[lammps-users] Nonlinear lattice


I am doing some simulations on nonlinear structures(triangular). I am using first fix/nvt to initialise them to a temperature say 300K. Then i switch to nve and pick two regions on both ends of a constant length say 10Angs. One side has about 25 atoms and other side has 140 or so. If i just ouput the temperature of those two regions, they oscillate about Tmean=300K.(I am not applying any velocity scaling, just measuring temp’s). But the magnitude of oscillation is high in the case of narrow end(less no of atoms) when compared to other end even though time avg of both is same.If you consider a symmetric structure and do a similar analysis, the magnitude of oscillation will be same on both sides. My question to you is that is there any way we can make the magnitude of oscillation same in the case of nonlinear structure??

I am using Gaussian distbn initially to assign velocities before NVT. Will that have an effect on this?


Temperatures oscillate. They oscillate more when averaged
over small numbers of atoms. If you don't want them
to oscillate, then equilibrate them well first, or run with
a thermostat like fix langevin that works on a per-atom basis.