On Kervaire and Murthy's conjecture. (Q1415010)

The conjecture is concerned with the natural exact sequence [see \textit{M. A. Kervaire} and \textit{M. P. Murthy}, Comment. Math. Helv. 52, 415--452 (1977; Zbl 0355.12009)], \(V_n\rightarrowtail \text{Pic} (\mathbb{Z} C_{n+1}) \twoheadrightarrow \text{Cl} (\mathbb{Q} (\zeta_{n+1})) \oplus \text{Pic} (\mathbb{Z} C_n)\) relating the Picard groups of the integral group rings \(\mathbb{Z} C_k\), \(C_k=\) cyclic of order \(p^k\), to the ideal class group of the cyclotomic field \(\mathbb{Q} (\zeta_{n+1})\) generated by a root of unity \(\zeta_{n+1}\) of order \(p^{n+1}\). Here, \(p\) is a semi-regular prime (meaning that \(p\nmid h_{\mathbb{Q} (\zeta_1+ \zeta_1^{-1})})\). With respect to the above appearing kernel \(V_n\) one has a splitting \(V_n= V_n^+\times V_n^-\), an explicit description of \(V_n^-\), and a canonical injection \(\widehat{V_n^+} \rightarrowtail (\text{Cl} (\mathbb{Q} (\zeta_n)))_p\) of the group of characters of \(V_n^+\) into the Sylow \(p\)-subgroup of \(\text{Cl} (\mathbb{Q} (\zeta_n))\). The conjecture says that this latter map is surjective. This is verified in the paper under review, provided that \(p\) is odd, semi-regular and, in addition, 2-regular (which means that \(B_{1,\omega^{i-1}} \not\equiv 0\bmod p^2\) and \(\frac{B_i}{i} \not\equiv \frac{B_{i+p-1}} {i+p-1}\bmod p^2\) where \(\omega\) is the Teichmüller character and \(i\) runs through the even indices between 2 and \(p-3\) with \(p\) dividing the numerator of \(B_i\)). The proof is based on certain norm maps \(\mathbb{Z} [\zeta_{k+i+1}]\to \mathbb{Z} [x]/ \frac{x^{p^{k-i}-1}} {x^{p^k}-1}\), on Mayer-Vietoris exact sequences for \(\text{Pic} (\mathbb{Z} C_n)\) when \(p\) is 2-regular, and, moreover, on ramification theory (but not on Iwasawa theory, as in Kervaire's and Murthy's paper).