abstract
- Diversity indicators are important for assessing Pareto front approximations in evolutionary multi-objective optimization. Weitzman diversity is a recursive set-based function defined on pairwise dissimilarities with strong theoretical properties. However, its exponential complexity makes exact computation infeasible for large point sets. In this paper, we reformulate Weitzman diversity as an optimization problem over removal permutations induced by its recursive definition. Then, we study three heuristics adapted from classical permutation strategies for the Traveling Salesman Problem and propose a new set-based method, the Global Max¿Min Heuristic (GMM). Experiments on three-objective Pareto front geometries under coverage and uniformity loss show that the proposed methods are generally consistent with Weitzman diversity while remaining computationally tractable. Overall, GMM achieves the highest induced values.