Section 4 — an adversary forces sets on the block family while
ProvedOnlineSetCover.LowerBound.block_adversaryLet be positive integers and let be the block family of Section 4 on disjoint blocks of elements. For every valid deterministic online algorithm for there is a nonempty arrival sequence of at most elements such that
- a single set of contains every element of , so , and
- the algorithm has chosen at least sets:
This is the paper's claim that "given any deterministic algorithm, an adversary can choose elements in , forcing the algorithm to pick sets from , while keeping the value of the optimum solution to be 1". It is the core of Proposition 4.2, which adds padding elements and sets to reach arbitrary and .
Formalization Note The paper's argument tacitly lets the algorithm add one set per arrival; here algorithms may add any number of sets per arrival, and the sequence then may be shorter than , hence "at most elements". The ground set is the blocks only; elements outside the blocks lie in no set and play no role.
import Mathlib import Definitions.Def_OnlineSetCover_LowerBound_Game import Definitions.Def_OnlineSetCover_LowerBound_BlockFamily
namespace OnlineSetCover.LowerBound
/-- The adversary claim of §4 (Alon et al. 2009, p. 369): for positive integers `k, r` and every
valid deterministic online algorithm `A` for the block family on the `k r²` blocks of `2^k`
elements, there is a nonempty arrival sequence of at most `k r` elements that a single set of
the family covers (so `OPT = 1`), on which `A` chooses at least `k r` sets. -/
theorem block_adversary (k r : ℕ) (hk : 0 < k) (hr : 0 < r)
(A : OnlineAlg (Fin (k * r ^ 2) × Fin (2 ^ k))) (hA : IsValid (blockFamily k r) A) :
∃ σ : List (Fin (k * r ^ 2) × Fin (2 ^ k)), σ ≠ [] ∧ σ.length ≤ k * r ∧
(∃ S ∈ blockFamily k r, ∀ x ∈ σ, x ∈ S) ∧ k * r ≤ cost A σ := by sorry
end OnlineSetCover.LowerBound
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.