My work sits at the interface of nonequilibrium statistical physics, optimization, and most recently, self-assembly.

Optimal control in complex, fluctuating systems

At Berkeley, much of my research has been devoted to measures of efficiency and optimality of nanoscale machines. Together with Gavin Crooks, I have been exploring the use and limitations of the thermodynamic length formalism. Thermodynamic geometry is an elegant, mathematical framework that allows nonequilibrium control problems to be converted into a purely geometric problem: The minimum dissipation protocol can be identified with the shortest path (or minimizing geodesic) on a Riemannian manifold. However, constructing the metric tensor for that particular manifold is not an easy problem, analytically or computationally.

Nonequilibrium control and self-assembly

While the idea of nonequilibrium control has a long history for cyclic processes, nonequilibrium protocols have not been widely used to study self-assembly. Self-assembly balances kinetics and thermodynamics, and therefore is inherently nonequilibrium. I have used ideas from my work on minimum dissipation control to identify perturbations that increase the fraction of successful assembly events. My goal is to, more broadly, develop systematic ways of analyzing the dynamical fluctuations that lead to particular self-assembly outcomes.

Large deviation theory in nonequilibrium statistical physics

The tools of large deviation theory have become an important part of nonequilibrium statistical mechanics. When a system is driven out of equilibrium, the important measurable quantities often depend not on the state, but an entire trajectory of the system evolving in time. I have investigated fluctuations in work, heat, and efficiency in models of stochastic engines using techniques of large deviation theory. Recently, I have been studying analogies between stochastic pumps, which are periodically driven, and nonequilibrium steady states by comparing their asymptotic fluctuations.