bernoulli_powerset_expectation_quadruple
Provedconcentration-inequalitiesmatrix-completionprobability
Quadruple linearity of the Bernoulli powerset expectation. For the finite Bernoulli observation measure on (each coordinate included independently with probability ), the expectation of a fourfold coordinate sum of functions of four inclusion indicators pushes through all four sums term by term: , where . This is the linearity step that expands the fourth moment of a statistic linear in the inclusion indicators (e.g. the centered sampling coefficient ) into a fourfold sum of coordinate expectations — the first step of the Rosenthal/Latała even-moment expansion (Boucheron–Lugosi–Massart, Concentration Inequalities, OUP 2013, Ch. 15) used in the scalar/noncommutative Bernstein moment bounds of Candès–Recht 2009 (arXiv:0805.4471, §6).
Preamble
import Definitions.Def_matrix_completion_neumann open MatrixCompletion open scoped BigOperators Classical
Formal statement
theorem bernoulli_powerset_expectation_quadruple {n₁ n₂ : ℕ} (p : ℝ)
(G : (Fin n₁ × Fin n₂) → (Fin n₁ × Fin n₂) → (Fin n₁ × Fin n₂) → (Fin n₁ × Fin n₂) →
ℝ → ℝ → ℝ → ℝ → ℝ) :
bernoulliExpectation p
(fun Omega => ∑ a : Fin n₁ × Fin n₂, ∑ b : Fin n₁ × Fin n₂,
∑ c : Fin n₁ × Fin n₂, ∑ d : Fin n₁ × Fin n₂,
G a b c d (if a ∈ Omega then 1 else 0) (if b ∈ Omega then 1 else 0)
(if c ∈ Omega then 1 else 0) (if d ∈ Omega then 1 else 0)) =
∑ a : Fin n₁ × Fin n₂, ∑ b : Fin n₁ × Fin n₂,
∑ c : Fin n₁ × Fin n₂, ∑ d : Fin n₁ × Fin n₂,
bernoulliExpectation p
(fun Omega => G a b c d (if a ∈ Omega then 1 else 0) (if b ∈ Omega then 1 else 0)
(if c ∈ Omega then 1 else 0) (if d ∈ Omega then 1 else 0)) := by sorrySource