Coexistence in two-type first-passage percolation models (Q1774221)

This paper studies a competition model which generalizes the two-type Richardson model on \(\mathbb{Z}^d\). In the former, mainly studied by Häggström and Pemantle among others, two infections starting from different sources compete to invade the sites of \(\mathbb{Z}^d\). Each infection can be characterized by a first passage percolation process governed by the same family of random variables \((t(e))_e\) indexed by the edges of the grid, and interact only because once a site is infected by one type, it remains infected for ever and cannot transmit. Whereas these standard first passage times are considered to be i.i.d. exponentials on the lattice, the authors deal with a general random field with a stationary and ergodic distribution on the infinite Bernoulli percolation cluster for any \(p>p_c(d)\). Their framework includes the standard case and most of the other previous known cases. The goal is to provide a criterion to distinguish the only two possible different evolutions: Either one type surrounds everything and kills the other, either the two types coexist. And more precisely, does coexistence occur with probability one? This question has not only an interest by its own, it provides also a powerful tool to study the existence of semi-infinite geodesics in first passage percolation models. The authors first recall results on the existence of a directional asymptotics speed in the standard first passage percolation model, and extend them in their context under reasonable smoothness conditions. Thereafter, they extend the techniques and tricks of Häggström et al. to establish the possible coexistence of species when the starting points are far enough. Other results on the competition model, mutual unbounded growth and the existence of two disjoint geodesics for diffuse passage time are also provided, and the results are nicely illustrated by an example of a discrete competing model.