Representation of Initial Temperature as a Function in Simulated Annealing Approach for Metal Nanoparticle Structures Modeling

Abstract

Very important in the study of the thermodynamic characteristics of nanostructures (melting/crystallization) is the structure of nanoclusters. The properties of the nanoparticles can be predicted if we know the mechanism of the formation and the dynamic of changes in the internal structure. The problem to find the stable structure of nanoparticle is NP-hard and it needs development of special methods coming from Artificial Intelligence to be solved. In this paper we apply Simulated Annealing Method to find approximate solution. The proposed algorithm is designed for metal nanoparticle structures optimization. This problem has an exceptional importance in studying the properties of nanomaterials. The problem is represented as a global optimization problem. The most important algorithm parameter is the temperature. The main focus in this paper is on representation of the initial temperature as a function. Thus the algorithm parameters will be closely related with the input data. The experiments are performed with real data as follows. One set of mono metal clusters is chosen for investigation: Silver (Ag) where the size of clusters for Ag varies from Ag150 (atoms) to Ag3000 (atoms). Several dependencies are derived between the number and configuration of atoms in the cluster on one hand, and temperature representation and stopping rule on the other hand.