Factors of generalised polynomials and automatic sequences (Q1749026)

\textit{N. Rampersad} and \textit{J. Shallit} have proved [``Common factors in automatic and Sturmian sequences'', Preprint, \url{arXiv:1802.00325}] that an automatic sequence and a Sturmian sequence cannot have arbitrarily long common factors. The authors of the paper under review generalize this result by proving that if an automatic sequence and a generalized poynomial sequence have arbitrarily long common factors, then there exists a periodic sequence coinciding with the generalized polynomial sequence out a set of upper Banach density zero, and having arbitrarily long common factors with the automatic sequence. Let us recall that a generalized polynomial is an expression involving real constants, additions and multiplications, and the (possibly iterated) use of the floor function. The proof, instead of using Kronecker's theorem as in the Sturmian case, needs the Bergelson-Leibman theorem of representation of generalized polynomials (see [\textit{V. Bergelson} and \textit{A. Leibman}, Acta Math. 198, No. 2, 155--230 (2007; Zbl 1137.37005)]). The authors also show how to replace automatic sequences with weakly periodic sequences: a sequence \(x\) is called weakly periodic if for any arithmetic subsequence \(x'\), say \(x'(n) := x(an+b)\) for some positive integer \(a\) and some non-negative integer \(b\), there exist a positive integer \(q\) and two distinct non-negative integers \(r\) and \(s\) such that \(x'(qn+r) = x'(qn+s)\) (see [the authors, ``Automatic sequences and generalised polynomials'', Preprint, \url{arxiv:1705.08979}]). Finally note that Reference [4] has appeared (see [\textit{J. Byszewski} and \textit{J. Konieczny}, Trans. Am. Math. Soc. 370, No. 11, 8081--8109 (2018; Zbl 1443.11136)]).