[MITgcm-devel] [MITgcm-cvs] MITgcm/pkg/ecco CVS Commit

[gael forget]

Dear all,

just to document this somewhat surprising behavior of TAF.

The objf_gencost re-initializations that I removed, appeared to confuse TAF.
It was resetting cost_generic to zero (no adjoint routine!) unless SST or SSS cost 
CPPs were defined (leading to a correct cost_generic_ad for gencost as well). 
It does make some sense that these re-initializations interfered with TAF. I am
not sure why (or if) the re-initializations at the beginning of cost_generic may 
be a good thing (as they are already done in ecco_cost_init_varia.F).

Cheers,
Gael

On Oct 18, 2014, at 10:17 AM, Gael Forget <gforget at mit.edu> wrote:

> Update of /u/gcmpack/MITgcm/pkg/ecco
> In directory baudelaire:/srv/scratch/gforget/MITgcm/pkg/ecco
> 
> Modified Files:
> 	cost_gencost_boxmean.F cost_gencost_seaicev4.F 
> 	cost_gencost_sshv4.F 
> Log Message:
> 
> - cost_gencost_boxmean.F, cost_gencost_seaicev4.F : fix gencost
>  for case of undef SST costs (... see devel posting later).
> - cost_gencost_sshv4.F : remove remaining CPP brackets
>  (ALLOW_GENCOST_SSHV4_OUTPUT, ALLOW_SSH_MEAN_COST_CONTRIBUTION,
>  ALLOW_PSBAR_MEAN) and rely on variables instead (added
>  useEtaMean; used gencost_outputlevel) .
> 
> 
> 
> _______________________________________________
> MITgcm-cvs mailing list
> MITgcm-cvs at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-cvs