Continuous wavelet transform and uncertainty principle related to the spherical mean operator (Q523674)

For any \(f\in\mathcal{C}_e(\mathbb{R}\times\mathbb{R}^n)\) (the spaces of continuous functions on \(\mathbb{R}\times\mathbb{R}^n\), even with respect to the first variable) and \((r,x)\in[0,\infty)\times\mathbb{R}^n\), the spherical mean operator is defined by \[ \mathcal{R}(f)(r,x):=\int_{S^n}f(r\eta,x+r\varphi)\,d\sigma_n(\eta,\varphi), \] where \(S^n\) is the unit sphere in \(\mathbb{R}^{n+1}\) and \(d\sigma_n\) is the normalized surface measure on \(S^n\). Let \(j_{\frac{n-1}2}\) be the modified Bessel function. Then, for any \(f\in L^1(d\nu)\) and \((\mu,\lambda)\in[0,\infty)\times\mathbb{R}^n\), \[ \widetilde{\mathcal{F}}(f)(\mu,\lambda):=\int_0^{\infty}\int_{\mathbb{R}^n} f(r,x)j_{\frac{n-1}2}(r\mu)e^{-i\lambda x}\,d\nu(r,x). \] A function \(\psi\in L^2(d\nu)\) is called an admissible wavelet if \[ 0<C_{\psi}:=c_{\frac{n-1}2}\int_0^{\infty}\int_{\mathbb{R}^n} \left|\widetilde{\mathcal{F}}(\psi)(\mu,\lambda_1, \cdots,\lambda_n)\right|^2\frac{d\mu}{\mu} \frac{d\lambda_1}{\lambda_1}\cdots\frac{d\lambda_n} {\lambda_n}<\infty, \] where \(c_{\frac{n-1}2}:=\frac1{2^{\frac{n-1}2} \Gamma(\frac{n+1}2)(2\pi)^{\frac{n}2}}\). Let \(D\) and \(\tau\) be the dilation operator and translation operator, respectively. For any admissible wavelet \(\psi\) and \(f\in L^2(d\nu)\), the continuous wavelet transform \(T_\psi\) associated with the sphere mean operator is defined by \[ T_\psi(f)(a,b,r,x):=\int_0^{\infty}\int_{\mathbb{R}^n}f(s,y) \overline{\psi_{a,b,r,x}(s,y)}\,d\nu(s,y), \] where \(\psi_{a,b,r,x}:=\tau_{(r,-x)}(D_{a,b}(\psi))\). In this paper, the authors prove the Plancherel theorem, the Parseval theorem and the reconstruction formula for \(T_\psi\). Precisely, they obtain the following conclusion. Let \(\psi\) be an admissible wavelet, then the following results hold true: (i) For any \(f\in L^2(d\nu)\), \[ \int_0^{\infty}\int_{\mathbb{R}^n}|f(r,x)|^2\,d\nu(r,x) \] is equal to \[ \frac1{C_\psi}\int_0^{\infty}\int_{\mathbb{R}^n}\int_0^{\infty} \int_{\mathbb{R}^n}|T_\psi(f)(a,b,r,x)|^2\,d\mu((a,b),(r,x)). \] (ii) For any \(f\) and \(g\in L^2(d \nu)\), \[ \int_0^{\infty}\int_{\mathbb{R}^n}f(r,x)\overline{g(r,x)}\,d\nu(r,x) \] is equal to \[ \frac1{C_\psi}\int_0^{\infty}\int_{\mathbb{R}^n}\int_0^{\infty} \int_{\mathbb{R}^n}T_\psi(f)(a,b,r,x)\overline{T_\psi(g)(a,b,r,x)} \,d\mu((a,b),(r,x)). \] (iii) Let \(\psi\) be an admissible wavelet such that \(|\psi|\) is also an admissible wavelet in \(L^2(\nu)\). Then, for any \(f\in L^2(\nu)\), \[ f(\cdot,\cdot) =\frac1{C_\psi}\int_0^{\infty}\int_{\mathbb{R}^n}\int_0^{\infty} \int_{\mathbb{R}^n}T_\psi(f)(a,b,r,x) \psi_{a,b,r,x}(\cdot,\cdot) \,d\mu((a,b),(r,x)) \] weakly in \(L^2(\nu)\). They also establish the Heisenberg uncertainty principle for the continuous wavelet transform and estimate the concentration of \(T_\psi\) on \((\mathbb{R}_+\times\mathbb{R}^n)^2\) of finite measure. Moreover, they obtain the local type uncertainty inequality for \(T_\psi\) and a general form of Heisenberg type inequality.