EUROPT 2024
Abstract Submission

303. On computing upper bounds in nonlinear problems involving disjunctive constraints

Invited abstract in session TC-7: Global optimization II / MINLP, stream Global optimization.

Thursday, 11:25 - 12:40
Room: M:I

Authors (first author is the speaker)

1. Sonia Cafieri
ENAC - Ecole Nationale d'Aviation Civile
2. Marcel Mongeau
ENAC

Abstract

The recently introduced continuous quadrant penalty formulation of logical disjunctive constraints constitutes a continuous-optimization alternative to the classical formulations (such as the bigM formulation) based on the introduction of binary variables.
We focus on nonlinear problems whose only combinatorial aspect comes from their logical constraints, and build on the continuous quadrant penalty, that yields to a continuous nonconvex problem, to derive an efficient computation of upper bounds to be used within Branch-and-Bound-based multi-step approaches.
We show on two problems, respectively from discrete geometry and arising in air traffic management optimization, that our approach is effective at speeding up the computational convergence.

Keywords

Status: accepted


Back to the list of papers