Faster and more precise free-energy estimation using noise

Free energy landscapes underpin our understanding of a wide range of physical and biological processes—from chemical reaction rates and protein folding to the design of responsive materials. Recent developments in stochastic thermodynamics have opened the possibility of estimating equilibrium free energies using nonequilibrium measurements, providing powerful new tools for experiments where equilibrium sampling is difficult or impossible. However, the practical utility of these methods is often limited by slow convergence: conventional algorithms require very large data sets, making precise estimation challenging in many experimental settings. In this work, based on [1] we explore a counterintuitive strategy for overcoming this bottleneck. By simultaneously rescaling the potential and introducing controlled noise—effectively raising the system's temperature—we accelerate relaxation and exponentially improve the precision of free-energy estimates. Our results suggest a new framework for designing stochastic thermodynamic measurements that harness, rather than suppress, fluctuations in experimental studies across soft matter and biophysics, where noise is ubiquitous and often unavoidable. From single-molecule manipulation experiments to the modeling of active and driven systems, and even the development of thermodynamic computers, our work provides a pathway to more practical and widely applicable techniques for system speed-up and free-energy estimation.