[molpro-user] Computation of the dipole moment with CCSD(T)

[Lorenzo Lodi]

Thanks everyone for the exaustive replies. I see that things are not 
completely straighforward and it'll take me a while to read some of the 
suggested literature.
By the way, I must specify that my goal is to calculate the dipole to an 
accuracy of 1.e-4 so that even "small" differences matter to me.

I am also doing calculations with the MRCI (and ACPF, AQCC) methods and 
I have some doubts here as well...
How are the dipole values reported as "!MRCI STATE 1.1 DIPOLE MOMENTS" 
calculated? Are they equivalent to the expectation value of the dipole 
on the MRCI wavefunction?

I calculated the dipole with the finite-field approach (near equilibrium 
geometry, aug-cc-pCVDZ basis set, 6-orbitals CAS) and I found for the 
MRCI dipole a difference with the value reported by molpro of about 
5.e-3 a.u.
(I thorougly adjusted the various threshold parameters and the field 
strength so the difference is not a numerical artefact)

Is there any theoretical reason why either of these values should be 
considered "sounder" than the other?
Would it make sense to take the difference between the two values as 
some estimate of the error in the dipole?

Thanks again