Molecular potential many-body expansion method

We are still in the situation where it is not possible to cover a polyatomic surface with ab initio points of sufficient accuracy that inaccuracies can be ignored the exceptions are sufficiently rare to prove the rule. For this reason potential functions which are based in part on empirical data will always be superior to those that are fully ab initio. However, it would be exceptional to derive a potential solely from empirical data. Spectroscopic and thermodynamic data generally only give information about minima on the surface often only the lowest minimum. Kinetic data generally only indicate barrier heights on reaction paths. Molecular beam scattering and equilibrium or transport gas phase data may provide sensitive tests of potentials but we cannot get the potential directly from the data. It is from the fusion of empirical and ab initio data that one obtains the best potential functions. This is, I believe, the reason for the success of LEPS fimctions [2], the DIM method [10] and the many-body expansion with empirical one and two-body terms [3]. 

An analogy may be drawn between the phase behavior of weakly attractive monodisperse dispersions and that of conventional molecular systems provided coalescence and Ostwald ripening do not occur. The similarity arises from the common form of the pair potential, whose dominant feature in both cases is the presence of a shallow minimum. The equilibrium statistical mechanics of such systems have been extensively explored. As previously explained, the primary difficulty in predicting equilibrium phase behavior lies in the many-body interactions intrinsic to any condensed phase. Fortunately, the synthesis of several methods (integral equation approaches, perturbation theories, virial expansions, and computer simulations) now provides accurate predictions of thermodynamic properties and phase behavior of dense molecular fluids or colloidal fluids [1]. 

---