Georgia Institute of Technology College of Sciences

In this talk I will begin with our recent results on non-equilibrium steady states (NESS) of a microscopic heat conduction model, which is a stochastic particle system coupled to unequal heat baths. This stochastic model is derived from a mechanical chain model (Eckmann and Young 2006) by randomizing certain quantities while retaining the other features. We proved various results including the existence and uniqueness of NESS and the exponential rate of mixing. Then I will follow with an energy dependent Kac-type model that is obtained from an improved version of randomization of the “local" dynamics. We rigorously proved that this Kac-type model has a mixing rate $\sim t^{-2}$. In the end, I will show that slow (polynomial) mixing rates appear in a large class of statistical mechanics models.