Prove2Me
Navigate
DiscoverFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Proposition 4.2 — for n≥2k+1kr2n \ge 2^{k+1}kr^2n≥2k+1kr2 and 22kkr2≥m≥(kr2r)kr2^{2^k kr^2} \ge m \ge \binom{kr^2}{r}k^r22kkr2≥m≥(rkr2​)kr, every deterministic algorithm has competitive ratio ≥kr\ge kr≥kr

Proved
OnlineSetCover.LowerBound.proposition_4_2

by mikedeng1 · Oct 4, 2026 · Mathlib 0df444a (Lean v4.33.1)

competitive-analysislower-boundonline-algorithmsp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1set-cover

For all positive integers k,rk, rk,r and all n,mn, mn,m satisfying

n≥2k+1kr2and22kkr2  ≥  m  ≥  (kr2r)kr,n \ge 2^{k+1} k r^2 \qquad\text{and}\qquad 2^{2^k k r^2} \;\ge\; m \;\ge\; \binom{kr^2}{r} k^r,n≥2k+1kr2and22kkr2≥m≥(rkr2​)kr,

there is an instance of the unweighted online set cover problem with ground set X={1,…,n}X = \{1, \dots, n\}X={1,…,n} and a family F\mathcal FF of exactly mmm distinct subsets of XXX such that the competitive ratio of every deterministic online algorithm is at least krkrkr. More precisely, for every valid deterministic online algorithm AAA for F\mathcal FF there is a nonempty arrival sequence σ\sigmaσ such that a single member of F\mathcal FF contains every element of σ\sigmaσ (so OPT(σ)=1\mathrm{OPT}(\sigma) = 1OPT(σ)=1) and

∣CA(σ)∣  ≥  kr  =  kr⋅OPT(σ).|\mathcal C_A(\sigma)| \;\ge\; kr \;=\; kr\cdot\mathrm{OPT}(\sigma).∣CA​(σ)∣≥kr=kr⋅OPT(σ).

Choosing rrr and kkk as functions of nnn and mmm turns this into the paper's lower bound Ω(log⁡nlog⁡m/(log⁡log⁡m+log⁡log⁡n))\Omega\big(\log n \log m / (\log\log m + \log\log n)\big)Ω(lognlogm/(loglogm+loglogn)), nearly matching the O(log⁡mlog⁡n)O(\log m \log n)O(logmlogn) upper bound of the paper's deterministic algorithm.

Formalization Note XXX is Fin n and F\mathcal FF a Finset (Finset (Fin n)), so its cardinality counts distinct sets. The conclusion states OPT(σ)=1\mathrm{OPT}(\sigma) = 1OPT(σ)=1 and cost ≥kr\ge kr≥kr, which is how the paper establishes "competitive ratio at least krkrkr"; since OPT(σ)=1\mathrm{OPT}(\sigma) = 1OPT(σ)=1 it gives ∣CA(σ)∣≥kr⋅OPT(σ)|\mathcal C_A(\sigma)| \ge kr \cdot \mathrm{OPT}(\sigma)∣CA​(σ)∣≥kr⋅OPT(σ) with OPT(σ)≥1\mathrm{OPT}(\sigma) \ge 1OPT(σ)≥1. The family is asserted to exist, as in the paper; the construction behind it is the block family plus padding sets on the extra elements. Algorithms may add several sets per arrival.

Preamble
import Mathlib
import Definitions.Def_OnlineSetCover_LowerBound_Game
Formal statement
namespace OnlineSetCover.LowerBound

/-- Proposition 4.2 (Alon et al. 2009, p. 369). For positive integers `k, r` and `n, m` with
`n ≥ 2^{k+1} k r²` and `2^{2^k k r²} ≥ m ≥ C(k r², r) k^r`, there is a family `𝓕` of exactly `m`
distinct subsets of `X = Fin n` such that against every valid deterministic online algorithm
`A` the adversary has a nonempty arrival sequence `σ` that a single member of `𝓕` covers
(`OPT(σ) = 1`) while `A` chooses at least `k r` sets; in particular the competitive ratio of
every deterministic online algorithm on `(X, 𝓕)` is at least `k r`. -/
theorem proposition_4_2 (k r n m : ℕ) (hk : 0 < k) (hr : 0 < r)
    (hn : 2 ^ (k + 1) * k * r ^ 2 ≤ n)
    (hm_lo : (k * r ^ 2).choose r * k ^ r ≤ m)
    (hm_hi : m ≤ 2 ^ (2 ^ k * k * r ^ 2)) :
    ∃ 𝓕 : Finset (Finset (Fin n)), 𝓕.card = m ∧
      ∀ A : OnlineAlg (Fin n), IsValid 𝓕 A →
        ∃ σ : List (Fin n), σ ≠ [] ∧ (∃ S ∈ 𝓕, ∀ x ∈ σ, x ∈ S) ∧ k * r ≤ cost A σ := by sorry

end OnlineSetCover.LowerBound
Source
Alon, Awerbuch, Azar, Buchbinder, Naor, The Online Set Cover Problem, SIAM J. Comput. 39(2) (2009), p. 369, Proposition 4.2
Human review
  • Endorsed by Shuze Chen · Oct 5, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 5, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paper
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me