I have no idea how to start, it looks like integration by parts won't work.


If someone could shed some light on this I'd be very thankful.

  • $\begingroup$ are you looking for exact value or just convergence? $\endgroup$ – Anurag A Jun 19 '14 at 9:37
  • $\begingroup$ With wolfram alpha I know that the answer is $\pi /8e$. I do not know how to come to a solution $\endgroup$ – user1043065 Jun 19 '14 at 9:41
  • $\begingroup$ Convergence is easy to prove since the function is bounded by an integrable function $\endgroup$ – user1043065 Jun 19 '14 at 9:44
  • $\begingroup$ I'm not so sure that it is bounded by an obvious integrable function. Perhaps I'm not thinking correctly. $\endgroup$ – Anurag A Jun 19 '14 at 9:45
  • $\begingroup$ M yields $\large{\pi \over 8{\rm e}} \approx 0.1445$. $\endgroup$ – Felix Marin Jun 20 '14 at 2:35

It is well known that $\displaystyle\int_0^\infty\frac{\sin x}{x}\,dx$ converges. Then $$ \int_0^{\infty}\frac{x^5\sin x}{(1+x^2)^3}\,dx=\int_0^{\infty}\Bigl(\frac{x^5}{(1+x^2)^3}-\frac1x\Bigr)\sin x\,dx+\int_0^{\infty}\frac{\sin(x)}{x}. $$ This implies that the integral converges, since the the first integral on the right hand side converges absolutely,

To compute the value use calculus of residues. Let $$ f(z)=\frac{z^5\,e^{iz}}{(1+z^2)^3}. $$ $f$ is meromorphic on an open set contaning the upper half plane, with a pole of order $3$ at $z=i$. For $R>1$ let $C_R$ be the semicircle $z=R\,e^{it}$, $0\le t\le\pi$. Then $$ \int_{-R}^R\frac{x^5e^{ix}}{(1+x^2)^3}\,dx+\int_{C_R}f(z)\,dz=2\,\pi\,i\,\text{Res}(f;i). $$ Now you have to:

  1. Prove that $\lim_{R\to\infty}\int_{C_R}f(z)\,dz=0$
  2. Compute the residue.
  • $\begingroup$ $f$ and $x^5\sin x/(1+x^2)^3 $ do not coincide on the reals. Is that no problem? $\endgroup$ – user1043065 Jun 19 '14 at 10:48
  • 1
    $\begingroup$ That the integral along the semi circle goes to 0 follows directl from the Jordan Lemma. $\endgroup$ – Zaid Alyafeai Jun 19 '14 at 12:01
  • $\begingroup$ I have used Lebesgue dominated convergence theorem for that, concluding that the function goes to zero $\lambda$ - almost everywhere. Would that also be correct? $\endgroup$ – user1043065 Jun 19 '14 at 12:05
  • $\begingroup$ I have edited the answer. The value of the integral will be the imaginary part of $2\,\pi\,i\,\text{Res}(f;i)$. $\endgroup$ – Julián Aguirre Jun 20 '14 at 11:51

Consider the function $f(t)=e^{\large-\sqrt a|t|}$, then the Fourier transform of $f(t)$ is given by $$ \begin{align} F(x)=\mathcal{F}[f(t)]&=\int_{-\infty}^{\infty}f(t)e^{-ix t}\,dt\\ &=\int_{-\infty}^{\infty}e^{-\sqrt a|t|}e^{-ix t}\,dt\\ &=\int_{-\infty}^{0}e^{\sqrt at}e^{-ix t}\,dt+\int_{0}^{\infty}e^{-\sqrt at}e^{-ix t}\,dt\\ &=\lim_{u\to-\infty}\left. \frac{e^{(\sqrt a-ix)t}}{\sqrt a-ix} \right|_{t=u}^0-\lim_{v\to\infty}\left. \frac{e^{-(\sqrt a+ix)t}}{\sqrt a+ix} \right|_{0}^{t=v}\\ &=\frac{1}{\sqrt a-ix}+\frac{1}{\sqrt a+ix}\\ &=\frac{2\sqrt a}{x^2+a}. \end{align} $$ Next, the inverse Fourier transform of $F(x)$ is $$ \begin{align} f(t)=\mathcal{F}^{-1}[F(x)]&=\frac{1}{2\pi}\int_{-\infty}^{\infty}F(x)e^{ix t}\,dx\\ e^{-\sqrt a|t|}&=\frac{1}{2\pi}\int_{-\infty}^{\infty}\frac{2\sqrt a}{x^2+a}e^{ix t}\,dx\\ \frac{\pi e^{-\sqrt a|t|}}{\sqrt a}&=\int_{-\infty}^{\infty}\frac{e^{ix t}}{x^2+a}\,dx,\tag1 \end{align} $$ where $(1)$ can be rewritten as $$ \int_{0}^{\infty}\frac{e^{ix t}}{x^2+a}\,dx=\frac{\pi e^{-\sqrt at}}{2\sqrt a}.\tag2 $$ Now differentiate $(2)$ with respect to $a$ twice and with respect to $t$ five times, take the real part, and set $a=t=1$ yields \begin{align} \Re\left[\int_{0}^{\infty}\frac{\partial^2}{\partial a^2}\frac{\partial^5}{\partial t^5}\left(\frac{e^{ix t}}{x^2+a}\right)\,dx\right]_{t=1,\,a=1}&=\left.\frac{\partial^2}{\partial a^2}\frac{\partial^5}{\partial t^5}\left(\frac{\pi e^{-\sqrt at}}{2\sqrt a}\right)\right|_{t=1,\,a=1}\\ -2\int_{0}^{\infty}\frac{x^5\sin x}{(x^2+1)^3}\,dx&=-\frac{\pi}{4e}\\ \int_{0}^{\infty}\frac{x^5\sin x}{(x^2+1)^3}\,dx&=\large\color{blue}{\frac{\pi}{8e}}. \end{align}

  • $\begingroup$ Forgot to mention that $a>0$ and $t>0$ in $(2)$. You may also refer to this link to obtain $(1)$ using double integral method (elementary way). $\endgroup$ – Tunk-Fey Jun 19 '14 at 11:59
  • $\begingroup$ Thank you very much for the effort! I am in a course of complex analysis so I think I'll use the other answer. I was still checking it so I did not yet check the answer acceptance button. sorry for that but still thanks for the effort! $\endgroup$ – user1043065 Jun 19 '14 at 12:04
  • $\begingroup$ @user1043065 It's OK, no hurt feeling from me. I just enjoy your problem, very nice. $\ddot\smile$ $\endgroup$ – Tunk-Fey Jun 19 '14 at 12:05

An alternative to the obvious canonical answer given by Julián Aguirre: it is easy to compute $$\int_{-\infty}^{\infty} \frac{e^{i b x}}{x^2+a}dx $$ for $a, b > 0$, because the integrand has simple poles only. Now differentiate with respect to $a$ twice and with respect to $b$ 5 times, and set $a = b = 1$. The theorem you need to justify the differentiation can be found in your textbook.


Note \begin{eqnarray} I&=&\int_0^\infty \frac{x[(x^2+1)-1]^2\sin x}{(1+x^2)^3}dx\\ &=&\int_0^\infty \frac{x\sin x}{1+x^2}dx-2\int_0^\infty \frac{x\sin x}{(1+x^2)^2}dx+\int_0^\infty \frac{x\sin x}{(1+x^2)^3}dx. \end{eqnarray} From $$ \int_0^\infty \frac{\cos(ax)}{b^2+x^2}dx=\frac{\pi}{2b}e^{-ab}, a>0, b>0, $$ we have \begin{eqnarray} \int_0^\infty \frac{x\sin(ax)}{b^2+x^2}dx&=&=-\frac{d}{da}\int_0^\infty \frac{\cos(ax)}{b^2+x^2}dt=\frac{\pi}{2e^{-ab}},\\ \int_0^\infty \frac{x\sin x}{(b^2+x^2)^2}dx&=&-\frac{1}{2b}\frac{d}{db}\int_0^\infty \frac{x\sin(ax)}{b^2+x^2}dt=\frac{a\pi}{4be^{-ab}},\\ \end{eqnarray} and \begin{eqnarray} \frac{d^2}{db^2}\int_0^\infty \frac{x\sin (ax)}{b^2+x^2}dx&=&8b^2\int_0^\infty \frac{x\sin(ax)}{(b^2+x^2)^3}dx-2\int_0^\infty \frac{x\sin(ax)}{(b^2+x^2)^2}dx. \end{eqnarray} The latter one implies \begin{eqnarray} \int_0^\infty \frac{x\sin(ax)}{(b^2+x^2)^3}dx=\frac{a(ab+1)\pi}{16b^3e^{-ab}}. \end{eqnarray} Thus $$ I=\frac{\pi}{8e}.$$


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.