How exactly Rayleigh, exponential and Chi-square distributions are related?

684 Views Asked by At

I have read on Wikipedia and understood that

  1. If $R$ is Rayleigh distributed with standard deviation = 1 then only $R^2$ has a chi-squared distribution.

  2. Also from various sources, I had seen that if $R$ is Rayleigh distributed with standard deviation other than 1 then $R^2$ has exponential distribution.

My query is that is my understanding given in above two points correct?

Any help in this regard will be highly appreciated.

1

There are 1 best solutions below

3
On BEST ANSWER

Start observing that the definition of a Reileigh $R(\sigma)$ is the following

$$R=\sqrt{X^2+Y^2}$$

where $X,Y$ are iid gaussian $N(0;\sigma^2)$


Proof:

$$f_{XY}(x,y)=\frac{1}{\sigma^22\pi}e^{-(x^2+y^2)/(2\sigma^2)}$$

passing in polars you get

$$f_{\Theta P}(\theta,\rho)=\frac{1}{2\pi}\mathbb{1}_{(0;2\pi)}(\theta)\times \frac{\rho}{\sigma^2} e^{-\rho^2/(2\sigma^2)}\mathbb{1}_{(0;\infty)}(\rho)$$

where $\Theta\sim U(0;2\pi)$ is the distribution of the angle and $P\sim\text{Rayleigh}(\sigma)$ is the distribution of the radius $R=\sqrt{X^2+Y^2}$


thus it is evident (and anyway easy to prove) that if $\sigma=1$, $R^2$ is the sum of two independent standard gaussian, that is

$$R^2(1)\sim \chi_{(2)}^2$$

It is not difficult to prove that

$$R^2(\sigma)\sim\text{Exp}\left( \frac{1}{2\sigma^2} \right)$$


Proof:

being $Y=X^2$ a monotonic transformation when $x\ge 0$ we have

$x=\sqrt{y}$; $|x'|=1/(2\sqrt{y})$ and thus

$$f_Y(y)=\frac{\sqrt{y}}{\sigma^2}\frac{1}{2\sqrt{y}}e^{-y/(2\sigma^2)}=\frac{1}{2\sigma^2}e^{-y/(2\sigma^2)}$$


Using MGF properties, immediately follows that

$$\Sigma_i R_i^2\sim\text{Gamma}\left(n; \frac{1}{2\sigma^2} \right)$$

You can find other related distribution on wiki here