An exact algorithm for a core/periphery bipartitioning problem

The discrete optimization problem associated with partitioning a set of actors into core and periphery subsets has typically been approached using approximate procedures such as exchange heuristics, genetic algorithms, and simulated annealing. Although these procedures are effective and scalable for large networks, they do not guarantee an optimal bipartition. In this paper, an exact algorithm is presented for a core/periphery bipartitioning problem. Unlike the approximate procedures in the extant literature, this new algorithm, which is based on the principles of branch-and-bound programming, affords a guarantee of an optimal bipartition. Computational results for empirical and simulated data sets reveal that the proposed algorithm is extremely efficient for networks with 40 or fewer actors, and is often scalable for networks with up to 60 actors.