dipole and tdm moment in MRCI; definitions, finite field vs. expec

[Marcin Gronowski]

Dear Molpro expert,

I have performed computations for for diatomic metal fluoride. The dipole moment computed as the expectation value from MRCI wavefunction is about 1 a.u. (with negligible sensitivity to the active space, basis set etc.). The same dipole moment obtained from the finite field at both CCSD(T) and MRCI is about 20% bigger (in agreement with the experiment and relaxed CCSD(T) value from the Cfour code).

I am asking you for help in understanding this difference. How Molpro calculates the dipole moment in MRCI? I assume that MRCI uses analytical energy derivatives rather than directly calculating the <1.1|mu|1.1> integral. Or so? Do we have any control over orbital relaxation? What are the differences between the procedures for calculating dipole moments and transition moments?

I will be grateful for your help,
All the best

Marcin

p.s. This is relevant part of the output

 !MRCI STATE 1.1 Energy              -136.519347820203

 Properties without orbital relaxation:

 !MRCI STATE 1.1 Dipole moment         -0.00000000     0.00000000    -1.02307971

(...)

 !MRCI trans          <2.1|DMY|1.1>    -2.388669344393 au =    -6.071391881739 Debye

 !MRCI trans          <3.1|DMX|1.1>    -2.388976457209 au =    -6.072172484655 Debye