|
Benchmark control problems in nonequilibrium statistical mechanics
S. Whitelam, C. Casert, M. Engel, I. Tamblyn Preprint (2025) Optimizing how we control physical systems that fluctuate and operate far from equilibrium is a fundamental challenge in statistical mechanics, with applications ranging from molecular machines to material self-assembly. In this work, we introduce NESTbench25, a collection of five benchmark problems designed to test and compare methods for finding optimal time-dependent control protocols in stochastic nonequilibrium systems. Each benchmark consists of a well-defined stochastic model with a clear optimization objective, implemented in both C++ and Python, lightweight enough to run on a standard laptop yet challenging enough to stress-test modern optimization algorithms. By providing the community with standardized tests, we aim to accelerate the development of better control strategies for nonequilibrium systems. |
