Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 1: for any μ ≥ 0, J_k(x) ≤ J̄_k(x, μ)

Proved
BidPriceControl.Asymptotic.lemma1_upper_bound

by mikedeng1 · Oct 3, 2026 · Mathlib 0df444a (Lean v4.33.1)

dynamic-programmingp2o-batch-rm-cafterp2o-gran-per-chapterp2o-plan-paperp2o-v1revenue-management

Under the standing assumptions of §1 on the request laws F1,…,FkF_1, \dots, F_kF1​,…,Fk​, for every bid-price vector μ≥0\mu \ge 0μ≥0 the optimal expected revenue of the Bellman equations (2)–(3) is bounded by the relaxed value (8):

Jk(x)≤Jˉk(x,μ)=∑t=1k∑j=1nE (Rtj−μAj)++μx.J_k(x) \le \bar J_k(x, \mu) = \sum_{t=1}^k\sum_{j=1}^n \mathbb{E}\,(R_t^j - \mu A^j)^+ + \mu x.Jk​(x)≤Jˉk​(x,μ)=t=1∑k​j=1∑n​E(Rtj​−μAj)++μx.

Minimizing the right side over μ≥0\mu \ge 0μ≥0 gives the upper bound vk(x)v_k(x)vk​(x) of (9), the benchmark against which the heuristic is measured.

Formalization Note. The paper defines Jˉk(x,μ)\bar J_k(x, \mu)Jˉk​(x,μ) as the value (7) of a relaxed control problem and evaluates it as (8). This item takes (8) as the definition, so it is Lemma 1 combined with the evaluation (8).

Preamble
import Mathlib
import Definitions.Def_BidPriceControl_Asymptotic_Model
Formal statement
namespace BidPriceControl.Asymptotic

open MeasureTheory ProbabilityTheory

/-- Lemma 1 (p. 1583), with the relaxed value in its evaluated form (8): under the standing
assumptions of §1, for any `μ ≥ 0`, `J_k(x) ≤ J̄_k(x, μ)`, where `J` is the optimal expected revenue
of the Bellman equations (2)–(3). -/
theorem lemma1_upper_bound {m n : ℕ} (A : Fin m → Fin n → ℕ) (F : ℕ → Measure (Fin n → ℝ))
    (k : ℕ) (x : Fin m → ℕ) (hF : ∀ t ∈ Finset.Icc 1 k, IsRequestLaw (F t))
    (μ : Fin m → ℝ) (hμ : 0 ≤ μ) :
    optValue A F k x ≤ relaxedValue A F k x μ := by sorry

end BidPriceControl.Asymptotic
Source
Talluri, van Ryzin, An Analysis of Bid-Price Controls for Network Revenue Management, Management Sci. 44(11), 1998, p. 1583, Lemma 1 (with the evaluation (8))
Read-back

What the Lean code literally says, in plain math · claude-opus-5-5

Data. The theorem takes a natural-number matrix A=(aij)∈Nm×nA=(a_{ij})\in\mathbb N^{m\times n}A=(aij​)∈Nm×n and a sequence of measures (Ft)t∈N(F_t)_{t\in\mathbb N}(Ft​)t∈N​ on Rn\mathbb R^nRn. It also takes k∈Nk\in\mathbb Nk∈N, a capacity x∈Nmx\in\mathbb N^mx∈Nm, and μ∈Rm\mu\in\mathbb R^mμ∈Rm with μi≥0\mu_i\ge0μi​≥0 for all iii.

Hypothesis. For every t∈{1,…,k}t\in\{1,\dots,k\}t∈{1,…,k}, FtF_tFt​ is a request law:

  • FtF_tFt​ is a probability measure.
  • FtF_tFt​-a.e. rrr has all coordinates ≥0\ge0≥0 with at most one strictly positive.
  • Ft(rj=s)=0F_t(r_j=s)=0Ft​(rj​=s)=0 for every jjj and every s>0s>0s>0.
  • Every coordinate rjr_jrj​ is FtF_tFt​-integrable.

No bounded-support hypothesis is made, and nothing is assumed about F0F_0F0​ or FtF_tFt​ for t>kt>kt>k.

The value Jk(x)J_k(x)Jk​(x). It is defined by the recursion

  • J0≡0J_0\equiv0J0​≡0,
  • Jt(y)=∫max⁡(Jt−1(y),max⁡j:Aj≤y(rj+Jt−1(y−Aj))) dFt(r)J_{t}(y)=\int\max\big(J_{t-1}(y),\max_{j:A^j\le y}(r_j+J_{t-1}(y-A^j))\big)\,dF_t(r)Jt​(y)=∫max(Jt−1​(y),maxj:Aj≤y​(rj​+Jt−1​(y−Aj)))dFt​(r).

In words: in each period you either reject, or accept one itinerary jjj whose column fits the remaining capacity, earning rjr_jrj​ and consuming AjA^jAj. Any non-integrable integrand counts as 000.

Conclusion.

Jk(x)  ≤  ∑t=1k∑j∫(rj−μAj)+ dFt(r)+∑iμixi,μAj=∑iμiaij.J_k(x)\;\le\;\sum_{t=1}^{k}\sum_{j}\int\big(r_j-\mu A^j\big)^+\,dF_t(r)+\sum_i\mu_ix_i,\qquad \mu A^j=\sum_i\mu_ia_{ij}.Jk​(x)≤t=1∑k​j∑​∫(rj​−μAj)+dFt​(r)+i∑​μi​xi​,μAj=i∑​μi​aij​.

Degenerate cases.

  • k=0k=0k=0: the claim is 0≤∑iμixi0\le\sum_i\mu_ix_i0≤∑i​μi​xi​.
  • m=0m=0m=0: every itinerary always fits and μAj=0\mu A^j=0μAj=0.
  • n=0n=0n=0: each FtF_tFt​ is the point mass on the single point of R0\mathbb R^0R0 and both sides are 000.
  • All-zero columns: an itinerary with Aj=0A^j=0Aj=0 fits every capacity and has bid-price sum 000.
Human review
  • Endorsed by Shuze Chen · Oct 21, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 21, 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