[gmx-users] oscillations in pressure (NPT)

[Erik Lindahl]

Valentin Gogonea wrote:

> Dear Gromacs users,
>
> I tried to run a NPT simulation (100 ps) of one of the demo examples 
> that come with gromacs distribution (cpeptide in 
> share/tutor/gmxdemo/). I got large oscillations of the pressure, I 
> searched the mailing list and following the suggestions given in an 
> earlier discussion I increased the tau_p from 0.5 to 2.0 and than 5.0 
> without success.
>
> I used the Berendsen barostat (I included the md.mdp file)
>
> I would appreciate if anybody can provide some advice of what to do to 
> remove these pressure oscillations. I included here the pres.xvg files 
> obtained with the g_energy program for the three simulations. 

Hi Valentin,

Those curves look quite OK to me.

Pressure is far from trivial on an atomic scale since it is an ensemble 
property. It will be very smooth when there as something like 6*10^26 
particles in your system, but the fluctuations at equilibrium will be 
inversely proportional to your system size, so you can imagine what 
happens in a system that is more than 10^20 times smaller...

Anyway, since it's an ensemble property it doesn't make much sense to 
look at the instantaneous pressure over picoseconds; run the system for 
a nanosecond and look at the average (or running average in xmgrace).

Cheers,

Erik