I am confused by two examples of spinor bundles over 4-manifolds, which I saw in various places:

(1) The spinor bundle $S = S_+ \oplus S_-$ associated to a spin or spinc structure of Riemannian four-manifold $M$

(2) If the manifold $M$ has almost complex structure, then the bundle $\tilde S \equiv \Lambda^{0,*}T^*M$ is also a spin bundle, which can also be split into $\tilde S_+ = \Lambda^{0,0}T^*M \oplus \Lambda^{0,2}T^*M$ and $\tilde S_- = \Lambda^{0,1}T^*M$.

My question is: are the two examples the same? In particular, I know that $S_+ \otimes S_+ = \Lambda_+^2 T^*M \oplus C^\infty(M)$ where $\Lambda_+^2T^*M$ denotes the bundle of self-dual two forms (corresponding to symmetric product of two $S_+$), I wonder if the same is true for $\tilde S_+$? I think it is true, since I can multiply by $(\sigma_{\mu \nu})^{\alpha \beta}$ to go from a bi-spinor to a self-dual 2-form. But I want to see if there is a more geometrical way (playing with the $(0,{\rm even})$-forms) to see it.


  • $\begingroup$ This can not be true, as the spin or spinc bundle is not unique. For expamle, if $S$ is a spinc bundle, $S\otimes L$ is still a spinc bundle where $L$ is any line bundle. $\endgroup$
    – shu
    Oct 30, 2013 at 11:49
  • $\begingroup$ Yes, but I think we can still use $\sigma_{\mu \nu}^{\alpha\beta}$ to obtain a self-dual 2-form charged under the line bundle. I am reading Liviu's note and hope to understand this. $\endgroup$
    – Lelouch
    Oct 30, 2013 at 15:58

1 Answer 1


First let me point out that the bundle $\Lambda^2_+ T^*M$ makes sense only when $\dim M=4$. Maybe you should add this assumption.

As for your question, I think that you can find the answer in Example 1.3.3 page 30 of these notes.

  • $\begingroup$ Yes, I have the assumption in the second line; maybe I should make it a global announcement. And I have a quick look at your note, which is amazing! Thanks. $\endgroup$
    – Lelouch
    Oct 30, 2013 at 15:25
  • $\begingroup$ I read the section you recommend, and I found the quadratic map $q$ gives a self-dual 2-form for each spinor $\psi \in S_+$, which is not exactly what I want. So I modify the $q$, such that for any two spinor $\psi_1$ and $\psi_2$, construct $\left| {{\psi _1}} \right\rangle \left\langle {{\psi _2}} \right| + \left| {{\psi _2}} \right\rangle \left\langle {{\psi _1}} \right| - \frac{1}{2}\left[ {\left\langle {{\psi _1}} \right|\left. {{\psi _2}} \right\rangle + \left\langle {{\psi _2}} \right|\left. {{\psi _1}} \right\rangle } \right]$. $\endgroup$
    – Lelouch
    Oct 31, 2013 at 20:19
  • $\begingroup$ I wonder if this is the map from $S_+\otimes S_+$ to $\Lambda^2_+T^*M$? It seems right but I am not sure. $\endgroup$
    – Lelouch
    Oct 31, 2013 at 20:21
  • $\begingroup$ Yes, that's the map. $\endgroup$ Nov 1, 2013 at 9:06

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.