Quantum mean-square predictors and thermodynamics

Thermodynamic quantities, such as heat and work, are not functions of state, but rather of the process undergone by a physical system. Assessing them can therefore be difficult, since it requires probing the system at least twice. This is even more so when these quantities are to be assessed at the stochastic level. In this paper we show how to obtain optimal estimates of thermodynamic quantities solely from indirect measurement unravelings of an auxiliary system. The method always yields the true average, and the mean-squared error of the prediction is directly proportional to how well the method estimates the variance. As an application, we study energy fluctuations in a driven system, and in an avoided crossing work protocol.