CR-EOMCCSD(T) correction question

Just Got Here

11:07:13 AM PST - Tue, Jan 7th 2014
Dear NWChem users and developers, According to the original paper [Kowalski and Piecuch. J. Chem. Phys. 120, 1715 (2004)], CR-EOMCCSD(T) adds correction in a state-selective manner, i.e. to the energy of each state rather than the excitation energy (please correct me if I am wrong). However in NWChem, according to the manual, it adds a correction to the EOMCCSD energy of an excited state and THE SAME correction to the excitation energy. Should the correction to the ground state also be taken into account if we want the correction to the excitation energy?

Clicked A Few Times

10:34:05 AM PST - Wed, Jan 8th 2014
The CR-EOMCCSD(T) approach as implemented in NWChem should be used to calculate vertical excitation energies (VEEs) only (VEEs are calculated with respect to the CCSD ground state energy). No ground-state corrections are added in calculating CR-EOMCCSD(T) excitation energies. As discussed in JCP 130, 194110 (2009) neglecting ground-state corrections is justified by assuming cancellation of certain types of contributions to the ground and excited state energies (see the nesting procedure). In general (this is my opinion) proper balancing of the ground- and excited-state correlation effects in non-iterative can be quite challenging. Therefore several non-iterative EOMCC methods accounting for triple excitations assume that the ground state is fully correlated at the CCSD level of approximation (please see the introduction of JCP 130, 194110 (2009) paper). Best, Karol