Information theory applied to time correlation functions in fluids

Abstract

Information theory is used to predict the reorientational time correlation functions for a linear molecule in a fluid given only the moment of inertia and the mean squared torque. Comparison with ‘exact’ computer simulation results for a variety of fluids show that the method is remarkably successful—even when the torque is small. It is just as successful, for instance, as a much more elaborate memory-function perturbation theory which uses the same information. The method is quite general and can be usefully applied to any time correlation function.