AMBER Archive (2009)

Subject: Re: [AMBER] difference in temperature

From: David A. Case (
Date: Wed May 06 2009 - 07:08:01 CDT

On Tue, May 05, 2009, Naser Alijabbari wrote:
> Below are the some of the eigenvectors for the second eigenfrequency from a
> MD simulation of E. coli’s thioredoxin. This is using AMBER10 running on Duo
> Core EM64T processor (serial simulation), with explicit solvation (8
> angstrom water shell, and 10 angstrom cutoff). There is very little
> difference between eigenvalues and eigenvectors when the temperature is 293K
> but there is a noticeable difference when 300K is used.

It's not clear what you did here. Are these quasi-harmonic modes? It is not
unusual that eigenvectors would be quite different with different simulations:
even if there is a conserved set of low-frequency motions, they would tend to
exchange among themselves, so that indiviudal eigenvector componenets would
change a lot from one simulation to the next.

> &cntrl
> imin = 0, irest = 1, ntx = 5,
> ntb = 1,
> cut = 10, ntr = 0,
> ntc = 2, ntf = 2, tol=.000001,
> tempi = 293.0, temp0 = 293.0, (these are 300K for 300K simulations)
> ntt = 0, nscm=0,

Note that with irest=1 and ntt=0, the values of tempi and temp0 are ignored.
You would need to examine the outputs to see what temperature you really have.

I also note that you have changed the defaults for a number of variables, such
as tol, dsum_tol, skinnb, etc. As the manual states, the defaults are
carefully chosen to be appropriate for most simulations. If you don't have a
good reason to do so, don't change them (especially by making the tolerances
less tight...)


AMBER mailing list