edu.stanford.nlp.optimization

Interface SparseOnlineFunction<K>

public interface SparseOnlineFunction<K>

An interface for functions over sparse parameters. the point is to run this online, and data control is passed to the optimizer K is probably a String or an int selectedData are the data points used in the current evaluation, which is more naturally handled by the minimizers instead of the implementation though if one prefers, one can implement that elsewhere, and make valueAt, derivativeAt independent of selectedData

Since:

1.0

Version:

1.0

Author:

Sida Wang

Method Summary

Modifier and Type	Method and Description
int	dataSize()
Counter<K>	derivativeAt(Counter<K> x, int[] selectedData)
double	valueAt(Counter<K> x, int[] selectedData) Returns the value of the function at a single point.

Method Detail

valueAt

double valueAt(Counter<K> x,
               int[] selectedData)

Returns the value of the function at a single point.

Parameters:

x - a double[] input

Returns:

the function value at the input

derivativeAt

Counter<K> derivativeAt(Counter<K> x,
                        int[] selectedData)

dataSize

int dataSize()