On Sat Sep 27 '14 3:18am, alex wrote
>I have some concerns about nbo module of firefly. The problem is that nbo’s starting orbitals doesn’t converge with mcscf functions. I did dozen of tests with my problem (CASSCF(6,7), NSTATE=2) and get neither converged after 2048 SCF steps if started from nbo guesses. I know about ‘look into orbitals’ and ’look again‘, but... The use of ‘MCSF optimized orbitals’ from that unconverged ffly runs resulted in perfect convergence (after less than 100 steps for 2 systems which I’ve tried)! Evidently my understanding of the chemical problem haven’t changed and I’ve selected active space in the same way (as far as it is possible). So the question is how to use nbo module for mcscf? Maybe one should pick some ‘wasted’ orbitals in addition to the active space to make nbo converging? Or maybe there is a need for preprocessing to convert nbo orbitals into something really useful?
>Thanks for sharing your experience,