I’m curious as to if it is possible for atoms that are inside a FENE potential to escape outside of it (either due to rounding errors or something like that). I am getting the error that a FENE bond is too long after a few dozen time steps and it eventually breaks. I have set my K = 30 kbT/sigma^2, R0 = 1.5sigma, sigma = 1, epsilon = 1. I double checked the broken bonds in the file dumps and manually calculated their displacement, all of which were within the finitely extensible limit before they broke. I was using a reflective wall, but changed that to a harmonic wall.
It is possible.
You need to set up small time steps if you don't want to decrease your system density.
在 Jul 12, 2011，2:05 PM， nickhil rokkam 写道：
I'll just add that the FENE potential blows up at its limit,
something like 1.5 sigma, i.e. the energy goes to infinity.
So LAMMPS has no choice but to first WARN, then give
an ERROR if your bond stretches that far. It means you
have a bad model if that happens.
Yeah, I’ve been trying to figure out what is causing it to blow up, since I’ve been seeing some particles have some abnormally high forces, and subsequently velocities, which have been pulling the bonds to their limit. I only have 4 potentials in my simulation, the bonds, pairwise interactions, a harmonic wall, and an electric field. I’ve tried turning all of them off but the bonds and it still breaks.
If you guys can think of any other troubleshooting tips, I’d love to hear them.
If you can't run stably with no external constratins, then something
is etiher wrong with your initial config, or too big a timestep.