Simulated Annealing and It’s Application in Molecular Structure Optimizations

Abstract

The main purpose of the article is to describe how simulated annealing concept can be applied in molecular structure optimizations. Simulated annealing is a metaheuristic optimization method that has applications in a variety of fields. This article introduces the reader to the concept of simulated annealing and what characteristics of simulated annealing differentiates it from other similar optimization methods like hill climbing. Individual steps of simulated annealing algorithm which include generation of successive position vectors, objective function evaluation and comparison, criteria for accepting new transitions, cooling schedule, and convergence criteria have been discussed in greater detail. Special focus has been placed on how each step of the algorithm is implemented from the point of view of a molecular structure optimization. This includes use of Monte Carlo methods and molecular mechanics for generation of successive position vectors, use of potential energy functions as objective functions and the use of convergence criteria for simulated annealing from a molecular simulation perspective. Different cooling schedules that are used in simulated annealing have also been discussed. A brief account on advantages and disadvantages of simulated annealing has also been provided at the end.