1

In this post, we have shown that $\sigma^{n-1}$ is the weak limit of the measures $\delta^{-1} \mathcal L^n\vert_{B(0, 1+\delta)\setminus B(0,1)}$ as $\delta\to 0$.


For a radial function $f\in L^1(\Bbb R^n)$, i.e., $f(x) = \psi(|x|)$ for some $\psi:[0,\infty)\to \Bbb C$, Mattila proves the following expression for the Fourier transform of $f$: $$\widehat f(x) = c(n)|x|^{-(n-2)/2} \int_0^\infty \psi(s)J_{(n-2)/2}(2\pi|x|s) s^{n/2}\, ds \tag{3.33}$$ where $J_m$'s are Bessel functions for $m > -1/2$.


Applying the formula $(3.33)$ to the characteristic function of the annulus $B(0,1+\delta)\setminus B(0,1)$ and letting $\delta\to 0$, we get $$\widehat{\sigma^{n-1}}(x) = c(n)|x|^{(2-n)/2} J_{(n-2)/2}(2\pi|x|) \tag{3.41}$$

Applying the formula $(3.33)$ to the characteristic function of the annulus $B(0,1+\delta)\setminus B(0,1)$, and observing that $$\chi_{B(0,1+\delta) \setminus B(0,1)}(x) = \chi_{\{s>0: 1 < s\le 1+\delta\}}(|x|) \tag{1}$$ we get $$\widehat{\chi_{B(0,1+\delta) \setminus B(0,1)}}(x) = c(n)|x|^{-(n-2)/2} \int_1^{1+\delta} J_{(n-2)/2}(2\pi|x|s) s^{n/2}\, ds \tag{2}$$ Dividing both sides by $\delta$, we have $$\frac{\widehat{\chi_{B(0,1+\delta) \setminus B(0,1)}}(x)}{\delta} = \frac{c(n)|x|^{-(n-2)/2} \int_1^{1+\delta} J_{(n-2)/2}(2\pi|x|s) s^{n/2}\, ds}{\delta} \tag{3}$$ Certainly, $$\frac{c(n)|x|^{-(n-2)/2} \int_1^{1+\delta} J_{(n-2)/2}(2\pi|x|s) s^{n/2}\, ds}{\delta} \xrightarrow{\delta\to 0} c(n)|x|^{-(n-2)/2}J_{(n-2)/2}(2\pi|x|)$$ However, to get $(3.41)$, we may have to show $$\lim_{\delta\to 0} \frac{\widehat{\chi_{B(0,1+\delta) \setminus B(0,1)}}(x)}{\delta} = \widehat{\sigma^{n-1}}(x)$$ How can we do that? Thank you!

  • I prove that to you in a previous posting to a question of yours. – Mittens May 31 '22 at 02:41
  • 1
    I think you are missing a factor $\delta^{-1}$. Recall that $\frac{1}{\delta}\int^{1+\delta}_1 h(s),ds\xrightarrow{\delta\rightarrow0}h(1)$ for $h$ continuous (or if $1$ is a Lebesgue point of $h$ more generally). – Mittens May 31 '22 at 03:32
  • Thanks! Please see my edit, I have divided by $\delta$ now. @OliverDíaz – stoic-santiago May 31 '22 at 06:29
  • The definition of $\hat{\sigma}$ is defined as a weak limit in the fashion of one of your previous postings – Mittens May 31 '22 at 10:24
  • @OliverDíaz Please see \begin{align}\lim_{\delta\to 0} \frac{\widehat{\chi_{B(0,1+\delta) \setminus B(0,1)}}(x)}{\delta} &= \lim_{\delta\to 0} \frac{1}{\delta} \int \chi_{B(0,1+\delta) \setminus B(0,1)}(z) e^{-2\pi ix\cdot z}, dz\ &= \lim_{\delta\to 0} \frac{1}{\delta} \int e^{-2\pi ix\cdot z}, d\mathcal L^n\vert_{B(0, 1+\delta)\setminus B(0,1)}(z)\ &{\color{red}{\stackrel{?}{=}}} \int e^{-2\pi ix\cdot \mathbf{u}}, d\sigma^{n-1}(\mathbf{u}) \&=\widehat{\sigma^{n-1}}(x) \end{align} How do you justify ${\color{red}{\stackrel{?}{=}}}$ Is $e^{-2\pi i x\cdot z} \in C_0(\Bbb R^n)$? – stoic-santiago May 31 '22 at 19:28
  • Notice that the result is local, you can modify the function smoothly outside a small shell around the unit sphere. – Mittens May 31 '22 at 19:50
  • Thanks! I understand the intuition. So, $h_1(z) := e^{-2\pi ix\cdot z}$ itself does not have compact support, but we can find modify it, say outside $B(0,2)$, to get a function, say $h_2$, of compact support, so that $$\int h_1(z) d\mathcal L^n\vert_{B(0, 1+\delta)\setminus B(0,1)}(z) = \int h_2(z) d\mathcal L^n\vert_{B(0, 1+\delta)\setminus B(0,1)}(z)$$ for all $\delta < 1$, right? Could you, if possible, help me with finding one such function $h_2$? – stoic-santiago May 31 '22 at 19:54
  • Exactly, notice that the limiting process only sees what is in the shell $1\leq |x|<1+\delta$, so, by multiplying by a radial mollifier supported in the shell, say $1/2\leq |x|<2$ that take value $1$ on $\frac34<|x|<\frac32$ you get what you need. – Mittens May 31 '22 at 20:10
  • @OliverDíaz This question may interest you! – stoic-santiago Jun 30 '22 at 20:39

0 Answers0