[gmx-users] Energy drift

[Mark Abraham]

On Tue, Aug 19, 2014 at 11:48 PM, Xiaobin Cao <xcao at lsu.edu> wrote:

> Thank you, Mark. I am following this paper: "Mutual diffusion coefficients
> of heptane isomers in nitrogen: A molecular dynamics study" (JCP, 2011).
> They use flexible model, so do I (please check my top and mdp files).
> Actually I already checked the time step of 0.5fs, and it does not work.
> Belows are their description for their system:
>
> They use the potential parameters I attached for N2,  but I am not sure if
> they assign charges on N atoms. Lennard-Jones 12-6 potential is used to
> describe interactions between molecules.
> NVT ensemble and PBC are used, and the cutoff is 18 A.
>  The initial velocities of the molecules were derived from the Boltzmann
> distribution.
> The Verlet leapfrog numerical integration algorithm was employed with a
> time step of 1.0 fs.
> The total simulation time was 14 ns and the sampling time for the velocity
> correlation was chosen to be 7 ns for the temperature range considered.
> The velocities and positions of the molecules were recorded every 50 time
> steps.
>

Given that their bond constant for N-N is ~4.5 times higher than their bond
constant for C-H, and both have comparable equilibrium lengths, and
unconstrained C-H bonds are normally treated with either 1fs or 0.5fs, and
that time step should be rather shorter than the period of bond
oscillation, I'd have serious reservations about the quality of that model,
and your results suggest reason for that. If you've got the parameters
right, then you should be able to suppress (post-equilibration) NVE drift
from integration inaccuracy by taking a time step that is small enough. How
small is small enough?

Mark

Thank you
>
> Date: Tue, 19 Aug 2014 14:35:51 -0500
> From: Mark Abraham <mark.j.abraham at gmail.com>
> To: Discussion list for GROMACS users <gmx-users at gromacs.org>
> Subject: Re: [gmx-users] Energy drift
> Message-ID:
>         <CAMNuMAQ5jiQeaFFBRk2sdp5s0mjtMpU5dfq=FPAno221co=
> Q2w at mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> Hopefully you are following some established protocol for your N2 model. If
> so, use its recommendation for time step and or bond constraints. Not using
> constraints can lead to your symptoms with 1fs time step - prefer 0.5fs.
>
> Mark
>
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>