system's center of mass drifting during BOMD NVT simulation

[Matt W]

Hi,

you can set COMVEL_TOL in the MD section to give a maximum velocity to the 
COM motion.

Temperature fluctuations are absolutely normal in a finite sized system. 
Temperature is only 'constant' in the limit of an infinitely large system, 
otherwise it fluctuates around the target value. The fluctuations with 
Nose-Hoover should be correct for a canonical ensemble (if all else is 
well).

Matt

On Monday, November 9, 2015 at 5:56:03 PM UTC, lar... at lbl.gov wrote:
>
> Dear all,
>
> I have been studying water systems of different sizes for a while now (I 
> have attached here an example input file ). In these simulations I 
> consistently observe the center of mass of the system drifting (i.e. the 
> whole system is moving).
>
> I'm running BOMD in the NVT ensemble using a Nose-Hoover massive 
> thermostat chain. My best guess to explain the drift of the system is that 
> the random velocities imposed by the thermostat (maybe during 
> initialization) are not canceling out exactly. Is there a way of 
> constraining the momentum of the system's center of mass to get rid of this 
> issue?
>
> I am also observing a sort of estrange long term behavior in the 
> temperature where the temperature vs. time curve exhibits a long-wave 
> fluctuation (I have attached a plot here, x-axis is number of steps and 
> y-axis is temperature in K). I'm afraid this could be an artifact of the 
> thermostat.
>
> If this is not an artifact of the thermostat I am not quite sure what 
> could be causing it, any other ideas?
>
> Thanks for your help!
> Luis
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20151110/f318e5e0/attachment.htm>