edu::tum::cs::bayesnets::inference::BackwardSamplingWithChildren::BackSamplingDistribution Class Reference

Inheritance diagram for edu::tum::cs::bayesnets::inference::BackwardSamplingWithChildren::BackSamplingDistribution:

[legend]

List of all members.

Public Member Functions
	BackSamplingDistribution (BackwardSamplingWithPriors sampler)
Protected Member Functions
void	construct (int i, int[] addr, CPF cpf, int[] nodeDomainIndices)
void	getProb (CPF cpf, int i, int[] addr, int[] nodeDomainIndices, MutableDouble ret)
double	getProb (CPF cpf, int[] nodeDomainIndices)

---

Detailed Description

Definition at line 29 of file BackwardSamplingWithChildren.java.

---

Constructor & Destructor Documentation

edu::tum::cs::bayesnets::inference::BackwardSamplingWithChildren::BackSamplingDistribution::BackSamplingDistribution

(

BackwardSamplingWithPriors

sampler

)

[inline]

Reimplemented from edu::tum::cs::bayesnets::inference::BackwardSamplingWithPriors::BackSamplingDistribution.

Definition at line 31 of file BackwardSamplingWithChildren.java.

---

Member Function Documentation

void edu::tum::cs::bayesnets::inference::BackwardSamplingWithChildren::BackSamplingDistribution::construct	(	int	i,
		int[]	addr,
		CPF	cpf,
		int[]	nodeDomainIndices
	)			[inline, protected]

recursively gets a distribution to backward sample from

Parameters:

	i	the node to instantiate next (as an index into the CPF's domain product)
	addr	the current setting of node indices of the CPF's domain product
	cpf	the conditional probability function of the node we are backward sampling

Reimplemented from edu::tum::cs::bayesnets::inference::BackwardSamplingWithPriors::BackSamplingDistribution.

Definition at line 42 of file BackwardSamplingWithChildren.java.

void edu::tum::cs::bayesnets::inference::BackwardSamplingWithChildren::BackSamplingDistribution::getProb	(	CPF	cpf,
		int	i,
		int[]	addr,
		int[]	nodeDomainIndices,
		MutableDouble	ret
	)			[inline, protected]

gets the probability indicated by the given CPF for the given domain indices, summing over all parents whose values are not set (i.e. set to -1) in nodeDomainIndices; i.e. computes the probability of the node whose CPF is provided given the evidence set in nodeDomainIndices

Parameters:

	cpf	the conditional probability function
	i	index of the next node to instantiate
	addr	the address (list of node domain indices relevant to the CPF)
	nodeDomainIndices	evidences (mapping of all nodes in the network to domain indices, -1 for no evidence)
	ret	variable in which to store the result (initialize to 0.0, because we are summing probability values)

Definition at line 149 of file BackwardSamplingWithChildren.java.

double edu::tum::cs::bayesnets::inference::BackwardSamplingWithChildren::BackSamplingDistribution::getProb	(	CPF	cpf,
		int[]	nodeDomainIndices
	)			[inline, protected]

Definition at line 101 of file BackwardSamplingWithChildren.java.

---

The documentation for this class was generated from the following file:

• BackwardSamplingWithChildren.java