On 4/5/21 3:33 AM, Konrad Simon wrote:
I stumbled over something in 2d meshes that confuses me a bit. Supoose I have
2 squares (a left and a right one) sharing one edge. If everything is fine all
lines and faces have standard orientation, i.e., line_orientation==true and
face_orientation==true for all lines and faces. Note that lines and faces are
NOT the same geometric entity here since their orientation is determined
differently (faces have normals and lines have tangents).
Now, if I rotate the right square by 90 degrees I get (on the shared edge)
non-matching face normals but all line orientations do match.
If I rotate the right square by 180 degrees I get (on the shared edge)
non-matching face normals and non-matching line orientations.
If I rotate the right square by 270 degrees I get (on the shared edge)
matching face normals and non-matching line orientations.
This is at least my intuition. When I query these information for each
rotation case I see that my intuition is confirmed for lines but the face
orientations are always true.
Is this a bug?
I am asking because it seems that 2d RaviartThomas elements (which make use of
face normals) are broken for some edge cases (3d RT I could fix).
It's possible this is indeed a bug. In most cases, we run the 2d meshes
through the mesh orienter, so we rarely see 2d meshes that are not correctly
oriented and it wouldn't surprise me if we have the kind of bug you describe
where things almost but not quite work correctly.
The questions then are (i) where does this need to be fixed, and (ii) how many
tests fail if we apply the fix? One of the things I learned (and I think you
did as well) is that for bugs in these kinds of layers, it is quite possible
that one ends up with 100 failing tests and then it takes a *lot* of work to
look through all of these to figure out whether the new output is correct or
not. But then, as mentioned above, it's also possible that only one or two
tests fail because we almost never see these kinds of meshes.
Best
W.
--
------------------------------------------------------------------------
Wolfgang Bangerth email: bange...@colostate.edu
www: http://www.math.colostate.edu/~bangerth/
--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see
https://groups.google.com/d/forum/dealii?hl=en
---
You received this message because you are subscribed to the Google Groups "deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/dealii/8f863ee5-302e-68e8-89fb-f100b8537bbb%40colostate.edu.