Some long-range dependence processes arising from fluctuations of particle systems (Q1776822)

The paper is mainly devoted to the discussion of processes (centered Gaussian, continuous, self-similar with long range-dependence property) with covariances \[ s^{h+1}+t^{h+1}-(t-s)^h(hs+t),\quad 0\leq s\leq t,\quad h\in (1,3/2]\tag{*} \] and \[ s^{h+1}+t^{h+1} -{1\over 4}(s+t)^{h+1}- {1\over 4}(t-s)^h[(2h-1)s+3t],\quad 0\leq s\leq t,\quad h\in(1,2). \tag{**} \] Existence of these processes is proved with the help of occupation processes of appropriate particle systems. In \({\mathbb R}^d\), consider a standard Poisson system of particles. Each particle moves, independently of the others, according to the standard \(\alpha\)-stable Lévy process. Moreover, new particles appear according to a standard space-time Poisson point process (and then they start moving as well). Let \(N_t(A)\) be the number of particles in the set \(A\subset {\mathbb R}^d\) at time \(t\). \(L(t)=\int_0^tN_sds\) is the occupation process. Define \[ X_T(t)={L(tT)-EL(tT)\over F_T}, \] where \(F_T\) is a suitable deterministic norming. It is shown that for appropriate \(d\) and \(\alpha\), the covariances of \(X_T\) converge, as \(T\to\infty\), to a functional involving (*), and if binary branching at some fixed rate is added to the model, then the limit involves (**). These results are closely related to a paper by \textit{T. Bojdecki, L. G. Gorostiza} and \textit{A. Talarczyk} [Stat. Probab. Lett. 69, 405--419 (2004; Zbl 1076.60027)], where systems without immigration were considered. Some other long range dependence processes related to particle systems are also indicated. Moreover, it is shown that different types of particle systems may lead to the same long range dependence processes. To this end the so called demographic variation process, introduced by \textit{L. G. Gorostiza}, [J. Multivariate Anal. 25, 174-200 (1998; Zbl 0657.60051)], is used.