My question is about the “shell-only optimization” tag in GULP. Does this tag take the coordinates and gradients of shells alone for optimization? Also, If I give the “opti” tag along with the “shell”, will the optimization include the contribution of core part?

What is the effect of using shell optimization for defect studies? And for what kind of problems the shell-only optimization will be useful? It would be helpful if any references on this are shared.

Shell only optimisation does indeed mean just that - only the shells are optimised and nothing else. Think of shells as being like the electrons in a system (it’s a crude mimic, but it helps). If you have a shell model then a shell-only calculation is like a single point energy in a QM method where you have to do SCF to converge the electron distribution in order to determine the total energy for a set of core positions. Same thing for shells.

I have one more question. The optimized values ( lattice parameters, bulk modulus etc.,) are different for the case where shell opti is used and when it is not used. Is this normal or there is problem with my potential parameters?

Also, is there any visualization software for GULP compatible with windows 10 available (non-commercial)?

It’s nothing to do with potential parameters. The bulk modulus and optimised lattice parameters are usually for a fully relaxed unit cell, so it makes no sense to compute them with only the shells relaxed.