The device sparse matrix class (compressed sparse row format). More...

Inheritance diagram for culite::csr::XxMatrix< T_Int, T_Scalar >:

[legend]

Constructors
	XxMatrix ()
	Default constructor.
	XxMatrix (T_Int nr, T_Int nc, T_Int nz, const Property &pr=Property::General())
	Dimension constructor.
	XxMatrix (T_Int nr, T_Int nc, T_Int rptr, T_Int cidx, T_Scalar *vals, bool bind, const Property &pr=Property::General())
	Auxiliary constructor.
	XxMatrix (const XxMatrix< T_Int, T_Scalar > &other)
	Copy constructor.
	XxMatrix (XxMatrix< T_Int, T_Scalar > &&other)
	Move constructor.
	~XxMatrix ()
	Destructor.

Operators
XxMatrix< T_Int, T_Scalar > &	operator= (const XxMatrix< T_Int, T_Scalar > &other)
	Copy assignment operator.
XxMatrix< T_Int, T_Scalar > &	operator= (XxMatrix< T_Int, T_Scalar > &&other)
	Move assignment operator.
alias::VirtualScal_csr< T_Int, T_Scalar >	operator- () const
	Unary negation operator.

Arguments
T_Int *	rowptr ()
	Access the row pointer array.
const T_Int *	rowptr () const
	Access the row pointer array.
T_Int *	colidx ()
	Access the column index array.
const T_Int *	colidx () const
	Access the column index array.
T_Int	nnz () const
	Get the number of non-zero elements.

Public Member Functions
void	clear ()
	Clear the device sparse matrix.
std::string	info (const std::string &header="") const
	Get information about the sparse matrix.
XxMatrix< T_Int, T_Scalar >	copy () const
	Create a deep copy of the device sparse matrix.
XxMatrix< T_Int, T_Scalar >	rcopy ()
	Create a reference copy (shallow copy) of the device sparse matrix.
Guard< XxMatrix< T_Int, T_Scalar > >	rcopy () const
	Create a guarded reference copy (shallow copy) of the device sparse matrix.
XxMatrix< T_Int, T_Scalar >	move ()
	Move the device sparse matrix's resources.
void	iscale (T_Scalar val)
	Scale the device sparse matrix in-place.
alias::VirtualTrans_csr< T_Int, T_Scalar >	transpose () const
	Compute the transpose.
alias::VirtualTrans_csr< T_Int, T_Scalar >	ctranspose () const
	Compute the conjugate transpose (Hermitian transpose).
alias::VirtualConj_csr< T_Int, T_Scalar >	conjugate () const
	Compute the complex conjugate.
void	iconjugate ()
	Conjugate the device sparse matrix in-place.
void	copyToHost (::cla3p::csr::XxMatrix< T_Cla3pInt, T_Cla3pScalar > &dest) const
	Copy the device sparse matrix to host memory.
void	copyFromHost (const ::cla3p::csr::XxMatrix< T_Cla3pInt, T_Cla3pScalar > &src)
	Copy a host sparse matrix to this device sparse matrix.

Creators/Generators
static Guard< XxMatrix< T_Int, T_Scalar > >	view (T_Int nr, T_Int nc, const T_Int rptr, const T_Int cidx, const T_Scalar *vals, const Property &pr=Property::General())
	Create a view of existing device CSR arrays.

Detailed Description

template<typename T_Int, typename T_Scalar>
class culite::csr::XxMatrix< T_Int, T_Scalar >

The device sparse matrix class (compressed sparse row format).

Represents a sparse matrix stored on the GPU device in CSR format.

Template Parameters

T_Int	The integer type for indexing (e.g., int32_t, int64_t).
T_Scalar	The scalar type (e.g., float, double, complex).

Constructor & Destructor Documentation

◆ XxMatrix() [1/5]

template<typename T_Int, typename T_Scalar>

culite::csr::XxMatrix< T_Int, T_Scalar >::XxMatrix ( )

Default constructor.

Creates an empty device sparse matrix with no allocated device memory.

◆ XxMatrix() [2/5]

template<typename T_Int, typename T_Scalar>

culite::csr::XxMatrix< T_Int, T_Scalar >::XxMatrix	(	T_Int	nr,
		T_Int	nc,
		T_Int	nz,
		const Property &	pr = Property::General() )

explicit

Dimension constructor.

Creates a device sparse matrix of the specified dimensions in CSR format and allocates device memory.

Parameters

[in]	nr	The number of rows.
[in]	nc	The number of columns.
[in]	nz	The number of non-zero elements.
[in]	pr	The matrix property (default: General).

◆ XxMatrix() [3/5]

template<typename T_Int, typename T_Scalar>

culite::csr::XxMatrix< T_Int, T_Scalar >::XxMatrix	(	T_Int	nr,
		T_Int	nc,
		T_Int *	rptr,
		T_Int *	cidx,
		T_Scalar *	vals,
		bool	bind,
		const Property &	pr = Property::General() )

explicit

Auxiliary constructor.

Creates a device sparse matrix using existing device memory CSR format arrays.

Parameters

[in]	nr	The number of rows.
[in]	nc	The number of columns.
[in]	rptr	Pointer to existing device row pointer array.
[in]	cidx	Pointer to existing device column index array.
[in]	vals	Pointer to existing device values array.
[in]	bind	If true, the matrix takes ownership of the device memory.
[in]	pr	The matrix property (default: General).

◆ XxMatrix() [4/5]

template<typename T_Int, typename T_Scalar>

culite::csr::XxMatrix< T_Int, T_Scalar >::XxMatrix ( const XxMatrix< T_Int, T_Scalar > & other )

Copy constructor.

Creates a new device sparse matrix by copying another device sparse matrix.

Parameters

[in] other The device sparse matrix to copy.

◆ XxMatrix() [5/5]

template<typename T_Int, typename T_Scalar>

culite::csr::XxMatrix< T_Int, T_Scalar >::XxMatrix ( XxMatrix< T_Int, T_Scalar > && other )

Move constructor.

Creates a new device sparse matrix by moving resources from another device sparse matrix.

Parameters

[in] other The device sparse matrix to move from.

◆ ~XxMatrix()

template<typename T_Int, typename T_Scalar>

culite::csr::XxMatrix< T_Int, T_Scalar >::~XxMatrix ( )

Destructor.

Destroys the device sparse matrix and releases allocated device memory.

Member Function Documentation

◆ operator=() [1/2]

template<typename T_Int, typename T_Scalar>

XxMatrix< T_Int, T_Scalar > & culite::csr::XxMatrix< T_Int, T_Scalar >::operator= ( const XxMatrix< T_Int, T_Scalar > & other )

Copy assignment operator.

Copies the contents of another device sparse matrix to this device sparse matrix.

Parameters

[in] other The device sparse matrix to copy.

Returns: Reference to this device sparse matrix.

◆ operator=() [2/2]

template<typename T_Int, typename T_Scalar>

XxMatrix< T_Int, T_Scalar > & culite::csr::XxMatrix< T_Int, T_Scalar >::operator= ( XxMatrix< T_Int, T_Scalar > && other )

Move assignment operator.

Moves resources from another device sparse matrix to this device sparse matrix.

Parameters

[in] other The device sparse matrix to move from.

Returns: Reference to this device sparse matrix.

◆ operator-()

template<typename T_Int, typename T_Scalar>

alias::VirtualScal_csr< T_Int, T_Scalar > culite::csr::XxMatrix< T_Int, T_Scalar >::operator- ( ) const

Unary negation operator.

Returns a negated copy of the device sparse matrix.

Returns: A device sparse matrix containing the negated elements.

◆ rowptr() [1/2]

template<typename T_Int, typename T_Scalar>

T_Int * culite::csr::XxMatrix< T_Int, T_Scalar >::rowptr ( )

Access the row pointer array.

Returns a pointer to the row pointer array (CSR format).

Returns: Pointer to the row pointer array.

◆ rowptr() [2/2]

template<typename T_Int, typename T_Scalar>

const T_Int * culite::csr::XxMatrix< T_Int, T_Scalar >::rowptr ( ) const

Access the row pointer array.

Returns a pointer to the row pointer array (CSR format).

Returns: Pointer to the row pointer array.

◆ colidx() [1/2]

template<typename T_Int, typename T_Scalar>

T_Int * culite::csr::XxMatrix< T_Int, T_Scalar >::colidx ( )

Access the column index array.

Returns a pointer to the column index array (CSR format).

Returns: Pointer to the column index array.

◆ colidx() [2/2]

template<typename T_Int, typename T_Scalar>

const T_Int * culite::csr::XxMatrix< T_Int, T_Scalar >::colidx ( ) const

Access the column index array.

Returns a pointer to the column index array (CSR format).

Returns: Pointer to the column index array.

◆ nnz()

template<typename T_Int, typename T_Scalar>

T_Int culite::csr::XxMatrix< T_Int, T_Scalar >::nnz ( ) const

Get the number of non-zero elements.

Returns the number of stored non-zero elements in the device sparse matrix.

Returns: The number of non-zero elements.

◆ clear()

template<typename T_Int, typename T_Scalar>

void culite::csr::XxMatrix< T_Int, T_Scalar >::clear ( )

Clear the device sparse matrix.

Releases all device memory and resets the matrix to an empty state.

◆ info()

template<typename T_Int, typename T_Scalar>

std::string culite::csr::XxMatrix< T_Int, T_Scalar >::info ( const std::string & header = "" ) const

Get information about the sparse matrix.

Returns a string containing information about the matrix's dimensions, non-zeros, and properties.

Parameters

[in] header Optional header string to prepend to the information.

Returns: A string containing matrix information.

◆ copy()

template<typename T_Int, typename T_Scalar>

XxMatrix< T_Int, T_Scalar > culite::csr::XxMatrix< T_Int, T_Scalar >::copy ( ) const

Create a deep copy of the device sparse matrix.

Creates a new device sparse matrix with its own device memory allocation and copies all elements from this matrix to the new matrix.

Returns: A new device sparse matrix containing a copy of this matrix's data.

◆ rcopy() [1/2]

template<typename T_Int, typename T_Scalar>

XxMatrix< T_Int, T_Scalar > culite::csr::XxMatrix< T_Int, T_Scalar >::rcopy ( )

Create a reference copy (shallow copy) of the device sparse matrix.

Creates a new device sparse matrix object that references the same device memory as this matrix. Changes to either matrix will affect both.

Returns: A device sparse matrix that shares device memory with this matrix.

◆ rcopy() [2/2]

template<typename T_Int, typename T_Scalar>

Guard< XxMatrix< T_Int, T_Scalar > > culite::csr::XxMatrix< T_Int, T_Scalar >::rcopy ( ) const

Create a guarded reference copy (shallow copy) of the device sparse matrix.

Creates a guarded device sparse matrix object that references the same device memory as this matrix. The guard ensures the reference is read-only.

Returns: A guarded device sparse matrix that shares device memory with this matrix.

◆ move()

template<typename T_Int, typename T_Scalar>

XxMatrix< T_Int, T_Scalar > culite::csr::XxMatrix< T_Int, T_Scalar >::move ( )

Move the device sparse matrix's resources.

Transfers ownership of the device memory to a new matrix object, leaving this matrix in an empty state.

Returns: A new device sparse matrix containing this matrix's device memory.

◆ iscale()

template<typename T_Int, typename T_Scalar>

void culite::csr::XxMatrix< T_Int, T_Scalar >::iscale ( T_Scalar val )

Scale the device sparse matrix in-place.

Multiplies all non-zero elements of the device sparse matrix by a scalar value.

Parameters

[in] val The scalar value to multiply by.

◆ transpose()

template<typename T_Int, typename T_Scalar>

alias::VirtualTrans_csr< T_Int, T_Scalar > culite::csr::XxMatrix< T_Int, T_Scalar >::transpose ( ) const

Compute the transpose.

Returns a virtual expression representing the transpose of the device sparse matrix. Rows and columns are swapped.

Returns: A virtual expression for the transposed matrix.

Warning: Explicit calculation is not supported. Evaluating the returned expression on an existing matrix will throw an exception.

◆ ctranspose()

template<typename T_Int, typename T_Scalar>

alias::VirtualTrans_csr< T_Int, T_Scalar > culite::csr::XxMatrix< T_Int, T_Scalar >::ctranspose ( ) const

Compute the conjugate transpose (Hermitian transpose).

Returns a virtual expression representing the conjugate transpose of the device sparse matrix. Rows and columns are swapped, and complex elements are conjugated. For real matrices, this is equivalent to transpose().

Returns: A virtual expression for the conjugate transposed matrix.

Warning: Explicit calculation is not supported. Evaluating the returned expression on an existing matrix will throw an exception.

◆ conjugate()

template<typename T_Int, typename T_Scalar>

alias::VirtualConj_csr< T_Int, T_Scalar > culite::csr::XxMatrix< T_Int, T_Scalar >::conjugate ( ) const

Compute the complex conjugate.

Returns a device matrix containing the complex conjugate of each non-zero element.

Returns: A device sparse matrix with conjugated elements.

◆ iconjugate()

template<typename T_Int, typename T_Scalar>

void culite::csr::XxMatrix< T_Int, T_Scalar >::iconjugate ( )

Conjugate the device sparse matrix in-place.

Replaces all non-zero elements with their complex conjugates.

◆ copyToHost()

template<typename T_Int, typename T_Scalar>

void culite::csr::XxMatrix< T_Int, T_Scalar >::copyToHost ( ::cla3p::csr::XxMatrix< T_Cla3pInt, T_Cla3pScalar > & dest ) const

Copy the device sparse matrix to host memory.

Transfers the device sparse matrix data to a host sparse matrix.

Parameters

[out] dest The host sparse matrix destination.

◆ copyFromHost()

template<typename T_Int, typename T_Scalar>

void culite::csr::XxMatrix< T_Int, T_Scalar >::copyFromHost ( const ::cla3p::csr::XxMatrix< T_Cla3pInt, T_Cla3pScalar > & src )

Copy a host sparse matrix to this device sparse matrix.

Transfers data from a host sparse matrix to this device sparse matrix.

Parameters

[in] src The host sparse matrix source.

◆ view()

template<typename T_Int, typename T_Scalar>

Guard< XxMatrix< T_Int, T_Scalar > > culite::csr::XxMatrix< T_Int, T_Scalar >::view	(	T_Int	nr,
		T_Int	nc,
		const T_Int *	rptr,
		const T_Int *	cidx,
		const T_Scalar *	vals,
		const Property &	pr = Property::General() )

static

Create a view of existing device CSR arrays.

Creates a guarded device sparse matrix that references existing device CSR format memory without taking ownership. The device memory must remain valid for the lifetime of the returned view.

Parameters

[in]	nr	The number of rows.
[in]	nc	The number of columns.
[in]	rptr	Pointer to the device row pointer array.
[in]	cidx	Pointer to the device column index array.
[in]	vals	Pointer to the device values array.
[in]	pr	The matrix property (default: General).

Returns: A guarded device sparse matrix that views the specified device CSR arrays.

Constructors

Operators

Arguments

Public Member Functions

Creators/Generators

Detailed Description

Constructor & Destructor Documentation

◆ XxMatrix() [1/5]

◆ XxMatrix() [2/5]

◆ XxMatrix() [3/5]

◆ XxMatrix() [4/5]

◆ XxMatrix() [5/5]

◆ ~XxMatrix()

Member Function Documentation

◆ operator=() [1/2]

◆ operator=() [2/2]

◆ operator-()

◆ rowptr() [1/2]

◆ rowptr() [2/2]

◆ colidx() [1/2]

◆ colidx() [2/2]

◆ nnz()

◆ clear()

◆ info()

◆ copy()

◆ rcopy() [1/2]

◆ rcopy() [2/2]

◆ move()

◆ iscale()

◆ transpose()

◆ ctranspose()

◆ conjugate()

◆ iconjugate()

◆ copyToHost()

◆ copyFromHost()

◆ view()