scientific article; zbMATH DE number 1453159

From MaRDI portal
Publication:4955441





Consider the two following systems of interacting diffusions: \[ d\zeta_i= \sum_{j\in S} p(i,j)(\zeta_j- \zeta_i)dt+ \sqrt{\zeta_i(1- \zeta_i)} dW_i,\quad i\in S,\tag{1} \] \[ d\eta_i= \sum_{j\in S} p(i,j)(\eta_j- \eta_i) dt+ \sqrt{\eta_i} dW_i,\quad i\in S,\tag{2} \] where \(S\) is a countable set and \(p(i,j)\) is an irreducible matrix with nonnegative elements such that \(\sum_{j\neq i}p(i, j)\), \(i\in S\), is bounded, and \(W_i\), \(i\in S\), are independent Brownian motions. The processes \(\zeta_i\) and \(\eta_i\) take values in \([0,1]\) and in \([0,\infty)\), respectively. The model (1) is a stepping stone system, and the model (2) is a system of interacting Feller diffusions. The two models belong to a class of models investigated by \textit{J. T. Cox}, \textit{K. Fleischmann} and \textit{A. Greven} [Probab. Theory Relat. Fields 105, No. 4, 513-528 (1996; Zbl 0853.60080)], from which follows that local fixation of (1), i.e. \[ {\mathcal L}(\zeta(t))\to (1- \theta)\delta_{\underline 0}+ \theta\delta_{\underline 1}\quad\text{as }t\to\infty\tag{3} \] for \(\zeta_i(0)= \theta\in (0,1)\), implies local extinction of (2), i.e., \[ {\mathcal L}(\eta(t))\to \delta_{\underline 0}\quad\text{as }t\to \infty\tag{4} \] for \(\eta_i(0)= c\in [0,\infty)\), where \(\underline 0= (0,0,\dots)\) and \(\underline 1= (1,1,\dots)\). The purpose of the present paper is to show that the converse implication does not hold; more precisely (Theorem 1.1): There exists an irreducible, symmetric, stochastic matrix \(p\) on \(S:= \mathbb{N}_0\) such that the system \(\eta\) given by (2) suffers local extinction, but the system \(\zeta\) given by (1) does not fixate locally. For the entire collection see [Zbl 0937.00066].











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4955441)