Distributive graph algorithms Global solutions from local data

Abstract

This paper deals with distributed graph algorithms. Processors reside in the vertices of a graph G and communicate only with their neighbors. The system is synchronous and reliable, there is no limit on message lengths and local computation is instantaneous. The results: A maximal independent set in an n-cycle cannot be found faster than Ω(log* n) and this is optimal by [CV]. The d-regular tree of radius r cannot be colored with fewer than √d colors in time 2r / 3. If Δ is the largest degree in G which has order n, then in time O(log*n) it can be colored with O(Δ2) colors.