On the arithmetic structure of the integers whose sum of digits is fixed

From MaRDI portal





Given an integer \(g\geq 2\), consider the \(g\)-additive function which assigns to each integer its sum of digits in base \(g\). The arithmetic structure of the integers whose sum of orbits belongs to a given residue class was studied by \textit{A. O. Gelfond} [Acta Arith. 13, 259-265 (1968; Zbl 0155.09003)] and continued by \textit{C. Mauduit} and \textit{A. Sárközy} [J. Number Theory 61, 25-38 (1996; Zbl 0868.11004)]. The aim of the paper under review is to obtain similar results for the sets of integers whose sum of digits is fixed, although these sets have a much smaller cardinality. Let \(V_k(N)\) (respectively \(U_{(m,r)} (N)\) denote the subset of \(\{1,\dots,N\}\) consisting of the numbers \(n\) satisfying \(S(n) =k\) (respectively \(S(n) \equiv r\pmod m)\). The authors give first sufficient conditions for \(V_k(N)\) to be well-distributed in the residue classes. In particular, this is the case for the residue classes of small moduli, if \(k\) tends towards infinity. The proof of this result requires a precise estimate of a uniform lower bound (in \(k)\) for the cardinality of \(V_k(N)\). Then the authors deduce \(V_k\) analogues from Gelfond's results [op. cit.]. Furthermore, they give an Erdős-Kac type theorem for \(V_k(N)\) when \(k\) is close to its mean value \({g-1 \over 2} {\log N \over \log g}\). This proof uses a nontrivial application of the large sieve and is more difficult than the proof of the corresponding theorem for the sets \(U_{(m,r)} (N)\). The authors do not obtain analogues to every theorem of their paper cited above, however they do state for each result a corresponding conjecture or a weaker version.




Cited in
(41)








This page was built for publication: On the arithmetic structure of the integers whose sum of digits is fixed

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4345851)