Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 4 — suggested matching achieves a tight 1−1/e1-1/e1−1/e factor

Proved
OnlineStochMatching.SuggestedMatching.theorem_4

by mikedeng1 · Oct 5, 2026 · Mathlib 0df444a (Lean v4.33.1)

bipartite-matchingonline-algorithmsp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1stochastic-matching

Fix the suggested matching algorithm on any finite integer-frequency online stochastic matching instance. For every ε>0\varepsilon>0ε>0, there exist δ>0\delta>0δ>0 and NNN independent of the instance, of which maximum integral flow is selected, and of its random-choice labelling such that, for n≥Nn\ge Nn≥N,

Pr⁡ ⁣[ALG≥(1−e−1)OPT−εn]≥1−e−δn.\Pr\!\left[\mathrm{ALG}\ge(1-e^{-1})\mathrm{OPT}-\varepsilon n\right]\ge1-e^{-\delta n}.Pr[ALG≥(1−e−1)OPT−εn]≥1−e−δn.

The factor is tight: on the complete bipartite family with n≥1n\ge1n≥1, every maximum expected-instance matching and every valid labelling have OPT=n\mathrm{OPT}=nOPT=n in every scenario and E[ALG]=n(1−(1−1/n)n)\mathbb E[\mathrm{ALG}]=n(1-(1-1/n)^n)E[ALG]=n(1−(1−1/n)n); the expected ratio tends to 1−e−11-e^{-1}1−e−1 as n→∞n\to\inftyn→∞.

Formalization Note The paper states a high-probability approximation factor. Its proof yields the displayed additive error without assuming OPT=Ω(n)\mathrm{OPT}=\Omega(n)OPT=Ω(n); a multiplicative ratio with an arbitrary small error follows when that further growth condition holds. The theorem keeps all integer ei≥0e_i\ge0ei​≥0 and represents both arrival and algorithm randomness by independent uniform draws of labelled copies.

Preamble
import Definitions.Def_OnlineStochMatching_SuggestedMatching_Algorithm
Formal statement
namespace OnlineStochMatching.SuggestedMatching

/-- Theorem 4: additive high-probability guarantee and a complete-graph family
whose exact expected ratio approaches `1 - 1/e`. -/
theorem theorem_4 :
    (∀ ε : ℝ, 0 < ε → ∃ δ : ℝ, 0 < δ ∧ ∃ N : ℕ,
      ∀ (A I : Type) [Fintype A] [Fintype I]
        (inst : Instance A I) (hn : N ≤ inst.n)
        (M : Finset (A × I)) (hM : IsMaxBMatching inst M)
        (label : inst.Copy → Option A) (hlabel : IsLabelling inst M label),
        runProbability inst (fun ω =>
          (1 - Real.exp (-1)) * (optimum inst ω : ℝ) - ε * (inst.n : ℝ) ≤
            (algorithm inst label ω : ℝ)) ≥
          1 - Real.exp (-δ * (inst.n : ℝ))) ∧
    (∀ (n : ℕ) (hn : 0 < n),
      let inst := completeInstance n hn
      ∀ (M : Finset (Fin n × Fin n)) (hM : IsMaxBMatching inst M)
        (label : inst.Copy → Option (Fin n)) (hlabel : IsLabelling inst M label),
        (∀ ω : Fin inst.n → inst.Copy, optimum inst ω = n) ∧
          expectedAlgorithm inst label =
            (n : ℝ) * (1 - (1 - 1 / (n : ℝ)) ^ n)) ∧
    Filter.Tendsto (fun n : ℕ => 1 - (1 - 1 / (n : ℝ)) ^ n)
      Filter.atTop (nhds (1 - Real.exp (-1))) := by sorry

end OnlineStochMatching.SuggestedMatching
Source
Feldman, Mehta, Mirrokni, Muthukrishnan, Online Stochastic Matching: Beating 1-1/e, arXiv:0905.4100v1, p. 6, Theorem 4 and §4.1, Tightness of the Analysis
Human review
  • Endorsed by Shuze Chen · Oct 6, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 6, 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