Retract rational fields (Q420665)

Let \(k\) be an infinite field. The main subject of the paper is the retract \(k\)-rationality. This notion is a descendant of the rationality problem which asks whether the fixed field \(k(V)^G\) is rational over \(k\) for a given finite group \(G\). Recall that a finitely generated extension \(L\) of \(k\) is called \(k\)-rational (or rational over \(k\)) if \(L\) is purely transcendental over \(k\), i.e., \(L\) is isomorphic to some rational function field over \(k\). An important particular case of the rationality problem is Noether's problem which asks, under what situation, the filed \(k(G)=k(x_g:g\in G)^G\) is \(k\)-rational. Note that, if \(k\) is an infinite field and \(k(G)\) is \(k\)-rational, then there exists a generic \(G\)-Galois extension over \(k\). On the other hand, when Hilbert's irreducibility theorem is valid for \(k\) (e.g. if \(k\) is any algebraic number field), it is not difficult to see that the existence of a generic \(G\)-Galois extension over \(k\) implies that there is a Galois field extension \(K\) over \(k\) such that \(\mathrm{Gal}(K/k)\simeq G\), i.e. the inverse Galois problem for the pair \((k, G)\) is solvable. In the study of generic Galois extensions and generic division algebras, Saltman was led to the notion of retract \(k\)-rationality. Let \(k\) be an infinite field and \(L\) be a field containing \(k\). \(L\) is called retract \(k\)-rational, if there are some affine domain \(A\) over \(k\) and \(k\)-algebra morphisms \(\varphi: A\to k[X_1, \dots, X_n][1/f], \psi: k[X_1, \dots, X_n][1/f]\to A\) where \(k[X_1, \dots , X_n]\) is a polynomial ring over \(k, f \in k[X_1, \dots , X_n]\setminus \{0\}\), satisfying that (i) \(L\) is the quotient field of \(A\), and (ii) \(\psi\circ\varphi = 1_A\) , the identity map on \(A\). It well known that ``\(k\)-rational'' \(\Rightarrow\) ``retract \(k\)-rational''. Moreover, Saltman proved that the retract \(k\)-rationality of \(k(G)\) is equivalent to the existence of a generic \(G\)-Galois extension over \(k\). In the paper under review, the author proves some new properties of the retract rational fields. He shows for example that these fields posses the transitivity property. Some of the results are related to \(G\)-lattices \(M\) and the retract rationality of \(k(M)^G\). The main result is that, if \(G\) is a finite group containing an abelian normal subgroup \(H\) such that \(G/H\) is a cyclic group, then \(k(V)^G\) is retract \(k\)-rational for any linear representation of \(G\) on a \(k\)-vector space \(V\), provided that \(k\) contains a primitive \(e\)-th root of unity for \(e=\exp(G)\).