Kummer's criterion for totally real number fields (Q1192428)

For a totally real number field \(k\), consider the extension \(K/k\) generated by the \(p\)th roots of 1, where \(p\) is an odd prime. Denote by \(\zeta(s,k)\) the Dedekind zeta function of \(k\) and by \(S\) the set of primes of \(K^ +\) (the maximal real subfield of \(K\)) above \(p\); assume that the primes in \(S\) do not split in \(K\). The author studies the equivalence of the following three conditions: (i) the \(p\)-primary part of the ideal class group of \(K\) has a non-trivial minus-part; (ii) \(p\) divides at least one of the numerators of the rational numbers \(\zeta(1- i,k)\) (\(i=2,4,\dots,d-2\)) and \(p^ \nu \zeta(1-d,k)\), where \(d=[K:k]\) and \(p^ \nu\) is the number of \(p\)-power roots of 1 contained in \(K\); (iii) \(K^ +\) possesses an extension of degree \(p\), unramified outside \(S\) but other than the first layer of the cyclotomic \(\mathbb{Z}_ p\)- extension of \(K^ +\). For \(k=\mathbb{Q}\), the equivalence of (i) and (ii) follows from the classical Kummer criterion. In the general case this equivalence was essentially proved by \textit{R. Greenberg} [Invent. Math. 21, 247--254 (1973; Zbl 0269.12005)]. The author makes use of \(p\)-adic \(L\)-functions to present a new proof for it, or in fact for a slightly more general result in which \(K\) is replaced by an intermediate field of \(K/k\). He also gives a proof, in the same spirit, for the equivalence of (ii) and (iii). As he points out, another proof for this equivalence follows from certain previous results which imply the equivalence of (i) and (iii).