Prove that $f:(X,d_X)\to (Y,d_Y)$ being continuous at $x\in X$ $\Leftrightarrow$ given any neighbourhood $N$ of $f(x)$, $f^{-1}(N)$ is a nbhd. of $x$

55 Views Asked by At

Context: We have not yet defined open and closed sets (in fact, their definition comes right after this proposition). I'll provide below the exact definitions we are using for this course:

Definition. Let $\left(X,d_X\right)$ and $\left(Y,d_Y\right)$ be metric spaces, and let $f:X\to Y$ be a function. We say $f$ is continuous at $x\in X$ if $\forall \varepsilon>0$, $\exists \delta > 0$ such that if $d_X\left(x,x'\right)<\delta$ then $d_Y\left(f(x),f\left(x'\right)\right)<\varepsilon$.

Definition. Let $(X,d)$ be a metric space, $x\in X$ and $r>0$. We define the open ball $B_{(X,d)}(x,r)$ in $X$ ,centred at $x$, and with radius $r$, in the metric $d$, to be the set \begin{equation*} B_{(X,d)}(x,r) := \{y\in X: d(x,y)<r\}. \end{equation*} The closed ball $B_{(X,d)}[x,r]$ is defined analogously: \begin{equation*} B_{(X,d)}[x,r] := \{y\in X: d(x,y)\leq r\}. \end{equation*}

Definition. A neighbourhood of a point $x\in X$ is a subset $N\subseteq X$ containing $B_r(x)$ for some $r>0$, i.e., $N\supseteq B_r(x)$.

The statement to be proven is

Statement. $f:X\to Y$ is continuous at $x\in X$ if and only if given any neighbourhood $N$ of $f(x)$ then $f^{-1}(N)$ is a neighbourhood of $x$.

My proof. $(\Longrightarrow)$. Suppose $f$ is continuous at $x\in X$ and that $N$ is a neighbourhood of $f(x)$. Then there exists an open ball $B_{\varepsilon}\left(f(x)\right)\subseteq N$. This means $\exists \varepsilon>0$ such that $y\in N$ if $d_Y \left(f(x),y\right)<\varepsilon$. Because $f$ is continuous at $x$, $\exists \delta > 0$ such that if $d_X \left(x,x'\right)<\delta$ then $d_Y \left(f(x),f\left(x'\right)\right)<\varepsilon$. Therefore $f^{-1}(N)$ is a neighbourhood of $x$, since it contains the ball $B_{\delta}(x)$.

$(\Longleftarrow)$. Suppose that given any neighbourhood $N$ of $f(x)$, $f^{-1}(N)$ is a neighbourhood of $x$. Then by the definition of neighbourhood, there exist (arbitrary) balls $B_{\varepsilon}\left(f(x)\right)$ and $B_{\delta}(x)$ in $N$ and $f^{-1}(N)$ respectively. In other words, $\forall \varepsilon>0$, $\exists \delta>0$ such that if $d_X\left(x,x'\right)<\delta$ then $d_Y\left(f(x),f\left(x'\right)\right)<\varepsilon$, which is the definition of continuity.

Question. Is my proof correct? And is there a better (clearer/more concise) way to prove this? Because I'm almost certain that this could be done much more elegantly (and potentially correctly).

Any help would be much appreciated.

1

There are 1 best solutions below

0
On BEST ANSWER

For the first implication, try to show more clearly why the ball $B_{\delta}(x)$ is contained in $f^{-1}(N)$. It can be done as follows

If $y \in B_{\delta}(x)$, then $f(y)\in f(B_{\delta}(x)) \underbrace{\subseteq}_{\text{continuity}}B_{\epsilon}(f(x))\subseteq N \iff y \in f^{-1}(N)$ and thus $B_{\delta}(x)\subseteq f^{-1}(N)$.

Another way of doing the second part is let $x \in X$ and let $\epsilon>0$. Take $N=B(f(x),\epsilon)$ which is clearly a neighborhood of $f(x)$, by hypothesis we conclude that $f^{-1}(N)$ is a neighborhood of $x$.

Thus, $\exists \delta>0$ such that $B_{\delta}(x)\subseteq f^{-1}(N)$. Then for all $y \in X$ with $d(x,y)<\delta$,

$$f(y) \in f(f^{-1}(N))\subseteq N=B(f(x),\epsilon) \iff d(f(x),f(y))<\epsilon$$ Which is the definition of continuity in $x$.