More on the Brownian force model: avalanche shapes, tip driven, higher $d$

Abstract: The Brownian force model (BFM) is the mean-field model for the avalanches of an elastic interface slowly driven in a random medium. It describes the spatio-temporal statistics of the velocity field, and, to some extent is analytically tractable. We extend our previous studies to obtain several observables for the BFM with short range elasticity, related to the local jump sizes $S(x)$ and to the avalanche spatial extension in $d=1$, or the avalanche span in $d>1$. In $d=1$ we consider both driving (i) by an imposed force (ii) by an imposed displacement "at the tip" and obtain in each case the mean spatial shape $\langle S(x) \rangle$ at fixed extension, or at fixed seed to edge distance. We find that near an edge $x_e$, $S(x) \simeq \sigma |x-x_e|^3$ where $\sigma$ has a universal distribution that we obtain. We also obtain the spatiotemporal shape near the edge. In $d>1$ we obtain (i) the mean shape $\langle S(x_1,x_\perp) \rangle$ for a fixed span, which exhibits a non-trivial dependence in the transverse distance to the seed $x_\perp$ (ii) the mean shape around a point which has not moved, $\langle S(x) \rangle_{S(0)=0}$, which vanishes at the center as $|x|^{b_d}$ with non trivial exponents, $b_1=4$, $b_2=2 \sqrt{2}$ and $b_3=\frac{1}{2} (\sqrt{17}-1)$. We obtain the probability distributions in any $d$ for the maximal radius of an avalanche and the minimal distance of approach to a given point, as well as the probability of not hitting a cone in $d=2$. These results equivalently apply to the continuum limit of some branching diffusions, as detailed in a companion paper.