“…They showed in particular that the integrality gap for Vertex Cover remains at least 2 − ε after Ω ε (log n) rounds of LS. Since then there has been a flurry of activity, proving larger gaps after fewer rounds for Vertex Cover and several other classical NP-hard optimization problems; see, e.g., [2,8,9,10,13,14,24,25,28]. Most of this work has focused on the LS and LS + hierarchies; exceptions are [10,13], which consider Sherali-Adams, and [24] which considers Lasserre.…”