Setting $$X_1,X_2,\ldots \overset{d}{\sim} \mathcal{N}(0,1)$$ $$V = X_1^2 + \ldots + X_n^2$$

Now I would like to find the expectation of $\sqrt{V}$. My biggest problem is I don't know how to write down the density of $\sqrt{V}$. Could someone help me derive it?


2 Answers 2


The distribution of $V$ is $\chi^2(n)$. The square root of a $\chi^2(n)$ random variable is a $\chi(n)$ random variable.

Let the random variable $V$ have the chi-square distribution with $n$ degrees of freedom with probability density function $$ f_V(v) = \begin{cases} \frac{v^{n/2-1} e^{-v/2}}{2^{n/2} \Gamma\left(\frac{n}{2}\right)}, & v \geq 0; \\ 0, & \text{otherwise}. \end{cases} $$ By the transformation technique, using the transformation $Y=g(V)=\sqrt{V}$, we have $$ f_Y(y)=f_V(g^{-1}(y))\left|\frac{dv}{dy}\right|=\frac{(y^2)^{n/2-1} e^{-y^2/2}}{2^{n/2} \Gamma\left(\frac{n}{2}\right)}|2y|=\frac{y^{n-1} e^{-y^2/2}}{2^{n/2-1} \Gamma\left(\frac{n}{2}\right)}\qquad y>0 $$ which is the probability density function of the chi distribution with $n$ degrees of freedom.

The expectation is $$ \Bbb{E}(Y)=\int_{-\infty}^{\infty}yf_Y(y)\textrm{d}y=\int_{0}^{\infty}\frac{y^{n} e^{-y^2/2}}{2^{n/2-1} \Gamma\left(\frac{n}{2}\right)}\textrm{d}y=\sqrt{2}\frac{\Gamma\left(\frac{n+1}{2}\right)}{\Gamma\left(\frac{n}{2}\right)} $$


First note that each $X^{2}_i\sim Gamma(\frac{1}{2},\frac{1}{2})$ for each $i$ (i.e. chi square with df=1) now to find distribution of $\sum_{i=1}^{n}X^2_{i}$ we see that using moment generating functions and fact that each are independent of each other that $$M_{\sum_{i=1}^{n}X^2_{i}}=M(t)^{n}=\left(\frac{\frac{1}{2}}{\frac{1}{2}-t}\right)^{\frac{n}{2}}$$ thus $\sum_{i=1}^{n}X^2_{i}\sim Gamma(\frac{n}{2},\frac{1}{2})$ thus $V=\sum_{i=1}^{n}X^2_{i}$ thus $$E(\sqrt{V})=\int_0^\infty \frac{\frac{1}{2}^{\frac{n}{2}}}{\Gamma(\frac{n}{2})}v^{\frac{1}{2}}v^{\frac{n}{2}-1}e^{-\frac{1}{2}x}dx=\frac{\frac{1}{2}^{\frac{n}{2}}}{\Gamma(\frac{n}{2})}\frac{\Gamma(\frac{n+1}{2})}{\frac{1}{2}^{\frac{n+1}{2}}}\int^\infty_0\frac{\frac{1}{2}^{\frac{n+1}{2}}}{\Gamma(\frac{n+1}{2})}v^{\frac{n+1}{2}-1}e^{-\frac{1}{2}x}dx=\frac{\sqrt{2}\Gamma(\frac{n+1}{2})}{\Gamma(\frac{n}{2})}$$

  • $\begingroup$ Care to explain why your end result is off by a factor of two from the above answer? $\endgroup$
    – will
    Feb 15, 2016 at 16:22
  • $\begingroup$ Nice catch, just simplified wrong second to last step it should be multiplied by $sqrt{2}$ at end $\endgroup$
    – Kamster
    Feb 15, 2016 at 16:28

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.