NSGAIIStudy2.java

//  This program is free software: you can redistribute it and/or modify
//  it under the terms of the GNU Lesser General Public License as published by
//  the Free Software Foundation, either version 3 of the License, or
//  (at your option) any later version.
//
//  This program 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 Lesser General Public License for more details.
//
//  You should have received a copy of the GNU Lesser General Public License
//  along with this program.  If not, see <http://www.gnu.org/licenses/>.

package org.uma.jmetal.experiment;

import org.uma.jmetal.algorithm.Algorithm;
import org.uma.jmetal.algorithm.multiobjective.nsgaii.NSGAIIBuilder;
import org.uma.jmetal.operator.impl.crossover.SBXCrossover;
import org.uma.jmetal.operator.impl.mutation.PolynomialMutation;
import org.uma.jmetal.problem.Problem;
import org.uma.jmetal.problem.multiobjective.zdt.*;
import org.uma.jmetal.qualityindicator.impl.*;
import org.uma.jmetal.qualityindicator.impl.hypervolume.PISAHypervolume;
import org.uma.jmetal.solution.DoubleSolution;
import org.uma.jmetal.util.JMetalException;
import org.uma.jmetal.util.experiment.Experiment;
import org.uma.jmetal.util.experiment.ExperimentBuilder;
import org.uma.jmetal.util.experiment.component.*;
import org.uma.jmetal.util.experiment.util.ExperimentAlgorithm;
import org.uma.jmetal.util.experiment.util.ExperimentProblem;

import java.io.IOException;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

/**
 * Example of experimental study based on solving the ZDT problems with four versions of NSGA-II, each
 * of them applying a different crossover probability (from 0.7 to 1.0).
 *
 * This experiment assumes that the reference Pareto front are not known, so the names of files containing
 * them and the directory where they are located must be specified.
 *
 * Six quality indicators are used for performance assessment.
 *
 * The steps to carry out the experiment are:
 * 1. Configure the experiment
 * 2. Execute the algorithms
 * 3. Generate the reference Pareto fronts
 * 4. Compute the quality indicators
 * 5. Generate Latex tables reporting means and medians
 * 6. Generate Latex tables with the result of applying the Wilcoxon Rank Sum Test
 * 7. Generate Latex tables with the ranking obtained by applying the Friedman test
 * 8. Generate R scripts to obtain boxplots
 *
 * @author Antonio J. Nebro <antonio@lcc.uma.es>
 */
public class NSGAIIStudy2 {
  private static final int INDEPENDENT_RUNS = 25 ;

  public static void main(String[] args) throws IOException {
    if (args.length != 1) {
      throw new JMetalException("Needed arguments: experimentBaseDirectory") ;
    }
    String experimentBaseDirectory = args[0] ;

    List<ExperimentProblem<DoubleSolution>> problemList = new ArrayList<>();
    problemList.add(new ExperimentProblem<>(new ZDT1()));
    problemList.add(new ExperimentProblem<>(new ZDT2()));
    problemList.add(new ExperimentProblem<>(new ZDT3()));
    problemList.add(new ExperimentProblem<>(new ZDT4()));
    problemList.add(new ExperimentProblem<>(new ZDT6()));

    List<ExperimentAlgorithm<DoubleSolution, List<DoubleSolution>>> algorithmList =
            configureAlgorithmList(problemList);

    Experiment<DoubleSolution, List<DoubleSolution>> experiment =
        new ExperimentBuilder<DoubleSolution, List<DoubleSolution>>("NSGAIIStudy2")
            .setAlgorithmList(algorithmList)
            .setProblemList(problemList)
            .setExperimentBaseDirectory(experimentBaseDirectory)
            .setOutputParetoFrontFileName("FUN")
            .setOutputParetoSetFileName("VAR")
            .setReferenceFrontDirectory(experimentBaseDirectory+"/referenceFronts")
            .setIndicatorList(Arrays.asList(
                new Epsilon<DoubleSolution>(), new Spread<DoubleSolution>(), new GenerationalDistance<DoubleSolution>(),
                new PISAHypervolume<DoubleSolution>(),
                new InvertedGenerationalDistance<DoubleSolution>(), new InvertedGenerationalDistancePlus<DoubleSolution>()))
            .setIndependentRuns(INDEPENDENT_RUNS)
            .setNumberOfCores(8)
            .build();

    new ExecuteAlgorithms<>(experiment).run();
    new GenerateReferenceParetoSetAndFrontFromDoubleSolutions(experiment).run();
    new ComputeQualityIndicators<>(experiment).run() ;
    new GenerateLatexTablesWithStatistics(experiment).run() ;
    new GenerateWilcoxonTestTablesWithR<>(experiment).run() ;
    new GenerateFriedmanTestTables<>(experiment).run();
    new GenerateBoxplotsWithR<>(experiment).setRows(3).setColumns(3).run() ;
  }

  /**
   * The algorithm list is composed of pairs {@link Algorithm} + {@link Problem} which form part of
   * a {@link ExperimentAlgorithm}, which is a decorator for class {@link Algorithm}. The {@link
   * ExperimentAlgorithm} has an optional tag component, that can be set as it is shown in this example,
   * where four variants of a same algorithm are defined.
   */
  static List<ExperimentAlgorithm<DoubleSolution, List<DoubleSolution>>> configureAlgorithmList(
          List<ExperimentProblem<DoubleSolution>> problemList) {
    List<ExperimentAlgorithm<DoubleSolution, List<DoubleSolution>>> algorithms = new ArrayList<>();

    for (int i = 0; i < problemList.size(); i++) {
      Algorithm<List<DoubleSolution>> algorithm = new NSGAIIBuilder<>(
              problemList.get(i).getProblem(),
              new SBXCrossover(1.0, 5),
              new PolynomialMutation(1.0 / problemList.get(i).getProblem().getNumberOfVariables(), 10.0))
              .setMaxEvaluations(25000)
              .setPopulationSize(100)
              .build();
      algorithms.add(new ExperimentAlgorithm<>(algorithm, "NSGAIIa", problemList.get(i).getTag()));
    }

    for (int i = 0; i < problemList.size(); i++) {
      Algorithm<List<DoubleSolution>> algorithm = new NSGAIIBuilder<>(
              problemList.get(i).getProblem(),
              new SBXCrossover(1.0, 20.0),
              new PolynomialMutation(1.0 / problemList.get(i).getProblem().getNumberOfVariables(), 20.0))
              .setMaxEvaluations(25000)
              .setPopulationSize(100)
              .build();
      algorithms.add(new ExperimentAlgorithm<>(algorithm, "NSGAIIb", problemList.get(i).getTag()));
    }

    for (int i = 0; i < problemList.size(); i++) {
      Algorithm<List<DoubleSolution>> algorithm = new NSGAIIBuilder<>(problemList.get(i).getProblem(), new SBXCrossover(1.0, 40.0),
              new PolynomialMutation(1.0 / problemList.get(i).getProblem().getNumberOfVariables(), 40.0))
              .setMaxEvaluations(25000)
              .setPopulationSize(100)
              .build();
      algorithms.add(new ExperimentAlgorithm<>(algorithm, "NSGAIIc", problemList.get(i).getTag()));
    }

    for (int i = 0; i < problemList.size(); i++) {
      Algorithm<List<DoubleSolution>> algorithm = new NSGAIIBuilder<>(problemList.get(i).getProblem(), new SBXCrossover(1.0, 80.0),
              new PolynomialMutation(1.0 / problemList.get(i).getProblem().getNumberOfVariables(), 80.0))
              .setMaxEvaluations(25000)
              .setPopulationSize(100)
              .build();
      algorithms.add(new ExperimentAlgorithm<>(algorithm, "NSGAIId", problemList.get(i).getTag()));
    }

    return algorithms;
  }

}