What causes the super-Arrhenius slowdown of supercooled liquids?

In molecular glass formers the structural relaxation time $\tau_\alpha(T)$ grows from about $10^{-12}\ \mathrm{s}$ in the high-temperature liquid to about $10^{2}\ \mathrm{s}$ at $T_g$, with an effective activation energy that increases on cooling; model liquids (Kob-Andersen Lennard-Jones mixture, polydisperse soft spheres) reproduce the onset of this growth. Competing explanations include random first-order transition theory (entropy-driven cooperative rearrangements), dynamic facilitation (kinetically constrained localized excitations) and elastic or frustration-based models. An answer is a theory that predicts $\tau_\alpha(T)$ and the associated growing length scales from the interaction potential and is singled out against the alternatives by equilibrium simulations near and below the experimental $T_g$. This is the root question of the topic; its working content is split into the sharper problems linked to it.

A theory whose quantitative predictions for $\tau_{\alpha}(T)$, $\xi(T)$ and their relation are confirmed, and those of rival theories refuted, in swap-Monte-Carlo-equilibrated simulations below the experimental $T_{g}$.