{SPAM}? {SPAM}? Re: [molpro-user] Localize virtual orbitals

[Gerald Knizia]

On Friday 30 May 2008 02:19, zhendong zhao wrote:
>>> I am trying to localize MOs using below input:
>>> [...]
>>> Finally the occupied MOs are well localized (this is default), but the
>>> virtual MOs does not. What should I do to fully localize the virtual
>>> MOs and examize the virtural MOs easily? 
>>> Appreciate your help.

> The gaussian program suites can generate well localized virtual
> orbitals, but I do not know how gaussian can do this work. The US gamess
> only generate well localized occupied orbitals too.

In general it is not possible to localize the virtual space in such a way that 
its orbitals are both orthogonal and local at the same time. This is an 
intrinsic property of the virtual orbitals, there is no way around that. 
This is why MOLPROs local methods excite directly into non-orthogonal (P)AOs 
instead of virtual orbitals: there is no other way to do it.
-- 
Gerald Knizia