[Mitgcm-support] moving tank experiements to c35

[mitgcm-support at dev.mitgcm.org]

Timour,
 
 Model still blew.

 I don't think I am going to have any time to help
out more this week. I am away early next week and
will be busy at the end of the week. 

 What you need to do is set up a bonanza spin-up run
with your old code decomposed the same way as the
new case

 e.g. SIZE.h : nPx=1, nPy=4

 and with the exact same parameters.
 Then you can play with the codes side by side and see
where they differ. 

 Jean-Michel has lots of experience with this and he is
using a little four-proc pair of machines like yours - but I
am not sure if he has any time either!
Perhaps you could see if he is willing to help out. I cc'd this
to him.

 JMC - if you do have time I have migrated the Helen numerical
setup from a code which dated back to July 1998 to c35. 
The setup is in theory very simple. Timour can explain the setup
and show you some notes and papers that Helen et. al. wrote on the
experiemnt.
Basically they have a circular rotating tank (like the one in the
lab/coffee room), a rotating disk putting in a stress at the surface
and heating over some smaller region. Initially they spin-up the
fluid without heating and with talpha=0. After some time
they turn on heating and switch talpha=2e-4.  The disk stress is 
constant throughout. Right now the model is blowing up during
the spin-up. The STDOUT stuff sort of looks OK until it blows up. 
Timour hasn't plotted any fields so I'm not sure about the rest 
of the stuff. In principle you should be able to do your number 
reproducing trick like with AIM etc... to make sure the setups
are truly the same. The results are also like the tank experiments
Larry Pratt described yesterday. The most likely thing is I have
missed something in the setup. One possibility (extremely unlikely)
is that becuase the rhsMax normalisation is off in the 
four-proc saint-tropez setup something strange happens. The
STDERR.NNNN trace looks odd.

To see the c35 setup 
 ssh saint.lcs.mit.edu
 cd /scratch/timour/c35/checkpoint35/verification/exp2/input
 code mods are in ../code
 ( note I splatted the true exp2 setup and reused the directory )
Timour can show you where the old setups are on bonanza.

If you don't have time don't worry, we will just have to figure out
some other way to get Timour going.

Chris