MUMPS : MUltifrontal Massively Parallel sparse direct Solver. More...

#include <mumpsoverlap.hh>

Inheritance diagram for concepts::MumpsOverlap< F >:

Public Types
typedef Cmplxtype< F >::type	c_type
	Real type of data type. More...

typedef Realtype< F >::type	r_type
	Real type of data type. More...

typedef F	type
	Type of data, e.g. matrix entries. More...

Public Member Functions
virtual void	apply_ ()
	Solver (call this solver on all other threads) More...

virtual void	apply_ (const Vector< F > &fncY, Vector< F > &fncX)
	Solver (call this solver on thread rank=0 (master)) More...

virtual const uint	dimX () const
	Returns the size of the image space of the operator (number of rows of the corresponding matrix) More...

virtual const uint	dimY () const
	Returns the size of the source space of the operator (number of columns of the corresponding matrix) More...

virtual void	hide_messages ()
	Function to hide solver messages. More...

	MumpsOverlap (const int NGlob, const Operator< F > &A, const int *LocToGLob, const int rank, const int comm=-987654, const bool CConvention=true)
	Constructor. More...

	MumpsOverlap (const Operator< F > &A, const int rank, const int comm=-987654, const bool CConvention=true)
	Constructor. More...

void	operator() ()
	Application method without second argument. Used for parallel solvers. More...

virtual void	operator() (const Function< c_type > &fncY, Function< c_type > &fncX)
	Application operator for complex function `fncY`. More...

virtual void	operator() (const Function< r_type > &fncY, Function< F > &fncX)
	Application operator for real function `fncY`. More...

void	operator() (const Matrix< c_type > &mX, Matrix< c_type > &mY)
	Application method to complex matrices. Calls apply_() More...

void	operator() (const Matrix< r_type > &mX, Matrix< F > &mY)
	Application method to real matrices. Calls function apply() More...

virtual void	operator() (const Vector< c_type > &fncY, Vector< c_type > &fncX)
	Application operator for complex function `fncY`. More...

virtual void	operator() (const Vector< r_type > &fncY, Vector< F > &fncX)
	Application operator for real vector `fncY`. More...

virtual void	show_messages ()
	Function to show solver messages. More...

virtual	~MumpsOverlap ()
	Destructor. More...

Protected Member Functions
virtual std::ostream &	info (std::ostream &) const

Protected Attributes
uint	dimX_
	Dimension of image space and the source space. More...

uint	dimY_

Private Attributes
void *	A_loc_
	The matrix of the operator in a special sparse notation (local). More...

int	comm_
	Get solver communicator (global) More...

void *	id_mumps
	Mumps internal state (used for the solver) More...

int *	irn_loc_
	Array of row indexes in a special sparse notation (local). More...

int *	jcn_loc_
	Array of column indexes in a special sparse notation (local). More...

uint	n_
	Dimension of the matrix (global) More...

uint	nz_loc_
	Number of non-zero elements (local) More...

int	rank_
	Get processor rank (local) More...

Detailed Description

template<class F>
class concepts::MumpsOverlap< F >

MUMPS : MUltifrontal Massively Parallel sparse direct Solver.

MUMPS is a general purpose library for the direct solution of large, sparse, nonsymmetric systems of linear equations on high perormances machines, for both real and complex matrices. The library routines will perform an partial factorization and Schur complement matrix (centralized or 2D block-cyclic).

MUMPS offers different ordering options for a column permutation .

MUMPS has also parallel implementation.

Warning: If MUMPS was compiled with MPI, then Concepts should be compiled with MPI as well (even if using one single node during the execution). As well, the C++ main function code should start with
MPI::Init(argc, argv);

and end with
MPI::Finalize();

return 0;

and the python code should start with
from mpi4py import MPI

Author: Patrick Amestoy; Alfredo Buttari; Abdou Guermouche; Jean-Yves L'Excellent; Bora Ucar

See also: homepage of MUMPS; P.R. Amestoy, I.S. Duff and J.-Y. L'Excellent, Multifrontal parallel distributed symmetric and unsymmetric solvers.Revised version appeared in Comput. Methods in Appl. Mech. Eng., 184, 501-520 (2000).

Examples: parallelizationTutorial.cc.

Definition at line 73 of file mumpsoverlap.hh.

Member Typedef Documentation

◆ c_type

template<class F >

typedef Cmplxtype<F>::type concepts::VecOperator< F >::c_type

inherited

Real type of data type.

Definition at line 120 of file compositions.hh.

◆ r_type

template<class F >

typedef Realtype<F>::type concepts::VecOperator< F >::r_type

inherited

Real type of data type.

Definition at line 118 of file compositions.hh.

◆ type

template<class F >

typedef F concepts::Operator< F >::type

inherited

Type of data, e.g. matrix entries.

Definition at line 45 of file compositions.hh.

Constructor & Destructor Documentation

◆ MumpsOverlap() [1/2]

template<class F >

concepts::MumpsOverlap< F >::MumpsOverlap	(	const int	NGlob,
		const Operator< F > &	A,
		const int *	LocToGLob,
		const int	rank,
		const int	comm = `-987654`,
		const bool	CConvention = `true`
	)

Constructor.

Parameters

NGlob	global number of degrees of freedom
A	Local part of the operator which should be inverted. Everything which is needed from `A` is copied in the constructor.
LocToGLob	Vector that contains local to global numbering
rank	Current thread rank.
comm	MUMPS internal communicator
CConvention	Define if

◆ MumpsOverlap() [2/2]

template<class F >

concepts::MumpsOverlap< F >::MumpsOverlap	(	const Operator< F > &	A,
		const int	rank,
		const int	comm = `-987654`,
		const bool	CConvention = `true`
	)

Constructor.

Parameters

A	Local part of the operator which should be inverted (of size NGlob times NGlob). Everything which is needed from `A` is copied in the constructor. We suppose that A is already scaled to the global matrix problem.
rank	Current thread rank.
comm	MUMPS internal communicator
CConvention	Define if

◆ ~MumpsOverlap()

template<class F >

virtual concepts::MumpsOverlap< F >::~MumpsOverlap ( )

virtual

Destructor.

Member Function Documentation

◆ apply_() [1/2]

template<class F >

virtual void concepts::MumpsOverlap< F >::apply_ ( )

virtual

Solver (call this solver on all other threads)

Implements concepts::VecOperator< F >.

◆ apply_() [2/2]

template<class F >

virtual void concepts::MumpsOverlap< F >::apply_	(	const Vector< F > &	fncY,
		Vector< F > &	fncX
	)

virtual

Solver (call this solver on thread rank=0 (master))

Implements concepts::VecOperator< F >.

◆ dimX()

template<class F >

virtual const uint concepts::Operator< F >::dimX ( ) const

inlinevirtualinherited

Returns the size of the image space of the operator (number of rows of the corresponding matrix)

Definition at line 93 of file compositions.hh.

◆ dimY()

template<class F >

virtual const uint concepts::Operator< F >::dimY ( ) const

inlinevirtualinherited

Returns the size of the source space of the operator (number of columns of the corresponding matrix)

Definition at line 98 of file compositions.hh.

◆ hide_messages()

template<class F >

virtual void concepts::MumpsOverlap< F >::hide_messages ( )

virtual

Function to hide solver messages.

◆ info()

template<class F >

virtual std::ostream& concepts::MumpsOverlap< F >::info ( std::ostream & ) const

protectedvirtual

Reimplemented from concepts::VecOperator< F >.

◆ operator()() [1/7]

template<class F >

void concepts::VecOperator< F >::operator() ( )

virtualinherited

Application method without second argument. Used for parallel solvers.

Reimplemented from concepts::Operator< F >.

◆ operator()() [2/7]

template<class F >

virtual void concepts::VecOperator< F >::operator()	(	const Function< c_type > &	fncY,
		Function< c_type > &	fncX
	)

virtualinherited

Application operator for complex function fncY.

Computes fncX = A(fncY) where A is this operator. fncX becomes complex.

In derived classes its enough to implement the operator() for complex Operator's. If a real counterpart is not implemented, the function fncY is splitted into real and imaginary part and the application operator for real functions is called for each. Then the result is combined.

If in a derived class the operator() for complex Operator's is not implemented, a exception is thrown from here.

Reimplemented from concepts::Operator< F >.

◆ operator()() [3/7]

template<class F >

virtual void concepts::VecOperator< F >::operator()	(	const Function< r_type > &	fncY,
		Function< F > &	fncX
	)

virtualinherited

Application operator for real function fncY.

Computes fncX = A(fncY) where A is this operator.

fncX becomes the type of the operator, for real data it becomes real, for complex data it becomes complex.

In derived classes its enough to implement the operator() for real Operator's. If a complex counterpart is not implemented, the function fncY is transformed to a complex function and then the application operator for complex functions is called.

If in a derived class the operator() for real Operator's is not implemented, a exception is thrown from here.

Reimplemented from concepts::Operator< F >.

◆ operator()() [4/7]

template<class F >

void concepts::VecOperator< F >::operator()	(	const Matrix< c_type > &	mX,
		Matrix< c_type > &	mY
	)

inherited

Application method to complex matrices. Calls apply_()

◆ operator()() [5/7]

template<class F >

void concepts::VecOperator< F >::operator()	(	const Matrix< r_type > &	mX,
		Matrix< F > &	mY
	)

inherited

Application method to real matrices. Calls function apply()

◆ operator()() [6/7]

template<class F >

virtual void concepts::VecOperator< F >::operator()	(	const Vector< c_type > &	fncY,
		Vector< c_type > &	fncX
	)

virtualinherited

Application operator for complex function fncY.

Computes fncX = A(fncY) where A is this operator. fncX becomes complex.

In derived classes its enough to implement the operator() for complex Operator's. If a real counterpart is not implemented, the vector fncY is splitted into real and imaginary part and the application operator for real vectors is called for each. Then the result is combined

If in a derived class the operator() for complex Operator's i not implemented, a exception is thrown from here.

◆ operator()() [7/7]

template<class F >

virtual void concepts::VecOperator< F >::operator()	(	const Vector< r_type > &	fncY,
		Vector< F > &	fncX
	)

virtualinherited

Application operator for real vector fncY.

Computes fncX = A(fncY) where A is this operator.

Type of fncX becomes that of the operator, for real data it becomes real, for complex data it becomes complex.

In derived classes its enough to implement the operator() for real Operator's. If a complex counterpart is not implemented, the vector fncY is transformed to a complex vector and then the application for complex vectors is called.

If in a derived class the operator() for real Operator's is not implemented, a exception is thrown from here.

◆ show_messages()

template<class F >

virtual void concepts::MumpsOverlap< F >::show_messages ( )

virtual

Function to show solver messages.

Reimplemented from concepts::Operator< F >.

Member Data Documentation

◆ A_loc_

template<class F >

void* concepts::MumpsOverlap< F >::A_loc_

private

The matrix of the operator in a special sparse notation (local).

Definition at line 119 of file mumpsoverlap.hh.

◆ comm_

template<class F >

int concepts::MumpsOverlap< F >::comm_

private

Get solver communicator (global)

Definition at line 131 of file mumpsoverlap.hh.

◆ dimX_

template<class F >

uint concepts::Operator< F >::dimX_

protectedinherited

Dimension of image space and the source space.

Definition at line 104 of file compositions.hh.

◆ dimY_

template<class F >

uint concepts::Operator< F >::dimY_

protectedinherited

Definition at line 104 of file compositions.hh.

◆ id_mumps

template<class F >

void* concepts::MumpsOverlap< F >::id_mumps

private

Mumps internal state (used for the solver)

Definition at line 133 of file mumpsoverlap.hh.

◆ irn_loc_

template<class F >

int* concepts::MumpsOverlap< F >::irn_loc_

private

Array of row indexes in a special sparse notation (local).

Definition at line 121 of file mumpsoverlap.hh.

◆ jcn_loc_

template<class F >

int* concepts::MumpsOverlap< F >::jcn_loc_

private

Array of column indexes in a special sparse notation (local).

Definition at line 123 of file mumpsoverlap.hh.

◆ n_

template<class F >

uint concepts::MumpsOverlap< F >::n_

private

Dimension of the matrix (global)

Definition at line 125 of file mumpsoverlap.hh.

◆ nz_loc_

template<class F >

uint concepts::MumpsOverlap< F >::nz_loc_

private

Number of non-zero elements (local)

Definition at line 127 of file mumpsoverlap.hh.

◆ rank_

template<class F >

int concepts::MumpsOverlap< F >::rank_

private

Get processor rank (local)

Definition at line 129 of file mumpsoverlap.hh.

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

operator/mumpsoverlap.hh

concepts::MumpsOverlap< F > Class Template ReferenceLinear Solvers

Public Types

Public Member Functions

Protected Member Functions

Protected Attributes

Private Attributes

Detailed Description

template<class F> class concepts::MumpsOverlap< F >

Member Typedef Documentation

◆ c_type

◆ r_type

◆ type

Constructor & Destructor Documentation

◆ MumpsOverlap() [1/2]

◆ MumpsOverlap() [2/2]

◆ ~MumpsOverlap()

Member Function Documentation

◆ apply_() [1/2]

◆ apply_() [2/2]

◆ dimX()

◆ dimY()

◆ hide_messages()

◆ info()

◆ operator()() [1/7]

◆ operator()() [2/7]

◆ operator()() [3/7]

◆ operator()() [4/7]

◆ operator()() [5/7]

◆ operator()() [6/7]

◆ operator()() [7/7]

◆ show_messages()

Member Data Documentation

◆ A_loc_

◆ comm_

◆ dimX_

◆ dimY_

◆ id_mumps

◆ irn_loc_

◆ jcn_loc_

◆ n_

◆ nz_loc_

◆ rank_

concepts::MumpsOverlap< F > Class Template Reference
Linear Solvers

template<class F>
class concepts::MumpsOverlap< F >