[gmx-users] g_energy averages don't match xmgrace values

Andre Farias de Moura andre at qt.dq.ufscar.br
Mon Aug 2 19:03:37 CEST 2004

dear berk,

I have just done what you suggested me, I have calculated
surface pressure over each individual edr file as well as
over the concatenated one. the results are summarized below:

(1) using g_energy (average and rmsd):

out01.edr: -578.658    4280.23
out02.edr:  206.244    4163.27
out03.edr:  201.476    4253.36
out04.edr: -145.052    4249.69
out05.edr:  194.143     4129.2

the average over these five subaverages is -24.369, which
is quite larger compared to the concatenated file result,
-8.12021 bar nm

(2) using grace (mean and stdev):

out01.edr: -559.569  4413.42
out02.edr:  116.704  4135.83
out03.edr:  201.288  4291.91
out04.edr: -104.37   4238.16
out05.edr:  230.695  3959.47

the average over these five subaverages is -23.050, in good
agreement with the concatenated file result, -22.825 bar nm.
if I use a simple fortran program to compute these averages
I get the same figures as obtained using grace.

it seems that something wrong happens to the averages when
energy files are concatenated, don't you think so?

I hope it helps to find out what is going on,

best regards,



    Dr. Andre' Farias de Moura


  Laboratorio de Quimica Teorica
Universidade Federal de Sao Carlos
     Sao Carlos - SP - Brasil


        Endereco Atual

    Grupo de Quimica Quantica
Instituto de Quimica de Sao Carlos
    Universidade de Sao Paulo
    Sao Carlos - SP - Brasil
    Telefone: 55-16-3373-8055


On Mon, 2 Aug 2004, Berk Hess wrote:

> >
> >just for you to be sure it's not a precision issue we are talking
> >about: another simulation gives average surface tension of 257.7
> >bar/nm using g_energy and 1804 bar/nm using grace and energy.xvg
> >file. that's certainly not a difference due to precision!
> This number seems a bit large.
> But the other numbers could easily be a precision effect considering the
> RMSD is 4000.
> You can check the results by manualy calculating the total average
> from the averages g_energy reports for each single energy file
> (of course taking into account different weights if the run times are
> different).
> I just recalled that the RMSD of concatenated files is not correct
> as eneconv -h will also tell you.
> But the averages should be correct.
> Berk.
> _________________________________________________________________
> Hotmail en Messenger on the move
> http://www.msn.nl/communicatie/smsdiensten/hotmailsmsv2/
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.

More information about the gromacs.org_gmx-users mailing list