On Wed, Nov 14, 2012 at 3:38 PM, Ismail, Mohd F. <farid.ou.edu> wrote:
> I have a solvent box of 400 small organic molecules. I use GAFF for the
> parameters. The box is equilibrated with NPT until the density converged,
> and I got good density comparable to experimental value. I then use NVE
> simulation to get the diffusion.
>
> I get the check COM velocity, temp: 0.000003 0.00(Removed) message
> all the time in the output file. But in this case, for the NVE run, the
> energy is drifting, i.e. it keeps increasing and doesn't seem like it is
> going to equilibrium. I tried doing NVT before going with NVE simulation,
> but after I returned to NVE the energy is still drifting.
>
If you are doing true NVE dynamics, then you should *maybe* see the note
about removing COM velocities once (at the beginning). After that velocity
is removed, conservation of momentum requires that the net translational
momentum remain zero.
The repeated removal of COM velocities does not actually cause the energy
drift you're seeing (since it would decrease the kinetic
energy/temperature, not increase it), but it is indicative of integration
errors that do cause energy drift. The obvious first steps are to check
your integration parameters. Specifically, use a shorter time step and
check that SHAKE is being applied. A 2 fs time step will require SHAKE for
decent energy conservation, although you will probably get better energy
conservation using a 1 fs time step (and I would SHAKE, unless you want to
drop to a 0.5 fs time step).
HTH,
Jason
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Nov 15 2012 - 05:00:02 PST
