search

mabarnes / moment_kinetics

Other
2 stars 4 forks source link

Feature gyroaverages #187

Closed mrhardman closed 5 months ago

mrhardman commented 7 months ago

To do list:

mrhardman commented 6 months ago

periodic_box_low_res.zip Low resolution runs with rhostar = 0.1 and box lengths of 1 in r and z. Neither run uses radial diffusion, but the run 2D-periodic-gk.toml uses a gyroaverage. Clear differences can be seen where the gyrokinetic run tends to have smoother features and smaller amplitudes in the peaks and troughs of the solution, whereas the run 2D-periodic-dk.toml obtains grid scale behaviour. A careful study of this 2D model might be useful for understanding the slab ITG instability. @johnomotani @mabarnes @LucasMontoya4. Further efficiency improvements in the GK operator are likely required to go to higher resolutions.

mrhardman commented 6 months ago

Marking this PR as ready for review as shared-memory MPI is now working, looking for feedback before implementing distributed-memory MPI support. @johnomotani @mabarnes

Edit: Some tests still seem to be failing on CI...

mrhardman commented 6 months ago

Commit https://github.com/mabarnes/moment_kinetics/pull/187/commits/cd85aa7ae8cdc5bf66340eb726dabe6238e74a1c passes all tests apart from tests in parallel on MacOS, which mysteriously time out after seeming to finish with the tests reportedly passing (although a different number of tests seem to be carried out as compared to the Ubuntu case.

johnomotani commented 6 months ago

Commit cd85aa7 passes all tests apart from tests in parallel on MacOS, which mysteriously time out after seeming to finish with the tests reportedly passing (although a different number of tests seem to be carried out as compared to the Ubuntu case.

The macOS parallel tests do this fairly frequently. Often just re-running the jobs lets them pass, so I've not investigated further.

The macOS parallel tests were being slow, so I cut them down to 1 test run (4 process, no --long). On the Linux job, we test 4 processes, 3 processes, and finally 2 processes. On the 2-processes test we use the --long flag - that's why there will be some more tests reported for that run than there are in the macOS job.

mrhardman commented 5 months ago

This issue can act as the tracker: https://github.com/mabarnes/moment_kinetics/issues/186

mrhardman commented 5 months ago

It looks like some of the parallel tests might have been broken, rerunning the tests as I don't understand how that can be.

johnomotani commented 5 months ago

It looks like parallel tests sometimes (often?) fail and sometimes pass, which is a bit suspicious. It would be good to add a 'debug check' that uses the gyroaverage - I'll have a look at that now.

johnomotani commented 5 months ago

Ah, the PR tests (in parallel) were failing because of changes from #203 (which had been merged into master. I've fixed that now (I think), but the CI runners are failing to install OpenMPI. I don't think that failure is anything to do with us, so hopefully it will sort itself out in a little while.