Deducing distributions using sample paths properties instead of the opposite

44 Views Asked by At

It is common in the theory of stochastic processes to "build" stochastic processes by imposing properties on the distribution of the process (see Brownian motion for obvious example: independent normal increments...) and then deduce properties of the sample paths of the process (for the case of Brownian motion, nowhere differentiability $a.s$).

What if we want to do it the other way ? We impose strong sample paths properties, and then ask what are the distributional properties of such a process ?

Here is an example.

Suppose we would like to learn about the distribution of a stochastic process $X_t$ on $(0,1)$ such that the sample paths of the process are $a.s$ continuous, non-decreasing, $X(0^+) = -\infty$, $X(1^-)=\infty$, $\int_0^1 X_t^2 \,dt < \infty$.

What are the distributional properties of such a process ? What are the tools at our disposal to study such a problem ?