Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 7.3 (second statement) — P(L(X)⊂A)≥(1−w(t,δ,T)) P(∃s≥t:X(s)∈U)P(L(X)\subset A)\ge(1-w(t,\delta,T))\,P(\exists s\ge t: X(s)\in U)P(L(X)⊂A)≥(1−w(t,δ,T))P(∃s≥t:X(s)∈U)

Proved
StochApproxDyn.Attractor.measure_limitSet_subset_ge

by mikedeng1 · Oct 4, 2026 · Mathlib 0df444a (Lean v4.33.1)

attractorconvergence-with-positive-probabilityp2o-batch-pfp1ap2o-gran-per-chapterp2o-plan-paperp2o-v1stochastic-approximation

Let Φ\PhiΦ be a semiflow on a locally compact metric space MMM and let A⊂MA\subset MA⊂M be an attractor of Φ\PhiΦ with basin B(A)B(A)B(A). Let U⊂MU\subset MU⊂M be an open set, relatively compact, with U‾⊂B(A)\overline U\subset B(A)U⊂B(A). Then there exist numbers T>0T>0T>0 and δ>0\delta>0δ>0, depending only on UUU (and on Φ\PhiΦ and AAA), such that for every process XXX on a probability space (Ω,F,P)(\Omega,\mathcal F,P)(Ω,F,P) satisfying the standing assumptions of Section 7 with a function www, and every t≥0t\ge0t≥0,

P(L(X)⊂A) ≥ (1−w(t,δ,T)) P(∃s≥t: X(s)∈U).P\big(L(X)\subset A\big)\ \ge\ \big(1-w(t,\delta,T)\big)\,P\big(\exists s\ge t:\ X(s)\in U\big).P(L(X)⊂A) ≥ (1−w(t,δ,T))P(∃s≥t: X(s)∈U).

This is the quantitative half of Theorem 7.3: the probability of converging to AAA is bounded below by the probability of visiting a compact part of the basin after time ttt, discounted by the probability bound w(t,δ,T)w(t,\delta,T)w(t,δ,T) in condition (24). When w(t,δ,T)≥1w(t,\delta,T)\ge1w(t,δ,T)≥1 the bound is trivial; it is informative for ttt large.

Formalization Note TTT and δ\deltaδ are chosen before the probability space, the process, www and ttt, so they cannot depend on any of them. Probabilities are in ℝ≥0∞, and the factor 1−w(t,δ,T)1-w(t,\delta,T)1−w(t,δ,T) is ENNReal.ofReal (1 - w t δ T), which is 000 when w>1w>1w>1 (the real inequality is then trivial as well). The event {L(X)⊂A}\{L(X)\subset A\}{L(X)⊂A} and the hitting event are measured by PPP as an outer measure; no measurability hypothesis on them is added. The inequality is asserted for every t≥0t\ge0t≥0.

Preamble
import Mathlib
import Definitions.Def_StochApproxDyn_Attractor_Dynamics
import Definitions.Def_StochApproxDyn_Attractor_Process

open scoped NNReal ENNReal
open MeasureTheory

universe u v
Formal statement
namespace StochApproxDyn.Attractor

/-- Theorem 7.3, second statement (Benaïm 1999, pp. 31–32). The numbers `T, δ` depend only on
`Φ`, `A` and `U`: they are chosen before the probability space, the process and `w`. -/
theorem measure_limitSet_subset_ge {M : Type u} [MetricSpace M] [LocallyCompactSpace M]
    [MeasurableSpace M] [BorelSpace M] (Φ : Flow ℝ≥0 M) (A : Set M) (hA : StochApproxDyn.LimitSet.IsAttractor Φ A)
    (U : Set M) (hU : IsOpen U) (hUc : IsCompact (closure U)) (hUB : closure U ⊆ StochApproxDyn.LimitSet.basin Φ A) :
    ∃ T : ℝ≥0, 0 < T ∧ ∃ δ : ℝ≥0, 0 < δ ∧
      ∀ (Ω : Type v) (m0 : MeasurableSpace Ω) (P : Measure Ω), IsProbabilityMeasure P →
        ∀ (ℱ : Filtration ℝ≥0 m0) (X : ℝ≥0 → Ω → M) (w : ℝ≥0 → ℝ≥0 → ℝ≥0 → ℝ≥0),
          SatisfiesStandingAssumption Φ P ℱ X w →
          ∀ t : ℝ≥0,
            ENNReal.ofReal (1 - (w t δ T : ℝ)) * P {ω | ∃ s : ℝ≥0, t ≤ s ∧ X s ω ∈ U} ≤
              P {ω | StochApproxDyn.LimitSet.limitSet (fun s => X s ω) ⊆ A} := by sorry

end StochApproxDyn.Attractor
Source
Benaïm, Dynamics of Stochastic Approximation Algorithms, Séminaire de Probabilités XXXIII, LNM 1709 (1999), pp. 31–32, Section 7.1, Theorem 7.3 (second statement)
Human review
  • Endorsed by Shuze Chen · Oct 5, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 5, 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