I am currently reading Introduction to Topology by Bert Mendelson, and I have some questions regarding the topic on Identification Topology in his book.

Let $(X,\tau)$ and $(Y,\gamma)$ be topological spaces.

Let $f:X\rightarrow Y$ be a continuous function.

Define $\sim_{f}$ to be the relation on $X$ by $x\sim_{f}x'$ if $f(x)=f(x')$ and $X/\sim_{f}$ be the set of equivalence classes.

Let $\pi_{f}:X\rightarrow X/\sim_{f}$ be the map that maps each $x\in X$ onto its equivalence class. And also define $\tau'$ to be the identification topology on $X/\sim_{f}$ determined by $\pi_{f}$.

It is true that for $x,x'\in X:$ $\pi_{f}(x)=\pi_{f}(x')$ iff $f(x)=f(x')$. Thus by the theorem, $f$ induces a continuous map $f^{*}:X/\sim_{f} \rightarrow Y$ such that $f=f^{*}\circ \pi_{f}$ by defining $f^{*}(u)=f(x)$ for $u=\pi_{f}(x)$. And one can also show $f^{*}$ is injective.

Define $f^{*}(\tau')=\{f^{*}(O)\subset Y : O\in\tau'\}$

Now my questions are: is it true that $\gamma\subset f^{*}(\tau')$? Will $f^{*}(\tau')$ form a topology on $Y$? If it would, will $f^{*}(\tau')$ be the finest or weakest topology on $Y$ such that $f$ is continuous?

In the original text, it says "Since $f^{*}$ is continuous, $f^{*-1}(\gamma)\subset \tau'$, or equivalently, since $f^{*}$ is one-one, $\gamma \subset f^{*}(\tau')$."..."Thus the topology $\tau'$ carried over to $Y$ by $f^*$ is the weakest or smallest topology such that f is continuous."(From chapter 3, Section 3: Identification Topologies, Page 103, last paragraph)

The implication that $f^{*}$ injective $\implies\gamma \subset f^{*}(\tau')$ was not obvious to me. And also, if that's the case, why would " $\tau'$ carried over to $Y$ by $f^*$ " (I'm assuming it's referring to $f^{*}(\tau')$) be the smallest topology?


  • 2
    $\begingroup$ Realize that the image of $f^*$ is the same as the image of $f$ . If $f$ is not surjective (and you don't demand that in your question) then $f(X)≠Y$ . This means that no set $O\in\tau'$ can exist with $f^*(O)=Y$ wich contradicts that $f^*(\tau')$ is a topology. $\endgroup$ – drhab Jul 30 '15 at 9:59
  • $\begingroup$ @drhab I see, thanks! So do you think it is a mistake in the book? I mean, maybe the author wanted to show some other concept but end up with having a typo in this paragraph, or do you think I should just let it go and ignore this part of the section? $\endgroup$ – HSea12345n Jul 30 '15 at 13:49
  • $\begingroup$ Was it explicitly said in the book that $f^*(\tau')$ is a topology in this context? I have my doubts about that (and have the book not at my disposal). I suspect this all has to do with writing the continuous $f$ as $f=i\circ s$ with $i$ injective and $s$ surjective (in your question $i=f^*$ and $s=\pi_f$). If $Z$ is domain of $i$ and codomain of $s$ then the question rises: what topologies on $Z$ make $i$ and $s$ both continuous? There is a finest (the quotienttopology=identificationtopology) and a coarsest (in your question the topology $f^{*-1}(\gamma)$). $\endgroup$ – drhab Jul 30 '15 at 14:34

I hope that this will make things more clear to you.

In this answer $i$ corresponds with the $f^*$ in your question and $s$ with the $\pi_f$ in your question.

Let $X$ and $Y$ be topological spaces and let $f:X\rightarrow Y$ be continuous. Looking at $f$ purely as a function we can write it as $f=i\circ s$ where $i$ is injective and $s$ is surjective. As codomain of $s$ and domain of $i$ we can take the set $Z$ that is denoted in your question as $X/\sim_{f}$.

Now we want $Z$ to be equipped with a topology such that $i$ and $s$ are both continuous.

Two extreme candidates are:

1) $\tau_{Z}^{\text{fine}}:=\left\{ O\in\wp\left(Z\right)\mid s^{-1}\left(O\right)\in\tau_{X}\right\} $.

This under the motto: "take every subset of $Z$ in the topology as long as you do not violate the continuity of $s$".

2) $\tau_{Z}^{\text{coarse}}:=\left\{ i^{-1}\left(U\right)\mid U\in\tau_{Y}\right\} $ or shortly $i^{-1}\left(\tau_{Y}\right)$.

This under the motto: "leave every subset of $Z$ out as long as you do not violate the continuity of $i$."

If $U\in\tau_{Y}$ then $s^{-1}\left(i^{-1}\left(U\right)\right)=f^{-1}\left(U\right)\in\tau_{X}$ since $f$ is continuous, showing that $i^{-1}\left(U\right)\in\tau_{Z}^{\text{fine}}$.

So we have $\tau_{Z}^{\text{coarse}}\subseteq\tau_{Z}^{\text{fine}}$.

The maps $i$ and $s$ are both continuous if and only the topology $\tau_{Z}$ on $Z$ satisfies:$$\tau_{Z}^{\text{coarse}}\subseteq\tau_{Z}\subseteq\tau_{Z}^{\text{fine}}$$


This actually gives two different factorizations of the continuous function $f=i\circ s$: $$\tau_Z=\tau_Z^{\text{fine}}\iff s\text{ is a quotientmap}$$ and: $$\tau_Z=\tau_Z^{\text{coarse}}\iff i\text{ is an embedding}$$

  • $\begingroup$ Thank you so much for the clear and insightful explanation! The book didn't explicitly say that $f^{*}(\tau')$ is a topology, even though it used some notation such as $(Y,f^{*}(\tau'))$. but now after reading this, I'm pretty sure it's a mistake on the book. I think the author's original intention was to show the comparison of different topologies on $X/\sim_{f}$ instead of on $Y$. Anyway, Thanks! $\endgroup$ – HSea12345n Jul 30 '15 at 16:13
  • $\begingroup$ Glad to help. Small addendum. $\endgroup$ – drhab Jul 30 '15 at 17:17

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.