Example usage for org.apache.commons.math3.optim MaxEval unlimited

Introduction

In this page you can find the example usage for org.apache.commons.math3.optim MaxEval unlimited.

Prototype

public static MaxEval unlimited() 

Source Link

Document

Factory method that creates instance of this class that represents a virtually unlimited number of evaluations.

Usage

From source file:edu.uchc.octane.GaussianFitAstigmatism.java

/**
 * calculate the Z coordinate based on astigmatism
 *///from   ww w  . ja v  a2 s. c  o m
void calculateZ() {

    if (calibration_ == null) {
        z_ = 0;
        return;
    }

    UnivariateFunction func = new UnivariateFunction() {
        @Override
        public double value(double z) {
            double sigmax = FastMath.sqrt(pvp_.getPoint()[3] / 2);
            double sigmay = FastMath.sqrt(pvp_.getPoint()[4] / 2);
            double vx = calibration_[0] + (z - calibration_[1]) * (z - calibration_[1]) * calibration_[2]
                    - sigmax;
            double vy = calibration_[3] + (z - calibration_[4]) * (z - calibration_[4]) * calibration_[5]
                    - sigmay;
            return vx * vx + vy * vy;
        }
    };

    BrentOptimizer optimizer = new BrentOptimizer(1e-4, 1e-4);

    UnivariatePointValuePair upvp = optimizer.optimize(new UnivariateObjectiveFunction(func), GoalType.MINIMIZE,
            MaxEval.unlimited(), new SearchInterval(2 * z0min_ - z0max_, 2 * z0max_ - z0min_));

    if (upvp.getValue() > errTol_) {
        throw (new ConvergenceException());
    }

    z_ = upvp.getPoint();
}

---

From source file:norbert.mynemo.core.selection.SvdRecommenderSelector.java

/**
 * Evaluates the given recommender type for several configurations. The number of iterations and
 * the number of features for each evaluation is chosen via an optimizer. The number of evaluation
 * performed depends of the number of steps necessary for the evaluation results to converge. The
 * number of evaluations performed is also bounded by an internal maximum.
 *
 * <p>/*from   ww  w .  ja va 2 s  .com*/
 * The given recommender type must be part of the SVD based family.
 *
 * @return all evaluations done during the selection process
 */
public Collection<RecommenderEvaluation> select(RecommenderType type, double minimumCoverage) {
    checkArgument(type.getFamily() == RecommenderFamily.SVD_BASED);

    // initialize necessary optimization data
    ConvergenceChecker<PointValuePair> checker = new MaxIterationChecker<PointValuePair>(CMAES_MAX_ITERATIONS);
    SvdRecommenderEvalFunction function = new SvdRecommenderEvalFunction(configuration, type, minimumCoverage);
    ObjectiveFunction objectiveFunction = new ObjectiveFunction(function);
    CMAESOptimizer optimizer = new CMAESOptimizer(CMAES_MAX_ITERATIONS, 1.0, true, 2, 0,
            new JDKRandomGenerator(), false, checker);

    MaxEval maxEval = MaxEval.unlimited();
    GoalType goalType = GoalType.MINIMIZE;
    SimpleBounds bounds = new SimpleBounds(new double[] { MIN_SVD_FEATURES, MIN_SVD_ITERATIONS },
            new double[] { MAX_SVD_FEATURES, MAX_SVD_ITERATIONS });
    InitialGuess initialGuess = new InitialGuess(
            new double[] { INITIAL_GUESS_FEATURES, INITIAL_GUESS_ITERATIONS });
    CMAESOptimizer.PopulationSize populationSize = new CMAESOptimizer.PopulationSize(16);
    CMAESOptimizer.Sigma sigma = new CMAESOptimizer.Sigma(
            new double[] { CMAES_SIGMA_FEATURE, CMAES_SIGMA_ITERATION });

    // run the optimizer
    optimizer.optimize(objectiveFunction, goalType, initialGuess, populationSize, sigma, bounds, maxEval);

    return function.getEvaluations();
}

---