Structural Optimization of Atomic Clusters

Abstract

This chapter exclusively addresses the algorithms employed to perform geometry optimization of clusters. These algorithms can be broadly classified into two groups: gradients-based algorithms and gradient-free algorithms. Gradient-based algorithms use the gradient of potential energy functions to give local minima. On the contrary, gradient-free algorithms are inspired by natural processes, which exploit some mathematical models, which lead to global minimum. Although there are a variety of gradient-free algorithms, some of the most popular ones include genetic algorithm, particle swarm, simulated annealing, etc. The strengths and weaknesses of all these algorithms have been also discussed.

Keywords: Artificial bee colony, Gradient, Genetic algorithm, Geometry optimization, Global minimum, Garticle swarm, Local minima, Neural network, Simulated annealing.