Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 3 — if v≺wv\prec wv≺w, the optimal profit of vvv is at most that of www, for πI\pi_IπI​ and πT\pi_TπT​

Proved
RetailVariety.Fashion.theorem_3

by mikedeng1 · Oct 6, 2026 · Mathlib 0df444a (Lean v4.33.1)

assortment-planningfashionmajorizationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Consider two merchandise categories with the same number nnn of variants, preference vectors v,w∈Rnv,w\in\mathbb R^nv,w∈Rn with nonnegative entries sorted decreasingly (v1≥⋯≥vn≥0v_1\ge\cdots\ge v_n\ge0v1​≥⋯≥vn​≥0, w1≥⋯≥wn≥0w_1\ge\cdots\ge w_n\ge 0w1​≥⋯≥wn​≥0), the same price and cost 0<c<p0<c<p0<c<p, the same demand volume λ>0\lambda>0λ>0, the same model constants σ>0\sigma>0σ>0, 0≤β<10\le\beta<10≤β<1, and equal no-purchase preferences v0=w0>0v_0=w_0>0v0​=w0​>0. Let r,l∈{0,…,n}r,l\in\{0,\dots,n\}r,l∈{0,…,n} be such that the prefix sets ArA_rAr​ and AlA_lAl​ are optimal:

π(Ar,v)=max⁡S⊆Nπ(S,v),π(Al,w)=max⁡S⊆Nπ(S,w).\pi(A_r,v)=\max_{S\subseteq N}\pi(S,v),\qquad \pi(A_l,w)=\max_{S\subseteq N}\pi(S,w).π(Ar​,v)=S⊆Nmax​π(S,v),π(Al​,w)=S⊆Nmax​π(S,w).

If v≺wv\prec wv≺w, that is, vvv is more fashionable than www, then

π(Ar,v)≤π(Al,w).\pi(A_r,v)\le\pi(A_l,w).π(Ar​,v)≤π(Al​,w).

This holds both for the independent population profit π=πI\pi=\pi_Iπ=πI​ of (7) and for the trend-following profit π=πT\pi=\pi_Tπ=πT​ of (8).

The theorem makes precise that fragmentation of preferences is costly: with all else equal, the optimal profit of a more fashionable category is no larger than that of a more basic one.

Formalization Note. The two models are combined in one statement as a conjunction. Optimality of ArA_rAr​ and AlA_lAl​ is stated as hypotheses π(S,⋅)≤π(A⋅,⋅)\pi(S,\cdot)\le\pi(A_\cdot,\cdot)π(S,⋅)≤π(A⋅​,⋅) for every SSS, which is how the theorem defines rrr and lll; their existence is Theorem 1 of the paper and is not used. r,lr,lr,l range over {0,…,n}\{0,\dots,n\}{0,…,n} with A0=∅A_0=\emptysetA0​=∅, since in the independent model the empty assortment can be optimal. The range 0≤β<10\le\beta<10≤β<1 is the model's (§2.3.1); the printed proof writes 0<β<10<\beta<10<β<1.

Preamble
import Mathlib
import Definitions.Def_RetailVariety_Fashion_Model
import Definitions.Def_RetailVariety_Fashion_Majorization
Formal statement
namespace RetailVariety.Fashion

/-- Theorem 3 (van Ryzin & Mahajan 1999, pp. 1505–1506). Two categories `v`, `w` with the same
`n` variants (sorted decreasingly, nonnegative preferences), the same `p, c, λ, σ, β` and equal
no-purchase preference `v_0 = w_0`. If `A_r` maximizes `π(·, v)`, `A_l` maximizes `π(·, w)` and
`v ≺ w`, then `π(A_r, v) ≤ π(A_l, w)`, for `π = π_I` (7) and for `π = π_T` (8). -/
theorem theorem_3 {n : ℕ} (p c lam σ β v0 : ℝ) (hc : 0 < c) (hcp : c < p) (hlam : 0 < lam)
    (hσ : 0 < σ) (hβ0 : 0 ≤ β) (hβ1 : β < 1) (hv0 : 0 < v0)
    (v w : Fin n → ℝ) (hv : ∀ j, 0 ≤ v j) (hw : ∀ j, 0 ≤ w j)
    (hva : Antitone v) (hwa : Antitone w) (r l : ℕ) (hr : r ≤ n) (hl : l ≤ n)
    (hvw : Majorized v w) :
    ((∀ S : Finset (Fin n), profitI p c lam σ β v v0 S ≤ profitI p c lam σ β v v0 (RetailVariety.Statics.A n r)) →
      (∀ S : Finset (Fin n), profitI p c lam σ β w v0 S ≤ profitI p c lam σ β w v0 (RetailVariety.Statics.A n l)) →
      profitI p c lam σ β v v0 (RetailVariety.Statics.A n r) ≤ profitI p c lam σ β w v0 (RetailVariety.Statics.A n l)) ∧
    ((∀ S : Finset (Fin n), profitT p c lam v v0 S ≤ profitT p c lam v v0 (RetailVariety.Statics.A n r)) →
      (∀ S : Finset (Fin n), profitT p c lam w v0 S ≤ profitT p c lam w v0 (RetailVariety.Statics.A n l)) →
      profitT p c lam v v0 (RetailVariety.Statics.A n r) ≤ profitT p c lam w v0 (RetailVariety.Statics.A n l)) := by sorry

end RetailVariety.Fashion
Source
van Ryzin & Mahajan, On the Relationship Between Inventory Costs and Variety Benefits in Retail Assortments, Management Science 45(11), 1999, pp. 1505–1506, Theorem 3
Human review
  • Endorsed by Shuze Chen · Oct 6, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 6, 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