Rescaled Lotka-Volterra models converge to super-Brownian motion (Q1781171)

Suppose that at each site of the \(d\)-dimensional integer lattice there is a plant of one of two types: 0 or 1. At random times plants die and are replaced by new plants, the times and types depending on the configuration of surrounding plants. Let \(\xi_t(x)\in\{0,1\}\) give the type of the plant at point \(x\) at time \(t\) and let \({\mathcal{N}}\) be a finite set not containing the origin and such that \(y\in {\mathcal{N}}\) implies \(-y\in {\mathcal{N}}\). Let \(f_i=f_i(x,\xi)\) be the frequency of type \(i\) in the neighborhood \(x+{\mathcal{N}}\) in configuration \(\xi.\) The dynamic of \(\xi_t(x)=i\) is described as follows. A plant of type \(i\) dies at rate \(f_i+\alpha_i f_{1-i}\) and is replaced by a plant whose type is \(j\) with probability \(f_j\). Here \(\alpha_i\) measures the strength of interspecific competition of type \(i\). The case \(\alpha_0 =\alpha_1=1\) gives the voter model. The authors consider a rescaled version of the described model by contracting the space \(l_N\) times, assigning a weight \(N^{\prime}\) to each site with value 1 and changing the parameters \(f_i\rightarrow Nf_i^N,\) and \(\alpha_i\rightarrow \alpha_i^N\). Assuming that \(N(\alpha_i-1)\rightarrow \theta_i\in (-\infty,\infty)\), they study the asymptotics for the obtained rescaled measure-valued process as \(N\to\infty\) in the cases when \(l_N\gg N\) for \(d=1\), \(l_N\gg \sqrt{N\ln N}\) for \(d=2\) and \(l_N\gg \sqrt{N}\) for \(d\geq 3\) and obtain super-Brownian motion with drift in the limit.