14
$\begingroup$

Hi I am trying to show this astonishing result$$ \int_0^1 \log\frac{\big(x+\sqrt{1-x^2}\big)^2}{\big(x-\sqrt{1-x^2}\big)^2} \frac{x\, dx}{1-x^2}=\frac{\pi^2}{2}. $$ Note we can to use $\ln(a/b)=\ln a-\ln b$ but that didn't help me much . After this I obtained integrals of the form $$ \int_0^1 \log \big[\big(x\pm\sqrt{1-x^2}\big)^2\big] \frac{x\, dx}{1-x^2} $$ which I am not sure how to handle. Thanks.

$\endgroup$
5
  • $\begingroup$ An approach I tried: let $x = \sin(t)$. Then the integral can be simplified to: $$\int_0^{\pi/2}\log^2(\tan(t+\pi/4))\tan(t)\;dt$$I don't know if that's helpful, but it's something different... $\endgroup$
    – apnorton
    May 6 '14 at 3:12
  • $\begingroup$ I am confused at your question. Is the expression $\log\big(\frac{x+\sqrt{1-x^2}}{x-\sqrt{1-x^2}} \big)^2$ or $[\log\big(\frac{x+\sqrt{1-x^2}}{x-\sqrt{1-x^2}} \big)]^2$? $\endgroup$
    – Rudstar
    May 6 '14 at 3:32
  • $\begingroup$ @Rudstar Sorry I made that explicitly clear with my notation now., just edited. I realize that could be confusing thank you for pointing that out to me. $\endgroup$ May 6 '14 at 3:38
  • $\begingroup$ OH! That makes a huge difference! :) Thanks. $\endgroup$
    – apnorton
    May 6 '14 at 3:39
  • $\begingroup$ @anorton Sorry about that. And thank you! $\endgroup$ May 6 '14 at 3:39
16
$\begingroup$

Let

$$$$

\begin{align} I(a)&=\int_0^1 2\, \log\frac{\big(x+a\, \sqrt{1-x^2}\big)}{\big(x-a\, \sqrt{1-x^2}\big)} \frac{x\, dx}{1-x^2} \tag 1\\ \therefore \frac{\partial}{\partial a}I(a) &= 2\, \int_0^1 \left(\frac{\sqrt{1-x^2}}{\big(x+a\, \sqrt{1-x^2}\big)}+\frac{\sqrt{1-x^2}}{\big(x-a\, \sqrt{1-x^2}\big)}\right) \frac{x}{1-x^2}\, dx\\ &= \int_0^{\pi/2} \frac{4\, \tan{(t)}^2}{\tan{(t)}^2-a^2}\, dt \hspace{100pt}\text{(subst. $x=\sin{t}$)}\\ &= \int_0^{\pi/2} \frac{4\, \tan{(t)}^2}{(\tan{(t)}^2-a^2)(1+\tan{(t)}^2)} \sec{(t)}^2 \, dt\\ &= \int_0^{\infty} \frac{4\, y^2}{(y^2-a^2)(1+y^2)} \, dy \hspace{82pt}\text{(subst. $\tan{t}=y$)}\\ &= \int_0^{\infty} \frac{1}{1+a^2}\left(\frac{2a}{y-a}-\frac{2a}{y+a}+\frac{4}{1+y^2}\right)\, dy\\ &= \frac{1}{1+a^2}\left(a\, \log\left(\frac{y-a}{y+a}\right)^2 +4\arctan{y}\right)\Bigg|_0^\infty\\ &= \frac{2\, \pi}{a^2+1}\\ \implies I(a)&=2\,\pi\arctan{a}+C \tag 2 \end{align}

Putting $a=0$ in $(1)$ and $(2)$, we find that $C=0$

Hence, $$I(a)=2\,\pi\arctan{a}$$

and the required integral $$I(1)=\frac{\pi^2}{2}$$

$\endgroup$
12
  • $\begingroup$ Do you mean $\partial_a$? $\endgroup$ May 6 '14 at 4:20
  • $\begingroup$ How do you do the integral in the line between (1) and (2), proof? Thank you for the clever solution, but this is my only question. $\endgroup$ May 6 '14 at 4:20
  • $\begingroup$ Yes, thanks, corrected it.. $\endgroup$
    – gar
    May 6 '14 at 4:20
  • $\begingroup$ I think a simple trigonometric substitution will do, like x=sin(t). Let me see. $\endgroup$
    – gar
    May 6 '14 at 4:22
  • 1
    $\begingroup$ You're welcome. I enjoy your integrals, keep them coming! $\endgroup$
    – gar
    May 6 '14 at 12:19
11
$\begingroup$

$\newcommand{\+}{^{\dagger}} \newcommand{\angles}[1]{\left\langle\, #1 \,\right\rangle} \newcommand{\braces}[1]{\left\lbrace\, #1 \,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\, #1 \,\right\rbrack} \newcommand{\ceil}[1]{\,\left\lceil\, #1 \,\right\rceil\,} \newcommand{\dd}{{\rm d}} \newcommand{\down}{\downarrow} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,{\rm e}^{#1}\,} \newcommand{\fermi}{\,{\rm f}} \newcommand{\floor}[1]{\,\left\lfloor #1 \right\rfloor\,} \newcommand{\half}{{1 \over 2}} \newcommand{\ic}{{\rm i}} \newcommand{\iff}{\Longleftrightarrow} \newcommand{\imp}{\Longrightarrow} \newcommand{\isdiv}{\,\left.\right\vert\,} \newcommand{\ket}[1]{\left\vert #1\right\rangle} \newcommand{\ol}[1]{\overline{#1}} \newcommand{\pars}[1]{\left(\, #1 \,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\pp}{{\cal P}} \newcommand{\root}[2][]{\,\sqrt[#1]{\vphantom{\large A}\,#2\,}\,} \newcommand{\sech}{\,{\rm sech}} \newcommand{\sgn}{\,{\rm sgn}} \newcommand{\totald}[3][]{\frac{{\rm d}^{#1} #2}{{\rm d} #3^{#1}}} \newcommand{\ul}[1]{\underline{#1}} \newcommand{\verts}[1]{\left\vert\, #1 \,\right\vert} \newcommand{\wt}[1]{\widetilde{#1}}$

$\ds{\int_{0}^{1} \ln\pars{\bracks{x + \root{1 - x^{2}}}^2 \over \bracks{x-\root{1 - x^2}}^{2}}\, {x\,\dd x \over 1 - x^{2}} = {\pi^{2} \over 2}:\ {\large ?}}$


With $\ds{x \equiv \sin\pars{\theta}}$: \begin{align}&\int_{0}^{1} \ln\pars{\bracks{x + \root{1 - x^{2}}}^2 \over \bracks{x-\root{1 - x^2}}^{2}}\, {x\,\dd x \over 1 - x^{2}} = -\int_{0}^{\pi/2} \ln\pars{\bracks{\tan\pars{\theta} - 1 \over \tan\pars{\theta} + 1 }^{2}}\, \tan\pars{\theta}\,\dd\theta \\[5mm]= &\ -\int_{0}^{\pi/2}\ln\pars{\tan^{2}\pars{\theta - {\pi \over 4}}}\, \tan\pars{\theta}\,\dd\theta =-\int_{-\pi/4}^{\pi/4}\ln\pars{\tan^{2}\pars{\theta}}\, \tan\pars{\theta + {\pi \over 4}}\,\dd\theta \\[5mm]= &\ -\int_{0}^{\pi/4}\ln\pars{\tan^{2}\pars{\theta}}\, \bracks{{1 + \tan\pars{\theta} \over 1 - \tan\pars{\theta}} + {1 - \tan\pars{\theta} \over 1 + \tan\pars{\theta}}}\,\dd\theta \\[3mm]&=-2\int_{0}^{\pi/4}\ln\pars{\tan\pars{\theta}}\, {2\sec^{2}\pars{\theta} \over 1 - \tan^{2}\pars{\theta}}\,\dd\theta =-4\ \overbrace{\int_{0}^{1}{\ln\pars{t} \over 1 - t^{2}}\,\dd t} ^{\ds{-\,{\pi^{2} \over 8}}} = \bbox[15px,#ffe,border:1px dotted navy]{\ds{\pi^{2} \over 2}} \end{align}

The last integral can be evaluated by expanding $\ds{\pars{1 - t^{2}}^{-1}}$ in powers of $\ds{t}$. Namely,

\begin{align}&\color{#c00000}{\int_{0}^{1}{\ln\pars{t} \over 1 - t^{2}}\,\dd t} =\sum_{n = 0}^{\infty}\int_{0}^{1}\ln\pars{t}t^{2n}\,\dd t =\left.\sum_{n = 0}^{\infty}\partiald{}{\mu}\int_{0}^{1}t^{2n + \mu}\,\dd t\, \right\vert_{\,\mu = 0} =-\sum_{n = 0}^{\infty}{1 \over \pars{2n + 1}^{2}} \\[3mm]&=-\sum_{n = 1}^{\infty}{1 \over n^{2}} + \sum_{n = 1}^{\infty}{1 \over \pars{2n}^{2}} =-\,{3 \over 4}\sum_{n = 1}^{\infty}{1 \over n^{2}} = -\,{3 \over 4}\,{\pi^{2} \over 6}=\color{#c00000}{-\,{\pi^{2} \over 8}} \end{align}

$\endgroup$
7
$\begingroup$

I hope it is not too late. Let $$ I(a)=\int_0^1 \log\frac{\big(x+a\sqrt{1-x^2}\big)^2}{\big(x-\sqrt{1-x^2}\big)^2} \frac{x\, dx}{1-x^2}, -1\le a\le 1. $$ Clearly $I(-1)=0$ and \begin{eqnarray} I'(a)&=&2\int_0^1 \frac{x}{(x+a\sqrt{1-x^2})\sqrt{1-x^2}}dx\\ &=&2\int_0^{\frac{\pi}{2}}\frac{\sin t}{\sin t+a\cos t}dt. \end{eqnarray} Define $$ A=\int_0^{\frac{\pi}{2}}\frac{\sin t}{\sin t+a\cos t}dt, B=\int_0^{\frac{\pi}{2}}\frac{\cos t}{\sin t+a\cos t}dt $$ and then $$ A+aB=\frac{\pi}{2}, B-aA=-\log |a|.$$ Thus $A=\frac{\frac{\pi}{2}+a\log |a|}{1+a^2}$. However $\frac{a\log |a|}{1+a^2}$ is an odd function of $a$ and hence $$ I(1)=2\int_{-1}^1\frac{\frac{\pi}{2}+a\log |a|}{1+a^2}da=2\int_{-1}^1\frac{\frac{\pi}{2}}{1+a^2}da=\frac{\pi^2}{2}.$$

$\endgroup$
2
  • $\begingroup$ I would like to point out that the integrand of your $I'(a)$ has a pole at $x = -a/\sqrt{1+a^2}$ if $a$ is negative. So this solution also shares the same issue with @gar's answer. Circumventin this problem, though, seems not hard. Oh, and +1 upvote, of course. $\endgroup$ Dec 10 '14 at 22:46
  • $\begingroup$ @sos440, Thank for pointing out this. I will fix it. $\endgroup$
    – xpaul
    Dec 11 '14 at 0:24
6
$\begingroup$

So here is a solution which follows the logic of both @gar's and @xpaul's but avoids the issue of poles.

Let $I(a)$ be defined by

$$ I(a) = \int_{0}^{1} \frac{2x}{1-x^{2}} \log \left| \frac{x + \sqrt{1-x^{2}}}{x + a\sqrt{1-x^{2}}} \right| \, dx. $$

The value we want to calculate is $I(-1)$. With the substitution $t = x/\sqrt{1-x^{2}}$ (this substitution is essentially the same as in @Felix's answer), it follows that

$$ I(a) = \int_{0}^{\infty} \frac{2t}{1+t^{2}} \log \left| \frac{1+t}{a+t} \right| \, dt = \Re \int_{0}^{\infty} \frac{2t}{1+t^{2}} \log \left( \frac{1+t}{a+t} \right) \, dt, $$

where the branch of logarithm is chosen to avoid the upper half-plane. Then by DCT,

$$ I(a) = \lim_{\epsilon \downarrow 0} \Re \int_{i\epsilon}^{\infty+i\epsilon} \frac{2t}{1+t^{2}} \log \left( \frac{1+t}{a+t} \right) \, dt =: \lim_{\epsilon \downarrow 0} I_{\epsilon}(a). $$

Now, for $0 < \epsilon \ll 1$, we have

\begin{align*} I_{\epsilon}'(a) &= - \Re \int_{i\epsilon}^{\infty+i\epsilon} \frac{2t}{(1+t^{2})(t+a)} \, dt \\ &= - \frac{2a}{1+a^{2}} \Re \int_{i\epsilon}^{\infty+i\epsilon} \left( \frac{1}{a} \frac{1}{1+t^{2}} + \frac{t}{1+t^{2}} - \frac{1}{t + a} \right) \, dt \\ &= - \frac{2a}{1+a^{2}} \Re \left( \frac{\pi}{2a} - \frac{1}{a}\arctan(i\epsilon) + \log(a+i\epsilon) - \log\sqrt{1-\epsilon^{2}} \right) \\ &= - \frac{2a}{1+a^{2}} \left( \frac{\pi}{2a} + \log\sqrt{a^{2}+\epsilon^{2}} - \log\sqrt{1-\epsilon^{2}} \right). \end{align*}

This gives, together with the initial condition $I_{\epsilon}(1) = 0$, that

$$ I_{\epsilon}(a) = \int_{a}^{1} \frac{2s}{1+s^{2}} \left( \frac{\pi}{2s} + \log\sqrt{s^{2}+\epsilon^{2}} - \log\sqrt{1-\epsilon^{2}} \right) \, ds. $$

Taking $\epsilon \downarrow$, again by DCT, we get

$$ I(a) = \int_{a}^{1} \frac{\pi + 2s\log|s|}{1+s^{2}} \, ds. $$

Plugging $a = -1$ we finally obtain the desired answer.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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