Feasible and infeasible region search for the maximally diverse grouping problem

Abstract

The maximally diverse grouping problem (MDGP) involves assigning elements into disjoint groups, and its objective is to maximize the total diversity of the groups where each pair of elements in a group has a certain diversity. MDGP has broad application and is known to be an NP-hard problem. In this paper, we integrate infeasible region search, exploration, and exploitation into a new algorithm called the feasible and infeasible region (FIFR) algorithm. The FIFR algorithm is significantly better than three state-of-the-art algorithms on 500 instances widely used in the literature.