/* * FunctionApproximator.java */package EDU.gatech.cc.is.learning;import java.io.*;import java.io.IOException;/** * Provides an abstract interface to  * various function approximator implementations. * <P> * Copyright (c)2000 Georgia Tech Research Corporation * * @author Juan Carlos Santamaria (carlos@cc.gatech.edu) * @version $Revision: 1.1 $ */public abstract class FunctionApproximator	implements Cloneable, Serializable{    /**     * The dimension of the domain space.     */    public final int domain_dim;    /**     * The dimension of the range space.     */    public final int range_dim;    /**     * Create an instance of a function approximator.  The approximator     * will map from <code>n</code> dimensions to <code>m</code> dimensions.     *     * @param      n   dimension of the domain space.     * @param      m   dimension of the range space.     */    public FunctionApproximator( int n, int m ) {        domain_dim = n;        range_dim  = m;    }    /**     * Create an instance of a function approximator according to the     * definition in a given file. The file format depends on the     * particular implementation.     *     * @param      filename   dimension of the domain space.     * @exception  FunctionApproximatorException  always.     */    public FunctionApproximator( String filename )    throws FunctionApproximatorException    {        // NOTE: This is a dummy body. It initializes the variables to some        // value to avoid a compilation error. The the subclass constructor        // is responsible for setting these variables.                   domain_dim = 0;        range_dim  = 0;        throw new FunctionApproximatorException(                      "FunctionApproximator: not implemented. "+                      "See documentation." );    }    /**     * Save a definition of this instance in a file.     *     * @param      filename   the file name.     * @exception  IOException                    if an I/O error occurs.     * @exception  FunctionApproximatorException  if something wrong occurs.     */    public abstract void saveDefinition( String filename )    throws FunctionApproximatorException, IOException;        /**     * Query the function approximator.     *     * @param      q   an n-dimensional array of doubles specifying a point     *                to evaluate.     *     * @return     an m-dimensional array of doubles representing the     *             value at the query point.     * @exception  FunctionApproximatorException  if something wrong occurs.     */    public abstract double[] query(double[] q)    throws FunctionApproximatorException;    /**     * Add a case to the approximator's learning set.  For some     * learners, this has no effect.     *     * @param      q   an n-dimensional array of doubles specifying a point     *                 in the domain space.     * @param      p   an m-dimensional array of doubles specifying the     *                 associate point in the range space.     * @exception  FunctionApproximatorException  if something wrong occurs.     */    public abstract void update(double[] q, double[] p)    throws FunctionApproximatorException;    }