Convergence Analysis of Douglas--Rachford Splitting Method for “Strongly + Weakly” Convex Programming

We consider the convergence of the Douglas--Rachford splitting method (DRSM) for minimizing the sum of a strongly convex function and a weakly convex function; this setting has various applications, especially in some sparsity-driven scenarios with the purpose of avoiding biased estimates which usually occur when convex penalties are used. Though the convergence of the DRSM has been well studied for the case where both functions are convex, its results for some nonconvex-function-involved cases, including the “strongly + weakly” convex case, are still in their infancy. In this paper, we prove the convergence of the DRSM for the “strongly + weakly” convex setting under relatively mild assumptions compared with some existing work in the literature. Moreover, we establish the rate of asymptotic regularity and the local linear convergence rate in the asymptotical sense under some regularity conditions. (A corrected version is attached.)