A consequence of Greenberg's generalized conjecture on Iwasawa invariants of \(\mathbb{Z}_p\)-extensions (Q344130)

Let \(p\) be a prime and let \(K\) be an algebraic extension of \(\mathbb Q\). Let \(X(K)\) be the Galois group over \(K\) of the maximal unramified abelian pro-\(p\)-extension of \(K\). If \(K\) is a \(\mathbb Z_p^d\)-extension of some number field \(k\), then \(X(K)\) is a finitely generated torsion \(\Lambda(K/k)\)-module, where \(\Lambda(K/k)\) denotes the Iwasawa algebra \(\mathbb Z_p[[\mathrm{Gal}(K/k)]]\). Now let \(\tilde k\) be the compositum of all \(\mathbb Z_p\)-extensions of \(k\). Then Greenberg's generalized conjecture asserts that \(X(\tilde k)\) is pseudo-null as a \(\Lambda(\tilde k/k)\)-module (see \textit{R. Greenberg} [Adv. Stud. Pure Math. 30, 335--385 (2001; Zbl 0998.11054)]). We will henceforth assume that this conjecture holds. If \(k\) is an imaginary quadratic field, \textit{M. Ozaki} [Adv. Stud. Pure Math. 30, 387--399 (2001; Zbl 1002.11078)] has shown that then the following holds for all but finitely many \(\mathbb Z_p\)-extensions \(K\) over \(k\): If one prime of \(k\) above \(p\) does not split in \(K/k\), then \(\mu(K/k) = 0\) and \(\lambda(K/k) = s\) (where \(s=1\) if \(p\) splits in \(k\), and \(s=0\) otherwise). Here, \(\mu(K/k)\) and \(\lambda(K/k)\) are the usual \(\lambda\)- and \(\mu\)-invariants of the Iwasawa module \(X(K)\). In the article under review, the author (partially) generalizes Ozaki's result to arbitrary number fields \(k\). Recall that the set \(\mathcal E(k)\) of all \(\mathbb Z_p\)-extensions of \(k\) is equipped with a compact Hausdorff topology. Two \(\mathbb Z_p\)-extensions \(K\) and \(K'\) are close if \([K \cap K' : k] = p^n\) with \(n\) large. For \(K \in \mathcal E(k)\) let \(s(K/k)\) be the \(\mathbb Z_p\)-rank of the \(\mathrm{Gal}(K/k)\)-coinvariants of \(X(K)\). Moreover, let \(s(k)\) be the minimum of the \(s(K/k)\), where \(K\) runs through all \(\mathbb Z_p\)-extensions of \(k\) such that every \(p\)-adic place of \(k\) ramifies in \(K/k\). For instance, one can show that \(s(k) = 0\) if \(p\) does not split in \(k\). If \(p\) splits completely in \(k\), one has \(s(k) = d(k) - 1\), where \(d(k)\) is the \(\mathbb Z_p\)-rank of \(\mathrm{Gal}(\tilde k/k)\). In particular, when \(p\) splits completely in an imaginary abelian field \(k\), then \(s(k) = [k:\mathbb Q] / 2\). In a similar way the author defines a second integer \(s'(k)\) which conjecturally always vanishes. Assuming \(s'(k) = 0\) and Greenberg's generalized conjecture, the author shows the following for generic \(\mathbb Z_p\)-extensions of \(k\) (where `generic' means that there is a closed subset \(E\) of \(\mathcal E(k)\) of measure \(0\) which contains the set of all \(K\) for which the following statement is not true): If every \(p\)-adic place of \(k\) does not split in \(K/k\), then \(\mu(K/k) = 0\) and \(\lambda(K/k) = s(k)\).