Alex Granovsky
gran@classic.chem.msu.su
This is what should be expected. CASCI expansion coefficients depend
on the rotations in the subspace of active MOs while CASSCF/CASCI
energy does not. Orbitals used during geometry optimization are not
the same as orbitals found in the punch file. The former more or less
arbitrary while the latter are either natural orbitals or
canonicalized MOs.
Kind regards,
Alex Granovsky
On Thu Mar 3 '16 4:04pm, GrEv wrote
-----------------------------------
>Hello,
>I encountered some strange inconsistency in the MCSCF WF coefficients printed out during a geometry optimization and at a single point calculation. I mean I performed a geom. optimization using SA2-CAS(2,2) and got a stationary point where the CAS WFs for two states look the following:
> STATE # 1 ENERGY = -2511.331945658
> CSF COEF OCCUPANCY (IGNORING CORE)
> --- ---- --------- --------- ----
> 1 0.529262 20
> 2 -0.713775 11
> 3 0.458701 02
>
> STATE # 2 ENERGY = -2511.322269051
>
> CSF COEF OCCUPANCY (IGNORING CORE)
> --- ---- --------- --------- -----
> 1 0.758665 20
> 2 0.156070 11
> 3 -0.632511 02
>When I performed a single point calculation at the optimized geometry (to get the orbitals) I however got the WFS that differ from the above WFs (fortunately the energies are the same):
>
>
> STATE # 1 ENERGY = -2511.331945652
> CSF COEF OCCUPANCY (IGNORING CORE)
> --- ---- --------- --------- -----
> 1 0.991938 20
> 2 -0.126662 11
> 3 -0.003975 02
> STATE # 2 ENERGY = -2511.322269057
> CSF COEF OCCUPANCY (IGNORING CORE)
> --- ---- --------- --------- -----
> 1 -0.126661 20
> 2 -0.991946 11
> 3 0.000508 02
>I wonder why it is so, what is the reason of the difference.
>Thanx!
>Best regards,
>Evgeniy