Sliceable groups and towers of fields (Q285589)

This is an interesting paper dealing with two closely related questions -- existence (or not) of intowers of Kronecker equivalent fields and the expressibility of profinite groups as unions of conjugates of finitely many closed subgroups of infinite index. The elementary statement that a finite group is not a union of the conjugates of a proper subgroup generalizes easily to profinite groups and closed proper subgroups. However, two immediate generalizations of this question are nontrivial:{\parindent = 0.6 cm \begin{itemize}\item[(i)] Does there exist a profinite group \(G\) admitting finitely many closed subgroups \(H_1, \dots, H_n\) of infinite index in it such that \[ G = \bigcup_{g \in G} \bigcup_{i=1}^n \{gH_ig^{-1} \}? \] \item[(ii)] Does there exist a profinite group \(G\) and a closed subgroup \(H\) of infinite index in it such that \(\bigcup_{g \in G} gHg^{-1}\) contains an open subgroup of \(G\)? \end{itemize}} The questions are related to towers of number fields which are Kronecker equivalent. Two number fields for which the same set of prime numbers (up to a finite number) have a degree \(1\) divisor are said to be Kronecker equivalent. It is well-known and easy to see that this is equivalent to the assertion that for a finite Galois extension \(N\) containing both fields \(L,L'\), \[ \bigcup_{g \in\mathrm{Gal}(N/\mathbb{Q})} \mathrm{Gal}(N/L)^g =\bigcup_{g \in \mathrm{Gal}(N/\mathbb{Q})} \mathrm{Gal}(N/L')^g. \] Thus, the existence of an infinite tower of Kronecker-equivalent number fields implies (and is, implied by, in case the profinite group \(G\) there is realizable as a Galois group over a number field) a positive answer to question (ii) mentioned above. The answer to question (ii) is unknown as yet. However, a generalization of these questions is given in this paper as follows. A profinite group \(G\) is finitely sliceable if there are finitely many closed subgroups \(H_1, \dots, H_n\) of infinite index answering question (i) mentioned above (in this case, \(G\) is said to be \(n\)-sliceable). As remarked, no profinite group is \(1\)-sliceable. The authors define analogously the concept of an \(n\)-sliceable infinite tower of Kronecker-equivalent number fields. The authors prove two contrasting results on sliceability of profinite groups. On the one hand, they prove that no open subgroup of \(\mathrm{GL}_2(\mathbb{Z}_l)\) (\(l\) odd prime) is finitely sliceable. As these groups occur as Galois groups over \(\mathbb{Q}\), this shows that the corresponding extension is not finitely sliceable. Incidentally, the authors also show that open subgroups of \(\mathrm{GL}_2(\mathbb{Z}_l)\) are infinitely sliceable, whose proof they analyse carefully to show the negative result on non-(finite sliceability). The other type of positive result shows that over any local field \(K\) of characteristic \(0\), for any central division algebra \(D\), any open subgroup of the group \(D^{\ast}/K^{\ast}\) is finitely sliceable. However, it is as yet unknown if any of these profinite groups occur as Galois groups over \(\mathbb{Q}\). The paper is lucidly written with simple and beautiful ideas.