Lemma 3.3 —
ProvedComplementFreeCA.XOSRounding.expected_max_clause_lower_boundLet , let every bidder have an XOS valuation with admissible clause set , and let be an XOS oracle for ; write for the maximizing clause of in . Let be a feasible solution of the LP relaxation and let be drawn by randomized rounding from . For an item let . Then for every item ,
Summed over the items, the right-hand side is , which yields the approximation ratio of Theorem 3.2.
Formalization Note The paper's display of Lemma 3.3 prints the sum over all pairs . As printed it is false: a maximizing clause assigns values to items outside (two bidders, items , both with the single clause , give ). The last line of the paper's proof restricts to , which is what is stated here. Only feasibility of is assumed, not optimality.
import Mathlib import Definitions.Def_ComplementFreeCA_XOSRounding_Model import Definitions.Def_ComplementFreeCA_XOSRounding_Rounding import Definitions.Def_ComplementFreeCA_XOSRounding_Algorithm
namespace ComplementFreeCA.XOSRounding
open Finset
theorem expected_max_clause_lower_bound {n m : ℕ} (hn : 0 < n)
(W : Fin n → Finset (Fin m → ℝ)) (v : Fin n → Finset (Fin m) → ℝ)
(hv : ∀ i, IsXOSWith (v i) (W i))
(cl : Fin n → Finset (Fin m) → Fin m → ℝ) (hcl : ∀ i, IsXOSOracle (v i) (W i) (cl i))
(x : Fin n → Finset (Fin m) → ℝ) (hx : IsLPFeasible x) (j : Fin m) :
(1 - (1 - 1 / (n : ℝ)) ^ n) *
∑ i, ∑ S ∈ univ.filter (fun S : Finset (Fin m) => j ∈ S), x i S * cl i S j ≤
roundingExpectation x (fun σ => maxClauseEntry hn cl σ j) := by sorry
end ComplementFreeCA.XOSRounding
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.