Re: Localized orbitals/CAS

Posted by Björn Roos on May 29, 2002 at 07:59:16:

In Reply to: Localized orbitals/CAS posted by Pablo Wessig on May 02, 2002 at 14:37:43:

: Does anybody know how to localize SCF orbitals for a subsequent RASSCF calculation? If the wavefaunction is defined by localized orbitals it is much easier to choose the correct active space.
: Thanks for any help.
: Pablo

Dear Dr. Wessig,
Interesting question. There is no localization code in MOLCAS. If anybody is interested to write one we would certainly include it. I once used the follwoing trick for ThF4+ with one hole in an F2p orbital. A charge was placed outside one of the F atoms. The calculation led automatically to localized orbitals. The charge was then removed and the calculation was repeated. The orbitals stayed localized with the 2p hole on one F atom only! I repested this for all four atoms and then ran a RASSI to reclaim symmetry.

So, the general answer is: destroy symmetry and hope it helps (it will not always of course). Apart from this special case, I have never felt that I needed to localize the orbitals to be able to defined the active space.

Follow Ups:

Post a Followup

Name:
E-Mail:

Subject:

Comments:
: : Does anybody know how to localize SCF orbitals for a subsequent RASSCF calculation? If the wavefaunction is defined by localized orbitals it is much easier to choose the correct active space. : : Thanks for any help. : : Pablo : Dear Dr. Wessig, : Interesting question. There is no localization code in MOLCAS. If anybody is interested to write one we would certainly include it. I once used the follwoing trick for ThF4+ with one hole in an F2p orbital. A charge was placed outside one of the F atoms. The calculation led automatically to localized orbitals. The charge was then removed and the calculation was repeated. The orbitals stayed localized with the 2p hole on one F atom only! I repested this for all four atoms and then ran a RASSI to reclaim symmetry. : So, the general answer is: destroy symmetry and hope it helps (it will not always of course). Apart from this special case, I have never felt that I needed to localize the orbitals to be able to defined the active space.