Package net.sf.tweety.math.opt.solver

Class HessianGradientDescent

java.lang.Object

net.sf.tweety.math.opt.Solver

net.sf.tweety.math.opt.solver.HessianGradientDescent

public class HessianGradientDescent
extends Solver

This class implements a gradient descent involving Hessian correction for solving unconstrained optimization problems.

Author:

Matthias Thimm

Field Summary

Fields

Modifier and Type	Field	Description
private org.slf4j.Logger	log	Logger.
private static double	PRECISION
private java.util.Map<Variable,Term>	startingPoint	The starting point for the solver.

Constructor Summary

Constructors

Constructor	Description
HessianGradientDescent(java.util.Map<Variable,Term> startingPoint)

Method Summary

Modifier and Type	Method	Description
private double[]	bestGuess(double[] currentGuess, double[] dir, java.util.List<Term> gradient, java.util.List<Variable> variables)	Find the best guess.
private double[]	getDirection(double[][] approxHessian, double[] evaluatedGradient)	Find the best direction.
static boolean	isInstalled()
java.util.Map<Variable,Term>	solve(ConstraintSatisfactionProblem problem)	Computes a solution to the given constraint satisfaction or optimization problem, i.e.

Methods inherited from class net.sf.tweety.math.opt.Solver

getDefaultGeneralSolver, getDefaultIntegerLinearSolver, getDefaultLinearSolver, hasDefaultGeneralSolver, hasDefaultIntegerLinearSolver, hasDefaultLinearSolver, setDefaultGeneralSolver, setDefaultIntegerLinearSolver, setDefaultLinearSolver

Methods inherited from class java.lang.Object

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

Field Detail

log

private org.slf4j.Logger log

Logger.

PRECISION

private static final double PRECISION

See Also:

Constant Field Values

startingPoint

private java.util.Map<Variable,Term> startingPoint

The starting point for the solver.

Constructor Detail

HessianGradientDescent

public HessianGradientDescent(java.util.Map<Variable,Term> startingPoint)

Method Detail

solve

public java.util.Map<Variable,Term> solve(ConstraintSatisfactionProblem problem)
                                         throws GeneralMathException

Description copied from class: Solver

Computes a solution to the given constraint satisfaction or optimization problem, i.e. a mapping from variables of the problem to terms.

Specified by:

solve in class Solver

Parameters:

problem - the actual problem

Returns:

a mapping from variables of the problem to terms.

Throws:

GeneralMathException - if something went wrong.

bestGuess

private double[] bestGuess(double[] currentGuess,
                           double[] dir,
                           java.util.List<Term> gradient,
                           java.util.List<Variable> variables)

Find the best guess.

Parameters:

currentGuess - the current guess

dir - the direction

gradient - the gradient

variables - the variables.

Returns:

the best guess

getDirection

private double[] getDirection(double[][] approxHessian,
                              double[] evaluatedGradient)

Find the best direction.

Parameters:

approxHessian - the approximated Hessian

evaluatedGradient - the evaluated gradient

Returns:

the direction

isInstalled

public static boolean isInstalled()
                           throws java.lang.UnsupportedOperationException

Throws:

java.lang.UnsupportedOperationException