Title:Nested Primal-dual Gradient Algorithms for Distributed Constraint-coupled Optimization

Abstract:We study a class of general distributed constraint-coupled optimization problems. A novel nested primal-dual gradient algorithm (NPGA) is designed from the dual perspective, which can achieve linear convergence under the weakest condition and the corresponding theoretical convergence rate is better than all existing results. More importantly, NPGA is not only an algorithm but also a unified algorithmic framework, which can cover existing algorithms as special cases. By designing different network matrices, we can conveniently obtain many variants of NPGA, which offers us a chance to design more efficient algorithms. Finally, the convergence rates of NPGA and existing algorithms are compared in numerical experiments.