ON THE CHOICE OF ORBITALS FOR SYMMETRY-BREAKING PROBLEMS WITH APPLICATION TO NO3

The suitability of using approximate Brueckner determinants in coupled-cluster (CC) studies of orbital symmetry-breaking problems is explored. At least for the cases considered here [NO3(2A2'), N2(2SIGMAg+) and NO2(2B2)], the Brueckner reference function, which by definition does not mix with single excitations in the CC wave function, preserves symmetry exactly. Furthermore, these solutions do not appear to be sensitive to the initial guess, as the same symmetric solutions are obtained starting with either symmetry-broken or symmetry-constrained initial guesses. The same behavior is not observed for Brueckner determinants obtained from configuration interaction calculations, where symmetry-broken solutions are found. An evaluation of the merits of basing CC calculations on Brueckner determinants [B-CC] and quasirestricted Hartree-Fock [QRHF] reference functions (which also can be chosen to preserve symmetry exactly) is presented. Calculations carried out on the NO3 radical predict a C2v equilibrium geometry at both the B-CC and QRHF-CC levels when triple excitations are not included. However, all methods which include an approximate treatment of triple excitation effects predict an extremely flat potential surface which slightly favors the symmetric [D3h] form of the molecule.