On \(p\)-rationality of \(\mathbb{Q}(\zeta_{2l+1})^+\) for Sophie Germain primes \(l\) (Q2239162)

Let \(p\) be an odd prime and let \(F\) be a number field. The Galois group of the maximal abelian pro-\(p\) extension of \(F\) that is unramified outside \(p\) is isomorphic to a direct sum of finitely many copies of \(\mathbb Z_p\) and a torsion group. The field \(F\) is \(p\)-rational if Leopoldt's conjecture holds for \(F\) and this torsion group is trivial. The author uses Iwasawa's conjecture for totally real fields to deduce the following extension of a result of \textit{J. Assim} and \textit{Z. Bouazzaoui} [Funct. Approximatio. Comment. Math. 63, No. 2, 201--213 (2020; Zbl. 1473.11196)]: Assume \(F\) is a totally real abelian extension of \(\mathbb Q\) and that \(F\) does not contain the first layer \(\mathbb B_1\) of the \(\mathbb Z_p\)-extension of \(\mathbb Q\). Then \(F\) is \(p\)-rational if and only if the \(L\)-function value \(L(2-p,\chi)\) is prime to \(p\) for each non-trivial Dirichlet character \(\chi\) associated to \(F\). From this criterion, the following is deduced: Let \(\ell\) be a Sophie Germain prime, which means that \(q=2\ell + 1\) is prime. Suppose \(p< 2q-2\) is an odd prime that is a primitive root mod \(\ell\). Then \(\mathbb Q(\zeta_q)^+\) is \(p\)-rational. A corollary is that in this situation, \(p\) does not divide the class number of \(\mathbb Q\zeta_q)^+\). A conjecture of Gras states that a number field should be \(p\)-rational for all but finitely many \(p\). The present paper gives some data on the uniform distribution of Bernoulli numbers mod \(p\) and gives a heuristic argument supporting the truth of Gras's conjecture.