Suppose $f_n(x), f(x)$ are continuous on $[0,1]$ and $\lim_{n\to \infty}f_n(x)=f(x)$ for all $x\in [0,1]$. Since the convergence is not necessarily uniform, there must exist a counterexample to the fact that $$\lim_{n\to\infty} \int_0^1 f_n(x)dx=\int_0^1f(x)dx$$ but I couldn't come up with any. I don't even understand what properties should $f_n$ have to begin with (except that this sequence is not uniformly convergent).

After I finished typing this question, I found this in similar questions. But I'm still wondering how could one come up with such a counterexample or how to come up with a different counterexample? What should I bear in mind to construct it? Obviously the key idea is that the limit of the integral is not the integral of the limit, but what subideas can I exploit?


Let $f_n$ be the tent function with support $[0, 1/n]$ and with $f_n(1/(2n)) = 2n$. Then it's clear that \begin{align} \int^1_0 f_n(x)\ dx = \frac{1}{2}\text{height }\cdot \text{ base} = 1. \end{align} Moreover, $f_n\rightarrow 0$ pointwise everywhere. Fix $x \in (0, 1]$, then there exists $N$ such that $x>N^{-1}$. In particular, we see that for all $n>N$ we have that $f_n(x) = 0$ since $x>\frac{1}{N}>\frac{1}{n}$. Of course, this is not uniform.

  • $\begingroup$ @Winther Ops. My silly mistake. I have edited the post. $\endgroup$ – Jacky Chong Aug 10 '18 at 2:07
  • $\begingroup$ Sorry, I can't see why this argument at the end implies that the convergence is not uniform. The negation of the definition is: there is $\epsilon$ such that for all $N$ there is $n>N$ and $x_0\in[0,1]$ with $|f_n(x_0)-0|\ge \epsilon$. What does it have with the inequalities $x > 1/N > 1/n$? $\endgroup$ – user531232 Aug 10 '18 at 2:38
  • $\begingroup$ Oh, I see. You used $x > 1/N > 1/n$ to prove pointwise convergence... But how do I see that there is no uniform convergence directly? Of course I can deduce it from the fact that the limit and the integral do not "commute" and appeal to the corresponding theorem, but perhaps it's useful to see it directly. $\endgroup$ – user531232 Aug 10 '18 at 2:59
  • $\begingroup$ Or can I just deduce that the convergence is not uniform from the fact that $\sup_{x\in [0,1]}|f_n(x)|=2n\to +\infty$? $\endgroup$ – user531232 Aug 10 '18 at 3:07
  • 2
    $\begingroup$ Exactly as you said. $\endgroup$ – Jacky Chong Aug 10 '18 at 4:02

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.