[molpro-user] CCSD density matrix/natural orbitals

[Terry Frankcombe]

Hi folks

I'd like to calculate approximate natural orbitals for an open shell CCSD
wave function (triplet and open shell singlet states).  These would then
be used as starting orbitals for an MCSCF calculation.

The manual says no.  (Or at least says "only works for closed-shell MP2
and QCISD.)

However, I note that the MATROP facility can calculate natural orbitals
from a density matrix.  The manual also seems to suggest that I can't get
the density matrix, but there are some rather cryptic comments about
EXPEC,DM directives.  (All this is from the closed-shell CCSD section of
the manual anyway... the open-shell CCSD section of the manual is rather
bereft of details.)

Are there any back-door ways to convince molpro to give me open shell CCSD
natural orbitals/density matrices?  If so, can anyone point to some
fully-worked example code?

Regards
Terry