Let $f(x)=e^{\frac{1}{x}}$. In which of the following domains is it uniformly continuous?

a) (0,1)

b) (1,$\infty$)

c) (1,2)

My Attempt: For uniform continuity, we have a given $\epsilon$>0 and we need to find an $\delta$>0 such that whenever $|x-y|<\delta$, the following holds : $|f(x)-f(y)|<\epsilon$. In case (b) and (c), clearly the function $e^{\frac{1}{x}}$ is bounded. So we can always find such an $\delta$, where the above result holds true. But in case (a) when $x$ takes values very close to $0$, the function $f(x)$ diverges. So we can say it is not uniformly continuous.

Is my approach and justification correct? Can someone please give proper reasoning and mathematical explanation because I am not completely satisfied with what I have presented above.

Thanks in advance.


On intervals where your function has a Lipschitz constant (essentially, a bound on the first derivative) it will be uniformly continuous. Your function has a bounded derivative on both $(1,2)$ and $(1,\infty)$, so it is UC there. It is not UC on $(0,1)$ since you have arbitrarily large slopes close to $0$. In other words you can pick two points close to zero in such a way that $$ \frac{f(x_1)-f(x_2)}{x_1-x_2}\quad (*) $$ is as large as you want. Given an $\varepsilon$, no uniform $\delta$ will satisfy the definition of UC, you would have to choose a smaller and smaller $\delta$ the closer you get to zero, because of $(*)$.

| cite | improve this answer | |
  • $\begingroup$ I have not yet come across a Lipschitz function. Thanks for the help. +1 $\endgroup$ – s1mple Oct 7 '19 at 2:35
  • $\begingroup$ Can you please say something about my approach? the idea is it correct? etc $\endgroup$ – s1mple Oct 7 '19 at 2:36
  • 1
    $\begingroup$ About your approach: being bounded is not enough to be able to choose a universal $\delta$. For example, take $f(x)=\sin(1/x)$ on $(0,1)$. This function is bounded but it is not uniformly continuous because it has oscillations with arbitrarily large frequency close to zero. What you need is to have a controlled slope. Your argument for part $a)$ seems correct: if your function has limit $\pm\infty$ at the end-point, it cannot be UC. $\endgroup$ – GReyes Oct 7 '19 at 3:14
  • $\begingroup$ Thank you for the help. +1 $\endgroup$ – s1mple Oct 7 '19 at 3:16

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.