Suppose I want to integrate this (I chose an easy one): $$\int \frac {dx}{\sqrt{x^2-4}}$$ Method 1: (Trig Substitution) $x=2\sec(\theta)$ $$\int \frac {dx}{\sqrt{x^2-4}}=\int \frac{\sec(\theta)\tan{(\theta)}}{|\tan(\theta)|}d\theta$$

Now removing the absolute value I find is tough. The inverse secant function splits out values of $\theta$ between $0$ and $\pi/2$, and $\pi/2$ to $\pi$. How can I resolve this? Tangent is positive from $0$ and $\pi/2$ - first quadrant, and negative from $\pi/2$ to $\pi$. I know most books overlook this, but I want to be most accurate.

I say "Method 1" because I also have another question right after this.

  • $\begingroup$ Why the absolute value? That is not common place in these types of problems. I suggest dropping the absolute value and cancelling the tangents. This leaves you with just an integral of secant. For that , I highly recommend looking up the answer. That thing is just about impossible to integrate without a ridiculously clever trick. $\endgroup$ – graydad Sep 2 '14 at 3:09
  • $\begingroup$ Are you trying to find the antiderivative for $x > 2$ or for $x < -2$? $\endgroup$ – JimmyK4542 Sep 2 '14 at 3:10
  • $\begingroup$ @JimmyK4542, not sure what you mean, but I do want to split the integral from when $\tan(\theta)>0$ and $\tan(\theta)<0$ user166967, we have an absolute value, one does not simply ignore it. Doing that, you are losing half of the solution. I have $|\tan(\theta))|$, which CAN be POSITIVE and NEGATIVE depending on the domain. $\endgroup$ – CalculusNerd Sep 2 '14 at 3:16
  • 1
    $\begingroup$ That being said, I do know the antiderivative of secant, which you can get by multiplying top and bottom by secantx+tangentx. $\endgroup$ – CalculusNerd Sep 2 '14 at 3:18
  • 1
    $\begingroup$ I think you can rewrite this integral by using pythagorean theorem. It eventually became the integral of sec(x). I never encountered a need for absolute value bars. $\endgroup$ – 123 Sep 2 '14 at 3:20

Notice that the integrand $\dfrac{1}{\sqrt{x^2-4}}$ is only defined on $(-\infty,-2)\cup(2,\infty)$.

Hence, we can consider the antiderivative on $x > 2$ and on $x < -2$ seperately.

If $x > 2$, then we substitute $x = 2\sec \theta$ where $0 < \theta < \tfrac{\pi}{2}$.

Over this range, $\tan \theta > 0$. So, $\sqrt{x^2-4} = 2\tan \theta$, and thus,

$\displaystyle\int\dfrac{dx}{\sqrt{x^2-4}} = \int \dfrac{2\sec \theta \tan \theta}{|2\tan \theta|} \,d\theta = \int \dfrac{\sec \theta \tan \theta}{\tan \theta} \,d\theta = \int \sec \theta \,d\theta$

$= \ln|\sec\theta + \tan \theta|+C = \ln\left|\frac{x}{2}+\frac{\sqrt{x^2-4}}{2}\right|+C = \ln\left(\frac{x}{2}+\frac{\sqrt{x^2-4}}{2}\right)+C$

If $x < -2$, then we substitute $x = 2\sec \theta$ where $\tfrac{\pi}{2} < \theta < \pi$.

Over this range, $\tan \theta < 0$. So, $\sqrt{x^2-4} = -2\tan \theta$, and thus,

$\displaystyle\int\dfrac{dx}{\sqrt{x^2-4}} = \int \dfrac{2\sec \theta \tan \theta}{|2\tan \theta|} \,d\theta = \int \dfrac{\sec \theta \tan \theta}{-\tan \theta} \,d\theta = -\int \sec \theta \,d\theta$

$= -\ln|\sec\theta + \tan \theta|+C = -\ln\left|\frac{x}{2}-\frac{\sqrt{x^2-4}}{2}\right|+C = \ln\left|\dfrac{1}{\frac{x}{2}-\frac{\sqrt{x^2-4}}{2}}\right|+C$.

$= \ln\left|\dfrac{\frac{x}{2}+\frac{\sqrt{x^2-4}}{2}}{\left(\frac{x}{2}-\frac{\sqrt{x^2-4}}{2}\right)\left(\frac{x}{2}+\frac{\sqrt{x^2-4}}{2}\right)}\right|+C = \ln\left|\dfrac{\frac{x}{2}+\frac{\sqrt{x^2-4}}{2}}{\frac{x^2}{4}-\frac{x^2-4}{4}}\right|+C = \ln\left|\frac{x}{2}+\frac{\sqrt{x^2-4}}{2}\right|+C$

As you can see, the two expressions are the same. So, in this case, we don't lose anything by carelessly writing $\sqrt{x^2-4} = 2\tan \theta$ instead of $|2\tan \theta|$.

Alternatively, if $x < -2$, then substitute $x = 2\sec \theta$ where $\pi < \theta < \frac{3\pi}{2}$. Then $\tan \theta > 0$ over this range, and so, we don't need to worry about the absolute value signs.

  • $\begingroup$ Jeez, that was perfect, and the answer I was looking for :D $\endgroup$ – CalculusNerd Sep 2 '14 at 4:05
  • $\begingroup$ But just a quite question...how does x>2 imply that $0<\theta <\pi/2$? $\endgroup$ – CalculusNerd Sep 2 '14 at 4:08
  • $\begingroup$ We need to pick a range for $\theta$ such that for every $x > 2$ we can write $x = 2\sec \theta$ for exactly one value of $\theta$ in our range. One such range is $0 < \theta < \pi/2$, although there are several other choices. $\endgroup$ – JimmyK4542 Sep 2 '14 at 4:12
  • $\begingroup$ I understand that, but I don't see how you figured out that the range must be $0 < \theta < \pi/2$. How did you work that out? $\endgroup$ – CalculusNerd Sep 2 '14 at 4:18
  • $\begingroup$ Try looking at the graph of $x = 2\sec \theta$. You'll see that as $\theta$ ranges from $0$ to $\pi/2$, $x$ increases from $2$ to $\infty$. $\endgroup$ – JimmyK4542 Sep 2 '14 at 4:23

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.