Single energy measurement Integral Fluctuation theorem and non-projective measurements

We study a Jarzysnki type equality for work in systems that are monitored using non-projective unsharp measurements. The information acquired by the observer from the outcome $f$ of an energy measurement, and the subsequent conditioned normalized state $\hat ρ(t,f)$ evolved up to a final time $t$ are used to define work, as the difference between the final expectation value of the energy and the result $f$ of the measurement. The Jarzynski equality obtained depends on the coherences that the state develops during the process, the characteristics of the meter used to measure the energy, and the noise it induces into the system. We analyze those contributions in some detail to unveil their role. We show that in very particular cases, but not in general, the effect of such noise gives a factor multiplying the result that would be obtained if projective measurements were used instead of non-projective ones. The unsharp character of the measurements used to monitor the energy of the system, which defines the resolution of the meter, leads to different scenarios of interest. In particular, if the distance between neighboring elements in the energy spectrum is much larger than the resolution of the meter, then a similar result to the projective measurement case is obtained, up to a multiplicative factor that depends on the meter. A more subtle situation arises in the opposite case in which measurements may be non-informative, i.e. they may not contribute to update the information about the system. In this case, a correction to the relation obtained in the non-overlapping case appears. We analyze the conditions in which such a correction becomes negligible. We also study the coherences, in terms of the relative entropy of coherence developed by the evolved post-measurement state. We illustrate the results by analyzing a two-level system monitored by a simple meter.

💡 Research Summary

The paper investigates how the Jarzynski equality, a cornerstone of nonequilibrium statistical mechanics, is modified when work is defined through a single, non‑projective (unsharp) energy measurement. Traditional approaches use two projective energy measurements, which collapse the system onto energy eigenstates and erase any quantum coherences generated during the subsequent unitary dynamics. This “two‑point” scheme therefore neglects genuinely quantum features of work fluctuations.

To overcome this limitation, the authors adopt a “single‑measurement” protocol. At an initial time (t_0) a continuous‑outcome measurement of the system Hamiltonian (\hat H(t_0)) is performed. The measurement is described by a set of positive operators (\hat G^{1/2}(f|\hat H)) that depend on a real outcome (f). The conditional probability density (G(f|a)) (centered at the eigenvalue (a) with width (\sigma)) models the finite resolution of the meter. After obtaining outcome (f) the post‑measurement state is \