Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Peterfalvi (12.3)–(12.5): orthogonality and decomposition data (first half)

Definition
OddOrder_Peterfalvi_S14_MaximalI_RhoConstancyDecomposition

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

feit-thompsonfinite-groupsgroup-theoryishida-odd-order

The first half of the ρ\rhoρ-constancy development for type I maximal subgroups: orthogonality and the character decomposition data, with the support properties of S\mathcal{S}S and its τ\tauτ-images.

Main declarations: Sset_vanishes_off_H (Peterfalvi (12.4)/(12.5) input); Sset_diff_support_subset_A1 (The full-member difference is A₁-supported); Sset_diff_tau_support_subset_ftThickenedA1 (The τ-image of a full-member difference is Ã₁-supported); constituentDiff_tau_support_subset_ftThickenedA (The τ-image of a constituent difference is Ã-supported).

The bundle also proves 3 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/RhoConstancyDecomposition.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_Hypothesis

namespace OddOrder.Peterfalvi.S14

open OddOrder.GroupTheory

open OddOrder.RepresentationTheory

open OddOrder.Isaacs

open scoped Pointwise

variable {G : Type*} [Group G]

/-- The difference `φ − φ̄` of a constituent is supported in `A(L)` (each constituent is supported
in
`A(L) ∪ {1}` by `data.supported`, and `(φ − φ̄)(1) = 0` by equal degree — `φ(1)` is real). -/
theorem constituentDiff_support_subset {L : Subgroup G} {hyp : Hypothesis L}
    {chi : ClassFunction ↥L ℂ} (data : CharacterDecompositionData hyp chi)
    {φ : IrreducibleCharacter ↥L} (hφ : φ ∈ data.constituents) :
    ((φ : ClassFunction ↥L ℂ) - (φ : ClassFunction ↥L ℂ).conj).support ⊆
      OddOrder.Peterfalvi.S04.supportInSubgroup hyp.ambientA L := by
  have := hyp.finiteG
  have hsupp_eq : (φ : ClassFunction ↥L ℂ).conj.support = (φ : ClassFunction ↥L ℂ).support := by
    ext y
    simp only [ClassFunction.mem_support, ne_eq, ClassFunction.conj_apply, star_eq_zero]
  intro x hx
  have hx0 : ((φ : ClassFunction ↥L ℂ) - (φ : ClassFunction ↥L ℂ).conj) x ≠ 0 := hx
  have hxsupp : x ∈ (φ : ClassFunction ↥L ℂ).support := by
    have hxU := ClassFunction.support_sub_subset _ _ hx
    rwa [hsupp_eq, Set.union_self] at hxU
  rcases data.supported φ hφ hxsupp with h | h
  · exact h
  · exfalso
    rw [Set.mem_singleton_iff] at h
    subst h
    obtain ⟨d, _, hd⟩ := irreducibleCharacter_apply_one_eq_pos_natCast φ
    exact hx0 (by rw [ClassFunction.sub_apply, ClassFunction.conj_apply, hd, star_natCast,
      sub_self])

/-- Evaluation of a finite sum of class functions at a point (the eval map is additive). -/
theorem classFunction_sum_apply {H : Type*} [Group H] {ι : Type*} (s : Finset ι)
    (F : ι → ClassFunction H ℂ) (g : H) : (∑ i ∈ s, F i) g = ∑ i ∈ s, (F i) g := by
  classical
  induction s using Finset.induction with
  | empty => simp
  | insert i s hi ih =>
      rw [Finset.sum_insert hi, Finset.sum_insert hi, ClassFunction.add_apply, ih]

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.4)/(12.5) input**: each member `χ = Ind_H^L θ` of `S` vanishes on `L − H`.
`H = L_F` is normal in `L` (`maxNilpotentNormalHall_subgroupOf_normal`, the Fitting subgroup `L_F`),
so the induced character is supported on `H` (`ClassFunction.induce_eq_zero_of_not_mem_normal`).
This
is the "the elements of `S` vanish on `L − H`" step of the constant-on-coset conclusions of
(12.4)/(12.5) (`ψ(xh) = β(xh) + γ(xh) = γ(x)`, the `β ∈ ℂ[S]` part vanishing off `H`). -/
theorem Sset_vanishes_off_H {L : Subgroup G} (hyp : Hypothesis L) {χ : ClassFunction ↥L ℂ}
    (hχ : χ ∈ hyp.Sset) {x : ↥L} (hxH : (x : G) ∉ hyp.H) : χ x = 0 := by
  have := hyp.finiteG
  obtain ⟨θ, _, hχ_eq⟩ := hχ
  have hnormal : ((hyp.typeI.typeF.H).subgroupOf L).Normal := by
    rw [hyp.typeI.typeF.H_eq]
    exact OddOrder.BG.Ch4.S15.maxNilpotentNormalHall_subgroupOf_normal L
  have hxmem : x ∉ (hyp.typeI.typeF.H).subgroupOf L :=
    fun hcon => hxH (Subgroup.mem_subgroupOf.mp hcon)
  rw [hχ_eq]
  exact ClassFunction.induce_eq_zero_of_not_mem_normal _ hxmem

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **The full-member difference is `A₁`-supported**: for `χ ∈ S` (`χ = Ind_H^L θ`, `H = L_F`),
`supp(χ − χ̄) ⊆ A₁(L) = (L_F)^#` — `χ` vanishes off the normal `H` (`Sset_vanishes_off_H`) and
`χ(1) = ∑_φ φ(1)` is real, so the difference also vanishes at `1`.  The `Ã₁`-side support input
of the mixed (8.18.c) application in (12.3). -/
theorem Sset_diff_support_subset_A1 {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    {chi : ClassFunction ↥L ℂ} (data : CharacterDecompositionData hyp chi) :
    (chi - chi.conj).support ⊆
      OddOrder.Peterfalvi.S04.supportInSubgroup (A1 L PeterfalviType.I) L := by
  have := hyp.finiteG
  intro x hx
  rw [ClassFunction.mem_support] at hx
  rw [OddOrder.Peterfalvi.S04.mem_supportInSubgroup]
  by_contra hxA1
  apply hx
  rw [ClassFunction.sub_apply, ClassFunction.conj_apply]
  by_cases hxH : (x : G) ∈ hyp.typeI.typeF.H
  · -- inside `H` but not in `A₁ = H^#`: forced `x = 1`, where the difference cancels.
    have hx1 : x = 1 := by
      by_contra hx1
      refine hxA1 ((Set.mem_sdiff _).mpr ⟨SetLike.mem_coe.mpr ?_, fun h => ?_⟩)
      · change (x : G) ∈ maxNilpotentNormalHall L
        rw [← hyp.typeI.typeF.H_eq]
        exact hxH
      · rw [Set.mem_singleton_iff] at h
        exact hx1 (Subtype.ext h)
    subst hx1
    have hreal : star (chi 1) = chi 1 := by
      rw [data.decomp, classFunction_sum_apply, star_sum]
      refine Finset.sum_congr rfl fun φ _ => ?_
      obtain ⟨d, _, hd⟩ := irreducibleCharacter_apply_one_eq_pos_natCast φ
      rw [hd, star_natCast]
    rw [hreal, sub_self]
  · rw [Sset_vanishes_off_H hyp data.chi_mem hxH, star_zero, sub_zero]

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **The `τ`-image of a full-member difference is `Ã₁`-supported** (the (2.11) restriction
computation): `supp((χ−χ̄)^τ) ⊆ Ã₁(L) = ftThickenedSupport L A₁`.  The difference is
`A₁`-supported (`Sset_diff_support_subset_A1`), so `τ` agrees with the `A₁`-restricted Dade map
(`Hypothesis.dadeMap_restrict_apply`), whose image vanishes off the restricted Dade support —
which is exactly the faithful thickened `A₁`-support (the per-point kernels agree,
`ftSupportKernel_restrict`). -/
theorem Sset_diff_tau_support_subset_ftThickenedA1 {L : Subgroup G} [Finite G]
    (hyp : Hypothesis L) {chi : ClassFunction ↥L ℂ}
    (data : CharacterDecompositionData hyp chi) :
    (hyp.tau (chi - chi.conj)).support ⊆
      OddOrder.Peterfalvi.S10.ftThickenedSupport L (A1 L PeterfalviType.I) := by
  have := hyp.finiteG
  have hA₁A : A1 L PeterfalviType.I ⊆ typeIA L hyp.typeI :=
    OddOrder.Peterfalvi.S10.A1_subset_typeIA L hyp.typeI
  have hA₁norm : ∀ (l : ↥L) ⦃a : G⦄, a ∈ A1 L PeterfalviType.I →
      (l : G) * a * (l : G)⁻¹ ∈ A1 L PeterfalviType.I := fun l _ ha =>
    OddOrder.Peterfalvi.S10.A1_conj_mem L OddOrder.GroupTheory.PeterfalviType.I l.2 ha
  have hsuppA1 := Sset_diff_support_subset_A1 hyp data
  have hsuppA : (chi - chi.conj).support ⊆
      OddOrder.Peterfalvi.S04.supportInSubgroup hyp.ambientA L := fun x hx => hA₁A (hsuppA1 hx)
  -- `τ(χ−χ̄)` is the (A₁-restricted) Dade map value on the A₁-supported difference
  have htau_eq : hyp.tau (chi - chi.conj)
      = (hyp.dadeData.dade.restrict hA₁A hA₁norm).dadeMap (k := ℂ)
          ⟨chi - chi.conj, (ClassFunction.mem_supportedSubmodule).mpr hsuppA1⟩ := by
    rw [OddOrder.Peterfalvi.S04.Hypothesis.dadeMap_restrict_apply]
    exact OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_apply_of_support hyp.dadeData.dade
      (hyp.dadeData.dade.fullDadeIsometryData) hsuppA
  -- the restricted Dade support is the faithful thickened `A₁`-support
  have hsub : (hyp.dadeData.dade.restrict hA₁A hA₁norm).dadeSupport ⊆
      OddOrder.Peterfalvi.S10.ftThickenedSupport L (A1 L PeterfalviType.I) := by
    intro g hg
    obtain ⟨a, h, hh, hconj⟩ :=
      (hyp.dadeData.dade.restrict hA₁A hA₁norm).mem_dadeSupport_iff.mp hg
    obtain ⟨c, hc⟩ := isConj_iff.mp hconj
    refine ⟨a.1, a.2, ?_⟩
    rw [OddOrder.GroupTheory.mem_conjClassSet]
    refine ⟨a.1 * h, ⟨h, ?_, rfl⟩, c, hc⟩
    rw [SetLike.mem_coe, OddOrder.Peterfalvi.S10.ftSupportKernel_restrict hA₁A a.2,
      ← hyp.dadeData.H_eq_ftSupportKernel ⟨a.1, hA₁A a.2⟩]
    exact hh
  intro g hg
  rw [ClassFunction.mem_support, htau_eq] at hg
  by_contra hgnot
  refine hg (OddOrder.Peterfalvi.S04.IsDadeMap.map_eq_zero_of_not_mem_dadeSupport
    ((hyp.dadeData.dade.restrict hA₁A hA₁norm).isDadeMap_dadeMap) _ g
    (fun hmem => hgnot (hsub hmem)))

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **The `τ`-image of a constituent difference is `Ã`-supported**:
`supp((φ−φ̄)^τ) ⊆ Ã(L) = ftThickenedSupport L A(L)`.  The difference is `A`-supported
(`constituentDiff_support_subset`), `τ` is the Dade map there, and the (8.15) faithful Dade
support is `Ã(L)` (`dadeSupport_eq_ftThickenedSupport`). -/
theorem constituentDiff_tau_support_subset_ftThickenedA {L : Subgroup G} [Finite G]
    (hyp : Hypothesis L) {chi : ClassFunction ↥L ℂ}
    (data : CharacterDecompositionData hyp chi)
    {φ : IrreducibleCharacter ↥L} (hφ : φ ∈ data.constituents) :
    (hyp.tau ((φ : ClassFunction ↥L ℂ) - (φ : ClassFunction ↥L ℂ).conj)).support ⊆
      OddOrder.Peterfalvi.S10.ftThickenedSupport L (typeIA L hyp.typeI) := by
  have := hyp.finiteG
  have hsupp := constituentDiff_support_subset data hφ
  have htau_eq : hyp.tau ((φ : ClassFunction ↥L ℂ) - (φ : ClassFunction ↥L ℂ).conj)
      = hyp.dadeData.dade.dadeMap (k := ℂ)
          ⟨(φ : ClassFunction ↥L ℂ) - (φ : ClassFunction ↥L ℂ).conj,
            (ClassFunction.mem_supportedSubmodule).mpr hsupp⟩ :=
    OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_apply_of_support
      hyp.dadeData.dade (hyp.dadeData.dade.fullDadeIsometryData) hsupp
  intro g hg
  rw [ClassFunction.mem_support, htau_eq] at hg
  by_contra hgnot
  refine hg (OddOrder.Peterfalvi.S04.IsDadeMap.map_eq_zero_of_not_mem_dadeSupport
    hyp.dadeData.dade.isDadeMap_dadeMap _ g ?_)
  rw [hyp.dadeData.dadeSupport_eq_ftThickenedSupport]
  exact hgnot

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (8.18.c), the (12.3) instantiation**: for non-conjugate type-I maximals
`L₁, L₂`, either `Ã(L₁) ∩ Ã₁(L₂) = ∅` or `Ã(L₂) ∩ Ã₁(L₁) = ∅` — the mixed asymmetric
disjointness, from the S10 `ftThickenedSupport_mixed_disjoint_of_nonconjugate` (issue 0096). -/
theorem nonconjugate_thickened_mixed_disjoint_or_swap {L1 L2 : Subgroup G} [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hyp1 : Hypothesis L1) (hyp2 : Hypothesis L2)
    (hnot_conj : ¬ ∃ g : G, MulAut.conj g • L1 = L2) :
    Disjoint (OddOrder.Peterfalvi.S10.ftThickenedSupport L1 (typeIA L1 hyp1.typeI))
        (OddOrder.Peterfalvi.S10.ftThickenedSupport L2 (A1 L2 PeterfalviType.I)) ∨
      Disjoint (OddOrder.Peterfalvi.S10.ftThickenedSupport L2 (typeIA L2 hyp2.typeI))
        (OddOrder.Peterfalvi.S10.ftThickenedSupport L1 (A1 L1 PeterfalviType.I)) := by
  have := hyp1.finiteG
  have hncTS : ¬ OddOrder.BG.Ch4.S14.IsConjugateSubgroup L2 L1 := fun hc =>
    hnot_conj hc.symm
  rcases OddOrder.Peterfalvi.S10.ftThickenedSupport_mixed_disjoint_of_nonconjugate hG
    hyp2.maximal hyp1.maximal hyp2.typeI hyp1.typeI hncTS with h | h
  · exact Or.inl h.symm
  · exact Or.inr h.symm

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Mixed inner vanishing from the (8.18.c) disjointness**: when `Ã(L₁) ∩ Ã₁(L₂) = ∅`, every
`L₁`-constituent difference image is orthogonal to every `L₂`-full-member difference image.  The
two support computations + disjointly-supported orthogonality. -/
theorem constituent_fullDiff_inner_zero_of_disjoint {L1 L2 : Subgroup G} [Finite G]
    (hyp1 : Hypothesis L1) (hyp2 : Hypothesis L2)
    (hdisj : Disjoint (OddOrder.Peterfalvi.S10.ftThickenedSupport L1 (typeIA L1 hyp1.typeI))
      (OddOrder.Peterfalvi.S10.ftThickenedSupport L2 (A1 L2 PeterfalviType.I)))
    {chi1 : ClassFunction ↥L1 ℂ} (data1 : CharacterDecompositionData hyp1 chi1)
    {φ1 : IrreducibleCharacter ↥L1} (hφ1 : φ1 ∈ data1.constituents)
    {chi2 : ClassFunction ↥L2 ℂ} (data2 : CharacterDecompositionData hyp2 chi2) :
    ClassFunction.inner
        (hyp1.tau ((φ1 : ClassFunction ↥L1 ℂ) - (φ1 : ClassFunction ↥L1 ℂ).conj))
        (hyp2.tau (chi2 - chi2.conj)) = 0 := by
  have := hyp1.finiteG
  exact ClassFunction.inner_eq_zero_of_disjoint_support
    (Disjoint.mono (constituentDiff_tau_support_subset_ftThickenedA hyp1 data1 hφ1)
      (Sset_diff_tau_support_subset_ftThickenedA1 hyp2 data2) hdisj)

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **The Dade `τ` commutes with complex conjugation on `A(L)`-supported functions** — the
coefficientwise form of `S07.dadeIntegralCharacterMap_mapRingEquiv_comm` (the map's value at a
Dade-support point is an evaluation, and `0` elsewhere, so conjugating coefficients commutes).
The τ/conj commutation input of the (12.3) bar-trick ((5.9.b) via
`CharacterDifferenceImage.nu_eq_mu_conj`). -/
theorem tau_conj_of_supported {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    {f : ClassFunction ↥L ℂ}
    (hf : f.support ⊆ OddOrder.Peterfalvi.S04.supportInSubgroup hyp.ambientA L) :
    hyp.tau f.conj = (hyp.tau f).conj := by
  have := hyp.finiteG
  have h1 : f.conj = ClassFunction.mapRingEquiv Complex.conjAe.toRingEquiv f := by
    ext g
    rw [ClassFunction.conj_apply, ClassFunction.mapRingEquiv_apply]
    rfl
  have h2 : (hyp.tau f).conj
      = ClassFunction.mapRingEquiv Complex.conjAe.toRingEquiv (hyp.tau f) := by
    ext g
    rw [ClassFunction.conj_apply, ClassFunction.mapRingEquiv_apply]
    rfl
  rw [h1, h2]
  exact OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_mapRingEquiv_comm
    hyp.dadeData.dade _ _ hf

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **(12.2.b) conjugate pairing of the difference-image block `R₁(φ)`**: `ν_φ = μ̄_φ`, so
`(φ − φ̄)^τ = ε·(μ_φ − μ̄_φ)`.  The (5.9.b) pairing `CharacterDifferenceImage.nu_eq_mu_conj`,
with the τ/conj commutation supplied by `tau_conj_of_supported` on the `A(L)`-supported
difference (`constituentDiff_support_subset`). -/
theorem R1cdi_nu_eq_mu_conj {L : Subgroup G} [Finite G] {hyp : Hypothesis L}
    {chi : ClassFunction ↥L ℂ} (data : CharacterDecompositionData hyp chi)
    {φ : IrreducibleCharacter ↥L} (hφ : φ ∈ data.constituents) :
    (R1cdi data hφ).nuClassFunction = (R1cdi data hφ).muClassFunction.conj := by
  have := hyp.finiteG
  exact OddOrder.Peterfalvi.S07.CharacterDifferenceImage.nu_eq_mu_conj _
    (tau_conj_of_supported hyp (constituentDiff_support_subset data hφ))

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **(12.2.b) cross-constituent image orthogonality**: for distinct constituents `φ ≠ φ'` of
one `χ ∈ S`, the images `(φ−φ̄)^τ` and `(φ'−φ̄')^τ` are orthogonal.  The Dade isometry
(`dadeIntegralCharacterMap_inner_eq_on_supported_span`) reduces to the `L`-side pairing
`⟨φ−φ̄, φ'−φ̄'⟩`, where the four irreducibles are pairwise distinct: `φ ≠ φ'` is given,
conjugation is injective, and `φ̄ ≠ φ'`, `φ ≠ φ̄'` are `data.conj_not_mem` — so all four
`Irr L` deltas vanish. -/
theorem constituentDiff_tau_inner_eq_zero_of_ne {L : Subgroup G} [Finite G]
    {hyp : Hypothesis L} {chi : ClassFunction ↥L ℂ}
    (data : CharacterDecompositionData hyp chi)
    {φ φ' : IrreducibleCharacter ↥L} (hφ : φ ∈ data.constituents)
    (hφ' : φ' ∈ data.constituents) (hne : φ ≠ φ') :
    ClassFunction.inner
        (hyp.tau ((φ : ClassFunction ↥L ℂ) - (φ : ClassFunction ↥L ℂ).conj))
        (hyp.tau ((φ' : ClassFunction ↥L ℂ) - (φ' : ClassFunction ↥L ℂ).conj)) = 0 := by
  have := hyp.finiteG
  classical
  -- The two differences form a supported generating family.
  have hSsupp : ∀ s ∈ ({((φ : ClassFunction ↥L ℂ) - (φ : ClassFunction ↥L ℂ).conj),
      ((φ' : ClassFunction ↥L ℂ) - (φ' : ClassFunction ↥L ℂ).conj)} :
        Set (ClassFunction ↥L ℂ)), s.support ⊆
      OddOrder.Peterfalvi.S04.supportInSubgroup hyp.ambientA L := by
    intro s hs
    rcases hs with rfl | hs
    · exact constituentDiff_support_subset data hφ
    · rw [Set.mem_singleton_iff] at hs
      subst hs
      exact constituentDiff_support_subset data hφ'
  -- The Dade isometry transports the pairing to `L`.
  refine Eq.trans (OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_inner_eq_on_supported_span
    hyp.dadeData.dade hSsupp
    (Submodule.subset_span (Set.mem_insert _ _))
    (Submodule.subset_span (Set.mem_insert_of_mem _ rfl))) ?_
  -- `L`-side: expand into four `Irr L` deltas, all off-diagonal.
  have hcross : ∀ a c : IrreducibleCharacter ↥L,
      ClassFunction.inner (a : ClassFunction ↥L ℂ) (c : ClassFunction ↥L ℂ) =
        if a = c then (1 : ℂ) else 0 :=
    fun a c => OddOrder.RepresentationTheory.irreducibleCharacter_inner a c
  have h2 : φ ≠ OddOrder.Peterfalvi.S07.conjIrreducibleCharacter (L := ↥L) φ' := by
    intro h
    exact data.conj_not_mem φ' hφ' φ hφ
      (congrArg (fun c : IrreducibleCharacter ↥L => (c : ClassFunction ↥L ℂ)) h).symm
  have h3 : OddOrder.Peterfalvi.S07.conjIrreducibleCharacter (L := ↥L) φ ≠ φ' := by
    intro h
    exact data.conj_not_mem φ hφ φ' hφ'
      (congrArg (fun c : IrreducibleCharacter ↥L => (c : ClassFunction ↥L ℂ)) h)
  have h4 : OddOrder.Peterfalvi.S07.conjIrreducibleCharacter (L := ↥L) φ
      ≠ OddOrder.Peterfalvi.S07.conjIrreducibleCharacter (L := ↥L) φ' := by
    intro h
    have hcf := congrArg (fun c : IrreducibleCharacter ↥L => (c : ClassFunction ↥L ℂ)) h
    simp only [OddOrder.Peterfalvi.S07.coe_conjIrreducibleCharacter] at hcf
    exact hne (IrreducibleCharacter.ext
      (by rw [← ClassFunction.conj_conj (φ : ClassFunction ↥L ℂ), hcf,
        ClassFunction.conj_conj]))
  rw [ClassFunction.inner_sub_left, ClassFunction.inner_sub_right,
    ClassFunction.inner_sub_right,
    ← OddOrder.Peterfalvi.S07.coe_conjIrreducibleCharacter (L := ↥L) φ,
    ← OddOrder.Peterfalvi.S07.coe_conjIrreducibleCharacter (L := ↥L) φ',
    hcross φ φ', hcross φ _, hcross _ φ', hcross _ _,
    if_neg hne, if_neg h2, if_neg h3, if_neg h4]
  norm_num

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Two-family variant of the constituent-difference block orthogonality.**  Same `L`, but the two
constituents may come from *different* members `χ₁, χ₂ ∈ S`: given `φ ∈ S(χ₁)`, `φ' ∈ S(χ₂)` with
the
distinctness conditions `φ ≠ φ'`, `φ ≠ φ̄'`, `φ̄ ≠ φ'` (a caller supplies these from
`χ₁ ∉ {χ₂, χ̄₂}`
+ disjoint constituents), the Dade images of the signed differences are orthogonal.

Generalizes `constituentDiff_tau_inner_eq_zero_of_ne` (same `χ`, where the conditions come from
`data.conj_not_mem`) — the cross-family input for the same-`L` `R(χ₁) ⊥ R(χ₂)` orthogonality.  The
proof is identical: the supports of both differences lie in `A(L)`
(`constituentDiff_support_subset`,
one per data), the Dade map is an isometry there, and the `L`-side pairing expands into four
off-diagonal `Irr L` deltas. -/
theorem constituentDiff_tau_inner_eq_zero_of_ne_across {L : Subgroup G} [Finite G]
    {hyp : Hypothesis L} {chi1 chi2 : ClassFunction ↥L ℂ}
    (data1 : CharacterDecompositionData hyp chi1) (data2 : CharacterDecompositionData hyp chi2)
    {φ φ' : IrreducibleCharacter ↥L} (hφ : φ ∈ data1.constituents)
    (hφ' : φ' ∈ data2.constituents) (hne : φ ≠ φ')
    (h2 : φ ≠ OddOrder.Peterfalvi.S07.conjIrreducibleCharacter (L := ↥L) φ')
    (h3 : OddOrder.Peterfalvi.S07.conjIrreducibleCharacter (L := ↥L) φ ≠ φ') :
    ClassFunction.inner
        (hyp.tau ((φ : ClassFunction ↥L ℂ) - (φ : ClassFunction ↥L ℂ).conj))
        (hyp.tau ((φ' : ClassFunction ↥L ℂ) - (φ' : ClassFunction ↥L ℂ).conj)) = 0 := by
  have := hyp.finiteG
  classical
  have hSsupp : ∀ s ∈ ({((φ : ClassFunction ↥L ℂ) - (φ : ClassFunction ↥L ℂ).conj),
      ((φ' : ClassFunction ↥L ℂ) - (φ' : ClassFunction ↥L ℂ).conj)} :
        Set (ClassFunction ↥L ℂ)), s.support ⊆
      OddOrder.Peterfalvi.S04.supportInSubgroup hyp.ambientA L := by
    intro s hs
    rcases hs with rfl | hs
    · exact constituentDiff_support_subset data1 hφ
    · rw [Set.mem_singleton_iff] at hs
      subst hs
      exact constituentDiff_support_subset data2 hφ'
  refine Eq.trans (OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_inner_eq_on_supported_span
    hyp.dadeData.dade hSsupp
    (Submodule.subset_span (Set.mem_insert _ _))
    (Submodule.subset_span (Set.mem_insert_of_mem _ rfl))) ?_
  have hcross : ∀ a c : IrreducibleCharacter ↥L,
      ClassFunction.inner (a : ClassFunction ↥L ℂ) (c : ClassFunction ↥L ℂ) =
        if a = c then (1 : ℂ) else 0 :=
    fun a c => OddOrder.RepresentationTheory.irreducibleCharacter_inner a c
  have h4 : OddOrder.Peterfalvi.S07.conjIrreducibleCharacter (L := ↥L) φ
      ≠ OddOrder.Peterfalvi.S07.conjIrreducibleCharacter (L := ↥L) φ' := by
    intro h
    have hcf := congrArg (fun c : IrreducibleCharacter ↥L => (c : ClassFunction ↥L ℂ)) h
    simp only [OddOrder.Peterfalvi.S07.coe_conjIrreducibleCharacter] at hcf
    exact hne (IrreducibleCharacter.ext
      (by rw [← ClassFunction.conj_conj (φ : ClassFunction ↥L ℂ), hcf,
        ClassFunction.conj_conj]))
  rw [ClassFunction.inner_sub_left, ClassFunction.inner_sub_right,
    ClassFunction.inner_sub_right,
    ← OddOrder.Peterfalvi.S07.coe_conjIrreducibleCharacter (L := ↥L) φ,
    ← OddOrder.Peterfalvi.S07.coe_conjIrreducibleCharacter (L := ↥L) φ',
    hcross φ φ', hcross φ _, hcross _ φ', hcross _ _,
    if_neg hne, if_neg h2, if_neg h3, if_neg h4]
  norm_num

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
open OddOrder.Peterfalvi.S07.CharacterDifferenceImage in
/-- **The (12.3) bar-trick descent, one-sided core.**  If `Ã(L₁) ∩ Ã₁(L₂) = ∅` ((8.18.c)),
each `L₁`-constituent image is orthogonal to each `L₂`-constituent image.

Peterfalvi's (12.3) argument: with `X = (χ₂−χ̄₂)^{τ₂}`, the support disjointness gives
`⟨(φ₁−φ̄₁)^{τ₁}, X⟩ = 0` (`constituent_fullDiff_inner_zero_of_disjoint`); writing
`(φ₁−φ̄₁)^{τ₁} = ε·(μ₁ − μ̄₁)` ((5.9.b), `R1cdi_nu_eq_mu_conj`) and using `X̄ = −X`
(`tau_conj_of_supported`) plus the integrality of the Fourier coefficients of `X ∈ ℤ[Irr G]`
(`mem_ZIrr_inner_int`) yields `⟨μ₁, X⟩ = ⟨μ̄₁, X⟩ = −conj ⟨μ₁, X⟩ = −⟨μ₁, X⟩`, so
`⟨μ₁, X⟩ = 0`.  Expanding `X = ∑_{φ∈S(χ₂)} (φ−φ̄)^{τ₂}`, the cross-`φ` block orthogonality
(`constituentDiff_tau_inner_eq_zero_of_ne` + the (4.1) member lemma) lets `μ₁` pair
nontrivially with at most one block, so every single summand `⟨μ₁, (φ₂−φ̄₂)^{τ₂}⟩` vanishes,
not just the sum — and likewise for `ν₁ = μ̄₁`. -/
theorem constituent_diffImage_inner_zero_of_disjoint {L1 L2 : Subgroup G} [Finite G]
    (hyp1 : Hypothesis L1) (hyp2 : Hypothesis L2)
    (hdisj : Disjoint (OddOrder.Peterfalvi.S10.ftThickenedSupport L1 (typeIA L1 hyp1.typeI))
      (OddOrder.Peterfalvi.S10.ftThickenedSupport L2 (A1 L2 PeterfalviType.I)))
    {chi1 : ClassFunction ↥L1 ℂ} (data1 : CharacterDecompositionData hyp1 chi1)
    {φ1 : IrreducibleCharacter ↥L1} (hφ1 : φ1 ∈ data1.constituents)
    {chi2 : ClassFunction ↥L2 ℂ} (data2 : CharacterDecompositionData hyp2 chi2)
    {φ2 : IrreducibleCharacter ↥L2} (hφ2 : φ2 ∈ data2.constituents) :
    ClassFunction.inner
        (hyp1.tau ((φ1 : ClassFunction ↥L1 ℂ) - (φ1 : ClassFunction ↥L1 ℂ).conj))
        (hyp2.tau ((φ2 : ClassFunction ↥L2 ℂ) - (φ2 : ClassFunction ↥L2 ℂ).conj)) = 0 := by
  have := hyp1.finiteG
  classical
  set X : ClassFunction G ℂ := hyp2.tau (chi2 - chi2.conj) with hX
  -- (a) `χ₂ − χ̄₂ = ∑_{φ∈S(χ₂)} (φ − φ̄)`, hence `X = ∑ (φ−φ̄)^{τ₂}` and `X ∈ ℤ[Irr G]`.
  have hdiff_eq : chi2 - chi2.conj = ∑ φ ∈ data2.constituents,
      ((φ : ClassFunction ↥L2 ℂ) - (φ : ClassFunction ↥L2 ℂ).conj) := by
    have hconj : (∑ φ ∈ data2.constituents, (φ : ClassFunction ↥L2 ℂ)).conj
        = ∑ φ ∈ data2.constituents, (φ : ClassFunction ↥L2 ℂ).conj := by
      ext g
      rw [ClassFunction.conj_apply, classFunction_sum_apply, classFunction_sum_apply, star_sum]
      exact Finset.sum_congr rfl fun φ _ => by rw [ClassFunction.conj_apply]
    conv_lhs => rw [data2.decomp]
    rw [hconj, ← Finset.sum_sub_distrib]
  have hXsum : X = ∑ φ ∈ data2.constituents,
      hyp2.tau ((φ : ClassFunction ↥L2 ℂ) - (φ : ClassFunction ↥L2 ℂ).conj) := by
    rw [hX, hdiff_eq, map_sum]
  have hfsupp : (chi2 - chi2.conj).support ⊆
      OddOrder.Peterfalvi.S04.supportInSubgroup hyp2.ambientA L2 := by
    rw [hdiff_eq]
    intro x hx
    rw [ClassFunction.mem_support, classFunction_sum_apply] at hx
    obtain ⟨φ, hφm, hne0⟩ := Finset.exists_ne_zero_of_sum_ne_zero hx
    exact constituentDiff_support_subset data2 hφm (ClassFunction.mem_support.mpr hne0)
  have hXZIrr : X ∈ OddOrder.RepresentationTheory.ZIrr G := by
    rw [hX]
    refine OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_mem_ZIrr_of_supported
      hyp2.dadeData.dade hfsupp ?_
    rw [hdiff_eq]
    exact Submodule.sum_mem _ fun φ _ => Submodule.sub_mem _ φ.mem_ZIrr
      (OddOrder.Peterfalvi.S07.conjIrreducibleCharacter (L := ↥L2) φ).mem_ZIrr
  -- (b) `⟨a, X⟩` is real for every irreducible `a` (integer Fourier coefficient).
  have hreal : ∀ a : IrreducibleCharacter G,
      star (ClassFunction.inner (a : ClassFunction G ℂ) X)
        = ClassFunction.inner (a : ClassFunction G ℂ) X := by
    intro a
    obtain ⟨m, hm⟩ := OddOrder.RepresentationTheory.mem_ZIrr_inner_int a hXZIrr
    rw [OddOrder.RepresentationTheory.inner_conj_symm X (a : ClassFunction G ℂ), hm,
      star_star, star_intCast]
  -- (c) `X̄ = −X` (τ₂ commutes with conjugation on the supported difference).
  have hXconj : X.conj = -X := by
    rw [hX, ← tau_conj_of_supported hyp2 hfsupp, ClassFunction.conj_sub,
      ClassFunction.conj_conj, ← neg_sub, map_neg]
  -- (d) the bar-trick: `⟨μ₁, X⟩ = ⟨ν₁, X⟩ = 0`.
  have hfull : ClassFunction.inner
      (hyp1.tau ((φ1 : ClassFunction ↥L1 ℂ) - (φ1 : ClassFunction ↥L1 ℂ).conj)) X = 0 :=
    constituent_fullDiff_inner_zero_of_disjoint hyp1 hyp2 hdisj data1 hφ1 data2
  rw [(R1cdi data1 hφ1).image_eq_signedDifference] at hfull
  simp only [OddOrder.Peterfalvi.S07.CharacterDifferenceImage.signedDifference,
    OddOrder.Peterfalvi.S07.CharacterDifferenceImage.difference] at hfull
  rw [← Int.cast_smul_eq_zsmul ℂ, ClassFunction.inner_smul_left,
    ClassFunction.inner_sub_left] at hfull
  have hs1 : (((R1cdi data1 hφ1).sign : ℤ) : ℂ) ≠ 0 :=
    Int.cast_ne_zero.mpr (R1cdi data1 hφ1).sign_ne_zero
  have heq : ClassFunction.inner ((R1cdi data1 hφ1).muClassFunction) X
      = ClassFunction.inner ((R1cdi data1 hφ1).nuClassFunction) X :=
    sub_eq_zero.mp ((mul_eq_zero.mp hfull).resolve_left hs1)
  have hnux : ClassFunction.inner ((R1cdi data1 hφ1).nuClassFunction) X
      = -star (ClassFunction.inner ((R1cdi data1 hφ1).muClassFunction) X) := by
    rw [R1cdi_nu_eq_mu_conj data1 hφ1]
    conv_lhs => rw [← ClassFunction.conj_conj X]
    rw [OddOrder.RepresentationTheory.inner_conj_conj, hXconj,
      ClassFunction.inner_neg_right, star_neg]
  have hmux : ClassFunction.inner ((R1cdi data1 hφ1).muClassFunction) X = 0 := by
    have h1 : ClassFunction.inner ((R1cdi data1 hφ1).muClassFunction) X
        = -(ClassFunction.inner ((R1cdi data1 hφ1).muClassFunction) X) := by
      conv_lhs => rw [heq, hnux, hreal (R1cdi data1 hφ1).mu]
    linear_combination h1 / 2
  have hnux0 : ClassFunction.inner ((R1cdi data1 hφ1).nuClassFunction) X = 0 :=
    heq.symm.trans hmux
  -- (e) descent: an irreducible orthogonal to `X` is orthogonal to the `φ₂`-summand.
  have hcrossG : ∀ a c : IrreducibleCharacter G,
      ClassFunction.inner (a : ClassFunction G ℂ) (c : ClassFunction G ℂ) =
        if a = c then (1 : ℂ) else 0 :=
    fun a c => OddOrder.RepresentationTheory.irreducibleCharacter_inner a c
  have hdescend : ∀ a : IrreducibleCharacter G,
      ClassFunction.inner (a : ClassFunction G ℂ) X = 0 →
      ClassFunction.inner (a : ClassFunction G ℂ)
        (hyp2.tau ((φ2 : ClassFunction ↥L2 ℂ) - (φ2 : ClassFunction ↥L2 ℂ).conj)) = 0 := by
    intro a haX
    by_contra ht2
    -- a nonzero pairing with the block of `φ'` puts `a` in its two-element image set.
    have hmem : ∀ φ' (hm : φ' ∈ data2.constituents),
        ClassFunction.inner (a : ClassFunction G ℂ)
          (hyp2.tau ((φ' : ClassFunction ↥L2 ℂ) - (φ' : ClassFunction ↥L2 ℂ).conj)) ≠ 0 →
        (a : ClassFunction G ℂ) ∈ (R1cdi data2 hm).imageSet := by
      intro φ' hm hne0
      rw [(R1cdi data2 hm).image_eq_signedDifference] at hne0
      simp only [OddOrder.Peterfalvi.S07.CharacterDifferenceImage.signedDifference,
        OddOrder.Peterfalvi.S07.CharacterDifferenceImage.difference] at hne0
      rw [← Int.cast_smul_eq_zsmul ℂ, OddOrder.RepresentationTheory.inner_smul_right,
        ClassFunction.inner_sub_right] at hne0
      have hsub : ClassFunction.inner (a : ClassFunction G ℂ)
            ((R1cdi data2 hm).muClassFunction)
          - ClassFunction.inner (a : ClassFunction G ℂ)
            ((R1cdi data2 hm).nuClassFunction) ≠ 0 :=
        fun h => hne0 (by rw [h, mul_zero])
      rw [OddOrder.Peterfalvi.S07.CharacterDifferenceImage.mem_imageSet_iff]
      by_cases hc1 : a = (R1cdi data2 hm).mu
      · exact Or.inl (congrArg (fun c : IrreducibleCharacter G =>
          (c : ClassFunction G ℂ)) hc1)
      by_cases hc2 : a = (R1cdi data2 hm).nu
      · exact Or.inr (congrArg (fun c : IrreducibleCharacter G =>
          (c : ClassFunction G ℂ)) hc2)
      exfalso
      exact hsub (by rw [hcrossG a _, hcrossG a _, if_neg hc1, if_neg hc2, sub_zero])
    have hmem2 := hmem φ2 hφ2 ht2
    have hothers : ∀ φ' ∈ data2.constituents, φ' ≠ φ2 →
        ClassFunction.inner (a : ClassFunction G ℂ)
          (hyp2.tau ((φ' : ClassFunction ↥L2 ℂ) - (φ' : ClassFunction ↥L2 ℂ).conj)) = 0 := by
      intro φ' hm' hne'
      by_contra htne'
      have hmem' := hmem φ' hm' htne'
      have hsd : ClassFunction.inner ((R1cdi data2 hm').signedDifference)
          ((R1cdi data2 hφ2).signedDifference) = 0 := by
        rw [← (R1cdi data2 hm').image_eq_signedDifference,
          ← (R1cdi data2 hφ2).image_eq_signedDifference]
        exact constituentDiff_tau_inner_eq_zero_of_ne data2 hm' hφ2 hne'
      have hcontra := inner_eq_zero_of_signedDifference_inner_zero_of_mem
        (R1cdi data2 hm') (R1cdi data2 hφ2) hsd hmem' hmem2
      rw [hcrossG a a, if_pos rfl] at hcontra
      exact one_ne_zero hcontra
    have hsum0 : ∑ φ' ∈ data2.constituents,
        ClassFunction.inner (a : ClassFunction G ℂ)
          (hyp2.tau ((φ' : ClassFunction ↥L2 ℂ) - (φ' : ClassFunction ↥L2 ℂ).conj)) = 0 := by
      rw [← OddOrder.RepresentationTheory.inner_sum_right, ← hXsum]
      exact haX
    rw [Finset.sum_eq_single_of_mem φ2 hφ2 hothers] at hsum0
    exact ht2 hsum0
  -- (f) conclude: expand `(φ₁−φ̄₁)^{τ₁} = ε·(μ₁ − ν₁)` and apply the two descents.
  rw [(R1cdi data1 hφ1).image_eq_signedDifference]
  simp only [OddOrder.Peterfalvi.S07.CharacterDifferenceImage.signedDifference,
    OddOrder.Peterfalvi.S07.CharacterDifferenceImage.difference]
  rw [← Int.cast_smul_eq_zsmul ℂ, ClassFunction.inner_smul_left,
    ClassFunction.inner_sub_left, hdescend (R1cdi data1 hφ1).mu hmux,
    hdescend (R1cdi data1 hφ1).nu hnux0, sub_zero, mul_zero]

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (8.18.c) difference-image orthogonality** (the geometric obligation of (12.3)).
For non-conjugate type-I maximals `L1, L2` and constituents `φ_i`, the Dade difference images
`τ_i(φ_i − φ̄_i)` are orthogonal.

**Genuine proof** (2026-07-03, loop¹⁰⁰): the mixed asymmetric `Ã(L₁) ∩ Ã₁(L₂) = ∅ ∨ swap` is
`nonconjugate_thickened_mixed_disjoint_or_swap` (S10 (8.18.c), three §16 pins); on the
disjoint side the bar-trick descent `constituent_diffImage_inner_zero_of_disjoint` closes the
per-constituent orthogonality, and the swap side follows by conjugate symmetry of the inner
product.  Hub issue 9003 Cluster B. -/
theorem nonconjugate_diffImage_inner_zero {L1 L2 : Subgroup G} [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hyp1 : Hypothesis L1) (hyp2 : Hypothesis L2)
    (hnot_conj : ¬ ∃ g : G, MulAut.conj g • L1 = L2)
    {chi1 : ClassFunction ↥L1 ℂ} (data1 : CharacterDecompositionData hyp1 chi1)
    {φ1 : IrreducibleCharacter ↥L1} (hφ1 : φ1 ∈ data1.constituents)
    {chi2 : ClassFunction ↥L2 ℂ} (data2 : CharacterDecompositionData hyp2 chi2)
    {φ2 : IrreducibleCharacter ↥L2} (hφ2 : φ2 ∈ data2.constituents) :
    ClassFunction.inner
        (hyp1.tau ((φ1 : ClassFunction ↥L1 ℂ) - (φ1 : ClassFunction ↥L1 ℂ).conj))
        (hyp2.tau ((φ2 : ClassFunction ↥L2 ℂ) - (φ2 : ClassFunction ↥L2 ℂ).conj)) = 0 := by
  have := hyp1.finiteG
  rcases nonconjugate_thickened_mixed_disjoint_or_swap hG hyp1 hyp2 hnot_conj with h | h
  · exact constituent_diffImage_inner_zero_of_disjoint hyp1 hyp2 h data1 hφ1 data2 hφ2
  · have hswap := constituent_diffImage_inner_zero_of_disjoint hyp2 hyp1 h data2 hφ2 data1 hφ1
    rw [OddOrder.RepresentationTheory.inner_conj_symm
        (hyp2.tau ((φ2 : ClassFunction ↥L2 ℂ) - (φ2 : ClassFunction ↥L2 ℂ).conj))
        (hyp1.tau ((φ1 : ClassFunction ↥L1 ℂ) - (φ1 : ClassFunction ↥L1 ℂ).conj)),
      hswap, star_zero]

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.3)**: for non-conjugate type-I maximal subgroups `L₁, L₂`, the families
`R(χ₁) = Rset data1` and `R(χ₂) = Rset data2` are mutually orthogonal.

Proof: a member `α ∈ R(χ₁)` lies in `R₁(φ₁) = (R1cdi data1 hφ₁).toOrthonormalImage` for some
constituent `φ₁`, and likewise `β ∈ R₁(φ₂)`.  The cross-`L` (4.1) orthogonality
`toOrthonormalImage_inner_eq_zero_across` reduces `⟨α, β⟩ = 0` to the orthogonality of the signed
differences `⟨(φ₁−φ̄₁)^{τ₁}, (φ₂−φ̄₂)^{τ₂}⟩ = 0` (`image_eq_signedDifference`), which is the
geometric
obligation `nonconjugate_diffImage_inner_zero` ((8.18.c): the supports lie in disjoint `Ã(L₁)`,
`Ã₁(L₂)`). -/
theorem nonconjugate_typeI_R_orthogonal {L1 L2 : Subgroup G} [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hyp1 : Hypothesis L1) (hyp2 : Hypothesis L2)
    (hnot_conj : ¬ ∃ g : G, MulAut.conj g • L1 = L2)
    {chi1 : ClassFunction ↥L1 ℂ} (data1 : CharacterDecompositionData hyp1 chi1)
    {chi2 : ClassFunction ↥L2 ℂ} (data2 : CharacterDecompositionData hyp2 chi2) :
    ∀ α ∈ Rset data1, ∀ β ∈ Rset data2, ClassFunction.inner α β = 0 := by
  intro α hαm β hβm
  obtain ⟨φ1, hφ1, hα⟩ := hαm
  obtain ⟨φ2, hφ2, hβ⟩ := hβm
  refine OddOrder.Peterfalvi.S07.CharacterDifferenceImage.toOrthonormalImage_inner_eq_zero_across
    (R1cdi data1 hφ1) (R1cdi data2 hφ2) ?_ hα hβ
  rw [← OddOrder.Peterfalvi.S07.CharacterDifferenceImage.image_eq_signedDifference
        (R1cdi data1 hφ1),
    ← OddOrder.Peterfalvi.S07.CharacterDifferenceImage.image_eq_signedDifference (R1cdi data2 hφ2)]
  exact nonconjugate_diffImage_inner_zero hG hyp1 hyp2 hnot_conj data1 hφ1 data2 hφ2

/-- **Difference-uniqueness for signed irreducible-character differences** (Peterfalvi §3, the
reconciliation core of (1.4)).  If two *signed* differences of distinct irreducible characters
coincide, `s • (a − b) = t • (c − d)` with `a ≠ b`, `c ≠ d` and a nonzero left scalar `s`, then the
unordered pairs agree, with the sign tracking the orientation: either `a = c, b = d, s = t`
(same orientation) or `a = d, b = c, s = −t` (reversed).

This is the lemma by which the per-constituent families `R₁(φ)` (built from `τ(φ̄ − φ)`) are
reconciled with the global signed family of (1.4) in pin (a) `constituent_diff_tau_mem_span`:
the two presentations of `τ(φ − φ̄)` as a signed difference must share their underlying irreducible
pair, so each `μ_φ` lands in `ℤ[R(χ)]`.

Proof: pairing the hypothesis on the *left* with the irreducible `a` (resp. `b`) and using
orthonormality `⟨χ, ψ⟩ = δ_{χ,ψ}` (`irreducibleCharacter_inner_eq_ite`; `ClassFunction.inner` is
ℂ-linear in the left slot) gives `s = t·([c=a] − [d=a])` and `−s = t·([c=b] − [d=b])`.  Since
`s ≠ 0`, exactly one of `c = a`, `d = a` holds (the two cases are the two orientations), and the
`b`-pairing pins the remaining equality. -/
theorem irreducibleCharacter_signed_difference_uniqueness [Finite G]
    {a b c d : IrreducibleCharacter G} (hab : a ≠ b) (hcd : c ≠ d)
    {s t : ℂ} (hs : s ≠ 0)
    (h : s • ((a : ClassFunction G ℂ) - (b : ClassFunction G ℂ))
        = t • ((c : ClassFunction G ℂ) - (d : ClassFunction G ℂ))) :
    (a = c ∧ b = d ∧ s = t) ∨ (a = d ∧ b = c ∧ s = -t) := by
  classical
  have : Fintype G := Fintype.ofFinite G
  have : Invertible (Nat.card G : ℂ) :=
    invertibleOfNonzero (Nat.cast_ne_zero.mpr Nat.card_pos.ne')
  have hba : b ≠ a := fun he => hab he.symm
  -- Pair the equation on the left with an arbitrary irreducible `e`; orthonormality turns each
  -- coercion-inner into a Kronecker delta.
  have key : ∀ e : IrreducibleCharacter G,
      s * ((if a = e then (1 : ℂ) else 0) - (if b = e then 1 else 0))
        = t * ((if c = e then (1 : ℂ) else 0) - (if d = e then 1 else 0)) := by
    intro e
    have h2 := congrArg (fun f => ClassFunction.inner f (e : ClassFunction G ℂ)) h
    simpa only [ClassFunction.inner_smul_left, ClassFunction.inner_sub_left,
      irreducibleCharacter_inner_eq_ite] using h2
  -- Evaluate at `a`: `s = t·([c=a] − [d=a])`.
  have ka : s = t * ((if c = a then (1 : ℂ) else 0) - (if d = a then 1 else 0)) := by
    have hka := key a
    rwa [if_pos rfl, if_neg hba, sub_zero, mul_one] at hka
  -- Evaluate at `b`: `−s = t·([c=b] − [d=b])`.
  have kb : -s = t * ((if c = b then (1 : ℂ) else 0) - (if d = b then 1 else 0)) := by
    have hkb := key b
    rwa [if_neg hab, if_pos rfl, zero_sub, mul_neg_one] at hkb
  by_cases hca : c = a
  · -- Orientation A: `c = a`, hence `d ≠ a`, and `ka` collapses to `s = t`.
    have hda : d ≠ a := fun he => hcd (hca.trans he.symm)
    rw [if_pos hca, if_neg hda, sub_zero, mul_one] at ka
    have hcb : c ≠ b := fun he => hab (hca.symm.trans he)
    rw [if_neg hcb, zero_sub, mul_neg] at kb
    by_cases hdb : d = b
    · exact Or.inl ⟨hca.symm, hdb.symm, ka⟩
    · rw [if_neg hdb, mul_zero, neg_zero] at kb
      exact absurd (neg_eq_zero.mp kb) hs
  · by_cases hda : d = a
    · -- Orientation B: `c ≠ a`, `d = a`, and `ka` collapses to `s = −t`.
      rw [if_neg hca, if_pos hda, zero_sub, mul_neg_one] at ka
      have hdb : d ≠ b := fun he => hab (hda.symm.trans he)
      rw [if_neg hdb, sub_zero] at kb
      by_cases hcb : c = b
      · exact Or.inr ⟨hda.symm, hcb.symm, ka⟩
      · rw [if_neg hcb, mul_zero] at kb
        exact absurd (neg_eq_zero.mp kb) hs
    · -- Neither: `ka` forces `s = 0`, contradiction.
      rw [if_neg hca, if_neg hda, sub_zero, mul_zero] at ka
      exact absurd ka hs

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.4) pin (a), piece 3** (span membership): the two underlying irreducibles
`μ_φ, ν_φ` of the difference image `R1cdi data hφ` lie in `ℤ[R(χ)]`.  The orthonormal block
`R₁(φ).imageSet = {ε·μ_φ, −ε·ν_φ} ⊆ R(χ)` with `ε = ±1`, so `μ_φ = ε·(ε·μ_φ)` and
`ν_φ = (−ε)·(−ε·ν_φ)` are integer multiples of `R(χ)` members. -/
theorem R1cdi_muNu_mem_span_Rset {L : Subgroup G} [Finite G] {hyp : Hypothesis L}
    {chi : ClassFunction ↥L ℂ} (data : CharacterDecompositionData hyp chi)
    {φ : IrreducibleCharacter ↥L} (hφ : φ ∈ data.constituents) :
    (R1cdi data hφ).muClassFunction ∈ Submodule.span ℤ (Rset data) ∧
      (R1cdi data hφ).nuClassFunction ∈ Submodule.span ℤ (Rset data) := by
  have := hyp.finiteG
  classical
  have himg : (R1 data hφ).imageSet
      = ({(R1cdi data hφ).sign • (R1cdi data hφ).muClassFunction,
          (-(R1cdi data hφ).sign) • (R1cdi data hφ).nuClassFunction} :
            Finset (ClassFunction G ℂ)) := rfl
  have hsq : (R1cdi data hφ).sign * (R1cdi data hφ).sign = 1 := (R1cdi data hφ).sign_mul_self
  have hμRset : (R1cdi data hφ).sign • (R1cdi data hφ).muClassFunction ∈ Rset data :=
    ⟨φ, hφ, by rw [himg]; exact Finset.mem_insert_self _ _⟩
  have hνRset : (-(R1cdi data hφ).sign) • (R1cdi data hφ).nuClassFunction ∈ Rset data :=
    ⟨φ, hφ, by rw [himg]; exact Finset.mem_insert_of_mem (Finset.mem_singleton_self _)⟩
  refine ⟨?_, ?_⟩
  · have hμ : (R1cdi data hφ).muClassFunction
        = (R1cdi data hφ).sign • ((R1cdi data hφ).sign • (R1cdi data hφ).muClassFunction) := by
      rw [smul_smul, hsq, one_smul]
    rw [hμ]
    exact Submodule.smul_mem _ _ (Submodule.subset_span hμRset)
  · have hν : (R1cdi data hφ).nuClassFunction
        = (-(R1cdi data hφ).sign) •
          ((-(R1cdi data hφ).sign) • (R1cdi data hφ).nuClassFunction) := by
      rw [smul_smul, neg_mul_neg, hsq, one_smul]
    rw [hν]
    exact Submodule.smul_mem _ _ (Submodule.subset_span hνRset)

open scoped Classical OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.4) pin (a), piece 1** (global (1.4) coherence): the constituent set `S(χ)`
together with its complex conjugates forms a single coherent family under the Dade isometry `τ`.
There is a uniform sign `ε = ±1` and an injection `μ` of the conjugate-closed set
`T = S(χ) ∪ S(χ)‾` into `Irr G` with `τ(α − β) = ε·(μ α − μ β)` for all `α, β ∈ T`.

This is the §3 (1.4) keystone `isometry_difference_pair_structure` applied to the constant-degree
family `T` (every member supported in `A(L) ∪ {1}`, so member differences are `A(L)`-supported and
the three Dade-isometry hypotheses hold by `dadeIntegralCharacterMap_{mem_ZIrr_of_supported,
apply_one_eq_zero,inner_eq_on_supported_span}`). -/
theorem exists_uniform_image_of_constituents {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    {chi : ClassFunction ↥L ℂ} (data : CharacterDecompositionData hyp chi) :
    ∃ (ε : ℤ) (μ : IrreducibleCharacter ↥L → IrreducibleCharacter G),
      (ε = 1 ∨ ε = -1) ∧
      Set.InjOn μ ↑(data.constituents ∪
          data.constituents.image (IrreducibleCharacter.conjPerm ↥L)) ∧
      ∀ α ∈ data.constituents ∪ data.constituents.image (IrreducibleCharacter.conjPerm ↥L),
        ∀ β ∈ data.constituents ∪ data.constituents.image (IrreducibleCharacter.conjPerm ↥L),
          hyp.tau ((α : ClassFunction ↥L ℂ) - (β : ClassFunction ↥L ℂ))
            = ε • ((μ α : ClassFunction G ℂ) - (μ β : ClassFunction G ℂ)) := by
  have := hyp.finiteG
  classical
  set T := data.constituents ∪ data.constituents.image (IrreducibleCharacter.conjPerm ↥L) with hTdef
  obtain ⟨φref, hφref⟩ := data.constituents_nonempty
  -- (1) every member of `T` is supported in `A(L) ∪ {1}`.
  have hTsupp : ∀ x ∈ T, (x : ClassFunction ↥L ℂ).support ⊆
      OddOrder.Peterfalvi.S04.supportInSubgroup hyp.ambientA L ∪ {1} := by
    intro x hx
    rw [hTdef, Finset.mem_union] at hx
    rcases hx with hx | hx
    · exact data.supported x hx
    · rw [Finset.mem_image] at hx
      obtain ⟨φ, hφ, rfl⟩ := hx
      rw [IrreducibleCharacter.conjPerm_apply_coe]
      have hconjsupp : (φ : ClassFunction ↥L ℂ).conj.support
          = (φ : ClassFunction ↥L ℂ).support := by
        ext y; simp only [ClassFunction.mem_support, ne_eq, ClassFunction.conj_apply, star_eq_zero]
      rw [hconjsupp]; exact data.supported φ hφ
  -- (2) every member of `T` has the reference degree `φref(1)`.
  have hTdeg : ∀ x ∈ T, ((x : ClassFunction ↥L ℂ) : ↥L → ℂ) 1
      = ((φref : ClassFunction ↥L ℂ) : ↥L → ℂ) 1 := by
    intro x hx
    rw [hTdef, Finset.mem_union] at hx
    rcases hx with hx | hx
    · exact data.equal_degree x hx φref hφref
    · rw [Finset.mem_image] at hx
      obtain ⟨φ, hφ, rfl⟩ := hx
      rw [IrreducibleCharacter.conjPerm_apply_coe]
      obtain ⟨d, _, hd⟩ := irreducibleCharacter_apply_one_eq_pos_natCast φ
      rw [ClassFunction.conj_apply, hd, star_natCast, ← hd]
      exact data.equal_degree φ hφ φref hφref
  -- (3) enumerate `T` as a `Fin n` family.
  set n := T.card with hndef
  have hφrefT : φref ∈ T := Finset.mem_union_left _ hφref
  have hconjrefT : IrreducibleCharacter.conjPerm ↥L φref ∈ T :=
    Finset.mem_union_right _ (Finset.mem_image_of_mem _ hφref)
  have hrefne : φref ≠ IrreducibleCharacter.conjPerm ↥L φref := fun hcon =>
    data.not_real φref hφref ((IrreducibleCharacter.conjPerm_eq_self_iff φref).mp hcon.symm)
  have hn2 : 2 ≤ n := Finset.one_lt_card.mpr ⟨φref, hφrefT, _, hconjrefT, hrefne⟩
  have : NeZero n := ⟨by omega⟩
  set fam : Fin n → IrreducibleCharacter ↥L :=
    fun i => (T.equivFin.symm i : IrreducibleCharacter ↥L)
    with hfamdef
  have hfam_mem : ∀ i, fam i ∈ T := fun i => (T.equivFin.symm i).2
  have hfam_inj : Function.Injective fam :=
    fun i j h => T.equivFin.symm.injective (Subtype.ext h)
  -- (4) each member difference `fam i − fam 0` is `A(L)`-supported.
  have hdiff_supp : ∀ i, (irreducibleCharacterDifference fam i).support ⊆
      OddOrder.Peterfalvi.S04.supportInSubgroup hyp.ambientA L := by
    intro i y hy
    have hne1 : y ≠ 1 := by
      rintro rfl
      refine (ClassFunction.mem_support.mp hy) ?_
      change (fam i : ClassFunction ↥L ℂ) 1 - (fam 0 : ClassFunction ↥L ℂ) 1 = 0
      rw [hTdeg (fam i) (hfam_mem i), hTdeg (fam 0) (hfam_mem 0), sub_self]
    rcases ClassFunction.support_sub_subset _ _ hy with h | h
    · rcases hTsupp _ (hfam_mem i) h with h2 | h2
      · exact h2
      · exact absurd (Set.mem_singleton_iff.mp h2) hne1
    · rcases hTsupp _ (hfam_mem 0) h with h2 | h2
      · exact h2
      · exact absurd (Set.mem_singleton_iff.mp h2) hne1
  have hSsupp : ∀ s ∈ Set.range (irreducibleCharacterDifference fam),
      s.support ⊆ OddOrder.Peterfalvi.S04.supportInSubgroup hyp.ambientA L := by
    rintro s ⟨i, rfl⟩; exact hdiff_supp i
  -- (5) the three (1.4) hypotheses for the Dade isometry, via the supported-span lemmas.
  have hsame_deg : ∀ i, ((fam i : ClassFunction ↥L ℂ) : ↥L → ℂ) 1
      = ((fam 0 : ClassFunction ↥L ℂ) : ↥L → ℂ) 1 := fun i =>
    (hTdeg (fam i) (hfam_mem i)).trans (hTdeg (fam 0) (hfam_mem 0)).symm
  have hvirtual : IsometryDifferenceImagesAreVirtual hyp.tau fam := by
    intro i
    change hyp.tau (irreducibleCharacterDifference fam i) ∈ ZIrr G
    exact OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_mem_ZIrr_of_supported
      hyp.dadeData.dade (hdiff_supp i)
      (Submodule.sub_mem _ (fam i).mem_ZIrr (fam 0).mem_ZIrr)
  have hzero : IsometryDifferenceImagesVanishAtOne hyp.tau fam := by
    intro i
    change hyp.tau (irreducibleCharacterDifference fam i) (1 : G) = 0
    exact OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_apply_one_eq_zero
      hyp.dadeData.dade (hdiff_supp i)
  have hisom : ∀ i j, ClassFunction.inner (isometryDifferenceImage hyp.tau fam i)
      (isometryDifferenceImage hyp.tau fam j)
      = ClassFunction.inner (irreducibleCharacterDifference fam i)
          (irreducibleCharacterDifference fam j) := by
    intro i j
    change ClassFunction.inner (hyp.tau (irreducibleCharacterDifference fam i))
      (hyp.tau (irreducibleCharacterDifference fam j)) = _
    exact OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_inner_eq_on_supported_span
      hyp.dadeData.dade hSsupp (Submodule.subset_span ⟨i, rfl⟩)
      (Submodule.subset_span ⟨j, rfl⟩)
  -- (6) apply the (1.4) keystone.
  obtain ⟨sdf, himage⟩ :=
    isometry_difference_pair_structure hn2 fam hfam_inj hsame_deg hyp.tau hvirtual hzero hisom
  -- (7) read off `ε` and `μ`.
  refine ⟨sdf.sign, fun x => if hx : x ∈ T then sdf.mu (T.equivFin ⟨x, hx⟩) else sdf.mu 0,
    sdf.sign_eq, ?_, ?_⟩
  · -- InjOn
    intro x hx y hy hxy
    have hxT : x ∈ T := Finset.mem_coe.mp hx
    have hyT : y ∈ T := Finset.mem_coe.mp hy
    simp only [dif_pos hxT, dif_pos hyT] at hxy
    exact Subtype.ext_iff.mp (T.equivFin.injective (sdf.injective hxy))
  · -- the pair relation
    intro α hα β hβ
    have hαT : α ∈ T := hα
    have hβT : β ∈ T := hβ
    -- key: for `x ∈ T`, `τ(x − fam 0) = ε·(μ x − μ₀)`.
    have key : ∀ x (hx : x ∈ T), hyp.tau ((x : ClassFunction ↥L ℂ) - (fam 0 : ClassFunction ↥L ℂ))
        = sdf.sign • ((sdf.mu (T.equivFin ⟨x, hx⟩) : ClassFunction G ℂ)
            - (sdf.mu 0 : ClassFunction G ℂ)) := by
      intro x hx
      have hfx : fam (T.equivFin ⟨x, hx⟩) = x := by
        simp only [hfamdef, Equiv.symm_apply_apply]
      have hLHS : (x : ClassFunction ↥L ℂ) - (fam 0 : ClassFunction ↥L ℂ)
          = irreducibleCharacterDifference fam (T.equivFin ⟨x, hx⟩) := by
        change (x : ClassFunction ↥L ℂ) - (fam 0 : ClassFunction ↥L ℂ)
          = (fam (T.equivFin ⟨x, hx⟩) : ClassFunction ↥L ℂ) - (fam 0 : ClassFunction ↥L ℂ)
        rw [hfx]
      rw [hLHS]
      change isometryDifferenceImage hyp.tau fam (T.equivFin ⟨x, hx⟩) = _
      rw [himage (T.equivFin ⟨x, hx⟩),
        SignedIrreducibleDifferenceFamily.signedDifference_apply,
        SignedIrreducibleDifferenceFamily.difference_apply,
        SignedIrreducibleDifferenceFamily.classFunction_apply,
        SignedIrreducibleDifferenceFamily.classFunction_apply]
    have hsub : (α : ClassFunction ↥L ℂ) - (β : ClassFunction ↥L ℂ)
        = ((α : ClassFunction ↥L ℂ) - (fam 0 : ClassFunction ↥L ℂ))
          - ((β : ClassFunction ↥L ℂ) - (fam 0 : ClassFunction ↥L ℂ)) := by abel
    rw [hsub, map_sub, key α hαT, key β hβT, ← smul_sub]
    congr 1
    simp only [dif_pos hαT, dif_pos hβT]
    abel

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/RhoConstancyDecomposition.lean

View graph

Get started

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

About Prove2Me

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

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me