Optimizing non-ergodic feedback engines

Maxwell's demon is a special case of a feedback controlled system, where information gathered by measurement is utilized by driving a system along a thermodynamic process that depends on the measurement outcome. The demon illustrates that with feedback one can design an engine that performs work by extracting energy from a single thermal bath. Besides the fundamental questions posed by the demon - the probabilistic nature of the Second Law, the relationship between entropy and information, etc. - there are other practical problems related to feedback engines. One of those is the design of optimal engines, protocols that extract the maximum amount of energy given some amount of information. A refinement of the second law to feedback systems establishes a bound to the extracted energy, a bound that is met by optimal feedback engines. It is also known that optimal engines are characterized by time reversibility. As a consequence, the optimal protocol given a measurement is the one that, run in reverse, prepares the system in the post-measurement state (preparation prescription). In this paper we review these results and analyze some specific features of the preparation prescription when applied to non-ergodic systems.