baseline_example.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example using Python Script\n",
    "\n",
    "This notebook uses the `investment_euler.py` provided in this repository and is not entirely self-contained."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\jesse\\anaconda3\\envs\\pytorch-sandbox\\lib\\site-packages\\tqdm\\auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import sys\n",
    "import torch\n",
    "import pytorch_lightning as pl\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from pytorch_lightning.cli import LightningCLI\n",
    "from investment_euler import InvestmentEuler"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "  The `args` allows us to override some of the default values.  Everything else is in the `investment_euler_simple.yaml` case, which does not have logging enabled.  See `investment_euler.py` and related docs for a full version."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Global seed set to 123\n",
      "GPU available: False, used: False\n",
      "TPU available: False, using: 0 TPU cores\n",
      "IPU available: False, using: 0 IPUs\n",
      "HPU available: False, using: 0 HPUs\n",
      "c:\\Users\\jesse\\anaconda3\\envs\\pytorch-sandbox\\lib\\site-packages\\pytorch_lightning\\callbacks\\model_checkpoint.py:604: UserWarning: Checkpoint directory c:\\Users\\jesse\\Documents\\GitHub\\symmetry_dynamic_programming\\checkpoints exists and is not empty.\n",
      "  rank_zero_warn(f\"Checkpoint directory {dirpath} exists and is not empty.\")\n",
      "\n",
      "  | Name     | Type    | Params\n",
      "-------------------------------------\n",
      "0 | ml_model | DeepSet | 199 K \n",
      "-------------------------------------\n",
      "199 K     Trainable params\n",
      "0         Non-trainable params\n",
      "199 K     Total params\n",
      "0.798     Total estimated model params size (MB)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 2: 100%|██████████| 96/96 [00:04<00:00, 21.98it/s, loss=3.24e-07, val_loss=1.95e-7, val_u_rel_error=0.000907, val_u_abs_error=2.93e-5]\n"
     ]
    }
   ],
   "source": [
    "args = {\"trainer.max_epochs\": 10,\n",
    "        \"model.N\": 128,\n",
    "        \"model.nu\": 1, # LQ example nested if nu=1\n",
    "        } # override some defaults\n",
    "sys.argv = [\"dummy.py\"]  # workaround since LightningCLI doesn't like being called from a notebook\n",
    "\n",
    "cli = LightningCLI(\n",
    "    InvestmentEuler,\n",
    "    run=False,\n",
    "    seed_everything_default=123,\n",
    "    save_config_callback=None,\n",
    "    parser_kwargs={\"default_config_files\": [\"investment_euler_simple.yaml\"]}, # note using a \"simple\" version without logging\n",
    "    args = args\n",
    ")\n",
    "\n",
    "# Fit the model.  The construction created internal training and validation points\n",
    "cli.trainer.fit(cli.model)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After the model has been trained, we can test it against simulated values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Testing DataLoader 0: 100%|██████████| 128/128 [00:02<00:00, 51.14it/s]\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<pre style=\"white-space:pre;overflow-x:auto;line-height:normal;font-family:Menlo,'DejaVu Sans Mono',consolas,'Courier New',monospace\">┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┓\n",
       "┃<span style=\"font-weight: bold\">        Test metric        </span>┃<span style=\"font-weight: bold\">       DataLoader 0        </span>┃\n",
       "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━┩\n",
       "│<span style=\"color: #008080; text-decoration-color: #008080\">         test_loss         </span>│<span style=\"color: #800080; text-decoration-color: #800080\">  2.0451582827263337e-07   </span>│\n",
       "│<span style=\"color: #008080; text-decoration-color: #008080\">     test_u_abs_error      </span>│<span style=\"color: #800080; text-decoration-color: #800080\">   2.93638640869176e-05    </span>│\n",
       "│<span style=\"color: #008080; text-decoration-color: #008080\">     test_u_rel_error      </span>│<span style=\"color: #800080; text-decoration-color: #800080\">   0.0009061677264980972   </span>│\n",
       "└───────────────────────────┴───────────────────────────┘\n",
       "</pre>\n"
      ],
      "text/plain": [
       "┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┓\n",
       "┃\u001b[1m \u001b[0m\u001b[1m       Test metric       \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1m      DataLoader 0       \u001b[0m\u001b[1m \u001b[0m┃\n",
       "┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━┩\n",
       "│\u001b[36m \u001b[0m\u001b[36m        test_loss        \u001b[0m\u001b[36m \u001b[0m│\u001b[35m \u001b[0m\u001b[35m 2.0451582827263337e-07  \u001b[0m\u001b[35m \u001b[0m│\n",
       "│\u001b[36m \u001b[0m\u001b[36m    test_u_abs_error     \u001b[0m\u001b[36m \u001b[0m│\u001b[35m \u001b[0m\u001b[35m  2.93638640869176e-05   \u001b[0m\u001b[35m \u001b[0m│\n",
       "│\u001b[36m \u001b[0m\u001b[36m    test_u_rel_error     \u001b[0m\u001b[36m \u001b[0m│\u001b[35m \u001b[0m\u001b[35m  0.0009061677264980972  \u001b[0m\u001b[35m \u001b[0m│\n",
       "└───────────────────────────┴───────────────────────────┘\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[{'test_loss': 2.0451582827263337e-07,\n",
       "  'test_u_rel_error': 0.0009061677264980972,\n",
       "  'test_u_abs_error': 2.93638640869176e-05}]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cli.trainer.test(cli.model) # test the model on generated values, storing the results in model.test_results"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are three types of points here, each with a separate `loss`:\n",
    "1. `train_loss` is loss applied to the generated points where we are attempting to interpolate.  Given the large number of parameters, having a low `train_loss` just means it successfully interpolated, not that it will necessarily generalize.\n",
    "2. `val_loss` is that applied to a set of generated points which are not trained on, but which are used within the optimization process (e.g., deciding when to stop training, adjusting learning rates).  This gives a better sense of generalization because they are \"out of sample\", but when they are used to guide the optimization process they become somewhat contaminated and a low `val_loss` might hide poor generalization.\n",
    "3. `test_loss` is applied to a simulated set of points that aren't used in the training process at all.  It is the most representative of generalization, and a low `test_loss` is the ultimate goal of the process.\n",
    "\n",
    "Here, the default stopping criteria is when the `val_loss` becomes sufficiently low relative to numerical precision."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'Time($t$)')"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model = cli.model\n",
    "# Plot the results of the previous cell\n",
    "df = model.test_results[\n",
    "    model.test_results[\"ensemble\"] == 0\n",
    "]  # first ensemble in dataframe from model on test data\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "ax.plot(df[\"t\"], df[\"u_hat\"], label=r\"$u(X_t)$, $\\phi($ReLU$)$\")\n",
    "if model.hparams.nu == 1.0:  # only add reference line if linear\n",
    "    ax.plot(df[\"t\"], df[\"u_reference\"], dashes=[10, 5, 10, 5], label=r\"$u(X_t)$, LQ\")\n",
    "ax.legend()\n",
    "ax.set_title(r\"$u(X_t)$ with $\\phi($ReLU$)$ : Equilibrium Path\")\n",
    "ax.set_xlabel(r\"Time($t$)\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[0.0234],\n",
      "        [0.0210]])\n"
     ]
    }
   ],
   "source": [
    "# Example to evaluate  model after fitting it, can do at multiple points at same time:\n",
    "model.eval()\n",
    "X_points = torch.stack((model.X_0 + 0.001 * torch.randn(model.hparams.N),\n",
    "                        model.X_0 + 0.05 * torch.ones(model.hparams.N)))\n",
    "# evaluate the policy.  Doing it with a `no_grad` can speed things up \n",
    "with torch.no_grad():\n",
    "    u_X = model(X_points) # evaluating like a function provides the policy because of the .forward() method\n",
    "    print(u_X)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "pytorch-sandbox",
   "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": "3.9.15"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "535f64b549f1c1a44024bff541c70f1e607ec1cdf54c7c015722119882ca2e26"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}