Self-similarity and Lamperti convergence for families of stochastic processes (Q392772)

In this paper, the authors introduce an extended definition of self-similarity for families of stochastic processes. In more detail, according to their new definition, a family of stochastic processes \(X(t)\), \(t \geq 0\), is called self-similar with Hurst exponent \(H \in \mathbb R\) and rate parameter \(\mu \in \overline{\Omega}\) if \[ \qquad X(\mu c^{H-1} ; ct) \overset {d} {=} c^H X(\mu; t), \] for \(t \geq 0, c > 0\) and \(\mu \in \Omega\). The authors discuss some relevant families of stochastic processes with finite variance that are self-similar in the sense of the generalized definition given above. In particular, they consider the Hougaard-Lévy processes, that include, for different values of the Hurst exponent \(H\), well-known stochastic processes such as Poisson processes, gamma compound Poisson processes and inverse Gaussian processes. Moreover, a new class of fractional Hougaard-Lévy motions is also introduced by means of a stochastic integration with respect to a Hougard-Lévy process. The authors also show that relevant properties of classical self-similar processes have analogues for general self-similar processes such as the covariance stucture. In this framework, they consider in detail H-SSSI processes with power variance function \(V (\mu) = \sigma^2 \mu^p\). Finally, an extension of the Lamperti transformation and Lamperti's limit theorem for general self-similarity is discussed.