Using pair_style hybrid/overlay in the lammps Python module

Dear LAMMPS developers,

I am using the LAMMPS Python module to perform energy minimization calculations for a system containing two types of atoms (Carbon). I am employing the ‘rebo’ and ‘drip’ pairwise potentials to relax my system.

To start, I compared the initial forces and per-atom potential energy of my calculation to an existing reference result. Both calculations use exactly the same input file and structure, although there is a difference in the LAMMPS versions. I am using the LAMMPS (3 Nov 2022) version Python module, while the reference result was obtained using an older version of LAMMPS (23 Jun 2022) on a Linux system.

When I perform the calculation using either the ‘rebo’ or ‘drip’ pairwise potential individually with the following commands:

For ‘rebo’:
pair_style rebo
pair_coeff * * CH.rebo C C

For ‘drip’:
pair_style hybrid/overlay drip
pair_coeff * * drip C.drip C C

The initial forces and per-atom potential energy match exactly with the reference. However, when I attempt to use both pairwise potentials together using the following commands:

pair_style hybrid/overlay rebo drip
pair_coeff * * rebo CH.rebo C C
pair_coeff * * drip C.drip C C

The calculated result is far from reality.

I would like to know whether I need to install any additional supporting libraries for my LAMMPS Python module to ensure that the pairwise potentials are correctly read when used together.

Thank you in advance.

There is no difference in execution whether you run LAMMPS from Python or as a standalone executable, for as long as you use the exact same input and the exact same files. In fact, it is the same binary code that will be executed in the same way.

I am not aware of any significant changes to the pair styles mentioned that would explain the discrepancy you are describing, so in all likelihood, the cause is somewhere else. There is insufficient information to make any assessment of what that could be. Most certainly, there is nothing required to install for python.

Please also note, that the “gold standard” is now the 2 August 2023 version.