Sparse deconvolution of seismic data with a regularized norm ratio Audrey Repetti, Mai Quyen Pham, Laurent Duval, Emilie Chouzenoux, Jean-Christophe Pesquet Sparse blind seismic deconvolution aims at jointly estimating an unknown sparse signal (reflectivity) and an unknown impulse response (seismic wavelet). The main difficulty stems from the non-uniqueness of the solutions. They may be regularized by sparsity enforcing norm ratios (Gray, 1978) which are non convex. In this work, we propose an alternating preconditioned method, based on forward-backward iterations, to solve this type of problem, involving a regularized norm ratio, assorted with theoretical convergence results.