Spectral representation of semistable processes, and semistable laws on Banach spaces (Q1089984): Difference between revisions

The spectral representation of symmetric stable processes are useful in studying sample path properties as well as prediction and estimation problems; cf. \textit{C. D. Hardin} jun., ibid. 12, 385-401 (1982; Zbl 0493.60046). In this paper the authors give a definition of stochastic integrals for a suitable class of functions relative to semistable and stable random measures. Moreover they obtain spectral representations for arbitrary (not necessarily symmetric) semistable and stable processes of index \(0<\alpha <2\), \(\alpha\neq 1.\) A condition for independence of infinitely divisible Banach-valued random variables, in terms of their Lévy measure, is given. As a consequence a criterion for independence of stochastic integrals is deduced. Reviewer's remark: The spectral representation theorem for general (i.e., possibly skewed or asymmetric) stable processes is also due to \textit{C. D. Hardin} jun., Univ. North Carolina, Center for Stochastic Processes, Technical Report {\#}79 (September 1984).