Involutions, odd degree extensions and generic splitting. (Q2349111): Difference between revisions

Let \(F\) be a field of characteristic \(\neq 2\). A famous result due to E.~Artin and, independently, T.~A.~Springer states that if a quadratic form over \(F\) is isotropic over an odd-degree extension of \(F\), then it is isotropic already over \(F\) itself. It is easy to show that the above statement remains true if one replaces `isotropic' by `hyperbolic'. As a further consequence, one obtains that if two quadratic forms over \(F\) are isometric (resp. similar) over an odd-degree extension of \(F\), then they are isometric (resp. similar) already over \(F\) itself. In this survey, the authors consider analogous problems for algebras with involutions motivated by the fact that for the endomorphism algebra \(\mathrm{End}_F(V)\) of an \(F\)-vector space \(V\), the questions on isotropy, hyperbolicity and isomorphism of orthogonal involutions on \(\mathrm{End}_F(V)\) can be translated into the question on isotropy, hyperbolicity and similarity of quadratic forms on \(V\) via the adjoint involution. More precisely, they ask the following three questions: Let \(A\) be a finite-dimensional central simple algebra over \(F\) with involutions \(\sigma,\sigma'\), and let \(L/F\) be an odd-degree field extension. (i) If \(\sigma\) is isotropic over \(L\), is it so over \(F\)? (ii) If \(\sigma\) is hyperbolic over \(L\), is it so over \(F\)? (iii) If \(\sigma\) and \(\sigma'\) are isomorphic over \(L\), are they so over \(F\)? Question (ii) has a positive answer as follows from work by \textit{E. Bayer-Fluckiger} and \textit{H. W. Lenstra} jun. [Am. J. Math. 112, No. 3, 359-373 (1990; Zbl 0729.12006)]. Question (iii) also has an affirmative answer as shown by \textit{D. W. Lewis} [Semigroup Forum 60, No. 1, 80-92 (2000; Zbl 0957.16014)] in the orthogonal and symplectic case, and by \textit{P. B. Barquero-Salavert} [Commun. Algebra 34, No. 2, 625-632 (2006; Zbl 1111.16034)] in the unitary case. Generally, the answer to (i) is negative as pointed out by \textit{R. Parimala} et al. [J. Algebra 243, No. 2, 780-789 (2001; Zbl 1003.11012)] who gave a counterexample in case of a unitary involution and they proposed a modified question (i'), in which `odd-degree' is replaced by `degree prime to \(2\mathrm{ind}(A)\). (i) and (i') are equivalent for orthogonal and symplectic involutions and for unitary involutions on algebras of 2-power exponent. At present, there are no counterexamples to (i'), and a positive answer is known only under additional assumptions on the algebra with involution, for instance concerning its decomposability. The survey presents all these results and shows methods to attack the problems, for example by invoking the theory of generic index reduction fields. The last section of the paper contains some new results. It is shown that (i') has a positive answer for orthogonal (resp. unitary) involutions on algebras of degree \(12\) (resp. degree \(6\) and exponent \(2\)), providing further evidence that (i') has a positive answer in general.