4

Consider a function $f\in L^1([a,b])$ and define $F(x):=\int_a^x f(y)dy$. I should prove that $V_a^bF=||f||_{L^1([a,b])}$.

My attempt was to proceed by approximation with a test function $\phi$, but I’m not getting it. Can someone please help me, please?

Frederik vom Ende
  • 5,187
  • 1
  • 11
  • 39

2 Answers2

3

Consider a partition of $[a,b]$ which splits $F(x)$ into positive and negative parts (continuity of F). Taking limits, this should give you your lower bound. The upper bound is an exercise in using the triangle inequality.

fGDu94
  • 4,072
  • About the part of the lower bound, I’m not understanding why I can take a partition who splits $F$ into its positive and negative part. I mean, there could be infinite point of $[a,b]$ were $F$ changes its sign –  Jun 01 '24 at 06:47
  • Yes but you can take limits – fGDu94 Jun 01 '24 at 14:11
1

Take any partion $\Pi = \{a = t_0 \leq \ldots \leq t_n = b\}$. Then \begin{align*} V_a^b(F, \Pi) = \sum_{i=1}^n|F(t_i) - F(t_{i-1})| &= \sum_{i=1}^n\Big|\int_{t_{i-1}}^{t_i}f(y)dy \Big| \\ &\overset{*}{\leq} \sum_{i=1}^n\int_{t_{i-1}}^{t_i}|f(y)|dy \\ &= \int_a^b|f(y)|dy \\ &= ||f||_{L^1([a,b])}, \end{align*} where we used Jensen's inequality at the $*$-step. Hence, $V_a^b(F) \leq ||f||_{L^1([a,b])}$. For the lower bound, you can use the fundamental theorem of calculus for Lebesgue integrals. That is, $F$ is differentiable on a set $N$ of Lebesgue measure $1$, and for $x \in N$, $F'(x) = f(x)$. For simplicity, we will assume $F$ to be everywhere differentiable, and $f$ to be Riemann integrable. (You fill in the rest of the details.) Now, take any sequence of partitions $\Pi^n = \{a=t_0^n \leq \ldots \leq t_{k_n}^n = b\}$ such that $\max_{1\leq i \leq k_n^n} |t_{i}^n-t_{i-1}^n| \to 0$ as $n \to \infty$. Then, \begin{align*} V_a^b(F, \Pi^n) = \sum_{i=1}^{k_n}|F(t_i^n) - F(t_{i-1}^n)| &= \sum_{i=1}^{k_n}\frac{|F(t_i^n) - F(t_{i-1}^n)|}{(t_i^n-t_{i-1}^n)}(t_i^n-t_{i-1}^n)\\ &= \sum_{i=1}^{k_n}|f(c_i^n)|(t_{i}^n - t_{i-1}^n) \\ &\to \int_a^b|f(y)|dy \\ &= ||f||_{L^1([a,b])}. \end{align*} as $n \to \infty$. Here, the values $c_i^n$ are determined by the mean value theorem. So indeed, $V_a^b(F) = ||f||_{L^1([a,b])}$.

MrTheOwl
  • 634
  • I take your answer, thank you very much. The only problem I steel have is what argument I should use to generalize it if $f$ il not Riemann integrable. Maybe I can try by an approximation topic with a continuous function, but this was the way I tried before and I was not successful. –  Jun 01 '24 at 11:15
  • Could you accept the answer then? Under very mild conditions, like $f$ only having a countable number of discontinuities, Riemann Integration and Lebesgue Integration coincide. How is $L^1$ defined in your exercise? What are the other assumptions? – MrTheOwl Jun 01 '24 at 11:46
  • There are not other conditions, the exercise says $f\in L^1((a,b))$. I studied that a function is Riemann integrable if and only if it is continuous almost everywhere, and in that case Riemann integral is equal to the Lebesgue one. –  Jun 01 '24 at 12:49
  • Anyway, maybe this proof is enough and it is very clear, so thank you very much –  Jun 01 '24 at 12:50