Mfem: SU/PG implementation

[tbarani]

Dear MFEM developers,
I am writing you to ask your opinion on the feasibility of implementing the streamline upwind PG stabilization for standard G-FEM. I am working on the solution of the advection/diffusion equation via the continuos G-FEM, which is known to be prone to instabilities. For a number of reasons, we want to stick at classic continuous G formulation, so I cannot exploit the DG method for a more stable solution.

Do you have any suggestion/recommendation/example from which I could start to approach the problem?

Thanks in advance!

BR
Tommaso

[tzanio]

I believe @tangqi and @KazemKamran in https://github.com/mfem/mfem/tree/pumi-NS have implemented SUPG and may be able to provide advice.

Other users and developers may also be able to help.

[tangqi]

Yes, I implemented SUPG for my MHD system following that pumi-NS branch. Most SUPG operators are already available at https://github.com/mfem/mfem/blob/pumi-NS/fem/NSnonlininteg.cpp

The difference between my branch and the pumi-NS branch is I did Jacobian by myself. But that pumi-NS repo should be a good starting point for your purpose. Let me know if you have more questions

[tbarani]

Thanks a lot guys! I will have a look into this branch and get in touch if needed. I am closing the issue for the moment.