Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I wonder whether there are natural norms for the space $V$ of vector-valued functions that map $\mathbb R^m$ into $\mathbb R^n$.

Formally, let's define $V$ as the set of $f$ such that $f: \mathbb R^m \rightarrow \mathbb R^n$. If the answers restrict $V$ to only continuous and/or bounded functions, that is fine for me.

I have tried to extend the usual norms for functions that map into $\mathbb R$, but I can never show the triangle inequality. $||f+g||\leq ||f||+||g||$.

I have tried (for the case with $n=2$ and denoting $f=(f_1(\cdot),f_2(\cdot))$:

  1. $||f||=\sup_{x \in \mathbb R^m} \left\{\max\left\{|f_1|,|f_2| \right\} \right\}$
  2. $||f||=\max\{\sup_{x\in\mathbb R^m}|f_1|,\sup_{x\in\mathbb R^m}|f_2|\}$

None of them seem to work.

I asked the question generally, but I am particularly interested in the case with $m=n=2$. Thanks for your suggestions.

Note: This question arises after the helpful comments in this other question.

share|improve this question
    
What do you mean by "multivalued"? –  Qiaochu Yuan May 11 '11 at 2:37
    
@QiaochuYuan I mean that the functions map into $\mathbb R^n$ instead of simply $\mathbb R$. Please feel free to edit the title if you think it is confusing. –  hulp10 May 11 '11 at 2:49
1  
It is confusing, since "multivalued function" is reserved for another thing, see en.wikipedia.org/wiki/Multivalued_function –  lentic catachresis May 11 '11 at 3:14
    
@BrunoStonek Thanks for the edit and sorry for the confusion. –  hulp10 May 11 '11 at 3:16

2 Answers 2

up vote 1 down vote accepted

The two norms you suggest are actually equal and do satisfy the triangle inequality. To see that they are equal, consider that taking a supremum of a finite set (i.e. $|f_1(x)|$ and $|f_2(x)|$ for a fixed $x$), is equal to taking the maximum, so one could rewrite both as

$\|f\| = \sup_{x \in \mathbb{R}^m, i = 1,2} |f_i(x)|.$

To see that it satisfies the triangle inequality, first check that the triangle inequality is satisfies for each fixed $x \in \mathbb{R}^m$. Then this also holds for the supremum over $x$, as inequalities are preserved under taking suprema.

Note that the space of bounded, continuous functions with this norm is a standard Banach space.

share|improve this answer
    
Thanks for the answer. I was doing something wrong when proving the triangle inequality. –  hulp10 May 12 '11 at 2:13

For any $m$, considering $\mathbb{R}^2\simeq \mathbb{C}$ then the space of measurable functions $\mathbb{R}^m \to \mathbb{C}$ which are essentially bounded, i.e. such that

$\|f\|_\infty := \inf \{ C\ge 0 : |f(x)| \le C \mbox{ for almost every } x\} <\infty$

can be equipped with the $\infty$-seminorm defined above.

If you take the quotient by the kernel of $\|\cdot\|$, i.e. if you identify functions equal almost everywhere, then you get a norm on the space of essentially bounded measurable functions with any $m$ and $n=2$, denoted $L^\infty_\mathbb{C}(\mathbb{R}^m)$

share|improve this answer
    
It's too late right now to even think straight, but I suspect that if we take $\inf \{ C\ge 0 : \|f(x)\| \le C \mbox{ for almost every } x\}$ and any $n$, then the same thing works. I'll think it over tomorrow. –  lentic catachresis May 11 '11 at 4:26
    
Thanks for the answer. For my application I only need continuous and bounded functions, but your answer it is still helpful. –  hulp10 May 12 '11 at 2:15

Your Answer

 
discard

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.