The generalized log-derivative method for evaluation of second-order transition amplitudes

Abstract

The log-derivative method of Johnson is generalized to calculate matrix elements of multichannel Green’s functions—second-order transition amplitudes—which arise from description of a variety of physical processes involving weak interactions of initial and final (bound) states with a set of strongly coupled continuum and/or bound intermediate states. A purely approximate-solution algorithm and two hybrid approximate-solution approximate-potential versions, based on the use of piecewise constant reference potentials, are presented and tested on problems concerning investigations of nonadiabatic effects in the spectroscopy of H2. A comparison with the renormalized Numerov method, extended to calculation of considered transition amplitudes, is made and superior efficiency of the hybrid log-derivative algorithms is demonstrated. It is shown both practically and theoretically that discretization errors of the hybrid algorithms grow linearly with increasing energy in calculations, whereas cubic growth of errors with energy is characteristic for the purely approximate-solution log-derivative and Numerov algorithms.