Fragility in Glassy Liquids: A Structural Approach Based on Machine Learning

The rapid growth of viscosity or relaxation time upon supercooling is universal hallmark of glassy liquids. The temperature dependence of the viscosity, however, is quite non-universal for glassy liquids and is characterized by the system's ``fragility,'' with liquids with a nearly Arrhenius temperature-dependent viscosities referred to as strong liquids and those with strongly super-Arrhenius behavior referred to as fragile liquids. What makes some liquids strong and others fragile is still not well understood. Here we explore this question in a family of glassy liquids that range from extremely strong to extremely fragile, using ``softness,'' a structural variable identified by machine learning to be highly correlated with dynamical rearrangements. We use a support vector machine to identify softness as a linear combination of structural quantities, and show that the same linear combination is successful in predicting rearrangements across the entire family of glassy liquids. We find that fragility is reflected in the softness-dependence of energy barriers.