Operators

Author:

Antonio J. Nebro

Date:

2022-12-1

Evolutionary algorithms use selection and variation operators for getting the mating pool and for generate a descendant population from this pool, respectively. In the case of genetic algorithms, variation operators are mutation and crossover.

The Operator interface (package org.uma.jmetal.operator) belongs the jmetal-core sub-project, and it is defined as follows:

package org.uma.jmetal.operator;

/** Interface representing an operator
 *
 * @author Antonio J. Nebro <antonio@lcc.uma.es>
 * @version 1.0

 * @param <Source> Source Class of the object to be operated with
 * @param <Result> Result Class of the result obtained after applying the operator
 */
public interface Operator<Source, Result> extends Serializable {
  /**
   * @param source The data to process
   */
  Result execute(Source source) ;
}

According to this interface, an operator is a generic entity providing an execute() method that is applied to a source and returns a result. The source and result are typically a solution or a list of solutions, depending on the operator.

Four classes of operators are currently included:

• Crossover. From a set of parent solutions, recombine them somehow to produce a number of children.

• Mutation. Mutates a solution.

• Selection. Select a number of solutions from a list according to some criterion.

• LocalSearchOperator. Applies a local search method to improve a solution.

In general, operators are related to the types of solutions to which they apply. For example, BLXAlphaCrossover is applied to DoubleSolution objects and its declaration is as follows:

public class BLXAlphaCrossover implements CrossoverOperator<DoubleSolution>

The currently implemented operators are included in the next figures: