Hello, I am trying to run the p4est Taylor Green Vortex test case on multiple processors but I am facing really weird instabilities. After the first iteration, all values of density, velocity and pressure are Nan. This is with the test case code untouched. I was wondering if someone had managed to made it work ?
After some experimentation it seems to come from the viscous fluxes or the Navier Stokes diffusion equation because all simulations and test cases with CompressibleEulerEquations are working fine but not the Lid Cavity test case. It may also come from the parallelization because the test cases with one process is running as expected.
So to clarify: Are you trying to run this with MPI? As of now, we have unfortunately not yet MPI support for the viscous solvers.
Yes, i have a mpi installation with openmpi and I followed the guidelines on the Trixi.jl documentation for parallelization but unfortunatly I am getting Nan values for the viscous test cases.
I also found that even without the CompressibleNavierStokesDiffusion3D, the Taylor Green Vortex doesn’t run more than a few dozens of iterations (20 before getting Nan values) after blowing up, do you think it could be related to my mpi ?
No, MPI for Navier-Stokes-Fourier is unfortunately not yet supported, only inviscid problems.
The relevant solvers carry the _parralel subscript.
Okay, thanks for the response !