Existence and percolation results for stopped germ-grain models with unbounded velocities

STOCHASTIC PROCESSES AND THEIR APPLICATIONS

卷:142

Article

Coupier, David¹  Dereudre, David²  Le Stum, Simon² 

[1] Inst Mines Telecom Nord Europe, Paris, France

[2] Univ Lille, Lille, France

关键词: Continuum percolation;    Geometric random graph;    Lilypond model;    Brownian dynamic;   

DOI  :  10.1016/j.spa.2021.08.008

来源: Elsevier

【 摘 要 】

We investigate the existence and first percolation properties of general stopped germ-grain models. They are defined via a random set of germs generated by a homogeneous planar Poisson point process in R-2. From each germ, a grain, composed of a random number of branches, grows. This grain stops growing whenever one of its branches hits another grain. The classical and historical example is the line segment model for which the grains are segments growing in a random direction in [0, 2 pi) with random velocity. In the bilateral line segment model the segments grow in both directions. Other examples are considered here such as the Brownian model where the branches are simply given by independent Brownian motions in R-2. The existence of such dynamics for an infinite number of germs is not obvious and our first result ensures it in a very general setting. In particular the existence of the line segment model is proved as soon as the random velocity admits a moment of order 4 which extends the result by Daley et al. (Theorem 4.3 in Daley et al. (2014)) for bounded velocity. Our result covers also the Brownian dynamic model. In the second part of the paper, we show that the line segment model with random velocity admitting a super exponential moment does not percolate. This improves a recent result (Theorem 3.2 in Coupier et al.(2020)) in the case of bounded velocity. (C) 2021 Elsevier B.V. All rights reserved.

【 授权许可】

Free