Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proposition 3 — v^DLP − v^FR = O(√T), uniformly in the capacities

Proved
ResolvingNRM.FRUpper.fr_dlp_gap_sqrt

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

network-revenue-managementp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1re-solvingregret-bound

Consider the network revenue management model with nnn customer classes and mmm resources: Poisson arrival rates λj>0\lambda_j > 0λj​>0, revenues rj≥0r_j \ge 0rj​≥0, and a nonnegative bill-of-materials matrix A=(alj)A = (a_{lj})A=(alj​). For an integer horizon T≥1T \ge 1T≥1 and initial capacity C≥0C \ge 0C≥0, let vDLP(T,C)=T v(C/T)v^{\mathrm{DLP}}(T, C) = T\, v(C/T)vDLP(T,C)=Tv(C/T) be the value of the deterministic LP (2), and vFR(T,C)v^{\mathrm{FR}}(T, C)vFR(T,C) the expected revenue of the Frequent Re-solving policy (Algorithm 2), which re-solves the LP at every integer time and uses any optimal solution. There is a constant MMM, depending only on λ\lambdaλ, rrr and AAA, such that for every optimal-solution selector, every T≥1T \ge 1T≥1 and every C≥0C \ge 0C≥0,

vDLP(T,C)−vFR(T,C) ≤ MT.v^{\mathrm{DLP}}(T, C) - v^{\mathrm{FR}}(T, C) \ \le\ M \sqrt{T}.vDLP(T,C)−vFR(T,C) ≤ MT​.

Since vDLPv^{\mathrm{DLP}}vDLP upper-bounds the expected revenue of every admissible policy, the revenue loss of FR is O(T)O(\sqrt{T})O(T​) without any nondegeneracy assumption on the DLP. The paper combines this with Proposition 2, an Ω(T)\Omega(\sqrt{T})Ω(T​) lower bound on a degenerate instance, to conclude v∗−vFR=Θ(T)v^* - v^{\mathrm{FR}} = \Theta(\sqrt{T})v∗−vFR=Θ(T​) there.

Formalization Note. "=O(T)= O(\sqrt{T})=O(T​)" is formalized with the constant chosen after (n,m,λ,r,A)(n, m, \lambda, r, A)(n,m,λ,r,A) and before the selector, TTT and CCC; this is the paper's statement that the constant "does not depend on the starting capacity ClC_lCl​". The paper's threshold T1T_1T1​ in the definition of O(⋅)O(\cdot)O(⋅) (p. 7) is dropped: the bound is stated for all T≥1T \ge 1T≥1, which is equivalent because 0≤vFR0 \le v^{\mathrm{FR}}0≤vFR and vDLP≤T∑jrjλjv^{\mathrm{DLP}} \le T \sum_j r_j \lambda_jvDLP≤T∑j​rj​λj​, so the gap is bounded on any finite range of TTT uniformly in CCC. λj>0\lambda_j > 0λj​>0, r≥0r \ge 0r≥0, alj≥0a_{lj} \ge 0alj​≥0, C≥0C \ge 0C≥0 are the model's standing assumptions (p. 7).

Preamble
import Mathlib
import Definitions.Def_RLPBidPrice_Unbiased_Model
import Definitions.Def_ResolvingNRM_FRUpper_Model

open RLPBidPrice.Unbiased Matrix
Formal statement
namespace ResolvingNRM.FRUpper

/-- Proposition 3, p. 19: `v^DLP − v^FR = O(√T)`, with a constant that depends on `λ`, `r` and `A`
but not on the starting capacity `C` (nor on the optimal LP solutions chosen). -/
theorem fr_dlp_gap_sqrt {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) :
    ∃ M : ℝ, ∀ sel : (Fin m → ℝ) → (Fin n → ℝ), ResolvingNRM.IRT.IsDLPSelector A r lam sel →
      ∀ T : ℕ, 1 ≤ T → ∀ C : Fin m → ℝ, 0 ≤ C →
        ResolvingNRM.IRT.dlpValue A r lam T C - frValue A r lam sel T C ≤ M * Real.sqrt T := by sorry

end ResolvingNRM.FRUpper
Source
Bumpensanti, Wang, A Re-solving Heuristic with Uniformly Bounded Loss for Network Revenue Management, arXiv:1802.06192v3, Proposition 3, p. 19 (proof App. C.2, pp. 34–36)
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