Greedy obtains a approximation for a monotone submodular objective
ProvedCustAssort.AugGreedy.greedy_approximationapproximation-algorithmgreedy-algorithmp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1submodular
Let be a monotone, submodular set function on a finite ground set , with . Greedy begins with the empty set, repeatedly adds an available element giving the largest next value, and stops when it reaches cardinality or exhausts . For every output , including every tie-breaking choice, and every feasible comparison set with ,
This is the classical bound cited for the truncated objective in the proof of Theorem 4.2.
Formalization Note The nonnegative empty-set value is needed for the multiplicative form. It holds with equality for the paper's truncated objective.
Preamble
import Mathlib import Definitions.Def_CustAssort_AugGreedy_Setting
Formal statement
namespace CustAssort.AugGreedy
/-- The Nemhauser et al. (1978) cardinality-constrained greedy bound used on pp. 10 and 12. -/
theorem greedy_approximation {α : Type*} [DecidableEq α] (g : Finset α → ℝ)
(P : Finset α) (k : ℕ) (Δ : Finset α)
(hmono : ∀ A B : Finset α, A ⊆ B → B ⊆ P → g A ≤ g B)
(hsub : SubmodularOn g P) (hzero : 0 ≤ g ∅)
(hΔ : IsGreedyOutput g P k Δ) :
∀ S : Finset α, S ⊆ P → S.card ≤ k →
(1 - Real.exp (-1)) * g S ≤ g Δ := by sorry
end CustAssort.AugGreedy
Source
El Housni & Topaloglu, Joint Assortment Optimization and Customization under a Mixture of Multinomial Logit Models: Value of Personalized Assortments, SSRN 3830082, https://ssrn.com/abstract=3830082 (version of December 7, 2021), §4.1, p. 10; proof of Theorem 4.2, p. 12, citing Nemhauser et al. (1978)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.