Improved Distributed Expander Decomposition and Nearly Optimal Triangle Enumeration

Improved Distributed Expander Decomposition and Nearly Optimal Triangle Enumeration
复制标题

改进的分布式扩展器分解和近乎最优的三角形枚举

DOI:
10.1145/3293611.3331618
复制
发表时间:
2019
期刊:
Proceedings 38th Symposium on Principles of Distributed Computing
影响因子:
--
通讯作者:
Saranurak, Thatchaphol
Saranurak, Thatchaphol
中科院分区:
--
文献类型:
--
作者:
Chang, Yi-Jun;Saranurak, Thatchaphol

文献摘要

参考文献

被引文献

相似文献

图G=(V,E)的一个(ε,φ)-扩展分解是顶点V=V1的一个聚类,使得(1)每个聚类Vi诱导的子图的电导至少为φ,(2)聚类间的边数至多为ε| E|.本文给出了一种改进的分布式扩展分解,并在CONGEST模型下得到了一个近似最优的分布式三角形枚举算法,具体地说,对于任意的ε ∈(0,1)和正整数k,我们构造了一个(ε,φ)-扩展分解,其中φ=(ε/logn)2 O(k),循环次数为O(n2/k poly(1/φ,logn)).例如,(1/no(1),1/no(1))-扩展器分解只需要O(no(1))轮来计算,这对于子多项式因子是最优的,并且对于任何任意小的常数γ > 0,(0.01,1/poly log n)-扩展器分解可以在O(nγ)轮中计算。以前,Chang,Pettie和Zhang的算法可以对任何δ > 0使用n(n1-δ)轮来构造(1/6,1/poly log n)-扩展器分解,但需要注意的是,该算法允许将一组边丢弃到形成荫度至多为nδ的子图的额外部分中。通过对Ghaffari,Kuhn和Su [PODC'17]提出的扩展器上的分布式路由算法稍加修改,我们得到了一个使用n(n1/3)轮的三角枚举算法。这与Izumi和LeGall [PODC'17]以及Pandurangan,罗宾逊和Scquizzato [SPAA'18]的下限(n1/3)相匹配,即使在CONGESTED-CLIQUE模型中也是如此。据我们所知,这提供了第一个非平凡的例子,一个分布式的问题,具有基本上相同的复杂性(多对数因子)在CONGEST和CONGESTED-CLIQUE。在我们的证明中的关键技术是第一个分布式近似算法,找到一个低电导切割,是尽可能平衡。以前的分布式稀疏切割算法没有这种几乎最平衡的保证。
An(ε,φ)-expander decomposition of a graph G=(V,E) is a clustering of the vertices V=V1∪…∪ Vxsuch that (1) each cluster Viinduces subgraph with conductance at least φ, and (2) the number of inter-cluster edges is at most ε|E|. In this paper, we give an improved distributed expander decomposition, and obtain a nearly optimal distributed triangle enumeration algorithm in the CONGEST model.Specifically, we construct an (ε,φ)-expander decomposition with φ=(ε/log n)2 O(k)in O(n2/k⋅ poly (1/φ, log n))rounds for any ε ∈(0,1) and positive integer k. For example, a (1/no(1), 1/no(1))-expander decomposition only requires O(no(1)) rounds to compute, which is optimal up to subpolynomial factors, and a (0.01,1/poly log n)-expander decomposition can be computed in O(nγ) rounds, for any arbitrarily small constant γ > 0. Previously, the algorithm by Chang, Pettie, and Zhang can construct a (1/6,1/poly log n)-expander decomposition using Õ (n1-δ) rounds for any δ > 0, with a caveat that the algorithm is allowed to throw away a set of edges into an extra part which form a subgraph with arboricity at most nδ. Our algorithm does not have this caveat.By slightly modifying the distributed algorithm for routing on expanders by Ghaffari, Kuhn and Su [PODC'17], we obtain a triangle enumeration algorithm using Õ(n1/3) rounds. This matches the lower bound by Izumi and LeGall [PODC'17] and Pandurangan, Robinson and Scquizzato [SPAA'18] of Ø(n1/3) which holds even in the CONGESTED-CLIQUE model. To the best of our knowledge, this provides the first non-trivial example for a distributed problem that has essentially the same complexity (up to a polylogarithmic factor) in both CONGEST and CONGESTED-CLIQUE.The key technique in our proof is the first distributed approximation algorithm for finding a low conductance cut that is as balanced as possible. Previous distributed sparse cut algorithms do not have this nearly most balanced guarantee.
拥塞团模型中的稀疏矩阵乘法和三角形列表
DOI: --
发表时间: 2018
期刊: International Conference on Principles of Distributed Systems
影响因子: --
作者:
K. Censor;Dean Leitersdorf;Elia Turner
通讯作者: Elia Turner
具有次多项式最坏情况更新时间的动态最小生成森林
DOI: 10.1109/focs.2017.92
发表时间: 2017
期刊: 2017 IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS)
影响因子: --
作者:
Danupon Nanongkai;Thatchaphol Saranurak;Christian Wulff
通讯作者: Christian Wulff
DOI: 10.1145/3210377.3210409
发表时间: 2016-02
期刊: Proceedings of the 30th on Symposium on Parallelism in Algorithms and Architectures
影响因子: --
作者:
Gopal Pandurangan;Peter Robinson;Michele Scquizzato
通讯作者: Gopal Pandurangan;Peter Robinson;Michele Scquizzato
分布式子图检测的可能性和不可能性
DOI: --
发表时间: 2018
期刊: ACM Symposium on Parallelism in Algorithms and Architectures
影响因子: --
作者:
O. Fischer;T. Gonen;F. Kuhn;R. Oshman
通讯作者: R. Oshman
DOI: 10.1145/3087801.3087827
发表时间: 2017-07
期刊: Proceedings of the ACM Symposium on Principles of Distributed Computing
影响因子: --
作者:
M. Ghaffari;F. Kuhn;Hsin-Hao Su
通讯作者: M. Ghaffari;F. Kuhn;Hsin-Hao Su