-
The dynamics of machine-learned "softness" in supercooled liquids describe dynamical heterogeneity
Authors:
Sean A. Ridout,
Andrea J. Liu
Abstract:
The dynamics of supercooled liquids slow down and become increasingly heterogeneous as they are cooled. Recently, local structural variables identified using machine learning, such as "softness", have emerged as predictors of local dynamics. Here we construct a model using softness to describe the structural origins of dynamical heterogeneity in supercooled liquids. In our model, the probability o…
▽ More
The dynamics of supercooled liquids slow down and become increasingly heterogeneous as they are cooled. Recently, local structural variables identified using machine learning, such as "softness", have emerged as predictors of local dynamics. Here we construct a model using softness to describe the structural origins of dynamical heterogeneity in supercooled liquids. In our model, the probability of particles to rearrange is determined by their softness, and each rearrangement induces changes in the softness of nearby particles, describing facilitation. We show how to ensure that these changes respect the underlying time-reversal symmetry of the liquid's dynamics. The model reproduces the salient features of dynamical heterogeneity, and demonstrates how long-ranged dynamical correlations can emerge at long time scales from a relatively short softness correlation length.
△ Less
Submitted 9 June, 2024;
originally announced June 2024.
-
Machine learning that predicts well may not learn the correct physical descriptions of glassy systems
Authors:
Arabind Swain,
Sean Alexander Ridout,
Ilya Nemenman
Abstract:
The complexity of glasses makes it challenging to explain their dynamics. Machine Learning (ML) has emerged as a promising pathway for understanding glassy dynamics by linking their structural features to rearrangement dynamics. Support Vector Machine (SVM) was one of the first methods used to detect such correlations. Specifically, a certain output of SVMs trained to predict dynamics from structu…
▽ More
The complexity of glasses makes it challenging to explain their dynamics. Machine Learning (ML) has emerged as a promising pathway for understanding glassy dynamics by linking their structural features to rearrangement dynamics. Support Vector Machine (SVM) was one of the first methods used to detect such correlations. Specifically, a certain output of SVMs trained to predict dynamics from structure, the distance from the separating hyperplane, was interpreted as being linearly related to the activation energy for the rearrangement. By numerical analysis of toy models, we explore under which conditions it is possible to infer the energy barrier to rearrangements from the distance to the separating hyperplane. We observe that such successful inference is possible only under very restricted conditions. Typical tests, such as the apparent Arrhenius dependence of the probability of rearrangement on the inferred energy and the temperature, or high cross-validation accuracy do not guarantee success. We propose practical approaches for measuring the quality of the energy inference and for modifying the inferred model to improve the inference, which should be usable in the context of realistic datasets.
△ Less
Submitted 1 March, 2024;
originally announced March 2024.
-
Building a "trap model" of glassy dynamics from a local structural predictor of rearrangements
Authors:
Sean A. Ridout,
Indrajit Tah,
Andrea J. Liu
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 ener…
▽ More
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.
△ Less
Submitted 15 August, 2023;
originally announced August 2023.
-
Universal scaling function ansatz for finite-temperature jamming
Authors:
Sean A. Ridout,
Andrea J. Liu,
James P. Sethna
Abstract:
We cast a nonzero-temperature analysis of the jamming transition into the framework of a scaling ansatz. We show that four distinct regimes for scaling exponents of thermodynamic derivatives of the free energy such as pressure, bulk and shear moduli, can be consolidated by introducing a universal scaling function with two branches. Both the original analysis and the scaling theory assume that the…
▽ More
We cast a nonzero-temperature analysis of the jamming transition into the framework of a scaling ansatz. We show that four distinct regimes for scaling exponents of thermodynamic derivatives of the free energy such as pressure, bulk and shear moduli, can be consolidated by introducing a universal scaling function with two branches. Both the original analysis and the scaling theory assume that the system always resides in a single basis in the energy landscape. The two branches are separated by a line $T^*(Δφ)$ in the $T-Δφ$ plane, where $Δφ=φ-φ_c^Λ$ is the deviation of the packing fraction from its critical, jamming value, $φ_c^Λ$, for that basin. The branch for $T<T^*(Δφ)$ reduces at $T=0$ to an earlier scaling ansatz that is restricted to $T=0$, $Δφ\ge 0$, while the branch for $T>T^*(Δφ)$ reproduces exponents observed for thermal hard spheres. In contrast to the usual scenario for critical phenomena, the two branches are characterized by different exponents. We suggest that this unusual feature can be resolved by the existence of a dangerous irrelevant variable $u$, which can appear to modify exponents if the leading $u=0$ term is sufficiently small in the regime described by one of the two branches of the scaling function.
△ Less
Submitted 27 March, 2024; v1 submitted 21 April, 2023;
originally announced April 2023.
-
Machine learning-informed structuro-elastoplasticity predicts ductility of disordered solids
Authors:
Hongyi Xiao,
Ge Zhang,
Entao Yang,
Robert J. S. Ivancic,
Sean A. Ridout,
Robert Riggleman,
Douglas J. Durian,
Andrea J. Liu
Abstract:
All solids yield under sufficiently high mechanical loads. Below yield, the mechanical responses of all disordered solids are nearly alike, but above yield every different disordered solid responds in its own way. Brittle systems can shatter without warning, like ordinary window glass, or exhibit strain localization prior to fracture, like metallic or polymeric glasses. Ductile systems, e.g. foams…
▽ More
All solids yield under sufficiently high mechanical loads. Below yield, the mechanical responses of all disordered solids are nearly alike, but above yield every different disordered solid responds in its own way. Brittle systems can shatter without warning, like ordinary window glass, or exhibit strain localization prior to fracture, like metallic or polymeric glasses. Ductile systems, e.g. foams like shaving cream or emulsions like mayonnaise, can flow indefinitely with no strain localization. While there are empirical strategies for tuning the degree of strain localization, there is no framework that explains their effectiveness or limitations. We show that Structuro-Elastoplastic (StEP) models provide microscopic understanding of how strain localization depends on the interplay of structure, plasticity and elasticity.
△ Less
Submitted 22 March, 2023;
originally announced March 2023.
-
Jammed solids with pins: Thresholds, Force networks and Elasticity
Authors:
Andy L. Zhang,
Sean A. Ridout,
Celia Parts,
Aarushi Sachdeva,
Cacey S. Bester,
Katharina Vollmayr-Lee,
Brian C. Utter,
Ted Brzinski,
Amy L. Graves
Abstract:
The role of fixed degrees of freedom in soft/granular matter systems has broad applicability and theoretical interest. Here we address questions of the geometrical role that a scaffolding of fixed particles plays in tuning the threshold volume fraction and force network in the vicinity of jamming. Our 2d simulated system consists of soft particles and fixed "pins", both of which harmonically repel…
▽ More
The role of fixed degrees of freedom in soft/granular matter systems has broad applicability and theoretical interest. Here we address questions of the geometrical role that a scaffolding of fixed particles plays in tuning the threshold volume fraction and force network in the vicinity of jamming. Our 2d simulated system consists of soft particles and fixed "pins", both of which harmonically repel overlaps. On one hand, we find that many of the critical scalings associated with jamming in the absence of pins continue to hold in the presence of even dense pin latices. On the other hand, the presence of pins lowers the jamming threshold, in a universal way at low pin densities and a geometry-dependent manner at high pin densities, producing packings with lower densities and fewer contacts between particles. The onset of strong lattice dependence coincides with the development of bond-orientational order. Furthermore, the presence of pins dramatically modifies the network of forces, with both unusually weak and unusually strong forces becoming more abundant. The spatial organization of this force network depends on pin geometry and is described in detail. Using persistent homology we demonstrate that pins modify the topology of the network. Finally, we observe clear signatures of this develo** bond-orientational order and broad force distribution in the elastic moduli which characterize the linear response of these packings to strain.
△ Less
Submitted 25 August, 2022; v1 submitted 29 May, 2022;
originally announced May 2022.
-
Structuro-elasto-plasticity (StEP) model for plasticity in disordered solids
Authors:
Ge Zhang,
Hongyi Xiao,
Entao Yang,
Robert J. S. Ivancic,
Sean A. Ridout,
Robert A. Riggleman,
Douglas J. Durian,
Andrea J. Liu
Abstract:
Elastoplastic lattice models for the response of solids to deformation typically incorporate structure only implicitly via a local yield strain that is assigned to each site. However, the local yield strain can change in response to a nearby or even distant plastic event in the system. This interplay is key to understanding phenomena such as avalanches in which one plastic event can trigger anothe…
▽ More
Elastoplastic lattice models for the response of solids to deformation typically incorporate structure only implicitly via a local yield strain that is assigned to each site. However, the local yield strain can change in response to a nearby or even distant plastic event in the system. This interplay is key to understanding phenomena such as avalanches in which one plastic event can trigger another, leading to a cascade of events, but typically is neglected in elastoplastic models. To include the interplay one could calculate the local yield strain for a given particulate system and follow its evolution, but this is expensive and requires knowledge of particle interactions, which is often hard to extract from experiments. Instead, we introduce a structural quantity, "softness," obtained using machine learning to correlate with imminent plastic rearrangements. We show that softness also correlates with local yield strain. We incorporate softness to construct a "structuro-elasto-plasticity" model that reproduces particle simulation results quantitatively for several observable quantities, confirming that we capture the influence of the interplay of local structure, plasticity, and elasticity on material response.
△ Less
Submitted 17 May, 2022;
originally announced May 2022.
-
Fragility in Glassy Liquids: A Structural Approach Based on Machine Learning
Authors:
Indrajit Tah,
Sean A. Ridout,
Andrea J. Liu
Abstract:
The rapid rise 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 nearly Arrhenius temperature-dependent relaxation times referred to as strong liquids and those with super-Arrhenius behavior refe…
▽ More
The rapid rise 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 nearly Arrhenius temperature-dependent relaxation times referred to as strong liquids and those with 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 order parameter identified by machine learning to be highly correlated with dynamical rearrangements. We use a support vector machine to identify softness as the same linear combination of structural quantities across the entire family of liquids studied. We then use softness to identify the factors controlling fragility.
△ Less
Submitted 15 May, 2022;
originally announced May 2022.
-
Correlation of plastic events with local structure in jammed packings across spatial dimensions
Authors:
Sean A. Ridout,
Jason W. Rocks,
Andrea J. Liu
Abstract:
In jammed packings, it is usually thought that local structure only plays a significant role in specific regimes. The standard deviation of the relative excess coordination, $σ_Z/ Z_\mathrm{c}$, decays like $1/\sqrt{d}$, so that local structure should play no role in high spatial dimensions. Furthermore, in any fixed dimension $d \geq 2$, there are diverging length scales as the pressure vanishes…
▽ More
In jammed packings, it is usually thought that local structure only plays a significant role in specific regimes. The standard deviation of the relative excess coordination, $σ_Z/ Z_\mathrm{c}$, decays like $1/\sqrt{d}$, so that local structure should play no role in high spatial dimensions. Furthermore, in any fixed dimension $d \geq 2$, there are diverging length scales as the pressure vanishes approaching the unjamming transition, again suggesting that local structure should not be sufficient to describe response. Here we challenge the assumption that local structure does not matter in these cases. In simulations of jammed packings under athermal, quasistatic shear, we use machine learning to identify a local structural variable, softness, that correlates with rearrangements in dimensions $d=2$ to $d=5$. We find that softness - and even just the coordination number $Z$ - are quite predictive of rearrangements over a wide range of pressures, all the way down to unjamming, in all $d$ studied. This result provides direct evidence that local structure can play a role in higher spatial dimensions.
△ Less
Submitted 25 November, 2020;
originally announced November 2020.
-
Learning-based approach to plasticity in athermal sheared amorphous packings: Improving softness
Authors:
Jason W. Rocks,
Sean A. Ridout,
Andrea J. Liu
Abstract:
The plasticity of amorphous solids undergoing shear is characterized by quasi-localized rearrangements of particles. While many models of plasticity exist, the precise relationship between plastic dynamics and the structure of a particle's local environment remains an open question. Previously, machine learning was used to identify a structural predictor of rearrangements, called "softness." Altho…
▽ More
The plasticity of amorphous solids undergoing shear is characterized by quasi-localized rearrangements of particles. While many models of plasticity exist, the precise relationship between plastic dynamics and the structure of a particle's local environment remains an open question. Previously, machine learning was used to identify a structural predictor of rearrangements, called "softness." Although softness has been shown to predict which particles will rearrange with high accuracy, the method can be difficult to implement in experiments where data is limited and the combinations of descriptors it identifies are often difficult to interpret physically. Here we address both of these weaknesses, presenting two major improvements to the standard softness method. First, we present a natural representation of each particle's observed mobility, allowing for the use of statistical models which are both simpler and provide greater accuracy in limited data sets. Second, we employ persistent homology as a systematic means of identifying simple, topologically-informed, structural quantities that are easy to interpret and measure experimentally. We test our methods on two-dimensional athermal packings of soft spheres under quasi-static shear. We find that the same structural information which predicts small variations in the response is also predictive of where plastic events will localize. We also find that excellent accuracy is achieved in athermal sheared packings using simply a particle's species and the number of nearest neighbor contacts.
△ Less
Submitted 27 October, 2020;
originally announced October 2020.
-
Mean-Field Predictions of Scaling Prefactors Match Low-Dimensional Jammed Packings
Authors:
James D Sartor,
Sean A. Ridout,
Eric I. Corwin
Abstract:
No known analytic framework precisely explains all the phenomena observed in jamming. The replica theory for glass and jamming is a mean field theory which attempts to do so by working in the limit of infinite dimensions, such that correlations between neighbors are negligible. As such, results from this mean field theory are not guaranteed to be observed in finite dimensions. However, many result…
▽ More
No known analytic framework precisely explains all the phenomena observed in jamming. The replica theory for glass and jamming is a mean field theory which attempts to do so by working in the limit of infinite dimensions, such that correlations between neighbors are negligible. As such, results from this mean field theory are not guaranteed to be observed in finite dimensions. However, many results in mean field for jamming have been shown to be exact or nearly exact in low dimensions. This suggests that the infinite dimensional limit is not necessary to obtain these results. In this paper, we perform precision measurements of jamming scaling relationships between pressure, excess packing fraction, and number of excess contacts from dimensions 2-10 in order to extract the prefactors to these scalings. While these prefactors should be highly sensitive to finite dimensional corrections, we find the mean field predictions for these prefactors to be exact in low dimensions. Thus the mean field approximation is not necessary for deriving these prefactors. We present an exact, first principles derivation for one, leaving the other as an open question.
△ Less
Submitted 20 April, 2021; v1 submitted 16 June, 2020;
originally announced June 2020.
-
Structured randomness: Jamming of soft discs and pins
Authors:
Prairie Wentworth-Nice,
Sean A. Ridout,
Brian Jenike,
Ari Liloia,
Amy L. Graves
Abstract:
Simulations are used to find the zero temperature jamming threshold, $φ_j$, for soft, bidisperse disks in the presence of small fixed particles, or "pins", arranged in a lattice. The presence of pins leads, as one expects, to a decrease in $φ_j$. Structural properties of the system near the jamming threshold are calculated as a function of the pin density. While the correlation length exponent rem…
▽ More
Simulations are used to find the zero temperature jamming threshold, $φ_j$, for soft, bidisperse disks in the presence of small fixed particles, or "pins", arranged in a lattice. The presence of pins leads, as one expects, to a decrease in $φ_j$. Structural properties of the system near the jamming threshold are calculated as a function of the pin density. While the correlation length exponent remains $ν= 1/2$ at low pin densities, the system is mechanically stable with more bonds, yet fewer contacts than the Maxwell criterion implies in the absence of pins. In addition, as pin density increases, novel bond orientational order and long-range spatial order appear, which are correlated with the square symmetry of the pin lattice.
△ Less
Submitted 29 April, 2020; v1 submitted 9 April, 2020;
originally announced April 2020.
-
Predicting plasticity in disordered solids from structural indicators
Authors:
D. Richard,
M. Ozawa,
S. Patinet,
E. Stanifer,
B. Shang,
S. A. Ridout,
B. Xu,
G. Zhang,
P. K. Morse,
J. -L. Barrat,
L. Berthier,
M. L. Falk,
P. Guan,
A. J. Liu,
K. Martens,
S. Sastry,
D. Vandembroucq,
E. Lerner,
M. L. Manning
Abstract:
Amorphous solids lack long-range order. Therefore identifying structural defects -- akin to dislocations in crystalline solids -- that carry plastic flow in these systems remains a daunting challenge. By comparing many different structural indicators in computational models of glasses, under a variety of conditions we carefully assess which of these indicators are able to robustly identify the str…
▽ More
Amorphous solids lack long-range order. Therefore identifying structural defects -- akin to dislocations in crystalline solids -- that carry plastic flow in these systems remains a daunting challenge. By comparing many different structural indicators in computational models of glasses, under a variety of conditions we carefully assess which of these indicators are able to robustly identify the structural defects responsible for plastic flow in amorphous solids. We further demonstrate that the density of defects changes as a function of material preparation and strain in a manner that is highly correlated with the macroscopic material response. Our work represents an important step towards predicting how and when an amorphous solid will fail from its microscopic structure.
△ Less
Submitted 23 November, 2020; v1 submitted 25 March, 2020;
originally announced March 2020.
