use of dr.math.distributions.GaussianProcessRandomGenerator in project beast-mcmc by beast-dev.
the class EllipticalSliceOperatorParser method parseXMLObject.
public Object parseXMLObject(XMLObject xo) throws XMLParseException {
final double weight = xo.getDoubleAttribute(MCMCOperator.WEIGHT);
final Parameter variable = (Parameter) xo.getChild(Parameter.class);
boolean drawByRowTemp = false;
if (xo.hasAttribute(DRAW_BY_ROW)) {
drawByRowTemp = xo.getBooleanAttribute(DRAW_BY_ROW);
}
final boolean drawByRow = drawByRowTemp;
boolean signal = xo.getAttribute(SIGNAL_CONSTITUENT_PARAMETERS, true);
if (!signal) {
Parameter possiblyCompound = variable;
if (variable instanceof MaskedParameter) {
possiblyCompound = ((MaskedParameter) variable).getUnmaskedParameter();
}
if (!(possiblyCompound instanceof CompoundParameter)) {
signal = true;
}
}
double bracketAngle = xo.getAttribute(BRACKET_ANGLE, 0.0);
boolean translationInvariant = xo.getAttribute(TRANSLATION_INVARIANT, false);
boolean rotationInvariant = xo.getAttribute(ROTATION_INVARIANT, false);
GaussianProcessRandomGenerator gaussianProcess = (GaussianProcessRandomGenerator) xo.getChild(GaussianProcessRandomGenerator.class);
if (gaussianProcess == null) {
final MultivariateDistributionLikelihood likelihood = (MultivariateDistributionLikelihood) xo.getChild(MultivariateDistributionLikelihood.class);
if (!(likelihood.getDistribution() instanceof GaussianProcessRandomGenerator)) {
throw new XMLParseException("Elliptical slice sampling only works for multivariate normally distributed random variables");
}
if (likelihood.getDistribution() instanceof MultivariateNormalDistribution)
gaussianProcess = (MultivariateNormalDistribution) likelihood.getDistribution();
if (likelihood.getDistribution() instanceof MultivariateNormalDistributionModel)
gaussianProcess = (MultivariateNormalDistributionModel) likelihood.getDistribution();
}
EllipticalSliceOperator operator = new EllipticalSliceOperator(variable, gaussianProcess, drawByRow, signal, bracketAngle, translationInvariant, rotationInvariant);
operator.setWeight(weight);
return operator;
}
use of dr.math.distributions.GaussianProcessRandomGenerator in project beast-mcmc by beast-dev.
the class CompoundGaussianProcessParser method parseXMLObject.
public Object parseXMLObject(XMLObject xo) throws XMLParseException {
List<GaussianProcessRandomGenerator> gpList = new ArrayList<GaussianProcessRandomGenerator>();
List<Likelihood> likelihoodList = new ArrayList<Likelihood>();
List<Integer> copyList = new ArrayList<Integer>();
for (int i = 0; i < xo.getChildCount(); ++i) {
Object obj = xo.getChild(i);
GaussianProcessRandomGenerator gp = null;
Likelihood likelihood = null;
int copies = -1;
if (obj instanceof DistributionLikelihood) {
DistributionLikelihood dl = (DistributionLikelihood) obj;
if (!(dl.getDistribution() instanceof GaussianProcessRandomGenerator)) {
throw new XMLParseException("Not a Gaussian process");
}
likelihood = dl;
gp = (GaussianProcessRandomGenerator) dl.getDistribution();
copies = 0;
for (Attribute<double[]> datum : dl.getDataList()) {
// Double draw = (Double) gp.nextRandom();
// System.err.println("DL: " + datum.getAttributeName() + " " + datum.getAttributeValue().length + " " + "1");
copies += datum.getAttributeValue().length;
}
} else if (obj instanceof MultivariateDistributionLikelihood) {
MultivariateDistributionLikelihood mdl = (MultivariateDistributionLikelihood) obj;
if (!(mdl.getDistribution() instanceof GaussianProcessRandomGenerator)) {
throw new XMLParseException("Not a Gaussian process");
}
likelihood = mdl;
gp = (GaussianProcessRandomGenerator) mdl.getDistribution();
copies = 0;
double[] draw = (double[]) gp.nextRandom();
for (Attribute<double[]> datum : mdl.getDataList()) {
// System.err.println("ML: " + datum.getAttributeName() + " " + datum.getAttributeValue().length + " " + draw.length);
copies += datum.getAttributeValue().length / draw.length;
}
} else if (obj instanceof GaussianProcessRandomGenerator) {
gp = (GaussianProcessRandomGenerator) obj;
likelihood = gp.getLikelihood();
copies = 1;
} else {
throw new XMLParseException("Not a Gaussian process");
}
gpList.add(gp);
likelihoodList.add(likelihood);
copyList.add(copies);
}
// System.exit(-1);
return new CompoundGaussianProcess(gpList, likelihoodList, copyList);
}
Aggregations