The parameterized complexity of Strong Conflict-Free Vertex-Connection Colorability

This paper continues the study of a new variant of graph coloring with a connectivity constraint recently introduced by Hsieh et al. [COCOON 2024]. A path in a vertex-colored graph is called conflict-free if there is a color that appears exactly once on its vertices. A connected graph is said to be strongly conflict-free vertex-connection $k$-colorable if it admits a (proper) vertex $k$-coloring such that any two distinct vertices are connected by a conflict-free shortest path. Among others, we show that deciding, for a given graph $G$ and an integer $k$, whether $G$ is strongly conflict-free $k$-colorable is fixed-parameter tractable when parameterized by the vertex cover number. But under the standard complexity-theoretic assumption NP $\not\subseteq$ coNP/poly, deciding, for a given graph $G$, whether $G$ is strongly conflict-free $3$-colorable does not admit a polynomial kernel, even for bipartite graphs. This kernel lower bound is in stark contrast to the ordinal $k$-Coloring problem which is known to admit a polynomial kernel when parameterized by the vertex cover number.

💡 Research Summary

This paper presents a comprehensive parameterized complexity analysis of the Strong Conflict-Free Vertex-Connection Coloring (strong CFVC) problem. This is a recently introduced variant of graph coloring where, given a connected graph G and an integer k, one asks if there exists a proper vertex k-coloring with an additional strong connectivity constraint: for every pair of distinct vertices u and v, there must exist a shortest u,v-path that is “conflict-free.” A path is conflict-free if there is a color that appears exactly once on its vertices.

The authors investigate the problem through the lens of structural parameterization, focusing primarily on the vertex cover number. They establish two main, contrasting results:

1. Fixed-Parameter Tractability (FPT) for the General Problem: They prove that the decision problem strong CFVC number (where k is part of the input) is fixed-parameter tractable when parameterized by the vertex cover number. They provide an FPT algorithm that first applies a safe reduction rule: for any subset S of the vertex cover X, if there are more than k+1 independent set vertices whose neighborhood in X is exactly S, all but k+1 of them can be removed. After exhaustively applying this rule, the kernelized instance has size bounded by a function of |X|. Checking all possible k-colorings on this kernel leads to an overall runtime of f(|X|) * poly(n), establishing FPT membership.

2. Polynomial Kernel Lower Bound for the Fixed Case: In a stark contrast, they show that for the specific case of k=3 (strong CFVC 3-coloring), the problem does not admit a polynomial kernel when parameterized by the vertex cover number, unless NP ⊆ coNP/poly. This lower bound holds even when the input graph is restricted to be bipartite and of bounded diameter. The proof involves a polynomial parameter transformation from the classical NP-complete problem Positive Not-All-Equal SAT (NAE SAT) to strong CFVC 3-coloring. They construct a bipartite graph G from a NAE SAT instance such that G is strong CFVC 3-colorable if and only if the Boolean formula has a not-all-equal assignment. Since NAE SAT parameterized by the number of variables is known not to have a polynomial kernel under the stated complexity assumption, the transformation implies the same hardness for strong CFVC 3-coloring parameterized by vertex cover. This result highlights a key difference from the ordinary k-Coloring problem, which admits a polynomial kernel under the same parameterization.

Key technical insights include the non-monotonicity of strong CFVC colorings (deleting vertices can turn a yes-instance into a no-instance), the careful handling of twin vertices in the FPT algorithm, and the intricate gadget construction for the kernel lower bound that encodes the NAE condition into the requirement for conflict-free shortest paths.

In conclusion, the paper delineates a precise parameterized complexity landscape for the strong CFVC problem. It demonstrates that while the general problem is FPT by vertex cover, the fixed-parameter case (k=3) resists efficient data reduction (kernelization), revealing a nuanced complexity dichotomy driven by the interplay of coloring and strong connectivity constraints.