You have an issue in your boundary worker.
The boundary worker is called passing a cell, a face_no, scratch, and copy.
Yet, in your code, you loop again on the faces of the cell:
for (const auto &face : cell->face_iterators())
{
if ( ((face->boundary_id() == 0) || (face->boundary_id() == 1)))
this should become
if((cell->face(face_no)->boundary_id() == 0) ||
(cell->face(face_no)->boundary_id() == 1) {
…
}
otherwise, you are going to assemble the face terms twice on the four corners.
Best,
Luca.
> On 11 Jan 2023, at 12:05, Abbas <abbas.ballou...@gmail.com> wrote:
>
> Hello,
>
> Setting:
> 1) Unit square domain,
> 2) Simple scalar Laplacian with no source term.
> 3) 0 and 1 Dirichlet on opposite ends of the square and zero Neumann on the
> other two edges.
>
> I am trying to solve the above problem using SIPG by adopting step 74. The
> expected solution should be perfectly linear with a constant gradient as the
> problem reduces to a pseudo 1-D problem.
>
> That is the case if I use CG, but with DG I do not get a perfectly linear
> solution and that is causing me problems.
>
> I have attached below a single image file that includes:
> 1) The solution.
> 2) The solution gradient with weird behaviour at the corners
> 3) Line plot of the solution across the lower wall.
>
> and the code I am using which is based on step 74.
>
> Things I tried/looked into:
> - I tried to adapt the code to CG and I get the expected results. I can't
> straight up impose BCs strongly with DG here.
> - I am sure this isn't a Paraview issue because I tried integrating across
> the boundary within deal and I am getting mesh dependent results.
> - The weak form is correct. The Neuman BC shows up as a boundary integral in
> the rhs which has no contribution. (Neumann zero at these walls).
> - I looked into the distributed LDG method within the dealii gallary and
> adopted it to these BCs and I didn't have this problem.
>
> The only thing I haven't tried yet is to do a simple problem by hand and
> compare the dealii and my outputs. Before doing that, any ideas on how I can
> debug this? Is this something associated with imposing BCs weakly with SIPG
> and will happen no matter what?
>
> best,
> Abbas
>
>
>
> --
> 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/df72814c-0cdc-4215-819c-46f5a8d05e84n%40googlegroups.com.
> <Screenshot from 2023-01-11 12-39-11.png><SIPG.cc>
--
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/09F64793-8B26-4AD7-9B65-799A6CE3133C%40gmail.com.
