Entropy filtering algorithms, e-func

Hello everyone,

I understand that entropy filtering algorithms work well for excitation capture, see these two papers(10.1016/j.jcp.2022.111501),

but I don’t understand what is the difference between e-func = numerical/physical in the latest version of PyFr?

In my calculations, I found that when using e-func = numerical to calculate divergence, switching to e-func = physical allows me to continue the calculation, perhaps due to the presence of strong excitations in the flow field? (Mach number is 3 at this time)

Regards, Wgbb.