Cornell University
Cornell University: Counting Solution Clusters in Graph Coloring Problems
This paper proposes one of the first scalable methods for estimating the number of clusters of solutions of graph coloring problems using a belief propagation like algorithm.