Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Observation 1, p. 10 β€” 𝔣_d is jointly convex, and the resulting Jensen bound on E[π“˜*]

Proved
LostSalesExpGap.ConstOrder.observation_1

by mikedeng1 Β· Oct 7, 2026 Β· Mathlib 0df444a (Lean v4.33.1)

convexityjensenlost-salesp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1

Fix Lβ‰₯1L\ge1Lβ‰₯1. For d∈RL\mathbf d\in\mathbb R^Ld∈RL and Ο‡1,…,Ο‡L,I∈R\chi_1,\dots,\chi_L,I\in\mathbb RΟ‡1​,…,Ο‡L​,I∈R let

fd(Ο‡1,…,Ο‡L,I)=max⁑j=0,…,L(βˆ‘i=1j(Ο‡L+1βˆ’iβˆ’dL+1βˆ’i)+Ξ΄j,LI).\mathfrak f_{\mathbf d}(\chi_1,\dots,\chi_L,I)=\max_{j=0,\dots,L}\Big(\sum_{i=1}^j(\chi_{L+1-i}-d_{L+1-i})+\delta_{j,L}I\Big).fd​(Ο‡1​,…,Ο‡L​,I)=j=0,…,Lmax​(i=1βˆ‘j​(Ο‡L+1βˆ’iβ€‹βˆ’dL+1βˆ’i​)+Ξ΄j,L​I).
  1. For each fixed d\mathbf dd, fd\mathfrak f_{\mathbf d}fd​ is jointly convex on RL+1\mathbb R^{L+1}RL+1.
  2. Let D1,D2,…D_1,D_2,\dotsD1​,D2​,… be i.i.d. nonnegative demands with finite mean, and let (Ο‡βˆ—,Iβˆ—)(\boldsymbol\chi^*,\mathcal I^*)(Ο‡βˆ—,Iβˆ—) be a random vector, independent of the demands, such that (i) it is nonnegative with probability one and has finite mean; (ii) E[Iβˆ—]=E[f(D1,…,DL)(Ο‡βˆ—,Iβˆ—)]\mathbb E[\mathcal I^*]=\mathbb E\big[\mathfrak f_{(D_1,\dots,D_L)}(\boldsymbol\chi^*,\mathcal I^*)\big]E[Iβˆ—]=E[f(D1​,…,DL​)​(Ο‡βˆ—,Iβˆ—)]; (iii) E[Ο‡iβˆ—]=E[Ο‡1βˆ—]≀E[D]\mathbb E[\chi^*_i]=\mathbb E[\chi^*_1]\le\mathbb E[D]E[Ο‡iβˆ—β€‹]=E[Ο‡1βˆ—β€‹]≀E[D] for all iii. Then
E[Iβˆ—]β‰₯E[max⁑j=0,…,L(j E[Ο‡1βˆ—]βˆ’βˆ‘i=1jDi+Ξ΄j,LE[Iβˆ—])].\mathbb E[\mathcal I^*]\ge\mathbb E\Big[\max_{j=0,\dots,L}\Big(j\,\mathbb E[\chi^*_1]-\sum_{i=1}^jD_i+\delta_{j,L}\mathbb E[\mathcal I^*]\Big)\Big].E[Iβˆ—]β‰₯E[j=0,…,Lmax​(jE[Ο‡1βˆ—β€‹]βˆ’i=1βˆ‘j​Di​+Ξ΄j,L​E[Iβˆ—])].

Conditions (i)–(iii) are those of the paper's Theorem 2, which constructs such a vector from a stationary optimal policy; this observation is how the optimal policy is compared with the constant-order policy ordering E[Ο‡1βˆ—]\mathbb E[\chi^*_1]E[Ο‡1βˆ—β€‹].

Formalization Note The law of (Ο‡βˆ—,Iβˆ—)(\boldsymbol\chi^*,\mathcal I^*)(Ο‡βˆ—,Iβˆ—) is an arbitrary probability measure Ξ½\nuΞ½ on RLΓ—R\mathbb R^L\times\mathbb RRLΓ—R satisfying (i)–(iii) as hypotheses; independence from the demands is the product measure Ξ½βŠ—DβŠ—N\nu\otimes\mathcal D^{\otimes\mathbb N}Ξ½βŠ—DβŠ—N. Theorem 2 itself is not part of this mission.

Preamble
import Mathlib
import Definitions.Def_LostSalesExpGap_ConstOrder_ConstantOrder

open MeasureTheory ProbabilityTheory
open scoped ENNReal NNReal BigOperators
Formal statement
namespace LostSalesExpGap.ConstOrder

/-- Observation 1, p. 10. (a) For each `d ∈ ℝ^L`, `𝔣_d(Ο‡_1, …, Ο‡_L, I)` is jointly convex on
`ℝ^{L+1}`. (b) For every law `Ξ½` of `(Ο‡*, π“˜*)` with Theorem 2.(i)–(iii) (nonnegative, finite mean,
independent of the demands, `E[π“˜*] = E[𝔣_{(D_1,…,D_L)}(Ο‡*, π“˜*)]`, `E[Ο‡*_i] = E[Ο‡*_1] ≀ E[D]`),
`E[π“˜*] β‰₯ E[max_{j=0,…,L} (j E[Ο‡*_1] βˆ’ Ξ£_{i=1}^j D_i + Ξ΄_{j,L} E[π“˜*])]`. -/
theorem observation_1 (P : CappedBaseStock.DemandLaw) (L : β„•) (hL : 1 ≀ L) :
    (βˆ€ d : Fin L β†’ ℝ, ConvexOn ℝ Set.univ (fun v : (Fin L β†’ ℝ) Γ— ℝ => fObs L d v.1 v.2)) ∧
    (βˆ€ Ξ½ : Measure ((Fin L β†’ ℝ) Γ— ℝ), IsProbabilityMeasure Ξ½ β†’
      (βˆ€α΅ v βˆ‚Ξ½, (βˆ€ i, 0 ≀ v.1 i) ∧ 0 ≀ v.2) β†’
      (βˆ€ i, Integrable (fun v : (Fin L β†’ ℝ) Γ— ℝ => v.1 i) Ξ½) β†’
      Integrable (fun v : (Fin L β†’ ℝ) Γ— ℝ => v.2) Ξ½ β†’
      ∫ v, v.2 βˆ‚Ξ½ =
        ∫ w, fObs L (firstDemands L w.2) w.1.1 w.1.2
          βˆ‚(Ξ½.prod (CappedBaseStock.demandPathLaw P)) β†’
      (βˆ€ i, ∫ v, v.1 i βˆ‚Ξ½ = ∫ v, v.1 ⟨0, hL⟩ βˆ‚Ξ½) β†’
      ∫ v, v.1 ⟨0, hL⟩ βˆ‚Ξ½ ≀ CappedBaseStock.mean P β†’
      ∫ d, (Finset.range (L + 1)).sup' Finset.nonempty_range_add_one (fun j =>
          (j : ℝ) * ∫ v, v.1 ⟨0, hL⟩ βˆ‚Ξ½ - βˆ‘ i ∈ Finset.Icc 1 j, (d (i - 1) : ℝ) +
            if j = L then ∫ v, v.2 βˆ‚Ξ½ else 0) βˆ‚CappedBaseStock.demandPathLaw P
        ≀ ∫ v, v.2 βˆ‚Ξ½) := by sorry

end LostSalesExpGap.ConstOrder
Source
Xin, Goldberg, Optimality gap of constant-order policies decays exponentially in the lead time for lost sales models, arXiv:1409.1499v2, Observation 1, p. 10 (with Theorem 2.(i)–(iii), pp. 9–10)
Human review
  • Endorsed by Shuze Chen Β· Oct 8, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 Β· Oct 8, 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
Β© 2026 Prove2Me