arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 0912.1200cs.DScs.DC
收藏

同步分布环境下Karger随机最小割算法的推广

Extending Karger's randomized min-cut Algorithm for a Synchronous Distributed setting

S. Shine, K. Murali Krishnan

摘要:

在网络中,分离顶点s和t的最小割是一个最小权的边集,去除它将断开s和t。这个问题是著名的s-t最大流问题的对偶。最小割问题的几种算法是基于最大流计算,虽然最快的已知最小割算法不是基于流的。著名的Karger的最小割随机算法是一种基于非流的方法,用于解决(全局)最小割问题,即在加权无向图中的所有顶点对s,t上找到最小s-t割。本文提出了一种适应Karger算法的同步分布式设置,每个节点只允许执行本地计算。本文主要解决的技术细节涉及规避的限制,由分布式设置的工作卡格的算法。虽然正确性证明直接遵循Karger算法,但复杂性分析有很大不同。该算法以O(mn^{2})的消息复杂度和O(n^{2})的时间复杂度实现了与原始算法相同的成功概率,其中n和m表示图中的顶点和边的数量。

英文摘要:

A min-cut that seperates vertices s and t in a network is an edge set of minimum weight whose removal will disconnect s and t. This problem is the dual of the well known s-t max-flow problem. Several algorithms for the min-cut problem are based on max-flow computation although the fastest known min-cut algorithms are not flow based. The well known Karger's randomized algorithm for min-cut is a non-flow based method for solving the (global) min-cut problem of finding the min s-t cut over all pair of vertices s,t in a weighted undirected graph. This paper presents an adaptation of Karger's algorithm for a synchronous distributed setting where each node is allowed to perform only local computations. The paper essentially addresses the technicalities involved in circumventing the limitations imposed by a distributed setting to the working of Karger's algorithm. While the correctness proof follows directly from Karger's algorithm, the complexity analysis differs significantly. The algorithm achieves the same probability of success as the original algorithm with O(mn^{2}) message complexity and O(n^{2}) time complexity, where n and m denote the number of vertices and edges in the graph.

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。

↑