I don't do plasma physics but other sorts of HPC numerical physics. The answer is a myriad of ways: Check small numerical cases on an already-validated code.

I don't do plasma physics but other sorts of HPC numerical physics. The answer is a myriad of ways: Check small numerical cases on an already-validated code. Check problems for which you have an analytic solution (in this case, maybe many laminar flow scenarios). Check problems where the 'weird part' (ie the turbulence in this case) is limited to one small part of the simulation. Compare only moderately complicated simulations to experiments that are already entirely understood.