3
$\begingroup$

Define H(a) to be the space of holomorphic functions $f(z)$ on $S_a:=\{z:|\Re z| < a\}$ with $$ ||f||^2_a:=\int_{S_a} |f(x+iy)|^2 (1+|x+iy|)^{100} dx \, dy < \infty. $$ Two questions:

  1. Is H(2) dense in H(1)?
  2. Are the entire functions dense in H(1)?

Is there a nice kernel to convolve things with?

This occurs in the following form: When building an equality of integral transforms, I need to use holomorphy of the test function out to some large domain (wide vertical strip), but the final equality converges absolutely even when the test function is holomorphic on a much smaller domain. I want to know if I can relax the assumptions on the test function. (Awfully sure the answer is yes, but why?)

One note: I mean the natural inclusion of H(2) into H(1), so no composing with conformal maps.

$\endgroup$
7
  • 2
    $\begingroup$ The $|x+iy|^{100}$ is rather curious... May I ask why are you interested in this question? $\endgroup$ Commented May 25, 2017 at 23:40
  • $\begingroup$ Ya. You don't gain any functions with that condition: you only lose them. So you are looking at the subspace of the full Bergman space of the strips which have a zero of order $50$ at the origin. Computing these Bergman spaces explicitly shouldn't be too hard because you can actually compute the Riemann map to the disk. Am I missing something? $\endgroup$ Commented May 25, 2017 at 23:57
  • $\begingroup$ Ah, did not mean to introduce a zero, corrected. $\endgroup$ Commented May 27, 2017 at 0:35
  • $\begingroup$ $H(a)$ is still the same as all functions $f$ for which $(z+2a)^{50}f(z)$ is in the unweighted Bergman space, so the weight doesn't seem to have much of an effect. $\endgroup$ Commented May 27, 2017 at 6:15
  • $\begingroup$ Yes. Does that help any? $\endgroup$ Commented May 29, 2017 at 18:01

1 Answer 1

0
$\begingroup$

It's possible I'm over-thinking this: If $f \in H(1)$, let $$ f_\delta(z) = \frac{1}{\sqrt{\pi\delta}} \int_{(0)} f(w) \exp ((w-z)^2/\delta) dw, $$ then $$ f_\delta(x+iy) = \frac{i\exp (x^2/\delta)}{\sqrt{\pi\delta}} \int_{-\infty}^\infty f(i(t+y)) e^{-2itx/\delta} \exp (-t^2/\delta) dt, $$ so $f_\delta$ is entire (integral converges rapidly for all $z$), and $f \in H(b)$ for all $b>1$, by interchanging integrals in the norm and using $$ (1+|(x+iy)-it|) \le (1+|t|)(1+|x+iy|). $$ Also, on $S_a$, $$ f_\delta(z) = \frac{1}{\sqrt{\pi\delta}} \int_{(0)} f(z+w) \exp (w^2/\delta) dw, $$ by contour shifting. Then $f_\delta \to f$ as $\delta \to 0$, by the usual approximation to the identity argument.

This works if we change the norm to $$ ||f||^2_a:=\sup_{|\sigma|<a} \int_{-\infty}^\infty |f(\sigma+it)|^2 (1+|t|)^{100} dt. $$

$\endgroup$

You must log in to answer this question.

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