R/clarabel.R
clarabel_control.Rd
Control parameters with default values and types in parenthesis
clarabel_control(
max_iter = 200L,
time_limit = Inf,
verbose = TRUE,
max_step_fraction = 0.99,
tol_gap_abs = 1e-08,
tol_gap_rel = 1e-08,
tol_feas = 1e-08,
tol_infeas_abs = 1e-08,
tol_infeas_rel = 1e-08,
tol_ktratio = 1e-06,
reduced_tol_gap_abs = 5e-05,
reduced_tol_gap_rel = 5e-05,
reduced_tol_feas = 1e-04,
reduced_tol_infeas_abs = 5e-05,
reduced_tol_infeas_rel = 5e-05,
reduced_tol_ktratio = 1e-04,
equilibrate_enable = TRUE,
equilibrate_max_iter = 10L,
equilibrate_min_scaling = 1e-04,
equilibrate_max_scaling = 10000,
linesearch_backtrack_step = 0.8,
min_switch_step_length = 0.1,
min_terminate_step_length = 1e-04,
direct_kkt_solver = TRUE,
direct_solve_method = c("qdldl", "mkl", "cholmod"),
static_regularization_enable = TRUE,
static_regularization_constant = 1e-08,
static_regularization_proportional = .Machine$double.eps * .Machine$double.eps,
dynamic_regularization_enable = TRUE,
dynamic_regularization_eps = 1e-13,
dynamic_regularization_delta = 2e-07,
iterative_refinement_enable = TRUE,
iterative_refinement_reltol = 1e-13,
iterative_refinement_abstol = 1e-12,
iterative_refinement_max_iter = 10L,
iterative_refinement_stop_ratio = 5,
presolve_enable = TRUE,
chordal_decomposition_enable = FALSE,
chordal_decomposition_merge_method = c("none", "parent_child", "clique_graph"),
chordal_decomposition_compact = FALSE,
chordal_decomposition_complete_dual = FALSE
)
maximum number of iterations (200L
)
maximum run time (seconds) (Inf
)
verbose printing (TRUE
)
maximum interior point step length (0.99
)
absolute duality gap tolerance (1e-8
)
relative duality gap tolerance (1e-8
)
feasibility check tolerance (primal and dual) (1e-8
)
absolute infeasibility tolerance (primal and dual) (1e-8
)
relative infeasibility tolerance (primal and dual) (1e-8
)
KT tolerance (1e-7
)
reduced absolute duality gap tolerance (5e-5
)
reduced relative duality gap tolerance (5e-5
)
reduced feasibility check tolerance (primal and dual) (1e-4
)
reduced absolute infeasibility tolerance (primal and dual) (5e-5
)
reduced relative infeasibility tolerance (primal and dual) (5e-5
)
reduced KT tolerance (1e-4
)
enable data equilibration pre-scaling (TRUE
)
maximum equilibration scaling iterations (10L
)
minimum equilibration scaling allowed (1e-4
)
maximum equilibration scaling allowed (1e+4
)
linesearch backtracking (0.8
)
minimum step size allowed for asymmetric cones with PrimalDual scaling (1e-1
)
minimum step size allowed for symmetric cones && asymmetric cones with Dual scaling (1e-4
)
use a direct linear solver method (required true) (TRUE
)
direct linear solver ("qdldl"
, "mkl"
or "cholmod"
) ("qdldl"
)
enable KKT static regularization (TRUE
)
KKT static regularization parameter (1e-8
)
additional regularization parameter w.r.t. the maximum abs diagonal term (.Machine.double_eps^2
)
enable KKT dynamic regularization (TRUE
)
KKT dynamic regularization threshold (1e-13
)
KKT dynamic regularization shift (2e-7
)
KKT solve with iterative refinement (TRUE
)
iterative refinement relative tolerance (1e-12
)
iterative refinement absolute tolerance (1e-12
)
iterative refinement maximum iterations (10L
)
iterative refinement stalling tolerance (5.0
)
whether to enable presolvle (TRUE
)
whether to enable chordal decomposition for SDPs (FALSE
)
chordal decomposition merge method, one of 'none'
, 'parent_child'
or 'clique_graph'
, for SDPs ('none'
)
a boolean flag for SDPs indicating whether to assemble decomposed system in compact form for SDPs (FALSE
)
a boolean flag indicating complete PSD dual variables after decomposition for SDPs
a list containing the control parameters.