#include <MatrixOperations.h>

Inheritance diagram for AMP::LinearAlgebra::MatrixOperations:

[legend]

Public Member Functions
virtual void	axpy (AMP::Scalar alpha, const MatrixData &X, MatrixData &Y)=0
	Compute the linear combination of two matrices.

virtual void	copy (const MatrixData &x, MatrixData &y)=0
	Set this matrix with the same non-zero and distributed structure as x and copy the coefficients.

virtual void	copyCast (const MatrixData &x, MatrixData &y)
	Set this matrix with the same non-zero and distributed structure as x and copy the coefficients after up/down casting.

virtual void	extractDiagonal (MatrixData const &A, std::shared_ptr< Vector > buf)=0
	Extract the diagonal values into a vector.

uint64_t	getID () const
	Get a unique id hash for the vector operation.

virtual void	getRowSums (MatrixData const &A, std::shared_ptr< Vector > buf)
	Extract the row sums into a vector.

virtual void	getRowSumsAbsolute (MatrixData const &A, std::shared_ptr< Vector > buf, const bool remove_zeros=false)
	Extract the absolute row sums into a vector.

virtual AMP::Scalar	LinfNorm (const MatrixData &X) const =0
	Compute the maximum row sum.

virtual void	matMatMult (std::shared_ptr< MatrixData > A, std::shared_ptr< MatrixData > B, std::shared_ptr< MatrixData > C)=0
	Compute the product of two matrices.

virtual void	mult (std::shared_ptr< const Vector > x, MatrixData const &A, std::shared_ptr< Vector > y)=0
	Matrix-vector multiplication.

virtual void	multTranspose (std::shared_ptr< const Vector > in, MatrixData const &A, std::shared_ptr< Vector > out)=0
	Matrix transpose-vector multiplication.

virtual void	registerChildObjects (AMP::IO::RestartManager *manager) const
	Register any child objects.

virtual void	scale (AMP::Scalar alpha, MatrixData &A)=0
	Scale the matrix by a scalar.

virtual void	scale (AMP::Scalar alpha, std::shared_ptr< const Vector > D, MatrixData &A)=0
	Scale the matrix by a scalar and diagonal matrix.

virtual void	scaleInv (AMP::Scalar alpha, std::shared_ptr< const Vector > D, MatrixData &A)=0
	Scale the matrix by a scalar and inverse of diagonal matrix.

virtual void	setDiagonal (std::shared_ptr< const Vector > in, MatrixData &A)=0
	Set the diagonal to the values in a vector.

virtual void	setIdentity (MatrixData &A)=0
	Set the matrix to the identity matrix.

virtual void	setScalar (AMP::Scalar alpha, MatrixData &A)=0
	Set the non-zeros of the matrix to a scalar.

virtual std::string	type () const
	Return the type of the matrix operations class.

virtual void	writeRestart (int64_t fid) const
	Write restart data to file.

virtual void	zero (MatrixData &A)=0
	Set the non-zeros of the matrix to zero.

virtual	~MatrixOperations ()
	Destructor.

Protected Member Functions
	MatrixOperations ()

Protected Attributes
uint64_t	d_hash = 0
	unique hash for object

Detailed Description

Definition at line 16 of file MatrixOperations.h.

Constructor & Destructor Documentation

◆ ~MatrixOperations()

virtual AMP::LinearAlgebra::MatrixOperations::~MatrixOperations ( )

inlinevirtual

Destructor.

Definition at line 20 of file MatrixOperations.h.

◆ MatrixOperations()

AMP::LinearAlgebra::MatrixOperations::MatrixOperations ( )

protected

Member Function Documentation

◆ axpy()

virtual void AMP::LinearAlgebra::MatrixOperations::axpy	(	AMP::Scalar	alpha,
		const MatrixData &	X,
		MatrixData &	Y
	)

pure virtual

Compute the linear combination of two matrices.

Parameters

[in]	alpha	scalar
[in]	X	matrix
[out]	Y	Output matrix Y

Compute \(\mathbf{THIS} = \alpha\mathbf{X} + \mathbf{THIS}\)

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ copy()

virtual void AMP::LinearAlgebra::MatrixOperations::copy	(	const MatrixData &	x,
		MatrixData &	y
	)

pure virtual

Set this matrix with the same non-zero and distributed structure as x and copy the coefficients.

Parameters

[in]	x	matrix data to copy from
[in]	y	matrix data to copy to

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ copyCast()

virtual void AMP::LinearAlgebra::MatrixOperations::copyCast	(	const MatrixData &	x,
		MatrixData &	y
	)

virtual

Set this matrix with the same non-zero and distributed structure as x and copy the coefficients after up/down casting.

Parameters

[in]	x	matrix data to copy from
[in]	y	matrix data to copy to after up/down casting the coefficients

Reimplemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, and AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >.

◆ extractDiagonal()

virtual void AMP::LinearAlgebra::MatrixOperations::extractDiagonal	(	MatrixData const &	A,
		std::shared_ptr< Vector >	buf
	)

pure virtual

Extract the diagonal values into a vector.

Parameters

[in]	A	The matrix to get the diagonal from
[in]	buf	Vector to store the diagonal to

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ getID()

uint64_t AMP::LinearAlgebra::MatrixOperations::getID ( ) const

Get a unique id hash for the vector operation.

◆ getRowSums()

virtual void AMP::LinearAlgebra::MatrixOperations::getRowSums	(	MatrixData const &	A,
		std::shared_ptr< Vector >	buf
	)

virtual

Extract the row sums into a vector.

Parameters

[in]	A	The matrix to get the row sums from
[in]	buf	Vector to store the row sums to

Reimplemented in AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, and AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >.

◆ getRowSumsAbsolute()

virtual void AMP::LinearAlgebra::MatrixOperations::getRowSumsAbsolute	(	MatrixData const &	A,
		std::shared_ptr< Vector >	buf,
		const bool	remove_zeros = `false`
	)

virtual

Extract the absolute row sums into a vector.

Parameters

[in]	A	The matrix to get the row sums from
[in]	buf	Vector to store the row sums to
[in]	remove_zeros	If true zero values in sum are replaced with ones

Reimplemented in AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, and AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >.

◆ LinfNorm()

virtual AMP::Scalar AMP::LinearAlgebra::MatrixOperations::LinfNorm ( const MatrixData & X ) const

pure virtual

Compute the maximum row sum.

Returns: The L-infinity norm of the matrix

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ matMatMult()

virtual void AMP::LinearAlgebra::MatrixOperations::matMatMult	(	std::shared_ptr< MatrixData >	A,
		std::shared_ptr< MatrixData >	B,
		std::shared_ptr< MatrixData >	C
	)

pure virtual

Compute the product of two matrices.

Parameters

[in]	A	A multiplicand
[in]	B	A multiplicand
[out]	C	The product \(\mathbf{AB}\).

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ mult()

virtual void AMP::LinearAlgebra::MatrixOperations::mult	(	std::shared_ptr< const Vector >	x,
		MatrixData const &	A,
		std::shared_ptr< Vector >	y
	)

pure virtual

Matrix-vector multiplication.

Parameters

[in]	x	The vector to multiply
[in]	A	The input matrix A
[out]	y	The resulting vectory

Compute \(\mathbf{Ax} = \mathbf{y}\).

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ multTranspose()

virtual void AMP::LinearAlgebra::MatrixOperations::multTranspose	(	std::shared_ptr< const Vector >	in,
		MatrixData const &	A,
		std::shared_ptr< Vector >	out
	)

pure virtual

Matrix transpose-vector multiplication.

Parameters

[in]	in	The vector to multiply
[in]	A	The input matrix A
[out]	out	The resulting vectory

Compute \(\mathbf{A}^T\mathbf{in} = \mathbf{out}\).

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ registerChildObjects()

virtual void AMP::LinearAlgebra::MatrixOperations::registerChildObjects ( AMP::IO::RestartManager * manager ) const

virtual

Register any child objects.

This function will register child objects with the manager

Parameters

manager Restart manager

◆ scale() [1/2]

virtual void AMP::LinearAlgebra::MatrixOperations::scale	(	AMP::Scalar	alpha,
		MatrixData &	A
	)

pure virtual

Scale the matrix by a scalar.

Parameters

[in]	alpha	The value to scale by
[in]	A	The input matrix A

Compute \(\mathbf{A} = \alpha\mathbf{A}\)

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ scale() [2/2]

virtual void AMP::LinearAlgebra::MatrixOperations::scale	(	AMP::Scalar	alpha,
		std::shared_ptr< const Vector >	D,
		MatrixData &	A
	)

pure virtual

Scale the matrix by a scalar and diagonal matrix.

Parameters

[in]	alpha	The value to scale by
[in]	D	A vector representing the diagonal matrix
[in]	A	The input matrix A

Compute \(\mathbf{A} = \alpha\mathbf{D}\mathbf{A}\)

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ scaleInv()

virtual void AMP::LinearAlgebra::MatrixOperations::scaleInv	(	AMP::Scalar	alpha,
		std::shared_ptr< const Vector >	D,
		MatrixData &	A
	)

pure virtual

Scale the matrix by a scalar and inverse of diagonal matrix.

Parameters

[in]	alpha	The value to scale by
[in]	D	A vector representing the diagonal matrix
[in]	A	The input matrix A

Compute \(\mathbf{A} = \alpha\mathbf{D}^{-1}\mathbf{A}\)

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ setDiagonal()

virtual void AMP::LinearAlgebra::MatrixOperations::setDiagonal	(	std::shared_ptr< const Vector >	in,
		MatrixData &	A
	)

pure virtual

Set the diagonal to the values in a vector.

Parameters

[in]	in	The values to set the diagonal to
[out]	A	The matrix A to set

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ setIdentity()

virtual void AMP::LinearAlgebra::MatrixOperations::setIdentity ( MatrixData & A )

pure virtual

Set the matrix to the identity matrix.

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ setScalar()

virtual void AMP::LinearAlgebra::MatrixOperations::setScalar	(	AMP::Scalar	alpha,
		MatrixData &	A
	)

pure virtual

Set the non-zeros of the matrix to a scalar.

Parameters

[in]	alpha	The value to set the non-zeros to
[out]	A	Matrix to set

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

◆ type()

virtual std::string AMP::LinearAlgebra::MatrixOperations::type ( ) const

inlinevirtual

Return the type of the matrix operations class.

Reimplemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, and AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >.

Definition at line 143 of file MatrixOperations.h.

◆ writeRestart()

virtual void AMP::LinearAlgebra::MatrixOperations::writeRestart ( int64_t fid ) const

virtual

Write restart data to file.

This function will write the mesh to an HDF5 file

Parameters

fid	File identifier to write

Reimplemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, and AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >.

◆ zero()

virtual void AMP::LinearAlgebra::MatrixOperations::zero ( MatrixData & A )

pure virtual

Set the non-zeros of the matrix to zero.

May not deallocate space.

Implemented in AMP::LinearAlgebra::CSRMatrixOperationsDefault< Config >, AMP::LinearAlgebra::DenseSerialMatrixOperations, AMP::LinearAlgebra::CSRMatrixOperationsDevice< Config >, AMP::LinearAlgebra::NativePetscMatrixOperations, and AMP::LinearAlgebra::EpetraMatrixOperations.

Member Data Documentation

◆ d_hash

uint64_t AMP::LinearAlgebra::MatrixOperations::d_hash = 0

protected

unique hash for object

Definition at line 169 of file MatrixOperations.h.

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

src/matrices/operations/MatrixOperations.h

Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

◆ ~MatrixOperations()

◆ MatrixOperations()

Member Function Documentation

◆ axpy()

◆ copy()

◆ copyCast()

◆ extractDiagonal()

◆ getID()

◆ getRowSums()

◆ getRowSumsAbsolute()

◆ LinfNorm()

◆ matMatMult()

◆ mult()

◆ multTranspose()

◆ registerChildObjects()

◆ scale() [1/2]

◆ scale() [2/2]

◆ scaleInv()

◆ setDiagonal()

◆ setIdentity()

◆ setScalar()

◆ type()

◆ writeRestart()

◆ zero()

Member Data Documentation

◆ d_hash