Low-Mach correction for HLLC solver

[BenWibking]

Includes the low-mach correction for the pressure in the Riemann solver. Does not include carbuncle fix. Based on Minoshima & Miyoshi (2021).

This is an updated version of PR #57.

Closes #18.

[BenWibking]

@pgrete @forrestglines I'd like to get this merged in, so I can run my precipitator simulations with upstream AthenaPK and Parthenon. Please let me know what you need from me to do this.

[pgrete]

This is not the full implementation of the LHLLC solver, i.e., it just contains the fix for the pressure at the contact surface and not the shock fix (based on the velocity divergence), isn't it?

[BenWibking]

This is not the full implementation of the LHLLC solver, i.e., it just contains the fix for the pressure at the contact surface and not the shock fix (based on the velocity divergence), isn't it?

Yes, that's right.

[pgrete]

Given that this boils down to a 5 line change (with regard to the HLLC solver), I'm wondering if there's a smart (constexpr if way?) around this to reduce the code duplication.

Otherwise, it'd be great if

• the current version is mentioned in the docs (with a note that it includes only one part of the solver described in the paper)
• there's a Changelog entry

[BenWibking]

Given that this boils down to a 5 line change (with regard to the HLLC solver), I'm wondering if there's a smart (constexpr if way?) around this to reduce the code duplication.

The template metaprogramming to do this while keeping it as a runtime parameter is not trivial. For instance, this is the AMReX implementation: https://github.com/AMReX-Codes/amrex/blob/b78849057c0f533466da6b517a59d2eddc99c1a8/Src/Base/AMReX_CTOParallelForImpl.H#L143

I can add it to CHANGELOG + docs.

[pgrete]

Given that this boils down to a 5 line change (with regard to the HLLC solver), I'm wondering if there's a smart (constexpr if way?) around this to reduce the code duplication.

The template metaprogramming to do this while keeping it as a runtime parameter is not trivial. For instance, this is the AMReX implementation: https://github.com/AMReX-Codes/amrex/blob/b78849057c0f533466da6b517a59d2eddc99c1a8/Src/Base/AMReX_CTOParallelForImpl.H#L143

Interesting. Might be worth to keep that in mind for the futures.

I can add it to CHANGELOG + docs.

Yes, that'd be great.
And then also a tiny test case (I suggest one more here https://github.com/parthenon-hpc-lab/athenapk/blob/main/tst/regression/test_suites/convergence/convergence.py#L37 and one more here https://github.com/parthenon-hpc-lab/athenapk/blob/main/tst/regression/test_suites/riemann_hydro/riemann_hydro.py#L27C28-L27C69 mirroring the "fiducial" VL2/PLM case).