If a random process $\{Y_t\}_{t\ge0}$ whose moment of first order is uniformly bounded, does $Y_t/t$ converges to $0$ a.s.?

51 Views Asked by At

I have problem in reading the paper in the proof Theorem 3.6 (in Page 14):

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.386.7227&rep=rep1&type=pdf

I am puzzled here: it says $X(t)/t$ converges a.s. to $v$, and $X(t)-R(t)$ is tight, then concluding that $R(t)/t$ converges a.s. to $v$.

I think the author want to say that he concluded $(X(t)-R(t))/t \rightarrow 0$ a.s. by the tightness, but I failed.

  • There is one counterexample for uniformly bounded first order moment process $Y_n$ whose time is discrete and $Y_n/n$ doesn't converges to $0$ a.s.: $Y_n$ are mutually independent, and $P(Y_n=n)=1/n, P(Y_n=0)=1-1/n$.

    Therefore, I think I need some property of the trajectory to have it, but the proof of the paper is really not clear to me.

  • However, it is easy to see uniformly finite first moment obtains the convergence to $0$ in $L^1$.

It might cost you some time to read the paper, and it is very grateful for your help!