reco_bias_stan_basic_mvn_prior_full_rank.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1. , 0.5, 0. , 0. , 0. ],\n",
       "       [0.1, 1. , 0. , 0. , 0. ],\n",
       "       [0. , 0. , 1. , 0.1, 0. ],\n",
       "       [0. , 0. , 0.1, 1. , 0. ],\n",
       "       [0. , 0. , 0. , 0. , 1. ]])"
      ]
     },
     "execution_count": 272,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "from scipy.special import expit\n",
    "from pylab import *\n",
    "from itertools import product\n",
    "city =   0\n",
    "drone =  1\n",
    "rice = 2\n",
    "couscous = 3\n",
    "beer = 4\n",
    "name_map = {city: 'city phone', drone: 'drone phone', rice: 'rice', couscous: 'couscous', beer: 'beer'}\n",
    "\n",
    "# An embedding provided that gives similarities between products\n",
    "Psi = array([[1.,0.5,0,0,0],[.1,1,0,0,0],[0,0,1,.1,0],[0,0,0.1,1,0],[0,.0,0,0,1]])\n",
    "P, K = Psi.shape\n",
    "\n",
    "Psi"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1.25, 0.6 , 0.  , 0.  , 0.  ],\n",
       "       [0.6 , 1.01, 0.  , 0.  , 0.  ],\n",
       "       [0.  , 0.  , 1.01, 0.2 , 0.  ],\n",
       "       [0.  , 0.  , 0.2 , 1.01, 0.  ],\n",
       "       [0.  , 0.  , 0.  , 0.  , 1.  ]])"
      ]
     },
     "execution_count": 273,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "matmul(Psi,Psi.T)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 294,
   "metadata": {},
   "outputs": [],
   "source": [
    "P = 5 # number of products\n",
    "\n",
    "\n",
    "data = [{'a': city, 'h': [city], 'I': 1000, 'C': 10},\n",
    "        {'a': drone, 'h': [city], 'I': 1000, 'C': 20},\n",
    "        {'a': rice, 'h': [city], 'I': 0, 'C': 0},        \n",
    "        {'a': couscous, 'h': [city], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [city], 'I': 0, 'C': 0},        \n",
    "        {'a': city, 'h': [drone], 'I': 1000, 'C': 10},\n",
    "        {'a': drone, 'h': [drone], 'I': 1000, 'C': 10},\n",
    "        {'a': rice, 'h': [drone], 'I': 0, 'C': 0},                          \n",
    "        {'a': couscous, 'h': [drone], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [drone], 'I': 0, 'C': 0},\n",
    "        {'a': city, 'h': [rice], 'I': 0, 'C': 0},\n",
    "        {'a': drone, 'h': [rice], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [rice], 'I': 1000, 'C': 10},        \n",
    "        {'a': couscous, 'h': [rice], 'I': 1000, 'C': 20},\n",
    "        {'a': beer, 'h': [rice], 'I': 0, 'C': 0},\n",
    "        {'a': city, 'h': [couscous], 'I': 0, 'C': 0},\n",
    "        {'a': drone, 'h': [couscous], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [couscous], 'I': 1000, 'C': 20},        \n",
    "        {'a': couscous, 'h': [couscous], 'I': 1000, 'C': 10},\n",
    "        {'a': beer, 'h': [couscous], 'I': 0, 'C': 0},\n",
    "        {'a': city, 'h': [beer], 'I': 0, 'C': 0},\n",
    "        {'a': drone, 'h': [beer], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [beer], 'I': 0, 'C': 0},        \n",
    "        {'a': couscous, 'h': [beer], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [beer], 'I': 1000, 'C': 20}]\n",
    "\n",
    "\n",
    "# unlucky in the sense that we get 1/1 click when h=beer and a=city\n",
    "data_unlucky = [{'a': city, 'h': [city], 'I': 1000, 'C': 10},\n",
    "        {'a': drone, 'h': [city], 'I': 1000, 'C': 20},\n",
    "        {'a': rice, 'h': [city], 'I': 0, 'C': 0},        \n",
    "        {'a': couscous, 'h': [city], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [city], 'I': 0, 'C': 0},        \n",
    "        {'a': city, 'h': [drone], 'I': 1000, 'C': 10},\n",
    "        {'a': drone, 'h': [drone], 'I': 1000, 'C': 10},\n",
    "        {'a': rice, 'h': [drone], 'I': 0, 'C': 0},                          \n",
    "        {'a': couscous, 'h': [drone], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [drone], 'I': 0, 'C': 0},\n",
    "        {'a': city, 'h': [rice], 'I': 0, 'C': 0},\n",
    "        {'a': drone, 'h': [rice], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [rice], 'I': 1000, 'C': 10},        \n",
    "        {'a': couscous, 'h': [rice], 'I': 1000, 'C': 20},\n",
    "        {'a': beer, 'h': [rice], 'I': 0, 'C': 0},\n",
    "        {'a': city, 'h': [couscous], 'I': 0, 'C': 0},\n",
    "        {'a': drone, 'h': [couscous], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [couscous], 'I': 1000, 'C': 20},        \n",
    "        {'a': couscous, 'h': [couscous], 'I': 1000, 'C': 10},\n",
    "        {'a': beer, 'h': [couscous], 'I': 0, 'C': 0},\n",
    "        {'a': city, 'h': [beer], 'I': 1, 'C': 1},\n",
    "        {'a': drone, 'h': [beer], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [beer], 'I': 0, 'C': 0},        \n",
    "        {'a': couscous, 'h': [beer], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [beer], 'I': 1000, 'C': 20}]\n",
    "\n",
    "\n",
    "data_empty = [{'a': city, 'h': [city], 'I': 0, 'C': 0},\n",
    "        {'a': drone, 'h': [city], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [city], 'I': 0, 'C': 0},        \n",
    "        {'a': couscous, 'h': [city], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [city], 'I': 0, 'C': 0},        \n",
    "        {'a': city, 'h': [drone], 'I': 0, 'C': 0},\n",
    "        {'a': drone, 'h': [drone], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [drone], 'I': 0, 'C': 0},                          \n",
    "        {'a': couscous, 'h': [drone], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [drone], 'I': 0, 'C': 0},\n",
    "        {'a': city, 'h': [rice], 'I': 0, 'C': 0},\n",
    "        {'a': drone, 'h': [rice], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [rice], 'I': 0, 'C': 0},        \n",
    "        {'a': couscous, 'h': [rice], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [rice], 'I': 0, 'C': 0},\n",
    "        {'a': city, 'h': [couscous], 'I': 0, 'C': 0},\n",
    "        {'a': drone, 'h': [couscous], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [couscous], 'I': 0, 'C': 0},        \n",
    "        {'a': couscous, 'h': [couscous], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [couscous], 'I': 0, 'C': 0},\n",
    "        {'a': city, 'h': [beer], 'I': 0, 'C': 0},\n",
    "        {'a': drone, 'h': [beer], 'I': 0, 'C': 0},\n",
    "        {'a': rice, 'h': [beer], 'I': 0, 'C': 0},        \n",
    "        {'a': couscous, 'h': [beer], 'I': 0, 'C': 0},\n",
    "        {'a': beer, 'h': [beer], 'I': 0, 'C': 0}]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'a': 0, 'h': [0], 'I': 1000, 'C': 10},\n",
       " {'a': 1, 'h': [0], 'I': 1000, 'C': 20},\n",
       " {'a': 2, 'h': [0], 'I': 0, 'C': 0},\n",
       " {'a': 3, 'h': [0], 'I': 0, 'C': 0},\n",
       " {'a': 4, 'h': [0], 'I': 0, 'C': 0},\n",
       " {'a': 0, 'h': [1], 'I': 1000, 'C': 10},\n",
       " {'a': 1, 'h': [1], 'I': 1000, 'C': 10},\n",
       " {'a': 2, 'h': [1], 'I': 0, 'C': 0},\n",
       " {'a': 3, 'h': [1], 'I': 0, 'C': 0},\n",
       " {'a': 4, 'h': [1], 'I': 0, 'C': 0},\n",
       " {'a': 0, 'h': [2], 'I': 0, 'C': 0},\n",
       " {'a': 1, 'h': [2], 'I': 0, 'C': 0},\n",
       " {'a': 2, 'h': [2], 'I': 1000, 'C': 10},\n",
       " {'a': 3, 'h': [2], 'I': 1000, 'C': 20},\n",
       " {'a': 4, 'h': [2], 'I': 0, 'C': 0},\n",
       " {'a': 0, 'h': [3], 'I': 0, 'C': 0},\n",
       " {'a': 1, 'h': [3], 'I': 0, 'C': 0},\n",
       " {'a': 2, 'h': [3], 'I': 1000, 'C': 20},\n",
       " {'a': 3, 'h': [3], 'I': 1000, 'C': 10},\n",
       " {'a': 4, 'h': [3], 'I': 0, 'C': 0},\n",
       " {'a': 0, 'h': [4], 'I': 0, 'C': 0},\n",
       " {'a': 1, 'h': [4], 'I': 0, 'C': 0},\n",
       " {'a': 2, 'h': [4], 'I': 0, 'C': 0},\n",
       " {'a': 3, 'h': [4], 'I': 0, 'C': 0},\n",
       " {'a': 4, 'h': [4], 'I': 1000, 'C': 20}]"
      ]
     },
     "execution_count": 275,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 276,
   "metadata": {},
   "outputs": [],
   "source": [
    "def expand_data_binomial(d):\n",
    "    a = d['a']\n",
    "    omega = vstack([Psi[h,:] for h in d['h']]).mean(0)\n",
    "    c = d['C']\n",
    "    i = d['I']\n",
    "    return omega, a, c, i"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 338,
   "metadata": {},
   "outputs": [],
   "source": [
    "def do_plot(s):\n",
    "    med = s['ctr'].mean(0).flatten()\n",
    "    low = med - percentile(s['ctr'],2.5,axis=0).flatten()\n",
    "    up = percentile(s['ctr'],97.5,axis=0).flatten() - med\n",
    "\n",
    "\n",
    "    xnames = ['h=' + h+' a='+a for h,a in product(name_map.values(),name_map.values())]\n",
    "    figure(figsize=(10,8))\n",
    "\n",
    "    errorbar(xnames,y=med.flatten(),yerr=vstack((low,up)),fmt='.',ecolor='rrrrrgggggmmmmmccccckkkkk')\n",
    "    plot(xnames,med.flatten(),'go')\n",
    "    _ = xticks(rotation=90)\n",
    "\n",
    "    plot(xnames,(click)/(impres),'x')\n",
    "\n",
    "    \n",
    "def print_best_rec(s):\n",
    "    best_rec = s['ctr'].mean(0).argmax(1)\n",
    "\n",
    "    for ii in range(P):\n",
    "        print('when hist is ' + list(name_map.values())[ii] + ' recommend ' + list(name_map.values())[best_rec[ii]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 339,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:pystan:COMPILING THE C++ CODE FOR MODEL anon_model_db65b353cae582ef7639d53ee6044e5e NOW.\n"
     ]
    }
   ],
   "source": [
    "import pystan\n",
    "\n",
    "\n",
    "script = \"\"\"\n",
    "data {\n",
    "  int P;\n",
    "  int N;\n",
    "  int K;\n",
    "  int click[N];\n",
    "  int impres[N];  \n",
    "  int a[N];\n",
    "  vector[K] omega[N];\n",
    "  matrix[P,K] Psi;\n",
    "  real w1;\n",
    "  real w2;\n",
    "}\n",
    "\n",
    "parameters {\n",
    "  matrix[K,K] eps;\n",
    "  real offset;\n",
    "  real rho[P];\n",
    "} transformed parameters{\n",
    "  matrix[P,K] beta;\n",
    "  \n",
    "  beta = w1 * Psi + w2 * Psi * eps;\n",
    "}\n",
    "\n",
    "model {\n",
    "  rho ~ normal(0, 0.1);\n",
    "  offset ~ normal(-8,2);\n",
    "  for (k in 1:K) {\n",
    "    eps[k] ~ normal(0, 1);\n",
    "  }\n",
    "  for (n in 1:N) {\n",
    "    click[n] ~ binomial_logit(impres[n],  dot_product(omega[n], beta[a[n],:]) + offset + rho[a[n]]);\n",
    "  }\n",
    "} generated quantities {\n",
    "  matrix[P,P] ctr;\n",
    "  for (p in 1:P) {\n",
    "     ctr = inv_logit(Psi * beta' + offset + rho[p]);\n",
    "  }\n",
    "}\n",
    "\"\"\"\n",
    "\n",
    "model = pystan.StanModel(model_code=script)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's try the model using only the organic embeddings and no bandit data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 340,
   "metadata": {},
   "outputs": [],
   "source": [
    "# fit to the empty data set\n",
    "flat_data = vstack([hstack(expand_data_binomial(d)) for d in data_empty])\n",
    "omega = flat_data[:,0:K]\n",
    "action = array(flat_data[:,K],dtype=int)\n",
    "click = array(flat_data[:,K+1],dtype=int)\n",
    "impres = array(flat_data[:,K+2],dtype=int)\n",
    "\n",
    "N = omega.shape[0]\n",
    "fit=model.sampling(data={'w1': 1., 'w2': .5, 'click': click, 'impres': impres, 'N': N, 'P': P, 'a': action+1, 'K': K, 'Psi': Psi, 'omega': omega})\n",
    "s = fit.extract()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 341,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "do_plot(s)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 342,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "when hist is city phone recommend city phone\n",
      "when hist is drone phone recommend drone phone\n",
      "when hist is rice recommend rice\n",
      "when hist is couscous recommend couscous\n",
      "when hist is beer recommend beer\n"
     ]
    }
   ],
   "source": [
    "print_best_rec(s)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's see if we get an advantage by using the bandit signal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 343,
   "metadata": {},
   "outputs": [],
   "source": [
    "flat_data = vstack([hstack(expand_data_binomial(d)) for d in data])\n",
    "omega = flat_data[:,0:K]\n",
    "action = array(flat_data[:,K],dtype=int)\n",
    "click = array(flat_data[:,K+1],dtype=int)\n",
    "impres = array(flat_data[:,K+2],dtype=int)\n",
    "\n",
    "N = omega.shape[0]\n",
    "fit=model.sampling(data={'w1': 1., 'w2': .5, 'click': click, 'impres': impres, 'N': N, 'P': P, 'a': action+1, 'K': K, 'Psi': Psi, 'omega': omega})\n",
    "s = fit.extract()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 344,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1. , 0.5, 0. , 0. , 0. ],\n",
       "       [1. , 0.5, 0. , 0. , 0. ],\n",
       "       [1. , 0.5, 0. , 0. , 0. ],\n",
       "       [1. , 0.5, 0. , 0. , 0. ],\n",
       "       [1. , 0.5, 0. , 0. , 0. ],\n",
       "       [0.1, 1. , 0. , 0. , 0. ],\n",
       "       [0.1, 1. , 0. , 0. , 0. ],\n",
       "       [0.1, 1. , 0. , 0. , 0. ],\n",
       "       [0.1, 1. , 0. , 0. , 0. ],\n",
       "       [0.1, 1. , 0. , 0. , 0. ],\n",
       "       [0. , 0. , 1. , 0.1, 0. ],\n",
       "       [0. , 0. , 1. , 0.1, 0. ],\n",
       "       [0. , 0. , 1. , 0.1, 0. ],\n",
       "       [0. , 0. , 1. , 0.1, 0. ],\n",
       "       [0. , 0. , 1. , 0.1, 0. ],\n",
       "       [0. , 0. , 0.1, 1. , 0. ],\n",
       "       [0. , 0. , 0.1, 1. , 0. ],\n",
       "       [0. , 0. , 0.1, 1. , 0. ],\n",
       "       [0. , 0. , 0.1, 1. , 0. ],\n",
       "       [0. , 0. , 0.1, 1. , 0. ],\n",
       "       [0. , 0. , 0. , 0. , 1. ],\n",
       "       [0. , 0. , 0. , 0. , 1. ],\n",
       "       [0. , 0. , 0. , 0. , 1. ],\n",
       "       [0. , 0. , 0. , 0. , 1. ],\n",
       "       [0. , 0. , 0. , 0. , 1. ]])"
      ]
     },
     "execution_count": 344,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "omega"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 345,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1. , 0.5, 0. , 0. , 0. ],\n",
       "       [0.1, 1. , 0. , 0. , 0. ],\n",
       "       [0. , 0. , 1. , 0.1, 0. ],\n",
       "       [0. , 0. , 0.1, 1. , 0. ],\n",
       "       [0. , 0. , 0. , 0. , 1. ]])"
      ]
     },
     "execution_count": 345,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Psi"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 346,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 2, 3, 4, 0, 1, 2, 3, 4, 0, 1, 2, 3, 4, 0, 1, 2, 3, 4, 0, 1,\n",
       "       2, 3, 4])"
      ]
     },
     "execution_count": 346,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "action"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 347,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "do_plot(s)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 348,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "when hist is city phone recommend drone phone\n",
      "when hist is drone phone recommend drone phone\n",
      "when hist is rice recommend couscous\n",
      "when hist is couscous recommend rice\n",
      "when hist is beer recommend beer\n"
     ]
    }
   ],
   "source": [
    "print_best_rec(s)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 349,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 5.89387942e-01,  2.03171013e-01, -1.58208296e-04,\n",
       "        -2.02576606e-03,  4.62290168e-03],\n",
       "       [ 5.56494149e-01,  6.20540073e-01, -2.15638392e-03,\n",
       "        -7.77683884e-03, -1.48289830e-03],\n",
       "       [-1.00325975e-02,  1.02344360e-02,  5.58135096e-01,\n",
       "         8.36018505e-01, -6.37191044e-03],\n",
       "       [-1.05851799e-03, -1.48730767e-03,  8.32166545e-01,\n",
       "         5.56179754e-01, -6.54396408e-03],\n",
       "       [-5.33803573e-03,  3.94383046e-03,  8.06728823e-03,\n",
       "         2.46213989e-03,  1.08446023e+00]])"
      ]
     },
     "execution_count": 349,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s['beta'].mean(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 350,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x126ca0080>"
      ]
     },
     "execution_count": 350,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "imshow(s['eps'].std(0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 351,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-1.34496659e+00, -2.25471630e-01,  1.93680771e-03,\n",
       "         3.92137550e-03,  1.12933702e-02],\n",
       "       [ 1.04748496e+00, -7.36372690e-01, -4.50644861e-03,\n",
       "        -1.59458152e-02, -4.09513362e-03],\n",
       "       [-2.00540318e-02,  2.09760946e-02, -1.04056881e+00,\n",
       "         1.57656673e+00, -1.15505334e-02],\n",
       "       [-1.11632815e-04, -5.07222480e-03,  1.56838997e+00,\n",
       "        -1.04529716e+00, -1.19328748e-02],\n",
       "       [-1.06760715e-02,  7.88766092e-03,  1.61345765e-02,\n",
       "         4.92427978e-03,  1.68920466e-01]])"
      ]
     },
     "execution_count": 351,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s['eps'].mean(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 352,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.01315499, 0.01565424, 0.00659663, 0.00662353, 0.00659238,\n",
       "       0.00856248, 0.01301041, 0.00660026, 0.00658529, 0.006606  ,\n",
       "       0.00663931, 0.00660368, 0.01257981, 0.01596818, 0.006657  ,\n",
       "       0.00657613, 0.00652991, 0.01609873, 0.01247901, 0.00666108,\n",
       "       0.0066999 , 0.00653875, 0.00656142, 0.00655246, 0.01938878])"
      ]
     },
     "execution_count": 352,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "percentile(s['ctr'],50,axis=0).flatten()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 353,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.01359939, 0.0160602 , 0.00796129, 0.00798125, 0.00794511],\n",
       "       [0.00894887, 0.01333716, 0.00783742, 0.00773378, 0.00776053],\n",
       "       [0.00792146, 0.00769193, 0.01292846, 0.0164434 , 0.00775082],\n",
       "       [0.00802039, 0.00764824, 0.01648802, 0.01290713, 0.00774781],\n",
       "       [0.00798996, 0.00773889, 0.00765946, 0.00764434, 0.01969873]])"
      ]
     },
     "execution_count": 353,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s['ctr'].mean(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 354,
   "metadata": {},
   "outputs": [],
   "source": [
    "# fit to the unlucky data set\n",
    "flat_data = vstack([hstack(expand_data_binomial(d)) for d in data_unlucky])\n",
    "omega = flat_data[:,0:K]\n",
    "action = array(flat_data[:,K],dtype=int)\n",
    "click = array(flat_data[:,K+1],dtype=int)\n",
    "impres = array(flat_data[:,K+2],dtype=int)\n",
    "\n",
    "N = omega.shape[0]\n",
    "fit=model.sampling(data={'w1': 1., 'w2': .5, 'click': click, 'impres': impres, 'N': N, 'P': P, 'a': action+1, 'K': K, 'Psi': Psi, 'omega': omega})\n",
    "s = fit.extract()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 355,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0, 0.04)"
      ]
     },
     "execution_count": 355,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "do_plot(s)\n",
    "ylim([0,0.04])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 356,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x1239c97f0>"
      ]
     },
     "execution_count": 356,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "imshow(s['eps'].mean(0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 357,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x122549b70>"
      ]
     },
     "execution_count": 357,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "imshow(s['eps'].std(0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 358,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0, 0.04)"
      ]
     },
     "execution_count": 358,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAJrCAYAAADeVPamAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzdfZxdZX3v/c8vE5MISEQSKRiSYMBSNNpIikjxMJVa8baKrVCjQ0VLTUVp7eHWSk8qrZzm3KL1sQI2iEphOKBYj1FRasWkRRpKIOiIhmN4igFFngwgQpjkd/+x1pDNMDGTWWuv2Tv783695pW916z9y7X3ta89373WtdaKzESSJEnNmDLZDZAkSeolhi9JkqQGGb4kSZIaZPiSJElqkOFLkiSpQYYvSZKkBo0rfEXEcRFxc0RsiIgzxvj99Ii4rPz9tRExf9Tv50bEwxHx7vHWlCRJ2h3tNHxFRB9wDvAq4DDgjRFx2KjVTgEeyMyDgY8CZ4/6/UeAr+9iTUmSpN3OeLZ8HQFsyMxbM3MLcClw/Kh1jgcuLG9fDhwbEQEQEa8DbgNu2sWakiRJu53xhK/nAD9uub+pXDbmOpk5DGwG9o2IvYD3Au+fQE1JkqTdztQ21/874KOZ+XC5IWyXRcRSYCnAnnvuefihhx5aX+skSZLa5Prrr783M2ePXj6e8HUncGDL/TnlsrHW2RQRU4GZwH3AS4ATIuKDwDOBbRHxKHD9OGoCkJkrgBUAixcvzrVr146jyZIkSZMrIu4Ya/l4wtd1wCERcRBFQFoCvGnUOiuBk4H/BE4Arsriit0va2nA3wEPZ+Yny4C2s5qSJEm7nZ2Gr8wcjojTgCuBPuAzmXlTRJwFrM3MlcAFwEURsQG4nyJM7XLNis9FkiSp40Wxgao7uNtRkiR1i4i4PjMXj17uGe4lSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWqQ4UuSJKlBhi9JkqQGGb4kSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWqQ4UuSJKlBhi9JkqQGGb4kSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWqQ4UuSJKlBhi9JkqQGGb4kSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWrQuMJXRBwXETdHxIaIOGOM30+PiMvK318bEfPL5UdExI3lz3cj4g9aHnN7RAyVv1tb1xOSJEnqZFN3tkJE9AHnAK8ANgHXRcTKzPxBy2qnAA9k5sERsQQ4G3gD8H1gcWYOR8T+wHcj4iuZOVw+7ncy8946n5AkSVInG8+WryOADZl5a2ZuAS4Fjh+1zvHAheXty4FjIyIy85GWoDUDyDoaLUmS1K3GE76eA/y45f6mctmY65RhazOwL0BEvCQibgKGgLe3hLEE/jUiro+IpTv6zyNiaUSsjYi199xzz3iekyRJUsdq+4T7zLw2M58P/Bbw1xExo/zV0Zn5YuBVwDsj4r/t4PErMnNxZi6ePXt2u5srSZLUVuMJX3cCB7bcn1MuG3OdiJgKzATua10hM38IPAy8oLx/Z/nvz4AvUezelCRJ2q2NJ3xdBxwSEQdFxDRgCbBy1DorgZPL2ycAV2Vmlo+ZChAR84BDgdsjYs+IeEa5fE/g9ygm50uSJO3Wdnq0Y3mk4mnAlUAf8JnMvCkizgLWZuZK4ALgoojYANxPEdAAjgbOiIjHgW3AOzLz3oh4LvCliBhpwyWZ+Y26n5wkSVKniczuOQBx8eLFuXatpwSTJEmdLyKuz8zFo5d7hntJkqQGGb4kSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWqQ4UuSJKlBhi9JkqQGGb4kSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWqQ4UuSJKlBhi9JkqQGGb4kSZIaZPiSJElqkOFLkqQO1t/fT39//2Q3QzUyfEmSJDXI8CVJktQgw5ckSVKDDF+SJEkNMnxJkiQ1yPAlSZLUIMOXJElSgwxfkiRJDTJ8SZIkNcjwJUmS1CDDlyRJUoMMX5IkSQ0yfEmSJDXI8CVJktQgw5ckSVKDDF+SJEkNMnxJkiQ1aFzhKyKOi4ibI2JDRJwxxu+nR8Rl5e+vjYj55fIjIuLG8ue7EfEH460pSZK0O9pp+IqIPuAc4FXAYcAbI+KwUaudAjyQmQcDHwXOLpd/H1icmb8JHAf8U0RMHWdNSZKk3c54tnwdAWzIzFszcwtwKXD8qHWOBy4sb18OHBsRkZmPZOZwuXwGkLtQU5IkabcznvD1HODHLfc3lcvGXKcMW5uBfQEi4iURcRMwBLy9/P14akqSJO122j7hPjOvzcznA78F/HVEzNiVx0fE0ohYGxFr77nnnvY0UpIkqSHjCV93Age23J9TLhtznYiYCswE7mtdITN/CDwMvGCcNUcetyIzF2fm4tmzZ4+juZIkSZ1rPOHrOuCQiDgoIqYBS4CVo9ZZCZxc3j4BuCozs3zMVICImAccCtw+zpqSJEm7nak7WyEzhyPiNOBKoA/4TGbeFBFnAWszcyVwAXBRRGwA7qcIUwBHA2dExOPANuAdmXkvwFg1a35ukiRJHWen4QsgM68Arhi17MyW248CJ47xuIuAi8ZbU5IkaXfnGe4lSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWqQ4UuSJPWM/v5++vv7J7UNhi9JkqQGGb4kSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWqQ4UuSJKlBhi9JkqQGGb4kSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWqQ4UuSJKlBhi9JkqQGGb4kSZIaZPiSJHW0/nXr6F+3brKbIdXG8CVJktQgw5ckSVKDDF+SJEkNMnxJUs3W9a9jXb9zlCSNzfAlSZLUIMOXJElSgwxfkiRJDRpX+IqI4yLi5ojYEBFnjPH76RFxWfn7ayNifrn8FRFxfUQMlf++vOUxq8qaN5Y/z67rSUmSJHWqqTtbISL6gHOAVwCbgOsiYmVm/qBltVOABzLz4IhYApwNvAG4F3hNZt4VES8ArgSe0/K4gcxcW9NzkSRJ6njj2fJ1BLAhM2/NzC3ApcDxo9Y5HriwvH05cGxERGauy8y7yuU3AU+PiOl1NFySJKkbjSd8PQf4ccv9TTx569WT1snMYWAzsO+odV4P3JCZj7Us+2y5y/F9ERG71HJJkqQu1MiE+4h4PsWuyD9rWTyQmQuBl5U/f7yDxy6NiLURsfaee+5pf2MlSZLaaDzh607gwJb7c8plY64TEVOBmcB95f05wJeAN2fmLSMPyMw7y38fAi6h2L35FJm5IjMXZ+bi2bNnj+c5SZIkdazxhK/rgEMi4qCImAYsAVaOWmclcHJ5+wTgqszMiHgm8DXgjMz8zsjKETE1ImaVt58G/D7w/WpPpUP19xc/kiRJjCN8lXO4TqM4UvGHwOcz86aIOCsiXluudgGwb0RsAE4HRk5HcRpwMHDmqFNKTAeujIjvATdSbDk7v84nJkmS1Il2eqoJgMy8Arhi1LIzW24/Cpw4xuP+Hvj7HZQ9fPzNlCRJ2j14hntJkqQGGb4kSZIaZPiSJElqkOFLkiSpQYYvSZJ6TH9/P/2eBmnSGL4kSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4UsELgEuS1AjDlyRJUoMMX5IkSQ0yfEmSJDXI8KX2ci6ZJElPYviSJElqkOFLUu3W9a9jXf+6yW6GJHUkw5ckSVKDDF/dyHlUkiR1LcOXJElSgwxfkiRJDTJ8SZIkNcjwJUk9qn/dOvrXeVSq1DTDlyRJUoMMX5IkSQ0yfEmSJDXI8CVJktQgw5ckSVKDDF+SJEkNMnxJkiQ1yPAlSZLUIMNXKy9YrQ7X/7l++j/XP9nNkCRVYPiSJElqkOFLkiSpQYYvCXfnSZKaY/iSJElqkOFLkiRV1t/fT78HrY2L4UuSJKlBhi+px/X6fLd1/etY179uspshqYcYviRJPal/3Tr61xm81TzDlyRJUoMMX5Kk2rg1Sdo5w5ckSVKDDF+SuoaT4yXtDgxfkiRJDTJ8SVIXcKuftPsYV/iKiOMi4uaI2BARZ4zx++kRcVn5+2sjYn65/BURcX1EDJX/vrzlMYeXyzdExCciIup6Up3gU6tv4Zpb7n3SsmtuuZdPrb5lklokqd0c9/Xy9VSdZ83vpPfTTsNXRPQB5wCvAg4D3hgRh41a7RTggcw8GPgocHa5/F7gNZm5EDgZuKjlMecBbwMOKX+Oq/A8Os4L58zktEvWcc3eBwJFB592yTpeOGfmJLdMUruMjPuhvR8DHPdVjbyev7y7e15Pj/bsXE+8nzrg7/LUcaxzBLAhM28FiIhLgeOBH7Ssczzwd+Xty4FPRkRkZus78Cbg6RExHXgWsHdmrilr/jPwOuDrFZ5LRzlqwSw++aZFnPbzn3PS3Tdy8SXr+OSbFnHUglmT3TRJbTIy7t/+8+t45d178C3HfSUjr+dJ/3wdz1iwB6fd4eupiTtqwSz+5hcrOWL1NzjgwXu568W/wd+cuoSjFryi8baMZ7fjc4Aft9zfVC4bc53MHAY2A/uOWuf1wA2Z+Vi5/qad1AQgIpZGxNqIWHvPPfeMo7md46gFszjp7hv5xJyjOOklc/3A6DG9ftmeXnXUglm88u49+MKchx33NThqwSyesWAPNv/A11PVXP2Bd/DKj3ySu/eaxXlHnsjde83ilR/5JFd/4B2Nt6WRCfcR8XyKXZF/tquPzcwVmbk4MxfPnj27/sa10TW33MvF+/0mf7HpGi6+duNT9jVL2v1cc8u9XLnfI5y4aS/HfQ2uueVeHrrlEWYe5uupauZ/cAXrZx/KwJLlfPhlJzGwZDnrZx/K/A+uaLwt4wlfdwIHttyfUy4bc52ImArMBO4r788BvgS8OTNvaVl/zk5qdrWRfcmf/NFKTt/0nWIX5CXr/OCQdmMj4/7dP9qHN23a23Ff0cjrOful+7DPQl9PVXPAA1tZM3chW/qmsm1KH49P6WPN3IUc8MDWxtsynvB1HXBIRBwUEdOAJcDKUeuspJhQD3ACcFVmZkQ8E/gacEZmfmdk5cz8CfBgRBxZHuX4ZuDLFZ9LR/neps3F3IQHiz22I3MXvrdp8yS3TFK7jIz7hQ9OBxz3VY28nk/fz9dT1d01c0+O3DjEtK3D9G0d5mnbtnLkxiHumrln423Z6YT7zByOiNOAK4E+4DOZeVNEnAWszcyVwAXARRGxAbifIqABnAYcDJwZEWeWy34vM38GvAP4HPB0ion2u81ke4C3H7PgKcuOWjDL+QrSbmxk3K9rmSbruJ+4Jz5H1/l6qrrbz3gzi848j8FLl7Fm7kKO3DjEofesZ91Zpz5pV1wTxnO0I5l5BXDFqGVnttx+FDhxjMf9PfD3O6i5FnjBrjRWkiRpIo4+41yuBg78/87j1DXruWufPtaddSpHn3Fu420ZV/iSJEnqdkefcS793yjOlLVq1arGt3iN8PJCkiRJDTJ8SZIkNcjwJUmS1CDDlyRJUoMMX5IkSQ0yfEmSJDXI8CVJktQgw5ckSVKDDF+SJEkNMnxJkiQ1yPAlSZLUIMOXJElSgwxfkiRJDTJ8SZIkNcjwJUmS1CDDlyRJUoMMX5IkSQ0yfEmSJDXI8CVJktQgw5ckSVKDDF+SJEkNMnxJkiQ1yPAlSZLUIMOXJElSgwxfkiRJDTJ8SZIkNcjwJUmS1CDDlyRJUoMMX5KkjjU4NMiab76a1SsPZ/7H5jM4NDjZTZIqM3xJUo2u/sA7mLVuMS9a/WI2PWsqV3/gHZPdpK41ODTI0q8sJR+eyd6Pn8BP7n86S7+y1ACmrmf4kqSaXP2Bd7DozPP42V7P47wjT+Tupx/CojPPM4BN0LJvLePlN81l/i+Ws++Wk5j/i+W8/Ka5LPvWsslumlSJ4UuSajL/gytYP/tQBpYs58MvO4mBJctZP/tQ5n9wxWQ3rSv99n/cwbG3L+Txvqlsm9LH8JQ+jr19Ib/9H3dMdtOkSgxfbTQ4NMj8I9cwpX+1cxWkHnDAA1tZM3chW8qw8PiUPtbMXcgBD2yd7KZ1pbO/3cfLbh9i2tZh+rYO87RtW3nZ7UOc/e2+yW6aVMnUyW7A7mpwaJB/+59v5ZzvLuC2Zy3koPuHuPyat8L7YGDhwGQ3T1Ib3LVPH0duLMLC45k8bdtWjtw4xF379DFnshvXhZ7z863MeWA9g5cuY83chRy5cYjD71pPxmS3TKrG8NUm1579Lv702gWccsJytvRNZdrWYS64fBmXnf0uBi42fEm7o9v/aimLzjzvSWHh0HvWs+6sUw1fExBz58Edd3D4Xes5/K71T14udTF3O7bJ6V+9jxsPePLuhxsPWMjpX71vspsmqU2OPuNc1p11Ks9++P9y6povsN8vf8S6s07l6DPOneymdafly2GPPZ68bI89iuVSF3PLV5vM3cyYux/mbp7slklqp6PPOJd133gb9wKLVi1yi1cVA8Vegp++5z08+6c/ZcrcuUXwGnDvgbqb4atNHtl/Xw6/66lzFR7ef1/2muzGSVK3GBhgyWGHAbBq0aJJboxUD8NXm+z1oY8z/Kd/8qS5CsMzprHXhz4+yS2TJEmTyTlf7TIwwNRPfwamTy/uz5tX3HdzuSRJPc0tX+00MADnn1/cXrVqUpsiSZI6g1u+JEmSGmT4kiRJapDhS5IkqUHjCl8RcVxE3BwRGyLijDF+Pz0iLit/f21EzC+X7xsR346IhyPik6Mes6qseWP58+w6npAkSVIn2+mE+4joA84BXgFsAq6LiJWZ+YOW1U4BHsjMgyNiCXA28AbgUeB9wAvKn9EGMnNtxecgSZLUNcaz5esIYENm3pqZW4BLgeNHrXM8cGF5+3Lg2IiIzPxFZl5NEcIkSZJ63njC13OAH7fc31QuG3OdzBwGNgP7jqP2Z8tdju+LCK9TL0mSdnuTOeF+IDMXAi8rf/54rJUiYmlErI2Itffcc0+jDZQkSarbeMLXncCBLffnlMvGXCcipgIzgft+VdHMvLP89yHgEordm2OttyIzF2fm4tmzZ4+judrd9X+un/7P9U92MyRJmpDxhK/rgEMi4qCImAYsAVaOWmclcHJ5+wTgqszMHRWMiKkRMau8/TTg94Hv72rjJUmSus1Oj3bMzOGIOA24EugDPpOZN0XEWcDazFwJXABcFBEbgPspAhoAEXE7sDcwLSJeB/wecAdwZRm8+oB/A86v9ZlJkiR1oHFd2zEzrwCuGLXszJbbjwIn7uCx83dQ9vDxNVGSJGn34RnuJUmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWqQ4UuSJKlBhi9JkqQGGb5Kg0ODzD9yDVP6VzP/Y/MZHBqc7CZJkqTd0LjOcL+7Gxwa5N/+51s557sLuO1ZCzno/iEuv+at8D4YWDgw2c1Tmw0ODbJm0xoe2/oY8z82n+XHLrffJUltY/gCrj37XfzptQs45YTlbOmbyrStw1xw+TIuO/tdDFzsH+Hd2Ujw/qLBW5LUEMMXcPpX7+Nrv/FytvRNZduUPh7P5MYDFnL6V78w2U1Tmxm8JUlNc84XMHczHLlxiGlbh+nbOszTtm3lyI1DzN082S1Tu53+1fu48YCF24P3lL4yeN832U2TJO2m3PIFPLL/vhx+13oGL13GmrkLOXLjEIfftZ6H99+XvSa7cWqr1uD9eKbBW5LUdm75Avb60McZnjGNw+9azzvXfIHD71rP8Ixp7PWhj09209RmrcH79KsHGbx0GYfftZ5H9t93spumBgwODfLqI1/N4f2He5RzjxkcGmTNN1/N6pX2vZrnli+AgYHihTjlFHjsMZg3j6nLl8OAc352d3t96OMM/+mfcPhd6zn8rvUABu8eMXKwxfnfXcBtzzragy16yJMPtLHv1TzD14iBATj//OL2qlWT2hQ1qAzeP33XKTz7vseYYvDuGR5s0bvse002w5c0MMCSx4vgveotqya3LWqMRzn3Lvtek83wJaknebBF77LvNdkMX5J6kkc59y77vret6oCpRYYvST3Jgy16l32vyeapJiT1poEBpn76M2yZvj9JFEc5f/ozHmzRC8q+/+n++7Mt7Hs1zy1fknrXwAA3nX8YAItWLZrkxqhRAwMsOazo+1WL7Hs1yy1fkiRJDTJ8SZIkNcjwJUmS1CDDlyRJUoMMX5IkSQ0yfEmSJDXI8CVJktQgw5ckSR1qcGiQNUeuYXX/auZ/bD6DQ4OT3STVwJOsSpLUgQaHBln6laXktIPYe3ghP7l/iKVfWQrAwELPxt/N3PIlSVIHWvatZQw/Opf9tiznmcMnsd+W5Qw/Opdl31o22U1TRYYvSZI60MbNG5mxbSHBVII+gj5mbFvIxs0bJ7tpqsjwJUlSB5o7cy6PThkiGS5/tvLolCHmzpw72U1TRYYvSZI60PJjlzN1xkbunraMn08d5O5py5g6YyPLj10+2U1TRU64lySpA41Mqj9l8BQenLGeec+cx/JjVzjZfjdg+JIkqUMNLBzg/DXnA7Bq1arJbYxq425HSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWrQuMJXRBwXETdHxIaIOGOM30+PiMvK318bEfPL5ftGxLcj4uGI+OSoxxweEUPlYz4REVHHE5IkSepkOw1fEdEHnAO8CjgMeGNEHDZqtVOABzLzYOCjwNnl8keB9wHvHqP0ecDbgEPKn+Mm8gTq8KnVt3DNLfc+adk1t9zLp1bfMkkt0lja0U/2fe+y73uXfa/JNp4tX0cAGzLz1szcAlwKHD9qneOBC8vblwPHRkRk5i8y82qKEPaEiNgf2Dsz12RmAv8MvK7KE6nihXNmctol67hm7wOBYhCedsk6Xjhn5mQ1SWMY6adf/qK+fmpHTXWHkb4f2vsxwL7vJU+M+7vte02O8YSv5wA/brm/qVw25jqZOQxsBvbdSc1NO6nZmKMWzOKTb1rEaYe8lo/M+W1Ou2Qdn3zTIo5aMGuymqQxjPTTPXe9lgfuqaef2lFT3WGk7//hkAe4ZM6D9n0PeWLc/+cDPDBk36t5HT/hPiKWRsTaiFh7zz33tO3/OWrBLE66+0Y+MecoTnrJXAdhhzpqwSye8cwb2Xxfff3UjprqDkctmMUr796DL8x52L7vMUctmMUzFuzB5h90eN8PDnLpmjVctXo1zJ8Pg4OT3SLVYDzh607gwJb7c8plY64TEVOBmcB9O6k5Zyc1AcjMFZm5ODMXz549exzNnZhrbrmXi/f7Tf5i0zVcfO3Gp8wHUGe45pZ7eejnv8nMfevrp3bUVHe45pZ7uXK/Rzhx0172fY+55pZ7eeiWR5h5WAf3/eAgLF3Krz32WPHH+o47YOlSA9huYDzh6zrgkIg4KCKmAUuAlaPWWQmcXN4+AbiqnMs1psz8CfBgRBxZHuX4ZuDLu9z6mozs7//kj1Zy+qbvFLsgL1nXmYOxh4300+wDVrLP7Hr6qR011R1G+v7dP9qHN23a277vIU+M+5fuwz4LO7jvly2DRx558rJHHimWq6vtNHyVc7hOA64Efgh8PjNvioizIuK15WoXAPtGxAbgdOCJ01FExO3AR4C3RMSmliMl3wF8GtgA3AJ8vZ6ntOu+t2lzsb//wWJq28h8gO9t2jxZTdIYRvrp6XvW10/tqKnuMNL3Cx+cDtj3veSJcb9fh/f9xo27tnycBocGWXPkGlb3r2b+x+YzOOSWtKZNHc9KmXkFcMWoZWe23H4UOHEHj52/g+VrgReMt6Ht9PZjFjxl2VELZnXuHIAe9UQ//cf2ZVX7qR011R1G+n5dy/FEdfX9olWLKtdQ+zwx7tfV3/e1mju32NU41vIJGhwaZOlXlpLTDmLv4YX85P4hln5lKQADCwcmXFe7ZlzhSxWsWjXZLZCkrrZqUY+G2eXLizlerbse99ijWD5By761jOFH57LfluUEU0mGuZtlLPvWMsNXgzr+aEdJknrSwACsWMFPp09nG8C8ebBiRbF8gjZu3siMbQsJphL0EfQxY9tCNm6utitTu8YtX5IkdaqBAZacfz4Aq2rYkzJ35lx+cv8QyTCQJFt5dMoQc2dOfFemdp1bviRJ6hHLj13O1BkbuXvaMn4+dZC7py1j6oyNLD924rsytesMX+oqV3/gHVz4l6u56q2r2fSsqVz9gXdMdpMkqWsMLBxgxWtWEFtu48GpX2D/Z/2SFa9Z4XyvhrnbUV3j6g+8g0Vnnsf62Yfy1d9YyJEbh1h05nlcDRx9xrmT3TxJ6goDCwc4f019uzK16wxf6hrzP7iC9bMPZWDJcrb0TWXa1mEGL13G/A+uAMOXJKlLuNtRXeOAB7ayZu5CtvRNZduUPh6f0seauQs54IGtk900SZLGzfAlBocGmX/kGqZ0+NmO79qnjyM3DjFt6zB9W4d52ratHLlxiLv26ZvspkmSNG6Grx43crbjn0w7iGcMn8hP7n86S7+ytJYAVneou/2vlnLoPesZvHQZp189yOClyzj0nvXc/ldLK7dV9RkcGuTVR76aw/sP7+gw3+sGhwZZ881Xs3ql/SQ1zfDVZeoONK1nO37m8Enst2U5w4/OZdm3ql24tR2h7ugzzmXdWacy6xfrOXXNF9jvlz9i3VmnOtm+g4z0+33TZvKM4RNqDfOqzxOXmHl4Jns/bj/1nNnoSJsAACAASURBVMFBLl2zhqtWr4b582HQfm+a4auLtCPQtOtsx+0KdUefcS4nf+wYXv7ZY5hz/3BPBa/BoUHWbFrD6js6d/dwu/od3KJWp3b2kzrc4CAsXcqvPfZYEQDuuKO4hFHFAObFuneN4auLtOMDc+7MuTw6pTjbcfFTz9mOvYRFvZ7YUrHlIPZ+vL7dw3UHunb1e69vUas7eLazn3p5V2ZXPP9ly558rUgo7i+b+N+RJ1+su97pK7srw1cXaccHZrvOdtyuUNct6g417Qje7Qh07er3Xt5S047g2Y5+6vVdmV3z/Dfu4O/FjpaPQy+Pz4kyfHWRdnxgjpzteP8tt/FQjWc77uVLWLQj1LQjeLfjA7Nd/d7LW1K7pZ96/Q9w1zz/uTv4e7Gj5ePQ1i2pu+muTMNXF2nXH7aBhQPcvuZItq06htv/8vZaLjPRrlDXDbpl93A7PjBH+n3fLZt5aOrltfV7L29JbWc/xV6befBp9fRTLwdk6KLnv3w57LHHk5ftsUexfILauiV1N92V6Rnuu8jIB+OywVPYOGM9c585j+XHdm6gGVg4wEB5CQt66BIWGzdv5BnbTnjiQxiy/BC+fMI1lx+7nKVfWcrdLGPGtoU8OmWoDN4rJlxz7sy5/OT+4gMTsrZAM7BwgMPWHAbAolWLKtUa0Y7n3y3a2U/nDxf9tGpR9X5qVzu7Rdc8/4Hi78VPTzmFZz/2GFPmzSuC18DE/460Y3y2fokNppIMczfLWPatZR37N29XGL66TK8Gmm7Sjg/hkQ+bU758Cg9uXc+8mdWDdzcFmpHn+Z7B9/DTGZcz95lzO/qLR526pZ+6pZ3t0lXPf2CAJefXd23HJz6fBk/hwRnrmVfDhoF2fIntJIYvqWbt+hAeWDjA+deXH5hvWVW5ne0IdO3Uji1q3aBbgucT76evv4cHf3k582Z2Zjvbxedf78W6u2ZL4gQZvqSadVOoqTvQqT26JXjWvSuz2/T6869TV21JnADDl0T9wcNQI0kT145dmZ3E8CVJkjpO3bsyO4nhq9Vu1rmSJKkDZWbX/Bx++OHZ8y6+OHP69EzInDevuN+JNdtY95jPHpPHfPaYWmq1zcUX50/2nZ5ba+6nbqn52PT9cxtR+/upHXVvOOaGvOGYG2qp1TbteO4XX5w/2X//3BodXrONjrnhhjzmhhr7vk3P/5hjjsljjjmmllptq3nxxfmT6W34fKq7ZsOAtTlGnpn0QLUrPz0fvi6+OHOPPYpuG/nZY49qb8h21Gxn3eyC8NUt/dQtNdtZN7sgfHVLP7Wxj7pCOz/zOj18+X7aoR2Fryh+1x0WL16ca9eunexmTJ7584sr0I82bx7cfnvn1GxnXaD/c/1AB09k75Z+6paa7awLrOtfB3TwUYTd0k9t7KOu0M7PvP5+oN55T7XW9P20QxFxfWYuHr3cywt1kzZcELUtNdtZtxt0Sz91S8121u0G3dJPvdxH0NvP3/fTLjN8dZM2XBC1LTXbWbcbdEs/dUvNdtbtBt3ST73cR9Dbz9/30y4zfHWTNlwQtS0121m3G3RLP3VLzXbWpdjd2LG7HKF7+qmXxzz09vP3/bTrxpoI1qk/PT/hPtOjHbMLJtxndtWRib1+tGNX8GjH7uDRjh7tOApOuN+NlBMlaz0vWTtqtqlux0+4L7Wjnd1Ss12T2Dt+cnwbteO5968ratZ5KZx21OwmbXlNO33CfZfVbJIT7iVJkjqA4UuSJKlBhi9JkqQGGb4kSZIaZPiSJElqkOFLkiSpQYYvSZKkBhm+JEmSGmT4kiRJapDhS5IkqUGGL0mSpAYZviRJkhpk+JIkSWqQ4UuSJKlBhi9JkqQGjSt8RcRxEXFzRGyIiDPG+P30iLis/P21ETG/5Xd/XS6/OSJe2bL89ogYiogbI2JtHU9GkiSp003d2QoR0QecA7wC2ARcFxErM/MHLaudAjyQmQdHxBLgbOANEXEYsAR4PnAA8G8R8bzM3Fo+7ncy894an48kSVJHG8+WryOADZl5a2ZuAS4Fjh+1zvHAheXty4FjIyLK5Zdm5mOZeRuwoawnSZLUk8YTvp4D/Ljl/qZy2ZjrZOYwsBnYdyePTeBfI+L6iFi6602XJEnqPjvd7dhGR2fmnRHxbOCbEbE+M/999EplMFsKMHfu3KbbKEmSVKvxbPm6Eziw5f6cctmY60TEVGAmcN+vemxmjvz7M+BL7GB3ZGauyMzFmbl49uzZ42iuJElS5xpP+LoOOCQiDoqIaRQT6FeOWmclcHJ5+wTgqszMcvmS8mjIg4BDgP+KiD0j4hkAEbEn8HvA96s/HUmSpM62092OmTkcEacBVwJ9wGcy86aIOAtYm5krgQuAiyJiA3A/RUCjXO/zwA+AYeCdmbk1IvYDvlTMyWcqcElmfqMNz0+7oVVvWTXZTZAkacLGNecrM68Arhi17MyW248CJ+7gscuB5aOW3Qq8aFcbK0mS1O08w70kSVKDDF+SJEkNMnxJkiQ1yPAlSZLUIMOXJElSgwxfkiRJDTJ8SZIkNcjwJUmS1CDDlyRJUoMMX5IkSQ0yfEmSJDXI8CVJktQgw5ckSVKDDF+SJEkNMnxJkiQ1yPAlSZLUIMOXJElSgwxfkiRJDZo62Q2QJEk7tmrVqslugmrmli9JkqQGueVLkqQe49a0yeWWL0mSpAYZviRJkhrkbkdJ6gKLVi2a7CZoHFYt6t1+clfm+Bm+pDZZ9ZZVPVuzXQwgna+Xw4fqt7sGOsOXpNoZkrqDQUmaHIYvFdr17WI3/dYiSdJEGb66kYFGkqSu5dGOkiRJDTJ8SZIkNcjwJUmS1CDDlyRJUoMMX5IkSQ0yfEmSJDXI8CVJktQgw5ckSVKDDF+SJEkNMnxJkiQ1yPAlSZLUIMOXJElSgwxfkiRJDTJ8SZIkNcjwJUmS1CDDlyRJUoMMX5IkSQ0aV/iKiOMi4uaI2BARZ4zx++kRcVn5+2sjYn7L7/66XH5zRLxyvDUlSZJ2RzsNXxHRB5wDvAo4DHhjRBw2arVTgAcy82Dgo8DZ5WMPA5YAzweOA86NiL5x1pQkSdrtjGfL1xHAhsy8NTO3AJcCx49a53jgwvL25cCxERHl8ksz87HMvA3YUNYbT01JkqTdznjC13OAH7fc31QuG3OdzBwGNgP7/orHjqemJEnSbmfqZDdgZyJiKbC0vPtwRNzc5v9yFnCvNTu+rjV7s2a76lqzN2u2q641e7PmWOaNtXA84etO4MCW+3PKZWOtsykipgIzgft28tid1QQgM1cAK8bRzlpExNrMXGzNzq5rzd6s2a661uzNmu2qa83erLkrxrPb8TrgkIg4KCKmUUygXzlqnZXAyeXtE4CrMjPL5UvKoyEPAg4B/mucNSVJknY7O93ylZnDEXEacCXQB3wmM2+KiLOAtZm5ErgAuCgiNgD3U4QpyvU+D/wAGAbemZlbAcaqWf/TkyRJ6izjmvOVmVcAV4xadmbL7UeBE3fw2OXA8vHU7BDt2MXZyzXbVdeavVmzXXWt2Zs121XXmr1Zc9yi2DsoSZKkJnh5IUmSpAYZvlSriJgSEX802e3QjpV9dNRkt0PaHTieNBE9H76icFJEnFnenxsRR9RQ9+iIeGt5e3Z5tGeVevtFxAUR8fXy/mERcUrFmgsiYnp5uz8i/iIinlmlZmZuA/6qSo0dqfs1LevMi4jfLW8/PSKeUbHeHhHxvog4v7x/SET8fsWae0bElPL28yLitRHxtInWK/vonCptGks73qNlndpf09baddQpa5048v6JiL+JiH+JiBdXqNcXEf9QV/tG1e7JsVTW6dnx5FjqHD0fvoBzgZcCbyzvP0TFgRQRfwu8F/jrctHTgIur1AQ+R3F06AHl/f8L/GXFml8EtkbEwRSTDw8ELqlYE+DfIuLdEXFgRDxr5KdKwXa8phHxNorLYf1TuWgO8H+q1AQ+CzxG8Z6C4vx1f1+x5r8DMyLiOcC/An9M8X6o4lsR8fqIiIp1Wn2O+t+j0IbXNCKOiogfAOvL+y+KiHMrtRLel5kPRcTRwO9SHAV+3kSLlUeGH12xTU/R42MJens8OZY6hOELXpKZ7wQeBcjMB4BpFWv+AfBa4BdlzbuASt8CgVmZ+XlgW1lzGNhasea2ss4fAP+Yme8B9q9YE+ANwDspPuSuL3/WVqzZjtf0ncBvAw+WNX8EPLtizQWZ+UHg8bLmI0DVD+Qo6/whcG5mnkhxsfoq/gz4ArAlIh6MiIci4sGKNdvxHoX2vKYfBV5JcTJoMvO7wH+rWHPkub4aWJGZX6P6Z8m6iFgZEX8cEX848lOxZi+PJejt8eRYqncsTVjHX16oAY9HRB+QUGyCp3yzV7AlMzMiRmruWbEewC8iYl+2t/NIimtoVvF4RLyR4gS5rymXTXjz+4jMrLwLYwzteE0fy8wtI19Wo7g6Q9XDf7dExNPZ3k8LKL5pVhER8VJgABjZ7dBXpWBmVv1jO5Z2vEehPa8pmfnjURsqqv5huzMi/gl4BXB2FLv0q37BnUHxR+3lLcsS+JcKNXt5LJWlenY8OZbqHUsTl5k9/UMxAFdSXNx7OXAzcGLFmu+m2Px+K/A24D+BP69Y88XAdygG33coNkG/sGLNw4BPAG8s7x8EvLeG13QP4G8ovrFAcWWD3+/A1/SDwP+g2Fz+CuBLwPKKNV8BrAbuAQaB24H+ijWPKd+j7y3vPxf4RMWaAZxEsXkfil3OR3Tae7SNr+nlwFHADRRfON4NXFqx5h4UW1MOKe/vD/xe1edf908vj6Wybs+OJ8dS5/x4ni8gIg4FjqUYQN/KzB/WUPMVwO+VNa/MzG/WUHMq8OtlzZsz8/GqNdshIi6j2NX45sx8QTkR85rM/M2KdWt9TctJt6e01gQ+nRUHRflt9ciy5prMbOLirbskIs6j2ML78sz8jYjYB/jXzPytinXb8h6t+zWNiFnAxynmkwTF3J93ZeZ9FWrOHWt5Zm6sUPN5FHNd9ivH0guB12Zm1Xk6jqUaddN4cizVO5Ym3B7DV3EkBLAfLbthq3RyO0TEO4HBzPx5eX8fii1WE57YGBG3Mcaugcx87oQbyvYLlkbEusxcVC77bma+qErdbhARf0BxbdPN5f1nUnyznPDk44j4NmP308vHWH28NW/IzBfX3UdRHHI/nyePpX+uUrOs+xxg3qi6/161bp0iYoiin4JiF8dBFH8wJzyfKCJWA+8B/qmln76fmS+oockdrR1jqazT0+PJsdQZY6nn53xFxJ8DfwvcTbGfOig6/YUVav4hcDbFhNMYqZmZe1do6tsy84mjMDPzgfIIoypHlbRe0X0GxSWiKh2VWKp9XkE7XtOI+G3g79j+QTRSs0r4/NvM/NLIncz8eXl0WZU/GO9uuT0DeD3FtVKrqH2uY0RcBCwAbmT7nI8Eqv6xOJviII6bWtqYFAd0TLTmhRTfzlu/zHw4M/9kojUzc+Go/+PFwDsmWq+0R2b+16j5NJX6vsfHEvTweHIs1TuWquj58AW8C/j1KptIx/BB4DV17L5s0RcRMbIZvxzolY7+GOM5fywirgfOHGv9XfC3wDeAAyNikOIoqLdUrNmO1/QC4L9T7CKt46g8GHtSaKVxlpnXj1r0nYj4ryo1Keb6fQnYLyKWAydQzNOrYjFwWNVdTWN4HcUYrWOy9YgXjvyxgCe+zCyqsT6ZeUNEvKRimXvLLy8j4/4E4CcVa/bsWIKeH0+OpXrH0oQZvuDH1HNEVqu7a/5ggyLMXFYeAQLFoc3fqFIwnnzSuikUg72OD7dvRsQNbJ9X8K4a5mq04zXdnJlfr7nm2oj4CNvPFfdOij9IExZPPkfaFOBwYGaVmpk5WAbtkbmOr6vh9f0+8GvU/4F2K8VE3jr/YEyJiH2yOLXMyGtc6b0fEae31qeYMH1XlZoU758VwKERcSdwG8XE7ip6dixBz48nx1K9Y2nCDF/Fm3FVRHyNljdkZn6kQs215aTz/zOqZpVDWt9LEbhOLe9/E/h0hXoAH265PUxx5EvlSwNFsV33VcBzM/OsKK8akJlVvl224zX9dkR8iOJQ49aaN1So+efA+4DLyvvfpBj0VVzP9vkPwxQfGpXPHA/MAh7JzM9GeZbzzLytYr0flFsRWl/P11Zs5yPAjRHxrVF1/6JCzQ8D/xkRX6B4XU+gONq5itbTDQwDX6M4kfGEZeatwO9GcTqIKZn5UJV6pV4eS9Db48mxVO9YmrCen3BfziF4isx8f4Wanx275MT3gXeTdhz5047XtJx4O1bNCU+87Rbl+34xxS6I50XEAcAXMvO3K9Q8Zqzlmbl6ojXLuifvoO6FFesexvZz/lyVmT+oUq+l7l4AmflwDbX2A/4XcEBmvqps80sz84IKNR1LNeuW8eRYqncsVWpPr4evEXV2cp0i4vOZ+UctR388SWZWOTBgJsX8rJGzEa8Gzho5uqhC3bYc+dPJIuJjmfmXEfEVxu6nCX9bjeK6c6eyvZ9WURyxM+HDziPiRmARcENLH32vyvuprLEfMBKy/yszf1alXkvdacDzyrsTPuQ+IvbOzAdjB5e7ysz7K7TxBcBFbD9o5V7g5Mz8foWaX6e4JMyyzHxRFKceWDd6QvLupJ1jqazf0+PJsdQZY6nndzuO7uSIuJfi/FQ3Vag5B/hHionmAP9BMe9p0wTKvav8t5aLn47yGYp5BSO7Gv+Y4s1Z9ZIL7Tjyp87XdKRmneHzovLfdly89TyKeRojR7b+cbnsTyvUrP0s5xHxR8CHKP6YBfCPEfGezLy8Yt1+4EKK3eJBcSDHyTmxw+MvoRhLI7uenvhvyvtVjs5bAZyemd+GJ9q9guIElBM1KzM/HxF/DcUlZiKi0oT2Hh9L0MPjybFU71iqJDvgTK+T+QNcA/xOy/1+ihOCVqn5TeCtFOF2KsWRft+sUK8P+HYbnvuN41k2gbojVw24k/quGlDra1rW/CLwfooPiedS/PH4l4r9NNiGfvrueJbtYs12nOX8u8CzW+7PrtrOss71FLtzRu4/D7i+Qr0A5nZJP60C9qXYogLFQSyrK9bs2bHUxn7qivHkWKp3LFX56fktX8CeWaZrgMxcVcO3ltmZ2Tqv4nMRMeGr0Wfm1ojYFhEzs+IuwVF+GRFHZ+bV8MS5en5ZtWg++cgfqOfIn1pf09KCzHx9y/33l7sPJqTsp3kRMS0zt1RsW6utEbEgM28BiIjnUvFw/sz8hyjOcv4gxQfwmVn9KgxT8sm7Re6j+vXYAJ6WmTeP3MnM/1vuOpqQzMzyAJu6dzfcGhHvY/uWm5Mo/hhXcTrFF5kFEfEdij/AJ1Ss2ctjCXp7PDmW6h1LE2b4ak8n3xcRJwH/u7z/RsorvlfwMDAUEd8EfjGyMKsdpXIqcGG5ywDgAaqfj2vEHhTfXhN4eg312vGatiN83kpx3qCVPLmfqhw9+x6Ko8lupfimOY9iy0VVQxR9k+Xtqr4REVeyvY/eAFxRQ921EfFp4OLy/gCwtmLNGyLitzLzuop1Wv0Jxdaff6F4Tf+jXDZhWZzf6BjqvcRML48l6O3x5FjqkEv09fyE+yiOxHs/cHS56D+Av8vynCUTrDmPYk7FSyneONcAf5HVrkvVlqNUytp7l7UerFqrrHcmxdnyv0h5zhuKI38mfA2tNr2mL6I4W/TMsp33A2/JzO9WqFn70bNl3ekUHxpQfGhUvWLAn1KcTPcqiud+DMUcnc9UrPt6WuYSZcsZyivUnE5xioHWMXpuldcgItYDBwN3UPxhHzkje6UJ0nWLiBkUZ/Y+mu1/hD6VmY9WqNnTY6ms3ZPjybFU71iq1J5eD1+9LCL+F/DBfPJlIf7fzKx0ZuaIuBl40cibOopLDd2Ymb/+qx85OeoOn3WL9lzX82bgqCyvchDFxXav6dQ+qlsZQJ4iM++oUPObFHMbW/vp0sx8ZYWanwceYvuWijcBz8zMEydas506fSyB46lujqWJ6fndjlFc6fzdPPXipVUusjqbYtLl6JqVNpvG9kOwP5aZVedoALwqM//HyJ0sLgvx/1D9shh3UVwzbeQbxXSKyfcT1o7XtPwW+PqRmlFe8yszz6rQVCLirzLzgyP/VqlVasd1Pe+j+CAa8RAVdz1Fe65pOrIL6+946sWAJ3w0VZU/DL/CrHzqZVaeXbHmCzLzsJb7346ISudQ6vGxBD08nhxL9Y6lKno+fAFfAD5Fcbb4ug47/TLFJs1/q7EmbD+Me8yT701AX0RMH9nkXG6hmj7RYhHxjxSbczcDN5XfXhJ4BVD12mnteE2/TNHW66n3chtLKK6fN/JvVbVd1zO2X7ZjA3BtRHyZoo+OB75XsZ3tuGYgtOe6gUTE1Zl59Mi/NZTcFhFzR3bflVsEqu5auCEijszMNWXNl1B9jk4vjyXo7fHkWKp3LE2Y4QuGM/O8mmvukZnvrblmOwwC34rtZ7x+K8U5YCZq5I18PcVFZkesqlBzRDte0zmZeVzNNVtFTXXqvK7nyGU7bil/Rnx5gvVateOagdCe6wZCcVAIQOVzMpWWAVdHxGqKvn8ZsHQihWL7SZWfBlwTERvL+/OA9RXb2ctjCXp7PDmW6h1LE9azc75i+1l5/wL4GUVYaL3WVZWz8/49xf7+Oo70aq37lDPH11DzOOB3y7vfzMwr66hbt3a8phGxAvjHzKzjyKTWurX2U0RMofjgeaKfgE9n5uSdILBFuXsEii2yv0ZN1wyM7Rd+/yOKI2frvG5gu8bTLIrzBwGsyQleUH5H82hGVJxP07NjqazZc+PJsbRjbdptulO9HL5uY/vFVUfLiewDj4iHWmruCWwBRg5lrWPuS91/1PcEfpmZ2yLi1ymO/vl61nD4bV3z09rxmrZ8E5oKHEJxSPtj1HSUTjs+iFpqP4tiK0PVXRq1zaeJsa8VOCInOpcoxr5eYGvdStcNbMN4+m2KA0t+EcWpHF4MfHyyPtxHcyyNWb8nxpNjqfP07G7HzDyoDTWfsfO1Osq/Ay8rjyT5BsVuwzdQnPulqlrmp7XpNW3HpZraJiJWAa+lGK/XAz+LiGsy879XLF3LfJrMrOMcSWPV/Z121G2j84AXRXHahdMp5tf8MxXHQF3zaRxLhV4cT46lQhvmpk1YHWef7moRMSMiTo+If4mIL0bEX0ZxPpCqdf8wIj4SER+OiNfV0VaKa2lBMVerDpGZj1Bcy/G8LA65fX5NtWtX12uamXeM/FBcbuJ4ig/jfWv6ZrWq/PdXfdvcFTOzOHT/D4F/zsyXsP3qAXWoZT5NRDw3Ir4SEfdExM8i4ssRUflLTkTsGxGfiIgbIuL6iPh4FIfyVy5dQ41Ww+Uk7uOBc7I4oq6OwFP3fJpeHkvQw+PJsVT/WJqong9fFGn6+RQnHfxkefuiX/mInYiIc4G3U5zl+PvA2yPinF/9qJ3LzH9o/bcGEREvpdjS9bVyWV9NtWvVjtc0ipPBXkjxR2MW8NmIqHqaDTLz9NZ/azA1IvanmK/x1ZpqtsMlwOeB/YEDKI4kvrSGupcC91CcyuCE8vZlNdT976P+reqhKC7aexLwtXJu0YQv3dIuPT6WoLfHk2OpU+QkXVSyU36AH4xn2S7WXE85n668PwX4YQ1tfTXwVxRnUj6T4vphVeodQ3Gtq/eW958LfKKm13Xk4qXraqpX+2tKccHvGS33n05xtuuqbX0BxQf7m0d+KtY7keKw9XNb+umLHdhH3xtjWR0X1v7+GMuG6mhznT8Uk6NPB15W3p9bte/b1E89O5bKmj07nhxL9fZRlZ+enfPVoh3n/thA8WYZ2ex+YLlswiLiUxSbTH+H4pxkJ1Dx3FmZuRpY3XL/VoqjPztR7a8p7TkZ7N8C/cBhFNdhexVwNcUW1gnJzC9QfOsduX8rxTfXTvP1iDiD4tt1Ul6LrpzUTE78COJ/jYglFFsBoHjvVzoqN4oTjb6Xop+emGaQFSYeZ+ZPgY+03N9IhX5vo54dS9Dz48mx1CEMX3A428/9AcWH0s0jR/HkxI7WeQbww4j4L4pBcwTFBU1XUhR97QRqHpWZL4yI72Xm+yPiw0Cl87WUR8A85XDXKoOmRd3z09rxmo55MtiI+ERZcyJB9ATgRRTfrN4aEfux/XIWE1Ie/TRWP1W6YgL1z6f5o/LfPxu1fAlF+yd6Fu23AX/J9ukAfcAvIuLPmPhRxIMUu1teTbEL7mSKXTAT1nI0IRQn7Xwa8HBmztzxo8ZXuuLjR+vZsQQ9P54cSx3C8AXtODHgmW2o+cvy30ci4gCKS1fsX7Hmu1tuz6D49jdcsSbQlvlp7XhNv0T9J4MdOXXHcBTXufsZxZaFKlrnpcwA/oBiS0MlWfN8mmzDEcRl3XYcpbdvZl4QEe8a2QIcEddVKdjazogIisnCR+74EeNW93yaXh5L0MPjybFU+1iasJ4PX9mG84aUb8C6fTUingl8CLiB4lvB+VUKZub1oxZ9p/w2XFlEvJri4IXWzdATvs5bO17TzKxyNv8dWVv20/kUh7E/DPxnlYKZ+cXW+xHxvyl2v1QSES/gqbsKdqtN+7/CyPmtflK+V+8CnvUr1t8lWUws+T/lrrMzKtZa1fpvVb08lsDx1AY9O5aq6NmTrHazKC5iOyMzN1es0zpAplDsgv1EZv56xbpjzk/LzFOq1O02ETEf2DtrOIHjqLq/DnwtMw+uUGPM+TSZeUItjexwEfH7FNc3PJDiSOe9gfdn5soKNf+w5e4UYDFwTGa+tELN2ufTdKN2jaWytuOpAsfSxPT8lq9ulMWFsOu4eO31bD/j9TBwG1BHQKp9flo3yszb66gTTz4zeQI/pfgQqaIt82m6RWaO7HraTPEloQ6vabk9DNxOsbukitrn03SjusYSOJ7q5liamJ4PXxHx58DFmflAzXWfDszNzJvrrFunds3RoQ3z0yLiNRTfTrdVbdwYtffI4mSz5yEcNQAAIABJREFUHalN8zRqn08TEf9CcSbqr9fdTxFxNHBIZn62/Aa7V2beVuf/UVW250z/tc+n6eWxBI4nx1J9Y6kKT7IK+wHXRcTnI+L/b+/Mw+2oqrz9/ogIMgkoojYyisgsggyKLUpjAy20oEAzqZEPaaAlfAjNA6gIDrQI6ifNIDKq6CMIONAyRCSARAXCEIiK0mCj4tDYtiCIEFjfH2tXbt2Tm4TU3uuec3L2+zz1nFuVnHX3rTqrzqq911q/nVNyXxbp5nYXLtmDpNc0lUSDhKSlJR0h6etp+xdJJRrZ9ean/YKx6seu7AP8XNKpkl6daQsASa+X9GOSsr2kzeUNKAcOSbtLOi1tJSRdevNp7iA/n+YsYD/8Ov1bWs7JJi3pHAsclw4tzQDOKkhaQ9KV8m7kv5crZqyRaXZcPo2kLcjPpxlpX4LR9afqS0BZX+rORM2/Rm3Dp5//Hu+ncj/wCWC9DHuzgBfSauRGZiO7NMYDSI1V8ZYYW2faPA/vSv2WtF0InFf43C6Dy3mUsLUSXnb9Q/zG9j5gxQx7P8KfTtvXab4mhB3sbg9MTT+vBqyTae/fgOuB96ZtOvCJgtdobWCzgvZeiE/r/xKYCUwFls6wd1f6/Lev03wNKPu9pesyFV9ReB7wHmB6ps23pfO5Cd7CYBawe4GxDrwvAesBy6Sfd8B7EK5c4G8fWX8q7Uv4BM7epf7Wlt2h8aXO4+nXLx60DV+v/yz+5HY2cCdwakdbP0yvxb4s0pjOJHWiBlYBbsu0OV+35ImODdKGy5ccic+mXQ38HHh/R1s/muA65XaQPhH4NvCztP9y4JZMm7OBpVr7U3I/T8HXZxreqPhb+CzLGcCMDJu3ptemO/XyBfxpdcaWdMCTcA/KtHnXczk2KNsQ+NJd6Yv3lcDP8Jn07xT4u0fWn4J86faAv3uofKnLNvLLjpKmSZqFq9DfAmxqZofilX9dux7PkbQfMEXS+pLOwJ9YctjGzA4ndZA2z1F7fqbNZySt1+xIWhd4JtNmCJL+UdKVeP+gpfFZv13woPkDHc3+UtLrAUtLsEcDP8kc6h64sPDjAGb2MGUEYVdu/ZzbaDCEdH1uxitddzOz3c3sa2b2fmCFDNOXSvo8sLKkg4HvktlmBbgI7+z98rT/MzwQyeEPkg6QNCVtB+D5jgPFEPnSs2Y2F/epM8zsGPJ7GzaMqj9F+NJ3JR0t6RWSVm22TJtD4UtZ9Dv66/cGnASstYB/27CjzeWAjwO34U8sH6ele9bR5o/wJ7TmiWU1MvWpgB2Bh/Cb8I34E/Cb+31NFjDWi4G/XdDf0dHmi/EKmN/hCbJfxpMyc8YZ8WS5Ly4Fc1E6Dw8C+/T7mkwwzl0nOLZMIds74TMfpwE7FbB3W3ptz9RkPVkDa+GzE/+dPk/fwItu+n5tesY5LL70o/TZv5e0dE+ZtICR9qcAX3pwgu2BTJtD4UtZf2O/B9DvDfjScznW7w3YP30Yf5WCufuAvQrYXQbYLG2lvigj8tM++VyO9XvDVQM+DzyAS3n8gI5LOT12X4bPqO0OvLTQWEvnpt3xXI4NwoY/cLyIsSB5W+DGfo9rkv72YfGljYDPAfum/XWAYwvZLupPBOSnDZM/DfpGUG5azjbyTVYl3WFmr23tT8GT4zfKsPkq/Et4bVrtPCyzmVuqTNoRD26uN7OsaX1JhwOXmNn/pv1V8BtdVpWSpLOBZ4G3mNmGye51Zva6DJvjrlM6Ntu6aW82718ND5DWZvx1ytJ4k7QT8Fb8Ol1rZtMz7e0BfM9SU91UVbWDmX0jw+aJeOPCDczsVfKWIJeZ2Rs62Hop8Df4bMd+jOmnrQScY2ZZFXXyhoufBF6SbIvuOnSNzdfiuTOb4DMrqwHvtIwmnpIuBqb1+NPpOZ8neb+oTwAvN7NdJG0EbGdm52fYHBpfiiDIn+7C/WltvMnqN4GNzWzXDrbC/CnIl5YDjsJnpt4naX38vnLVIt66MJsRvnS7mW3V9f2lGdngS9JxwPHAC4CmL42Ap4Bzzey4Bb33Odi+GzgHr6aYl0Nl88v5LK7dKXiicPvm9tCC37FIe3eZ2Wt6jt1pZlt0H+XYzb1tS9LdZrZ5B1uHAofhT5b3t/5pRTyR/YCMcc7Ecyp6r9PlC3xTH4i4TunLYgv8Sbq5Rp2+gCW9G69G2gpfZm94DLjIzK7oOs5k/3485yU3h6jX7vOADXC/v8/Mnl7EWxZlb75rUuA6XY1XIZ9gZpunMd9pZpt2sDVUviTpQSYWwO4q0N7YjfCn5p53DPCkmZ3R1WakP0X4kqSv4df9XWa2SQrGZvae48W0GeFL/wY8gjdafbw5bmb/09VmDiPbZNXMTgFOkXRKTqC1AOaa2dklDcqbwZ6I51Q8w1h35s5Pq3hBgCxF4Cm4y03iB3g62WrsrobPhHXhK3gl1imM1/V6rIDTLGdmuZ2txxHxZMnE/fhyffcpMzNJzTVavqshc12/iyW9Iyhw/V1A4NXM+s5J+6tIyp31XUrSKpYaNqek49zr9GIzuzQ9LGJmcyV1LYoZKl/Cg4+GZYG9KNOXKcKfnpa0L941venO3qlnYrA/FfclvC3TPunvx8yekLL7ZUb40j7p9fDWMQOygvmujGzwJenVZvZT4LK0BDEOM7sjw/y3JR0GXElLBijzBjcNn8otWfFxDfC1VP0C3vfnmgJ2P4f/7S+R9HFceuODHW2Zmf0ifVmOQ9Kqmef0Kkm7mtl3Mmz0cirlZ2lul/RpvNUI+M0jaxaV+aue3kvHqidJB5jZl4G1JR3V++9m9um8oXJ7err+BuP9KWdG7WAza84nZvbHdB5ygq/TgR9Iuizt74XnZ+bwuKQXMfYgsy0u49KFofKlCe51n5VXpn8403SEP03Fe3F93MwelLQO8KUuhoL9KcKXnpIrujSf0fXIl78r7ksWp+jSiVFedjw3rU/fMME/W05+Vpoun8hm5wg7jXMn89LrIkhaCg+4dkyHpuNNVrPbTahQfpqkq8zsba0liPYTVe45fQyvRnyKse7HufkPt3TJm1qEzeWBDwF/lw5dh9/kH1/wu56T3SK5aZIOMbPPpzyy+TCzkzKGiaQLJzablf9xD94Isz3rO9vMNu5qM9nZmDF9u++Z2Y8z7RXLTRtCX2o/FDfiyod2SV/osRviT6WI9KcgX9oJf7jeCD+XbwDeY2YzutpMdkv7UvHctKzxjGrwNWxIOh/PT/kPxj+x5M4qhFA6P21YkPT/gJdS9smyUhhJn8LL2duzvr80s659rsIonZs2LPQ8GDfiyqfZAOrlRuWnDQtpdnZb/DP6QzN7pM9Dmo+I3LQcapNV6XB5tUuzv0paMsyxGaGZ+BA+M/V8PEG22bKQ9Nn2awlSftrv8PFehQeMWU8XkvaQ9MLW/sqS3p41ULdTWuNtJbyA46147sduuKxFFpL+tf1awN6ekn4u6U+SHpX0mKRHM21ePIEvXVBgrBE6b8fiEiOHpu16IPvcSvp++7WAvcNx4eM5ZnYvsEKB+9NQ+JKZvbm17WRmB5cKvEr7Ez4r97q0vRFPvcjSTIzwpwhfkiRgF2DLNIu0nKStc2wmu0V9Cc9NO5U0M2suAJ+t5dwZG4B+F/3cmFjGILd5aZhmIt7ZeIWCf3/T5yjrb+6xeT+ZDRYn6TpNpPF2SslxB1ynIn1+0jXq1ER4ITbnux4lPlcE6LxNwnUq4k9Bn/uh8CW88/yn8Yq/2/E8oFI6sUX9aQG/Y1bm+4v7U4QvESB913ONSvnSTLy7QWN3PVJT7H5sI5tw3yKi4u91Nj4v4Xvy9hOdkbQJnsC5atp/BJ8+nZNjN4hf0j0peEFEVCjtCrzGzJ4Fmt4ydwI5bUbWwHN0mryvm/F+Nb/KHOu8X1HITkTVU0SFEsBqZtbOVblIUicpIEmXmtneKedromWinOrhCCLuT0PhS8AFeJ7b3mn/QPxBds8Mm70U8acF5KflntMIfyrmSy22sdRaCOYVr5Somi/NiXhB2SskXULKTevXYGrwFVPx94yk9czsPwFURjPxXOAoM7sh2dwBr057fabdCB4AZkgqmZ8WUaEErvHWVHmV0Hi7EC/p3yvtH5CO7VTAdkkiqp4iqv0g6bwBX037+9Jd521aei2xxDwZRNyfhsWX1jOztr7uSfL+dIPI6a2fm/y0vSf+r4tls7Q/lfSlhpKthcIws+mS7mAsN22a9TE3rQZfnvtxCJ73AaniL9PmMcANkh7AL/Ja+FRvDss3gReAmc1QRm+mYB5K2/Mp0zcM4P14hdLX0v51jO/X0oVTgDtTYq+Av2V8/6MuRDxZRtDOTWswoHPwZWZflLcCaCqU9rTMCqXEe/HZxM/gY5xJR38ys9+kL4qLzOzNi3xD/4m4Pw2LL/1F0vZm1uT+vAH4S6bNECI+S0H+VMyXWjSthVZXfmuhMFq5aeua2cmS1pS0tZnd2o/xjHzwlabJz05bNvL2DX8B1scrlMArlHL7njwg6UOM9Y45AJ9hGjgslUJLWiHt/7mAzcfJv5nPIzni9/GnoEb26Fgz+22m6Ygny+KYWe4Nd0F25wDFlsJToLSnme1eyqaZPSPpWUkvtCQxU5CiCbyl70/J5rD40qF4s9FmFu2P9HGZaGGkMZ6IB50ANwIn536+SvpThC8BmNklKUhsWha9vVBKQ+lk+LNIsnfAybhiwOWMfWYnlZGvdoSyFX/pZnmmmf3VzGanLTfwAn9iWQ2fmbgi/VxCN+0r6fWSArYAz09L6/9zgDmSZsl7tuTaLVahlHJovmNmvzGzb6Ut98sC/JrsDfwW+A3+FFgi0JmRXifqS7fYBFUQFq9QMu85t28JWz38GbhH0vmSPtdsBez+357XTki6NL3eI2l275Y7yGHwJTO7K+XObob3ZNvCzLJyZ1vMSK9F/AnPT3sM9/29gUfxdIMsSvpToC8BLAdMwWOKFxSyWcSXWmxjZocDT4LnplFuZWaxqX2+AE2gRZhp7zTgB8AVNoInWK7zdkJPftonzCwrP611neYTBu5o72Lg383stlxbw4ak6Xjg3Z5J3d/MsnLTSvtSsvkZXKqlV5OtswqFXD9vPsylXfqOpJelJdK1Jvp3M/uvTPsD70uSPgGcauPFlT9gZoO4pDWRXuR8xzrYLf3dFOFLH8bz0S7HZ6veDlxmZh/LG21ZJP0Iz5G+LZ3T1YDrSt2nFpeRX3YM4hC8k+5cSU9CEeX4VwFHA2szvnFp5078ye4/ABvj2mmNzZNzbBKfn1ZqOnobYH9J/4XfiJrr1LniLTn0wcx/nbJmKeXVrhsx/jp9McPksOSmATRfYO3PpeHLB52ICLLStT+W+a/TYo9zEnPTBtaXgF3M7PhmJ1XR7UqBfKIAfxqW/LTivgTsD2xuZk8CyAWs7wI6B18lfanFQOWm1eArADPLbn46AZcB5+DJttnyPwCSzsGni9+c7L4TKJF8OCz5aX8fYPObeHuJ71LuOp0I7IDfiL6DJ41+H8j5shiK3DSISWYGTzMwsyOb1wImL8FnFP4B1/l7N/DfXY0F56aVJsKXpkhapknbkOsHLpNrNMifhiI/LciXHsYDpCfT/jLArzNtFvUlCM1N60QNvgqiCQS62+RM7QJzzaxY0m3i9Wa2maTZZnaSpNOBqwvYfS9wEmOVczdTJj+tCPJ+OeA5GqVZzsyOLWzzncDmeLPBqZJWJ7N7NjFVT0XRBKLCbSxfWqtJjn5Tpp2GF5nZ+ZKmmdmNwI2Scpfhmty06YxfJjoi024Rgn3pEuB6jekRTsWbV+dS3J/M7C5gc0krpf0stYjSRPiSpDPwe8ef8Nze6Wl/J/If4iN8CcZy04xyuWmdqMFXWZpeL8viTfbuxqffN8M7NG+3uAZbN7dvy2VFrmR8X6b/mfCNz41mWvwJSS/HZz5elmGvGdMfgYH4clgAsxgTFl4Tf0oV3qfoIWCdDNtXSdrVzL6TPcox/mJmz0qam27uvwdekWMw5QwVrXpKlKxQamaQN8Arkr6V9nejzAxtaRrNxd+k5fyHSU2RM2gKbAaVMF8ys0/Km1M3AtgfNbNr84YLBPhTYH5aKX+K8KXb0+ss/HupYUZHe22K+9IEuWkXSupbbloNvpwiFX/NlK6kK4DXmtk9aX8T4CMdzbZvbuA9xOb9SiBHuPUquXbYp4A7kr0vZNgD4vLTKFShZGbrAEj6AnBlEyhJ2gVPFl1sJD3G2HU6XtJTjN1AsvL98KaYK+PXZhY+G/KDDHthuWkUrFCysZYlN+H+9Fja/wiuFzpofCwtO30An1VciczzEFgAMCO9DpwvNaQ80evM7BpJGwAbSFra8oXFi/sTcflpRfwpwpeCi1OK+xIBuWk51GrHACTNMbONF3VskJC0DLBsidyS9LR6Dn5jm5f3ZGYlumgXQ9I9Zrbpoo4NEpLWBlYys6x2A6ki9Wbmv0aX59iNQNJ9eKuBJvdnGWC2mW2w8Hcu0m7xyswIAnLTihPhSyk/5424VuD38ZmWp8xs/6zBjv8da1PGn2bjsnLt/LTbB+2eH+hLw/AZvQHYozU7uTLekSB3UqATdeaLkIq/2ZLOYyyPYH8g17mXBQ4DtsdnWG4Gzmmi+FySM5boRwYx+WkRFUoPS/og46/Twxn2AJC0J63rZGbfyLXZYGa/KGSqeG5aUIUSeCL0rZKapY23Axdl2hwmSuemDYsvycyekHQQcLaZnarC8kIF/al4flqQP0X5UvHPaCmCc9M6M/LBV1DF31S8+qXRkbuJ/A7VX8STWs9I+/vh1YR7LfAdk0xkflpQhdK+eFfq5kZ0I5lNCCWdBbySsSrCf5a0k3lzv0EiIjeteIUSgJl9XNI1eEALMNXM7sy1S0CD4WFgWHwJkKTt8EDuoHRsSqbNEILy0yIq/qJ8aZCJzE3rzMgvO6ZKv81arysAV5vZG/s9tjaSfmxmGy3qWD+R9CDj89PamJl1zk+TdA9jFUqbNxVKltkUtDSSfgpsaMmx5HJTc8xsw/6OzOnJTVseKJabJmmWmW3Z+FI6dpuZ9UW+Y0mh9PLoEPnSm/Ccn1tScLMucOSgVHq2SflpTSL/Bnhi+9U5+WnD5E8lP6PpnvlOM7u00PAGkiovNH/F39MUqPhTQcmixB2Stm3Z34axiL4Tcg5IVSAoCY12tWdm65jZuum1d8spDIB0Y8Mb1xapUAKKyqwk7servhpekY5lIWl7SVPTz6tJ6lRFZmYrmtlK6XUpM1s2/bxiZlEA9FQoSdqC/Go/AFRYtkjSMpL2k3S8pA83W6bN1eVyRVen/Y3SktmgMRS+ZGY3mtnuZvbJtP9AicBL0nop1wlJO0g6IuX/5HATsKykvwGuAQ4kfzkvxJ9K+1Jp0mez1P14HAHfy52pwdf8FX+/YGw5IofSa+BbAjMl/ULSL/DqnNcpab91tHkW3v6iWR54DDgzd6CSlpV0lKQr5JqBR6actRx6K5TuIL9CCeCfel5zWRH4iaQZKcHzx8BKkr4l6VuLeO+EpGWiY4Hj0qGlye/zhaQ9JX1a0umSsirTEu0KpaPxZfxSumzLpddSSgnfBP4RmIv3z2q2HC4CrgVenvZ/Bgxi8vFQ+JKkGyR9r3crYPpy4BlJrwTOxQPP3Hu+zOwJYE88P20vPI84hyh/Ku1LEXxX0tGSXiFp1WYrYHdgctNGPufLzD6afrxc0lUUqvgLYOcAm9s0U8Uwrzy6hNBo8fw0Mzss/XhOylnIrlDqoVQ/nazZkwWwB7AF/iWJmT0sKUtFISI3zcyuSj/+Cc+hHGTWMLPSPvViM7tU0nEAZjZXUgmVg6K5aUPkS0e3fl4WeAceLOfybLo2ewBnmNkZzT0wA6lwftqQ+VPp/Ml90mv7fpTbWmmgGPngq03hir+iWKaQ7gJ4Wq4f1+QnrQY8W8DuJj25aDdI+nEBu0DRCqXimHdjLs1TZmaSmutU4on1LYzPTbsYmFPA7rAwU9KmlnrxFeJxSS9izJ+2xb84szCz09qvJRlwX+ptTXOLpBLVaU9L2hdPYN8tHVs60+aR+Mz0lWY2J+WnZfVQGyZKf0Yt9Y9bkqnB12jTCI2+RGWFRu+QtK2Z/RDK5KeNOJdK+jywsqSDcWmg3Ga4TW5aE9QXyU0bdOTJ5obf+6ZKegB/4CohBH0U3jl8PUm3AKvhPpUz3mXwGZ+1Gd8MN6cVzlDQs8y0FJ568cIF/PfFYSpePfhxM3sw5U9+aRHvWSjpoevG1v4DDLbKRzEiPqOSlsP9aU0ze5+k9YENWrOBQ08NvkYYGy80KsoJjTb5aQ+l/TWB+5ovvswvuJHDzE6TtBPwKF5F9WEzm55ptslNuxUPRrbGc4G+lX5nhPRQDqWWst5WyM58mNkd8gq9DfDx3pdT7Zb4Jj57NosBnZUPpK3uMRd4kLElvc6Y2Y9pBUZm9iDwyRybKb9zvtYBXXtyKbbir6QEGMR8Ri9M9l6f9n8NXAbU4GtJQZLwdfp1zexkSWsCLzWz3Ontomvgkt6Pl4P/sYS9Fj/Hv9Sfl37Pmmb20MLfskgi8tOQtD2wvpldmJZIV0g3zhxmpNdiSwTy7tZrmtl9pWymYCs34GpTPDdN3rLgE8DLzWwXSRsB25nZ+QXMl5JZiVi+B0DS4cAlZjYn7a8iaV8zOyvDbPHcNEnrAb8ys79K2gHXnv2ipc7fGcxIr0V8KWrpSWMtcXp/X04+UdH8tNSy4l+BiOCrmARYIiJ/cj0z2yctD2PebLdE0Dgwvf1qny/pbDzP6S1mtqFcEPW6QeulIuljeBXRHcAFwLWWefFSQHci8DtcYqbE0ksIqeJvK3zq+VXytiCXmdkb+jy0cUjaDTgNeL6ZrSPpNcDJOTNJ8o75nwRegl+j5jrltoYoirzFwoXACeb9o56H95IaWLmmkki6y8xe03Msq++RpHPxpPBiuWnyLvFb4ctE38FnLjY2s11L/Y4SSFoab1bdVKjNAD6fO5uY8vIalsULgVY1s6IPJJJuNbPOrXvk2oOP4I1W51XiWkaz6giCPqMz8RWZW1JR2HrAV3PO56BRg68JmsNJutvMNs+0W1qyqJmleyues7AV/lR0vpn9Z0d79+MVj3/IGddkkL4wtgDuaF2n2bmBogrLrKRl3LcAM1rjzNW4ux/YrdCScBhKDSB7fGm+gKSD3SjZoqKkZfXNWkUMU3DdvMVuOdCTm7Y+UCw3rXXPOwZ4sqn2ywkSk93SvnQengjfyPQcCDxjZv8nZ5wL+F2zzGzLjPdPlJ/2OcvQTEwzdL1YzgxdSV8K/ozuhOcfbwRcB7wBeI+ZzciwOVD5kyO/7EhAxZ9iJItIFW+/BX6LT2mvAnxd0nQz69KU7pcUqMaaJIpX/ClGZuVpM/tTzwx57hPO7wY98EqEVPsRJFsUwDXA11JxBMAh6VgXwnLTCKj2C/Kl1/U8BH9PLuGThaTXtnaXwh9kc78Li+enBS27lvSlyPzJ6ZLuALbFz+k0M3sk0+xA5U/W4Cum4u/1NiZZdJKk04GrcwxKmga8C5+GPg84xsyeTomZP6dbR+AHgBmS/oPxGoyfzhxrRH5aRMXfOxmTWZmacpZym5fOkbQfMCVV6BwBzMy0ebukrwHfYPx1uiLHaEBuWvFqv8SLzOx8SdOaqjJJtxWwW5pj8YDr0LQ/HffVxSYyN42Aaj9ifOkZSes1M/vy9g0l+qad3vp5Lt5Ye+8cgxGBUlDFXzFfCs6fFB7Az8vFlrR1Zi52RG5aZ0Y++LKYir9eyaI/kC9ZtCqwZ+8HPiVmdn0CeShtz09bKVYHbktPLkXy0yym4q/RYisps/J+4AQ8SPoq3vH8owt9x6JZCXgCX3JuMKBz8NXOTQOK5KZZTLUf9MisAA9TSLaoJOayKGenbWCxgGo/YnzpGLxH4AP452ktPHDMwsyKNywNyk+LqPgbCl/C1VeexVM4Tsabdl8O5ORiR/T268zI53zBvNyM1Rm/Dty54k/Sh/Du7jvicj0GfCEnoVPSl8zswEUd62h7BQAz+3OurZbNovlpEci7vB+PFzJ8APgzcJeZZd/gB52g3LSm2u9/0/4qQG61H+nh4mb8y/wMPBg9ycw6yTWVRtKlZrZ3KwdmHINWwKKAar8oX0p5Ok3e1H3mjbCzkEv2nMhYoHQj/uDReYk8Ij9N0u1mtlXJfORB96WGkrnYkblpOYx88KXgir9088iWLGo+jK39KcA9Nr6T/OLa3ARfbmiefB4B3mWpVD4XSZvjwdfOePn5tkCn/DQFV/xJWpsCMiuSXoWXna/N+GC+c3K4pDXwG2VT2XkzngPxqwybPzSzbXtublkFDBMl15dI5B50JL3MzH4jaa2J/j14CXGxUXC1X0FfigrmLwfuZXygtLmZ7Zlhc77AoECgtMRX/C0IST/CZ/xuS3/7angXgsW+lyzILxv65p9mNtIb3tX7Rf0ex0LGdxw+5ToXX3J7NO3/ATgl0/ZM4M2t/R2AmQXGPA2fLr8Wv7EvnY4vBfxnxnXasN/X4zmM8258+WFrvOJpS2DLTJvT8SD2eWl7Dx7E5tg8H9fcnI0/DZ4BnJNp8x7SA13anwLM6fc1maTrPgW4od/jyBj/rH6PYYIx3TXBsTuD7M53bDFt3oH3pmr218Urs3Ns7oTPyv03nij/C2CHfl+XSbr2++P5o78GPg7cB+zV73GV3EY+54sBr/gzs1OAUySdYmbHFTa/vJnNa4hoZjNKVBESk582LBV/c82sdM7PamZ2YWv/IklHZtqMyE0rWe03VJjZM5KelfRCy5zljiao2i+CKZJk6ds4zfaXyE39i6Ttzez7ye4bGMvT7Urx/DSLqfgbCmx8LjaUU18ZGAbR4SabkIq/Ukh6tZn9FLis56YJeJJzhvkHUn5aU+l0AH4+clm3N/Bq8tMyHCik4i+Ab0s6DK+gbY8zpzHiHyQdgAdJAPviM5+dMbMjasgFAAAVdUlEQVQn8ODrhBw7PRSr9htS/gzcI2k645tiDprGX/FqvyCigvlDgYtT7hfAH/HZ5M6Y2fVNNWI6lJ2fVrriT7GSRREsh88oG/CCPo+lODXny/vTzIeZnZRhs5hkkaRzzcuMJ5LsMMvLJVoFOAnYPh26GfiIZbaICMpPu3CCw2Zm7+1qM9ktKlmkmMaIa+HLgtvhN6KZwBGWVxRSPDctCsXKFhVD0rsnOm5mF090fElCAZJFKVg4hLHZj+nAeWZWot0EqSoTM3u0gK3i+WkKUF9pkvi7vn+ykPRhPGXlclIXAlzR5GN9HVhBRj74aihZ8RfhNMOApOPwiqcX4K0RwB3nKeDcgGXTLDQkkkURyJtVnoPn5s37MjOzWR1shVb7acRli0oTVO03FJJFAJI+AZzaEyh9wMw693eMKDYpWfHXsjkskkX34UUQT6b9F+B5eZ0VAwaNpfo9gH4jaRNJdwJz8AaZsyQtthxID9uY2eHAkwBpJikrV0HS4ZJWbu2vkpa3cmy+StK5kq6T9L1m62rPzE4xsxWBT5nZSmlb0cxelBt4SVpD0pWSfp+2y1MVYA57ALuTbkJm9jCwYuY4l5Z0hKSvp+1f5D2AcmyuJun4dK0uaLYcm6TcNDO71cxmNVtHW9PS69vwjum9Wy4vTkslzwKY2VzKNNssjqTPtl8HlAvwop290/YoHtzm8Gy6LnvgOn/HkN/bMOp87tKekUv359wgcUpa8QCK5acVV18B9gEOB27CH7xmAbdn2ozgYVryR8AyePL9EkPN+YJzgaOaxPM0Zf4FxhrbdSHCaQ42szObHTP7o7zTe07Z9WX47Md5FPgyC85PuxBXpN8r7R+Qju2UYbO4ZBHeYHNpxq7LgelYjh7dN/El4e9SLugolptm3mZhCnCRBTSwJE62KIJmNulNfR3FwlnPzN7R2j8pzVzlUFyyKBFxPqdIWqbJyUqzKstk2ozIT2vUV1ZXIfUVi5EsKoakM3A//xM+GTI97e9EAYm+QaIGXzEVfxGSRRGVP6Ur844C3sf4hN4Gw5t6diWi4i9CsihCj245Mzs200YvTX7SMa1jhpfILzbB1X5RskWjSkS1X4RkURSXANe38kinMtbzqyvFi00iKv4UI1lUkmYWbhb+HdowY/KHEsvI53xJuhLv0dKu+NvSzPbItPtqxiSLri/gNJ/Cy5fbT1a/NLMPdLDVNFU9ApcBKVmZF4Kk6/GZrnbF31Qz23HB73pOdnfCO/ELl0HKkiySl4bvZeP16L7eLkDoYPNjeP+17+SMLRpJ3wS2wL94ilb7pTyv0rJFxZkoT2fQkEtJXQyMq/Yzs2zR6tJEnU9JOwN/l3anm9m1pWyXJK0gbI8/GN2SuXqAvGJ8Ft5Me5MUjM3szVerxFODr7iKv9KSRcUqfzQmL6IJ/jmrMi/Zj6j8KV7xF4GkHfEgcVy/n/bs6mLYeoyx67Q8XrjQBB1mGd39FaBFp6Bqv4jPUxTDEHw1qGy1X3HJomS3+PlMKxuNFuUGeFB/dW5AL+mzZnZk81pgnMUr/hQgWRRF6fM5aIx88BWBgiWLBh0NicyMCksWpQB5W/zJsqgeXWkUoEUXxbB8nmA4gi/FVPuFSBYFBV+zgDcCqwDfx5e6njKz/TPtFh2rAir+NESSRcPgSzmMfM6XYvodTcPX0bMaYfYS8GS1LHAYY9PaN+MSM09mmi6en5aKFg5m/uuU0+frVGC33CXh1lielXRmulFk6dr1kgLFedfJzL6RaTIiNy3qaTWq03kEX0mvl/R1FAtnFzM7vtlJxTu7kpGXOsG97rMpyMnVi4w4nzKzJyQdBJxtZqcWKDiIoKn4a+7HJSr+TsQLAV4h6RJcL/Y9mTYrHRj54IvCFX+JKMmi0pU/X8RLzs9I+/vhuW97LfAdz42Iyp+Iir8IyaLrJb0DuKIJFnKRdBbwSsby3f5Z0k7m7Uy68oyk9Xpy00qc14jqtKGRLTKz09qvA0rxaj8FSRYFnU9J2g5vhH1QOjaloP0sIiv+bIQliwaNGnzFaPENtGRRi01sfNf5GyT9uIDdCJmZiIq/CMmiQ/BqormSniRzKTPxFlxUvJn5uRjvS5dDcS26QIZCtkjSMsA7mH929uR+jWkBRFT7FZcsCjyfRwLHAVea2Zz04LHYOZmBhFX8SWUliyrdGdngq1XxF6HF91Dans/gLo8A3CFpWzP7IYCkbSjQcM/MnsV7W5UMaq+StGvhir+V8E78b20dM6Bz8GXeZLY09wNrAo1e5ivSsU6k3LS/AEW16KII+jxF8E18tmIWrXvJoGFmn0xLzE2130dzq/0spr9byPk0sxvxrv7N/gN45fdAkFugsgjOIqmvACfjKx+XA0u0+sogMrLBF+7Q7Yq/Iv2OYEwXUgUli4LYEpgpqakYXBO4T0kmJqdAoFTuT0/F3/GSilX8mVmxmR5N0FS253fllIivCPxE0q34udgan7X7VrK9++IYi8xNK4mCZYsCWMPMdu73IBZFqva7zsyuaar9JC2dWelaXLKIoPMp18md6POUq2taPD8tIIdymyaJHebl+w3qBMEw5E92ZmSDLwvs9CtpEzx3atW0/wjeVyV3qag0kV8URXJ/gmaSAJDLE52BJ52C55RNM7NfdTDXLLssi+e73I0HjJvhs4nbZQw1N2l5IornpgXQli0aBmZK2tTM7un3QBbBTcAbU5XjNfjncx88B6orFwD3MrbUeCDecmXPDJtR5/Po1s/L4kubc3ONBuWnlc6hjFBfCWFI8ic7M7LBV0NQxV+EZBEUfhIws/9a9P8aHAIq/opJFjXLLpKuAF7bfGGkQPwjOYNMyySlichNg4KfUYuXLSpCa2buecDUlEf3Vwa3xUxEtV8xyaLo82nza5jekmaVOzNE+X7FJYsiGKLz2ZmRD76IqfiLkCxa4p8EFkZQxV+EZNEG7Sd1M7tX0oaZNosTNaNY+jNqsbJFpRiWmbmGiGq/kpJFoeezle8LXpm5JWPd/rsyLPl+xSWLghiK85lDDb5iKv4ekPQhxksWPZBjcBSeBBZBRMXfHyQdwHjJotzebLPlDUy/nPb3Z4DyqiJz0wI/o38G7kkl90Vli0owbDPIxFT7HQpcnHK/IEkWdTE0Ceezne87F3iQsSC0K0OR75dYDg+2DXhBn8eyIIbpfHaiBl8xFX/vxSWLmqq5m9OxHIo/Ccg78X/ZMqWUJomiFX+J9+Iznp9hTLIoNwl/Kv5F1OQr3USBKr3Ui2lNM7sv01RkblrU0+oVZFSgVsYTUe1nZncBm6ugZFEUQfm+Q5Hvp/kliy6UlCVZFMRQnM8cRl5eSNJP8HL7cRV/+BPRwORrSLrXzDYpbPNjwD/hwuIX4MLSpRqDHm1mpzWvBezdiJdDj6v4IzWzXdyKv2FC0m7AacDzzWwduTDyyTl/c8pNO7E3N83M3plhs/hntFKeiGo/BUgWRaGCuqY9+Wnr4yscxfLTAu6jxSWLShJ9PgeJGny5YPMC6TIFrgDJIknnAmeUfhJITffeis/YbAVcCpxvqfP5oCBpodU+XZLSFSNZFCEDNQtfdp1hY2K495jZphk255jZxos6tpg2Qz6jyfYSLbI7mUjasrU7r9rPzP41w+Z8+ntK2nxdbUahgrqmEd8fkaTAe49WkLwyXvGc22ajCMN2PnMY+WXHoItZTLJoEip/TNJvgd/is32rAF+XNL3rzTgi9yeo4i9CsggCysPN7E8eJ88j96mpWG7aJFX7RcgWjSQR1X4ESBYFUkzXNDIYKHkfVaBkUUmWpOBqUYx88BVEScmisMofSdOAdwGP4IHiMWb2tLwD+s+Brk/Cw1KpEiFZFMEcSfvhX3Dr4/k5MzNtlsxNG7Zqv5EmqNovQrIoiihd09KUvI+GSRZVujHyy44lad3UjgB+T1nJouJIOgm4YKKnDUkbdi1BHpbcn5TzNtPKShbNW26ZaCmmo73lgBPw5WEB1+KSMDm96IaK0ud0lJH0IPNX+53ctInIsLszY5JF0y1TsigKSTvi/fzG6Zq22wMNAsNyH610Y+RnvgpX/IVJFgWxbm/gJelLZnZgZu+XkEqVUhV/CpQsisDMnsCDrxNK2q15VKNJRLWfAiSLojCz69MM8qDrmkbdR6vfDwAjH3wBqwO3Scqu+AsqYY6kN+F6Cr4E0YnI3J92xR+QVfFngZJFEUQUcCRqHtUIUrLar0WEZFEIkg4HLjGz2Wl/FUkHmdlZfR4aMCk5lNXvB4CRD77M7IPyhqhNxd+/S8qq+FOMZFExJB0HHA+8QFLTj0fAU7g0Ulcic38+greXmAHeV0hSdrCr8pJFUF4QtlgBxxCzRIvsTjJn49V+TbBxYDq22NV+LSIki6I42MzObHbMxaUPZux89JuaQzkCjHzwBSEVfxGSRcUws1OAUySdYmbHFbQbWalSvOJPMZJFETJQJQs4hpKAczrKFKv2ayGVlyyKYookNSscacb/+X0e0zxGqeJvlBn54Cuo4i9CsqgYkl5tZj8FLtMEcjOWITETSETFX3HJoog2G8C3JR3GgBdwRBF0TkeZiGq/CMmiKK4Bvibp82n/kHSsUpk0Rj74AlYF9ux92jCzZyV1nf6NkCwqyVHA+xiTmmljeFAyaLwfTzj/Kz5TdS3w0UybEZJFEW023p1eB72AI4phaV0yLByDPxCOq/bLMWgBkkWBHIsHXIem/en4g3elMmmMfKuJprpvUccW0+ZQSBaNOgqQLBqm8nAVli6JYpjO6bCQZhOLVfspQLKoEsOw+P2STg2+eiQw0vr/PT3LhotrcygkElpVP209tn0HpeqnTUTFn2Iki4pL7ARVpw0NEed0lInwewVIFkVS2y1U+s3IBl/tij/gieYwqeKvZCL6oCLpLjN7Tc+xgWximRKCz8GXnublp9j8Uil9QYGCsCqoRdeyOfB5VJHndJSZLL+XdKuZbV3SZilGtWnvMPj9qDCyOV9RFX9DxkBX/fQw6BV/keXhEdVpw5BHVUvuYyju94qRLKqUZxj8fiQY2eBrSCv+SjNMVT8DXfEXvJQcUZ22hpntnD+0OAZleX4JJMLv2+oejWTRQQt9R6UfDLzfjwqjvOx4rpm9LyWK9mKZuUQlJYvCSO00DgF2TIemA+eZ2cA18pTr0fViZpZV8adCkkWRKECLruZRjS7D5PdRjPCyY/X7AWFkg69I5ILN/wRkSxZV4mhLFplZlmRRFOmLclt8ZiG7Oq3mUVUiGLaikFELvqrfDx4jH3xFVfzJW7E3kkVbAVmSRVEMS9VPxM1d0iy8p9mM5gYs6R4z2zRzuEUp+QUxLJW4lVhK+31EUUgko9Zuofr94FGDr8DKH0mb48HXzni3522BrpJFIQzLE2BQxd8PzWzb9t8uafagPQVKOg34AXBFnUGtlKC030u6u6coZMJjlUrFGdmE+xYRlT8RkkWjTkTFX4RkUQSH4KoEcyU9ydhSwUr9HValMo+IopAQaruFyiBQg6+Yyp8IyaJRJ+LmHiFZVBwzW7HfY6hUFkFxyaJAaruFSt+pwVeMzte6vYFXI1lkZj/JtD2qROjRPYEHXyfkD688E7VAaTMi7VAqQ4CZXZ9mj4tJFgVS2y1U+s7IB19m9ixwdtpKsXF7Jy1lbrmA/1tZBGm59i94lU5JPbrikkWFaYTPl8WLNu7GA8/NcA3K7fo0rkplHK3CpdlpfxVJB+UWLgUxU9Kmtd1CpZ8s1e8BDAKSPtt+zbBznKTHgM0kPZq2x4Df41Pdg8hX0uslfR3FQkgB8plm9lczm522Ek/VlwF3Ah/EZ9aabSAwszeb2ZuB3wCvNbOtzGxLYAvg1/0dXWXIKe33BzcV4wCpx+HBhWwXQdI9kmYD2wN3SLpP0uzW8Upl0hj5akcIqfwZZcmiECIq/iTNSsHMQCNpjpn1zqbOd6xS6Repj9RmPYVLswfpM1rbLVQGiRp8US74aiSLFpSrM2g5OsNU9ZNmEJfHpUuKVPxJ+gg+KzmQkkUNkr4KPA58OR3aH1jBzPbt36gqw0qE30v6FJ6H2S5c+qWZfaD7SCuVJZcafFE0+AqTLIpA0jWMVf3Mqxw0s9MX+KYliCjJotJIWpbxDWZvAs42syf7N6rKsBLh91WyqFJZPGrwxfA0Gi2NpHvNbJN+j2Nh1Iq/SqUsw+D3lcqSTk24D0DS4ZJWbu2vIumwfo5pAcyUNFBSOhNwetrOBH4EnAt8If18Zo5hSUtLOkLS19P2L0nGaOAoVRRSqRDk9/UzWqk8d0a+1UQiovJnXmBgZn+UdDAwEGXXPSKrU1PvrIEUWU3Vfki6Aq/4uyftbwJ8JNP82bhkUXNdDkzHBlGPrllyfFNfR1EZWibB7+tntFJ5jtTgC2iEVQsKrBaXLCrMMHbZ36Ddl8fM7pW0YabNCMmiSmVQGUa/r1SWSEY++Aqq+IuQLCrGkJZUz07i2u2Kv9zePEOjR1ep5DKkfl+pLJGMfPBFjM5XhGTRqDMVP5/T0v5N5KsSDJMeXaVSqVSWEGrwFaDzFSRZNNKktgqfSVs2UZJFlUqlUqksilrtWCt/hoaS5zRQsiiKgZeBqow89TNaqTxHRrbPV0/lz/pA0cqfUe0dFkmADFRxyaJKpVKpVBbFKC871sqfyiHAUcBcSUUkiyIYJhmoymhSP6OVyuIxssFXrfypmNmK/R7DcySiKKRSKUn9jFYqi8HIBl+V0WUIJYuKF4VUKoWpn9FKZTGowVdlFGkEhJcFtgLuxpccNwNuB7br07gWxExJm7abzFYqA0b9jFYqi0ENvuKolT/lKXJOgyWLijFMMlCV0aR+RiuVboxstWOlImmOmW28qGP9QtJaC/v3mrdY6Tf1M1qpdKMGXwHUyp/yRJxTSV8FHme8ZNEKZrZv95FWKpVKpbJw6rJjDLXypzwR5zRCsqhSqVQqlYVSZ74CkHSvmW3S73EsSdRzWqlUKpUlhSovFEOIZNGIU2WgKpVKpbJEUGe+ChItWTSKVBmoSqVSqSxp1JyvslTJovLUc1qpVCqVJYoafBWkllWXp57TSqVSqSxp1JyvSqVSqVQqlUmkBl+VSqVSqVQqk0gNviqjTpWBqlQqlcqkUqsdK5VKpVKpVCaRmnBfGVmqDFSlUqlU+kENviqjTJWBqlQqlcqkU5cdKyNLlSyqVCqVSj+oCfeVUabKQFUqlUpl0qkzX5WRo8pAVSqVSqWf1OCrMnJIWmth/1676lcqlUolkhp8VSqVSqVSqUwiNeerUqlUKpVKZRKpwVelUqlUKpXKJFKDr0qlUqlUKpVJpAZflUqlUqlUKpNIDb4qlUqlUqlUJpH/DxJJB9/k8dxJAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# maximum likelihood also works quite well..  \n",
    "\n",
    "gg = model.optimizing(data={'w1': 1., 'w2': .5, 'click': click, 'impres': impres, 'N': N, 'P': P, 'a': action+1, 'K': K, 'Psi': Psi, 'omega': omega})\n",
    "xnames = ['h=' + h+' a='+a for h,a in product(name_map.values(),name_map.values())]\n",
    "\n",
    "\n",
    "do_plot(s)\n",
    "plot(xnames,gg['ctr'].flatten(),'ro')\n",
    "\n",
    "ylim([0,0.04])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}