Discovering governing equations from data by sparse identification of nonlinear dynamical systems
By Steven L. Brunton, Joshua L. Proctor, J. Nathan Kutz. 2015
5,211 citations recorded by the source corpus.
Significance Understanding dynamic constraints and balances in nature has facilitated rapid development of knowledge and enabled technology, including aircraft, combustion engines, satellites, and electrical power. This work develops a novel framework to discover governing equations underlying a dynamical system simply from data measurements, leveraging advances in sparsity techniques and machine learning. The resulting models are parsimonious, balancing model complexity with descriptive ability while avoiding overfitting. There are many critical data-driven problems, such as understanding cognition from neural recordings, inferring climate patterns, determining stability of financial markets, predicting and suppressing the spread of disease, and controlling turbulence for greener transportation and energy. With abundant data and elusive laws, data-driven discovery of dynamics will continue to play an important role in these efforts. Extracting governing equations from data is a central challenge in many diverse areas of science and engineering. Data are abundant whereas models often remain elusive, as in climate science, neuroscience, ecology, finance, and epidemiology, to name only a few examples. In this work, we combine sparsity-promoting techniques and machine learning with nonlinear dynamical systems to discover governing equations from noisy measurement data. The only assumption about the structure of the model is that there are only a few important terms that govern the dynamics, so that the equations are sparse in the space of possible functions; this assumption holds for many physical systems in an appropriate basis. In particular, we use sparse regression to determine the fewest terms in the dynamic governing equations required to accurately represent the data. This results in parsimonious models that balance accuracy with model complexity to avoid overfitting. We demonstrate the algorithm on a wide range of problems, from simple canonical systems, including linear and nonlinear oscillators and the chaotic Lorenz system, to the fluid vortex shedding behind an obstacle. The fluid example illustrates the ability of this method to discover the underlying dynamics of a system that took experts in the community nearly 30 years to resolve. We also show that this method generalizes to parameterized systems and systems that are time-varying or have external forcing.
Subfields: Artificial Intelligence, Control and Systems Engineering, Statistical and Nonlinear Physics, Causal Inference & Discovery, AI for Physical Sciences
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- G. Em Karniadakis
- Steven L. Brunton
- Paris Perdikaris
- Maizar Raissi
- J. Nathan Kutz
- Yang Liu
- Liu Yang
- Lu Lu
In Control and Systems Engineering
- Steven L. Brunton
- J. Nathan Kutz
- G. Em Karniadakis
- Maizar Raissi
- Paris Perdikaris
- Eurika Kaiser
- Joshua L. Proctor
- Bingni W. Brunton
In Statistical and Nonlinear Physics
- Steven L. Brunton
- G. Em Karniadakis
- J. Nathan Kutz
- Maizar Raissi
- Paris Perdikaris
- Bingni W. Brunton
- Yang Liu
- Lu Lu
In Causal Inference & Discovery
- Steven L. Brunton
- Maizar Raissi
- J. Nathan Kutz
- Stefan Bauer
- G. Em Karniadakis
- Niklas Pfister
- Paris Perdikaris
- Joshua L. Proctor
In AI for Physical Sciences
- G. Em Karniadakis
- Maizar Raissi
- Paris Perdikaris
- Steven L. Brunton
- J. Nathan Kutz
- Jens Berg
- Kaj Nyström
- Lu Lu
Citations in the corpus
- Interpretable Machine Learning: Fundamental Principles and 10 Grand Challenges (2021, methodology): “Several additional works [167, 43, 259] used sparse regression and dynamic mode decomposition to discover the governing equa-”
- Learning stable and predictive structures in kinetic systems (2018, methodology): “, 2014, Brunton et al., 2016] and can also be used as screening procedures in our method. Furthermore, we also compare with a more involved method called adaptive integral matching (AIM) introduced by Mikkelsen and Hansen [2017]. Rather than only fitting the target equation it fits an entire system ”
- fPINNs: Fractional Physics-Informed Neural Networks (2018, background): “In this paper, we focus on the identification of the parameters of fractional PDEs whose overall form is known but some coefficients and most importantly the operators themselves are not known.”
- Deep Hidden Physics Models: Deep Learning of Nonlinear Partial Differential Equations (2018, methodology): “More recently, sparsity [23] has been used to determine the governing dynamical system [24, 25, 26, 27, 28, 29, 30, 31, 32, 33].”
- Multistep Neural Networks for Data-driven Discovery of Nonlinear Dynamical Systems (2018, methodology): “More recently, approaches based on symbolic regression [15], sparse regression, and compressive sensing [16, 17] were able to go beyond estimating a black-box approximation of the dynamics given by f , and return more interpretable models that can uncover the full parametric form of an underlying…”
- Data-driven discovery of partial differential equations (2016, methodology): “Both the evolutionary (14) and sparse (16) symbolic regression methods avoid overfitting by selecting parsimonious models that balance model accuracy with complexity via Pareto analysis.”
- Physics Informed Deep Learning (Part II): Data-driven Discovery of Nonlinear Partial Differential Equations (2017, methodology): “This is the key highlight of our discrete time formulation for identification problems, setting it apart from competing approaches [7, 13].”
- PDE-Net: Learning PDEs from Data (2017, background): “Recently, Brunton et al. (2016), Schaeffer (2017), Rudy et al. (2017) and Wu & Zhang (2017) propose an alternative approach using sparse regression.”
- Machine learning of linear differential equations using Gaussian processes (2017, background): “However, regularization still remains an important factor even in cases where data is abundant as seen in the recently growing literature on discovering ordinary and partial differential equations from data using sparse regression techniques [7,41].”
- Integrating Scientific Knowledge with Machine Learning for Engineering and Environmental Systems (2020, methodology): “More recently, works have used sparse regression built on a dictionary of functions and partial derivatives to construct governing equations [43, 220, 234].”
- Inferring Biological Networks by Sparse Identification of Nonlinear Dynamics (2016, methodology): “This method has been recently used to successfully infer nonlinear dynamical systems [11].”
- Machine learning and applications in ultrafast photonics (2020, background): “There are of course many ways machine-learning techniques can be exploited, and we discuss below some possible future directions of research and challenges to overcome.”
- Machine Learning of Space-Fractional Differential Equations (2018, methodology): “In [5], a more parametric method for learning nonlinear autonomous systems was developed in which the candidate equation is build from a linear combination of elements from a user-defined “dictionary.”
- Extended Physics-informed Neural Networks (XPINNs): A Generalized Space-Time Domain Decomposition based Deep Learning Framework for Nonlinear Partial Differential Equations (2020, background): “[34] proposed the SINDy framework for dictionary learning of dynamical systems.”
- SINDy-PI: a robust algorithm for parallel implicit sparse identification of nonlinear dynamics (2020, methodology): “It is also possible to sweep through the parameter λ and candidate functions θj , and then choose the Pareto optimal model from a family of models on the Pareto front balancing accuracy and efficiency; this is the approach in the original SINDy work [28] and in earlier work leveraging genetic progra”