diamond.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "# Learning a GNN potential for diamond\n",
    "\n",
    "In this notebook, we will learn a graph neural network (GNN) potential from experimental stiffness tensor data.\n",
    "We will directly apply the DiffTRe library. If you are interested in the working mechanisms of DiffTRe, please have a look at\n",
    "the double well notebook.\n",
    "\n",
    "We will define all components necessary to initialize DiffTRe: experimental data, simulation box, GNN potential with prior, simulator, loss function and optimizer.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from DiffTRe import io, custom_space, custom_energy, custom_quantity, difftre, custom_simulator\n",
    "from jax_md import partition, simulate, space\n",
    "import optax\n",
    "from jax import jit, random, device_get, checkpoint, tree_util\n",
    "import jax.numpy as jnp\n",
    "from functools import partial\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import time, pickle\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')  # disable warnings about float64 usage\n",
    "\n",
    "energy_pickle_file_path = 'output/saved_models/Diamond_GNN.pkl'\n",
    "optimization_pickle_file_path = 'output/saved_optimization_results/Diamond_optimization.pkl'\n",
    "long_traj_pickle_file_path = 'output/saved_optimization_results/Diamond_long_traj.pkl'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### Experimental observables\n",
    "\n",
    "It is often useful to start with the experimental target data. In this example, these are elastic stiffness tensor values from an\n",
    "experiment conducted in the paper _The Elastic Stiffness Moduli of Diamond_ [(McSkimin et. al., 1972)](https://doi.org/10.1063/1.1661318).\n",
    "Due to symmetries in cubic crystals, the stiffness tensor consists of three distinct stiffness moduli $c_{11}$,\n",
    "$c_{12}$ and $c_{44}$ (in Voigt notation).</br>\n",
    "In the un-strained state, we assume the crystal to be in a stress-free state.\n",
    "The second target observable is therefore an initial stress tensor of $\\mathbf{0}$."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "c_11_target = 1079.  # in GPa\n",
    "c_12_target = 124.\n",
    "c_44_target = 578.\n",
    "\n",
    "convert_from_GPa_to_kJ_mol_nm_3 = 10**3 / 1.66054\n",
    "stiffness_targets = jnp.array([c_11_target, c_12_target, c_44_target]) * convert_from_GPa_to_kJ_mol_nm_3\n",
    "\n",
    "target_dict = {'stress': jnp.zeros((3, 3)), 'stiffness': stiffness_targets}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### Simulation setup\n",
    "\n",
    "Next, we need to define the parameters of the simulation.\n",
    "Some of these parameters are already defined by the experiment, e.g. temperature and density of the material during\n",
    "the experiment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model Density: 3511.9998 kg/m^3. Experimental density: 3512 kg/m^3\n"
     ]
    }
   ],
   "source": [
    "system_temperature = 298.15  # Kelvin = 25 deg. celsius\n",
    "Boltzmann_constant = 0.0083145107  # in kJ / mol K\n",
    "kbT = system_temperature * Boltzmann_constant\n",
    "\n",
    "experimental_density = 3512  # kg / m^3\n",
    "mass = 12.011  # mass of carbon atoms in u\n",
    "\n",
    "file_loc = 'data/confs/Diamond.pdb'\n",
    "R_init, _, box = io.load_configuration(file_loc)  # initial configuration of a diamond crystal\n",
    "N = R_init.shape[0]\n",
    "density = mass * N * 1.66054 / jnp.prod(box)\n",
    "box *= (density / experimental_density) ** (1 / 3)  # adjust box size to experiment\n",
    "density = mass * N * 1.66054 / jnp.prod(box)\n",
    "print('Model Density:', density, 'kg/m^3. Experimental density:', experimental_density, 'kg/m^3')\n",
    "\n",
    "# simulation times in ps\n",
    "dt = 0.5e-3\n",
    "total_time = 70.\n",
    "t_equilib = 10.  # discard all states within the first 5 ps as equilibration\n",
    "print_every = 0.025  # save states every 0.1 ps for use in averaging\n",
    "timing_struct = custom_simulator.process_printouts(dt, total_time, t_equilib, print_every)\n",
    "\n",
    "# we need to take derivatives wrt. the simulation box to compute the stiffness tensor\n",
    "# we therefore use a differentiable box in fractal coordinates: all atom positions are mapped onto a unit cube\n",
    "R_init, box_tensor = custom_space.scale_to_fractional_coordinates(R_init, box)\n",
    "displacement, shift = space.periodic_general(box_tensor)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### GNN Potential\n",
    "\n",
    "Now we define the GNN as well as the prior potential functioning as an advantageous\n",
    "starting point for the optimization. As prior potential, we select the Stillinger-Weber potential originally developed\n",
    "for silicon and only roughly adjust the length scale $\\sigma$ and energy scale $\\epsilon$ such that it is compatible\n",
    "with diamond. After defining both potentials, we can combine them by simply adding computed\n",
    "potential energy values. This is done inside the `energy_fn_template` defining a combined `energy` function from the\n",
    "current GNN weights. This scheme enables updating GNN weights during the optimization without compromising\n",
    "\"jitability\". You can define a custom template by creating a function that takes current learnable weights and returns a\n",
    "corresponding `energy_fn`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "key = random.PRNGKey(0)  # define random seed for initialization of model and simulation\n",
    "model_init_key, simuation_init_key = random.split(key, 2)\n",
    "\n",
    "# prior parameters\n",
    "sigma = 0.14  # nm\n",
    "eps = 200  # kJ / mol\n",
    "r_cut_SW = sigma * 1.8  # a = 1.8 in original SW potential\n",
    "r_cut_NN = 0.2\n",
    "\n",
    "# we use the same neighbor list for prior and GNN to avoid keeping track of 2 separate lists\n",
    "r_cut_nbrs = max(r_cut_SW, r_cut_NN)\n",
    "box_nbrs = jnp.ones(3)  # give hypercube box as we are using fractional coordinates\n",
    "neighbor_fn = partition.neighbor_list(displacement, box_nbrs, r_cut_nbrs, dr_threshold=0.05, capacity_multiplier=1.5, \n",
    "                                     disable_cell_list=True)\n",
    "nbrs_init = neighbor_fn(R_init)\n",
    "\n",
    "init_fn, GNN_energy = custom_energy.DimeNetPP_neighborlist(displacement, R_init, nbrs_init, r_cut_NN)\n",
    "init_params = init_fn(model_init_key, R_init, neighbor=nbrs_init)\n",
    "\n",
    "prior_fn = custom_energy.stillinger_weber_neighborlist(displacement, cutoff=r_cut_SW, sigma=sigma, epsilon=eps,\n",
    "                                                        initialize_neighbor_list=False)\n",
    "\n",
    "\n",
    "def energy_fn_template(energy_params):\n",
    "    gnn_energy = partial(GNN_energy, energy_params)\n",
    "    def energy(R, neighbor, **dynamic_kwargs):\n",
    "        return gnn_energy(R, neighbor=neighbor, **dynamic_kwargs) + prior_fn(R, neighbor=neighbor, **dynamic_kwargs)\n",
    "    return jit(energy)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loss function\n",
    "After defining the `energy_fn_template`, we can define the loss function. We compute the stiffness tensor\n",
    "via the 'stress-fluctuation method', where multiple instantaneous quantities need to be combined:\n",
    "$$C_{ijkl} = \\langle C^B_{ijkl} \\rangle - \\frac{V}{k_B T}\\left(\\langle \\sigma^B_{ij} \\sigma^B_{kl} \\rangle -\n",
    "\\langle \\sigma^B_{ij} \\rangle \\langle \\sigma^B_{kl} \\rangle\\right) + \\frac{N k_B T}{V}\\left( \\delta_{ik} \\delta_{jl} +\n",
    "\\delta_{il} \\delta_{jk} \\right)$$\n",
    "$$\\mathrm{with} \\quad \\sigma^B_{ij} = \\frac{1}{V}\\frac{\\partial U}{\\partial \\epsilon_{ij}} \\ ; \\quad  C^B_{ijkl} =\n",
    "\\frac{1}{V}\\frac{\\partial^2 U}{\\partial \\epsilon_{ij} \\partial \\epsilon_{kl}}$$\n",
    "\n",
    "We therefore need to compute instantaneous values of stresses $\\sigma_{ij}$, $\\sigma^B_{ij}$ and $C^B_{ijkl}$ and combine them to observables $\\langle \\sigma_{ij} \\rangle$ and $C_{ijkl}$.\n",
    "Therefore, we cannot use the default loss function in `DiffTRe_init` because it is only applicable to\n",
    "observables that are directly computed as averages of instantaneous quantities.\n",
    "\n",
    "We can compute instantaneous quantities from a trajectory via a `quantity_dict`. It requires for each instantaneous observable $O_k(S_i, U_{\\theta})$ a function to compute the respective observable for each state $S_i$. \n",
    "In the end, we combine loss contributions from stress and stiffness tensors by multiplication of each component with a scale $\\gamma$, which is useful to counteract the effect of different units / orders of magnitude of observables.\n",
    "\n",
    "When constructing a custom loss function, we need to adhere to the function signature required by the implementation of DiffTRe:\n",
    "`Loss_fn` takes as input the trajectory of computed instantaneous quantities saved in a dict under its respective key of the `quantitiy_dict`. Additionally, it receives corresponding weights $w_i$ to compute averages under the reweighting scheme. With these components, target observables can be computed.\n",
    "The output of the function is a tuple `(loss value, predicted ensemble averages)`. The latter is only necessary\n",
    "for post-processing of the optimization process. `loss_fn` below provides an example implementation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# initialize observable functions\n",
    "stress_fn = custom_quantity.init_virial_stress_tensor(energy_fn_template, box_tensor)\n",
    "\n",
    "born_stiffness_fn, sigma_born, sigma_tensor_prod, stiffness_tensor_fn = \\\n",
    "    custom_quantity.init_stiffness_tensor_stress_fluctuation(energy_fn_template, box_tensor, kbT, N)\n",
    "\n",
    "# add all functions that compute instantaneous values into quantity_dict\n",
    "# DiffTRe computes correspondig quantities for each state in the (sparse) trajectory\n",
    "# checkpoint is needed to avoid out-of-memory errors for expensive models, e.g.  GNNs\n",
    "quantity_dict = {}\n",
    "quantity_dict['stress'] = {'compute_fn': checkpoint(stress_fn)}\n",
    "quantity_dict['born_stiffness'] = {'compute_fn': checkpoint(born_stiffness_fn)}\n",
    "quantity_dict['born_stress'] = {'compute_fn': checkpoint(sigma_born)}\n",
    "\n",
    "gamma_stress = 5.e-8\n",
    "gamma_stiffness = 1.e-10\n",
    "\n",
    "def loss_fn(quantity_trajs, weights):\n",
    "    \n",
    "    def reweighting_average(quantity_snapshots):\n",
    "        weighted_snapshots = (quantity_snapshots.T * weights).T\n",
    "        return jnp.sum(weighted_snapshots, axis=0)\n",
    "    \n",
    "    # compute all contributions to stiffness tensor\n",
    "    born_stress_prod_snapshots = sigma_tensor_prod(quantity_trajs['born_stress'])\n",
    "    stress_product_born_mean = reweighting_average(born_stress_prod_snapshots)  # <sigma^B_ij sigma^B_kl>\n",
    "    born_stress_tensor_mean = reweighting_average(quantity_trajs['born_stress'])\n",
    "    born_stiffness_mean = reweighting_average(quantity_trajs['born_stiffness'])\n",
    "    stiffness_tensor = stiffness_tensor_fn(born_stiffness_mean, born_stress_tensor_mean, stress_product_born_mean)\n",
    "    \n",
    "    stiffness_pred = custom_quantity.stiffness_tensor_components_cubic_crystal(stiffness_tensor)\n",
    "    stress_tensor_mean = reweighting_average(quantity_trajs['stress'])\n",
    "    \n",
    "    loss = gamma_stress * difftre.mse_loss(stress_tensor_mean, target_dict['stress']) + \\\n",
    "           gamma_stiffness * difftre.mse_loss(stiffness_pred, target_dict['stiffness'])\n",
    "    \n",
    "    \n",
    "    predictions = {'stress': stress_tensor_mean, 'stiffness': stiffness_pred}\n",
    "    \n",
    "    return loss, predictions\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### Simulator\n",
    "\n",
    "Let's define a `simulator_template` taking the current potential energy function from the `energy_fn_template` as input\n",
    "and returning a NVT simulator for the current energy function. This is achieved by simply \"baking in\" all constant\n",
    "simulation parameters via `partial`. For this example, we select a Langevin thermostat.\n",
    "With a given simulator, we can generate the initial simulation state from which the first trajectory is sampled."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "energy_fn_init = energy_fn_template(init_params)\n",
    "simulator_template = partial(custom_simulator.nvt_langevin, shift=shift, dt=dt, kT=kbT, gamma=4.)\n",
    "init, _ = simulator_template(energy_fn_init)\n",
    "state = init(simuation_init_key, R_init, mass=mass, neighbor=nbrs_init)\n",
    "init_sim_state = (state, nbrs_init)  # store neighbor list together with current simulation state"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### Optimizer\n",
    "\n",
    "The last missing component for DiffTRe is the optimizer. Here, we select an Adam optimizer with exponential learning\n",
    "rate decay."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "num_updates = 700\n",
    "initial_step_size = 0.002\n",
    "lr_schedule = optax.exponential_decay(-initial_step_size, 500, 0.01)\n",
    "optimizer = optax.chain(\n",
    "    optax.scale_by_adam(0.1, 0.4),\n",
    "    optax.scale_by_schedule(lr_schedule)\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### Initialize DiffTRe\n",
    "\n",
    "With all individual components defined, we can initialize DiffTRe. `DiffTRe_init` defines the `update_fn` and generates the initial reference trajectory starting from the initial state defined above. `update_fn` can be called to update the potential\n",
    "parameters via a single optimizer step. `update_fn` handles the whole DiffTRe pipeline internally, including\n",
    "trajectory generation whenever necessary, reweighting, computing the gradient, updating weights and outputting\n",
    "predictions at each optimization step for post-processing."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using custom loss function. Ignoring 'gamma' and 'target' in  \"quantities\".\n",
      "Time for a single trajectory generation: 7.887696393330892 mins\n"
     ]
    }
   ],
   "source": [
    "simulation_funs = (simulator_template, energy_fn_template, neighbor_fn)\n",
    "update_fn, trajectory_state = difftre.DiffTRe_init(simulation_funs, timing_struct, quantity_dict, \n",
    "                                                   kbT, init_params, init_sim_state, optimizer, loss_fn=loss_fn)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### Learning the GNN potential\n",
    "\n",
    "Now we can call `update_fn` iteratively to update the GNN weights to learn a GNN potential for diamond. We save intermediate results to visualize the\n",
    "learning process. </br>\n",
    "This optimization takes some time. You can skip it and load our trained model in the next cell."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Step 0 in 756.67 sec Loss =  8.139043 \n",
      "\n",
      "Step 1 in 814.52 sec Loss =  7.2388716 \n",
      "\n",
      "Step 2 in 807.51 sec Loss =  5.9584727 \n",
      "\n",
      "Step 3 in 813.94 sec Loss =  5.4823027 \n",
      "\n",
      "Step 4 in 811.10 sec Loss =  5.9287357 \n",
      "\n",
      "Step 5 in 811.66 sec Loss =  4.87702 \n",
      "\n",
      "Step 6 in 812.32 sec Loss =  4.50622 \n",
      "\n",
      "Step 7 in 810.79 sec Loss =  4.783046 \n",
      "\n",
      "Step 8 in 811.81 sec Loss =  4.6467557 \n",
      "\n",
      "Step 9 in 817.50 sec Loss =  5.0279284 \n",
      "\n",
      "Step 10 in 807.80 sec Loss =  4.852529 \n",
      "\n",
      "Step 11 in 815.73 sec Loss =  4.4231853 \n",
      "\n",
      "Step 12 in 816.60 sec Loss =  4.518403 \n",
      "\n",
      "Step 13 in 816.93 sec Loss =  4.472864 \n",
      "\n",
      "Step 14 in 807.81 sec Loss =  4.5960045 \n",
      "\n",
      "Step 15 in 811.00 sec Loss =  4.976228 \n",
      "\n",
      "Step 16 in 807.98 sec Loss =  4.5518923 \n",
      "\n",
      "Step 17 in 804.64 sec Loss =  4.419735 \n",
      "\n",
      "Step 18 in 812.28 sec Loss =  4.632913 \n",
      "\n",
      "Step 19 in 813.31 sec Loss =  4.434315 \n",
      "\n",
      "Step 20 in 811.55 sec Loss =  4.8194156 \n",
      "\n",
      "Step 21 in 813.04 sec Loss =  4.4227962 \n",
      "\n",
      "Step 22 in 811.35 sec Loss =  4.533052 \n",
      "\n",
      "Step 23 in 810.89 sec Loss =  4.502104 \n",
      "\n",
      "Step 24 in 810.22 sec Loss =  4.576209 \n",
      "\n",
      "Step 25 in 814.05 sec Loss =  4.103753 \n",
      "\n",
      "Step 26 in 810.22 sec Loss =  4.310746 \n",
      "\n",
      "Step 27 in 815.79 sec Loss =  4.6321707 \n",
      "\n",
      "Step 28 in 812.16 sec Loss =  4.5009255 \n",
      "\n",
      "Step 29 in 814.52 sec Loss =  4.3878074 \n",
      "\n",
      "Step 30 in 810.82 sec Loss =  4.324175 \n",
      "\n",
      "Step 31 in 811.50 sec Loss =  4.2517767 \n",
      "\n",
      "Step 32 in 814.76 sec Loss =  4.292004 \n",
      "\n",
      "Step 33 in 813.43 sec Loss =  4.21386 \n",
      "\n",
      "Step 34 in 806.51 sec Loss =  4.476225 \n",
      "\n",
      "Step 35 in 814.66 sec Loss =  4.4092493 \n",
      "\n",
      "Step 36 in 802.20 sec Loss =  4.231749 \n",
      "\n",
      "Step 37 in 813.22 sec Loss =  4.2225757 \n",
      "\n",
      "Step 38 in 811.75 sec Loss =  4.260197 \n",
      "\n",
      "Step 39 in 812.81 sec Loss =  4.1365914 \n",
      "\n",
      "Step 40 in 811.94 sec Loss =  4.0742335 \n",
      "\n",
      "Step 41 in 813.19 sec Loss =  4.1706796 \n",
      "\n",
      "Step 42 in 811.36 sec Loss =  4.257217 \n",
      "\n",
      "Step 43 in 811.28 sec Loss =  3.9575782 \n",
      "\n",
      "Step 44 in 810.17 sec Loss =  3.6062124 \n",
      "\n",
      "Step 45 in 811.40 sec Loss =  3.1852903 \n",
      "\n",
      "Step 46 in 811.42 sec Loss =  3.5170825 \n",
      "\n",
      "Step 47 in 810.36 sec Loss =  7.233748 \n",
      "\n",
      "Step 48 in 811.12 sec Loss =  3.1053462 \n",
      "\n",
      "Step 49 in 809.13 sec Loss =  3.6752365 \n",
      "\n",
      "Step 50 in 800.92 sec Loss =  3.2473123 \n",
      "\n",
      "Step 51 in 810.31 sec Loss =  3.0568283 \n",
      "\n",
      "Step 52 in 810.40 sec Loss =  3.048824 \n",
      "\n",
      "Step 53 in 811.67 sec Loss =  3.0881252 \n",
      "\n",
      "Step 54 in 810.43 sec Loss =  4.2268147 \n",
      "\n",
      "Step 55 in 811.32 sec Loss =  3.11845 \n",
      "\n",
      "Step 56 in 814.59 sec Loss =  2.9532542 \n",
      "\n",
      "Step 57 in 812.97 sec Loss =  2.569557 \n",
      "\n",
      "Step 58 in 808.19 sec Loss =  4.4822764 \n",
      "\n",
      "Step 59 in 810.07 sec Loss =  2.7716558 \n",
      "\n",
      "Step 60 in 809.27 sec Loss =  2.2149909 \n",
      "\n",
      "Step 61 in 808.70 sec Loss =  2.5894442 \n",
      "\n",
      "Step 62 in 801.75 sec Loss =  3.6035476 \n",
      "\n",
      "Step 63 in 799.52 sec Loss =  2.9360747 \n",
      "\n",
      "Step 64 in 800.69 sec Loss =  2.2498329 \n",
      "\n",
      "Step 65 in 810.24 sec Loss =  1.9332138 \n",
      "\n",
      "Step 66 in 810.36 sec Loss =  1.894294 \n",
      "\n",
      "Step 67 in 808.81 sec Loss =  5.2549677 \n",
      "\n",
      "Step 68 in 802.11 sec Loss =  2.2553713 \n",
      "\n",
      "Step 69 in 804.89 sec Loss =  1.7777412 \n",
      "\n",
      "Step 70 in 809.50 sec Loss =  2.1236405 \n",
      "\n",
      "Step 71 in 811.35 sec Loss =  3.2477236 \n",
      "\n",
      "Step 72 in 807.70 sec Loss =  2.1186037 \n",
      "\n",
      "Step 73 in 811.99 sec Loss =  1.8738232 \n",
      "\n",
      "Step 74 in 809.35 sec Loss =  1.6649158 \n",
      "\n",
      "Step 75 in 810.76 sec Loss =  1.7348868 \n",
      "\n",
      "Step 76 in 808.13 sec Loss =  5.95714 \n",
      "\n",
      "Step 77 in 808.46 sec Loss =  1.6045856 \n",
      "\n",
      "Step 78 in 808.18 sec Loss =  1.8891168 \n",
      "\n",
      "Step 79 in 809.23 sec Loss =  2.8973267 \n",
      "\n",
      "Step 80 in 813.20 sec Loss =  1.6634332 \n",
      "\n",
      "Step 81 in 809.18 sec Loss =  1.6585845 \n",
      "\n",
      "Step 82 in 812.20 sec Loss =  1.8185482 \n",
      "\n",
      "Step 83 in 807.56 sec Loss =  3.255161 \n",
      "\n",
      "Step 84 in 807.85 sec Loss =  1.7099329 \n",
      "\n",
      "Step 85 in 811.32 sec Loss =  1.5613618 \n",
      "\n",
      "Step 86 in 809.82 sec Loss =  1.7984 \n",
      "\n",
      "Step 87 in 809.31 sec Loss =  1.9572549 \n",
      "\n",
      "Step 88 in 809.16 sec Loss =  2.1043005 \n",
      "\n",
      "Step 89 in 810.17 sec Loss =  1.8324776 \n",
      "\n",
      "Step 90 in 810.13 sec Loss =  1.7281823 \n",
      "\n",
      "Step 91 in 799.91 sec Loss =  1.5633512 \n",
      "\n",
      "Step 92 in 812.52 sec Loss =  2.4105449 \n",
      "\n",
      "Step 93 in 811.73 sec Loss =  1.6620413 \n",
      "\n",
      "Step 94 in 800.21 sec Loss =  1.5722971 \n",
      "\n",
      "Step 95 in 798.95 sec Loss =  1.4447557 \n",
      "\n",
      "Step 96 in 800.47 sec Loss =  1.7995977 \n",
      "\n",
      "Step 97 in 798.95 sec Loss =  2.3756726 \n",
      "\n",
      "Step 98 in 799.77 sec Loss =  1.6422353 \n",
      "\n",
      "Step 99 in 814.14 sec Loss =  1.2689725 \n",
      "\n",
      "Step 100 in 816.07 sec Loss =  1.3296047 \n",
      "\n",
      "Step 101 in 803.94 sec Loss =  1.4096031 \n",
      "\n",
      "Step 102 in 817.06 sec Loss =  3.3303976 \n",
      "\n",
      "Step 103 in 806.62 sec Loss =  1.4143269 \n",
      "\n",
      "Step 104 in 816.93 sec Loss =  1.2214224 \n",
      "\n",
      "Step 105 in 813.01 sec Loss =  1.2032791 \n",
      "\n",
      "Step 106 in 811.85 sec Loss =  1.3317786 \n",
      "\n",
      "Step 107 in 809.73 sec Loss =  1.0923284 \n",
      "\n",
      "Step 108 in 814.30 sec Loss =  2.5167146 \n",
      "\n",
      "Step 109 in 814.21 sec Loss =  1.4753189 \n",
      "\n",
      "Step 110 in 816.93 sec Loss =  1.2693405 \n",
      "\n",
      "Step 111 in 812.75 sec Loss =  1.0478408 \n",
      "\n",
      "Step 112 in 813.05 sec Loss =  2.1628356 \n",
      "\n",
      "Step 113 in 818.15 sec Loss =  1.2480483 \n",
      "\n",
      "Step 114 in 812.42 sec Loss =  1.0782292 \n",
      "\n",
      "Step 115 in 815.43 sec Loss =  1.3324782 \n",
      "\n",
      "Step 116 in 813.92 sec Loss =  1.6451356 \n",
      "\n",
      "Step 117 in 812.57 sec Loss =  1.4487369 \n",
      "\n",
      "Step 118 in 815.09 sec Loss =  1.2436303 \n",
      "\n",
      "Step 119 in 812.87 sec Loss =  0.78863037 \n",
      "\n",
      "Step 120 in 804.99 sec Loss =  0.5864739 \n",
      "\n",
      "Step 121 in 813.04 sec Loss =  0.71333724 \n",
      "\n",
      "Step 122 in 811.76 sec Loss =  2.1325867 \n",
      "\n",
      "Step 123 in 815.60 sec Loss =  1.2425417 \n",
      "\n",
      "Step 124 in 812.82 sec Loss =  0.33692324 \n",
      "\n",
      "Step 125 in 814.27 sec Loss =  1.7145358 \n",
      "\n",
      "Step 126 in 813.50 sec Loss =  0.5659748 \n",
      "\n",
      "Step 127 in 807.80 sec Loss =  0.8387066 \n",
      "\n",
      "Step 128 in 804.73 sec Loss =  2.305757 \n",
      "\n",
      "Step 129 in 812.58 sec Loss =  1.0783179 \n",
      "\n",
      "Step 130 in 816.16 sec Loss =  0.9268168 \n",
      "\n",
      "Step 131 in 810.54 sec Loss =  0.60262406 \n",
      "\n",
      "Step 132 in 813.26 sec Loss =  0.6571957 \n",
      "\n",
      "Step 133 in 805.48 sec Loss =  1.0533885 \n",
      "\n",
      "Step 134 in 812.44 sec Loss =  1.0392909 \n",
      "\n",
      "Step 135 in 809.88 sec Loss =  0.83004177 \n",
      "\n",
      "Step 136 in 813.30 sec Loss =  1.5562885 \n",
      "\n",
      "Step 137 in 810.73 sec Loss =  0.91223526 \n",
      "\n",
      "Step 138 in 810.39 sec Loss =  0.706782 \n",
      "\n",
      "Step 139 in 811.77 sec Loss =  0.7141918 \n",
      "\n",
      "Step 140 in 810.63 sec Loss =  0.56833094 \n",
      "\n",
      "Step 141 in 810.68 sec Loss =  1.0488358 \n",
      "\n",
      "Step 142 in 811.10 sec Loss =  0.9985217 \n",
      "\n",
      "Step 143 in 809.80 sec Loss =  0.76327044 \n",
      "\n",
      "Step 144 in 810.14 sec Loss =  0.9175396 \n",
      "\n",
      "Step 145 in 809.23 sec Loss =  0.9113608 \n",
      "\n",
      "Step 146 in 811.38 sec Loss =  0.71112144 \n",
      "\n",
      "Step 147 in 812.03 sec Loss =  0.4142552 \n",
      "\n",
      "Step 148 in 809.76 sec Loss =  0.2733998 \n",
      "\n",
      "Step 149 in 814.37 sec Loss =  1.1980814 \n",
      "\n",
      "Step 150 in 814.71 sec Loss =  0.64929163 \n",
      "\n",
      "Step 151 in 810.62 sec Loss =  0.5088478 \n",
      "\n",
      "Step 152 in 810.68 sec Loss =  0.37908915 \n",
      "\n",
      "Step 153 in 808.30 sec Loss =  0.4678311 \n",
      "\n",
      "Step 154 in 811.49 sec Loss =  1.1628175 \n",
      "\n",
      "Step 155 in 810.90 sec Loss =  0.63874954 \n",
      "\n",
      "Step 156 in 811.48 sec Loss =  0.45262963 \n",
      "\n",
      "Step 157 in 810.90 sec Loss =  0.65261996 \n",
      "\n",
      "Step 158 in 812.76 sec Loss =  0.5990062 \n",
      "\n",
      "Step 159 in 809.12 sec Loss =  0.711463 \n",
      "\n",
      "Step 160 in 808.63 sec Loss =  0.6998807 \n",
      "\n",
      "Step 161 in 810.70 sec Loss =  0.62127465 \n",
      "\n",
      "Step 162 in 811.05 sec Loss =  0.37184352 \n",
      "\n",
      "Step 163 in 813.66 sec Loss =  0.4626568 \n",
      "\n",
      "Step 164 in 812.05 sec Loss =  0.6551612 \n",
      "\n",
      "Step 165 in 807.53 sec Loss =  0.6442493 \n",
      "\n",
      "Step 166 in 809.19 sec Loss =  0.4310265 \n",
      "\n",
      "Step 167 in 810.78 sec Loss =  0.50313455 \n",
      "\n",
      "Step 168 in 810.72 sec Loss =  0.32183462 \n",
      "\n",
      "Step 169 in 809.68 sec Loss =  0.6342432 \n",
      "\n",
      "Step 170 in 812.19 sec Loss =  0.5316545 \n",
      "\n",
      "Step 171 in 810.65 sec Loss =  0.3682515 \n",
      "\n",
      "Step 172 in 810.69 sec Loss =  0.41897961 \n",
      "\n",
      "Step 173 in 813.17 sec Loss =  0.22258815 \n",
      "\n",
      "Step 174 in 810.09 sec Loss =  0.2524571 \n",
      "\n",
      "Step 175 in 810.55 sec Loss =  1.0428222 \n",
      "\n",
      "Step 176 in 812.96 sec Loss =  0.49177262 \n",
      "\n",
      "Step 177 in 810.45 sec Loss =  0.52423614 \n",
      "\n",
      "Step 178 in 811.77 sec Loss =  0.21188867 \n",
      "\n",
      "Step 179 in 813.39 sec Loss =  0.3284093 \n",
      "\n",
      "Step 180 in 813.66 sec Loss =  0.6169038 \n",
      "\n",
      "Step 181 in 813.74 sec Loss =  0.6536175 \n",
      "\n",
      "Step 182 in 813.36 sec Loss =  0.21686032 \n",
      "\n",
      "Step 183 in 811.06 sec Loss =  0.19160938 \n",
      "\n",
      "Step 184 in 800.69 sec Loss =  0.9048075 \n",
      "\n",
      "Step 185 in 813.51 sec Loss =  0.437994 \n",
      "\n",
      "Step 186 in 810.34 sec Loss =  0.269091 \n",
      "\n",
      "Step 187 in 810.91 sec Loss =  0.08298595 \n",
      "\n",
      "Step 188 in 810.90 sec Loss =  0.227147 \n",
      "\n",
      "Step 189 in 810.28 sec Loss =  0.7089037 \n",
      "\n",
      "Step 190 in 813.47 sec Loss =  0.7136225 \n",
      "\n",
      "Step 191 in 813.80 sec Loss =  0.19137692 \n",
      "\n",
      "Step 192 in 816.87 sec Loss =  0.13383655 \n",
      "\n",
      "Step 193 in 816.01 sec Loss =  0.07783808 \n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Step 194 in 811.97 sec Loss =  0.23979855 \n",
      "\n",
      "Step 195 in 812.98 sec Loss =  0.8937613 \n",
      "\n",
      "Step 196 in 814.34 sec Loss =  0.22204863 \n",
      "\n",
      "Step 197 in 813.07 sec Loss =  0.110779524 \n",
      "\n",
      "Step 198 in 812.21 sec Loss =  0.32804453 \n",
      "\n",
      "Step 199 in 813.33 sec Loss =  0.57818943 \n",
      "\n",
      "Step 200 in 810.98 sec Loss =  0.2512654 \n",
      "\n",
      "Step 201 in 810.37 sec Loss =  0.2712513 \n",
      "\n",
      "Step 202 in 811.25 sec Loss =  0.32985651 \n",
      "\n",
      "Step 203 in 810.48 sec Loss =  0.23214789 \n",
      "\n",
      "Step 204 in 810.19 sec Loss =  0.25845745 \n",
      "\n",
      "Step 205 in 812.12 sec Loss =  0.29916364 \n",
      "\n",
      "Step 206 in 819.26 sec Loss =  0.37343884 \n",
      "\n",
      "Step 207 in 814.46 sec Loss =  0.22290358 \n",
      "\n",
      "Step 208 in 812.84 sec Loss =  0.24401449 \n",
      "\n",
      "Step 209 in 813.03 sec Loss =  0.23336855 \n",
      "\n",
      "Step 210 in 805.77 sec Loss =  0.4090982 \n",
      "\n",
      "Step 211 in 818.26 sec Loss =  0.1580763 \n",
      "\n",
      "Step 212 in 815.27 sec Loss =  0.21757708 \n",
      "\n",
      "Step 213 in 811.32 sec Loss =  0.24474171 \n",
      "\n",
      "Step 214 in 814.27 sec Loss =  0.21282014 \n",
      "\n",
      "Step 215 in 810.99 sec Loss =  0.2793643 \n",
      "\n",
      "Step 216 in 818.76 sec Loss =  0.25849915 \n",
      "\n",
      "Step 217 in 816.44 sec Loss =  0.20898508 \n",
      "\n",
      "Step 218 in 816.32 sec Loss =  0.09201809 \n",
      "\n",
      "Step 219 in 815.78 sec Loss =  0.20035931 \n",
      "\n",
      "Step 220 in 821.22 sec Loss =  0.25513092 \n",
      "\n",
      "Step 221 in 817.05 sec Loss =  0.19281119 \n",
      "\n",
      "Step 222 in 807.34 sec Loss =  0.19293861 \n",
      "\n",
      "Step 223 in 804.48 sec Loss =  0.16144496 \n",
      "\n",
      "Step 224 in 803.95 sec Loss =  0.09911586 \n",
      "\n",
      "Step 225 in 814.42 sec Loss =  0.12063274 \n",
      "\n",
      "Step 226 in 815.01 sec Loss =  0.18080994 \n",
      "\n",
      "Step 227 in 816.38 sec Loss =  0.27694356 \n",
      "\n",
      "Step 228 in 815.57 sec Loss =  0.19006042 \n",
      "\n",
      "Step 229 in 815.48 sec Loss =  0.22312477 \n",
      "\n",
      "Step 230 in 819.87 sec Loss =  0.09888011 \n",
      "\n",
      "Step 231 in 819.15 sec Loss =  0.093121655 \n",
      "\n",
      "Step 232 in 818.30 sec Loss =  0.109954536 \n",
      "\n",
      "Step 233 in 823.47 sec Loss =  0.3711906 \n",
      "\n",
      "Step 234 in 816.73 sec Loss =  0.11181969 \n",
      "\n",
      "Step 235 in 818.56 sec Loss =  0.07219711 \n",
      "\n",
      "Step 236 in 812.00 sec Loss =  0.119260564 \n",
      "\n",
      "Step 237 in 816.25 sec Loss =  0.15213446 \n",
      "\n",
      "Step 238 in 814.47 sec Loss =  0.2370423 \n",
      "\n",
      "Step 239 in 816.33 sec Loss =  0.23819512 \n",
      "\n",
      "Step 240 in 816.59 sec Loss =  0.09635575 \n",
      "\n",
      "Step 241 in 816.65 sec Loss =  0.101699926 \n",
      "\n",
      "Step 242 in 815.53 sec Loss =  0.10260765 \n",
      "\n",
      "Step 243 in 822.28 sec Loss =  0.19187334 \n",
      "\n",
      "Step 244 in 814.54 sec Loss =  0.14379553 \n",
      "\n",
      "Step 245 in 807.51 sec Loss =  0.059320044 \n",
      "\n",
      "Step 246 in 812.30 sec Loss =  0.10872312 \n",
      "\n",
      "Step 247 in 812.39 sec Loss =  0.14123932 \n",
      "\n",
      "Step 248 in 812.46 sec Loss =  0.19482651 \n",
      "\n",
      "Step 249 in 812.22 sec Loss =  0.18889341 \n",
      "\n",
      "Step 250 in 812.13 sec Loss =  0.043066226 \n",
      "\n",
      "Step 251 in 805.83 sec Loss =  0.18802433 \n",
      "\n",
      "Step 252 in 802.08 sec Loss =  0.12948301 \n",
      "\n",
      "Step 253 in 807.73 sec Loss =  0.22560701 \n",
      "\n",
      "Step 254 in 810.51 sec Loss =  0.007714843 \n",
      "\n",
      "Step 255 in 812.29 sec Loss =  0.024746137 \n",
      "\n",
      "Step 256 in 811.76 sec Loss =  0.16825473 \n",
      "\n",
      "Step 257 in 807.57 sec Loss =  0.47941196 \n",
      "\n",
      "Step 258 in 814.20 sec Loss =  0.049219687 \n",
      "\n",
      "Step 259 in 813.79 sec Loss =  0.07292671 \n",
      "\n",
      "Step 260 in 813.84 sec Loss =  0.07790531 \n",
      "\n",
      "Step 261 in 814.18 sec Loss =  0.16747248 \n",
      "\n",
      "Step 262 in 814.03 sec Loss =  0.14848706 \n",
      "\n",
      "Step 263 in 814.70 sec Loss =  0.04549484 \n",
      "\n",
      "Step 264 in 814.21 sec Loss =  0.05143469 \n",
      "\n",
      "Step 265 in 810.92 sec Loss =  0.11255021 \n",
      "\n",
      "Step 266 in 811.54 sec Loss =  0.2909665 \n",
      "\n",
      "Step 267 in 812.33 sec Loss =  0.17370567 \n",
      "\n",
      "Step 268 in 811.91 sec Loss =  0.36034518 \n",
      "\n",
      "Step 269 in 814.81 sec Loss =  0.081326865 \n",
      "\n",
      "Step 270 in 813.50 sec Loss =  0.033472583 \n",
      "\n",
      "Step 271 in 811.82 sec Loss =  0.20935667 \n",
      "\n",
      "Step 272 in 811.49 sec Loss =  0.014155017 \n",
      "\n",
      "Step 273 in 821.45 sec Loss =  0.020650445 \n",
      "\n",
      "Step 274 in 811.44 sec Loss =  0.028084438 \n",
      "\n",
      "Step 275 in 817.66 sec Loss =  0.15994468 \n",
      "\n",
      "Step 276 in 811.05 sec Loss =  0.14660577 \n",
      "\n",
      "Step 277 in 811.84 sec Loss =  0.04147406 \n",
      "\n",
      "Step 278 in 817.54 sec Loss =  0.0118384175 \n",
      "\n",
      "Step 279 in 818.24 sec Loss =  0.015953906 \n",
      "\n",
      "Step 280 in 817.89 sec Loss =  0.12746465 \n",
      "\n",
      "Step 281 in 818.08 sec Loss =  0.17191899 \n",
      "\n",
      "Step 282 in 817.61 sec Loss =  0.12530634 \n",
      "\n",
      "Step 283 in 810.03 sec Loss =  0.048934363 \n",
      "\n",
      "Step 284 in 815.63 sec Loss =  0.08582967 \n",
      "\n",
      "Step 285 in 810.77 sec Loss =  0.2264878 \n",
      "\n",
      "Step 286 in 811.30 sec Loss =  0.068444 \n",
      "\n",
      "Step 287 in 811.38 sec Loss =  0.066020615 \n",
      "\n",
      "Step 288 in 814.96 sec Loss =  0.099369824 \n",
      "\n",
      "Step 289 in 815.04 sec Loss =  0.034507662 \n",
      "\n",
      "Step 290 in 813.62 sec Loss =  0.04843926 \n",
      "\n",
      "Step 291 in 801.77 sec Loss =  0.06997776 \n",
      "\n",
      "Step 292 in 808.60 sec Loss =  0.032011535 \n",
      "\n",
      "Step 293 in 813.97 sec Loss =  0.06910199 \n",
      "\n",
      "Step 294 in 813.66 sec Loss =  0.035255868 \n",
      "\n",
      "Step 295 in 813.89 sec Loss =  0.047899473 \n",
      "\n",
      "Step 296 in 815.04 sec Loss =  0.05480195 \n",
      "\n",
      "Step 297 in 816.49 sec Loss =  0.036071036 \n",
      "\n",
      "Step 298 in 816.41 sec Loss =  0.039996747 \n",
      "\n",
      "Step 299 in 811.73 sec Loss =  0.08776679 \n",
      "\n",
      "Step 300 in 811.46 sec Loss =  0.22885403 \n",
      "\n",
      "Step 301 in 811.90 sec Loss =  0.2749272 \n",
      "\n",
      "Step 302 in 811.43 sec Loss =  0.06003152 \n",
      "\n",
      "Step 303 in 815.04 sec Loss =  0.03226099 \n",
      "\n",
      "Step 304 in 800.94 sec Loss =  0.1984036 \n",
      "\n",
      "Step 305 in 810.11 sec Loss =  0.0409223 \n",
      "\n",
      "Step 306 in 810.84 sec Loss =  0.04677035 \n",
      "\n",
      "Step 307 in 810.61 sec Loss =  0.07655132 \n",
      "\n",
      "Step 308 in 812.10 sec Loss =  0.048000887 \n",
      "\n",
      "Step 309 in 813.66 sec Loss =  0.046818532 \n",
      "\n",
      "Step 310 in 816.00 sec Loss =  0.05183845 \n",
      "\n",
      "Step 311 in 814.29 sec Loss =  0.029313076 \n",
      "\n",
      "Step 312 in 814.92 sec Loss =  0.007565308 \n",
      "\n",
      "Step 313 in 815.76 sec Loss =  0.070875965 \n",
      "\n",
      "Step 314 in 815.20 sec Loss =  0.16102968 \n",
      "\n",
      "Step 315 in 817.58 sec Loss =  0.084703505 \n",
      "\n",
      "Step 316 in 815.56 sec Loss =  0.001282054 \n",
      "\n",
      "Step 317 in 340.38 sec Loss =  0.006942253 \n",
      "\n",
      "Step 318 in 815.45 sec Loss =  0.09295459 \n",
      "\n",
      "Step 319 in 815.52 sec Loss =  0.1121505 \n",
      "\n",
      "Step 320 in 817.29 sec Loss =  0.26407102 \n",
      "\n",
      "Step 321 in 816.18 sec Loss =  0.029728647 \n",
      "\n",
      "Step 322 in 816.24 sec Loss =  0.033841442 \n",
      "\n",
      "Step 323 in 818.20 sec Loss =  0.025795065 \n",
      "\n",
      "Step 324 in 818.23 sec Loss =  0.0018107662 \n",
      "\n",
      "Step 325 in 340.85 sec Loss =  0.010181842 \n",
      "\n",
      "Step 326 in 815.88 sec Loss =  0.12729554 \n",
      "\n",
      "Step 327 in 818.75 sec Loss =  0.020531166 \n",
      "\n",
      "Step 328 in 815.81 sec Loss =  0.023031136 \n",
      "\n",
      "Step 329 in 814.38 sec Loss =  0.06908893 \n",
      "\n",
      "Step 330 in 811.89 sec Loss =  0.036511566 \n",
      "\n",
      "Step 331 in 812.88 sec Loss =  0.01497109 \n",
      "\n",
      "Step 332 in 340.44 sec Loss =  0.008157325 \n",
      "\n",
      "Step 333 in 808.70 sec Loss =  0.020472001 \n",
      "\n",
      "Step 334 in 811.74 sec Loss =  0.008586716 \n",
      "\n",
      "Step 335 in 818.63 sec Loss =  0.045646034 \n",
      "\n",
      "Step 336 in 813.15 sec Loss =  0.04620637 \n",
      "\n",
      "Step 337 in 813.69 sec Loss =  0.016417278 \n",
      "\n",
      "Step 338 in 813.06 sec Loss =  0.05074438 \n",
      "\n",
      "Step 339 in 812.92 sec Loss =  0.04511464 \n",
      "\n",
      "Step 340 in 816.40 sec Loss =  0.005515982 \n",
      "\n",
      "Step 341 in 812.31 sec Loss =  0.018696493 \n",
      "\n",
      "Step 342 in 815.92 sec Loss =  0.062359694 \n",
      "\n",
      "Step 343 in 815.16 sec Loss =  0.03438572 \n",
      "\n",
      "Step 344 in 812.05 sec Loss =  0.041338526 \n",
      "\n",
      "Step 345 in 812.41 sec Loss =  0.020504614 \n",
      "\n",
      "Step 346 in 816.02 sec Loss =  0.005987329 \n",
      "\n",
      "Step 347 in 339.91 sec Loss =  0.010116523 \n",
      "\n",
      "Step 348 in 812.52 sec Loss =  0.043086007 \n",
      "\n",
      "Step 349 in 812.26 sec Loss =  0.057401612 \n",
      "\n",
      "Step 350 in 809.36 sec Loss =  0.120309986 \n",
      "\n",
      "Step 351 in 810.00 sec Loss =  0.023212574 \n",
      "\n",
      "Step 352 in 340.81 sec Loss =  0.006041672 \n",
      "\n",
      "Step 353 in 812.17 sec Loss =  0.027123196 \n",
      "\n",
      "Step 354 in 814.83 sec Loss =  0.019628694 \n",
      "\n",
      "Step 355 in 816.01 sec Loss =  0.056528144 \n",
      "\n",
      "Step 356 in 814.98 sec Loss =  0.015534872 \n",
      "\n",
      "Step 357 in 815.18 sec Loss =  0.020495031 \n",
      "\n",
      "Step 358 in 808.73 sec Loss =  0.030490885 \n",
      "\n",
      "Step 359 in 814.24 sec Loss =  0.17718759 \n",
      "\n",
      "Step 360 in 811.26 sec Loss =  0.028679287 \n",
      "\n",
      "Step 361 in 811.65 sec Loss =  0.03820393 \n",
      "\n",
      "Step 362 in 810.27 sec Loss =  0.04452473 \n",
      "\n",
      "Step 363 in 812.11 sec Loss =  0.0056880866 \n",
      "\n",
      "Step 364 in 340.17 sec Loss =  0.013845634 \n",
      "\n",
      "Step 365 in 812.09 sec Loss =  0.042978797 \n",
      "\n",
      "Step 366 in 809.99 sec Loss =  0.02466457 \n",
      "\n",
      "Step 367 in 810.62 sec Loss =  0.0026476607 \n",
      "\n",
      "Step 368 in 339.56 sec Loss =  0.007511374 \n",
      "\n",
      "Step 369 in 811.24 sec Loss =  0.023924394 \n",
      "\n",
      "Step 370 in 815.88 sec Loss =  0.022984792 \n",
      "\n",
      "Step 371 in 811.85 sec Loss =  0.022404237 \n",
      "\n",
      "Step 372 in 812.66 sec Loss =  0.014369173 \n",
      "\n",
      "Step 373 in 812.72 sec Loss =  0.032658253 \n",
      "\n",
      "Step 374 in 812.78 sec Loss =  0.018160602 \n",
      "\n",
      "Step 375 in 814.52 sec Loss =  0.011147162 \n",
      "\n",
      "Step 376 in 812.20 sec Loss =  0.012905631 \n",
      "\n",
      "Step 377 in 815.01 sec Loss =  0.030229073 \n",
      "\n",
      "Step 378 in 812.01 sec Loss =  0.009255626 \n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Step 379 in 812.60 sec Loss =  0.035169788 \n",
      "\n",
      "Step 380 in 811.92 sec Loss =  0.026180396 \n",
      "\n",
      "Step 381 in 815.35 sec Loss =  0.0063589583 \n",
      "\n",
      "Step 382 in 811.96 sec Loss =  0.005056066 \n",
      "\n",
      "Step 383 in 340.22 sec Loss =  0.0009957834 \n",
      "\n",
      "Step 384 in 813.14 sec Loss =  0.017528951 \n",
      "\n",
      "Step 385 in 811.57 sec Loss =  0.023748416 \n",
      "\n",
      "Step 386 in 811.68 sec Loss =  0.057891063 \n",
      "\n",
      "Step 387 in 813.90 sec Loss =  0.018952196 \n",
      "\n",
      "Step 388 in 815.19 sec Loss =  0.032416858 \n",
      "\n",
      "Step 389 in 814.70 sec Loss =  0.06848748 \n",
      "\n",
      "Step 390 in 814.96 sec Loss =  0.006065098 \n",
      "\n",
      "Step 391 in 340.33 sec Loss =  0.0032631385 \n",
      "\n",
      "Step 392 in 340.66 sec Loss =  0.006101937 \n",
      "\n",
      "Step 393 in 340.34 sec Loss =  0.01256637 \n",
      "\n",
      "Step 394 in 340.32 sec Loss =  0.012216397 \n",
      "\n",
      "Step 395 in 814.08 sec Loss =  0.014924498 \n",
      "\n",
      "Step 396 in 812.30 sec Loss =  0.018459685 \n",
      "\n",
      "Step 397 in 812.08 sec Loss =  0.01979603 \n",
      "\n",
      "Step 398 in 812.05 sec Loss =  0.0050871796 \n",
      "\n",
      "Step 399 in 339.77 sec Loss =  0.010633731 \n",
      "\n",
      "Step 400 in 339.77 sec Loss =  0.0139770275 \n",
      "\n",
      "Step 401 in 339.70 sec Loss =  0.0074055702 \n",
      "\n",
      "Step 402 in 339.83 sec Loss =  0.005820624 \n",
      "\n",
      "Step 403 in 340.28 sec Loss =  0.007345798 \n",
      "\n",
      "Step 404 in 339.80 sec Loss =  0.011424547 \n",
      "\n",
      "Step 405 in 339.75 sec Loss =  0.011985106 \n",
      "\n",
      "Step 406 in 339.73 sec Loss =  0.0043131514 \n",
      "\n",
      "Step 407 in 340.04 sec Loss =  0.004586369 \n",
      "\n",
      "Step 408 in 340.18 sec Loss =  0.010914659 \n",
      "\n",
      "Step 409 in 339.52 sec Loss =  0.012477567 \n",
      "\n",
      "Step 410 in 339.97 sec Loss =  0.0066774245 \n",
      "\n",
      "Step 411 in 340.20 sec Loss =  0.0044831764 \n",
      "\n",
      "Step 412 in 339.91 sec Loss =  0.0061334744 \n",
      "\n",
      "Step 413 in 340.09 sec Loss =  0.009406679 \n",
      "\n",
      "Step 414 in 340.16 sec Loss =  0.008447813 \n",
      "\n",
      "Step 415 in 339.81 sec Loss =  0.007988447 \n",
      "\n",
      "Step 416 in 339.81 sec Loss =  0.0059409207 \n",
      "\n",
      "Step 417 in 811.63 sec Loss =  0.0438721 \n",
      "\n",
      "Step 418 in 811.68 sec Loss =  0.0325698 \n",
      "\n",
      "Step 419 in 812.27 sec Loss =  0.002590376 \n",
      "\n",
      "Step 420 in 339.82 sec Loss =  0.0005070772 \n",
      "\n",
      "Step 421 in 339.97 sec Loss =  0.00031985986 \n",
      "\n",
      "Step 422 in 339.75 sec Loss =  0.0006132081 \n",
      "\n",
      "Step 423 in 339.76 sec Loss =  0.0071813427 \n",
      "\n",
      "Step 424 in 340.10 sec Loss =  0.011853375 \n",
      "\n",
      "Step 425 in 340.34 sec Loss =  0.006087719 \n",
      "\n",
      "Step 426 in 339.76 sec Loss =  0.0032191288 \n",
      "\n",
      "Step 427 in 339.70 sec Loss =  0.0030851592 \n",
      "\n",
      "Step 428 in 339.75 sec Loss =  0.0053926245 \n",
      "\n",
      "Step 429 in 339.78 sec Loss =  0.009714267 \n",
      "\n",
      "Step 430 in 339.81 sec Loss =  0.004209397 \n",
      "\n",
      "Step 431 in 339.80 sec Loss =  0.0025729265 \n",
      "\n",
      "Step 432 in 339.65 sec Loss =  0.0036243498 \n",
      "\n",
      "Step 433 in 339.80 sec Loss =  0.005744423 \n",
      "\n",
      "Step 434 in 339.74 sec Loss =  0.005101392 \n",
      "\n",
      "Step 435 in 339.96 sec Loss =  0.0031039761 \n",
      "\n",
      "Step 436 in 339.91 sec Loss =  0.0030202768 \n",
      "\n",
      "Step 437 in 813.25 sec Loss =  0.016855106 \n",
      "\n",
      "Step 438 in 812.15 sec Loss =  0.059180677 \n",
      "\n",
      "Step 439 in 811.78 sec Loss =  0.056902368 \n",
      "\n",
      "Step 440 in 812.23 sec Loss =  0.008906919 \n",
      "\n",
      "Step 441 in 340.16 sec Loss =  0.0042347484 \n",
      "\n",
      "Step 442 in 339.78 sec Loss =  0.0010666687 \n",
      "\n",
      "Step 443 in 339.74 sec Loss =  0.00042540304 \n",
      "\n",
      "Step 444 in 339.90 sec Loss =  0.0017153416 \n",
      "\n",
      "Step 445 in 811.39 sec Loss =  0.027159762 \n",
      "\n",
      "Step 446 in 811.65 sec Loss =  0.036141876 \n",
      "\n",
      "Step 447 in 816.28 sec Loss =  0.0056619733 \n",
      "\n",
      "Step 448 in 339.38 sec Loss =  0.0014729542 \n",
      "\n",
      "Step 449 in 339.05 sec Loss =  0.00076405104 \n",
      "\n",
      "Step 450 in 340.22 sec Loss =  0.0023462446 \n",
      "\n",
      "Step 451 in 339.17 sec Loss =  0.008732836 \n",
      "\n",
      "Step 452 in 339.12 sec Loss =  0.005089377 \n",
      "\n",
      "Step 453 in 340.24 sec Loss =  0.0020554983 \n",
      "\n",
      "Step 454 in 339.39 sec Loss =  0.0020234582 \n",
      "\n",
      "Step 455 in 814.45 sec Loss =  0.010761302 \n",
      "\n",
      "Step 456 in 811.87 sec Loss =  0.012935897 \n",
      "\n",
      "Step 457 in 340.03 sec Loss =  0.010596852 \n",
      "\n",
      "Step 458 in 340.00 sec Loss =  0.004574056 \n",
      "\n",
      "Step 459 in 340.12 sec Loss =  0.0046484713 \n",
      "\n",
      "Step 460 in 340.07 sec Loss =  0.0067111123 \n",
      "\n",
      "Step 461 in 340.07 sec Loss =  0.0076478827 \n",
      "\n",
      "Step 462 in 340.10 sec Loss =  0.0035276285 \n",
      "\n",
      "Step 463 in 813.45 sec Loss =  0.0035015312 \n",
      "\n",
      "Step 464 in 340.29 sec Loss =  0.0027995864 \n",
      "\n",
      "Step 465 in 340.65 sec Loss =  0.0035989606 \n",
      "\n",
      "Step 466 in 340.19 sec Loss =  0.009256745 \n",
      "\n",
      "Step 467 in 341.10 sec Loss =  0.0039447453 \n",
      "\n",
      "Step 468 in 340.44 sec Loss =  0.0022725558 \n",
      "\n",
      "Step 469 in 340.69 sec Loss =  0.0026389435 \n",
      "\n",
      "Step 470 in 340.64 sec Loss =  0.0053587747 \n",
      "\n",
      "Step 471 in 340.47 sec Loss =  0.005238665 \n",
      "\n",
      "Step 472 in 813.31 sec Loss =  0.23164569 \n",
      "\n",
      "Step 473 in 340.36 sec Loss =  0.13645682 \n",
      "\n",
      "Step 474 in 813.38 sec Loss =  0.023252621 \n",
      "\n",
      "Step 475 in 814.60 sec Loss =  0.01137584 \n",
      "\n",
      "Step 476 in 340.66 sec Loss =  0.0049979617 \n",
      "\n",
      "Step 477 in 340.49 sec Loss =  0.0040728683 \n",
      "\n",
      "Step 478 in 339.72 sec Loss =  0.0036609224 \n",
      "\n",
      "Step 479 in 340.63 sec Loss =  0.003923628 \n",
      "\n",
      "Step 480 in 339.87 sec Loss =  0.0057870885 \n",
      "\n",
      "Step 481 in 813.10 sec Loss =  0.014242579 \n",
      "\n",
      "Step 482 in 340.46 sec Loss =  0.0088074645 \n",
      "\n",
      "Step 483 in 340.23 sec Loss =  0.004909058 \n",
      "\n",
      "Step 484 in 813.90 sec Loss =  0.005465339 \n",
      "\n",
      "Step 485 in 340.45 sec Loss =  0.0019074938 \n",
      "\n",
      "Step 486 in 340.85 sec Loss =  0.0014402937 \n",
      "\n",
      "Step 487 in 340.63 sec Loss =  0.002739212 \n",
      "\n",
      "Step 488 in 340.12 sec Loss =  0.003756021 \n",
      "\n",
      "Step 489 in 340.05 sec Loss =  0.0018724513 \n",
      "\n",
      "Step 490 in 340.04 sec Loss =  0.0013606068 \n",
      "\n",
      "Step 491 in 340.72 sec Loss =  0.0016061106 \n",
      "\n",
      "Step 492 in 340.78 sec Loss =  0.0026729242 \n",
      "\n",
      "Step 493 in 340.40 sec Loss =  0.0027455825 \n",
      "\n",
      "Step 494 in 340.40 sec Loss =  0.002421549 \n",
      "\n",
      "Step 495 in 340.37 sec Loss =  0.0033328056 \n",
      "\n",
      "Step 496 in 340.36 sec Loss =  0.0024270588 \n",
      "\n",
      "Step 497 in 340.66 sec Loss =  0.0014566303 \n",
      "\n",
      "Step 498 in 340.29 sec Loss =  0.0019197826 \n",
      "\n",
      "Step 499 in 340.52 sec Loss =  0.002595311 \n",
      "\n",
      "Step 500 in 339.96 sec Loss =  0.0021146834 \n",
      "\n",
      "Step 501 in 340.10 sec Loss =  0.0017421562 \n",
      "\n",
      "Step 502 in 339.80 sec Loss =  0.0019371605 \n",
      "\n",
      "Step 503 in 339.82 sec Loss =  0.001700643 \n",
      "\n",
      "Step 504 in 340.29 sec Loss =  0.001480402 \n",
      "\n",
      "Step 505 in 340.47 sec Loss =  0.0016319706 \n",
      "\n",
      "Step 506 in 340.23 sec Loss =  0.0018409996 \n",
      "\n",
      "Step 507 in 340.36 sec Loss =  0.0028626733 \n",
      "\n",
      "Step 508 in 340.17 sec Loss =  0.0025317392 \n",
      "\n",
      "Step 509 in 340.30 sec Loss =  0.0015022514 \n",
      "\n",
      "Step 510 in 340.66 sec Loss =  0.0012700155 \n",
      "\n",
      "Step 511 in 340.17 sec Loss =  0.0016499296 \n",
      "\n",
      "Step 512 in 340.29 sec Loss =  0.001852886 \n",
      "\n",
      "Step 513 in 340.27 sec Loss =  0.001677203 \n",
      "\n",
      "Step 514 in 340.16 sec Loss =  0.0016949584 \n",
      "\n",
      "Step 515 in 340.24 sec Loss =  0.0024356777 \n",
      "\n",
      "Step 516 in 340.31 sec Loss =  0.00217365 \n",
      "\n",
      "Step 517 in 340.24 sec Loss =  0.0013089826 \n",
      "\n",
      "Step 518 in 340.32 sec Loss =  0.0012259893 \n",
      "\n",
      "Step 519 in 340.27 sec Loss =  0.0018792525 \n",
      "\n",
      "Step 520 in 340.27 sec Loss =  0.001904289 \n",
      "\n",
      "Step 521 in 340.53 sec Loss =  0.001448288 \n",
      "\n",
      "Step 522 in 340.32 sec Loss =  0.001297741 \n",
      "\n",
      "Step 523 in 340.35 sec Loss =  0.0015616424 \n",
      "\n",
      "Step 524 in 340.43 sec Loss =  0.0016846578 \n",
      "\n",
      "Step 525 in 340.31 sec Loss =  0.0013445553 \n",
      "\n",
      "Step 526 in 340.24 sec Loss =  0.0011971339 \n",
      "\n",
      "Step 527 in 339.68 sec Loss =  0.0015467004 \n",
      "\n",
      "Step 528 in 339.92 sec Loss =  0.0018664048 \n",
      "\n",
      "Step 529 in 339.77 sec Loss =  0.0018687472 \n",
      "\n",
      "Step 530 in 340.05 sec Loss =  0.0013716719 \n",
      "\n",
      "Step 531 in 339.88 sec Loss =  0.0011893451 \n",
      "\n",
      "Step 532 in 340.40 sec Loss =  0.0013341326 \n",
      "\n",
      "Step 533 in 339.83 sec Loss =  0.001666571 \n",
      "\n",
      "Step 534 in 339.77 sec Loss =  0.0014469762 \n",
      "\n",
      "Step 535 in 339.91 sec Loss =  0.0012071655 \n",
      "\n",
      "Step 536 in 340.29 sec Loss =  0.00121526 \n",
      "\n",
      "Step 537 in 340.04 sec Loss =  0.0014463264 \n",
      "\n",
      "Step 538 in 340.54 sec Loss =  0.0012304027 \n",
      "\n",
      "Step 539 in 339.33 sec Loss =  0.0010984814 \n",
      "\n",
      "Step 540 in 339.38 sec Loss =  0.0012316939 \n",
      "\n",
      "Step 541 in 340.11 sec Loss =  0.0015744207 \n",
      "\n",
      "Step 542 in 340.32 sec Loss =  0.0016484254 \n",
      "\n",
      "Step 543 in 340.63 sec Loss =  0.0011433607 \n",
      "\n",
      "Step 544 in 339.25 sec Loss =  0.0009403443 \n",
      "\n",
      "Step 545 in 340.95 sec Loss =  0.0012007074 \n",
      "\n",
      "Step 546 in 339.47 sec Loss =  0.0014938634 \n",
      "\n",
      "Step 547 in 340.43 sec Loss =  0.0011747815 \n",
      "\n",
      "Step 548 in 339.15 sec Loss =  0.0010010194 \n",
      "\n",
      "Step 549 in 339.12 sec Loss =  0.0010944633 \n",
      "\n",
      "Step 550 in 340.25 sec Loss =  0.0012317725 \n",
      "\n",
      "Step 551 in 339.44 sec Loss =  0.0012077847 \n",
      "\n",
      "Step 552 in 339.66 sec Loss =  0.0009932776 \n",
      "\n",
      "Step 553 in 339.70 sec Loss =  0.00096325413 \n",
      "\n",
      "Step 554 in 339.73 sec Loss =  0.0010963138 \n",
      "\n",
      "Step 555 in 339.79 sec Loss =  0.0012433898 \n",
      "\n",
      "Step 556 in 340.53 sec Loss =  0.0012720024 \n",
      "\n",
      "Step 557 in 340.10 sec Loss =  0.0011868208 \n",
      "\n",
      "Step 558 in 339.12 sec Loss =  0.0009278611 \n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Step 559 in 340.26 sec Loss =  0.00093674473 \n",
      "\n",
      "Step 560 in 340.17 sec Loss =  0.0011697481 \n",
      "\n",
      "Step 561 in 340.32 sec Loss =  0.0011022585 \n",
      "\n",
      "Step 562 in 340.33 sec Loss =  0.0009525009 \n",
      "\n",
      "Step 563 in 340.23 sec Loss =  0.0009850116 \n",
      "\n",
      "Step 564 in 339.06 sec Loss =  0.0010230722 \n",
      "\n",
      "Step 565 in 340.04 sec Loss =  0.00094449974 \n",
      "\n",
      "Step 566 in 339.69 sec Loss =  0.0009949586 \n",
      "\n",
      "Step 567 in 339.66 sec Loss =  0.0010297833 \n",
      "\n",
      "Step 568 in 339.65 sec Loss =  0.0009758394 \n",
      "\n",
      "Step 569 in 339.70 sec Loss =  0.0010219365 \n",
      "\n",
      "Step 570 in 339.61 sec Loss =  0.0008710043 \n",
      "\n",
      "Step 571 in 339.68 sec Loss =  0.00083667407 \n",
      "\n",
      "Step 572 in 339.64 sec Loss =  0.00085706054 \n",
      "\n",
      "Step 573 in 339.77 sec Loss =  0.0010681126 \n",
      "\n",
      "Step 574 in 340.20 sec Loss =  0.0010251626 \n",
      "\n",
      "Step 575 in 340.17 sec Loss =  0.00091476535 \n",
      "\n",
      "Step 576 in 340.09 sec Loss =  0.0007902483 \n",
      "\n",
      "Step 577 in 340.53 sec Loss =  0.0007785416 \n",
      "\n",
      "Step 578 in 812.77 sec Loss =  0.006335583 \n",
      "\n",
      "Step 579 in 340.00 sec Loss =  0.0025469312 \n",
      "\n",
      "Step 580 in 339.96 sec Loss =  0.0013319998 \n",
      "\n",
      "Step 581 in 339.68 sec Loss =  0.0012503642 \n",
      "\n",
      "Step 582 in 339.72 sec Loss =  0.0014041025 \n",
      "\n",
      "Step 583 in 340.07 sec Loss =  0.0018035936 \n",
      "\n",
      "Step 584 in 340.11 sec Loss =  0.0014673022 \n",
      "\n",
      "Step 585 in 340.42 sec Loss =  0.001351603 \n",
      "\n",
      "Step 586 in 340.19 sec Loss =  0.0014023167 \n",
      "\n",
      "Step 587 in 340.12 sec Loss =  0.0014893505 \n",
      "\n",
      "Step 588 in 340.06 sec Loss =  0.001456274 \n",
      "\n",
      "Step 589 in 340.17 sec Loss =  0.0014658883 \n",
      "\n",
      "Step 590 in 340.00 sec Loss =  0.0014441897 \n",
      "\n",
      "Step 591 in 340.08 sec Loss =  0.0014154997 \n",
      "\n",
      "Step 592 in 340.42 sec Loss =  0.0013854225 \n",
      "\n",
      "Step 593 in 340.13 sec Loss =  0.0013131629 \n",
      "\n",
      "Step 594 in 340.03 sec Loss =  0.0013404373 \n",
      "\n",
      "Step 595 in 339.97 sec Loss =  0.0014296977 \n",
      "\n",
      "Step 596 in 340.04 sec Loss =  0.0014716054 \n",
      "\n",
      "Step 597 in 340.04 sec Loss =  0.0013918793 \n",
      "\n",
      "Step 598 in 340.09 sec Loss =  0.0012821066 \n",
      "\n",
      "Step 599 in 340.01 sec Loss =  0.001286207 \n",
      "\n",
      "Step 600 in 340.04 sec Loss =  0.0013615303 \n",
      "\n",
      "Step 601 in 340.07 sec Loss =  0.0013965757 \n",
      "\n",
      "Step 602 in 340.09 sec Loss =  0.0013412065 \n",
      "\n",
      "Step 603 in 340.27 sec Loss =  0.0013022919 \n",
      "\n",
      "Step 604 in 340.32 sec Loss =  0.0012684792 \n",
      "\n",
      "Step 605 in 339.90 sec Loss =  0.0012473012 \n",
      "\n",
      "Step 606 in 339.75 sec Loss =  0.0012668603 \n",
      "\n",
      "Step 607 in 339.84 sec Loss =  0.001342481 \n",
      "\n",
      "Step 608 in 814.12 sec Loss =  0.0074380753 \n",
      "\n",
      "Step 609 in 340.80 sec Loss =  0.006067649 \n",
      "\n",
      "Step 610 in 339.78 sec Loss =  0.0050595677 \n",
      "\n",
      "Step 611 in 339.90 sec Loss =  0.004175059 \n",
      "\n",
      "Step 612 in 340.41 sec Loss =  0.0034553218 \n",
      "\n",
      "Step 613 in 340.01 sec Loss =  0.0029138853 \n",
      "\n",
      "Step 614 in 340.07 sec Loss =  0.0025564372 \n",
      "\n",
      "Step 615 in 340.04 sec Loss =  0.0022272386 \n",
      "\n",
      "Step 616 in 340.24 sec Loss =  0.001895494 \n",
      "\n",
      "Step 617 in 340.01 sec Loss =  0.0017286105 \n",
      "\n",
      "Step 618 in 340.12 sec Loss =  0.0016249612 \n",
      "\n",
      "Step 619 in 339.94 sec Loss =  0.0015989917 \n",
      "\n",
      "Step 620 in 340.34 sec Loss =  0.0015111287 \n",
      "\n",
      "Step 621 in 339.92 sec Loss =  0.0014448396 \n",
      "\n",
      "Step 622 in 339.85 sec Loss =  0.0014097363 \n",
      "\n",
      "Step 623 in 339.87 sec Loss =  0.0014097419 \n",
      "\n",
      "Step 624 in 340.25 sec Loss =  0.0014037899 \n",
      "\n",
      "Step 625 in 340.02 sec Loss =  0.0013737166 \n",
      "\n",
      "Step 626 in 339.87 sec Loss =  0.0013332444 \n",
      "\n",
      "Step 627 in 339.68 sec Loss =  0.0013400813 \n",
      "\n",
      "Step 628 in 339.42 sec Loss =  0.0013592502 \n",
      "\n",
      "Step 629 in 339.33 sec Loss =  0.0013352107 \n",
      "\n",
      "Step 630 in 339.61 sec Loss =  0.0013117144 \n",
      "\n",
      "Step 631 in 339.36 sec Loss =  0.0013066955 \n",
      "\n",
      "Step 632 in 339.68 sec Loss =  0.0013120441 \n",
      "\n",
      "Step 633 in 339.41 sec Loss =  0.0013087217 \n",
      "\n",
      "Step 634 in 339.65 sec Loss =  0.0012885305 \n",
      "\n",
      "Step 635 in 339.31 sec Loss =  0.0012767096 \n",
      "\n",
      "Step 636 in 339.52 sec Loss =  0.001266934 \n",
      "\n",
      "Step 637 in 339.45 sec Loss =  0.0012796775 \n",
      "\n",
      "Step 638 in 339.42 sec Loss =  0.001260896 \n",
      "\n",
      "Step 639 in 339.36 sec Loss =  0.0012304891 \n",
      "\n",
      "Step 640 in 339.56 sec Loss =  0.0012197002 \n",
      "\n",
      "Step 641 in 339.42 sec Loss =  0.0012552437 \n",
      "\n",
      "Step 642 in 340.12 sec Loss =  0.0012786238 \n",
      "\n",
      "Step 643 in 339.65 sec Loss =  0.0012232165 \n",
      "\n",
      "Step 644 in 339.91 sec Loss =  0.0011913624 \n",
      "\n",
      "Step 645 in 339.67 sec Loss =  0.001219279 \n",
      "\n",
      "Step 646 in 339.73 sec Loss =  0.0012259769 \n",
      "\n",
      "Step 647 in 340.21 sec Loss =  0.001191026 \n",
      "\n",
      "Step 648 in 339.76 sec Loss =  0.0011743188 \n",
      "\n",
      "Step 649 in 339.50 sec Loss =  0.0011709447 \n",
      "\n",
      "Step 650 in 339.48 sec Loss =  0.0011741972 \n",
      "\n",
      "Step 651 in 339.57 sec Loss =  0.0011676628 \n",
      "\n",
      "Step 652 in 339.93 sec Loss =  0.0011615981 \n",
      "\n",
      "Step 653 in 340.09 sec Loss =  0.0011672624 \n",
      "\n",
      "Step 654 in 340.43 sec Loss =  0.0011623882 \n",
      "\n",
      "Step 655 in 339.93 sec Loss =  0.001172686 \n",
      "\n",
      "Step 656 in 339.86 sec Loss =  0.0011447917 \n",
      "\n",
      "Step 657 in 339.98 sec Loss =  0.0011208801 \n",
      "\n",
      "Step 658 in 809.83 sec Loss =  0.03662092 \n",
      "\n",
      "Step 659 in 339.77 sec Loss =  0.0281059 \n",
      "\n",
      "Step 660 in 339.83 sec Loss =  0.021798331 \n",
      "\n",
      "Step 661 in 339.92 sec Loss =  0.016844789 \n",
      "\n",
      "Step 662 in 340.20 sec Loss =  0.012756479 \n",
      "\n",
      "Step 663 in 340.32 sec Loss =  0.009616473 \n",
      "\n",
      "Step 664 in 340.25 sec Loss =  0.00698751 \n",
      "\n",
      "Step 665 in 339.97 sec Loss =  0.0051040514 \n",
      "\n",
      "Step 666 in 813.56 sec Loss =  0.0054163644 \n",
      "\n",
      "Step 667 in 339.81 sec Loss =  0.003262161 \n",
      "\n",
      "Step 668 in 340.36 sec Loss =  0.0018615983 \n",
      "\n",
      "Step 669 in 339.70 sec Loss =  0.0010299357 \n",
      "\n",
      "Step 670 in 339.73 sec Loss =  0.00060357235 \n",
      "\n",
      "Step 671 in 340.24 sec Loss =  0.00044315233 \n",
      "\n",
      "Step 672 in 339.91 sec Loss =  0.00038964013 \n",
      "\n",
      "Step 673 in 339.82 sec Loss =  0.00036804745 \n",
      "\n",
      "Step 674 in 339.89 sec Loss =  0.0005017714 \n",
      "\n",
      "Step 675 in 339.90 sec Loss =  0.0005171201 \n",
      "\n",
      "Step 676 in 339.64 sec Loss =  0.0003839467 \n",
      "\n",
      "Step 677 in 339.96 sec Loss =  0.00035512322 \n",
      "\n",
      "Step 678 in 339.31 sec Loss =  0.0004100202 \n",
      "\n",
      "Step 679 in 339.96 sec Loss =  0.00047414354 \n",
      "\n",
      "Step 680 in 338.94 sec Loss =  0.00038090814 \n",
      "\n",
      "Step 681 in 338.91 sec Loss =  0.00036505028 \n",
      "\n",
      "Step 682 in 339.93 sec Loss =  0.00036857554 \n",
      "\n",
      "Step 683 in 339.99 sec Loss =  0.0003697134 \n",
      "\n",
      "Step 684 in 339.31 sec Loss =  0.0003888649 \n",
      "\n",
      "Step 685 in 340.23 sec Loss =  0.00038872857 \n",
      "\n",
      "Step 686 in 340.03 sec Loss =  0.0003343069 \n",
      "\n",
      "Step 687 in 339.41 sec Loss =  0.00032657507 \n",
      "\n",
      "Step 688 in 339.63 sec Loss =  0.00032620243 \n",
      "\n",
      "Step 689 in 340.13 sec Loss =  0.00036201027 \n",
      "\n",
      "Step 690 in 339.53 sec Loss =  0.00036776456 \n",
      "\n",
      "Step 691 in 339.49 sec Loss =  0.0003299206 \n",
      "\n",
      "Step 692 in 339.39 sec Loss =  0.00033367734 \n",
      "\n",
      "Step 693 in 340.41 sec Loss =  0.0003549386 \n",
      "\n",
      "Step 694 in 341.46 sec Loss =  0.00034639792 \n",
      "\n",
      "Step 695 in 340.04 sec Loss =  0.00031016886 \n",
      "\n",
      "Step 696 in 340.17 sec Loss =  0.0003008096 \n",
      "\n",
      "Step 697 in 340.66 sec Loss =  0.00030555011 \n",
      "\n",
      "Step 698 in 339.63 sec Loss =  0.0003224953 \n",
      "\n",
      "Step 699 in 339.81 sec Loss =  0.00032945303 \n",
      "\n"
     ]
    }
   ],
   "source": [
    "loss_history, times_per_update, predicted_quantities = [], [], []\n",
    "import jax_md\n",
    "opt_state = optimizer.init(init_params)\n",
    "params = init_params\n",
    "for step in range(num_updates):\n",
    "    start_time = time.time()\n",
    "    params, opt_state, trajectory_state, loss_val, predictions = update_fn(step, params, opt_state, trajectory_state)\n",
    "    loss_val.block_until_ready()\n",
    "    step_time = time.time() - start_time\n",
    "\n",
    "    times_per_update.append(step_time)\n",
    "    loss_history.append(loss_val)\n",
    "    predicted_quantities.append(predictions)\n",
    "    print(\"Step {} in {:0.2f} sec\".format(step, step_time), 'Loss = ', loss_val, '\\n')\n",
    "    \n",
    "\n",
    "    if np.isnan(loss_val):  # stop learning if optimization diverged\n",
    "        print('Loss is NaN. This was likely caused by divergence of the optimization or a bad model setup '\n",
    "              'causing a NaN trajectory.')\n",
    "        break\n",
    "\n",
    "init_sim_state = trajectory_state[0]\n",
    "        \n",
    "# save optimized energy_params for re-use\n",
    "final_energy_params = device_get(params)\n",
    "with open(energy_pickle_file_path, 'wb') as f:\n",
    "    pickle.dump(final_energy_params, f)\n",
    "\n",
    "\n",
    "with open(optimization_pickle_file_path, 'wb') as f:\n",
    "    pickle.dump([loss_history, times_per_update, predicted_quantities, init_sim_state], f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# load results, if necessary\n",
    "final_energy_params = pickle.load(open(energy_pickle_file_path, \"rb\" ))\n",
    "final_energy_params = tree_util.tree_map(jnp.array, final_energy_params)\n",
    "loss_history, times_per_update, predicted_quantities, init_sim_state = pickle.load(open(optimization_pickle_file_path, \"rb\" ))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "## Visualize training process\n",
    "\n",
    "Let's start post-processing by plotting the loss and compute-time per update. In this application, most of the update steps do not require a re-computation of the trajectory, reducing computation time significantly. Computing the stiffness tensor is rather expensive. The speed-up that results from re-using reference trajectories is therefore reduced compared to cheap observables, e.g. RDFs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax1 = plt.subplots()\n",
    "ax1.set_xlabel('Update Step')\n",
    "ax1.set_ylabel('MSE Loss')\n",
    "ax1.semilogy(loss_history, color='#3c5488ff', label='Loss')\n",
    "ax2 = ax1.twinx()\n",
    "ax2.plot(times_per_update, label='Time per update', color='#4dbbd5ff')\n",
    "ax2.set_ylabel('$t$ in s')\n",
    "fig.legend(loc=\"upper right\", bbox_to_anchor=(1,1), bbox_transform=ax1.transAxes)\n",
    "plt.savefig('Train_history_diamond.png')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "Next, we visualize convergence of our observables, the stress and stiffness tensor values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted stress tensor: [[-0.19772905 -0.1174123   0.1630551 ]\n",
      " [ 0.01021396 -0.18555078 -0.05879211]\n",
      " [ 0.01427229 -0.05110607  0.16001081]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "ax.set_xlabel('Update Step')\n",
    "ax.set_ylabel('Hydrostatic stress in GPa')\n",
    "\n",
    "if 'stress' in predicted_quantities[0]:\n",
    "    pressure_series = [jnp.trace(prediction_dict['stress']) / convert_from_GPa_to_kJ_mol_nm_3 / 3. for\n",
    "                       prediction_dict in predicted_quantities]\n",
    "    ax.plot(pressure_series, label='$\\sigma_h$', color='#3c5488ff')\n",
    "    ax.axhline(y=0., linestyle='--', color='k')\n",
    "    print('Predicted stress tensor:', predicted_quantities[-1]['stress'] / convert_from_GPa_to_kJ_mol_nm_3)\n",
    "\n",
    "if 'stiffness' in predicted_quantities[0]:\n",
    "    stiffness_list = [prediction_dict['stiffness'] for prediction_dict in predicted_quantities]\n",
    "    stiffness_array = jnp.stack(stiffness_list) / convert_from_GPa_to_kJ_mol_nm_3  # back to GPa\n",
    "    ax1 = ax.twinx()\n",
    "    ax1.set_ylabel('$C_{ij}$ in GPa')\n",
    "    ax1.set_ylim([0, 1200])\n",
    "    ax1.plot(stiffness_array[:, 0], label='$c_{11}$', color='#00a087ff')\n",
    "    ax1.plot(stiffness_array[:, 1], label='$c_{12}$', color='#91d1c2ff')\n",
    "    ax1.axhline(y=c_12_target, linestyle='--', color='k')\n",
    "    ax1.plot(stiffness_array[:, 2], label='$c_{44}$', color='#8491b4ff')\n",
    "    ax1.axhline(y=c_44_target, linestyle='--', color='k')\n",
    "    ax1.axhline(y=c_11_target, linestyle='--', label='targets', color='k')\n",
    "\n",
    "\n",
    "fig.legend(loc=\"lower right\", bbox_to_anchor=(1.4, 0.), bbox_transform=ax1.transAxes)\n",
    "plt.savefig('Stress_stiffness_history.png')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "## Validate simulation results\n",
    "\n",
    "Given that we've trained the model on rather short trajectories, let's validate the model on a longer trajectory.\n",
    "This allows to detect if the model overfitted to initial conditions or drifts away from the targets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted stress tensor: [[ 0.2783271   0.02886999 -0.01478336]\n",
      " [-0.00710172  0.29656953  0.00122136]\n",
      " [-0.00677423  0.02811435  0.3048873 ]]\n",
      "Predicted c_11: 1069.523 GPa; Target: 1079.0 GPa\n",
      "Predicted c_12: 114.45794 GPa; Target: 124.0 GPa\n",
      "Predicted c_44: 559.6833 GPa; Target: 578.0 GPa\n"
     ]
    }
   ],
   "source": [
    "total_time_long = 11000.  # 11 ns\n",
    "t_equilib_long = 1000.  # 1 ns\n",
    "print_every = 0.1\n",
    "\n",
    "\n",
    "long_trajectory_struct = custom_simulator.process_printouts(dt, total_time_long, t_equilib_long, print_every)\n",
    "trajectory_generator = difftre.trajectory_generator_init(simulation_funs, long_trajectory_struct)\n",
    "long_traj_state = trajectory_generator(final_energy_params, init_sim_state)\n",
    "quantity_traj = difftre.compute_quantity_traj(long_traj_state, quantity_dict, neighbor_fn, final_energy_params)\n",
    "\n",
    "stress_tensor = jnp.mean(quantity_traj['stress'], axis=0)\n",
    "print('Predicted stress tensor:', stress_tensor / convert_from_GPa_to_kJ_mol_nm_3)\n",
    "\n",
    "born_stress_prod_snapshots = sigma_tensor_prod(quantity_traj['born_stress'])\n",
    "stress_product_born_mean = jnp.mean(born_stress_prod_snapshots, axis=0)\n",
    "born_stress_tensor_mean = jnp.mean(quantity_traj['born_stress'], axis=0)\n",
    "born_stiffness_mean = jnp.mean(quantity_traj['born_stiffness'], axis=0)\n",
    "stiffness_tensor = stiffness_tensor_fn(born_stiffness_mean, born_stress_tensor_mean, stress_product_born_mean)\n",
    "\n",
    "stiffness_components = custom_quantity.stiffness_tensor_components_cubic_crystal(stiffness_tensor)\n",
    "print('Predicted c_11:', stiffness_components[0] / convert_from_GPa_to_kJ_mol_nm_3, 'GPa; Target:', c_11_target, 'GPa')\n",
    "print('Predicted c_12:', stiffness_components[1] / convert_from_GPa_to_kJ_mol_nm_3, 'GPa; Target:', c_12_target, 'GPa')\n",
    "print('Predicted c_44:', stiffness_components[2] / convert_from_GPa_to_kJ_mol_nm_3, 'GPa; Target:', c_44_target, 'GPa')\n",
    "\n",
    "\n",
    "with open(long_traj_pickle_file_path, 'wb') as f:\n",
    "    pickle.dump([stress_tensor, stiffness_components], f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# you can skip the long run above and just load results\n",
    "stress_tensor, stiffness_components = pickle.load(open(long_traj_pickle_file_path, \"rb\" ))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "final_epoch = num_updates\n",
    "x_position = [final_epoch for i in range(3)]\n",
    "stiffnesses_GPa = stiffness_components / convert_from_GPa_to_kJ_mol_nm_3\n",
    "\n",
    "ax.scatter(final_epoch, jnp.trace(stress_tensor) / convert_from_GPa_to_kJ_mol_nm_3 / 3., marker='x', \n",
    "           color='k', label='10 ns')\n",
    "ax1.scatter(x_position, stiffnesses_GPa, marker='x', color='k')\n",
    "\n",
    "fig.legend(loc=\"lower right\", bbox_to_anchor=(1.4, 0.), bbox_transform=ax1.transAxes)\n",
    "fig.savefig('Stress_stiffness_history.png')\n",
    "fig"
   ]
  }
 ],
 "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}