Timestep in GRMHD simulation

[Santhiya-PS]

For the GRMHD simulations, the documenation mentions that the size of a timestep is set by the shortest light crossing time in the grid. It does not change even if the sound (hydrodynamics) or fast magnetosonic (MHD) speeds are considerably slower than unity.Is there a specific reason for enforcing this stricter (light-speed-based) CFL condition?

In other codes, the fast magnetosonic speed is used to determine the timestep, which often allows for larger time steps and still maintains numerical stability. Such choices seem to serve their purpose well.

Could someone please clarify the rationale for this more conservative approach in Athena++?

[tomidakn]

Unfortunately, the developer who wrote this part already left the field, and I myself do not use relativistic hydro, so I am sorry but I do not really know. I guess, however, he implemented it as the safest choice, as you pointed out. Feel free to modify the code and relax it (Hydro::NewBlockTimeStep in src/hydro/new_blockdt.cpp).

[smressle]

I am guessing for simplicity. Using the local speed of light is much simpler than having to calculate the MHD wave speeds and for most "standard" GRMHD problems it doesn't matter because the fast magnetosonic speed almost always reaches ~ c somewhere in the region with the smallest cell. But obviously that is not always true.