On torsion subgroups of Whitehead groups of division algebras. (Q2376341)

Let \(F\) be a field, \(D\) a finite-dimensional central division \(F\)-algebra, \(D^*\) the multiplicative group of \(D\), and \(T(D)\) the maximal torsion subgroup of the \(K\)-group \(K_1(D)\cong D^*/[D^*,D^*]\). The paper under review shows that if the degree \(\deg(D)\) is a prime number, then \(T(D)\) is a locally cyclic group, i.e., finitely-generated subgroups of \(T(D)\) are cyclic (or, equivalently, \(T(D)\) is isomorphic to a subgroup of the quotient group \(\mathbb Q/\mathbb Z\) of the additive group of rational numbers by the subgroup of integers). Conversely, if \(A\) is a torsion locally cyclic group, then there is a field \(\Phi\) possessing a central division algebra \(\Delta\), such that \(\deg(\Delta)\) is prime and \(T(\Delta)\) is isomorphic to \(A\). These results can be considered as a noncommutative version of May's theorem concerning the embeddability of any Abelian group \(A'\) with a locally cyclic maximal torsion subgroup \(T(A')\) into the multiplicative group of some field \(F(A')\). The proof relies essentially on May's theorem and on two theorems about the multiplicative structure of tame (central) division algebras over a Henselian field \((K,v)\). These include a version of Platonov's Congruence Theorem, and a description of \(T(\Theta)\) in the following two special cases: \(\Theta\) is unramified over \(K\); \(\Theta/K\) is totally ramified (in both cases, \(T(\Theta)\) is described under the extra hypothesis that \(\text{char}(K)=\text{char}(\overline K)\), where \(\overline K\) is the residue field of \((K,v)\)). The paper also contains information shedding light on the generality of the main result. For example, a Primary Decomposition Theorem for \(T(D)\), and arbitrary \(D\), reduces the study of \(T(D)\) to the special case where \(\deg(D)\) is a power of a prime number. The author points out that, generally, \(T(D)\) need not be locally cyclic. Finally, he observes that if \(\text{ind}(D)=n\), then \(T(D)\) is (\(n\)-torsion)-by-locally cyclic. Hence, he observes that \(T(D)\) cannot be isomorphic to the direct sum \(\mathbb Q/\mathbb Z\oplus\mathbb Q/\mathbb Z\), for any \(n\).