NumericalDerivative.java

/***********************************************************************

	This file is part of KEEL-software, the Data Mining tool for regression, 
	classification, clustering, pattern mining and so on.

	Copyright (C) 2004-2010
	
	F. Herrera (herrera@decsai.ugr.es)
    L. S�nchez (luciano@uniovi.es)
    J. Alcal�-Fdez (jalcala@decsai.ugr.es)
    S. Garc�a (sglopez@ujaen.es)
    A. Fern�ndez (alberto.fernandez@ujaen.es)
    J. Luengo (julianlm@decsai.ugr.es)

	This program is free software: you can redistribute it and/or modify
	it under the terms of the GNU General Public License as published by
	the Free Software Foundation, either version 3 of the License, or
	(at your option) any later version.

	This program is distributed in the hope that it will be useful,
	but WITHOUT ANY WARRANTY; without even the implied warranty of
	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
	GNU General Public License for more details.

	You should have received a copy of the GNU General Public License
	along with this program.  If not, see http://www.gnu.org/licenses/
  
**********************************************************************/

/*
 * NumericalDerivative.java
 *
 * Copyright (C) 2002-2006 Alexei Drummond and Andrew Rambaut
 *
 * This file is part of BEAST.
 * See the NOTICE file distributed with this work for additional
 * information regarding copyright ownership and licensing.
 *
 * BEAST is free software; you can redistribute it and/or modify
 * it under the terms of the GNU Lesser General Public License as
 * published by the Free Software Foundation; either version 2
 * of the License, or (at your option) any later version.
 *
 *  BEAST is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *  GNU Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with BEAST; if not, write to the
 * Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
 * Boston, MA  02110-1301  USA
 */

package keel.Algorithms.Preprocess.Missing_Values.EM.util;


/**
 * approximates numerically the first and second derivatives of a
 * function of a single variable and  approximates gradient and
 * diagonal of Hessian for multivariate functions
 *
 * Known bugs and limitations:
 * - the sparse number of function evaluations used can potentially
 *
 * @author Korbinian Strimmer
 */
public class NumericalDerivative
{
	//
	// Public stuff
	//


	/**
	 * determine first derivative
	 *
	 * @param f univariate function
	 * @param x argument
	 *
	 * @return first derivate at x
	 */
	public static double firstDerivative(UnivariateFunction f, double x)
	{	
		double h = MachineAccuracy.SQRT_EPSILON*(Math.abs(x) + 1.0);

		// Centered first derivative
		return (f.evaluate(x + h) - f.evaluate(x - h))/(2.0*h);
	}

	/**
	 * determine second derivative
	 *
	 * @param f univariate function
	 * @param x argument
	 *
	 * @return second derivate at x
	 */
	public static double secondDerivative(UnivariateFunction f, double x)
	{
		double h = MachineAccuracy.SQRT_SQRT_EPSILON*(Math.abs(x) + 1.0);
	
		// Centered second derivative
		return (f.evaluate(x + h) - 2.0*f.evaluate(x) + f.evaluate(x - h))/(h*h);
	}

	
	/**
	 * determine gradient
	 *
	 * @param f multivariate function
	 * @param x argument vector
	 *
	 * @return gradient at x
	 */
	public static double[] gradient(MultivariateFunction f, double[] x)
	{	
		double[] result = new double[x.length];

		gradient(f, x, result);
		
		return result;
	}

	/**
	 * determine gradient
	 *
	 * @param f multivariate function
	 * @param x argument vector
	 * @param grad vector for gradient
	 */
	public static void gradient(MultivariateFunction f, double[] x, double[] grad)
	{	
		for (int i = 0; i < f.getNumArguments(); i++)
		{
			double h = MachineAccuracy.SQRT_EPSILON*(Math.abs(x[i]) + 1.0);
		
			double oldx = x[i];
			x[i] = oldx + h;
			double fxplus = f.evaluate(x);
			x[i] = oldx - h;
			double fxminus = f.evaluate(x);
			x[i] = oldx;

			// Centered first derivative
			grad[i] = (fxplus-fxminus)/(2.0*h);
		}
	}

	/**
	 * determine diagonal of Hessian
	 *
	 * @param f multivariate function
	 * @param x argument vector
	 *
	 * @return vector with diagonal entries of Hessian
	 */
	public static double[] diagonalHessian(MultivariateFunction f, double[] x)
	{
		int len = f.getNumArguments();
		double[] result = new double[len];

		for (int i = 0; i < len; i++)
		{
			double h = MachineAccuracy.SQRT_SQRT_EPSILON*(Math.abs(x[i]) + 1.0);
		
			double oldx = x[i];
			x[i] = oldx + h;
			double fxplus = f.evaluate(x);
			x[i] = oldx - h;
			double fxminus = f.evaluate(x);
			x[i] = oldx;
			double fx = f.evaluate(x);

			// Centered second derivative
			result[i] = (fxplus - 2.0*fx + fxminus)/(h*h);
		}
		
		return result;
	}
}