Home Debugging & ATPESC

Debugging & ATPESC

MHD Debugging

I found and fixed the last hydro but in my MHD code this week. There was a yid-1 when it should have been zid-1 in the PCM reconstruction function. After that fix, and thorough testing, hydro problems now work fine when run with the MHD algorithms. There’s still bug(s) in the magnetic field part but since it’s only when magnetic fields are on and the HLLD solver has been fully tested the bug is likely in the CT electric fields kernel, which isn’t a suprise due to its complexity. I’m working on tracking that bug down.

I also modified the 3D constant test case to include magnetic fields and tested that it’s actually constant using Athena++. The MHD constant test case is the same as the hydro one but with \( \vec{B} = \left[ 1e-5, 2e-5, 3e-5 \right] \).

I found a second bug that now makes the constant MHD case work. The VL+CT Algorithm requires at least 2 fully working ghost cells that have magnetic fields on both sides. That requires that we have at least 3 ghost cells in Cholla so that both edges have two fully working ghost cells. I added some logic to double check and set that and now the constant MHD case works. Compared to Athena++ it computes exactly the same time step and runs in exactly the same number of time steps. I then tested the Brio & Wu shock tube and that isn’t working yet. However, it looks like that is mostly due to a large oscillation in the \( B_z \) field so hopefully that is a simple sign error that I need to track down.

Argonne Training Program for Extreme Scale Computing (ATPESC)

I found out this week that I got into the Argonne Training Program for Extreme Scale Computing (ATPESC) which is a summer school for HPC that is in August. I spent some time this week getting paperwork sorted out for that.

This post is licensed under CC BY 4.0 by the author.