Modeling and simulation with operator scaling (Q608214)

Self-similar stochastic processes are widely used in various applications in science, e.g., electrical engineering, image processing, network traffic, finance and physics, because of their scaling properties. This paper deals with several aspects of operator self-similar processes, i.e., (multi-dimensional) self-similar processes where the scaling factor varies with each coordinate. In particular, the authors consider an operator self-similar Lévy process and develop a simulation algorithm based on a series representation of the process. The big jumps of the processes form a compound Poisson process and can thus be simulated exactly, while a Gaussian approximation is used for the small jumps, closely following \textit{S. Cohen} and \textit{J. Rosiński} [Bernoulli 13, No.~1, 195--210 (2007; Zbl 1121.60049)]. The authors illustrate the different aspects of operator self-similar processes through various simulation examples. Then, they present a classification of operator self-similar processes in dimension 2, in terms of the exponent and the spectral measure. This classification is important for model selection.