Package opennlp.tools.ml.maxent.quasinewton

Class ParallelNegLogLikelihood

java.lang.Object

opennlp.tools.ml.maxent.quasinewton.NegLogLikelihood

opennlp.tools.ml.maxent.quasinewton.ParallelNegLogLikelihood

All Implemented Interfaces:

Function

public class ParallelNegLogLikelihood
extends NegLogLikelihood

Evaluates negative log-likelihood and its gradient in parallel.

See Also:

• Function

Constructor Summary

Constructors

Constructor	Description
ParallelNegLogLikelihood(DataIndexer indexer, int threads)

Method Summary

Modifier and Type	Method	Description
double[]	gradientAt(double[] x)	Computes the gradient for x.
double	valueAt(double[] x)	Computes the function value for x.

Methods inherited from class opennlp.tools.ml.maxent.quasinewton.NegLogLikelihood

getDimension, getInitialPoint

Methods inherited from class java.lang.Object

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

Constructor Details

ParallelNegLogLikelihood

public ParallelNegLogLikelihood(DataIndexer indexer,
 int threads)

Parameters:

indexer - The DataIndexer to use as input provider.

threads - The number of thread to compute with in parallel. Must be greater than 0.

Throws:

IllegalArgumentException - Thrown if parameters were invalid.

Method Details

valueAt

public double valueAt(double[] x)

Computes the function value for x.

Specified by:

valueAt in interface Function

Overrides:

valueAt in class NegLogLikelihood

Parameters:

x - The input.

Returns:

Returns the computed negative log-likelihood.

Throws:

IllegalArgumentException - Thrown if parameters were invalid.

gradientAt

public double[] gradientAt(double[] x)

Computes the gradient for x.

Specified by:

gradientAt in interface Function

Overrides:

gradientAt in class NegLogLikelihood

Parameters:

x - The input.

Returns:

Returns the computed gradient.

Throws:

IllegalArgumentException - Thrown if parameters were invalid.