4

I have been looking for the generalized version of the below integral,

$$I(n)=\int_0^1 \frac{\log(1-x)\log^n(1+x)}{1+x} \, dx$$

So far I have derived the results only for $n=1$,

$$I(1)=\int_0^1 \frac{\log(1-x)\log(1+x)}{1+x} \, dx = \frac{1}{3}\log^3(2)-\frac{\pi^2}{12}\log(2)+\frac{\zeta(3)}{8}$$

Now trying for $n=2,3,4$ using all variations of identities led to integrals which do not seem to have any nice closed forms

I am aware of a similar generalization performed here

$$ \int_0^1 \frac{\log^n (1-x) \log^{n-1} (1+x)}{1+x} dx=\frac{1}{2} \left(\sum _{m=1}^n \sum _{k=1}^m +\sum _{m=n+1}^{2 n-1} \sum _{k=m-n+1}^n\right) \frac{\binom{n}{k}\binom{n-1}{m-k} \log ^{-m+2 n-1}(2) }{-k+m+1}\left(k \underset{a\to 0}{\text{lim}}\underset{b\to 1}{\text{lim}}\frac{\partial ^{m}B(a,b)}{\partial a^{k-1}\, \partial b^{-k+m+1}}+(-1)^m \log ^{m+1}(2)\right)+\frac{\log ^{2 n}(2)}{n}$$

Is there any alternate method to compute for $n>1$, and build a generalization ?

Amrut Ayan
  • 8,887

3 Answers3

6

\begin{align} I(n)=&\int_0^1 \frac{\ln(1-x)\ln^n(1+x)}{1+x} \, dx\\ \overset{ibp}=&\ \frac1{n+1} \int_0^1 \frac{\ln^{n+1}(1+x)-\ln^{n+1}2}{1-x} \, \overset{x\to 1-x}{dx}\\ =&\ \frac1{n+1} \sum_{k=1}^{n+1}\binom {n+1}k (\ln2)^{n+1-k}\int_0^1\frac{\ln^k(1-\frac x2)}x dx\\ \end{align} Note that \begin{align} J(k)=&\int_0^1\frac{\ln^k(1-\frac x2)}x \overset{t= 1-\frac x2 }{dx} =\int_0^{1}\frac{\ln^k t}{1-t} dt -\int_0^{1/2}\frac{\ln^k t}{1-t}\overset{t\to \frac12t}{ dt}\\ =& \ (-1)^k k!\ \text{Li}_{k+1}(1) - \sum_{j=0}^k \binom kj (\ln2)^{k-j}(-1)^j j! \ \text{Li}_{j+1}(\frac12) \end{align} Thus \begin{align} I(n) = &\ \frac1{n+1} \sum_{k=1}^{n+1}\binom {n+1}k (\ln2)^{n+1-k}J(k)\\ =&\ \sum_{k=1}^{n+1}\frac{n!\ln^{n+1}2}{(n+1-k)!} \bigg[\ln^k\frac12\ \text{Li}_{k+1}(1)- \sum_{j=0}^k \frac{\ln^j\frac12\ \text{Li}_{j+1}(\frac12)}{(k-j)!}\bigg] \end{align}

Quanto
  • 120,125
3

Change of variable:

$$y=1+x$$

$$I(n)=\int_1^2 \frac{\ln(2-y)\ln^ny}{y} \, dy$$

Now consider

$$I(a)=\int_1^2y^a\ln(2-y)\, dy$$

We get back to the original integral by differentiating $I(a)$ with respect to $a$ $n$ times and then taking $a=-1$

Using expansion

$$\ln\left (1-\frac{y}{2} \right )=-\sum_{k=1}^{\infty}\frac{y^k}{k2^k}$$

we get after a simple integration:

$$I(a)=\ln 2\frac{2^{a+1}-1}{a+1}+\sum_{k=1}^{\infty}\frac{1}{k2^k(k+a+1)}-\sum_{k=1}^{\infty}\frac{2^{k+a+1}}{k2^k(k+a+1)}$$

Now, by differentiating $I(a)$ with respect to $a$ $n$ times and then taking $a=-1$ we reach the desired result:

$$I(n)=\frac{(\ln 2)^{n+2}}{n+1}+(-1)^nn!\left [\text{Li}_{n+2} \left ( \frac{1}{2} \right ) -\zeta(n+2)-\sum_{i=1}^{n}(-1)^{i}\frac{(\ln 2)^{i}}{i!}\zeta(n+2-i)\right ]$$

An addition

An alternative form that does not use the polylogarithm and zeta function:

$$I(n)=\frac{(\ln 2)^{n+2}}{n+1}-(-1)^nn!\sum_{k=1}^{\infty} \frac{1- \frac{1}{2^k}+f(k,n)}{k^{n+2}}$$

where $$f(k,n)=\sum_{i=1}^{n}(-1)^{i}\frac{(k\ln 2)^{i}}{i!}$$ The previous result follows directly from this last form

Martin Gales
  • 7,927
  • Are you sure about the last formula?, I tried running it for $n=3$, it did not match – Amrut Ayan Dec 27 '24 at 11:10
  • 1
    @AmrutAyan So when I get time, I'll check it right away. It is possible that the signs are wrong somewhere. I did everything with paper and pencil. – Martin Gales Dec 27 '24 at 17:03
  • Yes sir, I too believe the sign went wrong somewhere, because when I checked for n=3, the value came out to be much more positive, thanks! – Amrut Ayan Dec 27 '24 at 17:05
  • 1
    @AmrutAyan I added an intermediate result from which the final result follows directly. It seems to be correct. Indeed there was one false minus sign. However, let me know if the problems persist. Thanks a lot for pointing out the bugs. – Martin Gales Dec 27 '24 at 18:31
0

First, assume $u = 1 + x$. Then, solve it with integration by parts, as $v=log(2-u)$ and $w=\frac{log^n(u)}{u}$. The result of indefinite integral is $\frac{1}{n+1}\left(log(2-u).log^{n+1}(u)+\int{\frac{log^{n+1}(u)}{2-u}du} \right)$. You can apply the definite version of it to get the result of generalized $I(n)$.