On the binary additive divisor problem (Q1661305)

The binary additive divisor problem is the study of the sum \[ \sum_{n=1}^Md(n)d(n+f)=MT(M,f)+E(M,f), \] where \(d(n)\) stands for the number of positive divisors of \(n\), \(f\) is a positive integer, \(T(M,f)\) is an explicit main term, closely related to a certain quadratic polynomial in \(\ln M\) with coefficients depending on \(f\), and \(E(M,f)\) is a remainder term. The authors first recall the historical development of this study, mentioning in particular the contributions [\textit{Y. Motohashi}, Ann. Sci. Éc. Norm. Supér. (4), 529--572 (1994; Zbl 0819.11038)] and [\textit{T. Meurman}, in Proceedings of the Turku symposium on number theory in memory of Kustaa Inkeri, Turku, Finland. Berlin: de Gruyter, 223--246 (2001; Zbl 0967.11039)], which give uniform estimates of \(E(M,f)\), the best ones to this day. Next, having observed that Meurman's statement is more precise than Motohashi's, the authors revisit the work of the latter, and prove that it can be sharpened to yield precisely Meurman's result. Their main contribution is a careful analysis of an integral used by Motohashi.