Documentation

TauCeti.NumberTheory.Multiquadratic.CandidateGenusField.Relative.Ramification

The candidate genus field is unramified over ℚ(√d) at the finite places #

For a squarefree integer d, candidateGenusField hd is the compositum of the quadratic fields attached to the prime discriminants dividing fundamentalDiscriminant d, and candidateGenusFieldBase hd is the copy of K = ℚ(√d) inside it. This file proves that, for such a d that is moreover not a rational square, every prime of 𝓞 K is unramified in the candidate genus field: it is an unramified extension of ℚ(√d) at all finite places. This is the defining property of the genus field at the finite places, and the half of the genus-field identification the roadmap flags as the trap, since it is exactly where ramification at 2 and at the primes dividing d has to be controlled.

The work is done by the general criterion TauCeti.Multiquadratic.isUnramifiedIn_of_forall_ramificationIdx_eq_two: a subfield of a prime-discriminant compositum in which every prime belonging to a factor already has ramification index 2 carries all the ramification, so the compositum is unramified over it. Here that subfield is the quadratic base, and the hypothesis is the total ramification of ℚ(√d) at the primes dividing its discriminant, which are exactly the primes of the prime-discriminant factors.

An imaginary d is in particular not a rational square (not_isSquare_of_neg), and the base is then totally complex, hence unramified at the infinite places too (isUnramifiedAtInfinitePlaces_candidateGenusField), so for d < 0 the two results together give unramifiedness at every place: the candidate genus field of an imaginary quadratic field is an everywhere-unramified abelian extension of it, so it lies inside the Hilbert class field. What remains for the genus-field identification is maximality — that it is the largest such extension abelian over ℚ — and the isomorphism Gal(K_gen/K) ≅ Cl(K)/Cl(K)².

The description of the genus field as the prime-discriminant compositum is classical; see D. A. Cox, Primes of the Form x² + ny², and F. Lemmermeyer, Reciprocity Laws.

Main result #

The candidate genus field is unramified over ℚ(√d) at every finite place. For squarefree d that is not a rational square, every prime of the ring of integers of the embedded base ℚ(√d) is unramified in the ring of integers of candidateGenusField hd.

This is the finite-place half of the genus-field property of candidateGenusField hd.