A polynomial bound on the pathwidth of graphs edge-coverable by $k$ shortest paths
Julien Baste, Lucas De Meyer, Ugo Giocanti, Etienne Objois, Timothé Picavet
TL;DR
This work strengthens the understanding of how edge-coverings by shortest paths constrain a graph's width parameters. The authors prove a polynomial bound $pw(G)=O(k^4)$ for graphs edge-coverable by $k$ shortest paths, improving the previous exponential guarantees, and they establish exact bounds $pw(G)\le k$ for $k\le 3$. A key technical contribution is a separator-based framework built around a $6k$-layered substructure that yields controlled pathwidth after removing a polynomial-size set. Additionally, they show that if $G$ is edge-coverable by two isometric trees, then $tw(G)\le 2$, contrasting with known vertex-cover phenomena. These results have potential algorithmic impact for Isometric Path Cover problems and related graph-routing questions, and they leave intriguing open questions about linear bounds and vertex-cover analogues.
Abstract
Dumas, Foucaud, Perez and Todinca (2024) recently proved that every graph whose edges can be covered by $k$ shortest paths has pathwidth at most $O(3^k)$. In this paper, we improve this upper bound on the pathwidth to a polynomial one; namely, we show that every graph whose edge set can be covered by $k$ shortest paths has pathwidth $O(k^4)$, answering a question from the same paper. Moreover, we prove that when $k\leq 3$, every such graph has pathwidth at most $k$ (and this bound is tight). Finally, we show that even though there exist graphs with arbitrarily large treewidth whose vertex set can be covered by $2$ isometric trees, every graph whose set of edges can be covered by $2$ isometric trees has treewidth at most $2$.
