Re: [AMBER] pmf -Equilibration termination

From: Jason Swails <jason.swails.gmail.com>
Date: Mon, 17 Mar 2014 08:56:00 -0400

On Mon, 2014-03-17 at 12:25 +0000, Rasha Alqus wrote:
> Dear Amber users,
>
> I am doing PMF calculation and i have problem with using restrain, I
> need to restrain system in Z direction and allow it to move only in xy
> direction, I have used distance restrain only at begning but my system
> was moving in z direction, then i applied an additional plane to point
> angle restrain using an angle 90 degree. I perform the minimization
> which goes normally, but when i perform equilibration (which include
> distance and point to plane restrain) only 8000 step is done out
> 250000 step using NVT or NPT and then stop. I did not see any error in
> the output file. I have checked for overlap using petraj and i did not
> see any, and i have checked the trajectories which did not show any
> strange behaviour.
>
> I have attached the out file and error massage .
>
> Could someone help me to solve this problem?

It's hard to know for sure what the actual problem is, but I suspect it
is the cluster you are running on in this case. The following line in
the stderr output:

[node514][[32747,1],9][btl_tcp_frag.c:213:mca_btl_tcp_frag_recv] mca_btl_tcp_frag_recv: readv failed: Connection reset by peer (104)

suggests to me that there's some kind of network or communication
problem. Maybe the machine lost power? Maybe there's a faulty switch?
It's hard to say for certain. I would run the simulation again (you can
restart from the last restart file written) and then keep track of which
nodes my calculation is running on to see if certain nodes tend to fail
repeatedly.

You can also retry the calculation with exactly the same settings and
see if it fails in the same place. If so, then it's likely an Amber
issue (although you'll have to re-run with ig set to a specific value so
the runs will be reproducible).

Good luck,
Jason

-- 
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Mar 17 2014 - 06:00:06 PDT
Custom Search