Settings

The settings are defined in the include/structs.h file.

Name	Type	Description	Allowed values	Default value
max_iters	QOCOInt	Maximum number of iterations	\((0, \infty)\)	200
ruiz_iters	QOCOInt	Number of Ruiz equilibration iterations performed	\((0, \infty)\)	0
max_ir_iters	QOCOInt	Maximum number of iterative refinement iterations	\((0, \infty)\)	5
ir_tol	QOCOFloat	Iterative refinement stopping tolerance: stop when \(\lVert Kx - b\rVert <\) ir_tol	\((0, \infty)\)	1e-6
kkt_static_reg_P	QOCOFloat	Static regularization for the (1,1) P block of the KKT system. Added to the diagonal of P before factorization to ensure the block remains positive definite.	\((0, \infty)\)	1e-13
kkt_static_reg_A	QOCOFloat	Static regularization for the (2,2) A block of the KKT system. Subtracted from the diagonal of the equality constraint block to give it a definite sign.	\((0, \infty)\)	1e-8
kkt_static_reg_G	QOCOFloat	Static regularization for the (3,3) G block of the KKT system. Subtracted from the diagonal of the NT scaling block to give it a definite sign.	\((0, \infty)\)	1e-13
kkt_dynamic_reg	QOCOFloat	Additional regularization applied dynamically during KKT factorization to any diagonal entry whose absolute value is below this threshold. Improves robustness on near-degenerate problems.	\((0, \infty)\)	1e-11
abstol	QOCOFloat	absolute tolerance	\((0, \infty)\)	1e-7
reltol	QOCOFloat	relative tolerance	\((0, \infty)\)	1e-7
abstol_inacc	QOCOFloat	low accuracy absolute tolerance	\((0, \infty)\)	1e-5
reltol_inacc	QOCOFloat	low accuracy relative tolerance	\((0, \infty)\)	1e-5
verbose	unsigned_char	should the solver print progress	0 or 1	0
dump_problem (Python only)	str or bool	(Python only) Dump the problem data to a binary .bin file. If a string, used as the filename. If True, defaults to "problem.bin".	any valid filepath or True	None