# simple integration question

I've tried but I cannot tell whether the following is true or not. Let $f:[0,1]\rightarrow \mathbb{R}$ be a nondecreasing and continuous function. Is it true that I can find a Lebesgue integrable function $h$ such that $$f(x)=f(0)+\int_{0}^{x}h(x)dx$$ such that $f'=h$ almost everywhere?

Any hint on how to proceed is really appreciated it!

-
@Andrew : The whole point is that we can't assume it is. We use the hypothesis to show that it is differentiable almost everywhere. – Patrick Da Silva Jul 21 '12 at 4:56
@Cristian: If the above formula holds, then you must have $f'(x) = h(x)$ at every Lebesgue point of $f$, which is a.e. Since $f'(x) = 0$ a.e., we have that $h$ is essentially zero. But the Cantor function is not constant. Absolute continuity is required. – copper.hat Jul 21 '12 at 6:13
For example, see Rudin's "Real & Complex Analysis" Theorem 7.11 and Section 7.16 for details. – copper.hat Jul 21 '12 at 6:18
thanks @cooper. You should put your comment as an answer to my question. It is the kind of argument I was looking for....ohhhh, and thanks for the reference to rudin's – Cristian Jul 21 '12 at 6:22
@copper.hat: I'm just repeating Cristian's above suggestion to post as an answer because he misspelled your username in the ping. – joriki Jul 21 '12 at 6:33

The Cantor function (call it $f$) is a counterexample. It is monotone, continuous, non-constant and has a zero derivative a.e.
If the above integral formula holds, then you have $f'(x) = h(x)$ at every Lebesgue point of $f$, which is a.e. $x \in [0,1]$. Since $f'(x) = 0$ a.e., we have that $h$ is essentially zero. Since $f$ is not constant, we have a contradiction.