halfdan/eva2
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merging mk branch revs. 543:546 (except niching constructors in OptimizerFactory)

Browse Source
This commit is contained in:
Marcel Kronfeld
2010-05-31 12:03:42 +00:00
parent b735e9402b
commit 89975518ef
4 changed files with 90 additions and 67 deletions
Download Patch File Download Diff File Expand all files Collapse all files

121
src/eva2/OptimizerFactory.java
View File

@@ -10,6 +10,7 @@ import eva2.server.go.IndividualInterface;
import eva2.server.go.InterfacePopulationChangedEventListener;
import eva2.server.go.InterfaceTerminator;
import eva2.server.go.enums.DETypeEnum;
import eva2.server.go.enums.MutateESCrossoverTypeEnum;
import eva2.server.go.enums.PSOTopologyEnum;
import eva2.server.go.enums.PostProcessMethod;
import eva2.server.go.individuals.AbstractEAIndividual;
@@ -144,11 +145,7 @@ public class OptimizerFactory {
problem.initProblem();
AbstractEAIndividual tmpIndi = problem.getIndividualTemplate();
tmpIndi.setCrossoverOperator(new NoCrossover());
tmpIndi.setCrossoverProbability(0.0);
tmpIndi.setMutationOperator(new NoMutation());
tmpIndi.setMutationProbability(0.0);
setTemplateOperators(problem, new NoMutation(), 0, new NoCrossover(), 0);
DifferentialEvolution de = new DifferentialEvolution();
de.SetProblem(problem);
@@ -261,11 +258,7 @@ public class OptimizerFactory {
problem.initProblem();
AbstractEAIndividual tmpIndi = problem.getIndividualTemplate();
tmpIndi.setCrossoverOperator(cross);
tmpIndi.setCrossoverProbability(pc);
tmpIndi.setMutationOperator(mut);
tmpIndi.setMutationProbability(pm);
setTemplateOperators(problem, mut, pm, cross, pc);
GeneticAlgorithm ga = new GeneticAlgorithm();
ga.SetProblem(problem);
@@ -374,13 +367,7 @@ public class OptimizerFactory {
problem.initProblem();
MutateESFixedStepSize mutator = new MutateESFixedStepSize();
mutator.setSigma(0.2); // mutations step size
AbstractEAIndividual tmpIndi = problem.getIndividualTemplate();
tmpIndi.setMutationOperator(mutator);
tmpIndi.setMutationProbability(1.0);
tmpIndi.setCrossoverOperator(new NoCrossover());
tmpIndi.setCrossoverProbability(0);
setTemplateOperators(problem, new MutateESFixedStepSize(0.2), 1., new NoCrossover(), 0);
HillClimbing hc = new HillClimbing();
hc.getPopulation().setTargetSize(pop);
@@ -395,7 +382,7 @@ public class OptimizerFactory {
/**
* This method performs a Monte Carlo Search with the given number of
* fitnesscalls.
* fitness calls.
*
* @param problem
* @param popsize
@@ -407,11 +394,7 @@ public class OptimizerFactory {
InterfacePopulationChangedEventListener listener) {
problem.initProblem();
AbstractEAIndividual tmpIndi = problem.getIndividualTemplate();
tmpIndi.setMutationOperator(new NoMutation());
tmpIndi.setMutationProbability(0);
tmpIndi.setCrossoverOperator(new NoCrossover());
tmpIndi.setCrossoverProbability(0);
setTemplateOperators(problem, new NoMutation(), 0, new NoCrossover(), 0);
MonteCarloSearch mc = new MonteCarloSearch();
mc.getPopulation().setTargetSize(popsize);
@@ -426,7 +409,7 @@ public class OptimizerFactory {
/**
* This method performs a particle swarm optimization. Standard topologies are
* linear (0), grid (1) and star (2).
* linear, grid and star.
*
* @param problem
* @param mut
@@ -447,11 +430,7 @@ public class OptimizerFactory {
problem.initProblem();
AbstractEAIndividual tmpIndi = problem.getIndividualTemplate();
tmpIndi.setCrossoverOperator(new NoCrossover());
tmpIndi.setCrossoverProbability(0.0);
tmpIndi.setMutationOperator(new NoMutation());
tmpIndi.setMutationProbability(0.0);
setTemplateOperators(problem, new NoMutation(), 0, new NoCrossover(), 0);
ParticleSwarmOptimization pso = new ParticleSwarmOptimization();
pso.SetProblem(problem);
@@ -492,11 +471,7 @@ public class OptimizerFactory {
problem.initProblem();
AbstractEAIndividual tmpIndi = problem.getIndividualTemplate();
tmpIndi.setCrossoverOperator(new NoCrossover());
tmpIndi.setCrossoverProbability(0.0);
tmpIndi.setMutationOperator(mut);
tmpIndi.setMutationProbability(1.0);
setTemplateOperators(problem, mut, 1, new NoCrossover(), 0);
SimulatedAnnealing sa = new SimulatedAnnealing();
sa.setAlpha(alpha);
@@ -927,19 +902,19 @@ public class OptimizerFactory {
String outputFilePrefix) {
OptimizerRunnable runnable = optimize(new OptimizerRunnable(params,
outputFilePrefix));
return runnable.getSolutionSet();
return runnable.getResultPopulation();
}
public static Population optimizeToPop(final int optType,
AbstractOptimizationProblem problem, String outputFilePrefix) {
OptimizerRunnable runnable = optimize(optType, problem,
outputFilePrefix);
return (runnable != null) ? runnable.getSolutionSet() : null;
return (runnable != null) ? runnable.getResultPopulation() : null;
}
public static Population optimizeToPop(OptimizerRunnable runnable) {
optimize(runnable);
return (runnable != null) ? runnable.getSolutionSet() : null;
return (runnable != null) ? runnable.getResultPopulation() : null;
}
///////////////////////////// post processing
@@ -985,7 +960,7 @@ public class OptimizerFactory {
runnable.run(); // this run will not set the lastRunnable -
// postProcessing
// starts always anew
return runnable.getSolutionSet();
return runnable.getResultPopulation();
}
public static Vector<BitSet> postProcessBinVec(int steps, double sigma,
@@ -1354,17 +1329,10 @@ public class OptimizerFactory {
es.setLambda(50);
es.setPlusStrategy(false);
AbstractEAIndividual indyTemplate = problem.getIndividualTemplate();
if ((indyTemplate != null)
&& (indyTemplate instanceof InterfaceESIndividual)) {
// Set CMA operator for mutation
AbstractEAIndividual indy = (AbstractEAIndividual) indyTemplate;
MutateESCovarianceMatrixAdaption cmaMut = new MutateESCovarianceMatrixAdaption();
cmaMut.setCheckConstraints(true);
AbstractEAIndividual.setOperators(indy, cmaMut, 1., new CrossoverESDefault(), 0.);
} else {
System.err
.println("Error, CMA-ES is implemented for ES individuals only (requires double data types)");
if (assertIndyType(problem,InterfaceESIndividual.class)) {
setTemplateOperators(problem, new MutateESCovarianceMatrixAdaption(true), 1, new CrossoverESDefault(), 0);
} else {
System.err.println("Error, CMA-ES is implemented for ES individuals only (requires double data types)");
return null;
}
@@ -1417,6 +1385,12 @@ public class OptimizerFactory {
return makeParams(nms, 50, problem, randSeed, makeDefaultTerminator());
}
/**
* This constructs a standard DE variant with current-to-best/1 scheme, F=0.8,
* k=0.6, lambda=0.6. The population size is 50 by default.
* @param problem
* @return
*/
public static final GOParameters standardDE(
AbstractOptimizationProblem problem) {
DifferentialEvolution de = new DifferentialEvolution();
@@ -1424,7 +1398,7 @@ public class OptimizerFactory {
de.setF(0.8);
de.setK(0.6);
de.setLambda(0.6);
de.setMt(0.05);
de.setMt(0.05); // this is not really employed for currentToBest
return makeParams(de, 50, problem, randSeed, makeDefaultTerminator());
}
@@ -1435,18 +1409,12 @@ public class OptimizerFactory {
es.setLambda(50);
es.setPlusStrategy(false);
AbstractEAIndividual indy = problem.getIndividualTemplate();
if ((indy != null) && (indy instanceof InterfaceESIndividual)) {
// Set CMA operator for mutation
indy.setMutationOperator(new MutateESGlobal());
indy.setCrossoverOperator(new CrossoverESDefault());
} else {
System.err
.println("Error, standard ES is implemented for ES individuals only (requires double data types)");
if (assertIndyType(problem,InterfaceESIndividual.class)) {
setTemplateOperators(problem, new MutateESGlobal(0.2, MutateESCrossoverTypeEnum.intermediate), 0.9, new CrossoverESDefault(), 0.2);
} else {
System.err.println("Error, standard ES is implemented for ES individuals only (requires double data types)");
return null;
}
return makeESParams(es, problem);
}
@@ -1458,6 +1426,12 @@ public class OptimizerFactory {
return makeParams(ga, 100, problem, randSeed, makeDefaultTerminator());
}
/**
* A standard constricted PSO instance with phi1=phi2=2.05, grid topology of range 1 and
* a population size of 30. The chi parameter is set to approx. 0.73 automatically.
* @param problem
* @return
*/
public static final GOParameters standardPSO(
AbstractOptimizationProblem problem) {
ParticleSwarmOptimization pso = new ParticleSwarmOptimization();
@@ -1471,4 +1445,31 @@ public class OptimizerFactory {
public static final GOParameters tribes(AbstractOptimizationProblem problem) {
return makeParams(new Tribes(), 1, problem, randSeed, makeDefaultTerminator());
}
/**
* Check if a given problem instance has a template individual matching the given class.
*
* @param prob
* @param cls
* @return
*/
private static boolean assertIndyType(AbstractOptimizationProblem prob,
Class<InterfaceESIndividual> cls) {
return cls.isAssignableFrom(prob.getIndividualTemplate().getClass());
}
/**
* Set the evolutionary operators of the individual template of the given problem instance.
* @param prob
* @param mute
* @param pMut
* @param cross
* @param pCross
*/
public static void setTemplateOperators(AbstractOptimizationProblem prob, InterfaceMutation mute, double pMut, InterfaceCrossover cross, double pCross) {
AbstractEAIndividual indy = prob.getIndividualTemplate();
if ((indy != null)) {
indy.setOperators(mute, pMut, cross, pCross);
}
}
}

19
src/eva2/OptimizerRunnable.java
View File

@@ -4,9 +4,9 @@ import java.io.PrintWriter;
import java.io.StringWriter;
import java.util.BitSet;
import eva2.server.go.IndividualInterface;
import eva2.server.go.InterfaceGOParameters;
import eva2.server.go.InterfaceNotifyOnInformers;
import eva2.server.go.InterfaceTerminator;
import eva2.server.go.individuals.InterfaceDataTypeBinary;
import eva2.server.go.individuals.InterfaceDataTypeDouble;
@@ -14,12 +14,13 @@ import eva2.server.go.individuals.InterfaceDataTypeInteger;
import eva2.server.go.operators.postprocess.InterfacePostProcessParams;
import eva2.server.go.operators.postprocess.PostProcessParams;
import eva2.server.go.populations.Population;
import eva2.server.go.populations.SolutionSet;
import eva2.server.modules.GOParameters;
import eva2.server.modules.Processor;
import eva2.server.stat.AbstractStatistics;
import eva2.server.stat.InterfaceStatistics;
import eva2.server.stat.StatisticsDummy;
import eva2.server.stat.InterfaceTextListener;
import eva2.server.stat.StatisticsDummy;
import eva2.server.stat.StatisticsStandalone;
import eva2.server.stat.StatsParameter;
import eva2.tools.jproxy.RemoteStateListener;
@@ -87,6 +88,10 @@ public class OptimizerRunnable implements Runnable {
public OptimizerRunnable(GOParameters params, InterfaceStatistics stats, boolean restart) {
rnblID = cntID;
cntID++;
if (stats.getStatisticsParameter() instanceof InterfaceNotifyOnInformers) {
// addition for the statistics revision with selectable strings - make sure the go parameters are represented within the statistics
params.addInformableInstance((InterfaceNotifyOnInformers)(stats.getStatisticsParameter()));
}
proc = new Processor(stats, null, params);
if (proc.getStatistics() instanceof AbstractStatistics) ((AbstractStatistics)proc.getStatistics()).setSaveParams(false);
doRestart = restart;
@@ -121,6 +126,10 @@ public class OptimizerRunnable implements Runnable {
InterfaceGOParameters params = proc.getGOParams();
proc = new Processor(stats, null, params);
if (proc.getStatistics() instanceof AbstractStatistics) ((AbstractStatistics)proc.getStatistics()).setSaveParams(false);
if (stats.getStatisticsParameter() instanceof InterfaceNotifyOnInformers) {
// addition for the statistics revision with selectable strings - make sure the go parameters are represented within the statistics
params.addInformableInstance((InterfaceNotifyOnInformers)(stats.getStatisticsParameter()));
}
}
public void setTextListener(InterfaceTextListener lsnr) {
@@ -184,10 +193,14 @@ public class OptimizerRunnable implements Runnable {
return proc.getStatistics().getBestSolution();
}
public Population getSolutionSet() {
public Population getResultPopulation() {
return proc.getResultPopulation();
}
public SolutionSet getSolutionSet() {
return (SolutionSet)proc.getGOParams().getOptimizer().getAllSolutions();
}
public void setPostProcessingParams(InterfacePostProcessParams ppp) {
proc.getGOParams().setPostProcessParams(ppp);
}

11
src/eva2/server/go/operators/mutation/MutateESCovarianceMatrixAdaption.java
View File

@@ -65,7 +65,16 @@ public class MutateESCovarianceMatrixAdaption implements InterfaceMutation, java
if (mutator.m_Eigenvalues != null) this.m_Eigenvalues = (double[]) mutator.m_Eigenvalues.clone();
}
/** This method will enable you to clone a given mutation operator
/**
* Constructor allowing to set the range check option.
*
* @param doCheckConstraints if true, the range constraints are enforced on mutation.
*/
public MutateESCovarianceMatrixAdaption(boolean doCheckConstraints) {
setCheckConstraints(doCheckConstraints);
}
/** This method will enable you to clone a given mutation operator
* @return The clone
*/
public Object clone() {

2
src/eva2/server/go/problems/WaitForEvARunnable.java
View File

@@ -40,7 +40,7 @@ class WaitForEvARunnable implements Runnable {
mp.log("\n");
// write results back to matlab
mp.exportResultToMatlab(runnable);
mp.exportResultPopulationToMatlab(runnable.getSolutionSet());
mp.exportResultPopulationToMatlab(runnable.getResultPopulation());
mp.log("reported results.\n");
mp.notifyFinished();
mp.log("notified finish.\n");