Preface

The swarm intelligence optimization method is used to study bioor non-bioinspired and population-based iterative algorithms for seeking the intrinsic cooperative mechanism in a swarm. It has recently attracted a great deal of attention from researchers from different fields and diverse domains. Many novel algorithms and their efficient improvements have been proposed continuously. The swarm intelligence optimization method is increasingly becoming one of the hottest and most important paradigms under the big umbrella of evolutionary computation (EC). Inspired by the fireworks explosion in the night sky, fireworks algorithm, abbreviated as FWA, was proposed by the author in 2010. FWA is a swarm intelligence optimization algorithm, which seems effective at finding a good enough solution to the global optimum of a complex optimization problem. In FWA, as a firework explodes, a shower of sparks is shown in the adjacent area. These sparks explode again and generate other showers of sparks in a smaller area. Gradually, the sparks search the whole solution space in a fine structure and focus on a small region to eventually find (a) good enough solution(s). To my memory, on the night of the Eve of the 2006 Chinese Lunar Year, as the municipality authorities of Beijing lifted the ban on fireworks during that Spring Festival, people in Beijing set off a large amount of fireworks with sparks of diverse colors which lighted up the dark sky in a variety of beautiful patterns. While I stared at the glorious scene and colorful patterns for a long time, suddenly an idea came to my mind that the way fireworks explode may be an efficient and effective way or strategy to search for a potential good solution in a vast solution space. Such a search strategy would be different from the established ones in the EC community. Since then I began to study this explosion-like search method. Like other practical optimization algorithms, FWA is able to fulfill three user requirements given by Storn and Price in 1997. First of all, FWA can process linear, nonlinear, and multi-model test functions. Second, FWA can be parallelized for tackling complicated real-world problems. Third, FWA has good convergence properties and can always find good enough solution(s) for a global minimization problem.