[MITgcm-support] Re: obcs counters correctly defined??

[Suneet Dwivedi]

Hi Matthew,
Thankyou very much for your reply and for sending read write vector
files. I followed your suggestion and recompiled and rerun the code
and optim.x, however, still the problem is not fixed and I am getting
the same error on running the optim.x. Are you aware of any other
changes required in files in the optim/lsopt folder when obcs control
vectors are allowed? Can you suggest me other places in the model code
that may be related to the problem?
Thanks again,
Suneet

On Mon, May 12, 2008 at 5:50 PM, Matthew Mazloff <mmazloff at mit.edu> wrote:
> Hi Suneet,
>
>  less sure that this is a model problem, could be with the
> linesearch....anyway, here are my first thoughts...
>  1)  try using the attached (modified for singlecpuio) read and write vector
> scripts, as i think the packing routines use them
>  2)  recompile optim.x because you have added files and I'm not sure if it
> can get messed up by this
>
>  neither of these suggestions may fix the problem, sorry i cant be more of a
> help....but these are the first two things i would try,
>  Matt
>
>
>
>
>
>
>  On May 12, 2008, at 3:44 PM, Suneet Dwivedi wrote:
>
>
> > Hi Everyone,
> > When I tried to run the assimilation experiment with my adjoint code
> > using lsopt and optim, my optim.x run stops with the following error
> > message:
> >
> --------------------------------------------------------------------------------------------------------
> > forrtl: severe (67): input statement requires too much data, unit 20,
> > file /tmp/mitgcm_so_suneet/JHU_ctrl _MIT_CE_000.opt0000
> >
> --------------------------------------------------------------------------------------------------------
> > Earlier, when i was running the same experiment without obcs package,
> > optim.x gave no problem and ran absolutely fine. This error is coming
> > only after I start using obcs (N,S,E,W) controls.
> >
> > I checked optim_readdata.F and found that when obcs controls are
> > defined, it looks for the wet point counts at Northern, Southern,
> > Eastern and Western boundaries (nWetobcsnGlo, nWetobcssGlo and so on)
> > in the control file (JHU_ctrl _MIT_CE_000.opt0000) for optimization,
> > in addition to the nWetcGlobal, nWetsGlobal, nWetwGlobal and
> > nWetvGlobal.
> >
> > I noticed that in my STDOUT.0000 file, there are "global nWet C/S/W/V"
> > values for different vertical levels but no "global nWet obcs T/S/U/V"
> > values; instead I found "global obcs N/S/E/W T,S,U,V" and "global
> > SUM(K) obcsN/S/E/W T,S,U,V".
> >
> > I wonder is there any bug in the code in the way it writes the output
> > of the obcs files, especially, in the nomenclature, in which it
> > denotes the number of wet point counters at different boundaries. What
> > may me the other reasons for the abovesaid error message? Please help
> > me resolve this problem.
> > Hoping for reply,
> > Cheers,
> > Suneet
> >
>
>