Kronecker-Weber, tame step: removal of the ramification at a prime from an abelian field of -power degree
OpenNumberField.exists_isUnramifiedIn_le_sup_of_prime_neLet be primes, let be an algebraically closed field of characteristic zero (an algebra over ), and let be a finite abelian extension of of degree . The statement asserts that there are subfields such that
- is a finite abelian extension of of degree a power of ,
- is unramified in ,
- each prime that is unramified in is unramified in ,
- embeds into the cyclotomic field , and
In words: the ramification of at the prime comes from a subfield of , and no new ramification appears.
Proof idea. Let be the exact power of that divides , let be the subfield of of degree , and let . The field is abelian of -power degree, so its inertia group at is tame, and divides , hence . The field is totally ramified at , so maps onto and . Let be the fixed field of . Then is unramified in , , and a degree count gives . A prime that is unramified in is unramified in , hence in , hence in . If is already unramified in (for example , or ), take and .
Use. Induction on the number of ramified primes different from reduces the Kronecker-Weber theorem for abelian fields of -power degree to the case of fields unramified outside . This is a child of Leopoldt.exists_algHom_cyclotomicField_of_isCyclic_primePow; the compositum is handled by NumberField.exists_algHom_cyclotomicField_of_sup.
Formalization Note. Fields are IntermediateField ℚ Ω; "abelian" is the Mathlib class IsAbelianGalois ℚ E; " unramified in " is Algebra.IsUnramifiedIn (𝓞 E) (Ideal.span {(ℓ : ℤ)}); the existential carries FiniteDimensional ℚ E' and IsAbelianGalois ℚ E' as propositions. Mathlib (at this revision) has inertia groups (Ideal.inertia, Ideal.card_inertia_eq_ramificationIdxIn), the inertia field (IsInertiaField), ramification in cyclotomic fields (IsCyclotomicExtension.Rat.ramificationIdxIn_eq_of_prime_pow, ramificationIdxIn_eq_of_not_dvd) and IsCyclotomicExtension.Rat.galEquivZMod. It has no higher ramification groups and no tame character , ; the divisibility for an abelian extension of with is the main missing piece. A second missing piece is that a prime unramified in two fields is unramified in their compositum.
import Mathlib open NumberField
theorem NumberField.exists_isUnramifiedIn_le_sup_of_prime_ne {Ω : Type*} [Field Ω] [Algebra ℚ Ω]
[IsAlgClosed Ω] (p q : ℕ) (hp : p.Prime) (hq : q.Prime) (hqp : q ≠ p)
(E : IntermediateField ℚ Ω) [FiniteDimensional ℚ E] [IsAbelianGalois ℚ E]
(k : ℕ) (hE : Module.finrank ℚ E = p ^ k) :
∃ E' F : IntermediateField ℚ Ω, FiniteDimensional ℚ E' ∧ IsAbelianGalois ℚ E' ∧
(∃ m : ℕ, Module.finrank ℚ E' = p ^ m) ∧
Algebra.IsUnramifiedIn (𝓞 E') (Ideal.span {(q : ℤ)}) ∧
(∀ ℓ : ℕ, ℓ.Prime → Algebra.IsUnramifiedIn (𝓞 E) (Ideal.span {(ℓ : ℤ)}) →
Algebra.IsUnramifiedIn (𝓞 E') (Ideal.span {(ℓ : ℤ)})) ∧
Nonempty (F →ₐ[ℚ] CyclotomicField q ℚ) ∧ E ≤ E' ⊔ F := by sorry