Thermodynamic control of non-equilibrium systems

We study the thermodynamic cost associated with driving systems between different non-equilibrium steady states. In particular, we combine a linear-response framework for non-equilibrium Markov systems with Lagrangian techniques to minimize the dissipation associated with driving processes. We then apply our framework to a simple toy model. Our results show several remarkable properties for the optimal protocol, such as diverging parameters and finite entropy production in the slow-driving limit.