3
$\begingroup$

The space $C^{\omega}(\Omega)$ of real-valued real analytic functions on the open bounded set $\Omega\subset \mathbb R^n$ does not have any obvious or natural metric which would make it a Fréchet space.

I am wondering if there is a way to define a norm on the space of real analytic real-valued functions defined on $\mathbb{R}^n$, even if not in a canonical way.

$\endgroup$
2
  • 4
    $\begingroup$ The space or real analytic functions with a suitable locally convex topology has been studied in detail by functional analysts, one reason being that its lc structure doesn't fit into the familiar special classes (Fréchet, Silva,...). Rather than giving explicit references, I would suggest you check out some of the main actors, e.g., André Martineau, Pawel Domański and Dietmar Vogt. $\endgroup$ Commented Sep 17, 2023 at 5:21
  • 2
    $\begingroup$ If $\Omega$ is connected and $K$ is closed ball contained in $\Omega$, then $\|f\|_K=\sup\{|f(x)|:x\in K\}$ is a norm on $C^\omega(\Omega)$. As mentioned in terceira's comment, the correct locally convex topology was introduced by Martineau. $\endgroup$ Commented Sep 17, 2023 at 7:17

1 Answer 1

2
$\begingroup$

I have no patient to write out all details, so I will just give something based on intuition

Define distance (not norm) on power series space $\text{dist}(A,B) = \sup |A_n - B_n|^{\frac{1}{n}}$ so that $\text{dist}(A(x),B(x)) < \epsilon \implies |\sum_{n \ge 1} A_n (v^n) - \sum_{n \ge 1} B_n (v^n)| \le \sum_{n \ge 1} |A_n - B_n| |v|^n = \frac{1}{1 - \epsilon |v|} - 1 \to 0 \text{ as } \epsilon \to 0$

Remark (off topic): This inspired us to define analytic type Sobolev space, i.e. define the integral distance $\sup_n \int |\frac{1}{n!} d^n f - \frac{1}{n!} d^n g|^{1/n}$

Change the base point of power series $A_m (x + \Delta) = \sum_{n \ge m} \binom{n}{m} A_n \Delta^{n-m}$

Compare at point $x + \Delta$

$$|A_m (x + \Delta) - B_m (x + \Delta)| \le \sum_{n \ge m} \binom{n}{m, n-m} |A_n - B_n| |\Delta|^{n-m}$$

Use $|A_n - B_n| < \epsilon^n$ and $n = p+m$ and definition of Taylor series

$$|A_m (x + \Delta) - B_m (x + \Delta)| \le \sum_{p \ge 0} \binom{p+m}{m,p} \epsilon^m |\epsilon \Delta|^p = \epsilon^m \frac{1}{(1 - |\epsilon \Delta|)^{m+1}}$$

$$|A_m (x + \Delta) - B_m (x + \Delta)|^{\frac{1}{m}} \le \epsilon \frac{1}{(1 - |\epsilon \Delta|)^{1 + \frac{1}{m}}} \le \frac{1}{(1 - |\epsilon \Delta|)^{2}}$$

Let $r < \min{R_A (x), R_B (x)}$ where $R(x)$ is radius of convergence at $x$, we have control

$$\lim_{\text{dist}(A(x),B(x)) \to 0} \sup_{|\Delta| \le r} |\sum |A_n (x + \Delta) - B_n (x + \Delta)| |v|^n| = 0$$

After finite step $x_i = x + \Delta_1 + \cdots + \Delta_i$, in compact set $\cup_{i = 1 .. N} \overline{\mathbb{B}} (x_i, r_i)$, it can still be control by the distance $\text{dist}(A(x),B(x))$ at $x$

Let $f : D_f \to \mathbb{R}$ be analytic, where $D_f$ is maximal continuation

Approximate $f$ in by

forall $\epsilon$, forall compact $D \subset D_f$, exist analytic $g : D_g \to \mathbb{R}$ ($D_g$ is maximal continuation) with $D \subset D_g$, so that $\sup_{x \in D} \text{dist} (\frac{1}{n!} d^n f (x) , \frac{1}{n!} d^n g (x)) < \epsilon$

This will define a net for space of analytic function

This is also similar to the definition of compact-open topology of space of continuous function

$\endgroup$
5
  • $\begingroup$ Is this an AI generated answer? $\endgroup$ Commented Jun 19 at 11:44
  • 1
    $\begingroup$ @DanieleTampieri no $\endgroup$ Commented Jun 19 at 12:06
  • 2
    $\begingroup$ @DanieleTampieri: Definitely not, because the English used is a bit hesitant. This looks very human to me. Also, I find it sketchy in the same way in which I myself would sketch a proof. $\endgroup$ Commented Jun 19 at 12:06
  • $\begingroup$ @AlexM. thanks. As you pointed out and also the OP their self declare, this is a sketchy answer possibly machine translated (AI translated). I retracted also my flag: apologies to ecbc. $\endgroup$ Commented Jun 19 at 13:27
  • $\begingroup$ Thank you for your effort! $\endgroup$ Commented Jun 22 at 8:41

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.