14

show that

$$\int_{0}^{\infty} \frac {\sin^3(x)}{x^3}dx=\frac{3\pi}{8}$$

using different ways

thanks for all

mnsh
  • 5,875

8 Answers8

17

Let $$f(y) = \int_{0}^{\infty} \frac{\sin^3{yx}}{x^3} \mathrm{d}x$$ Then, $$f'(y) = 3\int_{0}^{\infty} \frac{\sin^2{yx}\cos{yx}}{x^2} \mathrm{d}x = \frac{3}{4}\int_{0}^{\infty} \frac{\cos{yx} - \cos{3yx}}{x^2} \mathrm{d}x$$ $$f''(y) = \frac{3}{4}\int_{0}^{\infty} \frac{-\sin{yx} + 3\sin{3yx}}{x} \mathrm{d}x$$ Therefore, $$f''(y) = \frac{9}{4} \int_{0}^{\infty} \frac{\sin{3yx}}{x} \mathrm{d}x - \frac{3}{4} \int_{0}^{\infty} \frac{\sin{yx}}{x} \mathrm{d}x$$

Now, it is quite easy to prove that $$\int_{0}^{\infty} \frac{\sin{ax}}{x} \mathrm{d}x = \frac{\pi}{2}\mathop{\mathrm{signum}}{a}$$

Therefore, $$f''(y) = \frac{9\pi}{8} \mathop{\mathrm{signum}}{y} - \frac{3\pi}{8} \mathop{\mathrm{signum}}{y} = \frac{3\pi}{4}\mathop{\mathrm{signum}}{y}$$ Then, $$f'(y) = \frac{3\pi}{4} |y| + C$$ Note that, $f'(0) = 0$, therefore, $C = 0$. $$f(y) = \frac{3\pi}{8} y^2 \mathop{\mathrm{signum}}{y} + D$$ Again, $f(0) = 0$, therefore, $D = 0$.

Hence, $$f(1) = \int_{0}^{\infty} \frac{\sin^3{x}}{x^3} = \frac{3\pi}{8}$$

TenaliRaman
  • 3,846
15

Use Parseval's theorem:

$$\int_{-\infty}^{\infty} dx \, f(x) g^*(x) = \frac{1}{2 \pi} \int_{-\infty}^{\infty} dk \, F(k) G^*(k)$$

where $f$, $g$ and $F$, $G$ are respective Fourier transform pairs, e.g.,

$$F(k) = \int_{-\infty}^{\infty} dx \, f(x) \, e^{i k x}$$

etc. If $f(x) = \sin{x}/x$, then

$$F(k) = \begin{cases} \pi & |k| \le 1\\0 & |k| \gt 1 \end{cases}$$

Further, if $g(x) = \sin^2{x}/x^2$, then

$$G(k) = \begin{cases}\pi \left (1-\frac{|k|}{2} \right ) & |k| \le 2 \\ 0& |k| \gt 2\end{cases}$$

Then

$$\int_{-\infty}^{\infty} dx \,\frac{\sin^3{x}}{x^3} = \frac{1}{2 \pi} \int_{-1}^1 dk \, \pi^2 \left (1-\frac{|k|}{2} \right ) = \pi - \frac{\pi}{2} \int_0^1 dk \,k = \pi-\frac{\pi}{4}$$

Therefore

$$\int_{0}^{\infty} dx \,\frac{\sin^3{x}}{x^3} = \frac{3 \pi}{8}$$

ADDENDUM

You can also use contour integration techniques. For the integral

$$\int_0^{\infty} dt \frac{\sin^3{ \pi t}}{(\pi t)^3} \cos{u t}$$

I have derived a complete solution to the problem of its evaluation here using both contour integral techniques as well as the convolution theorem. You will see that the results agree for $u=0$ by a simple rescaling of the integral.

Ron Gordon
  • 138,521
  • thanks but sorry I don't know alot about this theorem – mnsh Jul 27 '13 at 01:29
  • Fundamental for Fourier transforms; allows evaluation of a wide range of integrals that are otherwise difficult. See https://en.wikipedia.org/wiki/Fourier_transform#Plancherel_theorem_and_Parseval.27s_theorem – Ron Gordon Jul 27 '13 at 01:32
9

In this answer, the more general integral $$ \int_0^\infty\left(\frac{\sin(x)}{x}\right)^n\,\mathrm{d}x $$ is calculated.

Your integral is that integral for $n=3$.


A Different Way

In a fashion similar to this answer, we will use the equation $$ \frac{\mathrm{d}^2}{\mathrm{d}x^2}\frac{\sin^3(kx)}{k^3} =\frac{9\sin(3kx)-3\sin(kx)}{4k}\tag{1} $$ and the series for $0\lt x\le\pi$, $$ \sum_{k=1}^\infty\frac{\sin(kx)}{k}=\frac{\pi-x}{2}\tag{2} $$ Using $(2)$, we get $$ \begin{align} \sum_{k=1}^\infty\frac{9\sin(3kx)-3\sin(kx)}{4k} &=\frac94\frac{\pi-3x}{2}-\frac34\frac{\pi-x}{2}\\ &=\frac{3\pi}{4}-3x\tag{3} \end{align} $$ Integrating from $0$ twice to back out the derivatives taken in $(1)$ yields $$ \sum_{k=1}^\infty\frac{\sin^3(kx)}{k^3}=\frac{3\pi}{8}x^2-\frac12x^3\tag{4} $$ Set $x=1/n$ and multiply by $n^2$ to get $$ \sum_{k=1}^\infty\frac{\sin^3(k/n)}{k^3/n^3}\frac1n=\frac{3\pi}{8}-\frac1{2n}\tag{5} $$ and $(5)$ is a Riemann sum for $$ \int_0^\infty\frac{\sin^3(x)}{x^3}\,\mathrm{d}x=\frac{3\pi}{8}\tag{6} $$

robjohn
  • 345,667
  • where is the wrong in my step.

    we know that $$\frac{b-a}{n}\sum_{k=0}^{n-1} f\left(a+k\frac{b-a}n\right) \approx \int_a^b f(x)\ dx$$

    put a=0 b=1 and take $n\rightarrow \infty$

    $$\lim_{n\rightarrow \infty}(\frac{1}{n}\sum_{k=0}^{n-1} f\left(k\frac{1}n\right)) = \int_0^1 f(x)\ dx$$

    $$f(x)=(\frac {\sin x} {x} )^3$$

    $$\lim_{n\rightarrow \infty}(\sum_{k=0}^{\infty} \frac{1}{n} (\frac {\sin \frac{k}{n}} {\frac{k}{n}} )^3 = \int_0^1 (\frac {\sin x} {x} )^3\ dx\neq \int_0^\infty (\frac {\sin x} {x} )^3\ dx$$

    – mnsh Jul 27 '13 at 15:32
  • 3
    @hmedan.mnsh: $$ \lim_{n\to\infty}\sum_{k=0}^\infty\frac1n\left(\frac {\sin\left(\frac{k}{n}\right)} {\frac{k}{n}}\right)^3\ne\lim_{n\to\infty}\sum_{k=0}^n\frac1n\left(\frac {\sin\left(\frac{k}{n}\right)} {\frac{k}{n}}\right)^3 $$ – robjohn Jul 27 '13 at 18:09
  • 1
    @hmedan.mnsh: You are correct in the formula for the Riemann Sum on a finite interval, but there are conditions under which the same idea can be extended to infinite intervals (improper Riemann Sums). The condition that applies most easily here is that $$ \sum_{k=0}^\infty\sup_{x\in[k,k+1]}|f'(x)|\lt\infty $$ – robjohn Jul 27 '13 at 18:45
8

Well, there's a lot of creative answers but it seems that no one bothered to put here the "follow your nose" one, so I'll add it here for completeness.

Recall the identity: $$\sin^3x = \frac{3 \sin x - \sin(3x)}{4}.$$

So we have: $$\int_{-\infty}^{+\infty} \left(\frac{\sin x}{x}\right)^3\,{\rm d}x = \int_{-\infty}^{\infty} \frac{3 \sin x - \sin (3x)}{4x^3}\,{\rm d}x. $$Let $0 < r < R$. Consider $\gamma_1$ the line segment joining $r$ to $R$, $\gamma_2$ the circular arc oriented counterclockwise joining $R$ to $-R$, $\gamma_3$ the line segment joining $-R$ to $-r$ and $\gamma_4$ the circular arc oriented clockwise joining $-r$ to $r$. Consider $\gamma = \gamma_1 \ast \gamma_2 \ast \gamma_3 \ast \gamma_4$ the concatenation. Sketch:

enter image description here

Consider the function: $$ f(z) = \frac{3e^{iz}-e^{3iz}}{4z^3}. $$The only singularity is $z = 0$ (triple pole). Expanding in Laurent: $$ f(z) = \frac{1}{4z^3}\left(3\sum_{n \geq 0}\frac{i^nz^n}{n!} - \sum_{n \geq 0}\frac{3^ni^nz^n}{n!}\right) = \sum_{n \geq 0} \left(\frac{(3-3^n)i^n}{4\cdot n!}\right)z^{n-3}. $$ Since $f$ is holomorphic in $\gamma$ and inside it, by Cauchy-Goursat we get: $$ \oint_{\gamma} f(z)\,{\rm d}z = \int_{\gamma_1} f(z)\,{\rm d}z + \int_{\gamma_2} f(z)\,{\rm d}z + \int_{\gamma_3}f(z)\,{\rm d}z + \int_{\gamma_4}f(z)\,{\rm d}z = 0.$$ Let's analyze everything sistematically.

Parametrizing $\gamma_1(x) = x$, with $r \leq x \leq R$, we have: $$ \int_{\gamma_1}f(z)\,{\rm d}z = \int_r^R \frac{3e^{ix}-e^{3ix}}{4x^3}\,{\rm d}x. $$For $\gamma_2$, we have that its lenght is $\pi R$, $|e^{iz}| = e^{{\rm Re}(iz)} = e^{-{\rm Im}(z)} < 1$, and similarly $|e^{3iz}| < 1$, since for all $z$ in $\gamma_2$ we have ${\rm Im}(z) > 0$. Hence: $$ \left|\int_{\gamma_2} f(z)\,{\rm d}z\right| \leq \frac{\pi R(3+4)}{4 R^3} = \frac{\pi}{R^2} \stackrel{R\, \to \,+\infty}{\longrightarrow} 0. $$

Parametrizing $\gamma_3^-(x) = -x$, with $r \leq x \leq R$, we have: $$ \int_{\gamma_3}f(z)\,{\rm d}z =- \int_r^R \frac{3e^{-ix}-e^{-3ix}}{4x^3}\,{\rm d}x, $$once the signs in $(-x^3) = -x^3$ and ${\rm d}z = -{\rm d}x$ cancel each other. Notice here that: $$ \int_{\gamma_1}f(z)\,{\rm d}z + \int_{\gamma_3}f(z)\,{\rm d}z = 2i\int_r^R \left(\frac{\sin x}{x}\right)^3\,{\rm d}x.$$

For $\gamma_4$, we have: $$\begin{align} \int_{\gamma_4}f(z)\,{\rm d}z &= \int_{\gamma_4} \sum_{n \geq 0}\left(\frac{(3-3^n)i^n}{4 \cdot n!}\right)z^{n-3}\,{\rm d}z \\ &= \int_{\gamma_4}\frac{1}{2z^3}\,{\rm d}z + \int_{\gamma_4}\frac{3}{4z}\,{\rm d}z + \int_{\gamma_4} \sum_{n \geq 3} \left(\frac{(3-3^n)i^n}{4 \cdot n!}\right)z^{n-3}\,{\rm d}z \end{align}$$

We have: $$ \int_{\gamma_4} \frac{1}{2z^3}\,{\rm d}z = -\frac{1}{4z^2}\Bigg|_{-r}^{r} = 0, \quad \int_{\gamma_4} \frac{3}{4z}\,{\rm d}z = \frac{3}{4}\int_{\gamma_4} -i \frac{{\rm d}z}{-iz} = \frac{3}{4}\int_0^\pi -i\,{\rm d}t = -\frac{3\pi i}{4}, $$ and: $$ \left|\int_{\gamma_4} \sum_{n \geq 3}\frac{(3-3^n)i^n}{4\cdot n!}z^{n-3}\right| \leq \pi r \sum_{n \geq 3}\frac{3^n-3}{4\cdot n!}r^{n-3} = \sum_{n \geq 3}\frac{\pi(3^n-3)}{4 \cdot n!}r^{n-2} \stackrel{r \to 0}{\longrightarrow} 0. $$

Making first $r \to 0$, and then $R \to +\infty$, in $\oint_\gamma f = 0$, we get: $$ 2i\int_0^{+\infty}\left(\frac{\sin x}{x}\right)^3\,{\rm d}x - \frac{3\pi i}{4} = 0 \implies \int_0^{+\infty}\left(\frac{\sin x}{x}\right)^3\,{\rm d}x = \frac{3\pi}{8}, $$as desired.

Ivo Terek
  • 77,665
3

Related technique. You can use the Laplace transform technique. Recalling the Laplace transform

$$F(s)= \int_{0}^{\infty} f(x) e^{-sx}dx. $$

Taking $ f(x) = \frac{\sin(x)^3}{x^3} $ gives

$$ F(s)= \frac{\pi \,{s}^{2}}{8}+\frac{3\,\pi}{8}- \frac{3( {s}^{2}-1) }{8}\,\arctan \left( s \right) +\frac{( {s}^{2}-9)}{8}\,\arctan \left( \frac{s}{3} \right) $$ $$+\frac{3s}{8}\, \left( -\ln \left( {s}^{2}+9 \right) +\ln \left( {s} ^{2}+1 \right) \right). $$

Taking the limit as $s\to 0$ gives the desired result $\frac{3\pi}{8}$.

Another Laplace transform approach: Referring to the problem, we can use the following relation

$$ \begin{align} \int_0^\infty F(u)g(u) \, du & = \int_0^\infty f(u)G(u) \, du \\[6pt] L[f(t)] & = F(s) \\[6pt] L[g(t)] & = G(s)\end{align} $$

Let

$$ G(u)=\frac{1}{u^3} \implies g(u)=\frac{u^2}{2!}, $$

and

$$ f(u)= \sin(u)^3 \implies F(u) = {\frac {6}{ \left( {u}^{2}+1 \right) \left( {u}^{2}+9 \right) }}. $$

Now,

$$ \int_0^\infty \frac{\sin^3 x}{x^3} \, dx = \frac{6}{2}\int_0^\infty \frac{u^2}{\left( {u}^{2}+1 \right) \left( {u}^{2}+9 \right)} \, du = \frac{3\pi}{8}$$.

  • 1
    thanks but can you show me step by step the laplace transform of $\frac{\sin(x)^3}{x^3}$ – mnsh Jul 27 '13 at 14:41
  • @hmedan.mnsh: The link has the technique. – Mhenni Benghorbal Jul 27 '13 at 15:48
  • I under stand the technique but I faild to know the laplace transform of $(\frac{\sin x} {x} )^3$ I started with $ \sin x =\frac{e^{ix} -e^{-ix}} {2i} $ but it become complicated – mnsh Jul 27 '13 at 16:33
  • @hmedan.mnsh: I think what Mhenni is trying to say is that you take the 3rd derivative of the LT expression to get the LT of $\sin^3{x}$, which may be evaluated as a sum of simple integrals. Of course, then you need to integrate the resulting expression three times and make sure you include all (implied) initial conditions. The result is presumably what Mhenni has provided, although I am not going to check it for correctness. – Ron Gordon Jul 27 '13 at 23:06
  • @RonGordon oh thanks alot MR GORDON AND MR BENGHORBAL – mnsh Jul 27 '13 at 23:40
  • @RonGordon I got $$F'''(s)=-\frac{6}{(1+s^2)(9+s^2)}$$

    so I integrated 3 times and take $s\rightarrow 0$ and the answer is 0 and I dont know where is the wrong and note that every time i integrate I take th constant=0 and every thing I did was true but it equal to 0

    – mnsh Jul 27 '13 at 23:57
  • 1
    @hmedan.mnsh: Mhenni has some explaining to do. Your expression is right (save for the sign, but never mind for now). Integrating this three times gives an expression that goes to zero as $s \to 0$. The integration constants provide a quadratic in $s$ which goes to the constant term from the last integration as $s \to 0$. Thus, the integration constant is...the original integral! In other words, as far as I can see, this method is completely useless in deriving the desired result. This is not the first time that Mhenni has provided a "hint" that was poorly thought out. – Ron Gordon Jul 28 '13 at 02:05
  • 1
    @MhenniBenghorbal: you owe the OP an explanation of how you got the $3 \pi/8$ term. I worked the integral out according to your technique specified in the link, and the limit as $s \to 0$ of the result produces an integration constant whose value is...the result we seek. Please explain how you got this term, because I cannot; as far as I can see, this method simply is of no use in deriving the result. – Ron Gordon Jul 28 '13 at 02:11
  • @RonGordon: This is what you think. Thanks for the comment. – Mhenni Benghorbal Jul 28 '13 at 09:23
  • @hmedan.mnsh: I recommend you to take your time to learn Laplace transform transform and then start looking at this problem. I posted this answer because you asked for different ways. The answer is correct. – Mhenni Benghorbal Jul 28 '13 at 09:26
  • @Downvoter: It is an honesty issue to identify your self. The answer is correct. – Mhenni Benghorbal Jul 28 '13 at 09:27
  • @RonGordon: By the way, thanks for answering the OP. – Mhenni Benghorbal Jul 28 '13 at 10:06
  • 1
    @MhenniBenghorbal Iam up voter not down voter and the way is very beautiful and for that iam going to learn more and more about laplace transform – mnsh Jul 28 '13 at 12:15
  • 1
    @MhenniBenghorbal: did you read my comment to you? I went through the methodology that I explained to the OP for you (something you should have done), and failed to get a meaningful answer. In my opinion, the $3 \pi/8$ in your expression is a fudge. I may be wrong, but I just do not see how you get your expression without knowing the answer first. Please point out where I am wrong, if I am. – Ron Gordon Jul 28 '13 at 13:58
  • @hmedan.mnsh: I wish you a good luck. Always, take your time to understand things. By the way I answered another question for you where I used the Mellin transform. – Mhenni Benghorbal Jul 28 '13 at 14:00
  • @RonGordon: Why do you think it is a fudge. Just because you do not see it. As I said we need to take our time to understand things and we should not rush to judge things. Thanks for the comment. – Mhenni Benghorbal Jul 28 '13 at 14:04
  • 3
    @MhenniBenghorbal: Then prove me wrong. How did you derive this Laplace transform without needing to evaluate the original integral we were supposed to solve? I did this and came to that the the only way to get your result was to write $3 \pi/8$ for the result of the integral without deriving it. I would love to be proven wrong, but you seem to not want to do this. By the way, the downvote is not about the right answer: the OP could just go to Wolfram|Alpha for that. It is about a useful technique, which this does not seem to be. Please show us something. – Ron Gordon Jul 28 '13 at 14:12
  • @hmedan.mnsh: I agree that Laplace transform techniques can be beautiful, but this is not such a case. LT techniques usually involve inverting the transform of accumulated functions to obtain a new function. In this case, the LT was simply plopped down without justification, not a valid technique. This is not an answer, and unless Mhenni can show how one evaluates the LT without needing to evaluate your integral, I am going to recommend that it be flagged as a very low quality answer. – Ron Gordon Jul 28 '13 at 14:39
  • @MhenniBenghorbal: the other technique is way, way better and actually very useful. You should still delete your first one, though. And you may wish to show the OP how to evaluate that integral. – Ron Gordon Jul 29 '13 at 14:12
  • 1
    (+1) The second technique is amazing and makes the whole integral brilliantly simple! – TenaliRaman Aug 04 '13 at 22:38
  • 1
    @TenaliRaman: Thank you very much for the comment. I really appreciate it. – Mhenni Benghorbal Aug 05 '13 at 02:36
  • @TenaliRaman: Thanks for editing my post. – Mhenni Benghorbal Aug 05 '13 at 02:41
2

$$ \begin{aligned} \int_{0}^{\infty} \frac{\sin ^{3} x}{x^{3}} d x =& \frac{1}{4} \int_{0}^{\infty} \frac{3 \sin x-\sin 3 x}{x^{3}} d x \\ \stackrel{IBP }{=} & \frac{1}{8} \int_{0}^{\infty} \frac{(3 \sin x-\sin 3 x)^{(2)}}{x} d x\\ =&\frac{1}{8} \int_{0}^{\infty} \frac{-3 \sin x+9 \sin 3 x}{x} d x \\ =&\frac{1}{8}\left(-\frac{3 \pi}{2}+\frac{9 \pi}{2}\right) \\ =&\frac{3 \pi}{8} \end{aligned} $$

Lai
  • 20,421
1

Define $\displaystyle{% {\cal F}\left(\mu\right) \equiv \int_{-\infty}^{\infty}{\sin^{3}\left(\mu x\right) \over x^{3}}\,{\rm d}x\,, \quad ? = {1 \over 2}\,{\cal F}\left(1\right)}$

\begin{align} {\cal F}'\left(\mu\right) &= \int_{-\infty}^{\infty} {3\sin^{2}\left(\mu x\right)\cos\left(\mu x\right)x \over x^{3}}\,{\rm d}x = {3 \over 2}\int_{-\infty}^{\infty} {\cos\left(\mu x\right) - \cos\left(2\mu x\right)\cos\left(\mu x\right) \over x^{2}}\,{\rm d}x \\[3mm]&= {3 \over 2}\int_{-\infty}^{\infty} {\cos\left(\mu x\right) - \left\lbrack\cos\left(3\mu x\right) + \cos\left(\mu x\right)\right\rbrack/2 \over x^{2}}\,{\rm d}x = {3 \over 4}\int_{-\infty}^{\infty} {\cos\left(\mu x\right) - \cos\left(3\mu x\right) \over x^{2}}\,{\rm d}x \end{align} \begin{align} -&------------------------------ \end{align} \begin{align} {\cal F}''\left(\mu\right) &= {3 \over 4}\int_{-\infty}^{\infty} {-\sin\left(\mu x\right)x + \sin\left(3\mu x\right)\left(3x\right) \over x^{2}} \,{\rm d}x = {3 \over 2}\,{\rm sgn}\left(\mu\right)\int_{-\infty}^{\infty}{\sin\left(x\right) \over x} \,{\rm d}x \\[3mm]&= {3 \over 2}\,{\rm sgn}\left(\mu\right) \int_{-\infty}^{\infty} \left({1 \over 2}\int_{-1}^{1}{\rm e}^{{\rm i}kx}\,{\rm d}k\right) \,{\rm d}x = {3 \over 2}\,\pi\,{\rm sgn}\left(\mu\right) \int_{-1}^{1}\left(\int_{-\infty}^{\infty}{\rm e}^{{\rm i}kx} \,{{\rm d}x \over 2\pi}\right){\rm d}k \\[3mm]&= {3 \over 2}\,\pi\,{\rm sgn}\left(\mu\right) \int_{-1}^{1}\delta\left(k\right){\rm d}k = {3 \over 2}\,\pi\,{\rm sgn}\left(\mu\right) = {3 \over 2}\,\pi\,{{\rm d}\left\vert \mu\right\vert \over {\rm d}\mu} \end{align} \begin{align} -&------------------------------ \end{align} \begin{align} {\cal F}'\left(\mu\right) & = {3 \over 2}\,\pi\,\left\vert\mu\right\vert\ \Longrightarrow\ {\cal F}\left(\mu\right) = {3 \over 2}\,\pi\int_{0}^{\mu}\left\vert\mu'\right\vert\,{\rm d}\mu' = {3 \over 2}\,\pi\,{\rm sgn}\left(\mu\right)\int_{0}^{\mu}\mu'\,{\rm d}\mu' = {3 \over 4}\,\pi\,\left\vert\mu\right\vert\mu \end{align}

$$ \begin{array}{|c|}\hline\\ \color{#ff0000}{\large\quad% \int_{0}^{\infty}{\sin^{3}\left(x\right) \over x^{3}}\,{\rm d}x = {1 \over 2}\,{\cal F}\left(1\right) = {3\pi \over 8}\quad} \\ \\ \hline \end{array} $$

Felix Marin
  • 89,464
0

Using the formula found in my answer, $$ \begin{aligned} \int_{0}^{\infty} \frac{\sin ^{3} x}{x^{3}} &=\frac{\pi}{2^{3} \cdot 2 !}\left[\left(\begin{array}{l} 3 \\ 0 \end{array}\right) 3^{2}-\left(\begin{array}{l} 3 \\ 1 \end{array}\right) 1^{2}\right] \\ &=\frac{3 \pi}{8} \end{aligned} $$

Lai
  • 20,421