Non-periodic boundary

[npiroozan]

Hello,

I hope you are doing well. I'm familiar with how to train a model with non-periodic boundary conditions in one direction.

Is this needed if one wants to run a LAMMPS simulation with a fixed boundary condition in one direction, using DeePMD as the pair_style?

Thank you very much. Just looking for clarification on this point.

[wanghan-iapcm]

It is not supported at the moment. One may work around to increase the vacuum layer in the direction of non-pbc.

[npiroozan]

I understand. My apologies, but how would one increase the vacuum layer?

[njzjz]

Just give a large box, e.g. 999, in that direction.

When running LAMMPS simulations, the boundary condition is handled by LAMMPS.

[npiroozan]

Understood, I thought perhaps something needed to change with respect to the training phase. Thank you!