4
$\begingroup$

So I was just wondering whether I can calculate the integral $\int_{-\infty}^{\infty} e^{-x^2}$ without the famous trick of the use of double integrals. '

If it helps, this problem is an exercise in a book I bought about “definite integrals, integration techniques and integration of series”.

Thank you in advance for any help.

$\endgroup$
3
  • $\begingroup$ Please use MathJax $\endgroup$ Apr 20 '20 at 20:25
  • 2
    $\begingroup$ You can use the Taylor series for $e^u$, then do the integral over $[0,N]$ term by term and let $N\to\infty$, and over $[-N,0]$ and let $N\to\infty$... $\endgroup$ Apr 20 '20 at 20:26
  • $\begingroup$ I don't know of a simple way, but you can relate it to $\Gamma(\frac12)$ by substituting $x=\sqrt{u}$, and then the reflection formula $\Gamma(x)\Gamma(1-x) = \frac{\pi}{\sin{\pi x}}$ can be used to find $\Gamma(1/2)$. Of course, the reflection formula is non-trivial to prove in its own right, but I think you could do it without ever invoking a double integral. $\endgroup$ Apr 20 '20 at 20:27
9
$\begingroup$

I believe you can use Feynman's differentiation under the integral sign trick. Keith Conrad's notes do a good job, as per usual, explaining this method (https://kconrad.math.uconn.edu/blurbs/analysis/diffunderint.pdf). First, notice that the integrand, $e^{-x^2}$, is even, so you can instead study $I = \int_{0}^{\infty}e^{-x^2}dx$, and then double the result. Second, let: $$ F(t) = \int_{0}^{\infty}\frac{e^{-t^2(1+x^2)}}{1+x^2}dx $$ Then, $$F(0) = \int_{0}^{\infty}\frac{1}{1+x^2}dx = \left.\tan^{-1}(x)\vphantom{$\dfrac12$}\right\vert_{0}^{\infty}=\frac{\pi}{2} $$ At $t=\infty$, at every point the integrand becomes $0$. Calculating the derivative of $F(t)$, we arrive at: $$ F'(t) = \int_{0}^{\infty}\frac{e^{-t^2(1+x^2)}(-2t(1+x^2))}{1+x^2} = -2te^{-t^2}\int_{0}^{\infty}e^{(-tx)^2}dx$$ Letting $y = tx \implies dy = tdx$, where $y(0)=0$, and $y(\infty)=\infty$. Then we arrive at: $$\begin{aligned} &= -2te^{-t^2}\int_{0}^{\infty}e^{-y^2}\frac{1}{t}dy\\ &= -2e^{-t^2}\int_{0}^{\infty}e^{-y^2}dy\\ F'(t) &= -2e^{-t^2}I \end{aligned}$$ Then, using the FTC: $$ \begin{aligned} \int_{0}^{\infty}F'(t)dt &=\int_{0}^{\infty} -2e^{-t^2}Idt\\ F(\infty) - F(0) &= -2I \int_{0}^{\infty}e^{-t^2}dt \end{aligned}$$ Seeing that $I$ appears again on the right side, we have $$0 - \frac{\pi}{2} = -2I^2 \implies I^2 = \frac{\pi}{4} \implies I = \frac{\sqrt{\pi}}{2}$$ Thus, after doubling the result,

$$\int_{\mathbb{R}}^{}e^{-x^2}dx = \sqrt{\pi}$$

$\endgroup$
4
  • $\begingroup$ Hey. Thank you very much for spending time in answering this. I will have to give a lot of effort to completely understand everything but you have helped me a lot. $\endgroup$
    – Grigoris
    Apr 21 '20 at 14:43
  • $\begingroup$ Of course, the link I provided is a good starting place, and there are plenty of other references that talk about this technique in general @Grigoris $\endgroup$
    – teddy
    Apr 21 '20 at 14:44
  • $\begingroup$ Well I just sat down studied the notes you attached and with some sneak peeks of your solution I managed to do it! It’s the most interesting think I’ve ever done as a high school student thanks!!! $\endgroup$
    – Grigoris
    Apr 21 '20 at 15:49
  • $\begingroup$ You're welcome! $\endgroup$
    – teddy
    Apr 21 '20 at 15:52

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.