Mass is not conserved for reflecting boundary conditions - Sod tube problem

[mugalino]

Hello! We are currently doing some tests with Parthenon in another code in development and thought to test AthenaPK as well if the problem occurs there. It seems that both mass and energy is not conserved for a Sod tube problem with periodic boundary conditions along x2 and x3, and reflecting along the x1 direction. We also encounter this problem for an RT test we wrote for a private code we have where we set reflecting boundary conditions along the x1 direction as well.

We compile the code using these commands
cmake -DKokkos_ENABLE_OPENMP=off -DKokkos_ARCH_BDW=on ..
and
make -j

and run the Sod tube problem with the parameter file attached. Is there a fix for this problem or is it an unresolved issue for Parthenon? I run with commit 540deb9 in the main branch.

Parameter file used
sod.txt

Run with ix1_bc and ox1_bc set to periodic
parthenon_periodic.txt

Run with ix1_bc and ox1_bc set to reflecting
parthenon_reflecting.txt

Any insight about this would be really helpful!

[BenWibking]

There is a partial fix in this work-in-progress PR: #58