edu.jhu.bme.smile.commons.optimize
Class FunctionNumericNDDifferentiation

java.lang.Object
  edu.jhu.bme.smile.commons.optimize.FunctionNumericNDDifferentiation

All Implemented Interfaces:

OptimizableNDContinuous, OptimizableNDContinuousDifferentiable

public class FunctionNumericNDDifferentiation
extends java.lang.Object
implements OptimizableNDContinuousDifferentiable

This class does numeric differentiation in ND. There are lots of methods of doing so, and depending on how many points are used, enter that value in for the method variable. Also computes 2nd derivatives and hessians Defaults to 3-point method.

Author:

Yufeng Guo, Hanlin Wan

Constructor Summary

FunctionNumericNDDifferentiation(OptimizableNDContinuous function)
Creates an object which can differentiate an ND function with default values.

FunctionNumericNDDifferentiation(OptimizableNDContinuous function, double step, int method)
Creates an object which can differentiate an ND function in various ways.

Method Summary

double[][]	get2ndDerivative(double[] x) Creates the hessian matrix
double	get2ndDerivative(double[] x, int dim1, int dim2) Gets 2nd partial derivative for the dimensions given
double[]	getDerivative(double[] x) Gets the gradient
double[]	getDomainMax() Gets the domain maximum
double[]	getDomainMin() Gets the domain minimum
double	getDomainTolerance() Gets the tolerance
int	getMethod() Gets the method used
int	getNumberOfDimensions() Gets the number of dimensions
double	getStep() Gets the step size used
double	getValue(double[] x) Gets the value

Methods inherited from class java.lang.Object

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

Constructor Detail

FunctionNumericNDDifferentiation

public FunctionNumericNDDifferentiation(OptimizableNDContinuous function)

Creates an object which can differentiate an ND function with default values. Step is 1e-5 and method is 3-point.

Parameters:

function - the function to be differentiated

FunctionNumericNDDifferentiation

public FunctionNumericNDDifferentiation(OptimizableNDContinuous function,
                                        double step,
                                        int method)

Creates an object which can differentiate an ND function in various ways.

Parameters:

function - the function to be differentiated

step - the step size

method - the differentiating method requested: 0,1,3,5,7, or 9. Default is 3.

Method Detail

get2ndDerivative

public double get2ndDerivative(double[] x,
                               int dim1,
                               int dim2)

Gets 2nd partial derivative for the dimensions given

Specified by:

get2ndDerivative in interface OptimizableNDContinuousDifferentiable

Parameters:

x - a double array that contains the location of the derivative required

dim1 - the first dimension of the derivative

dim2 - the second dimension of the derivative

Returns:

the 2nd derivative

get2ndDerivative

public double[][] get2ndDerivative(double[] x)

Creates the hessian matrix

Specified by:

get2ndDerivative in interface OptimizableNDContinuousDifferentiable

Parameters:

x - location to evaluate derivative at

Returns:

hessian matrix at location x

getDerivative

public double[] getDerivative(double[] x)

Gets the gradient

Specified by:

getDerivative in interface OptimizableNDContinuousDifferentiable

Parameters:

x - location to evaluate derivative at

Returns:

gradient at location x

getMethod

public int getMethod()

Gets the method used

Returns:

method

getStep

public double getStep()

Gets the step size used

Returns:

step size

getDomainMax

public double[] getDomainMax()

Gets the domain maximum

Specified by:

getDomainMax in interface OptimizableNDContinuous

Returns:

domain max

getDomainMin

public double[] getDomainMin()

Gets the domain minimum

Specified by:

getDomainMin in interface OptimizableNDContinuous

Returns:

domain min

getDomainTolerance

public double getDomainTolerance()

Gets the tolerance

Specified by:

getDomainTolerance in interface OptimizableNDContinuous

Returns:

tolerance

getValue

public double getValue(double[] x)

Gets the value

Specified by:

getValue in interface OptimizableNDContinuous

Parameters:

x - location to get value at

Returns:

value at location x

getNumberOfDimensions

public int getNumberOfDimensions()

Gets the number of dimensions

Specified by:

getNumberOfDimensions in interface OptimizableNDContinuous

Returns:

number of dimensions