Mathematics > Optimization and Control
[Submitted on 22 Dec 2023]
Title:A Gradient-Based Optimization Method Using the Koopman Operator
View PDF HTML (experimental)Abstract:In this paper, we propose a novel approach to solving optimization problems by reformulating the optimization problem into a dynamical system, followed by the adaptive spectral Koopman (ASK) method. The Koopman operator, employed in our approach, approximates the evolution of an ordinary differential equation (ODE) using a finite number of eigenfunctions and eigenvalues. We begin by providing a brief overview of the Koopman operator and the ASK method. Subsequently, we adapt the ASK method for solving a general optimization problem. Moreover, we provide an error bound to aid in understanding the performance of the proposed approach, marking the initial step in a more comprehensive numerical analysis. Experimentally, we demonstrate the applicability and accuracy of our method across a diverse range of optimization problems, including min-max problems. Our approach consistently yields smaller gradient norms and higher success rates in finding critical points compared to state-of-the-art gradient-based methods. We also observe the proposed method works particularly well when the dynamical properties of the system can be effectively modeled by the system's behaviors in a neighborhood of critical points.
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.