search

AMReX-Astro / Castro

Castro (Compressible Astrophysics): An adaptive mesh, astrophysical compressible (radiation-, magneto-) hydrodynamics simulation code for massively parallel CPU and GPU architectures.
http://amrex-astro.github.io/Castro
Other
301 stars 98 forks source link

is EOS call in MHD HLLD needed? #955

Closed zingale closed 4 years ago

zingale commented 4 years ago

In HLLD, we come in with qleft/qright, and then we convert to convert to conservative variables and compute the fluxes from the primitive state.

For the primitive state, we call eos(eos_input_rp, ...) but in every case I am aware of, we already have a valid QREINT in the primitive input state (either from the reconstruction or in the ct_update where we did a constoprim call before the Riemann.

So I think these EOS calls can be removed.

It is also the case that in a lot of places we already have the conservative state that we could send in as input, potentially eliminating the need to do the PtoC calls here (although we'd still need to compute the pressure for the flux)

guadabsb15 commented 4 years ago

we can definitely remove the EOS call from PtoC, but I think the other two we might still need them for gam1? But we could probably get that, on the plm (when we get q(i,j,k,QREINT)) and on the ctu routines that convert from conservative to primitive

guadabsb15 commented 4 years ago

If we were to remove PtoC, and send the conservative states, would we compute the pressure using an EOS in hlld? If that is the case, by doing that could we lose some information from the reconstructed pressure coming from the plm?

zingale commented 4 years ago

indeed, so perhaps we need to think more carefully about this.

zingale commented 4 years ago

some of this cleaning was done in #969