OAI/Analysis/KadisonKastler/IndependentOrbitMoments.lean
3 declarations, 0 contain sorry, 0 declared as axioms
Fourth moments for independently sampled operator coefficients
The diagonal correction in the sign expansion is retained before the
independent samples are averaged. Consequently its coefficient is
1/m, while the two off-diagonal pairings have coefficient 1-1/m.
- theorem
OAI.KadisonKastler.expect_sampledSignSum_vector_fourth_leline 24Source documentation
Exact diagonal counting and the crossed-pairing inequality give the unnormalized sampled fourth-moment estimate.
theorem expect_sampledSignSum_vector_fourth_le (Ω : H) (a : G → H →L[ℂ] H) : (𝔼 ω : I → G, 𝔼 ε : I → Bool, realVectorState Ω ((star (finiteSignSum (fun i => a (ω i)) ε) * finiteSignSum (fun i => a (ω i)) ε) ^ 2)) ≤ (Fintype.card I : ℝ) * (𝔼 g, realVectorState Ω ((star (a g) * a g) ^ 2)) + (Fintype.card I : ℝ) * ((Fintype.card I : ℝ) - 1) * ((𝔼 g, 𝔼 h, realVectorState Ω ((star (a g) * a g) * (star (a h) * a h))) + 2 * (𝔼 g, 𝔼 h, realVectorState Ω ((star (a g) * a h) * (star (a h) * a g)))) := by …Mentions: OAI.KadisonKastler.realVectorState, OAI.KadisonKastler.finiteSignSum
- theorem
OAI.KadisonKastler.vector_fourth_smulline 79 - theorem
OAI.KadisonKastler.expect_sampledNormalizedSignSum_vector_fourth_leline 90Source documentation
Normalizing by the sample size yields the exact
1/mdiagonal cost and the two independent off-diagonal moments.theorem expect_sampledNormalizedSignSum_vector_fourth_le (Ω : H) (a : G → H →L[ℂ] H) : (𝔼 ω : I → G, 𝔼 ε : I → Bool, realVectorState Ω ((star (normalizedFiniteSignSum (fun i => a (ω i)) ε) * normalizedFiniteSignSum (fun i => a (ω i)) ε) ^ 2)) ≤ (𝔼 g, realVectorState Ω ((star (a g) * a g) ^ 2)) / Fintype.card I + (1 - (Fintype.card I : ℝ)⁻¹) * ((𝔼 g, 𝔼 h, realVectorState Ω ((star (a g) * a g) * (star (a h) * a h))) + 2 * (𝔼 g, 𝔼 h, realVectorState Ω ((star (a g) * a h) * (star (a h) * a g)))) := by …Mentions: OAI.KadisonKastler.realVectorState, OAI.KadisonKastler.normalizedFiniteSignSum