Example usage for org.apache.commons.math3.fitting.leastsquares LevenbergMarquardtOptimizer optimize

Introduction

In this page you can find the example usage for org.apache.commons.math3.fitting.leastsquares LevenbergMarquardtOptimizer optimize.

Prototype

public Optimum optimize(final LeastSquaresProblem problem) 

Source Link

Usage

From source file:eu.crisis_economics.abm.markets.clearing.heterogeneous.LevenbergMarquardtClearingAlgorithm.java

@Override
public double applyToNetwork(final MixedClearingNetwork network) {
    Preconditions.checkNotNull(network);

    final VectorCostFunction function = super.getVectorCostFunction(network);
    final MultivariateJacobianFunction model = LeastSquaresFactory.model(function,
            super.getJacobianMatrixFunction(network));
    final RealVector observed = new ArrayRealVector(super.calculateTarget(network)),
            start = new ArrayRealVector(network.getNumberOfEdges());
    for (int i = 0; i < network.getNumberOfEdges(); ++i)
        start.setEntry(i, network.getEdges().get(i).getMaximumRateAdmissibleByBothParties());
    start.set(1.0);//from  w  w w  .j av a 2  s  .co  m

    final ConvergenceChecker<LeastSquaresProblem.Evaluation> evaluationChecker = LeastSquaresFactory
            .evaluationChecker(new SimpleVectorValueChecker(relErrorTarget, absErrorTarget));
    final LeastSquaresProblem problem = LeastSquaresFactory.create(model, observed, start, evaluationChecker,
            maximumEvaluations, maximumIterations);

    final LevenbergMarquardtOptimizer optimizer = new LevenbergMarquardtOptimizer();
    final Optimum result = optimizer.optimize(problem);

    final double residualCost = result.getRMS();
    System.out.println("Network cleared: residual cost: " + residualCost + ".");

    return residualCost;
}

---

From source file:de.bund.bfr.math.LeastSquaresOptimization.java

@Override
public Result optimize(int nParameterSpace, int nOptimizations, boolean stopWhenSuccessful,
        Map<String, Double> minStartValues, Map<String, Double> maxStartValues, int maxIterations,
        DoubleConsumer progressListener, ExecutionContext exec) throws CanceledExecutionException {
    if (exec != null) {
        exec.checkCanceled();//from  www. j  a v a2 s.  c o  m
    }

    progressListener.accept(0.0);

    List<ParamRange> ranges = MathUtils.getParamRanges(parameters, minStartValues, maxStartValues,
            nParameterSpace);
    RealVector targetVector = new ArrayRealVector(Doubles.toArray(targetValues));
    List<StartValues> startValuesList = MathUtils.createStartValuesList(ranges, nOptimizations,
            values -> targetVector
                    .getDistance(new ArrayRealVector(optimizerFunction.value(Doubles.toArray(values)))),
            progress -> progressListener.accept(0.5 * progress), exec);
    LevenbergMarquardtOptimizer optimizer = new LevenbergMarquardtOptimizer();
    Result result = new Result();
    AtomicInteger count = new AtomicInteger(0);

    for (StartValues startValues : startValuesList) {
        if (exec != null) {
            exec.checkCanceled();
        }

        progressListener.accept(0.5 * count.get() / startValuesList.size() + 0.5);

        try {
            LeastSquaresBuilder builder = createLeastSquaresBuilder(startValues.getValues(), maxIterations);

            builder.checker((iteration, previous, current) -> {
                double currentProgress = (double) iteration / (double) maxIterations;

                if (exec != null) {
                    try {
                        exec.checkCanceled();
                    } catch (CanceledExecutionException e) {
                        return true;
                    }
                }

                progressListener.accept(0.5 * (count.get() + currentProgress) / startValuesList.size() + 0.5);
                return iteration == maxIterations;
            });

            LeastSquaresOptimizer.Optimum optimizerResults = optimizer.optimize(builder.build());

            if (exec != null) {
                exec.checkCanceled();
            }

            double cost = optimizerResults.getCost();

            if (result.sse == null || cost * cost < result.sse) {
                result = getResults(optimizerResults);

                if (result.sse == 0.0) {
                    break;
                }

                if (result.r2 != null && result.r2 > 0.0 && stopWhenSuccessful) {
                    break;
                }
            }
        } catch (TooManyEvaluationsException | TooManyIterationsException | ConvergenceException e) {
        }

        count.incrementAndGet();
    }

    return result;
}

---