Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Finiteness of the ramification locus of an extension of Dedekind domains

Definition
TauCeti_RingTheory_DedekindDomain_RamificationLocus

by riccardo.brasca · Sep 29, 2026 · Mathlib 0df444a (Lean v4.33.1)

number-theorytauceti-chebotarev

A module-finite torsion-free extension of Dedekind domains with separable fraction-field extension ramifies at only finitely many primes. Equivalently, the complement of its unramified locus is finite. This supplies the finite exceptional sets in the density argument.

Formalization Note. These foundational declarations are transplanted from the Tau Ceti contributors' original source (Apache-2.0, commit 948fe4751b1fe528b6d580c522ca5d743d47f185), with compatibility adaptations for Lean 4.33.1. Mathematical proofs requiring separate theorem nodes are published separately.

Definition code
/- Transplanted from https://github.com/TauCetiProject/TauCeti at 948fe4751b1fe528b6d580c522ca5d743d47f185.
Original source copyright/license notices are retained below.
Generated exclusively from compiler declaration, command, and reference facts. -/
import Mathlib.FieldTheory.Separable
import Mathlib.RingTheory.DedekindDomain.Basic
import Mathlib.RingTheory.DedekindDomain.Different
import Mathlib.RingTheory.DedekindDomain.Factorization
import Mathlib.RingTheory.Unramified.Locus

section
set_option autoImplicit true
/-
Copyright (c) 2026 The Tau Ceti contributors. All rights reserved.
Released under Apache 2.0 license as described in the file LICENSE.
Authors: The Tau Ceti contributors
-/
/-!
# Finiteness of the ramification locus of an extension of Dedekind domains

Let `B` be a Dedekind domain, module-finite and torsion-free over a Dedekind domain `A`, with the
extension of fraction fields separable. Mathlib's `Algebra.unramifiedLocus A B` is the set of
primes of `B` at which `B` is unramified over `A`, and Mathlib knows it is open. Here it is shown
that its complement — the ramification locus — is *finite*.

The tool is the different ideal. Mathlib's `not_dvd_differentIdeal_iff` says a prime is unramified
exactly when it does not divide `differentIdeal A B`, and `differentIdeal_ne_bot` says that ideal
is nonzero; a nonzero ideal of a Dedekind domain has only finitely many divisors.

## Main results

* `Algebra.finite_compl_unramifiedLocus`: the ramification locus is finite.

Stated over Dedekind domains directly rather than in an AKLB tower: no ambient fraction fields
`K` and `L` and no `IsIntegralClosure B A L` are needed, since the separability hypothesis can be
carried by `FractionRing A` and `FractionRing B` themselves, which is the form
`not_dvd_differentIdeal_iff` already takes.

This supports Layer 1 of `TauCetiRoadmap/EllipticCurves/README.md`, whose
**separable-implies-unramified** milestone asks that a separable isogeny have `e_w = 1` at *every*
place, "by translation-invariance of the ramification locus". That argument needs the ramification
locus to be finite before translation-invariance can force it to be empty; the finiteness is what
this file supplies. Layer 2's `E[N]` and Layer 3's Hasse kernel count consume that milestone.

## Provenance

Ported from the AINTLIB `HasseWeil` project (`github.com/CBirkbeck/AINTLIB`, Apache-2.0, pinned by
that roadmap at `dev/hasse-weil @ 513e83879e2f`), `HasseWeil/Curves/RamificationFinite.lean`,
declarations `finite_setOf_dvd_differentIdeal`, `finite_setOf_ramificationIdx_ne_one` and
`finite_setOf_under_ramified`.

Changes from the source. The source works in an AKLB tower and rebuilds the standard instances
along the way, including `differentIdeal_ne_bot`, which Mathlib now proves; those are dropped, and
with them the tower. The ramification locus is named by Mathlib's `Algebra.unramifiedLocus` and
the criterion by Mathlib's `Algebra.IsUnramifiedAt`, in place of the source's hand-rolled
`ramifiedUnderLocus` and its `Ideal.ramificationIdx _ _ ≠ 1` spelling. The source's
divisor-finiteness step is Mathlib's `Ideal.finite_factors`, so it neither is rebuilt from the
`UniqueFactorizationMonoid` plumbing nor becomes public API here. Sixteen declarations become one.
-/

 section

open scoped nonZeroDivisors

attribute [local instance] FractionRing.liftAlgebra FractionRing.isScalarTower_liftAlgebra

namespace Algebra

variable (A B : Type*) [CommRing A] [IsDedekindDomain A] [CommRing B] [Algebra A B]
  [IsDedekindDomain B] [Module.IsTorsionFree A B] [Module.Finite A B]
  [Algebra.IsSeparable (FractionRing A) (FractionRing B)]

/-- **A module-finite, torsion-free extension of Dedekind domains whose fraction-field extension
is separable ramifies at only finitely many primes**: the complement of
`Algebra.unramifiedLocus` is finite. A ramified prime divides the different ideal, which is
nonzero and so, being an ideal of a Dedekind domain, has finitely many divisors. -/
theorem finite_compl_unramifiedLocus : (unramifiedLocus A B)ᶜ.Finite := by
  refine Set.Finite.of_finite_image (f := PrimeSpectrum.asIdeal) ?_
    fun _ _ _ _ h => PrimeSpectrum.ext h
  refine ((Ideal.finite_factors (differentIdeal_ne_bot (A := A) (B := B))).image
    _root_.IsDedekindDomain.HeightOneSpectrum.asIdeal).subset ?_
  rintro _ ⟨p, hp, rfl⟩
  have : p.asIdeal.IsPrime := p.isPrime
  have hdvd : p.asIdeal ∣ differentIdeal A B := dvd_differentIdeal_iff.mpr hp
  -- a ramified prime is nonzero: `⊥` divides only `⊥`, and the different ideal is not `⊥`
  have hbot : p.asIdeal ≠ ⊥ := fun h =>
    differentIdeal_ne_bot (A := A) (B := B) (eq_bot_iff.mpr (h ▸ Ideal.le_of_dvd hdvd))
  exact ⟨_root_.IsDedekindDomain.HeightOneSpectrum.ofPrime
    (Ideal.prime_of_isPrime hbot p.isPrime), hdvd, rfl⟩

end Algebra

end

end
Source
https://github.com/TauCetiProject/TauCeti/blob/948fe4751b1fe528b6d580c522ca5d743d47f185/TauCeti/RingTheory/DedekindDomain/RamificationLocus.lean

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me