43
$\begingroup$

I calculated the derivative of $\arctan\left(\frac{1+x}{1-x}\right)$ to be $\frac{1}{1+x^2}$. This is the same as $(\arctan)'$. Why is there no $c$ that satisfies $\arctan\left(\frac{1+x}{1-x}\right) = \arctan(x) +c$?

$\endgroup$
  • 2
    $\begingroup$ Why do you say that there is no $c$? $\endgroup$ – Thomas Andrews Apr 8 '15 at 2:21
  • 1
    $\begingroup$ This seems like one of those cases where trig identities involving constants are not obvious... $\endgroup$ – TravisJ Apr 8 '15 at 2:22
  • 5
    $\begingroup$ Plotting the two clearly shows they don't differ by a constant but also provides a hint as to why not. $\endgroup$ – Cascabel Apr 8 '15 at 18:23
  • $\begingroup$ Piecewise, they do differ by a constant. Just with a discontinuity at x=1. And we explicitly define arctan(t) to have a discontinuity at t=+/-inf $\endgroup$ – smci Apr 9 '15 at 11:38
65
$\begingroup$

The problem is that $\arctan \frac{1+x}{1-x}$ isn't defined at $x = 1$, and in particular isn't differentiable there. In fact, we have $$ \arctan \frac{1+x}{1-x} - \arctan x = \begin{cases} \frac{\pi}{4} & x < 1, \\ -\frac{3\pi}{4} & x > 1. \end{cases} $$ So the difference is piecewise constant.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ While formally you are right in the first sentence, note that $\frac x{x^2}$ is not defined at $0$, but it has a continuation there, which is also differentiable at $0$. Here the discontinuity is not a removable one, which is why this doesn't work. $\endgroup$ – Asaf Karagila Apr 8 '15 at 11:23
  • 1
    $\begingroup$ @AsafKaragila: I don't think you are talking about the function which you meant to talk about. The function you mentioned cannot be extended continuously at $x=0$. $\endgroup$ – Marc van Leeuwen Apr 8 '15 at 13:06
  • 1
    $\begingroup$ @Marc: You're right, I wrote that comment from my phone, so MathJax mistakes are easy to come by. And of course, that I meant $\frac{x^2}{x}$. Thanks for pointing that out! $\endgroup$ – Asaf Karagila Apr 8 '15 at 13:13
  • $\begingroup$ d00d, what a kewl constant! It's like not even constant! $\endgroup$ – bjb568 Apr 8 '15 at 21:19
4
$\begingroup$

Hint: $$\tan(x+y)=\frac{\tan x + \tan y}{1-\tan x\tan y}$$

What happens when $\tan y=1$?

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ The problem is that $\arctan$ always returns an answer in $(-\pi/2,\pi/2)$. $\endgroup$ – Yuval Filmus Apr 8 '15 at 2:24
1
$\begingroup$

Piecewise, they do differ by a constant. Just with a discontinuity at x=1.

And we explicitly define arctan(t) to have that discontinuity at t=+/-inf

| cite | improve this answer | |
$\endgroup$

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.