arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1607.08754math.COcs.DS
收藏

枚举公平着色算法的一种基于流的剪枝方案

A flow based pruning scheme for enumerative equitable coloring algorithms

Arie Koster, Robert Schweidweiler, Martin Tieves

摘要:

图的均匀染色是指图G的正常点染色,其中色类的大小相差不超过1。均匀色数是使G允许均匀k-染色的最小数k。我们专注于枚举算法的公平着色数的计算,并提出了一个通用的计划,以获得修剪规则:我们展示了如何通过网络流的部分着色到公平着色的可扩展性建模。因此,我们得到的修剪规则,可以通过流算法进行检查。计算实验表明,这些规则可以显着减少枚举算法的搜索树的大小,在大多数情况下,这种天真的方法,甚至产生一个更快的算法。此外,稳定性,即,在给定的时间限制内解决的实例的数量大大提高。由于在搜索树的每个节点上执行流算法是耗时的,我们从网络模型中推导出算术剪枝规则(广义霍尔条件)。将这些规则添加到枚举算法中会产生更大的运行时改进。

英文摘要:

An equitable graph coloring is a proper vertex coloring of a graph G where the sizes of the color classes differ by at most one. The equitable chromatic number is the smallest number k such that G admits such equitable k-coloring. We focus on enumerative algorithms for the computation of the equitable coloring number and propose a general scheme to derive pruning rules for them: We show how the extendability of a partial coloring into an equitable coloring can be modeled via network flows. Thus, we obtain pruning rules which can be checked via flow algorithms. Computational experiments show that the search tree of enumerative algorithms can be significantly reduced in size by these rules and, in most instances, such naive approach even yields a faster algorithm. Moreover, the stability, i.e., the number of solved instances within a given time limit, is greatly improved. Since the execution of flow algorithms at each node of a search tree is time consuming, we derive arithmetic pruning rules (generalized Hall-conditions) from the network model. Adding these rules to an enumerative algorithm yields an even larger runtime improvement.

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。

↑