Nonequilibrium Statistical Physics: The Principle of Maximum Caliber

There has been an aspiration to find a principle of nonequilibrium statistical physics that is as general and generative as the principle of Maximum Entropy and the Second Law of Thermodynamics have been for equilibria. We want it to guide the making of microscopic models that apply far from equilibrium, apply to nonlinear force-flow situations, don’t require local detailed balance assumptions, and don’t require the concept of a heat bath, so as to treat other types of flows like traffic or goods on networks. I will describe extensive work of ours and others on the principle of Maximum Caliber, a term coined by ET Jaynes and put onto a solid foundation by Shore and Johnson. The critical insight is that the entropy to be maximized is over paths, not over microstates. I’ll describe some historical context and implications.