All Downloads are FREE. Search and download functionalities are using the official Maven repository.

org.cicirello.search.operators.reals.GaussianMutation Maven / Gradle / Ivy

Go to download

Chips-n-Salsa is a Java library of customizable, hybridizable, iterative, parallel, stochastic, and self-adaptive local search algorithms. The library includes implementations of several stochastic local search algorithms, including simulated annealing, hill climbers, as well as constructive search algorithms such as stochastic sampling. Chips-n-Salsa now also includes genetic algorithms as well as evolutionary algorithms more generally. The library very extensively supports simulated annealing. It includes several classes for representing solutions to a variety of optimization problems. For example, the library includes a BitVector class that implements vectors of bits, as well as classes for representing solutions to problems where we are searching for an optimal vector of integers or reals. For each of the built-in representations, the library provides the most common mutation operators for generating random neighbors of candidate solutions, as well as common crossover operators for use with evolutionary algorithms. Additionally, the library provides extensive support for permutation optimization problems, including implementations of many different mutation operators for permutations, and utilizing the efficiently implemented Permutation class of the JavaPermutationTools (JPT) library. Chips-n-Salsa is customizable, making extensive use of Java's generic types, enabling using the library to optimize other types of representations beyond what is provided in the library. It is hybridizable, providing support for integrating multiple forms of local search (e.g., using a hill climber on a solution generated by simulated annealing), creating hybrid mutation operators (e.g., local search using multiple mutation operators), as well as support for running more than one type of search for the same problem concurrently using multiple threads as a form of algorithm portfolio. Chips-n-Salsa is iterative, with support for multistart metaheuristics, including implementations of several restart schedules for varying the run lengths across the restarts. It also supports parallel execution of multiple instances of the same, or different, stochastic local search algorithms for an instance of a problem to accelerate the search process. The library supports self-adaptive search in a variety of ways, such as including implementations of adaptive annealing schedules for simulated annealing, such as the Modified Lam schedule, implementations of the simpler annealing schedules but which self-tune the initial temperature and other parameters, and restart schedules that adapt to run length.

There is a newer version: 7.0.1
Show newest version
/*
 * Chips-n-Salsa: A library of parallel self-adaptive local search algorithms.
 * Copyright (C) 2002-2022 Vincent A. Cicirello
 *
 * This file is part of Chips-n-Salsa (https://chips-n-salsa.cicirello.org/).
 *
 * Chips-n-Salsa is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * Chips-n-Salsa is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see .
 */

package org.cicirello.search.operators.reals;

import java.util.function.DoubleBinaryOperator;
import java.util.function.IntFunction;
import org.cicirello.math.rand.RandomSampler;
import org.cicirello.math.rand.RandomVariates;
import org.cicirello.search.representations.RealValued;
import org.cicirello.util.Copyable;

/**
 * This class implements Gaussian mutation. Gaussian mutation is for mutating floating-point values.
 * This class can be used to mutate objects of any of the classes that implement the {@link
 * RealValued} interface, including both univariate and multivariate function input objects.
 *
 * 

In a Gaussian mutation, a value v is mutated by adding a randomly generated m such that m is * drawn from a Gaussian distribution with mean 0 and standard deviation sigma. It is commonly * employed in Evolution Strategies when mutating real valued parameters. * *

This mutation operator also implements the {@link RealValued} interface to enable * implementation of metaheuristics that mutate their own mutation parameters. That is, you can pass * a GaussianMutation object to the {@link #mutate} method of a GaussianMutation object. * *

To construct a GaussianMutation, you must use one of the factory methods. See the various * {@link #createGaussianMutation} methods. * *

Gaussian mutation was introduced in the following article:
* Hinterding, R. 1995. Gaussian mutation and self-adaption for numeric genetic algorithms. In IEEE * CEC. IEEE Press. 384–389. * * @param The specific RealValued type. * @author Vincent A. Cicirello, https://www.cicirello.org/ */ public class GaussianMutation extends AbstractRealMutation implements Copyable> { /* * Internal constructor. Constructs a Gaussian mutation operator. * Otherwise, must use the factory methods. * * @param sigma The standard deviation of the Gaussian. * * @param transformer The functional transformation of the mutation. */ GaussianMutation(double sigma, DoubleBinaryOperator transformer) { super(sigma, transformer); } /* * Internal constructor. Constructs a Gaussian mutation operator. * Otherwise, must use the factory methods. * * @param sigma The standard deviation of the Gaussian. * * @param transformer The functional transformation of the mutation. * * @param selector Chooses the indexes for a partial mutation. */ GaussianMutation(double sigma, DoubleBinaryOperator transformer, IntFunction selector) { super(sigma, transformer, selector); } /* * internal copy constructor */ GaussianMutation(GaussianMutation other) { super(other); } /** * Creates a Gaussian mutation operator with standard deviation equal to 1. * * @param The specific RealValued type. * @return A Gaussian mutation operator. */ public static GaussianMutation createGaussianMutation() { return createGaussianMutation(1.0); } /** * Creates a Gaussian mutation operator. * * @param sigma The standard deviation of the Gaussian. * @param The specific RealValued type. * @return A Gaussian mutation operator. */ public static GaussianMutation createGaussianMutation(double sigma) { return new GaussianMutation(sigma, (old, param) -> old + RandomVariates.nextGaussian(param)); } /** * Creates a Gaussian mutation operator, such that the mutate method constrains each mutated real * value to lie in the interval [lowerBound, upperBound]. * * @param sigma The standard deviation of the Gaussian. * @param lowerBound A lower bound on the result of a mutation. * @param upperBound An upper bound on the result of a mutation. * @param The specific RealValued type. * @return A Gaussian mutation operator. */ public static GaussianMutation createGaussianMutation( double sigma, double lowerBound, double upperBound) { if (upperBound < lowerBound) throw new IllegalArgumentException("upperBound must be at least lowerBound"); return new GaussianMutation( sigma, (old, param) -> { double mutated = old + RandomVariates.nextGaussian(param); if (mutated <= lowerBound) return lowerBound; if (mutated >= upperBound) return upperBound; return mutated; }); } /** * Create a Gaussian mutation operator. * * @param sigma The standard deviation of the Gaussian mutation. * @param k The number of input variables that the {@link #mutate} method changes when called. The * k input variables are chosen uniformly at random from among all subsets of size k. If there * are less than k input variables, then all are mutated. * @param The specific RealValued type. * @return A Gaussian mutation operator * @throws IllegalArgumentException if k < 1 */ public static GaussianMutation createGaussianMutation( double sigma, int k) { if (k < 1) throw new IllegalArgumentException("k must be at least 1"); return new GaussianMutation( sigma, (old, param) -> old + RandomVariates.nextGaussian(param), n -> RandomSampler.sample(n, k < n ? k : n, (int[]) null)); } /** * Create a Gaussian mutation operator. * * @param sigma The standard deviation of the Gaussian mutation. * @param p The probability that the {@link #mutate} method changes an input variable. If there * are n input variables, then n*p input variables will be mutated on average during a single * call to the {@link #mutate} method. * @param The specific RealValued type. * @return A Gaussian mutation operator * @throws IllegalArgumentException if p ≤ 0 */ public static GaussianMutation createGaussianMutation( double sigma, double p) { if (p <= 0) throw new IllegalArgumentException("p must be positive"); if (p >= 1) { return createGaussianMutation(sigma); } return new GaussianMutation( sigma, (old, param) -> old + RandomVariates.nextGaussian(param), n -> RandomSampler.sample(n, p)); } @Override public GaussianMutation split() { return new GaussianMutation(this); } /** * Creates an identical copy of this object. * * @return an identical copy of this object */ @Override public GaussianMutation copy() { return new GaussianMutation(this); } }





© 2015 - 2025 Weber Informatics LLC | Privacy Policy