The Fourier transform of multiradial functions (Q405732): Difference between revisions

Let \(n_1, n_2,...,n_m \geq 1\) \((m \geq 1)\) be integers. The Fourier transform of a function \(\Phi\) in \(L^1(\mathbb{R}^{n_1+...+n_m})\) is defined as \[ F_{n_1,...,n_m}(\Phi)(\xi_1,...,\xi_m)=\int_{\mathbb{R}^{n_m}}...\int_{\mathbb{R}^{n_1}}\Phi(x_1,..x_m)e^{-2\pi i(x_1.\xi_1+...+x_m.\xi_m)}dx_1...dx_m \] The function \(\Phi\) is a multiradial function, if there exists some function \(\phi\) on \((\mathbb{R}^{+} \cup \{0\})^{m}\) such that \[ \Phi(x_1,...,x_m) = \phi(|x_1|,...,|x_m|)\;\text{for all}\;x_i \in \mathbb{R}^{n_i}, \] where \(|x_j|\) denotes the Euclidean norm of \(x_j\). \textit{L. Grafakos} and \textit{G. Teschl} [``On Fourier transforms of radial functions and distributions'', J. Fourier Anal. Appl. 19, 167--179 (2013)] proved an explicit formula for the Fourier transform of a radial function \(\Phi(x) = \phi(|x|)\) in terms of the one-dimensional Fourier transform of \(\phi\) or the two-dimensional Fourier transform of \((t, s) \rightarrow \phi(|(t, s)|)\). In the paper under review, the authors extend the formula to multiradial functions.