Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 8.1.4 — Donsker's theorem

Proved
DurrettProbability.donsker

by naimengye · Sep 19, 2026 · Mathlib 0df444a (Lean v4.33.1)

analysisprobabilitystochastic-systems

Let X0,X1,…X_0,X_1,\dotsX0​,X1​,… be measurable real random variables on a probability space, jointly independent, each with the law of X0X_0X0​, satisfying

EX0=0,EX02=1,\mathbb E X_0=0,\qquad \mathbb E X_0^2=1,EX0​=0,EX02​=1,

with X02X_0^2X02​ integrable. Let Sm=X0+⋯+Xm−1S_m=X_0+\dots+X_{m-1}Sm​=X0​+⋯+Xm−1​, let S(u)S(u)S(u) be the polygonal path agreeing with SmS_mSm​ at integer times and linear in between, and set

Wn(t)=S(nt)n,t∈[0,1],W_n(t)=\frac{S(nt)}{\sqrt n},\qquad t\in[0,1],Wn​(t)=n​S(nt)​,t∈[0,1],

a random element of C[0,1]C[0,1]C[0,1].

Let BBB be a Brownian motion on a second probability space, every one of whose paths is continuous, and let B(⋅)B(\cdot)B(⋅) be its restriction to [0,1][0,1][0,1], also a random element of C[0,1]C[0,1]C[0,1].

Then

Wn ⟹ B(⋅),W_n\ \Longrightarrow\ B(\cdot),Wn​ ⟹ B(⋅),

meaning that the laws of WnW_nWn​ on C[0,1]C[0,1]C[0,1] converge weakly to the law of B(⋅)B(\cdot)B(⋅).

The space C[0,1]C[0,1]C[0,1] carries the topology of uniform convergence and its Borel σ\sigmaσ-algebra, so this is convergence of measures on an infinite-dimensional function space, not convergence of finite-dimensional marginals. That is precisely the content beyond the central limit theorem, which is the case of evaluation at t=1t=1t=1.

Formalization Note Indexing runs from zero, so S0=0S_0=0S0​=0 and the walk's mmm-th partial sum uses X0,…,Xm−1X_0,\dots,X_{m-1}X0​,…,Xm−1​; independence is joint independence of the whole family, not pairwise. Identical distribution is stated as each XkX_kXk​ having the same law as X0X_0X0​, and the mean and variance are moments of X0X_0X0​.

The interpolation is a finite sum of clamped linear pieces, so no continuity side condition accompanies the definition of WnW_nWn​; it is an element of C[0,1]C[0,1]C[0,1] by construction. At n=0n=0n=0 the rescaling divides by 0=0\sqrt0=00​=0, giving the zero path, which does not affect a limit along n→∞n\to\inftyn→∞.

Every path of BBB is assumed continuous, not merely almost every one. This is what makes B(⋅)B(\cdot)B(⋅) a genuine C[0,1]C[0,1]C[0,1]-valued random variable, and it is no restriction: a Brownian motion may be modified on a null set to have all paths continuous, and Durrett's construction on C[0,∞)C[0,\infty)C[0,∞) has the property by definition.

Existence of a Brownian motion is a hypothesis here, not a claim; producing one is the content of Durrett's Theorem 7.1.1, which the ambient library does not supply.

Preamble
import Mathlib
import Definitions.Def_DurrettProbability_Brownian
import Definitions.Def_DurrettProbability_Donsker

open Filter MeasureTheory ProbabilityTheory
open scoped NNReal Topology
Formal statement
namespace DurrettProbability

theorem donsker {Ω : Type*} [MeasurableSpace Ω] {P : Measure Ω} [IsProbabilityMeasure P]
    {X : ℕ → Ω → ℝ} (hmeas : ∀ k, Measurable (X k)) (hindep : iIndepFun X P)
    (hident : ∀ k, IdentDistrib (X k) (X 0) P P)
    (hint : Integrable (fun ω => X 0 ω ^ 2) P)
    (hmean : ∫ ω, X 0 ω ∂P = 0) (hvar : ∫ ω, X 0 ω ^ 2 ∂P = 1)
    {Ω' : Type*} [MeasurableSpace Ω'] {P' : Measure Ω'} [IsProbabilityMeasure P']
    {B : ℝ≥0 → Ω' → ℝ} (hB : IsBrownianReal B P') (hBc : ∀ ω, Continuous fun t => B t ω) :
    TendstoInDistribution (fun n : ℕ => walkPath X n) atTop (brownianPath B)
      (fun _ => P) P' := by sorry

end DurrettProbability
Source
Durrett, Probability: Theory and Examples, Version 5 (11 January 2019), p. 392 (PDF p. 400), Theorem 8.1.4: 'Donsker''s theorem. Under the hypotheses of Theorem 8.1.2, S(n.)/sqrt(n) => B(.), i.e., the associated measures on C[0,1] converge weakly.' The hypotheses of Theorem 8.1.2, p. 391 (PDF p. 399): 'Let X_1, X_2, ... be i.i.d. with a distribution F, which has mean 0 and variance 1, and let S_n = X_1 + ... + X_n.' The interpolation, same page: 'Let N be the nonnegative integers and let S(u) = S_k if u = k in N, linear on [k, k+1] for k in N.' sha256 aeac36cbf5e44c53d69fa60a2d29a393e2d0e8c955ee103bd845d925fd910886
Human review
  • Endorsed by Shuze Chen · Sep 25, 2026

    Confirmed by the moderator at approval.

  • Endorsed by naimengye · Sep 25, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me