Hello regarding this integral: Integral $\int_0^1 \frac{\sqrt x \ln x} {x^2 - x+1}dx$ The following conjecture comes: $$\psi_1\left(\frac{11}{12}\right)-\psi_1\left(\frac{5}{12}\right)=4\sqrt 3 \pi^2-80G$$ Where $G$ is Catalan's constant. How can we show this? Wolfram-alpha agrees on this: https://www.wolframalpha.com/input/?i=trigamma(11%2F12)-trigamma(5%2F12)%3D4sqrt3pi%5E2-80Catalan

up vote 1 down vote accepted

\begin{align} \psi_1(\tfrac{11}{12}) -\psi_1(\tfrac{5}{12}) &=4\sqrt 3 \pi^2-80G \tag{1}\label{1} . \end{align}

The constant $G$ - Catalan's constant has known to appear in relations \begin{align} \psi_1(\tfrac14)&=\pi^2+8G ,\\ \psi_1(\tfrac34)&=\pi^2-8G , \end{align}

so we can try to start from $\psi_1(\tfrac14)$.

Applying triplication identity

\begin{align} 9\psi_1(3x) &= \psi_1(x) +\psi_1(x+\tfrac13) +\psi_1(x+\tfrac23) \tag{2}\label{2} \end{align}

to $\psi_1(\tfrac14)=\psi_1(3\cdot\tfrac1{12})$, we get

\begin{align} 9\psi_1(\tfrac14) &= \psi_1(\tfrac1{12}) +\psi_1(\tfrac5{12}) +\psi_1(\tfrac34) ,\\ \psi_1(\tfrac1{12}) +\psi_1(\tfrac5{12}) &=9\psi_1(\tfrac14)-\psi_1(\tfrac34) \\ &=10\psi_1(\tfrac14)-(\psi_1(\tfrac14)+\psi_1(\tfrac34)) \\ &=10\psi_1(\tfrac14)-2\pi^2 \\ &=10(\pi^2+8G)-2\pi^2 \\ \psi_1(\tfrac1{12}) +\psi_1(\tfrac5{12}) &=8\pi^2+80G , \end{align} and we are almost there.

Now, applying the identity

\begin{align} \psi_1(1-x)&= \frac{\pi^2}{\sin(\pi x)^2} -\psi_1(x) \end{align}

to $\psi_1(\tfrac1{12})=\psi_1(1-\tfrac{11}{12})$,

we get \begin{align} \psi_1(\tfrac1{12}) &=\frac{\pi^2}{\sin(\tfrac\pi{12})^2} -\psi_1(\tfrac{11}{12}) ,\\ &=8\pi^2+4\sqrt3\pi^2 -\psi_1(\tfrac{11}{12}) , \end{align}

and \eqref{1} follows.

  • could you give me a link maybe from where to find that triplication formula? – Zacky Jun 3 at 18:57
  • @Zacky: Consider $9\sum_{n=0}^\infty\frac1{(3z+n)^2}=\sum_{n=0}^\infty\frac1{(z+\frac{n}3)^2}$ $=\sum_{k=0}^\infty\frac1{(z+\frac{3k}3)^2} +\frac1{(z+\frac{3k+1}3)^2}+\frac1{(z+\frac{3k+2}3)^2}=\dots$. – g.kov Jun 4 at 1:12
  • Sir, Is there a problem if I share your solution on other groups? – Zacky Aug 8 at 21:37
  • @Zacky: Not sure, what do you mean, this whole site is for sharing solutions (and questions), so if you want to share it - that's great. – g.kov Aug 8 at 22:25

$$\psi_1(z)=\sum_{n=0}^\infty\frac1{(z+n)^2}.$$

Consider the series $$S_1=\sum_{k=0}^\infty\left(\frac1{(12k+1)^2} +\frac1{(12k+5)^2}-\frac1{(12k+7)^2}-\frac1{(12k+11)^2}\right)$$ and $$S_2=\sum_{k=0}^\infty\left(\frac1{(12k+1)^2} -\frac1{(12k+5)^2}-\frac1{(12k+7)^2}+\frac1{(12k+11)^2}\right).$$ Then $$\psi_1(5/12)-\psi_1(11/12)=\frac{12^2(S_1-S_2)}2.$$

The first series is $$S_1=\left(1+\frac1{3^2}\right)\sum_{k=0}^\infty\left(\frac1{(4k+1)^2}-\frac1{(4k+3)^2}\right),$$ a rational multiple of Catalan's constant.

The second series is $L(\chi,2)$ where $\chi$ is the even Dirichlet character of conductor $12$. This can be evaluated using the functional equation of Dirichlet's L-functions (see for instance Washington's book on cyclotomic fields) or else via the infinite series for $\cot^2$.

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.