[MITgcm-devel] Re: lab_sea

[Jean-Michel Campin]

Hi Dimitri,

Thanks for your suggestions: it works !
I put LSR_ERROR=1.e-40 and get very tiny differences after 2 iters,
and after 10 iters, same magnitude as without dynamics.

In order to get a fair testing, I propose to change lab_sea exp. 
by setting LSR_ERROR=1.e-12 in input/data.seaice
(no big differences between 1.e-12 & 1.e-40 anyway)
And after this, we can change SIZE.h to use 2 tiles (like SIZE.h_mpi)
instead of 1, like all the other exp. do.
What do you think of this ?

Jean-Michel