Lemma 1: and are quasi-convex in on
ProvedRetailVariety.Structure.lemma_1assortment-planninginventorymultinomial-logitnewsvendorp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let be positive preferences (), the no-purchase preference, , , , , , and let be any assortment. With , , as in (9)–(11), the functions
are both quasi-convex on : every sublevel set is an interval.
and are the store profits after a variant of preference is added to . Quasi-convexity means the maximum over an interval of is attained at an end point, which drives the exchange argument of Theorem 1.
Formalization Note The paper's is the first variant, v ⟨0, hn⟩; the paper's is v0. carries on both terms; as printed in (11) multiplies only the sum, which does not change quasi-convexity.
Preamble
import Mathlib import Definitions.Def_RetailVariety_Structure_Model
Formal statement
namespace RetailVariety.Structure
/-- Lemma 1, p. 1503: for any `S`, `h_I = g_I / f` and `h_T = g_T / f` are quasi-convex in
`δ` on `0 ≤ δ ≤ v_1` (`v_1 = v ⟨0, hn⟩`, the most popular variant; `v0` is the no-purchase
preference `v_0`). -/
theorem lemma_1 (n : ℕ) (hn : 0 < n) (v : Fin n → ℝ) (v0 : ℝ) (hv : ∀ j, 0 < v j)
(hv0 : 0 < v0) (hanti : Antitone v) (p c lam σ β : ℝ) (hc : 0 < c) (hcp : c < p)
(hlam : 0 < lam) (hσ : 0 < σ) (hβ0 : 0 ≤ β) (hβ1 : β < 1)
(S : Finset (Fin n)) :
QuasiconvexOn ℝ (Set.Icc 0 (v ⟨0, hn⟩)) (hI p c lam σ β v v0 S) ∧
QuasiconvexOn ℝ (Set.Icc 0 (v ⟨0, hn⟩)) (hT p c lam v v0 S) := by sorry
end RetailVariety.Structure
Source
van Ryzin & Mahajan, On the Relationship Between Inventory Costs and Variety Benefits in Retail Assortments, Management Science 45(11), 1999, p. 1503, Lemma 1, eqs. (9)–(11)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.