Résumé
Computer simulations of fluids in thermal equilibrium can yield information on transport coefficients such as self—diffusion and interdiffusion coefficients, viscosity, and thermal conductivity. While the estimation of self—diffusion coefficients from the mean square displacements of the respective particles is rather straightforward, the estimation of other transport coefficients is less straightforward, and can be based on either an analysis of time correlation functions of the appropriate collective variables, or on nonequilibrium techniques where the linear response to appropriate perturbations is measured.
To concretize ideas, we first discuss a very simple example, interdiffusion in a binary (AB) noninteracting lattice gas, based on Monte Carlo simulation of diffusion in the framework of the vacancy mechanism. Measuring fluxes induced by chem?ical potential gradients, Onsager coefficients are deduced, while the interdiffusion coefficient is measured from an instationary Nonequilibrium Monte Carlo (NEMC) method, studying the decay of concentration waves. Then Nonequilibrium Molecular Dynamics (NEMD) methods are described, where fluid flow between plates is used to extract both viscosity and thermal conductivity. As an alternative method, the use of the Kubo formula for the viscosity and the analysis of time—dependent specific heats (to estimate the thermal conductivity) are mentioned. Application examples range from molten SiO2 to polymer melts.