A convenience class that builds a GmMatrixSet with the types given by GmpFemMatrixTypes and also associates a global matrix to each element matrix stored in the set (it can also associate a single global matrix, dissociated with matrix types) More...

#include <gmpFemMatrixSet.h>

Inheritance diagram for GmpFemMatrixSet:

[legend]

Collaboration diagram for GmpFemMatrixSet:

[legend]

Classes
struct	GmpFemMatrixTypeData
	Aux structure storing the matrix associated with a matrix type. More...

Public Types
enum	GmpFemMatrixMode { MATRIX_MATCH_MODE, MATRIX_SINGLE_MODE, MATRIX_SINGLE_DEBUG_MODE }
	Control mode specifying if the matrix set should have one global matrix per local matrix (MATCH_MODE), a single global matrix independing of the number of local matrices (SINGLE_MODE) or both: a single global matrix + one global matrix per configured local matrix. This last mode is usefull for debuging purposes only. More...

Public Member Functions
	GmpFemMatrixSet ()
	Constructor.

	~GmpFemMatrixSet ()
	Destructor.

void	clear ()
	Clears the set, restoring it to the state after construction.

bool	initTypes (const GmNumSolver *solver, GmpFemMatrixMode mode, int numMatrices,...)
	Informs the matrix set about the local matrix types that will be stored on it. It also creates the set of global matrices tied to the element matrices in the set. The number of effectively stored global matrices follows the given mode. More...

bool	adjustSizes (int globalNumDofs, int maxElemDofs, bool sparseLayoutChanged)
	Adjusts the sizes of local and global matrices according to the given number of dofs. Can be called more than once to update matrix sizes on model changes. More...

void	setNumThreads (int nthreads)
	Sets the number of threads used by sparse matrices in batch/layout operations. Default is the number of threads configured by the thread manager (-1). Should be called before calls to adjustSizes().

void	setTransposedDofVector (GmpFemMatrixTypes type, GmpFemVectorTypes vectorType)
	Registers the vector type that should be used to store transposed values when reducing a matrix of the given matrix type, due to fixed dofs. More...

void	clearGlobalMatrices (bool keepSparseLayout)
	Clears global matrices (or the single matrix), filling them with zeros and marking them as symmetric. More...

bool	supportsBatchInsert () const
	Returns true if the matrix set supports batch inserts.

bool	beginBatchInsert (size_t expectedEntries=0)
	Initializes a batch insert process for all ACTIVE global matrices + the print matrices if necessary. More...

bool	endBatchInsert (bool discardData)
	Finalizes a batch insert process for all ACTIVE global matrices + the print matrices if necessary. More...

bool	supportsSparseLayouts () const
	Returns true if the matrix set supports sparse layouts.

GmSparseMatrixLayoutBuilder *	layoutBuilder () const
	If the matrices in this matrix set support sparse layouts, returns a layout builder object capable of filling the shared matrices layout. Otherwise, returns NULL.

bool	emptyLayout () const
	If the matrices in this matrix set support sparse layouts and the shared layout is empty, returns true. Otherwise, returns false.

bool	supportsParallelAdd () const
	Returns true if the solver matrices support parallel assembling (supportedParallelAddMode() != NO_SUPPORT)

bool	needsLocking () const
	Returns true if the solver matrices needs external locking for parallel assembling (supportedParallelAddMode() == REENTRANT_SUPPORT)

GmSolverMatrix *	globalMatrix (GmpFemMatrixTypes type) const
	Returns a global matrix identified by its type in the set. More...

GmSolverMatrix *	globalMatrixFromIndex (int index) const
	Returns a global matrix identified by its index in the set. Can NOT be used for "single" or "single debug" modes.

int	transposedDofVector (GmpFemMatrixTypes type) const
	Returns the registered vector type for storing transposed values when reducing a matrix due to fixed dofs for the given matrix type. Will be equal to the value given in the call to setTransposedDofVector() or -1 if that function was not called. More...

int	transposedDofVectorFromIndex (int index) const
	Returns the registered vector type for storing transposed values when reducing a matrix due to fixed dofs for the given matrix index. Will be equal to the value given in the call to setTransposedDofVector() or -1 if that function was not called. Can NOT be used for "single" or "single debug" modes. More...

GmSolverMatrix *	globalDebugPrintMatrix (GmpFemMatrixTypes type) const
	Returns the auxiliar matrix used for storing global matrices matching local ones for debug printing purposes when operating in "single debug" mode or the regular matrix when operating in "match" mode. More...

GmSolverMatrix *	globalDebugPrintMatrixFromIndex (int index) const
	Returns the auxiliar matrix used for storing global matrices matching local ones for debug printing purposes when operating in "single debug" mode or the regular matrix when operating in "match" mode.

GmpFemMatrixMode	mode () const
	Returns the mode defining the global matrices stored by the set.

GmpFemMatrixTypes	femTypeFromIndex (int index) const
	Given a matrix index, returns its type as a GmpFemMatrixTypes. Equivalent to GmMatrixSet::typeFromIndex()

Public Member Functions inherited from GmMatrixSet
	GmMatrixSet (int maxMatrixTypes, const QStringList &typeNames)

bool	init (int maxNodes, int numMatrices,...)

bool	adjustSizes (int maxNodes)

void	clear ()

void	prepareSet (int nnodes)

bool	containsType (int type) const

void	setTypeEnabled (int type, bool enabled)

int	numEnabledTypes () const

GmMatrix &	matrix (int type)

GmMatrix &	useMatrix (int type)

GmMatrix &	useMatrix (int type, bool fill, bool sym)

void	setSymmetric (int type, bool symmetric)

void	setFilled (int type, bool filled)

bool	symmetric (int type) const

bool	filledSymmetric () const

bool	filled (int type) const

int	index (int type) const

int	typeFromIndex (int index) const

QString	typeName (int type) const

int	numMatrices () const

const GmMatrix &	matrixFromIndex (int index) const

bool	symmetricFromIndex (int index) const

bool	filledFromIndex (int index) const

Protected Member Functions
int	tindex (GmpFemMatrixTypes type) const
	Returns the index in _globalMatrices for the given type, accounting for the selected matrix mode.

Protected Member Functions inherited from GmMatrixSet
bool	initTypes (int numMatrices, va_list typeList)

Protected Attributes
GmpFemMatrixMode	_mode
	The mode defining how many global matrices are stored.

const GmNumSolver *	_solver
	The solver used to create global matrices.

int	_nthreads
	The configured number of threads that should be used by sparse matrices in batch mode/layout operations.

bool	_parallelSupport
	Do solver matrices support parallel assembling ?

bool	_needsLock
	Do solver matrices needs external locking for parallel assembling ?

GmpFemMatrixTypeData *	_globalMatrices
	Vector storing data for global matrices. Will store one data set per included type in the matrix set when _mode equals MATRIX_MATCH_MODE or a single entry otherwise (even in single debug mode).

GmSolverMatrix **	_printMatrices
	Vector with debug global matrices associated to element local matrices when _mode equals MATRIX_SINGLE_DEBUG_MODE.

Protected Attributes inherited from GmMatrixSet
int	_nTypes

int *	_typeIndex

int *	_indexType

bool *	_typeEnabled

int	_nEnabled

QStringList	_typeNames

int	_nMatrices

int	_maxNodes

GmTLBuffer< double, true >	_matrixMemory

GmTLBuffer< GmMatrix, true >	_matrices

GmTLBuffer< bool, true >	_symmetric

GmTLBuffer< bool, true >	_filled

Detailed Description

A convenience class that builds a GmMatrixSet with the types given by GmpFemMatrixTypes and also associates a global matrix to each element matrix stored in the set (it can also associate a single global matrix, dissociated with matrix types)

If the type of global matrices returned by the numeric solver supports sparse layouts, they will be used and shared between all global matrices in the set.

Member Enumeration Documentation

◆ GmpFemMatrixMode

enum GmpFemMatrixSet::GmpFemMatrixMode

Control mode specifying if the matrix set should have one global matrix per local matrix (MATCH_MODE), a single global matrix independing of the number of local matrices (SINGLE_MODE) or both: a single global matrix + one global matrix per configured local matrix. This last mode is usefull for debuging purposes only.

Enumerator
MATRIX_MATCH_MODE	Set stores one global matrix per local matrix.
MATRIX_SINGLE_MODE	Set stores a single "equivalent" global matrix.
MATRIX_SINGLE_DEBUG_MODE	Set stores an "equivalent" global matrix + another matrix per local matrix for debug purposes.

Member Function Documentation

◆ adjustSizes()

bool GmpFemMatrixSet::adjustSizes	(	int	globalNumDofs,
		int	maxElemDofs,
		bool	sparseLayoutChanged
	)

Adjusts the sizes of local and global matrices according to the given number of dofs. Can be called more than once to update matrix sizes on model changes.

Must be called at least once. If the number of dofs hasn't changed and the layout also hasn't changed, the matrix remains the same. Otherwise it will be cleared and layout data removed.

Parameters

globalNumDofs	The number of free dofs in the system, defining the size of the global matrices.
maxElemDofs	The maximum number of dofs needed by a model element.
sparseLayoutChanged	If true means that the global sparse matrix layout (position of non zero entries) has changed

Returns: Returns true when successfull, false on errors.

◆ beginBatchInsert()

bool GmpFemMatrixSet::beginBatchInsert ( size_t expectedEntries = 0 )

Initializes a batch insert process for all ACTIVE global matrices + the print matrices if necessary.

If the set does not support batch insert, does nothing (but returns true).

◆ clearGlobalMatrices()

void GmpFemMatrixSet::clearGlobalMatrices ( bool keepSparseLayout )

Clears global matrices (or the single matrix), filling them with zeros and marking them as symmetric.

If keepSparseLayout is set to true or if the matrix type does not support layouts, only ACTIVE matrices will be cleared and sparse layout information will be kept (if any). Otherwise, all matrices will be cleared and layout information will be removed.

◆ endBatchInsert()

bool GmpFemMatrixSet::endBatchInsert ( bool discardData )

Finalizes a batch insert process for all ACTIVE global matrices + the print matrices if necessary.

If the set does not support batch insert, does nothing (but returns true). If discardData is set to true, the assembly process did not finished well and the current data is incomplete and should be discarded.

◆ globalDebugPrintMatrix()

GmSolverMatrix* GmpFemMatrixSet::globalDebugPrintMatrix ( GmpFemMatrixTypes type ) const

inline

Returns the auxiliar matrix used for storing global matrices matching local ones for debug printing purposes when operating in "single debug" mode or the regular matrix when operating in "match" mode.

IMPORTANT: The type MUST belong to the set. Can be used only in "single debug" mode or "match" modes.

◆ globalMatrix()

GmSolverMatrix* GmpFemMatrixSet::globalMatrix ( GmpFemMatrixTypes type ) const

inline

Returns a global matrix identified by its type in the set.

IMPORTANT: The type MUST belong to the set. In match mode, it can be any of the local matrix types added to the matrix set. In single and single debug modes, type MUST be equal to GmpFemMatrix_Eq.

◆ initTypes()

bool GmpFemMatrixSet::initTypes	(	const GmNumSolver *	solver,
		GmpFemMatrixMode	mode,
		int	numMatrices,
			...
	)

Informs the matrix set about the local matrix types that will be stored on it. It also creates the set of global matrices tied to the element matrices in the set. The number of effectively stored global matrices follows the given mode.

This function does NOT allocate memory for the matrices. It only prepares the global used structure. Memory allocation is deferred to the adjustSizes() call. Returns false on errors.

IMPORTANT: The matrix set initialization must be followed by a call to adjustSizes().

Parameters

mode	The mode specifying which global matrices should be stored by the set in correspondence to the given set of local matrices.
numMatrices	The number of local matrices stored in the set
...	The variable list of parameters should contain the types of the stored matrices. The number of parameters should be equal to numMatrices. Each value should be a unique type number (between 0 and the maximum number of types given in the constructor)

Returns: Returns true when successfull, false on errors.

◆ setTransposedDofVector()

void GmpFemMatrixSet::setTransposedDofVector	(	GmpFemMatrixTypes	type,
		GmpFemVectorTypes	vectorType
	)

Registers the vector type that should be used to store transposed values when reducing a matrix of the given matrix type, due to fixed dofs.

IMPORTANT: The type MUST belong to the set. In match mode, it can be any of the local vector types added to the vector set. In single and single debug modes, type MUST be equal to GmpFemVector_Eq.

◆ transposedDofVector()

int GmpFemMatrixSet::transposedDofVector ( GmpFemMatrixTypes type ) const

inline

Returns the registered vector type for storing transposed values when reducing a matrix due to fixed dofs for the given matrix type. Will be equal to the value given in the call to setTransposedDofVector() or -1 if that function was not called.

If the function result is different from -1, it can be safely casted to a GmpFemVectorTypes.

IMPORTANT: The type MUST belong to the set. In match mode, it can be any of the local matrix types added to the matrix set. In single and single debug modes, type MUST be equal to GmpFemMatrix_Eq.

◆ transposedDofVectorFromIndex()

int GmpFemMatrixSet::transposedDofVectorFromIndex ( int index ) const

inline

Returns the registered vector type for storing transposed values when reducing a matrix due to fixed dofs for the given matrix index. Will be equal to the value given in the call to setTransposedDofVector() or -1 if that function was not called. Can NOT be used for "single" or "single debug" modes.

If the function result is different from -1, it can be safely casted to a GmpFemVectorTypes.

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

inc/gmpFemMatrixSet.h
src/gmpFemMatrixSet.cpp

Classes

Public Types

Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

Member Enumeration Documentation

◆ GmpFemMatrixMode

Member Function Documentation

◆ adjustSizes()

◆ beginBatchInsert()

◆ clearGlobalMatrices()

◆ endBatchInsert()

◆ globalDebugPrintMatrix()

◆ globalMatrix()

◆ initTypes()

◆ setTransposedDofVector()

◆ transposedDofVector()

◆ transposedDofVectorFromIndex()