Avalanche Size and Interevent Time Statistics in Amorphous Systems

arXiv:2609.24462 · cond-mat.stat-mech · Submitted 2026-09-21 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.

Kai: We're looking at the paper: "Avalanche Size and Interevent Time Statistics in Amorphous Systems".

Mira: Here's the gist of the paper. The intermittent, stick-slip flow of amorphous systems is characterized by avalanches whose statistical properties display robust power-laws.

Kai: This paper introduces a simple model for avalanche dynamics in soft-glassy systems where both exponents α and γ can be computed theoretically.

Mira: The model successfully reproduces the observed scaling properties of many amorphous systems.

Kai: Model Formulation The study utilizes a field theoretical model of a soft-glassy system under shear, where the order parameter is the fluidity related to the local rate of plastic events.

Title and authors: Kai: So we're looking at this paper called "Avalanche Size and Interevent Time Statistics in Amorphous Systems". It's written by Thomas Muhren, Roberto Benzi, Mauro Sbragaglia, Jeannot Trampert, and Federico Toschi.

Mira: That sounds like a fairly broad title for something that is actually quite specific to soft-glassy systems under shear. We need to figure out what this paper is actually building on.

Lev: From what I can see from the initial setup, they are using a field theoretical model of a soft-glassy system under shear, where the order parameter they're tracking is the fluidity, which relates to how fast plastic events happen locally.

Kai: Right, so it’s not just some abstract math; it’s rooted in modeling real physical processes in these complex materials. The authors are trying to get a better handle on the statistics of avalanches in things like amorphous solids or soft-glasses.

Mira: And they're focusing on two key statistics: the size of the avalanche, S, and the time between those events, ti. It seems their main goal is to link these two distributions together through scaling laws.

Lev: I'm interested in what this means for running simulations on real hardware. If this model works well, it suggests that we can predict scaling properties for stress overshoot as a function of some parameter called gamma-dot-one-eight, which is important when you're testing things experimentally <ref:2609.24462#pg2>.

Kai: So the paper is suggesting a way to calculate these exponents theoretically, instead of just guessing them from data. It’s about deriving the scaling laws for both S and ti directly from their equations.

Mira: And that’s what’s interesting because it allows them to test these relationships across different values of delta, which seems like a very useful parameter in this context.

The paper's summary: Kai: So the core finding here is that they have a simple model for avalanche dynamics where both exponents, alpha and gamma, can be computed theoretically from the equations of motion they set up. They claim this model successfully reproduces the scaling properties observed in many different amorphous systems.

Mira: They predict that the probability distribution for avalanche size S follows a power law PS proportional to S to the negative power of five-over-four plus delta over two, which is written as one over S to the power of five-over four plus delta over two, holding in the limit of large avalanches and ignoring exponential cutoffs.

Lev: That specific prediction for alpha, alpha equals five-over four plus delta over two, seems like a concrete target. But I wonder if that holds up when you introduce the noise terms they have in there.

Kai: The paper addresses that by incorporating microstructural fluctuations through multiplicative noise, which they say is crucial for triggering those abrupt avalanche-like fluidization events. They are using a field theoretical model involving stress and fluidity variables with terms like the Maxwell equation and a nonlinear diffusion equation to describe the system's behavior <ref:2609.24462#pg1>.

Mira: And they also look at the interevent time ti, finding that its distribution has two possible limiting situations depending on how big some other parameter, sigma-zero, compares to another term involving ti and gamma-dot-tau.

Lev: I need to know what those two limits actually mean for the scaling of ti. Is it just a simple power law in both cases?

Kai: They find that in case one, when sigma zero is much bigger than the term involving ti and gamma-dot-tau, the probability distribution for interevent time ti follows a power law proportional to t to the negative power of three-over two.

Mira: But in case two, when sigma zero is much smaller than that term, the distribution for interevent time ti becomes proportional to t to the negative power of one-half minus delta over gamma dot tau.

The paper's improvements: Lev: So they’ve got alpha and gamma derived from those limits, but the real meat seems to be in how they connect them. They find a very specific scaling relation that links alpha and gamma together.

Kai: Exactly, and this is the part I think is most significant for understanding these systems. They show that the two scaling exponents, alpha equals five-over four plus delta over two for the size distribution S, and gamma equals one-half minus delta for the interevent time ti, satisfy a relation where gamma equals three minus two times alpha.

Mira: That relationship is derived by checking that a variable called chi which is one over t to the power of two times i has the exact same probability distribution as S with extremely good agreement.

Lev: If chi has the same distribution as S, then this scaling relation for gamma must be robust because it’s linked to alpha through that shared variable. But what does that tell us about the coarse-graining transformation?

Kai: It implies that the scaling properties of Ptilambda s zero are invariant when you coarse-grain in avalanche size, which means the exponent gamma should be linked directly to alpha.

Mira: They also point out that this is different from what happens when you change the threshold Mt used to define an avalanche. Under a change in Mt, the functional form of PtiMt changes when lambda s zero is much bigger than one, becoming exponential in that limit <ref:2609.24462#pg3>.

Conclusion: Kai: So to wrap up the main points of "Avalanche Size and Interevent Time Statistics in Amorphous Systems," this paper provides a simple model where both avalanche size and interevent times are power-law distributed, and it successfully links their scaling exponents through the relation gamma equals three minus two alpha.

Mira: They show that this relationship is robust because they verify it by showing that the variable chi has the same probability distribution as S, which gives them strong evidence for linking gamma to alpha.

Lev: From an engineering standpoint, if we're building hardware, knowing these exponents allows us to predict how the statistics behave when you change the threshold or when you coarse-grain data. It makes predicting long-tail behavior in interevent times much more tractable.

Kai: It confirms that this Gamma distribution accurately captures the behavior of Ptilambda s zero in the large sigma zero regime, which is what they needed to do for many physical systems.

Mira: This model is built from just a Maxwell element and a dynamical equation for the relaxation time, which is quite fundamental. It’s a nice way to derive these scaling properties without having to postulate them upfront.

Lev: The paper notes that the analysis of interevent times in granular systems shows no long tail when flow is possible, which is an important piece of information for those applications where you're dealing with sand piles or similar things <ref:2609.24462#pg3>.

Kai: So this work gives us a concrete theoretical framework for understanding the statistics of stick-slip flow in these amorphous systems and how size and time are related. We'll keep an eye on how this model performs when we start applying it to more complex spatial models in the future.

Department of Physics, Eindhoven University of Technology · Sino-Europe Complex Science Center, School of Mathematics, North University of China · Department of Physics, University of Tor Vergata and INFN · Department of Earth Sciences, Utrecht University

cond-mat.stat-mech

Submitted: 2026-09-21

Updated: 2026-11-10

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 83/100

The gist: The intermittent, stick-slip flow of amorphous systems is characterized by avalanches whose statistical properties display robust power-laws.

Key concepts

Avalanche Size Distribution P[S]
This describes the probability of observing an avalanche of a certain size. The model predicts this follows a power law, $P[S] \sim S^{-(5/4 + \delta/2)}$, which is characteristic of critical phenomena in systems like earthquakes or sand piles.
Interevent Time Distribution P[ti]
This describes the probability of the time interval between two successive avalanches. The model shows this also follows a power law, $P[t_i] \sim t_i^{-(1/2 - \delta)}$, and its scaling exponent $\gamma$ is directly related to the size exponent $\alpha$ via $\gamma = 1/2 - \delta.
Scaling Relation γ = 3 - 2α
This is a key finding where the exponents governing avalanche size ($\alpha$) and interevent time ($\gamma$) are mathematically connected. This relationship confirms the consistency of the model across different statistical measures, suggesting a deep underlying physical link between how large events occur and how often they occur.
Fluidity as Order Parameter
In this model for soft-glassy systems, the 'fluidity' is used as the main variable describing the system's state. This fluidity relates directly to how fast local plastic events are happening, serving as a measure of the material's internal activity and driving the avalanche dynamics.

Terminology

Summary

The intermittent, stick-slip flow of amorphous systems is characterized by avalanches whose statistical properties display robust power-laws. p. 0 This paper introduces a simple model for avalanche dynamics in soft-glassy systems where both exponents α and γ can be computed theoretically. p. 0 The model successfully reproduces the observed scaling properties of many amorphous systems. p. 0

Model Formulation

The study utilizes a field theoretical model of a soft-glassy system under shear, where the order parameter is the fluidity related to the local rate of plastic events. The equations of motion are presented in terms of stress and fluidity, involving terms like the Maxwell equation (Eq. p. 0 1) and a nonlinear diffusion equation (Eq. p. 1 3). The model incorporates microstructural fluctuations through multiplicative noise, which is crucial for triggering abrupt, avalanche-like fluidization. p. 1

Avalanche Size Statistics

The probability distribution of the avalanche size is found to be P[S] ∼ 1/S(5/4+δ/2) in the limit of large avalanches and neglecting exponential cut-off effects. This prediction matches numerical results for various values of δ, including the mean-field prediction P[S] ∼ S(-1. p. 25) [3, 4, 8]. The scaling exponent α is found to be 5/4 + δ/2. p. 0

Interevent Time Statistics

The probability distribution of the interevent time ti exhibits two limiting possible situations: a) when Σ0 ≫ tiΓ˙τ and b) when Σ0 ≪ tiΓ˙τ. In case a), the probability distribution is given by Pa[ti] ∼ 1/t(3/2) i. p. 6 In case b), the distribution is derived as Pb[ti] ∼ 1/t(1/2-δ) i. p. 9 The scaling exponent γ for the interevent time ti is found to be 1/2 - δ. p. 8

Scaling Relations and Robustness

A remarkable finding is that the two scaling exponents α = 5/4 + δ/2 for the size distribution S and γ = 1/2 - δ for the interevent time ti satisfy the relation γ = 3 - 2α. This relation is directly checked by observing that the variable χ ≡ 1/t 2i has the same probability distribution as S with extremely good agreement. p. 3 The scaling properties of P[tiλS0] are invariant under coarse-graining in avalanche size, which implies that the scaling exponent γ should be linked to the scaling exponent α. p. 0

Generalizability and Conclusion

The model demonstrates that the scaling properties of both avalanche size distribution and interevent times depend solely on a single parameter δ. The analysis confirms that the Gamma distribution accurately captures the behavior of P[tiλS0] in the large λS0 regime. p. 3 The results show excellent agreement with experimental observations and numerical simulations of granular systems and soft-glasses. p. 3 This model is the first to derive – rather than postulate – the Gamma distribution of interevent times observed across diverse physical systems. p. 3

Supplementary Material Details

The supplementary material details the derivation of the correlation length lc = p/l0(1-1/(2δ)) (Eq. 4). The critical value δ = 1/2 corresponds to a point where large fluidity fluctuations start happening. p. 8 Furthermore, the analysis shows that the probability distribution P[tiMt] is not invariant when the definition of the avalanche is changed, suggesting that changing from Mt → λMt is not the same kind of coarse-graining transformation as seen in the main part. p. 9 The spatial model results show that for granular systems, no long tail is present in interevent time statistics when flow is possible. p. 8 The analysis confirms that the same analysis as done for the simplified fluidity model can be generalized when taking spatial fluctuations in the fluidity into account. p. 8 The study concludes by showing that these results are in qualitative and almost quantitative agreement with what is observed in experimental observations and numerical simulations of granular systems and soft-glasses. p. 3 The model is derived from just two fundamental ingredients: a Maxwell element and a dynamical equation for the relaxation time. p. 4 The paper is supported by numerous references including [1] through [24]. p. 1 The acknowledgments include support from DeepNL and INFN through project FIELDTURB. p. 4

--- Page 1 ---

Avalanche Size and Interevent Time Statistics in Amorphous Systems

Thomas Muhren1, Roberto Benzi2,3† Mauro Sbragaglia3†, Jeannot Trampert4, and Federico Toschi1.

Department of Physics, Eindhoven University of Technology. p. 0

Sino-Europe Complex Science Center, School of Mathematics, North University of China, Shanxi 030051, Taiyuan, China. p. 0

Department of Physics, University of Tor Vergata and INFN. p. 0

Department of Earth Sciences, Utrecht University. p. 0

(Dated: March 2026)

The intermittent, stick-slip flow of amorphous systems is characterized by avalanches whose statistical properties display robust power-laws. In this paper, we study avalanche statistics in a field theoretical model of a soft-glassy system under shear. p. 0 The order parameter in our model is the fluidity, which is related to the local rate of plastic events in a soft-glass. p. 0 While avalanche sizes S have been extensively characterized, the statistics of interevent times ti have remained largely unexplored. p. 0 Here, we find that both S and ti are power-law distributed and that their scaling exponents satisfy a precise relation. p. 0 The model enables the analytical derivation of the scaling exponents and the reproduction of the behavior observed across a wide class of amorphous systems. p. 0

The dynamics of avalanches can be observed in many different physical systems. p. 0 p. 0 Examples include amorphous solids, soft-glasses, sand piles, solar flares, earthquakes and the random-field Ising model [1–5]. p. 0 p. 0 The peculiar signature common to all of these systems is the power-law behavior of the avalanche size S distribution P[S] ∼ S(-α). p. 0 p. 0 where usually S is defined using some threshold Mt depending on the physical property of the system [6, 7]. p. 0 p. 0 The exponent α depends on the dimensionality of the system and, obviously, on the physical mechanism underlying the dynamics [8]. p. 0 p. 0 Many theoretical studies have been conducted to compute this exponent, and many investigations have identified power-law behavior as criticality, following the pioneering work on self-organized criticality by Bak et al. p. 0 p. 0 [9]. Another interesting quantity is the avalanche duration time, tE, which is always correlated with avalanche size. p. 0 p. 0 Few approaches have been developed to study the interevent time ti between two successive avalanches. p. 0 p. 0 Within the framework of self-organized criticality, ti is independent of the avalanche size, S, and is Poisson distributed [9]. p. 0 p. 0 Conversely, experimental and numerical results have shown that ti can exhibit a powerlaw distribution [6, 7, 10]. p. 0 p. 0 A notable finding across many studies is that the power-law behavior of the probability distribution of ti remains unchanged by coarse-graining in avalanche size [11, 12]. p. 0 This means that by considering only events where S > λS0, we can compute the interevent time distribution, denoted by P[tiλS0], which exhibits power-law behavior with the same exponent as for λ = 1. p. 0 p. 0 This differs from the statistics of ti as the threshold Mt changes, where the functional form of the probability distribution P[tiMt] changes under the transformation Mt → λMt, becoming exponential in the limit λ ≫ 1 [13]. p. 0 p. 0 In a recent paper [14] it has been argued that, if the scaling P[tiλS0] ∼ t−γi is invariant for λ > 1, then the scaling exponent γ should be linked to the scaling exponent α of avalanche distribution, in excellent agreement with published results [11, 12]. p. 0 Yet, no theoretical approach has been developed to compute the exponent γ. p. 0 p. 0

--- Page 2 ---

dΣ/dt = 1/τΓ˙ −⟨ϕ 2⟩Σ (1)

∂ϕ/∂t = 2ξ 2ϕ 2∆ϕ + 2ξ 2ϕ(∇phi) squared + mphi cubed − phi 3phi (Eq. 5)

+√ϵ0η1(y, t) + pϵ1⟨ϕ 2⟩η t (Eq. 6)

where Σ is the space averaged stress, and σy is the yield stress of the system. In Eq. (1) corresponds to the Maxwell equation which describes a viscoelastic material with a stress relaxation part whose characteristic inverse time comes from another dedicated Eq. p. 0 (2). In the latter equation, ξ is the cooperative length related to the non-locality of the flow, η1(y, t) is a delta correlated Gaussian noise in space and time and η2(t) is a Gaussian noise delta correlated in time. p. 0

--- Page 3 ---

fine the dimensionless stress as Σ ≡ σ/σy, where σ is the space averaged stress, and σy is the yield stress of the system. In Eq. (2) the quantity m is defined as m 2(Σ) = (Σ-1)/n Σ Θ(Σ − 1), with Θ being the Heaviside function and n the Herschel-Bulkley exponent. p. 1 In this continuum framework, the stochastic term proportional to epsilon0 accounts for microstructural fluctuations. p. 1 Crucially, long-range correlations are enhanced by the multiplicative noise through the epsilon1-dependent term, triggering abrupt, avalanche-like fluidization. p. 1

--- Page 7 ---

To obtain an analytical prediction for the probability distribution P[ϕ, t], we can use the Fokker-Planck equation related to the SDE given in Eq. (S19). To obtain a probability distribution in terms of f = ϕ squared, we do a change of variables, obtaining P[f] = Cbar f(1/2)(1 + Af)(1+δ) exp (N(f)). p. 7 For f ≪ 1/A, we find that P[f] ∼ 1/f(1/2) which is independent of the parameters used. p. 7 For f >> 1/A, we find, in the limit of small epsilon0, P[f] ∼ 1/f(3/2+δ) exp (m f(-2/3) epsilon1 + O(epsilon0)). p. 7

--- Page 8 ---

FIG. S1. Probability density P[f] for the simplified fluidity model described by Eqs. p. 7 (1) and (5) for different values of δ. p. 2 Black dashed lines are the analytical prediction given by Eq. p. 7 (S20).

--- Page 9 ---

FIG. S3. Probability density P[tiλS0] for δ = 1/2 and different values of λS0, for the model described by Eqs. p. 7 (1) and (5). The black dashed line is the Gamma distribution with γ = 0. p. 70 (exponential). p. 0

--- Page 10 ---

FIG. S4. A quantitative check for our theoretical approach. p. 8 The same slope of −1. p. 833 can be observed between ti ∼ 10−3 and 10−1, agreeing with the theory. p. 8

The dynamics of avalanches can be observed in many different physical systems. Examples include amorphous solids, soft-glasses, sand piles, solar flares, earthquakes and the random-field Ising model [1–5]. The peculiar signature common to all of these systems is the power-law behavior of the avalanche size S distribution P[S] ∼ S(-α). where usually S is defined using some threshold Mt depending on the physical property of the system [6, 7]. The exponent α depends on the dimensionality of the system and, obviously, on the physical mechanism underlying the dynamics [8]. Many theoretical studies have been conducted to compute this exponent, and many investigations have identified power-law behavior as criticality, following the pioneering work on self-organized criticality by Bak et al. [9]. Another interesting quantity is the avalanche duration time, tE, which is always correlated with avalanche size. Few approaches have been developed to study the interevent time ti between two successive avalanches. Within the framework of self-organized criticality, ti is independent of the avalanche size, S, and is Poisson distributed [9]. Conversely, experimental and numerical results have shown that ti can exhibit a powerlaw distribution [6, 7, 10]. A notable finding across many studies is that the power-law behavior of the probability distribution of ti remains unchanged by coarse-graining in avalanche size [11, 2609. p. 024463pg1]. This means that by considering only events where S > λS0, we can compute the interevent time distribution, denoted by P[tiλS0], which exhibits power-law behavior with the same exponent as for λ = 1. This differs from the statistics of ti as the threshold Mt changes, where the functional form of the probability distribution P[tiMt] changes under the transformation Mt → λMt, becoming exponential in the limit λ ≫ 1 [13]. In a recent paper [14] it has been argued that, if the scaling P[tiλS0] ∼ t−γi is invariant for λ > 1, then the scaling exponent γ should be linked to the scaling exponent α of avalanche distribution, in excellent agreement with published results [11, 2609. p. 024463pg1]. Yet, no theoretical approach has been developed to compute the exponent γ.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. AI systems can predict avalanche size distributions S using a formula derived from Eq (8) which is P[S] ∼ 1/S 5/4+δ/2 in the limit of large avalanches and neglecting the exponential cut-off.

  2. AI systems can predict interevent time distributions ti by applying the scaling relation gamma = 3 - 2alpha, which links the interevent time exponent gamma to the avalanche size exponent alpha.

  3. AI systems can determine if a coarse-graining transformation on avalanche size S is equivalent to changing the threshold Mt, as suggested by Fig S9 showing that P[ti|Mt] is not invariant under this specific change in definition.

  4. AI systems can generalize the analysis of interevent times to spatial dependent models by using the time averaged value delta, which allows for a consistent prediction of avalanche sizes and long-tail behavior in the interevent time statistics.

Related papers