1
$\begingroup$

I have some doubts regarding some different versions of Markov's inequality.

From Wikipedia's definition of Markov's inequality we have:

If $X$ is a nonnegative random variable and $a>0$, then $\mathbb{P}(X\geq a)\leq\frac{E(X)}{a}$.

How can we use that definition of Markov's inequality to derive a more general one:

Let $Y$ be a real random variable and $f:[0,\infty)\to[0,\infty)$ an increasing function. Then, for all $\epsilon>0$ with $f(\epsilon)>0$, $$\mathbb{P}(|Y|\geq\epsilon)\leq\frac{\mathbb{E}(f\circ|Y|)}{f(\epsilon)}$$

And some of my other questions are:
1. What are the purpose of introducing the function $f$?
2. Why $f$ has to be increasing?
3. How can we interpret $f(\epsilon)$?

Thanks!

$\endgroup$
2
$\begingroup$

It is easy to derive the more general form, starting from the standard Markov inequality.

$f$ being an increasing function, we have $\{|Y|\geq\epsilon\}=\{f\circ |Y|\geq f(\epsilon)\}$. Thus $P(|Y|\geq\epsilon)=P(f\circ |Y|\geq f(\epsilon))$. Then using the Markov inequality, $f\circ |Y|$ being non-negative: $$P(f\circ |Y|\geq f(\epsilon))\leq\frac{E(f\circ |Y|))}{f(\epsilon)}$$ So we have $$P(|Y|\geq\epsilon)\leq\frac{E(f\circ |Y|))}{f(\epsilon)}$$

The most classic example of this property is the Chebychev inequality, which corresponds to $Y=|X-E(X)|$ and $f(x)=x^2$. We get: $$P(|X-E(X)|\geq\epsilon)\leq \frac{Var(X)}{\epsilon^2}$$

So if we choose wisely the function $f$, we can get interesting inequalities.

$\endgroup$
  • 1
    $\begingroup$ where did you use that f is increasing? $\endgroup$ – bunny Oct 16 '17 at 16:21

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.