Hi All,
I am doing some calculations on the properties of alkanes. In my NPT run, the condition is set to 300K, 0Pa, using the Nose-Hoover thermostat. The simulation was run for 200,000 steps with a step size of 1 fs. The pressure was obtained by averaging the last 100,000 steps, the standard deviation of it was about 600 at.!
Besides, in my NVE+langevin runs, the pressure as well as the all its tensors also show significant fluctuation! I would like to know, if this is normal? Or how can I make the fluctuation smaller? I searched previous archives, but did not get enough information on this.
Thanks a lot.
Konzern
Konzern,
Yes this is normal. You didn't mention how big your system is.
Typically, the larger your system, the smaller the fluctuations will
be due to a bigger spatial averaging.
This concern actually comes up a lot and is perhaps the single most
frequently asked question on this forum. It is not a LAMMPS-specific
issue, but rather a result of trying to extract macroscopic
information like pressure from a microscopic system, and due to the
fact that in the simulation of liquids, pressure just so happens to
fluctuate a lot.
Another way to decrease the fluctuation is to use time-averaged blocks
and take more samples --- i.e. compute the pressure every timestep and
time average the results.
Paul