Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Peterfalvi (12.4): constancy of ρ\rhoρ (coherence pins onward)

Definition
OddOrder_Peterfalvi_S14_MaximalI_RhoConstancy

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

feit-thompsonfinite-groupsgroup-theoryishida-odd-order

The second half of the ρ\rhoρ-constancy development for type I maximal subgroups, starting from the coherence pins of Peterfalvi (12.4), including (5.5) for coherent extensions.

Main declarations: constituent_diff_tau_mem_span (Peterfalvi (12.4), pin (a)); coherent_extension_mem_span_imageFamily (Peterfalvi (5.5)); coherent_extension_constituent_mem_span_Rset (Peterfalvi (5.5) → (12.2.b), constituent form); coherent_extension_constituent_orthogonal_Rset_of_nonconjugate (Peterfalvi (5.5) + (12.3): the L-side Dade character ψ = χ_L^{τ₁} is orthogonal to R(χ_M)).

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/RhoConstancy.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_RhoConstancyDecomposition

namespace OddOrder.Peterfalvi.S14

open OddOrder.GroupTheory

open OddOrder.RepresentationTheory

open OddOrder.Isaacs

open scoped Pointwise

variable {G : Type*} [Group G]

open scoped Classical OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.4), pin (a)** (coherence): for constituents `φ₁, φ₂ ∈ S(χ)`, the Dade image
`(φ₁ − φ₂)^τ` lies in `ℤ[R(χ)]`.

Proof (the (1.4) coherence content, now genuine): the conjugate-closed constituent set `T` is a
single coherent family under `τ` (`exists_uniform_image_of_constituents`), giving a uniform sign
`ε` and injection `μ : T → Irr G` with `τ(φ₁ − φ₂) = ε·(μ φ₁ − μ φ₂)`.  For each constituent `φ`,
the two presentations of `τ(φ − φ̄)` — the global `ε·(μ φ − μ φ̄)` and the per-`φ` block
`R₁(φ)`'s `ε_φ·(μ_φ − ν_φ)` (`R1cdi.image_eq`) — must share their irreducible pair
(`irreducibleCharacter_signed_difference_uniqueness`), so `μ φ ∈ {μ_φ, ν_φ} ⊆ ℤ[R(χ)]`
(`R1cdi_muNu_mem_span_Rset`).  Hence `μ φ₁, μ φ₂ ∈ ℤ[R(χ)]` and `τ(φ₁ − φ₂) ∈ ℤ[R(χ)]`. -/
theorem constituent_diff_tau_mem_span {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    {chi : ClassFunction ↥L ℂ} (dχ : CharacterDecompositionData hyp chi)
    {φ₁ φ₂ : IrreducibleCharacter ↥L} (h₁ : φ₁ ∈ dχ.constituents) (h₂ : φ₂ ∈ dχ.constituents) :
    hyp.tau ((φ₁ : ClassFunction ↥L ℂ) - (φ₂ : ClassFunction ↥L ℂ)) ∈
      Submodule.span ℤ (Rset dχ) := by
  have := hyp.finiteG
  classical
  obtain ⟨ε, μ, hε, hμinj, hμrel⟩ := exists_uniform_image_of_constituents hyp dχ
  set T := dχ.constituents ∪ dχ.constituents.image (IrreducibleCharacter.conjPerm ↥L) with hTdef
  -- reconciliation: every `μ φ` (for a constituent `φ`) lies in `ℤ[R(χ)]`.
  have hmu_mem : ∀ φ ∈ dχ.constituents, (μ φ : ClassFunction G ℂ) ∈ Submodule.span ℤ (Rset dχ) := by
    intro φ hφ
    have hφT : φ ∈ T := Finset.mem_union_left _ hφ
    have hconjT : IrreducibleCharacter.conjPerm ↥L φ ∈ T :=
      Finset.mem_union_right _ (Finset.mem_image_of_mem _ hφ)
    set cdi := R1cdi dχ hφ with hcdi
    -- global vs per-`φ` presentation of `τ(φ − φ̄)`.
    have hglob := hμrel φ hφT (IrreducibleCharacter.conjPerm ↥L φ) hconjT
    rw [IrreducibleCharacter.conjPerm_apply_coe] at hglob
    have hcomb : ε • ((μ φ : ClassFunction G ℂ)
          - (μ (IrreducibleCharacter.conjPerm ↥L φ) : ClassFunction G ℂ))
        = cdi.sign • ((cdi.muClassFunction) - (cdi.nuClassFunction)) :=
      hglob.symm.trans cdi.image_eq
    have hcombℂ : (ε : ℂ) • ((μ φ : ClassFunction G ℂ)
          - (μ (IrreducibleCharacter.conjPerm ↥L φ) : ClassFunction G ℂ))
        = (cdi.sign : ℂ) • (((cdi.mu : ClassFunction G ℂ)) - ((cdi.nu : ClassFunction G ℂ))) := by
      rw [Int.cast_smul_eq_zsmul, Int.cast_smul_eq_zsmul]; exact hcomb
    -- the two distinct-pair hypotheses.
    have hφne : φ ≠ IrreducibleCharacter.conjPerm ↥L φ := fun h =>
      dχ.not_real φ hφ ((IrreducibleCharacter.conjPerm_eq_self_iff φ).mp h.symm)
    have hab : μ φ ≠ μ (IrreducibleCharacter.conjPerm ↥L φ) := fun h =>
      hφne (hμinj (Finset.mem_coe.mpr hφT) (Finset.mem_coe.mpr hconjT) h)
    have hs : (ε : ℂ) ≠ 0 := by rcases hε with h | h <;> simp [h]
    rcases irreducibleCharacter_signed_difference_uniqueness hab cdi.distinct hs hcombℂ with
      ⟨h1, _, _⟩ | ⟨h1, _, _⟩
    · rw [h1]; exact (R1cdi_muNu_mem_span_Rset dχ hφ).1
    · rw [h1]; exact (R1cdi_muNu_mem_span_Rset dχ hφ).2
  -- assemble.
  have hφ₁T : φ₁ ∈ T := Finset.mem_union_left _ h₁
  have hφ₂T : φ₂ ∈ T := Finset.mem_union_left _ h₂
  rw [hμrel φ₁ hφ₁T φ₂ hφ₂T]
  exact Submodule.smul_mem _ _ (Submodule.sub_mem _ (hmu_mem φ₁ h₁) (hmu_mem φ₂ h₂))

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (5.5)** (`χ^{τ₁} ∈ ℤ[R(χ)]` for a coherent extension).  For the coherent
extension `coh` of the type-I family `S` of a `Hypothesis L`, and a non-real member `χ ∈ S`
(as an `IrreducibleCharacter`) whose difference `χ − χ̄` is supported in the Dade domain `A(L)`
and which is orthogonal to its conjugate (`⟨χ, χ̄⟩ = 0`), the Dade character `ψ = χ^{τ₁} =
coh.extension χ` lies in the integral span of the orthonormal image family
`R(χ) = dadeOrthonormalCharacterImageFamilyOfDiff … χ`.

This is the `ψ = 0` case of the (5.4) decomposition `(χ − ψ)^{τ₁} = X − Y`.  Taking the coherent
extension as the auxiliary isometry `τ₁`, its `ZIrr`-codomain (`extension_mem_ZIrr`, the
virtual-character property the general **unsupported** `X`-family `Ind θ` lacks — `χ(1) ≠ 0`)
supplies the single number-theoretic input to `CharacterPsiDecomposition.ofProjection`; then
`eq_sum_of_psi_eq_zero` forces `Y = 0`, so `χ^{τ₁} = X = ∑_{α ∈ E ⊆ R(χ)} α ∈ ℤ[R(χ)]`.  This is
the L-side `ψ ∈ ℤ[R(χ_L)]` which, combined with the (12.3) cross-`L` orthogonality
`R(χ_L) ⊥ R(χ_M)` (`nonconjugate_typeI_R_orthogonal`), yields `ψ ⊥ R(χ_M)` — the `horth` input of
the (12.14) coset-constancy `psi_constant_on_xK`. -/
theorem coherent_extension_mem_span_imageFamily {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    (coh : OddOrder.Peterfalvi.S07.IsCoherent hyp.tau hyp.Sset hyp.A)
    (χ : IrreducibleCharacter ↥L)
    (hχmem : (χ : ClassFunction ↥L ℂ) ∈ hyp.Sset)
    (hχreal : ¬ ClassFunction.IsReal (χ : ClassFunction ↥L ℂ))
    (hdiffsupp : ((χ : ClassFunction ↥L ℂ).conj - (χ : ClassFunction ↥L ℂ)).support ⊆
      OddOrder.Peterfalvi.S04.supportInSubgroup (typeIA L hyp.typeI) L)
    (hχχbar : ClassFunction.inner (χ : ClassFunction ↥L ℂ) (χ : ClassFunction ↥L ℂ).conj = 0) :
    coh.extension (χ : ClassFunction ↥L ℂ) ∈
      Submodule.span ℤ
        ((OddOrder.Peterfalvi.S07.dadeOrthonormalCharacterImageFamilyOfDiff
          hyp.dadeData.dade χ hχreal hdiffsupp).imageSet :
          Set (ClassFunction G ℂ)) := by
  classical
  set imF := OddOrder.Peterfalvi.S07.dadeOrthonormalCharacterImageFamilyOfDiff
    hyp.dadeData.dade χ hχreal hdiffsupp with himF
  -- membership of `χ`, `χ̄`, and `χ − χ̄` in the coherent lattice `ℤ[S]`.
  have hχ_zSpan : (χ : ClassFunction ↥L ℂ) ∈ OddOrder.Peterfalvi.S07.zSpan hyp.Sset :=
    Submodule.subset_span hχmem
  have hχbar_mem : (χ : ClassFunction ↥L ℂ).conj ∈ hyp.Sset :=
    (Sset_closedUnderConjugate hyp).conj_mem hχmem
  have hχbar_zSpan : (χ : ClassFunction ↥L ℂ).conj ∈ OddOrder.Peterfalvi.S07.zSpan hyp.Sset :=
    Submodule.subset_span hχbar_mem
  have hdiff_zSpan : (χ : ClassFunction ↥L ℂ) - (χ : ClassFunction ↥L ℂ).conj ∈
      OddOrder.Peterfalvi.S07.zSpan hyp.Sset :=
    Submodule.sub_mem _ hχ_zSpan hχbar_zSpan
  -- `χ − χ̄` is supported in `A(L)` (sign flip of the given `χ̄ − χ` support).
  have hdiffsupp' : ((χ : ClassFunction ↥L ℂ) - (χ : ClassFunction ↥L ℂ).conj).support ⊆
      hyp.A := by
    have heq : (χ : ClassFunction ↥L ℂ) - (χ : ClassFunction ↥L ℂ).conj
        = -((χ : ClassFunction ↥L ℂ).conj - (χ : ClassFunction ↥L ℂ)) := by abel
    rw [heq, ClassFunction.support_neg]
    exact hdiffsupp
  -- the sublattice `ℤ[χ − χ̄, χ − 0]` sits inside `ℤ[S]`.
  have hsub : OddOrder.Peterfalvi.S07.zSpan (L := ↥L)
        {(χ : ClassFunction ↥L ℂ) - (χ : ClassFunction ↥L ℂ).conj,
          (χ : ClassFunction ↥L ℂ) - 0} ≤
      OddOrder.Peterfalvi.S07.zSpan hyp.Sset := by
    change Submodule.span ℤ _ ≤ Submodule.span ℤ _
    rw [Submodule.span_le]
    intro x hx
    simp only [Set.mem_insert_iff, Set.mem_singleton_iff] at hx
    rcases hx with rfl | rfl
    · exact hdiff_zSpan
    · rw [sub_zero]; exact hχ_zSpan
  -- build the (5.4) decomposition with `ψ = 0` and `τ₁ = coh.extension`, forcing `Y = 0`.
  obtain ⟨-, hτ1χ, E, hEsub, hXsum, -⟩ :=
    OddOrder.Peterfalvi.S07.CharacterPsiDecomposition.eq_sum_of_psi_eq_zero
      (OddOrder.Peterfalvi.S07.CharacterPsiDecomposition.ofProjection (ψ := 0) imF coh.extension
        (fun φ ζ hφ hζ => coh.extension_inner_eq φ ζ (hsub hφ) (hsub hζ))
        (coh.extends_on_supported
          ((χ : ClassFunction ↥L ℂ) - (χ : ClassFunction ↥L ℂ).conj)
          ⟨hdiff_zSpan, hdiffsupp'⟩)
        (by rw [sub_zero]; exact coh.extension_mem_ZIrr _ hχ_zSpan)
        (ClassFunction.inner_zero_right _)
        (ClassFunction.inner_zero_right _)
        hχχbar)
  -- `coh.extension χ = χ^{τ₁} = X = ∑_{α ∈ E} α ∈ ℤ[R(χ)]`.
  have hgoal : coh.extension (χ : ClassFunction ↥L ℂ) = ∑ α ∈ E, α := hτ1χ.trans hXsum
  rw [hgoal]
  exact Submodule.sum_mem _ fun α hα => Submodule.subset_span (Finset.mem_coe.mpr (hEsub hα))

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (5.5) → (12.2.b), constituent form**: the coherent Dade image `χ^{τ₁} =
coh.extension φ` of a constituent `φ ∈ S(χ)` (`data.constituents`) that is itself a member of the
family `S` lies in `ℤ[R(χ)] = Submodule.span ℤ (Rset data)`.

The (5.5) image family `dadeOrthonormalCharacterImageFamilyOfDiff … φ` is *definitionally* the block
`R₁(φ) = R1 data hφ` — both are the `toOrthonormalImage` of the same
`dadeCharacterDifferenceImageOfDiff hyp.dadeData.dade φ (data.not_real φ hφ)
(R1_diffsupp data hφ)` — which is a subfamily of `R(χ) = Rset data`, so
`coherent_extension_mem_span_imageFamily` lands in `ℤ[R(χ)]` after `span_mono`.  The orthogonality
`⟨φ, φ̄⟩ = 0` comes for free from `data.not_real φ hφ` (a non-real irreducible is orthogonal to its
conjugate).  This is the L-side `ψ ∈ ℤ[R(χ_L)]` for the (12.16) witness: the distinguished
`χ_L = Ind θ` is irreducible (Frobenius), so its unique constituent is itself, in `S`. -/
theorem coherent_extension_constituent_mem_span_Rset {L : Subgroup G} [Finite G]
    (hyp : Hypothesis L) (coh : OddOrder.Peterfalvi.S07.IsCoherent hyp.tau hyp.Sset hyp.A)
    {chi : ClassFunction ↥L ℂ} (data : CharacterDecompositionData hyp chi)
    {φ : IrreducibleCharacter ↥L} (hφ : φ ∈ data.constituents)
    (hφmem : (φ : ClassFunction ↥L ℂ) ∈ hyp.Sset) :
    coh.extension (φ : ClassFunction ↥L ℂ) ∈ Submodule.span ℤ (Rset data) := by
  classical
  -- a non-real irreducible is orthogonal to its complex conjugate.
  have hφne : φ ≠ IrreducibleCharacter.conjPerm ↥L φ := fun h =>
    data.not_real φ hφ ((IrreducibleCharacter.conjPerm_eq_self_iff φ).mp h.symm)
  have hχχbar :
      ClassFunction.inner (φ : ClassFunction ↥L ℂ) (φ : ClassFunction ↥L ℂ).conj = 0 := by
    have h0 := irreducibleCharacter_inner_eq_ite φ (IrreducibleCharacter.conjPerm ↥L φ)
    rw [if_neg hφne] at h0
    rwa [IrreducibleCharacter.conjPerm_apply_coe] at h0
  -- (5.5): `coh.extension φ ∈ ℤ[R₁(φ)]`; and `R₁(φ) ⊆ R(χ) = Rset data`.
  have h55 := coherent_extension_mem_span_imageFamily hyp coh φ hφmem
    (data.not_real φ hφ) (R1_diffsupp data hφ) hχχbar
  refine Submodule.span_mono ?_ h55
  intro α hα
  exact ⟨φ, hφ, Finset.mem_coe.mp hα⟩

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (5.5) + (12.3): the L-side Dade character `ψ = χ_L^{τ₁}` is orthogonal to
`R(χ_M)`.**  For two non-conjugate type-I maximal subgroups `L`, `M` and a constituent
`φ_L ∈ S(χ_L)` of the L-side family that is itself a member of `S` (the Frobenius witness case,
where `χ_L = Ind θ` is irreducible), the coherent Dade image `ψ = coh_L.extension φ_L` is orthogonal
to every element of `R(χ_M) = Rset data_M`.

Two ingredients combine: (5.5) `coherent_extension_constituent_mem_span_Rset` puts
`ψ ∈ ℤ[R(χ_L)]`, and (12.3) `nonconjugate_typeI_R_orthogonal` gives the cross-`L` orthogonality
`R(χ_L) ⊥ R(χ_M)`; since `⟨·,·⟩` is conjugate-symmetric and additive, orthogonality of `ψ` to all
of `R(χ_M)` follows from `inner_eq_zero_of_mem_zSpan`.  This is precisely the per-`χ_M` piece of the
`horth` hypothesis that the (12.4)/(12.14) coset-constancy chain (`Sset_coeff_equal`,
`psi_constant_on_xK`) consumes: `ψ` restricted to the `M`-structure has equal coefficients across
`S(χ_M)`, forcing `ψ` constant on the `M_F`-cosets. -/
theorem coherent_extension_constituent_orthogonal_Rset_of_nonconjugate {L M : Subgroup G}
    [Finite G] (hG : OddOrder.BG.IsMinimalSimpleOdd G)
    (hyp_L : Hypothesis L)
    (coh_L : OddOrder.Peterfalvi.S07.IsCoherent hyp_L.tau hyp_L.Sset hyp_L.A)
    {chi_L : ClassFunction ↥L ℂ} (data_L : CharacterDecompositionData hyp_L chi_L)
    {φ_L : IrreducibleCharacter ↥L} (hφ_L : φ_L ∈ data_L.constituents)
    (hφ_L_mem : (φ_L : ClassFunction ↥L ℂ) ∈ hyp_L.Sset)
    (hyp_M : Hypothesis M) (hnot_conj : ¬ ∃ g : G, MulAut.conj g • L = M)
    {chi_M : ClassFunction ↥M ℂ} (data_M : CharacterDecompositionData hyp_M chi_M) :
    ∀ α ∈ Rset data_M,
      ClassFunction.inner (coh_L.extension (φ_L : ClassFunction ↥L ℂ)) α = 0 := by
  -- (5.5): `ψ = coh_L.extension φ_L ∈ ℤ[R(χ_L)]`.
  have h55 := coherent_extension_constituent_mem_span_Rset hyp_L coh_L data_L hφ_L hφ_L_mem
  -- (12.3): `R(χ_L) ⊥ R(χ_M)`.
  have horth := nonconjugate_typeI_R_orthogonal hG hyp_L hyp_M hnot_conj data_L data_M
  intro α hα
  -- `α ⊥ R(χ_L)` (conjugate-swap of (12.3)), hence `α ⊥ ℤ[R(χ_L)] ∋ ψ`; conjugate back.
  have hαperp : ∀ β ∈ Rset data_L, ClassFunction.inner α β = 0 := by
    intro β hβ
    rw [inner_conj_symm β α, horth β hβ α hα, star_zero]
  have h0 : ClassFunction.inner α (coh_L.extension (φ_L : ClassFunction ↥L ℂ)) = 0 :=
    OddOrder.Peterfalvi.S07.IntegralCharacterMap.inner_eq_zero_of_mem_zSpan hαperp h55
  rw [inner_conj_symm α (coh_L.extension (φ_L : ClassFunction ↥L ℂ)), h0, star_zero]

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.5) support input** (the `A1xi12` step of the Coq `o_rpsi_S` proof): the
difference `χ₁ − χ₂` of two **equal-degree** members of `S` vanishes off `H^# = H ∖ {1}` — i.e. at
every `x` with `(x : G) ∉ H` (both `Ind_H^L`-characters vanish off the normal `H = L_F`,
`Sset_vanishes_off_H`) or `x = 1` (equal degree, so `(χ₁ − χ₂)(1) = 0`).  This is the
`xi1 − xi2 ∈ CF(L, H^#)` support hypothesis under which the type-I Dade isometry `τ` acts on the
difference (feeding `constituent_diff_tau_eq_induce` / the `chiRho_adjoint` reciprocity of the
(12.5) Fact-A rebuild). -/
theorem Sset_diff_vanishes_off_H_sharp {L : Subgroup G} (hyp : Hypothesis L)
    {χ₁ χ₂ : ClassFunction ↥L ℂ} (hχ₁ : χ₁ ∈ hyp.Sset) (hχ₂ : χ₂ ∈ hyp.Sset)
    (hdeg : χ₁ (1 : ↥L) = χ₂ (1 : ↥L)) {x : ↥L}
    (hx : (x : G) ∉ hyp.H ∨ x = 1) : (χ₁ - χ₂) x = 0 := by
  rw [ClassFunction.sub_apply]
  rcases hx with hxH | hx1
  · rw [Sset_vanishes_off_H hyp hχ₁ hxH, Sset_vanishes_off_H hyp hχ₂ hxH, sub_zero]
  · subst hx1; rw [hdeg, sub_self]

open scoped Classical in
/-- **General TI-induction self-value** (Isaacs 7.x / Peterfalvi (3.2.c) value half), generalized
from `TICyclicHypothesis.induce_apply_eq_self_of_mem_V` to an arbitrary TI subset.  For a TI subset
`A` relative to `L` (`L ⊆ N_G(A)`, `A ⊆ L`, `IsTISubset A L`) and a class function `α` of `L`
supported in `A`, the induced class function `Ind_L^G α` agrees with `α` on `A`: only the `|L|`
conjugators `x ∈ L` contribute to the induction sum (the others land outside `A`, where `α`
vanishes,
by the TI property), each with value `α(a)`.  This is **pin (b), step 1** — the value-half of the
"Dade map = Ind on the trivial-`H` part" bridge. -/
theorem induce_apply_eq_self_of_mem_tiSubset {A : Set G} {L : Subgroup G}
    [Fintype G] [Invertible (Nat.card L : ℂ)]
    (hAL : A ⊆ (L : Set G))
    (hnorm : ∀ x ∈ L, ∀ a ∈ A, x⁻¹ * a * x ∈ A)
    (hTI : OddOrder.GroupTheory.IsTISubset A L)
    (α : ClassFunction ↥L ℂ)
    (hαsupp : α.support ⊆ OddOrder.Peterfalvi.S04.supportInSubgroup A L)
    {a : G} (ha : a ∈ A) :
    ClassFunction.induce L α a = α ⟨a, hAL ha⟩ := by
  classical
  have : Fintype ↥L := Fintype.ofFinite _
  have haL : a ∈ L := hAL ha
  have hterm : ∀ x : G, ClassFunction.induceTerm L α x a
      = if x ∈ L then α ⟨a, haL⟩ else 0 := by
    intro x
    by_cases hx : x ∈ L
    · rw [if_pos hx]
      have haxL : x⁻¹ * a * x ∈ L := hAL (hnorm x hx a ha)
      rw [ClassFunction.induceTerm_of_mem _ haxL]
      have harg : (⟨x⁻¹ * a * x, haxL⟩ : ↥L)
          = ⟨x⁻¹, L.inv_mem hx⟩ * ⟨a, haL⟩ * ⟨x⁻¹, L.inv_mem hx⟩⁻¹ := by
        apply Subtype.ext; simp [inv_inv]
      rw [harg]
      exact α.conj_eq ⟨a, haL⟩ ⟨x⁻¹, L.inv_mem hx⟩
    · rw [if_neg hx]
      by_cases hax : x⁻¹ * a * x ∈ L
      · rw [ClassFunction.induceTerm_of_mem _ hax]
        have hnotA : x⁻¹ * a * x ∉ A := fun hV =>
          hx (by simpa using L.inv_mem (hTI x⁻¹ ⟨a, ha, by simpa using hV⟩))
        by_contra hne
        exact hnotA ((OddOrder.Peterfalvi.S04.mem_supportInSubgroup).mp
          (hαsupp (ClassFunction.mem_support.mpr hne)))
      · rw [ClassFunction.induceTerm_of_not_mem _ hax]
  rw [ClassFunction.induce_apply, Finset.sum_congr rfl (fun x _ => hterm x),
    ← Finset.sum_filter, Finset.sum_const]
  have hcard : (Finset.univ.filter (· ∈ L)).card = Nat.card ↥L := by
    rw [Nat.card_eq_fintype_card]; exact (Fintype.card_subtype _).symm
  rw [hcard, nsmul_eq_mul, ← mul_assoc, invOf_mul_self, one_mul]

open scoped Classical in
/-- **Peterfalvi (12.4) pin (b), step 2**: for a Dade hypothesis with all trivial stabilizers
`∀ a, H(a) = ⊥`, induction `Ind_L^G` (restricted to `CF(L, A)`) **is** the Dade map.  Generalizes
`TICyclicHypothesis.isDadeMap_inducedDadeMap`: the value half is the step-1 self-value
`induce_apply_eq_self_of_mem_tiSubset` (the coset condition collapses to `IsConj a g` since
`h ∈ ⊥`),
the support half is `induce_eq_zero_of_not_conjugatesIntoSet` (induced functions vanish off the
`A`-conjugates, which are the Dade support when `H = ⊥`).  Via `IsDadeMap.unique` this pins the
abstract Dade map to `Ind_L^G` on `CF(L, A)`. -/
theorem isDadeMap_induce_of_forall_H_eq_bot {A : Set G} {L : Subgroup G}
    [Fintype G] [Invertible (Nat.card L : ℂ)]
    (hyp : OddOrder.Peterfalvi.S04.Hypothesis G A L)
    (hH : ∀ a, hyp.H a = ⊥) :
    OddOrder.Peterfalvi.S04.IsDadeMap hyp
      (fun α => ClassFunction.induce L (α : ClassFunction ↥L ℂ)) where
  map_eq_of_isConj_hCoset := by
    intro α g a h hh hconj
    have hh1 : h = 1 := Subgroup.mem_bot.mp (by rw [← hH a]; exact hh)
    subst hh1
    have hga : IsConj a.1 g := by simpa using hconj
    change ClassFunction.induce L (α : ClassFunction ↥L ℂ) g = _
    rw [← (ClassFunction.induce L (α : ClassFunction ↥L ℂ)).of_isConj hga]
    exact induce_apply_eq_self_of_mem_tiSubset hyp.subset_L
      (fun x hx a' ha' => by simpa using hyp.L_normalizes_A ⟨x⁻¹, L.inv_mem hx⟩ ha')
      (hyp.isTISubset_of_forall_H_eq_bot hH) _
      (ClassFunction.mem_supportedSubmodule.mp α.2) a.2
  map_eq_zero_of_not_mem_dadeSupport := by
    intro α g hg
    change ClassFunction.induce L (α : ClassFunction ↥L ℂ) g = 0
    refine ClassFunction.induce_eq_zero_of_not_conjugatesIntoSet
      (ClassFunction.mem_supportedSubmodule.mp α.2) (fun hgin => hg ?_)
    rw [hyp.dadeSupport_eq_conjugatesOfSet_of_forall_H_eq_bot hH]
    obtain ⟨x, hx, hxV⟩ := hgin
    rw [OddOrder.Peterfalvi.S04.mem_supportInSubgroup] at hxV
    exact Group.mem_conjugatesOfSet_iff.mpr ⟨x⁻¹ * g * x, hxV, isConj_iff.mpr ⟨x, by group⟩⟩

/-- **Peterfalvi (12.4) pin (b), step 3** (restriction assembly): if a sub-support `A₁ ⊆ A` carries
only trivial Dade stabilizers (`(hyp.restrict …).H a = ⊥`), then on `A₁`-supported functions the
abstract Dade map of `hyp` **is** induction `Ind_L^G`.  The restricted hypothesis has `H = ⊥`, so
its
Dade map is `Ind_L^G` (step 2 + `IsDadeMap.unique`); `Hypothesis.dadeMap_restrict_apply` identifies
it with `hyp.dadeMap` of the included function. -/
theorem dadeMap_eq_induce_of_supported_on_trivial_H {A : Set G} {L : Subgroup G}
    [Fintype G] [Invertible (Nat.card L : ℂ)]
    (hyp : OddOrder.Peterfalvi.S04.Hypothesis G A L) {A₁ : Set G} (hA₁A : A₁ ⊆ A)
    (hA₁norm : ∀ (l : L) ⦃a : G⦄, a ∈ A₁ → (l : G) * a * (l : G)⁻¹ ∈ A₁)
    (hH₁ : ∀ a, (hyp.restrict hA₁A hA₁norm).H a = ⊥)
    (α : OddOrder.Peterfalvi.S04.SupportedClassFunctions (G := G) ℂ A₁ L) :
    hyp.dadeMap (OddOrder.Peterfalvi.S04.SupportedClassFunctions.inclusion
        (G := G) (k := ℂ) (L := L) hA₁A α)
      = ClassFunction.induce L (α : ClassFunction ↥L ℂ) := by
  have h1 := OddOrder.Peterfalvi.S04.IsDadeMap.unique
    ((hyp.restrict hA₁A hA₁norm).isDadeMap_dadeMap (k := ℂ))
    (isDadeMap_induce_of_forall_H_eq_bot (hyp.restrict hA₁A hA₁norm) hH₁)
  rw [← hyp.dadeMap_restrict_apply hA₁A hA₁norm α]
  exact congrFun h1 α

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.4) pin (b), type-I bridge**: for a type-I maximal `L`, on a class function `f`
supported in a trivial-`H` sub-support `A₁ ⊆ A(L)` (an `L`-invariant subset on which the type-I Dade
stabilizers vanish), the Dade isometry `τ` acts as induction `Ind_L^G`.  This instantiates the
general step-3 bridge `dadeMap_eq_induce_of_supported_on_trivial_H` at the type-I Dade map `hyp.tau`
(via `dadeIntegralCharacterMap_apply_of_support`, and `inclusion` to widen the support from
`A₁`). -/
theorem typeI_tau_eq_induce_of_supported_trivial_H {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    {A₁ : Set G} (hA₁A : A₁ ⊆ hyp.ambientA)
    (hA₁norm : ∀ (l : L) ⦃a : G⦄, a ∈ A₁ → (l : G) * a * (l : G)⁻¹ ∈ A₁)
    (hH₁ : ∀ a, (hyp.dadeData.dade.restrict hA₁A hA₁norm).H a = ⊥)
    {f : ClassFunction ↥L ℂ}
    (hf : f.support ⊆ OddOrder.Peterfalvi.S04.supportInSubgroup A₁ L) :
    hyp.tau f = ClassFunction.induce L f := by
  have := hyp.finiteG
  have hfA : f.support ⊆ OddOrder.Peterfalvi.S04.supportInSubgroup hyp.ambientA L :=
    hf.trans (OddOrder.Peterfalvi.S04.supportInSubgroup_mono hA₁A)
  have h1 : hyp.tau f = hyp.dadeData.dade.dadeMap (k := ℂ)
      ⟨f, (ClassFunction.mem_supportedSubmodule).mpr hfA⟩ :=
    OddOrder.Peterfalvi.S07.dadeIntegralCharacterMap_apply_of_support hyp.dadeData.dade
      (hyp.dadeData.dade.fullDadeIsometryData) hfA
  rw [h1]
  -- `⟨f, hfA⟩` is defeq to `inclusion hA₁A ⟨f, hf⟩` (same carrier `f`), so step 3 applies directly.
  exact dadeMap_eq_induce_of_supported_on_trivial_H hyp.dadeData.dade hA₁A hA₁norm hH₁
    ⟨f, (ClassFunction.mem_supportedSubmodule).mpr hf⟩

/-- The escaping-centralizer set `{a ∈ X : ¬ C_G(a) ≤ M}` is `M`-conjugation invariant when `X` is
(`C_G(gag⁻¹) = g·C_G(a)·g⁻¹` and `g ∈ M` normalizes `M`).  The `L`-invariance of the trivial-`H`
sub-support `A(L) ∖ escaping` rests on this. -/
private theorem escaping_conj_mem_iff {M : Subgroup G} {X : Set G} {g x : G}
    (hg : g ∈ M) (hmem : g * x * g⁻¹ ∈ X ↔ x ∈ X) :
    g * x * g⁻¹ ∈ escapingCentralizerSet M X ↔ x ∈ escapingCentralizerSet M X := by
  have hcent : (Subgroup.centralizer ({g * x * g⁻¹} : Set G) ≤ M)
      ↔ (Subgroup.centralizer ({x} : Set G) ≤ M) := by
    rw [← conj_smul_centralizer_singleton]
    conv_lhs => rw [← (conj_smul_eq_self_of_mem_normalizer (Subgroup.le_normalizer hg) :
      MulAut.conj g • M = M)]
    exact Subgroup.pointwise_smul_le_pointwise_smul_iff
  simp only [escapingCentralizerSet, Set.mem_ofPred_eq, hmem, hcent]

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Constituents of `χ = Ind_K^L θ` (`K = (L_F).subgroupOf L ⊴ L`) have equal restriction to `K`**
(the multiplicity-one case of Clifford's theorem [Is] 6.2, character level).  Two constituents
`φ₁, φ₂ ∈ S(χ)` both occur in `χ = Ind_K^L θ` with multiplicity one, so both lie over `θ`; by
Clifford single-orbit (`restrictionConstituentsSingleOrbit_of_isIrreducible`) the constituents of
`Res_K φᵢ` are exactly the `L`-orbit of `θ`, each with the common multiplicity one, whence
`Res_K φ₁ = Res_K φ₂`.  Computed at the inner-product level via Fourier: for every `ψ ∈ Irr K`,
`⟨Res_K φᵢ, ψ⟩ = ⟨φᵢ, Ind_K ψ⟩` (Frobenius) is `1` when `ψ` is `L`-conjugate to `θ`
(`Ind_K ψ = Ind_K θ = χ`, multiplicity one) and `0` otherwise (single-orbit), independently of `i`.
This is the [Is] 6.2 input Peterfalvi (12.4) cites for `Supp(φ₁ − φ₂) ⊆ A(L) − H^#`. -/
theorem restrict_eq_of_mem_constituents {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    {chi : ClassFunction ↥L ℂ} (dχ : CharacterDecompositionData hyp chi)
    {φ₁ φ₂ : IrreducibleCharacter ↥L} (h₁ : φ₁ ∈ dχ.constituents) (h₂ : φ₂ ∈ dχ.constituents) :
    ClassFunction.restrict ((hyp.typeI.typeF.H).subgroupOf L) (φ₁ : ClassFunction ↥L ℂ)
      = ClassFunction.restrict ((hyp.typeI.typeF.H).subgroupOf L) (φ₂ : ClassFunction ↥L ℂ) := by
  have := hyp.finiteG
  classical
  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
  set K := (hyp.typeI.typeF.H).subgroupOf L with hKdef
  obtain ⟨θ, hθ_ne, hchi_eq⟩ := dχ.chi_mem
  -- Frobenius reciprocity: `⟨φ, Ind_K ψ⟩ = ⟨Res_K φ, ψ⟩`.
  have hfrob : ∀ (φ : IrreducibleCharacter ↥L) (ψ : IrreducibleCharacter ↥K),
      ClassFunction.inner (φ : ClassFunction ↥L ℂ) (ClassFunction.induce K (ψ : ClassFunction ↥K ℂ))
        = ClassFunction.inner (ClassFunction.restrict K (φ : ClassFunction ↥L ℂ))
          (ψ : ClassFunction ↥K ℂ) := by
    intro φ ψ
    rw [OddOrder.RepresentationTheory.inner_conj_symm,
      ClassFunction.inner_induce_eq_inner_restrict,
      OddOrder.RepresentationTheory.inner_conj_symm, star_star]
  -- multiplicity-one: `⟨φ, χ⟩ = 1` for a constituent `φ`.
  have hmult : ∀ φ ∈ dχ.constituents, ClassFunction.inner (φ : ClassFunction ↥L ℂ) chi = 1 := by
    intro φ hφ
    rw [dχ.decomp, inner_sum_right,
      Finset.sum_eq_single_of_mem φ hφ (fun φ' _ hne => by
        rw [irreducibleCharacter_inner, if_neg (Ne.symm hne)]),
      irreducibleCharacter_inner, if_pos rfl]
  -- per-`ψ` value of `⟨Res_K φ, ψ⟩`, independent of the constituent `φ`.
  have hval : ∀ φ ∈ dχ.constituents, ∀ ψ : IrreducibleCharacter ↥K,
      ClassFunction.inner (ClassFunction.restrict K (φ : ClassFunction ↥L ℂ))
          (ψ : ClassFunction ↥K ℂ)
        = if (∃ g : ↥L, IrreducibleCharacter.conjBy g θ = ψ) then (1 : ℂ) else 0 := by
    intro φ hφ ψ
    rw [← hfrob φ ψ]
    by_cases hc : ∃ g : ↥L, IrreducibleCharacter.conjBy g θ = ψ
    · rw [if_pos hc]
      obtain ⟨g, rfl⟩ := hc
      rw [IrreducibleCharacter.coe_conjBy, ClassFunction.induce_conjBy_eq, ← hchi_eq]
      exact hmult φ hφ
    · rw [if_neg hc]
      by_contra hne
      refine hc ?_
      have hoθ : IrreducibleCharacter.LiesOver K φ θ := by
        rw [IrreducibleCharacter.LiesOver, ClassFunction.restrictionMultiplicity_def,
          ← hfrob φ θ, ← hchi_eq, hmult φ hφ]
        exact one_ne_zero
      have hoψ : IrreducibleCharacter.LiesOver K φ ψ := by
        rw [IrreducibleCharacter.LiesOver, ClassFunction.restrictionMultiplicity_def, ← hfrob φ ψ]
        exact hne
      exact (restrictionConstituentsSingleOrbit_of_isIrreducible φ).exists_conj hoθ hoψ
  -- Fourier: equal inner products with every irreducible `ψ ∈ Irr K` force equality.
  rw [← OddOrder.RepresentationTheory.sum_inner_irreducibleCharacter_smul
        (ClassFunction.restrict K (φ₁ : ClassFunction ↥L ℂ)),
      ← OddOrder.RepresentationTheory.sum_inner_irreducibleCharacter_smul
        (ClassFunction.restrict K (φ₂ : ClassFunction ↥L ℂ))]
  refine Finset.sum_congr rfl fun ψ _ => ?_
  rw [hval φ₁ h₁ ψ, hval φ₂ h₂ ψ]

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.4), the §8 support obligation** ([Is] 6.2 + (8.12.a)): for constituents
`φ₁, φ₂ ∈ S(χ)`, the difference `φ₁ − φ₂` is supported on the **non-escaping** part of `A(L)`,
`A₁ = {a ∈ A(L) : C_G(a) ≤ L}` (= `A(L) − H^#`, exactly where the type-I Dade stabilizers vanish).
By [Is] 6.2 `Res_H φ₁ = Res_H φ₂` (`restrict_eq_of_mem_constituents`, the multiplicity-one Clifford
restriction), so `φ₁ − φ₂` vanishes on `H`; each `φᵢ` is supported on `A(L) ∪ {1}` (carrier
`supported`) and the difference cancels the value at `1` (equal degree).  The escaping points of
`A(L)` lie in `A₁ = H^#` ((8.13.b) `escaping_typeIA_mem_A1`), so the difference vanishes there. -/
theorem constituent_diff_support_subset_nonescaping [Finite G] {L : Subgroup G}
    (hG : OddOrder.BG.IsMinimalSimpleOdd G) (hyp : Hypothesis L)
    {chi : ClassFunction ↥L ℂ} (dχ : CharacterDecompositionData hyp chi)
    {φ₁ φ₂ : IrreducibleCharacter ↥L} (h₁ : φ₁ ∈ dχ.constituents) (h₂ : φ₂ ∈ dχ.constituents) :
    ((φ₁ : ClassFunction ↥L ℂ) - (φ₂ : ClassFunction ↥L ℂ)).support ⊆
      OddOrder.Peterfalvi.S04.supportInSubgroup
        (hyp.ambientA \ escapingCentralizerSet L hyp.ambientA) L := by
  have := hyp.finiteG
  classical
  have hres := restrict_eq_of_mem_constituents hyp dχ h₁ h₂
  intro x hx
  rw [ClassFunction.mem_support] at hx
  rw [OddOrder.Peterfalvi.S04.mem_supportInSubgroup, Set.mem_sdiff]
  -- `x` lies in the support of `φ₁` or `φ₂`, hence in `A(L) ∪ {1}`.
  have hxsupp : x ∈ OddOrder.Peterfalvi.S04.supportInSubgroup hyp.ambientA L ∪ ({1} : Set ↥L) := by
    rcases ne_or_eq ((φ₁ : ClassFunction ↥L ℂ) x) 0 with h | h
    · exact dχ.supported φ₁ h₁ (ClassFunction.mem_support.mpr h)
    · refine dχ.supported φ₂ h₂ (ClassFunction.mem_support.mpr ?_)
      intro h2
      exact hx (by rw [ClassFunction.sub_apply, h, h2, sub_zero])
  -- `x ≠ 1`: the difference vanishes at `1` by equal degree.
  have hx1 : x ≠ 1 := by
    rintro rfl
    exact hx (by rw [ClassFunction.sub_apply, ← dχ.equal_degree φ₁ h₁ φ₂ h₂, sub_self])
  have hxAmem : (x : G) ∈ hyp.ambientA := by
    rcases hxsupp with h | h
    · exact OddOrder.Peterfalvi.S04.mem_supportInSubgroup.mp h
    · exact absurd (Set.mem_singleton_iff.mp h) hx1
  refine ⟨hxAmem, fun hesc => ?_⟩
  -- an escaping point of `A(L)` lies in `A₁ = H^#`, so in `H`, where the two restrictions agree.
  have hxA1 : (x : G) ∈ A1 L PeterfalviType.I :=
    OddOrder.Peterfalvi.S10.escaping_typeIA_mem_A1 hG hyp.maximal hyp.typeI hesc
  have hxH : (x : G) ∈ hyp.typeI.typeF.H := by
    rw [hyp.typeI.typeF.H_eq]
    have hmem : (x : G) ∈ OddOrder.GroupTheory.sharpSubgroup (maxNilpotentNormalHall L) := hxA1
    exact ((Set.mem_sdiff _).mp hmem).1
  have hxK : x ∈ (hyp.typeI.typeF.H).subgroupOf L := Subgroup.mem_subgroupOf.mpr hxH
  refine hx ?_
  have hev : ClassFunction.restrict ((hyp.typeI.typeF.H).subgroupOf L)
        (φ₁ : ClassFunction ↥L ℂ) ⟨x, hxK⟩
      = ClassFunction.restrict ((hyp.typeI.typeF.H).subgroupOf L)
        (φ₂ : ClassFunction ↥L ℂ) ⟨x, hxK⟩ := by rw [hres]
  rw [ClassFunction.restrict_apply, ClassFunction.restrict_apply] at hev
  rw [ClassFunction.sub_apply, hev, sub_self]

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
set_option backward.isDefEq.respectTransparency false in
/-- **Peterfalvi (12.4), pin (b)** ([Is] 7.7 + (8.12.c) + [Is] 6.2): for constituents
`φ₁, φ₂ ∈ S(χ)`,
the Dade isometry acts as induction on the difference, `(φ₁ − φ₂)^τ = Ind_L^G(φ₁ − φ₂)`.

Proof (now genuine, modulo the §8 support obligation): `φ₁ − φ₂` is supported on the non-escaping
part `A₁ = {a ∈ A(L) : C_G(a) ≤ L}` (`constituent_diff_support_subset_nonescaping`), which is
`L`-invariant (`escaping_conj_mem_iff` + `A(L)` `L`-invariant) and carries only trivial Dade
stabilizers (`ftSupportKernel = ⊥` off the escaping set, via `H_eq_ftSupportKernel`).  On such a
trivial-`H` support the type-I Dade isometry coincides with `Ind_L^G`
(`typeI_tau_eq_induce_of_supported_trivial_H`, i.e. pin (b) steps 1–3 + the restriction
assembly). -/
theorem constituent_diff_tau_eq_induce {L : Subgroup G} [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G) (hyp : Hypothesis L)
    {chi : ClassFunction ↥L ℂ} (dχ : CharacterDecompositionData hyp chi)
    {φ₁ φ₂ : IrreducibleCharacter ↥L} (h₁ : φ₁ ∈ dχ.constituents) (h₂ : φ₂ ∈ dχ.constituents) :
    hyp.tau ((φ₁ : ClassFunction ↥L ℂ) - (φ₂ : ClassFunction ↥L ℂ)) =
      ClassFunction.induce L ((φ₁ : ClassFunction ↥L ℂ) - (φ₂ : ClassFunction ↥L ℂ)) := by
  have hmem : ∀ (l : L) (a : G), ((l : G) * a * (l : G)⁻¹ ∈ hyp.ambientA ↔ a ∈ hyp.ambientA) := by
    intro l a
    refine ⟨fun h => ?_, fun h => hyp.dadeData.dade.L_normalizes_A l h⟩
    have h2 := hyp.dadeData.dade.L_normalizes_A l⁻¹ h
    have h3 : a ∈ typeIA L hyp.typeI := by simpa [Subgroup.coe_inv, mul_assoc] using h2
    exact h3
  have hA₁A : hyp.ambientA \ escapingCentralizerSet L hyp.ambientA ⊆ hyp.ambientA :=
    Set.sdiff_subset
  have hA₁norm : ∀ (l : L) ⦃a : G⦄,
      a ∈ hyp.ambientA \ escapingCentralizerSet L hyp.ambientA →
      (l : G) * a * (l : G)⁻¹ ∈ hyp.ambientA \ escapingCentralizerSet L hyp.ambientA := by
    intro l a ha
    exact ⟨hyp.dadeData.dade.L_normalizes_A l ha.1,
      fun hesc => ha.2 ((escaping_conj_mem_iff l.2 (hmem l a)).mp hesc)⟩
  have hH₁ : ∀ a, (hyp.dadeData.dade.restrict hA₁A hA₁norm).H a = ⊥ := by
    intro a
    rw [OddOrder.Peterfalvi.S04.Hypothesis.restrict_H, hyp.dadeData.H_eq_ftSupportKernel]
    exact OddOrder.Peterfalvi.S10.ftSupportKernel_eq_bot_of_not_escaping a.2.2
  exact typeI_tau_eq_induce_of_supported_trivial_H hyp hA₁A hA₁norm hH₁
    (constituent_diff_support_subset_nonescaping hG hyp dχ h₁ h₂)

open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.4), coherence → coefficient-equality bridge** (genuine).  If `ψ ⊥ R(χ)`, then
`Res_L ψ` has the *same* coefficient on every constituent of `χ`: `⟨Res_L ψ, φ₁⟩ = ⟨Res_L ψ, φ₂⟩`
for `φ₁, φ₂ ∈ S(χ)`.  Proof: `⟨Res_L ψ, φ₁ − φ₂⟩ = ⟨ψ, Ind_L^G(φ₁ − φ₂)⟩ = ⟨ψ, (φ₁ − φ₂)^τ⟩`
(Frobenius `inner_induce_eq_inner_restrict` + conjugate symmetry + pin (b)), and this is `0` because
`(φ₁ − φ₂)^τ ∈ ℤ[R(χ)]` (pin (a)) and `ψ ⊥ R(χ)` (`inner_eq_zero_of_mem_zSpan`).  This is the
genuine
content by which `ψ ⊥ R(χ)` forces the `∪S(χ)`-part of `Res_L ψ` to be `β = ∑_χ c_χ·χ ∈ ℂ[S]`. -/
theorem Sset_coeff_equal {L : Subgroup G} [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G) (hyp : Hypothesis L)
    {chi : ClassFunction ↥L ℂ} (dχ : CharacterDecompositionData hyp chi)
    {psi : ClassFunction G ℂ} (horth : ∀ α ∈ Rset dχ, ClassFunction.inner psi α = 0)
    {φ₁ φ₂ : IrreducibleCharacter ↥L} (h₁ : φ₁ ∈ dχ.constituents) (h₂ : φ₂ ∈ dχ.constituents) :
    ClassFunction.inner (ClassFunction.restrict L psi) (φ₁ : ClassFunction ↥L ℂ)
      = ClassFunction.inner (ClassFunction.restrict L psi) (φ₂ : ClassFunction ↥L ℂ) := by
  have := hyp.finiteG
  set f : ClassFunction ↥L ℂ :=
    (φ₁ : ClassFunction ↥L ℂ) - (φ₂ : ClassFunction ↥L ℂ) with hf
  -- `⟨ψ, τ f⟩ = 0`: `τ f ∈ ℤ[R(χ)]` (pin a) and `ψ ⊥ R(χ)`.
  have hψτ : ClassFunction.inner psi (hyp.tau f) = 0 :=
    OddOrder.Peterfalvi.S07.IntegralCharacterMap.inner_eq_zero_of_mem_zSpan horth
      (constituent_diff_tau_mem_span hyp dχ h₁ h₂)
  -- `⟨f, Res_L ψ⟩ = ⟨Ind_L^G f, ψ⟩ = ⟨τ f, ψ⟩ = star⟨ψ, τ f⟩ = 0`.
  have hfres : ClassFunction.inner f (ClassFunction.restrict L psi) = 0 := by
    rw [← ClassFunction.inner_induce_eq_inner_restrict L f psi,
      ← constituent_diff_tau_eq_induce hG hyp dχ h₁ h₂,
      inner_conj_symm psi (hyp.tau f), hψτ, star_zero]
  -- `⟨Res_L ψ, f⟩ = star⟨f, Res_L ψ⟩ = 0`, then split the difference.
  have hresf : ClassFunction.inner (ClassFunction.restrict L psi) f = 0 := by
    rw [inner_conj_symm f (ClassFunction.restrict L psi), hfres, star_zero]
  rw [hf, ClassFunction.inner_sub_right] at hresf
  exact sub_eq_zero.mp hresf

/-- The "`H ⊆ ker φ`" predicate: the Fitting subgroup `H = L_F` lies in the character kernel of the
irreducible character `φ` of `L`.  The `γ`-components of `Res_L ψ` in (12.4) are exactly those `φ`
with `InHKernel`; they are constant on `H`-cosets (`apply_mul_eq_of_mem_characterKernel`). -/
def InHKernel {L : Subgroup G} (hyp : Hypothesis L) (φ : IrreducibleCharacter ↥L) : Prop :=
  ((hyp.typeI.typeF.H).subgroupOf L : Set ↥L) ⊆
    OddOrder.Peterfalvi.S03.characterKernel (φ : ClassFunction ↥L ℂ)

open scoped Classical OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Toward (12.4) pin (c'), the off-kernel direction** (genuine): every constituent `φ ∈ S(χ)` of
a
member `χ = Ind_H^L θ ∈ S` (`θ ≠ 1_H`) is off-kernel, `H ⊄ ker φ`.  If `H ⊆ ker φ` then `Res_H φ` is
constant `= φ(1)` on `H` (`= φ(1)·1_H`), so by Frobenius
`⟨χ, φ⟩ = ⟨θ, Res_H φ⟩ = star(φ(1))·⟨θ, 1_H⟩
= 0` (`θ ≠ 1_H`); but `φ` a constituent of `χ` gives `⟨χ, φ⟩ = 1`.  This is the `⊇` inclusion
`S(χ) ⊆ {φ : H ⊄ ker φ}` of the partition `exists_offKernel_constituent_partition`. -/
theorem constituents_not_inHKernel {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    {chi : ClassFunction ↥L ℂ} (hχ : chi ∈ hyp.Sset) (dχ : CharacterDecompositionData hyp chi)
    {φ : IrreducibleCharacter ↥L} (hφ : φ ∈ dχ.constituents) : ¬ InHKernel hyp φ := by
  have := hyp.finiteG
  classical
  obtain ⟨θ, hθ_ne, hchi_eq⟩ := hχ
  intro hker
  set K := (hyp.typeI.typeF.H).subgroupOf L with hKdef
  -- `Res_K φ = φ(1) · 1_K` (since `H ⊆ ker φ`, `φ` is constant `= φ(1)` on `K`).
  have hrestrict : ClassFunction.restrict K (φ : ClassFunction ↥L ℂ)
      = OddOrder.Peterfalvi.S03.characterDegree (φ : ClassFunction ↥L ℂ) •
        (trivialIrreducibleCharacter ↥K : ClassFunction ↥K ℂ) := by
    ext k
    rw [ClassFunction.restrict_apply, ClassFunction.smul_apply]
    have hmem : (↑k : ↥L) ∈ OddOrder.Peterfalvi.S03.characterKernel (φ : ClassFunction ↥L ℂ) :=
      hker k.2
    rw [OddOrder.Peterfalvi.S03.mem_characterKernel] at hmem
    simp only [hmem, IrreducibleCharacter.coe_trivialIrreducibleCharacter,
      trivialClassFunction_apply, mul_one]
  -- `⟨χ, φ⟩ = ⟨θ, Res_K φ⟩ = star(φ(1)) · ⟨θ, 1_K⟩ = 0` (`θ ≠ 1_K`).
  have hzero : ClassFunction.inner chi (φ : ClassFunction ↥L ℂ) = 0 := by
    rw [hchi_eq, ClassFunction.inner_induce_eq_inner_restrict, hrestrict,
      OddOrder.RepresentationTheory.inner_smul_right, irreducibleCharacter_inner, if_neg hθ_ne,
      mul_zero]
  -- `⟨χ, φ⟩ = 1` (multiplicity-one constituent), contradiction.
  have hone : ClassFunction.inner chi (φ : ClassFunction ↥L ℂ) = 1 := by
    rw [dχ.decomp, inner_sum_left,
      Finset.sum_eq_single_of_mem φ hφ (fun φ' _ hne => by
        rw [irreducibleCharacter_inner, if_neg hne]),
      irreducibleCharacter_inner, if_pos rfl]
  rw [hone] at hzero
  exact one_ne_zero hzero

open scoped Classical OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Toward (12.4) pin (c'), the capturing direction** (genuine): every off-kernel irreducible `φ`
(`H ⊄ ker φ`) is a constituent of some `χ ∈ S`.  By `exists_constituent_not_subset_characterKernel`
([Is] 6.5 / constituent transitivity), `Res_H φ` has a constituent `θ ≠ 1_H`; then
`χ := Ind_H^L θ ∈ S`
and `⟨φ, χ⟩ = ⟨Res_H φ, θ⟩ ≠ 0` (Frobenius `inner_induce_eq_inner_restrict` + conjugate symmetry),
so
`φ ∈ S(χ)`.  This is the `⊆` inclusion `{φ : H ⊄ ker φ} ⊆ ⋃ S(χ)` of the partition
`exists_offKernel_constituent_partition`. -/
theorem not_inHKernel_imp_mem_constituents {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    (data : ∀ χ ∈ hyp.Sset, CharacterDecompositionData hyp χ)
    {φ : IrreducibleCharacter ↥L} (hφ : ¬ InHKernel hyp φ) :
    ∃ (χ : ClassFunction ↥L ℂ) (hχ : χ ∈ hyp.Sset), φ ∈ (data χ hχ).constituents := by
  have := hyp.finiteG
  classical
  obtain ⟨θ, hlo, hθker⟩ :=
    exists_constituent_not_subset_characterKernel
      (le_refl ((hyp.typeI.typeF.H).subgroupOf L)) φ hφ
  -- `θ ≠ 1`: else `K = K.subgroupOf K ⊆ ker θ = univ`, contradicting `hθker`.
  have hθ_ne : θ ≠ trivialIrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf L) := by
    rintro rfl
    exact hθker (by
      simp only [IrreducibleCharacter.coe_trivialIrreducibleCharacter,
        OddOrder.Peterfalvi.S03.characterKernel_trivialClassFunction, Set.subset_univ])
  -- `θ` is a genuine constituent of `Res_K φ`.
  have hlo' : ClassFunction.inner
      (ClassFunction.restrict ((hyp.typeI.typeF.H).subgroupOf L) (φ : ClassFunction ↥L ℂ))
      (θ : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ) ≠ 0 := by
    rw [IrreducibleCharacter.LiesOver, ClassFunction.restrictionMultiplicity_def] at hlo
    exact hlo
  refine ⟨ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L)
    (θ : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ), ⟨θ, hθ_ne, rfl⟩, ?_⟩
  -- `⟨φ, Ind_K θ⟩ = ⟨Res_K φ, θ⟩ ≠ 0` (Frobenius + conjugate symmetry).
  have hinner_ne : ClassFunction.inner (φ : ClassFunction ↥L ℂ)
      (ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L)
        (θ : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ)) ≠ 0 := by
    rw [OddOrder.RepresentationTheory.inner_conj_symm,
      ClassFunction.inner_induce_eq_inner_restrict,
      OddOrder.RepresentationTheory.inner_conj_symm, star_star]
    exact hlo'
  by_contra hnotin
  apply hinner_ne
  rw [(data _ ⟨θ, hθ_ne, rfl⟩).decomp, inner_sum_right]
  refine Finset.sum_eq_zero fun φ' hφ' => ?_
  have hne : φ ≠ φ' := by rintro rfl; exact hnotin hφ'
  rw [irreducibleCharacter_inner, if_neg hne]

open scoped Classical OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Toward (12.4) pin (c'), disjointness** (genuine, Clifford single-orbit): a `φ` that is a
constituent of both `χ, χ' ∈ S` forces `χ = χ'`.  Writing `χ = Ind_H^L θ`, `χ' = Ind_H^L θ'`, both
witnesses `θ, θ'` lie under `φ` (`⟨Res_H φ, θ⟩ = ⟨φ, χ⟩ ≠ 0`, Frobenius); by Clifford single-orbit
(`restrictionConstituentsSingleOrbit_of_isIrreducible` + `.exists_conj`) they are `L`-conjugate,
`conjBy g θ = θ'`, so `Ind θ = Ind θ'` (`induce_conjBy_eq`, Peterfalvi (1.5.a)).  This is the
`PairwiseDisjoint` content of `exists_offKernel_constituent_partition`. -/
theorem constituents_eq_of_mem {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    {χ χ' : ClassFunction ↥L ℂ} (hχ : χ ∈ hyp.Sset) (hχ' : χ' ∈ hyp.Sset)
    (dχ : CharacterDecompositionData hyp χ) (dχ' : CharacterDecompositionData hyp χ')
    {φ : IrreducibleCharacter ↥L} (hmem : φ ∈ dχ.constituents) (hmem' : φ ∈ dχ'.constituents) :
    χ = χ' := by
  have := hyp.finiteG
  classical
  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 ⟨θ, hθ_ne, rfl⟩ := hχ
  obtain ⟨θ', hθ'_ne, rfl⟩ := hχ'
  -- `θ`, `θ'` lie under `φ`: `⟨Res_K φ, η⟩ = ⟨φ, Ind_K η⟩ = 1 ≠ 0` for a constituent.
  have key : ∀ (η : IrreducibleCharacter ↥((hyp.typeI.typeF.H).subgroupOf L))
      (dη : CharacterDecompositionData hyp
        (ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L) (η : ClassFunction _ ℂ)))
      (_ : φ ∈ dη.constituents),
      IrreducibleCharacter.LiesOver ((hyp.typeI.typeF.H).subgroupOf L) φ η := by
    intro η dη hη
    rw [IrreducibleCharacter.LiesOver, ClassFunction.restrictionMultiplicity_def]
    have hval : ClassFunction.inner (φ : ClassFunction ↥L ℂ)
        (ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L) (η : ClassFunction _ ℂ)) = 1 := by
      rw [dη.decomp, inner_sum_right,
        Finset.sum_eq_single_of_mem φ hη (fun φ' _ hne => by
          rw [irreducibleCharacter_inner, if_neg (Ne.symm hne)]),
        irreducibleCharacter_inner, if_pos rfl]
    have hrel : ClassFunction.inner (φ : ClassFunction ↥L ℂ)
        (ClassFunction.induce ((hyp.typeI.typeF.H).subgroupOf L) (η : ClassFunction _ ℂ))
        = ClassFunction.inner
          (ClassFunction.restrict ((hyp.typeI.typeF.H).subgroupOf L) (φ : ClassFunction ↥L ℂ))
          (η : ClassFunction _ ℂ) := by
      rw [OddOrder.RepresentationTheory.inner_conj_symm,
        ClassFunction.inner_induce_eq_inner_restrict,
        OddOrder.RepresentationTheory.inner_conj_symm, star_star]
    rw [← hrel, hval]; exact one_ne_zero
  obtain ⟨g, hg⟩ :=
    (restrictionConstituentsSingleOrbit_of_isIrreducible φ).exists_conj
      (key θ dχ hmem) (key θ' dχ' hmem')
  rw [← hg]
  exact (ClassFunction.induce_conjBy_eq (H := (hyp.typeI.typeF.H).subgroupOf L) g
    (θ : ClassFunction ↥((hyp.typeI.typeF.H).subgroupOf L) ℂ)).symm

open scoped Classical OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.4), pin (c)** ([Is] 6.2 capturing + (1.5.a)/(1.2)): the off-kernel irreducible
characters `{φ : H ⊄ ker φ}` partition into the constituent-sets `S(χ)`, `χ ∈ S`.  By [Is] 6.2,
`H ⊄ ker φ ⟹ Res_H φ` has a non-trivial constituent `θ`, so `φ ∈ S(Ind_H^L θ)`; the orbit `θ`
determines `χ = Ind θ` uniquely ((1.5.a)/(1.2)), so the `S(χ)` are pairwise disjoint and cover the
off-kernel irreducibles.  This is the genuine cross-section content (the [Is] 6.2 partition); the
`β`-vanishing regroup `Sset_offKernel_vanishes_off_H` is proved from it. -/
theorem exists_offKernel_constituent_partition {L : Subgroup G} [Finite G] (hyp : Hypothesis L)
    (data : ∀ χ ∈ hyp.Sset, CharacterDecompositionData hyp χ) :
    ∃ parts : Finset {χ : ClassFunction ↥L ℂ // χ ∈ hyp.Sset},
      Finset.univ.filter (fun φ => ¬ InHKernel hyp φ) =
        parts.biUnion (fun χ => (data χ.1 χ.2).constituents) ∧
      (↑parts : Set {χ : ClassFunction ↥L ℂ // χ ∈ hyp.Sset}).PairwiseDisjoint
        (fun χ => (data χ.1 χ.2).constituents) := by
  have := hyp.finiteG
  classical
  by_cases hne : (Finset.univ.filter (fun φ => ¬ InHKernel hyp φ)).Nonempty
  · -- The off-kernel filter is nonempty, so `S` is nonempty; build the capturing map.
    obtain ⟨φ0, hφ0⟩ := hne
    rw [Finset.mem_filter] at hφ0
    obtain ⟨χ0, hχ0, -⟩ := not_inHKernel_imp_mem_constituents hyp data hφ0.2
    have : Nonempty {χ : ClassFunction ↥L ℂ // χ ∈ hyp.Sset} := ⟨⟨χ0, hχ0⟩⟩
    let cap : IrreducibleCharacter ↥L → {χ : ClassFunction ↥L ℂ // χ ∈ hyp.Sset} :=
      fun φ => if h : ¬ InHKernel hyp φ then
        ⟨(not_inHKernel_imp_mem_constituents hyp data h).choose,
         (not_inHKernel_imp_mem_constituents hyp data h).choose_spec.choose⟩
      else Classical.arbitrary _
    refine ⟨(Finset.univ.filter (fun φ => ¬ InHKernel hyp φ)).image cap, ?_, ?_⟩
    · ext φ
      simp only [Finset.mem_filter, Finset.mem_univ, true_and, Finset.mem_biUnion,
        Finset.mem_image]
      constructor
      · intro hφ
        refine ⟨cap φ, ⟨φ, hφ, rfl⟩, ?_⟩
        have hcapφ : cap φ = ⟨(not_inHKernel_imp_mem_constituents hyp data hφ).choose,
            (not_inHKernel_imp_mem_constituents hyp data hφ).choose_spec.choose⟩ := dif_pos hφ
        rw [hcapφ]
        exact (not_inHKernel_imp_mem_constituents hyp data hφ).choose_spec.choose_spec
      · rintro ⟨χs, ⟨φ', _, rfl⟩, hmem⟩
        exact constituents_not_inHKernel hyp (cap φ').2 (data (cap φ').1 (cap φ').2) hmem
    · intro χs _ χs' _ hne_s
      simp only [Function.onFun, Finset.disjoint_left]
      intro φ hmem hmem'
      exact hne_s (Subtype.ext (constituents_eq_of_mem hyp χs.2 χs'.2
        (data χs.1 χs.2) (data χs'.1 χs'.2) hmem hmem'))
  · -- The off-kernel filter is empty; the empty partition works.
    rw [Finset.not_nonempty_iff_eq_empty] at hne
    exact ⟨∅, by rw [hne, Finset.biUnion_empty], by
      rw [Finset.coe_empty]; exact Set.pairwiseDisjoint_empty⟩

open scoped Classical OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.4), the off-kernel regroup** (genuine, from the [Is] 6.2 partition pin): the
off-kernel Fourier part `β = ∑_{φ : H ⊄ ker φ} ⟨Res_L ψ, φ⟩·φ` of `Res_L ψ` vanishes on `L − H`.
Regroup the off-kernel irreducibles by the partition into `S(χ)`
(`exists_offKernel_constituent_partition`); on each `S(χ)` the coefficient `⟨Res_L ψ, φ⟩` is
constant
(`Sset_coeff_equal`, from `ψ ⊥ R(χ)`), so the `S(χ)`-block is `c_χ·∑_{φ ∈ S(χ)} φ = c_χ·χ`
(`decomp`), which vanishes at `g ∈ L − H` (`Sset_vanishes_off_H`). -/
theorem Sset_offKernel_vanishes_off_H {L : Subgroup G} [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G) (hyp : Hypothesis L)
    (data : ∀ χ ∈ hyp.Sset, CharacterDecompositionData hyp χ) {psi : ClassFunction G ℂ}
    (horth : ∀ χ (hχ : χ ∈ hyp.Sset), ∀ α ∈ Rset (data χ hχ), ClassFunction.inner psi α = 0)
    {g : ↥L} (hg : (g : G) ∉ hyp.H) :
    (∑ φ ∈ Finset.univ.filter (fun φ => ¬ InHKernel hyp φ),
      ClassFunction.inner (ClassFunction.restrict L psi) (φ : ClassFunction ↥L ℂ) •
        (φ : ClassFunction ↥L ℂ)) g = 0 := by
  obtain ⟨parts, hpart, hdisj⟩ := exists_offKernel_constituent_partition hyp data
  rw [classFunction_sum_apply, hpart, Finset.sum_biUnion hdisj]
  refine Finset.sum_eq_zero fun χ _ => ?_
  -- The `S(χ)`-block: `∑_{φ ∈ S(χ)} ⟨Res_L ψ, φ⟩ · φ(g) = c_χ · χ(g) = 0`.
  obtain ⟨φ₀, hφ₀⟩ := (data χ.1 χ.2).constituents_nonempty
  have hblock : ∑ φ ∈ (data χ.1 χ.2).constituents,
      (ClassFunction.inner (ClassFunction.restrict L psi) (φ : ClassFunction ↥L ℂ) •
        (φ : ClassFunction ↥L ℂ)) g
      = ClassFunction.inner (ClassFunction.restrict L psi) (φ₀ : ClassFunction ↥L ℂ) *
        ∑ φ ∈ (data χ.1 χ.2).constituents, (φ : ClassFunction ↥L ℂ) g := by
    rw [Finset.mul_sum]
    refine Finset.sum_congr rfl fun φ hφ => ?_
    rw [ClassFunction.smul_apply,
      Sset_coeff_equal hG hyp (data χ.1 χ.2) (horth χ.1 χ.2) hφ hφ₀]
  have hdecomp : ∑ φ ∈ (data χ.1 χ.2).constituents, (φ : ClassFunction ↥L ℂ) g = χ.1 g := by
    rw [← classFunction_sum_apply, ← (data χ.1 χ.2).decomp]
  rw [hblock, hdecomp, Sset_vanishes_off_H hyp χ.2 hg, mul_zero]

open scoped Classical OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (12.4)**: a class function `ψ` orthogonal to every type-I family `R(χ)` (`χ ∈ S`)
is
constant on each coset `xH` with `x ∈ L − H`.

Proof: write `Res_L ψ = γ + β` by the Fourier expansion (`sum_inner_irreducibleCharacter_smul`),
splitting `Irr L` into `{H ⊆ ker φ}` (= `γ`) and `{H ⊄ ker φ}` (= `β`).  The kernel part `γ` is
constant on `H`-cosets (`apply_mul_eq_of_mem_characterKernel`, each `φ` with `H ⊆ ker φ`); the
off-kernel part `β` vanishes on `L − H` (`Sset_offKernel_vanishes_off_H`: by [Is] 6.2 + the
coefficient bridge `Sset_coeff_equal`, `β ∈ ℂ[S]` vanishes off `H`).  Hence
`ψ(xh) = γ(xh) + β(xh) = γ(x) + 0 = γ(x) + β(x) = ψ(x)`. -/
theorem orthogonal_character_constant_on_coset {L : Subgroup G} [Finite G]
    (hG : OddOrder.BG.IsMinimalSimpleOdd G) (hyp : Hypothesis L)
    (data : ∀ χ ∈ hyp.Sset, CharacterDecompositionData hyp χ) {psi : ClassFunction G ℂ}
    (horth : ∀ χ (hχ : χ ∈ hyp.Sset), ∀ α ∈ Rset (data χ hχ), ClassFunction.inner psi α = 0)
    {x : G} (hxL : x ∈ L) (hxH : x ∉ hyp.H) :
    ∀ h : G, h ∈ hyp.H → psi (x * h) = psi x := by
  have := hyp.finiteG
  classical
  intro h hh
  have hhL : h ∈ L := hyp.typeI.typeF.H_le hh
  set xL : ↥L := ⟨x, hxL⟩ with hxLdef
  set hL : ↥L := ⟨h, hhL⟩ with hhLdef
  set gf : ClassFunction ↥L ℂ := ClassFunction.restrict L psi with hgf
  -- Fourier split of `Res_L ψ = γ + β`.
  set γ : ClassFunction ↥L ℂ := ∑ φ ∈ Finset.univ.filter (fun φ => InHKernel hyp φ),
    ClassFunction.inner gf (φ : ClassFunction ↥L ℂ) • (φ : ClassFunction ↥L ℂ) with hγ
  set β : ClassFunction ↥L ℂ := ∑ φ ∈ Finset.univ.filter (fun φ => ¬ InHKernel hyp φ),
    ClassFunction.inner gf (φ : ClassFunction ↥L ℂ) • (φ : ClassFunction ↥L ℂ) with hβ
  have hsplit : gf = γ + β := by
    rw [hγ, hβ, Finset.sum_filter_add_sum_filter_not]
    exact (sum_inner_irreducibleCharacter_smul gf).symm
  -- `hL` lies in `H` (as a subgroup of `L`).
  have hLmem : hL ∈ ((hyp.typeI.typeF.H).subgroupOf L : Set ↥L) :=
    Subgroup.mem_subgroupOf.mpr hh
  -- `γ` is constant on the `H`-coset `xL · hL`.
  have hγconst : γ (xL * hL) = γ xL := by
    rw [hγ, classFunction_sum_apply, classFunction_sum_apply]
    refine Finset.sum_congr rfl fun φ hφ => ?_
    rw [ClassFunction.smul_apply, ClassFunction.smul_apply,
      apply_mul_eq_of_mem_characterKernel φ.isIrreducible
        ((Finset.mem_filter.mp hφ).2 hLmem) xL]
  -- `β` vanishes on `L − H` (off-kernel part is in `ℂ[S]`).
  have hxhH : x * h ∉ hyp.H := fun hcon => hxH (by
    have hmem : x * h * h⁻¹ ∈ hyp.H := mul_mem hcon (inv_mem hh)
    rwa [mul_inv_cancel_right] at hmem)
  have hβxh : β (xL * hL) = 0 := by
    rw [hβ]
    exact Sset_offKernel_vanishes_off_H hG hyp data horth (g := xL * hL)
      (by rw [Subgroup.coe_mul]; exact hxhH)
  have hβx : β xL = 0 := by
    rw [hβ]; exact Sset_offKernel_vanishes_off_H hG hyp data horth (g := xL) hxH
  -- Assemble: `ψ(xh) = γ(xh) + β(xh) = γ(x) + 0 = γ(x) + β(x) = ψ(x)`.
  have key : gf (xL * hL) = gf xL := by
    simp only [hsplit, ClassFunction.add_apply, hγconst, hβxh, hβx, add_zero]
  have hgxh : gf (xL * hL) = psi (x * h) := by
    rw [hgf, ClassFunction.restrict_apply, Subgroup.coe_mul]
  have hgx : gf xL = psi x := by rw [hgf, ClassFunction.restrict_apply]
  rw [← hgxh, ← hgx]; exact key

/-- **Commutator bridge for the (12.5) core.**  For `H ≤ L` (subgroups of `G`), an element `x` of
`↥(H.subgroupOf L)` lies in the derived subgroup `[G_core, G_core]` of `G_core := ↥(H.subgroupOf L)`
iff its underlying `G`-element lies in `derivedInG H = [H, H]`.  Via the `MulEquiv`
`subgroupOfEquivOfLe : ↥(H.subgroupOf L) ≃* ↥H` (which preserves the commutator subgroup) and
`(derivedInG H).subgroupOf H = commutator ↥H`.  Lets the generic `DpsiH` core (whose `H_core` is
`commutator G_core`) translate its `x ∉ H_core` conclusion back to `h ∉ Hprime`. -/
theorem mem_commutator_subgroupOf_iff {L H : Subgroup G} (hHL : H ≤ L)
    (x : ↥(H.subgroupOf L)) :
    x ∈ commutator ↥(H.subgroupOf L) ↔ ((x : ↥L) : G) ∈ derivedInG H := by
  have hcomm_H : (derivedInG H).subgroupOf H = commutator ↥H := by
    rw [derivedInG, Subgroup.subgroupOf,
      Subgroup.comap_map_eq_self_of_injective H.subtype_injective]
  set e := Subgroup.subgroupOfEquivOfLe hHL with he
  have hmap : commutator ↥H = (commutator ↥(H.subgroupOf L)).map e.toMonoidHom := by
    rw [commutator, commutator, Subgroup.map_commutator,
      Subgroup.map_top_of_surjective _ e.surjective]
  have hcoe : ((e x : ↥H) : G) = ((x : ↥L) : G) := rfl
  have hstep1 : x ∈ commutator ↥(H.subgroupOf L) ↔ e x ∈ commutator ↥H := by
    rw [hmap]
    exact (Subgroup.mem_map_iff_mem e.injective).symm
  have hstep2 : e x ∈ commutator ↥H ↔ ((e x : ↥H) : G) ∈ derivedInG H := by
    rw [← hcomm_H, Subgroup.mem_subgroupOf]
  rw [hstep1, hstep2, hcoe]

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/RhoConstancy.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