I was studying the construction of the modular lambda function and I started thinking about the following question. Suppose that $\Omega\subset \mathbb{C}$ is an open connected set and $f:\Omega\to \mathbb{C}$ is continuous on $\Omega$ and holomorphic except on a set of measure zero (with respect to 2-dimensional Lebesgue measure). Is $f$ actually holomorphic on all of $\Omega$ ?

If this statement is false, how much does one have to weaken the measure zero condition? For example if the set of measure zero is a $C^1$ curve the above follows from Morera's theorem.


The following is too long for a comment. I will suppose here that your set of measure zero is compact.

In this case, Your question is closely related to so-called continuous analytic capacity. Let $K$ be a compact set in the plane, and let $\Omega$ be the complement of $K$ with respect to $\mathbb{C}_\infty$. The continuous analytic capacity $\alpha(K)$ is defined as

$$\alpha(K):= \sup \{ |f'(\infty)| : f \in A(\Omega), \|f\|_{\infty} \leq 1 \},$$

where $A(\Omega)$ is the set of all functions holomorphic in $\Omega$ that extend continuously to $\mathbb{C}_\infty$.

It is not difficult to prove that $$\alpha(K)=0$$ if and only if for every open set $U$ with $K \subseteq U$, every $f$ holomorphic on $U \setminus K$, continuous on $U$, extend analytically to the whole of $U$.

So what you're looking for is a characterization of the compact sets $K$ with $\alpha(K)=0$. This is certainly very difficult, since it took more than a hundred years to solve a similar problem (but with analytic capacity $\gamma$ instead of $\alpha$), the so-called Painlevé's problem.

It is easy to prove however that if $\alpha(K)=0$, then the area of $K$ must be zero. You're asking if this condition is sufficient. Most likely it is not, but I don't have a counterexample right now.. Usually, for problems related to removable sets for holomorphic functions, it is more useful to consider hausdorff measure instead.

Anyway, I suggest you look in Garnett's book "Analytic Capacity and measure". See also the recent book by Dudziak, "Vitushkin's conjecture for removable sets."

EDIT (Answer to the question) Every compact set $K$ with Hausdorff dimension strictly between $1$ and $2$ is a counterexample. An easy way to see this is to use Frostman's lemma. Indeed, using this lemma, there exists a nonzero Borel measure $\mu$ supported on $K$ with growth $\mu(B(z,r)) \leq C r^{1+\epsilon}$. The growth of $\mu$ implies that the Cauchy Transform $$f(z):=\int_{K} \frac{1}{z-\zeta}d\mu(\zeta)$$ is continuous on $\mathbb{C}$ and holomorphic on $\mathbb{C} \setminus K$. Furthermore, $f(z)\rightarrow 0$ as $z \rightarrow \infty$ and so in particular, $f$ is bounded on $\mathbb{C}$. However, $f$ is non-constant because $zf(z) \rightarrow \mu(K) \neq 0$. Therefore $f$ does not extend analytically to all of $\mathbb{C}$, for otherwise it would be constant by Liouville's Theorem.

  • 4
    $\begingroup$ I know nothing about this subject. However, if I understand correctly the MSE post linked by Claudio Gorodski above: if $K$ is removable for continuous functions, it is also removable for Hölder $C^\alpha$ functions for every $\alpha>0$, hence its $(1+\alpha)$-dimensional Hausdorff measure is $0$. In other words, a necessary condition is that $K$ has Hausdorff dimension at most $1$. Thus, any compact set of Hausdorff dimension strictly between $1$ and $2$ is a counterexample. $\endgroup$ Jan 28 '13 at 17:04
  • $\begingroup$ @EmilJeřábek : I didn't know about the removability criterion for the Hölder functions. This is very interesting. $\endgroup$ Jan 28 '13 at 17:44

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.