I want to show that $(L^\infty,\|\cdot\|_\infty)$ is a normed vector space. I understand that there are two things to show; firstly, that $L^\infty(X,\mu)$ is a linear space and secondly that $\|\cdot\|_\infty$ defines a norm on this space.

I don't have solutions to check whether what I have done here is correct or not - could somebody verify for me and point out anything I have done wrong, or that could be done better? In particular:

  1. In part (iii) of (b), how is it that taking out the factor of $|c|$ out of the $\inf$ doesn't affect the value of $M\in\mathbb R^+$ which bounds $|cf(x)|$?

  2. Similarly, in part (iv) of (b), should there be a $\le$" sign on the jump from line two to three when taking the $\inf$ of the sum?

(a) $L^\infty(X,\mu)$ is a linear space:

Take $f+g\in L^\infty(X,\mu)$ then $\exists\,M,N\ge 0:|f(x)|\le M\,,\forall x\in X\setminus A$ and $|g(x)|\le N\,,\forall x\in X\setminus B$, where $A,B$ are measurable sets of measure zero.

Consider then $|(f+g)(x)|\le|f(x)|+|g(x)|\le M+N,\,\forall x\in X\setminus(A\cup B)$. So, defining $P:=M+N\ge0$ we have shown that $|(f+g)(x)|\le P$. Thus, $f+g\in L^\infty(X,\mu)$.

Take then $f\in L^\infty(X,\mu)$ and some $c\in\mathbb R$. Then consider $|cf(x)|=|c||f(x)|\le|c|M=:Q$ where, clearly, $Q\ge0$. And so, it follows that $cf\in L^\infty(X,\mu)$.

Hence, $L^\infty(X,\mu)$ is closed under the addition and scalar multiplication of it's elements, and so is a linear space.

(b) $(L^\infty,\|\cdot\|_\infty)$ is a normed vector space:

We recall that $\|f\|_\infty=\inf\{M:|f(x)|\le M,\,\forall x\in X\setminus A\}$, where $f\in L^\infty(X,\mu)$.

(i) Suppose $f\in L^\infty(X,\mu)$ and consider $\|f\|_\infty=\inf\{M:|f(x)|\le M,\,\forall x\in X\setminus A\}$. By definition, such an $M$ is greater than or equal to zero, hence $\|f\|_\infty\ge0$, showing that our norm is nonnegative.

(ii) Consider: $\|f\|_\infty=0\iff\inf\{M:|f(x)|\le M,\,\forall x\in X\setminus A\}=0$

$\iff\forall x\in X\setminus A,\,|f(x)|\le0\iff\forall x\in X\setminus A,\,0\le f(x)\le0$ which means that $f\equiv0$ on $X\setminus A$. This establishes nondegeneracy of the norm.

(iii) For $c\in\mathbb R$ consider $\|cf\|_\infty=\inf\{M:|cf(x)|\le M,\,\forall x\in X\setminus A\}=\inf\{M:|c||f(x)|\le M,\,\forall x\in X\setminus A\}$

$=|c|\inf\{M:|f(x)|\le M,\,\forall x\in X\setminus A\}=|c|\|f\|_\infty$, showing multiplicativity of our norm.

(iv)Lastly, for $f,g\in L^\infty(X,\mu)$ consider,

$$\|f+g\|_\infty=\inf\{P:|f(x)+g(x)|\le P,\,\forall x\in X\setminus (A\cup B)\}$$

$$\le\inf\{M+N:|f(x)|+|g(x)|\le M+N,\,\forall x\in X\setminus (A\cup B)\}$$

$$=\inf\{M:|f(x)|\le M,\,\forall x\in X\setminus A\}+\inf\{N:|g(x)|\le N,\,\forall x\in X\setminus B\}$$

$$=\|f\|_\infty +\|g\|_\infty$$

So showing the triangle inequality.


I think you could be more careful with how you are treating the statements involving things that only hold on the complement of a measurable set.

E.g. I'd like better:

$\|f\|_\infty=\inf\{M:\exists A, \mu(A)=0 \colon \forall x\in X\setminus A\ |f(x)|\le M,\}$

Otherwise I might ask myself what the set $A$ is supposed to be. But this is just notation...

To (b) iii:

$\inf\{M\colon |c||f(x)|\leq M, ... \}=\inf\{M\colon |f(x)|\leq \frac{M}{|c|}, ... \}=\inf\{|c|\tilde{M}\colon |f(x)|\leq \tilde{M}, ... \}=|c|\cdot\inf\{\tilde{M}\colon |f(x)|\leq \tilde{M}, ... \}=\|f\|_\infty$

(at least for $c\neq 0$)

To (b) iv: yes I think you should use an $\leq$ sign

Otherwise I don't have concerns about your solutions, seems okay to me.

  • $\begingroup$ Thanks for the feedback. In the second line under (a) I mention that $A$ is a measurable set where $\mu(A)=0$; perhaps I should have noted it again in part (b) if that's what you mean(?). Also, $M$ isn't a set, it's a number which bounds from above the values of $|f(x)|$ when $x\in X\setminus A$, so I don't know what you mean when you write that you would rather see $\|f\|_\infty=\inf\{M:\exists M, \mu(M)=0 \colon \forall x\in X\setminus A\ |f(x)|\le M,\}$. Could you clarify, please? $\endgroup$ – Jeremy Jeffrey James Mar 30 '18 at 15:42
  • $\begingroup$ yeah with M you are totally right it was a typo I meant A I'll correct my answer $\endgroup$ – user412810 Mar 30 '18 at 15:43
  • 1
    $\begingroup$ I understand perfectly what is going on with (b) part (iii) now; thanks for spelling that out for me! $\endgroup$ – Jeremy Jeffrey James Mar 30 '18 at 15:45
  • $\begingroup$ see the problem is (I think) it's not necessarily always the same set $A$, rather all these things hold for some set $A$ with Lebesgue measure zero, you're welcome $\endgroup$ – user412810 Mar 30 '18 at 15:45
  • $\begingroup$ You're right that it's not necessarily the same set $A$ for different $f,g\in L^\infty(X,\mu)$. But given a particular $f\in L^\infty(X,\mu)$ you can say that there is some $A$ which has measure zero for which we have $|f(x)|\le M\,,\forall x\in X\setminus A$. $\endgroup$ – Jeremy Jeffrey James Mar 30 '18 at 15:49

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.