Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§3.2 — K≤1/ϵK\le1/\epsilonK≤1/ϵ and the stable buckets (corrected: half-open jumps, Figure 1 indexing)

Proved
FreeOrderEPTAS.bucket_facts

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

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

Let σ\sigmaσ be a permutation with Vt∗=RtV^*_t=R_tVt∗​=Rt​ and 1/ϵ1/\epsilon1/ϵ a positive integer. For the corrected buckets:

  1. K≤1/ϵK\le1/\epsilonK≤1/ϵ;
  2. every stable bucket has BaseVal(Bi+1)−BaseVal(Bi)≤ϵV1∗\mathrm{BaseVal}(B_{i+1})-\mathrm{BaseVal}(B_i)\le\epsilon V^*_1BaseVal(Bi+1​)−BaseVal(Bi​)≤ϵV1∗​, strictly if V1∗>0V^*_1>0V1∗​>0;
  3. every jump bucket holds exactly one variable;
  4. every variable lies in one of the 2K+12K+12K+1 buckets;
  5. on a nonempty bucket, BaseVal(Bi)=Vmax⁡{t:Xσ(t)∈Bi}+1∗\mathrm{BaseVal}(B_i)=V^*_{\max\{t:X_{\sigma(t)}\in B_i\}+1}BaseVal(Bi​)=Vmax{t:Xσ(t)​∈Bi​}+1∗​;
  6. BaseVal(B1)=0\mathrm{BaseVal}(B_1)=0BaseVal(B1​)=0, and at the index above the top bucket BaseVal=V1∗\mathrm{BaseVal}=V^*_1BaseVal=V1∗​.

These are the facts about the partition that Claims 3.4 and 3.5 use.

Differences from the source:

  • correction: jump iff the closed interval [Vt∗,Vt−1∗][V^*_t,V^*_{t-1}][Vt∗​,Vt−1∗​] contains a multiple of ϵV1∗\epsilon V^*_1ϵV1∗​; B2={Xσ∗(TK)}B_2=\{X_{\sigma^*(T_K)}\}B2​={Xσ∗(TK​)​}, p.17 → jump variable at sss iff (Vs+1∗,Vs∗](V^*_{s+1},V^*_s](Vs+1∗​,Vs∗​] contains a multiple; buckets as in Figure 1; reason: ϵ=1/2\epsilon=1/2ϵ=1/2, values (1,0,0,0,0)(1,0,0,0,0)(1,0,0,0,0), σ∗\sigma^*σ∗ the identity: closed intervals give K≥4>1/ϵK\ge4>1/\epsilonK≥4>1/ϵ; values (1,0.6,0.6,0)(1,0.6,0.6,0)(1,0.6,0.6,0): the text’s stable bucket has step 0.6>ϵOPT0.6>\epsilon\mathrm{OPT}0.6>ϵOPT, against Figure 1 and the proof of Claim 3.4
  • generalization: an optimal permutation σ∗\sigma^*σ∗ → every permutation σ\sigmaσ; reason: no step uses optimality; σ∗\sigma^*σ∗ is a permutation
  • reading: BaseVal(B2K+2)\mathrm{BaseVal}(B_{2K+2})BaseVal(B2K+2​), undefined → V1∗V^*_1V1∗​ at the index above the top bucket; reason: the value (15) needs at the top bucket; OPT\mathrm{OPT}OPT for optimal σ\sigmaσ
  • encoding: times and buckets 1,…,n1,\dots,n1,…,n and 1,…,2K+11,\dots,2K+11,…,2K+1 → 0-based Fin n, Fin (2K+1); even index stable, odd index jump; reason: a shift of indices; same objects
Preamble
import Mathlib
import Definitions.Def_FreeOrderEPTAS_bucket
Formal statement
namespace FreeOrderEPTAS

/-- §3.2, pp.17–18 (arXiv:2007.13121v3), for the corrected buckets of any permutation `σ` with
`1/ϵ ∈ ℕ`, where `V*_s = R I σ s` (0-based) and `K = numJumps`:
1. `K ≤ 1/ϵ`;
2. every stable (even) bucket has `BaseVal(B_{i+1}) - BaseVal(B_i) ≤ ϵ V*_0`, strictly when `V*_0 > 0`;
3. every jump (odd) bucket consists of exactly one variable;
4. the buckets are `0, …, 2K`;
5. `BaseVal(B_i) = V*_{max{t : X_{σ(t)} ∈ B_i}+1}` on a nonempty bucket;
6. `BaseVal(B_1) = 0`, and at the index `2K + 1` above the top bucket `BaseVal = V*_0`. -/
theorem bucket_facts {n : ℕ} (I : Instance n) (σ : Equiv.Perm (Fin n)) (ϵ : ℝ) (q : ℕ) (hq : 0 < q)
    (hϵ : ϵ = 1 / (q : ℝ)) :
    (numJumps I σ ϵ : ℝ) ≤ 1 / ϵ ∧
    (∀ i, i < 2 * numJumps I σ ϵ + 1 → Even i →
      BaseVal I σ ϵ (i + 1) - BaseVal I σ ϵ i ≤ ϵ * R I σ 0 ∧
        (0 < R I σ 0 → BaseVal I σ ϵ (i + 1) - BaseVal I σ ϵ i < ϵ * R I σ 0)) ∧
    (∀ i, i < 2 * numJumps I σ ϵ + 1 → Odd i → ∃! s, s < n ∧ bucket I σ ϵ s = i) ∧
    (∀ s, s < n → bucket I σ ϵ s < 2 * numJumps I σ ϵ + 1) ∧
    (∀ s, s < n → (∀ s', s' < n → bucket I σ ϵ s' = bucket I σ ϵ s → s' ≤ s) →
      BaseVal I σ ϵ (bucket I σ ϵ s) = R I σ (s + 1)) ∧
    BaseVal I σ ϵ 0 = 0 ∧
    BaseVal I σ ϵ (2 * numJumps I σ ϵ + 1) = R I σ 0 := by sorry

end FreeOrderEPTAS
Source
Segev & Singla, Efficient Approximation Schemes for Stochastic Probing and Selection-Stopping Problems, arXiv:2007.13121v3, §3.2, pp.17–18, Figure 1
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