Indeed the serial versions of sander and pmemd do give similar results using ntt=3, my error was to use the MPI versions to run small jobs.
This occurred when I changed from explicit solvent(water) to the generalised Born approximation.
Steve
Dr S Bailey
Department of Physics
Lancaster University UK
Email: s.bailey.lancaster.ac.uk
Tel: +441524592844
-----Original Message-----
From: David A Case [mailto:case.biomaps.rutgers.edu]
Sent: 18 June 2013 16:16
To: AMBER Mailing List
Subject: Re: [AMBER] PEG
On Tue, Jun 18, 2013, Bailey, Steven wrote:
>
> I have just changed my executable from pmemd.MPI to sander.MPI.
> This resulted in a fairly obvious error: Must have more residues than
> processors. As I have only 3 molecules in this test case then in
> retrospect I would have expected this. Nevertheless this error did not
> show up on the pmemd.MPI run.
This is expected: pmemd handles parallelization in a different way than sander, and has different restrictions.
>
> I then ran the job in serial mode sander and low and behold have
> 'jiggling' molecules.
We would need more information: be sure you are using ntt=3 if you just have three molecules in your system. Did you rms fit your trajectory before visualizing it? If so, try it without doing that. But without knowing how you really ran the two simulations, it is hard to provide any real answers:
generally speaking, pmemd and sander should give equivalent results.
...dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jun 18 2013 - 08:30:14 PDT
