ml_collisions_colab.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "full_both_sp.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HXmOg0bOo5Cl",
        "colab_type": "text"
      },
      "source": [
        "# from cloud"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QfapmkeGeBNG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# from google.colab import auth\n",
        "# auth.authenticate_user()\n",
        "\n",
        "# project_id = 'peaceful-impact-247117'\n",
        "# bucket_name = 'ml_collisions-data1'\n",
        "\n",
        "# !gcloud config set project {project_id}\n",
        "# !gsutil -m cp -r gs://ml-collisions-data1/hdf5_data/ /content/"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iTsbA-HofFwk",
        "colab_type": "text"
      },
      "source": [
        "# set vars"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iL0V2-G_exrU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from __future__ import print_function, division\n",
        "import torch\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import torch.nn as nn\n",
        "import torch.optim as optim\n",
        "import timeit\n",
        "import h5py\n",
        "from torch.utils.data import Dataset, DataLoader\n",
        "from scipy import stats\n",
        "from google.colab import files\n",
        "from torch.nn import functional as F\n",
        "from torch.autograd import Variable"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XczYn-LMey8z",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "batch_size = 32\n",
        "lr = 1e-7\n",
        "# !mkdir /content/checkpoints/1e-4\n",
        "momentum = 0.99\n",
        "num_epochs = 25\n",
        "percentage_train = 0.8\n",
        "percentage_val = 0.1\n",
        "lr_decay = 0.1\n",
        "step_size = 2\n",
        "# loss_weights = [1,0.5,0.5,0.5]\n",
        "loss_weights = [0,0.5,0.5,0.5,0.5]\n",
        "nphi = 1\n",
        "plot_rate = 250\n",
        "output_rate = 500\n",
        "val_rate = 2000\n",
        "\n",
        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yCU2Up06fEmm",
        "colab_type": "text"
      },
      "source": [
        "# nets"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UpwwVXSWe1C4",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def conv3x3(in_planes, out_planes):\n",
        "    return nn.Conv2d(in_planes, out_planes, kernel_size=3, padding=1, bias=True)\n",
        "\n",
        "class UnetDownBlock(nn.Module):\n",
        "   \n",
        "    def __init__(self, inplanes, planes, predownsample_block):\n",
        "        \n",
        "        super(UnetDownBlock, self).__init__()\n",
        "        \n",
        "        self.predownsample_block = predownsample_block\n",
        "        self.conv1 = conv3x3(inplanes, planes)\n",
        "        self.relu = nn.ReLU(inplace=True)\n",
        "        self.conv2 = conv3x3(planes, planes)\n",
        "        \n",
        "    def forward(self, x):\n",
        "        \n",
        "        x = self.predownsample_block(x)\n",
        "        x = self.conv1(x)\n",
        "        x = self.relu(x)\n",
        "        x = self.conv2(x)\n",
        "        \n",
        "        return x\n",
        "    \n",
        "class UnetUpBlock(nn.Module):\n",
        "   \n",
        "    def __init__(self, inplanes, planes, postupsample_block=None):\n",
        "        \n",
        "        super(UnetUpBlock, self).__init__()\n",
        "        \n",
        "        self.conv1 = conv3x3(inplanes, planes)\n",
        "        self.relu = nn.ReLU(inplace=True)\n",
        "        self.conv2 = conv3x3(planes, planes)\n",
        "        \n",
        "        if postupsample_block is None: \n",
        "            \n",
        "            self.postupsample_block = nn.ConvTranspose2d(in_channels=planes,\n",
        "                                                         out_channels=planes//2,\n",
        "                                                         kernel_size=2,\n",
        "                                                         stride=2)\n",
        "            \n",
        "        else:\n",
        "            \n",
        "            self.postupsample_block = postupsample_block\n",
        "        \n",
        "    def forward(self, x):\n",
        "        \n",
        "        x = self.conv1(x)\n",
        "        x = self.relu(x)\n",
        "        x = self.conv2(x)\n",
        "        x = self.postupsample_block(x)\n",
        "        \n",
        "        return x\n",
        "    \n",
        "    \n",
        "class Unet(nn.Module):\n",
        "    \n",
        "    def __init__(self):\n",
        "        \n",
        "        super(Unet, self).__init__()\n",
        "        \n",
        "        self.predownsample_block = nn.MaxPool2d(kernel_size=2, stride=2)\n",
        "        \n",
        "        self.identity_block = nn.Sequential()\n",
        "        \n",
        "        self.block1 = UnetDownBlock(\n",
        "                                    predownsample_block=self.identity_block,\n",
        "                                    inplanes=2, planes=64\n",
        "                                    )\n",
        "        \n",
        "        self.block2_down = UnetDownBlock(\n",
        "                                         predownsample_block=self.predownsample_block,\n",
        "                                         inplanes=64, planes=128\n",
        "                                         )\n",
        "        \n",
        "        self.block3_down = UnetDownBlock(\n",
        "                                         predownsample_block=self.predownsample_block,\n",
        "                                         inplanes=128, planes=256\n",
        "                                         )\n",
        "\n",
        "        self.block4_down = UnetDownBlock(\n",
        "                                         predownsample_block=self.predownsample_block,\n",
        "                                         inplanes=256, planes=512\n",
        "                                         )\n",
        "        \n",
        "        self.block5_down = UnetDownBlock(\n",
        "                                         predownsample_block=self.predownsample_block,\n",
        "                                         inplanes=512, planes=1024\n",
        "                                         )\n",
        "        \n",
        "        self.block1_up = nn.ConvTranspose2d(in_channels=1024, out_channels=512,\n",
        "                                                  kernel_size=2, stride=2)\n",
        "        \n",
        "        self.block2_up = UnetUpBlock(\n",
        "                                     inplanes=1024, planes=512\n",
        "                                     )\n",
        "        \n",
        "        self.block3_up = UnetUpBlock(\n",
        "                                     inplanes=512, planes=256\n",
        "                                     )\n",
        "        \n",
        "        self.block4_up = UnetUpBlock(\n",
        "                                     inplanes=256, planes=128\n",
        "                                     )\n",
        "        \n",
        "        self.block5 = UnetUpBlock(\n",
        "                                  inplanes=128, planes=64,\n",
        "                                  postupsample_block=self.identity_block\n",
        "                                  )\n",
        "        \n",
        "        self.logit_conv = nn.Conv2d(\n",
        "                                    in_channels=64, out_channels=2, kernel_size=1,\n",
        "                                    )\n",
        "        \n",
        "        \n",
        "    def forward(self, x):\n",
        "        \n",
        "        features_1s_down = self.block1(x)\n",
        "        features_2s_down = self.block2_down(features_1s_down)\n",
        "        features_4s_down = self.block3_down(features_2s_down)\n",
        "        features_8s_down = self.block4_down(features_4s_down)\n",
        "        \n",
        "        features_16s = self.block5_down(features_8s_down)\n",
        "        \n",
        "        features_8s_up = self.block1_up(features_16s)\n",
        "        features_8s_up = torch.cat([features_8s_down, features_8s_up],dim=1)\n",
        "        \n",
        "        features_4s_up = self.block2_up(features_8s_up)\n",
        "        features_4s_up = torch.cat([features_4s_down, features_4s_up],dim=1)\n",
        "        \n",
        "        features_2s_up = self.block3_up(features_4s_up)\n",
        "        features_2s_up = torch.cat([features_2s_down, features_2s_up],dim=1)\n",
        "        \n",
        "        features_1s_up = self.block4_up(features_2s_up)\n",
        "        features_1s_up = torch.cat([features_1s_down, features_1s_up],dim=1)\n",
        "        \n",
        "        features_final = self.block5(features_1s_up)\n",
        "        \n",
        "        logits = self.logit_conv(features_final)\n",
        "       \n",
        "        return logits"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iQIYoJMie2S2",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class VGG(nn.Module):\n",
        "    \n",
        "    def __init__(self, features, num_classes=1000, init_weights=True):\n",
        "        super(VGG, self).__init__()\n",
        "        self.features = features\n",
        "        self.avgpool = nn.AdaptiveAvgPool2d((7, 7))\n",
        "        self.classifier = nn.Sequential(\n",
        "            nn.Linear(512 * 7 * 7, 4096),\n",
        "            nn.ReLU(True),\n",
        "            nn.Dropout(),\n",
        "            nn.Linear(4096, 4096),\n",
        "            nn.ReLU(True),\n",
        "            nn.Dropout(),\n",
        "            nn.Linear(4096, num_classes),\n",
        "        )\n",
        "        if init_weights:\n",
        "            self._initialize_weights()\n",
        "\n",
        "    def forward(self, x):\n",
        "        x = self.features(x)\n",
        "        x = self.avgpool(x)\n",
        "        x = x.view(x.size(0), -1)\n",
        "        x = self.classifier(x)\n",
        "        return x\n",
        "\n",
        "    def _initialize_weights(self):\n",
        "        for m in self.modules():\n",
        "            if isinstance(m, nn.Conv2d):\n",
        "                nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')\n",
        "                if m.bias is not None:\n",
        "                    nn.init.constant_(m.bias, 0)\n",
        "            elif isinstance(m, nn.BatchNorm2d):\n",
        "                nn.init.constant_(m.weight, 1)\n",
        "                nn.init.constant_(m.bias, 0)\n",
        "            elif isinstance(m, nn.Linear):\n",
        "                nn.init.normal_(m.weight, 0, 0.01)\n",
        "                nn.init.constant_(m.bias, 0)\n",
        "\n",
        "\n",
        "def make_layers(cfg, batch_norm=False):\n",
        "    layers = []\n",
        "    in_channels = 2\n",
        "    for v in cfg:\n",
        "        if v == 'M':\n",
        "            layers += [nn.MaxPool2d(kernel_size=2, stride=2)]\n",
        "        else:\n",
        "            conv2d = nn.Conv2d(in_channels, v, kernel_size=3, padding=1)\n",
        "            if batch_norm:\n",
        "                layers += [conv2d, nn.BatchNorm2d(v), nn.ReLU(inplace=True)]\n",
        "            else:\n",
        "                layers += [conv2d, nn.ReLU(inplace=True)]\n",
        "            in_channels = v\n",
        "    return nn.Sequential(*layers)\n",
        "\n",
        "cfgs = {\n",
        "    'A': [64, 'M', 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],\n",
        "    'B': [64, 64, 'M', 128, 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],\n",
        "    'D': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 'M', 512, 512, 512, 'M', 512, 512, 512, 'M'],\n",
        "    'E': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 256, 'M', 512, 512, 512, 512, 'M', 512, 512, 512, 512, 'M'],\n",
        "}\n",
        "\n",
        "def _vgg(arch, cfg, batch_norm, pretrained, progress, **kwargs):\n",
        "    model = VGG(make_layers(cfgs[cfg], batch_norm=batch_norm), **kwargs)\n",
        "    return model\n",
        "\n",
        "def vgg16(pretrained=False, progress=True, **kwargs):\n",
        "    \"\"\"VGG 16-layer model (configuration \"D\")\n",
        "\n",
        "    Args:\n",
        "        pretrained (bool): If True, returns a model pre-trained on ImageNet\n",
        "        progress (bool): If True, displays a progress bar of the download to stderr\n",
        "    \"\"\"\n",
        "    return _vgg('vgg16', 'D', False, pretrained, progress, **kwargs)\n",
        "\n",
        "  \n",
        "class VGG16(nn.Module):\n",
        "    \n",
        "    def __init__(self, n_layers, usegpu=True):\n",
        "        super(VGG16,self).__init__()\n",
        "        \n",
        "        self.cnn = vgg16(pretrained=False)\n",
        "        self.cnn = nn.Sequential(*list(self.cnn.children())[0])\n",
        "        self.cnn = nn.Sequential(*list(self.cnn.children())[:n_layers])\n",
        "        \n",
        "    def __get_outputs(self,x):\n",
        "        \n",
        "        outputs = []\n",
        "        for i, layer in enumerate(self.cnn.children()):\n",
        "            x = layer(x)\n",
        "            outputs.append(x)\n",
        "            \n",
        "        return outputs\n",
        "    \n",
        "    def forward(self,x):\n",
        "        outputs = self.__get_outputs(x)\n",
        "        \n",
        "        return outputs[-1]\n",
        "   \n",
        "    \n",
        "class SkipVGG16(nn.Module):\n",
        "    \n",
        "    def __init__(self, usegpu=True):\n",
        "        super(SkipVGG16, self).__init__()\n",
        "        \n",
        "        self.outputs = [3,8]\n",
        "        self.n_filters = [64,128]\n",
        "        \n",
        "        self.model = VGG16(n_layers=16, usegpu=usegpu)\n",
        "        \n",
        "    def forward(self,x):\n",
        "        \n",
        "        out = []\n",
        "        for i, layer in enumerate(list(self.model.children())[0]):\n",
        "            x = layer(x)\n",
        "            if i in self.outputs:\n",
        "                out.append(x)\n",
        "        out.append(x)\n",
        "        \n",
        "        return out\n",
        "    \n",
        "    \n",
        "class ReNet(nn.Module):\n",
        "    \n",
        "    def __init__(self, n_input, n_units, patch_size=(1, 1), usegpu=True):\n",
        "        super(ReNet, self).__init__()\n",
        "        \n",
        "        self.usegpu=usegpu\n",
        "        \n",
        "        self.patch_size_height = int(patch_size[0])\n",
        "        self.patch_size_width = int(patch_size[1])\n",
        "        \n",
        "        assert self.patch_size_height >= 1\n",
        "        assert self.patch_size_width >= 1\n",
        "        \n",
        "        self.tiling = False if ((self.patch_size_height == 1) and (\n",
        "            self.patch_size_width == 1)) else True\n",
        "                \n",
        "        rnn_hor_n_inputs = n_input * self.patch_size_height * \\\n",
        "            self.patch_size_width\n",
        "            \n",
        "        self.rnn_hor = nn.GRU(rnn_hor_n_inputs, n_units,\n",
        "                              num_layers=1, batch_first=True,\n",
        "                              bidirectional=True)\n",
        "        \n",
        "        self.rnn_ver = nn.GRU(n_units * 2, n_units,\n",
        "                              num_layers=1, batch_first=True,\n",
        "                              bidirectional=True)\n",
        "        \n",
        "    def __tile(self,x):\n",
        "\n",
        "        if (x.size(2) % self.patch_size_height) == 0:\n",
        "            n_height_padding = 0\n",
        "        else:\n",
        "            n_height_padding = self.patch_size_height - \\\n",
        "                x.size(2) % self.patch_size_height\n",
        "        if (x.size(3) % self.patch_size_width) == 0:\n",
        "            n_width_padding = 0\n",
        "        else:\n",
        "            n_width_padding = self.patch_size_width - \\\n",
        "                x.size(3) % self.patch_size_width\n",
        "\n",
        "        n_top_padding = n_height_padding / 2\n",
        "        n_bottom_padding = n_height_padding - n_top_padding\n",
        "\n",
        "        n_left_padding = n_width_padding / 2\n",
        "        n_right_padding = n_width_padding - n_left_padding\n",
        "\n",
        "        x = F.pad(x, (n_left_padding, n_right_padding,\n",
        "                      n_top_padding, n_bottom_padding))\n",
        "\n",
        "        b, n_filters, n_height, n_width = x.size()\n",
        "\n",
        "        assert n_height % self.patch_size_height == 0\n",
        "        assert n_width % self.patch_size_width == 0\n",
        "\n",
        "        new_height = n_height / self.patch_size_height\n",
        "        new_width = n_width / self.patch_size_width\n",
        "\n",
        "        x = x.view(b, n_filters, new_height, self.patch_size_height,\n",
        "                   new_width, self.patch_size_width)\n",
        "        x = x.permute(0, 2, 4, 1, 3, 5)\n",
        "        x = x.contiguous()\n",
        "        x = x.view(b, new_height, new_width, self.patch_size_height *\n",
        "                   self.patch_size_width * n_filters)\n",
        "        x = x.permute(0, 3, 1, 2)\n",
        "        x = x.contiguous()\n",
        "\n",
        "        return x\n",
        "                \n",
        "    def __swap_hw(self, x):\n",
        "\n",
        "        # x : b, nf, h, w\n",
        "        x = x.permute(0, 1, 3, 2)\n",
        "        x = x.contiguous()\n",
        "        #  x : b, nf, w, h\n",
        "\n",
        "        return x\n",
        "    \n",
        "    def rnn_forward(self, x, hor_or_ver):\n",
        "\n",
        "        # x : b, nf, h, w\n",
        "        assert hor_or_ver in ['hor', 'ver']\n",
        "\n",
        "        if hor_or_ver == 'ver':\n",
        "            x = self.__swap_hw(x)\n",
        "\n",
        "        x = x.permute(0, 2, 3, 1)\n",
        "        x = x.contiguous()\n",
        "        b, n_height, n_width, n_filters = x.size()\n",
        "        # x : b, h, w, nf\n",
        "\n",
        "        x = x.view(b * n_height, n_width, n_filters)\n",
        "        # x : b * h, w, nf\n",
        "        if hor_or_ver == 'hor':\n",
        "            x, _ = self.rnn_hor(x)\n",
        "        elif hor_or_ver == 'ver':\n",
        "            x, _ = self.rnn_ver(x)\n",
        "            \n",
        "        x = x.contiguous()\n",
        "        x = x.view(b, n_height, n_width, -1)\n",
        "        # x : b, h, w, nf\n",
        "\n",
        "        x = x.permute(0, 3, 1, 2)\n",
        "        x = x.contiguous()\n",
        "        # x : b, nf, h, w\n",
        "\n",
        "        if hor_or_ver == 'ver':\n",
        "            x = self.__swap_hw(x)\n",
        "\n",
        "        return x\n",
        "    \n",
        "    def forward(self, x):\n",
        "\n",
        "        # x : b, nf, h, w\n",
        "        if self.tiling:\n",
        "            x = self.__tile(x)\n",
        "\n",
        "        x = self.rnn_forward(x, 'hor')\n",
        "        x = self.rnn_forward(x, 'ver')\n",
        "\n",
        "        return x\n",
        "        \n",
        "\n",
        "class ReSeg(nn.Module):\n",
        "    \n",
        "    def __init__(self, usegpu=True):\n",
        "        super(ReSeg, self).__init__()\n",
        "        \n",
        "        self.cnn = SkipVGG16(usegpu=usegpu)\n",
        "        \n",
        "        self.renet1 = ReNet(256, 100, usegpu=usegpu)\n",
        "        self.renet2 = ReNet(200, 100, usegpu=usegpu)\n",
        "        \n",
        "        self.upsampling1 = nn.ConvTranspose2d(200, 100,\n",
        "                                              kernel_size=2,stride=2)\n",
        "        self.relu1 = nn.ReLU()\n",
        "        self.upsampling2 = nn.ConvTranspose2d(100 + self.cnn.n_filters[1], 100,\n",
        "                                              kernel_size=2,stride=2)\n",
        "        self.relu2 = nn.ReLU()\n",
        "        \n",
        "        self.final = nn.Conv2d(\n",
        "                                in_channels = 100 + self.cnn.n_filters[0], \n",
        "                                out_channels = 1,\n",
        "                                kernel_size=1,stride=1)\n",
        "        \n",
        "    def forward(self, x):\n",
        "        \n",
        "        first_skip, second_skip, x_enc = self.cnn(x)\n",
        "        x_enc = self.renet1(x_enc)\n",
        "        x_enc = self.renet2(x_enc)\n",
        "        x_dec = self.relu1(self.upsampling1(x_enc))\n",
        "        x_dec = torch.cat((x_dec, second_skip), dim=1)\n",
        "        x_dec = self.relu2(self.upsampling2(x_dec))\n",
        "        x_dec = torch.cat((x_dec, first_skip), dim=1)\n",
        "        x_out = self.final(x_dec)\n",
        "        \n",
        "        return x_out"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XNvBqABHe47S",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class VGG(nn.Module):\n",
        "    \n",
        "    def __init__(self, features, num_classes=1000, init_weights=True):\n",
        "        super(VGG, self).__init__()\n",
        "        self.features = features\n",
        "        self.avgpool = nn.AdaptiveAvgPool2d((7, 7))\n",
        "        self.classifier = nn.Sequential(\n",
        "            nn.Linear(512 * 7 * 7, 4096),\n",
        "            nn.ReLU(True),\n",
        "            nn.Dropout(),\n",
        "            nn.Linear(4096, 4096),\n",
        "            nn.ReLU(True),\n",
        "            nn.Dropout(),\n",
        "            nn.Linear(4096, num_classes),\n",
        "        )\n",
        "        if init_weights:\n",
        "            self._initialize_weights()\n",
        "\n",
        "    def forward(self, x):\n",
        "        x = self.features(x)\n",
        "        x = self.avgpool(x)\n",
        "        x = x.view(x.size(0), -1)\n",
        "        x = self.classifier(x)\n",
        "        return x\n",
        "\n",
        "    def _initialize_weights(self):\n",
        "        for m in self.modules():\n",
        "            if isinstance(m, nn.Conv2d):\n",
        "                nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')\n",
        "                if m.bias is not None:\n",
        "                    nn.init.constant_(m.bias, 0)\n",
        "            elif isinstance(m, nn.BatchNorm2d):\n",
        "                nn.init.constant_(m.weight, 1)\n",
        "                nn.init.constant_(m.bias, 0)\n",
        "            elif isinstance(m, nn.Linear):\n",
        "                nn.init.normal_(m.weight, 0, 0.01)\n",
        "                nn.init.constant_(m.bias, 0)\n",
        "\n",
        "\n",
        "def make_layers(cfg, batch_norm=False):\n",
        "    layers = []\n",
        "    in_channels = 2\n",
        "    for v in cfg:\n",
        "        if v == 'M':\n",
        "            layers += [nn.MaxPool2d(kernel_size=2, stride=2)]\n",
        "        else:\n",
        "            conv2d = nn.Conv2d(in_channels, v, kernel_size=3, padding=1)\n",
        "            if batch_norm:\n",
        "                layers += [conv2d, nn.BatchNorm2d(v), nn.ReLU(inplace=True)]\n",
        "            else:\n",
        "                layers += [conv2d, nn.ReLU(inplace=True)]\n",
        "            in_channels = v\n",
        "    return nn.Sequential(*layers)\n",
        "\n",
        "cfgs = {\n",
        "    'A': [64, 'M', 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],\n",
        "    'B': [64, 64, 'M', 128, 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],\n",
        "    'D': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 'M', 512, 512, 512, 'M', 512, 512, 512, 'M'],\n",
        "    'E': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 256, 'M', 512, 512, 512, 512, 'M', 512, 512, 512, 512, 'M'],\n",
        "}\n",
        "\n",
        "def _vgg(arch, cfg, batch_norm, pretrained, progress, **kwargs):\n",
        "    model = VGG(make_layers(cfgs[cfg], batch_norm=batch_norm), **kwargs)\n",
        "    return model\n",
        "\n",
        "def vgg16(pretrained=False, progress=True, **kwargs):\n",
        "    \"\"\"VGG 16-layer model (configuration \"D\")\n",
        "\n",
        "    Args:\n",
        "        pretrained (bool): If True, returns a model pre-trained on ImageNet\n",
        "        progress (bool): If True, displays a progress bar of the download to stderr\n",
        "    \"\"\"\n",
        "    return _vgg('vgg16', 'D', False, pretrained, progress, **kwargs)\n",
        "\n",
        "  \n",
        "class VGG16(nn.Module):\n",
        "    \n",
        "    def __init__(self, n_layers, usegpu=True):\n",
        "        super(VGG16,self).__init__()\n",
        "        \n",
        "        self.cnn = vgg16(pretrained=False)\n",
        "        self.cnn = nn.Sequential(*list(self.cnn.children())[0])\n",
        "        self.cnn = nn.Sequential(*list(self.cnn.children())[:n_layers])\n",
        "        \n",
        "    def __get_outputs(self,x):\n",
        "        \n",
        "        outputs = []\n",
        "        for i, layer in enumerate(self.cnn.children()):\n",
        "            x = layer(x)\n",
        "            outputs.append(x)\n",
        "            \n",
        "        return outputs\n",
        "    \n",
        "    def forward(self,x):\n",
        "        outputs = self.__get_outputs(x)\n",
        "        \n",
        "        return outputs[-1]\n",
        "  \n",
        "    \n",
        "class ReNet(nn.Module):\n",
        "    \n",
        "    def __init__(self, n_input, n_units, patch_size=(1, 1), usegpu=True):\n",
        "        super(ReNet, self).__init__()\n",
        "        \n",
        "        self.usegpu=usegpu\n",
        "        \n",
        "        self.patch_size_height = int(patch_size[0])\n",
        "        self.patch_size_width = int(patch_size[1])\n",
        "        \n",
        "        assert self.patch_size_height >= 1\n",
        "        assert self.patch_size_width >= 1\n",
        "        \n",
        "        self.tiling = False if ((self.patch_size_height == 1) and (\n",
        "            self.patch_size_width == 1)) else True\n",
        "                \n",
        "        rnn_hor_n_inputs = n_input * self.patch_size_height * \\\n",
        "            self.patch_size_width\n",
        "            \n",
        "        self.rnn_hor = nn.GRU(rnn_hor_n_inputs, n_units,\n",
        "                              num_layers=1, batch_first=True,\n",
        "                              bidirectional=True)\n",
        "        \n",
        "        self.rnn_ver = nn.GRU(n_units * 2, n_units,\n",
        "                              num_layers=1, batch_first=True,\n",
        "                              bidirectional=True)\n",
        "        \n",
        "    def __tile(self,x):\n",
        "\n",
        "        if (x.size(2) % self.patch_size_height) == 0:\n",
        "            n_height_padding = 0\n",
        "        else:\n",
        "            n_height_padding = self.patch_size_height - \\\n",
        "                x.size(2) % self.patch_size_height\n",
        "        if (x.size(3) % self.patch_size_width) == 0:\n",
        "            n_width_padding = 0\n",
        "        else:\n",
        "            n_width_padding = self.patch_size_width - \\\n",
        "                x.size(3) % self.patch_size_width\n",
        "\n",
        "        n_top_padding = n_height_padding / 2\n",
        "        n_bottom_padding = n_height_padding - n_top_padding\n",
        "\n",
        "        n_left_padding = n_width_padding / 2\n",
        "        n_right_padding = n_width_padding - n_left_padding\n",
        "\n",
        "        x = F.pad(x, (n_left_padding, n_right_padding,\n",
        "                      n_top_padding, n_bottom_padding))\n",
        "\n",
        "        b, n_filters, n_height, n_width = x.size()\n",
        "\n",
        "        assert n_height % self.patch_size_height == 0\n",
        "        assert n_width % self.patch_size_width == 0\n",
        "\n",
        "        new_height = n_height / self.patch_size_height\n",
        "        new_width = n_width / self.patch_size_width\n",
        "\n",
        "        x = x.view(b, n_filters, new_height, self.patch_size_height,\n",
        "                   new_width, self.patch_size_width)\n",
        "        x = x.permute(0, 2, 4, 1, 3, 5)\n",
        "        x = x.contiguous()\n",
        "        x = x.view(b, new_height, new_width, self.patch_size_height *\n",
        "                   self.patch_size_width * n_filters)\n",
        "        x = x.permute(0, 3, 1, 2)\n",
        "        x = x.contiguous()\n",
        "\n",
        "        return x\n",
        "                \n",
        "    def __swap_hw(self, x):\n",
        "\n",
        "        # x : b, nf, h, w\n",
        "        x = x.permute(0, 1, 3, 2)\n",
        "        x = x.contiguous()\n",
        "        #  x : b, nf, w, h\n",
        "\n",
        "        return x\n",
        "    \n",
        "    def rnn_forward(self, x, hor_or_ver):\n",
        "\n",
        "        # x : b, nf, h, w\n",
        "        assert hor_or_ver in ['hor', 'ver']\n",
        "\n",
        "        if hor_or_ver == 'ver':\n",
        "            x = self.__swap_hw(x)\n",
        "\n",
        "        x = x.permute(0, 2, 3, 1)\n",
        "        x = x.contiguous()\n",
        "        b, n_height, n_width, n_filters = x.size()\n",
        "        # x : b, h, w, nf\n",
        "\n",
        "        x = x.view(b * n_height, n_width, n_filters)\n",
        "        # x : b * h, w, nf\n",
        "        if hor_or_ver == 'hor':\n",
        "            x, _ = self.rnn_hor(x)\n",
        "        elif hor_or_ver == 'ver':\n",
        "            x, _ = self.rnn_ver(x)\n",
        "            \n",
        "        x = x.contiguous()\n",
        "        x = x.view(b, n_height, n_width, -1)\n",
        "        # x : b, h, w, nf\n",
        "\n",
        "        x = x.permute(0, 3, 1, 2)\n",
        "        x = x.contiguous()\n",
        "        # x : b, nf, h, w\n",
        "\n",
        "        if hor_or_ver == 'ver':\n",
        "            x = self.__swap_hw(x)\n",
        "\n",
        "        return x\n",
        "    \n",
        "    def forward(self, x):\n",
        "\n",
        "        # x : b, nf, h, w\n",
        "        if self.tiling:\n",
        "            x = self.__tile(x)\n",
        "\n",
        "        x = self.rnn_forward(x, 'hor')\n",
        "        x = self.rnn_forward(x, 'ver')\n",
        "\n",
        "        return x\n",
        "\n",
        "\n",
        "class ConvGRUCell(nn.Module):\n",
        "    \n",
        "    def __init__(self, input_size, hidden_size, kernel_size, usegpu=True):\n",
        "        super(ConvGRUCell, self).__init__()\n",
        "        \n",
        "        self.input_size = input_size\n",
        "        self.hidden_size = hidden_size\n",
        "        self.kernel_size = kernel_size\n",
        "        self.usegpu = usegpu\n",
        "        \n",
        "        _n_inputs = self.input_size + self.hidden_size\n",
        "        self.conv_gates = nn.Conv2d(_n_inputs,\n",
        "                                    2 * self.hidden_size,\n",
        "                                    self.kernel_size,\n",
        "                                    padding=self.kernel_size // 2)\n",
        "\n",
        "        self.conv_ct = nn.Conv2d(_n_inputs, self.hidden_size,\n",
        "                                 self.kernel_size,\n",
        "                                 padding=self.kernel_size // 2)\n",
        "        \n",
        "    def forward(self, x, hidden):\n",
        "        \n",
        "        batch_size, _, height, width = x.size()\n",
        "        \n",
        "        if hidden is None:\n",
        "            size_h = [batch_size, self.hidden_size, height, width]\n",
        "            hidden = Variable(torch.zeros(size_h))\n",
        "\n",
        "            if self.usegpu:\n",
        "                hidden = hidden.cuda()\n",
        "                \n",
        "        c1 = self.conv_gates(torch.cat((x, hidden), dim=1))\n",
        "        rt, ut = c1.chunk(2, 1)\n",
        "\n",
        "        reset_gate = torch.sigmoid(rt)\n",
        "        update_gate = torch.sigmoid(ut)\n",
        "\n",
        "        gated_hidden = torch.mul(reset_gate, hidden)\n",
        "\n",
        "        ct = torch.tanh(self.conv_ct(torch.cat((x, gated_hidden), dim=1)))\n",
        "\n",
        "        next_h = torch.mul(update_gate, hidden) + (1 - update_gate) * ct\n",
        "\n",
        "        return next_h\n",
        "    \n",
        "\n",
        "class RecurrentHourglass(nn.Module):\n",
        "    \n",
        "    def __init__(self, input_n_filters, hidden_n_filters, kernel_size,\n",
        "                 n_levels, embedding_size, usegpu=True):\n",
        "        super(RecurrentHourglass, self).__init__()\n",
        "            \n",
        "        assert n_levels >= 1\n",
        "    \n",
        "        self.input_n_filters = input_n_filters\n",
        "        self.hidden_n_filters = hidden_n_filters\n",
        "        self.kernel_size = kernel_size\n",
        "        self.n_levels = n_levels\n",
        "        self.embedding_size = embedding_size\n",
        "        self.usegpu = usegpu\n",
        "        \n",
        "        self.convgru_cell = ConvGRUCell(self.hidden_n_filters,\n",
        "                                        self.hidden_n_filters,\n",
        "                                        self.kernel_size,\n",
        "                                        self.usegpu)\n",
        "        \n",
        "        self.__generate_pre_post_convs()\n",
        "        \n",
        "    def __generate_pre_post_convs(self):\n",
        "        \n",
        "        def __get_conv(input_n_filters, output_n_filters):\n",
        "            return nn.Conv2d(input_n_filters, output_n_filters,\n",
        "                             self.kernel_size,\n",
        "                             padding=self.kernel_size // 2)\n",
        "            \n",
        "        self.pre_conv_layers = [__get_conv(self.input_n_filters,\n",
        "                                           self.hidden_n_filters), ]\n",
        "    \n",
        "        for _ in range(self.n_levels - 1):\n",
        "            self.pre_conv_layers.append(__get_conv(self.hidden_n_filters,\n",
        "                                                   self.hidden_n_filters))\n",
        "        self.pre_conv_layers = nn.ModuleList(self.pre_conv_layers)\n",
        "    \n",
        "        self.post_conv_layers = [__get_conv(self.hidden_n_filters,\n",
        "                                            self.embedding_size), ]\n",
        "        for _ in range(self.n_levels - 1):\n",
        "            self.post_conv_layers.append(__get_conv(self.hidden_n_filters,\n",
        "                                                    self.hidden_n_filters))\n",
        "        self.post_conv_layers = nn.ModuleList(self.post_conv_layers)\n",
        "    \n",
        "    def forward_encoding(self, x):\n",
        "        \n",
        "        convgru_outputs = []\n",
        "        hidden = None\n",
        "        for i in range(self.n_levels):\n",
        "            x = F.relu(self.pre_conv_layers[i](x))\n",
        "            hidden = self.convgru_cell(x, hidden)\n",
        "            convgru_outputs.append(hidden)\n",
        "            \n",
        "        return convgru_outputs\n",
        "    \n",
        "    def forward_decoding(self, convgru_outputs):\n",
        "        \n",
        "        _last_conv_layer = self.post_conv_layers[self.n_levels - 1]\n",
        "        _last_output = convgru_outputs[self.n_levels - 1]\n",
        "        \n",
        "        post_feature_map = F.relu(_last_conv_layer(_last_output))\n",
        "        for i in range(self.n_levels - 1)[::-1]:\n",
        "            post_feature_map += convgru_outputs[i]\n",
        "            post_feature_map = self.post_conv_layers[i](post_feature_map)\n",
        "            post_feature_map = F.relu(post_feature_map)\n",
        "            \n",
        "        return post_feature_map\n",
        "    \n",
        "    def forward(self, x):\n",
        "        \n",
        "        x = self.forward_encoding(x)\n",
        "        x = self.forward_decoding(x)\n",
        "        \n",
        "        return x\n",
        "    \n",
        "\n",
        "class StackedRecurrentHourglass(nn.Module):\n",
        "    \n",
        "    def __init__(self, usegpu=True):\n",
        "        super(StackedRecurrentHourglass, self).__init__()\n",
        "        \n",
        "        self.usegpu = usegpu\n",
        "        \n",
        "        self.base_cnn = self.__generate_base_cnn()\n",
        "        \n",
        "        self.enc_stacked_hourglass = self.__generate_enc_stacked_hg(64,3)\n",
        "        \n",
        "        self.stacked_renet = self.__generate_stacked_renet(64,2)\n",
        "        \n",
        "        self.decoder = self.__generate_decoder(64)\n",
        "        \n",
        "    def __generate_base_cnn(self):\n",
        "        \n",
        "        base_cnn = VGG16(n_layers=4, usegpu=self.usegpu)\n",
        "        \n",
        "        return base_cnn\n",
        "    \n",
        "    def __generate_enc_stacked_hg(self, input_n_filters, n_levels):\n",
        "        \n",
        "        stacked_hourglass = nn.Sequential()\n",
        "        stacked_hourglass.add_module('Hourglass_1',\n",
        "                                     RecurrentHourglass(\n",
        "                                         input_n_filters=input_n_filters,\n",
        "                                         hidden_n_filters=64,\n",
        "                                         kernel_size=3,\n",
        "                                         n_levels=n_levels,\n",
        "                                         embedding_size=64,\n",
        "                                         usegpu=self.usegpu))\n",
        "        stacked_hourglass.add_module('pool_1',\n",
        "                                     nn.MaxPool2d(2, stride=2))\n",
        "        stacked_hourglass.add_module('Hourglass_2',\n",
        "                                     RecurrentHourglass(\n",
        "                                         input_n_filters=64,\n",
        "                                         hidden_n_filters=64,\n",
        "                                         kernel_size=3,\n",
        "                                         n_levels=n_levels,\n",
        "                                         embedding_size=64,\n",
        "                                         usegpu=self.usegpu))        \n",
        "        stacked_hourglass.add_module('pool_2',\n",
        "                                     nn.MaxPool2d(2, stride=2))    \n",
        "\n",
        "        return stacked_hourglass\n",
        "\n",
        "    def __generate_stacked_renet(self, input_n_filters, n_renets):\n",
        "\n",
        "        assert n_renets >= 1\n",
        "        \n",
        "        renet = nn.Sequential()\n",
        "        renet.add_module('ReNet_1', ReNet(input_n_filters, 32,\n",
        "                                          patch_size=(1, 1),\n",
        "                                          usegpu=self.usegpu))\n",
        "        for i in range(1, n_renets):\n",
        "            renet.add_module('ReNet_{}'.format(i + 1),\n",
        "                             ReNet(32 * 2, 32, patch_size=(1, 1),\n",
        "                                   usegpu=self.usegpu))\n",
        "            \n",
        "        return renet\n",
        "    \n",
        "    def __generate_decoder(self, input_n_filters):\n",
        "        \n",
        "        decoder = nn.Sequential()\n",
        "        decoder.add_module('ConvTranspose_1',\n",
        "                           nn.ConvTranspose2d(input_n_filters,\n",
        "                                              64,\n",
        "                                              kernel_size=(2, 2),\n",
        "                                              stride=(2, 2)))\n",
        "        decoder.add_module('ReLU_1', nn.ReLU())\n",
        "        decoder.add_module('ConvTranspose_2',\n",
        "                           nn.ConvTranspose2d(64,\n",
        "                                              64,\n",
        "                                              kernel_size=(2, 2),\n",
        "                                              stride=(2, 2)))\n",
        "        decoder.add_module('ReLU_2', nn.ReLU())\n",
        "        decoder.add_module('Final',\n",
        "                           nn.ConvTranspose2d(64,\n",
        "                                              1,\n",
        "                                              kernel_size=(1, 1),\n",
        "                                              stride=(1, 1)))\n",
        "        \n",
        "        return decoder\n",
        "            \n",
        "    def forward(self, x):\n",
        "        \n",
        "        x = self.base_cnn(x)\n",
        "        x = self.enc_stacked_hourglass(x)\n",
        "        x = self.stacked_renet(x)\n",
        "        x = self.decoder(x)\n",
        "        \n",
        "        return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uKFjxUWSfCSG",
        "colab_type": "text"
      },
      "source": [
        "# choose"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b5l_M-uve7TK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#net = Unet().to(device)\n",
        "net = ReSeg().to(device)\n",
        "#net = StackedRecurrentHourglass().to(device)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ethgDxJtfWDG",
        "colab_type": "text"
      },
      "source": [
        "# load data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iJxsAlK4fVIg",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def load_data_hdf(iphi):\n",
        "  \n",
        "  hf_f = h5py.File('/content/hdf5_data/hdf_f.h5','r')\n",
        "  hf_df = h5py.File('/content/hdf5_data/hdf_df.h5','r')\n",
        "  \n",
        "  e_f = hf_f['e_f'][iphi]  \n",
        "  i_f = hf_f['i_f'][iphi]\n",
        "  e_df = hf_df['e_df'][iphi] \n",
        "  i_df = hf_df['i_df'][iphi]\n",
        "  \n",
        "  ind1,ind2,ind3 = i_f.shape\n",
        "  \n",
        "  f = np.zeros([ind2,2,ind1,ind1])\n",
        "  df = np.zeros([ind2,2,ind1,ind1])\n",
        "\n",
        "  for n in range(ind2):\n",
        "    f[n,0,:,:-1] = e_f[:,n,:]\n",
        "    f[n,1,:,:-1] = i_f[:,n,:]\n",
        "    df[n,0,:,:-1] = e_df[:,n,:]\n",
        "    df[n,1,:,:-1] = i_df[:,n,:]\n",
        "\n",
        "    f[n,0,:,-1] = e_f[:,n,-1]\n",
        "    f[n,1,:,-1] = i_f[:,n,-1]\n",
        "    df[n,0,:,-1] = e_df[:,n,-1]\n",
        "    df[n,1,:,-1] = i_df[:,n,-1]\n",
        "    \n",
        "  del i_f,e_f,i_df,e_df\n",
        "  df+=f\n",
        "\n",
        "  # instantiate variables for conservation properties and for normalization\n",
        "  hf_cons = h5py.File('/content/hdf5_data/hdf_cons_fullvol.h5','r')\n",
        "  hf_vol = h5py.File('/content/hdf5_data/hdf_vol.h5')\n",
        "  cons = conservation_variables(hf_cons,hf_vol)\n",
        "  \n",
        "  hf_stats = h5py.File('/content/hdf5_data/hdf_stats.h5','r')\n",
        "  zvars = stats_variables(hf_stats)\n",
        "  \n",
        "  for n in range(ind2):\n",
        "    f[n] = (f[n]-zvars.mean_f)/zvars.std_f\n",
        "#     df[n] = (df[n]-zvars.mean_df)/zvars.std_df\n",
        "    df[n] = (df[n]-zvars.mean_fdf)/zvars.std_fdf\n",
        "    \n",
        "  zvars.mean_f = zvars.mean_f[np.newaxis]\n",
        "  zvars.mean_df = zvars.mean_df[np.newaxis]\n",
        "  zvars.mean_fdf = zvars.mean_fdf[np.newaxis]\n",
        "  zvars.std_f = zvars.std_f[np.newaxis]\n",
        "  zvars.std_df = zvars.std_df[np.newaxis]\n",
        "  zvars.std_fdf = zvars.std_fdf[np.newaxis]\n",
        "  \n",
        "  for i in range(int(np.ceil(np.log(batch_size)/np.log(2)))):\n",
        "    zvars.mean_f = np.concatenate((zvars.mean_f,zvars.mean_f),axis=0)\n",
        "    zvars.mean_df = np.concatenate((zvars.mean_df,zvars.mean_df),axis=0)\n",
        "    zvars.mean_fdf = np.concatenate((zvars.mean_fdf,zvars.mean_fdf),axis=0)\n",
        "    zvars.std_f = np.concatenate((zvars.std_f,zvars.std_f),axis=0)\n",
        "    zvars.std_df = np.concatenate((zvars.std_df,zvars.std_df),axis=0)  \n",
        "    zvars.std_fdf = np.concatenate((zvars.std_fdf,zvars.std_fdf),axis=0)\n",
        "  \n",
        "  zvars.mean_f = torch.from_numpy(zvars.mean_f).to(device).float()\n",
        "  zvars.mean_df = torch.from_numpy(zvars.mean_df).to(device).float()\n",
        "  zvars.mean_fdf = torch.from_numpy(zvars.mean_fdf).to(device).float()\n",
        "  zvars.std_f = torch.from_numpy(zvars.std_f).to(device).float()\n",
        "  zvars.std_df = torch.from_numpy(zvars.std_df).to(device).float()\n",
        "  zvars.std_fdf = torch.from_numpy(zvars.std_fdf).to(device).float()\n",
        "       \n",
        "  return f,df,ind2,zvars,cons"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uTgTZGvuqexO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class stats_variables():\n",
        "  \n",
        "  def __init__(self, hf_stats):\n",
        "    self.std_f = hf_stats['std_f'][...]\n",
        "    self.std_df = hf_stats['std_df'][...]\n",
        "    self.std_fdf = hf_stats['std_fdf'][...]\n",
        "    self.mean_f = hf_stats['mean_f'][...]\n",
        "    self.mean_df = hf_stats['mean_df'][...]\n",
        "    self.mean_fdf = hf_stats['mean_fdf'][...]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eiJ6jMtop2HQ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class conservation_variables():\n",
        "\n",
        "  def __init__(self, hf_cons, hf_vol):\n",
        "    self.f0_dsmu = hf_cons['f0_dsmu'][...]\n",
        "    self.f0_dvp = hf_cons['f0_dvp'][...]\n",
        "    self.f0_nvp = hf_cons['f0_nvp'][...]\n",
        "    self.f0_nmu = hf_cons['f0_nmu'][...]\n",
        "    self.ptl_mass = hf_cons['ptl_mass'][...]\n",
        "    self.sml_ev2j = 1.6022e-19\n",
        "    \n",
        "    self.temp = hf_cons['f0_T_ev'][...]\n",
        "    self.vol = np.zeros([self.temp.shape[0],self.f0_nmu+1,self.temp.shape[1]])\n",
        "    self.vol[0] = hf_vol['vole'][0]\n",
        "    self.vol[1] = hf_vol['voli'][0]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G5uuRPsRfZzQ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class DistFuncDataset(Dataset):\n",
        "\n",
        "    def __init__(self, f_array, df_array, temp_array, vol_array):\n",
        "        self.data = torch.from_numpy(f_array).float()\n",
        "        self.target = torch.from_numpy(df_array).float()\n",
        "        self.temp = torch.from_numpy(temp_array).float()\n",
        "        self.vol = torch.from_numpy(vol_array).float()\n",
        "        \n",
        "    def __len__(self):\n",
        "        return len(self.data)\n",
        "      \n",
        "    def __getitem__(self, index):\n",
        "        a = self.data[index]\n",
        "        b = self.target[index]\n",
        "        b = b.view(-1,32,32)\n",
        "        c = self.temp[index]\n",
        "        d = self.vol[index]\n",
        "            \n",
        "        return a, b, c, d"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2pGWpo_efdta",
        "colab_type": "text"
      },
      "source": [
        "# split data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qI4upfi6fdOJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def split_data(f,df,cons,num_nodes):\n",
        "    \n",
        "    inds = np.arange(num_nodes)\n",
        "    #np.random.seed(0)\n",
        "    np.random.shuffle(inds) \n",
        "    \n",
        "    num_train = int(np.floor(percentage_train*num_nodes))\n",
        "    num_val = int(np.floor(percentage_val*num_nodes))\n",
        "    \n",
        "    train_inds = inds[:num_train]\n",
        "    val_inds = inds[num_train:num_train+num_val]\n",
        "    test_inds = inds[num_train+num_val:]\n",
        " \n",
        "    f_train = f[train_inds]\n",
        "    f_val = f[val_inds]\n",
        "    f_test = f[test_inds]\n",
        "        \n",
        "    del f  \n",
        "      \n",
        "    df_train = df[train_inds]\n",
        "    df_val = df[val_inds]\n",
        "    df_test = df[test_inds]\n",
        "    \n",
        "    del df\n",
        "        \n",
        "    # temperature and volume separate arrays here\n",
        "    temp = np.einsum('ij -> ji', cons.temp)   \n",
        "    temp_train = temp[train_inds]\n",
        "    temp_val = temp[val_inds]\n",
        "    temp_test = temp[test_inds]\n",
        "    \n",
        "    del temp\n",
        "    \n",
        "    vol = np.einsum('ijk -> kji', cons.vol)\n",
        "    vol_train = vol[train_inds]\n",
        "    vol_val = vol[val_inds]\n",
        "    vol_test = vol[test_inds]\n",
        "    \n",
        "    \n",
        "    trainset = DistFuncDataset(f_train, df_train, temp_train, vol_train)\n",
        "    \n",
        "    trainloader = DataLoader(trainset, batch_size=batch_size, \n",
        "                             shuffle=True, pin_memory=True, num_workers=4)\n",
        "    \n",
        "    del f_train, df_train, temp_train, vol_train\n",
        "    \n",
        "    valset = DistFuncDataset(f_val, df_val, temp_val, vol_val)\n",
        "    \n",
        "    valloader = DataLoader(valset, batch_size=batch_size, \n",
        "                           shuffle=True, pin_memory=True, num_workers=4)\n",
        "        \n",
        "    return trainloader, valloader, f_test, df_test, temp_test, vol_test"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "P2UY0jNifkTg",
        "colab_type": "text"
      },
      "source": [
        "# check props"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b-kcnQUAfcq_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# same procedure as col_f_convergence_eval\n",
        "def check_properties_each(f_slice, cons, temp, vol, sp):\n",
        "    \n",
        "    f_slice = f_slice.float()\n",
        "       \n",
        "    if len(f_slice.shape) == 2:\n",
        "      nperp, npar = f_slice.shape\n",
        "      nbatch = 1\n",
        "    elif len(f_slice.shape) == 3:  \n",
        "      nbatch,nperp,npar = f_slice.shape\n",
        "      \n",
        "    vth = torch.sqrt(temp*cons.sml_ev2j/cons.ptl_mass[sp]) \n",
        "    \n",
        "    vpar = np.linspace(-cons.f0_nvp,cons.f0_nvp,2*cons.f0_nvp+1)*cons.f0_dvp\n",
        "    vperp = np.linspace(0,cons.f0_nmu,cons.f0_nmu+1)*cons.f0_dsmu\n",
        "    \n",
        "    \n",
        "#     print('vth',vth*torch.from_numpy(cons.f0_dsmu))\n",
        "#     print('vpar',torch.from_numpy(vpar)*vth[0])\n",
        "#     print('vperp',torch.from_numpy(vperp)*vth[0])\n",
        "    \n",
        "    vpar = torch.tensor(vpar).float().to(device)\n",
        "    vperp = torch.tensor(vperp).float().to(device)\n",
        "        \n",
        "    mass = cons.ptl_mass[sp]\n",
        "    conv_factor_notemp = 1/np.sqrt((2*np.pi*cons.sml_ev2j/mass)**3)\n",
        "    temp_factor = 1/torch.sqrt(temp)\n",
        "    \n",
        "    smu_n = cons.f0_dsmu/3 # smu_n = f0_dsmu/f0_mu0_factor, f0_mu0_factor = 3\n",
        "    f_slice_norm = torch.einsum('ijk,i -> ijk',f_slice,temp_factor)*conv_factor_notemp/smu_n\n",
        "      \n",
        "    ones_tensor = torch.ones(nbatch,nperp,npar).float().to(device)\n",
        "      \n",
        "    vol_tensor = torch.einsum('ijk,ij -> ijk',ones_tensor,vol)\n",
        "    vperp_tensor = torch.einsum('ijk,i,j -> ijk',ones_tensor,vth,vperp)\n",
        "    vpar_tensor = torch.einsum('ijk,i,k -> ijk',ones_tensor,vth,vpar)\n",
        "    \n",
        "    mass_tensor = vol_tensor\n",
        "    mom_tensor = vpar_tensor*cons.ptl_mass[sp]*vol_tensor\n",
        "    energy_tensor = (vpar_tensor**2 + vperp_tensor**2)*cons.ptl_mass[sp]*vol_tensor\n",
        "        \n",
        "    mass_tensor, mom_tensor, energy_tensor = \\\n",
        "    mass_tensor.to(device), mom_tensor.to(device), energy_tensor.to(device)                   \n",
        "         \n",
        "    mass = torch.sum(f_slice_norm*mass_tensor, dim = (1,2))\n",
        "    momentum = torch.sum(f_slice_norm*mom_tensor, dim = (1,2))\n",
        "    energy = torch.sum(f_slice_norm*energy_tensor, dim = (1,2))\n",
        "                \n",
        "    return mass, momentum, energy"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cmbXaXVefmHR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# makes calls to individual df/f property calculation\n",
        "# computes more useful quantities as in col_f_core_m after the calls to col_f_convergence_eval \n",
        "def check_properties_main(f,df,temp,vol,cons):\n",
        "  \n",
        "  masse = torch.from_numpy(np.array([cons.ptl_mass[0]])).to(device).float()\n",
        "  massi = torch.from_numpy(np.array([cons.ptl_mass[1]])).to(device).float()\n",
        "  \n",
        "  #print('df')\n",
        "  dne,dpe,dwe = check_properties_each(df[:,0],cons,temp[:,0],vol[:,:,0],0)\n",
        "  dni,dpi,dwi = check_properties_each(df[:,1],cons,temp[:,1],vol[:,:,1],1)\n",
        "  \n",
        "  #print('f')  \n",
        "  ne,mome,ene = check_properties_each(f[:,0],cons,temp[:,0],vol[:,:,0],0)\n",
        "  ni,momi,eni = check_properties_each(f[:,1],cons,temp[:,1],vol[:,:,1],1) \n",
        "      \n",
        "  dne_n = torch.abs(dne/ne)\n",
        "  dni_n = torch.abs(dni/ni)\n",
        "  \n",
        "  dp_p = torch.abs(dpi + dpe)/torch.max(torch.abs(momi + mome),1e-3*torch.max(massi,masse)*ne)\n",
        "  \n",
        "  dw_w = torch.abs((dwi + dwe)/(eni + ene))\n",
        "  \n",
        "  return dne_n,dni_n,dp_p,dw_w"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KUqPFCeSpwYi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# dataiter = iter(trainloader)\n",
        "# data, targets, temp, vol = dataiter.next()\n",
        "# data, targets, temp, vol = data.to(device), targets.to(device), temp.to(device), vol.to(device)     \n",
        "\n",
        "# outputs = net(data)\n",
        "# outputs = outputs.to(device)\n",
        "\n",
        "# nbatch = len(data)     \n",
        "\n",
        "# data_unnorm = data*zvars.std_f[:nbatch] + zvars.mean_f[:nbatch]\n",
        "# targets_unnorm = targets*zvars.std_fdf[:nbatch] + zvars.mean_fdf[:nbatch]\n",
        "# outputs_unnorm = outputs[:nbatch,0]*zvars.std_fdf[:nbatch,1] + zvars.mean_fdf[:nbatch,1]\n",
        "\n",
        "# targets_nof = targets_unnorm - data_unnorm\n",
        "# outputs_nof = outputs_unnorm[:nbatch] - data_unnorm[:nbatch,1]\n",
        "\n",
        "# outputs_nof_to_cat = outputs_nof[:nbatch].unsqueeze(1)\n",
        "# targets_nof_to_cat = targets_nof[:nbatch,0].unsqueeze(1)\n",
        "\n",
        "# outputs_nof = torch.cat((outputs_nof_to_cat,targets_nof_to_cat),1)\n",
        "# check_properties_main(data_unnorm[:,:,:,:-1],outputs_nof[:,:,:,:-1],temp,vol,cons)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8ic44p6mfp-4",
        "colab_type": "text"
      },
      "source": [
        "# train"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Zyc8x31qfnuw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def train(trainloader,valloader,sp_flag,epoch,end,zvars,cons):\n",
        "  \n",
        "    props_before = []\n",
        "    props_after = []\n",
        "    train_loss_vector = []\n",
        "    l2_loss_vector = []\n",
        "    cons_loss_vector = []\n",
        "    val_loss_vector = []\n",
        "    \n",
        "    running_loss = 0.0\n",
        "    running_l2_loss = 0.0\n",
        "    running_cons_loss = 0.0\n",
        "    timestart = timeit.default_timer()\n",
        "    for i, (data, targets, temp, vol) in enumerate(trainloader):\n",
        "        timeend = timeit.default_timer()\n",
        "        #print(timeend-timestart)\n",
        "     \n",
        "        data, targets, temp, vol = data.to(device), targets.to(device), temp.to(device), vol.to(device)     \n",
        "      \n",
        "        if sp_flag == 0:\n",
        "            optimizer.zero_grad()\n",
        "        else:\n",
        "            optimizer_e.zero_grad()\n",
        "\n",
        "        outputs = net(data)\n",
        "        outputs = outputs.to(device)\n",
        "        \n",
        "        nbatch = len(data)     \n",
        "        \n",
        "        data_unnorm = data*zvars.std_f[:nbatch] + zvars.mean_f[:nbatch]\n",
        "        targets_unnorm = targets*zvars.std_fdf[:nbatch] + zvars.mean_fdf[:nbatch]\n",
        "        outputs_unnorm = outputs[:,0]*zvars.std_fdf[:nbatch,1] + zvars.mean_fdf[:nbatch,1]\n",
        "                   \n",
        "        targets_nof = targets_unnorm - data_unnorm\n",
        "        outputs_nof = outputs_unnorm[:nbatch] - data_unnorm[:nbatch,1]\n",
        "        \n",
        "        outputs_nof_to_cat = outputs_nof[:nbatch].unsqueeze(1)\n",
        "        targets_nof_to_cat = targets_nof[:nbatch,0].unsqueeze(1)        \n",
        "        \n",
        "        # concatenate with actual dfe\n",
        "        outputs_nof = torch.cat((outputs_nof_to_cat,targets_nof_to_cat),1)\n",
        "                  \n",
        "        # updated calls to check_properties with correct arguments  \n",
        "        masse_b,massi_b,mom_b,energy_b = check_properties_main(data_unnorm[:,:,:,:-1],\\\n",
        "                                                               targets_nof[:,:,:,:-1],temp,vol,cons)\n",
        "\n",
        "        masse_a,massi_a,mom_a,energy_a = check_properties_main(data_unnorm[:,:,:,:-1],\\\n",
        "                                                               outputs_nof[:,:,:,:-1],temp,vol,cons)\n",
        "                       \n",
        "        props_before.append([(torch.sum(massi_b+masse_b)/nbatch).item(),\\\n",
        "                             (torch.sum(mom_b)/nbatch).item(),\\\n",
        "                             (torch.sum(energy_b)/nbatch).item()])\n",
        "        props_after.append([torch.sum((massi_a+massi_b)/nbatch).item(),\\\n",
        "                             torch.sum((mom_a)/nbatch).item(),\\\n",
        "                             torch.sum((energy_a)/nbatch).item()])\n",
        "        \n",
        "        masse_loss = torch.sum(masse_a)\n",
        "        massi_loss = torch.sum(massi_a)\n",
        "        mom_loss = torch.sum(mom_a)\n",
        "        energy_loss = torch.sum(energy_a)\n",
        "\n",
        "        # if i % 200 == 199:\n",
        "        #   print('outputs',masse_loss.item(),massi_loss.item(),mom_loss.item(),energy_loss.item())\n",
        "                \n",
        "        # masse_loss = torch.sum(torch.abs(masse_a - masse_b)).float()/nbatch\n",
        "        # massi_loss = torch.sum(torch.abs(massi_a - massi_b)).float()/nbatch\n",
        "        # mom_loss = torch.sum(torch.abs(mom_a - mom_b)).float()/nbatch\n",
        "        # energy_loss = torch.sum(torch.abs(energy_a - energy_b)).float()/nbatch    \n",
        "                \n",
        "        l2_loss = criterion(outputs[:,0],targets[:,1])  \n",
        "                  \n",
        "#         loss = l2_loss*loss_weights[0] \\\n",
        "#              + mass_loss*loss_weights[1] \\\n",
        "#              + mom_loss*loss_weights[2] \\\n",
        "#              + energy_loss*loss_weights[3]    \n",
        "        \n",
        "#         cons_loss = mass_loss*loss_weights[1] \\\n",
        "#                   + mom_loss*loss_weights[2] \\\n",
        "#                   + energy_loss*loss_weights[3] \\\n",
        "\n",
        "        loss = l2_loss*loss_weights[0]\\\n",
        "             + masse_loss*loss_weights[1]\\\n",
        "             + massi_loss*loss_weights[2]\\\n",
        "             + mom_loss*loss_weights[3]\\\n",
        "             + energy_loss*loss_weights[4]\n",
        "  \n",
        "        cons_loss = masse_loss*loss_weights[1]\\\n",
        "                  + massi_loss*loss_weights[2]\\\n",
        "                  + mom_loss*loss_weights[3]\\\n",
        "                  + energy_loss*loss_weights[4]\n",
        "  \n",
        "        loss.backward()\n",
        "        if sp_flag == 0:\n",
        "            optimizer.step()\n",
        "        else:\n",
        "            optimizer_e.step()\n",
        "\n",
        "        running_loss += loss.item()\n",
        "        running_l2_loss += l2_loss.item()\n",
        "        running_cons_loss += cons_loss.item()\n",
        "        \n",
        "        if i % output_rate == output_rate-1:\n",
        "            print('   [%d, %5d] loss: %.6f' %\n",
        "                  (epoch + 1, end + i + 1, running_loss / output_rate))\n",
        "            print('      L2 loss: %.6f' % (running_l2_loss / output_rate))\n",
        "            print('      conservation loss: %.6f' % (running_cons_loss / output_rate))\n",
        "            \n",
        "#             print('outputs',mass_a[0].item(),mom_a[0].item(),energy_a[0].item())\n",
        "#             print('targets',mass_b[0].item(),mom_b[0].item(),energy_b[0].item())\n",
        "          \n",
        "        if i % plot_rate == plot_rate-1:\n",
        "            train_loss_vector.append(running_loss / output_rate)\n",
        "            l2_loss_vector.append(running_l2_loss / output_rate)\n",
        "            cons_loss_vector.append(running_cons_loss / output_rate)\n",
        "            running_loss = 0.0\n",
        "            running_l2_loss = 0.0\n",
        "            running_cons_loss = 0.0\n",
        "            #plot_df(targets_unnorm[0,0,:,:-1],outputs_unnorm[0,0,:,:-1],epoch)\n",
        "        \n",
        "        if i % val_rate == val_rate-1:         \n",
        "          val_loss = validate(valloader,cons,zvars)\n",
        "          val_loss_vector.append(val_loss)\n",
        "        \n",
        "          is_best = False\n",
        "          if val_loss < np.min(val_loss_vector): ## check this\n",
        "            is_best = True \n",
        "\n",
        "          if i % (2*val_rate) == (2*val_rate-1):\n",
        "            save_checkpoint({\n",
        "                             'epoch': epoch+1,\n",
        "                             'state_dict': net.state_dict(),\n",
        "                             'val_loss': val_loss,\n",
        "                             'optimizer': optimizer.state_dict(),\n",
        "                             }, is_best, lr)\n",
        "\n",
        "        timestart = timeit.default_timer()  \n",
        "    end += i + 1\n",
        "    \n",
        "    cons_array = np.concatenate((np.array(props_before),np.array(props_after)),axis=1)\n",
        "    \n",
        "    return train_loss_vector, l2_loss_vector, cons_loss_vector, val_loss_vector, cons_array, end"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "G3Nupwr2fr6t",
        "colab_type": "text"
      },
      "source": [
        "# validate"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2W82aGzyfuJh",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def validate(valloader,cons,zvars):\n",
        "  \n",
        "  print('      Running validation set')\n",
        "  \n",
        "  running_loss = 0.0\n",
        "  \n",
        "  with torch.no_grad():\n",
        "    for i, (data, targets, temp, vol) in enumerate(valloader):\n",
        "      \n",
        "      data, targets, temp, vol = data.to(device), targets.to(device), temp.to(device), vol.to(device)\n",
        "      \n",
        "      outputs = net(data)\n",
        "      outputs = outputs.to(device)\n",
        "            \n",
        "      nbatch = len(data)     \n",
        "\n",
        "      data_unnorm = data*zvars.std_f[:nbatch] + zvars.mean_f[:nbatch]\n",
        "      targets_unnorm = targets*zvars.std_fdf[:nbatch] + zvars.mean_fdf[:nbatch]\n",
        "      outputs_unnorm = outputs[:nbatch,0]*zvars.std_fdf[:nbatch,1] + zvars.mean_fdf[:nbatch,1]\n",
        "\n",
        "      targets_nof = targets_unnorm - data_unnorm\n",
        "      outputs_nof = outputs_unnorm[:nbatch] - data_unnorm[:nbatch,1]\n",
        "      \n",
        "      outputs_nof_to_cat = outputs_nof[:nbatch].unsqueeze(1)\n",
        "      targets_nof_to_cat = targets_nof[:nbatch,0].unsqueeze(1)        \n",
        "        \n",
        "      # concatenate with actual dfe\n",
        "      outputs_nof = torch.cat((outputs_nof_to_cat,targets_nof_to_cat),1)\n",
        "\n",
        "      masse_b,massi_b,mom_b,energy_b = check_properties_main(data_unnorm[:,:,:,:-1],\\\n",
        "                                                 targets_nof[:,:,:,:-1],temp,vol,cons) \n",
        "      masse_a,massi_a,mom_a,energy_a = check_properties_main(data_unnorm[:,:,:,:-1],\\\n",
        "                                                 outputs_nof[:,:,:,:-1],temp,vol,cons)  \n",
        "      \n",
        "      masse_loss = torch.sum(torch.abs(masse_a - masse_b)).float()/nbatch\n",
        "      massi_loss = torch.sum(torch.abs(massi_a - massi_b)).float()/nbatch\n",
        "      mass_loss = massi_loss + masse_loss\n",
        "      mom_loss = torch.sum(torch.abs(mom_a - mom_b)).float()/nbatch\n",
        "      energy_loss = torch.sum(torch.abs(energy_a - energy_b)).float()/nbatch   \n",
        "     \n",
        "      l2_loss = criterion(outputs[:,0],targets[:,1])  \n",
        "      \n",
        "      loss = l2_loss*loss_weights[0] \\\n",
        "           + mass_loss*loss_weights[1] \\\n",
        "           + mom_loss*loss_weights[2] \\\n",
        "           + energy_loss*loss_weights[3] \n",
        "      \n",
        "      running_loss += loss.item()\n",
        "  \n",
        "  #print(i+nbatch/batch_size)\n",
        "  avg_loss = running_loss/(i+1)\n",
        "  \n",
        "  print('         Validation loss: %.3f' % (avg_loss))\n",
        "\n",
        "  return avg_loss\n",
        "      "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y3rgOOERfvK6",
        "colab_type": "text"
      },
      "source": [
        "# test"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9We2eU6dfwg5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def test(f_test,df_test,temp_test,vol_test):\n",
        "  \n",
        "    testset = DistFuncDataset(f_test, df_test, temp_test, vol_test)\n",
        "    \n",
        "    testloader = DataLoader(testset, batch_size=batch_size, \n",
        "                            shuffle=True, num_workers=4)\n",
        "      \n",
        "    props_before = []\n",
        "    props_after = []\n",
        "    \n",
        "    l2_error=[]\n",
        "    lt1=0\n",
        "    gt1=0\n",
        "    with torch.no_grad():\n",
        "        for (data, targets, temp, vol) in testloader:\n",
        "\n",
        "            data, targets, temp, vol = data.to(device), targets.to(device), temp.to(device), vol.to(device)\n",
        "          \n",
        "            outputs = net(data)\n",
        "            outputs = outputs.to(device)   \n",
        "            \n",
        "            nbatch = len(data)     \n",
        "\n",
        "            data_unnorm = data*zvars.std_f[:nbatch] + zvars.mean_f[:nbatch]\n",
        "            targets_unnorm = targets*zvars.std_fdf[:nbatch] + zvars.mean_fdf[:nbatch]\n",
        "            outputs_unnorm = outputs[:nbatch,0]*zvars.std_fdf[:nbatch,1] + zvars.mean_fdf[:nbatch,1]\n",
        "\n",
        "            targets_nof = targets_unnorm - data_unnorm\n",
        "            outputs_nof = outputs_unnorm[:nbatch] - data_unnorm[:nbatch,1]\n",
        "\n",
        "            outputs_nof_to_cat = outputs_nof[:nbatch].unsqueeze(1)\n",
        "            targets_nof_to_cat = targets_nof[:nbatch,0].unsqueeze(1)        \n",
        "\n",
        "            # concatenate with actual dfe\n",
        "            outputs_nof = torch.cat((outputs_nof_to_cat,targets_nof_to_cat),1)\n",
        "  \n",
        "            props_before.append([torch.sum(each_prop).item()\\\n",
        "                                 for each_prop in check_properties_main(data_unnorm[:,:,:,:-1],\\\n",
        "                                                                   targets_nof[:,:,:,:-1],temp,vol,cons)])         \n",
        "            props_after.append([torch.sum(each_prop).item()\\\n",
        "                                for each_prop in check_properties_main(data_unnorm[:,:,:,:-1],\\\n",
        "                                                                  outputs_nof[:,:,:,:-1],temp,vol,cons)])\n",
        "                                \n",
        "            loss = criterion(outputs, targets)\n",
        "            l2_error.append(loss.item()*100)\n",
        "            if loss.item()*100 < 1:\n",
        "                lt1+=1\n",
        "            else:\n",
        "                gt1+=1\n",
        "    \n",
        "    cons_array = np.concatenate((np.array(props_before),np.array(props_after)),axis=1)\n",
        "\n",
        "    num_error = len(cons_array)\n",
        "    cons_error = np.zeros([3,num_error])\n",
        "    \n",
        "    cons_error[0] = np.abs(((cons_array[:,4]+cons_array[:,5])-(cons_array[:,0]+cons_array[:,1]))/(cons_array[:,0]+cons_array[:,1]))\n",
        "    cons_error[1] = np.abs((cons_array[:,6]-cons_array[:,2])/cons_array[:,2])\n",
        "    cons_error[2] = np.abs((cons_array[:,7]-cons_array[:,3])/cons_array[:,3])\n",
        "    \n",
        "    print('Finished testing')\n",
        "    print('Percentage with MSE<1: %d %%' % (\n",
        "            100 * lt1/(lt1+gt1)))\n",
        "    print('Percent error in conservation properties:\\nmass: \\\n",
        "            %d %%\\nmomentum: %d %%\\nenergy: %d %%' % ( \n",
        "            100*cons_error[0].max(), \n",
        "            100*cons_error[1].max(), \n",
        "            100*cons_error[2].max()))\n",
        "    \n",
        "    return l2_error, cons_error, props_after"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ln6ueusnfyid",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def save_checkpoint(state, is_best, lr, filename='checkpoint.pth.tar'): \n",
        "#   torch.save(state,'/content/checkpoints/'+str(lr)+'/'+filename)\n",
        "  torch.save(state, filename)\n",
        "  if is_best:\n",
        "    shutil.copy(filename, 'model_best.pth.tar')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8l6stLMefzDO",
        "colab_type": "text"
      },
      "source": [
        "# plot"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H8SH-5oifzkx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plot_df(df_xgc,df_ml,epoch):\n",
        "  \n",
        "  df_xgc = df_xgc.cpu().detach().numpy()\n",
        "  df_ml = df_ml.cpu().detach().numpy()\n",
        "  \n",
        "  df_min = df_xgc.min().item()\n",
        "  df_max = df_xgc.max().item()\n",
        "  \n",
        "  cbarticks = np.linspace(df_min,df_max,10)\n",
        "  \n",
        "  fig = plt.figure()\n",
        "  fig.set_figheight(10)\n",
        "  fig.set_figwidth(10)\n",
        "\n",
        "  v_aspect=32/31\n",
        "  ax1 = fig.add_subplot(1,2,1,aspect=v_aspect)\n",
        "  ax2 = fig.add_subplot(1,2,2,aspect=v_aspect)\n",
        "  ctr = ax1.contourf(df_xgc, vmin=df_min, vmax=df_max)\n",
        "  ax2.contourf(df_ml)\n",
        "  \n",
        "  ax1.set_title('Actual df')\n",
        "  ax2.set_title('Predicted df')\n",
        "  \n",
        "  fig.subplots_adjust(right=0.8)\n",
        "  cbar_ax = fig.add_axes([0.85, 0.3, 0.05, 0.4])\n",
        "  fig.colorbar(ctr, cax=cbar_ax, ticks=cbarticks)\n",
        "  \n",
        "#   fig.savefig('figs/dfs_{}'.format(epoch+1))\n",
        "  \n",
        "  plt.show()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WzeFx7IKf1MP",
        "colab_type": "text"
      },
      "source": [
        "# main"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0sd0FuArf1qy",
        "colab_type": "code",
        "outputId": "5143d642-6a00-4e12-89f7-f3adeb0fffc5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "start = timeit.default_timer()\n",
        "criterion = nn.MSELoss()\n",
        "\n",
        "# optimizer = optim.Adam(net.parameters())\n",
        "optimizer = optim.SGD(net.parameters(), lr=lr, momentum=momentum)\n",
        "scheduler = optim.lr_scheduler.StepLR(optimizer,step_size=step_size,gamma=lr_decay)\n",
        "\n",
        "train_loss = []\n",
        "val_loss = []\n",
        "\n",
        "for epoch in range(num_epochs):\n",
        "\n",
        "  lr_epoch = [group['lr'] for group in optimizer.param_groups][0]\n",
        "\n",
        "  print('Epoch: {} (lr = {})'.format(epoch+1,lr_epoch)) \n",
        "  \n",
        "  epoch1 = timeit.default_timer() \n",
        "  end = 0\n",
        "  for iphi in range(nphi):\n",
        "\n",
        "    print('Beginning training iphi = {}'.format(iphi))\n",
        "    print('   Loading data')\n",
        "    load1 = timeit.default_timer()\n",
        "    f,df,num_nodes,zvars,cons = load_data_hdf(iphi)\n",
        "    load2 = timeit.default_timer()\n",
        "    print('      Loading time: %.3fs' % (load2-load1))\n",
        "\n",
        "    print('   Creating training set')\n",
        "    trainloader,valloader,f_test,df_test,temp_test,vol_test = split_data(f,df,cons,num_nodes)\n",
        "    del f,df\n",
        "    \n",
        "    train1 = timeit.default_timer()\n",
        "    ### gather testing data\n",
        "    if epoch == 0:\n",
        "      if iphi == 0:\n",
        "        f_all_test,df_all_test,temp_all_test,vol_all_test = f_test,df_test,temp_test,vol_test\n",
        "        del f_test,df_test,temp_test,vol_test\n",
        "\n",
        "        print('   Starting training')\n",
        "        train_loss, l2_loss, cons_loss, val_loss, cons_array, end = \\\n",
        "                                   train(trainloader,valloader,0,epoch,end,zvars,cons)\n",
        "\n",
        "      else:\n",
        "        f_all_test = np.vstack((f_all_test,f_test))\n",
        "        df_all_test = np.vstack((df_all_test,df_test))\n",
        "        temp_all_test = np.vstack((temp_all_test,temp_test))\n",
        "        vol_all_test = np.vstack((vol_all_test,vol_test))\n",
        "        del f_test,df_test,temp_test,vol_test\n",
        "\n",
        "        print('   Starting training')\n",
        "        train_loss_to_app, l2_loss_to_app, cons_loss_to_app, val_loss_to_app, cons_to_cat, end = \\\n",
        "                                   train(trainloader,valloader,0,epoch,end,zvars,cons)\n",
        "\n",
        "        for loss1 in train_loss_to_app:\n",
        "          train_loss.append(loss1)\n",
        "        for loss2 in val_loss_to_app:\n",
        "          val_loss.append(loss2)\n",
        "        cons_array = np.concatenate((cons_array, cons_to_cat), axis=1)\n",
        "    \n",
        "    else:\n",
        "      del f_test,df_test,temp_test,vol_test\n",
        "      print('   Starting training')\n",
        "      train_loss_to_app, l2_loss_to_app, cons_loss_to_app, val_loss_to_app, cons_to_cat, end = \\\n",
        "                                 train(trainloader,valloader,0,epoch,end,zvars,cons)\n",
        "\n",
        "      for loss1 in train_loss_to_app:\n",
        "          train_loss.append(loss1)\n",
        "      for loss2 in l2_loss_to_app:\n",
        "          l2_loss.append(loss2)         \n",
        "      for loss3 in cons_loss_to_app:\n",
        "          cons_loss.append(loss3)          \n",
        "      for loss4 in val_loss_to_app:\n",
        "          val_loss.append(loss2)      \n",
        "      cons_array = np.concatenate((cons_array, cons_to_cat), axis=0)\n",
        "         \n",
        "    train2 = timeit.default_timer()\n",
        "    print('Finished tranining iphi = {}'.format(iphi))\n",
        "    print('   Training time for iphi = %d: %.3fs' % (iphi,train2-train1))\n",
        "  \n",
        "  train_iterations = np.linspace(1,len(train_loss),len(train_loss))\n",
        "  val_iterations = np.linspace(2,len(train_loss),len(val_loss))\n",
        "  \n",
        "  plt.plot(train_iterations,train_loss,'-o',color='blue')\n",
        "  plt.plot(val_iterations,val_loss,'-o',color='orange')\n",
        "  plt.plot(train_iterations,l2_loss,'-o',color='red')\n",
        "  plt.plot(train_iterations,cons_loss,'-o',color='green')\n",
        "  plt.legend(['total','validation','l2','cons'])\n",
        "  plt.yscale('log')\n",
        "  plt.show()\n",
        "  \n",
        "  epoch2 = timeit.default_timer()\n",
        "  scheduler.step()\n",
        "  print('Epoch time: {}s\\n'.format(epoch2-epoch1))\n",
        "\n",
        "print('Starting testing')\n",
        "l2_error_i, cons_error_i = test(f_all_test,df_all_test,temp_all_test,vol_all_test)\n",
        "print('Finished testing')\n",
        "\n",
        "stop = timeit.default_timer()\n",
        "print('Runtime: %.3fmins' % ((stop-start)/60))"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch: 1 (lr = 1e-07)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 14.847s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [1,   500] loss: 333.019776\n",
            "      L2 loss: 0.560966\n",
            "      conservation loss: 333.019776\n",
            "   [1,  1000] loss: 137.090782\n",
            "      L2 loss: 0.544374\n",
            "      conservation loss: 137.090782\n",
            "   [1,  1500] loss: 196.938275\n",
            "      L2 loss: 0.547434\n",
            "      conservation loss: 196.938275\n",
            "   [1,  2000] loss: 374.549480\n",
            "      L2 loss: 0.534915\n",
            "      conservation loss: 374.549480\n",
            "      Running validation set\n",
            "         Validation loss: 14.811\n",
            "   [1,  2500] loss: 358.924520\n",
            "      L2 loss: 0.551838\n",
            "      conservation loss: 358.924520\n",
            "   [1,  3000] loss: 160.139454\n",
            "      L2 loss: 0.507608\n",
            "      conservation loss: 160.139454\n",
            "   [1,  3500] loss: 104.567833\n",
            "      L2 loss: 0.518627\n",
            "      conservation loss: 104.567833\n",
            "   [1,  4000] loss: 205.428652\n",
            "      L2 loss: 0.567529\n",
            "      conservation loss: 205.428652\n",
            "      Running validation set\n",
            "         Validation loss: 6.016\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 150.576s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\neZwAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 169.262580782s\n",
            "\n",
            "Epoch: 2 (lr = 1e-07)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 12.378s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [2,   500] loss: 178.772036\n",
            "      L2 loss: 0.515481\n",
            "      conservation loss: 178.772036\n",
            "   [2,  1000] loss: 139.278157\n",
            "      L2 loss: 0.503857\n",
            "      conservation loss: 139.278157\n",
            "   [2,  1500] loss: 124.467480\n",
            "      L2 loss: 0.611059\n",
            "      conservation loss: 124.467480\n",
            "   [2,  2000] loss: 190.732173\n",
            "      L2 loss: 0.575055\n",
            "      conservation loss: 190.732173\n",
            "      Running validation set\n",
            "         Validation loss: 37.788\n",
            "   [2,  2500] loss: 173.425773\n",
            "      L2 loss: 0.519265\n",
            "      conservation loss: 173.425773\n",
            "   [2,  3000] loss: 73.250795\n",
            "      L2 loss: 0.540075\n",
            "      conservation loss: 73.250795\n",
            "   [2,  3500] loss: 208.251645\n",
            "      L2 loss: 0.510554\n",
            "      conservation loss: 208.251645\n",
            "   [2,  4000] loss: 145.432248\n",
            "      L2 loss: 0.510467\n",
            "      conservation loss: 145.432248\n",
            "      Running validation set\n",
            "         Validation loss: 14.888\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 151.889s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 167.7557355570002s\n",
            "\n",
            "Epoch: 3 (lr = 1e-08)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 13.079s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [3,   500] loss: 88.054777\n",
            "      L2 loss: 0.536939\n",
            "      conservation loss: 88.054777\n",
            "   [3,  1000] loss: 30.747496\n",
            "      L2 loss: 0.532560\n",
            "      conservation loss: 30.747496\n",
            "   [3,  1500] loss: 29.328989\n",
            "      L2 loss: 0.522485\n",
            "      conservation loss: 29.328989\n",
            "   [3,  2000] loss: 35.845327\n",
            "      L2 loss: 0.512307\n",
            "      conservation loss: 35.845327\n",
            "      Running validation set\n",
            "         Validation loss: 0.822\n",
            "   [3,  2500] loss: 21.843931\n",
            "      L2 loss: 0.501555\n",
            "      conservation loss: 21.843931\n",
            "   [3,  3000] loss: 23.184527\n",
            "      L2 loss: 0.511686\n",
            "      conservation loss: 23.184527\n",
            "   [3,  3500] loss: 40.623841\n",
            "      L2 loss: 0.510342\n",
            "      conservation loss: 40.623841\n",
            "   [3,  4000] loss: 37.992102\n",
            "      L2 loss: 0.530271\n",
            "      conservation loss: 37.992102\n",
            "      Running validation set\n",
            "         Validation loss: 2.509\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 153.811s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXsAAAD4CAYAAAANbUbJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nN/xDI46DDAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 171.39405515499993s\n",
            "\n",
            "Epoch: 4 (lr = 1e-08)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 13.079s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [4,   500] loss: 12.749821\n",
            "      L2 loss: 0.542530\n",
            "      conservation loss: 12.749821\n",
            "   [4,  1000] loss: 30.348908\n",
            "      L2 loss: 0.519521\n",
            "      conservation loss: 30.348908\n",
            "   [4,  1500] loss: 18.355821\n",
            "      L2 loss: 0.509908\n",
            "      conservation loss: 18.355821\n",
            "   [4,  2000] loss: 19.225233\n",
            "      L2 loss: 0.512950\n",
            "      conservation loss: 19.225233\n",
            "      Running validation set\n",
            "         Validation loss: 3.345\n",
            "   [4,  2500] loss: 15.439230\n",
            "      L2 loss: 0.526918\n",
            "      conservation loss: 15.439230\n",
            "   [4,  3000] loss: 41.147213\n",
            "      L2 loss: 0.476475\n",
            "      conservation loss: 41.147213\n",
            "   [4,  3500] loss: 24.309136\n",
            "      L2 loss: 0.516866\n",
            "      conservation loss: 24.309136\n",
            "   [4,  4000] loss: 26.291674\n",
            "      L2 loss: 0.508234\n",
            "      conservation loss: 26.291674\n",
            "      Running validation set\n",
            "         Validation loss: 0.563\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 153.750s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYwboBowAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 171.37504063999995s\n",
            "\n",
            "Epoch: 5 (lr = 1e-09)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 13.004s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [5,   500] loss: 9.437390\n",
            "      L2 loss: 0.528262\n",
            "      conservation loss: 9.437390\n",
            "   [5,  1000] loss: 7.252842\n",
            "      L2 loss: 0.518396\n",
            "      conservation loss: 7.252842\n",
            "   [5,  1500] loss: 12.482608\n",
            "      L2 loss: 0.519703\n",
            "      conservation loss: 12.482608\n",
            "   [5,  2000] loss: 5.764842\n",
            "      L2 loss: 0.520462\n",
            "      conservation loss: 5.764842\n",
            "      Running validation set\n",
            "         Validation loss: 0.413\n",
            "   [5,  2500] loss: 6.050995\n",
            "      L2 loss: 0.547702\n",
            "      conservation loss: 6.050995\n",
            "   [5,  3000] loss: 18.672525\n",
            "      L2 loss: 0.488918\n",
            "      conservation loss: 18.672525\n",
            "   [5,  3500] loss: 7.124528\n",
            "      L2 loss: 0.458762\n",
            "      conservation loss: 7.124528\n",
            "   [5,  4000] loss: 6.312376\n",
            "      L2 loss: 0.517150\n",
            "      conservation loss: 6.312376\n",
            "      Running validation set\n",
            "         Validation loss: 0.391\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 153.897s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 172.36884372699933s\n",
            "\n",
            "Epoch: 6 (lr = 1e-09)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 26.368s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [6,   500] loss: 16.484596\n",
            "      L2 loss: 0.536938\n",
            "      conservation loss: 16.484596\n",
            "   [6,  1000] loss: 4.866044\n",
            "      L2 loss: 0.532767\n",
            "      conservation loss: 4.866044\n",
            "   [6,  1500] loss: 5.306158\n",
            "      L2 loss: 0.465569\n",
            "      conservation loss: 5.306158\n",
            "   [6,  2000] loss: 12.768801\n",
            "      L2 loss: 0.560707\n",
            "      conservation loss: 12.768801\n",
            "      Running validation set\n",
            "         Validation loss: 0.411\n",
            "   [6,  2500] loss: 6.067454\n",
            "      L2 loss: 0.481046\n",
            "      conservation loss: 6.067454\n",
            "   [6,  3000] loss: 8.285922\n",
            "      L2 loss: 0.507028\n",
            "      conservation loss: 8.285922\n",
            "   [6,  3500] loss: 6.546783\n",
            "      L2 loss: 0.499859\n",
            "      conservation loss: 6.546783\n",
            "   [6,  4000] loss: 7.109818\n",
            "      L2 loss: 0.496671\n",
            "      conservation loss: 7.109818\n",
            "      Running validation set\n",
            "         Validation loss: 0.522\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 154.703s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAD4CAYAAADrRI2NAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ncHYAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 185.35180956099975s\n",
            "\n",
            "Epoch: 7 (lr = 1.0000000000000002e-10)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 26.603s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [7,   500] loss: 4.688446\n",
            "      L2 loss: 0.542656\n",
            "      conservation loss: 4.688446\n",
            "   [7,  1000] loss: 5.438807\n",
            "      L2 loss: 0.511113\n",
            "      conservation loss: 5.438807\n",
            "   [7,  1500] loss: 4.650918\n",
            "      L2 loss: 0.537900\n",
            "      conservation loss: 4.650918\n",
            "   [7,  2000] loss: 6.384336\n",
            "      L2 loss: 0.490235\n",
            "      conservation loss: 6.384336\n",
            "      Running validation set\n",
            "         Validation loss: 0.320\n",
            "   [7,  2500] loss: 15.446346\n",
            "      L2 loss: 0.512323\n",
            "      conservation loss: 15.446346\n",
            "   [7,  3000] loss: 5.316881\n",
            "      L2 loss: 0.496054\n",
            "      conservation loss: 5.316881\n",
            "   [7,  3500] loss: 5.206316\n",
            "      L2 loss: 0.467232\n",
            "      conservation loss: 5.206316\n",
            "   [7,  4000] loss: 4.616143\n",
            "      L2 loss: 0.491150\n",
            "      conservation loss: 4.616143\n",
            "      Running validation set\n",
            "         Validation loss: 0.320\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 156.351s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nccIAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 187.57877148199987s\n",
            "\n",
            "Epoch: 8 (lr = 1.0000000000000002e-10)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 13.640s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [8,   500] loss: 4.692768\n",
            "      L2 loss: 0.495757\n",
            "      conservation loss: 4.692768\n",
            "   [8,  1000] loss: 4.200950\n",
            "      L2 loss: 0.586907\n",
            "      conservation loss: 4.200950\n",
            "   [8,  1500] loss: 4.621506\n",
            "      L2 loss: 0.515614\n",
            "      conservation loss: 4.621506\n",
            "   [8,  2000] loss: 6.232292\n",
            "      L2 loss: 0.445244\n",
            "      conservation loss: 6.232292\n",
            "      Running validation set\n",
            "         Validation loss: 0.353\n",
            "   [8,  2500] loss: 5.304156\n",
            "      L2 loss: 0.557987\n",
            "      conservation loss: 5.304156\n",
            "   [8,  3000] loss: 4.747273\n",
            "      L2 loss: 0.520641\n",
            "      conservation loss: 4.747273\n",
            "   [8,  3500] loss: 7.102835\n",
            "      L2 loss: 0.552159\n",
            "      conservation loss: 7.102835\n",
            "   [8,  4000] loss: 4.985974\n",
            "      L2 loss: 0.545364\n",
            "      conservation loss: 4.985974\n",
            "      Running validation set\n",
            "         Validation loss: 0.334\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 157.790s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 175.9930101889995s\n",
            "\n",
            "Epoch: 9 (lr = 1.0000000000000001e-11)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 13.317s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [9,   500] loss: 4.952860\n",
            "      L2 loss: 0.510233\n",
            "      conservation loss: 4.952860\n",
            "   [9,  1000] loss: 3.910159\n",
            "      L2 loss: 0.552963\n",
            "      conservation loss: 3.910159\n",
            "   [9,  1500] loss: 3.981690\n",
            "      L2 loss: 0.547783\n",
            "      conservation loss: 3.981690\n",
            "   [9,  2000] loss: 4.401717\n",
            "      L2 loss: 0.519437\n",
            "      conservation loss: 4.401717\n",
            "      Running validation set\n",
            "         Validation loss: 0.312\n",
            "   [9,  2500] loss: 4.885475\n",
            "      L2 loss: 0.500629\n",
            "      conservation loss: 4.885475\n",
            "   [9,  3000] loss: 5.927441\n",
            "      L2 loss: 0.525811\n",
            "      conservation loss: 5.927441\n",
            "   [9,  3500] loss: 6.957231\n",
            "      L2 loss: 0.516196\n",
            "      conservation loss: 6.957231\n",
            "   [9,  4000] loss: 4.418711\n",
            "      L2 loss: 0.534684\n",
            "      conservation loss: 4.418711\n",
            "      Running validation set\n",
            "         Validation loss: 0.312\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 156.295s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\njuM4I4g7d8dxnBHk/0hj3u+wz+Q6AAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 174.280698949s\n",
            "\n",
            "Epoch: 10 (lr = 1.0000000000000001e-11)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 13.338s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [10,   500] loss: 8.072563\n",
            "      L2 loss: 0.481597\n",
            "      conservation loss: 8.072563\n",
            "   [10,  1000] loss: 5.362490\n",
            "      L2 loss: 0.499132\n",
            "      conservation loss: 5.362490\n",
            "   [10,  1500] loss: 4.207320\n",
            "      L2 loss: 0.532395\n",
            "      conservation loss: 4.207320\n",
            "   [10,  2000] loss: 5.130080\n",
            "      L2 loss: 0.503683\n",
            "      conservation loss: 5.130080\n",
            "      Running validation set\n",
            "         Validation loss: 0.294\n",
            "   [10,  2500] loss: 6.110874\n",
            "      L2 loss: 0.496613\n",
            "      conservation loss: 6.110874\n",
            "   [10,  3000] loss: 4.306139\n",
            "      L2 loss: 0.513248\n",
            "      conservation loss: 4.306139\n",
            "   [10,  3500] loss: 5.627723\n",
            "      L2 loss: 0.532865\n",
            "      conservation loss: 5.627723\n",
            "   [10,  4000] loss: 8.630463\n",
            "      L2 loss: 0.516318\n",
            "      conservation loss: 8.630463\n",
            "      Running validation set\n",
            "         Validation loss: 0.297\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 155.853s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAcGlwAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 173.85279720499966s\n",
            "\n",
            "Epoch: 11 (lr = 1.0000000000000002e-12)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 13.169s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [11,   500] loss: 4.167761\n",
            "      L2 loss: 0.525977\n",
            "      conservation loss: 4.167761\n",
            "   [11,  1000] loss: 4.687988\n",
            "      L2 loss: 0.535662\n",
            "      conservation loss: 4.687988\n",
            "   [11,  1500] loss: 3.816621\n",
            "      L2 loss: 0.577303\n",
            "      conservation loss: 3.816621\n",
            "   [11,  2000] loss: 4.581349\n",
            "      L2 loss: 0.453842\n",
            "      conservation loss: 4.581349\n",
            "      Running validation set\n",
            "         Validation loss: 0.266\n",
            "   [11,  2500] loss: 6.167002\n",
            "      L2 loss: 0.543114\n",
            "      conservation loss: 6.167002\n",
            "   [11,  3000] loss: 4.099781\n",
            "      L2 loss: 0.543171\n",
            "      conservation loss: 4.099781\n",
            "   [11,  3500] loss: 4.443774\n",
            "      L2 loss: 0.502039\n",
            "      conservation loss: 4.443774\n",
            "   [11,  4000] loss: 7.094523\n",
            "      L2 loss: 0.503796\n",
            "      conservation loss: 7.094523\n",
            "      Running validation set\n",
            "         Validation loss: 0.266\n",
            "Finished tranining iphi = 0\n",
            "   Training time for iphi = 0: 157.158s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n7jiO00D+Bn3w1WQNZqWaAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch time: 175.10607417699975s\n",
            "\n",
            "Epoch: 12 (lr = 1.0000000000000002e-12)\n",
            "Beginning training iphi = 0\n",
            "   Loading data\n",
            "      Loading time: 13.434s\n",
            "   Creating training set\n",
            "   Starting training\n",
            "   [12,   500] loss: 4.102999\n",
            "      L2 loss: 0.514846\n",
            "      conservation loss: 4.102999\n",
            "   [12,  1000] loss: 4.704331\n",
            "      L2 loss: 0.455288\n",
            "      conservation loss: 4.704331\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-21-ffa4bfde4e09>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     59\u001b[0m       \u001b[0;32mdel\u001b[0m \u001b[0mf_test\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mdf_test\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtemp_test\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mvol_test\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     60\u001b[0m       \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'   Starting training'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 61\u001b[0;31m       \u001b[0mtrain_loss_to_app\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0ml2_loss_to_app\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcons_loss_to_app\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mval_loss_to_app\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcons_to_cat\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mend\u001b[0m \u001b[0;34m=\u001b[0m                                  \u001b[0mtrain\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrainloader\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mvalloader\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mepoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mend\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mzvars\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     62\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     63\u001b[0m       \u001b[0;32mfor\u001b[0m \u001b[0mloss1\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mtrain_loss_to_app\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-16-76f296fa84f4>\u001b[0m in \u001b[0;36mtrain\u001b[0;34m(trainloader, valloader, sp_flag, epoch, end, zvars, cons)\u001b[0m\n\u001b[1;32m     44\u001b[0m         \u001b[0mmasse_b\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mmassi_b\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mmom_b\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0menergy_b\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcheck_properties_main\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata_unnorm\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m                                                               \u001b[0mtargets_nof\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtemp\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mvol\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     45\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 46\u001b[0;31m         \u001b[0mmasse_a\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mmassi_a\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mmom_a\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0menergy_a\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcheck_properties_main\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata_unnorm\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m                                                               \u001b[0moutputs_nof\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtemp\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mvol\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     47\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     48\u001b[0m         \u001b[0mprops_before\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msum\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmassi_b\u001b[0m\u001b[0;34m+\u001b[0m\u001b[0mmasse_b\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0mnbatch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m                             \u001b[0;34m(\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msum\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmom_b\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0mnbatch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m                             \u001b[0;34m(\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msum\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0menergy_b\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0mnbatch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-14-ee2f3dc3e2bd>\u001b[0m in \u001b[0;36mcheck_properties_main\u001b[0;34m(f, df, temp, vol, cons)\u001b[0m\n\u001b[1;32m     10\u001b[0m   \u001b[0;31m#print('f')\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     11\u001b[0m   \u001b[0mne\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mmome\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mene\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcheck_properties_each\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtemp\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mvol\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 12\u001b[0;31m   \u001b[0mni\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mmomi\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0meni\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcheck_properties_each\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mf\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtemp\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mvol\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     13\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     14\u001b[0m   \u001b[0mdne_n\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mabs\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdne\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0mne\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-13-d56dfe6d81fa>\u001b[0m in \u001b[0;36mcheck_properties_each\u001b[0;34m(f_slice, cons, temp, vol, sp)\u001b[0m\n\u001b[1;32m     11\u001b[0m     \u001b[0mvth\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msqrt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtemp\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msml_ev2j\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mptl_mass\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0msp\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 13\u001b[0;31m     \u001b[0mvpar\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlinspace\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mf0_nvp\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mf0_nvp\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mf0_nvp\u001b[0m\u001b[0;34m+\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mf0_dvp\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     14\u001b[0m     \u001b[0mvperp\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlinspace\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mf0_nmu\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mf0_nmu\u001b[0m\u001b[0;34m+\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mcons\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mf0_dsmu\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<__array_function__ internals>\u001b[0m in \u001b[0;36mlinspace\u001b[0;34m(*args, **kwargs)\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/numpy/core/function_base.py\u001b[0m in \u001b[0;36mlinspace\u001b[0;34m(start, stop, num, endpoint, retstep, dtype, axis)\u001b[0m\n\u001b[1;32m    136\u001b[0m     \u001b[0mstop\u001b[0m  \u001b[0;34m=\u001b[0m \u001b[0masanyarray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mstop\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;34m*\u001b[0m \u001b[0;36m1.0\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    137\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 138\u001b[0;31m     \u001b[0mdt\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mresult_type\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mstart\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mstop\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfloat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnum\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    139\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mdtype\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    140\u001b[0m         \u001b[0mdtype\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdt\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<__array_function__ internals>\u001b[0m in \u001b[0;36mresult_type\u001b[0;34m(*args, **kwargs)\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sCPaWsmiuNIQ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                        ## used to see differences between voli/e and f0_grid_vol \n",
        "\n",
        "# fid1 = h5py.File('/content/hdf5_data/hdf_vol.h5','r')\n",
        "# fid2 = h5py.File('/content/hdf5_data/hdf_cons_fullvol.h5','r')\n",
        "\n",
        "# vole1 = fid1['vole'][0]\n",
        "# voli1 = fid1['vole'][0]\n",
        "\n",
        "# f0_grid_vol = fid2['f0_grid_vol'][...]\n",
        "# vole2 = f0_grid_vol[0]\n",
        "# voli2 = f0_grid_vol[1]"
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}