Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 4(a) — an optimal n-period (s, S) policy with s‾≤sn≤sˉ≤S‾≤Sn≤Sˉ\underline{s} \le s_n \le \bar{s} \le \underline{S} \le S_n \le \bar{S}s​≤sn​≤sˉ≤S​≤Sn​≤Sˉ

Proved
VeinottWagnerSS.Bounds.theorem4a_optimal_sS_within_bounds

by mikedeng1 · Sep 29, 2026 · Mathlib 0df444a (Lean v4.33.1)

dynamic-programminginventoryp2o-batch-p200bp2o-gran-per-chapterp2o-plan-paperp2o-v1ss-policy

Consider the nnn-period periodic-review inventory model with backlogging: i.i.d. demands with distribution φ\varphiφ on {0,1,… }\{0, 1, \dots\}{0,1,…} of finite mean, set-up cost K≥0K \ge 0K≥0, discount factor 0≤α≤10 \le \alpha \le 10≤α≤1, and one-period cost Gα:Z→RG_\alpha : \mathbb Z \to \mathbb RGα​:Z→R that is convex on the integers with Gα(y)→∞G_\alpha(y) \to \inftyGα​(y)→∞ as ∣y∣→∞|y| \to \infty∣y∣→∞. The cost of a policy YYY from initial level xxx is

fn(x∣Y)=∑t=1nαt−1[K E δ(Yt−Xt)+E Gα(Yt)].f_n(x \mid Y) = \sum_{t=1}^{n} \alpha^{t-1}\bigl[K\,E\,\delta(Y_t - X_t) + E\,G_\alpha(Y_t)\bigr].fn​(x∣Y)=t=1∑n​αt−1[KEδ(Yt​−Xt​)+EGα​(Yt​)].

Let s‾,sˉ,S‾,Sˉ\underline s, \bar s, \underline S, \bar Ss​,sˉ,S​,Sˉ be the numbers defined by (21)–(23).

Then for every n≥2n \ge 2n≥2 there is a policy YnY^nYn such that

  1. YnY^nYn is optimal: fn(x∣Yn)≤fn(x∣Y)f_n(x \mid Y^n) \le f_n(x \mid Y)fn​(x∣Yn)≤fn​(x∣Y) for every admissible (history-dependent) policy YYY and every initial level xxx;
  2. YnY^nYn is an (s,S)(s, S)(s,S) policy, i.e. it uses an (st,St)(s_t, S_t)(st​,St​) rule with st≤Sts_t \le S_tst​≤St​ in each of the nnn periods;
  3. its first-period rule (sn,Sn)(s_n, S_n)(sn​,Sn​) satisfies
s‾≤sn≤sˉ≤S‾≤Sn≤Sˉ.\underline{s} \le s_n \le \bar{s} \le \underline{S} \le S_n \le \bar{S}.s​≤sn​≤sˉ≤S​≤Sn​≤Sˉ.

The theorem localizes an optimal first-period policy of every finite-horizon model in a box computed from the one-period cost alone. The paper uses these bounds to restrict the search for optimal stationary policies.

Formalization Note The existence of an optimal (s,S)(s, S)(s,S) policy, which the paper takes from Scarf and from Zabel (p. 529), is part of the conclusion, not a hypothesis. The model is the reduced model of Eq. (2): GαG_\alphaGα​ is a primitive, and fnf_nfn​ is valued in the extended reals.

Preamble
import Mathlib
import Definitions.Def_VeinottWagnerSS_Bounds_Model
import Definitions.Def_VeinottWagnerSS_Bounds_CriticalNumbers

open scoped ENNReal
Formal statement
namespace VeinottWagnerSS.Bounds

/-- Theorem 4(a), p. 549. For the `n`-period model (`n = 2, 3, ⋯`) there exists an optimal
`(s, S)` policy whose first-period rule `(s_n, S_n)` satisfies
`s̲ ≤ s_n ≤ s̄ ≤ S̲ ≤ S_n ≤ S̄`. Optimality is against every admissible (history-dependent)
policy and for every initial level `x`. -/
theorem theorem4a_optimal_sS_within_bounds (G : ℤ → ℝ) (K α : ℝ) (hα₀ : 0 ≤ α) (hα₁ : α ≤ 1) (hK : 0 ≤ K)
    (hconv : ∀ y : ℤ, G (y + 1) - G y ≤ G (y + 2) - G (y + 1))
    (hG_top : Filter.Tendsto G Filter.atTop Filter.atTop)
    (hG_bot : Filter.Tendsto G Filter.atBot Filter.atTop)
    (φ : PMF ℕ) (hμ : ∑' k : ℕ, (k : ℝ≥0∞) * φ k ≠ ⊤)
    (n : ℕ) (hn : 2 ≤ n) :
    ∃ (Y : Policy) (sn Sn : ℤ), IsOptimal φ G K α n Y ∧ IsSSPolicy n Y ∧
      sn ≤ Sn ∧ UsesRule Y 0 sn Sn ∧
      sLow G K ≤ sn ∧ sn ≤ sHigh G K α ∧ sHigh G K α ≤ SLow G ∧
      SLow G ≤ Sn ∧ Sn ≤ SHigh G K α := by sorry

end VeinottWagnerSS.Bounds
Source
Veinott & Wagner, Computing Optimal (s, S) Inventory Policies, Management Sci. 11 (1965), p. 549, Theorem 4(a)
Human review
  • Endorsed by Shuze Chen · Oct 1, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 1, 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