Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Frobenius irreducibility of induced characters (Isaacs Thm 6.34)

Definition
OddOrder_GroupTheory_RepresentationTheory_InducedIrreducible

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

feit-thompsonfinite-groupsgroup-theoryishida-odd-order

For a normal subgroup H⊴GH\trianglelefteq GH⊴G with quotient W1=G/HW_1=G/HW1​=G/H, this file builds toward Isaacs, Character Theory of Finite Groups, Thm 6.34: when W1W_1W1​ acts freely on Irr H∖{1}\mathrm{Irr}\,H\setminus\{1\}IrrH∖{1} (the Frobenius/Dade situation of Peterfalvi §6), the induced character IndHGθ\mathrm{Ind}_H^G\thetaIndHG​θ of a nontrivial θ\thetaθ is irreducible of degree ∣W1∣θ(1)|W_1|\theta(1)∣W1​∣θ(1). This supplies the two facts Peterfalvi (6.8) needs: the constituents lie in Irr G\mathrm{Irr}\,GIrrG and share the common degree ∣W1∣|W_1|∣W1​∣.

Declarations include Mackey restriction as an orbit sum (card_smul_restrict_induce_eq_inertia_smul_orbitSum), orthogonality of induced characters from non-conjugate irreducibles, the norm of an induced irreducible character, induce_eq_induce_iff_conj, injectivity of induction at inertia-stable irreducibles, and orbit size equal to inertia index.

The bundle also proves 12 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/InducedIrreducible.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_InducedCharacter
import Definitions.Def_OddOrder_GroupTheory_RepresentationTheory_Inertia
import Definitions.Def_OddOrder_GroupTheory_RepresentationTheory_ZIrrFourier
import Definitions.Def_OddOrder_GroupTheory_RepresentationTheory_ColumnOrthogonality
import Definitions.Def_OddOrder_GroupTheory_RepresentationTheory_Clifford
import Definitions.Def_OddOrder_GroupTheory_RepresentationTheory_ConjugationBrauer
import Definitions.Def_OddOrder_GroupTheory_RepresentationTheory_BrauerPermutationUnconditional
import Definitions.Def_OddOrder_Isaacs_Ch06_FrobeniusActions_FrobeniusGroup

namespace OddOrder.RepresentationTheory

open ClassFunction

variable {G : Type*} [Group G] {H : Subgroup G} [hH : H.Normal]

variable {k : Type*} [CommRing k]

/-- Evaluation of a finite sum of class functions is the sum of the evaluations.

A pointwise companion to `ClassFunction.add_apply`/`zero_apply`, needed to unfold the
right-hand side of the Mackey restriction formula. -/
@[simp] theorem ClassFunction.finset_sum_apply {ι : Type*} (s : Finset ι)
    (f : ι → ClassFunction G k) (g : G) :
    (∑ i ∈ s, f i) g = ∑ i ∈ s, f i g := by
  classical
  induction s using Finset.induction with
  | empty => simp
  | @insert a s ha ih =>
      rw [Finset.sum_insert ha, Finset.sum_insert ha, ClassFunction.add_apply, ih]

section FrobeniusInertia

variable [Finite G] [Finite H]

/-- Frobenius-group specialization of the Brauer-conjugation inertia bridge.

If `G = H ⋊ W` is a Frobenius group with kernel `H`, every nontrivial irreducible character of
`H` has inertia subgroup exactly `H`.  This is the form needed in Peterfalvi (6.8), where
`[Is]` Theorem 6.34 is applied to `Ind_H^L θ` in the Frobenius case. -/
theorem inertia_eq_of_frobeniusGroup {W : Subgroup G}
    (hF : OddOrder.Isaacs.Ch06.IsFrobeniusGroup G H W)
    {θ : IrreducibleCharacter H} (hθ_ne : θ ≠ trivialIrreducibleCharacter H) :
    ClassFunction.inertia (G := G) (H := H) (θ : ClassFunction H ℂ) = H := by
  refine inertia_eq_of_freeAction (G := G) (H := H) ?_ hθ_ne
  intro h hh
  have hhG : (h : G) ≠ 1 := by
    intro h1
    exact hh (Subtype.ext h1)
  exact hF.centralizer_kernel_le (h : G) h.property hhG

end FrobeniusInertia

variable [Fintype G] [Invertible (Nat.card H : k)]

/-- **Mackey restriction formula** (normal-subgroup case), in unnormalized form.

For `H ⊴ G` and `θ : ClassFunction ↥H k`,
`|H| • Res_H (Ind_H^G θ) = ∑_{x ∈ G} θ^{x⁻¹}`,
where `θ^{x⁻¹} = ClassFunction.conjBy x⁻¹ θ`.

Summing over the *whole* group `G` (rather than a transversal) keeps the statement free of a
choice of coset representatives: each left coset `xH` contributes `|H|` equal summands
`θ^{x⁻¹}` — equal because `conjBy` is constant along the coset
(`ClassFunction.conjBy_eq_self_of_mem`) — so the global `|H|` factor exactly cancels the
`|H|⁻¹` built into `induce`. The classical transversal form
`Res_H (Ind_H^G θ) = ∑_{w ∈ G/H} θ^{w⁻¹}` is the same identity divided by `|H|`.

The proof is pointwise: for `h ∈ H` every conjugate `x⁻¹ h x` stays in `H` (normality), so
no induction term vanishes (`induceTerm_of_mem_normal`), and `θ(x⁻¹ h x)` is exactly
`(conjBy x⁻¹ θ)(h)`. -/
theorem card_smul_restrict_induce (θ : ClassFunction ↥H k) :
    (Nat.card H : k) • restrict H (induce H θ) = ∑ x : G, conjBy x⁻¹ θ := by
  ext h
  rw [ClassFunction.smul_apply, restrict_apply, induce_apply, ClassFunction.finset_sum_apply,
    ← mul_assoc, mul_invOf_self, one_mul]
  refine Finset.sum_congr rfl fun x _ => ?_
  rw [induceTerm_of_mem_normal (le_refl H) θ h.property x, conjBy_apply]
  exact congrArg θ (Subtype.ext (by group))

open scoped Classical in
/-- **Mackey restriction as an orbit sum** (normal-subgroup case): grouping the Mackey sum
`|H| • Res_H (Ind_H^G θ) = ∑_{x∈G} θ^{x⁻¹}` (`card_smul_restrict_induce`) by the value of the
conjugate gives

  `|H| • Res_H (Ind_H^G θ) = |I_G(θ)| • ∑_{ψ ∈ orbit(θ)} ψ`,

where the orbit is the `Finset` image of `x ↦ θ^{x⁻¹}` over `G` and each fibre
`{x | θ^{x⁻¹} = ψ}` is a right coset of the inertia group `I_G(θ)` (hence of constant size
`|I_G(θ)|`).  Combined with `|H|·‖Ind θ‖² = |I_G(θ)|`
(`card_mul_inner_self_induce_eq_card_inertia`) this is the classical
`Res_H (Ind_H^G θ) = ‖Ind θ‖² · (sum of the distinct conjugates)` — the "`(1/‖ζ_i‖²)·Res ζ_i`
is a character" step of Peterfalvi (13.5.a). -/
theorem card_smul_restrict_induce_eq_inertia_smul_orbitSum (θ : ClassFunction ↥H k) :
    (Nat.card H : k) • restrict H (induce H θ)
      = (Nat.card ↥(ClassFunction.inertia (G := G) (H := H) θ)) •
          ∑ ψ ∈ Finset.univ.image (fun x : G => ClassFunction.conjBy x⁻¹ θ), ψ := by
  classical
  rw [card_smul_restrict_induce, Finset.sum_comp (fun ψ : ClassFunction ↥H k => ψ)
    (fun x : G => ClassFunction.conjBy x⁻¹ θ)]
  -- Each fibre is a right coset of the inertia group.
  have hfiber : ∀ ψ ∈ Finset.univ.image (fun x : G => ClassFunction.conjBy x⁻¹ θ),
      (Finset.univ.filter (fun x : G => ClassFunction.conjBy x⁻¹ θ = ψ)).card
        = Nat.card ↥(ClassFunction.inertia (G := G) (H := H) θ) := by
    intro ψ hψ
    obtain ⟨x₀, -, rfl⟩ := Finset.mem_image.mp hψ
    have : Fintype ↥(ClassFunction.inertia (G := G) (H := H) θ) := Fintype.ofFinite _
    rw [Nat.card_eq_fintype_card, ← Finset.card_univ]
    -- Bijection `x ↦ x₀⁻¹·x` from the fibre (the left coset `x₀·I`) onto the inertia group.
    refine (Finset.card_bij' (fun x hx => (⟨x₀⁻¹ * x, ?_⟩ : ↥(ClassFunction.inertia
        (G := G) (H := H) θ))) (fun j _ => x₀ * (j : G)) ?_ ?_ ?_ ?_)
    · -- the fibre condition puts `x₀⁻¹·x` in the inertia group (`conjBy` composes
      -- contravariantly: `conjBy (a·b) = conjBy b ∘ conjBy a`)
      have hfib : ClassFunction.conjBy x⁻¹ θ = ClassFunction.conjBy x₀⁻¹ θ :=
        (Finset.mem_filter.mp hx).2
      rw [ClassFunction.mem_inertia]
      calc ClassFunction.conjBy (x₀⁻¹ * x) θ
          = ClassFunction.conjBy x (ClassFunction.conjBy x₀⁻¹ θ) := by
            rw [ClassFunction.conjBy_mul]
        _ = ClassFunction.conjBy x (ClassFunction.conjBy x⁻¹ θ) := by rw [hfib]
        _ = ClassFunction.conjBy (x⁻¹ * x) θ := by rw [ClassFunction.conjBy_mul]
        _ = θ := by rw [inv_mul_cancel, ClassFunction.conjBy_one]
    · -- membership of the image point (trivial: full finset)
      intro x hx
      exact Finset.mem_univ _
    · -- the reverse map lands in the fibre
      intro j _
      refine Finset.mem_filter.mpr ⟨Finset.mem_univ _, ?_⟩
      have hjinv : ClassFunction.conjBy ((j : G))⁻¹ θ = θ :=
        ClassFunction.mem_inertia.mp (Subgroup.inv_mem _ j.2)
      calc ClassFunction.conjBy (x₀ * (j : G))⁻¹ θ
          = ClassFunction.conjBy ((j : G)⁻¹ * x₀⁻¹) θ := by rw [mul_inv_rev]
        _ = ClassFunction.conjBy x₀⁻¹ (ClassFunction.conjBy ((j : G))⁻¹ θ) := by
            rw [ClassFunction.conjBy_mul]
        _ = ClassFunction.conjBy x₀⁻¹ θ := by rw [hjinv]
    · -- left inverse
      intro x hx
      group
    · -- right inverse
      intro j _
      exact Subtype.ext (by group)
  rw [Finset.sum_congr rfl (fun ψ hψ => by rw [hfiber ψ hψ]), ← Finset.smul_sum]

section Complex

variable [Fintype H] [Invertible (Nat.card G : ℂ)] [Invertible (Nat.card H : ℂ)]

/-- **Norm of an induced character via the Mackey sum** (any class function).

`|H| · ⟨Ind_H^G θ, Ind_H^G θ⟩ = ∑_{x ∈ G} ⟨θ, θ^{x⁻¹}⟩`.

This is Frobenius reciprocity combined with the Mackey restriction formula
`card_smul_restrict_induce`: pairing `θ` against `|H| • Res_H (Ind θ) = ∑_x θ^{x⁻¹}` and using
that the inner product is conjugate-linear on the right (`inner_smul_right`, with
`star (|H| : ℂ) = |H|`). No irreducibility is needed yet. -/
theorem card_mul_inner_self_induce (θ : ClassFunction ↥H ℂ) :
    (Nat.card H : ℂ) *
        ClassFunction.inner (induce H θ) (induce H θ) =
      ∑ x : G, ClassFunction.inner θ (conjBy x⁻¹ θ) := by
  have key : ClassFunction.inner θ ((Nat.card H : ℂ) • restrict H (induce H θ))
      = (Nat.card H : ℂ) * ClassFunction.inner (induce H θ) (induce H θ) := by
    rw [inner_smul_right, star_natCast, ← inner_induce_eq_inner_restrict]
  rw [← key, card_smul_restrict_induce, inner_sum_right]

/-- **Cross inner product of two induced characters via the Mackey sum** (any class functions).

`|H| · ⟨Ind_H^G θ, Ind_H^G ψ⟩ = ∑_{x ∈ G} ⟨θ, ψ^{x⁻¹}⟩`.  The two-argument generalization of
`card_mul_inner_self_induce`; the proof is identical, using Frobenius reciprocity
(`inner_induce_eq_inner_restrict`, general in the right argument) and the Mackey restriction
formula `card_smul_restrict_induce` applied to `ψ`. -/
theorem card_mul_inner_induce (θ ψ : ClassFunction ↥H ℂ) :
    (Nat.card H : ℂ) *
        ClassFunction.inner (induce H θ) (induce H ψ) =
      ∑ x : G, ClassFunction.inner θ (conjBy x⁻¹ ψ) := by
  have key : ClassFunction.inner θ ((Nat.card H : ℂ) • restrict H (induce H ψ))
      = (Nat.card H : ℂ) * ClassFunction.inner (induce H θ) (induce H ψ) := by
    rw [inner_smul_right, star_natCast, ← inner_induce_eq_inner_restrict]
  rw [← key, card_smul_restrict_induce, inner_sum_right]

omit [Fintype ↥H] in
/-- **Induced characters from non-conjugate irreducibles are orthogonal.**

If no `G`-conjugate of `θ` equals `ψ` (the two irreducibles of `H` lie in distinct `G`-orbits),
then `⟨Ind_H^G θ, Ind_H^G ψ⟩ = 0`.  Each Mackey summand `⟨θ, ψ^{x⁻¹}⟩` is an orthonormality
indicator (`irreducibleCharacter_inner_eq_ite`) that vanishes because `θ = ψ^{x⁻¹}` would give
`θ^x = ψ` (`conjBy_conjBy_inv`), contradicting the non-conjugacy hypothesis.  Used to prove the
`Y = S(H')` family `j ↦ Ind_H^L θ_j` injective. -/
theorem inner_induce_eq_zero_of_not_conj (θ ψ : IrreducibleCharacter H)
    (h : ∀ g : G, IrreducibleCharacter.conjBy g θ ≠ ψ) :
    ClassFunction.inner (induce H (θ : ClassFunction ↥H ℂ))
      (induce H (ψ : ClassFunction ↥H ℂ)) = 0 := by
  have : Fintype ↥H := Fintype.ofFinite ↥H
  classical
  have hcardH : (Nat.card H : ℂ) ≠ 0 := by rw [Nat.cast_ne_zero]; exact Nat.card_pos.ne'
  apply mul_left_cancel₀ hcardH
  rw [mul_zero, card_mul_inner_induce]
  refine Finset.sum_eq_zero fun x _ => ?_
  rw [← IrreducibleCharacter.coe_conjBy, irreducibleCharacter_inner_eq_ite, if_neg]
  intro hθ
  exact h x (by rw [hθ]; simp)

omit [Fintype ↥H] in
/-- **Norm of an induced irreducible character** ([Is] Thm 6.34, norm part).

For an irreducible character `θ` of `H ⊴ G`, `|H| · ‖Ind_H^G θ‖² = |I_G(θ)|`, equivalently
`‖Ind_H^G θ‖² = [I_G(θ) : H]`: the squared norm counts the cosets of `H` in the inertia group.

Each Mackey summand `⟨θ, θ^{x⁻¹}⟩` is `1` if `x ∈ I_G(θ)` and `0` otherwise, by orthonormality
of irreducible characters (`irreducibleCharacter_inner_eq_ite`) together with
`θ^{x⁻¹} = θ ⇔ x ∈ I_G(θ)`. Summing the indicator over `G` gives `|I_G(θ)|`. -/
theorem card_mul_inner_self_induce_eq_card_inertia (θ : IrreducibleCharacter H) :
    (Nat.card H : ℂ) *
        ClassFunction.inner (induce H (θ : ClassFunction ↥H ℂ))
          (induce H (θ : ClassFunction ↥H ℂ))
      = (Nat.card ↥(ClassFunction.inertia (θ : ClassFunction ↥H ℂ)) : ℂ) := by
  have : Fintype ↥H := Fintype.ofFinite ↥H
  classical
  rw [card_mul_inner_self_induce]
  have hterm : ∀ x : G,
      ClassFunction.inner (θ : ClassFunction ↥H ℂ) (conjBy x⁻¹ (θ : ClassFunction ↥H ℂ))
        = if x ∈ ClassFunction.inertia (θ : ClassFunction ↥H ℂ) then (1 : ℂ) else 0 := by
    intro x
    rw [← IrreducibleCharacter.coe_conjBy, irreducibleCharacter_inner_eq_ite]
    have hcond : (θ = IrreducibleCharacter.conjBy x⁻¹ θ)
        ↔ x ∈ ClassFunction.inertia (θ : ClassFunction ↥H ℂ) := by
      rw [eq_comm, IrreducibleCharacter.ext_iff, IrreducibleCharacter.coe_conjBy,
        ← ClassFunction.mem_inertia, Subgroup.inv_mem_iff]
    simp only [hcond]
  rw [Finset.sum_congr rfl fun x _ => hterm x, Finset.sum_boole, ← Fintype.card_subtype,
    Nat.card_eq_fintype_card]

omit [Fintype ↥H] in
/-- **Induced irreducibles coincide iff the sources are `G`-conjugate.**  For `θ, ψ ∈ Irr H`
(`H ⊴ G`), `Ind_H^G θ = Ind_H^G ψ` exactly when some `G`-conjugate of `θ` equals `ψ`.

Forward: if no conjugate of `θ` is `ψ`, then `Ind θ ⊥ Ind ψ` (`inner_induce_eq_zero_of_not_conj`),
so `Ind θ = Ind ψ` would force `‖Ind ψ‖² = 0`, contradicting `|H| · ‖Ind ψ‖² = |I_G(ψ)| > 0`
(`card_mul_inner_self_induce_eq_card_inertia`).  Backward: induction is conjugation-invariant
(`induce_conjBy_eq`).  This is the fibre description underpinning the orbit-counting of `S(A)`
(Peterfalvi (6.2), step 4a): the map `θ ↦ Ind_K^L θ` has fibres exactly the `L`-conjugacy orbits. -/
theorem induce_eq_induce_iff_conj (θ ψ : IrreducibleCharacter H) :
    induce H (θ : ClassFunction ↥H ℂ) = induce H (ψ : ClassFunction ↥H ℂ) ↔
      ∃ g : G, IrreducibleCharacter.conjBy g θ = ψ := by
  have : Fintype ↥H := Fintype.ofFinite ↥H
  classical
  constructor
  · intro heq
    by_contra hcon
    push Not at hcon
    have h0 := inner_induce_eq_zero_of_not_conj θ ψ hcon
    rw [heq] at h0
    have hpos : ClassFunction.inner (induce H (ψ : ClassFunction ↥H ℂ))
        (induce H (ψ : ClassFunction ↥H ℂ)) ≠ 0 := by
      have hcard := card_mul_inner_self_induce_eq_card_inertia ψ
      have hI : (Nat.card ↥(ClassFunction.inertia (ψ : ClassFunction ↥H ℂ)) : ℂ) ≠ 0 := by
        rw [Nat.cast_ne_zero]; exact Nat.card_pos.ne'
      intro hzero
      rw [hzero, mul_zero] at hcard
      exact hI hcard.symm
    exact hpos h0
  · rintro ⟨g, rfl⟩
    have : Fintype G := Fintype.ofFinite G
    rw [IrreducibleCharacter.coe_conjBy, induce_conjBy_eq]

omit [Fintype ↥H] in
/-- **Induction is injective at an inertia-stable irreducible.**  If `θ ∈ Irr H` (`H ⊴ G`) is fixed
by every `G`-conjugation (`I_G(θ) = G`, i.e. its conjugation orbit is the singleton `{θ}`), then
`Ind_H^G θ = Ind_H^G ψ` forces `ψ = θ`: the fibre of `ψ ↦ Ind ψ` over `Ind θ` is that singleton
orbit (`induce_eq_induce_iff_conj`).  This is the injectivity behind the `𝒳 ↔ 𝒮` count when the
relevant sources are `G`-invariant — Peterfalvi (9.5)/(9.9)'s `Smu` uniqueness (the Coq `ResIndXmu`
route: `Res_H (Ind θ) = [G:H] · θ` for `G`-stable `θ`). -/
theorem induce_injective_of_inertia_stable {θ ψ : IrreducibleCharacter H}
    (hθ : ∀ g : G, IrreducibleCharacter.conjBy g θ = θ)
    (h : induce H (θ : ClassFunction ↥H ℂ) = induce H (ψ : ClassFunction ↥H ℂ)) : ψ = θ := by
  have : Fintype ↥H := Fintype.ofFinite ↥H
  obtain ⟨g, hg⟩ := (induce_eq_induce_iff_conj θ ψ).mp h
  rw [← hg]; exact hθ g

omit [Fintype G] in
omit [Fintype H] [Invertible (Nat.card G : ℂ)] [Invertible (Nat.card H : ℂ)] in
/-- **Orbit size = inertia index.**  The `G`-conjugation orbit of `θ ∈ Irr H` (`H ⊴ G`) has
cardinality `[G : I_G(θ)]`, via the coset parametrization `conjByOrbitEquivLeftCosets`.  Together
with `induce_eq_induce_iff_conj` (orbit = fibre of `θ ↦ Ind θ`) this gives the multiplicity with
which each `χ ∈ S(A)` is hit, for the Peterfalvi (6.2) step-4a orbit count. -/
theorem card_conjByOrbit_eq_index_inertia (θ : IrreducibleCharacter H) :
    Nat.card (IrreducibleCharacter.conjByOrbit (G := G) (H := H) θ) =
      (IrreducibleCharacter.inertia (G := G) (H := H) θ).index :=
  (Nat.card_congr (IrreducibleCharacter.conjByOrbitEquivLeftCosets (G := G) (H := H) θ)).symm

omit [Fintype G] in
omit [Fintype H] [Invertible (Nat.card G : ℂ)] [Invertible (Nat.card H : ℂ)] in
/-- **Conjugation preserves degree.**  `(θ^g)(1) = θ(1)`: conjugation evaluates `θ` at
`g · 1 · g⁻¹ = 1`. -/
@[simp] theorem conjBy_apply_one (g : G) (θ : IrreducibleCharacter H) :
    ((IrreducibleCharacter.conjBy (G := G) (H := H) g θ : IrreducibleCharacter H) :
        ClassFunction ↥H ℂ) 1 = (θ : ClassFunction ↥H ℂ) 1 := by
  rw [IrreducibleCharacter.coe_conjBy, ClassFunction.conjBy_apply]
  refine congrArg _ (Subtype.ext ?_)
  simp

open scoped Classical in
omit [Fintype ↥H] in
/-- **Fibre cardinality = inertia index.**  Inside a conjugation-invariant Finset `T` of
irreducibles, the fibre `{θ ∈ T | Ind θ = Ind θ₀}` of the induction map equals the whole
`G`-orbit of `θ₀` (by `induce_eq_induce_iff_conj` and `T`-invariance), so its cardinality is
`[G : I_G(θ₀)]` (`card_conjByOrbit_eq_index_inertia`).  This is the multiplicity input to the
Peterfalvi (6.2) step-4a orbit count. -/
theorem card_filter_induce_eq_index_inertia (T : Finset (IrreducibleCharacter H))
    (hT : ∀ θ ∈ T, ∀ g : G, IrreducibleCharacter.conjBy (G := G) (H := H) g θ ∈ T)
    (θ₀ : IrreducibleCharacter H) (hθ₀ : θ₀ ∈ T) :
    (T.filter fun θ => induce H θ.toClassFunction = induce H θ₀.toClassFunction).card
      = (IrreducibleCharacter.inertia (G := G) (H := H) θ₀).index := by
  have : Fintype ↥H := Fintype.ofFinite ↥H
  rw [← card_conjByOrbit_eq_index_inertia (G := G) (H := H) θ₀,
    Nat.card_coe_set_eq, ← Set.ncard_coe_finset]
  congr 1
  ext θ
  simp only [Finset.coe_filter, Set.mem_ofPred_eq,
    IrreducibleCharacter.mem_conjByOrbit, induce_eq_induce_iff_conj]
  constructor
  · rintro ⟨_, g, hg⟩
    exact ⟨g⁻¹, by rw [← hg, ← IrreducibleCharacter.conjBy_mul]; simp⟩
  · rintro ⟨g, hg⟩
    exact ⟨hg ▸ hT θ₀ hθ₀ g, g⁻¹, by rw [← hg, ← IrreducibleCharacter.conjBy_mul]; simp⟩

open scoped Classical in
omit [Fintype ↥H] in
/-- **Peterfalvi (6.2), step 4a — the orbit-counted degree-sum identity.**  For a conjugation-
invariant Finset `T ⊆ Irr H` (`H ⊴ G`), summing `χ(1)²/‖χ‖²` over the induced characters
`S = {Ind_H^G θ | θ ∈ T}` equals `[G:H] · ∑_{θ∈T} θ(1)²`.

Each `χ = Ind θ` is hit by the whole `G`-orbit of `θ` (`card_filter_induce_eq_index_inertia`), of
size `[G:I_G(θ)]`, on which the degree is constant (`conjBy_apply_one`).  Combined with the norm
`|H|·‖Ind θ‖² = |I_G(θ)|` (`card_mul_inner_self_induce_eq_card_inertia`) and the degree
`Ind θ(1) = [G:H]·θ(1)` (`induce_apply_one`), the per-`χ` term telescopes via `[G:H]·|H| = |G| =
[G:I]·|I|` to `[G:H]·([G:I]·θ(1)²)`, which is exactly `[G:H]` times the fibre's degree-sum. -/
theorem sum_div_normSq_induce_image_eq (T : Finset (IrreducibleCharacter H))
    (hT : ∀ θ ∈ T, ∀ g : G, IrreducibleCharacter.conjBy (G := G) (H := H) g θ ∈ T) :
    ∑ χ ∈ T.image (fun θ => induce H θ.toClassFunction),
        χ 1 ^ 2 / ClassFunction.inner χ χ
      = (H.index : ℂ) * ∑ θ ∈ T, θ.toClassFunction 1 ^ 2 := by
  have : Fintype ↥H := Fintype.ofFinite ↥H
  have hcH : (Nat.card ↥H : ℂ) ≠ 0 := by rw [Nat.cast_ne_zero]; exact Nat.card_pos.ne'
  rw [Finset.mul_sum, ← Finset.sum_fiberwise_of_maps_to
    (fun θ (hθ : θ ∈ T) => Finset.mem_image_of_mem (fun θ => induce H θ.toClassFunction) hθ)]
  refine Finset.sum_congr rfl fun χ hχ => ?_
  obtain ⟨θ₀, hθ₀T, rfl⟩ := Finset.mem_image.mp hχ
  have hcI : (Nat.card ↥(IrreducibleCharacter.inertia (G := G) (H := H) θ₀) : ℂ) ≠ 0 := by
    rw [Nat.cast_ne_zero]; exact Nat.card_pos.ne'
  -- The fibre's `[G:H]·θ(1)²`-sum collapses to `[G:H]·[G:I]·θ₀(1)²` (constant degree + fibre card).
  have hfib : ∑ θ ∈ T.filter (fun θ =>
        induce H θ.toClassFunction = induce H θ₀.toClassFunction),
        (H.index : ℂ) * θ.toClassFunction 1 ^ 2
      = (H.index : ℂ)
          * (IrreducibleCharacter.inertia (G := G) (H := H) θ₀).index * θ₀.toClassFunction 1 ^ 2 :=
              by
    have hconst : ∀ θ ∈ T.filter (fun θ =>
        induce H θ.toClassFunction = induce H θ₀.toClassFunction),
        (H.index : ℂ) * θ.toClassFunction 1 ^ 2 = (H.index : ℂ) * θ₀.toClassFunction 1 ^ 2 := by
      intro θ hθ
      rw [Finset.mem_filter] at hθ
      obtain ⟨g, hg⟩ := (induce_eq_induce_iff_conj θ θ₀).mp hθ.2
      rw [← hg, conjBy_apply_one]
    rw [Finset.sum_congr rfl hconst, Finset.sum_const,
      card_filter_induce_eq_index_inertia T hT θ₀ hθ₀T, nsmul_eq_mul]
    ring
  rw [hfib, induce_apply_one]
  have hinner : ClassFunction.inner (induce H θ₀.toClassFunction) (induce H θ₀.toClassFunction)
      = (Nat.card ↥(IrreducibleCharacter.inertia (G := G) (H := H) θ₀) : ℂ) / (Nat.card ↥H : ℂ) :=
          by
    rw [eq_div_iff hcH, mul_comm]
    exact card_mul_inner_self_induce_eq_card_inertia θ₀
  rw [hinner]
  have hHmul : (H.index : ℂ) * (Nat.card ↥H : ℂ) = (Nat.card G : ℂ) := by
    rw [← Nat.cast_mul, Subgroup.index_mul_card]
  have hImul : ((IrreducibleCharacter.inertia (G := G) (H := H) θ₀).index : ℂ)
      * (Nat.card ↥(IrreducibleCharacter.inertia (G := G) (H := H) θ₀) : ℂ) = (Nat.card G : ℂ) := by
    rw [← Nat.cast_mul, Subgroup.index_mul_card]
  have hAB : (H.index : ℂ) * (Nat.card ↥H : ℂ)
      = ((IrreducibleCharacter.inertia (G := G) (H := H) θ₀).index : ℂ)
          * (Nat.card ↥(IrreducibleCharacter.inertia (G := G) (H := H) θ₀) : ℂ) := hHmul.trans
              hImul.symm
  rw [div_div_eq_mul_div, div_eq_iff hcI]
  linear_combination (θ₀.toClassFunction 1 ^ 2 * (H.index : ℂ)) * hAB

/-- Integer-vector combinatorics: if nonzero integer coefficients on a finite set have squares
summing to `1`, the set is a singleton with coefficient `±1`. (The `∑ = 1` analogue of
`exists_pair_of_sum_sq_eq_two`.) -/
theorem exists_single_of_sum_sq_eq_one {ι : Type*} {s : Finset ι} {c : ι → ℤ}
    (hne : ∀ a ∈ s, c a ≠ 0) (hsum : ∑ a ∈ s, c a ^ 2 = 1) :
    ∃ α, s = {α} ∧ (c α = 1 ∨ c α = -1) := by
  classical
  have hge : ∀ a ∈ s, 1 ≤ c a ^ 2 := fun a ha => by
    have h := Int.one_le_abs (hne a ha)
    calc (1 : ℤ) = 1 ^ 2 := by ring
      _ ≤ |c a| ^ 2 := by gcongr
      _ = c a ^ 2 := sq_abs (c a)
  have hcard : s.card = 1 := by
    have hle : (s.card : ℤ) ≤ ∑ a ∈ s, c a ^ 2 := by
      calc (s.card : ℤ) = ∑ _a ∈ s, (1 : ℤ) := by rw [Finset.sum_const, nsmul_eq_mul, mul_one]
        _ ≤ ∑ a ∈ s, c a ^ 2 := Finset.sum_le_sum hge
    rw [hsum] at hle
    have hpos : 0 < s.card := by
      rcases Finset.eq_empty_or_nonempty s with rfl | h
      · simp at hsum
      · exact Finset.card_pos.mpr h
    omega
  obtain ⟨α, rfl⟩ := Finset.card_eq_one.mp hcard
  rw [Finset.sum_singleton] at hsum
  refine ⟨α, rfl, ?_⟩
  have hmul : c α * c α = 1 := by rw [← sq]; exact hsum
  rcases Int.eq_one_or_neg_one_of_mul_eq_one' hmul with ⟨h, _⟩ | ⟨h, _⟩
  · exact Or.inl h
  · exact Or.inr h

set_option backward.isDefEq.respectTransparency false in
/-- **A virtual character of squared norm `1` and positive degree is irreducible.**

If `φ ∈ ℤ[Irr G]` has `‖φ‖² = 1` and `φ(1)` is a positive natural number, then `φ` is an
irreducible character. By Parseval its Fourier coefficients are integers with `∑ cᵢ² = 1`, so
exactly one is `±1`; positivity of `φ(1)` rules out the `-1` case, leaving `φ = χ` for an
irreducible `χ`. (The norm-`1` analogue of `exists_irr_sub_irr_of_inner_self_two`.) -/
theorem isIrreducibleCharacter_of_inner_self_one_of_apply_one_pos
    {φ : ClassFunction G ℂ} (hφ : φ ∈ ZIrr G)
    (hnorm : ClassFunction.inner φ φ = 1)
    (hpos : ∃ d : ℕ, 0 < d ∧ (φ : G → ℂ) 1 = (d : ℂ)) :
    IsIrreducibleCharacter φ := by
  classical
  have : Finite G := Finite.of_fintype G
  obtain ⟨c, hsupp, hrepr, hsq⟩ := mem_ZIrr_inner_self_eq_sum_sq hφ
  have hsumC : ∑ a ∈ c.support, (c a : ℂ) ^ 2 = 1 := hsq.symm.trans hnorm
  have hsumZ : ∑ a ∈ c.support, c a ^ 2 = 1 := by exact_mod_cast hsumC
  have hne : ∀ a ∈ c.support, c a ≠ 0 := fun a ha => Finsupp.mem_support_iff.mp ha
  obtain ⟨α₀, hs, hcα⟩ := exists_single_of_sum_sq_eq_one hne hsumZ
  have hα₀ : α₀ ∈ irreducibleCharacters G := hsupp (by rw [hs]; simp)
  rw [hs, Finset.sum_singleton] at hrepr
  obtain ⟨d, hd, hφ1⟩ := hpos
  obtain ⟨e, he, hα1⟩ :=
    irreducibleCharacter_apply_one_eq_pos_natCast (⟨α₀, hα₀⟩ : IrreducibleCharacter G)
  simp only [IrreducibleCharacter.coe_mk] at hα1
  rcases hcα with hcα | hcα
  · -- c α₀ = 1: φ = α₀, which is irreducible
    rw [hrepr, hcα]
    simp only [Int.cast_one, one_smul]
    exact hα₀
  · -- c α₀ = -1: φ(1) = -e < 0, contradicting φ(1) = d > 0
    exfalso
    have hval : (φ : G → ℂ) 1 = -(e : ℂ) := by
      rw [hrepr, hcα]
      simp only [Int.cast_neg, Int.cast_one, ClassFunction.smul_apply, hα1]
      ring
    rw [hval] at hφ1
    have hsum0 : (d : ℂ) + (e : ℂ) = 0 := by linear_combination -hφ1
    rw [← Nat.cast_add] at hsum0
    have : d + e = 0 := by exact_mod_cast hsum0
    omega

/-- **A virtual character of squared norm `1` is `±` an irreducible character.**

The signed refinement of `isIrreducibleCharacter_of_inner_self_one_of_apply_one_pos`: without
the positivity input, Parseval still forces a single Fourier coefficient `±1`, so
`φ = ε • ξ` with `ε ∈ {1, -1}` and `ξ ∈ Irr G`.  This is the normalization step of
Peterfalvi (5.9.a) ("there is an integer `ε = ±1` such that `ε φ^{τ₁} ∈ Irr G`"). -/
theorem exists_zsmul_irreducibleCharacter_of_inner_self_one
    {φ : ClassFunction G ℂ} (hφ : φ ∈ ZIrr G)
    (hnorm : ClassFunction.inner φ φ = 1) :
    ∃ (ε : ℤ) (ξ : IrreducibleCharacter G),
      (ε = 1 ∨ ε = -1) ∧ φ = ε • (ξ : ClassFunction G ℂ) := by
  classical
  have : Finite G := Finite.of_fintype G
  obtain ⟨c, hsupp, hrepr, hsq⟩ := mem_ZIrr_inner_self_eq_sum_sq hφ
  have hsumC : ∑ a ∈ c.support, (c a : ℂ) ^ 2 = 1 := hsq.symm.trans hnorm
  have hsumZ : ∑ a ∈ c.support, c a ^ 2 = 1 := by exact_mod_cast hsumC
  have hne : ∀ a ∈ c.support, c a ≠ 0 := fun a ha => Finsupp.mem_support_iff.mp ha
  obtain ⟨α₀, hs, hcα⟩ := exists_single_of_sum_sq_eq_one hne hsumZ
  have hα₀ : α₀ ∈ irreducibleCharacters G := hsupp (by rw [hs]; simp)
  rw [hs, Finset.sum_singleton] at hrepr
  refine ⟨c α₀, ⟨α₀, hα₀⟩, hcα, ?_⟩
  rw [hrepr]
  exact Int.cast_smul_eq_zsmul ℂ (c α₀) α₀

omit [Fintype ↥H] in
/-- **[Is] Theorem 6.34** (Frobenius/Dade induced irreducibility).

For `H ⊴ G` and an irreducible character `θ` of `H` whose inertia group in `G` is just `H`
(the free-action / Frobenius condition — which holds for `θ ≠ 1` when `G/H` acts freely on
`Irr H ∖ {1}`), the induced character `Ind_H^G θ` is irreducible. Its degree is `[G : H] · θ(1)`
(`induce_apply_one`).

Proof: `Ind θ ∈ ℤ[Irr G]` (`induce_mem_ZIrr`); `‖Ind θ‖² = |I_G(θ)|/|H| = 1` from
`card_mul_inner_self_induce_eq_card_inertia` with `I_G(θ) = H`; and `Ind θ (1) = [G:H]·θ(1) > 0`.
Norm `1` with positive degree forces irreducibility. -/
theorem isIrreducibleCharacter_induce_of_inertia_eq (θ : IrreducibleCharacter H)
    (hfree : ClassFunction.inertia (θ : ClassFunction ↥H ℂ) = H) :
    IsIrreducibleCharacter (induce H (θ : ClassFunction ↥H ℂ)) := by
  have : Fintype ↥H := Fintype.ofFinite ↥H
  have hmem : induce H (θ : ClassFunction ↥H ℂ) ∈ ZIrr G :=
    induce_mem_ZIrr H θ.property.mem_ZIrr
  have hcardH : (Nat.card H : ℂ) ≠ 0 := by
    rw [Nat.cast_ne_zero]; exact Nat.card_pos.ne'
  have hnorm : ClassFunction.inner (induce H (θ : ClassFunction ↥H ℂ))
      (induce H (θ : ClassFunction ↥H ℂ)) = 1 := by
    have h := card_mul_inner_self_induce_eq_card_inertia θ
    rw [hfree] at h
    exact mul_left_cancel₀ hcardH (by rw [h, mul_one])
  have hpos : ∃ d : ℕ, 0 < d ∧ (induce H (θ : ClassFunction ↥H ℂ) : G → ℂ) 1 = (d : ℂ) := by
    have : Finite G := Finite.of_fintype G
    obtain ⟨e, he, hθ1⟩ := irreducibleCharacter_apply_one_eq_pos_natCast θ
    have hidx : 0 < H.index := Nat.pos_of_ne_zero fun h0 => by
      have hmc := H.index_mul_card
      rw [h0, zero_mul] at hmc
      exact (Nat.card_pos (α := G)).ne' hmc.symm
    refine ⟨H.index * e, Nat.mul_pos hidx he, ?_⟩
    rw [induce_apply_one, hθ1, Nat.cast_mul]
  exact isIrreducibleCharacter_of_inner_self_one_of_apply_one_pos hmem hnorm hpos

omit [Fintype ↥H] in
/-- Frobenius-group form of **[Is] Theorem 6.34**.

In a Frobenius group with kernel `H`, inducing any nontrivial irreducible character of `H`
to the ambient group gives an irreducible character.  The proof combines
`inertia_eq_of_frobeniusGroup` with `isIrreducibleCharacter_induce_of_inertia_eq`; the degree
formula remains `ClassFunction.induce_apply_one`. -/
theorem isIrreducibleCharacter_induce_of_frobeniusGroup {W : Subgroup G}
    (hF : OddOrder.Isaacs.Ch06.IsFrobeniusGroup G H W)
    (θ : IrreducibleCharacter H) (hθ_ne : θ ≠ trivialIrreducibleCharacter H) :
    IsIrreducibleCharacter (induce H (θ : ClassFunction H ℂ)) :=
  isIrreducibleCharacter_induce_of_inertia_eq θ
    (inertia_eq_of_frobeniusGroup hF hθ_ne)

omit [Fintype ↥H] in
/-- **Norm of a Frobenius-induced character is `1`.**  In a Frobenius group with kernel `H`, a
nontrivial irreducible `θ` of `H` induces with `‖Ind_H^G θ‖² = |I_G(θ)|/|H| = 1` — the inertia is
`H` (`inertia_eq_of_frobeniusGroup`), so `card_mul_inner_self_induce_eq_card_inertia` reads
`|H|·‖Ind θ‖² = |H|`.  This is the `‖ζ_0‖² = 1` input that the §7 (7.8) `Hypothesis78` norm
machinery needs for an `Ind`-distinguished family member (e.g. the §12 type-I coherent family). -/
theorem inner_self_induce_eq_one_of_frobeniusGroup {W : Subgroup G}
    (hF : OddOrder.Isaacs.Ch06.IsFrobeniusGroup G H W)
    (θ : IrreducibleCharacter H) (hθ_ne : θ ≠ trivialIrreducibleCharacter H) :
    ClassFunction.inner (induce H (θ : ClassFunction H ℂ))
      (induce H (θ : ClassFunction H ℂ)) = 1 := by
  have : Fintype ↥H := Fintype.ofFinite ↥H
  have hcardH : (Nat.card H : ℂ) ≠ 0 := by rw [Nat.cast_ne_zero]; exact Nat.card_pos.ne'
  have h := card_mul_inner_self_induce_eq_card_inertia θ
  rw [inertia_eq_of_frobeniusGroup hF hθ_ne] at h
  exact mul_left_cancel₀ hcardH (by rw [h, mul_one])

open scoped Classical in
omit [Fintype ↥H] in
/-- **Odd-order Frobenius: a nontrivial induced irreducible is orthogonal to its complex
conjugate.**  In a Frobenius group `G` of odd order with kernel `H`, for `θ ∈ Irr H`, `θ ≠ 1`,
the induced `Ind_H^G θ` is irreducible (`isIrreducibleCharacter_induce_of_frobeniusGroup`) and
nontrivial (`⟨Ind θ, 1_G⟩ = ⟨θ, 1_H⟩ = 0 ≠ 1`), hence — `G` odd — not real
(`not_isReal_of_ne_trivial_of_odd_card'`): `(Ind θ)‾ = Ind θ̄ ≠ Ind θ`.  Distinct irreducibles are
orthogonal, so `⟨Ind θ, Ind θ̄⟩ = 0`. -/
theorem inner_induce_conj_eq_zero_of_frobenius_of_odd {W : Subgroup G}
    (hodd : Odd (Nat.card G)) (hF : OddOrder.Isaacs.Ch06.IsFrobeniusGroup G H W)
    (θ : IrreducibleCharacter H) (hθ : θ ≠ trivialIrreducibleCharacter H) :
    ClassFunction.inner (ClassFunction.induce H (θ : ClassFunction H ℂ))
      (ClassFunction.induce H ((θ : ClassFunction H ℂ).conj)) = 0 := by
  have : Fintype ↥H := Fintype.ofFinite ↥H
  -- `θ̄` is again a nontrivial irreducible character of `H`.
  have hθbar_ne : (⟨(θ : ClassFunction H ℂ).conj, θ.isIrreducible.conj⟩ :
      IrreducibleCharacter H) ≠ trivialIrreducibleCharacter H := by
    intro h
    apply hθ
    have hcoe : (θ : ClassFunction H ℂ).conj = trivialClassFunction H := by
      have h2 := congrArg (fun c : IrreducibleCharacter H => (c : ClassFunction H ℂ)) h
      simpa using h2
    apply Subtype.ext
    change (θ : ClassFunction H ℂ) = trivialClassFunction H
    rw [← ClassFunction.conj_conj (θ : ClassFunction H ℂ), hcoe]
    exact trivialClassFunction_isReal
  have hirr := isIrreducibleCharacter_induce_of_frobeniusGroup hF θ hθ
  have hirr' := isIrreducibleCharacter_induce_of_frobeniusGroup hF
    (⟨(θ : ClassFunction H ℂ).conj, θ.isIrreducible.conj⟩ : IrreducibleCharacter H) hθbar_ne
  -- `Ind θ` is nontrivial: `⟨Ind θ, 1_G⟩ = ⟨θ, 1_H⟩ = 0`, but `⟨1_G, 1_G⟩ = 1`.
  have hne_triv : (⟨ClassFunction.induce H (θ : ClassFunction H ℂ), hirr⟩ :
      IrreducibleCharacter G) ≠ trivialIrreducibleCharacter G := by
    intro h
    have hrestrict : ClassFunction.restrict H
          (trivialIrreducibleCharacter G : ClassFunction G ℂ)
        = (trivialIrreducibleCharacter H : ClassFunction H ℂ) := by
      ext x
      simp [ClassFunction.restrict_apply, IrreducibleCharacter.coe_trivialIrreducibleCharacter,
        trivialClassFunction_apply]
    have hzero : ClassFunction.inner (ClassFunction.induce H (θ : ClassFunction H ℂ))
        (trivialIrreducibleCharacter G : ClassFunction G ℂ) = 0 := by
      rw [ClassFunction.inner_induce_eq_inner_restrict, hrestrict,
        irreducibleCharacter_inner_eq_ite, if_neg hθ]
    have hcf : ClassFunction.induce H (θ : ClassFunction H ℂ)
        = (trivialIrreducibleCharacter G : ClassFunction G ℂ) :=
      congrArg (fun c : IrreducibleCharacter G => (c : ClassFunction G ℂ)) h
    rw [hcf, irreducibleCharacter_inner_eq_ite, if_pos rfl] at hzero
    exact one_ne_zero hzero
  -- Odd order gives `(Ind θ)‾ ≠ Ind θ`.
  have hnotreal := not_isReal_of_ne_trivial_of_odd_card' hodd hne_triv
  have hconj_eq : (ClassFunction.induce H (θ : ClassFunction H ℂ)).conj
      = ClassFunction.induce H ((θ : ClassFunction H ℂ).conj) :=
    ClassFunction.induce_conj H (θ : ClassFunction H ℂ)
  have hne : ClassFunction.induce H (θ : ClassFunction H ℂ)
      ≠ ClassFunction.induce H ((θ : ClassFunction H ℂ).conj) :=
    fun heq => hnotreal (hconj_eq.trans heq.symm)
  -- Distinct irreducibles are orthogonal.
  have hii := irreducibleCharacter_inner_eq_ite
    (⟨ClassFunction.induce H (θ : ClassFunction H ℂ), hirr⟩ : IrreducibleCharacter G)
    ⟨ClassFunction.induce H ((θ : ClassFunction H ℂ).conj), hirr'⟩
  rwa [if_neg (fun h => hne (congrArg (fun c : IrreducibleCharacter G =>
    (c : ClassFunction G ℂ)) h))] at hii

omit hH [Fintype G] [Invertible (Nat.card G : ℂ)] in
/-- **Degree from restriction multiplicities** (the degree side of Clifford's theorem).  For
`H ⊴ G` and any class function `χ` of `G`, the degree `χ(1)` equals `∑_θ ⟨Res χ, θ⟩ · θ(1)` over
the irreducible characters `θ` of `H` — the Fourier expansion
(`sum_inner_irreducibleCharacter_smul`)
of `Res^G_H χ` evaluated at `1`.

Combined with single-orbit + common multiplicity
(`hasCommonRestrictionMultiplicity_of_singleOrbit`) and the orbit size
(`card_conjByOrbit_eq_index_inertia`), this is the degree statement of Clifford's theorem
`χ(1) = e · [G : I_G(θ)] · θ(1)`. -/
theorem apply_one_eq_sum_restrictionMultiplicity_mul [Fintype (IrreducibleCharacter ↥H)]
    (χ : ClassFunction G ℂ) :
    χ (1 : G) = ∑ θ : IrreducibleCharacter ↥H,
      ClassFunction.restrictionMultiplicity H χ (θ : ClassFunction ↥H ℂ)
        * (θ : ClassFunction ↥H ℂ) (1 : ↥H) := by
  classical
  have hsum : ∀ (s : Finset (IrreducibleCharacter ↥H))
      (F : IrreducibleCharacter ↥H → ClassFunction ↥H ℂ),
      (∑ θ ∈ s, F θ) (1 : ↥H) = ∑ θ ∈ s, (F θ) (1 : ↥H) := by
    intro s F
    induction s using Finset.induction_on with
    | empty => simp
    | insert a s ha ih =>
        rw [Finset.sum_insert ha, Finset.sum_insert ha, ClassFunction.add_apply, ih]
  have hexp := sum_inner_irreducibleCharacter_smul (G := ↥H) (ClassFunction.restrict H χ)
  have key := congrArg (fun f : ClassFunction ↥H ℂ => f (1 : ↥H)) hexp
  simp only [hsum, ClassFunction.smul_apply, ClassFunction.restrict_apply,
    OneMemClass.coe_one] at key
  simp only [ClassFunction.restrictionMultiplicity_def]
  exact key.symm

end Complex

section SignedIrreducible

variable {Γ : Type*} [Group Γ] [Fintype Γ] [Invertible (Nat.card Γ : ℂ)]

/-- A norm-`1` virtual character has nonzero degree: if `φ = ε • ξ` with `ε = ±1` and `ξ`
irreducible, then `φ(1) = ε · ξ(1) ≠ 0` (irreducible degrees are positive). -/
theorem apply_one_ne_zero_of_mem_ZIrr_of_inner_self_one
    {φ : ClassFunction Γ ℂ} (hφ : φ ∈ ZIrr Γ)
    (hφn : ClassFunction.inner φ φ = 1) : φ (1 : Γ) ≠ 0 := by
  obtain ⟨ε, ξ, hε, rfl⟩ := exists_zsmul_irreducibleCharacter_of_inner_self_one hφ hφn
  obtain ⟨d, hd, hd1⟩ := irreducibleCharacter_apply_one_eq_pos_natCast ξ
  rw [← Int.cast_smul_eq_zsmul ℂ ε (ξ : ClassFunction Γ ℂ), ClassFunction.smul_apply, hd1]
  refine mul_ne_zero ?_ ?_
  · rcases hε with h | h <;> subst h <;> norm_num
  · exact_mod_cast hd.ne'

/-- A norm-`1` virtual character has degree of absolute value at least `1`: `φ = ε • ξ` with
`ε = ±1` and `ξ` irreducible, so `‖φ(1)‖ = ξ(1) ≥ 1`.  The quantitative form of
`apply_one_ne_zero_of_mem_ZIrr_of_inner_self_one`, feeding the `‖φ(1)‖²/|G| ≥ 1/|G|` term of the
Peterfalvi (13.10.1)/(13.10.2) Parseval estimates. -/
theorem one_le_normSq_apply_one_of_mem_ZIrr_of_inner_self_one
    {φ : ClassFunction Γ ℂ} (hφ : φ ∈ ZIrr Γ)
    (hφn : ClassFunction.inner φ φ = 1) : 1 ≤ ‖φ (1 : Γ)‖ ^ 2 := by
  obtain ⟨ε, ξ, hε, rfl⟩ := exists_zsmul_irreducibleCharacter_of_inner_self_one hφ hφn
  obtain ⟨d, hd, hd1⟩ := irreducibleCharacter_apply_one_eq_pos_natCast ξ
  rw [← Int.cast_smul_eq_zsmul ℂ ε (ξ : ClassFunction Γ ℂ), ClassFunction.smul_apply, hd1,
    norm_mul]
  have hε1 : ‖(ε : ℂ)‖ = 1 := by rcases hε with h | h <;> subst h <;> norm_num
  rw [hε1, one_mul, Complex.norm_natCast]
  have hd' : (1 : ℝ) ≤ (d : ℝ) := by exact_mod_cast hd
  nlinarith

/-- Two signed irreducibles with nonzero inner product are equal up to that inner product:
`ψ = ⟨φ, ψ⟩ • φ`.  (Writing `φ = ε•a`, `ψ = ε'•b`, the inner product `εε'·(a = b)` is nonzero only
if `a = b`, in which case `⟨φ,ψ⟩ = εε'` and `εε'·ε = ε'` since `ε² = 1`.) -/
theorem eq_inner_smul_of_inner_ne_zero
    {φ ψ : ClassFunction Γ ℂ} (hφ : φ ∈ ZIrr Γ) (hψ : ψ ∈ ZIrr Γ)
    (hφn : ClassFunction.inner φ φ = 1) (hψn : ClassFunction.inner ψ ψ = 1)
    (hne : ClassFunction.inner φ ψ ≠ 0) :
    ψ = ClassFunction.inner φ ψ • φ := by
  classical
  obtain ⟨ε, a, hε, rfl⟩ := exists_zsmul_irreducibleCharacter_of_inner_self_one hφ hφn
  obtain ⟨ε', b, hε', rfl⟩ := exists_zsmul_irreducibleCharacter_of_inner_self_one hψ hψn
  have key : ClassFunction.inner (ε • (a : ClassFunction Γ ℂ)) (ε' • (b : ClassFunction Γ ℂ)) =
      (ε : ℂ) * (ε' : ℂ) *
        ClassFunction.inner (a : ClassFunction Γ ℂ) (b : ClassFunction Γ ℂ) := by
    rw [← Int.cast_smul_eq_zsmul ℂ ε (a : ClassFunction Γ ℂ),
      ← Int.cast_smul_eq_zsmul ℂ ε' (b : ClassFunction Γ ℂ),
      ClassFunction.inner_smul_left, inner_smul_right, star_intCast]
    ring
  have hab : a = b := by
    by_contra hab
    exact hne (by rw [key, irreducibleCharacter_inner, if_neg hab, mul_zero])
  subst hab
  rw [key, irreducibleCharacter_inner, if_pos rfl, mul_one,
    ← Int.cast_smul_eq_zsmul ℂ ε' (a : ClassFunction Γ ℂ),
    ← Int.cast_smul_eq_zsmul ℂ ε (a : ClassFunction Γ ℂ), smul_smul]
  congr 1
  rcases hε with h | h <;> subst h <;> push_cast <;> ring

/-- **Peterfalvi (4.1)** (mmd 04.6 L5), cross-orthogonality core.

For signed irreducibles `α, β, γ, δ ∈ ±Irr Γ` (norm-`1` elements of `ZIrr Γ`) and nonzero reals
`u, v`, if `(α,β) = (γ,δ) = 0`, the signed difference `(α − β, u•γ − v•δ) = 0`, and both signed
differences vanish at `1`, then `(α, γ) = 0`.

Proof (mmd): if `(α,γ) ≠ 0`, then `γ = εα` (`ε = ±1`); orthogonality of `γ,δ` forces `(α,δ) = 0`,
and `(α,β) = 0` forces `(β,γ) = 0`.  Expanding `(α−β, u•γ−v•δ) = 0` gives `uε + v(β,δ) = 0`, so
`(β,δ) ≠ 0` and `v•δ = −uε•β`.  Then `u•γ − v•δ = uε•(α+β)`, and evaluating at `1` (using
`(α−β)(1) = 0`, i.e. `α(1) = β(1)`) gives `0 = 2uεα(1)`, contradicting `u ≠ 0`, `ε ≠ 0`,
`α(1) ≠ 0`. -/
theorem inner_eq_zero_of_orthogonal_signedDifference
    {α β γ δ : ClassFunction Γ ℂ} {u v : ℝ} (hu : u ≠ 0) (_hv : v ≠ 0)
    (hα : α ∈ ZIrr Γ) (hαn : ClassFunction.inner α α = 1)
    (hβ : β ∈ ZIrr Γ) (hβn : ClassFunction.inner β β = 1)
    (hγ : γ ∈ ZIrr Γ) (hγn : ClassFunction.inner γ γ = 1)
    (hδ : δ ∈ ZIrr Γ) (hδn : ClassFunction.inner δ δ = 1)
    (hαβ : ClassFunction.inner α β = 0) (hγδ : ClassFunction.inner γ δ = 0)
    (hdiff : ClassFunction.inner (α - β) ((u : ℂ) • γ - (v : ℂ) • δ) = 0)
    (hα1 : (α - β) (1 : Γ) = 0)
    (hγδ1 : ((u : ℂ) • γ - (v : ℂ) • δ) (1 : Γ) = 0) :
    ClassFunction.inner α γ = 0 := by
  by_contra hαγ
  -- (1) `γ = ⟨α,γ⟩ • α`.
  have hγeq : γ = ClassFunction.inner α γ • α :=
    eq_inner_smul_of_inner_ne_zero hα hγ hαn hγn hαγ
  -- (2) `⟨α,δ⟩ = 0` (else `δ = ±α = ±γ`, contradicting `γ ⊥ δ`).
  have hαδ : ClassFunction.inner α δ = 0 := by
    by_contra hαδ
    have hδeq : δ = ClassFunction.inner α δ • α :=
      eq_inner_smul_of_inner_ne_zero hα hδ hαn hδn hαδ
    have hcontra : ClassFunction.inner γ δ =
        ClassFunction.inner α γ * star (ClassFunction.inner α δ) := by
      conv_lhs => rw [hγeq, hδeq]
      rw [ClassFunction.inner_smul_left, inner_smul_right, hαn, mul_one]
    exact (mul_ne_zero hαγ (star_ne_zero.mpr hαδ)) (hγδ ▸ hcontra).symm
  -- (3) `⟨β,γ⟩ = 0`.
  have hβγ : ClassFunction.inner β γ = 0 := by
    rw [hγeq, inner_smul_right, inner_conj_symm α β, hαβ, star_zero, mul_zero]
  -- (4) Expand the signed-difference orthogonality to `u⟨α,γ⟩ + v⟨β,δ⟩ = 0`.
  have hexpand : (u : ℂ) * ClassFunction.inner α γ + (v : ℂ) * ClassFunction.inner β δ = 0 := by
    rw [ClassFunction.inner_sub_left, ClassFunction.inner_sub_right, ClassFunction.inner_sub_right,
      inner_smul_right, inner_smul_right, inner_smul_right, inner_smul_right] at hdiff
    simp only [show star ((u : ℝ) : ℂ) = ((u : ℝ) : ℂ) from by simp,
      show star ((v : ℝ) : ℂ) = ((v : ℝ) : ℂ) from by simp, hαδ, hβγ, mul_zero] at hdiff
    linear_combination hdiff
  -- (5) `⟨β,δ⟩ ≠ 0`, so `δ = ⟨β,δ⟩ • β`.
  have hβδ : ClassFunction.inner β δ ≠ 0 := by
    intro h0
    rw [h0, mul_zero, add_zero] at hexpand
    exact (mul_ne_zero (Complex.ofReal_ne_zero.mpr hu) hαγ) hexpand
  have hδeq : δ = ClassFunction.inner β δ • β :=
    eq_inner_smul_of_inner_ne_zero hβ hδ hβn hδn hβδ
  -- (6) Evaluate at `1`.  `α(1) = β(1)` from `(α−β)(1) = 0`.
  have hαβ1 : α (1 : Γ) = β (1 : Γ) := by
    have h := hα1; rw [ClassFunction.sub_apply, sub_eq_zero] at h; exact h
  have hγ1 : γ (1 : Γ) = ClassFunction.inner α γ * α (1 : Γ) := by
    conv_lhs => rw [hγeq]
    rw [ClassFunction.smul_apply]
  have hδ1 : δ (1 : Γ) = ClassFunction.inner β δ * β (1 : Γ) := by
    conv_lhs => rw [hδeq]
    rw [ClassFunction.smul_apply]
  rw [ClassFunction.sub_apply, ClassFunction.smul_apply, ClassFunction.smul_apply, hγ1, hδ1,
    ← hαβ1] at hγδ1
  -- `hγδ1 : u·(⟨α,γ⟩·α(1)) − v·(⟨β,δ⟩·α(1)) = 0`.
  have hfinal : 2 * (u : ℂ) * ClassFunction.inner α γ * α (1 : Γ) = 0 := by
    linear_combination hγδ1 + α (1 : Γ) * hexpand
  -- (7) Contradiction: every factor is nonzero.
  refine (mul_ne_zero (mul_ne_zero (mul_ne_zero ?_ ?_) hαγ)
    (apply_one_ne_zero_of_mem_ZIrr_of_inner_self_one hα hαn)) hfinal
  · norm_num
  · exact Complex.ofReal_ne_zero.mpr hu

/-- **Peterfalvi (4.1)** (mmd 04.6 L5).  Signed irreducibles `α, β, γ, δ ∈ ±Irr Γ` with
`(α,β) = (γ,δ) = 0`, orthogonal signed difference `(α − β, u•γ − v•δ) = 0` (`u, v` nonzero reals)
and both signed differences vanishing at `1` are **pairwise orthogonal**: all four cross inner
products `(α,γ), (α,δ), (β,γ), (β,δ)` vanish.

Each follows from the cross-orthogonality core
`inner_eq_zero_of_orthogonal_signedDifference` applied to a sign-flipped / swapped instance
(`α ↔ β`, `γ ↔ δ` with `u ↔ v`); the permuted hypotheses are produced by conjugate symmetry
(`inner_conj_symm`) and `neg_sub`. -/
theorem pairwise_inner_eq_zero_of_orthogonal_signedDifference
    {α β γ δ : ClassFunction Γ ℂ} {u v : ℝ} (hu : u ≠ 0) (hv : v ≠ 0)
    (hα : α ∈ ZIrr Γ) (hαn : ClassFunction.inner α α = 1)
    (hβ : β ∈ ZIrr Γ) (hβn : ClassFunction.inner β β = 1)
    (hγ : γ ∈ ZIrr Γ) (hγn : ClassFunction.inner γ γ = 1)
    (hδ : δ ∈ ZIrr Γ) (hδn : ClassFunction.inner δ δ = 1)
    (hαβ : ClassFunction.inner α β = 0) (hγδ : ClassFunction.inner γ δ = 0)
    (hdiff : ClassFunction.inner (α - β) ((u : ℂ) • γ - (v : ℂ) • δ) = 0)
    (hα1 : (α - β) (1 : Γ) = 0)
    (hγδ1 : ((u : ℂ) • γ - (v : ℂ) • δ) (1 : Γ) = 0) :
    ClassFunction.inner α γ = 0 ∧ ClassFunction.inner α δ = 0 ∧
      ClassFunction.inner β γ = 0 ∧ ClassFunction.inner β δ = 0 := by
  -- Conjugate-symmetric companions of the orthogonality hypotheses.
  have hδγ : ClassFunction.inner δ γ = 0 := by rw [inner_conj_symm γ δ, hγδ, star_zero]
  have hβα : ClassFunction.inner β α = 0 := by rw [inner_conj_symm α β, hαβ, star_zero]
  -- The `β ↔ α` and `γ,u ↔ δ,v` sign-flipped signed differences (orthogonal + vanishing at `1`).
  have hdiffYX : ClassFunction.inner (α - β) ((v : ℂ) • δ - (u : ℂ) • γ) = 0 := by
    rw [← neg_sub ((u : ℂ) • γ) ((v : ℂ) • δ), ClassFunction.inner_neg_right, hdiff, neg_zero]
  have hdiffBA : ClassFunction.inner (β - α) ((u : ℂ) • γ - (v : ℂ) • δ) = 0 := by
    rw [← neg_sub α β, ClassFunction.inner_neg_left, hdiff, neg_zero]
  have hdiffBAYX : ClassFunction.inner (β - α) ((v : ℂ) • δ - (u : ℂ) • γ) = 0 := by
    rw [← neg_sub α β, ← neg_sub ((u : ℂ) • γ) ((v : ℂ) • δ),
      ClassFunction.inner_neg_left, ClassFunction.inner_neg_right, neg_neg, hdiff]
  have h1BA : (β - α) (1 : Γ) = 0 := by
    rw [← neg_sub α β, ClassFunction.neg_apply, hα1, neg_zero]
  have h1YX : ((v : ℂ) • δ - (u : ℂ) • γ) (1 : Γ) = 0 := by
    rw [← neg_sub ((u : ℂ) • γ) ((v : ℂ) • δ), ClassFunction.neg_apply, hγδ1, neg_zero]
  refine ⟨?_, ?_, ?_, ?_⟩
  · exact inner_eq_zero_of_orthogonal_signedDifference hu hv hα hαn hβ hβn hγ hγn hδ hδn
      hαβ hγδ hdiff hα1 hγδ1
  · exact inner_eq_zero_of_orthogonal_signedDifference hv hu hα hαn hβ hβn hδ hδn hγ hγn
      hαβ hδγ hdiffYX hα1 h1YX
  · exact inner_eq_zero_of_orthogonal_signedDifference hu hv hβ hβn hα hαn hγ hγn hδ hδn
      hβα hγδ hdiffBA h1BA hγδ1
  · exact inner_eq_zero_of_orthogonal_signedDifference hv hu hβ hβn hα hαn hδ hδn hγ hγn
      hβα hδγ hdiffBAYX h1BA h1YX

end SignedIrreducible

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/InducedIrreducible.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