When I try to find the derivative of $f(x) = \sqrt[3]{x} \sin(x)$ at $x=0$, using the formal definition of first derivative, I get this:

$$ f'(0) = \lim\limits_{x \to 0} \frac{\sqrt[3]{x} \sin(x)-0}{x-0},$$ which gives zero.

However, when I use derivative rules I get that:

$$ f'(x) = {\sin(x) \frac{1}{3\sqrt[3]{x^2}}+\cos(x)\sqrt[3]{x}} $$

and thus $f'(0)$ doesn't exist.

Why does this happen? what's the reason behind it?

  • 6
    I'm not sure I follow. Note that $\sin(x)\sim x$ around $x=0$, and so $\sin(x)/x^{2/3}\sim x^{1/3}$, which goes to $0$ as $x\to0$. So $f'(x)\to0$, as expected. – AccidentalFourierTransform Nov 8 at 16:56
  • 1
    @AccidentalFourierTransform your comment actually constitutes an answer and shouldn't be just a comment. – Ister Nov 9 at 9:17
  • +1 for a good question. – Randall Nov 9 at 13:25
up vote 43 down vote accepted

The rule $(fg)'=f'g+fg'$ works where $f$ and $g$ are differentiable. And $\sqrt[3]x$ is not differentiable at $x=0$.

  • Oh, alright I see. But, why is the first derivative able to find it? I mean, I have never really understood the differences between direct derivative rules and formal definition. – Just_Cause Nov 8 at 15:16
  • 17
    It is possible that $g$ is not differentiable and $fg$ is. – ajotatxe Nov 8 at 15:18
  • 2
    It makes sense now. – Just_Cause Nov 8 at 15:19

$$ f'(x) = {\sin(x) \frac{1}{3\sqrt[3]{x^2}}+\cos(x)\sqrt[3]{x}} $$

and thus $f'(0)$ doesn't exist.

You have $\frac{\sin(x)}{3\sqrt[3]{x^2}}$, which is $\frac 00$, or indeterminate. If you express sine in terms of its Taylor Series, and divide each term by ${3\sqrt[3]{x^2}}$, you will see that you get a series that evaluates to zero at $x=0$. The lesson here is just because you can put something in an indeterminate form, doesn't mean that it doesn't exist. For instance, one can rewrite $x^2$ as $\frac {x^3} x$; that doesn't mean that $x^2$ doesn't exist when $x=0$.

  • 1
    Your argument is appealing but incorrect. One can write $x^2$ as $x^3/x$ only when $x\neq 0$. What you are trying to do here is to find limit $\lim_{x\to 0}f'(x)$ and show that it equals $f'(0) $. This works here only because $f'$ is continuous in this case. Consider another example where $g(x) =x^2\sin(1/x),g(0)=0$ then $g'(0)=0$ but $\lim_{x\to 0}g'(x)$ does not exist. – Paramanand Singh Nov 8 at 18:53
  • 2
    @ParamanandSingh And the OP's application of the product rule is valid for $x \neq 0$. So my example is analogous. – Acccumulation Nov 8 at 18:56
  • 2
    I think the intent of your answer is to explain how we can make sense of the formula obtained by product rule even when $x=0$. This is something which I believe is not the right way to think about the situation. – Paramanand Singh Nov 8 at 18:58
  • Wait @Paramanand Singh, how's $f'(x)$ continous here? And in your example why is $g(0) = 0$ ? I'm a bit confused – Stephen Alexander Nov 9 at 5:55
  • @StephenAlexander: note that $f'(0)=0$ and $\lim_{x\to 0}f'(x)=0 $ (check this!) so that makes $f'$ continuous. For my example I have chosen $g(0)=0$ this ensures that $g$ is continuous at $0$. – Paramanand Singh Nov 9 at 5:59

When I try to find the derivative of $f(x) = \sqrt[3]{x} \sin(x)$ at $x=0$, using the formal definition of first derivative, I get this: $$ f'(0) = \lim\limits_{x \to 0} \frac{\sqrt[3]{x} \sin(x)-0}{x-0}=0 $$

This is correct since: $$ \lim_{x \to 0} \frac{\sqrt[3]{x} \sin(x)-0}{x-0} =\lim_{x \to 0} \sqrt[3]{x}\cdot \frac{ \sin(x)}{x}=0.\tag{1} $$

However, when I use derivative rules I get that: $$ f'(x) = {\sin(x) \frac{1}{3\sqrt[3]{x^2}}+\cos(x)\sqrt[3]{x}}\tag{2} $$

(2) does not contradict (1) since (2) is only valid for $x\neq 0$.

and thus $f'(0)$ doesn't exist.

This implication is false:

[Edited later (thanks to comments by MMASRP63 and Paramanand Singh)] (2) only implies that the limit $\lim_{x\to 0}f'(x)$ does not exists. In other words, $f'(x)$ is not continuous at $x=0$. $$ \lim_{x\to 0} {\sin(x) \frac{1}{3\sqrt[3]{x^2}}+\cos(x)\sqrt[3]{x}} =\lim_{x\to 0}\frac{\sin x}{x}\frac{x}{3\sqrt[3]{x^2}} +\lim_{x\to 0}\cos(x)\sqrt[3]{x}=1\cdot 0+ 1\cdot 0=0\tag{3} $$ which together with (1) implies that $f'$ is actually continuous at $x=0$.

  • 2
    I believe that $f'$ is continuous at $x=0$ in this instance? – MMASRP63 Nov 8 at 18:13
  • 1
    Your last two sentences are incorrect and one should observe that using continuity of $f'$ at $0$ to evaluate $f'(0)$ is a very silly/roundabout method. – Paramanand Singh Nov 8 at 18:56
  • @MMASRP63: I should really not do the calculation in my head. What a silly mistake. Thanks for pointing it out! – user587192 Nov 8 at 19:49
  • @ParamanandSingh: thank you for pointing it out! – user587192 Nov 8 at 19:49

I want to add something to user587192's answer.

Indeed, there is no contradiction, and

$$\lim_{x \to 0} f'(x) = 0$$

as shown before. However, if you don't want to evaluate the derivative using the formal definition, you can use a theorem:

Suppose $f'(x)$ exists in a deleted neighbourhood of $a$ and $\lim_{x \to a} f'(x) $ exists and equals $L$. Then $f'(a)$ exists and equals $\lim_{x \to a} f'(x)$.

This is an immediate consequence of L'Hospital's Rule. Notice that both $f(x)-a$ and $x-a$ are differentiable on a deleted neighbourhood of $a$, and $\lim_{x \to a} \frac{f'(x)}{1}$ exists and equals $L$, we conclude that $\lim_{x \to a} \frac{f(x)-a}{x-a}$ exists and equals $L$. Hence $f'(a)=L$.

Indeed, in your example, it would be much faster to calculate $f'(0)$ using the formal definition. However, in some cases the above theorem does help.

Your Answer

 

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.