From: Sterling Paramore (paramore_at_hec.utah.edu)
Date: Tue Apr 03 2007 - 20:44:36 CDT
Well, it's hard to tell from the graph, but it looks like your
decorrelation time is less than 100ps. If it is 100ps, then it means
you have ~ 100 independent pressure measurements (since your total
simulation time was 10ns). You stated in an earlier e-mail that the
mean pressure was 80 with a standard deviation of 360. With 100
measurements, the standard deviation of the mean is 36, meaning that
the pressure is 80+/- 70 at the 95% confidence interval. If the
confidence interval had been greater than 80, then I would have said
you didn't have to worry because the average pressure would be
indistinguishable from 0 (or 1) to within your measurement. However,
it looks like you are actually measuring a pressure that is not equal
to 1 (especially since the estimated CI is an upper bound). So I'm
not sure what's going on here.....
Actually, in the previous e-mail it looks like your x and y pressures
were significantly different than the z. Was this an NPAT simulation
where only the z-direction was couple to the barostat? If so, the z
pressure is NOT statistically distinguishable from 1, in which case
everything looks OK.
On Apr 3, 2007, at 10:39 AM, Morad Alawneh wrote:
> Dear all,
> I am back to this problem with a graph showing the normalized
> autocorrelation function of the instantaneous pressure. And
> according to this graph, there is not that much of correlation ,
> since its values are close to zero, and the pressure is stable that
> is consistent with pressure vs time.
> Looking forward to hearing any suggestion.
> Morad Alawneh
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:44:32 CST