Theorem 1, PDF p. 11 — under simultaneous recommendation the profile (S̃, …, S̃) is optimal
ProvedMultiStoreAssort.SMR.theorem1_smr_optimalA seller operates stores and sells products with revenues and MNL attraction values . A fraction of customers visits store , with . A customer who visits store and buys product from store suffers a utility discount , with . Under the simultaneous recommendation (SMR) model, the customer of store facing the profile chooses product with probability given by (1). The seller's problem is
Let be the smallest revenue-ordered optimal solution of the single-store problem (3). Then is an optimal solution of (SMR): for every profile ,
Under simultaneous recommendation, cross-store coordination never beats running every store as a stand-alone MNL store. The resulting revenue is the paper's benchmark, which the sequential strategy is compared against.
Formalization Note. "The optimal solution is " is formalized as " is a global maximizer". Optimal solutions need not be unique. enters through the predicate IsStilde, and the milestone stilde_exists shows such an exists. The positivity is the series' standing assumption (revenues are prices). Zero discounts between different stores are allowed, as in the paper.
import Mathlib import Definitions.Def_MultiStoreAssort_SMR_Model
namespace MultiStoreAssort.SMR
/-- Theorem 1, p. 11: the optimal solution to the SMR problem is `S* = (S̃, S̃, …, S̃)`. -/
theorem theorem1_smr_optimal {m n : ℕ} (lam : Fin m → ℝ) (r v : Fin n → ℝ)
(c : Fin m → Fin m → Fin n → ℝ)
(hlam : ∀ i, 0 ≤ lam i) (hlam1 : ∑ i, lam i = 1) (hr : StrictAnti r) (hrpos : ∀ k, 0 < r k)
(hv : ∀ k, 0 < v k) (hc : ∀ i j k, 0 ≤ c i j k) (hcii : ∀ i k, c i i k = 0)
(St : Finset (Fin n)) (hSt : IsStilde r v St) :
∀ S : Fin m → Finset (Fin n), revM lam r v c S ≤ revM lam r v c (fun _ => St) := by sorry
end MultiStoreAssort.SMR
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.