On the a.s. convergence of certain random series to a fractional random field in \(\mathcal D'(\mathbb R^{d})\) (Q2387328)

The authors provide a method for constructing a random series of functions which converges a.s. to a generalized random field on \(\mathbb{R}^d\) with a prescribed covariance structure. The series resembles the Karhunen-Loève expansion. The covergence is in the sense of the Schwartz space of distributions \({\mathcal D}'(\mathbb{R}^d)\). The method is useful for constructing a series which converges to a random field \(X\) with a spectral density of the form \[ \Phi_X(\omega)= (1+ |\omega|^2)^{-\gamma} |\omega|^{-\alpha},\quad \gamma> 0,\;0< \alpha< d/2. \] This result is relevant in connection with long-range dependence (long memory) random fields.