On the recurrence of some time inhomogeneous Markov processes (Q4373940)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1112231
Language Label Description Also known as
default for all languages
No label defined
    English
    On the recurrence of some time inhomogeneous Markov processes
    scientific article; zbMATH DE number 1112231

      Statements

      On the recurrence of some time inhomogeneous Markov processes (English)
      0 references
      0 references
      8 September 1998
      0 references
      time change
      0 references
      Markov processes
      0 references
      Radon measures
      0 references
      criterion for strong recurrence
      0 references
      Let \(M=(X_t, P_{(t,x)})\) and \(\widehat M= (\widehat X_t, \widehat P_{(x,t)})\) be two time-inhomogeneous continuous Markov processes in duality w.r.t. the family of Radon measures \((m_t (\cdot))_{t\in R_+}\), i.e., \[ \int_X \int_X p(s,x;t,dy) \varphi(x) \psi(y) dm_s(x) =\int_X \int_X \widehat p(t,x;s,dx) \varphi(x) \psi(y) dm_t(y), \] where \((p)\) of \((\widehat p)\) denote respectively the transition-densities of \(M\) and \(\widehat M\), \(\varphi(x)\) and \(\psi (y)\) are nonnegative, measurable r.v.'s over the state-space \(X\). The author says that \(M\) is recurrent in the strong sense (resp. weak sense) provided that for each relatively compact subset \(C\subset X\) the sojourn time (resp. hitting time) of trajectories of \(X\) for the set \(C\) is a.s. infinite (resp. finite). In other words, \[ P_{(s,x)} \left\{\int^\infty_0 I_C(X_t)dt =+\infty \right\} =1 \quad \bigl(\text{resp. } P_{(s,x)} \{\sigma_C <+\infty\} =1\bigr) \text{ a.e. } x\in X. \tag{1} \] The above terminologies are justified by Lemma 1, which shows that a process recurrent in the strong sense has indeed the same property in the weak sense. The main result of the paper is Theorem 1, stated as follows. Consider the CFA \(\widehat A= \{\widehat A_t= \int^t_0 I_C(\widehat X \widehat \tau(s)) ds\}\) and denote by \(\widehat W^\alpha_{\beta,C}\) the \(\alpha\)-resolvent of the \(\exp (-\beta A_t)\)-subprocess of \(\widehat M\) and put \[ I(u:f): =\varepsilon (u,\widehat W^0_{\beta,C} f)= \lim_{\alpha \to\infty} \alpha (u,\widehat W^0_{\beta,C} f-\alpha G_\alpha \widehat W^0_{\beta,C} f)_\eta,\tag{2} \] where \((\cdot, \cdot)_\eta\) denotes the integration w.r.t. \(m_t(dx) dt\) over \(R\times X\), resp. \[ J(u:f): =\varepsilon (u,\widehat V_Cf)= \lim_{\beta \to \infty} \varepsilon (u,\widehat W^0_{\beta,C} f).\tag \(2'\) \] Theorem 1. \(M\) is recurrent in the strong sense (resp. weak sense) provided that for each \(C\subset X\) relatively compact there exists a sequence \((u_n)\) of nonnegative, bounded r.v.'s converging to 1 a.e. such that, if \(f\) is a nonnegative, bounded, with compact support r.v. in \(R_+\times C\), then \(\lim_nI(u_n;f)=0\) (see (2)) (resp. \(\lim_n J(u_n;f)=0\) (see \((2')))\). In the last subsection 3 the author gives three examples which show that his general criterion for strong recurrence could be apply to the cases of ``drifted recurrent diffusion processes'' as well of ``time change Brownian motions'', which are typical time-inhomogeneous Markov processes.
      0 references

      Identifiers