AMBER Archive (2007)

Subject: Re: AMBER: Heat capacity, diffusivity way off in this simulation of SPC/E water?

From: David A. Case (case_at_scripps.edu)
Date: Tue Oct 09 2007 - 09:09:02 CDT

On Mon, Oct 08, 2007, David Cerutti wrote:
>
> What I've got is a simulation that seems to produce very wrong heat
> capacity and diffusivity for SPC/E water. I've attached the topology and
> initial coordinate files for your perusal. In my calculations, I was
> careful to let the system come together (it's kind-of gaseous to
> begin with) before I started computing the diffusivity and heat capacity.
> In my calculations, I DO get the correct density and also I believe Hvap.
> But, the heat capacity I compute by varying the temperature to get
> (dH/dT)p is about 30% too low, and the diffusivity is also about 30% too
> low. I have re-run the problem with my own MD code, and I get correct
> results for SPC/E using the same analysis methods I applied to the AMBER
> outputs, so it would seem there truly is something strange about the
> AMBER simulation results this time. (Doubly so because I think I've seen
> published data using AMBER that gets all the right answers.)
>

To get dynamical properties, you should run NVE. So, get the system
equilibrated, then turn off the thermostat and barostat. I'd also recommend
setting tol to 0.0000005.

Don't know about the heat capacity problem...did you look at the fluctuations
at a given T, in addition to changing the temperature?

....dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu