TransWikia.com

Max of $2$ independent random variables

Mathematics Asked by zestiria on December 17, 2020

Let :

  • $X_1,X_2$ independent, with same law
  • $ Var(X_1)= sigma^2$, $E(X_1)=0$
  • $G$ is their cumulative function
  • Let $X= max(X_1,X_2)$ with cumulative function $F$

We want to show that $E(X)= int_{-infty}^{infty} [1-G(t)]G(t) dt $


My attempt :

$F(t)=P(X_1<t, X_2 <t)=P(X_1<t)P(X_2 <t)= G(t)^2$ so $E(X)= int_{-infty}^{infty} t 2 g(t) G(t)$ then do an integration by parts.

Or we can use $EZ= int_{0}^{infty}P(Z>t)dt$ but it works only if $Z>0$

$((G-G^2)^{-1})’ =- (g -2gG) frac{1}{ (G-G^2) ^2 } $

2 Answers

Let's prove it backwards:

$$mathbb{E}[T]=int_{-infty}^{infty}G(t)[1-G(t)]dt=int_{-infty}^{infty}G(t)dt-int_{-infty}^{infty}G^2(t)dt=$$

$$tG(t)Bigg|_{-infty}^{infty}-int_{-infty}^{infty}tg(t)dt-tG^2(t)Bigg|_{-infty}^{infty}+int_{-infty}^{infty}2t g(t)G(t)dt=$$

$$underbrace{t[G(t)-G^2(t)]Bigg|_{-infty}^{infty}}_{=0}-underbrace{mathbb{E}[X_1]}_{=0}+int_{-infty}^{infty}2t g(t)G(t)dt$$

  1. To prove that the first addend is zero you have only to calculate the limits...it's easy

  2. the second addend is zero as per initial statement

  3. the third addend is the result. It matches with the same result you correctly calculated by another way...the proof is finished.

Edit: further details on limit calculus:

$$lim_{t rightarrow +infty}frac{t}{frac{1}{G-G^2}} rightarrow-frac{(G-G)^2}{g-2gG}=0$$

Answered by tommik on December 17, 2020

Let $X^+ = max(X,0)$ and $X^- = max(-X,0)$. Then $X^+ = displaystyle int_0^{infty} mathbb{1}(X>t)dt = max(a ge : X ge a)$ and $X^- = displaystyle int_{-infty}^0 mathbb{1}(X le t) dt = min(a le 0: X > a)$.

Now we have: $$mathbb{E}(X) = mathbb{E}(X^+ - X^-) = mathbb{E}(X^+) - mathbb{E}(X^-) = ldots = int_0^{infty} (1-F(x))dx - int_{-infty}^0F(x)dx$$

Now for $F(x) = G^2(x)$ we have:

$$mathbb{E}(X) = mathbb{E}(max(X_1,X_2)) = int_0^{infty}(1-G(x))(1+G(x)) dx - int_{-infty}^0 G^2(x)dx$$

From this point it's easy to get your fact.

$Hint$: we have that $$mathbb{E}(X) = int_0^infty (1-G(x))G(x) dx + int_0^infty (1-G(x))dx -int_0^infty G^2(x)dx +$$ $$+int_{-infty}^0(1-G(x))G(x)dx -int_{-infty}^0(1-G(x))G(x)dx = $$ $$=int_{-infty}^{infty} (1-G(x))G(x)dx + R$$

And you need to prove: $$R = int_0^{infty}(1-G(x) )dx - int_0^{infty} G^2(x)dx-int_{-infty}^{0} (1-G(x))G(x)dx equiv 0$$

Answered by openspace on December 17, 2020

Add your own answers!

Ask a Question

Get help from others!

© 2024 TransWikia.com. All rights reserved. Sites we Love: PCI Database, UKBizDB, Menu Kuliner, Sharing RPP