vortex_equation/scheduled_alpha-decrease2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import sympy\n",
    "import numpy as np\n",
    "import torch\n",
    "torch.manual_seed(12)\n",
    "torch.cuda.manual_seed(12)\n",
    "torch.cuda.manual_seed_all\n",
    "\n",
    "dtype = torch.float\n",
    "device = torch.device(\"cuda:0\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "sys.path.append('/u/wangnisn/devel/machine_learning_kinetics/ml_kinetics')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from mlk import learning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Set up model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "vortex = {'eqn_type':'vortex',\n",
    "        'fcn':'exp(-1/2*(x-cos(t))**2-1/2*(y-sin(t))**2)',\n",
    "        'domain':{'x':[-2,2],'y':[-2,2],'t':[0,2]},\n",
    "        \n",
    "        'err_vec':[0,0,0,1,-1,0,0]}\n",
    "\n",
    "u_t,u_x,u_y,u,x,y = sympy.symbols('u_t u_x u_y u x y')\n",
    "\n",
    "dictionary = (u_x,u_y,x*u_x,y*u_x,x*u_y,y*u_y,u)\n",
    "vortex['dictionary'] = dictionary\n",
    "pde = vortex\n",
    "domain =pde['domain']\n",
    "fcn = pde['fcn']\n",
    "n_points = 50000\n",
    "x = sympy.symbols([x for x in domain.keys()])\n",
    "u_exact = sympy.lambdify(x,sympy.sympify(fcn),'numpy')\n",
    "inputs = {}\n",
    "for key in domain.keys():\n",
    "    min_d = domain[key][0]\n",
    "    max_d = domain[key][1]\n",
    "    #iputs[key] = ((max_d-min_d)*torch.rand((n_points,1),device=device,dtype=dtype,requires_grad=True)+min_d)\n",
    "    tt = np.random.uniform(min_d,max_d,(n_points,1))\n",
    "    inputs[key] = torch.tensor(tt,device=device,dtype=dtype,requires_grad=True)\n",
    "u_exact = sympy.lambdify(x,sympy.sympify(fcn),'numpy')\n",
    "u = u_exact(*[i.detach().cpu() for i in inputs.values()])\n",
    "pde = {}\n",
    "u = u.type(dtype).to(device)\n",
    "pde['inputs'] = inputs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch:    0,  Loss u: 1.369e-01, Loss pde: 9.829e-04, Loss_norm: 7.488e+00, Loss tot: 1.369e-01\n",
      "  coefs: -5.6304e-01   -2.8970e+00    1.4370e-01   -6.2471e-01   -1.4866e+00    1.3500e+00    4.2288e-01  \n",
      "Epoch:   50,  Loss u: 1.584e-02, Loss pde: 3.436e-01, Loss_norm: 7.488e+00, Loss tot: 1.584e-02\n",
      "  coefs: -5.6304e-01   -2.8970e+00    1.4370e-01   -6.2471e-01   -1.4866e+00    1.3500e+00    4.2288e-01  \n",
      "Epoch:  100,  Loss u: 2.492e-03, Loss pde: 9.029e-01, Loss_norm: 7.488e+00, Loss tot: 2.492e-03\n",
      "  coefs: -5.6304e-01   -2.8970e+00    1.4370e-01   -6.2471e-01   -1.4866e+00    1.3500e+00    4.2288e-01  \n",
      "Epoch:  150,  Loss u: 1.326e-03, Loss pde: 9.321e-01, Loss_norm: 7.488e+00, Loss tot: 1.326e-03\n",
      "  coefs: -5.6304e-01   -2.8970e+00    1.4370e-01   -6.2471e-01   -1.4866e+00    1.3500e+00    4.2288e-01  \n",
      "Epoch:  200,  Loss u: 8.976e-04, Loss pde: 9.346e-01, Loss_norm: 7.488e+00, Loss tot: 8.976e-04\n",
      "  coefs: -5.6304e-01   -2.8970e+00    1.4370e-01   -6.2471e-01   -1.4866e+00    1.3500e+00    4.2288e-01  \n",
      "Epoch:  250,  Loss u: 6.862e-04, Loss pde: 9.649e-01, Loss_norm: 7.488e+00, Loss tot: 6.862e-04\n",
      "  coefs: -5.6304e-01   -2.8970e+00    1.4370e-01   -6.2471e-01   -1.4866e+00    1.3500e+00    4.2288e-01  \n",
      "Epoch:  300,  Loss u: 4.190e-04, Loss pde: 9.469e-01, Loss_norm: 7.488e+00, Loss tot: 4.190e-04\n",
      "  coefs: -5.6304e-01   -2.8970e+00    1.4370e-01   -6.2471e-01   -1.4866e+00    1.3500e+00    4.2288e-01  \n",
      "Epoch:  350,  Loss u: 2.946e-04, Loss pde: 9.466e-01, Loss_norm: 7.488e+00, Loss tot: 2.946e-04\n",
      "  coefs: -5.6304e-01   -2.8970e+00    1.4370e-01   -6.2471e-01   -1.4866e+00    1.3500e+00    4.2288e-01  \n",
      "Epoch:  400,  Loss u: 2.423e-04, Loss pde: 9.465e-01, Loss_norm: 7.488e+00, Loss tot: 2.423e-04\n",
      "  coefs: -5.6304e-01   -2.8970e+00    1.4370e-01   -6.2471e-01   -1.4866e+00    1.3500e+00    4.2288e-01  \n",
      "Epoch:  450,  Loss u: 7.368e-04, Loss pde: 9.414e-01, Loss_norm: 7.488e+00, Loss tot: 7.368e-04\n",
      "  coefs: -5.6304e-01   -2.8970e+00    1.4370e-01   -6.2471e-01   -1.4866e+00    1.3500e+00    4.2288e-01  \n",
      "Epoch:  500,  Loss u: 1.661e-04, Loss pde: 2.490e-03, Loss_norm: 1.915e+00, Loss tot: 2.384e-02\n",
      "  coefs:  1.3489e-01   -1.0560e-01    1.3597e-02    8.3663e-01   -8.1464e-01   -2.6838e-03    6.7564e-03  \n",
      "Epoch:  550,  Loss u: 1.721e-02, Loss pde: 1.019e-04, Loss_norm: 1.048e+00, Loss tot: 1.819e-02\n",
      "  coefs:  3.6879e-01   -3.9653e-01   -8.8406e-02    6.1094e-02   -8.8329e-02   -5.9651e-03   -3.8801e-02  \n",
      "Epoch:  600,  Loss u: 4.367e-03, Loss pde: 9.144e-05, Loss_norm: 1.244e+00, Loss tot: 5.239e-03\n",
      "  coefs:  4.6795e-01   -3.9765e-01   -6.9009e-02    1.4653e-01   -1.3349e-01   -4.8866e-03   -2.4229e-02  \n",
      "Epoch:  650,  Loss u: 1.759e-03, Loss pde: 3.769e-05, Loss_norm: 1.345e+00, Loss tot: 2.125e-03\n",
      "  coefs:  5.2662e-01   -3.9243e-01   -5.7511e-02    1.7895e-01   -1.7310e-01   -3.7023e-03   -1.2986e-02  \n",
      "Epoch:  700,  Loss u: 1.364e-03, Loss pde: 2.391e-05, Loss_norm: 1.404e+00, Loss tot: 1.600e-03\n",
      "  coefs:  5.5567e-01   -3.9217e-01   -4.7941e-02    1.9638e-01   -2.0364e-01   -0.0000e+00   -7.7126e-03  \n",
      "Epoch:  750,  Loss u: 1.197e-03, Loss pde: 2.203e-05, Loss_norm: 1.453e+00, Loss tot: 1.415e-03\n",
      "  coefs:  5.8350e-01   -3.8104e-01   -3.9751e-02    2.1453e-01   -2.2619e-01   -2.3208e-03   -5.7550e-03  \n",
      "Epoch:  800,  Loss u: 9.668e-04, Loss pde: 1.691e-05, Loss_norm: 1.480e+00, Loss tot: 1.137e-03\n",
      "  coefs:  5.8943e-01   -3.7319e-01   -3.3567e-02    2.3311e-01   -2.4435e-01   -3.4456e-03   -2.8588e-03  \n",
      "Epoch:  850,  Loss u: 8.838e-04, Loss pde: 1.589e-05, Loss_norm: 1.507e+00, Loss tot: 1.044e-03\n",
      "  coefs:  5.9050e-01   -3.6509e-01   -2.8095e-02    2.5320e-01   -2.6405e-01   -4.3355e-03   -1.5114e-03  \n",
      "Epoch:  900,  Loss u: 9.660e-04, Loss pde: 1.982e-05, Loss_norm: 1.505e+00, Loss tot: 1.161e-03\n",
      "  coefs:  6.0577e-01   -3.6265e-01   -2.3807e-02    2.5216e-01   -2.5314e-01   -6.0075e-03   -1.5252e-03  \n",
      "Epoch:  950,  Loss u: 8.490e-04, Loss pde: 1.363e-05, Loss_norm: 1.514e+00, Loss tot: 9.875e-04\n",
      "  coefs:  5.8062e-01   -3.5766e-01   -2.0897e-02    2.7310e-01   -2.7604e-01   -4.5239e-03   -1.1064e-03  \n",
      "Epoch: 1000,  Loss u: 7.904e-04, Loss pde: 1.374e-05, Loss_norm: 1.526e+00, Loss tot: 9.293e-04\n",
      "  coefs:  5.6412e-01   -3.5107e-01   -1.7177e-02    2.9297e-01   -2.9537e-01   -4.1330e-03   -9.1409e-04  \n",
      "Epoch: 1050,  Loss u: 7.413e-04, Loss pde: 1.379e-05, Loss_norm: 1.541e+00, Loss tot: 8.802e-04\n",
      "  coefs:  5.5050e-01   -3.4644e-01   -1.4350e-02    3.1127e-01   -3.1416e-01   -3.7296e-03   -5.4884e-04  \n",
      "Epoch: 1100,  Loss u: 1.032e-03, Loss pde: 1.996e-05, Loss_norm: 1.584e+00, Loss tot: 1.226e-03\n",
      "  coefs:  5.5375e-01   -3.7028e-01   -1.4939e-02    3.1738e-01   -3.2656e-01    4.9309e-04   -6.9008e-04  \n",
      "Epoch: 1150,  Loss u: 6.792e-04, Loss pde: 1.383e-05, Loss_norm: 1.584e+00, Loss tot: 8.174e-04\n",
      "  coefs:  5.3150e-01   -3.4948e-01   -1.0767e-02    3.4068e-01   -3.4718e-01   -3.1085e-03   -8.1091e-04  \n",
      "Epoch: 1200,  Loss u: 6.298e-04, Loss pde: 1.385e-05, Loss_norm: 1.591e+00, Loss tot: 7.676e-04\n",
      "  coefs:  5.1686e-01   -3.3265e-01   -9.2486e-03    3.6049e-01   -3.6648e-01   -4.1336e-03   -8.0330e-04  \n",
      "Epoch: 1250,  Loss u: 7.419e-04, Loss pde: 1.474e-05, Loss_norm: 1.610e+00, Loss tot: 8.871e-04\n",
      "  coefs:  5.5618e-01   -2.7863e-01   -1.1235e-02    3.7391e-01   -3.7090e-01   -1.4928e-02   -4.4431e-03  \n",
      "Epoch: 1300,  Loss u: 6.302e-04, Loss pde: 1.322e-05, Loss_norm: 1.619e+00, Loss tot: 7.614e-04\n",
      "  coefs:  5.2748e-01   -2.8043e-01   -1.0052e-02    3.9474e-01   -3.8918e-01   -1.2403e-02   -5.0055e-03  \n",
      "Epoch: 1350,  Loss u: 7.409e-04, Loss pde: 1.943e-05, Loss_norm: 1.688e+00, Loss tot: 9.258e-04\n",
      "  coefs:  5.3469e-01   -3.2820e-01   -9.5710e-03    4.0277e-01   -4.0659e-01   -2.9496e-03   -3.3721e-03  \n",
      "Epoch: 1400,  Loss u: 8.110e-04, Loss pde: 1.801e-05, Loss_norm: 1.774e+00, Loss tot: 9.836e-04\n",
      "  coefs:  5.5932e-01   -3.5229e-01   -7.2413e-03    4.1605e-01   -4.3133e-01   -2.8629e-03   -5.3579e-03  \n",
      "Epoch: 1450,  Loss u: 7.892e-04, Loss pde: 1.652e-05, Loss_norm: 1.793e+00, Loss tot: 9.484e-04\n",
      "  coefs:  5.6493e-01   -3.2298e-01   -2.7633e-03    4.3907e-01   -4.4648e-01   -7.3492e-03   -9.3420e-03  \n",
      "Epoch: 1500,  Loss u: 6.742e-04, Loss pde: 1.469e-05, Loss_norm: 1.775e+00, Loss tot: 8.167e-04\n",
      "  coefs:  5.3684e-01   -3.0975e-01   -3.5014e-03    4.5380e-01   -4.5597e-01   -6.2040e-03   -8.7891e-03  \n",
      "Epoch: 1550,  Loss u: 5.922e-04, Loss pde: 1.457e-05, Loss_norm: 1.754e+00, Loss tot: 7.328e-04\n",
      "  coefs:  5.0415e-01   -3.0049e-01   -3.3305e-03    4.6825e-01   -4.6596e-01   -4.8145e-03   -7.3519e-03  \n",
      "Epoch: 1600,  Loss u: 6.554e-04, Loss pde: 1.424e-05, Loss_norm: 1.729e+00, Loss tot: 7.923e-04\n",
      "  coefs:  4.8830e-01   -2.5007e-01   -1.2526e-03    4.9070e-01   -4.7810e-01   -1.1419e-02   -8.9059e-03  \n",
      "Epoch: 1650,  Loss u: 4.774e-04, Loss pde: 1.227e-05, Loss_norm: 1.732e+00, Loss tot: 5.972e-04\n",
      "  coefs:  4.6717e-01   -2.4780e-01   -2.2064e-03    5.0522e-01   -4.9334e-01   -8.9326e-03   -7.7958e-03  \n",
      "Epoch: 1700,  Loss u: 4.256e-04, Loss pde: 1.183e-05, Loss_norm: 1.726e+00, Loss tot: 5.410e-04\n",
      "  coefs:  4.4233e-01   -2.4638e-01   -1.6581e-03    5.1842e-01   -5.0559e-01   -5.6460e-03   -6.2731e-03  \n",
      "Epoch: 1750,  Loss u: 5.228e-04, Loss pde: 1.758e-05, Loss_norm: 1.678e+00, Loss tot: 6.846e-04\n",
      "  coefs:  4.1013e-01   -1.8951e-01   -1.4427e-03    5.4218e-01   -5.1421e-01   -1.4145e-02   -6.7530e-03  \n",
      "Epoch: 1800,  Loss u: 3.921e-04, Loss pde: 1.094e-05, Loss_norm: 1.696e+00, Loss tot: 4.988e-04\n",
      "  coefs:  3.9430e-01   -2.0326e-01   -2.2391e-03    5.5359e-01   -5.3020e-01   -7.2506e-03   -5.5090e-03  \n",
      "Epoch: 1850,  Loss u: 3.470e-04, Loss pde: 1.077e-05, Loss_norm: 1.717e+00, Loss tot: 4.520e-04\n",
      "  coefs:  3.8250e-01   -2.2006e-01   -1.8616e-03    5.6191e-01   -5.4309e-01   -2.6751e-03   -4.6673e-03  \n",
      "Epoch: 1900,  Loss u: 3.511e-04, Loss pde: 1.197e-05, Loss_norm: 1.734e+00, Loss tot: 4.653e-04\n",
      "  coefs:  3.5281e-01   -2.4297e-01   -4.3161e-03    5.6661e-01   -5.5823e-01    5.3524e-03   -3.7599e-03  \n",
      "Epoch: 1950,  Loss u: 3.721e-04, Loss pde: 1.358e-05, Loss_norm: 1.813e+00, Loss tot: 4.996e-04\n",
      "  coefs:  3.8102e-01   -2.5416e-01   -5.3264e-03    5.7727e-01   -5.8462e-01    6.1479e-03   -4.8371e-03  \n",
      "Epoch: 2000,  Loss u: 3.173e-04, Loss pde: 1.383e-05, Loss_norm: 1.799e+00, Loss tot: 4.459e-04\n",
      "  coefs:  3.6898e-01   -2.1184e-01   -4.7629e-03    6.0184e-01   -6.0207e-01   -2.5380e-03   -7.1228e-03  \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 2050,  Loss u: 3.277e-04, Loss pde: 1.314e-05, Loss_norm: 1.786e+00, Loss tot: 4.500e-04\n",
      "  coefs:  3.6469e-01   -2.1090e-01   -3.7411e-03    6.0658e-01   -5.8898e-01   -4.7340e-03   -6.0332e-03  \n",
      "Epoch: 2100,  Loss u: 3.026e-04, Loss pde: 1.123e-05, Loss_norm: 1.768e+00, Loss tot: 4.090e-04\n",
      "  coefs:  3.4282e-01   -2.1627e-01   -3.0241e-03    6.1261e-01   -5.8859e-01   -1.1488e-03   -3.6505e-03  \n",
      "Epoch: 2150,  Loss u: 2.647e-04, Loss pde: 9.600e-06, Loss_norm: 1.765e+00, Loss tot: 3.578e-04\n",
      "  coefs:  3.2355e-01   -2.0872e-01   -6.0681e-04    6.2659e-01   -6.0381e-01   -4.6241e-07   -1.6018e-03  \n",
      "Epoch: 2200,  Loss u: 2.547e-04, Loss pde: 1.086e-05, Loss_norm: 1.778e+00, Loss tot: 3.572e-04\n",
      "  coefs:  3.1619e-01   -1.9663e-01   -2.5119e-03    6.3557e-01   -6.2348e-01   -5.2139e-04   -3.3015e-03  \n",
      "Epoch: 2250,  Loss u: 6.961e-04, Loss pde: 2.658e-05, Loss_norm: 1.827e+00, Loss tot: 9.204e-04\n",
      "  coefs:  3.1881e-01   -2.3182e-01   -6.7490e-03    6.2841e-01   -6.3026e-01    7.7193e-03   -3.4403e-03  \n",
      "Epoch: 2300,  Loss u: 2.369e-04, Loss pde: 9.446e-06, Loss_norm: 1.834e+00, Loss tot: 3.280e-04\n",
      "  coefs:  3.1002e-01   -2.1514e-01   -5.3243e-03    6.4435e-01   -6.5227e-01    4.2759e-03   -2.4289e-03  \n",
      "Epoch: 2350,  Loss u: 2.377e-04, Loss pde: 1.191e-05, Loss_norm: 1.802e+00, Loss tot: 3.468e-04\n",
      "  coefs:  2.8800e-01   -1.7844e-01   -5.5450e-03    6.6000e-01   -6.6470e-01   -1.3886e-03   -3.8342e-03  \n",
      "Epoch: 2400,  Loss u: 2.248e-04, Loss pde: 1.155e-05, Loss_norm: 1.808e+00, Loss tot: 3.307e-04\n",
      "  coefs:  2.8367e-01   -1.7879e-01   -8.0131e-03    6.6300e-01   -6.7127e-01    0.0000e+00   -3.6776e-03  \n",
      "Epoch: 2450,  Loss u: 1.901e-04, Loss pde: 8.412e-06, Loss_norm: 1.804e+00, Loss tot: 2.716e-04\n",
      "  coefs:  2.6702e-01   -1.7538e-01   -5.6111e-03    6.7618e-01   -6.7809e-01    1.0385e-03   -9.6580e-04  \n",
      "Epoch: 2500,  Loss u: 1.967e-04, Loss pde: 9.430e-06, Loss_norm: 1.835e+00, Loss tot: 2.858e-04\n",
      "  coefs:  2.6918e-01   -1.7567e-01   -6.8041e-03    6.8344e-01   -6.9573e-01    2.1278e-03   -1.7008e-03  \n",
      "Epoch: 2550,  Loss u: 1.867e-04, Loss pde: 9.269e-06, Loss_norm: 1.871e+00, Loss tot: 2.744e-04\n",
      "  coefs:  2.7007e-01   -1.8672e-01   -9.4311e-03    6.8669e-01   -7.1010e-01    6.2921e-03   -1.4888e-03  \n",
      "Epoch: 2600,  Loss u: 1.895e-04, Loss pde: 9.547e-06, Loss_norm: 1.873e+00, Loss tot: 2.789e-04\n",
      "  coefs:  2.6449e-01   -1.8138e-01   -9.6547e-03    6.9528e-01   -7.1593e-01    4.1840e-03   -1.9929e-03  \n",
      "Epoch: 2650,  Loss u: 1.690e-04, Loss pde: 7.837e-06, Loss_norm: 1.867e+00, Loss tot: 2.452e-04\n",
      "  coefs:  2.5262e-01   -1.7807e-01   -8.6703e-03    7.0022e-01   -7.2207e-01    4.5449e-03   -4.2669e-04  \n",
      "Epoch: 2700,  Loss u: 2.056e-04, Loss pde: 1.027e-05, Loss_norm: 1.830e+00, Loss tot: 2.989e-04\n",
      "  coefs:  2.3468e-01   -1.5067e-01   -6.4669e-03    7.1399e-01   -7.2337e-01    1.2178e-03    6.8077e-05  \n",
      "Epoch: 2750,  Loss u: 1.395e-04, Loss pde: 7.677e-06, Loss_norm: 1.842e+00, Loss tot: 2.136e-04\n",
      "  coefs:  2.2914e-01   -1.4233e-01   -6.4907e-03    7.2445e-01   -7.3805e-01    4.2132e-04   -6.6207e-04  \n",
      "Epoch: 2800,  Loss u: 2.558e-04, Loss pde: 1.096e-05, Loss_norm: 1.921e+00, Loss tot: 3.540e-04\n",
      "  coefs:  2.5173e-01   -1.8314e-01   -1.0160e-02    7.1559e-01   -7.5153e-01    7.3798e-03   -1.8886e-03  \n",
      "Epoch: 2850,  Loss u: 1.467e-04, Loss pde: 7.377e-06, Loss_norm: 1.877e+00, Loss tot: 2.182e-04\n",
      "  coefs:  2.2622e-01   -1.5905e-01   -4.2950e-03    7.3051e-01   -7.5575e-01    6.7121e-04   -5.3600e-04  \n",
      "Epoch: 2900,  Loss u: 1.373e-04, Loss pde: 7.529e-06, Loss_norm: 1.878e+00, Loss tot: 2.095e-04\n",
      "  coefs:  2.1982e-01   -1.4939e-01   -2.9363e-03    7.3924e-01   -7.6518e-01   -9.8269e-04   -6.0907e-04  \n",
      "Epoch: 2950,  Loss u: 1.298e-04, Loss pde: 7.510e-06, Loss_norm: 1.850e+00, Loss tot: 2.013e-04\n",
      "  coefs:  2.0028e-01   -1.2814e-01   -7.2663e-04    7.5139e-01   -7.6586e-01   -3.8930e-03   -5.8498e-05  \n",
      "Epoch: 3000,  Loss u: 1.174e-04, Loss pde: 6.846e-06, Loss_norm: 1.846e+00, Loss tot: 1.838e-04\n",
      "  coefs:  1.9262e-01   -1.1700e-01   -1.3159e-03    7.6050e-01   -7.7232e-01   -2.2089e-03   -1.1243e-04  \n",
      "Epoch: 3050,  Loss u: 1.369e-04, Loss pde: 7.440e-06, Loss_norm: 1.827e+00, Loss tot: 2.068e-04\n",
      "  coefs:  1.7891e-01   -9.6370e-02   -2.2682e-03    7.7452e-01   -7.7388e-01   -1.2433e-03   -1.3330e-04  \n",
      "Epoch: 3100,  Loss u: 1.226e-04, Loss pde: 8.033e-06, Loss_norm: 1.846e+00, Loss tot: 1.965e-04\n",
      "  coefs:  1.7704e-01   -9.5912e-02   -9.6084e-04    7.8498e-01   -7.8369e-01   -2.2363e-03   -8.4895e-04  \n",
      "Epoch: 3150,  Loss u: 1.057e-04, Loss pde: 6.492e-06, Loss_norm: 1.858e+00, Loss tot: 1.687e-04\n",
      "  coefs:  1.6928e-01   -1.0293e-01    2.1820e-03    7.8708e-01   -7.9198e-01   -3.9037e-03   -3.5576e-04  \n",
      "Epoch: 3200,  Loss u: 1.093e-04, Loss pde: 7.278e-06, Loss_norm: 1.886e+00, Loss tot: 1.776e-04\n",
      "  coefs:  1.7598e-01   -1.1441e-01   -1.4875e-03    7.8993e-01   -8.0312e-01    1.2164e-03   -1.0639e-04  \n",
      "Epoch: 3250,  Loss u: 9.044e-05, Loss pde: 6.729e-06, Loss_norm: 1.894e+00, Loss tot: 1.548e-04\n",
      "  coefs:  1.7059e-01   -1.1252e-01   -0.0000e+00    7.9726e-01   -8.1252e-01   -8.7198e-04   -2.1331e-04  \n",
      "Epoch: 3300,  Loss u: 9.803e-05, Loss pde: 8.094e-06, Loss_norm: 1.879e+00, Loss tot: 1.710e-04\n",
      "  coefs:  1.6003e-01   -9.4693e-02   -0.0000e+00    8.0663e-01   -8.1303e-01   -3.4156e-03   -7.9034e-04  \n",
      "Epoch: 3350,  Loss u: 8.933e-05, Loss pde: 6.150e-06, Loss_norm: 1.869e+00, Loss tot: 1.489e-04\n",
      "  coefs:  1.5227e-01   -9.0386e-02   -9.2742e-04    8.0970e-01   -8.1541e-01   -6.5612e-04   -0.0000e+00  \n",
      "Epoch: 3400,  Loss u: 1.228e-04, Loss pde: 7.766e-06, Loss_norm: 1.918e+00, Loss tot: 1.933e-04\n",
      "  coefs:  1.6532e-01   -1.1329e-01   -1.1228e-03    8.0783e-01   -8.2849e-01    2.3543e-03    0.0000e+00  \n",
      "Epoch: 3450,  Loss u: 7.933e-05, Loss pde: 6.248e-06, Loss_norm: 1.905e+00, Loss tot: 1.393e-04\n",
      "  coefs:  1.5468e-01   -1.0113e-01   -5.5581e-04    8.1630e-01   -8.3210e-01   -2.4720e-04   -2.7958e-04  \n",
      "Epoch: 3500,  Loss u: 7.808e-05, Loss pde: 6.591e-06, Loss_norm: 1.901e+00, Loss tot: 1.399e-04\n",
      "  coefs:  1.4886e-01   -9.3595e-02    0.0000e+00    8.2191e-01   -8.3445e-01   -1.8514e-03   -3.5931e-04  \n",
      "Epoch: 3550,  Loss u: 8.146e-05, Loss pde: 5.827e-06, Loss_norm: 1.876e+00, Loss tot: 1.378e-04\n",
      "  coefs:  1.3500e-01   -7.7713e-02   -1.3905e-03    8.2941e-01   -8.3213e-01   -4.9074e-04    1.7780e-05  \n",
      "Epoch: 3600,  Loss u: 8.764e-05, Loss pde: 6.860e-06, Loss_norm: 1.909e+00, Loss tot: 1.506e-04\n",
      "  coefs:  1.4221e-01   -9.0099e-02   -2.8123e-04    8.3215e-01   -8.4336e-01    6.2122e-04    1.5101e-06  \n",
      "Epoch: 3650,  Loss u: 6.781e-05, Loss pde: 6.056e-06, Loss_norm: 1.914e+00, Loss tot: 1.254e-04\n",
      "  coefs:  1.3912e-01   -9.0339e-02   -2.0734e-04    8.3581e-01   -8.4886e-01   -9.8919e-06   -1.2932e-04  \n",
      "Epoch: 3700,  Loss u: 7.042e-05, Loss pde: 6.401e-06, Loss_norm: 1.917e+00, Loss tot: 1.299e-04\n",
      "  coefs:  1.3828e-01   -8.7503e-02   -2.4956e-06    8.3839e-01   -8.5172e-01   -1.0673e-03   -3.6338e-04  \n",
      "Epoch: 3750,  Loss u: 6.897e-05, Loss pde: 5.590e-06, Loss_norm: 1.920e+00, Loss tot: 1.231e-04\n",
      "  coefs:  1.3644e-01   -9.2720e-02   -0.0000e+00    8.3624e-01   -8.5404e-01   -3.3449e-04    9.2201e-05  \n",
      "Epoch: 3800,  Loss u: 9.804e-05, Loss pde: 5.663e-06, Loss_norm: 1.959e+00, Loss tot: 1.527e-04\n",
      "  coefs:  1.4724e-01   -1.1175e-01   -2.7503e-04    8.3169e-01   -8.6593e-01    1.7139e-03    1.3539e-05  \n",
      "Epoch: 3850,  Loss u: 7.414e-05, Loss pde: 5.314e-06, Loss_norm: 1.942e+00, Loss tot: 1.262e-04\n",
      "  coefs:  1.4011e-01   -9.9044e-02   -1.5823e-03    8.3707e-01   -8.6454e-01    1.0074e-05   -0.0000e+00  \n",
      "Epoch: 3900,  Loss u: 6.211e-05, Loss pde: 5.186e-06, Loss_norm: 1.931e+00, Loss tot: 1.131e-04\n",
      "  coefs:  1.3347e-01   -8.6553e-02   -4.1405e-03    8.4211e-01   -8.6287e-01    1.4842e-03   -0.0000e+00  \n",
      "Epoch: 3950,  Loss u: 6.406e-05, Loss pde: 6.159e-06, Loss_norm: 1.912e+00, Loss tot: 1.204e-04\n",
      "  coefs:  1.2321e-01   -7.0852e-02   -7.8060e-04    8.5154e-01   -8.6274e-01   -2.6605e-03   -5.4162e-05  \n",
      "Epoch: 4000,  Loss u: 6.219e-05, Loss pde: 5.420e-06, Loss_norm: 1.904e+00, Loss tot: 1.137e-04\n",
      "  coefs:  1.1701e-01   -6.6509e-02   -1.5943e-03    8.5472e-01   -8.6278e-01   -9.1682e-04    0.0000e+00  \n",
      "Epoch: 4050,  Loss u: 5.645e-05, Loss pde: 5.080e-06, Loss_norm: 1.911e+00, Loss tot: 1.058e-04\n",
      "  coefs:  1.1487e-01   -6.8122e-02   -4.7672e-04    8.5928e-01   -8.6711e-01   -7.6539e-04    1.7838e-05  \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 4100,  Loss u: 5.905e-05, Loss pde: 5.323e-06, Loss_norm: 1.941e+00, Loss tot: 1.099e-04\n",
      "  coefs:  1.1991e-01   -8.1680e-02    1.3517e-03    8.5972e-01   -8.7753e-01   -8.7783e-04    1.2454e-04  \n",
      "Epoch: 4150,  Loss u: 5.433e-05, Loss pde: 4.951e-06, Loss_norm: 1.937e+00, Loss tot: 1.027e-04\n",
      "  coefs:  1.1898e-01   -7.8408e-02   -8.8820e-04    8.6123e-01   -8.7762e-01   -1.7307e-05   -5.5109e-05  \n",
      "Epoch: 4200,  Loss u: 6.248e-05, Loss pde: 5.134e-06, Loss_norm: 1.905e+00, Loss tot: 1.113e-04\n",
      "  coefs:  1.0446e-01   -5.6643e-02   -1.6019e-03    8.6904e-01   -8.7250e-01   -5.2093e-04    2.9919e-04  \n",
      "Epoch: 4250,  Loss u: 5.114e-05, Loss pde: 5.176e-06, Loss_norm: 1.920e+00, Loss tot: 1.001e-04\n",
      "  coefs:  1.0575e-01   -6.0609e-02    3.4115e-04    8.7239e-01   -8.7892e-01   -1.8445e-03   -1.8415e-04  \n",
      "Epoch: 4300,  Loss u: 5.793e-05, Loss pde: 5.154e-06, Loss_norm: 1.950e+00, Loss tot: 1.068e-04\n",
      "  coefs:  1.1424e-01   -7.9576e-02    6.4205e-04    8.6857e-01   -8.8700e-01    0.0000e+00   -5.5752e-05  \n",
      "Epoch: 4350,  Loss u: 4.971e-05, Loss pde: 4.671e-06, Loss_norm: 1.943e+00, Loss tot: 9.554e-05\n",
      "  coefs:  1.1034e-01   -7.3865e-02   -5.3248e-04    8.7123e-01   -8.8722e-01   -9.5892e-05   -0.0000e+00  \n",
      "Epoch: 4400,  Loss u: 5.847e-05, Loss pde: 4.792e-06, Loss_norm: 1.908e+00, Loss tot: 1.044e-04\n",
      "  coefs:  9.4606e-02   -4.9788e-02   -1.9660e-03    8.7962e-01   -8.8129e-01   -2.9844e-04    3.4773e-04  \n",
      "Epoch: 4450,  Loss u: 4.484e-05, Loss pde: 4.728e-06, Loss_norm: 1.925e+00, Loss tot: 9.034e-05\n",
      "  coefs:  9.7315e-02   -5.5209e-02    2.0407e-05    8.8308e-01   -8.8841e-01   -1.2296e-03   -1.1560e-04  \n",
      "Epoch: 4500,  Loss u: 5.437e-05, Loss pde: 5.547e-06, Loss_norm: 1.955e+00, Loss tot: 1.044e-04\n",
      "  coefs:  1.0479e-01   -7.0873e-02    1.6367e-03    8.8118e-01   -8.9619e-01   -6.8735e-04    0.0000e+00  \n",
      "Epoch: 4550,  Loss u: 4.653e-05, Loss pde: 4.541e-06, Loss_norm: 1.948e+00, Loss tot: 9.076e-05\n",
      "  coefs:  1.0346e-01   -6.9585e-02   -0.0000e+00    8.7960e-01   -8.9509e-01   -4.4765e-04   -1.4620e-05  \n",
      "Epoch: 4600,  Loss u: 4.339e-05, Loss pde: 4.475e-06, Loss_norm: 1.945e+00, Loss tot: 8.700e-05\n",
      "  coefs:  1.0113e-01   -6.5319e-02   -2.5204e-04    8.8194e-01   -8.9572e-01   -4.6467e-04   -0.0000e+00  \n",
      "Epoch: 4650,  Loss u: 4.686e-05, Loss pde: 4.566e-06, Loss_norm: 1.919e+00, Loss tot: 9.048e-05\n",
      "  coefs:  8.9127e-02   -4.7326e-02   -2.0128e-03    8.8840e-01   -8.9167e-01    1.5074e-04    0.0000e+00  \n",
      "Epoch: 4700,  Loss u: 4.113e-05, Loss pde: 4.705e-06, Loss_norm: 1.931e+00, Loss tot: 8.538e-05\n",
      "  coefs:  8.9772e-02   -5.0517e-02   -9.3744e-05    8.9319e-01   -8.9724e-01   -5.2421e-04   -0.0000e+00  \n",
      "Epoch: 4750,  Loss u: 4.341e-05, Loss pde: 4.775e-06, Loss_norm: 1.951e+00, Loss tot: 8.799e-05\n",
      "  coefs:  9.5007e-02   -6.3289e-02    7.3200e-05    8.9007e-01   -9.0251e-01   -6.4601e-05   -0.0000e+00  \n",
      "Epoch: 4800,  Loss u: 3.991e-05, Loss pde: 4.259e-06, Loss_norm: 1.951e+00, Loss tot: 8.157e-05\n",
      "  coefs:  9.4370e-02   -6.2158e-02   -1.1405e-05    8.9067e-01   -9.0349e-01   -2.9980e-04    0.0000e+00  \n",
      "Epoch: 4850,  Loss u: 4.353e-05, Loss pde: 4.718e-06, Loss_norm: 1.928e+00, Loss tot: 8.711e-05\n",
      "  coefs:  8.3951e-02   -4.5756e-02   -6.1590e-04    8.9642e-01   -9.0040e-01   -1.2214e-03   -0.0000e+00  \n",
      "Epoch: 4900,  Loss u: 3.963e-05, Loss pde: 4.399e-06, Loss_norm: 1.928e+00, Loss tot: 8.134e-05\n",
      "  coefs:  7.9916e-02   -4.1664e-02    2.8395e-04    9.0174e-01   -9.0293e-01   -1.5430e-03   -1.7613e-04  \n",
      "Epoch: 4950,  Loss u: 5.112e-05, Loss pde: 5.332e-06, Loss_norm: 1.968e+00, Loss tot: 9.772e-05\n",
      "  coefs:  9.3234e-02   -6.3356e-02   -0.0000e+00    8.9865e-01   -9.1104e-01    1.0233e-03   -2.5180e-04  \n",
      "Epoch: 5000,  Loss u: 3.633e-05, Loss pde: 4.163e-06, Loss_norm: 1.954e+00, Loss tot: 7.668e-05\n",
      "  coefs:  8.6740e-02   -5.7378e-02    0.0000e+00    8.9948e-01   -9.1015e-01   -1.2873e-04   -3.5971e-05  \n",
      "Epoch: 5050,  Loss u: 3.402e-05, Loss pde: 4.055e-06, Loss_norm: 1.951e+00, Loss tot: 7.360e-05\n",
      "  coefs:  8.4656e-02   -5.3646e-02   -0.0000e+00    9.0135e-01   -9.1074e-01   -3.4858e-04   -0.0000e+00  \n",
      "Epoch: 5100,  Loss u: 4.122e-05, Loss pde: 4.296e-06, Loss_norm: 1.923e+00, Loss tot: 8.151e-05\n",
      "  coefs:  7.2049e-02   -3.4840e-02   -1.5679e-03    9.0810e-01   -9.0685e-01    3.3733e-05    0.0000e+00  \n",
      "Epoch: 5150,  Loss u: 3.416e-05, Loss pde: 4.427e-06, Loss_norm: 1.939e+00, Loss tot: 7.503e-05\n",
      "  coefs:  7.3428e-02   -3.8585e-02    6.7375e-05    9.1282e-01   -9.1295e-01   -9.6709e-04   -3.3637e-04  \n",
      "Epoch: 5200,  Loss u: 3.428e-05, Loss pde: 4.618e-06, Loss_norm: 1.957e+00, Loss tot: 7.602e-05\n",
      "  coefs:  7.9718e-02   -5.1504e-02   -6.6626e-05    9.0878e-01   -9.1696e-01    0.0000e+00   -2.5652e-04  \n",
      "Epoch: 5250,  Loss u: 3.155e-05, Loss pde: 3.972e-06, Loss_norm: 1.956e+00, Loss tot: 6.998e-05\n",
      "  coefs:  7.8746e-02   -5.1022e-02    0.0000e+00    9.0896e-01   -9.1758e-01   -1.2836e-04   -1.8098e-05  \n",
      "Epoch: 5300,  Loss u: 3.019e-05, Loss pde: 4.011e-06, Loss_norm: 1.950e+00, Loss tot: 6.854e-05\n",
      "  coefs:  7.4819e-02   -4.4761e-02   -3.0215e-05    9.1175e-01   -9.1771e-01   -6.2599e-04   -0.0000e+00  \n",
      "Epoch: 5350,  Loss u: 3.590e-05, Loss pde: 4.131e-06, Loss_norm: 1.930e+00, Loss tot: 7.441e-05\n",
      "  coefs:  6.3646e-02   -2.9876e-02    2.7838e-05    9.1900e-01   -9.1627e-01   -1.1699e-03   -1.5057e-04  \n",
      "Epoch: 5400,  Loss u: 3.125e-05, Loss pde: 4.556e-06, Loss_norm: 1.951e+00, Loss tot: 7.172e-05\n",
      "  coefs:  6.8492e-02   -3.7782e-02    0.0000e+00    9.2141e-01   -9.2205e-01   -4.6026e-04   -4.7442e-04  \n",
      "Epoch: 5450,  Loss u: 2.893e-05, Loss pde: 4.123e-06, Loss_norm: 1.961e+00, Loss tot: 6.730e-05\n",
      "  coefs:  7.1739e-02   -4.6607e-02   -3.7366e-05    9.1789e-01   -9.2418e-01    8.5515e-05   -2.6690e-04  \n",
      "Epoch: 5500,  Loss u: 2.734e-05, Loss pde: 3.828e-06, Loss_norm: 1.959e+00, Loss tot: 6.416e-05\n",
      "  coefs:  7.0716e-02   -4.5000e-02    0.0000e+00    9.1840e-01   -9.2494e-01   -2.0681e-04   -6.0327e-05  \n",
      "Epoch: 5550,  Loss u: 3.128e-05, Loss pde: 4.132e-06, Loss_norm: 1.943e+00, Loss tot: 6.910e-05\n",
      "  coefs:  6.3298e-02   -3.3308e-02   -3.3896e-04    9.2201e-01   -9.2314e-01   -8.8054e-04   -5.1064e-06  \n",
      "Epoch: 5600,  Loss u: 3.314e-05, Loss pde: 4.004e-06, Loss_norm: 1.935e+00, Loss tot: 7.010e-05\n",
      "  coefs:  5.5310e-02   -2.3996e-02    4.0049e-04    9.2899e-01   -9.2435e-01   -1.5137e-03   -2.7346e-04  \n",
      "Epoch: 5650,  Loss u: 3.434e-05, Loss pde: 4.790e-06, Loss_norm: 1.971e+00, Loss tot: 7.490e-05\n",
      "  coefs:  6.7195e-02   -4.2538e-02    2.4602e-04    9.2831e-01   -9.3209e-01    1.4583e-04   -6.6261e-04  \n",
      "Epoch: 5700,  Loss u: 2.517e-05, Loss pde: 3.925e-06, Loss_norm: 1.963e+00, Loss tot: 6.168e-05\n",
      "  coefs:  6.4106e-02   -4.0880e-02    1.6780e-04    9.2651e-01   -9.3120e-01   -1.5672e-04   -2.9910e-04  \n",
      "Epoch: 5750,  Loss u: 2.394e-05, Loss pde: 3.726e-06, Loss_norm: 1.962e+00, Loss tot: 5.940e-05\n",
      "  coefs:  6.3586e-02   -3.9839e-02   -0.0000e+00    9.2674e-01   -9.3162e-01   -1.0352e-04   -6.0600e-05  \n",
      "Epoch: 5800,  Loss u: 2.934e-05, Loss pde: 4.083e-06, Loss_norm: 1.945e+00, Loss tot: 6.594e-05\n",
      "  coefs:  5.6195e-02   -2.8113e-02   -3.9109e-04    9.2998e-01   -9.2968e-01   -9.0731e-04   -8.3946e-05  \n",
      "Epoch: 5850,  Loss u: 3.171e-05, Loss pde: 3.916e-06, Loss_norm: 1.937e+00, Loss tot: 6.733e-05\n",
      "  coefs:  4.7902e-02   -1.8758e-02    4.4880e-04    9.3707e-01   -9.3085e-01   -1.4808e-03   -3.0128e-04  \n",
      "Epoch: 5900,  Loss u: 2.956e-05, Loss pde: 4.553e-06, Loss_norm: 1.973e+00, Loss tot: 6.796e-05\n",
      "  coefs:  5.9244e-02   -3.6810e-02    5.1458e-04    9.3694e-01   -9.3859e-01    0.0000e+00   -6.8260e-04  \n",
      "Epoch: 5950,  Loss u: 2.205e-05, Loss pde: 3.822e-06, Loss_norm: 1.966e+00, Loss tot: 5.720e-05\n",
      "  coefs:  5.7155e-02   -3.5967e-02    1.9109e-04    9.3473e-01   -9.3769e-01   -1.7126e-04   -3.7621e-04  \n",
      "Epoch: 6000,  Loss u: 2.100e-05, Loss pde: 3.634e-06, Loss_norm: 1.965e+00, Loss tot: 5.518e-05\n",
      "  coefs:  5.6958e-02   -3.5303e-02   -0.0000e+00    9.3447e-01   -9.3790e-01   -7.6116e-05   -8.9650e-05  \n",
      "Epoch: 6050,  Loss u: 2.281e-05, Loss pde: 3.798e-06, Loss_norm: 1.955e+00, Loss tot: 5.736e-05\n",
      "  coefs:  5.2439e-02   -2.7872e-02   -1.4553e-04    9.3641e-01   -9.3698e-01   -7.4897e-04   -1.2342e-04  \n",
      "Epoch: 6100,  Loss u: 3.150e-05, Loss pde: 3.850e-06, Loss_norm: 1.937e+00, Loss tot: 6.589e-05\n",
      "  coefs:  4.0843e-02   -1.3667e-02    4.7323e-04    9.4402e-01   -9.3644e-01   -1.4590e-03   -2.6205e-04  \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 6150,  Loss u: 2.276e-05, Loss pde: 4.147e-06, Loss_norm: 1.963e+00, Loss tot: 5.836e-05\n",
      "  coefs:  4.7616e-02   -2.5218e-02    5.4327e-04    9.4625e-01   -9.4300e-01   -3.1041e-04   -5.4070e-04  \n",
      "Epoch: 6200,  Loss u: 1.950e-05, Loss pde: 3.805e-06, Loss_norm: 1.969e+00, Loss tot: 5.366e-05\n",
      "  coefs:  5.0787e-02   -3.1522e-02    7.7182e-05    9.4264e-01   -9.4380e-01   -0.0000e+00   -4.9269e-04  \n",
      "Epoch: 6250,  Loss u: 1.848e-05, Loss pde: 3.570e-06, Loss_norm: 1.968e+00, Loss tot: 5.155e-05\n",
      "  coefs:  5.0796e-02   -3.1228e-02    0.0000e+00    9.4171e-01   -9.4382e-01   -5.6090e-05   -1.7742e-04  \n",
      "Epoch: 6300,  Loss u: 1.801e-05, Loss pde: 3.578e-06, Loss_norm: 1.966e+00, Loss tot: 5.091e-05\n",
      "  coefs:  4.9721e-02   -2.8984e-02    0.0000e+00    9.4220e-01   -9.4402e-01   -5.2454e-04   -2.0909e-04  \n",
      "Epoch: 6350,  Loss u: 4.728e-05, Loss pde: 3.954e-06, Loss_norm: 1.923e+00, Loss tot: 8.094e-05\n",
      "  coefs:  3.1354e-02   -2.2912e-03   -1.9045e-03    9.4852e-01   -9.3766e-01   -9.6238e-04    0.0000e+00  \n",
      "Epoch: 6400,  Loss u: 2.491e-05, Loss pde: 3.756e-06, Loss_norm: 1.947e+00, Loss tot: 5.790e-05\n",
      "  coefs:  3.4396e-02   -1.0979e-02    6.2822e-05    9.5446e-01   -9.4551e-01   -7.8820e-04   -4.5527e-04  \n",
      "Epoch: 6450,  Loss u: 1.798e-05, Loss pde: 4.042e-06, Loss_norm: 1.977e+00, Loss tot: 5.210e-05\n",
      "  coefs:  4.4498e-02   -2.7850e-02    9.1068e-04    9.5182e-01   -9.5132e-01   -0.0000e+00   -6.1816e-04  \n",
      "Epoch: 6500,  Loss u: 1.628e-05, Loss pde: 3.574e-06, Loss_norm: 1.971e+00, Loss tot: 4.850e-05\n",
      "  coefs:  4.4426e-02   -2.6898e-02    0.0000e+00    9.4945e-01   -9.4981e-01   -0.0000e+00   -2.9644e-04  \n",
      "Epoch: 6550,  Loss u: 1.566e-05, Loss pde: 3.497e-06, Loss_norm: 1.971e+00, Loss tot: 4.743e-05\n",
      "  coefs:  4.4708e-02   -2.6661e-02    0.0000e+00    9.4879e-01   -9.4999e-01   -2.8631e-04   -2.2074e-04  \n",
      "Epoch: 6600,  Loss u: 1.545e-05, Loss pde: 3.569e-06, Loss_norm: 1.970e+00, Loss tot: 4.729e-05\n",
      "  coefs:  4.4638e-02   -2.5667e-02    0.0000e+00    9.4881e-01   -9.5023e-01   -6.6890e-04   -3.0785e-04  \n",
      "Epoch: 6650,  Loss u: 8.139e-05, Loss pde: 5.768e-06, Loss_norm: 1.929e+00, Loss tot: 1.200e-04\n",
      "  coefs:  2.4356e-02    5.4543e-03   -2.7187e-03    9.5290e-01   -9.4143e-01   -1.6876e-03   -3.7995e-04  \n",
      "Epoch: 6700,  Loss u: 3.110e-05, Loss pde: 3.552e-06, Loss_norm: 1.938e+00, Loss tot: 6.221e-05\n",
      "  coefs:  2.5022e-02   -3.3383e-03    2.2304e-04    9.6057e-01   -9.4802e-01   -9.1493e-04   -1.6035e-04  \n",
      "Epoch: 6750,  Loss u: 2.138e-05, Loss pde: 3.925e-06, Loss_norm: 1.984e+00, Loss tot: 5.399e-05\n",
      "  coefs:  3.6825e-02   -2.4288e-02    1.8980e-03    9.6235e-01   -9.5860e-01    1.4940e-04   -3.4352e-04  \n",
      "Epoch: 6800,  Loss u: 1.372e-05, Loss pde: 3.519e-06, Loss_norm: 1.975e+00, Loss tot: 4.474e-05\n",
      "  coefs:  3.6378e-02   -2.1447e-02    1.6979e-04    9.5926e-01   -9.5714e-01   -3.2236e-04   -6.6765e-04  \n",
      "Epoch: 6850,  Loss u: 1.375e-05, Loss pde: 3.635e-06, Loss_norm: 1.975e+00, Loss tot: 4.495e-05\n",
      "  coefs:  3.7994e-02   -2.1558e-02    0.0000e+00    9.5725e-01   -9.5681e-01   -4.3771e-04   -4.8381e-04  \n",
      "Epoch: 6900,  Loss u: 1.383e-05, Loss pde: 3.703e-06, Loss_norm: 1.977e+00, Loss tot: 4.509e-05\n",
      "  coefs:  4.0125e-02   -2.3576e-02    7.9725e-06    9.5563e-01   -9.5692e-01   -5.8108e-04   -4.7757e-04  \n",
      "Epoch: 6950,  Loss u: 4.537e-05, Loss pde: 4.211e-06, Loss_norm: 2.013e+00, Loss tot: 7.835e-05\n",
      "  coefs:  5.4197e-02   -4.4816e-02    1.2945e-03    9.5095e-01   -9.6117e-01    3.9343e-04   -1.8370e-04  \n",
      "Epoch: 7000,  Loss u: 3.535e-05, Loss pde: 3.460e-06, Loss_norm: 2.013e+00, Loss tot: 6.586e-05\n",
      "  coefs:  6.2437e-02   -5.2146e-02   -8.6733e-04    9.3948e-01   -9.5681e-01    1.3395e-03   -0.0000e+00  \n",
      "Epoch: 7050,  Loss u: 2.624e-05, Loss pde: 3.266e-06, Loss_norm: 1.999e+00, Loss tot: 5.587e-05\n",
      "  coefs:  6.2432e-02   -4.8061e-02   -6.4173e-04    9.3521e-01   -9.5253e-01   -0.0000e+00    1.1811e-04  \n",
      "Epoch: 7100,  Loss u: 1.693e-05, Loss pde: 3.117e-06, Loss_norm: 1.985e+00, Loss tot: 4.582e-05\n",
      "  coefs:  5.7644e-02   -4.0269e-02   -1.4339e-03    9.3586e-01   -9.4930e-01   -0.0000e+00    1.8964e-04  \n",
      "Epoch: 7150,  Loss u: 1.563e-05, Loss pde: 3.129e-06, Loss_norm: 1.960e+00, Loss tot: 4.416e-05\n",
      "  coefs:  4.5204e-02   -2.4113e-02   -2.3964e-03    9.4185e-01   -9.4686e-01   -2.0131e-05    7.1390e-06  \n",
      "Epoch: 7200,  Loss u: 2.393e-05, Loss pde: 3.267e-06, Loss_norm: 1.947e+00, Loss tot: 5.255e-05\n",
      "  coefs:  3.2763e-02   -1.1213e-02   -1.2686e-03    9.5178e-01   -9.4906e-01   -6.0121e-04   -0.0000e+00  \n",
      "Epoch: 7250,  Loss u: 4.959e-05, Loss pde: 4.004e-06, Loss_norm: 1.971e+00, Loss tot: 8.031e-05\n",
      "  coefs:  3.1663e-02   -1.7245e-02    1.3422e-03    9.6165e-01   -9.5816e-01   -4.4962e-04    1.4665e-04  \n",
      "Epoch: 7300,  Loss u: 2.052e-05, Loss pde: 3.421e-06, Loss_norm: 2.001e+00, Loss tot: 4.977e-05\n",
      "  coefs:  4.8359e-02   -3.8556e-02   -6.4413e-04    9.5271e-01   -9.5814e-01    2.6921e-03   -4.8391e-05  \n",
      "Epoch: 7350,  Loss u: 1.294e-05, Loss pde: 3.107e-06, Loss_norm: 1.985e+00, Loss tot: 4.103e-05\n",
      "  coefs:  4.6343e-02   -3.2419e-02   -1.1277e-04    9.4928e-01   -9.5645e-01   -0.0000e+00    3.6950e-05  \n",
      "Epoch: 7400,  Loss u: 1.134e-05, Loss pde: 3.071e-06, Loss_norm: 1.981e+00, Loss tot: 3.914e-05\n",
      "  coefs:  4.4315e-02   -2.9682e-02   -1.3174e-04    9.5036e-01   -9.5655e-01   -0.0000e+00    4.8025e-07  \n",
      "Epoch: 7450,  Loss u: 1.034e-05, Loss pde: 3.068e-06, Loss_norm: 1.980e+00, Loss tot: 3.796e-05\n",
      "  coefs:  4.2226e-02   -2.7717e-02   -0.0000e+00    9.5236e-01   -9.5737e-01   -2.1682e-05    0.0000e+00  \n",
      "Epoch: 7500,  Loss u: 9.475e-06, Loss pde: 3.083e-06, Loss_norm: 1.977e+00, Loss tot: 3.695e-05\n",
      "  coefs:  3.9265e-02   -2.4486e-02   -7.2420e-06    9.5469e-01   -9.5820e-01   -1.6204e-04    0.0000e+00  \n",
      "Epoch: 7550,  Loss u: 1.880e-05, Loss pde: 3.346e-06, Loss_norm: 1.961e+00, Loss tot: 4.661e-05\n",
      "  coefs:  2.9796e-02   -1.2381e-02   -8.3438e-04    9.5896e-01   -9.5834e-01   -4.9886e-04   -1.2264e-04  \n",
      "Epoch: 7600,  Loss u: 1.281e-04, Loss pde: 4.420e-06, Loss_norm: 2.013e+00, Loss tot: 1.588e-04\n",
      "  coefs:  4.2149e-02   -3.3706e-02    4.0220e-03    9.6432e-01   -9.6815e-01   -9.6814e-04    1.2306e-04  \n",
      "Epoch: 7650,  Loss u: 6.595e-05, Loss pde: 4.108e-06, Loss_norm: 2.044e+00, Loss tot: 9.605e-05\n",
      "  coefs:  6.7827e-02   -5.8724e-02    1.4443e-03    9.4766e-01   -9.6712e-01    7.3695e-04   -2.3734e-04  \n",
      "Epoch: 7700,  Loss u: 3.957e-05, Loss pde: 4.040e-06, Loss_norm: 2.016e+00, Loss tot: 6.903e-05\n",
      "  coefs:  6.8219e-02   -5.1472e-02   -0.0000e+00    9.3732e-01   -9.5814e-01   -7.4098e-04    6.2967e-05  \n",
      "Epoch: 7750,  Loss u: 2.473e-05, Loss pde: 3.337e-06, Loss_norm: 2.002e+00, Loss tot: 5.226e-05\n",
      "  coefs:  6.4594e-02   -4.7533e-02   -3.9635e-04    9.3495e-01   -9.5297e-01   -4.9895e-04    5.6162e-04  \n",
      "Epoch: 7800,  Loss u: 1.522e-05, Loss pde: 3.027e-06, Loss_norm: 1.990e+00, Loss tot: 4.178e-05\n",
      "  coefs:  5.7741e-02   -4.0606e-02   -7.7699e-04    9.3859e-01   -9.5180e-01   -1.1534e-04    5.2824e-04  \n",
      "Epoch: 7850,  Loss u: 1.590e-05, Loss pde: 3.052e-06, Loss_norm: 1.968e+00, Loss tot: 4.214e-05\n",
      "  coefs:  4.3966e-02   -2.5605e-02   -1.7533e-03    9.4499e-01   -9.5194e-01   -9.0411e-05   -0.0000e+00  \n",
      "Epoch: 7900,  Loss u: 1.271e-05, Loss pde: 3.055e-06, Loss_norm: 1.970e+00, Loss tot: 3.883e-05\n",
      "  coefs:  3.4937e-02   -1.9292e-02    0.0000e+00    9.5684e-01   -9.5838e-01   -3.3688e-04    0.0000e+00  \n",
      "Epoch: 7950,  Loss u: 4.176e-05, Loss pde: 3.553e-06, Loss_norm: 1.993e+00, Loss tot: 6.898e-05\n",
      "  coefs:  3.7086e-02   -2.5659e-02    1.6420e-03    9.6273e-01   -9.6561e-01   -3.1744e-04   -7.4496e-05  \n",
      "Epoch: 8000,  Loss u: 1.787e-05, Loss pde: 3.438e-06, Loss_norm: 1.952e+00, Loss tot: 4.427e-05\n",
      "  coefs:  1.7316e-02   -1.5063e-03    0.0000e+00    9.6819e-01   -9.6272e-01   -1.7653e-03   -4.1510e-04  \n",
      "Epoch: 8050,  Loss u: 4.635e-05, Loss pde: 4.006e-06, Loss_norm: 2.042e+00, Loss tot: 7.458e-05\n",
      "  coefs:  5.3740e-02   -4.8100e-02    2.5906e-03    9.6205e-01   -9.7392e-01    1.0130e-03    3.3785e-04  \n",
      "Epoch: 8100,  Loss u: 1.728e-05, Loss pde: 3.234e-06, Loss_norm: 2.004e+00, Loss tot: 4.346e-05\n",
      "  coefs:  4.9241e-02   -3.7285e-02   -2.5203e-04    9.5191e-01   -9.6495e-01   -3.4469e-04    1.3055e-04  \n",
      "Epoch: 8150,  Loss u: 1.105e-05, Loss pde: 3.064e-06, Loss_norm: 1.994e+00, Loss tot: 3.665e-05\n",
      "  coefs:  4.5343e-02   -3.2204e-02   -4.2492e-04    9.5299e-01   -9.6244e-01   -9.4194e-05    2.2808e-04  \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 8200,  Loss u: 7.917e-06, Loss pde: 3.006e-06, Loss_norm: 1.983e+00, Loss tot: 3.316e-05\n",
      "  coefs:  3.8568e-02   -2.5150e-02   -5.5294e-04    9.5642e-01   -9.6235e-01   -1.2093e-04    0.0000e+00  \n",
      "Epoch: 8250,  Loss u: 6.016e-06, Loss pde: 3.028e-06, Loss_norm: 1.978e+00, Loss tot: 3.110e-05\n",
      "  coefs:  3.0836e-02   -1.8648e-02    0.0000e+00    9.6303e-01   -9.6572e-01   -2.1082e-04   -0.0000e+00  \n",
      "Epoch: 8300,  Loss u: 5.427e-06, Loss pde: 3.078e-06, Loss_norm: 1.983e+00, Loss tot: 3.049e-05\n",
      "  coefs:  2.8299e-02   -1.7175e-02    1.1756e-04    9.6799e-01   -9.6921e-01   -1.8351e-04   -7.1655e-05  \n",
      "Epoch: 8350,  Loss u: 4.950e-06, Loss pde: 3.098e-06, Loss_norm: 1.983e+00, Loss tot: 2.989e-05\n",
      "  coefs:  2.5518e-02   -1.5103e-02    0.0000e+00    9.7063e-01   -9.7141e-01   -1.8564e-04   -1.0125e-05  \n",
      "Epoch: 8400,  Loss u: 5.593e-05, Loss pde: 3.911e-06, Loss_norm: 1.973e+00, Loss tot: 8.192e-05\n",
      "  coefs:  2.1541e-02   -1.1631e-02   -4.7527e-05    9.7029e-01   -9.6929e-01   -0.0000e+00   -1.6782e-04  \n",
      "Epoch: 8450,  Loss u: 9.661e-06, Loss pde: 3.554e-06, Loss_norm: 1.977e+00, Loss tot: 3.494e-05\n",
      "  coefs:  2.0214e-02   -1.1882e-02   -1.1368e-03    9.7132e-01   -9.7167e-01    4.9228e-04   -2.3321e-04  \n",
      "Epoch: 8500,  Loss u: 7.851e-06, Loss pde: 3.166e-06, Loss_norm: 1.974e+00, Loss tot: 3.234e-05\n",
      "  coefs:  1.8423e-02   -8.9936e-03   -1.2183e-03    9.7249e-01   -9.7272e-01    0.0000e+00   -3.4713e-04  \n",
      "Epoch: 8550,  Loss u: 5.564e-06, Loss pde: 3.061e-06, Loss_norm: 1.984e+00, Loss tot: 2.984e-05\n",
      "  coefs:  1.8978e-02   -1.1178e-02   -4.1013e-04    9.7708e-01   -9.7602e-01    0.0000e+00    0.0000e+00  \n",
      "Epoch: 8600,  Loss u: 5.468e-06, Loss pde: 3.144e-06, Loss_norm: 1.986e+00, Loss tot: 2.973e-05\n",
      "  coefs:  1.8467e-02   -1.1931e-02   -0.0000e+00    9.7781e-01   -9.7772e-01    0.0000e+00   -1.6494e-04  \n",
      "Epoch: 8650,  Loss u: 4.193e-06, Loss pde: 3.055e-06, Loss_norm: 1.986e+00, Loss tot: 2.818e-05\n",
      "  coefs:  1.8014e-02   -1.1360e-02   -0.0000e+00    9.7807e-01   -9.7812e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 8700,  Loss u: 3.875e-05, Loss pde: 3.685e-06, Loss_norm: 2.004e+00, Loss tot: 6.358e-05\n",
      "  coefs:  2.7122e-02   -2.1675e-02    1.1771e-03    9.7533e-01   -9.7831e-01   -1.6602e-04   -0.0000e+00  \n",
      "Epoch: 8750,  Loss u: 1.426e-05, Loss pde: 3.376e-06, Loss_norm: 1.982e+00, Loss tot: 3.831e-05\n",
      "  coefs:  1.7115e-02   -8.2526e-03   -4.3889e-04    9.7860e-01   -9.7698e-01   -5.2101e-04    5.6087e-05  \n",
      "Epoch: 8800,  Loss u: 5.468e-06, Loss pde: 3.221e-06, Loss_norm: 1.989e+00, Loss tot: 2.922e-05\n",
      "  coefs:  1.7612e-02   -1.1473e-02    4.0350e-05    9.7941e-01   -9.8006e-01   -0.0000e+00   -7.9192e-05  \n",
      "Epoch: 8850,  Loss u: 5.036e-06, Loss pde: 3.207e-06, Loss_norm: 1.986e+00, Loss tot: 2.858e-05\n",
      "  coefs:  1.5353e-02   -8.0970e-03   -5.0989e-04    9.8100e-01   -9.8050e-01   -2.1496e-04    0.0000e+00  \n",
      "Epoch: 8900,  Loss u: 3.514e-06, Loss pde: 3.076e-06, Loss_norm: 1.988e+00, Loss tot: 2.678e-05\n",
      "  coefs:  1.5616e-02   -9.6662e-03    0.0000e+00    9.8136e-01   -9.8138e-01   -4.2347e-05   -0.0000e+00  \n",
      "Epoch: 8950,  Loss u: 5.776e-05, Loss pde: 4.302e-06, Loss_norm: 1.980e+00, Loss tot: 8.208e-05\n",
      "  coefs:  1.2179e-02   -5.5790e-03   -1.1077e-03    9.8098e-01   -9.7972e-01    2.8699e-04   -3.0419e-04  \n",
      "Epoch: 9000,  Loss u: 9.860e-06, Loss pde: 3.197e-06, Loss_norm: 1.993e+00, Loss tot: 3.299e-05\n",
      "  coefs:  1.6803e-02   -1.1776e-02   -0.0000e+00    9.8226e-01   -9.8235e-01    1.0640e-04   -4.5373e-05  \n",
      "Epoch: 9050,  Loss u: 3.822e-06, Loss pde: 3.131e-06, Loss_norm: 1.990e+00, Loss tot: 2.670e-05\n",
      "  coefs:  1.4523e-02   -8.9819e-03   -0.0000e+00    9.8339e-01   -9.8334e-01   -8.6498e-05    0.0000e+00  \n",
      "Epoch: 9100,  Loss u: 2.620e-06, Loss pde: 3.115e-06, Loss_norm: 1.990e+00, Loss tot: 2.533e-05\n",
      "  coefs:  1.2883e-02   -8.1288e-03   -0.0000e+00    9.8473e-01   -9.8452e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9150,  Loss u: 6.265e-05, Loss pde: 4.601e-06, Loss_norm: 1.973e+00, Loss tot: 8.630e-05\n",
      "  coefs:  4.7920e-03    1.2572e-03   -1.9812e-03    9.8328e-01   -9.8069e-01    1.0459e-03   -4.2717e-04  \n",
      "Epoch: 9200,  Loss u: 4.924e-06, Loss pde: 3.280e-06, Loss_norm: 1.992e+00, Loss tot: 2.747e-05\n",
      "  coefs:  9.4249e-03   -5.5923e-03    3.8055e-04    9.8904e-01   -9.8704e-01   -2.3550e-04    0.0000e+00  \n",
      "Epoch: 9250,  Loss u: 2.353e-06, Loss pde: 3.199e-06, Loss_norm: 1.993e+00, Loss tot: 2.468e-05\n",
      "  coefs:  1.0410e-02   -6.7058e-03   -7.3408e-05    9.8790e-01   -9.8784e-01    0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9300,  Loss u: 2.138e-06, Loss pde: 3.210e-06, Loss_norm: 1.993e+00, Loss tot: 2.431e-05\n",
      "  coefs:  9.8432e-03   -6.3263e-03   -0.0000e+00    9.8828e-01   -9.8815e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9350,  Loss u: 1.986e-06, Loss pde: 3.231e-06, Loss_norm: 1.993e+00, Loss tot: 2.402e-05\n",
      "  coefs:  9.3953e-03   -6.0611e-03   -0.0000e+00    9.8885e-01   -9.8870e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9400,  Loss u: 1.141e-05, Loss pde: 4.074e-06, Loss_norm: 1.977e+00, Loss tot: 3.363e-05\n",
      "  coefs:  5.8442e-03   -1.4245e-03   -1.6619e-04    9.8556e-01   -9.8399e-01    0.0000e+00   -1.2025e-04  \n",
      "Epoch: 9450,  Loss u: 2.184e-06, Loss pde: 3.353e-06, Loss_norm: 1.995e+00, Loss tot: 2.398e-05\n",
      "  coefs:  9.5025e-03   -6.2544e-03   -0.0000e+00    9.8945e-01   -9.8954e-01   -0.0000e+00    0.0000e+00  \n",
      "Epoch: 9500,  Loss u: 1.924e-06, Loss pde: 3.342e-06, Loss_norm: 1.994e+00, Loss tot: 2.354e-05\n",
      "  coefs:  8.3387e-03   -5.3698e-03   -0.0000e+00    9.9013e-01   -9.9008e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9550,  Loss u: 1.753e-06, Loss pde: 3.377e-06, Loss_norm: 1.994e+00, Loss tot: 2.322e-05\n",
      "  coefs:  7.5711e-03   -4.8906e-03   -0.0000e+00    9.9104e-01   -9.9098e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9600,  Loss u: 1.617e-06, Loss pde: 3.417e-06, Loss_norm: 1.995e+00, Loss tot: 2.294e-05\n",
      "  coefs:  6.8919e-03   -4.4783e-03   -0.0000e+00    9.9183e-01   -9.9174e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9650,  Loss u: 1.621e-06, Loss pde: 3.462e-06, Loss_norm: 1.995e+00, Loss tot: 2.279e-05\n",
      "  coefs:  6.1830e-03   -4.0311e-03   -0.0000e+00    9.9252e-01   -9.9241e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9700,  Loss u: 6.987e-06, Loss pde: 4.150e-06, Loss_norm: 2.001e+00, Loss tot: 2.825e-05\n",
      "  coefs:  1.0777e-02   -7.7831e-03   -0.0000e+00    9.9094e-01   -9.9126e-01   -0.0000e+00    2.2338e-04  \n",
      "Epoch: 9750,  Loss u: 1.677e-06, Loss pde: 3.705e-06, Loss_norm: 1.996e+00, Loss tot: 2.257e-05\n",
      "  coefs:  5.9244e-03   -3.7976e-03   -0.0000e+00    9.9308e-01   -9.9309e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9800,  Loss u: 1.478e-06, Loss pde: 3.721e-06, Loss_norm: 1.997e+00, Loss tot: 2.219e-05\n",
      "  coefs:  5.1047e-03   -3.3090e-03   -0.0000e+00    9.9406e-01   -9.9415e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9850,  Loss u: 1.347e-06, Loss pde: 3.795e-06, Loss_norm: 1.997e+00, Loss tot: 2.189e-05\n",
      "  coefs:  4.3338e-03   -2.8415e-03   -0.0000e+00    9.9494e-01   -9.9501e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9900,  Loss u: 1.242e-06, Loss pde: 3.899e-06, Loss_norm: 1.998e+00, Loss tot: 2.161e-05\n",
      "  coefs:  3.6201e-03   -2.3997e-03   -0.0000e+00    9.9578e-01   -9.9584e-01   -0.0000e+00   -0.0000e+00  \n",
      "Epoch: 9950,  Loss u: 1.154e-06, Loss pde: 4.038e-06, Loss_norm: 1.998e+00, Loss tot: 2.134e-05\n",
      "  coefs:  2.9463e-03   -1.9776e-03   -0.0000e+00    9.9661e-01   -9.9667e-01   -0.0000e+00   -0.0000e+00  \n"
     ]
    }
   ],
   "source": [
    "params = {'n_epochs':10000,\n",
    "          'alpha_pde_start':10,\n",
    "           'alpha_pde_end':0.001,\n",
    "           'alpha_l1':1e-5, \n",
    "          'warmup_nsteps':500,\n",
    "          'linearRegInterval':5,\n",
    "          'linearRegression':True,\n",
    "          'width':50,\n",
    "          'layers':8,\n",
    "          'lr':0.002,\n",
    "          'update_coef_in_dl':False,\n",
    "          'logfile':'alpha_decrease2.txt'}\n",
    "\n",
    "model = learning(inputs=inputs, u=u, dictionary=dictionary,**params)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "log = pd.read_csv('alpha_decrease2.txt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Epoch</th>\n",
       "      <th>Loss_u</th>\n",
       "      <th>Loss_pde</th>\n",
       "      <th>Loss_l1</th>\n",
       "      <th>Loss_tot</th>\n",
       "      <th>p0</th>\n",
       "      <th>p1</th>\n",
       "      <th>p2</th>\n",
       "      <th>p3</th>\n",
       "      <th>p4</th>\n",
       "      <th>p5</th>\n",
       "      <th>p6</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0.136880</td>\n",
       "      <td>0.000983</td>\n",
       "      <td>7.4879</td>\n",
       "      <td>0.136880</td>\n",
       "      <td>-0.56304</td>\n",
       "      <td>-2.897000</td>\n",
       "      <td>0.14370</td>\n",
       "      <td>-0.62471</td>\n",
       "      <td>-1.48660</td>\n",
       "      <td>1.350000</td>\n",
       "      <td>0.422880</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.129730</td>\n",
       "      <td>0.000562</td>\n",
       "      <td>7.4879</td>\n",
       "      <td>0.129730</td>\n",
       "      <td>-0.56304</td>\n",
       "      <td>-2.897000</td>\n",
       "      <td>0.14370</td>\n",
       "      <td>-0.62471</td>\n",
       "      <td>-1.48660</td>\n",
       "      <td>1.350000</td>\n",
       "      <td>0.422880</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0.118250</td>\n",
       "      <td>0.000711</td>\n",
       "      <td>7.4879</td>\n",
       "      <td>0.118250</td>\n",
       "      <td>-0.56304</td>\n",
       "      <td>-2.897000</td>\n",
       "      <td>0.14370</td>\n",
       "      <td>-0.62471</td>\n",
       "      <td>-1.48660</td>\n",
       "      <td>1.350000</td>\n",
       "      <td>0.422880</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0.098827</td>\n",
       "      <td>0.004477</td>\n",
       "      <td>7.4879</td>\n",
       "      <td>0.098827</td>\n",
       "      <td>-0.56304</td>\n",
       "      <td>-2.897000</td>\n",
       "      <td>0.14370</td>\n",
       "      <td>-0.62471</td>\n",
       "      <td>-1.48660</td>\n",
       "      <td>1.350000</td>\n",
       "      <td>0.422880</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0.072141</td>\n",
       "      <td>0.023201</td>\n",
       "      <td>7.4879</td>\n",
       "      <td>0.072141</td>\n",
       "      <td>-0.56304</td>\n",
       "      <td>-2.897000</td>\n",
       "      <td>0.14370</td>\n",
       "      <td>-0.62471</td>\n",
       "      <td>-1.48660</td>\n",
       "      <td>1.350000</td>\n",
       "      <td>0.422880</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9995</th>\n",
       "      <td>9995</td>\n",
       "      <td>0.000096</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>1.9905</td>\n",
       "      <td>0.000116</td>\n",
       "      <td>0.00021</td>\n",
       "      <td>0.002557</td>\n",
       "      <td>0.00171</td>\n",
       "      <td>0.99160</td>\n",
       "      <td>-0.99305</td>\n",
       "      <td>-0.001308</td>\n",
       "      <td>-0.000039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9996</th>\n",
       "      <td>9996</td>\n",
       "      <td>0.000005</td>\n",
       "      <td>0.000012</td>\n",
       "      <td>1.9905</td>\n",
       "      <td>0.000025</td>\n",
       "      <td>0.00021</td>\n",
       "      <td>0.002557</td>\n",
       "      <td>0.00171</td>\n",
       "      <td>0.99160</td>\n",
       "      <td>-0.99305</td>\n",
       "      <td>-0.001308</td>\n",
       "      <td>-0.000039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9997</th>\n",
       "      <td>9997</td>\n",
       "      <td>0.000054</td>\n",
       "      <td>0.000018</td>\n",
       "      <td>1.9905</td>\n",
       "      <td>0.000074</td>\n",
       "      <td>0.00021</td>\n",
       "      <td>0.002557</td>\n",
       "      <td>0.00171</td>\n",
       "      <td>0.99160</td>\n",
       "      <td>-0.99305</td>\n",
       "      <td>-0.001308</td>\n",
       "      <td>-0.000039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9998</th>\n",
       "      <td>9998</td>\n",
       "      <td>0.000143</td>\n",
       "      <td>0.000009</td>\n",
       "      <td>1.9905</td>\n",
       "      <td>0.000163</td>\n",
       "      <td>0.00021</td>\n",
       "      <td>0.002557</td>\n",
       "      <td>0.00171</td>\n",
       "      <td>0.99160</td>\n",
       "      <td>-0.99305</td>\n",
       "      <td>-0.001308</td>\n",
       "      <td>-0.000039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9999</th>\n",
       "      <td>9999</td>\n",
       "      <td>0.000130</td>\n",
       "      <td>0.000027</td>\n",
       "      <td>1.9905</td>\n",
       "      <td>0.000150</td>\n",
       "      <td>0.00021</td>\n",
       "      <td>0.002557</td>\n",
       "      <td>0.00171</td>\n",
       "      <td>0.99160</td>\n",
       "      <td>-0.99305</td>\n",
       "      <td>-0.001308</td>\n",
       "      <td>-0.000039</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10000 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Epoch    Loss_u  Loss_pde  Loss_l1  Loss_tot       p0        p1  \\\n",
       "0         0  0.136880  0.000983   7.4879  0.136880 -0.56304 -2.897000   \n",
       "1         1  0.129730  0.000562   7.4879  0.129730 -0.56304 -2.897000   \n",
       "2         2  0.118250  0.000711   7.4879  0.118250 -0.56304 -2.897000   \n",
       "3         3  0.098827  0.004477   7.4879  0.098827 -0.56304 -2.897000   \n",
       "4         4  0.072141  0.023201   7.4879  0.072141 -0.56304 -2.897000   \n",
       "...     ...       ...       ...      ...       ...      ...       ...   \n",
       "9995   9995  0.000096  0.000006   1.9905  0.000116  0.00021  0.002557   \n",
       "9996   9996  0.000005  0.000012   1.9905  0.000025  0.00021  0.002557   \n",
       "9997   9997  0.000054  0.000018   1.9905  0.000074  0.00021  0.002557   \n",
       "9998   9998  0.000143  0.000009   1.9905  0.000163  0.00021  0.002557   \n",
       "9999   9999  0.000130  0.000027   1.9905  0.000150  0.00021  0.002557   \n",
       "\n",
       "           p2       p3       p4        p5        p6  \n",
       "0     0.14370 -0.62471 -1.48660  1.350000  0.422880  \n",
       "1     0.14370 -0.62471 -1.48660  1.350000  0.422880  \n",
       "2     0.14370 -0.62471 -1.48660  1.350000  0.422880  \n",
       "3     0.14370 -0.62471 -1.48660  1.350000  0.422880  \n",
       "4     0.14370 -0.62471 -1.48660  1.350000  0.422880  \n",
       "...       ...      ...      ...       ...       ...  \n",
       "9995  0.00171  0.99160 -0.99305 -0.001308 -0.000039  \n",
       "9996  0.00171  0.99160 -0.99305 -0.001308 -0.000039  \n",
       "9997  0.00171  0.99160 -0.99305 -0.001308 -0.000039  \n",
       "9998  0.00171  0.99160 -0.99305 -0.001308 -0.000039  \n",
       "9999  0.00171  0.99160 -0.99305 -0.001308 -0.000039  \n",
       "\n",
       "[10000 rows x 12 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "log"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2b9b57223690>]"
      ]
     },
     "execution_count": 18,
     "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(log['Epoch'].values, log['p3'].values)\n",
    "plt.plot([0,10000],[1,1])\n",
    "plt.ylim(-1,1.2)\n",
    "plt.plot([0,10000],[-1,-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2b9b5729b0d0>]"
      ]
     },
     "execution_count": 19,
     "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(log['Epoch'], log['p4'])\n",
    "plt.plot([0,10000],[1,1])\n",
    "plt.ylim(-2,-0.4)\n",
    "plt.plot([0,10000],[-1,-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2b5caff68c10>]"
      ]
     },
     "execution_count": 213,
     "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.semilogy(stat['epoch'], stat['loss_u'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2b5d17cb8590>]"
      ]
     },
     "execution_count": 214,
     "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.semilogy(stat['epoch'], stat['loss_f'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2b5d7a777410>]"
      ]
     },
     "execution_count": 215,
     "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.semilogy(stat['epoch'], stat['loss_tot'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "t_mesh = dt*np.arange(nsteps)\n",
    "x_mesh = dx*np.arange(nx)\n",
    "y_mesh = dy*np.arange(ny)\n",
    "t_mesh, x_mesh,y_mesh = np.meshgrid(t_mesh, x_mesh, y_mesh,indexing='ij')\n",
    "t_mesh = np.reshape(t_mesh, (-1,1))\n",
    "x_mesh = np.reshape(x_mesh, (-1,1))\n",
    "y_mesh = np.reshape(y_mesh, (-1,1))\n",
    "x_mesh /= x_max\n",
    "y_mesh /= y_max\n",
    "t_mesh /= t_max\n",
    "prediction = model(torch.cat([torch.tensor(x_mesh), torch.tensor(y_mesh), torch.tensor(t_mesh)],dim=1).double().cuda(0))"
   ]
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}