This class implements Conjugate Gradient method for monolithic and block linear operators. More...

#include <CgSolver.hpp>

Inheritance diagram for geos::CgSolver< VECTOR >:

Public Types
using	Base = KrylovSolver< VECTOR >
	Alias for base type.

using	Vector = typename Base::Vector
	Alias for template parameter.

Public Types inherited from geos::KrylovSolver< VECTOR >
using	Base = LinearOperator< VECTOR >
	Base type.

using	Vector = typename Base::Vector
	Alias for template parameter.

Public Types inherited from geos::LinearOperator< VECTOR >
using	Vector = VECTOR
	Alias for template parameter.

Public Member Functions
Constructor/Destructor Methods
	CgSolver (LinearSolverParameters params, LinearOperator< Vector > const &A, LinearOperator< Vector > const &M)
	Constructor. More...

KrylovSolver interface
virtual void	solve (Vector const &b, Vector &x) const override final
	Solve preconditioned system. More...

virtual string	methodName () const override final
	Get name of the Krylov subspace method. More...

Public Member Functions inherited from geos::KrylovSolver< VECTOR >
	KrylovSolver (LinearSolverParameters params, LinearOperator< Vector > const &matrix, LinearOperator< Vector > const &precond)
	Constructor. More...

virtual	~KrylovSolver () override=default
	Virtual destructor.

virtual void	apply (Vector const &src, Vector &dst) const override final
	Apply operator to a vector. More...

virtual globalIndex	numGlobalRows () const override final
	Get the number of global rows. More...

virtual globalIndex	numGlobalCols () const override final
	Get the number of global columns. More...

virtual localIndex	numLocalRows () const override final
	Get the number of local rows. More...

virtual localIndex	numLocalCols () const override final
	Get the number of local columns. More...

virtual MPI_Comm	comm () const override final
	Get the MPI communicator the matrix was created with. More...

LinearSolverParameters const &	parameters () const

LinearSolverResult const &	result () const

arrayView1d< real64 const >	history () const
	Get convergence history of a linear solve. More...

Public Member Functions inherited from geos::LinearOperator< VECTOR >
	LinearOperator ()=default
	Constructor.

virtual	~LinearOperator ()=default
	Destructor.

virtual void	residual (Vector const &x, Vector const &b, Vector &r) const
	Compute residual `r = b - this(x)`. More...

Protected Types
using	VectorTemp = typename KrylovSolver< VECTOR >::VectorTemp
	Alias for vector type that can be used for temporaries.

Protected Types inherited from geos::KrylovSolver< VECTOR >
using	VectorTemp = typename VectorStorageHelper< VECTOR >::type
	Alias for vector type that can be used for temporaries.

Protected Member Functions
void	logProgress () const
	Output iteration progress (called by implementations). More...

void	logResult () const
	Output convergence result (called by implementations).

Protected Member Functions inherited from geos::KrylovSolver< VECTOR >
void	logProgress () const
	Output iteration progress (called by implementations). More...

void	logResult () const
	Output convergence result (called by implementations).

Static Protected Member Functions
static VectorTemp	createTempVector (Vector const &src)
	Helper function to create temporary vectors based on a source vector. More...

Static Protected Member Functions inherited from geos::KrylovSolver< VECTOR >
static VectorTemp	createTempVector (Vector const &src)
	Helper function to create temporary vectors based on a source vector. More...

Protected Attributes
LinearSolverParameters	m_params
	parameters of the solver

LinearOperator< Vector > const &	m_operator
	reference to the operator to be solved

LinearOperator< Vector > const &	m_precond
	reference to the preconditioning operator

LinearSolverResult	m_result
	results of a solve

array1d< real64 >	m_residualNorms
	Absolute residual norms at each iteration (if available)

Protected Attributes inherited from geos::KrylovSolver< VECTOR >
LinearSolverParameters	m_params
	parameters of the solver

LinearOperator< Vector > const &	m_operator
	reference to the operator to be solved

LinearOperator< Vector > const &	m_precond
	reference to the preconditioning operator

LinearSolverResult	m_result
	results of a solve

array1d< real64 >	m_residualNorms
	Absolute residual norms at each iteration (if available)

Additional Inherited Members
Static Public Member Functions inherited from geos::KrylovSolver< VECTOR >
static std::unique_ptr< KrylovSolver< VECTOR > >	create (LinearSolverParameters const &parameters, LinearOperator< VECTOR > const &matrix, LinearOperator< VECTOR > const &precond)
	Factory method for instantiating Krylov solver objects. More...

Detailed Description

template<typename VECTOR>
class geos::CgSolver< VECTOR >

This class implements Conjugate Gradient method for monolithic and block linear operators.

Template Parameters

VECTOR type of vectors this solver operates on.

Note: The notation is consistent with "Iterative Methods for Linear and Non-Linear Equations" from C.T. Kelley (1995) and "Iterative Methods for Sparse Linear Systems" from Y. Saad (2003).

Definition at line 38 of file CgSolver.hpp.

Constructor & Destructor Documentation

◆ CgSolver()

template<typename VECTOR >

geos::CgSolver< VECTOR >::CgSolver	(	LinearSolverParameters	params,
		LinearOperator< Vector > const &	A,
		LinearOperator< Vector > const &	M
	)

Constructor.

Parameters

[in]	params	parameters for the solver
[in]	A	reference to the system matrix.
[in]	M	reference to the preconditioning operator.

Member Function Documentation

◆ createTempVector()

template<typename VECTOR >

static VectorTemp geos::KrylovSolver< VECTOR >::createTempVector

inlinestaticprotected

Helper function to create temporary vectors based on a source vector.

Parameters

src	the source vector, whose size and parallel distribution will be used

Returns: the new vector

The main purpose is to deal with BlockVector/View/Wrapper hierarchy.

Definition at line 191 of file KrylovSolver.hpp.

◆ logProgress()

template<typename VECTOR >

void geos::KrylovSolver< VECTOR >::logProgress

protected

Output iteration progress (called by implementations).

Note: must be called after pushing the most recent residual into m_residualNorms

◆ methodName()

template<typename VECTOR >

virtual string geos::CgSolver< VECTOR >::methodName ( ) const

inlinefinaloverridevirtual

Get name of the Krylov subspace method.

Returns: the abbreviated name of the method

Implements geos::KrylovSolver< VECTOR >.

Definition at line 77 of file CgSolver.hpp.

◆ solve()

template<typename VECTOR >

virtual void geos::CgSolver< VECTOR >::solve	(	Vector const &	b,
		Vector &	x
	)		const

finaloverridevirtual

Solve preconditioned system.

Parameters

[in]	b	system right hand side.
[in,out]	x	system solution (input = initial guess, output = solution).

Implements geos::KrylovSolver< VECTOR >.

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

coreComponents/linearAlgebra/solvers/CgSolver.hpp

Public Types

Public Member Functions

Protected Types

Protected Member Functions

Static Protected Member Functions

Protected Attributes

Additional Inherited Members

Detailed Description

template<typename VECTOR> class geos::CgSolver< VECTOR >

Constructor & Destructor Documentation

◆ CgSolver()

Member Function Documentation

◆ createTempVector()

◆ logProgress()

◆ methodName()

◆ solve()

template<typename VECTOR>
class geos::CgSolver< VECTOR >