I finished up implementing MHD boundary conditions with some advice from Alwin. To test it I learned the basics of how to use cuda-gdb on the command line which was actually a lot easier than expected, unfortunately required though since the cuda-gdb VS Code integration is very poor at the time of writing. I implemented the MHD boundary conditions by, if
MHD is defined, generating two indices; one for the hydro variables and one for the magnetic fields. Then I loop through the fields and choose which index to use based on which field is being copied or modified.
As we add more systems tests and their associated data the main Cholla repo could end up being fairly large which is an issue for non-developers who want to clone the repo and run simulations. To alleviate this I’m working on making a seperated repo, cholla-tests-data, that is a submodule of Cholla and uses Git LFS to manage the HDF5 files. All the planning and research for that is done, I’m just working on implementing it now. I did look into using DVC but at this time I think it’s overkill for what we’re doing.
- Prepped for the testing hack day I’m leading this week
- Fixed an issue with MPI not installing properly on GitHub Actions builds
- Some reading on dust polarization