I would like to run a MCSCF procedure with maximasing rotations in the configuration space and minimising the rotations of the orbitals. The aim is to approximate non-adiabatic couplings with numerical differenciation while neglecting orbital rotations.
Maybe it is possible to project the final set of active orbitals (after MCSCF convergence) on the initial set (initial guess) and apply the corresponding "backrotation" in the configuration space.
This procedure is also known as orbital diabatization in some other codes.
Is there something similar in Firefly?
Then, is there a way to output the full vector of the CI coefficients?
I also have the same type of questions with XMCQDPT2 on top of the MCSCF.