🤖 AI Summary
This paper investigates the structural properties of graphs covered by “fat shortest paths”—i.e., graphs in which every vertex lies within distance at most ρ of some shortest path. We introduce the ρ-fault-tolerant path cover model and establish quantitative coarse-geometric relationships between it and classical graph parameters: such graphs are coarsely quasi-isometric to graphs of pathwidth $k^{O(k)}$ and maximum degree $O(k)$. Furthermore, we construct a path decomposition with spherical covering property: each bag is covered by $k^{O(k)}$ balls of radius at most $2
ho$. This decomposition paradigm bridges coarse geometry and path decomposition theory for the first time, yielding a foundational framework for designing efficient approximation algorithms for distance-constrained combinatorial optimization problems—including Maximum Distance-Independent Set and Minimum Distance-Dominating Set.
📝 Abstract
Dumas, Foucaud, Perez, and Todinca [SIAM J. Disc. Math., 2024] proved that if the vertex set of a graph $G$ can be covered by $k$ shortest paths, then the pathwidth of $G$ is bounded by $mathcal{O}(k cdot 3^k)$. We prove a coarse variant of this theorem: if in a graph $G$ one can find~$k$ shortest paths such that every vertex is at distance at most $
ho$ from one of them, then $G$ is $(3,12
ho)$-quasi-isometric to a graph of pathwidth $k^{mathcal{O}(k)}$ and maximum degree $mathcal{O}(k)$, and $G$ admits a path-partition-decomposition whose bags are coverable by $k^{mathcal{O}(k)}$ balls of radius at most $2
ho$ and vertices from non-adjacent bags are at distance larger than $2
ho$. We also discuss applications of such decompositions in the context of algorithms for finding maximum distance independent sets and minimum distance dominating sets in graphs.