0

I want to prove the following claim:

If $\{X_t: t\in \mathbb{R_{\geq 0}}\}$ convergences to $X_1$ in probability when $t\to 1$, then there is a sequence $\{t_n\}\subset \mathbb{R_{\geq 0}}$ converging to 1 such that $\{X_{t_n}\}$ converges to $X_1$ almost surely.

Since $\{X_t: t\in \mathbb{R_{\geq 0}}\}$ convergences to $X_1$ in probability, we have $$ \forall\ m\in \mathbb N,\ \exists\ \{t_{m,n}\}_n \text{ such that } \lim_{n\to \infty}\mathbb{P}(|X_{t_{m,n}}-X_1|>\frac{1}{m})=0 $$ I guess that $t_{n,n}$ is the desired sequence, but I don't know how to prove it. Any ideas?

bernard
  • 826

1 Answers1

1

Let $\varepsilon \gt0$. Choose $t_n $ so that $$ \mathbb{P}(|X_{t_{n}}-X_1|>\varepsilon)\leq \frac{1}{n^2}. $$ Apply the first Borel-Cantelli lemma to conclude.

cqfd
  • 12,974