A Kähler–Hodge manifold $M$ can be defined as a Kähler manifold whose Kähler form $\omega$ is integral, namely $\omega\in H^{2}(M,\mathbb{Z})$. It is known then that there always exists a Hermitian line bundle $L\to M$ whose Chern class satisfies $c_{1}(L) = [\omega]$. I have the following questions:

  • Does anyone know a reference where the statements above are proven?

  • How can this line bundle $L$, given a particular integral Kähler form $\omega$, be explicitly constructed?

  • What happens if $\omega$ is not integral? Can one still construct some kind of $\mathbb{R}$-bundle whose curvature is $\omega$?



2 Answers 2


At least when $M$ is compact, which I assume here, the standard reference for this subject is probably the book

  • Algebraic Geometry, Griffiths & Harris

but the following books are excellent references as well :

A proof could run as follows:

  • Hodge's theorem implies that there exists a complex line bundle $L$ on $M$ with first Chern class $c_1(L)=\omega$;
  • Endow $L$ with an hermitian metric $(L,\|\cdot\|_0)$; its curvature form $c_1(L,\|\cdot\|_0)$ is a $(1,1)$-form cohomologous to $\omega$. By the $dd^c$-lemma, there exists a function $\phi$ such that $\omega= c_1(L,\|\cdot\|_0)+dd^c\phi$; then $\|\cdot\|=e^{-\phi}\|\cdot\|_0$ is a hermitian metric on $L$ with curvature form equal to $\omega$.

For an explicit construction of $L$: the standard proof of Hodge's theorem is cohomological, and uses a long exact sequence in cohomology. Tracing down the arguments gives an actual construction if one is given an open covering of $M$ by contractible open sets.

The arguments really require that $\omega$ be integral. A generic complex torus of dimension $>1$ carries no complex line bundle, but has many Kähler forms.

  • $\begingroup$ It seems that you are assuming that $M$ is compact, a restriction which was not in the question. $\endgroup$
    – abx
    Commented Feb 18, 2015 at 10:52
  • $\begingroup$ @abx: Indeed... I'll edit the answer. $\endgroup$
    – ACL
    Commented Feb 19, 2015 at 10:02
  • $\begingroup$ Thanks for the explanation. How crucial is that $M$ is a compact manifold? Does a similar result hold in for a non-compact base $M$? $\endgroup$
    – Bilateral
    Commented Mar 14, 2015 at 23:00
  • $\begingroup$ Dear @ACL, I have difficulty to understand why a generic complex torus of dimension $>1$ has no complex line bundle. Let $T^n$ be a complex torus, then $Pic^0(T)$ is the dual torus of $T$ of complex dimension $n$. In particular, there exist many non-isomorphic topologically trivial holomorphic line bundles. $\endgroup$
    – Doug Liu
    Commented Jan 16, 2023 at 17:38
  • $\begingroup$ You're perfectly right. What I had in mind is that the Neron-Severi group is in general trivial, so that all complex line bundles are topologically trivial. $\endgroup$
    – ACL
    Commented Jan 17, 2023 at 20:44

If we consider a usual sympleetic manifold with $[\omega]\in H^2(M,\mathbb Z)$ integral then the Weil's theorem guarantees that there exists a Hermitian line bundle $L$ over $X$ with a conneetion $V$ such that $V$ is compatible with the Hermitan strueture on $L$ and induces the symplectic form by $\omega = curv(V)$.

For the second part of your question , The proof is in section "12.2 Integral closed 2-forms and line bundles" http://www.math.toronto.edu/~jeffrey/mat1312/lec14.gq.pdf

In third part of your question: Manifolds which are not quantizable then there is no such line bundle $L$ which first chern class be proportional to kahler class


Your Answer

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