Stop running 0th time step #2084
Open
Task list completed / task-list-completed
Started
2024-11-22 02:25:18
ago
1 / 4 tasks completed
3 tasks still to be completed
Details
Required Tasks
| Task | Status |
|---|---|
| Generating new baselines and rerunning the mosart/rtm test suites, because he expects no diffs from the code changes to those components. | Incomplete |
| A F2000 simulation as confirmation that all this works correctly in coupled mode now that CAM has eliminated the 0th time step, as well: | |
| Updated my cesm to cesm3_0_alpha05a | |
./create_newcase --compset 2000_CAM70_CLM60%SP_CICE%PRES_DOCN%DOM_MOSART_SGLC_SWAV --res ne30pg3_t232 --case /glade/u/home/slevis/cases_LMWG_dev/f2000.ne30_t232.SP --run-unsupported |
|
./case.build currently returns this error (even after the Forums suggested conda activate ctsm_pylib): |
|
ERROR: Cannot modify case, read_only. Case must be opened with read_only=False and can only be modified within a context manager |
|
| Creating the case from Cecile's checkout of cesm3_0_beta04 WORKED, so now I need to use the case's /SourceMods. | |
| BUT it complicates things that this tag points to ctsm5.3.007, mosart1.1.02, rtm1_0_80. | |
| I now cloned my own checkout of cesm3.0-alphabranch, which points to the same component versions as Cecile's. | Incomplete |
| 2-day versus 1+1-day simulations are b4b. | Incomplete |
| branch? | Incomplete |
| Answers changed from the baseline due to removal of 0th time step (expected) | Incomplete |
| Answers also changed in restart (or branch) but just for one field described as constant | Incomplete |
| Answers in the next PR, which includes the code from this PR, didn't change and restart passed | Incomplete |
| So I'm inclined to ignore the answer change in restart in this PR for now | Incomplete |
| Great, thanks for this change. I agree with this and we can remove the above BACKWARDS_COMPATIBILITY comment. | Completed |
| Generally: As you know, this PR removes references to the 0th time step | Incomplete |
| Specifically: Keith removed the ".and. (nstep /= 0)" part of this if-statement | Incomplete |
| In the latest version, the if-statement became "time_to_reset" and the nstep became "effective_nstep" | Incomplete |
| My question: Is it ok that I removed the effective_nstep part of the line, the same way that Keith originally removed the nstep part? | Incomplete |
I'm pretty sure this change should be reverted. effective_nstep refers not to the number of timesteps since run start/restart, but rather to the number of timesteps since the accumulator field was initialized/reset. |
Incomplete |
Loading