0
$\begingroup$

I've been trying to prove that every Hilbert space has a Hilbert basis, and think I have succeeded with the proof below, but I am not sure since this would seem to suggest that it in fact has an orthonormal basis in the standard sense (that every element is a finite linear combination of basis elements), and moreover this proof only uses the inner product and not completeness. Is the proof correct and do the consequent results hold?

Let $P$ be the set of orthonormal subsets of $V,$ a pre-Hilbertian vector space and define on $P$ the partial ordering $T_1\leq T_2$ for $T_1\subset T_2.$ Then any chain $M$ in $P$ has a maximal element by the union of elements of $M,$ so Zorn's Lemma implies existence of a maximal element $B$ of $P.$

Suppose for contradiction that $B$ is not a basis, so there exists $v\in V\setminus\operatorname{span}(B).$ Choose $u\in B,$ which is non-empty for $V$ non-empty, and set $\tilde{v}=\frac{v-\langle v,u\rangle u}{\|v-\langle v,u\rangle u\|}$. Then $\|\tilde{v}\|=1$ and $\langle u,\tilde{v}\rangle =0,$ so $\{u,\tilde{v}\}$ is orthonormal, so contained in $B$ by maximality. But then $v=(\|v-\langle v,u\rangle u\|)\tilde{v}+\langle u,\tilde{v}\rangle u\in\operatorname{span}(B)$. Contradiction.

p.s. I am aware similar questions to this have been asked but I have not found my answer in any of those I have read. Standard proofs of the result I was trying to prove seem to use the orthogonal decomposition in a Hilbert space, which indeed uses the properties I avoided above.

$\endgroup$
2
$\begingroup$

There is an error in your proof, when you write that $\left\{u,\tilde v\right\}\subset B$ “by maximality”. The fact that $B$ is maximal doesn't imply that it contains every orthonormal set. In $\mathbb R^2$, endowed with its usual inner product, $\{(1,0),(0,1)\}$ is a maximal orthonormal set, $\left\{\frac35,\frac45\right\}$ is orthonormal, but $\{(1,0),(0,1)\}\not\supset\left\{\frac35,\frac45\right\}.$

$\endgroup$
  • $\begingroup$ Ah of course, I am not too used to the definitions in Zorn's lemma so that's really helpful in general, thank you! $\endgroup$ – user294388 Sep 16 at 21:12
  • 2
    $\begingroup$ You haven't answered the question in the subject line. $\endgroup$ – Rob Arthan Sep 16 at 21:40

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.