[lammps-users] pair energy

Dear all,
In my simulations for the polymer and for the nanotube, I get very large number for the pair potential and nan right after the first step. It was working fine with same input criteria for similar systems with different morphology.

My first guess was that the atoms were too close, but they are not overlapping. Any idea why I get huge numbers for pair potential? I applied a soft potential (worked fine) and then switched to lj/cut but it again printed nan’s. So now, I wanted to try lj/class2 but it complains of illegal pair style. Below is my input file.

units real
atom_style molecular
read_data cnt.data

pair_style lj/class2 10.0
pair_coeff 1 1 0.10174 3.35 10.0
pair_coeff 2 2 0.10174 3.35 10.0

bond_style harmonic
bond_coeff * 700.00 1.395

angle_style harmonic
angle_coeff * 39.39368 120

dihedral_style harmonic
dihedral_coeff * 1.00 -1 2

velocity all create 300.0 4928459 rot yes dist gaussian
fix 1 all npt 300.0 300.0 10000 xyz 1.0 1.0 10000 drag 1.0
thermo 10

dump config all custom 1000 coord tag type x y z
write_restart class2.cnt

timestep 0.1
run 1000

Can we not apply pair style lj/class2 without any charge (electrostatic charges of atoms) data?

Note: the above input file is for nanotube composed of only aromatic carbon atoms. So type 1 and 2 are actually carbon. It is just due to the way my pre-processing program works.

Thank you.

Can you send your cnt.data (if too big, send directly to me)?
Also, your input script that gives NAN on the 1st step (presumably
for some other pair potential, not lj/class2).


With the data file you sent, and the input script below, I get
the following behavior. Both pair potentials give a huge LJ
energy on step 0 (initial config) with periodic BC. Both
give an OK energy with non-perioidic BC, and both run a
few steps of dynamics OK in that case.

So I think your initial config with periodic BC puts
some atoms nearly on top of others across the boundaries.

I see no issues with lj/class2 vs lj/cut ...