Pareto Front Tracing in Higher Dimensions Using Numerical Integration
Lara Löhken, Matthias Bolten, Hanno Gottschalk, Kathrin Klamroth
Abstract
We propose a numerical integration approach for tracing the Pareto front of unconstrained continuous multi-objective optimization problems. By stating optimality conditions for weighted sum scalarizations, we derive parametric equations for Pareto critical points, parameterized by the weights. Sensitivity analysis with respect to the weights yields a system of partial differential equations (PDEs). We prove local existence and uniqueness of solutions and derive conditions under which a solution extends to the entire weight space and yields Pareto optimal solutions. A tracing strategy for the resulting PDE system is developed that uses numerical integration along individual weight coordinates using Runge--Kutta methods. We provide an error analysis considering inexact initial solutions, perturbations of the right-hand sides, and numerical integration errors. We investigate the accuracy and computational efficiency of the tracing strategy on convex quadratic problems and WFG examples, where the latter demonstrate the applicability to problems with Pareto fronts consisting of convex and non-convex parts. Comparisons with repeated weighted sum scalarizations and a predictor--corrector method illustrate the computational benefits of the proposed strategy.
Create a lesson
Related papers
Optimal Scheduling of Multi-Site AI Data Centers in Day-Ahead Markets
Xinyang Zhou, Guido Cavraro
Minimization of free energy in a Poisson--Boltzmann problem
Shahid Ali, Klemen Bohinc, Angeles Martinez et al.
Lyapunov Theory for Output-to-State Stability of Nonlinear Infinite-Dimensional Systems
Qiaoling Chen, Andrii Mironchenko, Fabian Wirth
Semi-global Convergence and Finite-Sample Error Bounds for Model Predictive Path Integral Control
Christian Fiedler, Giacomo Borghi
When Suspicion Adds Up: A Stochastic Framework for Optimizing Honeypot-Based Insider-Threat Detection
Saharnaz Mehrani, Lujia Zhan, Mohsen Emadikhiav et al.
Hamiltonian Potential Optimization for Partially Observed Stochastic Optimal Control
Christian Bayer, Saifeddine Ben Naamia, Erik von Schwerin et al.