Dear Chan,
I would suggest you to define two systems, a real one whose wave-functions are
the one corresponding to the physical system you investigate and a second
auxiliary system in which you define the special two site dependent function as
a hopping in the Hamiltonian. Therefore, computing the current in the second
system using the wave functions of the first one should correspond to what you
wanted to do. For the 'onsite' function I think it doesn't accept two site
dependence.
I hope this makes sense,
Regards,
Ousmane
