Fireworks Algorithm versus Plant Propagation Algorithm

摘要

In recent years, the field of Evolutionary Algorithms has seen a tremendous increase in novel methods. While these algorithmic innovations often show excellent results on relatively limited domains, they are less often rigorously cross-tested or compared to other state-of-the-art developments. Two of these methods, quite similar in their appearance, are the Fireworks Algorithm and Plant Propagation Algorithm. This study compares the similarities and differences between these two algorithms, from both quantitative and qualitative perspectives, by comparing them on a set of carefully chosen benchmark functions. The Fireworks Algorithm outperforms the Plant Propagation Algorithm on the majority of these, but when the functions are shifted slightly, Plant Propagation gives better results. Reasons behind these surprising differences are presented, and comparison methods for evolutionary algorithms are discussed in a wider context. All source code, graphs, test functions, and algorithmic implementations have been made publicly available for reference and further reuse.