#include <cl_SolverMUMPS.hpp>

Inheritance diagram for belfem::solver::MUMPS:

Collaboration diagram for belfem::solver::MUMPS:

Public Member Functions
	MUMPS (const SolverParameters *aParams, const proc_t aMasterRank=0)
	~MUMPS () override
bool	supports_factorization_reuse () const override
void	freeze_factorization (const SpMatrix &aMatrix) override
	arm the frozen state: subsequent solves reuse the factorization that the LAST SUCCESSFUL solve built, instead of rebuilding it.
void	unfreeze_factorization () override
	drop the frozen state and return to ordinary re-factorizing solves.
bool	factorization_is_frozen () const override
	true while the frozen scope is armed
void	solve (SpMatrix &aMatrix, Vector< real > &aLHS, Vector< real > &aRHS) override
void	solve (SpMatrix &aMatrix, Matrix< real > &aLHS, Matrix< real > &aRHS) override
void	free () override
void	set_reordering (const MumpsSerialReodrdering aSerial, const MumpsParallelReodrdering aParallel)
void	set_block_low_ranking (const MumpsBlockLowRanking aBLK, const real aEpsilon)
void	set_error_analysis (const MumpsErrorAnalysis aSetting)
real	get_determinant () const override
	returns the determinant, if supported by the solver and computation was requested
real	get_cond1 () const override
	returns the conditioning number, if supported by the solver and computation was requested
real	get_cond2 () const override
	returns the conditioning number, if supported by the solver and computation was requested
real	get_forward_error () const override
	RINFOG( 9 ): estimated forward error of the last solve.
real	get_backward_error () const override
	RINFOG( 7 ) + RINFOG( 8 ): the componentwise backward error.
real	get_omega2 () const override
	RINFOG( 8 ) alone: omega2.
Public Member Functions inherited from belfem::solver::Wrapper
	Wrapper (const string &aLabel, const bool aUsesMPI)
virtual	~Wrapper ()
const string &	label () const
	returns the name of the solver as string
void	set_soft_fail (const bool aFlag)
	arm/disarm the soft-fail contract ( see the member note )
bool	failed () const
	true if the last factorization/solve failed softly
void	clear_failure ()
	clear a recorded soft failure before the next solve
bool	is_initialized () const
	tells if the initialitzation routine has already been called
proc_t	rank () const
	returns the communication rank of this proc
proc_t	comm_size () const
	returns the communication size of this proc
bool	uses_mpi () const
void	free () override
Public Member Functions inherited from belfem::CommunicationObject
	CommunicationObject ()
	~CommunicationObject ()
index_t	index () const

Protected Member Functions
void	initialize (SpMatrix &aMatrix, const SymmetryMode aSymmetryMode=SymmetryMode::Unsymmetric, const int_t aNumRhsColumns=1) override
string	error_message (const int_t *aInfo, const int_t &aN, const int_t &aNNZ)
void	warning_message (const int_t *aInfo, Cell< string > &aWarnings)
Protected Member Functions inherited from belfem::solver::Wrapper
bool	soft_fail () const
	subclass hooks for the soft-fail contract
void	flag_failure ()
virtual void	initialize ()
void	hatch_turtle ()
	print a warning ( the turtle ) when the MPI ranks sharing a node ask for more OpenMP threads than the node has PHYSICAL cores.
void	mat2vec (const Matrix< real > &aM, Vector< real > &aV)
void	vec2mat (const Vector< real > &aV, Matrix< real > &aM)
SpMatrix *	matrix ()
Vector< real > &	x ()
Vector< real > &	y ()

Additional Inherited Members
Protected Attributes inherited from belfem::solver::Wrapper
SpMatrix *	mMatrix = nullptr
Vector< real > *	mX = nullptr
Vector< real > *	mY = nullptr

Constructor & Destructor Documentation

◆ MUMPS()

belfem::solver::MUMPS::MUMPS	(	const SolverParameters *	aParams,
		const proc_t	aMasterRank = 0 )

◆ ~MUMPS()

belfem::solver::MUMPS::~MUMPS ( )

override

Member Function Documentation

◆ error_message()

string belfem::solver::MUMPS::error_message	(	const int_t *	aInfo,
		const int_t &	aN,
		const int_t &	aNNZ )

protected

◆ factorization_is_frozen()

bool belfem::solver::MUMPS::factorization_is_frozen ( ) const

overridevirtual

true while the frozen scope is armed

Reimplemented from belfem::solver::Wrapper.

◆ free()

void belfem::solver::MUMPS::free ( )

overridevirtual

Reimplemented from belfem::CommunicationObject.

◆ freeze_factorization()

void belfem::solver::MUMPS::freeze_factorization ( const SpMatrix & aMatrix )

overridevirtual

arm the frozen state: subsequent solves reuse the factorization that the LAST SUCCESSFUL solve built, instead of rebuilding it.

This is a scope, not a hint, and the caller owns the whole of it:

solve( A, x, b ) <- builds the factorization freeze_factorization( A ) <- arm, only after that succeeded solve( A, y, c ) ... <- reuse, values of A must NOT change unfreeze_factorization() <- always, including on every failure

Why a scope rather than a flag: if the values of A change while frozen, the solver returns a valid-LOOKING solution of the wrong matrix, with no error anywhere. Pointer identity cannot detect that – the wrapper already treats the same pointer as "same structure, different values" and deliberately re-factorizes for exactly this reason. So the contract is that NO assembly may happen inside the scope, the caller guarantees it, and the wrapper checks everything about the matrix it cheaply can ( structure and identity, not an O( nnz ) values scan on every solve ).

Checks are ALWAYS-ACTIVE errors rather than assertions: a silent stale-factor solve is precisely the failure that must not survive into a release build.

Rank behaviour: only the rank that owns the matrix can record its identity – on every other rank the matrix argument is a local submatrix or empty, and the distributed solvers never read it. So a worker records only that it is frozen. Uniformity comes from the caller arming and dropping the scope at the same point of the same code path on every rank, never from a rank-local decision.

The default implementation throws: a wrapper that does not support reuse must be asked via supports_factorization_reuse() first

Reimplemented from belfem::solver::Wrapper.

◆ get_backward_error()

real belfem::solver::MUMPS::get_backward_error ( ) const

overridevirtual

RINFOG( 7 ) + RINFOG( 8 ): the componentwise backward error.

MUMPS reports it in two halves, split by whether a row's denominator is trustworthy, and they sum

Reimplemented from belfem::solver::Wrapper.

◆ get_cond1()

real belfem::solver::MUMPS::get_cond1 ( ) const

overridevirtual

returns the conditioning number, if supported by the solver and computation was requested

Reimplemented from belfem::solver::Wrapper.

◆ get_cond2()

real belfem::solver::MUMPS::get_cond2 ( ) const

overridevirtual

returns the conditioning number, if supported by the solver and computation was requested

Reimplemented from belfem::solver::Wrapper.

◆ get_determinant()

real belfem::solver::MUMPS::get_determinant ( ) const

overridevirtual

returns the determinant, if supported by the solver and computation was requested

Reimplemented from belfem::solver::Wrapper.

◆ get_forward_error()

real belfem::solver::MUMPS::get_forward_error ( ) const

overridevirtual

RINFOG( 9 ): estimated forward error of the last solve.

Reimplemented from belfem::solver::Wrapper.

◆ get_omega2()

real belfem::solver::MUMPS::get_omega2 ( ) const

overridevirtual

RINFOG( 8 ) alone: omega2.

Zero means COND2's term drops out of the forward error. It does NOT prove MUMPS skipped the COND2 estimator – see the implementation for why the two differ

Reimplemented from belfem::solver::Wrapper.

◆ initialize()

void belfem::solver::MUMPS::initialize	(	SpMatrix &	aMatrix,
		const SymmetryMode	aSymmetryMode = SymmetryMode::Unsymmetric,
		const int_t	aNumRhsColumns = 1 )

overrideprotectedvirtual

Reimplemented from belfem::solver::Wrapper.

◆ set_block_low_ranking()

void belfem::solver::MUMPS::set_block_low_ranking	(	const MumpsBlockLowRanking	aBLK,
		const real	aEpsilon )

◆ set_error_analysis()

void belfem::solver::MUMPS::set_error_analysis ( const MumpsErrorAnalysis aSetting )

◆ set_reordering()

void belfem::solver::MUMPS::set_reordering	(	const MumpsSerialReodrdering	aSerial,
		const MumpsParallelReodrdering	aParallel )

◆ solve() [1/2]

void belfem::solver::MUMPS::solve	(	SpMatrix &	aMatrix,
		Matrix< real > &	aLHS,
		Matrix< real > &	aRHS )

overridevirtual

Reimplemented from belfem::solver::Wrapper.

◆ solve() [2/2]

void belfem::solver::MUMPS::solve	(	SpMatrix &	aMatrix,
		Vector< real > &	aLHS,
		Vector< real > &	aRHS )

overridevirtual

Reimplemented from belfem::solver::Wrapper.

◆ supports_factorization_reuse()

bool belfem::solver::MUMPS::supports_factorization_reuse ( ) const

overridevirtual

MUMPS keeps its factors in the saved DMUMPS_STRUC until JOB -2, so solving against them is a matter of asking for JOB 3

Reimplemented from belfem::solver::Wrapper.

◆ unfreeze_factorization()

void belfem::solver::MUMPS::unfreeze_factorization ( )

overridevirtual

drop the frozen state and return to ordinary re-factorizing solves.

Safe to call when not frozen, so a scope guard can call it unconditionally on the way out

Reimplemented from belfem::solver::Wrapper.

◆ warning_message()

void belfem::solver::MUMPS::warning_message	(	const int_t *	aInfo,
		Cell< string > &	aWarnings )

protected

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

src/sparse/cl_SolverMUMPS.hpp
src/sparse/cl_SolverMUMPS.cpp

Public Member Functions

Protected Member Functions

Additional Inherited Members

Constructor & Destructor Documentation

◆ MUMPS()

◆ ~MUMPS()

Member Function Documentation

◆ error_message()

◆ factorization_is_frozen()

◆ free()

◆ freeze_factorization()

◆ get_backward_error()

◆ get_cond1()

◆ get_cond2()

◆ get_determinant()

◆ get_forward_error()

◆ get_omega2()

◆ initialize()

◆ set_block_low_ranking()

◆ set_error_analysis()

◆ set_reordering()

◆ solve() [1/2]

◆ solve() [2/2]

◆ supports_factorization_reuse()

◆ unfreeze_factorization()

◆ warning_message()