Long-range competition on the torus
Bas Lodewijks,Neeladri Maitra
2024-03-09
Abstract:We study a competition between two growth models with long-range correlations on the torus $\mathbb T_n^d$ of size $n$ in dimension $d$. We append the edge set of the torus $\mathbb T_n^d$ by including all non-nearest-neighbour edges, and from two source vertices, two first-passage percolation (FPP) processes start flowing on $\mathbb T_n^d$ and compete to cover the sites. The FPP processes we consider are long-range first-passage percolation processes, as studied by Chaterjee and Dey. Here, we have two types, say Type-$1$ and Type-$2$, and the Type-$i$ transmission time of an edge $e$ equals $\lambda_i^{-1} \|e\|^{\alpha_i}E_e$ for $i\in\{1,2\}$, where $(E_e)_e$ is a family of i.i.d.\ rate-one exponential random variables, $\lambda_1,\lambda_2>0$ are the global rate parameters, and $\alpha_1,\alpha_2\geq 0$ are the long-range parameters. In particular, we consider the instantaneous percolation regime, where $\alpha_1,\alpha_2\in[0,d)$, and we allow all parameters to depend on $n$. We study \emph{coexistence}, the event that both types reach a positive proportion of the graph, and identify precisely when coexistence occurs. In the case of absence of coexistence, we outline several phase transitions in the size of the losing type, depending on the relation between the rates of both types. One of the consequences of our results is that for constant intensity competition, i.e.\ when the long-range parameters of the two processes are the same, while their rates differ by a constant multiplicative factor, coexistence of the two processes at the scale of the torus volume happen if and only if their global rates are equal. On the other hand, when the long-range parameters differ, it is possible for one of the types, e.g.\ Type-$2$, to reach a significant number of vertices, even when its global rate parameter $\lambda_2$ is much smaller than $\lambda_1$.
Probability