Typedefs
typedef struct scs_a_data_matrix	ScsAMatrix

typedef struct scs_private_data	ScsPrivWorkspace

Functions
ScsPrivWorkspace *	scs_init_priv (const ScsAMatrix A, const ScsSettings stgs)

scs_int	scs_solve_lin_sys (const ScsAMatrix A, const ScsSettings stgs, ScsPrivWorkspace p, scs_float b, const scs_float *s, scs_int iter)

void	scs_free_priv (ScsPrivWorkspace *p)

void	scs_accum_by_a_trans (const ScsAMatrix A, ScsPrivWorkspace p, const scs_float x, scs_float y)

void	scs_accum_by_a (const ScsAMatrix A, ScsPrivWorkspace p, const scs_float x, scs_float y)

scs_int	scs_validate_linsys (const ScsAMatrix *A)

char *	scs_get_linsys_method (const ScsAMatrix A, const ScsSettings stgs)

char *	scs_get_linsys_summary (ScsPrivWorkspace p, const ScsInfo info)

void	scs_normalize_a (ScsAMatrix A, const ScsSettings stgs, const ScsCone k, ScsScaling scal)

void	scs_unnormalize_a (ScsAMatrix A, const ScsSettings stgs, const ScsScaling *scal)

void	scs_free_a_matrix (ScsAMatrix *A)

scs_int	scs_linsys_is_indirect (void)

scs_int	scs_linsys_total_cg_iters (ScsPrivWorkspace *priv)

scs_float	scs_linsys_total_solve_time_ms (ScsPrivWorkspace *priv)

Typedef Documentation

typedef struct scs_a_data_matrix ScsAMatrix

Structure supporting sparse matrices in CSC format.

See Also: Sparse matrices documentation

typedef struct scs_private_data ScsPrivWorkspace

Stores the necessary private workspace, only the linear system solver interacts with this struct

Function Documentation

void scs_accum_by_a	(	const ScsAMatrix *	A,
		ScsPrivWorkspace *	p,
		const scs_float *	x,
		scs_float *	y
	)

Performs y += A*x

void scs_accum_by_a_trans	(	const ScsAMatrix *	A,
		ScsPrivWorkspace *	p,
		const scs_float *	x,
		scs_float *	y
	)

Performs y += A'*x

void scs_free_a_matrix ( ScsAMatrix * A )

Frees the memory allocated in ScsAMatrix.

void scs_free_priv ( ScsPrivWorkspace * p )

Frees scs_linsys_priv_workspace structure and allocated memory in it.

char* scs_get_linsys_method	(	const ScsAMatrix *	A,
		const ScsSettings *	stgs
	)

Returns string describing method, can return null, if not null free will be called on output

char* scs_get_linsys_summary	(	ScsPrivWorkspace *	p,
		const ScsInfo *	info
	)

Returns string containing summary information about linear system solves, can return null, if not null free will be called on output

ScsPrivWorkspace* scs_init_priv	(	const ScsAMatrix *	A,
		const ScsSettings *	stgs
	)

initialize scs_linsys_priv_workspace structure and perform any necessary preprocessing

scs_int scs_linsys_is_indirect ( void )

Whether the solution of the linear system is based on an indirect method.

Returns: 1 if indirect, 0 otherwise

scs_int scs_linsys_total_cg_iters ( ScsPrivWorkspace * priv )

Total number of CG iterations if an indirect method is used, -1 otherwise

Parameters

priv	private workspace structure

Returns: total CG iterations

scs_float scs_linsys_total_solve_time_ms ( ScsPrivWorkspace * priv )

Returns the total solve time for the linear system (in ms).

Parameters

priv	private workspace structure

Returns: solve time in ms

void scs_normalize_a	(	ScsAMatrix *	A,
		const ScsSettings *	stgs,
		const ScsCone *	k,
		ScsScaling *	scal
	)

normalizes A matrix, sets w->E and w->D diagonal scaling matrices,Anew = d->SCALE * (D^-1)*A*(E^-1) (different to paper which is D*A*E) D and E must be all positive entries, D must satisfy cone boundaries must set (w->meanNormRowA = mean of norms of rows of normalized A) THEN scale resulting A by d->SCALE.

scs_int scs_solve_lin_sys	(	const ScsAMatrix *	A,
		const ScsSettings *	stgs,
		ScsPrivWorkspace *	p,
		scs_float *	b,
		const scs_float *	s,
		scs_int	iter
	)

Solves [d->RHO_X * I A' ; A -I] x = b for x, stores result in b, s contains warm-start, iter is current scs iteration count

Parameters

A	sparse matrix A
stgs	user-specified settings
p	private structure
b	right hand side
s	initial guess
iter

Returns: on success, returns 0

void scs_unnormalize_a	(	ScsAMatrix *	A,
		const ScsSettings *	stgs,
		const ScsScaling *	scal
	)

unnormalizes A matrix, unnormalizes by w->D and w->E and d->SCALE.

scs_int scs_validate_linsys ( const ScsAMatrix * A )

Returns negative num if input data is invalid

Typedefs

Functions

Typedef Documentation

Function Documentation