[molpro-user] ROTATE orbitals directive in CASSCF, keeping a stable active space for RASSCF, disabling orbital optimisation in CASSCF

[Leon Freitag]

Bartosz,

> 2. perform CASSCF with ROTATE and switching off the orbital optimization
> (this would put the CAS back into place. Orbitals don't need
> re-optimization as they're taken from step #1. Only the CI-vector needs
> diagonalization, to properly swap the coefficients.)

If you're rotating between inactive and active orbitals, the orbitals surely 
do need reoptimisation, since you're allowing excitations from the active 
space only, and now that you've changed the orbitals in the active space and 
electrons from these orbitals are allowed to be correlated, but previously 
they were not. Due to the inclusion of the correlation the orbitals may very 
well change their shape.

I've had a similar problem for a while, and I couldn't solve it, just work 
around it by using a different active space -- you might want to use a larger 
active space, but for me actually a smaller active space worked too. You may 
also try a lower convergence threshold for the orbital rotations, although I 
don't know how accurate is the solution obtained this way. Finally you might 
want to try a different guess, if you can obtain one somehow (maybe from a 
close geometry)
> 
> * MRCI with NOEXC and ROTATE
> This was so far the best "working" option.

MRCI would disable orbital optimisation, i.e. all orbital rotations, but the 
energy obtained with it probably won't be consistent with the CASSCF values, 
just because of the disabled optimisation.

Leon
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20110720/b742a086/attachment.asc>