Parallelization of Enhanced Firework Algorithm using MapReduce

Swarm intelligence algorithms are inherently parallel since different individuals in the swarm perform independent computations at different positions simultaneously. Hence, these algorithms lend themselves well to parallel implementations thereby speeding up the optimization process. FireWorks Algorithm (FWA) is a recently proposed swarm intelligence algorithm for optimization. This work investigates the scalability of the parallelization of the Enhanced FireWorks Algorithm (EFWA), which is an improved version of FWA. The authors use the MapReduce platform for parallelizing EFWA, investigate its ability to scale, and report on the speedup obtained on different benchmark functions for increasing problem dimensions.