We address the problem of minimizing objectives from the class of piecewise differentiable functions whose nonsmoothness can be encapsulated in the absolute value function. They possess local piecewise linear approximations with a discrepancy that can be bounded by a quadratic proximal term. This overestimating local model is continuous but generally nonconvex. It can be generated in its abs-normal form by a minor extension of standard algorithmic differentiation tools. Here we demonstrate how the local model can be minimized by a bundle type method, which benefits from the availability of additional gray-box information via the abs-normal form. In the convex case our algorithm realizes the consistent steepest descent trajectory for which finite convergence was established in the book by Hiriart-Urruty and Lemarechal, specifically covering counter examples where steepest descent with exact line-search famously fails. The analysis of the abs-normal representation and the design of the optimization algorithm is geared towards the general case, whereas the convergence proof so far only covers the convex case.
Humboldt-Universität zu Berlin, June 2014