If $f = g$ a.e. in $E \in \mathfrak{M}$ (the Lebesgue measurable sets) and $f$ is continuous a.e. in $E$, is $g$ continuous a.e. in $E$?

I think this is true.

My “proof”:

Let us denote $D_1 = \{ x \in E: f(x) \text{ discontinuous}\}$, $m(D_1) = 0$ and $D_2 = \{ x \in E: f(x) \neq g(x)\}$, $m(D_2) = 0$.

Define $D_3 = \{ x \in E: g(x) \text{ discontinuous}\}$.

If $f$ is identically $g$, then it is clear that the result follows as $D_3 = D_1$.

Otherwise, we have $D_3 \subseteq D_1 \cup D_2$, and so $m^*(D_3) \leq m^*(D_1 \cup D_2) \leq m^*(D_1) + m^*(D_2) = 0$.

So, $m(D_3) = 0$ and hence $g$ is continuous almost everywhere.

Does this proof work?


Edit: for clarity, $m$ denotes the Lebesgue measure and $m^*$ the Lebesgue outer measure.

  • 2
    $\begingroup$ How do you justify $D_3 \subseteq D_1 \cup D_2$ ? $\endgroup$ – Yves Daoust Dec 12 '18 at 20:46
  • $\begingroup$ @YvesDaoust My reasoning was that if $f(x) = g(x)$ on $E\setminus D_1$ then since $f(x)$ is continuous there, $g(x)$ ought to be as well? $\endgroup$ – TuringTester69 Dec 12 '18 at 20:47
  • $\begingroup$ Continuity of $f$ in $E\setminus D_1$ does not imply continuity of $g$, because of $D_2$. $\endgroup$ – Yves Daoust Dec 12 '18 at 20:51
  • 4
    $\begingroup$ Consider the case where $f$ is identically 0, and $g$ is the characteristic function of the rationals. $\endgroup$ – Andrés E. Caicedo Dec 12 '18 at 20:52
  • $\begingroup$ @Andrés E. Caicedo, you're absolutely right I was just about to fix my mistake with that example. That should be the answer. $\endgroup$ – James Yang Dec 12 '18 at 20:54

The claim is not true. An easy counterexample is obtained by letting $f$ be identically 0 and $g$ be the characteristic function of the rationals. We have $f=g$ a.e., $f$ is everywhere continuous and $g$ is nowhere continuous.

(Clearly, the restriction of $g$ to the irrationals is continuous (being constant), but this is not enough to ensure that $g$ is continuous on the irrationals.)


$f$ continuous on $E\setminus D_1$ implies $f$ continuous on $E\setminus(D_1\cup D_2)$. And of course, $g$ continuous on $E\setminus(D_1\cup D_2)$. Then the measure of $D_3$ is at most that of $D_1\cup D_2$.

  • 1
    $\begingroup$ Not quite. The restriction of $g$ to a set $F$ may be continuous without $g$ itself being continuous on $F$. $\endgroup$ – Andrés E. Caicedo Dec 12 '18 at 20:57
  • 1
    $\begingroup$ @AndrésE.Caicedo: what difference do you make between "restricted to $F$" and "on $F$" ?? $\endgroup$ – Yves Daoust Dec 12 '18 at 20:59
  • $\begingroup$ The domain of $g$ restricted to $F$ is $F$. The domain of $g$ may be larger, and points not in $F$ may affect whether $g$ is continuous at some point in $F$. See my answer for an example. There, with $F$ the irrationals, $g|F$ is continuous because it is constant, but $g$, defined everywhere, is not continuous at any point. $\endgroup$ – Andrés E. Caicedo Dec 12 '18 at 21:02

No your proof does not work because the fact that $f(x)=g(x)$ on $E\setminus D_1$ implies $g(x)$ is continuous from the continuity of $f(x)$ tells you that:

$D_2\cap (E \setminus D_1) \subseteq (E \setminus D_3)$

which means:

$D_3 \subseteq (E \setminus D_2) \cup D_1$


Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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