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

Create JSON object with optimization details.

Browse Source 
ToDo:
- Store JSON to file
- Fix sync issue with generation 1 duplicate
This commit is contained in:
Fabian Becker 2013-12-10 11:50:37 +01:00
parent b005bb7a96
commit 30381f4c04
1 changed files with 74 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

87
src/eva2/cli/Main.java
View File

@ -8,9 +8,11 @@ import eva2.optimization.modules.OptimizationParameters;
import eva2.optimization.operator.terminators.CombinedTerminator;
import eva2.optimization.operator.terminators.EvaluationTerminator;
import eva2.optimization.operator.terminators.FitnessValueTerminator;
import eva2.optimization.population.Population;
import eva2.optimization.problems.AbstractOptimizationProblem;
import eva2.optimization.strategies.DifferentialEvolution;
import eva2.optimization.strategies.InterfaceOptimizer;
import com.google.gson.*;
import org.apache.commons.cli.*;
import org.reflections.Reflections;
import java.lang.reflect.Modifier;
@ -48,6 +50,9 @@ public class Main implements OptimizationStateListener, InterfacePopulationChang
private long seed = System.currentTimeMillis();
private AbstractOptimizationProblem problem;
private InterfaceOptimizer optimizer;
private JsonObject jsonObject;
private JsonArray optimizationRuns;
private JsonArray generationsArray;
/**
* Creates a set of default options used in all optimizations.
@ -84,6 +89,8 @@ public class Main implements OptimizationStateListener, InterfacePopulationChang
return opt;
}
@Override
public void performedStop() {
//To change body of implemented methods use File | Settings | File Templates.
@ -101,7 +108,7 @@ public class Main implements OptimizationStateListener, InterfacePopulationChang
@Override
public void updateProgress(int percent, String msg) {
//To change body of implemented methods use File | Settings | File Templates.
printProgressBar(percent);
}
public static void printProgressBar(int percent) {
@ -172,6 +179,9 @@ public class Main implements OptimizationStateListener, InterfacePopulationChang
optimizerParams = new String[]{};
}
this.jsonObject = new JsonObject();
this.optimizationRuns = new JsonArray();
/**
* Default command line options only require help or optimizer.
* Later we build extended command line options depending on
@ -234,6 +244,16 @@ public class Main implements OptimizationStateListener, InterfacePopulationChang
System.exit(-1);
}
}
this.jsonObject.addProperty("population_size", this.populationSize);
this.jsonObject.addProperty("number_of_runs", this.numberOfRuns);
this.jsonObject.addProperty("seed", this.seed);
JsonObject problemObject = new JsonObject();
problemObject.addProperty("name", this.problem.getName());
problemObject.addProperty("dimension", 30);
this.jsonObject.add("problem", problemObject);
}
/**
@ -290,6 +310,9 @@ public class Main implements OptimizationStateListener, InterfacePopulationChang
break;
case "GeneticAlgorithm":
System.out.println("Genetic Algorithm");
break;
case "ParticleSwarmOptimization":
break;
default:
throw new Exception("Unsupported Optimizer");
@ -310,23 +333,39 @@ public class Main implements OptimizationStateListener, InterfacePopulationChang
}
private void runOptimization() {
for(int i = 0; i < this.numberOfRuns; i++) {
// Terminate after 10000 function evaluations OR after reaching a fitness < 0.1
OptimizerFactory.setEvaluationTerminator(50000);
OptimizerFactory.addTerminator(new FitnessValueTerminator(new double[]{0.01}), CombinedTerminator.OR);
this.generationsArray = new JsonArray();
// Terminate after 10000 function evaluations OR after reaching a fitness < 0.1
OptimizerFactory.setEvaluationTerminator(50000);
OptimizerFactory.addTerminator(new FitnessValueTerminator(new double[]{0.01}), CombinedTerminator.OR);
LOGGER.log(Level.INFO, "Running {0}", "Differential Evolution");
LOGGER.log(Level.INFO, "Running {0}", "Differential Evolution");
OptimizationParameters params = OptimizerFactory.makeParams(optimizer, this.populationSize, this.problem);
double[] result = OptimizerFactory.optimizeToDouble(params);
OptimizationParameters params = OptimizerFactory.makeParams(optimizer, this.populationSize, this.problem);
double[] result = OptimizerFactory.optimizeToDouble(params);
// This is stupid - why isn't there a way to wait for the optimization to finish?
while(OptimizerFactory.terminatedBecause().equals("Not yet terminated")) {
// wait
// This is stupid - why isn't there a way to wait for the optimization to finish?
while(OptimizerFactory.terminatedBecause().equals("Not yet terminated")) {
// wait
}
JsonObject optimizationDetails = new JsonObject();
optimizationDetails.addProperty("total_time", 1.0);
optimizationDetails.addProperty("total_function_calls", optimizer.getPopulation().getFunctionCalls());
optimizationDetails.addProperty("termination_criteria", OptimizerFactory.terminatedBecause());
optimizationDetails.add("generations", this.generationsArray);
this.optimizationRuns.add(optimizationDetails);
System.out.println(OptimizerFactory.terminatedBecause());
System.out.println(optimizer.getPopulation().getFunctionCalls());
}
System.out.println(OptimizerFactory.terminatedBecause());
System.out.println(optimizer.getPopulation().getFunctionCalls());
this.jsonObject.add("runs", this.optimizationRuns);
Gson gson = new GsonBuilder().setPrettyPrinting().create();
System.out.println(gson.toJson(this.jsonObject));
}
private static void showOptimizerHelp() {
@ -349,6 +388,28 @@ public class Main implements OptimizationStateListener, InterfacePopulationChang
@Override
public void registerPopulationStateChanged(Object source, String name) {
//System.out.println(name);
if (name.equals("NextGenerationPerformed")) {
InterfaceOptimizer optimizer = (InterfaceOptimizer)source;
Population population = optimizer.getPopulation();
JsonObject newGeneration = new JsonObject();
newGeneration.addProperty("generation", population.getGeneration());
newGeneration.addProperty("function_calls", population.getFunctionCalls());
JsonArray bestFitness = new JsonArray();
for(double val : population.getBestFitness()) {
bestFitness.add(new JsonPrimitive(val));
}
newGeneration.add("best_fitness", bestFitness);
JsonArray meanFitness = new JsonArray();
for(double val : population.getMeanFitness()) {
meanFitness.add(new JsonPrimitive(val));
}
newGeneration.add("mean_fitness", meanFitness);
System.out.println(newGeneration.toString());
//this.generationsArray.add(newGeneration);
}
}
}