Proposition 3.2 — under MHR, is continuous and convex on the interior of
ProvedBilateralProfitMHR.proposition_3_2bilateral-trademonotone-hazard-rateonline-learningp2o-gran-per-chapterp2o-plan-paperp2o-v1posted-pricesregret-bounds
Let the seller and buyer values be independent with densities strictly positive on , and let both satisfy the monotone hazard rate property (Definition 2.3). Let be the expected profit, where . Then the loss
is continuous and convex on the interior of the triangle .
Convexity of is what turns profit maximization into a stochastic convex optimization problem, to which Corollary 3.1 applies.
Differences from the source:
- encoding: "under the MHR assumption" (with §2's standing independent laws with strictly positive densities on ) → an environment with supports satisfying MHR for both agents; reason: this is §2's class exactly; the reused environment adds no condition
- encoding: "the interior of " → the set ; reason: this set is the interior of in , and on it, so the logarithm is the true one
Preamble
import Mathlib import Definitions.Def_BilateralProfitMHR_Model
Formal statement
namespace BilateralProfitMHR
open MeasureTheory BilateralProfit MechanismDesign.BilateralTrade
/-- **Proposition 3.2**, Brüstle, di Gregorio, Fusco, Leonardi (manuscript, sha256 e91e8127…),
p.11 (PDF 13), restated with proof p.30 (PDF 32): under the MHR assumption, the loss function
`L(p, q) = −log µ(p, q)` is continuous and convex in the interior of `𝒯`.
Stated for every environment with independent seller and buyer values whose densities are
strictly positive on `[0, 1]` and both MHR (Definition 2.3), with `µ` the expected profit under
the product law `E.prior`.
**Formalization Note.** The interior of `𝒯 = {(p, q) ∈ [0,1]² : p ≤ q}` in `ℝ²` is
`intT = {0 < p < q < 1}`. On it `µ > 0` (positive densities), so `Real.log` is the true
logarithm there; no smoothness of the densities is assumed. -/
theorem proposition_3_2 (E : Environment) (hE : UnitSupport E) (hM : IsMHR E) :
ContinuousOn (lossL E.prior) intT ∧ ConvexOn ℝ intT (lossL E.prior) := by sorry
end BilateralProfitMHR
Source
Brüstle, di Gregorio, Fusco, Leonardi, Regret Bounds for Profit Maximization in Bilateral Trade: Learning the Best Posted-Price Mechanism (manuscript, STOC 2027 submission, sha256 e91e81279505b765c7c96c219d674c0fbbbd6ca47a00da8fe228d498ca25ac87), Proposition 3.2, p.11 (PDF 13); restated with proof p.30 (PDF 32)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.