Discovering Sparse Interpretable Dynamics from Partial Observations

Abstract

Identifying the governing equations of a nonlinear dynamical system is key to both understanding the physical features of the system and constructing an accurate model of the dynamics that generalizes well beyond the available data. We propose a machine learning framework for discovering these governing equations using only partial observations, combining an encoder for state reconstruction with a sparse symbolic model. Our tests show that this method can successfully reconstruct the full system state and identify the underlying dynamics for a variety of ODE and PDE systems.

10 pages, 6 figures (4 main text, 2 supplemental)

References in corpus (5)

• Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges
• Augmenting Physical Models with Deep Networks for Complex Dynamics Forecasting
• Model selection of chaotic systems from data with hidden variables using sparse data assimilation
• Automatic Differentiation to Simultaneously Identify Nonlinear Dynamics and Extract Noise Probability Distributions from Data
• OccamNet: A Fast Neural Model for Symbolic Regression at Scale

Cited by in corpus (5)

• Discovering Conservation Laws using Optimal Transport and Manifold Learning
• Interpretable structural model error discovery from sparse assimilation increments using spectral bias-reduced neural networks: A quasi-geostrophic turbulence test case
• Challenges in identifying simple pattern-forming mechanisms in the development of settlements using demographic data
• CTSR: Cartesian tensor-based sparse regression for data-driven discovery of high-dimensional invariant governing equations
• Data-driven discovery of dynamics from time-resolved coherent scattering