[lammps-users] Si energy with Stilinger-Weber

Dear users,

When I run a simple NVT simulation using silicon and the Stillinger Weber potential, the energy of my system continues to fluctuate greatly over long runs. The measured temperature, when set at 300K, fluctuates from 150K to 500K. I’m using lammps-26Jun08. Hopefully this is a trivial problem with my input file, but I cannot seem to find the cause of this issue. See attached.

Michael Sellers

in.SW-Si (669 Bytes)

SW-Si.out (367 KB)




Yes, it is a relatively trivial problem with your equilibration procedure. The Nose-Hoover thermostat used by the NVT fix does not behave well for systems that are initially far from equilibrium. In your case, starting with a perfect crystal, the initial potential energy is zero, whereas it should be ~ kT. Solutions include:

-First equilibrate with another method, such as Langevin dynamics
-Reinitialize velocities after initial equilibration
-Use “drag” option with NVT fix