On the Structural Parameterized Complexity of Defective Coloring

Abstract

In this paper, we consider the problem Defective Coloring. Given a graph G and two positive integers k and $\Delta ^*$, the objective is to determine whether it is possible to obtain a coloring (not necessarily proper) of the vertices of G using at most k colors such that each vertex in a color class c has at most $\Delta ^*$ neighbors in the same color class. Defective Coloring is a generalization of Graph Coloring with $\Delta ^*=0$. The optimization variant of this problem, which aims to find the minimum number of colors k, is known to be NP-hard even for split graphs and cographs.

Belmonte, Lampis, and Mitsou (SIDMA 2020) showed that Defective Coloring is W[1]-hard when parameterized by tree-width, path-width, tree-depth, or feedback vertex set. The problem is W[1]-hard parameterized by modular-width or clique-width as Defective Coloring is NP-hard on cographs. They asked as an open question whether Defective Coloring is fixed-parameter tractable ($\texttt {FPT}$) when parameterized by modular-width, clique-width or neighborhood diversity combined with either k or $\Delta ^*$. In an effort to address the question concerning modular-width, this study investigates the parameters neighborhood diversity and twin-cover, which are special cases of modular-width. We show that Defective Coloring is $\texttt {FPT}$ when parameterized by twin-cover, distance to disjoint paths, or the combined parameters neighborhood diversity and k. The latter result implies an FPT algorithm for complete-d-partite graphs, a subclass of cographs, parameterized by d. This provides a partial response to an open question raised in the above paper. We present an algorithm for graphs with bounded distance to d-degree and as a corollary we obtain an FPTalgorithm parameterized by distance to disjoint paths. Furthermore, the study also presents a 1-additive approximation algorithm for split graphs.