Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Isometry on orthonormal difference pairs

Definition
OddOrder_GroupTheory_RepresentationTheory_IsometryDifferencePair

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

feit-thompsonfinite-groupsgroup-theoryishida-odd-order

A structural lemma underlying Peterfalvi §3 (1.4), §5 (3.2), §6 (4.5) and §7 (5.6). Given n≥2n\ge2n≥2 distinct irreducible characters χ0,…,χn−1\chi_0,\dots,\chi_{n-1}χ0​,…,χn−1​ of a finite group HHH of the same degree, and a Z\mathbb{Z}Z-linear map τ\tauτ to class functions of GGG that is isometric on the differences χi−χ0\chi_i-\chi_0χi​−χ0​, there exist distinct irreducible characters μi\mu_iμi​ of GGG and a uniform sign ε\varepsilonε with τ(χi−χ0)=ε(μi−μ0)\tau(\chi_i-\chi_0)=\varepsilon(\mu_i-\mu_0)τ(χi​−χ0​)=ε(μi​−μ0​).

Main definitions: irreducibleCharacterDifference (the source difference χi−χ0\chi_i-\chi_0χi​−χ0​), isometryDifferenceImage (its image under an integral isometry), the hypotheses IsometryDifferenceImagesVanishAtOne and IsometryDifferenceImagesAreVirtual, IsometryDifferencePairNumerics (numeric data of the image differences) and SignedIrreducibleDifferenceFamily (a signed tuple of irreducibles of GGG as target shape).

The bundle also proves 25 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/GroupTheory/RepresentationTheory/IsometryDifferencePair.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_GroupTheory_RepresentationTheory_IrrIndexing
import Definitions.Def_OddOrder_GroupTheory_RepresentationTheory_RowOrthogonality
import Definitions.Def_OddOrder_GroupTheory_RepresentationTheory_SecondOrthogonality
import Definitions.Def_OddOrder_GroupTheory_RepresentationTheory_ZIrrFourier
import Mathlib.Data.Complex.Basic

namespace OddOrder.RepresentationTheory

variable {G H : Type*} [Group G] [Group H]

theorem irreducibleCharacter_inner_eq_if
    [Fintype G] [Invertible (Nat.card G : ℂ)]
    {n : ℕ} (χ : Fin n → IrreducibleCharacter G) (hχ : Function.Injective χ)
    (i j : Fin n) :
    ClassFunction.inner (χ i : ClassFunction G ℂ) (χ j : ClassFunction G ℂ) =
      if i = j then 1 else 0 := by
  by_cases hij : i = j
  · subst j
    simpa [characterTableRowPairing] using
      CharacterTableRowOrthogonality.diagonal (G := G) characterTableRowOrthogonality (χ i)
  · have hμ : χ i ≠ χ j := fun h => hij (hχ h)
    rw [if_neg hij]
    simpa [characterTableRowPairing] using
      CharacterTableRowOrthogonality.offDiagonal (G := G) characterTableRowOrthogonality hμ

theorem irreducibleCharacter_difference_inner_self_of_ne_zero
    [Fintype G] [Invertible (Nat.card G : ℂ)]
    {n : ℕ} [NeZero n] (χ : Fin n → IrreducibleCharacter G)
    (hχ : Function.Injective χ) {i : Fin n} (hi : i ≠ 0) :
    ClassFunction.inner
        ((χ i : ClassFunction G ℂ) - (χ 0 : ClassFunction G ℂ))
        ((χ i : ClassFunction G ℂ) - (χ 0 : ClassFunction G ℂ)) = 2 := by
  rw [ClassFunction.inner_sub_left, ClassFunction.inner_sub_right,
    ClassFunction.inner_sub_right]
  rw [irreducibleCharacter_inner_eq_if (G := G) χ hχ i i]
  rw [irreducibleCharacter_inner_eq_if (G := G) χ hχ i 0]
  rw [irreducibleCharacter_inner_eq_if (G := G) χ hχ 0 i]
  rw [irreducibleCharacter_inner_eq_if (G := G) χ hχ 0 0]
  simp [hi, eq_comm]
  norm_num

theorem irreducibleCharacter_difference_inner_of_ne_zero_of_ne
    [Fintype G] [Invertible (Nat.card G : ℂ)]
    {n : ℕ} [NeZero n] (χ : Fin n → IrreducibleCharacter G)
    (hχ : Function.Injective χ) {i j : Fin n}
    (hi : i ≠ 0) (hj : j ≠ 0) (hij : i ≠ j) :
    ClassFunction.inner
        ((χ i : ClassFunction G ℂ) - (χ 0 : ClassFunction G ℂ))
        ((χ j : ClassFunction G ℂ) - (χ 0 : ClassFunction G ℂ)) = 1 := by
  rw [ClassFunction.inner_sub_left, ClassFunction.inner_sub_right,
    ClassFunction.inner_sub_right]
  rw [irreducibleCharacter_inner_eq_if (G := G) χ hχ i j]
  rw [irreducibleCharacter_inner_eq_if (G := G) χ hχ i 0]
  rw [irreducibleCharacter_inner_eq_if (G := G) χ hχ 0 j]
  rw [irreducibleCharacter_inner_eq_if (G := G) χ hχ 0 0]
  simp [hi, hj, hij, eq_comm]

/-- The source-side difference `χ_i - χ_0` used in Peterfalvi §3 (1.4). -/
abbrev irreducibleCharacterDifference {n : ℕ} [NeZero n]
    (χ : Fin n → IrreducibleCharacter G) (i : Fin n) : ClassFunction G ℂ :=
  (χ i : ClassFunction G ℂ) - (χ 0 : ClassFunction G ℂ)

@[simp] theorem irreducibleCharacterDifference_zero
    {n : ℕ} [NeZero n] (χ : Fin n → IrreducibleCharacter G) :
    irreducibleCharacterDifference χ 0 = 0 := by
  simp [irreducibleCharacterDifference]

@[simp] theorem irreducibleCharacterDifference_apply_one
    {n : ℕ} [NeZero n] (χ : Fin n → IrreducibleCharacter G) (i : Fin n) :
    irreducibleCharacterDifference χ i (1 : G) =
      ((χ i : ClassFunction G ℂ) : G → ℂ) 1 -
        ((χ 0 : ClassFunction G ℂ) : G → ℂ) 1 :=
  rfl

theorem irreducibleCharacterDifference_inner_self_of_ne_zero
    [Fintype G] [Invertible (Nat.card G : ℂ)]
    {n : ℕ} [NeZero n] (χ : Fin n → IrreducibleCharacter G)
    (hχ : Function.Injective χ) {i : Fin n} (hi : i ≠ 0) :
    ClassFunction.inner (irreducibleCharacterDifference χ i)
        (irreducibleCharacterDifference χ i) = 2 := by
  exact irreducibleCharacter_difference_inner_self_of_ne_zero (G := G) χ hχ hi

theorem irreducibleCharacterDifference_inner_of_ne_zero_of_ne
    [Fintype G] [Invertible (Nat.card G : ℂ)]
    {n : ℕ} [NeZero n] (χ : Fin n → IrreducibleCharacter G)
    (hχ : Function.Injective χ) {i j : Fin n}
    (hi : i ≠ 0) (hj : j ≠ 0) (hij : i ≠ j) :
    ClassFunction.inner (irreducibleCharacterDifference χ i)
        (irreducibleCharacterDifference χ j) = 1 := by
  exact irreducibleCharacter_difference_inner_of_ne_zero_of_ne
    (G := G) χ hχ hi hj hij

/-- The image under an integral isometry of a source-side difference `χ_i - χ_0`. -/
abbrev isometryDifferenceImage {G H : Type*} [Group G] [Group H]
    {n : ℕ} [NeZero n]
    (τ : ClassFunction H ℂ →ₗ[ℤ] ClassFunction G ℂ)
    (χ : Fin n → IrreducibleCharacter H) (i : Fin n) : ClassFunction G ℂ :=
  τ (irreducibleCharacterDifference χ i)

@[simp] theorem isometryDifferenceImage_zero {G H : Type*} [Group G] [Group H]
    {n : ℕ} [NeZero n]
    (τ : ClassFunction H ℂ →ₗ[ℤ] ClassFunction G ℂ)
    (χ : Fin n → IrreducibleCharacter H) :
    isometryDifferenceImage τ χ 0 = 0 := by
  simp [isometryDifferenceImage]

/-- The degree-zero hypothesis for the images in Peterfalvi §3 (1.4).

In the textbook this comes from the map landing in the reduced virtual-character
lattice on `G#`; here it is kept as an explicit class-function condition until
that lattice target is part of the theorem statement. -/
def IsometryDifferenceImagesVanishAtOne {G H : Type*} [Group G] [Group H]
    {n : ℕ} [NeZero n]
    (τ : ClassFunction H ℂ →ₗ[ℤ] ClassFunction G ℂ)
    (χ : Fin n → IrreducibleCharacter H) : Prop :=
  ∀ i, isometryDifferenceImage τ χ i (1 : G) = 0

/-- The virtual-character hypothesis for the images in Peterfalvi §3 (1.4).

Peterfalvi's `τ : ℤ[X, H#] → ℤ[Irr G]` carries virtual characters to virtual
characters; the integer-linear formulation `τ : ClassFunction H ℂ →ₗ[ℤ]
ClassFunction G ℂ` is broader, so the virtual-character target needs to be
recorded as a separate hypothesis on the inputs we feed into the structure
theorem. -/
def IsometryDifferenceImagesAreVirtual {G H : Type*} [Group G] [Group H]
    {n : ℕ} [NeZero n]
    (τ : ClassFunction H ℂ →ₗ[ℤ] ClassFunction G ℂ)
    (χ : Fin n → IrreducibleCharacter H) : Prop :=
  ∀ i, isometryDifferenceImage τ χ i ∈ ZIrr G

theorem isometryDifferenceImage_inner_self_of_ne_zero
    [Fintype G] [Fintype H]
    [Invertible (Nat.card G : ℂ)] [Invertible (Nat.card H : ℂ)]
    {n : ℕ} [NeZero n] (χ : Fin n → IrreducibleCharacter H)
    (hχ : Function.Injective χ)
    (τ : ClassFunction H ℂ →ₗ[ℤ] ClassFunction G ℂ)
    (h_isom : ∀ i j,
        ClassFunction.inner (isometryDifferenceImage τ χ i)
          (isometryDifferenceImage τ χ j) =
        ClassFunction.inner (irreducibleCharacterDifference χ i)
          (irreducibleCharacterDifference χ j))
    {i : Fin n} (hi : i ≠ 0) :
    ClassFunction.inner (isometryDifferenceImage τ χ i)
        (isometryDifferenceImage τ χ i) = 2 := by
  rw [h_isom i i]
  exact irreducibleCharacterDifference_inner_self_of_ne_zero
    (G := H) χ hχ hi

theorem isometryDifferenceImage_inner_of_ne_zero_of_ne
    [Fintype G] [Fintype H]
    [Invertible (Nat.card G : ℂ)] [Invertible (Nat.card H : ℂ)]
    {n : ℕ} [NeZero n] (χ : Fin n → IrreducibleCharacter H)
    (hχ : Function.Injective χ)
    (τ : ClassFunction H ℂ →ₗ[ℤ] ClassFunction G ℂ)
    (h_isom : ∀ i j,
        ClassFunction.inner (isometryDifferenceImage τ χ i)
          (isometryDifferenceImage τ χ j) =
        ClassFunction.inner (irreducibleCharacterDifference χ i)
          (irreducibleCharacterDifference χ j))
    {i j : Fin n} (hi : i ≠ 0) (hj : j ≠ 0) (hij : i ≠ j) :
    ClassFunction.inner (isometryDifferenceImage τ χ i)
        (isometryDifferenceImage τ χ j) = 1 := by
  rw [h_isom i j]
  exact irreducibleCharacterDifference_inner_of_ne_zero_of_ne
    (G := H) χ hχ hi hj hij

/-- The numeric data carried by the image differences in Peterfalvi §3 (1.4).

This is the interface between character theory and the finite combinatorial
argument: the zero row, degree-zero condition, norm `2`, and mutual inner
product `1` for distinct nonzero differences. -/
structure IsometryDifferencePairNumerics {n : ℕ} [NeZero n] [Fintype G]
    [Invertible (Nat.card G : ℂ)]
    (v : Fin n → ClassFunction G ℂ) : Prop where
  zero : v 0 = 0
  degree_zero : ∀ i, v i (1 : G) = 0
  inner_self_of_ne_zero :
    ∀ ⦃i : Fin n⦄, i ≠ 0 → ClassFunction.inner (v i) (v i) = 2
  inner_of_ne_zero_of_ne :
    ∀ ⦃i j : Fin n⦄, i ≠ 0 → j ≠ 0 → i ≠ j →
      ClassFunction.inner (v i) (v j) = 1

namespace IsometryDifferencePairNumerics

variable {n : ℕ} [NeZero n] [Fintype G] [Invertible (Nat.card G : ℂ)]
    {v : Fin n → ClassFunction G ℂ}

theorem inner_self (hv : IsometryDifferencePairNumerics v)
    {i : Fin n} (hi : i ≠ 0) :
    ClassFunction.inner (v i) (v i) = 2 :=
  hv.inner_self_of_ne_zero hi

theorem inner_of_ne (hv : IsometryDifferencePairNumerics v)
    {i j : Fin n} (hi : i ≠ 0) (hj : j ≠ 0) (hij : i ≠ j) :
    ClassFunction.inner (v i) (v j) = 1 :=
  hv.inner_of_ne_zero_of_ne hi hj hij

end IsometryDifferencePairNumerics

/-- The image differences of an isometric family satisfy the finite numeric
conditions needed by the later combinatorial proof core. -/
theorem isometryDifferencePairNumerics_of_isometryDifferenceImage
    [Fintype G] [Fintype H]
    [Invertible (Nat.card G : ℂ)] [Invertible (Nat.card H : ℂ)]
    {n : ℕ} [NeZero n] (χ : Fin n → IrreducibleCharacter H)
    (hχ : Function.Injective χ)
    (τ : ClassFunction H ℂ →ₗ[ℤ] ClassFunction G ℂ)
    (h_image_degree_zero : IsometryDifferenceImagesVanishAtOne τ χ)
    (h_isom : ∀ i j,
        ClassFunction.inner (isometryDifferenceImage τ χ i)
          (isometryDifferenceImage τ χ j) =
        ClassFunction.inner (irreducibleCharacterDifference χ i)
          (irreducibleCharacterDifference χ j)) :
    IsometryDifferencePairNumerics
      (fun i => isometryDifferenceImage τ χ i) where
  zero := by
    simp
  degree_zero := h_image_degree_zero
  inner_self_of_ne_zero := by
    intro i hi
    exact isometryDifferenceImage_inner_self_of_ne_zero
      (G := G) (H := H) χ hχ τ h_isom hi
  inner_of_ne_zero_of_ne := by
    intro i j hi hj hij
    exact isometryDifferenceImage_inner_of_ne_zero_of_ne
      (G := G) (H := H) χ hχ τ h_isom hi hj hij

/-- A signed `n`-tuple of irreducible characters of `G`, used as the target
shape in Peterfalvi's difference-pair lemmas.

The theorem `isometry_difference_pair_structure` says that the image of
`χ_i - χ_0` is obtained from such a family as
`ε • (μ_i - μ_0)`. -/
structure SignedIrreducibleDifferenceFamily (G : Type*) [Group G] (n : ℕ) where
  mu : Fin n → IrreducibleCharacter G
  sign : ℤ
  sign_eq : sign = 1 ∨ sign = -1
  injective : Function.Injective mu

namespace SignedIrreducibleDifferenceFamily

variable {n : ℕ}

/-- The underlying class-function family. -/
abbrev classFunction (data : SignedIrreducibleDifferenceFamily G n)
    (i : Fin n) : ClassFunction G ℂ :=
  data.mu i

@[simp] theorem classFunction_apply
    (data : SignedIrreducibleDifferenceFamily G n) (i : Fin n) :
    data.classFunction i = (data.mu i : ClassFunction G ℂ) :=
  rfl

/-- The class-function family underlying a signed irreducible-difference family
is injective. -/
theorem classFunction_injective (data : SignedIrreducibleDifferenceFamily G n) :
    Function.Injective data.classFunction := by
  intro i j hij
  exact data.injective (IrreducibleCharacter.ext (by simpa [classFunction] using hij))

/-- Every member of a signed irreducible-difference family is irreducible. -/
@[simp] theorem classFunction_irreducible
    (data : SignedIrreducibleDifferenceFamily G n) (i : Fin n) :
    IsIrreducibleCharacter (data.classFunction i) :=
  (data.mu i).isIrreducible

/-- The difference `μ_i - μ_0` attached to a signed irreducible-difference
family. -/
abbrev difference (data : SignedIrreducibleDifferenceFamily G n)
    [NeZero n] (i : Fin n) : ClassFunction G ℂ :=
  data.classFunction i - data.classFunction 0

@[simp] theorem difference_apply
    (data : SignedIrreducibleDifferenceFamily G n) [NeZero n] (i : Fin n) :
    data.difference i = data.classFunction i - data.classFunction 0 :=
  rfl

@[simp] theorem difference_zero
    (data : SignedIrreducibleDifferenceFamily G n) [NeZero n] :
    data.difference 0 = 0 := by
  simp [difference]

/-- The signed target difference `ε • (μ_i - μ_0)`. -/
abbrev signedDifference (data : SignedIrreducibleDifferenceFamily G n)
    [NeZero n] (i : Fin n) : ClassFunction G ℂ :=
  data.sign • data.difference i

@[simp] theorem signedDifference_apply
    (data : SignedIrreducibleDifferenceFamily G n) [NeZero n] (i : Fin n) :
    data.signedDifference i = data.sign • data.difference i :=
  rfl

@[simp] theorem difference_apply_one
    (data : SignedIrreducibleDifferenceFamily G n) [NeZero n] (i : Fin n) :
    data.difference i (1 : G) =
      data.classFunction i (1 : G) - data.classFunction 0 (1 : G) :=
  rfl

@[simp] theorem signedDifference_zero
    (data : SignedIrreducibleDifferenceFamily G n) [NeZero n] :
    data.signedDifference 0 = 0 := by
  simp [signedDifference]

theorem sign_mul_self (data : SignedIrreducibleDifferenceFamily G n) :
    data.sign * data.sign = 1 := by
  rcases data.sign_eq with hsign | hsign <;> simp [hsign]

theorem classFunction_inner_eq_if
    [Fintype G] [Invertible (Nat.card G : ℂ)]
    (data : SignedIrreducibleDifferenceFamily G n) (i j : Fin n) :
    ClassFunction.inner (data.classFunction i) (data.classFunction j) =
      if i = j then 1 else 0 := by
  by_cases hij : i = j
  · subst j
    simpa [classFunction, characterTableRowPairing] using
      CharacterTableRowOrthogonality.diagonal (G := G) characterTableRowOrthogonality (data.mu i)
  · have hμ : data.mu i ≠ data.mu j := fun h => hij (data.injective h)
    rw [if_neg hij]
    simpa [classFunction, characterTableRowPairing] using
      CharacterTableRowOrthogonality.offDiagonal (G := G) characterTableRowOrthogonality hμ

end SignedIrreducibleDifferenceFamily

variable [Fintype G] [Fintype H]

open scoped Classical in
/-- Expanding `⟨α - β, γ - δ⟩` for irreducible characters: each summand is a Kronecker
delta by `irreducibleCharacter_inner_eq_ite`. -/
theorem irreducibleCharacter_inner_sub_sub_eq_ite
    [Invertible (Nat.card G : ℂ)]
    (α β γ δ : IrreducibleCharacter G) :
    ClassFunction.inner
        ((α : ClassFunction G ℂ) - (β : ClassFunction G ℂ))
        ((γ : ClassFunction G ℂ) - (δ : ClassFunction G ℂ)) =
      (if α = γ then (1 : ℂ) else 0) - (if α = δ then (1 : ℂ) else 0)
        - (if β = γ then (1 : ℂ) else 0) + (if β = δ then (1 : ℂ) else 0) := by
  rw [ClassFunction.inner_sub_left, ClassFunction.inner_sub_right,
      ClassFunction.inner_sub_right,
      irreducibleCharacter_inner_eq_ite α γ, irreducibleCharacter_inner_eq_ite α δ,
      irreducibleCharacter_inner_eq_ite β γ, irreducibleCharacter_inner_eq_ite β δ]
  ring

/-- **Cross-pair classification (Type A / Type B).**
If `⟨α - β, γ - δ⟩ = 1` for distinct irreducible characters `α ≠ β`, `γ ≠ δ`, then
exactly one of the following holds:

* (Type A) `α = γ`, with `α ≠ δ`, `β ≠ γ`, `β ≠ δ` — the "α positions agree";
* (Type B) `β = δ`, with `α ≠ γ`, `α ≠ δ`, `β ≠ γ` — the "β positions agree".

Both branches imply `α ≠ δ ∧ β ≠ γ`.

Proof: expanding the inner product into four Kronecker deltas and ruling out the
combinations that violate `α ≠ β` or `γ ≠ δ` leaves only sum patterns whose value is
`1` exactly when one of the two cases above holds.  This is the integer-equality core
of Peterfalvi §3 (1.4) uniform-sign argument. -/
theorem irreducibleCharacter_cross_pair_classify
    [Invertible (Nat.card G : ℂ)]
    {α β γ δ : IrreducibleCharacter G}
    (hαβ : α ≠ β) (hγδ : γ ≠ δ)
    (h_inner : ClassFunction.inner
        ((α : ClassFunction G ℂ) - (β : ClassFunction G ℂ))
        ((γ : ClassFunction G ℂ) - (δ : ClassFunction G ℂ)) = 1) :
    (α = γ ∧ α ≠ δ ∧ β ≠ γ ∧ β ≠ δ) ∨
      (α ≠ γ ∧ α ≠ δ ∧ β ≠ γ ∧ β = δ) := by
  classical
  rw [irreducibleCharacter_inner_sub_sub_eq_ite] at h_inner
  by_cases hαγ : α = γ
  · have hαδ : α ≠ δ := fun h => hγδ (hαγ.symm.trans h)
    have hβγ : β ≠ γ := fun h => hαβ (hαγ.trans h.symm)
    by_cases hβδ : β = δ
    · exfalso
      rw [if_pos hαγ, if_neg hαδ, if_neg hβγ, if_pos hβδ] at h_inner
      norm_num at h_inner
    · exact Or.inl ⟨hαγ, hαδ, hβγ, hβδ⟩
  · by_cases hβδ : β = δ
    · have hαδ : α ≠ δ := fun h => hαβ (h.trans hβδ.symm)
      have hβγ : β ≠ γ := fun h => hγδ (h.symm.trans hβδ)
      exact Or.inr ⟨hαγ, hαδ, hβγ, hβδ⟩
    · exfalso
      rw [if_neg hαγ, if_neg hβδ] at h_inner
      by_cases hαδ : α = δ
      · by_cases hβγ : β = γ
        · rw [if_pos hαδ, if_pos hβγ] at h_inner; norm_num at h_inner
        · rw [if_pos hαδ, if_neg hβγ] at h_inner; norm_num at h_inner
      · by_cases hβγ : β = γ
        · rw [if_neg hαδ, if_pos hβγ] at h_inner; norm_num at h_inner
        · rw [if_neg hαδ, if_neg hβγ] at h_inner; norm_num at h_inner

/-- **Orthonormal difference-pair structure under isometry**
([Peterfalvi §3 (1.4)] abstracted, also used in §5 (3.2), §6 (4.5), §7 (5.6)).

Given `n ≥ 2` distinct irreducible characters `χ : Fin n → Irr(H)`, all of
equal degree, and a `ℤ`-linear `τ : ClassFunction H ℂ → ClassFunction G ℂ` such
that, on each difference `χ_i - χ_0`, the image lies in `ZIrr G`, vanishes at
`1 : G`, and preserves the normalized inner product, the conclusion is the
existence of a `SignedIrreducibleDifferenceFamily G n` whose signed differences
coincide with the images.  Equivalently: the image is built from a new
orthonormal `n`-tuple `μ : Fin n → Irr(G)` times a uniform sign `ε = ±1`.

The three hypotheses on `τ` correspond to Peterfalvi's textbook setup:

* `h_image_virtual` — Peterfalvi's `τ : ℤ[X, H#] → ℤ[Irr G]` is virtual-character
  preserving; required to identify the image in `ZIrr G`.
* `h_image_degree_zero` — character differences vanish at `1` on the source,
  and the isometry preserves this.
* `h_isom` — the normalized inner product is preserved on the `χ_i - χ_0`.

The proof goes through `exists_irr_sub_irr_of_inner_self_two`
(`ZIrrFourier`): each image `τ (χ_i - χ_0) ∈ ZIrr G` of squared norm `2` vanishing at
`1` is a difference `α_i - β_i` of two named irreducibles.  The uniform sign emerges
from `irreducibleCharacter_cross_pair_classify` applied to a fixed index `i = 1`:
every distinct nonzero index falls into "Type A" (sharing the `α` slot with `1`) or
"Type B" (sharing the `β` slot with `1`), and a triangle-style argument across three
indices forces a single global type.  In Type A the family is built from the shared
`α_1` with sign `-1`; in Type B from the shared `β_1` with sign `+1`.
See `issues/0025-peterfalvi-isometry-difference-core.md` for the recipe.
Downstream consumers in §3 (1.4) / §5 (3.2) / §6 (4.5) / §7 (5.6) apply this
theorem by supplying the three hypotheses from their own contexts. -/
theorem isometry_difference_pair_structure
    [Invertible (Nat.card G : ℂ)] [Invertible (Nat.card H : ℂ)]
    {n : ℕ} [NeZero n] (_hn : 2 ≤ n)
    (χ : Fin n → IrreducibleCharacter H)
    (_h_distinct : Function.Injective χ)
    (_h_same_degree :
      ∀ i, ((χ i : ClassFunction H ℂ) : H → ℂ) 1 =
        ((χ 0 : ClassFunction H ℂ) : H → ℂ) 1)
    (τ : ClassFunction H ℂ →ₗ[ℤ] ClassFunction G ℂ)
    (h_image_virtual : IsometryDifferenceImagesAreVirtual τ χ)
    (h_image_degree_zero : IsometryDifferenceImagesVanishAtOne τ χ)
    (h_isom : ∀ i j,
        ClassFunction.inner (isometryDifferenceImage τ χ i)
          (isometryDifferenceImage τ χ j) =
        ClassFunction.inner (irreducibleCharacterDifference χ i)
          (irreducibleCharacterDifference χ j)) :
    ∃ data : SignedIrreducibleDifferenceFamily G n,
      ∀ i, isometryDifferenceImage τ χ i = data.signedDifference i := by
  classical
  -- Numerics: norm 2, distinct mutual inner product 1, zero at index 0, vanish at `1`
  have hnum : IsometryDifferencePairNumerics
      (fun i => isometryDifferenceImage τ χ i) :=
    isometryDifferencePairNumerics_of_isometryDifferenceImage χ _h_distinct τ
      h_image_degree_zero h_isom
  -- Bridge: each image `τ (χ_i - χ_0)` (i ≠ 0) is a difference of two named irreducibles
  have key : ∀ i : Fin n, i ≠ 0 →
      ∃ α β : IrreducibleCharacter G, α ≠ β ∧
        isometryDifferenceImage τ χ i =
          (α : ClassFunction G ℂ) - (β : ClassFunction G ℂ) := fun i hi =>
    exists_irr_sub_irr_of_inner_self_two
      (h_image_virtual i) (hnum.inner_self hi) (hnum.degree_zero i)
  -- Choice functions α_i, β_i for each i (default trivial for i = 0)
  let αFun : Fin n → IrreducibleCharacter G := fun i =>
    if h : i = 0 then trivialIrreducibleCharacter G
    else (key i h).choose
  let βFun : Fin n → IrreducibleCharacter G := fun i =>
    if h : i = 0 then trivialIrreducibleCharacter G
    else (key i h).choose_spec.choose
  have hα_def : ∀ i (hi : i ≠ 0), αFun i = (key i hi).choose := fun i hi => dif_neg hi
  have hβ_def : ∀ i (hi : i ≠ 0), βFun i = (key i hi).choose_spec.choose :=
    fun i hi => dif_neg hi
  have hαβFun : ∀ i (hi : i ≠ 0), αFun i ≠ βFun i := by
    intro i hi
    rw [hα_def i hi, hβ_def i hi]
    exact (key i hi).choose_spec.choose_spec.1
  have hvFun : ∀ i (hi : i ≠ 0),
      isometryDifferenceImage τ χ i =
        (αFun i : ClassFunction G ℂ) - (βFun i : ClassFunction G ℂ) := by
    intro i hi
    rw [hα_def i hi, hβ_def i hi]
    exact (key i hi).choose_spec.choose_spec.2
  -- Reference index `one : Fin n`
  let one : Fin n := ⟨1, by omega⟩
  have hone_val : (one : Fin n).val = 1 := rfl
  have hone_ne_zero : one ≠ 0 := by
    intro h
    have h1 : (1 : ℕ) = 0 := by
      have := congrArg Fin.val h
      rw [hone_val, Fin.val_zero] at this
      exact this
    exact absurd h1 (by omega)
  -- Cross inner product = 1 for distinct nonzero indices
  have hcross : ∀ i j : Fin n, i ≠ 0 → j ≠ 0 → i ≠ j →
      ClassFunction.inner
        ((αFun i : ClassFunction G ℂ) - (βFun i : ClassFunction G ℂ))
        ((αFun j : ClassFunction G ℂ) - (βFun j : ClassFunction G ℂ)) = 1 := by
    intro i j hi hj hij
    rw [← hvFun i hi, ← hvFun j hj]
    exact hnum.inner_of_ne hi hj hij
  -- Case split on global Type A (∃ k ≠ 0, k ≠ one, αFun one = αFun k) vs Type B
  by_cases hExA : ∃ k : Fin n, k ≠ 0 ∧ k ≠ one ∧ αFun one = αFun k
  · -- Type A globally: sign = -1, μ_0 := αFun one, μ_i := βFun i (i ≠ 0)
    obtain ⟨k₀, hk₀, hk₀ne, hAk₀⟩ := hExA
    -- (one, k₀) is Type A (since αFun one = αFun k₀)
    have h_oneK₀ :
        αFun one ≠ βFun k₀ ∧ βFun one ≠ αFun k₀ ∧ βFun one ≠ βFun k₀ := by
      have hcr := hcross one k₀ hone_ne_zero hk₀ hk₀ne.symm
      rcases irreducibleCharacter_cross_pair_classify
          (hαβFun one hone_ne_zero) (hαβFun k₀ hk₀) hcr with
        ⟨_, h1, h2, h3⟩ | ⟨hne, _, _, _⟩
      · exact ⟨h1, h2, h3⟩
      · exact absurd hAk₀ hne
    obtain ⟨hαβ_one_k₀, hβα_one_k₀, hβ_ne_k₀⟩ := h_oneK₀
    -- Global Type A: for all k ≠ 0, αFun k = αFun one
    have hαAll : ∀ k : Fin n, k ≠ 0 → αFun k = αFun one := by
      intro k hk
      by_cases hkOne : k = one
      · subst hkOne; rfl
      by_cases hkk₀ : k = k₀
      · subst hkk₀; exact hAk₀.symm
      have hk₀k : k₀ ≠ k := fun h => hkk₀ h.symm
      have hcr_onek := hcross one k hone_ne_zero hk (Ne.symm hkOne)
      rcases irreducibleCharacter_cross_pair_classify
          (hαβFun one hone_ne_zero) (hαβFun k hk) hcr_onek with
        ⟨hα_eq, _, _, _⟩ | ⟨hα_ne, _, _, hβ_eq⟩
      · exact hα_eq.symm
      -- Type B at (one, k): contradiction via (k₀, k)
      exfalso
      have hcr_k₀k := hcross k₀ k hk₀ hk hk₀k
      rcases irreducibleCharacter_cross_pair_classify
          (hαβFun k₀ hk₀) (hαβFun k hk) hcr_k₀k with
        ⟨hαk₀k, _, _, _⟩ | ⟨_, _, _, hβk₀k⟩
      · exact hα_ne (hAk₀.trans hαk₀k)
      · exact hβ_ne_k₀ (hβ_eq.trans hβk₀k.symm)
    -- μFun: μ_0 := αFun one, μ_i := βFun i for i ≠ 0
    let μFun : Fin n → IrreducibleCharacter G :=
      Function.update βFun 0 (αFun one)
    have hμ0 : μFun 0 = αFun one := Function.update_self _ _ _
    have hμne : ∀ i, i ≠ 0 → μFun i = βFun i := fun i hi =>
      Function.update_of_ne hi _ _
    -- Injectivity of μFun
    have hμInj : Function.Injective μFun := by
      intro i j hij
      by_cases hi : i = 0
      · by_cases hj : j = 0
        · exact hi.trans hj.symm
        exfalso
        rw [hi, hμ0, hμne j hj] at hij
        -- hij : αFun one = βFun j; but αFun j = αFun one and αFun j ≠ βFun j
        exact hαβFun j hj ((hαAll j hj).trans hij)
      by_cases hj : j = 0
      · exfalso
        rw [hμne i hi, hj, hμ0] at hij
        -- hij : βFun i = αFun one; but αFun i = αFun one and αFun i ≠ βFun i
        exact hαβFun i hi ((hαAll i hi).trans hij.symm)
      rw [hμne i hi, hμne j hj] at hij
      by_contra hne
      have hcr_ij := hcross i j hi hj hne
      have hα_eq : αFun i = αFun j := (hαAll i hi).trans (hαAll j hj).symm
      rcases irreducibleCharacter_cross_pair_classify
          (hαβFun i hi) (hαβFun j hj) hcr_ij with
        ⟨_, _, _, hβ_ne⟩ | ⟨hα_ne, _, _, _⟩
      · exact hβ_ne hij
      · exact hα_ne hα_eq
    -- Construct data
    refine ⟨{ mu := μFun, sign := -1, sign_eq := Or.inr rfl, injective := hμInj }, ?_⟩
    intro i
    by_cases hi : i = 0
    · subst hi
      change isometryDifferenceImage τ χ 0 =
        (-1 : ℤ) • ((μFun 0 : ClassFunction G ℂ) - (μFun 0 : ClassFunction G ℂ))
      rw [isometryDifferenceImage_zero, sub_self, smul_zero]
    · rw [hvFun i hi]
      have hα_i : αFun i = αFun one := hαAll i hi
      change (αFun i : ClassFunction G ℂ) - (βFun i : ClassFunction G ℂ) =
        (-1 : ℤ) • ((μFun i : ClassFunction G ℂ) - (μFun 0 : ClassFunction G ℂ))
      rw [hμne i hi, hμ0, hα_i, neg_one_zsmul, neg_sub]
  · -- Type B globally: sign = 1, μ_0 := βFun one, μ_i := αFun i (i ≠ 0)
    push Not at hExA
    have hβAll : ∀ k : Fin n, k ≠ 0 → βFun k = βFun one := by
      intro k hk
      by_cases hkOne : k = one
      · subst hkOne; rfl
      have hcr_onek := hcross one k hone_ne_zero hk (Ne.symm hkOne)
      rcases irreducibleCharacter_cross_pair_classify
          (hαβFun one hone_ne_zero) (hαβFun k hk) hcr_onek with
        ⟨hα_eq, _, _, _⟩ | ⟨_, _, _, hβ_eq⟩
      · exact absurd hα_eq (hExA k hk hkOne)
      · exact hβ_eq.symm
    let μFun : Fin n → IrreducibleCharacter G :=
      Function.update αFun 0 (βFun one)
    have hμ0 : μFun 0 = βFun one := Function.update_self _ _ _
    have hμne : ∀ i, i ≠ 0 → μFun i = αFun i := fun i hi =>
      Function.update_of_ne hi _ _
    have hμInj : Function.Injective μFun := by
      intro i j hij
      by_cases hi : i = 0
      · by_cases hj : j = 0
        · exact hi.trans hj.symm
        exfalso
        rw [hi, hμ0, hμne j hj] at hij
        -- hij : βFun one = αFun j; but βFun j = βFun one and αFun j ≠ βFun j
        exact hαβFun j hj (hij.symm.trans (hβAll j hj).symm)
      by_cases hj : j = 0
      · exfalso
        rw [hμne i hi, hj, hμ0] at hij
        exact hαβFun i hi (hij.trans (hβAll i hi).symm)
      rw [hμne i hi, hμne j hj] at hij
      by_contra hne
      have hcr_ij := hcross i j hi hj hne
      have hβ_eq : βFun i = βFun j := (hβAll i hi).trans (hβAll j hj).symm
      rcases irreducibleCharacter_cross_pair_classify
          (hαβFun i hi) (hαβFun j hj) hcr_ij with
        ⟨_, _, _, hβ_ne⟩ | ⟨hα_ne, _, _, _⟩
      · exact hβ_ne hβ_eq
      · exact hα_ne hij
    refine ⟨{ mu := μFun, sign := 1, sign_eq := Or.inl rfl, injective := hμInj }, ?_⟩
    intro i
    by_cases hi : i = 0
    · subst hi
      change isometryDifferenceImage τ χ 0 =
        (1 : ℤ) • ((μFun 0 : ClassFunction G ℂ) - (μFun 0 : ClassFunction G ℂ))
      rw [isometryDifferenceImage_zero, sub_self, smul_zero]
    · rw [hvFun i hi]
      have hβ_i : βFun i = βFun one := hβAll i hi
      change (αFun i : ClassFunction G ℂ) - (βFun i : ClassFunction G ℂ) =
        (1 : ℤ) • ((μFun i : ClassFunction G ℂ) - (μFun 0 : ClassFunction G ℂ))
      rw [hμne i hi, hμ0, hβ_i, one_zsmul]

end OddOrder.RepresentationTheory
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/GroupTheory/RepresentationTheory/IsometryDifferencePair.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