Round and Communication Efficient Graph Coloring
Yi-Jun Chang, Gopinath Mishra, Hung Thuan Nguyen, Farrel D Salim
TL;DR
This work studies graph coloring in a two-party edge-partition model, where edges are split between two players and the goal is to compute valid vertex or edge colorings with minimal communication and fast rounds. The authors develop a randomized $O(n)$-bit, round-efficient protocol for $(Δ+1)$-vertex coloring with $O(\log\log n\cdot\log Δ)$ rounds, and a deterministic $O(n)$-bit, $O(1)$-round protocol for $(2Δ-1)$-edge coloring, complemented by tight $Ω(n)$-bit communication lower bounds and a $W$-streaming space lower bound. Key techniques include the Random-Color-Trial framework, color sampling via Color-Sample, palette sparsification for D1LC, the use of Δ-perfect matchings, and a deferral-heavy strategy for edge coloring guided by Fournier-type results. These contributions yield near-optimal communication, meaningful round-efficiency, and foundational lower bounds that connect the two-party model to streaming and other distributed settings, with several open questions about further round-communication reductions and color-count relaxations.
Abstract
In the context of communication complexity, we explore protocols for graph coloring, focusing on the vertex and edge coloring problems in $n$-vertex graphs $G$ with a maximum degree $Δ$. We consider a scenario where the edges of $G$ are partitioned between two players. Our first contribution is a randomized protocol that efficiently finds a $(Δ+ 1)$-vertex coloring of $G$, utilizing $O(n)$ bits of communication in expectation and completing in $O(\log \log n \cdot \log Δ)$ rounds in the worst case. This advancement represents a significant improvement over the work of Flin and Mittal [Distributed Computing 2025], who achieved the same communication cost but required $O(n)$ rounds in expectation, thereby making a significant reduction in the round complexity. Our second contribution is a deterministic protocol to compute a $(2Δ- 1)$-edge coloring of $G$, which maintains the same $O(n)$ bits of communication and uses only $O(1)$ rounds. We complement the result with a tight $Ω(n)$-bit lower bound on the communication complexity of the $(2Δ-1)$-edge coloring problem, while a similar $Ω(n)$ lower bound for the $(Δ+1)$-vertex coloring problem has been established by Flin and Mittal [Distributed Computing 2025]. Our result implies a space lower bound of $Ω(n)$ bits for $(2Δ- 1)$-edge coloring in the $W$-streaming model, which is the first non-trivial space lower bound for edge coloring in the $W$-streaming model.