simulate_qutrit_readout.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import scipy.stats as st\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.utils import to_categorical"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Generate Simulation Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "class SingleShotTracesGenerator:\n",
    "    def __init__(self, model='inf_lifetime', snr=100, time=100, seed=0):\n",
    "        self.model = model\n",
    "        self.snr = snr\n",
    "        self.seed = seed\n",
    "        self.time = time\n",
    "        np.random.seed(seed)\n",
    "    \n",
    "    def generate(self, n, state, time=None, snr=None):\n",
    "        if time is None:\n",
    "            time = self.time\n",
    "        if snr is None:\n",
    "            snr = self.snr\n",
    "        model = SingleShotTracesGenerator.models[self.model]\n",
    "        return np.asarray([model(state, time, snr) for _ in range(n)])\n",
    "    \n",
    "    @staticmethod\n",
    "    def _wiener(mean, time, snr, init=0+0.j):\n",
    "    #def _wiener(mean, time, snr, init=np.array([0,0])):\n",
    "        process = [init]\n",
    "        for t in range(time - 1):\n",
    "            new = init + complex(np.random.normal(0., np.sqrt(1/snr)), np.random.normal(0., np.sqrt(1/snr)))\n",
    "            #new = init + np.array([np.random.normal(0., np.sqrt(1/snr)), np.random.normal(0., np.sqrt(1/snr))])\n",
    "            process.append(new)\n",
    "        return np.asarray(process)\n",
    "    \n",
    "    @staticmethod\n",
    "    def _inf_lifetime(state, time, snr):\n",
    "        if state == 0:\n",
    "            mean = 1. + 0.j\n",
    "            #mean = [1., 0.]\n",
    "        elif state == 1:\n",
    "            mean = 0 + 2.j\n",
    "            #mean = [0., 1.]\n",
    "        elif state == 2:\n",
    "            mean = -1. + 0.j\n",
    "            #mean = [-1., 0.]\n",
    "        return SingleShotTracesGenerator._wiener(mean, time, snr, init=mean)\n",
    "    \n",
    "    models = dict(inf_lifetime=_inf_lifetime.__func__)\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_traces = 500\n",
    "time = 100\n",
    "plot_mean = False\n",
    "\n",
    "sstg = SingleShotTracesGenerator(time=time)\n",
    "\n",
    "g = sstg.generate(n_traces, state=0)\n",
    "e = sstg.generate(n_traces, state=1)\n",
    "f = sstg.generate(n_traces, state=2)\n",
    "\n",
    "# if plot_mean:\n",
    "#     plt.plot(np.mean(g.real, axis=0), np.mean(g.imag, axis=0), color='b')\n",
    "#     plt.plot(np.mean(e.real, axis=0), np.mean(e.imag, axis=0), color='r')\n",
    "#     plt.plot(np.mean(f.real, axis=0), np.mean(f.imag, axis=0), color='g')\n",
    "# else:\n",
    "#     for i in range(n_traces):\n",
    "#         plt.plot(g[i].real, g[i].imag, color='b', alpha=0.01)\n",
    "#         plt.plot(e[i].real, e[i].imag, color='r', alpha=0.01)\n",
    "#         plt.plot(f[i].real, f[i].imag, color='g', alpha=0.01)\n",
    "        \n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Find Basis and Orthonormalize"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def normalize(v):\n",
    "        return v / np.sqrt(v.dot(v.conjugate()))\n",
    "\n",
    "def gs(A):\n",
    "    A[:, 0] = normalize(A[:, 0])\n",
    "\n",
    "    for i in range(1, A.shape[1]):\n",
    "        Ai = A[:, i]\n",
    "        for j in range(0, i):\n",
    "            Aj = A[:, j]\n",
    "            t = Ai.dot(Aj.conjugate())\n",
    "            Ai = Ai - t * Aj\n",
    "        A[:, i] = normalize(Ai)\n",
    "    return A"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2, 8000)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.vstack([np.mean(e, axis=0) - np.mean(g, axis=0), \n",
    "             np.mean(f, axis=0) - np.mean(g, axis=0)]).shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "w = gs(np.vstack([np.mean(e, axis=0) - np.mean(g, axis=0), \n",
    "             np.mean(f, axis=0) - np.mean(g, axis=0)]).transpose())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x28e8ecefeb8>]"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(1, figsize=[5,5])\n",
    "ax.plot(w[:,0].imag , marker='o')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "63.43494882292201"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.angle(1 + 2j, deg=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "116.56505117707799"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.angle(w[0,0], deg=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.20021576+0.40043152j, -0.01302901+0.25327451j],\n",
       "       [-0.19864795+0.39971228j, -0.22134655-0.1059649j ],\n",
       "       [-0.19826503+0.39924841j, -0.41047582-0.40874663j],\n",
       "       [-0.19885546+0.40063733j,  0.37030525-0.25609276j],\n",
       "       [-0.20073644+0.40159698j,  0.2696436 +0.5142743j ]])"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "w"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-1.        , -1.0278917 , -0.97919665, -2.        , -1.98915811,\n",
       "        -1.96758979],\n",
       "       [ 1.        ,  0.99299786,  0.97810687,  0.        ,  0.00933629,\n",
       "        -0.00838355]])"
      ]
     },
     "execution_count": 299,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.vstack([np.mean(e, axis=0) - np.mean(g, axis=0), \n",
    "             np.mean(f, axis=0) - np.mean(g, axis=0)]).transpose()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Wavenet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from WaveNetClassifier import WaveNetClassifier\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From c:\\users\\nathan\\onedrive\\github\\wavenet-classifier\\env\\lib\\site-packages\\tensorflow\\python\\framework\\op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "original_input (InputLayer)     (None, 8000, 2)      0                                            \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_1 (Conv1D)         (None, 8000, 10)     50          original_input[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_2_tanh (Conv1D)    (None, 8000, 10)     210         dilated_conv_1[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_2_sigm (Conv1D)    (None, 8000, 10)     210         dilated_conv_1[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "gated_activation_1 (Multiply)   (None, 8000, 10)     0           dilated_conv_2_tanh[0][0]        \n",
      "                                                                 dilated_conv_2_sigm[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "skip_1 (Conv1D)                 (None, 8000, 10)     110         gated_activation_1[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "residual_block_1 (Add)          (None, 8000, 10)     0           skip_1[0][0]                     \n",
      "                                                                 dilated_conv_1[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_4_tanh (Conv1D)    (None, 8000, 10)     210         residual_block_1[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_4_sigm (Conv1D)    (None, 8000, 10)     210         residual_block_1[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "gated_activation_2 (Multiply)   (None, 8000, 10)     0           dilated_conv_4_tanh[0][0]        \n",
      "                                                                 dilated_conv_4_sigm[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "skip_2 (Conv1D)                 (None, 8000, 10)     110         gated_activation_2[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "residual_block_2 (Add)          (None, 8000, 10)     0           skip_2[0][0]                     \n",
      "                                                                 residual_block_1[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_8_tanh (Conv1D)    (None, 8000, 10)     210         residual_block_2[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_8_sigm (Conv1D)    (None, 8000, 10)     210         residual_block_2[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "gated_activation_3 (Multiply)   (None, 8000, 10)     0           dilated_conv_8_tanh[0][0]        \n",
      "                                                                 dilated_conv_8_sigm[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "skip_3 (Conv1D)                 (None, 8000, 10)     110         gated_activation_3[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "residual_block_3 (Add)          (None, 8000, 10)     0           skip_3[0][0]                     \n",
      "                                                                 residual_block_2[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_16_tanh (Conv1D)   (None, 8000, 10)     210         residual_block_3[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_16_sigm (Conv1D)   (None, 8000, 10)     210         residual_block_3[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "gated_activation_4 (Multiply)   (None, 8000, 10)     0           dilated_conv_16_tanh[0][0]       \n",
      "                                                                 dilated_conv_16_sigm[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "skip_4 (Conv1D)                 (None, 8000, 10)     110         gated_activation_4[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "residual_block_4 (Add)          (None, 8000, 10)     0           skip_4[0][0]                     \n",
      "                                                                 residual_block_3[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_32_tanh (Conv1D)   (None, 8000, 10)     210         residual_block_4[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_32_sigm (Conv1D)   (None, 8000, 10)     210         residual_block_4[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "gated_activation_5 (Multiply)   (None, 8000, 10)     0           dilated_conv_32_tanh[0][0]       \n",
      "                                                                 dilated_conv_32_sigm[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "skip_5 (Conv1D)                 (None, 8000, 10)     110         gated_activation_5[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "residual_block_5 (Add)          (None, 8000, 10)     0           skip_5[0][0]                     \n",
      "                                                                 residual_block_4[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_64_tanh (Conv1D)   (None, 8000, 10)     210         residual_block_5[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_64_sigm (Conv1D)   (None, 8000, 10)     210         residual_block_5[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "gated_activation_6 (Multiply)   (None, 8000, 10)     0           dilated_conv_64_tanh[0][0]       \n",
      "                                                                 dilated_conv_64_sigm[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "skip_6 (Conv1D)                 (None, 8000, 10)     110         gated_activation_6[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "residual_block_6 (Add)          (None, 8000, 10)     0           skip_6[0][0]                     \n",
      "                                                                 residual_block_5[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_128_tanh (Conv1D)  (None, 8000, 10)     210         residual_block_6[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_128_sigm (Conv1D)  (None, 8000, 10)     210         residual_block_6[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "gated_activation_7 (Multiply)   (None, 8000, 10)     0           dilated_conv_128_tanh[0][0]      \n",
      "                                                                 dilated_conv_128_sigm[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "skip_7 (Conv1D)                 (None, 8000, 10)     110         gated_activation_7[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "residual_block_7 (Add)          (None, 8000, 10)     0           skip_7[0][0]                     \n",
      "                                                                 residual_block_6[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_256_tanh (Conv1D)  (None, 8000, 10)     210         residual_block_7[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_256_sigm (Conv1D)  (None, 8000, 10)     210         residual_block_7[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "gated_activation_8 (Multiply)   (None, 8000, 10)     0           dilated_conv_256_tanh[0][0]      \n",
      "                                                                 dilated_conv_256_sigm[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "skip_8 (Conv1D)                 (None, 8000, 10)     110         gated_activation_8[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "residual_block_8 (Add)          (None, 8000, 10)     0           skip_8[0][0]                     \n",
      "                                                                 residual_block_7[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_512_tanh (Conv1D)  (None, 8000, 10)     210         residual_block_8[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dilated_conv_512_sigm (Conv1D)  (None, 8000, 10)     210         residual_block_8[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "gated_activation_9 (Multiply)   (None, 8000, 10)     0           dilated_conv_512_tanh[0][0]      \n",
      "                                                                 dilated_conv_512_sigm[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "skip_9 (Conv1D)                 (None, 8000, 10)     110         gated_activation_9[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "skip_connections (Add)          (None, 8000, 10)     0           skip_1[0][0]                     \n",
      "                                                                 skip_2[0][0]                     \n",
      "                                                                 skip_3[0][0]                     \n",
      "                                                                 skip_4[0][0]                     \n",
      "                                                                 skip_5[0][0]                     \n",
      "                                                                 skip_6[0][0]                     \n",
      "                                                                 skip_7[0][0]                     \n",
      "                                                                 skip_8[0][0]                     \n",
      "                                                                 skip_9[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "activation_1 (Activation)       (None, 8000, 10)     0           skip_connections[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv_5ms (Conv1D)               (None, 8000, 10)     8010        activation_1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "downsample_to_200Hz (AveragePoo (None, 100, 10)      0           conv_5ms[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv_500ms (Conv1D)             (None, 100, 10)      10010       downsample_to_200Hz[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv_500ms_target_shape (Conv1D (None, 100, 3)       3003        conv_500ms[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "downsample_to_2Hz (AveragePooli (None, 1, 3)         0           conv_500ms_target_shape[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "final_conv (Conv1D)             (None, 1, 3)         12          downsample_to_2Hz[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "final_pooling (AveragePooling1D (None, 1, 3)         0           final_conv[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "reshape_1 (Reshape)             (None, 3)            0           final_pooling[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_2 (Activation)       (None, 3)            0           reshape_1[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 25,855\n",
      "Trainable params: 25,855\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "wnc = WaveNetClassifier((8000, 2), (3,), \n",
    "                        kernel_size=2, \n",
    "                        dilation_depth=9, \n",
    "                        n_filters=10, \n",
    "                        task='classification')\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1500, 8000, 2)\n",
      "(1500,)\n"
     ]
    }
   ],
   "source": [
    "dataset = np.array([])\n",
    "truths = [np.zeros(g.shape[0], dtype=np.int), np.ones(e.shape[0], dtype=np.int), np.ones(f.shape[0], dtype=np.int)*2]\n",
    "for state, truth in zip([g,e,f], truths):\n",
    "    temp = np.moveaxis(np.array([state.real,state.imag]), 0,-1)\n",
    "    if dataset.shape ==(0,):\n",
    "        dataset = temp\n",
    "    else:\n",
    "        dataset = np.vstack([dataset, temp])\n",
    "dataset_truths = np.hstack(truths)\n",
    "print(dataset.shape)\n",
    "print(dataset_truths.shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1125, 8000, 2) (1125, 3)\n"
     ]
    }
   ],
   "source": [
    "X_train, X_val, y_train, y_val = train_test_split(dataset, to_categorical(dataset_truths))\n",
    "#X_train, X_val, y_train , y_val = train_test_split(np.random.rand(15, 96000),to_categorical(np.random.randint(10, size=15)))\n",
    "print(X_train.shape, y_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From c:\\users\\nathan\\onedrive\\github\\wavenet-classifier\\env\\lib\\site-packages\\tensorflow\\python\\ops\\math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "Train on 1125 samples, validate on 375 samples\n",
      "Epoch 1/1\n",
      "1125/1125 [==============================] - 147s 130ms/step - loss: 0.2410 - acc: 0.9307 - val_loss: 1.1921e-07 - val_acc: 1.0000\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 0.00000, saving model to ./saved_wavenet_clasifier.h5\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x28ef8f0e048>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "wnc.fit(X_train, y_train, validation_data = (X_val, y_val), epochs=1, batch_size=32, optimizer='adam', save=True, save_dir='./')\n",
    "#y_pred = wnc.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.utils import plot_model\n",
    "plot_model(wnc.get_model(), to_file='model.png', show_shapes=True, show_layer_names=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "os.environ[\"PATH\"] += os.pathsep + 'C:/Program Files (x86)/Graphviz2.38/bin/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.0b5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}