Closed lconlon closed 1 year ago
@cbegeman, this is a placeholder until we figure out how to have shared yaml options
We ran cosine bell with this branch on anvil and it ran successfully
@lconlon Great. Let me know when this is ready for review. Seems like you might have changes to other files pertaining to the vertical grid that aren't added yet
@lconlon, it seems like @cbegeman's right. I looks like the changes to cosine_bell.cfg
didn't get committed and pushed to this branch yet. We'll do that tomorrow.
@cbegeman, we're waiting on #57, regardless so no need to review yet. These changes will be updated once that goes in.
We re-tested with the recent changes (and rebase) and the tests still pass.
@lconlon and @xylar I just wanted to make you aware that this commit doesn't allow users to run single layer with debug tracers https://github.com/E3SM-Project/E3SM/pull/5590/commits/73b90f7a392587840855cc6a05945a23e0209196. Here's the relevant comment: https://github.com/E3SM-Project/E3SM/pull/5590#issuecomment-1522526731.
I'm not quite sure why this case works with debug tracers but dam_break did not. I guess we should change MPAS_LOG_CRIT
to MPAS_LOG_WARN
.
@cbegeman, thanks for the heads up. I guess the current submodule must be from before this change or we wouldn't have been able to run cosine bell. Since there aren't any dynamics other than advection, it could be that this test just doesn't trigger whatever problem you saw in https://github.com/E3SM-Project/E3SM/pull/5590#issuecomment-1522526731.
It does seem like the change in https://github.com/E3SM-Project/E3SM/commit/73b90f7a392587840855cc6a05945a23e0209196 would make it impossible to run the cosine bell in single layer. I don't know how much time we want to spend investigating debugTracers
in RK4 but maybe we could allow debugTracers
when config_disable_tr_all_tend == .false.
but disallow them when config_disable_tr_all_tend == .true.
. We aren't disabling tracer tendencies here (because tracer advection is the whole point of the test case).
Maybe this is a combination you've tried already and caused crashes. Let's chat more about this tomorrow.
@xylar I've already tried that combination. I think the best path forward is just to change MPAS_LOG_CRIT
to MPAS_LOG_WARN
and continue disabling debug and active tracers for the wetting and drying tests.
Thanks @cbegeman. I can make the PR to E3SM if you like.
@lconlon I think the documentation for this test case does refer to 3 layers in the vertical_grid
section. If so, that should be updated.
@cbegeman, thanks for catching that. I believe I have it updated now, but let us know if I missed any multilayer references that you had seen.
@cbegeman, thanks for catching that. I believe I have it updated now, but let us know if I missed any multilayer references that you had seen.
That's the one I had in mind!
I'll update the E3SM submodule and re-test, then I hope we're good to go on this.
I ran both cosine_bell
test cases and they passed with this branch:
Test Runtimes:
06:26 PASS ocean/global_convergence/qu/cosine_bell
04:54 PASS ocean/global_convergence/icos/cosine_bell
Checklist
Testing
comment in the PR documents testing used to verify the changes