mfem::SparseMatrix Class Reference

Data type sparse matrix. More...

#include <sparsemat.hpp>

Inheritance diagram for mfem::SparseMatrix:

Collaboration diagram for mfem::SparseMatrix:

Public Member Functions
	SparseMatrix ()
	Create an empty SparseMatrix. More...
	SparseMatrix (int nrows, int ncols=-1)
	Create a sparse matrix with flexible sparsity structure using a row-wise linked list (LIL) format. More...
	SparseMatrix (int *i, int *j, double *data, int m, int n)
	Create a sparse matrix in CSR format. Ownership of i, j, and data is transferred to the SparseMatrix. More...
	SparseMatrix (int *i, int *j, double *data, int m, int n, bool ownij, bool owna, bool issorted)
	Create a sparse matrix in CSR format. Ownership of i, j, and data is optionally transferred to the SparseMatrix. More...
	SparseMatrix (int nrows, int ncols, int rowsize)
	Create a sparse matrix in CSR format where each row has space allocated for exactly rowsize entries. More...
	SparseMatrix (const SparseMatrix &mat, bool copy_graph=true, MemoryType mt=MemoryType::PRESERVE)
	Copy constructor (deep copy). More...
	SparseMatrix (const Vector &v)
	Create a SparseMatrix with diagonal v, i.e. A = Diag(v) More...
void	UseCuSparse (bool useCuSparse_=true)
SparseMatrix &	operator= (const SparseMatrix &rhs)
	Assignment operator: deep copy. More...
void	MakeRef (const SparseMatrix &master)
	Clear the contents of the SparseMatrix and make it a reference to master. More...
int	Size () const
	For backward compatibility, define Size() to be synonym of Height(). More...
void	Clear ()
	Clear the contents of the SparseMatrix. More...
void	ClearCuSparse ()
	Clear the CuSparse descriptors. This must be called after releasing the device memory of A. More...
bool	Empty () const
	Check if the SparseMatrix is empty. More...
int *	GetI ()
	Return the array I. More...
const int *	GetI () const
	Return the array I, const version. More...
int *	GetJ ()
	Return the array J. More...
const int *	GetJ () const
	Return the array J, const version. More...
double *	GetData ()
	Return the element data, i.e. the array A. More...
const double *	GetData () const
	Return the element data, i.e. the array A, const version. More...
Memory< int > &	GetMemoryI ()
const Memory< int > &	GetMemoryI () const
const int *	ReadI (bool on_dev=true) const
int *	WriteI (bool on_dev=true)
int *	ReadWriteI (bool on_dev=true)
const int *	HostReadI () const
int *	HostWriteI ()
int *	HostReadWriteI ()
Memory< int > &	GetMemoryJ ()
const Memory< int > &	GetMemoryJ () const
const int *	ReadJ (bool on_dev=true) const
int *	WriteJ (bool on_dev=true)
int *	ReadWriteJ (bool on_dev=true)
const int *	HostReadJ () const
int *	HostWriteJ ()
int *	HostReadWriteJ ()
Memory< double > &	GetMemoryData ()
const Memory< double > &	GetMemoryData () const
const double *	ReadData (bool on_dev=true) const
double *	WriteData (bool on_dev=true)
double *	ReadWriteData (bool on_dev=true)
const double *	HostReadData () const
double *	HostWriteData ()
double *	HostReadWriteData ()
int	RowSize (const int i) const
	Returns the number of elements in row i. More...
int	MaxRowSize () const
	Returns the maximum number of elements among all rows. More...
int *	GetRowColumns (const int row)
	Return a pointer to the column indices in a row. More...
const int *	GetRowColumns (const int row) const
	Return a pointer to the column indices in a row, const version. More...
double *	GetRowEntries (const int row)
	Return a pointer to the entries in a row. More...
const double *	GetRowEntries (const int row) const
	Return a pointer to the entries in a row, const version. More...
void	SetWidth (int width_=-1)
	Change the width of a SparseMatrix. More...
int	ActualWidth () const
	Returns the actual Width of the matrix. More...
void	SortColumnIndices ()
	Sort the column indices corresponding to each row. More...
void	MoveDiagonalFirst ()
	Move the diagonal entry to the first position in each row, preserving the order of the rest of the columns. More...
virtual double &	Elem (int i, int j)
	Returns reference to a_{ij}. More...
virtual const double &	Elem (int i, int j) const
	Returns constant reference to a_{ij}. More...
double &	operator() (int i, int j)
	Returns reference to A[i][j]. More...
const double &	operator() (int i, int j) const
	Returns reference to A[i][j]. More...
void	GetDiag (Vector &d) const
	Returns the Diagonal of A. More...
DenseMatrix *	ToDenseMatrix () const
	Produces a DenseMatrix from a SparseMatrix. More...
void	ToDenseMatrix (DenseMatrix &B) const
	Produces a DenseMatrix from a SparseMatrix. More...
virtual MemoryClass	GetMemoryClass () const
	Return the MemoryClass preferred by the Operator. More...
virtual void	Mult (const Vector &x, Vector &y) const
	Matrix vector multiplication. More...
void	AddMult (const Vector &x, Vector &y, const double a=1.0) const
	y += A * x (default) or y += a * A * x More...
void	MultTranspose (const Vector &x, Vector &y) const
	Multiply a vector with the transposed matrix. y = At * x. More...
void	AddMultTranspose (const Vector &x, Vector &y, const double a=1.0) const
	y += At * x (default) or y += a * At * x More...
void	BuildTranspose () const
	Build and store internally the transpose of this matrix which will be used in the methods AddMultTranspose() and MultTranspose(). More...
void	ResetTranspose () const
void	PartMult (const Array< int > &rows, const Vector &x, Vector &y) const
void	PartAddMult (const Array< int > &rows, const Vector &x, Vector &y, const double a=1.0) const
void	BooleanMult (const Array< int > &x, Array< int > &y) const
	y = A * x, treating all entries as booleans (zero=false, nonzero=true). More...
void	BooleanMultTranspose (const Array< int > &x, Array< int > &y) const
	y = At * x, treating all entries as booleans (zero=false, nonzero=true). More...
void	AbsMult (const Vector &x, Vector &y) const
	y = |A| * x, using entry-wise absolute values of matrix A More...
void	AbsMultTranspose (const Vector &x, Vector &y) const
	y = |At| * x, using entry-wise absolute values of the transpose of matrix A More...
double	InnerProduct (const Vector &x, const Vector &y) const
	Compute y^t A x. More...
void	GetRowSums (Vector &x) const
	For all i compute \( x_i = \sum_j A_{ij} \). More...
double	GetRowNorml1 (int irow) const
	For i = irow compute \( x_i = \sum_j | A_{i, j} | \). More...
virtual MatrixInverse *	Inverse () const
	This virtual method is not supported: it always returns NULL. More...
void	EliminateRow (int row, const double sol, Vector &rhs)
	Eliminates a column from the transpose matrix. More...
void	EliminateRow (int row, DiagonalPolicy dpolicy=DIAG_ZERO)
	Eliminates a row from the matrix. More...
void	EliminateCol (int col, DiagonalPolicy dpolicy=DIAG_ZERO)
	Eliminates the column col from the matrix. More...
void	EliminateCols (const Array< int > &cols, const Vector *x=NULL, Vector *b=NULL)
	Eliminate all columns i for which cols[i] != 0. More...
void	EliminateCols (const Array< int > &col_marker, SparseMatrix &Ae)
	Similar to EliminateCols + save the eliminated entries into Ae so that (*this) + Ae is equal to the original matrix. More...
void	EliminateRowCol (int rc, const double sol, Vector &rhs, DiagonalPolicy dpolicy=DIAG_ONE)
	Eliminate row rc and column rc and modify the rhs using sol. More...
void	EliminateRowColMultipleRHS (int rc, const Vector &sol, DenseMatrix &rhs, DiagonalPolicy dpolicy=DIAG_ONE)
	Similar to EliminateRowCol(int, const double, Vector &, DiagonalPolicy), but multiple values for eliminated unknowns are accepted, and accordingly multiple right-hand-sides are used. More...
void	EliminateRowColDiag (int rc, double value)
	Perform elimination and set the diagonal entry to the given value. More...
void	EliminateRowCol (int rc, DiagonalPolicy dpolicy=DIAG_ONE)
	Eliminate row rc and column rc. More...
void	EliminateRowCol (int rc, SparseMatrix &Ae, DiagonalPolicy dpolicy=DIAG_ONE)
	Similar to EliminateRowCol(int, DiagonalPolicy) + save the eliminated entries into Ae so that (*this) + Ae is equal to the original matrix. More...
void	SetDiagIdentity ()
	If a row contains only one diag entry of zero, set it to 1. More...
virtual void	EliminateZeroRows (const double threshold=1e-12)
	If a row contains only zeros, set its diagonal to 1. More...
void	Gauss_Seidel_forw (const Vector &x, Vector &y) const
	Gauss-Seidel forward and backward iterations over a vector x. More...
void	Gauss_Seidel_back (const Vector &x, Vector &y) const
double	GetJacobiScaling () const
	Determine appropriate scaling for Jacobi iteration. More...
void	Jacobi (const Vector &b, const Vector &x0, Vector &x1, double sc) const
void	DiagScale (const Vector &b, Vector &x, double sc=1.0) const
void	Jacobi2 (const Vector &b, const Vector &x0, Vector &x1, double sc=1.0) const
void	Jacobi3 (const Vector &b, const Vector &x0, Vector &x1, double sc=1.0) const
virtual void	Finalize (int skip_zeros=1)
	Finalize the matrix initialization, switching the storage format from LIL to CSR. More...
void	Finalize (int skip_zeros, bool fix_empty_rows)
	A slightly more general version of the Finalize(int) method. More...
bool	Finalized () const
	Returns whether or not CSR format has been finalized. More...
bool	ColumnsAreSorted () const
	Returns whether or not the columns are sorted. More...
void	Threshold (double tol, bool fix_empty_rows=false)
	Remove entries smaller in absolute value than a given tolerance tol. If fix_empty_rows is true, a zero value is inserted in the diagonal entry (for square matrices only) More...
void	GetBlocks (Array2D< SparseMatrix * > &blocks) const
void	GetSubMatrix (const Array< int > &rows, const Array< int > &cols, DenseMatrix &subm) const
void	SetColPtr (const int row) const
	Initialize the SparseMatrix for fast access to the entries of the given row which becomes the "current row". More...
void	ClearColPtr () const
	Reset the "current row" set by calling SetColPtr(). This method must be called between any two calls to SetColPtr(). More...
double &	SearchRow (const int col)
	Perform a fast search for an entry in the "current row". See SetColPtr(). More...
void	_Add_ (const int col, const double a)
	Add a value to an entry in the "current row". See SetColPtr(). More...
void	_Set_ (const int col, const double a)
	Set an entry in the "current row". See SetColPtr(). More...
double	_Get_ (const int col) const
	Read the value of an entry in the "current row". See SetColPtr(). More...
double &	SearchRow (const int row, const int col)
void	_Add_ (const int row, const int col, const double a)
void	_Set_ (const int row, const int col, const double a)
void	Set (const int i, const int j, const double a)
void	Add (const int i, const int j, const double a)
void	SetSubMatrix (const Array< int > &rows, const Array< int > &cols, const DenseMatrix &subm, int skip_zeros=1)
void	SetSubMatrixTranspose (const Array< int > &rows, const Array< int > &cols, const DenseMatrix &subm, int skip_zeros=1)
void	AddSubMatrix (const Array< int > &rows, const Array< int > &cols, const DenseMatrix &subm, int skip_zeros=1)
bool	RowIsEmpty (const int row) const
virtual int	GetRow (const int row, Array< int > &cols, Vector &srow) const
	Extract all column indices and values from a given row. More...
void	SetRow (const int row, const Array< int > &cols, const Vector &srow)
void	AddRow (const int row, const Array< int > &cols, const Vector &srow)
void	ScaleRow (const int row, const double scale)
void	ScaleRows (const Vector &sl)
	this = diag(sl) * this; More...
void	ScaleColumns (const Vector &sr)
	this = this * diag(sr); More...
SparseMatrix &	operator+= (const SparseMatrix &B)
	Add the sparse matrix 'B' to '*this'. This operation will cause an error if '*this' is finalized and 'B' has larger sparsity pattern. More...
void	Add (const double a, const SparseMatrix &B)
	Add the sparse matrix 'B' scaled by the scalar 'a' into '*this'. Only entries in the sparsity pattern of '*this' are added. More...
SparseMatrix &	operator= (double a)
SparseMatrix &	operator*= (double a)
void	Print (std::ostream &out=mfem::out, int width_=4) const
	Prints matrix to stream out. More...
void	PrintMatlab (std::ostream &out=mfem::out) const
	Prints matrix in matlab format. More...
void	PrintMM (std::ostream &out=mfem::out) const
	Prints matrix in Matrix Market sparse format. More...
void	PrintCSR (std::ostream &out) const
	Prints matrix to stream out in hypre_CSRMatrix format. More...
void	PrintCSR2 (std::ostream &out) const
	Prints a sparse matrix to stream out in CSR format. More...
void	PrintInfo (std::ostream &out) const
	Print various sparse matrix statistics. More...
double	IsSymmetric () const
	Returns max_{i,j} |(i,j)-(j,i)| for a finalized matrix. More...
void	Symmetrize ()
	(*this) = 1/2 ((*this) + (*this)^t) More...
virtual int	NumNonZeroElems () const
	Returns the number of the nonzero elements in the matrix. More...
double	MaxNorm () const
int	CountSmallElems (double tol) const
	Count the number of entries with |a_ij| <= tol. More...
int	CheckFinite () const
	Count the number of entries that are NOT finite, i.e. Inf or Nan. More...
void	SetGraphOwner (bool ownij)
	Set the graph ownership flag (I and J arrays). More...
void	SetDataOwner (bool owna)
	Set the data ownership flag (A array). More...
bool	OwnsGraph () const
	Get the graph ownership flag (I and J arrays). More...
bool	OwnsData () const
	Get the data ownership flag (A array). More...
void	LoseData ()
	Lose the ownership of the graph (I, J) and data (A) arrays. More...
void	Swap (SparseMatrix &other)
virtual	~SparseMatrix ()
	Destroys sparse matrix. More...
Type	GetType () const
Public Member Functions inherited from mfem::AbstractSparseMatrix
	AbstractSparseMatrix (int s=0)
	Creates a square matrix of the given size. More...
	AbstractSparseMatrix (int h, int w)
	Creates a matrix of the given height and width. More...
virtual	~AbstractSparseMatrix ()
	Destroys AbstractSparseMatrix. More...
Public Member Functions inherited from mfem::Matrix
	Matrix (int s)
	Creates a square matrix of size s. More...
	Matrix (int h, int w)
	Creates a matrix of the given height and width. More...
bool	IsSquare () const
	Returns whether the matrix is a square matrix. More...
virtual	~Matrix ()
	Destroys matrix. More...
Public Member Functions inherited from mfem::Operator
void	InitTVectors (const Operator *Po, const Operator *Ri, const Operator *Pi, Vector &x, Vector &b, Vector &X, Vector &B) const
	Initializes memory for true vectors of linear system. More...
	Operator (int s=0)
	Construct a square Operator with given size s (default 0). More...
	Operator (int h, int w)
	Construct an Operator with the given height (output size) and width (input size). More...
int	Height () const
	Get the height (size of output) of the Operator. Synonym with NumRows(). More...
int	NumRows () const
	Get the number of rows (size of output) of the Operator. Synonym with Height(). More...
int	Width () const
	Get the width (size of input) of the Operator. Synonym with NumCols(). More...
int	NumCols () const
	Get the number of columns (size of input) of the Operator. Synonym with Width(). More...
virtual Operator &	GetGradient (const Vector &x) const
	Evaluate the gradient operator at the point x. The default behavior in class Operator is to generate an error. More...
virtual void	AssembleDiagonal (Vector &diag) const
	Computes the diagonal entries into diag. Typically, this operation only makes sense for linear Operators. In some cases, only an approximation of the diagonal is computed. More...
virtual const Operator *	GetProlongation () const
	Prolongation operator from linear algebra (linear system) vectors, to input vectors for the operator. NULL means identity. More...
virtual const Operator *	GetRestriction () const
	Restriction operator from input vectors for the operator to linear algebra (linear system) vectors. NULL means identity. More...
virtual const Operator *	GetOutputProlongation () const
	Prolongation operator from linear algebra (linear system) vectors, to output vectors for the operator. NULL means identity. More...
virtual const Operator *	GetOutputRestrictionTranspose () const
	Transpose of GetOutputRestriction, directly available in this form to facilitate matrix-free RAP-type operators. More...
virtual const Operator *	GetOutputRestriction () const
	Restriction operator from output vectors for the operator to linear algebra (linear system) vectors. NULL means identity. More...
void	FormLinearSystem (const Array< int > &ess_tdof_list, Vector &x, Vector &b, Operator *&A, Vector &X, Vector &B, int copy_interior=0)
	Form a constrained linear system using a matrix-free approach. More...
void	FormRectangularLinearSystem (const Array< int > &trial_tdof_list, const Array< int > &test_tdof_list, Vector &x, Vector &b, Operator *&A, Vector &X, Vector &B)
	Form a column-constrained linear system using a matrix-free approach. More...
virtual void	RecoverFEMSolution (const Vector &X, const Vector &b, Vector &x)
	Reconstruct a solution vector x (e.g. a GridFunction) from the solution X of a constrained linear system obtained from Operator::FormLinearSystem() or Operator::FormRectangularLinearSystem(). More...
void	FormSystemOperator (const Array< int > &ess_tdof_list, Operator *&A)
	Return in A a parallel (on truedofs) version of this square operator. More...
void	FormRectangularSystemOperator (const Array< int > &trial_tdof_list, const Array< int > &test_tdof_list, Operator *&A)
	Return in A a parallel (on truedofs) version of this rectangular operator (including constraints). More...
void	FormDiscreteOperator (Operator *&A)
	Return in A a parallel (on truedofs) version of this rectangular operator. More...
void	PrintMatlab (std::ostream &out, int n=0, int m=0) const
	Prints operator with input size n and output size m in Matlab format. More...
virtual	~Operator ()
	Virtual destructor. More...
Type	GetType () const
	Return the type ID of the Operator class. More...

Protected Types
typedef MemAlloc< RowNode, 1024 >	RowNodeAlloc

Protected Member Functions
void	Destroy ()
void	SetEmpty ()
void	InitCuSparse ()
Protected Member Functions inherited from mfem::Operator
void	FormConstrainedSystemOperator (const Array< int > &ess_tdof_list, ConstrainedOperator *&Aout)
	see FormSystemOperator() More...
void	FormRectangularConstrainedSystemOperator (const Array< int > &trial_tdof_list, const Array< int > &test_tdof_list, RectangularConstrainedOperator *&Aout)
	see FormRectangularSystemOperator() More...
Operator *	SetupRAP (const Operator *Pi, const Operator *Po)
	Returns RAP Operator of this, using input/output Prolongation matrices Pi corresponds to "P", Po corresponds to "Rt". More...

Protected Attributes
RowNode **	Rows
	Array of linked lists, one for every row. This array represents the linked list (LIL) storage format. More...
int	current_row
int *	ColPtrJ
RowNode **	ColPtrNode
SparseMatrix *	At
	Transpose of A. Owned. Used to perform MultTranspose() on devices. More...
RowNodeAlloc *	NodesMem
bool	isSorted
	Are the columns sorted already. More...
bool	useCuSparse {true}
cusparseStatus_t	status
cusparseMatDescr_t	descr =0
bool	initBuffers {false}
cusparseSpMatDescr_t	matA_descr
cusparseDnVecDescr_t	vecX_descr
cusparseDnVecDescr_t	vecY_descr
cusparseMatDescr_t	matA_descr
Arrays used by the CSR storage format.
Memory< int >	I
	Array with size (height+1) containing the row offsets. More...
Memory< int >	J
	Array with size I[height], containing the column indices for all matrix entries, as indexed by the I array. More...
Memory< double >	A
	Array with size I[height], containing the actual entries of the sparse matrix, as indexed by the I array. More...
Protected Attributes inherited from mfem::Operator
int	height
	Dimension of the output / number of rows in the matrix. More...
int	width
	Dimension of the input / number of columns in the matrix. More...

Static Protected Attributes
static cusparseHandle_t	handle = nullptr
static int	SparseMatrixCount = 0
static size_t	bufferSize = 0
static void *	dBuffer = nullptr

Additional Inherited Members
Public Types inherited from mfem::Operator
enum	DiagonalPolicy { DIAG_ZERO, DIAG_ONE, DIAG_KEEP }
	Defines operator diagonal policy upon elimination of rows and/or columns. More...
enum	Type {   ANY_TYPE, MFEM_SPARSEMAT, Hypre_ParCSR, PETSC_MATAIJ,   PETSC_MATIS, PETSC_MATSHELL, PETSC_MATNEST, PETSC_MATHYPRE,   PETSC_MATGENERIC, Complex_Operator, MFEM_ComplexSparseMat, Complex_Hypre_ParCSR }
	Enumeration defining IDs for some classes derived from Operator. More...

Detailed Description

Data type sparse matrix.

Definition at line 41 of file sparsemat.hpp.

Member Typedef Documentation

typedef MemAlloc<RowNode, 1024> mfem::SparseMatrix::RowNodeAlloc

protected

Definition at line 74 of file sparsemat.hpp.

Constructor & Destructor Documentation

mfem::SparseMatrix::SparseMatrix ( )

inline

Create an empty SparseMatrix.

Definition at line 108 of file sparsemat.hpp.

mfem::SparseMatrix::SparseMatrix ( int  nrows, int  ncols = -1  )

explicit

Create a sparse matrix with flexible sparsity structure using a row-wise linked list (LIL) format.

New entries are added as needed by methods like AddSubMatrix(), SetSubMatrix(), etc. Calling Finalize() will convert the SparseMatrix to the more compact compressed sparse row (CSR) format.

Definition at line 72 of file sparsemat.cpp.

mfem::SparseMatrix::SparseMatrix	(	int *	i,
		int *	j,
		double *	data,
		int	m,
		int	n
	)

Create a sparse matrix in CSR format. Ownership of i, j, and data is transferred to the SparseMatrix.

Definition at line 98 of file sparsemat.cpp.

mfem::SparseMatrix::SparseMatrix	(	int *	i,
		int *	j,
		double *	data,
		int	m,
		int	n,
		bool	ownij,
		bool	owna,
		bool	issorted
	)

Create a sparse matrix in CSR format. Ownership of i, j, and data is optionally transferred to the SparseMatrix.

If the parameter data is NULL, then the internal A array is allocated by this constructor (initializing it with zeros and taking ownership, regardless of the parameter owna).

Definition at line 117 of file sparsemat.cpp.

mfem::SparseMatrix::SparseMatrix	(	int	nrows,
		int	ncols,
		int	rowsize
	)

Create a sparse matrix in CSR format where each row has space allocated for exactly rowsize entries.

SetRow() can then be called or the I, J, A arrays can be used directly.

Definition at line 150 of file sparsemat.cpp.

mfem::SparseMatrix::SparseMatrix	(	const SparseMatrix &	mat,
		bool	copy_graph = true,
		MemoryType	mt = MemoryType::PRESERVE
	)

Copy constructor (deep copy).

If mat is finalized and copy_graph is false, the I and J arrays will use a shallow copy (copy the pointers only) without transferring ownership. If mt is MemoryType::PRESERVE the memory type of the resulting SparseMatrix's I, J, and A arrays will be the same as mat, otherwise the type will be mt for those arrays that are deep copied.

Definition at line 173 of file sparsemat.cpp.

mfem::SparseMatrix::SparseMatrix

(

const Vector &

v

)

Create a SparseMatrix with diagonal v, i.e. A = Diag(v)

Definition at line 241 of file sparsemat.cpp.

virtual mfem::SparseMatrix::~SparseMatrix ( )

inlinevirtual

Destroys sparse matrix.

Definition at line 627 of file sparsemat.hpp.

Member Function Documentation

void mfem::SparseMatrix::_Add_ ( const int  col, const double  a  )

inline

Add a value to an entry in the "current row". See SetColPtr().

Definition at line 506 of file sparsemat.hpp.

void mfem::SparseMatrix::_Add_ ( const int  row, const int  col, const double  a  )

inline

Definition at line 515 of file sparsemat.hpp.

double mfem::SparseMatrix::_Get_ ( const int  col ) const

inline

Read the value of an entry in the "current row". See SetColPtr().

Definition at line 817 of file sparsemat.hpp.

void mfem::SparseMatrix::_Set_ ( const int  col, const double  a  )

inline

Set an entry in the "current row". See SetColPtr().

Definition at line 509 of file sparsemat.hpp.

void mfem::SparseMatrix::_Set_ ( const int  row, const int  col, const double  a  )

inline

Definition at line 517 of file sparsemat.hpp.

void mfem::SparseMatrix::AbsMult	(	const Vector &	x,
		Vector &	y
	)		const

y = |A| * x, using entry-wise absolute values of matrix A

Definition at line 933 of file sparsemat.cpp.

void mfem::SparseMatrix::AbsMultTranspose	(	const Vector &	x,
		Vector &	y
	)		const

y = |At| * x, using entry-wise absolute values of the transpose of matrix A

Definition at line 982 of file sparsemat.cpp.

int mfem::SparseMatrix::ActualWidth

(

)

const

Returns the actual Width of the matrix.

This method can be called for matrices finalized or not.

Definition at line 3257 of file sparsemat.cpp.

void mfem::SparseMatrix::Add	(	const int	i,
		const int	j,
		const double	a
	)

Definition at line 2577 of file sparsemat.cpp.

void mfem::SparseMatrix::Add	(	const double	a,
		const SparseMatrix &	B
	)

Add the sparse matrix 'B' scaled by the scalar 'a' into '*this'. Only entries in the sparsity pattern of '*this' are added.

Definition at line 2961 of file sparsemat.cpp.

void mfem::SparseMatrix::AddMult ( const Vector &  x, Vector &  y, const double  a = 1.0  ) const

virtual

y += A * x (default) or y += a * A * x

Implements mfem::AbstractSparseMatrix.

Definition at line 620 of file sparsemat.cpp.

void mfem::SparseMatrix::AddMultTranspose ( const Vector &  x, Vector &  y, const double  a = 1.0  ) const

virtual

y += At * x (default) or y += a * At * x

Implements mfem::AbstractSparseMatrix.

Definition at line 777 of file sparsemat.cpp.

void mfem::SparseMatrix::AddRow	(	const int	row,
		const Array< int > &	cols,
		const Vector &	srow
	)

Definition at line 2812 of file sparsemat.cpp.

void mfem::SparseMatrix::AddSubMatrix	(	const Array< int > &	rows,
		const Array< int > &	cols,
		const DenseMatrix &	subm,
		int	skip_zeros = 1
	)

Insert the DenseMatrix into this SparseMatrix at the specified rows and columns. If skip_zeros==0 , all entries from the DenseMatrix are added including zeros. If skip_zeros==2 , no zeros are added to the SparseMatrix regardless of their position in the matrix. Otherwise, the default skip_zeros behavior is to omit the zero from the SparseMatrix unless it would break the symmetric structure of the SparseMatrix.

Definition at line 2512 of file sparsemat.cpp.

void mfem::SparseMatrix::BooleanMult	(	const Array< int > &	x,
		Array< int > &	y
	)		const

y = A * x, treating all entries as booleans (zero=false, nonzero=true).

The actual values stored in the data array, A, are not used - this means that all entries in the sparsity pattern are considered to be true by this method.

Definition at line 880 of file sparsemat.cpp.

void mfem::SparseMatrix::BooleanMultTranspose	(	const Array< int > &	x,
		Array< int > &	y
	)		const

y = At * x, treating all entries as booleans (zero=false, nonzero=true).

The actual values stored in the data array, A, are not used - this means that all entries in the sparsity pattern are considered to be true by this method.

Definition at line 910 of file sparsemat.cpp.

void mfem::SparseMatrix::BuildTranspose

(

)

const

Build and store internally the transpose of this matrix which will be used in the methods AddMultTranspose() and MultTranspose().

If this method has been called, the internal transpose matrix will be used to perform the action of the transpose matrix in AddMultTranspose(), and MultTranspose().

Warning: any changes in this matrix will invalidate the internal transpose. To rebuild the transpose, call ResetTranspose() followed by a call to this method. If the internal transpose is already built, this method has no effect.

When any non-default backend is enabled, i.e. Device::IsEnabled() is true, the methods AddMultTranspose(), and MultTranspose(), require the internal transpose to be built. If that is not the case (i.e. the internal transpose is not built), these methods will raise an error with an appropriate message pointing to this method. When using the default backend, calling this method is optional.

This method can only be used when the sparse matrix is finalized.

Definition at line 822 of file sparsemat.cpp.

int mfem::SparseMatrix::CheckFinite

(

)

const

Count the number of entries that are NOT finite, i.e. Inf or Nan.

Definition at line 1518 of file sparsemat.cpp.

void mfem::SparseMatrix::Clear ( )

inline

Clear the contents of the SparseMatrix.

Definition at line 170 of file sparsemat.hpp.

void mfem::SparseMatrix::ClearColPtr ( ) const

inline

Reset the "current row" set by calling SetColPtr(). This method must be called between any two calls to SetColPtr().

Definition at line 771 of file sparsemat.hpp.

void mfem::SparseMatrix::ClearCuSparse

(

)

Clear the CuSparse descriptors. This must be called after releasing the device memory of A.

Definition at line 55 of file sparsemat.cpp.

bool mfem::SparseMatrix::ColumnsAreSorted ( ) const

inline

Returns whether or not the columns are sorted.

Definition at line 475 of file sparsemat.hpp.

int mfem::SparseMatrix::CountSmallElems

(

double

tol

)

const

Count the number of entries with |a_ij| <= tol.

Definition at line 1490 of file sparsemat.cpp.

void mfem::SparseMatrix::Destroy ( )

protected

Definition at line 3224 of file sparsemat.cpp.

void mfem::SparseMatrix::DiagScale	(	const Vector &	b,
		Vector &	x,
		double	sc = 1.0
	)		const

Definition at line 2427 of file sparsemat.cpp.

double & mfem::SparseMatrix::Elem ( int  i, int  j  )

virtual

Returns reference to a_{ij}.

Implements mfem::Matrix.

Definition at line 481 of file sparsemat.cpp.

const double & mfem::SparseMatrix::Elem ( int  i, int  j  ) const

virtual

Returns constant reference to a_{ij}.

Implements mfem::Matrix.

Definition at line 486 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateCol	(	int	col,
		DiagonalPolicy	dpolicy = DIAG_ZERO
	)

Eliminates the column col from the matrix.

• If dpolicy = DIAG_ZERO, all entries in the column will be set to 0.
• If dpolicy = DIAG_ONE (matrix must be square), the diagonal entry will be set equal to 1 and all other entries in the column to 0.
• The policy DIAG_KEEP is not supported.

Definition at line 1595 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateCols	(	const Array< int > &	cols,
		const Vector *	x = NULL,
		Vector *	b = NULL
	)

Eliminate all columns i for which cols[i] != 0.

Elimination of a column means that all entries in the column are set to zero. In addition, if the pointers x and b are not NULL, the eliminated matrix entries are multiplied by the corresponding solution value in *x and subtracted from the r.h.s. vector, *b.

Definition at line 1635 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateCols	(	const Array< int > &	col_marker,
		SparseMatrix &	Ae
	)

Similar to EliminateCols + save the eliminated entries into Ae so that (*this) + Ae is equal to the original matrix.

Definition at line 1674 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateRow	(	int	row,
		const double	sol,
		Vector &	rhs
	)

Eliminates a column from the transpose matrix.

Definition at line 1547 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateRow	(	int	row,
		DiagonalPolicy	dpolicy = DIAG_ZERO
	)

Eliminates a row from the matrix.

• If dpolicy = DIAG_ZERO, all the entries in the row will be set to 0.
• If dpolicy = DIAG_ONE (matrix must be square), the diagonal entry will be set equal to 1 and all other entries in the row to 0.
• The policy DIAG_KEEP is not supported.

Definition at line 1563 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateRowCol	(	int	rc,
		const double	sol,
		Vector &	rhs,
		DiagonalPolicy	dpolicy = DIAG_ONE
	)

Eliminate row rc and column rc and modify the rhs using sol.

Eliminates the column rc to the rhs, deletes the row rc and replaces the element (rc,rc) with 1.0; assumes that element (i,rc) is assembled if and only if the element (rc,i) is assembled. By default, elements (rc,rc) are set to 1.0, although this behavior can be adjusted by changing the dpolicy parameter.

Definition at line 1708 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateRowCol	(	int	rc,
		DiagonalPolicy	dpolicy = DIAG_ONE
	)

Eliminate row rc and column rc.

Definition at line 1928 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateRowCol	(	int	rc,
		SparseMatrix &	Ae,
		DiagonalPolicy	dpolicy = DIAG_ONE
	)

Similar to EliminateRowCol(int, DiagonalPolicy) + save the eliminated entries into Ae so that (*this) + Ae is equal to the original matrix.

Definition at line 2072 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateRowColDiag	(	int	rc,
		double	value
	)

Perform elimination and set the diagonal entry to the given value.

Definition at line 2009 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateRowColMultipleRHS	(	int	rc,
		const Vector &	sol,
		DenseMatrix &	rhs,
		DiagonalPolicy	dpolicy = DIAG_ONE
	)

Similar to EliminateRowCol(int, const double, Vector &, DiagonalPolicy), but multiple values for eliminated unknowns are accepted, and accordingly multiple right-hand-sides are used.

Definition at line 1804 of file sparsemat.cpp.

void mfem::SparseMatrix::EliminateZeroRows ( const double  threshold = 1e-12 )

virtual

If a row contains only zeros, set its diagonal to 1.

Implements mfem::AbstractSparseMatrix.

Definition at line 2177 of file sparsemat.cpp.

bool mfem::SparseMatrix::Empty ( ) const

inline

Check if the SparseMatrix is empty.

Definition at line 177 of file sparsemat.hpp.

virtual void mfem::SparseMatrix::Finalize ( int  skip_zeros = 1 )

inlinevirtual

Finalize the matrix initialization, switching the storage format from LIL to CSR.

This method should be called once, after the matrix has been initialized. Internally, this method converts the matrix from row-wise linked list (LIL) format into CSR (compressed sparse row) format.

Reimplemented from mfem::Matrix.

Definition at line 467 of file sparsemat.hpp.

void mfem::SparseMatrix::Finalize	(	int	skip_zeros,
		bool	fix_empty_rows
	)

A slightly more general version of the Finalize(int) method.

Definition at line 1177 of file sparsemat.cpp.

bool mfem::SparseMatrix::Finalized ( ) const

inline

Returns whether or not CSR format has been finalized.

Definition at line 473 of file sparsemat.hpp.

void mfem::SparseMatrix::Gauss_Seidel_back	(	const Vector &	x,
		Vector &	y
	)		const

Definition at line 2280 of file sparsemat.cpp.

void mfem::SparseMatrix::Gauss_Seidel_forw	(	const Vector &	x,
		Vector &	y
	)		const

Gauss-Seidel forward and backward iterations over a vector x.

Definition at line 2196 of file sparsemat.cpp.

void mfem::SparseMatrix::GetBlocks

(

Array2D< SparseMatrix * > &

blocks

)

const

Split the matrix into M x N blocks of sparse matrices in CSR format. The 'blocks' array is M x N (i.e. M and N are determined by its dimensions) and its entries are overwritten by the new blocks.

Definition at line 1292 of file sparsemat.cpp.

double* mfem::SparseMatrix::GetData ( )

inline

Return the element data, i.e. the array A.

Definition at line 190 of file sparsemat.hpp.

const double* mfem::SparseMatrix::GetData ( ) const

inline

Return the element data, i.e. the array A, const version.

Definition at line 192 of file sparsemat.hpp.

void mfem::SparseMatrix::GetDiag

(

Vector &

d

)

const

Returns the Diagonal of A.

Definition at line 549 of file sparsemat.cpp.

int* mfem::SparseMatrix::GetI ( )

inline

Return the array I.

Definition at line 180 of file sparsemat.hpp.

const int* mfem::SparseMatrix::GetI ( ) const

inline

Return the array I, const version.

Definition at line 182 of file sparsemat.hpp.

int* mfem::SparseMatrix::GetJ ( )

inline

Return the array J.

Definition at line 185 of file sparsemat.hpp.

const int* mfem::SparseMatrix::GetJ ( ) const

inline

Return the array J, const version.

Definition at line 187 of file sparsemat.hpp.

double mfem::SparseMatrix::GetJacobiScaling

(

)

const

Determine appropriate scaling for Jacobi iteration.

Definition at line 2363 of file sparsemat.cpp.

virtual MemoryClass mfem::SparseMatrix::GetMemoryClass ( ) const

inlinevirtual

Return the MemoryClass preferred by the Operator.

This is the MemoryClass that will be used to access the input and output vectors in the Mult() and MultTranspose() methods.

For example, classes using the MFEM_FORALL macro for implementation can return the value returned by Device::GetMemoryClass().

The default implementation of this method in class Operator returns MemoryClass::HOST.

Reimplemented from mfem::Operator.

Definition at line 300 of file sparsemat.hpp.

Memory<double>& mfem::SparseMatrix::GetMemoryData ( )

inline

Definition at line 227 of file sparsemat.hpp.

const Memory<double>& mfem::SparseMatrix::GetMemoryData ( ) const

inline

Definition at line 228 of file sparsemat.hpp.

Memory<int>& mfem::SparseMatrix::GetMemoryI ( )

inline

Definition at line 195 of file sparsemat.hpp.

const Memory<int>& mfem::SparseMatrix::GetMemoryI ( ) const

inline

Definition at line 196 of file sparsemat.hpp.

Memory<int>& mfem::SparseMatrix::GetMemoryJ ( )

inline

Definition at line 211 of file sparsemat.hpp.

const Memory<int>& mfem::SparseMatrix::GetMemoryJ ( ) const

inline

Definition at line 212 of file sparsemat.hpp.

int mfem::SparseMatrix::GetRow ( const int  row, Array< int > &  cols, Vector &  srow  ) const

virtual

Extract all column indices and values from a given row.

If the matrix is finalized (i.e. in CSR format), cols and srow will simply be references to the specific portion of the J and A arrays. As required by the AbstractSparseMatrix interface this method returns:

• 0, if cols and srow are copies of the values in the matrix, i.e. when the matrix is open.
• 1, if cols and srow are views of the values in the matrix, i.e. when the matrix is finalized.

  Warning

  This method breaks the const-ness when the matrix is finalized because it gives write access to the J and A arrays.

Implements mfem::AbstractSparseMatrix.

Definition at line 2725 of file sparsemat.cpp.

int * mfem::SparseMatrix::GetRowColumns

(

const int

row

)

Return a pointer to the column indices in a row.

Definition at line 358 of file sparsemat.cpp.

const int * mfem::SparseMatrix::GetRowColumns

(

const int

row

)

const

Return a pointer to the column indices in a row, const version.

Definition at line 365 of file sparsemat.cpp.

double * mfem::SparseMatrix::GetRowEntries

(

const int

row

)

Return a pointer to the entries in a row.

Definition at line 372 of file sparsemat.cpp.

const double * mfem::SparseMatrix::GetRowEntries

(

const int

row

)

const

Return a pointer to the entries in a row, const version.

Definition at line 379 of file sparsemat.cpp.

double mfem::SparseMatrix::GetRowNorml1

(

int

irow

)

const

For i = irow compute \( x_i = \sum_j | A_{i, j} | \).

Definition at line 1092 of file sparsemat.cpp.

void mfem::SparseMatrix::GetRowSums

(

Vector &

x

)

const

For all i compute \( x_i = \sum_j A_{ij} \).

Definition at line 1069 of file sparsemat.cpp.

void mfem::SparseMatrix::GetSubMatrix	(	const Array< int > &	rows,
		const Array< int > &	cols,
		DenseMatrix &	subm
	)		const

Definition at line 2676 of file sparsemat.cpp.

Type mfem::SparseMatrix::GetType ( ) const

inline

Definition at line 652 of file sparsemat.hpp.

const double* mfem::SparseMatrix::HostReadData ( ) const

inline

Definition at line 235 of file sparsemat.hpp.

const int* mfem::SparseMatrix::HostReadI ( ) const

inline

Definition at line 203 of file sparsemat.hpp.

const int* mfem::SparseMatrix::HostReadJ ( ) const

inline

Definition at line 219 of file sparsemat.hpp.

double* mfem::SparseMatrix::HostReadWriteData ( )

inline

Definition at line 239 of file sparsemat.hpp.

int* mfem::SparseMatrix::HostReadWriteI ( )

inline

Definition at line 207 of file sparsemat.hpp.

int* mfem::SparseMatrix::HostReadWriteJ ( )

inline

Definition at line 223 of file sparsemat.hpp.

double* mfem::SparseMatrix::HostWriteData ( )

inline

Definition at line 237 of file sparsemat.hpp.

int* mfem::SparseMatrix::HostWriteI ( )

inline

Definition at line 205 of file sparsemat.hpp.

int* mfem::SparseMatrix::HostWriteJ ( )

inline

Definition at line 221 of file sparsemat.hpp.

void mfem::SparseMatrix::InitCuSparse ( )

protected

Definition at line 38 of file sparsemat.cpp.

double mfem::SparseMatrix::InnerProduct	(	const Vector &	x,
		const Vector &	y
	)		const

Compute y^t A x.

Definition at line 1028 of file sparsemat.cpp.

MatrixInverse * mfem::SparseMatrix::Inverse ( ) const

virtual

This virtual method is not supported: it always returns NULL.

Implements mfem::Matrix.

Definition at line 1542 of file sparsemat.cpp.

double mfem::SparseMatrix::IsSymmetric

(

)

const

Returns max_{i,j} |(i,j)-(j,i)| for a finalized matrix.

Definition at line 1382 of file sparsemat.cpp.

void mfem::SparseMatrix::Jacobi	(	const Vector &	b,
		const Vector &	x0,
		Vector &	x1,
		double	sc
	)		const

One scaled Jacobi iteration for the system A x = b. x1 = x0 + sc D^{-1} (b - A x0) where D is the diag of A.

Definition at line 2396 of file sparsemat.cpp.

void mfem::SparseMatrix::Jacobi2	(	const Vector &	b,
		const Vector &	x0,
		Vector &	x1,
		double	sc = 1.0
	)		const

x1 = x0 + sc D^{-1} (b - A x0) where \( D_{ii} = \sum_j |A_{ij}| \).

Definition at line 2464 of file sparsemat.cpp.

void mfem::SparseMatrix::Jacobi3	(	const Vector &	b,
		const Vector &	x0,
		Vector &	x1,
		double	sc = 1.0
	)		const

x1 = x0 + sc D^{-1} (b - A x0) where \( D_{ii} = \sum_j A_{ij} \).

Definition at line 2488 of file sparsemat.cpp.

void mfem::SparseMatrix::LoseData ( )

inline

Lose the ownership of the graph (I, J) and data (A) arrays.

Definition at line 622 of file sparsemat.hpp.

void mfem::SparseMatrix::MakeRef

(

const SparseMatrix &

master

)

Clear the contents of the SparseMatrix and make it a reference to master.

After this call, the matrix will point to the same data as master but it will not own its data. The master must be finalized.

Definition at line 280 of file sparsemat.cpp.

double mfem::SparseMatrix::MaxNorm

(

)

const

Definition at line 1465 of file sparsemat.cpp.

int mfem::SparseMatrix::MaxRowSize

(

)

const

Returns the maximum number of elements among all rows.

Definition at line 334 of file sparsemat.cpp.

void mfem::SparseMatrix::MoveDiagonalFirst

(

)

Move the diagonal entry to the first position in each row, preserving the order of the rest of the columns.

Definition at line 457 of file sparsemat.cpp.

void mfem::SparseMatrix::Mult ( const Vector &  x, Vector &  y  ) const

virtual

Matrix vector multiplication.

Implements mfem::AbstractSparseMatrix.

Definition at line 613 of file sparsemat.cpp.

void mfem::SparseMatrix::MultTranspose ( const Vector &  x, Vector &  y  ) const

virtual

Multiply a vector with the transposed matrix. y = At * x.

Implements mfem::AbstractSparseMatrix.

Definition at line 770 of file sparsemat.cpp.

int mfem::SparseMatrix::NumNonZeroElems ( ) const

virtual

Returns the number of the nonzero elements in the matrix.

Implements mfem::AbstractSparseMatrix.

Definition at line 1443 of file sparsemat.cpp.

double & mfem::SparseMatrix::operator()	(	int	i,
		int	j
	)

Returns reference to A[i][j].

Definition at line 491 of file sparsemat.cpp.

const double & mfem::SparseMatrix::operator()	(	int	i,
		int	j
	)		const

Returns reference to A[i][j].

Definition at line 514 of file sparsemat.cpp.

SparseMatrix & mfem::SparseMatrix::operator*=

(

double

a

)

Definition at line 3010 of file sparsemat.cpp.

SparseMatrix & mfem::SparseMatrix::operator+=

(

const SparseMatrix &

B

)

Add the sparse matrix 'B' to '*this'. This operation will cause an error if '*this' is finalized and 'B' has larger sparsity pattern.

Definition at line 2931 of file sparsemat.cpp.

SparseMatrix & mfem::SparseMatrix::operator=

(

const SparseMatrix &

rhs

)

Assignment operator: deep copy.

Definition at line 270 of file sparsemat.cpp.

SparseMatrix & mfem::SparseMatrix::operator=

(

double

a

)

Definition at line 2984 of file sparsemat.cpp.

bool mfem::SparseMatrix::OwnsData ( ) const

inline

Get the data ownership flag (A array).

Definition at line 619 of file sparsemat.hpp.

bool mfem::SparseMatrix::OwnsGraph ( ) const

inline

Get the graph ownership flag (I and J arrays).

Definition at line 616 of file sparsemat.hpp.

void mfem::SparseMatrix::PartAddMult	(	const Array< int > &	rows,
		const Vector &	x,
		Vector &	y,
		const double	a = 1.0
	)		const

Definition at line 862 of file sparsemat.cpp.

void mfem::SparseMatrix::PartMult	(	const Array< int > &	rows,
		const Vector &	x,
		Vector &	y
	)		const

Definition at line 836 of file sparsemat.cpp.

void mfem::SparseMatrix::Print ( std::ostream &  out = mfem::out, int  width_ = 4  ) const

virtual

Prints matrix to stream out.

Reimplemented from mfem::Matrix.

Definition at line 3034 of file sparsemat.cpp.

void mfem::SparseMatrix::PrintCSR

(

std::ostream &

out

)

const

Prints matrix to stream out in hypre_CSRMatrix format.

Definition at line 3124 of file sparsemat.cpp.

void mfem::SparseMatrix::PrintCSR2

(

std::ostream &

out

)

const

Prints a sparse matrix to stream out in CSR format.

Definition at line 3148 of file sparsemat.cpp.

void mfem::SparseMatrix::PrintInfo

(

std::ostream &

out

)

const

Print various sparse matrix statistics.

Definition at line 3173 of file sparsemat.cpp.

void mfem::SparseMatrix::PrintMatlab

(

std::ostream &

out = mfem::out

)

const

Prints matrix in matlab format.

Definition at line 3082 of file sparsemat.cpp.

void mfem::SparseMatrix::PrintMM

(

std::ostream &

out = mfem::out

)

const

Prints matrix in Matrix Market sparse format.

Definition at line 3102 of file sparsemat.cpp.

const double* mfem::SparseMatrix::ReadData ( bool  on_dev = true ) const

inline

Definition at line 229 of file sparsemat.hpp.

const int* mfem::SparseMatrix::ReadI ( bool  on_dev = true ) const

inline

Definition at line 197 of file sparsemat.hpp.

const int* mfem::SparseMatrix::ReadJ ( bool  on_dev = true ) const

inline

Definition at line 213 of file sparsemat.hpp.

double* mfem::SparseMatrix::ReadWriteData ( bool  on_dev = true )

inline

Definition at line 233 of file sparsemat.hpp.

int* mfem::SparseMatrix::ReadWriteI ( bool  on_dev = true )

inline

Definition at line 201 of file sparsemat.hpp.

int* mfem::SparseMatrix::ReadWriteJ ( bool  on_dev = true )

inline

Definition at line 217 of file sparsemat.hpp.

void mfem::SparseMatrix::ResetTranspose

(

)

const

Reset (destroy) the internal transpose matrix. See BuildTranspose() for more details.

Definition at line 830 of file sparsemat.cpp.

bool mfem::SparseMatrix::RowIsEmpty

(

const int

row

)

const

Definition at line 2704 of file sparsemat.cpp.

int mfem::SparseMatrix::RowSize

(

const int

i

)

const

Returns the number of elements in row i.

Definition at line 311 of file sparsemat.cpp.

void mfem::SparseMatrix::ScaleColumns

(

const Vector &

sr

)

this = this * diag(sr);

Definition at line 2903 of file sparsemat.cpp.

void mfem::SparseMatrix::ScaleRow	(	const int	row,
		const double	scale
	)

Definition at line 2844 of file sparsemat.cpp.

void mfem::SparseMatrix::ScaleRows

(

const Vector &

sl

)

this = diag(sl) * this;

Definition at line 2872 of file sparsemat.cpp.

double & mfem::SparseMatrix::SearchRow ( const int  col )

inline

Perform a fast search for an entry in the "current row". See SetColPtr().

If the matrix is not finalized and the entry is not found in the SparseMatrix, it will be added to the sparsity pattern initialized with zero. If the matrix is finalized and the entry is not found, an error will be generated.

Definition at line 790 of file sparsemat.hpp.

double & mfem::SparseMatrix::SearchRow ( const int  row, const int  col  )

inline

Definition at line 831 of file sparsemat.hpp.

void mfem::SparseMatrix::Set	(	const int	i,
		const int	j,
		const double	a
	)

Definition at line 2558 of file sparsemat.cpp.

void mfem::SparseMatrix::SetColPtr ( const int  row ) const

inline

Initialize the SparseMatrix for fast access to the entries of the given row which becomes the "current row".

Fast access to the entries of the "current row" can be performed using the methods: SearchRow(const int), Add(const int, const double), Set(const int, const double), and Get(const int).

Definition at line 736 of file sparsemat.hpp.

void mfem::SparseMatrix::SetDataOwner ( bool  owna )

inline

Set the data ownership flag (A array).

Definition at line 613 of file sparsemat.hpp.

void mfem::SparseMatrix::SetDiagIdentity

(

)

If a row contains only one diag entry of zero, set it to 1.

Definition at line 2166 of file sparsemat.cpp.

void mfem::SparseMatrix::SetEmpty ( )

protected

Definition at line 292 of file sparsemat.cpp.

void mfem::SparseMatrix::SetGraphOwner ( bool  ownij )

inline

Set the graph ownership flag (I and J arrays).

Definition at line 609 of file sparsemat.hpp.

void mfem::SparseMatrix::SetRow	(	const int	row,
		const Array< int > &	cols,
		const Vector &	srow
	)

Definition at line 2765 of file sparsemat.cpp.

void mfem::SparseMatrix::SetSubMatrix	(	const Array< int > &	rows,
		const Array< int > &	cols,
		const DenseMatrix &	subm,
		int	skip_zeros = 1
	)

Definition at line 2596 of file sparsemat.cpp.

void mfem::SparseMatrix::SetSubMatrixTranspose	(	const Array< int > &	rows,
		const Array< int > &	cols,
		const DenseMatrix &	subm,
		int	skip_zeros = 1
	)

Definition at line 2635 of file sparsemat.cpp.

void mfem::SparseMatrix::SetWidth

(

int

width_ = -1

)

Change the width of a SparseMatrix.

If width_ = -1 (DEFAULT), this routine will set the new width to the actual Width of the matrix awidth = max(J) + 1. Values 0 <= width_ < awidth are not allowed (error check in Debug Mode only)

This method can be called for matrices finalized or not.

Definition at line 386 of file sparsemat.cpp.

int mfem::SparseMatrix::Size ( ) const

inline

For backward compatibility, define Size() to be synonym of Height().

Definition at line 167 of file sparsemat.hpp.

void mfem::SparseMatrix::SortColumnIndices

(

)

Sort the column indices corresponding to each row.

Definition at line 424 of file sparsemat.cpp.

void mfem::SparseMatrix::Swap

(

SparseMatrix &

other

)

Definition at line 3970 of file sparsemat.cpp.

void mfem::SparseMatrix::Symmetrize

(

)

(*this) = 1/2 ((*this) + (*this)^t)

Definition at line 1424 of file sparsemat.cpp.

void mfem::SparseMatrix::Threshold	(	double	tol,
		bool	fix_empty_rows = false
	)

Remove entries smaller in absolute value than a given tolerance tol. If fix_empty_rows is true, a zero value is inserted in the diagonal entry (for square matrices only)

Definition at line 1116 of file sparsemat.cpp.

DenseMatrix * mfem::SparseMatrix::ToDenseMatrix

(

)

const

Produces a DenseMatrix from a SparseMatrix.

Definition at line 583 of file sparsemat.cpp.

void mfem::SparseMatrix::ToDenseMatrix

(

DenseMatrix &

B

)

const

Produces a DenseMatrix from a SparseMatrix.

Definition at line 596 of file sparsemat.cpp.

void mfem::SparseMatrix::UseCuSparse ( bool  useCuSparse_ = true )

inline

Definition at line 155 of file sparsemat.hpp.

double* mfem::SparseMatrix::WriteData ( bool  on_dev = true )

inline

Definition at line 231 of file sparsemat.hpp.

int* mfem::SparseMatrix::WriteI ( bool  on_dev = true )

inline

Definition at line 199 of file sparsemat.hpp.

int* mfem::SparseMatrix::WriteJ ( bool  on_dev = true )

inline

Definition at line 215 of file sparsemat.hpp.

Member Data Documentation

Memory<double> mfem::SparseMatrix::A

protected

Array with size I[height], containing the actual entries of the sparse matrix, as indexed by the I array.

Definition at line 59 of file sparsemat.hpp.

SparseMatrix* mfem::SparseMatrix::At

mutableprotected

Transpose of A. Owned. Used to perform MultTranspose() on devices.

Definition at line 71 of file sparsemat.hpp.

size_t mfem::SparseMatrix::bufferSize = 0

staticprotected

Definition at line 94 of file sparsemat.hpp.

int* mfem::SparseMatrix::ColPtrJ

mutableprotected

Definition at line 67 of file sparsemat.hpp.

RowNode** mfem::SparseMatrix::ColPtrNode

mutableprotected

Definition at line 68 of file sparsemat.hpp.

int mfem::SparseMatrix::current_row

mutableprotected

Definition at line 66 of file sparsemat.hpp.

void * mfem::SparseMatrix::dBuffer = nullptr

staticprotected

Definition at line 95 of file sparsemat.hpp.

cusparseMatDescr_t mfem::SparseMatrix::descr =0

protected

Definition at line 92 of file sparsemat.hpp.

cusparseHandle_t mfem::SparseMatrix::handle = nullptr

staticprotected

Definition at line 91 of file sparsemat.hpp.

Memory<int> mfem::SparseMatrix::I

protected

Array with size (height+1) containing the row offsets.

The data for row r, 0 <= r < height, is at offsets j, I[r] <= j < I[r+1]. The offsets, j, are indices in the J and A arrays. The first entry in this array is always zero, I[0] = 0, and the last entry, I[height], gives the total number of entries stored (at a minimum, all nonzeros must be represented) in the sparse matrix.

Definition at line 53 of file sparsemat.hpp.

bool mfem::SparseMatrix::initBuffers {false}

mutableprotected

Definition at line 96 of file sparsemat.hpp.

bool mfem::SparseMatrix::isSorted

protected

Are the columns sorted already.

Definition at line 79 of file sparsemat.hpp.

Memory<int> mfem::SparseMatrix::J

protected

Array with size I[height], containing the column indices for all matrix entries, as indexed by the I array.

Definition at line 56 of file sparsemat.hpp.

cusparseSpMatDescr_t mfem::SparseMatrix::matA_descr

mutableprotected

Definition at line 98 of file sparsemat.hpp.

cusparseMatDescr_t mfem::SparseMatrix::matA_descr

mutableprotected

Definition at line 102 of file sparsemat.hpp.

RowNodeAlloc* mfem::SparseMatrix::NodesMem

protected

Definition at line 75 of file sparsemat.hpp.

RowNode** mfem::SparseMatrix::Rows

protected

Array of linked lists, one for every row. This array represents the linked list (LIL) storage format.

Definition at line 64 of file sparsemat.hpp.

int mfem::SparseMatrix::SparseMatrixCount = 0

staticprotected

Definition at line 93 of file sparsemat.hpp.

cusparseStatus_t mfem::SparseMatrix::status

protected

Definition at line 90 of file sparsemat.hpp.

bool mfem::SparseMatrix::useCuSparse {true}

protected

Definition at line 84 of file sparsemat.hpp.

cusparseDnVecDescr_t mfem::SparseMatrix::vecX_descr

mutableprotected

Definition at line 99 of file sparsemat.hpp.

cusparseDnVecDescr_t mfem::SparseMatrix::vecY_descr

mutableprotected

Definition at line 100 of file sparsemat.hpp.

---

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

• linalg/sparsemat.hpp
• linalg/sparsemat.cpp