[Mitgcm-support] Re: butterflies

[mitgcm-support at dev.mitgcm.org]

Re. Ralf's "butterflies":

Ralf Giering wrote:

> Hi,
> 
> Working with ecco_c44_e22 I have found some butterflies by compiling and
> running the code with array boundary check. See the list below.
> 
> You might want to add an array boundary check as a standard test case of the 
> code.
> 
> Ralf
> 
> 
> Code version: ecco_c44_e22
> ==========================
> - iikey is set in thermodynamics but also used in dynamics
>   it should either be defined in ecco_the_model_main
>   or thermodynamics and dynamics should use different key variables.


This key is never used so didn't cause problems.
I'll fix it nevertheless, "just for the heck of it" as they say.


> - tmpvar used but not defined in ctrl_unpack.F


Is a dummy variable so doesn't cause problems.
I'll fix it nevertheless, "just ...


> - using undefined values of kapparv in mom_v_bottomdrag
>   mom_v_bottomdrag must have iMin,iMax,jMin,jMax as arguments
>   and only use values being calculated in calc_viscosity
>   There are probably many more routines in mom_fluxform and mom_vecin
>   using undefined values.


This is what caused the problems with the adjoint
of kapparu, etc. a few months ago.
We fixed it by extending the setting of the background
to the full overlap. That's a quick and dirty fix.
Maybe the jmin, etc. can be sorted out


That's it.

Patrick

-- 
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
Patrick Heimbach     Massachusetts Institute of Technology
FON: +1/617/253-5259                    EAPS, Room 54-1518
FAX: +1/617/253-4464               77 Massachusetts Avenue
mailto:heimbach at mit.edu                 Cambridge MA 02139
http://www.mit.edu/~heimbach/                       U.S.A.