io.improbable.keanu.algorithms.mcmc.nuts

Class NUTS.NUTSBuilder

java.lang.Object

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

Enclosing class:

NUTS

public static class NUTS.NUTSBuilder
extends java.lang.Object

Constructor Summary

Constructors

Constructor and Description
NUTSBuilder()

Method Summary

Modifier and Type	Method and Description
NUTS.NUTSBuilder	adaptCount(long adaptCount)
NUTS.NUTSBuilder	adaptPotentialEnabled(boolean adaptPotentialEnabled)
NUTS.NUTSBuilder	adaptStepSizeEnabled(boolean adaptEnabled)
NUTS	build()
NUTS.NUTSBuilder	initialStepSize(java.lang.Double initialStepSize)
NUTS.NUTSBuilder	maxEnergyChange(double maxEnergyChange)
NUTS.NUTSBuilder	maxTreeHeight(int maxTreeHeight)
NUTS.NUTSBuilder	potential(Potential potential)
NUTS.NUTSBuilder	random(KeanuRandom random)
NUTS.NUTSBuilder	saveStatistics(boolean saveStatistics)
NUTS.NUTSBuilder	targetAcceptanceProb(double targetAcceptanceProb)
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Constructor Detail

NUTSBuilder

public NUTSBuilder()

Method Detail

random

public NUTS.NUTSBuilder random(KeanuRandom random)

targetAcceptanceProb

public NUTS.NUTSBuilder targetAcceptanceProb(double targetAcceptanceProb)

Parameters:

targetAcceptanceProb - The targeted acceptance rate for the step size adaption. This defaults to 0.8.

Returns:

the builder for NUTS

adaptStepSizeEnabled

public NUTS.NUTSBuilder adaptStepSizeEnabled(boolean adaptEnabled)

Parameters:

adaptEnabled - Set to true if step size adaption is wanted.

Returns:

the builder for NUTS

adaptCount

public NUTS.NUTSBuilder adaptCount(long adaptCount)

Parameters:

adaptCount - The number of samples for which the step size and potential will be tuned. For the remaining samples in which it is not tuned, the step size will be frozen to its last calculated value.

Returns:

the builder for NUTS

initialStepSize

public NUTS.NUTSBuilder initialStepSize(java.lang.Double initialStepSize)

Parameters:

initialStepSize - The initial step size. If this is null then a step size will be calculated using heuristics.

Returns:

the builder for NUTS

adaptPotentialEnabled

public NUTS.NUTSBuilder adaptPotentialEnabled(boolean adaptPotentialEnabled)

Parameters:

adaptPotentialEnabled - Set to true if adapting of the mass matrix is wanted.

Returns:

the builder for NUTS

potential

public NUTS.NUTSBuilder potential(Potential potential)

Parameters:

potential - provides mass in velocity and energy calculations

Returns:

the builder for NUTS

maxEnergyChange

public NUTS.NUTSBuilder maxEnergyChange(double maxEnergyChange)

Parameters:

maxEnergyChange - The maximum energy change for a step to be considered divergent.

Returns:

the builder for NUTS

maxTreeHeight

public NUTS.NUTSBuilder maxTreeHeight(int maxTreeHeight)

Parameters:

maxTreeHeight - The maximum tree size for the sampler. This controls how long a sample walk can be before it terminates. This will set at a maximum approximately 2^treeSize number of logProb evaluations for a sample.

Returns:

the builder for NUTS

saveStatistics

public NUTS.NUTSBuilder saveStatistics(boolean saveStatistics)

Parameters:

saveStatistics - Set to true if sampling statistics are wanted. Sets whether or not to save debug STATISTICS. The STATISTICS available are: Step size, Log Prob, Mean Tree Acceptance Prob, Tree Size.

Returns:

the builder for NUTS

build

public NUTS build()

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object