Theorem 7.1.6 — Brownian paths are nowhere Lipschitz, hence nowhere differentiable
ProvedDurrettProbability.brownian_nowhere_lipschitzLet be a Brownian motion on a probability space. Then for -almost every , for every time and every real constant , the path is not Lipschitz at with constant : there is no for which
Since a function differentiable at is Lipschitz at for some constant, this says in particular that almost every Brownian path is differentiable at no point whatsoever.
Formalization Note The order of quantifiers is what makes this the Paley–Wiener–Zygmund theorem rather than a much weaker statement: the null set comes first, so a single path fails the Lipschitz condition simultaneously at every point. The weaker reading — for each fixed , almost surely not Lipschitz at — would follow from a one-point calculation and is not what is asserted here.
The constant ranges over all the reals. For the Lipschitz condition can hold only where the path is locally constant, so its negation is easy there and the content is at . At the condition is one-sided, the index set being , and the conclusion is asserted there too.
import Mathlib import Definitions.Def_DurrettProbability_Brownian open Filter MeasureTheory ProbabilityTheory open scoped NNReal Topology
namespace DurrettProbability
theorem brownian_nowhere_lipschitz {Ω : Type*} [MeasurableSpace Ω] {P : Measure Ω}
[IsProbabilityMeasure P] {B : ℝ≥0 → Ω → ℝ} (hB : IsBrownianReal B P) :
∀ᵐ ω ∂P, ∀ (s : ℝ≥0) (C : ℝ), ¬ LipschitzAtPoint (fun t => B t ω) s C := by sorry
end DurrettProbabilityConfirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.