Subscribe to Events

Download as iCal file

Mathematical Physics Seminar

What can Maxwell’s demon do? — An experimental and theoretical study of the factors that limit the performance of information-fueled engines

John Bechhoefer - Simon Fraser University

Location:  zoom
Date & time: Wednesday, 11 August 2021 at 10:45AM - 11:45AM


Burnaby, BC

Information engines exploit the statistical nature of the second law of thermodynamics to generate work by information-to-energy conversion. Experimental tests of these engines require high-precision measurements and fast feedback control. We designed a simple information engine using optical tweezers and feedback to raise a micron-sized trapped bead — a heavy mass — against gravity, without doing any external work. We find that the rate of work extraction (“power”) is maximized for larger beads and the directed velocity (“speed”) for smaller beads. Optimizing the experimental parameters, we record a maximum power of 1000 kT/s and a maximum speed of 190 ?m/s, numbers that are comparable to similar-sized biological systems such as bacteria. We also design a feedback algorithm for time-dependent control of the engine output. Finally, adding an optimally configured memory — a type of Kalman filter — allows the motor to retain a large fraction of its performance, even in situations where measurement noise is comparable to the observable fluctuations (signal). Our results show in detail how the availability of relevant information limits the performance of information-fueled engines.