Diversity-Guided Mutation in Genetic Algorithms: Cost, Parent, and Variable-Level Adaptation

This capsule contains the source code and reproduction data for the research paper **"Diversity-Guided Mutation in Genetic Algorithms: Cost, Parent, and Variable-Level Adaptation"**. ### Summary The study presents a dynamic, diversity-guided mutation strategy that balances exploration and exploitation by monitoring three complementary metrics: * **Cost Diversity** * **Parent Diversity** * **Variable Columnar Diversity** Unlike static or fitness-based adaptive approaches, this method utilizes these metrics to regulate mutation rates, select candidate solutions, and bias variable value alterations in real-time. ### Experimental Evaluation The approach was evaluated in both 2D and 7D configurations using a comprehensive suite of benchmark functions, including: Ackley, Beale, Booth, Griewank, Himmelblau, Rastrigin, Rosenbrock, Sphere, and Styblinski–Tang. **Performance Metrics:** * Average Fitness * Convergence Generations * Early Convergence * Area Under the Convergence Curve (AUCC) * Standard Deviation ### Results Across most experiments, the diversity-guided mutation demonstrated improved convergence stability, higher solution accuracy, and reduced variability across independent runs. The results indicate that diversity-regulated mutation effectively preserves population heterogeneity and supports efficient early progress. tags: - Genetic Algorithm - Evolutionary Computation - Optimization - Self-adaptive - Benchmarking - Python