[Paper Review] Mean-field forest-fire models and pruning of random trees

Xavier Bressaud, Nicolas Fournier|arXiv (Cornell University)|Jan 31, 2012

Stochastic processes and statistical mechanics24 references3 citations

TL;DR

This paper studies a mean-field coagulation-fragmentation model inspired by the one-dimensional forest-fire process, where particles coagulate at rate 2 and fragment into unit masses at rate proportional to (i−1)/n. As n→∞, the equilibrium distribution of typical particle mass converges to the size distribution of leaves in a critical binary Galton-Watson tree, while the size-biased typical particle converges to a limit profile involving the zeros of the Airy function and its derivative.

ABSTRACT

We consider a family of discrete coagulation-fragmentation equations closely related to the one-dimensional forest-fire model of statistical mechanics: each pair of particles with masses $i,j \in n$ merge together at rate 2 to produce a single particle with mass $i+j$, and each particle with mass $i$ breaks into $i$ particles with mass 1 at rate $(i-1)/n$. The (large) parameter $n$ controls the rate of ignition and there is also an acceleration factor (depending on the total number of particles) in front of the coagulation term. We prove that for each $n\in n$, such a model has a unique equilibrium state and study in details the asymptotics of this equilibrium as $n o \infty$: (I) the distribution of the mass of a typical particle goes to the law of the number of leaves of a critical binary Galton-Watson tree, (II) the distribution of the mass of a typical size-biased particle converges, after rescaling, to a limit profile, which we write explicitly in terms of the zeroes of the Airy function and its derivative. We also indicate how to simulate perfectly a typical particle and a size-biased typical particle, which allows us to give some probabilistic interpretations of the above results in terms of pruned Galton-Watson trees and pruned continuum random trees.

Motivation & Objective

• To establish the existence and uniqueness of equilibrium states for a mean-field coagulation-fragmentation model with ignition rate controlled by parameter n.
• To analyze the asymptotic behavior of the equilibrium distribution as n→∞, particularly focusing on the typical and size-biased particle mass distributions.
• To provide probabilistic interpretations of the limiting distributions using pruning procedures on Galton-Watson and continuum random trees.
• To derive exact expressions for the Laplace exponent of the inverse local time of a diffusion with drift related to the Airy function, linking stochastic processes to the scaling limit.

Proposed method

• Formalize a mean-field coagulation-fragmentation system (CF_n) where pairs of particles with masses i,j coagulate at rate 2, and particles of mass i fragment into i unit particles at rate (i−1)/n.
• Introduce an acceleration factor 1/∑c_l^n(t) in the coagulation term to account for spatial correlations in the original forest-fire model.
• Use analytic techniques to prove existence and uniqueness of the equilibrium measure for each finite n.
• Apply scaling limits and asymptotic analysis to study the behavior of the equilibrium as n→∞, particularly focusing on the mass distribution of typical and size-biased particles.
• Develop perfect simulation algorithms based on pruning procedures to sample from the equilibrium distribution of typical and size-biased particles.
• Establish a connection between the limiting distribution and the contour process of a critical binary Galton-Watson tree, and relate it to the continuum random tree (CRT) via a scaling limit.

Experimental results

Research questions

• RQ1 What is the asymptotic distribution of the mass of a typical particle in the equilibrium of the mean-field forest-fire model as n→∞?
• RQ2 How does the distribution of the mass of a size-biased typical particle behave in the limit n→∞, and what is its limiting profile?
• RQ3 Can the limiting distribution of the size-biased particle be expressed explicitly in terms of special functions, such as the Airy function and its derivative?
• RQ4 What is the connection between the equilibrium of the coagulation-fragmentation process and the pruning of Galton-Watson or continuum random trees?
• RQ5 Is there a stochastic process whose first passage time distribution matches the Laplace exponent derived from the limiting profile, and what is the nature of this process?

Key findings

• For each finite n, the mean-field coagulation-fragmentation system (CF_n) has a unique equilibrium state.
• As n→∞, the distribution of the mass of a typical particle converges in law to the distribution of the number of leaves in a critical binary Galton-Watson tree.
• The distribution of the mass of a size-biased typical particle, after appropriate rescaling, converges to a limit profile that can be explicitly written in terms of the zeros of the Airy function Ai and its derivative Ai′.
• A perfect simulation algorithm is constructed based on a pruning procedure, allowing exact sampling from the equilibrium distribution of both typical and size-biased particles.
• The limiting process for the size-biased particle is linked to a diffusion process with drift β(x) = −sign(x)α Ai′(α|x|+a₁′)/Ai(α|x|+a₁′), whose inverse local time has a Laplace exponent ψ(λ) = β(2λ/α³).
• Remarkably, when α = 2^{1/3}, the drift coefficient β and the Laplace exponent ψ(λ) of the inverse local time coincide, revealing a deep and surprising connection between two distinct stochastic objects.