Let $f$ be a bounded function on a compact interval $J$, and let $I(c,r)$ denote the open interval centered at $c$ of radius $r>0$. Let $osc(f,c,r)=\sup|f(x)-f(y)|$, where the supremum is taken over all $x,y\in J\cap I(c,r)$, and define the oscillation of $f$ at $c$ by $osc(f,c)=\underset{r\rightarrow 0}{\lim}osc(f,c,r)$. Clearly, $f$ is continuous at $c\in J$ if and only if $osc(f,c)=0$.

(a) For every $\epsilon>0$, the set of points $c$ in J such that $osc(f,c)\geq \epsilon$ is compact.

(b) If the set of discontinuities of $f$ has measure $0$, then $f$ is Riemann integrable.

I've already proved part (a) so this is my attempt at part (b)

My proof attempt:

Proof. Fix $\epsilon>0$. Define \begin{equation*} D_\epsilon=\{c\in J: osc(f,c)>\epsilon\} \end{equation*} Then $D_\epsilon$ is the set of discontinuities of $f$. By hypothesis, $D_\epsilon$ is measure zero. We will prove that $f$ is Riemann integrable by showing that there exists a partition $P$ such that $|U(P,f)-L(P,f)|<\epsilon$, where $U(P,f)$ and $L(P,f)$ are the upper and lower sums of $f$ over partition $P$, respectively.

By Observation 3 of the Exterior Measure in Stein, there exists an open set $\mathcal{O}$ such that $D_\epsilon\subset \mathcal{O}$ and $m(\mathcal{O})\leq \epsilon.$ By theorem 1.3, $\mathcal{O}$ can be written as a disjoint countable union of open intervals $(I_j)_j$. By Part (a), $D_\epsilon$ is compact. Since $(I_j)_j$ covers $D_\epsilon$, we need only a finite sub-cover to cover $D_\epsilon$, say $(I_j)^{k}_{1},$ and denote the collection as $\mathcal{F}$. Then \begin{equation*} m(D_\epsilon)\leq m(\underset{I\in \mathcal{F}}{\bigcup}I)\leq m(\mathcal{O})\leq \epsilon \end{equation*} Hence, we now have a finite cover of $D_\epsilon$ whose "total length" is $\leq \epsilon$.

Since $J$ is closed and $I\in \mathcal{F}$ is open, it follows that $J\setminus \underset{I\in \mathcal{F}}{\bigcup}I$ is closed. Since $J\setminus \underset{I\in \mathcal{F}}{\bigcup}I\subset J$, it is also compact. Any continuous $f$ over a compact interval is uniformly continuous. Therefore, there exists $\delta>0$ such that for all $x,y\in J\setminus \underset{I\in \mathcal{F}}{\bigcup}I$ satisfying \begin{equation*} |x-y|<\delta \implies |f(x)-f(y)|<\epsilon. \end{equation*} Let $P_1$ be a partition of $J$ such that every pair of consecutive points is within $\delta$ of each other, e.g. one way of constructing such a set is to partition $J$ into $\frac{|J|}{2^n}$ length intervals for large enough $n\in \mathbb{N}$. Now define $\hat{P}_1$ to be a partition of $J$ such that \begin{equation*} \hat{P}_1=P_1\setminus \{x\in P_1: x\in \underset{I\in \mathcal{F}}{\bigcup}I\}. \end{equation*} Next, define $P_2$ to be a partition of $J$ such that $P_2$ is the set of the end points of $I_1, I_2,...,I_k$, and $J$. Finally, let $P$ be the common refinement of $P_1$ and $P_2$, i.e. $P=P_1\cup P_2$. We will now show that $P$ is the desired partition of $J$.

Enumerate the points of $P$ as $x_1,x_2,...,x_N$. Denote $\Delta x_i=|x_{i+1}-x_i|$ for $i=1,...,N-1$. Denote as well the interval $[x_{i+1},x_i]$ as $\nabla x_i.$ If two consecutive points in $P$ are also in $P_2$, denote the length instead by $\Delta y_i$. Moreover, for each $\Delta x_i$, let $m_i$ and $M_i$ be the infimum and supremum of $f$ over $\nabla x_i$, respectively. Define \begin{equation} C:=2\cdot \sup\{|f(x)|:x\in J\}<\infty \quad \text{ Since $f$ is bounded.} \end{equation} Finally, observe \begin{align*} |U(P,f)-L(P,f)|&=|\underset{x_i\in P}{\sum} M_i\Delta x_i- \underset{x_i\in P}{\sum} m_i\Delta x_i| \\ &=|\underset{x_i\in P}{\sum} (M_i-m_i)\Delta x_i| \\ &=|\underset{x_i\in P_1}{\sum} (M_i-m_i)\Delta x_i + \underset{x_i\in P_2}{\sum} (M_i-m_i)\Delta y_i| \\ &\leq \underset{x_i\in P_1}{\sum} |(M_i-m_i)|\Delta x_i + \underset{x_i\in P_2}{\sum} |(M_i-m_i)|\Delta y_i \\ &\leq \epsilon|J| + C\epsilon \\ &\text{ By Uniform Continuity of $f$ over $J\setminus \underset{I\in \mathcal{F}}{\bigcup}I$} \\ &=\epsilon(|J|+C) \\ \end{align*} As niether $|J|$ nor $C$ depend on $\epsilon$, this completes our proof.

The definition of the Riemann Integrable I use is from Rudin's PMA. Any corrections of the proof or comments on style are highly appreciated.

I recognize the proof is really long and I appreciate the time of anybody who parses through it.


Your Answer

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

Browse other questions tagged or ask your own question.