Prob. 8, Sec. 21, in Munkres' TOPOLOGY, 2nd ed: If $f_n$ converges uniformly to $f$ and if $x_n$ converges to $x$, ...

678 Views Asked by At

Here is Prob. 8, Sec. 21, in the book Topology by James R. Munkres, 2nd edition:

Let $X$ be a topological space and let $Y$ be a metric space. Let $f_n \colon X \to Y$ be a sequence of continuous functions. Let $x_n$ be a sequence of points of $X$ converging to $x$. Show that if the sequence $\left( f_n \right)$ converges uniformly to $f$, then $\left( \ f_n \left( x_n \right) \ \right)$ converges to $f(x)$.

My Attempt:

As $f_n \colon X \to Y$ is a sequence of continuous functions converging uniformly to the function $f \colon X \to Y$, so $f$ is also continuous, by Theorem 21.6 in Munkres.

Let $\varepsilon > 0$ be given. Let $d$ be the metric for $Y$.

Now as the open ball $$ B_d \left( \ f(x) , \frac{\varepsilon}{2} \right) \colon= \left\{ \ y \in Y \ \colon \ d \left( y, f(x) \right) < \frac{\varepsilon}{2} \ \right\} $$ is an open set in $Y$, as $f(x) \in B_d \left( \ f(x) , \frac{\varepsilon}{2} \right) $, and as the function $f \colon X \to Y$ is continuous, so the inverse image $$ f^{-1} \left( \ B_d \left( \ f(x) \ , \ \frac{\varepsilon}{2} \right) \ \right) \colon= \left\{ \ u \in X \ \colon \ f(u) \in B_d \left( \ f(x) \ , \ \frac{\varepsilon}{2} \right) \ \right\} = \left\{ \ u \in X \ \colon \ d \left( \ f(u) \ ,\ f(x) \ \right) < \frac{\varepsilon}{2} \ \right\} $$ is an open set in $X$ and also $x \in f^{-1} \left( \ B_d \left( \ f(x) \ , \ \frac{\varepsilon}{2} \right) \ \right)$.

Let $U \colon= f^{-1} \left( \ B_d \left( \ f(x) \ , \ \frac{\varepsilon}{2} \right) \ \right)$.

Then $U$ is an open set in $X$, $x \in U$, and also $$ f ( U ) = f \left( \ f^{-1} \left( \ B_d \left( \ f(x) \ , \ \frac{\varepsilon}{2} \right) \ \right) \ \right) \subset B_d \left( \ f(x) \ , \ \frac{\varepsilon}{2} \right), $$ by Prob. 1 (b) in the Exercises following Sec. 2 in Munkres; that is, $$ d \left( \ f(u), f(x) \ \right) < \frac{\varepsilon}{2} \tag{1} $$ for all points $u \in U$.

Now as $\left( x_n \right)_{n \in \mathbb{N} }$ is a sequence in the topological space $X$ converging to the point $x \in X$ and as $U$ is an open set in $X$ such that $x \in U$, so there is a natural number $M$ such that $x_n \in U$ for all natural numbers $n > M$. Therefore (1) implies that $$ d \left( \ f \left( x_n \right) \ , \ f(x) \ \right) < \frac{\varepsilon}{2} \tag{2} $$ for any natural number $n > M$.

Moreover, as the sequence $f_n$ of functions converges uniformly to $f$ on $X$, so there exists a natural number $N$ such that $$ d \left( f_n(t), f(t) \right) < \frac{\varepsilon}{2} $$ for all natural numbers $n > N$ and for all $t \in X$. Therefore $$ d \left( f_n(t), f(t) \right) < \frac{\varepsilon}{2} $$ for all natural numbers $n > N$ and for all $t \in U$. And, as $x_n \in U$ for all natural numbers $n > M$, so we can conclude that $$ d \left( \ f_n \left(x_n \right) \ , \ f \left( x_n \right) \ \right) < \frac{\varepsilon}{2} \tag{3} $$ for all natural numbers $n$ such that $n > M$ and $n > N$.

So for any natural number $n > \max \{ \ M \ , \ N \ \}$, we see that $$ \begin{align} d \left( \ f_n \left( x_n \right) \ , \ f(x) \ \right) &\leq d \left( \ f_n \left( x_n \right) \ , \ f \left( x_n \right) \ \right) + d \left( \ f \left( x_n \right) \ , \ f(x) \ \right) \\ &< \frac{\varepsilon}{2} + \frac{\varepsilon}{2} \\ & \qquad \mbox{ [ using (2) and (3) above; we note that } \\ & \ \ \ \qquad \mbox{ since $n > \max\{ \ M \ , \ N \ \}$, therefore $n > M$ and $n > N$ ] } \\ &= \varepsilon. \end{align} $$

Thus we have shown that, corresponding to every real number $\varepsilon > 0$, there exists a natural number $K \colon= \max \{ M , N \}$ such that $$ d \left( \ f_n \left( x_n \right) \ , \ f(x) \ \right) < \varepsilon $$ for any natural number $n > K$.

Hence the sequence $\left( \ f_n \left( x_n \right) \ \right)_{n \in \mathbb{N} } $ converges in the metric sapce $(Y, d)$ to the point $f(x) \in Y$, as required.

Is this proof correct? If so, then is each and every step in my presentation clear enough too? If not, then where lies the problem?

1

There are 1 best solutions below

3
On

The whole essence of the proof is: for all $n$:

$$d(f_n(x_n), f(x)) \le d(f_n(x_n),f(x_n)) + d(f(x_n), f(x))$$

And both terms on the right can be made as small as we like: pick $n$ large enough for the uniform convergence definition for the first term, and $n$ also large enough from the fact that $f(x_n) \to f(x)$ by continuity of $f$.

A lot of your write-up is superfluous or repetitive, IMHO. Just convey the main idea.