Peterfalvi (3.6)–(3.8): grid-coefficient theory of the -grid
DefinitionOddOrder_Peterfalvi_S16_GridExpansionfeit-thompsonfinite-groupsgroup-theoryishida-odd-order
Formalizes Peterfalvi §3 (pp. 15-20) for the -grid of §13–§16: orthonormality, the relation (3.7), and the Bessel-type vanishing of (13.19.b).
Main declarations: eta_orthonormal (The η-grid is orthonormal); inner_eta_grid_relation (Peterfalvi (3.7)); inner_eta_eq_zero_of_vanish_of_inner_self_eq_two (Peterfalvi (13.19.b)-core); eta_diff_rigidity (Peterfalvi (3.8) difference rigidity for the η-grid).
The bundle also proves 5 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/S16_GridExpansion.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_S15_SAndT_Setup_HypothesisBasics
import Definitions.Def_OddOrder_Peterfalvi_S15_SAndT_Setup_HypothesisSwap
import Definitions.Def_OddOrder_Peterfalvi_S15_SAndT_Setup_DegreesFirstSplit
import Definitions.Def_OddOrder_Peterfalvi_S15_SAndT_Setup_PairStructure
import Definitions.Def_OddOrder_Peterfalvi_S15_SAndT_Setup_CountingLayer
import Definitions.Def_OddOrder_Peterfalvi_S15_SAndT_Setup_Canonicalization
import Definitions.Def_OddOrder_Peterfalvi_S13_ElementaryAbelianKernel
import Definitions.Def_OddOrder_BG_Ch2_Uniqueness_S07_Theorem74
import Definitions.Def_OddOrder_Peterfalvi_S15_SAndT_Setup_TSideDegrees
import Definitions.Def_OddOrder_Peterfalvi_S05_GridRigidity
namespace OddOrder.Peterfalvi.S16
open OddOrder.GroupTheory
open OddOrder.RepresentationTheory
variable {G : Type*} [Group G]
/-- **Peterfalvi (3.3), the trivial grid entry**: `ω₀₀ = 1_W`. Every `w ∈ W = W₁ ⊔ W₂`
splits as `w = x·y` with `x ∈ W₁`, `y ∈ W₂` (`W` cyclic hence abelian, so the join is the
product), and the row-`0`/column-`0` trivialities (issue-2033 grid semantics) kill both
factors. -/
theorem omega_principal_eq_trivial [Finite G]
(hyp : OddOrder.Peterfalvi.S15.Hypothesis (G := G)) :
hyp.omega ⟨0, hyp.q_prime.pos⟩ ⟨0, hyp.p_prime.pos⟩
= trivialClassFunction ↥hyp.W := by
let := hyp.W_cyclic
let : CommGroup ↥hyp.W := IsCyclic.commGroup
have hW1le : hyp.W1 ≤ hyp.W := by
rw [hyp.W_eq_join]; exact le_sup_left
have hW2le : hyp.W2 ≤ hyp.W := by
rw [hyp.W_eq_join]; exact le_sup_right
ext w
have hw : w ∈ (hyp.W1.subgroupOf hyp.W) ⊔ (hyp.W2.subgroupOf hyp.W) := by
have h1 : (hyp.W1 ⊔ hyp.W2).subgroupOf hyp.W = ⊤ := by
rw [← hyp.W_eq_join, Subgroup.subgroupOf_self]
rw [← Subgroup.subgroupOf_sup hW1le hW2le, h1]
exact Subgroup.mem_top w
obtain ⟨x, hxmem, y, hymem, hxy⟩ := Subgroup.mem_sup.mp hw
have hval : hyp.omega ⟨0, hyp.q_prime.pos⟩ ⟨0, hyp.p_prime.pos⟩ w = 1 := by
rw [← hxy, hyp.omega_mul,
hyp.omega_row_zero_apply_of_mem_W1 _ x (Subgroup.mem_subgroupOf.mp hxmem),
hyp.omega_col_zero_apply_of_mem_W2 _ y (Subgroup.mem_subgroupOf.mp hymem),
one_mul]
rw [hval, trivialClassFunction_apply]
/-- **Peterfalvi (3.3)/(13.1.d): the principal `η`-entry is the trivial character** —
`η₀₀ = τ₃(ω₀₀) = τ₃(1_W) = 1_G` (`omega_principal_eq_trivial` + `tau3_trivial`), as an
equality of class functions. -/
theorem eta_principal_eq_trivial [Finite G]
(hyp : OddOrder.Peterfalvi.S15.Hypothesis (G := G)) :
hyp.eta ⟨0, hyp.q_prime.pos⟩ ⟨0, hyp.p_prime.pos⟩ = trivialClassFunction G := by
rw [hyp.eta_eq_tau_omega, omega_principal_eq_trivial hyp, hyp.tau3_trivial]
/-- **Peterfalvi (3.3)/(13.1.d), pointwise form**: `η₀₀(g) = 1` for every `g ∈ G`. -/
theorem eta_principal_apply_eq_one [Finite G]
(hyp : OddOrder.Peterfalvi.S15.Hypothesis (G := G)) (g : G) :
hyp.eta ⟨0, hyp.q_prime.pos⟩ ⟨0, hyp.p_prime.pos⟩ g = 1 := by
rw [eta_principal_eq_trivial hyp, trivialClassFunction_apply]
/-- Each `η_ij = τ₃(ω_ij)` is a virtual character of `G` (`tau3_mem_ZIrr` on
`omega_mem_ZIrr`). -/
theorem eta_mem_ZIrr [Finite G]
(hyp : OddOrder.Peterfalvi.S15.Hypothesis (G := G))
(i : Fin hyp.q) (j : Fin hyp.p) :
hyp.eta i j ∈ ZIrr G := by
rw [hyp.eta_eq_tau_omega]
exact hyp.tau3_mem_ZIrr _ (hyp.omega_mem_ZIrr i j)
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **The `η`-grid is orthonormal**: `⟨η_ij, η_kl⟩ = δ_{(i,j),(k,l)}`, transported from the
`ω`-grid orthonormality (`omega_orthonormal`) through the `τ₃` isometry (`tau3_isometry`). -/
theorem eta_orthonormal [Finite G]
(hyp : OddOrder.Peterfalvi.S15.Hypothesis (G := G))
(i k : Fin hyp.q) (j l : Fin hyp.p) :
ClassFunction.inner (hyp.eta i j) (hyp.eta k l)
= if i = k ∧ j = l then 1 else 0 := by
rw [hyp.eta_eq_tau_omega, hyp.eta_eq_tau_omega,
hyp.tau3_isometry.inner_eq, hyp.omega_orthonormal]
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (3.7)** (the `(i',j') = (0,0)` instance, which generates the general four-index
relation by arithmetic): for a class function `φ` vanishing on the conjugacy saturation of the
regular set `Ŵ = W ∖ (W₁ ∪ W₂)`,
`⟨φ, η_ij⟩ + ⟨φ, η_00⟩ = ⟨φ, η_i0⟩ + ⟨φ, η_0j⟩`.
The four-corner combination `η_ij + η_00 − η_i0 − η_0j` vanishes off `Ŵ^G`
(`eta_fourcorner_vanish` + `η_00 = 1_G`), while `φ` vanishes on `Ŵ^G`, so every summand of the
inner product dies. -/
theorem inner_eta_grid_relation [Finite G]
(hyp : OddOrder.Peterfalvi.S15.Hypothesis (G := G)) {φ : ClassFunction G ℂ}
(hvanish : ∀ x ∈ conjClassSet
((hyp.W : Set G) \ ((hyp.W1 : Set G) ∪ (hyp.W2 : Set G))), φ x = 0)
(i : Fin hyp.q) (j : Fin hyp.p) :
ClassFunction.inner φ (hyp.eta i j)
+ ClassFunction.inner φ (hyp.eta ⟨0, hyp.q_prime.pos⟩ ⟨0, hyp.p_prime.pos⟩)
= ClassFunction.inner φ (hyp.eta i ⟨0, hyp.p_prime.pos⟩)
+ ClassFunction.inner φ (hyp.eta ⟨0, hyp.q_prime.pos⟩ j) := by
classical
by_cases hi : i = ⟨0, hyp.q_prime.pos⟩
· subst hi; ring
by_cases hj : j = ⟨0, hyp.p_prime.pos⟩
· subst hj; ring
-- the four-corner combination, supported on `Ŵ^G`
have hcombo : ∀ x : G, x ∉ conjClassSet
((hyp.W : Set G) \ ((hyp.W1 : Set G) ∪ (hyp.W2 : Set G))) →
(hyp.eta i j + hyp.eta ⟨0, hyp.q_prime.pos⟩ ⟨0, hyp.p_prime.pos⟩
- hyp.eta i ⟨0, hyp.p_prime.pos⟩ - hyp.eta ⟨0, hyp.q_prime.pos⟩ j) x = 0 := by
intro x hx
have h4 := hyp.eta_fourcorner_vanish i j hi hj x hx
rw [ClassFunction.sub_apply, ClassFunction.sub_apply, ClassFunction.add_apply,
eta_principal_apply_eq_one hyp x]
rw [hyp.eta_eq_tau_omega i j, hyp.eta_eq_tau_omega i ⟨0, hyp.p_prime.pos⟩,
hyp.eta_eq_tau_omega ⟨0, hyp.q_prime.pos⟩ j]
linear_combination h4
-- `⟨φ, c⟩ = 0`: every summand of the inner sum has a vanishing factor
have hinner : ClassFunction.inner φ
(hyp.eta i j + hyp.eta ⟨0, hyp.q_prime.pos⟩ ⟨0, hyp.p_prime.pos⟩
- hyp.eta i ⟨0, hyp.p_prime.pos⟩ - hyp.eta ⟨0, hyp.q_prime.pos⟩ j) = 0 := by
rw [ClassFunction.inner_eq_inv_card_mul_innerSum]
have hsum : ClassFunction.innerSum φ
(hyp.eta i j + hyp.eta ⟨0, hyp.q_prime.pos⟩ ⟨0, hyp.p_prime.pos⟩
- hyp.eta i ⟨0, hyp.p_prime.pos⟩ - hyp.eta ⟨0, hyp.q_prime.pos⟩ j) = 0 := by
rw [ClassFunction.innerSum]
refine Finset.sum_eq_zero fun x _ => ?_
by_cases hx : x ∈ conjClassSet
((hyp.W : Set G) \ ((hyp.W1 : Set G) ∪ (hyp.W2 : Set G)))
· rw [hvanish x hx, zero_mul]
· rw [hcombo x hx, star_zero, mul_zero]
rw [hsum, mul_zero]
-- expand by conjugate-linearity in the second argument
rw [ClassFunction.inner_sub_right, ClassFunction.inner_sub_right,
ClassFunction.inner_add_right] at hinner
linear_combination hinner
/-- **Peterfalvi (3.8), the small-support case** (pure combinatorics): a `q × p` grid `a`
(`3 ≤ q`, `3 ≤ p`) satisfying the (3.7) relation `a_ij + a_{i₀j₀} = a_{ij₀} + a_{i₀j}`
relative to some base point `(i₀, j₀)`, with at most `2` nonzero entries, is identically
zero. In Peterfalvi's trichotomy the row/column cases (3.8.b)/(3.8.c) force at least
`min q p ≥ 3` nonzero entries, so only case (3.8.a) survives.
The proof mirrors the rank-one shift structure: an all-zero row exists (else every row
contributes a nonzero entry, `NC ≥ q ≥ 3`), which makes every row constant
(`a_ij = a_{ij₀} − a_{i₁j₀}`); a row with nonzero constant would contribute `p ≥ 3` nonzero
entries. -/
theorem grid_eq_zero_of_relation_of_card_le_two {q p : ℕ} (hq : 3 ≤ q) (hp : 3 ≤ p)
{a : Fin q → Fin p → ℂ} (i0 : Fin q) (j0 : Fin p)
(hrel : ∀ i j, a i j + a i0 j0 = a i j0 + a i0 j)
(hNC : (Finset.univ.filter fun x : Fin q × Fin p => a x.1 x.2 ≠ 0).card ≤ 2) :
∀ i j, a i j = 0 := by
classical
set S : Finset (Fin q × Fin p) :=
Finset.univ.filter fun x : Fin q × Fin p => a x.1 x.2 ≠ 0 with hS
-- an all-zero row exists
have hrow : ∃ i₁ : Fin q, ∀ j, a i₁ j = 0 := by
by_contra hno
push Not at hno
choose f hf using hno
have hcard : q ≤ S.card := by
have hmaps : ∀ i ∈ (Finset.univ : Finset (Fin q)),
((i, f i) : Fin q × Fin p) ∈ S := fun i _ =>
Finset.mem_filter.mpr ⟨Finset.mem_univ _, hf i⟩
have := Finset.card_le_card_of_injOn
(fun i : Fin q => ((i, f i) : Fin q × Fin p)) hmaps
(fun i _ i' _ h => (Prod.ext_iff.mp h).1)
simpa using this
omega
obtain ⟨i₁, hi₁⟩ := hrow
-- every row is constant: `a i j = a i j₀ − a i₁ j₀`
have hconst : ∀ i j, a i j = a i j0 - a i₁ j0 := by
intro i j
have h1 := hrel i j
have h2 := hrel i₁ j
rw [hi₁ j] at h2
linear_combination h1 - h2
-- a row with nonzero constant contributes `p ≥ 3` nonzero entries
intro i j
rw [hconst i j]
by_contra hne
have hcard : p ≤ S.card := by
have hmaps : ∀ j' ∈ (Finset.univ : Finset (Fin p)),
((i, j') : Fin q × Fin p) ∈ S := fun j' _ =>
Finset.mem_filter.mpr ⟨Finset.mem_univ _, by rw [hconst i j']; exact hne⟩
have := Finset.card_le_card_of_injOn
(fun j' : Fin p => ((i, j') : Fin q × Fin p)) hmaps
(fun j' _ j'' _ h => (Prod.ext_iff.mp h).2)
simpa using this
omega
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (13.19.b)-core** ((3.8) applied through Bessel): a virtual character
`φ ∈ ℤ[Irr G]` with `⟨φ, φ⟩ = 2` vanishing on the conjugacy saturation of the regular set
`Ŵ = W ∖ (W₁ ∪ W₂)` is orthogonal to the entire `η`-grid.
This is Peterfalvi's `NC(ψ) ≤ ‖ψ‖² = 2 < 2q` step: the grid coefficients
`a_ij = ⟨φ, η_ij⟩` are integers (`inner_mem_ZIrr_int`), the orthogonal projection onto the
(orthonormal) grid satisfies Bessel `∑ a_ij² ≤ ⟨φ, φ⟩ = 2`, so at most two coefficients are
nonzero; the (3.7) relation (`inner_eta_grid_relation`) then forces all of them to vanish
(`grid_eq_zero_of_relation_of_card_le_two`, using `3 ≤ q` and `3 ≤ p`).
Consumed with `φ = (ψ − ψ̄)^τ` for an irreducible `ψ` of a type-I family (Peterfalvi
(13.19.b), the `L^{τ₁} ⊥ η_ij` fact behind the (14.11.2)/(14.16) signed expansions). -/
theorem inner_eta_eq_zero_of_vanish_of_inner_self_eq_two [Finite G]
(hyp : OddOrder.Peterfalvi.S15.Hypothesis (G := G)) {φ : ClassFunction G ℂ}
(hφZ : φ ∈ ZIrr G)
(hvanish : ∀ x ∈ conjClassSet
((hyp.W : Set G) \ ((hyp.W1 : Set G) ∪ (hyp.W2 : Set G))), φ x = 0)
(hnorm : ClassFunction.inner φ φ = 2) :
∀ (i : Fin hyp.q) (j : Fin hyp.p), ClassFunction.inner φ (hyp.eta i j) = 0 := by
classical
-- integer coefficient grid
have hint : ∀ (i : Fin hyp.q) (j : Fin hyp.p),
∃ m : ℤ, ClassFunction.inner φ (hyp.eta i j) = (m : ℂ) := fun i j =>
ClassFunction.inner_mem_ZIrr_int hφZ (eta_mem_ZIrr hyp i j)
choose m hm using hint
-- the grid projection and its residual
set X : ClassFunction G ℂ :=
∑ i : Fin hyp.q, ∑ j : Fin hyp.p, (m i j : ℂ) • hyp.eta i j with hX
set Y : ClassFunction G ℂ := φ - X with hY
-- `⟨X, η_kl⟩ = m k l` by orthonormality
have hXeta : ∀ (k : Fin hyp.q) (l : Fin hyp.p),
ClassFunction.inner X (hyp.eta k l) = (m k l : ℂ) := by
intro k l
rw [hX, inner_sum_left]
rw [Finset.sum_eq_single_of_mem k (Finset.mem_univ _) (fun i _ hik => ?_)]
· rw [inner_sum_left,
Finset.sum_eq_single_of_mem l (Finset.mem_univ _) (fun j _ hjl => ?_)]
· rw [ClassFunction.inner_smul_left, eta_orthonormal hyp,
if_pos ⟨rfl, rfl⟩, mul_one]
· rw [ClassFunction.inner_smul_left, eta_orthonormal hyp,
if_neg (by rintro ⟨-, rfl⟩; exact hjl rfl), mul_zero]
· rw [inner_sum_left]
refine Finset.sum_eq_zero fun j _ => ?_
rw [ClassFunction.inner_smul_left, eta_orthonormal hyp,
if_neg (by rintro ⟨rfl, -⟩; exact hik rfl), mul_zero]
-- `⟨Y, η_kl⟩ = 0`
have hYeta : ∀ (k : Fin hyp.q) (l : Fin hyp.p),
ClassFunction.inner Y (hyp.eta k l) = 0 := by
intro k l
rw [hY, ClassFunction.inner_sub_left, hXeta k l, hm k l, sub_self]
-- `⟨φ, X⟩ = ∑ m²` and `⟨X, X⟩ = ∑ m²`
have hsum_sq : ∀ ψ : ClassFunction G ℂ,
(∀ (k : Fin hyp.q) (l : Fin hyp.p),
ClassFunction.inner ψ (hyp.eta k l) = (m k l : ℂ)) →
ClassFunction.inner ψ X
= ((∑ i : Fin hyp.q, ∑ j : Fin hyp.p, (m i j) ^ 2 : ℤ) : ℂ) := by
intro ψ hψ
rw [hX, inner_sum_right]
push_cast
refine Finset.sum_congr rfl fun i _ => ?_
rw [inner_sum_right]
refine Finset.sum_congr rfl fun j _ => ?_
rw [ClassFunction.inner_smul_right, hψ i j, star_intCast]
ring
-- `⟨X, Y⟩ = 0` and `⟨Y, X⟩ = 0`
have hXY : ClassFunction.inner X Y = 0 := by
have h := hsum_sq X hXeta
have h2 : ClassFunction.inner X φ
= ((∑ i : Fin hyp.q, ∑ j : Fin hyp.p, (m i j) ^ 2 : ℤ) : ℂ) := by
rw [OddOrder.RepresentationTheory.inner_conj_symm, hsum_sq φ hm]
rw [star_intCast]
rw [hY, ClassFunction.inner_sub_right, h2, h, sub_self]
have hYX : ClassFunction.inner Y X = 0 := by
rw [OddOrder.RepresentationTheory.inner_conj_symm, hXY, star_zero]
-- Pythagorean split: `⟨φ, φ⟩ = ∑ m² + ⟨Y, Y⟩`
have hsplit : ClassFunction.inner φ φ
= ((∑ i : Fin hyp.q, ∑ j : Fin hyp.p, (m i j) ^ 2 : ℤ) : ℂ)
+ ClassFunction.inner Y Y := by
have hφXY : φ = X + Y := by rw [hY]; abel
calc ClassFunction.inner φ φ
= ClassFunction.inner (X + Y) (X + Y) := by rw [← hφXY]
_ = ClassFunction.inner X X + ClassFunction.inner X Y
+ (ClassFunction.inner Y X + ClassFunction.inner Y Y) := by
rw [ClassFunction.inner_add_left, ClassFunction.inner_add_right,
ClassFunction.inner_add_right]
_ = ((∑ i : Fin hyp.q, ∑ j : Fin hyp.p, (m i j) ^ 2 : ℤ) : ℂ)
+ ClassFunction.inner Y Y := by
rw [hXY, hYX, hsum_sq X hXeta]; ring
-- `⟨Y, Y⟩` is a nonnegative real, so `∑ m² ≤ 2` over `ℤ`
have hYY : ∃ r : ℝ, 0 ≤ r ∧ ClassFunction.inner Y Y = (r : ℂ) := by
refine ⟨(Nat.card G : ℝ)⁻¹ * ∑ g : G, Complex.normSq (Y g), ?_,
inner_self_eq_realCast Y⟩
have hsum : (0 : ℝ) ≤ ∑ g : G, Complex.normSq (Y g) :=
Finset.sum_nonneg fun g _ => Complex.normSq_nonneg _
positivity
obtain ⟨r, hr0, hr⟩ := hYY
have hsq_le : ∑ i : Fin hyp.q, ∑ j : Fin hyp.p, (m i j) ^ 2 ≤ 2 := by
have h2 : ((∑ i : Fin hyp.q, ∑ j : Fin hyp.p, (m i j) ^ 2 : ℤ) : ℂ) + (r : ℂ)
= (2 : ℂ) := by rw [← hr, ← hsplit, hnorm]
have hreal : ((∑ i : Fin hyp.q, ∑ j : Fin hyp.p, (m i j) ^ 2 : ℤ) : ℝ) + r
= 2 := by
have hre := congrArg Complex.re h2
simp only [Complex.add_re, Complex.intCast_re, Complex.ofReal_re] at hre
norm_num at hre
exact_mod_cast hre
have : ((∑ i : Fin hyp.q, ∑ j : Fin hyp.p, (m i j) ^ 2 : ℤ) : ℝ) ≤ 2 := by
linarith
exact_mod_cast this
-- at most two nonzero coefficients
have hNC : (Finset.univ.filter
fun x : Fin hyp.q × Fin hyp.p => ((m x.1 x.2 : ℂ) ≠ 0)).card ≤ 2 := by
have hcard_le : ((Finset.univ.filter
fun x : Fin hyp.q × Fin hyp.p => ((m x.1 x.2 : ℂ) ≠ 0)).card : ℤ)
≤ ∑ x : Fin hyp.q × Fin hyp.p, (m x.1 x.2) ^ 2 :=
calc ((Finset.univ.filter
fun x : Fin hyp.q × Fin hyp.p => ((m x.1 x.2 : ℂ) ≠ 0)).card : ℤ)
= ∑ x ∈ Finset.univ.filter
(fun x : Fin hyp.q × Fin hyp.p => ((m x.1 x.2 : ℂ) ≠ 0)), 1 := by
simp
_ ≤ ∑ x ∈ Finset.univ.filter
(fun x : Fin hyp.q × Fin hyp.p => ((m x.1 x.2 : ℂ) ≠ 0)),
(m x.1 x.2) ^ 2 := by
refine Finset.sum_le_sum fun x hx => ?_
have hne : m x.1 x.2 ≠ 0 := by
intro h0
exact (Finset.mem_filter.mp hx).2 (by rw [h0, Int.cast_zero])
nlinarith [Int.one_le_abs hne, sq_abs (m x.1 x.2)]
_ ≤ ∑ x : Fin hyp.q × Fin hyp.p, (m x.1 x.2) ^ 2 :=
Finset.sum_le_sum_of_subset_of_nonneg (Finset.filter_subset _ _)
(fun x _ _ => sq_nonneg _)
have hprod : ∑ x : Fin hyp.q × Fin hyp.p, (m x.1 x.2) ^ 2
= ∑ i : Fin hyp.q, ∑ j : Fin hyp.p, (m i j) ^ 2 := by
rw [Fintype.sum_prod_type]
omega
-- the (3.7) relation on the coefficient grid, then rigidity
have hrel : ∀ (i : Fin hyp.q) (j : Fin hyp.p),
(m i j : ℂ) + (m ⟨0, hyp.q_prime.pos⟩ ⟨0, hyp.p_prime.pos⟩ : ℂ)
= (m i ⟨0, hyp.p_prime.pos⟩ : ℂ) + (m ⟨0, hyp.q_prime.pos⟩ j : ℂ) := by
intro i j
have h := inner_eta_grid_relation hyp hvanish i j
rw [hm, hm, hm, hm] at h
exact h
have hzero := grid_eq_zero_of_relation_of_card_le_two
hyp.three_le_q hyp.three_le_p
⟨0, hyp.q_prime.pos⟩ ⟨0, hyp.p_prime.pos⟩ hrel hNC
intro i j
rw [hm i j, hzero i j]
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (3.8) difference rigidity for the `η`-grid** (`eq_signed_sub_cTIiso`, Coq
`PFsection3.v`). A virtual character `X ∈ ℤ[Irr G]` with `‖X‖² = 2` that agrees with a signed
row-difference `s·(η_{i₀j₁} − η_{i₀j₂})` (`s = ±1`, `j₁ ≠ j₂`) on the conjugacy saturation of the
regular set `Ŵ = W ∖ (W₁ ∪ W₂)` equals it: `X = s·(η_{i₀j₁} − η_{i₀j₂})`.
Where `inner_eta_eq_zero_of_vanish_of_inner_self_eq_two` handles the `NC ≤ 2` *all-zero* case, this
is the `NC(ψ) ≤ 4` *difference* case: it feeds the abstract engine
`S05.orthonormalGrid_diff_rigidity` (full (3.8) trichotomy, constant-row/column excluded) with the
`η`-grid data — orthonormality (`eta_orthonormal`), virtual-character membership (`eta_mem_ZIrr`),
the coprime-odd axis sizes `q, p ≥ 3`, and the (3.7) separability of
`ψ = X − s·(η_{i₀j₁} − η_{i₀j₂})` (assembled from four instances of `inner_eta_grid_relation`,
valid because `ψ` vanishes on `Ŵ^G`).
This is the row-`0` shape (`i₀ = 0`) behind the (13.18) `S`-side cross-relation
`τ_S(μ_{0j} − μ_{01}) = η_{0j} − η_{01}`, and the general row/column shape used by Peterfalvi
(4.8)/(10.5)/(10.10)/(11.8). -/
theorem eta_diff_rigidity [Finite G]
(hyp : OddOrder.Peterfalvi.S15.Hypothesis (G := G))
{X : ClassFunction G ℂ} (hXZ : X ∈ ZIrr G) (hX2 : ClassFunction.inner X X = 2)
(i0 : Fin hyp.q) {j1 j2 : Fin hyp.p} (hj : j1 ≠ j2) {s : ℤ} (hs : s = 1 ∨ s = -1)
(hvanish : ∀ x ∈ conjClassSet
((hyp.W : Set G) \ ((hyp.W1 : Set G) ∪ (hyp.W2 : Set G))),
(X - (s : ℂ) • (hyp.eta i0 j1 - hyp.eta i0 j2)) x = 0) :
X = (s : ℂ) • (hyp.eta i0 j1 - hyp.eta i0 j2) := by
classical
have hcardq : Nat.card (Fin hyp.q) = hyp.q :=
Nat.card_eq_fintype_card.trans (Fintype.card_fin _)
have hcardp : Nat.card (Fin hyp.p) = hyp.p :=
Nat.card_eq_fintype_card.trans (Fintype.card_fin _)
-- (3.7) separability of the `ψ`-coefficient grid, from four `inner_eta_grid_relation` instances.
have hsep : ∀ (i i' : Fin hyp.q) (j j' : Fin hyp.p),
ClassFunction.inner (X - (s : ℂ) • (hyp.eta i0 j1 - hyp.eta i0 j2))
((fun pq : Fin hyp.q × Fin hyp.p => hyp.eta pq.1 pq.2) (i, j))
+ ClassFunction.inner (X - (s : ℂ) • (hyp.eta i0 j1 - hyp.eta i0 j2))
((fun pq : Fin hyp.q × Fin hyp.p => hyp.eta pq.1 pq.2) (i', j'))
= ClassFunction.inner (X - (s : ℂ) • (hyp.eta i0 j1 - hyp.eta i0 j2))
((fun pq : Fin hyp.q × Fin hyp.p => hyp.eta pq.1 pq.2) (i, j'))
+ ClassFunction.inner (X - (s : ℂ) • (hyp.eta i0 j1 - hyp.eta i0 j2))
((fun pq : Fin hyp.q × Fin hyp.p => hyp.eta pq.1 pq.2) (i', j)) := by
intro i i' j j'
have h1 := inner_eta_grid_relation hyp hvanish i j
have h2 := inner_eta_grid_relation hyp hvanish i' j'
have h3 := inner_eta_grid_relation hyp hvanish i j'
have h4 := inner_eta_grid_relation hyp hvanish i' j
simp only
linear_combination h1 + h2 - h3 - h4
have hmain := OddOrder.Peterfalvi.S05.orthonormalGrid_diff_rigidity
(fun pq : Fin hyp.q × Fin hyp.p => hyp.eta pq.1 pq.2)
(fun pq => eta_mem_ZIrr hyp pq.1 pq.2)
(fun a => by simpa using eta_orthonormal hyp a.1 a.1 a.2 a.2)
(fun a b hab => by
rw [eta_orthonormal hyp a.1 b.1 a.2 b.2, if_neg ?_]
rintro ⟨h1, h2⟩; exact hab (Prod.ext h1 h2))
(by rw [hcardq]; exact hyp.three_le_q) (by rw [hcardp]; exact hyp.three_le_p)
(by rw [hcardq]; exact hyp.q_odd) (by rw [hcardp]; exact hyp.p_odd)
(by rw [hcardq, hcardp]
exact (Nat.coprime_primes hyp.q_prime hyp.p_prime).mpr (Ne.symm hyp.p_ne_q))
hXZ hX2 (P1 := (i0, j1)) (P2 := (i0, j2))
(by intro h; exact hj (Prod.ext_iff.mp h).2) hs hsep
simpa using hmain
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (3.8) difference rigidity, column form** (issue 2035 #86, the de-swap of the
`T`-side cross-relation): the same abstract engine as `eta_diff_rigidity`
(`S05.orthonormalGrid_diff_rigidity` takes arbitrary distinct grid points), instantiated at a
fixed *column* `j₀` with distinct rows `i₁ ≠ i₂`. This is the `T`-side
column-fixed/row-difference shape consumed by `tauT_nu_cross` directly at `hyp` — previously
obtained by transposing through an older asymmetric `Hypothesis.swap` interface. -/
theorem eta_diff_rigidity_col [Finite G]
(hyp : OddOrder.Peterfalvi.S15.Hypothesis (G := G))
{X : ClassFunction G ℂ} (hXZ : X ∈ ZIrr G) (hX2 : ClassFunction.inner X X = 2)
{i1 i2 : Fin hyp.q} (hi : i1 ≠ i2) (j0 : Fin hyp.p) {s : ℤ} (hs : s = 1 ∨ s = -1)
(hvanish : ∀ x ∈ conjClassSet
((hyp.W : Set G) \ ((hyp.W1 : Set G) ∪ (hyp.W2 : Set G))),
(X - (s : ℂ) • (hyp.eta i1 j0 - hyp.eta i2 j0)) x = 0) :
X = (s : ℂ) • (hyp.eta i1 j0 - hyp.eta i2 j0) := by
classical
have hcardq : Nat.card (Fin hyp.q) = hyp.q :=
Nat.card_eq_fintype_card.trans (Fintype.card_fin _)
have hcardp : Nat.card (Fin hyp.p) = hyp.p :=
Nat.card_eq_fintype_card.trans (Fintype.card_fin _)
have hsep : ∀ (i i' : Fin hyp.q) (j j' : Fin hyp.p),
ClassFunction.inner (X - (s : ℂ) • (hyp.eta i1 j0 - hyp.eta i2 j0))
((fun pq : Fin hyp.q × Fin hyp.p => hyp.eta pq.1 pq.2) (i, j))
+ ClassFunction.inner (X - (s : ℂ) • (hyp.eta i1 j0 - hyp.eta i2 j0))
((fun pq : Fin hyp.q × Fin hyp.p => hyp.eta pq.1 pq.2) (i', j'))
= ClassFunction.inner (X - (s : ℂ) • (hyp.eta i1 j0 - hyp.eta i2 j0))
((fun pq : Fin hyp.q × Fin hyp.p => hyp.eta pq.1 pq.2) (i, j'))
+ ClassFunction.inner (X - (s : ℂ) • (hyp.eta i1 j0 - hyp.eta i2 j0))
((fun pq : Fin hyp.q × Fin hyp.p => hyp.eta pq.1 pq.2) (i', j)) := by
intro i i' j j'
have h1 := inner_eta_grid_relation hyp hvanish i j
have h2 := inner_eta_grid_relation hyp hvanish i' j'
have h3 := inner_eta_grid_relation hyp hvanish i j'
have h4 := inner_eta_grid_relation hyp hvanish i' j
simp only
linear_combination h1 + h2 - h3 - h4
have hmain := OddOrder.Peterfalvi.S05.orthonormalGrid_diff_rigidity
(fun pq : Fin hyp.q × Fin hyp.p => hyp.eta pq.1 pq.2)
(fun pq => eta_mem_ZIrr hyp pq.1 pq.2)
(fun a => by simpa using eta_orthonormal hyp a.1 a.1 a.2 a.2)
(fun a b hab => by
rw [eta_orthonormal hyp a.1 b.1 a.2 b.2, if_neg ?_]
rintro ⟨h1, h2⟩; exact hab (Prod.ext h1 h2))
(by rw [hcardq]; exact hyp.three_le_q) (by rw [hcardp]; exact hyp.three_le_p)
(by rw [hcardq]; exact hyp.q_odd) (by rw [hcardp]; exact hyp.p_odd)
(by rw [hcardq, hcardp]
exact (Nat.coprime_primes hyp.q_prime hyp.p_prime).mpr (Ne.symm hyp.p_ne_q))
hXZ hX2 (P1 := (i1, j0)) (P2 := (i2, j0))
(by intro h; exact hi (Prod.ext_iff.mp h).1) hs hsep
simpa using hmain
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **A norm-one virtual character is a signed irreducible.** If `φ ∈ ℤ[Irr G]` has
`⟨φ, φ⟩ = 1`, then `φ = ε • χ` for a single irreducible `χ` and a sign `ε = ±1`. Its
integer Fourier coefficients sum of squares equals `1`, so exactly one is nonzero and it
is `±1`. This is the `dirr` primitive used to turn the (3.8) rigidity output
`⟨ψ^{τ₁} − ψ̄^{τ₁}, η_ij⟩ = 0` into the individual orthogonality `⟨η_ij, ψ^{τ₁}⟩ = 0`. -/
theorem exists_sign_irr_of_inner_self_one [Finite G] {φ : ClassFunction G ℂ}
(hφ : φ ∈ ZIrr G) (hnorm : ClassFunction.inner φ φ = 1) :
∃ (χ : IrreducibleCharacter G) (ε : ℤ), (ε = 1 ∨ ε = -1) ∧
φ = (ε : ℂ) • (χ : ClassFunction G ℂ) := by
classical
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
have hge : ∀ a ∈ c.support, 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 hcard1 : c.support.card = 1 := by
have hle : (c.support.card : ℤ) ≤ ∑ a ∈ c.support, c a ^ 2 :=
calc (c.support.card : ℤ) = ∑ _a ∈ c.support, (1 : ℤ) := by
rw [Finset.sum_const, nsmul_eq_mul, mul_one]
_ ≤ ∑ a ∈ c.support, c a ^ 2 := Finset.sum_le_sum hge
rw [hsumZ] at hle
have hpos : 0 < c.support.card := by
rw [Finset.card_pos]
by_contra hempty
rw [Finset.not_nonempty_iff_eq_empty] at hempty
rw [hempty, Finset.sum_empty] at hsumZ
exact absurd hsumZ (by norm_num)
omega
obtain ⟨α₀, hα0eq⟩ := Finset.card_eq_one.mp hcard1
have hα0mem : α₀ ∈ irreducibleCharacters G := hsupp (by rw [hα0eq]; simp)
rw [hα0eq, Finset.sum_singleton] at hrepr
rw [hα0eq, Finset.sum_singleton] at hsumZ
have hcα : c α₀ = 1 ∨ c α₀ = -1 := by
rw [pow_two] at hsumZ; exact mul_self_eq_one_iff.mp hsumZ
exact ⟨⟨α₀, hα0mem⟩, c α₀, hcα, by rw [hrepr]⟩
open scoped OddOrder.Peterfalvi.S12.FiniteInduce in
/-- **Peterfalvi (13.19.b), the cross-group grid orthogonality** (engine + `dirr` finish;
the Dade support-vanishing is the one named §13/§14 hypothesis).
Let `ψ^{τ₁}` (`psi`) and `ψ̄^{τ₁}` (`psiConj`) be two distinct norm-one virtual characters of
`G` (the `τ₁`-images of an irreducible `ψ ∈ ℒ` of a type-`I` maximal subgroup and its complex
conjugate; distinctness `⟨ψ^{τ₁}, ψ̄^{τ₁}⟩ = 0` reflects `ψ ≠ ψ̄`). If the difference
`ψ^{τ₁} − ψ̄^{τ₁}` vanishes on the conjugacy saturation of the regular set `Ŵ = W ∖ (W₁ ∪ W₂)`
(Peterfalvi (13.19.a): `Ã(L)` avoids `P^G ∪ W^G`, so `(ψ − ψ̄)^τ` is supported off `Ŵ^G`),
then `ψ^{τ₁}` is orthogonal to the whole `η`-grid.
The difference has squared norm `2`, so the (3.8) rigidity engine gives
`⟨ψ^{τ₁} − ψ̄^{τ₁}, η_ij⟩ = 0`; the `dirr` finish (each of `ψ^{τ₁}`, `ψ̄^{τ₁}`, `η_ij` is a
signed irreducible, `exists_sign_irr_of_inner_self_one`) upgrades this to the individual
orthogonality: a nonzero `⟨η_ij, ψ^{τ₁}⟩` would force `ψ^{τ₁}` and `ψ̄^{τ₁}` to share the
constituent `χ_{η_ij}`, contradicting `ψ^{τ₁} ≠ ψ̄^{τ₁}`. -/
theorem eta_orthogonal_of_norm_one_pair_vanish [Finite G]
(hyp : OddOrder.Peterfalvi.S15.Hypothesis (G := G))
{psi psiConj : ClassFunction G ℂ}
(hpsiZ : psi ∈ ZIrr G) (hconjZ : psiConj ∈ ZIrr G)
(hpsi1 : ClassFunction.inner psi psi = 1)
(hconj1 : ClassFunction.inner psiConj psiConj = 1)
(hcross : ClassFunction.inner psi psiConj = 0)
(hvanish : ∀ x ∈ conjClassSet
((hyp.W : Set G) \ ((hyp.W1 : Set G) ∪ (hyp.W2 : Set G))), (psi - psiConj) x = 0) :
∀ (i : Fin hyp.q) (j : Fin hyp.p),
ClassFunction.inner (hyp.eta i j) psi = 0 := by
classical
-- the difference `Ψ = ψ^{τ₁} − ψ̄^{τ₁}` is a norm-2 virtual character
have hPsiZ : psi - psiConj ∈ ZIrr G := (ZIrr G).sub_mem hpsiZ hconjZ
have hcross' : ClassFunction.inner psiConj psi = 0 := by
rw [OddOrder.RepresentationTheory.inner_conj_symm, hcross, star_zero]
have hPsi2 : ClassFunction.inner (psi - psiConj) (psi - psiConj) = 2 := by
rw [ClassFunction.inner_sub_left, ClassFunction.inner_sub_right,
ClassFunction.inner_sub_right, hpsi1, hconj1, hcross, hcross']
ring
-- engine: `Ψ ⊥ η_ij` for all `i, j`
have hengine := inner_eta_eq_zero_of_vanish_of_inner_self_eq_two hyp hPsiZ hvanish hPsi2
-- signed-irreducible decompositions of `ψ^{τ₁}`, `ψ̄^{τ₁}`
obtain ⟨χp, εp, hεp, hpeq⟩ := exists_sign_irr_of_inner_self_one hpsiZ hpsi1
obtain ⟨χc, εc, hεc, hceq⟩ := exists_sign_irr_of_inner_self_one hconjZ hconj1
-- distinct constituents: `χp ≠ χc`
have hp0 : (εp : ℂ) ≠ 0 := by rcases hεp with h | h <;> rw [h] <;> norm_num
have hc0 : (εc : ℂ) ≠ 0 := by rcases hεc with h | h <;> rw [h] <;> norm_num
have hχne : χp ≠ χc := by
intro heq
have hval2 : ClassFunction.inner psi psiConj = (εp : ℂ) * (εc : ℂ) := by
rw [hpeq, hceq, ClassFunction.inner_smul_left,
OddOrder.RepresentationTheory.inner_smul_right,
OddOrder.RepresentationTheory.irreducibleCharacter_inner_eq_ite, if_pos heq,
star_intCast, mul_one]
rw [hval2] at hcross
exact (mul_ne_zero hp0 hc0) hcross
-- individual orthogonality
intro i j
-- `η_ij` is a signed irreducible
obtain ⟨χη, εη, hεη, hηeq⟩ := exists_sign_irr_of_inner_self_one (eta_mem_ZIrr hyp i j)
(by simpa using eta_orthonormal hyp i i j j)
-- `⟨η_ij, ψ^{τ₁}⟩ = ε_η ε_p · [χ_η = χ_p]`
have hval : ∀ (χ : IrreducibleCharacter G) (ε : ℤ),
ClassFunction.inner (hyp.eta i j) ((ε : ℂ) • (χ : ClassFunction G ℂ))
= (εη : ℂ) * (ε : ℂ) * (if χη = χ then (1 : ℂ) else 0) := by
intro χ ε
rw [hηeq, ClassFunction.inner_smul_left,
OddOrder.RepresentationTheory.inner_smul_right,
OddOrder.RepresentationTheory.irreducibleCharacter_inner_eq_ite]
rw [star_intCast]; ring
-- engine relation at `(i, j)`: `⟨η_ij, ψ^{τ₁}⟩ = ⟨η_ij, ψ̄^{τ₁}⟩`
have hrel : ClassFunction.inner (hyp.eta i j) psi
= ClassFunction.inner (hyp.eta i j) psiConj := by
have h0 : ClassFunction.inner (hyp.eta i j) (psi - psiConj) = 0 := by
rw [OddOrder.RepresentationTheory.inner_conj_symm, hengine i j, star_zero]
rwa [ClassFunction.inner_sub_right, sub_eq_zero] at h0
by_contra hnz
rw [hpeq] at hnz hrel
rw [hceq] at hrel
rw [hval χp εp] at hnz hrel
rw [hval χc εc] at hrel
-- nonzero forces `χη = χp`; the relation then forces `χη = χc`; contradiction
have hp_eq : χη = χp := by
by_contra hpne; rw [if_neg hpne, mul_zero] at hnz; exact hnz rfl
rw [if_pos hp_eq] at hnz hrel
have hc_eq : χη = χc := by
by_contra hcne
rw [if_neg hcne, mul_zero] at hrel
exact hnz hrel
exact hχne (hp_eq ▸ hc_eq)
end OddOrder.Peterfalvi.S16
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/S16_GridExpansion.lean