3

Problem.src) Evaluate $\displaystyle \int_{-\infty}^{\infty} \frac{x \sin x}{x^2+4} \, \mathrm{d}x $

I know I am supposed to split it up like this

enter image description here

and $\Gamma(R)$ tends to zero and the other tends to my integral as $R$ tends to infinity?

I compute the residue at $2i$ which I think is $\frac{\sin(2i)}{2}$ ?

But I am a little stuck as to what to do now, I have never seen an example of this type of integral involving $\sin(x)$.

Usually we use the ML lemma for these types of problems, but because of the $x$ on top do I need to use Jordan's lemma?

Do I have to use $\sin(x)=\frac{e^{ix}-e^{-ix}}{2i}$ ? I am quite confused, any help would be appreciated.

Thanks.

Sangchul Lee
  • 167,468
bws
  • 327
  • What you should use is usually called the residue theorem. – AlexR May 26 '15 at 20:17
  • The roots of $x^2+4$ are $2i$ and $-2i$. Where are you getting $i\sqrt{2}$ from? – Spencer May 26 '15 at 20:19
  • sorry, when I said use ML lemma and Jordans lemma I meant to show that gamma(R) tends to zero (i think?) – bws May 26 '15 at 20:20
  • oh dear, I copied the question down wrong when I was trying to work it out, I will go through it again :( – bws May 26 '15 at 20:21

2 Answers2

1

HINT:

$$\int_{-\infty}^{\infty}\frac{x\sin x}{x^2+4}dx=\text{Im}\left(\int_{-\infty}^{\infty}\frac{xe^{ix}}{x^2+4}dx\right)$$

Then, evaluate

$$\text{Im}\left(\oint_C\frac{ze^{iz}}{z^2+4} dz\right)=2\pi i\,\text{Im}\left(\text{Res}\left(\frac{ze^{iz}}{z^2+4}\right),z=i2\right)$$


By request, we show that $\int_{C_R}\frac{ze^{iz}}{z^2+4} dz \to 0$ as $R\to \infty$. On $C_R$, we have $z=Re^{it}$ so that $dz=iRe^{it}dt$, with $0\le t\le \pi$. Then, for $R>2$ we have

$$\begin{align} \left| \int_{C_R}\frac{ze^{iz}}{z^2+4} dz\right|&=\left| \int_0^{\pi}\frac{Re^{it}e^{iRe^{it}}}{R^2e^{i2t}+4} iRe^{it}dt\right|\\\\ &=\left| \int_0^{\pi}\frac{R^2e^{iR\cos t}e^{-R\sin t}}{R^2e^{i2t}+4} dt\right|\\\\ &\le\int_0^{\pi}\left|\frac{R^2e^{-R\sin t}}{R^2e^{i2t}+4} \right|dt\\\\ &=\int_0^{\pi}\frac{R^2e^{-R\sin t}}{|R^2e^{i2t}+4|} dt\\\\ &\le \frac{2R^2}{R^2-4}\int_0^{\pi/2}e^{-R\sin t} dt\\\\ &\le\frac{2R^2}{R^2-4}\int_0^{\pi/2}e^{-R(2t/\pi)} dt\\\\ &=\frac{2R^2}{R^2-4}\frac{\pi}{2}\frac{1-e^{-R}}{R}\\\\ &\to 0\,\,\text{as} R\to \infty \end{align}$$

Mark Viola
  • 179,405
  • I think you are missing a factor of $z$ in the numerator of the integrand. – Spencer May 26 '15 at 20:30
  • @Spencer Thanks for the catch! Edited. – Mark Viola May 26 '15 at 20:37
  • do I need the residue at plus and minus $2i$ ? I still don't understand... :( – bws May 26 '15 at 20:46
  • @bws That was a typo. You are correct. The enclosure is in the upper-half plane with the pole at $z=+i2$. One could also enclose the lower-half plane, account for the reversed orientation (i.e., clock-wise in the lower-half plane) by multiplying by $-1$, and evaluate the residue at $z=-i2$. – Mark Viola May 26 '15 at 21:06
  • @Dr. MV Thank you :) How do I use Jordan's Lemma to show that $gamma(R)$ tends to $0$ as $R$ tends to infinity? – bws May 26 '15 at 21:44
  • @bws You're welcome. My pleasure. Please see my previous comment for a synopsis on Jordan's Lemma. – Mark Viola May 26 '15 at 22:01
  • @Dr. MV what previous comment? where? – bws May 26 '15 at 22:10
  • @bws The comment was embedded in an answer that was deleted by the person who posted it. So, I added a section that shows how to show $\int_{C_R}\to 0$ as $R\to \infty$. Let me know how I can improve my answer. I just want to give you the best answer I can. – Mark Viola May 27 '15 at 02:52
  • This isn't really using Jordan's lemma, as much as actually deriving it. – mrf May 27 '15 at 05:09
  • @mrf You're absolutely correct. I thought it would be instructive to prove that the integral goes to zero rather than state the general result as JL. – Mark Viola May 27 '15 at 14:39
1

Just to be a little bit more general, let's look at the integral

$$ J[a,b]=\int_{-\infty}^{\infty}\frac{e^{i a x}}{x^2+b^2}dx $$

where $a>0$ and $\Im[{b}]>0$.

it's clear that we can write our orignal integral as

$$ I=-\Re[\partial_a J[a,b]]_{a=1,b=2} $$

Consider now the complex function

$$ f(z)=\frac{e^{i a z}}{z^2+b^2} $$

Let's apply Cauchy's theorem to a large semicircle in the upper half plane as depicted in your question (This allowed, because the integral vanishs as $1/z$ for $z\rightarrow \infty$). We obtain

$$ J[a,b]=\int_{SC} dz f(z)=\int_{-\infty}^{\infty}f(x)dx=-2\pi i \times \text{res}[z=i b]=2\pi i \times\frac{e^{-a b}}{2 i b}=\pi \frac{e^{-a b}}{b} $$

Here $SC$ stands for semicircle, therefore $$ I=\pi \Re[e^{-ab}]_{a=1,b=2}=\frac{\pi}{e^2} $$

The advantage of this (maybe a little bit over-the-top looking) method is that you can now calclulate arbritary integrals of the form $\int_{-\infty}^{\infty}\frac{ x^n e^{i a x}}{(x^2+b^2)^m}$ (with $m>n$ for convergence reasons)by just taking appropriate derivatives w.r.t to $a$ and $b$ and then seperate into real and imaginary parts if you are interested in the trigonometric cases!

tired
  • 12,325
  • Feynman wouldn't have done it differently. – Mark Viola May 26 '15 at 21:08
  • That's a HUGE compliment! Thanks a lot my friend! :) – tired May 26 '15 at 21:09
  • You're welcome. Feynman was amazing. Have you seen his Path Integral Representation for the wave function? – Mark Viola May 26 '15 at 21:26
  • Yeah, i'm actually doing some research in Quantum Field Theory. It's amazing how such a bad defined object leads to so many correct results ^^ – tired May 26 '15 at 21:35
  • Ed Witten has said that QFT is by far the most difficult theory in modern physics. But I believe that the Path Integral was finally defined rigotously in the 1970s. Yet, prior to that, its formal use led to so many great insights. – Mark Viola May 26 '15 at 21:50
  • yeah you are right, for a huge class of hamiltonians the path integral is nowadays well defined. But i think still mathematicans do not fully understand it rigorously in all physical significant cases. http://physics.stackexchange.com/questions/169626/why-isnt-the-path-integral-rigorous – tired May 26 '15 at 22:12