Distribution of exponential(X/c)

153 Views Asked by At

Suppose $X \sim Exponential(\lambda)$. That is, the PDF for $X$ is

$f_X(x)=\lambda \cdot e^{-\lambda x}$, $x\ge 0$, and the CDF of $X$ is

$F_X (x)=\int_{-\infty}^x f_X(x)=1-e^{-\lambda x}$, $x\ge 0$.

Then what can we say about the distribution of $\frac Xc$ ($c>0$)?

I think it is just $\frac Xc \sim Exponential(c\lambda)$, since we are looking for $F_{\frac Xc} (x)=Pr(\frac Xc \le x)= Pr(X\le cx)=F_X (cx)$. Is this correct reasoning? If so, wouldn't this work for any distribution, not just the exponential distribution?

1

There are 1 best solutions below

0
On BEST ANSWER

You're nearly there, but not quite. You have to take the additional step to say that, if $Z=X/c$, then $$\tag{1}F_Z(x) = F_X(cx)=1-e^{-\lambda(cx)}=1-e^{-(c\lambda)x}=F_Y(x)$$ where $Y\sim\operatorname{Exp}(c\lambda)$, and therefore $Z$ and $Y$ are identically distributed.

More tersely, $$\tag{2}F_{\frac1c\operatorname{Exp}(\lambda)}(x)=F_{\operatorname{Exp}(\lambda)}(cx)=F_{\operatorname{Exp}(c\lambda)}(x)$$ so $$\tag{3}\frac1c\operatorname{Exp}(\lambda)\sim\operatorname{Exp}(c\lambda)$$

Note that it is the third equality in $(1)$ that is crucial here. Something like this isn't true in general, so you can't make the conclusion with just any distribution.