Slow and Fast push off - LAMMPS implementation doubt

Dear All,

I am trying to simulate a polymer system with 1000 chains (each with 40 monomers). To avoid blowing up of energy, I am using the methods like soft/fast push off (ref: Auhl-Plimpton, equilibration of long chain polymer melts in computer simulations) using LAMMPS. I found that pair style soft performs fast push off. I also found fix nve/limit as an alternate method for push off. But I have a doubt as to whether it really corresponds to slow push off? The reason is, from reading documentation it said the following

A limit is imposed on the maximum distance an atom can move in one timestep. Forces on atoms must still be computable (which typically means 2 atoms must have a separation distance > 0.0). But large velocities generated by large forces are reset to a value that corresponds to a displacement of length xmax in a single timestep. Xmax is specified in distance units; see the units command for details.

But in the paper, it is given by U® = (r-r_fc)*U’(r_fc) + U® for r< rfc and rfc is reduced from 2^1/6 sigma to 0.8 sigma. If xmax in LAMMPS equal r-r_fc, then is the reduction in rfc as told in the paper, automatically taken care of in the source code? Also if the velocities are scaled by a value corresponding to xmax, how exactly is the U® term (which is dependent on r and not a constant) incorporated?

Sorry for the big mail.

Thanks and Kind Regards

Vaidyanathan M S

Dept of Chemical Engg

UT Austin

How fast or slow either of the methods is,
depends on what parameters you set. You
can ramp the strength of the soft potential
as slowly as you want, over billions of
timesteps. You can set the nve/limits as
small as you want so you take tiny timesteps.

Steve

Dear Steve

So in principle, we can obtain the same equilibrated morphology with both slow and fast push off methods by choosing an intelligent value for each of those method.

Thanks for the information.

Thanks and Kind Regards

Vaidyanathan M S

I’m saying that if you equilibrate too fast (with either method)
then you won’t get a well-relaxed system.

Steve