VOPy: A Framework for Black-box Vector Optimization

  • 2024-12-09 15:53:02
  • Yaşar Cahit Yıldırım, Efe Mert Karagözlü, İlter Onat Korkmaz, Çağın Ararat, Cem Tekin
  • 0

Abstract

We introduce VOPy, an open-source Python library designed to addressblack-box vector optimization, where multiple objectives must be optimizedsimultaneously with respect to a partial order induced by a convex cone. VOPyextends beyond traditional multi-objective optimization (MOO) tools by enablingflexible, cone-based ordering of solutions; with an application scope thatincludes environments with observation noise, discrete or continuous designspaces, limited budgets, and batch observations. VOPy provides a modulararchitecture, facilitating the integration of existing methods and thedevelopment of novel algorithms. We detail VOPy's architecture, usage, andpotential to advance research and application in the field of vectoroptimization. The source code for VOPy is available athttps://github.com/Bilkent-CYBORG/VOPy.

 

Quick Read (beta)

loading the full paper ...