Théorique, numérique
Despite decades of intense study, the staggering increase in relaxation time in liquids supercooled below their freezing point (the glass transition), there is still no consensus around the underlying mechanism. In this way, the glass transition is a scientific revolution in the sense of Thomas Kuhn, in that mutually incompatible theories give equally good descriptions of the available data. Computer simulation (and experiments with colloids) promise the ability to resolve this challenge, due to the particle-resolved data which enables analysis not otherwise possible, capable of discriminating between theories. Yet, accessible timescales with both approaches limit the data obtained to a dynamical regime where relaxation is quite well—understood, the so-called mode-coupling regime. Recently this has begun to change. In particular, supercooling below the mode-coupling crossover has opened the possibility to probe the predictions of dynamic frustration and thermodynamic descriptions across many decades in time [5,6]. This project builds on such recent work and uses computer simulation to address the following outstanding questions: what is the nature of the excitations of dynamic facilitation at very deep supercooling? How do these couple to co-operatively re-arranging regions at larger length- and timescales? And how are excitations related to the dynamical phase transition that underlies the glass transition in the facilitation picture?