A dual method for minimizing a nonsmooth objective over one smooth inequality constraint (Q312667)

The authors consider a special class of convex optimization problems which contain a nonsmooth goal function and a single smooth constraint. They replace the feasible set with its inner approximation and take the dual problem at each iterate. Its sequential solution yields a proximal type method, which converges to a solution under the usual assumptions. The convergence rate \(O(1/k)\) is established. The results of computational tests and comparisons illustrate the performance of the proposed method.