Limit theorems for occupation time fluctuations of branching systems. II: Critical and large dimensions (Q2576955)

A system of particles in \(R^{d}\) is considered which starts off at time 0 from a Poisson field of intensity \(\lambda\). Particles move according to a symmetric \(\alpha\)-stable law and undergo critical binary branching with rate \(V>0\). Let \(N_{t}\) be the empirical measure of the particle system at time \(t\). The occupation time fluctuation process \((X_{T}(t))_{t\geq 0}\) is given by \[ (X_{T}(t),\varphi)= \frac{T}{F_{T}}\int_0^{t}((N_{Ts},\varphi)-(\lambda,\varphi))\,ds \] where \(\varphi\) belongs to a set of smooth rapidly decreasing functions on \(R^d\) and \(F_{T}\) is a scaling. In the previous paper of the same authors [Stochastic Processes Appl. 116, No. 1, 1--18 (2006; Zbl 1082.60024)] two functional limit theorems describing convergence \((X_{T}(t))_{0\leq t\leq\tau},\tau>0,\) to limiting objects in the cases \(1=d<\alpha\) and \(\alpha<d<2\alpha\) were proved. In the present paper the cases of critical and supercritical dimensions, \(d=2\alpha\) and \(d>2\alpha\), are investigated It is shown that in the critical dimension the limit as \(T\to\infty\) (with \(F_T=\sqrt{T\log T}\)) is \(C\lambda\beta\) where \(\beta=(\beta_t)_{t\geq 0}\) is real standard Brownian motion while for supercritical dimensions with norming \(F_T=\sqrt{T}\) the limit is a ``truly'' generalized Wiener process, i.e., it takes on values in a set of smooth rapidly decreasing functions.