1
$\begingroup$

I have a rough idea of approach to prove the Borwein integral in $(1)$ via Complex-Analytic techniques, is it valid ?

$(1)$

$$ \int_{0}^{\infty}\frac{\sin(x)}{x}\frac{\sin(x/3)}{x/3}dx = \frac{\pi}{2}.$$

From glancing at it seems one can consider an Indented Contour $\Gamma_{R}$, and define in $(1.2)$

$(1.2)$

$$\Gamma_{R}^{1}(t) = t \, \, \, \, \text{if} \, \, \, -R \leq t \leq -1/R $$

$$\Gamma_{R}^{2}(t) = e^{it}/R \, \, \, \, \text{if} \, \, \, \pi \leq t \leq 2 \pi$$

$$\Gamma_{R}^{3}(t) = e^{it}/R \, \, \, \, \text{if} \, \, \, \pi \leq t \leq 2 \pi$$

$$ \Gamma_{R}^{4}(t) = Re^{it} \, \, \, \, \text{if} \, \, \, 0 \leq t \leq \pi$$

Our original integral $\Psi(x) = \frac{\sin(x)}{x}\frac{\sin(x/3)}{x/3}$ can be defined on $\Gamma_{R}$ as $\Psi(z)=\frac{e^{iz}}{z}\frac{e^{iz}}{z/3}$ so our original integral now becomes

$$I = \oint_{\Gamma_{R}}\frac{e^{iz}}{z}\frac{e^{iz}}{z/3}dz; $$

Then it seems via Cauchy's Theorem one has in $(1.3)$

$(1.3)$ $$\, \, \, \, \, \, \lim_{R \rightarrow \infty} \bigg(\oint_{\Gamma_{R}^{1}} \Psi(x)dx + \oint_{\Gamma_{R}^{2}} \Psi(z)dz + \oint_{\Gamma_{R}^{3}}\Psi(x)dx + \oint_{\Gamma_{R}^{4}} \Psi(z)dz \bigg) = 0 $$

To finish the argument it seems one would have to invoke Jordan's Lemma on the last integral in $(1.3)$, then parametrize the integral over $\Gamma_{R}^{4}$ then in $(1.4)$ we have

$(1.4)$

$$\lim_{R \rightarrow \infty} \oint_{\Gamma_{R}}\frac{e^{iz}}{z}\frac{e^{iz}}{z/3}dz \rightarrow \int_{0}^{\pi}\frac{e^{ire^{i\theta}}}{re^{i\theta}}\frac{e^{ire^{i\theta}}}{re^{i\theta}/3}d\theta = \frac{\pi}{2}$$

$\endgroup$
  • $\begingroup$ I think the initial "sketch" is wrong can anyone tell me why ? $\endgroup$ – Zophikel Feb 25 '18 at 4:18
  • 2
    $\begingroup$ I do not get how $\sin(x)\sin(x/3)$ is converted into $e^{iz} e^{iz}$, for instance. $\endgroup$ – Jack D'Aurizio Feb 25 '18 at 17:53
  • $\begingroup$ Yes I believe that's where the issue is I think $sin(x)sin(x/3)$ would be $e^{iz}e^{iz/3}$ I think that would remedy the issue. $\endgroup$ – Zophikel Feb 25 '18 at 17:55
  • 1
    $\begingroup$ No, it does not fix it either. The imaginary part of $e^{iz} e^{iz/3}$ is not $\sin(z)\sin(z/3)$, but $\sin(z)\sin(z/3)$ is the real part of $\frac{1}{2}\left(e^{2iz/3}-e^{4iz/3}\right).$ $\endgroup$ – Jack D'Aurizio Feb 25 '18 at 17:58
  • $\begingroup$ Then what would it be I'm not sure how to define the integrand on the contour $\endgroup$ – Zophikel Feb 25 '18 at 18:00
3
$\begingroup$

As an alternative to finding a suitable integration contour, the Laplace transform often deserves a thought. For any $a,b>0$ we have $$\begin{eqnarray*} I(a,b)&=&\int_{0}^{+\infty}\frac{\sin(ax)\sin(bx)}{x^2}\,dx=\int_{0}^{+\infty}\frac{\cos((a-b)x)-\cos((a+b)x)}{2x^2}\,dx\\&=&J(a+b)-J(a-b)\end{eqnarray*} $$ where $$ J(c)=\int_{0}^{+\infty}\frac{1-\cos(cx)}{2x^2}\,dx = \int_{0}^{+\infty}\frac{\sin^2\left(\frac{c}{2}x\right)}{x^2}\,dx\stackrel{\mathcal{L},\mathcal{L}^{-1}}{=}\int_{0}^{+\infty}\frac{c^2}{2c^2+2s^2}\,ds=\frac{\pi}{4}|c| $$ ensures: $$ I(a,b) = \frac{\pi}{4}\left(|a+b|-|a-b|\right)=\color{red}{\frac{\pi}{2}\min(a,b).}$$

$\endgroup$
  • $\begingroup$ do you know what the error was in my attempted proof-outline I think I defined the integrand on the contour incorrectly $\endgroup$ – Zophikel Feb 25 '18 at 17:49
1
$\begingroup$

Perhaps you might be interested in seeing a real method used to evaluate this integral.

We begin by enforcing a substitution of $x \mapsto 3x$. This gives \begin{align*} \int_0^\infty \frac{\sin x}{x} \frac{\sin (x/3)}{x/3} \, dx &= \int_0^\infty \frac{\sin (3x) \sin x}{x^2} \, dx\\ &= 3 \int_0^\infty \frac{\sin^2 x}{x^2} \, dx - 4 \int_0^\infty \frac{\sin^4 x}{x^2} \, dx, \tag1 \end{align*} since $\sin (3x) = 3 \sin x - 4 \sin^3 x$.

For the first of these integrals, as is well-known $$\frac{\pi}{2} = \int_0^\infty \frac{\sin x}{x} \, dx,$$ (this is just the Dirichlet integral) enforcing a substitution of $x \mapsto 2x$ gives \begin{align*} \frac{\pi}{2} &= \int_0^\infty \frac{\sin (2x)}{x} \, dx = \int_0^\infty \frac{2 \sin x \cos x}{x} \, dx\ = \int_0^\infty \frac{(\sin^2 x)'}{x} \, dx\\ &= \left [\frac{\sin^2 x}{x} \right ]_0^\infty + \int_0^\infty \frac{\sin^2 x}{x^2} \, dx = \int_0^\infty \frac{\sin^2 x}{x^2} \, dx. \end{align*}

And for the second of the integrals, as we just found above $$\frac{\pi}{2} = \int_0^\infty \frac{\sin^2 x}{x^2}dx.$$ Enforcing a substitution of $x \mapsto 2x$ gives $$\frac{\pi}{2} = \frac{1}{2} \int_0^\infty \frac{\sin^2 (2x)}{x^2} \, dx.$$ Since $$\sin^2 (2x) = 4 \cos^2 x \sin^2 x = 4(1 - \sin^2 x)\sin^2 x = 4 \sin^2 x - 4 \sin^4 x,$$ we have $$\frac{\pi}{2} = 2 \int_0^\infty \frac{\sin^2 x}{x^2} \, dx - 2 \int_0^\infty \frac{\sin^4 x}{x^2} \, dx = 2 \cdot \frac{\pi}{2} - 2 \int_0^\infty \frac{\sin^4 x}{x^2} \, dx,$$ or $$\int_0^\infty \frac{\sin^4 x}{x^2}dx = \frac{\pi}{4}.$$

So returning to (1) we have $$\int_0^\infty \frac{\sin x}{x} \frac{\sin (x/3)}{x/3} \, dx = 3 \cdot \frac{\pi}{2} - 4 \cdot \frac{\pi}{4} = \frac{\pi}{2},$$ as required to show.

$\endgroup$
  • $\begingroup$ I'm interested in whether my proof sketch of $(1)$ is wrong I think I missed up defining the integrand on the $\Gamma_{R}$ $\endgroup$ – Zophikel Feb 25 '18 at 13:01

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.