[1] introduces a new class of proximal splitting algorithms that differ from existing methods in that they are both block-iterative and asynchronous, and allow for individual conditioning of the functions. We present the first numerical applications of the method to imaging, and discuss how its flexibility leads to extremely fast implementations. Extensive numerical simulations are provided. Joint work with L. Glaudin. [1] P. Combettes and J. Eckstein, Asynchronous block-iterative decomposition.., Mathematical Programming, 2017.

This presentation is part of Contributed Presentation “CP1 - Contributed session 1”

Authors:

Patrick Combettes (North Carolina State University)

Keywords:

image reconstruction, nonlinear optimization