Dolbeault operators and Dolbeault cohomology
Turn the bar-partial complex into global cohomology and state exactly which holomorphic sheaves its smooth resolutions compute.
On this page
The question for today
What you’ll understand
- Derive the squared-zero and anticommutation relations from type and d squared equal to zero.
- Compute closed and exact bar-partial forms with correct wedge signs.
- State the Dolbeault sheaf isomorphism for holomorphic forms and vector bundles without unnecessary compactness assumptions.
The operator bar-partial measures failure of holomorphicity and creates a complex whose cohomology detects global obstructions to solving an inhomogeneous equation. Locally, a closed positive-antiholomorphic-degree form has a primitive on a sufficiently small neighborhood. Globally, those primitives may not glue. Dolbeault's theorem expresses that obstruction as sheaf cohomology of holomorphic forms or holomorphic bundle sections. Neither Kähler geometry nor a metric is needed for this identification; those structures enter later when harmonic representatives are used to calculate the groups.
From type splitting to a differential complex
Use Wirtinger derivatives with one half in the definitions: differentiation with respect to z is half of real-x differentiation minus i times real-y differentiation, and conjugate differentiation uses the plus sign. On smooth functions, partial collects the dz terms and bar-partial collects conjugate differential terms. Extend by the graded Leibniz rule. On an integrable complex manifold the different bidegrees in the square of d must vanish separately. This yields partial squared zero, bar-partial squared zero, and anticommutation of the two operators.
For each fixed holomorphic degree p, bar-partial therefore forms a complex in the antiholomorphic degree q. Closed means the next bar-partial is zero; exact means the form is the bar-partial of a global smooth form of degree one less. The cohomology quotient distinguishes these conditions. Holomorphic p-forms occupy the zero-antiholomorphic-degree kernel. In positive degree, exactness always implies closedness but the converse is a local assertion only until global obstructions have been addressed.
Specify the sheaf theorem's scope
The local Dolbeault lemma says every smooth bar-partial-closed form with positive antiholomorphic degree is locally bar-partial-exact. Consequently the sheaves of smooth p-q forms give a fine resolution of the sheaf of holomorphic p-forms. Smooth partitions of unity make those resolution sheaves acyclic on the underlying paracompact manifold. Taking global sections computes sheaf cohomology and produces the displayed Dolbeault isomorphism. This explanation separates the local analytic input from the global topological mechanism.
More generally, let E be a holomorphic vector bundle of finite rank with its integrable bar-partial operator. Its smooth bundle-valued Dolbeault complex computes cohomology of the locally free sheaf of holomorphic sections, with the holomorphic p-form factor included. No compactness or Kähler assumption is needed for this sheaf identification. An arbitrary coherent sheaf need not itself be a vector bundle, so one cannot simply write the same bundle-valued smooth-form complex for it without additional resolutions or constructions.
Local primitives can fail to be periodic
On a flat complex torus, the constant conjugate one-form descends globally from the plane. It is bar-partial-closed. On the plane it has a simple primitive, the conjugate coordinate, but that coordinate is not periodic and hence is not a global function on the quotient. A global primitive on the torus would contradict integration against a holomorphic one-form: Stokes' theorem kills the integral of a global exact expression, while the constant volume pairing is nonzero. This is a concrete obstruction rather than merely a lack of an obvious formula.
Care with antisymmetric signs is equally important locally. Differentiating a coefficient creates a conjugate differential on the left. Reordering it past another conjugate differential changes the sign. Two terms can cancel for precisely that reason, while a superficially similar single term remains nonclosed. The scalar quadratic instrument checks the real-axis restriction of a simple complex potential; the full bar-partial derivative must still be calculated from both real variables, not inferred from one real plot alone.
Work it through
A closed pair whose individual summands are not closed
On complex two-space take . Compute its bar-partial derivative and a global primitive. Compare the single form .
Differentiation gives . The two terms cancel because swapping one-forms changes the wedge sign.
The smooth function satisfies by the product rule. This is a global primitive on the whole affine space, so the class of alpha is zero.
For the single form beta, . Thus beta is not even closed and cannot represent a Dolbeault class of degree zero-one.
The sign in the differentiated wedge distinguishes a genuine exact form from a nonclosed summand. Local solvability is consistent with the explicit global affine primitive.
The living notebook
Explore the model
Interactive lab · function
A real-axis derivative is only part of a complex derivative
Pause and Reset, set both , and predict quadratic value one quarter and real tangent slope one. For the separate complex function , compute , whose coefficient at that real point is one half. Reset and set both , predicting slope minus one and complex coefficient minus one half. The lab plots only , not the full complex function. Edits retain time; Reset clears both controls and clock.
- y at chosen x
- 0.608
- Tangent slope · 2ax
- 1.35
- Signed area · ax³/3
- 0.182
This lab uses only y = ax². Playback traces the curve within the fixed integration interval from 0 to x; it does not estimate numerical error or change the chosen point. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.
Look forWirtinger differentiation uses both real directions; the factor one half must not be lost when inspecting a real slice.
Pause & reason
Check your understanding
Your turn
Try it yourself
On the square torus , show that is closed but not globally bar-partial-exact. Use the integral of its wedge with dz over one fundamental square.
Take a moment to make a prediction or work it out on paper. Open a hint when you need one.
Give me a hint
If a periodic function f were a primitive, dz wedged with its bar-partial would be minus d of f dz. Compare with the nonzero constant area form.
Compare with the solution
The constant form has bar-partial zero. If for a global smooth f, then because the partial-f term wedges to zero. Its integral would vanish by Stokes on the compact boundaryless torus. But for the unit square, a contradiction. The affine primitive conjugate z does not descend to the torus.
Keep the idea
Take it with you
- Bar-partial squared zero creates a cohomology complex with essential wedge signs.
- Dolbeault cohomology computes holomorphic-form sheaf cohomology without a Kähler hypothesis.
- Local primitives need not glue into a global periodic primitive.
Read further
Original sources and deeper explanations to follow at your own pace.
Optional progress, saved in this browser when available. Revisit any lesson at any time.
Lesson not marked complete.