Can Nequip handle two unique atom types definition for the same atom (e.g Ow and Os)?

[asoyemi1]

Hello developers,

I am exploring summing a NequIP potential with a classical forcefield such as a simple LJ potential using the hybrid/overlay function in LAMMPS. Say, I have two kinds of oxygen atoms in the system (in water molecules and in a sulfate anion) and for this I defined Ow and Os in the data file. Can NequIP handle this?

[anjohan]

Hi,

Sorry for the delay.

Yes. If the chemical symbols defined in your training config have one type of oxygen, O, you specify the mapping from LAMMPS type to NequIP species with

pair_coeff * * model.pth H O O S ...

if hydrogen is type 1 and the two kinds of oxygen are types 2 and 3.