Documentation

Atlas.TheoryOfProbability.code.KolmogorovContinuity

The set of dyadic rationals in ℝ, i.e., real numbers of the form k / 2 ^ n for some integer k and natural number n. This is the dense countable set on which the Kolmogorov continuity theorem first establishes Hölder continuity.

Instances For
    theorem kolmogorov_continuity_theorem {Ω : Type u_1} {m : MeasurableSpace Ω} {μ : MeasureTheory.Measure Ω} [MeasureTheory.IsProbabilityMeasure μ] {X : ℝ → Ω → ℝ} {α β K : ℝ} (hα : 0 < α) (hβ : 0 < β) (hK : 0 ≤ K) (hkolm : ∀ (s t : ℝ), ∫ (ω : Ω), |X s ω - X t ω| ^ β ∂μ ≤ K * |s - t| ^ (1 + α)) (γ : ℝ) :
    γ < α / β → γ > 0 → ∃ (Y : ℝ → Ω → ℝ), (∀ (t : ℝ), ∀ᵐ (ω : Ω) ∂μ, Y t ω = X t ω) ∧ ∀ᵐ (ω : Ω) ∂μ, ∃ (C : ℝ), 0 < C ∧ ∀ (q r : ℝ), q ∈ dyadicRationals ∩ Set.Icc 0 1 → r ∈ dyadicRationals ∩ Set.Icc 0 1 → |Y q ω - Y r ω| ≤ C * |q - r| ^ γ

    Kolmogorov continuity theorem. Suppose E |X_s - X_t|^β ≤ K |t - s|^{1+α} where α, β > 0. Then for every γ < α/β with γ > 0, there exists a modification Y of the process X (i.e., Y t = X t almost surely for each t) such that with probability one there is a constant C(ω) > 0 with |Y(q,ω) - Y(r,ω)| ≤ C |q - r|^γ for all dyadic rationals q, r ∈ [0,1]. In other words, Y has Hölder-continuous sample paths of exponent γ on the dyadic rationals of the unit interval.