Showing 1–1 of 1 results for author: Kaasinen, J

An optimal local approximation algorithm for max-min linear programs

Authors: Patrik Floréen, Joel Kaasinen, Petteri Kaski, Jukka Suomela

Abstract: We present a local algorithm (constant-time distributed algorithm) for approximating max-min LPs. The objective is to maximise $ω$ subject to $Ax \le 1$, $Cx \ge ω1$, and $x \ge 0$ for nonnegative matrices $A$ and $C$. The approximation ratio of our algorithm is the best possible for any local algorithm; there is a matching unconditional lower bound. We present a local algorithm (constant-time distributed algorithm) for approximating max-min LPs. The objective is to maximise $ω$ subject to $Ax \le 1$, $Cx \ge ω1$, and $x \ge 0$ for nonnegative matrices $A$ and $C$. The approximation ratio of our algorithm is the best possible for any local algorithm; there is a matching unconditional lower bound. △ Less

Submitted 9 September, 2008; originally announced September 2008.

Comments: 16 pages, 3 figures