SOPPA with the Complete Third-Order A Matrix: Singlet Excitation Energies

Abstract

We present the derivation and implementation of the complete third-order correction of the A matrix in the Møller-Plesset polarization propagator approximation. A new contribution to the third-order correction was discovered and we thus report the first results obtained with the complete third-order A matrix. The performance of the Second Order Polarization Propagator Approximation including the third order A matrix (SOPPA-A3) method in calculation of vertical excitation energies was assessed by comparing to the reference methods FCI or CCSDTQP. A substantial improvement over SOPPA was found with the inclusion of the complete third-order A matrix as the SOPPA-A3 method yields excitation energies of a quality comparable to CC2. The inclusion of the new contribution results in a change in the individual excitation energies of varying size and direction depending on the molecule and state and the contribution should therefore be included for a formally complete third-order treatment of the A matrix.

arXiv comment33 pages, 2 figures