Peterfalvi (12.13)–(12.17) and the (8.8) case (b) dichotomy: the Dade contradiction
DefinitionOddOrder_Peterfalvi_S14_MaximalI_DadeContradictionfeit-thompsonfinite-groupsgroup-theoryishida-odd-order
Formalizes the final Dade calculation of Peterfalvi (12.13)–(12.17) for type I maximal subgroups and the (8.8) case (b) dichotomy.
Main declarations: DadeNotation (Peterfalvi (12.13)); dadeNotation_of_coherence (Peterfalvi (12.13), the Dade calculation realized from coherence); exists_distinguished_char (Peterfalvi (12.13), the distinguished character); exists_witness_placed_family (The placed induced family for the witness L).
The bundle also proves 23 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/DadeContradiction.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_MinimalCounterexample
namespace OddOrder.Peterfalvi.S14
open OddOrder.GroupTheory
open OddOrder.RepresentationTheory
open OddOrder.Isaacs
open scoped Pointwise
variable {G : Type*} [Group G]
/-- **Peterfalvi (12.13)**: notation for the final Dade calculation in the
minimal counterexample. -/
structure DadeNotation {L : Subgroup G} (hyp : Hypothesis L) where
e : ℕ
tau1 : OddOrder.Peterfalvi.S07.IntegralCharacterMap ↥L G
chi : ClassFunction ↥L ℂ
chi_mem : chi ∈ hyp.Sset
chi_degree_eq_e : chi 1 = (e : ℂ)
psi : ClassFunction G ℂ
psi_eq_tau1_chi : psi = tau1 chi
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.13), the Dade calculation realized from coherence**: given the (12.6) coherent
extension of `L`'s family `S` (`S07.IsCoherent`, supplied by `witness_L_coherent`) and a
distinguished
character `χ ∈ S` of degree `e`, the (12.13) `DadeNotation` is realized with `τ₁ =` the coherent
extension and `ψ = χ^{τ₁} = extension χ`.
This wires the coherent isometric extension into the `ψ`-construction backbone of (12.16): the
former opaque `tau1`/`psi` are now the genuine `coh.extension` and its value on `χ`. The remaining
input is the *selection* of the distinguished `χ` — a minimal-degree `Ind_H^L θ` with `θ` a
nontrivial linear character of `H = L_F`, so `χ(1) = [L:H] = e` — together with the (12.12) degree
bounds on `e`. (The former carried `Prop` fields `e_eq_index`/`rhoFormula`/`rhoMFormula` were
removed in the issue 0172 §12 audit: nothing consumed them, `rhoFormula`/`rhoMFormula` were filled
with `True`, and the genuine `e = [L:H]` identification is threaded explicitly as the `he_eq`
argument of `witness_value_norm_package`.) -/
noncomputable def dadeNotation_of_coherence {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
(coh : OddOrder.Peterfalvi.S07.IsCoherent hyp.tau hyp.Sset hyp.A)
(χ : ClassFunction ↥L ℂ) (hχ : χ ∈ hyp.Sset) (e : ℕ) (hdeg : χ 1 = (e : ℂ)) :
DadeNotation hyp where
e := e
tau1 := coh.extension
chi := χ
chi_mem := hχ
chi_degree_eq_e := hdeg
psi := coh.extension χ
psi_eq_tau1_chi := rfl
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.13), the distinguished character**: the family `S` of `L` contains a member of
minimal degree `[L : H]` (`H = L_F`), namely `Ind_H^L θ` for `θ` a nontrivial **linear** character
of `H`. Such `θ` exists because `H = L_F` is a nontrivial nilpotent group, so its commutator is
proper (`H` is not perfect);
`exists_irreducibleCharacter_ne_trivial_degree_one_of_commutator_ne_top`
then supplies a nontrivial degree-one `θ`, and `induce_apply_one` gives the induced degree
`[L:H]·θ(1) = [L:H]`. This is the distinguished `χ ∈ S` with `χ(1) = e = [L:H]` of the
(12.13)/(12.16)
Dade calculation — the input to `dadeNotation_of_coherence`. -/
theorem exists_distinguished_char {L : Subgroup G} [Finite G] (hyp : Hypothesis L) :
∃ χ ∈ hyp.Sset, χ (1 : ↥L) = (((hyp.typeI.typeF.H).subgroupOf L).index : ℂ) := by
have hHL : hyp.typeI.typeF.H ≤ L := hyp.typeI.typeF.H_le
have e : ↥((hyp.typeI.typeF.H).subgroupOf L) ≃* ↥(hyp.typeI.typeF.H) :=
Subgroup.subgroupOfEquivOfLe hHL
have : Nontrivial ↥(hyp.typeI.typeF.H) :=
(Subgroup.nontrivial_iff_ne_bot _).mpr hyp.typeI.typeF.H_nontrivial
have : Group.IsNilpotent ↥(hyp.typeI.typeF.H) :=
hyp.typeI.typeF.H_eq ▸ OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_isNilpotent L
have : Nontrivial ↥((hyp.typeI.typeF.H).subgroupOf L) := e.toEquiv.nontrivial
have : Group.IsSolvable ↥((hyp.typeI.typeF.H).subgroupOf L) :=
Group.isSolvable_of_isSolvable_injective (f := e.toMonoidHom) e.injective
have hcomm : commutator ↥((hyp.typeI.typeF.H).subgroupOf L) ≠ ⊤ :=
(Group.IsSolvable.commutator_lt_top_of_nontrivial _).ne
obtain ⟨θ, hθ_ne, hθ_deg⟩ :=
OddOrder.Peterfalvi.S08.exists_irreducibleCharacter_ne_trivial_degree_one_of_commutator_ne_top
hcomm
have hmem : ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L)
(θ : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ) ∈ hyp.Sset :=
⟨θ, hθ_ne, rfl⟩
refine ⟨_, hmem, ?_⟩
rw [ClassFunction.induce_apply_one, hθ_deg, mul_one]
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **The placed induced family for the witness `L`** (§12→§7 bridge, the `θ`/`ind1H` shape
`hypothesis78OfDade` consumes). Applies `exists_placed_induced_family` to the distinguished
`χ = Ind θ_lin ∈ S` of `exists_distinguished_char` (`θ_lin` nontrivial linear, so `χ ≠ Ind 1_K` by
`induce_ne_trivialChar_induce`): the distinguished char lands at index `0` with induced degree
`[L:K]` (`= e`), the trivial char `1_K` lands at some `ind1H ≠ 0`, and the family is
injective/covering. `K = (L_F).subgroupOf L` is normal in `L`
(`maxNilpotentNormalHall_..._normal`).
This is the family input to the witness-`L` `Hypothesis78`. -/
theorem exists_witness_placed_family {L : Subgroup G} [Finite G] (hyp : Hypothesis L) :
∃ (n : ℕ) (θ : Fin (n + 1) → IrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf L))
(ind1H : Fin (n + 1)),
ind1H ≠ 0 ∧
ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L)
(θ 0 : ClassFunction _ ℂ) (1 : ↥L) = (((hyp.typeI.typeF.H).subgroupOf L).index : ℂ) ∧
θ ind1H = trivialIrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf L) ∧
Function.Injective (fun i => ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L)
(θ i : ClassFunction _ ℂ)) ∧
∀ φ : IrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf L),
ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L) (φ : ClassFunction _ ℂ) ∈
Set.range (fun i => ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L)
(θ i : ClassFunction _ ℂ)) := by
have hKnormal : ((hyp.typeI.typeF.H).subgroupOf L).Normal := by
rw [hyp.typeI.typeF.H_eq]
exact OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_subgroupOf_normal L
obtain ⟨χ, hχ, hdeg⟩ := exists_distinguished_char hyp
obtain ⟨θlin, hθ_ne, hχ_eq⟩ := hχ
obtain ⟨n, θ, ind1H, hind, h0, htriv, hinj, hcov⟩ :=
OddOrder.Peterfalvi.S09.Cert.exists_placed_induced_family ((hyp.typeI.typeF.H).subgroupOf L) χ
⟨θlin, hχ_eq.symm⟩
(hχ_eq ▸ OddOrder.Peterfalvi.S09.Cert.induce_ne_trivialChar_induce
((hyp.typeI.typeF.H).subgroupOf L) θlin hθ_ne)
exact ⟨n, θ, ind1H, hind, by rw [h0]; exact hdeg, htriv, hinj, hcov⟩
/-- **Peterfalvi (12.13)/(12.16), the degree lower bound `e ≥ 3`**: the distinguished degree
`e = [L:H]` (`H = L_F`) of a type-I `Hypothesis` is at least `3`. It equals the order of the
Frobenius complement `U` (`H` complements `U` in `L`, `typeF.complement`), which is **nontrivial**
(`typeF.U_nontrivial`) and of **odd** order (a subgroup of the odd-order `G`); an odd integer `> 1`
is `≥ 3`. This discharges the `he : 3 ≤ e` field of `CounterexampleDadeData`. -/
theorem three_le_index {L : Subgroup G} [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
(hyp : Hypothesis L) :
3 ≤ ((hyp.typeI.typeF.H).subgroupOf L).index := by
-- `[L : H] = |U|` via the complement `H ⋊ U = L`.
have hUle : hyp.typeI.typeF.U ≤ L := hyp.typeI.typeF.U_le
have hidx_eq : ((hyp.typeI.typeF.H).subgroupOf L).index = Nat.card ↥(hyp.typeI.typeF.U) := by
rw [hyp.typeI.typeF.complement.symm.index_eq_card,
Nat.card_congr (Subgroup.subgroupOfEquivOfLe hUle).toEquiv]
-- `|U| > 1` (nontrivial) and `|U|` odd (divides `|G|` odd), so `|U| ≥ 3`.
have : Nontrivial ↥(hyp.typeI.typeF.U) :=
(Subgroup.nontrivial_iff_ne_bot _).mpr hyp.typeI.typeF.U_nontrivial
have hgt1 : 1 < ((hyp.typeI.typeF.H).subgroupOf L).index := by
rw [hidx_eq]; exact Finite.one_lt_card
have hodd : Odd ((hyp.typeI.typeF.H).subgroupOf L).index := by
have hdvd : ((hyp.typeI.typeF.H).subgroupOf L).index ∣ Nat.card G :=
(Subgroup.index_dvd_card _).trans (Subgroup.card_subgroup_dvd_card L)
rcases Nat.even_or_odd ((hyp.typeI.typeF.H).subgroupOf L).index with hev | ho
· exfalso
obtain ⟨d, hd⟩ := hG.odd
obtain ⟨m, hm⟩ := hev.two_dvd.trans hdvd
omega
· exact ho
obtain ⟨k, hk⟩ := hodd
omega
/-- **Peterfalvi (12.11)/(12.16), the index bound `|M| ≤ |K|·|H|`** (`H = L_F`): from the (12.11)
complement structure (`M ∩ L` complements `K` in `M`, and `M ∩ L ≤ L_F`), the order of `M`
factors as `|M| = |K|·|M ∩ L| ≤ |K|·|L_F|`. This discharges the `hM` field of
`CounterexampleDadeData` (cites the (12.11) `intersection_complement_structure`). -/
theorem card_M_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) :
Nat.card ↥ctr.M ≤ Nat.card ↥ctr.K * Nat.card ↥(maxNilpotentNormalHall data.L) := by
obtain ⟨hcompl, hsub⟩ := intersection_complement_structure hG hnoV data
have hKM : ctr.K ≤ ctr.M := ctr.K_eq_MF ▸ OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le ctr.M
-- `|M| = |K| · |M ∩ L|` from the complement `K ⋊ (M ∩ L) = M`.
have h1 : Nat.card ↥(ctr.K.subgroupOf ctr.M) * (ctr.K.subgroupOf ctr.M).index = Nat.card ↥ctr.M :=
Subgroup.card_mul_index _
have h2 : (ctr.K.subgroupOf ctr.M).index = Nat.card ↥((ctr.M ⊓ data.L).subgroupOf ctr.M) :=
hcompl.symm.index_eq_card
have h3 : Nat.card ↥(ctr.K.subgroupOf ctr.M) = Nat.card ↥ctr.K :=
Nat.card_congr (Subgroup.subgroupOfEquivOfLe hKM).toEquiv
have h4 : Nat.card ↥((ctr.M ⊓ data.L).subgroupOf ctr.M) = Nat.card ↥(ctr.M ⊓ data.L) :=
Nat.card_congr (Subgroup.subgroupOfEquivOfLe inf_le_left).toEquiv
have hMeq : Nat.card ↥ctr.M = Nat.card ↥ctr.K * Nat.card ↥(ctr.M ⊓ data.L) := by
rw [← h3, ← h4, ← h2, h1]
-- `|M ∩ L| ≤ |L_F|` since `M ∩ L ≤ L_F`.
have hle : Nat.card ↥(ctr.M ⊓ data.L) ≤ Nat.card ↥(maxNilpotentNormalHall data.L) :=
Nat.le_of_dvd Nat.card_pos (Subgroup.card_dvd_of_le hsub)
rw [hMeq]
exact Nat.mul_le_mul_left _ hle
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.14)**: the character `dade.psi` is constant on the coset `x·K`.
**Assembly** (the (12.4) coset-constancy applied to the counterexample `M`): since `M` is type-I
(`ctr.M_typeI`), it carries its own Hypothesis (`exists_typeI_hypothesis`), whose kernel is
`H_M = M_F = ctr.K` (`typeF.H_eq` + `K_eq_MF`). Applying (12.4)
(`orthogonal_character_constant_on_coset`) to this `Hypothesis M` with `x = witness.x ∈ P₀ ≤ M`
gives `dade.psi(x·g) = dade.psi(x)` for `g ∈ H_M = K`, provided the two inputs:
* `horth`: `dade.psi ⊥ R_M(χ)` for `χ ∈ S_M` — the cross-group orthogonality `L ≠ M`
(`coherent_extension_constituent_orthogonal_Rset_of_nonconjugate`, since
`dade.psi = coh.extension χ_L`
lies in `ℤ[R(χ_L)]` and `R(χ_L) ⊥ R(χ_M)`); needs the coherence `coh` and `L ≠ M` in scope;
* `hxK`: `x ∉ K` — `x` is a nontrivial `p`-element and `p ∤ |K| = |M_F|` (`K` is the `p'`-Hall
`M_F`). -/
theorem psi_constant_on_xK [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
{ctr : CounterexampleHypothesis (G := G)} {L : Subgroup G}
(hyp : Hypothesis L)
(coh : OddOrder.Peterfalvi.S07.IsCoherent hyp.tau hyp.Sset hyp.A)
(witness : RankTwoWitnessData ctr) (dade : DadeNotation hyp)
{chi0 : IrreducibleCharacter ↥L}
(data0 : CharacterDecompositionData hyp (chi0 : ClassFunction ↥L ℂ))
(hchi0 : chi0 ∈ data0.constituents)
(hchi0_mem : (chi0 : ClassFunction ↥L ℂ) ∈ hyp.Sset)
(hpsi : dade.psi = coh.extension (chi0 : ClassFunction ↥L ℂ))
(hLM : ¬ ∃ g : G, MulAut.conj g • L = ctr.M) :
∀ g : G, g ∈ ctr.K → dade.psi (witness.x * g) = dade.psi witness.x := by
classical
obtain ⟨hypM⟩ := exists_typeI_hypothesis hG ctr.M_maximal ctr.M_typeI
have hHK : hypM.H = ctr.K := hypM.typeI.typeF.H_eq.trans ctr.K_eq_MF.symm
have data_M : ∀ χ ∈ hypM.Sset, CharacterDecompositionData hypM χ :=
fun χ hχ => (character_decomposition_and_dade_domain hG hypM hχ).choose
-- (12.3)/(5.5) cross-group orthogonality `dade.psi ⊥ R_M` (the genuine content; `M ≠ L`):
-- `dade.psi = coh.extension χ₀ ∈ ℤ[R(χ₀)]` and `R(χ₀) ⊥ R_M` since `L ≠ M`.
have horth : ∀ χ (hχ : χ ∈ hypM.Sset), ∀ α ∈ Rset (data_M χ hχ),
ClassFunction.inner dade.psi α = 0 := by
intro χ hχ α hα
rw [hpsi]
exact coherent_extension_constituent_orthogonal_Rset_of_nonconjugate hG hyp coh data0 hchi0
hchi0_mem hypM hLM (data_M χ hχ) α hα
have hxM : witness.x ∈ ctr.M := ctr.P0_le_M witness.x_mem_P0
-- `x ∉ K`: nontrivial `p`-element, `p ∤ |K|` (K = M_F is the `p'`-Hall since `p ∣ [M:M_F]`).
have hxK : witness.x ∉ hypM.H := by
rw [hHK]
intro hxmem
have : Fact ctr.p.Prime := ⟨ctr.p_prime⟩
have hord : orderOf witness.x = ctr.p :=
orderOf_eq_prime witness.x_mem_omega1 witness.x_ne_one
-- `p = orderOf x ∣ |K|`.
have hpK : ctr.p ∣ Nat.card ↥ctr.K := by
have hd := orderOf_dvd_natCard (⟨witness.x, hxmem⟩ : ↥ctr.K)
have he : orderOf (⟨witness.x, hxmem⟩ : ↥ctr.K) = ctr.p := by
rw [← hord]
exact (orderOf_injective ctr.K.subtype ctr.K.subtype_injective _).symm
rwa [he] at hd
-- `Coprime |K| [M:K]` (Hall) and `p ∣ [M:K]` ⟹ contradiction.
have hKleM : ctr.K ≤ ctr.M :=
ctr.K_eq_MF ▸ OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le ctr.M
have hcard : Nat.card ↥(ctr.K.subgroupOf ctr.M) = Nat.card ↥ctr.K :=
Nat.card_congr (Subgroup.subgroupOfEquivOfLe hKleM).toEquiv
have hcop : Nat.Coprime (Nat.card ↥ctr.K) (ctr.K.relIndex ctr.M) := by
have h := (OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_isHall ctr.M).coprime_index
rw [← ctr.K_eq_MF, hcard] at h
rwa [Subgroup.relIndex]
exact Nat.Prime.not_dvd_one ctr.p_prime (hcop ▸ Nat.dvd_gcd hpK ctr.p_dvd_index)
intro g hg
exact orthogonal_character_constant_on_coset hG hypM data_M horth hxM hxK g (hHK ▸ hg)
/-- The witness `x` lies in `L` (via `x ∈ P₀ ⊆ L_F ≤ L`, Peterfalvi (12.10)). -/
theorem witness_x_mem_L [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) :
data.x ∈ data.L :=
OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_le data.L
(witness_P0_le_kernel hG hnoV data data.x_mem_P0)
/-- **Peterfalvi (12.1)/(12.10) for the witness `L`: `A(L) = L_F^#`** — the (12.7)-free form.
The witness `L` is Frobenius by (12.10) (`witness_L_frobenius`, proved from minimality alone), so
`typeIA_eq_sharp_of_frobenius` applies directly. The general `Hypothesis.typeIA_eq_sharp` routes
through (12.7) `typeI_frobenius` = `pi_empty` = the (12.16) contradiction itself, so it must not
be used in the (12.16) supply chain (circularity). -/
theorem witness_typeIA_eq_sharp [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) :
OddOrder.GroupTheory.typeIA data.L hyp.typeI = (hyp.typeI.typeF.H : Set G) \ {1} := by
obtain ⟨frob, -⟩ := witness_L_frobenius hG hnoV data
have hKf : frob.typeI.typeF.H = hyp.typeI.typeF.H := by
rw [frob.typeI.typeF.H_eq, hyp.typeI.typeF.H_eq]
exact hyp.typeIA_eq_sharp_of_frobenius (hKf ▸ frob.frobenius)
/-- The witness `x` lies in the type-I support `A(L) = L_F^#` (Peterfalvi (12.1)/(12.10)):
`x ∈ P₀ ⊆ L_F = typeF.H` and `x ≠ 1`. -/
theorem witness_x_mem_typeIA [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) :
data.x ∈ OddOrder.GroupTheory.typeIA data.L hyp.typeI := by
rw [witness_typeIA_eq_sharp hG hnoV data hyp]
refine Set.mem_sdiff_of_mem ?_ (by simpa using data.x_ne_one)
rw [SetLike.mem_coe, hyp.typeI.typeF.H_eq]
exact witness_P0_le_kernel hG hnoV data data.x_mem_P0
/-- **The witness maximal pin `𝓜(C_G(x)) = {M}`** (Peterfalvi (12.9)/(12.14), the `N[x] = M`
identification). The witness `x ∈ Ω₁(P₀)^# ⊆ L_F = L_σ` is a `σ`-sharp element of `L` escaping
`L` (`C_G(x) ⊄ L`, (12.9)), so the BG Theorem-D singleton
(`maximalSubgroupsContaining_centralizer_eq_singleton_of_sigmaSharp_escape`) pins a unique
maximal over `C_G(x)`; it is `M` because `C_G(x) ≤ N_G(⟨x⟩) ≤ M` ((12.9)). -/
theorem witness_maximalContaining_centralizer_eq_singleton [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) :
OddOrder.GroupTheory.maximalSubgroupsContaining
(Subgroup.centralizer ({data.x} : Set G)) = {ctr.M} := by
classical
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 hxσ : data.x ∈ OddOrder.BG.Ch4.S14.sigmaSharp data.L :=
⟨hLF_eq ▸ witness_P0_le_kernel hG hnoV data data.x_mem_P0, by simpa using data.x_ne_one⟩
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 ∈ OddOrder.GroupTheory.maximalSubgroupsContaining
(Subgroup.centralizer ({data.x} : Set G)) := ⟨ctr.M_maximal, hCM⟩
rw [hN₀] at hMin
exact hMin
rw [hN₀, hMN₀]
/-- **Peterfalvi (12.14), `R(x) = C_K(x) ⊆ K`** (Definition (8.14) at the witness): the faithful
Dade kernel of the witness `x` lands in `K = M_F`. On the escaping branch the kernel is
`R(x) = (N[x])_σ ⊓ C_G(x)` with `N[x] = M` (the singleton pin
`witness_maximalContaining_centralizer_eq_singleton`) and `M_σ = M_F = K` for the type-I
counterexample (`MF_eq_Msigma`); the non-escaping branch is `⊥`. Stated for an arbitrary
support set `A` (the escaping branch does not depend on it). -/
theorem witness_ftSupportKernel_le_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)
{A : Set G} :
OddOrder.Peterfalvi.S10.ftSupportKernel data.L A data.x ≤ ctr.K := by
classical
by_cases hesc : data.x ∈ OddOrder.GroupTheory.escapingCentralizerSet data.L A
· rw [OddOrder.Peterfalvi.S10.ftSupportKernel_eq_of_escaping hesc]
-- The branch condition of `FT_signalizerBase`: `1 < |𝓜_σ(x)|` and `𝓜(C_G(x)) ≠ ∅`.
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 hsing := witness_maximalContaining_centralizer_eq_singleton hG hnoV data
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)
have hne : (OddOrder.GroupTheory.maximalSubgroupsContaining
(Subgroup.centralizer ({data.x} : Set G))).Nonempty := by
rw [hsing]
exact ⟨ctr.M, rfl⟩
have hcond : 1 < (OddOrder.BG.Ch4.S14.maximalSigmaSubgroupsOfElement data.x).ncard ∧
(OddOrder.GroupTheory.maximalSubgroupsContaining
(Subgroup.centralizer ({data.x} : Set G))).Nonempty := ⟨hgt, hne⟩
-- `N[x] = M`: the chosen member of the singleton `𝓜(C_G(x)) = {M}`.
have hb : OddOrder.BG.Ch4.S16.FT_signalizerBase data.x = hcond.2.choose := dif_pos hcond
have hch : hcond.2.choose ∈ ({ctr.M} : Set (Subgroup G)) :=
(Set.ext_iff.mp hsing _).mp hcond.2.choose_spec
have hbase : OddOrder.BG.Ch4.S16.FT_signalizerBase data.x = ctr.M :=
hb.trans (Set.mem_singleton_iff.mp hch)
calc OddOrder.BG.Ch4.S16.FT_signalizer data.x
≤ OddOrder.BG.Ch3.S10.Msigma (OddOrder.BG.Ch4.S16.FT_signalizerBase data.x) :=
inf_le_left
_ = ctr.K := by rw [hbase, ← MF_eq_Msigma hG ctr]
· rw [OddOrder.Peterfalvi.S10.ftSupportKernel_eq_bot_of_not_escaping hesc]
exact bot_le
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.14), the `ψ^ρ(x) = ψ(x)` collapse**: any `χ ∈ CF(G)` constant on the coset
`x·K` (e.g. `ψ = dade.psi`, via `psi_constant_on_xK`) has `ρ`-average at the witness `x` equal to
its value: `χ^ρ(x) = χ(x)`. The local Dade kernel is `H(x) = R(x) ⊆ K`
(`H_eq_ftSupportKernel` + `witness_ftSupportKernel_le_K`), so the `x·K`-constancy restricts to
the `H(x)`-coset and the `ρ`-average collapses (`chiRho_apply_eq_of_forall_coset`). -/
theorem witness_chiRho_apply_eq_of_forall_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)
(hyp : Hypothesis data.L) (χ : ClassFunction G ℂ)
(hconst : ∀ g : G, g ∈ ctr.K → χ (data.x * g) = χ data.x)
(hxL : data.x ∈ data.L) :
hyp.toHypothesis71.chiRho χ ⟨data.x, hxL⟩ = χ data.x := by
have hxA : ((⟨data.x, hxL⟩ : ↥data.L) : G) ∈ OddOrder.GroupTheory.typeIA data.L hyp.typeI :=
witness_x_mem_typeIA hG hnoV data hyp
refine OddOrder.Peterfalvi.S09.Hypothesis71.chiRho_apply_eq_of_forall_coset
hyp.toHypothesis71 χ hxA ?_
intro y hy
refine hconst y (witness_ftSupportKernel_le_K hG hnoV data
(A := OddOrder.GroupTheory.typeIA data.L hyp.typeI) ?_)
rw [← hyp.dadeData.H_eq_ftSupportKernel ⟨data.x, hxA⟩]
exact hy
/-- **The witness `L` is not conjugate to `M`** ((12.14): "`p` divides `|L_s|` but not `|M_s|`").
`P₀ ≤ L_F = L_σ` gives `p ∣ |L_σ|`, while `|M_σ| = |K|` is prime to `p` (`p_not_dvd_card_K`);
a conjugation `L^g = M` would carry `L_σ` onto `M_σ` (`Msigma_conj_smul`), contradiction. -/
theorem witness_L_not_conj_M [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) :
¬ ∃ g : G, MulAut.conj g • data.L = ctr.M := by
rintro ⟨g, hg⟩
have : Fact ctr.p.Prime := ⟨ctr.p_prime⟩
-- `p ∣ |P₀| ∣ |L_σ|`.
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 hP0le : ctr.P0 ≤ OddOrder.BG.Ch3.S10.Msigma data.L :=
hLF_eq ▸ witness_P0_le_kernel hG hnoV data
have : Nontrivial ↥ctr.P0 := by
by_contra h
rw [not_nontrivial_iff_subsingleton] at h
exact ctr.P0_noncyclic isCyclic_of_subsingleton
obtain ⟨n, hn⟩ := IsPGroup.iff_card.mp ctr.P0_pGroup
have hpP0 : ctr.p ∣ Nat.card ↥ctr.P0 := by
rcases Nat.eq_zero_or_pos n with h0 | hpos
· rw [h0, pow_zero] at hn
exact absurd hn Finite.one_lt_card.ne'
· rw [hn]
exact dvd_pow_self _ hpos.ne'
have hpLσ : ctr.p ∣ Nat.card ↥(OddOrder.BG.Ch3.S10.Msigma data.L) :=
hpP0.trans (Subgroup.card_dvd_of_le hP0le)
-- Transport along the conjugation: `|L_σ| = |(L^g)_σ| = |M_σ| = |K|`.
have hcard : Nat.card ↥(OddOrder.BG.Ch3.S10.Msigma data.L)
= Nat.card ↥(OddOrder.BG.Ch3.S10.Msigma ctr.M) := by
rw [← hg, OddOrder.BG.Ch4.S14.Msigma_conj_smul]
exact Nat.card_congr
(Subgroup.equivSMul (MulAut.conj g) (OddOrder.BG.Ch3.S10.Msigma data.L)).toEquiv
rw [hcard, ← MF_eq_Msigma hG ctr] at hpLσ
exact p_not_dvd_card_K ctr hpLσ
/-- **The counterexample prime is at least `3`** (`p` is an odd prime: `p ∣ [M:K] ∣ |G|` odd). -/
theorem counterexample_three_le_p [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
(ctr : CounterexampleHypothesis (G := G)) : 3 ≤ ctr.p := by
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 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
/-- **Peterfalvi (12.14), the `p² ≤ h` size input**: `P₀ ⊆ H = L_F` is a noncyclic `p`-group, so
`p² ∣ |P₀| ∣ |H|`. -/
theorem witness_p_sq_le_card_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)
(hyp : Hypothesis data.L) :
ctr.p ^ 2 ≤ Nat.card ↥(hyp.typeI.typeF.H) := by
have : Fact ctr.p.Prime := ⟨ctr.p_prime⟩
obtain ⟨n, hn⟩ := IsPGroup.iff_card.mp ctr.P0_pGroup
-- Noncyclic forces `n ≥ 2` (order `1` and order `p` groups are cyclic).
have hn2 : 2 ≤ n := by
by_contra h
push Not at h
interval_cases n
· rw [pow_zero] at hn
have : Subsingleton ↥ctr.P0 := Nat.card_eq_one_iff_unique.mp hn |>.1
exact ctr.P0_noncyclic isCyclic_of_subsingleton
· rw [pow_one] at hn
exact ctr.P0_noncyclic (isCyclic_of_prime_card hn)
have hp2 : ctr.p ^ 2 ∣ Nat.card ↥ctr.P0 := hn ▸ pow_dvd_pow _ hn2
have hP0H : ctr.P0 ≤ hyp.typeI.typeF.H := by
rw [hyp.typeI.typeF.H_eq]
exact witness_P0_le_kernel hG hnoV data
calc ctr.p ^ 2 ≤ Nat.card ↥ctr.P0 := Nat.le_of_dvd Nat.card_pos hp2
_ ≤ Nat.card ↥(hyp.typeI.typeF.H) :=
Nat.le_of_dvd Nat.card_pos (Subgroup.card_dvd_of_le hP0H)
/-- **Peterfalvi (12.12), the `2e ≤ p + 1` bound in kernel-index form**: the witness complement
order `e = [L : H]` satisfies `2e ≤ p + 1` (`two_mul_card_complement_le`, transported along
`IsComplement'.index_eq_card` and the kernel identification `typeF.H = L_F`). -/
theorem witness_two_mul_index_le_p_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)
(hyp : Hypothesis data.L) :
2 * ((hyp.typeI.typeF.H).subgroupOf data.L).index ≤ ctr.p + 1 := by
obtain ⟨fdata, -⟩ := witness_L_frobenius hG hnoV data
have h2e := two_mul_card_complement_le hG hnoV data fdata
have hidx : ((fdata.typeI.typeF.H).subgroupOf data.L).index = Nat.card ↥fdata.complement :=
fdata.frobenius.isComplement.symm.index_eq_card
have hHeq : fdata.typeI.typeF.H = hyp.typeI.typeF.H := by
rw [fdata.typeI.typeF.H_eq, hyp.typeI.typeF.H_eq]
rw [hHeq] at hidx
omega
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
open OddOrder.Peterfalvi.S09 in
open OddOrder.Peterfalvi.S09.Cert in
/-- **Peterfalvi (7.8.a) for the witness `L`: `⟨ζ_0^ν, 1_G⟩ = 0`** (`hzeta0nu`, the last input to
the (7.8.b) bound `hB`). The abstract `IsCoherent` does not carry orthogonality to `1_G`, but it
is recovered from the **complex conjugate** `ζ̄_0 = Ind θ̄_0 ∈ S` (`Sset_closedUnderConjugate`) — a
second member of the *same degree*, distinct from `ζ_0` because `L` has odd order (no nontrivial
real irreducible, `not_isReal_of_ne_trivial_of_odd_card'`).
`coherence_extension_orthogonal_constOne`
then forces `⟨ν ζ_0, 1_G⟩ = 0`. Holds for **any** nontrivial `θ_0` (degree-`e`/linearity
unused). -/
theorem witness_L_hzeta0nu [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
{L : Subgroup G} (hyp : Hypothesis L) {C : Subgroup ↥L}
(hfrob : OddOrder.Isaacs.Ch06.IsFrobeniusGroup ↥L ((hyp.typeI.typeF.H).subgroupOf L) C)
(coh : OddOrder.Peterfalvi.S07.IsCoherent hyp.tau hyp.Sset hyp.A)
(hAH : typeIA L hyp.typeI = ((hyp.typeI.typeF.H) : Set G) \ {1})
(θ0 : IrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf L))
(hθ0 : θ0 ≠ trivialIrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf L)) :
ClassFunction.inner
(coh.extension (ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L)
(θ0 : ClassFunction _ ℂ))) (Hypothesis71.constOne G) = 0 := by
classical
have hodd : Odd (Nat.card ↥L) := hG.odd.of_dvd_nat (Subgroup.card_subgroup_dvd_card L)
have hKnormal : ((hyp.typeI.typeF.H).subgroupOf L).Normal := by
rw [hyp.typeI.typeF.H_eq]
exact OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_subgroupOf_normal L
-- The complex conjugate character `θ̄_0` is again nontrivial irreducible. Introduce it
-- **opaquely** (via `obtain`, not `let`), carrying only its coercion `↑θ̄_0 = (↑θ_0)‾`: a `let`
-- gets its coercion re-unfolded inside every `induce` coset sum, blowing the `whnf` budget.
obtain ⟨θ0', hθ0'coe⟩ :
∃ t : IrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf L),
(t : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ)
= (θ0 : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ).conj :=
⟨⟨(θ0 : ClassFunction _ ℂ).conj, θ0.isIrreducible.conj⟩, rfl⟩
have hθ0' : θ0' ≠ trivialIrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf L) := by
intro h
apply hθ0
have hcoe : (θ0 : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ).conj
= trivialClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) := by
rw [← hθ0'coe]
have h2 := congrArg
(fun c : IrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf L) =>
(c : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ)) h
simpa using h2
apply Subtype.ext
change (θ0 : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ)
= trivialClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L)
rw [← ClassFunction.conj_conj (θ0 : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ), hcoe]
exact trivialClassFunction_isReal
-- The two members `ζ_0 = Ind θ_0`, `ζ̄_0 = Ind θ̄_0 ∈ S`.
have hmem0 : ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L)
(θ0 : ClassFunction _ ℂ) ∈ hyp.Sset := ⟨θ0, hθ0, rfl⟩
have hmem0' : ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L)
(θ0' : ClassFunction _ ℂ) ∈ hyp.Sset := ⟨θ0', hθ0', rfl⟩
-- Norms `= 1` (Frobenius), orthogonality to `1_L`, irreducibility.
have hnorm0 := inner_self_induce_eq_one_of_frobeniusGroup hfrob θ0 hθ0
have hnorm0' := inner_self_induce_eq_one_of_frobeniusGroup hfrob θ0' hθ0'
have h1_0 := inner_induce_constOne_eq_zero ((hyp.typeI.typeF.H).subgroupOf L) θ0 hθ0
have h1_0' := inner_induce_constOne_eq_zero ((hyp.typeI.typeF.H).subgroupOf L) θ0' hθ0'
-- `⟨ζ_0, ζ̄_0⟩ = 0` (odd-order Frobenius: `ζ_0` non-real), from the reusable general helper
-- (`hθ0'coe : ↑θ̄_0 = (↑θ_0)‾` reindexes it to `θ_0'`).
have horth : ClassFunction.inner
(ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L) (θ0 : ClassFunction _ ℂ))
(ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L) (θ0' : ClassFunction _ ℂ)) = 0 := by
rw [hθ0'coe]
exact inner_induce_conj_eq_zero_of_frobenius_of_odd hodd hfrob θ0 hθ0
-- The equal-degree difference is `A(L) = H#`-supported.
have hdeg' : ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L) (θ0' : ClassFunction _ ℂ)
(1 : ↥L)
= 1 * ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L) (θ0 : ClassFunction _ ℂ)
(1 : ↥L) := by
rw [one_mul, ClassFunction.induce_apply_one, ClassFunction.induce_apply_one]
congr 1
rw [hθ0'coe, ClassFunction.conj_apply]
obtain ⟨n, -, hn⟩ := irreducibleCharacter_apply_one_eq_pos_natCast θ0
rw [hn, star_natCast]
have hsupp : (ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L) (θ0' : ClassFunction _ ℂ)
- ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L)
(θ0 : ClassFunction _ ℂ)).support ⊆ hyp.A := by
have hds := induce_diff_support (K := (hyp.typeI.typeF.H).subgroupOf L) θ0' θ0 1 hdeg'
rw [one_smul] at hds
intro x hx
have hxd := hds hx
rw [Set.mem_sdiff, SetLike.mem_coe, Set.mem_singleton_iff] at hxd
exact (mem_supportInSubgroup_sharp_subgroupOf_iff hyp.typeI.typeF.H hAH x).mpr ⟨hxd.1, hxd.2⟩
-- The Dade `⊥ 1_G` transport and `ℂ`-linearity of `τ = hyp.tau`.
have htau1 : ∀ φ : ClassFunction ↥L ℂ, φ.support ⊆ hyp.A →
ClassFunction.inner (hyp.tau φ) (Hypothesis71.constOne G)
= ClassFunction.inner φ (Hypothesis71.constOne L) := by
intro φ hφ
rw [show hyp.tau φ = hyp.dadeData.dade.dadeMap
⟨φ, (ClassFunction.mem_supportedSubmodule).mpr hφ⟩ from
OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_apply_of_support hyp.dadeData.dade
(hyp.dadeData.dade.fullDadeIsometryData) hφ]
exact inner_tau_supported_constOne hyp.toHypothesis71
⟨φ, (ClassFunction.mem_supportedSubmodule).mpr hφ⟩
have hτ_smul : ∀ (c : ℂ) (x : ClassFunction ↥L ℂ), hyp.tau (c • x) = c • hyp.tau x :=
dadeIntegralCharacterMap_smul_complex hyp.dadeData.dade
(hyp.dadeData.dade.fullDadeIsometryData)
exact coherence_extension_orthogonal_constOne coh hτ_smul htau1 hmem0 hmem0'
hnorm0 hnorm0' horth hsupp h1_0 h1_0'
set_option maxHeartbeats 800000 in
-- raised heartbeat budget for the heavy elaboration below
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
open OddOrder.Peterfalvi.S09 in
open OddOrder.Peterfalvi.S09.Cert in
/-- **Peterfalvi (12.14) for the witness, full form: `ψ(x) = χ(x)`.** The witness `L` of (12.9)
carries a (12.13) Dade calculation `dade` — with `χ = dade.chi = Ind θ_0` the distinguished
degree-`e` member and `ψ = dade.psi = χ^{τ₁}` its coherent extension — whose value at the witness
`x` is `ψ(x) = χ(x)`.
Peterfalvi's proof, assembled from the three proven pieces:
* `ψ` is constant on `x·K` ((12.3)+(12.4)+(5.5): `L ≁ M` by `witness_L_not_conj_M`, then
`psi_constant_on_xK`);
* `ψ^ρ(x) = ψ(x)` (the `ρ`-collapse: `R(x) = C_K(x) ⊆ K` per Definition (8.14), so constancy on
`x·K` collapses the `ρ`-average — `witness_chiRho_apply_eq_of_forall_K`);
* `ψ^ρ(x) = χ(x)` (the (7.8.a) `a = 0` counting and the two-sided (7.7.a) evaluation —
`chiRho_nu_zeta0_apply_eq_zeta0_ofDade`, fed by the witness `Hypothesis78` assembly of
`witness_L_hypothesis78`/`witness_L_zeta_bound` plus the size inputs `3 ≤ p`
(`counterexample_three_le_p`), `p² ≤ h` (`witness_p_sq_le_card_kernel`), and `2e ≤ p + 1`
(`witness_two_mul_index_le_p_add_one`)).
This is the (12.16) `h_psix` feed: with the `L`-side (1.10.a) congruence
`χ(x) ≡ χ(1) = e (mod 1 − ε)` it gives `ψ(x) ≡ e (mod 1 − ε)` without the coherent-extension
degree identity `ψ(1) = e`. -/
theorem witness_dade_psi_apply_x_eq_chi [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)
(coh : OddOrder.Peterfalvi.S07.IsCoherent hyp.tau hyp.Sset hyp.A)
(dade : DadeNotation hyp),
dade.psi = coh.extension dade.chi ∧
dade.e = ((hyp.typeI.typeF.H).subgroupOf data.L).index ∧
dade.psi ∈ ZIrr G ∧
dade.chi ∈ ZIrr ↥data.L ∧
dade.psi data.x = dade.chi ⟨data.x, witness_x_mem_L hG hnoV data⟩ := by
classical
obtain ⟨hyp, C, hC, hNonTI⟩ := witness_L_hypothesis_frobenius hG hnoV data
obtain ⟨coh⟩ : Nonempty (OddOrder.Peterfalvi.S07.IsCoherent hyp.tau hyp.Sset hyp.A) := by
rcases hyp.typeI.alternative with hTI | hab | hexp
· exact absurd hTI hNonTI
· exact frobenius_typeI_coherent_of_abelianKernel hG hyp ⟨C, hC⟩ hab
· exact frobenius_typeI_coherent_of_cyclicQuotient hG hyp ⟨C, hC⟩ hexp
have hHL : hyp.typeI.typeF.H ≤ data.L := hyp.typeI.typeF.H_le
have hKnormal : ((hyp.typeI.typeF.H).subgroupOf data.L).Normal := by
rw [hyp.typeI.typeF.H_eq]
exact OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_subgroupOf_normal data.L
have : (hyp.H.subgroupOf data.L).Normal := hKnormal
have hAH : typeIA data.L hyp.typeI = (hyp.typeI.typeF.H : Set G) \ {1} :=
witness_typeIA_eq_sharp hG hnoV data hyp
have hHnorm : ∀ (l : ↥data.L) {h : G}, h ∈ hyp.typeI.typeF.H →
(l : G) * h * (l : G)⁻¹ ∈ hyp.typeI.typeF.H := by
intro l h hh
have hhL : h ∈ data.L := hHL hh
have hmem : (⟨h, hhL⟩ : ↥data.L) ∈ (hyp.typeI.typeF.H).subgroupOf data.L :=
(Subgroup.mem_subgroupOf).mpr hh
have hconj := hKnormal.conj_mem ⟨h, hhL⟩ hmem l
rw [Subgroup.mem_subgroupOf] at hconj
simpa using hconj
obtain ⟨n, θ, ind1H, hind1H, hdeg0, htriv, hinj, hcover⟩ := exists_witness_placed_family hyp
have hSmem : ∀ i, i ≠ ind1H →
ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L) (θ i : ClassFunction _ ℂ)
∈ hyp.Sset := by
intro i hi
refine ⟨θ i, fun htriv_i => hi (hinj ?_), rfl⟩
change ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L) (θ i : ClassFunction _ ℂ)
= ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L) (θ ind1H : ClassFunction _ ℂ)
rw [htriv_i, htriv]
have hθ0_ne : θ 0 ≠ trivialIrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf data.L) := by
intro h
refine hind1H (hinj ?_).symm
change ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L) (θ 0 : ClassFunction _ ℂ)
= ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L) (θ ind1H : ClassFunction _ ℂ)
rw [h, htriv]
let d : Fin (n + 1) → ℂ :=
fun i => (θ i : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf data.L) ℂ)
(1 : ↥((hyp.typeI.typeF.H).subgroupOf data.L))
have hd : ∀ i, d i = (θ i : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf data.L) ℂ)
(1 : ↥((hyp.typeI.typeF.H).subgroupOf data.L)) := fun _ => rfl
have hdeg : ∀ i, ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ i : ClassFunction _ ℂ) (1 : ↥data.L)
= d i * ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ 0 : ClassFunction _ ℂ) (1 : ↥data.L) := by
intro i
rw [ClassFunction.induce_apply_one ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ i : ClassFunction _ ℂ), hdeg0, hd i]
ring
have hdeg_match : ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ 0 : ClassFunction _ ℂ) (1 : ↥data.L)
= ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ ind1H : ClassFunction _ ℂ) (1 : ↥data.L) := by
rw [hdeg0, htriv]
change (((hyp.typeI.typeF.H).subgroupOf data.L).index : ℂ)
= ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(trivialClassFunction ↥((hyp.typeI.typeF.H).subgroupOf data.L)) (1 : ↥data.L)
rw [induce_trivialChar_apply_eq_index _ (Subgroup.one_mem _)]
have psi_support : ∀ i, (ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ i : ClassFunction _ ℂ)
- d i • ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ 0 : ClassFunction _ ℂ)).support
⊆ OddOrder.Peterfalvi.S04.supportInSubgroup (typeIA data.L hyp.typeI) data.L := by
intro i
refine (induce_diff_support (θ i) (θ 0) (d i) (hdeg i)).trans ?_
intro x hx
rw [Set.mem_sdiff, SetLike.mem_coe, Set.mem_singleton_iff] at hx
exact (mem_supportInSubgroup_sharp_subgroupOf_iff hyp.typeI.typeF.H hAH x).mpr ⟨hx.1, hx.2⟩
have hnu_isometry : ∀ i j : Fin (n + 1), i ≠ ind1H → j ≠ ind1H →
ClassFunction.inner (coh.extension
(ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L) (θ i : ClassFunction _ ℂ)))
(coh.extension
(ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L) (θ j : ClassFunction _ ℂ)))
= ClassFunction.inner
(ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L) (θ i : ClassFunction _ ℂ))
(ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L) (θ j : ClassFunction _
ℂ)) :=
fun i j hi hj => coherence_extension_inner_eq_on_family coh (hSmem i hi) (hSmem j hj)
have hagree : ∀ i : Fin (n + 1), i ≠ 0 → i ≠ ind1H →
hyp.toHypothesis71.τ ⟨ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ i : ClassFunction _ ℂ)
- d i • ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ 0 : ClassFunction _ ℂ), psi_support i⟩
= coh.extension (ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ i : ClassFunction _ ℂ))
- d i • coh.extension (ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ 0 : ClassFunction _ ℂ)) := by
intro i _ hi_ind
obtain ⟨deg_i, -, hdeg_i_eq⟩ := irreducibleCharacter_apply_one_eq_pos_natCast (θ i)
exact coherence_hagree_dadeMap hyp.dadeData.dade coh
(hSmem i hi_ind) (hSmem 0 (Ne.symm hind1H)) (m0 := 1) (mi := deg_i) (by norm_num)
(by rw [hd i, hdeg_i_eq, Nat.cast_one, div_one]) (psi_support i)
-- The distinguished member `ζ_0 = Ind θ_0 ∈ S` and its (12.13) Dade calculation.
have hmem0 : ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ 0 : ClassFunction _ ℂ) ∈ hyp.Sset := hSmem 0 (Ne.symm hind1H)
set dade : DadeNotation hyp := dadeNotation_of_coherence hyp coh
(ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L) (θ 0 : ClassFunction _ ℂ))
hmem0 ((hyp.typeI.typeF.H).subgroupOf data.L).index hdeg0 with hdade
have hψZ : dade.psi ∈ ZIrr G := coh.extension_mem_ZIrr _ (Submodule.subset_span hmem0)
-- (12.14) constancy: `ψ` is constant on `x·K` (via `L ≁ M`).
have hchi0_irr : IsIrreducibleCharacter (ClassFunction.induce
((hyp.typeI.typeF.H).subgroupOf data.L) (θ 0 : ClassFunction _ ℂ)) :=
Sset_isIrreducibleCharacter hyp hC hmem0
obtain ⟨chi0, hchi0_coe⟩ : ∃ t : IrreducibleCharacter ↥data.L,
(t : ClassFunction ↥data.L ℂ) = ClassFunction.induce
((hyp.typeI.typeF.H).subgroupOf data.L) (θ 0 : ClassFunction _ ℂ) :=
⟨⟨_, hchi0_irr⟩, rfl⟩
have hchi0_mem : (chi0 : ClassFunction ↥data.L ℂ) ∈ hyp.Sset := by
rw [hchi0_coe]; exact hmem0
have data0 : CharacterDecompositionData hyp (chi0 : ClassFunction ↥data.L ℂ) :=
(character_decomposition_and_dade_domain hG hyp hchi0_mem).choose
have hchi0_cons : chi0 ∈ data0.constituents := by
obtain ⟨φ, hφcoe, hφmem⟩ := Sset_self_mem_constituents hyp hC hchi0_mem data0
have hφeq : φ = chi0 := Subtype.ext hφcoe
rwa [hφeq] at hφmem
have hpsi_chi0 : dade.psi = coh.extension (chi0 : ClassFunction ↥data.L ℂ) := by
rw [hchi0_coe, hdade]
rfl
have hconst : ∀ g : G, g ∈ ctr.K → dade.psi (data.x * g) = dade.psi data.x :=
psi_constant_on_xK hG hyp coh data dade data0 hchi0_cons hchi0_mem hpsi_chi0
(witness_L_not_conj_M hG hnoV data)
-- The `ρ`-collapse `ψ^ρ(x) = ψ(x)`.
have hxL : data.x ∈ data.L := witness_x_mem_L hG hnoV data
have hcollapse : hyp.toHypothesis71.chiRho dade.psi ⟨data.x, hxL⟩ = dade.psi data.x :=
witness_chiRho_apply_eq_of_forall_K hG hnoV data hyp dade.psi hconst hxL
-- The (7.8) input `a`: `(β, ζ_0^ν) + 1 ∈ ℤ`.
set H78 := hypothesis78OfDade hyp.toHypothesis71
(hyp.dadeData.dade.fullDadeIsometryData).toDadeIsometryData.isDadeIsometry
hyp.typeI.typeF.H hHL hHnorm hAH θ hinj hcover d psi_support hdeg ind1H hind1H htriv
hdeg_match coh.extension hnu_isometry hagree with hH78def
obtain ⟨a, ha⟩ := exists_betaDecomp_a H78
(Submodule.sub_mem _
(ClassFunction.induce_mem_ZIrr _ (θ ind1H).property.mem_ZIrr)
(ClassFunction.induce_mem_ZIrr _ (θ 0).property.mem_ZIrr))
(coh.extension_mem_ZIrr _ (Submodule.subset_span hmem0))
rw [hH78def] at ha
-- Size inputs: `3 ≤ p`, `p² ≤ h`, `2e ≤ p + 1`.
have hp3 : 3 ≤ ctr.p := counterexample_three_le_p hG ctr
have hph : (ctr.p : ℝ) ^ 2 ≤ (Nat.card ↥(hyp.typeI.typeF.H) : ℝ) := by
exact_mod_cast witness_p_sq_le_card_kernel hG hnoV data hyp
have h2e : 2 * ((((hyp.typeI.typeF.H).subgroupOf data.L).index : ℝ)) ≤ (ctr.p : ℝ) + 1 := by
exact_mod_cast witness_two_mul_index_le_p_add_one hG hnoV data hyp
-- The two-sided (7.7.a) evaluation `ψ^ρ(x) = ζ_0(x)`.
have hxA : ((⟨data.x, hxL⟩ : ↥data.L) : G) ∈ typeIA data.L hyp.typeI :=
witness_x_mem_typeIA hG hnoV data hyp
have heval := chiRho_nu_zeta0_apply_eq_zeta0_ofDade hyp.toHypothesis71
(hyp.dadeData.dade.fullDadeIsometryData).toDadeIsometryData.isDadeIsometry
hyp.typeI.typeF.H hHL hHnorm hAH θ hinj hcover d psi_support hdeg ind1H hind1H htriv
hdeg_match coh.extension hnu_isometry hagree
(witness_L_hzeta0nu hG hyp hC coh hAH (θ 0) hθ0_ne)
(inner_self_induce_eq_one_of_frobeniusGroup hC (θ 0) hθ0_ne)
a ha (Sset_isIrreducibleCharacter hyp hC hmem0) hp3 hph h2e hxA
refine ⟨hyp, coh, dade, rfl, rfl, hψZ, IsIrreducibleCharacter.mem_ZIrr hchi0_irr, ?_⟩
calc dade.psi data.x
= hyp.toHypothesis71.chiRho dade.psi ⟨data.x, hxL⟩ := hcollapse.symm
_ = ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf data.L)
(θ 0 : ClassFunction _ ℂ) ⟨data.x, hxL⟩ := heval
_ = dade.chi ⟨data.x, hxL⟩ := rfl
/-- Virtual-character values are algebraic integers (local copy of the `S05` lemma
`isIntegral_apply_of_mem_ZIrr`, which lives in an unimported leaf): each irreducible value is a
sum of roots of unity (`character_isIntegral`), and `IsIntegral ℤ` is closed under the `ℤ`-span. -/
private theorem isIntegral_apply_of_mem_ZIrr' {φ : ClassFunction G ℂ} [Finite G]
(hφ : φ ∈ ZIrr G) (g : G) : IsIntegral ℤ (φ g) := by
induction hφ using Submodule.span_induction with
| mem x hx =>
obtain ⟨V, _, _, _, ρ, hchar⟩ :=
IsIrreducibleCharacter.isCharacter (mem_irreducibleCharacters.mp hx)
rw [show x g = ρ.character g from congrFun hchar g]
exact OddOrder.RepresentationTheory.character_isIntegral ρ g
| zero => rw [ClassFunction.zero_apply]; exact isIntegral_zero
| add a b _ _ ha hb => rw [ClassFunction.add_apply]; exact ha.add hb
| smul n a _ ha =>
rw [ClassFunction.zsmul_apply, zsmul_eq_mul]
exact (isIntegral_algebraMap (x := n)).mul ha
/-- **Peterfalvi (12.15), the integrality clause** (`ψ(g) ∈ ℤ` for `g ∈ K − K′`): a virtual
character `ψ ∈ ℤ[Irr G]` that is **constant on `K − K′`** takes an integer value there.
**Honest reconstruction of the (12.15) integrality** (Coq `rhoM_psi`, final `Cint_rat_Aint` step).
Two facts combine:
* `ψ(g)` is an **algebraic integer** — the value of a virtual character
(`isIntegral_apply_of_mem_ZIrr`);
* `ψ(g)` is **rational** — from the class-function inner-product identity
`|K|·⟨Res_K ψ, 1_K⟩ = |K′|·⟨Res_{K′} ψ, 1_{K′}⟩ + |K − K′|·ψ(g)`. The two inner products are
integers (`inner_mem_ZIrr_int`, since `Res ψ` and `1` are virtual characters), and `ψ` is
constant `= ψ(g)` on the `|K − K′|` elements of `K − K′` (the `hconst` hypothesis, which is the
companion "`ψ` constant on `K − K′`" clause of (12.15), proven from (12.3)/(12.5)); so `ψ(g)` is a
`ℚ`-combination of integers, i.e. rational.
A rational algebraic integer is a rational integer (`exists_int_of_isIntegral_of_mem_range_rat`).
The **constancy** hypothesis `hconst` isolates the genuine input this integrality needs; the
`ψ ∈ ZIrr G` hypothesis (the Dade image is a virtual character, by (12.13)) makes the statement
sound — for a non-virtual `ψ` the value need not be an integer. This discharges the `h_psig_int`
field of `CounterexampleDadeData` once `ψ = dade.psi` and its `K − K′`-constancy are in place. -/
theorem rhoM_integer_values [Finite G]
{ctr : CounterexampleHypothesis (G := G)}
{ψ : ClassFunction G ℂ} (hψ : ψ ∈ ZIrr G)
(hconst : ∀ g₁ g₂ : G, g₁ ∈ ctr.K → g₁ ∉ ctr.Kprime →
g₂ ∈ ctr.K → g₂ ∉ ctr.Kprime → ψ g₁ = ψ g₂) :
∀ g : G, g ∈ ctr.K → g ∉ ctr.Kprime → ∃ z : ℤ, ψ g = (z : ℂ) := by
classical
intro g hgK hgK'
-- `K′ = [K, K] ≤ K`.
have hK'K : ctr.Kprime ≤ ctr.K := ctr.Kprime_eq ▸ Subgroup.map_subtype_le _
have : Fintype ↥ctr.K := Fintype.ofFinite _
have : Fintype ↥ctr.Kprime := Fintype.ofFinite _
have : Invertible (Nat.card ↥ctr.K : ℂ) :=
invertibleOfNonzero (by exact_mod_cast (Nat.card_pos (α := ↥ctr.K)).ne')
have : Invertible (Nat.card ↥ctr.Kprime : ℂ) :=
invertibleOfNonzero (by exact_mod_cast (Nat.card_pos (α := ↥ctr.Kprime)).ne')
-- The restrictions are virtual characters; `1` is a virtual character.
have hResK : ClassFunction.restrict ctr.K ψ ∈ ZIrr ↥ctr.K :=
ClassFunction.restrict_mem_ZIrr ctr.K hψ
have hResK' : ClassFunction.restrict ctr.Kprime ψ ∈ ZIrr ↥ctr.Kprime :=
ClassFunction.restrict_mem_ZIrr ctr.Kprime hψ
have h1K : trivialClassFunction ↥ctr.K ∈ ZIrr ↥ctr.K :=
(trivialClassFunction_isIrreducible (G := ↥ctr.K)).mem_ZIrr
have h1K' : trivialClassFunction ↥ctr.Kprime ∈ ZIrr ↥ctr.Kprime :=
(trivialClassFunction_isIrreducible (G := ↥ctr.Kprime)).mem_ZIrr
-- The two inner products are integers.
obtain ⟨a, ha⟩ := ClassFunction.inner_mem_ZIrr_int hResK h1K
obtain ⟨b, hb⟩ := ClassFunction.inner_mem_ZIrr_int hResK' h1K'
-- `∑_{z:↥K} ψ(z) = |K|·⟨Res_K ψ, 1⟩`.
have hsumK : (∑ z : ↥ctr.K, ψ (z : G)) = (Nat.card ↥ctr.K : ℂ) * (a : ℂ) := by
have := ClassFunction.card_mul_inner (ClassFunction.restrict ctr.K ψ)
(trivialClassFunction ↥ctr.K)
rw [ha] at this
rw [this]
simp only [ClassFunction.innerSum, ClassFunction.restrict_apply,
trivialClassFunction_apply, star_one, mul_one]
-- `∑_{z:↥K′} ψ(z) = |K′|·⟨Res_{K′} ψ, 1⟩`.
have hsumK' : (∑ z : ↥ctr.Kprime, ψ (z : G)) = (Nat.card ↥ctr.Kprime : ℂ) * (b : ℂ) := by
have := ClassFunction.card_mul_inner (ClassFunction.restrict ctr.Kprime ψ)
(trivialClassFunction ↥ctr.Kprime)
rw [hb] at this
rw [this]
simp only [ClassFunction.innerSum, ClassFunction.restrict_apply,
trivialClassFunction_apply, star_one, mul_one]
-- Split the `↥K` sum by membership in `K′` (as a predicate on `↥K`).
set p : ↥ctr.K → Prop := fun z => (z : G) ∈ ctr.Kprime with hp
have hsplit : (∑ z : ↥ctr.K, ψ (z : G)) =
(∑ z ∈ Finset.univ.filter p, ψ (z : G)) +
(∑ z ∈ Finset.univ.filter (fun z => ¬ p z), ψ (z : G)) :=
(Finset.sum_filter_add_sum_filter_not Finset.univ p (fun z => ψ (z : G))).symm
-- The `K′`-part (filtered `↥K` sum) reindexes to the `↥K′` sum, via `{z : ↥K // (z:G)∈K′} ≃ ↥K′`.
have hpart1 : (∑ z ∈ Finset.univ.filter p, ψ (z : G)) =
∑ z : ↥ctr.Kprime, ψ (z : G) := by
-- The bijection `{z : ↥K // (z:G) ∈ K′} ≃ ↥K′`, `⟨⟨z,-⟩, hz⟩ ↦ ⟨z, hz⟩`.
let φ : {z : ↥ctr.K // p z} ≃ ↥ctr.Kprime :=
{ toFun := fun z => ⟨((z : ↥ctr.K) : G), z.2⟩
invFun := fun z => ⟨⟨(z : G), hK'K z.2⟩, z.2⟩
left_inv := fun z => by ext; rfl
right_inv := fun z => by ext; rfl }
rw [Finset.sum_subtype (Finset.univ.filter p) (Finset.mem_filter_univ (p := p))
(fun z => ψ (z : G))]
exact Fintype.sum_equiv φ (fun z : {z : ↥ctr.K // p z} => ψ ((z : ↥ctr.K) : G))
(fun z : ↥ctr.Kprime => ψ (z : G)) (fun z => rfl)
-- The complement-part is constant `= ψ(g)`.
have hconst' : ∀ z ∈ Finset.univ.filter (fun z : ↥ctr.K => ¬ p z),
ψ (z : G) = ψ g := by
intro z hz
rw [Finset.mem_filter] at hz
exact hconst (z : G) g z.2 hz.2 hgK hgK'
set N : ℕ := (Finset.univ.filter (fun z : ↥ctr.K => ¬ p z)).card with hN
have hpart2 : (∑ z ∈ Finset.univ.filter (fun z : ↥ctr.K => ¬ p z), ψ (z : G))
= (N : ℂ) * ψ g := by
rw [Finset.sum_congr rfl hconst', Finset.sum_const, hN, nsmul_eq_mul]
-- Assemble the identity `|K|·a = |K′|·b + N·ψ(g)`.
have hident : (Nat.card ↥ctr.K : ℂ) * (a : ℂ) =
(Nat.card ↥ctr.Kprime : ℂ) * (b : ℂ) + (N : ℂ) * ψ g := by
have h := hsplit
rw [hsumK, hpart1, hsumK', hpart2] at h
exact h
-- `N ≠ 0`: `g` itself is such an element.
have hNpos : 0 < N := by
rw [hN]
refine Finset.card_pos.mpr ⟨⟨g, hgK⟩, ?_⟩
rw [Finset.mem_filter]
exact ⟨Finset.mem_univ _, hgK'⟩
have hNne : (N : ℂ) ≠ 0 := by exact_mod_cast hNpos.ne'
-- Solve for `ψ(g)` in `ℂ`.
have hsolve : ψ g =
((Nat.card ↥ctr.K : ℂ) * (a : ℂ) - (Nat.card ↥ctr.Kprime : ℂ) * (b : ℂ)) / (N : ℂ) := by
rw [eq_div_iff hNne]
linear_combination -hident
-- `ψ(g)` is rational (image of a `ℚ`) and an algebraic integer, hence a rational integer.
obtain ⟨z, hz⟩ := OddOrder.Algebra.exists_int_of_isIntegral_of_mem_range_rat
(isIntegral_apply_of_mem_ZIrr' hψ g)
⟨((Nat.card ↥ctr.K : ℚ) * (a : ℚ) - (Nat.card ↥ctr.Kprime : ℚ) * (b : ℚ)) / (N : ℚ), by
rw [hsolve]; push_cast; ring⟩
exact ⟨z, hz.symm⟩
/-- **Peterfalvi (12.16), the (1.10) congruence core**: the cyclotomic-congruence chain of the
(12.16) contradiction. Given the minimal-counterexample data — a virtual character `ψ ∈ ℤ[Irr G]`,
an order-`p` element `x` and a commuting `g` — together with the facts supplied by the surrounding
§12 machinery (`ψ` constant on the coset, `ψ(xg) = ψ(x)`, by (12.14); `ψ(x) ≡ e (mod 1-ε)`, from the
Dade value relation and (1.10.a) applied to `χ`; and `ψ(g) = mval ∈ ℤ`, by (12.15)), Peterfalvi
(1.10.a) (`exists_integral_apply_sub_of_commute`) and (1.10.b) (`int_dvd_of_one_sub_primRoot_dvd`)
yield `ψ(g) ≡ e (mod p)`, i.e. `p ∣ (mval - e)`.
This isolates the `(1.10)`-using arithmetic of (12.16) (now fully discharged); the remaining
contradiction is the norm/degree inequality (`2e ≤ p+1` of (12.12) together with (12.15)). -/
theorem psi_int_congr_e_mod_p [Finite G] {p : ℕ} (hp : p.Prime) {ε : ℂ}
(hε : IsPrimitiveRoot ε p) {ψ : ClassFunction G ℂ} (hψ : ψ ∈ ZIrr G) {x g : G}
(hx : x ^ p = 1) (hxg : Commute x g) {e mval : ℤ}
(h_const : ψ (x * g) = ψ x)
(h_psix : ∃ w : ℂ, IsIntegral ℤ w ∧ ψ x - (e : ℂ) = (1 - ε) * w)
(h_psig_int : ψ g = (mval : ℂ)) :
(p : ℤ) ∣ (mval - e) := by
obtain ⟨z, hz, hzeq⟩ :=
OddOrder.RepresentationTheory.exists_integral_apply_sub_of_commute hp.pos hε hψ hx hxg
obtain ⟨w, hw, hweq⟩ := h_psix
apply OddOrder.RepresentationTheory.int_dvd_of_one_sub_primRoot_dvd hp hε (hw.sub hz)
have h1 : ψ x - ψ g = (1 - ε) * z := by rw [← h_const]; exact hzeq
rw [h_psig_int] at h1
push_cast
linear_combination hweq - h1
/-- **Peterfalvi (12.16), the magnitude step**: an integer `mval ≡ e (mod p)` with `1 ≤ e` and
`2e ≤ p+1` (the degree bound (12.12)) satisfies `|mval| ≥ e - 1`. Indeed the integers `≡ e (mod p)`
nearest `0` are `e` (distance `e`) and `e - p` (distance `p - e ≥ e - 1`, by `2e ≤ p+1`), so every
such value has `|·| ≥ min(e, p-e) ≥ e - 1`. -/
theorem abs_ge_e_sub_one {p : ℕ} (hppos : 0 < p) {e mval : ℤ} (he : 1 ≤ e)
(h2e : 2 * e ≤ (p : ℤ) + 1) (hdvd : (p : ℤ) ∣ (mval - e)) :
e - 1 ≤ |mval| := by
obtain ⟨k, hk⟩ := hdvd
have hpZ : (0 : ℤ) < (p : ℤ) := by exact_mod_cast hppos
by_cases hk' : 0 ≤ k
· have hpk : 0 ≤ (p : ℤ) * k := mul_nonneg hpZ.le hk'
rw [abs_of_nonneg (by omega)]; omega
· have hk1 : k ≤ -1 := by omega
have hpk : (p : ℤ) * k ≤ -(p : ℤ) := by nlinarith [mul_le_mul_of_nonneg_left hk1 hpZ.le]
rw [abs_of_nonpos (by omega)]; omega
/-- **Peterfalvi (12.16), the value-magnitude conclusion**: chaining the `(1.10)` congruence core
(`psi_int_congr_e_mod_p`) with the degree bound `2e ≤ p+1` of (12.12) gives `|ψ(g)| ≥ e - 1` — the
lower bound on `|ψ(g)|` feeding the final norm inequality of (12.16). -/
theorem abs_psi_g_ge_e_sub_one [Finite G] {p : ℕ} (hp : p.Prime) {ε : ℂ}
(hε : IsPrimitiveRoot ε p) {ψ : ClassFunction G ℂ} (hψ : ψ ∈ ZIrr G) {x g : G}
(hx : x ^ p = 1) (hxg : Commute x g) {e mval : ℤ} (he : 1 ≤ e)
(h2e : 2 * e ≤ (p : ℤ) + 1)
(h_const : ψ (x * g) = ψ x)
(h_psix : ∃ w : ℂ, IsIntegral ℤ w ∧ ψ x - (e : ℂ) = (1 - ε) * w)
(h_psig_int : ψ g = (mval : ℂ)) :
e - 1 ≤ |mval| :=
abs_ge_e_sub_one hp.pos he h2e
(psi_int_congr_e_mod_p hp hε hψ hx hxg h_const h_psix h_psig_int)
/-- **Peterfalvi (12.16), the index/degree contradiction** (the heart of the final inequality): the
reduced inequality `(|K| - |K'|)(e-1)² < e·|K|` together with `4·|K'| ≤ |K|` (i.e. `[K:K'] ≥ 4`,
forced by the fixed-point-free order-`p` action of (8.1.c)) and `e ≥ 3` is contradictory. Indeed
`|K| - |K'| ≥ (3/4)|K|`, so `(3/4)(e-1)² < e`, i.e. `3(e-1)² < 4e`, i.e. `(3e-1)(e-3) < 0` — false
for `e ≥ 3`. This is the `e/(e-1)² ≤ 3/4 < 1 - |K'|/|K|` step of (12.16). -/
theorem index_ratio_contradiction {e kK kKp : ℝ} (he : 3 ≤ e) (hkKp : 0 < kKp)
(hidx : 4 * kKp ≤ kK) (hineq : (kK - kKp) * (e - 1) ^ 2 < e * kK) : False := by
have hkK : 0 < kK := by linarith
nlinarith [hineq, mul_nonneg (show (0:ℝ) ≤ kK / 4 - kKp by linarith) (sq_nonneg (e - 1)),
mul_nonneg hkK.le (mul_nonneg (show (0:ℝ) ≤ e - 3 by linarith)
(show (0:ℝ) ≤ 3 * e - 1 by linarith))]
/-- **Peterfalvi (12.16), the (12.11) reduction**: the final norm inequality
`((|K|-|K'|)/|M|)(e-1)² + 1 - e/|H| < 1` together with `|M| ≤ |K|·|H|` of (12.11) reduces to
`(|K|-|K'|)(e-1)² < e·|K|` (clear `|M|`, `|H|`, then bound `|M|` above). -/
theorem norm_ineq_reduce {e kK kKp kM kH : ℝ} (hkM : 0 < kM) (hkH : 0 < kH)
(he1 : 0 ≤ e) (hM : kM ≤ kK * kH)
(hnorm : ((kK - kKp) / kM) * (e - 1) ^ 2 + 1 - e / kH < 1) :
(kK - kKp) * (e - 1) ^ 2 < e * kK := by
have h1 : (kK - kKp) * (e - 1) ^ 2 / kM < e / kH := by
have e1 : (kK - kKp) * (e - 1) ^ 2 / kM = ((kK - kKp) / kM) * (e - 1) ^ 2 := by ring
rw [e1]; linarith [hnorm]
rw [div_lt_div_iff₀ hkM hkH] at h1
have h2 : e * kM ≤ e * (kK * kH) := mul_le_mul_of_nonneg_left hM he1
have h3 : (kK - kKp) * (e - 1) ^ 2 * kH < e * kK * kH := by nlinarith [h1, h2]
exact lt_of_mul_lt_mul_right h3 hkH.le
/-- **Peterfalvi (12.16), the closing contradiction** (norm-inequality endgame): given the final
norm bound `((|K|-|K'|)/|M|)(e-1)² + 1 - e/|H| < 1` (from (7.3)/(7.8.b)/(8.17) with `|ψ(g)| ≥ e-1`),
the index bound `|M| ≤ |K|·|H|` of (12.11), the degree bound `e ≥ 3`, and the fixed-point-free
`[K:K'] ≥ 4` of (8.1.c), the minimal counterexample is impossible. Combines `norm_ineq_reduce`
with `index_ratio_contradiction`. -/
theorem counterexample_closing {e kK kKp kM kH : ℝ} (he : 3 ≤ e) (hkKp : 0 < kKp)
(hkM : 0 < kM) (hkH : 0 < kH) (hidx : 4 * kKp ≤ kK) (hM : kM ≤ kK * kH)
(hnorm : ((kK - kKp) / kM) * (e - 1) ^ 2 + 1 - e / kH < 1) : False :=
index_ratio_contradiction he hkKp hidx (norm_ineq_reduce hkM hkH (by linarith) hM hnorm)
/-- **Peterfalvi (12.16), the middle (norm-bound) glue**: from `|ψ(g)| ≥ e-1` and the three §7/§8
norm bounds — `A`: `‖ψ^{ρM}‖² ≥ (|K-K'|/|M|)|ψ(g)|²` (from (12.15)), `B`: `‖ψ^ρ‖² ≥ 1 - e/|H|`
((7.8.b)), `C`: `‖ψ^{ρM}‖² + ‖ψ^ρ‖² < 1` ((7.3) with (8.17)) — the norm conclusion
`(|K-K'|/|M|)(e-1)² + 1 - e/|H| < 1` follows (`|ψ(g)|² = mval² ≥ (e-1)²`, then linear). -/
theorem norm_conclusion_glue {e mval : ℤ} {kK kKp kM kH normRhoM normRho : ℝ}
(he : 3 ≤ e) (hkKp : 0 < kKp) (hkM : 0 < kM) (hidx : 4 * kKp ≤ kK)
(hmag : (e : ℝ) - 1 ≤ |(mval : ℝ)|)
(hA : (kK - kKp) / kM * (mval : ℝ) ^ 2 ≤ normRhoM)
(hB : (1 : ℝ) - (e : ℝ) / kH ≤ normRho)
(hC : normRhoM + normRho < 1) :
(kK - kKp) / kM * ((e : ℝ) - 1) ^ 2 + 1 - (e : ℝ) / kH < 1 := by
have hsq : ((e : ℝ) - 1) ^ 2 ≤ (mval : ℝ) ^ 2 := by
have heR : (3 : ℝ) ≤ (e : ℝ) := by exact_mod_cast he
nlinarith [hmag, sq_abs (mval : ℝ), abs_nonneg (mval : ℝ)]
have hpos : 0 ≤ (kK - kKp) / kM := div_nonneg (by linarith) hkM.le
have hA' : (kK - kKp) / kM * ((e : ℝ) - 1) ^ 2 ≤ normRhoM :=
le_trans (mul_le_mul_of_nonneg_left hsq hpos) hA
linarith [hA', hB, hC]
/-- **Peterfalvi (12.16), the full assembly** (the entire argument, parameterized on its gated
upstream): the minimal counterexample is impossible. Combines the `(1.10)` congruence/magnitude
start (`abs_psi_g_ge_e_sub_one`: `|ψ(g)| ≥ e-1`), the §7/§8 norm-bound middle
(`norm_conclusion_glue`
from the three bounds `hA`/`hB`/`hC`), and the index/degree endgame (`counterexample_closing`).
Every hypothesis is a fact supplied by §7/§8/§12: `h_const` = (12.14), `h_psix` = Dade value
relation
with (1.10.a) on `χ`, `h_psig_int` = (12.15), `h2e` = (12.12); `hA` = (12.15)+`|ψ(g)|`, `hB` =
(7.8.b),
`hC` = (7.3)+(8.17); `hM` = (12.11), `hidx` = fpf `[K:K'] ≥ 4` of (8.1.c). The remaining work to
close
`counterexample_contradiction` is exactly the construction of these — the §7 `ρ`/`ρM` machinery. -/
theorem counterexample_contradiction_of_facts [Finite G]
{p : ℕ} (hp : p.Prime) {ε : ℂ} (hε : IsPrimitiveRoot ε p)
{ψ : ClassFunction G ℂ} (hψ : ψ ∈ ZIrr G) {x g : G} (hx : x ^ p = 1) (hxg : Commute x g)
{e mval : ℤ} (he : 3 ≤ e) (h2e : 2 * e ≤ (p : ℤ) + 1)
(h_const : ψ (x * g) = ψ x)
(h_psix : ∃ w : ℂ, IsIntegral ℤ w ∧ ψ x - (e : ℂ) = (1 - ε) * w)
(h_psig_int : ψ g = (mval : ℂ))
{kK kKp kM kH normRhoM normRho : ℝ}
(hkKp : 0 < kKp) (hkM : 0 < kM) (hkH : 0 < kH)
(hidx : 4 * kKp ≤ kK) (hM : kM ≤ kK * kH)
(hA : (kK - kKp) / kM * (mval : ℝ) ^ 2 ≤ normRhoM)
(hB : (1 : ℝ) - (e : ℝ) / kH ≤ normRho)
(hC : normRhoM + normRho < 1) :
False := by
have hmagZ : (e - 1 : ℤ) ≤ |mval| :=
abs_psi_g_ge_e_sub_one hp hε hψ hx hxg (by linarith) h2e h_const h_psix h_psig_int
have hmag : (e : ℝ) - 1 ≤ |(mval : ℝ)| := by rw [← Int.cast_abs]; exact_mod_cast hmagZ
have heR : (3 : ℝ) ≤ (e : ℝ) := by exact_mod_cast he
exact counterexample_closing heR hkKp hkM hkH hidx hM
(norm_conclusion_glue he hkKp hkM hidx hmag hA hB hC)
/-- **Peterfalvi (12.9), the centralizer witness extraction**: the rank-two witness datum records
`¬(C_G(x) ⊓ K ≤ K')`, which directly yields an element `g ∈ C_K(x)` with `g ∉ K'` — the `g`
commuting with `x` used throughout the (12.16) argument. (Pure `SetLike` extraction, ungated.) -/
theorem exists_witness_g {ctr : CounterexampleHypothesis (G := G)}
(witness : RankTwoWitnessData ctr) :
∃ g : G, Commute witness.x g ∧ g ∈ ctr.K ∧ g ∉ ctr.Kprime := by
obtain ⟨g, hgA, hgB⟩ := SetLike.not_le_iff_exists.mp witness.CKx_not_le_Kprime
rw [Subgroup.mem_inf] at hgA
exact ⟨g, Subgroup.mem_centralizer_iff.mp hgA.1 witness.x (Set.mem_singleton _), hgA.2, hgB⟩
/-- **Peterfalvi (7.8.b)/(12.12) size condition for an odd-order Frobenius group**: if a finite
Frobenius group has kernel `N` and complement `A` both of **odd** order, with `N ≠ ⊥`, then
`2|A| + 1 ≤ |N|` (equivalently `e ≤ (h-1)/2`, the `smallIndex` hypothesis of the §7 `(7.8.b)` norm
bound). The complement `A` acts freely on `N#`, so `|A| ∣ |N| - 1` (`card_kernel_modEq_one`,
Isaacs 6.1); as `|N|` is odd, `|N| - 1` is even, and an odd divisor of an even number is at most
half of it, so `|N| - 1 ≥ 2|A|`. This is the `2e_i + 1 ≤ h_i` shape consumed by
`localSmallIndex_of_family_cardinalities` for the witness `L` of (12.16). (General Frobenius fact,
hoistable to `Ch06`.) -/
theorem frobenius_two_mul_card_complement_add_one_le_card_kernel {Γ : Type*} [Group Γ] [Finite Γ]
{N A : Subgroup Γ} (hFrob : OddOrder.Isaacs.Ch06.IsFrobeniusGroup Γ N A)
(hNodd : Odd (Nat.card ↥N)) (hAodd : Odd (Nat.card ↥A)) (hNnt : N ≠ ⊥) :
2 * Nat.card ↥A + 1 ≤ Nat.card ↥N := by
have : Nontrivial ↥N := (Subgroup.nontrivial_iff_ne_bot N).mpr hNnt
have hN1 : 1 < Nat.card ↥N := Finite.one_lt_card
-- `|A| ∣ |N| - 1` from `|N| ≡ 1 [MOD |A|]` (Isaacs 6.1).
obtain ⟨m, hm⟩ : Nat.card ↥A ∣ Nat.card ↥N - 1 :=
(Nat.modEq_iff_dvd' hN1.le).mp hFrob.card_kernel_modEq_one.symm
-- `|N| - 1` is even (`|N|` odd), `|A|` is odd, so the cofactor `m` is even.
have hNm1_even : Even (Nat.card ↥N - 1) := Nat.Odd.sub_odd hNodd odd_one
have hm_even : Even m := by
rcases (Nat.even_mul.mp (hm ▸ hNm1_even)) with hA | hm
· exact absurd hA (Nat.not_even_iff_odd.mpr hAodd)
· exact hm
-- `m ≠ 0` (else `|N| = 1`), so `m ≥ 2`; hence `|N| - 1 = |A|·m ≥ 2|A|`.
have hApos : 0 < Nat.card ↥A := Nat.card_pos
have hm_pos : 0 < m := by
rcases Nat.eq_zero_or_pos m with h0 | hp
· rw [h0, Nat.mul_zero] at hm; omega
· exact hp
have hm2 : 2 ≤ m := Nat.le_of_dvd hm_pos hm_even.two_dvd
have hge : Nat.card ↥A * 2 ≤ Nat.card ↥A * m := Nat.mul_le_mul_left _ hm2
omega
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/DadeContradiction.lean