Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 4.1, (4.3) — empirical-law LLN for the percolated particle system

Open
GraphonMF.SparseLLN.eq_4_3

by tianyipeng · Nov 4, 2026 · Mathlib 0df444a (Lean v4.33.1)

graphoninteracting-diffusionslaw-of-large-numberspercolation

Consider the percolated nnn-particle system (4.1),

Xin(t)=Xi/n(0)+∫0t1nβn∑j=1nξijn b(Xin(s),Xjn(s)) ds+∫0tσ(Xin(s)) dBi/n(s),i=1,…,n,X_i^n(t)=X_{i/n}(0)+\int_0^t\frac1{n\beta_n}\sum_{j=1}^n\xi_{ij}^n\,b(X_i^n(s),X_j^n(s))\,ds+\int_0^t\sigma(X_i^n(s))\,dB_{i/n}(s),\qquad i=1,\dots,n,Xin​(t)=Xi/n​(0)+∫0t​nβn​1​j=1∑n​ξijn​b(Xin​(s),Xjn​(s))ds+∫0t​σ(Xin​(s))dBi/n​(s),i=1,…,n,

and the limiting graphon particle system (4.2),

Xu(t)=Xu(0)+∫0t ⁣∫I ⁣∫Rdb(Xu(s),x) G(u,v) μv,s(dx) dv ds+∫0tσ(Xu(s)) dBu(s),μu,t=L(Xu(t)),X_u(t)=X_u(0)+\int_0^t\!\int_I\!\int_{\mathbb R^d} b(X_u(s),x)\,G(u,v)\,\mu_{v,s}(dx)\,dv\,ds+\int_0^t\sigma(X_u(s))\,dB_u(s),\qquad \mu_{u,t}=\mathcal L(X_u(t)),Xu​(t)=Xu​(0)+∫0t​∫I​∫Rd​b(Xu​(s),x)G(u,v)μv,s​(dx)dvds+∫0t​σ(Xu​(s))dBu​(s),μu,t​=L(Xu​(t)),

driven by the same independent initial states and Brownian motions {Xu(0),Bu:u∈I}\{X_u(0),B_u: u\in I\}{Xu​(0),Bu​:u∈I}.

Assume Condition 2.2(a) (finitely many intervals covering III on each of which u↦L(Xu(0))u\mapsto\mathcal L(X_u(0))u↦L(Xu​(0)) is W2W_2W2​-continuous), Condition 4.1 (measurable initial laws with uniformly bounded second moments; bbb bounded and Lipschitz; σ\sigmaσ bounded, Lipschitz, invertible with bounded inverse; βn∈(0,1]\beta_n\in(0,1]βn​∈(0,1] and nβn→∞n\beta_n\to\inftynβn​→∞) and Condition 4.2 (GnG_nGn​ is a step graphon on the grid of mesh 1/n1/n1/n; ξijn=ξjin\xi^n_{ij}=\xi^n_{ji}ξijn​=ξjin​ are independent Bernoulli(βnGn(i/n,j/n))\mathrm{Bernoulli}(\beta_nG_n(i/n,j/n))Bernoulli(βn​Gn​(i/n,j/n)) for i≤ji\le ji≤j, independent of the noise; and ∥Gn−G∥□→0\|G_n-G\|_\square\to0∥Gn​−G∥□​→0). Let

μn:=1n∑i=1nδXin,μˉ:=∫IL(Xu) du.\mu^n:=\frac1n\sum_{i=1}^n\delta_{X_i^n},\qquad \bar\mu:=\int_I\mathcal L(X_u)\,du .μn:=n1​i=1∑n​δXin​​,μˉ​:=∫I​L(Xu​)du.

Then

μn⟶μˉin P(Cd) in probability as n→∞.(4.3)\mu^n\longrightarrow\bar\mu\quad\text{in }\mathcal P(\mathcal C_d)\text{ in probability as }n\to\infty. \tag{4.3}μn⟶μˉ​in P(Cd​) in probability as n→∞.(4.3)

Formalization notes. This is the first clause of Theorem 4.1, stated with exactly the same hypotheses and objects as the mission goal GraphonMF.SparseLLN.theorem_4_1 (Condition 2.2(b) is not assumed). Empirical measures are indexed by n+1n+1n+1 so that they are defined; convergence in probability is in the weak topology of P(Cd)\mathcal P(\mathcal C_d)P(Cd​): for every neighbourhood UUU of μˉ\bar\muμˉ​, P(μn∉U)→0\mathbb P(\mu^{n}\notin U)\to0P(μn∈/U)→0 (outer measure). In the paper this is proved (Section 7.2) by approximating GGG in operator norm by a continuous graphon G~\tilde GG~ (Lusin), using the stability estimate of Theorem 2.1(c), Lemma 6.2 for G~\tilde GG~, and Lemma 7.2 with Remark 2.1.

Preamble
import Definitions.Def_GraphonMF_SparseLLN_Setting

open MeasureTheory Filter Topology
open scoped ENNReal NNReal
set_option autoImplicit false
Formal statement
namespace GraphonMF.SparseLLN

/-- Theorem 4.1, display (4.3), pp. 3597–3598: under Conditions 2.2(a), 4.1 and 4.2 the
empirical law of the percolated particle system converges in probability to `∫_I Law(X_u) du`. -/
theorem eq_4_3
    {T : ℝ≥0} {d : ℕ} {Ω : Type*} [MeasurableSpace Ω]
    (hT : 0 < T) (hd : 0 < d)
    (P : Measure Ω) (X0 : GraphonMF.Stability.I → Ω → Vec d) (B : GraphonMF.Stability.I → ℝ≥0 → Ω → Vec d)
    (hnoise : NoiseSetting P X0 B)
    (G : GraphonMF.Stability.I → GraphonMF.Stability.I → ℝ) (hG : GraphonMF.Stability.IsGraphon G)
    (b : Vec d → Vec d → Vec d) (σ : Vec d → Matrix (Fin d) (Fin d) ℝ)
    (β : ℕ → ℝ) (h41 : Cond41 (initialLaw P X0) b σ β)
    (N : ℕ) (J : Fin N → Set GraphonMF.Stability.I)
    (h22a : Cond22a (initialLaw P X0) J)
    (Gs : ℕ → GraphonMF.Stability.I → GraphonMF.Stability.I → ℝ)
    (ξ : (n : ℕ) → Ω → Fin n → Fin n → ℝ)
    (hξ : ∀ n i j, Measurable (fun ω => ξ n ω i j))
    (h42 : Cond42 P X0 B ξ Gs β G)
    (X : GraphonMF.Stability.I → Ω → GraphonMF.Stability.Cd T d)
    (hX : IsGraphonSolution42 P X0 B hnoise G b σ X)
    (Xn : (n : ℕ) → Fin n → Ω → GraphonMF.Stability.Cd T d)
    (hXn : ∀ n, IsParticleSolution41 P X0 B hnoise ξ hξ β b σ n (Xn n)) :
    ConvergesInProbability P (empiricalSeq Xn) (solutionMixture hX) := by sorry

end GraphonMF.SparseLLN
Source
Bayraktar, Chakraborty, Wu, Graphon mean field systems, Ann. Appl. Probab. 33(5) (2023), pp. 3597–3598, Theorem 4.1, display (4.3); proof in Section 7.2, p. 3615; https://doi.org/10.1214/22-AAP1901 (arXiv:2003.13180)

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