Abstract
Using the general anharmonic oscillator as a case study, we examine the coupled-cluster method (CCM) in some detail. Emphasis is specially placed on the accuracy of the standard ground-state energy calculation and the excitation spectrum as derived via the Emrich ansatz. We are particularly interested in problems that can arise from large differences between the exact wave function and its model counterpart that forms the starting point of the CCM. To this end we begin with a variational Hartree approximation applied to three particular anharmonic-oscillator Hamiltonians that contain, respectively, pure quartic, equally weighted cubic and quartic, and double-well perturbations. These are chosen to provide increasingly stringent tests for the CCM in the above sense. Considerable attention is paid to the variational description of the double-well case, where the various possible solutions to the Hartree equations are considered and a further calculation on the energy-level splitting is performed. The CCM is then used to improve systematically upon our chosen starting point and is shown, particularly for the ground state, both to produce extremely accurate results and to be rather resilient to even gross deficiencies in the starting wave function. The main problem is the double-well system, where the CCM shows its lack of intrinsic inbuilt tunneling mechanisms via the absence of level splitting. Even here the CCM still produces very accurate average energies for very deep wells. While clearly needing some modification in such extreme cases, the CCM is shown to be quite adaptable and robust with regard to inaccurate starting wave functions.