Three-dimensional purely quasimonomial actions (Q2175457)

The paper under review aims to study the rationality problem of the multiplicative actions, i.e. the \(3\)-dimensional purely quasi-monomial actions. To begin with, we will provide an alternative definition of the purely quasi-monomial actions via \(G\)-lattices. Let \(G\) be a finite group. Recall that a finitely generated \(\mathbb{Z}[G]\)-module \(M\) is called a \(G\)-lattice if it is torsion-free as an abelian group. As an abelian group, \(M\) becomes a free abelian group, say, of rank \(n\); the same integer \(n\) is called the rank of the \(G\)-lattice \(M\). Let \(M=\bigoplus_{1\le i\le n} \mathbb{Z}\cdot e_i\) be a \(G\)-lattice of rank \(n\), \(K/k\) be a finite Galois extension field such that there is a surjection \(G\to \mathrm{Gal}(K/k)\). Thus \(G\) acts naturally on \(K\) by \(k\)-automorphisms. We define an action of \(G\) on \(K(M)=K(x_1,\ldots,x_n)\) the rational function field of \(n\) variables over \(K\), by \(\sigma\cdot x_j=\prod_{1\le i\le n} x_i^{a_{ij}}\) if \(\sigma\cdot e_j=\sum_{1\le i\le n} a_{ij} e_i \in M\), for any \(\sigma \in G\) (note that \(G\) acts on \(K\) as given before). The fixed field is denoted by \(K(M)^G\). The action of \(G\) on \(K(M)\) is called a purely quasi-monomial action of \(n\) variables in this paper. It is possible that \(G\) acts faithfully on \(K\) (the case of the function field of an algebraic torus) or trivially on \(K\) (the case \(K=k\); see \textit{D. J. Saltman}'s paper [J. Algebra 133, No. 2, 533--544 (1990; Zbl 0729.13006)] and \textit{M. Hajja} and \textit{M.-C. Kang}'s papers [J. Algebra 149, No. 1, 139--154 (1992; Zbl 0760.12004); ibid. 170, No. 3, 805--860 (1994; Zbl 0831.14003)]). The \(1\)-dimensional and the \(2\)-dimensional rationality problems of \(K(M)^G\) were solved by \textit{A. Hoshi} et al. [J. Algebra 403, 363--400 (2014; Zbl 1308.12005)]. It were shown that, \(K(M)^G\) were rational over \(k\) in some situations, and \(K(M)^G\) were rational over \(k\) if and only if some quaternion \(k\)-algebra splits in other situations. For details, see [loc. cit.] or Theorem 1.10 and Theorem 1.11 of this paper. Thus this paper focuses on the \(3\)-dimensional case. In the main results of this paper it is assumed that char \(k \neq 2\) (this assumption was not necessary in [loc. cit.]). Since the case of algebraic tori was solved by \textit{B. È. Kunyavskiĭ} [in: Issled. Teor. Chisel, Saratov 9, 90--111 (1987; Zbl 0681.14020)], it is also assumed that the action of \(G\) on \(K\) is not faithful in Theorem 1.15 and Theorem 1.16. Now we come to the main results of this paper, Theorem 1.15 and Theorem 1.16. There are \(73\) finite subgroups in \(\mathrm{GL}_3(\mathbb Z)\), up to conjugation. These groups give arise to all the \(G\)-lattices of rank \(3\) (for various groups \(G\)). In the book of \textit{H. Brown} et al. [Crystallographic groups of four-dimensional space. New York etc.: John Wiley \& Sons. (1978; Zbl 0381.20002)], these groups are classified into seven crystal systems and are designated by \(G_{i,j,k}\). The problem is solved in Theorem 1.15 if \(G\) is a group not belonging to the seventh crystal system. Theorem 1.16 takes care of the remaining cases, but the final result is incomplete. That is, if \(G\) belongs to the seventh crystal system and \(G\) is not equal to \(G_{7,j,3}\) (for \(j= 2,3,5\)), the problem is solved also. Hence there remain three groups for which the rationality problem are still unsettled.