Re: [AMBER] Question

From: Jason Swails <>
Date: Thu, 10 Jul 2014 14:11:50 -0700

On Thu, Jul 10, 2014 at 2:01 PM, Brittany Boykin <> wrote:

> Hello,
> I ran MD simulations for an apo protein for 10ns. After completing the
> calculation, I examined the out file and noticed that the calculation
> skipped in number of steps (59000 to 290000). Can someone possibly explain
> why this happened? Also, the calculation went to a million steps as stated
> but why did it read completed, if it skipped multiple steps?

‚ÄčI can guess, but that's the best I can do given the information you
provided. It's possible that you tried to run your simulation in parallel
but one of two things happened:

1) You used the serial executable instead of the parallel executable
(sander instead of sander.MPI or pmemd instead of pmemd.MPI).
2) The 'mpirun' you used to run in parallel did not come from the same MPI
installation that you used to compile Amber in the first place (this only
could have happened with sander.MPI or pmemd.cuda.MPI).

In both of these cases, what happened was you effectively launched a
handful of separate jobs that were each writing (and overwriting) to the
same files. Anything can happen there.

Some other possibilities I can think of:

1) Filesystem or other hardware problem
2) I'm misunderstanding your problem completely.


Of course, if this doesn't help you, you have to give us enough information
to help more (including, of course, the actual program and program version
you used...)

Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Thu Jul 10 2014 - 14:30:05 PDT
Custom Search