timestep and damping for langevin thermostat

Thanks Oleg

My system is just a bundle of cnts. There is nothing else interacting with
The atoms in the cnts are already at the equilibrium distance before I start

how did you determine that?

equilibriation. I then give them a velocity consistent with a temp of 0.5 K
and then I try and get a fix langevin to maintain this temp.

When I use velocity rescale the system seems to behave.

it doesn't, you just get fooled by temp/rescale. that is one of the
reasons why i keep saying all the time that temp/rescale is bad.

try running with fix nve and without *any* thermostat, just the
initial temperature assignment. if your system is properly relaxed and
minimized the final temperature should be about half the initial

the fact that it keeps exploding like seen in your log file, indicates
that something is *way* off.