io.improbable.keanu.algorithms.mcmc

Class MCMC

java.lang.Object

io.improbable.keanu.algorithms.mcmc.MCMC

public class MCMC
extends java.lang.Object

Class for choosing the appropriate sampling algorithm given a network. If the given network is differentiable, NUTS is proposed, otherwise Metropolis Hastings is chosen. Usage: PosteriorSamplingAlgorithm samplingAlgorithm = MCMC.withDefaultConfig().forNetwork(yourNetwork); samplingAlgorithm.getPosteriorSamples(...);

Constructor Summary

Constructors

Constructor and Description
MCMC()

Method Summary

Modifier and Type	Method and Description
PosteriorSamplingAlgorithm	forNetwork(BayesianNetwork bayesianNetwork)
static MCMC	withDefaultConfig()
static MCMC	withDefaultConfig(KeanuRandom random)

Methods inherited from class java.lang.Object

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

Constructor Detail

MCMC

public MCMC()

Method Detail

withDefaultConfig

public static MCMC withDefaultConfig()

withDefaultConfig

public static MCMC withDefaultConfig(KeanuRandom random)

forNetwork

public PosteriorSamplingAlgorithm forNetwork(BayesianNetwork bayesianNetwork)

Parameters:

bayesianNetwork - network for which to choose sampling algorithm.

Returns:

recommended sampling algorithm for this network.