{"id":"arxiv:2506.00428","kind":"preprint","title":"Faster negative length shortest paths by bootstrapping hop reducers","authors":["Yufan Huang","Peter Jin","Kent Quanrud"],"author_count":3,"year":2025,"published":"2025-05-31","venue":null,"cited_by":null,"doi":null,"arxiv_id":"2506.00428","nct_id":null,"source_url":"https://arxiv.org/abs/2506.00428","archive_url":"https://space-kitty.com/papers/arxiv:2506.00428","updated":"2025-12-13","pmid":null,"pmcid":null,"openalex_id":null,"pdf_url":"https://arxiv.org/pdf/2506.00428","abstract":"The textbook algorithm for real-weighted single-source shortest paths takes $O(m n)$ time on a graph with $m$ edges and $n$ vertices. The breakthrough algorithm by Fineman [Fin24] takes $\\tilde{O}(m n^{8/9})$ randomized time. The running time was subsequently improved to $\\tilde{O}(mn^{4/5})$ [HJQ25]. We build on [Fin24; HJQ25] to obtain an $\\tilde{O}(m n^{3/4} + m^{4/5} n)$ randomized running time. (Equivalently, $\\tilde{O}(mn^{3/4})$ for $m \\geq n^{5/4}$, and $\\tilde{O}(m^{4/5} n)$ for $m \\leq n^{5/4}$.) The main new technique replaces the hop-reducing auxiliary graph from [Fin24] with a bootstrapping process where constant-hop reducers for small subgraphs of the input graph are iteratively amplified and expanded until the desired polynomial-hop reduction is achieved over the entire graph.","abstract_license":"CC0-1.0 (arXiv descriptive metadata)","license":null,"is_open_access":false,"retracted":false,"peer_reviewed":false,"categories":["cs.DS"],"details":null,"provenance":"arxiv","attribution":"Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.","topics":[],"cited_in_dossiers":[]}