Imperfect restarts again, with OBCs
Closed this issue · 8 comments
The DEBUG information is pointing to uh and vh at Beginning of step_MOM. We must be missing more OBC code during initialization.
There is complicated logic in initialize_dyn_split_rk2 in which it sometimes initializes uh and vh, but only if CS%store_CAu is false or if it didn't read CAu and CAv from the restart file. The OBCs need uh and vh on restart as well.
Oh, and if you have non-zero initial velocity from a file, the first call to continuity during initialization spawns those redundant errors for uh, vh. It's during a restart that continuity isn't called at all, which may or may not be related to the pre-btstep accel DEBUG differences.
If the uh vh stuff is a red herring, surely a Post OBC avg u,v is not. Arg.
The phase speeds for the time filter of the radiation conditions are written to the restart file, but apparently aren't being read in. I see the logic for reading in CAu, CAv. How do I tell it to only read on restarts?
Actually, the restarts were working when I had DIABATIC_FIRST = False. Does it make sense that some restart fields would be read one way, not the other?
I watched in the debugger - and it behaved itself. This is with gnu 11.3.0.
Kate, was this issue addressed by PR #520?
Yes, it should be fine now.