Continuity and Sequences
\[\newcommand{\ds}{\displaystyle} \newcommand{\curlies}[1]{\left\lbrace #1 \right\rbrace} \newcommand{\T}{\mathcal T} \newcommand{\Ext}{\text{Ext}} \newcommand{\B}{\mathcal B}\]Continuity
Let $X$ and $Y$ be topological spaces, and $f: X \to Y$. We say $f$ is continuous if for every open set $V \subseteq Y$, $f^{-1}(V) \subseteq U$ is open.
Why this definition?
- in short, because it wokrs without additional assumptions
- Related definition: Open functions
- let $X$ and $Y$ be topological spaces, and $f: X \to Y$. $f$ is open if it preserves openness, aka for all open sets $U \subseteq X$, $f(U) \subseteq Y$ is open.
- (similarly, a closed function preserves closedness)
- This is not really what we want for continuity! It doesn’t necessarily work out.
$f$ is continuous iff for every closed $C$, $f^{-1}(C)$ is also closed
($\Rightarrow$) Suppose $f$ is continuous, $C \subseteq Y$ is closed, then $C^C$ is open. Then $f^{-1}(C^C)$ is open. $f^{-1}(C^C)^C$ is closed, and $f^{-1}(C^C)^C = f^{-1}(C)$, so $f^{-1}(C)$ is closed.
($\Leftarrow$) Suppose for every closed set $C \subseteq Y$, $f^{-1}(C)$ is closed. Then we can do the previous proof in reverse to find that $f$ is continuous.
Continuity and bases
If $f: X \to Y$ is a function and $\B$ is a basis for $Y$, then $f$ is continuous iff $f^{-1}(B)$ is open for every $B \in \B$.
($\Rightarrow$) Suppose $f$ is continuous. Then for each $B \in \B$, $B$ is open so $f^{-1}(B)$ is open.
($\Leftarrow$) Suppose for every $B \in \B$, $f^{-1}(B)$ is open. Suppose $U \subseteq Y$ is open, then $\ds U = \bigcup_{i \in I} B_i \text{ where } B_i \in \B$. Then,
\[f^{-1}(U) = f^{-1}\left(\bigcup_{i \in I} B_i\right) = \bigcup_{i \in I} f^{-1}(B)\]Thus, $f^{-1}(U)$ is the arbitrary union of open sets, so it is open. Since $U$ is arbitrary, this is true for every open $U$.
Example: metric spaces
Let $(X, d_X)$ and $(Y, d_Y)$ be metric spaces with their metric topologies, and $f: X \to Y$ be a map between them.
Using the proposition above, we know that $f$ is continuous iff the preimage of every open ball is open.
To specify a ball, we need to present a centre $y \in Y$ and a radius $\epsilon \in \R^+$
so $f$ is continuous iff for every ball $B_\epsilon(y) \subseteq Y$, $f^{-1}(B_\epsilon(y)) \subseteq X$ is open. $f^{-1}(B_\epsilon(y))$ is open if it is the union of open balls in $X$, aka if there is an open ball around every point in $f^{-1}(B_\epsilon(y))$.
Thus, $f$ is continuous iff for every $y \in Y$, $\epsilon \in \R^+$, and every point $x \in f^{-1}(B_\epsilon(y))$, there is a $\delta \in \R^+$ so that $B_\delta(x) \subseteq f^{-1}(B_\epsilon(y))$, aka $f(B_\delta(x)) \subseteq B_\epsilon(y)$.
Sequences
Limit points
Let $X$ be a topological space and $A \subseteq X$ be any subset. $p \in X$ is an accumulation point or limit point if every neighbourhood of $p$ has a point in $A$ different from $p$.
Limits
Let $X$ be a topological space and $x_1, x_2, x_3, …$ be a sequence of points in $X$. $x \in X$ is a limit of the sequence iff for every neighbourhood $U$ of $x$ there exists an $N \in \N$ such that if $n \geq N$, then $x_n \in U$.
Note that $x$ is a limit, not necessarily the limit of the sequence, because a sequence may have multiple limits in some topological spaces.
Unique convergence in metric spaces
Let $(X, d)$ be a metric space, and $x_1, x_2, x_3, …$ be a sequence with $x_i \to x$ and $x_i \to y$, then $x = y$.
Proof. Suppose $x \neq y$, then if $r = \vert x - y \vert / 2 \neq 0$, then $B_r(x)$ and $B_r(y)$ are disjoint.
If $x_i \to x$, then since $x \in B_r(x)$, there exists $N \in \N$ such that if $n \geq N$, then $x_n \in B_r(x)$, so $x_n \notin B_r(y)$.
Then it is clear that we cannot say that an infinite tail of the sequence can be contained in $B_r(y)$, so $x_i \not \to y$, which is a contradiction. Thus, $x = y$.
Note that this proof hinges on the ability to find disjoint open balls containing $x$ and $y$!
Hausdorff spaces
A topological space $X$ is Hausdorff if for every $x, y \in X$ where $x \neq y$, we can find open sets $U, V \subseteq X$ where $x \in U$, $u \in V$, and $U \cap V = \emptyset$.
Limits are unique in Hausdorff spaces! This can be shown by slightly modifying the proof above for metric spaces.
Even though non-Hausdorff spaces seem pathological, they arise naturally in some areas (e.g. algebraic geometry).
In Hausdorff spaces, singleton sets (sets like $\curlies{p}$) are closed
If $X$ is Hausdorff, every $q \neq p$ has a neighbourhood disjoint of $p$, so $\curlies{p}^C$, the union of all of these, is open.
Limits in first countable spaces
Let $X$ be a first countable topological space, let $A \subseteq X$ be any subset and $x \in X$.
- $x \in \overline A$ iff $x$ is a limit of a sequence of points in $A$
- $x \in A^O$ iff every sequence in $X$ converging to $x$ is eventually in $A$
- $A$ is closed in $X$ iff $A$ contains every limit of every convergent sequence of points in $A$
- $A$ is open in $X$ iff every sequence in $X$ converging to a point of $A$ is eventually in $A$
Proof of 1:
($\Rightarrow$) We know $\overline A = A \cup \partial A$, so if $x \in \overline A$ then $x \in A$ or $x \in \partial A$.
Suppose $x \in A$, then $x$ is the limit of the sequence $x, x, x, …$
Suppose $x \in \partial A$ but $x \notin A$, then since $X$ is first countable, there exists a nested countable basis at $x$: $U_1 \supseteq U_2 \supseteq U_3 \supseteq …$ where each $U_i$ is open and contains $x$. Since $x \in \partial A$, for each $n \in \N$, $U_n \cap A \neq \emptyset$ , so we can construct a sequence $\curlies{x_n}$ by choosing each $x_n \in U_n \cap A$. Clearly this sequence consists of points in $A$ and converges to $x$.
($\Leftarrow$) Suppose $x \notin \overline A$, then $x \in \Ext(A)$, and $x_1, x_2, x_3, …$ is a sequence of points in $A$.
$\Ext(A)$ is open and does not intersect $A$, so there is an open set $U \subseteq \Ext(A)$ where $x \in U$. For this $U$, there is no subsequence of $\lbrace x_n \rbrace$ that is eventually contained in $U$, so $\lbrace x_n \rbrace$ does not converge to $x$.
Note that this direction did not use first countability! So we know that limits of sequences in $A$ are always trapped in $\overline A$ in every topological space. However, the opposite direction is not true in general, and is pretty much a characterization of first countable spaces.
Proof of 3:
($\Rightarrow$) Suppose that $A$ is closed. If $x_n \to x$ with each $x_n \in A$, then by statement 1, $x$ is a limit of point of $A$
…
($\Leftarrow$) $A \subseteq \overline A$ so we only need to show that $\overline A \subseteq A$.
…
Continuity and sequence convergence
Let $X$ and $Y$ be topological spaces with $X$ first countable and $f: X \to Y$. Then $f$ is continuous iff $x_n \to x$ implies that $f(x_n) \to f(x)$.
Proof.
($\Rightarrow$) Suppose $x_n \to x$ implies $f(x_n) \to f(x)$.
Let $C \subseteq Y$ be a closed set. We will use part 3 of the previous proof.
Let $a_1, a_2, … \in f^{-1}(C)$ with $a_i \to a$. By hypothesis, $f(a_i) \to f(a)$. Moreover, $f(a_i) \in C$, so $f(a) \in \overline C = C$ since $C$ is closed. Thus, $a \in f^{-1}(C)$.
By part 3 the previous proof, we know that $f^{-1}(C)$ is closed. Therefore, $f$ is continuous.
Homeomorphisms
Let $X$ and $Y$ be topological spaces, and $f: X \to Y$ a bijection between them. Then $f$ is a homeomorphism if both $f$ and $f^{-1}$ are continuous.
- There exist continuous bijections whose inverses are not continuous!
- e.g. if $f : (X, \T_{discrete}) \to (X, \T_{trivial})$ is the identity map, then $f$ is continuous but its inverse is not!
$X$ and $Y$ are homeomorphic, written as $X \cong Y$, if there exists a homeomorphism between them.
Covers
Let $X$ be a topological space and $A = \lbrace A_\alpha \rbrace_{\alpha \in I}$ be a collection of subsets of $X$. Then $A$ is a cover of $X$ if
\[X = \bigcup_{\alpha \in I} A_\alpha\]- If $A_\alpha$ is open for each $\alpha \in I$, then $A$ is an open cover. We similarly define a closed cover. Bases are an example of open covers.
- We care about covers because they sometimes help us to examine local properties.
- Given a cover $A = \lbrace A_\alpha \rbrace_{\alpha \in I}$ of $X$ and a subset $J \subseteq I$, $B = \lbrace A_\beta \rbrace_{\beta \in J}$ is a subcover if $B$ is also a cover of $X$.
- Given two covers $A = \lbrace A_\alpha \rbrace_{\alpha \in I}$ and $B = \lbrace B_\beta \rbrace_{\beta \in J}$, $A$ is a refinement of $B$ if for all $\alpha \in I$, there exists $\beta \in J$ such that $A_\alpha \subseteq B_\beta$.
- Notice that subcovers are refinements.