io.improbable.keanu.vertices.dbl.nonprobabilistic

Class NonProbabilisticDouble

java.lang.Object

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

io.improbable.keanu.vertices.ContinuousVertex<DoubleTensor>

io.improbable.keanu.vertices.dbl.DoubleVertex

io.improbable.keanu.vertices.dbl.nonprobabilistic.NonProbabilisticDouble

All Implemented Interfaces:

io.improbable.keanu.kotlin.DoubleOperators<DoubleVertex>, io.improbable.keanu.kotlin.Operators<DoubleVertex>

Direct Known Subclasses:

CastDoubleVertex, ConstantDoubleVertex, DoubleBinaryOpLambda, DoubleBinaryOpVertex, DoubleIfVertex, DoubleReduceVertex, DoubleUnaryOpLambda, DoubleUnaryOpVertex

public abstract class NonProbabilisticDouble
extends DoubleVertex

Field Summary

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

ID_GENERATOR

Constructor Summary

Constructors

Constructor and Description
NonProbabilisticDouble()

Method Summary

Modifier and Type	Method and Description
java.util.Map<java.lang.Long,DoubleTensor>	dLogPdf(DoubleTensor value)
abstract DoubleTensor	getDerivedValue()
boolean	isProbabilistic()
double	logPdf(DoubleTensor value)
void	observe(DoubleTensor value) Observing non-probabilistic values of this type causes the probability of the graph to flatten to 0 in all places that doesn't exactly match the observation.
DoubleTensor	updateValue() This causes a non-probabilistic vertex to recalculate it's value based off it's parent's current values.

Methods inherited from class io.improbable.keanu.vertices.dbl.DoubleVertex

abs, acos, asin, atan, atan2, calculateDualNumber, ceil, cos, div, div, divideBy, divideBy, dLogPdf, dLogPdf, exp, floor, getDualNumber, lambda, log, logPdf, logPdf, minus, minus, multiply, multiply, observe, observe, plus, plus, pow, pow, setAndCascade, setAndCascade, setAndCascade, setAndCascade, setValue, setValue, sigmoid, sin, tan, times, times, unaryMinus

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

dLogProb, logProb

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

addChild, addParent, addParents, dLogProbAtValue, equals, exploreSetting, getChildren, getConnectedGraph, getId, getParents, getRawValue, getShape, getValue, hashCode, hasValue, isObserved, lazyEval, logProbAtValue, observeOwnValue, sample, sampleUsingDefaultRandom, setAndCascade, setAndCascade, setParents, setParents, setValue, unobserve

Methods inherited from class java.lang.Object

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

Constructor Detail

NonProbabilisticDouble

public NonProbabilisticDouble()

Method Detail

observe

public void observe(DoubleTensor value)

Observing non-probabilistic values of this type causes the probability of the graph to flatten to 0 in all places that doesn't exactly match the observation. This is so bad that it is actually prohibited by throwing an exception. This is not the case for all types of non-probabilistic observations.

Overrides:

observe in class Vertex<DoubleTensor>

Parameters:

value - the value to be observed

logPdf

public double logPdf(DoubleTensor value)

Specified by:

logPdf in class ContinuousVertex<DoubleTensor>

dLogPdf

public java.util.Map<java.lang.Long,DoubleTensor> dLogPdf(DoubleTensor value)

Specified by:

dLogPdf in class ContinuousVertex<DoubleTensor>

isProbabilistic

public boolean isProbabilistic()

Specified by:

isProbabilistic in class Vertex<DoubleTensor>

Returns:

True if the vertex is probabilistic, false otherwise. A probabilistic vertex is defined as a vertex whose value is not derived from it's parents. However, the probability of the vertex's value may be dependent on it's parents values.

updateValue

public DoubleTensor updateValue()

Description copied from class: Vertex

This causes a non-probabilistic vertex to recalculate it's value based off it's parent's current values.

Specified by:

updateValue in class Vertex<DoubleTensor>

Returns:

The updated value

getDerivedValue

public abstract DoubleTensor getDerivedValue()