Global geometry optimizations

The optimization task can be local (the highest or lowest function value in a finite neighborhood) or global (the highest or lowest function value of all). Even if some algorithms may fail to find the local minimum, local optimization problems are in principle solvable. It is rather obvious that global optimization is a much more difficult task than local optimization. Global optimization is conducted in the presence of a large number of local minima, aiming to find the best local minimum amongst them. This may suggest that global optimization problems are in principle solvable if enough time is given. For this reason, in order to obtain a solution to any global optimization problems within a reasonable time, the effort exerted to solve the problem must be scaled with problem size.

Simulated Annealing (SA) and Evolutionary Algorihtms (EA) are the most promising global optimization methods. In our group as well as developing new variants of these methods, we apply them to find the lowest-energy structure of clusters (e.g., silicon and molecular aggregates (e.g., acetylene, acetylene-benzene, benzene, acetylene-furan).