Lemma 2(b) — Finite-menu interval domination
ProvedFBMSP.lemma2b_finite_menu_intervalsFor a finite sorted prediction support with positive probability weights, a nonempty sorted nonnegative price menu, and independent integrable mean-zero MHR error, there exists a monotone maximizing assignment with interval fibers. It weakly improves revenue relative to every other assignment using that menu.
Repeated prices, unused menu indices, zero purchase probabilities, and ties are allowed. This is the finite computational specialization used by Theorem 2.
import Definitions.Def_FBMSP_Model
namespace FBMSP
/-- Tie-safe finite-menu form of Lemma 2(b): there exists a monotone row
maximizer, whose fibers are intervals and whose expected revenue dominates
every assignment using that same menu. Draft statement; proof remains open. -/
theorem lemma2b_finite_menu_intervals {n k : ℕ}
(η : MeasureTheory.Measure ℝ) (hnoise : ValidNoise η) (hmhr : HasMHR η)
(μ w : Fin n → ℝ) (hmarket : ValidFiniteMarket μ w)
(hk : 0 < k) (p : Fin k → ℝ)
(hp : Monotone p) (hp_nonneg : ∀ j, 0 ≤ p j) :
∃ a : Fin n → Fin k,
Monotone a ∧
(∀ i j, postedRevenue η (μ i) (p j) ≤
postedRevenue η (μ i) (p (a i))) ∧
(∀ i₁ i₂ i₃, i₁ ≤ i₂ → i₂ ≤ i₃ →
a i₁ = a i₃ → a i₂ = a i₁) ∧
∀ b : Fin n → Fin k,
finitePolicyRevenue η μ w b p ≤ finitePolicyRevenue η μ w a p := by sorry
end FBMSPRead-back
What the Lean code literally says, in plain math · gpt-6-astra
Let be a probability measure on such that the identity is integrable and . Put and . Assume also that is absolutely continuous with respect to Lebesgue measure and that is convex and is concave on , meaning for and ; zero survival values outside are allowed. For a natural , let real arrays satisfy , whenever , all , all , and . For any natural and nondecreasing real price vector , there exists an assignment that is nondecreasing and satisfies all three conclusions: for every and , ; for indices , equality implies ; and for every assignment into the same slots, . Thus every chosen index is a pointwise maximizer for its mean, the indices assigned to any fixed slot form an interval or the empty set, and the chosen assignment dominates all other assignments for this fixed price vector. Duplicate prices and ties are allowed and the assignment need not be unique; some slots may be unused, including when . Both and are excluded, and no condition is imposed.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.