I have a problem in the proof for variance of normally distributed random variable and my notes just report this:
\begin{align} & \frac{\sigma^2}{\sqrt{2\pi}}\int\limits_{-\infty}^\infty x^2e^{-\frac{x^2}{2}} \, dx = \frac{\sigma^2}{\sqrt{2\pi}}\int\limits_{-\infty}^\infty x e^{-\frac{x^2}{2}} (x \, dx) \\[10pt] = {} & \frac{\sigma^2}{\sqrt{2\pi}}\int\limits_{-\infty}^\infty -x \, d(e^{-\frac{x^2}{2}}) = \frac{\sigma^2}{\sqrt{2\pi}} \int\limits_{-\infty}^\infty -e^{\frac{x^2}{2}}\, dx = \sigma^2 \end{align}
I need to specifiy that we weren't taught how to compute Lebesgue-Stieltjes integrals and indeed we just faced its properties and basic definition. In this case I can't really understand how you build that chain of equalities, and what's the reason for them being equalities. I don't want to just write in my fair copy something I can't understand or justify...
EDIT: I forgot to say that I can tell why $\frac{1}{\sqrt{2\pi}}\int\limits_{-\infty}^{+\infty} e^{-\frac{x^2}{2}}dx = 1$ so I know how to get the last equality.
You don't need Lebesgue Steiltje's integrals for this. If $f(x)=e^{-\frac {x^{2}} 2}$ then $f'(x)= -xe^{-\frac {x^{2}} 2}$ by chain rule. Hence the given integral is nothing but $-\int_{-\infty}^{\infty} xf'(x)\, dx$. Now integrate by parts and you will get the answer easily.