On a complex manifold, if we are dealing with the $d$ operator, there's a pretty easy way of showing some form is not $d$-exact, simply by integrating in a closed loop. If you can find a loop that is closed, but gives a nonzero integral for your form, then you know the form is not $d$-exact.

However, I can't seem to find an equivalent test for $\overline\partial$-exact forms. My intuitive understanding of the complex vector space is a bit shaky, but it I think if we integrate $\overline\partial f$ in a closed loop, we will get $-\int\partial f$ in that loop, which we don't really know anything about. We can't come up with a loop on which $\int\partial f$ is necessarily zero, or necessarily anything specific.

Is there some way to salvage this method, or is there another way we can prove $\overline\partial$-exactness? Any overview about how you'd approach this sort of problem is also welcome.


Certainly if $\alpha$ is a $(0,1)$-form which is $\overline{\partial}$-exact, it must be $\overline{\partial}$-closed as well, since $\overline{\partial}^2 = 0$. Thus if $\alpha$ is not $\overline{\partial}$-closed, it cannot be $\overline{\partial}$-exact.

Assume then that $\alpha$ is $\overline{\partial}$-closed. In this case, my understanding is that your question is deep, and the answer depends greatly on the complex geometry of the manifold you are working on (whereas for the $d$-operator, knowing whether a form is exact is essentially topological).

If your manifold is $\mathbb{C}^n$ or a polydisk, the $\overline{\partial}$-poincare lemma says that $\alpha$ is necessarily $\overline{\partial}$-exact. I'm no expert in this field, but my understanding is that if your manifold is strongly pseudoconvex, then $\alpha$ is necessarily $\overline{\partial}$-exact (this is Cartan theorem B, I think).

A central theorem for finding solutions to $\overline{\partial}u = \alpha$ is Hormander's theorem, which gives conditions that allow you to find a solution which has some nice $L^2$ properties.

Bo Berndtsson has some notes called "An introduction to things $\overline{\partial}$" which I find to be pretty readable. Dror Varolin also has some notes called "Three Variations on a Theme in Complex Analytic Geometry" which could be useful. You may also want to look at this manuscript by Demailly, but it seems like it could be a bit harder.

Hope some of this helps.

  • $\begingroup$ The two sets of notes you mention are the first and third sections respectively of the book Analytic and Algebraic Geometry: Common Problems, Different Methods. $\endgroup$ – Michael Albanese Jun 12 '13 at 6:17
  • $\begingroup$ @froggie Further Details: The Dolbeault lemma states that, at least locally, every $\overline{\partial}$-closed form is $\overline{\partial}$-exact. See Chapter 1 of Gunning and Rossi's text for details of this. Determining whether these local representations may be glued together to form a global representation is measured by sheaf (or Dolbeault) cohomology. In the above reference, it is shown that if your domain is a simply connected polydomain in $\mathbb{C}^n$, then the Dolbeault cohomology groups vanish. In particular, there is always a global representation for domains of this type. $\endgroup$ – AmorFati Sep 10 '18 at 22:42
  • $\begingroup$ @froggie Cartan's theorem $B$ states exactly that $H^k(X, \mathscr{S})=0$ for all $k \geq 1$ and coherent analytic sheaves $\mathscr{S}$. Gunning and Rossi is the best text for a first look at this rather advanced material. A second look at this material is the classic text by Grauert and Remmert: Theory of Stein Spaces, but is substantially more difficult to read (but would still highly recommend). Moreover, if you would like more details than anyone could ask for, there is a great book by the Kaup brothers called Holomorphic Functions of Many Variables. $\endgroup$ – AmorFati Sep 10 '18 at 22:44

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.