3
$\begingroup$

I was viewing past questions of the MIT Integration Bee and came across the problem:

$$ \int \frac{x^2+1}{x^4-x^2+1}\,dx$$

According to MIT, first semester calculus should be sufficient enough to solve this problem.

I have tried many conventional methods, unsuccessfully, and would appreciate any and all help.

Edit: Changed LaTeX formatting

| cite | improve this question | |
$\endgroup$
3
$\begingroup$

Maybe not as elegant but more intuitive, there's always partial fractions.

$$\int\frac{x^2+1}{x^4-x^2+1}dx=\int\frac{x^2+1}{x^4+2x^2+1-3x^2}dx=\int\frac{x^2+1}{(x^2+1+x\sqrt3)(x^2+1-x\sqrt3)}dx=$$ $$\frac12\int\frac{dx}{(x+\frac{\sqrt3}2)^2+\frac14}+\frac12\int\frac{dx}{(x-\frac{\sqrt3}2)^2+\frac14}=$$ $$\int\frac{2dx}{(2x+\sqrt3)^2+1}+\int\frac{2dx}{(2x-\sqrt3)^2+1}=$$ $$\tan^{-1}(2x+\sqrt3)+\tan^{-1}(2x-\sqrt3)$$

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Thanks for the feedback. I enjoyed your solution because the answer you achieved would also work better if the original question were a definite integral, rather than an indefinite integral. This is because the other answer has a domain excluding 0, whereas your answer does not. So, for instance, if the question were a definite integral with 0 as one of the limits of integration, only your answer could be used to compute a real value. $\endgroup$ – A is for Ambition Jun 7 '14 at 20:43
4
$\begingroup$

$$F=\frac{x^2+1}{x^4-x^2+1}=\frac{1+\dfrac1{x^2}}{x^2-1+\frac1{x^2}}$$

Observe that $\displaystyle\int\left(1+\dfrac1{x^2}\right)=x-\frac1x$

So,$\displaystyle F=\frac{1+\dfrac1{x^2}}{\left(x-\frac1x\right)^2+2-1}$

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ While the other answer was the one that immediately came to my mind, this one is quite elegant. :) $\endgroup$ – apnorton Jun 7 '14 at 17:14
  • $\begingroup$ So the answer should then be arctan(x-(1/x)), right? $\endgroup$ – A is for Ambition Jun 7 '14 at 17:17
  • $\begingroup$ @user155812, Elementary formula, right? $\endgroup$ – lab bhattacharjee Jun 7 '14 at 17:20
  • $\begingroup$ Yes, it would seem like it. Also, WolframAlpha says the answer is arctan(x/(1-x^2)). Wolfram's answer and your answer are slightly different functions with different domains. In this case, which answer is the "correct" or acceptable answer? $\endgroup$ – A is for Ambition Jun 7 '14 at 17:22
  • 1
    $\begingroup$ @user155812, $$\arctan\left(x-\frac1x\right)=\arctan\frac{x^2-1}x=-\arctan\left(\frac{1-x^2}x\right)$$ $$=\text{arccot}\left(\frac{1-x^2}x\right)-\frac\pi2=\arctan\frac x{1-x^2}-\frac\pi2$$ $\endgroup$ – lab bhattacharjee Jun 8 '14 at 5:35

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.