0
$\begingroup$

Supposed we are given a random variable $X$, whose probability density function is: $$ f_X(x)=\begin{cases} λ \exp⁡(-λx), & x≥0 \\ 0, & x<0. \end{cases} $$ We wish to find an invertible function $g$, such that $Y=g(X)$ is uniformly distributed on the unit interval $[0,1]$. Find $g(x)$. My answer: Since $Y$ is uniformly distributed on $[0,1)$, $fy =1$. since $fy(y)=fx(x)/|d/dx(g(x))|;$ $d/dx(g(x))=fx(x)$. Therefore, $g(x)= \exp(-λx).$ I am not sure whether this is correct. If it is not, could anyone help me out? Thank you very much.

$\endgroup$
  • $\begingroup$ please use Mathjax to make your expressions readable. See Michael Hardy's edit to learn how. $\endgroup$ – spaceisdarkgreen Nov 3 '17 at 1:55
  • $\begingroup$ There is a reason why, in the notation $f_X(x),$ one distinguishes between capital $X$ and lower-case $x$ and one is careful about which is which. Without that, you could not understand $\Pr(X\le x),$ and a variety of other things. And you get sloppy about that. $\endgroup$ – Michael Hardy Nov 3 '17 at 1:56
  • $\begingroup$ Sorry about that. I am a new user and I will learn how to use Mathjax to ask questions clearly. Thank you very much. $\endgroup$ – Deepsea234 Nov 3 '17 at 2:38
1
$\begingroup$

Let $g:[0,\infty)\to[0,1]$ be the map $u\mapsto 1-e^{-\lambda x}$. Then for $0\leqslant u\leqslant 1$ we have \begin{align} \mathbb P(g(X)\leqslant u) &= \mathbb P\left(1-e^{-\lambda X}\leqslant u\right)\\ &= \mathbb P\left( e^{-\lambda X}\geqslant 1-u \right)\\ &= \mathbb P(-\lambda X\geqslant\log(1-u))\\ &= \mathbb P\left(X\leqslant \frac1\lambda \log(1-u) \right)\\ &= 1-e^{-\lambda\left(\frac1\lambda \log(1-u)\right)}\\ &= 1-(1-u)\\ &= u, \end{align} so that $g(X)$ has $U[0,1]$ distribution. The inverse of $g$ is the map $u\mapsto -\frac1\lambda\log(1-u)$ for $0\leqslant u\leqslant1$.

$\endgroup$
  • 1
    $\begingroup$ Thank you very much for the detailed explanation. I greatly appreciate it. $\endgroup$ – Deepsea234 Nov 3 '17 at 2:39

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.