JMSLTM Numerical Library 6.1

com.imsl.math
Class RadialBasis

java.lang.Object
  extended by com.imsl.math.RadialBasis
All Implemented Interfaces:
Serializable, Cloneable

public class RadialBasis
extends Object
implements Serializable, Cloneable

RadialBasis computes a least-squares fit to scattered data in {bf R}^d, where d is the dimension. More precisely, we are given data points

x_0,ldots,x_{n-1}in{bf R}^d

and function values

f_0,ldots,f_{n-1}in{bf R}^1

The radial basis fit to the data is a function F which approximates the above data in the sense that it minimizes the sum-of-squares error

sum_{i=0}^{n-1}w_ileft(F(x_i)-f_iright)^2

where w are the weights. Of course, we must restrict the functional form of F. Here we assume it is a linear combination of radial functions:

F(x)equivsum_{j=0}^{m-1}alpha_jphi(|x-c_j|)

The c_j are the centers.

A radial function, phi(r), maps [0,infty) into {bf R}^1. The default radial function is the Hardy multiquadric,

phi(r)equivsqrt{r^2+delta^2}

with delta=1. An alternate radial function is the Gaussian, e^{-ax^2}.

By default, the centers are points in a Faure sequence, scaled to cover the box containing the data.

Two update methods allow the user to specify weights for each data point in the approximation scheme. In this way the user can influence the fit of the radial basis function. For example, if weights are in the range [0,1] then 0-weighted points are effectively removed from computations and 1-weighted points will have more influence than any others. When the number of centers equals the number of data points, the RBF fit will be "exact", otherwise it will be an approximation (useful for large or noisy data sets). Provided the ratios of the weights are not too extreme, weights will not appreciably change the accuracy of the fit to the data, but they will affect the shape of the approximating function away from the data: Greater weights result in greater influence at greater distances.

See Also:
Example, Serialized Form

Nested Class Summary
static interface RadialBasis.Function
          Public interface for the user supplied function to the RadialBasis object.
static class RadialBasis.Gaussian
          The Gaussian basis function, e^{-ax^2}.
static class RadialBasis.HardyMultiquadric
          The Hardy multiquadric basis function, sqrt{r^2+delta^2}.
 
Constructor Summary
RadialBasis(int nDim, int nCenters)
          Creates a new instance of RadialBasis.
 
Method Summary
 ANOVA getANOVA()
          Returns the ANOVA statistics from the linear regression.
 RadialBasis.Function getRadialFunction()
          Returns the radial function.
 double[] gradient(double[] x)
          Returns the gradient of the radial basis approximation at a point.
 void setRadialFunction(RadialBasis.Function radialFunction)
          Sets the radial function.
 void update(double[][] x, double[] f)
          Adds a set of data points, all with weight = 1.
 void update(double[][] x, double[] f, double[] w)
          Adds a set of data points with user-specified weights.
 void update(double[] x, double f)
          Adds a data point with weight = 1.
 void update(double[] x, double f, double w)
          Adds a data point with a specified weight.
 double value(double[] x)
          Returns the value of the radial basis approximation at a point.
 double[] value(double[][] x)
          Returns the value of the radial basis at a point.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Constructor Detail

RadialBasis

public RadialBasis(int nDim,
                   int nCenters)
Creates a new instance of RadialBasis.

Parameters:
nDim - an int specifying the number of dimensions.
nCenters - an int specifying the number of centers.
Method Detail

getANOVA

public ANOVA getANOVA()
Returns the ANOVA statistics from the linear regression.

Returns:
an ANOVA table and related statistics
See Also:
LinearRegression, ANOVA

getRadialFunction

public RadialBasis.Function getRadialFunction()
Returns the radial function.

Returns:
a RadialBasis.Function which is the current radial function.

gradient

public double[] gradient(double[] x)
Returns the gradient of the radial basis approximation at a point.

Parameters:
x - is a double array containing the locations of the data point at which the approximation's gradient is to be computed.
Returns:
a double array, of length nDim containing the value of the gradient of the radial basis approximation at x.

setRadialFunction

public void setRadialFunction(RadialBasis.Function radialFunction)
Sets the radial function.

Parameters:
radialFunction - a RadialBasis.Function to be used in the approximation. The default is Hardy Multiquadric with delta=1.

update

public void update(double[][] x,
                   double[] f)
Adds a set of data points, all with weight = 1.

Parameters:
x - is a double matrix of size n by nDim containing the locations of the data points for each dimension.
f - is a double array containing the function values at the data points.

update

public void update(double[][] x,
                   double[] f,
                   double[] w)
Adds a set of data points with user-specified weights.

Parameters:
x - is a double matrix of size n by nDim containing the locations of the data points for each dimension.
f - is a double array containing the function values at the data points.
w - is a double array containing the weights associated with the data points.

update

public void update(double[] x,
                   double f)
Adds a data point with weight = 1.

Parameters:
x - is a double array containing the locations of the data point.
f - is a double containing the function value at the data point.

update

public void update(double[] x,
                   double f,
                   double w)
Adds a data point with a specified weight.

Parameters:
x - is a double array containing the locations of the data point.
f - is a double containing the function value at the data point.
w - is a double containing the weight of this data point.

value

public double value(double[] x)
Returns the value of the radial basis approximation at a point.

Parameters:
x - is a double array containing the locations of the data point at which the approximation is to be computed.
Returns:
a double containing the value of the radial basis approximation at x.

value

public double[] value(double[][] x)
Returns the value of the radial basis at a point.

Parameters:
x - a double matrix of size n by nDim containing the points at which the radial basis is to be evaluated.
Returns:
a double array giving the value of the radial basis at the point x

JMSLTM Numerical Library 6.1

Copyright © 1970-2010 Visual Numerics, Inc.
Built July 30 2010.