Answers are grounded in this paper and cite their sources.
This paper demonstrates novel ways to enhance reproduction efficiency in decomposition-based evolutionary multi-objective optimization. The foundation for all developments is the concept of local mini-archives associated with subproblems. These mini-archives preserve locally best contributing solutions and are designed to be computationally efficient. During steady-state evolution, the method exploits these mini-archives to perform advantageous gradient-based adjustments to offspring decision vectors constructed using regular crossover and mutation. The paper also introduces a dynamic parent selection mechanism that allows the method to shift from more neighborhood-based exploratory behavior at the beginning of the optimization to more archive-based exploitation at the end. All developments were thoroughly tested in both a posteriori and an interactive optimization context. The latter optimization is typically more demanding and requires several adaptations of the population within a single optimizer run. In both cases, the developments allowed the construction of more promising populations with fewer function evaluations.
No takes yet. Share an insight, caveat, or question.
Michał K. Tomczyk (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: