Particle RNG issue?

[brryan]

@Yurlungur's excellent tracer particles plot from #830 shows repeating particle positions which I bet indicates an RNG issue in the underlying particle solver, maybe per-block or something.

[Yurlungur]

@brryan probably it's in pgen for the particle tracers example. I think the rng seed may actually hand-fixed.

[brryan]

Ah yes you're right looks like each MPI rank gets its own seed. Well that's an easy fix (I can do this).

[pgrete]

I agree that RNG should be problem dependent. For example, in my (downstream) turbulence driver it was important that every rank has an identical seed (and rng sequence) and I had to dump its state to the restart outputs in order to get reproducible restarts.