Using WCA and LJ interactions in the same LAMMPS simulation — best practice?

LAMMPS LAMMPS Beginners
1

Dear LAMMPS users,

I am simulating a system with three atom types (type-1, type-2, type-3), and I want to apply:

  • Lennard-Jones (LJ) interactions between type-2 and type-3
  • WCA interactions (purely repulsive LJ) between all other pairs

I understand that WCA corresponds to a truncated and shifted LJ potential at the cutoff rc=21/6σ≈1.12246r_c = 2^{1/6} \sigma \approx 1.12246, which makes the potential zero at the cutoff.

Initially, I tried this:

pair_style lj/cut 2.5
pair_coeff 2 3 1.0 1.0 2.5
pair_modify shift yes
pair_coeff 1 3 1.0 1.0 1.12246

But I’m now unsure whether pair_modify shift yes affects the WCA interaction (type 1–3) in a way that breaks its intended form.

My questions:

  1. Does pair_modify shift yes apply globally to all lj/cut interactions, including those with a short cutoff (like 1.12246 for WCA)?
  2. Will using it in combination with WCA effectively “double-shift” the interaction and distort the WCA potential?
  3. Is the recommended approach in this case to use pair_style hybrid to separately define LJ and WCA interactions?
  4. Any potential issues or performance considerations with using pair_style hybrid for this case?

I’d really appreciate clarification or best practice suggestions for this mixed interaction setup.

Thank you,
Gokul Upadhyay

2

Yes. Keep in mind the shift must be computed and applied separately for each pair of atom types.

Why should that be? When shifting a potential, you only modify the absolute value of the energy by a constant. That constant is lost when taking the derivative to get the forces.

I think that question is already answered by the previous answers.

It should be obvious that using pair style hybrid introduces overhead.

3

Is this correct ?

pair_style lj/cut 2.5

LJ (attractive + repulsive)

pair_coeff 2 2 1.0 1.0 2.5
pair_coeff 2 3 1.0 1.0 2.5

WCA (purely repulsive)

pair_modify shift yes
pair_coeff 1 1 1.0 1.0 1.12246
pair_coeff 1 3 1.0 1.0 1.12246
pair_coeff 3 3 1.0 1.0 1.12246

2/2

4

Please note that this forum is not an input file validation service.

You have to decide what you want by yourself. After all, what if I answer but am wrong?