Skip to content

Latest commit

 

History

History
9 lines (7 loc) · 1.48 KB

README.md

File metadata and controls

9 lines (7 loc) · 1.48 KB

Example program for Best-of-Random Differential Evolution (BoRDE) optimization method

This Fortran program provides an example use case to demonstrate the application of the Best-of-Random Differential Evolution (BoRDE) method with adaptive parameter setting.

The main program entry point is in file Fit_FunctionParam.f90, while the main Differential Evolution algorithm is in file/subroutine BestOfRandomDE, which is called from subroutine MinBoRDE. The code is not particularly polished or documented, but should provide interested users with an example as a starting point.

Origin and some details

The fit method used is a self-adaptive Best-of-Random Differential Evolution (BoRDE) variant based on the paper by Lin et al. (2010, J. Glob. Optim., https://doi.org/10.1007/s10898-010-9535-7, with some modifications by Andreas Zuend, who introduced the self-adaptive feature; see information in appendix of Zuend et al. (2010, Atmos. Chem. Phys., https://doi.org/10.5194/acp-10-7795-2010). Differential Evolution (DE) as a derivative-free optimization method was first introduced by Storn, R., and Price, K.V., (1996): "Minimizing the real function of the ICEC'96 contest by differential evolution", IEEE conf. on Evolutionary Computation, 842-844. The Fortran 90 DE code used here is based on the version written and distributed by Dr. Feng-Sheng Wang, Department of Chemical Engineering, National Chung Cheng University, Chia-Yi 621, Taiwan. It has only been modified slightly and not fully modernized.