Class for manipulating matrices, and interfacing SparseChol. More...

#include <MatrixBlockIP.h>

Collaboration diagram for MatrixBlockIP:

Classes
struct	Order_ija
	Auxiliary struct for sorting matrices in ija format. More...

struct	Order_vector
	Auxiliary struct for sorting vectors. More...

Public Member Functions
	MatrixBlockIP (int blocks=1)
	Constructor.

	MatrixBlockIP (MatrixBlockIP *mbip)
	Copy constructor, does not copy SparseChol.

	~MatrixBlockIP ()
	Destructor.

void	copy (MatrixBlockIP *mbip)
	Copy, does not copy SparseChol.

void	reset ()
	Comes back to the initial state.

void	restore ()
	Comes back to the state after create the matrix.

void	create_general_matrix_row_wise (int m, int n, int nz, int &inirowa, int &icola, double *&a)
	Native method to create a general row-wise packed matrix.

void	create_general_matrix_column_wise (int m, int n, int nz, int &inicola, int &irowa, double *&a)
	Native method to create a general column-wise packed matrix.

void	create_network_matrix (int num_arcs, int num_nodes, int &src, int &dst, bool oriented=true)
	Native method to create a network matrix.

void	create_identity_matrix (int dim)
	Native method to create a identity matrix of dimension dim.

void	create_idty_idty_matrix (int nrows)
	Native method to create a identity-identity matrix with two identities [I I].

void	create_diagonal_matrix (int dim, double *&d)
	Native method to create a diagonal matrix D= diag(d) of dimension dim.

void	create_diag_diag_matrix (int nrows, double &d1, double &d2)
	Native method to create a diagonal-diagonal matrix D= [D1 D2].

void	create_general_matrix_format_ija (int m, int n, int nz, int &row_index, int &col_index, double *&values)
	Creates a general matrix in format ija.

void	compute_full_matrix (int numBlocks, bool sameN, MatrixBlockIP N[], bool sameL, MatrixBlockIP L[], int numActiveLnk=0, int *listActiveLnk=NULL)
	Builds a packed rowwise format structured matrix based on diagonal blocks and linking constraints blocks.

void	analyze_D (int numBlocks, bool sameL, MatrixBlockIP L[])

void	analyze_D_diagonal (int numBlocks, bool sameL, MatrixBlockIP L[])

void	analyze_D_gen_sym_uptr (int numBlocks, bool sameL, MatrixBlockIP L[])

void	compute_D (int numBlocks, bool sameL, MatrixBlockIP L[], bool isActive[], int iniTheta[], double Theta[])
	Compute D = Theta_(numBlocks+1) + sum{i..numBlocks} L_iTheta_iL_i'.

void	compute_D_diagonal (int numBlocks, bool sameL, MatrixBlockIP L[], bool isActive[], int iniTheta[], double Theta[])
	Compute D = Theta_(numBlocks+1) + sum{i in 1..numBlocks} L_iTheta_iL_i' when D is diagonal.

void	compute_D_gen_sym_uptr (int numBlocks, bool sameL, MatrixBlockIP L[], bool isActive[], int iniTheta[], double Theta[])
	Compute D = Theta_(numBlocks+1) + sum{i..numBlocks} L_iTheta_iL_i' when D is a symmetric general matrix.

void	native_to_general (bool delete_native_format=false)
	Calculates and stores the matrix in packed rowwise format.

void	ija_to_rowwise ()
	Calculates and stores the matrix in packed rowwise format.

void	network_to_general ()
	Calculates and stores the network matrix (either oriented or nonoriented) in packed rowwise format.

void	identity_to_general ()
	Calculates and stores the I matrix in packed rowwise format.

void	diagonal_to_general ()
	Calculates and stores the diagonal D matrix in packed rowwise format.

void	idty_idty_to_general ()
	Calculates and stores the matrix [I I] in packed rowwise format.

void	diag_diag_to_general ()
	Calculates and stores the matrix [D1 D2] in packed rowwise format.

void	network_to_ija_format ()
	Calculates and stores the network matrix in ija format. It considers both oriented and nonoriented cases.

void	identity_to_ija_format ()
	Calculates and stores the identity I matrix in ija format.

void	diagonal_to_ija_format ()
	Calculates and stores the diagonal D1 matrix in ija format.

void	idty_idty_to_ija_format ()
	Calculates and stores the matrix [I I] in ija format.

void	diag_diag_to_ija_format ()
	Calculates and stores the matrix [D1 D2] in ija format.

void	order_matrix ()
	Order the matrix when has been created in row-wise or column-wise format.

void	mul_Mv (double vout[], const double vin[])
	Matrix-vector product vout=M*vin (vout(m),vin(n)) (driver)

void	mul_Mtv (double vout[], const double vin[])
	MatrixTranspose-vector product vout=M(t)*vin (vout(n),vin(m) (driver)

void	add_mul_Mv (double vout[], const double vin[])
	Add matrix-vector product vout += M*vin (vout(m),vin(n)) (driver)

void	add_mul_Mtv (double vout[], const double vin[])
	Add matrixTranspose-vector product vout += M(t)*vin (vout(n),vin(m) (driver)

void	exist_var_in_row (int column, bool appear[], double coefs[])
	Given a column determines if exists an element for each row of the matrix, in case of exist returns the value.

void	add_new_column (int size, int irows[], double values[])
	Add a new column into the matrix.

void	add_new_columns (int num_columns, int size[], int irows[], double values[])
	Add new columns into the matrix.

void	change_columns_sign (int size, int columns[])
	Changes the sign of some columns.

void	change_rows_sign (int size, int rows[])
	Change the sign of some rows.

void	delete_rows (int size, int rows[])
	Delete some rows.

int	get_column (int column, int &irows, double &values, bool invert_sign=false)
	Gets a column of the matrix.

void	symbolic_fact_MMt (CHOL_SOLVER chslv=SPRSBLKLLT, int prov_pfa=NULL, int prov_ipfa=NULL, NUMBERING prov_pfa_numbering=NOT_COMPUTED)
	Interface routine to SparseChol symbolic factorization.

void	numeric_fact_MMt (double *Theta, int i_k=0)
	Interface routine to SparseChol numeric factorization.

void	numeric_solve_MMt (double *rhs, int i_k=0, WHO_PERMUTES whoperm=CHOLESKY)
	Interface routine to SparseChol numeric solve.

void	symbolic_fact_M (CHOL_SOLVER chslv=SPRSBLKLLT, int prov_pfa=NULL, int prov_ipfa=NULL, NUMBERING prov_pfa_numbering=NOT_COMPUTED)
	Interface routine to SparseChol symbolic factorization.

void	numeric_fact_M ()
	Interface routine to SparseChol numeric factorization.

void	numeric_solve_M (double *rhs, WHO_PERMUTES whoperm=CHOLESKY)
	Interface routine to SparseChol numeric solve.

int	get_pfa (int i)
	Interface routine to SparseChol get_pfa().

int	get_ipfa (int i)
	Interface routine to SparseChol get_ipfa().

int	get_maxlnz ()
	Interface routine to SparseChol get_maxlnz().

int	get_maxfillin ()
	Interface routine to SparseChol get_maxfillin().

int	get_njka ()
	Interface routine to SparseChol get_njka().

int	get_num_zero_pivots ()
	Interface routine to SparseChol get_num_zero_pivots().

int	get_num_semidef_matrix ()
	Interface routine to SparseChol get_semidef_matrix().

void	print_matrix (bool ija=true, bool start_one=true)
	Print the matrix.

void	print_matrix (ofstream &outfile, bool print_ija=true, bool start_one=true)
	Write the matrix into a file.

void	print_vector (double *v, int size, string name)
	Print some positions and name of a vector.

void	column_wise_to_row_wise_format ()
	Convert a matrix in column-wise format to row-wise format.

void	row_wise_to_column_wise_format ()
	Convert a matrix in column-wise format to row-wise format.

Public Attributes
TYPE_MATRIX	type
	Type of matrix.

int	m
	Number of rows.

int	n
	Number of columns.

bool	ija
	True if irowa, icola and a are stored.

bool	rowwise
	True if inirowa, icola and a are stored.

bool	columnwise
	True if the matrix is stored in column-wise packed format, incompatible with rowwise.

int	nz
	Number of non-zero elements.

double *	a
	Value of non-zero elements.

int *	inirowa
	Index to the first element of each row.

int *	icola
	Column position for each element.

int *	inicola
	Index to the first element of each column.

int *	irowa
	Row position for each element.

TYPE_ORIENTATION	type_orientation
	Type of orientation, by default, oriented.

int	num_arcs
	Number of arcs.

int	num_nodes
	Number of nodes.

int *	src
	Source of each arc.

int *	dst
	Destination of each arc.

double *	d1
	d1 for DIAGONAL and 1st submatrix of DIAG_DIAG

double *	d2
	d2 for 2nd submatrix of DIAG_DIAG

SparseChol	chol
	Sparse Cholesky class.

int	sizeL

bool	made_symbfct_MMt

bool	made_analyze_D

int	num_blocks

Private Member Functions
void	free_memory ()
	Free all the memory allocated by the application - not the user.

void	mul_Mv_row_wise (double vout[], const double vin[])
	Matrix-vector product vout=M*vin (vout(m),vin(n)) (general rowwise matrix)

void	mul_Mtv_row_wise (double vout[], const double vin[])
	MatrixTranspose-vector product vout=M(t)*vin (vout(n),vin(m) (general rowwise matrix)

void	add_mul_Mv_row_wise (double vout[], const double vin[])
	Add matrix-vector product vout += M*vin (vout(m),vin(n)) (general rowwise matrix)

void	add_mul_Mtv_row_wise (double vout[], const double vin[])
	Add matrixTranspose-vector product vout += M(t)*vin (vout(n),vin(m) (general rowwise matrix)

void	mul_Mv_column_wise (double vout[], const double vin[])
	Matrix-vector product vout=M*vin (vout(m),vin(n)) (general columnwise matrix)

void	mul_Mtv_column_wise (double vout[], const double vin[])
	MatrixTranspose-vector product vout=M(t)*vin (vout(n),vin(m) (general columnwise matrix)

void	add_mul_Mv_column_wise (double vout[], const double vin[])
	Add matrix-vector product vout += M*vin (vout(m),vin(n)) (general columnwise matrix)

void	add_mul_Mtv_column_wise (double vout[], const double vin[])
	Add matrixTranspose-vector product vout += M(t)*vin (vout(n),vin(m) (general columnwise matrix)

void	mul_Mv_network (double vout[], const double vin[])
	Matrix-vector product vout=M*vin (vout(m),vin(n)) (network matrix, either oriented or nonoriented)

void	mul_Mtv_network (double vout[], const double vin[])
	MatrixTranspose-vector product vout=M(t)*vin (vout(n),vin(m) (network matrix, either oriented or nonoriented)

void	add_mul_Mv_network (double vout[], const double vin[])
	Add matrix-vector product vout += M*vin (vout(m),vin(n)) (network matrix, either oriented or nonoriented)

void	add_mul_Mtv_network (double vout[], const double vin[])
	Add matrixTranspose-vector product vout += M(t)*vin (vout(n),vin(m) (network matrix, either oriented or nonoriented)

void	mul_Mv_identity (double vout[], const double vin[])
	Identity-vector product vout=M*vin (vout(m),vin(n))

void	add_mul_Mv_identity (double vout[], const double vin[])
	Add Identity-vector product vout += M*vin (vout(m),vin(n))

void	mul_Mv_diagonal (double vout[], const double vin[])
	Diagonal-vector product vout=D*vin (vout(m),vin(n)) D is diagonal, m=n.

void	add_mul_Mv_diagonal (double vout[], const double vin[])
	Add Diagonal-vector product vout += D*vin (vout(m),vin(n)) D is diagonal, m=n.

void	mul_Mv_idty_idty (double vout[], const double vin[])
	Matrix-vector product vout= [I I]vin (vout(m),vin(n)) (M= [I I] matrix, n= 2m)

void	mul_Mtv_idty_idty (double vout[], const double vin[])
	MatrixTranspose-vector product vout= [I I]'vin (vout(n),vin(m) (M= [I I] matrix, n= 2m)

void	add_mul_Mv_idty_idty (double vout[], const double vin[])
	Add matrix-vector product vout += [I I]vin (vout(m),vin(n)) (M= [I I] matrix, n= 2m)

void	add_mul_Mtv_idty_idty (double vout[], const double vin[])
	Add matrixTranspose-vector product vout += [I I]'vin (vout(n),vin(m) (M= [I I] matrix, n= 2m)

void	mul_Mv_diag_diag (double vout[], const double vin[])
	Matrix-vector product vout= [D1 D2]vin (vout(m),vin(n)) (M= [D1 D2] matrix, n= 2m)

void	mul_Mtv_diag_diag (double vout[], const double vin[])
	MatrixTranspose-vector product vout= [D1 D2]'vin (vout(n),vin(m) (M= [D1 D2] matrix, n= 2m)

void	add_mul_Mv_diag_diag (double vout[], const double vin[])
	Add matrix-vector product vout += [D1 D2]vin (vout(m),vin(n)) (M= [D1 D2] matrix, n= 2m)

void	add_mul_Mtv_diag_diag (double vout[], const double vin[])
	Add matrixTranspose-vector product vout += [D1 D2]'vin (vout(n),vin(m) (M= [D1 D2] matrix, n= 2m)

void	mul_Mv_gen_sym_uptr (double vout[], const double vin[])
	Matrix-vector product vout=M*vin (vout(m),vin(n)) (general symmetric upper triangular rowwise matrix)

void	add_mul_Mv_gen_sym_uptr (double vout[], const double vin[])
	Add matrix-vector product vout += M*vin (vout(m),vin(n)) (general symmetric upper triangular rowwise matrix)

void	order_packed_format (int begsize, int nz, int &beg, int &ind, double *&val)
	Order a matrix packed format (both column-wise and row-wise)

Detailed Description

Class for manipulating matrices, and interfacing SparseChol.

Constructor & Destructor Documentation

MatrixBlockIP::MatrixBlockIP ( int blocks = 1 )

Constructor.

Parameters

blocks number of blocks where this matrix will appear;

Note: if blocks unknown at construction time, do not pass this parameter to the constructor; it will be updated later, before symbolic factorization is performed, by the minimization algorithm

MatrixBlockIP::MatrixBlockIP ( MatrixBlockIP * mbip )

Copy constructor, does not copy SparseChol.

Parameters

mbip	MatrixBlockIP to copy

Member Function Documentation

void MatrixBlockIP::add_mul_Mtv	(	double	vout[],
		const double	vin[]
	)

Add matrixTranspose-vector product vout += M(t)*vin (vout(n),vin(m) (driver)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mtv_column_wise	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrixTranspose-vector product vout += M(t)*vin (vout(n),vin(m) (general columnwise matrix)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mtv_diag_diag	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrixTranspose-vector product vout += [D1 D2]'*vin (vout(n),vin(m) (M= [D1 D2] matrix, n= 2*m)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mtv_idty_idty	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrixTranspose-vector product vout += [I I]'*vin (vout(n),vin(m) (M= [I I] matrix, n= 2*m)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mtv_network	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrixTranspose-vector product vout += M(t)*vin (vout(n),vin(m) (network matrix, either oriented or nonoriented)

Note: It assumes vin has dimension m+1; last equation, related to last node, is redundant and not used, but it is needed for an efficient computation.; Element vin[m] is used (but backed-up at beginning and restored at ending)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mtv_row_wise	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrixTranspose-vector product vout += M(t)*vin (vout(n),vin(m) (general rowwise matrix)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mv	(	double	vout[],
		const double	vin[]
	)

Add matrix-vector product vout += M*vin (vout(m),vin(n)) (driver)

Parameters

vout	Result of the product
vin	Vector to product

Here is the caller graph for this function:

void MatrixBlockIP::add_mul_Mv_column_wise	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrix-vector product vout += M*vin (vout(m),vin(n)) (general columnwise matrix)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mv_diag_diag	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrix-vector product vout += [D1 D2]*vin (vout(m),vin(n)) (M= [D1 D2] matrix, n= 2*m)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mv_diagonal	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add Diagonal-vector product vout += D*vin (vout(m),vin(n)) D is diagonal, m=n.

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mv_gen_sym_uptr	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrix-vector product vout += M*vin (vout(m),vin(n)) (general symmetric upper triangular rowwise matrix)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::add_mul_Mv_identity	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add Identity-vector product vout += M*vin (vout(m),vin(n))

Identity matrix, m=n, just add input to output vector

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mv_idty_idty	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrix-vector product vout += [I I]*vin (vout(m),vin(n)) (M= [I I] matrix, n= 2*m)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mv_network	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrix-vector product vout += M*vin (vout(m),vin(n)) (network matrix, either oriented or nonoriented)

Note: It assumes vout has dimension m+1; last equation, related to last node, is redundant and not used, but it is needed for an efficient computation.; Element vout[m] is used (but backed-up at beginning and restored at ending)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_mul_Mv_row_wise	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Add matrix-vector product vout += M*vin (vout(m),vin(n)) (general rowwise matrix)

Parameters

vout	Result of the product
vin	Vector to product

void MatrixBlockIP::add_new_column	(	int	size,
		int	irows[],
		double	values[]
	)

Add a new column into the matrix.

Parameters

size	Number of non-zero elements of the new column
irows	Row position for each element, have to be ordered
values	Value of non-zero elements

void MatrixBlockIP::add_new_columns	(	int	num_columns,
		int	size[],
		int *	irows[],
		double *	values[]
	)

Add new columns into the matrix.

Parameters

num_columns	Number of columns to add
size	number of non-zero elements of the new column for each column
irows	Row position for each element and column, have to be ordered
values	Value of non-zero elements for each column

void MatrixBlockIP::analyze_D	(	int	numBlocks,
		bool	sameL,
		MatrixBlockIP	L[]
	)

inline

Create internal structure arrays to compute D = Theta_(numBlocks+1) + sum{i..numBlocks} L_i*Theta_i*L_i' It decides whether D is DIAGONAL or GEN_SYM_UPTR

Parameters

numBlocks	Number of diagonal blocks
sameL	Define if the same matrix is used for each L block
L	Linking constraints blocks

void MatrixBlockIP::analyze_D_diagonal	(	int	numBlocks,
		bool	sameL,
		MatrixBlockIP	L[]
	)

Create internal structure arrays to compute D = Theta_(numBlocks+1) + sum{1..numBlocks} L_i*Theta_i*L_i' when D is a diagonal

Parameters

numBlocks	Number of diagonal blocks
sameL	Define if the same matrix is used for each L block
L	Linking constraints blocks

Note: valD (of analyze_D_gen_sym_uptr()) is reused, then it is allocated with ALLOC (see analyze_D_gen_sym_uptr for an explanation)

void MatrixBlockIP::analyze_D_gen_sym_uptr	(	int	numBlocks,
		bool	sameL,
		MatrixBlockIP	L[]
	)

Create internal structure arrays to compute D = Theta_(numBlocks+1) + sum{i..numBlocks} L_i*Theta_i*L_i' when D is a general symmetric upper triangular matrix

Parameters

numBlocks	Number of diagonal blocks
sameL	Define if the same matrix is used for each L block
L	Linking constraints blocks

Note: icolLLt, inivalD, blockD, valD and icolD are allocated with ALLOC for performance purposes, so they have to be freed with FREE

void MatrixBlockIP::change_columns_sign	(	int	size,
		int	columns[]
	)

Changes the sign of some columns.

Parameters

size	Number of columns to change the sign
columns	Index to the columns to change the sign

void MatrixBlockIP::change_rows_sign	(	int	size,
		int	rows[]
	)

Change the sign of some rows.

Parameters

size	Number of rows to change the sign
rows	Index to the rows to change the sign

void MatrixBlockIP::compute_D	(	int	numBlocks,
		bool	sameL,
		MatrixBlockIP	L[],
		bool	isActive[],
		int	iniTheta[],
		double	Theta[]
	)

inline

Compute D = Theta_(numBlocks+1) + sum{i..numBlocks} L_i*Theta_i*L_i'.

Parameters

numBlocks	Number of diagonal blocks
sameL	Define if the same matrix is used for each L block
isActive	Says what linking constrains are active
iniTheta	Indice to the first element of the Theta matrix of each block
Theta	numBlocks diagonal matrices with dimension n_i x n_i

void MatrixBlockIP::compute_D_diagonal	(	int	numBlocks,
		bool	sameL,
		MatrixBlockIP	L[],
		bool	isActive[],
		int	iniTheta[],
		double	Theta[]
	)

Compute D = Theta_(numBlocks+1) + sum{i in 1..numBlocks} L_i*Theta_i*L_i' when D is diagonal.

Parameters

numBlocks	Number of diagonal blocks
sameL	Define if the same matrix is used for each L block
isActive	Says what linking constrains are active
iniTheta	Indice to the first element of the Theta matrix of each block
Theta	numBlocks diagonal matrices with dimension n_i x n_i

void MatrixBlockIP::compute_D_gen_sym_uptr	(	int	numBlocks,
		bool	sameL,
		MatrixBlockIP	L[],
		bool	isActive[],
		int	iniTheta[],
		double	Theta[]
	)

Compute D = Theta_(numBlocks+1) + sum{i..numBlocks} L_i*Theta_i*L_i' when D is a symmetric general matrix.

Parameters

numBlocks	Number of diagonal blocks
sameL	Define if the same matrix is used for each L block
isActive	Says what linking constrains are active
iniTheta	Indice to the first element of the Theta matrix of each block
Theta	numBlocks diagonal matrices with dimension n_i x n_i

void MatrixBlockIP::compute_full_matrix	(	int	numBlocks,
		bool	sameN,
		MatrixBlockIP	N[],
		bool	sameL,
		MatrixBlockIP	L[],
		int	numActiveLnk = `0`,
		int *	listActiveLnk = `NULL`
	)

Builds a packed rowwise format structured matrix based on diagonal blocks and linking constraints blocks.

Parameters

numBlocks	Number of diagonal blocks
sameN	Define if the same matrix is used for each N block
N	Diagonal blocks
sameL	Define if the same matrix is used for each L block
L	Linking constraints blocks
numActiveLnk	Number of active lnk
listActiveLnk	List of active linking, j=listActiveLnk[i], i=1..numActiveLnk, j in {1,..,l_link}) If listActiveLnk is NULL all L rows are used

Here is the call graph for this function:

Here is the caller graph for this function:

void MatrixBlockIP::copy ( MatrixBlockIP * mbip )

Copy, does not copy SparseChol.

Parameters

mbip	MatrixBlockIP to copy

void MatrixBlockIP::create_diag_diag_matrix	(	int	nrows,
		double *&	d1,
		double *&	d2
	)

Native method to create a diagonal-diagonal matrix D= [D1 D2].

where D1= diag(d1), D2= diag(d2) (dimension: nrows x (2nrows))

Parameters

nrows	Number of rows
d1	Values of the diagonal D1
d2	Values of the diagonal D2

void MatrixBlockIP::create_diagonal_matrix	(	int	dim,
		double *&	d
	)

Native method to create a diagonal matrix D= diag(d) of dimension dim.

Parameters

dim	Dimension of the matrix
d	Values of the diagonal elements

Here is the caller graph for this function:

void MatrixBlockIP::create_general_matrix_column_wise	(	int	m,
		int	n,
		int	nz,
		int *&	inicola,
		int *&	irowa,
		double *&	a
	)

Native method to create a general column-wise packed matrix.

Parameters

m	Number of rows
n	Number of columns
nz	Number of non-zero elements
inicola	Index to the first element of each column
irowa	Row position for each element
a	Value of non-zero elements

Note: If icola is not ordered order_matrix function must be called when the arrays are filled

Here is the caller graph for this function:

void MatrixBlockIP::create_general_matrix_format_ija	(	int	m,
		int	n,
		int	nz,
		int *&	row_index,
		int *&	col_index,
		double *&	values
	)

Creates a general matrix in format ija.

Parameters

m	Number of rows
n	Number of columns
nz	Number of non-zero elements
row_index	Row position for each element
col_index	Column position for each element
values	Value of non-zero elements

void MatrixBlockIP::create_general_matrix_row_wise	(	int	m,
		int	n,
		int	nz,
		int *&	inirowa,
		int *&	icola,
		double *&	a
	)

Native method to create a general row-wise packed matrix.

Parameters

m	Number of rows
n	Number of columns
nz	Number of non-zero elements
inirowa	Index to the first element of each row
icola	Column position for each element
a	Value of non-zero elements

Note: If icola is not ordered order_matrix function must be called when the arrays are filled

void MatrixBlockIP::create_identity_matrix ( int dim )

Native method to create a identity matrix of dimension dim.

Parameters

dim	Dimension of the matrix

void MatrixBlockIP::create_idty_idty_matrix ( int nrows )

Native method to create a identity-identity matrix with two identities [I I].

Dimension: nrows x (2nrows)

Parameters

nrows Number of rows

void MatrixBlockIP::create_network_matrix	(	int	num_arcs,
		int	num_nodes,
		int *&	src,
		int *&	dst,
		bool	oriented = `true`
	)

Native method to create a network matrix.

Parameters

num_arcs	Number of arcs
num_nodes	Number of nodes
src	Source of each arc. Dimension num_arcs
dst	Destination of each arc. Dimension num_arcs
oriented	For oriented/nonoriented networks

void MatrixBlockIP::delete_rows	(	int	size,
		int	rows[]
	)

Delete some rows.

Parameters

size	Number of rows to be deleted
rows	Index to the rows to be deleted, must be ordered

void MatrixBlockIP::exist_var_in_row	(	int	column,
		bool	appear[],
		double	coefs[]
	)

Given a column determines if exists an element for each row of the matrix, in case of exist returns the value.

Parameters

column	The column to examine
appear	The array that says if a element exists or not for each row. The user must allocate the space before call the function
coefs	The array that have the value of the element if exists, if not the content in that position is indeterminate. The user must allocate the space before call the function

int MatrixBlockIP::get_column	(	int	column,
		int *&	irows,
		double *&	values,
		bool	invert_sign = `false`
	)

Gets a column of the matrix.

Parameters

column	Index to the column
irows	Row index for each ealement. Must be freed with delete[]
values	Value for each non-zero element. Must be freed with delete[]
invert_sign	If true change the sign of each element in the column

Returns: Number of non-zero elements in the column

Here is the caller graph for this function:

int MatrixBlockIP::get_ipfa ( int i )

inline

Interface routine to SparseChol get_ipfa().

Returns: ipfa[i]

Note: The user must guarantee ipfa previously computed in call to symbolic_fact().

int MatrixBlockIP::get_maxfillin ( )

inline

Interface routine to SparseChol get_maxfillin().

Returns: maxfillin

Note: The user must guarantee maxfillin previously computed in call to symbolic_fact().

int MatrixBlockIP::get_maxlnz ( )

inline

Interface routine to SparseChol get_maxlnz().

Returns: maxlnz

Note: The user must guarantee maxlnz previously computed in call to symbolic_fact().

int MatrixBlockIP::get_njka ( )

inline

Interface routine to SparseChol get_njka().

Returns: njka

Note: The user must guarantee njka previously computed in call to symbolic_fact().

int MatrixBlockIP::get_num_semidef_matrix ( )

inline

Interface routine to SparseChol get_semidef_matrix().

Returns: num_semidef_matrix

Note: It will 0 if no numeric factorization made.

int MatrixBlockIP::get_num_zero_pivots ( )

inline

Interface routine to SparseChol get_num_zero_pivots().

Returns: num_zero_pivots

Note: It will 0 if no numeric factorization made.

int MatrixBlockIP::get_pfa ( int i )

inline

Interface routine to SparseChol get_pfa().

Returns: pfa[i]

Note: The user must guarantee pfa previously computed in call to symbolic_fact().

void MatrixBlockIP::ija_to_rowwise ( )

Calculates and stores the matrix in packed rowwise format.

Note: ija format is loaded

void MatrixBlockIP::mul_Mtv	(	double	vout[],
		const double	vin[]
	)

MatrixTranspose-vector product vout=M(t)*vin (vout(n),vin(m) (driver)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mtv_column_wise	(	double	vout[],
		const double	vin[]
	)

inlineprivate

MatrixTranspose-vector product vout=M(t)*vin (vout(n),vin(m) (general columnwise matrix)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mtv_diag_diag	(	double	vout[],
		const double	vin[]
	)

inlineprivate

MatrixTranspose-vector product vout= [D1 D2]'*vin (vout(n),vin(m) (M= [D1 D2] matrix, n= 2*m)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mtv_idty_idty	(	double	vout[],
		const double	vin[]
	)

inlineprivate

MatrixTranspose-vector product vout= [I I]'*vin (vout(n),vin(m) (M= [I I] matrix, n= 2*m)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mtv_network	(	double	vout[],
		const double	vin[]
	)

inlineprivate

MatrixTranspose-vector product vout=M(t)*vin (vout(n),vin(m) (network matrix, either oriented or nonoriented)

Note: It assumes vin has dimension m+1; last equation, related to last node, is redundant and not used, but it is needed for an efficient computation.; Element vin[m] is used (but backed-up at beginning and restored at ending)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mtv_row_wise	(	double	vout[],
		const double	vin[]
	)

inlineprivate

MatrixTranspose-vector product vout=M(t)*vin (vout(n),vin(m) (general rowwise matrix)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mv	(	double	vout[],
		const double	vin[]
	)

Matrix-vector product vout=M*vin (vout(m),vin(n)) (driver)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mv_column_wise	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Matrix-vector product vout=M*vin (vout(m),vin(n)) (general columnwise matrix)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mv_diag_diag	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Matrix-vector product vout= [D1 D2]*vin (vout(m),vin(n)) (M= [D1 D2] matrix, n= 2*m)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mv_diagonal	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Diagonal-vector product vout=D*vin (vout(m),vin(n)) D is diagonal, m=n.

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mv_gen_sym_uptr	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Matrix-vector product vout=M*vin (vout(m),vin(n)) (general symmetric upper triangular rowwise matrix)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mv_identity	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Identity-vector product vout=M*vin (vout(m),vin(n))

Identity matrix, m=n, just copy input to output vector

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mv_idty_idty	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Matrix-vector product vout= [I I]*vin (vout(m),vin(n)) (M= [I I] matrix, n= 2*m)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mv_network	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Matrix-vector product vout=M*vin (vout(m),vin(n)) (network matrix, either oriented or nonoriented)

Note: It assumes vout has dimension m+1; last equation, related to last node, is redundant and not used, but it is needed for an efficient computation.; Element vout[m] is used (but backed-up at beginning and restored at ending)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::mul_Mv_row_wise	(	double	vout[],
		const double	vin[]
	)

inlineprivate

Matrix-vector product vout=M*vin (vout(m),vin(n)) (general rowwise matrix)

Parameters

vout	Result of the product. Must be allocated
vin	Vector to product

void MatrixBlockIP::native_to_general ( bool delete_native_format = false )

inline

Calculates and stores the matrix in packed rowwise format.

Parameters

delete_native_format If true only packed rowwise format will be stored

Here is the caller graph for this function:

void MatrixBlockIP::order_packed_format	(	int	begsize,
		int	nz,
		int *&	beg,
		int *&	ind,
		double *&	val
	)

private

Order a matrix packed format (both column-wise and row-wise)

Parameters

begsize	Number of rows (if row-wise) or columns (if column-wise) + 1
nz	Number of nonzeros in the matrix.
beg	Rows (if row-wise) or columns (if column-wise) starts of the matrix.
ind	Columns (if row-wise) or rows (if column-wise) indices of nonzeros entries.
val	Values of the nonzero entries.

Note: Parameters beg, ind and val assumes to be of appropriate dimensions before the method is called, namely beg[begsize], ind[nz], val[nz]

void MatrixBlockIP::print_matrix	(	bool	print_ija = `true`,
		bool	start_one = `true`
	)

Print the matrix.

Parameters

print_ija	If print_ija is true then write triple (i,j,a) whenever possible (if matrix is in ija or packed rowwise format); if print_ija=false then writes (inirow, j, a) whenever possible (if matrix is in packed rowwise format).
start_one	Start vectors at position 1 (true) or zero (false)

void MatrixBlockIP::print_matrix	(	ofstream &	outfile,
		bool	print_ija = `true`,
		bool	start_one = `true`
	)

Write the matrix into a file.

Parameters

outfile Output file stream where the matrix will be printed

void MatrixBlockIP::print_vector	(	double *	v,
		int	size,
		string	name
	)

Print some positions and name of a vector.

Parameters

v	Vector to print
size	Number of positions to print
name	Vector name

Member Data Documentation

SparseChol MatrixBlockIP::chol

Sparse Cholesky class.

For sparse Cholesky

double* MatrixBlockIP::d1

d1 for DIAGONAL and 1st submatrix of DIAG_DIAG

Diagonal format attributes

int* MatrixBlockIP::inicola

Index to the first element of each column.

General column-wise packed format attributes

int* MatrixBlockIP::inirowa

Index to the first element of each row.

General row-wise packed format attributes

bool MatrixBlockIP::made_analyze_D

boolean to check whether analyze_D already made, needed to compute_D()

bool MatrixBlockIP::made_symbfct_MMt

boolean to check whether symbolic factorization already made

int MatrixBlockIP::num_blocks

number of replications of this matrix; this is needed to store space for num_blocks numerical factorization of M*Theta[i]*M', for different Theta[i], i=0,...,num_blocks-1 This value is set by the minimization algorithm, which needs the factorizations.

int MatrixBlockIP::nz

Number of non-zero elements.

Common packed format attributes

int MatrixBlockIP::sizeL

For D Matrix

TYPE_MATRIX MatrixBlockIP::type

Type of matrix.

Common in all types matrix

TYPE_ORIENTATION MatrixBlockIP::type_orientation

Type of orientation, by default, oriented.

Network format attributes

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

/home/jcastro2/intpoint/BlockIPPlatform/BlockIP/BlockIP/src_blockip-v2/MatrixBlockIP.h
/home/jcastro2/intpoint/BlockIPPlatform/BlockIP/BlockIP/src_blockip-v2/MatrixBlockIP.C

Classes

Public Member Functions

Public Attributes

Private Member Functions

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation