Re: [AMBER] imin=5, maxcyc>1 stalls

From: Jason Swails <>
Date: Tue, 8 Jul 2014 08:21:26 -0700

On Tue, Jul 8, 2014 at 6:59 AM, Allen, Scott Edward <>

> Hello All,
> I have a trajectory that contains 141 atoms (one residue- a ligand) over
> 5000 steps in implicit solvent that was obtained after processing REMD
> replicas with the script provided in Tutorial A7<
>>. I took the
> trajectories at the lowest temperature and I would like to minimize each
> step before doing a clustering analysis to carry the representative
> structures on to further steps.
> When I run sander with imin=5 and maxcyc any value above 1, the
> calculation stalls at a point that seems to depend on the value of maxcyc
> that I use: when I include more minimization cycles, the calculation stalls
> sooner. By stalling, I mean that sander continues to run, but no more data
> is written and no apparent error is given, and I need to kill the program
> manually. I'm running AMBER on a compute cluster, but the same thing
> happens if I bypass the LSF submission system. If I set maxcyc=1, the
> calculation finishes without any problem. My full sander submission command
> is:

​What happens if you run locally on one of your machines? Does it finish?
 Are you sure that your input trajectory is sane? (Have you visualized it?
Does it look OK?) Are you running in parallel? (Your command below
suggests that you are not). Does it always hang on the same frame (just on
different steps)? Have you tried minimizing the offending frames This is
highly unusual for not running in parallel.

> sander -O -i -o FrameMin.mdout -p 133resub.prmtop -c
> 133resub.inpcrd -ref 133resub.inpcrd -y 133resub_remd.Ttraj.267 -e
> FrameMin.mden -inf FrameMin.mdinfo -x FrameMin.mdcrd
> My full input file is:
> &cntrl
> imin = 5, ntx = 1, maxcyc=1000, ncyc=0, dx0=0.01,

​I suggest leaving ncyc at its default value. It's usually good to start
off with a few steps of steepest descent before switching to conjugate
gradient (at least in my experience).​

igb = 5,
> ntt = 1,
> tol=0.00001,
> ntb=0,
> ntc=1,
> ntf=1,
> ntwx = 1,
> ntwe = 1,
> ntwr = 1000,
> ntpr = 1,
> cut = 999.0,
> itgtmd=1,
> tgtrmsd=0.0,
> tgtmdfrc=0.000,
> tgtfitmask="@1,3,79,83,84,86",
> tgtrmsmask=".1,2,32,33,34,19,20,21,42,43,44,5,6,7,49,50,51,13,14,15,57,58,59,71,72,73,77,78,79",

​Why do you have targeted MD on in your minimization? Does it work if you
disable this?

The questions above are questions that I would ask myself if I had the same
problems, and should hopefully help with debugging.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Tue Jul 08 2014 - 08:30:04 PDT
Custom Search