burlap.behavior.functionapproximation

Class FunctionGradient.SparseGradient

java.lang.Object

burlap.behavior.functionapproximation.FunctionGradient.SparseGradient

All Implemented Interfaces:

FunctionGradient

Enclosing interface:

FunctionGradient

public static class FunctionGradient.SparseGradient
extends java.lang.Object
implements FunctionGradient

A sparse FunctionGradient that only explicitly stores the partial derivative for parameters with non-zero partial derivatives. The partial derivative for parameters not explicitly stored will have a value of zero returned.

Nested Class Summary

Nested classes/interfaces inherited from interface burlap.behavior.functionapproximation.FunctionGradient

FunctionGradient.PartialDerivative, FunctionGradient.SparseGradient

Field Summary

Fields

Modifier and Type	Field and Description
protected java.util.Map<java.lang.Integer,java.lang.Double>	gradient A map from weight identifiers to their partial derivative

Constructor Summary

Constructors

Constructor and Description
SparseGradient() Initializes with the gradient unspecified for any weights.
SparseGradient(int capacity) Initializes with the gradient unspecified, but reserves space for the given capacity

Method Summary

Modifier and Type	Method and Description
java.util.Set<FunctionGradient.PartialDerivative>	getNonZeroPartialDerivatives() Returns all non-zero partial derivatives.
double	getPartialDerivative(int parameterId) Returns the partial derivative for the given weight
int	numNonZeroPDs() Returns the number of non-zero partial derivatives
void	put(int parameterId, double partialDerivative) Adds the partial derivative for a given weight

Methods inherited from class java.lang.Object

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

Field Detail

gradient

protected java.util.Map<java.lang.Integer,java.lang.Double> gradient

A map from weight identifiers to their partial derivative

Constructor Detail

SparseGradient

public SparseGradient()

Initializes with the gradient unspecified for any weights.

SparseGradient

public SparseGradient(int capacity)

Initializes with the gradient unspecified, but reserves space for the given capacity

Parameters:

capacity - how much space to reserve for storing the gradient; i.e., the number of weights over which the gradient will be defined

Method Detail

put

public void put(int parameterId,
                double partialDerivative)

Description copied from interface: FunctionGradient

Adds the partial derivative for a given weight

Specified by:

put in interface FunctionGradient

Parameters:

parameterId - the parameter identifier for which the partial derivative is to be stored.

partialDerivative - the partial derivative value for the parameter

getPartialDerivative

public double getPartialDerivative(int parameterId)

Description copied from interface: FunctionGradient

Returns the partial derivative for the given weight

Specified by:

getPartialDerivative in interface FunctionGradient

Parameters:

parameterId - the parameter for which the partial derivative will be returned

Returns:

the partial derivative for the given weight; 0 if it is not stored.

getNonZeroPartialDerivatives

public java.util.Set<FunctionGradient.PartialDerivative> getNonZeroPartialDerivatives()

Description copied from interface: FunctionGradient

Returns all non-zero partial derivatives. Result is stored in a Map.Entry where the key is an Integer of the parameter identifier and the value is a double specifying its partial derivative.

Specified by:

getNonZeroPartialDerivatives in interface FunctionGradient

Returns:

all non-zero partial derivatives.

numNonZeroPDs

public int numNonZeroPDs()

Description copied from interface: FunctionGradient

Returns the number of non-zero partial derivatives

Specified by:

numNonZeroPDs in interface FunctionGradient

Returns:

the number of non-zero partial derivatives