New approximate distance oracles and their applications

Abstract

Let $G = (V, E)$ be an undirected graph with $n$ vertices and $m$ edges, and let $μ= m/n$. A \emph{distance oracle} is a data structure designed to answer approximate distance queries, with the goal of achieving low stretch, efficient space usage, and fast query time. While much of the prior work focused on distance oracles with constant query time, this paper presents a comprehensive study of distance oracles with non-constant query time. We explore the tradeoffs between space, stretch, and query time of distance oracles in various regimes. Specifically, we consider both weighted and unweighted graphs in the regimes of stretch $< 2$ and stretch $\ge 2$. Among our results, we present a new three-way trade-off between stretch, space, and query time, offering a natural extension of the classical Thorup-Zwick distance oracle [STOC'01 and JACM'05] to regimes with larger query time. Specifically, for any $0 < r < 1/2$ and integer $k \ge 1$, we construct a $(2k(1 - 2r) - 1)$-stretch distance oracle with $\tilde{O}(m + n^{1 + 1/k})$ space and $\tilde{O}(μn^r)$ query time. In addition, we demonstrate several applications of our new distance oracles to the $n$-Pairs Shortest Paths ($n$-PSP) problem and the All Nodes Shortest Cycles ($ANSC$) problem.