Category
JFL, Terrace Conference Room (001)
Description
Genetic algorithms (GAs) are a unique subset of optimization algorithms that are widely used in solving multi-objective engineering design optimization problems. Although a large number of GAs exist in literature, with many available in Python and C/C++ libraries, new algorithms are still needed to continuously improve the solution quality and/or computational efficiency. In the previously developed optimization software GimOPT [1], the non-dominated sorting genetic algorithm II (NSGA-II) [2] was implemented with a dominance-tree based sorting technique [3] that offers great computational efficiency over other implementations. In this research, NSGA-III [4] was implemented into GimOPT using the dominance-tree based sorting algorithm. The speedup of this implementation in plain C++ was evaluated by comparing it to existing libraries such as pymoo for NSGA-II and NSGA-III, and the solution quality of GimOPT's NSGA-II and NSGA-III were compared using benchmark problems. The results showed that this implementation of NSGA-III offers significant improvement on the computational speed of pymoo and simultaneously bolsters solutional quality in comparison to GimOPT's NSGA-II. This new implementation of NSGA-III will relieve researchers of prior computational restraints and empower them to tackle new and difficult multi- and many-objective optimization problems. References [1] H. Fang, M.F. Horstemeyer (2006). A generic optimizer interface for programming-free optimization systems. Advances in Engineering Software, 37(6), 360-369. (https://doi.org/10.1016/j.advengsoft.2005.10.002) [2] K. Deb, A. Pratab, S. Agarwal, T. Meyarivan (2002). A fast and elitist multi-objective genetic algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, 6(2), 182-197. (https://doi.org/10.1109/4235.996017) [3] H. Fang, Q. Wang, Y. Tu, M.F. Horstemeyer (2008). An efficient non-dominated sorting method for evolutionary algorithms. Evolutionary Computation, 16(3), 355-384. (https://doi.org/10.1162/evco.2008.16.3.355) [4] K. Deb, H. Jain (2014). An Evolutionary Many-Objective Optimization Algorithm Using Reference-Point-Based Nondominated Sorting Approach, Part I: Solving Problems with Box Constraints. IEEE Transactions on Evolutionary Computation, 18(4), 577-601. (https://doi.org/10.1109/TEVC.2013.2281535)
Speed Grows on Trees: NSGA-III & Design Optimization
JFL, Terrace Conference Room (001)
Genetic algorithms (GAs) are a unique subset of optimization algorithms that are widely used in solving multi-objective engineering design optimization problems. Although a large number of GAs exist in literature, with many available in Python and C/C++ libraries, new algorithms are still needed to continuously improve the solution quality and/or computational efficiency. In the previously developed optimization software GimOPT [1], the non-dominated sorting genetic algorithm II (NSGA-II) [2] was implemented with a dominance-tree based sorting technique [3] that offers great computational efficiency over other implementations. In this research, NSGA-III [4] was implemented into GimOPT using the dominance-tree based sorting algorithm. The speedup of this implementation in plain C++ was evaluated by comparing it to existing libraries such as pymoo for NSGA-II and NSGA-III, and the solution quality of GimOPT's NSGA-II and NSGA-III were compared using benchmark problems. The results showed that this implementation of NSGA-III offers significant improvement on the computational speed of pymoo and simultaneously bolsters solutional quality in comparison to GimOPT's NSGA-II. This new implementation of NSGA-III will relieve researchers of prior computational restraints and empower them to tackle new and difficult multi- and many-objective optimization problems. References [1] H. Fang, M.F. Horstemeyer (2006). A generic optimizer interface for programming-free optimization systems. Advances in Engineering Software, 37(6), 360-369. (https://doi.org/10.1016/j.advengsoft.2005.10.002) [2] K. Deb, A. Pratab, S. Agarwal, T. Meyarivan (2002). A fast and elitist multi-objective genetic algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, 6(2), 182-197. (https://doi.org/10.1109/4235.996017) [3] H. Fang, Q. Wang, Y. Tu, M.F. Horstemeyer (2008). An efficient non-dominated sorting method for evolutionary algorithms. Evolutionary Computation, 16(3), 355-384. (https://doi.org/10.1162/evco.2008.16.3.355) [4] K. Deb, H. Jain (2014). An Evolutionary Many-Objective Optimization Algorithm Using Reference-Point-Based Nondominated Sorting Approach, Part I: Solving Problems with Box Constraints. IEEE Transactions on Evolutionary Computation, 18(4), 577-601. (https://doi.org/10.1109/TEVC.2013.2281535)
Comments
Undergraduate