A statistical method for identifying transition states in high dimensional problems

Publisher Website
Google Scholar
Abstract
A statistical method based on Monte Carlo sampling of paths associated with Markovian stochastic processes is proposed for the location of transition states in molecularly complex reactions. It is discussed in what respects the proposed method would identify transition state regions, and an algorithm for Metropolis sampling of Metropolis paths (based on the Metropolis random walk process) is devised. Physical aspects of this new algorithm are examined, and the relation of this method to simulated annealing algorithms for physical design of circuits is noted.