Theorem 1, p. 15 — decreasing-price sequential revenue ≥ pathwise, and
ProvedAssortRotation.Concealment.theorem_1A retailer has products with nonnegative prices . Its consumer has realized valuation shocks , which are mutually independent, continuous, integrable random variables on a probability space , and a nonnegative, decreasing marginal utility . Let be the fixed-assortment revenue (4) of a subset of products. Let over all subsets, and the largest expected sequential revenue over all orders of all products.
- (a) Suppose that is an optimized fixed assortment, , with products. Let be the revenue of the sequential assortment that offers the same products in decreasing order of prices, . Then for every outcome
- (b) The retailer's value of concealment is (weakly) positive:
This is the paper's central result. Concealing products and revealing them one at a time is never worse for the retailer than a fixed assortment. The decreasing-price sequence of the optimal fixed assortment already earns at least as much.
Formalization Note. "With probability 1" in (a) is strengthened to every outcome . Any order among equal prices is allowed in (a). Prices are assumed nonnegative, which is implicit in the paper. Independence, continuity (null singletons for the law of each ), integrability and measurability are the paper's standing assumptions (§2(d) and Notation, p. 10). They are kept even though the inequality does not use them. is computed with the canonical Definition-1 order (ties broken by index), and other orders differ only on a null tie event. "Positive" is meant in the weak sense, the paper's convention (p. 10).
import Mathlib import Definitions.Def_AssortRotation_Concealment_Model
namespace AssortRotation.Concealment
open MeasureTheory ProbabilityTheory
/-- Theorem 1, p. 15. In the model of §2 (`N` products with nonnegative prices `p`,
independent, continuous, integrable valuation shocks `X`, marginal utility `v`):
(a) if `S` is an optimized fixed assortment, then for every outcome `ω` the sequential
revenue `R` of the same products shown in decreasing order of prices (any order among equal
prices) is at least `R^f_*(ω)`;
(b) the value of concealment is (weakly) positive: `E[R^f_*] ≤ E[R^s_*]`. -/
theorem theorem_1 {Ω : Type*} [MeasurableSpace Ω] (μ : Measure Ω) [IsProbabilityMeasure μ]
{N : ℕ} (v : ℕ → ℝ) (hv : MarginalUtility v) (p : Fin N → ℝ) (hp : ∀ i, 0 ≤ p i)
(X : Fin N → Ω → ℝ) (hXm : ∀ i, Measurable (X i)) (hXi : ∀ i, Integrable (X i) μ)
(hind : iIndepFun X μ) (hcont : ∀ i, NullSingletonClass (μ.map (X i))) :
(∀ S : Finset (Fin N), (∫ ω, RfSet v p X S ω ∂μ = ERfStar μ v p X) →
∀ ω : Ω, ∀ L : List (ℝ × ℝ), L.Perm (products p X S ω) →
L.Pairwise (fun a b => b.1 ≤ a.1) → RfSet v p X S ω ≤ seqRev v L 0) ∧
ERfStar μ v p X ≤ ERsStar μ v p X := by sorry
end AssortRotation.Concealment
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.