Input of potentials for two body interactions


I have a question about the input for the two body terms of the Stillinger-Weber potential (sw2) in GULP. Let’ say, we have a system with two atom types: Si and O for example. Now, we know the two body-terms for Si-O and O-Si interactions should be same as they are the same bonds.
My question is:

Should I enlist both Si-O and O-Si interactions in the input file, albeit them being of equal value? Or should I just list one of them only (either Si-O or O-Si interaction)?

Thanks and please let me know if you have any sugeestions.

Shoieb Ahmed CHowdhury

Stillinger-Weber (sw2) is a standard 2-body potential and so Si-O should only be listed once since adding O-Si would duplicate the interaction as there is no difference in the potential depending on the order.

Dear Julian,

Thanks for the clarification. Two follow up questions:

  1. If I list both Si-O and O-Si interactions (equal values), is that an error? Like would GULP be adding them together or simply copy them for both interactions? I was kind of playing with it and saw I get different results (e.g. elastic constants) if I list only Si-O interaction vs listing both Si-O and O-Si interactions (albeit being equal values).

  2. Does the same concept extend to 3-body terms. For example: if I have 3 types of atoms in the system (let’s say Si,O, and C ). For that system, should I list only the Si-O-C interaction terms once and the Si-C-O interaction (they are the same angle) would be automatically copied by GULP? Or should I list both Si-O-C and Si-C-O terms? I wonder which is the correct way.

  1. It’s not an error to duplicate terms - it’s a choice. As long you as you are consistent with however the model is fitted then that’s fine.
  2. Three-body terms depend on the pivot atom & so Si-O-C is not the same as Si-C-O. However, C-O-Si would be unnecessary duplication.

Thanks for the clarification. Just to be clear, in GULP when defining SW3 interaction with “atom1 atom2 atom3”, isn’t atom 1 the pivot (centre) atom?

In GULP input, atom 1 is the pivot. When you wrote the angles as Si-O-C I took the dashes to be bonds and so the middle atom to be the pivot.