Fractional spectral graph wavelet transform approximation in terms of modulus of continuity

From MaRDI portal





First the definition of several generalizations of the Fourier transform are recalled. The classical Fourier transform uses a basis of complex exponentials that are the eigenfunctions of the continuous Laplace operator. The fractional Fourier transform (FRFT) rotates the signal in the time-frequency plane over an angle \(\theta\) which need not be \(\pi/2\) as in the classical transform. In the (spectral) graph Fourier transform (SGFT), the function is given as values at the vertices of a graph. This is a kind of discrete transform where the vector of function values is expressed as a linear combination of the (unitary) eigenvectors of the (self-adjoint) Laplace matrix of the graph. The combination of both generalizations gives the spectral graph fractional Fourier transform (SGFRFT).\N\NA similar generalization can be defined for the wavelet transform where the Fourier basis is replaced by a wavelet frame defined by a generating wavelet that is dilated and a translated. Eventually this gives rise to the spectral graph fractional wavelet transform. (SGFRWT).\N\NBesides introducing this SGFRWT, the purpose of the paper is to derive several of its properties from which an upper and lower bound is obtained for the approximation error when the function \(f\in L^2(\mathbb{R})\) is approximated by a polynomial. These bounds depend on the modulus of continuity for the function.\N\NThe paper is concise and mixes many concepts. It contains some proofs, but also relies often on the literature. Unfortunately it has some typos which however do not influence the results.











This page was built for publication: Fractional spectral graph wavelet transform approximation in terms of modulus of continuity

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6969346)