Integer-valued polynomials over matrix rings of number fields (Q2666371)

For a natural number \(n\) and a number field \(K\) with ring of integers \(\mathcal{O}_{K}\), let \(M_{n}(\mathcal{O}_{K})\) (resp. \(M_{n}(\mathbb{Q})\)) be the set of \(n\times n\) matrices with entries in \(\mathcal{O}_{K}\) (resp. in \(\mathbb{Q}\)) and let \[ \mathrm{Int}_{M_{n}(\mathbb{Q})}(M_{n}(\mathcal{O}_{K})):=\{f\in M_{n}(\mathbb{Q})[x]\mid f(M_{n}(\mathcal{O}_{K}))\subseteq M_{n}(\mathcal{O}_{K})\}, \] where \(M_{n}(\mathbb{Q})[x]\) denotes as usual the polynomial ring in the variable \(x\) over \(M_{n}(\mathbb{Q})\). The domains \(\mathrm{Int}_{\mathbb{Q}}(\mathcal{O}_{K}):=\mathrm{Int}_{M_{1}(\mathbb{Q})}(M_{1}(\mathcal{O}_{K}))\) were introduced by \textit{K. A. Loper} and \textit{N. J. Werner} [J. Number Theory 132, No. 11, 2481--2490 (2012; Zbl 1281.13012)], and have been considered by several authors. In [J. Number Theory 171, 198--212 (2017; Zbl 1352.13014)], \textit{B. Heidaryan} et al. showed that a prime number \(p\) is a totally split prime in \(\mathcal{O}_{K}\) if and only if the polynomial \((x^{p}-x)/p\) belongs to \(\mathrm{Int}_{\mathbb{Q}}(\mathcal{O}_{K})\), and using this statement they also proved that if \(L\) and \(N\) are two finite normal extensions of \(\mathbb{Q}\) with \(\mathrm{Int}_{\mathbb{Q}}(\mathcal{O}_{L})=\mathrm{Int}_{\mathbb{Q}}(\mathcal{O}_{N})\), then \(L=N\). In the paper under review, the authors extend the two results mentioned above to the class \(\mathrm{Int}_{M_{n}(\mathbb{Q})}(M_{n}(\mathcal{O}_{K}))\), where \(n>1\). More precisely, they show that a prime number \(p\) is a totally split prime in \(\mathcal{O}_{K}\) if and only if \((x^{p^{n}}-x)(x^{p^{n-1}}-x)\cdots (x^{p}-x)/p \in \mathrm{Int}_{M_{n}(\mathbb{Q})}(M_{n}(\mathcal{O}_{K}))\) (if \(f(x)=a_{0}+a_{1}x+\cdots +a_{d}x^{d} \in \mathbb{Q}[x]\), then \(f\) may be identified with \(a_{0}I_{n}+a_{1}I_{n}x+\cdots +a_{d}I_{n}x^{d}\in M_{n}(\mathbb{Q})[x]\), where \(I_{n}\) is the identity matrix of size \(n\)). Using this equivalence and some related results from [\textit{J. V. Brawley} et al., Linear Algebra Appl. 10, 199--217 (1975; Zbl 0307.15004)], they also obtain that two finite normal extensions \(L\) and \(N\) of \(\mathbb{Q}\) are equal whenever \(\mathrm{Int}_{M_{n}(\mathbb{Q})}(M_{n}(\mathcal{O}_{L}))=\mathrm{Int}_{M_{n}(\mathbb{Q})}(M_{n}(\mathcal{O}_{N}))\). Furthermore, they show that for each number field \(K\) the ring \(\mathrm{Int}_{M_{n}(\mathbb{Q})}(M_{n}(\mathcal{O}_{K}))\) is not Noetherian, and from this they conclude that the sets \(\mathrm{Int}_{\mathbb{Q}}(\mathcal{O}_{K})\), sandwiched between \(\mathbb{Z}[x]\) and \(\mathrm{Int}_{\mathbb{Q}}(\mathcal{O}_{\mathbb{Q}})\), are non-Noetherian integral domains.