Two theorems on Hunt's hypothesis (H) for Markov processes (Q2039094): Difference between revisions

The abstract given by the authors describes the contents of the main results in a clear way as follows: ``Hunt's hypothesis (H) and the related Getoor's conjecture is one of the most important problems in the basic theory of Markov processes. In this paper, we investigate the invariance of Hunt's hypothesis (H) for Markov processes under two classes of transformations, which are change of measure and subordination. Our first theorem shows that for two standard processes \((X_t)\) and \((Y_t)\), if \((X_t)\) satisfies (H) and \((Y_t)\) is locally absolutely continuous with respect to \((X_t)\), then \((Y_t)\) satisfies (H). Our second theorem shows that a standard process \((X_t)\) satisfies (H) if and only if \((X_{\tau_t})\) satisfies (H) for some (and hence any) subordinator \((\tau_t)\) which is independent of \((X_t)\) and has a positive drift coefficient. Applications of the two theorems are given.'' The paper is written in a self-contained way, sometimes with too much known details, so the reader familiar with \textit{R. M. Blumenthal} and \textit{R. K. Getoor}'s book [Academic Press. X, 313 p. 140 s. (1968; Zbl 0169.49204)] could skip these preliminaries! The authors clarify the assumption ``\((Y_t)\) is locally absolutely continuous w.r.t. to \((X_t)\)'' in the Theorem 1.1. Let \(X=(\Omega,\mathcal{M}^X, \mathcal{M}^X_t, Z_t, \theta_t, P^x)\) and \(Y=(\Omega, \mathcal{M}^Y, \mathcal{M}^Y_t, Z_t, \theta_t, \mathcal{Q}^x)\) be two standard processes on \(E\) such that \(\mathcal{M}^X\cap \mathcal{M}^Y\supset \mathcal{F}^0\) and \(\mathcal{M}^X_t\cap \mathcal{M}^Y_t\supset \mathcal{F}^0_t\) for \(0\le t<\infty\). \begin{itemize} \item[(i)] Suppose that \(X\) satisfies (H) and for any \(x\in E\) and \(t>0\) the measure \(\mathcal{Q}^x|_{\mathcal{F}^0_t}\) is absolutely continuous with respect to the measure \(P^x|_{\mathcal{F}^0_t}\). Then \(Y\) satisfies (H). \item[(ii)] The same result is true if one replaces the axiom (H) by the axiom (H\(_m\)) defined by the authors. \end{itemize} Remarks by the reviewer: \begin{itemize} \item[1)] The authors use the coordinate projection \(Z_t(\omega)=\omega(t)\), \(\forall\omega\in\Omega\) to denote the sample paths of both standard processes \(X\) and \(Y\). Note that the assumptions \(\mathcal{M}^X\cap \mathcal{M}^Y\supset \mathcal{F}^0\) and \(\mathcal{M}^X_t\cap \mathcal{M}^Y_t\supset\mathcal{F}^0_t\) for \(0\le t<\infty\) are not automatically verified with two given standard processes, for instance the Brownian motion and the Poisson process do not verify the mentioned assumptions! \item[2)] The reviewer recognizes that the ``absolutely continuity assumption'' given in part (i) by the authors implies the process \((Y_t)\) being a subprocess of \((X_t)\) under very mild conditions (see Blumenthal and Getoor [loc. cit.], Chapter III, Theorem 2.3). Roughly speaking, this means \((Y_t)\) is obtained by ``killing'' \((X_t)\) with a suitable multiplicative functional, and so \(Y\) inherits automatically the axiom (H) (resp. (H\(_m)\)) from \(X\)! \end{itemize} Theorem 1.3. Let \(X =(X_t)\) be a standard process defined on \(E\). Then, \begin{itemize} \item[(i)] \((X_t)\) satisfies (H) if and only if the process \(X_\tau=(X_{\tau_t})\) satisfies (H) for some (and hence any) subordinator \(\tau=(\tau_t)\) which is independent of \(X=(X_t)\) and has a positive drift. \item[(ii)] The same result is true if one replaces the axiom (H) by the axiom (H\(_m\)) defined by the authors. \end{itemize} Remark by the reviewer: Let \(L=(L_t)\) be a real Lévy process with generating parameters \((a,\mathcal{Q}, \tau)\), in which \(\mathcal{Q}\) is called the diffusion coefficient of \(L\), \(\pi(dx)\) the Lévy measure of \(L\). It is the Lévy measure that determines the jumping mechanism of \(L\). A subordinator \(\tau=(\tau_t)\) is a real-valued Lévy process which only takes nonnegative values. By using the definition of a Lévy process it is immediate that a subordinator must be an increasing process. By using the Lévy-Ito formula one can show that for a subordinator the diffusion coefficient must be zero, the drift, i.e. \(d=(a+ \int^{+1}_{-1}x\pi(dx))\), must be nonnegative and the Lévy measure \(\pi(dx)\) cannot charge \((-\infty,0)\). And so, the assumption on the ``positivity of the drift'' of the subordinator given by the authors turns out to be superfluous if \(a\ge 0\)!