Package org.apache.commons.math4.linear

Class DefaultIterativeLinearSolverEvent

java.lang.Object

java.util.EventObject

org.apache.commons.math4.util.IterationEvent

org.apache.commons.math4.linear.IterativeLinearSolverEvent

org.apache.commons.math4.linear.DefaultIterativeLinearSolverEvent

All Implemented Interfaces:

java.io.Serializable

public class DefaultIterativeLinearSolverEvent
extends IterativeLinearSolverEvent

A default concrete implementation of the abstract class IterativeLinearSolverEvent.

See Also:

Serialized Form

Field Summary

Fields inherited from class java.util.EventObject

source

Constructor Summary

Constructors

Constructor	Description
DefaultIterativeLinearSolverEvent(java.lang.Object source, int iterations, RealVector x, RealVector b, double rnorm)	Creates a new instance of this class.
DefaultIterativeLinearSolverEvent(java.lang.Object source, int iterations, RealVector x, RealVector b, RealVector r, double rnorm)	Creates a new instance of this class.

Method Summary

Modifier and Type	Method	Description
double	getNormOfResidual()	Returns the norm of the residual.
RealVector	getResidual()	Returns the residual.
RealVector	getRightHandSideVector()	Returns the current right-hand side of the linear system to be solved.
RealVector	getSolution()	Returns the current estimate of the solution to the linear system to be solved.
boolean	providesResidual()	Returns true if IterativeLinearSolverEvent.getResidual() is supported.

Methods inherited from class org.apache.commons.math4.util.IterationEvent

getIterations

Methods inherited from class java.util.EventObject

getSource, toString

Methods inherited from class java.lang.Object

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

Constructor Details

DefaultIterativeLinearSolverEvent

public DefaultIterativeLinearSolverEvent(java.lang.Object source, int iterations, RealVector x, RealVector b, RealVector r, double rnorm)

Creates a new instance of this class. This implementation does not deep copy the specified vectors x, b, r. Therefore the user must make sure that these vectors are either unmodifiable views or deep copies of the same vectors actually used by the source. Failure to do so may compromise subsequent iterations of the source. If the residual vector r is null, then getResidual() throws a MathUnsupportedOperationException, and providesResidual() returns false.

Parameters:

source - the iterative solver which fired this event

iterations - the number of iterations performed at the time this event is created

x - the current estimate of the solution

b - the right-hand side vector

r - the current estimate of the residual (can be null)

rnorm - the norm of the current estimate of the residual

DefaultIterativeLinearSolverEvent

public DefaultIterativeLinearSolverEvent(java.lang.Object source, int iterations, RealVector x, RealVector b, double rnorm)

Creates a new instance of this class. This implementation does not deep copy the specified vectors x, b. Therefore the user must make sure that these vectors are either unmodifiable views or deep copies of the same vectors actually used by the source. Failure to do so may compromise subsequent iterations of the source. Callling getResidual() on instances returned by this constructor throws a MathUnsupportedOperationException, while providesResidual() returns false.

Parameters:

source - the iterative solver which fired this event

iterations - the number of iterations performed at the time this event is created

x - the current estimate of the solution

b - the right-hand side vector

rnorm - the norm of the current estimate of the residual

Method Details

getNormOfResidual

public double getNormOfResidual()

Returns the norm of the residual. The returned value is not required to be exact. Instead, the norm of the so-called updated residual (if available) should be returned. For example, the conjugate gradient method computes a sequence of residuals, the norm of which is cheap to compute. However, due to accumulation of round-off errors, this residual might differ from the true residual after some iterations. See e.g. A. Greenbaum and Z. Strakos, Predicting the Behavior of Finite Precision Lanzos and Conjugate Gradient Computations, Technical Report 538, Department of Computer Science, New York University, 1991 (available here).

Specified by:

getNormOfResidual in class IterativeLinearSolverEvent

Returns:

the norm of the residual, ||r||

getResidual

public RealVector getResidual()

Returns the residual. This is an optional operation, as all iterative linear solvers do not provide cheap estimate of the updated residual vector, in which case

• this method should throw a MathUnsupportedOperationException,
• IterativeLinearSolverEvent.providesResidual() returns false.

The default implementation throws a MathUnsupportedOperationException. If this method is overriden, then IterativeLinearSolverEvent.providesResidual() should be overriden as well.

This implementation throws an MathUnsupportedOperationException if no residual vector r was provided at construction time.

Overrides:

getResidual in class IterativeLinearSolverEvent

Returns:

the updated residual, r

getRightHandSideVector

public RealVector getRightHandSideVector()

Returns the current right-hand side of the linear system to be solved. This method should return an unmodifiable view, or a deep copy of the actual right-hand side vector, in order not to compromise subsequent iterations of the source IterativeLinearSolver.

Specified by:

getRightHandSideVector in class IterativeLinearSolverEvent

Returns:

the right-hand side vector, b

getSolution

public RealVector getSolution()

Returns the current estimate of the solution to the linear system to be solved. This method should return an unmodifiable view, or a deep copy of the actual current solution, in order not to compromise subsequent iterations of the source IterativeLinearSolver.

Specified by:

getSolution in class IterativeLinearSolverEvent

Returns:

the solution, x

providesResidual

public boolean providesResidual()

Returns true if IterativeLinearSolverEvent.getResidual() is supported. The default implementation returns false. This implementation returns true if a non-null value was specified for the residual vector r at construction time.

Overrides:

providesResidual in class IterativeLinearSolverEvent

Returns:

true if r != null