CMAExample1.java

package fr.inria.optimization.cmaes.examples;
import fr.inria.optimization.cmaes.CMAEvolutionStrategy;
import fr.inria.optimization.cmaes.fitness.IObjectiveFunction;

/** The very well-known Rosenbrock objective function to be minimized. 
 */
class Rosenbrock implements IObjectiveFunction { // meaning implements methods valueOf and isFeasible
	public double valueOf (double[] x) {
		double res = 0;
		for (int i = 0; i < x.length-1; ++i)
			res += 100 * (x[i]*x[i] - x[i+1]) * (x[i]*x[i] - x[i+1]) + 
			(x[i] - 1.) * (x[i] - 1.);
		return res;
	}
	public boolean isFeasible(double[] x) {return true; } // entire R^n is feasible
}

/** A very short example program how to use the class CMAEvolutionStrategy.  The code is given below, see also the code snippet in the documentation of class {@link CMAEvolutionStrategy}.  
 *  For implementation of restarts see {@link CMAExample2}.
<pre>
public class CMAExample1 {
	public static void main(String[] args) {
		IObjectiveFunction fitfun = new Rosenbrock();

		// new a CMA-ES and set some initial values
		CMAEvolutionStrategy cma = new CMAEvolutionStrategy();
		cma.readProperties(); // read options, see file CMAEvolutionStrategy.properties
		cma.setDimension(22); // overwrite some loaded properties
		cma.setInitialX(0.5); // in each dimension, also setTypicalX can be used
		cma.setInitialStandardDeviation(0.2); // also a mandatory setting 
		cma.options.stopFitness = 1e-9;       // optional setting

		// initialize cma and get fitness array to fill in later
		double[] fitness = cma.init();  // new double[cma.parameters.getPopulationSize()];

		// initial output to files
		cma.writeToDefaultFilesHeaders(0); // 0 == overwrites old files

		// iteration loop
		while(cma.stopConditions.getNumber() == 0) {

			// core iteration step 
			double[][] pop = cma.samplePopulation(); // get a new population of solutions
			for(int i = 0; i < pop.length; ++i) {    // for each candidate solution i
				while (!fitfun.isFeasible(pop[i]))   //    test whether solution is feasible,  
					pop[i] = cma.resampleSingle(i);  //       re-sample solution until it is feasible  
				fitness[i] = fitfun.valueOf(pop[i]); //    compute fitness value, where fitfun
			}	                                     //    is the function to be minimized
			cma.updateDistribution(fitness);         // pass fitness array to update search distribution

			// output to console and files
			cma.writeToDefaultFiles();
			int outmod = 150;
			if (cma.getCountIter() % (15*outmod) == 1)
				cma.printlnAnnotation(); // might write file as well
			if (cma.getCountIter() % outmod == 1)
				cma.println(); 
		}
		// evaluate mean value as it is the best estimator for the optimum
		cma.setFitnessOfMeanX(fitfun.valueOf(cma.getMeanX())); // updates the best ever solution 

		// final output
		cma.writeToDefaultFiles(1);
		cma.println();
		cma.println("Terminated due to");
		for (String s : cma.stopConditions.getMessages())
			cma.println("  " + s);
		cma.println("best function value " + cma.getBestFunctionValue() 
				+ " at evaluation " + cma.getBestEvaluationNumber());
			
		// we might return cma.getBestSolution() or cma.getBestX()

	} // main  
} // class
</pre>

 * 
 * @see CMAEvolutionStrategy
 * 
 * @author Nikolaus Hansen, released into public domain. 
 */
public class CMAExample1 {
	public static void main(String[] args) {
		IObjectiveFunction fitfun = new Rosenbrock();

		// new a CMA-ES and set some initial values
		CMAEvolutionStrategy cma = new CMAEvolutionStrategy();
		cma.readProperties(); // read options, see file CMAEvolutionStrategy.properties
		cma.setDimension(10); // overwrite some loaded properties
		cma.setInitialX(0.05); // in each dimension, also setTypicalX can be used
		cma.setInitialStandardDeviation(0.2); // also a mandatory setting 
		cma.options.stopFitness = 1e-14;       // optional setting

		// initialize cma and get fitness array to fill in later
		double[] fitness = cma.init();  // new double[cma.parameters.getPopulationSize()];

		// initial output to files
		cma.writeToDefaultFilesHeaders(0); // 0 == overwrites old files

		// iteration loop
		while(cma.stopConditions.getNumber() == 0) {

            // --- core iteration step ---
			double[][] pop = cma.samplePopulation(); // get a new population of solutions
			for(int i = 0; i < pop.length; ++i) {    // for each candidate solution i
            	// a simple way to handle constraints that define a convex feasible domain  
            	// (like box constraints, i.e. variable boundaries) via "blind re-sampling" 
            	                                       // assumes that the feasible domain is convex, the optimum is  
				while (!fitfun.isFeasible(pop[i]))     //   not located on (or very close to) the domain boundary,  
					pop[i] = cma.resampleSingle(i);    //   initialX is feasible and initialStandardDeviations are  
                                                       //   sufficiently small to prevent quasi-infinite looping here
                // compute fitness/objective value	
				fitness[i] = fitfun.valueOf(pop[i]); // fitfun.valueOf() is to be minimized
			}
			cma.updateDistribution(fitness);         // pass fitness array to update search distribution
            // --- end core iteration step ---

			// output to files and console 
			cma.writeToDefaultFiles();
			int outmod = 150;
			if (cma.getCountIter() % (15*outmod) == 1)
				cma.printlnAnnotation(); // might write file as well
			if (cma.getCountIter() % outmod == 1)
				cma.println(); 
		}
		// evaluate mean value as it is the best estimator for the optimum
		cma.setFitnessOfMeanX(fitfun.valueOf(cma.getMeanX())); // updates the best ever solution 

		// final output
		cma.writeToDefaultFiles(1);
		cma.println();
		cma.println("Terminated due to");
		for (String s : cma.stopConditions.getMessages())
			cma.println("  " + s);
		cma.println("best function value " + cma.getBestFunctionValue() 
				+ " at evaluation " + cma.getBestEvaluationNumber());
			
		// we might return cma.getBestSolution() or cma.getBestX()

	} // main  
} // class