A very efficient technique for the finite-basis-set calculation of logarithmic sums is introduced. The basis sets contain sequences of nonlinear parameters determined by the zeros of Laguerre polynomials. It is shown that one can easily obtain convergence to 12 digits in fast calculations involving small basis sets. The method is then applied to calculations of the asymptotic expansion of the Bethe logarithm for the Rydberg states of helium. As pertubation calculations of the sums involved diverge, the present method is used to obtain accurate results for atoms in dipole and quadrupole fields from which the adiabatic contributions are extracted. A discussion of gauges in the multipole case and the suppression of roundoK errors by working in a mixed gauge is presented. A strategy for handling the nonadiabatic contributions to sums is presented and used to obtain preliminary bounds on these contributions. PACS number(s): 31.30.Jv, 31.20.Di, 31.50.+w
I. THE BETHE LOGARITHMIn the past few years, experimental measurements [1] and theoretical predictions [2] of transition frequencies among the Rydberg states of helium have advanced substantially in accuracy to the point that they are sensitive to both /ED contributions and Casimir-Polder retardation corrections. The largest uncertainty in the theoretical predictions was due to the uncertainty in the selfenergy contribution to the energy, specifically the Bethelogarithm contribution to the Lamb shift. Results for the variational calculation of the electric dipole polarizability contribution to the Bethe logarithm were published recently [3], eliminating the gap between the uncertaintainties in the theoretical predictions and experimental results and pointing to discrepancies between both that cannot be accounted for by residual Casimir-Polder retardation corrections [4]. It is still necessary to add to that calculation the electric quadrupole and (of the same order) the nonadiabatic dipole polarizability contributions to the Bethe logarithm in order to have an estimate of the uncertainty in the theoretical results. In this paper we present the details of the difficult calculations that yield the dipole corrections to the Bethe logarithm, as well as results for the quadrupole contribution and an estimate of the nonadiabatic dipole correction with a discussion of the difficulties one encounters in this last case.Following Kabir and Salpeter [5], the lowest-order Lamb shift for a two-electron ion in an nI.S state can be written as b, EI", = n'Z D -+ In-[(crZ) '] 3' 30 + ln[Z A~/k(nLS, Z)]+2.296vra 2 + -CM where D = , (b(r, ) + b (r2)) is independent of Z in the hydrogenic limit and (p/M)CM involves finite-mass terms. There is in (1) a term that is difficult to calculate: the two-electron Bethe logarithm (BL from now on) defined for a state with energy Eo by