As is well known, the smoothness of a function of one variable is related to the behaviour of its Fourier transform for arbitrarily large frequency. Specifically (see here), if a function $f(t)$ is in $C^m$ its Fourier transform $\hat f(\omega)$ is $O(\omega^{-(m+2)})$ for $\omega \rightarrow \infty$.
Is there a similar result for the two-dimensional Fourier transform?
I am only interested in radial functions, for which the two-dimensional Fourier transform can be expressed as a Hankel transform. So a result for the latter would be useful too; in fact more so. Also, my initial function has bounded support.
For example:
- The "unit cylinder" function (which equals $1$ in the unit circle and is $0$ outside) is discontinuous. Its Fourier transform is $2\pi J_1(\omega_\mathrm{r}) / \omega_\mathrm{r}$, where $\omega_\mathrm{r}$ is the "radial frequency", and so it decays like $O(\omega_\mathrm{r}^{-3/2})$.
- Replacing the cylinder by a cone, which makes the function continuous, gives a (more complicated) transform that seems to decay like $O(\omega_\mathrm{r}^{-5/2})$.
So, also in two dimensions there seems to be a faster asymptotic decrease rate of the transform as the original function is smoother.