Groups
ADMM splits a hard optimization problem into two easier subproblems that communicate through simple averaging-like steps.
A proximal operator pulls a point x toward minimizing a function f while penalizing how far it moves, acting like a denoiser or projector depending on f.