Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 4.1, p. 294 — E[C(B)]=2∑t=1T−LE[Zt]E[\mathcal C(B)] = 2\sum_{t=1}^{T-L} E[Z_t]E[C(B)]=2∑t=1T−L​E[Zt​]

Proved
LeviBalancing.DualBalancing.expected_cost_eq_two_sum_balanced

by mikedeng1 · Oct 5, 2026 · Mathlib 0df444a (Lean v4.33.1)

approximation-algorithmsconditional-expectationinventory-controlp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Let BBB be a dual-balancing policy for an instance with horizon TTT and lead time LLL, under a demand process on a probability space with filtration (Ft)(\mathcal F_t)(Ft​). For t=1,…,T−Lt = 1, \dots, T - Lt=1,…,T−L let

Zt=E[HtB∣Ft]  (=E[ΠtB∣Ft] a.s.)Z_t = E\bigl[H^B_t \mid \mathcal F_t\bigr] \;(= E[\Pi^B_t \mid \mathcal F_t] \text{ a.s.})Zt​=E[HtB​∣Ft​](=E[ΠtB​∣Ft​] a.s.)

be the balanced cost of period ttt. Then the expected cost of BBB is twice the expected sum of the balanced costs:

E[C(B)]=2∑t=1T−LE[Zt].E[\mathcal C(B)] = 2 \sum_{t=1}^{T-L} E[Z_t].E[C(B)]=2t=1∑T−L​E[Zt​].

This is the first half of the amortization in the proof of Theorem 4.1: the cost of the dual-balancing policy is expressed through the quantities ZtZ_tZt​, which the comparison with an arbitrary policy then bounds.

Formalization Note Both sides are in [0,∞][0,\infty][0,∞]: E[C(B)]E[\mathcal C(B)]E[C(B)] is the lower Lebesgue integral of the realized cost, and E[Zt]E[Z_t]E[Zt​] is the lower Lebesgue integral of the positive part of a version of the conditional expectation (which is almost surely nonnegative).

Preamble
import Mathlib
import Definitions.Def_LeviBalancing_DualBalancing_Model
import Definitions.Def_LeviBalancing_DualBalancing_Policy
open Finset MeasureTheory
open scoped ENNReal
Formal statement
namespace LeviBalancing.DualBalancing

/-- Lemma 4.1, p. 294: the expected cost of a dual-balancing policy is twice the expected sum of
the balanced costs `Z_t = E[H_t^B | ℱ t] = E[Π_t^B | ℱ t]`, `t = 1, …, T − L`. -/
theorem expected_cost_eq_two_sum_balanced {Ω : Type*} [m : MeasurableSpace Ω] {μ : Measure Ω}
    [IsProbabilityMeasure μ] {ℱ : Filtration ℤ m} (I : Instance) (dem : DemandProcess ℱ μ I.T)
    (B : ℤ → Ω → ℝ) (hB : IsDualBalancing I ℱ μ dem.D B) :
    expectedCost I μ dem.D B =
      2 * ∑ t ∈ Icc (1 : ℤ) ((I.T : ℤ) - I.L),
        ∫⁻ ω, ENNReal.ofReal ((μ[randHolding I dem.D B t | ℱ t]) ω) ∂μ := by sorry

end LeviBalancing.DualBalancing
Source
Levi, Pál, Roundy & Shmoys, Approximation Algorithms for Stochastic Inventory Control Models, Math. Oper. Res. 32(2):284–302 (2007), DOI 10.1287/moor.1060.0205, p. 294 (PDF p. 11), Lemma 4.1
Human review
  • Endorsed by Shuze Chen · Oct 5, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 5, 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