8
$\begingroup$

I am studying topology, and the definition of a limit point on my book is:

Let $A$ be a subset of a topological space $X$ and $x$ a point of $X$, we say that $x$ is a limit point of $A$ if every neighborhood of $x$ intersects $A$ at some point other than $x$ itself.

The definition of a neighborhood is an open subset containing $x$.

My question is, why do we require the intersection to be at some point other than $x$ itself?

$\endgroup$
4
  • 1
    $\begingroup$ I agree with you, this is a strange requirement. Usually we say not nessarily $x$ itselt $\endgroup$
    – user58697
    Commented Sep 3, 2023 at 2:23
  • $\begingroup$ @user58697 I think that you are saying the definition of adherent points, not limit points. $\endgroup$
    – flowing
    Commented Sep 3, 2023 at 3:04
  • 1
    $\begingroup$ See this. Limit points of sets cannot be isolated. If we didn't add the "other than itself" requirement then a discrete set like $\{1,2\}$ would have two limit points. The concept of a limit point is related to the continuum, not to discrete sets. $\endgroup$
    – John Douma
    Commented Sep 3, 2023 at 12:24
  • 3
    $\begingroup$ This is really just a convention, there is no deep reason behind it. In fact, in French topology textbooks, instead of defining limit points, we define adherence points instead, which is almost exactly the same definition but without the "other than x itself" bit. You can easily switch between the two definitions: a limit point is "a non-isolated adherence point", and an adherence point is "a point which is either a limit point of the set or a point of the set". $\endgroup$
    – Stef
    Commented Sep 3, 2023 at 12:55

5 Answers 5

24
$\begingroup$

The reason is that if we didn't have the phrase "other than $x$ itself", then every point of $A$ would automatically be a limit point, which is not the desired situation.

The concept of a limit point is meant (among other things) to distinguish them from isolated points of $A$. For example, if $X=\Bbb R$ and $A=(0,1) \cup \{2\}$, the point $x=2$ is qualitatively different as a member of $A$: there is no sequence of other points of $A$ that converges to $2$. The book's definition of "limit point" gives us a way of describing this phenomenon. (Check that the set of limit points of this $A$ is $[0,1]$.)

$\endgroup$
3
  • $\begingroup$ Thank you! It seems to me that we add $\textit{points other than itself}$ because we don't want the point to be isolated. It also reminds of the definition of $\textit{continuity}$ in real analysis, that a function is $\textit{continuous}$ at an isolated point, why is it defined in this way? $\endgroup$
    – Coco
    Commented Sep 3, 2023 at 18:59
  • $\begingroup$ Obviously continuity and limit points are somewhat related, but in what way does the definition of limit points remind you of the definition of continuity? And what definition of continuity are you using? I think we usually say a function is continuous at an isolated point of its domain because it's a consequence of how continuity is defined and it doesn't make anything easier or better to change the definition just to exclude functions from being continuous at isolated points. $\endgroup$ Commented Sep 3, 2023 at 20:59
  • $\begingroup$ This is arguably true for limit points as well. Which is why, as others have pointed out, some people choose to talk about adherent points instead of limit points. Clearly someone at some point in time thought it was worth the hassle of separating out isolated points from limit points, but what really matters is consistency. $\endgroup$ Commented Sep 3, 2023 at 21:06
3
$\begingroup$

A point being a limit point of some set $A$ means that there is an infinite sequence of points in $A$ that approaches said point. Let $X=\mathbb{R}$ and $A= (0,1) \cup \{2\}$. $1$ is a limit point of $A$ because every neighborhood of $1$ intersects the open interval $(0,1)$ and we can use this to construct a sequence. Loosely, as you take smaller and smaller neighborhoods, you can pick a point in the intersection of the neighborhood and $A$ and this should give you a sequence of points in $A$ that do actually approach your limit point (in the topological sense).

But this idea falls flat (or at least becomes uninteresting) if you drop the condition that the intersection of the neighborhoods and $A$ contain a point other than the limit point. For example, take the point $2$, although every neighborhood of $2$ contains an element of $A$, namely $2$, the sequence we get from this is $2,2,2,\ldots$ which, in many senses, is not meaningful.

$\endgroup$
3
  • 1
    $\begingroup$ This characterization works in, say, metric spaces, but for an arbitrary topological space, a limit point need not be approached by any sequence that isn't eventually constant, e.g., an uncountable limit ordinal is not approached by any countable sequence in the order topology, yet is still a limit point of the ordinals below it. $\endgroup$
    – M W
    Commented Sep 3, 2023 at 3:28
  • $\begingroup$ @MW Ahh yes you are right, my intuition from metric spaces seeped in. Greg Martin’s answer is a more effective argument here then. $\endgroup$
    – CoArp
    Commented Sep 3, 2023 at 3:33
  • 4
    $\begingroup$ Happens to the best of us. If you replace "sequence" with "net" then your characterization is fine I think. Unfortunately nets don't seem to be as widely known. $\endgroup$
    – M W
    Commented Sep 3, 2023 at 3:37
1
$\begingroup$

I would like to explain this concept in Real numbers. There are two terms, one is the limit point of a set and the other is the Adherent point.

Let $X=\mathbb{R}$ and $A \subseteq \mathbb{R}$

Limit point: A real number $x$ is limit point of $A$ if every open neighbourhood containing $x,$ conatins atleast one point $A$ other than $x$.

Example $A=[0,1] \cup \{2,3,5,7\}$ then limit points of $A$ is $[0,1]$

Adherant points: A real number $x$ is said to be adherent of $A$ if every open neighborhood of $x$ containing $x$, contains atleast one point of $A.$

$A=[0,1] \cup \{2,3,5,7\}$ then adherent points of $A$ is $[0,1] \cup \{2,3,5,7\}$

See the difference in both of these definitions, For limit point there are always infinite numbers of elements that should be present no matter how small the neighborhood you choose that is why "other than $x$ " which may not in case for adherent points

The limit point is an adherent point but not conversely.

Hopefully, I give the answer to what you are asking.

$\endgroup$
1
$\begingroup$

$\newcommand{\Ball}[3][]{B_{#2}^{#1}(#3)}\newcommand{\Del}[2]{\Ball[\times]{#1}{#2}}$There are good answers already; posting in the hope this conceptual taxonomy of the situation of a point relative to a subset is helpful to posterity.


Fix a metric space $(X, d)$, a subset $A$ of $X$, and a point $x$ of $X$. If $r$ is a positive real number, let $\Ball{r}{x}$ denote the $d$-ball of radius $r$ about $x$, i.e., the set of $x'$ in $X$ such that $d(x, x') < r$.

Precisely one of the following occurs:

  1. There exists a positive real $r$ such that $\Ball{r}{x} \subset A$. (In this case we say $x$ is an interior point of $A$.)
  2. There exists a positive real $r$ such that $\Ball{r}{x} \cap A = \varnothing$, i.e., $\Ball{r}{x} \subset (X \setminus A)$. (In this case we say $x$ is an exterior point of $A$, i.e., an interior point of the complement $X \setminus A$.)
  3. For every positive real $r$, both intersections $\Ball{r}{x} \cap A$ and $\Ball{r}{x} \cap (X \setminus A)$ are non-empty. (In this case we say $x$ is a boundary point of $A$. Because these three conditions are a logical trichotomy, every set $A$ has the same boundary as its complement $X \setminus A$.)

Further, let $\Del{r}{x} = \Ball{r}{x} \setminus \{x\}$ denote the deleted $d$-ball of radius $r$ about $x$, i.e., the ball $\Ball{r}{x}$ with $x$ removed. Conceptually, just as an open ball "contains all points of $X$ sufficiently close to $x$," a deleted ball "contains all points of $X$ sufficiently close to, but distinct from, $x$."

With notation as above, precisely one of the following occurs:

  1. There exists a positive real $r$ such that $\Del{r}{x} \cap A = \varnothing$. (If $x \in A$, we say $x$ is an isolated point of $A$. Otherwise $x$ is an exterior point of $A$, as above.)
  2. For every positive real $r$, the intersection $\Del{r}{x} \cap A$ is non-empty (In this case we say $x$ is a limit point of $A$, whether or not $x \in A$.)
$\endgroup$
0
$\begingroup$

The most basic situation in which that definition is useful is one you are probably already familiar with: computing limits of real-valued functions.

If $E\subseteq\mathbb R$, $f:E\to\mathbb R$ is a function, $c$ is a limit point of $E,$ and $l\in\mathbb R,$ then $f(x)\to l$ as $x\to c$ means: for every $\varepsilon>0$, there is a $\delta>0$ such that, for all $x\in E,$ if $0<|x-c|<\delta$ then $|f(x)-f(c)|<\varepsilon.$

Notice that this definition applies in a very general context. For instance, it allows us to say that $\frac{\sin(1/x)}{\sin(1/x)}\to 1$ as $x\to 0,$ even though the function in question is undefined for values of $x$ which are arbitrarily close to $0$. Although we could define limits whenever $c\in\mathbb R$, this has an undesirable side-effect. If $c$ is not a limit point of $E$, then there is a $r>0$ such that $(c-r,c+r)$ does not contain any points of $E$, except possibly $c$. In that situation, for every $l\in\mathbb R$, it is vacuously true that $f(x)\to l$ as $x\to c$, and so we lose the uniqueness of limits.

In summary, the definition of a limit point is useful because it leads to us a very general definition of limits, but not so general as to be unusable.

$\endgroup$
1
  • $\begingroup$ Would the person who downvoted please comment? $\endgroup$
    – Joe
    Commented Sep 4, 2023 at 14:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .