Preconditioner - HYPRE - BOOMERAMG
Usage:
-pc_type hypre -pc_hypre_type boomeramg
To improve answers sometimes, choose to have PETSc run several V or W cycles,
-ksp_type richardson -pc_type hypre -pc_hypre_type boomeramg
Preonly causes boomerAMG to use only one V/W cycle.
Options:
-pc_hypre_boomeramg_max_levels
Number of levels (of grids) allowed. Default = 25
-pc_hypre_boomeramg_max_iter
Maximum iterations used (None). Default = 1
-pc_hypre_boomeramg_tol
Convergence tolerance (None). Default = 1e-07
-pc_hypre_boomeramg_truncfactor
Truncation factor (None). Default = 0
-pc_hypre_boomeramg_strong_threshold
Threshold for being strongly connected (None). Default = 0.25
-pc_hypre_boomeramg_max_row_sum
Maximum row sum (None). Default = 0.9
-pc_hypre_boomeramg_grid_sweeps_all
Number of sweeps for all grid levels (fine, up, and down) (None). Default = 1
-pc_hypre_boomeramg_grid_sweeps_fine
Number of sweeps for the fine level (None). Default = 1
-pc_hypre_boomeramg_grid_sweeps_down
Number of sweeps for the down cycles (None). Default = 1
-pc_hypre_boomeramg_grid_sweeps_up
Number of sweeps for the up cycles (None). Default = 1
-pc_hypre_boomeramg_grid_sweeps_coarse
Number of sweeps for the coarse level (None). Default = 1
-pc_hypre_boomeramg_relax_type_all
(choose one of) Default = SOR/Jacobi
Jacobi
sequential-Gauss-Seidel
SOR/Jacobi
backward-SOR/Jacobi
symmetric-SOR/Jacobi
Gaussian-elimination-pc_hypre_boomeramg_relax_type_fine
(choose one of) Default = SOR/Jacobi
Jacobi
sequential-Gauss-Seidel
SOR/Jacobi
backward-SOR/Jacobi
symmetric-SOR/Jacobi
Gaussian-elimination-pc_hypre_boomeramg_relax_type_down
(choose one of) Default = SOR/Jacobi
Jacobi
sequential-Gauss-Seidel
SOR/Jacobi
backward-SOR/Jacobi
symmetric-SOR/Jacobi
Gaussian-elimination-pc_hypre_boomeramg_relax_type_up
(choose one of) Default = SOR/Jacobi
Jacobi
sequential-Gauss-Seidel
SOR/Jacobi
backward-SOR/Jacobi
symmetric-SOR/Jacobi
Gaussian-elimination-pc_hypre_boomeramg_relax_type_coarse
(choose one of) Default = Gaussian-elimination
Jacobi
sequential-Gauss-Seidel
SOR/Jacobi
backward-SOR/Jacobi
symmetric-SOR/Jacobi
Gaussian-elimination-pc_hypre_boomeramg_relax_weight_all
Relaxation weight for all levels (0 = hypre estimates, -k = determined with k CG steps) (None). Default = 1
-pc_hypre_boomeramg_relax_weight_level
Set the relaxation weight for a particular level (weight,level) (None). Default = 1
-pc_hypre_boomeramg_outer_relax_weight_all
Outer relaxation weight for all levels ( -k = determined with k CG steps) (None). Default = 1
-pc_hypre_boomeramg_outer_relax_weight_level
Set the outer relaxation weight for a particular level (weight,level) (None). Default = 1
-pc_hypre_boomeramg_no_CF
Do not use CF-relaxation (None)
-pc_hypre_boomeramg_measure_type
(choose one of) Default = local
local
global-pc_hypre_boomeramg_coarsen_type
(choose one of) Default = Falgout
CLJP
Ruge-Stueben
modifiedRuge-Stueben
Falgout-pc_hypre_boomeramg_print_statistics
Print statistics (None)
Info: Implements multigrid. Due to setup times inherent in multigrid, this should only be used for larger problems, perhaps on the order of several hundreds of thousands or more. Since the hypre preconditioner can be utilized as a solver as well, sometimes this should be run with the option
-ksp_type preonly
Run with -ksp_view to see all the hypre options used.