Set_weights_1D.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "20cd927e-f47d-4dc1-a996-c9a55414a8db",
   "metadata": {},
   "source": [
    "## Different imports and setup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "3df341a7-ecaa-4f15-b519-0cf0bc18502f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Standard import\n",
    "import numpy as np\n",
    "from numpy.random import default_rng\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib widget"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "ffe6a666-f797-45c3-ad7b-4765c66aad39",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Machine learning libraries\n",
    "from tensorflow import keras\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.losses import MeanSquaredError\n",
    "\n",
    "from tqdm.keras import TqdmCallback"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "f98ec635-8382-463a-b00a-923a674199d6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Custom plots file and tensorflow models\n",
    "from Tools import Plot\n",
    "from Tools import Custom_models\n",
    "from Tools import Helpers"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b00aaa99-a4d3-4587-92ba-16e8d12597be",
   "metadata": {},
   "source": [
    "## Define the grid and the real solution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "ba65232b-674b-4218-96c6-187bdff86a5f",
   "metadata": {},
   "outputs": [],
   "source": [
    "N = 100\n",
    "h = 2/(N-1)\n",
    "\n",
    "k = 50   # steepness parameter\n",
    "x1 = -1/3  # location of transition parameter\n",
    "x2 = 1/3\n",
    "\n",
    "x = np.linspace(-1,1,N)\n",
    "u = 1/2*(1 + np.tanh(k*(x-x1))) - 1/2*(1 + np.tanh(k*(x-x2)))  # Function we want to approximate"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cee148e7-612f-4a24-8eb8-b491fb8c166a",
   "metadata": {},
   "source": [
    "## Create exact finite element weights "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "c4bd988f-a99e-4f71-b6c1-c23486038325",
   "metadata": {},
   "outputs": [],
   "source": [
    "core1 = np.ones(N)\n",
    "bias1 = -x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "f422968d-8a81-47a0-bd0b-be5d05ea64c9",
   "metadata": {},
   "outputs": [],
   "source": [
    "core2 = np.zeros(N)\n",
    "for i in range(2,N-2):\n",
    "    core2[i] = 1/h*(u[i-1] - 2*u[i] + u[i+1])\n",
    "core2[0] = u[1]/h\n",
    "core2[1] = 1/h*(-2*u[1] + u[2])\n",
    "core2[N-2] = 1/h*(u[N-3] - 2*u[N-2])\n",
    "core2[N-1] = u[N-2]/h\n",
    "\n",
    "bias2 = np.array([0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "98ab53d9-2abc-4699-b937-073be216e780",
   "metadata": {},
   "outputs": [],
   "source": [
    "core1 = np.expand_dims(core1, axis=0)\n",
    "core2 = np.expand_dims(core2, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "5867f4d3-d783-46f8-98a1-4409cd48d730",
   "metadata": {},
   "outputs": [],
   "source": [
    "weights1_exact = [core1, bias1]\n",
    "weights2_exact = [core2, bias2]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d082f018-bad9-480b-a98d-3b5aa4e0e543",
   "metadata": {},
   "source": [
    "## Set the seeds and extract the training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "baf54dec-b8ca-4abf-9aa6-56c1627a7217",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set the seed for reproducibility\n",
    "seed = 6\n",
    "np.random.seed(seed)\n",
    "tf.random.set_seed(seed*2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "513e84cd-4b44-4588-bf59-efc4817d6572",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load the data\n",
    "features_train, labels_train, features_val, labels_val, features_test, labels_test = Helpers.Load_and_split_val(\n",
    "    'Data/tanh_k50_negative.txt', D=1, seed=seed, ratio_test=0.1, ratio_val=0.1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "255467a9-0d96-4a14-a590-d41a4c78b14d",
   "metadata": {},
   "source": [
    "## Slightly modify the exact weights with Gaussian noise"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b611c353-39df-4fff-9e96-2d012a11120d",
   "metadata": {},
   "outputs": [],
   "source": [
    "rng = default_rng(seed=seed)\n",
    "std = 0.00001\n",
    "core1_rand = core1 + std*rng.standard_normal(core1.shape)\n",
    "bias1_rand = bias1 + std*rng.standard_normal(bias1.shape)\n",
    "core2_rand = core2 + std*rng.standard_normal(core2.shape)\n",
    "bias2_rand = bias2 + std*rng.standard_normal(bias2.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5f00f958-930e-4a34-a7b1-84c524afb801",
   "metadata": {},
   "outputs": [],
   "source": [
    "weights1_rand = [core1_rand, bias1_rand]\n",
    "weights2_rand = [core2_rand, bias2_rand]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a4c198f6-48b8-4495-a73c-c4a7a7cb47c7",
   "metadata": {},
   "source": [
    "## Create a model with the modified weights "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "6ab379a8-625c-419f-a5d9-ae8a39db01ce",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"Exact_FEM_weights_N_100\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense1 (Dense)               multiple                  200       \n",
      "_________________________________________________________________\n",
      "dense2 (Dense)               multiple                  101       \n",
      "=================================================================\n",
      "Total params: 301\n",
      "Trainable params: 301\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = Custom_models.Model_1D_1_layer(K1=N, K_output=1, name='Exact_FEM_weights_N_100')\n",
    "model.build((None, 1))\n",
    "model.get_layer(name='dense1').set_weights(weights1_exact)\n",
    "model.get_layer(name='dense2').set_weights(weights2_exact)\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "919d9bf2-8c8d-4083-8589-60b583cf3fbd",
   "metadata": {},
   "source": [
    "## Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "a88d34de-ffcc-45b5-8cd9-e9a5893d4a39",
   "metadata": {},
   "outputs": [],
   "source": [
    "learning_rate = 1e-2\n",
    "opt = Adam(learning_rate=learning_rate)\n",
    "\n",
    "model.compile(loss=MeanSquaredError(), optimizer=opt)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "d4a98f4e-e07b-42cb-a009-8154fa3d35d5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d9137ba02f2d4c5d8fc8ea2d9c626a80",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "0epoch [00:00, ?epoch/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "0batch [00:00, ?batch/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Train\n",
    "N_epochs = 2\n",
    "\n",
    "history = model.fit(features_train, labels_train, epochs=N_epochs, validation_split=0.1, verbose=0,\n",
    "                    shuffle=True, initial_epoch=0, callbacks=[TqdmCallback()])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d587c785-1ba3-4814-acf2-f7dbd6b4d55d",
   "metadata": {},
   "source": [
    "## Check the efficiency of training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "ee3bda25-f2b6-4498-81f4-58f1a26b4a80",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0c810852d2f7410ba1e4c923e564f149",
       "version_major": 2,
       "version_minor": 0
      },
      "image/png": "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",
      "text/plain": [
       "Canvas(toolbar=Toolbar(toolitems=[('Home', 'Reset original view', 'home', 'home'), ('Back', 'Back to previous …"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "filename = 'Figures/Exact_FEM_N_100_training_error.pdf'\n",
    "save = False\n",
    "\n",
    "plt.figure()\n",
    "plt.semilogy(history.epoch, np.array(history.history['loss']), label='Training')\n",
    "plt.semilogy(history.epoch, np.array(history.history['val_loss']), label = 'Validation')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Mean squared error')\n",
    "plt.xlim([0,200])\n",
    "plt.legend()\n",
    "plt.grid()\n",
    "plt.show()\n",
    "if (save):\n",
    "    plt.savefig(filename, bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "add8ed39-2a44-4ed3-b7e4-c2b6df6abe27",
   "metadata": {},
   "source": [
    "## Save the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "db291a55-e79c-4374-bc3e-2bb9632695b2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: Saved_models/Exact_FEM_N_100_trained_20/assets\n"
     ]
    }
   ],
   "source": [
    "# Save the model\n",
    "model_name = 'Saved_models/Exact_FEM_N_100_trained_20'\n",
    "save_model = False\n",
    "\n",
    "if (save_model):\n",
    "    model.save(model_name)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e83ba70-7b23-42fe-b1ca-29db0d53c869",
   "metadata": {},
   "source": [
    "## Check the results on the testing set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "49adef46-b0a1-4354-98ab-dfc7d5703f02",
   "metadata": {},
   "outputs": [],
   "source": [
    "sorting = np.argsort(features_testing)\n",
    "features_testing = features_testing[sorting]\n",
    "labels_testing = labels_testing[sorting]\n",
    "\n",
    "prediction = model.predict(features_testing)\n",
    "prediction = np.ravel(prediction)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "b44c3995-833b-4364-9efc-e9c2ab1ce0ba",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "de59403d2ca34165817f4d849b4e9b90",
       "version_major": 2,
       "version_minor": 0
      },
      "image/png": "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",
      "text/plain": [
       "Canvas(toolbar=Toolbar(toolitems=[('Home', 'Reset original view', 'home', 'home'), ('Back', 'Back to previous …"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.plot(features_testing, labels_testing, 'b-', label='true')\n",
    "plt.plot(features_testing, prediction, 'r--', label='Prediction')\n",
    "plt.xlabel('$x$')\n",
    "plt.ylabel('$y$')\n",
    "plt.legend()\n",
    "plt.grid()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fae7cbbf-c8a3-4e4c-b1d4-13856abdfd8f",
   "metadata": {},
   "source": [
    "## Retrieve the weights and assess the difference with the exact weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4557e627-4a89-4e8f-b224-9ed7653751a2",
   "metadata": {},
   "outputs": [],
   "source": [
    "weights1_after_train = model.get_layer(name='dense1').get_weights()\n",
    "weights2_after_train = model.get_layer(name='dense2').get_weights()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8431f942-fea1-4196-a8fc-7d505b2a15ec",
   "metadata": {},
   "outputs": [],
   "source": [
    "core1_after_train = weights1_after_train[0]\n",
    "bias1_after_train = weights1_after_train[1]\n",
    "core2_after_train = weights2_after_train[0]\n",
    "bias2_after_train = weights2_after_train[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cbdfbd31-e026-46cf-9710-a988394f6076",
   "metadata": {},
   "outputs": [],
   "source": [
    "bias1 - bias1_after_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3b6ed228-deeb-4540-a45e-887746bdbeeb",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}