[molpro-user] MRAQCC closed orbital occupancies

Lorenzo Lodi l.lodi at ucl.ac.uk
Tue Jun 24 16:05:41 BST 2014


Dear David,
an AQCC (or MRCI) wave function is a sum of Slater derminants, each of
them specified by orbital occupancies.
Natural orbital occupancies less than 2 indicate than in some of the
derminants those orbitals are not
doubly occupied; in practice an exact 2 occupancy implies those orbitals
are not correlated, ie, they are `CORE' orbitals. Orbitals which are
CLOSED are not included in the active space but *are* correlated at the
CISD level, i.e. determinants with single and double excitations out of
those orbitals *are* included in the AQCC expansion, hence their
occupancies will be less than 2.
Having virtual orbitals with non-zero occupancies is also normal and
indicates that some excited configuration in which the orbitals in
question are occupied has a relatively high weight in the AQCC expansion.

Lorenzo


On 17/06/14 02:32, David Scarborough wrote:
> Hello,
>  
> I'm running a MRAQCC single point energy calculation with the
> following input:
>  
> {mcscf,maxit=100
> occ,26
> closed,20
> wf,nelec=46,spin=0,charge=0,sym=1
> print,orbitals,virtuals
> cpmcscf,hess
> }
>  
> {aqcc,maxit=250
> core,6
> wf,nelec=46,spin=0,charge=0,sym=1
> natorb,print=20
> }
>  
> The natural orbitals from the AQCC step suggest that the six core
> orbitals have double occupancy but the remaining 14 closed orbitals do
> not. Also the virtual orbitals from number 27 onwards have non-zero
> occupancies. Is there an error in my input that is resulting in the
> closed orbitals not maintaining an occupancy of two? Why are the
> virtual orbitals outside the active space occupied?
>  
> I obtain the same solution whether I include the occ and closed
> commands in the AQCC block or not.
>  
> Thank you for your help.
> David.
> Postgraduate research student
> david.l.scarborough at gmail.com <mailto:david.l.scarborough at gmail.com>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20140624/c8736ea6/attachment.html>


More information about the Molpro-user mailing list