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.

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

Consider a normed vector space $V$. Suppose that for every pair of unit vectors $v,w$ there exists a linear isometry which sends $v$ to $w$ (and leaves the subspace spanned by $v$ and $w$ invariant).

Does it follow that $V$ is an inner product space?

By the formulation and the parallelogram law the problem immediately reduces to the two-dimensional case. I couldn't find a nice argument for that case. It would be nice to have a relatively elementary argument for this.

It seems to me that the answer for $\dim{V} = 2$ should be yes: the group $G$ of linear isometries of $V$ is a closed subgroup of $GL(2,\Bbb R)$ and since the unit sphere is compact, $G$ is compact. However, there aren't that many infinite compact subgroups of $GL(2,\Bbb R)$: up to conjugation, they are $O(2)$ and $SO(2)$. I would like this to tell me that I really have a circle as a unit sphere but I don't really know how to make this precise and how to conclude.

Could you please help me finishing this up or give me an alternative argument (or a counterexample)?

Bonus question: What happens if I drop the condition on leaving the subspace spanned by $v$ and $w$ invariant?


share|cite|improve this question
You're almost there! If $v$ is a unit vector, the unit sphere must be the orbit of $v$ under the action of $\text{O}(2)$ or $\text{SO}(2)$ which in either case is just the usual unit circle (up to scale). – Qiaochu Yuan Aug 6 '12 at 19:20
Thank you for confirming that I didn't hallucinate! – geezer Aug 6 '12 at 19:28
@geezer: The two accounts have been merged, and cookies from the most recent should give you access. Welcome to the forum, again! – mixedmath Aug 8 '12 at 9:38
@mixedmath: thank you very much for your help and the welcome! Having registered my account, I expect no further trouble with lost cookies :-) Sorry for the inconvenience. – geezer Aug 8 '12 at 11:17
up vote 3 down vote accepted

The answer to the bonus question in full generality is no. A Banach space is called transitive if for every unit vectors $u,v$ there is a surjective linear isometry mapping $u$ to $v$. There are nonseparable transitive Banach spaces which are not Hilbert spaces. The Banach-Mazur rotation problem asks whether every separable transitive Banach space is a Hilbert space. This remains unsolved. You can find a bunch of references by googling "transitive Banach space" or "Banach-Mazur problem". There is an MO thread with a proof for finite dimensions and a nonseparable counterexample.

share|cite|improve this answer
Thank you very much for this very interesting and helpful information, I wasn't aware that this was a well-known problem! – geezer Aug 7 '12 at 8:40

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.