Toward equivariant Iwasawa theory. III (Q2505573)

This paper is a continuation of two previous works of the same authors [Manuscr. Math. 109, No. 2, 131--146 (2002; Zbl 1014.11066) and Indag. Math., New Ser. 15, No. 4, 549--572 (2004; Zbl 1142.11369)]. Let \(\ell\) be a fixed odd prime number and \(K_ \infty/k\) a Galois extension of totally real number fields with \([k:{\mathbb Q}]<\infty\), \(k_\infty\) the cyclotomic \(\ell\)-extension of \(k\) and \([K_ \infty:k_ \infty]<\infty\). Let \(G_\infty={\roman{Gal}}(K_ \infty/k)\), \(\Lambda(G_\infty)= {\mathbb Z}_ \ell[[G_\infty]]\). Fix a finite set \(S\) of primes of \(k\) containing \(\ell\), \(\infty\) and the ramified primes in \(K_ \infty\). In part II [op. cit] the authors formulated a ``main conjecture'' of equivariant Iwasawa theory in the non-abelian case. The abelian case was settled in part I [op. cit.]. There is an element \(L_ {k, S}\) in a \(\roman{Hom}\) group which incorporates all of the \(\ell\)-adic \(S\)-truncated Artin \(L\)-functions of \(K_ \infty/k\). The conjecture consists of two parts: (1) \(L_{k,S}\) has a unique preimage \(\tilde{\Theta}_ S\) in \(K_ 1 (Q(G_ \infty))\) via a determinant map and (2) the natural connecting homomorphism \(\partial\colon K_ 1(Q(\Lambda(G_\infty)))\to K_ 0 T(\Lambda(G_\infty))\) satisfies \(\partial(\tilde{\Theta})=\mho_ S\), where \(\mho_ S\in K_ 0 T( \Lambda(G_ \infty))\) contains all the information of the \(\Lambda(G_ \infty)\)-module \(X_ \infty\) in classical Iwasawa theory. The main result of the paper establishes that if the \(\mu\) Iwasawa invariant vanishes, then the main conjecture holds up to a uniqueness of \(\tilde{\Theta}_ S\) if, and only if, \(L_ {k,S}\in \det K_ 1(\Lambda(G_ \infty)_{\bullet})\) where \(\Lambda(G_\infty)_{\bullet}\) is the localization of \(\Lambda( G_ \infty)\) at \(\ell\), meaning that all the elements which are regular modulo \(\ell\) are inverted. The main result is consequence of Theorem B which says that an element of the \(\roman{Hom}\) group which belongs to \(\det(K_ 1(\Lambda(G_\infty)_{\bullet}))\) already belongs to \(\det(K_ 1(\Lambda(G_\infty)))\). In this paper the authors need \(G_ \infty\) to be an \(\ell\)-group for Theorem B. In a sequel paper [Homology Homotopy Appl. 7, No. 3, Conf., 155--171 (2005; Zbl 1154.11037)] this restriction is eliminated.