Let us consider an orientable smooth 4-manifold $M$. Pick a vector bundle $T$ that's isomorphic to the tangent bundle $TM$. We then equip $T$ with a cotetrad (or coframe field) $e$ and (spin) connection $\omega$. This lets us describe a Riemannian metric on $M$ using a vector bundle map

$e: TM\rightarrow T$

We have that

$\omega\in so(4)$ (it's a $SO(4)$ connection) and $e\in \mathbb{R}^4$.

We can introduce now a $SO(5)$ connection $A$ such that

$A=\omega+e$ $\hspace{1.0cm}$(because $so(5)=so(4)\oplus \mathbb{R}^4$)

My question is: if I introduce a spin structure (spin bundle), can I define the following Dirac operator:

$D=\gamma^\mu(\partial_\mu+A_\mu)$ ?

Can I consider $A$ as a "pure" (gauge) connection as in the standard case?

  • 1
    $\begingroup$ In order for the isomorphism e : TM -> T to define a Riemannian metric you need to require in addition that the structure group of T is not all of GL(n), but is O(n). (Otherwise the condition is empty: just choose T := TM !) In other words, a Riemannian metric, is given by a smooth reduction of the structure group of the tangent bundle along O(n) --> GL(n) and any such is a vielbein. $\endgroup$ Jul 4, 2012 at 14:23

1 Answer 1


You have to be more precise. $A$ is an SO(5) connection but on what bundle? This is a formula for the Dirac operator which depends on local trivialization. I doubt that this defines a global operator in general (this works for a spin or $spin^c$ structures on TM and doesn't have to work on spin structure on every vetor bundle).

I will recommend you chapters "spin structures on vector bundles" and "dirac opetors" form book D.Salamon,Spin geometry and Seiberg-Witten invariants.

Hope it helped a little.

  • $\begingroup$ Hi! I would like define this Dirac operator only locally on $TM$. Thank you for your suggestion! $\endgroup$
    – Gian
    Jun 20, 2012 at 7:43
  • $\begingroup$ I have just another question: if now we consider $D=e^a\gamma_a^\mu(\partial+A)$, can I consider $A$ as an extended spin connection respect to $\omega$? $\endgroup$
    – Gian
    Jun 20, 2012 at 21:29
  • $\begingroup$ You need an SO(5)-principal bundle which admits a lift to a Spin(5)-principal bundle and then consider your Dirac operator on that. What remains is that you need to decide how you want that Spin(5)-bundle to be related to the original tangent bundle. Probably you want them to be related by reduction/extension? $\endgroup$ Jul 4, 2012 at 14:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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