Identifying Structural Flow Defects in Disordered Solids Using Machine-Learning Methods

Cubuk, Ekin D.; Schoenholz, Samuel S.; Rieser, Jennifer M.; Malone, Brad D.; Rottler, Jörg; Durian, Douglas J.; Kaxiras, Efthimios; Liu, Andrea J.

doi:10.1103/physrevlett.114.108001

Cited by 391 publications

(389 citation statements)

References 41 publications

Supporting

Mentioning

373

Contrasting

Unclassified

Order By: Relevance

“…There is a long tradition of associating glassy behavior with geometrical features associated with the arrangements of particles around a given particle, see e.g. [13][14][15][16][17], and our work on disks in a narrow channel is entirely consistent with these ideas. In our earlier work [6,7] a length scale ξ associated with zigzag order has been determined from the decay with s of the correlation function y i y i+s ∼ (−1) s exp(−s/ξ) .…”

Section: Fig 1: (Color Online)supporting

confidence: 74%

“…Our system is sufficiently simple that we can quantitatively relate its dynamical features to structural features. The threedimensional problem is much richer and success along these lines is probably only just starting [17].…”

Section: Discussionmentioning

confidence: 99%

“…2. In three dimensional systems the ordering associated with glassy behavior is complicated [17]. Furthermore, the ordering associated with glassy behavior is not apparent in In the top and bottom diagrams, the two blue-shaded disks are a defect in the zigzag arrangement of disks.…”

Section: Fig 1: (Color Online)mentioning

confidence: 99%

“…This marked difference with the behavior in three dimensions where the structure factor hardly alters near the glass transition indicates that glass behavior in three dimensions must, as suggested in Refs. [13][14][15][16][17], involve higher order correlations which have little impact on the structure factor, unlike the simple zigzag orientational order of the narrow channel system. In Sec.…”

Section: Fig 1: (Color Online)mentioning

confidence: 99%

See 3 more Smart Citations

Glasslike behavior of a hard-disk fluid confined to a narrow channel

2016

View full text Add to dashboard Cite

Disks moving in a narrow channel have many features in common with the glassy behavior of hard spheres in three dimensions. In this paper we study the caging behavior of the disks which sets in at characteristic packing fraction φ d . Four-point overlap functions similar to those studied when investigating dynamical heterogeneities have been determined from event driven molecular dynamics simulations and the time dependent dynamical length scale has been extracted from them. The dynamical length scale increases with time and, on the equilibration time scale, it is proportional to the static length scale associated with the zigzag ordering in the system, which grows rapidly above φ d . The structural features responsible for the onset of caging and the glassy behavior are easy to identify as they show up in the structure factor, which we have determined exactly from the transfer matrix approach.

show abstract

Section: Fig 1: (Color Online)supporting

confidence: 74%

Section: Discussionmentioning

confidence: 99%

Section: Fig 1: (Color Online)mentioning

confidence: 99%

Section: Fig 1: (Color Online)mentioning

confidence: 99%

See 2 more Smart Citations

Glasslike behavior of a hard-disk fluid confined to a narrow channel

2016

View full text Add to dashboard Cite

show abstract

“…Recent work by Cubuk et al [20] on the flow of jammed and glassy systems under stress has shown that regions that are susceptible to rearrangement can be discovered by machine-learning methods that combine information derived from several features of the local structure, such as the radial distribution of particles and the bond angles. Since our narrow-channel system with NNN contacts is relatively simple, we are able to identify the collective motions of the disks that lead to the apparent ideal glass behaviors.…”

Section: Introductionmentioning

confidence: 99%

Understanding the ideal glass transition: Lessons from an equilibrium study of hard disks in a channel

Godfrey

Moore

2015

Phys. Rev. E

View full text Add to dashboard Cite

We use an exact transfer-matrix approach to compute the equilibrium properties of a system of hard disks of diameter σ confined to a two-dimensional channel of width 1.95 σ at constant longitudinal applied force. At this channel width, which is sufficient for next-nearest-neighbor disks to interact, the system is known to have a great many jammed states. Our calculations show that the longitudinal force (pressure) extrapolates to infinity at a well-defined packing fraction φK that is less than the maximum possible φmax, the latter corresponding to a buckled crystal. In this quasi-onedimensional problem there is no question of there being any real divergence of the pressure at φK . We give arguments that this avoided phase transition is a structural feature -the remnant in our narrow channel system of the hexatic to crystal transition -but that it has the phenomenology of the (avoided) ideal glass transition. We identify a length scaleξ3 as our equivalent of the penetration length for amorphous order: In the channel system, it reaches a maximum value of around 15 σ at φK , which is larger than the penetration lengths that have been reported for three dimensional systems. It is argued that the α-relaxation time would appear on extrapolation to diverge in a Vogel-Fulcher manner as the packing fraction approaches φK .

show abstract

Machine Learning in Soft Matter: From Simulations to Experiments

Zhang,

Gong,

Jiang

2024

Adv Funct Materials

View full text Add to dashboard Cite

Soft matter with diverse functionalities that are easily designable has fascinated tremendous research interests in the past several decades. Nevertheless, the inherent confluence of time and length scale ubiquitous in soft matter immensely complicates the elucidation of the structure–property relationship and thereby severely impedes the function exploration of soft materials. Recently, the emergent machine learning (ML) techniques open new paradigms in property prediction and molecular design of functional materials, due to their extraordinarily distinguished performance in the aspect of trend identity and pattern extraction from data, and objective optimization by accelerating the guided search in high‐dimensional spaces. This review exclusively focuses on the current state‐of‐the‐art progress in the development of ML techniques applied in the realms of soft matter, ranging from coarse‐grained simulations to theoretical prediction on the structural formation and macroscopic properties, as well as the optimization and algorithm‐aided design in experiments. Finally, an outlook on the challenges and opportunities for this rapidly evolving field is discussed.

show abstract

Identifying Structural Flow Defects in Disordered Solids Using Machine-Learning Methods

Cited by 391 publications

References 41 publications

Glasslike behavior of a hard-disk fluid confined to a narrow channel

Glasslike behavior of a hard-disk fluid confined to a narrow channel

Understanding the ideal glass transition: Lessons from an equilibrium study of hard disks in a channel

Machine Learning in Soft Matter: From Simulations to Experiments

Contact Info

Product

Resources

About