5
$\begingroup$

Suppose $X \geq 0$ and $Y \geq 0$ are random variables and that $p\geq 0$

  1. Prove $$E((X+Y)^p)\leq 2^p (E(X^p)+E(Y^p))$$

    Proof

Since $(X+Y)^p \leq (2 \> \max\{X,Y\})^p=2^p \> \max \{X^p,Y^p\}\leq 2^p(X^p+Y^p)$ $ \implies E((X+Y)^p)\leq 2^p (E(X^p)+E(Y^p))$

  1. If $p>1$ the factor $2^p$ may be replaced by $2^{p-1}$
  2. If $0 \leq p \leq 1$ the factor $2^p$ can be replaced by $1$

Need help with part 2 and 3 any suggestions

$\endgroup$
4
  • 3
    $\begingroup$ 2. Use Jensen's inequality. 3. Use $(X+Y)^p\le X^p+Y^p$ $\endgroup$
    – A.S.
    Nov 17, 2015 at 4:13
  • $\begingroup$ How would you define the convex function. Im not seeing it @A.S. $\endgroup$
    – Josh
    Nov 17, 2015 at 4:23
  • 3
    $\begingroup$ 2. $f(x)=x^p$. Take $2^p$ over to the left to get $((X+Y)/2)^p$ to better see this. $\endgroup$
    – A.S.
    Nov 17, 2015 at 4:26
  • $\begingroup$ @Josh if you figure it out please post. I am interested $\endgroup$
    – Boby
    Nov 17, 2015 at 4:37

2 Answers 2

4
$\begingroup$

Let $f(x)=x^p$ By convexity $$f(\frac{X+Y}{2})\leq \frac{1}{2} f(X)+\frac{1}{2} f(Y) \implies \frac{(X+Y)^p}{2^p}\leq\frac{1}{2} X^p+\frac{1}{2}Y^p$$ The result follows.

$\endgroup$
1
  • $\begingroup$ For $p>1$, this is indeed true. $\endgroup$
    – Gordon
    Nov 19, 2015 at 16:18
3
$\begingroup$

If $p>1$, then by Holder inequality, \begin{align*} X+Y &\le (X^p+Y^p)^{\frac{1}{p}} 2^{1-\frac{1}{p}}. \end{align*} That is, \begin{align*} (X+Y)^p \le 2^{p-1} (X^p+Y^p). \end{align*} For $0 \le p \le 1$, we note that \begin{align*} \left(\frac{X}{X+Y} \right)^p + \left(\frac{Y}{X+Y} \right)^p \ge \frac{X}{X+Y}+ \frac{Y}{X+Y}=1, \end{align*} and then \begin{align*} (X+Y)^p \le X^p+Y^p. \end{align*}

$\endgroup$

Not the answer you're looking for? Browse other questions tagged .