For a better understanding of this question, please see the question and answer here.

In $Spin(8)$ there are plenty of copies of $Spin(7)$; consider, for instance, the antiimage of $SO(7)<SO(8)$ by the double cover for any $SO(7)<SO(8)$ obtained by fixing some direction in $\mathbb{R}^8$. So it is true, there are plenty of $Spin(7)<Spin(8)$ (let's call this subgroups primitives).

By the triality automorphism of $Spin(8)$, the $Spin(7)$ subgroups go to other $Spin(7)$ subgroups. But the hot point is that it is not a permutation between primtives subgroups: that is, if we perform triality on a $Spin(7)<Spin(8)$ obtained by fixing $e_1\in\mathbb{R}^8$, then we get a copy $Spin(7)<Spin(8)$ whose proyection onto $SO(8)$ is not a double cover of some subgroup $SO(7)<SO(8)$, but an isomorphism over some $Spin(7)<SO(8)$. And it happens that this $Spin(7)<SO(8)$ may be described as the stabilizer of a 4-form, as described in the link above.

So the corner question is: Why this $Spin(7)$ obtained by performing triality to a primitive $Spin(7)$ should in fact be so strangely described, as the stabilizer of some 4-form? Is there a nice way to understand this fact?

Any idea or suggestion is welcome.


I think that you want to look at the relevant passages in Spin Geometry by Lawson and Michelsohn, particularly Chapter IV, Sections 9 and 10, where they explain in general how the square of a spinor can be written as a sum of differential forms. This gives a general way to see how having a parallel spinor (in any dimension) defines a set of parallel forms and conversely. This may be the 'nice way to understand this fact' that you seek.

| cite | improve this answer | |
  • $\begingroup$ "and conversely": in general, must every invariant form be the square of some invariant spinor (say, wrt some invariant bilinear/sesquilinear form on complex spinors)? $\endgroup$ – Paul Reynolds Mar 5 '15 at 11:46
  • 1
    $\begingroup$ @PaulReynolds: Sorry, I wrote the above in some haste and it's perhaps not as clear as it could be. I didn't mean to suggest that a parallel form implies a parallel spinor in all dimensions. That's certainly not true in general. What I meant was that, sometimes, the group of linear transformations fixing some set of forms on a vector space can be characterized as the subgroup that fixes the square of a spinor. The reference I mentioned above explains this fully. $\endgroup$ – Robert Bryant Mar 5 '15 at 12:31
  • $\begingroup$ Right, thanks, of course, e.g. the volume form on any generic holonomy space. I have never really understood what's particularly special about the forms that do come from spinors vs those that don't, but I guess that's another question. $\endgroup$ – Paul Reynolds Mar 5 '15 at 12:55

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.