3
$\begingroup$

Suppose $\{a_n\}$ is a sequence of real numbers which converges to $a$, and $\{X_n\}$ is a sequence of random variables which converges to $X$ in probability. Prove that $a_nX_n \to aX$ in probability.

Through some work I deduced that $$P(|a_nX_n-aX| > \epsilon) \leq P(|a_n||X_n-X|>\epsilon/2)+P(|X||a_n-a|>\epsilon/2).$$

We know $|a_n|$ is bounded and so $P(|a_n||X_n-X|>\epsilon/2) \to 0$. However, it doesn't seem like we know anything about $|X|$ so I'm not sure what to do about the second term above. Any thoughts?

$\endgroup$
  • $\begingroup$ Note that $P(|X|> \epsilon/(2|a_n-a|))= P(|X|> k_n)$ where $k_n \to \infty$ and $X$ is a single random variable. Hence, this probability goes to $0$. [$P(|X|=\infty) = 0$] $\endgroup$ – Atbey Mar 8 '18 at 6:59
2
$\begingroup$

We can safely assume that $P(|X| = \infty) = 0$. Then $P\left(|X| > \frac{\epsilon}{2|a_n - a|}\right)$. If we ever have $|a_n - a| = 0$ then we'd have $P(0 > \epsilon/2) = 0$. Aside from those cases, though, $\frac{\epsilon}{2|a_n - a|} \to \infty$, and since the cdf $F(x)$ of any finite random variable converges to 1 as $x \to \infty$, we can say $P\left(|X| > \frac{\epsilon}{2|a_n - a|}\right) = 1 - P\left(|X| \leq \frac{\epsilon}{2|a_n - a|}\right) = 1 - F_{|X|}\left(\frac{\epsilon}{2|a_n - a|}\right) \to 0$ as $n \to \infty$.

This is what you needed. The rest of your argument checks out.

$\endgroup$
  • 1
    $\begingroup$ Another way is to note $\lim_n P(X > k_n) = P(\cap_n\{|X| > k_n\}) = P(|X| = \infty) = 0$. $\endgroup$ – Atbey Mar 8 '18 at 7:11

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.