Upgrade Multi-Objective Algorithm examples page

docs/source/example/moalg.ipynb

@@ -0,0 +1,164 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Multi-Objective Algorithm"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "In this notebook, we will use the Reference Vector Guided Evolutionary Algorithm (**RVEA**) to find the optimal solutions of the **DTLZ2** problem."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import time\n",
+    "import torch\n",
+    "from torch.profiler import ProfilerActivity, profile\n",
+    "\n",
+    "from evox.algorithms import RVEA\n",
+    "from evox.metrics import igd\n",
+    "from evox.problems.numerical import DTLZ2\n",
+    "from evox.workflows import StdWorkflow"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## (Optional) Use GPU to run the code\n",
+    "We often prefer to run our code on a GPU for faster execution. However, if a GPU is unavailable, running on a CPU is also acceptable."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "cuda:0\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Use GPU first to run the code.\n",
+    "torch.set_default_device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
+    "print(torch.get_default_device())"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Running example: RVEA on DTLZ2 problem\n",
+    "The following code is used to set up the DTLZ2 problem and the RVEA algorithm. More information about the problem and algorithm can be found in the corresponding section of the documentation."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Init the problem, algorithm and workflow.\n",
+    "prob = DTLZ2(m=3)\n",
+    "pf = prob.pf()\n",
+    "algo = RVEA(pop_size=100, n_objs=3, lb=-torch.zeros(12), ub=torch.ones(12))\n",
+    "workflow = StdWorkflow()\n",
+    "workflow.setup(algo, prob)\n",
+    "workflow.init_step()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "With this setup in place, we can now start to optimize. We set to let the multi-objective algorithm optimize for 100 steps on this problem"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "tensor(0.4746, device='cuda:0')\n",
+      "tensor(0.1650, device='cuda:0')\n",
+      "tensor(0.0902, device='cuda:0')\n",
+      "tensor(0.0653, device='cuda:0')\n",
+      "tensor(0.0586, device='cuda:0')\n",
+      "tensor(0.0564, device='cuda:0')\n",
+      "tensor(0.0559, device='cuda:0')\n",
+      "tensor(0.0551, device='cuda:0')\n",
+      "tensor(0.0546, device='cuda:0')\n",
+      "tensor(0.0544, device='cuda:0')\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Run the workflow for 100 steps\n",
+    "t = time.time()\n",
+    "with profile(activities=[ProfilerActivity.CPU, ProfilerActivity.CUDA], record_shapes=True, profile_memory=True) as prof:\n",
+    "    for i in range(100):\n",
+    "        workflow.step()\n",
+    "        fit = workflow.algorithm.fit\n",
+    "        fit = fit[~torch.isnan(fit).any(dim=1)]\n",
+    "        if i % 10 == 0:\n",
+    "          print(igd(fit, pf))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "(Optional) If needed, you can use the following code to get more information"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# print(prof.key_averages().table())\n",
+    "# torch.cuda.synchronize()\n",
+    "# print(time.time() - t)"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": ".venv",
+   "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": "3.12.0"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}