You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@pgrete FYI (not sure if you've seen it), first order flux correction is implemented to fix troubled cells. @BenWibking Yes, I've seen that. In my tests with Quokka, the shock flattener appeared necessary to suppress some post-shock oscillations in strong shocks. I remember this being a particular issue with the Zel'dovich spike in radiative shocks. Do you see any of these kinds of issues currently? I didn't do characteristic reconstruction in Quokka, so that might be a difference.
@pgrete I see. I didn't think of the post-shock oscillations. For those the flattener might be the better choice.
@BenWibking Is first-order flux correction implemented for RK2/RK3?
@pgrete We call first order flux correction in each stage at the moment (so not just at the very end). I'm not sure which is the better approach, but I went for this one as I think that "each stage" might be better because no neighboring cells will be updated during an entire cycle with potentially nonphysical data from a previous stage. @BenWibking That's what I did in Quokka as well. If I didn't do this at the end of each stage, then in the later stages, the Riemann solver would produce NANs from the imaginary sound speeds. So this makes sense to me.
@BenWibking reported on https://gitlab.com/theias/hpc/jmstone/athena-parthenon/athenapk/-/issues/14:
It would be very useful to implement the shock flattener from Miller & Colella (2002): https://ui.adsabs.harvard.edu/abs/2002JCoPh.183...26M/abstract
Thread:
The text was updated successfully, but these errors were encountered: