[Dune] Boundary conditions, segments and indexing

[Carsten Gräser]

Hi,

Am 24.08.20 um 11:18 schrieb Firmbach, Max:
[...]
> So for a minimal example, let's consider the domain given in the
> attached picture. This could be a slender beam or plate or something
> similar. On one side it is fixed at a wall (tag 1 for that B.C.), on the
> other
> end a loadvector, which could be a force is applied (tag 3 for that B.C.).
> The other boundary faces are tagged with 0, meaning that no B.C. should
> be applied there.
by a minimal example I mean a minimal peace of code that I can
compile and run to see the problem accompanied with a description
of the problem.
[...]
> so i guess they are just stored in the way appearing in the .msh file (
> consecutively ordered). Now I want to check how DUNE sees the boundary
I suspect that by "consecutively ordered" you mean that the tags are
ordered according to the boundary segments. The question is, in which
order do you insert the boundary segments?

[...]
> This generates the following output listing which index
> the actual boundary-segment has and lists the coordinates
> of the points that span the boundary face (hopefully they are
> printed in the right way):
[...]
> B.C. seg. index:  5
> Coordinate of 0 is: 0 0.1 0
> Coordinate of 1 is: 5 0.1 0
> Coordinate of 2 is: 0 0.1 0
> Coordinate of 3 is: 5 0.1 1
If this really is the output, then there is a severe
problem, because for boundary segment 5 the same
vertex shows up twice.

> Checking with the domain from above, the insertion index chooses
> the tag for the boundary faces. So boundary segment face
> with tag 1 corresponds to insertion index 1 that fits with the domain,
> looking at the corresponding coordinates.
> Now, tag 3 should correspond to insertion index 2, but the coordinates
> behind that boundary face are not right, they don't fit with the picture
> of the domain. So the insertion index, the boundary tag and the
But not being consistent with the problem is not necessarily
an issue. Again, the question is, if the insertion index/segment
index is consistent with respect to the ordering in which you
inserted the segments.

> corresponding
> boundary face coordinates don't match in the right way.
Match to what?

Best,
Carsten

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <https://lists.dune-project.org/pipermail/dune/attachments/20200824/bd11b04e/attachment.sig>