Avalanche shape and exponents beyond mean-field theory

TL;DR

This paper analytically computes the first-order correction to avalanche shapes in elastic systems beyond mean-field theory, revealing asymmetries and dependencies on critical exponents near the upper critical dimension.

Contribution

It provides the first analytical 1-loop correction to avalanche shapes at fixed duration for elastic systems, extending mean-field predictions to finite dimensions.

Findings

Asymmetry in avalanche shape skewed towards the end.

Analytical expression approximates the shape well across dimensions.

Shape at fixed size introduced and calculated.

Abstract

Elastic systems, such as magnetic domain walls, density waves, contact lines, and cracks, are all pinned by substrate disorder. When driven, they move via successive jumps called avalanches, with power law distributions of size, duration and velocity. Their exponents, and the shape of an avalanche, defined as its mean velocity as function of time, have recently been studied. They are known approximatively from experiments and simulations, and were predicted from mean-field models, such as the Brownian force model (BFM), where each point of the elastic interface sees a force field which itself is a random walk. As we showed in EPL 97 (2012) 46004, the BFM is the starting point for an $\epsilon = d_{\rm c}-d$ expansion around the upper critical dimension, with $d_{\rm c}=4$ for short-ranged elasticity, and $d_{\rm c}=2$ for long-ranged elasticity. Here we calculate analytically the ${\cal O}(\epsilon)$, i.e. 1-loop, correction to the avalanche shape at fixed duration $T$, for both types of elasticity. The exact expression is well approximated by $\left< \dot u(t=x T)\right>_T\simeq [ Tx(1-x)]^{\gamma-1} \exp\left( {\cal A}\left[\frac12-x\right]\right)$, $0<x<1$. The asymmetry ${\cal A}\approx - 0.336 (1-d/d_{\rm c})$ is negative for $d$ close to $d_{\rm c}$, skewing the avalanche towards its end, as observed in numerical simulations in $d=2$ and $3$. The exponent $\gamma=(d+\zeta)/z$ is given by the two independent exponents at depinning, the roughness $\zeta$ and the dynamical exponent $z$. We propose a general procedure to predict other avalanche exponents in terms of $\zeta$ and $z$. We finally introduce and calculate the shape at fixed avalanche size, not yet measured in experiments or simulations.