BG §15: -set and disjointness (TI-failure part)
DefinitionOddOrder_BG_Ch4_FamilyOfMaximal_S15_MF_PisetBetaDisjointfeit-thompsonfinite-groupsgroup-theoryishida-odd-order
A part of the TI-failure development of §15 of Bender–Glauberman, Local Analysis for the Odd Order Theorem, split from the TI-failure file. It consists of lemmas only: if and are disjoint then , the -core of the -complement in is cyclic with prime factors in , and forces .
The bundle consists of lemmas only.
The bundle also proves 27 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/Ch4_FamilyOfMaximal/S15_MF/PisetBetaDisjoint.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_Ch4_FamilyOfMaximal_S15_MF_Corollary155
namespace OddOrder.BG.Ch4.S15
open OddOrder.GroupTheory
open OddOrder.Isaacs
open OddOrder.BG.Ch3.S12
open OddOrder.BG.Ch4.S14
open scoped Pointwise
open scoped IsMulCommutative
open scoped commutatorElement
variable {G : Type*} [Group G]
/-- **BG Theorem 15.2(b), contrapositive form** (mmd L4185): if no prime divides `M_F` and lies
in `β(M)`, then `M_F = M_σ`. Theorem 15.2 shows that whenever `M_F ≠ M_σ`, the prime `q = |K*|`
satisfies `q ∈ π(M_F) ∩ β(M)`; the contrapositive gives the claim. A Hall `κ(M)`-subgroup `K`
needed to invoke 15.2 exists by solvability of `M` (Hall's theorem).
This is the `M_F = M_σ` endgame of Theorem 15.7(a): once the rank-theoretic core
`piSet_mf_inf_beta_disjoint_of_not_fittingIsTI` establishes `π(M_F) ∩ β(M) = ∅`, this lemma
delivers `M_F = M_σ` (equivalently, `M_σ` nilpotent). -/
theorem mf_eq_msigma_of_piSet_inf_beta_disjoint [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G)
(hdisj : ∀ q : ℕ, q ∈ S14.piSet (MF M) → q ∉ OddOrder.BG.Ch3.S10.beta M) :
MF M = OddOrder.BG.Ch3.S10.Msigma M := by
classical
have : Group.IsSolvable ↥M := hG.isSolvable_of_mem_maximalSubgroups hM
by_contra hne
-- A Hall `κ(M)`-subgroup `K` of `M` exists by Hall's theorem in the solvable group `↥M`.
obtain ⟨K', hK'⟩ := Ch03.hall_E_exists (G := ↥M) (S14.kappa M)
set K : Subgroup G := K'.map M.subtype with hKdef
have hKM : K ≤ M := hKdef ▸ Subgroup.map_subtype_le K'
have hKeq : K.subgroupOf M = K' :=
hKdef ▸ Subgroup.comap_map_eq_self_of_injective M.subtype_injective K'
have hK : Ch03.IsHallSubgroup (S14.kappa M) (K.subgroupOf M) := hKeq ▸ hK'
-- Theorem 15.2(b): for `M_F ≠ M_σ`, the prime `q = |K*|` lies in `π(M_F) ∩ β(M)`.
obtain ⟨_, _, _, _, _, _, _, _, _, hqπ, hqβ, _⟩ :=
(mf_ne_msigma_typeP1_structure hG hM hne hKM hK rfl).2
exact hdisj _ hqπ hqβ
/-- **BG Corollary 15.5(a)**: `O_{σ(M)'}(F(M))` is cyclic. Extracted from `fitting_decomposition`,
whose cyclic witness `Y` (a `τ₂(M) ⊆ σ(M)ᶜ`-group, normal in the nilpotent `F(M)` and complementing
`O_{σ(M)}(F(M)) = F(M_σ)`) equals `O_{σ(M)'}(F(M))` by modularity. -/
theorem opiCoreInG_sigmaCompl_fittingInAmbient_isCyclic [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G) :
IsCyclic ↥(opiCoreInG (OddOrder.BG.Ch3.S10.sigma M)ᶜ (fittingInAmbient M)) := by
classical
set σ := OddOrder.BG.Ch3.S10.sigma M with hσdef
set F := fittingInAmbient M with hFdef
obtain ⟨Y, hYcyc, hYtau2, hYleF, _, _, hF_eq, _hFmσ_Y_bot, hcomm, _, _, _⟩ :=
fitting_decomposition hG hM
-- `O_σ(F) = F(M_σ)`, so `F = O_σ(F) ⊔ Y`.
have hOσ : opiCoreInG σ F = fittingInAmbient (OddOrder.BG.Ch3.S10.Msigma M) :=
opiCoreInG_sigma_fittingInAmbient_eq_fittingInAmbient_Msigma
have hF_eq2 : F = opiCoreInG σ F ⊔ Y := by rw [hOσ]; exact hF_eq
-- `Y ⊴ F`: normalized by `Y` and by `F(M_σ)` (which centralizes it), and `F = F(M_σ) ⊔ Y`.
have hYnorm : (Y.subgroupOf F).Normal := by
refine (Subgroup.normal_subgroupOf_iff_le_normalizer hYleF).mpr ?_
rw [hF_eq]
refine sup_le ?_ Subgroup.le_normalizer
have h1 : fittingInAmbient (OddOrder.BG.Ch3.S10.Msigma M) ≤ Subgroup.centralizer (Y : Set G) :=
Subgroup.commutator_eq_bot_iff_le_centralizer.mp hcomm
exact h1.trans (Subgroup.centralizer_le_normalizer _)
-- `Y` is a `σᶜ`-group (`π(Y) ⊆ τ₂(M) ⊆ σᶜ`).
have hYpi : Subgroup.IsPiSubgroup σᶜ Y := fun r hr => tau2_subset_sigma_compl M (hYtau2 hr)
have hYle : Y ≤ opiCoreInG σᶜ F := le_opiCoreInG_of_normal_of_isPiSubgroup hYleF hYnorm hYpi
have hinf : opiCoreInG σ F ⊓ opiCoreInG σᶜ F = ⊥ :=
OddOrder.GroupTheory.inf_eq_bot_of_isPiSubgroup_of_isPiSubgroup_compl
(OddOrder.GroupTheory.isPiSubgroup_opiCoreInG σ F)
(OddOrder.GroupTheory.isPiSubgroup_opiCoreInG σᶜ F)
-- `O_σ(F) = F(M_σ)` centralizes `Y`, so it normalizes `Y`.
have hEnorm : opiCoreInG σ F ≤ Subgroup.normalizer (Y : Set G) := by
rw [hOσ]
exact (Subgroup.commutator_eq_bot_iff_le_centralizer.mp hcomm).trans
(Subgroup.centralizer_le_normalizer _)
-- `O_{σᶜ}(F) = Y` by the normality-aware Dedekind law, hence cyclic.
have hDF : opiCoreInG σᶜ F ≤ F := OddOrder.GroupTheory.opiCoreInG_le σᶜ F
have hDY : opiCoreInG σᶜ F = Y := by
have hkey := Subgroup.inf_sup_eq_of_le_normalizer_of_inf_eq_bot
(W := Y) (A := opiCoreInG σ F) (L := opiCoreInG σᶜ F) hYle hinf hEnorm
rw [← hkey, sup_comm, ← hF_eq2, inf_eq_right.mpr hDF]
rw [hDY]; exact hYcyc
/-- **BG Corollary 15.5(a), `τ₂`-membership form**: the `σ'`-part `O_{σ(M)'}(F(M))` of the Fitting
subgroup is a `τ₂(M)`-group. Companion to `opiCoreInG_sigmaCompl_fittingInAmbient_isCyclic`,
extracted from `fitting_decomposition`'s cyclic witness `Y` (which equals `O_{σ'}(F(M))` by the
modular argument). -/
theorem opiCoreInG_sigmaCompl_fittingInAmbient_primeFactors_subset_tau2 [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G) :
↑(Nat.card ↥(opiCoreInG (OddOrder.BG.Ch3.S10.sigma M)ᶜ (fittingInAmbient M))).primeFactors ⊆
tau2 M := by
classical
set σ := OddOrder.BG.Ch3.S10.sigma M with hσdef
set F := fittingInAmbient M with hFdef
obtain ⟨Y, _hYcyc, hYtau2, hYleF, _, _, hF_eq, _, hcomm, _, _, _⟩ := fitting_decomposition hG hM
have hOσ : opiCoreInG σ F = fittingInAmbient (OddOrder.BG.Ch3.S10.Msigma M) :=
opiCoreInG_sigma_fittingInAmbient_eq_fittingInAmbient_Msigma
have hF_eq2 : F = opiCoreInG σ F ⊔ Y := by rw [hOσ]; exact hF_eq
have hYnorm : (Y.subgroupOf F).Normal := by
refine (Subgroup.normal_subgroupOf_iff_le_normalizer hYleF).mpr ?_
rw [hF_eq]
refine sup_le ?_ Subgroup.le_normalizer
have h1 : fittingInAmbient (OddOrder.BG.Ch3.S10.Msigma M) ≤ Subgroup.centralizer (Y : Set G) :=
Subgroup.commutator_eq_bot_iff_le_centralizer.mp hcomm
exact h1.trans (Subgroup.centralizer_le_normalizer _)
have hYpi : Subgroup.IsPiSubgroup σᶜ Y := fun r hr => tau2_subset_sigma_compl M (hYtau2 hr)
have hYle : Y ≤ opiCoreInG σᶜ F := le_opiCoreInG_of_normal_of_isPiSubgroup hYleF hYnorm hYpi
have hinf : opiCoreInG σ F ⊓ opiCoreInG σᶜ F = ⊥ :=
OddOrder.GroupTheory.inf_eq_bot_of_isPiSubgroup_of_isPiSubgroup_compl
(OddOrder.GroupTheory.isPiSubgroup_opiCoreInG σ F)
(OddOrder.GroupTheory.isPiSubgroup_opiCoreInG σᶜ F)
have hEnorm : opiCoreInG σ F ≤ Subgroup.normalizer (Y : Set G) := by
rw [hOσ]
exact (Subgroup.commutator_eq_bot_iff_le_centralizer.mp hcomm).trans
(Subgroup.centralizer_le_normalizer _)
have hDF : opiCoreInG σᶜ F ≤ F := OddOrder.GroupTheory.opiCoreInG_le σᶜ F
have hDY : opiCoreInG σᶜ F = Y := by
have hkey := Subgroup.inf_sup_eq_of_le_normalizer_of_inf_eq_bot
(W := Y) (A := opiCoreInG σ F) (L := opiCoreInG σᶜ F) hYle hinf hEnorm
rw [← hkey, sup_comm, ← hF_eq2, inf_eq_right.mpr hDF]
rw [hDY]; exact hYtau2
/-- **BG Theorem 15.7(d), the `τ₃(M) = ∅` prime-set core**: if `M' ≤ F(M)` then `τ₃(M) = ∅`.
`F(M) = F(M_σ) × O_{σ'}(F(M))` (`fitting_decomposition`, an internal direct product) with `F(M_σ)`
a `σ(M)`-group (`= O_σ(F(M))`) and `O_{σ'}(F(M))` a `τ₂(M)`-group, so every prime of `M' ≤ F(M)` is
in `σ(M) ∪ τ₂(M)`. A `τ₃(M)`-prime is `∉ σ(M)` (hence `∈ τ₂(M)`, `r_p = 2`) yet lies in `π(M')`
with `r_p = 1` — contradiction.
This is the `defE`/`E3_1` prime-set half of Corollary 15.9's cyclic Frobenius complement (Coq
`nonTI_Fitting_structure` part (d)): once `τ₃(M) = ∅` the `τ₃`-Hall `E₃` of any `σ(M)'`-complement
is
trivial (`E3_eq_bot_of_tau3_eq_empty`), so `E = E₁E₂` and — with `τ₂(M) = ∅` (Theorem 15.8) —
`E = E₁`
is cyclic. The remaining upstream gate is the `¬FittingIsTI`-specific `M' ≤ F(M)` (`M'` nilpotency;
issue 2037). -/
theorem tau3_eq_empty_of_derivedInG_le_fittingInAmbient [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G)
(hM'F : derivedInG M ≤ fittingInAmbient M) : tau3 M = ∅ := by
classical
obtain ⟨Y, -, hYtau2, -, -, -, hF6, hF7, hF8, -, -, -⟩ := fitting_decomposition hG hM
set Fσ : Subgroup G := fittingInAmbient (OddOrder.BG.Ch3.S10.Msigma M) with hFσdef
-- `F(M) = F(M_σ) × Y` ⟹ `|F(M)| = |F(M_σ)| · |Y|`.
have hFσnormY : Fσ ≤ Subgroup.normalizer (Y : Set G) :=
(Subgroup.commutator_eq_bot_iff_le_centralizer.mp hF8).trans
(Subgroup.centralizer_le_normalizer _)
have hcard : Nat.card ↥(fittingInAmbient M) = Nat.card ↥Fσ * Nat.card ↥Y := by
rw [hF6]; exact card_sup_eq_mul_of_le_normalizer_of_disjoint hFσnormY hF7
-- `π(F(M_σ)) ⊆ σ(M)` (`F(M_σ) = O_σ(F(M))`, a `σ`-core).
have hFσσ : ∀ p ∈ (Nat.card ↥Fσ).primeFactors, p ∈ OddOrder.BG.Ch3.S10.sigma M := by
intro p hp
rw [hFσdef, ← opiCoreInG_sigma_fittingInAmbient_eq_fittingInAmbient_Msigma] at hp
exact OddOrder.GroupTheory.isPiSubgroup_opiCoreInG (OddOrder.BG.Ch3.S10.sigma M)
(fittingInAmbient M) p hp
-- No `τ₃`-prime survives.
ext p
simp only [Set.mem_empty_iff_false, iff_false]
rw [mem_tau3_iff]
rintro ⟨hpσ, hpM', hp1⟩
have hpprime : p.Prime := Nat.prime_of_mem_primeFactors hpM'
have hpF : p ∈ (Nat.card ↥(fittingInAmbient M)).primeFactors :=
Nat.mem_primeFactors.mpr ⟨hpprime,
(Nat.dvd_of_mem_primeFactors hpM').trans (Subgroup.card_dvd_of_le hM'F), Nat.card_pos.ne'⟩
rw [hcard, Nat.primeFactors_mul Nat.card_pos.ne' Nat.card_pos.ne', Finset.mem_union] at hpF
rcases hpF with h | h
· exact hpσ (hFσσ p h)
· have hp2 : pRank ↥M p = 2 := ((mem_tau2_iff M p).mp (hYtau2 (Finset.mem_coe.mpr h))).2
omega
/-- **`E₃ = ⊥` from `τ₃(M) = ∅`**: the `τ₃(M)`-Hall factor `E₃` of any `σ(M)'`-complement `E`-setup
is trivial when `M` has no `τ₃`-primes. A `{τ₃}`-Hall subgroup of the empty prime set has order
coprime to every prime, hence order `1`. Companion to
`tau3_eq_empty_of_derivedInG_le_fittingInAmbient` (BG Theorem 15.7(d), `E3_1`). -/
theorem E3_eq_bot_of_tau3_eq_empty [Finite G] {M E E₁ E₂ E₃ : Subgroup G}
(hsetup : OddOrder.BG.Ch3.S12.SubgroupESetup M E E₁ E₂ E₃) (htau3 : tau3 M = ∅) :
E₃ = ⊥ := by
classical
rw [← Subgroup.card_eq_one]
by_contra hne
obtain ⟨p, hp, hpdvd⟩ := Nat.exists_prime_and_dvd hne
have hpsub : p ∈ (Nat.card ↥(E₃.subgroupOf E)).primeFactors := by
rw [Nat.card_congr (Subgroup.subgroupOfEquivOfLe hsetup.E₃_le).toEquiv]
exact Nat.mem_primeFactors.mpr ⟨hp, hpdvd, Nat.card_pos.ne'⟩
have hmem : p ∈ tau3 M := hsetup.E₃_hall.1 p hpsub
rw [htau3] at hmem
exact hmem
/-- **`E₂ = ⊥` from `τ₂(M) = ∅`**: the `τ₂(M)`-Hall factor `E₂` of any
`σ(M)'`-complement `E`-setup is
trivial when `M` has no `τ₂`-*primes* (Theorem 15.8's `τ₂(M) = ∅` for the Corollary 15.9 escape, in
the `∀ p, p.Prime → p ∉ τ₂(M)` form output by `tau2_transfer_constraint`). `τ₂`-analogue of
`E3_eq_bot_of_tau3_eq_empty`. -/
theorem E2_eq_bot_of_tau2_eq_empty [Finite G] {M E E₁ E₂ E₃ : Subgroup G}
(hsetup : OddOrder.BG.Ch3.S12.SubgroupESetup M E E₁ E₂ E₃)
(htau2 : ∀ p : ℕ, p.Prime → p ∉ tau2 M) :
E₂ = ⊥ := by
classical
rw [← Subgroup.card_eq_one]
by_contra hne
obtain ⟨p, hp, hpdvd⟩ := Nat.exists_prime_and_dvd hne
have hpsub : p ∈ (Nat.card ↥(E₂.subgroupOf E)).primeFactors := by
rw [Nat.card_congr (Subgroup.subgroupOfEquivOfLe hsetup.E₂_le).toEquiv]
exact Nat.mem_primeFactors.mpr ⟨hp, hpdvd, Nat.card_pos.ne'⟩
exact htau2 p hp (hsetup.E₂_hall.1 p hpsub)
/-- **Type-`P₂` `M_F`-internal Fitting decomposition** (BG Corollary 15.5; the `M' = M_F × U`
form feeding Proposition 16.1's forward bridges). For a type-`P₂` maximal subgroup `M` with
`κ`-Hall `K` and `(κ ∪ σ)'`-Hall `U`, the derived subgroup `M'` is the internal direct product of
the Fitting kernel `M_F` and `U` (the complement `hDcompl`), the Fitting subgroup is
`F(M) = M_F ⊔ (U ⊓ C_M(M_F))` (`hFiteq`), and `M'' ≤ F(M)` (`hSDfit`).
This discharges the three genuinely-deep `M_F`-internal residuals (`hDcompl`/`hSDfit`/`hFiteq`)
of `typePData_of_isTypeP_of_inputs`, the shared linchpin of the `hP2II`/`hP1neIIIIV`/`hP1eqV`
forward bridges of `proposition_type_classification`.
Proof: `M_F = M_σ` (`M_σ` nilpotent for type `P₂`) and `M' = U ⊔ M_σ` with `U ⊓ M_σ = ⊥`
(Lemma 15.1(b)) give the complement. The Fitting identity routes through `Y := O_{σ'}(F(M))`,
the `τ₂`-part of `F(M)` (Corollary 15.5(a)): `Y` is a `(κ ∪ σ)'`-group normal in `M`
(`π(Y) ⊆ τ₂`, and `τ₂ ∩ κ = ∅` by rank), hence `Y ≤ U`, and `Y` centralizes `M_σ = M_F`, so
`F(M) = M_F ⊔ Y ≤ M_F ⊔ (U ⊓ C_M(M_F)) ≤ F(M)`. -/
theorem typeP2_mf_internal_fitting_decomposition [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M K U : Subgroup G}
(hM : M ∈ maximalSubgroups G) (hP2 : S14.IsTypeP2 M) (hKM : K ≤ M) (hUM : U ≤ M)
(hKne : K ≠ ⊥)
(hK : Ch03.IsHallSubgroup (S14.kappa M) (K.subgroupOf M))
(hU : Ch03.IsHallSubgroup ((S14.kappa M ∪ OddOrder.BG.Ch3.S10.sigma M)ᶜ) (U.subgroupOf M)) :
Subgroup.IsComplement' ((maxNilpotentNormalHall M).subgroupOf (derivedInG M))
(U.subgroupOf (derivedInG M)) ∧
secondDerivedInAmbient M ≤ maxNilpotentNormalHall M ⊔
(U ⊓ Subgroup.centralizer (maxNilpotentNormalHall M : Set G)) ∧
(OddOrder.Isaacs.Ch01.fitting ↥M).map M.subtype = maxNilpotentNormalHall M ⊔
(U ⊓ Subgroup.centralizer (maxNilpotentNormalHall M : Set G)) := by
classical
set σ := OddOrder.BG.Ch3.S10.sigma M with hσdef
set Mσ := OddOrder.BG.Ch3.S10.Msigma M with hMσdef
set F := fittingInAmbient M with hFdef
have hFnil : Group.IsNilpotent ↥F := OddOrder.BG.Ch2.S08.fittingInG_isNilpotent M
-- `M_F = M_σ`: `M_σ` is nilpotent for type `P₂`.
have hMσnil : Group.IsNilpotent ↥Mσ := S14.msigma_isNilpotent_of_isTypeP2 hG hM hP2
have hMFMσ : maxNilpotentNormalHall M = Mσ :=
(maxNilpotentNormalHall_eq_Msigma_iff_isNilpotent hG hM).mpr hMσnil
-- Lemma 15.1(b): `M' = U ⊔ M_σ`.
obtain ⟨hM'eq, _hUab⟩ := typeP_hall_derived_eq_and_abelian hG hM hKM hUM hKne hK hU
have hMσ_le_M' : Mσ ≤ derivedInG M := OddOrder.BG.Ch3.S10.Msigma_le_derived hG hM
have hM'_le_M : derivedInG M ≤ M := Subgroup.map_subtype_le _
have hMσ_le_M : Mσ ≤ M := OddOrder.BG.Ch3.S10.Msigma_le M
have hU_le_M' : U ≤ derivedInG M := by rw [hM'eq]; exact le_sup_left
-- `U ⊓ M_σ = ⊥` (coprime Hall orders).
have hUMσ_bot : U ⊓ Mσ = ⊥ := by
have hMσHall : Ch03.IsHallSubgroup σ (Mσ.subgroupOf M) :=
OddOrder.BG.Ch3.S10.Msigma_subgroupOf_isHall hG hM
have hcop : Nat.Coprime (Nat.card ↥U) (Nat.card ↥Mσ) := by
refine Ch03.Nat.coprime_of_isPiGroup_of_isPiGroup_compl
(π := (S14.kappa M ∪ σ)ᶜ) Nat.card_pos.ne' Nat.card_pos.ne' ?_ ?_
· intro p hp
exact hU.1 p (by rwa [Nat.card_congr (Subgroup.subgroupOfEquivOfLe hUM).toEquiv])
· intro p hp hpcompl
have hpMσ : p ∈ (Nat.card ↥(Mσ.subgroupOf M)).primeFactors := by
rwa [Nat.card_congr (Subgroup.subgroupOfEquivOfLe hMσ_le_M).toEquiv]
exact hpcompl (Or.inr (hMσHall.1 p hpMσ))
exact (Subgroup.disjoint_of_coprime_natCard hcop).eq_bot
-- `Y := O_{σ'}(F(M))`, the `τ₂`-part of `F(M)`.
set Y := opiCoreInG σᶜ F with hYdef
have hY_le_M : Y ≤ M :=
(OddOrder.GroupTheory.opiCoreInG_le σᶜ F).trans (OddOrder.BG.Ch2.S08.fittingInG_le M)
have hOσMσ : opiCoreInG σ F = Mσ := by
rw [opiCoreInG_sigma_fittingInAmbient_eq_fittingInAmbient_Msigma, ← hMσdef,
fittingInAmbient_eq_self_of_isNilpotent]
-- `Y` centralizes `M_σ`.
have hY_le_cent : Y ≤ Subgroup.centralizer (Mσ : Set G) := by
have hcomm : ⁅opiCoreInG σ F, Y⁆ = ⊥ := by
rw [hYdef]; exact OddOrder.BG.Ch2.S08.opiCoreInG_commutator_compl_eq_bot σ F
rw [hOσMσ, Subgroup.commutator_comm] at hcomm
exact Subgroup.commutator_eq_bot_iff_le_centralizer.mp hcomm
-- `Y ≤ U`: `Y` is a `(κ∪σ)'`-group normal in `M`, and `U` is the `(κ∪σ)'`-Hall.
have hY_le_U : Y ≤ U := by
have hY_pi : Ch03.Subgroup.IsPiGroup ((S14.kappa M ∪ σ)ᶜ) (Y.subgroupOf M) := by
intro p hp
have hpY : p ∈ (Nat.card ↥Y).primeFactors := by
rwa [Nat.card_congr (Subgroup.subgroupOfEquivOfLe hY_le_M).toEquiv] at hp
have hpσ : p ∈ σᶜ := OddOrder.GroupTheory.isPiSubgroup_opiCoreInG σᶜ F p hpY
have hpτ2 : p ∈ tau2 M :=
opiCoreInG_sigmaCompl_fittingInAmbient_primeFactors_subset_tau2 hG hM hpY
rw [Set.mem_compl_iff, Set.mem_union]
rintro (hpκ | hpσ')
· have hr1 : pRank ↥M p = 1 := by
rcases S14.kappa_subset_tau1_union_tau3 hpκ with h | h
· exact ((mem_tau1_iff M p).mp h).2.2
· exact ((mem_tau3_iff M p).mp h).2.2
have hr2 : pRank ↥M p = 2 := ((mem_tau2_iff M p).mp hpτ2).2
rw [hr1] at hr2; exact absurd hr2 (by norm_num)
· exact hpσ hpσ'
have hYnorm_M : (Y.subgroupOf M).Normal := by
rw [hYdef]; exact OddOrder.BG.Ch2.S08.opiCoreInG_fittingInG_subgroupOf_normal σᶜ M
have hsub_le : Y.subgroupOf M ≤ U.subgroupOf M :=
Ch03.Subgroup.IsPiGroup.normal_le_hall hY_pi hU
have hmap := Subgroup.map_mono (f := M.subtype) hsub_le
rwa [Subgroup.map_subgroupOf_eq_of_le hY_le_M, Subgroup.map_subgroupOf_eq_of_le hUM] at hmap
-- `F(M) = M_F ⊔ Y`.
have hF_split : F = maxNilpotentNormalHall M ⊔ Y := by
have hsplit : opiCoreInG σ F ⊔ Y = F := by
rw [hYdef]; exact opiCoreInG_sup_compl_eq_of_isNilpotent σ
rw [← hsplit, hOσMσ, hMFMσ]
-- `M_F ≤ F(M)`.
have hMF_le_F : maxNilpotentNormalHall M ≤ F := by
have : Group.IsNilpotent ↥(maxNilpotentNormalHall M) := maxNilpotentNormalHall_isNilpotent M
exact le_fittingInAmbient_of_subgroupOf_normal_of_isNilpotent
(maxNilpotentNormalHall_le M) (maxNilpotentNormalHall_subgroupOf_normal M)
-- `F(M) = M_F ⊔ (U ⊓ C_M(M_F))`.
have hFiteq : F = maxNilpotentNormalHall M ⊔
(U ⊓ Subgroup.centralizer (maxNilpotentNormalHall M : Set G)) := by
apply le_antisymm
· rw [hF_split]
refine sup_le_sup_left (le_inf hY_le_U ?_) _
rw [hMFMσ]; exact hY_le_cent
· refine sup_le hMF_le_F ?_
have hFcent : F = (Subgroup.centralizer (maxNilpotentNormalHall M : Set G) ⊓ M) ⊔
maxNilpotentNormalHall M := by
obtain ⟨-, -, -, -, -, h, -, -, -, -, -, -⟩ := fitting_decomposition hG hM
exact h
rw [hFcent]
exact (le_inf inf_le_right (inf_le_left.trans hUM)).trans le_sup_left
refine ⟨?_, ?_, hFiteq⟩
· -- `hDcompl`: `M_F = M_σ` complements `U` in `M'`.
rw [hMFMσ]
have hMσnorm' : ((Mσ).subgroupOf (derivedInG M)).Normal :=
(Subgroup.normal_subgroupOf_iff_le_normalizer hMσ_le_M').mpr
(hM'_le_M.trans (OddOrder.GroupTheory.le_normalizer_opiCoreInG σ M))
refine Subgroup.isComplement'_of_disjoint_and_mul_eq_univ ?_ ?_
· rw [Subgroup.disjoint_def]
intro x hxMσ hxU
rw [Subgroup.mem_subgroupOf] at hxMσ hxU
have hx : (x : G) ∈ U ⊓ Mσ := ⟨hxU, hxMσ⟩
rw [hUMσ_bot, Subgroup.mem_bot] at hx
exact Subtype.ext hx
· rw [← Subgroup.normal_mul, ← Subgroup.subgroupOf_sup hMσ_le_M' hU_le_M',
sup_comm, ← hM'eq, Subgroup.subgroupOf_self, Subgroup.coe_top]
· -- `hSDfit`: `M'' ≤ F(M) = M_F ⊔ (U ⊓ C_M(M_F))`.
have hM''F : secondDerivedInAmbient M ≤ F := by
obtain ⟨-, -, -, -, h, -, -, -, -, -, -, -⟩ := fitting_decomposition hG hM
exact h
exact hM''F.trans (le_of_eq hFiteq)
/-- In a finite cyclic group, a subgroup is determined by its cardinality (every subgroup is the
kernel of `x ↦ x ^ |A|`, which depends only on `|A|`). -/
theorem eq_of_card_eq_of_isCyclic {C : Type*} [Group C] [Finite C] [IsCyclic C]
{H K : Subgroup C} (h : Nat.card ↥H = Nat.card ↥K) : H = K := by
let : CommGroup C := IsCyclic.commGroup
have key : ∀ A : Subgroup C, A = (powMonoidHom (Nat.card ↥A) : C →* C).ker := by
intro A
have hcard : Nat.card ↥((powMonoidHom (Nat.card ↥A) : C →* C).ker) = Nat.card ↥A := by
rw [IsCyclic.card_powMonoidHom_ker, Nat.gcd_eq_right (Subgroup.card_subgroup_dvd_card A)]
refine Subgroup.eq_of_le_of_card_ge (fun a ha => ?_) (le_of_eq hcard)
rw [MonoidHom.mem_ker, powMonoidHom_apply]
have h1 : (⟨a, ha⟩ : ↥A) ^ Nat.card ↥A = 1 := pow_card_eq_one'
simpa only [SubmonoidClass.coe_pow, OneMemClass.coe_one] using congrArg Subtype.val h1
rw [key H, key K, h]
/-- Two subgroups of `G` of equal finite order, both contained in a cyclic subgroup `C`, are
equal. -/
theorem eq_of_le_isCyclic_of_card_eq [Finite G] {C H K : Subgroup G} [IsCyclic ↥C]
(hHC : H ≤ C) (hKC : K ≤ C) (h : Nat.card ↥H = Nat.card ↥K) : H = K := by
have hcardH : Nat.card ↥(H.subgroupOf C) = Nat.card ↥H :=
Nat.card_congr (Subgroup.subgroupOfEquivOfLe hHC).toEquiv
have hcardK : Nat.card ↥(K.subgroupOf C) = Nat.card ↥K :=
Nat.card_congr (Subgroup.subgroupOfEquivOfLe hKC).toEquiv
have hsub : H.subgroupOf C = K.subgroupOf C :=
eq_of_card_eq_of_isCyclic (by rw [hcardH, hcardK, h])
rw [← Subgroup.map_subgroupOf_eq_of_le hHC, ← Subgroup.map_subgroupOf_eq_of_le hKC, hsub]
/-- If `K` normalizes a cyclic subgroup `C` and `X ≤ C` is finite, then `K` normalizes `X`:
subgroups of a cyclic group are determined by their order, so the order-preserving `C`-conjugation
by elements of `K` fixes `X`. -/
theorem le_normalizer_of_le_isCyclic_normalized [Finite G] {C X K : Subgroup G} [IsCyclic ↥C]
(hXC : X ≤ C) (hKC : K ≤ Subgroup.normalizer (C : Set G)) :
K ≤ Subgroup.normalizer (X : Set G) := by
intro m hm
-- `m` conjugates `C` to itself.
have hmC : MulAut.conj m • C = C := by
ext x
rw [Subgroup.mem_pointwise_smul_iff_inv_smul_mem,
show ((MulAut.conj m)⁻¹ • x : G) = m⁻¹ * x * m by
simp [MulAut.smul_def, ← map_inv, MulAut.conj_apply]]
exact ((Subgroup.mem_normalizer_iff''.mp (hKC hm)) x).symm
-- `conj m • X ≤ C` and has the same order, so equals `X`.
have hconjle : MulAut.conj m • X ≤ C :=
hmC ▸ (Subgroup.pointwise_smul_le_pointwise_smul_iff.mpr hXC)
have hcard : Nat.card ↥(MulAut.conj m • X) = Nat.card ↥X :=
(Nat.card_congr (Subgroup.equivSMul (MulAut.conj m) X).toEquiv).symm
have hXeq : MulAut.conj m • X = X := eq_of_le_isCyclic_of_card_eq hconjle hXC hcard
rw [Subgroup.mem_normalizer_iff'']
intro h
have hiff : h ∈ MulAut.conj m • X ↔ m⁻¹ * h * m ∈ X := by
rw [Subgroup.mem_pointwise_smul_iff_inv_smul_mem,
show ((MulAut.conj m)⁻¹ • h : G) = m⁻¹ * h * m by
simp [MulAut.smul_def, ← map_inv, MulAut.conj_apply]]
rwa [hXeq] at hiff
/-- **Setup for BG Theorem 15.7(a)**: `¬FittingIsTI M` produces an element `g ∉ M` and a nontrivial
intersection `F(M) ⊓ F(M)^g`. Unfolding `¬IsTISubset (F(M)^#) (N_G(F(M)))`:
there is `g ∉ N_G(F(M))`
and `a ∈ F(M)^#` with `gag⁻¹ ∈ F(M)^#`; then `gag⁻¹ ∈ F(M) ⊓ (conj g • F(M))` is nontrivial, and
`g ∉ M` because `F(M) ⊴ M` forces `M ≤ N_G(F(M))`. -/
theorem exists_notMem_inf_conj_fitting_ne_bot_of_not_fittingIsTI {M : Subgroup G}
(hnotTI : ¬ FittingIsTI M) :
∃ g : G, g ∉ M ∧
(fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M : Subgroup G) ≠ ⊥ := by
rw [FittingIsTI] at hnotTI
simp only [OddOrder.GroupTheory.IsTISubset] at hnotTI
push Not at hnotTI
obtain ⟨g, ⟨a, haA, hgaA⟩, hgN⟩ := hnotTI
simp only [fittingSharp, sharpSubgroup, Set.mem_sdiff, SetLike.mem_coe,
Set.mem_singleton_iff] at haA hgaA
obtain ⟨haF, _ha1⟩ := haA
obtain ⟨hgaF, hga1⟩ := hgaA
-- `g ∉ M`: `F(M) ⊴ M` ⟹ `M ≤ N_G(F(M))`, but `g ∉ N_G(F(M))`.
have hMN : M ≤ Subgroup.normalizer ((fittingInAmbient M : Subgroup G) : Set G) :=
(Subgroup.normal_subgroupOf_iff_le_normalizer (OddOrder.BG.Ch2.S08.fittingInG_le M)).mp
(OddOrder.BG.Ch2.S08.fittingInG_subgroupOf_normal M)
refine ⟨g, fun h => hgN (hMN h), ?_⟩
-- `gag⁻¹` is a nontrivial element of `F(M) ⊓ conj g • F(M)`.
intro hbot
have hmem : g * a * g⁻¹ ∈
(fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M : Subgroup G) := by
refine Subgroup.mem_inf.mpr ⟨hgaF, ?_⟩
rw [Subgroup.mem_pointwise_smul_iff_inv_smul_mem]
simpa [MulAut.smul_def, MulAut.conj_apply, mul_assoc] using haF
rw [hbot, Subgroup.mem_bot] at hmem
exact hga1 hmem
/-- **Step 3 of BG Theorem 15.7(a)**: a prime `p` dividing the TI-failure intersection
`F(M) ⊓ F(M)^g` (with `g ∉ M`) lies in `σ(M)`.
If not, `O_{σ(M)'}(F(M))` is cyclic (`opiCoreInG_sigmaCompl_fittingInAmbient_isCyclic`), so any
order-`p` subgroup `X₁ ≤ F(M) ⊓ F(M)^g` (it is a `p`-subgroup of `F(M)`, hence
`≤ O_p(F(M)) ≤ O_{σ'}(F(M))`) is the unique one, hence normalized by both `M` and `M^g` (each
normalizes the relevant cyclic `O_{σ'}` and `X₁ = (X₁)` is order-preserved). Then
`normalizer_eq_of_normal_of_mem_maximal` gives `N_G(X₁) = M`, but also
`M = N_G(g⁻¹·X₁·g) = g⁻¹·M·g`,
forcing `g ∈ M` — contradiction. -/
theorem mem_sigma_of_prime_dvd_card_inf_conj_fitting [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G)
{g : G} (hgM : g ∉ M) {p : ℕ} (hp : p.Prime)
(hpdvd : p ∣ Nat.card ↥(fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M)) :
p ∈ OddOrder.BG.Ch3.S10.sigma M := by
classical
have : Fact p.Prime := ⟨hp⟩
by_contra hpσ
set F := fittingInAmbient M with hFdef
set σ := OddOrder.BG.Ch3.S10.sigma M with hσdef
have hFleM : F ≤ M := OddOrder.BG.Ch2.S08.fittingInG_le M
-- `O_{σᶜ}(F)` is cyclic and normalized by `M`.
have hcyc : IsCyclic ↥(opiCoreInG σᶜ F) := opiCoreInG_sigmaCompl_fittingInAmbient_isCyclic hG hM
have hMNOσc : M ≤ Subgroup.normalizer ((opiCoreInG σᶜ F : Subgroup G) : Set G) :=
(Subgroup.normal_subgroupOf_iff_le_normalizer
((OddOrder.GroupTheory.opiCoreInG_le σᶜ F).trans hFleM)).mp
(OddOrder.BG.Ch2.S08.opiCoreInG_fittingInG_subgroupOf_normal σᶜ M)
have hpσc : ({p} : Set ℕ) ⊆ σᶜ := by
intro q hq; rw [Set.mem_singleton_iff] at hq; rw [hq]; exact hpσ
-- Generic helper: a `p`-subgroup of `F` is normalized by `M` (via `X₁ ≤ O_{σᶜ}(F)` cyclic).
have hpnorm : ∀ Z : Subgroup G, Z ≤ F → IsPGroup p ↥Z →
M ≤ Subgroup.normalizer (Z : Set G) := by
intro Z hZF hZp
have hZOp : Z ≤ opiCoreInG ({p} : Set ℕ) F :=
OddOrder.BG.Ch2.S08.le_opiCoreInG_singleton_of_isPGroup_of_le_nilpotent
(OddOrder.BG.Ch2.S08.fittingInG_isNilpotent M) hZF hZp
have hZOσc : Z ≤ opiCoreInG σᶜ F :=
hZOp.trans (Subgroup.map_mono (OddOrder.Isaacs.Ch03.oPiCore_mono hpσc _))
exact le_normalizer_of_le_isCyclic_normalized hZOσc hMNOσc
-- An order-`p` subgroup `X₁` of `X = F ⊓ F^g`.
obtain ⟨x, hxord⟩ := exists_prime_orderOf_dvd_card'
(G := ↥(F ⊓ MulAut.conj g • F)) p hpdvd
set X₁ : Subgroup G := (Subgroup.zpowers x).map (F ⊓ MulAut.conj g • F).subtype with hX₁def
have hX₁leX : X₁ ≤ F ⊓ MulAut.conj g • F := hX₁def ▸ Subgroup.map_subtype_le _
have hX₁card : Nat.card ↥X₁ = p := by
rw [hX₁def, Subgroup.card_map_of_injective (F ⊓ MulAut.conj g • F).subtype_injective,
Nat.card_zpowers, hxord]
have hX₁ne : X₁ ≠ ⊥ := by
intro h; rw [h, Subgroup.card_bot] at hX₁card; exact hp.one_lt.ne' hX₁card.symm
have hX₁pg : IsPGroup p ↥X₁ := IsPGroup.of_card (n := 1) (by rw [hX₁card, pow_one])
have hX₁F : X₁ ≤ F := hX₁leX.trans inf_le_left
have hX₁cF : X₁ ≤ MulAut.conj g • F := hX₁leX.trans inf_le_right
-- `X₁ ⊴ M`, so `N_G(X₁) = M`.
have hMNX₁ : M ≤ Subgroup.normalizer (X₁ : Set G) := hpnorm X₁ hX₁F hX₁pg
have hNX₁ : Subgroup.normalizer (X₁ : Set G) = M :=
OddOrder.BG.Ch2.S08.normalizer_eq_of_normal_of_mem_maximal hG hM
((Subgroup.normal_subgroupOf_iff_le_normalizer (hX₁F.trans hFleM)).mpr hMNX₁)
hX₁ne (hX₁F.trans hFleM)
-- `g⁻¹·X₁·g ≤ F` is also a `p`-group, so `M ≤ N(g⁻¹·X₁·g) = g⁻¹·N(X₁)·g = g⁻¹·M·g`.
set X₁' : Subgroup G := MulAut.conj g⁻¹ • X₁ with hX₁'def
have hX₁'card : Nat.card ↥X₁' = p := by
rw [hX₁'def, ← Nat.card_congr (Subgroup.equivSMul (MulAut.conj g⁻¹) X₁).toEquiv, hX₁card]
have hX₁'F : X₁' ≤ F := by
have hle : X₁' ≤ MulAut.conj g⁻¹ • (MulAut.conj g • F) :=
hX₁'def ▸ Subgroup.pointwise_smul_le_pointwise_smul_iff.mpr hX₁cF
rwa [← mul_smul, ← map_mul, inv_mul_cancel, map_one, one_smul] at hle
have hX₁'pg : IsPGroup p ↥X₁' := IsPGroup.of_card (n := 1) (by rw [hX₁'card, pow_one])
have hMNX₁' : M ≤ Subgroup.normalizer (X₁' : Set G) := hpnorm X₁' hX₁'F hX₁'pg
-- `N(X₁') = g⁻¹ • N(X₁) = g⁻¹ • M`.
have hNX₁'eq : Subgroup.normalizer (X₁' : Set G) = MulAut.conj g⁻¹ • M := by
rw [hX₁'def, ← hNX₁]
exact (Subgroup.map_normalizer_eq_of_bijective X₁ (MulAut.conj g⁻¹).bijective).symm
-- So `M ≤ g⁻¹ • M`, equal cards ⟹ `M = g⁻¹ • M`, i.e. `g ∈ N(M) = M`.
rw [hNX₁'eq] at hMNX₁'
have hcardM : Nat.card ↥(MulAut.conj g⁻¹ • M) = Nat.card ↥M :=
(Nat.card_congr (Subgroup.equivSMul (MulAut.conj g⁻¹) M).toEquiv).symm
have hMeq : MulAut.conj g⁻¹ • M = M :=
(Subgroup.eq_of_le_of_card_ge hMNX₁' (le_of_eq hcardM)).symm
-- `g⁻¹·M·g = M ⟹ g⁻¹ ∈ N_G(M) ≤ M ⟹ g ∈ M`, contradicting `g ∉ M`.
have hg_inv_N : g⁻¹ ∈ Subgroup.normalizer (M : Set G) := by
rw [Subgroup.mem_normalizer_iff'']
intro h
have hiff : h ∈ MulAut.conj g⁻¹ • M ↔ g * h * g⁻¹ ∈ M := by
rw [Subgroup.mem_pointwise_smul_iff_inv_smul_mem,
show ((MulAut.conj g⁻¹)⁻¹ • h : G) = g * h * g⁻¹ by
simp [MulAut.smul_def, ← map_inv, MulAut.conj_apply]]
rw [hMeq] at hiff
rw [inv_inv]; exact hiff
have hg_inv_M : g⁻¹ ∈ M :=
OddOrder.BG.Ch3.S10.maximal_normalizer_le_self hG hM hg_inv_N
exact hgM (by simpa using inv_mem hg_inv_M)
/-- **Rank-3 lower bound on `M_F` for `β`-primes** (the `≥ 3` side of BG Theorem 15.7(a)'s rank
dichotomy): if a prime `r` divides `M_F` and lies in `β(M)`, then `r_r(M_F) ≥ 3`.
`M_F` is a Hall subgroup of `M` (`maxNilpotentNormalHall_isHall`), so `r ∈ π(M_F)` gives
`r ∤ [M : M_F]` and hence `r_r(M_F) = r_r(M)` (`pRank_eq_of_le_of_not_dvd_index`); and
`r ∈ β(M) ⊆ α(M)` gives `r_r(M) ≥ 3` by the definition of `α(M)`. Consumed by the rank core
`piSet_mf_inf_beta_disjoint_of_not_fittingIsTI` as the contradiction target against the `< 3`
bound coming from `C_{M_F}(X₁)`. -/
theorem three_le_pRank_mf_of_mem_beta [Finite G]
(_hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (_hM : M ∈ maximalSubgroups G)
{r : ℕ} (hrπ : r ∈ S14.piSet (MF M)) (hrβ : r ∈ OddOrder.BG.Ch3.S10.beta M) :
3 ≤ pRank ↥(MF M) r := by
have hrmem : r ∈ (Nat.card ↥(MF M)).primeFactors := hrπ
have hrp : r.Prime := Nat.prime_of_mem_primeFactors hrmem
have : Fact r.Prime := ⟨hrp⟩
-- `M_F` is `π(M_F)`-Hall in `M`; `r ∈ π(M_F)` ⟹ `r ∤ [M : M_F]`.
have hHall := maxNilpotentNormalHall_isHall (G := G) M
have hidx : ¬ r ∣ ((MF M).subgroupOf M).index := fun hdvd =>
hHall.2 r (Nat.mem_primeFactors.mpr ⟨hrp, hdvd, Subgroup.index_ne_zero_of_finite⟩) hrmem
-- `r_r(M_F) = r_r(M)`; and `r ∈ β(M) ⊆ α(M)` ⟹ `r_r(M) ≥ 3`.
rw [OddOrder.GroupTheory.pRank_eq_of_le_of_not_dvd_index (maxNilpotentNormalHall_le M) hidx]
exact (OddOrder.BG.Ch3.S10.beta_subset_alpha M hrβ).2
/-- A subgroup contained in two **distinct** maximal subgroups of a minimal simple odd group has
rank `< 3`. Contrapositive of `isUniquelyMaximal_of_three_le_rank_of_lt_top`: rank `≥ 3` would
force unique maximality, contradicting membership in two distinct coatoms. -/
theorem rank_lt_three_of_le_two_maximals [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
{C M N : Subgroup G} (hM : M ∈ maximalSubgroups G) (hN : N ∈ maximalSubgroups G)
(hMN : M ≠ N) (hCM : C ≤ M) (hCN : C ≤ N) : rank ↥C < 3 := by
by_contra h
have hCt : C < ⊤ := lt_of_le_of_lt hCM (OddOrder.GroupTheory.mem_maximalSubgroups.mp hM).lt_top
exact hMN ((OddOrder.BG.Ch2.S09.isUniquelyMaximal_of_three_le_rank_of_lt_top hG hCt
(not_lt.mp h)).eq_of_isCoatom_of_le (OddOrder.GroupTheory.mem_maximalSubgroups.mp hM) hCM
(OddOrder.GroupTheory.mem_maximalSubgroups.mp hN) hCN)
/-- **Order-`p` non-TI witness extraction** (shared infrastructure of BG Theorem 15.7(a)/(e)):
from `¬FittingIsTI M` produce an element `g ∉ M`, a prime `p ∈ σ(M)`, and an order-`p` subgroup `X₁`
of `M_σ` that is also contained in the conjugate `M_σ^g`, with `C_G(X₁) ⊄ M` and
`rank (M_F ⊓ C_G(X₁)) < 3`. This bundles the common prefix (steps 1, 3, 5–7) of
`piSet_mf_inf_beta_disjoint_of_not_fittingIsTI`, so that both the rank-core (`M_F = M_σ`,
Theorem 15.7(a)) and the type-`F` trichotomy (Theorem 15.7(e), `isTypeI_of_isTypeF`) can consume the
same witness `X₁`. The two conjugate-membership facts `X₁ ≤ M_σ` and `X₁ ≤ M_σ^g` drive the
`O_p(M_σ)`-noncyclicity argument (Coq `not_cycMp`), and `rank (M_F ⊓ C_G(X₁)) < 3` is the
`E1X_facts` rank bound.
Stated **for a fixed `g`**, exactly as BG and Coq do (`nonTI_Fitting_structure` takes
`g \notin M -> X :!=: 1 ->` as hypotheses): BG's argument works for *any* `g ∉ M` with
`X = F(M) ∩ F(M)^g ≠ 1`, and the `p = |X|` step of Theorem 15.7(e) needs `p` tied to that same `X`,
which an `∃ g` statement cannot express. The `∃ g` form is `exists_inf_conj_fitting_orderP_witness`
below, which just supplies a `g` from `¬FittingIsTI`. -/
theorem exists_orderP_witness_of_inf_conj_fitting_ne_bot [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G)
{g : G} (hgM : g ∉ M)
(hXne : (fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M : Subgroup G) ≠ ⊥) :
∃ (p : ℕ) (X₁ : Subgroup G),
p.Prime ∧ p ∈ OddOrder.BG.Ch3.S10.sigma M ∧
Nat.card ↥X₁ = p ∧
X₁ ≤ OddOrder.BG.Ch3.S10.Msigma M ∧
X₁ ≤ MulAut.conj g • OddOrder.BG.Ch3.S10.Msigma M ∧
¬ Subgroup.centralizer (X₁ : Set G) ≤ M ∧
rank ↥(MF M ⊓ Subgroup.centralizer (X₁ : Set G)) < 3 ∧
-- `X₁` sits inside the TI-failure intersection `X = F(M) ∩ F(M)ᵍ` itself. True by
-- construction (`X₁` is generated by an order-`p` element of `X`), but it was not exposed
-- until 2026-07-19; BG's `p = |X|` step needs `X₁` and `X` related in the statement.
X₁ ≤ (fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M : Subgroup G) := by
classical
have hXcard : Nat.card ↥(fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M) ≠ 1 :=
fun h => hXne (Subgroup.card_eq_one.mp h)
obtain ⟨p, hp, hpdvd⟩ := Nat.exists_prime_and_dvd hXcard
have : Fact p.Prime := ⟨hp⟩
have hpσ : p ∈ OddOrder.BG.Ch3.S10.sigma M :=
mem_sigma_of_prime_dvd_card_inf_conj_fitting hG hM hgM hp hpdvd
-- An order-`p` subgroup `X₁ ≤ X`.
obtain ⟨x, hxord⟩ := exists_prime_orderOf_dvd_card'
(G := ↥(fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M)) p hpdvd
set X₁ : Subgroup G :=
(Subgroup.zpowers x).map (fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M).subtype
with hX₁def
have hX₁leX : X₁ ≤ fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M :=
hX₁def ▸ Subgroup.map_subtype_le _
have hX₁card : Nat.card ↥X₁ = p := by
rw [hX₁def, Subgroup.card_map_of_injective
(fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M).subtype_injective,
Nat.card_zpowers, hxord]
have hX₁ne : X₁ ≠ ⊥ := fun h => hp.one_lt.ne' (by rw [← hX₁card, h, Subgroup.card_bot])
have hX₁pg : IsPGroup p ↥X₁ := IsPGroup.of_card (n := 1) (by rw [hX₁card, pow_one])
have hX₁F : X₁ ≤ fittingInAmbient M := hX₁leX.trans inf_le_left
have hX₁cF : X₁ ≤ MulAut.conj g • fittingInAmbient M := hX₁leX.trans inf_le_right
have hX₁M : X₁ ≤ M := hX₁F.trans (OddOrder.BG.Ch2.S08.fittingInG_le M)
have hpσ_sub : ({p} : Set ℕ) ⊆ OddOrder.BG.Ch3.S10.sigma M := by
intro q hq; rw [Set.mem_singleton_iff] at hq; rw [hq]; exact hpσ
-- Generic helper: a `p`-subgroup of `F(M)` lies in `M_σ`
-- (`O_p(F(M)) ≤ O_σ(F(M)) = F(M_σ) ≤ M_σ`).
have hMσ_of : ∀ Z : Subgroup G, Z ≤ fittingInAmbient M → IsPGroup p ↥Z →
Z ≤ OddOrder.BG.Ch3.S10.Msigma M := by
intro Z hZF hZp
have h1 : Z ≤ opiCoreInG ({p} : Set ℕ) (fittingInAmbient M) :=
OddOrder.BG.Ch2.S08.le_opiCoreInG_singleton_of_isPGroup_of_le_nilpotent
(OddOrder.BG.Ch2.S08.fittingInG_isNilpotent M) hZF hZp
have h2 : opiCoreInG ({p} : Set ℕ) (fittingInAmbient M)
≤ opiCoreInG (OddOrder.BG.Ch3.S10.sigma M) (fittingInAmbient M) :=
Subgroup.map_mono (OddOrder.Isaacs.Ch03.oPiCore_mono hpσ_sub _)
have h3' : opiCoreInG (OddOrder.BG.Ch3.S10.sigma M) (fittingInAmbient M)
= fittingInAmbient (OddOrder.BG.Ch3.S10.Msigma M) :=
opiCoreInG_sigma_fittingInAmbient_eq_fittingInAmbient_Msigma
exact (h1.trans h2).trans (h3' ▸ OddOrder.BG.Ch2.S08.fittingInG_le _)
have hX₁Mσ : X₁ ≤ OddOrder.BG.Ch3.S10.Msigma M := hMσ_of X₁ hX₁F hX₁pg
-- `X₁ ≤ M_σ^g`: pull `X₁` back by `g⁻¹` into `F(M)`, apply `hMσ_of`, push forward by `g`.
have hY₁ : MulAut.conj g⁻¹ • X₁ ≤ fittingInAmbient M := by
have hle : MulAut.conj g⁻¹ • X₁ ≤ MulAut.conj g⁻¹ • (MulAut.conj g • fittingInAmbient M) :=
Subgroup.pointwise_smul_le_pointwise_smul_iff.mpr hX₁cF
rwa [← mul_smul, ← map_mul, inv_mul_cancel, map_one, one_smul] at hle
have hY₁pg : IsPGroup p ↥(MulAut.conj g⁻¹ • X₁) := by
have hcard : Nat.card ↥(MulAut.conj g⁻¹ • X₁) = p := by
rw [← Nat.card_congr (Subgroup.equivSMul (MulAut.conj g⁻¹) X₁).toEquiv, hX₁card]
exact IsPGroup.of_card (n := 1) (by rw [hcard, pow_one])
have hX₁cMσ : X₁ ≤ MulAut.conj g • OddOrder.BG.Ch3.S10.Msigma M := by
have hY₁Mσ : MulAut.conj g⁻¹ • X₁ ≤ OddOrder.BG.Ch3.S10.Msigma M := hMσ_of _ hY₁ hY₁pg
have hpush := (Subgroup.pointwise_smul_le_pointwise_smul_iff (a := MulAut.conj g)).mpr hY₁Mσ
rwa [← mul_smul, ← map_mul, mul_inv_cancel, map_one, one_smul] at hpush
-- Step 6: `C_G(X₁) ⊄ M` (Theorem 10.1(e)).
have hconj_g_inv : MulAut.conj g⁻¹ • X₁ ≤ M := hY₁.trans (OddOrder.BG.Ch2.S08.fittingInG_le M)
have hCGnotM : ¬ Subgroup.centralizer (X₁ : Set G) ≤ M := by
intro hCG
have he := (OddOrder.BG.Ch3.S10.fusion_control_of_mem_sigma hG hM hpσ hX₁ne hX₁pg).2.2.2.2
exact hgM (by simpa using inv_mem (he hX₁M hCG g⁻¹ hconj_g_inv))
-- Step 7: `rank (M_F ⊓ C_G(X₁)) < 3` (`C_{M_F}(X₁)` lies in `M` and in a coatom `N ≠ M`).
obtain ⟨x₀, hx₀ne⟩ := Subgroup.ne_bot_iff_exists_ne_one.mp hX₁ne
have hCGlt : Subgroup.centralizer (X₁ : Set G) < ⊤ :=
lt_of_le_of_lt
(Subgroup.centralizer_le (Set.singleton_subset_iff.mpr x₀.2))
(OddOrder.BG.Ch2.S09.centralizer_singleton_lt_top hG
(fun h => hx₀ne (Subtype.ext h)))
obtain ⟨N, hNco, hCGN⟩ :=
(eq_top_or_exists_le_coatom (Subgroup.centralizer (X₁ : Set G))).resolve_left hCGlt.ne
have hNmax : N ∈ maximalSubgroups G := mem_maximalSubgroups.mpr hNco
have hNneM : M ≠ N := fun h => hCGnotM (h ▸ hCGN)
have hrank3 : rank ↥(MF M ⊓ Subgroup.centralizer (X₁ : Set G)) < 3 :=
rank_lt_three_of_le_two_maximals hG hM hNmax hNneM
(inf_le_left.trans (maxNilpotentNormalHall_le M)) (inf_le_right.trans hCGN)
exact ⟨p, X₁, hp, hpσ, hX₁card, hX₁Mσ, hX₁cMσ, hCGnotM, hrank3, hX₁leX⟩
/-- **BG Theorem 15.7(e) non-TI witness, `∃ g` form** — the packaging most consumers want. From
`¬FittingIsTI M` alone, `exists_notMem_inf_conj_fitting_ne_bot_of_not_fittingIsTI` produces some
`g ∉ M` with `X = F(M) ∩ F(M)^g ≠ 1`, and `exists_orderP_witness_of_inf_conj_fitting_ne_bot`
supplies the witness for it.
Prefer the fixed-`g` version when the conclusion has to mention `X` itself — notably BG's
`p = |X|` step, where the existential over `g` would decouple `p` from the intersection it is
supposed to measure. -/
theorem exists_inf_conj_fitting_orderP_witness [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G)
(hnotTI : ¬ FittingIsTI M) :
∃ (g : G) (p : ℕ) (X₁ : Subgroup G),
g ∉ M ∧ p.Prime ∧ p ∈ OddOrder.BG.Ch3.S10.sigma M ∧
Nat.card ↥X₁ = p ∧
X₁ ≤ OddOrder.BG.Ch3.S10.Msigma M ∧
X₁ ≤ MulAut.conj g • OddOrder.BG.Ch3.S10.Msigma M ∧
¬ Subgroup.centralizer (X₁ : Set G) ≤ M ∧
rank ↥(MF M ⊓ Subgroup.centralizer (X₁ : Set G)) < 3 ∧
X₁ ≤ (fittingInAmbient M ⊓ MulAut.conj g • fittingInAmbient M : Subgroup G) := by
obtain ⟨g, hgM, hXne⟩ := exists_notMem_inf_conj_fitting_ne_bot_of_not_fittingIsTI hnotTI
obtain ⟨p, X₁, h⟩ := exists_orderP_witness_of_inf_conj_fitting_ne_bot hG hM hgM hXne
exact ⟨g, p, X₁, hgM, h⟩
/-- **`O_p(M_F)` is noncyclic at a non-TI witness prime**
(Coq `nonTI_Fitting_structure`, `not_cycMp`):
if `M_F` contains an order-`p` subgroup `X₁` that is also contained in the conjugate `M_F^g` for
some
`g ∉ M`, then `O_p(M_F)` is not cyclic. Were it cyclic, `X₁` would be its unique
order-`p` subgroup,
hence characteristic in `O_p(M_F) ⊴ M`, giving `N_G(X₁) = M`; applied to `g⁻¹·X₁·g ≤ O_p(M_F)` it
gives
`N_G(g⁻¹·X₁·g) = g⁻¹·M·g`, so `M = g⁻¹·M·g`, forcing `g ∈ M` — contradiction. Both `X₁` and its
`g⁻¹`-conjugate land in the *same* cyclic `O_p(M_F)`, so no cyclic-conjugate transfer is needed.
This supplies the rank `≥ 2` lower bound of the abelian branch of Theorem 15.7(e)
(`isTypeI_of_isTypeF`): an abelian noncyclic `p`-group has `p`-rank `≥ 2`. -/
theorem not_isCyclic_opiCore_mf_of_orderP_le_conj [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G)
{g : G} {p : ℕ} {X₁ : Subgroup G} (hp : p.Prime) (hgM : g ∉ M)
(hX₁card : Nat.card ↥X₁ = p) (hX₁MF : X₁ ≤ MF M)
(hX₁cMF : X₁ ≤ MulAut.conj g • MF M) :
¬ IsCyclic ↥(opiCoreInG ({p} : Set ℕ) (MF M)) := by
classical
intro hcyc
have : Fact p.Prime := ⟨hp⟩
set C : Subgroup G := opiCoreInG ({p} : Set ℕ) (MF M) with hCdef
have : IsCyclic ↥C := hcyc
-- `M ≤ N(C)` since `C = O_p(M_F)` is characteristic in `M_F ⊴ M`.
have hMNC : M ≤ Subgroup.normalizer (C : Set G) :=
le_normalizer_opiCoreInG_of_le_normalizer _ (maxNilpotentNormalHall_le_normalizer M)
have hX₁ne : X₁ ≠ ⊥ := fun h => hp.one_lt.ne' (by rw [← hX₁card, h, Subgroup.card_bot])
have hX₁pg : IsPGroup p ↥X₁ := IsPGroup.of_card (n := 1) (by rw [hX₁card, pow_one])
have hMFnil : Group.IsNilpotent ↥(MF M) := maxNilpotentNormalHall_isNilpotent M
have hX₁C : X₁ ≤ C :=
OddOrder.BG.Ch2.S08.le_opiCoreInG_singleton_of_isPGroup_of_le_nilpotent hMFnil hX₁MF hX₁pg
have hX₁M : X₁ ≤ M := hX₁MF.trans (maxNilpotentNormalHall_le M)
-- `M ≤ N(X₁)`, hence `N(X₁) = M`.
have hMNX₁ : M ≤ Subgroup.normalizer (X₁ : Set G) :=
le_normalizer_of_le_isCyclic_normalized hX₁C hMNC
have hNX₁ : Subgroup.normalizer (X₁ : Set G) = M :=
OddOrder.BG.Ch2.S08.normalizer_eq_of_normal_of_mem_maximal hG hM
((Subgroup.normal_subgroupOf_iff_le_normalizer hX₁M).mpr hMNX₁) hX₁ne hX₁M
-- `X₁' := g⁻¹·X₁·g ≤ O_p(M_F)` too (it lies in `M_F` and is a `p`-group), so `M ≤ N(X₁')`.
set X₁' : Subgroup G := MulAut.conj g⁻¹ • X₁ with hX₁'def
have hX₁'card : Nat.card ↥X₁' = p := by
rw [hX₁'def, ← Nat.card_congr (Subgroup.equivSMul (MulAut.conj g⁻¹) X₁).toEquiv, hX₁card]
have hX₁'pg : IsPGroup p ↥X₁' := IsPGroup.of_card (n := 1) (by rw [hX₁'card, pow_one])
have hX₁'MF : X₁' ≤ MF M := by
have hle : X₁' ≤ MulAut.conj g⁻¹ • (MulAut.conj g • MF M) :=
hX₁'def ▸ Subgroup.pointwise_smul_le_pointwise_smul_iff.mpr hX₁cMF
rwa [← mul_smul, ← map_mul, inv_mul_cancel, map_one, one_smul] at hle
have hX₁'C : X₁' ≤ C :=
OddOrder.BG.Ch2.S08.le_opiCoreInG_singleton_of_isPGroup_of_le_nilpotent hMFnil hX₁'MF hX₁'pg
have hMNX₁' : M ≤ Subgroup.normalizer (X₁' : Set G) :=
le_normalizer_of_le_isCyclic_normalized hX₁'C hMNC
-- `N(X₁') = g⁻¹ • N(X₁) = g⁻¹ • M`.
have hNX₁'eq : Subgroup.normalizer (X₁' : Set G) = MulAut.conj g⁻¹ • M := by
rw [hX₁'def, ← hNX₁]
exact (Subgroup.map_normalizer_eq_of_bijective X₁ (MulAut.conj g⁻¹).bijective).symm
rw [hNX₁'eq] at hMNX₁'
have hcardM : Nat.card ↥(MulAut.conj g⁻¹ • M) = Nat.card ↥M :=
(Nat.card_congr (Subgroup.equivSMul (MulAut.conj g⁻¹) M).toEquiv).symm
have hMeq : MulAut.conj g⁻¹ • M = M :=
(Subgroup.eq_of_le_of_card_ge hMNX₁' (le_of_eq hcardM)).symm
-- `g⁻¹·M·g = M ⟹ g⁻¹ ∈ N_G(M) ≤ M ⟹ g ∈ M`, contradicting `g ∉ M`.
have hg_inv_N : g⁻¹ ∈ Subgroup.normalizer (M : Set G) := by
rw [Subgroup.mem_normalizer_iff'']
intro h
have hiff : h ∈ MulAut.conj g⁻¹ • M ↔ g * h * g⁻¹ ∈ M := by
rw [Subgroup.mem_pointwise_smul_iff_inv_smul_mem,
show ((MulAut.conj g⁻¹)⁻¹ • h : G) = g * h * g⁻¹ by
simp [MulAut.smul_def, ← map_inv, MulAut.conj_apply]]
rw [hMeq] at hiff
rw [inv_inv]; exact hiff
have hg_inv_M : g⁻¹ ∈ M :=
OddOrder.BG.Ch3.S10.maximal_normalizer_le_self hG hM hg_inv_N
exact hgM (by simpa using inv_mem hg_inv_M)
/-- **An abelian noncyclic `p`-group (`p` odd) has `p`-rank `≥ 2`** (additive analogue of mathcomp's
`abelian_rank1_cyclic`): a finite commutative `p`-group `R` with `p` odd that is not cyclic has
`2 ≤ pRank R p`. Noncyclicity gives `p < |Ω₁(R)|` (contrapositive of
`isCyclic_of_card_omega1_le_prime`); `Ω₁(R)` is elementary abelian
(`isElementaryAbelian_omega_one_of_comm`) and a `p`-group, so `p² ≤ |Ω₁(R)|` and hence
`2 ≤ log_p|Ω₁(R)| ≤ pRank R p` (`le_pRank`). Supplies the rank `≥ 2` lower bound of the abelian
branch of BG Theorem 15.7(e) (`isTypeI_of_isTypeF`). -/
theorem two_le_pRank_of_comm_isPGroup_not_isCyclic {R : Type*} [Group R] [Finite R]
{p : ℕ} [Fact p.Prime] (hp_odd : Odd p) (hcomm : ∀ x y : R, x * y = y * x)
(hR : IsPGroup p R) (hnc : ¬ IsCyclic R) : 2 ≤ pRank R p := by
have hp1 : 1 < p := (Fact.out : p.Prime).one_lt
-- `Ω₁(R)` is elementary abelian and a `p`-group.
have hΩea : (Omega R p 1).IsElementaryAbelian p :=
OddOrder.BG.Ch1_Preliminary.isElementaryAbelian_omega_one_of_comm hcomm
have hΩpg : IsPGroup p ↥(Omega R p 1) := fun g =>
(hR g.val).imp fun k hk =>
Subtype.ext (by rw [SubmonoidClass.coe_pow, OneMemClass.coe_one]; exact hk)
-- Noncyclicity ⟹ `p < |Ω₁(R)|`.
have hlt : p < Nat.card ↥(Omega R p 1) := by
by_contra hle
exact hnc (OddOrder.BG.Ch1.S04.isCyclic_of_card_omega1_le_prime hR hp_odd (not_lt.mp hle))
-- `|Ω₁(R)| = p^k` with `k ≥ 2`.
obtain ⟨k, hk⟩ := IsPGroup.iff_card.mp hΩpg
have hk2 : 2 ≤ k := by
by_contra h
push Not at h
interval_cases k
· simp only [pow_zero] at hk; rw [hk] at hlt; omega
· simp only [pow_one] at hk; rw [hk] at hlt; omega
have hcard : p ^ 2 ≤ Nat.card ↥(Omega R p 1) := by
rw [hk]; exact Nat.pow_le_pow_right (le_of_lt hp1) hk2
-- `2 ≤ log_p|Ω₁(R)| ≤ pRank R p`.
exact le_trans (Nat.le_log_of_pow_le hp1 hcard) (le_pRank (Omega R p 1) hΩea)
/-- **`C_{M_F}(X₁)` is not uniquely maximal** (Coq `nonTI_Fitting_structure`, `E1X_facts` clause
`C1 ∉ 'U`): if the centralizer `C_G(X₁)` of a nontrivial subgroup `X₁` is not contained in `M`, then
`C_{M_F}(X₁) = M_F ⊓ C_G(X₁)` is not uniquely maximal.
Were it uniquely maximal, then since `C_{M_F}(X₁) ≤ C_G(X₁) < ⊤`, the overgroup `C_G(X₁)` would also
be uniquely maximal (`IsUniquelyMaximal.of_le_of_lt_top`); and the unique maximal subgroup over
`C_{M_F}(X₁)` is `M` (a coatom containing it), so `C_G(X₁) ≤ M`, contradicting the hypothesis.
This is the `E1X_facts` input feeding the non-abelian branch of Theorem 15.7(e)
(`abelian C_{M_F}(X₁)` and `cyclic O_{p'}(M_F)`). -/
theorem not_isUniquelyMaximal_mf_inf_centralizer_of_not_le [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G)
{X₁ : Subgroup G} (hX₁ne : X₁ ≠ ⊥)
(hCGnotM : ¬ Subgroup.centralizer (X₁ : Set G) ≤ M) :
¬ OddOrder.GroupTheory.IsUniquelyMaximal (MF M ⊓ Subgroup.centralizer (X₁ : Set G)) := by
intro huniq
-- `C₁ ≤ M`, `C₁ ≤ C_G(X₁)`, and `C_G(X₁) < ⊤`.
have hC1M : MF M ⊓ Subgroup.centralizer (X₁ : Set G) ≤ M :=
inf_le_left.trans (maxNilpotentNormalHall_le M)
have hC1C : MF M ⊓ Subgroup.centralizer (X₁ : Set G) ≤ Subgroup.centralizer (X₁ : Set G) :=
inf_le_right
obtain ⟨x₀, hx₀ne⟩ := Subgroup.ne_bot_iff_exists_ne_one.mp hX₁ne
have hCGlt : Subgroup.centralizer (X₁ : Set G) < ⊤ :=
lt_of_le_of_lt (Subgroup.centralizer_le (Set.singleton_subset_iff.mpr x₀.2))
(OddOrder.BG.Ch2.S09.centralizer_singleton_lt_top hG (fun h => hx₀ne (Subtype.ext h)))
-- `C_G(X₁)` is uniquely maximal, with the same unique maximal subgroup `M` as `C₁`.
have huniqC : OddOrder.GroupTheory.IsUniquelyMaximal (Subgroup.centralizer (X₁ : Set G)) :=
huniq.of_le_of_lt_top hC1C hCGlt
have hMco : IsCoatom M := OddOrder.GroupTheory.mem_maximalSubgroups.mp hM
have heq : M = huniqC.uniqueMaximalSubgroup :=
huniq.eq_of_isCoatom_of_le hMco hC1M huniqC.uniqueMaximalSubgroup_isCoatom
(hC1C.trans huniqC.le_uniqueMaximalSubgroup)
exact hCGnotM (heq ▸ huniqC.le_uniqueMaximalSubgroup)
/-- **A finite nilpotent group with all Sylow subgroups abelian is abelian.** A finite nilpotent
group is the internal direct product of its (normal) Sylow subgroups
(`Sylow.directProductOfNormal`); if each factor is commutative the product is, and commutativity
transports back across the isomorphism. Used to prove `abelian C_{M_F}(X₁)` in the `E1X_facts`
input to BG Theorem 15.7(e): each Sylow of the nilpotent `C_{M_F}(X₁)` is abelian (a non-abelian
one would be uniquely maximal by `nonabelian_pgroup_isUniquelyMaximal`, contradicting
`not_isUniquelyMaximal_mf_inf_centralizer_of_not_le`). -/
theorem isMulCommutative_of_isNilpotent_of_sylow_comm {N : Type*} [Group N] [Finite N]
[Group.IsNilpotent N]
(h : ∀ (p : ℕ) [Fact p.Prime] (P : Sylow p N), IsMulCommutative ↥(P : Subgroup N)) :
IsMulCommutative N := by
classical
have : Fintype N := Fintype.ofFinite N
have hnorm : ∀ {p : ℕ} [Fact p.Prime] (P : Sylow p N), (↑P : Subgroup N).Normal :=
fun P => OddOrder.Isaacs.Ch01.Sylow.normal_of_isNilpotent P
let e := Sylow.directProductOfNormal (G := N) hnorm
-- The direct-product domain is commutative: each component `↥P` is.
have hDcomm : ∀ a b : (∀ p : (Nat.card N).primeFactors, ∀ P : Sylow p N, ↥(P : Subgroup N)),
a * b = b * a := by
intro a b
funext p P
have : Fact (p : ℕ).Prime := Fact.mk (Nat.prime_of_mem_primeFactors p.2)
exact (isMulCommutative_iff.mp (h p P)) (a p P) (b p P)
-- Transport commutativity across `e : (Π Sylows) ≃* N`.
refine isMulCommutative_iff.mpr fun x y => ?_
have hxy := congrArg e (hDcomm (e.symm x) (e.symm y))
rwa [map_mul, map_mul, e.apply_symm_apply, e.apply_symm_apply] at hxy
/-- **`C_{M_F}(X₁)` is abelian** (Coq `nonTI_Fitting_structure`, `E1X_facts` clause `abelian C1`):
if `C_G(X₁) ⊄ M` (with `X₁ ≠ 1`), the nilpotent subgroup `C_{M_F}(X₁) = M_F ⊓ C_G(X₁)` is abelian.
Every Sylow `P` of the nilpotent `C_{M_F}(X₁)` is abelian: a non-abelian one would be uniquely
maximal (`nonabelian_pgroup_isUniquelyMaximal`), and as `P ≤ C_{M_F}(X₁) < ⊤` this would force
`C_{M_F}(X₁)` itself to be uniquely maximal (`IsUniquelyMaximal.of_le_of_lt_top`), contradicting
`not_isUniquelyMaximal_mf_inf_centralizer_of_not_le`. Then
`isMulCommutative_of_isNilpotent_of_sylow_comm`
gives abelianness. This is the second `E1X_facts` input (with noncyclicity) to the non-abelian
branch of BG Theorem 15.7(e). -/
theorem isMulCommutative_mf_inf_centralizer_of_not_le [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G)
{X₁ : Subgroup G} (hX₁ne : X₁ ≠ ⊥)
(hCGnotM : ¬ Subgroup.centralizer (X₁ : Set G) ≤ M) :
IsMulCommutative ↥(MF M ⊓ Subgroup.centralizer (X₁ : Set G)) := by
classical
set C1 : Subgroup G := MF M ⊓ Subgroup.centralizer (X₁ : Set G) with hC1def
have hC1MF : C1 ≤ MF M := inf_le_left
-- `C1` is nilpotent (a subgroup of the nilpotent `M_F`).
have hMFnil : Group.IsNilpotent ↥(MF M) := maxNilpotentNormalHall_isNilpotent M
have : Group.IsNilpotent ↥(C1.subgroupOf (MF M)) := Subgroup.isNilpotent (C1.subgroupOf (MF M))
have hC1nil : Group.IsNilpotent ↥C1 :=
Group.nilpotent_of_surjective (Subgroup.subgroupOfEquivOfLe hC1MF).toMonoidHom
(Subgroup.subgroupOfEquivOfLe hC1MF).surjective
-- `C1 < ⊤` and `C1` is not uniquely maximal.
have hC1notU : ¬ OddOrder.GroupTheory.IsUniquelyMaximal C1 :=
not_isUniquelyMaximal_mf_inf_centralizer_of_not_le hG hM hX₁ne hCGnotM
have hC1lt : C1 < ⊤ := lt_of_le_of_lt (hC1MF.trans (maxNilpotentNormalHall_le M))
(OddOrder.GroupTheory.mem_maximalSubgroups.mp hM).lt_top
-- Each Sylow of `C1` is abelian (else `C1` is uniquely maximal).
apply isMulCommutative_of_isNilpotent_of_sylow_comm
intro r _ P
by_contra hnab
set P' : Subgroup G := (P : Subgroup ↥C1).map C1.subtype with hP'def
have hP'le : P' ≤ C1 := hP'def ▸ Subgroup.map_subtype_le _
have hequiv : ↥(P : Subgroup ↥C1) ≃* ↥P' :=
Subgroup.equivMapOfInjective _ C1.subtype C1.subtype_injective
have hP'p : IsPGroup r ↥P' := P.isPGroup'.of_equiv hequiv
have hP'nab : ¬ IsMulCommutative ↥P' := fun hcomm => hnab <|
isMulCommutative_iff.mpr fun a b => hequiv.injective <| by
rw [map_mul, map_mul]; exact (isMulCommutative_iff.mp hcomm) (hequiv a) (hequiv b)
exact hC1notU ((nonabelian_pgroup_isUniquelyMaximal hG hP'p hP'nab).of_le_of_lt_top hP'le hC1lt)
/-- **Frobenius complement order divides `|kernel| - 1`** (mathcomp `regular_norm_dvd_pred`, in the
`IsFrobeniusAction` form): if a finite group `A` acts on a finite group `N` by automorphisms with no
nonidentity element of `A` fixing a nonidentity element of `N` (`IsFrobeniusAction A N`), then
`|A| ∣ |N| - 1`. Immediate from `IsFrobeniusAction.card_modEq_one` (`|N| ≡ 1 [MOD |A|]`).
This is the divisibility crux of the exponent condition (conjunct A) of BG Theorem 15.7(e2):
applied to `N = Z_q = Ω₁(Z(O_q(M_σ)))` (order `q`) with `A = U₀` the Frobenius complement acting
fixed-point-freely on the kernel `M_σ`, it gives `|U₀| ∣ q - 1`, hence
`exp(U) = exp(U₀) ∣ q - 1`. -/
theorem card_dvd_sub_one_of_isFrobeniusAction {A N : Type*} [Group A] [Finite A] [Group N]
[Finite N] [MulDistribMulAction A N] (h : OddOrder.Isaacs.Ch06.IsFrobeniusAction A N) :
Nat.card A ∣ Nat.card N - 1 := by
have : Fintype A := Fintype.ofFinite A
have : Fintype N := Fintype.ofFinite N
have hmod : Nat.card N ≡ 1 [MOD Nat.card A] := by
simpa only [Nat.card_eq_fintype_card] using h.card_modEq_one
exact (Nat.modEq_iff_dvd' Nat.card_pos).mp hmod.symm
/-- **The join of two commuting commutative subgroups is commutative.** Writing `A ⊔ B` as the
closure of `↑A ∪ ↑B`, generators commute: two from `A` (abelian), two from `B` (abelian), or one of
each (they centralize each other, `A ≤ C_G(B)`). Used for `not_cPP` in BG Theorem 15.7(e): were
`O_p(M_F)` abelian, then `M_F = O_p(M_F) ⊔ O_{p'}(M_F)` would be abelian (the `p'`-core is abelian
and centralizes the `p`-core), contradicting non-abelianness of `M_F`. -/
theorem isMulCommutative_sup_of_le_centralizer {A B : Subgroup G}
(hA : IsMulCommutative ↥A) (hB : IsMulCommutative ↥B)
(hAB : A ≤ Subgroup.centralizer (B : Set G)) :
IsMulCommutative ↥(A ⊔ B) := by
rw [Subgroup.sup_eq_closure]
refine Subgroup.isMulCommutative_closure fun x hx y hy => ?_
rcases hx with hx | hx <;> rcases hy with hy | hy
· simpa using congrArg Subtype.val (isMulCommutative_iff.mp hA ⟨x, hx⟩ ⟨y, hy⟩)
· exact (Subgroup.mem_centralizer_iff.mp (hAB hx) y hy).symm
· exact Subgroup.mem_centralizer_iff.mp (hAB hy) x hx
· simpa using congrArg Subtype.val (isMulCommutative_iff.mp hB ⟨x, hx⟩ ⟨y, hy⟩)
/-- **A finite commutative group of odd order and rank `≤ 1` is cyclic** (additive converse of
mathcomp's `abelian_rank1_cyclic`). Each Sylow `q`-subgroup is an abelian `q`-group whose `pRank`
is at most `rank N ≤ 1`, hence cyclic (contrapositive of
`two_le_pRank_of_comm_isPGroup_not_isCyclic`, using that `q`, a divisor of the odd `|N|`, is odd);
so `N` is a `Z`-group, and a finite commutative (hence nilpotent) `Z`-group is cyclic
(`IsZGroup.exponent_eq_card` + `IsCyclic.of_exponent_eq_card`). This is the rank-1 ⇒ cyclic step of
the `cyclic O_{p'}(M_F)` conjunct of BG Theorem 15.7(e). -/
theorem isCyclic_of_isMulCommutative_of_rank_le_one {N : Type*} [Group N] [Finite N]
(hcomm : ∀ x y : N, x * y = y * x) (hodd : Odd (Nat.card N)) (hrank : rank N ≤ 1) :
IsCyclic N := by
classical
have hmc : IsMulCommutative N := isMulCommutative_iff.mpr hcomm
have hZ : _root_.IsZGroup N := by
refine ⟨fun q hq P => ?_⟩
have : Fact q.Prime := ⟨hq⟩
by_contra hPnc
rcases subsingleton_or_nontrivial ↥(P : Subgroup N) with _ | hnt
· exact hPnc inferInstance
· have hPcomm : ∀ x y : ↥(P : Subgroup N), x * y = y * x := fun x y =>
Subtype.ext (hcomm (x : N) (y : N))
have hqdvd : q ∣ Nat.card N := by
obtain ⟨n, hn⟩ := IsPGroup.iff_card.mp P.isPGroup'
have h1 : 1 < Nat.card ↥(P : Subgroup N) := Finite.one_lt_card_iff_nontrivial.mpr hnt
have hn0 : n ≠ 0 := by rintro rfl; rw [pow_zero] at hn; omega
have hqP : q ∣ Nat.card ↥(P : Subgroup N) := by rw [hn]; exact dvd_pow_self q hn0
exact hqP.trans (Subgroup.card_subgroup_dvd_card _)
have hqodd : Odd q := by
rcases hq.eq_two_or_odd' with rfl | h
· exact absurd (even_iff_two_dvd.mpr hqdvd) (Nat.not_even_iff_odd.mpr hodd)
· exact h
have h2 : 2 ≤ pRank ↥(P : Subgroup N) q :=
two_le_pRank_of_comm_isPGroup_not_isCyclic hqodd hPcomm P.isPGroup' hPnc
have hle : pRank ↥(P : Subgroup N) q ≤ rank N :=
le_trans (pRank_mono_of_le (P : Subgroup N)) (pRank_le_rank q)
omega
exact IsCyclic.of_exponent_eq_card (_root_.IsZGroup.exponent_eq_card N)
/-- **Shared "`O_p(M_F)` is non-abelian" step** for BG Theorem 15.7(e) (Coq `not_cPP`). From the
non-TI witness data (`X₁ ≤ M_F` of order `p`, `C_G(X₁) ⊄ M`) with a non-abelian Fitting subgroup
`M_F`, the `p`-core `P = O_p(M_F)` is non-abelian: the `p'`-core `R = O_{p'}(M_F)` centralizes `P`
(cores of a nilpotent group commute) and lies in the abelian `C₁ = C_{M_F}(X₁)`, so `R` is abelian;
were `P` abelian too, `M_F = P ⊔ R` would be a join of two commuting commutative subgroups, hence
abelian. Used by both conjunct B (`typeF_nonabelian_cyclic_opiCore_compl`) and the per-prime
witness `q = p` case. -/
theorem opiCore_singleton_not_isMulCommutative_of_witness [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G)
{p : ℕ} {X₁ : Subgroup G} (hp : p.Prime)
(hX₁card : Nat.card ↥X₁ = p) (hX₁MF : X₁ ≤ MF M)
(hCGnotM : ¬ Subgroup.centralizer (X₁ : Set G) ≤ M)
(hnab : ¬ IsMulCommutative ↥(MF M)) :
¬ IsMulCommutative ↥(opiCoreInG ({p} : Set ℕ) (MF M)) := by
classical
have : Fact p.Prime := ⟨hp⟩
have hX₁ne : X₁ ≠ ⊥ :=
fun h => hp.one_lt.ne' (by rw [← hX₁card, h, Subgroup.card_bot])
have hX₁pg : IsPGroup p ↥X₁ := IsPGroup.of_card (n := 1) (by rw [hX₁card, pow_one])
have hMFnil : Group.IsNilpotent ↥(MF M) := maxNilpotentNormalHall_isNilpotent M
set P : Subgroup G := opiCoreInG ({p} : Set ℕ) (MF M) with hPdef
set R : Subgroup G := opiCoreInG (({p} : Set ℕ)ᶜ) (MF M) with hRdef
have hX₁P : X₁ ≤ P :=
OddOrder.BG.Ch2.S08.le_opiCoreInG_singleton_of_isPGroup_of_le_nilpotent hMFnil hX₁MF hX₁pg
have hcomm : ⁅R, P⁆ = ⊥ := by
rw [hRdef, hPdef, Subgroup.commutator_comm]
exact OddOrder.BG.Ch2.S08.opiCoreInG_commutator_compl_eq_bot ({p} : Set ℕ) (MF M)
have hRcP : R ≤ Subgroup.centralizer (P : Set G) :=
Subgroup.commutator_eq_bot_iff_le_centralizer.mp hcomm
have hRMF : R ≤ MF M := opiCoreInG_le _ _
have hCPCX : Subgroup.centralizer (P : Set G) ≤ Subgroup.centralizer (X₁ : Set G) :=
Subgroup.centralizer_le (SetLike.coe_subset_coe.mpr hX₁P)
have hRC1 : R ≤ MF M ⊓ Subgroup.centralizer (X₁ : Set G) := le_inf hRMF (hRcP.trans hCPCX)
have hC1ab : IsMulCommutative ↥(MF M ⊓ Subgroup.centralizer (X₁ : Set G)) :=
isMulCommutative_mf_inf_centralizer_of_not_le hG hM hX₁ne hCGnotM
have hRab : ∀ x y : ↥R, x * y = y * x := fun x y =>
Subtype.ext (by
simpa using congrArg Subtype.val
(isMulCommutative_iff.mp hC1ab ⟨(x : G), hRC1 x.2⟩ ⟨(y : G), hRC1 y.2⟩))
have hPRsup : P ⊔ R = MF M := opiCoreInG_sup_compl_eq_of_isNilpotent ({p} : Set ℕ)
intro hPab
refine hnab (hPRsup ▸ isMulCommutative_sup_of_le_centralizer hPab ?_ ?_)
· exact isMulCommutative_iff.mpr hRab
· rw [Subgroup.commutator_comm] at hcomm
exact Subgroup.commutator_eq_bot_iff_le_centralizer.mp hcomm
/-- **BG Theorem 15.7(e), conjunct B — `cyclic O_{p'}(M_F)`** (Coq `nonTI_Fitting_structure`,
`cycHp'`): for the non-TI witness prime `p` (`X₁ ≤ M_F` of order `p`, `C_G(X₁) ⊄ M`) and a
non-abelian Fitting subgroup `M_F`, the `p'`-core `O_{p'}(M_F)` is cyclic.
`C₁ = C_{M_F}(X₁)` is abelian (`isMulCommutative_mf_inf_centralizer_of_not_le`) and not uniquely
maximal (`not_isUniquelyMaximal_mf_inf_centralizer_of_not_le`). The `p'`-core `R = O_{p'}(M_F)`
centralizes the `p`-core `P = O_p(M_F) ⊇ X₁` (`opiCoreInG_commutator_compl_eq_bot`), so `R ≤ C₁`,
hence `R` is abelian. `P` is non-abelian (else `M_F = P ⊔ R` would be abelian,
`isMulCommutative_sup_of_le_centralizer`), so `P ∈ 𝒰` (`nonabelian_pgroup_isUniquelyMaximal`).
Were `rank R ≥ 2`, then `D = M_F ⊓ C_G(P) ⊇ R` would have `rank ≥ 2` and lie in `C_G(P)`, so
`D ∈ 𝒰` (BG Corollary 9.2) and hence `C₁ ⊇ D` would be uniquely maximal — contradiction. Thus
`rank R ≤ 1`, and the odd abelian `R` is cyclic (`isCyclic_of_isMulCommutative_of_rank_le_one`). -/
theorem typeF_nonabelian_cyclic_opiCore_compl [Finite G]
(hG : OddOrder.BG.IsMinimalSimpleOdd G) {M : Subgroup G} (hM : M ∈ maximalSubgroups G)
{p : ℕ} {X₁ : Subgroup G} (hp : p.Prime)
(hX₁card : Nat.card ↥X₁ = p) (hX₁MF : X₁ ≤ MF M)
(hCGnotM : ¬ Subgroup.centralizer (X₁ : Set G) ≤ M)
(hnab : ¬ IsMulCommutative ↥(MF M)) :
p ∈ (Nat.card ↥(MF M)).primeFactors ∧
IsCyclic ↥(opiCoreInG (({p} : Set ℕ)ᶜ) (MF M)) := by
classical
have : Fact p.Prime := ⟨hp⟩
have hX₁ne : X₁ ≠ ⊥ := fun h => hp.one_lt.ne' (by rw [← hX₁card, h, Subgroup.card_bot])
have hX₁pg : IsPGroup p ↥X₁ := IsPGroup.of_card (n := 1) (by rw [hX₁card, pow_one])
have hMFnil : Group.IsNilpotent ↥(MF M) := maxNilpotentNormalHall_isNilpotent M
-- `p ∈ π(M_F)`: `p = |X₁| ∣ |M_F|`.
have hpπ : p ∈ (Nat.card ↥(MF M)).primeFactors :=
Nat.mem_primeFactors.mpr ⟨hp, hX₁card ▸ Subgroup.card_dvd_of_le hX₁MF, Nat.card_pos.ne'⟩
set P : Subgroup G := opiCoreInG ({p} : Set ℕ) (MF M) with hPdef
set R : Subgroup G := opiCoreInG (({p} : Set ℕ)ᶜ) (MF M) with hRdef
-- `X₁ ≤ P = O_p(M_F)`.
have hX₁P : X₁ ≤ P :=
OddOrder.BG.Ch2.S08.le_opiCoreInG_singleton_of_isPGroup_of_le_nilpotent hMFnil hX₁MF hX₁pg
-- `R = O_{p'}(M_F) ≤ C_G(P)` (the `p`- and `p'`-cores of the nilpotent `M_F` commute).
have hcomm : ⁅R, P⁆ = ⊥ := by
rw [hRdef, hPdef, Subgroup.commutator_comm]
exact OddOrder.BG.Ch2.S08.opiCoreInG_commutator_compl_eq_bot ({p} : Set ℕ) (MF M)
have hRcP : R ≤ Subgroup.centralizer (P : Set G) :=
Subgroup.commutator_eq_bot_iff_le_centralizer.mp hcomm
have hRMF : R ≤ MF M := opiCoreInG_le _ _
have hCPCX : Subgroup.centralizer (P : Set G) ≤ Subgroup.centralizer (X₁ : Set G) :=
Subgroup.centralizer_le (SetLike.coe_subset_coe.mpr hX₁P)
have hRC1 : R ≤ MF M ⊓ Subgroup.centralizer (X₁ : Set G) := le_inf hRMF (hRcP.trans hCPCX)
-- `C₁` abelian ⟹ `R` abelian.
have hC1ab : IsMulCommutative ↥(MF M ⊓ Subgroup.centralizer (X₁ : Set G)) :=
isMulCommutative_mf_inf_centralizer_of_not_le hG hM hX₁ne hCGnotM
have hRab : ∀ x y : ↥R, x * y = y * x := fun x y =>
Subtype.ext (by
simpa using congrArg Subtype.val
(isMulCommutative_iff.mp hC1ab ⟨(x : G), hRC1 x.2⟩ ⟨(y : G), hRC1 y.2⟩))
-- `P` non-abelian (`not_cPP`): else `M_F = P ⊔ R` is abelian.
have hPRsup : P ⊔ R = MF M := opiCoreInG_sup_compl_eq_of_isNilpotent ({p} : Set ℕ)
have hPnab : ¬ IsMulCommutative ↥P := by
intro hPab
refine hnab (hPRsup ▸ isMulCommutative_sup_of_le_centralizer hPab ?_ ?_)
· exact isMulCommutative_iff.mpr hRab
· rw [Subgroup.commutator_comm] at hcomm
exact Subgroup.commutator_eq_bot_iff_le_centralizer.mp hcomm
have hPpg : IsPGroup p ↥P := isPGroup_opiCoreInG_singleton (MF M)
have hPU : OddOrder.GroupTheory.IsUniquelyMaximal P :=
nonabelian_pgroup_isUniquelyMaximal hG hPpg hPnab
-- `rank R ≤ 1`: else `C₁ ∈ 𝒰` via `D = M_F ⊓ C_G(P)`, contradicting `nonuniqC1`.
have hC1notU : ¬ OddOrder.GroupTheory.IsUniquelyMaximal
(MF M ⊓ Subgroup.centralizer (X₁ : Set G)) :=
not_isUniquelyMaximal_mf_inf_centralizer_of_not_le hG hM hX₁ne hCGnotM
have hrankR : rank ↥R ≤ 1 := by
by_contra hr
have h2R : 2 ≤ rank ↥R := by omega
have hRD : R ≤ MF M ⊓ Subgroup.centralizer (P : Set G) := le_inf hRMF hRcP
have h2D : 2 ≤ rank ↥(MF M ⊓ Subgroup.centralizer (P : Set G)) :=
le_trans h2R (rank_le_of_injective (Subgroup.inclusion_injective hRD))
have hDU : OddOrder.GroupTheory.IsUniquelyMaximal (MF M ⊓ Subgroup.centralizer (P : Set G)) :=
OddOrder.BG.Ch2.S09.isUniquelyMaximal_of_le_centralizer_of_two_le_rank hG hPU inf_le_right h2D
have hDC1 : MF M ⊓ Subgroup.centralizer (P : Set G) ≤
MF M ⊓ Subgroup.centralizer (X₁ : Set G) := le_inf inf_le_left (inf_le_right.trans hCPCX)
have hC1lt : MF M ⊓ Subgroup.centralizer (X₁ : Set G) < ⊤ :=
lt_of_le_of_lt (inf_le_left.trans (maxNilpotentNormalHall_le M))
(mem_maximalSubgroups.mp hM).lt_top
exact hC1notU (hDU.of_le_of_lt_top hDC1 hC1lt)
-- `R` is odd, abelian, of rank `≤ 1`, hence cyclic.
have hRodd : Odd (Nat.card ↥R) := by
rcases Nat.even_or_odd (Nat.card ↥R) with he | ho
· exact absurd
(even_iff_two_dvd.mpr ((even_iff_two_dvd.mp he).trans (Subgroup.card_subgroup_dvd_card R)))
(Nat.not_even_iff_odd.mpr hG.odd)
· exact ho
exact ⟨hpπ, isCyclic_of_isMulCommutative_of_rank_le_one hRab hRodd hrankR⟩
/-- **BG Theorem D(2) — `M_σ ∩ M^g` is cyclic** (mmd L4317, BG Lemma 12.17 third clause, Coq
`sigma_compl_embedding` cyclic part): for `g ∉ M`, the intersection `M_σ ∩ M^g` is cyclic.
`M_σ ∩ M^g` is abelian (its derived subgroup lies in `(M_σ ∩ M^g) ⊓ M_σ' = 1`, the TI part
`S12.Msigma_inf_conj_inf_derived_eq_bot`), of odd order, and of rank `≤ 1`: any noncyclic elementary
abelian `A ≤ M_σ ∩ M^g` would, by `norm_noncyclic_sigma` (Corollary 12.4), satisfy `C_G(A) ≤ N_G(A)
≤ M`, contradicting the σ-uniqueness core `S12.centralizer_not_le_of_isPGroup_le_Msigma_inf_conj`.
A finite commutative odd group of rank `≤ 1` is cyclic
(`isCyclic_of_isMulCommutative_of_rank_le_one`). Supplies the `hD2` input of Theorem D.
*Placement*: this is §12 content (Lemma 12.17 third clause) and was originally stated next to
Theorem D in `S16_MainResults`; it lives here because its last step needs the §15 helper
`isCyclic_of_isMulCommutative_of_rank_le_one`, and because Theorem 15.7(b)
(`inf_conj_fitting_isCyclic`) consumes it upstream of §16. -/
theorem Msigma_inf_conj_isCyclic [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
{M : Subgroup G} (hM : M ∈ maximalSubgroups G) {g : G} (hg : g ∉ M) :
IsCyclic ↥(OddOrder.BG.Ch3.S10.Msigma M ⊓ MulAut.conj g • M) := by
classical
set K : Subgroup G := OddOrder.BG.Ch3.S10.Msigma M ⊓ MulAut.conj g • M with hKdef
have hKMσ : K ≤ OddOrder.BG.Ch3.S10.Msigma M := hKdef ▸ inf_le_left
-- **abelian**: `K' ≤ K ⊓ M_σ' = 1` (the TI part of Lemma 12.17).
have hTI : K ⊓ derivedInG (OddOrder.BG.Ch3.S10.Msigma M) = ⊥ := by
rw [hKdef]; exact OddOrder.BG.Ch3.S12.Msigma_inf_conj_inf_derived_eq_bot hG hM hg
have hder_bot : derivedInG K = ⊥ := by
rw [eq_bot_iff, ← hTI]
exact le_inf (Subgroup.map_subtype_le _)
(OddOrder.BG.Ch3.S12.derivedInG_le_derivedInG hKMσ)
have hcommK : commutator ↥K = ⊥ := by
have h := hder_bot
rwa [derivedInG, Subgroup.map_eq_bot_iff_of_injective _ K.subtype_injective] at h
have habelian : ∀ x y : ↥K, x * y = y * x := by
have hle : (⊤ : Subgroup ↥K) ≤ Subgroup.centralizer ((⊤ : Subgroup ↥K) : Set ↥K) :=
Subgroup.commutator_eq_bot_iff_le_centralizer.mp hcommK
intro x y
exact Subgroup.mem_centralizer_iff.mp (hle (Subgroup.mem_top y)) x (Subgroup.mem_top x)
-- **odd order**.
have hodd : Odd (Nat.card ↥K) := hG.odd.of_dvd_nat (Subgroup.card_subgroup_dvd_card K)
-- **rank ≤ 1**: a noncyclic elementary abelian `A ≤ K` contradicts the σ-uniqueness core.
have hrank : rank ↥K ≤ 1 := by
by_contra hcon
obtain ⟨p, hp, A, hAea, hAle, hAnc⟩ :=
exists_isElementaryAbelian_not_isCyclic_le_of_two_le_rank K (by omega)
have : Fact p.Prime := ⟨hp⟩
have hAmeet : A ≤ OddOrder.BG.Ch3.S10.Msigma M ⊓ MulAut.conj g • M := hKdef ▸ hAle
have hAMσ : A ≤ OddOrder.BG.Ch3.S10.Msigma M := hAmeet.trans inf_le_left
have hAM : A ≤ M := hAMσ.trans (OddOrder.BG.Ch3.S10.Msigma_le M)
have hAp : IsPGroup p ↥A := hAea.isPGroup
have hAne : A ≠ ⊥ := by rintro rfl; exact hAnc isCyclic_of_subsingleton
have hpdvdA : p ∣ Nat.card ↥A := by
obtain ⟨n, hn⟩ := hAp.exists_card_eq
rcases Nat.eq_zero_or_pos n with h0 | hpos
· exact absurd (Subgroup.card_eq_one.mp (by rw [hn, h0, pow_zero])) hAne
· exact hn ▸ dvd_pow_self p hpos.ne'
have hApσ : p ∈ OddOrder.BG.Ch3.S10.sigma M :=
OddOrder.BG.Ch3.S10.Msigma_isPiGroup M p (Nat.mem_primeFactors.mpr
⟨hp, hpdvdA.trans (Subgroup.card_dvd_of_le hAMσ), Nat.card_pos.ne'⟩)
have hNA : Subgroup.normalizer (A : Set G) ≤ M :=
OddOrder.BG.Ch3.S12.norm_noncyclic_sigma hG hM hApσ hAp hAM hAnc
have hCA : Subgroup.centralizer (A : Set G) ≤ M :=
(Subgroup.centralizer_le_normalizer _).trans hNA
exact OddOrder.BG.Ch3.S12.centralizer_not_le_of_isPGroup_le_Msigma_inf_conj
hG hM hg hApσ hAp hAne hAmeet hCA
exact isCyclic_of_isMulCommutative_of_rank_le_one habelian hodd hrank
end OddOrder.BG.Ch4.S15
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/Ch4_FamilyOfMaximal/S15_MF/PisetBetaDisjoint.lean