[MITgcm-support] Job died for unknown(?) reason with float (flt) package

[Yangxin He]

Hi guys,

I have been running MITgcm for a while with no problem. I simulate the interaction of barotropic tide and continental shelf, and I use obcs, pp81 and diaganostics packages. The runs were fine until a few days ago, I tried this float (flt) package. (By the way, I see the flt readme file has not been updated for many years and there is perhaps a mistake in the description of "kpart", and the parameter "kfloat" does not seem to matter, so I just set it to be 0.) Anyways, the flt package seems easy(?) to use.

However, when I included flt package in my simulation, my run gets killed after running a few hours. I needs at least 30 tidal periods of outputs and it can only produces up to about 3 or 4 tidal periods. (I ran my jobs on compute Canada, in the past, sometimes my runs will get killed for unknown reasons but if I re-submitted it, it will be fine.) As a result, I re-submitted my job for 6 times, except this time (with the flt package) the job always gets killed after running for a few hours. I checked the results, they seem fine. So I am pretty lost.

I have 100 particles, and I set max_npart_tile=300, and max_npart_exch=100 in FLT_SIZE.h.
I use iup= -1 for 3d advection. Flt_noise=0.

Any advice will be appreciated!

Thank you!

Yangxin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20210218/3774dc48/attachment.html>