Parallel solving of optimization problems on distributed systems by asynchronous differential evolution

Asynchronous Differential Evolution (ADE) is an Evolutionary algorithm to solve derivative-free global optimization problems. It provides an effective parallel realization. ADE with Restart (ADE-R) automatically restarts calculations with an increased population size while stagnation or population degeneration is diagnosed. Thus the probability of convergence can approach a unity. Performance and speed-up for parallel calculations of various ADE-R strategies is compared with the classical Differential Evolution strategies with Restart (DE-R).