Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Peterfalvi §14: witness Frobenius and intersection structure

Definition
OddOrder_Peterfalvi_S14_MaximalI_WitnessFrobenius

by burkh4rt · Nov 12, 2026 · Mathlib 0df444a (Lean v4.33.1)

feit-thompsonfinite-groupsgroup-theoryishida-odd-order

Proves that the witness subgroup is Frobenius and develops its intersection structure (Peterfalvi, Character Theory for the Odd Order Theorem, Section 14, pp. 69-74, (12.10)–(12.11)). The bundle consists of lemmas only.

The bundle also proves 10 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/Peterfalvi/S14_MaximalI/WitnessFrobenius.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_Peterfalvi_S14_MaximalI_WitnessSylowCyclic

namespace OddOrder.Peterfalvi.S14

open OddOrder.GroupTheory

open OddOrder.RepresentationTheory

open OddOrder.Isaacs

open scoped Pointwise

variable {G : Type*} [Group G]

/-- **Peterfalvi (12.10) obligation B, minimality core** (§8/(12.8) obligation pinned as an
explicit hypothesis, hub 9003 Cluster A; the upstream has since been proved): for the type-I
witness `L` of (12.9), every Sylow `q`-subgroup of `L` at a prime
`q` dividing `|U|` (`U =` the complement of `H = L_F`) is **cyclic**.

Peterfalvi's argument: a prime `q ∣ |L/H|` has `q < p` — in case (8.3.c) `q ∣ p−1`; in case
(8.3.b) a Sylow `p`-subgroup `P` of `H` is of rank `2` and (8.1.c) yields an order-`q` element of
`L` acting fixed-point-freely on `Ω₁(P)`, so `q ∣ p²−1`, hence `q ∣ p−1` or `q ∣ p+1`, giving
`q < p`.  By the minimality of `p` in (12.8) (no type-I maximal has a noncyclic Sylow `q`-subgroup
of its `M/M_F` for `q < p`), a Sylow `q`-subgroup of `L` is cyclic.

**Assembly** (proven): case (a) of the (8.3) alternative is excluded by
`witness_H_sharp_not_isTISubset_of_typeI`; case (b) is the counting core
`TypeFData.prime_dvd_sq_sub_one_of_abelian_kernel` (`q ∣ p² − 1`) followed by
`prime_lt_of_odd_dvd_sq_sub_one` (`q < p`, using that `p`, `q` are odd); case (c) pairs the
exponent bound `exp U ∣ p − 1` at the prime `p ∣ |H|` with a Cauchy order-`q` element of `U`.
With `q < p`, a noncyclic Sylow `q`-subgroup `Q` of `L` would witness `InPi q` (its `L`-image has
full `q`-order, and `q ∣ [L : L_F] = |U|`), contradicting the (12.8) minimality `minimal_p`. -/
theorem witness_L_sylow_cyclic_of_dvd_complement [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hnoV : ¬ ∃ M : Subgroup G, M ∈ maximalSubgroups G ∧ OddOrder.GroupTheory.IsTypeV M)
    {ctr : CounterexampleHypothesis (G := G)} (data : RankTwoWitnessData ctr)
    (typeI : TypeIData data.L) {q : ℕ} (hq : q.Prime)
    (hqU : q ∣ Nat.card ↥typeI.typeF.U) (Q : Sylow q ↥data.L) :
    IsCyclic ↥(Q : Subgroup ↥data.L) := by
  classical
  have : Fact q.Prime := ⟨hq⟩
  have : Fact ctr.p.Prime := ⟨ctr.p_prime⟩
  -- `P₀ ≤ H`, so `p ∣ |H|`.
  have hP0H : ctr.P0 ≤ typeI.typeF.H := by
    rw [typeI.typeF.H_eq]; exact witness_P0_le_kernel hG hnoV data
  have hP0ne : ctr.P0 ≠ ⊥ := fun h => ctr.P0_noncyclic (h ▸ inferInstance)
  have hpH : ctr.p ∣ Nat.card ↥typeI.typeF.H := by
    obtain ⟨k, hk⟩ := ctr.P0_pGroup.exists_card_eq
    have hk0 : k ≠ 0 := by
      rintro rfl
      exact hP0ne (Subgroup.card_eq_one.mp (by rw [hk, pow_zero]))
    exact (dvd_pow_self ctr.p hk0).trans (hk ▸ Subgroup.card_dvd_of_le hP0H)
  -- `p` and `q` are odd (divisors of the odd `|G|`).
  have hq_odd : Odd q :=
    hG.odd.of_dvd_nat (hqU.trans (Subgroup.card_subgroup_dvd_card _))
  have hp_odd : Odd ctr.p :=
    hG.odd.of_dvd_nat (hpH.trans (Subgroup.card_subgroup_dvd_card _))
  -- Step A: `q < p`, by the (8.3) alternative for the type-I witness `L`.
  have hqp : q < ctr.p := by
    rcases typeI.alternative with hTI | ⟨hab, hrank⟩ | ⟨hexp, _⟩
    · exact absurd hTI (witness_H_sharp_not_isTISubset_of_typeI hG hnoV data typeI)
    · exact prime_lt_of_odd_dvd_sq_sub_one ctr.p_prime hq hp_odd hq_odd
        (typeI.typeF.prime_dvd_sq_sub_one_of_abelian_kernel hab hrank.le hq hpH hqU)
    · have hpmem : ctr.p ∈ (Nat.card ↥typeI.typeF.H).primeFactors :=
        Nat.mem_primeFactors.mpr ⟨ctr.p_prime, hpH, Nat.card_pos.ne'⟩
      have : Fintype ↥typeI.typeF.U := Fintype.ofFinite _
      obtain ⟨u, hu⟩ := exists_prime_orderOf_dvd_card (G := ↥typeI.typeF.U) q
        (by rwa [← Nat.card_eq_fintype_card])
      have hqp1 : q ∣ ctr.p - 1 :=
        (hu ▸ Monoid.order_dvd_exponent u).trans (hexp ctr.p ctr.p_prime hpmem)
      have hp2 := ctr.p_prime.two_le
      have := Nat.le_of_dvd (by omega) hqp1
      omega
  -- Step B: minimality of `p` (12.8) — a noncyclic Sylow `q` of `L` would put `q ∈ π`.
  by_contra hnc
  refine absurd (ctr.minimal_p q hq ⟨data.L, data.L_maximal, ⟨typeI⟩,
    (Q : Subgroup ↥data.L).map data.L.subtype, Subgroup.map_subtype_le _,
    Q.2.map data.L.subtype, ?_, ?_, ?_⟩) (not_le.mpr hqp)
  · -- The image has full `q`-order in `L`: `¬ q ∣ [L : Q]`.
    rw [Subgroup.relIndex, Subgroup.subgroupOf,
      Subgroup.comap_map_eq_self_of_injective data.L.subtype_injective]
    exact Q.not_dvd_index
  · -- Noncyclicity transfers along `Q ≅ Q.map L.subtype`.
    intro hc
    have := hc
    exact hnc (isCyclic_of_surjective
      (Subgroup.equivMapOfInjective (Q : Subgroup ↥data.L) data.L.subtype
        data.L.subtype_injective).symm.toMonoidHom
      (Subgroup.equivMapOfInjective (Q : Subgroup ↥data.L) data.L.subtype
        data.L.subtype_injective).symm.surjective)
  · -- `q ∣ [L : L_F] = |U|`.
    rw [← typeI.typeF.H_eq, Subgroup.relIndex, typeI.typeF.complement.symm.index_eq_card,
      Nat.card_congr (Subgroup.subgroupOfEquivOfLe typeI.typeF.U_le).toEquiv]
    exact hqU

/-- **Peterfalvi (12.10) obligation B**: the type-I witness `L`'s complement `U` is a Z-group.

**Assembly** (`sorry`-free modulo the (8.3)/(8.1.c)/(12.8) minimality core): to show every Sylow
`q`-subgroup `P` of `U` is cyclic, distinguish `q ∣ |U|` from `q ∤ |U|`.  If `q ∤ |U|` then `P` is
trivial (its order is a `q`-power dividing `|U|`, forcing order `1`), hence cyclic.  If `q ∣ |U|`,
embed `U ↪ L` (via `U_le`): `P` becomes a `q`-subgroup of `L`, contained in a Sylow `q`-subgroup `Q`
of `L`, which is cyclic by the minimality core `witness_L_sylow_cyclic_of_dvd_complement`; a
subgroup
of a cyclic group is cyclic, so `P` is cyclic. -/
theorem witness_L_complement_isZGroup [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hnoV : ¬ ∃ M : Subgroup G, M ∈ maximalSubgroups G ∧ OddOrder.GroupTheory.IsTypeV M)
    {ctr : CounterexampleHypothesis (G := G)} (data : RankTwoWitnessData ctr)
    (typeI : TypeIData data.L) :
    _root_.IsZGroup ↥typeI.typeF.U := by
  classical
  rw [isZGroup_iff]
  intro q hq P
  have : Fact q.Prime := ⟨hq⟩
  by_cases hqU : q ∣ Nat.card ↥typeI.typeF.U
  · -- `q ∣ |U|`: embed `P` into `L`, contain it in a cyclic Sylow `q`-subgroup of `L`.
    -- `P`, pushed along `U ↪ L`, is a `q`-subgroup of `L`.
    have hincl : Function.Injective (Subgroup.inclusion typeI.typeF.U_le) :=
      Subgroup.inclusion_injective _
    set PL : Subgroup ↥data.L :=
      (P : Subgroup ↥typeI.typeF.U).map (Subgroup.inclusion typeI.typeF.U_le) with hPL
    have hPLpg : IsPGroup q ↥PL :=
      (P.2.map (Subgroup.inclusion typeI.typeF.U_le))
    obtain ⟨Q, hQle⟩ := hPLpg.exists_le_sylow
    -- The containing Sylow `q`-subgroup of `L` is cyclic (minimality core).
    have hQcyc : IsCyclic ↥(Q : Subgroup ↥data.L) :=
      witness_L_sylow_cyclic_of_dvd_complement hG hnoV data typeI hq hqU Q
    -- A subgroup of a cyclic group is cyclic; `PL ≤ Q ≅ P`.
    have : IsCyclic ↥PL := Subgroup.isCyclic_of_le hQle
    exact isCyclic_of_surjective _
      (Subgroup.equivMapOfInjective (P : Subgroup ↥typeI.typeF.U)
        (Subgroup.inclusion typeI.typeF.U_le) hincl).symm.surjective
  · -- `q ∤ |U|`: the Sylow `q`-subgroup is trivial, hence cyclic.
    have hcard : Nat.card ↥(P : Subgroup ↥typeI.typeF.U) ∣ Nat.card ↥typeI.typeF.U :=
      (P : Subgroup ↥typeI.typeF.U).card_subgroup_dvd_card
    obtain ⟨k, hk⟩ := P.2.exists_card_eq
    have hqk : q ^ k ∣ Nat.card ↥typeI.typeF.U := hk ▸ hcard
    have hk0 : k = 0 := by
      by_contra hk0
      exact hqU ((dvd_pow_self q hk0).trans hqk)
    have hcard1 : Nat.card ↥(P : Subgroup ↥typeI.typeF.U) = 1 := by rw [hk, hk0, pow_zero]
    have : Subsingleton ↥(P : Subgroup ↥typeI.typeF.U) :=
      (Finite.card_le_one_iff_subsingleton).mp (by omega)
    infer_instance

/-- **Peterfalvi (12.10)**: the maximal subgroup `L` supplied by (12.9) is Frobenius with kernel
`L_F`.  **Assembly** (`sorry`-free modulo the two (12.10) obligations): `L` is Type I
(`witness_L_isTypeI`) and its complement `U` is a Z-group (`witness_L_complement_isZGroup`), so the
(8.2.b) bridge `typeI_frobenius_of_isZGroup_complement` yields the Frobenius structure with kernel
`H = L_F` (`typeF.H_eq`). -/
theorem witness_L_frobenius [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hnoV : ¬ ∃ M : Subgroup G, M ∈ maximalSubgroups G ∧ OddOrder.GroupTheory.IsTypeV M)
    {ctr : CounterexampleHypothesis (G := G)} (data : RankTwoWitnessData ctr) :
    ∃ frob : TypeIFrobeniusData data.L, frob.kernel_eq_MF := by
  obtain ⟨typeI⟩ := witness_L_isTypeI hG hnoV data
  exact ⟨{ typeI := typeI
           complement := typeI.typeF.U.subgroupOf data.L
           kernel_eq_MF := typeI.typeF.H = maxNilpotentNormalHall data.L
           kernel_eq_MF_holds := typeI.typeF.H_eq
           frobenius := typeI_frobenius_of_isZGroup_complement typeI
             (witness_L_complement_isZGroup hG hnoV data typeI) },
         typeI.typeF.H_eq⟩

/-- **Peterfalvi (12.11), first assertion**: `M ∩ L` complements `K = M_F` in `M`.  This is the
"first assertion follows from (12.9) and (8.13.c1)" step, with the (8.13)/BG roles swapped from
the (12.11) notation: the witness `x ∈ Ω₁(P₀)^# ⊆ L_F = L_σ` (type I) is a `σ`-sharp element of
`L` **escaping `L`** (`C_G(x) ⊄ L`), and its supporting maximal is `M` (`C_G(x) ≤ N_G(⟨x⟩) ≤ M`
pins `M` in the singleton `𝓜(C_G(x))`).

**Assembly** (proven): BG Theorem D(4) tail — "`M ∩ N` is a complement of `N_σ` in `N`" — is the
`IsComplement'` conjunct of `signalizer_structure_of_mem_sigmaSharp` (its unique `N` is `M` by the
singleton), applied at `M' := L ∈ 𝓜_σ(x)`; `N_σ = M_σ = M_F = K` is the type-I identification
`MF_eq_Msigma`. -/
theorem intersection_complements_K [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hnoV : ¬ ∃ M : Subgroup G, M ∈ maximalSubgroups G ∧ OddOrder.GroupTheory.IsTypeV M)
    {ctr : CounterexampleHypothesis (G := G)} (data : RankTwoWitnessData ctr) :
    Subgroup.IsComplement' (ctr.K.subgroupOf ctr.M) ((ctr.M ⊓ data.L).subgroupOf ctr.M) := by
  classical
  -- The witness `L` is type I with `L_F = L_σ`, so `x ∈ P₀ ⊆ L_F` is `σ`-sharp in `L`.
  have hLtypeI : IsTypeI data.L := witness_L_isTypeI hG hnoV data
  have hLF_eq : maxNilpotentNormalHall data.L = OddOrder.BG.Ch3.S10.Msigma data.L :=
    (OddOrder.BG.Ch4.S16.proposition_type_classification hG data.L_maximal).2.2.2.2.2.mpr
      (Or.inl hLtypeI)
  have hxLσ : data.x ∈ OddOrder.BG.Ch3.S10.Msigma data.L :=
    hLF_eq ▸ witness_P0_le_kernel hG hnoV data data.x_mem_P0
  have hxσ : data.x ∈ OddOrder.BG.Ch4.S14.sigmaSharp data.L :=
    ⟨hxLσ, by simpa using data.x_ne_one⟩
  -- `C_G(x) ≤ M` (via `N_G(⟨x⟩) ≤ M`), so `M` is THE maximal subgroup over `C_G(x)`.
  have hCM : Subgroup.centralizer ({data.x} : Set G) ≤ ctr.M := by
    refine le_trans ?_ data.normalizer_closure_x_le_M
    rw [← Subgroup.centralizer_closure]
    exact Subgroup.centralizer_le_normalizer _
  obtain ⟨N₀, hN₀⟩ :=
    OddOrder.BG.Ch4.S16.maximalSubgroupsContaining_centralizer_eq_singleton_of_sigmaSharp_escape
      hG data.L_maximal hxσ data.centralizer_x_not_le_L
  have hMN₀ : ctr.M = N₀ := by
    have hMin : ctr.M ∈ maximalSubgroupsContaining
        (Subgroup.centralizer ({data.x} : Set G)) := ⟨ctr.M_maximal, hCM⟩
    rw [hN₀] at hMin
    exact hMin
  -- The signalizer structure at the escaping `σ`-sharp `x`; its unique `N` is `M`.
  have hgt : 1 < (OddOrder.BG.Ch4.S14.maximalSigmaSubgroupsOfElement data.x).ncard := by
    by_contra h
    push Not at h
    exact data.centralizer_x_not_le_L
      (OddOrder.BG.Ch4.S16.centralizer_le_of_maximalSigma_le_one hG data.L_maximal hxLσ
        data.x_ne_one h)
  obtain ⟨N, ⟨hNmax, hCN, -, -, -, -, hforall⟩, -⟩ :=
    OddOrder.BG.Ch4.S16.signalizer_structure_of_mem_sigmaSharp hG data.L_maximal hxσ hgt
  have hNM : N = ctr.M := by
    have hNin : N ∈ maximalSubgroupsContaining
        (Subgroup.centralizer ({data.x} : Set G)) := ⟨hNmax, hCN⟩
    rw [hN₀] at hNin
    rw [hNin, ← hMN₀]
  -- `L ∈ 𝓜_σ(x)`: apply the Theorem D(4) complement conjunct at `M' = L`.
  have hLin : data.L ∈ OddOrder.BG.Ch4.S14.maximalSigmaSubgroupsOfElement data.x :=
    ⟨data.L_maximal, hxLσ⟩
  obtain ⟨-, -, hcompl, -⟩ := hforall data.L hLin
  rw [hNM, ← MF_eq_Msigma hG ctr,
    show data.L ⊓ ctr.M = ctr.M ⊓ data.L from inf_comm .. ] at hcompl
  exact hcompl

/-- **`|M ∩ L|` is coprime to `|K|`** (from the first assertion (12.11) + `M_F` Hall).  `M ∩ L`
complements `K = M_F` in `M` (`intersection_complements_K`), so `|M ∩ L| = [M : K]`, which is
coprime to `|K|` because `K = M_F` is a Hall subgroup of `M`. -/
theorem card_MinfL_coprime_card_K [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hnoV : ¬ ∃ M : Subgroup G, M ∈ maximalSubgroups G ∧ OddOrder.GroupTheory.IsTypeV M)
    {ctr : CounterexampleHypothesis (G := G)} (data : RankTwoWitnessData ctr) :
    Nat.Coprime (Nat.card ↥(ctr.M ⊓ data.L)) (Nat.card ↥ctr.K) := by
  have hKM : ctr.K ≤ ctr.M := ctr.K_eq_MF ▸ OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le ctr.M
  have hcompl := intersection_complements_K hG hnoV data
  have hHall := (OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_isHall ctr.M).coprime_index
  rw [← ctr.K_eq_MF] at hHall
  -- `hHall : Coprime |K| [M : K]`;  `[M : K] = |M ∩ L|` by the complement.
  have hidx : (ctr.K.subgroupOf ctr.M).index = Nat.card ↥((ctr.M ⊓ data.L).subgroupOf ctr.M) :=
    hcompl.symm.index_eq_card
  rw [hidx, Nat.card_congr (Subgroup.subgroupOfEquivOfLe inf_le_left).toEquiv,
    Nat.card_congr (Subgroup.subgroupOfEquivOfLe hKM).toEquiv] at hHall
  exact hHall.symm

/-- **Peterfalvi (12.11), core of the second assertion**: a subgroup `A ≤ M ∩ L` meeting the
witness kernel `H = L_F` trivially (`A ⊓ H = ⊥`, i.e. of order prime to `|H|`) is trivial.

The genuine (12.11) argument, now fully assembled from the landed infrastructure.  Put
`P = O_p(H) ∩ M` (an `A`-invariant `p`-subgroup of `H` containing `P₀`, via the nilpotent core
`opiCoreInG`); then:
* `P` does not centralize `K` (`P₀_not_le_centralizer_K`, `P₀ ≤ P`);
* `P ⊔ A` is Frobenius with kernel `P` (from `L`'s Frobenius structure), acts coprimely on `K`
  (`P ⊔ A ≤ M ∩ L`, coprime to `|K|`), so by Wielandt (9.1)
  `exists_ne_one_centralized_by_complement_of_kernel_not_centralizes` gives `C_K(A) ≠ 1`;
* `C_K(x) ≠ 1` by (12.9) (`ctr.CKx_not_le_Kprime`);
* since `M ∩ L` complements `K` in `M` (first assertion), `A` and `x` land in a common abelian
  subgroup `W` (`exists_abelian_centralizer_le_of_isComplement` with `V = M ∩ L`), so `A`
  centralizes `x`;
* by `L`'s Frobenius condition (4) (`centralizer_kernel_le`, `x ∈ H^#`), `A ≤ C_L(x) ⊆ H`, so
  `A ⊆ H`, forcing `A = A ⊓ H = ⊥`. -/
theorem witness_MinfL_pprime_subgroup_eq_bot [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hnoV : ¬ ∃ M : Subgroup G, M ∈ maximalSubgroups G ∧ OddOrder.GroupTheory.IsTypeV M)
    {ctr : CounterexampleHypothesis (G := G)} (data : RankTwoWitnessData ctr) {A : Subgroup G}
    (hAML : A ≤ ctr.M ⊓ data.L) (hAH : A ⊓ maxNilpotentNormalHall data.L = ⊥) (hAne : A ≠ ⊥) :
    False := by
  classical
  have : Fact ctr.p.Prime := ⟨ctr.p_prime⟩
  set H : Subgroup G := maxNilpotentNormalHall data.L with hHdef
  have hAM : A ≤ ctr.M := hAML.trans inf_le_left
  have hAL : A ≤ data.L := hAML.trans inf_le_right
  have hHnilp : Group.IsNilpotent ↥H := OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_isNilpotent
      data.L
  have hHL : H ≤ data.L := OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le data.L
  have hHnorm : data.L ≤ Subgroup.normalizer (H : Set G) :=
    OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le_normalizer data.L
  -- Frobenius structure of `L` with kernel `H` (upstream of this theorem).
  obtain ⟨frob, _⟩ := witness_L_frobenius hG hnoV data
  have hHfrob : frob.typeI.typeF.H = H := frob.typeI.typeF.H_eq
  have hFrobL : ∃ C : Subgroup ↥data.L,
      OddOrder.Isaacs.Ch06.IsFrobeniusGroup ↥data.L (H.subgroupOf data.L) C :=
    ⟨frob.complement, hHfrob ▸ frob.frobenius⟩
  -- `x ∈ H^#` and `x ∈ M ∩ L`.
  have hxH : data.x ∈ H := witness_P0_le_kernel hG hnoV data data.x_mem_P0
  have hxML : data.x ∈ ctr.M ⊓ data.L := ⟨ctr.P0_le_M data.x_mem_P0, hHL hxH⟩
  -- `P = O_p(H) ∩ M` contains `P₀`, sits inside `H` and `M`.
  set P : Subgroup G := opiCoreInG ({ctr.p} : Set ℕ) H ⊓ ctr.M with hPdef
  have hP0_le_P : ctr.P0 ≤ P :=
    le_inf (pGroup_le_opiCoreInG_of_le_of_isNilpotent ctr.P0_pGroup (witness_P0_le_kernel hG hnoV
        data))
      ctr.P0_le_M
  have hP_le_H : P ≤ H := inf_le_left.trans (opiCoreInG_le _ _)
  have hP_le_M : P ≤ ctr.M := inf_le_right
  have hP0ne : ctr.P0 ≠ ⊥ := fun hb => ctr.P0_noncyclic (hb ▸ isCyclic_of_subsingleton)
  have hPne : P ≠ ⊥ := fun hb => hP0ne (le_bot_iff.mp (hb ▸ hP0_le_P))
  -- `A` normalises `P` (normalises `O_p(H)` and `M`).
  have hAnorm_opi : A ≤ Subgroup.normalizer (opiCoreInG ({ctr.p} : Set ℕ) H) :=
    le_normalizer_opiCoreInG_of_le_normalizer _ (hAL.trans hHnorm)
  have hAnorm_M : A ≤ Subgroup.normalizer (ctr.M : Set G) := hAM.trans Subgroup.le_normalizer
  have hAP : A ≤ Subgroup.normalizer (P : Set G) := by
    intro a ha
    rw [Subgroup.mem_normalizer_iff]
    intro y
    have hOpi := (Subgroup.mem_normalizer_iff.mp (hAnorm_opi ha)) y
    have hM := (Subgroup.mem_normalizer_iff.mp (hAnorm_M ha)) y
    simp only [hPdef, Subgroup.mem_inf]
    rw [hOpi, hM]
  -- `P ⊔ A ≤ N_G(K)` and `≤ M ∩ L`.
  have hMnorm_K : ctr.M ≤ Subgroup.normalizer (ctr.K : Set G) :=
    ctr.K_eq_MF ▸ OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le_normalizer ctr.M
  have hPAK : P ⊔ A ≤ Subgroup.normalizer (ctr.K : Set G) :=
    sup_le (hP_le_M.trans hMnorm_K) (hAM.trans hMnorm_K)
  have hPA_ML : P ⊔ A ≤ ctr.M ⊓ data.L :=
    sup_le (le_inf hP_le_M (hP_le_H.trans hHL)) hAML
  -- `K` is solvable (subgroup of the solvable maximal `M`).
  have hKM : ctr.K ≤ ctr.M := ctr.K_eq_MF ▸ OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le ctr.M
  have hMsolv : Group.IsSolvable ↥ctr.M := hG.isSolvable_of_mem_maximalSubgroups ctr.M_maximal
  have hKsolv : Group.IsSolvable ↥ctr.K :=
    Group.isSolvable_of_isSolvable_injective (Subgroup.inclusion_injective hKM)
  -- Coprimality of the `P ⊔ A`-action on `K`.
  have hcop : Nat.Coprime (Nat.card ↥ctr.K) (Nat.card ↥(P ⊔ A)) :=
    (card_MinfL_coprime_card_K hG hnoV data).symm.coprime_dvd_right (Subgroup.card_dvd_of_le hPA_ML)
  -- `P` does not centralize `K`.
  have hPnc : ¬ P ≤ Subgroup.centralizer (ctr.K : Set G) := fun hPc =>
    P0_not_le_centralizer_K hG ctr (hP0_le_P.trans hPc)
  -- **`C_K(A) ≠ 1`** (Wielandt (9.1) via the sub-Frobenius engine).
  obtain ⟨n, hnK, hn1, hnA⟩ := exists_ne_one_centralized_by_complement_of_kernel_not_centralizes
    hHL hFrobL hP_le_H hPne hAL hAH hAne hAP hPAK hKsolv hcop hPnc
  -- **`C_K(x) ≠ 1`** (Peterfalvi (12.9)).
  obtain ⟨n', hn'mem, hn'K'⟩ := SetLike.not_le_iff_exists.mp data.CKx_not_le_Kprime
  obtain ⟨hn'C, hn'K⟩ := Subgroup.mem_inf.mp hn'mem
  have hn'1 : n' ≠ 1 := fun h => hn'K' (by rw [h]; exact Subgroup.one_mem _)
  -- **`A` and `x` in a common abelian `W ≤ M ∩ L`** (step (8.1.b), `V = M ∩ L`).
  obtain ⟨typeIM⟩ := ctr.M_typeI
  have htypeFH : typeIM.typeF.H = ctr.K := ctr.K_eq_MF ▸ typeIM.typeF.H_eq
  obtain ⟨W, hWab, hWle⟩ := exists_abelian_centralizer_le_of_isComplement hMsolv typeIM.typeF
    (V := ctr.M ⊓ data.L) inf_le_left (htypeFH ▸ intersection_complements_K hG hnoV data)
  have hnFH : n ∈ typeIM.typeF.H := by rw [htypeFH]; exact hnK
  have hn'FH : n' ∈ typeIM.typeF.H := by rw [htypeFH]; exact hn'K
  have hA_W : A ≤ W := by
    intro a haA
    have haC : (a : G) ∈ Subgroup.centralizer ({n} : Set G) :=
      Subgroup.mem_centralizer_singleton_iff.mpr (mul_inv_eq_iff_eq_mul.mp (hnA a haA))
    exact hWle n hnFH hn1 ⟨hAML haA, haC⟩
  have hx_W : data.x ∈ W := by
    have hxC : data.x ∈ Subgroup.centralizer ({n'} : Set G) :=
      Subgroup.mem_centralizer_singleton_iff.mpr
        (Subgroup.mem_centralizer_singleton_iff.mp hn'C).symm
    exact hWle n' hn'FH hn'1 ⟨hxML, hxC⟩
  -- **`A` centralizes `x`** (both in abelian `W`), then **`A ⊆ H`** by Frobenius condition (4).
  have hA_H : A ≤ H := by
    intro a haA
    have hax : (a : G) * data.x = data.x * a :=
      congrArg Subtype.val (hWab.is_comm.comm ⟨a, hA_W haA⟩ ⟨data.x, hx_W⟩)
    have hxHfrob : (⟨data.x, hHL hxH⟩ : ↥data.L) ∈ frob.typeI.typeF.H.subgroupOf data.L := by
      rw [Subgroup.mem_subgroupOf, hHfrob]; exact hxH
    have hx1 : (⟨data.x, hHL hxH⟩ : ↥data.L) ≠ 1 :=
      fun h => data.x_ne_one (by simpa using congrArg Subtype.val h)
    have hcent := frob.frobenius.centralizer_kernel_le _ hxHfrob hx1
    have haC : (⟨a, hAL haA⟩ : ↥data.L) ∈
        Subgroup.centralizer ({(⟨data.x, hHL hxH⟩ : ↥data.L)} : Set ↥data.L) := by
      rw [Subgroup.mem_centralizer_singleton_iff]
      exact Subtype.ext (by simpa using hax)
    have haH := hcent haC
    rw [Subgroup.mem_subgroupOf, hHfrob] at haH
    exact haH
  exact hAne (le_bot_iff.mp (hAH ▸ le_inf le_rfl hA_H))

/-- **Peterfalvi (12.11), second assertion**: `M ∩ L ⊆ H = L_F`.  `M ∩ L` has no nontrivial
subgroup meeting `H` trivially (`witness_MinfL_pprime_subgroup_eq_bot`), so its order is coprime to
`[L : H]` (any common prime would give a nontrivial Sylow subgroup meeting `H` trivially), and the
normal-Hall reduction `le_of_coprime_card_index_of_normal` places `M ∩ L` inside `H`. -/
theorem intersection_le_kernel [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hnoV : ¬ ∃ M : Subgroup G, M ∈ maximalSubgroups G ∧ OddOrder.GroupTheory.IsTypeV M)
    {ctr : CounterexampleHypothesis (G := G)} (data : RankTwoWitnessData ctr) :
    ctr.M ⊓ data.L ≤ maxNilpotentNormalHall data.L := by
  classical
  set H : Subgroup G := maxNilpotentNormalHall data.L with hHdef
  have hHL : H ≤ data.L := OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le data.L
  have hMLL : ctr.M ⊓ data.L ≤ data.L := inf_le_right
  have hHnorm : (H.subgroupOf data.L).Normal :=
    OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_subgroupOf_normal data.L
  -- `|M ∩ L|` is coprime to `[L : H]`.
  have hcop : Nat.Coprime (Nat.card ↥((ctr.M ⊓ data.L).subgroupOf data.L))
      (H.subgroupOf data.L).index := by
    rw [Nat.card_congr (Subgroup.subgroupOfEquivOfLe hMLL).toEquiv, Nat.coprime_iff_gcd_eq_one]
    by_contra hgcd
    obtain ⟨q, hq, hqdvd⟩ := Nat.exists_prime_and_dvd hgcd
    have : Fact q.Prime := ⟨hq⟩
    rw [Nat.dvd_gcd_iff] at hqdvd
    obtain ⟨hqML, hqidx⟩ := hqdvd
    -- A Sylow `q`-subgroup `Q` of `M ∩ L` is nontrivial, meets `H` trivially, contradicts the core.
    obtain ⟨Q⟩ := (Sylow.nonempty : Nonempty (Sylow q ↥(ctr.M ⊓ data.L)))
    set A : Subgroup G := (Q : Subgroup ↥(ctr.M ⊓ data.L)).map (ctr.M ⊓ data.L).subtype with hAdef
    have hApg : IsPGroup q ↥A := Q.2.map _
    have hAML : A ≤ ctr.M ⊓ data.L := Subgroup.map_subtype_le _
    -- `q ∉ π(H)` (as `q ∣ [L : H]` and `H` is Hall in `L`), so `A ⊓ H = ⊥`.
    have hqH : ¬ q ∣ Nat.card ↥H := by
      have hHallL := (OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_isHall data.L).coprime_index
      intro hdvd
      have hdvd' : q ∣ Nat.card ↥(H.subgroupOf data.L) := by
        rw [Nat.card_congr (Subgroup.subgroupOfEquivOfLe hHL).toEquiv]; exact hdvd
      have hg : q ∣ Nat.gcd (Nat.card ↥(H.subgroupOf data.L)) (H.subgroupOf data.L).index :=
        Nat.dvd_gcd hdvd' hqidx
      rw [hHallL] at hg
      exact hq.one_lt.ne' (Nat.dvd_one.mp hg)
    have hAH : A ⊓ H = ⊥ := by
      rw [eq_bot_iff]
      intro z hz
      obtain ⟨hzA, hzH⟩ := Subgroup.mem_inf.mp hz
      rw [Subgroup.mem_bot]
      by_contra hzne
      apply hqH
      obtain ⟨k, hk⟩ := (IsPGroup.iff_orderOf.mp hApg) ⟨z, hzA⟩
      have hoz : orderOf z = q ^ k := by
        rw [← hk]; exact orderOf_injective A.subtype A.subtype_injective ⟨z, hzA⟩
      have hk0 : k ≠ 0 := fun h => hzne (orderOf_eq_one_iff.mp (by rw [hoz, h, pow_zero]))
      have hqoz : q ∣ Nat.card ↥(Subgroup.zpowers z) := by
        rw [Nat.card_zpowers, hoz]; exact dvd_pow_self q hk0
      exact hqoz.trans (Subgroup.card_dvd_of_le ((Subgroup.zpowers_le).mpr hzH))
    have hAne : A ≠ ⊥ := by
      have hqQ : q ∣ Nat.card ↥(Q : Subgroup ↥(ctr.M ⊓ data.L)) := by
        have hmul := Subgroup.card_mul_index (Q : Subgroup ↥(ctr.M ⊓ data.L))
        rcases (Nat.Prime.dvd_mul hq).mp (hmul ▸ hqML) with h | h
        · exact h
        · exact absurd h Q.not_dvd_index
      intro hb
      have hA1 : Nat.card ↥A = Nat.card ↥(Q : Subgroup ↥(ctr.M ⊓ data.L)) :=
        (Nat.card_congr (Subgroup.equivMapOfInjective _ _
          (ctr.M ⊓ data.L).subtype_injective).toEquiv).symm
      rw [hb, Subgroup.card_bot] at hA1
      rw [← hA1] at hqQ
      exact hq.one_lt.ne' (Nat.dvd_one.mp hqQ)
    exact witness_MinfL_pprime_subgroup_eq_bot hG hnoV data hAML hAH hAne
  -- Apply the normal-Hall reduction.
  have hle := Subgroup.le_of_coprime_card_index_of_normal hcop
  intro z hz
  have : (⟨z, hMLL hz⟩ : ↥data.L) ∈ H.subgroupOf data.L :=
    hle (Subgroup.mem_subgroupOf.mpr hz)
  exact Subgroup.mem_subgroupOf.mp this

/-- **Peterfalvi (12.11)**: `M ∩ L` complements `K` in `M` and lies in the Fitting kernel
`H = L_F` of the witness subgroup `L`.

**Assembly**: the two textbook assertions of (12.11) are `intersection_complements_K` (from (12.9)
and (8.13.c1)) and `intersection_le_kernel` (the (8.1.b/c)+(9.1)+(12.10) `A = 1` argument),
combined here. -/
theorem intersection_complement_structure [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hnoV : ¬ ∃ M : Subgroup G, M ∈ maximalSubgroups G ∧ OddOrder.GroupTheory.IsTypeV M)
    {ctr : CounterexampleHypothesis (G := G)} (data : RankTwoWitnessData ctr) :
    Subgroup.IsComplement' (ctr.K.subgroupOf ctr.M) ((ctr.M ⊓ data.L).subgroupOf ctr.M) ∧
      ctr.M ⊓ data.L ≤ maxNilpotentNormalHall data.L :=
  ⟨intersection_complements_K hG hnoV data, intersection_le_kernel hG hnoV data⟩

/-- **Peterfalvi (12.10), non-TI clause**: for the (12.9) witness `L`, its Frobenius kernel
`H = L_F` has `H^#` **not** a TI-subset of `G`.  This is the "By (12.9), `H^#` is not a TI-subset"
step of (12.10): the rank-two witness `x ∈ Ω₁(P₀)^#` has `C_G(x) ⊄ L`
(`data.centralizer_x_not_le_L`)
while `N_G(H) = L` (maximality of `L` + `H = L_F` normal); pick `g ∈ C_G(x) ∖ L`, then `g ∉ N_G(H)`
yet `g x g⁻¹ = x ∈ H^#`, witnessing the TI failure (`x ∈ H^# ∩ (H^#)^g`).

This is the honest (12.9)/(12.10) prerequisite of the *witness* coherence route: with it,
`witness_L_coherent` dispatches only through the (b)/(c) cases of (12.6) (which are `sorry`-free),
never the TI-only case (a) — so the witness coherence depends on this genuine (12.9) fact rather
than on the (8.18.c) geometry that case (a) (`sibleyTarget_frobI`) transitively needs.

Specialization of the `TypeIData`-form `witness_H_sharp_not_isTISubset_of_typeI` to the
Frobenius witness (whose `x ∈ H` route is the upstream `witness_P0_le_kernel`, not (12.11)). -/
theorem witness_H_sharp_not_isTISubset [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hnoV : ¬ ∃ M : Subgroup G, M ∈ maximalSubgroups G ∧ OddOrder.GroupTheory.IsTypeV M)
    {ctr : CounterexampleHypothesis (G := G)} (data : RankTwoWitnessData ctr)
    (frob : TypeIFrobeniusData data.L) :
    ¬ OddOrder.GroupTheory.IsTISubset
        (OddOrder.GroupTheory.sharpSubgroup frob.typeI.typeF.H)
        (Subgroup.normalizer (frob.typeI.typeF.H : Set G)) :=
  witness_H_sharp_not_isTISubset_of_typeI hG hnoV data frob.typeI

/-- **Peterfalvi (12.1) for the witness subgroup `L`, with its Frobenius witness**: the second
maximal subgroup `L` of (12.9) carries the (12.1) Hypothesis together with an explicit Frobenius
decomposition of its kernel `H = L_F`.  Since `L` is type I (Frobenius, by (12.10)
`witness_L_frobenius`), `hypothesis_of_typeIData` applied to the recovered `TypeIData` yields the
Hypothesis whose `typeI` is that very data, so the Frobenius group structure `frob.frobenius`
transfers to `hyp.H`.  This Frobenius witness is the structural input to coherence (12.6). -/
theorem witness_L_hypothesis_frobenius [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hnoV : ¬ ∃ M : Subgroup G, M ∈ maximalSubgroups G ∧ OddOrder.GroupTheory.IsTypeV M)
    {ctr : CounterexampleHypothesis (G := G)} (data : RankTwoWitnessData ctr) :
    ∃ hyp : Hypothesis data.L, ∃ C : Subgroup ↥data.L,
      OddOrder.Isaacs.Ch06.IsFrobeniusGroup ↥data.L (hyp.H.subgroupOf data.L) C ∧
      ¬ OddOrder.GroupTheory.IsTISubset
          (OddOrder.GroupTheory.sharpSubgroup hyp.typeI.typeF.H)
          (Subgroup.normalizer (hyp.typeI.typeF.H : Set G)) := by
  obtain ⟨frob, hker⟩ := witness_L_frobenius hG hnoV data
  obtain ⟨hyp, hhyp⟩ := hypothesis_of_typeIData hG data.L_maximal frob.typeI
  have hH : hyp.typeI.typeF.H = frob.typeI.typeF.H := by rw [hhyp]
  refine ⟨hyp, frob.complement, ?_, ?_⟩
  · rw [show hyp.H = hyp.typeI.typeF.H from rfl, hH]
    exact frob.frobenius
  · rw [hH]
    exact witness_H_sharp_not_isTISubset hG hnoV data frob

end OddOrder.Peterfalvi.S14
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/Peterfalvi/S14_MaximalI/WitnessFrobenius.lean

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me