Suppose a function $f$ is Riemann integrable over any interval $[0,b]$. By definition the improper integral is convergent if there is a real number $I$ such that

$$\lim_{b \to \infty}\int_0^b f(x) dx= I := \int_0^\infty f(x)dx.$$

I have shown that if $f$ is nonnegative then this is equivalent for $n \in \mathbb{N}$ to

$$\lim_{n \to \infty}\int_0^n f(x) dx = I.$$

Now I would like to know for general $f$ (not just nonnegative) if the following proposition holds.

Suppose the improper integral of $f$ over $[0,\infty)$ is convergent. Let $A_n \subset [0, \infty)$ be a nested sequence of compact sets such that $A_1 \subset A_2 \subset A_3 \subset ...$ and $\cup_n A_n =[0,\infty)$, and where the Riemann integral is defined on each $A_n$. Then

$$\lim_{n \to \infty} \int_{A_n}f(x) dx = \int_0^\infty f(x) dx.$$

This question is not about Lebesgue integration. Thank you for any help you can give me.

  • $\begingroup$ +1 for thinking of alternative definition of improper integral. $\endgroup$ – Paramanand Singh Feb 11 '17 at 3:50

If the improper integral converges conditionally and not absolutely, then the limit of $\int_{A_n} f$ need not exist. This is somewhat surprising since $A_n \subset A_{n+1}$ and $A_n \uparrow [0,\infty).$

For example, it is well known that the improper integral of $f(x) = \sin x / x$ converges conditionally with

$$\lim_{c \to \infty}\int_0^c \frac{\sin x}{x} \, dx = \frac{\pi}{2}.$$

A counterexample to your conjecture is provided by the following sequence $A_n$ where each set is a finite union of intervals with gaps, of the form

$$A_n = [0, 2n\pi - \pi] \cup \bigcup_{k=n}^{2n}[2k\pi,2k\pi + \pi].$$

It is easy to show that $A_n \subset A_{n+1}$ for all $n$. Furthermore for any $c > 0$ there exists $n$ such that $2n\pi - \pi > c$ and $[0,c] \subset A_n$. This implies $\cup_n A_n = [0,\infty)$.

The integral over $A_n$ is

$$\int_{A_n}\frac{\sin x }{x} \, dx = \int_0^{2n\pi - \pi}\frac{\sin x }{x} \, dx + \sum_{k=n}^{2n} \int_{2k \pi}^{2k \pi + \pi} \frac{\sin x} {x} \, dx,$$

which can be shown to converge to a value greater than $\pi/2 + \log 2 /\pi$.

The first integral on the right-hand side converges to $\pi/2$ and, since $\sin x \geqslant 0$ for $x \in [2k \pi,2k \pi + \pi]$, it follows that

$$\int_{2k \pi}^{2k \pi + \pi} \frac{\sin x} {x} \, dx > \frac{1}{2k\pi + \pi}\int_{2k \pi}^{2k \pi + \pi} \sin x \, dx = \frac{2}{2k\pi + \pi} > \frac{1}{\pi}\frac{1}{k+1}.$$


$$\lim_{n \to \infty}\int_{A_n} \frac{\sin x }{x} \, dx > \frac{\pi}{2} + \lim_{n \to \infty} \frac{1}{\pi} \sum_{k = n}^{2n}\frac{1}{k+1}. $$

There can be no unique value of the limit of the integral for every choice of sequence $(A_n).$

  • $\begingroup$ Very good answer to a very good question. +1 $\endgroup$ – Paramanand Singh Feb 11 '17 at 3:49
  • $\begingroup$ Is this still true if you replace Riemann integrals with Lebesgue integrals? $\endgroup$ – Keshav Srinivasan Oct 16 '18 at 4:37
  • $\begingroup$ @KeshavSrinivasan: Notice that the counterexample involves $\frac{\sin x}{x}$ which is not Lebesgue integrable on $[0,\infty)$. This is very much a consequence conditional convergence of an improper Riemann integral. Lebesgue integrable means absolutely integrable and this issue should not come up. $\endgroup$ – RRL Oct 16 '18 at 18:13

Not true. You can choose the $A_n$ in a biased way. Consider $f(x)=\sin(x)$ and $A_n=[0,n\pi/2]$.

  • $\begingroup$ Sorry. This is not helpful. The improper integral $\int_0^\infty \sin x \, dx$ does not exist. $\endgroup$ – SAS Feb 10 '17 at 21:15
  • $\begingroup$ Under this new hypothesis, it is still not true. See RRL's answer, it has a good explanation and example. $\endgroup$ – Fimpellizieri Feb 10 '17 at 21:33

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.