Documentation

TauCeti.Topology.Algebra.Group.Profinite.Free.RelationModule.Basic

Lyndon's theorem for the relation module of a profinite presentation #

Let G be a finite group, g : ι → G a generating family indexed by a finite type, F the free profinite group on ι, and R the kernel of the continuous homomorphism π = lift g : F → G sending the generators to the g i. This file identifies the abelian pro-p group R^ab(p) = abelianizationProP p F R, with the conjugation action of F ⧸ R, with the relation module relationModule ℤ_[p] G g of the family g, the kernel of ℤ_p[G]^ι → ℤ_p[G], e_i ↦ g i - 1. Through this identification a ℤ_p[G]-module map out of the relation module acts on the kernel R^ab(p) of the group extension F ⧸ K → G, for K the kernel of R → R^ab(p); this is how the relation module enters the computation of the number of generators of the absolute Galois group of a p-adic field (NSW (7.4.1)).

The comparison map is the Fox derivative read in ℤ_p[G]: the unique continuous map D : F → ℤ_p[G]^ι with D (f * f') = D f + π f • D f' and D (of i) = e_i (TauCeti.freeProfiniteGroup.foxDerivative). It exists because a continuous homomorphism from F to the semidirect product of G by ℤ_p^{G × ι} may take any values on the generators. Its values satisfy the fundamental formula of Fox calculus ∑ i, D f i * (g i - 1) = π f - 1, so on R it is an equivariant homomorphism into the relation module, and it factors through R^ab(p).

The inverse is written down on the Schreier generators: for a set-theoretic section s of π the elements r h i = s h * of i * s (h * g i)⁻¹ lie in R, and m ↦ ∑ (m i)(h) • [r h i] is inverse to the Fox derivative on the relation module. That it is a left inverse is the uniqueness of crossed homomorphisms on F with given values on the generators, applied to the crossed homomorphism f ↦ (x ↦ [s x⁻¹ * f * s (x⁻¹ * π f)⁻¹]) with values in the coinduced module G → R^ab(p); that it is a right inverse is a computation with the relation ∑ m i * (g i - 1) = 0.

Main definitions #

Main results #

References #

noncomputable def TauCeti.freeProfiniteGroup.foxDerivative (p : ℕ) [Fact (Nat.Prime p)] {G : Type u} [Group G] [Finite G] [TopologicalSpace G] [DiscreteTopology G] {ι : Type u} (g : ι → G) [DecidableEq ι] :

The Fox derivative along lift g, read in ℤ_p[G]: the unique continuous map D : F → ℤ_p[G]^ι with D (f * f') = D f + lift g f • D f' and D (of i) = e_i (foxDerivative_unique). Its i-th coordinate is the image in ℤ_p[G] of the Fox derivative ∂f/∂x_i of f.

Instances For
    theorem TauCeti.freeProfiniteGroup.continuous_coeff_foxDerivative (p : ℕ) [Fact (Nat.Prime p)] {G : Type u} [Group G] [Finite G] [TopologicalSpace G] [DiscreteTopology G] {ι : Type u} (g : ι → G) [DecidableEq ι] :
    Continuous fun (f : ↑(freeProfiniteGroup ι).toProfinite.toTop) (i : ι) (x : G) => (foxDerivative p g f i).coeff x

    The coefficients of the Fox derivative depend continuously on the element of F.

    The Leibniz rule of the Fox derivative: D (f * f') = D f + lift g f • D f'.

    @[simp]
    theorem TauCeti.freeProfiniteGroup.foxDerivative_of (p : ℕ) [Fact (Nat.Prime p)] {G : Type u} [Group G] [Finite G] [TopologicalSpace G] [DiscreteTopology G] {ι : Type u} (g : ι → G) [DecidableEq ι] (i : ι) :

    The Fox derivative of the generator of i is the standard basis vector e_i.

    @[simp]
    theorem TauCeti.freeProfiniteGroup.foxDerivative_one (p : ℕ) [Fact (Nat.Prime p)] {G : Type u} [Group G] [Finite G] [TopologicalSpace G] [DiscreteTopology G] {ι : Type u} (g : ι → G) [DecidableEq ι] :

    The Fox derivative of 1 is 0.

    The Fox derivative of an inverse: D f⁻¹ = -(lift g f)⁻¹ • D f.

    The Fox derivative of a quotient: D (f * f'⁻¹) = D f - lift g (f * f'⁻¹) • D f'.

    On the kernel of lift g the Fox derivative is additive.

    theorem TauCeti.freeProfiniteGroup.foxDerivative_conj (p : ℕ) [Fact (Nat.Prime p)] {G : Type u} [Group G] [Finite G] [TopologicalSpace G] [DiscreteTopology G] {ι : Type u} (g : ι → G) [DecidableEq ι] {r : ↑(freeProfiniteGroup ι).toProfinite.toTop} (hr : r ∈ (↑(lift g)).ker) (f : ↑(freeProfiniteGroup ι).toProfinite.toTop) :

    On the kernel of lift g the Fox derivative is equivariant for conjugation: D (f * r * f⁻¹) = lift g f • D r.

    theorem TauCeti.freeProfiniteGroup.foxDerivative_unique (p : ℕ) [Fact (Nat.Prime p)] {G : Type u} [Group G] [Finite G] [TopologicalSpace G] [DiscreteTopology G] {ι : Type u} (g : ι → G) [DecidableEq ι] (D : ↑(freeProfiniteGroup ι).toProfinite.toTop → ι → MonoidAlgebra ℤ_[p] G) (hD : Continuous fun (f : ↑(freeProfiniteGroup ι).toProfinite.toTop) (i : ι) (x : G) => (D f i).coeff x) (hmul : ∀ (f f' : ↑(freeProfiniteGroup ι).toProfinite.toTop), D (f * f') = D f + MonoidAlgebra.single ((lift g) f) 1 • D f') (hof : ∀ (i : ι), D (of i) = Pi.single i 1) :

    The Fox derivative is unique: a continuous map D : F → ℤ_p[G]^ι with D (f * f') = D f + lift g f • D f' and D (of i) = e_i is foxDerivative p g.

    theorem TauCeti.freeProfiniteGroup.sum_foxDerivative_mul_sub_one (p : ℕ) [Fact (Nat.Prime p)] {G : Type u} [Group G] [Finite G] [TopologicalSpace G] [DiscreteTopology G] {ι : Type u} (g : ι → G) [DecidableEq ι] [Fintype ι] (f : ↑(freeProfiniteGroup ι).toProfinite.toTop) :
    ∑ i : ι, foxDerivative p g f i * (MonoidAlgebra.single (g i) 1 - 1) = MonoidAlgebra.single ((lift g) f) 1 - 1

    The fundamental formula of Fox calculus, read in ℤ_p[G]: ∑ i, D f i * (g i - 1) = lift g f - 1.

    The Fox derivative maps the kernel of lift g into the relation module of g.

    The Fox derivative on R^ab(p) #

    The Schreier inverse #

    Lyndon's isomorphism #

    Lyndon's theorem for the relation module (NSW (5.6.6)). For a finite group G generated by the family g : ι → G, the maximal abelian pro-p quotient R^ab(p) of the kernel R of lift g : F → G is isomorphic to the relation module relationModule ℤ_[p] G g. The isomorphism sends the class of r ∈ R to its Fox derivative (coe_abelianizationProPEquivRelationModule_ofMul), it is ℤ_p-linear (abelianizationProPEquivRelationModule_padicPow), and it is equivariant (abelianizationProPEquivRelationModule_smul).

    Instances For
      @[simp]

      Lyndon's isomorphism sends the class of r ∈ R to the Fox derivative of r.

      @[simp]

      Lyndon's isomorphism is ℤ_p-linear: it sends the p-adic power a ^ c in R^ab(p) to c times the image of a.

      Lyndon's isomorphism as a ℤ_p-linear equivalence, for the module structure on R^ab(p) given by p-adic exponentiation. Its underlying map is abelianizationProPEquivRelationModule.

      Instances For
        @[simp]

        The linear form of Lyndon's isomorphism evaluates as its additive form.

        Lyndon's isomorphism is equivariant: conjugation by f ∈ F on R^ab(p), through F ⧸ R, corresponds to multiplication by lift g f on the relation module.