Negative-Weight Single-Source Shortest Paths in Near-linear Time

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

Documents

  • Fulltext

    Accepted author manuscript, 544 KB, PDF document

We present a randomized algorithm that computes single-source shortest paths (SSSP) in O(mlog(8)(n) logW) time when edge weights are integral and can be negative. This essentially resolves the classic negative-weight SSSP problem. The previous bounds are (O) over tilde ((m + n(1.5)) logW) [BLNPSSSW FOCS'20] and m(4/3+o(1)) logW [AMV FOCS'20]. Near-linear time algorithms were known previously only for the special case of planar directed graphs [Fakcharoenphol and Rao FOCS'01]. In contrast to all recent developments that rely on sophisticated continuous optimization methods and dynamic algorithms, our algorithm is simple: it requires only a simple graph decomposition and elementary combinatorial tools. In fact, ours is the first combinatorial algorithm for negative-weight SSSP to break through the classic (O) over tilde (m root n logW) bound from over three decades ago [Gabow and Tarjan SICOMP'89].

Original languageEnglish
Title of host publication2022 IEEE 63rd Annual Symposium on Foundations of Computer Science (FOCS)
PublisherIEEE
Publication date2022
Pages600-611
ISBN (Electronic)978-1-6654-5519-0
DOIs
Publication statusPublished - 2022
Event63rd Annual IEEE Symposium on Foundations of Computer Science (FOCS) - Denver, Colombia
Duration: 31 Oct 20223 Nov 2022

Conference

Conference63rd Annual IEEE Symposium on Foundations of Computer Science (FOCS)
LandColombia
ByDenver
Periode31/10/202203/11/2022
SeriesAnnual IEEE Symposium on Foundations of Computer Science
ISSN0272-5428

    Research areas

  • graphs and networks, path and circuit problems, graph algorithms, analysis of algorithms, SCALING ALGORITHMS, GRAPHS

ID: 337602711