Benchmark "lgeb" :
Laplacian problem in L-shaped domain
Is there a different behavior of adaptive F.E.M. for SINGULAR and NON-SINGULAR
solutions ?
|
|
Singular Solution | Non-singular Solution |
| |
Norm of Gradient of Singular Solution | Norm of Gradient of Non-singular Solution |
| |
Material coefficients: | a1 | a2 | c |
| 1. | 1. | 0. |
source term : | f |
|
special function |
Boundary conditions : |
Dirichlet - type: | 0 | (overall) |
Typical behavior of estimated relative error (linear postprocessing for linear elements):
(Legend)
for Singular Solution | for Non-singular Solution |
| |
Behavior of the exact relative error (linear postprocessing for linear elements):
for Singular Solution | for Non-singular Solution |
| |
Efficiency index (=ex.error / est.error) (linear postprocessing for linear elements):
for Singular Solution | for Non-singular Solution |
| |
Now Speciality: QUADRATIC postprocessing for linear macro-elements:
Typical behavior of estimated relative error (quadratic postprocessing for linear elements !!):
(Legend)
for Singular Solution | for Non-singular Solution |
| |
Behavior of the exact relative error (quadratic postprocessing for linear elements !!):
for Singular Solution | for Non-singular Solution |
| |
Efficiency index (=ex.error / est.error) (quadr. postprocessing for linear elements):
for Singular Solution | for Non-singular Solution |
| |