In this thesis, we mainly discuss some potential theory in the framework of Markov processes. Firstly, we study the properties of invariant sets and absorbing sets, introduce the concept of a-excessive function, a-excessive measure, a-recurrent and a-transient for right processes, where a < 0, and give a thorough investigation. Secondly we study the properties of moment generating functions of probability measures, calculate its sub-differential by the convex analysis, use it to characterize the quasi-symmetric probability. Particularly, if the probability measure u is symmetric or if the mean value is 0, then u is quasi-symmetric. Thirdly, we study the a-invariant Radon measures for random walks and Levy processes and give some nice ratio limit theorems that is relative to the a-invariant Radon measures. Fourthly we use the infimum of the moment generating function to study the a-transient and a-recurrent for random walks and Levy processes, give a Dichotomy theorem for not one-sided processes and prove that the process X is quasi-symmetric if and only if X is not a-recurrent for all a < 0 which is the probability explanation of quasi-symmetry. We also study some relationship between random walks and Levy processes. Finally, we give the concept of polynomial recurrent and polynomial transient for Levy processes and give a necessity condition for polynomial recurrence Levy processes which is also a sufficient condition for symmetric Levy processes.
|