Stochastic Optimal Control via Local Occupation Measures

Viewing stochastic processes through the lens of occupation measures has proved to be a powerful angle of attack for the theoretical and computational analysis of a wide range of stochastic optimal control problems. We present a simple modification of the traditional occupation measure framework der...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Holtorf, Flemming, Edelman, Alan, Rackauckas, Christopher
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Viewing stochastic processes through the lens of occupation measures has proved to be a powerful angle of attack for the theoretical and computational analysis of a wide range of stochastic optimal control problems. We present a simple modification of the traditional occupation measure framework derived from resolving the occupation measures locally on a partition of the control problem's space-time domain. This notion of local occupation measures provides fine-grained control over the construction of structured semidefinite programming relaxations for a rich class of stochastic optimal control problems with embedded diffusion and jump processes via the moment-sum-of-squares hierarchy. As such, it bridges the gap between discretization-based approximations to the solution of the Hamilton-Jacobi-Bellmann equations and approaches based on convex optimization and the moment-sum-of-squares hierarchy. We demonstrate with examples that this approach enables the computation of high quality bounds on the optimal value for a large class of stochastic optimal control problems with notable performance gains relative to the traditional occupation measure framework.
DOI:10.48550/arxiv.2211.15652