Moving boundaries for Ornstein-Uhlenbeck processes

155 Views Asked by At

Let $\tau(X_t)$ be the first-passing time to the moving boundary $a(t)$ for an Ornstein-Uhlenbeck process $X_t$. I wonder how general an $a$ can be allowed in order to guarantee that $\tau$ becomes continuous almost everywhere with respect to the measure induced by $X_t$ (Wiener, correct?). I would be very grateful for any references!

1

There are 1 best solutions below

0
On

Somewhat embarrassingly, it seems I found the answer (please correct me if I'm wrong).

An excellent book on the subject is Whitt: Stochastic Process Limits (Springer 2002). It provides nice intuition together with a useful toolbox for attacking these problems.

The keyword appears to be Skorokhod topologies. Elements in this space are continuous with jumps, or more precisely speaking, cadlag functions. A nice property of these functions is that they can be described as limits of sequences of piecewise constant functions with jumps. Whitt explains very nicely how properties such as first-passing time are analyzed by considering convergent sequences. It seems clear from this that cadlag boundaries will be fine, since OU processes are continuous (and thereby cadlag).