A two-dimensional rationality problem and intersections of two quadrics (Q2148100)

Let us first recall that if \(k\) is a field and \(L\) is a finitely generated extension field of \(k\), \(L\) is called \(k\)-rational (or rational over \(k\)) if \(L\) is purely transcendental over \(k\), i.e. \(L\) is \(k\)-isomorphic to the quotient field of some polynomial ring over \(k\). This problem is closely related to Noether's problem, which has been investigated for many groups in the last years. There also other variants, which are mentioned in the present paper: stably \(k\)-rational and \(k\)-unirational problems. The authors make the additional assumptions that \(a\in k\setminus k^2, L = k(\sqrt a)(x, y)\), where \(x, y\) are algebraically independent over \(k(\sqrt a)\). Then one can define a \(k\)-automorphism on \(L\) by \[\sigma:\sqrt a\to -\sqrt a, x\to x, y\to \frac{f(x)}{y},\] where \(f(x)\in k(x)\) is a non-zero rational function. Then the question arises: When the fixed subfield \(L^{\langle\sigma\rangle}\) will be \(k\)-rational? \textit{M. Hajja} et al. [J. Algebra 163, No. 2, 383--403 (1994; Zbl 0797.12005)] gave an answer to this question when \(f(x)\) is a polynomial of degree \(\leq 2\). The present paper finds a solution to this problem in terms of Hilbert symbols in the particular case when \(f(x)=c(x+b/x)+d\), where \(b, c, d \in k, b\ne 0\) and at least one of \(c, d\) is non-zero. As an appendix of the paper, the authors also give an alternative geometric proof of a part of the result which was provided to them by J.-L. Colliot-Thélène.