0
$\begingroup$

Let $f$ be a non-decreasing Lebesgue-integrable real function on $[a,b]$. I read in Kolmogorov-Fomin's Элементы теории функций и функционального анализа (p. 340 here) that $$\lim_{h\to 0^+}\Bigg(\frac{1}{h}\int_{[b,b+h]}fd\mu-\frac{1}{h}\int_{[a,a+h]}fd\mu\Bigg)=f(b)-f(a^+)$$where I have written $f(a^+)$ for $\lim_{h\to 0^+}f(a+h)$.

I know that for any summable function $f:[a,b]\to\mathbb{R}$ the equality $\frac{d}{dx}\int_{[a,x]}fd\mu=f(x)$ holds for almost all $x\in[a,b]$, but I do not see how it necessarily holds, at least in this particular case where $f$ is non-decreasing, for $x=a$ and $x=b$. Could anybody explain the reason why, in the case where $f$ is non-decreasing, $\lim_{h\to 0^+}(\frac{1}{h}\int_{[b,b+h]}fd\mu-\frac{1}{h}\int_{[a,a+h]}fd\mu)=f(b)-f(a^+)$ holds? I thank you all!!!

$\endgroup$
1
$\begingroup$

The first integral extends outside of the original domain of $f$. Presumably they are defining $f(x) = f(b)$ for all $x > b$. This implies $$ \frac 1h \int_{[b,b+h]} f \, d\mu = f(b)$$ for all $h > 0$.

On the other hand, $\lim_{x \to a^+} f(x)$ exists since $f$ is nondecreasing and equals, by definition, $f(a^+)$. Given $\epsilon > 0$ there exists $\delta > 0$ so that $a < x < a+ \delta$ implies that $|f(x) - f(a^+)| < \epsilon$. In particular, $$ 0 < h < \delta \implies \left| \frac 1h \int_{[a,a+h]} f(x) \, d\mu - f(a^+) \right| < \epsilon$$ which quantifies the limit you are looking for.

$\endgroup$
1

Your Answer

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

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