arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1401.5316cs.DScs.DC
收藏

一种分布式最小割近似格式

A Distributed Minimum Cut Approximation Scheme

Hsin-Hao Su

摘要:

本文研究了分布式消息传递模型CONGEST模型中最小割的近似问题。最小割问题已经在集中式算法的上下文中得到了很好的研究。然而,在Ghaffari和Kuhn最近的工作之前,在分布式模型中没有已知的非平凡算法。他们给出了分别在$O(D)+\tilde{O}(n^{1/2+ε})$轮和$\tilde{O}(D+\sqrt{n})$轮中寻找大小为$O(ε^{-1}λ)$和$(2+ε)λ$的割的算法,其中$λ$是最小割的大小。这与他们提供的下限匹配,直到多对数因子。然而,还没有已知的实现$(1+ε)$近似比的方案。我们给出了一个分布式算法,发现切割的大小$(1+ε)λ$在$\tilde{O}(D+\sqrt{n})$时间,这是最佳的多对数因子。

英文摘要:

In this paper, we study the problem of approximating the minimum cut in a distributed message-passing model, the CONGEST model. The minimum cut problem has been well-studied in the context of centralized algorithms. However, there were no known non-trivial algorithms in the distributed model until the recent work of Ghaffari and Kuhn. They gave algorithms for finding cuts of size $O(ε^{-1}λ)$ and $(2+ε)λ$ in $O(D)+\tilde{O}(n^{1/2+ε})$ rounds and $\tilde{O}(D+\sqrt{n})$ rounds respectively, where $λ$ is the size of the minimum cut. This matches the lower bound they provided up to a polylogarithmic factor. Yet, no scheme that achieves $(1+ε)$-approximation ratio is known. We give a distributed algorithm that finds a cut of size $(1+ε)λ$ in $\tilde{O}(D+\sqrt{n})$ time, which is optimal up to polylogarithmic factors.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑