Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm stuck on this problem, namely I cannot find a bounded subset in $L^\infty(0,1)$ such that it is not mapped by the canonical inclusion $$j: L^\infty(0,1)\to L^1(0,1)$$ onto a relatively compact subset in $L^1(0,1)$. Can anybody provide me an example? Really I don't see the point.

My thoughts are wondering on the fact that the ball of $L^\infty(0,1)$ is norm dense in $L^1(0,1)$ so the inclusion cannot be compact, however, as i said, no practical examples come to my mind.

Thank you very much in advance.

share|cite|improve this question
I changed the $L_1$ in your title to $L^1$. Hope you don't mind. – NKS Feb 24 '12 at 0:17
Maybe I'm missing something, but an operator is compact if the image of the closed ball is pre-compact and by your argument that would imply $L^1(0,\infty)=L^\infty(0,\infty)$. – azarel Feb 24 '12 at 0:19
The ball of $L^\infty$ is not norm dense in $L^1$. For instance, you cannot approximate the constant function 2 in $L^1$ norm by functions from the unit ball of $L^\infty$ (i.e. by functions bounded by 1). – Nate Eldredge Feb 24 '12 at 2:43
up vote 2 down vote accepted

Rademacher's functions are useful tools here. Let $\alpha, \beta \in \mathbb R$ be distinct real numbers. Define $f \in L^\infty(0,1)$ as $$ f(x) = \begin{cases} \alpha & 0 < x \le 1/2\\ \beta & 1/2 < x < 1 \end{cases} $$ Then set $u_n(x) = f(2^nx \pmod 1).\ $ You can compute directly that $(u_n)$ has no convergent subsequence in any $L^p$.

share|cite|improve this answer
It’s even easier to see if you set $u_n(x)=f\left(2^nx\bmod 1\right)$. – Brian M. Scott Feb 24 '12 at 0:28
Yes, in fact when I did this exercise I ended up doing precisely that. No reason not to change it, I guess. – NKS Feb 24 '12 at 0:31
Thanks a lot everybody – uforoboa Feb 24 '12 at 0:55

This is actually just a variant of a special case of NKS’s example, but it may be especially easy to visualize with this description.

For $n\in\mathbb{Z}^+$ and $x\in(0,1)$ let $f_n(x)$ be the $n$-th bit in the unique non-terminating binary expansion of $x$. Then $\|f_n\|_\infty=1$, but $\|f_n-f_m\|_1=\frac12$ whenever $n\ne m$.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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