Infinite rate mutually catalytic branching in infinitely many colonies: the longtime behavior (Q662425)

Following \textit{D. A. Dawson} and \textit{E. A. Perkins} [Ann. Probab. 26, No. 3, 1088--1138 (1998; Zbl 0938.60042)], the authors consider as a model for \textit{mutually catalytic branching} a system of coupled stochastic differential equations: \[ \begin{aligned} Y_{i,t}^{\gamma}(k) = Y_{i,0}^{\gamma}(k)+ \int_0^t \sum_{l\in S}{\mathcal A}(k,l)Y_{i,s}^{\gamma}(l)ds \\+\int_0^t(\gamma Y_{1,s}^{\gamma}(k) Y_{2,s}^{\gamma}(k))^{1/2}dW_{i,s}(k)\end{aligned} \] for \(t\geq 0\), \(k\in S\), \(i=1,2\). Here the site space \(S\) is a countable set, \(\gamma>0\) is a parameter, and the matrix \({\mathcal A}\) is defined by \( {\mathcal A}(k,l)={\mathbf A}(k,l) -{\mathbf 1}_{\{k=l\}}, \) where \({\mathbf A}\) is the transpose of a stochastic matrix \({\mathbf A}^T\). As \(\gamma \rightarrow \infty\), the limit in distribution of \(Y^{\gamma}\) is a processes \(X\) that is called the \textit{infinite rate mutually catalytic branching process} (IMUB). The two types cannot coexist in the limit \(\gamma \rightarrow \infty\) and so the proper state space at each site is \(E:=[0,\infty)^2\setminus (0,\infty)^2\). For finite initial conditions, the authors show that only one type globally survives in the long run \(t\rightarrow \infty\) if the interaction kernel \({\mathbf A}\) is recurrent. On the other hand, under a slightly stronger condition than transience, they show that both types can coexist.