#990 boundary segment indices in ALUSimplexGrid get mixed up
Metadata
| Property | Value |
|---|---|
| Reported by | Lars Lubkoll (lars.lubkoll@googlemail.com) |
| Reported at | Dec 12, 2011 08:57 |
| Type | Bug Report |
| Version | 2.0 |
| Operating System | Linux 64bit |
| Last edited by | Carsten Gräser (graeser@math.fu-berlin.de) |
| Last edited at | Dec 14, 2011 13:32 |
| Closed by | Carsten Gräser (graeser@math.fu-berlin.de) |
| Closed at | Dec 14, 2011 13:32 |
| Closed in version | Unknown |
| Resolution | Not a bug |
| Comment |
Description
I am working with Dune2.0 and compiling with gcc-4.3.2. I am solving PDE-problems with mixed boundary conditions. In order to detect the correct boundary condition I store an integer for each boundary segment in a vector sorted according to the level-0 boundary segment index (i.e. the order of insertion of boundary elements into the grid factory) I then use 'Intersection::boundarySegmentIndex()' to get the index and retrieve the stored integer from the created vector. This worked fine using UGGrid until I had to use 2 grids in one computation (the corresponding bug has been reported to Oliver Sander, who took care of it). Therefore I did try to use ALUSimplexGrid instead. It seems like the boundary segment indices get mixed up in the latter (also when using one grid). You will find attached 4 .png-files, that contain an example geometry, on which a scalar function, linear on each tetrahedron, has been defined in order to visualize the imposed boundary conditions. If everything works fine the result should look as in "boundaryConditions.png" and "boundaryConditions2.png, which has been created using one UGGrid. I then did exactly the same, only exchanging UGGrid with ALUSimplexGrid, and get the result visualized in "boundaryConditions_ALU_.png" and "boundaryConditions_ALU_2.png which is wrong:(