I would like to specify that english is not my primary language so I had to do some translation here, I do really hope every mathematical term I used is correct, if not please forgive me :)

Let's say I want to prove that $\lim_{x \to 2} x^2 = 4$, from what I've read, in order to do that, the solution to $|f(x)-l|<\epsilon$ must contain a neighborhood of $x_{0}$ such as $x_{0}-\delta<x<x_{0}+\delta$ which proves that $|x-x_{0}|<\delta$

So in this case $|x^2-4|<\epsilon$ must contain a neighborhood of $2$, so I proceed by solving that:




and this is as far as my book goes

The limit is proven true because $\sqrt{4-\epsilon}<x<\sqrt{4+\epsilon}$ is a neighborhood of $x_{0}$ which is 2, so when $x\to2$ then $y\to4$

but what I don't understand is: how do we know that $\sqrt{4-\epsilon}<x<\sqrt{4+\epsilon}$ is really a neighborhood of $2$? That doesn't seem obvious to me.

Let's say that instead of a "true" limit I wanted to prove (Or prove wrong) a "false" limit such as $\lim_{x \to 2} x^2 = 8$ which is not true. How do I do that? I can do the usual steps:




but how do I check if this is really a neighborhood of 2 so that $|x-x_{0}|<\delta$ or not? How do I know if this limit is really proven or not?

PS: This is what I've learned reading my book, if anything I said is wrong, please let me know! And if you know a faster/better/simpler method of proving a limit please tell me!

  • $\begingroup$ Unless the book worked out examples other than this, this is a crappy book. Nobody proves things like that. $\endgroup$
    – Cauchy
    Aug 12, 2017 at 15:22
  • $\begingroup$ @Cauchy translated word for word: "To prove that $\lim_{x\to x_{0}}=l$ we need to solve $|f(x)-l|<\epsilon$ and prove that the solutions contains a neighborhood of $x_{0}$ for every $\epsilon>0$" $\endgroup$ Aug 12, 2017 at 15:28
  • $\begingroup$ Maybe of interest: math.stackexchange.com/questions/195726/… $\endgroup$ Aug 12, 2017 at 17:30

2 Answers 2


I suppose by a neighborhood of $2$ you are talking about an open interval of center $2$. If so, then note that $0 < \varepsilon < 4$ implies that $\sqrt{4-\varepsilon} < 2 < \sqrt{4+\varepsilon}$. Let $a := \min \{ 2-\sqrt{4-\varepsilon}, \sqrt{4+\varepsilon}-2 \}$. Then $x \in (2-a,2+a)$ implies $|x^{2}-4| < \varepsilon$. Here the function $x \mapsto x^{2}$ is defined at $x=2$, so there is no need to exclude the possibility $x=2$. All in all, this is what the author tries to convey, I think. I am silent on whether the author succeeds though...

To see why this argument works, note that it shows that for every $0 < \varepsilon < 4$ we can construct some $\delta > 0$ (the $a$ above is an instance) such that $|x-2| < \delta$ implies $|x^{2}-4| < \varepsilon$. For every $\varepsilon \geq 4$, we use the same $\delta$, which suffices.

I am not in any conflict of interest with the book or the author or anything related to it :). I would like to say that the proof is an intention to make things easy. It is a subject matter if it is good or not. I would say that the choice of the method is with pedagogical consideration but the lack of explanation backfires.

You can do it with triangle inequality. Note that $|x^{2}-4| = |x-2||x+2|$ for all real $x$. If $|x-2| < 1$, then $|x| - 2 \leq |x-2| < 1$ by triangle inequality, so $|x| < 3$ and hence $|x+2| \leq |x| + 2 < 5$ by triangle inequality, so $|x-2||x+2| < 5|x-2|$. If $\varepsilon > 0$, then $5|x-2| < \varepsilon$ if $|x-2| < \varepsilon/5$. Combining all the majorants for $|x-2|$, we conclude that $|x-2| < \min \{ 1, \varepsilon/5 \}$ implies $|x^{2}-4| < \varepsilon$.

  • $\begingroup$ People are saying that this method of proving limits is bad, do you know where I can find a better one? $\endgroup$ Aug 12, 2017 at 15:44
  • $\begingroup$ @EliaPerantoni, Hold on, I got a minute :). $\endgroup$
    – Megadeth
    Aug 12, 2017 at 15:45
  • $\begingroup$ @EliaPerantoni, You may check my newest version now. $\endgroup$
    – Megadeth
    Aug 12, 2017 at 15:50
  • $\begingroup$ Thansk! It will take me some time to understand this since I don't know what triangle inequalities are but I'll try :) $\endgroup$ Aug 12, 2017 at 15:53
  • $\begingroup$ But as I've said in a comment, can't I just assume that $\epsilon$ is extremely small yet positive and see what happens? In this case we would have $\sqrt{4-0^+}<x<\sqrt{4+0^+}$ becomes $2^-<x<2^+$ ? I've tried this with different excersises and it seems to work pretty well! $\endgroup$ Aug 12, 2017 at 15:59

The square root function is increasing and continuous and $\sqrt{4} = 2$, hence for $\varepsilon>0$ we have that $\sqrt{4-\varepsilon}<2$ and $\sqrt{4+\varepsilon}>2$.

  • $\begingroup$ It is better to write ".......is increasing and continuous AT THE NEIGHBORHOOD OF $x=2$ and ......." $\endgroup$
    – Piquito
    Aug 12, 2017 at 15:34
  • $\begingroup$ Because $\epsilon$ can as small as we want as long as it is $\epsilon>0$ right? So if we choose an incredibly small $\epsilon$ so $\sqrt{4-\epsilon}$ and $\sqrt{4+\epsilon}$ are almost equal to $\sqrt{4}$, now we have $2^-<x<2^+$ so we can assume $x_{0}=2$ and the limit is proven? Is this correct? $\endgroup$ Aug 12, 2017 at 15:37
  • $\begingroup$ @EliaPerantoni Yes, that's basically correct. (I say "basically" because your intuition is correct, however it depends on how rigorously you want to prove it). $\endgroup$
    – Eff
    Aug 12, 2017 at 18:07

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.