A new method is presented for the computation of the lowest energy configurations of atomic clusters. It is based on recently developed set oriented numerical algorithms for the global optimization of nonlinear functions. Its underlying idea is to combine multilevel subdivision techniques for the computation of fixed points of dynamical systems with well-known branch and bound methods. We describe how this method can be used to find global minima of silicon nanoclusters in the self-consistent charge tight-binding-density-functional (SCC-DFTB) energy surface. Due to the insufficient experimental evidence of structures of silicon clusters, local minima that are near to the global minimum, are also important. (C) 2003 Elsevier B.V. All rights reserved.