LAUSR

We investigate the stochastic dynamics of a thermal machine realized by a fast-driven Otto cycle. By employing a stochastic approach, we find that system coherences strongly affect fluctuations depending on the thermodynamic current. Specifically, we observe an increment in the system instabilities when considering the heat exchanged with the cold bath. On the contrary, the cycle precision improves when the system couples with the hot bath, where thermodynamic fluctuations reduce below the classical Thermodynamic Uncertainty Relation bound. Violation of the classical bound holds even when a dephasing source couples with the system. We also find that coherence suppression not only restores the cycle cooling but also enhances the convergence of fluctuation relations by increasing the entropy production of the reversed process. An additional analysis unveiled that the stochastic sampling required to ensure good statistics increases for the cooling cycle while downsizes for the other protocols. Despite the simplicity of our model, our results provide further insight into thermodynamic relations at the stochastic level.