Better Bounds for Semi-Streaming Single-Source Shortest Paths

TL;DR

This paper introduces a new semi-streaming algorithm for approximating single-source shortest paths with fewer passes and space than previous methods, and establishes lower bounds showing the difficulty of constant-factor approximations.

Contribution

It presents a simple randomized semi-streaming algorithm for (1+ε)-approximate shortest paths with improved pass and space complexity, and proves lower bounds for constant approximation algorithms.

Findings

New algorithm achieves (1+ε)-approximation in O(1/ε * n log^3 n) space and O((log n / log log n)^2) passes.

Lower bounds show any constant approximation requires at least (log n / log log n) passes.

Results narrow the gap between upper and lower bounds for semi-streaming shortest path approximations.

Abstract

In the semi-streaming model, an algorithm must process any $n$-vertex graph by making one or few passes over a stream of its edges, use $O(n \cdot \text{polylog }n)$ words of space, and at the end of the last pass, output a solution to the problem at hand. Approximating (single-source) shortest paths on undirected graphs is a longstanding open question in this model. In this work, we make progress on this question from both upper and lower bound fronts: We present a simple randomized algorithm that for any $\epsilon > 0$, with high probability computes $(1+\epsilon)$-approximate shortest paths from a given source vertex in \[ O\left(\frac{1}{\epsilon} \cdot n \log^3 n \right)~\text{space} \quad \text{and} \quad O\left(\frac{1}{\epsilon} \cdot \left(\frac{\log n}{\log\log n} \right) ^2\right) ~\text{passes}. \] The algorithm can also be derandomized and made to work on dynamic streams at a cost of some extra $\text{poly}(\log n, 1/\epsilon)$ factors only in the space. Previously, the best known algorithms for this problem required $1/\epsilon \cdot \log^{c}(n)$ passes, for an unspecified large constant $c$. We prove that any semi-streaming algorithm that with large constant probability outputs any constant approximation to shortest paths from a given source vertex (even to a single fixed target vertex and only the distance, not necessarily the path) requires \[ \Omega\left(\frac{\log n}{\log\log n}\right) ~\text{passes}. \] We emphasize that our lower bound holds for any constant-factor approximation of shortest paths. Previously, only constant-pass lower bounds were known and only for small approximation ratios below two. Our results collectively reduce the gap in the pass complexity of approximating single-source shortest paths in the semi-streaming model from $\text{polylog } n$ vs $\omega(1)$ to only a quadratic gap.