On some functions defined by the canonical expansion of complex numbers (Q1820189): Difference between revisions

A Gaussian integer \(\vartheta\) is called a canonical number base if every Gaussian integer \(\alpha\) can be represented uniquely in the form \[ \alpha =a_ 0+a_ 1\vartheta +...+a_ r\vartheta^ r,\quad a_ 0,a_ 1,...,a_ r\in \{0,1,...,N(\vartheta)-1\}. \] \textit{I. Kátai} and \textit{J. Szabó} [Acta Sci. Math. 37, 255--260 (1975; Zbl 0297.12003)] proved that \(\vartheta\) is a canonical number base if and only if \(\vartheta\) is of the form \(\vartheta =-A\pm i\), \(A\in {\mathbb N}\). Let \(\vartheta\) be a canonical number base and denote by \(H\) the set of all complex numbers \(z\) that have at least one representation in the form \[ z=\sum^{\infty}_{k=1}a_ k \vartheta^{-k},\quad a_ k\in \{0,1,...,N(\vartheta)-1\}. \] The author proves the following result: Let \(F: H\to {\mathbb C}\), \(F(0)=0\) and assume that \[ F(z)=F\left(\sum^{\infty}_{k=1}a_ k \vartheta^{- k}\right)=\sum^{\infty}_{k=1}F\left(a_ k \vartheta^{-k}\right) \] holds for every \(z\in H\). Then there exist two constants \(c\) and \(d\) such that \(F(z)=cz+d\bar z\). The proof is elementary but very ingenious. In the introduction, a recent result of \textit{Z. Daróczy}, \textit{A. Járai} and \textit{I. Kátai} (see the preceding review Zbl 0614.10006) is misleadingly stated. It should be remarked that relation (1.3) holds for two different coprime numbers \(q\).