The SplitSystemSolver class allows to solve a split linear system. More...

Inheritance diagram for bitpit::SplitSystemSolver:

[legend]

Collaboration diagram for bitpit::SplitSystemSolver:

[legend]

Public Types
typedef SplitSystemMatrixAssembler	Assembler

typedef SystemSplitStrategy	splitStrategy

Public Types inherited from bitpit::SystemSolver
typedef SystemMatrixAssembler	Assembler

enum	FileFormat { FILE_BINARY , FILE_ASCII }

Public Member Functions
	SplitSystemSolver (bool debug=false)

	SplitSystemSolver (bool flatten, bool transpose, bool debug)

	SplitSystemSolver (bool transpose, bool debug)

	SplitSystemSolver (const std::string &prefix, bool debug=false)

	SplitSystemSolver (const std::string &prefix, bool flatten, bool transpose, bool debug)

	SplitSystemSolver (const std::string &prefix, bool transpose, bool debug)

void	assembly (const Assembler &assembler)

void	assembly (const Assembler &assembler, const SystemMatrixOrdering &reordering)

void	assembly (const SparseMatrix &matrix, SystemSplitStrategy splitStrategy, const std::vector< int > &splitSizes)

void	assembly (const SparseMatrix &matrix, SystemSplitStrategy splitStrategy, const std::vector< int > &splitSizes, const SystemMatrixOrdering &reordering)

void	exportMatrix (const std::string &filePath, FileFormat exportFormat=FILE_BINARY) const override

int	getBlockSize () const override

int	getSplitCount () const

KSPOptions &	getSplitKSPOptions (int split)

const KSPOptions &	getSplitKSPOptions (int split) const

const KSPStatus &	getSplitKSPStatus (int split) const

std::vector< int >	getSplitOffsets () const

std::vector< int >	getSplitSizes () const

SystemSplitStrategy	getSplitStrategy () const

	SystemSolver (bool debug=false)

	SystemSolver (bool flatten, bool transpose, bool debug)

	SystemSolver (bool transpose, bool debug)

	SystemSolver (const std::string &prefix, bool debug=false)

	SystemSolver (const std::string &prefix, bool flatten, bool transpose, bool debug)

	SystemSolver (const std::string &prefix, bool transpose, bool debug)

void	update (const Assembler &assembler)

void	update (const SparseMatrix &elements)

void	update (long nRows, const long *rows, const Assembler &assembler)

void	update (long nRows, const long *rows, const SparseMatrix &elements)

Public Member Functions inherited from bitpit::SystemSolver
	SystemSolver (bool debug=false)

	SystemSolver (bool flatten, bool transpose, bool debug)

	SystemSolver (bool transpose, bool debug)

	SystemSolver (const std::string &prefix, bool debug=false)

	SystemSolver (const std::string &prefix, bool flatten, bool transpose, bool debug)

	SystemSolver (const std::string &prefix, bool transpose, bool debug)

virtual	~SystemSolver ()

void	assembly (const Assembler &assembler)

void	assembly (const Assembler &assembler, const SystemMatrixOrdering &reordering)

void	assembly (const SparseMatrix &matrix)

void	assembly (const SparseMatrix &matrix, const SystemMatrixOrdering &reordering)

virtual void	clear ()

virtual void	clearWorkspace ()

void	dumpSystem (const std::string &header, const std::string &directory, const std::string &prefix="") const

void	enableForceConsistency (bool enable)

virtual void	exportRHS (const std::string &filePath, FileFormat exportFormat=FILE_BINARY) const

virtual void	exportSolution (const std::string &filePath, FileFormat exportFormat=FILE_BINARY) const

long	getColCount () const

long	getColElementCount () const

long	getColGlobalCount () const

long	getColGlobalElementCount () const

KSPOptions &	getKSPOptions ()

const KSPOptions &	getKSPOptions () const

const KSPStatus &	getKSPStatus () const

double *	getRHSRawPtr ()

const double *	getRHSRawPtr () const

const double *	getRHSRawReadPtr () const

long	getRowCount () const

long	getRowElementCount () const

long	getRowGlobalCount () const

long	getRowGlobalElementCount () const

double *	getSolutionRawPtr ()

const double *	getSolutionRawPtr () const

const double *	getSolutionRawReadPtr () const

bool	getTranspose () const

virtual void	importMatrix (const std::string &filePath)

virtual void	importRHS (const std::string &filePath)

virtual void	importSolution (const std::string &filePath)

bool	isAssembled () const

bool	isForceConsistencyEnabled () const

bool	isPartitioned () const

bool	isSetUp () const

void	restoreRHSRawPtr (double *raw_rhs)

void	restoreRHSRawReadPtr (const double *raw_rhs) const

void	restoreSolutionRawPtr (double *raw_solution)

void	restoreSolutionRawReadPtr (const double *raw_solution) const

void	restoreSystem (MPI_Comm communicator, bool redistribute, const std::string &directory, const std::string &prefix="")

virtual void	setNullSpace ()

void	setTranspose (bool transpose)

void	setUp ()

void	solve ()

void	solve (const std::vector< double > &rhs, std::vector< double > *solution)

void	unsetNullSpace ()

void	update (const Assembler &assembler)

void	update (const SparseMatrix &elements)

void	update (long nRows, const long *rows, const Assembler &assembler)

void	update (long nRows, const long *rows, const SparseMatrix &elements)

Protected Member Functions
void	assembly (const Assembler &assembler)

void	assembly (const Assembler &assembler, const SystemMatrixOrdering &reordering)

void	assembly (const SparseMatrix &matrix)

void	assembly (const SparseMatrix &matrix, const SystemMatrixOrdering &reordering)

template<typename DerivedSystemSolver >
void	assembly (const typename DerivedSystemSolver::Assembler &assembler, const SystemMatrixOrdering &reordering)

void	destroyKSPOptions () override

void	destroyKSPStatus () override

virtual void	destroySplitKSPOptions ()

virtual void	destroySplitKSPStatuses ()

void	dumpInfo (std::ostream &systemStream) const override

virtual void	exportMatrix (const std::string &filePath, FileFormat exportFormat=FILE_BINARY) const

void	exportMatrix (Mat matrix, const std::string &filePath, FileFormat fileFormat) const

virtual void	fillKSPStatus ()

void	fillKSPStatus () override

virtual void	fillKSPStatus (KSP ksp, KSPStatus *status) const

virtual void	fillSplitKSPStatuses ()

void	generateSplitIndexes (int split, long nItems, std::vector< std::size_t > *indexes) const

std::string	generateSplitPath (const std::string &path, const std::string &index) const

std::string	generateSplitPath (const std::string &path, int i) const

std::string	generateSplitPath (const std::string &path, int i, int j) const

void	generateSplitPermutation (long nItems, long itemGlobalOffset, IS *splitReordering) const

void	initializeKSPOptions () override

void	initializeKSPStatus () override

virtual void	initializeSplitKSPOptions ()

virtual void	initializeSplitKSPStatuses ()

void	matrixAssembly (const Assembler &assembler)

void	matrixDestroy () override

void	matrixDump (std::ostream &systemStream, const std::string &directory, const std::string &prefix) const override

void	matrixFill (const std::string &filePath) override

void	matrixRestore (std::istream &systemStream, const std::string &directory, const std::string &prefix, bool redistribute) override

void	matrixUpdate (long nRows, const long *rows, const Assembler &assembler)

void	postKSPSolveActions () override

void	preKrylovSetupActions () override

virtual void	resetKSPOptions (KSPOptions *options) const

virtual void	resetKSPStatus ()

void	resetKSPStatus () override

virtual void	resetKSPStatus (KSPStatus *status) const

virtual void	resetSplitKSPStatuses ()

void	restoreInfo (std::istream &systemStream) override

virtual void	setupKrylov ()

void	setupKrylov () override

virtual void	setupKrylov (KSP ksp, const KSPOptions &options) const

virtual void	setupPreconditioner ()

void	setupPreconditioner () override

virtual void	setupPreconditioner (PC pc, const KSPOptions &options) const

virtual void	setupSplitKrylovs ()

virtual void	setupSplitPreconditioners ()

void	update (const Assembler &assembler)

void	update (const SparseMatrix &elements)

void	update (long nRows, const long *rows, const Assembler &assembler)

void	update (long nRows, const long *rows, const SparseMatrix &elements)

template<typename DerivedSystemSolver >
void	update (long nRows, const long *rows, const typename DerivedSystemSolver::Assembler &assembler)

void	vectorsCreate () override

void	vectorsCreateSplitPermutations ()

void	vectorsDestroy () override

void	vectorsReorder (bool invert) override

void	vectorsRestore (std::istream &systemStream, const std::string &directory, const std::string &prefix) override

Protected Member Functions inherited from bitpit::SystemSolver
template<typename DerivedSystemSolver >
void	assembly (const typename DerivedSystemSolver::Assembler &assembler, const SystemMatrixOrdering &reordering)

void	clearReordering ()

virtual void	createKSP ()

void	createMatrix (int rowBlockSize, int colBlockSize, int nNestRows, int nNestCols, Mat subMatrices, Mat matrix) const

void	createMatrix (int rowBlockSize, int colBlockSize, Mat *matrix) const

void	createVector (int blockSize, int nestSize, Vec subVectors, Vec vector) const

void	createVector (int blockSize, Vec *vector) const

virtual void	destroyKSP ()

void	destroyMatrix (Mat *matrix) const

void	destroyVector (Vec *vector) const

void	dumpMatrix (Mat matrix, const std::string &directory, const std::string &name) const

void	dumpVector (Vec vector, const std::string &directory, const std::string &name) const

void	dumpVector (Vec vector, VectorSide side, const std::string &directory, const std::string &name) const

void	exportMatrix (Mat matrix, const std::string &filePath, FileFormat fileFormat) const

void	exportVector (Vec vector, const std::string &filePath, FileFormat fileFormat) const

void	exportVector (Vec vector, std::vector< double > *data) const

void	fillMatrix (Mat matrix, const std::string &filePath) const

void	fillVector (Vec vector, const std::string &filePath) const

void	fillVector (Vec vector, const std::vector< double > &data) const

virtual void	finalizeKSP ()

const MPI_Comm &	getCommunicator () const

std::string	getDataFilePath (const std::string &directory, const std::string &name) const

virtual int	getDumpVersion () const

std::string	getFilePath (const std::string &directory, const std::string &name) const

std::string	getInfoFilePath (const std::string &directory, const std::string &name) const

void	matrixAssembly (const Assembler &assembler)

void	matrixUpdate (long nRows, const long *rows, const Assembler &assembler)

virtual void	postKrylovSetupActions ()

virtual void	postPreconditionerSetupActions ()

virtual void	preKSPSolveActions ()

virtual void	prepareKSP ()

virtual void	prePreconditionerSetupActions ()

void	reorderVector (Vec vector, IS permutations, bool invert) const

void	restoreMatrix (const std::string &directory, const std::string &name, bool redistribute, Mat *matrix) const

void	restoreVector (const std::string &directory, const std::string &name, bool redistribute, Vec *vector) const

void	restoreVector (const std::string &directory, const std::string &name, Mat matrix, VectorSide side, Vec *vector) const

void	restoreVector (const std::string &directory, const std::string &name, std::size_t localSize, std::size_t globalSize, Vec *vector) const

void	setReordering (long nRows, long nCols, const SystemMatrixOrdering &reordering)

virtual void	solveKSP ()

template<typename DerivedSystemSolver >
void	update (long nRows, const long *rows, const typename DerivedSystemSolver::Assembler &assembler)

virtual void	vectorsDump (std::ostream &systemStream, const std::string &directory, const std::string &prefix) const

virtual void	vectorsFill (const std::string &rhsFilePath, const std::string &solutionFilePath)

virtual void	vectorsFill (const std::vector< double > &rhs, const std::vector< double > &solution)

Protected Attributes
std::vector< Mat >	m_splitAs

std::vector< KSPOptions >	m_splitKSPOptions

std::vector< KSPStatus >	m_splitKSPStatuses

SystemSplitStrategy	m_splitStrategy

Protected Attributes inherited from bitpit::SystemSolver
Mat	m_A

IS	m_colReordering

bool	m_convergenceMonitorEnabled

bool	m_flatten

KSP	m_KSP

bool	m_KSPDirty

KSPOptions	m_KSPOptions

KSPStatus	m_KSPStatus

Vec	m_rhs

IS	m_rowReordering

Vec	m_solution

bool	m_transpose

Friends
class	SystemSolver

Additional Inherited Members
Protected Types inherited from bitpit::SystemSolver
enum	VectorSide { VECTOR_SIDE_RIGHT , VECTOR_SIDE_LEFT }

Detailed Description

The SplitSystemSolver class allows to solve a split linear system.

Block matrices represent an important class of problems in numerical linear algebra and offer the possibility of far more efficient iterative solvers than just treating the entire matrix as black box. Following the common linear algebra definition, a block matrix is a matrix divided in a small, problem-size independent (two, three or so) number of very large blocks. These blocks arise naturally from the underlying physics or discretization of the problem, for example, they may be associated with different variables of a physical problem. Under a certain numbering of unknowns the matrix can be written as

                     [ A00  A01  A02 ]
                     [ A10  A11  A12 ]
                     [ A20  A21  A22 ]

where each A_ij is an entire block (see the paragraph "Solving Block Matrices" in the PETSc manual).

When assembling the matrix, a monolithic matrix should be provided. For example, assuming to group the elements of the matrix in five-by-five groups (here, five is the so-called block size of the matrix and usually rises when coupling variables with different meaning, for example pressure, the three components of the velocity and temperature) the assembler will provide the following system:

[           |           |          |           ]   [           ]     [           ]
[  (5 x 5)  |  (5 x 5)  |    ...   |  (5 x 5)  ]   [  (5 x 5)  ]     [  (5 x 5)  ]
[           |           |          |           ]   [           ]     [           ]
[----------------------------------------------]   [-----------]     [-----------]
[           |           |          |           ]   [           ]     [           ]
[    ...    |    ...    |    ...   |  (5 x 5)  ]   [    ...    ]  =  [    ...    ]
[           |           |          |           ]   [           ]     [           ]
[----------------------------------------------]   [-----------]     [-----------]
[           |           |          |           ]   [           ]     [           ]
[  (5 x 5)  |  (5 x 5)  |    ...   |  (5 x 5)  ]   [  (5 x 5)  ]     [  (5 x 5)  ]
[           |           |          |           ]   [           ]     [           ]

Internally, the monolithic matrix will be split into blocks. For example, considering two splits, the first one that group together the first four variables and the second one that holds the last variable (i.e., split sizes equal to [4, 1]), the internal split system will be:

  [                     |                     ]  [           ]     [           ]
  [  (4 * N) x (4 * M)  |  (4 * N) x (1 * M)  ]  [  (4 x M)  ]     [  (4 x N)  ]
  [                     |                     ]  [           ]     [           ]
  [-------------------------------------------]  [-----------]  =  [-----------]
  [                     |                     ]  [           ]     [           ]
  [  (1 * N) x (4 * M)  |  (1 * N) x (1 * M)  ]  [  (1 x M)  ]     [  (1 x N)  ]
  [                     |                     ]  [           ]     [           ]

where M and N are the number of rows and columns respectively.

The PETSc PCFIELDSPLIT preconditioner is used to solve the split system. There are different split strategies available:

1) DIAGONAL: this strategy assumes that the only non-zero blocks are the diagonal ones.

Considering a two-by-two block block matrix

                       [ A00    0  ]
                       [  0    A11 ],

the preconditioned problem will look like

      [ KSPSolve(A00)        0         ]  [ A00    0  ]
      [      0           KSPSolve(A00) ]  [  0    A11 ],

in other words the preconditioner is:

                             ( [ A00    0  ] )
         approximate inverse (               )
                             ( [  0    A11 ] ).

The system is solved efficiently by solving each block independently from the others.

Definition at line 66 of file system_solvers_split.hpp.

Member Typedef Documentation

◆ Assembler

typedef SplitSystemMatrixAssembler bitpit::SplitSystemSolver::Assembler

Definition at line 69 of file system_solvers_split.hpp.

◆ splitStrategy

typedef SystemSplitStrategy bitpit::SplitSystemSolver::splitStrategy

Definition at line 71 of file system_solvers_split.hpp.

Constructor & Destructor Documentation

◆ SplitSystemSolver() [1/6]

bitpit::SplitSystemSolver::SplitSystemSolver ( bool debug = false )

Blocks are solved using a flexible GMRES iterative method. If the system is partitioned each block is preconditioned using the (restricted) additive Schwarz method (ASM). On each block of the ASM preconditioner an incomplete LU factorization (ILU) is used. There is one ASM block per process. If the system is not partitioned it is preconditioned using the incomplete LU factorization (ILU).

2) LOWER: this strategy assumes that the only non-zero blocks are the ones on an below the diagonal.

Considering a two-by-two block block matrix

                       [ A00    0  ]
                       [ A01   A11 ],

the preconditioner is

                             ( [ A00    0  ] )
         approximate inverse (               )
                             ( [ A01   A11 ] ).

The system is solved efficiently by first solving with A00, then applying A01 to that solution, removing it from the right hand side of the second block and then solving with A11, in other words

          [ I      0   ]  [    I    0 ]  [  A00^-1  0 ]
          [ 0   A11^-1 ]  [ - A10   I ]  [   0      I ].

This strategy can be seen as a "block" Gauss-Seidel with the blocks being the splits.

Blocks are solved using a flexible GMRES iterative method. If the system is partitioned each block is preconditioned using the (restricted) additive Schwarz method (ASM). On each block of the ASM preconditioner an incomplete LU factorization (ILU) is used. There is one ASM block per process. If the system is not partitioned it is preconditioned using the incomplete LU factorization (ILU).

3) FULL: this strategy doesn't make assumptions on the structure of the blocks.

Considering a two-by-two block block matrix

                       [ A00   A01 ]
                       [ A11   A11 ],

the preconditioned problem will look like

      [ KSPSolve(A00)        0         ]  [ A00   A01 ]
      [      0           KSPSolve(A00) ]  [ A11   A11 ],

in other words the preconditioner is:

                             ( [ A00    0  ] )
         approximate inverse (               )
                             ( [  0    A11 ] )

The preconditioner is evaluated considering only the diagonal blocks and then the full system is solved.

The system is solved using a flexible GMRES iterative method. If the system is partitioned each diagonal block is preconditioned using the (restricted) additive Schwarz method (ASM). On each block of the ASM preconditioner an incomplete LU factorization (ILU) is used. There is one ASM block per process. If the system is not partitioned it is preconditioned using the incomplete LU factorization (ILU).

Constructor.

Parameters

debug if set to true, debug information will be printed

Definition at line 306 of file system_solvers_split.cpp.

◆ SplitSystemSolver() [2/6]

bitpit::SplitSystemSolver::SplitSystemSolver	(	bool	transpose,
		bool	debug )

Constuctor

Parameters

transpose	if set to true, transposed system will be solved
debug	if set to true, debug information will be printed

Definition at line 317 of file system_solvers_split.cpp.

◆ SplitSystemSolver() [3/6]

bitpit::SplitSystemSolver::SplitSystemSolver	(	bool	flatten,
		bool	transpose,
		bool	debug )

Constuctor

Parameters

flatten	if set to true, block size will not be taken into account when allocating the internal storage of the system matrix. Even when the internal storage is flat, block size information are available and can be used by the solver to speed up the solution of the system. However, since the internal storage doesn't take blocks into account, some low level operations (e.g., matrix-matrix multiplications) cannot use block information. Some algorithms for the solution of the system (e.g., multigrid) requires a flat storage.
transpose	if set to true, transposed system will be solved
debug	if set to true, debug information will be printed

Definition at line 334 of file system_solvers_split.cpp.

◆ SplitSystemSolver() [4/6]

bitpit::SplitSystemSolver::SplitSystemSolver	(	const std::string &	prefix,
		bool	debug = false )

Constructor.

Parameters

prefix	is the prefix string to prepend to all option requests
debug	if set to true, debug information will be printed

Definition at line 345 of file system_solvers_split.cpp.

◆ SplitSystemSolver() [5/6]

bitpit::SplitSystemSolver::SplitSystemSolver	(	const std::string &	prefix,
		bool	transpose,
		bool	debug )

Constructor.

Parameters

prefix	is the prefix string to prepend to all option requests
debug	if set to true, debug information will be printed

Definition at line 356 of file system_solvers_split.cpp.

◆ SplitSystemSolver() [6/6]

bitpit::SplitSystemSolver::SplitSystemSolver	(	const std::string &	prefix,
		bool	flatten,
		bool	transpose,
		bool	debug )

Constuctor

Parameters

prefix	is the prefix string to prepend to all option requests
flatten	if set to true, block size will not be taken into account when allocating the internal storage of the system matrix. Even when the internal storage is flat, block size information are available and can be used by the solver to speed up the solution of the system. However, since the internal storage doesn't take blocks into account, some low level operations (e.g., matrix-matrix multiplications) cannot use block information. Some algorithms for the solution of the system (e.g., multigrid) requires a flat storage.
transpose	if set to true, transposed system will be solved
debug	if set to true, debug information will be printed

Definition at line 374 of file system_solvers_split.cpp.

Member Function Documentation

◆ assembly() [1/9]

void bitpit::SplitSystemSolver::assembly ( const Assembler & assembler )

Assembly the system.

After assembying th system solver, its options will be reset.

Parameters

assembler is the matrix assembler

Definition at line 734 of file system_solvers_split.cpp.

◆ assembly() [2/9]

void bitpit::SystemSolver::assembly ( const Assembler & assembler )

protected

Assembly the system.

After assembying th system solver, its options will be reset.

Parameters

assembler is the matrix assembler

Definition at line 290 of file system_solvers_large.cpp.

◆ assembly() [3/9]

void bitpit::SplitSystemSolver::assembly	(	const Assembler &	assembler,
		const SystemMatrixOrdering &	reordering )

Assembly the system.

Parameters

assembler	is the matrix assembler
reordering	is the reordering that will be applied when assembling the system

Definition at line 745 of file system_solvers_split.cpp.

◆ assembly() [4/9]

void bitpit::SystemSolver::assembly	(	const Assembler &	assembler,
		const SystemMatrixOrdering &	reordering )

protected

Assembly the system.

After assembying th system solver, its options will be reset.

Parameters

assembler	is the matrix assembler
reordering	is the reordering that will be applied when assembling the system

Definition at line 291 of file system_solvers_large.cpp.

◆ assembly() [5/9]

void bitpit::SystemSolver::assembly ( const SparseMatrix & matrix )

protected

Assembly the system.

After assembying th system solver, its options will be reset.

Parameters

matrix is the matrix

Definition at line 288 of file system_solvers_large.cpp.

◆ assembly() [6/9]

void bitpit::SystemSolver::assembly	(	const SparseMatrix &	matrix,
		const SystemMatrixOrdering &	reordering )

protected

Assembly the system.

After assembying th system solver, its options will be reset.

Parameters

matrix	is the matrix
reordering	is the reordering that will be applied when assembling the system

Definition at line 289 of file system_solvers_large.cpp.

◆ assembly() [7/9]

void bitpit::SplitSystemSolver::assembly	(	const SparseMatrix &	matrix,
		SystemSplitStrategy	splitStrategy,
		const std::vector< int > &	splitSizes )

Assembly the system.

After assembying th system solver, its options will be reset.

Parameters

matrix	is the matrix
splitStrategy	the type of split that will be applied to the system
splitSizes	are the sizes of the splits

Definition at line 699 of file system_solvers_split.cpp.

◆ assembly() [8/9]

void bitpit::SplitSystemSolver::assembly	(	const SparseMatrix &	matrix,
		SystemSplitStrategy	splitStrategy,
		const std::vector< int > &	splitSizes,
		const SystemMatrixOrdering &	reordering )

Assembly the system.

Parameters

matrix	is the matrix
splitStrategy	the type of split that will be applied to the system
splitSizes	are the sizes of the splits
reordering	is the reordering that will be applied when assemblying the system

Definition at line 713 of file system_solvers_split.cpp.

◆ assembly() [9/9]

template<typename DerivedSystemSolver >

void bitpit::SystemSolver::assembly	(	const typename DerivedSystemSolver::Assembler &	assembler,
		const SystemMatrixOrdering &	reordering )

protected

Assembly the system.

After assembying th system solver, its options will be reset.

Parameters

assembler	is the matrix assembler
reordering	is the reordering that will be applied when assemblying the system

Definition at line 408 of file system_solvers_large.tpp.

◆ destroyKSPOptions()

void bitpit::SplitSystemSolver::destroyKSPOptions ( )

overrideprotectedvirtual

Destroy the options associated with the KSP.

Reimplemented from bitpit::SystemSolver.

Definition at line 548 of file system_solvers_split.cpp.

◆ destroyKSPStatus()

void bitpit::SplitSystemSolver::destroyKSPStatus ( )

overrideprotectedvirtual

Destroy the status of the KSP.

Reimplemented from bitpit::SystemSolver.

Definition at line 649 of file system_solvers_split.cpp.

◆ destroySplitKSPOptions()

void bitpit::SplitSystemSolver::destroySplitKSPOptions ( )

protectedvirtual

Destroy the options associated with the split KSPs.

Definition at line 557 of file system_solvers_split.cpp.

◆ destroySplitKSPStatuses()

void bitpit::SplitSystemSolver::destroySplitKSPStatuses ( )

protectedvirtual

Destroy the status of the split KSPs.

Definition at line 658 of file system_solvers_split.cpp.

◆ dumpInfo()

void bitpit::SplitSystemSolver::dumpInfo ( std::ostream & systemStream ) const

overrideprotectedvirtual

Dump system information.

Parameters

systemStream is the stream in which system information is written

Reimplemented from bitpit::SystemSolver.

Definition at line 669 of file system_solvers_split.cpp.

◆ exportMatrix() [1/3]

void bitpit::SystemSolver::exportMatrix	(	const std::string &	filePath,
		FileFormat	fileFormat = FILE_BINARY ) const

protectedvirtual

Export the matrix to the specified file.

Parameters

filePath	is the path of the file
fileFormat	is the file format that will be used for exporting the matrix, note that the ASCII format may not be able to handle large matrices

Reimplemented from bitpit::SystemSolver.

Definition at line 338 of file system_solvers_large.cpp.

◆ exportMatrix() [2/3]

void bitpit::SplitSystemSolver::exportMatrix	(	const std::string &	filePath,
		FileFormat	fileFormat = FILE_BINARY ) const

overridevirtual

Export the matrix to the specified file.

Parameters

filePath	is the path of the file
fileFormat	is the file format that will be used for exporting the matrix, note that the ASCII format may not be able to handle large matrices

Reimplemented from bitpit::SystemSolver.

Definition at line 1458 of file system_solvers_split.cpp.

◆ exportMatrix() [3/3]

void bitpit::SystemSolver::exportMatrix	(	Mat	matrix,
		const std::string &	filePath,
		FileFormat	fileFormat ) const

protected

Export the given matrix to the specified file.

Parameters

matrix	is the matrix that will be filled
filePath	is the path of the file that will contain matrix data
fileFormat	is the file format that will be used for exporting the matrix, note that the ASCII format may not be able to handle large matrices

Definition at line 455 of file system_solvers_large.cpp.

◆ fillKSPStatus() [1/3]

void bitpit::SystemSolver::fillKSPStatus ( )

protectedvirtual

Fill the status of the KSP.

Reimplemented from bitpit::SystemSolver.

Definition at line 439 of file system_solvers_large.cpp.

◆ fillKSPStatus() [2/3]

void bitpit::SplitSystemSolver::fillKSPStatus ( )

overrideprotectedvirtual

Fill the status of the KSP.

Reimplemented from bitpit::SystemSolver.

Definition at line 604 of file system_solvers_split.cpp.

◆ fillKSPStatus() [3/3]

void bitpit::SystemSolver::fillKSPStatus	(	KSP	ksp,
		KSPStatus *	status ) const

protectedvirtual

Fill the status of the specified KSP.

Parameters

ksp	is the KSP the status will be fetched from
status	on output will contain the status of the KSP

Reimplemented from bitpit::SystemSolver.

Definition at line 440 of file system_solvers_large.cpp.

◆ fillSplitKSPStatuses()

void bitpit::SplitSystemSolver::fillSplitKSPStatuses ( )

protectedvirtual

Fill the status of the split KSPs.

Definition at line 613 of file system_solvers_split.cpp.

◆ generateSplitIndexes()

void bitpit::SplitSystemSolver::generateSplitIndexes	(	int	split,
		long	nItems,
		std::vector< std::size_t > *	indexes ) const

protected

Generate split indexes.

Split indexes define the indexes of the matrix that are assigned to the specified split. The k-th element of a split matrix row/column corresponds to the indexes[k]-th element of the corresponding row/column of the full matrix.

Parameters

split	is the split
nItems	is the number of items in the row/column
indexes	on output will contain the split indexes

Definition at line 1680 of file system_solvers_split.cpp.

◆ generateSplitPath() [1/3]

std::string bitpit::SplitSystemSolver::generateSplitPath	(	const std::string &	path,
		const std::string &	index ) const

protected

Generate the split path associated with the specified indexes.

The split file path is generated adding "_split_ij" before the extension of the specified path. For example, given the path "data/matrix.dat", the split path associated with the index INDEX is "data/matrix_split_INDEX.dat".

Parameters

path	is the path
index	is the index associated with the split

Definition at line 1741 of file system_solvers_split.cpp.

◆ generateSplitPath() [2/3]

std::string bitpit::SplitSystemSolver::generateSplitPath	(	const std::string &	path,
		int	i ) const

protected

Generate the split path associated with the specified indexes.

The split file path is generated adding "_split_ij" before the extension of the specified path. For example, given the path "data/matrix.dat", the split path associated with the index 1 is "data/matrix_split_1.dat".

Parameters

path	is the path
i	is the index associated with the split

Definition at line 1710 of file system_solvers_split.cpp.

◆ generateSplitPath() [3/3]

std::string bitpit::SplitSystemSolver::generateSplitPath	(	const std::string &	path,
		int	i,
		int	j ) const

protected

Generate the split path associated with the specified indexes.

The split file path is generated adding "_split_ij" before the extension of the specified path. For example, given the path "data/matrix.dat", the split path associated with the indexes (1,0) is "data/matrix_split_10.dat".

Parameters

path	is the path
i	is the first index associated with the split
j	is the second index associated with the split

Definition at line 1726 of file system_solvers_split.cpp.

◆ generateSplitPermutation()

void bitpit::SplitSystemSolver::generateSplitPermutation	(	long	nItems,
		long	itemsGlobalOffset,
		IS *	permutation ) const

protected

Generate the split permutation.

Given a vector whose elements are ordered according to the full matrix, the split permutations allow to obtain a vector whose elements are ordered according to the split matrices.

Parameters

	nItems	is the number of items in the row/column
	itemsGlobalOffset	is the global offset of the items that belong to the current process
	split	is the split
[out]	permutations	on output will contain the permutations

Definition at line 1620 of file system_solvers_split.cpp.

◆ getBlockSize()

int bitpit::SplitSystemSolver::getBlockSize ( ) const

overridevirtual

Get the block size of the system.

Returns: The block size of the system.

Reimplemented from bitpit::SystemSolver.

Definition at line 385 of file system_solvers_split.cpp.

◆ getSplitCount()

int bitpit::SplitSystemSolver::getSplitCount ( ) const

Get the number of splits that will be applied.

Returns: The number of splits that will be applied.

Definition at line 412 of file system_solvers_split.cpp.

◆ getSplitKSPOptions() [1/2]

KSPOptions & bitpit::SplitSystemSolver::getSplitKSPOptions ( int split )

Get a reference to the options associated to the Krylov solver of the specified split.

The options associated with the Krylov solver can only be accessed after assembling the system.

Parameters

split is the split

Returns: A reference to the options associated to the Krylov solver of the specified split.

Definition at line 497 of file system_solvers_split.cpp.

◆ getSplitKSPOptions() [2/2]

const KSPOptions & bitpit::SplitSystemSolver::getSplitKSPOptions ( int split ) const

Get a constant reference to the options associated to the Krylov solver of the specified split.

The options associated with the Krylov solver can only be accessed after assembling the system.

Parameters

split is the split

Returns: A constant reference to the options associated to the Krylov solver of the specified split.

Definition at line 515 of file system_solvers_split.cpp.

◆ getSplitKSPStatus()

const KSPStatus & bitpit::SplitSystemSolver::getSplitKSPStatus ( int split ) const

Get a constant reference to the status of the Krylov solver of the specified split.

The status of the Krylov solver can only be accessed after assembling the system.

Parameters

split is the split

Returns: A constant reference to the status of the Krylov solver of the specified split.

Definition at line 570 of file system_solvers_split.cpp.

◆ getSplitOffsets()

std::vector< int > bitpit::SplitSystemSolver::getSplitOffsets ( ) const

Get the block offtest of the splits.

Returns: The block offsets of the splits.

Definition at line 467 of file system_solvers_split.cpp.

◆ getSplitSizes()

std::vector< int > bitpit::SplitSystemSolver::getSplitSizes ( ) const

Get the size of each split.

Each block of elements is split into pieces whose size is defined by the split sizes. The sum of the split sizes must be equal to the block size. Given a matrix with a block size equal to 5 and split sizes equal to [4, 1], the split matrix will look like this:

   [                     |                     ]
   [  (4 * N) * (4 * M)  |  (4 * N) * (1 * M)  ]
   [                     |                     ]
   [-------------------------------------------]
   [                     |                     ]
   [  (1 * N) * (4 * M)  |  (1 * N) * (1 * M)  ]
   [                     |                     ]

where M and N are the number of rows and columns respectively.

Returns: The size of each split.

Definition at line 443 of file system_solvers_split.cpp.

◆ getSplitStrategy()

SystemSplitStrategy bitpit::SplitSystemSolver::getSplitStrategy ( ) const

Get the split strategy will be used when solving the system.

Returns: The split strategy will be used when solving the system.

Definition at line 402 of file system_solvers_split.cpp.

◆ initializeKSPOptions()

void bitpit::SplitSystemSolver::initializeKSPOptions ( )

overrideprotectedvirtual

Initialize the options associated with the KSP.

Reimplemented from bitpit::SystemSolver.

Definition at line 527 of file system_solvers_split.cpp.

◆ initializeKSPStatus()

void bitpit::SplitSystemSolver::initializeKSPStatus ( )

overrideprotectedvirtual

Initialize the status of the split KSPs.

Reimplemented from bitpit::SystemSolver.

Definition at line 581 of file system_solvers_split.cpp.

◆ initializeSplitKSPOptions()

void bitpit::SplitSystemSolver::initializeSplitKSPOptions ( )

protectedvirtual

Initialize the options associated with the sub KSP.

Definition at line 536 of file system_solvers_split.cpp.

◆ initializeSplitKSPStatuses()

void bitpit::SplitSystemSolver::initializeSplitKSPStatuses ( )

protectedvirtual

Initialize the options associated with the split KSPs.

Definition at line 590 of file system_solvers_split.cpp.

◆ matrixAssembly()

void bitpit::SplitSystemSolver::matrixAssembly ( const Assembler & assembler )

protected

Assemble the matrix.

Parameters

assembler is the matrix assembler

Definition at line 818 of file system_solvers_split.cpp.

◆ matrixDestroy()

void bitpit::SplitSystemSolver::matrixDestroy ( )

overrideprotectedvirtual

Destroy the matrix.

Reimplemented from bitpit::SystemSolver.

Definition at line 1336 of file system_solvers_split.cpp.

◆ matrixDump()

void bitpit::SplitSystemSolver::matrixDump	(	std::ostream &	systemStream,
		const std::string &	directory,
		const std::string &	prefix ) const

overrideprotectedvirtual

Dump the matrix.

Parameters

systemStream	is the stream in which system information is written
directory	is the directory in which the matrix data file will be written.
prefix	is the prefix added to the name of the file containing matrix data

Reimplemented from bitpit::SystemSolver.

Definition at line 1260 of file system_solvers_split.cpp.

◆ matrixFill()

void bitpit::SplitSystemSolver::matrixFill ( const std::string & filePath )

overrideprotectedvirtual

Fill the matrix reading its contents form the specified file.

The input file should contain a compatible vector stored in PETSc binary format. It's up to the caller of this routine to make sure the loaded vector is compatible with the matrix. If the matrix file cannot be read an exception is thrown.

Parameters

filePath is the path of the file

Reimplemented from bitpit::SystemSolver.

Definition at line 1005 of file system_solvers_split.cpp.

◆ matrixRestore()

void bitpit::SplitSystemSolver::matrixRestore	(	std::istream &	systemStream,
		const std::string &	directory,
		const std::string &	prefix,
		bool	redistribute )

overrideprotectedvirtual

Restore the matrix.

Parameters

systemStream	is the stream from which system information is read
directory	is the directory from which the matrix data file will be read
prefix	is the prefix that will be was added to the files during
redistribute	if set to true, the matrix will be redistributed among the available processes, allowing to restore the matrix with a different number of processes than those used to dump it

Reimplemented from bitpit::SystemSolver.

Definition at line 1300 of file system_solvers_split.cpp.

◆ matrixUpdate()

void bitpit::SplitSystemSolver::matrixUpdate	(	long	nRows,
		const long *	rows,
		const Assembler &	assembler )

protected

Update the specified rows of the matrix.

The contents of the specified rows will be replaced by the data provided by the given assembler. If the matrix has not been assembled yet, both the pattern and the values of the matrix will be updated. After the matrix has been assembled only the values will be updated.

The block size of the assembler should be equal to the block size of the matrix.

Parameters

nRows	is the number of rows that will be updated
rows	are the local indices of the rows that will be updated, if a null pointer is passed, the rows that will be updated are the rows from 0 to (nRows - 1).
assembler	is the matrix assembler for the rows that will be updated

Definition at line 1039 of file system_solvers_split.cpp.

◆ postKSPSolveActions()

void bitpit::SplitSystemSolver::postKSPSolveActions ( )

overrideprotectedvirtual

Post-solve actions.

Reimplemented from bitpit::SystemSolver.

Definition at line 480 of file system_solvers_split.cpp.

◆ preKrylovSetupActions()

void bitpit::SplitSystemSolver::preKrylovSetupActions ( )

overrideprotectedvirtual

Perform actions before Krylov subspace method setup.

Reimplemented from bitpit::SystemSolver.

Definition at line 1578 of file system_solvers_split.cpp.

◆ resetKSPOptions()

void bitpit::SystemSolver::resetKSPOptions ( KSPOptions * options ) const

protectedvirtual

Reset the specified KSP options.

Parameters

options are the options that will be reset

Reimplemented from bitpit::SystemSolver.

Definition at line 433 of file system_solvers_large.cpp.

◆ resetKSPStatus() [1/3]

void bitpit::SystemSolver::resetKSPStatus ( )

protectedvirtual

Reset the status of the KSP.

Reimplemented from bitpit::SystemSolver.

Definition at line 437 of file system_solvers_large.cpp.

◆ resetKSPStatus() [2/3]

void bitpit::SplitSystemSolver::resetKSPStatus ( )

overrideprotectedvirtual

Reset the status of the KSP.

Reimplemented from bitpit::SystemSolver.

Definition at line 630 of file system_solvers_split.cpp.

◆ resetKSPStatus() [3/3]

void bitpit::SystemSolver::resetKSPStatus ( KSPStatus * status ) const

protectedvirtual

Reset the status of the specified KSP.

Parameters

status on output will contain the status of the KSP

Reimplemented from bitpit::SystemSolver.

Definition at line 438 of file system_solvers_large.cpp.

◆ resetSplitKSPStatuses()

void bitpit::SplitSystemSolver::resetSplitKSPStatuses ( )

protectedvirtual

Reset the status of the split KSPs.

Definition at line 639 of file system_solvers_split.cpp.

◆ restoreInfo()

void bitpit::SplitSystemSolver::restoreInfo ( std::istream & systemStream )

overrideprotectedvirtual

Restore system information.

Parameters

systemStream is the stream from which system information is read

Reimplemented from bitpit::SystemSolver.

Definition at line 683 of file system_solvers_split.cpp.

◆ setupKrylov() [1/3]

void bitpit::SystemSolver::setupKrylov ( )

protectedvirtual

Set up the Krylov subspace method used to solve the system.

Reimplemented from bitpit::SystemSolver.

Definition at line 423 of file system_solvers_large.cpp.

◆ setupKrylov() [2/3]

void bitpit::SplitSystemSolver::setupKrylov ( )

overrideprotectedvirtual

Set up the Krylov subspace method used to solve the system.

Reimplemented from bitpit::SystemSolver.

Definition at line 1529 of file system_solvers_split.cpp.

◆ setupKrylov() [3/3]

void bitpit::SystemSolver::setupKrylov	(	KSP	ksp,
		const KSPOptions &	options ) const

protectedvirtual

Set up the Krylov subspace method using the given options.

This function is in charge of only setting the properties of the Krylov subspace method that will be used to solve the system. There is a dedicated function to set up the preconditioner.

Parameters

ksp	is the KSP whose Krylov subspace method will be setup
options	are the options that will be used to set up the KSP

Reimplemented from bitpit::SystemSolver.

Definition at line 424 of file system_solvers_large.cpp.

◆ setupPreconditioner() [1/3]

void bitpit::SystemSolver::setupPreconditioner ( )

protectedvirtual

Set up the preconditioner.

Reimplemented from bitpit::SystemSolver.

Definition at line 418 of file system_solvers_large.cpp.

◆ setupPreconditioner() [2/3]

void bitpit::SplitSystemSolver::setupPreconditioner ( )

overrideprotectedvirtual

Set up the preconditioner.

Reimplemented from bitpit::SystemSolver.

Definition at line 1477 of file system_solvers_split.cpp.

◆ setupPreconditioner() [3/3]

void bitpit::SystemSolver::setupPreconditioner	(	PC	pc,
		const KSPOptions &	options ) const

protectedvirtual

Set up the specified preconditioner using the given options.

Parameters

pc	is the preconditioner to set up
options	are the options that will be used to set up the preconditioner

Reimplemented from bitpit::SystemSolver.

Definition at line 419 of file system_solvers_large.cpp.

◆ setupSplitKrylovs()

void bitpit::SplitSystemSolver::setupSplitKrylovs ( )

protectedvirtual

Set up the Krylov subspace method used to solve the system.

Definition at line 1551 of file system_solvers_split.cpp.

◆ setupSplitPreconditioners()

void bitpit::SplitSystemSolver::setupSplitPreconditioners ( )

protectedvirtual

Set up the Krylov subspace method used to solve the system.

Definition at line 1508 of file system_solvers_split.cpp.

◆ SystemSolver() [1/6]

bitpit::SystemSolver::SystemSolver ( bool debug = false )

Constructor.

Parameters

debug if set to true, debug information will be printed

Definition at line 274 of file system_solvers_large.cpp.

◆ SystemSolver() [2/6]

bitpit::SystemSolver::SystemSolver	(	bool	flatten,
		bool	transpose,
		bool	debug )

Constuctor

Parameters

flatten	if set to true, block size will not be taken into account when allocating the internal storage of the system matrix. Even when the internal storage is flat, block size information are available and can be used by the solver to speed up the solution of the system. However, since the internal storage doesn't take blocks into account, some low level operations (e.g., matrix-matrix multiplications) cannot use block information. Some algorithms for the solution of the system (e.g., multigrid) requires a flat storage.
transpose	if set to true, transposed system will be solved
debug	if set to true, debug information will be printed

Definition at line 276 of file system_solvers_large.cpp.

◆ SystemSolver() [3/6]

bitpit::SystemSolver::SystemSolver	(	bool	transpose,
		bool	debug )

Constuctor

Parameters

transpose	if set to true, transposed system will be solved
debug	if set to true, debug information will be printed

Definition at line 275 of file system_solvers_large.cpp.

◆ SystemSolver() [4/6]

bitpit::SystemSolver::SystemSolver	(	const std::string &	prefix,
		bool	debug = false )

Constructor.

Parameters

prefix	is the prefix string to prepend to all option requests
debug	if set to true, debug information will be printed

Definition at line 277 of file system_solvers_large.cpp.

◆ SystemSolver() [5/6]

bitpit::SystemSolver::SystemSolver	(	const std::string &	prefix,
		bool	flatten,
		bool	transpose,
		bool	debug )

Constuctor

Parameters

prefix	is the prefix string to prepend to all option requests
flatten	if set to true, block size will not be taken into account when allocating the internal storage of the system matrix. Even when the internal storage is flat, block size information are available and can be used by the solver to speed up the solution of the system. However, since the internal storage doesn't take blocks into account, some low level operations (e.g., matrix-matrix multiplications) cannot use block information. Some algorithms for the solution of the system (e.g., multigrid) requires a flat storage.
transpose	if set to true, transposed system will be solved
debug	if set to true, debug information will be printed

Definition at line 279 of file system_solvers_large.cpp.

◆ SystemSolver() [6/6]

bitpit::SystemSolver::SystemSolver	(	const std::string &	prefix,
		bool	transpose,
		bool	debug )

Constructor.

Parameters

prefix	is the prefix string to prepend to all option requests
debug	if set to true, debug information will be printed

Definition at line 278 of file system_solvers_large.cpp.

◆ update() [1/8]

void bitpit::SplitSystemSolver::update ( const Assembler & assembler )

Update all the rows of the system.

Only the values of the system matrix can be updated, once the system is assembled its pattern cannot be modified.

Parameters

assembler is the matrix assembler for the rows that will be updated

Definition at line 793 of file system_solvers_split.cpp.

◆ update() [2/8]

void bitpit::SystemSolver::update ( const Assembler & assembler )

protected

Update all the rows of the system.

Only the values of the system matrix can be updated, once the system is assembled its pattern cannot be modified.

Parameters

assembler is the matrix assembler for the rows that will be updated

Definition at line 295 of file system_solvers_large.cpp.

◆ update() [3/8]

void bitpit::SplitSystemSolver::update ( const SparseMatrix & elements )

Update all the rows of the system.

Only the values of the system matrix can be updated, once the system is assembled its pattern cannot be modified.

Parameters

elements are the elements that will be used to update the rows

Definition at line 758 of file system_solvers_split.cpp.

◆ update() [4/8]

void bitpit::SplitSystemSolver::update	(	long	nRows,
		const long *	rows,
		const Assembler &	assembler )

Update the system.

Only the values of the system matrix can be updated, once the system is assembled its pattern cannot be modified.

Parameters

nRows	is the number of rows that will be updated
rows	are the indices of the rows that will be updated
assembler	is the matrix assembler for the rows that will be updated

Definition at line 808 of file system_solvers_split.cpp.

◆ update() [5/8]

void bitpit::SystemSolver::update	(	long	nRows,
		const long *	rows,
		const Assembler &	assembler )

protected

Update the system.

Only the values of the system matrix can be updated, once the system is assembled its pattern cannot be modified.

Parameters

nRows	is the number of rows that will be updated
rows	are the indices of the rows that will be updated
assembler	is the matrix assembler for the rows that will be updated

Definition at line 296 of file system_solvers_large.cpp.

◆ update() [6/8]

void bitpit::SplitSystemSolver::update	(	long	nRows,
		const long *	rows,
		const SparseMatrix &	elements )

Update the system.

Only the values of the system matrix can be updated, once the system is assembled its pattern cannot be modified.

Parameters

nRows	is the number of rows that will be updated
rows	are the indices of the rows that will be updated
elements	are the elements that will be used to update the rows

Definition at line 773 of file system_solvers_split.cpp.

◆ update() [7/8]

void bitpit::SystemSolver::update	(	long	nRows,
		const long *	rows,
		const SparseMatrix &	elements )

protected

Update the system.

Only the values of the system matrix can be updated, once the system is assembled its pattern cannot be modified.

Parameters

nRows	is the number of rows that will be updated
rows	are the indices of the rows that will be updated
elements	are the elements that will be used to update the rows

Definition at line 294 of file system_solvers_large.cpp.

◆ update() [8/8]

template<typename DerivedSystemSolver >

void bitpit::SystemSolver::update	(	long	nRows,
		const long *	rows,
		const typename DerivedSystemSolver::Assembler &	assembler )

protected

Update the system.

Only the values of the system matrix can be updated, once the system is assembled its pattern cannot be modified.

Parameters

nRows	is the number of rows that will be updated
rows	are the indices of the rows that will be updated
assembler	is the matrix assembler for the rows that will be updated

Definition at line 411 of file system_solvers_large.tpp.

◆ vectorsCreate()

void bitpit::SplitSystemSolver::vectorsCreate ( )

overrideprotectedvirtual

Create RHS and solution vectors.

Vectors will be created, but they will not be initialized.

Note that MatNest does not depend on VecNest, and in general, Jed Brown (one of the main developers of PETSc) recommend not using VecNest. For this reason, VecNest will not be used.

Reimplemented from bitpit::SystemSolver.

Definition at line 1353 of file system_solvers_split.cpp.

◆ vectorsCreateSplitPermutations()

void bitpit::SplitSystemSolver::vectorsCreateSplitPermutations ( )

protected

Create the split permutation for RHS and solution vectors.

Given a vector whose elements are ordered according to the full matrix, the split permutations allow to obtain a vector whose elements are ordered according to the split matrices.

Definition at line 1398 of file system_solvers_split.cpp.

◆ vectorsDestroy()

void bitpit::SplitSystemSolver::vectorsDestroy ( )

overrideprotectedvirtual

Destroy RHS and solution vectors.

Reimplemented from bitpit::SystemSolver.

Definition at line 1436 of file system_solvers_split.cpp.

◆ vectorsReorder()

void bitpit::SplitSystemSolver::vectorsReorder ( bool invert )

overrideprotectedvirtual

Reorder RHS and solution vectors to match the order of the system matrix.

Parameters

invert is a flag for inverting the ordering

Reimplemented from bitpit::SystemSolver.

Definition at line 1367 of file system_solvers_split.cpp.

◆ vectorsRestore()

void bitpit::SplitSystemSolver::vectorsRestore	(	std::istream &	systemStream,
		const std::string &	directory,
		const std::string &	prefix )

overrideprotectedvirtual

Restore RHS and solution vectors.

Parameters

systemStream	is the stream from which system information is read
directory	is the directory from which the vector data file will be read
prefix	is the prefix added to the name of the file containing vectors data

Reimplemented from bitpit::SystemSolver.

Definition at line 1382 of file system_solvers_split.cpp.

Friends And Related Symbol Documentation

◆ SystemSolver

friend class SystemSolver

friend

Definition at line 73 of file system_solvers_split.hpp.

Member Data Documentation

◆ m_splitAs

std::vector<Mat> bitpit::SplitSystemSolver::m_splitAs

protected

Definition at line 112 of file system_solvers_split.hpp.

◆ m_splitKSPOptions

std::vector<KSPOptions> bitpit::SplitSystemSolver::m_splitKSPOptions

protected

Definition at line 114 of file system_solvers_split.hpp.

◆ m_splitKSPStatuses

std::vector<KSPStatus> bitpit::SplitSystemSolver::m_splitKSPStatuses

protected

Definition at line 115 of file system_solvers_split.hpp.

◆ m_splitStrategy

SystemSplitStrategy bitpit::SplitSystemSolver::m_splitStrategy

protected

Definition at line 110 of file system_solvers_split.hpp.

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

src/LA/system_solvers_split.hpp
src/LA/system_solvers_split.cpp

Public Types

Public Member Functions

Protected Member Functions

Protected Attributes

Friends

Additional Inherited Members

Detailed Description

Member Typedef Documentation

◆ Assembler

◆ splitStrategy

Constructor & Destructor Documentation

◆ SplitSystemSolver() [1/6]

◆ SplitSystemSolver() [2/6]

◆ SplitSystemSolver() [3/6]

◆ SplitSystemSolver() [4/6]

◆ SplitSystemSolver() [5/6]

◆ SplitSystemSolver() [6/6]

Member Function Documentation

◆ assembly() [1/9]

◆ assembly() [2/9]

◆ assembly() [3/9]

◆ assembly() [4/9]

◆ assembly() [5/9]

◆ assembly() [6/9]

◆ assembly() [7/9]

◆ assembly() [8/9]

◆ assembly() [9/9]

◆ destroyKSPOptions()

◆ destroyKSPStatus()

◆ destroySplitKSPOptions()

◆ destroySplitKSPStatuses()

◆ dumpInfo()

◆ exportMatrix() [1/3]

◆ exportMatrix() [2/3]

◆ exportMatrix() [3/3]

◆ fillKSPStatus() [1/3]

◆ fillKSPStatus() [2/3]

◆ fillKSPStatus() [3/3]

◆ fillSplitKSPStatuses()

◆ generateSplitIndexes()

◆ generateSplitPath() [1/3]

◆ generateSplitPath() [2/3]

◆ generateSplitPath() [3/3]

◆ generateSplitPermutation()

◆ getBlockSize()

◆ getSplitCount()

◆ getSplitKSPOptions() [1/2]

◆ getSplitKSPOptions() [2/2]

◆ getSplitKSPStatus()

◆ getSplitOffsets()

◆ getSplitSizes()

◆ getSplitStrategy()

◆ initializeKSPOptions()

◆ initializeKSPStatus()

◆ initializeSplitKSPOptions()

◆ initializeSplitKSPStatuses()

◆ matrixAssembly()

◆ matrixDestroy()

◆ matrixDump()

◆ matrixFill()

◆ matrixRestore()

◆ matrixUpdate()

◆ postKSPSolveActions()

◆ preKrylovSetupActions()

◆ resetKSPOptions()

◆ resetKSPStatus() [1/3]

◆ resetKSPStatus() [2/3]

◆ resetKSPStatus() [3/3]

◆ resetSplitKSPStatuses()

◆ restoreInfo()

◆ setupKrylov() [1/3]

◆ setupKrylov() [2/3]

◆ setupKrylov() [3/3]

◆ setupPreconditioner() [1/3]

◆ setupPreconditioner() [2/3]

◆ setupPreconditioner() [3/3]

◆ setupSplitKrylovs()

◆ setupSplitPreconditioners()

◆ SystemSolver() [1/6]

◆ SystemSolver() [2/6]