Â鶹ÊÓƵ

Synopsis

Microscopic systems experience random thermal forces from their surroundings, while nonequilibrium processes, such as energy-consuming activity within living cells, can generate additional fluctuations.  Can we deliberately introduce such noise into nonliving microscopic systems and harness it to improve their performance?  In this talk, I will describe how, by adding noise forces, we realized an effective thermal bath above 3000 K without physically heating the surrounding water.  Using feedback-controlled optical tweezers, we explore two applications of these engineered fluctuations.  First, by simultaneously scaling the deterministic forces and proportionally increasing the noise, we accelerate Langevin dynamics without altering the system’s physical properties.  This offers a route to faster thermodynamic computing, which aims to perform computations using little energy, and could complement other optimization methods across a broad range of stochastic thermodynamic operations.  Second, we investigate a Szilard information engine in which geometric constraints make only a fraction of measurements useful, limiting conventional deterministic feedback strategies.  Using a three-state memory, we demonstrate how a probabilistic policy, combined an effectively hotter bath, optimally uses information and outperforms deterministic alternatives.