3
$\begingroup$

I'm looking for references for the following fact, to pass on to a grad student. I think I can prove it, but it is a bit sloppy and I'd rather have something which is already written up.

Let $X$ be a complex manifold. Let $E$ be a smooth, complex vector bundle. Let $C^{\infty}(E)$ be the sheaf of smooth sections of $E$. A dee-bar connection is a $\mathbb{C}$-linear map $\nabla : C^{\infty}(E) \to C^{\infty}(E) \otimes \Omega^{0,1}$, obeying the Liebnitz rule $$\nabla(f \sigma) = f \nabla(\sigma) + \bar{\partial}f \otimes \sigma$$ Just as with connection in the $C^{\infty}$ setting, we can define the curvature; it is a section of $\Omega^{0,2} \otimes \mathrm{End}(E)$. Suppose that the curvature is zero.

Let $\mathrm{Ker}(\nabla)$ be the kernel of $\nabla : C^{\infty}(E) \to C^{\infty}(E) \otimes \Omega^{0,1}$. Let $\mathcal{O}$ be the sheaf of homolomorphic functions. It is easy to see that $\mathrm{Ker}(\nabla)$ is a $\mathcal{O}$-module. I want a reference for the fact that:

$\mathrm{Ker}(\nabla)$ is a locally free $\mathcal{O}$-module of rank $\dim E$.

The analogous fact for smooth connections is (one direction of) the Riemann-Hilbert correspondence; see, for example, Theorem 2.6 here.

$\endgroup$
3
$\begingroup$

This is Theorem 2.1.53 in Donaldson--Kronheimer's Geometry of 4-manifolds. They prove it from scratch via the contraction mapping principle but there are arguments from the Newlander--Nirenberg theorem also.

$\endgroup$
1
  • $\begingroup$ Thanks! This isn't really a proof from scratch; large parts of it are copying the proof of Dolbeault's lemma, so I can just think "plug in the proof of Dolbeault here". $\endgroup$ – David E Speyer Jul 17 '13 at 19:55
2
$\begingroup$

Kobayashi, "Differential geometry of complex vector bundles", Proposition 3.7. He does it using the Frobenius theorem.

$\endgroup$

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.