Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 3.2 — (IPProphets) is feasible (corrected: guess grid, buckets)

Proved
FreeOrderEPTAS.lemma_3_2

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

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

Let σ∗\sigma^*σ∗ be optimal and 1/ϵ1/\epsilon1/ϵ a positive integer. For all guesses satisfying (14) and (15), (IPProphets) admits a feasible solution. Moreover, if ϵ<1\epsilon<1ϵ<1 and E∈[(1−ϵ)OPT,OPT]E\in[(1-\epsilon)\mathrm{OPT},\mathrm{OPT}]E∈[(1−ϵ)OPT,OPT], then guesses satisfying (14) and (15) exist on the grid {kϵ2E:0≤k≤⌈1/(ϵ2(1−ϵ))⌉}\{k\epsilon^2E:0\le k\le\lceil1/(\epsilon^2(1-\epsilon))\rceil\}{kϵ2E:0≤k≤⌈1/(ϵ2(1−ϵ))⌉}.

This is why the enumeration of §3.3 reaches a feasible Santa Claus instance.

Differences from the source:

  • correction: grid {0,ϵ2E,…,(1−ϵ2)E}\{0,\epsilon^2E,\dots,(1-\epsilon^2)E\}{0,ϵ2E,…,(1−ϵ2)E}, p.18 → grid {kϵ2E:0≤k≤⌈1/(ϵ2(1−ϵ))⌉}\{k\epsilon^2E:0\le k\le\lceil1/(\epsilon^2(1-\epsilon))\rceil\}{kϵ2E:0≤k≤⌈1/(ϵ2(1−ϵ))⌉}; reason: BaseVal\mathrm{BaseVal}BaseVal can exceed EEE: ϵ=1/2\epsilon=1/2ϵ=1/2, x=(1,0.9)x=(1,0.9)x=(1,0.9), σ∗\sigma^*σ∗ the identity, E=0.5E=0.5E=0.5, where (14) needs a guess in [0.775,0.9][0.775,0.9][0.775,0.9], above the printed top 0.3750.3750.375
  • 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
  • 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
  • encoding: the guesses are read from the enumerated grid → non-negativity is a clause of IsGuess; reason: the grid is non-negative; p.20 uses "the non-negativity of BaseGuess"
  • reading: BaseVal(B2K+2)\mathrm{BaseVal}(B_{2K+2})BaseVal(B2K+2​), needed in (15) for the top bucket, undefined → V1∗V^*_1V1∗​ (BaseVal at the index 2K+12K+12K+1); reason: the value (15) needs at the top bucket; it is OPT\mathrm{OPT}OPT for optimal σ∗\sigma^*σ∗
Preamble
import Mathlib
import Definitions.Def_FreeOrderEPTAS_IsGuess
import Definitions.Def_FreeOrderEPTAS_IPProphets
import Definitions.Def_FreeOrderEPTAS_OPT
import Definitions.Def_FreeOrderEPTAS_freeOrder
Formal statement
namespace FreeOrderEPTAS

/-- Lemma 3.2, p.19 (arXiv:2007.13121v3), for `σ*` optimal and `1/ϵ ∈ ℕ`: for guesses satisfying (14)
and (15), (IPProphets) admits a feasible solution. Also, when `E ∈ [(1 - ϵ) OPT, OPT]` with
`ϵ < 1`, such guesses exist on the corrected grid `{k ϵ²E : 0 ≤ k ≤ ⌈1/(ϵ²(1 - ϵ))⌉}`. -/
theorem lemma_3_2 {n : ℕ} (I : Instance n) (σs : Equiv.Perm (Fin n)) (hopt : V I σs = OPT I)
    (ϵ E : ℝ) (q : ℕ) (hq : 0 < q) (hϵ : ϵ = 1 / (q : ℝ)) :
    (∀ b Δ : Fin (2 * numJumps I σs ϵ + 1) → ℝ, IsGuess I σs ϵ E b Δ →
      IPProphetsFeasible I b Δ) ∧
    (ϵ < 1 → (1 - ϵ) * OPT I ≤ E → E ≤ OPT I →
      ∃ b Δ : Fin (2 * numJumps I σs ϵ + 1) → ℝ, IsGuess I σs ϵ E b Δ ∧
        ∀ i, b i ∈ guessGrid ϵ E ⌈1 / (ϵ ^ 2 * (1 - ϵ))⌉₊ ∧
          Δ i ∈ guessGrid ϵ E ⌈1 / (ϵ ^ 2 * (1 - ϵ))⌉₊) := by sorry

end FreeOrderEPTAS
Source
Segev & Singla, Efficient Approximation Schemes for Stochastic Probing and Selection-Stopping Problems, arXiv:2007.13121v3, Lemma 3.2, p.19; (14), (15), p.18
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