Lower bounds for dominating set reconfiguration on sparse (directed) graphs

TL;DR

This paper investigates the computational complexity of reconfiguring dominating sets in sparse (directed) graphs, establishing hardness results and parameterized complexity classifications for various graph classes and reconfiguration variants.

Contribution

It proves W[2]-hardness of dominating set reconfiguration with token sliding on sparse graphs and explores complexity in directed graphs, extending known results.

Findings

W[2]-hardness for dominating set reconfiguration with token sliding on graphs with bounded pathwidth

NP-hardness of directed reconfiguration on DAGs of treewidth 5

W[2]-hardness for directed reconfiguration on DAGs with bounded depth and pathwidth

Abstract

In a graph, a vertex dominates itself and its neighbors, and a dominating set is a set of vertices that together dominate the entire graph. Given a graph and two dominating sets of equal size $k$, the {\em Dominating Set Reconfiguration with Token sliding} (DSR-TS) problem asks whether one can, by iteratively replacing a vertex by an adjacent one, transform the first set into the second one, while ensuring that every set during the reconfiguration process is a dominating set. The token jumping variant, where a vertex can be replaced by a non-adjacent one, is known to be efficiently solvable on many graph classes such as planar, bounded treewidth, and the very broad notion of nowhere-dense classes of graphs. Alternatively, some algorithms also exist for the reconfiguration of independent sets in the token sliding paradigm for graph classes with bounded degree or large girth. We show that DSR-TS is W[2]-hard when parameterized $k$, the pathwidth of the instance, and the iteration of the reconfiguration sequence (a recently introduced parameter). This is a setting where both the token jumping and the independent set variants are fixed parameter tractable. Not restricting the iteration yields W[2] hardness already on graphs with treewidth 9 and pathwidth 13. In the directed variant (DSR-DTS), we are only allowed to replace a vertex with an out-neighbor. We show that DSR-DTS is NP-hard on DAGs of treewidth 5 and W[2]-hard for both the case of DAGs of depth 3 parameterized by $k$, and the case of DAGs when parameterized by $k$ and the pathwidth of the instance (independent set reconfiguration is again FPT in both settings).