Quadratic maps to 𝔽₂ on finite elementary abelian 2-groups #
The paper's §6 determinant obstructions are quadratic maps q : V → 𝔽₂ on finite elementary
abelian 2-groups (equivalently 𝔽₂-vector spaces), with their polar forms
B(v,w) = q(v+w) + q(v) + q(w), nonsingularity (trivial polar radical), the Arf
invariant, and the Wall doubling q_U(x) = q(x) + B(x, U⁻¹x) = q(x) + B(x, Ux) of an
orthogonal operator U (char 2: U⁻¹x and Ux give the same doubling since B is
U-invariant; we take the paper's (83) shape directly).
Encoding decisions (docs/section67-extraction.md §D1):
- Modules are the project's plain classes (
[AddCommGroup V]+ aPropfor exponent 2 where needed) — noModule (ZMod 2)bundling, matchingGQ2/DiscreteModule.lean's conventions. Quadratic maps are plain functionsq : V → ZMod 2with the predicateIsQuadraticFp2(normalization + biadditive polar), mirroring howGQ2/Cohomology.leantreats cochains. arfis the democratic (majority) invariant (Browder):arf q = 0iff the zeros ofqare the strict majority. For a nonsingular form on a finite even-dimensional space this agrees with the classical Arf invariant (#q⁻¹(0) = 2^{d−1} + (−1)^{arf} 2^{d/2−1}), and it is the only reading §§6–9 consume (the base determinant Gauss sums of Props 6.9/6.18). Mathlib has no Arf invariant; this definition is total and choice-free. Deviation flagged in the extraction note.
No sorry in this file — everything here is definitional or elementary.
Every element of 𝔽₂ = ZMod 2 is 0 or 1 (the shared case-split helper; consolidates
the per-file zmod2_cases copies).
The polar form of q : V → 𝔽₂: B(v,w) = q(v+w) + q(v) + q(w).
Equations
- GQ2.QuadraticFp2.polar q v w = q (v + w) + q v + q w
Instances For
q is a quadratic map to 𝔽₂: it is normalized (q 0 = 0) and its polar form is
biadditive. (Over 𝔽₂ there is no scalar condition; on an exponent-2 group this is the standard
notion of a quadratic form, cf. paper §6 and Serre CiA.)
- map_zero : q 0 = 0
Instances For
The polar form is symmetric (immediately from commutativity of +).
Polarization at equal arguments computes q(2v); on an exponent-2 group it vanishes, i.e.
the polar form is alternating (B(v,v) = 0), because q(2v) = q(0) = 0.
Nonsingularity: the polar radical is trivial — every nonzero vector pairs nontrivially
with something. (For finite V this is the paper's nondegeneracy; no dual-space bundling.)
Equations
- GQ2.QuadraticFp2.Nonsingular q = ∀ (v : V), v ≠ 0 → ∃ (w : V), GQ2.QuadraticFp2.polar q v w ≠ 0
Instances For
Invariance of q under an action of C on V.
Equations
- GQ2.QuadraticFp2.IsInvariant C q = ∀ (c : C) (v : V), q (c • v) = q v
Instances For
The zero count #q⁻¹(0) of a form on a finite group — the (unsigned) content of the
paper's base determinant Gauss sums (Props 6.9/6.18, eqs. (91)/(115)).
Equations
- GQ2.QuadraticFp2.zeroCount q = Nat.card { v : V // q v = 0 }
Instances For
The Arf invariant, democratically (Browder): 0 iff the zeros form a strict majority.
Agrees with the classical Arf invariant for nonsingular forms on finite even-dimensional spaces
(where #q⁻¹(0) = 2^{d−1} ± 2^{d/2−1} ≠ 2^{d−1}); total and choice-free in general.
Equations
- GQ2.QuadraticFp2.arf q = if 2 * GQ2.QuadraticFp2.zeroCount q > Nat.card V then 0 else 1
Instances For
The Wall doubling of q by an operator U : V → V (paper Lemma 6.6 and eq. (83)):
q_U(x) = q(x) + B(x, Ux). (Char 2, B U-invariant: same as the paper's B(x, U⁻¹x).)
Equations
- GQ2.QuadraticFp2.qDouble q U x = q x + GQ2.QuadraticFp2.polar q x (U x)
Instances For
Paper-tag ledger (auto-generated by paperforge; do not edit) #
- eq. (83) = ⟦eq-QAcases⟧
- eq. (91) = ⟦eq-candidatezeros⟧
- Lemma 6.6 = ⟦lem-wall⟧