I have started topology from Munkres. Here in section $14$, I am stuck up in a definition. It says,

Let $X$ be a set with a simple order relation; assume $X$ has more than one element. Let B be the collection of all sets of following types:

  1. All open intervals $(a, b)$ in $X$.
  2. All intervals of form $[a, b)$ of $X$
  3. All intervals of form $(a, b]$ of $X$.

The collection B is a basis for order topology on $X$.

But here we have omitted $[a, b]$. Why?

I have read about standard topology, lower limit topology, upper limit topology. But set of all closed interval don't generate topology. My intuition says that these two are somehow linked.


2 Answers 2


Read Munkres carefully. He writes:

Let $X$ be a set with a simple order relation and $|X|>1$. Let $\mathcal{B}$ be a collection of the following:

  1. Intervals of the form $(a,b)$ in $X$.

  2. Intervals of the form $[a_0,b)$, for $a_0$ the smallest element of $X$, if one exists.

  3. Intervals of the form $(a,b_0]$ for $b_0$ the largest, if one exists.

Then $\mathcal{B}$ is a basis for the order topology on $X$.

It does not say intervals of the form $[a,b)$ and $(a,b]$ for any $a,b$, but only when the endpoints are the extrema of the order relation.

For example, take $[0,1]$ with the subspace topology in $\mathbb{R}$ with the standard topology. Then open sets which form a basis are of the form $(a,b), 0<a<b<1$, $[0,a)$ and $(b,1]$. This is what he means.

The order topology is meant to mimic our understanding of the order topology on $\mathbb{R}$. We want sets to be open which are unions of "open intervals", i.e. not containing their endpoints.

Edit: I'll add a bit more.

If you want the question as initially stated, then in fact every subset is open, so this is the discrete topology. To see this, if $[a,M)$ and $(-M,a]$ are in the basis for each $a, |a|<M$, then $\{a\}=[a,M)\cap (-M,a]$ is an open set for any $a$.

An interesting note, though this is a bit more advanced, is that you CAN include one of those half open intervals, either $[a,b)$ or $(a,b]$, and get a topology which is not trivial or the standard order topology. When we take $\mathbb{R}$ and include sets of the form $[a,b)$ in the basis we get the so called lower limit topology. This has many interesting properties: it is non-metrizable, it doesn't have a countable basis, and is totally disconnected, all properties we normally associate with $\mathbb{R}$ with the order topology.

  • $\begingroup$ I couldn't find notation for a0 and b0 so I wrote it that way in hope that it would be edited. $\endgroup$
    – divya
    Aug 17, 2015 at 18:00
  • $\begingroup$ Ok. This is nice. Thank you. $\endgroup$
    – divya
    Aug 17, 2015 at 18:09
  • $\begingroup$ You said "The equivalent statement of $2,3$ in $\mathbb{R}$ are $(-\infty,a)$ and $(b,\infty)$" but I can't interpret this any way that makes it true. $\Bbb R$ has no largest or smallest element, so there is no analog of (2) or (3). In the extended real numbers, there is $\infty$ but then the sets added by (3) look like $(b, \infty]$, not like $(b, \infty)$. In $\Bbb R$, the unbounded interval $(b, \infty)$ simply is not a basic open set in this basis. $\endgroup$
    – MJD
    Nov 20, 2017 at 15:21
  • $\begingroup$ Also you claimed that $[a,b) \cup (a,b] = [a,b]$ and that $[a,b] =\{a\}$ when $a=b$. But when $a=b$, the first claim is wrong because both intervals are empty. $\endgroup$
    – MJD
    Nov 20, 2017 at 15:26
  • $\begingroup$ To your first, that's exactly the point. There is no largest/smallest element, so this is the analog of 2,3 in $\mathbb{R}$ (really, it is the union of all basis elements $(a,b)$, see Munkres for the exact details. $(a,\infty)$ need not be included in a basis for this reason, but it is a subbasis for this basis). For your second comment, yes, it only makes sense when $a<b$. The real reason is $(-\infty,a] \cap [a,\infty) = \{a\}$ is open. I will amend. $\endgroup$
    – Moya
    Nov 20, 2017 at 16:47

I assume your space is connected.You can't include $[a,b]$ ,for arbitrary $ a,b$ in your basis.As $[a,b]$ is closed set,being complement of open set $[-m,a)\cup (b,M]$,where $m$ and $M$ are smallest and largest elements respectively(if exist).

As space is connected by assumption,so $[a,b]$ can't be open,otherwise space will be disconnected.As elements of basis elements are open sets.So $[a,b]$ can't belongs to basis,being a closed subset.

  • $\begingroup$ Note that $[a, b]$ is open if it is the entire set. $\endgroup$ Aug 18, 2015 at 15:17

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.