The clang tools PR (PR #211) has been merged! At the beginning of this week I rebased it for the other PRs that have been added and I wrote a script to help summarize the output of clang-tidy (
I think I’ve narrowed the issue with the Sod shock tube in MHD to the HLLD solver. The issues I’m seeing go away when I replace the HLLD solver with the HLLC solver as you can see in this plot.
The left four figures are heatmaps of the x=0 face of the density at time step 34 running with 1, 2, 4, and 1 MPI ranks respectively. The rightmost figure is the same time in the hydro simulation. You can see a clear pattern around the edge where the HLLD solver is generating different values than the rest o the slice when it shouldn’t. I haven’t found the issue yet but I’m working on it
The Pitt CRC has finished their Jenkins setup so we met with them this week about getting it setup. I think it’s setup to run and I just need to get the Jenkinsfile working.