Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Explicit bound: J^H_θk(θx) ≥ θ v_k(x) − (C/2)√θ Σ_i √(Σ_t Σ_j Var(Y_ijt))

Proved
BidPriceControl.Asymptotic.explicit_lower_bound

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

p2o-batch-rm-cafterp2o-gran-per-chapterp2o-plan-paperp2o-v1probabilityrevenue-management

Assume the standing assumptions of §1 and Rtj≤CR_t^j \le CRtj​≤C almost surely, and let μ∗\mu^*μ∗ be an optimal solution of (9) for (x,k)(x, k)(x,k), so that vk(x)=Jˉk(x,μ∗)v_k(x) = \bar J_k(x, \mu^*)vk​(x)=Jˉk​(x,μ∗). Then for every integer θ≥1\theta \ge 1θ≥1 the fixed-bid-price heuristic with bid prices μ∗\mu^*μ∗ earns, in the θ\thetaθ-scaled problem,

JθkH(θx)≥θ vk(x)−C2θ∑i=1m∑t=1k∑j=1nVar⁡(Yijt),J^H_{\theta k}(\theta x) \ge \theta\,v_k(x) - \frac{C}{2}\sqrt{\theta}\sum_{i=1}^m\sqrt{\sum_{t=1}^k\sum_{j=1}^n \operatorname{Var}(Y_{ijt})},JθkH​(θx)≥θvk​(x)−2C​θ​i=1∑m​t=1∑k​j=1∑n​Var(Yijt​)​,

where Yijt=1{Rtj>μ∗Aj} aijY_{ijt} = \mathbf{1}\{R_t^j > \mu^* A^j\}\,a_{ij}Yijt​=1{Rtj​>μ∗Aj}aij​ and Var⁡(Yijt)=aij2 ptj(1−ptj)\operatorname{Var}(Y_{ijt}) = a_{ij}^2\,p_{tj}(1 - p_{tj})Var(Yijt​)=aij2​ptj​(1−ptj​) with ptj=P(Rtj>μ∗Aj)p_{tj} = P(R_t^j > \mu^* A^j)ptj​=P(Rtj​>μ∗Aj). This is the paper's JθkH(θx)≥θvk(x)−O(θ)J^H_{\theta k}(\theta x) \ge \theta v_k(x) - O(\sqrt\theta)JθkH​(θx)≥θvk​(x)−O(θ​) with the explicit constant of p. 1586.

Formalization Note. The constant is carried exactly as printed. Var⁡(Yijt)\operatorname{Var}(Y_{ijt})Var(Yijt​) is the Bernoulli closed form under the original laws FtF_tFt​, t=1,…,kt = 1, \dots, kt=1,…,k.

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

open MeasureTheory ProbabilityTheory

/-- The last display of the proof of Theorem 1 (p. 1586): if the revenues are bounded by `C` (a.s.)
and `μ*` is an optimal solution of (9), then for every integer `θ ≥ 1`
`J^H_{θk}(θx) ≥ θ v_k(x) − (C/2) √θ ∑_i √(∑_{t=1}^k ∑_j Var(Y_ijt))`,
where `v_k(x) = J̄_k(x, μ*)` and `Y_ijt = 1{R_t^j > μ*A^j} a_ij`. -/
theorem explicit_lower_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))
    (C : ℝ) (hC : ∀ t ∈ Finset.Icc 1 k, ∀ᵐ r ∂(F t), ∀ j, r j ≤ C)
    (μstar : Fin m → ℝ) (hmin : IsRelaxedMinimizer A F k x μstar) (θ : ℕ) (hθ : 1 ≤ θ) :
    (θ : ℝ) * relaxedValue A F k x μstar
        - C / 2 * Real.sqrt (θ : ℝ) *
          ∑ i, Real.sqrt (∑ t ∈ Finset.Icc 1 k, ∑ j, indicatorVar A F μstar i t j)
      ≤ heuristicValue A (scaledLaw F θ) (θ * k) (θ • x) μstar := 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. 1586, proof of Theorem 1, last display
Read-back

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

Data. The theorem takes A=(aij)∈Nm×nA=(a_{ij})\in\mathbb N^{m\times n}A=(aij​)∈Nm×n, measures (Ft)t(F_t)_t(Ft​)t​ on Rn\mathbb R^nRn, k∈Nk\in\mathbb Nk∈N, x∈Nmx\in\mathbb N^mx∈Nm, a real CCC, μ∗∈Rm\mu^*\in\mathbb R^mμ∗∈Rm and θ∈N\theta\in\mathbb Nθ∈N with θ≥1\theta\ge1θ≥1.

Hypotheses.

  • For t∈{1,…,k}t\in\{1,\dots,k\}t∈{1,…,k}, FtF_tFt​ is a request law: a probability measure, a.s. nonnegative with at most one positive coordinate, no positive-value atoms, and integrable coordinates.
  • For each such ttt, FtF_tFt​-a.e. rrr has rj≤Cr_j\le Crj​≤C for every jjj.
  • μ∗≥0\mu^*\ge0μ∗≥0 minimizes Jˉk(x,μ)=∑t=1k∑j∫(rj−μAj)+dFt+∑iμixi\bar J_k(x,\mu)=\sum_{t=1}^k\sum_j\int(r_j-\mu A^j)^+dF_t+\sum_i\mu_ix_iJˉk​(x,μ)=∑t=1k​∑j​∫(rj​−μAj)+dFt​+∑i​μi​xi​ over μ≥0\mu\ge0μ≥0, where μAj=∑iμiaij\mu A^j=\sum_i\mu_ia_{ij}μAj=∑i​μi​aij​.

Notation.

  • ptj=Ft({r:rj>μ∗Aj})p_{tj}=F_t(\{r:r_j>\mu^*A^j\})ptj​=Ft​({r:rj​>μ∗Aj}).
  • JHJ^HJH is the expected revenue of the following heuristic under the product measure on (Rn)θk(\mathbb R^n)^{\theta k}(Rn)θk whose chronological coordinate sss has law F⌈(θk−s)/θ⌉F_{\lceil(\theta k-s)/\theta\rceil}F⌈(θk−s)/θ⌉​. It starts from capacity θx\theta xθx and processes periods in order. In each period it sells the smallest jjj with rj>μ∗Ajr_j>\mu^*A^jrj​>μ∗Aj that fits the current capacity, earning rjr_jrj​ and subtracting AjA^jAj. Otherwise it sells nothing.

Conclusion.

θ Jˉk(x,μ∗)  −  C2 θ ∑i∑t=1k∑jaij2 ptj(1−ptj)  ≤  JH.\theta\,\bar J_k(x,\mu^*)\;-\;\frac C2\,\sqrt\theta\,\sum_i\sqrt{\sum_{t=1}^k\sum_j a_{ij}^2\,p_{tj}(1-p_{tj})}\;\le\;J^H.θJˉk​(x,μ∗)−2C​θ​i∑​t=1∑k​j∑​aij2​ptj​(1−ptj​)​≤JH.

Degenerate cases.

  • No minimizer: if none exists, the statement is vacuous.
  • Sign of CCC: C≥0C\ge0C≥0 is forced when k,n≥1k,n\ge1k,n≥1. When k=0k=0k=0 or n=0n=0n=0, CCC is unconstrained.
  • k=0k=0k=0: the claim is θ∑iμi∗xi≤0\theta\sum_i\mu^*_ix_i\le0θ∑i​μi∗​xi​≤0.
  • n=0n=0n=0: the claim is also θ∑iμi∗xi≤0\theta\sum_i\mu^*_ix_i\le0θ∑i​μi∗​xi​≤0.
  • Square roots: every argument is nonnegative, since ptj∈[0,1]p_{tj}\in[0,1]ptj​∈[0,1].
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