Iterative Fourier Reconstruction
Stefan Kunis (Univ. Lübeck)
(joint work with Daniel Potts)
Without doubt, the fast Fourier transform belongs to most popular algorithms in
applied mathematics. Due to the fact that sampling nodes in many
applications are not equally spaced, these fast algorithms for the particular
matrix-vector-multiplication with the Fourier matrix
have been generalised for arbitrary nodes .
Here, we consider consistent linear systems and their minimal norm solutions,
We show that the condition number of A is uniformly bounded for norms
with smooth weights under the condition that the sampling nodes
are separated with respect to the polynomial degree N.