Peterfalvi §14: minimal-counterexample representation bounds
DefinitionOddOrder_Peterfalvi_S14_MaximalI_MinimalCounterexamplefeit-thompsonfinite-groupsgroup-theoryishida-odd-order
Formalizes (12.12) and its minimal-counterexample consequences (Peterfalvi, Character Theory for the Odd Order Theorem, Section 14, pp. 69-74). The bundle consists of lemmas only.
The bundle also proves 15 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/MinimalCounterexample.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_WitnessFrobenius
namespace OddOrder.Peterfalvi.S14
open OddOrder.GroupTheory
open OddOrder.RepresentationTheory
open OddOrder.Isaacs
open scoped Pointwise
variable {G : Type*} [Group G]
/-- **(12.12) Case A core.** A finite group `E` acting faithfully on a one-dimensional
`𝔽_p`-space `V` is cyclic, with `|E| ∣ |V| - 1 = p - 1`. This is the reducible / rank-one case
of Peterfalvi (12.12): `End_{𝔽_p}(V) ≅ 𝔽_p` (every endomorphism of a line is a homothety), so
`E ↪ End(V)ˣ ≅ (ℤ/p)ˣ`, a cyclic group of order `p - 1`. -/
theorem isCyclic_and_card_dvd_of_faithful_one_dim
{p : ℕ} [Fact p.Prime] {E V : Type*} [Group E] [Finite E]
[AddCommGroup V] [Module (ZMod p) V] [Finite V]
(ρ : Representation (ZMod p) E V) (hfaith : Function.Injective ρ)
(hdim : Module.finrank (ZMod p) V = 1) :
IsCyclic E ∧ Nat.card E ∣ Nat.card V - 1 := by
classical
have : Nontrivial V := Module.nontrivial_of_finrank_eq_succ hdim
have : Module.Finite (ZMod p) V := Module.Finite.of_finite
-- `End_{𝔽_p}(V) ≅ 𝔽_p` via `algebraMap` (bijective in dimension one: every endo is `c • id`).
have hsurj : Function.Surjective (algebraMap (ZMod p) (Module.End (ZMod p) V)) := by
intro u
obtain ⟨c, hc, -⟩ := LinearMap.existsUnique_eq_smul_id_of_finrank_eq_one hdim u
exact ⟨c, by rw [Algebra.algebraMap_eq_smul_one, hc, Module.End.one_eq_id]⟩
have hinj : Function.Injective (algebraMap (ZMod p) (Module.End (ZMod p) V)) :=
(algebraMap (ZMod p) (Module.End (ZMod p) V)).injective
let eRing : ZMod p ≃+* Module.End (ZMod p) V := RingEquiv.ofBijective _ ⟨hinj, hsurj⟩
-- `E ↪ End(V)ˣ ≃ (ℤ/p)ˣ`.
let φ : E →* (ZMod p)ˣ :=
(Units.mapEquiv eRing.toMulEquiv).symm.toMonoidHom.comp (MonoidHom.toHomUnits ρ)
have hφinj : Function.Injective φ := by
intro a b hab
apply hfaith
have h1 : (MonoidHom.toHomUnits ρ) a = (MonoidHom.toHomUnits ρ) b :=
(Units.mapEquiv eRing.toMulEquiv).symm.injective (by simpa [φ] using hab)
simpa using congrArg (Units.val) h1
have : IsCyclic (ZMod p)ˣ := inferInstance
have : IsCyclic φ.range := inferInstance
have hcardV : Nat.card V = p := by
rw [Module.natCard_eq_pow_finrank (K := ZMod p), hdim, pow_one, Nat.card_eq_fintype_card,
ZMod.card]
refine ⟨isCyclic_of_surjective (MonoidHom.ofInjective hφinj).symm.toMonoidHom
(MonoidHom.ofInjective hφinj).symm.surjective, ?_⟩
rw [hcardV]
calc Nat.card E = Nat.card φ.range := Nat.card_congr (MonoidHom.ofInjective hφinj).toEquiv
_ ∣ Nat.card (ZMod p)ˣ := Subgroup.card_subgroup_dvd_card _
_ = p - 1 := by
rw [Nat.card_eq_fintype_card, ZMod.card_units_eq_totient, Nat.totient_prime (Fact.out)]
/-- **(12.12) irreducible-case core.** An odd-order group `E` acting faithfully and
irreducibly on a two-dimensional `𝔽_p`-space `V` (with `p ∤ |E|`) is cyclic, with
`|E| ∣ |V| - 1 = p² - 1`. This is the rank-two irreducible case of Peterfalvi (12.12):
BG Theorem 2.6(a) (`odd_two_dim_abelian`) abelianizes `E`, and the commutativity-free Singer
mechanism (`isCyclic_and_card_dvd_of_faithful_irreducible_comm`) then realizes `E` inside the
units of the Singer field `𝔽_p[E] ⧸ I ≅ 𝔽_{p²}`. -/
theorem isCyclic_and_card_dvd_of_odd_two_dim_irreducible
{p : ℕ} [Fact p.Prime] {E V : Type*} [Group E] [Finite E] (hodd : Odd (Nat.card E))
[AddCommGroup V] [Module (ZMod p) V] [Finite V]
(ρ : Representation (ZMod p) E V) (hfaith : Function.Injective ρ)
(hirr : Representation.IsIrreducible ρ)
(hdim : Module.finrank (ZMod p) V = 2) (hp_ndvd : ¬ p ∣ Nat.card E) :
IsCyclic E ∧ Nat.card E ∣ Nat.card V - 1 := by
classical
have : Module.Finite (ZMod p) V := Module.Finite.of_finite
-- BG 2.6(a): a faithful odd two-dimensional representation has abelian image.
have hchar : ∀ q : ℕ, q.Prime → q ∣ Nat.card E → ¬ CharP (ZMod p) q := fun q _ hqdvd hcharq =>
hp_ndvd ((CharP.eq (ZMod p) hcharq (ZMod.charP p)) ▸ hqdvd)
have hcomm : ∀ a b : E, a * b = b * a :=
(OddOrder.BG.Ch1.S02.odd_two_dim_abelian hodd hdim ρ hfaith hchar).comm
-- Give `V` the `𝔽_p[E]`-module structure of the representation *directly* (this is
-- definitionally `ρ.asModule`'s instance, but stated on `V` so that instance synthesis does
-- not choke on the `ρ.asModule` notation — which it does once `IsIrreducible ρ` is around).
let : Module (MonoidAlgebra (ZMod p) E) V := Module.compHom V (ρ.asAlgebraHom).toRingHom
have hsmul : ∀ (e : E) (x : V), MonoidAlgebra.of (ZMod p) E e • x = ρ e x := fun e x => by
change (ρ.asAlgebraHom) (MonoidAlgebra.of (ZMod p) E e) x = ρ e x
rw [Representation.asAlgebraHom_of]
have : IsSimpleModule (MonoidAlgebra (ZMod p) E) V :=
(Representation.irreducible_iff_isSimpleModule_asModule ρ).mp hirr
have hfaith' : ∀ e : E, (∀ x : V, MonoidAlgebra.of (ZMod p) E e • x = x) → e = 1 := by
intro e he
apply hfaith
ext v
rw [map_one, Module.End.one_apply, ← hsmul e v]
exact he v
exact isCyclic_and_card_dvd_of_faithful_irreducible_comm (M := V) hcomm hfaith'
/-- **(12.12) `p + 1` refinement, irreducible case.** An odd-order group `E` (`p ∤ |E|`) acting
faithfully and irreducibly on a two-dimensional `𝔽_p`-space `V`, with **no nontrivial element
acting as an `𝔽_p`-scalar** (`hnonscalar`), is cyclic with `|E| ∣ p + 1`.
This is the rank-two refinement of Peterfalvi (12.12): the plain irreducible core
(`isCyclic_and_card_dvd_of_odd_two_dim_irreducible`) only bounds `|E| ∣ p² - 1`. The Singer
realization places `E` inside the cyclic group `𝔽_{p²}ˣ` (order `p² - 1`), where the non-scalar
hypothesis makes it meet the scalar subgroup `𝔽_pˣ` (order `p - 1`) trivially, so
`coprime_card_sub_one_of_faithful_irreducible_comm_nonscalar` gives `Coprime |E| (p - 1)`. Together
with `|E| ∣ p² - 1 = (p - 1)(p + 1)`, coprimality to the first factor forces `|E| ∣ p + 1`. -/
theorem isCyclic_and_card_dvd_add_one_of_two_dim_irreducible_nonscalar
{p : ℕ} [Fact p.Prime] {E V : Type*} [Group E] [Finite E] (hodd : Odd (Nat.card E))
[AddCommGroup V] [Module (ZMod p) V] [Finite V]
(ρ : Representation (ZMod p) E V) (hfaith : Function.Injective ρ)
(hirr : Representation.IsIrreducible ρ)
(hdim : Module.finrank (ZMod p) V = 2) (hp_ndvd : ¬ p ∣ Nat.card E)
(hnonscalar : ∀ e : E, (∃ n : ℕ, ∀ x : V, ρ e x = n • x) → e = 1) :
IsCyclic E ∧ Nat.card E ∣ p + 1 := by
classical
have : Module.Finite (ZMod p) V := Module.Finite.of_finite
-- `|E| ∣ p² - 1` and cyclicity from the irreducible core.
obtain ⟨hcyc, hdvd_sq⟩ :=
isCyclic_and_card_dvd_of_odd_two_dim_irreducible hodd ρ hfaith hirr hdim hp_ndvd
refine ⟨hcyc, ?_⟩
have hcardV : Nat.card V = p ^ 2 := by
rw [Module.natCard_eq_pow_finrank (K := ZMod p), hdim, Nat.card_eq_fintype_card, ZMod.card]
rw [hcardV] at hdvd_sq
-- Singer non-scalar core ⟹ `Coprime |E| (p - 1)`. Reuse the `𝔽ₚ[E]`-module setup of the core.
have hchar : ∀ q : ℕ, q.Prime → q ∣ Nat.card E → ¬ CharP (ZMod p) q := fun q _ hqdvd hcharq =>
hp_ndvd ((CharP.eq (ZMod p) hcharq (ZMod.charP p)) ▸ hqdvd)
have hcomm : ∀ a b : E, a * b = b * a :=
(OddOrder.BG.Ch1.S02.odd_two_dim_abelian hodd hdim ρ hfaith hchar).comm
let : Module (MonoidAlgebra (ZMod p) E) V := Module.compHom V (ρ.asAlgebraHom).toRingHom
have hsmul : ∀ (e : E) (x : V), MonoidAlgebra.of (ZMod p) E e • x = ρ e x := fun e x => by
change (ρ.asAlgebraHom) (MonoidAlgebra.of (ZMod p) E e) x = ρ e x
rw [Representation.asAlgebraHom_of]
have : IsSimpleModule (MonoidAlgebra (ZMod p) E) V :=
(Representation.irreducible_iff_isSimpleModule_asModule ρ).mp hirr
have hfaith' : ∀ e : E, (∀ x : V, MonoidAlgebra.of (ZMod p) E e • x = x) → e = 1 := by
intro e he
apply hfaith
ext v
rw [map_one, Module.End.one_apply, ← hsmul e v]
exact he v
have hns' : ∀ e : E,
(∃ n : ℕ, ∀ x : V, MonoidAlgebra.of (ZMod p) E e • x = n • x) → e = 1 := by
rintro e ⟨n, hn⟩
exact hnonscalar e ⟨n, fun x => by rw [← hsmul e x]; exact hn x⟩
have hcop : Nat.Coprime (Nat.card E) (p - 1) :=
OddOrder.RepresentationTheory.coprime_card_sub_one_of_faithful_irreducible_comm_nonscalar
hcomm hfaith' hns'
-- `|E| ∣ (p - 1)(p + 1) = p² - 1` and `Coprime |E| (p - 1)` force `|E| ∣ p + 1`.
have hpq : (p - 1) * (p + 1) = p ^ 2 - 1 := by
obtain ⟨n, rfl⟩ : ∃ n, p = n + 2 := ⟨p - 2, by have := (Fact.out (p := p.Prime)).two_le; omega⟩
change (n + 1) * (n + 3) = (n + 2) ^ 2 - 1
have hexp : (n + 2) ^ 2 = (n + 1) * (n + 3) + 1 := by ring
omega
rw [← hpq] at hdvd_sq
exact hcop.dvd_of_dvd_mul_left hdvd_sq
/-- **(12.12) rep-theory core (dichotomy form).** A finite odd-order group `E` (`p ∤ |E|`)
acting **fixed-point-freely** on an `𝔽_p`-space `V` of dimension `1` or `2` is **cyclic**, and
either `|E| ∣ p − 1`, or the action is `2`-dimensional **irreducible** with `|E| ∣ p² − 1`.
Dim 1 and the reducible dim-2 case (an `E`-invariant line) go through Case A
(`isCyclic_and_card_dvd_of_faithful_one_dim`); irreducible dim 2 is Case B
(`isCyclic_and_card_dvd_of_odd_two_dim_irreducible`). The FPF hypothesis makes `E` faithful on
every nonzero invariant subspace. The dichotomy (rather than the combined `|E| ∣ |V| − 1`)
retains the irreducibility needed by the (12.12) `p + 1` refinement
(`isCyclic_and_card_dvd_add_one_of_two_dim_irreducible_nonscalar`). -/
theorem isCyclic_and_card_dvd_dichotomy_of_fpf_dim_le_two
{p : ℕ} [Fact p.Prime] {E V : Type*} [Group E] [Finite E] (hodd : Odd (Nat.card E))
[AddCommGroup V] [Module (ZMod p) V] [Finite V]
(ρ : Representation (ZMod p) E V)
(hfpf : ∀ e : E, e ≠ 1 → ∀ v : V, ρ e v = v → v = 0)
(hdim : Module.finrank (ZMod p) V = 1 ∨ Module.finrank (ZMod p) V = 2)
(hp_ndvd : ¬ p ∣ Nat.card E) :
IsCyclic E ∧ (Nat.card E ∣ p - 1 ∨
(Module.finrank (ZMod p) V = 2 ∧ Representation.IsIrreducible ρ ∧
Nat.card E ∣ p ^ 2 - 1)) := by
classical
have : Module.Finite (ZMod p) V := Module.Finite.of_finite
-- Step 1: the FPF hypothesis makes `ρ` faithful.
-- `V` is nontrivial because `finrank V ≥ 1`, so it has a nonzero vector; a nontrivial element in
-- the kernel would fix that vector, contradicting `hfpf`.
have hVnt : Nontrivial V := by
rcases hdim with h | h
· exact Module.nontrivial_of_finrank_eq_succ h
· exact Module.nontrivial_of_finrank_eq_succ (n := 1) (by rw [h])
have hfaith : Function.Injective ρ := by
intro a b hab
by_contra hne
have hba : b⁻¹ * a ≠ 1 := fun h => hne (inv_mul_eq_one.mp h).symm
obtain ⟨v, hv⟩ := exists_ne (0 : V)
refine hv (hfpf (b⁻¹ * a) hba v ?_)
rw [map_mul, Module.End.mul_apply, hab, ← Module.End.mul_apply, ← map_mul, inv_mul_cancel,
map_one, Module.End.one_apply]
rcases hdim with hd1 | hd2
· -- dim 1: Case A; `|V| = p`.
obtain ⟨hcyc, hdvd⟩ := isCyclic_and_card_dvd_of_faithful_one_dim ρ hfaith hd1
have hcardV : Nat.card V = p := by
rw [Module.natCard_eq_pow_finrank (K := ZMod p), hd1, pow_one, Nat.card_eq_fintype_card,
ZMod.card]
exact ⟨hcyc, Or.inl (by rwa [hcardV] at hdvd)⟩
· -- dim 2.
by_cases hirr : Representation.IsIrreducible ρ
· -- irreducible: Case B; `|V| = p²`.
obtain ⟨hcyc, hdvd⟩ :=
isCyclic_and_card_dvd_of_odd_two_dim_irreducible hodd ρ hfaith hirr hd2 hp_ndvd
have hcardV : Nat.card V = p ^ 2 := by
rw [Module.natCard_eq_pow_finrank (K := ZMod p), hd2, Nat.card_eq_fintype_card, ZMod.card]
exact ⟨hcyc, Or.inr ⟨hd2, hirr, by rwa [hcardV] at hdvd⟩⟩
· -- reducible: a proper nonzero invariant line `W` exists; Case A on `W.toRepresentation`.
have hbnt : (⊥ : Subrepresentation ρ) ≠ ⊤ := fun h =>
bot_ne_top (congrArg Subrepresentation.toSubmodule h)
have : Nontrivial (Subrepresentation ρ) := ⟨⊥, ⊤, hbnt⟩
have hnotall : ¬ ∀ W : Subrepresentation ρ, W = ⊥ ∨ W = ⊤ := fun H =>
hirr { eq_bot_or_eq_top := H }
push Not at hnotall
obtain ⟨W, hWbot, hWtop⟩ := hnotall
-- `W` is a proper nonzero subrepresentation; its submodule has `finrank = 1`.
have hWsub_bot : W.toSubmodule ≠ ⊥ := fun h =>
hWbot (Subrepresentation.toSubmodule_injective (by rw [h]; rfl))
have hWsub_top : W.toSubmodule ≠ ⊤ := fun h =>
hWtop (Subrepresentation.toSubmodule_injective (by rw [h]; rfl))
have : Finite ↥W.toSubmodule := Subtype.finite
have : Module.Finite (ZMod p) ↥W.toSubmodule := Module.Finite.of_finite
have hpos : 0 < Module.finrank (ZMod p) ↥W.toSubmodule := by
have := Submodule.finrank_lt_finrank_of_lt (s := (⊥ : Submodule (ZMod p) V))
(t := W.toSubmodule) (lt_of_le_of_ne bot_le (Ne.symm hWsub_bot))
simpa using this
have hlt : Module.finrank (ZMod p) ↥W.toSubmodule < Module.finrank (ZMod p) V :=
Submodule.finrank_lt hWsub_top
have hWdim : Module.finrank (ZMod p) ↥W.toSubmodule = 1 := by
rw [hd2] at hlt; omega
-- faithfulness of `W.toRepresentation` from `hfpf` restricted to `W`.
have hfaithW : Function.Injective W.toRepresentation := by
intro a b hab
by_contra hne
have hba : b⁻¹ * a ≠ 1 := fun h => hne (inv_mul_eq_one.mp h).symm
-- every element of `W` is fixed by `ρ (b⁻¹ a)`, hence is `0` by `hfpf`; so `W = ⊥`.
apply hWsub_bot
rw [Submodule.eq_bot_iff]
intro w hw
have hfix : W.toRepresentation (b⁻¹ * a) ⟨w, hw⟩ = ⟨w, hw⟩ := by
rw [map_mul, Module.End.mul_apply, hab, ← Module.End.mul_apply, ← map_mul,
inv_mul_cancel, map_one, Module.End.one_apply]
have hfixV : ρ (b⁻¹ * a) w = w := by
have := congrArg Subtype.val hfix
simpa [Subrepresentation.toRepresentation, LinearMap.coe_restrict_apply] using this
exact hfpf (b⁻¹ * a) hba w hfixV
-- Case A on `W.toRepresentation` gives `IsCyclic E ∧ |E| ∣ p - 1`.
obtain ⟨hcyc, hdvd⟩ :=
isCyclic_and_card_dvd_of_faithful_one_dim W.toRepresentation hfaithW hWdim
have hcardW : Nat.card ↥W.toSubmodule = p := by
rw [Module.natCard_eq_pow_finrank (K := ZMod p), hWdim, pow_one, Nat.card_eq_fintype_card,
ZMod.card]
exact ⟨hcyc, Or.inl (by rwa [hcardW] at hdvd)⟩
/-- **(12.12) rep-theory core (combined).** A finite odd-order group `E` (`p ∤ |E|`) acting
**fixed-point-freely** (no nonzero vector is fixed by a nontrivial element) on an `𝔽_p`-space `V`
of dimension `1` or `2` is **cyclic**, with `|E| ∣ |V| - 1`. Forgetful form of the dichotomy
`isCyclic_and_card_dvd_dichotomy_of_fpf_dim_le_two` (`p − 1` divides `|V| − 1` in both dims). -/
theorem isCyclic_and_card_dvd_of_fpf_dim_le_two
{p : ℕ} [Fact p.Prime] {E V : Type*} [Group E] [Finite E] (hodd : Odd (Nat.card E))
[AddCommGroup V] [Module (ZMod p) V] [Finite V]
(ρ : Representation (ZMod p) E V)
(hfpf : ∀ e : E, e ≠ 1 → ∀ v : V, ρ e v = v → v = 0)
(hdim : Module.finrank (ZMod p) V = 1 ∨ Module.finrank (ZMod p) V = 2)
(hp_ndvd : ¬ p ∣ Nat.card E) :
IsCyclic E ∧ Nat.card E ∣ Nat.card V - 1 := by
classical
have : Module.Finite (ZMod p) V := Module.Finite.of_finite
obtain ⟨hcyc, hdvd⟩ :=
isCyclic_and_card_dvd_dichotomy_of_fpf_dim_le_two hodd ρ hfpf hdim hp_ndvd
refine ⟨hcyc, ?_⟩
have hsub_dvd : (p - 1 : ℕ) ∣ p ^ 2 - 1 := by
have hp1 : 1 ≤ p := (Fact.out (p := p.Prime)).one_le
obtain ⟨k, rfl⟩ : ∃ k, p = k + 1 := ⟨p - 1, by omega⟩
refine ⟨k + 2, ?_⟩
have hsq : (k + 1) ^ 2 = k * (k + 2) + 1 := by ring
rw [hsq, Nat.add_sub_cancel, Nat.add_sub_cancel]
rcases hdim with hd1 | hd2
· have hcardV : Nat.card V = p := by
rw [Module.natCard_eq_pow_finrank (K := ZMod p), hd1, pow_one, Nat.card_eq_fintype_card,
ZMod.card]
rw [hcardV]
rcases hdvd with h1 | ⟨hd2, -, -⟩
· exact h1
· rw [hd1] at hd2; omega
· have hcardV : Nat.card V = p ^ 2 := by
rw [Module.natCard_eq_pow_finrank (K := ZMod p), hd2, Nat.card_eq_fintype_card, ZMod.card]
rw [hcardV]
rcases hdvd with h1 | ⟨-, -, hsq⟩
· exact h1.trans hsub_dvd
· exact hsq
/-- **(12.12) rep-theory bridge (abstract `MulDistribMulAction` form).** A finite odd-order group
`E` (`p ∤ |E|`) acting **fixed-point-freely** on an elementary abelian `p`-group `M` — encoded by
a `ZMod p`-module structure on `Additive M` — of `𝔽_p`-dimension `1` or `2` is **cyclic**, with
`|E| ∣ |M| - 1`.
This lifts `isCyclic_and_card_dvd_of_fpf_dim_le_two` from an abstract `Representation` to a
`MulDistribMulAction` (via `Representation.ofDistribMulAction`), which is the form that a
conjugation action of `E` on an elementary abelian subgroup supplies. -/
theorem isCyclic_and_card_dvd_of_fpf_mulDistribMulAction
{p : ℕ} [Fact p.Prime] {E M : Type*} [Group E] [Finite E] (hodd : Odd (Nat.card E))
[CommGroup M] [Finite M] [Module (ZMod p) (Additive M)] [MulDistribMulAction E M]
(hp_ndvd : ¬ p ∣ Nat.card E)
(hdim : Module.finrank (ZMod p) (Additive M) = 1 ∨
Module.finrank (ZMod p) (Additive M) = 2)
(hfpf : ∀ e : E, e ≠ 1 → ∀ m : M, e • m = m → m = 1) :
IsCyclic E ∧ Nat.card E ∣ Nat.card M - 1 := by
classical
have : Finite (Additive M) := inferInstanceAs (Finite M)
-- The fixed-point-free hypothesis, transported to the additive representation `ρ = e ↦ (e • ·)`.
have hfpf' : ∀ e : E, e ≠ 1 → ∀ v : Additive M,
(Representation.ofDistribMulAction (ZMod p) E (Additive M)) e v = v → v = 0 := by
intro e he v hv
rw [Representation.ofDistribMulAction_apply_apply] at hv
-- `e • v = Additive.ofMul (e • v.toMul)` definitionally; pass to the multiplicative action.
change Additive.ofMul (e • Additive.toMul v) = v at hv
have hev : e • Additive.toMul v = Additive.toMul v := by
have := congrArg Additive.toMul hv; simpa using this
have hm1 : Additive.toMul v = 1 := hfpf e he _ hev
exact Additive.toMul.injective (by simp [hm1])
obtain ⟨hcyc, hdvd⟩ := isCyclic_and_card_dvd_of_fpf_dim_le_two hodd
(Representation.ofDistribMulAction (ZMod p) E (Additive M)) hfpf' hdim hp_ndvd
exact ⟨hcyc, by rwa [Nat.card_congr (Additive.toMul (α := M))] at hdvd⟩
/-- **(12.12) rep-theory bridge (conjugation form).** Let a finite group `E` normalize an
elementary abelian `p`-subgroup `T` of order `p` or `p²` of `G`, with `|E|` odd and coprime to `p`,
and let `E` act **fixed-point-freely on `T` by conjugation** (no nontrivial element of `T` is fixed
by a nontrivial element of `E`). Then `E` is **cyclic** and `|E|` divides `p - 1` or `p² - 1`.
This is the `§8`-free structural core of Peterfalvi (12.12): there `T = Ω₁(Z(O_p(H)))` is the
rank `≤ 2` elementary abelian subgroup and `E` the Frobenius complement of `L`, acting FPF on `T`
by (12.10). The `p + 1` refinement of (12.12) is separate (it consumes (12.9)/(12.11)). -/
theorem isCyclic_and_card_dvd_of_fpf_conj_elemAbelian
{G : Type*} [Group G] [Finite G] {p : ℕ} [Fact p.Prime]
{T E : Subgroup G} (hT : IsElementaryAbelian p ↥T)
(hEnorm : E ≤ Subgroup.normalizer (T : Set G))
(hodd : Odd (Nat.card ↥E)) (hp_ndvd : ¬ p ∣ Nat.card ↥E)
(hT_card : Nat.card ↥T = p ∨ Nat.card ↥T = p ^ 2)
(hfpf : ∀ e : G, e ∈ E → e ≠ 1 → ∀ t : G, t ∈ T → e * t * e⁻¹ = t → t = 1) :
IsCyclic ↥E ∧ (Nat.card ↥E ∣ p - 1 ∨ Nat.card ↥E ∣ p ^ 2 - 1) := by
classical
let : CommGroup ↥T := hT.subgroupCommGroup
let : Module (ZMod p) (Additive ↥T) := hT.subgroupZmodModule
let act : MulDistribMulAction ↥E ↥T :=
MulDistribMulAction.compHom (M := ↥(Subgroup.normalizer (T : Set G))) ↥T
(Subgroup.inclusion hEnorm)
-- The conjugation action's coercion: `(ε • τ : G) = ε * τ * ε⁻¹`.
have hsmul_coe : ∀ (ε : ↥E) (τ : ↥T), ((ε • τ : ↥T) : G) = (ε : G) * (τ : G) * (ε : G)⁻¹ :=
fun _ _ => rfl
-- `dim_{𝔽_p} (Additive T) ∈ {1, 2}` from `|T| ∈ {p, p²}`.
have hcard_pow : p ^ Module.finrank (ZMod p) (Additive ↥T) = Nat.card ↥T := by
rw [FiniteField.pow_finrank_eq_natCard p (Additive ↥T),
Nat.card_congr (Additive.toMul (α := ↥T))]
have h2le := (Fact.out (p := p.Prime)).two_le
have hdim : Module.finrank (ZMod p) (Additive ↥T) = 1 ∨
Module.finrank (ZMod p) (Additive ↥T) = 2 := by
rcases hT_card with h | h
· have e1 : p ^ Module.finrank (ZMod p) (Additive ↥T) = p ^ 1 := by rw [hcard_pow, h, pow_one]
exact Or.inl (Nat.pow_right_injective h2le e1)
· have e2 : p ^ Module.finrank (ZMod p) (Additive ↥T) = p ^ 2 := by rw [hcard_pow, h]
exact Or.inr (Nat.pow_right_injective h2le e2)
-- The conjugation FPF hypothesis, transported to the `MulDistribMulAction` form.
have hfpf' : ∀ ε : ↥E, ε ≠ 1 → ∀ τ : ↥T, ε • τ = τ → τ = 1 := by
intro ε hεne τ hτ
have hc : (ε : G) * (τ : G) * (ε : G)⁻¹ = (τ : G) := by
rw [← hsmul_coe]; exact congrArg Subtype.val hτ
exact OneMemClass.coe_eq_one.mp
(hfpf (ε : G) ε.2 (mt OneMemClass.coe_eq_one.mp hεne) (τ : G) τ.2 hc)
obtain ⟨hcyc, hdvd⟩ :=
isCyclic_and_card_dvd_of_fpf_mulDistribMulAction hodd hp_ndvd hdim hfpf'
refine ⟨hcyc, ?_⟩
rcases hT_card with h | h
· exact Or.inl (by rwa [h] at hdvd)
· exact Or.inr (by rwa [h] at hdvd)
/-- **(12.12) rep-theory bridge (abstract `MulDistribMulAction` form), `p ± 1` refinement.**
As in `isCyclic_and_card_dvd_of_fpf_mulDistribMulAction`, but with the **non-scalar** input in
the rank-two case — no nontrivial `e : E` acts on `M` as a uniform power `m ↦ m ^ n` — which
upgrades the rank-two branch to `|E| ∣ p + 1` (Singer,
`isCyclic_and_card_dvd_add_one_of_two_dim_irreducible_nonscalar`); the rank-one and reducible
branches give `|E| ∣ p − 1` outright (dichotomy core,
`isCyclic_and_card_dvd_dichotomy_of_fpf_dim_le_two`). -/
theorem isCyclic_and_card_dvd_sub_or_add_one_of_fpf_mulDistribMulAction
{p : ℕ} [Fact p.Prime] {E M : Type*} [Group E] [Finite E] (hodd : Odd (Nat.card E))
[CommGroup M] [Finite M] [Module (ZMod p) (Additive M)] [MulDistribMulAction E M]
(hp_ndvd : ¬ p ∣ Nat.card E)
(hdim : Module.finrank (ZMod p) (Additive M) = 1 ∨
Module.finrank (ZMod p) (Additive M) = 2)
(hfpf : ∀ e : E, e ≠ 1 → ∀ m : M, e • m = m → m = 1)
(hnonscalar : Module.finrank (ZMod p) (Additive M) = 2 →
∀ e : E, (∃ n : ℕ, ∀ m : M, e • m = m ^ n) → e = 1) :
IsCyclic E ∧ (Nat.card E ∣ p - 1 ∨ Nat.card E ∣ p + 1) := by
classical
have : Finite (Additive M) := inferInstanceAs (Finite M)
-- The FPF hypothesis, transported to the additive representation `ρ = e ↦ (e • ·)`.
have hfpf' : ∀ e : E, e ≠ 1 → ∀ v : Additive M,
(Representation.ofDistribMulAction (ZMod p) E (Additive M)) e v = v → v = 0 := by
intro e he v hv
rw [Representation.ofDistribMulAction_apply_apply] at hv
change Additive.ofMul (e • Additive.toMul v) = v at hv
have hev : e • Additive.toMul v = Additive.toMul v := by
have := congrArg Additive.toMul hv; simpa using this
have hm1 : Additive.toMul v = 1 := hfpf e he _ hev
exact Additive.toMul.injective (by simp [hm1])
obtain ⟨hcyc, hdvd⟩ := isCyclic_and_card_dvd_dichotomy_of_fpf_dim_le_two hodd
(Representation.ofDistribMulAction (ZMod p) E (Additive M)) hfpf' hdim hp_ndvd
refine ⟨hcyc, ?_⟩
rcases hdvd with h1 | ⟨hd2, hirr, -⟩
· exact Or.inl h1
· -- Irreducible rank-two branch: supply the non-scalar input and apply Singer.
have hfaith :
Function.Injective (Representation.ofDistribMulAction (ZMod p) E (Additive M)) := by
intro a b hab
by_contra hne
have hba : b⁻¹ * a ≠ 1 := fun h => hne (inv_mul_eq_one.mp h).symm
have : Nontrivial (Additive M) :=
Module.nontrivial_of_finrank_eq_succ (n := 1) (by rw [hd2])
obtain ⟨v, hv⟩ := exists_ne (0 : Additive M)
refine hv (hfpf' (b⁻¹ * a) hba v ?_)
rw [map_mul, Module.End.mul_apply, hab, ← Module.End.mul_apply, ← map_mul,
inv_mul_cancel, map_one, Module.End.one_apply]
have hns : ∀ e : E, (∃ n : ℕ, ∀ v : Additive M,
(Representation.ofDistribMulAction (ZMod p) E (Additive M)) e v = n • v) → e = 1 := by
rintro e ⟨n, hn⟩
refine hnonscalar hd2 e ⟨n, fun m => ?_⟩
have h1 := hn (Additive.ofMul m)
rw [Representation.ofDistribMulAction_apply_apply] at h1
have := congrArg Additive.toMul h1
rw [show Additive.toMul (e • Additive.ofMul m) = e • m from rfl] at this
simpa using this
obtain ⟨-, hp1⟩ :=
isCyclic_and_card_dvd_add_one_of_two_dim_irreducible_nonscalar hodd
(Representation.ofDistribMulAction (ZMod p) E (Additive M)) hfaith hirr hd2 hp_ndvd hns
exact Or.inr hp1
/-- **(12.12) rep-theory bridge (conjugation form), `p ± 1` refinement.** As in
`isCyclic_and_card_dvd_of_fpf_conj_elemAbelian`, but with the **non-scalar** input — no
nontrivial `e ∈ E` conjugates every `t ∈ T` to the same power `t^n` — which upgrades the
`|T| = p²` branch to `|E| ∣ p + 1` (Singer,
`isCyclic_and_card_dvd_add_one_of_two_dim_irreducible_nonscalar`); the reducible/`|T| = p`
branches give `|E| ∣ p − 1` outright (dichotomy core). -/
theorem isCyclic_and_card_dvd_sub_or_add_one_of_fpf_conj_elemAbelian
{G : Type*} [Group G] [Finite G] {p : ℕ} [Fact p.Prime]
{T E : Subgroup G} (hT : IsElementaryAbelian p ↥T)
(hEnorm : E ≤ Subgroup.normalizer (T : Set G))
(hodd : Odd (Nat.card ↥E)) (hp_ndvd : ¬ p ∣ Nat.card ↥E)
(hT_card : Nat.card ↥T = p ∨ Nat.card ↥T = p ^ 2)
(hfpf : ∀ e : G, e ∈ E → e ≠ 1 → ∀ t : G, t ∈ T → e * t * e⁻¹ = t → t = 1)
(hnonscalar : Nat.card ↥T = p ^ 2 →
∀ e : G, e ∈ E → (∃ n : ℕ, ∀ t : G, t ∈ T → e * t * e⁻¹ = t ^ n) → e = 1) :
IsCyclic ↥E ∧ (Nat.card ↥E ∣ p - 1 ∨ Nat.card ↥E ∣ p + 1) := by
classical
let : CommGroup ↥T := hT.subgroupCommGroup
let : Module (ZMod p) (Additive ↥T) := hT.subgroupZmodModule
let act : MulDistribMulAction ↥E ↥T :=
MulDistribMulAction.compHom (M := ↥(Subgroup.normalizer (T : Set G))) ↥T
(Subgroup.inclusion hEnorm)
-- The conjugation action's coercion: `(ε • τ : G) = ε * τ * ε⁻¹`.
have hsmul_coe : ∀ (ε : ↥E) (τ : ↥T), ((ε • τ : ↥T) : G) = (ε : G) * (τ : G) * (ε : G)⁻¹ :=
fun _ _ => rfl
-- `dim_{𝔽_p} (Additive T) ∈ {1, 2}` from `|T| ∈ {p, p²}`.
have hcard_pow : p ^ Module.finrank (ZMod p) (Additive ↥T) = Nat.card ↥T := by
rw [FiniteField.pow_finrank_eq_natCard p (Additive ↥T),
Nat.card_congr (Additive.toMul (α := ↥T))]
have h2le := (Fact.out (p := p.Prime)).two_le
have hdim : Module.finrank (ZMod p) (Additive ↥T) = 1 ∨
Module.finrank (ZMod p) (Additive ↥T) = 2 := by
rcases hT_card with h | h
· have e1 : p ^ Module.finrank (ZMod p) (Additive ↥T) = p ^ 1 := by rw [hcard_pow, h, pow_one]
exact Or.inl (Nat.pow_right_injective h2le e1)
· have e2 : p ^ Module.finrank (ZMod p) (Additive ↥T) = p ^ 2 := by rw [hcard_pow, h]
exact Or.inr (Nat.pow_right_injective h2le e2)
-- The conjugation FPF hypothesis, transported to the `MulDistribMulAction` form.
have hfpfM : ∀ ε : ↥E, ε ≠ 1 → ∀ τ : ↥T, ε • τ = τ → τ = 1 := by
intro ε hεne τ hτ
have hc : (ε : G) * (τ : G) * (ε : G)⁻¹ = (τ : G) := by
rw [← hsmul_coe]; exact congrArg Subtype.val hτ
exact OneMemClass.coe_eq_one.mp
(hfpf (ε : G) ε.2 (mt OneMemClass.coe_eq_one.mp hεne) (τ : G) τ.2 hc)
-- The conjugation non-scalar hypothesis, transported to the `MulDistribMulAction` form.
have hnsM : Module.finrank (ZMod p) (Additive ↥T) = 2 →
∀ ε : ↥E, (∃ n : ℕ, ∀ τ : ↥T, ε • τ = τ ^ n) → ε = 1 := by
rintro hd2 ε ⟨n, hn⟩
have hcardT : Nat.card ↥T = p ^ 2 := by rw [← hcard_pow, hd2]
refine Subtype.ext (hnonscalar hcardT (ε : G) ε.2 ⟨n, fun t ht => ?_⟩)
have hcoe := congrArg Subtype.val (hn ⟨t, ht⟩)
rw [hsmul_coe] at hcoe
simpa using hcoe
exact isCyclic_and_card_dvd_sub_or_add_one_of_fpf_mulDistribMulAction hodd hp_ndvd hdim
hfpfM hnsM
/-- **Peterfalvi (12.12), the `p + 1` refinement**: if the witness Frobenius complement's order
`e = |E|` divides `p² − 1`, then it divides `p − 1` or `p + 1`.
Peterfalvi's argument (the second half of the (12.12) proof): `E` is cyclic acting on
`T = Ω₁(P₀)` of order `p²`; identifying `T ⋊ E ↪ 𝔽_{p²} ⋊ 𝔽_{p²}^*` (Schur, as in (9.7.b)), the
subgroup `A ≤ E` of order `gcd(e, p−1)` lands in `𝔽_p^* `, so it normalizes every order-`p`
subgroup of `T` — in particular `⟨x⟩` for the (12.9) witness `x ∈ T`. Then `A ⊆ N_G(⟨x⟩) ⊆ M`
by (12.9), so `A ⊆ M ∩ L ⊆ H` by (12.11), while `A ≤ E` meets `H` trivially — `A = 1`. Hence
`gcd(e, p−1) = 1` and `e ∣ p + 1`.
**Assembly** (proven): instead of the Singer-cyclic identification, the rep-theory core
`isCyclic_and_card_dvd_sub_or_add_one_of_fpf_conj_elemAbelian` reduces the dichotomy to the
**nonscalar** hypothesis — no `e ∈ E` acts as a power map `t ↦ tⁿ` on a `T` of order `p²` —
which is exactly Peterfalvi's `A = 1` argument, elementwise: a scalar-acting `e` normalizes
every subgroup of `T`; the bookkeeping `|T| = p²` forces `T = Ω₁(P₀) ∋ x` (both are elementary
abelian sandwiched in the rank-`2` abelian `P₀`, so `T ⊆ Ω₁(P₀)` with `|Ω₁(P₀)| ≤ p² = |T|`),
so `e ∈ N_G(⟨x⟩) ⊆ M` by (12.9), hence `e ∈ M ∩ L ⊆ H = L_F` by (12.11)
(`intersection_le_kernel`), and the Frobenius disjointness `H ∩ E = 1` gives `e = 1`. -/
theorem witness_complement_dvd_p_sub_or_add_one [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)
{T : Subgroup G} (hTelem : T.IsElementaryAbelian ctr.p) (hTP0 : T ≤ ctr.P0) (hTne : T ≠ ⊥)
(hEnorm : frob.complement.map data.L.subtype ≤ Subgroup.normalizer (T : Set G))
(hfpf : ∀ e : G, e ∈ frob.complement.map data.L.subtype → e ≠ 1 →
∀ t : G, t ∈ T → e * t * e⁻¹ = t → t = 1) :
Nat.card ↥frob.complement ∣ ctr.p - 1 ∨ Nat.card ↥frob.complement ∣ ctr.p + 1 := by
classical
have : Fact ctr.p.Prime := ⟨ctr.p_prime⟩
have hEcard : Nat.card ↥(frob.complement.map data.L.subtype) = Nat.card ↥frob.complement :=
Subgroup.card_map_of_injective (K := frob.complement) data.L.subtype_injective
-- `|T| ∈ {p, p²}` from `T ≤ P₀` of rank `2`.
have hTcard : Nat.card ↥T = ctr.p ∨ Nat.card ↥T = ctr.p ^ 2 :=
OddOrder.GroupTheory.card_eq_prime_or_sq_of_isElementaryAbelian_le hTelem hTP0
(counterexample_P0_K_structure hG ctr).2.le hTne
-- Odd order of the realized complement `E' = E.map L.subtype`.
have hodd : Odd (Nat.card ↥(frob.complement.map data.L.subtype)) :=
hG.odd.of_dvd_nat (Subgroup.card_subgroup_dvd_card _)
-- `T ≤ H` through `P₀ ≤ L_F` (12.10).
have hHeq : frob.typeI.typeF.H = maxNilpotentNormalHall data.L := frob.typeI.typeF.H_eq
have hTH : T ≤ frob.typeI.typeF.H := by
refine hTP0.trans ?_
rw [hHeq]
exact witness_P0_le_kernel hG hnoV data
-- `p ∤ |E'|`: Frobenius kernel/complement coprimality with `p ∣ |T| ∣ |H|`.
have hp_ndvd : ¬ ctr.p ∣ Nat.card ↥(frob.complement.map data.L.subtype) := by
have hpT : ctr.p ∣ Nat.card ↥T := by
rcases hTcard with h | h
· rw [h]
· rw [h]; exact dvd_pow_self ctr.p two_ne_zero
have hcopLL : Nat.Coprime (Nat.card ↥(frob.typeI.typeF.H.subgroupOf data.L))
(Nat.card ↥frob.complement) := frob.frobenius.coprime_card_kernel_complement
have hpHL : ctr.p ∣ Nat.card ↥(frob.typeI.typeF.H.subgroupOf data.L) := by
rw [Nat.card_congr (Subgroup.subgroupOfEquivOfLe frob.typeI.typeF.H_le).toEquiv]
exact hpT.trans (Subgroup.card_dvd_of_le hTH)
rw [hEcard]
intro hpE
exact ctr.p_prime.not_dvd_one (hcopLL ▸ Nat.dvd_gcd hpHL hpE)
-- The **nonscalar** hypothesis — Peterfalvi's (12.11) `A = 1` argument, elementwise.
have hnonscalar : Nat.card ↥T = ctr.p ^ 2 →
∀ e : G, e ∈ frob.complement.map data.L.subtype →
(∃ n : ℕ, ∀ t : G, t ∈ T → e * t * e⁻¹ = t ^ n) → e = 1 := by
intro hT2 e heE hsc
obtain ⟨n, hn⟩ := hsc
-- Bookkeeping: `|T| = p²` forces `T = Ω₁(P₀)`, so the (12.9) witness `x` lies in `T`.
obtain ⟨hP0ab, hP0rank⟩ := counterexample_P0_K_structure hG ctr
have hP0comm : ∀ y ∈ ctr.P0, ∀ z ∈ ctr.P0, y * z = z * y := fun y hy z hz =>
congrArg Subtype.val (hP0ab.is_comm.comm (⟨y, hy⟩ : ↥ctr.P0) ⟨z, hz⟩)
have hxOm : data.x ∈ OddOrder.GroupTheory.omega1OfAbelian G ctr.P0 ctr.p hP0comm :=
⟨data.x_mem_P0, data.x_mem_omega1⟩
have hTOm : T ≤ OddOrder.GroupTheory.omega1OfAbelian G ctr.P0 ctr.p hP0comm := fun t ht =>
⟨hTP0 ht, by simpa using congrArg Subtype.val (hTelem.2 (⟨t, ht⟩ : ↥T))⟩
have hOmne : OddOrder.GroupTheory.omega1OfAbelian G ctr.P0 ctr.p hP0comm ≠ ⊥ := fun hbot =>
data.x_ne_one (Subgroup.mem_bot.mp (hbot ▸ hxOm))
have hOmcard := OddOrder.GroupTheory.card_eq_prime_or_sq_of_isElementaryAbelian_le
OddOrder.GroupTheory.omega1OfAbelian_isElementaryAbelian
OddOrder.GroupTheory.omega1OfAbelian_le hP0rank.le hOmne
have hTeq : T = OddOrder.GroupTheory.omega1OfAbelian G ctr.P0 ctr.p hP0comm := by
refine Subgroup.eq_of_le_of_card_ge hTOm ?_
rw [hT2]
rcases hOmcard with h | h
· rw [h]; exact Nat.le_self_pow two_ne_zero ctr.p
· rw [h]
have hxT : data.x ∈ T := by rw [hTeq]; exact hxOm
-- `orderOf x = p`; the scalar exponent `n` is prime to `p` (else conjugation kills `x`).
have hordx : orderOf data.x = ctr.p := orderOf_eq_prime data.x_mem_omega1 data.x_ne_one
have hexn : e * data.x * e⁻¹ = data.x ^ n := hn data.x hxT
have hxnne : data.x ^ n ≠ 1 := fun h1 => by
have : e * data.x * e⁻¹ = 1 := by rw [hexn, h1]
exact data.x_ne_one (by
calc data.x = e⁻¹ * (e * data.x * e⁻¹) * e := by group
_ = 1 := by rw [this]; group)
have hordxn : orderOf (data.x ^ n) = ctr.p := by
have hdvd : orderOf (data.x ^ n) ∣ ctr.p :=
orderOf_dvd_iff_pow_eq_one.mpr (by
rw [← pow_mul, mul_comm n ctr.p, pow_mul, data.x_mem_omega1, one_pow])
rcases (Nat.Prime.eq_one_or_self_of_dvd ctr.p_prime _ hdvd) with h1 | hp
· exact absurd (orderOf_eq_one_iff.mp h1) hxnne
· exact hp
-- `zpowers (xⁿ) = zpowers x` (both have order `p`), so `e` normalizes `⟨x⟩`.
have hzp : Subgroup.zpowers (data.x ^ n) = Subgroup.zpowers data.x := by
refine Subgroup.eq_of_le_of_card_ge
(Subgroup.zpowers_le.mpr (Subgroup.pow_mem _ (Subgroup.mem_zpowers data.x) n)) ?_
rw [Nat.card_zpowers, Nat.card_zpowers, hordx, hordxn]
have heN : e ∈ Subgroup.normalizer
((Subgroup.closure ({data.x} : Set G) : Subgroup G) : Set G) := by
rw [← Subgroup.zpowers_eq_closure, Subgroup.mem_set_normalizer_iff]
intro h
constructor
· intro hh
obtain ⟨k, rfl⟩ := Subgroup.mem_zpowers_iff.mp (SetLike.mem_coe.mp hh)
refine SetLike.mem_coe.mpr ?_
rw [← conj_zpow, hexn]
exact hzp ▸ Subgroup.mem_zpowers_iff.mpr ⟨k, rfl⟩
· intro hh
have hh' : e * h * e⁻¹ ∈ Subgroup.zpowers (data.x ^ n) := by
rw [hzp]; exact SetLike.mem_coe.mp hh
obtain ⟨k, hk⟩ := Subgroup.mem_zpowers_iff.mp hh'
rw [← hexn, conj_zpow] at hk
exact SetLike.mem_coe.mpr (Subgroup.mem_zpowers_iff.mpr
⟨k, mul_left_cancel (mul_right_cancel hk)⟩)
-- (12.9): `e ∈ N_G(⟨x⟩) ⊆ M`; with `e ∈ L`, (12.11) places `e ∈ H = L_F`.
have heM : e ∈ ctr.M := data.normalizer_closure_x_le_M heN
have heL : e ∈ data.L := Subgroup.map_subtype_le _ heE
have heH : e ∈ frob.typeI.typeF.H := by
rw [hHeq]
exact intersection_le_kernel hG hnoV data ⟨heM, heL⟩
-- Frobenius disjointness: `e ∈ E ∩ H = 1`.
obtain ⟨a, haC, rfl⟩ := heE
have haH : a ∈ frob.typeI.typeF.H.subgroupOf data.L :=
Subgroup.mem_subgroupOf.mpr heH
have ha1 : a = 1 :=
Subgroup.disjoint_def.mp frob.frobenius.isComplement.disjoint haH haC
rw [ha1, map_one]
-- The proven rep-theory dichotomy, with the scalar case excluded.
obtain ⟨-, hdvd⟩ := isCyclic_and_card_dvd_sub_or_add_one_of_fpf_conj_elemAbelian
hTelem hEnorm hodd hp_ndvd hTcard hfpf hnonscalar
rwa [hEcard] at hdvd
/-- **Peterfalvi (12.12), structural input from (12.9)/(12.10)/(12.11)** (§8/§9 obligation pinned
as an explicit hypothesis, hub 9003 Cluster A; the upstream has since been proved). For the
(12.9) witness `L` (type-I Frobenius, kernel `H = L_F`),
with `E := frob.complement.map L.subtype` the Frobenius complement realized in `G`, there is a
subgroup `T ≤ G` — Peterfalvi's `T = Ω₁Z(O_p(H))` — that is
* **elementary abelian** of order `p` or `p²` (`P₀` is abelian of rank `2` by (12.9), so `Ω₁Z(P)`
has order `p` or `p²`);
* **normalized by `E`** (`E` normalizes `O_p(H)`, its center, and the `Ω₁`);
* on which `E` acts **fixed-point-freely by conjugation** (Peterfalvi (12.10): as `L` is Frobenius
with kernel `H`, the complement `E` fixes no nonidentity element of `H`, a fortiori none of
`T ⊆ H`),
and, encoding the `p+1` refinement of (12.12) (the (12.11) step `A ⊆ M ⟹ A = 1` for `A ≤ E` of
order dividing `p-1`), if `|E|` divides `p² - 1` then in fact `|E|` divides `p - 1` or `p + 1`. We
also record `T ≤ H` (`Ω₁Z(O_p(H)) ⊆ H`), used to see `p ∣ |H|`.
**Assembly** (proven, modulo the `p+1` refinement pin): `P := O_p(H)` contains `P₀` (nilpotent
`H`), and `T := Ω₁(Z(P))` is elementary abelian (`omega1OfAbelian`). The elided order bound is
the (12.9) control: `T ⊆ Z(P) ⊆ C_G(x) ≤ N_G(⟨x⟩) ≤ M` and `T` centralizes `P₀ ≤ P`, so the
abelian `p`-subgroup `T ⊔ P₀ ≤ M` lies in the full `p`-part `P₀`, whence `T ⊆ P₀` (abelian of
rank `2`) and `|T| ∈ {p, p²}` (`card_eq_prime_or_sq_of_isElementaryAbelian_le`). `E` normalizes
`T` through the characteristic chain `N(H) ≤ N(O_p(H)) ≤ N(Z(O_p(H))) ≤ N(Ω₁(...))`, and acts
fixed-point-freely on `T ⊆ H` by the Frobenius structure. The `p+1` refinement is the pinned
`witness_complement_dvd_p_sub_or_add_one`. -/
theorem exists_center_omega1_elemAbelian_fpf_of_witness [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) :
∃ T : Subgroup G, IsElementaryAbelian ctr.p ↥T ∧
(frob.complement.map data.L.subtype ≤ Subgroup.normalizer (T : Set G)) ∧
(Nat.card ↥T = ctr.p ∨ Nat.card ↥T = ctr.p ^ 2) ∧
T ≤ frob.typeI.typeF.H ∧
(∀ e : G, e ∈ frob.complement.map data.L.subtype → e ≠ 1 →
∀ t : G, t ∈ T → e * t * e⁻¹ = t → t = 1) ∧
(Nat.card ↥frob.complement ∣ ctr.p ^ 2 - 1 →
Nat.card ↥frob.complement ∣ ctr.p - 1 ∨ Nat.card ↥frob.complement ∣ ctr.p + 1) := by
classical
have : Fact ctr.p.Prime := ⟨ctr.p_prime⟩
-- `H = L_F` is nilpotent and contains `P₀`; set `P := O_p(H) ⊇ P₀`.
have hHeq : frob.typeI.typeF.H = maxNilpotentNormalHall data.L := frob.typeI.typeF.H_eq
have hHnilp : Group.IsNilpotent ↥frob.typeI.typeF.H := by
rw [hHeq]; exact OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_isNilpotent data.L
have hP0H : ctr.P0 ≤ frob.typeI.typeF.H := by
rw [hHeq]; exact witness_P0_le_kernel hG hnoV data
set P : Subgroup G := opiCoreInG ({ctr.p} : Set ℕ) frob.typeI.typeF.H with hPdef
have hP0P : ctr.P0 ≤ P := pGroup_le_opiCoreInG_of_le_of_isNilpotent ctr.P0_pGroup hP0H
have hPH : P ≤ frob.typeI.typeF.H := opiCoreInG_le _ _
have hPp : IsPGroup ctr.p ↥P := isPGroup_opiCoreInG_singleton _
-- `Z := Z(P)` in `G` (abelian), `T := Ω₁(Z)`.
set Z : Subgroup G := (Subgroup.center ↥P).map P.subtype with hZdef
have hZcomm : ∀ x ∈ Z, ∀ y ∈ Z, x * y = y * x := fun x hx y hy =>
((Subgroup.mem_center_map_subtype_iff.mp hx).2 y
(Subgroup.mem_center_map_subtype_iff.mp hy).1).symm
set T : Subgroup G := OddOrder.GroupTheory.omega1OfAbelian G Z ctr.p hZcomm with hTdef
have hTZ : T ≤ Z := OddOrder.GroupTheory.omega1OfAbelian_le
have hZP : Z ≤ P := hZdef ▸ Subgroup.map_subtype_le _
have hTH : T ≤ frob.typeI.typeF.H := hTZ.trans (hZP.trans hPH)
have hTelem : T.IsElementaryAbelian ctr.p :=
OddOrder.GroupTheory.omega1OfAbelian_isElementaryAbelian
-- `T ≠ ⊥`: the nontrivial `p`-group `P` has nontrivial center, whose Cauchy `p`-element
-- lies in `Ω₁(Z)`.
have hP0ne : ctr.P0 ≠ ⊥ := fun h => ctr.P0_noncyclic (h ▸ inferInstance)
have hPne : P ≠ ⊥ := fun h => hP0ne (le_bot_iff.mp (h ▸ hP0P))
have hTne : T ≠ ⊥ := by
have : Nontrivial ↥P := P.nontrivial_iff_ne_bot.mpr hPne
have : Nontrivial (Subgroup.center ↥P) := hPp.center_nontrivial
have hZne : Z ≠ ⊥ := by
intro hbot
obtain ⟨z, hz1⟩ := exists_ne (1 : Subgroup.center ↥P)
refine hz1 (Subtype.ext (Subtype.ext ?_))
have hzZ : ((z : ↥P) : G) ∈ Z := hZdef ▸ ⟨z, z.2, rfl⟩
rw [hbot, Subgroup.mem_bot] at hzZ
exact hzZ
have hZp : IsPGroup ctr.p ↥Z := hPp.to_le hZP
obtain ⟨k, hk⟩ := hZp.exists_card_eq
have hkpos : k ≠ 0 := by
rintro rfl
exact hZne (Subgroup.card_eq_one.mp (by rw [hk, pow_zero]))
have : Fintype ↥Z := Fintype.ofFinite _
obtain ⟨g, hg⟩ := exists_prime_orderOf_dvd_card (G := ↥Z) ctr.p
(by rw [← Nat.card_eq_fintype_card, hk]; exact dvd_pow_self _ hkpos)
have hg_ord : orderOf (g : G) = ctr.p := by
rw [← hg]; exact orderOf_injective Z.subtype Z.subtype_injective g
have hgT : (g : G) ∈ T := ⟨g.2, by rw [← hg_ord]; exact pow_orderOf_eq_one _⟩
intro hbot
rw [hbot, Subgroup.mem_bot] at hgT
rw [hgT, orderOf_one] at hg_ord
exact ctr.p_prime.one_lt.ne' hg_ord.symm
-- The (12.9) control: `T ⊆ Z(P) ⊆ C_G(x) ≤ M`, and `T` centralizes `P₀`, so the abelian
-- `p`-subgroup `T ⊔ P₀ ≤ M` lies in the full `p`-part `P₀`; hence `T ⊆ P₀` of rank `2`.
have hxP : data.x ∈ P := hP0P data.x_mem_P0
have hTCx : T ≤ Subgroup.centralizer ({data.x} : Set G) := fun t ht =>
Subgroup.mem_centralizer_singleton_iff.mpr
((Subgroup.mem_center_map_subtype_iff.mp (hTZ ht)).2 data.x hxP).symm
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 _
have hTM : T ≤ ctr.M := hTCx.trans hCM
have hTcent : T ≤ Subgroup.centralizer (ctr.P0 : Set G) := fun t ht =>
Subgroup.mem_centralizer_iff.mpr fun w hw =>
(Subgroup.mem_center_map_subtype_iff.mp (hTZ ht)).2 w (hP0P hw)
have hTp : IsPGroup ctr.p ↥T := hPp.to_le (hTZ.trans hZP)
have hsup_p : IsPGroup ctr.p ↥(T ⊔ ctr.P0) :=
IsPGroup.to_sup_of_normal_right' hTp ctr.P0_pGroup
(hTcent.trans (Subgroup.centralizer_le_normalizer _))
have hsup_eq : T ⊔ ctr.P0 = ctr.P0 := by
have hle_M : T ⊔ ctr.P0 ≤ ctr.M := sup_le hTM ctr.P0_le_M
obtain ⟨m, hm⟩ := hsup_p.exists_card_eq
have hdvd_M : (ctr.p : ℕ) ^ m ∣ Nat.card ↥ctr.M := by
rw [← hm]; exact Subgroup.card_dvd_of_le hle_M
have hMsplit : Nat.card ↥ctr.M = Nat.card ↥ctr.P0 * ctr.P0.relIndex ctr.M := by
rw [Subgroup.relIndex, ← Nat.card_congr
(Subgroup.subgroupOfEquivOfLe ctr.P0_le_M).toEquiv]
exact ((ctr.P0.subgroupOf ctr.M).card_mul_index).symm
have hcop : Nat.Coprime (ctr.p ^ m) (ctr.P0.relIndex ctr.M) :=
Nat.Coprime.pow_left m
(ctr.p_prime.coprime_iff_not_dvd.mpr ctr.P0_sylow)
have hdvd_P0 : (ctr.p : ℕ) ^ m ∣ Nat.card ↥ctr.P0 :=
hcop.dvd_of_dvd_mul_right (hMsplit ▸ hdvd_M)
refine (Subgroup.eq_of_le_of_card_ge le_sup_right ?_).symm
rw [hm]
exact Nat.le_of_dvd Nat.card_pos hdvd_P0
have hTP0 : T ≤ ctr.P0 := hsup_eq ▸ le_sup_left
have hTcard : Nat.card ↥T = ctr.p ∨ Nat.card ↥T = ctr.p ^ 2 :=
OddOrder.GroupTheory.card_eq_prime_or_sq_of_isElementaryAbelian_le hTelem hTP0
(counterexample_P0_K_structure hG ctr).2.le hTne
-- `E` normalizes `T`: through `N(H) ≤ N(O_p(H)) ≤ N(Z(O_p(H))) ≤ N(Ω₁(Z))`.
have hEnorm : frob.complement.map data.L.subtype ≤ Subgroup.normalizer (T : Set G) := by
have hchain : Subgroup.normalizer ((frob.typeI.typeF.H : Subgroup G) : Set G) ≤
Subgroup.normalizer ((T : Subgroup G) : Set G) := by
intro g hg
have hgP : g ∈ Subgroup.normalizer ((P : Subgroup G) : Set G) :=
le_normalizer_opiCoreInG_of_le_normalizer ({ctr.p} : Set ℕ) le_rfl hg
have hgZ : g ∈ Subgroup.normalizer ((Z : Subgroup G) : Set G) :=
hZdef ▸ Subgroup.mem_normalizer_center_map_of_mem_normalizer hgP
exact OddOrder.GroupTheory.mem_normalizer_omega1OfAbelian hgZ
refine le_trans ?_ hchain
intro e he
obtain ⟨a, -, rfl⟩ := he
have hLN : data.L ≤ Subgroup.normalizer ((frob.typeI.typeF.H : Subgroup G) : Set G) := by
rw [hHeq]
exact OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le_normalizer data.L
exact hLN a.2
-- Fixed-point-freeness on `T ⊆ H` from the Frobenius structure of `L`.
have hfpf : ∀ e : G, e ∈ frob.complement.map data.L.subtype → e ≠ 1 →
∀ t : G, t ∈ T → e * t * e⁻¹ = t → t = 1 := by
intro e he hene t ht hconj
obtain ⟨a, haC, rfl⟩ := he
have htH : t ∈ frob.typeI.typeF.H := hTH ht
have htL : t ∈ data.L := frob.typeI.typeF.H_le htH
by_contra htne
have hane : a ≠ 1 := fun h => hene (by rw [h]; rfl)
exact frob.frobenius.conj_frobenius a haC hane ⟨t, htL⟩
(Subgroup.mem_subgroupOf.mpr htH) (fun h => htne (congrArg Subtype.val h))
(Subtype.ext hconj)
exact ⟨T, hTelem, hEnorm, hTcard, hTH, hfpf, fun _ =>
witness_complement_dvd_p_sub_or_add_one hG hnoV data frob hTelem hTP0 hTne hEnorm hfpf⟩
/-- **Peterfalvi (12.12)**: the Frobenius complement `E` in the (12.9) witness subgroup `L` is
cyclic, with order `e = |E|` dividing `p - 1` or `p + 1`.
**Assembly** (`sorry`-free modulo the (12.9)/(12.10)/(12.11) structural package): from
`exists_center_omega1_elemAbelian_fpf_of_witness` we obtain `T = Ω₁Z(O_p(H))` — elementary abelian
of order `p` or `p²`, normalized by `E` (realized in `G` as `E' = frob.complement.map L.subtype`),
with `E'` acting fixed-point-freely on `T` by conjugation. The proven rep-theory core
`isCyclic_and_card_dvd_of_fpf_conj_elemAbelian` then gives
`IsCyclic E' ∧ (|E'| ∣ p-1 ∨ |E'| ∣ p²-1)`
(the `§8`-free Singer/Case-A+B mechanism). Transporting cyclicity back along `L.subtype` (`E ≅ E'`)
and applying the packaged `p+1` refinement to the `p²-1` branch yields the (12.12) conclusion. -/
theorem complement_cyclic_order_dvd [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) :
IsCyclic ↥frob.complement ∧
((Nat.card ↥frob.complement ∣ ctr.p - 1) ∨
(Nat.card ↥frob.complement ∣ ctr.p + 1)) := by
classical
have : Fact ctr.p.Prime := ⟨ctr.p_prime⟩
-- The Frobenius complement, realized as a subgroup `E'` of the ambient `G`.
set E' : Subgroup G := frob.complement.map data.L.subtype with hE'
-- `E ≅ E'` (injective image), so cardinalities agree.
have hEcard : Nat.card ↥E' = Nat.card ↥frob.complement :=
Subgroup.card_map_of_injective (K := frob.complement) data.L.subtype_injective
-- The (12.9)/(12.10)/(12.11) structural package for the witness complement.
obtain ⟨T, hTelem, hEnorm, hTcard, hTleH, hfpf, hrefine⟩ :=
exists_center_omega1_elemAbelian_fpf_of_witness hG hnoV data frob
-- Odd order of `E'` (a subgroup of the odd-order `G`).
have hodd : Odd (Nat.card ↥E') :=
hG.odd.of_dvd_nat (Subgroup.card_subgroup_dvd_card E')
-- `p ∤ |E'|`: `E'` is a Frobenius complement, coprime to the kernel `H ⊇ T` which has order
-- divisible by `p` (`|T| = p` or `p²`). Concretely `|T| ∣ |kernel|` and `p ∣ |T|`, while
-- `Coprime |kernel| |complement|`, so `p ∤ |E'|`.
have hp_ndvd : ¬ ctr.p ∣ Nat.card ↥E' := by
-- `p ∣ |T|` (order `p` or `p²`).
have hpT : ctr.p ∣ Nat.card ↥T := by
rcases hTcard with h | h
· rw [h]
· rw [h]; exact dvd_pow_self ctr.p (by norm_num)
-- `T ≤ H` (`T = Ω₁Z(O_p(H)) ⊆ H`); realize via the FPF hypothesis: `T`'s elements are moved by
-- every nontrivial element of `E'`, and `E'`, `H` are Frobenius-coprime. We use the abstract
-- coprimality of the Frobenius pair on `↥L`.
have hcopLL : Nat.Coprime (Nat.card ↥(frob.typeI.typeF.H.subgroupOf data.L))
(Nat.card ↥frob.complement) := frob.frobenius.coprime_card_kernel_complement
-- It suffices that `p ∣ |H|` and `Coprime |H| |E'|` (via `|E'| = |E|`), then `p ∤ |E'|`.
-- `|H_L| = |H|` where `H_L = H.subgroupOf L`.
have hHcard : Nat.card ↥(frob.typeI.typeF.H.subgroupOf data.L)
= Nat.card ↥frob.typeI.typeF.H :=
Nat.card_congr (Subgroup.subgroupOfEquivOfLe frob.typeI.typeF.H_le).toEquiv
-- `p ∣ |H|`: `T ≤ H` (`hTleH`, from the package), and `p ∣ |T| ∣ |H|`.
have hpH : ctr.p ∣ Nat.card ↥frob.typeI.typeF.H :=
hpT.trans (Subgroup.card_dvd_of_le hTleH)
have hpHL : ctr.p ∣ Nat.card ↥(frob.typeI.typeF.H.subgroupOf data.L) := by
rw [hHcard]; exact hpH
rw [hEcard]
intro hpE
exact ctr.p_prime.not_dvd_one (hcopLL ▸ Nat.dvd_gcd hpHL hpE)
-- The proven rep-theory core: `E'` cyclic and `|E'| ∣ p-1 ∨ |E'| ∣ p²-1`.
obtain ⟨hcycE', hdvdE'⟩ :=
isCyclic_and_card_dvd_of_fpf_conj_elemAbelian hTelem hEnorm hodd (hEcard ▸ hp_ndvd) hTcard hfpf
-- Transport cyclicity `E' ≅ E` back to `E`.
have hcyc : IsCyclic ↥frob.complement :=
isCyclic_of_surjective _
(Subgroup.equivMapOfInjective frob.complement data.L.subtype
data.L.subtype_injective).symm.surjective
refine ⟨hcyc, ?_⟩
-- Rewrite `|E'| = |E|` in the divisibility and apply the `p+1` refinement.
rw [hEcard] at hdvdE'
rcases hdvdE' with h | h
· exact Or.inl h
· exact hrefine h
/-- **Peterfalvi (12.12) → (12.16) numeric input: `2e ≤ p + 1`** ("Also `2e ≤ p+1` by (12.12)").
The witness complement order `e = |E|` is odd (a subgroup of the odd-order `G`) and divides
`p − 1` or `p + 1` by (12.12); an odd divisor of the even number `p ∓ 1` divides its half, so
`2e ≤ p ∓ 1 ≤ p + 1`. This is the `h2e` field of `CounterexampleDadeData`, in `ℕ` form. -/
theorem two_mul_card_complement_le [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) :
2 * Nat.card ↥frob.complement ≤ ctr.p + 1 := by
classical
have : Fact ctr.p.Prime := ⟨ctr.p_prime⟩
obtain ⟨-, hdvd⟩ := complement_cyclic_order_dvd hG hnoV data frob
-- `e = |E|` is odd.
have hodd : Odd (Nat.card ↥frob.complement) :=
hG.odd.of_dvd_nat ((Subgroup.card_subgroup_dvd_card frob.complement).trans
(Subgroup.card_subgroup_dvd_card data.L))
-- `p` is odd and `≥ 3` (`p ∣ |G|` odd).
have hpG : ctr.p ∣ Nat.card G := by
have h1 : ctr.p ∣ (ctr.K.subgroupOf ctr.M).index := by
have := ctr.p_dvd_index
rwa [Subgroup.relIndex] at this
exact (h1.trans (Subgroup.index_dvd_card _)).trans
(Subgroup.card_subgroup_dvd_card ctr.M)
have hpodd : Odd ctr.p := hG.odd.of_dvd_nat hpG
have hp3 : 3 ≤ ctr.p := by
have h2 := ctr.p_prime.two_le
have hne : ctr.p ≠ 2 := fun h => by
rw [h] at hpodd
exact (by decide : ¬ Odd 2) hpodd
omega
-- An odd divisor of an even number divides its half.
have he_mod : Nat.card ↥frob.complement % 2 = 1 := Nat.odd_iff.mp hodd
have key : ∀ m : ℕ, 0 < m → Even m → Nat.card ↥frob.complement ∣ m →
2 * Nat.card ↥frob.complement ≤ m := by
intro m hm hme hdvd'
obtain ⟨k, hk⟩ := hme
have hm2k : m = 2 * k := by omega
have hcop : Nat.Coprime (Nat.card ↥frob.complement) 2 :=
(Nat.prime_two.coprime_iff_not_dvd.mpr (fun h => by omega)).symm
have hdvd_k : Nat.card ↥frob.complement ∣ k :=
Nat.Coprime.dvd_of_dvd_mul_left hcop (hm2k ▸ hdvd')
have hkpos : 0 < k := by omega
have := Nat.le_of_dvd hkpos hdvd_k
omega
rcases hdvd with h | h
· -- `e ∣ p − 1`: `2e ≤ p − 1 ≤ p + 1`.
have := key (ctr.p - 1) (by omega) (Nat.Odd.sub_odd hpodd odd_one) h
omega
· -- `e ∣ p + 1`: `2e ≤ p + 1`.
exact key (ctr.p + 1) (by omega) (Odd.add_one hpodd) h
/-- **The counterexample `M` is not a Frobenius group with kernel `M_σ = K`** (the (12.8)
non-Frobenius content, from `P₀` noncyclic): a Frobenius decomposition `M = M_σ ⋊ E` with
**cyclic** `E` would make the quotient `M/M_σ ≅ E` cyclic, and the noncyclic `p`-group `P₀`
(with `p ∤ |K| = |M_σ|`) injects into that quotient — contradiction.
Stated to refute exactly the `IsTypeP2 N` branch package of BG Theorem D(4)
(`exists_RData_escape_structure`), which is the Peterfalvi (8.13.c4) "furthermore" clause:
a type-`P₂` signalizer neighbour `N[g]` forces `M` Frobenius over `M_σ` with cyclic
complement. -/
theorem counterexample_not_frobenius_MF [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
(ctr : CounterexampleHypothesis (G := G)) :
¬ ∃ E : Subgroup G, E ≤ ctr.M ∧ IsCyclic ↥E ∧
Subgroup.IsComplement' ((OddOrder.BG.Ch3.S10.Msigma ctr.M).subgroupOf ctr.M)
(E.subgroupOf ctr.M) ∧
OddOrder.Isaacs.Ch06.IsFrobeniusGroup ↥ctr.M
((OddOrder.BG.Ch3.S10.Msigma ctr.M).subgroupOf ctr.M) (E.subgroupOf ctr.M) := by
classical
rintro ⟨E, hEM, hEcyc, hcompl, -⟩
have : Fact ctr.p.Prime := ⟨ctr.p_prime⟩
set Q : Subgroup ↥ctr.M := (OddOrder.BG.Ch3.S10.Msigma ctr.M).subgroupOf ctr.M with hQ
have hQnormal : Q.Normal := by
rw [hQ, OddOrder.BG.Ch3.S10.Msigma_subgroupOf]
infer_instance
-- `|Q| = |K|`, so `|P₀| = pⁿ` is coprime to `|Q|` (`p ∤ |K|`).
have hQcard : Nat.card ↥Q = Nat.card ↥ctr.K := by
rw [hQ, ← MF_eq_Msigma hG ctr]
exact Nat.card_congr (Subgroup.subgroupOfEquivOfLe
(ctr.K_eq_MF ▸ OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le ctr.M)).toEquiv
set P' : Subgroup ↥ctr.M := ctr.P0.subgroupOf ctr.M with hP'
have hP'card : Nat.card ↥P' = Nat.card ↥ctr.P0 :=
Nat.card_congr (Subgroup.subgroupOfEquivOfLe ctr.P0_le_M).toEquiv
have hPQbot : P' ⊓ Q = ⊥ := by
have hcop : Nat.Coprime (Nat.card ↥P') (Nat.card ↥Q) := by
obtain ⟨n, hn⟩ := IsPGroup.iff_card.mp ctr.P0_pGroup
rw [hP'card, hQcard, hn]
exact ((Nat.Prime.coprime_iff_not_dvd ctr.p_prime).mpr (p_not_dvd_card_K ctr)).pow_left n
exact disjoint_iff.mp (Subgroup.disjoint_of_coprime_natCard hcop)
-- The quotient `M/Q ≅ E` is cyclic.
have hEsub_cyc : IsCyclic ↥(E.subgroupOf ctr.M) := by
have := hEcyc
exact isCyclic_of_surjective (Subgroup.subgroupOfEquivOfLe hEM).symm.toMonoidHom
(Subgroup.subgroupOfEquivOfLe hEM).symm.surjective
have hquot_cyc : IsCyclic (↥ctr.M ⧸ Q) := by
refine isCyclic_of_surjective
((QuotientGroup.mk' Q).comp (E.subgroupOf ctr.M).subtype) ?_
intro q
obtain ⟨m, rfl⟩ := QuotientGroup.mk'_surjective Q q
obtain ⟨⟨q0, e0⟩, hqe⟩ := (hcompl.existsUnique m).exists
refine ⟨⟨(e0 : ↥ctr.M), SetLike.mem_coe.mp e0.2⟩, ?_⟩
have hq0 : (QuotientGroup.mk' Q) (q0 : ↥ctr.M) = 1 :=
(QuotientGroup.eq_one_iff _).mpr (SetLike.mem_coe.mp q0.2)
calc ((QuotientGroup.mk' Q).comp (E.subgroupOf ctr.M).subtype)
⟨(e0 : ↥ctr.M), SetLike.mem_coe.mp e0.2⟩
= (QuotientGroup.mk' Q) (e0 : ↥ctr.M) := rfl
_ = (QuotientGroup.mk' Q) (q0 : ↥ctr.M) * (QuotientGroup.mk' Q) (e0 : ↥ctr.M) := by
rw [hq0, one_mul]
_ = (QuotientGroup.mk' Q) ((q0 : ↥ctr.M) * (e0 : ↥ctr.M)) := by rw [map_mul]
_ = (QuotientGroup.mk' Q) m := by rw [hqe]
-- `P₀` injects into the cyclic quotient, so it is cyclic — contradiction with `P0_noncyclic`.
have hinj : Function.Injective ((QuotientGroup.mk' Q).comp P'.subtype) := by
rw [← MonoidHom.ker_eq_bot_iff, eq_bot_iff]
intro x hx
have hxQ : (x : ↥ctr.M) ∈ Q :=
(QuotientGroup.eq_one_iff _).mp (MonoidHom.mem_ker.mp hx)
have hxPQ : (x : ↥ctr.M) ∈ P' ⊓ Q := ⟨x.2, hxQ⟩
rw [hPQbot, Subgroup.mem_bot] at hxPQ
exact Subgroup.mem_bot.mpr (Subtype.ext hxPQ)
have hP'cyc : IsCyclic ↥P' :=
isCyclic_of_surjective (MonoidHom.ofInjective hinj).symm.toMonoidHom
(MonoidHom.ofInjective hinj).symm.surjective
exact ctr.P0_noncyclic (isCyclic_of_surjective
(Subgroup.subgroupOfEquivOfLe ctr.P0_le_M).toMonoidHom
(Subgroup.subgroupOfEquivOfLe ctr.P0_le_M).surjective)
/-- **The witness `L` is not conjugate to a maximal `N` with a non-kernel element centralizing
the kernel** (the (12.15) `N ≁ L` step): if some `g ∈ N ∖ N_F` has `C_{N_F}(g) ≠ 1`, then `N`
is not a Frobenius group with kernel `N_F` — but every conjugate of the witness `L` is: the
Frobenius structure of `L` ((12.10), `witness_L_frobenius`) transports along `conj c`
(`maxNilpotentNormalHall_pointwise_smul`), and non-kernel elements of a Frobenius group act
fixed-point-freely on the kernel
(`IsFrobeniusGroup.centralizer_inf_kernel_eq_bot_of_not_mem`). -/
theorem witness_L_not_conj_of_kernel_centralizer_ne_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)
{N : Subgroup G} {g : G} (hgN : g ∈ N)
(hgNF : g ∉ maxNilpotentNormalHall N)
(hR : maxNilpotentNormalHall N ⊓ Subgroup.centralizer ({g} : Set G) ≠ ⊥) :
¬ ∃ c : G, MulAut.conj c • data.L = N := by
rintro ⟨c, hc⟩
obtain ⟨frob, -⟩ := witness_L_frobenius hG hnoV data
-- kernel transport: `N_F = (conj c) • L_F`.
have hNF : maxNilpotentNormalHall N
= MulAut.conj c • maxNilpotentNormalHall data.L := by
rw [← hc, maxNilpotentNormalHall_pointwise_smul]
-- pull `g` back to `L`.
set g' : G := (MulAut.conj c)⁻¹ • g with hg'
have hg'L : g' ∈ data.L :=
Subgroup.mem_pointwise_smul_iff_inv_smul_mem.mp (hc ▸ hgN)
have hg'LF : g' ∉ maxNilpotentNormalHall data.L := by
intro hmem
refine hgNF ?_
rw [hNF]
have hsm : MulAut.conj c • g' ∈ MulAut.conj c • maxNilpotentNormalHall data.L :=
Subgroup.smul_mem_pointwise_smul _ _ _ hmem
rwa [hg', smul_inv_smul] at hsm
-- pull a nontrivial centralizing kernel element back to `L_F`.
have hex : ∃ x ∈ maxNilpotentNormalHall N ⊓ Subgroup.centralizer ({g} : Set G), x ≠ 1 := by
by_contra h
push Not at h
exact hR (eq_bot_iff.mpr fun x hx => Subgroup.mem_bot.mpr (h x hx))
obtain ⟨x, hxmem, hxne⟩ := hex
obtain ⟨hxNF, hxcent⟩ := Subgroup.mem_inf.mp hxmem
set x' : G := (MulAut.conj c)⁻¹ • x with hx'
have hx'LF : x' ∈ maxNilpotentNormalHall data.L :=
Subgroup.mem_pointwise_smul_iff_inv_smul_mem.mp (hNF ▸ hxNF)
have hx'ne : x' ≠ 1 := by
intro h1
apply hxne
have : MulAut.conj c • x' = MulAut.conj c • (1 : G) := by rw [h1]
rwa [hx', smul_inv_smul, MulAut.smul_def, map_one] at this
have hcomm : x' * g' = g' * x' := by
have hgx : g * x = x * g := (Subgroup.mem_centralizer_singleton_iff.mp hxcent).symm
rw [hg', hx', ← smul_mul', ← smul_mul', hgx]
-- contradiction with the Frobenius fixed-point-freeness of `L`.
have hg'L' : (⟨g', hg'L⟩ : ↥data.L) ∉ frob.typeI.typeF.H.subgroupOf data.L := by
intro hmem
exact hg'LF (frob.typeI.typeF.H_eq ▸ Subgroup.mem_subgroupOf.mp hmem)
have hFPF := IsFrobeniusGroup.centralizer_inf_kernel_eq_bot_of_not_mem frob.frobenius hg'L'
have hx'L : x' ∈ data.L :=
OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le data.L hx'LF
have hxLmem : (⟨x', hx'L⟩ : ↥data.L) ∈
Subgroup.centralizer ({(⟨g', hg'L⟩ : ↥data.L)} : Set ↥data.L) ⊓
frob.typeI.typeF.H.subgroupOf data.L := by
refine Subgroup.mem_inf.mpr ⟨?_, ?_⟩
· rw [Subgroup.mem_centralizer_singleton_iff]
apply Subtype.ext
push_cast
exact hcomm
· exact Subgroup.mem_subgroupOf.mpr (frob.typeI.typeF.H_eq ▸ hx'LF)
rw [hFPF, Subgroup.mem_bot] at hxLmem
exact hx'ne (by simpa using congrArg (fun z : ↥data.L => (z : G)) hxLmem)
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/MinimalCounterexample.lean