gov.sandia.cognition.learning.algorithm.minimization.matrix

Class SteepestDescentMatrixSolver

java.lang.Object

gov.sandia.cognition.learning.algorithm.minimization.matrix.IterativeMatrixSolver<MatrixVectorMultiplier>

gov.sandia.cognition.learning.algorithm.minimization.matrix.SteepestDescentMatrixSolver

All Implemented Interfaces:

AnytimeAlgorithm<InputOutputPair<Vector,Vector>>, IterativeAlgorithm, StoppableAlgorithm, BatchLearner<MatrixVectorMultiplier,InputOutputPair<Vector,Vector>>, FunctionMinimizer<Vector,Vector,MatrixVectorMultiplier>, CloneableSerializable, java.io.Serializable, java.lang.Cloneable

@PublicationReference(author="Jonathan Richard Shewchuk",
                      title="An Introduction to the Conjugate Gradient Method Without the Agonizing Pain",
                      type=WebPage,
                      year=1994,
                      url="http://www.cs.cmu.edu/~quake-papers/painless-conjugate-gradient.pdf\u200e")
public final class SteepestDescentMatrixSolver
extends IterativeMatrixSolver<MatrixVectorMultiplier>

Implements a basic Steepest Descent iterative solver for linear systems of equations.

Since:

4.0.0

Author:

Jeremy D. Wendt

See Also:

Serialized Form

Field Summary

Fields inherited from class gov.sandia.cognition.learning.algorithm.minimization.matrix.IterativeMatrixSolver

iterationCounter, listeners, maxIterations, rhs, shouldStop, tolerance, x0

Constructor Summary

Constructors

Constructor and Description
SteepestDescentMatrixSolver(Vector x0, Vector rhs) Initializes a steepest-descent solver with the minimum values
SteepestDescentMatrixSolver(Vector x0, Vector rhs, double tolerance) Initializes a steepest-descent solver with some additional parameters
SteepestDescentMatrixSolver(Vector x0, Vector rhs, double tolerance, int maxIterations) Initializes a steepest-descent solver with all user-definable parameters

Method Summary

Modifier and Type	Method and Description
CloneableSerializable	clone() Creates a new clone (shallow copy) of this object.
protected InputOutputPair<Vector,Vector>	completeSolver() Called after the final iteration.
boolean	equals(java.lang.Object o)
int	hashCode()
protected void	initializeSolver(MatrixVectorMultiplier function) Called before iterations begin in learn.
protected double	iterate() Called during each step of the iterative solver.

Methods inherited from class gov.sandia.cognition.learning.algorithm.minimization.matrix.IterativeMatrixSolver

addIterativeAlgorithmListener, getInitialGuess, getIteration, getMaxIterations, getResult, getTolerance, isResultValid, learn, removeIterativeAlgorithmListener, setInitialGuess, setMaxIterations, setTolerance, stop

Methods inherited from class java.lang.Object

finalize, getClass, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

SteepestDescentMatrixSolver

public SteepestDescentMatrixSolver(Vector x0,
                                   Vector rhs)

Initializes a steepest-descent solver with the minimum values

Parameters:

x0 - The initial guess for x

rhs - The "b" to solve to

SteepestDescentMatrixSolver

public SteepestDescentMatrixSolver(Vector x0,
                                   Vector rhs,
                                   double tolerance)

Initializes a steepest-descent solver with some additional parameters

Parameters:

x0 - The initial guess for x

rhs - The "b" to solve to

tolerance - The minimum tolerance that this must reach before stopping (unless maxIterations is exceeded).

SteepestDescentMatrixSolver

public SteepestDescentMatrixSolver(Vector x0,
                                   Vector rhs,
                                   double tolerance,
                                   int maxIterations)

Initializes a steepest-descent solver with all user-definable parameters

Parameters:

x0 - The initial guess for x

rhs - The "b" to solve to

tolerance - The minimum tolerance that this must reach before stopping (unless maxIterations is exceeded).

maxIterations - The maximum number of iterations to make

Method Detail

initializeSolver

protected final void initializeSolver(MatrixVectorMultiplier function)

Description copied from class: IterativeMatrixSolver

Called before iterations begin in learn. Iterative solvers can solve for initial state and should store function away.

Specified by:

initializeSolver in class IterativeMatrixSolver<MatrixVectorMultiplier>

Parameters:

function - The matrix wrapper to save for iterate.

iterate

protected final double iterate()

Description copied from class: IterativeMatrixSolver

Called during each step of the iterative solver. Take one step forward in the algorithm.

Specified by:

iterate in class IterativeMatrixSolver<MatrixVectorMultiplier>

Returns:

the residual after this step.

completeSolver

protected final InputOutputPair<Vector,Vector> completeSolver()

Description copied from class: IterativeMatrixSolver

Called after the final iteration. The solver should clean up any intermediate results and return the final results.

Specified by:

completeSolver in class IterativeMatrixSolver<MatrixVectorMultiplier>

Returns:

the final results of the algorithm.

clone

public final CloneableSerializable clone()

Description copied from interface: CloneableSerializable

Creates a new clone (shallow copy) of this object.

Specified by:

clone in interface CloneableSerializable

Specified by:

clone in class IterativeMatrixSolver<MatrixVectorMultiplier>

Returns:

A new clone (shallow copy) of this object.

See Also:

CloneableSerializable.clone()

equals

public boolean equals(java.lang.Object o)

Overrides:

equals in class IterativeMatrixSolver<MatrixVectorMultiplier>

hashCode

public int hashCode()

Overrides:

hashCode in class IterativeMatrixSolver<MatrixVectorMultiplier>

See Also:

Object.hashCode()