Solves convex cone programs via operator splitting. Can solve: linear programs ('LPs'), second-order cone programs ('SOCPs'), semidefinite programs ('SDPs'), exponential cone programs ('ECPs'), and power cone programs ('PCPs'), or problems with any combination of those cones. 'SCS' uses 'AMD' (a set of routines for permuting sparse matrices prior to factorization) and 'LDL' (a sparse 'LDL' factorization and solve package) from 'SuiteSparse' (< http://www.suitesparse.com>).
Parameter | Type | Description | Default Value |
---|---|---|---|
max_iters |
integer | maximum number of iterations | 5000 |
normalize |
boolean | turn heuristic data rescaling on and off | TRUE |
verbose |
boolean | write out progress | FALSE |
cg_rate |
numeric | for indirect, tolerance goes down like (1 / iter)^cg_rate |
2 |
scale |
numeric | if normalized, rescales by this factor | 1 |
rho_x |
numeric | x equality constraint scaling | 1e-3 |
alpha |
numeric | relaxation parameter | 1.5 |
eps |
numeric | convergence tolerance | 1e-5 |
https://doi.org/10.1007/s10957-016-0892-3
.