Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 3.1, (3.3), p. 3595 — μⁿ → μ̄ in 𝒫(𝒞_d) in probability

Open
GraphonMF.DenseLLN.theorem_3_1_eq_3_3

by tianyipeng · Nov 5, 2026 · Mathlib 0df444a (Lean v4.33.1)

graphoninteracting-diffusionslaw-of-large-numberspropagation-of-chaos

Let I=[0,1]I=[0,1]I=[0,1], T>0T>0T>0, and let the initial states Xu(0)∼μu(0)X_u(0)\sim\mu_u(0)Xu​(0)∼μu​(0) and the standard Brownian motions BuB_uBu​, u∈Iu\in Iu∈I, be mutually independent on one probability space. Assume

  1. Condition 2.1: u↦μu(0)u\mapsto\mu_u(0)u↦μu​(0) is measurable, sup⁡uE∣Xu(0)∣2+ε<∞\sup_u\mathbb E|X_u(0)|^{2+\varepsilon}<\inftysupu​E∣Xu​(0)∣2+ε<∞ for some ε>0\varepsilon>0ε>0, and b,σb,\sigmab,σ are Lipschitz;
  2. Condition 2.2(a): for a finite cover of III by intervals I1,…,INI_1,\dots,I_NI1​,…,IN​, u↦μu(0)u\mapsto\mu_u(0)u↦μu​(0) is W2W_2W2​-continuous on each IiI_iIi​;
  3. Condition 3.1: each GnG_nGn​ is a step graphon as in (3.2); the weights satisfy either ξijn=Gn(i/n,j/n)\xi^n_{ij}=G_n(i/n,j/n)ξijn​=Gn​(i/n,j/n), or ξijn=ξjin∼Bernoulli(Gn(i/n,j/n))\xi^n_{ij}=\xi^n_{ji}\sim\mathrm{Bernoulli}(G_n(i/n,j/n))ξijn​=ξjin​∼Bernoulli(Gn​(i/n,j/n)) independently for i≤ji\le ji≤j and independently of the noise; and ∥Gn−G∥□→0\|G_n-G\|_\square\to0∥Gn​−G∥□​→0 for a graphon GGG.

Let X=(Xu)u∈IX=(X_u)_{u\in I}X=(Xu​)u∈I​ solve the graphon particle system (2.1) for GGG, with path laws μu\mu_uμu​, and let Xn=(X1n,…,Xnn)X^n=(X^n_1,\dots,X^n_n)Xn=(X1n​,…,Xnn​) solve the nnn-particle system (3.1), driven by Xi/n(0)X_{i/n}(0)Xi/n​(0) and Bi/nB_{i/n}Bi/n​. Then

μn:=1n∑i=1nδXin ⟶ μˉ:=∫Iμu duin P(Cd) in probability(3.3)\mu^n:=\frac1n\sum_{i=1}^n\delta_{X^n_i}\ \longrightarrow\ \bar\mu:=\int_I\mu_u\,du\quad\text{in }\mathcal P(\mathcal C_d)\text{ in probability}\tag{3.3}μn:=n1​i=1∑n​δXin​​ ⟶ μˉ​:=∫I​μu​duin P(Cd​) in probability(3.3)

as n→∞n\to\inftyn→∞.

This is the first assertion of Theorem 3.1; it needs no regularity of GGG (Condition 2.2(b) is not assumed). It is stated separately because the second assertion (3.4) follows directly from Lemma 6.1 and Remark 2.1, while (3.3) needs the continuous-graphon approximation of §6.2.

Formalization Note Same conventions as the parent theorem GraphonMF.DenseLLN.theorem_3_1: convergence in probability means that for every neighbourhood UUU of μˉ\bar\muμˉ​ in the weak topology, P(μn∉U)→0\mathbb P(\mu^n\notin U)\to0P(μn∈/U)→0 with P\mathbb PP applied as an outer measure; μˉ\bar\muμˉ​ is the mixture ∫Iμu du\int_I\mu_u\,du∫I​μu​du; each ξn\xi^nξn is assumed measurable. The hypotheses are exactly those of the parent theorem.

Preamble
import Mathlib
import Definitions.Def_GraphonMF_DenseLLN_Setting

open MeasureTheory ProbabilityTheory Filter Topology
open scoped ENNReal NNReal
Formal statement
namespace GraphonMF.DenseLLN
theorem theorem_3_1_eq_3_3 {Ω : Type*} [MeasurableSpace Ω] {T : ℝ≥0} {d N : ℕ} {P : Measure Ω}
    {μ0 : GraphonMF.Stability.I → Measure (GraphonMF.Stability.State d)} {X0 : GraphonMF.Stability.I → Ω → GraphonMF.Stability.State d} {B : GraphonMF.Stability.I → ℝ≥0 → Ω → GraphonMF.Stability.State d}
    (noise : NoiseSetting P μ0 X0 B) (hT : 0 < T)
    (b : GraphonMF.Stability.State d → GraphonMF.Stability.State d → GraphonMF.Stability.State d) (σ : GraphonMF.Stability.State d → GraphonMF.Stability.State d → Matrix (Fin d) (Fin d) ℝ)
    (ε : ℝ) (h21 : GraphonMF.Stability.Cond21 ε μ0 b σ)
    (J : Fin N → Set GraphonMF.Stability.I) (h22a : Cond22a μ0 J)
    (G : GraphonMF.Stability.I → GraphonMF.Stability.I → ℝ) (hG : GraphonMF.Stability.IsGraphon G)
    (Gs : ℕ → GraphonMF.Stability.I → GraphonMF.Stability.I → ℝ) (h31a : ∀ n, 0 < n → GraphonMF.Stability.IsStepGraphon n (Gs n))
    (ξ : (n : ℕ) → Ω → Fin n → Fin n → ℝ) (hξ : ∀ n, Measurable (ξ n))
    (h31b : Cond31b P X0 B ξ Gs)
    (h31c : Tendsto (fun n => GraphonMF.Stability.cutNorm (fun u v => Gs n u v - G u v)) atTop (𝓝 0))
    (X : GraphonMF.Stability.I → Ω → GraphonMF.Stability.Cd T d) (hX : IsGraphonSolution noise G b σ X)
    (Xn : (n : ℕ) → Fin n → Ω → GraphonMF.Stability.Cd T d)
    (hXn : ∀ n, IsParticleSolution noise b σ n (ξ n) (hξ n) (Xn n)) :
    TendstoInProb P (fun n ω => empiricalPM (fun i => Xn n i ω)) (mixturePM P X) := by sorry
end GraphonMF.DenseLLN
Source
Bayraktar, Chakraborty, Wu, Graphon mean field systems, Ann. Appl. Probab. 33(5) (2023), p. 3595, Theorem 3.1, display (3.3); proof in §6.2, p. 3608

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