“…A full-splitting, ADMM algorithm was proposed in (Bot et al 2019), exploiting the proximal mapping of g, h, and the linear operator K, and the gradient ∇f (x, y), separately. The convergence of the proposed algorithm requires that K is full row rank (surjective), a common assumption shared by other ADMM algorithms for dealing with nonsmooth composite functions, see e.g., Pong 2015, Sun et al 2019).…”