4

Problem somewhat related to this post, and I refer to the same exercise in John Lee's Introduction to Smooth Manifolds.

My issue is trying to understand the relationship between a general subspace topology, and the topology given to an immersed submanifold.

Let $ M$ be a smooth manifold and $S \subseteq M$ be an immersed submanifold of $M.$ Following the definition outlined in Lee, it is equipped with a topology such that $S$ is a topological smooth manifold and the inclusion map $ \iota :S \hookrightarrow M$ is a smooth immersion.

The problem: We want to show that open sets in $S$ with respect to the subspace topology are also open with respect to the given submanifold topology.

My understanding is as follows: For the subspace topology, a set $U \subseteq S$ is open if it can be expressed as $U = S \cap V$ for some open set $V \in M$. Following guidance from the linked post, since the inclusion map $\iota$ is smooth, it is continuous, and hence the preimage preserves open sets. If $V$ is open in $M$ (with respect to $M$'s topology), then $\iota^{-1}(V) = V \cap S$ is open; but here is where I am confused.

My questions: With respect to what topology would $V \cap S$ would it be open in?

Is showing that $U =V \cap S$ is open as preimage of an open set under the inclusion map what is meant by being "open" with respect to the submanifold topology (in particular, for an immersed submanifold)?

If not, what mechanism do we use to prove that a set is open in this topology (for instance, in the subspace topology, we must show there exists an open $V$ in $M$ such that $U = S\cap V.$ How does one do this in the submanifold topology, in order to prove the statement?)

Like how the open sets of the subspace topology can be found by intersecting $S$ with every open set in $V$, that is

$\mathcal{O} = \{U \subseteq S : \text{exists an open set } V \subseteq M \text{ such that } U = V \cap S\}$,

can one write something similar for the submanifold topology?

I would appreciate any clarification on this!

cmperez024
  • 47
  • 7
  • It's the topology that makes $\iota\colon S\to M$ an immersion (makes $S$ an immersed submanifold). Note that $V\cap S$ is always open in the subspace topology of $S$ by definition of subspace topology. – Alex Ortiz May 22 '25 at 18:29
  • I see. But in the case of the proof, is the inclusion map being continuous enough to imply that the set $U = V \cap S$ is also open in $S$ respect to the submanifold topology? That is to say, is preserving openness under the preimage the mathematical version of what we mean by openness in the immersed submanifold topology? – cmperez024 May 22 '25 at 19:00
  • When we say the inclusion is smooth (hence continuous), we mean implicitly that the inclusion is smooth (hence continuous) with respect to the immersed submanifold topology on $S$, which is the only topology on $S$ in context that makes $S$ a smooth immersed submanifold. – Alex Ortiz May 22 '25 at 19:28
  • I understand that much. Sorry for my confusion. I suppose I should rephrase: how exactly does one prove that a set is open in with respect to that topology? As in the original post, proving $U$ is open in $S$ in the subspace topology requires exhibiting the existence of an open set $V \in M$ such that $U = S \cap V.$ Under what mechanism do we show that something is open in the submanifold topology? I edited my original post to better reflect this. – cmperez024 May 22 '25 at 20:36

3 Answers3

4

$S$ is a subset of $M$. Then the set $S$ is endowed with a topology $\tau_S$ and smooth structure, a priori independent from $M$. This topology $\tau_S$ is the submanifold topology.

As $S$ is a subset of $M$ there is the inclusion map $\iota: S \to M$. By definition of immersed submanifold $\iota$ is now a smooth immersion and in particular continuous. Thus $\iota^{-1}(\tau_M) \subseteq \tau_S$, where $\tau_M$ is the topology of $M$.

But the subspace topology $\tau_{S\subseteq M}$ is, by definition, $\iota^{-1}(\tau_M) = \tau_{S\subseteq M}$. Hence $\tau_{S\subseteq M} \subseteq \tau_S$.

psl2Z
  • 5,772
  • 1
    I recently learned that the subspace topology is the coarsest topology for $S$ such that the inclusion map $\iota : S \hookrightarrow M$ is continuous. Seeing that reaffirmed here in more detail is helpful! – cmperez024 May 22 '25 at 22:25
4
  • First remark: let $X$ be a set, $(Y,\tau_Y)$ a topological space and $f:X\to Y$ any function. Then, the collection $f^*(\tau_Y):= \{f^{-1}[V]\,:\, V\in \tau_Y\}$ is a topology on $X$. This is an easy enough exercise for you to verify. In particular, if $X$ is a subset of $Y$ and $f=\iota:X\hookrightarrow Y$ is the canonical inclusion map, then $\iota^*(\tau_Y)$ is called the subset topology on $X$ inherited from $Y$.

  • Second remark: Let $(X,\tau_X), (Y,\tau_Y)$ be topological spaces and $f:X\to Y$ a given function. Then, by definition, we will have that $f$ is continuous if and only if $f^*(\tau_Y)\subseteq \tau_X$. But note that the inclusion could very well be strict. We say that $f$ is a topological embedding if $f$ is injective and $f^*(\tau_Y)=\tau_X$ (this is equivalent to saying the target-restricted map $f:(X,\tau_X)\to (f[X], \tau_{\text{subset},\,\, f[X]\subset Y})$ is a homeomorphism, or equivalently, that it has a continuous inverse).

Now, let $M$ be a smooth manifold and $S\subseteq M$ an immersed submanifold. By definition this sentence means we have $(M,\mathscr{A}_M,\tau_M)$, a set a maximal smooth atlas and a topology on $M$, and likewise $(S,\mathscr{A}_S,\tau_S)$ for $S$, such that the inclusion map $\iota:S\hookrightarrow M$ is a smooth immersion. Note that the choices of $\mathscr{A}_S,\tau_S$ are not uniquely determined by $(M,\mathscr{A}_M,\tau_M)$, meaning that you have to specify this for us before we can proceed with the discussion; see here.

Since $\iota$ is smooth, it is in particular continuous, so by our previous two remarks, we have \begin{align} \tau_{\text{subset},\, S\subseteq M}:=\iota^*(\tau_M)\subseteq \tau_{S}. \end{align} So, the fact that every set which is open in the subset topology is also open relative to the given immersed submanifold topology is pretty much trivially true by definition.


At this stage, I should remark that the concept of an immersed submanifold is unnecessarily confusing. I find it much cleaner to just speak of immersions/ injective immersions.

For example, the preceding discussion could have been phrased as follows: suppose $(X,\mathscr{A}_X,\tau_X), (Y,\mathscr{A}_Y,\tau_Y)$ are smooth manifolds and $f:X\to Y$ is a smooth injective immersion. Then, we have \begin{align} f^*(\tau_Y)\subseteq \tau_X, \end{align} but the inclusion may be strict (if they happen to be equal, then $f$ is said to be a smoth embedding).

peek-a-boo
  • 65,833
3

Im a beginner myself so i could be wrong, $\iota$ is a smooth immersion so $S$ is already equipped with the submanifold topology as the submanifold topology is just the topology that makes $\iota$ into a smooth immersion. Because $\iota$ is continuous, if a set $V\cap S$ is open in the subspace topology, its preimage under $\iota$ is open in the submanifold topology.

cmperez024
  • 47
  • 7
hermit1010
  • 61
  • 3