English

Voting algorithms for unique games on complete graphs

Data Structures and Algorithms 2022-11-09 v2

Abstract

An approximation algorithm for a constraint satisfaction problem is called robust if it outputs an assignment satisfying a (1f(ϵ))(1 - f(\epsilon))-fraction of the constraints on any (1ϵ)(1-\epsilon)-satisfiable instance, where the loss function ff is such that f(ϵ)0f(\epsilon) \rightarrow 0 as ϵ0\epsilon \rightarrow 0. Moreover, the runtime of a robust algorithm should not depend in any way on ϵ\epsilon. In this paper, we present such an algorithm for Min-Unique-Games on complete graphs with qq labels. Specifically, the loss function is f(ϵ)=(ϵ+cϵϵ2)f(\epsilon) = (\epsilon + c_{\epsilon} \epsilon^2), where cϵc_{\epsilon} is a constant depending on ϵ\epsilon such that limϵ0cϵ=16\lim_{\epsilon \rightarrow 0} c_{\epsilon} = 16. The runtime of our algorithm is O(qn3)O(qn^3) (with no dependence on ϵ\epsilon) and can run in time O(qn2)O(qn^2) using a randomized implementation with a slightly larger constant cϵc_{\epsilon}. Our algorithm is combinatorial and uses voting to find an assignment. It can furthermore be used to provide a PTAS for Min-Unique-Games on complete graphs, recovering a result of Karpinski and Schudy with a simpler algorithm and proof. We also prove NP-hardness for Min-Unique-Games on complete graphs and (using a randomized reduction) even in the case where the constraints form a cyclic permutation, which is also known as Min-Linear-Equations-mod-qq on complete graphs.

Keywords

Cite

@article{arxiv.2110.11851,
  title  = {Voting algorithms for unique games on complete graphs},
  author = {Antoine Méot and Arnaud de Mesmay and Moritz Mühlenthaler and Alantha Newman},
  journal= {arXiv preprint arXiv:2110.11851},
  year   = {2022}
}
R2 v1 2026-06-24T07:06:33.956Z