[molpro-user] May the RCCSD(T) read orbitals from the preceding CASSF calculation ?

[Gerald Knizia]

On Monday 13 December 2010 11:28, Peter Knowles wrote:
> Yes, this can be done:
>
> {rhf;occ,7,1}
> {casscf; occ,8,1;  wf, 15,1,1;state,2 ;  wf, 15,2,1}
> {Rccsd(t);orbital,ignore_error; occ,8,1; wf,15,1,1 }
>
> ... although I would consider it to be inadvisable since the (T) correction
> assumes that the reference wavefunction is RHF.

I'm just wondering: Doesn't UCCSD(T)/RCCSD(T) make semi-canonical orbitals for 
the triples correction? If yes, would this assumption still be a problem?
-- 
Gerald Knizia