A performance study on synchronous and asynchronous updates in particle swarm optimization
Juan Rada-Vilela, Mengjie Zhang, Winston K.G. Seah · 2011
This work provides a further study on the difference between synchronous and asynchronous updates in Particle Swarm Optimization with different neighborhood sizes ranging from local best to global best. Ten well-known functions are used as benchmarks on both variants. Statistical tests performed on the results provide strong evidence to claim that synchronous updates yield in general better results with similar or even faster speed of convergence than its asynchronous counterpart, contrary to observations and conclusions of previous studies based solely on descriptive statistics.