Variational inverse design for nonequilibrium colloidal self-assembly

Nonequilibrium conditions lead to novel effective interactions and dynamics during colloidal self-assembly, resulting in new structural and fluxional phases of functional materials. Automated inverse design algorithms for self-assembly protocols out-of-equilibrium are constrained due to the corresponding configurational probability distributions being non-Boltzmann. We have developed an optimization algorithm to discover nonequilibrium inverse design principles based on a trajectory ensemble formalism. The optimization procedure uses explicit stochastic gradients within a nonequilibrium steady-state. We use this algorithm to discover optimal control forces in molecular dynamics simulations of DNA-labeled colloids, for the self-assembly of various target structures and dynamical states.