On soluble-by-finite subgroups of division algebras. (Q2581397)

In this delightful paper the author analyses the structure of a soluble-by-finite subgroup \(G\) of the multiplicative group of a division algebra \(D\) of finite degree \(m\). For example, he proves that \(G\) contains an Abelian normal subgroup \(A\) of finite index dividing \(cmt\), where \(t<m\) is a divisor of \(m\) and \(c\) is \(1\), \(12\) or \(30\). (Thus in particular the index \((G:A)\) divides \(60m^2\).) Also frequently \(c\) can be taken to be \(1\); for example it can if \(\text{char\,}D>0\), or if \(m\) is odd, or if \(G\) is torsion-free. An important step in the author's analysis is the careful determination of the normalizer \(H\) of \(G\) in \(D^*\) where \(G\) is nilpotent of class \(\leq 2\) and \(G\) and \(C_D(G)\) generate \(D\). The answer is precise and depends upon the structure of the maximal periodic normal subgroup of \(G\). In fact \(H=G.C_{D^*}(G)\) in most cases and in all the exceptions the index \((H:G.C_{D^*}(G))\) is \(2\) or \(6\). Finally, the author considers whether the subalgebra generated by \(G\) (again \(G\) is soluble-by-finite) is a crossed product. A word of warning for those who have seen a preprint of this paper -- as far as I can see the content of my copy of the preprint is identical to that of the printed version but the numbering of every single result has changed.