If we assume that the derivative is continuous the answer is clearly yes. I suspect however that now the answer is no. This question arose when I wanted to investigate for which non-decreasing, right-continuous functions $g: \mathbb{R} \rightarrow \mathbb{R}$ the Lebesgue-Stjeltes measure associated to $g$, $\mu_g$, is given by $\mu_g(A)=\int_A g' d\mu$ for all Borel sets $A$ (1). If such a $g$ is automatically absolutely continuous then the second theorem of calculus for Lebesgue integration would imply that $\int_{[a,b]}g'(x)d\mu(x)=g(b)-g(a)$ and the uniqueness in Caratheodory's theorem would imply (1).
-
Also equivalent to this previous question – Mike Earnest Feb 27 '22 at 15:35
1 Answers
Contrary to your suspicion, the answer is Yes. We appeal to the following characterization (Banach-Zaretsky theorem) of absolutely continuous functions:
$f: [a,b] \to \Bbb R$ is absolutely continuous if and only if it is continuous, is of bounded variation, and has the Luzin N property.$\color{blue}{^1}$
Suppose $f: [a,b]\to \Bbb R$ is monotone, and differentiable. Certainly, $f$ is continuous$\color{blue}{^2}$ and has bounded variation.$\color{blue}{^3}$ So it remains to check if $f$ satisfies the Luzin N property.
Luzin-N Property. A function $f:[a,b]\to\Bbb R$ has the Luzin N property if for all $N\subset [a,b]$ such that $\lambda(N) = 0$, we have $\lambda (f(N)) = 0$. Here, $\lambda$ denotes the Lebesgue measure.
It is known that any differentiable function has the Luzin N property (see this or Lemma $7.25$ of Rudin's Real and Complex Analysis). Therefore, $f$ is absolutely-continuous.
$\color{blue}{1.}$ See this paper on "A new proof for the Banach-Zarecki theorem: A light on integrability and continuity" by Ali Mahdipour-Shirayeh, Homayoon Eshraghi.
$\color{blue}{2.}$ Differentiability implies continuity.
$\color{blue}{3.}$ Monotone functions on closed intervals have bounded variation. For more details, see these notes.
- 14,877
-
1Very nice. I actually encountered this Luzin (N) property long ago on an exam, so cool to see it again. – Jonathan Hole Feb 05 '22 at 14:03
-
1Moisej A. Zaretsky (1903–1930) was a Russian mathematician who died all too young. He proved this result independent of Banach. Most textbooks use the Polish transliteration of his name and you will see, more often, this cited as the Banach–Zarecki theorem. Does anyone have a reference for his paper? Luzin introduced that property in 1915. – B. S. Thomson Feb 05 '22 at 18:32