Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Eqs. (6)–(8) — E[w(V_1 ∪ C)] ≤ w(V_1) + (1 − k/t)w(V_{1/2}) ≤ 2(1 − k/t)(w(V_1) + ½w(V_{1/2})) ≤ (2 − 2/(t/k)) OPT

Proved
SingleMachinePrec.Framework.eqs_6_8

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

approximationlp-roundingp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1vertex-cover

Let SSS be an instance of 1∣prec∣∑wjCj1|\mathrm{prec}|\sum w_jC_j1∣prec∣∑wj​Cj​ whose precedence order PPP is not a linear order, let L1,…,LtL_1,\dots,L_tL1​,…,Lt​ be a k:tk:tk:t-realizer of PPP, and let xxx be a half-integral optimal solution of [CS-LP], with Va={u:xu=a}V_a = \{u : x_u = a\}Va​={u:xu​=a}. For each iii let Ci=V1∪(V1/2∖I1/2(Li))C_i = V_1 \cup (V_{1/2} \setminus I_{1/2}(L_i))Ci​=V1​∪(V1/2​∖I1/2​(Li​)), where I1/2(Li)I_{1/2}(L_i)I1/2​(Li​) is the set of pairs of V1/2V_{1/2}V1/2​ reversed in LiL_iLi​. Then

1t∑i=1tw(Ci)≤w(V1)+(1−kt)w(V1/2)(6)≤2(1−kt)(w(V1)+12w(V1/2))(7)≤(2−2t/k) OPT,(8)\begin{aligned} \frac1t\sum_{i=1}^t w(C_i) &\le w(V_1) + \Bigl(1-\frac kt\Bigr) w(V_{1/2}) &(6)\\ &\le 2\Bigl(1-\frac kt\Bigr)\Bigl(w(V_1) + \frac12 w(V_{1/2})\Bigr) &(7)\\ &\le \Bigl(2 - \frac{2}{t/k}\Bigr)\,\mathrm{OPT}, &(8) \end{aligned}t1​i=1∑t​w(Ci​)​≤w(V1​)+(1−tk​)w(V1/2​)≤2(1−tk​)(w(V1​)+21​w(V1/2​))≤(2−t/k2​)OPT,​(6)(7)(8)​

where OPT\mathrm{OPT}OPT is the minimum weight of a vertex cover of GPSG^S_PGPS​.

This chain is the cost estimate of Theorem 5.1. Step (7) uses t/k≥2t/k \ge 2t/k≥2, which holds because PPP is not a linear order; step (8) uses that w(V1)+12w(V1/2)w(V_1) + \tfrac12 w(V_{1/2})w(V1​)+21​w(V1/2​) is the optimal value of [CS-LP].

Formalization Note The expectation is the uniform average over the ttt indices. The paper's assumption fdim⁡(P)≥2\operatorname{fdim}(P) \ge 2fdim(P)≥2 is stated as "PPP is not a linear order" (equivalent by p. 655). k/tk/tk/t and 2/(t/k)2/(t/k)2/(t/k) are real divisions.

Preamble
import Mathlib
import Definitions.Def_SingleMachinePrec_Framework_CSLP
Formal statement
namespace SingleMachinePrec.Framework

/-- **Eqs. (6)–(8)** (p. 659). Let `P` (the precedence order of `S`) not be a linear order, let
`L_1, …, L_t` be a `k : t`-realizer of `P`, and let `x` be a half-integral optimal solution of
[CS-LP]. With `C_i = V_1 ∪ (V_{1/2} \ I_{1/2}(L_i))`:
(6) `(1/t) ∑_i w(C_i) ≤ w(V_1) + (1 - k/t) w(V_{1/2})`;
(7) `w(V_1) + (1 - k/t) w(V_{1/2}) ≤ 2 (1 - k/t) (w(V_1) + ½ w(V_{1/2}))`;
(8) `2 (1 - k/t) (w(V_1) + ½ w(V_{1/2})) ≤ (2 - 2/(t/k)) · OPT`. -/
theorem eqs_6_8 {N : Type*} [Fintype N] [DecidableEq N] (S : Instance N)
    (hP : ¬ IsLinearOrder N S.P) (k t : ℕ)
    (L : Fin t → LinearExtension S.P) (hL : IsKFoldRealizer S.P k t L)
    (x : IncPair S.P → ℝ) (hx : IsCSLPOptimal S x) (hhalf : IsHalfIntegral x) :
    (1 / (t : ℝ)) * ∑ i, weight S (roundedCover x (L i)) ≤
        weight S (levelSet x 1) + (1 - (k : ℝ) / t) * weight S (levelSet x (1 / 2)) ∧
      weight S (levelSet x 1) + (1 - (k : ℝ) / t) * weight S (levelSet x (1 / 2)) ≤
        2 * (1 - (k : ℝ) / t) *
          (weight S (levelSet x 1) + 1 / 2 * weight S (levelSet x (1 / 2))) ∧
      2 * (1 - (k : ℝ) / t) *
          (weight S (levelSet x 1) + 1 / 2 * weight S (levelSet x (1 / 2))) ≤
        (2 - 2 / ((t : ℝ) / k)) * OPT S := by sorry

end SingleMachinePrec.Framework
Source
Ambühl, Mastrolilli, Mutsanas, Svensson, On the Approximability of Single-Machine Scheduling with Precedence Constraints, Math. Oper. Res. 36(4) (2011), p. 659, Eqs. (6), (7), (8) (proof of Theorem 5.1)
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