Dini's Theorem states that

Given a sequence of real-valued continuous functions $(f_n)$ on a compact set $E\subseteq \mathbb{R},$ if $(f_n)$ decreases to a continuous function $f$ pointwise on $E$, then $(f_n)$ converges to $f$ uniformly on $E$.

Egoroff's Theorem states that

Given a measure space $(E,\mathcal{A},\mu)$ where $E \subseteq \mathbb{R}$ has finite measure. $\mathcal{A}$ is an $\sigma$-algebra and $\mu$ is a measure on $E.$ If a sequence of measurable functions $(f_n)$ converges pointwise to $f$ almost everywhere, then for every $\varepsilon>0,$ there exists a measurable subset $F\subseteq E$ with $\mu(E\setminus F) <\varepsilon$ such that $(f_n)$ converges uniformly to $f$ on $F.$

Question: Do there exist theorems, other than Dini and Egoroff, which give sufficient conditions for pointwise convergence to be uniform convergence?

I would like to see techniques involved in proving those theorems. Dini and Egoroff used similar technique, which is to define a set containing elements such that $f_n$ and $f$ are 'closed to each other'. If possible, I would like to know other technique to show pointwise convergence implies uniform convergence.


Here's one that I rather like that uses equicontinuity and compactness as the engine for moving from pointwise to uniform convergence. For the sake of clarity, let's recall what equicontinuity means.

Definition: Let $X$ and $Y$ be metric spaces. A set $E$ of continuous functions from $X$ to $Y$ (i.e. $E \subseteq C(X,Y)$ ) is equicontinuous if for every $x \in X$ and every $\varepsilon>0$ there exists $\delta >0$ such that $$ y \in X \text{ and } d_X(x,y) < \delta \Rightarrow d_Y(f(x),f(y)) < \varepsilon \text{ for all } f \in E. $$

With this definition in hand, we can state the result.

Theorem: Let $X$ and $Y$ be metric spaces with $X$ compact. Suppose that for $n \in \mathbb{N}$ the function $f_n: X \to Y$ is continuous and that $\{f_n | n \in \mathbb{N}\}$ is equicontinuous. Let $f : X \to Y$ and suppose that $f_n \to f$ pointwise as $n \to \infty$. Then $f_n \to f$ uniformly as $n \to \infty$. In particular, this means that $f$ is continuous.


Let $\varepsilon >0$. By the equicontinuity property, for each $x \in X$ we can find $\delta_x >0$ such that $$ y \in B_X(x,\delta_x) \Rightarrow d_Y(f_n(x),f_n(y)) < \frac{\varepsilon}{4} \text{ for all } n \in \mathbb{N}. $$ The collection of open balls $\{B_X(x,\delta_x)\}_{x \in X}$ is an open cover of $X$. The compactness of $X$ allows us to find a finite subcover, namely $x_1,\dotsc,x_k \in X$ such that $X = \bigcup_{i=1}^k B_X(x_i,\delta_{x_i})$. Due to the pointwise convergence $f_n \to f$ (or really the pointwise Cauchy property), for each $i=1,\dotsc,k$ we can choose $N_i \in \mathbb{N}$ such that $$ m,n \ge N_i \Rightarrow d_Y(f_n(x_i),f_m(x_i)) < \frac{\varepsilon}{4}. $$

Set $N = \max \{N_1,\dotsc,N_m\} \in \mathbb{N}$. For each $x \in X$ there exists $i\in \{1,\dotsc,k\}$ such that $x \in B(x_i,\delta_{x_i})$. Then for $m \ge n \ge N$ we have the estimate $$ d_Y(f_m(x),f_n(x)) \le d_Y(f_m(x),f_m(x_i)) + d_Y(f_m(x_i),f_n(x_i)) + d_Y(f_n(x_i),f_n(x)) \\ < \frac{\varepsilon}{4} + \frac{\varepsilon}{4} + \frac{\varepsilon}{4} = \frac{3\varepsilon}{4}. $$ Hence for $n \ge N$, $$ d_Y(f(x),f_n(x)) = \lim_{m \to \infty} d_Y(f_m(x),f_n(x)) \le \frac{3\varepsilon}{4}. $$ This holds for all $x \in X$, so $$ n \ge N \Rightarrow \sup_{x \in X} d_Y(f(x),f_n(x)) \le \frac{3\varepsilon}{4} < \varepsilon, $$ and we deduce that $f_n \to f$ uniformly as $n \to \infty$.

  • $\begingroup$ Oops, how can i forgot compactness and equicontonuity. $\endgroup$ – Idonknow Nov 6 '17 at 23:34
  • $\begingroup$ So your equicontinuity refers to pointwise equicontinuity? $\endgroup$ – Idonknow Nov 7 '17 at 6:00
  • $\begingroup$ Yes, this is really pointwise equicontinuity, as opposed to uniform equicontinuity. For this result only the pointwise version is needed. $\endgroup$ – Glitch Nov 7 '17 at 11:55
  • $\begingroup$ Can this be considered as a corollary to the generalized Arzela-Ascoli theorem? $\endgroup$ – Hashimoto Aug 2 '18 at 20:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.