[lammps-users] Regarding fix rigid/nvt for MD Simulations

LAMMPS LAMMPS Mailing List Mirror
1

Dear LAMMPS users,

I am doing two-dimensional MD simulations of LJ particles in NVE and NVT. Next, we put a dumbbell-shaped tracer in the system. We are making the dumbbell by fusing two disks using fix rigid/NVE and rigid/NVT for NVE and NVT respectively. Before putting the dumbbell-shaped tracer, the trajectories of the particles for NVE are similar to the trajectories of the particles for NVT. But after putting the dumbbell-shaped tracer, the trajectories of the particles for NVE are drastically different from the trajectories of the particles for NVT. Can you please help with that matter? We are attaching the trajectories of a tagged particle for NVE and NVT before and after putting the dumbbell.

particle_NVE.png

particle_NVT.png

particle_rigid-NVE.png

particle_rigid-NVT.png

2

There is no convincing proof of any problem here. MD simulations are “chaotic” the smallest changes can result in exponential divergence. So rather having similar trajectories with different time integration would be the unusual case and not the expected behavior.
Axel