Discrete approximation of a stable self-similar stationary increments process (Q605854)

In contrast to the Gaussian case, in general there exist many different \(\alpha\)-stable random processes with stationary increments and which are \(H\)-self-similar for some \(H\in(0,1/\alpha)\) if \(0<\alpha\leq 1\) or \(H\in(0,1)\) if \(1<\alpha<2\). Recall that in the Gaussian case, i.e., if \(\alpha=2\), the only processes possessing these properties are the fractional Brownian motions. The aim of the present paper is to investigate a special class of \(\beta\)--stable processes with stationary increments and being \(H\)-self-similar with \((\beta+1)/2\beta\leq H<1/\beta\) if \(0<\beta<1\), with \(H=1\) if \(\beta=1\) and with \(1/\beta<H\leq (\beta+1)/2\beta\) if \(1<\beta<2\). The construction of those processes goes via a discrete approximation following the ideas of \textit{S. Cohen} and \textit{G. Samorodnitsky} [Ann. Appl. Probab. 16, No. 3, 1432--1461 (2006; Zbl 1133.60016)]. Strong relationships with Kesten's and Spitzer's random walk in in random sceneries are shown. Finally, path properties, as e.g.~Hölder continuity, are derived from the approximation.