com.bayesserver.statistics

Class KullbackLeibler

java.lang.Object

com.bayesserver.statistics.KullbackLeibler

public final class KullbackLeibler
extends Object

Calculate the Kullback–Leibler divergence between 2 distributions with the same variables, D(P||Q). A value of 0 indicates that the two distributions behave in a very similar or identical way.

Method Summary

Modifier and Type	Method and Description
static double	divergence(Distribution priorQ, Distribution posteriorP, LogarithmBase logarithm) Calculates the Kullback-Leibler divergence D(P||Q).

Methods inherited from class java.lang.Object

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

Method Detail

divergence

public static double divergence(Distribution priorQ,
                                Distribution posteriorP,
                                LogarithmBase logarithm)

Calculates the Kullback-Leibler divergence D(P||Q). Supports multivariate distributions. Supports discrete or continuous distributions.

Parameters:

priorQ - The distribution Q in D(P||Q), e.g. the Prior.

posteriorP - The distribution P in D(P||Q), e.g. the Posterior.

logarithm - The base of the computation.

Returns:

The Kullback-Leibler divergence in NATS if the logarithm is natural, BITS if the logarithm is base 2.