Missing diffusive fluxes in FOFC

[jmstone]

In GitLab by @changgoo1 on Jan 20, 2023, 09:43

In Hydro::FOFC and MHD::FOFC, the fluxes of a cell that hits floors are replaced. When diffusion processes are turned on, the original fluxes include the fluxes from the diffusion processes. So, this has to be added back for consistency.

[jmstone]

In GitLab by @changgoo1 on Jan 20, 2023, 09:51

In Athena++, it was pretty straightforward as it keeps diffusive fluxes in separate arrays. It seems that this is not the case here. I feel that creating inline functions for single state diffusive fluxes may be one way to solve this.

[jmstone]

In GitLab by @pdmullen on Jan 20, 2023, 10:01

Nice catch! What if we moved

  // Add viscous, resistive, heat-flux, etc fluxes
  if (pvisc != nullptr) {
    pvisc->IsotropicViscousFlux(w0, pvisc->nu, peos->eos_data, uflx);
  }
  if ((presist != nullptr) && (peos->eos_data.is_ideal)) {
    presist->OhmicEnergyFlux(b0, uflx);
  }
  if (pcond != nullptr) {
    pcond->AddHeatFlux(w0, peos->eos_data, uflx);
  }

to after the FOFC call? If we elected for this option, then diffusive fluxes would not inform the FOFC prediction (only the standard non-diffusive fluxes would). As a point of comparison: in GRMHD, we do not include coordinate source terms in the prediction of whether a cell will be floored or not.

[jmstone]

In GitLab by @changgoo1 on Jan 20, 2023, 10:05

Yes, that can be an option. The current FOFC prediction is anyway not the perfect prediction as it misses a few source terms.