Comment on triples in BCCD(T)

[Andreas Hesselmann]

Hello,

I would like to mention a somewhat awkward
point concerning Brueckner calculations with
triples:
it seems that Molpro simply uses the diagonal
elements of the Fock matrix without diagonalizing
the partial Fock matrices f_occ,occ and
f_vir,vir as proposed by Handy et al.
(Chem. Phys. Lett. 164 (1989) 185).
I thus believe that one should use this scheme
with care, though we have found no
dramatic differences between the two
alternatives.
However, a more natural way to calculate
the triples would be to use canonical
Brueckner orbitals as proposed by 
Scuseria (Chem. Phys. Lett. 226 (1994) 251).
Though the generalized Fock matrix
appears as intermediate in the CC program,
I am not sure if one can make the program
work the way as in the above reference.
Any support is welcome...

Andreas



*********************************************************** 
* Andreas Hesselmann                                        *
* Institut fuer Theoretische Chemie                        *
* Heinrich-Heine Universitaet                              *
* Universitaetsstrasse 1 / 40225 Duesseldorf          *
* Phone:  +49-211 / 8111439                              *
* Fax:      +49-211 / 8113466                              *
* E-Mail:  andreas at theochem.uni-duesseldorf.de  *
***********************************************************