2
$\begingroup$

Being a novice to this topic, I am not being able to properly write down a step by step solution to this problem.

For each integer $n$, let $X_n$ be a non negative random variable with finite mean $\mu _n$ . Prove that if $\underset{n \rightarrow > \infty}{\lim}{\mu _n}=0 $ then $X_n \mbox{ converges in probability to }0$.

Update: Please don't down vote if my solution is wrong. Of course you are welcome to edit and correct it...

This is a solution I tried following @whuber’s instructions. I don’t know whether I have gone wrong somewhere or not.

Assume $$ X_n \space doesnot\space converge\space in \space probability \space to\space 0$$ $$\therefore\space there\space exists\space \epsilon \gt 0\space such \space that$$ $$\lim_{n\to\infty} P\left[|X_n-\mu_n|\ge\epsilon\right]\rightarrow 1$$ $$\implies \lim_{n\to\infty} P\left[X_n \ge\mu_n +\epsilon\right]\rightarrow 1$$ $$\implies \lim_{n \to\infty} \frac{\left(E\left[X_n\right]=\mu_n\right)}{\mu_n + \epsilon}\rightarrow 1$$ $ \implies\lim_{n\to\infty} \mu_n\to\mu_n + \epsilon$ which is a contradiction as $\epsilon$ is a number greater than 0. Hence our initial assumption was wrong. Hence proved.

$\endgroup$
  • 7
    $\begingroup$ For intuition, you might contemplate proving the contrapositive. Suppose $X_n$ did not converge in probability to $0$. Given that these variables are nonnegative, what would that imply about the limit of the means? Intuition is converted into a proof by applying the definitions... . $\endgroup$ – whuber Dec 30 '15 at 19:31
9
$\begingroup$

You already have a hint from @whuber on how to solve the problem and I encourage you to work through it.

I am going to offer another approach that is based on an application of a famous inequality relating probabilities and expectations. First, recall the definition of convergence in probability.

We say that a sequence of random variables $X_n$ converges in probability to a random variable $X$ if for every $\epsilon>0$, it holds that $$\lim_{n\to \infty} P\left[|X_n - X| < \epsilon \right] \to 1 $$

We will work with the complement of the above probability. Pick an $\epsilon>0$ and let's try to compute

$$\lim_{n \to \infty} P\left[|X_n - 0| \geq \epsilon \right] = \lim_{n \to \infty} P\left[X_n \geq \epsilon \right]$$

due to the nonnegativity. Applying Markov's inequality, we get

$$\lim_{n \to \infty} P\left[X_n \geq \epsilon \right] \leq \lim_{n \to \infty} \frac{ \mathbb{E}\left[X_n \right]}{\epsilon} \to 0$$

by assumption. Since the probability of the complement goes to zero and $\epsilon$ was arbitrary, the probability of the event $\left\{ X_n < \epsilon \right\}$ goes to one, which is what we wanted to show. Thus, your random variable converges in probability to zero. $\square$

$\endgroup$

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.