arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2004.11666cs.DSmath.CO
收藏

更快的并行多端切割

Faster Parallel Multiterminal Cuts

Monika Henzinger, Alexander Noe, Christian Schulz

摘要:

我们给出了一个改进的分支定界求解器的多端切割问题的基础上,最近的工作Henzinger等人。我们贡献了新的,高效的数据简化规则,将图转换为一个更小的等效实例。此外,我们提出了一个局部搜索算法,可以显着提高一个给定的解决方案的多端切割问题。与Henzinger等人的最新算法相比,我们的精确算法能够对更多更难的问题给出精确解;并为超过三分之二的问题提供更好的解决方案,这些问题太大而无法解决最优性。此外,我们给出了一个不精确的启发式算法,计算高质量的解决方案非常困难的情况下,在合理的时间。

英文摘要:

We give an improved branch-and-bound solver for the multiterminal cut problem, based on the recent work of Henzinger et al.. We contribute new, highly effective data reduction rules to transform the graph into a smaller equivalent instance. In addition, we present a local search algorithm that can significantly improve a given solution to the multiterminal cut problem. Our exact algorithm is able to give exact solutions to more and harder problems compared to the state-of-the-art algorithm by Henzinger et al.; and give better solutions for more than two third of the problems that are too large to be solved to optimality. Additionally, we give an inexact heuristic algorithm that computes high-quality solutions for very hard instances in reasonable time.

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。

↑