betaFeedback

WorksPhysics and Astronomy

Yielding and memory in a driven mean-field model of glasses

Abstract

Glassy systems reveal a wide variety of generic behaviors, which lack a unified theoretical description. Here, we study a mean-field model, recently shown to reproduce the universal non-phononic vibrational spectra of glasses, under oscillatory driving forces. The driven mean-field model, featuring a disordered Hamiltonian structure, naturally predicts the salient dynamical phenomena in cyclically deformed glasses. Specifically, it features an oscillatory yielding transition, characterized by an absorbing-to-diffusive transition in the system's microscopic trajectories and large-scale hysteresis. The model also reveals dynamic slowing-down from both sides of the transition, as well as mechanical and thermal annealing effects that mirror their glass counterparts. Finally, we demonstrate a non-equilibrium ensemble equivalence between the driven post-yielding dynamics at fixed quenched disorder and quenched disorder averages of the non-driven system, along with memory formation.

Background and motivation

Glasses are formed by rapidly cooling (quenching) equilibrium liquids such that crystallization is avoided [1, 2, 3, 4]. Glasses are intrinsically non-equilibrium and disordered materials, revealing generic properties and dynamical behaviors that differ from their ordered crystalline counterparts. These include universal properties of the low-frequency vibrational spectra of glasses, which affect low temperature thermodynamic and transport properties [5, 6, 7], and strongly driven thermo-mechanical phenomena such as thermal and mechanical annealing, yielding transitions from quiescent to plastically-flowing states, and memory formation [8, 9, 10, 11, 12, 13]. All of these are inherently linked to the multiplicity of energy minima of glassy systems, manifested in their complex energy landscapes [1, 2, 3, 4, 14]. Despite their basic importance, along with a broad range of engineering applications [15, 16, 17], we currently lack a unified and predictive theory that includes both sets of features described above.

Recent progress established that the vibrational density of states (VDoS) D(ω)\mathcal{D}(\omega) of glasses, experimentally probed through various scattering techniques [18, 19], features a non-phononic contribution DG(ω)\mathcal{D}_{\mathrm{G}}(\omega) for small vibrational frequencies ω\omega, i.e., of soft modes [20, 21, 22, 23, 24, 25, 26, 27]. That is, in addition to extended low-frequency phonons, which exist in both glassy and crystalline solids, and follow Debye’s VDoS, glasses also feature localized, non-phononic modes that follow a universal quartic VDoS DG(ω)∼ω4\mathcal{D}_{\mathrm{G}}(\omega)\sim\omega^{4} for ω→0\omega\to0 [24, 25, 26, 27]. Moreover, DG(ω)\mathcal{D}_{\mathrm{G}}(\omega) features a generic peak [27, 28, 29, 30] at slightly higher frequency ωp\omega_{\mathrm{p}} (typically in the THz regime), intrinsically related to the well-known “boson peak” [31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41]. DG(ω)\mathcal{D}_{\mathrm{G}}(\omega) reflects statistical properties of minima of the underlying complex energy landscape, a manifestation of self-organization during glass formation. Furthermore, dissipative deformation in driven glasses — i.e., transitions between different minima of an energy landscape — corresponds to irreversible (plastic) rearrangements of the localized, non-phononic modes [42, 43, 44, 45, 46, 47, 48].

In parallel, extensive work has demonstrated that glasses driven by oscillatory forces undergo a plastic yielding transition with increasing driving amplitude [49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70]. The observed yielding transition revealed various generic properties, including an absorbing-to-diffusive transition in the system’s microscopic trajectories and large-scale plastic hysteresis [49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66], yet again related to dynamics in the underlying complex energy landscape. In addition, it has been shown that cyclically deformed glasses can be trained to store mechanical memories, which can be subsequently extracted through well-defined reading protocols [71, 72, 73, 74, 75, 76, 77, 78, 79].

Here, we show that a Hamiltonian mean-field model of glasses, which remarkably reproduced the observed properties of the non-phononic VDoS DG(ω)\mathcal{D}_{\mathrm{G}}(\omega) [30, 80, 81], naturally predicts the salient features of the yielding transition and mechanical memory formation extensively observed in particle-based glasses under driving, with no free parameters relative to the non-driven regime. The model relates statistical-structural properties of quiescent glasses to their elasto-plastic driven dynamics, and opens the way for gaining additional insight into such relations. Taken together, the Hamiltonian mean-field model studied here constitutes a theoretical framework that unifies a broad range of static and dynamic glass properties, and offers a platform for studying additional glass phenomena that currently lack a complete theoretical understanding.

Model formulation

The mean-field model, originally introduced in [80, 81] and inspired by [22, 82] (denoted by the acronym ‘KHGPS’ based on the authors’ surnames), is defined by the following Hamiltonian for NN anharmonic oscillators

H=12∑iκixi2+A4!∑ixi4+∑i<jJijxixj−f(t)∑ixi .(1)H = \frac{1}{2}\sum_i \kappa_i x_i^2 + \frac{A}{4!}\sum_i x_i^4 + \sum_{i<j}J_{ij}x_i x_j - f(t)\sum_i x_i \, . \tag*{(1)}

where i,j=1i,j=1–NN. The oscillators, each characterized by a continuous scalar displacement xix_i, initially represent localized, mesoscopic regions inside an equilibrium, high-temperature liquid, featuring random stiffnesses κi\kappa_i that are uniformly distributed over [0,κ0][0,\kappa_0] and a deterministic quartic anharmonicity of amplitude AA. κi\kappa_i represents instantaneous (short time) and local elasticity in the liquid state, which can be negative as well, hence we take it to attain values down to zero. Upon quenching the liquid, a self-organization process which is mimicked in the framework of the model by the minimization of the Hamiltonian HH into one of its local minima, long-range elasticity builds up.

The latter is represented by bilinear interactions between the oscillators, with disordered coupling coefficients JijJ_{ij} that are i.i.d. Gaussian random variables, with zero mean and variance J2/NJ^2/N for i≠ji\ne j. The mean-field (infinite-dimensional) nature of the model is encapsulated in the disordered interaction coefficients JijJ_{ij}, which feature no spatial dependence and couple each oscillator to all other oscillators. It is important to note that as κi>0\kappa_i>0 and A>0A>0, all nontrivial properties of the model’s glassy landscape emerge from the self-reorganization process induced by the disordered interactions [80, 81].

The force f(t)f(t), which has not been considered previously, is a time-dependent external field that is coupled to the mean oscillators position, ⟨x⟩≡N−1∑ixi\langle x\rangle\equiv N^{-1}\sum_i x_i. In previous analyses of the model [30, 80, 81, 83, 84], which focused on the statistical properties of the eigenvalues at minima of HH and on the localization of the corresponding eigenmodes, f(t)=hf(t)=h has been used, with hh being constant and small. These analyses gave rise to the non-phononic VDoS DG(ω)D_G(\omega) discussed above. Here, we consider the driven regime, where f(t)f(t) is taken to be an oscillatory force of amplitude f0f_0. We study the model in the athermal quasi-static (AQS) limit, where the temperature and driving rate are vanishingly small. Consequently, we do not specify the period of f(t)f(t), but rather consider the cyclic sequence 0→f0→0→−f0→⋯0\to f_0\to0\to-f_0\to\cdots, applied through small increments Δf\Delta f, each followed by a minimization of HH (see Appendix). Therefore, we will quantify the evolution of the system in terms of the number of driving cycles nn. We measure energy in units of κ02/A\kappa_0^2/A, force in units of κ03/2/A1/2\kappa_0^{3/2}/A^{1/2} and length in units of κ01/2/A1/2\kappa_0^{1/2}/A^{1/2}. Consequently, the driven model features only two parameters (to be used hereafter in their dimensionless form), the strength of quenched disorder JJ and the amplitude of the periodic driving f0f_0.

The yielding transition

A large body of recent work, both computational and experimental [49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70], demonstrated that glasses under oscillatory shear deformation γ(t)\gamma(t) undergo a well-defined yielding transition as a function of the amplitude γ0\gamma_0 of the applied deformation. The analog of γ0\gamma_0 in Eq. (1) is the amplitude f0f_0 of the driving force f(t)f(t). Consequently,

our first goal is to test whether the mean-field model features a similar yielding transition as a function of f0f_0. We fix the only other parameter in the model to J=0.5J=0.5, a choice that will be further discussed below, and drive the system to a steady oscillatory state (i.e., over a large enough number of cycles nn) at various f0f_0 values.

In Figure 1a, we plot the average potential energy U(t)≡H(t)+f(t)∑ixiU(t)\equiv H(t)+f(t)\sum_i x_i per oscillator as a function of f(t)f(t) in a steady-state cycle for 4 values of f0f_0 (see legend). It is observed that the U(f)U(f) relation is elastic, i.e., revealing no hysteresis, at small f0f_0, but develops an elasto-plastic hysteretic response with increasing f0f_0, indicating an underlying yielding transition. For comparison, we present in Figure 1b the corresponding quantities obtained in a particle-based computer glass under AQS oscillatory shear deformation (see figure caption and Appendix), revealing striking qualitative similarities. Note also that U(f=0)U(f=0), and likewise U(γ=0)U(\gamma=0), vary non-monotonically with the periodic driving amplitude, which is also a signature of an underlying yielding, as will be further discussed below.

Figure 1. (a) The energy per oscillator U(t)/NU(t)/N vs. the periodic driving f(t)f(t) in steady-state at various amplitudes f0f_0 (see legend), see text for discussion. Here, we used N=8192N=8192 and J=0.5J=0.5, and the results are averaged over more than M=20M=20 realizations of quenched disorder and more than 10 steady-state cycles. The curves are displaced vertically for visual clarity, but maintain their order (see Appendix). (b) The same as (a), but for a computer glass driven by an oscillatory shear γ(t)\gamma(t) (see Appendix for details). (c) Representative stroboscopic trajectories of the dynamics of the mean-field model at various amplitudes of f0f_0 (see legend). Shown is the energy at the end of each cycle, Uf(t)=0U_{f(t)=0}, vs. the number of cycles nn, see text for discussion. (d) The corresponding results for the computer glass. Note that at the intermediate f0f_0 value, the model’s trajectory reveals multiperiodicity, cf. panel (c), not observed in the corresponding trajectory of the computer glass, cf. panel (d).

In glassy systems undergoing a yielding transition, the emergence of large-scale hysteresis, as observed in Figs. 1a-b, is accompanied by an absorbing-to-diffusive transition in the system’s microscopic trajectories [49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66]. That is, below the transition the system periodically visits exactly the same microscopic state after an integer numbers of cycles, i.e., an absorbing state accompanied by no or minute dissipation, while above the transition the behavior is diffusive, characterized by non-periodic trajectories and significant dissipation. In Figure 1c, we plot the (stroboscopic) energy at the end of each deformation cycle as a function of nn, for 3 of the values of f0f_0 used in Figure 1a (see legend). It is observed that the microscopic trajectories in the model undergo an absorbing-to-diffusive transition, indicating that an yielding transition occurs at a driving amplitude slightly larger than f0=0.16f_0 = 0.16. In Figure 1d, we present the corresponding results for the computer glass, yet again revealing striking similarities.

In Figure 2a, we plot the average steady-state stroboscopic energy UssU_{\rm ss} in the model (at f(t)=0f(t)=0, per oscillator) as a function of f0f_0. As implied by Figure 1a-Figure 1b, Uss(f0)U_{\rm ss}(f_0) initially decreases with increasing f0f_0, indicating mechanical annealing, a process in which the system visits deeper energy minima, as extensively observed [49, 53, 55, 56, 57, 58, 59, 60, 61, 62]. Mechanical annealing is maximal at f0≃0.2f_0 \simeq0.2, close to the above-mentioned estimate of the yielding amplitude, above which Uss(f0)U_{\rm ss}(f_0) increases. In Figure 2b, we plot the average fraction of reversible/periodic (marked by R\mathcal{R}) and irreversible/non-periodic (marked by I\mathcal{I}) steady-states as

Figure 2. (a) The steady-state stroboscopic energy Uss(f0)U_{\rm ss}(f_0), evaluated at the end of each cycle, for the calculations reported on in Figure 1a. (b) The fraction of R\mathcal{R} and I\mathcal{I} steady-states vs. f0f_0 (R\mathcal{R} denotes reversible/periodic steady-states, while I\mathcal{I} denotes irreversible/non-periodic ones). (c) The steady-state cycle-to-cycle mean square displacement MSDss\mathrm{MSD}_{\rm ss}, defined in Eq. (2), vs. f0f_0. (d) The characteristic accumulated driving force facc∗f_{\rm acc}^* needed to reach a steady-state as a function of f0f_0, obtained by fitting a stretched exponential function Uss+ΔUexp⁡[−(facc/facc∗)1/2]U_{\rm ss}+\Delta U\exp[-(f_{\rm acc}/f_{\rm acc}^*)^{1/2}] (where facc≡4f0nf_{\rm acc}\equiv4f_0n) to stroboscopic trajectories (see Appendix for details). The dashed lines are guides to the eye.

a function of f0f_0. A crossover between the two types of steady-states occurs close to the position of the minimum of Uss(f0)U_{\rm ss}(f_0), which we hereafter denote as the yielding amplitude fy(J=0.5)≃0.2f_y(J=0.5)\simeq0.2. We verified that this yielding amplitude essentially remains unchanged when NN is increased.

Next, we consider fluctuations quantified through the cycle-to-cycle (stroboscopic) mean square displacement of the individual oscillators

MSD(n)≡N−1∑i[xi(n)−xi(n−1)]2.(2)\mathrm{MSD}(n)\equiv N^{-1}\sum_i [x_i(n)-x_i(n-1)]^2 . \tag*{(2)}

In Figure 2c, we plot the steady-state value of MSD(n)\mathrm{MSD}(n), denoted as MSDss\mathrm{MSD}_{\rm ss} (the approach to steady-state is discussed next), against f0f_0. MSDss(f0)\mathrm{MSD}_{\rm ss}(f_0) significantly grows from zero near the yielding amplitude fy(J=0.5)≃0.2f_y(J=0.5)\simeq0.2, an observation that mirrors the corresponding behavior of the particle diffusion coefficient in particle-based computer glasses [49, 52, 55, 56, 57]. The latter also reveal a clear signature of the yielding transition in transient (non-steady-state) dynamics, where the number of cycles required to reach steady-state grows significantly as the yielding transition is approached from both sides. To test this in the mean-field model, we define facc≡4f0nf_{\rm acc}\equiv4f_0n and extract the characteristic accumulated driving force facc∗f_{\rm acc}^* needed to reach steady-state (see Appendix). facc∗(f0)f_{\rm acc}^*(f_0) is presented in Figure 2d, revealing dynamic slowing-down as facc∗→fyf_{\rm acc}^* \to f_y, as in computer glasses.

Overall, Figure 1-Figure 2 establish that the same mean-field model that properly predicts the non-phononic VDos DG(ω)D_{\rm G}(\omega) of glasses in the non-driven regime, undergoes an oscillatory yielding transition that closely resembles that of glassy systems in the driven regime, involving no additional free parameters. Next, we will use the model to shed additional light on the yielding transition. Specifically, we will explore the relations between the statistical-mechanical properties of configurations visited by a single system during the post-yielding driven dynamics and their non-driven counterparts, corresponding to ensemble averaging over quenched disorder. We will also study the effect of the strength of quenched disorder on the yielding transition and its relations to thermal annealing. In Appendix, we demonstrate mechanical memory formation, in which memories can be stored in the model and subsequently extracted.

Statistical-mechanical properties of configurations explored during post-yielding dynamics

Some earlier theoretical works suggested that the yielding transition is intrinsically related to the generation of configurational entropy [85, 86, 87, 88]. That is, it has been suggested that upon yielding a large number of glassy configurations in the energy landscape become dynamically accessible. We can use the unifying nature of the mean-field model under discussion to push this line of thinking further and ask: do the statistical-mechanical properties of the minima explored by a driven system of finite size NN in the post-yielding regime identify with the corresponding known properties of ensemble averages over realizations of the quenched disorder in the non-driven system?

To make this question and the procedure to address it as concrete and quantitative as possible, we consider the non-phononic VDoS DG(ω)\mathcal{D}_{G}(\omega) and in particular its universal quartic tail DG(ω)∼ω4\mathcal{D}_{G}(\omega)\sim\omega^{4} of soft modes. For a moderate value of NN, say N=8192N=8192 (as used in Figure 1-Figure 2), the vibrational spectrum of an energy minimum of the non-driven system of a single realization of the quenched disorder does not reveal the universal quartic tail DG(ω)∼ω4\mathcal{D}_{G}(\omega)\sim\omega^{4}. This is demonstrated in Figure 3a, where DG(ω,n=0)\mathcal{D}_{G}(\omega,n=0) is plotted (here, n=0n=0 corresponds to a non-driven, single realization of the quenched disorder with J=0.5J=0.5). The reason for this is that the probability to find a low-frequency eigenmode in the ω4\omega^{4} tail is very small. The statistically convergent power-law tail DG(ω)∼ω4\mathcal{D}_{G}(\omega)\sim\omega^{4}, down to an NN-dependent minimal vibrational frequency, is obtained only by averaging over a large enough ensemble of independent realizations of the quenched disorder. This is demonstrated in Figure 3a by superposing DG(ω)\mathcal{D}_{G}(\omega) obtained through M=1000M=1000 realizations of the quenched disorder, clearly revealing the universal quartic ω4\omega^{4} tail.

Figure 3. (a) The non-phononic VDoS DG(ω)∼ω4\mathcal{D}_{G}(\omega) \sim\omega^{4} (see line and power-law triangle) obtained by averaging over M=1000M=1000 realizations (with N=8192N=8192) of the quenched disorder with J=0.5J=0.5 (large gray squares) and by periodically driving a single realization with f0=0.28>fy(J=0.5)f_{0}=0.28>f_{y}(J=0.5) over sufficiently large number of cycles nn (see legend). It is observed that DG(ω;n)\mathcal{D}_{G}(\omega;n) entirely lacks the ω4\omega^{4} tail initially (n=0n=0) and that the latter progressively builds up with increasing nn, see text for additional discussion. The arrows mark the edge of the vibrational spectrum per nn. (b) The number density of the lowest 100 eigenfrequencies, denoted as DG(ωmin⁡;n)\mathcal{D}_{G}(\omega_{\min};n), with increasing nn (see color bar on the right, using the same color code of panel (a)). The progressive build up of the ω4\omega^{4} tail is highlighted by the power-law triangle.

The question then, rephrasing the above-posed one, is whether driving a single realization with f0>fyf_{0}>f_{y} for a sufficiently large number of cycles nn would result in DG(ω,n)\mathcal{D}_{G}(\omega,n) converging to DG(ω)\mathcal{D}_{G}(\omega) at fixed NN. That is, we ask whether a single system in the “chaotic” post-yielding regime is dynamically self-averaging, at least as far as the low-frequency tail of DG(ω)\mathcal{D}_{G}(\omega) is concerned. In Figure 3a, we superpose results for DG(ω,n)\mathcal{D}_{G}(\omega,n) for increasing values of driving cycles nn (see legend), demonstrating progressive convergence towards DG(ω)∼ω4\mathcal{D}_{G}(\omega)\sim\omega^{4}. The upward pointing arrows indicate the edges of the resolved ω4\omega^{4} tail, clearly demonstrating its systematic shift towards lower frequencies along the power-law tail with increasing nn. To reiterate the point, we plot in Figure 3b the number density of the lowest 100 eigenfrequencies, denoted by DG(ωmin⁡;n)\mathcal{D}_{G}(\omega_{\min};n), which further highlights the progressive build up of the quartic tail with increasing nn. These results demonstrate the equivalence of the two protocols in the present context, where a larger/smaller MM in the non-driven system implies a larger/smaller nn in the single driven system at a fixed NN.

Variable quenched disorder and thermal annealing

The analysis so far focused on a fixed strength of quenched disorder, characterized by J=0.5J=0.5. Our next goal is to understand the dependence of the yielding transition under oscillatory driving forces on JJ. In glasses of fixed composition and interparticle interactions, the state of disorder can be varied through thermal annealing [58, 59, 89, 90, 91, 92, 93, 94, 95], resulting in self-generated positional — and consequently also interactional/mechanical — disorder. We discuss thermal annealing later, but first take advantage of the fact that in the dimensionless formulation of the model, quenched disorder is varied by a single parameter JJ.

In Appendix (specifically, see Sect. E therein), we show that the number of soft modes up to ωp\omega_{\mathrm{p}} (defined above), quantified by ∫0ωpDG(ω) dω\int_{0}^{\omega_{\mathrm{p}}}\mathcal{D}_{G}(\omega)\,d\omega, is small at small JJ and increases with JJ until it levels off above J≃0.2J\simeq0.2. Since the number of soft modes is related to accessible minima in the energy landscape and to plastic rearrangements, we expect small values of JJ to give rise to little mechanical annealability. In this regime, we expect a “brittle-like” behavior with no plastic yielding, which may be analogous to the so-called “brittle yielding” observed in particulate systems in finite dimensions [58, 59]. On the other hand, at larger JJ values, we expect to observe an increasing degree of mechanical annealability and a “ductile-like” yielding transition (already established above for J=0.5J=0.5). See additional discussion in Appendix.

In testing these expectations, one should bear in mind that while under thermal annealing the typical energy scale remains the same (inherited from interparticle interactions), this is not the case upon varying JJ in the model. To stress this point, we plot in the inset of Figure 4a the variation of the average energy per oscillator U0/NU_{0}/N of the non-driven model with JJ. It varies from U0/N=−2.65×10−4U_{0}/N=-2.65\times10^{-4} to U0/N=−2.68U_{0}/N=-2.68, i.e. by 4 orders of magnitude, when JJ is varied by an order of magnitude. Consequently, in the oscillatory driven case to be considered, we focus on the energy difference ΔUss\Delta U_{\mathrm{ss}} relative to the minimal steady-state energy as a function of f0f_0. The results for ΔUss(f0)\Delta U_{\mathrm{ss}}(f_0) are presented in Figure 4a, for J=0.1−1.1J=0.1-1.1 (see legend in Figure 4b).

Figure 4. (a) ΔUss(f0)\Delta U_{\mathrm{ss}}(f_0), similarly to Figure 2a but for various levels of quenched disorder quantified by JJ (see legend in panel (b)). Here, ΔUss(f0)\Delta U_{\mathrm{ss}}(f_0) is measured relative to the minimum of Uss(f0)U_{\mathrm{ss}}(f_0) per JJ, with N=4096N=4096 and M=16M=16. (inset) The energy U0U_0 of the non-driven system vs. JJ. (b) The same as panel (a), but vs. f0/Jf_0/J, see text for discussion. (c) ΔUss(f0)\Delta U_{\mathrm{ss}}(f_0) for thermally annealed states quantified by the “parent temperature” TpT_{\mathrm{p}}, see text for discussion and Appendix for technical details.

An immediate observation in Figure 4a is that the yielding amplitude fy(J)f_{\mathrm{y}}(J) is an increasing function of JJ. While we do not discuss here theoretical considerations that may predict this dependence (and in fact indicate a stronger than linear dependence), we replot the results as ΔUss(f0/J)\Delta U_{\mathrm{ss}}(f_0/J) in Figure 4b. The latter supports the expectations stated above that the degree of mechanical annealibility increases with JJ and for sufficiently small JJ a “brittle-like” behavior with no yielding emerges. These findings are qualitatively similar to the corresponding phenomenology observed as a function of the degree of thermal annealing [53, 56, 58, 59, 60].

It has been shown that under a variable degree of thermal annealing, a glass is sensitive to the initial state of disorder for f0<fyf_0 < f_{\mathrm{y}}, but is independent of it for f0≥fyf_0 \ge f_{\mathrm{y}} [53, 56, 58, 59, 60]. Indeed, the different ΔUss(f0/J)\Delta U_{\mathrm{ss}}(f_0/J) curves in Figure 4b nearly collapse sufficiently above yielding, yet the variability with JJ does not allow to directly map the effect of quenched disorder to that of thermal annealing. Consequently, and while we do not extensively discuss here thermal dynamics of the model, we thermalized it and generated zero temperature initial states at different degrees of thermal annealing. The latter are quantified by the “parent temperature” TpT_{\mathrm{p}} from which zero temperature states are quenched (see Appendix). In Figure 4c, we present Uss(f0)U_{\mathrm{ss}}(f_0) for 4 values of TpT_{\mathrm{p}}, obtained by subjecting the thermally annealed states to oscillatory mechanical driving. The pre-yielding dependence on thermal annealing and its disappearance in the post-yielding regime are observed. Note, though, that for the range of thermal annealing considered here (see Appendix), a “brittle-like” behavior is not observed.

Discussion

We studied a Hamiltonian mean-field model of glasses, with continuous degrees of freedom, under oscillatory driving forces. In the non-driven regime, the model has been recently shown to correctly predict the low-frequency properties of the non-phononic vibrational density of states Dg(ω)D_{\mathrm{g}}(\omega) of glasses, including its universal quartic tail, the boson peak and the localization of vibrational modes [30, 80, 81]. We showed that in the periodically driven regime, the model undergoes a yielding transition as a function of the amplitude of the oscillatory force, revealing a wide range of dynamical behaviors that closely mirror observations in particle-based glasses. We also demonstrated mechanical memory formation (see Appendix), similarly to glasses.

Our analysis treats the model as a predictive theory in the sense that no additional free parameters are introduced relative to the non-driven version, where physically-relevant non-phononic excitations are defined. The latter, missing in discrete Hamiltonian models (e.g., [96, 97, 98]) and in non-Hamiltonian ones (e.g., [61, 62, 63, 99, 100, 101]) emerge from the physically-motivated Hamiltonian structure of the model and its continuous degrees of freedom. Overall, the results indicate that the model naturally unifies a broad range glassy phenomena, linked to the underlying complex energy landscape, and offers a generic framework to study the physics of glasses.

Acknowledgments

E.B. acknowledges support from the Israel Science Foundation (ISF Grant No. 403/24), the Ben May Center for Chemical Theory and Computation and the Harold Perlman Family. M.S. acknowledges support from the International Joint Graduate Program in Earth and Environmental Sciences. Part of the computations were performed using supercomputing resources at the Cyberscience Center, Tohoku University.

Appendix

The goal of this Appendix is to provide technical details regarding the results presented in the main text, and to offer additional supporting results, most notably about memory formation in Sect. B and about the relation between the strength of disordered interaction JJ and the number of soft non-phononic modes, and a “brittle-like” behavior, in Sect. E.

Numerical implementation of the model and additional details

The mean-field model we studied is formulated through the Hamiltonian HH defined in Eq. (1), where the parameters and nondimensionalization procedure are also discussed. In most of the results reported in the main text and in this Appendix (specifically, in Figure 1-Figure 3, Figure 4a-Figure 4b and Figure A4 therein), samples were prepared by performing energy minimization in the absence of an applied force, f(t)=0f(t)=0, starting from random oscillator configurations where each xix_i is drawn from a uniform distribution over [−ϵ/2,ϵ/2][-\epsilon/2,\epsilon/2] with ϵ=10−2\epsilon=10^{-2}, independently of JJ. We verified that ⟨x2⟩≫ϵ2\langle x^2\rangle\gg\epsilon^2 after minimization (for all JJ values employed), demonstrating that the value of ϵ\epsilon used does not affect the results. The system size was N=8192N=8192 for the results presented in Figure 1-Figure 3, Figure A4 and N=4096N=4096 for those presented in Figure 4. We prepared M≥20M\geq20 independent realizations of quenched disorder for Figure 1-Figure 2, Figure A4, and M=16M=16 realizations for Figure 4a-Figure 4b. We prepared M=1M=1 realizations for the driven minima-sampling presented in Figure 3a-Figure 3b, and M=1000M=1000 realizations for the “as-realized” data presented in Figure 3a.

Figure A4. The parallel read protocol, demonstrating memory formation in the mean-field model. MSDread(fread)\mathrm{MSD}_{\mathrm{read}}(f_{\mathrm{read}}) is plotted for various training cycles ntrn_{\mathrm{tr}} (see legend in panel (a)) and 4 different training amplitudes (marked by the vertical dashed line), (a) ftr=0.08f_{\mathrm{tr}}=0.08, (b) ftr=0.18f_{\mathrm{tr}}=0.18, (c) ftr=0.28f_{\mathrm{tr}}=0.28 and (d) ftr=0.36f_{\mathrm{tr}}=0.36. See text for definitions and discussion.

The oscillatory driving force

We have carried out oscillatory driving simulations in the athermal quasi-static (AQS) limit, starting from initial zero-force configurations (whose preparation protocol is detailed in Sec. A). We increased the applied force f(t)f(t) by small increments Δf\Delta f and performed energy minimization using the conjugate-gradient method after each increment. The force increment Δf\Delta f was selected depending on the value of JJ to keep the ratio Δf/J\Delta f/J at most 2×10−32\times10^{-3}. This choice ensures the proper resolution of the oscillators’ displacement Δx\Delta x in response to a force increment Δf\Delta f such that one can accurately detect plastic rearrangements, i.e., transitions between different minima in the complex energy landscape. Moreover, this choice ensures the numerical convergence of our results in the range of parameters under consideration. The values of Δf\Delta f per JJ are provided in Table 1. Note that the particular values were chosen such that f0/Δff_0/\Delta f is an integer. The driving was applied in the sequence 0→f0→0→−f0→⋯0\to f_0\to0\to-f_0\to\cdots and persisted until the potential energy U≡H+f∑ixiU\equiv H+f\sum_i x_i and mean-squared-displacement MSD(n)≡N−1∑i[xi(n)−xi(n−1)]2\mathrm{MSD}(n)\equiv N^{-1}\sum_i[x_i(n)-x_i(n-1)]^2 reached steady-states, where U(n)U(n) and MSD(n)\mathrm{MSD}(n) were calculated at the end of each cycle as functions of the number of driving cycles nn. Unless otherwise noted, all reported data points result from averaging over MM independent realizations, and for steady-state macroscopic quantities, data points are also averaged over ≥10\geq10 steady-state cycles.

Interaction strength JJForce increment Δf\Delta f
0.15×10−55\times10^{-5}
0.32×10−42\times10^{-4}
0.51×10−31\times10^{-3}
0.71×10−31\times10^{-3}
0.91.5×10−31.5\times10^{-3}
1.12×10−32\times10^{-3}

Table 1. Numerical force increment Δf(J)\Delta f(J)

Screening multiperiodic steady-state trajectories

At driving amplitudes f0f_0 close to the yielding amplitude fyf_y, the mean-field model can settle into one-periodic, multiperiodic or non-periodic steady-states. Figure A1a-Figure A1c display these representative stroboscopic trajectories in terms of MSD(n)\mathrm{MSD}(n). One- and multiperiodic steady-states form closed loop trajectories, but in the latter the periodicity is a multiple integer of the driving period which results in a finite value of MSDss\mathrm{MSD}_{\mathrm{ss}} [50, 51, 97, 102, 103]. Consequently, averaging over an ensemble containing many such trajectories could slightly obscure the emergence of non-periodic steady-states in our analysis. This is not the case in particulate glasses where the particle diffusion is measured, to which multiperiodic trajectories do not contribute. Consequently, in the MSDss\mathrm{MSD}_{\mathrm{ss}} results presented in Figure 2b, we screened multiperiodic trajectories. For completeness, we present in Figure A1d both the “multiperoidicity screened” data of Figure 2b and the corresponding unscreened data. It is observed, as expected, that differences exist only near fyf_y.

Figure A1. (a) One-periodic, (b) multiperiodic, and (c) non-periodic steady-states observed in J=0.5J=0.5 (and N=8192N=8192) systems driven at f0=0.18f_0=0.18, presented in terms of MSD(n)\mathrm{MSD}(n). (d) The steady-state mean-squared-displacement MSDss\mathrm{MSD}_{\mathrm{ss}} as a function of the driving amplitude f0f_0, plotted here for the unscreened (green diamonds) and multiperiodicity screened (brown-red squares) values. See text for definitions and discussion.

In addition, we note that the question of which glass (a coarse-grained model, a molecular computer glass or a laboratory glass) features multiperiodic steady-states

near yielding — and/or under what conditions — is an interesting open question for future investigation.

Extraction of the characteristic accumulated driving force

In Figure 2d, we report on the characteristic accumulated driving force facc∗f_{\mathrm{acc}}^{*} needed to reach a steady-state as a function of f0f_0. Here, we provide detail of the extraction procedure, which was performed as follows: we first plot the stroboscopic trajectories of the potential energy per oscillator U(n)/NU(n)/N, and then fit a stretched exponential function of the form Uss+ΔUexp⁡[−(facc/facc∗)1/2]U_{\mathrm{ss}}+\Delta U\exp[-(f_{\mathrm{acc}}/f_{\mathrm{acc}}^{*})^{1/2}], where facc≡4f0nf_{\mathrm{acc}}\equiv4f_0n. Figure A2a-Figure A2b demonstrate this procedure, where the symbols are the data points and the solid lines are the fitted curves. We fixed UssU_{\mathrm{ss}} and ΔU\Delta U, and extracted facc∗f_{\mathrm{acc}}^{*} as the sole fitting parameter, which is reported in Figure 2d.

Figure A2. (a) The energy per oscillator Uf(t)=0/NU_{f(t)=0}/N at the end of each cycle vs. facc=4f0nf_{\mathrm{acc}}=4f_0n (where nn is the number of cycles) for f0≤0.2f_0\leq0.2 (markers) and fits to a stretched exponential function (solid lines, see text), plotted for J=0.5J=0.5 and N=8192N=8192 systems. (b) Same as (a), but for f0>0.2f_0>0.2.

Undisplaced full-cycle trajectories

In Figure 1a-Figure 1b, we present the energy of the system vs. the periodic driving in steady-state at various driving amplitudes, for both the mean-field model (panel (a)) and computer glasses (panel (b), details about the latter are provided below in Sect. C). As mentioned therein, the curves are displaced vertically for visual clarity, but maintain their order. For completeness, we present in Figure A3a-Figure A3b the undisplaced curves.

Figure A3. The same data as in Figure 1a-Figure 1b, but without the vertical displacement used therein for visual clarity. In panel (a), we also present a zoom in on the small force portion of each curve.

Memory formation

Glasses are known to reveal rather intriguing memory effects related to their underlying complex landscapes (see, e.g., [12, 13] for recent reviews). Our goal here is to demonstrate that the mean-field model defined in Eq. (1) reveals mechanical memory effects, further strengthening the model’s “glassiness”. Specifically, we set J=0.5J=0.5 (for which fy≃0.2f_y\simeq0.2) and follow a procedure known as parallel read [78]. In this protocol, the system is first trained at a specific amplitude ftrf_{\mathrm{tr}} over ntrn_{\mathrm{tr}} training cycles. Then, identical copies of the trained system are subjected to a single “read cycle” at an amplitude freadf_{\mathrm{read}}. Memory is quantified through the MSD between configurations before and after a read cycle, i.e., through MSDread=N−1∑i[xi(fread)−xi(ftr,ntr)]2\mathrm{MSD}_{\mathrm{read}}=N^{-1}\sum_i[x_i(f_{\mathrm{read}})-x_i(f_{\mathrm{tr}},n_{\mathrm{tr}})]^2.

For ftr<fyf_{\mathrm{tr}}<f_y and a large enough ntrn_{\mathrm{tr}}, we expect a memory of the amplitude ftrf_{\mathrm{tr}} to be stored in the system and robustly extracted from it. This is indeed demonstrated in Figure A4a, where we plot MSDread(fread)\mathrm{MSD}_{\mathrm{read}}(f_{\mathrm{read}}) for ftr=0.08<fy≃0.2f_{\mathrm{tr}}=0.08<f_y\simeq0.2, manifested in a strong suppression of MSDread(fread)\mathrm{MSD}_{\mathrm{read}}(f_{\mathrm{read}}) for fread≤ftrf_{\mathrm{read}}\leq f_{\mathrm{tr}} and a sharp transition at fread=ftrf_{\mathrm{read}}=f_{\mathrm{tr}}. A similar pre-yielding behavior is observed for ftr=0.18<fy≃0.2f_{\mathrm{tr}}=0.18<f_y\simeq0.2 in Figure A4b.

For ftr>fyf_{\mathrm{tr}}>f_y and a large enough ntrn_{\mathrm{tr}}, we expect a single reading cycle in the range fread<fyf_{\mathrm{read}}<f_y to lead to a non-negligible MSDread\mathrm{MSD}_{\mathrm{read}}, significantly larger than the corresponding MSDread\mathrm{MSD}_{\mathrm{read}} in Figure A4a-Figure A4b, in view of the accessibility of a large number of different configurations in the post-yielding regime. Yet, it is a priori unclear whether a signature of the training amplitude ftr>fyf_{\mathrm{tr}}>f_y is entirely washed out as freadf_{\mathrm{read}} is further increased or not. In fact, some systems revealed the latter behavior [71, 72, 73, 74, 76, 79] and others the former [78].

In Figure A4c-Figure A4d, we present results for the parallel read protocol with ftr=0.28, 0.36>fy≃0.2f_{\mathrm{tr}}=0.28,\,0.36>f_y\simeq0.2, respectively. First, it is observed — as anticipated — that MSDread\mathrm{MSD}_{\mathrm{read}} is indeed significantly larger than its counterpart in Figure A4a-Figure A4b for fread<fyf_{\mathrm{read}}<f_y. Moreover, MSDread\mathrm{MSD}_{\mathrm{read}} appears to reach an ftrf_{\mathrm{tr}}-dependent plateau for fread≃fyf_{\mathrm{read}}\simeq f_y. However, a deviation from the plateau is observed at fread≃ftrf_{\mathrm{read}}\simeq f_{\mathrm{tr}} (marked by the vertical dashed line), indicating that some memory of the training amplitude remains encoded in the system also in the post-yielding regime. These results strengthen the findings discussed in the main text, demonstrating that the model naturally captures a broad range of glassy behaviors.

Computer glasses

In Figure 1 (right column panels), we present results on computer glasses subjected to oscillatory shear, highlighting the similarities to the model observations (presented on the left column panels). Here, for completeness, we provide technical details about these computer glasses, which are extensively used in the literature (see, for example, [104]). We employed a generic glass-forming model in 3D, which consists of a 50:50 binary mixture of N=16,000N=16{,}000 “large” and “small” particles of equal mass mm enclosed in a cubic box of linear size LL. The particles interact via a radially symmetric purely repulsive inverse power-law pairwise potentials, given by

φ(rij)={ε[(λijrij)n+∑l=0qc2l(rijλij)2l],rijλij≤xc,0,rijλij>xc.(3)\varphi(r_{ij}) = \begin{cases} \varepsilon\left[\left(\frac{\lambda_{ij}}{r_{ij}}\right)^n+\sum_{l=0}^{q}c_{2l}\left(\frac{r_{ij}}{\lambda_{ij}}\right)^{2l}\right], & \frac{r_{ij}}{\lambda_{ij}}\leq x_c,\\ 0, & \frac{r_{ij}}{\lambda_{ij}}>x_c. \end{cases} \tag*{(3)}

where rijr_{ij} is the distance between the iith and jjth particles, ε\varepsilon is an energy scale, and xcx_c is the dimensionless distance for which φ\varphi vanishes continuously up to qq derivatives. Distances are measured in terms of the interaction lengthscale λ\lambda between two “small” particles, and the rest are chosen to be λij=1.18λ\lambda_{ij}=1.18\lambda for one “small” and one “large” particle, and λij=1.4λ\lambda_{ij}=1.4\lambda for two “large” particles. The coefficients c2lc_{2l} are chosen such that φ\varphi vanishes continuously up to qq derivatives, viz.

c2l=(−1)l+1(2q−2l)!!(2l)!!(n+2q)!!(n−2)!!(n+2l)xc−(n+2l).(4)c_{2l}=\frac{(-1)^{l+1}}{(2q-2l)!!(2l)!!}\frac{(n+2q)!!}{(n-2)!!(n+2l)}x_c^{-(n+2l)}. \tag*{(4)}

We choose the parameters xc=1.48x_c=1.48, n=10n=10, and q=3q=3. The density was set to be N/V=0.82λ−3N/V=0.82\lambda^{-3}, where V≡L3V\equiv L^3 is the volume. Temperatures are expressed in terms of ε/kB\varepsilon/k_{\mathrm{B}}, with kBk_{\mathrm{B}} being the Boltzmann constant, and time in terms of mλ2/ε\sqrt{m\lambda^2/\varepsilon}. This system undergoes a computer glass transition at Tg≈0.5T_{\mathrm{g}}\approx0.5 [104].

For Figure 1, we prepared M≥50M\geq50 samples by performing a continuous quench down to zero temperature at a cooling rate T˙=3×10−3\dot{T}=3\times10^{-3}, starting from equilibrium configurations at T=1.0T=1.0. Oscillatory shear AQS deformation is applied as follows; starting from a shear-free configuration, each driving step imposes a uniform shear of Δγ=2×10−4\Delta\gamma=2\times10^{-4}, followed by energy minimization via the conjugate gradient method. All simulations were performed under Lees-Edwards periodic boundary conditions. The driving was applied through the sequence of 0→γ0→0→−γ0→⋯0\to\gamma_0\to0\to-\gamma_0\to\cdots, and was continued until the potential energy U≡∑i<jφ(rij)U\equiv\sum_{i<j}\varphi(r_{ij}), measured at the end of each cycle, reached a steady-state.

configurations at T=1.0T=1.0. Oscillatory shear AQS deformation is applied as follows; starting from a shear-free configuration, each driving step imposes a uniform shear of Δγ=2×10−4\Delta\gamma=2\times10^{-4}, followed by energy minimization via the conjugate gradient method. All simulations were performed under Lees-Edwards periodic boundary conditions. The driving was applied through the sequence of 0→γ0→0→−γ0→⋯0\to\gamma_0\to0\to-\gamma_0\to\cdots, and was continued until the potential energy U≡∑i<jφ(rij)U\equiv\sum_{i<j}\varphi(r_{ij}), measured at the end of each cycle, reached a steady-state.

Thermal annealing

Most of the analysis in this work is devoted to applying mechanical driving to mean-field model states (minima of the Hamiltonian HH) obtained from realizations of the quenched disorder. Yet, in Figure 4c, we present results obtained by mechanically driving thermally annealed mean-field model states. Here, we briefly explained how the latter have been obtained.

We considered the following overdamped Langevin dynamics

γdxidt=−∂H∂xi+2kBTγ ζi(t),(5)\gamma\frac{dx_i}{dt}=-\frac{\partial H}{\partial x_i}+\sqrt{2k_{\mathrm{B}}T\gamma}\,\zeta_i(t), \tag*{(5)}

where HH is defined in Eq. (1), kBk_{\mathrm{B}} is the Boltzmann constant as above, γ\gamma is a damping coefficient, ζi(t)\zeta_i(t) is a stationary Gaussian process with ⟨ζi(t)⟩=0\langle\zeta_i(t)\rangle=0 and ⟨ζi(t)ζj(t′)⟩=δijδ(t−t′)\langle\zeta_i(t)\zeta_j(t')\rangle=\delta_{ij}\delta(t-t'), where δij\delta_{ij} is Kronecker’s delta, and Roman indices label different oscillators. The external field f(t)f(t) is set to zero during the thermal calculations. Temperatures are measured in terms of κ02/kBA\kappa_0^2/k_{\mathrm{B}}A, and time in terms of γ/κ0\gamma/\kappa_0. We numerically integrate Eq. (5) by considering a small time increment Δt=10−2\Delta t=10^{-2}. This step size is chosen based on several preliminary J=0.5J=0.5 simulations at T=3.0T=3.0 in which integrations were performed at systematically varied Δt\Delta t, ranging from 10−110^{-1} to 10−410^{-4}, until the N=4096N=4096 samples reached statistically stationary states. The changes in the resulting steady-states are negligible below Δt=3×10−2\Delta t=3\times10^{-2}, and hence we choose Δt=10−2\Delta t=10^{-2}.

Thermally annealed initial zero-force configurations were prepared through the following protocol: systems are initially driven to statistically stationary states at T=2.0T=2.0, characterized by the time-translational invariance of macroscopic observables. We then subject the samples to an instantaneous quench to the “parent temperature” Tp<2.0T_{\mathrm{p}}<2.0, and continue the simulation for a duration of 10510^5 time units (i.e., 10710^7 time steps) for Tp<1.0T_{\mathrm{p}}<1.0, and for a duration of 10310^3 for Tp≥1.0T_{\mathrm{p}}\geq1.0. Finally, we quench all samples to zero temperature by performing energy minimization using the conjugate gradient method. To obtain the results presented in Figure 4c, M=16M=16 independent quenched disorder realizations with N=4096N=4096 were prepared for 0.2≤Tp≤2.00.2\leq T_{\mathrm{p}}\leq2.0. From those, we subjected samples annealed at Tp=0.2T_{\mathrm{p}}=0.2, Tp=0.6T_{\mathrm{p}}=0.6, Tp=0.8T_{\mathrm{p}}=0.8, and Tp=2.0T_{\mathrm{p}}=2.0 to oscillatory driving, the results of which are reported on in Figure 4c.

In Figure A5a, we present H/NH/N computed at different temperatures TT just before the final instantaneous quench to zero temperature. The energy after the quench, denoted as HT=0/NH_{T=0}/N, is plotted in Figure A5b against TpT_p, where the latter corresponds to TT of panel (a) to which instantaneous T=0T=0 quenches are applied. It is important to note that we employed a simple, well-defined thermalization procedure, which does not guarantee reaching equilibrium at the full range of temperatures considered for the simulation times used. This issue is interesting on its own and should be studied separately.

Figure A5. (a) The energy per oscillator H/NH/N computed at different temperatures TT just before the final instantaneous quench to zero temperature. (b) The energy after a quench to zero temperature, HT=0/NH_{T=0}/N, plotted against the parent temperature TpT_p (the temperature TT appearing on panel (a) from which the system was quenched to T=0T=0). Colored symbols indicate the samples subjected to oscillatory driving, using the same symbols and colors as in Figure 4c.

The relation between JJ and the number of soft non-phononic modes, and a ``brittle-like'' behavior

As stressed in the main text, in the dimensionless formulation of the model, quenched disorder is varied by a single parameter JJ. Yet, various physical observables can be affected differently by varying JJ. Of particular interest is the effect of JJ on the underlying complex energy landscape of the model, which dictates many of its emergent statistical and dynamical properties. Specifically, one is interested in computing the multiplicity of energy landscape minima and the barriers that separate them, which is in general a daunting task. Soft non-phononic modes, as quantified by the vibrational density of states (VDoS) DG(ω)D_G(\omega) discussed in the main text, may serve as a proxy for these landscape properties. The reason is that they quantify “soft directions” in the energy landscape that are dynamically accessible to the system once some driving forces, e.g., mechanical and/or thermal, are applied. Here, we explore this possibility.

As mentioned in the introduction, and as extensively discussed in [30], DG(ω)D_G(\omega) features a generic peak at a frequency ωp\omega_p, which is intrinsically related to the well-known “boson peak” in glasses. In Figure A6a, we present the VDoS DG(ω)D_G(\omega) for the model over the entire range of JJ values considered in the main text, i.e., J=0.1J=0.1–1.11.1 (see legend), on a double-logarithmic scale. The universal quartic tail, DG(ω)∼ω4D_G(\omega)\sim\omega^4, is evident (see the lines going through the small ω\omega data and the 4:1 power-law triangle). The same data are presented in Figure A6b on a linear scale, clearly revealing a peak at ωp\omega_p, marked by an arrow on each curve (using the same color code as that of Figure A6a). The dependence of ωp\omega_p on JJ is presented in the inset over a broader range of JJ; it is important to note that for small JJ, e.g., J=0.1J=0.1 in the main panel, ωp\omega_p corresponds to the first local maximum of DG(ω)D_G(\omega), where the predominantly linear DG(ω)∼ωD_G(\omega)\sim\omega for ω>ωp\omega>\omega_p corresponds to initial liquid-like modes that did not undergo interaction-induced “mixing/reconstruction”, as discussed in [30].

Figure A6. (a) The non-phononic VDoS DG(ω)D_{\mathrm{G}}(\omega) for various JJ values (see legend), plotted on a double-logarithmic scale. The superposed straight lines are fits to Agω4A_{\mathrm{g}}\omega^{4} in the low-frequency regime. (inset) The extracted quartic prefactor AgA_{\mathrm{g}} as a function of JJ (the Ag(J)∼J−3A_{\mathrm{g}}(J)\sim J^{-3} behavior in the small JJ regime is highlighted by the superposed line and the power-law triangle). (b) The same as panel (a), but on a linear scale. The arrows mark the first peak frequency ωp\omega_{\mathrm{p}} (following the same color code of panel (a)). (inset) ωp\omega_{\mathrm{p}} vs. JJ (the ωp(J)∼J\omega_{\mathrm{p}}(J)\sim J behavior in the small JJ regime is highlighted by the superposed line and the power-law triangle). (c) The number of soft non-phononic modes up to the frequency ωp\omega_{\mathrm{p}}, ∫0ωpDG(ω) dω\int_{0}^{\omega_{\mathrm{p}}}D_{\mathrm{G}}(\omega)\,\mathrm{d}\omega, plotted against JJ (brown-red squares). ∫0ωpAgω4 dω=Agωp5/5\int_{0}^{\omega_{\mathrm{p}}}A_{\mathrm{g}}\omega^{4}\,\mathrm{d}\omega=A_{\mathrm{g}}\omega_{\mathrm{p}}^{5}/5 vs. JJ is plotted as well (green diamonds). The ∼J2\sim J^{2} behavior in the small JJ regime is highlighted by the lines superposed on both datasets and the power-law triangle. The non-phononic VDoSs in the non-driven regime analyzed and presented here have been obtained by averaging over M=2000M=2000 realizations of the quenched disorder in a system size of N=8192N=8192. See text for the discussion of the results.

A natural quantity of interest is the number of soft modes up to ωp\omega_p, as quantified by

∫0ωpDG(ω) dω(6)\int_0^{\omega_p} D_G(\omega)\,d\omega \tag*{(6)}

which may be viewed as a quantifier of disorder [26], and its variation with JJ. The quantity defined in Eq. (6) is plotted in Figure A6c (brown-red squares). It is observed that ∫0ωpDG(ω) dω\int_0^{\omega_p}D_G(\omega)\,d\omega is very small for small JJ, increases as ∼J2\sim J^2 up to J≃0.2J\simeq0.2 and then levels off at larger JJ values.

These observations can be understood in the following way; first, consider the universal quartic tail DG(ω)=Agω4D_G(\omega)=A_g\omega^4, where the non-universal prefactor AgA_g [26] is extracted in Figure A6a. It is presented in the inset therein and observed to follow Ag(J)∼J−3A_g(J)\sim J^{-3} at small JJ. Combining the latter with the inset of Figure A6b, which indicates that ωp(J)∼J\omega_p(J)\sim J at small JJ, implies ∫0ωpAgω4 dω=Agωp5/5∼J2\int_0^{\omega_p}A_g\omega^4\,d\omega=A_g\omega_p^5/5\sim J^2, exactly as observed for ∫0ωpDG(ω) dω\int_0^{\omega_p}D_G(\omega)\,d\omega at small JJ. Agωp5/5A_g\omega_p^5/5 is superposed on Figure A6c (green diamonds), revealing the J2J^2 power-law at small JJ (albeit up shifted, as expected). Moreover, it also reveals a quasi-plateau above J≃0.2J\simeq0.2, similar to ∫0ωpDG(ω) dω\int_0^{\omega_p}D_G(\omega)\,d\omega.

The plateau in ∫0ωpDG(ω) dω\int_0^{\omega_p}D_G(\omega)\,d\omega above J≃0.2J\simeq0.2 indicates that the fraction of soft modes in the frequency range 0≤ω≤ωp(J)0\leq\omega\leq\omega_p(J) becomes nearly JJ independent for J≳0.2J\gtrsim0.2. The crossover between the J2J^2 power-law and the plateau at J≃0.2J\simeq0.2 corresponds to the strength of disordered interactions JJ at which all initial liquid-like modes undergo interaction-induced “mixing/reconstruction” [26, 30]. The small number of soft modes in the J≲0.2J\lesssim0.2 regime appears to indicate a reduced number of accessible energy landscape minima, possibly also echoed in the Replica Method solution of the non-driven model [81] in which the mean-field “effective potential” features double-wells with a low probability at small JJ values.

It is precisely this regime, i.e., J≲0.2J\lesssim0.2, where the driven model (cf. Figure 4a-Figure 4b) reveals little mechanical annealability and no yielding. We interpret this regime in the driven model to correspond to a “brittle-like” behavior. It is important to stress that the “brittle-like” behavior in the model is not accompanied by yielding, unlike the corresponding observations in computer glasses. The reason for this is clear from the structure of the mean-field model in which the Hamiltonian features only the leading order x4x^{4} nonlinearity/anharmonicity; that is, in the absence of accessible energy landscape minima for plastic rearrangements, the energy will simply be controlled by the x4x^{4} contributions.

Finally, in the plateau J≳0.2J\gtrsim0.2 regime in Figure A6c, other physical quantity may feature JJ dependence. For example, it is evident from Figure A6b that modes undergo progressive stiffening with increasing JJ, though their total number up to ωp(J)\omega_{\mathrm{p}}(J) is essentially constant. All of these important issues should be carefully addressed in a follow-up investigation.

References

References

  1. [1]M. Goldstein, Viscous liquids and the glass transition: A potential energy barrier picture, J. Chem. Phys. 51, 3728 (1969).DOI
  2. [2]P. G. Debenedetti and F. H. Stillinger, Supercooled liquids and the glass transition, Nature 410, 259 (2001).pubmed.ncbi.nlm.nih.gov/11258381
  3. [3]M. D. Ediger and P. Harrowell, Perspective: Supercooled liquids and glasses, J. Chem. Phys. 137, 080901 (2012).DOI
  4. [4]A. Cavagna, Supercooled liquids for pedestrians, Phys. Rep. 476, 51 (2009).DOI
  5. [5]R. C. Zeller and R. O. Pohl, Thermal conductivity and specific heat of noncrystalline solids, Phys. Rev. B 4, 2029 (1971).DOI
  6. [6]W. A. Phillips, Amorphous solids: Low-temperature properties, 1981st ed., edited by W. A. Phillips (Springer, Berlin, Germany, 1981).DOI
  7. [7]R. O. Pohl, X. Liu, and E. Thompson, Low-temperature thermal conductivity and acoustic attenuation in amorphous solids, Rev. Mod. Phys. 74, 991 (2002).DOI
  8. [8]M. Utz, P. G. Debenedetti, and F. H. Stillinger, Atomistic simulation of aging and rejuvenation in glasses, Phys. Rev. Lett. 84, 1471 (2000).pubmed.ncbi.nlm.nih.gov/11017545
  9. [9]V. Viasnoff and F. Lequeux, Rejuvenation and overaging in a colloidal glass under shear, Phys. Rev. Lett. 89, 065701 (2002).DOI
  10. [10]D. J. Lacks and M. J. Osborne, Energy landscape picture of overaging and rejuvenation in a sheared glass, Phys. Rev. Lett. 93, 255501 (2004).
  11. [11]L. Berthier, G. Biroli, L. Manning, and F. Zamponi, Yielding and plasticity in amorphous solids, Nat. Rev. Phys. 7, 313 (2025).DOI
  12. [12]N. C. Keim, J. D. Paulsen, Z. Zeravcic, S. Sastry, and S. R. Nagel, Memory formation in matter, Rev. Mod. Phys. 91, 035002 (2019).DOI
  13. [13]J. D. Paulsen and N. C. Keim, Mechanical memories in solids, from disorder to design, Annu. Rev. Condens. Matter Phys. 16, 61 (2025).DOI
  14. [14]P. K. Gupta and W. Kob, Basis glass states: New insights from the potential energy landscape, J. Non-Cryst. Solids: X 3, 100031 (2019).arxiv.org/abs/1907.10232
  15. [15]M. I. Ojovan and W. E. Lee, Glassy wasteforms for nuclear waste immobilization, Metall. Mater. Trans. A 42, 837 (2011).DOI
  16. [16]R. E. Simpson, P. Fons, A. V. Kolobov, T. Fukaya, M. Krbal, T. Yagi, and J. Tominaga, Interfacial phase-change memory, Nat. Nanotechnol. 6, 501 (2011).DOI
  17. [17]J. R. Jones, Review of bioactive glass: from hench to hybrids, Acta Biomater. 9, 4457 (2013).DOI
  18. [18]U. Buchenau, Neutron and x-ray scattering from glasses, AIP Conf. Proc. 489, 3–23 (1999).DOI
  19. [19]B. S. Hudson, Vibrational spectroscopy using inelastic neutron scattering: Overview and outlook, Vib. Spectrosc. 42, 25 (2006).
  20. [20]U. Buchenau, Y. M. Galperin, V. L. Gurevich, and H. R. Schober, Anharmonic potentials and vibrational localization in glasses, Phys. Rev. B 43, 5039 (1991).pubmed.ncbi.nlm.nih.gov/9997881
  21. [21]V. Gurarie and J. T. Chalker, Bosonic excitations in random media, Phys. Rev. B 68, 134207 (2003).DOI
  22. [22]V. L. Gurevich, D. A. Parshin, and H. R. Schober, Anharmonicity, vibrational instability, and the boson peak in glasses, Phys. Rev. B 67, 094203 (2003).DOI
  23. [23]D. A. Parshin, H. R. Schober, and V. L. Gurevich, Vibrational instability, two-level systems, and the boson peak in glasses, Phys. Rev. B 76, 064206 (2007).DOI
  24. [24]E. Lerner, G. Düring, and E. Bouchbinder, Statistics and properties of low-frequency vibrational modes in structural glasses, Phys. Rev. Lett. 117, 035501 (2016).DOI
  25. [25]H. Mizuno, H. Shiba, and A. Ikeda, Continuum limit of the vibrational properties of amorphous solids, Proc. Natl. Acad. Sci. U. S. A. 114, E9767 (2017).DOI
  26. [26]E. Lerner and E. Bouchbinder, Low-energy quasilocalized excitations in structural glasses, J. Chem. Phys. 155, 200901 (2021).DOI
  27. [27]A. Moriel, E. Lerner, and E. Bouchbinder, Experimental evidence for the ω⁴ tail of the nonphononic spectra of glasses, J. Appl. Phys. 136, 225105 (2024).
  28. [28]A. Kalampounias, S. Yannopoulos, and G. Papatheodorou, A low-frequency raman study of glassy, supercooled and molten silica and the preservation of the boson peak in the equilibrium liquid state, J. Non-Cryst. Solids 352, 4619 (2006).DOI
  29. [29]S. Yannopoulos, K. Andrikopoulos, and G. Ruocco, On the analysis of the vibrational boson peak and low-energy excitations in glasses, J. Non-Cryst. Solids 352, 4541 (2006).DOI
  30. [30]A. Moriel, E. Lerner, and E. Bouchbinder, Boson peak in vibrational spectra of glasses, Phys. Rev. Res. 6, 023053 (2024).DOI
  31. [31]V. Malinovsky and A. Sokolov, The nature of boson peak in raman scattering in glasses, Solid State Commun. 57, 757 (1986).DOI
  32. [32]M. A. A. Ramos, S. Vieira, F. J. Bermejo, J. Dawidowski, H. E. Fischer, H. Schober, M. A. González, C. K. Loong, and D. L. Price, Quantitative assessment of the effects of orientational and positional disorder on glassy dynamics, Phys. Rev. Lett. 78, 82 (1997).DOI
  33. [33]A. Wischnewski, U. Buchenau, A. J. Dianoux, W. A. Kamitakahara, and J. L. Z. Z. and, Neutron scattering analysis of low-frequency modes in silica, Philos. Mag. B 77, 579 (1998).
  34. [34]N. Surovtsev, A. Shebanin, and M. Ramos, Density of states and light-vibration coupling coefficient in B₂O₃ glasses with different thermal history, Phys. Rev. B 67, 024203 (2003).DOI
  35. [35]G. Parisi, On the origin of the boson peak, J. Phys. Condens. Matter 15, S765 (2003).DOI
  36. [36]M. Wyart, S. R. Nagel, and T. A. Witten, Geometric origin of excess low-frequency vibrational modes in weakly connected amorphous solids, Europhys. Lett. 72, 486 (2005).DOI
  37. [37]A. Monaco, A. Chumakov, Y.-Z. Yue, G. Monaco, L. Comez, D. Fioretto, W. Crichton, and R. Rüffer, Density of vibrational states of a hyperquenched glass, Phys. Rev. Lett. 96, 205502 (2006).pubmed.ncbi.nlm.nih.gov/16803182
  38. [38]G. Baldi, A. Fontana, and G. Monaco, Vibrational dynamics of non-crystalline solids, in Low-Temperature Thermal and Vibrational Properties of Disordered Solids: A Half-Century of Universal “Anomalies” of Glasses (World Scientific, 2023) pp. 177–226.arxiv.org/abs/2011.10415
  39. [39]A. Marruzzo, W. Schirmacher, A. Fratalocchi, and G. Ruocco, Heterogeneous shear elasticity of glasses: the origin of the boson peak, Sci. Rep. 3, 1407 (2013).DOI
  40. [40]E. DeGiuli, A. Laversanne-Finot, G. During, E. Lerner, and M. Wyart, Effects of coordination and pressure on sound attenuation, boson peak and elasticity in amorphous solids, Soft Matter 10, 5628 (2014).arxiv.org/abs/1401.6563
  41. [41]M. González-Jiménez, T. Barnard, B. A. Russell, N. V. Tukachev, U. Javornik, L.-A. Hayes, A. J. Farrell, S. Guinane, H. M. Senn, A. J. Smith, M. Wilding, G. Mali, M. Nakano, Y. Miyazaki, P. McMillan, G. C. Sosso, and K. Wynne, Understanding the emergence of the boson peak in molecular glasses, Nat. Commun. 14, 215 (2023).
  42. [42]D. L. Malandro and D. J. Lacks, Relationships of shear-induced changes in the potential energy landscape to the mechanical properties of ductile glasses, J. Chem. Phys. 110, 4593 (1999).DOI
  43. [43]C. Maloney and A. Lemaître, Universal breakdown of elasticity at the onset of material failure, Phys. Rev. Lett. 93, 195501 (2004).DOI
  44. [44]A. Tanguy, B. Mantisi, and M. Tsamados, Vibrational modes as a predictor for plasticity in a model glass, Europhys. Lett. 90, 16004 (2010).DOI
  45. [45]M. L. Manning and A. J. Liu, Vibrational modes identify soft spots in a sheared disordered packing, Phys. Rev. Lett. 107, 108302 (2011).pubmed.ncbi.nlm.nih.gov/21981537
  46. [46]J. Ding, S. Patinet, M. L. Falk, Y. Cheng, and E. Ma, Soft spots and their structural signature in a metallic glass, Proc. Natl. Acad. Sci. U. S. A. 111, 14052 (2014).DOI
  47. [47]E. Lerner, Micromechanics of nonlinear plastic modes, Phys. Rev. E 93, 053004 (2016).DOI
  48. [48]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, and M. L. Manning, Predicting plasticity in disordered solids from structural indicators, Phys. Rev. Mater. 4, 113609 (2020).DOI
  49. [49]D. Fiocco, G. Foffi, and S. Sastry, Oscillatory athermal quasistatic deformation of a model glass, Phys. Rev. E 88, 020301 (2013).pubmed.ncbi.nlm.nih.gov/24032763
  50. [50]I. Regev, T. Lookman, and C. Reichhardt, Onset of irreversibility and chaos in amorphous solids under periodic shear, Phys. Rev. E 88, 062401 (2013).DOI
  51. [51]I. Regev, J. Weber, C. Reichhardt, K. A. Dahmen, and T. Lookman, Reversibility and criticality in amorphous solids, Nat. Commun. 6, 8805 (2015).DOI
  52. [52]T. Kawasaki and L. Bertier, Macroscopic yielding in jammed solids is accompanied by a nonequilibrium first-order transition in particle trajectories, Phys. Rev. E. 94, 022615 (2016).DOI
  53. [53]P. Leishangthem, A. D. S. Parmar, and S. Sastry, The yielding transition in amorphous solids under oscillatory shear deformation, Nat. Commun. 8, 14653 (2017).DOI
  54. [54]Y. Jin, P. Urbani, F. Zamponi, and H. Yoshino, A stability-reversibility map unifies elasticity, plasticity, yielding, and jamming in hard sphere glasses, Sci. Adv. 4, eaati6387 (2018).DOI
  55. [55]A. D. S. Parmar, S. Kumar, and S. Sastry, Strain localization above the yielding point in critically deformed glasses, Phys. Rev. X. 9, 021018 (2019).DOI
  56. [56]E. Schinasi-Lemberg and I. Regev, Annealing and rejuvenation in a two-dimensional model amorphous solid under oscillatory shear, Phys. Rev. E. 101, 012603 (2020).
  57. [57]P. Das, H. A. Vinutha, and S. Sastry, Unified phase diagram of reversible-irreversible, jamming, and yielding transitions in cyclically sheared soft-sphere packings, Proc. Natl. Acad. Sci. U. S. A. 117, 10203 (2020).
  58. [58]W.-T. Yeh, M. Ozawa, K. Miyazaki, T. Kawasaki, and L. Berthier, Glass stability changes the nature of yielding under oscillatory shear, Phys. Rev. Lett. 124, 225502 (2020).DOI
  59. [59]H. Bhaumik, G. Foffi, and S. Sastry, The role of annealing in determining the yielding behavior of glasses under cyclic shear deformation, Proc. Natl. Acad. Sci. U. S. A. 118, e2100227118 (2021).DOI
  60. [60]Y. Goswami, G. V. Shivashankar, and S. Sastry, Yielding behaviour of active particles in bulk and in confinement, Nat. Phys. 21, 817 (2025).
  61. [61]C. Liu, E. E. Ferrero, E. A. Jagla, K. Martens, A. Rosso, and L. Talon, The fate of shear-oscillated amorphous solids, J. Chem. Phys. 156, 104902 (2022).
  62. [62]J. T. Parley, S. Sastry, and P. Sollich, Mean-field theory of yielding under oscillatory shear, Phys. Rev. Lett. 128, 198001 (2022).DOI
  63. [63]J. O. Cochran, G. L. Callaghan, M. J. G. Caven, and S. M. Fielding, Slow fatigue and highly delayed yielding via shear banding in oscillatory shear, Phys. Rev. Lett. 132, 168202 (2024).DOI
  64. [64]E. D. Knowlton, D. J. Pine, and L. Cipelletti, A microscopic view of the yielding transition in concentrated emulsions, Soft Matter 10, 6931 (2014).DOI
  65. [65]K. Hima Nagamanasa, S. Gokhale, A. K. Sood, and R. Ganapathy, Experimental signatures of a nonequilibrium phase transition governing the yielding of a soft glass, Phys. Rev. E 89, 062308 (2014).
  66. [66]A. Ghosh, J. Radhakrishnan, P. M. Chaikin, D. Levine, and S. Ghosh, Coupled dynamical phase transitions in driven disk packings, Phys. Rev. Lett. 129, 188002 (2022).DOI
  67. [67]K. van der Vaart, Y. Rahmani, R. Zargar, Z. Hu, D. Bonn, and P. Schall, Rheology of concentrated soft and hard-sphere suspensions, J. Rheol. 57, 1195 (2013).DOI
  68. [68]D. V. Denisov, M. T. Dang, B. Struth, A. Zaccone, G. H. Wegdam, and P. Schall, Sharp symmetry-change marks the mechanical failure transition of glasses, Sci. Rep. 5, 14359 (2015).DOI
  69. [69]A. Ghosh, Z. Budrikis, V. Chikkadi, A. L. Sellerio, S. Zapperi, and P. Schall, Direct observation of percolation in the yielding transition of colloidal glasses, Phys. Rev. Lett. 118, 148001 (2017).DOI
  70. [70]S. Aime, D. Truzzolillo, D. Pine, L. Ramos, and L. Cipelletti, A unified state diagram for the yielding transition of soft colloids, Nat. Phys. 19, 1673 (2022).arxiv.org/abs/2212.08863
  71. [71]N. C. Keim, J. D. Paulsen, and S. R. Nagel, Multiple transient memories in sheared suspensions: robustness, structure, and routes to plasticity, Phys. Rev. E 88, 032306 (2013).DOI
  72. [72]F. Arceri, E. I. Corwin, and V. F. Hagh, Marginal stability in memory training of jammed solids, Phys. Rev. E 104, 044907 (2021).DOI
  73. [73]S. Mukherji, N. Kandula, A. K. Sood, and R. Ganapathy, Strength of mechanical memories is maximal at the yield point of a soft glass, Phys. Rev. Lett. 122, 158001 (2019).DOI
  74. [74]K. L. Galloway, E. G. Teich, X. G. Ma, C. Kammer, I. R. Graham, N. C. Keim, C. Reina, D. J. Jerolmack, A. G. Yodh, and P. E. Arratia, Relationships between structure, memory and flow in sheared disordered materials, Nat. Phys. 18, 565 (2022).
  75. [75]D. Shohat and Y. Lahini, Dissipation indicates memory formation in driven disordered systems, Phys. Rev. Lett. 130, 048202 (2023).DOI
  76. [76]Y. Chen, S. A. Rogers, S. Narayanan, J. L. Harden, and R. L. Leheny, Microstructural and rheological training and memory of nanocolloid soft glasses under cyclic shear, Phys. Rev. Mater. 9, 025601 (2025).DOI
  77. [77]D. Fiocco, G. Foffi, and S. Sastry, Encoding of memory in sheared amorphous solids, Phys. Rev. Lett. 112, 025702 (2014).pubmed.ncbi.nlm.nih.gov/24484027
  78. [78]M. Adhikari and S. Sastry, Memory formation in cyclically deformed amorphous solids and sphere assemblies, Eur. Phys. J. E 41, 105 (2018).arxiv.org/abs/1805.09113
  79. [79]M. Mungan, D. Kumar, S. Patinet, and D. Vandembroucq, Self-organization and memory in a disordered solid subject to random driving, Phys. Rev. Lett. 134, 178203 (2025).DOI
  80. [80]C. Rainone, P. Urbani, F. Zamponi, E. Lerner, and E. Bouchbinder, Mean-field model of interacting quasilocalized excitations in glasses, SciPost Phys. Core 4, 008 (2021).DOI
  81. [81]E. Bouchbinder, E. Lerner, C. Rainone, P. Urbani, and F. Zamponi, Low-frequency vibrational spectrum of mean-field disordered systems, Phys. Rev. B 103, 174202 (2021).DOI
  82. [82]R. Kühn and U. Horstmann, Random matrix approach to glassy physics: Low temperatures and beyond, Phys. Rev. Lett. 78, 4067 (1997).DOI
  83. [83]G. Folena and P. Urbani, Marginal stability of soft anharmonic mean field spin glasses, J. Stat. Mech. 2022, 053301 (2022).DOI
  84. [84]T. Maimbourg, Two-level systems and harmonic excitations in a mean-field anharmonic quantum glass, Phys. Rev. B 110, 064203 (2024).DOI
  85. [85]E. Bouchbinder and J. S. Langer, Nonequilibrium thermodynamics of driven amorphous materials. III. shear-transformation-zone plasticity, Phys. Rev. E 80, 031133 (2009).pubmed.ncbi.nlm.nih.gov/19905088
  86. [86]M. L. Falk and J. S. Langer, Deformation and failure of amorphous, solidlike materials, Annu. Rev. Condens. Matter Phys. 2, 353 (2011).DOI
  87. [87]P. K. K. Shal, I. Procaccia, C. Rainone, and M. Singh, Mechanical yield in amorphous solids: A first-order phase transition, Phys. Rev. Lett. 116, 085501 (2016).
  88. [88]G. Parisi, I. Procaccia, C. Rainone, and M. Singh, Shear bands as manifestation of a criticality in yielding amorphous solids, Proc. Natl. Acad. Sci. U. S. A. 114, 5577 (2017).DOI
  89. [89]Y. Shi and M. L. Falk, Strain localization and percolation of stable structure in amorphous solids, Phys. Rev. Lett. 95, 095502 (2005).pubmed.ncbi.nlm.nih.gov/16197224
  90. [90]Y. Shi, M. B. Katz, H. Li, and M. L. Falk, Evaluation of the disorder temperature and free-volume formalisms via simulations of shear banding in amorphous solids, Phys. Rev. Lett. 98, 185505 (2007).pubmed.ncbi.nlm.nih.gov/17501586
  91. [91]M. Vasoya, C. H. Rycroft, and E. Bouchbinder, Notch fracture toughness of glasses: Dependence on rate, age, and geometry, Phys. Rev. Appl. 6, 024008 (2016).
  92. [92]M. Ozawa, L. Berthier, G. Biroli, A. Rosso, and G. Tarjus, Random critical point separates brittle and ductile yielding transitions in amorphous materials, Proc. Natl. Acad. Sci. U.S.A. 115, 6656 (2018).DOI
  93. [93]J. Ketkaew, W. Chen, H. Wang, A. Datye, M. Fan, G. Pereira, U. D. Schwarz, Z. Liu, R. Yamada, W. Dmowski, M. D. Shattuck, C. S. O’Hern, T. Egami, E. Bouchbinder, and J. Schroers, Mechanical glass transition revealed by the fracture toughness of metallic glasses, Nat. Commun. 9, 3271 (2018).
  94. [94]D. Richard, E. Lerner, and E. Bouchbinder, Brittle-to-ductile transitions in glasses: Roles of soft defects and loading geometry, MRS Bull. 46, 902 (2021).DOI
  95. [95]D. F. Castellanos, S. Roux, and S. Patinet, History dependent plasticity of glass: A mapping between atomistic and elasto-plastic models, Acta Mater. 241, 118405 (2022).
  96. [96]D. Fiocco, G. Foffi, and S. Sastry, Memory effects in schematic models of glasses subjected to oscillatory deformation, J. Phys. Condens. Matter 27, 194130 (2015).DOI
  97. [97]N. C. Keim and J. D. Paulsen, Multiperiodic orbits from interacting soft spots in cyclically sheared amorphous solids, Sci. Adv. 7, eabg6785 (2021).DOI
  98. [98]C. W. Lindeman, T. R. Jalowiec, and N. C. Keim, Generalizing multiple memories from a single drive: The hysteron latch, Sci. Adv. 11, eadr5933 (2025).DOI
  99. [99]M. Mungan and T. A. Witten, Cyclic annealing as an iterated random map, Phys. Rev. E. 99, 052132 (2019).DOI
  100. [100]M. Mungan and S. Sastry, Metastability as a mechanism for yielding in amorphous solids under cyclic shear, Phys. Rev. Lett. 127, 248002 (2021).DOI
  101. [101]D. Sarkar, J. N. Nampoothiri, M. Mungan, J. T. Parley, P. Sollich, and S. Sastry, Coarse grained descriptions of the dynamics of yielding of amorphous solids under cyclic shear, arXiv [cond-mat.stat-mech] (2025).arxiv.org/abs/2505.14912
  102. [102]C. F. Schreck, R. S. Hoy, M. D. Shattuck, and C. S. O’Hern, Particle-scale reversibility in athermal particulate media below jamming, Phys. Rev. E 88, 052205 (2013).DOI
  103. [103]A. Szulc, M. Mungan, and I. Regev, Cooperative effects driving the multi-periodic dynamics of cyclically sheared amorphous solids, J. Chem. Phys. 156, 164506 (2022).DOI
  104. [104]E. Lerner and E. Bouchbinder, A characteristic energy scale in glasses, J. Chem. Phys. 148, 214502 (2018).DOI

Paper details

vlnc:W_zxpmptbx
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents