Princeton University PCTS Special Physics Talk

Abstract: Thermal quantum memories can preserve quantum information for exponentially long times simply by cooling the system. Such memories exist in three spatial dimensions, but the known construction relies on an iterative procedure, motivating the search for simpler models. We present a no-go result suggesting that three-dimensional thermal quantum memories must break translation invariance. More generally, we introduce a criterion for fast convergence to the Gibbs state under local thermalization dynamics, resolving open problems in models such as the classical Potts model. Our approach connects the dynamical problem of thermalization to the static characterization of thermal phases.