Added examples from the EvA2Doc

resources/MatlabInterface/JEInterface-examples.m

@@ -1,11 +1,12 @@
-clear all
-clear classes
+%% Usage examples for the EvA2 to Matlab interface.
+%  Author: Marcel Kronfeld, Chair for Cognitive Systems, University of Tuebingen, Germany
+%  URL: http://www.ra.cs.uni-tuebingen.de/software/EvA2/
 
 % adapt the path settings!
-addpath '/home/mkron/workspace/JE2Base/resources/MatlabInterface'
-javaaddpath '/home/mkron/workspace/JE2Base/build'
-addpath 'C:\Dokumente und Einstellungen\mkron\workspace\JE2Base\resources\MatlabInterface'
-javaaddpath 'C:\Dokumente und Einstellungen\mkron\workspace\JE2Base\build'
+addpath '/home/user/workspace/MatlabInterface' 	% .. directory containing @JEInterface
+javaaddpath '/home/user/workspace/EvA2Base.jar'  	% .. the EvA2 base package
+% addpath 'C:\Dokumente und Einstellungen\user\workspace\MatlabInterface'	% Windows will look differently
+% javaaddpath 'C:\Dokumente und Einstellungen\user\workspace\EvA2Base.jar'	% Windows will look differently
 
 % real valued case
 R=[-5 -5 -5; 5 5 5];
@@ -28,4 +29,3 @@ JI=JEInterface(@testfun, 'int', R, initR, 5, 'Display', 'iter');
 JI=optimize(JI, 3);
 [sol, fit]=getResult(JI);
 finalPop=getMultipleResults(JI);
-

resources/MatlabInterface/testfun.m

@@ -1,3 +1,7 @@
+%% Usage examples for the EvA2 to Matlab interface.
+%  Author: Marcel Kronfeld, Chair for Cognitive Systems, University of Tuebingen, Germany
+%  URL: http://www.ra.cs.uni-tuebingen.de/software/EvA2/
+
 function z = testfun(x, y)
 switch y
     case 1 % modulated parabola

src/eva2/examples/TerminatorExample.java

@@ -0,0 +1,31 @@
+package eva2.examples;
+import eva2.OptimizerFactory;
+import eva2.server.go.problems.F1Problem;
+import eva2.server.go.operators.terminators.*;
+import eva2.server.go.operators.terminators.PopulationMeasureTerminator.*;
+
+public class TerminatorExample {
+	public static void main(String[] args) {
+		F1Problem f1 = new F1Problem(); 
+		double[] sol;
+		// A combined terminator for fitness and phenotype convergence
+		CombinedTerminator convT = new CombinedTerminator(
+				// fitness-based stagnation period, absolute threshold, consider stagnation 
+				// in both direction (per dim.) or w.r.t. minimization only
+				new FitnessConvergenceTerminator(0.0001, 1000, StagnationTypeEnum.fitnessCallBased, ChangeTypeEnum.absoluteChange, DirectionTypeEnum.decrease),
+				new PhenotypeConvergenceTerminator(0.0001, 1000, StagnationTypeEnum.fitnessCallBased, ChangeTypeEnum.absoluteChange, DirectionTypeEnum.bidirectional),
+				CombinedTerminator.AND);
+		// Adding an evaluation terminator with OR to the convergence criterion
+		OptimizerFactory.setTerminator(new CombinedTerminator(
+				new EvaluationTerminator(20000),
+				convT,
+				CombinedTerminator.OR));
+		sol = OptimizerFactory.optimizeToDouble(OptimizerFactory.PSO, f1, null);
+		System.out.println(OptimizerFactory.lastEvalsPerformed() 
+				+ " evals performed. "
+				+ OptimizerFactory.terminatedBecause()
+				+ " Found solution: ");
+		for (int i=0; i<f1.getProblemDimension(); i++) System.out.print(sol[i] + " ");
+		System.out.println();
+	};
+}

src/eva2/examples/TestingCbnPostProc.java

@@ -0,0 +1,39 @@
+package eva2.examples;
+import java.util.Vector;
+import eva2.OptimizerFactory;
+import eva2.server.go.individuals.AbstractEAIndividual;
+import eva2.server.go.operators.postprocess.PostProcessParams;
+import eva2.server.go.operators.terminators.EvaluationTerminator;
+import eva2.server.go.problems.FM0Problem;
+import eva2.server.modules.GOParameters;
+
+public class TestingCbnPostProc {
+	public static void main(String[] args) {
+		// a simple bimodal target function, two optima near (1.7,0) and (-1.44/0)
+		FM0Problem fm0 = new FM0Problem();              
+		AbstractEAIndividual best;              
+		Vector<AbstractEAIndividual> ppSols;
+
+		GOParameters esParams = OptimizerFactory.standardCbnES(fm0);            
+		esParams.setTerminator(new EvaluationTerminator(2000));                 
+		esParams.setSeed(0);
+		best = (AbstractEAIndividual)OptimizerFactory.optimizeToInd(esParams, null);
+
+		System.out.println(esParams.getTerminator().lastTerminationMessage() + "\nFound solution: "                           
+				+ AbstractEAIndividual.getDefaultDataString(best));     
+
+		// post-process with clustering only            
+		ppSols = OptimizerFactory.postProcessIndVec(new PostProcessParams(0, 0.1, 5));          
+		System.out.println("After clustering: ");               
+		for (AbstractEAIndividual indy : ppSols) {                      
+			System.out.println(AbstractEAIndividual.getDefaultDataString(indy));            
+		}       
+
+		// post-process with clustering and hill climbing               
+		ppSols = OptimizerFactory.postProcessIndVec(new PostProcessParams(1000, 0.1, 5));               
+		System.out.println("After clustering and local refinement: ");          
+		for (AbstractEAIndividual indy : ppSols) {                      
+			System.out.println(AbstractEAIndividual.getDefaultDataString(indy));            
+		}
+	};
+}

src/eva2/examples/TestingF1PSO.java

@@ -0,0 +1,15 @@
+package eva2.examples;
+import eva2.OptimizerFactory;
+import eva2.server.go.problems.F1Problem;
+
+public class TestingF1PSO {
+	public static void main(String[] args) {
+		F1Problem f1 = new F1Problem(); 
+		// start a PSO with a runtime of 50000 evaluations
+		OptimizerFactory.setEvaluationTerminator(50000);
+		double[] sol = OptimizerFactory.optimizeToDouble(OptimizerFactory.PSO, f1, null);
+		System.out.println(OptimizerFactory.terminatedBecause() + "\nFound solution: ");
+		for (int i=0; i<f1.getProblemDimension(); i++) System.out.print(sol[i] + " ");
+		System.out.println();
+	};
+}

src/eva2/examples/TestingGAB1.java

@@ -0,0 +1,35 @@
+package eva2.examples;
+import java.util.BitSet;
+import eva2.OptimizerFactory;
+import eva2.server.go.operators.selection.SelectXProbRouletteWheel;
+import eva2.server.go.operators.terminators.EvaluationTerminator;
+import eva2.server.go.populations.Population;
+import eva2.server.go.problems.B1Problem;
+import eva2.server.go.strategies.GeneticAlgorithm;
+import eva2.server.modules.GOParameters;
+
+public class TestingGAB1 {
+	public static void main(String[] args) {
+		B1Problem b1 = new B1Problem();
+		BitSet sol;
+		// default go-parameter instance with a GA
+		GOParameters gaParams = OptimizerFactory.standardGA(b1);                
+		// add an evaluation terminator
+		gaParams.setTerminator(new EvaluationTerminator(1000));
+		// set a specific random seed
+		gaParams.setSeed(2342);
+
+		// access the GA
+		GeneticAlgorithm ga = (GeneticAlgorithm)gaParams.getOptimizer();
+		ga.setElitism(false);
+		ga.setParentSelection(new SelectXProbRouletteWheel()); // roulette wheel selection
+		ga.setPopulation(new Population(150));	// population size 150
+
+		// run optimization and print intermediate results to a file with given prefix
+		sol = OptimizerFactory.optimizeToBinary(gaParams, "ga-opt-results");
+		System.out.println(OptimizerFactory.terminatedBecause() + "\nFound solution: ");
+		for (int i=0; i<b1.getProblemDimension(); i++) System.out.print(sol.get(i)+" ");
+		System.out.println();
+	};
+}
+

src/eva2/examples/TestingPlusCmaEs.java

@@ -0,0 +1,40 @@
+package eva2.examples;
+import eva2.OptimizerFactory;
+import eva2.server.go.individuals.AbstractEAIndividual;
+import eva2.server.go.operators.crossover.CrossoverESDefault;
+import eva2.server.go.operators.mutation.MutateESCovarianceMatrixAdaption;
+import eva2.server.go.operators.terminators.EvaluationTerminator;
+import eva2.server.go.problems.FM0Problem;
+import eva2.server.go.strategies.EvolutionStrategies;
+import eva2.server.modules.GOParameters;
+
+public class TestingPlusCmaEs {
+	public static void main(String[] args) {
+		// a simple bimodal target function, two optima near (1.7,0) and (-1.44/0)
+		FM0Problem fm0 = new FM0Problem();              
+		AbstractEAIndividual bestIndy;          
+		// create standard ES parameters                
+		GOParameters esParams = OptimizerFactory.standardES(fm0);               
+		esParams.setTerminator(new EvaluationTerminator(2000));                 
+		// set a random seed based on system time               
+		esParams.setSeed(0);
+
+		// set evolutionary operators and probabilities                 
+		AbstractEAIndividual.setOperators(                              
+				fm0.getIndividualTemplate(),                            
+				new MutateESCovarianceMatrixAdaption(true), 0.9,                            
+				new CrossoverESDefault(), 0.1);         
+
+		// access the ES                
+		EvolutionStrategies es = (EvolutionStrategies)esParams.getOptimizer();          
+		// set a (1+5) selection strategy               
+		es.setMu(1);            
+		es.setLambda(5);                
+		es.setPlusStrategy(true);       
+
+		// run optimization and retrieve winner individual              
+		bestIndy = (AbstractEAIndividual)OptimizerFactory.optimizeToInd(esParams, null);    
+		System.out.println(esParams.getTerminator().lastTerminationMessage() + "\nFound solution: " 
+				+ AbstractEAIndividual.getDefaultDataString(bestIndy));
+	};
+}