cern.colt.matrix.tfloat.algo.solver

Class DefaultFloatIterationMonitor

java.lang.Object

cern.colt.matrix.tfloat.algo.solver.AbstractFloatIterationMonitor

cern.colt.matrix.tfloat.algo.solver.DefaultFloatIterationMonitor

All Implemented Interfaces:

FloatIterationMonitor

Direct Known Subclasses:

CGLSFloatIterationMonitor, MRNSDFloatIterationMonitor

public class DefaultFloatIterationMonitor
extends AbstractFloatIterationMonitor

Default iteration monitor. This tester checks declares convergence if the absolute value of the residual norm is sufficiently small, or if the relative decrease is small. Divergence is decleared if too many iterations are spent, or the residual has grown too much. NaNs will also cause divergence to be flagged.

Constructor Summary

Constructors

Constructor and Description
DefaultFloatIterationMonitor() Constructor for DefaultIterationMonitor.
DefaultFloatIterationMonitor(int maxIter, float rtol, float atol, float dtol) Constructor for DefaultIterationMonitor

Method Summary

Modifier and Type	Method and Description
float	getAbsoluteTolerance() Returns the absolute convergence tolerance
float	getDivergenceTolerance() Returns the relative divergence tolerance
int	getMaxIterations() Returns maximum number of iterations to permit
float	getRelativeTolerance() Returns the relative convergence tolerance
void	setAbsoluteTolerance(float atol) Sets the absolute convergence tolerance
void	setDivergenceTolerance(float dtol) Sets the relative divergence tolerance
void	setMaxIterations(int maxIter) Sets maximum number of iterations to permit
void	setRelativeTolerance(float rtol) Sets the relative convergence tolerance

Methods inherited from class cern.colt.matrix.tfloat.algo.solver.AbstractFloatIterationMonitor

converged, converged, converged, converged, getIterationReporter, getNormType, isFirst, iterations, next, residual, setFirst, setIterationReporter, setNormType

Methods inherited from class java.lang.Object

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

Constructor Detail

DefaultFloatIterationMonitor

public DefaultFloatIterationMonitor(int maxIter,
                                    float rtol,
                                    float atol,
                                    float dtol)

Constructor for DefaultIterationMonitor

Parameters:

maxIter - Maximum number of iterations

rtol - Relative convergence tolerance (to initial residual)

atol - Absolute convergence tolerance

dtol - Relative divergence tolerance (to initial residual)

DefaultFloatIterationMonitor

public DefaultFloatIterationMonitor()

Constructor for DefaultIterationMonitor. Default is 100000 iterations at most, relative tolerance of 1e-3, absolute tolerance of 1e-25 and a divergence tolerance of 1e+3.

Method Detail

setMaxIterations

public void setMaxIterations(int maxIter)

Sets maximum number of iterations to permit

Parameters:

maxIter - Maximum number of iterations

getMaxIterations

public int getMaxIterations()

Returns maximum number of iterations to permit

setRelativeTolerance

public void setRelativeTolerance(float rtol)

Sets the relative convergence tolerance

Parameters:

rtol - relative convergence tolerance (to initial residual)

getRelativeTolerance

public float getRelativeTolerance()

Returns the relative convergence tolerance

Returns:

relative convergence tolerance (to initial residual)

setAbsoluteTolerance

public void setAbsoluteTolerance(float atol)

Sets the absolute convergence tolerance

Parameters:

atol - absolute convergence tolerance

getAbsoluteTolerance

public float getAbsoluteTolerance()

Returns the absolute convergence tolerance

Returns:

absolute convergence tolerance

setDivergenceTolerance

public void setDivergenceTolerance(float dtol)

Sets the relative divergence tolerance

Parameters:

dtol - relative divergence tolerance (to initial residual)

getDivergenceTolerance

public float getDivergenceTolerance()

Returns the relative divergence tolerance

Returns:

relative divergence tolerance (to initial residual)