{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T09:39:07.625154Z",
     "start_time": "2018-12-24T09:39:07.214436Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The autoreload extension is already loaded. To reload it, use:\n",
      "  %reload_ext autoreload\n",
      "Dataset has been loaded\n",
      "x-train (180000, 16)\n",
      "x-test  (20000, 16)\n",
      "y-train (180000, 128)\n",
      "y-test  (20000, 128)\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import h5py\n",
    "from sklearn.model_selection import train_test_split\n",
    "#import jtplot submodule from jupyterthemes\n",
    "from jupyterthemes import jtplot\n",
    "#currently installed theme will be used to\n",
    "#set plot style if no arguments provided\n",
    "jtplot.style()\n",
    "\n",
    "#now load this dataset \n",
    "h5f = h5py.File('./datasets/s16_d_siti_2.h5','r')\n",
    "X = h5f['sizes'][:]\n",
    "Y = h5f['spectrum'][:]\n",
    "\n",
    "Y = Y / 100.0\n",
    "X = X*2 - 1\n",
    "\n",
    "\n",
    "#get the ranges of the loaded data\n",
    "num_layers = X.shape[1]\n",
    "num_lpoints = Y.shape[1]\n",
    "\n",
    "\n",
    "#this information is not given in the dataset\n",
    "lam_min = 400\n",
    "lam_max = 800\n",
    "lams = np.linspace(lam_min, lam_max, num_lpoints, endpoint=True)\n",
    "\n",
    "#create a train - test split of the dataset\n",
    "x_train, x_test, y_train, y_test = train_test_split(X, Y, test_size=0.1, random_state=42)\n",
    "\n",
    "print(\"Dataset has been loaded\")\n",
    "print(\"x-train\", x_train.shape)\n",
    "print(\"x-test \", x_test.shape)\n",
    "print(\"y-train\", y_train.shape)\n",
    "print(\"y-test \", y_test.shape)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T09:39:28.273798Z",
     "start_time": "2018-12-24T09:39:28.138281Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.93503673, -0.63129639, -0.32914863, -0.39642248,  0.6034896 ,\n",
       "       -0.42785669,  0.40668602,  0.12134943,  0.16664638, -0.08531775,\n",
       "       -0.55238342,  0.36449522,  0.65747872, -0.57561007,  0.04765327,\n",
       "        0.96556356])"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 460.8x403.2 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lam_inv = np.linspace(1/400.0, 1/800.0, num=num_lpoints, endpoint=True)\n",
    "lams = 1.0/lam_inv\n",
    "\n",
    "plt.plot( lams, y_train[ np.random.randint(0, 9000)    ]  )\n",
    "plt.ylim(  [0, 1])\n",
    "\n",
    "x_train[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T11:29:50.954548Z",
     "start_time": "2018-12-24T11:29:50.363676Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "first (Dense)                (None, 128)               2176      \n",
      "_________________________________________________________________\n",
      "activation_35 (Activation)   (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "H0 (Dense)                   (None, 128)               16512     \n",
      "_________________________________________________________________\n",
      "activation_36 (Activation)   (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "H1 (Dense)                   (None, 128)               16512     \n",
      "_________________________________________________________________\n",
      "activation_37 (Activation)   (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "H2 (Dense)                   (None, 128)               16512     \n",
      "_________________________________________________________________\n",
      "activation_38 (Activation)   (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "last (Dense)                 (None, 128)               16512     \n",
      "_________________________________________________________________\n",
      "activation_39 (Activation)   (None, 128)               0         \n",
      "=================================================================\n",
      "Total params: 68,224\n",
      "Trainable params: 68,224\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "import scnets as scn\n",
    "model = scn.fullycon(in_size=16, \n",
    "                     out_size=128, \n",
    "                     batch_size=64,\n",
    "                     N_hidden=3, \n",
    "                     N_neurons=128, \n",
    "                     N_gpus=3)\n",
    "\n",
    "model.summary()    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T12:14:55.706347Z",
     "start_time": "2018-12-24T11:30:04.408558Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 180000 samples, validate on 20000 samples\n",
      "Epoch 1/200\n",
      "  2048/180000 [..............................] - ETA: 22s - loss: 1067.9223 - calc_mre_K: 0.3911"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/hegder/anaconda3/lib/python3.7/site-packages/mxnet/module/bucketing_module.py:408: UserWarning: Optimizer created manually outside Module but rescale_grad is not normalized to 1.0/batch_size/num_workers (1.0 vs. 0.015625). Is this intended?\n",
      "  force_init=force_init)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "180000/180000 [==============================] - 14s 76us/step - loss: 586.6970 - calc_mre_K: 0.2149 - val_loss: 486.2231 - val_calc_mre_K: 0.1782\n",
      "Epoch 2/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 425.6830 - calc_mre_K: 0.1559 - val_loss: 387.3521 - val_calc_mre_K: 0.1420\n",
      "Epoch 3/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 356.6724 - calc_mre_K: 0.1306 - val_loss: 323.5688 - val_calc_mre_K: 0.1186\n",
      "Epoch 4/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 303.3858 - calc_mre_K: 0.1111 - val_loss: 283.9960 - val_calc_mre_K: 0.1041\n",
      "Epoch 5/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 269.8481 - calc_mre_K: 0.0988 - val_loss: 264.3996 - val_calc_mre_K: 0.0969\n",
      "Epoch 6/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 250.0728 - calc_mre_K: 0.0916 - val_loss: 243.8528 - val_calc_mre_K: 0.0894\n",
      "Epoch 7/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 235.4887 - calc_mre_K: 0.0862 - val_loss: 231.1335 - val_calc_mre_K: 0.0847\n",
      "Epoch 8/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 224.0921 - calc_mre_K: 0.0821 - val_loss: 222.4504 - val_calc_mre_K: 0.0815\n",
      "Epoch 9/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 214.7462 - calc_mre_K: 0.0786 - val_loss: 213.2136 - val_calc_mre_K: 0.0781\n",
      "Epoch 10/200\n",
      "180000/180000 [==============================] - 14s 77us/step - loss: 207.1419 - calc_mre_K: 0.0759 - val_loss: 207.0157 - val_calc_mre_K: 0.0759\n",
      "Epoch 11/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 200.7301 - calc_mre_K: 0.0735 - val_loss: 203.3895 - val_calc_mre_K: 0.0745\n",
      "Epoch 12/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 195.3061 - calc_mre_K: 0.0715 - val_loss: 195.5112 - val_calc_mre_K: 0.0716\n",
      "Epoch 13/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 190.7506 - calc_mre_K: 0.0699 - val_loss: 189.3304 - val_calc_mre_K: 0.0694\n",
      "Epoch 14/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 186.5810 - calc_mre_K: 0.0683 - val_loss: 186.5654 - val_calc_mre_K: 0.0684\n",
      "Epoch 15/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 182.8832 - calc_mre_K: 0.0670 - val_loss: 185.1371 - val_calc_mre_K: 0.0679\n",
      "Epoch 16/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 179.4296 - calc_mre_K: 0.0657 - val_loss: 180.5062 - val_calc_mre_K: 0.0661\n",
      "Epoch 17/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 176.5398 - calc_mre_K: 0.0647 - val_loss: 177.7483 - val_calc_mre_K: 0.0651\n",
      "Epoch 18/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 173.6522 - calc_mre_K: 0.0636 - val_loss: 176.4632 - val_calc_mre_K: 0.0647\n",
      "Epoch 19/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 170.8636 - calc_mre_K: 0.0626 - val_loss: 174.8414 - val_calc_mre_K: 0.0641\n",
      "Epoch 20/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 168.4388 - calc_mre_K: 0.0617 - val_loss: 169.3205 - val_calc_mre_K: 0.0621\n",
      "Epoch 21/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 166.2766 - calc_mre_K: 0.0609 - val_loss: 168.7226 - val_calc_mre_K: 0.0618\n",
      "Epoch 22/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 164.4004 - calc_mre_K: 0.0602 - val_loss: 165.1768 - val_calc_mre_K: 0.0605\n",
      "Epoch 23/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 162.5570 - calc_mre_K: 0.0595 - val_loss: 163.8222 - val_calc_mre_K: 0.0600\n",
      "Epoch 24/200\n",
      "180000/180000 [==============================] - 13s 72us/step - loss: 160.8853 - calc_mre_K: 0.0589 - val_loss: 164.4745 - val_calc_mre_K: 0.0603\n",
      "Epoch 25/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 159.4165 - calc_mre_K: 0.0584 - val_loss: 162.8385 - val_calc_mre_K: 0.0597\n",
      "Epoch 26/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 157.9590 - calc_mre_K: 0.0579 - val_loss: 158.0162 - val_calc_mre_K: 0.0579\n",
      "Epoch 27/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 156.6144 - calc_mre_K: 0.0574 - val_loss: 158.1233 - val_calc_mre_K: 0.0579\n",
      "Epoch 28/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 155.3613 - calc_mre_K: 0.0569 - val_loss: 155.5807 - val_calc_mre_K: 0.0570\n",
      "Epoch 29/200\n",
      "180000/180000 [==============================] - 14s 77us/step - loss: 154.2818 - calc_mre_K: 0.0565 - val_loss: 157.3408 - val_calc_mre_K: 0.0577\n",
      "Epoch 30/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 153.0406 - calc_mre_K: 0.0560 - val_loss: 156.1291 - val_calc_mre_K: 0.0572\n",
      "Epoch 31/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 151.9722 - calc_mre_K: 0.0557 - val_loss: 153.7511 - val_calc_mre_K: 0.0564\n",
      "Epoch 32/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 151.0765 - calc_mre_K: 0.0553 - val_loss: 151.9524 - val_calc_mre_K: 0.0557\n",
      "Epoch 33/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 150.2245 - calc_mre_K: 0.0550 - val_loss: 157.4623 - val_calc_mre_K: 0.0577\n",
      "Epoch 34/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 149.3630 - calc_mre_K: 0.0547 - val_loss: 154.7732 - val_calc_mre_K: 0.0567\n",
      "Epoch 35/200\n",
      "180000/180000 [==============================] - 14s 77us/step - loss: 148.6071 - calc_mre_K: 0.0544 - val_loss: 152.2949 - val_calc_mre_K: 0.0558\n",
      "Epoch 36/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 147.6697 - calc_mre_K: 0.0541 - val_loss: 150.3888 - val_calc_mre_K: 0.0551\n",
      "Epoch 37/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 147.0172 - calc_mre_K: 0.0538 - val_loss: 154.8937 - val_calc_mre_K: 0.0568\n",
      "Epoch 38/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 146.0331 - calc_mre_K: 0.0535 - val_loss: 149.7763 - val_calc_mre_K: 0.0549\n",
      "Epoch 39/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 145.3110 - calc_mre_K: 0.0532 - val_loss: 148.9532 - val_calc_mre_K: 0.0546\n",
      "Epoch 40/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 144.4204 - calc_mre_K: 0.0529 - val_loss: 145.0329 - val_calc_mre_K: 0.0531\n",
      "Epoch 41/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 143.7615 - calc_mre_K: 0.0527 - val_loss: 147.1543 - val_calc_mre_K: 0.0539\n",
      "Epoch 42/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 142.8767 - calc_mre_K: 0.0523 - val_loss: 144.3887 - val_calc_mre_K: 0.0529\n",
      "Epoch 43/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 142.3874 - calc_mre_K: 0.0521 - val_loss: 145.7458 - val_calc_mre_K: 0.0534\n",
      "Epoch 44/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 141.8685 - calc_mre_K: 0.0520 - val_loss: 147.2032 - val_calc_mre_K: 0.0539\n",
      "Epoch 45/200\n",
      "180000/180000 [==============================] - 14s 77us/step - loss: 141.1346 - calc_mre_K: 0.0517 - val_loss: 144.7660 - val_calc_mre_K: 0.0531\n",
      "Epoch 46/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 140.7750 - calc_mre_K: 0.0516 - val_loss: 144.0061 - val_calc_mre_K: 0.0528\n",
      "Epoch 47/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 140.1142 - calc_mre_K: 0.0513 - val_loss: 141.1734 - val_calc_mre_K: 0.0517\n",
      "Epoch 48/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 139.8645 - calc_mre_K: 0.0512 - val_loss: 142.9880 - val_calc_mre_K: 0.0524\n",
      "Epoch 49/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 139.3830 - calc_mre_K: 0.0510 - val_loss: 143.6409 - val_calc_mre_K: 0.0526\n",
      "Epoch 50/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 138.8140 - calc_mre_K: 0.0508 - val_loss: 143.2184 - val_calc_mre_K: 0.0525\n",
      "Epoch 51/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 138.3973 - calc_mre_K: 0.0507 - val_loss: 139.8861 - val_calc_mre_K: 0.0513\n",
      "Epoch 52/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "180000/180000 [==============================] - 14s 75us/step - loss: 138.0892 - calc_mre_K: 0.0506 - val_loss: 140.6725 - val_calc_mre_K: 0.0516\n",
      "Epoch 53/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 137.7028 - calc_mre_K: 0.0504 - val_loss: 138.8589 - val_calc_mre_K: 0.0509\n",
      "Epoch 54/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 137.3089 - calc_mre_K: 0.0503 - val_loss: 138.8965 - val_calc_mre_K: 0.0509\n",
      "Epoch 55/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 136.9568 - calc_mre_K: 0.0502 - val_loss: 138.5185 - val_calc_mre_K: 0.0508\n",
      "Epoch 56/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 136.6009 - calc_mre_K: 0.0500 - val_loss: 138.7215 - val_calc_mre_K: 0.0508\n",
      "Epoch 57/200\n",
      "180000/180000 [==============================] - 13s 72us/step - loss: 136.0806 - calc_mre_K: 0.0498 - val_loss: 138.3350 - val_calc_mre_K: 0.0507\n",
      "Epoch 58/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 135.8345 - calc_mre_K: 0.0497 - val_loss: 137.7190 - val_calc_mre_K: 0.0505\n",
      "Epoch 59/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 135.6358 - calc_mre_K: 0.0497 - val_loss: 138.6744 - val_calc_mre_K: 0.0508\n",
      "Epoch 60/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 135.3754 - calc_mre_K: 0.0496 - val_loss: 139.0814 - val_calc_mre_K: 0.0510\n",
      "Epoch 61/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 134.8890 - calc_mre_K: 0.0494 - val_loss: 136.9085 - val_calc_mre_K: 0.0502\n",
      "Epoch 62/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 134.6396 - calc_mre_K: 0.0493 - val_loss: 137.3877 - val_calc_mre_K: 0.0503\n",
      "Epoch 63/200\n",
      "180000/180000 [==============================] - 13s 72us/step - loss: 134.3992 - calc_mre_K: 0.0492 - val_loss: 135.2938 - val_calc_mre_K: 0.0496\n",
      "Epoch 64/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 134.3122 - calc_mre_K: 0.0492 - val_loss: 139.1176 - val_calc_mre_K: 0.0510\n",
      "Epoch 65/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 133.7135 - calc_mre_K: 0.0490 - val_loss: 135.2479 - val_calc_mre_K: 0.0496\n",
      "Epoch 66/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 133.7190 - calc_mre_K: 0.0490 - val_loss: 135.7683 - val_calc_mre_K: 0.0498\n",
      "Epoch 67/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 133.5231 - calc_mre_K: 0.0489 - val_loss: 136.1366 - val_calc_mre_K: 0.0499\n",
      "Epoch 68/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 133.2074 - calc_mre_K: 0.0488 - val_loss: 135.8328 - val_calc_mre_K: 0.0498\n",
      "Epoch 69/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 132.9496 - calc_mre_K: 0.0487 - val_loss: 134.4080 - val_calc_mre_K: 0.0493\n",
      "Epoch 70/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 132.7247 - calc_mre_K: 0.0486 - val_loss: 137.4605 - val_calc_mre_K: 0.0504\n",
      "Epoch 71/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 132.4558 - calc_mre_K: 0.0485 - val_loss: 139.4707 - val_calc_mre_K: 0.0511\n",
      "Epoch 72/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 132.4104 - calc_mre_K: 0.0485 - val_loss: 137.1305 - val_calc_mre_K: 0.0503\n",
      "Epoch 73/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 132.1399 - calc_mre_K: 0.0484 - val_loss: 134.0655 - val_calc_mre_K: 0.0491\n",
      "Epoch 74/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 131.9301 - calc_mre_K: 0.0483 - val_loss: 134.4354 - val_calc_mre_K: 0.0493\n",
      "Epoch 75/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 131.7246 - calc_mre_K: 0.0482 - val_loss: 134.5573 - val_calc_mre_K: 0.0493\n",
      "Epoch 76/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 131.5269 - calc_mre_K: 0.0482 - val_loss: 134.4077 - val_calc_mre_K: 0.0493\n",
      "Epoch 77/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 131.2656 - calc_mre_K: 0.0481 - val_loss: 134.7935 - val_calc_mre_K: 0.0494\n",
      "Epoch 78/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 125.7911 - calc_mre_K: 0.0461 - val_loss: 120.3471 - val_calc_mre_K: 0.0441\n",
      "Epoch 79/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 119.3826 - calc_mre_K: 0.0437 - val_loss: 123.1155 - val_calc_mre_K: 0.0451\n",
      "Epoch 80/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 119.3031 - calc_mre_K: 0.0437 - val_loss: 122.2213 - val_calc_mre_K: 0.0448\n",
      "Epoch 81/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 118.9789 - calc_mre_K: 0.0436 - val_loss: 120.6334 - val_calc_mre_K: 0.0442\n",
      "Epoch 82/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 118.9969 - calc_mre_K: 0.0436 - val_loss: 120.7862 - val_calc_mre_K: 0.0443\n",
      "Epoch 83/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 118.7809 - calc_mre_K: 0.0435 - val_loss: 120.8567 - val_calc_mre_K: 0.0443\n",
      "Epoch 84/200\n",
      "180000/180000 [==============================] - 14s 79us/step - loss: 118.4634 - calc_mre_K: 0.0434 - val_loss: 123.2169 - val_calc_mre_K: 0.0452\n",
      "Epoch 85/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 118.3166 - calc_mre_K: 0.0433 - val_loss: 120.7881 - val_calc_mre_K: 0.0443\n",
      "Epoch 86/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 118.2228 - calc_mre_K: 0.0433 - val_loss: 122.8423 - val_calc_mre_K: 0.0450\n",
      "Epoch 87/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 118.1214 - calc_mre_K: 0.0433 - val_loss: 120.5271 - val_calc_mre_K: 0.0442\n",
      "Epoch 88/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 117.8085 - calc_mre_K: 0.0431 - val_loss: 120.9597 - val_calc_mre_K: 0.0443\n",
      "Epoch 89/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 117.7032 - calc_mre_K: 0.0431 - val_loss: 121.4759 - val_calc_mre_K: 0.0445\n",
      "Epoch 90/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 117.5624 - calc_mre_K: 0.0431 - val_loss: 119.1851 - val_calc_mre_K: 0.0437\n",
      "Epoch 91/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 117.2560 - calc_mre_K: 0.0429 - val_loss: 122.3540 - val_calc_mre_K: 0.0448\n",
      "Epoch 92/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 117.1679 - calc_mre_K: 0.0429 - val_loss: 119.8552 - val_calc_mre_K: 0.0439\n",
      "Epoch 93/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 117.0217 - calc_mre_K: 0.0429 - val_loss: 120.0495 - val_calc_mre_K: 0.0440\n",
      "Epoch 94/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 117.0185 - calc_mre_K: 0.0429 - val_loss: 117.3088 - val_calc_mre_K: 0.0430\n",
      "Epoch 95/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 116.8251 - calc_mre_K: 0.0428 - val_loss: 120.4652 - val_calc_mre_K: 0.0441\n",
      "Epoch 96/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 116.6205 - calc_mre_K: 0.0427 - val_loss: 120.3900 - val_calc_mre_K: 0.0441\n",
      "Epoch 97/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 116.4496 - calc_mre_K: 0.0426 - val_loss: 118.7385 - val_calc_mre_K: 0.0435\n",
      "Epoch 98/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 116.1749 - calc_mre_K: 0.0425 - val_loss: 120.9634 - val_calc_mre_K: 0.0443\n",
      "Epoch 99/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 116.2886 - calc_mre_K: 0.0426 - val_loss: 119.9227 - val_calc_mre_K: 0.0439\n",
      "Epoch 100/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 116.0157 - calc_mre_K: 0.0425 - val_loss: 117.9213 - val_calc_mre_K: 0.0432\n",
      "Epoch 101/200\n",
      "180000/180000 [==============================] - 13s 71us/step - loss: 115.8638 - calc_mre_K: 0.0424 - val_loss: 118.1739 - val_calc_mre_K: 0.0433\n",
      "Epoch 102/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 115.8557 - calc_mre_K: 0.0424 - val_loss: 118.0157 - val_calc_mre_K: 0.0432\n",
      "Epoch 103/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "180000/180000 [==============================] - 14s 77us/step - loss: 115.8097 - calc_mre_K: 0.0424 - val_loss: 118.3356 - val_calc_mre_K: 0.0434\n",
      "Epoch 104/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 115.4539 - calc_mre_K: 0.0423 - val_loss: 118.5177 - val_calc_mre_K: 0.0434\n",
      "Epoch 105/200\n",
      "180000/180000 [==============================] - 13s 72us/step - loss: 115.3549 - calc_mre_K: 0.0422 - val_loss: 117.6739 - val_calc_mre_K: 0.0431\n",
      "Epoch 106/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 115.3671 - calc_mre_K: 0.0423 - val_loss: 117.9273 - val_calc_mre_K: 0.0432\n",
      "Epoch 107/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 115.1121 - calc_mre_K: 0.0422 - val_loss: 118.1231 - val_calc_mre_K: 0.0433\n",
      "Epoch 108/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 115.0414 - calc_mre_K: 0.0421 - val_loss: 117.0860 - val_calc_mre_K: 0.0429\n",
      "Epoch 109/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 115.0065 - calc_mre_K: 0.0421 - val_loss: 115.8543 - val_calc_mre_K: 0.0425\n",
      "Epoch 110/200\n",
      "180000/180000 [==============================] - 14s 77us/step - loss: 114.7209 - calc_mre_K: 0.0420 - val_loss: 117.3827 - val_calc_mre_K: 0.0430\n",
      "Epoch 111/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 114.7564 - calc_mre_K: 0.0420 - val_loss: 118.1904 - val_calc_mre_K: 0.0433\n",
      "Epoch 112/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 114.5286 - calc_mre_K: 0.0419 - val_loss: 115.2011 - val_calc_mre_K: 0.0422\n",
      "Epoch 113/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 114.3709 - calc_mre_K: 0.0419 - val_loss: 120.3179 - val_calc_mre_K: 0.0441\n",
      "Epoch 114/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 114.4362 - calc_mre_K: 0.0419 - val_loss: 116.7370 - val_calc_mre_K: 0.0428\n",
      "Epoch 115/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 114.4016 - calc_mre_K: 0.0419 - val_loss: 118.0783 - val_calc_mre_K: 0.0433\n",
      "Epoch 116/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 114.2110 - calc_mre_K: 0.0418 - val_loss: 118.7904 - val_calc_mre_K: 0.0435\n",
      "Epoch 117/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 114.1403 - calc_mre_K: 0.0418 - val_loss: 117.1699 - val_calc_mre_K: 0.0429\n",
      "Epoch 118/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 113.9420 - calc_mre_K: 0.0417 - val_loss: 118.5807 - val_calc_mre_K: 0.0435\n",
      "Epoch 119/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 113.8674 - calc_mre_K: 0.0417 - val_loss: 116.5409 - val_calc_mre_K: 0.0427\n",
      "Epoch 120/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 113.7845 - calc_mre_K: 0.0417 - val_loss: 115.9241 - val_calc_mre_K: 0.0425\n",
      "Epoch 121/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 113.7896 - calc_mre_K: 0.0417 - val_loss: 116.5504 - val_calc_mre_K: 0.0427\n",
      "Epoch 122/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 113.6858 - calc_mre_K: 0.0416 - val_loss: 117.1755 - val_calc_mre_K: 0.0429\n",
      "Epoch 123/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 113.5650 - calc_mre_K: 0.0416 - val_loss: 115.6700 - val_calc_mre_K: 0.0424\n",
      "Epoch 124/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 113.5974 - calc_mre_K: 0.0416 - val_loss: 115.9245 - val_calc_mre_K: 0.0425\n",
      "Epoch 125/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 113.2688 - calc_mre_K: 0.0415 - val_loss: 115.1209 - val_calc_mre_K: 0.0422\n",
      "Epoch 126/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 113.2184 - calc_mre_K: 0.0415 - val_loss: 121.6078 - val_calc_mre_K: 0.0446\n",
      "Epoch 127/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 113.1851 - calc_mre_K: 0.0415 - val_loss: 115.0124 - val_calc_mre_K: 0.0421\n",
      "Epoch 128/200\n",
      "180000/180000 [==============================] - 14s 77us/step - loss: 113.1473 - calc_mre_K: 0.0414 - val_loss: 115.1346 - val_calc_mre_K: 0.0422\n",
      "Epoch 129/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 112.9800 - calc_mre_K: 0.0414 - val_loss: 116.6493 - val_calc_mre_K: 0.0427\n",
      "Epoch 130/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 112.8253 - calc_mre_K: 0.0413 - val_loss: 116.9615 - val_calc_mre_K: 0.0429\n",
      "Epoch 131/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 112.8448 - calc_mre_K: 0.0413 - val_loss: 117.5529 - val_calc_mre_K: 0.0431\n",
      "Epoch 132/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 112.6031 - calc_mre_K: 0.0412 - val_loss: 113.0558 - val_calc_mre_K: 0.0414\n",
      "Epoch 133/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 112.8572 - calc_mre_K: 0.0413 - val_loss: 122.2617 - val_calc_mre_K: 0.0448\n",
      "Epoch 134/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 112.6042 - calc_mre_K: 0.0412 - val_loss: 113.9315 - val_calc_mre_K: 0.0417\n",
      "Epoch 135/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 112.5802 - calc_mre_K: 0.0412 - val_loss: 115.1420 - val_calc_mre_K: 0.0422\n",
      "Epoch 136/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 112.4730 - calc_mre_K: 0.0412 - val_loss: 116.4512 - val_calc_mre_K: 0.0427\n",
      "Epoch 137/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 112.2334 - calc_mre_K: 0.0411 - val_loss: 115.2554 - val_calc_mre_K: 0.0422\n",
      "Epoch 138/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 112.2175 - calc_mre_K: 0.0411 - val_loss: 114.1861 - val_calc_mre_K: 0.0418\n",
      "Epoch 139/200\n",
      "180000/180000 [==============================] - 14s 78us/step - loss: 112.3223 - calc_mre_K: 0.0411 - val_loss: 116.3959 - val_calc_mre_K: 0.0427\n",
      "Epoch 140/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 112.2103 - calc_mre_K: 0.0411 - val_loss: 115.8674 - val_calc_mre_K: 0.0425\n",
      "Epoch 141/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 112.0046 - calc_mre_K: 0.0410 - val_loss: 118.0436 - val_calc_mre_K: 0.0433\n",
      "Epoch 142/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 111.9950 - calc_mre_K: 0.0410 - val_loss: 117.0975 - val_calc_mre_K: 0.0429\n",
      "Epoch 143/200\n",
      "180000/180000 [==============================] - 14s 78us/step - loss: 111.8503 - calc_mre_K: 0.0410 - val_loss: 112.9747 - val_calc_mre_K: 0.0414\n",
      "Epoch 144/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 111.7370 - calc_mre_K: 0.0409 - val_loss: 115.8392 - val_calc_mre_K: 0.0424\n",
      "Epoch 145/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 111.7339 - calc_mre_K: 0.0409 - val_loss: 113.2684 - val_calc_mre_K: 0.0415\n",
      "Epoch 146/200\n",
      "180000/180000 [==============================] - 14s 77us/step - loss: 111.6293 - calc_mre_K: 0.0409 - val_loss: 115.6064 - val_calc_mre_K: 0.0424\n",
      "Epoch 147/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 111.4826 - calc_mre_K: 0.0408 - val_loss: 112.9015 - val_calc_mre_K: 0.0414\n",
      "Epoch 148/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 111.4659 - calc_mre_K: 0.0408 - val_loss: 113.5482 - val_calc_mre_K: 0.0416\n",
      "Epoch 149/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 111.2826 - calc_mre_K: 0.0408 - val_loss: 114.0267 - val_calc_mre_K: 0.0418\n",
      "Epoch 150/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 111.2027 - calc_mre_K: 0.0407 - val_loss: 113.9669 - val_calc_mre_K: 0.0418\n",
      "Epoch 151/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 111.2015 - calc_mre_K: 0.0407 - val_loss: 114.8176 - val_calc_mre_K: 0.0421\n",
      "Epoch 152/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 111.1044 - calc_mre_K: 0.0407 - val_loss: 115.5316 - val_calc_mre_K: 0.0423\n",
      "Epoch 153/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 111.0018 - calc_mre_K: 0.0407 - val_loss: 113.1085 - val_calc_mre_K: 0.0414\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 154/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 110.8464 - calc_mre_K: 0.0406 - val_loss: 115.1649 - val_calc_mre_K: 0.0422\n",
      "Epoch 155/200\n",
      "180000/180000 [==============================] - 14s 77us/step - loss: 110.7531 - calc_mre_K: 0.0406 - val_loss: 112.3627 - val_calc_mre_K: 0.0412\n",
      "Epoch 156/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 110.7424 - calc_mre_K: 0.0406 - val_loss: 113.4726 - val_calc_mre_K: 0.0416\n",
      "Epoch 157/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 110.7128 - calc_mre_K: 0.0405 - val_loss: 115.1285 - val_calc_mre_K: 0.0422\n",
      "Epoch 158/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 110.7458 - calc_mre_K: 0.0406 - val_loss: 115.5896 - val_calc_mre_K: 0.0424\n",
      "Epoch 159/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 110.7272 - calc_mre_K: 0.0406 - val_loss: 115.1904 - val_calc_mre_K: 0.0422\n",
      "Epoch 160/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 110.4099 - calc_mre_K: 0.0404 - val_loss: 114.0765 - val_calc_mre_K: 0.0418\n",
      "Epoch 161/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 110.5111 - calc_mre_K: 0.0405 - val_loss: 115.1958 - val_calc_mre_K: 0.0422\n",
      "Epoch 162/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 110.2853 - calc_mre_K: 0.0404 - val_loss: 114.3131 - val_calc_mre_K: 0.0419\n",
      "Epoch 163/200\n",
      "180000/180000 [==============================] - 14s 77us/step - loss: 110.4154 - calc_mre_K: 0.0404 - val_loss: 115.1023 - val_calc_mre_K: 0.0422\n",
      "Epoch 164/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 110.2121 - calc_mre_K: 0.0404 - val_loss: 112.0128 - val_calc_mre_K: 0.0410\n",
      "Epoch 165/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 110.2456 - calc_mre_K: 0.0404 - val_loss: 113.3032 - val_calc_mre_K: 0.0415\n",
      "Epoch 166/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 110.0765 - calc_mre_K: 0.0403 - val_loss: 112.6374 - val_calc_mre_K: 0.0413\n",
      "Epoch 167/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 110.0118 - calc_mre_K: 0.0403 - val_loss: 111.5512 - val_calc_mre_K: 0.0409\n",
      "Epoch 168/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 110.0178 - calc_mre_K: 0.0403 - val_loss: 114.4487 - val_calc_mre_K: 0.0419\n",
      "Epoch 169/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 110.0767 - calc_mre_K: 0.0403 - val_loss: 114.3352 - val_calc_mre_K: 0.0419\n",
      "Epoch 170/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 109.9391 - calc_mre_K: 0.0403 - val_loss: 111.9408 - val_calc_mre_K: 0.0410\n",
      "Epoch 171/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 109.7299 - calc_mre_K: 0.0402 - val_loss: 112.9176 - val_calc_mre_K: 0.0414\n",
      "Epoch 172/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 109.8765 - calc_mre_K: 0.0402 - val_loss: 111.6814 - val_calc_mre_K: 0.0409\n",
      "Epoch 173/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 109.6485 - calc_mre_K: 0.0402 - val_loss: 113.3964 - val_calc_mre_K: 0.0416\n",
      "Epoch 174/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 109.6828 - calc_mre_K: 0.0402 - val_loss: 113.9606 - val_calc_mre_K: 0.0418\n",
      "Epoch 175/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 109.6594 - calc_mre_K: 0.0402 - val_loss: 111.0966 - val_calc_mre_K: 0.0407\n",
      "Epoch 176/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 109.5734 - calc_mre_K: 0.0401 - val_loss: 112.2511 - val_calc_mre_K: 0.0411\n",
      "Epoch 177/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 109.4159 - calc_mre_K: 0.0401 - val_loss: 111.1301 - val_calc_mre_K: 0.0407\n",
      "Epoch 178/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 109.4449 - calc_mre_K: 0.0401 - val_loss: 110.6236 - val_calc_mre_K: 0.0405\n",
      "Epoch 179/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 109.3654 - calc_mre_K: 0.0401 - val_loss: 111.6166 - val_calc_mre_K: 0.0409\n",
      "Epoch 180/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 109.2524 - calc_mre_K: 0.0400 - val_loss: 112.9461 - val_calc_mre_K: 0.0414\n",
      "Epoch 181/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 109.2225 - calc_mre_K: 0.0400 - val_loss: 110.1173 - val_calc_mre_K: 0.0404\n",
      "Epoch 182/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 109.1709 - calc_mre_K: 0.0400 - val_loss: 114.7229 - val_calc_mre_K: 0.0420\n",
      "Epoch 183/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 109.0046 - calc_mre_K: 0.0399 - val_loss: 113.1834 - val_calc_mre_K: 0.0415\n",
      "Epoch 184/200\n",
      "180000/180000 [==============================] - 14s 75us/step - loss: 109.0439 - calc_mre_K: 0.0399 - val_loss: 111.2284 - val_calc_mre_K: 0.0408\n",
      "Epoch 185/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 109.0802 - calc_mre_K: 0.0399 - val_loss: 113.1321 - val_calc_mre_K: 0.0415\n",
      "Epoch 186/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 108.9818 - calc_mre_K: 0.0399 - val_loss: 111.1894 - val_calc_mre_K: 0.0407\n",
      "Epoch 187/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 108.9389 - calc_mre_K: 0.0399 - val_loss: 111.3882 - val_calc_mre_K: 0.0408\n",
      "Epoch 188/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 108.9094 - calc_mre_K: 0.0399 - val_loss: 111.0407 - val_calc_mre_K: 0.0407\n",
      "Epoch 189/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 108.8313 - calc_mre_K: 0.0399 - val_loss: 111.8095 - val_calc_mre_K: 0.0410\n",
      "Epoch 190/200\n",
      "180000/180000 [==============================] - 14s 77us/step - loss: 108.7589 - calc_mre_K: 0.0398 - val_loss: 111.1458 - val_calc_mre_K: 0.0407\n",
      "Epoch 191/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 108.7291 - calc_mre_K: 0.0398 - val_loss: 111.6015 - val_calc_mre_K: 0.0409\n",
      "Epoch 192/200\n",
      "180000/180000 [==============================] - 14s 79us/step - loss: 108.7573 - calc_mre_K: 0.0398 - val_loss: 109.6390 - val_calc_mre_K: 0.0402\n",
      "Epoch 193/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 108.6145 - calc_mre_K: 0.0398 - val_loss: 114.1445 - val_calc_mre_K: 0.0418\n",
      "Epoch 194/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 108.6406 - calc_mre_K: 0.0398 - val_loss: 110.2498 - val_calc_mre_K: 0.0404\n",
      "Epoch 195/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 108.5555 - calc_mre_K: 0.0398 - val_loss: 111.8910 - val_calc_mre_K: 0.0410\n",
      "Epoch 196/200\n",
      "180000/180000 [==============================] - 13s 73us/step - loss: 108.4471 - calc_mre_K: 0.0397 - val_loss: 110.8580 - val_calc_mre_K: 0.0406\n",
      "Epoch 197/200\n",
      "180000/180000 [==============================] - 14s 76us/step - loss: 108.4934 - calc_mre_K: 0.0397 - val_loss: 110.3960 - val_calc_mre_K: 0.0405\n",
      "Epoch 198/200\n",
      "180000/180000 [==============================] - 13s 75us/step - loss: 108.3710 - calc_mre_K: 0.0397 - val_loss: 111.3316 - val_calc_mre_K: 0.0408\n",
      "Epoch 199/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 108.4080 - calc_mre_K: 0.0397 - val_loss: 110.5910 - val_calc_mre_K: 0.0405\n",
      "Epoch 200/200\n",
      "180000/180000 [==============================] - 13s 74us/step - loss: 108.4204 - calc_mre_K: 0.0397 - val_loss: 111.7881 - val_calc_mre_K: 0.0410\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(x_train, y_train,\n",
    "                    batch_size=64,\n",
    "                    epochs=200, \n",
    "                    verbose=1,\n",
    "                    validation_data=(x_test, y_test))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T08:43:50.327911Z",
     "start_time": "2018-12-24T08:43:50.304722Z"
    }
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T13:55:51.147176Z",
     "start_time": "2018-12-24T13:55:51.009996Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-0.8568308   0.95530558  0.55021432  0.57709433  0.51667057 -0.13719201\n",
      "  0.10193289 -0.09344975  0.67611094 -0.57865021 -0.89294536 -0.11672836\n",
      " -0.23931468  0.93205126 -0.5196566   0.98282012]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 460.8x403.2 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "numbr = np.random.randint(0,2000)\n",
    "lams = np.linspace(400, 800, endpoint=True, num=128)\n",
    "yz = model.predict(x_test[numbr:numbr+3])\n",
    "plt.plot( lams,  yz[0])\n",
    "plt.plot( lams, y_test[numbr])\n",
    "plt.ylim([0,1])\n",
    "print(x_test[numbr])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T12:24:54.693850Z",
     "start_time": "2018-12-24T12:24:54.672379Z"
    }
   },
   "outputs": [
    {
     "ename": "SyntaxError",
     "evalue": "invalid syntax (<ipython-input-111-d8901bf549c6>, line 1)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;36m  File \u001b[0;32m\"<ipython-input-111-d8901bf549c6>\"\u001b[0;36m, line \u001b[0;32m1\u001b[0m\n\u001b[0;31m    test = np.array([  1.55886594  48.2155035   85.86266944  40.18463303  41.39958257\u001b[0m\n\u001b[0m                                            ^\u001b[0m\n\u001b[0;31mSyntaxError\u001b[0m\u001b[0;31m:\u001b[0m invalid syntax\n"
     ]
    }
   ],
   "source": [
    "test = np.array([  1.55886594,  48.2155035,   85.86266944,  40.18463303,  41.39958257,\n",
    "  40.9130793   90.7680843   50.48189625 100.          74.36306673\n",
    "  99.93251134  99.99088077  99.88496324  67.78092403 100.\n",
    "  92.1395581 ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T13:41:43.224099Z",
     "start_time": "2018-12-24T13:41:43.202555Z"
    }
   },
   "outputs": [],
   "source": [
    "\n",
    "pt = np.array([ 0.79899497,  1.        , -0.79899497, -0.95979899,  0.42713568,\n",
    "       -1.        ,  1.        ,  1.        ,  0.63819095, -1.        ,\n",
    "        1.        , -0.59798995, -0.86934673,  0.68844221, -0.34673367,\n",
    "       -0.77889447])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T14:18:25.945782Z",
     "start_time": "2018-12-24T14:18:25.926975Z"
    }
   },
   "outputs": [],
   "source": [
    "ttt = np.tile(pt, (15,1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T14:18:27.982296Z",
     "start_time": "2018-12-24T14:18:27.810376Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0, 10)"
      ]
     },
     "execution_count": 186,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 460.8x403.2 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "zg = model.predict(ttt)\n",
    "plt.plot(lams, 100*zg[0])\n",
    "plt.ylim([0,10] )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T14:03:37.067486Z",
     "start_time": "2018-12-24T14:03:37.022150Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MXNet Backend: Successfully exported the model as MXNet model!\n",
      "MXNet symbol file -  siti_fully-symbol.json\n",
      "MXNet params file -  siti_fully-0000.params\n",
      "\n",
      "\n",
      "Model input data_names and data_shapes are: \n",
      "data_names :  ['/first_input9']\n",
      "data_shapes :  [DataDesc[/first_input9,(3, 16),float32,NCHW]]\n",
      "\n",
      "\n",
      "Note: In the above data_shapes, the first dimension represent the batch_size used for model training. \n",
      "You can change the batch_size for binding the module based on your inference batch_size.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(['/first_input9'], [DataDesc[/first_input9,(3, 16),float32,NCHW]])"
      ]
     },
     "execution_count": 170,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Import the save_mxnet_model API\n",
    "from keras.models import save_mxnet_model\n",
    "save_mxnet_model(model=model, prefix='siti_fully', epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T14:07:14.828378Z",
     "start_time": "2018-12-24T14:07:14.805617Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import mxnet as mx\n",
    "\n",
    "# Step1: Load the model in MXNet\n",
    "\n",
    "# Use the same prefix and epoch parameters we used in save_mxnet_model API.\n",
    "sym, arg_params, aux_params = mx.model.load_checkpoint(prefix='siti_fully', epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T14:18:39.620655Z",
     "start_time": "2018-12-24T14:18:39.590332Z"
    }
   },
   "outputs": [],
   "source": [
    "# We use the data_names and data_shapes returned by save_mxnet_model API.\n",
    "mod = mx.mod.Module(symbol=sym, \n",
    "                    data_names=['/first_input9'], \n",
    "                    context=mx.gpu(), \n",
    "                    label_names=None)\n",
    "mod.bind(for_training=False, \n",
    "         data_shapes=[('/first_input9', (3,16))], \n",
    "         label_shapes=mod._label_shapes)\n",
    "mod.set_params(arg_params, aux_params, allow_missing=True)\n",
    "\n",
    "#data_iter = mx.io.NDArrayIter(x_test, None, 1)\n",
    "data_iter = mx.io.NDArrayIter(ttt, None, 1)\n",
    "res2 = mod.predict(data_iter).asnumpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-12-24T14:19:27.200347Z",
     "start_time": "2018-12-24T14:19:27.079370Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0, 100)"
      ]
     },
     "execution_count": 189,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAFWCAYAAABHDstlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzt3Xl0VOed5vHnvVWl0oaEECCBWCTwBrENBgy2MQRjjA3BjrdJ4k4mnXTH3Wecnp7u2DM9czonp3vOmU7nzKSTzun2kHG8JbETdyd2HNtgVmMwtmVjzGZkGzCrJMSmfSmVqt75o7CAgAR6q6S6Vf5+zvFx6b513/f9CdBT9773Xpk5879gBQDAAHnpngAAIDMRIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADASfBS3jTz+pt0zfSZGl02RrVHDumZp5b3toVycrRk2X26/MqpivX0aNv772jD2pWX3A4AyEyBcROv+LuLvamouFhHDh1QS3OTioqHa+e2Lb1tS5bdq8LCYfrFE49q5473dMuipZKkutrDl9QOAMhMl3QK66OaXfqoZpfa21vP2R4MhfS5a67ThvWvqqurU81Njare/LqmXTf7ktoBAJnrkk5h9WVE6SgFAgE1HK3r3Xa0/ohGjS6TMeai7dZe+CkqufmF6ol2JzM1AECKBEM56upoO397Mp2Gc8KKRqOy8Xjvtq6uLnleQMFQ6KLt0e7zQyI3v1BL7/1mMtMCAKTYiuefPC9EkgqQSHdEoVBIxvN6QyI3N1fxeEw90ehF2y/k0yOPlS88pZ6eC7+nP8bzNHbilao7+NE5wZWJqMWfqMW/sqkev9QSDIa05J5vXPCsUFIBcurkccViMZWVjdHR+lpJUtmYCp04fkzW2ou29yfaHXE6jWU8T7FYTNFIV1b8BaIW/6EW/8qmevxSS39jX9IiuvE8BYJBeV5AxpjE60BAPdGoPtj5vj6/8A6Fc3NVPLxEc276vLZtrZaki7YDADLXJR2B3Dz/Vs1bsLj367/57vd18MA+PfPUcq1Z+Xvdsexe/cVf/61isZi2ba3WlurNve+9WDsAIDNdUoBs2rBGmzasuWBbd3dEv3/+V33ue7F2AEBm4lEmAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwEkxFJ4XDirR4yd2aUDlJknTk0AGtWvGCWluaZTxPixYv09XTZsoYow9379CqV15QLBZLxdAAgDRJyRHI7V+4R17A06M//r7+5Uf/S9Fot5beeb8kae68hZpQOVmPPfpDLf/JDzRyVLkWLFqaimEBAGmUkgApKSlVza7t6u6OqCca1a4dWzWqrFySNG3GbG3euFZtrS3q6GjXpg2rde30WZIxqRgaAJAmKTmFVf3WRk353DTt3VMjG7e6ZtpM7fv4Q4Vzc1VcXKKj9XW9722or1VeXr6KiorV0tzUZ5/G82S8gefbp/u47Os31OJP1OJf2VSPX2rpb/yUBMiRQwc0/brr9Z2/+XtZKx1rqNOzT/8/5eSEJUmRSGfve7u6Eq/D4XC/fVZUTUlqnaSiaqrzvn5DLf5ELf6VTfWku5ZAINBnW/IBYowe+PqDqvlgh371y59JkuYtWKwvf+1P9evTX4fDeers6JAk5ebmSZIikUi/3dbur1FPtHvg0/E8VVRNVe3+3bLx+ID39xNq8Sdq8a9sqscvtQRDOZox6+YLtyXbeV5enoYPH6Et1W+oJxqVJG2pfkM3zl0gz/PU3NyosvKxamo8KUkqG1Ohzs4OtbQ099uvjceT+qYlu7+fUIs/UYt/ZVM96a6lv7GTPrnW2dGhUyePa+b1NykQDCoQDGrWnJvV0tykzo4Obd/6jubOX6jCYUXKzy/QvAW3ace2LZK1yQ4NAEijlKyB/ObXT2nR7XfpP3/nuzLGqOFonf79V09KkjZvWq+8/AI9+NDDvfeBbFi3MhXDAgDSKCUBcuL4sd71jj9k43GtWfmi1qx8MRVDAQB8IvOvdQMApAUBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwEU9XRZVdM0fxbFmtE6Sh1RyKqfmujqt98XcbztGjxMl09baaMMfpw9w6teuUFxWKxVA0NAEiDlATIpMuu1JJl9+mlF36tgwc/USgUUnHxcEnS3HkLNaFysh579IeKx2K6/4FvasGipVq36qVUDA0ASJOUnMKaf8vt2rxxnQ7s3ysbj6s7EtHxYw2SpGkzZmvzxrVqa21RR0e7Nm1YrWunz5KMScXQAIA0SfoIJBQKaczYCu3etU1/9u1HlJeXryOHD2rNqy8qEulScXGJjtbX9b6/ob5WeXn5KioqVktzU5/9Gs+T8Qaeb5/u47Kv31CLP1GLf2VTPX6ppb/xkw6Q3Lx8GePp2umz9Nwzj6u9vU233X6X7v3S1/Xb556WJEUinb3v7+pKvA6Hw/32W1E1Jal1koqqqc77+g21+BO1+Fc21ZPuWgKBQJ9tSQdIdyQiSXq3+g01NzVKkjasX6m//m9/L2utJCkczlNnR4ckKTc3T5IUOb1fX2r316gn2j3g+RjPU0XVVNXu3y0bjw94fz+hFn+iFv/Kpnr8UkswlKMZs26+cFuynUciXWpqOiWdDgtJ0lkvm5sbVVY+Vk2NJyVJZWMq1NnZoZaW5n77tfF4Ut+0ZPf3E2rxJ2rxr2yqJ9219Dd2Sk6uvb/lbV1/wzwNKypWIBjU/IW3q672sNpaW7R96zuaO3+hCocVKT+/QPMW3KYd27acGzgAgIyTkst439q8Qbm5efrTP/8ryRgdOXRAz//bzyVJmzetV15+gR586OHe+0A2rFuZimEBAGmUmhsJrdVra1fotbUrzm+Kx7Vm5Ytas/LFlAwFAPCHzL/WDQCQFgQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcEKAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBwQoAAAJwQIAAAJwQIAMAJAQIAcBJMaWfBoL710MPKzy/QP/3j9yRJxvO0aPEyXT1tpowx+nD3Dq165QXFYrFUDg0AGGIpPQKZf8vtam1pPmfb3HkLNaFysh579Ida/pMfaOSoci1YtDSVwwIA0iBlRyBlYyo0+fKrtHbVS7rnP3ytd/u0GbO1btVLamttkSRt2rBad9//Va1b/bJkbZ/9Gc+T8Qaeb5/u47Kv31CLP1GLf2VTPX6ppb/xUxIgxvO09M77tWrFC+dsD+fmqri4REfr63q3NdTXKi8vX0VFxWppbuqzz4qqKUmd5qqomuq8r99Qiz9Ri39lUz3priUQCPTZlpIAmXPjfB1rqNehA59oQuWk3u05OWFJUiTS2butqyvxOhwO99tn7f4a9US7BzwX43mqqJqq2v27ZePxAe/vJ9TiT9TiX9lUj19qCYZyNGPWzRduS7bz4SWlmjn7Jj2x/MfntXV3RyRJ4XCeOjs6JEm5uXmSpEgk0m+/Nh5P6puW7P5+Qi3+RC3+lU31pLuW/sZOOkDGT6xSfn6hHvz2I5IShzvhcFh/+cj39NvnnlZzc6PKyseqqfGkpMRaSWdnh1r+YLEdAJBZkg6Qml3b9Mnej3q/Hjd+or7wxS/p8eU/Umdnh7ZvfUdz5y9U7ZGDisdimrfgNu3YtqXfBXQAgP8lHSA9PT3qaWvt/bqzM3Gqqv30ts2b1isvv0APPvRw730gG9atTHZYAECapfRGQkk6dOCT3psIpcT5szUrX9SalS+meigAQBpl/sXSAIC0IEAAAE4IEACAEwIEAOCEAAEAOCFAAABOCBAAgBMCBADghAABADghQAAATggQAIATAgQA4IQAAQA4IUAAAE4IEACAEwIEAOCEAAEAOCFAAABOCBAAgBMCBADghAABADghQAAATggQAIATAgQA4IQAAQA4IUAAAE4IEACAEwIEAOCEAAEAOCFAAABOCBAAgBMCBADghAABADghQAAATggQAIATAgQA4IQAAQA4IUAAAE4IEACAEwIEAOCEAAEAOCFAAABOCBAAgBMCBADghAABADghQAAATggQAIATAgQA4IQAAQA4IUAAAE4IEACAk2CyHQQCAS1eeo8qJ12mgoJCtbW26L1339S7b78hSTKep0WLl+nqaTNljNGHu3do1SsvKBaLJT15AED6JB0gnuepva1Vv/7FY2psPKXRZWP0wNe+pbbWVtV8sF1z5y3UhMrJeuzRHyoei+n+B76pBYuWat2ql1IxfwBAmiQdINFoVBtfW9X79bGjddq7p0bjJlSq5oPtmjZjttatekltrS2SpE0bVuvu+7+qdatflqzts1/jeTLewM+wfbqPy75+Qy3+RC3+lU31+KWW/sZPOkAuNNi4CVV6e/MGhXNzVVxcoqP1db3tDfW1ysvLV1FRsVqam/rsp6JqSlKnuSqqpjrv6zfU4k/U4l/ZVE+6awkEAn22pTxAFi/5oiJdXdq5/T0VFBRKkiKRzt72rq7E63A43G8/tftr1BPtHvD4xvNUUTVVtft3y8bjA97fT6jFn6jFv7KpHr/UEgzlaMasmy/clsqBbl28TOMnVOmZp5crHoupuzsiSQqH89TZ0SFJys3NkyRFIpF++7LxeFLftGT39xNq8Sdq8a9sqifdtfQ3dspOri264y5VTb5Czz79096wiHR1qbm5UWXlY3vfVzamQp2dHWppaU7V0ACANEhJgNy25IuqmnS5nn36p+roaD+nbfvWdzR3/kIVDitSfn6B5i24TTu2bel3AR0A4H9Jn8IqKh6u6+fcrJ6eqB76q//Ru/3wwf167pnHtXnTeuXlF+jBhx7uvQ9kw7qVyQ4LAEizpAOkpblJ//B3/7XPdhuPa83KF7Vm5YvJDgUAGADr5che8YDM/pdkIqdS3n/mXywNALggO+YmKZArdfd9y0QyCBAAyEI2p0gqmyNzZJ2MHZyruAgQAMhCdtxCqfWQ1Lx30MYgQAAgy9iCCqlkiszhtTKDOA4BAgBZxEqy4xdJx7fJdB0f1LEIEADIJsOvlPJGy9RtHPShCBAAyBJWRnbcLVJDtUxP+8V3SBIBAgDZYuR0KZArc/TtIRmOAAGALGC9kOzYeTL1b8jEB/4kcxcECABkAVt+gxSPSse3DtmYBAgAZDgbGiaV3TioNw1eCAECABnOjlsotR+Rmj4e0nEJEADIYLagQhoxddBvGryQlP9KWwDA4LLGk0qmyA6/UiqeLB1/X6bz2JDPgwABgAxiwyWyk+6Wcoqlxg9l9v1GajmQlrkQIACQIeyIq2UnLpGa98l8/CuZWFda50OAAEAGsOU3yo6dL3PoVenE9iFf77gQAgQAfMxKio9bKI2aKbP332Ra9qd7Sr0IEADwMTtmvjRyuszHz8i016V7OufgMl4A8KmOYLls2Q0ye//dd+EhESAA4Es2d6RO5F8nU7tOpu1wuqdzQQQIAPiMDRYoPuk+5UfrZY6/l+7p9IkAAYA0syZw5nUgT/aKP5K6W1TaucMXV1v1hUV0AEgjGyyQvebbsl3HZU59IDvic1I8Iu+T38hUXZnu6fWLIxAASKeSK6Vom0zTx7Ijp0vxHpk9z8nEo+me2UVxBAIAaWRLpkiNu2XqN8vUbz7T4Pn/873/ZwgAWcoGC6RhE2VO7U73VJwQIACQLiVXSZFTUhqepJsKBAgApIkdMUU6VePrK636Q4AAQBrYUKFUOEGmMTNPX0ksogPAkLKSFC6RHT1L6jopdR5P95ScESAAMERsTrHslf9RChdLXadkal/P2NNXEgECAEPGlt8gdTfL1Dwp09Oe7ukkjTUQABgCNpAnlU6TOfpmVoSHRIAAwNAYPUPqbpKa96V7JilDgADAILMmIDt6lszR6oxe8/hDBAgADCIrSaOuS7w4tSvNs0ktFtEBYBDY8HDZsjlS8WVSeLjMwVdlbCzd00opAgRAxrLhEbJj5kqxiEztazJK/w9oawJS+Q2JebUelqnbJDXvy5qF87MRIAAygvWCspd/RQqXSO1HpXi3NGJqYlG6YIxs8bdkD/w+3dOUveIBKWe4zCe/k2n6ON3TGVQECADfs5Js5V1SsEDm8DrZ/DIpZ5jMhz+Xaa+VNUHZcQsVv+obOhzvUvyK66RIoxRpkok0So01MvGewZ9nfrlUOF5mx7/IRFsHfbx0I0DQy+aOlMIlMs170j0V4Bx27HypaKJMzVMykcbznh9lbI/M4dXSia0qrbxOJxpbZHOKpfzyxC9pKpst7f2NTHfz4M5z5DSp5ZPPRHhIBAhOs16O7OVfkgJ50vYfZ91iHzKPzSmWSqbIFk9OfKrf82ziaKIfJnJK+T0N8o7vko3HE/14ObJVd8lO/RPp4KuJH/CxSOLqqNyRkheS6ahPfr4mKI24WubAy0n3lSkIEEiS7MQ7pFi35IWl4VdIjTXpnhI+w6wXkp3yTSnaJjXtkTm8VqazwakvE++W9v1GGjNXtvILUiAs23VSChZIwVzJxqVDa2SOb7m0uYWGSaHC80On5ErJxqTP0BE8AQLZEVdLJVfJ7H5CdvRM2dJrZAgQpNPI6VI8KlPzhIyNJ92dkaT6zVL9m1LeSCl/rNTTJrXVSoXjZSffKxkjc+zdfvuxUuK9BWOlgytkTmw/0zZyunRyZ0rmmykIkM8Qm1Mse9XXpe4WqaNBkk38Q8grkzm8WqbrhHRih+yUb8gGC7LyskP4nzUB2fIbZOrfTPkPYyObeHz62Y9Qb94j7fut7OT7ZMPFMrUbE0ctF1J6rZQ7UubQatkJd8iGS2Sa90qhYVJRpcyhV1M6X78jQD5DbPmNUneLTGNN4moRGZlTH0htq2Ta6xJv6qiXuk4lLo+8yKcxYFCUXiMZTzrr0/1gM817pY+flZ24RLZkqnR47XkL9TYQlh23UKbudZnj70mdxxKhU3aDFOuSTuyQ6To5ZHP2AwIkS9mcItmx82UOrZaJdyd++9nIaTJ7npNpPdDn83iMJJ3cKTvy2osezmcKawKSjWXVM4iylZWRLb9RpuEdGTv4l92ezbQdlnb/TBp9vWzVMtm8Upm6Tb1/b+zYzyfWZI69d+b923/8mf57xbOwspQdfb00clri6hMp8UiFjgap9cDFdz65U8ork80bPdjTHHRWSpySG7cw3VPBpSi9Rgrm9/6QHmrGxmUaqmU+/rVUdoNsxS2yuaMUn3SvNHqmzKFVidNgn74/LbP0DwIkC1kTkEqvlTnymjRsvOy4W6VRMxK/h+AS9jfRVql5T+KX32S6okmJSzXL5mRFIGYzm1MkO/42mdoNfa9BDBHTdkjm42elUTNkr/6zxAL77scTRx3oxSmsQWBHTpctHC9z4OVzPq0MmZIrJVmpoVqmvS7xaIWuE9IAHqtg6jfLXvXHsnWbLnrtvZ/ZsjmJc+leKHGp8oc//8x/avSjxJ3md0rttdLx9Bx9/CHTXivVPCl5QZnOY+meji8NSYAYz9Oixct09bSZMsbow907tOqVFxSLZd/Naja/XHbC7VK8R3bcLTJH1g/9HEbNkE7uSNwM2HpA2ve8FG0d0A9O014n27JftvwmmYOvDNZUB5XNGy0VVcnsWiXFIrJX/7lUOk06OXSLs7hEZbOl/NEyHzzmq4A3kVPpnoKvDckprLnzFmpC5WQ99ugPtfwnP9DIUeVasGjpUAw9pKyXIzvpHunENpk9v04sxo24emjnkFsqDZsoc3xb7zbT9NGZq6wGwNS/IZVek7gjOAPZstlS8x6ZyCmZnnaZ2g2y4xfJll6TjuNC/AHr5ciWXqv4VV+XHbdI5sAKmWhbuqeFARiSI5BpM2Zr3aqX1NbaIknatGG17r7/q1q3+mXJXvifcignLOMNPN9sUZU6whMVGJMrGx/aHxO2eJKkqLxjb8jYmOJ1a2QnLZUpvVyKNEk97X3WeyHGM+oIlw2oFltUKXUcVEAdUjjXrZBPRU8o1nVEqrpDpuWTpLpyqSXJAeWNnCrvk+dlTn8fbMtu2ZywbOVtUvl1Mo0fnv/n4eVIOcOkUL7UE5GirVKsU2cnzpDXMojSUosXkh1WKRWOk2JdMqc+kDmySibanPTfWeN5CgQCCoVzex9lkqn8UkswGOq7bbAHD+fmqri4REfrz3wCbqivVV5evoqKitXS3HTuhEI5kqQl93wjyZErk9w/CdP/5KwvOiWNO/2fq8qB7zLjwSTGu5BULUBXpqifS9EpXbXkAtujkoolzUmy/8ok9/eTyjSM2Xn6/1NP/5c6M2bdnNL+0skvtQRDOeqJnntxw6AHSE5OWJIUiXT2buvqSrwOh8Pnvb+ro00rnn/yvIkCANIjGMpRV8f5pxcHPUC6uyOSpHA4T50dHZKk3Nw8SVIkErngPheaKAAgPfr6QD/oi+iRri41NzeqrHxs77ayMRXq7OxQS8vgPpsfADB4huQqrO1b39Hc+QtVOKxI+fkFmrfgNu3YtmVAC8oAAH8ZkquwNm9ar7z8Aj340MO994FsWLdyKIYGAAwSM2f+FzgMAAAMWFY9yiRT73gPBAJavPQeVU66TAUFhWprbdF7776pd99+Q1Lm1hUMBvWthx5Wfn6B/ukfvycpM2u57Iopmn/LYo0oHaXuSETVb21U9ZuvZ1wthcOKtHjJ3ZpQOUmSdOTQAa1a8YJaW5p9X8vM62/SNdNnanTZGNUeOaRnnlre2xbKydGSZffp8iunKtbTo23vv6MNa1decvtQ66uW/IICLbr9Lk2YOEnh3Fy1NDdp86Z12r3zzE3BfqslMG7iFX+XttFT7Ob5t2ry5Vfp50/8q959e5NmzZmn0pGjtX/fpT8DKh2CwaDKyiu0fvXLWr9mhQ4d3K+ly+5TS3OTThxvyNi6Fty6RDnhsAoLh+mtN16TlHl/RpMuu1JL77xfq155QStfeV7vv/e2Ojva1NHennG13HnvV+R5nn755P9V9VsbNfnyqzTlc9P0wc73fV9LUXGxjhw6oJbmJhUVD9fObWd+/eySZfeqsHCYfvHEo9q54z3dcvopF3W1hy+pfaj1VUtB4TDl5uVr7arfa8O6lTp54pjuvu+rOrh/n1pPX3Dkt1qy6mm802bM1uaNa9XW2qKOjnZt2rBa106fJRk/PV3nfNFoVBtfW6XGUycla3XsaJ327qnRuAmVkjKzrrIxFZp8+VV6c9O5zwLLtFrm33K7Nm9cpwP798rG4+qORHT8WOJ3c2daLSUlparZtV3d3RH1RKPatWOrRpWVS/J/LR/V7NJHNbvU3t56zvZgKKTPXXOdNqx/VV1dnWpualT15tc17brZl9SeDn3V0tR4StVvvt4bFgf371Nd7WFVjJ8oyZ+1ZE2AXOyO90xiPE/jJlTpWEN9RtZlPC/xqX3FC4rFzvxSoEyrJRQKaczYCgVDIf3Ztx/Rf3nke7rvy3+souLhGVeLJFW/tVFTPjdN4dxc5eSEdc20mdr38YcZWcunRpSOUiAQUMPRM3M/Wn9Eo0aXyRhz0XY/C4dzVVY+Vsca6iVdvNZ0yJoAGegd7362eMkXFenq0s7t72VkXXNunK9jDfU6dODc52dlWi25efkyxtO102fpuWce17/+8/fV0d6me7/09YyrRUqseeTl5ek7f/P3+s5//58aUTpSr61dkZG1fCqcE1Y0Gj3nWVFdXV3yvICCodBF2/3KeJ7uvPcrOnxwvw58skfSxWtNh6wJkLPveP/Uxe5496NbFy/T+AlVeu6Znykei2VcXcNLSjVz9k1av/rl89oyrZbu03N6t/oNNTc1qica1Yb1KzW2Yrzs6XuYMqUWGaMHvv6gjhw5pP/9D9/V//n+d7X/k7368tf+NOP+XM4W6Y4oFAqd8+DV3NxcxeMx9USjF233I+N5+uK9DygnlKPf/eaXvdv9WEvWBEg23PG+6I67VDX5Cj379E97H/uSaXWNn1il/PxCPfjtR/SXp0/5hMNh/eUj39PIUWUZVUsk0qWmplPn3vB61stMqiUvL0/Dh4/Qluo31BONqica1ZbqN1QxbqI8z8uoWs526uRxxWIxlZWN6d1WNqZCJ44fk7X2ou1+YzxPd9/3R8ovKNS//epJ9fScOQXsx1qyJkCkzL7j/bYlX1TVpMv17NM/VUdH+zltmVRXza5tevSfv6/Hl/9Ijy//kVb8/t8ViUT0+PIfqb7uSEbVIknvb3lb198wT8OKihUIBjV/4e2qqz2sttaWjKqls6NDp04e18zrb1IgGFQgGNSsOTerpblJnR0dvq/FeJ4CwaA8LyBjTOJ1IKCeaFQf7Hxfn194R2ItZ3iJ5tz0eW3bWi1JF233Uy2e5+me+7+WCI9nnzjvqMKXtWTTjYTG87To9jt19bUzzlxI3ymeAAAA5klEQVTLvuJ3ip2V4n5UVDxcf/HXf6uenqjiZ53fPHxwv5575vGMrUuSJlRO0v1f+ca594FkUi3G6JZbl2jadddLxujIoQNavfJ3amluyrhaRo4arUW336XyseNkjFHD0TqtW/WSGo7W+b6WeQtu07wFi8/ZdvDAPj3z1HLl5IR1x7J7E/dGxGLatrX6nHsjLtY+1PqqZdNrq/W1b/6nxDqHPfNzYNeOrXr15ecl+a+WrAoQAMDQyapTWACAoUOAAACcECAAACcECADACQECAHBCgAAAnBAgAAAnBAgAwAkBAgBw8v8BHBx+7JeGBIEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 460.8x403.2 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(100*res2[0])\n",
    "plt.ylim ( [0,100])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.0"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}