AMBER Archive (1999)

Subject: Re: question

From: darden (darden_at_gamera.niehs.nih.gov)
Date: Thu Dec 09 1999 - 14:32:18 CST

dear valeria
THe good energy conservation comes in NVE simulations. Then you should
notice that the mean square fluctuation of total energy is much less
(~1-2%) of the mean square fluctuation in potential or kinetic energy,
which in turn should be nearly the same.
However if you run with NVT or NPT using the Berendsen coupling technique,
the mean square fluctuation of total energy is close to that of potential
or kinetic energy, and in turn the ratio of mean square fluctuation of
total energy to average total energy is pretty high. The reason is that
the Berendsen techniique doesn't conserve anything... its a crude velocity
scaling designed to stabilize things.
Tom Darden

On Thu, 9 Dec 1999, Valeria Costa wrote:

>
>
> I have used roar to simulate the dynamics of a peptide
> in NPT conditions. I obtained as result a standard deviation
> of total energy around 10^-1. Is it possible?
> I was expecting a conservation energy with a standard deviation
> about 10^-4.
> I thank you in advance for your carfully attention and help,
>
> Valeria Costa
>
>
>