Lusch_Autoencoder_linearity_loss_aux_network.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf\n",
    "from keras.layers import Input, Dense, Subtract, Add, Lambda, Dot, Concatenate\n",
    "from keras.models import Model\n",
    "from keras import layers\n",
    "from keras import regularizers\n",
    "from keras import initializers\n",
    "from keras import backend as K\n",
    "from keras.models import load_model\n",
    "from keras.constraints import UnitNorm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# the training data comes from a \n",
    "pend_data = np.load('pend_dat.npy')\n",
    "\n",
    "d_train=pend_data[:,0:10000,:,:]\n",
    "d_test=pend_data[:,10000:-1,:,:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1280000, 51, 2)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#d_train=d_train[:,:,0,:]\n",
    "d_train=d_train.reshape((128*10000, 51,2))\n",
    "np.shape(d_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1280000, 1, 2)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "zeros=np.zeros((np.shape(d_train)[0],1,2))\n",
    "np.shape(zeros)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "d_train=np.concatenate((d_train, zeros), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1280000, 52, 2)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.shape(d_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the size of our encoded representations i.e. the width of the layers\n",
    "encoding_dim = 80\n",
    "koop_dim = 2\n",
    "# this is our input placeholder for the initial condition\n",
    "input_dat = Input(shape=(2,))\n",
    "\n",
    "# and we need another fifty placeholders for the trajectory\n",
    "traject = []\n",
    "for i in range(51):\n",
    "    traject.append(Input(shape=(2,)))\n",
    "    \n",
    "# prepares the input list\n",
    "combined_in = traject\n",
    "combined_in.insert(0, input_dat)\n",
    "\n",
    "# gives weights to l2 regularization\n",
    "regularizer_weight = 1e-14\n",
    "\n",
    "# initialization of weights following Lusch\n",
    "a=1/np.sqrt(10)\n",
    "init= initializers.TruncatedNormal(mean=0.0, stddev=a, seed=None)\n",
    "\n",
    "# \"encoded\" is the encoded representation of the input, and we add hidden layers\n",
    "HL_en1 = Dense(encoding_dim, \n",
    "                activation='relu',\n",
    "                kernel_regularizer=regularizers.l2(regularizer_weight),\n",
    "                #kernel_initializer=initializers.TruncatedNormal(mean=0.0, stddev=1/2.8, seed=None),\n",
    "                bias_initializer='zeros',\n",
    "                use_bias=True)\n",
    "\n",
    "HL_en2 = Dense(encoding_dim, \n",
    "                activation='relu',\n",
    "                kernel_regularizer=regularizers.l2(regularizer_weight),\n",
    "                #kernel_initializer=init,\n",
    "                bias_initializer='zeros',\n",
    "                use_bias=True)\n",
    "\n",
    "encoded = Dense(koop_dim, \n",
    "                activation='linear',\n",
    "                kernel_regularizer=regularizers.l2(regularizer_weight),\n",
    "                kernel_initializer=init,\n",
    "                bias_initializer='zeros',\n",
    "                use_bias=True)\n",
    "\n",
    "# the koopman operator is supposed to just be a matrix, so biases are disabled\n",
    "# also, if regularizers are turned on, eigenvalues on average take absolute values smaller than unity.\n",
    "koopman = Dense(koop_dim,\n",
    "                activation='linear',\n",
    "                #kernel_constraint= UnitNorm(axis=0),\n",
    "                #kernel_regularizer=regularizers.l2(regularizer_weight),\n",
    "                #kernel_initializer=init,\n",
    "                use_bias=False)\n",
    "\n",
    "# add hidden layers for decoding\n",
    "HL_de1 = Dense(encoding_dim, \n",
    "              activation='relu',\n",
    "              kernel_regularizer=regularizers.l2(regularizer_weight),\n",
    "              #kernel_initializer=init,\n",
    "              bias_initializer='zeros',\n",
    "              use_bias=True)\n",
    "\n",
    "HL_de2 = Dense(encoding_dim, \n",
    "              activation='relu',\n",
    "              kernel_regularizer=regularizers.l2(regularizer_weight),\n",
    "              #kernel_initializer=init,\n",
    "              bias_initializer='zeros',\n",
    "              use_bias=True)\n",
    "\n",
    "# \"decoded\" is the reconstruction of the input\n",
    "decoded = Dense(2,\n",
    "                activation='linear',\n",
    "                kernel_regularizer=regularizers.l2(regularizer_weight),\n",
    "                use_bias=True)\n",
    "\n",
    "redim = Dense(2,\n",
    "                activation='linear',\n",
    "                #kernel_regularizer=regularizers.l2(regularizer_weight),\n",
    "                kernel_initializer=initializers.Ones(),\n",
    "                use_bias=False,\n",
    "                trainable=False)\n",
    "\n",
    "\n",
    "aux1=Dense(140, \n",
    "          activation='relu',\n",
    "          use_bias=True)\n",
    "\n",
    "aux2=Dense(1, \n",
    "          activation='linear',\n",
    "          use_bias=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "merged = layers.concatenate([encoded(HL_en2(HL_en1(input_dat))), aux2(aux1(encoded(HL_en2(HL_en1(input_dat)))))])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "50"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# calculates the 1st, 2nd, ..., nth application of the Koopman layer (koopman^i(encoded))\n",
    "# this corresponds to the arguments of L_pred in Lusch, 2018\n",
    "\n",
    "K_ton=[encoded(HL_en2(HL_en1(input_dat)))]\n",
    "#K_ton=[merged]\n",
    "#K_ton.append(koopman(K_ton[-1]))\n",
    "for i in range(50):\n",
    "    K_ton.append(koopman(layers.concatenate([K_ton[-1], aux2(aux1(encoded(HL_en2(HL_en1(input_dat)))))])))\n",
    "\n",
    "\n",
    "# remove the first element which is just decoded(encoded(input))\n",
    "K_ton.pop(0)\n",
    "\n",
    "# decodes these data points to compare with actual trajectory\n",
    "DK_ton=[]\n",
    "for e in K_ton:\n",
    "    DK_ton.append(decoded(HL_de2(HL_de1(e))))\n",
    "len(DK_ton)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tf.Tensor 'input_52:0' shape=(None, 2) dtype=float32>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "combined_in[-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tf.Tensor 'dense_7_50/BiasAdd:0' shape=(None, 2) dtype=float32>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "decoded(HL_de2(HL_de1(encoded(HL_en2(HL_en1(input_dat))))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "52"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "# we need an output list of the same dimension as inputs for the autoencoder to train\n",
    "# the outputs include the initial condition encoded and decoded, as well as the outcome of \n",
    "# up to fifty applications of the koopman operator\n",
    "combined_out=DK_ton\n",
    "\n",
    "def absolute(x):\n",
    "        absol_x = K.abs(x)\n",
    "        return absol_x\n",
    "\n",
    "if len(combined_out)==50:\n",
    "    combined_out.insert(0,decoded(HL_de2(HL_de1(encoded(HL_en2(HL_en1(input_dat)))))))\n",
    "    \n",
    "    linsum=Lambda(absolute)(Subtract()([encoded(HL_en2(HL_en1(combined_in[1]))), koopman(layers.concatenate([encoded(HL_en2(HL_en1(combined_in[0]))),aux2(aux1(encoded(HL_en2(HL_en1(input_dat)))))]))]))\n",
    "    #linsum=Subtract()([encoded(HL_en2(HL_en1(combined_in[1]))),koopman(encoded(HL_en2(HL_en1(combined_in[0]))))])\n",
    "    for i in range(50-2):\n",
    "        #linsum=Add()([linsum, Lambda(absolute)(Subtract()([encoded(HL_en2(HL_en1(combined_in[i+2]))),koopman(encoded(HL_en2(HL_en1(combined_in[i+1]))))]))])\n",
    "        #linsum=Add()([linsum, Subtract()([encoded(HL_en2(HL_en1(combined_in[i+2]))),koopman(encoded(HL_en2(HL_en1(combined_in[i+1]))))])])\n",
    "        linsum=Add()([linsum, Lambda(absolute)(Subtract()([encoded(HL_en2(HL_en1(combined_in[i+2]))),koopman(layers.concatenate([encoded(HL_en2(HL_en1(combined_in[0]))),aux2(aux1(encoded(HL_en2(HL_en1(combined_in[i+1])))))]))]))])\n",
    "    combined_out.insert(len(combined_out), redim(linsum))\n",
    "\n",
    "len(combined_out)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "#tf.constant([tf.math.reduce_sum(linsum),tf.math.reduce_sum(linsum)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tf.Tensor 'dense_8/MatMul:0' shape=(None, 2) dtype=float32>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "combined_out[-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "Dimensions must be equal, but are 20 and 80 for 'Mul_54' (op: 'Mul') with input shapes: [?,20], [2,80].",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mInvalidArgumentError\u001b[0m                      Traceback (most recent call last)",
      "\u001b[0;32m/usr/local/lib/python3.7/site-packages/tensorflow_core/python/framework/ops.py\u001b[0m in \u001b[0;36m_create_c_op\u001b[0;34m(graph, node_def, inputs, control_inputs)\u001b[0m\n\u001b[1;32m   1618\u001b[0m   \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1619\u001b[0;31m     \u001b[0mc_op\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mc_api\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_FinishOperation\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mop_desc\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1620\u001b[0m   \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mInvalidArgumentError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mInvalidArgumentError\u001b[0m: Dimensions must be equal, but are 20 and 80 for 'Mul_54' (op: 'Mul') with input shapes: [?,20], [2,80].",
      "\nDuring handling of the above exception, another exception occurred:\n",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-98-e901c4095ada>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m#a=Dot(1)([linsum, tensor_ones])\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mtf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mreduce_sum\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmultiply\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlinsum\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtensor_ones\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maxis\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m/usr/local/lib/python3.7/site-packages/tensorflow_core/python/util/dispatch.py\u001b[0m in \u001b[0;36mwrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    178\u001b[0m     \u001b[0;34m\"\"\"Call target, and fall back on dispatchers if there is a TypeError.\"\"\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    179\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 180\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mtarget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    181\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mTypeError\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mValueError\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    182\u001b[0m       \u001b[0;31m# Note: convert_to_eager_tensor currently raises a ValueError, not a\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.7/site-packages/tensorflow_core/python/ops/math_ops.py\u001b[0m in \u001b[0;36mmultiply\u001b[0;34m(x, y, name)\u001b[0m\n\u001b[1;32m    332\u001b[0m \u001b[0;34m@\u001b[0m\u001b[0mdispatch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0madd_dispatch_support\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    333\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mmultiply\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 334\u001b[0;31m   \u001b[0;32mreturn\u001b[0m \u001b[0mgen_math_ops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmul\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    335\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    336\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.7/site-packages/tensorflow_core/python/ops/gen_math_ops.py\u001b[0m in \u001b[0;36mmul\u001b[0;34m(x, y, name)\u001b[0m\n\u001b[1;32m   6123\u001b[0m   \u001b[0;31m# Add nodes to the TensorFlow graph.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   6124\u001b[0m   _, _, _op, _outputs = _op_def_library._apply_op_helper(\n\u001b[0;32m-> 6125\u001b[0;31m         \"Mul\", x=x, y=y, name=name)\n\u001b[0m\u001b[1;32m   6126\u001b[0m   \u001b[0m_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_outputs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   6127\u001b[0m   \u001b[0;32mif\u001b[0m \u001b[0m_execute\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmust_record_gradient\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.7/site-packages/tensorflow_core/python/framework/op_def_library.py\u001b[0m in \u001b[0;36m_apply_op_helper\u001b[0;34m(op_type_name, name, **keywords)\u001b[0m\n\u001b[1;32m    740\u001b[0m       op = g._create_op_internal(op_type_name, inputs, dtypes=None,\n\u001b[1;32m    741\u001b[0m                                  \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mscope\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput_types\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0minput_types\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 742\u001b[0;31m                                  attrs=attr_protos, op_def=op_def)\n\u001b[0m\u001b[1;32m    743\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    744\u001b[0m     \u001b[0;31m# `outputs` is returned as a separate return value so that the output\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.7/site-packages/tensorflow_core/python/framework/func_graph.py\u001b[0m in \u001b[0;36m_create_op_internal\u001b[0;34m(self, op_type, inputs, dtypes, input_types, name, attrs, op_def, compute_device)\u001b[0m\n\u001b[1;32m    593\u001b[0m     return super(FuncGraph, self)._create_op_internal(  # pylint: disable=protected-access\n\u001b[1;32m    594\u001b[0m         \u001b[0mop_type\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minputs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtypes\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput_types\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mattrs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mop_def\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 595\u001b[0;31m         compute_device)\n\u001b[0m\u001b[1;32m    596\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    597\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0mcapture\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtensor\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mshape\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.7/site-packages/tensorflow_core/python/framework/ops.py\u001b[0m in \u001b[0;36m_create_op_internal\u001b[0;34m(self, op_type, inputs, dtypes, input_types, name, attrs, op_def, compute_device)\u001b[0m\n\u001b[1;32m   3320\u001b[0m           \u001b[0minput_types\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0minput_types\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3321\u001b[0m           \u001b[0moriginal_op\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_default_original_op\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 3322\u001b[0;31m           op_def=op_def)\n\u001b[0m\u001b[1;32m   3323\u001b[0m       \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_create_op_helper\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mret\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcompute_device\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcompute_device\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3324\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mret\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.7/site-packages/tensorflow_core/python/framework/ops.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, node_def, g, inputs, output_types, control_inputs, input_types, original_op, op_def)\u001b[0m\n\u001b[1;32m   1784\u001b[0m           op_def, inputs, node_def.attr)\n\u001b[1;32m   1785\u001b[0m       self._c_op = _create_c_op(self._graph, node_def, grouped_inputs,\n\u001b[0;32m-> 1786\u001b[0;31m                                 control_input_ops)\n\u001b[0m\u001b[1;32m   1787\u001b[0m       \u001b[0mname\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_str\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnode_def\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mname\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1788\u001b[0m     \u001b[0;31m# pylint: enable=protected-access\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/usr/local/lib/python3.7/site-packages/tensorflow_core/python/framework/ops.py\u001b[0m in \u001b[0;36m_create_c_op\u001b[0;34m(graph, node_def, inputs, control_inputs)\u001b[0m\n\u001b[1;32m   1620\u001b[0m   \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mInvalidArgumentError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1621\u001b[0m     \u001b[0;31m# Convert to ValueError for backwards compatibility.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1622\u001b[0;31m     \u001b[0;32mraise\u001b[0m \u001b[0mValueError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mstr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1623\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1624\u001b[0m   \u001b[0;32mreturn\u001b[0m \u001b[0mc_op\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mValueError\u001b[0m: Dimensions must be equal, but are 20 and 80 for 'Mul_54' (op: 'Mul') with input shapes: [?,20], [2,80]."
     ]
    }
   ],
   "source": [
    "#a=Dot(1)([linsum, tensor_ones])\n",
    "\n",
    "tf.math.reduce_sum(tf.math.multiply(linsum,tensor_ones), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this model maps a trajectory to its projection based on the initial condition\n",
    "autoencoder = Model(inputs=combined_in, outputs=combined_out)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[5, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 0.02, 2.5]\n"
     ]
    }
   ],
   "source": [
    "# to compile, prepare the weights given to different loss terms, \n",
    "# these are 1 for L_recon, 1/50 for each term in L_pred\n",
    "alpha=1\n",
    "weights=[alpha/50]*50\n",
    "weights.insert(0, alpha*5)\n",
    "weights.insert(len(weights), 1/koop_dim)\n",
    "print(weights)\n",
    "\n",
    "# now we compile with this list of weights\n",
    "autoencoder.compile(optimizer='adagrad', loss='mse', loss_weights=weights)#, metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(52, 1280000, 2)"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# before training, massage the data format a bit, as the model expects a list with 51\n",
    "# arrays of dimension (trajectory_number, 2) each\n",
    "t=d_train.transpose((1,0,2))\n",
    "d=list(t)\n",
    "# this means each element of the list d represents one time step. Each element combines all trajectories and\n",
    "# two dimensions\n",
    "np.shape(d)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 640000 samples, validate on 640000 samples\n",
      "Epoch 1/5\n",
      "640000/640000 [==============================] - 318s 497us/step - loss: 2.8906 - dense_7_loss: 1.6631 - dense_8_loss: 0.5786 - dense_7_accuracy: 0.9356 - dense_7_accuracy_1: 0.9564 - dense_7_accuracy_2: 0.9672 - dense_7_accuracy_3: 0.9569 - dense_7_accuracy_4: 0.9372 - dense_7_accuracy_5: 0.9200 - dense_7_accuracy_6: 0.9072 - dense_7_accuracy_7: 0.8946 - dense_7_accuracy_8: 0.8780 - dense_7_accuracy_9: 0.8596 - dense_7_accuracy_10: 0.8335 - dense_7_accuracy_11: 0.7990 - dense_7_accuracy_12: 0.7548 - dense_7_accuracy_13: 0.6976 - dense_7_accuracy_14: 0.6340 - dense_7_accuracy_15: 0.5754 - dense_7_accuracy_16: 0.5137 - dense_7_accuracy_17: 0.4502 - dense_7_accuracy_18: 0.4236 - dense_7_accuracy_19: 0.4377 - dense_7_accuracy_20: 0.4788 - dense_7_accuracy_21: 0.5182 - dense_7_accuracy_22: 0.5423 - dense_7_accuracy_23: 0.5609 - dense_7_accuracy_24: 0.5773 - dense_7_accuracy_25: 0.5921 - dense_7_accuracy_26: 0.6044 - dense_7_accuracy_27: 0.6154 - dense_7_accuracy_28: 0.6242 - dense_7_accuracy_29: 0.6314 - dense_7_accuracy_30: 0.6376 - dense_7_accuracy_31: 0.6436 - dense_7_accuracy_32: 0.6487 - dense_7_accuracy_33: 0.6534 - dense_7_accuracy_34: 0.6576 - dense_7_accuracy_35: 0.6583 - dense_7_accuracy_36: 0.6521 - dense_7_accuracy_37: 0.6384 - dense_7_accuracy_38: 0.6194 - dense_7_accuracy_39: 0.5964 - dense_7_accuracy_40: 0.5739 - dense_7_accuracy_41: 0.5527 - dense_7_accuracy_42: 0.5331 - dense_7_accuracy_43: 0.5149 - dense_7_accuracy_44: 0.4983 - dense_7_accuracy_45: 0.4829 - dense_7_accuracy_46: 0.4688 - dense_7_accuracy_47: 0.4543 - dense_7_accuracy_48: 0.4411 - dense_7_accuracy_49: 0.4292 - dense_7_accuracy_50: 0.4185 - dense_8_accuracy: 1.0000 - val_loss: 1.4150 - val_dense_7_loss: 1.5834 - val_dense_8_loss: 0.0812 - val_dense_7_accuracy: 0.9423 - val_dense_7_accuracy_1: 0.9664 - val_dense_7_accuracy_2: 0.9855 - val_dense_7_accuracy_3: 0.9890 - val_dense_7_accuracy_4: 0.9787 - val_dense_7_accuracy_5: 0.9644 - val_dense_7_accuracy_6: 0.9483 - val_dense_7_accuracy_7: 0.9337 - val_dense_7_accuracy_8: 0.9174 - val_dense_7_accuracy_9: 0.8952 - val_dense_7_accuracy_10: 0.8660 - val_dense_7_accuracy_11: 0.8260 - val_dense_7_accuracy_12: 0.7773 - val_dense_7_accuracy_13: 0.7206 - val_dense_7_accuracy_14: 0.6373 - val_dense_7_accuracy_15: 0.5669 - val_dense_7_accuracy_16: 0.4978 - val_dense_7_accuracy_17: 0.4191 - val_dense_7_accuracy_18: 0.4001 - val_dense_7_accuracy_19: 0.4560 - val_dense_7_accuracy_20: 0.5328 - val_dense_7_accuracy_21: 0.5584 - val_dense_7_accuracy_22: 0.5770 - val_dense_7_accuracy_23: 0.5893 - val_dense_7_accuracy_24: 0.6089 - val_dense_7_accuracy_25: 0.6243 - val_dense_7_accuracy_26: 0.6362 - val_dense_7_accuracy_27: 0.6453 - val_dense_7_accuracy_28: 0.6521 - val_dense_7_accuracy_29: 0.6573 - val_dense_7_accuracy_30: 0.6618 - val_dense_7_accuracy_31: 0.6658 - val_dense_7_accuracy_32: 0.6698 - val_dense_7_accuracy_33: 0.6728 - val_dense_7_accuracy_34: 0.6745 - val_dense_7_accuracy_35: 0.6744 - val_dense_7_accuracy_36: 0.6660 - val_dense_7_accuracy_37: 0.6495 - val_dense_7_accuracy_38: 0.6272 - val_dense_7_accuracy_39: 0.6017 - val_dense_7_accuracy_40: 0.5771 - val_dense_7_accuracy_41: 0.5548 - val_dense_7_accuracy_42: 0.5334 - val_dense_7_accuracy_43: 0.5140 - val_dense_7_accuracy_44: 0.4963 - val_dense_7_accuracy_45: 0.4805 - val_dense_7_accuracy_46: 0.4656 - val_dense_7_accuracy_47: 0.4516 - val_dense_7_accuracy_48: 0.4383 - val_dense_7_accuracy_49: 0.4261 - val_dense_7_accuracy_50: 0.4158 - val_dense_8_accuracy: 1.0000\n",
      "Epoch 2/5\n",
      "640000/640000 [==============================] - 306s 477us/step - loss: 1.3645 - dense_7_loss: 1.6211 - dense_8_loss: 0.0715 - dense_7_accuracy: 0.9473 - dense_7_accuracy_1: 0.9707 - dense_7_accuracy_2: 0.9891 - dense_7_accuracy_3: 0.9888 - dense_7_accuracy_4: 0.9771 - dense_7_accuracy_5: 0.9634 - dense_7_accuracy_6: 0.9489 - dense_7_accuracy_7: 0.9353 - dense_7_accuracy_8: 0.9206 - dense_7_accuracy_9: 0.9010 - dense_7_accuracy_10: 0.8709 - dense_7_accuracy_11: 0.8307 - dense_7_accuracy_12: 0.7787 - dense_7_accuracy_13: 0.7107 - dense_7_accuracy_14: 0.6370 - dense_7_accuracy_15: 0.5685 - dense_7_accuracy_16: 0.4696 - dense_7_accuracy_17: 0.3947 - dense_7_accuracy_18: 0.3831 - dense_7_accuracy_19: 0.4254 - dense_7_accuracy_20: 0.4924 - dense_7_accuracy_21: 0.5485 - dense_7_accuracy_22: 0.5793 - dense_7_accuracy_23: 0.5997 - dense_7_accuracy_24: 0.6162 - dense_7_accuracy_25: 0.6313 - dense_7_accuracy_26: 0.6434 - dense_7_accuracy_27: 0.6536 - dense_7_accuracy_28: 0.6609 - dense_7_accuracy_29: 0.6661 - dense_7_accuracy_30: 0.6703 - dense_7_accuracy_31: 0.6740 - dense_7_accuracy_32: 0.6769 - dense_7_accuracy_33: 0.6795 - dense_7_accuracy_34: 0.6810 - dense_7_accuracy_35: 0.6770 - dense_7_accuracy_36: 0.6634 - dense_7_accuracy_37: 0.6425 - dense_7_accuracy_38: 0.6180 - dense_7_accuracy_39: 0.5914 - dense_7_accuracy_40: 0.5671 - dense_7_accuracy_41: 0.5447 - dense_7_accuracy_42: 0.5242 - dense_7_accuracy_43: 0.5052 - dense_7_accuracy_44: 0.4878 - dense_7_accuracy_45: 0.4718 - dense_7_accuracy_46: 0.4575 - dense_7_accuracy_47: 0.4431 - dense_7_accuracy_48: 0.4302 - dense_7_accuracy_49: 0.4188 - dense_7_accuracy_50: 0.4089 - dense_8_accuracy: 1.0000 - val_loss: 1.3175 - val_dense_7_loss: 1.5770 - val_dense_8_loss: 0.0619 - val_dense_7_accuracy: 0.9500 - val_dense_7_accuracy_1: 0.9709 - val_dense_7_accuracy_2: 0.9894 - val_dense_7_accuracy_3: 0.9881 - val_dense_7_accuracy_4: 0.9759 - val_dense_7_accuracy_5: 0.9632 - val_dense_7_accuracy_6: 0.9497 - val_dense_7_accuracy_7: 0.9373 - val_dense_7_accuracy_8: 0.9256 - val_dense_7_accuracy_9: 0.9058 - val_dense_7_accuracy_10: 0.8751 - val_dense_7_accuracy_11: 0.8361 - val_dense_7_accuracy_12: 0.7801 - val_dense_7_accuracy_13: 0.7088 - val_dense_7_accuracy_14: 0.6422 - val_dense_7_accuracy_15: 0.5809 - val_dense_7_accuracy_16: 0.5193 - val_dense_7_accuracy_17: 0.4485 - val_dense_7_accuracy_18: 0.4347 - val_dense_7_accuracy_19: 0.4898 - val_dense_7_accuracy_20: 0.5358 - val_dense_7_accuracy_21: 0.5555 - val_dense_7_accuracy_22: 0.5669 - val_dense_7_accuracy_23: 0.5883 - val_dense_7_accuracy_24: 0.6064 - val_dense_7_accuracy_25: 0.6212 - val_dense_7_accuracy_26: 0.6323 - val_dense_7_accuracy_27: 0.6408 - val_dense_7_accuracy_28: 0.6470 - val_dense_7_accuracy_29: 0.6516 - val_dense_7_accuracy_30: 0.6558 - val_dense_7_accuracy_31: 0.6592 - val_dense_7_accuracy_32: 0.6629 - val_dense_7_accuracy_33: 0.6656 - val_dense_7_accuracy_34: 0.6669 - val_dense_7_accuracy_35: 0.6676 - val_dense_7_accuracy_36: 0.6640 - val_dense_7_accuracy_37: 0.6506 - val_dense_7_accuracy_38: 0.6305 - val_dense_7_accuracy_39: 0.6070 - val_dense_7_accuracy_40: 0.5827 - val_dense_7_accuracy_41: 0.5604 - val_dense_7_accuracy_42: 0.5392 - val_dense_7_accuracy_43: 0.5200 - val_dense_7_accuracy_44: 0.5026 - val_dense_7_accuracy_45: 0.4870 - val_dense_7_accuracy_46: 0.4721 - val_dense_7_accuracy_47: 0.4582 - val_dense_7_accuracy_48: 0.4450 - val_dense_7_accuracy_49: 0.4331 - val_dense_7_accuracy_50: 0.4230 - val_dense_8_accuracy: 1.0000\n",
      "Epoch 3/5\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "640000/640000 [==============================] - 305s 476us/step - loss: 1.2983 - dense_7_loss: 1.6330 - dense_8_loss: 0.0603 - dense_7_accuracy: 0.9524 - dense_7_accuracy_1: 0.9721 - dense_7_accuracy_2: 0.9903 - dense_7_accuracy_3: 0.9884 - dense_7_accuracy_4: 0.9759 - dense_7_accuracy_5: 0.9632 - dense_7_accuracy_6: 0.9501 - dense_7_accuracy_7: 0.9374 - dense_7_accuracy_8: 0.9246 - dense_7_accuracy_9: 0.9047 - dense_7_accuracy_10: 0.8733 - dense_7_accuracy_11: 0.8333 - dense_7_accuracy_12: 0.7792 - dense_7_accuracy_13: 0.7099 - dense_7_accuracy_14: 0.6380 - dense_7_accuracy_15: 0.5707 - dense_7_accuracy_16: 0.4816 - dense_7_accuracy_17: 0.4004 - dense_7_accuracy_18: 0.3862 - dense_7_accuracy_19: 0.4276 - dense_7_accuracy_20: 0.4958 - dense_7_accuracy_21: 0.5509 - dense_7_accuracy_22: 0.5807 - dense_7_accuracy_23: 0.6007 - dense_7_accuracy_24: 0.6169 - dense_7_accuracy_25: 0.6322 - dense_7_accuracy_26: 0.6445 - dense_7_accuracy_27: 0.6545 - dense_7_accuracy_28: 0.6618 - dense_7_accuracy_29: 0.6674 - dense_7_accuracy_30: 0.6715 - dense_7_accuracy_31: 0.6752 - dense_7_accuracy_32: 0.6782 - dense_7_accuracy_33: 0.6807 - dense_7_accuracy_34: 0.6820 - dense_7_accuracy_35: 0.6764 - dense_7_accuracy_36: 0.6611 - dense_7_accuracy_37: 0.6392 - dense_7_accuracy_38: 0.6141 - dense_7_accuracy_39: 0.5876 - dense_7_accuracy_40: 0.5638 - dense_7_accuracy_41: 0.5417 - dense_7_accuracy_42: 0.5214 - dense_7_accuracy_43: 0.5027 - dense_7_accuracy_44: 0.4856 - dense_7_accuracy_45: 0.4698 - dense_7_accuracy_46: 0.4557 - dense_7_accuracy_47: 0.4416 - dense_7_accuracy_48: 0.4290 - dense_7_accuracy_49: 0.4178 - dense_7_accuracy_50: 0.4080 - dense_8_accuracy: 1.0000 - val_loss: 1.2713 - val_dense_7_loss: 1.5952 - val_dense_8_loss: 0.0555 - val_dense_7_accuracy: 0.9540 - val_dense_7_accuracy_1: 0.9717 - val_dense_7_accuracy_2: 0.9902 - val_dense_7_accuracy_3: 0.9884 - val_dense_7_accuracy_4: 0.9754 - val_dense_7_accuracy_5: 0.9632 - val_dense_7_accuracy_6: 0.9505 - val_dense_7_accuracy_7: 0.9377 - val_dense_7_accuracy_8: 0.9257 - val_dense_7_accuracy_9: 0.9053 - val_dense_7_accuracy_10: 0.8738 - val_dense_7_accuracy_11: 0.8351 - val_dense_7_accuracy_12: 0.7838 - val_dense_7_accuracy_13: 0.7162 - val_dense_7_accuracy_14: 0.6428 - val_dense_7_accuracy_15: 0.5777 - val_dense_7_accuracy_16: 0.5168 - val_dense_7_accuracy_17: 0.4415 - val_dense_7_accuracy_18: 0.4141 - val_dense_7_accuracy_19: 0.4629 - val_dense_7_accuracy_20: 0.5336 - val_dense_7_accuracy_21: 0.5593 - val_dense_7_accuracy_22: 0.5735 - val_dense_7_accuracy_23: 0.5893 - val_dense_7_accuracy_24: 0.6074 - val_dense_7_accuracy_25: 0.6225 - val_dense_7_accuracy_26: 0.6342 - val_dense_7_accuracy_27: 0.6436 - val_dense_7_accuracy_28: 0.6506 - val_dense_7_accuracy_29: 0.6559 - val_dense_7_accuracy_30: 0.6604 - val_dense_7_accuracy_31: 0.6644 - val_dense_7_accuracy_32: 0.6681 - val_dense_7_accuracy_33: 0.6711 - val_dense_7_accuracy_34: 0.6725 - val_dense_7_accuracy_35: 0.6729 - val_dense_7_accuracy_36: 0.6635 - val_dense_7_accuracy_37: 0.6462 - val_dense_7_accuracy_38: 0.6239 - val_dense_7_accuracy_39: 0.5990 - val_dense_7_accuracy_40: 0.5751 - val_dense_7_accuracy_41: 0.5531 - val_dense_7_accuracy_42: 0.5322 - val_dense_7_accuracy_43: 0.5132 - val_dense_7_accuracy_44: 0.4960 - val_dense_7_accuracy_45: 0.4806 - val_dense_7_accuracy_46: 0.4660 - val_dense_7_accuracy_47: 0.4524 - val_dense_7_accuracy_48: 0.4395 - val_dense_7_accuracy_49: 0.4277 - val_dense_7_accuracy_50: 0.4178 - val_dense_8_accuracy: 1.0000\n",
      "Epoch 4/5\n",
      "640000/640000 [==============================] - 300s 469us/step - loss: 1.2445 - dense_7_loss: 1.6160 - dense_8_loss: 0.0549 - dense_7_accuracy: 0.9533 - dense_7_accuracy_1: 0.9722 - dense_7_accuracy_2: 0.9903 - dense_7_accuracy_3: 0.9878 - dense_7_accuracy_4: 0.9734 - dense_7_accuracy_5: 0.9589 - dense_7_accuracy_6: 0.9436 - dense_7_accuracy_7: 0.9273 - dense_7_accuracy_8: 0.9082 - dense_7_accuracy_9: 0.8838 - dense_7_accuracy_10: 0.8502 - dense_7_accuracy_11: 0.8073 - dense_7_accuracy_12: 0.7540 - dense_7_accuracy_13: 0.6888 - dense_7_accuracy_14: 0.6140 - dense_7_accuracy_15: 0.5368 - dense_7_accuracy_16: 0.4546 - dense_7_accuracy_17: 0.3552 - dense_7_accuracy_18: 0.2933 - dense_7_accuracy_19: 0.3134 - dense_7_accuracy_20: 0.3803 - dense_7_accuracy_21: 0.4515 - dense_7_accuracy_22: 0.5124 - dense_7_accuracy_23: 0.5601 - dense_7_accuracy_24: 0.5964 - dense_7_accuracy_25: 0.6236 - dense_7_accuracy_26: 0.6435 - dense_7_accuracy_27: 0.6575 - dense_7_accuracy_28: 0.6670 - dense_7_accuracy_29: 0.6738 - dense_7_accuracy_30: 0.6789 - dense_7_accuracy_31: 0.6832 - dense_7_accuracy_32: 0.6867 - dense_7_accuracy_33: 0.6895 - dense_7_accuracy_34: 0.6900 - dense_7_accuracy_35: 0.6812 - dense_7_accuracy_36: 0.6636 - dense_7_accuracy_37: 0.6395 - dense_7_accuracy_38: 0.6127 - dense_7_accuracy_39: 0.5856 - dense_7_accuracy_40: 0.5613 - dense_7_accuracy_41: 0.5389 - dense_7_accuracy_42: 0.5183 - dense_7_accuracy_43: 0.4994 - dense_7_accuracy_44: 0.4819 - dense_7_accuracy_45: 0.4657 - dense_7_accuracy_46: 0.4510 - dense_7_accuracy_47: 0.4364 - dense_7_accuracy_48: 0.4232 - dense_7_accuracy_49: 0.4113 - dense_7_accuracy_50: 0.4007 - dense_8_accuracy: 1.0000 - val_loss: 1.2100 - val_dense_7_loss: 1.5503 - val_dense_8_loss: 0.0516 - val_dense_7_accuracy: 0.9538 - val_dense_7_accuracy_1: 0.9709 - val_dense_7_accuracy_2: 0.9898 - val_dense_7_accuracy_3: 0.9877 - val_dense_7_accuracy_4: 0.9727 - val_dense_7_accuracy_5: 0.9571 - val_dense_7_accuracy_6: 0.9405 - val_dense_7_accuracy_7: 0.9220 - val_dense_7_accuracy_8: 0.8996 - val_dense_7_accuracy_9: 0.8723 - val_dense_7_accuracy_10: 0.8374 - val_dense_7_accuracy_11: 0.7925 - val_dense_7_accuracy_12: 0.7360 - val_dense_7_accuracy_13: 0.6683 - val_dense_7_accuracy_14: 0.5932 - val_dense_7_accuracy_15: 0.5115 - val_dense_7_accuracy_16: 0.4263 - val_dense_7_accuracy_17: 0.3421 - val_dense_7_accuracy_18: 0.2562 - val_dense_7_accuracy_19: 0.2548 - val_dense_7_accuracy_20: 0.3503 - val_dense_7_accuracy_21: 0.4416 - val_dense_7_accuracy_22: 0.5108 - val_dense_7_accuracy_23: 0.5534 - val_dense_7_accuracy_24: 0.5925 - val_dense_7_accuracy_25: 0.6170 - val_dense_7_accuracy_26: 0.6319 - val_dense_7_accuracy_27: 0.6439 - val_dense_7_accuracy_28: 0.6532 - val_dense_7_accuracy_29: 0.6606 - val_dense_7_accuracy_30: 0.6675 - val_dense_7_accuracy_31: 0.6736 - val_dense_7_accuracy_32: 0.6793 - val_dense_7_accuracy_33: 0.6839 - val_dense_7_accuracy_34: 0.6874 - val_dense_7_accuracy_35: 0.6876 - val_dense_7_accuracy_36: 0.6761 - val_dense_7_accuracy_37: 0.6571 - val_dense_7_accuracy_38: 0.6330 - val_dense_7_accuracy_39: 0.6066 - val_dense_7_accuracy_40: 0.5815 - val_dense_7_accuracy_41: 0.5586 - val_dense_7_accuracy_42: 0.5365 - val_dense_7_accuracy_43: 0.5163 - val_dense_7_accuracy_44: 0.4975 - val_dense_7_accuracy_45: 0.4804 - val_dense_7_accuracy_46: 0.4638 - val_dense_7_accuracy_47: 0.4480 - val_dense_7_accuracy_48: 0.4331 - val_dense_7_accuracy_49: 0.4194 - val_dense_7_accuracy_50: 0.4072 - val_dense_8_accuracy: 1.0000\n",
      "Epoch 5/5\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "640000/640000 [==============================] - 301s 470us/step - loss: 1.1986 - dense_7_loss: 1.5761 - dense_8_loss: 0.0505 - dense_7_accuracy: 0.9553 - dense_7_accuracy_1: 0.9718 - dense_7_accuracy_2: 0.9899 - dense_7_accuracy_3: 0.9867 - dense_7_accuracy_4: 0.9715 - dense_7_accuracy_5: 0.9555 - dense_7_accuracy_6: 0.9384 - dense_7_accuracy_7: 0.9190 - dense_7_accuracy_8: 0.8954 - dense_7_accuracy_9: 0.8661 - dense_7_accuracy_10: 0.8293 - dense_7_accuracy_11: 0.7827 - dense_7_accuracy_12: 0.7258 - dense_7_accuracy_13: 0.6595 - dense_7_accuracy_14: 0.5828 - dense_7_accuracy_15: 0.4990 - dense_7_accuracy_16: 0.4111 - dense_7_accuracy_17: 0.2984 - dense_7_accuracy_18: 0.2083 - dense_7_accuracy_19: 0.2092 - dense_7_accuracy_20: 0.2710 - dense_7_accuracy_21: 0.3512 - dense_7_accuracy_22: 0.4336 - dense_7_accuracy_23: 0.5027 - dense_7_accuracy_24: 0.5560 - dense_7_accuracy_25: 0.5964 - dense_7_accuracy_26: 0.6260 - dense_7_accuracy_27: 0.6467 - dense_7_accuracy_28: 0.6607 - dense_7_accuracy_29: 0.6707 - dense_7_accuracy_30: 0.6782 - dense_7_accuracy_31: 0.6847 - dense_7_accuracy_32: 0.6899 - dense_7_accuracy_33: 0.6950 - dense_7_accuracy_34: 0.6979 - dense_7_accuracy_35: 0.6918 - dense_7_accuracy_36: 0.6755 - dense_7_accuracy_37: 0.6519 - dense_7_accuracy_38: 0.6250 - dense_7_accuracy_39: 0.5971 - dense_7_accuracy_40: 0.5719 - dense_7_accuracy_41: 0.5486 - dense_7_accuracy_42: 0.5272 - dense_7_accuracy_43: 0.5073 - dense_7_accuracy_44: 0.4885 - dense_7_accuracy_45: 0.4707 - dense_7_accuracy_46: 0.4544 - dense_7_accuracy_47: 0.4380 - dense_7_accuracy_48: 0.4230 - dense_7_accuracy_49: 0.4090 - dense_7_accuracy_50: 0.3964 - dense_8_accuracy: 1.0000 - val_loss: 1.1795 - val_dense_7_loss: 1.5479 - val_dense_8_loss: 0.0471 - val_dense_7_accuracy: 0.9576 - val_dense_7_accuracy_1: 0.9706 - val_dense_7_accuracy_2: 0.9894 - val_dense_7_accuracy_3: 0.9866 - val_dense_7_accuracy_4: 0.9715 - val_dense_7_accuracy_5: 0.9558 - val_dense_7_accuracy_6: 0.9392 - val_dense_7_accuracy_7: 0.9204 - val_dense_7_accuracy_8: 0.8976 - val_dense_7_accuracy_9: 0.8697 - val_dense_7_accuracy_10: 0.8337 - val_dense_7_accuracy_11: 0.7875 - val_dense_7_accuracy_12: 0.7321 - val_dense_7_accuracy_13: 0.6651 - val_dense_7_accuracy_14: 0.5875 - val_dense_7_accuracy_15: 0.5049 - val_dense_7_accuracy_16: 0.4171 - val_dense_7_accuracy_17: 0.3314 - val_dense_7_accuracy_18: 0.2373 - val_dense_7_accuracy_19: 0.2356 - val_dense_7_accuracy_20: 0.3251 - val_dense_7_accuracy_21: 0.4205 - val_dense_7_accuracy_22: 0.4930 - val_dense_7_accuracy_23: 0.5421 - val_dense_7_accuracy_24: 0.5850 - val_dense_7_accuracy_25: 0.6117 - val_dense_7_accuracy_26: 0.6276 - val_dense_7_accuracy_27: 0.6402 - val_dense_7_accuracy_28: 0.6497 - val_dense_7_accuracy_29: 0.6575 - val_dense_7_accuracy_30: 0.6646 - val_dense_7_accuracy_31: 0.6710 - val_dense_7_accuracy_32: 0.6773 - val_dense_7_accuracy_33: 0.6828 - val_dense_7_accuracy_34: 0.6871 - val_dense_7_accuracy_35: 0.6896 - val_dense_7_accuracy_36: 0.6796 - val_dense_7_accuracy_37: 0.6620 - val_dense_7_accuracy_38: 0.6383 - val_dense_7_accuracy_39: 0.6120 - val_dense_7_accuracy_40: 0.5865 - val_dense_7_accuracy_41: 0.5634 - val_dense_7_accuracy_42: 0.5413 - val_dense_7_accuracy_43: 0.5213 - val_dense_7_accuracy_44: 0.5027 - val_dense_7_accuracy_45: 0.4852 - val_dense_7_accuracy_46: 0.4685 - val_dense_7_accuracy_47: 0.4524 - val_dense_7_accuracy_48: 0.4371 - val_dense_7_accuracy_49: 0.4223 - val_dense_7_accuracy_50: 0.4092 - val_dense_8_accuracy: 1.0000\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.callbacks.History at 0x2307fc6d8>"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "autoencoder.fit(d,\n",
    "                d,\n",
    "                epochs=5,\n",
    "                batch_size=128,\n",
    "                shuffle=False,\n",
    "                validation_split=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "test=(d_test[-1,0:10,:,:])\n",
    "zeros=np.zeros((np.shape(test)[0],1,2))\n",
    "test=np.concatenate((test, zeros), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(52, 10, 2)"
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "\n",
    "test=list(test[:,:,:].transpose(1,0,2))\n",
    "#test=np.zeros((51,1,2))\n",
    "#test[:,:,1]=1.9\n",
    "#test=list(test)\n",
    "np.shape(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [],
   "source": [
    "#autoencoder.evaluate(test, test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "p=autoencoder.predict(test)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'x1: position, x2:velocity')"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.array(p)[0:51,0,:])\n",
    "#plt.plot(np.array(p[:,0,:]).transpose())\n",
    "plt.legend(['prediction_x1', 'prediction_x2', 'actual_x1', 'actual_x2'])\n",
    "plt.xlabel(\"time[steps]\")\n",
    "plt.ylabel('x1: position, x2:velocity')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'x1: position, x2:velocity')"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "which=3\n",
    "#np.array(p).transpose(1,0,2)[0,:,:]\n",
    "plt.plot(np.array(p).transpose(1,0,2)[which,0:51,:])\n",
    "plt.plot(d_test[-1,which,0:51,:])\n",
    "#print(np.array(p).transpose(1,0,2)[which,0:51,:])\n",
    "plt.legend(['prediction_x1', 'prediction_x2', 'actual_x1', 'actual_x2'])\n",
    "plt.xlabel(\"time[steps]\")\n",
    "plt.ylabel('x1: position, x2:velocity')\n",
    "#print(np.array(p).transpose(1,0,2)[0,:,:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[array([[ 0.26136222, -0.9738055 ],\n",
      "       [-0.20607422,  0.7039781 ],\n",
      "       [ 0.646448  ,  0.8588275 ]], dtype=float32)]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.49965330958366394"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Koop=koopman.get_weights()\n",
    "print(Koop)\n",
    "Koop=np.array(Koop)\n",
    "Koop=Koop[0,0:2,:]\n",
    "sigma=np.linalg.eig(Koop)\n",
    "np.sum(np.absolute(np.array(sigma[0])))/len(sigma[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(sigma[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x23075ab70>"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "until_where=18\n",
    "t = np.linspace(0,2*np.pi,101)\n",
    "plt.plot(np.cos(t),np.sin(t))\n",
    "plt.scatter(sigma[0][0:until_where].real,sigma[0][0:until_where].imag)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# why the short time interval lusch et al use? \n",
    "\n",
    "# do decoded() and encoded() count as hidden layers in the sense that \n",
    "# lusch et al say they use two hidden layers in the encoder and decoder?\n",
    "\n",
    "# any more elegant way than the lists I used for inputs and outputs?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "autoencoder.save('10epochs_savemodel_linloss_luschparam.h5')\n",
    "autoencoder.save_weights('10epochs_saveweights_linloss_luschparam.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "\n",
    "old_ae=load_model('10epochs_savemodel_linloss_luschparam.h5')\n",
    "old_ae.load_weights('10epochs_saveweights_linloss_luschparam.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "test=(d_test[-1,0:10,:,:])\n",
    "test=list(test[:,:,:].transpose(1,0,2))\n",
    "np.shape(test)\n",
    "#old_ae.summary()\n",
    "p=old_ae.predict(np.array([[0.5,0.2],[0.4,0.1]]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "k=koopman.get_weights()\n",
    "k=np.array(k)\n",
    "k=k[0,:,:]\n",
    "k.shape\n",
    "pigma=np.linalg.eig(k)\n",
    "np.array(pigma[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "L=old_ae.layers[4].get_weights()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "L"
   ]
  },
  {
   "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": "3.7.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}