there is nothing wrong with your calculations.
You was getting identical MCSCF energies because MCSCF stage
of calculations was identical in all three cases. Naturally,
with constant state-averaging, the energies of MCSCF or CI
states do not depend on the number of requested states.
Note, the state-averaging of MCSCF runs is controlled by
wstate array of the $det input group (using aldet code) or
the $gugdm2 group (guga code). The wstate and avecoe arrays
of $xmcqdpt group are only used to control state-averaging
in constructing semi-canonical Fock orbitals and orbital
energies for PT.
Hope this helps.
On Sat Jul 9 '11 11:31pm, mark huntress wrote
>I am running mcscf with xmcqdpt2, with 3 roots. Then I decided to do calculations with 2 and 5 roots, using the same orbitals from the original run, that I knew were good (I used moread, and the $vec group at the end). So when I ran the 2 root job, I only changed NSTATE=2, WSTATE(1)=1, 1 and KSTATE(1)=1, 1. The rest of the input was the same as the 3 root job, with the orbitals at the end. But I used inorb=0 for all cases. Depending on the number of roots I use, I get different xmcqdpt2 energies, but identical mcscf energies. Am I doing something wrong? How is it that the mcscf energies are identical, no matter how many states I include?