Babuska-Rheinboldt refinement criterion. Determines the refinement thresholds such that all cells with an error contribution exceeding the expected error contribution of the worst cell after refinement will be marked. This tends to equilibrate the error contributions quite fast. More...
#include <errorest.hh>
Detailed Description
Babuska-Rheinboldt refinement criterion. Determines the refinement thresholds such that all cells with an error contribution exceeding the expected error contribution of the worst cell after refinement will be marked. This tends to equilibrate the error contributions quite fast.
Note that the local convergence order should be estimated (important in the vicinity of singularities) but is here assumed to be the fixed specified order.
Definition at line 134 of file errorest.hh.
Public Member Functions | |
| BabuskaRheinboldtCriterion (std::vector< int > const &order) | |
| std::vector< double > | threshold (boost::multi_array< double, 2 > const &normalizedErrors) const |
| Computes the thresholds for refinement. More... | |
Constructor & Destructor Documentation
◆ BabuskaRheinboldtCriterion()
| Kaskade::BabuskaRheinboldtCriterion::BabuskaRheinboldtCriterion | ( | std::vector< int > const & | order | ) |
Member Function Documentation
◆ threshold()
|
inherited |
Computes the thresholds for refinement.
- Parameters
-
normalizedErrors a twodimensional array containing the normalized error contribution for each cell and each variable, i.e. normalizedErrors[i][j] contains the error contribution of cell i to the error in variable j, divided by the tolerance, such that the sum over i should not exceed one (for acceptance)
- Returns
- an array containing a threshold value such that any cell i for which normalizedErrors[i][j] > return[j] for any j is to be refined, or an empty array in case all variables are accurate enough
Referenced by Kaskade::EmbeddedErrorEstimator< VariableSetDescription, Scaling >::estimate().
The documentation for this class was generated from the following file:
