Post

Particle Debugging & MHD Prep

Posted Updated
By Robert Caddy 1 min read

Particle System Tests

I finished hunting (I hope) all the particle related bugs in the system test runner class and found some bugs with how particle IDs are currently implemented in Cholla. So I submitted a PR and got my stuff merged into Cholla and will do more thorough testing once Cholla supports global particle IDs on GPUs.

MHD Prep

I’m now preparing to implement MHD into Cholla, starting with the HLLD solver. To this end I spent some time reviewing the HLLD solver, and Riemann Solvers in general, wrote a bunch of the boilerplate code that MHD will require, updated my MHD GitHub Issue with current plans, and started prepping testing for the future HLLD solver.

Getting known correct data from the HLLD algorithm is tricky since it’s not exact so we can’t rely on analytic results; therefore I need a known good HLLD solver to compare with my HLLD solver. I extracted the HLLD solver from Athena1 and built a program to call it with different values and print the results at full precision. To find reasonable inputs I decided to use values from various MHD Riemann problems and so researched and wrote a blog post on them: MHD Riemann Problems. Next I plan to use Athena++ to simulate all the Riemann problems then use L/R values that span each wave as my test cases. The extracted HLLD solver, results from Athena++, and analysis notebook can all be found on in my hydro sandbox.

Other

Footnotes

  1. I used the older Athena HLLD solver since Athena++’s HLLD solver is much more tightly integrated and depends on many different classes that I didn’t want to risk trying to mock up. The Athena HLLD solver is less tightly integrated and I could easily mock or copy the structs and functions required. ↩

MHD, Testing
Testing Particles HLLD
This post is licensed under CC BY 4.0 by the author.