The average density of super-Brownian motion (Q5929092)

To treat irregular measures, we can recall the notion of average density introduced by \textit{T. Bedford} and \textit{A. M. Fisher} [Proc. Lond. Math. Soc., III. Ser. 64, No. 1, 95-124 (1992; Zbl 0706.28009)]. The average density of order \(n\) (\(n \geq 2\)) of measure \(\mu\) at \(x\) is defined as \[ \text{Ad}_x(\mu, n, \psi) = \lim_{k \rightarrow \infty} \frac{1}{k} \int_0^k \frac{ \mu( B(x, 1 / \exp^{(n-1)} (a))}{(1 / \exp^{(n-1)} (a))^{\alpha} } da \] with gauge function \(\varphi\), where \(\exp^{(n)}\) is the \(n\)th iterate of the exponential function and \(B(x,r)\) is a closed ball of radius \(r\) with the center point \(x\). In line with this, the average density of the Brownian path has been investigated by \textit{K. J. Falconer} and \textit{Y. M. Xiao} [Stochastic Processes Appl. 55, No. 2, 271-283 (1995; Zbl 0819.60038)] and by the author [ibid. 74, No. 1, 133-149 (1998; Zbl 0937.60031)]. On the other hand, the path of a Brownian motion and the support of a super-Brownian motion possess the same exact Hausdorff dimension gauge as random sets. So that, it is quite natural and important, not to study simple measurements of the size of random sets, but to compare them by using a parameter characterizing their features such as local density of mass and geometric regularity. Therefore, the purpose of this paper consists in investigating the average density of the super-Brownian motion. The author proves that, for a super-Brownian motion \(\{ Z_t \}\) in \(\mathbb R^d\) (\(d \geq 3\)) with arbitrary finite initial mass \(\mu\) \(\in\) \(M_F(\mathbb R^d)\) and constant branching rate \(\gamma > 0\), at a fixed time \(t> 0\), the average density \(\text{Ad}_x(Z_t, 2, \psi_0)\) of order two exists with probability one, at \(Z_t\)-almost every \(x \in \mathbb R^d\) and is equal to a constant \(D(d, \gamma)\) \(=\) \(\gamma / (2 d - 4)\), where the density gauge is given by \(\psi_0(r)\) \(=\) \(r^2\). This constant can also be interpreted in terms of the equilibrium random measure \(Z_{\infty}\) on the space \(M_p(\mathbb R^d)\) of the super-Brownian motion. Even for the case of dimension \(d=2\), although some subtle averaging argument is required, he shows, too, that at a fixed time \(t>0\), the average density \(\text{Ad}_x(Z_t, 3,\psi_1)\) of order three exists and is constant at \(Z_t\)-almost every point \(x\), where the gauge is given by \(\psi_1(r)\) \(=\) \(r^2 \log (1 / r)\). It is interesting to note that this constant coincides with the average density of the Brownian occupation measure of the same dimension exactly if \(\gamma = 4\). The proofs are essentially due to precise estimates on the decay of correlation between the mass of concentric balls as the radii move apart. There are some other interesting papers treating different fractal parameters to compare the support of super-Brownian motion and the Brownian path, see e.g. \textit{J.-F. Le Gall, E. A. Perkins} and \textit{S. J. Taylor} [ibid. 59, No. 1, 1-20 (1995; Zbl 0848.60078)] on the exact packing dimension gauge, and \textit{E. A. Perkins} and \textit{S. J. Taylor} [Ann. Inst. Henri Poincaré, Probab. Stat. 34, No. 1, 97-138 (1998; Zbl 0905.60031)] on the multifractal spectrum of super-Brownian motion.