Implementation namespace for sparse approximate inverse preconditioner. More...

Data Structures
class	block_matrix
	Represents a contigious matrices on GPU. More...
class	block_vector
	Represents a contigious vector on GPU. More...
class	fspai_tag
	A tag for FSPAI. Experimental. Contains values for the algorithm. Must be passed to spai_precond constructor. More...
struct	CompareSecond
class	spai_tag
	A tag for SPAI Contains values for the algorithm. Must be passed to spai_precond constructor. More...
class	sparse_vector
	Represents sparse vector based on std::map<unsigned int, ScalarType> More...
Typedefs
typedef std::pair< unsigned int, double >	PairT
Functions
template<typename MatrixType , typename ScalarType >
void	sym_sparse_matrix_to_stl (MatrixType const &A, std::vector< std::map< unsigned int, ScalarType > > &STL_A)
template<typename MatrixType >
void	generateJ (MatrixType const &A, std::vector< std::vector< std::size_t > > &J)
template<typename ScalarType , typename MatrixType , typename VectorType >
void	fill_blocks (std::vector< std::map< unsigned int, ScalarType > > &A, std::vector< MatrixType > &blocks, std::vector< std::vector< std::size_t > > const &J, std::vector< VectorType > &Y)
template<typename MatrixType >
void	cholesky_decompose (MatrixType &A)
template<typename MatrixType , typename VectorType >
void	cholesky_solve (MatrixType const &L, VectorType &b)
template<typename MatrixType , typename VectorType1 >
void	computeL (MatrixType const &A, MatrixType &L, MatrixType &L_trans, std::vector< VectorType1 > &Y, std::vector< std::vector< std::size_t > > &J)
template<typename MatrixType >
void	computeFSPAI (MatrixType const &A, MatrixType const &PatternA, MatrixType &L, MatrixType &L_trans, fspai_tag)
template<typename T , typename InputIterator >
void	Print (std::ostream &ostr, InputIterator it_begin, InputIterator it_end)
template<typename VectorType , typename MatrixType >
void	write_to_block (VectorType &con_A_I_J, unsigned int start_ind, const std::vector< unsigned int > &I, const std::vector< unsigned int > &J, MatrixType &m)
template<typename VectorType >
void	print_continious_matrix (VectorType &con_A_I_J, std::vector< cl_uint > &blocks_ind, const std::vector< std::vector< unsigned int > > &g_I, const std::vector< std::vector< unsigned int > > &g_J)
template<typename VectorType >
void	print_continious_vector (VectorType &con_v, std::vector< cl_uint > &block_ind, const std::vector< std::vector< unsigned int > > &g_J)
void	compute_blocks_size (const std::vector< std::vector< unsigned int > > &g_I, const std::vector< std::vector< unsigned int > > &g_J, unsigned int &sz, std::vector< cl_uint > &blocks_ind, std::vector< cl_uint > &matrix_dims)
	************************************** BLOCK FUNCTIONS **********************************//
template<typename SizeType >
void	get_size (const std::vector< std::vector< SizeType > > &inds, SizeType &size)
	Computes size of particular container of index set.
template<typename SizeType >
void	init_start_inds (const std::vector< std::vector< SizeType > > &inds, std::vector< cl_uint > &start_inds)
	Initializes start indices of particular index set.
template<typename MatrixType , typename ScalarType >
void	dot_prod (const MatrixType &A, unsigned int beg_ind, ScalarType &res)
	Dot prod of particular column of martix A with it's self starting at a certain index beg_ind.
template<typename MatrixType , typename VectorType , typename ScalarType >
void	custom_inner_prod (const MatrixType &A, const VectorType &v, unsigned int col_ind, unsigned int start_ind, ScalarType &res)
	Dot prod of particular matrix column with arbitrary vector: A(:, col_ind)
template<typename MatrixType , typename VectorType >
void	copy_vector (const MatrixType &A, VectorType &v, const unsigned int beg_ind)
	Copying part of matrix column.
template<typename MatrixType , typename VectorType , typename ScalarType >
void	householder_vector (const MatrixType &A, unsigned int j, VectorType &v, ScalarType &b)
	Coputation of Householder vector, householder reflection c.f. Gene H. Golub, Charles F. Van Loan "Matrix Computations" 3rd edition p.210.
template<typename MatrixType , typename VectorType , typename ScalarType >
void	apply_householder_reflection (MatrixType &A, unsigned int iter_cnt, VectorType &v, ScalarType b)
	Inplace application of Householder vector to a matrix A.
template<typename MatrixType , typename VectorType >
void	store_householder_vector (MatrixType &A, unsigned int ind, VectorType &v)
	Storage of vector v in column(A, ind), starting from ind-1 index of a column.
template<typename MatrixType , typename VectorType >
void	single_qr (MatrixType &R, VectorType &b_v)
	Inplace QR factorization via Householder reflections c.f. Gene H. Golub, Charles F. Van Loan "Matrix Computations" 3rd edition p.224.
template<typename SizeType >
void	get_max_block_size (const std::vector< std::vector< SizeType > > &inds, SizeType max_size)
	Getting max size of rows/columns from container of index set.
template<typename MatrixType , typename VectorType , typename ScalarType >
void	custom_dot_prod (const MatrixType &A, const VectorType &v, unsigned int ind, ScalarType &res)
	Dot_prod(column(A, ind), v) starting from index ind+1.
template<typename MatrixType , typename VectorType >
void	apply_q_trans_vec (const MatrixType &R, const VectorType &b_v, VectorType &y)
	Recovery Q from matrix R and vector of betas b_v.
template<typename MatrixType , typename VectorType >
void	apply_q_trans_mat (const MatrixType &R, const VectorType &b_v, MatrixType &A)
	Multiplication of Q'*A, where Q is in implicit for lower part of R and vector of betas - b_v.
template<typename ScalarType >
void	block_qr (std::vector< std::vector< unsigned int > > &g_I, std::vector< std::vector< unsigned int > > &g_J, block_matrix &g_A_I_J_vcl, block_vector &g_bv_vcl, std::vector< cl_uint > &g_is_update)
	Inplace QR factorization via Householder reflections c.f. Gene H. Golub, Charles F. Van Loan "Matrix Computations" 3rd edition p.224 performed on GPU.
template<typename MatrixType >
void	make_rotation_matrix (MatrixType &mat, std::size_t new_size, std::size_t off_diagonal_distance=4)
template<typename MatrixType >
double	determinant (boost::numeric::ublas::matrix_expression< MatrixType > const &mat_r)
template<typename MatrixType >
void	composeNewR (const MatrixType &A, const MatrixType &R_n, MatrixType &R)
	Composition of new matrix R, that is going to be used in Least Square problem solving.
template<typename VectorType >
void	composeNewVector (const VectorType &v_n, VectorType &v)
	Composition of new vector of coefficients beta from QR factorizations(necessary for Q recovery)
template<typename SparseVectorType , typename ScalarType >
void	sparse_norm_2 (const SparseVectorType &v, ScalarType &norm)
	Computation of Euclidean norm for sparse vector.
template<typename SparseVectorType , typename ScalarType >
void	sparse_inner_prod (const SparseVectorType &v1, const SparseVectorType &v2, ScalarType &res_v)
	Dot product of two sparse vectors.
template<typename SparseVectorType , typename ScalarType >
bool	buildAugmentedIndexSet (const std::vector< SparseVectorType > &A_v_c, const SparseVectorType &res, std::vector< unsigned int > &J, std::vector< unsigned int > &J_u, const spai_tag &tag)
	Building a new set of column indices J_u, cf. Kallischko dissertation p.31.
template<typename SparseVectorType >
void	buildNewRowSet (const std::vector< SparseVectorType > &A_v_c, const std::vector< unsigned int > &I, const std::vector< unsigned int > &J_n, std::vector< unsigned int > &I_n)
	Building a new indices to current set of row indices I_n, cf. Kallischko dissertation p.32.
template<typename MatrixType >
void	QRBlockComposition (const MatrixType &A_I_J, const MatrixType &A_I_J_u, MatrixType &A_I_u_J_u)
	Composition of new block for QR factorization cf. Kallischko dissertation p.82, figure 4.7.
template<typename SparseMatrixType , typename SparseVectorType , typename DenseMatrixType , typename VectorType >
void	block_update (const SparseMatrixType &A, const std::vector< SparseVectorType > &A_v_c, std::vector< SparseVectorType > &g_res, std::vector< bool > &g_is_update, std::vector< std::vector< unsigned int > > &g_I, std::vector< std::vector< unsigned int > > &g_J, std::vector< VectorType > &g_b_v, std::vector< DenseMatrixType > &g_A_I_J, spai_tag const &tag)
	CPU-based dynamic update for SPAI preconditioner.
template<typename ScalarType >
void	block_q_multiplication (const std::vector< std::vector< unsigned int > > &g_J_u, const std::vector< std::vector< unsigned int > > &g_I, block_matrix &g_A_I_J_vcl, block_vector &g_bv_vcl, block_matrix &g_A_I_J_u_vcl, std::vector< cl_uint > &g_is_update)
	Performs multiplication Q'*A(I, \tilde J) on GPU.
template<typename SizeType >
void	assemble_qr_row_inds (const std::vector< std::vector< SizeType > > &g_I, const std::vector< std::vector< SizeType > > g_J, const std::vector< std::vector< SizeType > > &g_I_u, std::vector< std::vector< SizeType > > &g_I_q)
	Assembly of container of index row sets: I_q, row indices for new "QR block".
template<typename ScalarType >
void	assemble_qr_block (const std::vector< std::vector< unsigned int > > &g_J, const std::vector< std::vector< unsigned int > > &g_I, const std::vector< std::vector< unsigned int > > &g_J_u, const std::vector< std::vector< unsigned int > > &g_I_u, std::vector< std::vector< unsigned int > > &g_I_q, block_matrix &g_A_I_J_u_vcl, viennacl::ocl::handle< cl_mem > &matrix_dimensions, block_matrix &g_A_I_u_J_u_vcl, std::vector< cl_uint > &g_is_update, const bool is_empty_block)
	Performs assembly for new QR block.
template<typename ScalarType >
void	assemble_r (std::vector< std::vector< unsigned int > > &g_I, std::vector< std::vector< unsigned int > > &g_J, block_matrix &g_A_I_J_vcl, block_matrix &g_A_I_J_u_vcl, block_matrix &g_A_I_u_J_u_vcl, block_vector &g_bv_vcl, block_vector &g_bv_vcl_u, std::vector< cl_uint > &g_is_update)
	Performs assembly for new R matrix on GPU.
template<typename ScalarType , unsigned int MAT_ALIGNMENT, typename SparseVectorType >
void	block_update (const viennacl::compressed_matrix< ScalarType, MAT_ALIGNMENT > &A, const std::vector< SparseVectorType > &A_v_c, std::vector< cl_uint > &g_is_update, std::vector< SparseVectorType > &g_res, std::vector< std::vector< unsigned int > > &g_J, std::vector< std::vector< unsigned int > > &g_I, block_matrix &g_A_I_J_vcl, block_vector &g_bv_vcl, spai_tag const &tag)
	GPU-based block update.
template<typename SizeType >
bool	isInIndexSet (const std::vector< SizeType > &J, SizeType ind)
	Determines if element ind is in set {J}.
template<typename VectorType , typename SparseVectorType >
void	fanOutVector (const VectorType &m_in, const std::vector< unsigned int > &J, SparseVectorType &m)
	Projects solution of LS problem onto original column m.
template<typename MatrixType , typename VectorType >
void	backwardSolve (const MatrixType &R, const VectorType &y, VectorType &x)
	Solution of linear:R*x=y system by backward substitution.
template<typename VectorType , typename ScalarType >
void	projectI (const std::vector< unsigned int > &I, VectorType &y, unsigned int ind)
	Perform projection of set I on the unit-vector.
template<typename SparseVectorType >
void	buildColumnIndexSet (const SparseVectorType &v, std::vector< unsigned int > &J)
	Builds index set of projected columns for current column of preconditioner.
template<typename SparseMatrixType >
void	initPreconditioner (const SparseMatrixType &A, SparseMatrixType &M)
	Initialize preconditioner with sparcity pattern = p(A)
template<typename SparseVectorType >
void	projectRows (const std::vector< SparseVectorType > &A_v_c, const std::vector< unsigned int > &J, std::vector< unsigned int > &I)
	Row projection for matrix A(:,J) -> A(I,J), building index set of non-zero rows.
template<typename SparseVectorType >
void	print_sparse_vector (const SparseVectorType &v)
template<typename DenseMatrixType >
void	print_matrix (DenseMatrixType &m)
template<typename SparseVectorType , typename ScalarType >
void	add_sparse_vectors (const SparseVectorType &v, const ScalarType b, SparseVectorType &res_v)
	Add two sparse vectors res_v = b*v.
template<typename SparseVectorType , typename ScalarType >
void	compute_spai_residual (const std::vector< SparseVectorType > &A_v_c, const SparseVectorType &v, const unsigned int ind, SparseVectorType &res)
	Computation of residual res = A*v - e.
template<typename SparseVectorType >
void	build_index_set (const std::vector< SparseVectorType > &A_v_c, const SparseVectorType &v, std::vector< unsigned int > &J, std::vector< unsigned int > &I)
	Setting up index set of columns and rows for certain column.
template<typename SparseMatrixType , typename DenseMatrixType >
void	initProjectSubMatrix (const SparseMatrixType &A_in, const std::vector< unsigned int > &J, std::vector< unsigned int > &I, DenseMatrixType &A_out)
	Initializes a dense matrix from a sparse one.
template<typename SparseMatrixType , typename DenseMatrixType , typename SparseVectorType , typename VectorType >
void	block_set_up (const SparseMatrixType &A, const std::vector< SparseVectorType > &A_v_c, const std::vector< SparseVectorType > &M_v, std::vector< std::vector< unsigned int > > &g_I, std::vector< std::vector< unsigned int > > &g_J, std::vector< DenseMatrixType > &g_A_I_J, std::vector< VectorType > &g_b_v)
	Setting up blocks and QR factorizing them on CPU.
template<typename SparseVectorType >
void	index_set_up (const std::vector< SparseVectorType > &A_v_c, const std::vector< SparseVectorType > &M_v, std::vector< std::vector< unsigned int > > &g_J, std::vector< std::vector< unsigned int > > &g_I)
	Setting up index set of columns and rows for all columns.
template<typename ScalarType , unsigned int MAT_ALIGNMENT, typename SparseVectorType >
void	block_set_up (const viennacl::compressed_matrix< ScalarType, MAT_ALIGNMENT > &A, const std::vector< SparseVectorType > &A_v_c, const std::vector< SparseVectorType > &M_v, std::vector< cl_uint > g_is_update, std::vector< std::vector< unsigned int > > &g_I, std::vector< std::vector< unsigned int > > &g_J, block_matrix &g_A_I_J, block_vector &g_bv)
	Setting up blocks and QR factorizing them on GPU.
template<typename ScalarType , typename SparseVectorType >
void	custom_fan_out (const std::vector< ScalarType > &m_in, unsigned int start_m_ind, const std::vector< unsigned int > &J, SparseVectorType &m)
	Elicitation of sparse vector m for particular column from m_in - contigious vector for all columns.
template<typename SparseVectorType , typename ScalarType >
void	least_square_solve (std::vector< SparseVectorType > &A_v_c, std::vector< SparseVectorType > &M_v, std::vector< std::vector< unsigned int > > &g_I, std::vector< std::vector< unsigned int > > &g_J, block_matrix &g_A_I_J_vcl, block_vector &g_bv_vcl, std::vector< SparseVectorType > &g_res, std::vector< cl_uint > &g_is_update, const spai_tag &tag)
	Solution of Least square problem on GPU.
template<typename SparseVectorType , typename DenseMatrixType , typename VectorType >
void	least_square_solve (const std::vector< SparseVectorType > &A_v_c, std::vector< DenseMatrixType > &g_R, std::vector< VectorType > &g_b_v, std::vector< std::vector< unsigned int > > &g_I, std::vector< std::vector< unsigned int > > &g_J, std::vector< SparseVectorType > &g_res, std::vector< bool > &g_is_update, std::vector< SparseVectorType > &M_v, const spai_tag &tag)
	Solution of Least square problem on CPU.
template<typename VectorType >
bool	is_all_update (VectorType &parallel_is_update)
template<typename SparseMatrixType , typename SparseVectorType >
void	vectorize_column_matrix (const SparseMatrixType &M_in, std::vector< SparseVectorType > &M_v)
	Solution of Least square problem on CPU.
template<typename SparseMatrixType , typename SparseVectorType >
void	vectorize_row_matrix (const SparseMatrixType &M_in, std::vector< SparseVectorType > &M_v)
template<typename SizeType >
void	write_set_to_array (const std::vector< std::vector< SizeType > > &ind_set, std::vector< cl_uint > &a)
template<typename ScalarType , unsigned int MAT_ALIGNMENT>
void	block_assembly (const viennacl::compressed_matrix< ScalarType, MAT_ALIGNMENT > &A, const std::vector< std::vector< unsigned int > > &g_J, const std::vector< std::vector< unsigned int > > &g_I, block_matrix &g_A_I_J_vcl, std::vector< cl_uint > &g_is_update, bool &is_empty_block)
	Assembly of blocks on GPU by a gived set of row indices: g_I and column indices: g_J.
template<typename SparseMatrixType , typename SparseVectorType >
void	insert_sparse_columns (const std::vector< SparseVectorType > &M_v, SparseMatrixType &M, bool is_right)
	Insertion of vectorized matrix column into original sparse matrix.
template<typename MatrixType >
void	sparse_transpose (const MatrixType &A_in, MatrixType &A)
	Transposition of sparse matrix.
template<typename MatrixType >
void	computeSPAI (const MatrixType &A, MatrixType &M, spai_tag &tag)
	Construction of SPAI preconditioner on CPU.
template<typename ScalarType , unsigned int MAT_ALIGNMENT>
void	computeSPAI (const viennacl::compressed_matrix< ScalarType, MAT_ALIGNMENT > &A, const boost::numeric::ublas::compressed_matrix< ScalarType > &cpu_A, boost::numeric::ublas::compressed_matrix< ScalarType > &cpu_M, viennacl::compressed_matrix< ScalarType, MAT_ALIGNMENT > &M, const spai_tag &tag)
	Construction of SPAI preconditioner on GPU.

Detailed Description

Implementation namespace for sparse approximate inverse preconditioner.

Typedef Documentation

typedef std::pair<unsigned int, double> PairT

Function Documentation

void viennacl::linalg::detail::spai::add_sparse_vectors	(	const SparseVectorType &	v,
		const ScalarType	b,
		SparseVectorType &	res_v
	)

Add two sparse vectors res_v = b*v.

Parameters:

v	initial sparse vector
b	scalar
res_v	output vector

void viennacl::linalg::detail::spai::apply_householder_reflection	(	MatrixType &	A,
		unsigned int	iter_cnt,
		VectorType &	v,
		ScalarType	b
	)

Inplace application of Householder vector to a matrix A.

Parameters:

A	init matrix
iter_cnt	current iteration
v	Householder vector
b	beta

void viennacl::linalg::detail::spai::apply_q_trans_mat	(	const MatrixType &	R,
		const VectorType &	b_v,
		MatrixType &	A
	)

Multiplication of Q'*A, where Q is in implicit for lower part of R and vector of betas - b_v.

Parameters:

R	input matrix
b_v	vector of betas
A	output matrix

void viennacl::linalg::detail::spai::apply_q_trans_vec	(	const MatrixType &	R,
		const VectorType &	b_v,
		VectorType &	y
	)

Recovery Q from matrix R and vector of betas b_v.

Parameters:

R	input matrix
b_v	vector of betas
y	output vector

void viennacl::linalg::detail::spai::assemble_qr_block	(	const std::vector< std::vector< unsigned int > > &	g_J,
		const std::vector< std::vector< unsigned int > > &	g_I,
		const std::vector< std::vector< unsigned int > > &	g_J_u,
		const std::vector< std::vector< unsigned int > > &	g_I_u,
		std::vector< std::vector< unsigned int > > &	g_I_q,
		block_matrix &	g_A_I_J_u_vcl,
		viennacl::ocl::handle< cl_mem > &	matrix_dimensions,
		block_matrix &	g_A_I_u_J_u_vcl,
		std::vector< cl_uint > &	g_is_update,
		const bool	is_empty_block
	)

Performs assembly for new QR block.

Parameters:

g_J	container of column indices
g_I	container of row indices
g_J_u	container of new column indices
g_I_u	container of new row indices
g_I_q	container of row indices for new QR blocks
g_A_I_J_u_vcl	blocks of Q'*A(I, \tilde J)
matrix_dimensions	array with matrix dimensions for all blocks
g_A_I_u_J_u_vcl	blocks A(\tilde I, \tilde J)
g_is_update	container with update indicators
is_empty_block	indicator if all previous blocks A(\tilde I, \tilde J) - are empty, in case if they are empty kernel with smaller number of arguments is used

void viennacl::linalg::detail::spai::assemble_qr_row_inds	(	const std::vector< std::vector< SizeType > > &	g_I,
		const std::vector< std::vector< SizeType > >	g_J,
		const std::vector< std::vector< SizeType > > &	g_I_u,
		std::vector< std::vector< SizeType > > &	g_I_q
	)

Assembly of container of index row sets: I_q, row indices for new "QR block".

Parameters:

g_I	container of row indices
g_J	container of column indices
g_I_u	container of new row indices
g_I_q	container of row indices for new QR blocks

void viennacl::linalg::detail::spai::assemble_r	(	std::vector< std::vector< unsigned int > > &	g_I,
		std::vector< std::vector< unsigned int > > &	g_J,
		block_matrix &	g_A_I_J_vcl,
		block_matrix &	g_A_I_J_u_vcl,
		block_matrix &	g_A_I_u_J_u_vcl,
		block_vector &	g_bv_vcl,
		block_vector &	g_bv_vcl_u,
		std::vector< cl_uint > &	g_is_update
	)

Performs assembly for new R matrix on GPU.

Parameters:

g_I	container of row indices
g_J	container of column indices
g_A_I_J_vcl	container of block matrices from previous update
g_A_I_J_u_vcl	container of block matrices Q'*A(I, \tilde J)
g_A_I_u_J_u_vcl	container of block matrices QR factored on current iteration
g_bv_vcl	block of beta vectors from previous iteration
g_bv_vcl_u	block of updated beta vectors got after recent QR factorization
g_is_update	container with identificators that shows which block should be modified

void viennacl::linalg::detail::spai::backwardSolve	(	const MatrixType &	R,
		const VectorType &	y,
		VectorType &	x
	)

Solution of linear:R*x=y system by backward substitution.

Parameters:

R	uppertriangular matrix
y	right handside vector
x	solution vector

void viennacl::linalg::detail::spai::block_assembly	(	const viennacl::compressed_matrix< ScalarType, MAT_ALIGNMENT > &	A,
		const std::vector< std::vector< unsigned int > > &	g_J,
		const std::vector< std::vector< unsigned int > > &	g_I,
		block_matrix &	g_A_I_J_vcl,
		std::vector< cl_uint > &	g_is_update,
		bool &	is_empty_block
	)

Assembly of blocks on GPU by a gived set of row indices: g_I and column indices: g_J.

Parameters:

A	intial sparse matrix
g_J	container of column index set
g_I	container of row index set
g_A_I_J_vcl	contigious blocks A(I, J) using GPU memory
g_is_update	container with indicators which blocks are active
is_empty_block	parameter that indicates if no block were assembled

void viennacl::linalg::detail::spai::block_q_multiplication	(	const std::vector< std::vector< unsigned int > > &	g_J_u,
		const std::vector< std::vector< unsigned int > > &	g_I,
		block_matrix &	g_A_I_J_vcl,
		block_vector &	g_bv_vcl,
		block_matrix &	g_A_I_J_u_vcl,
		std::vector< cl_uint > &	g_is_update
	)

Performs multiplication Q'*A(I, \tilde J) on GPU.

Parameters:

g_J_u	container of sets of new column indices
g_I	container of row indices
g_A_I_J_vcl	block matrix composed from previous blocks, they are blocks of R
g_bv_vcl	block of beta vectors
g_A_I_J_u_vcl	block of matrices A(I, \tilde J)
g_is_update	indicators, that show if a certain block should be processed

void viennacl::linalg::detail::spai::block_qr	(	std::vector< std::vector< unsigned int > > &	g_I,
		std::vector< std::vector< unsigned int > > &	g_J,
		block_matrix &	g_A_I_J_vcl,
		block_vector &	g_bv_vcl,
		std::vector< cl_uint > &	g_is_update
	)

Inplace QR factorization via Householder reflections c.f. Gene H. Golub, Charles F. Van Loan "Matrix Computations" 3rd edition p.224 performed on GPU.

Parameters:

g_I	container of row indices
g_J	container of column indices
g_A_I_J_vcl	contigious matrices, GPU memory is used
g_bv_vcl	contigios vectors beta, GPU memory is used
g_is_update	container of indicators that show active blocks

void viennacl::linalg::detail::spai::block_set_up	(	const SparseMatrixType &	A,
		const std::vector< SparseVectorType > &	A_v_c,
		const std::vector< SparseVectorType > &	M_v,
		std::vector< std::vector< unsigned int > > &	g_I,
		std::vector< std::vector< unsigned int > > &	g_J,
		std::vector< DenseMatrixType > &	g_A_I_J,
		std::vector< VectorType > &	g_b_v
	)

Setting up blocks and QR factorizing them on CPU.

Parameters:

A	initial sparse matrix
A_v_c	column major vectorized initial sparse matrix
M_v	initialized preconditioner
g_I	container of row indices
g_J	container of column indices
g_A_I_J	container of dense matrices -> R matrices after QR factorization
g_b_v	container of vectors beta, necessary for Q recovery

void viennacl::linalg::detail::spai::block_set_up	(	const viennacl::compressed_matrix< ScalarType, MAT_ALIGNMENT > &	A,
		const std::vector< SparseVectorType > &	A_v_c,
		const std::vector< SparseVectorType > &	M_v,
		std::vector< cl_uint >	g_is_update,
		std::vector< std::vector< unsigned int > > &	g_I,
		std::vector< std::vector< unsigned int > > &	g_J,
		block_matrix &	g_A_I_J,
		block_vector &	g_bv
	)

Setting up blocks and QR factorizing them on GPU.

Parameters:

A	initial sparse matrix
A_v_c	column major vectorized initial sparse matrix
M_v	initialized preconditioner
g_is_update	container that indicates which blocks are active
g_I	container of row indices
g_J	container of column indices
g_A_I_J	container of dense matrices -> R matrices after QR factorization
g_bv	container of vectors beta, necessary for Q recovery

void viennacl::linalg::detail::spai::block_update	(	const SparseMatrixType &	A,
		const std::vector< SparseVectorType > &	A_v_c,
		std::vector< SparseVectorType > &	g_res,
		std::vector< bool > &	g_is_update,
		std::vector< std::vector< unsigned int > > &	g_I,
		std::vector< std::vector< unsigned int > > &	g_J,
		std::vector< VectorType > &	g_b_v,
		std::vector< DenseMatrixType > &	g_A_I_J,
		spai_tag const &	tag
	)

CPU-based dynamic update for SPAI preconditioner.

Parameters:

A	initial sparse matrix
A_v_c	vectorized column-wise initial matrix
g_res	container of residuals for all columns
g_is_update	container with identificators that shows which block should be modified
g_I	container of row index sets for all columns
g_J	container of column index sets for all columns
g_b_v	container of vectors of beta for Q recovery(cf. Golub Van Loan "Matrix Computations", 3rd edition p.211)
g_A_I_J	container of block matrices from previous update
tag	SPAI configuration tag

void viennacl::linalg::detail::spai::block_update	(	const viennacl::compressed_matrix< ScalarType, MAT_ALIGNMENT > &	A,
		const std::vector< SparseVectorType > &	A_v_c,
		std::vector< cl_uint > &	g_is_update,
		std::vector< SparseVectorType > &	g_res,
		std::vector< std::vector< unsigned int > > &	g_J,
		std::vector< std::vector< unsigned int > > &	g_I,
		block_matrix &	g_A_I_J_vcl,
		block_vector &	g_bv_vcl,
		spai_tag const &	tag
	)

GPU-based block update.

Parameters:

A	sparse matrix
A_v_c	vectorized column-wise initial matrix
g_is_update	container with identificators that shows which block should be modified
g_res	container of residuals for all columns
g_J	container of column index sets for all columns
g_I	container of row index sets for all columns
g_A_I_J_vcl	container of block matrices from previous update
g_bv_vcl	block of beta vectors from previous iteration
tag	SPAI configuration tag

void viennacl::linalg::detail::spai::build_index_set	(	const std::vector< SparseVectorType > &	A_v_c,
		const SparseVectorType &	v,
		std::vector< unsigned int > &	J,
		std::vector< unsigned int > &	I
	)

Setting up index set of columns and rows for certain column.

Parameters:

A_v_c	column major vectorized initial sparse matrix
v	current column of preconditioner matrix
J	set of column indices
I	set of row indices

bool viennacl::linalg::detail::spai::buildAugmentedIndexSet	(	const std::vector< SparseVectorType > &	A_v_c,
		const SparseVectorType &	res,
		std::vector< unsigned int > &	J,
		std::vector< unsigned int > &	J_u,
		const spai_tag &	tag
	)

Building a new set of column indices J_u, cf. Kallischko dissertation p.31.

Parameters:

A_v_c	vectorized column-wise initial matrix
res	residual vector
J	set of column indices
J_u	set of new column indices
tag	SPAI tag with parameters

void viennacl::linalg::detail::spai::buildColumnIndexSet	(	const SparseVectorType &	v,
		std::vector< unsigned int > &	J
	)

Builds index set of projected columns for current column of preconditioner.

Parameters:

v	current column of preconditioner
J	output - index set of non-zero columns

void viennacl::linalg::detail::spai::buildNewRowSet	(	const std::vector< SparseVectorType > &	A_v_c,
		const std::vector< unsigned int > &	I,
		const std::vector< unsigned int > &	J_n,
		std::vector< unsigned int > &	I_n
	)

Building a new indices to current set of row indices I_n, cf. Kallischko dissertation p.32.

Parameters:

A_v_c	vectorized column-wise initial matrix
I	set of previous determined row indices
J_n	set of new column indices
I_n	set of new indices

void viennacl::linalg::detail::spai::cholesky_decompose ( MatrixType & A )

void viennacl::linalg::detail::spai::cholesky_solve	(	MatrixType const &	L,
		VectorType &	b
	)

void viennacl::linalg::detail::spai::composeNewR	(	const MatrixType &	A,
		const MatrixType &	R_n,
		MatrixType &	R
	)

Composition of new matrix R, that is going to be used in Least Square problem solving.

Parameters:

A	matrix Q'*A(I, \tilde J), where \tilde J - set of new column indices
R_n	matrix A_Iu_J_u after QR factorization
R	previously composed matrix R

void viennacl::linalg::detail::spai::composeNewVector	(	const VectorType &	v_n,
		VectorType &	v
	)

Composition of new vector of coefficients beta from QR factorizations(necessary for Q recovery)

Parameters:

v_n	new vector from last QR factorization
v	composition of previous vectors from QR factorizations

void viennacl::linalg::detail::spai::compute_blocks_size	(	const std::vector< std::vector< unsigned int > > &	g_I,
		const std::vector< std::vector< unsigned int > > &	g_J,
		unsigned int &	sz,
		std::vector< cl_uint > &	blocks_ind,
		std::vector< cl_uint > &	matrix_dims
	)		`[inline]`

**************************************** BLOCK FUNCTIONS ************************************//

Computes size of elements, start indices and matrix dimensions for a certain block

Parameters:

g_I	container of row indices
g_J	container of column indices
sz	general size for all elements in a certain block
blocks_ind	start indices in a certain
matrix_dims	matrix dimensions for each block

void viennacl::linalg::detail::spai::compute_spai_residual	(	const std::vector< SparseVectorType > &	A_v_c,
		const SparseVectorType &	v,
		const unsigned int	ind,
		SparseVectorType &	res
	)

Computation of residual res = A*v - e.

Parameters:

A_v_c	column major vectorized input sparse matrix
v	sparse vector, in this case new column of preconditioner matrix
ind	index for current column
res	residual

void viennacl::linalg::detail::spai::computeFSPAI	(	MatrixType const &	A,
		MatrixType const &	PatternA,
		MatrixType &	L,
		MatrixType &	L_trans,
		fspai_tag
	)

void viennacl::linalg::detail::spai::computeL	(	MatrixType const &	A,
		MatrixType &	L,
		MatrixType &	L_trans,
		std::vector< VectorType1 > &	Y,
		std::vector< std::vector< std::size_t > > &	J
	)

void viennacl::linalg::detail::spai::computeSPAI	(	const MatrixType &	A,
		MatrixType &	M,
		spai_tag &	tag
	)

Construction of SPAI preconditioner on CPU.

Parameters:

A	initial sparse matrix
M	output preconditioner
tag	spai tag

void viennacl::linalg::detail::spai::computeSPAI	(	const viennacl::compressed_matrix< ScalarType, MAT_ALIGNMENT > &	A,
		const boost::numeric::ublas::compressed_matrix< ScalarType > &	cpu_A,
		boost::numeric::ublas::compressed_matrix< ScalarType > &	cpu_M,
		viennacl::compressed_matrix< ScalarType, MAT_ALIGNMENT > &	M,
		const spai_tag &	tag
	)

Construction of SPAI preconditioner on GPU.

Parameters:

A	initial sparse matrix
cpu_A	copy of initial matrix on CPU
cpu_M	output preconditioner on CPU
M	output preconditioner
tag	SPAI tag class with parameters

void viennacl::linalg::detail::spai::copy_vector	(	const MatrixType &	A,
		VectorType &	v,
		const unsigned int	beg_ind
	)

Copying part of matrix column.

Parameters:

A	init matrix
v	output vector
beg_ind	start index for copying

void viennacl::linalg::detail::spai::custom_dot_prod	(	const MatrixType &	A,
		const VectorType &	v,
		unsigned int	ind,
		ScalarType &	res
	)

Dot_prod(column(A, ind), v) starting from index ind+1.

Parameters:

A	input matrix
v	input vector
ind	index
res	result value

void viennacl::linalg::detail::spai::custom_fan_out	(	const std::vector< ScalarType > &	m_in,
		unsigned int	start_m_ind,
		const std::vector< unsigned int > &	J,
		SparseVectorType &	m
	)

Elicitation of sparse vector m for particular column from m_in - contigious vector for all columns.

Parameters:

m_in	contigious sparse vector for all columns
start_m_ind	start index of particular vector
J	column index set
m	sparse vector for particular column

void viennacl::linalg::detail::spai::custom_inner_prod	(	const MatrixType &	A,
		const VectorType &	v,
		unsigned int	col_ind,
		unsigned int	start_ind,
		ScalarType &	res
	)

Dot prod of particular matrix column with arbitrary vector: A(:, col_ind)

Parameters:

A	init matrix
v	input vector
col_ind	starting column index
start_ind	starting index inside column
res	result of dot product

double viennacl::linalg::detail::spai::determinant ( boost::numeric::ublas::matrix_expression< MatrixType > const & mat_r )

void viennacl::linalg::detail::spai::dot_prod	(	const MatrixType &	A,
		unsigned int	beg_ind,
		ScalarType &	res
	)

Dot prod of particular column of martix A with it's self starting at a certain index beg_ind.

Parameters:

A	init matrix
beg_ind	starting index
res	result of dot product

void viennacl::linalg::detail::spai::fanOutVector	(	const VectorType &	m_in,
		const std::vector< unsigned int > &	J,
		SparseVectorType &	m
	)

Projects solution of LS problem onto original column m.

Parameters:

m_in	solution of LS
J	set of non-zero columns
m	original column of M

void viennacl::linalg::detail::spai::fill_blocks	(	std::vector< std::map< unsigned int, ScalarType > > &	A,
		std::vector< MatrixType > &	blocks,
		std::vector< std::vector< std::size_t > > const &	J,
		std::vector< VectorType > &	Y
	)

void viennacl::linalg::detail::spai::generateJ	(	MatrixType const &	A,
		std::vector< std::vector< std::size_t > > &	J
	)

void viennacl::linalg::detail::spai::get_max_block_size	(	const std::vector< std::vector< SizeType > > &	inds,
		SizeType	max_size
	)

Getting max size of rows/columns from container of index set.

Parameters:

inds	container of index set
max_size	max size that corresponds to that container

void viennacl::linalg::detail::spai::get_size	(	const std::vector< std::vector< SizeType > > &	inds,
		SizeType &	size
	)

Computes size of particular container of index set.

Parameters:

inds	container of index sets
size	output size

void viennacl::linalg::detail::spai::householder_vector	(	const MatrixType &	A,
		unsigned int	j,
		VectorType &	v,
		ScalarType &	b
	)

Coputation of Householder vector, householder reflection c.f. Gene H. Golub, Charles F. Van Loan "Matrix Computations" 3rd edition p.210.

Parameters:

A	init matrix
j	start index for computations
v	output Householder vector
b	beta

void viennacl::linalg::detail::spai::index_set_up	(	const std::vector< SparseVectorType > &	A_v_c,
		const std::vector< SparseVectorType > &	M_v,
		std::vector< std::vector< unsigned int > > &	g_J,
		std::vector< std::vector< unsigned int > > &	g_I
	)

Setting up index set of columns and rows for all columns.

Parameters:

A_v_c	column major vectorized initial sparse matrix
M_v	initialized preconditioner
g_J	container of column indices
g_I	container of row indices

void viennacl::linalg::detail::spai::init_start_inds	(	const std::vector< std::vector< SizeType > > &	inds,
		std::vector< cl_uint > &	start_inds
	)

Initializes start indices of particular index set.

Parameters:

inds	container of index sets
start_inds	output index set

void viennacl::linalg::detail::spai::initPreconditioner	(	const SparseMatrixType &	A,
		SparseMatrixType &	M
	)

Initialize preconditioner with sparcity pattern = p(A)

Parameters:

A	input matrix
M	output matrix - initialized preconditioner

void viennacl::linalg::detail::spai::initProjectSubMatrix	(	const SparseMatrixType &	A_in,
		const std::vector< unsigned int > &	J,
		std::vector< unsigned int > &	I,
		DenseMatrixType &	A_out
	)

Initializes a dense matrix from a sparse one.

Parameters:

A_in	Riginal sparse matrix
J	Set of column indices
I	Set of row indices
A_out	dense matrix output

void viennacl::linalg::detail::spai::insert_sparse_columns	(	const std::vector< SparseVectorType > &	M_v,
		SparseMatrixType &	M,
		bool	is_right
	)

Insertion of vectorized matrix column into original sparse matrix.

Parameters:

M_v	column-major vectorized matrix
M	original sparse matrix
is_right	indicates if matrix should be transposed in the output

bool viennacl::linalg::detail::spai::is_all_update ( VectorType & parallel_is_update )

bool viennacl::linalg::detail::spai::isInIndexSet	(	const std::vector< SizeType > &	J,
		SizeType	ind
	)

Determines if element ind is in set {J}.

Parameters:

J	current set
ind	current element

void viennacl::linalg::detail::spai::least_square_solve	(	std::vector< SparseVectorType > &	A_v_c,
		std::vector< SparseVectorType > &	M_v,
		std::vector< std::vector< unsigned int > > &	g_I,
		std::vector< std::vector< unsigned int > > &	g_J,
		block_matrix &	g_A_I_J_vcl,
		block_vector &	g_bv_vcl,
		std::vector< SparseVectorType > &	g_res,
		std::vector< cl_uint > &	g_is_update,
		const spai_tag &	tag
	)

Solution of Least square problem on GPU.

Parameters:

A_v_c	column-major vectorized initial sparse matrix
M_v	column-major vectorized sparse preconditioner matrix
g_I	container of row set indices
g_J	container of column set indices
g_A_I_J_vcl	contigious matrix that consists of blocks A(I_k, J_k)
g_bv_vcl	contigious vector that consists of betas, necessary for Q recovery
g_res	container of residuals
g_is_update	container with indicators which blocks are active
tag	spai tag

void viennacl::linalg::detail::spai::least_square_solve	(	const std::vector< SparseVectorType > &	A_v_c,
		std::vector< DenseMatrixType > &	g_R,
		std::vector< VectorType > &	g_b_v,
		std::vector< std::vector< unsigned int > > &	g_I,
		std::vector< std::vector< unsigned int > > &	g_J,
		std::vector< SparseVectorType > &	g_res,
		std::vector< bool > &	g_is_update,
		std::vector< SparseVectorType > &	M_v,
		const spai_tag &	tag
	)

Solution of Least square problem on CPU.

Parameters:

A_v_c	column-major vectorized initial sparse matrix
g_R	blocks for least square solution
g_b_v	vectors beta, necessary for Q recovery
g_I	container of row index set for all columns of matrix M
g_J	container of column index set for all columns of matrix M
g_res	container of residuals
g_is_update	container with indicators which blocks are active
M_v	column-major vectorized sparse matrix, final preconditioner
tag	spai tag

void viennacl::linalg::detail::spai::make_rotation_matrix	(	MatrixType &	mat,
		std::size_t	new_size,
		std::size_t	off_diagonal_distance = `4`
	)

void viennacl::linalg::detail::spai::Print	(	std::ostream &	ostr,
		InputIterator	it_begin,
		InputIterator	it_end
	)

void viennacl::linalg::detail::spai::print_continious_matrix	(	VectorType &	con_A_I_J,
		std::vector< cl_uint > &	blocks_ind,
		const std::vector< std::vector< unsigned int > > &	g_I,
		const std::vector< std::vector< unsigned int > > &	g_J
	)

void viennacl::linalg::detail::spai::print_continious_vector	(	VectorType &	con_v,
		std::vector< cl_uint > &	block_ind,
		const std::vector< std::vector< unsigned int > > &	g_J
	)

void viennacl::linalg::detail::spai::print_matrix ( DenseMatrixType & m )

void viennacl::linalg::detail::spai::print_sparse_vector ( const SparseVectorType & v )

void viennacl::linalg::detail::spai::projectI	(	const std::vector< unsigned int > &	I,
		VectorType &	y,
		unsigned int	ind
	)

Perform projection of set I on the unit-vector.

Parameters:

I	set of non-zero rows
y	result vector
ind	index of unit vector

void viennacl::linalg::detail::spai::projectRows	(	const std::vector< SparseVectorType > &	A_v_c,
		const std::vector< unsigned int > &	J,
		std::vector< unsigned int > &	I
	)

Row projection for matrix A(:,J) -> A(I,J), building index set of non-zero rows.

Parameters:

A_v_c	input matrix
J	set of non-zero rows
I	output matrix

void viennacl::linalg::detail::spai::QRBlockComposition	(	const MatrixType &	A_I_J,
		const MatrixType &	A_I_J_u,
		MatrixType &	A_I_u_J_u
	)

Composition of new block for QR factorization cf. Kallischko dissertation p.82, figure 4.7.

Parameters:

A_I_J	previously composed block
A_I_J_u	matrix Q'*A(I, \tilde J), where \tilde J - set of new column indices
A_I_u_J_u	is composition of lower part A(I, \tilde J) and A(\tilde I, \tilde J) - new block for QR decomposition

void viennacl::linalg::detail::spai::single_qr	(	MatrixType &	R,
		VectorType &	b_v
	)

Inplace QR factorization via Householder reflections c.f. Gene H. Golub, Charles F. Van Loan "Matrix Computations" 3rd edition p.224.

Parameters:

R	input matrix
b_v	vector of betas

void viennacl::linalg::detail::spai::sparse_inner_prod	(	const SparseVectorType &	v1,
		const SparseVectorType &	v2,
		ScalarType &	res_v
	)

Dot product of two sparse vectors.

Parameters:

v1	initial sparse vector
v2	initial sparse vector
res_v	scalar that represents dot product result

void viennacl::linalg::detail::spai::sparse_norm_2	(	const SparseVectorType &	v,
		ScalarType &	norm
	)

Computation of Euclidean norm for sparse vector.

Parameters:

v	initial sparse vector
norm	scalar that represents Euclidean norm

void viennacl::linalg::detail::spai::sparse_transpose	(	const MatrixType &	A_in,
		MatrixType &	A
	)

Transposition of sparse matrix.

Parameters:

A_in	intial sparse matrix
A	output transposed matrix

void viennacl::linalg::detail::spai::store_householder_vector	(	MatrixType &	A,
		unsigned int	ind,
		VectorType &	v
	)

Storage of vector v in column(A, ind), starting from ind-1 index of a column.

Parameters:

A	init matrix
ind	index of a column
v	vector that should be stored

void viennacl::linalg::detail::spai::sym_sparse_matrix_to_stl	(	MatrixType const &	A,
		std::vector< std::map< unsigned int, ScalarType > > &	STL_A
	)

void viennacl::linalg::detail::spai::vectorize_column_matrix	(	const SparseMatrixType &	M_in,
		std::vector< SparseVectorType > &	M_v
	)

Solution of Least square problem on CPU.

Parameters:

M_in	input sparse, boost::numeric::ublas::compressed_matrix
M_v	array of sparse vectors

void viennacl::linalg::detail::spai::vectorize_row_matrix	(	const SparseMatrixType &	M_in,
		std::vector< SparseVectorType > &	M_v
	)

void viennacl::linalg::detail::spai::write_set_to_array	(	const std::vector< std::vector< SizeType > > &	ind_set,
		std::vector< cl_uint > &	a
	)

void viennacl::linalg::detail::spai::write_to_block	(	VectorType &	con_A_I_J,
		unsigned int	start_ind,
		const std::vector< unsigned int > &	I,
		const std::vector< unsigned int > &	J,
		MatrixType &	m
	)

Data Structures

Typedefs

Functions

Detailed Description

Typedef Documentation

Function Documentation