Why Directly Learning Periodic Trajectories Can Fail

DOI10.48550/arXiv.2609.32254MaRDI QIDQ7360747arXivFDO

Authors Anita T. Layton, Kaixin Zheng, Anita Layton

Publication date 26 September 2026

Published in arXiv

Tldr We investigate why directly learning periodic trajectories can fail when varying system parameters; we explain through frequency mismatch, derive bounds on performance, and propose an effective workaround by learning period and waveform separately.

Full work available at URL https://doi.org/10.48550/arXiv.2609.32254

arXiv classification

cs.AImath.CAmath.NA

Cites work

• A neural network kernel decomposition for learning multiple steady states in parameterized dynamical systems
• An alternative unsteady adaptive stochastic finite element formulation based on interpolation at constant phase
• Asynchronous time integration for polynomial chaos expansion of uncertain periodic dynamics
• Belousov-Zhabotinsky type reactions: the non-linear behavior of chemical systems
• Curve Registration
• Data-driven model identification using forcing-induced limit cycles
• Deep Equilibrium Based Neural Operators for Steady-State PDEs
• Definition and data-driven reconstruction of asymptotic phase and amplitudes of stochastic oscillators via Koopman operator theory
• Fourier Neural Operator for Parametric Partial Differential Equations
• Functional data analysis of amplitude and phase variation
• Generative models for functional data using phase and amplitude separation
• Multilayer feedforward networks are universal approximators
• Neural Green's operators for parametric partial differential equations
• Regularisation of neural networks by enforcing Lipschitz continuity
• Reliable extrapolation of deep neural operators informed by physics or sparse observations
• Sensitivity and Nonlinearity of Thermoacoustic Oscillations
• Six decades of the FitzHugh-Nagumo model: A guide through its spatio-temporal dynamics and influence across disciplines
• Surrogate Models for Oscillatory Systems Using Sparse Polynomial Chaos Expansions and Stochastic Time Warping

This page was built for publication: Why Directly Learning Periodic Trajectories Can Fail

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7360747)