[lammps-users] LJ Overlap in (i) & (i+2) neighbors in a flexible polymer system

Dear All,

I have a simulation setup where I want to model a flexible polymer with no angles. However, in these simulations, we find that a fraction of atom pairs that are i and i+2 neighbors show an overlap that is well less than sigma (often pairwise distances closer than sigma/2).

These overlaps go away if I employ an angle_style with low bending stiffness. Are the LJ forces being weighted differently for i & i+2 atom pairs?

Would appreciate help troubleshooting this.

The script is as follows (see lammps data file attached)

conf1.dat (10.4 KB)

it is not possible to reproduce your claim with the provided input.
point 1) is particularly irritating. how can you expect help if you cannot even provide a proper input that would run without having to modify it?

  1. your input cannot run without modifications. it will complain about too many atom sorting bins unless a proper atom_modify command is added and then it will complain about neighbor list bins unless they are set explicitly or the nsq neighbor list build is used
  2. when running a properly updated input on your data file, i cannot see many close contacts and most certainly none with as short a distance as you claim there would be.
  3. for pairwise distance analysis i have added the following lines to your input

compute 1 all pair/local dist
compute 2 all property/local patom1 patom2
dump 1 all local 10000 pairs.dump index c_2[1] c_2[2] c_1

and then run for 10000000 steps

the visualized trajectory looks very reasonable and when checking the local dump file with pairwise distances using

awk ‘/^[01-9]+ [0-9]+ [0-9]+ [0-9]+/ { if (($4 < 0.92) && (($2-$3)*($2-$3) > 1) {print} }’ pairs.dump

I find very few pairs with short distances and all of them are larger than 0.9 sigma.

29 98 100 0.907273
40 123 126 0.915548
6 25 27 0.913829
58 194 196 0.919012
42 94 96 0.918039