7

First of all, hello everyone, i am a new in MSE community. I hope you guys all well. I found this integral on a web:

$$\int_{0}^{1}\frac{x\log{x}\log^2{(1-x)}}{1+x^2}dx$$ I tried to IBP with $u=\log{x}\log^2{(1-x)}$ and $dv=\frac{x}{1+x^2}$ it to get: $$\int_{0}^{1}\frac{\log{x}\log{(1-x)}\log{(1+x^2)}}{1-x}dx-\frac{1}{2}\int_{0}^{1}\frac{\log{(x^2+1)}\log^2{(1-x)}}{x}dx$$

From here I got stuck, i know may be I misdirected. I tried to search our forum and found this link seems similar.

The technique is unique, but I don't know how to apply it for my case. Can you guys give some advices or ideas? Thank you so much.

user97357329
  • 5,319
  • 3
    You might want to try the easier one first, if you haven't: $$\int_0^1 \frac{x\log x \log(1-x)}{1+x^2} dx = \frac{41 \zeta (3)}{64}-\frac{3}{32} \pi ^2 \log (2)$$ the case when $\log(1-x)$ has exponent $2,3$ or $4$ have similar results, except now one need to use $\text{Li}_4, \text{Li}_5, \text{Li}_6$. – pisco Jul 21 '22 at 08:58
  • @pisco Thank, so with your suggestion, what will i need to solve? Can I use series and derivative of Beta function? –  Jul 21 '22 at 09:08
  • @pisco Would you mind giving me some hints? –  Jul 22 '22 at 06:09
  • I don't have an elementary solution in mind. You might consider editing the question so it gets more attention. – pisco Jul 22 '22 at 07:25
  • @pisco Still thank you for what you suggested. –  Jul 22 '22 at 10:50
  • 2
    For answer checking: $$\small \int_{0}^{1}\frac{x\log{x}\log^2{(1-x)}}{1+x^2}dx = -\frac{15 \text{Li}4\left(\frac{1}{2}\right)}{8}+\frac{167 \pi ^4}{23040}-\frac{5}{64}\log ^4(2)+\frac{1}{32} \pi ^2 \log ^2(2)$$ and $$\small \int{0}^{1}\frac{x\log{x}\log^3{(1-x)}}{1+x^2}dx = \frac{45 \text{Li}_5\left(\frac{1}{2}\right)}{16}-\frac{35 \pi ^2 \zeta (3)}{64}+\frac{13287 \zeta (5)}{2048}-\frac{3}{128}\log ^5(2)+\frac{1}{64} \pi ^2 \log ^3(2)-\frac{23 \pi ^4 \log (2)}{1024}$$ with similar expression for $\log^4(1-x)$. The difficulty, of course, increases exponentially. – pisco Jul 22 '22 at 12:35

1 Answers1

4

The posed question is very similar in some sense to the one treated in my previous post (and also the answers might look similar to some extent, in some sense).


A solution in (large steps) by Cornel Ioan Valean

In the following, the general line of a straightforward solution (it can be counted so if we have some key and extremely hard to derive series at hand, and we have them!) will be given. More details appearing on the way will be found in the sequel of (Almost) Impossible Integrals, Sums, and Series.

We may start directly from the main integral as posed and turn it into a sum of series by using that $\displaystyle \sum_{n=1}^{\infty} (-1)^{n-1} x^{2n-1}=\frac{x}{1+x^2}, \ \int_0^1 x^{n-1} \log(x)\log^2(1-x)\textrm{d}x=2\zeta(3)\frac{1}{n}+2\zeta(2)\frac{H_n}{n}-\frac{H_n^{(2)}}{n^2}-2\frac{H_n^{(3)}}{n}-2\frac{H_n H_n^{(2)}}{n}-\frac{H_n^2}{n^2},$ where the latter is immediately obtained by differentiating once with respect to $n$ the well-known logarithmic integral, $\displaystyle \int_0^1 x^{n-1}\log^2(1-x)\textrm{d}x=\frac{(\psi(n+1)+\gamma)^2+\zeta(2)-\psi^{(1)}(n+1)}{n}$ (observe the Beta function structure). When $n$ is a positive integer, the previous logarithmic integral result immediately turns into $\displaystyle \int_0^1 x^{n-1}\log^2(1-x)\textrm{d}x=\frac{H_n^2+H_n^{(2)}}{n}.$ An elementary derivation of the latter special case may be found in (Almost) Impossible Integrals, Sums, and Series, pages $60-62$.

So, returning to the main integral, and using the results above, we have $$\int_0^1\frac{x\log(x)\log^2(1-x)}{1+x^2}\textrm{d}x=\sum_{n=1}^{\infty}(-1)^{n-1}\int_0^1 x^{2n-1}\log{(x)}\log^2(1-x)\textrm{d}x$$ $$=\zeta(3)\underbrace{\sum_{n=1}^{\infty} (-1)^{n-1}\frac{1}{n}}_{\displaystyle \log(2)}+\zeta(2)\underbrace{\sum_{n=1}^{\infty} (-1)^{n-1}\frac{H_{2n}}{n}}_{\displaystyle 5/48 \pi^2-\log^2(2)/4}-\frac{1}{4}\sum_{n=1}^{\infty} (-1)^{n-1}\frac{H_{2n}^2}{n^2}-\frac{1}{4}\sum_{n=1}^{\infty} (-1)^{n-1}\frac{H_{2n}^{(2)}}{n^2}$$ $$-\sum_{n=1}^{\infty} (-1)^{n-1}\frac{H_{2n} H_{2n}^{(2)}}{n}-\sum_{n=1}^{\infty} (-1)^{n-1}\frac{H_{2n}^{(3)}}{n}$$ $$=\frac{167}{256}\zeta(4)+\frac{3}{16}\log^2(2)\zeta(2)-\frac{5}{64}\log^4(2)-\frac{15}{8}\operatorname{Li}_4\left(\frac{1}{2}\right).\tag1$$

Now, a first important point to mention is that during the extraction of some advanced harmonic series appearing above we need the well-known polylogarithmic relations on this page, and the special tetralogarithmic values, $$\Re\biggr \{\operatorname{Li}_4\left(\frac{1\pm i}{2}\right)\biggr\}=\frac{343}{1024}\zeta(4)-\frac{5}{128}\log^2(2)\zeta(2)+\frac{1}{96}\log^4(2)+\frac{5}{16}\operatorname{Li}_4\left(\frac{1}{2}\right);$$ $$\Re\{\operatorname{Li}_4(1\pm i)\}=\frac{485}{512}\zeta(4)+\frac{1}{8}\log^2(2)\zeta(2)-\frac{5}{384}\log^4(2)-\frac{5}{16}\operatorname{Li}_4\left(\frac{1}{2}\right).$$

In order to extract the tetralogarithmic values we need the alternating harmonic series $\displaystyle \sum_{n=1}^{\infty}(-1)^{n-1}\frac{H_{2n}}{n^3}$ you may find calculated in a very simple and elegant way in this answer. Other steps are trivial as explained in this 2014 answer, where we use the classical generating function $\displaystyle \sum_{n=1}^{\infty}x^n \frac{H_n}{n^3}$ and the tetralogarithmic inverse relation in order to make the connections between that series and the special tetralogarithmic values. Full details about the extraction process of these results (and other related ones) will be found in the sequel of (Almost) Impossible Integrals, Sums, and Series.

The fourth series in $(1)$ is calculated in this awesome answer. The third and the fifth series may be extracted immediately by exploiting their corresponding generating functions $$\sum_{n=1}^{\infty} x^n \frac{H_n^2}{n^2}$$ $$=2\zeta(4)-\frac{1}{3}\log(x)\log^3(1-x)-\log^2(1-x)\operatorname{Li}_2(1-x)+\frac{1}{2}(\operatorname{Li}_2(x))^2$$ $$+2\log(1-x)\operatorname{Li}_3(1-x)+\operatorname{Li}_4(x)-2\operatorname{Li}_4(1-x), \ |x|\le1 \land \ x\neq1,$$ and $$\sum_{n=1}^{\infty} x^n \frac{H_n H_n^{(2)}}{n}$$ $$=-\zeta(4)+\frac{1}{6}\log(x)\log^3(1-x)-\frac{1}{24}\log^4(1-x)+\frac{1}{2}\log^2(1-x)\operatorname{Li}_2(1-x)$$ $$-\log(1-x)\operatorname{Li}_3(1-x)+\operatorname{Li}_4(1-x)-\operatorname{Li}_4\left(\frac{x}{x-1}\right), \ |x|\le1 \land \ x\neq1.$$ leading to $$\sum_{n=1}^{\infty} (-1)^{n-1} \frac{H_{2n}^2}{n^2}$$ $$=2 G^2-\log(2)\pi G+\frac{231}{32}\zeta(4)-\frac{35}{16}\log(2)\zeta(3)+\log^2(2)\zeta(2)-\frac{5}{48}\log^4(2)$$ $$-2 \pi \Im\biggr\{\operatorname{Li}_3\left(\frac{1+i}{2}\right)\biggr\}-\frac{5}{2}\operatorname{Li}_4\left(\frac{1}{2}\right)$$ and $$\sum_{n=1}^{\infty} (-1)^{n-1} \frac{H_{2n}H_{2n}^{(2)}}{n}$$ $$ =\frac{1}{4}\log(2)\pi G-\frac{137}{128}\zeta(4)+\frac{35}{64}\log(2)\zeta(3)-\frac{3}{8}\log^2(2)\zeta(2)+\frac{5}{96}\log^4(2)$$ $$+\frac{\pi}{2}\Im\biggr\{\operatorname{Li}_3\left(\frac{1+i}{2}\right)\biggr\}+\frac{5}{4}\operatorname{Li}_4\left(\frac{1}{2}\right).$$

Both series above require tetralogarithmic values with a complex argument as previously mentioned.

And the final series in $(1)$ is a particular case of the more general result, $$\sum_{n=1}^{\infty} (-1)^{n-1}\frac{H_{2n}^{(m)}}{n}=\frac{(-1)^m}{(m-1)!}\int_0^1\frac{\displaystyle \log^{m-1}(x)\log\left(\frac{1+x^2}{2}\right)}{1-x}\textrm{d}x$$ $$=m\zeta (m+1)- 2^{-m} \left(1-2^{-m+1}\right) \log(2 ) \zeta (m) -\sum_{k=0}^{m-1}\beta(k+1)\beta(m-k)$$ $$-\sum_{k=1}^{m-2}2^{- m-1}(1-2^{-k})(1-2^{-m+k+1}) \zeta (k+1)\zeta (m-k),$$ where $H_n^{(m)}=1+\frac{1}{2^m}+\cdots+\frac{1}{n^m}$ represents the $n$th generalized harmonic number of order $m$,
$\zeta$ denotes the Riemann zeta function and $\beta$ designates the Dirichlet beta function.

A very simple solution may be found in this answer that exploits the symmetry in double integrals.

End of story

user97357329
  • 5,319
  • 1
    While the $\text{Li}_4$ is already nontrivial, I would still like to get your attention to the $\text{Li}_5$ case, which is related to OP's integral for $\log^3(1-x)$: $$\Re \text{Li}_5((1+i)/2) = \frac{5 \text{Li}_5\left(\frac{1}{2}\right)}{32}+\frac{2139 \zeta (5)}{4096}-\frac{1}{960} \log ^5(2)+\frac{5 \pi ^2 \log ^3(2)}{4608}-\frac{343 \pi ^4 \log (2)}{184320}$$ the only proof I am aware of uses multiple zeta values theory. There is also a deep conjecture regarding the nature of values $\Re \text{Li}_n((1+i)/2)$. I hope your elementary-based mindset could offer some new insights. – pisco Jul 22 '22 at 19:39
  • @pisco Thank you for sharing. It will be done elementarily over some time. – user97357329 Jul 22 '22 at 19:44
  • @user97357329 Thank you for your help. Just another question if it is ok, would you mind showing me how to find the generating function of two alternating harmonic series and how to derive with $2n$ you mentioned? –  Jul 23 '22 at 08:21
  • 1
    @PelikanSH037 They are immediately built upon the generating functions $\sum_{n=1}^{\infty} x^n H_n^2=\frac{1}{1-x}(\log ^2(1-x)+\text{Li}2(x))$ and $\sum{n=1}^{\infty} x^n H_n H_n^{(2)}=\frac{1}{1-x}\biggr(\frac{1}{2}\log(x) \log^2(1-x)+\operatorname{Li}_3(x)+\operatorname{Li}_3(1-x)-\zeta(2)\log(1-x)-\zeta(3)\biggr)$ which both appear in (Almost) Impossible Integrals, Sums, and Series. The full (clever) derivations with detailed steps are already included in the sequel of (Almost) Impossible Integrals, Sums, and Series. I'll make an update with the reference after its publishing. – user97357329 Jul 23 '22 at 08:36
  • 1
    @PelikanSH037 They are also presented and derived in Ali Shadhar's book, An Introduction To The Harmonic Series And Logarithmic Integrals: For High School Students Up To Researchers. – user97357329 Jul 23 '22 at 08:37
  • 1
    @user97357329 you are so nice, i wish i could give you +100, thank you so much. –  Jul 23 '22 at 08:41
  • @PelikanSH037 Welcome! All the best! =) – user97357329 Jul 23 '22 at 08:43
  • 1
    The importance of using harmonic series appears in such solution. Very nice (+1). – Ali Shadhar Jul 24 '22 at 17:17
  • @pisco do you have a generalization for such integral with $\ln^n(1-x)$? – Ali Shadhar Jul 24 '22 at 17:20
  • @AliShadhar No. Then general pattern for $\log^n(1-x)$ in terms of $\text{Li}_n$ is intractable, that's why I said the difficulty explodes for larger $n$, like $n=3$ or $4$. Alternatively, the following all have, non-trivially, similar results: $$\int_0^1 \frac{ x \log^a x \log^b (1-x) \log^c(1+x)}{1+x^2} dx \qquad a+b+c\leq 5$$ – pisco Jul 24 '22 at 17:21
  • @pisco makes sense. I thought you do since you use very powerful techniques. – Ali Shadhar Jul 24 '22 at 17:23
  • @AliShadhar Actually, there is no generalization to $\log^n(1-x)$: theoretical evidence says the $I(a,b,c) = \int_0^1 \frac{ x \log^a x \log^b (1-x) \log^c(1+x)}{1+x^2} dx$, when weight $\geq 7$, cannot be expressed in terms of $\text{Li}_n$ alone. Nonetheless, I do have a conjecture on the nature of $I(a,b,c)$, this conjecture seems inaccessible even with my techniques. Therefore, it would be nice if you and Cornel could offer some insights, since your viewpoint is very different from mine. – pisco Jul 24 '22 at 17:47
  • @pisco if it can't be expressed in terms of polylog using your techniques, no way I can do it . – Ali Shadhar Jul 24 '22 at 18:07
  • @AliShadhar It can be expressed when weight $\leq 6$, so there is a chance you guys come up with a nice solution. Conrad's this answer treats the case weight $=4$. – pisco Jul 24 '22 at 18:23