Résumé
Computer simulations give precious insight into the microscopic behavior of disordered and amorphous materials, but their typical time scales are orders of magnitude shorter than the experimentally relevant ones. In particular, simulations of supercooled liquids cover at most 4-5 decades of viscous slowing down, which falls far short of the 13 decades commonly accessible in experimental studies. We close this enormous gap for a class of realistic models of liquids, which we successfully equilibrate beyond laboratory time scales by means of the swap Monte Carlo algorithm. We show that combined optimization of selected features of the interaction potential, such as particle softness, polydispersity and non-additivity, leads to computer models with excellent glass-forming ability. For such models, we achieve over 10 orders of magnitude speedup in equilibration time scale. This numerical advance allows us to address some outstanding questions concerning glass formation, such as the role of local structure and the relevance of an entropy crisis, in a dynamical range that remains inaccessible in experiments. Our results support the view that non-trivial static correlations continue to build up steadily in supercooled liquids even below the laboratory glass temperature.