muller-gan.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/je714/miniconda3/envs/dl/lib/python3.6/site-packages/sklearn/cross_validation.py:41: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.\n",
      "  \"This module will be removed in 0.20.\", DeprecationWarning)\n",
      "/Users/je714/miniconda3/envs/dl/lib/python3.6/site-packages/sklearn/grid_search.py:42: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. This module will be removed in 0.20.\n",
      "  DeprecationWarning)\n",
      "/Users/je714/miniconda3/envs/dl/lib/python3.6/site-packages/statsmodels/compat/pandas.py:56: FutureWarning: The pandas.core.datetools module is deprecated and will be removed in a future version. Please use the pandas.tseries module instead.\n",
      "  from pandas.core import datetools\n",
      "/Users/je714/miniconda3/envs/dl/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from msmbuilder.example_datasets import MullerPotential\n",
    "from msmbuilder.preprocessing import MinMaxScaler\n",
    "%matplotlib inline\n",
    "from __future__ import print_function, division\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "import keras\n",
    "import keras.backend as K\n",
    "from keras.layers import Input, Dense, Activation, LeakyReLU, BatchNormalization\n",
    "from keras.models import Sequential\n",
    "from keras.optimizers import Adam\n",
    "from sklearn.model_selection import train_test_split\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "from utils import *\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This dataset consists of 10 trajectories simulated with Brownian dynamics\n",
    "    on the Muller potential, a two-dimensional, three-well potential energy\n",
    "    surface."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "trajs = MullerPotential().get().trajectories"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We scale the trajectories to range -1 to 1. We can go back to the original space with `sc.inverse_transform`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MinMaxScaler(copy=True, feature_range=(-1, 1))"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sc = MinMaxScaler(feature_range=(-1 , 1))\n",
    "sc.fit(trajs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "txx = np.concatenate(trajs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "a = sc.fit_transform([txx])[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12eaee5f8>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX8AAAD8CAYAAACfF6SlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJztnX2UXHV98D/fmZ0Ns4nNZkmssEkg1JRWIBDYCpac1hIKIhJiqoug9RXyULEUbQPxlELk8ZEAp2J5HlsMyEEKIiuGEEQPVNDa0BMlMW+iUhE0yYbWhGRXIUN2dub3/HHnzt65c99m5s7793POZmfv3Ln3d3PvfH/f3/dVjDEoiqIo3UWi2QNQFEVRGo8Kf0VRlC5Ehb+iKEoXosJfURSlC1HhryiK0oWo8FcURelCVPgriqJ0ISr8FUVRuhAV/oqiKF1IT7MH4Mfs2bPN8ccf3+xhKIqitBVbt249YIyZE7Zfywr/448/ni1btjR7GIqiKG2FiPwqyn5q9lEURelCVPgriqJ0ISr8FUVRuhAV/oqiKF2ICn9FUZQuRIW/oihKF1Kz8BeReSLyXRH5qYg8JyJ/47GPiMgdIvKCiOwUkdNrPa+iKIpSPXFo/pPA3xpj/hA4C7hKRN7i2ucCYGHhZyXwLzGcV1Fal50jcPvJsKbf+r1zpNkjUpQSahb+xpiXjTE/Krz+LfBTYNC128XAfcZiM9AvIsfUem5FaUl2jsBjV8P4HsBYvx+7uvoJQCcSpQ7EavMXkeOBxcAPXG8NAnscf++lfIJARFaKyBYR2bJ///44h9aa6Je6/Yhyz566CbKZ0m3ZDDxyZeX3OO6JRFEKxFbeQURmAN8ArjHG/Mb9tsdHTNkGY9YB6wCGhobK3u8o7C+1LSTsLzXAouHmjUux+OanYOu9YHIgSTjjwzD/rPJ7tv4K+PZ1cMEtU/dtfK/3MU3O+vzuzfDzJ639Zs6FpTf433O/ieSpm/Q5UWpCjKldxopICvgm8IQx5vMe738J+J4x5sHC388DbzfGvOx3zKGhIdPRtX1uP7mgzbmYOQ8++ePGj6dT2TliCefMwcIGAYz1/2wL3Z0jljAd3wvpWZAZB3Llx0pNh+xr/ueSBJzxkYJg97i3UztSqvt4jMlmTT8eepL1mTVjAedQuhUR2WqMGQrbr2bNX0QE+DLwUy/BX2Aj8AkR+RpwJjAeJPi7Aj/t0L3dKZjCtMR2oVHXtHMEHr0KchOOjQVBOr4H1q+0NHenMC5OEh4ECX4Ak4ctX4ZELyR7Xect2dH7b6/V38y5PkrC3OCxKEoIcdj8zwb+EjhHRLYXft4pIleKyJWFfb4FvAi8ANwFfDyG87Y3fl9e5/ZOtPc28pqeuilAAMOUEI7ZwpifAGOslUCl2CYdm6U3QCpduk8qbW1XlBqoWfM3xmzC26bv3McAV9V6ro5i6Q2l9mMo/1J3or03zmsKW0H4ra4aQT5bMBMdxtPEE4Rz3Pb1dNrqT2k6LVvPv+OJ8qWOahpqJ8KuyW2jTw+UOlNtvBzm66+wTDkYy0kbt0ZfKZ5moghjcq8KFw2rsFdiR4V/Mwn7UneivTfomnaOwIaPW1qzTeagZbeHqf+rnSNW2KTxcMrawtXzvTYgkVKTjtIQtLZPK9OJ9t6ga3rqplLBb5ObsN7bOQK3LLA0/HYV7mFMe4Nq+UpDUOHfaCpJ7Fo0DBfdYYUAItbvi+5ob+GwaBhOvaxglsH6fepl1vYgc9b4HmsFEBSN0wlkDjV7BEqXEEucfz3oyDh/t50aLK233QU6+Dtfi9v3WILe5PCNc5eEFS7ZzWieh1IjDYvzVyqgE6N3wD9befdm2PHVqe1FU41PnHu3C/5kb3ub9JS2Qs0+tVBpbZ5OjN4B/0lt673l2xVv0gNw8RfbWwlQ2grV/Kulmto8rRy9U0vWbVAtGyUYScKNHe7HUFoS1fyrJciE40erRu/UmnXbCpNXu6ITpNIkVPhXSzUmnFaN3qlmInOy9AbLXq1Ujh31pCgNRs0+1VKtCacVszV9J7I9lvYfZbz5yXjH1C1I0vIZadkGpcGo5l8t9TThNLrJS9CEFWb+sU1G3R6pUy35CTqmaJ/SVqjwr5Z6mXCaUcnTayKzCTP/eJmMlOqwu31pZzelAajZpxbcJhxbY6+l+mIzcgHs466/wvt9p1nIHRUU2LREqRjbAayd3ZQ6o5p/XARp7JWYccLs77WO0W8ci4atWHMvbLOQ1zUq9aMSp7uiVIhq/nHhp7F/+zqYzETPBwjSpmvRBMPyEnaOwJHfln/OmXWqJp74SE23CtZ5FbJz0u4JgErLopp/XPh9STMHKwujrMX+HoTf5PTIlVaz8keu9BZEvTPCG5MrldMzDZb/85TPyC/kU3MolDqhwj8uKv2SOpuXOE0xYDmOwz5XKUFZuFu+7J9s5KwymZ5V3bmVcuzqpJ/8sdWI/d13tmYCoNKxqPCPC7/QzyA7up+fAAoaoc/nqqGWz9l19Du9nHKjcUZxtWoCoNKxaEnnOPGqjwP+ZZztUsduZs7z7vGLwNBH4V2fj2ccSvPREs5KzEQt6azCvxE4hbFtOskcIrCf65pxyxa/5Z7S/cLq/wf1DICA9odKcxDL7KMoMRFV+KvZpxEsGra0u6GPWkI/c5BAwW87/37+ZPl+1SRdZTNWDP8jV8LxS7QOTyuhfhSlSWioZ6PYOVKuxftRTPSponhc0HsmBy/9e/j5FUXpeGLR/EXkHhH5tYh4Gi9F5O0iMi4i2ws/3RPCYEfzrL+CSIIfppy9fk7aIOethga2CDL10s/pD9qzV2kacZl97gXeEbLPfxhjTiv8dEfaYkk0T0Sc4X1hxeO8MnaD8gSUxpBKw4p1lt9mzThc91L80VuKUiOxCH9jzPcBjQN0U3FGrMCpl005c4PC/4LCRC+6Q+vEN42kt0O+VRv5KF1LI23+bxORHcA+4O+MMc818NwWtbQqrIaKE7JMwcnrwK/+f1A5ietegvUrKzy3EgtJx6Trft5Ovcy6v416/hQlgEYJ/x8BxxljXhWRdwIbgIXunURkJbASYP78+fGOoJqeu7VSTdXLqBNGUDmJNTMrO6cSH7kJawKG8udtx1c1cUtpGRoS6mmM+Y0x5tXC628BKRGZ7bHfOmPMkDFmaM6cOfEOotZWhdXgt9RfcVftNmC1FbcumYPNed4UpQIaIvxF5E0iIoXXby2c95VGnLtINWGTtWAv+bOZKfu702Zfqw1YbcWtTaOfN0WpkFjMPiLyIPB2YLaI7AVuBFIAxpg7gfcAfyUik0AGeJ9pdGpxtT13q8FtYjK5KcHudOZC5T4Ipx1ZEto+sWkIvqG76QHond64501RqqB7yjsElT2I2wZ7+8neX/z0gOWMrRava1AaT3oALrjFer3h46WlsBMpq1QzNO55UxQHUcs7dE+Gb7WadjUEOWN3jkQ/pztaZOI1FfytgHsCD3qmGhldpigV0D2afyPx0/whehVH1fJbk1pXb4pSZ7SwWzMJcsYGOfycGbuPXKmCvxU58tvaeykrSgugwr8eRGmG7sadsatll1uTfNaamJ0lNRSlDVHhXy8uuKWyUM5vX6eafrtgcpSU1NAJQGlDVPjXiyht+YpmnpnaIrFd0cQtpU3pnmifZuBXlwfUodtJaOKW0oao5t8s1MzTmgRVQ/V7TxO3lDZENf9msHNEzTytxNDH4F2fL9/uzrNYeJ5VnM2duKWlNpQ2RDX/ZqA24tZhxV3+gt/dL2HHV62yzEF+HEVpE1TzbwZqI24dbMEdJZs6m7Hq8UdJ0lOUFkc1/1rxaqUYhtqIWwsvLd/PLKcTt9IhqPCvBb9WimETwNIbrAJgSmtQSbtNnbiVDkGFfy0EtVIMYtGwVfkxNb1+Y1PCsbOwo2rz6txVOgi1+UfBr/dvlOqdzs+mZxXePzT1WmkeJ73b+u3X66F3ujVBaFVOpQNR4R9GUO/foB69dkSP87NOO7KGejafHV+F+WdZQv3Rq6z+u04mJ1TgKx2Lmn3CCOrFGla9sxJbstJ47Pu4aBh6Z5S/n89qWK7SsajwDyOoF2tY9U6NDGl97HuUORT8vqJ0GCr8w/CL7rC3B1Xv1MiQ1se+R2H3WVE6DBX+YSy9Ibg0s7t6Z3oAetKwfqWVKKS0Ls77GHafFaXDUOEfRpTSzIuGrazPFetgMlNw5hp16rYSqbRVw8fvPka5z4rSQWgP30rxC/vcOWJ1eNIOXK3JirtUkCtdQdQevhrq6cZPuNvvucM+H70KHrsGsmriaVlmzqtd8Ac9F4rShqjwdxIU079o2Dt0MzdRHh+utA5x2O3DngtFaUNisfmLyD0i8msR8Sx3KBZ3iMgLIrJTRE6P47yxExTTDxr2126kB4Lt9lGL8oU9F4rShsSl+d8L/D/gPp/3LwAWFn7OBP6l8Lu18I3p32MJB1rTP9JVpNLREufcNv6wxixB2nzQc7Gmv+5moA3bRrntiefZN5bh2P40q84/keWLB+v+WaWziUXzN8Z8HwgKbbkYuM9YbAb6ReSYOM4dK74x3eJfxkFpHHYEzsx54fu5Bb+7+uqWe6Jr84Gx/hVUcw1gw7ZRzl77NAtWP87Za59mw7bR4vZPr9/F6FgGA4yOZfjkQ9u5fsOuSMd0f/bT63cVj610N40K9RwEnNJzb2Fb/amk3r5XrDeCavwtQHrACqddNOxznwp42fg9y2z43FMvLT/ofDY1mIGCBPxtTzxPJlsaQWaA+zfvZvFNT5ZNFk68PpvJ5rjtieerGqfSWTTK4Sse28q+fSKyElgJMH/+/NrPWqmjzt7mNA+oxt8aHPntVKXUkvu0x2qsbnKWxu9lfqnEV+Ol5bufC5+JIz++lxNWP178e1ZfihsvOgmgaHo5KpUgk82XfC7tsc0W8EEcOpwFpiaLax7azqDDtLNvzNs85rdd6S5ii/MXkeOBbxpjTvZ470vA94wxDxb+fh54uzHmZb/jxRLnf/vJ3sJ75rzorfj8jqHUh2Sv1ejGK3S2kvvmxOce5g0knGpJKh0tscvneHvzs1kycUfl44uZdCrJzStO4bYnnmfUQ9AP9qd5ZvU5TRiZ0giixvk3yuyzEfhgIernLGA8SPDHRlBRtiCcpqKJ17TrViPpnQHZw97vVRtt5WG2OWx6+dfcuezNzyZvhMPpY8IFv/1cjO/BvZg9bHq5dbI1wj5t086q808knUqWvJdOJVl1/olNGpnSSsRi9hGRB4G3A7NFZC9wI5ACMMbcCXwLeCfwAnAY+Egc5w3Fz2wT5MBzm4oyBy1tND1gVX6UhGbx1pPMoeruWxAOs01+fC/78kdz6+QwG/NLuLGwy2A6zarciaz5zJOMZSxzim22Wb54EHaOMPnoX9OTe73wCYPBMs/sy88uHq9V2DeWKUb1aLSP4kVnl3dwC3KwNMBTL4OfP+mdreln5kkPWJ2d1ARUX2y7vdd9i6HWzoLVj/u671MJIZsvfTeVFG57z6mc9+RS+jLli9VWMfW4UdNO96LlHcDbgRsU323/7UXmoBZqqzd2pI7XfYspjv7Y/rSnHTwp5YIfIJsz3PbE8yx7/b+9jyevBJ5vWWIT1/aMcKwcYJ9pzApBQE07Siidrfl7EaTZT2a081az8IvUiZkN20ZZ9fAOsrmp5z6VlJK/vdjUezVzEwfKtgdp/ssSm1ibups+mSr/cdj0sjp7ed0ngF+uvbCux1daF9X8/Qhquq40h2qjeIJwZfQ++3t/zRXbFhTt+U7CBD/ArZPDnoI8yMl7bc9Iyf4AfTLBtT0jbJyon/Af7A/JSVAUukX4f/NTsPVeddS2KjHXTHp245c4+Uf/QJojhePv4aSt1/Mn2cvZSHVCd2N+CWQpmHBeYZ85OtSEc6yUrxSs7cGmolpQk48Slc4X/t/8FGz5crNH0fmkB+DIOOQDJlhJgMmXb68kiiektPKGbaP80dZbScuRko9Vq3FXYrN373vIzOBoebVsv33m6IrGEBUB3n/WfI3mUSLR+cJ/670V7KylHKoikbJ6GX/7ugDzmcAZHyl1tkNoyWVnYbIPzfgh15s7p8ItPTK2b3vief6DeDRut81+rhxgbepuyFI2AXjtO2F6OGKSTJOpCbGe+QC3X3KaCn4lMp3fxrEiU48K/qrIZ60uZoF+EwPv+nxFrRLdNW8un7jfEWdfwFVTZ99Yhn1mtvcwEV6cdhmbeq9mWWJT6GUF2eyj7Nsrk6TI8Up+Bnkj7M3Prquztyj4K6lnVcm+SkfRuZq/bR6oCNX8qyZskrUrcbrDOO175JwACj6ai02OdyUSPNBzDlvzv8+gjw3dGb11bH+aW39T7pw1BnrEMjkFafA2yxKbfM/ntYLws+8nBN7A61yT/auSc8UdApqUQsZxJfWs4mxSo53O2o7ODPX0Su5SmkeyFy7+4lSv40evKu1+5nzfw0djDORIFIV3OQIr1sGiYa7fsIv7N+8uEa4CiEdpQWeYpnN/g7Uk9voMwEEzg8PmKI6VA+RJkCQfMr7yc9UjBHSwP82/ycc9k9E8I6riqH0F/smUMSTlKZUTNdSzM4V/UDE2ScLRC+HAz6ofnBKK/VhN9PYz7aLbpoTALQu8zUOp6dA3UH0GtSR5dvHNfPDZ40rKGPvF54NV2G2fmc2gLfB9hL0T+7q8JgZj/CeMvBFOOPJA4JjCcgairBRenPZ+EuLznZ45r1QzX78S75WuwJox72M42TkS7OepRwivEkp3x/n7hg4KvPvOwkOv1JMcCR7IncP52R28af1KS0iAv6DIvgbjHpU8o2JynLT1elabP2Fp7/aikPQ1FWEZ+WwhHEHuW/sF7Bj0Xh5hWWITG/NLfE1EzrE6hf0hM4M3yOv0yiRQbrZy7psnQcJPoNsT6/gea/XlR1j0VZjQt9G2py1Nd2n+tt1Z6/PEg11H34cgTbheuM/ZjDH4YUf/zJAjnmPKG7gm+3GAMrOQF3sLBeW8/Bulx6/AlxVmrqnEpKqaf1NotZLOjcWr85IdUhimjUgy+P22p3ZJmDG9PHv6rd4x+84zxSR0DZAz0Q7mPqfIlKmm2fTKJG9IeAt+sMxO1/aMeEYOeTEoB/h86s6yfe1rzhv4b+YQWfBLMtxO79kVzYOQEF6l+XSm8F807B9SmJ4V/FmTJw4B2bpUJwlzRorhitdlL+eanyysvsRyBRgDLzKPyRoe1VbR/KMwKAcCTVVODPg6mUUgT4I3mv3RT27y3oK/GA46M9qq2Z5EQMNIW5jOtPlDabs/m50jMFGecVmCLdDUNFQkZ4RPukIVZSwDl3mUXo4ZETjOjAZG0nQSlaxUwhzUFf+feU3mlUbOJVJw+gfhsWtKu7HVEkaq1IXO1Pz9eOqm0hBDLxaeV1iutpG62AT6+1Kc/a3Z/M1rH6l7ZkSS7hD8Nk1bqYzvsaKxnBp6VDOPzXF/DD+6z7sNZw1N7pX46S7hHyX64Ef3FV60iKG4BUiKYW3q7mJWbCopvPr6JKNjGR7NL6n7f1U7mW3ansxBKxLIngAqjdh56d+tjG8/NAKoZehcs4+NM/MwSgvGfBbWX9GYsbURfTLB51N3Ml16OJydZJUj5tyg66SOIjdhfQeeuglSfd5afLU0wE+kRKOzhb/bXqklnWuiR/J8ztzBZI+QKiQSzZUDLRNNo8RM7H4vqU8EkJaWqIrONvtUaq9sZYY+NpWn0EQ9W4Si4HduU5RQhj4av1C2FbzxPYCZcixrZFEonS38O8W+OHOeVRHzkz+GFXeFh6sqSquRHrB+xx366aXgqWM5Ep1t9pk5twNCNsWKQIKWLljXSpm0SguSOVhasG98j1VmZfdmmH9W9WYbPwWvUxS/OtLZmr9Xpq+T0GzeCqRZ7/To+1aEsRqg2HbNFhT8MCX41fyvRMdYE8KGj1dvtvFzIKtjOZTOFv52pq+95CxB4Pgl/pODJKxIh0gILHqf9RnfXWr4r7aXsSHaTCs4XlX5VyrGHRrqNNuENZsJKuWiBBKL8BeRd4jI8yLygois9nj/wyKyX0S2F34uj+O8kVg0DNe9BAv+1PWGsWKSJeE9OZh8BSFuBe3cq9ZNKm3Z6X+nxvZ69pI4eBQtMQEoSs2M743mzA0q5aIEUnNVTxFJAv8F/DmwF3gWuNQY8xPHPh8Ghowxn4h63JqqetoUQ8BC7P6pNJCoPp7Zr7qlJK0S0ouGLc2lFqNI8Rz+FRqNgVfNNN6QOOL5vqK0DUEVeLVaaCCNrOr5VuAFY8yLxpgJ4GvAxTEctzZKtIYQspnqBX+y1z9/wFkoq1YbZPEc/hOICGQlxaRXBcxazE6K0khSaSvIwe+7q87cWIhDIgwCzru0t7DNzV+IyE4ReVhE5nm8j4isFJEtIrJl//4KqhF60SjnaO8MR/y9C1vg7xyBiRizJAOYxavk3Lc1kQotv6wozaWgsMycB6deZplR/VBnbizEIfy9fHxu9fQx4HhjzCLgO8BXvA5kjFlnjBkyxgzNmTOntlE1SjvIHAp2OtkrkLCuR0Vqc5nmSDBNXCuRfLbiPgXqOlAai7Ge0fG9sPVef8Ut2avO3JiIQ/jvBZyq71xgn3MHY8wrxhjbEH0XcEYM5w2mUdpBetbUKsMWsE6nU8UrkBrEbirtX8bX5ILDXl1o1I7ScEwOMMFlWHpnqDM3JuIQ/s8CC0VkgYj0Au8DNjp3EJFjHH8uA34aw3mDCYvxj8rMeYWsWq9w0YTVH8C2TdoC1pmk0qgVSHoAegKuNz3giIpAfQBKe5I51OwRdAw1SwBjzCTwCeAJLKE+Yox5TkRuEpFlhd2uFpHnRGQHcDXw4VrPG0pJCFgE0gP+pptFw3DBLR6mk3x5fwB3anmjViCTmXDT0qLhqUlRfQBKO1LN9yksV6BL6cwG7m78Gro7WXGX9dsvzfyWBZXZ7deMWS93jtS/RHRII/WSMUX5v4gBLfegxE6yFxb/Jfz8yeilILxKooQ1qW9zooZ6dnZtH5vQpu2JqQfB74GILPgp1U4WDcO3rwv+vC28Z86rXDCn0tF8CsX2lPU3Q7WoPqG0O7ks/OheyBcUnfE9VuMZ8P/eBhV+61DhH5XuMPyGLRXP+Eh85/JKLT/p3fi6UFNpKxFszbiVuOLpW/DBbcePMqYGmKFEVOtX6oGZEvw2uQlLufJDC7/50h2a/9Ibgk0v7/p8+DHSA97ae+906z2/ZejOkULMso86XIsWct1LU6/Lqn0WMoFnzisd09IbrGqKGsypdAqZg/CZWZYfSxJW4EP2sPV9TM/y/t5qrkCXCP8g00tUh/AFt8CGK0s1j0QS3vWFYMEdJdTTqYVEjWZwOp/t80cpi7toGLbdb9U1UpROwQ5gcNbk8jOhauE3oFuEP1jC28vxU9FDkAByrr9DiLK8dGohUXsQuB28i4ajrR52jsDeH4bvpyidSHrAkgVdbu+HbrH5Q+3V/566qbz0bD4b3jEobHnpnoCi5idEXbG4+fZ1viuRWg1BxlTu7FXnsNJQeqer4C/QPZo/RNeOvajWcbT0huj2eHuMMGXCSc+CzBjgiMtPpKpbtu4cCYw6isNHe4QkR1EeduoX+qmOYaWh2KWiteF7lwn/WvAzx4Rp9pXY452fsd/fOWKFszmTyaqVmBH6mtYSny8C00yOrJGSJu+HTS9HMaElI5Tm09tXGvBgt5Ncf4W3MtbBdEeSVzW4tYOF51lRO/VKFvHTRvySsqqpaR6hp0DOCONMp5/X2GeOZs60LNOy4xWdxhg4xIziMW6dHObanhHmJg5UNl5FiZNkb3lGvicCQx+NFgXYgmiSVxT8BK47K3B8jyX4T72ssuzCSsbhPt9jVxdexxin7Bf25iAphqPMBNdk/4qN+SVcmtjM58wdFa0GRCg5BgCTsDZ1N30S/OWzdRE1BymxE0nwQ7G3MLTtBBCF7tX8g9K+/bp/VdtBKMzGGKTdQzxj8TIfBbA3P5slE3cA8KNpKxmQV6Ofy+MYAMsSm/in1D+rYFfaA0nCjRVk9rcIjezk1Z4EpX3HqW1H6UMadL64GlQ/dZOn4Peb+4+VV4qv12Q/yIQpXySGRfc4jwGwMb+EUTM72ngVpdmYXEcXgute4R8kcP2cuNVkBQZNMmHHnTk3vgbVFU5c+8zRxdcb80v4u+xKXjXTSoS9CEwivhOA8xg2t04O1y28s0UXsUo7M74HHvlfUxNAB1UI7V7hHyRw49K2IdoqIux8i4YtE8+Kddbf61eGP3juhzQ9y3O3Q8xgwpSWqj5iktw6WTq5bMwvYcy8ocxkkxLDIWZw2PSWbM8bGJQDbOq9mmWJTSXHuS93bl0EtZqTlLpg8lMTQNgqvo3oXuEfJHDj0rYh2ioiyvkqefC89p141coPcHDY9PLN3NtwR/hL4e9liU1s6r2aF6ddxqbeqzlWvKN1+nmN1dnL2ZufTd5Ygj9RKO42N3GAtam7SyaAGyc/yt9kP17YXzhoZnDEuHslTDFphLyZMjPlVcNXGo3JW+GgXqv4DVc1Z0w10r0OX2hMske19cTdY5t4zb82kdvx6+NAPmhmMCFpfpcDTKR+h9cm8szit54a8yv5GaRloiQ6xxbqbpyO3U29V3uGdLqdv26WJTZxbc8Ix8orjDEdY2CWTIWKFqOGCvtGiRxSlIax4E/hQxsd39s9paXaF55Xn0hBDzTUMwq1ZPxWQo+j5n6U2iJeoZ9+eJmVfExN/bzGCa+v4z29/8n/zq5jIEB4DsirZZNCQsongMOmt8RE5Lc6cDt/3WzML2HjxJLAfZz7koVre0YYlANq7lGaz0v/Xv69NY6+A3boqP23HcrdxISy7jX7NAL7YXBq7JMRGq9U0vTdy6zkY2qyHbDX8DXSVKc1CxTNNXvzs1mdvbxEK9/nE83j5fythY35JSyZuKOqekQtuthV2p31K6N/b91BH01AhX89iRLp40XUyBw/J7SHP8Opoftp58598z7FGHIkWDJxBycceYAlE3eUCH6wonnczl/36qAa3P4H24fgN9kE+QVE1G+g1IMKH6omN5RR4V9Pqs0X8HMSpweiOaEdDuQ85Rq6n8A0huK+4vMgJ50F5jzYmF/icP56rw4qxbbxz00cIOFyIt/x/7faAAAYuklEQVQ6OezpLM6T4JX8DF8t317BVFOJVFFiockNZbrb5l9vqi0G51UJNJX29BVs2DbKbU88z76xDMf2p1l1/oksXzxY9Gds3DbKp9fvIuNoQnPr5HCZw/Sw6S0R0teaEeZ6rBAMljAOEuaV2O+jcG3PSJlzt08muLZnhFsnh+nxmJB6JE/GHMWoOcrzOkaN5YDe2ruSoxOVZy8rSk20QEMZ1fzrSbX5AhFDTTcUBPvoWAYDjI5l+PT6XWzYNlrcZ/niQW5ecQqD/emiISeKdn7r5LCnaSQhljBuJP5OZGsFkBRv1f1YeSXQDLUsscm3bIWuBpS6EldByBro7lDPRlDHcNKz1z7N6Fi5g2mwP80zq88p2WavELz29+PFaZd5hnbmjXDCkQcqHm+1+IWPTpoEPeJvhrLDS51hpPvM0TyVP42lie0aKaQ0jzWVVcqthIaGeorIO4B/ApLA3caYta73pwH3AWcArwCXGGN+Gce5W546hpPu8xHk7u32CiGTLW+yEnh8M9vTZBJ35E4QyxKb6JPXy/oMHDa9gRFLTiez0wylOQJKS3D7yU3vHVCz2UdEksAXgQuAtwCXishbXLt9DDhkjHkzcDtwS63nVeDYfu92jwkRFqx+nLPXPl3U+CsV/FC/yJ2o2ILamXNgjJWAtjp7uW+RuEmT8HUye/kPFKXhuDP0m1AzKA6b/1uBF4wxLxpjJoCvARe79rkY+Erh9cPAUhFdcNfKqvNPJJ0qj3TJGVPiA6jE1OOkHpE7leAlqEUgw1FszC/xnZw+lb3Sd4xhYa4tagVVOhE77LtJNYPiMPsMAs6Qlr3AmX77GGMmRWQcOBrQ1k41sHzxIEAx2ichQs4lvarR+J3EHblTCWHZws5MX6c9/9qeEb4g/8w+M7usNISfKcvGSDy9jBUlEuN7g/OB6mgWikPz9/quuPWnKPsgIitFZIuIbNm/f38MQ+t8li8e5JnV5/DS2gvJd5jaGiVb2M70PeHIA9w6Ocx7k9/3zAew8VotOEk4/o2doY/V57hK+yIJ//ItdU4Ci+Mp3wvMc/w9F9jnt4+I9AAzgbIqZcaYdcaYIWPM0Jw5c2IYWudz/YZd/N6nv8Xxqx+vqtRBK1OpzyEoH8DGNmVNGp9Hf+Y8WPElK6EuTpK9VkvA3unxHldpb0zAyrzOSWBxCP9ngYUiskBEeoH3ARtd+2wEPlR4/R7gadOqMaZtxPUbdnH/5t1lpp5OoVKfQ9SichvzS/hU9sryFYCzpPd1L8GacZ49/VbPLmYV0zvD+v2uL1jantKdpAcAsSp+BtGAJLCan+qCDf8TwBNYoZ73GGOeE5GbgC3GmI3Al4F/FZEXsDT+99V6XgUe/EFAtc8OoRKfQ9TQVKHcX/B635vou6DcxnrNTxZyRnZl7b2HM4es34uGefaXhzhj67WeORRKB2OXX985YvUGCNqvAWGgscT5G2O+BXzLte0Gx+vXgffGcS5lik7V+KvFr2yF20z0/rPm892f7eexsSVs7fvzYkmMDdtGuW3t0yWlMvaNZRhliW+5i8g4lvDX/GQhK3Pn8sHkdzTJrJtYesNUZI8fXv056oTW9mlTnCUcFAuv6B93tM+svhRDxw3wzR0vF8NhP/PYc2z51UEe+uEesoWaFqNjGVZ9fQfpVILD2bznxEKylyOJNKmJcQT/NpI56eGX/Wczfc2beaPZz0NmNn3J11XwdxO2r+eRK/3t/IlUQ+v9aHmHNqTajN1uJ5UULvmjeSVCvhLcZSK+mLiMFafPZehH13qHsxmre9o382fx3uT3SyYOd8ay0gWk0sH1/tMDlq+pRrSTVwdTbcZuNyMC2Zzh/s27qz6Gl//hqi1X+/pvDcIZE+vY1Hu1Z7Ka0kVIMrzRi+0XahAadtCG+NX0Ufyp1wI3KGPYdjSHZRUrHU4qHRzSadPg+v4q/NsQv5o+SuMJ6iRmO5r99lE6HUc59pnzgndtQn1/Ff5tiFdNn3QqSTrVvbez0VaUz/TcwwvTPsCgHChbVeQN/Gvu3KKjOSyr2IkxkGtNN5xSCekBWDNmRe4sGvbu7WE/tUFd+eqI2vzbEHdNn2P70/zZH8zhoWc7P+7fi8HC9X9j62hDfCGf6bmnLEzTngBGPeoJ2VFIa1L3MYtXA+39IjBuZmDyFBvNxOEfUAdzA0n2Wl33nNiCvU69PapBo306BL/GLp2MCNw+fFpxMtywbZRrHtpe9/O+MO0Dnk1kJk2CNx+5P/CzUxFDB3zDQ53NcvzO5cT+CvsJ98Oml6/n/oSLkpuZRXwTSkeSnAa5I8H7pAfg9XFvO74k4d13Nleoa7RPd1BNh65OwRi45qHtfH3Lbn75SqZhjnC/JvZhze2hNGJoU+/VoRnJfse0Bf4hZrAm+0Gu7fFOQnP2Nrhx8qOB53Ufu2smiFS63OyycwS+fR1kDpbva2v1Xn22W6A9Y1S610jcATh7+HYzz/ziYLGPcSPI+Xxt/Lb7EaVwnV/DmlEzmwVHvsrpR9ZV3NsgzAcxamZzX+5czx7OHUd6wFtg2/WdVtzl3Us7Yp/tVkY1/zZG4/2bwwO5czxt/g/kzvH/kAdRMpKjlqyIciz3vl4+CPvYG/NL2Jr/fa7tGWGwsErw8nG8xlGkyDJNciXvOffNmymHfB5BCtN0YG0ju6pq5hCkZ029njkXBk4g/8v/QPJ58ljNfabzOjkSJCWPzJwHC8+Dnz9p2dfTs2DiVcg5cy0Ehj5qVVoNIqgNax1btDYCtfm3MQtavIyz4NG0oUP4TM89vD/5NEny5EjwQO6colklbtyZxX5CvZ7HDtrP/d5T+dNYmtgeeMxliU38Q/ph5uR+bdnJTa6igmZ+Pq7B/jTPrPaYhHeOtJSztZ5Etfmr8G9jWtnJmxTh0jOrL6WgdDa9SSGVTPDahLVi6Esl6O1JMp7JFovq2Y58L/wUHwFeWnthfQbdJkQV/mrzb2P8evi2AjljeGDzbhX8iicTOVMU/ACHs3nGMtmS3tNBxQv9Eh01ATI6KvzbmOWLB7l5xSkkWzQsQ8W+Ui2ZbI7bnnje932/RMdV559Y76F1DOrwbXOWLx5ky68OehYsSyUgGx59qCgtyb6xTDGU2dljYfniQc9ExzBTkVKK2vzbmG6O8Vc6n/50iiOT+ZKItnQqyc0rTlEhH4Da/DscjfFXOpl0KsnEZK4slDnMHKRER80+bYrG+CudSkIIfLa1pHk8qObfpjTqC9CarmSlkwkLENOInnhQ4d+mNOIL0MlJWkr7ohE98aDCv02pJcZ/Vl+Kvgi1/w00NIw0nUowqy/VsPMp7UmYs3fDtlHOXvs0C1Y/ztlrnw7MF+hmVPi3KXaMf3+6MmGZTiW5cNExkZOvcsY0zPRzVCpJX6+6oRR/wp53ZyBE1ISxbkWFfxuzfPEg06eFC8tZfSkEq+7JzStO4bs/2082Yruowf50w0w/Y4ez6sxTAsnm8oEavVcghEYIeVOTmiUiA8BDwPHAL4FhY0xZC3oRyQG7Cn/uNsYsq+W8yhRhwjIhsO2G80q2fTJiwxM7Y/Izjz3HocNZz32SIuRiyhWx/Rgavqr4YZeEGB3LsOrhHUCpGcjv+6BKRTm1rrFXA08ZY9aKyOrC39d57JcxxpxW47kUD47tTwcKy8vOnF/RZ5Ii5I0pZkwCvPr6pOe+qYRw23tPBWDV13fUVMfHmZrfiG5cSvuTzRmueWh7yfPip4xohFA5tZp9Lga+Unj9FWB5jcdTKiQs8uGzy0/x/EwqWW7JTwj84/CpvLT2Qp5ZfQ7LFw9y2xPP+wr1bN5w2xPPs3zxIDOOCtcjEoKnk3pWX6qYtbl88SDTe1uzWJ3S+ngJfq35402twv93jTEvAxR+v9Fnv6NEZIuIbBYR3wlCRFYW9tuyf//+GofWHSxfPOjrBBv00XaWLx7kkj+aV7bdK7InbLlsvz/mYxZykjdw84pTGOxPF30QX7jkNLbdcF5JH96JSS1IpMRDfzql5SB8CFXXROQ7wJs83vr7Cs4z3xizT0ROAJ4WkV3GmF+4dzLGrAPWgVXbp4LjdzVrlp3Ep9fvKquBEqTtfPdn5ZOrU5O3CTMr2cvpsP3AEvbOolye17LxOS0DrcTG9Gk9Kvh9CNX8jTHnGmNO9vh5FPgfETkGoPD71z7H2Ff4/SLwPWBxbFegFMM+nRp1mLYT1TEWlE/gnGDC8g5SCQldem/YNspYJnwFoShRGR3LaMy/D7U6fDcCHwLWFn4/6t5BRGYBh40xR0RkNnA2cGuN51VchGnUbvw0dbdjzFk6d3QsU3SoDbpK6Nq//3Zkh6fddcZR4RqYhuMp9cB+zkfHMkUHsfv57UZqFf5rgRER+RiwG3gvgIgMAVcaYy4H/hD4kojksVYaa40xP6nxvEoAfjXQndv7+1KkElJiYvEzFQVNLO5z+YV9RvEJRA3Hqya8dLA/zWtHJnVloQBTE8HXt+zmgSve1uzhNIWahL8x5hVgqcf2LcDlhdf/CZSHnCh1wc5wtO3/dobjll8d5BtbR4vbDx3OkkoK/elU5L6pUc7lVw8oSqjdzHQqVDinEsIlb53n2bzGD+ek5vaNKN3NM784yPUbdnlGxXU6mkvfYfhlOD74gz1l2nI2Z5g+rYftN5YmgbnxW0l4nctQXhAuaqhdlDJC2bzhuz/bzwfOmh95ArAzPJ9ZfQ6geQRKKfdv3s3QcQNdZwJS4d9h+JlO/MwkYaYWv5VE0GcNlpml0vZ6UUxD9nltTS3qBGCPdfniQV+/hNK9XPPQdrb86iBDxw10TWtIre3TYfiZV/yqc4aZY4Jqpfh9drA/zTOrz+H2S6yk7k8+tN030sJZgTERsYKofV6vcFU/EiLF8196ZnmOg6Lcv3k3q76+o2uKwqnw7zC8Qi7TqSSXnjnPc3uYOSYoJNTvXKvOPzFSdUX3PlG18bHDEyxY/XhFNYByxhTP/9nlp3D27w1E/qzSPbhzTDLZHGs2Ptek0dQXNft0GM7QTPfStZolbVBIqPtc/X0pjLE0/YRHRI69YnB+zsv5GtZExi7u5cesvhS/yUwGnv+BK95W9GUETSIJCe8spXQ2Y5ks12/YxeM7Xy4WOEynEhyVSjJ2uLpgiVZATIvaPoeGhsyWLVuaPYyu5/oNu3hg8+4yB647icztG/BDgJfWXgjAgtWPx14u2h7bJx/a7nls5/mdbNg2WlK9tD+dYs2yk0rKTgRVN1W6G6/vRLMQka3GmKGw/VTzV3zZsG2Ub2wdLRGiAvzFGeVx/1Ebyjv9BFFKQlTCoCsSKUoSm01Ykpx9TBX+ihfuVW07oMJf8cUvlNPL0RolQcvtY1h1/omxxd3bTuagY1da3dEd4qp9BpQg2q1ngAp/xZdKGmP4CUd3fwCnZuTlnzg8MemrXfcHJIG5xxTk+4hCJQlsigLt1zNAhb/iS9T6P+CvaYfZQd3mFi/fgfM4b/mHb3M4W17yud+j8Xul9Y6c+K16FMWPdusZoKGeii9BoZxuqqks6kXYcXp7vCuHxh23UM0S3v1/1ajG90pr0E72flDNXwmgUtNJLZp21OOM+5h9/LZXw4Zto56hqkHYzmb7/ypKnSJFaSYq/JVA4hLocVGJKaoabLOTl+D3s/kLFCdFu3qqXQJD6Q5EKCYwtkt5CDX7KG1FJaaoavALWU2K8P6z5nuadt5/1vySL3jUsFelczAGVj28o63KQ6jwV9qKuHwLfvjZ+vPG8Nnlp5Scuz+dor8vxQObd5fULmq3kD8lHrI541keolWbFKnZR2k76mmKCjMruU07XtVONSdAcdKqyoBq/oriIKpZKaja6arzT/SN9OlLJSL1LQiiN6lxRO1Eq8b/q/BXFAdRzUpBCXDLFw/y/rPml00A6VSSz61YxEs3X8gHzppf9Rj7entI6Te3LYjTHxU3avZRFBdRzEph5qHPLj8lsIpqpc1onIxnstx+yWnakawNaJVib15oVU9FqYKwTOSonL326Yr9A3bJjJ4EeCQ7Kw0klYDJvHcIsLveVKOIWtVTF4+KUgVxRR15+RjCyBmDQQV/K5D1EfwAf/YHcxo6lkpRs4+iVEkcUUfuLOpKM4uV1qWSNqPNQIW/ojQZ5ySyYPXjTR6NEhetGuJpU5PZR0TeKyLPiUheRHxtTCLyDhF5XkReEJHVtZxTUToZv7BAO6lMaR9aNcTTplab/4+BFcD3/XYQkSTwReAC4C3ApSLylhrPqygdx/UbdrFvvFxbtEtIrFl2UsX+AaU+DPanmeVRRtymlUM8bWoy+xhjfgogwVkrbwVeMMa8WNj3a8DFwE9qObeidBLXb9jlGfbZl0rwuRWLSnwL2ku4eczqS7HthvMA/77Vs/pS3HjRSS0b4mnTCJv/ILDH8fde4MwGnFdR2oYHf7DHc/uRSVPW/cwuL7Fm43NaNrqBCHDjRScV/661W1yzCRX+IvId4E0eb/29MebRCOfwWhZ4hjOIyEpgJcD8+dVnQCpKu+EX4eO33Z4E/PIEBmusL/SBs+bz4A/2dGTk0ay+FBcuOoYHNu/2LdHt3u5VvRVar+R5JYTa/I0x5xpjTvb4iSL4wdL05zn+ngvs8znXOmPMkDFmaM6c1o6RVZQ4SfqYTv222wTVIhoMcDgO9qd9Hcj96RSfXX4K/zh8qm+NonSb1pewzTafXX6KbwmO2y85jS9cclpJDsftl5xWzMruFBph9nkWWCgiC4BR4H3AZQ04r6K0DZeeOc/T5n/pmfM89p4izPQQlIXsl6W8ZtlJxWNv+dXBMg3ZPgbA3z+yi9cm6tO7QMSqk58s5D4M9qc5/ug0m188VLIisd/3a7Zjk0pKidkmrARHu2r0UampvIOIvBv4v8AcYAzYbow5X0SOBe42xryzsN87gS8ASeAeY8z/CTu2lndQuo3rN+wqmlqSIlx65ryatc0N20YDbdJh79dyDOf2mekUIjB2OOu5z+hYpkTIV2M7d59vYjLH4UIadLs4YeMgankHre2jKIrSQWhtH0VRFMUXFf6KoihdiAp/RVGULkSFv6IoSheiwl9RFKULUeGvKIrShbRsqKeI7Ad+VeNhZgMHYhhOK9MN1wjdcZ3dcI3QHdfZzGs8zhgTWiKhZYV/HIjIlijxru1MN1wjdMd1dsM1QndcZztco5p9FEVRuhAV/oqiKF1Ipwv/dc0eQAPohmuE7rjObrhG6I7rbPlr7Gibv6IoiuJNp2v+iqIoigcdJfxF5L0i8pyI5EXE19MuIr8UkV0isl1E2qp0aAXX+A4ReV5EXhCR1Y0cYxyIyICI/JuI/Lzwe5bPfrnCfdwuIhsbPc5qCLs3IjJNRB4qvP8DETm+8aOsnQjX+WER2e+4f5c3Y5y1ICL3iMivReTHPu+LiNxR+D/YKSKnN3qMfnSU8Ad+DKwAvh9h3z8zxpzW6uFYHoReo4gkgS8CFwBvAS4Vkbc0ZnixsRp4yhizEHiq8LcXmcJ9PM0Ys6xxw6uOiPfmY8AhY8ybgduBWxo7ytqp4Bl8yHH/7m7oIOPhXuAdAe9fACws/KwE/qUBY4pERwl/Y8xPjTHPN3sc9STiNb4VeMEY86IxZgL4GnBx/UcXKxcDXym8/gqwvIljiZMo98Z57Q8DS0VC+jm2Hp3wDIZijPk+cDBgl4uB+4zFZqBfRI5pzOiC6SjhXwEGeFJEthaaxncag8Aex997C9vaid81xrwMUPj9Rp/9jhKRLSKyWUTaYYKIcm+K+xhjJoFx4OiGjC4+oj6Df1EwhzwsIsE9K9uTlv0uNqKHb6yIyHeAN3m89fcVNJU/2xizT0TeCPybiPysMIO3BDFco5eW2HJhXUHXWcFh5hfu5QnA0yKyyxjzi3hGWBei3Ju2uH8hRLmGx4AHjTFHRORKrNXOOXUfWWNp2XvZdsLfGHNuDMfYV/j9axF5BGuJ2jLCP4Zr3As4tai5wL4ajxk7QdcpIv8jIscYY14uLJN/7XMM+16+KCLfAxYDrSz8o9wbe5+9ItIDzCTYtNCKhF6nMeYVx5930Ya+jQi07Hex68w+IjJdRN5gvwbOw3KidhLPAgtFZIGI9ALvA9oiEsbBRuBDhdcfAspWPCIyS0SmFV7PBs4GftKwEVZHlHvjvPb3AE+b9kvICb1Ol+17GfDTBo6vUWwEPliI+jkLGLfNmU3HGNMxP8C7sWbaI8D/AE8Uth8LfKvw+gRgR+HnOSxTStPHHuc1Fv5+J/BfWFpwW11jYfxHY0X5/Lzwe6CwfQi4u/D6j4FdhXu5C/hYs8cd8drK7g1wE7Cs8Poo4OvAC8APgROaPeY6XefNhe/gDuC7wB80e8xVXOODwMtAtvC9/BhwJXBl4X3Binr6ReEZHWr2mO0fzfBVFEXpQrrO7KMoiqKo8FcURelKVPgriqJ0ISr8FUVRuhAV/oqiKF2ICn9FUZQuRIW/oihKF6LCX1EUpQv5//o5iZjkbvn7AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12eae5c18>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = scatter(a)\n",
    "scatter(txx, ax=ax)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "train, test = train_test_split(a)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(75000, 2)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25000, 2)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# hyperparameters\n",
    "sample_size     = 100     # latent sample size (i.e., 100 random numbers)\n",
    "g_hidden_size   = 128\n",
    "d_hidden_size   = 128\n",
    "leaky_alpha     = 0.01\n",
    "g_learning_rate = 0.0001  # learning rate for the generator\n",
    "d_learning_rate = 0.001   # learning rate for the discriminator\n",
    "epochs          = 100\n",
    "batch_size      = 64      # train batch size\n",
    "eval_size       = 16      # evaluate size\n",
    "smooth          = 0.1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create a GAN, a generator and a discriminator\n",
    "# comes from utils.py file\n",
    "gan, generator, discriminator = make_2dtraj_GAN(\n",
    "    sample_size, \n",
    "    g_hidden_size, \n",
    "    d_hidden_size, \n",
    "    leaky_alpha, \n",
    "    g_learning_rate,\n",
    "    d_learning_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_3 (Dense)              (None, 128)               384       \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_2 (LeakyReLU)    (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 1)                 129       \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 1)                 0         \n",
      "=================================================================\n",
      "Total params: 513\n",
      "Trainable params: 513\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "discriminator.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train loop"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/je714/miniconda3/envs/dl/lib/python3.6/site-packages/keras/engine/training.py:973: UserWarning: Discrepancy between trainable weights and collected trainable weights, did you set `model.trainable` without calling `model.compile` after ?\n",
      "  'Discrepancy between trainable weights and collected trainable'\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch:   1/100 Discriminator Loss: 0.9941 Generator Loss: 1.1476\n",
      "Epoch:   2/100 Discriminator Loss: 0.9933 Generator Loss: 1.1991\n",
      "Epoch:   3/100 Discriminator Loss: 1.2541 Generator Loss: 0.8986\n",
      "Epoch:   4/100 Discriminator Loss: 1.2476 Generator Loss: 0.8771\n",
      "Epoch:   5/100 Discriminator Loss: 1.3937 Generator Loss: 0.8026\n",
      "Epoch:   6/100 Discriminator Loss: 1.3151 Generator Loss: 0.8253\n",
      "Epoch:   7/100 Discriminator Loss: 1.3521 Generator Loss: 0.8122\n",
      "Epoch:   8/100 Discriminator Loss: 1.2572 Generator Loss: 0.8885\n",
      "Epoch:   9/100 Discriminator Loss: 1.4035 Generator Loss: 0.7815\n",
      "Epoch:  10/100 Discriminator Loss: 1.3382 Generator Loss: 0.7781\n",
      "Epoch:  11/100 Discriminator Loss: 1.2382 Generator Loss: 0.8809\n",
      "Epoch:  12/100 Discriminator Loss: 1.3815 Generator Loss: 0.8317\n",
      "Epoch:  13/100 Discriminator Loss: 1.2778 Generator Loss: 0.8039\n",
      "Epoch:  14/100 Discriminator Loss: 1.3257 Generator Loss: 0.8603\n",
      "Epoch:  15/100 Discriminator Loss: 1.3076 Generator Loss: 0.8346\n",
      "Epoch:  16/100 Discriminator Loss: 1.4450 Generator Loss: 0.8322\n",
      "Epoch:  17/100 Discriminator Loss: 1.2919 Generator Loss: 0.8799\n",
      "Epoch:  18/100 Discriminator Loss: 1.2403 Generator Loss: 0.8316\n",
      "Epoch:  19/100 Discriminator Loss: 1.4177 Generator Loss: 0.7953\n",
      "Epoch:  20/100 Discriminator Loss: 1.2685 Generator Loss: 0.8186\n",
      "Epoch:  21/100 Discriminator Loss: 1.2484 Generator Loss: 0.8778\n",
      "Epoch:  22/100 Discriminator Loss: 1.2431 Generator Loss: 0.8258\n",
      "Epoch:  23/100 Discriminator Loss: 1.1482 Generator Loss: 0.8704\n",
      "Epoch:  24/100 Discriminator Loss: 1.2847 Generator Loss: 0.7583\n",
      "Epoch:  25/100 Discriminator Loss: 1.5298 Generator Loss: 0.7477\n",
      "Epoch:  26/100 Discriminator Loss: 1.4003 Generator Loss: 0.7830\n",
      "Epoch:  27/100 Discriminator Loss: 1.3448 Generator Loss: 0.8989\n",
      "Epoch:  28/100 Discriminator Loss: 1.4007 Generator Loss: 0.8588\n",
      "Epoch:  29/100 Discriminator Loss: 1.2959 Generator Loss: 0.8391\n",
      "Epoch:  30/100 Discriminator Loss: 1.4135 Generator Loss: 0.8014\n",
      "Epoch:  31/100 Discriminator Loss: 1.2975 Generator Loss: 0.8620\n",
      "Epoch:  32/100 Discriminator Loss: 1.2854 Generator Loss: 0.8437\n",
      "Epoch:  33/100 Discriminator Loss: 1.3213 Generator Loss: 0.8411\n",
      "Epoch:  34/100 Discriminator Loss: 1.3551 Generator Loss: 0.8341\n",
      "Epoch:  35/100 Discriminator Loss: 1.2330 Generator Loss: 0.8954\n",
      "Epoch:  36/100 Discriminator Loss: 1.3467 Generator Loss: 0.7931\n",
      "Epoch:  37/100 Discriminator Loss: 1.3177 Generator Loss: 0.8746\n",
      "Epoch:  38/100 Discriminator Loss: 1.1145 Generator Loss: 0.8576\n",
      "Epoch:  39/100 Discriminator Loss: 1.2941 Generator Loss: 0.8554\n",
      "Epoch:  40/100 Discriminator Loss: 1.3436 Generator Loss: 0.8472\n",
      "Epoch:  41/100 Discriminator Loss: 1.1702 Generator Loss: 0.8294\n",
      "Epoch:  42/100 Discriminator Loss: 1.2269 Generator Loss: 0.8350\n",
      "Epoch:  43/100 Discriminator Loss: 1.2756 Generator Loss: 0.8398\n",
      "Epoch:  44/100 Discriminator Loss: 1.2349 Generator Loss: 0.8454\n",
      "Epoch:  45/100 Discriminator Loss: 1.3035 Generator Loss: 0.8694\n",
      "Epoch:  46/100 Discriminator Loss: 1.2793 Generator Loss: 0.8500\n",
      "Epoch:  47/100 Discriminator Loss: 1.3373 Generator Loss: 0.8255\n",
      "Epoch:  48/100 Discriminator Loss: 1.3432 Generator Loss: 0.8315\n",
      "Epoch:  49/100 Discriminator Loss: 1.3036 Generator Loss: 0.7900\n",
      "Epoch:  50/100 Discriminator Loss: 1.3724 Generator Loss: 0.8249\n",
      "Epoch:  51/100 Discriminator Loss: 1.3944 Generator Loss: 0.7690\n",
      "Epoch:  52/100 Discriminator Loss: 1.3019 Generator Loss: 0.8773\n",
      "Epoch:  53/100 Discriminator Loss: 1.2957 Generator Loss: 0.8244\n",
      "Epoch:  54/100 Discriminator Loss: 1.3953 Generator Loss: 0.8037\n",
      "Epoch:  55/100 Discriminator Loss: 1.4014 Generator Loss: 0.7446\n",
      "Epoch:  56/100 Discriminator Loss: 1.2558 Generator Loss: 0.7929\n",
      "Epoch:  57/100 Discriminator Loss: 1.2248 Generator Loss: 0.8762\n",
      "Epoch:  58/100 Discriminator Loss: 1.1879 Generator Loss: 0.8732\n",
      "Epoch:  59/100 Discriminator Loss: 1.3231 Generator Loss: 0.8211\n",
      "Epoch:  60/100 Discriminator Loss: 1.3951 Generator Loss: 0.8020\n",
      "Epoch:  61/100 Discriminator Loss: 1.3091 Generator Loss: 0.8023\n",
      "Epoch:  62/100 Discriminator Loss: 1.4474 Generator Loss: 0.7927\n",
      "Epoch:  63/100 Discriminator Loss: 1.4520 Generator Loss: 0.7861\n",
      "Epoch:  64/100 Discriminator Loss: 1.2476 Generator Loss: 0.8613\n",
      "Epoch:  65/100 Discriminator Loss: 1.3415 Generator Loss: 0.8513\n",
      "Epoch:  66/100 Discriminator Loss: 1.3982 Generator Loss: 0.8553\n",
      "Epoch:  67/100 Discriminator Loss: 1.3027 Generator Loss: 0.8193\n",
      "Epoch:  68/100 Discriminator Loss: 1.3997 Generator Loss: 0.7575\n",
      "Epoch:  69/100 Discriminator Loss: 1.3279 Generator Loss: 0.8024\n",
      "Epoch:  70/100 Discriminator Loss: 1.3396 Generator Loss: 0.7996\n",
      "Epoch:  71/100 Discriminator Loss: 1.2934 Generator Loss: 0.8144\n",
      "Epoch:  72/100 Discriminator Loss: 1.3203 Generator Loss: 0.7958\n",
      "Epoch:  73/100 Discriminator Loss: 1.2850 Generator Loss: 0.8462\n",
      "Epoch:  74/100 Discriminator Loss: 1.3577 Generator Loss: 0.8360\n",
      "Epoch:  75/100 Discriminator Loss: 1.2975 Generator Loss: 0.8954\n",
      "Epoch:  76/100 Discriminator Loss: 1.2402 Generator Loss: 0.8378\n",
      "Epoch:  77/100 Discriminator Loss: 1.3793 Generator Loss: 0.7960\n",
      "Epoch:  78/100 Discriminator Loss: 1.3490 Generator Loss: 0.7887\n",
      "Epoch:  79/100 Discriminator Loss: 1.2960 Generator Loss: 0.8209\n",
      "Epoch:  80/100 Discriminator Loss: 1.3519 Generator Loss: 0.8393\n",
      "Epoch:  81/100 Discriminator Loss: 1.3857 Generator Loss: 0.8341\n",
      "Epoch:  82/100 Discriminator Loss: 1.3025 Generator Loss: 0.7642\n",
      "Epoch:  83/100 Discriminator Loss: 1.3909 Generator Loss: 0.8400\n",
      "Epoch:  84/100 Discriminator Loss: 1.2971 Generator Loss: 0.8413\n",
      "Epoch:  85/100 Discriminator Loss: 1.3403 Generator Loss: 0.8335\n",
      "Epoch:  86/100 Discriminator Loss: 1.3882 Generator Loss: 0.8042\n",
      "Epoch:  87/100 Discriminator Loss: 1.3438 Generator Loss: 0.7635\n",
      "Epoch:  88/100 Discriminator Loss: 1.3831 Generator Loss: 0.7605\n",
      "Epoch:  89/100 Discriminator Loss: 1.3728 Generator Loss: 0.7634\n",
      "Epoch:  90/100 Discriminator Loss: 1.3349 Generator Loss: 0.8152\n",
      "Epoch:  91/100 Discriminator Loss: 1.4277 Generator Loss: 0.7292\n",
      "Epoch:  92/100 Discriminator Loss: 1.4118 Generator Loss: 0.7543\n",
      "Epoch:  93/100 Discriminator Loss: 1.3866 Generator Loss: 0.7611\n",
      "Epoch:  94/100 Discriminator Loss: 1.3793 Generator Loss: 0.7588\n",
      "Epoch:  95/100 Discriminator Loss: 1.3925 Generator Loss: 0.7447\n",
      "Epoch:  96/100 Discriminator Loss: 1.4150 Generator Loss: 0.7515\n",
      "Epoch:  97/100 Discriminator Loss: 1.3816 Generator Loss: 0.7696\n",
      "Epoch:  98/100 Discriminator Loss: 1.3857 Generator Loss: 0.7540\n",
      "Epoch:  99/100 Discriminator Loss: 1.3985 Generator Loss: 0.7393\n",
      "Epoch: 100/100 Discriminator Loss: 1.3756 Generator Loss: 0.7482\n"
     ]
    }
   ],
   "source": [
    "losses = []\n",
    "for e in range(epochs):\n",
    "    for i in range(len(train)//batch_size):\n",
    "        # real x y coords\n",
    "        X_batch_real = train[i*batch_size:(i+1)*batch_size]\n",
    "        \n",
    "\n",
    "        # latent samples and the generated x y coords\n",
    "        latent_samples = make_latent_samples(batch_size, sample_size)\n",
    "        X_batch_fake = generator.predict_on_batch(latent_samples)\n",
    "        \n",
    "        # train the discriminator to detect real and fake images\n",
    "        make_trainable(discriminator, True)\n",
    "        discriminator.train_on_batch(X_batch_real, y_train_real * (1 - smooth))\n",
    "        discriminator.train_on_batch(X_batch_fake, y_train_fake)\n",
    "        \n",
    "        # train the generator via GAN\n",
    "        make_trainable(discriminator, False)\n",
    "        gan.train_on_batch(latent_samples, y_train_real)\n",
    "        \n",
    "        \n",
    "    # evaluate\n",
    "    X_eval_real = test[np.random.choice(len(test), eval_size, replace=False)]\n",
    "    \n",
    "    latent_samples = make_latent_samples(eval_size, sample_size)\n",
    "    X_eval_fake = generator.predict_on_batch(latent_samples)\n",
    "\n",
    "    d_loss  = discriminator.test_on_batch(X_eval_real, y_eval_real)\n",
    "    d_loss += discriminator.test_on_batch(X_eval_fake, y_eval_fake)\n",
    "    g_loss  = gan.test_on_batch(latent_samples, y_eval_real) # we want the fake to be realistic!\n",
    "    \n",
    "    losses.append((d_loss, g_loss))\n",
    "    \n",
    "    print(\"Epoch: {:>3}/{} Discriminator Loss: {:>6.4f} Generator Loss: {:>6.4f}\".format(\n",
    "        e+1, epochs, d_loss, g_loss))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x136f5d320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "losses = np.array(losses)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "plt.plot(losses.T[0], label='Discriminator')\n",
    "plt.plot(losses.T[1], label='Generator')\n",
    "plt.title(\"Training Losses\")\n",
    "plt.legend()\n",
    "plt.show()\n",
    "fig.savefig('ayy.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "latent_samples = make_latent_samples(50000, sample_size)\n",
    "generated_coords = generator.predict(latent_samples)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1310d1550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "f, ax = plt.subplots()\n",
    "ax = scatter(txx, scatter_kws={'label':'real'}, ax=ax)\n",
    "scatter(sc.inverse_transform(generated_coords), ax, {'label':'fake'})\n",
    "ax.legend()\n",
    "f.savefig('first_test.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The GAN recovers quite well the simple 2D surface"
   ]
  }
 ],
 "metadata": {
  "hide_input": false,
  "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}