Methods for Non-smooth Non-convex Optimization
摘要
This chapter deals with the task of minimizing a class of locally Lipschitz functions over closed sets. The focus is given to local-solution methods for structured non-smooth and non-convex problems. Structures of interest include Difference-of-Convex (DC) and difference of Convex and weakly Convex (CwC) optimization problems. The chapter starts by revisiting the generalized Frank-Wolfe method, passing through the DC algorithm to more sophisticated bundle methods for problems whose objective and non-linear constraint can be expressed as CwC functions.