In playing around with a simple lennard jones fluid at constant NVT. I’ve encountered an annoying oscillatory behavior in the potential and kinetic energies. The documentation suggests adding a drag term in the fix nvt command to alleviate this problem, but I would like to simulatate using true Nose-Hoover dynamics.
Does anyone know why or under what situations this oscillatory behavior arises (initial conditions, system size…)? I’m guessing one fix would be to introduce NH thermostat chains, but it doesn’t look like lammps currently has that functionality built in.
I’ve part of my input script, just in case it could be helpful.
In playing around with a simple lennard jones fluid at constant NVT.
I've encountered an annoying oscillatory behavior in the potential
and kinetic energies. The documentation suggests adding a drag term
in the fix nvt command to alleviate this problem, but I would like
to simulatate using true Nose-Hoover dynamics.
Does anyone know why or under what situations this oscillatory
behavior arises (initial conditions, system size...)? I'm guessing
one fix would be to introduce NH thermostat chains, but it doesn't
look like lammps currently has that functionality built in.
this frequently happens during equilibration. so a simpler fix would be
to run for a while with drag enabled and then turn it off or use a
much smaller value.
whether N-H chains are a solution depends largely on what property
you are after and what system size you want to run. if your system
is going to be small, then (long?) N-H chains are more important
than for larger ones.