EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

3054. A bundle-type method based on descent-ascent directions for solving nonsmooth DC programs

Invited abstract in session TC-41: Subgradient-based methods, stream Nonsmooth Optimization.

Tuesday, 12:30-14:00
Room: 97 (building: 306)

Authors (first author is the speaker)

1. Giovanni Giallombardo
DIMES, University of Calabria
2. Manlio Gaudioso
DIMES, University of Calabria
3. Giovanna Miglionico
DIMES, Università della Calabria

Abstract

We present a bundle method for the unconstrained minimization of nonsmooth difference-of-convex functions, which is based on the calculation of descent-ascent directions. In fact, we define a descent-ascent direction as a direction which is expected to simultaneously provide a descent for the minuend component function, and an increase for the subtrahend component function. The algorithm only requires evaluations of the minuend component function at each iterate-point, and it can be considered as a parsimonious bundle method, as accumulation of information takes place only in case the descent-ascent direction does not provide a sufficient decrease. Hence, bundle-resets take place every time a sufficient decrease in the objective function is achieved. No line search is performed, while proximity control is pursued independent of whether the decrease in the objective function is achieved. Computational performance of the algorithm, which is proved to terminate at a point satisfying a weak criticality condition, are reported relative to a set of benchmark DC instances, ranging from small to large size.

Keywords

Status: accepted


Back to the list of papers