Variational learning of Euler-Lagrange dynamics from data

DOI10.1016/J.CAM.2022.114780zbMATH Open1498.65222arXiv2112.12619OpenAlexW4294946167MaRDI QIDQ2095136FDOQ2095136

Christian Offen, Sina Ober-Blöbaum

Publication date: 9 November 2022

Published in: Journal of Computational and Applied Mathematics (Search for Journal in Brave)

Abstract: The principle of least action is one of the most fundamental physical principle. It says that among all possible motions connecting two points in a phase space, the system will exhibit those motions which extremise an action functional. Many qualitative features of dynamical systems, such as the presence of conservation laws and energy balance equations, are related to the existence of an action functional. Incorporating variational structure into learning algorithms for dynamical systems is, therefore, crucial in order to make sure that the learned model shares important features with the exact physical system. In this paper we show how to incorporate variational principles into trajectory predictions of learned dynamical systems. The novelty of this work is that (1) our technique relies only on discrete position data of observed trajectories. Velocities or conjugate momenta do {em not} need to be observed or approximated and {em no} prior knowledge about the form of the variational principle is assumed. Instead, they are recovered using backward error analysis. (2) Moreover, our technique compensates discretisation errors when trajectories are computed from the learned system. This is important when moderate to large step-sizes are used and high accuracy is required. For this, we introduce and rigorously analyse the concept of inverse modified Lagrangians by developing an inverse version of variational backward error analysis. (3) Finally, we introduce a method to perform system identification from position observations only, based on variational backward error analysis.

Full work available at URL: https://arxiv.org/abs/2112.12619

zbMATH Keywords

variational integratorsmodified LagrangianLagrangian learningphysics informed learningvariational backward error analysis

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05) Numerical methods for Hamiltonian systems including symplectic integrators (65P10) System identification (93B30)

Cites Work

• Symplectic Gaussian process regression of maps in Hamiltonian systems
• Discrete mechanics and variational integrators
• Numerical integrators based on modified differential equations
• Backward error analysis for variational discretisations of PDEs
• Modified equations for variational integrators
• Sympnets: intrinsic structure-preserving symplectic networks for identifying Hamiltonian systems

Cited In (5)

• Learning discrete Lagrangians for variational PDEs from data and detection of travelling waves
• Learning the Dynamics of Objects by Optimal Functional Interpolation
• Backward error analysis for conjugate symplectic methods
• Learning of discrete models of variational PDEs from data
• Learning Hamiltonian systems with mono-implicit Runge-Kutta methods

Uses Software

• GitHub
• SympGPR

This page was built for publication: Variational learning of Euler-Lagrange dynamics from data

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2095136)