Iwasawa theory for Artin representations. I (Q2042502)

This paper is concerned with two important aspects of Artin representations \(\rho: \Delta \to \mathrm{GL}_d(\bar{\mathbb Q})\) attached to a Galois extension \(K/\mathbb Q\) with Galois group \(\Delta\): firstly, the possibility of attaching a \(p\)-adic \(L\)-function to \(\rho\), and secondly, finiteness results for appropriately defined Selmer groups associated with \(\rho\). It is known that one can construct a complex-valued Artin \(L\)-function \(L_\infty(s,\chi)\), where \(\chi\) is the character of \(\rho\). (Any representation over a field of characteristic zero is uniquely determined by its character.) The second question (finiteness of Selmer groups) is discussed first, in sections \S2--\S4. Various notations and hypotheses are required. Fix a prime \(p\) and an embedding \(\bar{\mathbb Q} \to \bar{\mathbb Q}_p\). This defines a place \(\mathfrak p\) above \(p\). The authors introduce three hypotheses, which are assembled into what they call Hypothesis A. In detail: (1) The degree \([K:\mathbb Q]\) is coprime to \(p\). (2) The plus eigenspace of complex conjugation under \(\rho\) is one-dimensional (authors' notation: \(d^+=1\)). (3) The restriction of \(\rho\) to the decomposition group \(\Delta_{\mathfrak p}\) contains a one-dimensional representation \(\varepsilon_{\mathfrak p}\), which moreover occurs only with multiplicity one. Using this setup, the authors manufacture a Selmer group denoted \( S_{\chi,\varepsilon}(\mathbb Q)\) as a subgroup of H\(^1(G_{\mathbb Q},D)\) defined by local conditions; the condition at \(p\) explicitly uses the character \(\varepsilon_{\mathfrak p}\). (The Galois module \(D\) is constructed as usual: take a number field \(\mathcal F\) over which \(\rho\) is realized, \(V\) the representation space of \(\rho\) over \(\mathcal F\), \(T\) an \(O_{\mathcal F}\)-lattice invariant under the action of \(G_{\bar{\mathbb Q}}\), and finally take \(D=V/T\).) One of the main theorems is the finiteness of this Selmer group. In fact a stronger result is proved (Prop.~3.1), replacing \(\mathbb Q\) by \(\mathbb Q_n\), the \(n\)-th stage in the \(p\)-cyclotomic tower. This then permits to prove (Prop.~4.2) that an Iwasawa-theoretic analogue denoted \(S_{\chi,\varepsilon}(\mathbb Q_\infty)\) is again ``small'', where now of course ``small'' cannot mean ``finite'' in general; the correct analog is ``cotorsion as a \(\Lambda\)-module''. The arguments are representation-theoretic and akin to ideas that have been used in order to prove Leopoldt's conjecture for certain number fields whose Galois group over the rationals is nonabelian but rather small. See for example the paper of [\textit{M. Emsalem} et al., J. Number Theory 19, 384--391 (1984; Zbl 0547.12003)]. We remind the reader that Leopoldt's conjecture is proved for abelian extensions of \(\mathbb Q\). The gist of these old arguments, and also of the approach in this paper, seems to be as follows. The main obstacle in the nonabelian case is the circumstance that certain representations (e.g., coming from global units of a field) will usually contain higher-degree representations of the Galois group with multiplicity greater than one, similarly as we observe in the regular representation. The point is to eliminate the problems caused by these multiplicities. In the paper under review, this is done using the clever definition of the local condition at \(p\). All relevant steps and ingredients in this context, and in particular the connection with Leopoldt's conjecture, are explained very clearly. Section 5 of the paper contains a long discussion of the second issue, the existence of \(p\)-adic \(L\)-functions attached to this kind of Artin representations. Here \(d=2\) is assumed. The contents of this section is even more sophisticated than what came before, and (to the reviewer's eye) sketchy in places. The main tools appear to be Hida theory, and a remarkable uniqueness result due to Greenberg-Vatsal and Bellaïche-Dimitrov, independently. The \(p\)-adic \(L\)-function to be constructed is called \(L_p(s,\chi,\varepsilon)\); it should come from a certain Iwasawa power series \(\theta^{an}_{\chi,\varepsilon}\). On the algebraic side, Theorem 3 states the existence of an algebraic analog \(\theta_{\chi,\varepsilon}\), and a ``main conjecture'' is mentioned, which postulates that \(\theta^{an}_{\chi,\varepsilon}\) and \(\theta_{\chi,\varepsilon}\) generate the same principal ideal. For more details on these difficult matters, the reader is referred to a sequel paper. As far as the reviewer can see, this second paper is not yet available. Reviewer's remarks: In line 6 of p.265, \(h-2\) is a typo and should read \(h_2\). The umlaut-like diacritic in the name Bellaïche is missing (references) or misplaced (Thm.~5.2). The ``characteristic series'' is written \(\theta_{\chi,\varepsilon}\) in Theorem 3 (p.261) but \(\theta^{al}_{\chi,\varepsilon}\) (presumably ``al'' for ``algebraic'') in the Conjecture two pages later. From the constructions in section 5, it is hard to extract a formal definition of the \(p\)-adic \(L\)-series that was announced in the introduction, but from context it appears that \(L_p(s,\chi,\varepsilon)\) is \(L_p(\mathfrak F)(\phi_1)\). The reviewer apologizes for the case that an explicit definition was overlooked. For the entire collection see [Zbl 1462.11006].