3
$\begingroup$

I have a simple question: transitivity of $Spin(7)$ in triples of orthogonal vectors. Let $Spin(7)\subset SO(8)$ act on $\mathbb{R}^8$, and $e_1,e_2,e_3$, $v_1,v_2,v_3$ be two triples of mutually orthogonal vectors.

Is there a (unique?) transformation in $Spin(7)$ that takes $e_i$ to $v_i$, $i=1,2,3$?

If so, which is the easiest way to convince oneself? If not, why? And what about quadruples of vectors?

Thank you very much for you attention and answers.

$\endgroup$
  • $\begingroup$ If such a transformation always exists, it cannot be unique. Indeed the dimension of the manifold $M$ of triples of mutually orthogonal vectors is 18, while $\dim Spin(7)=28$. If $Spin(7)$ acts transitively on $M$ then the dimension of stabilizer of each triple is $28-18=10$. $\endgroup$ – orbits Apr 16 '16 at 17:30
4
$\begingroup$

We know that $\operatorname{Spin}(7)$ acts transitively on unit vectors, with stabilizer $G_2$, so we need only prove that $G_2$ acts transitively on pairs of orthogonal unit vectors. But then apply the same argument, since $G_2$ acts transitively on the unit sphere in $\mathbb{R}^7$, with stabilizer $SU(3)$, which also acts transitively on the unit sphere in $\mathbb{R}^6$. So $\operatorname{Spin}(7)$ acts transitively on triples of mutually orthogonal unit vectors. But the dimension count ensures that it is not unique, apparently, following Semyon Alesker's argument.

$\endgroup$
  • 2
    $\begingroup$ But I think that $\operatorname{Spin}(7)$ has dimension 21. $\endgroup$ – Ben McKay Apr 16 '16 at 18:28
  • $\begingroup$ Yes of course, and the stabiliser of $SU(3)$ acting on $S^5$ is $SU(2)$, which has dimension $3$. In particular, Spin(7) does not act transitively on quadruples. $\endgroup$ – Sebastian Goette Apr 16 '16 at 18:31
  • 4
    $\begingroup$ In fact $\operatorname{Spin}(7)$ acts on the orthogonal quadruples preserving the Cayley 4-planes. $\endgroup$ – Ben McKay Apr 16 '16 at 18:34
  • $\begingroup$ @BenMcKay : you are right. The dimension is 21. Sorry for the mistake. $\endgroup$ – orbits Apr 16 '16 at 18:41

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.