Canonical: https://knowlpedia.com/differential-geometry/fundamental-form-almost-hermitian/

# Fundamental 2-form of an almost-Hermitian manifold

The nondegenerate real 2-form obtained by pairing an almost-complex structure with its compatible metric.

Let \((M,J,g)\) be an [almost-Hermitian manifold](https://knowlpedia.com/differential-geometry/almost-hermitian-manifold/). Its **fundamental \(2\)-form** is the differential form \(\omega\) defined by
\[
\omega(X,Y)=g(JX,Y)
\]
for [tangent vectors](https://knowlpedia.com/fiber-bundles/tangent-space-at-a-point/) \(X,Y\) at the same point. Compatibility of \(g\) with \(J\) makes \(\omega\) skew-symmetric, while positivity of \(g\) makes it nondegenerate. Thus \(\omega\) is a smooth, nondegenerate real [\(2\)-form](https://knowlpedia.com/fiber-bundles/differential-k-form/), but it need not be closed. The displayed order fixes the sign convention: interchanging \(X\) and \(JY\) changes the sign.

## Recovering the compatible data

The form and [almost-complex structure](https://knowlpedia.com/differential-geometry/almost-complex-structure/) recover the metric by
\[
g(X,Y)=\omega(X,JY).
\]
Conversely, compatible pairs among \(J\), \(g\), and \(\omega\), together with the appropriate positivity condition, determine the third object. This is pointwise Hermitian linear algebra applied smoothly to the [tangent bundle](https://knowlpedia.com/fiber-bundles/tangent-bundle/).

## Closedness and integrability

If \(d\omega=0\), the almost-Hermitian structure is called almost Kähler, and \(\omega\) is a symplectic form. If \(J\) is [integrable](https://knowlpedia.com/differential-geometry/integrable-almost-complex-structure/), the structure is Hermitian. It is Kähler when both conditions hold. Neither [closedness](https://knowlpedia.com/fiber-bundles/closed-differential-form/) of \(\omega\) nor integrability of \(J\) follows from compatibility alone.

## Conventions and scope

**Warning.** Many authors instead define the fundamental form by \(\omega(X,Y)=g(X,JY)\), which is the negative of the convention here. Either convention is valid, but formulas involving \(d^c\), curvature, and Hamiltonian signs must be adjusted consistently.

## References

1. Shoshichi Kobayashi and Katsumi Nomizu, *Foundations of Differential Geometry*, vol. II, Wiley, 1969. [Publisher record](https://www.wiley.com/en-us/Foundations+of+Differential+Geometry%2C+Volume+2-p-9780471157328). Relevant: Chapter IX, §1, almost-Hermitian structures and their fundamental forms.
2. Daniel Huybrechts, *Complex Geometry: An Introduction*, Springer, 2005. [DOI record](https://doi.org/10.1007/b137952). Relevant: Chapter 3, Hermitian and Kähler geometry.
