From resolvents to càdlàg processes through compact excessive functions and applications to singular SDE on Hilbert spaces (Q645954)

The authors of this paper consider one of the main techniques in constructing strong Markov processes with right continuous sample paths with left limits (càdlàg, for short); the one which is based on a given resolvent of kernels. The authors' purpose is to develop a general approach, although they are guided by specific examples. In addition to Lévy processes on infinite dimensional spaces, their main class of examples consists of the solutions of stochastic differential equations for Hilbert-space-valued processes, driven by a cylindrical Brownian motion, possessing non-continuous drift. They emphasize problems that one encounters in applying known SDE and martingale techniques to such examples. SDE coefficients are singular, and the issue of proper starting points is difficult, since the known constructions usually require modifications which produce exceptional sets (they are small in the sense of being polar sets), where one can define the corresponding solution only in a trivial manner. The authors' approach overcomes these problems. From the abstract point of view, their main theorem is based on four conditions, labelled (H4) -- (H7). The first two could be considered known and rely to some extent on building a topology which ensures the existence of excessive functions with compact level sets. The last two conditions are new and rather technical. Roughly, they ensure that there is a control, via specific potentials, of the exceptional sets (on which the generating family of excessive functions attains infinite values).