On arithmetically equivalent fields with distinct \(p\)-class numbers. (Q1427353)

Let \(K\), \(K'\) denote two number fields. If their Dedekind zeta-functions agree, then \(K\) and \(K'\) have the same degree, the same Galois closure, the same discriminant, and the same products of class number and regulator. In 1994, the author and \textit{R. Perlis} [Bull. Am. Math. Soc., New Ser. 31, No. 2, 213--215 (1994; Zbl 0814.11053)] discovered the first example of number fields with the same zeta function and different \(2\)-class numbers. These examples had degree \(8\); in [Algorithmic number theory, ANTS III, Portland, 392--399 (1998; Zbl 0914.11055)] the author constructed examples of degree \(7\). He also found examples with different \(p\)-class number and degrees \(2p+2\) for the primes \(p = 3\) and \(p = 5\). In this article it is shown that this is not a coincidence: if two number fields with the same zeta function and different \(p\)-class numbers exist, then their degree is \(\geq 2p+2\). Moreover, if examples with degree \(2p+2\) exist, then they are certain subfields of a Galois extension of \(\mathbb Q\) with Galois group GL\(_2(\mathbb F_p)/\mathbb F_p^{\times\,2}\). This result is easily reduced to a group theoretical statement, which is then proved using results from representation theory, the classification of Zassenhaus groups, and group cohomology.