From: Dudo (dudomail_at_gmail.com)
Date: Thu Sep 15 2011 - 15:23:15 CDT

for an RDF a thousand structures is far behind necessary.
for computing an MSD autocorrelation function you would need more
structures,
time step depends on settings of your potential.
if you constrain vibrations, saving structures each 10-20 fs should be
sufficient
you can extract a period of the fastest movements in your simulation
by computing VACF. from this, by using FFT, you obtain a power spectra
comparable to IR.
this will give you an information on your fastest movements, then you adjust
a time step for computing MSD.
your question was about computing RDF though, so the structure is in
principle changed
after a few picoseconds (in glass it won't change either after microseconds
of simulation,
in liquid the structure changes within 1 ns), in any case for computing
RDF's
a thousand structures from a nanosecond run would be sufficient to provide
accurate data
to compare with the experiment.

On Thu, Sep 15, 2011 at 5:38 PM, Axel Kohlmeyer <akohlmey_at_gmail.com> wrote:

> On Thu, 2011-09-15 at 09:25 -0400, Goldsmith, Jacob wrote:
> > I echo yes, but you might want you output frequency to be small to get a
> smooth function ... say 1-20 fs should do it.
>
> i disagree. those frames are highly correlated and thus do not
> contribute much to the accuracy, but dumping such many frames
> is a waste of disk space and time. frames spaced 1ps apart are
> frequent enough. the smoothness of the g(r) also depends on the
> number of selected atoms and the structural features.
>
> cheers,
> axel.
>
>
> > -Jacob
> > ________________________________________
> > From: owner-vmd-l_at_ks.uiuc.edu [owner-vmd-l_at_ks.uiuc.edu] On Behalf Of
> Axel Kohlmeyer [akohlmey_at_gmail.com]
> > Sent: Wednesday, September 14, 2011 12:51 AM
> > To: siladitya mukherjee
> > Cc: vmd-l_at_ks.uiuc.edu
> > Subject: Re: vmd-l: RDF calculation for production steps
> >
> > On Tue, Sep 13, 2011 at 4:15 PM, siladitya mukherjee
> > <mukherjeesiladitya_at_gmail.com> wrote:
> > > Hi,
> > > I am running a simulation with 2500000 number of timesteps but I want
> to
> > > calculate the RDF only for the production steps, may be for last 500000
> or
> > > 1000000 timesteps. Is it possible to do that in VMD?
> >
> > yes. certainly.
> >
> > axel.
> >
> > > Regards,
> > > Siladitya
> >
> >
> >
> > --
> > Dr. Axel Kohlmeyer
> > akohlmey_at_gmail.com http://goo.gl/1wk0
> >
> > Institute for Computational Molecular Science
> > Temple University, Philadelphia PA, USA.
> >
>
>
>

-- 
____________________
Ing. Dusan Racko, PhD
Polymer Institute of the Slovak Academy of Sciences
Dubravska cesta 3
845 41 Bratislava, Slovak Republic
tel: +421 2 3229 4321