Shortest Path Separators in Unit Disk Graphs
Elfarouk Harb, Zhengcheng Huang, Da Wei Zheng
TL;DR
This work resolves an open problem by showing that every unit-disk graph admits a shortest path $1$-neighborhood separator: removing the $1$-neighborhood of two shortest paths plus the $1$-neighborhood of two additional vertices yields a balanced cut. The authors develop a novel framework based on non-crossing path systems formed from Delaunay triangulation edges and extend Lipton–Tarjan’s planar separator ideas to triangulated perturbations of these path systems. Key technical components include the definition and non-crossing properties of Delaunay paths, a Path Extension Lemma to preserve non-crossing extensions, and the crominating property of Delaunay edges. The result improves previous $3$-neighborhood separator bounds, is optimal in the sense that a $0$-neighborhood separator can fail for cliques, and provides a foundation for further divide-and-conquer approaches in unit-disk graphs with practical implications for geometric network design. The separator can be constructed in $O(n^2)$ time, with the path system itself of size $O(n^2)$, and the perturbation-based LT separator lies at the core of the approach.
Abstract
We introduce a new balanced separator theorem for unit-disk graphs involving two shortest paths combined with the 1-hop neighbours of those paths and two other vertices. This answers an open problem of Yan, Xiang and Dragan [CGTA '12] and improves their result that requires removing the 3-hop neighborhood of two shortest paths. Our proof uses very different ideas, including Delaunay triangulations and a generalization of the celebrated balanced separator theorem of Lipton and Tarjan [J. Appl. Math. '79] to systems of non-intersecting paths.
