Fixing bug in boundary discrete model #651
Conversation
The current implementation of BoundaryDiscreteModelsTests.jl does not allow for a different ordering of the cells than the one in the model from which it is created
were actually not well defined. This commit defines appropiately this set of tests.
in the portion the numbering of the faces of dimension d < D (as before), but sticks into the user-provided ordering for the faces of dimension d=D (this is the actual fix)
Codecov Report
@@ Coverage Diff @@
## master #651 +/- ##
=======================================
Coverage 87.72% 87.73%
=======================================
Files 134 134
Lines 14350 14360 +10
=======================================
+ Hits 12589 12599 +10
Misses 1761 1761
Continue to review full report at Codecov.
|
|
Hi @amartinhuertas, can you briefly comment on what the problem was? |
Sure. The previous-to-the-fix function aimed to preserve the relative order in the numbering of the objects of the background model in the numbering of the objects filtered for the BoundaryDiscreteModel. This is not necessarily OK for the objects of the highest dimension, i.e., the cells in the |
…in_boundary_discrete_model
|
@fverdugo ... FYI ... the following test (that I added on porpuse) was broken before the bug fix |
Ok. In any case BoundaryDiscreteModel is likely to be repaced by something else in the future to simplify the handling of mixed dimensional PDEs. |
Yes, please, :-) |
No description provided.