Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

It is known that $$ \int_{-\infty}^\infty e^{-x^2} dx=\sqrt{\pi}.$$ What about $$\int_{-\infty}^\infty e^{-(x-ti)^2} dx, $$ where $t \in \mathbb{R}$, $i= \sqrt{-1}$.


share|cite|improve this question
Please avoid titles that are entirely in $\LaTeX$. Thanks. – J. M. Mar 24 '12 at 15:26
Consider a counter-clockwised rectangle with four corners $\pm R$ and $\pm R - it$. With a simple application of Cauchy integral formula, you will reach the conclusion: $$ \int_{-\infty}^{\infty} e^{-(x - it)^2} \; dx = \sqrt{\pi}.$$ – Sangchul Lee Mar 24 '12 at 15:27
There is a one-line proof that the answer is $\sqrt\pi$ for every real $t$ if complex analysis is allowed, and a somewhat longer one if only real analysis tools are allowed. Here is a typical example where answers to the questions what do you know? what did you try? where are you stuck? should be made clear in the post. – Did Mar 24 '12 at 15:27
To substitute (formally, as they say) $x-it$ in the formula one gets for real $x$ and to pretend that $x-it$ is real, or that it is not but hey, who cares? is not what I call a proof. So no, this is not the one-line proof mentioned in my comment. – Did Mar 24 '12 at 15:54
up vote 22 down vote accepted

Note that $$ \frac{d}{dt}\int_{-\infty}^\infty e^{-(x-it)^2}\,dx = -i\int_{-\infty}^\infty -2(x-it)e^{-(x-it)^2}\,dx = -i e^{-(x-it)^2}\big|_{x=-\infty}^\infty = 0. $$ So the integral is constant as a function of $t$, and you can set $t=0$ to find the constant.

share|cite|improve this answer
I'm just astounded... – Sangchul Lee Mar 24 '12 at 16:40
Beautiful Solution! – Daniel Montealegre Mar 25 '12 at 7:29

We start with $$ \int_{-\infty}^\infty \mathrm{e}^{-(x-ti)^2} \mathrm{d}x = \mathrm{e}^{t^2} \cdot \int_{-\infty}^\infty \mathrm{e}^{-x^2} \cdot \cos(2 t x) \mathrm{d}x $$ Now, let $\mathcal{I}(t) = \int_{-\infty}^\infty \mathrm{e}^{-x^2} \cdot \cos(2 t x) \mathrm{d}x$. Then $$ \mathcal{I}^{\prime}(t) = -2\int_{-\infty}^\infty x \cdot \mathrm{e}^{-x^2} \cdot \sin(2 t x) \mathrm{d}x = \int_{-\infty}^\infty \sin(2 t x) \cdot \mathrm{d} \left( \mathrm{e}^{-x^2} \right) \stackrel{\text{by parts}}{=}\\ -2t \int_{-\infty}^\infty \cos(2 t x) \cdot \mathrm{e}^{-x^2} \mathrm{d} x = -2 t \mathcal{I}(t) $$ Thus $$ \mathcal{I}(t) = \mathcal{I}(0) \cdot \mathrm{e}^{-t^2} = \sqrt{\pi} \cdot \mathrm{e}^{-t^2} $$

Hence $$ \int_{-\infty}^\infty \mathrm{e}^{-(x-ti)^2} \mathrm{d}x = \sqrt{\pi} $$

share|cite|improve this answer
Very thanks for answer. – Richard Mar 24 '12 at 15:55
I usually preferred real analysis methods to complex analysis ones when calculating integrals on real lines. Now I'm a bit alarmed by your answer, since it gives me a feeling that I began to forget real analysis methods. I threw you an upvote for that. – Sangchul Lee Mar 24 '12 at 16:01
There is no need to decompose the exponential into a sine and cosine part. Write the integral on the right side at the top as $\int_{-\infty}^{\infty} e^{-x^2+2xti}dx$ and call it $I(t)$. Then $I'(t) = -2tI(t)$ using integration by parts (Sasha, you lost a factor of 2 when you get just $-tI(t)$) with $u = -ie^{2xti}$ and $dv = -2xe^{-x^2}dx$. – KCd Mar 24 '12 at 16:04
@sos440: why would you prefer real analysis methods to complex analysis methods to evaluate real integrals, after seeing how incredible the residue theorem is for such computations? – KCd Mar 24 '12 at 16:06
Why even expand out the $(x-ti)^2$? If $J(t) = \int_{-\infty}^\infty e^{-(x-ti)^2}\,dt$, then $$J'(t) = -i\int_{-\infty}^\infty -2(x-ti)e^{-(x-ti)^2}\,dt = -i e^{-(x-ti)^2}\big|_{x=-\infty}^{\infty} = 0.$$ So $J(t)$ is constant. – Nick Strehlke Mar 24 '12 at 16:20

First show that $$ F(z) := \int_{-\infty}^\infty e^{-(x-z)^2} dx $$ exists for all $z \in \mathbb C$, and that it is analytic in $z$. Then evaluate it for some convenient values (say $z$ real) that have a limit point. Then you may conclude the value for all $z$.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.