The Stopping criterion namespace. More...

Classes
class	AbsoluteResidualNorm
	The AbsoluteResidualNorm class is a stopping criterion which stops the iteration process when the residual norm is below a certain threshold, i.e. More...

class	Combined
	The Combined class is used to combine multiple criterions together through an OR operation. More...

class	Criterion
	The Criterion class is a base class for all stopping criteria. More...

struct	CriterionArgs
	This struct is used to pass parameters to the EnableDefaultCriterionFactoryCriterionFactory::generate() method. More...

class	ImplicitResidualNorm
	The ImplicitResidualNorm class is a stopping criterion which stops the iteration process when the implicit residual norm is below a certain threshold relative to. More...

class	Iteration
	The Iteration class is a stopping criterion which stops the iteration process after a preset number of iterations. More...

class	RelativeResidualNorm
	The RelativeResidualNorm class is a stopping criterion which stops the iteration process when the residual norm is below a certain threshold relative to the norm of the right-hand side, i.e. More...

class	ResidualNorm
	The ResidualNorm class is a stopping criterion which stops the iteration process when the actual residual norm is below a certain threshold relative to. More...

class	ResidualNormBase
	The ResidualNormBase class provides a framework for stopping criteria related to the residual norm. More...

class	ResidualNormReduction
	The ResidualNormReduction class is a stopping criterion which stops the iteration process when the residual norm is below a certain threshold relative to the norm of the initial residual, i.e. More...

class	Time
	The Time class is a stopping criterion which stops the iteration process after a certain amount of time has passed. More...

Typedefs
using	CriterionFactory = AbstractFactory< Criterion, CriterionArgs >
	Declares an Abstract Factory specialized for Criterions.

template<typename ConcreteFactory , typename ConcreteCriterion , typename ParametersType , typename PolymorphicBase = CriterionFactory>
using	EnableDefaultCriterionFactory = EnableDefaultFactory< ConcreteFactory, ConcreteCriterion, ParametersType, PolymorphicBase >
	This is an alias for the EnableDefaultFactory mixin, which correctly sets the template parameters to enable a subclass of CriterionFactory. More...

Enumerations
enum	mode { absolute, initial_resnorm, rhs_norm }
	The mode for the residual norm criterion. More...

Functions
template<typename FactoryContainer >
std::shared_ptr< const CriterionFactory >	combine (FactoryContainer &&factories)
	Combines multiple criterion factories into a single combined criterion factory. More...

deferred_factory_parameter< const Iteration::Factory >	max_iters (size_type count)
	Creates the precursor to an Iteration stopping criterion factory, to be used in conjunction with `.with_criteria(...)` function calls when building a solver factory. More...

deferred_factory_parameter< const CriterionFactory >	min_iters (size_type count, deferred_factory_parameter< const CriterionFactory > criterion)
	Creates the precursor to an MinimumIteration stopping criterion factory, to be used in conjunction with `.with_criteria(...)` function calls when building a solver factory. More...

template<typename... Args>
deferred_factory_parameter< const CriterionFactory >	min_iters (size_type count, Args &&... criteria)

deferred_factory_parameter< CriterionFactory >	absolute_residual_norm (double tolerance)
	Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with `.with_criteria(...)` function calls when building a solver factory. More...

deferred_factory_parameter< CriterionFactory >	relative_residual_norm (double tolerance)
	Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with `.with_criteria(...)` function calls when building a solver factory. More...

deferred_factory_parameter< CriterionFactory >	initial_residual_norm (double tolerance)
	Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with `.with_criteria(...)` function calls when building a solver factory. More...

deferred_factory_parameter< CriterionFactory >	absolute_implicit_residual_norm (double tolerance)
	Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with `.with_criteria(...)` function calls when building a solver factory. More...

deferred_factory_parameter< CriterionFactory >	relative_implicit_residual_norm (double tolerance)
	Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with `.with_criteria(...)` function calls when building a solver factory. More...

deferred_factory_parameter< CriterionFactory >	initial_implicit_residual_norm (double tolerance)
	Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with `.with_criteria(...)` function calls when building a solver factory. More...

deferred_factory_parameter< Time::Factory >	time_limit (std::chrono::nanoseconds duration)
	Creates the precursor to a Time stopping criterion factory, to be used in conjunction with `.with_criteria(...)` function calls when building a solver factory. More...

Detailed Description

The Stopping criterion namespace.

Stopping criteria

Typedef Documentation

◆ EnableDefaultCriterionFactory

template<typename ConcreteFactory , typename ConcreteCriterion , typename ParametersType , typename PolymorphicBase = CriterionFactory>

using gko::stop::EnableDefaultCriterionFactory = typedef EnableDefaultFactory<ConcreteFactory, ConcreteCriterion, ParametersType, PolymorphicBase>

This is an alias for the EnableDefaultFactory mixin, which correctly sets the template parameters to enable a subclass of CriterionFactory.

Template Parameters

ConcreteFactory	the concrete factory which is being implemented [CRTP parameter]
ConcreteCriterion	the concrete Criterion type which this factory produces, needs to have a constructor which takes a const ConcreteFactory , and a const CriterionArgs as parameters.
ParametersType	a subclass of enable_parameters_type template which defines all of the parameters of the factory
PolymorphicBase	parent of ConcreteFactory in the polymorphic hierarchy, has to be a subclass of CriterionFactory

Function Documentation

◆ absolute_implicit_residual_norm()

deferred_factory_parameter<CriterionFactory> gko::stop::absolute_implicit_residual_norm ( double tolerance )

Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with .with_criteria(...) function calls when building a solver factory.

This stopping criterion will stop the iteration after the residual norm has decreased below the specified value or by the specified amount.

Full usage example: Stop after 100 iterations or when the absolute residual norm is below $10^{-10}$ , whichever happens first.

auto factory = gko::solver::Cg<double>::build()
                   .with_criteria(
                       gko::stop::max_iters(100),
                       gko::stop::absolute_residual_norm(1e-10))
                   .on(exec);

Parameters

tolerance the value the residual norm needs to be below. With residual , initial guess , right-hand side , matrix , absolute means the exact value of the norm , relative means the norm relative to the right-hand side , initial means the norm relative to the initial residual . An implicit stopping criterion is only available with some solvers, and refers to either the energy norm in short-recurrence solvers like Cg or the euclidian norm in solvers like GMRES. Implicit residual norms are cheaper to compute, but may be less precise due to accumulating rounding errors.

Returns: a deferred_factory_parameter that can be passed to the with_criteria function when building a solver.

◆ absolute_residual_norm()

deferred_factory_parameter<CriterionFactory> gko::stop::absolute_residual_norm ( double tolerance )

Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with .with_criteria(...) function calls when building a solver factory.

This stopping criterion will stop the iteration after the residual norm has decreased below the specified value or by the specified amount.

Full usage example: Stop after 100 iterations or when the absolute residual norm is below $10^{-10}$ , whichever happens first.

auto factory = gko::solver::Cg<double>::build()
                   .with_criteria(
                       gko::stop::max_iters(100),
                       gko::stop::absolute_residual_norm(1e-10))
                   .on(exec);

Parameters

tolerance the value the residual norm needs to be below. With residual , initial guess , right-hand side , matrix , absolute means the exact value of the norm , relative means the norm relative to the right-hand side , initial means the norm relative to the initial residual . An implicit stopping criterion is only available with some solvers, and refers to either the energy norm in short-recurrence solvers like Cg or the euclidian norm in solvers like GMRES. Implicit residual norms are cheaper to compute, but may be less precise due to accumulating rounding errors.

Returns: a deferred_factory_parameter that can be passed to the with_criteria function when building a solver.

◆ initial_implicit_residual_norm()

deferred_factory_parameter<CriterionFactory> gko::stop::initial_implicit_residual_norm ( double tolerance )

Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with .with_criteria(...) function calls when building a solver factory.

This stopping criterion will stop the iteration after the residual norm has decreased below the specified value or by the specified amount.

Full usage example: Stop after 100 iterations or when the absolute residual norm is below $10^{-10}$ , whichever happens first.

auto factory = gko::solver::Cg<double>::build()
                   .with_criteria(
                       gko::stop::max_iters(100),
                       gko::stop::absolute_residual_norm(1e-10))
                   .on(exec);

Parameters

tolerance the value the residual norm needs to be below. With residual , initial guess , right-hand side , matrix , absolute means the exact value of the norm , relative means the norm relative to the right-hand side , initial means the norm relative to the initial residual . An implicit stopping criterion is only available with some solvers, and refers to either the energy norm in short-recurrence solvers like Cg or the euclidian norm in solvers like GMRES. Implicit residual norms are cheaper to compute, but may be less precise due to accumulating rounding errors.

Returns: a deferred_factory_parameter that can be passed to the with_criteria function when building a solver.

◆ initial_residual_norm()

deferred_factory_parameter<CriterionFactory> gko::stop::initial_residual_norm ( double tolerance )

Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with .with_criteria(...) function calls when building a solver factory.

This stopping criterion will stop the iteration after the residual norm has decreased below the specified value or by the specified amount.

Full usage example: Stop after 100 iterations or when the absolute residual norm is below $10^{-10}$ , whichever happens first.

auto factory = gko::solver::Cg<double>::build()
                   .with_criteria(
                       gko::stop::max_iters(100),
                       gko::stop::absolute_residual_norm(1e-10))
                   .on(exec);

Parameters

tolerance the value the residual norm needs to be below. With residual , initial guess , right-hand side , matrix , absolute means the exact value of the norm , relative means the norm relative to the right-hand side , initial means the norm relative to the initial residual . An implicit stopping criterion is only available with some solvers, and refers to either the energy norm in short-recurrence solvers like Cg or the euclidian norm in solvers like GMRES. Implicit residual norms are cheaper to compute, but may be less precise due to accumulating rounding errors.

Returns: a deferred_factory_parameter that can be passed to the with_criteria function when building a solver.

◆ max_iters()

deferred_factory_parameter<const Iteration::Factory> gko::stop::max_iters ( size_type count )

Creates the precursor to an Iteration stopping criterion factory, to be used in conjunction with .with_criteria(...) function calls when building a solver factory.

This stopping criterion will stop the iteration after count iterations of the solver have finished.

Full usage example: Stop after 100 iterations or when the relative residual norm is below $10^{-10}$ , whichever happens first.

auto factory = gko::solver::Cg<double>::build()
                   .with_criteria(
                       gko::stop::max_iters(100),
                       gko::stop::relative_residual_norm(1e-10))
                   .on(exec);

Parameters

count the number of iterations after which to stop

Returns: a deferred_factory_parameter that can be passed to the with_criteria function when building a solver.

◆ min_iters()

deferred_factory_parameter<const CriterionFactory> gko::stop::min_iters	(	size_type	count,
		deferred_factory_parameter< const CriterionFactory >	criterion
	)

Creates the precursor to an MinimumIteration stopping criterion factory, to be used in conjunction with .with_criteria(...) function calls when building a solver factory.

This stopping criterion wraps another stopping criterion inside, which only starts getting checked after the first count iterations finished.

Full usage example: Stop when the relative residual norm is below $10^{-10}$ , but with at least 100 iterations.

auto factory = gko::solver::Cg<double>::build()
                   .with_criteria(gko::stop::min_iters(100,
                       gko::stop::relative_residual_norm(1e-10)))
                   .on(exec);

Parameters

count	the number of iterations after which to start checking the inner criterion
criterion	the inner criterion, which will not be checked until `count` iterations finished, afterwards the min_iters stopping criterion behaves like the inner criterion.

Returns: a deferred_factory_parameter that can be passed to the with_criteria function when building a solver.

◆ relative_implicit_residual_norm()

deferred_factory_parameter<CriterionFactory> gko::stop::relative_implicit_residual_norm ( double tolerance )

Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with .with_criteria(...) function calls when building a solver factory.

This stopping criterion will stop the iteration after the residual norm has decreased below the specified value or by the specified amount.

Full usage example: Stop after 100 iterations or when the absolute residual norm is below $10^{-10}$ , whichever happens first.

auto factory = gko::solver::Cg<double>::build()
                   .with_criteria(
                       gko::stop::max_iters(100),
                       gko::stop::absolute_residual_norm(1e-10))
                   .on(exec);

Parameters

tolerance the value the residual norm needs to be below. With residual , initial guess , right-hand side , matrix , absolute means the exact value of the norm , relative means the norm relative to the right-hand side , initial means the norm relative to the initial residual . An implicit stopping criterion is only available with some solvers, and refers to either the energy norm in short-recurrence solvers like Cg or the euclidian norm in solvers like GMRES. Implicit residual norms are cheaper to compute, but may be less precise due to accumulating rounding errors.

Returns: a deferred_factory_parameter that can be passed to the with_criteria function when building a solver.

◆ relative_residual_norm()

deferred_factory_parameter<CriterionFactory> gko::stop::relative_residual_norm ( double tolerance )

Creates the precursor to a ResidualNorm stopping criterion factory, to be used in conjunction with .with_criteria(...) function calls when building a solver factory.

This stopping criterion will stop the iteration after the residual norm has decreased below the specified value or by the specified amount.

Full usage example: Stop after 100 iterations or when the absolute residual norm is below $10^{-10}$ , whichever happens first.

auto factory = gko::solver::Cg<double>::build()
                   .with_criteria(
                       gko::stop::max_iters(100),
                       gko::stop::absolute_residual_norm(1e-10))
                   .on(exec);

Parameters

tolerance the value the residual norm needs to be below. With residual , initial guess , right-hand side , matrix , absolute means the exact value of the norm , relative means the norm relative to the right-hand side , initial means the norm relative to the initial residual . An implicit stopping criterion is only available with some solvers, and refers to either the energy norm in short-recurrence solvers like Cg or the euclidian norm in solvers like GMRES. Implicit residual norms are cheaper to compute, but may be less precise due to accumulating rounding errors.

Returns: a deferred_factory_parameter that can be passed to the with_criteria function when building a solver.

◆ time_limit()

deferred_factory_parameter<Time::Factory> gko::stop::time_limit ( std::chrono::nanoseconds duration )

Creates the precursor to a Time stopping criterion factory, to be used in conjunction with .with_criteria(...) function calls when building a solver factory.

This stopping criterion will stop the iteration after the specified amount of time since the start of the solver run has elapsed.

Full usage example: Stop after 1 second or when the relative residual norm is below $10^{-10}$ , whichever happens first.

auto factory = gko::solver::Cg<double>::build()
                   .with_criteria(
                       gko::stop::time_limit(std::chrono::seconds(1)),
                       gko::stop::relative_residual_norm(1e-10))
                   .on(exec);

Parameters

duration the time limit after which to stop the iteration. Thanks to std::chrono's converting constructors, you can specify any time units, and they will be converted to nanoseconds automatically.

Returns: a deferred_factory_parameter that can be passed to the with_criteria function when building a solver.

Classes

Typedefs

Enumerations

Functions

Detailed Description

Typedef Documentation

◆ EnableDefaultCriterionFactory

Function Documentation

◆ absolute_implicit_residual_norm()

◆ absolute_residual_norm()

◆ initial_implicit_residual_norm()

◆ initial_residual_norm()

◆ max_iters()

◆ min_iters()

◆ relative_implicit_residual_norm()

◆ relative_residual_norm()

◆ time_limit()