Additive functionals of several Lévy processes and intersection local times (Q1577732)

The paper is about sufficient conditions for the space-time continuity of: (i) intersection local times for \(n\) independent strongly symmetric (i.e., admitting a symmetric transition density) Lévy processes; (ii) continuous additive functionals of several independent strongly symmetric Lévy processes; (iii) \(n\)-fold ``near'' intersection local times of one strongly symmetric Lévy process. The conditions are given in terms of the associated Gaussian processes and their second-order chaos processes. Using the present authors' results [J. Theor. Probab. 12, No. 2, 489-522 (1999)] and known regularity conditions for Gaussian processes [e.g., \textit{M. Talagrand}, Acta Math. 159, No. 1/2, 99-149 (1987; Zbl 0712.60044)] yields concrete continuity criteria for the above functionals in terms of the covariance functions of the associated Gaussian processes (which are the 1-potential densities of the Lévy processes). The main tool of the paper are several new variants of the Dynkin isomorphism theorem, cf. \textit{E. B. Dynkin} [J. Funct. Anal. 55, 344-376 (1984; Zbl 0533.60061)] and the authors [Ann. Probab. 20, No. 4, 1603-1684 (1992; Zbl 0762.60068)] for the basic statement, relating certain functionals of strongly symmetric Markov processes with Gaussian processes. This allows to deduce many probabilistic properties of one class of random processes from the other class. The connection between the classes is achieved via the 1-potential density of the Markov process which is used as covariance function defining a unique associated Gaussian process. The exposition is necessarily technical and not totally self-contained. In particular, (alterations of) proofs contained in the authors' papers [J. Theor. Probab. 12, No. 2, 489-522 (1999) and Ann. Probab. 24, No. 3, 1130-1177 (1996; Zbl 0862.60065)] by the present authors are quite heavily used.