Fancy Summation

Calculus Level 5

Let x 1 , x 2 , x 3 , x_1,\, x_2,\, x_3,\,\cdots denote all the positive solutions of the equation tan x = x \tan{x}=x .

Find n = 1 cos 2 x n \displaystyle \sum_{n=1}^{\infty}\cos^2{x_n}

e 2 2 e + 1 30 \dfrac{e^2-2e+1}{30} e + 1 e 3 e+\dfrac{1}{e}-3 e 2 7 4 \dfrac{e^2-7}{4} e 2 8 \dfrac{e-2}{8}

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

Sangchul Lee
Mar 19, 2019

Here is an Euler-themed solution. First of all, notice that cos 2 x k = 1 1 + tan 2 x k = 1 1 + x k 2 \cos^2 x_k = \frac{1}{1+\tan^2 x_k} = \frac{1}{1+x_k^2} , and so, k = 1 cos 2 ( x k ) = k = 1 1 1 + x k 2 . \sum_{k=1}^{\infty} \cos^2(x_k) = \sum_{k=1}^{\infty} \frac{1}{1+x_k^2}. Next, introduce the function f ( z ) = sin z z cos z f(z) = \frac{\sin\sqrt{z}}{\sqrt{z}} - \cos\sqrt{z} . This is an entire function of order 1/2 . Moreover, a careful inspection shows that the only zeros of f f are either 0 0 or x k 2 x_k^2 for some k N k \in \mathbb{N} , all of which are simple. So by the Hadamard factorization theorem , f ( z 1 ) = A ( z 1 ) k = 1 ( 1 z x k 2 + 1 ) . f(z-1) = A(z-1)\prod_{k=1}^{\infty} \left( 1 - \frac{z}{x_k^2 + 1} \right). (Notice that the zeros of f f are either 1 1 or x k 2 + 1 x_k^2+1 for some k N k \in \mathbb{N} .) Plugging z = 0 z = 0 determines the value of A A as 1 / e 1/e , and so, k = 1 ( 1 z x k 2 + 1 ) = e f ( z 1 ) z 1 = 1 e 2 7 4 z + O ( z 2 ) . \prod_{k=1}^{\infty} \left( 1 - \frac{z}{x_k^2 + 1} \right) = e\frac{f(z-1)}{z-1} = 1 - \frac{e^2-7}{4} z + \mathcal{O}(z^2). Now comparing the coefficient of z z in both sides, we obtain k = 1 1 1 + x k 2 = e 2 7 4 . \sum_{k=1}^{\infty} \frac{1}{1+x_k^2} = \frac{e^2-7}{4}.

This is nice!!!

Aaghaz Mahajan - 2 years, 2 months ago

Are you the same Sangchul Lee from MSE? If yes, I am a great fan of your integration skills. Your answers there are quite amazing. As far as this solution goes, I could just say one thing- Perfection at its best!!!!

Rohan Shinde - 2 years, 2 months ago

Log in to reply

Thank you! I never expected anyone to recognize me, so I'm glad to see someone who knows me :)

Sangchul Lee - 2 years, 2 months ago

Log in to reply

Genius always outshines the others... I know you from a quite long time. You have even answered some of my favourite questions very easily on MSE..... So how could I forget you....

Rohan Shinde - 2 years, 2 months ago
Mark Hennings
Feb 19, 2019

The function f ( z ) = z cos z sin z f(z) \; = \; z\cos z - \sin z has a triple zero at z = 0 z=0 , and simple zeros at z = ± x n z = \pm x_n for n N n \in \mathbb{N} . Moreover f ( x ) = z sin z f'(x) \; = \; -z\sin z . Thus the function g ( z ) = z sin z f ( z ) g(z) \; = \; \frac{z\sin z}{f(z)} is an odd function with a simple pole at z = 0 z=0 and simple poles at z = ± x n z = \pm x_n for n N n \in \mathbb{N} . Moreover R e s z = 0 g ( z ) = 3 R e s z = ± x n g ( z ) = 1 \mathrm{Res}_{z=0} g(z) \; = \; -3 \hspace{2cm} \mathrm{Res}_{z=\pm x_n}g(z) \; = \; -1 If we let C N C_N be the square with vertices N π + i N π , N π i N π , N π i N π , N π + i N π N\pi+iN\pi,N\pi-iN\pi,-N\pi-iN\pi,-N\pi+iN\pi , then we can show that there exists K > 0 K > 0 such that g ( z ) K |g(z)| \le K for all z C N z \in C_N for all positive integers N N , and this implies that lim N C N g ( z ) z 2 + 1 d z = 0 \lim_{N \to \infty}\int_{C_N} \frac{g(z)}{z^2+1}\,dz \; = \; 0 Now R e s z = i g ( z ) z 2 + 1 = g ( i ) 2 i = 1 4 ( e 2 1 ) R e s z = i g ( z ) z 2 + 1 = g ( i ) 2 i = 1 4 ( e 2 1 ) \mathrm{Res}_{z=i}\frac{g(z)}{z^2+1} \; = \; \frac{g(i)}{2i} \; = \; \tfrac14(e^2-1) \hspace{2cm} \mathrm{Res}_{z=-i} \frac{g(z)}{z^2+1} \; = \; \frac{g(-i)}{-2i} \; = \; \tfrac14(e^2-1) and so 1 2 π i C N g ( z ) z 2 + 1 d z = R e s z = 0 g ( z ) z 2 + 1 + R e s z = i g ( z ) z 2 + 1 + R e s z = i g ( z ) z 2 + 1 + 2 n = 1 N R e s z = x n g ( z ) z 2 + 1 = 3 + 1 4 ( e 2 1 ) + 1 4 ( e 2 1 ) 2 n = 1 N 1 x n 2 + 1 = 1 2 ( e 2 7 ) 2 n = 1 N cos 2 x n \begin{aligned} \frac{1}{2\pi i}\int_{C_N} \frac{g(z)}{z^2+1}\,dz & = \; \mathrm{Res}_{z=0}\frac{g(z)}{z^2+1} + \mathrm{Res}_{z=i}\frac{g(z)}{z^2+1} + \mathrm{Res}_{z=-i}\frac{g(z)}{z^2+1} + 2\sum_{n=1}^N \mathrm{Res}_{z=x_n}\frac{g(z)}{z^2+1} \\ & = \; -3 + \tfrac14(e^2-1) + \tfrac14(e^2-1) - 2\sum_{n=1}^N \frac{1}{x_n^2+1} \\ & = \; \tfrac12(e^2-7) - 2\sum_{n=1}^N \cos^2x_n \end{aligned} Letting N N \to \infty tells us that n = 1 cos 2 x n = 1 4 ( e 2 7 ) \sum_{n=1}^\infty \cos^2x_n \; = \; \boxed{\tfrac14(e^2-7)}

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...