The leading computational QED uncertainty of the Lamb shift in the $1s$ state in the hydrogen atom and helium ion is due to two-loop pure self-energy contributions. An exact evaluation in $Z\ensuremath{\alpha}$ has not yet been performed for $Z=1,2$, while a calculation based on the $Z\ensuremath{\alpha}$ expansion has a limited accuracy. To improve it, we calculate a leading logarithmic contribution in order ${\ensuremath{\alpha}}^{2}{(Z\ensuremath{\alpha})}^{8}m$ and estimate the nonleading subterms in orders ${\ensuremath{\alpha}}^{2}{(Z\ensuremath{\alpha})}^{7}m$ and ${\ensuremath{\alpha}}^{2}{(Z\ensuremath{\alpha})}^{8}m$. To further improve the accuracy, we take advantage of a combined evaluation of the numerical results of the exact calculations in $Z\ensuremath{\alpha}$ performed at $Z=10--30$, and the results of the $Z\ensuremath{\alpha}$ expansion in order to reach the most accurate result, which can be done by fitting the data. The quality of the fit, including its reliability and accuracy, strongly depends on a theoretical input, which is the main target of this paper. Strictly speaking, we develop not a fit, but a theoretical approximation for the pure self-energy contribution, which may serve as a base for a robust fit. It is also important to estimate a deviation of the truncated function from the true one. For this purpose, we calculate the leading logarithmic term and estimate the subleading ones in order ${\ensuremath{\alpha}}^{2}{(Z\ensuremath{\alpha})}^{9}m$. That allows us to find a reliable result which is more accurate than both fits over the numerical data and the results from the $Z\ensuremath{\alpha}$ expansion, which have been previously considered in literature. Performing the fit procedures, we focus our attention on the theoretical input which plays an essential role in the reliability and accuracy of the fit. Our results for the two-loop pure self-energy contribution to the $1s$ Lamb shift are expressed in a standard notation as ${B}_{60}=\ensuremath{-}72(7),\phantom{\rule{0.28em}{0ex}}{G}_{60}(Z=1)=\ensuremath{-}80(6)$, and ${G}_{60}(Z=2)=\ensuremath{-}83(5)$. That significantly improves the theoretical accuracy of the $1s$ Lamb shift in hydrogen, deuterium, and in the helium ions.
Read full abstract