Noether's problem for cyclic groups of prime order (Q679720): Difference between revisions

Let \(K\) be a field, \(p\) a prime number, and \(K(x_0, x_1, \ldots , x_{p-1})\) the rational function field in \(p\) algebraically independent variables over \(K\). Suppose that \(G = \langle \sigma : \sigma^p = 1\rangle \) is a group of order \(p\), acting on \(K(x_0, \ldots , x_{p-1})\) by \(K\)-automorphisms as \(\sigma: x_0 \to x_1 \to \ldots \to x_{p-1} \to x_0\). Denote by \(K(x_0, \ldots , x_{p-1})^G\) the fixed field \(\{f \in K(x_0, \ldots , x_{p-1}): \sigma f = f\}\) of \(G\). Noether's problem for the group \(G\) asks whether \(K(x_0, \ldots , x_{p-1})^G\) is a rational (= purely transcendental) extension of \(K\). The main result of the paper under review gives a negative solution to this problem in the important special case where \(K\) is an algebraic number field (of finite degree over the field \(\mathbb Q\) of rational numbers). This result shows that then \(K(x_0, \ldots , x_{p-1})^G/K\) is not stably rational, for any prime \(p\) with finitely many exceptions. The exceptional set of prime numbers is described more precisely as follows: let \(P\) be the set of prime numbers, \(P_K\) the set of prime divisors of the absolute discriminant \(D_K\) (i.e. the set of all \(p \in P\) ramifying in the ring of algebraic integers in \(K\)), and \(P_0\) the set of those \(p \in P\), for which the field \(\mathbb Q(\xi_{p-1})\) is of class number \(1\), where \(\xi_{p-1}\) is a primitive complex root of unity of degree \(p - 1\). The set \(P_K\) is clearly finite (it is well-known that \(D_K \neq 0\)) and the fact that \(P_0\) is also finite has been established by \textit{J. M. Masley} and \textit{H. L. Montgomery} [J. Reine Angew. Math. 286/287, 248--256 (1976; Zbl 0335.12013)]. More precisely, they have proved that \(P_0 = \{p \in P: p \le 43\} \cup \{61, 67, 71\}\). The role of the set \(P_0\) in the solution of Noether's problem has been revealed by Plans; his solution for the case of \(K = \mathbb Q\) states that \(\mathbb Q(x_0, \ldots , x_{p-1}) ^ G/\mathbb Q\) is rational if and only if \(p \in P_0\). The main result of the paper under review states that \(K(x_0, \ldots , x_{p-1})^G/K\) is not stably rational, for any \(p \in P\) not lying in the union \(P_0 \cup P_K\). The author has generalized his main result to the case where \(K/\mathbb Q\) is a field extension, such that the relative algebraic closure \(K_0\) of \(\mathbb Q\) in \(K\) is a finite extension of \(\mathbb Q\) (this holds, for example, when \(K/\mathbb Q\) is a finitely-generated extension). It is not difficult to see that, in this case, the set \(P' = \{p \in \mathbb P: [K_0(\xi _ p): K_0] < p - 1\}\) is finite (here \(\xi_p\) denotes a primitive complex \(p\)-th root of unity). The paper proves that \(K(x_0, \ldots , x_{p-1})^G/K\) is not rational if and only if \(p \in P \setminus (P_0 \cup P')\). Furthermore, it shows that if \(K/\mathbb Q\) is finitely-generated, then \(K(x_0, \ldots , x_{p-1})^G/K\) is not stably rational, for any \(p \notin P _ 0 \cup P'\).