MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

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

I'm working with Kahler manifolds at the moment and looking at their spin$^c$ structure. I really don't know much about spin$^c$ structures in general and don't have enough time to learn it all at present, so I was hoping someone might be able to show me a shortcut in this case.

As far as I understand, for the usual spin$^c$ structure on an $N$-dimensional Kahler manifold $M$, the spinor bundle $S$ is given by the the direct sum of all the anti-holomorphic, that is $S = \bigoplus_{i=1}^N \Omega^{(0,i)}(M)$. Let $$ \nabla^s:S \to S \otimes \Omega^{(1,1)}, ~~~~~~~~~ s \mapsto \sum s_i \otimes \omega_i $$ be the spin$^c$ connection.

Is there a simple direct algebraic description of the Clifford action $$ c:S \otimes \Omega^{(1,1)} \to S $$ in this case? For example, an initial stupid guess might be, for $\omega_i = \omega_i^h + \omega_i^{ah}$, with $\omega_i^{h} \in \Omega^{(1,0)}$ and $\omega_i^{ah} \in \Omega^{(0,1)}$, we would have $$ c(\sum_i s_i \otimes \omega_i) = \sum_i s_i\omega^{ah}_i. $$ I am quite sure this is complete rubbish, but it illustrates the kind of result-from-heaven I'm hoping exists

share|cite|improve this question
I think you probably mean Spin-c structure, not Spin structure. The complex projective plane has $w_2\neq 0$. – Tim Perutz Feb 5 '11 at 20:02
Yes, of course. I've edited accordingly. – Ago Szekeres Feb 6 '11 at 17:12
@Ago: I think your summation $\oplus_{i=1}^N$ should read $\oplus_{i=0}^N$. See, for example Friedrich's book. – Jean Delinez Feb 11 '11 at 18:22
up vote 4 down vote accepted

You ask for a short-cut, and I'm going to interpret this as asking "How can I see that on a Kaehler manifold $X$ the forms $\Omega^{0,\bullet}_X$ are a complex Clifford module, without first going through all that stuff about spin groups and their representations?"

Warning: My scalar factors are probably wrong.

I'm going to start with the Hodge Laplacian $$\Delta = 2(\overline{\partial}\;\overline{\partial}^\ast+\overline{\partial}^\ast\overline{\partial})$$ acting on $S:=\Omega^{0,\bullet}_X$.

What makes $\Delta$ a Laplacian is that it's a second order differential operator whose symbol $\sigma \colon T^\ast X \to \mathrm{End}(S)$ is the quadratic map $\lambda \mapsto - \| \lambda\|^2 \mathrm{Id}_S$. (The symbol is just the coefficient matrix of the leading (i.e. second) derivative terms in $\Delta$, written invariantly.) The symbol of $\overline{\partial}$ is $\lambda\mapsto \lambda^{0,1}\wedge \cdot$, that of $\overline{\partial}^\ast$ is $\lambda\mapsto \iota(g(\lambda))$ where $g\colon T^\ast X \to TX$ is the metric tensor, and from this you can check the symbol of $\Delta$.

$\Delta$ has an evident square root, $D=\sqrt{2}(\overline{\partial}+\overline{\partial}^\ast)$, and this is a Dirac operator: a first order differential operator whose square is a Laplacian. The (complexified) symbol $\rho\colon T^*X\otimes \mathbb{C}\to \mathrm{End}(S)$ of a Dirac operator $D\colon \Gamma(S)\to \Gamma(S)$ makes $S$ into a Clifford module. The natural Clifford multiplication on $(0,\bullet)$-forms is the symbol of $D$, that is $$ \rho(\lambda) = \sqrt{2}(\iota(g(\lambda))+\lambda^{0,1}\wedge \cdot ) \in \mathrm{End}(S). $$ In fact, this Clifford module is a $Spin^c$-structure, which means that pointwise on $X$ it's a isomorphic to a standard Clifford module.

share|cite|improve this answer
Great, this seems to be what I'm looking for. Just some questions about notation though: What do you mean by $\lambda^{(1,0)}$? Also, I assume that by $g(\lambda)$ you mean $g(\lambda, \cdot )$, but what does $\iota$ stand for? – Ago Szekeres Feb 7 '11 at 19:25
Also, what happens when we tensor $S$ with the square root of the canonical bundle $\Omega^{(N,0)}(X)$. And in this case is the spin connection the Levi-Civita connection? – Ago Szekeres Feb 7 '11 at 19:34
By $\lambda^{0,1}$ I mean the $(0,1)$-component of $\lambda\in T^\ast X$ under the decomposition $T^\ast X\otimes \mathbb{C}=T^\ast_{1,0}X \oplus T^\ast_{0,1}X$ into $dz$'s and $d\overline{z}$'s. By $\iota(v)$ I mean the contraction of forms by the vector field $v$. – Tim Perutz Feb 7 '11 at 19:59
Other spin-c structures differ from the canonical one $s_{can}$ by tensoring with line bundles (the Clifford multiplication is given by the same formula tensored with the identity map on the line bundle). And if we have a square root $L$ of $K_X$ then $L\otimes s_{can}$ arises from a spin structure. Not quite sure what you're asking about spin connections... Bear in mind that there's not a unique spin connection in the spinor bundle of a spin-c structure. – Tim Perutz Feb 8 '11 at 0:46
Thanks a lot for your help. Just one last question: So there's no unique spin connection, but will the Levi-Civita connection a spin connection? – Ago Szekeres Feb 8 '11 at 14:45

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.