Source-linked AI summary
Identifying structural flow defects in disordered solids using machine learning methods
Ekin D. Cubuk, Samuel S. Schoenholz, Jennifer M. Rieser, Brad D. Malone, Joerg Rottler, Douglas J. Durian, Efthimios Kaxiras, Andrea J. Liu
TL;DR
The paper asks whether flow-defect particles in disordered solids can be identified from local structure, addressing limitations of prior structural and vibrational-mode approaches. It applies machine learning to granular and Lennard-Jones systems, finding robust identification across temperatures and dimensions while revealing distinguishing radial and bond-angle features.
Problem
Flow defects in disordered solids are hypothesized to host localized rearrangements, but prior structural analyses have been insufficient to identify them a priori and vibrational-mode methods are not broadly applicable.
Method
The paper represents each particle’s local environment with radial-density and bond-orientation structure functions, then trains a support vector machine to classify particles as soft or hard from rearrangement labels.
Results
The method identifies rearrangement-prone particles in granular pillars and two- and three-dimensional Lennard-Jones glasses; soft particles capture 80%, 73%, and 72% of rearrangements in the three systems, respectively.
Takeaways & Limitations
Local structural geometry contains subtle features associated with heterogeneous dynamics, and machine learning can identify flow-defect populations directly from experimental or simulated snapshots.
Takeaways & Limitations
The method identifies populations likely to rearrange over short timescales, but cannot predict which specific particles will rearrange later.
Abstract
from arXiv · showhide
We use machine learning methods on local structure to identify flow defects - or regions susceptible to rearrangement - in jammed and glassy systems. We apply this method successfully to two disparate systems: a two dimensional experimental realization of a granular pillar under compression, and a Lennard-Jones glass in both two and three dimensions above and below its glass transition temperature. We also identify characteristics of flow defects that differentiate them from the rest of the sample. Our results show it is possible to discern subtle structural features responsible for heterogeneous dynamics observed across a broad range of disordered materials.