVdew6Zib+6DBo\nYUpkzpinWfSsTPvig5S6xzwDHVPg/RhlE5QXIZn8c8m/Q4wtwrlpWcC1a8CpU95rn/50K5LoHWeX\noYKBAWggB01TkANfdarNTjVZCD8tTInUTHjASk/KtJOpp+trZ1lArQbFdXGjUsNv/ryFp3YaPl9O\nt4uYMV/8EJNGp0ES1iAZAG65xWs1c+QIf86yoLqNVnTvEg7hpv9QxvSWmPNHfVbzfFHH0cKUSEWv\nfoG0ymQIirsnZRpn6unp2hWLrTQAhbmYPlTE9BH/W9JcqwlfEBHjSpLVaNTrR454SlRgmnByOpy6\njTp0/JFSxl1bDEwfT+kE1TQe8CRScyIGLy1MicQElcXKSnL/qJDNWo3L5tGjwJYt8QvQITj0e1Km\ncaaeRD6VKC23sdEK/Yeq8seBw5JeKwqUIEaWToMkjWOyOZauHF3A6c9v4FnXxMWCgc+ZXXwXkc+a\nNsKPIKKQlYWmAcvL3P8v5ng58TlI01LZKrF56lT8MUNy6PekTONMPR19KnFazjT5hYo4OM21okAJ\nYmTpNEiSOialsXSLruMjj1Zw44aB3zZTyHpwspN3puQQJXpFVhZXrgCPPeav/Ro3OQdrxXY6Rn6/\npg1MfnsOQIoy9XT0qcRpuWCYtMiXM8Lz4kwzepNLgRLEyNJpkCR1TAbGUmnDQimJaTfus8R5yTdC\nZIVQFtVq8pxSMbEfPQp8/vNcxhlrPybo7xd52cH87D7S19SYWJ9KJy0nDgzZvYaN+6hNLgVKECNN\nJ8djEsdkVivGsNy/ABR/QPRM0pzSoPXy0Uf5e4PHBN83M8OtKozxf8fBzNsTSbRczO5VXuScOOEt\ncsJ2/hQoQUw0A1oxUvwBkRlJJuXg/L+xEV5qK/g+YCjmyOEWbeh0QTusuOUgr0FXoCKIkWIAK0aK\nPyAGSlKLS1g9zG4KEfTI6FZAAtpW3FUYsOa96yMGt1Ckd9zBd6k0wAmiSS922cCxFH9ADJQwf978\nfLssR1lmwszGotB70gLUKRhtZQq0VtxhJqbg4CZFSkw8aYo8AN3bZUMGnGEYFH9ADBbZnxcny50s\nM9Uq8P73e23elpYyN60MrTl42gEZZmI6fpyCi4hNRNoiDzMz4YnycuBGVPRuhE2X4g+IodCrj0Ec\nL6jXx1+ZBsf7CwtVlDasNm0YVLhRJiYa3MSmIW2Rh6tXeWqAqnqJ8iJ3dGEBmJuLrniUJv+MIPpN\nrz4GcbzYmebzmcv0wJWpPN531qq49f4pwOUXaH2hgq9tGCgWvXEuL8BpF0psatIUedA04MknvcT1\nffuAs2c9RXvmTHzFo6DZB6BQXmJ49KoADAM4d87vMwUylemBK1N5vH9AsZBzbMB1wGo2/uw+CyeZ\nAVX1xri8AKddKLGpSVPkQa4yoyjA1q1+Rbx/P7C66t+Z1uv832LRO5/4jPl5CuUlhkuvCiB4fMYy\nPXBlKo/3u4omlDk+wBuqjmcdE47rFbhQFIoaJAgfSYs8iCozcekCpZL3eH0duP9+PrHMzfHXqJQY\nMclkLNNDCUDy5gMDKHHN+lrRxMtzBjTbc+kMutk6QUwMSdIFZMVsWV4h8ajKJ+RnISZGR7CgAAAB\n5ElEQVSJjGV6+KkxzQFdAlAp0VgliMxIYxZLskonPwsxaWQo08NXphI0VgkiWxIHK9LOkyB6YqSU\nKUEQ2ZG6li6tZgmia9RhfwGCIPpDWFoqQRD9gZQpQUwowg2qaSMcgFut8hSFanXY34SYFIYkU2Tm\nJYgxJs4n2tENmkX1l14L6VMhCCJLhihTY6dMqaIZQXCSzBuhblDRPWNpiduAu510ep24qKcbkTWy\nTF2/zuV8QDI1VmZeMXYffJD/S5YhYjPTlU9UDKLFxd4dqr06ZcfCDk2MFabJ5Qng1X+WlwemKMZK\nmVJABUF4dKWLxCBijD/upcxYr8pQ2KEfeohMvEQ2GAZw8CCXa4A3bxiQohgrMy9VNCMIj65SQ+VB\nlMsBBw503yg5i9xUSschsqZc9pfSHJCiGCtlSnnlBOEntS7KehCRMiRGjSEpirFSpgCNXYLoGRpE\nxKQzBBkfK58pQRAEQYwipEwJgiAIokcUJqL6krxZUf43gDf793WIMWIbY+wd/f4QkjkiAMkdMWgS\nyVwqZUoQBEEQRDtk5iUIgiCIHiFlShAEQRA9QsqUIAiCIHqElClBEARB9AgpU4IgCILoEVKmBEEQ\nBNEjpEwJgiAIokdImRIEQRBEj5AyJQiCIIge+f8qAno3EtLOBAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd105fa7080>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Examples (2d)\n",
    "\n",
    "for i in range(9):\n",
    "    plt.subplot(3,3,i+1)\n",
    "    train,test,labels = mkData(200,200)\n",
    "\n",
    "    l0 = labels==0\n",
    "    l1 = labels==1\n",
    "\n",
    "    plt.plot(test[l0,0],test[l0,1],'b.')\n",
    "    plt.plot(test[l1,0],test[l1,1],'r.')\n",
    "    plt.xticks([])\n",
    "    plt.yticks([])\n",
    "plt.gcf().set_size_inches((8,8))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "invar = T.tensor3()\n",
    "invar2 = T.tensor3()\n",
    "targ = T.matrix()\n",
    "\n",
    "train_input = lasagne.layers.InputLayer((None,None,DIM+1), input_var = invar)\n",
    "\n",
    "inp_shape = invar.shape[1]\n",
    "test_shape = invar2.shape[1]\n",
    "\n",
    "test_input = lasagne.layers.InputLayer((None,None,DIM+1), input_var = invar2)\n",
    "\n",
    "train_shuf = lasagne.layers.DimshuffleLayer(train_input,(0,2,1))\n",
    "test_shuf = lasagne.layers.DimshuffleLayer(test_input,(0,2,1))\n",
    "\n",
    "# This block computes the population mean of some function of the input and concatenates that\n",
    "# to the individual values.\n",
    "def mkBlock(inp):\n",
    "    nin1 = lasagne.layers.NINLayer(inp, num_units = 64)\n",
    "    nin2 = lasagne.layers.NINLayer(nin1, num_units = 64)\n",
    "    pool = lasagne.layers.GlobalPoolLayer(nin2)\n",
    "    res = lasagne.layers.ReshapeLayer(pool, (-1,64,1))\n",
    "    rep = RepeatLayer(res, n=inp_shape)\n",
    "    return lasagne.layers.ConcatLayer([inp,rep])\n",
    "\n",
    "cur = train_shuf\n",
    "for i in range(3):\n",
    "    cur = mkBlock(cur)\n",
    "\n",
    "# This final pooling layer gives the learned embedding of the task inferred from the dataset\n",
    "nin_final = lasagne.layers.NINLayer(cur, num_units = 256)\n",
    "embedding = lasagne.layers.GlobalPoolLayer(nin_final)\n",
    "\n",
    "# Make a copy for each test sample and concatenate...\n",
    "emb_reshape = lasagne.layers.ReshapeLayer(embedding, (-1,256,1))\n",
    "emb_repeat = RepeatLayer(emb_reshape,n=test_shape)\n",
    "\n",
    "test_stack = lasagne.layers.ConcatLayer([test_shuf, emb_repeat],axis=1)\n",
    "\n",
    "# Now process each example forwards\n",
    "classify1 = lasagne.layers.NINLayer(test_stack, num_units = 256)\n",
    "classify2 = lasagne.layers.NINLayer(classify1, num_units = 256)\n",
    "classify3 = lasagne.layers.NINLayer(classify2, num_units = 256)\n",
    "# Use skip connections for better gradient propagation\n",
    "classify4 = lasagne.layers.NINLayer(lasagne.layers.ConcatLayer([classify1,classify3]), num_units = 256)\n",
    "\n",
    "final = lasagne.layers.NINLayer(classify4, num_units=1,nonlinearity=lasagne.nonlinearities.sigmoid)\n",
    "output = lasagne.layers.DimshuffleLayer(final,(0,2,1))\n",
    "\n",
    "out, embedding = lasagne.layers.get_output([output,embedding])\n",
    "params = lasagne.layers.get_all_params(output,trainable=True)\n",
    "\n",
    "# To avoid NaNs when the network receives an easy problem\n",
    "out = T.clip(out[:,:,0],1e-5,1-1e-5)\n",
    "elemloss = T.mean((-targ*T.log(out) - (1-targ)*T.log(1-out) ),axis=1)\n",
    "loss = T.mean(elemloss)\n",
    "\n",
    "updates = lasagne.updates.adam(loss, params, learning_rate = 1e-3)\n",
    "\n",
    "train = theano.function([invar, invar2, targ], loss, updates=updates, allow_input_downcast=True)\n",
    "process = theano.function([invar, invar2], [out,embedding], allow_input_downcast = True)\n",
    "test = theano.function([invar, invar2, targ], loss,  allow_input_downcast=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAD8CAYAAAB9y7/cAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\neBGRJkoBLyLSRP1/ZdRW/WixG+UAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd0f29cf978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-19-4b7b070e6ab8>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     13\u001b[0m     \u001b[0mtlabels\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     14\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1600\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 15\u001b[0;31m         \u001b[0md\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtd\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0ml\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmkData\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mN\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mNT\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     16\u001b[0m         \u001b[0mdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0md\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     17\u001b[0m         \u001b[0mtdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtd\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-12-38a8565ed77b>\u001b[0m in \u001b[0;36mmkData\u001b[0;34m(N, Nt)\u001b[0m\n\u001b[1;32m     50\u001b[0m         \u001b[0mlbl\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandom\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     51\u001b[0m         \u001b[0mexample\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmatmul\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m==\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0msig1\u001b[0m\u001b[0;34m+\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m==\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0msig2\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mmean\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mvec1\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mvec2\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 52\u001b[0;31m         \u001b[0mexample\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhstack\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mexample\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     53\u001b[0m         \u001b[0mtdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mexample\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     54\u001b[0m         \u001b[0mtlabels\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/numpy/core/shape_base.py\u001b[0m in \u001b[0;36mhstack\u001b[0;34m(tup)\u001b[0m\n\u001b[1;32m    281\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    282\u001b[0m     \"\"\"\n\u001b[0;32m--> 283\u001b[0;31m     \u001b[0marrs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0matleast_1d\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_m\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0m_m\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mtup\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    284\u001b[0m     \u001b[0;31m# As a special case, dimension 0 of 1-dimensional arrays is \"horizontal\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    285\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0marrs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndim\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/numpy/core/shape_base.py\u001b[0m in \u001b[0;36m<listcomp>\u001b[0;34m(.0)\u001b[0m\n\u001b[1;32m    281\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    282\u001b[0m     \"\"\"\n\u001b[0;32m--> 283\u001b[0;31m     \u001b[0marrs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0matleast_1d\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_m\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0m_m\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mtup\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    284\u001b[0m     \u001b[0;31m# As a special case, dimension 0 of 1-dimensional arrays is \"horizontal\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    285\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0marrs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndim\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/numpy/core/shape_base.py\u001b[0m in \u001b[0;36matleast_1d\u001b[0;34m(*arys)\u001b[0m\n\u001b[1;32m     48\u001b[0m     \u001b[0mres\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     49\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mary\u001b[0m \u001b[0;32min\u001b[0m \u001b[0marys\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 50\u001b[0;31m         \u001b[0mary\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0masanyarray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mary\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     51\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mary\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     52\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mary\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mreshape\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.5/dist-packages/numpy/core/numeric.py\u001b[0m in \u001b[0;36masanyarray\u001b[0;34m(a, dtype, order)\u001b[0m\n\u001b[1;32m    581\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    582\u001b[0m     \"\"\"\n\u001b[0;32m--> 583\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ma\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcopy\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0morder\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0morder\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msubok\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    584\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    585\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAD8CAYAAAB9y7/cAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\neBGRJkoBLyLSRP1/ZdRW/WixG+UAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd0f29cf978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from IPython import display\n",
    "import time\n",
    "\n",
    "errs = []\n",
    "\n",
    "# Train the classifier network\n",
    "for epoch in range(1500):\n",
    "    NT = 40\n",
    "    N = 40\n",
    "    data = []\n",
    "    tdata = []\n",
    "    tlabels = []\n",
    "    for i in range(1600):\n",
    "        d,td,l = mkData(N,NT)\n",
    "        data.append(d)\n",
    "        tdata.append(td)\n",
    "        tlabels.append(l)\n",
    "    data = np.array(data)\n",
    "    tdata = np.array(tdata)\n",
    "    tlabels = np.array(tlabels)\n",
    "\n",
    "    e0 = train(data,tdata,tlabels)\n",
    "    errs.append(e0)\n",
    "    \n",
    "    if epoch%10 == 0:\n",
    "        plt.clf()\n",
    "        plt.plot(errs)\n",
    "        display.clear_output(wait=True)\n",
    "        display.display(plt.gcf())\n",
    "        time.sleep(0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.290264699218\n",
      "0.316139911628\n"
     ]
    }
   ],
   "source": [
    "from sklearn.svm import SVC\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "err = 0\n",
    "for i in range(1600):\n",
    "    d,td,l = mkData(N,NT)\n",
    "    \n",
    "    data = d[:,0:DIM]\n",
    "    labels = d[:,DIM]\n",
    "    \n",
    "    svc = SVC(C=0.5,probability = True)\n",
    "    svc.fit(data,labels)\n",
    "    preds = np.clip(svc.predict_proba(td[:,0:DIM]),1e-5,1-1e-5)\n",
    "    \n",
    "    err += np.mean(-l*np.log(preds[:,1])-(1-l)*np.log(preds[:,0]))/1600.0\n",
    "    \n",
    "print(err)\n",
    "\n",
    "lerr = 0\n",
    "for i in range(1600):\n",
    "    d,td,l = mkData(N,NT)\n",
    "    \n",
    "    data = d[:,0:DIM]\n",
    "    labels = d[:,DIM]\n",
    "    \n",
    "    svc = LogisticRegression(penalty='l2',C=50)\n",
    "    svc.fit(data,labels)\n",
    "    preds = np.clip(svc.predict_proba(td[:,0:DIM]),1e-5,1-1e-5)\n",
    "    \n",
    "    lerr += np.mean(-l*np.log(preds[:,1])-(1-l)*np.log(preds[:,0]))/1600.0\n",
    "    \n",
    "print(lerr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAY4AAAEKCAYAAAAFJbKyAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIiKSFSUOERHJihKHiIhkRYlDRESy8v8BxenQTw5VMBoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd0f23bed30>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(errs,'b-',label=\"2d trained\")\n",
    "plt.plot(np.arange(0,1500), err*np.ones(1500), 'r-', label=\"kSVC\")\n",
    "plt.plot(np.arange(0,1500), lerr*np.ones(1500), 'm-', label=\"Linear\")\n",
    "plt.xlim(0,1400)\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.ylabel(\"Log-loss\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeYAAAHVCAYAAAAkfVjMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd0f27543c8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(9):\n",
    "    plt.subplot(3,3,i+1)\n",
    "    d,td,l = mkData(N,NT)\n",
    "\n",
    "    xx,yy = np.meshgrid(np.arange(-5,5,0.1), np.arange(-5,5,0.1))\n",
    "    testgrid = np.array([xx.ravel(),yy.ravel(),np.zeros(10000)]).T\n",
    "    preds, embs = process(d.reshape(1,40,3),testgrid.reshape(1,10000,3))\n",
    "\n",
    "    plt.contourf(xx,yy,preds.reshape((100,100)))\n",
    "    plt.plot(d[d[:,2]==-1,0],d[d[:,2]==-1,1],'b.')\n",
    "    plt.plot(d[d[:,2]==1,0],d[d[:,2]==1,1],'r.')\n",
    "plt.gcf().set_size_inches((8,8))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Higher-dimensional case\n",
    "DIM = 64\n",
    "\n",
    "train_input = lasagne.layers.InputLayer((None,None,DIM+1), input_var = invar)\n",
    "\n",
    "inp_shape = invar.shape[1]\n",
    "test_shape = invar2.shape[1]\n",
    "\n",
    "test_input = lasagne.layers.InputLayer((None,None,DIM+1), input_var = invar2)\n",
    "\n",
    "train_shuf = lasagne.layers.DimshuffleLayer(train_input,(0,2,1))\n",
    "test_shuf = lasagne.layers.DimshuffleLayer(test_input,(0,2,1))\n",
    "\n",
    "# This block computes the population mean of some function of the input and concatenates that\n",
    "# to the individual values.\n",
    "def mkBlock(inp):\n",
    "    nin1 = lasagne.layers.NINLayer(inp, num_units = 64)\n",
    "    nin2 = lasagne.layers.NINLayer(nin1, num_units = 64)\n",
    "    pool = lasagne.layers.GlobalPoolLayer(nin2)\n",
    "    res = lasagne.layers.ReshapeLayer(pool, (-1,64,1))\n",
    "    rep = RepeatLayer(res, n=inp_shape)\n",
    "    return lasagne.layers.ConcatLayer([inp,rep])\n",
    "\n",
    "cur = train_shuf\n",
    "for i in range(3):\n",
    "    cur = mkBlock(cur)\n",
    "\n",
    "# This final pooling layer gives the learned embedding of the task inferred from the dataset\n",
    "nin_final = lasagne.layers.NINLayer(cur, num_units = 256)\n",
    "embedding = lasagne.layers.GlobalPoolLayer(nin_final)\n",
    "\n",
    "# Make a copy for each test sample and concatenate...\n",
    "emb_reshape = lasagne.layers.ReshapeLayer(embedding, (-1,256,1))\n",
    "emb_repeat = RepeatLayer(emb_reshape,n=test_shape)\n",
    "\n",
    "test_stack = lasagne.layers.ConcatLayer([test_shuf, emb_repeat],axis=1)\n",
    "\n",
    "# Now process each example forwards\n",
    "classify1 = lasagne.layers.NINLayer(test_stack, num_units = 256)\n",
    "classify2 = lasagne.layers.NINLayer(classify1, num_units = 256)\n",
    "classify3 = lasagne.layers.NINLayer(classify2, num_units = 256)\n",
    "# Use skip connections for better gradient propagation\n",
    "classify4 = lasagne.layers.NINLayer(lasagne.layers.ConcatLayer([classify1,classify3]), num_units = 256)\n",
    "\n",
    "final = lasagne.layers.NINLayer(classify4, num_units=1,nonlinearity=lasagne.nonlinearities.sigmoid)\n",
    "output = lasagne.layers.DimshuffleLayer(final,(0,2,1))\n",
    "\n",
    "out, embedding = lasagne.layers.get_output([output,embedding])\n",
    "params = lasagne.layers.get_all_params(output,trainable=True)\n",
    "\n",
    "# To avoid NaNs when the network receives an easy problem\n",
    "out = T.clip(out[:,:,0],1e-5,1-1e-5)\n",
    "elemloss = T.mean((-targ*T.log(out) - (1-targ)*T.log(1-out) ),axis=1)\n",
    "loss = T.mean(elemloss)\n",
    "\n",
    "updates = lasagne.updates.adam(loss, params, learning_rate = 1e-3)\n",
    "\n",
    "train = theano.function([invar, invar2, targ], loss, updates=updates, allow_input_downcast=True)\n",
    "process = theano.function([invar, invar2], [out,embedding], allow_input_downcast = True)\n",
    "test = theano.function([invar, invar2, targ], loss,  allow_input_downcast=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAD8CAYAAACb4nSYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/wOjtsxfITmUwAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd10ad43d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-79-3ea78132e3c7>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     17\u001b[0m     \u001b[0mtlabels\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m100\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 19\u001b[0;31m         \u001b[0md\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtd\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0ml\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmkData\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mN\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mNT\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     20\u001b[0m         \u001b[0mdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0md\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     21\u001b[0m         \u001b[0mtdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtd\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-12-38a8565ed77b>\u001b[0m in \u001b[0;36mmkData\u001b[0;34m(N, Nt)\u001b[0m\n\u001b[1;32m     40\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     41\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mN\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 42\u001b[0;31m         \u001b[0ms\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandom\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mDIM\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     43\u001b[0m         \u001b[0mlbl\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandom\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     44\u001b[0m         \u001b[0mexample\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmatmul\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m==\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0msig1\u001b[0m\u001b[0;34m+\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m==\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0msig2\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mmean\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mvec1\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mvec2\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAD8CAYAAACb4nSYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/wOjtsxfITmUwAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd10ad43d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "errs0 = []\n",
    "\n",
    "# The actual problem we want to solve (6-sparse) is too hard to start with, \n",
    "# so we first pretrain on much easier problems and then make it progressively\n",
    "# sparser\n",
    "\n",
    "SPARSE = 24\n",
    "MINSEP = 5\n",
    "MAXSEP = 10\n",
    "\n",
    "# Pretrain the classifier network\n",
    "for epoch in range(500):\n",
    "    NT = 40\n",
    "    N = 40\n",
    "    data = []\n",
    "    tdata = []\n",
    "    tlabels = []\n",
    "    for i in range(100):\n",
    "        d,td,l = mkData(N,NT)\n",
    "        data.append(d)\n",
    "        tdata.append(td)\n",
    "        tlabels.append(l)\n",
    "    data = np.array(data)\n",
    "    tdata = np.array(tdata)\n",
    "    tlabels = np.array(tlabels)\n",
    "\n",
    "    e0 = train(data,tdata,tlabels)\n",
    "    errs0.append(e0)\n",
    "    \n",
    "    plt.clf()\n",
    "    plt.plot(errs0)\n",
    "    display.clear_output(wait=True)\n",
    "    display.display(plt.gcf())\n",
    "    time.sleep(0.01)\n",
    "\n",
    "SPARSE = 16\n",
    "MINSEP = 2\n",
    "MAXSEP = 6\n",
    "\n",
    "for epoch in range(500):\n",
    "    NT = 40\n",
    "    N = 40\n",
    "    data = []\n",
    "    tdata = []\n",
    "    tlabels = []\n",
    "    for i in range(100):\n",
    "        d,td,l = mkData(N,NT)\n",
    "        data.append(d)\n",
    "        tdata.append(td)\n",
    "        tlabels.append(l)\n",
    "    data = np.array(data)\n",
    "    tdata = np.array(tdata)\n",
    "    tlabels = np.array(tlabels)\n",
    "\n",
    "    e0 = train(data,tdata,tlabels)\n",
    "    errs0.append(e0)\n",
    "    \n",
    "    plt.clf()\n",
    "    plt.plot(errs0)\n",
    "    display.clear_output(wait=True)\n",
    "    display.display(plt.gcf())\n",
    "    time.sleep(0.01)\n",
    "    \n",
    "SPARSE = 10\n",
    "MINSEP = 1\n",
    "MAXSEP = 4\n",
    "\n",
    "for epoch in range(1000):\n",
    "    NT = 40\n",
    "    N = 40\n",
    "    data = []\n",
    "    tdata = []\n",
    "    tlabels = []\n",
    "    for i in range(100):\n",
    "        d,td,l = mkData(N,NT)\n",
    "        data.append(d)\n",
    "        tdata.append(td)\n",
    "        tlabels.append(l)\n",
    "    data = np.array(data)\n",
    "    tdata = np.array(tdata)\n",
    "    tlabels = np.array(tlabels)\n",
    "\n",
    "    e0 = train(data,tdata,tlabels)\n",
    "    errs0.append(e0)\n",
    "    \n",
    "    plt.clf()\n",
    "    plt.plot(errs0)\n",
    "    display.clear_output(wait=True)\n",
    "    display.display(plt.gcf())\n",
    "    time.sleep(0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAD8CAYAAACb4nSYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niiH/D+M9Dd888PsFAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd0f235df98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-114-2325ecb672db>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     13\u001b[0m     \u001b[0mtlabels\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     14\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m200\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 15\u001b[0;31m         \u001b[0md\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtd\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0ml\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmkData\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mN\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mNT\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     16\u001b[0m         \u001b[0mdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0md\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     17\u001b[0m         \u001b[0mtdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtd\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-12-38a8565ed77b>\u001b[0m in \u001b[0;36mmkData\u001b[0;34m(N, Nt)\u001b[0m\n\u001b[1;32m     41\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mN\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     42\u001b[0m         \u001b[0ms\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandom\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mDIM\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 43\u001b[0;31m         \u001b[0mlbl\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandom\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrandint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     44\u001b[0m         \u001b[0mexample\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmatmul\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m==\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0msig1\u001b[0m\u001b[0;34m+\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m==\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0msig2\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mmean\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mvec1\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mvec2\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     45\u001b[0m         \u001b[0mexample\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhstack\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mexample\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mlbl\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAD8CAYAAACb4nSYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niiH/D+M9Dd888PsFAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd0f235df98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "SPARSE = 6\n",
    "MINSEP = 1\n",
    "MAXSEP = 2\n",
    "\n",
    "#errs1 = []\n",
    "\n",
    "# Train the classifier network\n",
    "for epoch in range(5000):\n",
    "    NT = 40\n",
    "    N = 40\n",
    "    data = []\n",
    "    tdata = []\n",
    "    tlabels = []\n",
    "    for i in range(200):\n",
    "        d,td,l = mkData(N,NT)\n",
    "        data.append(d)\n",
    "        tdata.append(td)\n",
    "        tlabels.append(l)\n",
    "    data = np.array(data)\n",
    "    tdata = np.array(tdata)\n",
    "    tlabels = np.array(tlabels)\n",
    "\n",
    "    e0 = train(data,tdata,tlabels)\n",
    "    errs1.append(e0)\n",
    "    \n",
    "    plt.clf()\n",
    "    plt.plot(errs1)\n",
    "    L = len(errs1)\n",
    "    plt.plot(np.arange(0,L),np.ones(L)*err,'r-',label=\"SVC\")\n",
    "    plt.legend()\n",
    "    display.clear_output(wait=True)\n",
    "    display.display(plt.gcf())\n",
    "    time.sleep(0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.271511943652\n",
      "0.458407727986\n"
     ]
    }
   ],
   "source": [
    "err = 0\n",
    "for i in range(1600):\n",
    "    d,td,l = mkData(N,NT)\n",
    "    \n",
    "    data = d[:,0:DIM]\n",
    "    labels = d[:,DIM]\n",
    "    \n",
    "    svc = SVC(C=5,probability = True)\n",
    "    svc.fit(data,labels)\n",
    "    preds = np.clip(svc.predict_proba(td[:,0:DIM]),1e-5,1-1e-5)\n",
    "    \n",
    "    err += np.mean(-l*np.log(preds[:,1])-(1-l)*np.log(preds[:,0]))/1600.0\n",
    "    \n",
    "print(err)\n",
    "\n",
    "lerr = 0\n",
    "for i in range(1600):\n",
    "    d,td,l = mkData(N,NT)\n",
    "    \n",
    "    data = d[:,0:DIM]\n",
    "    labels = d[:,DIM]\n",
    "    \n",
    "    svc = LogisticRegression(penalty='l1',C=50)\n",
    "    svc.fit(data,labels)\n",
    "    preds = np.clip(svc.predict_proba(td[:,0:DIM]),1e-5,1-1e-5)\n",
    "    \n",
    "    lerr += np.mean(-l*np.log(preds[:,1])-(1-l)*np.log(preds[:,0]))/1600.0\n",
    "    \n",
    "print(lerr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZUAAAEKCAYAAADaa8itAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd0ee7dddd8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(errs1,'b-',label=\"64d trained\")\n",
    "plt.plot(np.arange(0,2000), err*np.ones(2000), 'r-', label=\"kSVC\")\n",
    "plt.plot(np.arange(0,2000), lerr*np.ones(2000), 'm-', label=\"Linear\")\n",
    "plt.xlim(0,2000)\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.ylabel(\"Log-loss\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = []\n",
    "tdata = []\n",
    "tlabels = []\n",
    "for i in range(400):\n",
    "    d,td,l = mkData(N,NT)\n",
    "    data.append(d)\n",
    "    tdata.append(td)\n",
    "    tlabels.append(l)\n",
    "data = np.array(data)\n",
    "tdata = np.array(tdata)\n",
    "tlabels = np.array(tlabels)\n",
    "\n",
    "preds, embs = process(data,tdata)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.manifold import TSNE\n",
    "from sklearn.decomposition import PCA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [],
   "source": [
    "pca = PCA()\n",
    "pembs = pca.fit_transform(embs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd105869e80>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.cumsum(pca.explained_variance_ratio_))\n",
    "plt.plot(np.arange(256),np.ones(256)*0.9,'r--')\n",
    "plt.xlabel(\"Principal component\")\n",
    "plt.ylabel(\"Cumulative variance\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAEKCAYAAAASByJ7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nxmV45JDJAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd0ed568320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(pembs[:,0], pembs[:,1],'b.')\n",
    "plt.xlabel(\"PC1\")\n",
    "plt.ylabel(\"PC2\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "tsn = TSNE()\n",
    "tembs = tsn.fit_transform(embs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAD8CAYAAAB0IB+mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJztnX2MHdd53p93d7lboSkgey3IiqStBERxK0VBUi2ELJA/\n1iEjS0ZQ2aZTyEG6siyIICCnIRKg1pZwTUA1addAzKJ246UgKlxAiBDUdaQYMlRR0MIK9jrNynVs\nWTJj2kFoqZKtbOt8IBAlkqd/nDve4XDmzjlzPubMzPMDLu7uvXPnfMyZ57znPe85I0opEEII6T9T\nbWeAEEJIHCj4hBAyECj4hBAyECj4hBAyECj4hBAyECj4hBAyECj4hBAyECj4hBAyECj4hBAyEGba\nzkCed7zjHeq6665rOxuEENIpnn/++b9RSl1Rd1xSgn/ddddha2ur7WwQQkinEJG/NjmOLh1CCBkI\nFHxCCBkIFHxCCBkIFHxCCBkIFHxCCBkIFHxCCBkIFHwCABiNgCNH9DshpJ8kFYdP2mE0AnbvBt58\nE5idBZ55BlhaajtXhBDf0MLvEaZWevG4jQ0t9ufP6/eNjdA5JYS0AS38nmBqpZcdt7ys/84+W16O\nnXtCSAxo4fcEUyu97LilJS38Dz5Idw4hfYYWfk8wtdKrjltaiiP0o5HuZJaX2bEQEhsKfk/IrPQ6\nMTU9LgRdmhxmx0T6CAW/wxRFydRKj2XNF1lfB954A1DqYndSanSpYyLEBgp+R+maKI1GwCOPaLEH\ngOnpdCeHq+Y5COk6nLTtKF0LpdzYAM6d03+LAB/5SLoims1zTE8zaon0C1r4HaVroZTF/K6smP2u\nDV96m/MchIREVDbGToDFxUU15Cde2YpbG2LokmaT8nXJbUVIW4jI80qpxbrjaOE3oEy4XMW3ibjF\nnnx1FWDb/Nr40hlVQ0g9FHxLykQPcLdEuzBRGDuPpm6r2CMB350LOysSC2fBF5F/AuCrAObG5/vv\nSqlPiMj1AB4DMA/geQD/Vin1pmt6bVM1WeoqhF3wycfOo6kvPWZH5LtzoduKxMSHhX8WwK8opf5B\nRHYB+FMR+QqA3wHwWaXUYyLyBQD3Avh9D+m1SpXouQphyhOFeQs0dh5N3EAxOyLfnUsXRnakPzgL\nvtKzvv8w/nfX+KUA/AqA3xh/fgLAIfRA8KuE2YcQtrUgahJlFujqqtv5fHcYMTtL351LF0Z2pD94\n8eGLyDS02+ZnAHwewPcA/FgpNY68xssArvaRVgqUCXMqYu1bUH1aoCHdF/n69zGBXvV7351LaiM7\nzif0Gy+Cr5Q6D+AXRORyAF8C8C9Mfysi+wDsA4CFhQUf2WlMyMYe40YKIag+LdAY7gvXOjD5ve/O\nfdL5Ygow5xP6j9coHaXUj0XkWQBLAC4XkZmxlX8NgFcqfnMMwDFAx+H7zI8NIRt7rBsphKD6tEBj\nuC9c68BXHfoQ6tgCzPmE/uMjSucKAG+Nxf4yAL8K4NMAngXwQehInbsBPO6aVkhCNnbTc7uKRChB\n9WXRVnUePq1Y1zrwUYe+hDrVMFjSXXxY+FcBODH2408B+COl1JdF5EUAj4nIfwLwvwE87CGtYDRp\n7KZCZXJuHyKRmj+4jCxPWTgrUF3uJh2Bax34qENfQu0iwG3UHekASqlkXrfccotqk81NpQ4f1u8m\nx152mVLT0/q97jd15z58WJ8L0O+HD9vnvwsU623//vJy29ZvSvjMu02bDJE+6QYAtpSBxnKlbY7M\ndZE95NvnYp86t0jM4XSbK0WL9QaUl7vL/mSflnITd1qX646EhYJfwNS14lugfYiEifBWla9pJ2Dr\niirbNXNl5dK0u+5PbjNMt+t1R8JBwS9gah2F8He6iISp8JaVD2g+f9BkpFNWb7bx7l2KF6/Ka6gy\n0BdPqqDgF7CxjlJZbAWYC29Z+TY2gLNngQsX9LuNC6CJNWlab1XHtRUv3kSgJ42oyjbh87mgy+Qc\nXeo4iTsU/AJtW0dNb0BT4S0r37e+pcUe0O/z8+bptlFfbfiom3YyVXktfr6+Dpw4EbcT89lxsuPo\nBhT8EkJZ7nU3hcsNaCO8xfJtbwNTU1rsp6b0/zaY1FdKsfa2jEbAoUM7oyCbTqYqr8XPgfidmK+O\nkyt0u0OvBL9NK8OHmLvegJNcIJPytrwMzM2FE1DfgpDv3Obnd+YhQnXSu3fviP3UlF0dTZqzyH8O\nXGzhx5ho9dVxMiqoQ5jEbsZ6ucThh4w9rouFLku7+BuTOPsQZTA9p2m8d5O4cJs1BiHXQjRJJ5/3\nqSmlbrstXFx7k7pNIU3G/bcPDOPwWxf5/MtF8EMtXDJpzMW09+8v7wB8Cq8pPutlUhkm5dum7DbC\n0bRsNumYlHltLb5Q+8a13bXRWZEdTAW/Ny6dUH5dk+GqiT92ddXMx+57/iDGbpd1LhvT+QWTus67\np+rKVuXKsnFBVOW9zNUzN9dN/7WvbT1s129wkrcFTHqFWC/XrRVCWBlNLPO2l9YXWVvTroi1tebn\nyPJSVi5fo4i6ejNxnZmcy8f1yZc5e5mWPTVrOPa2HnQB+QdDc+mEpOl+Jin4Rn3fXJub2mW1f//O\nuWJ1cDbCVHesDxfGZZdpv37m3zfdU6msrtrsBGIL8FD2jYqJqeD3xqUTkiZulqaumfxQt2n0g49z\nTCKLJjlxYmf4H2PvGBv3VN2xrq6zYqTQ9nazvYTqVjrHcH3EXkvBrR/aY/CCn5IvsehLPXq02ZbN\n+XP81m8BIvbhhFVUdSCx1i7YrDUIvTdRkzJXrXRuMjfik5irxtte3DhkBin42Y08Pw8cOJDOgpHi\njb+9bX9j5M9x9izw2c/qScXpad2BdGnRU5Xg2SxGc1k5GkJsq8SurE5NR2cpGS2mNLk2XSxnagxO\n8PM38tSUvplsV0+GokxMbW+M/DlEdsonYr+CtoyY1lmbC3pCpl28plV1avrgnOVl4K23gF272m/D\nGSG24OZqXncGJ/j5G1kpLfoiafgSfYhp0becH8H4Kl+M4f9oBJw5o0cmQPzrE9vPXFande1hNALu\nvXfnuQLZnjxtC2EIceZqXj8MTvCLN/LRo+YTbjGoE1MTyyl/jptvrt/yoWkHE2qInReMmRngvvv0\nnvkxr08qfuZJ22UsL++IfUqEEGdO9PqhF4JvIzyp3MhNaGI5TepAXCyxSb917QjygpH/LCtPE2zy\nlD92ddX+9zHY2NBunDzT07pjbJsQ4tzl+zYpTGI3Y72axOGnvIgj5W0SXM63uakXcmUx6E2eRWu6\nFcPcnFKzs/HWIlQt7kqtjW1u6nrJL/pyXVjnk9QWl/UdDCUOP7RvL9aj/0zyMD/v13JqYonV7R5p\nuj2C6VYMZ84ADz1kd33z1wyw29q4Kk7epo3Fip3f2NA+eyCsy6tJeWKGeRILTHqFSS8A1wJ4FsCL\nAL4N4LfHn78dwNMAvjt+f1vduVKz8F3OHWq7Ad8bddlaYnW7RzbZbK5u90yba1A2OqhaDVtW9kkW\n/tSUUrt2TbakUxgN+LSuXcpDKz8eiLW1AoCrAPyr8d//DMBfArgRwH8G8MD48wcAfLruXE23VgjV\nsFxE29eNn9oydJNy1V2PJiJuen3z9SWiX2Wdk8kumPnP1taUmpmp30Kh6nrFEj/fHU6MHUmJO6aC\n7+zSUUq9CuDV8d9/LyIvAbgawJ0AlseHnQCwAeBjrumVEWL4mIUFzoxryNaF4muSyXUCzLd7waRc\nddfDtm5srm++vmZmtIf7/Hn92aFD+pgjR/S1rXLTlKW3va3PVecWKrteMWPIfbs4TdtfsZ0xjDJN\nvPrwReQ6AL8I4M8AXDnuDADgNQBX+kwrJPkbdHq6eVigqVBNEmWXjqOp0ITYUsDmHC6dVLG+gIv/\nzod72sT4mwpf2fU6cqR66wTfvv5iPufndfo2+/3UladIWTtjGGWimAwDTF4AfgrA8wA+MP7/x4Xv\n/1/F7/YB2AKwtbCw4DSs8TVsjulGCTX0nRRJ00Z+TAmZftmDamzai0n7sp0XCPmEtrW1yTt6hr5f\n6MOPB2JujwxgF4CnAPxO7rNTAK5SO37+U3XnSeURhzFFz/TRh7bC1GTrXtP8hCRk+qGvq828QIx6\nnrRnv89ggLaNBBLRhy8iAuBhAC8ppX4v99UTAO4G8Knx++OuaU3Cp88w5iIPk6c22bplsrrIwib3\n7NH+a5Ny5PMzPa193aOR3YIsl3rz6Qoo5iX0dc23wTfeuHibg6ILq2k5beo3S6MshDaf17Nngfvv\n111CkzmG4nYerovkJpHaArjOYdIrTHoB+GUACsA3AXxj/HovgHkAz0CHZZ4E8Pa6c6Vi4cfG10M/\n8udzqYvNTe3umJuzO4eva+DDFdBGe9jc1HWWWdOzs/bun7rz25Yp797Jp5U/165d9q4/X/lL6fxd\nBhGjdP4UgFR8vdv1/KZ0een1pAnMJpaga11kURbnztmNmFxGWWXWuAtFC/bQIfNRTlOWloB77gHW\n1naigybVgW05TevXpC6LVrmPTfaaXH8bi52RPx4w6RVivXw/4rAvk0ZtlKOpNdnEAgthubnMY/hI\n1+cjJW2elexyDWKPqkIfPyQwlK0VqujT/tk+wjubpGk7Smg6sqiy3MrKY1rGLC+HDgEnT8Z55kGW\nN187sJa14br6bWoF+wq1tbn+tnnt8ig+GUx6hVgvnxa+7yiI1EcLXbZ+TMMWY446fJTBFR/zN763\n4vBJ1TVONb8pg6Fb+L6jPWxGC21EEsTyb4Yom+liJcC+jLGsQtv6N6lH1/mb1B7hWaRskVxfRuXJ\nYtIrxHql6sMPudmXL/Lpzs7qKJsuR0nEXqzkik3ebI9t2obbXlNhS9fymxIYuoUP+Ntjx8bSaiuS\nILOW1teBRx7RWwqfOOHXSrIpm+tIoMoyT9WHazOSsKlHlzbsY5TrMo9iC7djCE+vBd8XpvuJZMPo\nthpt03BKU2w20vIxNC8TO1+deBEfImaat1jC5urOKruOQDi3Cydlw0PBN6Rus6/8TdDmc3JDionp\nDZlavHSdmMeO6AopbGUx+E3PX3YdgbDXNlSHTjQUfA8Ub4zt7Z1nocYmtJVkckMWtyi23Z7BJyZi\n3kYHFULYfHdcVcaDD4OCWyS0AwXfA/k9S6amtFvHhhB71rd5E+XnE44fDzOfYIqJmIfydTc5xgXf\nHVeoeZQ+rZHpHCYzu7FevqN0YmL6RKQiKUeeuJJC1IVp/bpEw4RcAWuSts1K3BRIoV30DTBKJy6m\nT0Qq0uX4+Tp8zSf4fCBKiGgYk2sY4jo3WYlrU5ejUZiHpDMap0VMeoVYry5b+CntI+MjDV9rGFzP\n0wWr1WTFaOiVuGUPlDfJ56RjZ2fVT3b+nJtrtutq1bXnilq/gBZ+XJpOlsYIRWuyCtSXjzWznEcj\nvXrWtoypRfyUYbpi1Pd1Lu53f/Ik8Nxz1dfLpi43NoC33tr5v8nup5PaUNvzTEOFgu+Rpo04dOO3\nHUIXhWF9vb1Juqq859c9tBUCmyd/DaueYev7OmediOkGcTbtYHlZR1hloq+UXTBCFzrqIULBHwC2\n1mUxrPL4cX3jNrX2XW7+srxnHUj+SU5zc+lEe8T0US8tacF/7rlmD1ifdN577wW+8AX9/9SU7lhN\noZ8+TSj4A8HGuswLw5kzOqzSxVJzvfmLec86kAsX9P8xtj62IfaKUVshN83PyooOp21y3bhqNk1E\n+/vTYHFxUW1tbbWdDZLDlz/fZ5RQbAt/yIuEhlz2LiEizyulFmuPo+CTOlK86WP58LlIiHQBU8Gn\nS4fUEnJSuWln4jtPVfkY0uRjih078QsFn7RGKtbzpHwMZfIxlWtBwjLl4yQiclxEfiQiL+Q+e7uI\nPC0i3x2/v81HWqQZWRz8aNR2Tnao2o0xpXxkk48PPthvEUzlWpCw+LLw/wDA5wCs5z57AMAzSqlP\nicgD4/8/5ik9YkGq1lsq1nNdPny5j1J2maRyLUhYvAi+UuqrInJd4eM7ASyP/z4BYAMU/FZI1Q+d\nSuhejHyk2ulmpHItSFhC+vCvVEq9Ov77NQBXBkyLTCBl6y2VJfah85Fqp5snlWtBwhFl0lYppUSk\nNP5TRPYB2AcACwsLMbIzOGi9tU/KnS4ZDt7i8McunS8rpX5u/P8pAMtKqVdF5CoAG0qpd006B+Pw\nSZ9J2YdPuk0KcfhPALgbwKfG748HTIuQ5KHLhLSNr7DMPwQwAvAuEXlZRO6FFvpfFZHvAtgz/p8Q\nQkhL+IrS+VDFV7t9nJ8QQog7Xix8Qggh6UPBJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQgUDB\nJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQ\ngUDBJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQgUDBJ4SQgRBc8EXkdhE5JSKnReSB0OkRQggpJ6jgi8g0\ngM8DuAPAjQA+JCI3hkyTXMpoBBw5ot9J+/B6kLaYCXz+WwGcVkp9HwBE5DEAdwJ4MXC6ZMxoBOze\nDbz5JjA7CzzzDLC01HaudhiNgI0NYHk5rXyFIvXrQfpNaJfO1QB+kPv/5fFnJBIbG1pczp/X7xsb\nbedoh0z8Pv5x/T4Eizfl60H6T+uTtiKyT0S2RGTr9ddfbzs7vWN5WVuS09P6fXm57RztEEr8UnaZ\npHw9SP8J7dJ5BcC1uf+vGX/2E5RSxwAcA4DFxUUVOD+DY2lJuw1SdJtk4pe5N3yIX+ouk5SvB+k/\noQX/zwHcICLXQwv9XQB+I3CapMDSUprCEkL8ykYNqZU91etB+k9QwVdKnRORjwJ4CsA0gONKqW+H\nTDMlhjYh2QTf4hdi1JAKbE/EldAWPpRSTwJ4MnQ6qZG6ayEEKQhSX10msdpTCteQhCO44A+VLrgW\nfJJSB9dHl0mM9uR6DdlZpE/rUTp9pYvRGC7RLQw3vBSf0UJN2pNt+i7XcIghtl2EFn4guuZacLXu\nQvvOY1qPPtLyPeKxbU9N0ne5hkMb0XYVCn5AuuRacL1hm3RwpsLqWzwnpesrrRACaNOemqTvYqT0\nebK8T1DwCQA/N6yNINkIq0/xrEu3aVrFTqQtAczyMT/fLP2mRkrXRrRDhYJPAMS/YW2E1ad41qXb\nJK2qTiS2ABbzcfQosL0dL/0ujWiHCgWf/ISYN+z8PDA1BShVL6w+xbNO0JukVdWJ+KhPm/mEYj62\nt4HVVbf0Sb+g4JPojEbAgQNamKamtCVq4l+uOsZGFE0E3VaoQ7lvbOcT6EcndVDwO0oXY56zPJ85\no0XpwgVARFuiLue0nWT1PZIJ5b6xnU/okh+9i+23D1DwO0hKi5xMyed5ZkbHkwPhffKxCOEOM7XY\nRyNgfV3/vbLix41TJsi+RLqL7bcvUPA7iA+Ri21h5fMMAPfdBywshPfJ21JVL21YpCYW+2gEvPvd\nwNmz+v/jx907vTJBBvyJdOhOmqOHaij4HcRV5GwsLF83TzHPKyt+bsambowqC7asXtq0SOtGDpl4\nZrz1lruAVq24tRHpSe0m5FwDRw+ToeAnTtmN4+qrNbWwfN48If3Ltu6UqnJV1Usoi9RHZ5qJZ2bh\n79plJ6BleagSZFORrms3IdtCKi6+VKHgJ8ykG8fFZ2xqYfm+eVKJ064qV148p6Z06ChQXl+uYu2r\nM11aAp599mIfvuvitypBNhVpk3YTqi0wUmkyFPxINBGIUNaKqYXV15unqlxLSzpE9P77dZ0fOADc\nfPOl9QW4i7XPa9tUPKvyUNVWTdNps910KVKpDSj4EWhqzYW8cUxu3jZvnpATb5PKtb2tF4NduFC9\niOrIEXexTqEzrRq5uHZmbYtuKiPJFKHgR6CpNdfWjVMU29g3T8iJt3zZysIXTYTY175DbVuiZXkw\n7czqOuRURJcROwWUUsm8brnlFtVHNjeVuuwypaan9fvmZts5qqZpXjc3lTp82E/ZDh/W6QP6/fBh\n93MqZV42k7L4LG9KmNSRa3uOVXdduu9cAbClDDSWFn4EQlpzvi0Ym9FItuDntdeAJ5/Uv/FhkYdy\nd5iWzdTdFTqUtQ1M2qrL/IPt6M2lLhmxcykU/EiEGOKGcH3YrO5cXr44BhzwF80TooOM4Tfv2rNn\nq8J+i+fMH+dSj7YGRcoP5ekiFPwW8HWzhrBgTMV2Y0Mv8skjcnE4o2s+Yk7W+iKGVemrUzE9T9lx\nMR6U4lqXKcyTpIaT4IvIrwM4BOBfArhVKbWV+24VwL0AzgP4d0qpp1zS6gs+LcBQFoyJ2C4v60U+\nmYU/M6OjW4rhjBmpuDlCTybGsCp9dSqm5yk7bnW1+bqBTITn53dW8Zady9fkOIV+B1cL/wUAHwCw\nlv9QRG4EcBeAmwD8NICTIvKzSqnzjul1Ht/x1z4tmDpRLn6/sbGz4AcAHnro0nDG7He+3RyxOxDT\n9OquSXGjsyZ599WpmJ6neNz8vI7maVr32W/q2gQt9ACYzOzWvQBsAFjM/b8KYDX3/1MAlurOk1qU\nju9ogs1NpfbvV2p2tr3Igaoy1UU0uHzvM+qmjTr0Fbm0uanzrcdCSs3N6c+atDNfbdP0PNlxa2t+\nIl9CRWJV0deoqgwYRumEEvzPAfjN3P8PA/hg3XlSEnybm9w0jC8739ycFq02xL6pKJvcoE07E9v8\ni+yIZgyxaCJOZWU+fPjivIvoduAzxDG0sNXVhU0HEitk0jatLnYOpoJf69IRkZMA3lny1UGl1OMu\no4vx+fcB2AcACwsLrqfzhqnrxdRdkT8foLcGjj1EnVSmuuG9yfC/yl/qa2ie5V/bEHqSOEb0RRMX\nSlldF+c9Zmf1u68Qx6NH9fxJyAihSXVh47qL6a6JGRmUOrWCr5Ta0+C8rwC4Nvf/NePPys5/DMAx\nAFhcXFQN0gqC7w3GUggRm5SHuhvQ9Qb1MXmWz//MDHDPPf62WZ5Ek7KX1XVx3mNlRb+fONHMP15s\ne1/8YvgIoUl1YTs/FWtCNWZkUPKYDAPqXrjUpXMTgL8AMAfgegDfBzBdd56UXDpK2btqfLh+XBjC\nCtGy/Md2a5gS2j9ebHt1v4/R/ny57nznM0VXk08Qw4cP4P0AXgZwFsAPATyV++4ggO8BOAXgDpPz\npSb4pqQgMKk11LaWz/uaVMzOHfO6Np0rMOnsqtpHiMAEm07OJJ8hr4OJAdEFogi+71dXBT8FfE2m\n+SBm51Ms92232Ylm6IlmG0KmWdY+2jISbIIHXCe1m+aja5gK/lQ03xEJSuannJ6unkz7+Mf1+2jk\nP/3RSPues3j1oh80FMVy791bXQ9lea6qF5cy5OvChsw//uCD/icLy9pHzOuUZ1K6xXwCFx+7vt6s\nbm3z0Ve4tUJP8DmZBtgtbCqLFrGdoG66kKqs3Dff7P5kpqaT7K5RHqH2XNrY0Ndle/viemkjkMAm\neADYmdSemdEPaU99k76UoeB7ILuh5ucvvaFiUiUWtg3bVrSKwrm9bRfV4lskTUXTJWqpitSiPOoe\nk9nGSlaTiLD8Z9mxZ87o1dw+H7k5uJW8Jn6fWK8u+vAzP+DUlPY7Tk019wfGnpyqwnby0NUXGnvV\nZZ4Qk5Yp+YXbrFvfpFa3KQHuhx+HzKK7cEH/X7aXjAmhF3zYuApsRwSullKbQ2vfLpTUrMYU3Ba+\n9j3yUbcht5XuBCa9QqzXkC381Cyx2KFpvtLrYkidC6mvvTCxyl3zFzvGPsWRBmjhx6G43WtTH36s\nh3OYWiWxt5X1kV7fl8UXMS1vjGtZ1bbq5jRcr5nN733Nr6Q2T2MDBd8DPm4ol+GqiZD3SQybikvd\n70Pmz2Tr6UnbJpf9PhXhqWpbo5GeaJ0Zq0yZIeNaBpvfx95WOklMhgGxXl106bjiYzhrMmS+7bYd\nt1MKLiNTylaSVpXXtC5CDserVorWbXlQtm1yXZ5TcS2UuSPX1pSamdFtbna2endY1zLY/r6vrkPQ\npZM+Pqxu0yHz2bN6QnlqqjtWSVn9TCqvySip+Pv1db/WftVinknXaGPj4sdFFo+pKnMqE8RlD0j5\n6EeBc+f09+fOVe8O62NTPpvf+3JvxXZ5+oKC3yI+huR1w8t8FNHUFLBnD3DoULjFPT6Fp6x+6spb\ndyPmfz89DTzyiBYkX26uqvxNynPZtsn5Y+rWC7jm+dgxvdPm3r3Avn32vy+K7sbGzjbggG53kwyM\nYhls21JXxbcVTIYBsV5Dc+nE2F3Q57Df574zptElZeddW9MuqrU1t3Ls3x8mMqrJhlybmzo/k1wf\nIVwIa2s7riRAqfe9z30NST5ybXra7pz5az43p39r+rCg1NwsNrjmHdw8rRvEaKQ+0qjb8Coflupz\noZaND99nmYbCbbddLPiAfV1UzVs0eRRlfj4g/yrOa5jkIUWqjAHXvJsKPjdPC4jJJlpLS8Dqatgh\nqY80qnzTgPbZ5heezc83P1dd3m1+W0fminDZrKzqGjfdQC02e/de+pltvVa5JhcW9Gc212p5eSeq\nxyZPPttFKKo264uZd/rwA9GnMEhgsh95e1v7abN5gu3tyX5Yl7A23yFxLv7fSeGIvq59sR5NQ3BN\nfeCZz/7hh4Gvf13b076uSdNrpZR+fKXKPf+u7vddCJWs6hij5t1kGBDr1SeXTmorZ31g6sM3eQhJ\n0edr43JKxVdbdY19Xfum9drUPeBSr5Pahs0583U3NaXUrbc28+GHbCNNz10XUkwffofpik/RF/kG\nayN4Xa6n0PHxTR7u0nVDw0fdhWxTPtYNhOiITAWfLp1ApBIjHYuia8R0iJrKatFJVLlIqq6xr2tf\nHOrv3Qs899zkeu2Ca2MSPuouZJtyPXfbIaQU/ICEuLgxdulzTcPmps0E6uxZ7f+vm/CNTZ0/vuoa\nh9puo+7hLl01NIptziXfITu9rneorbtx8q8+uXRc8Bnv3iTt2VmlRPR7DBdLfhl+jAdX21D1LFhX\nf3Uq5XMldMhv0zRS8OHHvMagD7+b1MW7h/bP7t+vLop/3r/ffxpFYj642hbTiVPbtQVV5+hSJxBq\nriLfrk0y18lbAAAKaUlEQVTTiFV3NmIfsw2bCr5THL6IfEZEviMi3xSRL4nI5bnvVkXktIicEpH3\nuI5EhsKkmNxsOGnygO4uUSwXYB+XHCruvRirv71dnjebWOqyY2M8aN43vuLHJ7VrkzRi1Z1NOqmu\nC3D14T8NYFUpdU5EPg1gFcDHRORGAHcBuAnATwM4KSI/q5Q6P+FcvcXGJx7iOas2rKzo/WWy9FdW\n/KeRp+wB28DOg6tNOrbYTwtzjTkvOzaVyWtfbdWGSe3aJI1YdWeTTnHPpjNndN22PqdiMgwweQF4\nP4BHx3+vQncE2XdPAViqO0cfXTpNhna+/KK+Y6p94ysuOXYoYggfflMXgM9rFbOt+pzviOU+KUtn\nUt42N7V7cm4ufN4Q24cP4E8A/Ob4789lf4//fxjAB+vO0UfBbyMuuiux7aEWKKVa3jpsxdN3uWO1\n1RDXqw0fvkk5YtWpqeDXunRE5CSAd5Z8dVAp9fj4mIMAzgF41HaEISL7AOwDgIWFBdufJ08bYVyp\nuAfqcKmbouuhi6GIRWzDEX1f51htNUT7jBXfnk/nyJH6cqQWxlkr+EqpPZO+F5EPA/g1ALvHPQ0A\nvALg2txh14w/Kzv/MQDHAGBxcVGVHdNl2hCjkI3M5zqApnVT5bPvqtA3JcS+QjHaqs98x1iXUpWW\nSTmSM0ZMhgFVLwC3A3gRwBWFz28C8BcA5gBcD+D7AKbrztdHl05bhBjihl6ynqrPPhQmZTbxX8cO\n5UxljslkbyFfpB5Kixg+fACnAfwAwDfGry/kvjsI4HsATgG4w+R8FPy0CSW0th1JWWx8CjfdJJpM\n0qY4N9F2nvLp79oV7znNrm0/dMdgKvhOYZlKqZ+Z8N0nAXzS5fwkLUK5imx9uvlh8vw8cOBA2ttQ\n2z6bNyN/zBtv6Ofvtl22tueH8ukrpbfjEPHTHkNt6Z3SVuncS2fANHl26CR/ZFN/apObKfPZm0yc\nlRHT91smkiZlXl7WMdyZuD3yiF4X4fL8V1d8dPoueS6mn1+/4VJ+kz2Tmvri2+4kL8JkGBDrRZdO\nPHwPzV3P5xLPbeveiR3z7uL/3b9f72vksu2Ab1z97z62P/btHgk5LxTjOoHbI5NJ+LY6XM/XNMqm\niXunSV5dhuWTtlGuO8fKSvWq47YsxybXKrPqz5xxz3OIiKyQkW0pRepQ8AdK5i64cEG/uzbwNuON\nbd07TfKaQofWZNuBFMh3ljMzur0B4fLcxGUUWpRTCRum4A8YkYvfXUjBijEVwCZ5TaFDK/s8dp03\nEdN8ZwkA992nH3AeIs+uI7EURDkkFPyBsrEBnDunJwPPnevWasdJ6ZsKoG1eU+jQyohZ503FtNhZ\nFieefZLUBGmCUPAHSlfcAbaEFMCm544dSRMKFzG9+279HlLsgf62a19Q8AdKqharLamLaUox2K40\nEdNi+UNvt92Xdh0KCv6AadsF40oXxLSrLoayjrSJmLZR/q6365BQ8Eln6YKYxnoQhs+RzqSO1FZM\n6WJJC6dHHBLSJl145GNmFd93n46Geugh/4/h8/2Iv7KOtClZ+bNHRFZ1FqEeUUkuhhY+MSY1f3kq\n/tq6ella2omKCjEa8T3SCbHt8qT8dME11xco+MSIVG/Ktv21pvUS0rXR1X3xM7rgmusLFHxiBG/K\nckzrJaSIhjh3zI6Ufv54UPCJEbwpy7GplzIR9eUma3ukk8f3LqzEH6JUOk8VXFxcVFtbW21ng1SQ\nmg8/FZrWS6puMhfKygSw3YRGRJ5XSi3WHUcLnxiTkhVpS8jOqmm99NFNVizT+vrFu332oVPrMhR8\n0ntStaT76CYrlgnoX6fWZSj4pLOYWu2pWtJ99F0XywRU7+dP4kPBJ53ExmpP2ZLuspusimKZ+tap\ndRkKPukkNlZ7Hy3pLtHHTq2rOAm+iDwI4E4AFwD8CMCHlVL/R0QEwH8B8F4A/zj+/OuumSUkw9Zq\np+gQ4r6XzmeUUj+vlPoFAF8G8B/Hn98B4Ibxax+A33dMh5CLMN2jhRCyg5OFr5T6u9y//xRAFtR/\nJ4D18dPUvyYil4vIVUqpV13SIyQPrXZC7HD24YvIJwGsAPhbAO8ef3w1gB/kDnt5/BkFnxBCWqLW\npSMiJ0XkhZLXnQCglDqolLoWwKMAPmqbARHZJyJbIrL1+uuv25eARIHb1xLSfWotfKXUHsNzPQrg\nSQCfAPAKgGtz310z/qzs/McAHAP01gqGaZGIpLpwiRBih9OkrYjckPv3TgDfGf/9BIAV0fwSgL+l\n/767+HwgBiGkPVx9+J8SkXdBh2X+NYD948+fhA7JPA0dlnmPYzqkRVJeuEQIMcc1SmdvxecKwP0u\n5ybpwIVLhPQDrrQlRjAEkpDuw4eYE0LIQKDgE0LIQKDgE0LIQKDgE0LIQKDgE0LIQKDgE0LIQBAd\nMp8GIvI69AKuELwDwN8EOnfXYF3swLrYgXWh6WI9/HOl1BV1ByUl+CERkS2l1GLb+UgB1sUOrIsd\nWBeaPtcDXTqEEDIQKPiEEDIQhiT4x9rOQEKwLnZgXezAutD0th4G48MnhJChMyQLnxBCBk3vBV9E\nPiMi3xGRb4rIl0Tk8tx3qyJyWkROich72sxnDETk10Xk2yJyQUQWC98NrS5uH5f1tIg80HZ+YiIi\nx0XkRyLyQu6zt4vI0yLy3fH729rMYyxE5FoReVZEXhzfG789/ryX9dF7wQfwNICfU0r9PIC/BLAK\nACJyI4C7ANwE4HYA/01EplvLZRxeAPABAF/Nfzi0uhiX7fMA7gBwI4APjetgKPwB9HXO8wCAZ5RS\nNwB4Zvz/EDgH4HeVUjcC+CUA94/bQi/ro/eCr5T6n0qpc+N/vwb9fF1AP5LxMaXUWaXUX0E/nevW\nNvIYC6XUS0qpUyVfDa0ubgVwWin1faXUmwAeg66DQaCU+iqA/1v4+E4AJ8Z/nwDwvqiZagml1KtK\nqa+P//57AC8BuBo9rY/eC36BjwD4yvjvqwH8IPfdy+PPhsjQ6mJo5TXhytxzp18DcGWbmWkDEbkO\nwC8C+DP0tD568cQrETkJ4J0lXx1USj0+PuYg9PDt0Zh5i41JXRAyCaWUEpFBhe+JyE8B+CKAA0qp\nvxORn3zXp/roheArpfZM+l5EPgzg1wDsVjtxqK8AuDZ32DXjzzpNXV1U0Mu6mMDQymvCD0XkKqXU\nqyJyFYAftZ2hWIjILmixf1Qp9T/GH/eyPnrv0hGR2wH8ewD/Win1j7mvngBwl4jMicj1AG4A8L/a\nyGMCDK0u/hzADSJyvYjMQk9YP9FyntrmCQB3j/++G8AgRoOiTfmHAbyklPq93Fe9rI/eL7wSkdMA\n5gBsjz/6mlJq//i7g9B+/XPQQ7mvlJ+lH4jI+wH8VwBXAPgxgG8opd4z/m5odfFeAEcBTAM4rpT6\nZMtZioaI/CGAZehdIX8I4BMA/hjAHwFYgN6x9t8opYoTu71DRH4ZwHMAvgXgwvjj/wDtx+9dffRe\n8AkhhGh679IhhBCioeATQshAoOATQshAoOATQshAoOATQshAoOATQshAoOATQshAoOATQshA+P8i\nenaPpB2JigAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fd0ee1f3908>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(tembs[:,0],tembs[:,1],'b.')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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": "3.5.2+"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}