Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Appendix §5.1 display — ΔshA(q)\Delta_s h^{\mathcal A}(q)Δs​hA(q) as an integral, last line corrected to 1λsP(τs(qs+1)≤τA∖{s}∧T)\frac1{\lambda_s}\mathbb P(\tau_s(q_s+1)\le\tau_{\mathcal A\setminus\{s\}}\wedge T)λs​1​P(τs​(qs​+1)≤τA∖{s}​∧T)

Proved
FastFashion.Structure.delta_hA_integral_formula

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

inventoryp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1poisson-processstopping-time

Let (Ns)s∈S(N_s)_{s \in \mathcal S}(Ns​)s∈S​ be an independent Poisson family with rates λs>0\lambda_s > 0λs​>0 on a probability space, let T>0T > 0T>0, let A⊆S\mathcal A \subseteq \mathcal SA⊆S be a set of sizes, s∈As \in \mathcal As∈A, and q∈NSq \in \mathbb N^{\mathcal S}q∈NS. With hA(q)=E[τA∧T]h^{\mathcal A}(q) = \mathbb E[\tau_{\mathcal A} \wedge T]hA(q)=E[τA​∧T] and ΔshA(q)=hA(q+es)−hA(q)\Delta_s h^{\mathcal A}(q) = h^{\mathcal A}(q + e_s) - h^{\mathcal A}(q)Δs​hA(q)=hA(q+es​)−hA(q),

ΔshA(q)=∫0T[P(τs(qs+1)>t)−P(τs(qs)>t)]∏s′∈A∖{s}P(τs′(qs′)>t) dt=∫0TP(Ns(t)=qs)∏s′∈A∖{s}P(τs′(qs′)>t) dt=1λs P(τs(qs+1)≤τA∖{s}∧T).\begin{aligned} \Delta_s h^{\mathcal A}(q) &= \int_0^T \big[\mathbb P(\tau_s(q_s+1) > t) - \mathbb P(\tau_s(q_s) > t)\big] \prod_{s' \in \mathcal A \setminus \{s\}} \mathbb P(\tau_{s'}(q_{s'}) > t)\,dt \\ &= \int_0^T \mathbb P(N_s(t) = q_s) \prod_{s' \in \mathcal A \setminus \{s\}} \mathbb P(\tau_{s'}(q_{s'}) > t)\,dt \\ &= \frac{1}{\lambda_s}\,\mathbb P\big(\tau_s(q_s+1) \le \tau_{\mathcal A \setminus \{s\}} \wedge T\big). \end{aligned}Δs​hA(q)​=∫0T​[P(τs​(qs​+1)>t)−P(τs​(qs​)>t)]s′∈A∖{s}∏​P(τs′​(qs′​)>t)dt=∫0T​P(Ns​(t)=qs​)s′∈A∖{s}∏​P(τs′​(qs′​)>t)dt=λs​1​P(τs​(qs​+1)≤τA∖{s}​∧T).​

The display expresses the marginal value of one more unit of size sss as the (scaled) probability that this unit is demanded before the display is removed and before the end of the period; its monotonicity in qqq gives the discrete concavity and the cross-monotonicity of hAh^{\mathcal A}hA.

Formalization Note The paper prints the last line as P(τs(qs)≤τA∖{s}∧T)\mathbb P(\tau_s(q_s) \le \tau_{\mathcal A\setminus\{s\}} \wedge T)P(τs​(qs​)≤τA∖{s}​∧T), which is false: the integral has the dimension of a time, and for A={s}\mathcal A = \{s\}A={s}, qs=0q_s = 0qs​=0, λs=T=1\lambda_s = T = 1λs​=T=1 the printed value is 111 while ΔshA(q)=1−e−1\Delta_s h^{\mathcal A}(q) = 1 - e^{-1}Δs​hA(q)=1−e−1. Since τs(qs+1)\tau_s(q_s+1)τs​(qs​+1) has density λsP(Ns(t)=qs)\lambda_s \mathbb P(N_s(t) = q_s)λs​P(Ns​(t)=qs​), the correct value is the one stated here; the first two lines are as printed. The hypothesis s∈As \in \mathcal As∈A is implicit in the display (the factor for sss is taken out of the product; for s∉As \notin \mathcal As∈/A one has ΔshA=0\Delta_s h^{\mathcal A} = 0Δs​hA=0). P(τ>t)\mathbb P(\tau > t)P(τ>t) is written with τ∈R∪{+∞}\tau \in \mathbb R \cup \{+\infty\}τ∈R∪{+∞} and τ∅=+∞\tau_{\emptyset} = +\inftyτ∅​=+∞; the integrals are interval integrals over (0,T](0, T](0,T] of bounded measurable functions of ttt, and P(E)\mathbb P(E)P(E) is the real number (P E).toReal.

Preamble
import Mathlib
import Definitions.Def_FastFashion_Structure_Model
Formal statement
namespace FastFashion.Structure

open MeasureTheory ProbabilityTheory

theorem delta_hA_integral_formula {S Ω : Type*} [Fintype S] [DecidableEq S] [MeasurableSpace Ω]
    (P : Measure Ω) [IsProbabilityMeasure P] (lam : S → ℝ) (N : S → ℝ → Ω → ℕ) (T : ℝ)
    (hN : IsPoissonFamily lam N P) (hlam : ∀ s, 0 < lam s) (hT : 0 < T)
    (A : Finset S) (q : S → ℕ) (s : S) (hs : s ∈ A) :
    delta (hA N P T A) s q =
        ∫ t in (0)..T,
          ((P {ω | ((t : ℝ) : WithTop ℝ) < stockoutTime N s (q + Pi.single s 1) ω}).toReal
            - (P {ω | ((t : ℝ) : WithTop ℝ) < stockoutTime N s q ω}).toReal)
          * ∏ s' ∈ A.erase s, (P {ω | ((t : ℝ) : WithTop ℝ) < stockoutTime N s' q ω}).toReal ∧
    (∫ t in (0)..T,
          ((P {ω | ((t : ℝ) : WithTop ℝ) < stockoutTime N s (q + Pi.single s 1) ω}).toReal
            - (P {ω | ((t : ℝ) : WithTop ℝ) < stockoutTime N s q ω}).toReal)
          * ∏ s' ∈ A.erase s, (P {ω | ((t : ℝ) : WithTop ℝ) < stockoutTime N s' q ω}).toReal) =
        ∫ t in (0)..T,
          (P {ω | N s t ω = q s}).toReal
          * ∏ s' ∈ A.erase s, (P {ω | ((t : ℝ) : WithTop ℝ) < stockoutTime N s' q ω}).toReal ∧
    (∫ t in (0)..T,
          (P {ω | N s t ω = q s}).toReal
          * ∏ s' ∈ A.erase s, (P {ω | ((t : ℝ) : WithTop ℝ) < stockoutTime N s' q ω}).toReal) =
        (1 / lam s) *
          (P {ω | stockoutTime N s (q + Pi.single s 1) ω
                    ≤ min (tauMin N (A.erase s) q ω) ((T : ℝ) : WithTop ℝ)}).toReal := by sorry

end FastFashion.Structure
Source
Caro & Gallien, Inventory Management of a Fast-Fashion Retail Network, working paper (August 2, 2007), p. 31, Appendix §5.1, Proof of Proposition 1, displayed equation for Δ_s h^A(q) (last line corrected)
Human review
  • Endorsed by Shuze Chen · Oct 7, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 7, 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