Rational approximation to values of \(G\)-functions, and their expansions in integer bases (Q1702681)

The authors' main result reads as follows. Assume \(F\in\mathbb{Q}[[z]]\setminus\mathbb{Q}(z)\) to be a \(G\)-function and \(t\in\mathbb{R}_{\geq0}\). Then there exist constants \(c_1,c_2,c_3,c_4\in\mathbb{R}_{>0}\) depending only on \(F\) (and \(t\) as well for \(c_3\)) such that the subsequent property holds. Let \(a\neq0\) and \(b,B\geq1\) be integers satisfying \(b>(c_1|a|)^{c_2}\) and \(B\leq b^t\). Then \(F(a/b)\) is irrational and, moreover, for any \(n,m\in\mathbb{Z}\) with \(m\geq c_3(\log b)/\log(|a|+1)\), the inequality \((\ast)\!: |F(a/b)-n/(Bb^m)|\geq1/(Bb^m(|a|+1)^{c_4m})\) holds. It should be pointed out that the constants \(c_1,\ldots,c_4\) are effectively computable since the proof of the theorem is based on an effective zero estimate of \textit{Y. André} [G-functions and geometry. Wiesbaden etc.: Friedr. Vieweg \&| Sohn (1989; Zbl 0688.10032)]. It should be also noted that \((\ast)\) implies an effective irrationality measure for \(F(a/b)\). As another consequence of their main theorem, the authors show, under the same hypotheses on \(F\), that for any \(\varepsilon\in\mathbb{R}_{>0}\) and any \(b,m\in\mathbb{Z}_{>0}\) large enough with respect to \(a,\varepsilon,F\), the inequality \(|F(a/b)-n/b^m|\geq1/b^{m(1+\varepsilon)}\) holds. This enables one to obtain a new and effective result on repetition of patterns in the \(b\)-ary expansion of \(F(a/b^s)\) for any \(b\in\mathbb{Z}_{\geq2}\) (if the integer \(s\) is large). In the erratum to the main article, the authors correct a minor mistake in their proof of the main theorem having no consequence on the rest of the conclusion.