A new variant of the Genetic Algorithm (GA) inspired by monogamy mating system is put forward. The Monogamous Pairs Genetic Algorithm (MopGA) incorporates two important operations: pair bonding and infidelity at a small probability. With pair bonding, parents continue to mate at each iteration until their bond expires. In the meantime, infidelity generates variety and promotes diversity via mating with extrapair. We evaluate the algorithm's performance using various parametrizations and making comparisons to the Standard Genetic Algorithm (SGA) based on the Hierarchical If-and-Only-If (HIFF) and Deceptive (DP) functions. Empirical results show that incorporating pair bonding is a practical move. Improvement in performance in terms of solution quality and computational efforts have been observed for all test problems. Additionally, we also report the effectiveness of MopGA in handling easy and difficult sudoku puzzles.
展开▼
