I’m unclear on how the CEPA(0) implementation considers singles. Is it analogous to Szabo’s MBPT equations:

-B^\dagger D B = E_{corr}

B = \braket{\Psi_0|H|\Psi_{ij}^{ab}}

D = \braket{\Psi_{kl}^{cd}|H|\Psi_{ij}^{ab}}

but with D including single/single and single/double terms and with B including single terms? The Wennmohs & Neese paper mentions an ambiguity in the treatment of singles, but I believe that ambiguity should go away if all the shifts are 0 anyway, right?