RE: AMBER: output file format in SANDER.MPI & PMEMD

From: Ross Walker <>
Date: Thu, 28 Aug 2008 10:09:38 -0700

Hi Fatima,

> I run Umbrella sampling simulation in both SANDER.MPI and PMEMD. I
> realized that
> the values of the restraint are different (diff. range 0.05 - 1.00)
> although the
> simulation system is literally the same.

What do you mean by the word 'restraint' here? Do you mean that you
physically set the restraints the same in the input file but pmemd used a
different restraint force? I.e. it reads it wrong from the input?

Or did you set the restraint forces different in the same simulations? In
which case what is the problem?

Or do you means that the PMF comes out different?

> I suspect this to be inherent to SANDER and PMEMD numerical implementation

Chances are your simulations are not converged - they should converge to the
same answer but only if you have run a long enough sampling.

Rounding differences in the calculations will mean that the trajectories
diverge over time, the first 500 steps or so should match well, beyond this
they will begin to decorrelate and will ultimately end up totally
decorellated. This is the nature of newtonian dynamics which is ultimately
what would be called a chaotic system. In theory the path is analytical but
an infinitely small change in the starting conditions will ultimately end up
with a set of uncorrelated trajectories.

What you ultimately need to compare is what the averages come out to and
what your PMF's from multiple runs look like. These should be equivalent in
PMEMD and Sander providing you have sampled sufficiently. Comparing a single
snapshot or say the final energies will not provide you with any useful
information here.

Good luck,

|\oss Walker

| Assistant Research Professor |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- |
| | PGP Key available on request |

Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
Received on Sun Aug 31 2008 - 06:07:36 PDT
Custom Search