Abstract: By combining the genetic algorithm, fast simulated annealing and conjugated gradient algorithm, a novel global optimization method, termed as fast hybrid global optimized algorithm (FHGOA), was proposed and applied to global geometry optimization of water clusters (H2O)n. In the case of TIP3P, we found lower energy water clusters compared with those reported in literature and the computational efficiency by using FHGOA has been greatly improved. In the case of TIPS2, the global optimized geometries of water clusters (H2O)n in the size range of 2 to 27 were found. The water clusters are all-surface structures when n < 17,and the cage structure with one interior water molecule and two interior water molecules appear when n=17 and n=25, respectively. This transformation from all-surface structure to the cage structure with one interior water molecule and then to the cage structure with two interior water molecules is in agreement with the case of TTM2-F.The results of this work show that TIPS2 has the ability of modeling large water clusters and is a suitable potential function for MD and MC simulation.

Key words: Fast hybrid global optimization algorithm, Water clusters, Geometry optimization