Involutions and stable subalgebras (Q1675085)

Let \(F\) be an arbitrary field, \(A\) a central simple \(F\)-algebra (of finite dimension) with involution \(\sigma \), and \(K\) a commutative \(F\)-algebra. We say that \(K\) is an étale algebra, if it is isomorphic to a finite direct product of finite separable field extensions of \(F\). An étale \(F\)-algebra is called split, if it is isomorphic to \(F ^ n\), for some \(n \in \mathbb N\). The paper under review investigates étale subalgebras of \(A\) contained in the subspace Sym\((\sigma ) = \{x \in A: \sigma (x) = x\}\) of symmetric elements of \((A, \sigma )\). Its research method emphasizes the similarities between the different types of involution in a way that allows to avoid unnecessary case distinctions, according to the characteristic and the type of involution. To follow more easily the desired unified approach, the authors introduce the notions of capacity and of neat subalgebra. Also, they use the set Sym\(^ *(\sigma )\) defined to be Sym\((\sigma )\) if \(\sigma \) is orthogonal or unitary, and to be equal to the set Symd\((\sigma ) = \{x + \sigma (x): x \in A\}\) in case \(\sigma \) is symplectic (note that Sym\(^ *(\sigma ) = \text{Symd}(\sigma )\) except when char\((F) = 2\) and \(\sigma \) is orthogonal). The capacity cap\((A, \sigma )\) is defined to be the degree deg\((A)\) if \(\sigma \) is orthogonal or unitary, and to be deg\((A)/2\) if \(\sigma \) is symplectic. By a neat subalgebra of \((A, \sigma )\), the authors mean an étale \(F\)-subalgebra \(L\) of \(A\), such that \(A\) is free as a left \(L\)-module, \(L \subseteq \text{Sym}(\sigma )\), and for each nonzero idempotent \(e\) of \(L\), the \(F\)-algebra with involution \((eAe, \sigma | _ {eAe})\) has the same type as \((A, \sigma )\). They observe that separable field extensions of \(F\) contained in Sym\((\sigma )\) are neat in \((A, \sigma )\). The reviewed paper shows that any étale \(F\)-subalgebra of \(A\) contained in Sym\((\sigma )\) is included in Sym\(^ *(\sigma )\); in addition, the maximal possible dimension \([L: K]\) of such an algebra \(L\) is equal to cap\((A, \sigma )\), and in this case, \(L\) is neat in \((A, \sigma )\). The paper proves that a commutative \(F\)-subalgebra \(K\) of \((A, \sigma )\) is neat if and only if \(K \subseteq L \subseteq \text{Sym}(\sigma )\), for some étale \(F\)-subalgebra \(L\) of \(A\) with \([L: F] = \text{cap}(A, \sigma )\) and such that \(L\) is free as a \(K\)-module; moreover, if \(K\) and \((A, \sigma )\) are split, then \(L\) can be chosen to be split as well. As an application, a conceptual proof of a theorem of Rowen is obtained, which asserts that every division algebra of exponent 2 and degree 8 contains a maximal subfield that is a triquadratic separable extension of the centre. In fact, the authors provide a refined version of this result which says that any symplectic involution on a central simple algebra of degree 8 stabilizes some triquadratic étale extension of the centre.