NumericsSparseMatrix.h¶

Go to the source code of this file

Data structures and functions for sparse matrices

Typedefs

typedef struct linalg_data_t linalg_data_t¶

typedef void (*freeNSLSP)(void *p)¶

typedef NumericsSparseOrigin NSM_t¶

Enums

enum linalg_data_id¶

Values:

enumerator SN_LINALG_UNKNOWN¶

enumerator SN_LINALG_MKL¶

enum NSM_linear_solver¶

NumericsSparseMatrix.h id for linear algebra solvers.

Values:

enumerator NSM_CSPARSE¶

enumerator NSM_MUMPS¶

enumerator NSM_UMFPACK¶

enumerator NSM_MKL_PARDISO¶

enumerator NSM_SUPERLU¶

enumerator NSM_SUPERLU_MT¶

enumerator NSM_HSL¶

enum NumericsSparseTypesNZ¶

value of nz for some matrix storage type

Values:

enumerator NSM_CS_CSC¶

enumerator NSM_CS_CSR¶

enum NumericsSparseOrigin¶

NumericsSparseMatrix.h matrix storage types.

Values:

enumerator NSM_UNKNOWN¶

enumerator NSM_TRIPLET¶

enumerator NSM_CSC¶

enumerator NSM_CSR¶

enumerator NSM_HALF_TRIPLET¶

Functions

void NSM_null(NumericsSparseMatrix *A)¶

Initialize the fields of a NumericsSparseMatrix.

Parameters: A – the sparse matrix

NumericsSparseMatrix *NSM_new(void)¶

New and empty NumericsSparseMatrix with correctly initialized fields.

Returns: a pointer on the allocated space.

NumericsSparseMatrix *NSM_triplet_eye(unsigned int size)¶

NumericsSparseMatrix *NSM_clear(NumericsSparseMatrix *A)¶

Free allocated space for a NumericsSparseMatrix.

Parameters: A – a NumericsSparseMatrix
Returns: NULL on success

void NSM_version_copy(const NumericsSparseMatrix *const A, NumericsSparseMatrix *B)¶

Copy NumericsSparseMatrix version.

Parameters

A – a NumericsSparseMatrix
B – a NumericsSparseMatrix

void NSM_copy(NumericsSparseMatrix *A, NumericsSparseMatrix *B)¶

Copy a NumericsSparseMatrix.

Parameters

A – a NumericsSparseMatrix
B – a NumericsSparseMatrix

void NSM_clear_p(void *p)¶

Free a workspace related to a LU factorization.

Parameters: p – the structure to free

double *NSM_data(NumericsSparseMatrix *A)¶

Get the data part of sparse matrix.

Parameters: A – the sparse matrix
Returns: a pointer to the data array

static inline void *NSM_linear_solver_data(NSM_linear_solver_params *p)¶

Get the LU factors for cs_lusol.

Parameters: p – the structure holding the data for the solver

static inline double *NSM_workspace(NSM_linear_solver_params *p)¶

Get the workspace for the sparse solver.

Parameters: p – the structure holding the data for the solver
Returns: the (double) workspace

size_t NSM_nnz(const CSparseMatrix *const A)¶

get the number of non-zero (nnz) in a sparse matrix

Parameters: A – the matrix
Returns: the number of non-zero elements in the matrix

CS_INT *NSM_diag_indices(NumericsMatrix *M)¶

return the set of indices corresponding to the diagonal elements of the matrix

Warning

should be better tested

Parameters: M – the matrix
Returns: the list of indices for the diagonal elements

void NSM_extract_block(NumericsMatrix *M, double *blockM, size_t pos_row, size_t pos_col, size_t block_row_size, size_t block_col_size)¶

Extract a block from a sparse matrix.

Parameters

M – matrix
blockM – dense storage for the block
pos_row – starting row for the block
pos_col – starting column for the block
block_row_size – block width
block_col_size – block height

NSM_linear_solver_params *NSM_linearSolverParams_free(NSM_linear_solver_params *p)¶

Free allocated space for NSM_linear_solver_params.

Parameters: p – a NSM_linear_solver_params
Returns: NULL on success

NSM_linear_solver_params *NSM_linearSolverParams_new(void)¶

New and empty NSM_linear_solver_params.

Returns: a pointer on the allocated space.

NSM_linear_solver_params *NSM_linearSolverParams(NumericsMatrix *A)¶

Get linear solver parameters with initialization if needed.

Parameters: A – [inout] a NumericsMatrix.
Returns: a pointer on parameters.

void NSM_fix_csc(CSparseMatrix *A)¶

Check and fix a matrix, if needed.

Parameters: A – the matrix to check, modified if necessary to have ordered indices

void NSM_sort_csc(CSparseMatrix *A)¶

unsigned NSM_origin(const NumericsSparseMatrix *M)¶

return the origin of a sparse part of a matrix

Parameters: M – the matrix
Returns: -1 if the matrix has no sparse representation, the origin otherwise

CSparseMatrix *NSM_get_origin(const NumericsSparseMatrix *M)¶

return the sparse matrix that has the original label

Parameters: M – the matrix
Returns: the sparse matrix that is at the origin, or NULL if an error occur

void NSM_write_in_file(const NumericsSparseMatrix *m, FILE *file)¶

NumericsSparseMatrix *NSM_new_from_file(FILE *file)¶

New and empty NumericsSparseMatrix with correctly initialized fields.

Returns: a pointer on the allocated space.

int NSM_to_dense(const NumericsSparseMatrix *const A, double *B)¶

version_t NSM_version(const NumericsSparseMatrix *M, NSM_t type)¶

Get current version of a type of csparse matrix.

Parameters

M – the NumericsSparseMatrix,
type – the type of sparse storage from NumericsSparseOrigin

Returns

a comparable version.

version_t NSM_max_version(const NumericsSparseMatrix *M)¶

Get the maximum of versions of csparse matrices.

Parameters: M – the NumericsSparseMatrix,
Returns: a comparable version.

void NSM_set_version(NumericsSparseMatrix *M, NSM_t type, version_t value)¶

Set the version of a NumericsSparseMatrix.

Parameters

M – the NumericsSparseMatrix,
type – the NumericsSparseOrigin of storage,
value – the new version.

void NSM_reset_versions(NumericsSparseMatrix *M)¶

Reset all versions of a NumericsSparseMatrix.

Parameters: M – the NumericsSparseMatrix.

void NSM_reset_version(NumericsSparseMatrix *M, NSM_t type)¶

Reset version of a sparse storage.

Parameters

M – the NumericsSparseMatrix,
type – the NumericsSparseOrigin of storage.

void NSM_inc_version(NumericsSparseMatrix *M, NSM_t type)¶

Increment the version of a NumericsSparseMatrix.

Parameters

M – the NumericsSparseMatrix,
type – the NumericsSparseOrigin of storage

NSM_t NSM_latest_id(const NumericsSparseMatrix *M)¶

Get the NumericsSparseOrigin with the latest version.

Parameters: M – the NumericsSparseMatrix
Returns: the NumericsSparseOrigin.

CSparseMatrix *NSM_latest(const NumericsSparseMatrix *M)¶

Get most recent CSparseMatrix.

Parameters: M – the NumericsSparseMatrix
Returns: a pointer on a CSparseMatrix.

struct linalg_data_t

#include <>

generic data struct for linear algebra operations

Public Members

int id¶

void (*free_fn)(struct linalg_data_t*)¶

struct NSM_linear_solver_params¶

#include <>

solver-specific parameters

Public Members

NSM_linear_solver solver¶

NSM_linear_solver LDLT_solver¶

void *linear_solver_data¶: solver-specific data (or workspace)

freeNSLSP solver_free_hook¶: solver-specific hook to free linear_solver_data

int *iWork¶: integer work vector array (internal)

int iWorkSize¶: size of integer work vector array

double *dWork¶

int dWorkSize¶

linalg_data_t *linalg_data¶: data for the linear algebra

struct NumericsSparseMatrix¶

#include <>

Sparse matrix representation in Numerics.

The supported format are: triplet (aka coordinate, COO), CSC (via CSparse) and CSR if MKL is used

Public Members

CSparseMatrix *triplet¶: triplet format, aka coordinate

CSparseMatrix *half_triplet¶: half triplet format for symmetric matrices

CSparseMatrix *csc¶: csc matrix

CSparseMatrix *trans_csc¶: transpose of a csc matrix (used by CSparse)

CSparseMatrix *csr¶: csr matrix, only supported with mkl

CS_INT *diag_indx¶

indices for the diagonal terms.

Very useful for the proximal perturbation

NSM_t origin¶: original format of the matrix

NSM_linear_solver_params *linearSolverParams¶: solver-specific parameters

NumericsDataVersion versions[5]¶