2026

Journal

Evolutionary Intelligence 19 (1), 9

Genetic Algorithms (GAs) are widely used for solving complex optimization problems thanks to their ability to explore vast solution spaces and adapt to diverse constraints. Among the key components of GAs, the crossover operator critically influences the balance between exploration and exploitation, by combining promising genetic material and exploring new regions of the solution space. However, traditional crossover operators typically apply a single recombination per parent pair, which limits their ability to deeply exploit high-quality gene patterns and often leads to premature convergence. To address this limitation, we propose the Decuple Crossover Scheme (DCS), a novel hierarchical crossover scheme that intensifies intra-generational recombination. In DCS, a pair of parents initially produce two offspring, which then undergo multiple recombinations with the original parents, resulting in promising candidate offspring. From this pool, the two best individuals are selected for the next generation, enabling deeper exploitation of genetic material and improved population diversity. The effectiveness of DCS is demonstrated through a proof-of-concept application on the Traveling Salesman Problem (TSP) and the Vehicle Routing Problem (VRP), showcasing its potential for enhancing solution quality and convergence speed. These results highlight the broader applicability of DCS to diverse optimization challenges, making it a promising tool for advancing the field of evolutionary computation.

@article{derouiche2026decuple,
  title={A decuple crossover scheme in genetic algorithms: a step toward deep evolution},
  author={Derouiche, Hana and Elarbi, Maha and Bechikh, Slim},
  journal={Evolutionary Intelligence},
  volume={19},
  number={1},
  pages={9},
  year={2026},
  publisher={Springer}
}