Example_1D.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "academic-green",
   "metadata": {},
   "source": [
    "## 1-D Problem: Friction Factor of Flow in Smooth Pipes\n",
    "\n",
    "The flow in smooth pipes is a classical fluid mechanics problem where the friction factor (f) is a sole function of the Reynolds number (Re). Three distinct physical regimes exist over the entire range of the Reynolds number: laminar flow (Re ≲ 2300), transitional flow (2300 ≲ Re ≲ 4000), and turbulent flow (Re ≳ 4000). "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "modular-sample",
   "metadata": {},
   "source": [
    "### 1. Import the taining dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "surrounded-packaging",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# import the training dataset (should be saved in .csv)\n",
    "# the csv file should include a header (1st row) with the names of the input and output variables\n",
    "df_train = pd.read_csv(\"data/data_smooth.csv\")\n",
    "SIZE = len(df_train)\n",
    "\n",
    "col_x = ['Re'] # list of the names of the input variables\n",
    "col_y = ['f'] # list of the names of the output variables\n",
    "\n",
    "# you may visualize the training data\n",
    "# this is possible for 1D and 2D problem, but may be difficult for multidimensional data \n",
    "fig, ax = plt.subplots()\n",
    "ax.scatter(df_train.Re, df_train.f, s=0.1)  \n",
    "ax.set_xscale('log')\n",
    "ax.set_yscale('log')\n",
    "ax.set_xlabel('Re')\n",
    "ax.set_ylabel('f')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "comparable-hobby",
   "metadata": {},
   "source": [
    "### 2. Train DimNet models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "reduced-motel",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Repeat #0\n",
      "Training model for hidden-layer config of [2, 2] with 250 data samples\n",
      "Resume from the last epoch 80000\n",
      "\n",
      "     EPOCH        LOSS   LOSS_AVE    IMPROVE?   MAPE     TIME/s        ELAPSED       CHANGE\n",
      "     90000     0.0002152006  2.15313e-04   YES    18.00%     0.00     0:00:21.658706   -0.0010%\n",
      "Training completed!\n",
      "To continue, run this cell again.\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Repeat #1\n",
      "Training model for hidden-layer config of [2, 2] with 250 data samples\n",
      "No parameters loaded. Train from scratch.\n",
      "Created a new log.txt\n",
      "\n",
      "     EPOCH        LOSS   LOSS_AVE    IMPROVE?   MAPE     TIME/s        ELAPSED       CHANGE\n",
      "     10000     0.0004792524  4.97420e-04   YES    69.84%     0.00     0:00:21.442096   -0.0750%\n",
      "Training completed!\n",
      "To continue, run this cell again.\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Repeat #2\n",
      "Training model for hidden-layer config of [2, 2] with 250 data samples\n",
      "No parameters loaded. Train from scratch.\n",
      "Created a new log.txt\n",
      "\n",
      "     EPOCH        LOSS   LOSS_AVE    IMPROVE?   MAPE     TIME/s        ELAPSED       CHANGE\n",
      "     10000     0.0001380768  1.38527e-04   YES    21.78%     0.00     0:00:21.491510   -0.0563%\n",
      "Training completed!\n",
      "To continue, run this cell again.\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAagAAAEYCAYAAAAJeGK1AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAAY1klEQVR4nO3dfbAldX3n8fcnw6MYiDLAKjBcWCDlaHwc8SGCihrBiOBDCeiqMVRYSflAslpCdDdhEzWQJ2sWas24sMiagPjIGNFIyAJJRHBQ1EGXdUAsRogM62YQZRHwu390XzjenAv3nnvuPX37vl9VXaf7d/r0fPs7p+p7f92/8+tUFZIkdc0vTDoASZKGsUBJkjrJAiVJ6iQLlCSpkyxQkqROskBJkjrJAiVJ6qQdJh3AYli9enVNTU1NOgxJ0hxcd911d1bVXjPbO1+gkhwH/DqwN3BOVX3xkT4zNTXFpk2bFjs0SdIYJPnesPaJXOJLcl6SO5JsntF+VJIbk2xJchpAVX2mqn4L+A3g+AmEK0magEndgzofOGqwIckq4BzgaGAtcGKStQO7vLd9X5K0AkykQFXVVcAPZzQfBmypqpur6qfARcCxaZwJfL6qvjrbMZOcnGRTkk3btm1bvOAlSUuiS6P49gVuHdje2ra9DXgx8Jokb5ntw1W1oarWVdW6vfb6V/faJEnLTJcGSWRIW1XVemD9UgcjSZqsLvWgtgL7D2zvB9w2oVgkSRPWpQL1FeCQJAcm2Qk4Adg4nwMkOSbJhu3bty9KgJKkpTOpYeYXAlcDv5xka5KTqup+4K3A3wLfBi6uqhvmc9yq+mxVnbzHHnuMP2hJ0pKayD2oqjpxlvZLgUuXOBxJUgd16RKfJEkPskBJkjqpVwXKQRKS1B+9KlAOkpCk/uhVgZIk9YcFSpLUSRYoSVIn9apAOUhCkvqjVwXKQRKS1B+9KlCSpP6wQEmSOskCJUnqpF4VKAdJSFJ/9KpAOUhCkvqjVwVKktQfIz0PKsmr5rDb/2uf7yRJ0ryN+sDCDwOXAHmYfY7Ahw9KkkY0aoH6fFX95sPtkOSjIx5bkqTR7kFV1b8bxz6SJM1mLIMkkhyc5KNJPpnkOeM45ohxOMxcknpipAKVZJcZTX8I/GfgNOC/LjSoUTnMXJL6Y9Qe1GeTvGFg+z5gql0eWGBMkiSNXKCOAvZI8oUkhwPvpBm1dzTw+nEFJ0lauUYaxVdVDwBnJ/kfwH8CHgf8x6q6aZzBSZJWrlF/qPss4F3AT4H3A/cA70uyFfjDqnKUgiRpQUb9HdSHgNcAjwb+sqp+FTghyfOBi4GXjik+SdIKNWqBeoBmQMSjaHpRAFTVlcCVCw9LkrTSjVqgXgf8e5ri9MbxhSNJUmPUQRL/G/gPY45lwZIcAxxz8MEHTzoUSdICjfpD3b8Zxz7j5g91Jak/Rr3E97wkGx/m/QBrRzy2JEkjF6hj57DPTx95F0mShhv1HpQj9SRJi8pHvkuSOskCJUnqpFFH8e3+MO+tGT0cSZIao/agrpheSXL5jPc+M2owkiRNG7VAZWD9sQ/zniRJIxm1QNUs68O2l4yPfJek/hj1d1B7J/ldmt7S9Drt9l5jiWwEVfVZ4LPr1q37rUnFIEkaj1EL1IeBXxyyDvDfFhSRJEmM/kPdMwCSrK6qO8cbkiRJow8zf3mSbcA3kmxN8twxxyVJWuFGHSTxfuDwqno88GrgA+MLSZKk0QvU/VX1vwCq6hp+/h6UJEkLttBRfEO3q+rPFxaWJGmlG8covpnbE/sdlCSpPxY0im+YJM8cPRxJkhqj9qB+TpK1wAnAicB2YN04jitJWrlGLlBJDqApSCcC9wMHAOuq6pbxhCZJWslG/R3Ul4BLgR2B11TVM4AfWZwkSeMy6jDzbTSDIvbhobn3HBwhSRqbkQpUVR0L/ArwVeCMJN8FHpPksHEGJ0lauUa+B1VV24HzgPOS7AMcD3wwyf5Vtf+4ApQkrUyjXuL7OVX1g6paX1XPBZ43jmOOwudBSVJ/jNSDSrLxEXZ5xSjHXSifByVJ/THqJb7nALcCFwLX4GPeJUljNmqB+jfAS2h+A/U64HPAhVV1w7gCkyStbKOO4nugqr5QVW8Cng1sAa5I8raxRidJWrEWMpPEzsCv0/SipoD1wKfGE5YkaaUbdZDER4AnAZ8HzqiqzWONSpK04o3ag3oD8GPgUODtyYNjJAJUVe0+htgkSSvYqI/bGMvvpyRJmo2FRpLUSRYoSVInWaAkSZ1kgZIkdZIFSpLUSRYoSVInWaAkSZ1kgZIkdZIFSpLUSRYoSVInWaAkSZ1kgZIkdZIFSpLUSRYoSVIndb5AJTkoyblJPjHpWCRJS2ciBSrJeUnuSLJ5RvtRSW5MsiXJaQBVdXNVnTSJOCVJkzOpHtT5wFGDDUlWAecARwNrgROTrF360CRJXTCRAlVVVwE/nNF8GLCl7TH9FLgIOHaux0xycpJNSTZt27ZtjNFKkiahS/eg9gVuHdjeCuybZM8kHwKeluT02T5cVRuqal1Vrdtrr70WO1ZJ0iLbYdIBDMiQtqqq/wO8ZamDkSRNVpd6UFuB/Qe29wNum1AskqQJ61IP6ivAIUkOBL4PnAC8bj4HSHIMcAxwV5LvjD/EJbcauHPSQXSAeWiYh4eYi0Zf8nDAsMZU1VIHQpILgRfQJPcHwO9X1blJXgZ8EFgFnFdV71vy4DokyaaqWjfpOCbNPDTMw0PMRaPveZhID6qqTpyl/VLg0iUOR5LUQV26ByVJ0oMsUN22YdIBdIR5aJiHh5iLRq/zMJF7UJIkPRJ7UJKkTrJASZI6yQK1xJK8I8nmJDckOXWg/W3tTO43JDlroP30dnb3G5O8dKD9GUm+2b63PsmwmTg6a1gekjw1yZeTXN/Oq3jYwP69yMOwmfyTPDbJZUm+074+ZuC9eZ13kp2TfKxtvybJ1JKe4DzMJxdJXpLkuvacr0ty5MBnlnUu5vudaN9fk+TuJO8caFvWeRiqqlyWaAGeBGwGHkUzxP/vgEOAF7brO7f77d2+rgW+DuwMHAjcBKxq37sWeA7NFFGfB46e9PmNIQ9fnD4P4GXAFX3LA3AE8HRg80DbWcBp7fppwJmjnjfw28CH2vUTgI9N+pzHlIunAY8f+P58f+AzyzoX88nDwPufBD4OvLMveRi22INaWk8AvlxVP6mq+4ErgVcCpwB/XFX3AlTVHe3+xwIXVdW9VfVdYAtwWJLHAbtX1dXVfOsuAI5b4nNZiNnyUMDu7T578NBUV73JQw2fyf9Y4CPt+kd46BxGOe/BY30CeFFXe5XzyUVVfa2qpr8PNwC7tD2DZZ+LeX4nSHIccDNNHqbbln0ehrFALa3NwBFpZmh/FE0vYX/gUODwtvt9ZZJntvsPneG9XbYOaV8uZsvDqcCfJLkV+FNgevb6vuZh2j5VdTtA+7p32z7KeT/4mbb4bwf2XLTIx2+2XAx6NfC19g+6vuZiaB6S7Aa8Gzhjxv69zEOX5uLrvar6dpIzgcuAu2ku39xP8//wGODZwDOBi5McxCwzvD9M+7LwMHk4BfidqvpkktcC5wIvpqd5mINRzrvXOUnyROBM4Nemm4bs1udcnAH8RVXdPaMT1Ms82INaYlV1blU9vaqOoOnWf4fmr51PVeNa4Gc08xTONsP71nZ9ZvuyMUse3gR8qt3l4zQPsYQe56H1g/YSzfSlmulLvKOc94OfSbIDzaXSmZePumy2XJBkP+DTwBur6qa2ua+5mC0PzwLOSnILzRWH30vyVnqaBwvUEksy3VVfA7wKuBD4DHBk234osBPNDMUbgRPaa+0H0gwkuLbt8v8oybPba8lvBC5Z6nNZiFnycBvw/HaXI2mKFvQ4D62NNMWZ9vWSgfb5nvfgsV4D/H17T2K5GJqLJL8EfA44var+aXrnHudiaB6q6vCqmqqqKZqJtd9fVWf3Ng+THqWx0hbgH4Bv0VzWelHbthPwUZp7M18FjhzY/z00o7duZGCEGrCu3f8m4GzaWUGWyzJLHp4HXNe2XQM8o295oCnEtwP30fxlexLN/YDLaQry5cBjRz1vYBea3ucWmlFdB036nMeRC+C9wI+B6weW6dGuyzoX8/1ODHzuD/j5UXzLOg/DFqc6kiR1kpf4JEmdZIGSJHWSBUqS1EkWKElSJ1mgJEmdZIGSlliSB9LM2D69nDbGY08NzootLWdOdSQtvXuq6qmTDkLqOntQUkckuSXJmUmubZeD2/YDklye5Bvt65q2fZ8kn07y9XZ5bnuoVUk+nOZZW19MsuvETkpaAAuUtPR2nXGJ7/iB9+6qqsNoZgL4YNt2NnBBVT0Z+Ctgfdu+Hriyqp5C8zyh6ccvHAKcU1VPBP6FZvZvadlxJglpiSW5u6oePaT9Fppprm5OsiPwz1W1Z5I7gcdV1X1t++1VtTrJNmC/ap8j1h5jCrisqg5pt98N7FhVf7QEpyaNlT0oqVtqlvXZ9hnm3oH1B/Bes5YpC5TULccPvF7drn+J5lHdAK8H/rFdv5zmGVokWZVk+mnEUi/4l5W09HZNcv3A9heqanqo+c5JrqH54/HEtu3twHlJ3gVsA97ctr8D2JDkJJqe0ik0s2JLveA9KKkj2ntQ66rqzknHInWBl/gkSZ1kD0qS1En2oCRJnWSBkiR1kgVKktRJFihJUidZoCRJnWSBkiR1kgVKktRJFihJUidZoCRJnWSBkiR1Ui9nM1+9enVNTU1NOgxJ0hxcd911d1bVXjPbe1mgpqam2LRp06TDkCTNQZLvDWvv/CW+JE9I8qEkn0hyyqTjkSQtjUUtUEnOS3JHks0z2o9KcmOSLUlOm+3zAFX17ap6C/BaYN1ixitJ6o7F7kGdDxw12JBkFXAOcDSwFjgxydokv5Lkb2Yse7efeQXNY64vX+R4JUkdsaj3oKrqqiRTM5oPA7ZU1c0ASS4Cjq2qDwAvn+U4G4GNST4H/PWwfZKcDJwMsGbNmvGcgCRpYiYxSGJf4NaB7a3As2bbOckLgFcBOwOXzrZfVW0ANgCsW7fOpzBK0jI3iQKVIW2zFpSqugK4YrGCkSR10yRG8W0F9h/Y3g+4bQJxSJI6bBIF6ivAIUkOTLITcAKwcRwHTnJMkg3bt28fx+EkSRO02MPMLwSuBn45ydYkJ1XV/cBbgb8Fvg1cXFU3jOPfq6rPVtXJe+yxxzgOJ0maoMUexXfiLO2X8jADHiRJ6vxMEpKklalXBcp7UJLUH70qUN6DkqT+6FWBkiT1hwVKktRJFihJUif1qkA5SEKS+qNXBcpBEpLUH70qUJKk/rBASZI6yQIlSeokC5QkqZN6VaAcxSdJ/dGrAuUoPknqj14VKElSf1igJEmdZIGSJHWSBUqS1EkWKElSJ/WqQDnMXJL6o1cFymHmktQfvSpQkqT+sEBJkjrJAiVJ6iQLlCSpkyxQkqROskBJkjqpVwXK30FJUn/MqUAl2S3JL7TrhyZ5RZIdFze0+fN3UJLUH3PtQV0F7JJkX+By4M3A+YsVlCRJcy1QqaqfAK8C/ktVvRJYu3hhSZJWujkXqCTPAV4PfK5t22FxQpIkae4F6lTgdODTVXVDkoOA/7loUUmSVrw59YKq6krgSoB2sMSdVfX2xQxMkrSyzXUU318n2T3JbsC3gBuTvGtxQ5MkrWRzvcS3tqruAo4DLgXWAG9YrKAkSZprgdqx/d3TccAlVXUfUIsWlSRpxZtrgfpL4BZgN+CqJAcAdy1WUJIkzalAVdX6qtq3ql5Wje8BL1zk2ObNqY4kqT/mOkhijyR/nmRTu/wZTW+qU5zqSJL6Y66X+M4DfgS8tl3uAv77YgUlSdJcZ4P4t1X16oHtM5JcvwjxSJIEzL0HdU+S501vJPlV4J7FCUmSpLn3oN4CXJBk+ubO/wXetDghSZI096mOvg48Jcnu7fZdSU4FvrGIsUmSVrB5PVG3qu5qZ5QA+N1FiEeSJGBhj3zP2KKQJGmGhRQopzqSJC2ah70HleRHDC9EAXZdlIgkSeIRClRV/eJSBSJJ0qCFXOKTJGnRWKAkSZ1kgZIkdVKvCpSP25Ck/uhVgfJxG5LUH70qUJKk/rBASZI6yQIlSeokC5QkqZMsUJKkTrJASZI6yQIlSeokC5QkqZMsUJKkTrJASZI6yQIlSeokC5QkqZMsUJKkTrJASZI6yQIlSeokC5QkqZMsUJKkTrJASZI6yQIlSeokC5QkqZMsUJKkTloWBSrJbkmuS/LyScciSVoai1qgkpyX5I4km2e0H5XkxiRbkpw2h0O9G7h4caKUJHXRDot8/POBs4ELphuSrALOAV4CbAW+kmQjsAr4wIzP/ybwZOBbwC6LHKskqUMWtUBV1VVJpmY0HwZsqaqbAZJcBBxbVR8A/tUlvCQvBHYD1gL3JLm0qn42ZL+TgZMB1qxZM9bzkCQtvcXuQQ2zL3DrwPZW4Fmz7VxV7wFI8hvAncOKU7vfBmADwLp162pcwUqSJmMSBSpD2h6xoFTV+eMPRZLUVZMYxbcV2H9gez/gtgnEIUnqsEkUqK8AhyQ5MMlOwAnAxnEcOMkxSTZs3759HIeTJE1Qqhbvdk2SC4EXAKuBHwC/X1XnJnkZ8EGakXvnVdX7xvzvbgO+N85jTshq4M5JB9EB5qFhHh5iLhp9ycMBVbXXzMZFLVBamCSbqmrdpOOYNPPQMA8PMReNvudhWcwkIUlaeSxQkqROskB124ZJB9AR5qFhHh5iLhq9zoP3oCRJnWQPSpLUSRYoSVInWaCWWJJ3JNmc5IYkpw60v619BMkNSc4aaD+9fSzJjUleOtD+jCTfbN9bn2TYFFKdNSwPSZ6a5MtJrk+yKclhA/v3Ig/DHkGT5LFJLkvynfb1MQPvzeu8k+yc5GNt+zVDJmvujPnkIslL2mfCfbN9PXLgM8s6F/P9TrTvr0lyd5J3DrQt6zwMVVUuS7QATwI2A4+imQfx74BDgBe26zu3++3dvq4Fvg7sDBwI3ASsat+7FngOzdyGnweOnvT5jSEPX5w+D+BlwBV9ywNwBPB0YPNA21nAae36acCZo5438NvAh9r1E4CPTfqcx5SLpwGPH/j+fH/gM8s6F/PJw8D7nwQ+DryzL3kYttiDWlpPAL5cVT+pqvuBK4FXAqcAf1xV9wJU1R3t/scCF1XVvVX1XWALcFiSxwG7V9XV1XzrLgCOW+JzWYjZ8lDA7u0+e/DQHI29yUNVXQX8cEbzscBH2vWP8NA5jHLeg8f6BPCirvYq55OLqvpaVU1/H24Adml7Bss+F/P8TpDkOOBmmjxMty37PAxjgVpam4EjkuyZ5FE0vYT9gUOBw9vu95VJntnuP+zRJPu2y9Yh7cvFbHk4FfiTJLcCfwqc3u7f1zxM26eqbgdoX/du20c57wc/0xb/7cCeixb5+M2Wi0GvBr7W/kHX11wMzUOS3WieMH7GjP17mYdJPG5jxaqqbyc5E7gMuJvm8s39NP8PjwGeDTwTuDjJQcz+aJKRHlnSFQ+Th1OA36mqTyZ5LXAu8GJ6moc5GOW8e52TJE8EzgR+bbppyG59zsUZwF9U1d0zOkG9zIM9qCVWVedW1dOr6giabv13aP7a+VQ1rgV+RjMJ5GyPJtnars9sXzZmycObgE+1u3yc5unL0OM8tH7QXqKZvlQzfYl3lPN+8DNJdqC5VDrz8lGXzZYLkuwHfBp4Y1Xd1Db3NRez5eFZwFlJbqG54vB7Sd5KT/NggVpiSaa76muAVwEXAp8BjmzbDwV2opmheCNwQnut/UCagQTXtl3+HyV5dnst+Y3AJUt9LgsxSx5uA57f7nIkTdGCHuehtZGmONO+XjLQPt/zHjzWa4C/b+9JLBdDc5Hkl4DPAadX1T9N79zjXAzNQ1UdXlVTVTVF80SI91fV2b3Nw6RHaay0BfgH4Fs0l7Ve1LbtBHyU5t7MV4EjB/Z/D83orRsZGKEGrGv3vwk4m3ZWkOWyzJKH5wHXtW3XAM/oWx5oCvHtwH00f9meRHM/4HKagnw58NhRzxvYhab3uYVmVNdBkz7nceQCeC/wY+D6gWV6tOuyzsV8vxMDn/sDfn4U37LOw7DFqY4kSZ3kJT5JUidZoCRJnWSBkiR1kgVKktRJFihJUidZoKQlluSBNDO2Ty+njfHYU4OzYkvLmVMdSUvvnqp66qSDkLrOHpTUEUluSXJmkmvb5eC2/YAklyf5Rvu6pm3fJ8mnk3y9XZ7bHmpVkg+nedbWF5PsOrGTkhbAAiUtvV1nXOI7fuC9u6rqMJqZAD7Ytp0NXFBVTwb+Cljftq8Hrqyqp9A8T2j68QuHAOdU1ROBf6GZ/VtadpxJQlpiSe6uqkcPab+FZpqrm5PsCPxzVe2Z5E7gcVV1X9t+e1WtTrIN2K/a54i1x5gCLquqQ9rtdwM7VtUfLcGpSWNlD0rqlpplfbZ9hrl3YP0BvNesZcoCJXXL8QOvV7frX6J5VDfA64F/bNcvp3mGFklWJZl+GrHUC/5lJS29XZNcP7D9haqaHmq+c5JraP54PLFteztwXpJ3AduAN7ft7wA2JDmJpqd0Cs2s2FIveA9K6oj2HtS6qrpz0rFIXeAlPklSJ9mDkiR1kj0oSVInWaAkSZ1kgZIkdZIFSpLUSRYoSVIn/X+r5dfCRixRdAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Repeat #3\n",
      "Training model for hidden-layer config of [2, 2] with 250 data samples\n",
      "No parameters loaded. Train from scratch.\n",
      "Created a new log.txt\n",
      "\n",
      "     EPOCH        LOSS   LOSS_AVE    IMPROVE?   MAPE     TIME/s        ELAPSED       CHANGE\n",
      "     10000     0.0003213300  3.25192e-04   YES    23.30%     0.00     0:00:21.617637   -0.0248%\n",
      "Training completed!\n",
      "To continue, run this cell again.\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Repeat #4\n",
      "Training model for hidden-layer config of [2, 2] with 250 data samples\n",
      "No parameters loaded. Train from scratch.\n",
      "Created a new log.txt\n",
      "\n",
      "     EPOCH        LOSS   LOSS_AVE    IMPROVE?   MAPE     TIME/s        ELAPSED       CHANGE\n",
      "     10000     0.0001740271  1.74469e-04   YES    19.10%     0.00     0:00:21.479216   -0.0053%\n",
      "Training completed!\n",
      "To continue, run this cell again.\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from core import train_DimNet\n",
    "from datetime import datetime\n",
    "import os\n",
    "from joblib import dump\n",
    "import torch\n",
    "\n",
    "# config the widths of the two hidden layers of the DimNet, respectively\n",
    "N_HIDDEN = [2,2]\n",
    "\n",
    "# set the number of repeats for training DimNets \n",
    "# note: each repeat uses a different seed for initialization\n",
    "N_REPEAT = 50\n",
    "\n",
    "# set the starting number of the current run of repeats; default = 0\n",
    "# As you may run this cell for several times, every time you run this cell, \n",
    "# set N_START = 1 + # of the last repeat in the last run, so that the previous output files won't be overriden\n",
    "N_START = 0\n",
    "\n",
    "# Create a name for the job. \n",
    "# The output results, models, etc. will be saved under a folder with this name\n",
    "NAME = \"smooth\"\n",
    "\n",
    "mape_ = []\n",
    "init_seed_ = []\n",
    "\n",
    "index_ = list(range(N_START,N_START+N_REPEAT))\n",
    "\n",
    "for i in index_:\n",
    "    print(f\"Repeat #{i}\")\n",
    "    print(f\"Training model for hidden-layer config of {N_HIDDEN} with {SIZE} data samples\")  \n",
    "    _,mape,init_seed = train_DimNet(df_train,col_x,col_y,NAME,N_HIDDEN,i_repeat=i,\n",
    "                                        BATCH_SIZE = len(df_train),\n",
    "                                        LR = 0.001,\n",
    "                                        MIN_EPOCH = 20000,\n",
    "                                        MAX_EPOCH = 100000,\n",
    "                                        INTERVAL_PRINT = 10000,\n",
    "                                        INTERVAL_SAVE = 100,\n",
    "                                        WIN_LEN = 100,\n",
    "                                        CHANGE_RATE_TOLERANCE = 1e-6,\n",
    "                                        SOLVER = 'Adam',\n",
    "                                        SHUFFLE = True)\n",
    "    mape_.append(mape)\n",
    "    init_seed_.append(init_seed)\n",
    "\n",
    "sorted_list = sorted(zip(index_,mape_,init_seed_), key=lambda x:x[1])\n",
    "i_best = sorted_list[0][0]\n",
    "mape_best = sorted_list[0][1]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "compliant-camcorder",
   "metadata": {},
   "source": [
    "### 3. Print & log training information"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "statewide-medicare",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==========2022-04-15 12:09:29.858598==========\n",
      "\n",
      "size of training set = 250\n",
      "hidden layer config = [2, 2]\n",
      "solver = Adams\n",
      "\n",
      "BEST_REPEAT = 0, with lowest MAPE = 18.00%\n"
     ]
    }
   ],
   "source": [
    "path_log = os.path.join(NAME,'log.txt')\n",
    "with open(path_log, 'a') as f_log:\n",
    "    f_log.write('\\n\\n')\n",
    "    f_log.write('='*10 + f\"{datetime.now()}\" + '='*10)\n",
    "    f_log.write(f'\\nsize of training set = {SIZE}\\n'+\n",
    "                f'hidden layer config = {str(N_HIDDEN)}\\n'+\n",
    "                'solver = Adams\\n'+\n",
    "                '\\nREPEAT#  MAPE    SEED\\n')\n",
    "    \n",
    "    print('='*10 + f\"{datetime.now()}\" + '='*10)\n",
    "    print(f'\\nsize of training set = {SIZE}\\n'+\n",
    "          f'hidden layer config = {str(N_HIDDEN)}\\n'+\n",
    "          'solver = Adams')\n",
    "    \n",
    "    for t in sorted_list:\n",
    "        f_log.write(f'{t[0]}    {t[1]*100:.2f}%    {t[2]}\\n')\n",
    "\n",
    "    f_log.write(f'\\nBEST_REPEAT = {i_best}, with lowest MAPE = {mape_best*100:.2f}%')\n",
    "    print(f'\\nBEST_REPEAT = {i_best}, with lowest MAPE = {mape_best*100:.2f}%')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "revolutionary-southeast",
   "metadata": {},
   "source": [
    "### 4. Pick and load the best DimNet model\n",
    "\n",
    "* Prior to run the cell below, you need to manually type the BEST_REPEAT.\n",
    "* If you've run the training cell more than once, you may compare different runs and pick the very best BEST_REPEAT\n",
    "* more info can be found in log.txt the {NAME} directory. In this case, the path is smooth/log.txt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "boring-practitioner",
   "metadata": {},
   "outputs": [],
   "source": [
    "BEST_REPEAT = 1\n",
    "\n",
    "# load the correpsonding .model file corresponding to the best REPEAT #\n",
    "from joblib import load\n",
    "\n",
    "path = f'{NAME}/{NAME}_{BEST_REPEAT}.model'\n",
    "# You may comment out the code above and explicitly write the path for the model \n",
    "# path = 'smooth/smooth_0.model'\n",
    "\n",
    "model = load(path)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "developmental-income",
   "metadata": {},
   "source": [
    "### 5. You may skip the steps 2-4 and load a pre-trained model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "collected-transcript",
   "metadata": {},
   "outputs": [],
   "source": [
    "from joblib import load\n",
    "\n",
    "path = 'data/Noisy_Smooth_200_[2, 2]_75.model'\n",
    "# path = 'data/Noisy_Smooth_200_[2, 3]_77.model'\n",
    "# path = 'data/Noisy_Smooth_200_[3, 1]_17.model'\n",
    "# path = 'data/Noisy_Smooth_200_[3, 2]_98.model'\n",
    "# path = 'data/Noisy_Smooth_200_[3, 3]_27.model'\n",
    "\n",
    "model = load(path)\n",
    "\n",
    "# note: you may ignore the UserWarning about unpicke estimator StandardScaler/MaxAbsScaler"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "adjusted-springer",
   "metadata": {},
   "source": [
    "### 6. Evaluate the model performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "lightweight-partner",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 400x400 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred = model.predict(df_train[col_x].values)\n",
    "y_label = df_train.f.values\n",
    "\n",
    "from core import compute_metrics\n",
    "mape, p30, p50 = compute_metrics(y_label,y_pred, verbose=False)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=[4,4], dpi=100)\n",
    "ax.plot(y_label, y_pred, marker='o', markersize=2, ls='')\n",
    "ax.annotate(f\"MAPE={mape*100:.1f}%\\nP₃₀={p30*100:.1f}%\\n\", \n",
    "            xy=(0.05,0.95), xycoords='axes fraction', \n",
    "            horizontalalignment='left', verticalalignment='top', \n",
    "            fontsize='medium')\n",
    "\n",
    "x = np.linspace(0,1,100)\n",
    "lw = 0.75\n",
    "ax.plot(x,x,'-.',c='k',linewidth=lw)\n",
    "ax.plot(x,1.3*x,'--',c='k',linewidth=lw)\n",
    "ax.plot(x,0.7*x,'--',c='k',linewidth=lw)\n",
    "ax.annotate('+30%', xy=(0.7,0.95), xycoords='axes fraction',\n",
    "            horizontalalignment='left', verticalalignment='top', fontsize='medium')\n",
    "ax.annotate('-30%', xy=(0.88,0.78), xycoords='axes fraction', \n",
    "            horizontalalignment='left', verticalalignment='top', fontsize='medium')\n",
    "\n",
    "ax.set_xlabel('label')\n",
    "ax.set_ylabel('prediction')\n",
    "ax.set_xscale('log')\n",
    "ax.set_yscale('log')\n",
    "\n",
    "ax.set_xlim(0.005,0.1)\n",
    "ax.set_ylim(0.005,0.1)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "unexpected-parallel",
   "metadata": {},
   "source": [
    "### 7. Convert the model to an algebraic function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "inclusive-compound",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--------------------------------------------------------------------------------\n",
      "DimNet Config: 1 - 2 - 2 - 1\n",
      "--------------------------------------------------------------------------------\n",
      "Input variable(s): Re\n",
      "Output variable(s): f\n",
      "--------------------------------------------------------------------------------\n",
      "Algebraic Correlation:\n",
      "\n",
      "z1 = -1.147e+00 + ln(Re**0.25)\n",
      "z2 = 4.570e+00 + ln(Re**-0.57)\n",
      " \n",
      "if z1 > 0 and z2 > 0:\n",
      "    f = -1.483e-03 + 9.070e-12 * (Re**2.69) + 6.605e+01 * (Re**-1.01) \n",
      "if z1 > 0 and z2 <= 0:\n",
      "    f = -1.483e-03 + 6.716e-02 * (Re**-0.12) + 2.275e+00 * (Re**-0.59) \n",
      "if z1 <= 0 and z2 > 0:\n",
      "    f = -1.483e-03 + 5.218e-12 * (Re**2.81) + 4.450e+00 * (Re**-0.42) \n",
      "if z1 <= 0 and z2 <= 0:\n",
      "    f = -1.483e-03 + 3.863e-02 * (Re**0.00) + 1.532e-01 * (Re**0.00)\n",
      "--------------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "from convert import dimnet_to_correlation\n",
    "\n",
    "dimnet_to_correlation(model, col_x=col_x, col_y=col_y, verbose=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "freelance-brazilian",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}