Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Eq. (33) — the DLP value changes by at most Σ_l r^l_max (b_l − b'_l)⁺ when the capacity right-hand side drops

Proved
ResolvingNRM.FRUpper.lp_rhs_sensitivity

by mikedeng1 · Oct 6, 2026 · Mathlib 0df444a (Lean v4.33.1)

linear-programmingnetwork-revenue-managementp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1sensitivity-analysis

Let A∈Rm×nA \in \mathbb{R}^{m\times n}A∈Rm×n have nonnegative entries alja_{lj}alj​, let r≥0r \ge 0r≥0 and λj>0\lambda_j > 0λj​>0, and for b≥0b \ge 0b≥0 let

v(b)=max⁡{∑j=1nrjxj : ∑j=1nAjxj≤b, 0≤xj≤λj}.v(b) = \max\Big\{ \sum_{j=1}^n r_j x_j \ :\ \sum_{j=1}^n A_j x_j \le b,\ 0 \le x_j \le \lambda_j \Big\}.v(b)=max{j=1∑n​rj​xj​ : j=1∑n​Aj​xj​≤b, 0≤xj​≤λj​}.

With rmax⁡l=max⁡j∈[n]{rj I(alj>0)/alj}r^l_{\max} = \max_{j \in [n]} \{ r_j \, \mathbb{I}(a_{lj} > 0)/a_{lj} \}rmaxl​=maxj∈[n]​{rj​I(alj​>0)/alj​}, for all capacity vectors b,b′≥0b, b' \ge 0b,b′≥0,

v(b)−v(b′)≤∑l=1mrmax⁡l (bl−bl′)+.v(b) - v(b') \le \sum_{l=1}^m r^l_{\max} \, (b_l - b'_l)^+ .v(b)−v(b′)≤l=1∑m​rmaxl​(bl​−bl′​)+.

Applied with b=C/Tb = C/Tb=C/T and b′=b(t)b' = b(t)b′=b(t), the average remaining capacity of the FR policy in period ttt, this is eq. (33) of the paper: it converts the per-period gap between the DLP and the LP that FR solves into the positive parts (bl−bl(t))+(b_l - b_l(t))^+(bl​−bl​(t))+, which Lemma 8 controls. The number rmax⁡lr^l_{\max}rmaxl​ bounds the dual price of resource lll.

Formalization Note. The paper states (33) for the two LPs it compares (bbb and b(t)b(t)b(t)); the statement here is for arbitrary right-hand sides b,b′≥0b, b' \ge 0b,b′≥0, as the paper's justification ("only differ in the right hand side") is. The printed sum runs over l=1,…,Ll = 1, \dots, Ll=1,…,L; the paper has mmm resources, so it is read as l∈[m]l \in [m]l∈[m]. λj>0\lambda_j > 0λj​>0, r≥0r \ge 0r≥0 and alj≥0a_{lj} \ge 0alj​≥0 are the model's standing assumptions (p. 7). With b,b′≥0b, b' \ge 0b,b′≥0 the LP is feasible and bounded, so piValue is its maximum.

Preamble
import Mathlib
import Definitions.Def_RLPBidPrice_Unbiased_Model
import Definitions.Def_ResolvingNRM_FRUpper_Model

open RLPBidPrice.Unbiased Matrix
Formal statement
namespace ResolvingNRM.FRUpper

/-- Eq. (33), App. C.2, p. 35: two DLPs (2) that differ only in the capacity right-hand side
satisfy `v(b) − v(b') ≤ ∑_l r^l_max (b_l − b'_l)⁺`. -/
theorem lp_rhs_sensitivity {m n : ℕ} (A : Matrix (Fin m) (Fin n) ℝ) (r lam : Fin n → ℝ)
    (hlam : ∀ j, 0 < lam j) (hr : 0 ≤ r) (hA : ∀ l j, 0 ≤ A l j)
    (b b' : Fin m → ℝ) (hb : 0 ≤ b) (hb' : 0 ≤ b') :
    piValue A r b lam - piValue A r b' lam ≤ ∑ l, rmaxRes A r l * max (b l - b' l) 0 := by sorry

end ResolvingNRM.FRUpper
Source
Bumpensanti, Wang, A Re-solving Heuristic with Uniformly Bounded Loss for Network Revenue Management, arXiv:1802.06192v3, App. C.2, eq. (33), p. 35
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