Distributed simulation of coloring graph vertices

The problem of coloring graph vertices, known to be NP-complete, is discussed. The authors present an alternative solution to this problem as a distributed simulation which uses a parallel randomized heuristic algorithm for making local decisions to color graph vertices. The algorithm is based on an inter-particle system from statistical mechanics. Since the algorithm works locally, it is likely to be highly parallel. The simulation of a coloring process on n vertices of a graph can be seen as a set of n distributed processes running in parallel. The simulation algorithm is implemented in Occam 2 language and runs on a transputer system.<>