Speaker
Patrick Potts
Description
Patrick Potts
Certifying the non-classicality of fluctuations
Fluctuations have an important role in quantum thermodynamics. For instance, the output of a quantum thermal machine will in general be a fluctuating quantity. Due to the unavoidable measurement back-action in quantum mechanics, a measurement of multiple observables can in general not be described by a probability distribution that does not include the measurement apparatus explicitly. An example which has received considerable attention is provided by work fluctuations. In a classical scenario, a measurement might also act back on the measured system. However, this is in principle avoidable. Nevertheless, it is non-trivial to infer from measured data if the underlying observables can be described by a positive probability distribution or not. In this talk, I will show how one can rule out a description using only positive probability distributions (i.e., a classical description) by making a few natural assumptions on the measurement apparatus. This is achieved by obtaining an experimentally accessible inequality. In quantum theory, this inequality can be violated if and only if the Keldysh quasi-probabiliy distribution, a distribution that has been used before to describe work fluctuations, becomes negative. Importantly, all assumptions on the detectors can be fulfilled in quantum theory, allowing for the conclusion that the measured data cannot be obtained from any classical system.
Certifying the non-classicality of fluctuations
Fluctuations have an important role in quantum thermodynamics. For instance, the output of a quantum thermal machine will in general be a fluctuating quantity. Due to the unavoidable measurement back-action in quantum mechanics, a measurement of multiple observables can in general not be described by a probability distribution that does not include the measurement apparatus explicitly. An example which has received considerable attention is provided by work fluctuations. In a classical scenario, a measurement might also act back on the measured system. However, this is in principle avoidable. Nevertheless, it is non-trivial to infer from measured data if the underlying observables can be described by a positive probability distribution or not. In this talk, I will show how one can rule out a description using only positive probability distributions (i.e., a classical description) by making a few natural assumptions on the measurement apparatus. This is achieved by obtaining an experimentally accessible inequality. In quantum theory, this inequality can be violated if and only if the Keldysh quasi-probabiliy distribution, a distribution that has been used before to describe work fluctuations, becomes negative. Importantly, all assumptions on the detectors can be fulfilled in quantum theory, allowing for the conclusion that the measured data cannot be obtained from any classical system.
Primary author
Prof.
Patrick Potts
