Characterization of rational matrices that admit finite digit representations (Q1790477)

Let \(A\) and \(\digamma\) denote, respectively, an \(n\times n\) matrix with rational entries, and a finite subset of the \(\mathbb{Z}\)-module \[\mathbb{Z}^n[A]:=\bigcup_{k\in \mathbb{N}}(\mathbb{Z}^n+A\mathbb{Z}^{n}+\cdots +A^{k-1}\mathbb{Z}^n).\] Then, the pair \((A,\digamma)\) is said to satisfy the finiteness representation property, in short FRP, if each element \(v\) of \(\mathbb{Z}^n[A]\) may be written \(v=d_0+Ad_1+\cdots +A^{k-1}d_{k-1}\), for some \(\{d_0, d_1,\ldots,d_{k-1}\}\subset \digamma\) and \(k\in \mathbb{N}\). In [Congr. Numerantium 98, 199--212 (1993; Zbl 0801.05023); SIAM J. Discrete Math. 6, No 3, 501--521 (1993; Zbl 0826.52019)], \textit{A. Vince} considered the particular case where \(A\) has integer entries, and showed that there is a (resp. there is no) subset \(\digamma \) of \(\mathbb{Z}^n[A]=\mathbb{Z}^n\) such that \((A,\digamma )\) satisfies FRP, whenever \(A\) has no (resp. has an) eigenvalue in the closed (resp. in the open) unit disc. Using a related theorem about algebraic numbers satisfying the height reducing property [Indag. Math., New Ser. 26, No. 1, 24--27 (2015; Zbl 1308.11096)], the authors of the paper under review complete totally Vince's result, by proving that for a fixed matrix \(A,\) there exists a set \(\digamma\) such that that \((A,\digamma)\) satisfies FRP if and only if the eigenvalues of \(A\) are all of modulus at least \(1\). A simplified proof of this theorem may be obtained, without mentioning the notion of block-digit systems.