On numbers with given digit distributions (Q1104968)

For \(\alpha\) \(\in [0,1)\) and \(g\geq 2\), \(g\in {\mathbb{N}}\), let \(\Gamma _ g(\alpha)\) denote the set of all distribution measures of the base g digit expansion of \(\alpha\). If \(\alpha\) is of the form \(\alpha =.\alpha _ 1\alpha _ 2...\) for some increasing sequence \(a_ 1,a_ 2,..\). of natural numbers (expressed as base g digit blocks) let \(\alpha\) * be obtained by substituting \(a\) \(*_ n=c_ na_ n\) for \(a_ n\), where \(c_ 1,c_ 2,..\). belong to a given finite set \(M\subseteq {\mathbb{N}}.\) The authors prove a rather general result on the relation between \(\Gamma _ g(\alpha)\) and \(\Gamma _ g(\alpha\) *). If \(\alpha\) is normal, the theorem reduces to a result of \textit{J.-M. Dumont} and \textit{A. Thomas} [Ann. Fac. Sci. Toulouse, V. Sér., Math. 8(1986/87), No.3, 367-373 (1987; Zbl 0642.10049)] in which the same is true for \(\alpha\) *.