Building a "trap model" of glassy dynamics from a local structural predictor of rearrangements

TL;DR

This paper develops a trap model for glassy dynamics using a machine learning-derived local structural variable called softness, successfully capturing key behaviors but also revealing limitations due to neglected correlations.

Contribution

It introduces a novel trap model based on softness, linking local structure to dynamics in supercooled liquids, and highlights the importance of correlations in such models.

Findings

Model reproduces qualitative features of softness.

Predicts dependence of fragility on density.

Fails to account for correlations in softness dynamics.

Abstract

Here we introduce a variation of the trap model of glasses based on softness, a local structural variable identified by machine learning, in supercooled liquids. Softness is a particle-based quantity that reflects the local structural environment of a particle and characterizes the energy barrier for the particle to rearrange. As in the trap model, we treat each particle's softness, and hence energy barrier, as evolving independently. We show that such a model reproduces many qualitative features of softness, and therefore makes qualitatively reasonable predictions of behaviors such as the dependence of fragility on density in a model supercooled liquid. We also show failures of this simple model, indicating features of the dynamics of softness that may only be explained by correlations.