Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Luzin's theorem states that: let $f:[a,b]\rightarrow R$ be an a.e. finite function, $f$ is measurable iff $\forall \epsilon \geq 0: \exists \phi_\epsilon$ continuous on $[a,b]$ and $\mu\{x: f(x)\neq \phi_\epsilon (x)\} \leq \epsilon $

Why doesn't it imply that a measurable function $f$ equals a.e. to a continuous function $\phi_\epsilon$?

share|cite|improve this question
$\phi_\epsilon$ might be different for every $\epsilon$. How would you use them to pick a single function $\phi$? I think you'll find that your technique doesn't preserve continuity. – Nate Eldredge Oct 8 '12 at 3:44
@NateEldredge I don't necessarily mean to pick a single $\phi$. – Anita Oct 8 '12 at 3:51
Well, each $\phi_\epsilon$ could be different from $f$ on a set of positive measure (but at most $\epsilon$). I guess I don't understand why you think Luzin's theorem could possibly imply that $f$ is a.e. equal to some continuous function. And of course, you must know of counterexamples. – Nate Eldredge Oct 8 '12 at 3:57
Anna: In addition to the answer Seirios gave, you might find some things of interest in my essay on Luzin's Theorem. – Dave L. Renfro Oct 8 '12 at 20:17

You can illustrate it with a simple example :

Let $f : [0,1] \to [0,1]$ such that $f(x)= 0$ if $x \leq 1/2$ and $f(x)=1$ otherwise. Then, let $f_{\epsilon} : [0,1] \to [0,1]$ be such that $f(x)=0$ if $0 \leq x \leq \frac{1}{2}- \epsilon$, $f(x)= 1$ if $\frac{1}{2}+ \epsilon \leq x \leq 1$ and $f_{\epsilon}$ affine on $[1/2- \epsilon, 1/2+ \epsilon]$.

So $f_{\epsilon}$ is continuous and $\mu \{x : f(x) \neq f_{\epsilon}(x) \}=2 \epsilon$.

However, there is no continuous function $\phi$ such that $f=\phi$ almost everywhere. Indeed, if it was the case, there would be sequences $x_n < 1/2$ and $y_n > 1/2$ converging to $1/2$ and such that $f(x_n)=\phi(x_n)$, $\phi(y_n)=f(y_n)$ (otherwise $\{ x : f(x) \neq \phi(x) \}$ would contain an interval); for $n \to + \infty$, you find $0= \phi(1/2)=1$.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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