A cage model of liquids supported by molecular-dynamics simulations .1. The cage variables

Stochastic cage models require a choice for the cage variables suitable to describe the restoring forces generated by the solvent on the solute. A set of cage variables is introduced from the parametrization of the cage potential which is defined as the solute-solvent interaction energy considered as a function of the solute position for a fixed solvent configuration. This is an operative definition of cage variables that allows their identification at each time step of molecular dynamics simulations. Therefore, quantitative information about the equilibrium properties and the dynamics of cage variables can be extracted from molecular dynamics simulations. This procedure is applied to liquid argon near the triple point, in order to recognize the different processes contributing to the cage diffusion. The equilibrium distribution and the characteristic correlation times are derived as ingredients for the stochastic cage model developed in part II of the work.