Uniform threshold for fixation of the stochastic sandpile model on the line (Q2659823)

This paper proves that if the density of the initial random configuration of particles in a one dimensional abelian stochastic sandpile model, in which sites are independently toppled at rate \(1\) when they have more than one particle and send exactly two particles to randomly selected neighbors, is less than \(1/2\), then the model fixates almost surely. Sandpile models describe dynamics which involve particles at sites, in which toppling takes place if there are too many particles at one site; their main interest is that they display self organized criticality [\textit{A. A. Járai}, Probab. Surv. 15, 243--306 (2018; Zbl 1409.60144)]. They appear in various other contexts, including the analysis of algorithms [\textit{J. Ding} et al., ``LLL and stochastic sandpile models'', Preprint, \url{arXiv:1804.03285}]. The sandpile model under consideration in this paper consists of a randomly selected initial configuration \(\eta_0(x)\) of particles for each site \(x\) of \(\mathbb Z\). At any time \(t\), each site \(x\) is declared unstable if \(\eta_t(x)>1\); in this situation, an exponential time is determined for each unstable site, independently from all other sites, and at the given time the site \textit{topples} sending out two particles: each particle selects independently one neighbor (the right one with probability \(q\) and the left one with probability \(1-q\)) and moves there. The model is called \textit{abelian} as the number of particles moving from a toppling site is always equal to \(2\) independently of the current status of the configuration \(\eta_t(x)\), and it is \textit{stochastic} as the destination of the moving particles is random. The dynamics consists thus of particles moving from toppling sites, and occasionally getting trapped when landing on an empty site. It could be that every site changes only a finite number of times, in which case we say that the model \textit{stabilizes}, or continue changing forever. One can see that if each site has a particle except one site having an extra one, than the each site is toppled forever: such (deterministic) initial configuration has density \(1\). The authors build up on a previous paper [\textit{L. T. Rolla} and \textit{V. Sidoravicius}, Invent. Math. 188, No. 1, 127--150 (2012; Zbl 1242.60104)] that showed that if the initial density is less than \(q(1-q)\), then the model fixates almost surely. The current paper improves the bound by showing that for any initial density less than \(1/2\) the model fixates. A natural conjecture could be that the model fixates for any density less than \(1\).