7
$\begingroup$

Can you prove or disprove the following claim:

Claim: $$\frac{\sqrt{3} \pi}{24}=\displaystyle\sum_{n=0}^{\infty}\frac{1}{(6n+1)(6n+5)}$$

The SageMath cell that demonstrates this claim can be found here.

$\endgroup$
6
  • $\begingroup$ I think it should be possible to rewrite the sum in terms of a doubly infinite sum with n ranging over $\mathbb{Z}$. From there, I think the methods of math.stackexchange.com/questions/1322086/… will get you to a closed form. $\endgroup$ Aug 2 at 3:54
  • 14
    $\begingroup$ Write this as $\frac{1}{4} \sum_{n = 0}^{\infty} \left(\frac{1}{6n + 1} - \frac{1}{6n + 5}\right) = \frac{1}{4} \sum_{n = 1}^{\infty} \frac{\chi_6(n)}{n} = \frac{1}{4} L(1,\chi_6)$, where $\chi_6$ is the quadratic Dirichlet character modulo $6$, which satisfies $\chi_6(n) = 1$ if $n \equiv 1 \pmod{6}$, $\chi_6(n) = -1$ if $n \equiv 5 \pmod{6}$, and $\chi_6(n) = 0$ if $(n,6) \neq 1$. Then Dirichlet's class number formula implies the result. $\endgroup$ Aug 2 at 3:58
  • 1
    $\begingroup$ Another idea, related to @PeterHumphries answer, is to write his first step as the limit $s\to1$ of two Hurwitz zeta functions, and then use the known expression of this limit in terms of the digamma function evaluated at $1/6$ and $5/6$, which are related to $\sqrt{3}\pi$. $\endgroup$
    – EFinat-S
    Aug 2 at 4:09
  • 1
    $\begingroup$ FYI, a large number of such summations, nearly all of which are much more complicated, can be found in Chapters IX-X (pp. 147-200) of An Introduction to the Operations with Series by Isaac Joachim Schwatt (1924; reprinted with corrections by Chelsea Publishing Company in 1962). A google-books .pdf file is freely available where I'm at. $\endgroup$ Aug 2 at 13:56
  • 2
    $\begingroup$ Yet one more way, close to GH from MO's answer: Let $\omega$ be a primitive $6$-th root of unity and write this as $\sum_{n=1}^{\infty} \tfrac{\omega^n + \omega^{2n} - \omega^{4n} - \omega^{5n}}{n (8 \sqrt{-3})}$. Then write this as the limit as $r \to 1^-$ of $\tfrac{-1}{8 \sqrt{-3}} \left( \log(1-r \omega) + \log(1-r \omega^2) - \log(1- r \omega^4) - \log(1-r \omega^5) \right)$. (There are probably some typos in this.) $\endgroup$ Aug 2 at 17:49
27
$\begingroup$

Here is an elementary proof. We rewrite the series as $$\frac{1}{4}\int_0^1\frac{1-x^4}{1-x^6}\,dx=\frac{1}{8}\int_0^1\frac{dx}{1-x+x^2}+\frac{1}{8}\int_0^1\frac{dx}{1+x+x^2}.$$ It is straightforward to show that \begin{align*} \int_0^1\frac{dx}{1-x+x^2}&=\frac{2\pi}{3\sqrt{3}},\\ \int_0^1\frac{dx}{1+x+x^2}&=\frac{\pi}{3\sqrt{3}}, \end{align*} so we are done.

$\endgroup$
12
$\begingroup$

Let the sum be $S$. First of all, it is easy to see that $$S=\frac{1}{2}\sum_{n=-\infty}^\infty f(n),\quad\mbox{where}\quad f(z)=\frac{1}{(6z+1)(6z+5)}.$$ This is true because $f(z)=f(-1-z)$. Then by the summation formula (see any undergraduate Complex Variables textbook) $$\sum_{n=-\infty}^\infty f(n)=-\sum_a{\mathrm{res}}_a \left(f(z)\pi\cot\pi z\right),$$ where summation is over all poles of $f$, that is $a_1=-1/6$ and $a_2=-5/6$. Computing the residues (each of them equals $-\pi\sqrt{3}/24$) we obtain the result.

$\endgroup$
7
$\begingroup$

Write the sum as $S_{o}=\sum_{n=1}^{\infty} \frac{1}{(3(2n+1)-2)(3(2n+1)+2)}=\sum_{n=\text{odd}} \frac{1}{(3n)^2-(2)^2}$.

Hence, $S_o=S-S_e$. Where, for $S$ , $n$ are all natural numbers in the above expression and for $S_e$, $n$ takes only positive even numbers.

Hence, from the identity $\frac{\pi\text{cot}(x)}{x}=\frac{1}{x^2}+\sum_{n=1}^{\infty} \frac{2}{x^2-n^2}$, we get $S=\frac{9-6\pi\text{cot}(\frac{2\pi}{3})}{72}$

And, $S_e=\frac{1}{4}\sum_{n=1}^{\infty} \frac{1}{(3n)^2-1}$. This similarly gives $S_e=\frac{9-3\pi\text{cot}(\frac{\pi}{3})}{72}$.

Hence, $S_o=\frac{3\pi\text{cot}(\frac{\pi}{3})-6\pi\text{cot}(\frac{2\pi}{3})}{72}=\frac{\sqrt{3}\pi}{24}$

$\endgroup$

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.