[gmx-users] Statistical uncertainty in gromacs

[Juliette N.]

Hi Matthew,

Thanks for your reply. I tried g_analyze as you suggested:

1) I am wondering why the average given by g_energy and g_analyze are not
identical. I tried the following:

g_energy -f  Potential. edr -o Potential.xvg and extracted the average then
provided Potential.xvg as input to g_analyze ( I did this since g_analyze
seems to read xvg files only). and issued: g_analyze –f.Potential.xvg–av.

However the averages are not identical.


2) Also I tried g_energy -f  Potential. edr -o Potential.xvg and from
g_energy I can see  Energy                      Average   *Err.Est.*
RMSD  Tot-Drift


are given by default. Then tried g_analyze -ee and noticed that the *
Err.Est.*  reported by g_energy and the one from g_analyze -ee are not
identical!

Could you please explain why is this happening and how  *Err.Est.* is
estimated? .

g_energy says:
An error estimate of the average is given based on a block averages over 5
blocks using the full-precision averages. The error estimate can be
performed over multiple block lengths with the options -nbmin and -nbmax. *
Note* that in most cases the energy files contains averages over all MD
steps, or over many more points than the number of frames in energy file.
This makes the g_energy statistics output more accurate than the
.xvg<http://manual.gromacs.org/online/xvg.html>output

I cant seem to understand how error is obtained.:( If I am reading the
frames from say , -b 4000 -5000, what happens if I dont provide blocks
etc...

Appreciate your comments,
Thank you,

Cheers
J.





*
*

On 21 February 2013 10:48, Matthew Zwier <mczwier at gmail.com> wrote:

> Hi,
>
> I don't know of a GROMACS tool to do this. g_analyze may work (see manual
> page, option "-ee"), if you can generate a time series of A to look at.
>
> That said, what you've described is a classic propagation of error problem.
> If uncertainties are small and likely to be symmetric about the mean, then
> you can analytically calculate the uncertainty in A in terms of the
> uncertainties on the input quantities. If the uncertainties are not small
> or are very asymmetric about the mean, then something like bootstrapping
> may be in order. Information on both of these is readily available on the
> web (now that you have a few keywords to search for).
>
> Cheers,
> Matt Zwier
>
>
> On Thu, Feb 21, 2013 at 2:09 AM, Juliette N. <joojoojooon at gmail.com>
> wrote:
>
> > Hi everyone,
> >
> > Does anyone know of a gmx tool which calculates the statistical
> > uncertainty?
> >
> > and if there is such a tool, how can one calculate the statistical
> > uncertainty for a quantity A (not extractable directly by gromcas), which
> > is expressed in terms of 3 other quantities, e.g. property A=(Potential
> of
> > gas phase - potential of liquid phase)*(density of liquid phase).
> >
> > I am interested in statistical uncertainty of A.
> >
> > Appreciate your help,
> > --
> > Thanks,
> > Best,
> > --
> > gmx-users mailing list    gmx-users at gromacs.org
> > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > * Please search the archive at
> > http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> > * Please don't post (un)subscribe requests to the list. Use the
> > www interface or send it to gmx-users-request at gromacs.org.
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>



-- 
Thanks,
J. N.