97. Derivative-free Penalty-IPM for nonsmooth constrained optimization
Invited abstract in session MD-1: Derivative-Free Optimization Methods for challenging applications: Handling Nonsmoothness and Constraints, stream Zeroth and first-order optimization methods.
Monday, 16:30-18:30Room: B100/1001
Authors (first author is the speaker)
| 1. | Andrea Brilli
|
| Department of Computer Control and Management Engineering “A. Ruberti”, Sapienza University of Rome | |
| 2. | Youssef Diouane
|
| Mathematics and Industrial Engineering, Polytechnique Montréal | |
| 3. | Sébastien Le Digabel
|
| Polytechnique Montréal | |
| 4. | Giampaolo Liuzzi
|
| DIAG - Sapienza University of Rome | |
| 5. | Christophe Tribes
|
| Mathematics and Industrial Engineering, Polytechnique Montréal |
Abstract
We propose an Penalty-Interior Point method to solve nonsmooth constrained optimization problems. The unconstrained subproblems are solved using a derivative-free Mesh Adaptive Direct Search approach. Assuming local Lipschitz continuity of the limit points, we prove various stationarity properties using regularity conditions for the constraints. The approach has been implemented into the NOMAD solver, and we have performed experiments to compare the new strategy against existing ones.
Keywords
- Derivative-free optimization
- Non-smooth optimization
- Black-box optimization
Status: accepted
Back to the list of papers