Driven particle in a one dimensional periodic potential with feedback control: efficiency and power optimization (2312.16962v1)

Abstract: A Brownian particle moving in a staircase-like potential with feedback control offers a way to implement Maxwell's demon. An experimental demonstration of such a system using sinusoidal periodic potential carried out by Toyabe et al. has shown that information about the particle's position can be converted to useful work. In this paper, we carry out a numerical study of a similar system using Brownian dynamics simulation. A Brownian particle moving in a periodic potential under the action of a constant driving force is made to move against the drive by measuring the position of the particle and effecting feedback control by altering potential. The work is extracted during the potential change and from the movement of the particle against the external drive. These work extractions come at the cost of information gathered during the measurement. Efficiency and work extracted per cycle of this information engine are optimized by varying control parameters as well as feedback protocols. Both these quantities are found to crucially depend on the amplitude of the periodic potential as well as the width of the region over which the particle is searched for during the measurement phase. For the case when potential flip ($i.e.$, changing the phase of the potential by $180$ degrees) is used as the feedback mechanism, we argue that the square potential offers a more efficient information-to-work conversion. The control over the numerical parameters and averaging over large number of trial runs allow one to study the non-equilibrium work relations with feedback for this process with precision. It is seen that the generalized integral fluctuation theorem for error free measurements holds to within the accuracy of the simulation.