Two-direction subgradient method for non-differentiable optimization problems (Q581239)

The authors consider the problem of minimizing a convex, not necessarily differentiable real function in \(R^ n\) with a `known' optimal objective value. Based on the analysis of the projection onto the intersection of two half spaces, the authors propose an algorithm using two subgradients at a time. Numerical examples show that this new algorithm converges faster than the classical subgradient method.