Re: Acceptable pressure values

From: Peter Freddolino (petefred_at_ks.uiuc.edu)
Date: Tue May 27 2008 - 14:33:00 CDT

Hi Fabio,
if you run for long enough you should eventually reach an average of 1
atmosphere. However, 5 ns may not be long enough; especially for a small
system, the fluctuations in pressure are expected to be quite large (see
http://www.ks.uiuc.edu/Research/namd/2.6/ug/node32.html). You still
didn't mention whether or not you've excluded a slice at the beginning
of the simulation for equilibration; what is much more meaningful is to
consider the pressure values excluding the first ns (for example) to
allow the system to equilibrate. You should also check whether your
volume has converged (keeping in mind that the stdev in volume indicates
alterations on the order of less than 0.1 angstroms in each dimension of
the box length). So long as your volume has converged and your pressures
are averaging near 1 atm with standard deviations like those you're
quoting, this is pretty normal. In practice, for any particular run (of,
say, 1 ns) I usually see a mean GPRESSAVG within 0.1 atm of the target
(keeping in mind that there's a large standard deviation, and the runs
themselves aren't terribly long).

Best,
Peter

So it's not really clear to me what sort of answer you're after.

Fabio Zegarra wrote:
> Dear Peter,
>
> Thank you for your answer. These are the mean values that I got from a
> 5ns simulation:
>
> TEMPAVG: 299.3 (std=2.07)
> PRESSURE: -28.75 (std=299.7)
> GPRESSURE: 3.515 (std=299.6)
> VOLUME: 82020 (std=420.5)
> PRESSAVG: 1.554 (std=23.7)
> GPRESSAVG: 1.555 (std=23.7)
>
> *std: standard deviation
>
> You can observe that the mean values of PRESSURE and GPRESSURE are far
> from the 1 atm target, but PRESSAVG and GPRESSAVG are much closer.
>
> I would like to know if you have similar values for the GPRESSAVG
> using Langevin Piston (langevinPistonTarget 1.01325). I am still
> thinking that it could be too much, because there is 50% difference.
>
> Sincerely,
> Fabio

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:49:30 CST