I have a function $f(x)=\frac{1}{x}+\sin(\frac{1}{x})$ which I have proved that $\lim_{x \to0^+}f(x)=\infty$, and I need to prove that $\lim_{x \to0^+}f'(x) \ne -\infty$.

I have got that $f'(x)=-\frac{1}{x^2}-\frac{\cos(\frac{1}{x})}{x^2}$

I am trying to prove that:

$$\lim_{x \to 0^+}-\frac{1}{x^2}-\frac{\cos(\frac{1}{x})}{x^2}$$

Does not equal to $-\infty$ (or to prove that it doesn't exists at all).

I have tried to assume by contradiction that $$\lim_{x \to 0^+}-\frac{1}{x^2}-\frac{\cos(\frac{1}{x})}{x^2}=-\infty$$

So there exists $\delta_{1}>0$ such that for every $0<x<\delta_{1}$ we say that $$f(x)<\square $$

We choose $\delta=\delta_{1}$, and let $0<x<\delta$ so:

$$f(x)=-\frac{1}{x^2}-\frac{\cos(\frac{1}{x})}{x^2}=-\frac{1}{x^2}(1+\cos(\frac{1}{x}))\le -\frac{2}{x^2}<\frac{2}{\delta^2}=k$$

I do not know what to choose instead of $\square$, and if what I did is correct.

Appreciate your help!

I have also tried to think about the continuity of $f$, and saying that it is not bounded so it can't be continuous, so the limit doesn't exists.

Thanks a lot!

  • 1
    $\begingroup$ Big hint - $\cos(\pi+2k\pi) =-1$ for all $k$ $\endgroup$ – Calvin Khor Jan 16 at 12:15
  • $\begingroup$ I have try thinking about if for a while, but I do not know what to do with your hint... $\endgroup$ – MathLover Jan 16 at 13:30

Scrap the whole proof by contradiction thing, it's not necessary.

Hint: Find the values of $x$ such that $\frac{1}{x} = 2k\pi$, and $\frac{1}{x} = (2k+1)\pi,$ where $k \in \mathbb{Z}$.

Edit: In fact, as Calvin Khor pointed out in the comments to this answer, you only need one of the above two sequences to show that $\lim_{x \to0^+}f'(x) \ne -\infty$, which was the original question.

Further edit: I think OP doesn't properly understand what is meant by $lim_{x \to c^+}g(x) = -\infty$. I also needed a sanity check, which is why I posted this question.

If $g(x)$ is a real function, then a definition of $\lim_{x \to c^+}g(x) = -\infty$ is:

for each $\gamma \in \mathbb{R}, \exists \delta > 0 $ such that $x \in (c,c+\delta) \implies g(x) < \gamma$.

So an equivalent definition of $\ \neg\left(\lim_{x \to c^+}g(x) = -\infty\ \right) $ is:

there exists $\beta\in\mathbb{R}$ such that for every $\delta>0$, there exists $x'\in(c,c+\delta)$ with $g(x')\ge \beta$

If no such $\beta$ exists, then $\ \lim_{x \to c^+}g(x) = -\infty\ .$

Note that $g$ need not be continuous at $c$ in either of my definitions.

  • $\begingroup$ Your hint gives one way to show its not convergent, but you need only the second half to show that it does not converge to $-\infty$ $\endgroup$ – Calvin Khor Jan 16 at 12:22
  • $\begingroup$ Yes that's true. $\endgroup$ – Adam Rubinson Jan 16 at 12:23
  • $\begingroup$ Thank you, what do you mean by finding $x$'s values? $\endgroup$ – MathLover Jan 16 at 13:33
  • $\begingroup$ What do you think about showing that $f'$ isn't bounded? By choosing $n \in \mathbb{N}$ such that $x=\frac{1}{\pi +2 \pi n}$? $\endgroup$ – MathLover Jan 16 at 13:40
  • $\begingroup$ $k$ represents an integer (other than $0$). Do you know how to rearrange equations? And do you know what the graph of $\cos$ looks like? Then you should try to draw the graph of $\cos\left(\frac{1}{x}\right)$ using my hint. $\endgroup$ – Adam Rubinson Jan 16 at 13:41

Because there exists a subsequence converging to $0$:

if you pick $\{x_k\}$ to be such that $x_k=\frac{1}{\pi + 2\pi k}$, then $$f^\prime (x_k) = -(\pi + 2\pi k)^2 \Big(1+cos(\pi + 2\pi k)\Big) = 0,$$ since $$cos(\pi + 2\pi k)=-1 \mbox{, } \forall k;$$ Also the sequence of $x_k$ converges to $0$ as you can observe letting $k\to +\infty$.

  • $\begingroup$ Beat you to it by about 20 seconds. I'm getting quicker at this. $\endgroup$ – Adam Rubinson Jan 16 at 12:18
  • $\begingroup$ Lol, I'm cool with taking my time. $\endgroup$ – Measure me Jan 16 at 12:19
  • $\begingroup$ Why can I choose this $x$? What is the benefit of showing that $f'(x_{k})=0$? $\endgroup$ – MathLover Jan 16 at 13:32
  • $\begingroup$ If you apply the definition of limit knowing this you see it doesn't work. The idea is that you always have an $x_k$ for some $k$ such that it is as close to $0^+$ as you want, so the limit can't be $-\infty$: in particular this shows that, if it exists, it is $0$. $\endgroup$ – Measure me Jan 16 at 13:39
  • $\begingroup$ Cool! Thank you for that! $\endgroup$ – MathLover Jan 16 at 13:43

With $z:=\dfrac1x$, rewrite as

$$\lim_{z\to\infty}-z^2(1+\cos z).$$

Then it is clear that the function alternates between $0$ and $-2z^2$.

  • $\begingroup$ Why is it clear? $\endgroup$ – MathLover Jan 16 at 18:32
  • $\begingroup$ @MathLover: do you really need a tutorial on the cosine function ? $\endgroup$ – Yves Daoust Jan 16 at 18:47
  • $\begingroup$ No, I know that $cos(?)$ is between -1 and 1, but $-z^2 \to -\infty$ so I do not understand why is it clear... $\endgroup$ – MathLover Jan 16 at 19:43
  • $\begingroup$ @MathLover: isn't $1+\cos z$ between $0$ and $2$ ? $\endgroup$ – Yves Daoust Jan 16 at 20:53
  • 1
    $\begingroup$ @MathLover: does $0$ tend to infinity ? $\endgroup$ – Yves Daoust Jan 16 at 21:07

Part 1:

You can try making it a bit "easier" by substitution.

$$\lim_{x \to 0^+}-\frac{1}{x^2}-\frac{cos(\frac{1}{x})}{x^2}\underset{t=\frac{1}{x}}{\Rightarrow}$$ $$\lim_{t\to\infty}-t^{2}\left(1+\cos t\right)$$ Now it should be easier to negate by definition, as you can easily see that no matter how big $t$ is, and how small $f'$ is for those big $t$'s, you can choose $t<t_1$ such that $f'(t_1)=0$

If you'll still be struggling, make sure to leave a comment and I will help you further.

Part 2:

Assuming, in contradiction that $\lim_{t\to\infty}-t^{2}\left(1+\cos t\right)=-\infty$

Then there exists $M_{1}<0$ such that for every $t<M_{1}$ we have:$$f\left(t\right)<-1$$ Choosing $M_{1}>t_{1}=\lfloor M_{1}\rfloor\cdot2\pi-\pi$ , we get: $$f\left(t\right)=0>-1$$ Contradiction!

  • $\begingroup$ Thank you for that, I have already tried it before actually... $\endgroup$ – MathLover Jan 16 at 14:10
  • $\begingroup$ Look at the edit $\endgroup$ – GuyPago Jan 16 at 14:32
  • $\begingroup$ Awesome! Thank you so much! $\endgroup$ – MathLover Jan 16 at 14:46

Assume by contradiction that for every $K<0$ there exists a $\delta>0$ such that for every $0<x<\delta$, we get $f'(x)<K$.

We choose $n \in \mathbb{N}$ such that $n>K$, and we choose $x_{n}= \frac{1}{\pi +2 \pi n}$.

Note that $0<\frac{1}{\pi +2 \pi n}<1$. (So in this case $\delta=1$)

And: $f'(x)=-(\pi +2\pi n)(\cos(\pi+2\pi n)+1)=0>K$

So $f'(x)>K$, in contradiction!

Is that correct?


  • $\begingroup$ Your first line, "Assume by contradiction that there exists $K<0$ such that $f'(x)<K$", is meaningless. Do you mean: Assume by contradiction that there exists $K<0$ such that $f'(x)<K $ for all $x$ ? Or: Assume by contradiction that there exists $K<0$ such that $f'(x)<K$ for all $ x>c$? Or something else? $\endgroup$ – Adam Rubinson Jan 16 at 14:49
  • $\begingroup$ I meant $x>0$ not $x>c$. $\endgroup$ – Adam Rubinson Jan 16 at 15:11
  • $\begingroup$ That for every $\delta>0$ there exists $0<x<1$, in this case, $\delta=1$ and we chose $x$ to be $x_{n}$ $\endgroup$ – MathLover Jan 16 at 15:12
  • $\begingroup$ I don't understand what you mean. If you think you should edit your answer then please do so. $\endgroup$ – Adam Rubinson Jan 16 at 15:39
  • $\begingroup$ I have edited the answer :) Thanks a lot for your time my friend! $\endgroup$ – MathLover Jan 16 at 15:56

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.