Effective mean value estimates for complex multiplicative functions

From MaRDI portal





The need for a good estimate for the mean value \((1/x)\sum_{n\leq x}g(n)\) of a multiplicative function \(g\) arises in problems in various branches of number theory. Sharp and effective bounds have been established for certain classes of real non-negative functions \(g\); see, for example, Chapter 0 of the authors' book [Divisors (Cambridge, 1988; Zbl 0653.10001)]. In this paper, the authors consider the analogous problem for complex- valued multiplicative functions \(g\) such that \(| g(n)| \leq 1\) for all \(n\) and, for all primes \(p\), \(g(p)\) lies in an elliptical region \[ E(\delta,\phi)=\{z\in \mathbb{C} : (\hbox{Im}(e^{-i\phi}z))^ 2\leq \delta^ 2(1-\hbox{Re}(e^{-i\phi}z))^ 2)\} \] where \(0\leq \delta \leq 1\), \(0\leq \phi < \pi\). It is shown that, for such functions \(g\), \[ (1/x)\sum_{n\leq x}g(n)\ll \exp\bigl(-K\sum_{p\leq x}{1-\hbox{Re }g(p)\over p}\bigr) \] where \(K=K(\delta,\phi)\geq 0\) is defined uniquely and shown to be sharp in a sense that is made precise, and \(K>0\) if and only if \(0\leq \delta <1\). It is pointed out that the example \(g(n)=n^ i\) shows that the inequality above (with \(K>0\)) fails when the second condition on \(g\) is omitted.











This page was built for publication: Effective mean value estimates for complex multiplicative functions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3982362)