Improved Analysis of a Max Cut Algorithm Based on Spectral Partitioning
Data Structures and Algorithms
2014-12-03 v2
Abstract
Trevisan [SICOMP 2012] presented an algorithm for Max-Cut based on spectral partitioning techniques. This is the first algorithm for Max-Cut with an approximation guarantee strictly larger than 1/2 that is not based on semidefinite programming. Trevisan showed that its approximation ratio is of at least 0.531. In this paper we improve this bound up to 0.614247. We also define and extend this result for the more general Maximum Colored Cut problem.
Cite
@article{arxiv.0910.0504,
title = {Improved Analysis of a Max Cut Algorithm Based on Spectral Partitioning},
author = {José Soto},
journal= {arXiv preprint arXiv:0910.0504},
year = {2014}
}
Comments
9 pages, 2 figures. Final version