8
$\begingroup$

How can we show that

$$\lim_{n\to\infty}\int_{-\infty}^{\infty}{\sin(n+0.5)x\over \sin(x/2)}\cdot{\mathrm dx\over 1+x^2}=\pi\cdot{e+1\over e-1}\tag1$$

$(1)$, substitution doesn't work, either integration by parts.

We know $(2)$

$$\int_{-\infty}^{\infty}{\mathrm dx\over 1+x^2}=\pi\tag2$$

$${\sin(n+0.5)x\over \sin(x/2)}={\sin(nx)\cos(x/2)+\sin(x/2)\cos(nx)\over \sin(x/2)}\tag3$$

Simplified to

$$=\sin(nx)\coth(x/2)+\cos(nx)\tag4$$

$$\lim_{n\to\infty}\int_{-\infty}^{\infty}\sin(nx)\cot(x/2)\cdot{\mathrm dx\over 1+x^2}+\int_{-\infty}^{\infty}\cos(nx)\cdot{\mathrm dx\over 1+x^2}=\pi\cdot{e+1\over e-1}\tag5$$

$$\lim_{n\to\infty}\int_{-\infty}^{\infty}\sin(nx)\cot(x/2)\cdot{\mathrm dx\over 1+x^2}+{\pi\over e^n}=\pi\cdot{e+1\over e-1}\tag6$$

$$\lim_{n\to\infty}\int_{-\infty}^{\infty}\sin(nx)\cot(x/2)\cdot{\mathrm dx\over 1+x^2}=\pi\cdot{e+1\over e-1}\tag7$$

I am not sure how to continue

$\endgroup$
  • $\begingroup$ do you know the method of stationary phase? $\endgroup$ – tired Feb 3 '17 at 20:20
12
+100
$\begingroup$

Here is one quick way. Note that

$${\sin(n+0.5)x\over \sin(x/2)}$$

is a Dirichlet kernel and we may use the identity (which can easily be proven)

$${\sin(n+0.5)x\over \sin(x/2)} = 1 + 2\sum_{k=1}^n \cos (kx)$$ to rewrite the integral as

\begin{align} I&= \lim_{n\to\infty}\int_{-\infty}^{\infty}{\sin(n+0.5)x\over \sin(x/2)}\cdot{\mathrm dx\over 1+x^2}=\lim_{n\to\infty}\int_{-\infty}^{\infty}\left[1 + 2\sum_{k=1}^n \cos (kx)\right]\cdot{\mathrm dx\over 1+x^2}\\ &=\int_{-\infty}^{\infty}{\mathrm dx\over 1+x^2} + 2\sum_{k=1}^{\infty }\int_{-\infty}^{\infty}{\cos (kx) \over (1+x^2)} \end{align}

The latter integral can be evaluated using residues to get

$$I = \pi + 2\pi \sum_{k=1}^{\infty }e^{-k}=\pi + \frac{2\pi}{e-1}=\pi \frac{e+1}{e-1}$$

$\endgroup$

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.