[molpro-user] Selection of certain orbitals to correlate

[Peterson, Kirk]

Dear Quan,

the much more straightforward route is to use the rotate command from within the ccsd(t) calculation itself and leave your hf input unchanged, e.g.,

{ccsd(t)
core,16,9,9,2,9,2,2,0
rotate,9.1,16.1
rotate,4.2,9.2
rotate,4.3,9.3
rotate,4.5,9.5
maxit,50
shift,1
}


best regards,

-Kirk


> On Oct 31, 2014, at 2:18 PM, Quan Phung <manhquank9 at gmail.com> wrote:
> 
> Dear Molpro users,
> 
> Let say I have HF orbitals of a molecule and I want to to CCSD(T) of this molecule.
> Now I would like to correlate just some orbitals. including some valance (but not all valance) and some low-lying core orbitals. So I have to rotate the core orbitals with the valance orbitals that I do not want to correlate and do only one iteration. For example:
> 
> {hf
> wf,108,1,0
> shift,-1.0,-0.5
> maxit,1
> start,2101,2
> closed,18,9,9,3,9,3,3,0
> occ,   18,9,9,3,9,3,3,0
> rotate,9.1,16.1,90
> rotate,4.2,9.2,90
> rotate,4.3,9.3,90
> rotate,4.5,9.5,90
> 
> }
> where 9.1, 4.2, 4.3, 4.5 are core orbitals and 16.1, 9.2, 9.3 and 9.5 are valance orbitals that I do not want to correlate.
> After that I do a CCSD(T) calculation based on this HF calculation.
> {ccsd(t)
> core,16,9,9,2,9,2,2,0
> maxit,50
> shift,1
> }
> 
> I would like to ask if it is make sense to do this? 
> I also notice that after rotation and do the HF calculation, the core orbitals (9.1, 4.2, 4.3, 4.5)  are now high-lying and have the energy equal the the energy of the valence orbitals that were rotated. I am not sure if this is correct. To be clear:
> Before rotation (full HF calculation):
> ORBITAL  9.1 (core)    eig=   -4.4491
> ORBITAL 16.1 (valance) eig=   -0.9945
> 
> 
> After rotation, HF one iteration:
> ORBITAL  9.1 (valance)  eig=   -4.4491
> ORBITAL 16.1 (core)     eig=   -0.9945
> 
> 
> Best regards,
> Quan 
> 
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user