Hydro vs. MHD in Hydro Problems

Debugging

The hydro tests weren’t working anymore in MHD so I set about working on fixing them. I started with the simplier sound wave test. After some digging, the new Arm debugger was very helpful, I found that the issue was in how I refactored the sound wave initial conditions. Since the values of each element of the right eigenvectors defaulted to 1 it was giving me tiny magnetic waves even when there was no magnetic field. Changing the default to zero fixed it and I added the sound wave test to the list of MHD tests, with some small tweaks to fix some bugs in the testing tools for those tests.

The Sod tests appear to not pass mostly becaus the hydro version uses the piecewise parabolic method in primitive variables (PPMP) and the MHD version currently only has piecewise constant in primitive variables (PCM) implemented. I’ve temporarily created an MHD only Sod test that uses PCM; that test currently passes with 1 or 2 ranks but not 4, I’ll figure that out next week.

To help with finding the issues I wrote a simple notebook to compare the hydro and MHD results which is available upon request.

Other

• Read Tacconi et al. 2020