Attempt 1:

If we apply substitution $x=1/u$, then our integral will be like following;


But this doesn't make any sense:


Attempt 2:

$$\dfrac{1}{(x^2+1)(x^t+1)}=\dfrac{Ax+B}{(x^2+1)}+\dfrac{h_{t-1}x^{t-1}+h_{t-2}x^{t-2}+...+h_1x+h_0}{(x^t+1)}$$$$\to$$ $$A+h_{t-1}=0\\B+h_{t-2}=0\\B+h_0=0\\A+h_1=0$$ other $h_i$ are $0$; $$\displaystyle\int_{1/t}^t\dfrac{dx}{(x^2+1)(x^t+1)}=\displaystyle\int_{1/t}^t\left(\dfrac{Ax+B}{(x^2+1)}+\dfrac{h_{t-1}x^{t-1}+h_{t-2}x^{t-2}+...+h_1x+h_0}{(x^t+1)}a\right)$$$$\to$$$$ \displaystyle\int_{1/t}^t\dfrac{dx}{(x^2+1)(x^t+1)}=\displaystyle\int_{1/t}^t\left(\dfrac{Ax+B}{(x^2+1)}-\dfrac{Ax^{t-1}+Bx^{t-2}+Ax+B}{(x^t+1)}\right)$$

After this, I tried to apply substitution of trigonometric forms etc., but I failed.


Hint: You've shown $$I = \int_{1/t}^t \frac{x^t \, \mathrm{d}x}{(x^2+1)(x^t+1)} = \int_{1/t}^t \frac{\mathrm{d}x}{(x^2+1)(x^t+1)}$$

Now add them together to get $$2I = \int_{1/t}^t \frac{x^t + 1}{(x^2+1)(x^t+1)} \, \mathrm{d}x = \int_{1/t}^t \frac{\mathrm{d}x}{1+x^2}$$

As per Dr. MV's excellent comment, this technique works for any integral of the form $\int_{1/a}^a \frac{\mathrm{d}x}{(x^2+1)(x^b+1)}$, that is having both the upper limit be $t$ and the exponent in the integrand be $t$ as well has no relevance.

In fact, your (clever) substitution is what shows us that the value of the exponent in $(x^t+1)$ is entirely irrelevant to the problem.

  • 1
    $\begingroup$ Just beat me :) but yes, this is definitely the way to go. Apparently the presence of the $x^t$ in the numerator scared the OP? $\endgroup$ – Brevan Ellefsen May 3 '17 at 21:59
  • 1
    $\begingroup$ Indeed, I can understand why. @OP It may seem counter intuitive that $\int_a^b \frac{f}{g} = \int_a^b \frac{1}{g}$, but remember that two different functions can integrate (over a region) to the same thing, your result isn't wrong in any sense. $\endgroup$ – Zain Patel May 3 '17 at 22:02
  • 3
    $\begingroup$ (+1) You might consider adding a note that exposes that the having $t$ be both the exponent and the upper integration limit is a red herring. The approach works for any exponent on $x$ provided the limits are reciprocals of each other (i.e., $\int_a^{1/a} \frac{1}{(x^2+1)(x^b+1)}\,dx$). $\endgroup$ – Mark Viola May 3 '17 at 22:57
  • $\begingroup$ @Dr.MV Done, thanks very much for the suggestion! $\endgroup$ – Zain Patel May 3 '17 at 23:01
  • $\begingroup$ Zain, appreciative of the reference. ;-) $\endgroup$ – Mark Viola May 4 '17 at 2:35

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.