Optimal stopping of strong Markov processes (Q1940245): Difference between revisions

The paper deals with an optimal stopping problem: find a function \(V\) and a stopping time \(\tau^\star\) such that \[ V(x) := \sup_{\tau\in\mathcal M}\operatorname{E}_x(e^{-\beta \tau}G(X_\tau))=\operatorname{E}_x(e^{-\beta {\tau^\star}}G(X_{\tau^\star})), \] where the function \(G\) has some regularity properties and \(\beta > 0\) is a constant. On \(\{\tau=\infty\}\), \(e^{-\beta \tau}G(X_\tau))=\lim\sup_{t\rightarrow\infty}e^{-\beta t}G(X_t))\) is adopted. The main result of the paper characterizes the solution of this optimal stopping problem for general strong Markov processes (cf. specific results by \textit{G. Deligiannidis, H. Le} and \textit{S. Utev} [J. Appl. Probab. 46, No. 4, 1130--1145 (2009; Zbl 1213.60082)], \textit{A. A. Novikov} and \textit{A. N. Shiryaev} [Theory Probab. Appl. 49, No. 2, 344--354 (2005); translation from Teor. Veroyatn. Primen. 49, No. 2, 373--382 (2004; Zbl 1092.60018)] and \textit{B. A. Surya} [Stochastics 79, No. 3--4, 337--361 (2007; Zbl 1156.60026)]). It is shown that the optimal rule has threshold form. It has been shown that, in some problem, the optimal rule is a one-side one. There are also characterized two-side cases (i.e., \(\tau^\star=\inf\{t\geq 0: X_t\notin [x_\star,x^\star]\}\)).