Biquaternion division algebras and fourth power-central elements (Q2065632)

This paper concerns the conditions under which a bi-quaternion algebra is a cyclic algebra (of degree 4) over its center. When \(F\) is a field of \(\operatorname{char}(F)=2\), it is known that every bi-quaternion algebra is a cyclic algebra [\textit{A. A. Albert}, Am. J. Math. 56, 75--86 (1934; Zbl 0008.24202)]. For a field \(F\) of \(\operatorname{char}(F)\neq 2\) containing a square root of \(-1\), it was proven [\textit{M. Rost} et al., C. R., Math., Acad. Sci. Paris 342, No. 2, 83--87 (2006; Zbl 1110.16014)] that \(A=(\alpha,\beta)_{2,F} \otimes (\gamma,\delta)_{2,F}\) is cyclic if and only if \(\langle \! \langle \alpha,\beta,\gamma,\delta \rangle \! \rangle\) is hyperbolic. Note that the latter is independent of the choice of decomposition of \(A\), and thus makes for a cohomological invariant of \(A\) in \(H^4(F,\mu_2)\). The author of the paper under discussion tackles the situation where \(F\) does not contain \(\sqrt{-1}\). In this case, he proves that \(A\) is cyclic if and only if there exists a decomposition \(A=(\alpha,\beta)_{2,F} \otimes (\gamma,\delta)_{2,F}\) such that \(\langle \! \langle -\alpha,\beta,\gamma,\delta \rangle \! \rangle\) is hyperbolic. The paper contains some other nice but more technical results concerning the behavior of bi-quaternion algebras, and a few open questions.