Implicit solver for electrons (and more?)

lanl / Multi-BGK

Conservative multispecies kinetic equation solver

Other

4 stars 11 forks source link

Implicit solver for electrons (and more?) #3

Open berselius opened 6 years ago

berselius commented 6 years ago

One can treat the electrons as a species in the code. If one sets the input parameter ecouple=2 then the code treats the first species specially as electrons and uses different collision rates derived from Landau-Spitzer and/or NRL. However, the timestep requirements are excessive in this case. If we can do a semi-implicit solve for the electrons we can step over that restriction. Simple lagged collision rates have proved to be unstable.

berselius commented 5 years ago

A linear (lagged) implicit solve has been implemented. Caveats include

Only works for first order solve in time/space
Will not work with collision rates specified from MD
I've seen some squirrley-ness when electric fields are on - probably CFL related

Still to do is integrating the nonlinear solve, which needs to call collision rates