Probability Density of Mapped $y=ax^2$ to Normal Distribution

284 Views Asked by At

Let $x$ be a scalar random variable and let $y$ be a scalar quantity such that $y = ax^2$.

If

$$p(x) = \frac 1 {\sigma\sqrt{2\pi}}\, e^{-x^2/(2\sigma^2)}$$

a) Find the probability density of $y$

b) Find expected value of vector $y$;

c) Find covariance of $y$.

Perhaps I am missing something here, but for part a would we simply $x^2 = y/a$ and put that into the probability density of $p(x)$ to get get $p(y)$.

For part b, would we just take the integral of $\int_{-\infty}^\infty p(x)\,dx$.

For the rest, I am not exactly sure how to tackle the problem.

Thank you very much in advance.

1

There are 1 best solutions below

0
On

One should not use the same character to refer to both the random variable and the argument to the density or to the c.d.f. Thus in the expression $F_X(x)=\Pr(X\le x)$, the capital $X$ and the lower-case $x$ have different roles.

\begin{align} f_Y(y) & =\frac{d}{dy} \Pr(Y\le y) = \frac{d}{dy} \Pr\left(-\sqrt{\frac y a}\le X\le\sqrt{\frac y a}\,\right) \\[10pt] & = \frac d {dy} 2\Pr\left(0\le X\le\sqrt{\frac y a}\right) = 2\frac d {dy} \int_0^\sqrt{y/a} \frac 1 {\sigma\sqrt{2\pi}} e^{-x^2/(2\sigma^2)} \, dx \\[10pt] & = 2 \frac d {du} \int_0^u \frac 1 {\sigma\sqrt{2\pi}} e^{-x^2/(2\sigma^2)} \, dx \cdot \frac{du}{dy} = 2 \frac 1 {\sigma\sqrt{2\pi}} e^{-u^2/(2\sigma^2)} \cdot\frac d {dy} \sqrt{\frac y a} \\[10pt] & = 2 \frac 1 {\sigma\sqrt{2\pi}} e^{-u^2/(2\sigma^2)} \cdot \frac 1 2 \sqrt{\frac a y} \cdot\frac 1 a = 2 \frac 1 {\sigma\sqrt{2\pi}} e^{-y/(2a\sigma^2)} \cdot \frac 1 2 \sqrt{\frac a y} \cdot\frac 1 a \end{align} and do routine simplifications from there.

For part $(b)$, you can find $$ \int_{-\infty}^\infty ax^2 f_X(x)\,dx = 2 \int_0^\infty ax^2 f_X(x)\,dx. $$