Genetic algorithms (GAs) are among the most widely used stochastic optimization algorithms. They have been applied to solve many real-life optimization problems belonging to different domains.
This paper proposes a new variant of genetic algorithms called multi-offspring genetic algorithms (MO-GAs). The proposed approach involving multiple offspring is inspired by the process of natural evolution. In this algorithm, the possibility of having several offspring has been explored. For testing the performance of MO-GA, the authors have applied it to the following six different benchmark instances of the traveling salesman problem (TSP) of different sizes, ranging from 14 cities to 3,038 cities: “(1) wxp20 with 20 cities [...], (2) burma14 in TSPLIB with 14 cities, (3) eil51 in TSPLIB with 51 cities, (4) kroB100 in TSPLIB with 100 cities, (5) pr1002 in TSPLIB with 1,002 cities, and (6) pcb3038 in TSPLIB with 3,038 cities.” The authors have performed the simulations using MATLAB R2013a.
It is evident from the results that MO-GA performs better than the basic genetic algorithm (BGA). The authors have compared the performance of MO-GA only with the BGA; it could also have been compared with the other recently proposed state-of-the-art variants of GA for establishing its supremacy over them.
The paper is written nicely and is well organized. It is highly recommended for students and researchers working in the fields of GA and soft computing.