Replication data for: Investigating the impacts of different crossover and mutation operators in decomposition-based multi-objective expensive discrete optimization

The filenames follow the pattern: igd_"crossover_operator"_"mutation_operator"_"size".txt. For example, igd_uniform_crossover_bitflip_mutation_50.txt contains the IGD values for each of the 31 runs of MOEA/D using the uniform crossover and bit-flip mutation operators with 50 variables. The UNISIM_II_D subfolder contains folders with the MOEA/D run results for each operator combination.

For instance, in the twopointC_bitflipM folder, files contain iteration-wise results for that operator combination. Additionally, this subfolder includes a non_dominated.xlsx file with the non-dominated solutions obtained for each combination, and a reference_set.csv file containing the values used as the reference set for plotting the solution frontiers. source_code.zip: contains the source code (in Python) for reproducing the experiments.

The code includes: (i) implementations of the MOEA/D algorithms; (ii) test instances of the 0-1 KP benchmark problems with different dimensionalities (50, 100, 200, 300, and 400); (iii) reference sets (refsets); (iv) .txt files containing the IGD values of the 31 independent runs for all conducted experiments, named according to the pattern: igd_"crossover_operator"_"mutation_operator"_"size".txt. Usage instructions are provided in the README.txt file included in the package.