Sur la notion de fonction multiplicative et quelques problèmes qui lui sont associés. II. (On the notion of multiplicative function and some problems associated with it. II) (Q1090705): Difference between revisions

[Part I, cf. ibid. 61, 228-231 (1985; Zbl 0579.10023).] The author uses some results from his book ''Intégration et théorie des nombres'' (Paris 1986) to give proofs for the following results on multiplicative functions \(f: {\mathcal N}\to {\mathbb{C}}\), defined on a set \({\mathcal N}\) of Beurling numbers (for which \(Z(S)=\sum_{n\in {\mathcal N}}N(n)^{-s}\) has abscissa of convergence equal to 1). (1) If \[ (*)\quad \lim_{k\to \infty}\limsup_{\sigma \to 1+}\quad Z(\sigma)^{-1} \cdot \sum_{| f(n)| >K}\frac{| f(n)|}{N(n)^{\sigma}} = 0 \] and if \[ (**)\quad \limsup_{\sigma \to 1+}\quad Z(\sigma)^{-1} \cdot \sum_{n\in {\mathcal N}}\frac{| f(n)|}{N(n)^{\sigma}} > 0 \] then f is equi-summable \(\{\) that means: \(\forall \epsilon >0\exists \eta >0\) such that \(\forall a {\mathcal N}\to {\mathbb{C}}\) \[ \lim_{\sigma \to 1}Z(\sigma)^{-1} \cdot \sum \frac{| \gamma (n)|}{N(n)^{\sigma}} < \eta \text{ implies }\limsup_{\sigma \to 1}Z(\sigma)^{-1} \cdot \sum\frac{| \gamma (n)| \cdot | a(n)|}{N(n)^{\sigma}} < \epsilon, \] where \(\gamma (n)=\sum_{d}a(d)\cdot I_ d(n)\), \(I_ d(n)=1\) if d \(| n\), \(I_ d(n)=0\) otherwise\}. (2) If for any \(c>1\) \[ \sup_{x>0}\sum_{e^ x\leq N(p)\leq e^{cx}}1/N(p)< \infty, \] and if f satisfies (*) and (**), and if \(\lim_{s\to 1}Z(s)^{-1}\cdot \sum f(n)/N(n)^ s\neq 0\) exists, then four Elliott-type series like \[ \sum \frac{1-f(p)}{N(p)},\quad \sum_{| f(p)| <2}\frac{| 1-f(p)|^ 2}{N(p)},... \] are convergent. (3) An Elliott-Daboussi theorem on the existence of mean-values.