-
Notifications
You must be signed in to change notification settings - Fork 35
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
providing marked_vertices
for InterfaceCoupling decreases robustness
#438
Comments
Interesting result! (From my current understanding, I would have expected the boundary vertices to be enough.) @kronbichler @peterrum maybe we could exchange ideas / discuss this topic in person, given that we had to adapt the functions finding (all) active cells around points in deal.II already in the past. My opinion is that |
To me it looks like that there are functions that use the vertices of the triangulation directly instead of the mapped ones, which would be returned by calls There are other instances in the namespace I guess why providing no |
I tried to figure out where exactly we are arriving and ended up in this function here: https://github.com/dealii/dealii/blob/d47537fa246886e5308d5aecbf42358d73a00774/source/grid/grid_tools.cc#L2927-L2945 where I could not find the function in
I might post more if I see more later. |
I guess I pointed out the wrong |
I agree, there are too many functions with similar tasks and too many wrinkles in the code that jump back and forth. It's just that I do not see a simple solution, either. What I would like to do, in case we identify the problem, is to check if the overall complexities are ok or if there is something to be done. I fear that those functions do a lot of useless work sometimes, but it is hard to look at it without a specific hint. |
I think the easiest approach for testing would be to take @richardschu Does this make sense to you. Could you try this out? |
Thanks for the help everyone! I think we managed to resolve that problem. In all my tests, the discretization error in the map compared to the tolerance is the cause if RPE is not succesful in finding any points. This is expected behavior and with a PR I will push to deal.II soon, this problem is fixed. |
So I suggest we close this issue if you agree? |
Yes, this sounds good to me. I am relieved that things work as they should now (at least when the comments are fixed). Should we wait with closing until the relevant fixes are in deal.II? |
yes, let's wait until the problem is really resolved, i.e. dealii/master has been updated accordingly. |
this issue seems to be resolved now? |
Working on #433 I realized, that omitting the
marked_vertices
allowed finding all the points on the interface (using the test presented in #433). Providing the vertices is optional though, and the Docu from RPE notes the argument functionmarked_vertices
as a"Function that marks relevant vertices to make search of active cells around point more efficient."
get_marked_vertices_via_boundary_ids
marks vertices on faces depending on some boundary ID, which are passed to InterfaceCoupling::setup().I would argue that the "relevant vertices" are the ones on the whole cell -- depending on the implamentation details which I did not check (yet). Maybe the prarallel layout plays a role as well ... I will adapt the vertices in an alternative function and let you know.
The text was updated successfully, but these errors were encountered: