Documentation

TauCeti.Topology.Algebra.Group.Profinite.ProP.TateDuality

Dévissage of Tate's duality maps for a pro-p group #

Let G be a pro-p group and N a discrete G-module which is a ℤ/nℤ-module satisfying Baer's criterion, such that H²(G, N) satisfies Baer's criterion over ℤ/nℤ as well. For a finite discrete G-module M write M' = InternalHom G M N for its N-dual, and αᵢ : Hⁱ(G, M) → Hom(H²⁻ⁱ(G, M'), H²(G, N)), i = 0, 1, 2, for Tate's duality maps (TauCeti.ContCohomology.dualityMap0, dualityMap1, dualityMap2). This file runs Tate's dévissage argument: if α₀ is surjective and α₁ injective on every discrete G-module of order p with trivial action killed by n, then the same holds on every finite p-primary discrete G-module killed by n (TauCeti.IsProP.dualityMap0_surjective_dualityMap1_injective), and likewise for α₁ surjective and α₂ injective (TauCeti.IsProP.dualityMap1_surjective_dualityMap2_injective); the two dévissages run side by side in TauCeti.IsProP.dualityMap0_surjective_dualityMap1_bijective_dualityMap2_injective, from a base case stating all three properties at once. The induction is TauCeti.IsProP.finite_pPrimary_induction along the trivial filtration, and the inductive step is the four lemmas along a short exact sequence 0 → A → B → C → 0 killed by n (TauCeti.ContCohomology.DiscreteShortExact.dualityMap0_surjective and its companions), whose N-dual sequence exists because N is Baer, and whose Hom(-, H²(G, N))-dual bottom row is exact because H²(G, N) is Baer. The four lemmas pair off: surjectivity of α₀ and injectivity of α₁ on the two ends of an extension give both on its middle, and so do surjectivity of α₁ and injectivity of α₂.

For an infinite profinite pro-p group, α₀ is moreover injective (TauCeti.IsProP.dualityMap0_injective): H⁰ is co-effaceable, the trace from a coinduced module along a deep enough open subgroup vanishing on invariants (TauCeti.IsProP.exists_isOpen_trace_eq_zero_of_mem_H0), and α₁ is injective on the kernel of the trace by the dévissage of α₀ surjective and α₁ injective, so only those two base cases are needed. Together with the dévissage this gives all three maps at once (TauCeti.IsProP.dualityMap0_bijective_dualityMap1_bijective_dualityMap2_injective): α₀ and α₁ bijective, α₂ injective.

The two instances of this argument are the coefficient system of a Demushkin group: N = 𝔽_p with n = p, where H²(G, 𝔽_p) is an 𝔽_p-vector space, and the twisted coefficients N = I(χ)/pⁱ of the canonical character with n = pⁱ, where H²(G, I(χ)/pⁱ) ≅ ℤ/pⁱ is self-injective. The remaining half of Tate's duality, the bijectivity of α₂, is a counting argument on top of this dévissage that uses the double duality of M with respect to N, and is not part of this file.

Main results #

References #

theorem TauCeti.IsProP.dualityMap0_surjective_dualityMap1_injective {p : ℕ} [Fact (Nat.Prime p)] {G : Type u} [Group G] [TopologicalSpace G] [IsTopologicalGroup G] (hG : IsProP p G) {n : ℕ} (N : Type v) [AddCommGroup N] [TopologicalSpace N] [DiscreteTopology N] [DistribMulAction G N] [ContinuousSMul G N] [Module (ZMod n) N] (hN : Module.Baer (ZMod n) N) (hH2 : Module.Baer (ZMod n) (ContCohomology.H2 G N)) (h : ∀ (A : Type u) [inst : AddCommGroup A] [inst_1 : TopologicalSpace A] [inst_2 : DiscreteTopology A] [inst_3 : DistribMulAction G A] [inst_4 : ContinuousSMul G A] [inst_5 : Finite A], Nat.card A = p → (∀ (g : G) (a : A), g • a = a) → (∀ (a : A), n • a = 0) → Function.Surjective ⇑(ContCohomology.dualityMap0 G A N) ∧ Function.Injective ⇑(ContCohomology.dualityMap1 G A N)) (M : Type u) [AddCommGroup M] [TopologicalSpace M] [DiscreteTopology M] [DistribMulAction G M] [ContinuousSMul G M] [Finite M] (hM : IsAddPGroup p M) (hMn : ∀ (m : M), n • m = 0) :

The dévissage of α₀ surjective and α₁ injective. Let G be a pro-p group and N a discrete G-module which is a Baer ℤ/nℤ-module with H²(G, N) a Baer ℤ/nℤ-module. If Tate's duality maps α₀ and α₁ into H²(G, N) are respectively surjective and injective on every discrete G-module of order p with trivial action killed by n, then they are so on every finite p-primary discrete G-module M killed by n: the pair of properties passes through extensions by the four lemmas DiscreteShortExact.dualityMap0_surjective and dualityMap1_injective.

theorem TauCeti.IsProP.dualityMap0_injective {p : ℕ} [Fact (Nat.Prime p)] {G : Type u} [Group G] [TopologicalSpace G] [IsTopologicalGroup G] (hG : IsProP p G) {n : ℕ} (N : Type v) [AddCommGroup N] [TopologicalSpace N] [DiscreteTopology N] [DistribMulAction G N] [ContinuousSMul G N] [Module (ZMod n) N] (hN : Module.Baer (ZMod n) N) (hH2 : Module.Baer (ZMod n) (ContCohomology.H2 G N)) (h : ∀ (A : Type u) [inst : AddCommGroup A] [inst_1 : TopologicalSpace A] [inst_2 : DiscreteTopology A] [inst_3 : DistribMulAction G A] [inst_4 : ContinuousSMul G A] [inst_5 : Finite A], Nat.card A = p → (∀ (g : G) (a : A), g • a = a) → (∀ (a : A), n • a = 0) → Function.Surjective ⇑(ContCohomology.dualityMap0 G A N) ∧ Function.Injective ⇑(ContCohomology.dualityMap1 G A N)) (M : Type u) [AddCommGroup M] [TopologicalSpace M] [DiscreteTopology M] [DistribMulAction G M] [ContinuousSMul G M] [Finite M] (hM : IsAddPGroup p M) (hMn : ∀ (m : M), n • m = 0) [CompactSpace G] [TotallyDisconnectedSpace G] [Infinite G] :

Injectivity of α₀ for an infinite pro-p group. Under the hypotheses of the dévissage of α₀ surjective and α₁ injective, if G is an infinite profinite pro-p group, then Tate's duality map α₀ : H⁰(G, M) → Hom(H²(G, M'), H²(G, N)) is injective on every finite p-primary discrete G-module M killed by n: the trace Coind_V^G M → M along a deep enough open subgroup V vanishes on invariants, and α₁ is injective on its kernel.

theorem TauCeti.IsProP.dualityMap1_surjective_dualityMap2_injective {p : ℕ} [Fact (Nat.Prime p)] {G : Type u} [Group G] [TopologicalSpace G] [IsTopologicalGroup G] (hG : IsProP p G) {n : ℕ} (N : Type v) [AddCommGroup N] [TopologicalSpace N] [DiscreteTopology N] [DistribMulAction G N] [ContinuousSMul G N] [Module (ZMod n) N] (hN : Module.Baer (ZMod n) N) (hH2 : Module.Baer (ZMod n) (ContCohomology.H2 G N)) (h : ∀ (A : Type u) [inst : AddCommGroup A] [inst_1 : TopologicalSpace A] [inst_2 : DiscreteTopology A] [inst_3 : DistribMulAction G A] [inst_4 : ContinuousSMul G A] [inst_5 : Finite A], Nat.card A = p → (∀ (g : G) (a : A), g • a = a) → (∀ (a : A), n • a = 0) → Function.Surjective ⇑(ContCohomology.dualityMap1 G A N) ∧ Function.Injective ⇑(ContCohomology.dualityMap2 G A N)) (M : Type u) [AddCommGroup M] [TopologicalSpace M] [DiscreteTopology M] [DistribMulAction G M] [ContinuousSMul G M] [Finite M] (hM : IsAddPGroup p M) (hMn : ∀ (m : M), n • m = 0) :

The dévissage of α₁ surjective and α₂ injective. Under the hypotheses of the dévissage, if Tate's duality maps α₁ and α₂ into H²(G, N) are respectively surjective and injective on every discrete G-module of order p with trivial action killed by n, then they are so on every finite p-primary discrete G-module M killed by n: the pair of properties passes through extensions by the four lemmas DiscreteShortExact.dualityMap1_surjective and dualityMap2_injective.

The two dévissages of Tate's duality maps, side by side. Under the hypotheses of the dévissage, if on every discrete G-module of order p with trivial action killed by n Tate's duality maps α₀, α₁, α₂ into H²(G, N) are respectively surjective, bijective and injective, then they are so on every finite p-primary discrete G-module M killed by n.

The dévissage of Tate's duality maps for an infinite pro-p group. Under the hypotheses of IsProP.dualityMap0_surjective_dualityMap1_bijective_dualityMap2_injective, if G is an infinite profinite pro-p group, then on every finite p-primary discrete G-module M killed by n, α₀ and α₁ are bijective and α₂ is injective: α₀ is moreover injective by IsProP.dualityMap0_injective.