4
$\begingroup$

Could you explain to me why

$$\frac{\sin x}{x} =\left(1-\frac{x^2}{\pi ^2}\right)\left(1-\frac{x^2}{(2 \pi) ^2}\right)\left(1-\frac{x^2}{(3 \pi )^2}\right)\cdots$$

I've read in this article http://twoplusonet.wordpress.com/2011/06/24/an-elegant-result/

that firstly we write

$$\frac{\sin x}{x}\text{ as a polynomial }k\left(1-\frac{x}{a_1}\right)\left(1-\frac{x}{a_2}\right)\left(1-\frac{x}{a_3}\right)\cdots\tag{*}$$

Then, due to the fact that sin$x$ has roots at $_-^+ \pi, _-^+ 2\pi,\ldots$, we can write

$$k\left(1-\frac{x}{\pi}\right)\left(1+\frac{x}{\pi}\right)\left(1-\frac{x}{2 \pi}\right)\cdots\tag{**}$$

Since $\lim _{x\rightarrow 0} \frac {\sin x}{x}=1$, we have $k=1$ . Using the difference of two squares we get the formula above.

My question is: (why) is that enough? In the article mentioned before the author says that between writing $\frac{\sin x}{x}$ as a polynomial * and (**) we make a "little jump of faith (and let the analysts deal with the consequences)". Does that mean something has been omitted here?

$\endgroup$
  • 5
    $\begingroup$ Polynomials, by definition, have a finite number of roots. You are trying to get an infinite product expansion for $\sin$, and there are many details that have been left out in the blog post you linked to. So yes, much has been omitted. $\endgroup$ – Potato Feb 21 '13 at 9:07
  • 3
    $\begingroup$ What Potato said is very true. But, also, shouldn't the "difference of two squares" mean that every $x$ in the first expression should really be an $x^2$? $\endgroup$ – Patch Feb 21 '13 at 9:35
  • $\begingroup$ Yes, you are right. I'll correct it right now. $\endgroup$ – Hagrid Feb 21 '13 at 10:49
  • $\begingroup$ Thank you for your comments. Do you know where I could find a more detailed version of this? $\endgroup$ – Hagrid Feb 21 '13 at 11:42
  • $\begingroup$ @Potato : Is that really "by definition"? $\endgroup$ – Michael Hardy Feb 21 '13 at 15:14
7
$\begingroup$

The result is actually quite deep.

The function $\sin(x)$ extends to the complex plane as the function $z\mapsto \sin z = \frac{\exp(iz) - \exp(-iz)}{2}$ where $\exp$ is the complex exponential. This function is in particular holomorphic on the entire complex plane.

Since it has a simple zero at $z = 0$, we can divide $\sin(z) / z$ to recover another holomorphic function on the complex plane. Then we can apply the Weierstrass factorization theorem (in fact, the Hadamard form), using that by definition $\sin z$ has order 1 as an entire function, which implies that

$$ \frac{\sin(z)}{z} = G(z) \cdot \prod_{k = 1}^\infty \left( 1 - \frac{z^2}{(k\pi)^2}\right)$$

where $G(z) = \exp (az + b)$ is a nowhere vanishing entire function. Observing that $\sin z / z$ is an even function of $z$ (that is: $\sin(-z)/(-z) = \sin z / z$) we see that $G(z)$ must be even, requiring that $a = 0$. The constant $b$ can be fixed by evaluating both sides at $z = 0$.

$\endgroup$

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.