io.improbable.keanu.vertices.generic.probabilistic.discrete

Class CategoricalVertex<CATEGORY>

java.lang.Object

io.improbable.keanu.vertices.Vertex<T>

io.improbable.keanu.vertices.generic.GenericVertex<GenericTensor<T>>

io.improbable.keanu.vertices.generic.GenericTensorVertex<CATEGORY>

io.improbable.keanu.vertices.generic.probabilistic.discrete.CategoricalVertex<CATEGORY>

All Implemented Interfaces:

Variable<GenericTensor<CATEGORY>,VertexState<GenericTensor<CATEGORY>>>, NonSaveableVertex, Observable<GenericTensor<CATEGORY>>, Probabilistic<GenericTensor<CATEGORY>>, Samplable<GenericTensor<CATEGORY>>

public class CategoricalVertex<CATEGORY>
extends GenericTensorVertex<CATEGORY>
implements Probabilistic<GenericTensor<CATEGORY>>, NonSaveableVertex

Constructor Summary

Constructors

Constructor and Description
CategoricalVertex(long[] tensorShape, java.util.Map<CATEGORY,DoubleVertex> selectableValues)
CategoricalVertex(java.util.Map<CATEGORY,DoubleVertex> selectableValues)

Method Summary

Modifier and Type	Method and Description
java.util.Map<Vertex,DoubleTensor>	dLogProb(GenericTensor<CATEGORY> value, java.util.Set<? extends Vertex> withRespectTo) The partial derivatives of the natural log prob.
java.util.Map<CATEGORY,DoubleVertex>	getSelectableValues()
double	logProb(GenericTensor<CATEGORY> value) This is the natural log of the probability at the supplied value.
static CategoricalVertex<java.lang.Integer>	of(DirichletVertex vertex)
static <CATEGORY> CategoricalVertex<CATEGORY>	of(DirichletVertex vertex, java.util.List<CATEGORY> categories)
static <CATEGORY> CategoricalVertex<CATEGORY>	of(java.util.Map<CATEGORY,java.lang.Double> selectableValues)
GenericTensor<CATEGORY>	sample(KeanuRandom random)

Methods inherited from class io.improbable.keanu.vertices.generic.GenericTensorVertex

slice, take

Methods inherited from class io.improbable.keanu.vertices.Vertex

addChild, addParent, addParents, equals, eval, getChildren, getConnectedGraph, getDegree, getId, getIndentation, getLabel, getObservedValue, getParents, getRank, getReference, getShape, getState, getValue, hashCode, hasValue, isDifferentiable, isObserved, isProbabilistic, lazyEval, loadValue, observe, observeOwnValue, print, print, removeLabel, save, saveValue, setAndCascade, setLabel, setLabel, setParents, setParents, setState, setValue, toString, unobserve

Methods inherited from class java.lang.Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Methods inherited from interface io.improbable.keanu.vertices.Probabilistic

dLogProb, dLogProbAtValue, dLogProbAtValue, getValue, keepOnlyProbabilisticVertices, logProbAtValue

Methods inherited from interface io.improbable.keanu.vertices.Observable

getObservedValue, isObserved, observe, unobserve

Methods inherited from interface io.improbable.keanu.vertices.Samplable

sample

Constructor Detail

CategoricalVertex

public CategoricalVertex(long[] tensorShape,
                         java.util.Map<CATEGORY,DoubleVertex> selectableValues)

CategoricalVertex

public CategoricalVertex(java.util.Map<CATEGORY,DoubleVertex> selectableValues)

Method Detail

of

public static <CATEGORY> CategoricalVertex<CATEGORY> of(java.util.Map<CATEGORY,java.lang.Double> selectableValues)

of

public static <CATEGORY> CategoricalVertex<CATEGORY> of(DirichletVertex vertex,
                                                        java.util.List<CATEGORY> categories)

of

public static CategoricalVertex<java.lang.Integer> of(DirichletVertex vertex)

getSelectableValues

public java.util.Map<CATEGORY,DoubleVertex> getSelectableValues()

sample

public GenericTensor<CATEGORY> sample(KeanuRandom random)

Specified by:

sample in interface Samplable<GenericTensor<CATEGORY>>

Parameters:

random - source of randomness

Returns:

a sample from the vertex's distribution. For non-probabilistic vertices, this will always be the same value.

logProb

public double logProb(GenericTensor<CATEGORY> value)

Description copied from interface: Probabilistic

This is the natural log of the probability at the supplied value. In the case of continuous vertices, this is actually the log of the density, which will differ from the probability by a constant.

Specified by:

logProb in interface Probabilistic<GenericTensor<CATEGORY>>

Parameters:

value - The supplied value.

Returns:

The natural log of the probability function at the supplied value. For continuous variables this is called the PDF (probability density function). For discrete variables this is called the PMF (probability mass function).

dLogProb

public java.util.Map<Vertex,DoubleTensor> dLogProb(GenericTensor<CATEGORY> value,
                                                   java.util.Set<? extends Vertex> withRespectTo)

Description copied from interface: Probabilistic

The partial derivatives of the natural log prob.

Specified by:

dLogProb in interface Probabilistic<GenericTensor<CATEGORY>>

Parameters:

value - at a given value

withRespectTo - list of parents to differentiate with respect to

Returns:

the partial derivatives of the log of the probability function at the supplied value. For continuous variables this is called the PDF (probability density function). For discrete variables this is called the PMF (probability mass function).