Multistep Neural Networks for Data-driven Discovery of Nonlinear Dynamical Systems
This addresses the challenge of automated model discovery from data in science and engineering, though it appears incremental as it blends existing numerical and machine learning tools.
The authors tackled the problem of automatically identifying nonlinear dynamical systems from data by combining multi-step time-stepping schemes with deep neural networks, achieving accurate learning of dynamics, forecasting, and identification of basins of attraction for benchmark problems like the Lorenz system and fluid flow behind a cylinder.
The process of transforming observed data into predictive mathematical models of the physical world has always been paramount in science and engineering. Although data is currently being collected at an ever-increasing pace, devising meaningful models out of such observations in an automated fashion still remains an open problem. In this work, we put forth a machine learning approach for identifying nonlinear dynamical systems from data. Specifically, we blend classical tools from numerical analysis, namely the multi-step time-stepping schemes, with powerful nonlinear function approximators, namely deep neural networks, to distill the mechanisms that govern the evolution of a given data-set. We test the effectiveness of our approach for several benchmark problems involving the identification of complex, nonlinear and chaotic dynamics, and we demonstrate how this allows us to accurately learn the dynamics, forecast future states, and identify basins of attraction. In particular, we study the Lorenz system, the fluid flow behind a cylinder, the Hopf bifurcation, and the Glycoltic oscillator model as an example of complicated nonlinear dynamics typical of biological systems.