[gmx-users] Re: g_velacc

rams rams rams.crux at gmail.com
Sat Aug 30 18:02:11 CEST 2008


Thanks to David van der Spoel and Vitaly Chaban for their valuable
suggestions regarding the calculation of diffusion constants.

I did used the option -acflen with default values and focussed on the region
close to t=0. This time my plot looks fine though the decay is not smooth
(rather it is a sharp decay and then parallel to time access). Now I would
like to know how to get the diffusion constant from this.

It appears to me like I need to integrate this correlation funtion to obtain
the diffusion constant (does -integrate option in g_analyze helpful ?) or
any other way I can get this.

I also found a reasonably good value this time (in comparison to experiments
for my diffusion constatns using g_msd). I want to make a comparison between
the values obtained by einstein and Green-kubo relations.

Thanks in advance.


On Fri, Aug 29, 2008 at 5:32 PM, rams rams <rams.crux at gmail.com> wrote:

> Dear Users,
> I am trying to calculate the velocity correlation functions to estimate the
> diffusion constant of my protein which is having nearly 50 residues. I am
> using the following command:
> g_velacc -f .trr -s .tpr -n .ndx -o .xvg -fitfn exp
> the output is the following:
> COR:
> COR: Correlation time (plain integral from  0.000 to 2500.000 ps) =
> 0.90300 ps
> COR: Relaxation times are computed as fit to an exponential:
> COR:   y = exp(-x/a1)
> COR: Fit to correlation function from  0.000 ps to 2500.000 ps, results in
> a
> COR:   Fit from   Integral Tail Value   Sum (ps)    a1
> (ps)
> COR: 0.0000e+00 0.0000e+00 1.0458e-01 1.0458e-01  1.0458e-01
> I have no idea about the output to understand. I understood it is trying to
> fit to the equation it displayed but have no idea about the number 0.90300
> ps.
> The .xvg plot I got is so surprising its just a straight line parallel to
> time axis. I am sure I am making some mistake as I am using g_velacc for the
> first time.
> From this if we want  to get the Diffusion coefficient we need to integrate
> the correlation time from 0 to infinitive as it is described in the manual.
> Is there any way we can do it by using any of the options or we need to do
> it separately by using any other tools ? If so please let me know.
> I also wanted to know the following things:
> What does the following options do:
> -normalize
> -integrate (in g_analyze)
> -fitfn (there are couple of options are given if any one have idea about
> the sources for them ?)
> I calculated the diffusion coefficients using g_msd also, it gave the
> diffusion coefficient value as 6.0*10^-6. Which is quite large compared to
> the values reported in the literature which are around 2.0*10^-6. Whats
> going wrong here also it is not generating the output .xvg file. (It is
> complaining input/output fatal error).
> Thanks in advance,
> Ram.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20080830/170b676f/attachment.html>

More information about the gromacs.org_gmx-users mailing list