Generalizing Brooks' theorem via Partial Coloring is Hard Classically and Locally
Abstract
We investigate the classical and distributed complexity of \emph{-partial -coloring} where , a natural generalization of Brooks' theorem where each vertex should be colored from the palette such that it must have at least neighbors colored differently. Das, Fraigniaud, and Ros{\'{e}}n~[OPODIS 2023] showed that the problem of -partial -coloring admits efficient centralized and distributed algorithms and posed an open problem about the status of the distributed complexity of -partial -coloring. We show that the problem becomes significantly harder when the number of colors is reduced from to for every constant . In the classical setting, we prove that deciding whether a graph admits a -partial -coloring is NP-complete for every constant , revealing a sharp contrast with the linear-time solvable -color case. For the distributed LOCAL model, we establish an -round lower bound for computing -partial -colorings, even when the graph is guaranteed to be -partial -colorable. This demonstrates an exponential separation from the -round algorithms known for -colorings. Our results leverage novel structural characterizations of ``hard instances'' where partial coloring reduces to proper coloring, and we construct intricate graph gadgets to prove lower bounds via indistinguishability arguments.
Keywords
Cite
@article{arxiv.2508.16308,
title = {Generalizing Brooks' theorem via Partial Coloring is Hard Classically and Locally},
author = {Jan Bok and Avinandan Das and Anna Gujgiczer and Nikola Jedličková},
journal= {arXiv preprint arXiv:2508.16308},
year = {2025}
}