AMBER Archive (2008)

Subject: Re: AMBER: self-diffusion coefficient and Berendsen

From: David A. Case (
Date: Sun May 25 2008 - 23:09:32 CDT

On Sun, May 25, 2008, wrote:

> I had put the value of NSCM greater than the number of steps because I didn't
> know why it can be useful and it was creating discontinuities on the total
> energy profile. But after reading the 'Flying Ice Cube', I put it back to its
> default value and now the diffusion coefficient is ok with Berendsen thermostat

I still think something is very wrong here.

How are you calculating the diffusion constants? Is your system really
equilibrated before you begin the production calculation? What value of tautp
are you using?

The value of NSCM should not be causing discontinuities in total energy. And,
you *do* want to turn it off (set it to a large value) for computing
diffusion. I just have a fear that you are getting what seems like the
"right" result for the wrong reason.

> As the diffusion coefficient depends upon the temperature, isn't it a
> problem to calculate it from a NVE simulation ?

No: you equilibrate the system to the temperature of interest, but dynamical
properties should be derived from Newton's equations. Thermostats disrupt
dynamics (although ntt=1 can be a minimal disruption).


The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)