I fixed the bug in the rotation when generating rotated waves and now I believe grid aligned waves in any direction are working. Non-grid aligned waves are trickier. Since the initial conditions aren’t lined up as neatly it gives a non-zero divergence unless you initialize with the magnetic vector potential or perform divergence cleaning. I’ve set this issue aside for now and am just doing grid aligned waves but will likely implement one or the other in the future. For some reason I get slightly different timesteps when running in the different directions. The results are still correct though so I’m not worried about it at this time.
I added a test for MPI versions of the slow wave and it is working correctly.
I have been using periodic boundary conditions for the linear waves. I tried the linear waves with reflecting and transmissive boundaries and both led to non-zero divergence’s before the first timestep. I’ll work on this next week.
- Present on Shankar et al. 2022