Scaling functions generating fractional Hilbert transforms of a wavelet function (Q500103)

The authors define the fractional Hilbert transform by \(\mathcal H_c= \cos(c\pi ) I + \sin(c\pi)\mathcal H\) where \(\mathcal H\) is the standard Hilbert transform and \(c\in \mathbb R\). Furthermore they define a modified translation operator \(T^\dagger_c\) using Fourier transforms by \(\widehat {T^\dagger_c f}(\xi)= \mathrm{e}^{-\mathrm{i} c\tau(\xi)}\hat f(\xi)\) where \(\tau(\xi)= \xi\) when \(|\xi| \leq 2\pi\), \(\tau(\xi)= \tau(\xi+ 2\pi)\) when \(\xi <-2\pi\) and \(\tau(\xi)= \tau(\xi-2\pi)\) when \(\xi >2\pi\). Then they show that if \(\phi\) is the scaling function of a multiresolution analysis and \(\psi\) is the wavelet function naturally associated with \(\phi\), then \(T^\dagger_c \phi\) is a scaling function and the wavelet function naturally associated with it is \(\mathcal H_c \psi\). As a special case they thus get the result that if the support of \(\hat \phi\) is contained in \([-2\pi,2\pi]\), then \(T_c\phi\) (with \((T_c\phi)(x)=\phi(x-c)\)) is a scaling function and the wavelet function naturally associated with it is \(H_c\phi\).