Let X be a normed space over $\mathbb{C}$ and Y be a finite-dimensional space over $\mathbb{C}$ with supremum norm. If K is a subspace of X and $f:K\rightarrow Y$ is a bounded linear operator, prove that f can be extended to a bounded linear operator F from X into Y such that $||F|| =||f||$.

For each $x \in K$, $f(x) \in Y$. Let $\lbrace e_1,\cdots , e_n \rbrace$ be a basis for Y then $f(x)=\sum_{i=1}^n a_i e_i$, then $$||f||=\sup \lbrace |f(x)|\; ; \; ||x||_X=1\rbrace=sup|\sum_{i=1}^n a_i e_i|\leq \sum_{i=1}^n a_i \sup{e_i}$$

Then I put $F=f(x)$ when $x \in K$ and $F(x)=0$ otherwise.

  • $\begingroup$ If you define it like that it will not be linear. Take $x \in K$ and $y \notin K$. Then $x+y \notin K$. But $0=F(x+y)\neq F(x)+F(y) =F(x)$. $\endgroup$
    – Maik Pickl
    Jan 13 '17 at 13:26
  • $\begingroup$ You can assume that $Y=\mathbb{C}^n$, then you consider projections $\pi_j:\mathbb{C}^n\to\mathbb{C}$, then you apply Hahn-Banach for each composition $\pi_j\circ f$ and finally you just combine results with inclusions $i_j:\mathbb{C}\to\mathbb{C}^n$. $\endgroup$
    – freakish
    Jan 13 '17 at 13:27

As freakish said in a comment, the key to solution is that the norm on $Y$ is the supremum norm, which implies $\|F\|=\max\|F_j\|$ when a linear operator $F$ is written out in components, $F=(F_1,\dots, F_n)$. So it suffices to get $F_j$ such that $\|F_j\|\le \|f_j\|$ for all $j$, but this this precisely what the Hahn-Banach theorem delivers.

The result would not be true for any other $\ell^p$ norm on $Y$ (with the exception of 2-dimensional case, when $\ell^1$ is isometric to $\ell^\infty$).


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.