de.jstacs.sequenceScores.statisticalModels.trainable.hmm.states

Interface DifferentiableState

All Superinterfaces:

State

All Known Implementing Classes:

SimpleDifferentiableState

public interface DifferentiableState
extends State

This interface declares a method that allows to evaluate the gradient which is essential for numerical optimization.

Author:

Jens Keilwagen

Method Summary

Methods

Modifier and Type	Method and Description
double	getLogScoreAndPartialDerivation(int startPos, int endPos, IntList indices, DoubleList partDer, Sequence seq) This method allows to compute the logarithm of the score and the gradient for the given subsequences.

Methods inherited from interface de.jstacs.sequenceScores.statisticalModels.trainable.hmm.states.State

getGraphvizNodeOptions, getLogScoreFor, isSilent, toString

Method Detail

getLogScoreAndPartialDerivation

double getLogScoreAndPartialDerivation(int startPos,
                                     int endPos,
                                     IntList indices,
                                     DoubleList partDer,
                                     Sequence seq)
                                       throws OperationNotSupportedException,
                                              WrongLengthException

This method allows to compute the logarithm of the score and the gradient for the given subsequences.

Parameters:

startPos - the start position (inclusive)

endPos - the end position (inclusive)

indices - a list for the parameter indices

partDer - a list for the partial derivations

seq - the Sequence(s)

Returns:

the logarithm of the score

Throws:

WrongLengthException - if the length can not be modeled

OperationNotSupportedException - if the reverse complement of the sequence can not be computed