5

Let $q:X\to Y$ be a surjective map, where $(X,\tau_X)$ is a topological space. The quotient topology $\tau_q$ on $Y$ is given as $U\in \tau_q$ iff $q^{-1}(U)\in \tau_X$.

Suppose that there is another topology $\tau_d$ on $Y$ such that for any topological space $Z$ is holds that $f:Y\to Z$ is continuous iff $(f\circ q):X\to Z$ is continuous. I should prove that then $\tau_d = \tau_q$.

My idea is to consider $f_1 = \operatorname{Id}_Y:Y_d\to Y_q$ and $f_2 = \operatorname{Id}_Y:Y_q\to Y_d$ to use the characteristic property. Then I have that since $q:X\to Y_q$ is a quotient map, it is continuous hence $f_2$ is continuous. Here $Y_q = (Y,\tau_q) $ and $Y_d = (Y,\tau_d)$.

But I don't have an idea how to prove that $f_1$ is continuous since I don't know that $q:X\to Y_d$ is continuous. To prove the last fact I consider $f_3 = \operatorname{Id}_Y:Y_d\to Y_d$ which is continuous so is $q = (f_3\circ q):X\to Y_d$. Hence, $q:X\to Y_d$ is continuous so is $f_1$.

Could you please help me to realize if my proof is correct?

SBF
  • 36,041

2 Answers2

4

To avoid any confusion use distinct symbols for the mapping $q$ depending on whether the image is $Y_q$ or $Y_d$. So denote $q'=f_2\circ q:X\to Y_d$.

To show that $f_1$ is continuous use the property of $\tau_d$. We know that the quotient map $q=f_1\circ q':X\to Y_q$ is continuous. Hence $f_1$ is continuous by the backward direction of the property of $\tau_d$.

To show that $f_2$ is continuous first use the forward direction of the property of $\tau_d$ to show that $q'=f_3\circ q'$ is continuous (as you have done). Then noting that $q'=f_2\circ q$ use the characteristic property of the quotient space $Y_q$ to conclude that $f_2$ is continuous.

Finally the continuity of $f_1$ and $f_2$ implies that $\tau_d=\tau_q$.

LostInMath
  • 4,528
  • why does the continuity of $f_1$ and $f_2$ imply that $\tau_d = \tau_q$? (I'm taking a first course in topology right now) – william_grisaitis Feb 03 '22 at 04:20
  • You can prove that two topologies are equivalent if and only if the identity map from one topology to the other, as well as its inverse, are continuous. In this case, $f_1$ is the identity map from $\tau_d$ to $\tau_q$, and $f_2$ is its inverse. – username2384 Feb 11 '22 at 14:03
3

(Your proof is essentially correct, but there might be a clearer way to present it.)

I'd say the clearest way is to separate this into two problems.

(1) There can be at most one topology on $Y$, $\tau$, such that for all spaces $Z$, a function $g:Y\rightarrow Z$ is continuous if and only if $g\circ f:X\rightarrow Z$ is continuous.

This essentially proves that ${\tau}_d$ in your problem is unique if it exists.

(2) ${\tau}_q$ has the property described in (1).

Proof of (1): If ${\tau}$ has this property, then, since $id_\tau:(Y,\tau)\rightarrow(Y,\tau)$ is continuous, $f_\tau:X\rightarrow(Y,\tau)$ is continuous. Now, given $\tau$ and $\tau'$ topologies on $Y$ with this property, we need to show that ${id}_1:(Y,\tau)\rightarrow (Y,\tau')$ and ${id}_2:(Y,\tau')\rightarrow (Y,\tau)$ are continuous. But a function from $(Y,\tau)$ is continuous if any only if the composition with $f_\tau$ is continuous, and we know that $f_\tau \circ {id}_1 = f_{\tau'}$, which we know is continuous. Similarly, ${id}_2$ is continuous, so $\tau=\tau'$.

Proof of (2): If $g:(Y,{\tau}_q) \rightarrow Z$ is continuous, then $V$ is open in $Z$ implies that $g^{-1}(V)\in {\tau}_q$ which implies that $f^{-1}(g^{-1})(V)$ is open in $X$. But that means that $g\circ f$ is continuous.

On the other hand, given $g:Y\rightarrow Z$, if $g \circ f$ is continuous, then for any open $V$ in $Z$, $(g\circ f)^{-1}(V) = f^{-1}(g^{-1}(V))$ is open in $X$. But, by the definition of ${\tau}_q$, that means that $g^{-1}(V)\in {\tau}_q$, so $g:(Y,\tau_q)\rightarrow Z$ is continuous.

So $\tau_q$ has the property from (1).

Note that this proof is a stronger result than the original problem, because the original problem essentially says the only candidate topology for (1) is $\tau_q$. It might be that no topology satisfies (1), in which case, your original problem would still be true.

Thomas Andrews
  • 177,126