Gradient Avalanche Dynamics

• Gradient avalanche dynamics is defined as cascade propagation governed by explicit local slope thresholds as well as emergent, heterogeneous distance-to-threshold fields.
• The framework unites diverse models—including the Kadanoff Sand Pile, depinning interfaces, and excitable networks—to reveal deterministic rules and statistical scaling in avalanche behavior.
• Studies highlight practical insights such as finite-size scaling, temporal asymmetry in burst profiles, and dynamically regenerated instability fields, which are crucial for analyzing material failure.

Gradient avalanche dynamics, as represented in the current literature, denotes avalanche propagation governed either directly by thresholded local gradients or indirectly by spatial variation in local distance-to-threshold, self-generated stress heterogeneity, rate-controlled switching, or dynamically evolving unstable response fields. The resulting corpus is not a single canonical theory but a family of closely related formalisms spanning sandpiles, elastic interfaces, heterogeneous threshold models, excitable networks, sheared amorphous matter, and architected metamaterials (Perrot et al., 2011, Zhu et al., 2017, Biswas et al., 2018, Jin et al., 3 Jul 2025).

1. Scope, observables, and modes of relevance

Across these models, an avalanche is a threshold-activated cascade with a well-defined onset, a finite or critical propagation regime, and eventual arrest. The primary observables differ by subfield: avalanche size may be the number of failed fibers $\Delta$ in a fiber bundle (Danku et al., 2019), the number of failed sites $S$ in a long-range threshold model (Biswas et al., 2018), the integrated local advance $S(x)=u_2(x)-u_1(x)$ of a pinned interface (Zhu et al., 2017), or the spatially integrated instantaneous velocity $\dot u(t)$ in depinning (Kolton et al., 2019). Duration is denoted $W$ in localized-load-sharing fracture (Danku et al., 2019) and $T$ in network and depinning formulations (Gleeson et al., 2016, Dobrinevski et al., 2014).

The literature separates naturally into direct gradient-threshold models, models of imposed or embedded heterogeneity, and models where gradients are emergent rather than externally prescribed. That distinction is essential because several influential papers are relevant only indirectly to gradient dynamics: they analyze self-generated heterogeneous stress fields or spatially extended statistical signatures rather than an explicit gradient field (Danku et al., 2019, Biswas et al., 2018).

Mode of relevance	Representative formulations	Salient control
Direct local gradients	KSPM slopes $\sigma_i=h_i-h_{i+1}$ (Perrot et al., 2011); edge profile $S(x)$ (Zhu et al., 2017)	local slope, boundary scaling
Spatial heterogeneity	embedded patch with $s_1$ vs $s_2$ (Biswas et al., 2018); mean-field gain $S$0 (Rastegar et al., 2019)	distance-to-threshold, branching gain
Emergent or dynamic gradients	localized load sharing (Danku et al., 2019); race conditions (Jin et al., 3 Jul 2025); unstable soft spots (Stanifer et al., 2021)	stress concentration, switching time, unstable spectrum

2. Direct gradient-threshold dynamics and spatial front structure

The most explicit gradient-threshold formulation appears in the one-dimensional Kadanoff Sand Pile Model. There the preferred variables are slopes

$S$1

and column $S$2 is firable iff $S$3. In gradient variables, a firing obeys

$S$4

This makes avalanche propagation a directly gradient-threshold phenomenon: motion is controlled by local slope excess, not absolute height (Perrot et al., 2011).

Several structural results in KSPM are especially important. In a single avalanche each column fires at most once. Avalanches advance through peaks, defined as firings that extend the rightmost reached column, and then backfill holes behind the front. Beyond a dense fired block of $S$5 consecutive columns, propagation becomes pseudo-local: for $S$6,

$S$7

Thus a new front advance occurs exactly where the pre-avalanche slope is one below threshold. For $S$8, the transient region has size $S$9, after which peaks are exactly the sites with slope $S(x)=u_2(x)-u_1(x)$0 and the remainder of the avalanche is a deterministic local backfilling sequence (Perrot et al., 2011).

A complementary spatial perspective is provided by the Brownian Force Model analysis of avalanche shape at fixed extent $S(x)=u_2(x)-u_1(x)$1. For an interface avalanche with local advance $S(x)=u_2(x)-u_1(x)$2, the mean shape satisfies

$S(x)=u_2(x)-u_1(x)$3

Near the avalanche boundary,

$S(x)=u_2(x)-u_1(x)$4

and in the $S(x)=u_2(x)-u_1(x)$5 BFM, where $S(x)=u_2(x)-u_1(x)$6,

$S(x)=u_2(x)-u_1(x)$7

The integrated displacement profile therefore turns on smoothly at the edge, with vanishing first derivative there; the front is not cusp-like in $S(x)=u_2(x)-u_1(x)$8 but has a universal cubic boundary layer in the BFM (Zhu et al., 2017).

Taken together, these works identify two distinct notions of gradient in avalanche dynamics. In KSPM, the gradient is the primitive threshold variable itself. In the BFM spatial-shape problem, the salient object is the spatial profile of integrated motion and its universal edge regularity. The first is a local instability rule; the second is a continuum spatial signature of the resulting burst.

3. Heterogeneity, local distance-to-threshold, and nonlocal statistical signatures

A different route to gradient avalanche dynamics arises when avalanche statistics respond to spatial variation in local margin to failure. In the heterogeneity-mapping model on an $S(x)=u_2(x)-u_1(x)$9 square lattice, thresholds are assigned by

$\dot u(t)$0

with $\dot u(t)$1. The parameter $\dot u(t)$2 is the excess capacity above load: smaller $\dot u(t)$3 means the system is closer to failure. Load redistribution is long-ranged, with kernel proportional to $\dot u(t)$4, and the cumulative avalanche-size distribution obeys

$\dot u(t)$5

As $\dot u(t)$6 decreases, $\dot u(t)$7 decreases, so larger avalanches become relatively more probable (Biswas et al., 2018).

The model does not impose a smooth field $\dot u(t)$8; instead it embeds a square inclusion with excess capacity $\dot u(t)$9 in a background $W$0, with contrast $W$1. Even this sharp defect generates a spatially extended crossover in the measured exponent,

$W$2

because long-range redistribution makes the observed local statistics nonlocal averages rather than pointwise indicators. The detection time of the heterogeneity scales as

$W$3

with distinct exponent pairs for weaker and stronger inclusions: $W$4

$W$5

and

$W$6

The asymmetry between weak and strong anomalies is therefore intrinsic to the detection problem (Biswas et al., 2018).

A mathematically cleaner, though nonspatial, version of state-dependent avalanche amplification appears in the homogeneous excitable-network model equivalent to Longini’s endemic Reed-Frost process. The exact update law is

$W$7

and under $W$8, the normalized activity $W$9 converges to

$T$0

Near zero, $T$1, so $T$2 is the threshold separating decay from growth (Rastegar et al., 2019). This suggests a generalized gradient interpretation in which a local amplification field would replace the scalar gain $T$3, although that extension is not a result of the paper.

4. Self-generated stress gradients, localized load sharing, and externally controlled cutoffs

In higher-dimensional localized-load-sharing fracture, no explicit gradient is imposed. The fiber-bundle model uses quenched exponential thresholds,

$T$4

with perfectly brittle fibers and nearest-neighbor redistribution on hypercubic lattices from $T$5 to $T$6. The coordination number

$T$7

grows with dimension, so the same dropped load is spread more thinly as $T$8 increases. The consequence is a continuous crossover from localized-load-sharing behavior to the mean-field equal-load-sharing class (Danku et al., 2019).

Macroscopic and avalanche observables approach mean-field exponentially with dimension. The fracture stress satisfies

$T$9

with

$\sigma_i=h_i-h_{i+1}$0

and the burst-size exponent obeys

$\sigma_i=h_i-h_{i+1}$1

with

$\sigma_i=h_i-h_{i+1}$2

The temporal profile evolves from strongly right-skewed at low $\sigma_i=h_i-h_{i+1}$3 to a symmetric parabola at high $\sigma_i=h_i-h_{i+1}$4. The crucial point for gradient dynamics is that this asymmetry is attributed not to an imposed gradient but to self-organized stress concentration and heterogeneous stress fields along crack perimeters (Danku et al., 2019).

An externally controlled cutoff appears in avalanche-driven interface growth. In forced-flow imbibition, the characteristic correlation length obeys

$\sigma_i=h_i-h_{i+1}$5

so decreasing the mean interface velocity $\sigma_i=h_i-h_{i+1}$6 increases the avalanche extent and drives the system toward critical, scale-free activity. The same scaling theory links avalanche kinetics,

$\sigma_i=h_i-h_{i+1}$7

to local activity statistics and global velocity correlations (Lopez et al., 2010). In this setting, the “gradient” is not a local slope threshold but the driving condition that sets the finite-scale cutoff.

These two cases delimit an important conceptual boundary. Localized fracture shows how gradient-like heterogeneity can be generated internally by redistribution, whereas imbibition shows how a control parameter can impose a cutoff length on avalanche propagation without specifying a local gradient field.

5. Temporal profiles, velocity fields, and asymmetry classes

Temporal profile analysis demonstrates that avalanche dynamics cannot be reduced to size distributions alone. In branching-process theory for network cascades, the average shape of avalanches of fixed duration $\sigma_i=h_i-h_{i+1}$8 is

$\sigma_i=h_i-h_{i+1}$9

where $S(x)$0 is the offspring generating function and $S(x)$1 is the extinction probability. At criticality, if $S(x)$2, the shape is asymptotically symmetric and parabolic; if instead

$S(x)$3

then

$S(x)$4

with peak at

$S(x)$5

so the profile is left-skewed (Gleeson et al., 2016).

By contrast, the one-loop depinning theory beyond mean field predicts a fixed-duration center-of-mass shape

$S(x)$6

with

$S(x)$7

Near $S(x)$8, $S(x)$9, which skews the avalanche toward its end rather than toward its beginning (Dobrinevski et al., 2014).

Mean-field elastic-interface theory adds a third layer. At or above upper critical dimension, the center of mass reduces to the ABBM model, while the full interface is described by the Brownian-force model. Tree-level theory yields explicit PDFs for the instantaneous velocity and exact mean-field shape functions, and finite-$s_1$0 correlations are asymmetric under time reversal even when center-of-mass observables are not (Doussal et al., 2013). On the velocity side, functional-renormalization and numerical work in $s_1$1 short-range elasticity give

$s_1$2

with measured

$s_1$3

in excellent agreement with the prediction $s_1$4 (Kolton et al., 2019).

Frequency-domain analysis in sheared disordered solids shows yet another decomposition of temporal structure. The power spectrum of kinetic energy has three regimes: a low-frequency power-law rise $s_1$5 indicating anticorrelation, an intermediate white-noise plateau, and a high-frequency decay $s_1$6 reflecting intra-avalanche structure. At finite strain rate, the largest avalanche scale is controlled by

$s_1$7

and

$s_1$8

(Clemmer et al., 2021).

Taken together, these results suggest that temporal asymmetry is mechanism-dependent rather than sign-universal. Right-skew in localized fracture, left-skew in heavy-tailed branching, end-skew in one-loop depinning, and finite-$s_1$9 mean-field asymmetry are distinct phenomena generated by different microscopic or mesoscopic constraints.

6. Dynamic pathway selection, race conditions, and evolving local susceptibility

A qualitatively different class of gradient-like avalanche dynamics appears when switching times, damping, and propagation delays select the avalanche pathway. In serially coupled hysteretic metamaterials, the state is $s_2$0 with binary element states, and static interacting-hysteron models assign state-dependent switching thresholds $s_2$1. The paper shows that this static picture is incomplete. In a two-element system, the gap

$s_2$2

would forbid an avalanche for $s_2$3 in a static sequential model, but low-damping experiments exhibit avalanches even when $s_2$4, because underdamped overshoot drives the trajectory past an intermediate statically stable state (Jin et al., 3 Jul 2025).

The dynamic equations explicitly include masses and damping,

$s_2$5

$s_2$6

In three-element systems, race conditions arise when one flip destabilizes multiple elements nearly simultaneously. Changing only the damping of one branch can redirect the avalanche, for example from

$s_2$7

to

$s_2$8

The designed threshold hierarchy

$s_2$9

therefore selects what can become unstable, while damping, switching time, and propagation delay select what actually flips first (Jin et al., 3 Jul 2025).

A closely related but particle-resolved picture emerges in sheared athermal packings. There, avalanches are decomposed into localized bursts of nonaffine motion using a space-time persistent-homology-inspired clustering of

$S$00

These bursts account for

$S$01

of the nonaffine motion while occupying only

$S$02

of spacetime volume; their mean duration is $S$03 natural time units (Stanifer et al., 2021).

The unstable Hessian during an avalanche has a rapidly changing spectrum with as many as $S$04 or $S$05 negative eigenvalues, so the most unstable eigenvector is generally a poor predictor of the actual dynamics. A modified unstable-system indicator, non-affine vibrality $S$06, built from the nonaffine content of the lowest few eigenmodes without $S$07 weighting, identifies soft spots that overlap

$S$08

of localized bursts. Only

$S$09

of soft spots already exist before avalanche onset; most are generated during the avalanche itself (Stanifer et al., 2021).

These two studies show that gradient avalanche dynamics can be controlled by dynamically evolving timing and susceptibility fields, not merely by static threshold landscapes. A plausible implication is that propagation in many disordered systems is governed by a continuously regenerated local instability field.

7. Exact combinatorial results, critical exponents, and conceptual limits

Exact counting results provide a complementary backbone for the subject. Denker and Rodrigues show that avalanche histories can be indexed by compositions $S$10 of the total size into generations, with the central identity

$S$11

This yields exact finite-$S$12 avalanche-size laws and the critical asymptotic

$S$13

for the homogeneous case (Denker et al., 2011).

A rigorous first-wave counterpart appears in the Abelian sandpile on the expanded cactus graph. Using the Dhar-Majumdar framework together with a filling method for recurrent configurations, the cell-wise first-wave avalanche probability obeys

$S$14

The result is exact for threshold-driven transport on a branching geometry with local loops, but it is not a continuum gradient theory; the relevant transport variable is a toppling threshold rather than a smooth field gradient (Gauthier, 2011).

The main conceptual limit of the broader literature is terminological. Several widely cited contributions are only indirectly about gradients: the higher-dimensional fiber-bundle study imposes no explicit spatial gradient, and the heterogeneity-mapping study uses a localized inclusion rather than a smooth field (Danku et al., 2019, Biswas et al., 2018). Likewise, temporal-profile and depinning papers often resolve velocity, duration, or spatial shape without specifying an external gradient field. This suggests that “gradient avalanche dynamics” is best understood as an umbrella for multiple mechanisms—direct slope thresholds, spatially varying distance-to-threshold, self-generated stress concentration, rate-controlled switching, and dynamically regenerated soft spots—rather than as a single established universality class.