How to set-up a wave already inside plasma (eg Alfven wave).

[lyutikov-bot]

I am less concerned with shifted spacial grid for particles and fields, more how the initial drift velocity is fed into Maxwell’s equation. For example, for subluminal waves (eg Alfven) there is a frame where E-field is zero. In that frame curl-B=j. But since B and j are calculated at different times, I cannot just specify initial analytical expressions for B-fields and currents (as product of density and drift velocities), and expect that E-field remains zero - it will not.

[lyutikov-bot]

The problem is: initial velocity is not used for current computation until the first update is done. So, instead of curl-B=J (and E=0), the code treats J=0 and curl-B= d_t E, generating vacuum-like E-field. In principle, this can be fixed with a half-step–back of velocity. That's tricky. Or: can the initial velocity be included as a current?