lama::DenseMatrix< T > Class Template Reference

Class for dense matrices where rows are distributed among rows and columns are splitted according to a column distribution. More...

#include <DenseMatrix.hpp>

Inheritance diagram for lama::DenseMatrix< T >:

Public Types
typedef T	ValueType
typedef boost::shared_ptr < DenseStorage< T > >	DenseStoragePtr
enum	SyncKind { ASYNCHRONOUS, SYNCHRONOUS }
	SyncKind describes if the communication and computation should be done synchronously or asynchronously. More...
enum	MatrixKind { DENSE, SPARSE }
	MatrixKind describes if a matrix is dense or sparse. More...
typedef const Matrix &	ExpressionMemberType
	ExpressionMemberType is the type that is used the template Expression to store a Vector.
Public Member Functions
	DenseMatrix ()
	Default constructor.
	DenseMatrix (const IndexType numRows, const IndexType numColumns)
	Constructor of a replicated dense matrix.
	DenseMatrix (DistributionPtr rowDist, DistributionPtr colDist)
	Constructor of a distributed dense matrix.
	DenseMatrix (DistributionPtr dist)
	Construct a square unity matrix.
	DenseMatrix (const DenseMatrix< ValueType > &other)
	Overwrite default copy constructor so it uses other copy constructor.
	DenseMatrix (const Matrix &other)
	Construct a dense matrix from any other matrix that can be of a different type.
	DenseMatrix (const Matrix &other, DistributionPtr distribution, DistributionPtr colDistribution)
	Construct a dense matrix from any other matrix with new distributions.
	DenseMatrix (const DenseMatrix< ValueType > &matrix, DistributionPtr distribution, DistributionPtr colDistribution)
	DenseMatrix (const _MatrixStorage &localData, DistributionPtr rowDist, DistributionPtr colDist)
	Constructor of a dense matrix by local storage.
template<typename OtherValueType >
	DenseMatrix (const IndexType numRows, const IndexType numColumns, const IndexType numNoneZeros, const IndexType *const ia, const IndexType *const ja, const OtherValueType *const values)
	Creates a replicated dense matrix from the passed csr sparse matrix.
	DenseMatrix (const Expression< Expression< Scalar, Expression< Matrix, Matrix, Times >, Times >, Expression< Scalar, Matrix, Times >, Plus > expression)
	DenseMatrix (const Expression< Matrix, Matrix, Times > expression)
	DenseMatrix (const Expression< Scalar, Expression< Matrix, Matrix, Times >, Times > expression)
	DenseMatrix (const Expression< Scalar, Matrix, Times > expression)
	DenseMatrix (const std::string &filename)
	Constructor of a replicated dense matrix by reading the matrix data from a file.
virtual	~DenseMatrix ()
DenseMatrix &	operator= (const DenseMatrix &matrix)
	Override default assignment operator needed to guarantee deep copy.
DenseMatrix &	operator= (const Matrix &matrix)
	the assignment operator for matrix.
DenseMatrix &	operator= (const Expression< Scalar, Matrix, Times > expression)
	the assignment operator for a scalar matrix multiplication.
DenseMatrix &	operator= (const Expression< Matrix, Matrix, Times > expression)
	the assignment operator for a matrix matrix multiplication.
DenseMatrix &	operator= (const Expression< Scalar, Expression< Matrix, Matrix, Times >, Times > expression)
	the assignment operator for a scalar matrix matrix multiplication.
DenseMatrix &	operator= (const Expression< Expression< Scalar, Expression< Matrix, Matrix, Times >, Times >, Expression< Scalar, Matrix, Times >, Plus > expression)
	the assignment operator for a GEMM expression.
const char *	getTypeName () const
	Implementation for Matrix::getTypeName()
Matrix::MatrixKind	getMatrixKind () const
	Each derived matrix must give info about its kind (DENSE or SPARSE).
virtual void	setContext (const ContextPtr context)
	specifies on which compute back end the matrix operations should take place.
virtual ContextPtr	getContextPtr () const
	Getter routine for the context.
virtual void	setIdentity ()
	Operator that sets the matrix to the identity matrix.
template<typename OtherValueType >
void	setRawDenseData (const IndexType numRows, const IndexType numColumns, const OtherValueType values[], const OtherValueType eps=0.0)
	Set matrix with global dense data.
virtual void	buildCSRData (LAMAArray< IndexType > &rowIA, LAMAArray< IndexType > &rowJA, _LAMAArray &rowValues) const
	Implementation of pure method for the dense storage format.
virtual void	setCSRData (const LAMAArray< IndexType > &rowIA, const LAMAArray< IndexType > &rowJA, const _LAMAArray &rowValues, DistributionPtr rowDistribution, DistributionPtr colDistribution)
	Implementation of pure method.
void	setCSRDataLocal (const LAMAArray< IndexType > &rowIA, const LAMAArray< IndexType > &rowJA, const _LAMAArray &rowValues) const
	Local version of setCSRData .
virtual void	clear ()
	Implementation of pure method.
virtual void	allocate (const IndexType numRows, const IndexType numColumns)
	Reallocate this matrix to a replicated zero-matrix of the given shape.
virtual void	allocate (DistributionPtr rowDistribution, DistributionPtr colDistribution)
	Reallocate this matrix to a distributed zero-matrix by the given distributions.
virtual void	assign (const Matrix &other)
	This method assigns a matrix to this matrix with automatic conversion to the needed matrix type and the current row / column distribution of.
void	swap (DenseMatrix< ValueType > &other)
	Swap will swap all member variables of the two dense matrices.
void	assignSparse (const _SparseMatrix &other)
void	assign (const _MatrixStorage &storage)
	Assign replicated storage for a replicated matrix.
void	assign (const _MatrixStorage &storage, DistributionPtr rowDist, DistributionPtr colDist)
	Assign local storage for a distributed matrix.
void	assignLocal (const _MatrixStorage &other)
virtual void	buildLocalStorage (_MatrixStorage &storage) const
	Implementation of Matrix::buildLocalStorage.
void	redistribute (DistributionPtr rowDistribution, DistributionPtr colDistribution)
	Implementation of pure method.
virtual void	getDiagonal (Vector &diagonal) const
	This method returns the diagonal.
virtual void	getRow (Vector &row, const IndexType globalRowIndex) const
	This method returns one row of the matrix.
virtual void	setDiagonal (const Vector &diagonal)
	This method replaces the diagonal.
virtual void	setDiagonal (const Scalar diagonalValue)
	This method replaces the diagonal by a diagonal value.
virtual void	scale (const Vector &values)
	This method scales all values with a vector.
virtual void	scale (const Scalar value)
	This method scales all matrix values with a scalar.
Scalar	getValue (lama::IndexType i, lama::IndexType j) const
	returns a copy of the value at the passed global indexes.
void	matrixTimesScalar (const Matrix &other, const Scalar alpha)
	matrixTimesScalar computes this = alpha * other
void	matrixTimesVectorImpl (DenseVector< ValueType > &denseResult, const ValueType alphaValue, const DenseVector< ValueType > &denseX, const ValueType betaValue, const DenseVector< ValueType > &denseY) const
	Matrix times vector with same values types and correct distributions.
virtual void	invert (const Matrix &other)
	Set this matrix with the inverse of another square matrix.
void	invert ()
	Invert in place.
Scalar	maxDiffNorm (const Matrix &other) const
	Implementation of Matrix::maxDiffNorm for dense matrices.
ValueType	maxDiffNormImpl (const DenseMatrix< ValueType > &other) const
	Get the maximal difference between two elements for dense matrices of same type.
void	matrixTimesMatrix (Matrix &result, const Scalar alpha, const Matrix &x, const Scalar beta, const Matrix &y) const
	Implementation of pure method Matrix::matrixTimesMatrix.
virtual void	prefetch () const
	Prefetch matrix data to its 'preferred' context location.
void	prefetch (lama::ContextPtr loc) const
void	wait () const
	wait for a possibly running prefetch.
const DenseStorage< ValueType > &	getLocalStorage () const
	This method returns the storage containing the local data regarding row/col distribution.
DenseStorage< ValueType > &	getLocalStorage ()
virtual IndexType	getLocalNumValues () const
	Implementation of Matrix::getLocalNumValues, getLocalNumRows, getLocalNumColumns.
virtual IndexType	getLocalNumRows () const
	Getter routine for the local number of rows.
virtual IndexType	getLocalNumColumns () const
	Getter routine for the local number of rows and columns.
virtual IndexType	getNumValues () const
	Get the total number of non-zero values in the matrix.
std::vector< DenseStoragePtr > &	getCyclicLocalValues ()
const std::vector < DenseStoragePtr > &	getCyclicLocalValues () const
virtual bool	hasDiagonalProperty () const
	hasDiagonalProperty returns the diagonalProperty of the local storage.
virtual void	resetDiagonalProperty ()
	resetDiagonalProperty rechecks the storages for their diagonal property
virtual void	writeAt (std::ostream &stream) const
	Write some information about this to the passed stream.
virtual Scalar::ScalarType	getValueType () const
	Query the value type of the matrix elements, e.g.
void	writeToFile (const std::string &fileName, const File::FileType fileType=File::BINARY, const File::DataType dataType=File::INTERNAL, const File::IndexDataType indexDataTypeIA=File::INT, const File::IndexDataType indexDataTypeJA=File::INT) const
	Method writes dense matrix to a file.
virtual std::auto_ptr< Matrix >	create () const
	Constructing a new object by the default constructor.
virtual std::auto_ptr< Matrix >	copy () const
	Constructing a new object by the copy constructor.
virtual size_t	getMemoryUsage () const
	getMemoryUsage returns the global memory that is allocated to hold this matrix.
const std::vector< PartitionId > &	getOwners () const
template<typename OtherValueType >
void	getDiagonalImpl (DenseVector< OtherValueType > &diagonal) const
template<>
const char *	typeName ()
template<>
const char *	typeName ()
template<typename OtherValueType >
void	copyDenseMatrix (const DenseMatrix< OtherValueType > &other)
void	matrixTimesVector (Vector &result, const Scalar alpha, const Vector &x, const Scalar beta, const Vector &y) const
	matrixTimesVector computes result = alpha * this * x + beta * y
std::auto_ptr< _LAMAArray >	createArray () const
	This routine creates a LAMA array with the same value type as the matrix.
Scalar	operator() (IndexType i, IndexType j) const
	returns a copy of the value at the passed global indexes.
IndexType	getNumRows () const
	Returns the number of global rows.
IndexType	getNumColumns () const
	Returns the number of columns.
double	getSparsityRate () const
virtual void	matrix2CSRGraph (IndexType *xadj, IndexType *adjncy, IndexType *vwgt, CommunicatorPtr comm, const IndexType *globalRowIndices=NULL, IndexType *vtxdist=NULL) const
	transformation from matrix type to a csr graph
const Distribution &	getColDistribution () const
	gets a constant reference to the column distribution.
DistributionPtr	getColDistributionPtr () const
	gets a pointer to the column distribution.
virtual void	setContext (const ContextPtr localContext, const ContextPtr haloContext)
	Only sparse matrices will override this method, others will ignore second argument.
const Context &	getContext () const
SyncKind	getCommunicationKind () const
	get the communication kind.
SyncKind	getComputeKind () const
	get the compute kind.
void	setCommunicationKind (SyncKind communicationKind)
	set the communication kind.
virtual void	setComputeKind (SyncKind computeKind)
	set the compute kind.
void	inheritAttributes (const Matrix &other)
	Inherit context and kind arguments from another matrix.
std::auto_ptr< Matrix >	create (const IndexType numRows, const IndexType numColumns) const
	Constructor creates a replicated matrix of same type as a given matrix.
std::auto_ptr< Matrix >	create (const IndexType size) const
	Constructor creates a distributed zero matrix of same type as a given matrix.
std::auto_ptr< Matrix >	create (DistributionPtr rowDistribution, DistributionPtr colDistribution) const
	Constructor creates a distributed zero matrix of same type as a given matrix.
std::auto_ptr< Matrix >	create (DistributionPtr distribution) const
	Constructor creates a distributed zero matrix of same type as a given matrix.
VectorPtr	createDenseVector (DistributionPtr distribution, const Scalar value) const
	Constructor creates a distributed dense vector of same type as a given matrix.
const Distribution &	getDistribution () const
DistributionPtr	getDistributionPtr () const
Static Public Member Functions
static const char *	typeName ()
	Getter for the type name of the class.
Data Fields
std::vector< boost::shared_ptr < DenseStorage< ValueType > > >	mData
	local data is allocated in chunks according to column distribution
Protected Member Functions
void	setReplicatedMatrix (const IndexType numRows, const IndexType numColumns)
	Set the global/local size of replicated matrix.
void	setDistributedMatrix (DistributionPtr distribution, DistributionPtr colDistribution)
	Set the global and local size of distributed matrix.
void	checkSettings ()
void	swapMatrix (Matrix &other)
void	swap (Distributed &other)
void	setDistributionPtr (DistributionPtr distributionPtr)
Protected Attributes
std::vector< PartitionId >	mOwners
DistributionPtr	mColDistribution
IndexType	mNumRows
IndexType	mNumColumns
Private Member Functions
void	allocateData ()
	Allocate of storage for the column blocks.
template<typename otherT >
void	copyDenseMatrix (const DenseMatrix< otherT > &other)
	Copy a dense matrix with different data type; inherits sizes and distributions.
template<typename OtherT >
void	getDiagonalImpl (DenseVector< OtherT > &diagonal) const
	Optimized implementation for dense vectors as diagonal.
void	redistributeRows (DistributionPtr rowDistribution)
void	splitColumns (DistributionPtr colDistribution)
	Split the replicated columns into chunks according to the column distribution.
void	getRow (DenseVector< ValueType > &row, const IndexType i) const
	LAMA_LOG_DECL_STATIC_LOGGER (logger)
void	computeOwners ()
void	invertCyclic ()
	Special implementation of invert in place for a cyclic distributed matrix.
Static Private Member Functions
static void	joinColumnData (DenseStorage< ValueType > &result, const std::vector< boost::shared_ptr< DenseStorage< ValueType > > > &chunks, const std::vector< IndexType > &columnOwners)
	Join dense storage of column distributed data.
static void	splitColumnData (std::vector< boost::shared_ptr< DenseStorage< ValueType > > > &chunks, const DenseStorage< ValueType > &columnData, const PartitionId numChunks, const std::vector< IndexType > &columnOwners)
	Split dense storage data according to a distribution into chunks.
static void	localize (DenseStorage< ValueType > &local, const DenseStorage< ValueType > &global, const Distribution &rowDistribution)
	Restrict dense storage of a replicated matrix to its local part according to row distribution.
Private Attributes
LAMAArray< ValueType >	mSendValues
LAMAArray< ValueType >	mReceiveValues

---

Detailed Description

template<typename T>
class lama::DenseMatrix< T >

Class for dense matrices where rows are distributed among rows and columns are splitted according to a column distribution.

The local rows are splitted according to a column distribution into different blocks that are stored in a vector of shared pointers. A copy of this vector will not make deep copies of the blocks so default copy constructor and assignment operator must be overridden.

---

Member Typedef Documentation

template<typename T>

typedef boost::shared_ptr<DenseStorage<T> > lama::DenseMatrix< T >::DenseStoragePtr

typedef const Matrix& lama::Matrix::ExpressionMemberType [inherited]

ExpressionMemberType is the type that is used the template Expression to store a Vector.

template<typename T>

typedef T lama::DenseMatrix< T >::ValueType

---

Member Enumeration Documentation

enum lama::Matrix::MatrixKind [inherited]

MatrixKind describes if a matrix is dense or sparse.

Enumerator:

DENSE	matrix kind for a dense matrix
SPARSE	matrix kind for a sparse matrix

enum lama::Matrix::SyncKind [inherited]

SyncKind describes if the communication and computation should be done synchronously or asynchronously.

Enumerator:

ASYNCHRONOUS
SYNCHRONOUS

---

Constructor & Destructor Documentation

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix

(

)

Default constructor.

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix	(	const IndexType	numRows,
		const IndexType	numColumns
	)

Constructor of a replicated dense matrix.

References lama::DenseMatrix< T >::computeOwners(), and lama::DenseMatrix< T >::mData.

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix	(	DistributionPtr	rowDist,
		DistributionPtr	colDist
	)

Constructor of a distributed dense matrix.

References lama::DenseMatrix< T >::allocateData(), and lama::DenseMatrix< T >::computeOwners().

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix

(

DistributionPtr

dist

)

[explicit]

Construct a square unity matrix.

References lama::DenseMatrix< T >::computeOwners(), lama::Distributed::getDistribution(), lama::Distribution::getLocalSize(), lama::Distribution::getNumPartitions(), lama::DenseMatrix< T >::mData, and lama::DenseMatrix< T >::mOwners.

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix

(

const DenseMatrix< ValueType > &

other

)

Overwrite default copy constructor so it uses other copy constructor.

Note: Default copy constructor would not make deep copies of the storages, so it must be overridden

References lama::DenseMatrix< T >::assign().

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix

(

const Matrix &

other

)

Construct a dense matrix from any other matrix that can be of a different type.

New dense matrix has the same size and the same distribution.

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix	(	const Matrix &	other,
		DistributionPtr	distribution,
		DistributionPtr	colDistribution
	)

Construct a dense matrix from any other matrix with new distributions.

The following codes are equivalent:

      DenseMatrix dense( other, rowDist, colDist );
      // same as
      DenseMatrix dense( other );
      dense->redistribute( rowDist, colDist );

The constructor with distributions is more convenient and might be more efficient due to less memory allocations as less temporary data is needed.

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix	(	const DenseMatrix< ValueType > &	matrix,
		DistributionPtr	distribution,
		DistributionPtr	colDistribution
	)

References lama::DenseMatrix< T >::computeOwners(), lama::Matrix::getColDistribution(), lama::Distribution::getCommunicator(), lama::Distributed::getDistribution(), lama::Distribution::getLocalSize(), lama::Matrix::getNumColumns(), lama::Distribution::getNumPartitions(), LAMA_THROWEXCEPTION, lama::DenseMatrix< T >::localize(), lama::DenseMatrix< T >::mData, lama::DenseMatrix< T >::mOwners, and lama::DenseMatrix< T >::splitColumnData().

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix	(	const _MatrixStorage &	localData,
		DistributionPtr	rowDist,
		DistributionPtr	colDist
	)

Constructor of a dense matrix by local storage.

Parameters:

[in]	localData	contains local rows of the distributed matrix
[in]	rowDist	is distribution of localData
[in]	colDist	specifies how to split columns of local rows

This constructor works also fine if localData is the full global matrix; in this case only local rows will be taken on this processor.

template<typename ValueType >

template<typename OtherValueType >

lama::DenseMatrix< ValueType >::DenseMatrix	(	const IndexType	numRows,
		const IndexType	numColumns,
		const IndexType	numNoneZeros,
		const IndexType *const	ia,
		const IndexType *const	ja,
		const OtherValueType *const	values
	)

Creates a replicated dense matrix from the passed csr sparse matrix.

Parameters:

[in]	numRows	the number of rows of the matrix
[in]	numColumns	the number of columns of the matrix
[in]	numNoneZeros	the number of none zeros of the matrix
[in]	ia	row pointer of the input csr sparse matrix
[in]	ja	column indexes of the input csr sparse matrix
[in]	values	the none zero values of the input csr sparse matrix

References lama::DenseMatrix< T >::computeOwners(), lama::DenseMatrix< T >::mData, lama::Matrix::mNumColumns, and lama::Matrix::mNumRows.

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix

(

const Expression< Expression< Scalar, Expression< Matrix, Matrix, Times >, Times >, Expression< Scalar, Matrix, Times >, Plus >

expression

)

References lama::Matrix::operator=().

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix

(

const Expression< Matrix, Matrix, Times >

expression

)

References lama::Matrix::operator=().

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix

(

const Expression< Scalar, Expression< Matrix, Matrix, Times >, Times >

expression

)

References lama::Matrix::operator=().

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix

(

const Expression< Scalar, Matrix, Times >

expression

)

References lama::Matrix::operator=().

template<typename ValueType >

lama::DenseMatrix< ValueType >::DenseMatrix

(

const std::string &

filename

)

Constructor of a replicated dense matrix by reading the matrix data from a file.

Next releases will also support distributed/parallel I/O. In the meantime this constructor should be used with a following call of the redistribute method.

References lama::Matrix::setReplicatedMatrix().

template<typename ValueType >

lama::DenseMatrix< ValueType >::~DenseMatrix

(

)

[virtual]

---

Member Function Documentation

template<typename ValueType >

void lama::DenseMatrix< ValueType >::allocate	(	const IndexType	numRows,
		const IndexType	numColumns
	)		[virtual]

Reallocate this matrix to a replicated zero-matrix of the given shape.

Implements lama::Matrix.

References LAMA_ASSERT_ERROR, and lama::Matrix::setReplicatedMatrix().

Referenced by lama::SparseMatrix< T >::matrixTimesVectorNImpl().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::allocate	(	DistributionPtr	rowDistribution,
		DistributionPtr	colDistribution
	)		[virtual]

Reallocate this matrix to a distributed zero-matrix by the given distributions.

Implements lama::Matrix.

References LAMA_THROWEXCEPTION, and lama::Matrix::setDistributedMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::allocateData

(

)

[private]

Allocate of storage for the column blocks.

References LAMA_ASSERT_DEBUG, and LAMA_ASSERT_EQUAL_DEBUG.

Referenced by lama::DenseMatrix< T >::DenseMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::assign

(

const Matrix &

other

)

[virtual]

This method assigns a matrix to this matrix with automatic conversion to the needed matrix type and the current row / column distribution of.

Parameters:

[in]

other

must be matrix with same global sizes

Implements lama::Matrix.

References lama::Matrix::DENSE, lama::Scalar::DOUBLE, lama::Scalar::FLOAT, lama::Matrix::getColDistributionPtr(), lama::Distributed::getDistributionPtr(), lama::Matrix::getMatrixKind(), lama::Matrix::getValueType(), LAMA_THROWEXCEPTION, and lama::Matrix::setDistributedMatrix().

Referenced by lama::DenseMatrix< T >::DenseMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::assign

(

const _MatrixStorage &

storage

)

[virtual]

Assign replicated storage for a replicated matrix.

Implements lama::Matrix.

References lama::_MatrixStorage::getNumColumns(), lama::_MatrixStorage::getNumRows(), and lama::Matrix::setReplicatedMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::assign	(	const _MatrixStorage &	storage,
		DistributionPtr	rowDist,
		DistributionPtr	colDist
	)		[virtual]

Assign local storage for a distributed matrix.

Implements lama::Matrix.

References lama::DenseStorageView< T >::assign(), lama::DENSE, lama::_MatrixStorage::getFormat(), lama::_MatrixStorage::getNumRows(), lama::_MatrixStorage::getValueType(), LAMA_ASSERT_ERROR, LAMA_THROWEXCEPTION, lama::MatrixStorage< T >::localize(), and lama::Matrix::setDistributedMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::assignLocal

(

const _MatrixStorage &

other

)

References lama::_MatrixStorage::buildCSRData().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::assignSparse

(

const _SparseMatrix &

other

)

References lama::Matrix::getColDistribution(), lama::Matrix::getColDistributionPtr(), lama::Distributed::getDistributionPtr(), lama::_SparseMatrix::getLocalStorage(), lama::Matrix::getNumColumns(), lama::Distribution::isReplicated(), LAMA_ASSERT_DEBUG, and lama::Matrix::setDistributedMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::buildCSRData	(	LAMAArray< IndexType > &	rowIA,
		LAMAArray< IndexType > &	rowJA,
		_LAMAArray &	rowValues
	)		const [virtual]

Implementation of pure method for the dense storage format.

References lama::_LAMAArray::getValueType(), and LAMA_THROWEXCEPTION.

template<typename ValueType >

void lama::DenseMatrix< ValueType >::buildLocalStorage

(

_MatrixStorage &

storage

)

const [virtual]

Implementation of Matrix::buildLocalStorage.

Implements lama::Matrix.

void lama::Matrix::checkSettings

(

)

[protected, inherited]

Reimplemented in lama::SparseMatrix< T >, and lama::SparseMatrix< ValueType >.

References lama::Matrix::getColDistribution(), lama::Distributed::getDistribution(), LAMA_THROWEXCEPTION, lama::Matrix::mColDistribution, lama::Matrix::mNumColumns, and lama::Matrix::mNumRows.

Referenced by lama::Matrix::Matrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::clear

(

)

[virtual]

Implementation of pure method.

Implements lama::Matrix.

References lama::Matrix::setReplicatedMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::computeOwners

(

)

[private]

References lama::Distribution::computeOwners().

Referenced by lama::DenseMatrix< T >::DenseMatrix().

template<typename ValueType >

std::auto_ptr< Matrix > lama::DenseMatrix< ValueType >::copy

(

)

const [virtual]

Constructing a new object by the copy constructor.

Implements lama::Matrix.

template<typename T>

template<typename otherT >

void lama::DenseMatrix< T >::copyDenseMatrix

(

const DenseMatrix< otherT > &

other

)

[private]

Copy a dense matrix with different data type; inherits sizes and distributions.

template<typename T>

template<typename OtherValueType >

void lama::DenseMatrix< T >::copyDenseMatrix

(

const DenseMatrix< OtherValueType > &

other

)

References lama::Matrix::getColDistributionPtr(), lama::Distributed::getDistributionPtr(), lama::DenseMatrix< T >::getOwners(), LAMA_ASSERT_ERROR, lama::DenseMatrix< T >::mData, and lama::Matrix::setDistributedMatrix().

template<typename ValueType >

std::auto_ptr< Matrix > lama::DenseMatrix< ValueType >::create

(

)

const [virtual]

Constructing a new object by the default constructor.

Implements lama::Matrix.

std::auto_ptr< Matrix > lama::Matrix::create	(	const IndexType	numRows,
		const IndexType	numColumns
	)		const [inherited]

Constructor creates a replicated matrix of same type as a given matrix.

References lama::Matrix::create().

std::auto_ptr< Matrix > lama::Matrix::create

(

const IndexType

size

)

const [inherited]

Constructor creates a distributed zero matrix of same type as a given matrix.

References lama::Matrix::create().

std::auto_ptr< Matrix > lama::Matrix::create	(	DistributionPtr	rowDistribution,
		DistributionPtr	colDistribution
	)		const [inherited]

Constructor creates a distributed zero matrix of same type as a given matrix.

References lama::Matrix::create().

std::auto_ptr< Matrix > lama::Matrix::create

(

DistributionPtr

distribution

)

const [inherited]

Constructor creates a distributed zero matrix of same type as a given matrix.

References lama::Matrix::create().

std::auto_ptr< _LAMAArray > lama::Matrix::createArray

(

)

const [inherited]

This routine creates a LAMA array with the same value type as the matrix.

: returns an auto pointer to the LAMA array.

Same as _LAMAArray::create( this.getValueType() )

Value type is known only at runtime, so pointer to the base class is returned. Auto pointer indicates that calling routine takes ownership of the allocated array.

References lama::Matrix::create(), and lama::Matrix::getValueType().

VectorPtr lama::Matrix::createDenseVector	(	DistributionPtr	distribution,
		const Scalar	value
	)		const [inherited]

Constructor creates a distributed dense vector of same type as a given matrix.

References lama::Scalar::DOUBLE, lama::Scalar::FLOAT, lama::Scalar::getValue(), lama::Matrix::getValueType(), and LAMA_THROWEXCEPTION.

const Distribution & lama::Matrix::getColDistribution

(

)

const [inline, inherited]

gets a constant reference to the column distribution.

Returns:

a constant reference to the column distribution.

References LAMA_ASSERT_ERROR, and lama::Matrix::mColDistribution.

Referenced by lama::DenseMatrix< T >::assignSparse(), lama::Matrix::checkSettings(), lama::LUSolver::computeLUFactorization(), lama::InverseSolver::decompose(), lama::DenseMatrix< T >::DenseMatrix(), lama::DenseMatrixOps::invert(), lama::InverseSolver::invert(), lama::DenseMatrixOps::invertReplicated(), lama::SpecializedJacobi::iterateTyped(), lama::Matrix::Matrix(), lama::DenseMatrix< T >::matrixTimesMatrix(), lama::SparseMatrix< T >::matrixTimesMatrixImpl(), lama::CRTPMatrix< DenseMatrix< T >, T >::matrixTimesVector(), lama::SparseMatrix< T >::matrixTimesVectorNImpl(), lama::DenseMatrix< T >::maxDiffNorm(), lama::SparseMatrix< T >::maxDiffNorm(), lama::DenseMatrix< T >::maxDiffNormImpl(), lama::SparseMatrix< T >::maxDiffNormImpl(), lama::Matrix::operator=(), lama::LUSolver::solve(), and lama::Solver::solveInit().

DistributionPtr lama::Matrix::getColDistributionPtr

(

)

const [inline, inherited]

gets a pointer to the column distribution.

Returns:

a pointer to the column distribution.

References LAMA_ASSERT_ERROR, and lama::Matrix::mColDistribution.

Referenced by lama::SparseMatrix< T >::assign(), lama::DenseMatrix< T >::assign(), lama::DenseMatrix< T >::assignSparse(), lama::SparseMatrix< T >::assignTransposeImpl(), lama::DenseMatrix< T >::copyDenseMatrix(), lama::MatrixCreator< T >::fillRandom(), lama::DenseMatrixOps::invert(), lama::DenseMatrix< T >::invert(), lama::DenseMatrix< T >::matrixTimesMatrix(), lama::SparseMatrix< T >::matrixTimesMatrixImpl(), and lama::SparseMatrix< T >::matrixTimesVectorNImpl().

Matrix::SyncKind lama::Matrix::getCommunicationKind

(

)

const [inline, inherited]

get the communication kind.

Returns:

the communication kind.

References lama::Matrix::mCommunicationKind.

Referenced by lama::Matrix::inheritAttributes(), lama::SpecializedJacobi::iterateTyped(), and lama::DenseMatrix< T >::matrixTimesVectorImpl().

Matrix::SyncKind lama::Matrix::getComputeKind

(

)

const [inline, inherited]

get the compute kind.

Returns:

the compute kind.

References lama::Matrix::mComputeKind.

Referenced by lama::Matrix::inheritAttributes().

const Context& lama::Matrix::getContext

(

)

const [inline, inherited]

Reimplemented in lama::SparseMatrix< T >, and lama::SparseMatrix< ValueType >.

template<typename T>

virtual ContextPtr lama::DenseMatrix< T >::getContextPtr

(

)

const [inline, virtual]

Getter routine for the context.

Note: Only for SparseMatrix the context of the halo can be queried.

Implements lama::Matrix.

Referenced by lama::DenseMatrixOps::invertCyclic().

template<typename ValueType >

std::vector< typename DenseMatrix< ValueType >::DenseStoragePtr > & lama::DenseMatrix< ValueType >::getCyclicLocalValues

(

)

Referenced by lama::LUSolver::computeLUFactorization(), and lama::LUSolver::ptrsm().

template<typename ValueType >

const std::vector< typename DenseMatrix< ValueType >::DenseStoragePtr > & lama::DenseMatrix< ValueType >::getCyclicLocalValues

(

)

const

template<typename ValueType >

void lama::DenseMatrix< ValueType >::getDiagonal

(

Vector &

diagonal

)

const [virtual]

This method returns the diagonal.

Parameters:

[out]

diagonal

is the destination array

Calculations are dependent to the diagonal property.

Implements lama::Matrix.

References lama::Vector::assign(), lama::Scalar::DOUBLE, lama::Scalar::FLOAT, lama::Vector::getValueType(), and LAMA_THROWEXCEPTION.

template<typename T>

template<typename OtherT >

void lama::DenseMatrix< T >::getDiagonalImpl

(

DenseVector< OtherT > &

diagonal

)

const [private]

Optimized implementation for dense vectors as diagonal.

template<typename T>

template<typename OtherValueType >

void lama::DenseMatrix< T >::getDiagonalImpl

(

DenseVector< OtherValueType > &

diagonal

)

const

References lama::DenseVector< T >::allocate(), and lama::DenseVector< T >::getLocalValues().

const Distribution & lama::Distributed::getDistribution

(

)

const [inline, inherited]

References lama::Distributed::mDistribution.

Referenced by lama::SparseMatrix< T >::assignTransposeImpl(), lama::Matrix::checkSettings(), lama::LUSolver::computeLUFactorization(), lama::InverseSolver::decompose(), lama::DenseMatrix< T >::DenseMatrix(), lama::DenseVector< T >::DenseVector(), lama::DenseVector< T >::dotProduct(), lama::MatrixCreator< T >::fillRandom(), lama::SparseMatrix< T >::getLocalRow(), lama::SparseMatrix< T >::getRow(), lama::DenseMatrix< T >::getRow(), lama::DenseMatrixOps::invert(), lama::InverseSolver::invert(), lama::DenseMatrixOps::invertCyclic(), lama::DenseMatrixOps::invertReplicated(), lama::Matrix::Matrix(), lama::DenseMatrix< T >::matrixTimesMatrix(), lama::SparseMatrix< T >::matrixTimesMatrixImpl(), lama::CRTPMatrix< DenseMatrix< T >, T >::matrixTimesVector(), lama::SparseMatrix< T >::matrixTimesVector(), lama::SparseMatrix< T >::matrixTimesVectorNImpl(), lama::DenseMatrix< T >::maxDiffNorm(), lama::SparseMatrix< T >::maxDiffNorm(), lama::DenseMatrix< T >::maxDiffNormImpl(), lama::SparseMatrix< T >::maxDiffNormImpl(), lama::Matrix::operator=(), lama::SparseMatrix< T >::scale(), lama::SparseMatrix< T >::setDiagonal(), lama::LUSolver::solve(), lama::Solver::solveInit(), lama::Vector::Vector(), and lama::Vector::~Vector().

DistributionPtr lama::Distributed::getDistributionPtr

(

)

const [inline, inherited]

References lama::Distributed::mDistribution.

Referenced by lama::SparseMatrix< T >::assign(), lama::DenseMatrix< T >::assign(), lama::DenseVector< T >::assign(), lama::DenseMatrix< T >::assignSparse(), lama::SparseMatrix< T >::assignTransposeImpl(), lama::DenseMatrix< T >::copyDenseMatrix(), lama::DenseVector< T >::DenseVector(), lama::MatrixCreator< T >::fillRandom(), lama::SingleGridSetup::initialize(), lama::CG::initialize(), lama::DefaultJacobi::initialize(), lama::GMRES::initialize(), lama::DenseMatrixOps::invert(), lama::DenseMatrix< T >::invert(), lama::DenseMatrixOps::invertCyclic(), lama::GMRES::iterate(), lama::SparseMatrix< T >::matrixTimesMatrixImpl(), lama::CRTPMatrix< DenseMatrix< T >, T >::matrixTimesVector(), lama::Vector::operator=(), lama::Vector::size(), and lama::Vector::writeAt().

template<typename ValueType >

IndexType lama::DenseMatrix< ValueType >::getLocalNumColumns

(

)

const [virtual]

Getter routine for the local number of rows and columns.

Implements lama::Matrix.

template<typename ValueType >

IndexType lama::DenseMatrix< ValueType >::getLocalNumRows

(

)

const [virtual]

Getter routine for the local number of rows.

Implements lama::Matrix.

template<typename ValueType >

IndexType lama::DenseMatrix< ValueType >::getLocalNumValues

(

)

const [virtual]

Implementation of Matrix::getLocalNumValues, getLocalNumRows, getLocalNumColumns.

Implements lama::Matrix.

template<typename ValueType >

const DenseStorage< ValueType > & lama::DenseMatrix< ValueType >::getLocalStorage

(

)

const

This method returns the storage containing the local data regarding row/col distribution.

References LAMA_ASSERT_EQUAL_ERROR, and LAMA_ASSERT_ERROR.

Referenced by lama::InverseSolver::decompose(), lama::InverseSolver::invert(), lama::DenseMatrixOps::invertCyclic(), lama::DenseMatrixOps::invertReplicated(), and lama::SparseMatrix< T >::matrixTimesVectorNImpl().

template<typename ValueType >

DenseStorage< ValueType > & lama::DenseMatrix< ValueType >::getLocalStorage

(

)

References LAMA_ASSERT_EQUAL_ERROR, and LAMA_ASSERT_ERROR.

template<typename T>

Matrix::MatrixKind lama::DenseMatrix< T >::getMatrixKind

(

)

const [inline, virtual]

Each derived matrix must give info about its kind (DENSE or SPARSE).

Implements lama::Matrix.

References lama::DENSE.

template<typename ValueType >

size_t lama::DenseMatrix< ValueType >::getMemoryUsage

(

)

const [virtual]

getMemoryUsage returns the global memory that is allocated to hold this matrix.

getMemoryUsage returns the global memory that is allocated to hold this matrix. For a distributed matrix all partitions are summed together.

Returns:

the memory consumption of this matrix.

Implements lama::Matrix.

IndexType lama::Matrix::getNumColumns

(

)

const [inline, inherited]

Returns the number of columns.

Returns:

the number of columns.

References lama::Matrix::mNumColumns.

Referenced by lama::DenseMatrix< T >::assignSparse(), lama::LUSolver::computeLUFactorization(), lama::InverseSolver::decompose(), lama::DenseMatrix< T >::DenseMatrix(), lama::MatrixCreator< T >::fillRandom(), lama::SparseMatrix< T >::getLocalNumColumns(), lama::InverseSolver::invert(), lama::DenseMatrix< T >::invert(), lama::DenseMatrixOps::invertCyclic(), lama::MatrixConfigGrammar::MatrixConfigGrammar(), lama::Matrix::operator=(), lama::LUSolver::solve(), and lama::Solver::solveInit().

IndexType lama::Matrix::getNumRows

(

)

const [inline, inherited]

Returns the number of global rows.

Returns:

the number of rows.

References lama::Matrix::mNumRows.

Referenced by lama::InverseSolver::computeInverse(), lama::LUSolver::computeLUFactorization(), lama::InverseSolver::decompose(), lama::SparseMatrix< T >::getLocalNumRows(), lama::LUSolver::initialize(), lama::DenseMatrixOps::invert(), lama::InverseSolver::invert(), lama::DenseMatrix< T >::invert(), lama::DenseMatrixOps::invertCyclic(), lama::SpecializedJacobi::iterateTyped(), lama::MatrixConfigGrammar::MatrixConfigGrammar(), lama::Matrix::operator=(), lama::LUSolver::solve(), and lama::Solver::solveInit().

template<typename ValueType >

IndexType lama::DenseMatrix< ValueType >::getNumValues

(

)

const [virtual]

Get the total number of non-zero values in the matrix.

An element is considered to be non-zero if its absolute value is greater equal than mEpsilon. Zero diagonal elements are also counted if this->hasDiagonalProperty() is given.

This routine does not count zero elements even if they are stored (e.g. for dense or dia storage data).

Returns:

the number of non-zero values in this matrix.

Implements lama::Matrix.

template<typename T>

const std::vector<PartitionId>& lama::DenseMatrix< T >::getOwners

(

)

const [inline]

Referenced by lama::DenseMatrix< T >::copyDenseMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::getRow	(	Vector &	row,
		const IndexType	globalRowIndex
	)		const [virtual]

This method returns one row of the matrix.

Parameters:

[out]

row

is a replicated vector with all values of the row

Note: the value type of the vector should be the same type as the matrix (otherwise conversion) and it should be a replicated vector (otherwise reallocation)

Unclear: should the distribution always be unchanged ?

Implements lama::Matrix.

References lama::Vector::getValueType(), LAMA_ASSERT_DEBUG, and LAMA_THROWEXCEPTION.

template<typename ValueType >

void lama::DenseMatrix< ValueType >::getRow	(	DenseVector< ValueType > &	row,
		const IndexType	i
	)		const [private]

References lama::Communicator::bcast(), lama::Distributed::getDistribution(), lama::DenseVector< T >::getLocalValues(), lama::Communicator::getRank(), lama::Distribution::isReplicated(), LAMA_ASSERT_ERROR, LAMA_THROWEXCEPTION, lama::nIndex, and lama::Communicator::sum().

double lama::Matrix::getSparsityRate

(

)

const [inherited]

References lama::Matrix::getNumValues(), lama::Matrix::mNumColumns, and lama::Matrix::mNumRows.

template<typename ValueType >

const char * lama::DenseMatrix< ValueType >::getTypeName

(

)

const [virtual]

Implementation for Matrix::getTypeName()

Implements lama::Matrix.

template<typename ValueType >

Scalar lama::DenseMatrix< ValueType >::getValue	(	lama::IndexType	i,
		lama::IndexType	j
	)		const [virtual]

returns a copy of the value at the passed global indexes.

Parameters:

[in]	i	the global row index
[in]	j	the global column index

Returns:

a copy of the value at the passed global position.

As this operation requires communication ins SPMD mode it can be very inefficient in some situations.

Implements lama::Matrix.

References lama::Distribution::getCommunicator(), lama::Distribution::getNumPartitions(), lama::Communicator::getRank(), lama::Distribution::global2local(), lama::Distribution::isLocal(), LAMA_ASSERT_ERROR, lama::nIndex, and lama::Communicator::sum().

template<typename ValueType >

Scalar::ScalarType lama::DenseMatrix< ValueType >::getValueType

(

)

const [virtual]

Query the value type of the matrix elements, e.g.

DOUBLE or FLOAT.

Implements lama::Matrix.

template<typename ValueType >

bool lama::DenseMatrix< ValueType >::hasDiagonalProperty

(

)

const [virtual]

hasDiagonalProperty returns the diagonalProperty of the local storage.

Returns:

if the diagonal property is full filled.

Implements lama::Matrix.

void lama::Matrix::inheritAttributes

(

const Matrix &

other

)

[inherited]

Inherit context and kind arguments from another matrix.

This routine will also be used by copy constructors in base classes.

References lama::Matrix::getCommunicationKind(), lama::Matrix::getComputeKind(), lama::Matrix::getContextPtr(), lama::Matrix::setCommunicationKind(), lama::Matrix::setComputeKind(), and lama::Matrix::setContext().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::invert

(

const Matrix &

other

)

[virtual]

Set this matrix with the inverse of another square matrix.

This matrix will have afterwards the same distribution as the other matrix.

Implements lama::Matrix.

References lama::Matrix::getColDistributionPtr(), lama::Distribution::getCommunicatorPtr(), lama::Distributed::getDistributionPtr(), lama::Matrix::getNumColumns(), lama::Matrix::getNumRows(), and LAMA_ASSERT_ERROR.

Referenced by lama::SparseMatrix< T >::invert().

template<typename T>

void lama::DenseMatrix< T >::invert

(

)

[inline]

Invert in place.

References lama::DenseMatrix< T >::invert().

Referenced by lama::DenseMatrix< T >::invert().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::invertCyclic

(

)

[private]

Special implementation of invert in place for a cyclic distributed matrix.

References lama::CyclicDistribution::chunkSize(), lama::WriteAccess< T >::get(), lama::DenseStorageView< T >::getData(), lama::_MatrixStorage::getNumRows(), lama::LAMAInterface::getSCALAPACKInterface(), LAMA_ASSERT_EQUAL_ERROR, LAMA_ASSERT_ERROR, LAMA_REGION, LAMA_THROWEXCEPTION, and lama::_LAMAArray::size().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::joinColumnData	(	DenseStorage< ValueType > &	result,
		const std::vector< boost::shared_ptr< DenseStorage< ValueType > > > &	chunks,
		const std::vector< IndexType > &	columnOwners
	)		[static, private]

Join dense storage of column distributed data.

Parameters:

[out]	result	will be the joined storage
[in]	chunks	is a vector of all chunks, one chunk for each partition
[in]	columnOnwers	vector with owner for each column

Note: the column distribution itself is given implicitly by the vector of owners

References lama::DenseStorageView< T >::getData(), lama::_MatrixStorage::getNumColumns(), lama::_MatrixStorage::getNumRows(), LAMA_ASSERT_EQUAL_ERROR, and LAMA_ASSERT_ERROR.

template<typename T>

lama::DenseMatrix< T >::LAMA_LOG_DECL_STATIC_LOGGER

(

logger

)

[private]

Reimplemented from lama::Matrix.

template<typename ValueType >

void lama::DenseMatrix< ValueType >::localize	(	DenseStorage< ValueType > &	local,
		const DenseStorage< ValueType > &	global,
		const Distribution &	rowDistribution
	)		[static, private]

Restrict dense storage of a replicated matrix to its local part according to row distribution.

Parameters:

[out]	local	is the local part of the dense storage for this partition
[in]	global	is the replicated dense storage
[in]	rowDistribution	distribution used to localize the dense storage.

References lama::DenseStorageView< T >::allocate(), lama::DenseStorageView< T >::getData(), lama::Distribution::getGlobalSize(), lama::Distribution::getLocalSize(), lama::_MatrixStorage::getNumColumns(), lama::_MatrixStorage::getNumRows(), LAMA_ASSERT_EQUAL_ERROR, and lama::Distribution::local2global().

Referenced by lama::DenseMatrix< T >::DenseMatrix().

void lama::Matrix::matrix2CSRGraph	(	IndexType *	xadj,
		IndexType *	adjncy,
		IndexType *	vwgt,
		CommunicatorPtr	comm,
		const IndexType *	globalRowIndices = NULL,
		IndexType *	vtxdist = NULL
	)		const [virtual, inherited]

transformation from matrix type to a csr graph

transformation from matrix type to a csr graph, so that it (Par)Metis can work with it.

Parameters:

[out]	xadj	the ia array of the csr graph
[out]	adjncy	the ja array of the csr graph

References LAMA_THROWEXCEPTION.

template<typename ValueType >

void lama::DenseMatrix< ValueType >::matrixTimesMatrix	(	Matrix &	result,
		const Scalar	alpha,
		const Matrix &	x,
		const Scalar	beta,
		const Matrix &	y
	)		const [virtual]

Implementation of pure method Matrix::matrixTimesMatrix.

Implements lama::Matrix.

References lama::Matrix::getColDistribution(), lama::Matrix::getColDistributionPtr(), lama::Distributed::getDistribution(), lama::Scalar::getValue(), lama::Distribution::isReplicated(), LAMA_ASSERT_ERROR, LAMA_THROWEXCEPTION, lama::DenseMatrix< T >::mData, and lama::DenseMatrix< T >::prefetch().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::matrixTimesScalar	(	const Matrix &	other,
		const Scalar	alpha
	)		[virtual]

matrixTimesScalar computes this = alpha * other

Parameters:

[out]	other	the Matrix to mutliply
[in]	alpha	the Scalar of the expression

matrixTimesScalar computes this = alpha * other.

Implements lama::Matrix.

References lama::Scalar::getValue().

void lama::CRTPMatrix< DenseMatrix< T > , T >::matrixTimesVector	(	Vector &	result,
		const Scalar	alpha,
		const Vector &	x,
		const Scalar	beta,
		const Vector &	y
	)		const [inline, virtual, inherited]

matrixTimesVector computes result = alpha * this * x + beta * y

Parameters:

[out]	result	the Vector to store the result to
[in]	alpha	the Scalar alpha of the expression
[in]	x	the Vector x of the expression
[in]	beta	the Scalar beta of the expression
[in]	y	the Vector y of the expression

matrixTimesVector computes result = alpha * this * x + beta * y. If result == x or result == y new storage is allocated to store the result.

Implements lama::Matrix.

References lama::Matrix::getColDistribution(), lama::Distributed::getDistribution(), lama::Distributed::getDistributionPtr(), lama::Scalar::getValue(), LAMA_ASSERT, LAMA_ASSERT_EQUAL, LAMA_REGION, LAMA_THROWEXCEPTION, and lama::Vector::resize().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::matrixTimesVectorImpl	(	DenseVector< ValueType > &	denseResult,
		const ValueType	alphaValue,
		const DenseVector< ValueType > &	denseX,
		const ValueType	betaValue,
		const DenseVector< ValueType > &	denseY
	)		const

Matrix times vector with same values types and correct distributions.

Note: Matrix::matrixTimesMatrix is implemented in the CRTPMatrix class. that requires this method.

Note: all vectors must have the right distribution.

References lama::Matrix::ASYNCHRONOUS, lama::Matrix::getCommunicationKind(), lama::Distribution::getCommunicator(), lama::DenseVector< T >::getLocalValues(), lama::Communicator::getNeighbor(), lama::Distribution::getNumPartitions(), lama::Communicator::getRank(), LAMA_REGION, lama::DenseStorageView< T >::matrixTimesVector(), lama::Communicator::max(), lama::DenseVector< T >::prefetch(), lama::WriteAccess< T >::resize(), lama::Communicator::shift(), lama::Communicator::shiftAsync(), lama::_LAMAArray::size(), lama::SyncToken::wait(), and lama::zero.

template<typename ValueType >

Scalar lama::DenseMatrix< ValueType >::maxDiffNorm

(

const Matrix &

other

)

const [virtual]

Implementation of Matrix::maxDiffNorm for dense matrices.

Implements lama::Matrix.

References lama::Matrix::getColDistribution(), lama::Distributed::getDistribution(), lama::Matrix::getValueType(), LAMA_ASSERT_DEBUG, and LAMA_UNSUPPORTED.

template<typename ValueType >

ValueType lama::DenseMatrix< ValueType >::maxDiffNormImpl

(

const DenseMatrix< ValueType > &

other

)

const

Get the maximal difference between two elements for dense matrices of same type.

References lama::Matrix::getColDistribution(), lama::Distributed::getDistribution(), LAMA_ASSERT_EQUAL_ERROR, lama::Communicator::max(), and lama::DenseMatrix< T >::mData.

Scalar lama::Matrix::operator()	(	IndexType	i,
		IndexType	j
	)		const [inherited]

returns a copy of the value at the passed global indexes.

Parameters:

[in]	i	the global row index
[in]	j	the global column index

Returns:

a copy of the value at the passed global position.

As this operator requires communication ins SPMD mode it can be very inefficient in some situations.

References lama::Matrix::getValue().

template<typename ValueType >

DenseMatrix< ValueType > & lama::DenseMatrix< ValueType >::operator=

(

const DenseMatrix< T > &

matrix

)

Override default assignment operator needed to guarantee deep copy.

template<typename ValueType >

DenseMatrix< ValueType > & lama::DenseMatrix< ValueType >::operator=

(

const Matrix &

other

)

the assignment operator for matrix.

Parameters:

[in]

other

is the input matrix.

The assignment operator will make a deep copy of the input matrix. Size and distributions are inherited, but there might be implicit conversions regarding storage format and/or value type of the matrix elements.

Reimplemented from lama::Matrix.

template<typename ValueType >

DenseMatrix< ValueType > & lama::DenseMatrix< ValueType >::operator=

(

const Expression< Scalar, Matrix, Times >

exp

)

the assignment operator for a scalar matrix multiplication.

Reimplemented from lama::Matrix.

References lama::Matrix::operator=().

template<typename ValueType >

DenseMatrix< ValueType > & lama::DenseMatrix< ValueType >::operator=

(

const Expression< Matrix, Matrix, Times >

exp

)

the assignment operator for a matrix matrix multiplication.

Reimplemented from lama::Matrix.

References lama::Matrix::operator=().

template<typename ValueType >

DenseMatrix< ValueType > & lama::DenseMatrix< ValueType >::operator=

(

const Expression< Scalar, Expression< Matrix, Matrix, Times >, Times >

exp

)

the assignment operator for a scalar matrix matrix multiplication.

the assignment operator for a matrix matrix multiplication.

Reimplemented from lama::Matrix.

References lama::Matrix::operator=().

template<typename ValueType >

DenseMatrix< ValueType > & lama::DenseMatrix< ValueType >::operator=

(

const Expression< Expression< Scalar, Expression< Matrix, Matrix, Times >, Times >, Expression< Scalar, Matrix, Times >, Plus >

exp

)

the assignment operator for a GEMM expression.

Reimplemented from lama::Matrix.

References lama::Matrix::operator=().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::prefetch

(

)

const [virtual]

Prefetch matrix data to its 'preferred' context location.

Implements lama::Matrix.

Referenced by lama::DenseMatrix< T >::matrixTimesMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::prefetch

(

lama::ContextPtr

loc

)

const

template<typename ValueType >

void lama::DenseMatrix< ValueType >::redistribute	(	DistributionPtr	rowDistribution,
		DistributionPtr	colDistribution
	)		[virtual]

Implementation of pure method.

Implements lama::Matrix.

References LAMA_REGION, and lama::Matrix::mColDistribution.

Referenced by lama::DenseMatrixOps::invert().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::redistributeRows

(

DistributionPtr

rowDistribution

)

[private]

References lama::DenseStorageView< T >::getData(), lama::_MatrixStorage::getNumColumns(), LAMA_ASSERT_EQUAL_DEBUG, lama::replicate(), and lama::DenseStorageView< T >::swap().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::resetDiagonalProperty

(

)

[virtual]

resetDiagonalProperty rechecks the storages for their diagonal property

Returns:

if the diagonal property is full filled.

Implements lama::Matrix.

template<typename ValueType >

void lama::DenseMatrix< ValueType >::scale

(

const Vector &

scaling

)

[virtual]

This method scales all values with a vector.

Parameters:

[in]

values

is the source array

row wise calculations

Implements lama::Matrix.

References lama::Scalar::DOUBLE, lama::Scalar::FLOAT, lama::DenseVector< T >::getLocalValues(), and LAMA_THROWEXCEPTION.

template<typename ValueType >

void lama::DenseMatrix< ValueType >::scale

(

const Scalar

scaling

)

[virtual]

This method scales all matrix values with a scalar.

Parameters:

[in]

value

is the source value

Implements lama::Matrix.

References LAMA_THROWEXCEPTION.

void lama::Matrix::setCommunicationKind

(

SyncKind

communicationKind

)

[inherited]

set the communication kind.

Parameters:

[in]

communicationKind

the communication kind.

References lama::Matrix::mCommunicationKind.

Referenced by lama::Matrix::inheritAttributes().

void lama::Matrix::setComputeKind

(

SyncKind

computeKind

)

[virtual, inherited]

set the compute kind.

Parameters:

[in]

computeKind

the compute kind.

Reimplemented in lama::SparseMatrix< T >, and lama::SparseMatrix< ValueType >.

References lama::Matrix::mComputeKind.

Referenced by lama::Matrix::inheritAttributes().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::setContext

(

const ContextPtr

context

)

[virtual]

specifies on which compute back end the matrix operations should take place.

Parameters:

[in]

context

the compute back to use for calculations with matrix

Note: Only for sparse matrices it is possible to specify separate locations for local and halo computations.

Implements lama::Matrix.

void lama::Matrix::setContext	(	const ContextPtr	localContext,
		const ContextPtr	haloContext
	)		[virtual, inherited]

Only sparse matrices will override this method, others will ignore second argument.

Reimplemented in lama::SparseMatrix< T >, and lama::SparseMatrix< ValueType >.

References lama::Matrix::setContext().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::setCSRData	(	const LAMAArray< IndexType > &	rowIA,
		const LAMAArray< IndexType > &	rowJA,
		const _LAMAArray &	rowValues,
		DistributionPtr	rowDistribution,
		DistributionPtr	colDistribution
	)		[virtual]

Implementation of pure method.

template<typename ValueType >

void lama::DenseMatrix< ValueType >::setCSRDataLocal	(	const LAMAArray< IndexType > &	rowIA,
		const LAMAArray< IndexType > &	rowJA,
		const _LAMAArray &	rowValues
	)		const

Local version of setCSRData .

References lama::_LAMAArray::size().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::setDiagonal

(

const Vector &

diagonal

)

[virtual]

This method replaces the diagonal.

Parameters:

[in]

diagonal

is the source array

Calculations are dependent to the diagonal property

Implements lama::Matrix.

References lama::Scalar::DOUBLE, lama::Scalar::FLOAT, lama::DenseVector< T >::getLocalValues(), and LAMA_THROWEXCEPTION.

template<typename ValueType >

void lama::DenseMatrix< ValueType >::setDiagonal

(

const Scalar

scalar

)

[virtual]

This method replaces the diagonal by a diagonal value.

Parameters:

[in]

scalar

is the source value

Calculations are dependent to the diagonal property

Implements lama::Matrix.

References LAMA_THROWEXCEPTION.

void lama::Matrix::setDistributedMatrix	(	DistributionPtr	distribution,
		DistributionPtr	colDistribution
	)		[protected, inherited]

Set the global and local size of distributed matrix.

Parameters:

[in]	distribution	is distribution for rows
[in]	colDistribution	is distribution for columns

Global and local size is given implicitly by the distributions itself.

References LAMA_ASSERT_ERROR, lama::Matrix::mColDistribution, lama::Matrix::mNumColumns, lama::Matrix::mNumRows, and lama::Distributed::setDistributionPtr().

Referenced by lama::DenseMatrix< T >::allocate(), lama::SimpleStorageStrategy< ValueType >::applyStrategy(), lama::DenseMatrix< T >::assign(), lama::DenseMatrix< T >::assignSparse(), lama::COOSparseMatrix< T >::COOSparseMatrix(), lama::DenseMatrix< T >::copyDenseMatrix(), lama::CSRSparseMatrix< T >::CSRSparseMatrix(), lama::DIASparseMatrix< T >::DIASparseMatrix(), lama::ELLSparseMatrix< T >::ELLSparseMatrix(), lama::JDSSparseMatrix< T >::JDSSparseMatrix(), lama::Matrix::Matrix(), lama::Matrix::setReplicatedMatrix(), lama::DenseMatrix< T >::splitColumns(), and lama::XXXSparseMatrix< T >::XXXSparseMatrix().

void lama::Distributed::setDistributionPtr

(

DistributionPtr

distributionPtr

)

[protected, inherited]

References lama::Distributed::mDistribution.

Referenced by lama::Vector::resize(), and lama::Matrix::setDistributedMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::setIdentity

(

)

[virtual]

Operator that sets the matrix to the identity matrix.

 void sub( ..., Matrix& a, ... )
 ...
 LAMA_ASSERT_EQUAL_DEBUG( a.getNumRows(), a.getNumColumns() );
 A.setIdentity();

Implements lama::Matrix.

References lama::Communicator::getRank(), lama::Communicator::getSize(), LAMA_ASSERT_EQUAL_DEBUG, and LAMA_THROWEXCEPTION.

template<typename ValueType >

template<typename OtherValueType >

void lama::DenseMatrix< ValueType >::setRawDenseData	(	const IndexType	numRows,
		const IndexType	numColumns,
		const OtherValueType	values[],
		const OtherValueType	eps = 0.0
	)

Set matrix with global dense data.

void lama::Matrix::setReplicatedMatrix	(	const IndexType	numRows,
		const IndexType	numColumns
	)		[protected, inherited]

Set the global/local size of replicated matrix.

Parameters:

[in]	numRows	number of rows, must be non-negative.
[in]	numColumns	number of columns, must be non-negative.

References lama::Matrix::setDistributedMatrix().

Referenced by lama::DenseMatrix< T >::allocate(), lama::DenseMatrix< T >::assign(), lama::DenseMatrix< T >::clear(), and lama::DenseMatrix< T >::DenseMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::splitColumnData	(	std::vector< boost::shared_ptr< DenseStorage< ValueType > > > &	chunks,
		const DenseStorage< ValueType > &	columnData,
		const PartitionId	numChunks,
		const std::vector< IndexType > &	columnOwners
	)		[static, private]

Split dense storage data according to a distribution into chunks.

Parameters:

[out]	chunks	vector of shared pointer to the new allocated chunks
[in]	columnData	is the dense storage to split
[in]	numChunks	is the number of chunks, same as #partitions of distribution
[in]	columnOnwers	owner for each column

References lama::DenseStorageView< T >::getData(), lama::_MatrixStorage::getNumColumns(), lama::_MatrixStorage::getNumRows(), LAMA_ASSERT_DEBUG, and LAMA_ASSERT_EQUAL_ERROR.

Referenced by lama::DenseMatrix< T >::DenseMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::splitColumns

(

DistributionPtr

colDistribution

)

[private]

Split the replicated columns into chunks according to the column distribution.

References LAMA_ASSERT_EQUAL_ERROR, and lama::Matrix::setDistributedMatrix().

void lama::Distributed::swap

(

Distributed &

other

)

[protected, inherited]

References lama::Distributed::mDistribution.

Referenced by lama::Matrix::swapMatrix().

template<typename ValueType >

void lama::DenseMatrix< ValueType >::swap

(

DenseMatrix< ValueType > &

other

)

Swap will swap all member variables of the two dense matrices.

This operation might be useful in iteration loops where a dense matrix is updated each iteration. It is more convenient than a solution that is based on using pointers in the application.

References lama::DenseMatrix< T >::mData, lama::DenseMatrix< T >::mOwners, and lama::Matrix::swapMatrix().

void lama::Matrix::swapMatrix

(

Matrix &

other

)

[protected, inherited]

References lama::Matrix::mColDistribution, lama::Matrix::mNumColumns, lama::Matrix::mNumRows, and lama::Distributed::swap().

Referenced by lama::DenseMatrix< T >::swap().

template<typename T>

static const char* lama::DenseMatrix< T >::typeName

(

)

[static]

Getter for the type name of the class.

template<>

const char * lama::DenseMatrix< float >::typeName

(

)

template<>

const char * lama::DenseMatrix< double >::typeName

(

)

template<typename ValueType >

void lama::DenseMatrix< ValueType >::wait

(

)

const [virtual]

wait for a possibly running prefetch.

Implements lama::Matrix.

template<typename ValueType >

void lama::DenseMatrix< ValueType >::writeAt

(

std::ostream &

stream

)

const [virtual]

Write some information about this to the passed stream.

Parameters:

[out]

stream

the stream to write to.

Reimplemented from lama::Matrix.

template<typename ValueType >

void lama::DenseMatrix< ValueType >::writeToFile	(	const std::string &	fileName,
		const File::FileType	fileType = File::BINARY,
		const File::DataType	dataType = File::INTERNAL,
		const File::IndexDataType	indexDataTypeIA = File::INT,
		const File::IndexDataType	indexDataTypeJA = File::INT
	)		const

Method writes dense matrix to a file.

Writing is only supported for a replicated matrix.

References lama::Communicator::getRank(), LAMA_THROWEXCEPTION, and lama::Communicator::synchronize().

---

Field Documentation

DistributionPtr lama::Matrix::mColDistribution [protected, inherited]

Referenced by lama::Matrix::checkSettings(), lama::Matrix::getColDistribution(), lama::Matrix::getColDistributionPtr(), lama::DenseMatrix< T >::redistribute(), lama::Matrix::setDistributedMatrix(), and lama::Matrix::swapMatrix().

template<typename T>

std::vector<boost::shared_ptr<DenseStorage<ValueType> > > lama::DenseMatrix< T >::mData

local data is allocated in chunks according to column distribution

Referenced by lama::DenseMatrix< T >::copyDenseMatrix(), lama::DenseMatrix< T >::DenseMatrix(), lama::DenseMatrix< T >::matrixTimesMatrix(), lama::DenseMatrix< T >::maxDiffNormImpl(), and lama::DenseMatrix< T >::swap().

IndexType lama::Matrix::mNumColumns [protected, inherited]

Referenced by lama::Matrix::checkSettings(), lama::DenseMatrix< T >::DenseMatrix(), lama::Matrix::getNumColumns(), lama::Matrix::getSparsityRate(), lama::Matrix::Matrix(), lama::Matrix::setDistributedMatrix(), lama::Matrix::swapMatrix(), and lama::Matrix::writeAt().

IndexType lama::Matrix::mNumRows [protected, inherited]

Referenced by lama::Matrix::checkSettings(), lama::DenseMatrix< T >::DenseMatrix(), lama::Matrix::getNumRows(), lama::Matrix::getSparsityRate(), lama::Matrix::Matrix(), lama::Matrix::setDistributedMatrix(), lama::Matrix::swapMatrix(), and lama::Matrix::writeAt().

template<typename T>

std::vector<PartitionId> lama::DenseMatrix< T >::mOwners [protected]

Referenced by lama::DenseMatrix< T >::DenseMatrix(), and lama::DenseMatrix< T >::swap().

template<typename T>

LAMAArray<ValueType> lama::DenseMatrix< T >::mReceiveValues [mutable, private]

template<typename T>

LAMAArray<ValueType> lama::DenseMatrix< T >::mSendValues [mutable, private]

---

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

• /home/brandes/workspace/LAMA/src/lama/matrix/DenseMatrix.hpp
• /home/brandes/workspace/LAMA/src/lama/matrix/DenseMatrix.cpp