io.improbable.keanu.algorithms.mcmc.nuts

Class NUTS

java.lang.Object

io.improbable.keanu.algorithms.mcmc.nuts.NUTS

All Implemented Interfaces:

PosteriorSamplingAlgorithm

public class NUTS
extends java.lang.Object
implements PosteriorSamplingAlgorithm

NUTS with multinomial sampling

References:

Algorithm 6: "No-U-Turn Sampler with Dual Averaging". The No-U-Turn Sampler: Adaptively Setting Path Lengths in Hamiltonian Monte Carlo https://arxiv.org/pdf/1111.4246.pdf

A Conceptual Introduction to Hamiltonian Monte Carlo by Michael Betancourt https://arxiv.org/pdf/1701.02434.pdf

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	NUTS.Metrics
static class	NUTS.NUTSBuilder

Constructor Summary

Constructors

Constructor and Description
NUTS()

Method Summary

Modifier and Type	Method and Description
static NUTS.NUTSBuilder	builder()
NetworkSamplesGenerator	generatePosteriorSamples(ProbabilisticModel model, java.util.List<? extends Variable> fromVariables)
NetworkSamples	getPosteriorSamples(ProbabilisticModel model, java.util.List<? extends Variable> variablesToSampleFrom, int sampleCount) Sample from the posterior of a probabilistic model using the No-U-Turn-Sampling algorithm
Statistics	getStatistics()

Methods inherited from class java.lang.Object

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

Methods inherited from interface io.improbable.keanu.algorithms.PosteriorSamplingAlgorithm

getPosteriorSamples, getPosteriorSamples

Constructor Detail

NUTS

public NUTS()

Method Detail

builder

public static NUTS.NUTSBuilder builder()

getPosteriorSamples

public NetworkSamples getPosteriorSamples(ProbabilisticModel model,
                                          java.util.List<? extends Variable> variablesToSampleFrom,
                                          int sampleCount)

Sample from the posterior of a probabilistic model using the No-U-Turn-Sampling algorithm

Specified by:

getPosteriorSamples in interface PosteriorSamplingAlgorithm

Parameters:

model - the probabilistic model to sample from

variablesToSampleFrom - the variables inside the probabilistic model to sample from

sampleCount - the number of samples to take

Returns:

Samples taken with NUTS

generatePosteriorSamples

public NetworkSamplesGenerator generatePosteriorSamples(ProbabilisticModel model,
                                                        java.util.List<? extends Variable> fromVariables)

Specified by:

generatePosteriorSamples in interface PosteriorSamplingAlgorithm

getStatistics

public Statistics getStatistics()