source: src/main/java/agents/uk/ac/soton/ecs/gp4j/util/GaussianProcessUtils.java

package agents.uk.ac.soton.ecs.gp4j.util;

import agents.Jama.CholeskyDecomposition;
import agents.Jama.Matrix;
import agents.org.apache.commons.math.random.RandomData;
import agents.org.apache.commons.math.random.RandomDataImpl;
import agents.uk.ac.soton.ecs.gp4j.gp.covariancefunctions.CovarianceFunction;

public class GaussianProcessUtils {
        public static Matrix drawSample(CovarianceFunction function,
                        double[] logHyper, Matrix testX, RandomData data) {
                Matrix covarianceMatrix = function.calculateCovarianceMatrix(logHyper,
                                testX);

                return drawSample(covarianceMatrix, testX, data);
        }

        public static Matrix drawSample(CovarianceFunction function,
                        double[] logHyper, Matrix testX) {
                return drawSample(function, logHyper, testX, new RandomDataImpl());
        }

        public static Matrix drawSample(Matrix covarianceMatrix, Matrix testX,
                        RandomData data) {
                Matrix randomVector = new Matrix(testX.getRowDimension(), 1);

                for (int i = 0; i < randomVector.getRowDimension(); i++) {
                        randomVector.set(i, 0, data.nextGaussian(0, 1));
                }

                CholeskyDecomposition chol = covarianceMatrix.chol();

                return chol.getL().times(randomVector);
        }

        public static Matrix drawSample(Matrix covarianceMatrix, Matrix testX) {
                return drawSample(covarianceMatrix, testX, new RandomDataImpl());
        }
}