Issue with Tabulated Potential in LAMMPS - "Pair distance < table inner cutoff" Error

LAMMPS LAMMPS Beginners
1

Hi,
I’m trying to simulate a 16-atom MD system in LAMMPS using a tabulated potential.
The tabular potential I use in this example is actually obtained from LAMMPS itself using the pair_write command.

I first use the built-in Lennard-Jones (LJ) potential with these parameters:

pair_style     lj/cut 10.0   # LJ with 10.0 Å cutoff
pair_modify    tail yes      # Analytic long-range tail corrections
pair_coeff     * * 0.0715 3.31
pair_write     1 1 1600 r 1.0 13.9 table_pot1.txt mypot

Then, I read back this same tabulated potential for the simulation.

However, when I run the simulation, I get the following error:

15 atoms in group xenon
1 atoms in group rubiduim
Generated 0 of 1 mixed pair_coeff terms from geometric mixing rule
Setting up Verlet run ...
  Unit style    : real
  Current step  : 0
  Time step     : 2
ERROR on proc 0: Pair distance < table inner cutoff: ijtype 1 1 dist 0.772383999

This suggests that two Xenon atoms are 0.77 Å apart, which is unphysical.
But when I manually check the distance using a Python script, the minimum actual distance is 6.51 Å. The simulation runs fine if I use the built-in LJ potential, but it fails when I read in the same potential as a tabulated file.

Here is my LAMMPS input script: LAMMPS_INPUT_FILE - Pastebin.com
Here is the atomic configuration File (Rb1Xe15.cfg): Rb1Xe15.cfg - Pastebin.com

What I Have Tried

Manually checked interatomic distances:

  • Used a Python script to compute pairwise distances.
  • Minimum distance = 6.51 Å, far from the 0.77 Å reported by LAMMPS.

Checked for periodic boundary effects: Issue persists even with PBC off.

Checked for simulation box size issues:

  • Box size is large (55×55×55 ų).
  • No atoms appear too close to the box boundaries.

Checked tabulated potential format:

  • Used pair_write to generate a tabulated potential from LAMMPS itself.
  • Ensured same parameters for both lj/cut and table.
  • Issue only occurs when using the tabulated potential.

My Question

  • Why does LAMMPS report an unphysical 0.77 Å separation only when using the tabulated potential?
  • The same LJ potential runs fine when used directly but fails when read from a table.
  • What could be causing this discrepancy?

I want to learn how to properly use tabulated potentials for a future project, so understanding this issue is critical.

Any help or suggestions would be greatly appreciated!

img
img1400×800 152 KB

Here is the tabulated potential: # DATE: 2025-03-11 UNITS: real Created by pair_write# Pair potential lj/cut fo - Pastebin.com

2

The problem has nothing to do with the tabulation, but rather with how your data file is formatted.
You are using atom style full, so the format for the Atoms section is:

atom-ID type mol-ID charge x y z

So column 4 is read and assigned to the per-atom charge and thus all your atoms have a z coordinate of 0 (last column).

If you add the following lines to your input, you can track which are the pair distances LAMMPS computes:

compute 1 all property/local patom1 patom2
compute 2 all pair/local dist
dump 1 all local 1 distances.dump c_1[*] c_2

From that output you can see that atoms 10 and 11 are at the close distance.

10  1  1  14.986145  2.524285  11.407135  0
11  1  1  7.366026  2.666401  12.166332  0

And indeed the coordinates (2.524285 11.407135 0) and (7.366026 2.666401 12.166332 0)
are less than 1 angstrom apart.

From the context it seems that you want the charge to be zero, but for that you have to add the column with charges right after the column with the molecule IDs and not as the last column.

1 Like
3

Thank you, this fixed the problem. I placed the charge values at the end by mistake. Thank you for giving me the correct format.