In this paper we explore the possibility of using stochastic optimizers, namely Simulated Annealing (SA) in locating critical points (global minima, local minima, and first order saddle points) in argon noble gas clusters. The atomic interaction potential is the Lennard-Jones potential. We also try to see if a continuous transformation in geometry during the search process can lead to a realization of a kind of minimum energy path (MEP) for transformation from one minimum geometry to another through a transition state (first order saddle point). We try our recipe for three sizes of argon cluster, namely (Ar)8, (Ar)10 and (Ar)12. © 2010 Elsevier B.V. All rights reserved.