Galois groups of tamely ramified \(p\)-extensions (Q2642754)

For a prime \(p\) and a number field \(K,\) let us denote \(G_{K,S}(p)\) the Galois group of the maximal pro-\(p\)-extension of \(K\) unramified outside a finite set of primes \(S.\) The cohomological properties of \(G_{K,S}(p)\) have been much studied in the case when \(S\) contains all \(p\)-primes (see e.g. the book by [\textit{H. Koch}, Galois theory of \(p\)-extensions. Springer Monographs in Mathematics. Berlin: Springer (2002; Zbl 1023.11002)]. On the contrary, not much is known in the case (let us call it ``case \((T)\)'') when \(S\) contains no \(p\)-prime. In this paper, the author describes some methods to compute \(G_{K,S}(p)\) in case \((T)\) when it is finite and conjectural properties of it when it is infinite. Let us for example review the developments around the Fontaine-Mazur conjecture. This famous conjecture asserts that under \((T),\) \(G_{K,S}(p)\) has no infinite analytic quotient. The author proposes a stronger refinement, called \((VGS)\) : in case \((T),\) any infinite quotient of \(G_{K,S}(p)\) has an open subgroup which is Golod-Shafarevich (a pro-\(p\)-group \(G\) is called Golod-Shafarevich if \(\dim\;H^2(G, {\mathbb F}_p) \leq {1 \over 4}(\dim\;H^1(G, {\mathbb F}_p))^2).\) Together with the so-called ``conjecture \(\varepsilon\)'', namely that non abelian free pro-\(p\)-groups are not linear, conjecture \(VGS\) implies the Fontaine-Mazur conjecture under \((T).\) More precisely, it implies that under \((T)\) any just-infinite quotient of \(G_{K,S}(p)\) is a branch group. Here, a pro-\(p\)-group is called just-infinite if its only infinite quotient is itself; a branch pro-\(p\)-group is a certain kind of subgroup of the automorphisms of a locally finite, rooted tree. This implies that there should exist actions with infinite image of \(G_{K,S}(p)\) on such trees. The author expresses the hope that there should be a theory of ``arboreal'' Galois representations which would help in investigating \(G_{K, S}(p)\) in case \((T)\) (contrary to the established theory of \(p\)-adic Galois representations, as predicted by Fontaine-Mazur).