1
$\begingroup$

Many series (in fact, "most" Taylor series) have a natural boundary at the unit circle. This boundary is only 1 dimensional, in the sense that only along the unit circle is it true that there is a dense set of singularities around every point. Do there exist any nice examples of series that have a 2-dimensional space of natural boundaries?

The best I've been able to come up with so far is something like $$\sum_{n=1}^\infty \frac{x^n}{\frac{n}{\alpha} - \beta-x^n} $$ In the complex plane, this looks like: enter image description here

This is close to a solution, since it has many singularities outside of just a circle, however, these singularities don't become dense except along a circle.

I'm thinking there might be some way to craft a function like this by doing something along the lines of: $$\sum_{n=1}^\infty \frac{1}{n^2(\sin(f(x,n)) - \cos(g(x,n)))} $$ However, I can't think of a good way to use this to create dense singularities in some places, and not end up having singularities everywhere.

Any help or ideas would be appreciated!

$\endgroup$
1
  • $\begingroup$ This (very old) question, to which I never got an answer, might be worth revisiting in relation to the present question. $\endgroup$
    – Gro-Tsen
    Commented Nov 18, 2021 at 14:50

2 Answers 2

4
$\begingroup$

Existence. (Maybe not a "nice example" as requested, though.)
Take any simple closed curve $S$ with Hausdorff dimension $2$. (I am assuming you mean Hausdorff dimension when you say "dimension".) Take any function $F(z)$ on the unit disk with the unit circle $T$ as its natural boundary. By the Riemann mapping theorem, we get a conformal equivalence $\varphi$ from the region inside $S$ onto the unit disk.
Our new function is $G(z) = F(\varphi(z))$.

$\endgroup$
1
  • 3
    $\begingroup$ Assuming that your curve is nowhere analytic, you can just take the Riemann map itself (no need for the additional function F). Moreover, I think it could happen that the function G as you describe it doesn't have natural boundary S (e.g. when F is the inverse of phi). $\endgroup$ Commented Nov 18, 2021 at 13:47
3
$\begingroup$

It is easy to prove that for every region $D$ there exists a function $f$ analytic in $D$ such that $\partial D$ is the "natural boundary" that is $f$ does not have an analytic continuation into any larger region. So whatever you mean by ``$\partial D$ is $2$-dimensional'', you can always have such an example. Such function is easy to construct: zeros of an analytic function can be arbitrarily assigned, so choose a sequence of zeros whose limit set is dense in $\partial D$.

Another even simpler method is to take a dense countable set $\{ z_k\}$ in $\partial D$ and construct a series $$\sum_k\frac{c_k}{z-z_k}$$ where $c_k$ are so small that the series converges on every compact subset of $D$.

$\endgroup$

You must log in to answer this question.

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