Cyclic Quartic Fields and Genus Theory of Their Subfields
Z XIANKE
DOI: https://doi.org/10.1016/0022-314x(84)90067-2
IF: 0.679
1984-01-01
Journal of Number Theory
Abstract:Let k = Q(√u) (u ≠ 1 squarefree), K any possible cyclic quartic field containing k. A close relation is established between K and the genus group of k. In particular: (1) Each K can be written uniquely as K = Q(√vwη), where η is fixed in k and satisfies η ⪢ 1, (η) = U2√u, |U2| = |(√u)|, (v, u) = 1, v ∈ Z is squarefree, w|u, 0 < w < √u. Thus if u ≠ a2 + b2, there is no K ⊃ k. If u = a2 + b2 then for each fixed v there are 2g − 1K ⊃ k, where g is the number of prime divisors of u. (2) Kk has a relative integral basis (RIB) (i.e., OK is free over Ok) iff N(ε0) = −1 and w = 1, where ε0 is the fundamental unit of k, (or, equivalently, iff K = Q(√vε0√u), (v, u) = 1). (3) A RIB is constructed explicitly whenever it exists. (4) disc(K) is given. In particular, the following results are special cases of (2): (i) Narkiewicz showed in 1974 that Kk has a RIB if u is a prime; (ii) Edgar and Peterson (J. Number Theory 12 (1980), 77–83) showed that for u composite there is at least one K ⊃ k having no RIB. Besides, it follows from (4) that the classification and integral basis of K given by Albert (Ann. of Math. 31 (1930), 381–418) are wrong.