5
$\begingroup$

For a Kahler manifold $M$, and a smooth vector bundle $E$ over $M$, let us denote by $A^{(p,q)} := \Omega^{(p,q)} \otimes E$ the bundle of forms with values in $E$. Now with respect to a choice of connection on $E$, we can extend $d$ to a mapping d$ _E: A^{(p,q)} \to A^{(p+1,q+1)}$. Moreover, we can extend $\partial$, and $\overline{\partial}$, to mappings $\partial_E: A^{(p,q)} \to A^{(p+1,q)}$, and $\overline{\partial}_E: A^{(p,q)} \to A^{(p,q+1)}$ respectively. As is well-known, if we also assume that $E$ is holomorphic, then we can choose a connection such that $\overline{\partial}_E: A^{(0,\bullet)} \to A^{(0,\bullet)}$ is a chain complex. However, the same cannot be said for the sequences ${\partial}_E: A^{(\bullet,0)} \to A^{(\bullet,0)}$, and d$ _E: A^{(\bullet,\bullet)} \to A^{(\bullet,\bullet)}$, meaning that in general we have no analogue for holomorphic Dolbeault, or de Rham, cohomologies. We do have, however, natural analogues of the Kahler identities. What I would like to know is, are these still of any real interest, given that they can no longer be used to prove equality of holomorphic, anti-holomorphic, and de Rham cohomologies? For example, in Voisin and Huybrecht's books they do not even appear.

$\endgroup$

1 Answer 1

6
$\begingroup$

They can be used to prove the vanishing theorems (and more). Kodaira-Nakano vanishing theorem and Kodaira embedding theorem follow from these identities. One proves that the difference of the $\partial$ and $\bar\partial$-Laplacians is a commutator of Hodge $\Lambda$ operator and the curvature. This commutator happens to be positive or negative when the bundle is positive or negative; vanishing of cohomology follows immediately, because the Laplacians themselves are positive operators.

$\endgroup$
1

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.