Bender–Glauberman §2F: fixed submodules and quotients (main layer)
DefinitionOddOrder_BG_Ch1_Preliminary_S02_FixedQuotientMachineryfeit-thompsonfinite-groupsgroup-theoryishida-odd-order
The main layer of the machinery behind the proof of Bender–Glauberman Theorem 2.6, built on fixedOnSubmoduleAndQuotientSubgroup: scalar characters, determinant-kernel line analysis, the stay-or-swap dichotomy, and the commutator bounds, up to but excluding the final assembly layer.
Main definitions: representationToGeneralLinearGroup (a representation as a homomorphism into ), determinantCharacterOfRepresentation (the determinant character) and determinantKernelSubgroup (the determinant-kernel subgroup of BG Thm 2.6).
The bundle also proves 40 supporting lemmas used by later statements.
Definition code
/-
Copyright (c) 2026 Yawara Ishida. All rights reserved.
Released under Apache 2.0 license as described in the file LICENSE.
Authors: Yawara Ishida
-/
-- From github.com/yawara/odd-order @ 20a6fccbb63c (Apache-2.0; LICENSE at that commit):
-- OddOrder/BG/Ch1_Preliminary/S02_FixedQuotientMachinery.lean
-- Generated for Prove2Me by deleting declarations; the remaining commands are verbatim except:
-- `private` dropped where another file uses the declaration, and a solution's main theorem
-- renamed `_root_.solution` (platform convention).
import Definitions.Def_OddOrder_BG_Ch1_Preliminary_S02_RepresentationsBasic
namespace OddOrder.BG.Ch1.S02
open scoped Pointwise
open OddOrder.RepresentationTheory (baseChangeRepresentation baseChangeRepresentation_apply_tmul
baseChangeRepresentation_faithful)
private theorem mem_fixedOnSubmoduleAndQuotientSubgroup
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
{W : Submodule F V} {ρ : Representation F G V} {g : G} :
g ∈ fixedOnSubmoduleAndQuotientSubgroup W ρ ↔
(∀ w ∈ W, ρ g w = w) ∧ (∀ v, ρ g v - v ∈ W) :=
Iff.rfl
/-- If `g` acts trivially on `W` and on `V/W`, then `ρ g - 1` squares to zero.
This is the unipotent calculation behind BG Thm 2.6(b), q = p. -/
private theorem fixedOnSubmoduleAndQuotientSubgroup_sub_pow_two_eq_zero
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(W : Submodule F V) (ρ : Representation F G V)
{g : G} (hg : g ∈ fixedOnSubmoduleAndQuotientSubgroup W ρ) :
(((ρ g : Module.End F V) - 1) ^ 2 : Module.End F V) = 0 := by
rw [pow_two]
ext v
rw [Module.End.mul_apply]
change ρ g (ρ g v - v) - (ρ g v - v) = 0
have hmem := (mem_fixedOnSubmoduleAndQuotientSubgroup.mp hg).2 v
have hfix := (mem_fixedOnSubmoduleAndQuotientSubgroup.mp hg).1 (ρ g v - v) hmem
rw [hfix, sub_self]
/-- In characteristic `p`, every element of `C_G(W) ∩ C_G(V/W)` acts with
p-th power identity under the representation. -/
private theorem fixedOnSubmoduleAndQuotientSubgroup_rep_pow_prime_eq_one
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] {V : Type*} [AddCommGroup V] [Module F V]
[Nontrivial V]
(W : Submodule F V) (ρ : Representation F G V)
{g : G} (hg : g ∈ fixedOnSubmoduleAndQuotientSubgroup W ρ) :
(ρ g : Module.End F V) ^ p = 1 := by
have : CharP (Module.End F V) p := IsPGroup.charP_End_of_field
have hsq :
(((ρ g : Module.End F V) - 1) ^ 2 : Module.End F V) = 0 :=
fixedOnSubmoduleAndQuotientSubgroup_sub_pow_two_eq_zero W ρ hg
have hpow_zero :
(((ρ g : Module.End F V) - 1) ^ p : Module.End F V) = 0 :=
pow_eq_zero_of_le (Nat.Prime.two_le (Fact.out : p.Prime)) hsq
have hsub :
(((ρ g : Module.End F V) - 1) ^ p : Module.End F V) =
(ρ g : Module.End F V) ^ p - 1 := by
rw [sub_pow_char_of_commute p (Commute.one_right (ρ g : Module.End F V)), one_pow]
exact sub_eq_zero.mp (hsub ▸ hpow_zero)
/-- Faithfulness turns the previous representation-level p-torsion into
group-level p-torsion. -/
private theorem fixedOnSubmoduleAndQuotientSubgroup_pow_prime_eq_one
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] {V : Type*} [AddCommGroup V] [Module F V]
[Nontrivial V]
(W : Submodule F V) (ρ : Representation F G V)
(hfaithful : Function.Injective ρ)
{g : G} (hg : g ∈ fixedOnSubmoduleAndQuotientSubgroup W ρ) :
g ^ p = 1 := by
apply hfaithful
simpa [map_pow] using
(fixedOnSubmoduleAndQuotientSubgroup_rep_pow_prime_eq_one
(p := p) W ρ hg)
/-- In a faithful representation over characteristic `p`, the subgroup acting
trivially on a submodule and on the quotient is a p-subgroup. -/
private theorem fixedOnSubmoduleAndQuotientSubgroup_isPGroup_of_faithful
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] {V : Type*} [AddCommGroup V] [Module F V]
[Nontrivial V]
(W : Submodule F V) (ρ : Representation F G V)
(hfaithful : Function.Injective ρ) :
IsPGroup p (fixedOnSubmoduleAndQuotientSubgroup W ρ) := by
intro g
refine ⟨1, Subtype.ext ?_⟩
simpa using
(fixedOnSubmoduleAndQuotientSubgroup_pow_prime_eq_one
(p := p) W ρ hfaithful g.property)
/-- If a p-subgroup acts on a rank-one invariant submodule and rank-one quotient,
then it lies in `C_G(W) ∩ C_G(V/W)`. -/
private theorem subgroup_le_fixedOnSubmoduleAndQuotientSubgroup_of_rank_one_subquotients
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] {V : Type*} [AddCommGroup V] [Module F V]
(H : Subgroup G) (W : Submodule F V) [Module.Free F W] [Module.Free F (V ⧸ W)]
(ρ : Representation F G V) (hH : IsPGroup p H)
(hW : ∀ h : H, W ≤ W.comap (ρ h))
(hdimW : Module.finrank F W = 1) (hdimQ : Module.finrank F (V ⧸ W) = 1) :
H ≤ fixedOnSubmoduleAndQuotientSubgroup W ρ := by
intro g hg
rw [mem_fixedOnSubmoduleAndQuotientSubgroup]
let h : H := ⟨g, hg⟩
constructor
· exact isPGroup_rank_one_submodule_action_trivial_of_charP hH W hdimW (ρ.comp H.subtype)
hW h
· exact isPGroup_rank_one_quotient_action_trivial_of_charP hH W hdimQ
(ρ.comp H.subtype) hW h
/-- Commutativity consequence of
`subgroup_le_fixedOnSubmoduleAndQuotientSubgroup_of_rank_one_subquotients`. -/
private theorem subgroup_commutative_of_rank_one_subquotients
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] {V : Type*} [AddCommGroup V] [Module F V]
(H : Subgroup G) (W : Submodule F V) [Module.Free F W] [Module.Free F (V ⧸ W)]
(ρ : Representation F G V) (hfaithful : Function.Injective ρ) (hH : IsPGroup p H)
(hW : ∀ h : H, W ≤ W.comap (ρ h))
(hdimW : Module.finrank F W = 1) (hdimQ : Module.finrank F (V ⧸ W) = 1) :
Std.Commutative (· * · : H → H → H) := by
apply subgroup_commutative_of_faithful_representation_fixed_on_submodule_and_quotient
H W ρ hfaithful
intro h
exact mem_fixedOnSubmoduleAndQuotientSubgroup.mp
(subgroup_le_fixedOnSubmoduleAndQuotientSubgroup_of_rank_one_subquotients
H W ρ hH hW hdimW hdimQ h.property)
/-- A scalar character into a field's unit group kills the commutator subgroup. -/
private theorem commutator_le_ker_of_units_character
{F : Type*} [Field F] {G : Type*} [Group G] (φ : G →* Fˣ) :
commutator G ≤ φ.ker := by
rw [_root_.commutator_def, Subgroup.commutator_le]
intro x _hx y _hy
rw [MonoidHom.mem_ker, map_commutatorElement]
exact commutatorElement_eq_one_iff_mul_comm.mpr (mul_comm (φ x) (φ y))
/-- A representation as a group homomorphism into `GL(V)`.
This is the determinant-facing form of a `Representation`; the inverse of
`ρ g` is supplied by `ρ g⁻¹`. -/
def representationToGeneralLinearGroup
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(ρ : Representation F G V) : G →* LinearMap.GeneralLinearGroup F V where
toFun g :=
{ val := ρ g
inv := ρ g⁻¹
val_inv := by
rw [← map_mul, mul_inv_cancel, map_one]
inv_val := by
rw [← map_mul, inv_mul_cancel, map_one] }
map_one' := by
apply Units.ext
ext v
simp
map_mul' g h := by
apply Units.ext
ext v
simp [map_mul]
/-- The determinant character attached to a representation. BG writes its
kernel as `G* = G ∩ SL(V, F)`. -/
noncomputable def determinantCharacterOfRepresentation
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(ρ : Representation F G V) : G →* Fˣ :=
(LinearEquiv.det : (V ≃ₗ[F] V) →* Fˣ).comp
((LinearMap.GeneralLinearGroup.generalLinearEquiv F V).toMonoidHom.comp
(representationToGeneralLinearGroup ρ))
/-- Determinant-kernel subgroup `G*` from BG Thm 2.6. -/
noncomputable def determinantKernelSubgroup
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(ρ : Representation F G V) : Subgroup G :=
(determinantCharacterOfRepresentation ρ).ker
theorem mem_determinantKernelSubgroup
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
{ρ : Representation F G V} {g : G} :
g ∈ determinantKernelSubgroup ρ ↔
determinantCharacterOfRepresentation ρ g = 1 :=
Iff.rfl
theorem determinantKernelSubgroup_normal
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(ρ : Representation F G V) :
(determinantKernelSubgroup ρ).Normal := by
dsimp [determinantKernelSubgroup]
infer_instance
/-- On an invariant rank-one submodule and rank-one quotient, the determinant
character is the product of the two scalar characters. -/
private theorem determinantCharacter_eq_scalarCharacter_mul_quotient
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(W : Submodule F V) [Module.Free F W] [Module.Free F (V ⧸ W)]
(ρ : Representation F G V) (hW : ∀ g : G, W ≤ W.comap (ρ g))
(hdimW : Module.finrank F W = 1) (hdimQ : Module.finrank F (V ⧸ W) = 1)
(g : G) :
determinantCharacterOfRepresentation ρ g =
scalarCharacterOfFinrankEqOne hdimW (ρ.subrepresentation W hW) g *
scalarCharacterOfFinrankEqOne hdimQ (ρ.quotient W hW) g := by
ext
have hdet :
LinearMap.det (ρ g) =
LinearMap.det ((ρ g).restrict (hW g)) *
LinearMap.det (W.mapQ W (ρ g) (hW g)) :=
LinearMap.det_eq_det_mul_det W (ρ g) (hW g)
have hdetW :
LinearMap.det ((ρ g).restrict (hW g)) =
(scalarCharacterOfFinrankEqOne hdimW (ρ.subrepresentation W hW) g : F) := by
apply det_eq_scalar_of_finrank_eq_one hdimW
intro w
simpa [Representation.subrepresentation] using
scalarCharacterOfFinrankEqOne_apply_smul hdimW
(ρ.subrepresentation W hW) g w
have hdetQ :
LinearMap.det (W.mapQ W (ρ g) (hW g)) =
(scalarCharacterOfFinrankEqOne hdimQ (ρ.quotient W hW) g : F) := by
apply det_eq_scalar_of_finrank_eq_one hdimQ
intro v
simpa [Representation.quotient] using
scalarCharacterOfFinrankEqOne_apply_smul hdimQ (ρ.quotient W hW) g v
calc
(determinantCharacterOfRepresentation ρ g : F) = LinearMap.det (ρ g) := by
simp [determinantCharacterOfRepresentation, representationToGeneralLinearGroup]
_ = LinearMap.det ((ρ g).restrict (hW g)) *
LinearMap.det (W.mapQ W (ρ g) (hW g)) := hdet
_ = (scalarCharacterOfFinrankEqOne hdimW (ρ.subrepresentation W hW) g : F) *
(scalarCharacterOfFinrankEqOne hdimQ (ρ.quotient W hW) g : F) := by
rw [hdetW, hdetQ]
/-- Membership in the determinant kernel forces the two rank-one scalar
characters to have product one. -/
private theorem scalarCharacters_mul_eq_one_of_mem_determinantKernel
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(W : Submodule F V) [Module.Free F W] [Module.Free F (V ⧸ W)]
(ρ : Representation F G V) (hW : ∀ g : G, W ≤ W.comap (ρ g))
(hdimW : Module.finrank F W = 1) (hdimQ : Module.finrank F (V ⧸ W) = 1)
{g : G} (hg : g ∈ determinantKernelSubgroup ρ) :
scalarCharacterOfFinrankEqOne hdimW (ρ.subrepresentation W hW) g *
scalarCharacterOfFinrankEqOne hdimQ (ρ.quotient W hW) g = 1 := by
have hdet :=
determinantCharacter_eq_scalarCharacter_mul_quotient W ρ hW hdimW hdimQ g
have hgdet : determinantCharacterOfRepresentation ρ g = 1 :=
mem_determinantKernelSubgroup.mp hg
simpa [hdet] using hgdet
/-- On a preserved complement, the scalar character of the quotient `V/W`
is the same as the scalar character on the complementary line.
This is the bridge needed in BG Thm 2.6 q ≠ p to replace the determinant
formula's quotient scalar by the actual scalar on the second Maschke line. -/
private theorem scalarCharacter_quotient_eq_complement_of_isCompl
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(W U : Submodule F V) [Module.Free F U] [Module.Finite F U]
[Module.Free F (V ⧸ W)]
(ρ : Representation F G V)
(hcompl : IsCompl W U)
(hW : ∀ g : G, W ≤ W.comap (ρ g))
(hU : ∀ g : G, U ≤ U.comap (ρ g))
(hdimU : Module.finrank F U = 1) (hdimQ : Module.finrank F (V ⧸ W) = 1)
(g : G) :
scalarCharacterOfFinrankEqOne hdimQ (ρ.quotient W hW) g =
scalarCharacterOfFinrankEqOne hdimU (ρ.subrepresentation U hU) g := by
let φQ : G →* Fˣ := scalarCharacterOfFinrankEqOne hdimQ (ρ.quotient W hW)
let φU : G →* Fˣ := scalarCharacterOfFinrankEqOne hdimU (ρ.subrepresentation U hU)
have hU_ne_bot : U ≠ ⊥ := by
rw [← Submodule.one_le_finrank_iff]
omega
rcases Submodule.nonzero_mem_of_bot_lt (bot_lt_iff_ne_bot.mpr hU_ne_bot) with
⟨u, hu_ne_zero⟩
have hmk_ne_zero :
(Submodule.Quotient.mk (u : V) : V ⧸ W) ≠ 0 := by
intro hmk
have huW : (u : V) ∈ W := by
simpa [Submodule.Quotient.mk_eq_zero] using hmk
have huInf : (u : V) ∈ W ⊓ U := ⟨huW, u.2⟩
have huBot : (u : V) ∈ (⊥ : Submodule F V) := by
simpa [hcompl.inf_eq_bot] using huInf
exact hu_ne_zero (Subtype.ext (by simpa using huBot))
have hquot :=
scalarCharacterOfFinrankEqOne_apply_smul hdimQ (ρ.quotient W hW) g
(Submodule.Quotient.mk (u : V) : V ⧸ W)
change
(Submodule.Quotient.mk (ρ g (u : V)) : V ⧸ W) =
(φQ g : F) • (Submodule.Quotient.mk (u : V) : V ⧸ W) at hquot
have hsub :=
scalarCharacterOfFinrankEqOne_apply_smul hdimU (ρ.subrepresentation U hU) g u
have hsubV : ρ g (u : V) = (φU g : F) • (u : V) :=
congrArg Subtype.val hsub
have hscalar :
(φU g : F) • (Submodule.Quotient.mk (u : V) : V ⧸ W) =
(φQ g : F) • (Submodule.Quotient.mk (u : V) : V ⧸ W) := by
simpa [hsubV] using hquot
ext
have hdiff :
((φU g : F) - (φQ g : F)) •
(Submodule.Quotient.mk (u : V) : V ⧸ W) = 0 := by
rw [sub_smul, sub_eq_zero]
exact hscalar
have hcoeff : (φU g : F) - (φQ g : F) = 0 := by
exact (smul_eq_zero.mp hdiff).resolve_right hmk_ne_zero
exact (sub_eq_zero.mp hcoeff).symm
/-- Determinant-kernel elements have inverse scalar characters on two preserved
complementary rank-one lines. -/
private theorem scalarCharacters_complement_mul_eq_one_of_mem_determinantKernel
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(W U : Submodule F V) [Module.Free F W] [Module.Free F U] [Module.Free F (V ⧸ W)]
(ρ : Representation F G V)
(hcompl : IsCompl W U)
(hW : ∀ g : G, W ≤ W.comap (ρ g))
(hU : ∀ g : G, U ≤ U.comap (ρ g))
(hdimW : Module.finrank F W = 1)
(hdimU : Module.finrank F U = 1) (hdimQ : Module.finrank F (V ⧸ W) = 1)
{g : G} (hg : g ∈ determinantKernelSubgroup ρ) :
scalarCharacterOfFinrankEqOne hdimW (ρ.subrepresentation W hW) g *
scalarCharacterOfFinrankEqOne hdimU (ρ.subrepresentation U hU) g = 1 := by
have hprod :=
scalarCharacters_mul_eq_one_of_mem_determinantKernel W ρ hW hdimW hdimQ hg
have hquot_eq :=
scalarCharacter_quotient_eq_complement_of_isCompl W U ρ hcompl hW hU
hdimU hdimQ g
simpa [hquot_eq] using hprod
/-- A nontrivial odd-order determinant-kernel element has distinct scalars on
two preserved complementary rank-one lines.
This is the BG Thm 2.6 q ≠ p eigenvalue step: if the two scalars were equal,
their determinant product would make the common scalar square to one; odd order
then forces both scalars to be one, and faithfulness makes the element trivial. -/
private theorem scalarCharacters_ne_of_mem_determinantKernel_of_ne_one
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(W U : Submodule F V) [Module.Free F W] [Module.Free F U]
[Module.Finite F U] [Module.Free F (V ⧸ W)]
(ρ : Representation F G V)
(hfaithful : Function.Injective ρ)
(hcompl : IsCompl W U)
(hW : ∀ g : G, W ≤ W.comap (ρ g))
(hU : ∀ g : G, U ≤ U.comap (ρ g))
(hdimW : Module.finrank F W = 1)
(hdimU : Module.finrank F U = 1) (hdimQ : Module.finrank F (V ⧸ W) = 1)
{g : G} (hgdet : g ∈ determinantKernelSubgroup ρ)
(hoddg : Odd (orderOf g)) (hg_ne_one : g ≠ 1) :
scalarCharacterOfFinrankEqOne hdimW (ρ.subrepresentation W hW) g ≠
scalarCharacterOfFinrankEqOne hdimU (ρ.subrepresentation U hU) g := by
let φW : G →* Fˣ := scalarCharacterOfFinrankEqOne hdimW (ρ.subrepresentation W hW)
let φU : G →* Fˣ := scalarCharacterOfFinrankEqOne hdimU (ρ.subrepresentation U hU)
intro hsame
have hprod :
φW g * φU g = 1 :=
scalarCharacters_complement_mul_eq_one_of_mem_determinantKernel
W U ρ hcompl hW hU hdimW hdimU hdimQ hgdet
have hsq : (φW g) ^ 2 = 1 := by
simpa [φW, φU, pow_two, hsame] using hprod
have hpow_order : (φW g) ^ orderOf g = 1 := by
rw [← map_pow, pow_orderOf_eq_one, map_one]
have horder_dvd_two : orderOf (φW g) ∣ 2 := by
rw [orderOf_dvd_iff_pow_eq_one]
exact hsq
have horder_dvd_g : orderOf (φW g) ∣ orderOf g :=
orderOf_dvd_of_pow_eq_one hpow_order
have hodd_scalar : Odd (orderOf (φW g)) :=
hoddg.of_dvd_nat horder_dvd_g
have horder_pos : 0 < orderOf (φW g) := Odd.pos hodd_scalar
have horder_le_two : orderOf (φW g) ≤ 2 :=
Nat.le_of_dvd (by decide) horder_dvd_two
have horder_ne_two : orderOf (φW g) ≠ 2 := by
intro htwo
rw [htwo] at hodd_scalar
exact (by norm_num : ¬ Odd (2 : ℕ)) hodd_scalar
have horder_one : orderOf (φW g) = 1 := by omega
have hφW_one : φW g = 1 := orderOf_eq_one_iff.mp horder_one
have hφU_one : φU g = 1 := by
simpa [φW, φU, hsame] using hφW_one
have hg_one : g = 1 := by
apply hfaithful
ext v
obtain ⟨w, u, hv, _huniq⟩ := Submodule.existsUnique_add_of_isCompl hcompl v
have hw_fix : ρ g (w : V) = (w : V) := by
have hw_scalar :=
scalarCharacterOfFinrankEqOne_apply_smul hdimW
(ρ.subrepresentation W hW) g w
calc
ρ g (w : V) = (φW g : F) • (w : V) := congrArg Subtype.val hw_scalar
_ = (w : V) := by simp [hφW_one]
have hu_fix : ρ g (u : V) = (u : V) := by
have hu_scalar :=
scalarCharacterOfFinrankEqOne_apply_smul hdimU
(ρ.subrepresentation U hU) g u
calc
ρ g (u : V) = (φU g : F) • (u : V) := congrArg Subtype.val hu_scalar
_ = (u : V) := by simp [hφU_one]
calc
ρ g v = ρ g ((w : V) + (u : V)) := by rw [hv]
_ = ρ g (w : V) + ρ g (u : V) := map_add (ρ g) (w : V) (u : V)
_ = (w : V) + (u : V) := by rw [hw_fix, hu_fix]
_ = v := hv
_ = (ρ 1) v := by simp
exact hg_ne_one hg_one
/-- A nontrivial odd-order determinant-kernel element with two complementary
rank-one eigenspaces exhausts the rank-one subrepresentations.
This packages the BG Thm 2.6 q ≠ p step that the distinct eigenvalues of
`x ∈ K#` make the two Maschke lines the only one-dimensional `K`-submodules. -/
private theorem rank_one_subrepresentation_eq_left_or_right_of_determinantKernel_element
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(ρ : Representation F G V)
(hfaithful : Function.Injective ρ)
(W U : Subrepresentation ρ)
[Module.Free F W.toSubmodule] [Module.Free F U.toSubmodule]
[Module.Finite F U.toSubmodule] [Module.Free F (V ⧸ W.toSubmodule)]
(hcompl : IsCompl W.toSubmodule U.toSubmodule)
(hdimW : Module.finrank F W.toSubmodule = 1)
(hdimU : Module.finrank F U.toSubmodule = 1)
(hdimQ : Module.finrank F (V ⧸ W.toSubmodule) = 1)
{x : G} (hxdet : x ∈ determinantKernelSubgroup ρ)
(hoddx : Odd (orderOf x)) (hx_ne_one : x ≠ 1) :
∀ L : Subrepresentation ρ,
Module.finrank F L.toSubmodule = 1 → L = W ∨ L = U := by
have hW : ∀ g : G, W.toSubmodule ≤ W.toSubmodule.comap (ρ g) := by
intro g v hv
exact W.apply_mem_toSubmodule g hv
have hU : ∀ g : G, U.toSubmodule ≤ U.toSubmodule.comap (ρ g) := by
intro g v hv
exact U.apply_mem_toSubmodule g hv
let φW : G →* Fˣ :=
scalarCharacterOfFinrankEqOne hdimW (ρ.subrepresentation W.toSubmodule hW)
let φU : G →* Fˣ :=
scalarCharacterOfFinrankEqOne hdimU (ρ.subrepresentation U.toSubmodule hU)
have hdistinct_units : φW x ≠ φU x :=
scalarCharacters_ne_of_mem_determinantKernel_of_ne_one
W.toSubmodule U.toSubmodule ρ hfaithful hcompl hW hU hdimW hdimU hdimQ
hxdet hoddx hx_ne_one
have hdistinct : (φW x : F) ≠ (φU x : F) := by
intro hval
exact hdistinct_units (Units.ext hval)
intro L hLdim
have hLstable : L.toSubmodule ≤ L.toSubmodule.comap (ρ x) := by
intro v hv
exact L.apply_mem_toSubmodule x hv
have hWscalar : ∀ w ∈ W.toSubmodule, ρ x w = (φW x : F) • w := by
intro w hw
have hw_scalar :=
scalarCharacterOfFinrankEqOne_apply_smul hdimW
(ρ.subrepresentation W.toSubmodule hW) x ⟨w, hw⟩
exact congrArg Subtype.val hw_scalar
have hUscalar : ∀ u ∈ U.toSubmodule, ρ x u = (φU x : F) • u := by
intro u hu
have hu_scalar :=
scalarCharacterOfFinrankEqOne_apply_smul hdimU
(ρ.subrepresentation U.toSubmodule hU) x ⟨u, hu⟩
exact congrArg Subtype.val hu_scalar
rcases rank_one_invariant_submodule_eq_left_or_right_of_distinct_scalars
W.toSubmodule U.toSubmodule L.toSubmodule (ρ x)
hcompl hdimW hdimU hLdim hLstable (φW x : F) (φU x : F)
hdistinct hWscalar hUscalar with hLW | hLU
· left
exact Subrepresentation.toSubmodule_injective hLW
· right
exact Subrepresentation.toSubmodule_injective hLU
/-- Ambient form of
`rank_one_subrepresentation_eq_left_or_right_of_determinantKernel_element`.
If `K ≤ G*` and `G` has odd order, a nontrivial element of `K` supplies the
distinct-scalar witness for the restricted `K`-representation. This is the
form needed before applying the normal-conjugate line bridge. -/
private theorem rank_one_subrepresentation_eq_left_or_right_of_determinantKernel_subgroup
{F : Type*} [Field F] {G : Type*} [Group G] [Finite G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(ρ : Representation F G V) (hfaithful : Function.Injective ρ)
(hodd : Odd (Nat.card G))
(K : Subgroup G) (hKle : K ≤ determinantKernelSubgroup ρ)
(W U : Subrepresentation (ρ.comp K.subtype))
[Module.Free F W.toSubmodule] [Module.Free F U.toSubmodule]
[Module.Finite F U.toSubmodule] [Module.Free F (V ⧸ W.toSubmodule)]
(hcompl : IsCompl W.toSubmodule U.toSubmodule)
(hdimW : Module.finrank F W.toSubmodule = 1)
(hdimU : Module.finrank F U.toSubmodule = 1)
(hdimQ : Module.finrank F (V ⧸ W.toSubmodule) = 1)
{x : K} (hx_ne_one : x ≠ 1) :
∀ L : Subrepresentation (ρ.comp K.subtype),
Module.finrank F L.toSubmodule = 1 → L = W ∨ L = U := by
have hρK_faithful : Function.Injective (ρ.comp K.subtype) := by
intro a b hab
apply Subtype.ext
exact hfaithful (by simpa using hab)
have hxdet : x ∈ determinantKernelSubgroup (ρ.comp K.subtype) := by
have hxdetG : (x : G) ∈ determinantKernelSubgroup ρ := hKle x.2
rw [mem_determinantKernelSubgroup] at hxdetG ⊢
simpa [determinantCharacterOfRepresentation, representationToGeneralLinearGroup] using hxdetG
have hoddxG : Odd (orderOf (x : G)) :=
hodd.of_dvd_nat (orderOf_dvd_natCard (x : G))
have hoddx : Odd (orderOf x) := by
simpa [Subgroup.orderOf_coe x] using hoddxG
exact rank_one_subrepresentation_eq_left_or_right_of_determinantKernel_element
(ρ.comp K.subtype) hρK_faithful W U hcompl hdimW hdimU hdimQ
hxdet hoddx hx_ne_one
/-- Determinant-kernel uniqueness forces every ambient conjugate of one Maschke
line to be one of the two Maschke lines.
This is the next BG Thm 2.6 q ≠ p bridge after choosing a nontrivial
`x : K`: normality makes the conjugate a `K`-subrepresentation, and the
distinct-scalar uniqueness pins it to `W` or `U`. -/
private theorem conjugateSubrepresentation_eq_left_or_right_of_determinantKernel_subgroup
{F : Type*} [Field F] {G : Type*} [Group G] [Finite G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(ρ : Representation F G V) (hfaithful : Function.Injective ρ)
(hodd : Odd (Nat.card G))
(K : Subgroup G) (hKnormal : K.Normal) (hKle : K ≤ determinantKernelSubgroup ρ)
(W U : Subrepresentation (ρ.comp K.subtype))
[Module.Free F W.toSubmodule] [Module.Free F U.toSubmodule]
[Module.Finite F U.toSubmodule] [Module.Free F (V ⧸ W.toSubmodule)]
(hcompl : IsCompl W.toSubmodule U.toSubmodule)
(hdimW : Module.finrank F W.toSubmodule = 1)
(hdimU : Module.finrank F U.toSubmodule = 1)
(hdimQ : Module.finrank F (V ⧸ W.toSubmodule) = 1)
{x : K} (hx_ne_one : x ≠ 1) (g : G) :
conjugateSubrepresentationOfNormal K hKnormal ρ W g = W ∨
conjugateSubrepresentationOfNormal K hKnormal ρ W g = U := by
exact conjugateSubrepresentation_eq_left_or_right_of_rank_one_unique
K hKnormal ρ W U hdimW
(rank_one_subrepresentation_eq_left_or_right_of_determinantKernel_subgroup
ρ hfaithful hodd K hKle W U hcompl hdimW hdimU hdimQ hx_ne_one)
g
/-- The two conjugate-line alternatives are coherent: an ambient element either
preserves both Maschke lines or swaps them.
The impossible mixed cases would make the two conjugate complementary lines
land on the same nonzero rank-one line. -/
private theorem conjugateSubrepresentations_stay_or_swap_of_determinantKernel_subgroup
{F : Type*} [Field F] {G : Type*} [Group G] [Finite G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(ρ : Representation F G V) (hfaithful : Function.Injective ρ)
(hodd : Odd (Nat.card G))
(K : Subgroup G) (hKnormal : K.Normal) (hKle : K ≤ determinantKernelSubgroup ρ)
(W U : Subrepresentation (ρ.comp K.subtype))
[Module.Free F W.toSubmodule] [Module.Free F U.toSubmodule]
[Module.Finite F W.toSubmodule] [Module.Finite F U.toSubmodule]
[Module.Free F (V ⧸ W.toSubmodule)] [Module.Free F (V ⧸ U.toSubmodule)]
(hcompl : IsCompl W.toSubmodule U.toSubmodule)
(hdimW : Module.finrank F W.toSubmodule = 1)
(hdimU : Module.finrank F U.toSubmodule = 1)
(hdimQW : Module.finrank F (V ⧸ W.toSubmodule) = 1)
(hdimQU : Module.finrank F (V ⧸ U.toSubmodule) = 1)
{x : K} (hx_ne_one : x ≠ 1) (g : G) :
(conjugateSubrepresentationOfNormal K hKnormal ρ W g = W ∧
conjugateSubrepresentationOfNormal K hKnormal ρ U g = U) ∨
(conjugateSubrepresentationOfNormal K hKnormal ρ W g = U ∧
conjugateSubrepresentationOfNormal K hKnormal ρ U g = W) := by
have hW_ne_bot : W.toSubmodule ≠ ⊥ := by
rw [← Submodule.one_le_finrank_iff]
omega
have hU_ne_bot : U.toSubmodule ≠ ⊥ := by
rw [← Submodule.one_le_finrank_iff]
omega
have hcomplg :
IsCompl
(conjugateSubrepresentationOfNormal K hKnormal ρ W g).toSubmodule
(conjugateSubrepresentationOfNormal K hKnormal ρ U g).toSubmodule :=
isCompl_conjugateSubrepresentationOfNormal K hKnormal ρ W U g hcompl
rcases conjugateSubrepresentation_eq_left_or_right_of_determinantKernel_subgroup
ρ hfaithful hodd K hKnormal hKle W U hcompl hdimW hdimU hdimQW
hx_ne_one g with hWW | hWU
· rcases conjugateSubrepresentation_eq_left_or_right_of_determinantKernel_subgroup
ρ hfaithful hodd K hKnormal hKle U W hcompl.symm hdimU hdimW hdimQU
hx_ne_one g with hUU | hUW
· exact Or.inl ⟨hWW, hUU⟩
· exfalso
have hW_bot : W.toSubmodule = ⊥ := by
have hleft := congrArg Subrepresentation.toSubmodule hWW
have hright := congrArg Subrepresentation.toSubmodule hUW
simpa [hleft, hright] using hcomplg.inf_eq_bot
exact hW_ne_bot hW_bot
· rcases conjugateSubrepresentation_eq_left_or_right_of_determinantKernel_subgroup
ρ hfaithful hodd K hKnormal hKle U W hcompl.symm hdimU hdimW hdimQU
hx_ne_one g with hUU | hUW
· exfalso
have hU_bot : U.toSubmodule = ⊥ := by
have hleft := congrArg Subrepresentation.toSubmodule hWU
have hright := congrArg Subrepresentation.toSubmodule hUU
simpa [hleft, hright] using hcomplg.inf_eq_bot
exact hU_ne_bot hU_bot
· exact Or.inr ⟨hWU, hUW⟩
/-- Coherent `comap` form of the two-line alternatives: each ambient element
either preserves both lines or swaps them. -/
private theorem both_lines_le_comap_stay_or_swap_of_determinantKernel_subgroup
{F : Type*} [Field F] {G : Type*} [Group G] [Finite G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(ρ : Representation F G V) (hfaithful : Function.Injective ρ)
(hodd : Odd (Nat.card G))
(K : Subgroup G) (hKnormal : K.Normal) (hKle : K ≤ determinantKernelSubgroup ρ)
(W U : Subrepresentation (ρ.comp K.subtype))
[Module.Free F W.toSubmodule] [Module.Free F U.toSubmodule]
[Module.Finite F W.toSubmodule] [Module.Finite F U.toSubmodule]
[Module.Free F (V ⧸ W.toSubmodule)] [Module.Free F (V ⧸ U.toSubmodule)]
(hcompl : IsCompl W.toSubmodule U.toSubmodule)
(hdimW : Module.finrank F W.toSubmodule = 1)
(hdimU : Module.finrank F U.toSubmodule = 1)
(hdimQW : Module.finrank F (V ⧸ W.toSubmodule) = 1)
(hdimQU : Module.finrank F (V ⧸ U.toSubmodule) = 1)
{x : K} (hx_ne_one : x ≠ 1) (g : G) :
(W.toSubmodule ≤ W.toSubmodule.comap (ρ g) ∧
U.toSubmodule ≤ U.toSubmodule.comap (ρ g)) ∨
(W.toSubmodule ≤ U.toSubmodule.comap (ρ g) ∧
U.toSubmodule ≤ W.toSubmodule.comap (ρ g)) := by
rcases conjugateSubrepresentations_stay_or_swap_of_determinantKernel_subgroup
ρ hfaithful hodd K hKnormal hKle W U hcompl hdimW hdimU hdimQW hdimQU
hx_ne_one g with hstay | hswap
· exact Or.inl
⟨le_comap_of_conjugateSubrepresentation_eq K hKnormal ρ W W hstay.1,
le_comap_of_conjugateSubrepresentation_eq K hKnormal ρ U U hstay.2⟩
· exact Or.inr
⟨le_comap_of_conjugateSubrepresentation_eq K hKnormal ρ W U hswap.1,
le_comap_of_conjugateSubrepresentation_eq K hKnormal ρ U W hswap.2⟩
/-- An injective character into `Fˣ` forces the source group to be abelian. -/
private theorem commutative_of_injective_units_character
{F : Type*} [Field F] {G : Type*} [Group G] (φ : G →* Fˣ)
(hφ : Function.Injective φ) :
Std.Commutative (· * · : G → G → G) := by
constructor
intro x y
apply hφ
calc
φ (x * y) = φ x * φ y := map_mul φ x y
_ = φ y * φ x := by rw [mul_comm]
_ = φ (y * x) := (map_mul φ y x).symm
private theorem determinantCharacter_injective_of_kernel_eq_bot
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(ρ : Representation F G V) (hker : determinantKernelSubgroup ρ = ⊥) :
Function.Injective (determinantCharacterOfRepresentation ρ) :=
(MonoidHom.ker_eq_bot_iff _).mp (by
simpa [determinantKernelSubgroup] using hker)
/-- BG Thm 2.6 determinant endpoint: if `G* = ker(det ∘ ρ)` is trivial, then
the determinant character embeds `G` into the abelian group `Fˣ`. -/
theorem commutative_of_determinantKernel_eq_bot
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(ρ : Representation F G V) (hker : determinantKernelSubgroup ρ = ⊥) :
Std.Commutative (· * · : G → G → G) :=
commutative_of_injective_units_character (determinantCharacterOfRepresentation ρ)
(determinantCharacter_injective_of_kernel_eq_bot ρ hker)
/-- If `G` preserves a rank-one submodule and rank-one quotient, then `G'`
acts trivially on both.
This is the `G' ≤ C_G(W) ∩ C_G(V/W)` half of BG Thm 2.6, q = p. Unlike the
p-subgroup bridge above, it does not use characteristic `p`: scalar characters
to `Fˣ` kill commutators because `Fˣ` is abelian. -/
private theorem commutator_le_fixedOnSubmoduleAndQuotientSubgroup_of_rank_one_subquotients
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(W : Submodule F V) [Module.Free F W] [Module.Free F (V ⧸ W)]
(ρ : Representation F G V) (hW : ∀ g : G, W ≤ W.comap (ρ g))
(hdimW : Module.finrank F W = 1) (hdimQ : Module.finrank F (V ⧸ W) = 1) :
commutator G ≤ fixedOnSubmoduleAndQuotientSubgroup W ρ := by
let φW : G →* Fˣ :=
scalarCharacterOfFinrankEqOne hdimW (ρ.subrepresentation W hW)
let φQ : G →* Fˣ :=
scalarCharacterOfFinrankEqOne hdimQ (ρ.quotient W hW)
have hkerW : commutator G ≤ φW.ker :=
commutator_le_ker_of_units_character φW
have hkerQ : commutator G ≤ φQ.ker :=
commutator_le_ker_of_units_character φQ
intro g hg
rw [mem_fixedOnSubmoduleAndQuotientSubgroup]
constructor
· intro w hw
have hφW : φW g = 1 := MonoidHom.mem_ker.mp (hkerW hg)
have hsub :=
scalarCharacterOfFinrankEqOne_apply_smul hdimW
(ρ.subrepresentation W hW) g ⟨w, hw⟩
calc
ρ g w = (φW g : F) • w := congrArg Subtype.val hsub
_ = w := by simp [hφW]
· intro v
have hφQ : φQ g = 1 := MonoidHom.mem_ker.mp (hkerQ hg)
have hq :=
scalarCharacterOfFinrankEqOne_apply_smul hdimQ (ρ.quotient W hW) g
(Submodule.Quotient.mk v : V ⧸ W)
change Submodule.Quotient.mk (ρ g v) =
(φQ g : F) • (Submodule.Quotient.mk v : V ⧸ W) at hq
have hq_one :
(Submodule.Quotient.mk (ρ g v) : V ⧸ W) =
(Submodule.Quotient.mk v : V ⧸ W) := by
simpa [hφQ] using hq
simpa [Submodule.Quotient.eq] using hq_one
/-- An element acting trivially on a submodule and on the quotient is the identity
if it also preserves a complementary submodule.
This is the diagonal-form bridge for BG Thm 2.6, q ≠ p: once the two
one-dimensional `K`-submodules are both `G`-invariant, a commutator that is
trivial on one line and on the quotient is trivial on the complementary line
as well. -/
private theorem eq_one_of_mem_fixedOnSubmoduleAndQuotientSubgroup_of_preserves_complement
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(W U : Submodule F V) (ρ : Representation F G V)
(hfaithful : Function.Injective ρ) (hcompl : IsCompl W U)
(hU : ∀ g : G, U ≤ U.comap (ρ g))
{g : G} (hg : g ∈ fixedOnSubmoduleAndQuotientSubgroup W ρ) :
g = 1 := by
apply hfaithful
ext v
obtain ⟨w, u, hv, _huniq⟩ := Submodule.existsUnique_add_of_isCompl hcompl v
have hgW : ∀ w ∈ W, ρ g w = w :=
(mem_fixedOnSubmoduleAndQuotientSubgroup.mp hg).1
have hgQ : ∀ v, ρ g v - v ∈ W :=
(mem_fixedOnSubmoduleAndQuotientSubgroup.mp hg).2
have hfixU : ρ g (u : V) = (u : V) := by
have hdiffW : ρ g (u : V) - (u : V) ∈ W := hgQ u
have hdiffU : ρ g (u : V) - (u : V) ∈ U :=
U.sub_mem (hU g u.2) u.2
have hdiff_bot : ρ g (u : V) - (u : V) ∈ (⊥ : Submodule F V) := by
simpa [hcompl.inf_eq_bot] using (show ρ g (u : V) - (u : V) ∈ W ⊓ U from
⟨hdiffW, hdiffU⟩)
exact sub_eq_zero.mp (by simpa using hdiff_bot)
calc
ρ g v = ρ g ((w : V) + (u : V)) := by rw [hv]
_ = ρ g (w : V) + ρ g (u : V) := map_add (ρ g) (w : V) (u : V)
_ = (w : V) + (u : V) := by rw [hgW (w : V) w.2, hfixU]
_ = v := hv
_ = (ρ 1) v := by simp
/-- If a faithful representation preserves complementary one-dimensional
submodules, then the group is abelian.
In the q≠p branch of BG Thm 2.6, Maschke and algebraic closedness give
`V = W₁ ⊕ W₂` with both lines `G`-invariant. The two rank-one scalar
characters kill commutators on `W₁` and on `V/W₁`; the complement prevents any
remaining unipotent shear. -/
private theorem commutative_of_faithful_representation_preserves_rank_one_complement
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(W U : Submodule F V) [Module.Free F W] [Module.Free F (V ⧸ W)]
(ρ : Representation F G V) (hfaithful : Function.Injective ρ)
(hcompl : IsCompl W U) (hW : ∀ g : G, W ≤ W.comap (ρ g))
(hU : ∀ g : G, U ≤ U.comap (ρ g))
(hdimW : Module.finrank F W = 1)
(hdimQ : Module.finrank F (V ⧸ W) = 1) :
Std.Commutative (· * · : G → G → G) := by
constructor
intro x y
rw [← commutatorElement_eq_one_iff_mul_comm]
apply eq_one_of_mem_fixedOnSubmoduleAndQuotientSubgroup_of_preserves_complement
W U ρ hfaithful hcompl hU
exact commutator_le_fixedOnSubmoduleAndQuotientSubgroup_of_rank_one_subquotients
W ρ hW hdimW hdimQ
(Subgroup.commutator_mem_commutator (Subgroup.mem_top x) (Subgroup.mem_top y))
/-- Odd order turns a stay/swap dichotomy for two complementary lines into
actual preservation of both lines.
This is the line-permutation step in BG Thm 2.6, q ≠ p, stated without
choosing an explicit `Fin 2` action. If an element swapped the two lines, then
its square would preserve them; since the element has odd order, the element
itself is a power of its square, contradiction with the nonzero first line and
the direct-sum decomposition. -/
private theorem preserves_of_stay_or_swap_rank_one_complement_of_odd
{F : Type*} [Field F] {G : Type*} [Group G] [Finite G]
{V : Type*} [AddCommGroup V] [Module F V]
(W U : Submodule F V) [Module.Finite F W]
(ρ : Representation F G V)
(hodd : Odd (Nat.card G)) (hcompl : IsCompl W U)
(hstay_swap : ∀ g : G,
(W ≤ W.comap (ρ g) ∧ U ≤ U.comap (ρ g)) ∨
(W ≤ U.comap (ρ g) ∧ U ≤ W.comap (ρ g)))
(hdimW : Module.finrank F W = 1) :
∀ g : G, W ≤ W.comap (ρ g) ∧ U ≤ U.comap (ρ g) := by
intro g
rcases hstay_swap g with hstay | hswap
· exact hstay
· exfalso
have hW_ne_bot : W ≠ ⊥ := by
rw [← Submodule.one_le_finrank_iff]
omega
have hg2W : W ≤ W.comap (ρ (g ^ 2)) := by
intro w hw
rw [pow_two, map_mul]
exact hswap.2 (hswap.1 hw)
have hpowW : ∀ n : ℕ, W ≤ W.comap (ρ ((g ^ 2) ^ n)) := by
intro n
induction n with
| zero =>
intro w hw
simpa using hw
| succ n ih =>
intro w hw
rw [pow_succ, map_mul]
exact ih (hg2W hw)
have hoddg : Odd (orderOf g) := hodd.of_dvd_nat (orderOf_dvd_natCard g)
rcases hoddg with ⟨k, hk⟩
have hg_pow_square : g = (g ^ 2) ^ (k + 1) := by
calc
g = g ^ (orderOf g + 1) := by
rw [pow_succ, pow_orderOf_eq_one, one_mul]
_ = g ^ (2 * (k + 1)) := by
congr 1
omega
_ = (g ^ 2) ^ (k + 1) := by
rw [pow_mul]
have hgW : W ≤ W.comap (ρ g) := by
rw [hg_pow_square]
exact hpowW (k + 1)
rcases Submodule.exists_mem_ne_zero_of_ne_bot hW_ne_bot with ⟨w, hwW, hw_ne_zero⟩
have hρgw_bot : ρ g w ∈ (⊥ : Submodule F V) := by
have hρgw_inf : ρ g w ∈ W ⊓ U := ⟨hgW hwW, hswap.1 hwW⟩
simpa [hcompl.inf_eq_bot] using hρgw_inf
have hρgw_zero : ρ g w = 0 := by
simpa using hρgw_bot
have hw_zero : w = 0 := by
have hleft : ρ g⁻¹ (ρ g w) = w := by
calc
ρ g⁻¹ (ρ g w) = ((ρ g⁻¹) * (ρ g)) w := rfl
_ = ρ (g⁻¹ * g) w := by rw [map_mul]
_ = w := by simp
simpa [hleft] using congrArg (ρ g⁻¹) hρgw_zero
exact hw_ne_zero hw_zero
/-- Stay/swap form of the odd-order two-line bridge.
This avoids building a separate `Fin 2` action when the proof has already
produced the concrete dichotomy that every element preserves both lines or
swaps them. -/
private theorem commutative_of_faithful_representation_stay_or_swap_rank_one_complement_of_odd
{F : Type*} [Field F] {G : Type*} [Group G] [Finite G]
{V : Type*} [AddCommGroup V] [Module F V]
(W U : Submodule F V) [Module.Free F W] [Module.Finite F W]
[Module.Free F (V ⧸ W)]
(ρ : Representation F G V) (hfaithful : Function.Injective ρ)
(hodd : Odd (Nat.card G)) (hcompl : IsCompl W U)
(hstay_swap : ∀ g : G,
(W ≤ W.comap (ρ g) ∧ U ≤ U.comap (ρ g)) ∨
(W ≤ U.comap (ρ g) ∧ U ≤ W.comap (ρ g)))
(hdimW : Module.finrank F W = 1)
(hdimQ : Module.finrank F (V ⧸ W) = 1) :
Std.Commutative (· * · : G → G → G) := by
have hpreserve :=
preserves_of_stay_or_swap_rank_one_complement_of_odd
W U ρ hodd hcompl hstay_swap hdimW
exact commutative_of_faithful_representation_preserves_rank_one_complement
W U ρ hfaithful hcompl
(fun g => (hpreserve g).1) (fun g => (hpreserve g).2) hdimW hdimQ
/-- Determinant-kernel line-pair bridge for the q≠p branch.
Once Maschke has produced two complementary rank-one `K`-submodules and a
nontrivial element of the normal subgroup `K ≤ G*`, the determinant-kernel
uniqueness argument gives a stay/swap dichotomy for every ambient group
element. Odd order removes the swap case, so the faithful diagonal bridge
makes `G` abelian. -/
theorem commutative_of_determinantKernel_subgroup_rank_one_complement
{F : Type*} [Field F] {G : Type*} [Group G] [Finite G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(ρ : Representation F G V) (hfaithful : Function.Injective ρ)
(hodd : Odd (Nat.card G))
(K : Subgroup G) (hKnormal : K.Normal) (hKle : K ≤ determinantKernelSubgroup ρ)
(W U : Subrepresentation (ρ.comp K.subtype))
[Module.Free F W.toSubmodule] [Module.Free F U.toSubmodule]
[Module.Finite F W.toSubmodule] [Module.Finite F U.toSubmodule]
[Module.Free F (V ⧸ W.toSubmodule)] [Module.Free F (V ⧸ U.toSubmodule)]
(hcompl : IsCompl W.toSubmodule U.toSubmodule)
(hdimW : Module.finrank F W.toSubmodule = 1)
(hdimU : Module.finrank F U.toSubmodule = 1)
(hdimQW : Module.finrank F (V ⧸ W.toSubmodule) = 1)
(hdimQU : Module.finrank F (V ⧸ U.toSubmodule) = 1)
{x : K} (hx_ne_one : x ≠ 1) :
Std.Commutative (· * · : G → G → G) := by
exact
commutative_of_faithful_representation_stay_or_swap_rank_one_complement_of_odd
W.toSubmodule U.toSubmodule ρ hfaithful hodd hcompl
(fun g =>
both_lines_le_comap_stay_or_swap_of_determinantKernel_subgroup
ρ hfaithful hodd K hKnormal hKle W U hcompl hdimW hdimU hdimQW hdimQU
hx_ne_one g)
hdimW hdimQW
/-- Two-dimensional form of
`commutator_le_fixedOnSubmoduleAndQuotientSubgroup_of_rank_one_subquotients`. -/
private theorem commutator_le_fixedOnSubmoduleAndQuotientSubgroup_of_finrank_two
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(W : Submodule F V) [Module.Free F W] [Module.Free F (V ⧸ W)]
(ρ : Representation F G V) (hW : ∀ g : G, W ≤ W.comap (ρ g))
(hdim : Module.finrank F V = 2) (hW_ne_bot : W ≠ ⊥) (hW_ne_top : W ≠ ⊤) :
commutator G ≤ fixedOnSubmoduleAndQuotientSubgroup W ρ := by
rcases rank_one_subquotients_of_finrank_two W hdim hW_ne_bot hW_ne_top with
⟨hdimW, hdimQ⟩
exact commutator_le_fixedOnSubmoduleAndQuotientSubgroup_of_rank_one_subquotients
W ρ hW hdimW hdimQ
/-- A normal p-subgroup is contained in every Sylow p-subgroup.
This is the Sylow-conjugacy bridge used in BG Thm 2.6(b) after a normal
p-subgroup containing `G'` has been constructed. -/
private theorem normal_pSubgroup_le_sylow
{p : ℕ} [Fact p.Prime] {G : Type*} [Group G] [Finite (Sylow p G)]
(N : Subgroup G) (hNnormal : N.Normal) (hN : IsPGroup p N) (P : Sylow p G) :
N ≤ (P : Subgroup G) := by
have : N.Normal := hNnormal
obtain ⟨Q, hNQ⟩ := hN.exists_le_sylow
obtain ⟨g, hgQ⟩ := MulAction.exists_smul_eq G Q P
calc (N : Subgroup G)
= MulAut.conj g • N := (Subgroup.Normal.conj_smul_eq_self g N).symm
_ ≤ MulAut.conj g • (Q : Subgroup G) :=
Subgroup.pointwise_smul_le_pointwise_smul_iff.mpr hNQ
_ = ↑(g • Q) := Sylow.coe_subgroup_smul.symm
_ = ↑P := by rw [hgQ]
/-- If `G'` lies in a normal p-subgroup, then `G'` lies in every Sylow
p-subgroup. -/
private theorem commutator_le_sylow_of_le_normal_pSubgroup
{p : ℕ} [Fact p.Prime] {G : Type*} [Group G] [Finite (Sylow p G)]
(N : Subgroup G) (hNnormal : N.Normal) (hN : IsPGroup p N)
(hcomm : commutator G ≤ N) (P : Sylow p G) :
commutator G ≤ (P : Subgroup G) :=
hcomm.trans (normal_pSubgroup_le_sylow N hNnormal hN P)
/-- Two-dimensional wrapper for
`subgroup_commutative_of_rank_one_subquotients`.
This is the form needed in BG Thm 2.6, q = p, after constructing a nonzero
proper invariant fixed-space `W = C_V(K)`. -/
private theorem subgroup_commutative_of_finrank_two_invariant_submodule
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] {V : Type*} [AddCommGroup V] [Module F V]
[Module.Finite F V]
(H : Subgroup G) (W : Submodule F V) [Module.Free F W] [Module.Free F (V ⧸ W)]
(ρ : Representation F G V) (hfaithful : Function.Injective ρ) (hH : IsPGroup p H)
(hW : ∀ h : H, W ≤ W.comap (ρ h))
(hdim : Module.finrank F V = 2) (hW_ne_bot : W ≠ ⊥) (hW_ne_top : W ≠ ⊤) :
Std.Commutative (· * · : H → H → H) := by
rcases rank_one_subquotients_of_finrank_two W hdim hW_ne_bot hW_ne_top with
⟨hdimW, hdimQ⟩
exact subgroup_commutative_of_rank_one_subquotients
H W ρ hfaithful hH hW hdimW hdimQ
/-- If a normal p-subgroup has a proper fixed space in a faithful two-dimensional
representation over characteristic `p`, then every p-subgroup preserving that
fixed space is abelian.
In BG Thm 2.6, q = p, this is applied to
`K = Ω₁(Z(O_p(G^*)))` and `W = C_V(K)`. Nontriviality of `W` is supplied by
`IsPGroup.invariants_ne_bot`; normality of `K` supplies `G`-invariance. -/
private theorem subgroup_commutative_of_normal_p_fixed_space_proper
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] [Finite G] {V : Type*} [AddCommGroup V] [Module F V]
[Module.Finite F V]
(K H : Subgroup G) [K.Normal] (ρ : Representation F G V)
(hfaithful : Function.Injective ρ) (hK : IsPGroup p K) (hH : IsPGroup p H)
(hdim : Module.finrank F V = 2)
(hfixed_ne_top : Representation.invariants (ρ.comp K.subtype) ≠ ⊤) :
Std.Commutative (· * · : H → H → H) := by
let W : Submodule F V := Representation.invariants (ρ.comp K.subtype)
have hV_ne_bot : (⊤ : Submodule F V) ≠ ⊥ := by
intro hbot
have htop : Module.finrank F (⊤ : Submodule F V) = 2 := by
simp [hdim]
have hzero : Module.finrank F (⊤ : Submodule F V) = 0 := by
rw [hbot, finrank_bot]
omega
have hW_ne_bot : W ≠ ⊥ := by
simpa [W] using hK.invariants_ne_bot (ρ.comp K.subtype) hV_ne_bot
have hW_ne_top : W ≠ ⊤ := by
simpa [W] using hfixed_ne_top
have hW_invariant : ∀ h : H, W ≤ W.comap (ρ h) := by
intro h
exact Representation.le_comap_invariants ρ K h
exact subgroup_commutative_of_finrank_two_invariant_submodule
H W ρ hfaithful hH hW_invariant hdim hW_ne_bot hW_ne_top
/-- In a faithful representation, a nontrivial subgroup cannot fix the whole
space pointwise. -/
private theorem invariants_ne_top_of_faithful_subgroup_ne_bot
{F : Type*} [Field F] {G : Type*} [Group G]
{V : Type*} [AddCommGroup V] [Module F V]
(K : Subgroup G) (ρ : Representation F G V) (hfaithful : Function.Injective ρ)
(hK_ne_bot : K ≠ ⊥) :
Representation.invariants (ρ.comp K.subtype) ≠ ⊤ := by
intro htop
apply hK_ne_bot
ext g
constructor
· intro hg
let k : K := ⟨g, hg⟩
have hk_fixed : ∀ v : V, ρ g v = v := by
intro v
have hv : v ∈ Representation.invariants (ρ.comp K.subtype) := by
rw [htop]
exact Submodule.mem_top
simpa [k] using (Representation.mem_invariants (ρ.comp K.subtype) v).mp hv k
have hρg : ρ g = 1 := by
ext v
exact hk_fixed v
have hg_one : g = 1 := by
apply hfaithful
simp [hρg]
simp [hg_one]
· intro hg
have hg_one : g = 1 := by
simpa using hg
simp [hg_one]
/-- Nontrivial-normal-subgroup version of
`subgroup_commutative_of_normal_p_fixed_space_proper`.
This is the closest current Lean entrypoint to BG Thm 2.6, q = p: once the
nontrivial normal p-subgroup `K` is constructed, faithful two-dimensionality
forces its fixed space to be nonzero and proper, hence every p-subgroup is
abelian. -/
private theorem subgroup_commutative_of_nontrivial_normal_p_fixed_space
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] [Finite G] {V : Type*} [AddCommGroup V] [Module F V]
[Module.Finite F V]
(K H : Subgroup G) [K.Normal] (ρ : Representation F G V)
(hfaithful : Function.Injective ρ) (hK : IsPGroup p K) (hH : IsPGroup p H)
(hdim : Module.finrank F V = 2) (hK_ne_bot : K ≠ ⊥) :
Std.Commutative (· * · : H → H → H) :=
subgroup_commutative_of_normal_p_fixed_space_proper K H ρ hfaithful hK hH hdim
(invariants_ne_top_of_faithful_subgroup_ne_bot K ρ hfaithful hK_ne_bot)
/-- Nontrivial-normal-p-subgroup form of
`commutator_le_fixedOnSubmoduleAndQuotientSubgroup_of_finrank_two`.
This packages the `G' ≤ C_G(W) ∩ C_G(V/W)` bridge for the same
`W = C_V(K)` used in BG Thm 2.6, q = p. -/
private theorem
commutator_le_fixedOnSubmoduleAndQuotientSubgroup_of_nontrivial_normal_p_fixed_space
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] [Finite G] {V : Type*} [AddCommGroup V] [Module F V]
[Module.Finite F V]
(K : Subgroup G) [K.Normal] (ρ : Representation F G V)
(hfaithful : Function.Injective ρ) (hK : IsPGroup p K)
(hdim : Module.finrank F V = 2) (hK_ne_bot : K ≠ ⊥) :
commutator G ≤
fixedOnSubmoduleAndQuotientSubgroup
(Representation.invariants (ρ.comp K.subtype)) ρ := by
let W : Submodule F V := Representation.invariants (ρ.comp K.subtype)
have hV_ne_bot : (⊤ : Submodule F V) ≠ ⊥ := by
intro hbot
have htop : Module.finrank F (⊤ : Submodule F V) = 2 := by
simp [hdim]
have hzero : Module.finrank F (⊤ : Submodule F V) = 0 := by
rw [hbot, finrank_bot]
omega
have hW_ne_bot : W ≠ ⊥ := by
simpa [W] using hK.invariants_ne_bot (ρ.comp K.subtype) hV_ne_bot
have hW_ne_top : W ≠ ⊤ := by
simpa [W] using invariants_ne_top_of_faithful_subgroup_ne_bot K ρ hfaithful hK_ne_bot
have hW_invariant : ∀ g : G, W ≤ W.comap (ρ g) := by
intro g
exact Representation.le_comap_invariants ρ K g
exact commutator_le_fixedOnSubmoduleAndQuotientSubgroup_of_finrank_two
W ρ hW_invariant hdim hW_ne_bot hW_ne_top
/-- Nontrivial-normal-p-subgroup normality form for the same
`W = C_V(K)` used in BG Thm 2.6, q = p. -/
private theorem
fixedOnSubmoduleAndQuotientSubgroup_normal_of_nontrivial_normal_p_fixed_space
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] [Finite G] {V : Type*} [AddCommGroup V] [Module F V]
[Module.Finite F V]
(K : Subgroup G) [K.Normal] (ρ : Representation F G V)
(hfaithful : Function.Injective ρ) (hK : IsPGroup p K)
(hdim : Module.finrank F V = 2) (hK_ne_bot : K ≠ ⊥) :
(fixedOnSubmoduleAndQuotientSubgroup
(Representation.invariants (ρ.comp K.subtype)) ρ).Normal :=
Subgroup.Normal.of_commutator_le
(G := G)
(H := fixedOnSubmoduleAndQuotientSubgroup
(Representation.invariants (ρ.comp K.subtype)) ρ)
(commutator_le_fixedOnSubmoduleAndQuotientSubgroup_of_nontrivial_normal_p_fixed_space
K ρ hfaithful hK hdim hK_ne_bot)
/-- Nontrivial-normal-p-subgroup route to the Sylow containment `G' ≤ P`.
This is the current q = p endpoint: after constructing a nontrivial normal
p-subgroup `K`, its fixed space supplies the invariant submodule `W = C_V(K)`,
and the fixed-subquotient subgroup carries `G'` into every Sylow p-subgroup. -/
private theorem commutator_le_sylow_of_nontrivial_normal_p_fixed_space
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] [Finite G] [Finite (Sylow p G)]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(K : Subgroup G) [K.Normal] (ρ : Representation F G V)
(hfaithful : Function.Injective ρ) (hK : IsPGroup p K)
(hdim : Module.finrank F V = 2) (hK_ne_bot : K ≠ ⊥)
(P : Sylow p G) :
commutator G ≤ (P : Subgroup G) := by
let W : Submodule F V := Representation.invariants (ρ.comp K.subtype)
have hVpos : 0 < Module.finrank F V := by
rw [hdim]
norm_num
have : Nontrivial V := Module.nontrivial_of_finrank_pos (R := F) (M := V) hVpos
exact commutator_le_sylow_of_le_normal_pSubgroup
(fixedOnSubmoduleAndQuotientSubgroup W ρ)
(by
simpa [W] using
(fixedOnSubmoduleAndQuotientSubgroup_normal_of_nontrivial_normal_p_fixed_space
K ρ hfaithful hK hdim hK_ne_bot))
(fixedOnSubmoduleAndQuotientSubgroup_isPGroup_of_faithful
(p := p) W ρ hfaithful)
(by
simpa [W] using
(commutator_le_fixedOnSubmoduleAndQuotientSubgroup_of_nontrivial_normal_p_fixed_space
K ρ hfaithful hK hdim hK_ne_bot))
P
/-- q = p endpoint for BG Thm 2.6(b), conditional on the construction of a
nontrivial normal p-subgroup `K`.
The remaining theorem-level task is to construct the `K` supplied by the BG
argument; once it exists, this lemma gives exactly the Sylow conclusion. -/
theorem sylow_commutative_and_commutator_le_of_nontrivial_normal_p_fixed_space
{p : ℕ} [Fact p.Prime] {F : Type*} [Field F] [CharP F p]
{G : Type*} [Group G] [Finite G] [Finite (Sylow p G)]
{V : Type*} [AddCommGroup V] [Module F V] [Module.Finite F V]
(K : Subgroup G) [K.Normal] (ρ : Representation F G V)
(hfaithful : Function.Injective ρ) (hK : IsPGroup p K)
(hdim : Module.finrank F V = 2) (hK_ne_bot : K ≠ ⊥)
(P : Sylow p G) :
Std.Commutative (· * · : P → P → P) ∧
commutator G ≤ (P : Subgroup G) := by
constructor
· exact subgroup_commutative_of_nontrivial_normal_p_fixed_space
K (P : Subgroup G) ρ hfaithful hK P.2 hdim hK_ne_bot
· exact commutator_le_sylow_of_nontrivial_normal_p_fixed_space
K ρ hfaithful hK hdim hK_ne_bot P
end OddOrder.BG.Ch1.S02
Source
Yawara Ishida, Lean 4 formalization of the Feit–Thompson (odd order) theorem, github.com/yawara/odd-order @ 20a6fccbb63c (Apache-2.0), https://github.com/yawara/odd-order/blob/20a6fccbb63c0b138acda9f7fafd798c42f9441b/OddOrder/BG/Ch1_Preliminary/S02_FixedQuotientMachinery.lean