The following integral may seem easy to evaluate ...

$$ \int_{0}^{\Large\frac{\pi}{2}} \arctan \left(2 \tan^2 x\right) \mathrm{d}x = \pi \arctan \left( \frac{1}{2} \right). $$

Could you prove it?

  • $\begingroup$ Have you tried the variable $u=2.tan^2(x)$ ? $\endgroup$ – D.L. Jul 22 '14 at 9:59
  • 4
    $\begingroup$ Are you challenging people to prove it? If so, do you have your own solution? You can always post it using >! before to hide it. $\endgroup$ – user37238 Jul 22 '14 at 10:15
  • $\begingroup$ @user37238 Thanks! I did not know this functionality! Where can I have more explanations about it? Yes, you have to see this integral as a small challenge. $\endgroup$ – Olivier Oloa Jul 22 '14 at 11:06
  • $\begingroup$ Have you tried compuer algebra systems? $\endgroup$ – Mhenni Benghorbal Jul 22 '14 at 11:08
  • $\begingroup$ @Mhenni Benghorbal I didn't test it, but you are right, Mathematica finds it! >$\,\frac{\pi}{2} \text{ArcTan}\left[\frac{4}{3}\right]$ $\endgroup$ – Olivier Oloa Jul 22 '14 at 11:12

My answer is different from that you gave. Let $$ I(a)=\int_0^{\frac{\pi}{2}}\arctan(a\tan^2x)dx. $$ Than $I(0)=0$ and \begin{eqnarray} I'(a)&=&\int_0^{\frac{\pi}{2}}\frac{\tan^2x}{1+a^2\tan^4x}dx\\ &=&\int_0^\infty\frac{u^2}{(1+u^2)(1+a^2u^4)}du\\ &=&\frac{1}{1+a^2}\int_0^\infty\left(-\frac{1}{1+u^2}+\frac{1+a^2u^2}{1+a^2u^4}\right)du\\ &=&\frac{1}{1+a^2}\left(-\int_0^\infty\frac{1}{1+u^2}du+\int_0^\infty\frac{1}{1+a^2u^4}du+\int_0^\infty\frac{a^2u^2}{1+a^2u^4}\right)du\\ &=&\frac{1}{1+a^2}\left(-\frac{\pi}{2}+\frac{\pi}{2\sqrt{2}\sqrt{a}}+\frac{\sqrt{a}\pi}{2\sqrt{2}}\right) \end{eqnarray} and hence $$ I=\int_0^2I'(a)da=\pi\arctan(1+\sqrt{2a})\Big]_0^2=-\frac{\pi^2}{4}+\pi\arctan 3=\frac{\pi}{2}\arctan\frac{4}{3}.$$

| cite | improve this answer | |
  • $\begingroup$ Thanks. Our $I$ are not the same. What is your final result for your $I$? $\endgroup$ – Olivier Oloa Jul 22 '14 at 14:34
  • $\begingroup$ why you can take the differentiation inside the integral? Can you justify it? $\endgroup$ – student Jul 22 '14 at 15:09
  • $\begingroup$ @OlivierOloa,I fixed the problem. $\endgroup$ – xpaul Jul 22 '14 at 15:41
  • $\begingroup$ @QinfengLi, I just revised the answer. I made a mistake before. $\endgroup$ – xpaul Jul 22 '14 at 15:42
  • $\begingroup$ @xpaul Please, I don't understand your antepenultimate equality. $\endgroup$ – Olivier Oloa Jul 22 '14 at 15:47

Here is a result avoiding differentiation with respect to a parameter.

Set$$ I(\alpha):= \int_{0}^{\Large\frac{\pi}{2}} \arctan \left(\frac{2\alpha \:\sin^2 x}{\alpha^2-1+\cos^2 x}\right)\: \mathrm{d}x, \quad \alpha>0. $$ Then $$ I(\alpha)= \pi \arctan \left(\frac{1}{2\alpha}\right) \quad ({\star}) $$

With $ \alpha:=1$, we get $$ \int_{0}^{\Large\frac{\pi}{2}} \arctan \left(2 \tan^2 x\right) \mathrm{d}x = \pi \arctan \left( \frac{1}{2} \right). $$ To obtain $({\star})$ use the standard evaluation extended to complex numbers

$$ \int_{0}^{\Large\frac{\pi}{2}} \log \left(1+ t \sin^2 x\right) \mathrm{d}x = \pi \log \left( \frac{1+\sqrt{1+t}}{2} \right) $$

and observe that $$ \arctan (z) = \frac{i}{2} \left(\log (1-i z)-\log (1+i z)\right), \quad\Re z \neq 0. $$

| cite | improve this answer | |
  • $\begingroup$ I don't understand your last two equalities. Could you please show some more details? Cause at first glance I don't think the last two equalities are easier to be obtained than the original question. $\endgroup$ – student Jul 22 '14 at 15:12

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.