gov.sandia.cognition.statistics.method

Class ImportanceSampling

java.lang.Object

gov.sandia.cognition.statistics.method.ImportanceSampling

@PublicationReference(author="Wikipedia",
                      title="Importance Sampling",
                      type=WebPage,
                      year=2009,
                      url="http://en.wikipedia.org/wiki/Importance_sampling")
public class ImportanceSampling
extends java.lang.Object

Importance sampling is a technique for estimating properties of a target distribution, while only having samples generated from an "importance" distribution rather than the target distribution. Typically, the importance distribution is easy to sample from, while the target distribution is difficult to sample from, and the importance distribution has support everywhere that the target distribution has support. Then, this results in an weighted set of samples that are an unbiased sampling of the target distribution.

Since:

3.0

Author:

Kevin R. Dixon

Constructor Summary

Constructors

Constructor and Description
ImportanceSampling()

Method Summary

Modifier and Type	Method and Description
static <ValueType> java.util.ArrayList<DefaultWeightedValue<ValueType>>	sample(ProbabilityDensityFunction<ValueType> importanceDistribution, Evaluator<ValueType,java.lang.Double> targetDistribution, java.util.Random random, int numSamples) Importance sampling is a technique for estimating properties of a target distribution, while only having samples generated from an "importance" distribution rather than the target distribution.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

ImportanceSampling

public ImportanceSampling()

Method Detail

sample

public static <ValueType> java.util.ArrayList<DefaultWeightedValue<ValueType>> sample(ProbabilityDensityFunction<ValueType> importanceDistribution,
                                                                                      Evaluator<ValueType,java.lang.Double> targetDistribution,
                                                                                      java.util.Random random,
                                                                                      int numSamples)

Importance sampling is a technique for estimating properties of a target distribution, while only having samples generated from an "importance" distribution rather than the target distribution. Typically, the importance distribution is easy to sample from, while the target distribution is difficult to sample from, and the importance distribution has support everywhere that the target distribution has support. Then, this results in an weighted set of samples that are an unbiased sampling of the target distribution.

Type Parameters:

ValueType - Domain type of the distributions.

Parameters:

importanceDistribution - Easy-to-sample-from distribution that will generate the samples.

targetDistribution - The hard-to-sample-from distribution that is desired.

random - Random number generator.

numSamples - Number of samples to create.

Returns:

Weighted samples that are an unbiased estimate of the target distribution.