4
$\begingroup$

I'm curious whether lacunary functions (functions in the complex plane that are holomorphic in some open ball about the origin, but cannot be analytically extended past that ball) are typical or the exception.

Are lacunary functions dense in the space of functions that are holomorphic on the open unit ball?

Improve this question
$\endgroup$

2 Answers 2

Reset to default
4
$\begingroup$

I'm going to prove density in compact-open topology, i.e. that for any $f$ holomorphic on the disk there is a sequence of lacunary functions convergent to $f$ uniformly on compact sets.

Take any function $g$ lacunary in the unit disk satisfying $g(0)=0$. Then it's easy to see that for $g_n(x)=g(x^n)$, $g_n(x)$ is lacunary again and $g_n$ converges to $0$ uniformly on compacts. Let $f_n$ be the $n$-th Taylor polynomial of $f$, then $f_n$ converges to $f$ uniformly on compacts. It follows $f_n+g_n$ is a sequence of lacunary functions convergent to $f$ uniformly on compacts.

Improve this answer
$\endgroup$
4
  • $\begingroup$ It is perhaps even simpler to consider $f((1+\frac{1}{n})z) + \frac{1}{n} g(z)$ instead. $\endgroup$
    – Mateusz Kwaśnicki
    Commented Jan 8, 2019 at 21:00
  • $\begingroup$ @MateuszKwaśnicki That's true! I was thinking of $g(z)/n$ originally, but for some reason thought it doesn't work because $g$ is unbounded... of course, it doesn't matter on compact subsets. $\endgroup$
    – Wojowu
    Commented Jan 8, 2019 at 21:02
  • $\begingroup$ Thanks for the example! As a follow up, I am curious if "most" functions are lacunary in the following sense: is there a natural measure to study the space of holomorphic functions in the ball, and if so, would the lacunary functions be measure 0? $\endgroup$
    – DJA
    Commented Jan 8, 2019 at 22:12
  • 1
    $\begingroup$ @DylanAltschuler: I do not know if there is a reasonable measure structure on the class $\mathcal{L}$ of lacunary functions, but one can ask a more natural question: is this class residual? The answer seems to be "yes": the class $\mathcal{H}_{\alpha,r}$ of holomorphic functions in the unit disk $D$ that extend to $D \cup D(e^{i \alpha},r)$ for a fixed $\alpha$ and $r > 0$ is nowhere dense, and the complement of $\mathcal{L}$ is the union of $\mathcal{H}_{\alpha,r}$ over all rational $\alpha$ and $r$. $\endgroup$
    – Mateusz Kwaśnicki
    Commented Jan 9, 2019 at 9:02
3
$\begingroup$

Much more is true: the "non-lacunary" series are very rare, from the point of view of Baire's cathegory, in various natural topologies, and from the point of view of probability. These results go back to Polya and Hausdorff, and a nice survey of them is contained in chapter 4 of the book of Bieberbach, Analytische Fortsetzung, Springer 1955.

Improve this answer
$\endgroup$

You must log in to answer this question.

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