de.up.ling.irtg.sampling

Class ProposalSumComputer

java.lang.Object

de.up.ling.irtg.sampling.ProposalSumComputer

public class ProposalSumComputer
extends Object

This class is used to compute the cumulative probability of proposing a given tree in case of an ambiguous automaton. If there are more than one ways of generating a tree from the underlying automaton, the we need to compute the sum of their probabilities. If the computer is used again after changing the underlying rule weights, then it must be reset, because it stores inside values that have already been computed.

Author:

christoph

Constructor Summary

Constructors

Constructor and Description
ProposalSumComputer()

Method Summary

Modifier and Type	Method and Description
double	computeInside(Tree<Rule> input, int state, RuleWeighting weights) Computes the proposal probability for the given tree assuming the given rule weights and assuming that we start at the given state.
double	computeInside(Tree<Rule> input, RuleWeighting weights) Computes the proposal probability for the given tree assuming the given rule weights.
void	fillInsides(TreeSample<Rule> sample, RuleWeighting rw) Computes the proposal probability for each tree in the sample assuming the given rule weightings and sets it as the log sum weight of the sample.
void	reset() This resets the cache of the inside values.

Methods inherited from class java.lang.Object

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

Constructor Detail

ProposalSumComputer

public ProposalSumComputer()

Method Detail

fillInsides

public void fillInsides(TreeSample<Rule> sample,
                        RuleWeighting rw)

Computes the proposal probability for each tree in the sample assuming the given rule weightings and sets it as the log sum weight of the sample.

Parameters:

sample -

rw -

computeInside

public double computeInside(Tree<Rule> input,
                            RuleWeighting weights)

Computes the proposal probability for the given tree assuming the given rule weights.

Parameters:

input -

weights -

Returns:

computeInside

public double computeInside(Tree<Rule> input,
                            int state,
                            RuleWeighting weights)

Computes the proposal probability for the given tree assuming the given rule weights and assuming that we start at the given state.

Parameters:

input -

state -

weights -

Returns:

reset

public void reset()

This resets the cache of the inside values.