Hyperaccurate bounds in discrete-state Markovian systems

Generalized empirical currents represent a vast class of thermodynamic observables of mesoscopic systems. Their fluctuations satisfy the thermodynamic uncertainty relations (TURs), as they can be bounded by the average entropy production. Here, we derive a general closed expression for the hyperaccurate current in discrete-state Markovian systems, i.e. the one with the least fluctuations, for both discrete- and continuous-time evolution. We show that its associated hyperaccurate bound is generally much tighter than the one given by the TURs, and might be crucial to providing a reliable estimation of the average entropy production. We also show that one-loop systems (rings) exhibit a hyperaccurate current only for finite times, highlighting the importance of short-time observations. Additionally, we derive two novel bounds for the efficiency of work-to-work converters, solely as a function of either the input or the output power. Finally, our theoretical results are employed to analyze a six-state model network for kinesin, and a chemical system in a thermal gradient exhibiting a dissipation-driven selection of states.