Adaptive refinement algorithm. More...

#include <adaptive_refinement.hh>

Detailed Description

Adaptive refinement algorithm.

Implementation uses the strategy pattern, and should be flexible to use. In order to make it work one has to provide implementations of the following interfaces:

AbstractLinearSolver : responsible for the solution of the linear discrete problems
AbstractGrid : coarse grained interface to a grid, mostly mark/refine

Definition at line 67 of file adaptive_refinement.hh.

Inheritance diagram for Kaskade::SimpleAdaptiveRefinement:

Public Member Functions
	SimpleAdaptiveRefinement (AbstractNewtonDirection &als, AbstractErrorEstimator &aee, AbstractAdaptiveGrid &ag, AbstractNorm &an, AdaptiveParameters &p_)
	Construction. More...

virtual int	runAlgorithm ()
	To be used as an algorithm. More...

virtual	~SimpleAdaptiveRefinement ()

virtual void	setRelativeAccuracy (double accuracy)
	set relative accuracy More...

virtual void	setAbsoluteAccuracy (double accuracy)
	set absolute accuracy More...

double	getRelativeAccuracy ()
	get achieved relative accuracy More...

double	getAbsoluteAccuracy ()
	get achieved absolute accuracy More...

bool	improvementPossible ()

virtual void	doSolve (AbstractFunctionSpaceElement &corr, AbstractLinearization &lin)
	solve problem More...

virtual void	doResolve (AbstractFunctionSpaceElement &correction, AbstractLinearization const &lin, AbstractLinearization const &olin) const
	Solve simplified Newton system: scorrection = +F'(iterate)^{-1}F(trialIterate) More...

virtual void	doResolve (AbstractFunctionSpaceElement &correction, AbstractLinearization const &lin) const
	Solve simplified Newton system: scorrection = +F'(iterate)^{-1}F(trialIterate) More...

virtual bool	changedGrid ()

void	performTiming (bool doit)

void	reportOnIteration (int level)

void	ordinary (AbstractFunctionSpaceElement &correction, AbstractLinearization &linearization)

void	simplified (AbstractFunctionSpaceElement &correction, AbstractLinearization const &linearization, AbstractLinearization const &oldlinearization) const

void	simplified (AbstractFunctionSpaceElement &correction, AbstractLinearization const &linearization) const

Public Attributes
AdaptiveParameters &	p

bool	alwaysestimate

boost::signals2::signal< void()>	changed

Protected Member Functions
virtual bool	convergenceTest (AbstractErrorEstimate const &estimate, AbstractAdaptiveGrid const &)

virtual double	getBulk (int step) const

virtual void	initialize ()

virtual void	finalize (int flag)

virtual void	terminationMessage (int flag)

int	algorithmWrapper ()
	Run algorithm, completely with initialization and finalization. More...

int	oneStepWrapper ()
	Run one step of algorithm. More...

Protected Attributes
std::shared_ptr< AbstractErrorEstimate >	estimate

AbstractNewtonDirection &	fixedSolver

AbstractErrorEstimator &	errorEstimator

AbstractNorm &	norm

AbstractAdaptiveGrid &	grid

AbstractLinearization *	linearization

AbstractFunctionSpaceElement *	correction

std::unique_ptr< AbstractFunctionSpaceElement >	oldcorrection

double	noldold

int	report

Constructor & Destructor Documentation

◆ SimpleAdaptiveRefinement()

Kaskade::SimpleAdaptiveRefinement::SimpleAdaptiveRefinement	(	AbstractNewtonDirection &	als,
		AbstractErrorEstimator &	aee,
		AbstractAdaptiveGrid &	ag,
		AbstractNorm &	an,
		AdaptiveParameters &	p_
	)

inline

Construction.

Definition at line 72 of file adaptive_refinement.hh.

◆ ~SimpleAdaptiveRefinement()

virtual Kaskade::SimpleAdaptiveRefinement::~SimpleAdaptiveRefinement ( )

inlinevirtual

Definition at line 82 of file adaptive_refinement.hh.

Member Function Documentation

◆ algorithmWrapper()

int Kaskade::Algorithm::algorithmWrapper ( )

protectedinherited

Run algorithm, completely with initialization and finalization.

◆ changedGrid()

virtual bool Kaskade::SimpleAdaptiveRefinement::changedGrid ( )

inlinevirtual

Reimplemented from Kaskade::AbstractNewtonDirection.

Definition at line 115 of file adaptive_refinement.hh.

◆ convergenceTest()

virtual bool Kaskade::SimpleAdaptiveRefinement::convergenceTest	(	AbstractErrorEstimate const &	estimate,
		AbstractAdaptiveGrid const &
	)

protectedvirtual

◆ doResolve() [1/2]

virtual void Kaskade::SimpleAdaptiveRefinement::doResolve	(	AbstractFunctionSpaceElement &	correction,
		AbstractLinearization const &	lin
	)		const

inlinevirtual

Solve simplified Newton system: scorrection = +F'(iterate)^{-1}F(trialIterate)

Implements Kaskade::AbstractNewtonDirection.

Definition at line 110 of file adaptive_refinement.hh.

◆ doResolve() [2/2]

virtual void Kaskade::SimpleAdaptiveRefinement::doResolve	(	AbstractFunctionSpaceElement &	correction,
		AbstractLinearization const &	lin,
		AbstractLinearization const &	olin
	)		const

inlinevirtual

Solve simplified Newton system: scorrection = +F'(iterate)^{-1}F(trialIterate)

Implements Kaskade::AbstractNewtonDirection.

Definition at line 105 of file adaptive_refinement.hh.

◆ doSolve()

virtual void Kaskade::SimpleAdaptiveRefinement::doSolve	(	AbstractFunctionSpaceElement &	corr,
		AbstractLinearization &	lin
	)

virtual

solve problem

Implements Kaskade::AbstractNewtonDirection.

◆ finalize()

virtual void Kaskade::SimpleAdaptiveRefinement::finalize ( int flag )

protectedvirtual

Reimplemented from Kaskade::Algorithm.

◆ getAbsoluteAccuracy()

double Kaskade::SimpleAdaptiveRefinement::getAbsoluteAccuracy ( )

inlinevirtual

get achieved absolute accuracy

Implements Kaskade::AbstractNewtonDirection.

Definition at line 98 of file adaptive_refinement.hh.

◆ getBulk()

virtual double Kaskade::SimpleAdaptiveRefinement::getBulk ( int step ) const

protectedvirtual

Reimplemented in Kaskade::FinalAdaptiveRefinement.

◆ getRelativeAccuracy()

double Kaskade::SimpleAdaptiveRefinement::getRelativeAccuracy ( )

inlinevirtual

get achieved relative accuracy

Implements Kaskade::AbstractNewtonDirection.

Definition at line 96 of file adaptive_refinement.hh.

◆ improvementPossible()

bool Kaskade::SimpleAdaptiveRefinement::improvementPossible ( )

inlinevirtual

Implements Kaskade::AbstractNewtonDirection.

Definition at line 100 of file adaptive_refinement.hh.

◆ initialize()

virtual void Kaskade::SimpleAdaptiveRefinement::initialize ( )

protectedvirtual

Reimplemented from Kaskade::Algorithm.

◆ oneStepWrapper()

int Kaskade::Algorithm::oneStepWrapper ( )

protectedinherited

Run one step of algorithm.

◆ ordinary()

void Kaskade::AbstractNewtonDirection::ordinary	(	AbstractFunctionSpaceElement &	correction,
		AbstractLinearization &	linearization
	)

inlineinherited

This method is not const, since the internal state of the solver may be modified Computes an undamped ordinary Newton step, i.e., \(correction=-F'(x)^{-1}F(x)\)

Definition at line 275 of file abstract_interface.hh.

Referenced by Kaskade::NewtonsMethod::getSearchDirection().

◆ performTiming()

void Kaskade::Algorithm::performTiming ( bool doit )

inlineinherited

Definition at line 181 of file algorithm_base.hh.

◆ reportOnIteration()

void Kaskade::Algorithm::reportOnIteration ( int level )

inlineinherited

Definition at line 182 of file algorithm_base.hh.

◆ runAlgorithm()

virtual int Kaskade::SimpleAdaptiveRefinement::runAlgorithm ( )

virtual

To be used as an algorithm.

Implements Kaskade::Algorithm.

Reimplemented in Kaskade::FixedSolverWithErrorEstimate.

◆ setAbsoluteAccuracy()

virtual void Kaskade::SimpleAdaptiveRefinement::setAbsoluteAccuracy ( double accuracy )

inlinevirtual

set absolute accuracy

Reimplemented from Kaskade::AbstractNewtonDirection.

Reimplemented in Kaskade::FinalAdaptiveRefinement.

Definition at line 94 of file adaptive_refinement.hh.

◆ setRelativeAccuracy()

virtual void Kaskade::SimpleAdaptiveRefinement::setRelativeAccuracy ( double accuracy )

inlinevirtual

set relative accuracy

Implements Kaskade::AbstractNewtonDirection.

Reimplemented in Kaskade::FinalAdaptiveRefinement.

Definition at line 92 of file adaptive_refinement.hh.

◆ simplified() [1/2]

void Kaskade::AbstractNewtonDirection::simplified	(	AbstractFunctionSpaceElement &	correction,
		AbstractLinearization const &	linearization
	)		const

inlineinherited

Computes an undamped simplified Newton step, i.e., correction=-F'(xold)^{-1}F(x) where F'(xold)^{-1} has been kept from an earlier ordinary() call

Definition at line 287 of file abstract_interface.hh.

◆ simplified() [2/2]

void Kaskade::AbstractNewtonDirection::simplified	(	AbstractFunctionSpaceElement &	correction,
		AbstractLinearization const &	linearization,
		AbstractLinearization const &	oldlinearization
	)		const

inlineinherited

Computes an undamped simplified Newton step, i.e., correction=-F'(xold)^{-1}F(x)

Definition at line 280 of file abstract_interface.hh.

Referenced by Kaskade::NewtonsMethod::resolve().

◆ terminationMessage()

virtual void Kaskade::SimpleAdaptiveRefinement::terminationMessage ( int flag )

protectedvirtual

Reimplemented from Kaskade::Algorithm.

Member Data Documentation

◆ alwaysestimate

bool Kaskade::SimpleAdaptiveRefinement::alwaysestimate

Definition at line 133 of file adaptive_refinement.hh.

◆ changed

boost::signals2::signal<void()> Kaskade::AbstractNewtonDirection::changed

mutableinherited

Definition at line 319 of file abstract_interface.hh.

◆ correction

AbstractFunctionSpaceElement* Kaskade::SimpleAdaptiveRefinement::correction

protected

Definition at line 125 of file adaptive_refinement.hh.

Referenced by doResolve().

◆ errorEstimator

AbstractErrorEstimator& Kaskade::SimpleAdaptiveRefinement::errorEstimator

protected

Definition at line 121 of file adaptive_refinement.hh.

◆ estimate

std::shared_ptr<AbstractErrorEstimate> Kaskade::SimpleAdaptiveRefinement::estimate

protected

Definition at line 119 of file adaptive_refinement.hh.

◆ fixedSolver

AbstractNewtonDirection& Kaskade::SimpleAdaptiveRefinement::fixedSolver

protected

Definition at line 120 of file adaptive_refinement.hh.

Referenced by doResolve().

◆ grid

AbstractAdaptiveGrid& Kaskade::SimpleAdaptiveRefinement::grid

protected

Definition at line 123 of file adaptive_refinement.hh.

◆ linearization

AbstractLinearization* Kaskade::SimpleAdaptiveRefinement::linearization

protected

Definition at line 124 of file adaptive_refinement.hh.

◆ noldold

double Kaskade::SimpleAdaptiveRefinement::noldold

protected

Definition at line 127 of file adaptive_refinement.hh.

◆ norm

AbstractNorm& Kaskade::SimpleAdaptiveRefinement::norm

protected

Definition at line 122 of file adaptive_refinement.hh.

◆ oldcorrection

std::unique_ptr<AbstractFunctionSpaceElement> Kaskade::SimpleAdaptiveRefinement::oldcorrection

protected

Definition at line 126 of file adaptive_refinement.hh.

◆ p

AdaptiveParameters& Kaskade::SimpleAdaptiveRefinement::p

Definition at line 132 of file adaptive_refinement.hh.

Referenced by changedGrid(), getAbsoluteAccuracy(), getRelativeAccuracy(), improvementPossible(), setAbsoluteAccuracy(), Kaskade::FinalAdaptiveRefinement::setAbsoluteAccuracy(), setRelativeAccuracy(), and Kaskade::FinalAdaptiveRefinement::setRelativeAccuracy().

◆ report

int Kaskade::Algorithm::report

protectedinherited

Definition at line 193 of file algorithm_base.hh.

Referenced by Kaskade::Algorithm::reportOnIteration().

The documentation for this class was generated from the following file:

adaptive_refinement.hh

Detailed Description

Public Member Functions

Public Attributes

Protected Member Functions

Protected Attributes

Constructor & Destructor Documentation

◆ SimpleAdaptiveRefinement()

◆ ~SimpleAdaptiveRefinement()

Member Function Documentation

◆ algorithmWrapper()

◆ changedGrid()

◆ convergenceTest()

◆ doResolve() [1/2]

◆ doResolve() [2/2]

◆ doSolve()

◆ finalize()

◆ getAbsoluteAccuracy()

◆ getBulk()

◆ getRelativeAccuracy()

◆ improvementPossible()

◆ initialize()

◆ oneStepWrapper()

◆ ordinary()

◆ performTiming()

◆ reportOnIteration()

◆ runAlgorithm()

◆ setAbsoluteAccuracy()

◆ setRelativeAccuracy()

◆ simplified() [1/2]

◆ simplified() [2/2]

◆ terminationMessage()

Member Data Documentation

◆ alwaysestimate

◆ changed

◆ correction

◆ errorEstimator

◆ estimate

◆ fixedSolver

◆ grid

◆ linearization

◆ noldold

◆ norm

◆ oldcorrection

◆ p

◆ report