4
$\begingroup$

Let $\mathbb{D}$ be the unit disk on the plane and let $U,V\subset \mathbb{D}$ be open and such that $U\cup V=\mathbb{D}$.

Is there a holomorphic map $\varphi:\mathbb{D}\times \mathbb{D}\to \mathbb{D}$ such that $\varphi(U\times V)=\mathbb{D}$?

Of course one can ask the same question for "ambient domains" other than $\mathbb{D}$.

$\endgroup$

2 Answers 2

4
$\begingroup$

Edit2: everything works, updating the answer.

Yes. Consider two cases.

Case 1. There is a point on the boundary contained both in the closure of $U$ and $V$. Do a Mobius transform transforming our disk into upper half-plane $\mathbb{H}$ and putting this point to $0$.

Now take the map $(x, y) \mapsto x+y$.

It is open, and its image (from $U \times V$) contains arbitrarily small translations of $U$ and $V$, hence contains $U$ and $V$, hence is $\mathbb{H}$.

Case 2. One of the subsets (say $U$ without loss of generality) contains $\partial\mathbb{D}$ in it's closure. Then, it contains a small annulus $1-\varepsilon<|r|<1$. There is a map from disk to itself which is surjective from this annulus: one can do such an automorphism of a disk that $0$ is in the image of the annulus and then use $z \mapsto z^2$. Composing this map with the projection on the first component we obtain the desired result.

$\endgroup$
4
  • $\begingroup$ I am not sure what you mean in the edit: since $U\cup V=D$, the union of the closures contains $\partial D$, and so $\partial U\cup \partial V=\partial D$. Since the latter is connected, $\partial U\cap \partial V\ne\varnothing$. Or did you mean how to drop the condition $U\cup V=D$? I would be very much interested in merely having $U\cap V=\varnothing$.. $\endgroup$
    – erz
    Commented Mar 8, 2020 at 10:29
  • $\begingroup$ Annulus and concentric circle. $\endgroup$ Commented Mar 8, 2020 at 10:32
  • $\begingroup$ Actually my intuition says this might be a counterexample, but I'm not sure yet. $\endgroup$ Commented Mar 8, 2020 at 10:33
  • $\begingroup$ answer updated to full generality $\endgroup$ Commented Mar 8, 2020 at 14:39
0
$\begingroup$

Certainly not. If you take U, V to be small neighborhoods of distinct points this will fail. Are you sure this is what you need?

$\endgroup$
3
  • $\begingroup$ I've made them bigger. Thank you $\endgroup$
    – erz
    Commented Mar 8, 2020 at 3:33
  • $\begingroup$ @erz better, but you probably want to choose $\phi$ first and allow $U, V$ to be arbitrary with $U\cup V = \mathbb{D}$? Also is $\mathbb{D}$ open or closed for you? $\endgroup$ Commented Mar 8, 2020 at 5:33
  • $\begingroup$ I am not sure I understand your first question. $\mathbb{D}$ is both open and closed in itself. The motivation for my question is mathoverflow.net/questions/354133/generating-h-inftyx I am allowed to take a lot of functions, and compose them with a function of two variables. This is a natural extension in my opinion $\endgroup$
    – erz
    Commented Mar 8, 2020 at 5:52

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .