0
$\begingroup$

Consider a random process $\{X_n\}_{n=1}^\infty$ where each random variable is continuous. Assume that the sequence of random variables converges almost surely to $\alpha >0$ i.e., \begin{equation} X_n \xrightarrow[n\rightarrow\infty]{\text{a.s.}} \alpha. \end{equation} This implies that there exists $N$ such that \begin{equation} \mathbb{P}(X_n >0, \text{ for all } n>N) =1. \end{equation} I am trying to show that \begin{equation} \mathbb{P}(X_n >0, \text{ for all } n>N | X_1>0 ,\dots,X_N>0) =1. \end{equation} Here we assume that $\mathbb{P}(X_1>0,\dots,X_N >0 ) >0.$ My intuition is that this should hold because of the convergence result but I am not sure how to prove it formally. Any help would be really appreciated.

$\endgroup$
1
$\begingroup$

Note that $(A\cap B)\cup (A^c\cap B)=B$ where $A^c$ is the complementary event to $A$. Therefore $$\mathbb P(A\cap B)=\mathbb P(B)-\mathbb P(A^c\cap B).$$ Next, $\mathbb P(A^c\cap B)\leq \mathbb P(A^c)=0$ if $\mathbb P(A)=1$. Then $$\mathbb P(A\cap B)=\mathbb P(B)-\mathbb P(A^c\cap B)\geq \mathbb P(B).$$ Finally $$ \mathbb P(A\mid B) = \frac{\mathbb{P}(A \cap B)}{\mathbb{P}(B)} \geq \frac{\mathbb{P}(B)}{\mathbb{P}(B)} =1. $$

Small comment to the text of your question. Note that $N$ in your statement

This implies that there exists $N$ such that \begin{equation}\mathbb{P}(X_n >0, \text{ for all } n>N) =1.\end{equation}

should depend on elementary event $\omega$: for almost every $\omega$ there exists $N=N(\omega)$ such that \begin{equation}\mathbb{P}(X_n >0, \text{ for all } n>N(\omega)) =1.\end{equation} Indeed, take $X_n=1+\frac{X}{n}$ where $X$ is standard normal. Then for each non-random $N$ $$ \mathbb{P}(X_n >0, \text{ for all } n>N) = \mathbb P(X>-N)<1. $$

$\endgroup$
  • $\begingroup$ Regarding the comment to the text of the question. Would your note be the same if $X_n \rightarrow \infty$ almost surely? $\endgroup$ – SpawnKilleR Sep 7 '18 at 20:23
  • $\begingroup$ @SpawnKilleR Yes, it will be the same. Take $X_n=n+X$ and the last string remains unchanged. $\endgroup$ – NCh Sep 7 '18 at 23:43
0
$\begingroup$

Assuming that the sequence is i.i.d. and denoting $A = \{ X_n > 0: \forall n > N \}$ and $B = \{ X_1 > 0, \dots, X_N > 0 \}$ we have $$ \mathbb{P}(A ~|~ B) = \frac{\mathbb{P}(A \cap B)}{\mathbb{P}(B)} = \frac{\mathbb{P}(A)\mathbb{P}(B)}{\mathbb{P}(B)} = \mathbb{P}(A) = 1, $$ as desired.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.