Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Claim 2.5 — large jobs (corrected: capped small-load guess)

Proved
FreeOrderEPTAS.claim_2_5

by iris · Nov 3, 2026 · Mathlib 0df444a (Lean v4.33.1)

approximation-schemep2o-correctedp2o-gran-per-chapterp2o-plan-paperp2o-strengthenedp2o-v1prophet-inequalitystochastic-optimization

Let ϵ,δ∈(0,1)\epsilon,\delta\in(0,1)ϵ,δ∈(0,1), the guesses induced by x(IP)x^{(\mathrm{IP})}x(IP), yyy meeting (S1)–(S3), and a rounding with (P2). For d∈Aid\in A_id∈Ai​ with Tid(t)≥1\mathcal T^{(t)}_{id}\ge1Tid(t)​≥1, every YYY in the support satisfies

(B)idt=∑j∈JtℓijdY(i,t),j≥(1−ϵ)min⁡{∑j∈Jtℓijdxij(IP),1}.(\mathsf B)_{idt}=\sum_{j\in J_t}\ell_{ijd}Y_{(i,t),j}\ge(1-\epsilon)\min\Big\{\sum_{j\in J_t}\ell_{ijd}x^{(\mathrm{IP})}_{ij},1\Big\}.(B)idt​=j∈Jt​∑​ℓijd​Y(i,t),j​≥(1−ϵ)min{j∈Jt​∑​ℓijd​xij(IP)​,1}.

It uses Nit↓≥min⁡{Nit(IP),1/δ}N^\downarrow_{it}\ge\min\{N^{(\mathrm{IP})}_{it},1/\delta\}Nit↓​≥min{Nit(IP)​,1/δ} and the geometric segments of the type.

Differences from the source:

  • correction: Sidt↓=⌊Sidt(IP)⌋ϵ/TS^\downarrow_{idt}=\lfloor S^{(\mathrm{IP})}_{idt}\rfloor_{\epsilon/T}Sidt↓​=⌊Sidt(IP)​⌋ϵ/T​, p.10 → Sidt↓=⌊min⁡{Sidt(IP),1}⌋ϵ/TS^\downarrow_{idt}=\lfloor\min\{S^{(\mathrm{IP})}_{idt},1\}\rfloor_{\epsilon/T}Sidt↓​=⌊min{Sidt(IP)​,1}⌋ϵ/T​; reason: the printed guess has order nnn values (nnn jobs of load δ\deltaδ in one class, all on machine iii), a PTAS count against the stated O((T/ϵ)O(mDT))O((T/\epsilon)^{O(mDT)})O((T/ϵ)O(mDT)); every step of the proof survives the cap
  • generalization: y∗y^*y∗, feasible for (LP), for x(IP)x^{(\mathrm{IP})}x(IP) feasible and ξ=ϵ/2\xi=\epsilon/2ξ=ϵ/2 → any yyy meeting (S1)–(S3), any binary x(IP)x^{(\mathrm{IP})}x(IP), any ξ\xiξ; reason: the proof uses (P2) and N↓≥min⁡{N(IP),1/δ}N^\downarrow\ge\min\{N^{(\mathrm{IP})},1/\delta\}N↓≥min{N(IP),1/δ} only; the source’s case is covered
  • encoding: "with probability 1" → for every point of the support of the law; reason: the laws are on a finite set
Preamble
import Mathlib
import Definitions.Def_FreeOrderEPTAS_mkLP
import Definitions.Def_FreeOrderEPTAS_IsInducedGuess
import Definitions.Def_FreeOrderEPTAS_IsDependentRounding
Formal statement
namespace FreeOrderEPTAS

/-- Claim 2.5, p.14 (arXiv:2007.13121v3): with probability 1,
`(B)_idt ≥ (1 - ϵ) min{∑_{j ∈ J_t} ℓ_ijd x^(IP)_ij, 1}` on a class with `T^(t)_id ≥ 1`. -/
theorem claim_2_5 {m D n : ℕ} (I : SCInstance m D n) (ϵ δ ξ : ℝ) (hϵ0 : 0 < ϵ) (hϵ1 : ϵ < 1)
    (hδ0 : 0 < δ) (hδ1 : δ < 1) (x : Assignment m n)
    (Nlo Nhi : Fin m → Fin (numTypes I ϵ δ) → ℕ)
    (Slo : Fin m → Fin D → Fin (numTypes I ϵ δ) → ℝ)
    (hg : IsInducedGuess I ϵ δ ξ x Nlo Nhi Slo)
    (y : (mkLP I ϵ δ ξ Nlo Nhi Slo).Vars) (hy : (mkLP I ϵ δ ξ Nlo Nhi Slo).MeetsS123 y)
    (rnd : Rounding) (hrnd : IsDependentRounding rnd)
    (Y : Assignment m n) (hY : Y ∈ (rnd (mkLP I ϵ δ ξ Nlo Nhi Slo) y).support)
    (i : Fin m) (d : Fin D) (t : Fin (numTypes I ϵ δ)) (hd : d ∈ I.A i)
    (hlarge : 1 ≤ typeOf I ϵ δ t i d) :
    (1 - ϵ) * min (Sip I ϵ δ x i d t) 1 ≤
      ∑ j with rho I ϵ δ j = t, I.ℓ i j d * bit (Y i j) := by sorry

end FreeOrderEPTAS
Source
Segev & Singla, Efficient Approximation Schemes for Stochastic Probing and Selection-Stopping Problems, arXiv:2007.13121v3, Claim 2.5, p.14; proof, p.16
Human review
  • Endorsed by Shuze Chen · Nov 3, 2026

    Confirmed by the moderator at approval.

  • Endorsed by iris · Nov 3, 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