Appendix D, p. 42 — optimal DLP allocation on the two-class instance
ProvedResolvingNRM.FRLower.lp_solutionlinear-programmingnetwork-revenue-managementp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1
Let and consider the two-class instance with one unit of resource consumed by either accepted customer. For any nonnegative average resource capacity , every optimal DLP allocation satisfies
The first coordinate is the admission probability for class 1 used in the paper's phase-two argument; the second determines the class-2 admission probability when capacity exceeds one.
Formalization Note The page states the formula explicitly; the formula is the unique optimal companion under and is used implicitly in (64)–(65). Nonnegative is the domain of the policy's LP.
Preamble
import Mathlib import Definitions.Def_ResolvingNRM_FRLower_Instance
Formal statement
namespace ResolvingNRM.FRLower
/-- The unique optimal DLP allocation of the two-class instance. The first equality is
stated on p. 42; the second is the implicit companion used in (64)–(65). -/
theorem lp_solution (r₁ r₂ : ℝ) (hr₂ : 0 < r₂) (hr : r₂ < r₁)
(b : Fin 1 → ℝ) (hb : 0 ≤ b 0) (x : Fin 2 → ℝ)
(hx : IsDLPOptimal (twoClass r₁ r₂ (le_of_lt hr₂) (le_of_lt hr)) b x) :
x 0 = min (b 0) 1 ∧ x 1 = min (max (b 0 - 1) 0) 1 := by sorry
end ResolvingNRM.FRLower
Source
Bumpensanti, Wang, A Re-solving Heuristic with Uniformly Bounded Loss for Network Revenue Management, arXiv:1802.06192v3, Appendix D, p. 42, paragraph before (61) and (64)–(65)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.