Conditions for Markov process not to reach point at infinity

35 Views Asked by At

My question concerns the book Lectures from Markov Processes to Brownian Motion by Kai Lai Chung, more precisely the remark at the bottom of page 76:

We prove later in paragraph 3.3 that on $\{ t < \zeta \}$ we have $X_{t-} \neq \partial$, namely $X_{t-} \in \mathbf{E}$. For a Feller process this is implied by theorem 7 of paragraph 2.2.

To put things in context, the book studies Markov processes with values in a metric space $\mathbf{E}$, to which one adjoins a point at infinity $\partial$, yielding the one-point (Alexandrov) compactification $\mathbf{E}_{\partial}$. $\zeta$ is defined in the statement of theorem 7, paragraph 2.2:

$$\zeta(\omega) \equiv \inf\left\{ t \geq 0 \,;\, X_{t-}(\omega) = \partial\textrm{ or }X_t(\omega) = \partial \right\}$$

The statement of the aforementioned theorem 7 is as follows:

Let $\{X_t, \mathcal{F}_t\}$ be a Feller process with right-continuous paths having left limits. Then we have almost surely $X(\zeta + t) = \partial$ for all $t \geq 0$, on the set $\{ \zeta < \infty \}$.

I am not sure I understand why we need to invoke theorem 7 to prove the statement in the remark. Indeed, if $X_{t-} = \partial$ for some $t < \zeta$, doesn't it simply contradict the definition of $\zeta$? Could you shed some light on this point?

1

There are 1 best solutions below

0
On BEST ANSWER

Checking another reference on the general theory of Hunt processes: Fukushima, Dirichlet Forms and Markov Processes the life-time is defined there as

$$\zeta(\omega) \equiv \inf\{ t \geq 0 \,;\, X_t(\omega) = \partial \}$$

It seems the remark now makes sense if the authors thought about this definition rather than the one given in theorem 7. I therefore assume the problem comes from an inconsistency in the definitions/notations within the book.