cc.kave.repackaged.jayes.factor

Class SparseFactor

java.lang.Object

cc.kave.repackaged.jayes.factor.AbstractFactor

cc.kave.repackaged.jayes.factor.SparseFactor

All Implemented Interfaces:

Cloneable

public class SparseFactor
extends AbstractFactor

Field Summary

Fields inherited from class cc.kave.repackaged.jayes.factor.AbstractFactor

cut, dimensionIDs, dimensions, selections, values

Constructor Summary

Constructors

Constructor and Description
SparseFactor()

Method Summary

Modifier and Type	Method and Description
SparseFactor	clone()
void	copyValues(IArrayWrapper other)
void	fill(double d)
static SparseFactor	fromFactor(AbstractFactor f)
int	getOverhead()
protected int	getRealPosition(int virtualPosition)
static boolean	isSuitable(int futureLength, AbstractFactor... multiplicationCandidates) approximates whether creating a sparse factor will save memory compared to a dense factor
void	multiplyCompatible(AbstractFactor compatible) multiply the factors.
int[]	prepareMultiplication(AbstractFactor compatible)
void	setDimensions(int... dimensions)
void	sparsify(AbstractFactor... compatible) Prepares the factor in the sense that it's internal structures are optimized according to the zero/non-zero structure of the factors that will be multiplied in.

Methods inherited from class cc.kave.repackaged.jayes.factor.AbstractFactor

getDimensionFromID, getDimensionIDs, getDimensions, getValue, getValues, isLogScale, marginalizeAllBut, multiplyCompatibleToLog, multiplyPrepared, resetSelections, select, setDimensionIDs, setLogScale, setValues, sumPrepared, validateCut

Methods inherited from class java.lang.Object

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

Constructor Detail

SparseFactor

public SparseFactor()

Method Detail

setDimensions

public void setDimensions(int... dimensions)

Overrides:

setDimensions in class AbstractFactor

copyValues

public void copyValues(IArrayWrapper other)

Specified by:

copyValues in class AbstractFactor

multiplyCompatible

public void multiplyCompatible(AbstractFactor compatible)

Description copied from class: AbstractFactor

multiply the factors. Only compatible factors are allowed, meaning ones that have a subset of the variables of this factor (assume consistent Dimension ID / size pairs)

Overrides:

multiplyCompatible in class AbstractFactor

Parameters:

compatible - input factor

prepareMultiplication

public int[] prepareMultiplication(AbstractFactor compatible)

Specified by:

prepareMultiplication in class AbstractFactor

sparsify

public void sparsify(AbstractFactor... compatible)

Prepares the factor in the sense that it's internal structures are optimized according to the zero/non-zero structure of the factors that will be multiplied in. This method should always and only be called once before any call that modifies this factor's values

Parameters:

compatible - a Factor with compatible dimensions

getRealPosition

protected int getRealPosition(int virtualPosition)

Specified by:

getRealPosition in class AbstractFactor

fill

public void fill(double d)

Specified by:

fill in class AbstractFactor

getOverhead

public int getOverhead()

Specified by:

getOverhead in class AbstractFactor

Returns:

approximated memory requirements on top of the value array

clone

public SparseFactor clone()

Overrides:

clone in class AbstractFactor

isSuitable

public static boolean isSuitable(int futureLength,
                                 AbstractFactor... multiplicationCandidates)

approximates whether creating a sparse factor will save memory compared to a dense factor

Parameters:

futureLength - the length that the new factor's value array would have in a dense factor

multiplicationCandidates - array of candidates

Returns:

resulting assessment

fromFactor

public static SparseFactor fromFactor(AbstractFactor f)