English

Subexponential LPs Approximate Max-Cut

Data Structures and Algorithms 2020-04-20 v2

Abstract

We show that for every ε>0\varepsilon > 0, the degree-nεn^\varepsilon Sherali-Adams linear program (with exp(O~(nε))\exp(\tilde{O}(n^\varepsilon)) variables and constraints) approximates the maximum cut problem within a factor of (12+ε)(\frac{1}{2}+\varepsilon'), for some ε(ε)>0\varepsilon'(\varepsilon) > 0. Our result provides a surprising converse to known lower bounds against all linear programming relaxations of Max-Cut, and hence resolves the extension complexity of approximate Max-Cut for approximation factors close to 12\frac{1}{2} (up to the function ε(ε)\varepsilon'(\varepsilon)). Previously, only semidefinite programs and spectral methods were known to yield approximation factors better than 12\frac 12 for Max-Cut in time 2o(n)2^{o(n)}. We also show that constant-degree Sherali-Adams linear programs (with poly(n)\text{poly}(n) variables and constraints) can solve Max-Cut with approximation factor close to 11 on graphs of small threshold rank: this is the first connection of which we are aware between threshold rank and linear programming-based algorithms. Our results separate the power of Sherali-Adams versus Lov\'asz-Schrijver hierarchies for approximating Max-Cut, since it is known that (12+ε)(\frac{1}{2}+\varepsilon) approximation of Max Cut requires Ωε(n)\Omega_\varepsilon (n) rounds in the Lov\'asz-Schrijver hierarchy. We also provide a subexponential time approximation for Khot's Unique Games problem: we show that for every ε>0\varepsilon > 0 the degree-(nεlogq)(n^\varepsilon \log q) Sherali-Adams linear program distinguishes instances of Unique Games of value 1ε\geq 1-\varepsilon' from instances of value ε\leq \varepsilon', for some ε(ε)>0\varepsilon'( \varepsilon) >0, where qq is the alphabet size. Such guarantees are qualitatively similar to those of previous subexponential-time algorithms for Unique Games but our algorithm does not rely on semidefinite programming or subspace enumeration techniques.

Keywords

Cite

@article{arxiv.1911.10304,
  title  = {Subexponential LPs Approximate Max-Cut},
  author = {Samuel B. Hopkins and Tselil Schramm and Luca Trevisan},
  journal= {arXiv preprint arXiv:1911.10304},
  year   = {2020}
}
R2 v1 2026-06-23T12:25:04.319Z