[MITgcm-devel] testing restart and initialisation

[Jean-Michel Campin]

Hi,

Using my script "do_test_2+2" and "tst_2+2" (both checked-in in
 MITgcm_contrib/jmc_script ), I was able to check the restart
process for almost all our verification experiments.
The 2 exceptions are tutorial_cfc_offline & fizhi-cs-32x32x40
for which the restart is not "standard" (from my point of view).
The 2 scripts (do_test_2+2 & tst_2+2) are now easy to use: a copy 
of both in MITgcm/verification and after running a full testreport, 
just typing "do_test_2+2" will check the restart of all exp.

most of the experiment give a perfect restart, except:
 dome & tutorial_plume_on_slope <- orlanski
 fizhi-cs-aqualev20             <- ?
 tutorial_global_oce_biogeo	<- but only dic fields: pH computation ?

Nothing to do with restart:
I also run a testreport checking for un-initialized variables
(with all the tricks Constantinos gave me):
Apart from 3 exp. (fizhi*) which fail to compile, and MLAdjust.1.leith
which requires MNC to start, only 4 experiments stop with a 
"Floating point exception":
 global_ocean.cs32x15.viscA4
 hs94.cs-32x32x5
 hs94.cs-32x32x5.impIGW
 tutorial_held_suarez_cs
and I found this not too bad. Will try to investigate those 4 cases.

Cheers,
Jean-Michel