Does the embedded atom model have predictive power?

Potassium, rubidium, aluminum, iron, nickel, and tin embedded atom models (EAMs) have been used as examples to ascertain how well the properties of a metal are described by EAM potentials calculated from the shape of shock adiabats and/or static compression data (from a function of cold pressure). Verification of the EAM potential implies an evaluation of its predictive power and an analysis of the agreement with experiment both at 0 or 298 K and under shock compression. To obtain consistent results, all contributions of collectivized electrons to energy and pressure need to be taken into consideration, especially in transition metals. Taking account of or ignoring electron contributions has little effect on the calculated melting lines of the models, self-diffusion coefficients, and viscosity. The shape of the melting line is sensitive to the behavior of the repulsive branch of the pair contribution to the EAM potential at small distances. (reviews of topical problems)