Using Hartree-Fock pseudopotentials in GW calculations

The issue of including shallow ``semi-core'' states as valence has recently resurfaced in the context of self-consistent GW calculations.\footnote{F. Bruneval \textit{et al}., Phys. Rev. Lett. \textbf{97}, 267601 (2006).} Supposing that semi-core-valence exchange is the dominant process necessitating the inclusion of semi-cores, we have investigated whether the use Hartree-Fock pseudopotentials\footnote{W. A. Al-Saidi, E. J. Walter, and A. M. Rappe, Phys. Rev. B \textbf{77}, 075122 (2008).} instead of density-functional psp's might obviate the need for semi-cores. The answers to this question appear to be ``yes'' for the case of CuCl (filled d shell), and ``semi-cores don't matter anyway'' for ScN (empty d shell). Opportunity permitting, additional examples will be discussed.