[MITgcm-support] Re: MITgcm-support Digest, Vol 10, Issue 12

[Ed Hill]

On Mon, 2004-04-19 at 17:17, sju at eos.ubc.ca wrote:
> Dear MITgcm support,
> 
> I will try to refrase my last question. I downloaded the new version of
> MITgcm, did some changes for my particular problem (modification of a
> mixture between exp 4, and flt_example), like in I had in the old version,
> changed CPP_OPTIONS.h, similar to what I see in exp 4, and compiled it.
> Although, the compilation is different, I manage to do it with no errors.
> Now I run the model, but it goes unstable very fast. Could this be due to
> the way I do the compilation:
> 
> ../tools/genmake 2 -makefile
> make depend
> make
> 
> I changed pkg_default to include the flt package, but I let genmake2
> choose the build options file by itself.
> 
> Or should it be due to some new parameters in this version?
> 
> I am puzzled, because I have the same settings and the old version using
> genmake runs perfect.


Hi Sergio,

Code in "pkg/flt" and "verification/flt_example" have not been tested
continuously through the past few months.  Its possible that some bugs
or slightly different behavior have been introduced.  But then other
explanations are more likely (see bellow).

So the first thing to check is whether the "flt" package itself is
working with the latest code.  Using the latest MITgcm from CVS, I ran
the following commands:

  cd MITgcm/verification/flt_example
  mkdir run
  mkdir build
  cd build
  ../../../tools/genmake2 -enable=flt -mods=../code
  make depend
  make
  cd ../run
  ln -s ../input/* .
  ../build/mitgcmuv 

and was able to build and run the "flt_example", which is good news.  I
then looked at the output using MatLAB:

  matlab -nojvm
    flts=read_flt_traj('float_trajectories');
    plot( flts(3).time, flts(3).x/1e3 )
    for k=1:126;plot(flts(k).x/1e3,flts(k).y/1e3);hold on;end
    hold off

and got what appeared to be reasonable results.  Note that I am not
familiar enough with the "flt" package to confirm that they are indeed
the correct answers, but they do look reasonable.  Could you verify that
the "flt_example" results with the latest MITgcm from CVS are the same
as your "old" version?  That would be a good test.  And I can't do it
because I don't know which version is your "old" one.

The next thing to try is your input files.  Are you sure that your code
modifications (especially "CPP_OPTIONS.h") are compatible with the
latest version of MITgcm?  A number of things have been moved out of
"CPP_OPTIONS.h" since the "genmake2-merge" as you can see from the diffs
at:

http://dev.mitgcm.org/cgi-bin/viewcvs.cgi/MITgcm/verification/exp4/code/CPP_OPTIONS.h

so you might want to double-check that file.

The last thing one can do is a file-by-file comparison between the two
versions to see what, exactly, has changed.  Thats a tedious process so
it should be a last resort.

Good luck!
Ed

-- 
Edward H. Hill III, PhD
office:  MIT Dept. of EAPS;  Room 54-1424;  77 Massachusetts Ave.
            Cambridge, MA 02139-4307
emails:  eh3 at mit.edu  ed at eh3.com
URLs:    http://web.mit.edu/eh3/  http://eh3.com/
phone:   617-253-0098
fax:     617-253-4464
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20040419/f1d937f5/attachment.sig>