8

Prove the following

$$\int^1_0 \frac{\log(x)^3}{\sqrt{x(1-x)}}dx= -\pi (12 \zeta(3) + \log^3(4) + \pi^2 \log(4))$$

The basic approach is differentiating the beta function three times which I am trying to avoid here.A contour method would be nice.

Question 1

Hence how to solve the integral without having to differentiate the beta function ?

My question can be generalized to any problem involving higher order logarithms.

Question 2

Is there a general formula for

$$\frac{\partial^n \partial^m}{\partial x^n \partial y^m} B(x,y)$$

Zaid Alyafeai
  • 14,343
  • @Dr.MV, I already illustrated that i a m trying to avoid differentiation of the beta function. – Zaid Alyafeai Jan 29 '17 at 00:00
  • @Dr.MV, I am not sure if the rsulting sum will be easy to evaluate. Also we still have to differntiate three time which is a lengthy process. – Zaid Alyafeai Jan 29 '17 at 00:11
  • @Dr.MV, already tried it , nowhere to go from there. – Zaid Alyafeai Jan 29 '17 at 00:17
  • what could work: transform into a trigonometric representation, expand $\log(\sin(x))$ in a clever way, integrate, think about sums like $$\sum_{n,m,l\geq1}\frac{1}{mnl(m+n+l)}$$... – tired Jan 29 '17 at 03:16
  • By substituting $x = e^y$, you get $\int\limits_{-\infty}^{0} \frac{y^3}{1-e^y} \sqrt{e^y(1-e^y)} dy$ The first term should contribute to $ \zeta(3)$ by definition of zeta function. I wasn't able to solve the integral tough. – Srini Jan 29 '17 at 03:21
  • @tired, seems promising. – Zaid Alyafeai Jan 29 '17 at 03:35
  • Integral is equal to $\displaystyle -48\int_0^1 \dfrac{(\log x)^2\arcsin x}{x}dx$. – FDP Jan 29 '17 at 10:07
  • Therefore, the integral is equal to $\displaystyle -96\sum_{n=0}^{+\infty} \dfrac{(2n)!}{4^n(n!)^2(2n+1)^4}$ – FDP Jan 29 '17 at 11:52
  • @ZaidAlyafeai Have you made any progress on this? This one has been a challenge. – Mark Viola Feb 04 '17 at 05:15
  • @Dr.MV, no idea so far. – Zaid Alyafeai Feb 04 '17 at 14:14
  • @ZaidAlyafeai the method i proposed will work in the end since it is known how to calculate the kind of sums i mentioned: https://arxiv.org/pdf/1205.0037v1.pdf they even have a special name: mordell-tornheim sums – tired Feb 04 '17 at 14:47
  • @tired, i am not sure how you reached to that sum. – Zaid Alyafeai Feb 04 '17 at 15:09
  • As @tired commented earlier, expand the exponential as $$\log^3(x)=-\sum_{\ell \ge 1}\sum_{m\ge 1}\sum_{n\ge 1}\frac{(-1)^{\ell+m+n}(x-1)^\ell,(x-1)^m,(x-1)^n}{\ell,m,n}$$Then, let $x=\sin^2(\theta)$. Of course, there might be other ways to proceed. – Mark Viola Feb 09 '17 at 06:46
  • @Dr. MV what ihat originally in mind was the Fourier expansion of $\log(\sin(x))$... – tired Feb 09 '17 at 09:37
  • @tired, contour integration ? – Zaid Alyafeai Feb 10 '17 at 19:54
  • @ZaidAlyafeai use the aforementioned fourier expansion and integrate term by term using the orthogonality properties of the trigonometric functions. afterwards you have to perform a few sums which should relate to different zeta values by using for example the paper i linked above – tired Feb 11 '17 at 12:36
  • @Dr.MV here we go – tired Feb 11 '17 at 18:59

2 Answers2

5

Ok, let's see... The integral in question reads

$$ I=\int_0^1dx\frac{\log^3(x)}{\sqrt{x}\sqrt{1-x}} $$

The basic idea is to exploit the Fourier expansion of $\log(\sin(x))$ so first of all we have to transform in an appropriate form. Namely set $x=y^2=\sin^2(\theta)$ to get

$$ I=2^4\int_0^1\frac{\log^3(y)}{\sqrt{1-y^2}}dy=16\int_0^{\pi/2}d\theta\log^3(\sin{\theta}) $$

we now can make use of the aformentioned Fourier expansion, yielding

$$ -\frac{1}{16}I=\color{red}{I_1}+3\color{blue}{I_2}+3\color{green}{I_3}+\color{orange}{I_4}\quad (\star) $$

with $$\color{red}{I_1}=\log^3(2)\int_0^{\pi/2}d\theta=\color{red}{\frac{\pi}{2}\log^3(2)}\\ \color{blue}{I_2}=\log^2(2)\sum_{n\geq1}\frac{1}{n}\int_0^{\pi/2}\cos(2n\theta)d\theta=\log^2(2)\sum_{n\geq1}\frac{\sin(\pi n)}{n^2}=\color{blue}{0}$$

the remaining terms need a bit more work, we have (we use the orthogonatity relations for the trigonometric functions)

$$ \color{green}{I_3}=\log(2)\sum_{n,m\geq1}\frac{1}{nm}\int_{0}^{\pi/2}\cos(2n\theta)\cos(2m\theta)d\theta=\\\frac{\log(2)}{4}\sum_{n,m\geq1}\frac{1}{mn}\left(\frac{\sin(\pi(m-n))}{m-n}+\frac{\sin(\pi(m+n))}{m+n}\right) $$

since $m,n\in \mathbb{N}^+$ the $\sin(\pi (m\mp n))$ terms can only contribute if their argument is zero, where we can use the limit $\sin(ax)/x\sim_0a$. This is only possible if $m=n$ in the first term and impossible in the second one ($m+n\neq0$ !). This means

$$ \color{green}{I_3}=\frac{\pi\log(2)}{4}\sum_{n\geq 1}\frac{1}{n^2}=\color{green }{\frac{\pi^3\log(2)}{24}} $$

for the last remaining sum we obtain

$$ \color{orange}{I_4}=\sum_{n,m,l\geq1}\frac{1}{nml}\int_{0}^{\pi/2}\cos(2n\theta)\cos(2m\theta)\cos(2l\theta)d\theta=\\ \sum_{n,m,l\geq1}\frac{1}{8nml}\left(\frac{3\sin(\pi(l-m-n))}{l-m-n}+\frac{\sin(\pi(l+m+n))}{l+m+n}\right) $$

where we used the symmetry of the first term under cyclic relabelings of $(l,m,n)$ .Now the considerations concerning $\color{green}{I_2}$ also apply here and we find

$$ \color{orange}{I_4}=\frac{3\pi}{8}\sum_{n,m \geq1}\frac{1}{nm(m+l)}=\frac{3\pi}{8}S $$

now following the steps in @r9m nice answer we can finally state that $$ \color{orange}{I_4}=\frac{3\pi}{8}2\zeta(3)=\color{orange}{\frac{3\pi}{4}\zeta(3)} $$

Putting everything we have into $(\star)$ we obtain

$$ I=-16\left(\color{red}{\frac{\pi}{2}\log^3(2)}+3\cdot\color{blue}{0}+3\color{green}{\frac{\pi^3\log(2)}{24}}+\color{orange}{\frac{3\pi}{4}\zeta(3)}\right)=\\ -\pi(8\log^3(2)+2\pi^2\log(2)+12\zeta(3)) $$

as announced


For an alternative evaluation of the sum contained in $\color{orange}{I_4}$ we can use $\int_{0}^{\infty}e^{-(m+n)x}=\frac{1}{m+n}$ which yields

$$ S=\int_0^{\infty}\log^2(1-e^{-x})dx\underbrace{=}_{e^{-x}=z}\int_0^{1}\frac{\log^2(1-z)}{z}dz \quad(\star\star) $$

Now, using integration by parts (alternativly we could recognize the connection to the generating function of the harmonic number which gives us another nice oppurtunity to evaluate this series) we find that

$$ S=-2\int_0^1dz\frac{\log(1-z)\log(z)}{1-z} $$

another integration by parts yields, using the integral representation of the polylogarithm

$$ S=2\int_0^1dz\frac{\text{Li}_2(1-z)}{1-z}=-2\text{Li}_3(1-z)|_0^1=2\text{Li}_3(1)=2\zeta(3) $$


Recognized too late, since wasted: Embarking from $(\star\star)$, setting $1-z=q$ and expanding the resulting geometric series, gives a near trivial evalution of $S$ which might be generalized easily. This enables us, in the end, to evaluate the whole family of sums

$$ S_r=\sum_{m_1,..m_r\geq1}\frac{1}{m_1...m_r}\frac{1}{m_1+...+m_r} $$

and as a consequence the integrals

$$ I_{r+1}=\int_0^1dx\frac{\log^{r+1}(x)}{\sqrt{x}\sqrt{1-x}} $$

by the technique shown above

tired
  • 12,325
  • 1
    Wow! You must really be tired. I'd like to give you a (+1000), but all I've got is a measly (+1). – Mark Viola Feb 11 '17 at 19:38
  • @Dr.MV thanks a lot for the kind words! i'm glad you like my answer – tired Feb 11 '17 at 19:42
  • Well, I've spent quite a bit of effort on this one, trying to get to a form where $\text{Li}_3(1)$ appears. I tried multiple subs (one was yours), Feynman's trick, IBPs, integral representations of thr logarithm (e.g., Frullani's integral), etc. And I got nuttin' to show for it. Anyway, I'm glad you were able to crack this with rpm's answer to evaluate the multiple sum. – Mark Viola Feb 11 '17 at 19:53
  • How did you adapt rpm's approach to odd integers? It seems that that approach worked only for even powers. – Mark Viola Feb 11 '17 at 20:07
  • @Dr. MV to be honest i didn't go through all the details but R9m states in the comments that also odd intgers work (and also gave a explicit formula). i will add some more content tomorrow! – tired Feb 11 '17 at 20:26
  • That sounds good. Did you notice that r9m posted two solutions? Which did you leverage? – Mark Viola Feb 11 '17 at 20:39
  • @Dr.MV i was using the first one – tired Feb 11 '17 at 20:43
  • @Dr.MV since you where interested in a connection with polylogs i added an alternative evaluation of the sum in question making use of them – tired Feb 12 '17 at 13:26
  • 1
    @ZaidAlyafeai since you where interested in a contour approach: you can also integrate $\log^m(1-e^{-2iz})$ around a rectangle $(0,\pi/2,\pi/2-i\infty,-i\infty)$ and relate the result to the integral in question – tired Feb 12 '17 at 14:27
  • @tired, how do you come up with the function to integrate ? by experience ? – Zaid Alyafeai Feb 12 '17 at 15:40
  • i was rethinking my evaluation of $S$ and noticed that $\log^2(1-e^{-z})=\log^2(1-e^{(-i)(-iz)})$ from that point everything became clear (i remember that i have used this approach before: http://math.stackexchange.com/questions/1164183/how-to-evaluate-i-int-limits-0-pi-2-fracx-log-sinx-sinx-dx/1172031#1172031 ) – tired Feb 12 '17 at 16:10
2

$\newcommand{\bbx}[1]{\,\bbox[8px,border:1px groove navy]{\displaystyle{#1}}\,} \newcommand{\braces}[1]{\left\lbrace\,{#1}\,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\,{#1}\,\right\rbrack} \newcommand{\dd}{\mathrm{d}} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,\mathrm{e}^{#1}\,} \newcommand{\ic}{\mathrm{i}} \newcommand{\mc}[1]{\mathcal{#1}} \newcommand{\mrm}[1]{\mathrm{#1}} \newcommand{\pars}[1]{\left(\,{#1}\,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\root}[2][]{\,\sqrt[#1]{\,{#2}\,}\,} \newcommand{\totald}[3][]{\frac{\mathrm{d}^{#1} #2}{\mathrm{d} #3^{#1}}} \newcommand{\verts}[1]{\left\vert\,{#1}\,\right\vert}$ \begin{align} &\int_{0}^{1}{\ln^{3}\pars{x} \over \root{x\pars{1 - x}}}\,\dd x \,\,\,\stackrel{x\ =\ \sin^{2}\pars{\theta}}{=}\,\,\, 16\int_{0}^{\pi/2}\ln^{3}\pars{\sin\pars{\theta}}\,\dd\theta \\[5mm] = &\ \left.16\,\Re\int_{0}^{\pi/2}\ln^{3}\pars{{1 - z^{2} \over 2z}\ic} \,{\dd z \over \ic z}\,\right\vert_{\ z\ =\ \exp\pars{\ic\theta}} = \left.16\,\Im\int_{0}^{\pi/2}\ln^{3}\pars{{1 - z^{2} \over 2z}\ic} \,{\dd z \over z}\,\right\vert_{\ z\ =\ \exp\pars{\ic\theta}} \\[5mm] \stackrel{\mrm{as}\ \epsilon\ \to\ 0^{+}}{\sim} & -16\,\Im\int_{\pi/2}^{0} \bracks{-\ln\pars{2\epsilon} + \pars{{\pi \over 2} - \theta}\ic}^{\,3}\ \ic\,\dd\theta - 16\,\Im\int_{\epsilon}^{1} \bracks{\ln\pars{1 - x^{2} \over 2x} + {\pi \over 2}\,\ic}^{3}\,{\dd x \over x} \\[5mm] = &\ 16\int_{0}^{\pi/2}\bracks{-\ln^{3}\pars{2\epsilon} + 3\ln\pars{2\epsilon}\,\theta^{2}}\,\dd\theta - 16\int_{\epsilon}^{1}\bracks{3\ln^{2}\pars{1 - x^{2} \over 2x}\,{\pi \over 2} -{\pi^{3} \over 8}}\,{\dd x \over x} \\[1cm] = &\ -8\pi\ln^{3}\pars{2\epsilon} + \color{#f00}{2\pi^{3}\ln\pars{2\epsilon}} - 24\pi\int_{\epsilon}^{1}{\ln^{2}\pars{1 - x^{2}} - 2\ln\pars{1 - x^{2} }\ln\pars{2x} + \ln^{2}\pars{2x} \over x}\,\dd x \\[5mm] &\ \color{#f00}{\mbox{} - 2\pi^{3}\ln\pars{\epsilon}} \\ \stackrel{\mrm{as}\ \epsilon\ \to\ 0^{+}}{\sim} &\ \color{#f00}{-8\pi\ln^{3}\pars{2\epsilon}} + 2\pi^{3}\ln\pars{2} - 12\pi\ \overbrace{\int_{0}^{1}{\ln^{2}\pars{1 - x} \over x}\,\dd x} ^{\ds{2\zeta\pars{3}}}\ +\ 24\pi\ln\pars{2}\ \overbrace{\int_{0}^{1}{\ln\pars{1 - x} \over x}\,\dd x} ^{\ds{-\,\mrm{Li}_{2}\pars{1} = -\,{\pi^{2} \over 6}}} \\[5mm] &\ \mbox{} + 12\pi\ \underbrace{\int_{0}^{1}{\ln\pars{1 - x}\ln\pars{x} \over x}\,\dd x} _{\ds{\zeta\pars{3}}}\ -\ 24\pi\ \underbrace{\int_{2\epsilon}^{2}{\ln^{2}\pars{x} \over x}\,\dd x} _{\ds{{\ln^{3}\pars{2} \over 3} - \color{#f00}{\ln^{3}\pars{2\epsilon} \over 3}}} \label{1}\tag{1} \end{align}

Note that

$$ \left\{\begin{array}{rcl} \ds{\int_{0}^{1}{\ln\pars{1 - x}\ln\pars{x} \over x}\,\dd x} & \ds{=} & \ds{-\int_{0}^{1}\mrm{Li}_{2}'\pars{x}\ln\pars{x}\,\dd x = \int_{0}^{1}{\mrm{Li}_{2}\pars{x} \over x}\,\dd x = \int_{0}^{1}\mrm{Li}_{3}'\pars{x}\,\dd x} \\[1mm] & \ds{=} & \ds{\mrm{Li}_{3}\pars{1} = \bbx{\ds{\zeta\pars{3}}}} \\[5mm] \ds{\int_{0}^{1}{\ln^{2}\pars{1 - x} \over x}\,\dd x} & \ds{=} & \ds{\int_{0}^{1}{\ln^{2}\pars{x} \over 1 - x}\,\dd x = 2\int_{0}^{1}{\ln\pars{1 - x}\ln\pars{x} \over x}\,\dd x = \bbx{\ds{2\,\zeta\pars{3}}}} \end{array}\right. $$

Expression \eqref{1} becomes

\begin{align} &\int_{0}^{1}{\ln^{3}\pars{x} \over \root{x\pars{1 - x}}}\,\dd x \,\,\,\stackrel{\mrm{as}\ \epsilon\ \to\ 0^{+}}{\to}\,\,\, 2\pi^{3}\ln\pars{2} - 24\pi\,\zeta\pars{3} - 4\pi^{3}\ln\pars{2} + 12\pi\,\zeta\pars{3} - 8\pi\ln^{3}\pars{2} \\[5mm] = &\,\,\, \bbox[15px,#ffe,border:1px dotted navy]{\ds{-2\pi\bracks{\vphantom{\Large A}6\,\zeta\pars{3} + 4\ln^{3}\pars{2} + \pi^{2}\ln\pars{2}}}} \approx -96.6701 \end{align}

Felix Marin
  • 89,464