Reproduce “One-shot learning for the complex dynamical behaviors of weakly nonlinear forced oscillators” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Teng Ma, Luca Rosafalco, Wei Cui, Lin Zhao, Attilio Frangi
Researchers at Tongji University and Politecnico di Milano developed MEv-SINDy, a one-shot learning method, to identify explicit governing equations for weakly nonlinear forced oscillators. This approach accurately predicts global frequency-response curves from a single excitation time history, enabling extrapolative analysis of complex dynamics in systems like MEMS with significantly reduced data requirements.
Key insights
- Transforming the dynamics of weakly nonlinear forced oscillators into an evolutionary domain via GHB allows sparse regression methods to effectively identify critical, low-magnitude nonlinear terms often obscured in raw time-domain data.
- One-shot learning, which infers global system-level behaviors from a single excitation time history, is a viable strategy for data-efficient scientific machine learning in complex dynamical systems.
- Identifying explicit governing equations provides strong physical interpretability and enables superior extrapolative prediction capabilities for system behavior compared to purely data-driven, black-box models.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.