Galois subfields of tame division algebras. (Q273106)

Let \(F\) be a field and \(d(F)\) the class of associative finite-dimensional central division \(F\)-algebras. It is a well-known consequence of the classical Skolem-Noether theorem that an algebra \(D\in d(F)\) is a crossed product if and only if \(D\) possesses a maximal subfield which is a Galois extension of \(F\). When \(D\) does not possess such a maximal subfield, it is called a noncrossed product. By class field theory, \(D\) is necessarily a crossed product (more precisely, a cyclic \(F\)-algebra), if \(F\) is a global or local field. The question of whether there exist fields admitting noncrossed products had been open for about 40 years, when it was answered affirmatively by \textit{S. A. Amitsur} [Isr. J. Math. 12, 408-420 (1972; Zbl 0248.16006)]. This was specified by \textit{L. J. Risman} [Isr. J. Math. 28, 113-128 (1977; Zbl 0366.16006)], and in any prescribed characteristic by \textit{M. M. Schacher}, \textit{L. W. Small} [J. Algebra 24, 100-103 (1973; Zbl 0256.16013)] and \textit{D. J. Saltman} [J. Algebra 52, 302-314 (1978; Zbl 0391.13002)]. Subsequently, the existence of noncrossed products over more familiar fields was found by a number of authors, including Brussel, McKinnie, Tengan, Tignol, Jacob, Wadsworth, Hanke (references to their work can be found in the reviewed paper). The paper under review solves the crossed product problem in the case where \(F\) is endowed with a Henselian valuation \(v\), and \(D/F\) is tamely ramified (i.e. there is a tamely ramified finite extension of \(F\), which is a splitting field of \(D\)). Note that the Henselity of \(v\) ensures its extendability to a unique, up-to an equivalence, valuation \(v_\Delta\), on each finite-dimensional associative division \(F\)-algebra \(\Delta\). The main result of the reviewed paper solves the considered problem by showing that \(D\) is a crossed product if and only if the residue division \(\overline F\)-algebra \(\overline D\) of \((D,v_D)\), where \(\overline F\) is the residue field of \((F,v)\), contains as a maximal subfield a Galois extension of \(\overline F\). The proof of the ``only if'' implication relies on the theory of graded division algebras (presented by \textit{J.-P. Tignol} and \textit{A. R. Wadsworth} [Value functions on simple algebras, and associated graded rings. Cham: Springer (2015; Zbl 1357.16002)]). The authors deduce from the main result that \(d(F)\) consists of crossed products, provided that \(\overline F\) is a local field. The same follows in case the absolute Galois group of \(\overline F\) is a projective profinite group. When \(\overline F\) is a global field, the paper gives a description of the location of noncrossed products among tame division algebras, and leads to their discovery in new parts of the Brauer group.