On a higher class number formula of \(\mathbb Z_p\)-extensions (Q2566541)

For a \({\mathbb Z}_{p}\)-extension \(K = \bigcup_{n}\;k_{n}\) of a number field \(k,\) let \(\widetilde L\) (resp. \(\widetilde L_{n})\) be the \(p\)-class field tower of \(K\) (resp. \(k_{n})\) and \(\widetilde G = \;\text{ Gal}\;(\widetilde L/K)\) (resp. \(\widetilde G_{n} =\;\text{ Gal}\;(\widetilde L_{n}/\widetilde k_{n})).\) For \(i \geq 1,\) consider the quotients \(X^{(i)} = C_{i}(\widetilde G)/C_{i+1}(\widetilde G)\) in the descending central series of \(\widetilde G,\) and similarly \(X^{(i)}_{n} = C_{i} (\widetilde G_{n})/C_{i+1} (\widetilde G_{n}).\) Suppose that the classical invariant \(\mu (K/k)\) vanishes and put \(\lambda^{(i)} (K/k) = \lambda^{(i)} =\;\text{ rank}_{{\mathbb Z}_{p}} X^{(i)}.\) \underbar{Question} : Does there exist an integer \(\nu^{(i)} (K/k) = \nu^{(i)}\) such that \(\sharp\;X_{n}^{(i)} = p^{\lambda^{(i)}n + \nu^{(i)}}\) for all \(n >> 0\;?\) \smallskip\noindent In this paper the author shows that this question admits an affirmative answer for \(i = 2\) if \(p\) does not split in \(K.\) Note that \textit{M. Ozaki} has recently shown (``Non-abelian Iwasawa theory of \({\mathbb Z}_{p}\)-extensions'', preprint) that the answer is always affirmative.