 |
Reference documentation for deal.II version 8.5.1
|
|
Reference documentation for deal.II version 8.5.1
|
#include <deal.II/lac/trilinos_precondition.h>
Public Member Functions | |
| AdditionalData (const bool elliptic=true, const bool higher_order_elements=false, const unsigned int n_cycles=1, const bool w_cyle=false, const double aggregation_threshold=1e-4, const std::vector< std::vector< bool > > &constant_modes=std::vector< std::vector< bool > >(0), const unsigned int smoother_sweeps=2, const unsigned int smoother_overlap=0, const bool output_details=false, const char *smoother_type="Chebyshev", const char *coarse_type="Amesos-KLU") | |
Public Attributes | |
| bool | elliptic |
| bool | higher_order_elements |
| unsigned int | n_cycles |
| bool | w_cycle |
| double | aggregation_threshold |
| std::vector< std::vector< bool > > | constant_modes |
| unsigned int | smoother_sweeps |
| unsigned int | smoother_overlap |
| bool | output_details |
| const char * | smoother_type |
| const char * | coarse_type |
A data structure that is used to control details of how the algebraic multigrid is set up. The flags detailed in here are then passed to the Trilinos ML implementation. A structure of the current type are passed to the constructor of PreconditionAMG.
Definition at line 1364 of file trilinos_precondition.h.
| TrilinosWrappers::PreconditionAMG::AdditionalData::AdditionalData | ( | const bool | elliptic = true, |
| const bool | higher_order_elements = false, |
||
| const unsigned int | n_cycles = 1, |
||
| const bool | w_cyle = false, |
||
| const double | aggregation_threshold = 1e-4, |
||
| const std::vector< std::vector< bool > > & | constant_modes = std::vector<std::vector<bool> > (0), |
||
| const unsigned int | smoother_sweeps = 2, |
||
| const unsigned int | smoother_overlap = 0, |
||
| const bool | output_details = false, |
||
| const char * | smoother_type = "Chebyshev", |
||
| const char * | coarse_type = "Amesos-KLU" |
||
| ) |
Constructor. By default, we pretend to work on elliptic problems with linear finite elements on a scalar equation.
Definition at line 78 of file trilinos_precondition_ml.cc.
| bool TrilinosWrappers::PreconditionAMG::AdditionalData::elliptic |
Determines whether the AMG preconditioner should be optimized for elliptic problems (ML option smoothed aggregation SA, using a Chebyshev smoother) or for non-elliptic problems (ML option non- symmetric smoothed aggregation NSSA, smoother is SSOR with underrelaxation).
Definition at line 1389 of file trilinos_precondition.h.
| bool TrilinosWrappers::PreconditionAMG::AdditionalData::higher_order_elements |
Determines whether the matrix that the preconditioner is built upon is generated from linear or higher-order elements.
Definition at line 1395 of file trilinos_precondition.h.
| unsigned int TrilinosWrappers::PreconditionAMG::AdditionalData::n_cycles |
Defines how many multigrid cycles should be performed by the preconditioner.
Definition at line 1401 of file trilinos_precondition.h.
| bool TrilinosWrappers::PreconditionAMG::AdditionalData::w_cycle |
Defines whether a w-cycle should be used instead of the standard setting of a v-cycle.
Definition at line 1407 of file trilinos_precondition.h.
| double TrilinosWrappers::PreconditionAMG::AdditionalData::aggregation_threshold |
This threshold tells the AMG setup how the coarsening should be performed. In the AMG used by ML, all points that strongly couple with the tentative coarse-level point form one aggregate. The term strong coupling is controlled by the variable aggregation_threshold, meaning that all elements that are not smaller than aggregation_threshold times the diagonal element do couple strongly.
Definition at line 1418 of file trilinos_precondition.h.
| std::vector<std::vector<bool> > TrilinosWrappers::PreconditionAMG::AdditionalData::constant_modes |
Specifies the constant modes (near null space) of the matrix. This parameter tells AMG whether we work on a scalar equation (where the near null space only consists of ones, and default value is OK) or on a vector-valued equation. For vector-valued equation problem with n_component, the provided constant_modes should fulfill the following requirements:
n_component ic][id] == "idth DoF is corresponding to component ic Definition at line 1436 of file trilinos_precondition.h.
| unsigned int TrilinosWrappers::PreconditionAMG::AdditionalData::smoother_sweeps |
Determines how many sweeps of the smoother should be performed. When the flag elliptic is set to true, i.e., for elliptic or almost elliptic problems, the polynomial degree of the Chebyshev smoother is set to smoother_sweeps. The term sweeps refers to the number of matrix-vector products performed in the Chebyshev case. In the non-elliptic case, smoother_sweeps sets the number of SSOR relaxation sweeps for post-smoothing to be performed.
Definition at line 1448 of file trilinos_precondition.h.
| unsigned int TrilinosWrappers::PreconditionAMG::AdditionalData::smoother_overlap |
Determines the overlap in the SSOR/Chebyshev error smoother when run in parallel.
Definition at line 1454 of file trilinos_precondition.h.
| bool TrilinosWrappers::PreconditionAMG::AdditionalData::output_details |
If this flag is set to true, then internal information from the ML preconditioner is printed to screen. This can be useful when debugging the preconditioner.
Definition at line 1461 of file trilinos_precondition.h.
| const char* TrilinosWrappers::PreconditionAMG::AdditionalData::smoother_type |
Determines which smoother to use for the AMG cycle. Possibilities for smoother_type are the following:
Definition at line 1497 of file trilinos_precondition.h.
| const char* TrilinosWrappers::PreconditionAMG::AdditionalData::coarse_type |
Determines which solver to use on the coarsest level. The same settings as for the smoother type are possible.
Definition at line 1503 of file trilinos_precondition.h.
1.8.13