[gmx-users] Re: energyfile questions

Justin A. Lemkul jalemkul at vt.edu
Fri Sep 5 19:42:36 CEST 2008

Matteus Lindgren wrote:
> Thanks Justin for the reply. 
> Gmxcheck of the .edr file does not tell me the spacing of the frames,
> probably because it is inconsistent. I do get a lot of the timesteps don´t
> match errors though. It seems all my .edr files that have been calculated on
> the cluster have extra frames in the output, while on my workstation it
> works fine fro the same .tpr file. Something with data flush on the cluster?

Over extremely long timeframes, I have noticed that rounding errors can be 
introduced.  I have also noticed that tpbconv is more prone to propagating these 
(giving weird fractional times like you've shown below), when compared to 
restarting runs using grompp.

It may be a machine precision issue, i.e. how accurate the computations are, and 
thus how much error is propagated.  I find it more convenient to run shorter 
simulations and continue using grompp - it seems to introduce a lot less 
rounding issues.

> Since this run was the first I used grompp to make the .tpr but it shouldn’t
> matter if it’s a cluster only problem.
> Checking energy file s6wu.1.edr
> Opened s6wu.1.edr as single precision energy file
> frame:      0 (index      0), t:      0.000
> Reading frame    200 time  100.000
> Timesteps at t=120.5 don't match (0.5, 0.00999451)
> Timesteps at t=120.51 don't match (0.00999451, 0.490005)
> Reading frame    400 time  199.500
> Timesteps at t=237 don't match (0.5, 0.0699921)
> Timesteps at t=237.07 don't match (0.0699921, 0.430008)
> Reading frame    700 time  349.000
> Timesteps at t=354 don't match (0.5, 0.0299988)
> And so on...
> About the protein temperature: Is 50K fluctuation in temperature really ok
> for nose-hoover thermostat on a 100 residue protein????

Well, I don't know anything about how you equilibrated your system, or what its 
doing, so it's hard to say.  That is quite a bit of fluctuation.  I typically 
see about a 15 K fluctuation for even smaller peptides.  It might be related to 
the machine precision problem, but that would be a wild guess.


> Thanks
> /Matteus 
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/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/mailing_lists/users.php


Justin A. Lemkul
Graduate Research Assistant
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080


More information about the gromacs.org_gmx-users mailing list