Applying manybody potential in Hybrid pair style

Dear Lammps user,
I am trying to simulate hybrid potential for binary alloy. I want to take A-A interaction from eam/alloy whereas A-B, and B-B interaction from eam/fs. It is always recommended not to mix and match manybody potential. Even though I am just testing few setup. Kindly comment whether below methodology is genuine.

pair_style hybrid eam/fs eam/alloy
pair_coeff * * eam/fs AB.eam.fs A B
pair_coeff 1 1 none
pair_coeff * * eam/alloy A.eam.alloy A NULL

Regards

Pranav Kumar

It is bogus.

The only way to make this work is to build a new .eam.alloy potential file that combines the properties as needed.

Thanks @akohlmey for quick reply, It will be great if you can quickly comment about this also, in case single potential is unavailable.

pair_style hybrid eam/fs eam/alloy lj/cut 10
pair_coeff * * eam/fs B.eam.fs NULL B
pair_coeff * * eam/alloy A.eam.alloy A NULL
pair_coeff 1 2 lj/cut 1.0 1.0

Bogus, too.

You are ignoring core properties of the EAM model: every time you use multiple EAM pair styles in a hybrid setup, the embedding term will be incorrectly computed. Each instance will compute and apply only the contributions from its assigned atom types, but you would need the sum across all instances and then apply that. This is not possible. If you want to have potential files with combinations of elements that don’t exist, you have to learn how to generate them. I am so adamant in telling people not even to try this kind of mix-and-matching of potentials for good reasons, not because I enjoy making people’s lives uncomfortable.

Or in short: there ain’t no escape from the blues.

Thanks again and I agree with you in all front !