Let $E\subseteq \mathbb{R}$ and $f:E\to \mathbb{R}$ be a function. When we define the limit of $f(x)$ as $x\to x_0$ we require $x_0$ to be the limit point of $E$.

Why do we require $x_0$ to be the limit point of $E$?

The definition says: We say that $A=\lim \limits_{x\to x_0} f(x)$ iff $\forall \varepsilon>0$ $\exists \delta=\delta(\varepsilon)>0$ : $\forall x\in E$ with $0<|x-x_0|<\delta$ $\Rightarrow$ $|f(x)-A|<\varepsilon$.

My thoughts: If $x_0$ is the limit of point of $E$ then $\{x\in E:0<|x-x_0|<\delta\}\neq \varnothing$ for any $\delta>0$. Probably this is one of the reasons but I may be wrong!

But what if $x_0$ is not a limit of $E$? For example, let's take some function $f(x)$ defined on $\{0\}\cup (1,2)$, where $x_0=0$.

What difficulties do we have?

Would be thankful if someone provide detailed answer with examples.

  • $\begingroup$ $0$ is a limit point of your $E$ $\endgroup$ – Exodd Jun 17 at 19:43
  • 1
    $\begingroup$ "A" limit point, not "the" limit point. $\endgroup$ – Stefan Jun 17 at 19:46
  • 3
    $\begingroup$ "Probably this is one of the reasons": This is exactly the reason. If we take your definition literally, every function would converge to all values simultaneously at a non-limit-point. $\endgroup$ – Stefan Jun 17 at 19:49
  • 1
    $\begingroup$ @ZFR: There is no right or wrong. It is just not useful. Imagine everytime you in a theorem or proof that you need to exclude or discuss the case that $x_0$ not a limit point. $\endgroup$ – user251257 Jun 17 at 22:00
  • 2
    $\begingroup$ Maybe nothing, but at least the term $\lim_{x\rightarrow x_0}f(x)$ is only well-defined if there is a unique limit. $\endgroup$ – Stefan Jun 17 at 22:01

The def'n is flawed. It implies that if $x_0$ is not a limit point of $E$ then $$A=\lim_{x\to x_0}f(x)$$ is true for any and all $A,$ e.g. $0=\lim_{x\to x_0}f(x)=1.$

If $S$ is a sentence and $\forall y\; (T)\;$ (or, respectively $\exists y\; (T)\;$) occurs in $S$, where $(T)$ is the rest of the sentence, then in the negation of $S$ this part will change to $\exists y (\neg T)\;$ (respectively $\forall y\;(\neg T)\;$).

So the def'n says $$A\ne \lim_{x\to x_0} f(x) \text { iff } \exists e>0\;\forall d>0\; \exists x\, (0<|x-x_0|<d \land |f(x)-f(x_0)\ge e).$$ Now if $0<|x-x_0|<d$ then $f(x)$ does not exist unless $x\in E,$ so a consequence of the def'n is

$ A\ne \lim_{x\to x_0}f(x)\implies x_0$ is a limit point of $E,$


if $x_0$ is not a limit point of $E$ then $A=\lim_{x\to x_0}f(x).$

The def'n should be amended by inserting "$x_0$ is a limit point of $E$ and" just after the bold-face "iff".

| cite | improve this answer | |
  • $\begingroup$ In some contexts it is convenient to also allow $f(x_0)=\lim_{x\to x_0} f(x)$ when $x_0$ is an isolated point of $E$. $\endgroup$ – DanielWainfleet Jun 17 at 21:21

Probably, Heine's definition (that is equivalent to epsilon-delta definition) of the limit of a function may answer your question:

For functions on the real line, one way to define the limit of a function is in terms of the limit of sequences. In this setting: $$\lim _{x\to x_0}f(x)=A$$ if and only if for all sequences $x_{n}$ (with $x_{n} \not = x_0$, $\forall n$) converging to $x_0$ the sequence $f(x_n)$ converges to $A$.

Therefore, if $x_0$ would not be a limit point of $\mathcal{D}(f)$, then how would such $x_n$ exist?

| cite | improve this answer | |

I like to work with the following definition of the limit of a function between metric spaces:

Let $(X,d_{X})$ and $(Y,d_{Y})$ be metric spaces such that $f:X\to Y$ is a function. Consider as well that we are given a set $E\subseteq X$ such that $x_{0}\in X$ is an adherent point of $E$ and $L\in Y$. Then we say that $f$ converges to $L$ as $x$ approaches $x_{0}$ along $E$ iff for every $\varepsilon > 0$, there corresponds a $\delta > 0$ such that, for every $x\in E$, \begin{align*} d_{X}(x,x_{0}) < \delta \Rightarrow d_{Y}(f(x),L) <\varepsilon \end{align*}

It is slightly different from your definition, because we demand that $d_{X}(x,x_{0}) < \delta$ instead of $0 < d_{X}(x,x_{0}) < \varepsilon$. However your definition can be considered a particular case from the definition I have mentioned. Indeed, it suffices to consider $E\backslash\{x_{0}\}$ and both of them are the same. The set $E$ tells us how we are approaching $x_{0}$.

Still, the answer to your question has not been given. More precisely, why do we need to consider $x_{0}\in X$ to be an adherent point of $E$? In order to answer it, remember that the set of adherent points of $E$ equals $\overline{E} = \text{int}(E)\cup\partial(E)$. Consequently, if $x\not\in\overline{E}$, then $x\in\text{ext}(E)$. This means there exists a positive real number $r > 0$ such that $B(x_{0},r)\cap E = \varnothing$. In other words, for a $\delta$ small enough $(\delta \leq r)$, there is no $x\in E$ such that $d_{X}(x,x_{0}) < \delta$, which makes the definition of limit vacuous.

That's why we require $x_{0}\in X$ to be an adherent point of $E$.

| cite | improve this answer | |

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.