cuDenseAffineMatrixFunction< DataType > Class Template Reference

#include <cu_dense_affine_matrix_function.h>

Inheritance diagram for cuDenseAffineMatrixFunction< DataType >:

Collaboration diagram for cuDenseAffineMatrixFunction< DataType >:

Public Member Functions
	cuDenseAffineMatrixFunction (const DataType *A_, const FlagType A_is_row_major_, const LongIndexType num_rows_, const LongIndexType num_colums_, const int num_gpu_devices_)
	Constructor. Matrix B is assumed to be the identity matrix. More...
	cuDenseAffineMatrixFunction (const DataType *A_, const FlagType A_is_row_major_, const LongIndexType num_rows_, const LongIndexType num_columns_, const DataType *B_, const FlagType B_is_row_major_, const int num_gpu_devices_)
virtual	~cuDenseAffineMatrixFunction ()
virtual void	dot (const DataType *vector, DataType *product)
	Computes the matrix vector product: More...
virtual void	transpose_dot (const DataType *vector, DataType *product)
	Computes the matrix vector product: More...
Public Member Functions inherited from cuAffineMatrixFunction< DataType >
	cuAffineMatrixFunction ()
	Constructor. More...
virtual	~cuAffineMatrixFunction ()
	Virtual destructor. More...
void	set_parameters (DataType *t)
DataType	get_eigenvalue (const DataType *known_parameters, const DataType known_eigenvalue, const DataType *inquiry_parameters) const
	This function defines an analytic relationship between a given set of parameters and the corresponding eigenvalue of the operator. Namely, given a set of parameters and a known eigenvalue of the operator for that specific set of parameters, this function obtains the eigenvalue of the operator for an other given set of parameters. More...
Public Member Functions inherited from cuLinearOperator< DataType >
	cuLinearOperator ()
	cuLinearOperator (int num_gpu_devices_)
	Constructor with setting num_rows and num_columns. More...
virtual	~cuLinearOperator ()
cublasHandle_t	get_cublas_handle () const
	This function returns a reference to the cublasHandle_t object. The object will be created, if it is not created already. More...
Public Member Functions inherited from cLinearOperator< DataType >
	cLinearOperator ()
	Default constructor. More...
	cLinearOperator (const LongIndexType num_rows_, const LongIndexType num_columns_)
	Constructor with setting num_rows and num_columns. More...
virtual	~cLinearOperator ()
LongIndexType	get_num_rows () const
LongIndexType	get_num_columns () const
void	set_parameters (DataType *parameters_)
	Sets the scalar parameter this->parameters. Parameter is initialized to NULL. However, before calling dot or transpose_dot functions, the parameters must be set. More...
IndexType	get_num_parameters () const
FlagType	is_eigenvalue_relation_known () const
	Returns a flag that determines whether a relation between the parameters of the operator and its eigenvalue(s) is known. More...

Protected Attributes
cuDenseMatrix< DataType >	A
cuDenseMatrix< DataType >	B
Protected Attributes inherited from cuAffineMatrixFunction< DataType >
bool	B_is_identity
Protected Attributes inherited from cuLinearOperator< DataType >
int	num_gpu_devices
bool	copied_host_to_device
cublasHandle_t *	cublas_handle
cusparseHandle_t *	cusparse_handle
Protected Attributes inherited from cLinearOperator< DataType >
const LongIndexType	num_rows
const LongIndexType	num_columns
FlagType	eigenvalue_relation_known
DataType *	parameters
IndexType	num_parameters

Additional Inherited Members
Protected Member Functions inherited from cuAffineMatrixFunction< DataType >
void	_add_scaled_vector (const DataType *input_vector, const LongIndexType vector_size, const DataType scale, DataType *output_vector) const
	Performs the operation \( \boldsymbol{c} = \boldsymbol{c} + \alpha * \boldsymbol{b} \), where \( \boldsymbol{b} \) is an input vector scaled by \( \alpha \) and \( \boldsymbol{c} \) it the output vector. More...
Protected Member Functions inherited from cuLinearOperator< DataType >
int	query_gpu_devices () const
	Before any numerical computation, this method chechs if any gpu device is available on the machine, or notifies the user if nothing was found. More...
void	initialize_cublas_handle ()
	Creates a cublasHandle_t object, if not created already. More...
void	initialize_cusparse_handle ()
	Creates a cusparseHandle_t object, if not created already. More...

Detailed Description

template<typename DataType>
class cuDenseAffineMatrixFunction< DataType >

Definition at line 30 of file cu_dense_affine_matrix_function.h.

Constructor & Destructor Documentation

cuDenseAffineMatrixFunction() [1/2]

template<typename DataType >

cuDenseAffineMatrixFunction< DataType >::cuDenseAffineMatrixFunction	(	const DataType *	A_,
		const FlagType	A_is_row_major_,
		const LongIndexType	num_rows_,
		const LongIndexType	num_colums_,
		const int	num_gpu_devices_
	)

Constructor. Matrix B is assumed to be the identity matrix.

Definition at line 30 of file cu_dense_affine_matrix_function.cu.

                                   :
 
    // Base class constructor
    cLinearOperator<DataType>(num_rows_, num_columns_),
 
    // Initializer list
    A(A_, num_rows_, num_columns_, A_is_row_major_, num_gpu_devices_)
{
    // This constructor is called assuming B is identity
    this->B_is_identity = true;
 
    // When B is identity, the eigenvalues of A+tB are known for any t
    this->eigenvalue_relation_known = 1;
 
    // Set gpu device
    this->initialize_cublas_handle();
}

References cuAffineMatrixFunction< DataType >::B_is_identity, cLinearOperator< DataType >::eigenvalue_relation_known, and cuLinearOperator< DataType >::initialize_cublas_handle().

Here is the call graph for this function:

cuDenseAffineMatrixFunction() [2/2]

template<typename DataType >

cuDenseAffineMatrixFunction< DataType >::cuDenseAffineMatrixFunction	(	const DataType *	A_,
		const FlagType	A_is_row_major_,
		const LongIndexType	num_rows_,
		const LongIndexType	num_columns_,
		const DataType *	B_,
		const FlagType	B_is_row_major_,
		const int	num_gpu_devices_
	)

Definition at line 59 of file cu_dense_affine_matrix_function.cu.

                                   :
 
    // Base class constructor
    cLinearOperator<DataType>(num_rows_, num_columns_),
 
    // Initializer list
    A(A_, num_rows_, num_columns_, A_is_row_major_, num_gpu_devices_),
    B(B_, num_rows_, num_columns_, B_is_row_major_, num_gpu_devices_)
{
    // Matrix B is assumed to be non-zero. Check if it is identity or generic
    if (this->B.is_identity_matrix())
    {
        this->B_is_identity = true;
        this->eigenvalue_relation_known = 1;
    }
 
    // Set gpu device
    this->initialize_cublas_handle();
}

References cuDenseAffineMatrixFunction< DataType >::B, cuAffineMatrixFunction< DataType >::B_is_identity, cLinearOperator< DataType >::eigenvalue_relation_known, and cuLinearOperator< DataType >::initialize_cublas_handle().

Here is the call graph for this function:

~cuDenseAffineMatrixFunction()

template<typename DataType >

cuDenseAffineMatrixFunction< DataType >::~cuDenseAffineMatrixFunction

virtual

Definition at line 92 of file cu_dense_affine_matrix_function.cu.

{
}

Member Function Documentation

dot()

template<typename DataType >

void cuDenseAffineMatrixFunction< DataType >::dot ( const DataType *  vector, DataType *  product  )

virtual

Computes the matrix vector product:

\[ \boldsymbol{c} = (\mathbf{A} + t \mathbf{B}) \boldsymbol{b}. \]

Parameters

[in]	vector	The input vector :math:\\boldsymbol{b} is given by vector. If \( \mathbf{A} \) and \( \mathbf{B} \) are \( m \times n \) matrices, the length of input c vector is n.
[out]	product	The output of the product, \( \boldsymbol{c} \), is written in-place into this array. Let m be the number of rows of \( \mathbf{A} \) and \( \mathbf{B} \), then, the output vector product is 1D column array of length m.

Implements cLinearOperator< DataType >.

Definition at line 118 of file cu_dense_affine_matrix_function.cu.

{
    // Matrix A times vector
    this->A.dot(vector, product);
    LongIndexType min_vector_size;
 
    // Matrix B times vector to be added to the product
    if (this->B_is_identity)
    {
        // Check parameter is set
        ASSERT((this->parameters != NULL), "Parameter is not set.");
 
        // Find minimum of the number of rows and columns
        min_vector_size = \
            (this->num_rows < this->num_columns) ? \
            this->num_rows : this->num_columns;
 
        // Adding input vector to product
        this->_add_scaled_vector(vector, min_vector_size,
                                 this->parameters[0], product);
    }
    else
    {
        // Check parameter is set
        ASSERT((this->parameters != NULL), "Parameter is not set.");
 
        // Adding parameter times B times input vector to the product
        this->B.dot_plus(vector, this->parameters[0], product);
    }
}

References ASSERT.

transpose_dot()

template<typename DataType >

void cuDenseAffineMatrixFunction< DataType >::transpose_dot ( const DataType *  vector, DataType *  product  )

virtual

Computes the matrix vector product:

\[ \boldsymbol{c} = (\mathbf{A} + t \mathbf{B})^{\intercal} \boldsymbol{b}. \]

Parameters

[in]	vector	The input vector \( \boldsymbol{b} \) is given by vector. If \( \mathbf{A} \) and \( \mathbf{B} \) are \( m \times n \) matrices, the length of input vector is n.
[out]	product	The output of the product, \( \boldsymbol{c} \), is written in-place into this array. Let n be the number of columns of \( \mathbf{A} \) and \( \mathbf{B} \), then, the output vector product is 1D column array of length m.

Implements cLinearOperator< DataType >.

Definition at line 174 of file cu_dense_affine_matrix_function.cu.

{
    // Matrix A times vector
    this->A.transpose_dot(vector, product);
    LongIndexType min_vector_size;
 
    // Matrix B times vector to be added to the product
    if (this->B_is_identity)
    {
        // Check parameter is set
        ASSERT((this->parameters != NULL), "Parameter is not set.");
 
        // Find minimum of the number of rows and columns
        min_vector_size = \
            (this->num_rows < this->num_columns) ? \
            this->num_rows : this->num_columns;
 
        // Adding input vector to product
        this->_add_scaled_vector(vector, min_vector_size,
                                 this->parameters[0], product);
    }
    else
    {
        // Check parameter is set
        ASSERT((this->parameters != NULL), "Parameter is not set.");
 
        // Adding "parameter * B * input vector" to the product
        this->B.transpose_dot_plus(vector, this->parameters[0], product);
    }
}

References ASSERT.

Member Data Documentation

A

template<typename DataType >

cuDenseMatrix<DataType> cuDenseAffineMatrixFunction< DataType >::A

protected

Definition at line 64 of file cu_dense_affine_matrix_function.h.

B

template<typename DataType >

cuDenseMatrix<DataType> cuDenseAffineMatrixFunction< DataType >::B

protected

Definition at line 65 of file cu_dense_affine_matrix_function.h.

Referenced by cuDenseAffineMatrixFunction< DataType >::cuDenseAffineMatrixFunction().

---

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

• /home/runner/work/imate/imate/imate/_cu_linear_operator/cu_dense_affine_matrix_function.h
• /home/runner/work/imate/imate/imate/_cu_linear_operator/cu_dense_affine_matrix_function.cu