Augmenting Physical Models with Deep Networks for Complex Dynamics Forecasting

Abstract

Forecasting complex dynamical phenomena in settings where only partial knowledge of their dynamics is available is a prevalent problem across various scientific fields. While purely data-driven approaches are arguably insufficient in this context, standard physical modeling based approaches tend to be over-simplistic, inducing non-negligible errors. In this work, we introduce the APHYNITY framework, a principled approach for augmenting incomplete physical dynamics described by differential equations with deep data-driven models. It consists in decomposing the dynamics into two components: a physical component accounting for the dynamics for which we have some prior knowledge, and a data-driven component accounting for errors of the physical model. The learning problem is carefully formulated such that the physical model explains as much of the data as possible, while the data-driven component only describes information that cannot be captured by the physical model, no more, no less. This not only provides the existence and uniqueness for this decomposition, but also ensures interpretability and benefits generalization. Experiments made on three important use cases, each representative of a different family of phenomena, i.e. reaction-diffusion equations, wave equations and the non-linear damped pendulum, show that APHYNITY can efficiently leverage approximate physical models to accurately forecast the evolution of the system and correctly identify relevant physical parameters. Code is available at https://github.com/yuan-yin/APHYNITY .

arXiv commentAccepted at ICLR 2021 (Oral)

References in corpus (14)

• Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting
• Discovering governing equations from data: Sparse identification of nonlinear dynamical systems
• DGM: A deep learning algorithm for solving partial differential equations
• Unsupervised Learning of Video Representations using LSTMs
• Hamiltonian Neural Networks
• PredRNN++: Towards A Resolution of the Deep-in-Time Dilemma in Spatiotemporal Predictive Learning
• Deep Learning for Physical Processes: Incorporating Prior Scientific Knowledge
• PDE-Net: Learning PDEs from Data
• A critical review of statistical calibration/prediction models handling data inconsistency and model inadequacy
• Symplectic Recurrent Neural Networks
• Learning Dynamical Systems from Partial Observations
• HybridNet: Integrating Model-based and Data-driven Learning to Predict Evolution of Dynamical Systems
• Recurrent Kalman Networks: Factorized Inference in High-Dimensional Deep Feature Spaces
• PDE-Driven Spatiotemporal Disentanglement

Cited by in corpus (11)

• A Review of Machine Learning Methods Applied to Structural Dynamics and Vibroacoustic
• Discovering Sparse Interpretable Dynamics from Partial Observations
• Artificial Intelligence for Science in Quantum, Atomistic, and Continuum Systems
• Continuous PDE Dynamics Forecasting with Implicit Neural Representations
• Learning complexity to guide light-induced self-organized nanopatterns
• Neural Dynamical Systems: Balancing Structure and Flexibility in Physical Prediction
• Physics-Integrated Variational Autoencoders for Robust and Interpretable Generative Modeling
• Deep KKL: Data-driven Output Prediction for Non-Linear Systems
• Guiding Video Prediction with Explicit Procedural Knowledge
• CONFIDE: Contextual Finite Differences Modelling of PDEs
• A Grid-Structured Model of Tubular Reactors