1
$\begingroup$

When given a piecewise distribution function for a continuous random variable $X$, how do you find the distribution function for $X^2$? Is it that you just square each of the pieces or do you have to use $F_X(x^{1/2})$? I've been given two contrasting pieces of advice. Thank you.

$\endgroup$
1
$\begingroup$

\begin{align} F_{X^2}(x) & = \Pr(X^2\le x) = \Pr(-\sqrt x \le X \le \sqrt x\,) = F_X(\sqrt x\,) - F_X(-\sqrt x\,). \tag 1 \\[10pt] f_{X^2}(x) & = \frac d {dx} F_{X^2}(x) = \frac d {dx} F_X(\sqrt x\,) - \frac d {dx} F_X( - \sqrt x\,) \tag 2 \\[10pt] & = f_X(\sqrt x\,) \cdot \frac d {dx} \sqrt x - f_X(-\sqrt x\,) \cdot \frac d {dx} (-\sqrt x\,). \tag 2 \end{align} In line $(3)$, you have two minus signs cancelling each other.

In line $(1),$ in cases where the distribution of $X$ is symmetric about $0,$ you have $F_X(-\sqrt x \,) = 1 - F_X(\sqrt x\,),$ so you can simplify a bit further.

In cases where the distribution of $X$ is symmetric about $0,$ in line $(3)$ you have $f_X(\sqrt x\,) = f_X(-\sqrt x\,),$ so you can do some simplifying.

If you just want $\operatorname E(X^2)$ rather than a complete characterization of the distribution of $X^2,$ then instead of computing $$ \int_0^\infty x f_{X^2}(x)\,dx $$ you can instead compute $$ \int_{-\infty}^\infty x^2 f_X(x)\,dx $$ and it will come to the same thing, and the integral may be less work to compute.

$\endgroup$

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.