add emulator ipynb example

examples/hybrid_emulator_demo.ipynb

@@ -0,0 +1,259 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "a4a3f852",
+   "metadata": {},
+   "source": [
+    "# Ice Sheet Emulator Demo"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "e7ebd4e4",
+   "metadata": {},
+   "source": [
+    "### Installation\n",
+    "\n",
+    "To install, run this is your terminal. This will create an environment, download ISE, and activate a jupyter notebook instance. Open up this notebook to run the model.  \n",
+    "\n",
+    "```conda create -n ise -y```  \n",
+    "```conda activate ise```  \n",
+    "```conda install nb_conda ipykernel -y```  \n",
+    "```pip install git+https://github.com/Brown-SciML/ise```  \n",
+    "```jupyter notebook```  "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "a13a05eb",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import numpy as np\n",
+    "\n",
+    "from ise.models.hybrid import HybridEmulator, DeepEnsemble, NormalizingFlow\n",
+    "from sklearn.preprocessing import StandardScaler"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "0882b6e8",
+   "metadata": {},
+   "source": [
+    "### Synthetic Data and Scaling"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "a814d0bd",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "X = np.array([[1,2,3], [4,5,6], [7,8,9]], dtype=int)\n",
+    "y = np.array([[1], [2], [3],], dtype=int)\n",
+    "\n",
+    "scaler_y = StandardScaler().fit(y)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "8847c8d1",
+   "metadata": {},
+   "source": [
+    "### Create emulator Model\n",
+    "Create the model with the Deep Ensemble as the Predictor and the Normalizing Flow as the uncertainty quantifier."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "0beb86de",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "input_shape = X.shape[1]\n",
+    "output_shape = y.shape[1]\n",
+    "num_ensemble_members = 2\n",
+    "\n",
+    "predictor = DeepEnsemble(num_predictors=num_ensemble_members, forcing_size=input_shape, sle_size=output_shape)\n",
+    "uncertainty_quantifier = NormalizingFlow(forcing_size=input_shape, sle_size=output_shape)\n",
+    "emulator = HybridEmulator(deep_ensemble=predictor, normalizing_flow=uncertainty_quantifier)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "bf7a5dc3",
+   "metadata": {
+    "scrolled": false
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Training Normalizing Flow (10 epochs):\n",
+      "Epoch 1, Loss: 4.753732204437256\n",
+      "Epoch 2, Loss: 4.68987512588501\n",
+      "Epoch 3, Loss: 4.626246929168701\n",
+      "Epoch 4, Loss: 4.562777042388916\n",
+      "Epoch 5, Loss: 4.499433517456055\n",
+      "Epoch 6, Loss: 4.436193943023682\n",
+      "Epoch 7, Loss: 4.373051166534424\n",
+      "Epoch 8, Loss: 4.310001850128174\n",
+      "Epoch 9, Loss: 4.247047424316406\n",
+      "Epoch 10, Loss: 4.1841888427734375\n",
+      "\n",
+      "Training Deep Ensemble (15 epochs):\n",
+      "Training Weak Predictor 1 of 2:\n",
+      "Epoch 1, Average Batch Loss: 4.4524664878845215\n",
+      "Epoch 2, Average Batch Loss: 4.2489094734191895\n",
+      "Epoch 3, Average Batch Loss: 4.11221170425415\n",
+      "Epoch 4, Average Batch Loss: 3.971240282058716\n",
+      "Epoch 5, Average Batch Loss: 3.7946555614471436\n",
+      "Epoch 6, Average Batch Loss: 3.5632998943328857\n",
+      "Epoch 7, Average Batch Loss: 3.2641429901123047\n",
+      "Epoch 8, Average Batch Loss: 2.8817074298858643\n",
+      "Epoch 9, Average Batch Loss: 2.406568765640259\n",
+      "Epoch 10, Average Batch Loss: 1.8449459075927734\n",
+      "Epoch 11, Average Batch Loss: 1.2372466325759888\n",
+      "Epoch 12, Average Batch Loss: 0.6458789706230164\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "C:\\Users\\pvank\\miniconda3\\envs\\ise\\Lib\\site-packages\\ise\\data\\dataclasses.py:33: UserWarning: Full projections of 86 timesteps are not present in the dataset. This may lead to unexpected behavior.\n",
+      "  warnings.warn(\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 13, Average Batch Loss: 0.20140044391155243\n",
+      "Epoch 14, Average Batch Loss: 0.067240871489048\n",
+      "Epoch 15, Average Batch Loss: 0.31725022196769714\n",
+      "\n",
+      "Training Weak Predictor 2 of 2:\n",
+      "Epoch 1, Average Batch Loss: 1.4070473909378052\n",
+      "Epoch 2, Average Batch Loss: 1.3644334077835083\n",
+      "Epoch 3, Average Batch Loss: 1.327694296836853\n",
+      "Epoch 4, Average Batch Loss: 1.2840057611465454\n",
+      "Epoch 5, Average Batch Loss: 1.2289267778396606\n",
+      "Epoch 6, Average Batch Loss: 1.1562410593032837\n",
+      "Epoch 7, Average Batch Loss: 1.0581227540969849\n",
+      "Epoch 8, Average Batch Loss: 0.9263723492622375\n",
+      "Epoch 9, Average Batch Loss: 0.7507588863372803\n",
+      "Epoch 10, Average Batch Loss: 0.5248854756355286\n",
+      "Epoch 11, Average Batch Loss: 0.26091358065605164\n",
+      "Epoch 12, Average Batch Loss: 0.05477694049477577\n",
+      "Epoch 13, Average Batch Loss: 0.053589772433042526\n",
+      "Epoch 14, Average Batch Loss: 0.23185868561267853\n",
+      "Epoch 15, Average Batch Loss: 0.30263322591781616\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "emulator.fit(X, y, nf_epochs=10, de_epochs=15)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "b70b8513",
+   "metadata": {},
+   "source": [
+    "### Predict on unseen data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "a525e483",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "predictions, uncertainties = emulator.predict(np.array([[10, 11, 12]]), output_scaler=scaler_y)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "2318cea9",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "array([[4.0753965]], dtype=float32)"
+      ]
+     },
+     "execution_count": 6,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "predictions"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "64d26fc6",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "{'total': array([[135.88687]], dtype=float32),\n",
+       " 'epistemic': array([[0.06036377]], dtype=float32),\n",
+       " 'aleatoric': array([[135.8265]], dtype=float32)}"
+      ]
+     },
+     "execution_count": 7,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "uncertainties"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "f62b427f",
+   "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": "3.11.5"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}