The rationality problem for purely monomial group actions (Q657366)

Let \(G\) be a finite subgroup of \(\text{GL}_n(\mathbb Z)\) (the general linear group over the ring of integers) and \(k\) be a field. The group \(G\) induces an action on \(k(x_1, \cdots, x_n)\) (the rational function field of \(n\) variables over the field \(k\)) by defining \(\sigma (x_j)=\prod_{1 \leq i \leq n} x_i^{a_{i,j}}\) where \(\sigma=(a_{i,j}) \in G \subset \text{GL}_n(\mathbb Z)\) and \(G\) acts trivially on the base field \(k\). Such a multiplicative action is called a purely monomial action of the finite subgroup \(G\) of \(\text{GL}_n(\mathbb Z)\) on \(k(x_1, \cdots, x_n)\). The main question is to determine whether the fixed field \(k(x_1, \cdots, x_n)^G :=\{\alpha \in k(x_1, \cdots, x_n): \sigma (\alpha)= \alpha\) for any \(\sigma \in G \}\) is rational (i.e. purely transcendental) over \(k\). This question is ubiquitous in the study of various rationality problems in algebra, algebraic geometry and algebraic number theory. Unfortunately our knowledge about the answer of the above question is rather limited. It is known that, if \(n \leq 3\), then \(k(x_1, \cdots, x_n)^G\) is rational for any finite subgroup \(G\) in \(\text{GL}_n(\mathbb Z)\) (see the papers by Hajja, Hajja and Kang, Hoshi and Rikuna in the references of this paper). The paper under review provides another result about this question. Let \(n\) be any positive integer, \(G\) be a finite subgroup of \(\text{GL}_n(\mathbb Z)\) such that \(G\) is isomorphic to the direct product of \(n\) copies of \(C_2\) (the cyclic group of order two) as abstract groups. Note that we use the same integer \(n\) in the description of the group \(G\). Here is the main result of this paper. Theorem: If \(k\) is any field with char\(k \neq 2\), then \(k(x_1, \cdots, x_n)^G\) is rational. The idea of the proof is as follows. Assume that \(G_1\) and \(G_2\) are two such groups which are conjugate within \(\text{GL}_n(\mathbb Q)\). It is possible that they are not conjugate within \(\text{GL}_n(\mathbb Z)\). Hence we may write \(G_2=P^{-1}G_1 P\) for some \(P \in \text{GL}_n(\mathbb Q)\). We may also assume that all the entries of \(P\) are integers. In considering the fixed field \(k(x_1, \cdots, x_n)^{G_2}\), a subfield \(k(y_1, \cdots, y_n)\) of \(k(x_1, \cdots, x_n)\) is constructed with the aid of the transformation matrix \(P\) such that \(k(x_1, \cdots, x_n)^{G_2}\) is isomorphic to \(k(y_1, \cdots, y_n)^{G_1}\) (see Theorem 2.4). The matrix \(P\) may be simplified furthermore (see Lemma 3.1); it turns out that the field \(k(y_1, \cdots, y_n)\) constructed above is the fixed field of \(k(x_1, \cdots, x_n)\) under another group \(S\). Thus the rationality problem of \(k(x_1, \cdots, x_n)^{G_2}\) is reduced to that of \(k(x_1, \cdots, x_n)^{G_1}\). For details, see the proof on page 101.