Evolutionary Computing and Swarm Intelligence

Efficient variants of the Differential Evolution algorithm have been derived to provide elegant solutions of dynamic single and multi-objective optimization problems, where the nature of the functional landscape changes with time. Inter-agent communication, search dynamics and the chaotic dynamical characteristics of certain simulated swarms have been investigated both analytically and experimentally to gain better insight into the coordinated swarm control observed in nature. Some of the devised optimization algorithms have also been applied to solve some challenging signal estimation problems. Many real world optimization scenarios demand identification of all possible optimal solutions on the run. We have developed very efficient niching algorithms in the framework of the Differential Evolution algorithm. We are using the neighbourhood based mutation models to avoid basin to basin transfer in the evolutionary process. Our algorithm can detect all the local and global optima of several benchmark and (practically) multi-modal functions efficiently.