Initial layers of \(\mathbb{Z}_p\)-extensions and Greenberg's conjecture (Q1294052)

The subject of this paper is a close-up on a criterion for Greenberg's conjecture, applicable to real quadratic fields \(k\) in which the prime \(p\not=2\) does not split and \(p\) divides the class number of \(k\) just once. The criterion says: \(\lambda_p(k)=0\) (this is the so-called Greenberg conjecture) iff there exists some \(n\geq 0\) and \(\alpha_n\in K_n=k(\zeta_{p^{n+1}})\) satisfying a specific condition linking it to the class group of \(k\), such that \(\root p \of \alpha_n\) starts a \({\mathbb{Z}}_p\)-extension of \(K_n\). Call this whole condition \((J_n)\). The main theorem gives an exact value of \(n\) for which \((J_n)\) begins to hold, provided it begins to hold some time. But the value for \(n\) is given in terms of the size of \(A_\infty(k)\), and the very finiteness of \(A_\infty(k)\) is tantamount to \(\lambda_p(k)=0\), so this extra knowledge concerning \(n\) is probably of no help in proving the Greenberg conjecture. Also, the main theorem imposes a very strict condition, namely that a certain Iwasawa polynomial is of degree one. (The author's previous paper in J. Math. Soc. Japan 49, 689-711 (1997; Zbl 0907.11038) only assumed it to be irreducible.) One principal point in the proof is furnished by a result of \textit{R. Greenberg} [Am. J. Math. 98, 263-284 (1976; Zbl 0334.12013)] which explicitly describes, in the given situation, pieces of the composite of all \({\mathbb{Z}}_p\)-extensions of all \(K_n\) as Kummerian extensions obtained by extraction of \(p\)-power roots of units. Capitulation from level zero to level \(n\) in the cyclotomic tower also plays a role. Some numerical data conclude the article under review.