3

Any family of subsets of a set $X$ is a subbase for some topology on $X$ and the topology which results is the smallest topology containing the given collection of sets.

I'm really not sure how to start with this one. I let $S$ be the family of subsets of $X$, and take the intersection of all the topology containing $S$, then this clearly is a topology right? And by definition, $S$ is a subbase.

How do I show that this is the smallest topology that contains the given collection of sets?

Thank you for any help.

JKnecht
  • 6,543
Akaichan
  • 3,424
  • 4
  • 34
  • 60

2 Answers2

5

If you define $\tau(\mathscr{S})$ to be the intersection of all topologies $\tau$ on $X$ such that $\mathscr{S}\subseteq\tau$, then it’s trivial that $\tau(\mathscr{S})$ is the smallest topology on $X$ containing $X$: by definition if $\tau$ is any topology on $X$ with $\mathscr{S}\subseteq\tau$, then $\tau(\mathscr{S})\subseteq\tau$.

What’s not quite trivial is to prove that if $$\mathscr{B}=\left\{\bigcap\mathscr{F}:\mathscr{F}\subseteq\mathscr{S}\text{ is finite}\right\}$$ and $$\tau(\mathscr{B})=\left\{\bigcup\mathscr{U}:\mathscr{U}\subseteq\mathscr{B}\right\}\;,$$

then $\tau(\mathscr{B})=\tau(\mathscr{S})$, i.e., that the process of closing $\mathscr{S}$ under finite intersections and then under arbitrary unions yields the smallest topology containing $\mathscr{S}$. (This is the other common definition of the topology generated by an arbitrary $\mathscr{S}\subseteq\wp(X)$.)

Added: I’ll show that $\tau(\mathscr{B})=\tau(\mathscr{S})$. The first step is to show that $\mathscr{S}\subseteq\tau(\mathscr{B})$.

$\mathscr{B}$ is the set of all intersections of finite subsets of $\mathscr{S}$, so in particular $\mathscr{S}\subseteq\mathscr{B}$. And $\tau(\mathscr{B})$ is the set of all unions of subsets of $\mathscr{B}$, so in particular $\mathscr{B}\subseteq\tau(\mathscr{B})$. Thus, $\mathscr{S}\subseteq\mathscr{B}\subseteq\tau(\mathscr{B})$, i.e., $\mathscr{S}\subseteq\tau(\mathscr{B})$.

The next step is to show that $\tau(\mathscr{B})$ is actually a topology on $X$.

  • By definition each $U\in\tau(\mathscr{B})$ is a union of members of $\mathscr{B}$, and a union of unions of members of $\mathscr{B}$ is itself a union of members of $\mathscr{B}$ and therefore a member of $\tau(\mathscr{B})$, so $\tau(\mathscr{B})$ is closed under taking arbitrary unions.

  • Suppose that $U,V\in\tau(\mathscr{B})$; then there are $\mathscr{U}\subseteq\mathscr{B}$ and $\mathscr{V}\subseteq\mathscr{B}$ such that $U=\bigcup\mathscr{U}$ and $V=\bigcup\mathscr{V}$. (That is, $U$ and $V$ are both unions of members of $\mathscr{B}$.) Then $$\begin{align*}U\cap V&=\left(\bigcup\mathscr{U}\right)\cap\left(\bigcup\mathscr{V}\right)\\&=\bigcup_{G\in\mathscr{U}}\left(G\cap\bigcup\mathscr{V}\right)\\&=\bigcup_{G\in\mathscr{U}}\bigcup_{H\in\mathscr{V}}(G\cap H)\tag{1}\;,\end{align*}$$ where each of the sets $G\cap H$ is the intersection of two members of $\mathscr{B}$. Each member of $\mathscr{B}$ is an intersection of finitely many members of $\mathscr{S}$; if you intersect two such sets, you still have an intersection of finitely many members of $\mathscr{S}$ and hence a member of $\mathscr{B}$. That is, each of the sets $G\cap H$ appearing in the big union in $(1)$ is a member of $\mathscr{B}$, so by definition their union, $U\cap V$, is a member of $\tau(\mathscr{B})$.

  • The only thing that remains to be checked is that $\varnothing,X\in\tau(\mathscr{B})$. Certainly $\varnothing\subseteq\mathscr{B}$, so $\bigcup\varnothing\in\tau(\mathscr{B})$ by definition, and $\bigcup\varnothing=\varnothing$. In words, $\varnothing$ is the union of the empty subfamily of $\mathscr{B}$. Finally, $\varnothing$ is a finite subfamily of $\mathscr{S}$, and it’s vacuously true that $$\bigcap\varnothing=\{x\in X:x\in S\text{ for all }S\in\varnothing\}=X\;.$$

We now know that $\tau(\mathscr{B})$ is a topology on $X$ with $\mathscr{S}\subseteq\tau(\mathscr{B})$, so by the first part $\tau(\mathscr{S})\subseteq\tau(\mathscr{B})$. On the other hand, if $\tau$ is a topology on $X$ such that $\mathscr{S}\subseteq\tau$, then it must be true that $\mathscr{B}\subseteq\tau$, because $\tau$ is closed under taking finite intersections. And then it must also be true that $\tau(\mathscr{B})\subseteq\tau$, since $\tau$ is closed under taking arbitrary unions. Thus,

$$\tau(\mathscr{S})\subseteq\tau(\mathscr{B})\subseteq\bigcap\left\{\tau:\tau\text{ is a topology on }X\text{ and }\mathscr{S}\subseteq\tau\right\}=\tau(\mathscr{S})\;,$$

and it follows immediately that $\tau(\mathscr{B})=\tau(\mathscr{S})$.

Brian M. Scott
  • 616,228
  • Sorry, that is my fault, I know the answer right when I submitted it, that's why it has been deleted. So how do I show that $\tau(\mathscr{B})=\tau(\mathscr{S})$? Do I suppose that $\mathscr{S}\subset \mathscr{p}(X)$? – Akaichan Mar 12 '13 at 22:52
  • I really do not know why I figured all this out right after I asked it. But I got this, thank you very much. – Akaichan Mar 12 '13 at 22:54
  • @IvordesGreenleaf: Do you mean that you’ve figured out why $\tau(\mathscr{B})=\tau(\mathscr{S})$? Or do you still need a bit of help with that? (And you’re welcome.) – Brian M. Scott Mar 12 '13 at 22:56
  • Yes, I did $\tau(\mathscr{B})=\tau(\bigcap\mathscr{F'})=\bigcup(\bigcap \mathscr{F'})= \bigcap (\bigcup\mathscr{F'})$ since union distributes over intersection $=\bigcap\mathscr{S}=\tau(\mathscr{S})$ – Akaichan Mar 12 '13 at 23:54
  • @IvordesGreenleaf: Whoa! You have some serious difficulties with the notation: $\bigcap\mathscr{S}$ is just a single subset of $X$, not a topology on $X$. Similarly, if $\mathscr{F}'$ is some collection of subsets of $X$, then $\bigcup\left(\bigcap\mathscr{F}'\right)$ is just a subset of $X$. – Brian M. Scott Mar 12 '13 at 23:58
  • whoops, I meant to write just $\mathscr{F}$ – Akaichan Mar 13 '13 at 00:08
  • I have another question, what is the subbase for this topology on $X$? – Akaichan Mar 13 '13 at 00:15
  • @IvordesGreenleaf: Making it $\mathscr{F}$ doesn’t help: what you wrote simply doesn’t make sense. It’ll take me a little while, but let me add that proof to my answer. // This topology has many subbases: $\mathscr{S},\mathscr{B}$, and the topology itself are all subbases for it. – Brian M. Scott Mar 13 '13 at 00:17
2

The topology generated by $S$ is, according to what you say (correctly), the collection $$<S>=\bigcap_{S\subseteq \tau }\tau, $$ where $\tau $ ranges over all topologies that contain $S$. Showing that it is a topology is straightforward (if done correctly).

Now, to show that $<S>$ is the smallest topology containing $S$, first note that it contains $S$ (since each $\tau $ contains $S$). Now, suppose that $\tau '$ is some topology that contains $S$. Then $\tau '$ participates in the intersection defining $<S>$, and thus $<S>\subseteq \tau '$. So, $<S>$ is a topology, it contains $S$, and it is contained in any topology containing $S$. So, $<S>$ is the smallest topology containing $S$, and by definition of subbsase $S$ is a subbase of $<S>$.

Stefan Hamcke
  • 27,733
Ittay Weiss
  • 79,840
  • 7
  • 141
  • 236