3
$\begingroup$

I have a simple question and I would appreciate if anyone could clarify for me, please.

I understand that a function can fail to differentiate when you have a "corner", a vertical assymptote or a non-removable discontinuity on a given point. But what about a function with a removable discontinuity?

Let's say that you have the function

$$y = \frac{x-1}{x-1}$$

for all $$x \neq 1$$

and

$$y=2$$

when

$$x = 1$$

Can you take the derivative of that function on $$x = 1$$

?

Thank you for your time,

Best Regards,

Bruno

$\endgroup$
  • $\begingroup$ So, in short you have $$y(x)=\begin{cases}x&\text{for }x\ne 1\\2&\text{for }x=1\end{cases}$$? That's quite discontinuous at $x=1$, and you can only "remove" that by considering a different function instead. $\endgroup$ – hmakholm left over Monica Aug 7 '14 at 18:42
  • $\begingroup$ The definition of the derivative forces an answer of no. And there is no reason to wish it were not so: saying that $y$ is $2$ at $x=1$ is a deliberate act. The situation is different if the function is simply left undefined at $x=1$. The definition of most calculus books would leave the answer at no, but it would be sensible to quietly close up and view $y(1)$ as being equal to $1$. $\endgroup$ – André Nicolas Aug 7 '14 at 18:45
  • $\begingroup$ Henning, the function OP mentioned HAS a removable discontinuity point..isnt it? $\endgroup$ – luka5z Aug 7 '14 at 18:48
  • $\begingroup$ @luka5z: No, he explicitly defined the value at $1$ to be $2$. "Removable" implies that you can make the function nice by extending it, not by changing values where it is already defined. $\endgroup$ – hmakholm left over Monica Aug 7 '14 at 18:51
  • 1
    $\begingroup$ (Also, it looks like I'm unable to divide $x-1$ by itself ...) $\endgroup$ – hmakholm left over Monica Aug 7 '14 at 18:52
0
$\begingroup$

Edited to be correct (I hope):

As Daniel Fischer points out in the comments, the left- (resp. right-) derivative of a function $f$ at $x$ is not defined unless $f$ is left- (resp. right-) continuous at $x$. So I was wrong to claim that your function $y$ is left- and right-differentiable everywhere. It is simply not differentiable at $x=1$, because it is not continuous there.

Sorry for the confusion.

$\endgroup$
  • $\begingroup$ Thanks Tony, your answer makes sense. Best Regards! $\endgroup$ – bru1987 Aug 7 '14 at 19:54
  • $\begingroup$ Just a note: right-differentiability at $a$ implies right-continuity at $a$. Analogously for left-differentability. Hence (in the one-dimensional case) semi-differentiability at $a$ implies continuity at $a$ (see third bullet point in the section). $\endgroup$ – Daniel Fischer Aug 7 '14 at 19:59
  • $\begingroup$ Oh, and neither the left nor the right derivative of the function $y$ from the question exists at $1$. That part is incorrect. $\endgroup$ – Daniel Fischer Aug 7 '14 at 20:01
  • $\begingroup$ @DanielFischer: Thank you for pointing that out! I have edited my answer accordingly. $\endgroup$ – TonyK Aug 7 '14 at 20:12
  • $\begingroup$ Thanks @TonyK for being honest and Daniel Fischer for the answer. Best regards for both! $\endgroup$ – bru1987 Aug 7 '14 at 20:32
0
$\begingroup$

The function $f(x) = \frac{x-1}{x-1}$ is really shorthand for the constant function $1$ with domain $\mathbb{R} \setminus \{1\}$. This function cannot have a derivative at $x = 1$ because $x = 1$ is not part of its domain. However, if you "remove" the discontinuity (as one often does), you can arrive at a corresponding function $g(x) = 1$ which is differentiable at $x = 1$.

Similarly a function like $h: \mathbb{R} \to \mathbb{R}$ $$h(x) = \begin{cases} 1 & \text{if } x \ne 1 \\ 2 & \text{if } x = 1 \end{cases}$$ is not differentiable at $1$ but can be made differentiable by changing the value of the function at a single point.

That $h$ is not differentiable is a result of the definition of the derivative: $$ \lim_{\Delta \to 0} \frac{h(1 + \Delta) - h(1)}{\Delta} = \lim_{\Delta \to 0} \frac{2 - 1}{\Delta} = \lim_{\Delta \to 0} \frac{1}{\Delta} $$ which does not exist.

$\endgroup$
  • $\begingroup$ I get it Goos, thanks. But let's say that I leave the function like I originally posted, with the Domain R (all real numbers). Can I take the derivative of that function when $$x = 1$$? The limit surely exists... $\endgroup$ – bru1987 Aug 7 '14 at 18:59
  • $\begingroup$ No it doesn't; your limit becomes $\lim_{h \to 0} 1/h$. $\endgroup$ – Ian Aug 7 '14 at 19:09
  • $\begingroup$ @bru1987 No, that is my function $h$ and the limit does not exist. I will clarify in my answer. $\endgroup$ – 6005 Aug 7 '14 at 19:15
  • $\begingroup$ Thanks for your explanation Goos. Best Regards! $\endgroup$ – bru1987 Aug 7 '14 at 19:55

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.