Determination of conductors from Galois module structure (Q1849679)

This interesting and well-written paper determines the conductors of cyclic Kummer extensions \(K/F\) of odd prime power degree in the following setting: the base field is \(F=E(\zeta_{p^n})\) with \(E\) any unramified extension of \(\mathbb{Q}_p\), and \(K\) is Galois (i.e. normal) over \(E\). Note that \(K\) is then always a solvable extension of \(E\), but it certainly does not have to be Abelian. In fact, \(K_a=F(a^{1/p^n})\) is normal over \(E\) iff there is an integer \(r\) such that \(\text{Gal}(E/F)\) acts on \(\bar a \in F^*/(F^*)^{p^n}\) via the \(r\)th power of the cyclotomic character; let \(F^r\) be the set of all \(a\in F^*\) such that \(\bar a\) has this property. The point is now to evaluate sufficiently many Hilbert symbols \((a,y)_{n,F}\) for \(a\in F^r\), \(y\in F^*\) in order to pin down the conductor of \(K_a\) over \(F\). In this, it is important to have some control on \(F^r\). Actually it is possible to give a presentation for a certain Galois module \(D_r\) which is related to \(F_r\); this uses results of the reviewer [J. Reine Angew. Math. 479, 1-37 (1996; Zbl 0856.11051)]. If \(r\) is not congruent to 0 or 1 modulo \(p-1\), one Galois generator suffices for \(D_r\), which simplifies the argument. The result in this case is that if \(a\) is not a \(p\)th power, then the conductor of \(K_a\) is \(p^{n-1}t\) with \(t\) the smallest positive integer \(\equiv 2-r\) modulo \(p-1\). If \(r\) is 0 or 1 modulo \(p-1\), the argument and the results become substantially more complex; however, these cases are of particular interest since the case \(r=0\) means that \(\bar a\) comes from the base field \(E\), and the case \(r=1\) means exactly that \(K\) is Abelian over \(E\). (Note that \(r\) is defined modulo \(\phi(p^n)\), not only modulo \(p-1\).) It is nice that the author fully covers these cases as well.