de.up.ling.irtg.sampling

Class Proposal

java.lang.Object

de.up.ling.irtg.sampling.Proposal

public class Proposal
extends Object

Generates a sample of trees according the proposal probabilities of a given rule weighting. It also keeps track of the log of the proposal probabilities for the the generated trees.

Author:

christoph

Constructor Summary

Constructors

Constructor and Description
Proposal() Create a new instance with a WELL RNG with the current time as the seed.
Proposal(long seed) Create a new instance with a WELL RNG with the given seed.
Proposal(org.apache.commons.math3.random.RandomGenerator rg) Create a new instance with the given random number sequence.

Method Summary

Modifier and Type	Method and Description
TreeSample<Integer>	getRawTreeSample(RuleWeighting guide, int sampleSize) Get a sample of trees with label IDs for the rules sampled.
TreeSample<String>	getStringTreeSample(RuleWeighting guide, int sampleSize) Returns a sample of string trees by transforming sampled rule trees.
<Type> TreeSample<Type>	getTreeSample(BiFunction<Rule,TreeAutomaton,Type> mapping, RuleWeighting guide, int numberOfSamples) This method samples a number of rule trees and transforms them into the desired type of tree by applying the mapping to the tree and the automaton the guide returns.
TreeSample<Rule>	getTreeSample(RuleWeighting guide, int sampleSize) Returns a sample of rule trees.

Methods inherited from class java.lang.Object

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

Constructor Detail

Proposal

public Proposal(org.apache.commons.math3.random.RandomGenerator rg)

Create a new instance with the given random number sequence.

Parameters:

rg -

Proposal

public Proposal(long seed)

Create a new instance with a WELL RNG with the given seed.

Parameters:

seed -

Proposal

public Proposal()

Create a new instance with a WELL RNG with the current time as the seed.

Method Detail

getRawTreeSample

public TreeSample<Integer> getRawTreeSample(RuleWeighting guide,
                                            int sampleSize)

Get a sample of trees with label IDs for the rules sampled.

Parameters:

guide -

sampleSize -

Returns:

getTreeSample

public TreeSample<Rule> getTreeSample(RuleWeighting guide,
                                      int sampleSize)

Returns a sample of rule trees.

Parameters:

guide -

sampleSize -

Returns:

getStringTreeSample

public TreeSample<String> getStringTreeSample(RuleWeighting guide,
                                              int sampleSize)

Returns a sample of string trees by transforming sampled rule trees.

Parameters:

guide -

sampleSize -

Returns:

getTreeSample

public <Type> TreeSample<Type> getTreeSample(BiFunction<Rule,TreeAutomaton,Type> mapping,
                                             RuleWeighting guide,
                                             int numberOfSamples)

This method samples a number of rule trees and transforms them into the desired type of tree by applying the mapping to the tree and the automaton the guide returns. After calling this methods return.getLogPropWeight will return the log proposal probability for the rules from which the tree is derived.

Type Parameters:

Type - The label type of the trees.

Parameters:

mapping -

guide -

numberOfSamples -

Returns: