[MITgcm-support] rbcs and adjoint

Patrick Heimbach heimbach at MIT.EDU
Wed Jul 12 11:19:49 EDT 2006


Hi,

just to close this thread, 
problem is fixed
(adjoint level-N checkpointing 
was not implemented for pkg/rbcs/).

-Patrick



On Mon, 2006-07-10 at 14:01, Patrick Heimbach wrote:
> Hi Lars,
> 
> that sounds like an extensive recomputation problem,
> likely a part of the rbcs state isn't properly
> stored for lev3 and lev2,
> a guess in the absence of availability of your config
> (would need to look into taf_ad.log).
> 
> Best way to test and fix is if you could send me
> a self-consistent code/ dir. with usual
> header files, and I'll look into it.
> 
> -Patrick
> 
> 
> 
> On Mon, 2006-07-10 at 13:35, Lars Czeschel wrote:
> > Hi all,
> > 
> > Does anybody use the rbcs package in the adjoint model?
> > If I use rbcs the model writes out tapelevel3 as normal,
> > but afterwards the model seems to be trapped in an 
> > infinite loop, i.e. the whole forward run is repeated again
> > and again without writing out tapelevel2.
> > The same setup works fine without rbcs.
> > 
> > Thanks for your help,
> > Lars
> > 
> > --------------------------
> > Lars Czeschel
> > Department of Meteorology
> > University of Reading
> > PO Box 243
> > Reading RG6 6BB
> > Phone: +44 (0)118 378 7997
> > 
> > 
> > _______________________________________________
> > MITgcm-support mailing list
> > MITgcm-support at mitgcm.org
> > http://mitgcm.org/mailman/listinfo/mitgcm-support
-- 
--------------------------------------------------------
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/                        USA




More information about the MITgcm-support mailing list