Fourier synthesis via partially finite convex programming

A dual algorithm for problems of Fourier Synthesis is proposed. Partially finite convex programming provides tools for a formulation which enables to elude static pixelization of the object to be reconstructed. This leads to a regularized reconstruction-interpolation formula for problems in which finitely many and possibly irregularly spaced samples of the Fourier transform of the unknown object are known, as is the case in Magnetic Resonance Imaging with non-Cartesian and sparse acquisitions.