22
7

A Distributed Palette Sparsification Theorem

Abstract

The celebrated palette sparsification result of [Assadi, Chen, and Khanna SODA'19] shows that to compute a Δ+1\Delta+1 coloring of the graph, where Δ\Delta denotes the maximum degree, it suffices if each node limits its color choice to O(logn)O(\log n) independently sampled colors in {1,2,,Δ+1}\{1, 2, \dots, \Delta+1\}. They showed that it is possible to color the resulting sparsified graph -- the spanning subgraph with edges between neighbors that sampled a common color, which are only O~(n)\tilde{O}(n) edges -- and obtain a Δ+1\Delta+1 coloring for the original graph. However, to compute the actual coloring, that information must be gathered at a single location for centralized processing. We seek instead a local algorithm to compute such a coloring in the sparsified graph. The question is if this can be achieved in poly(logn)\operatorname{poly}(\log n) distributed rounds with small messages. Our main result is an algorithm that computes a Δ+1\Delta+1-coloring after palette sparsification with O(log2n)O(\log^2 n) random colors per node and runs in O(log2Δ+log3logn)O(\log^2 \Delta + \log^3 \log n) rounds on the sparsified graph, using O(logn)O(\log n)-bit messages. We show that this is close to the best possible: any distributed Δ+1\Delta+1-coloring algorithm that runs in the LOCAL model on the sparsified graph, given by palette sparsification, for any poly(logn)\operatorname{poly}(\log n) colors per node, requires Ω(logΔ/loglogn)\Omega(\log \Delta / \log\log n) rounds. This distributed palette sparsification result leads to the first poly(logn)\operatorname{poly}(\log n)-round algorithms for Δ+1\Delta+1-coloring in two previously studied distributed models: the Node Capacitated Clique, and the cluster graph model.

View on arXiv
Comments on this paper