REALISTIC MASTER EQUATION MODELING OF RELAXATION ON COMPLETE POTENTIAL-ENERGY SURFACES - PARTITION-FUNCTION MODELS AND EQUILIBRIUM RESULTS

To elucidate the role that potential surface topography plays in shapi ng the evolution of a cluster toward equilibrium, entire sets of kinet ically accessible bound-state configurations and transition states on the model potential energy surfaces of (KCl)(5) and Ar-9 are mapped an d compared. To describe the stochastic dynamics on these surfaces in t erms of transition-state theory, we require adequate approximations of the partition functions of the minima and transition states. In this paper we introduce several partition function models derived from harm onic and anharmonic approximations and compare their predicted equilib rium population distributions with those determined from canonical-ens emble molecular dynamics. We perform this comparison for both (KCl)(5) and Ar-9 in order to evaluate the relative performance of the models for two different types of potential surfaces. For each system, partic ular models are found to give results that agree better with simulatio n than do the results using the simple harmonic approximation. However , no one unparameterized model gives acceptable results for all minima , and the best parameter-free strategies differ for (KCl)(5) and Ar-9. Nevertheless, a one-parameter version of one of the models is shown t o give the best agreement with simulation for both systems. In an acco mpanying paper, the best partition function models are used to constru ct a stochastic master equation which makes predictions of relaxation behavior. These predictions are compared with results from molecular d ynamics. (C) 1998 American Institute of Physics. [S0021 -9606(98)02243 -0].