{ "cells": [ { "cell_type": "markdown", "metadata": { "id": "OGBwo5rUCf4s" }, "source": [ "# One-Max Problem\n", "\n", "[![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/google/pyglove/blob/main/docs/notebooks/evolution/onemax.ipynb)\n", "\n", "\n", "One-Max is a simple problem often used to introduce the concept of genetic algorithms. The problem boils down to one question: **what is the maximum sum of a bitstring (a string consisting of only 1s and 0s) of length N?**\n", "\n", "This tutorial will show how this problem can be solved in PyGlove's way, with an option to use user-defined genome encoding.\n", "\n", "*(Disclaimer: PyGlove is designed to be used with problems with evaluation time greater than sub-second, since the onemax problem can be evaluated in nano/micro-seconds, you may find slowdown in PyGlove's solution. Bear in mind that the overhead from PyGlove for each iteration is in sub-seconds)*" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "Ij_a3QK_CZRG" }, "outputs": [], "source": [ "!pip install pyglove" ] }, { "cell_type": "markdown", "metadata": { "id": "wGLk5t4ZDSVq" }, "source": [ "## Solving One-Max with `pg.oneof`\n", "\n", "First we show that the search space can be expressed with `pg.oneof`, a PyGlove's built-in primitive for search space specification." ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "height": 282 }, "executionInfo": { "elapsed": 6604, "status": "ok", "timestamp": 1656370912687, "user": { "displayName": "Daiyi Peng", "userId": "00212084952560362045" }, "user_tz": 420 }, "id": "t2S9AiHLDRy2", "outputId": "7581f7b8-ff25-4c66-b6e2-4311bfc16533" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Best sequence: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1] (sum=10)\n" ] }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXAAAAD4CAYAAAD1jb0+AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90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\nGCansAFnGIbJKWzAGYZhcsr/B8u078eA2dpHAAAAAElFTkSuQmCC\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "import pyglove as pg\n", "import matplotlib.pyplot as plt\n", "\n", "def one_max(search_space, search_algorithm, num_trials=200) -> None:\n", " best_sequence, best_reward = None, None\n", " history = []\n", " for sequence, feedback in pg.sample(\n", " search_space, search_algorithm,\n", " num_examples=num_trials):\n", " reward = sum(sequence)\n", " if best_reward is None or best_reward < reward:\n", " best_sequence, best_reward = sequence, reward\n", " feedback(reward)\n", " history.append(reward)\n", " print(f'Best sequence: {list(best_sequence)} (sum={best_reward})')\n", " plt.plot(list(range(num_trials)), history)\n", " plt.show()\n", "\n", "N = 10\n", "search_space = pg.List([pg.oneof([0, 1])] * N)\n", "search_algorithm = pg.evolution.regularized_evolution(\n", " population_size=20, tournament_size=10)\n", "\n", "one_max(search_space, search_algorithm)" ] }, { "cell_type": "markdown", "metadata": { "id": "P11qylOOSy7p" }, "source": [ "## Solving the same problem with custom genome encoding\n", "\n", "At times, users might want to encode the search space without using PyGlove's built-in search primitives, when they have a more efficient genome representation, or to support new semantics of the search space (e.g. variable number of decision points). In such cases, PyGlove allows the user to define their own search primitives by extending `pg.hyper.CustomHyper`. The user needs to implement `custom_decode` method, which takes a `pg.DNA` with string value and returns its corresponding client object for evaluation. In this case, the client object is a list of binary integers. \n", "\n", "The user needs to also implement a mutator for such custom-encoded DNA, and provides a DNA generator for generating the initial population. The custom mutator can be combined with PyGlove's built-in evolution operations to create new algorithms. " ] }, { "cell_type": "code", "execution_count": null, "metadata": { "colab": { "height": 282 }, "executionInfo": { "elapsed": 999, "status": "ok", "timestamp": 1656371233008, "user": { "displayName": "Daiyi Peng", "userId": "00212084952560362045" }, "user_tz": 420 }, "id": "cIOWBxvMTPaY", "outputId": "3bb32077-0de9-442d-f1d0-06c630153d84" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Best sequence: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1] (sum=10)\n" ] }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXAAAAD4CAYAAAD1jb0+AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90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\nYII=\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "import random\n", "\n", "class BitString(pg.hyper.CustomHyper):\n", "\n", " def custom_decode(self, dna: pg.DNA):\n", " assert isinstance(dna.value, str)\n", " bitstr = dna.value\n", " return [int(x) for x in bitstr]\n", "\n", "class MutateOneBit(pg.evolution.Mutator):\n", "\n", " def mutate(self, dna: pg.DNA):\n", " bitstr = dna.value\n", " assert len(bitstr) > 0\n", " index = random.randint(0, len(dna.value) - 1)\n", " new_bitstr = (bitstr[:index]\n", " + ('0' if bitstr[index] == '1' else '1')\n", " + bitstr[index + 1:])\n", " return pg.DNA(new_bitstr)\n", "\n", "def init_population(population_size, n):\n", " @pg.geno.dna_generator\n", " def initializer(dna_spec):\n", " for _ in range(population_size):\n", " bits = [str(random.randint(0, 1)) for _ in range(n)]\n", " yield pg.DNA(''.join(bits))\n", " return initializer()\n", " \n", "search_space = BitString()\n", "search_algorithm = pg.evolution.Evolution(\n", " (pg.evolution.selectors.Random(10)\n", " >> pg.evolution.selectors.Top(1)\n", " >> MutateOneBit()),\n", " population_init=init_population(10, N),\n", " population_update=pg.evolution.selectors.Last(20))\n", "\n", "one_max(search_space, search_algorithm)" ] } ], "metadata": { "colab": { "collapsed_sections": [], "last_runtime": { "build_target": "//learning/deepmind/dm_python:dm_notebook3", "kind": "private" }, "name": "One-Max Problem", "provenance": [ { "file_id": "1NwzvqHrZPVNfaLChubFtIGO_SOKYnSpd", "timestamp": 1656372916534 } ] }, "kernelspec": { "display_name": "Python 3", "name": "python3" }, "language_info": { "name": "python" } }, "nbformat": 4, "nbformat_minor": 0 }