Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§2.2, p. 5 — (√N[ĝ_N(x) − g(x)])_{x∈T} ⇒ (Y(x))_{x∈T}, a centered Gaussian vector with the autocovariance of G(x, W)

Proved
KleywegtSAA.ValueCLT.joint_clt

by mikedeng1 · Oct 7, 2026 · Mathlib 0df444a (Lean v4.33.1)

central-limit-theoremmultivariate-gaussianp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1sample-average-approximation

Let S\mathcal SS be a nonempty finite set, G(x,⋅)G(x, \cdot)G(x,⋅) measurable for x∈Sx \in \mathcal Sx∈S, and W1,W2,…W^1, W^2, \dotsW1,W2,… an i.i.d. sample of WWW. Let T⊆ST \subseteq \mathcal ST⊆S be a set of points such that the variance σ2(x)=Var⁡{G(x,W)}\sigma^2(x) = \operatorname{Var}\{G(x, W)\}σ2(x)=Var{G(x,W)} of (2.7) exists for every x∈Tx \in Tx∈T. Let Y=(Y(x))x∈TY = (Y(x))_{x \in T}Y=(Y(x))x∈T​ be a Gaussian random vector with mean zero and covariance

Cov⁡(Y(x),Y(x′))=Cov⁡(G(x,W),G(x′,W)),x,x′∈T.\operatorname{Cov}\big(Y(x), Y(x')\big) = \operatorname{Cov}\big(G(x, W), G(x', W)\big), \qquad x, x' \in T.Cov(Y(x),Y(x′))=Cov(G(x,W),G(x′,W)),x,x′∈T.

Then, as N→∞N \to \inftyN→∞,

(N [g^N(x)−g(x)])x∈T  ⇒  (Y(x))x∈T,\Big(\sqrt N\,\big[\hat g_N(x) - g(x)\big]\Big)_{x \in T} \;\Rightarrow\; \big(Y(x)\big)_{x \in T},(N​[g^​N​(x)−g(x)])x∈T​⇒(Y(x))x∈T​,

where ⇒\Rightarrow⇒ denotes convergence in distribution.

This is the central limit theorem step of the proof of Proposition 2.3; with T=S∗T = \mathcal S^*T=S∗ it supplies the Gaussian limit in (2.8).

Formalization Note The page states the CLT for T=ST = \mathcal ST=S under finite variances on S\mathcal SS, in two sentences: marginal convergence of each coordinate to N(0,σ2(x))N(0, \sigma^2(x))N(0,σ2(x)) and the identification of the covariance of the limit. The proof of (2.8) uses the joint convergence of the vector, which is what is stated here, for every subfamily T⊆ST \subseteq \mathcal ST⊆S with finite variances (so it covers both T=ST = \mathcal ST=S and T=S∗T = \mathcal S^*T=S∗). The limit is any random vector YYY, on any probability space, whose law is the multivariate Gaussian measure with mean 000 and the autocovariance matrix of G(⋅,W)G(\cdot, W)G(⋅,W) on TTT; that matrix is positive semidefinite under the variance hypothesis, so the Gaussian law is the genuine one. The integrability hypothesis of (1.1) is implied by the variance hypothesis on TTT and is not repeated.

Preamble
import Mathlib
import Definitions.Def_KleywegtSAA_ValueCLT_Setting
Formal statement
namespace KleywegtSAA.ValueCLT

open MeasureTheory ProbabilityTheory Filter Topology

open Classical in
/-- Kleywegt–Shapiro, §2.2, p. 5 (the CLT step before Proposition 2.3): for a finite family
`T ⊆ S` of points at which `G(x, W)` has finite variance, the random vector
`(√N [ĝ_N(x) − g(x)])_{x ∈ T}` converges in distribution to a centered Gaussian vector `Y` whose
covariance matrix is the autocovariance `Cov(G(x, W), G(x', W))`, `x, x' ∈ T`. -/
theorem joint_clt
    {X : Type*} (S : Finset X)
    {𝒲 : Type*} [MeasurableSpace 𝒲] (G : X → 𝒲 → ℝ) (hG : ∀ x ∈ S, Measurable (G x))
    {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω) [IsProbabilityMeasure P]
    (W : ℕ → Ω → 𝒲) (hWm : ∀ n, Measurable (W n)) (hind : iIndepFun W P)
    (hid : ∀ n, IdentDistrib (W n) (W 0) P P)
    (T : Finset X) (hT : T ⊆ S)
    (hT2 : ∀ x ∈ T, MemLp (fun ω => G x (W 0 ω)) 2 P) :
    ∀ {Ω' : Type*} [MeasurableSpace Ω'] (Q : Measure Ω') [IsProbabilityMeasure Q]
      (Y : Ω' → EuclideanSpace ℝ T),
      HasLaw Y (multivariateGaussian 0 (covMat G P W T)) Q →
      TendstoInDistribution (fun (N : ℕ) ω => scaledDev G P W T N ω) atTop Y (fun _ => P) Q := by sorry

end KleywegtSAA.ValueCLT
Source
Kleywegt & Shapiro, The sample average approximation method for stochastic discrete optimization, preprint (two-author version, sha256 56657748…), p. 5, §2.2, (2.7) and the two sentences following it ("Then it follows by the Central Limit Theorem (CLT) …")
Human review
  • Endorsed by Shuze Chen · Oct 7, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 7, 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