3
$\begingroup$

Recently while working on an interesting problem, I'm stuck on evaluating the following daunting but interesting integral:

$$\int_0^{1/\sqrt{3}}\dfrac{\arctan(x)\ln(1-3x^2)}{1+x^2}\,\mathrm{d}x$$

I have been working on it since quite some time, I've tried some trivial substitutions and integration by parts but it didn't get me anywhere. Any help would be highly appreciated.

Thanks for reading.

$\endgroup$
6
  • $\begingroup$ Isn’t this improper because of the log term? $\endgroup$
    – Randall
    Jul 4, 2021 at 3:22
  • $\begingroup$ Also, doesn’t this fail to converge since the log term goes to $-\infty$? $\endgroup$
    – Randall
    Jul 4, 2021 at 3:27
  • 1
    $\begingroup$ @Randall logarithmic singularitities are integrable $\endgroup$ Jul 4, 2021 at 3:35
  • $\begingroup$ @NinadMunshi ah you are right. $\endgroup$
    – Randall
    Jul 4, 2021 at 3:43
  • $\begingroup$ Let $x=\tan u$ and that clears the denominator. Then note that $$1-3\tan^2 u = \frac{2\cos(2u)-1}{\cos^2 u}.$$ This should be able to split up the log. $\endgroup$
    – Mr Pie
    Jul 4, 2021 at 4:25

2 Answers 2

4
$\begingroup$

Define the function $\mathcal{I}:\mathbb{R}_{>0}\rightarrow\mathbb{R}$ via the (convergent) improper integral

$$\mathcal{I}{\left(a\right)}:=-8\int_{0}^{a}\mathrm{d}x\,\frac{\arctan{\left(x\right)}\ln{\left(1-a^{-2}x^{2}\right)}}{1+x^{2}}.\tag{1}$$

This answer will consider the problem of the general evaluation of $\mathcal{I}$ and obtain a closed-form expression for it in terms of elementary functions and polylogarithms.

Note that this includes the OP's integral as the following particular value:

$$-\frac18\mathcal{I}{\left(\frac{1}{\sqrt{3}}\right)}=\int_{0}^{\frac{1}{\sqrt{3}}}\mathrm{d}x\,\frac{\arctan{\left(x\right)}\ln{\left(1-3x^{2}\right)}}{1+x^{2}}\approx-0.124354.$$


Recall that the standard Clausen functions are defined by the following Fourier series:

$$\operatorname{Cl}_{2m}{\left(\theta\right)}:=\sum_{n=1}^{\infty}\frac{\sin{\left(n\theta\right)}}{n^{2m}};~~~\small{m\in\mathbb{N}\land\theta\in\mathbb{R}},\tag{2a}$$

$$\operatorname{Cl}_{2m+1}{\left(\theta\right)}:=\sum_{n=1}^{\infty}\frac{\cos{\left(n\theta\right)}}{n^{2m+1}};~~~\small{m\in\mathbb{N}\land\theta\in\mathbb{R}}.\tag{2b}$$

The Clausen function of order $2$ may also be given by the integral representation

$$\operatorname{Cl}_{2}{\left(\theta\right)}=-\int_{0}^{\theta}\mathrm{d}\varphi\,\ln{\left(\left|2\sin{\left(\frac{\varphi}{2}\right)}\right|\right)};~~~\small{\theta\in\mathbb{R}}.$$

The following pair of special values and pair of derivatives are obtained directly from the Fourier series definitions $(2)$:

$$\operatorname{Cl}_{2m}{\left(\theta\right)}=0;~~~\small{m\in\mathbb{N}},$$

$$\operatorname{Cl}_{2m+1}{\left(0\right)}=\sum_{n=1}^{\infty}\frac{1}{n^{2m+1}}=\zeta{\left(2m+1\right)};~~~\small{m\in\mathbb{N}},$$

$$\frac{d}{d\theta}\operatorname{Cl}_{2m+1}{\left(\theta\right)}=-\operatorname{Cl}_{2m}{\left(\theta\right)};~~~\small{m\in\mathbb{N}\land\theta\in\mathbb{R}},$$

$$\frac{d}{d\theta}\operatorname{Cl}_{2m+2}{\left(\theta\right)}=\operatorname{Cl}_{2m+1}{\left(\theta\right)};~~~\small{m\in\mathbb{N}\land\theta\in\mathbb{R}}.$$

Then, the following integration formulas for higher order Clausen functions follow quite simply from the fundamental theorem of calculus:

$$\int_{0}^{\theta}\mathrm{d}\varphi\,\operatorname{Cl}_{2m}{\left(\varphi\right)}=\zeta{\left(2m+1\right)}-\operatorname{Cl}_{2m+1}{\left(\theta\right)};~~~\small{m\in\mathbb{N}\land\theta\in\mathbb{R}},$$

$$\int_{0}^{\theta}\mathrm{d}\varphi\,\operatorname{Cl}_{2m+1}{\left(\varphi\right)}=\operatorname{Cl}_{2m+2}{\left(\theta\right)};~~~\small{m\in\mathbb{N}\land\theta\in\mathbb{R}}.$$


Suppose $a\in\mathbb{R}_{>0}$, and set $\alpha:=\arctan{\left(a\right)}$. Then, $0<\alpha<\frac{\pi}{2}\land a=\tan{\left(\alpha\right)}$, and we find

$$\begin{align} \mathcal{I}{\left(a\right)} &=-8\int_{0}^{a}\mathrm{d}x\,\frac{\arctan{\left(x\right)}\ln{\left(1-a^{-2}x^{2}\right)}}{1+x^{2}}\\ &=-8\int_{0}^{\tan{\left(\alpha\right)}}\mathrm{d}x\,\frac{\arctan{\left(x\right)}\ln{\left(1-\cot^{2}{\left(\alpha\right)}x^{2}\right)}}{1+x^{2}}\\ &=-8\int_{0}^{\alpha}\mathrm{d}\varphi\,\varphi\ln{\left(1-\cot^{2}{\left(\alpha\right)}\tan^{2}{\left(\varphi\right)}\right)};~~~\small{\left[\arctan{\left(x\right)}=\varphi\right]}\\ &=-8\int_{0}^{\alpha}\mathrm{d}\varphi\,\varphi\ln{\left(\frac{\sin^{2}{\left(\alpha\right)}\cos^{2}{\left(\varphi\right)}-\cos^{2}{\left(\alpha\right)}\sin^{2}{\left(\varphi\right)}}{\sin^{2}{\left(\alpha\right)}\cos^{2}{\left(\varphi\right)}}\right)}\\ &=-8\int_{0}^{\alpha}\mathrm{d}\varphi\,\varphi\ln{\left(\csc^{2}{\left(\alpha\right)}\sec^{2}{\left(\varphi\right)}\sin{\left(\alpha-\varphi\right)}\sin{\left(\alpha+\varphi\right)}\right)}\\ &=8\int_{0}^{\alpha}\mathrm{d}\varphi\,\varphi\left[\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+\ln{\left(\cos^{2}{\left(\varphi\right)}\right)}-\ln{\left(\sin{\left(\alpha-\varphi\right)}\right)}-\ln{\left(\sin{\left(\alpha+\varphi\right)}\right)}\right]\\ &=8\int_{0}^{\alpha}\mathrm{d}\varphi\,\varphi\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}\\ &~~~~~+8\int_{0}^{\alpha}\mathrm{d}\varphi\,\varphi\left[2\ln{\left(\cos{\left(\varphi\right)}\right)}-\ln{\left(\sin{\left(\alpha-\varphi\right)}\right)}-\ln{\left(\sin{\left(\alpha+\varphi\right)}\right)}\right]\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}\\ &~~~~~+8\int_{0}^{\alpha}\mathrm{d}\varphi\int_{0}^{\varphi}\mathrm{d}\vartheta\,\left[2\ln{\left(\cos{\left(\varphi\right)}\right)}-\ln{\left(\sin{\left(\alpha-\varphi\right)}\right)}-\ln{\left(\sin{\left(\alpha+\varphi\right)}\right)}\right]\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}\\ &~~~~~+8\int_{0}^{\alpha}\mathrm{d}\vartheta\int_{\vartheta}^{\alpha}\mathrm{d}\varphi\,\left[2\ln{\left(\cos{\left(\varphi\right)}\right)}-\ln{\left(\sin{\left(\alpha-\varphi\right)}\right)}-\ln{\left(\sin{\left(\alpha+\varphi\right)}\right)}\right].\\ \end{align}$$

Then,

$$\begin{align} \mathcal{I}{\left(a\right)} &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}\\ &~~~~~+8\int_{0}^{\alpha}\mathrm{d}\vartheta\int_{\vartheta}^{\alpha}\mathrm{d}\varphi\,\left[2\ln{\left(\cos{\left(\varphi\right)}\right)}-\ln{\left(\sin{\left(\alpha-\varphi\right)}\right)}-\ln{\left(\sin{\left(\alpha+\varphi\right)}\right)}\right]\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}\\ &~~~~~+8\int_{0}^{\alpha}\mathrm{d}\vartheta\int_{\vartheta}^{\alpha}\mathrm{d}\varphi\,\left[-\ln{\left(2\sin{\left(\alpha-\varphi\right)}\right)}-\ln{\left(2\sin{\left(\alpha+\varphi\right)}\right)}+2\ln{\left(2\cos{\left(\varphi\right)}\right)}\right]\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}\\ &~~~~~+8\int_{0}^{\alpha}\mathrm{d}\vartheta\,\bigg{[}-\int_{\vartheta}^{\alpha}\mathrm{d}\varphi\,\ln{\left(2\sin{\left(\alpha-\varphi\right)}\right)}-\int_{\vartheta}^{\alpha}\mathrm{d}\varphi\,\ln{\left(2\sin{\left(\alpha+\varphi\right)}\right)}\\ &~~~~~+2\int_{\vartheta}^{\alpha}\mathrm{d}\varphi\,\ln{\left(2\cos{\left(\varphi\right)}\right)}\bigg{]}\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}\\ &~~~~~+8\int_{0}^{\alpha}\mathrm{d}\vartheta\,\bigg{[}-\int_{0}^{\alpha-\vartheta}\mathrm{d}\varphi\,\ln{\left(2\sin{\left(\varphi\right)}\right)}-\int_{\alpha+\vartheta}^{2\alpha}\mathrm{d}\varphi\,\ln{\left(2\sin{\left(\varphi\right)}\right)}\\ &~~~~~+2\int_{\frac{\pi}{2}-\alpha}^{\frac{\pi}{2}-\vartheta}\mathrm{d}\varphi\,\ln{\left(2\sin{\left(\varphi\right)}\right)}\bigg{]}\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+4\int_{0}^{\alpha}\mathrm{d}\vartheta\,\bigg{[}-\int_{0}^{2\alpha-2\vartheta}\mathrm{d}\varphi\,\ln{\left(2\sin{\left(\frac{\varphi}{2}\right)}\right)}\\ &~~~~~-\int_{2\alpha+2\vartheta}^{4\alpha}\mathrm{d}\varphi\,\ln{\left(2\sin{\left(\frac{\varphi}{2}\right)}\right)}+2\int_{\pi-2\alpha}^{\pi-2\vartheta}\mathrm{d}\varphi\,\ln{\left(2\sin{\left(\frac{\varphi}{2}\right)}\right)}\bigg{]};~~~\small{\left[\varphi\mapsto\frac{\varphi}{2}\right]}\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+4\int_{0}^{\alpha}\mathrm{d}\vartheta\,\bigg{[}\operatorname{Cl}_{2}{\left(2\alpha-2\vartheta\right)}\\ &~~~~~+\operatorname{Cl}_{2}{\left(4\alpha\right)}-\operatorname{Cl}_{2}{\left(2\alpha+2\vartheta\right)}+2\operatorname{Cl}_{2}{\left(\pi-2\alpha\right)}-2\operatorname{Cl}_{2}{\left(\pi-2\vartheta\right)}\bigg{]},\\ \end{align}$$

and then,

$$\begin{align} \mathcal{I}{\left(a\right)} &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+4\int_{0}^{\alpha}\mathrm{d}\vartheta\,\bigg{[}\operatorname{Cl}_{2}{\left(2\alpha-2\vartheta\right)}\\ &~~~~~+2\operatorname{Cl}_{2}{\left(2\alpha\right)}-\operatorname{Cl}_{2}{\left(2\alpha+2\vartheta\right)}-2\operatorname{Cl}_{2}{\left(\pi-2\vartheta\right)}\bigg{]}\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+8\alpha\operatorname{Cl}_{2}{\left(2\alpha\right)}\\ &~~~~~+4\int_{0}^{\alpha}\mathrm{d}\vartheta\,\left[\operatorname{Cl}_{2}{\left(2\alpha-2\vartheta\right)}-\operatorname{Cl}_{2}{\left(2\alpha+2\vartheta\right)}-2\operatorname{Cl}_{2}{\left(\pi-2\vartheta\right)}\right]\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+8\alpha\operatorname{Cl}_{2}{\left(2\alpha\right)}\\ &~~~~~+2\int_{0}^{2\alpha}\mathrm{d}\vartheta\,\left[\operatorname{Cl}_{2}{\left(2\alpha-\vartheta\right)}-\operatorname{Cl}_{2}{\left(2\alpha+\vartheta\right)}-2\operatorname{Cl}_{2}{\left(\pi-\vartheta\right)}\right];~~~\small{\left[\vartheta\mapsto\frac{\vartheta}{2}\right]}\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+8\alpha\operatorname{Cl}_{2}{\left(2\alpha\right)}\\ &~~~~~+2\left[\int_{0}^{2\alpha}\mathrm{d}\vartheta\,\operatorname{Cl}_{2}{\left(2\alpha-\vartheta\right)}-\int_{0}^{2\alpha}\mathrm{d}\vartheta\,\operatorname{Cl}_{2}{\left(2\alpha+\vartheta\right)}-2\int_{0}^{2\alpha}\mathrm{d}\vartheta\,\operatorname{Cl}_{2}{\left(\pi-\vartheta\right)}\right]\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+8\alpha\operatorname{Cl}_{2}{\left(2\alpha\right)}\\ &~~~~~+2\left[\int_{0}^{2\alpha}\mathrm{d}\vartheta\,\operatorname{Cl}_{2}{\left(\vartheta\right)}-\int_{2\alpha}^{4\alpha}\mathrm{d}\vartheta\,\operatorname{Cl}_{2}{\left(\vartheta\right)}-2\int_{\pi-2\alpha}^{\pi}\mathrm{d}\vartheta\,\operatorname{Cl}_{2}{\left(\vartheta\right)}\right]\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+8\alpha\operatorname{Cl}_{2}{\left(2\alpha\right)}\\ &~~~~~+2\left[\operatorname{Cl}_{3}{\left(0\right)}-\operatorname{Cl}_{3}{\left(2\alpha\right)}+\operatorname{Cl}_{3}{\left(4\alpha\right)}-\operatorname{Cl}_{3}{\left(2\alpha\right)}+2\operatorname{Cl}_{3}{\left(\pi\right)}-2\operatorname{Cl}_{3}{\left(\pi-2\alpha\right)}\right]\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+8\alpha\operatorname{Cl}_{2}{\left(2\alpha\right)}\\ &~~~~~+2\operatorname{Cl}_{3}{\left(0\right)}+4\operatorname{Cl}_{3}{\left(\pi\right)}+2\operatorname{Cl}_{3}{\left(4\alpha\right)}-4\operatorname{Cl}_{3}{\left(2\alpha\right)}-4\operatorname{Cl}_{3}{\left(\pi-2\alpha\right)}\\ &=4\alpha^{2}\ln{\left(\sin^{2}{\left(\alpha\right)}\right)}+8\alpha\operatorname{Cl}_{2}{\left(2\alpha\right)}-\zeta{\left(3\right)}+\operatorname{Cl}_{3}{\left(4\alpha\right)}.\blacksquare\\ \end{align}$$


For the particular case in which $a=\frac{1}{\sqrt{3}}$, we then have $\alpha=\arctan{\left(\frac{1}{\sqrt{3}}\right)}=\frac{\pi}{6}$, and hence

$$\begin{align} \mathcal{I}{\left(\frac{1}{\sqrt{3}}\right)} &=4\left(\frac{\pi}{6}\right)^{2}\ln{\left(\sin^{2}{\left(\frac{\pi}{6}\right)}\right)}+8\cdot\frac{\pi}{6}\operatorname{Cl}_{2}{\left(2\cdot\frac{\pi}{6}\right)}-\zeta{\left(3\right)}+\operatorname{Cl}_{3}{\left(4\cdot\frac{\pi}{6}\right)}\\ &=-\frac{2\pi^{2}}{9}\ln{\left(2\right)}+\frac{4\pi}{3}\operatorname{Cl}_{2}{\left(\frac{\pi}{3}\right)}-\zeta{\left(3\right)}+\operatorname{Cl}_{3}{\left(\frac{2\pi}{3}\right)}\\ &=\frac{4\pi}{3}G-\frac{13}{9}\zeta{\left(3\right)}-\frac{2\pi^{2}}{9}\ln{\left(2\right)},\\ \end{align}$$

where here $G$ denotes Gieseking's constant.


$\endgroup$
1
$\begingroup$

This is not a full answer.

Consider using Feynman trick $$I(a)=\int_0^{\frac 1 {\sqrt 3}}\frac{ \tan ^{-1}(a x)\log \left(1-3 x^2\right)}{1+x^2}\,dx$$ $$I'(a)=\int_0^{\frac 1 {\sqrt 3}}\frac{x \log \left(1-3 x^2\right)}{\left(x^2+1\right) \left(a^2 x^2+1\right)}\,dx=\frac{\text{Li}_2\left(\frac{1}{4}\right)-\text{Li}_2\left(\frac{a^2}{a^2+3}\right)}{ 2 \left(a^2-1\right)}$$ We cannot split the next integral but a CAS finds the antiderivative (very long !).

The computation of $$\int_0^1 \frac{\text{Li}_2\left(\frac{1}{4}\right)-\text{Li}_2\left(\frac{a^2}{a^2+3}\right)}{2 \left(a^2-1\right)}\,da$$ took a very long time because we need to take the limits (not just apply the formula which gives indeterminate at both bounds). I have an awful result containing a bunch of polylogarithms with complex argements.

The last integrand is quite nice and, by the end, I used a series expansion of it built around $a=0$. This gives $$\frac{\text{Li}_2\left(\frac{1}{4}\right)-\text{Li}_2\left(\frac{a^2}{a^2+3}\right)}{2 \left(a^2-1\right)}=k+\sum_{n=1}^p \left({b_n}+k\right)\, a^{2n}+O(a^{2p+2}) \qquad \text{with} \qquad k=-\frac{1}{2}\,\text{Li}_2\left(\frac{1}{4}\right)$$ where the $b_n$'s are $$\left\{\frac{1}{6},\frac{1}{8},\frac{265}{1944},\frac{115}{864},\frac{78173}{583200}, \frac{234029}{1749600},\frac{11474681}{85730400},\frac{137669537}{1028764800},\frac {33455693611}{249989846400},\cdots\right\}$$

Now, as a function of $p$, the value of the integral $$\left( \begin{array}{ccc} 1 & \frac{1}{18}-\frac{2 }{3}\text{Li}_2\left(\frac{1}{4}\right)& -0.1228795372 \\ 2 & \frac{29}{360}-\frac{23 }{30}\text{Li}_2\left(\frac{1}{4}\right) & -0.1246448011 \\ 3 & \frac{3403}{34020}-\frac{88 }{105}\text{Li}_2\left(\frac{1}{4}\right) & -0.1242890078 \\ 4 & \frac{31249}{272160}-\frac{563 }{630}\text{Li}_2\left(\frac{1}{4}\right) & -0.1243695042 \\ 5 & \frac{178228}{1403325}-\frac{3254}{3465} \text{Li}_2\left(\frac{1}{4}\right) & -0.1243499428 \\ 6 & \frac{240448777}{1751349600}-\frac{88069 }{90090} \text{Li}_2\left(\frac{1}{4}\right) & -0.1243549342 \\ 7 & \frac{3361000121}{22986463500}-\frac{45536 }{45045}\text{Li}_2\left(\frac{1}{4}\right) & -0.1243536145 \\ 8 & \frac{1926817784779}{12504636144000}-\frac{1593269 }{1531530}\text{Li}_2\left(\frac{1}{4}\right) & -0.1243539730 \\ 9 & \frac{2325692917041587}{14433476269212000}-\frac{15518938 }{14549535} \text{Li}_2\left(\frac{1}{4}\right) & -0.1243538735 \\ 10 & \frac{203084487005661923}{1212412006613808000}-\frac{31730711 }{29099070} \text{Li}_2\left(\frac{1}{4}\right)& -0.1243539016 \\ 11 & \frac{1661411146332456037}{9585632427290419500}-\frac{372177944 }{334639305}\text{Li}_2\left(\frac{1}{4}\right) & -0.1243538935 \\ 12 & \frac{822107289750189315107}{4601103565099401360000}-\frac{3788707301 }{3346393050}\text{Li}_2\left(\frac{1}{4}\right) & -0.1243538959 \end{array} \right)$$

$\endgroup$

You must log in to answer this question.

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