14
$\begingroup$

I ran across an interesting series in a paper written by J.W.L. Glaisher. Glaisher mentions that it is a known formula but does not indicate how it can be derived.

I think it is difficult.

$$\sum_{k=2}^\infty \frac{\log(k)}{k}\sin(2k \mu \pi) = \pi \left(\log(\Gamma(\mu)) +\frac{1}{2}\log \sin(\pi \mu)-(1-\mu)\log(\pi)- \left(\frac{1}{2}-\mu\right)(\gamma+\log 2)\right)$$

Can someone suggest a method of attack?

$\gamma$ is the Euler-Mascheroni Constant.

Thank You!

$\endgroup$
  • $\begingroup$ $\mu$ is an arbitrary constant? $\endgroup$ – J. M. is a poor mathematician Apr 8 '13 at 13:34
  • $\begingroup$ @J.M: Yes, its an arbitrary constant. $\endgroup$ – Shobhit Bhatnagar Apr 8 '13 at 13:36
  • 2
    $\begingroup$ You might want to link to it in the question... $\endgroup$ – J. M. is a poor mathematician Apr 8 '13 at 13:47
  • 2
    $\begingroup$ Maybe it can be done like this: the right-hand side is a certain function of $\mu$. Show that the left-hand side is the Fourier series of that function. $\endgroup$ – GEdgar Apr 10 '13 at 12:59
  • 1
    $\begingroup$ Note it is valid only for $0 < \mu < 1$. The left side has period $1$, but the right side doesn't. $\endgroup$ – GEdgar Apr 12 '13 at 16:01
13
+300
$\begingroup$

It suffices to do these integrals: $$ \begin{align} \int_0^1 \log(\Gamma(s))\;ds &= \frac{\log(2\pi)}{2} \tag{1a}\\ \int_0^1 \log(\Gamma(s))\;\cos(2k \pi s)\;ds &= \frac{1}{4k},\qquad k \ge 1 \tag{1b}\\ \int_0^1 \log(\Gamma(s))\;\sin(2k \pi s)\;ds &= \frac{\gamma+\log(2k\pi)}{2k\pi},\qquad k \ge 1 \tag{1c} \\ \int_0^1 \frac{\log(\sin(\pi s))}{2}\;ds &= \frac{-\log 2}{2} \tag{2a} \\ \int_0^1 \frac{\log(\sin(\pi s))}{2}\;\cos(2k \pi s)\;ds &= \frac{-1}{4k},\qquad k \ge 1 \tag{2b} \\ \int_0^1 \frac{\log(\sin(\pi s))}{2}\;\sin(2k \pi s)\;ds &= 0,\qquad k \ge 1 \tag{2c} \\ \int_0^1 1 \;ds &= 1 \tag{3a} \\ \int_0^1 1 \cdot \cos(2k \pi s)\;ds &= 0,\qquad k \ge 1 \tag{3b} \\ \int_0^1 1 \cdot \sin(2k \pi s)\;ds &= 0,\qquad k \ge 1 \tag{3c} \\ \int_0^1 s \;ds &= \frac{1}{2} \tag{4a} \\ \int_0^1 s \cdot \cos(2k \pi s)\;ds &= 0,\qquad k \ge 1 \tag{4b} \\ \int_0^1 s \cdot \sin(2k \pi s)\;ds &= \frac{-1}{2k\pi},\qquad k \ge 1 \tag{4c} \end{align} $$ Then for $f(s) = \pi \left(\log(\Gamma(s)) +\frac{1}{2}\log \sin(\pi s)-(1-s)\log(\pi)- \left(\frac{1}{2}-s\right)(\gamma+\log 2)\right)$, we get $$ \begin{align} \int_0^1 f(s)\;ds &= 0 \\ 2\int_0^1f(s) \cos(2k\pi s)\;\;ds &= 0,\qquad k \ge 1 \\ 2\int_0^1f(s) \sin(2k\pi s)\;\;ds &= \frac{\log k}{k},\qquad k \ge 1 \end{align} $$ and the formula follows as a Fourier series: $$ f(s) = \sum_{k=1}^\infty \frac{\log k}{k}\;\sin(2 k\pi s),\qquad 0 < s < 1. $$

reference

Gradshteyn & Ryzhik, Table of Integrals Series and Products

(1a) 6.441.2
(1b) 6.443.3
(1c) 6.443.1
(2a) 4.384.3
(2b) 4.384.3
(2c) 4.384.1

$\endgroup$
0
$\begingroup$

That is the well known Fourier-expansion of $\log\Gamma(x)$ doing by Kummer in 1847. The article of Glaisher is from 1893. See https://de.wikipedia.org/wiki/Gammafunktion

$\endgroup$

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.