A study on the role of memory in heuristic search. The case for discrete dispersion problems
A. Martínez Gavara, H. Zheng, R. Martí, F. Glover
This work focuses on the design of effective solution methodologies for a combinatorial optimization problem that integrates diversity maximization with group fairness constraints. In particular, we investigate two metaheuristic paradigms, the Greedy Randomized Adaptive Search Procedure (GRASP) and Probabilistic Tabu Search (PTS), and analyze how different memory mechanisms influence search performance.
We propose two novel PTS variants that incorporate memory in distinct phases: one in the improvement phase through tabu and frequency-based structures, and another in the construction phase via probabilistic biasing guided by elite solutions and entropy-based control. We further introduce an advanced hybrid framework that integrates memory-based perturbation with strategic oscillation to effectively balance intensification and diversification. Results highlight the key role of memory in improving performance.
Keywords: metaheuristics, combinatorial optimization
Scheduled
GT Heurísticas y metaheurísticas II
September 2, 2026 5:40 PM
Aula 29
Other papers in the same session
M. Alcaide Catalán, A. Martínez Gavara, S. Pérez Peló
R. Martín Santamaría, S. Salazar Cárdenas, J. M. Colmenar Verdugo, E. García Pardo
J. Sánchez-Díaz, S. Cavero, E. García Pardo, L. Hernando