[gmx-users] Statistical errors

[David]

On Sat, 2003-02-08 at 15:08, Pedro Alexandre Lapido Loureiro wrote:
> Hi,
> 
> Apart from systematic errors (choice of forcefield, use of cut-offs etc) and
> statistical ones (undersampling) can we say there is "noise" in the measurement
> of some average derived from a simulation?
If you assume you are at equilibrium there are fluctuations. These are
determined by the kind of property. The usually scale with
1/sqrt(natoms)

> For instance, do the temperature and pressure baths introduce "noise"?
No, the Berendsen baths actually reduce the fluctuations. Nose-Hoover
doesn't reduce the size of fluctuations, but does introduce artificial
periodicity in the fluctuations.

> Another point is the propagation of sound waves in systems with periodic
> boundary conditions: it would introduce noise in time correlation functions. How
> do I recognize such phenomena?
Not completely sure, but I think that with non-equilibrium MD (imposing
a velocity field?) you can measure the speed of sound. How it influences
other properties I don't know. Check literature...


-- 
Groeten, David.
________________________________________________________________________
Dr. David van der Spoel, 	Dept. of Cell and Molecular Biology
Husargatan 3, Box 596,  	75124 Uppsala, Sweden
phone:	46 18 471 4205		fax: 46 18 511 755
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://xray.bmc.uu.se/~spoel
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++