I am trying to construct a holomorphic function $U \to L^2(\mathbb R)$ for some open $U \subseteq \mathbb C$, which is not locally bounded by an $L^2$ function. (I believe this should exist.)

To do so, I was hoping that there exist entire functions $\mathbb C \to \mathbb C$ that are simultaneously:

  1. bounded (say by $1$) outside an arbitrarily small angular region $D_\alpha = \{0 \leq \arg z \leq \alpha\}$
  2. of arbitrarily large absolute value at a point of modulus $1$ in that region, say of size $y_\alpha$
  3. bounded by $M y_\alpha$ in arbitrarily large disks $B(0, R_\alpha)$, where $M > 0$ does not depend on $\alpha$

Do they exist?

Here the "arbitrary large/small" can be dependent on each other, i.e. I'm looking for such a sequence of functions with $(\alpha, y_\alpha, R_\alpha) \to(0, \infty, \infty)$.

The first two conditions can be satisfied, e.g. by Mittag-Leffler functions as mentioned here: https://mathoverflow.net/questions/29734

The additional difficulty is the controlled growth in large disks. I would be happy too if these functions are only defined on $B(0,R_\alpha)$.


To give more context: if $f_n$ is such a sequence with $\alpha_n$ of size $1/n^2$ (say) so that the angular regions $D_n$ can be chosen pairwise disjoint, and take a subsequence for which the $y_{\alpha_n}$ are at distance $\geq 1$ from each other, then

$$g(y, s) = \sum_{n=1}^\infty \rho(y-y_{\alpha_n})\frac{f_n(s R_{\alpha_n})}{y}$$

with $\rho$ a bump function such that $\rho(y-y_{\alpha_n})$ have disjoint support, is a holomorphic $L^2(\mathbb R_{>1})$-valued function on $B(0,1)$ which is not bounded by an $L^2$-function in any neighborhood of $0$.


It turns out that this is not possible, by Montel's theorem:

Call those functions $f_\alpha$, and w.l.o.g. assume that they attain $y_\alpha$ at the point $s=1$, so that the angular regions $D_\alpha$ all contain the positive real ray. Then $f_\alpha/y_\alpha$ is uniformly bounded on (say) $B(0,2)$ hence is a normal family. But the uniform limit of any subsequence is $0$ outside $[0,2]$ and equals $1$ at $s=1$, hence cannot be continuous (let alone holomorphic).

In fact, from $L^{\infty}$ norm is bounded by $L^2$ norm for holomorphic functions it follows that it is true, that holomorphic $U \to L^2$ are locally bounded by an $L^2$ function.


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.