We present further evidence that closed-shell single-reference coupled cluster theory can be used as a reliable predictive tool even when multiple chemical bonds are broken, provided that a near-variational, rather than a projective, ansatz is used. Building on the Optimized-orbital Quasi-Variational Coupled Cluster Doubles (OQVCCD) method by adding the standard perturbative treatment of triple excitations, the OQVCCD(T) method provides outstanding accuracy for the dissociation of multiply bonded molecules and other problems involving strong nondynamic correlation of the electrons. We find that in the case of singly bonded molecules, OQVCCD and OQVCCD(T) perform similarly to the equivalent Brueckner Coupled Cluster Doubles approaches, BCCD and BCCD(T). However, when multiple bonds are broken, such as in acetylene and dicarbon, OQVCCD(T) is capable of predicting both qualitatively and quantitatively accurate potential energy curves, unlike the standard methods based on traditional coupled cluster theory, and for approximately the same computational cost.

We enhance the recently proposed Optimized-orbital Quasi-Variational Coupled Cluster Doubles (OQVCCD) method for the calculation of ground-state molecular electronic structure by augmenting it with the standard perturbative (T) correction for the effects of connected triple excitations. We demonstrate the OQVCCD(T) ansatz to be outstandingly robust and accurate in the description of the breaking of the triple bond in diatomic nitrogen, N2, where traditional CCSD and CCSD(T) completely fail, yet with a computational cost that is nearly the same as that of CCSD(T). This result provides insight into the failure of CCSD(T) and related methods and how it may be overcome.

We enhance the recently proposed Optimized-orbital Quasi-Variational Coupled Cluster Doubles (OQVCCD) method for the calculation of ground-state molecular electronic structure by augmenting it with the standard perturbative (T) correction for the effects of connected triple excitations. We demonstrate the OQVCCD(T) ansatz to be outstandingly robust and accurate in the description of the breaking of the triple bond in diatomic nitrogen, N2, where traditional CCSD and CCSD(T) completely fail, yet with a computational cost that is nearly the same as that of CCSD(T). This result provides insight into the failure of CCSD(T) and related methods and how it may be overcome.