Documentation

Atlas.BooleanFunctions.code.KKL

theorem BooleanFourier.expect_sq_eq_one {n : ℕ} (f : BoolFn n) (hf : ∀ (x : Fin n → Bool), f x = 1 ∨ f x = -1) :
(expect fun (x : Fin n → Bool) => f x ^ 2) = 1
theorem BooleanFourier.variance_le_one {n : ℕ} (f : BoolFn n) (hf : ∀ (x : Fin n → Bool), f x = 1 ∨ f x = -1) :
theorem BooleanFourier.kkl_theorem :
∃ c > 0, ∀ n ≥ 2, ∀ (f : BoolFn n), (∀ (x : Fin n → Bool), f x = 1 ∨ f x = -1) → ∃ (i : Fin n), fourierInfluence f i ≥ c * variance f * Real.log ↑n / ↑n
theorem BooleanFourier.variance_balanced_eq_one {n : ℕ} (f : BoolFn n) (hf : ∀ (x : Fin n → Bool), f x = 1 ∨ f x = -1) (hbal : expect f = 0) :
theorem BooleanFourier.kkl_balanced :
∃ c > 0, ∀ n ≥ 2, ∀ (f : BoolFn n), (∀ (x : Fin n → Bool), f x = 1 ∨ f x = -1) → expect f = 0 → ∃ (i : Fin n), fourierInfluence f i ≥ c * Real.log ↑n / ↑n