2
$\begingroup$

From quite sometime I've been struggling on proving

$$\int_{0}^{1}\dfrac{\log(x)\log(1 + 2x)}{x(1+x)}\,dx = -\dfrac{7}{6}\zeta(3)$$

I have tried partial fraction decomposition which produces a simple integral and a difficult one. I'm quite wondering if we can evaluate the integral without using the Linear combination of polylogarithms and logarithms. Any approach including complex analysis is most welcomed.

Any help would be appreciated. Thanks for reading.

$\endgroup$
4
  • $\begingroup$ There is an explicit antiderivative. $\endgroup$ Commented May 13, 2021 at 9:47
  • 1
    $\begingroup$ @ClaudeLeibovici could you please elaborate a bit more? $\endgroup$
    – BookWick
    Commented May 13, 2021 at 9:48
  • $\begingroup$ Yes,but it involves polylogarithms and logarithms $\endgroup$ Commented May 13, 2021 at 9:50
  • $\begingroup$ @ClaudeLeibovici I would love to see it at least even if it involves polylogarithms and logarithms :) $\endgroup$
    – BookWick
    Commented May 13, 2021 at 9:51

2 Answers 2

7
$\begingroup$

Note

\begin{align} &\int_{0}^{1}\dfrac{\ln x\ln(1 + 2x)}{x(1+x)}dx =\int_0^1 \int_0^2 \frac{\ln x}{(1+x)(1+t x)}dtdx\\ =&\int_0^2 \frac{dt}{1-t}\int_0^1 \frac{\ln x}{1+x} -\frac{t\ln x}{1+ t x}\>dx =\int_0^2 \frac{Li_2(-1)- Li_2(-t)}{1-t}dt\\ =& \int_0^1 \frac{Li_2(-1)- Li_2(-t)}{1-t}dt + \int_0^1 \frac{Li_2(-1-t)- Li_2(-1)}{t}dt\\ \overset{IBP}=& \int_0^1\frac{\ln(1-t)\ln(1+t)}t dt+ \int_0^1\frac{\ln t\ln(2+t)}{1+t}dt \end{align} where $\int_0^1\frac{\ln t\ln(2+t)}{1+t}dt= -\frac{13}{24}\zeta(3)$ and \begin{align} \int_0^1 \frac{\ln(1-t)\ln(1+t)}{t}\,dx =&\frac14 \int_0^1 \overset{t^2\to t\to 1-t}{\frac{\ln^2(1-t^2)}t}dt -\frac14 \int_0^1 \frac{\ln^2\overset{\to t}{\frac{1-t}{1+t}}}t dt\\ =& \left( \frac1{8} -\frac7{16}\right)\int_0^1 \frac{\ln^2 t}{1-t}dt =-\frac5{8}\zeta(3) \end{align} Thus $$\int_{0}^{1}\frac{\ln x\ln(1 + 2x)}{x(1+x)}dx=-\frac58\zeta(3) -\frac{13}{24}\zeta(3)=-\frac76\zeta(3)$$

$\endgroup$
1
$\begingroup$

Assuming $0 \leq x\leq 1$ a CAS gives for the antiderivative $$\text{Li}_3\left(1+\frac{1}{2 x}\right)-\text{Li}_3\left(2+\frac{1}{x}\right)+\text{Li}_3(-2 x-1)+\text{Li}_3(-2 x)+\text{Li}_3(-x)+\left(\text{Li}_2\left(2+\frac{1}{x}\right)-\text{Li}_2\left(1+ \frac{1}{2 x}\right)\right) \log \left(\frac{1}{x}+2\right)-\text{Li}_2(-2 x-1) \log (2 x+1)-(\text{Li}_2(-2 x)+\text{Li}_2(-x)) \log (x)+\frac{1}{2} \log (2) \left(\log ^2(x)-\pi ^2\right)-\log (x) \log (2 (x+1)) \log (2 x+1)$$ I must confess that using the bounds I have a quite ugly expression $$\text{Li}_3(-3)+\text{Li}_3(-2)+\text{Li}_3\left(\frac{3}{2}\right)-\text{Li}_3(3)- \text{Li}_2(-3) \log (3)+2 \text{Li}_2(3) \log (3)-\frac{\log ^3(3)}{2}+(\log (2)+i \pi ) \log ^2(3)+\frac{1}{6} \left(-\log ^3(2)+3 i \pi \log ^2(2)+2\pi ^2 \log (2)\right)+\frac{1}{2} \left(\log \left(\frac{3}{2}\right)+i \pi \right)^2 \log (3)$$ which numerically is the rhs.

If we use partial fraction decomposition, $$\int \frac{\log (x) \log (2 x+1)}{x (x+1)}\,dx=\int \frac{\log (x) \log (2 x+1)}{x }\,dx-\int \frac{\log (x) \log (2 x+1)}{ (x+1)}\,dx$$ we have for the simplest one $$int \frac{\log (x) \log (2 x+1)}{x }\,dx=\text{Li}_3(-2 x)-\text{Li}_2(-2 x) \log (x)$$ but, as you wrote, the second is not pleasant $$\int \frac{\log (x) \log (2 x+1)}{ (x+1)}\,dx=-\text{Li}_3\left(1+\frac{1}{2 x}\right)+\text{Li}_3\left(2+\frac{1}{x}\right)-\text{Li}_3(-2 x-1)-\text{Li}_3(-x)+\left(\text{Li}_2\left(1+\frac{1}{2 x}\right)-\text{Li}_2\left(2+\frac{1}{x}\right)\right) \log \left(\frac{1}{x}+2\right)+\text{Li}_2(-2 x-1) \log (2 x+1)+\text{Li}_2(-x) \log (x)+\frac{1}{2} \log (2) \left(\pi ^2-\log ^2(x)\right)+\log (x) \log (2 (x+1)) \log (2 x+1)$$ and still the same problem when using the bounds.

$\endgroup$
5
  • $\begingroup$ Claude, this really seems very ugly :( $\endgroup$
    – BookWick
    Commented May 13, 2021 at 12:16
  • $\begingroup$ @BookWick. Even more that ugly ! Shall we agree on awful ? $\endgroup$ Commented May 13, 2021 at 12:17
  • $\begingroup$ Claude, haha, this integral is really painful though. I don't think if real analytics methods would give us any nice solution. What's your opinion on using residue theorem, complex analysis or Mellin transforms? $\endgroup$
    – BookWick
    Commented May 13, 2021 at 12:19
  • $\begingroup$ or maybe even contour integration? what do you think? $\endgroup$
    – BookWick
    Commented May 13, 2021 at 12:20
  • $\begingroup$ @BookWick. Probably but these are among the many areas I am very bad (probably $Z^[--}$. Sorry for that ! Please, ping me when you have an answer. Cheers :-) $\endgroup$ Commented May 13, 2021 at 12:25

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .